survival/0000755000175100001440000000000012545156350012151 5ustar hornikuserssurvival/inst/0000755000175100001440000000000012545002416013117 5ustar hornikuserssurvival/inst/CITATION0000644000175100001440000000111312470203267014254 0ustar hornikusersbibentry(bibtype="Manual", title = "A Package for Survival Analysis in S", author= person(c("Terry M"), "Therneau"), year =2015, note ="version 2.38", url="http://CRAN.R-project.org/package=survival", key= "survival-package" ) bibentry(bibtype= "Book", title="Modeling Survival Data: Extending the {C}ox Model", author=c(person(c("Terry M.", "Therneau")), person(c("Patricia M.", "Grambsch"))), year = "2000", publisher= "Springer", address = "New York", isbn = "0-387-98784-3", key = "survival-book" ) survival/inst/NEWS.Rd.orig0000644000175100001440000014400412514444605015132 0ustar hornikusers\name{NEWS} \title{NEWS file for the survival package} \section{Changes in version 2.38-2}{ \itemize{ \item Fix an error in the tmerge function. Using the options argument would generate a spurious error. \item Add the number of observations used and deleted due to missing to summary.pyears. \item Allow the combination of a null coxph model (~1 on the right) and the exact calculation for tied times. No one had ever asked for this before. (2015/3/25) \item Shorten the default printout for survfit. The records, n.max and n.start columns are often the same: if so suppress duplicates. \item Move the anova.coxphlist function from the survival package to coxme. (2015/3/3) \item Change the logLik method for coxph models so that the nobs component is the number of events rather than the number of rows in the data. This is superior for follow on methods such as AIC. }} \section{Changes in version 2.38-1}{ \itemize{ \item Fix an error discovered by CRAN, which triggered a core dump for them on a particular manual page (but never for me). The linear predictors from a frailty model contained NA values (incorrect), leading to failure in survConcordance.fit. (2015/2/16). \item An error was found in the mgus data set (a progression after death). Now corrected, and added a little more follow-up time for some subjects. \item Add error check for infitinte weights or offsets. This in respose to a bug report where someone did this on purpose, trying to mimic cure fractions, and then complained that survfit.coxph failed. \item Robust variance is not supported for a coxph model with the "exact" approximation. (Rarely requested and a lot of work to add.) Add an error message to clogit(), so users get a more useful notice of the issue rather than a late error from residuals.coxph. \item Update the rats data set: it now includes both female and male litters so as to match the documentation. \item The term frailty(x) would fail if x were a factor, and not all levels were present. Pointed out by Theodor Balan. \item Fix error of "abs" instead of "fabs" in the agfit4.c code; pointed out to me by CRAN. \item Replace all instances of the obsolete prmatrix function. \item Modify pyears to allow cbind(time, count) as the response, giving a cumulative sum of counts, when the counts per observation may be other than 0/1. \item The lines.survfit function was incorrect for data sets that used the start.time option and xscale (it neglected to rescale the start time.) \item An increasingly common error is for user to put the time variable on both sides of a coxph equation in the mistaken belief that this is a way to create time-dependent coefficients. Generate a warning message for this case. \item Update the basehaz function to a simple alias for "survfit". Prior versions called surfit but then only returned part of the object. Update 2/2015: reverted the change. It turns out that 6 different packages that depend on survival also depended on the old behavior. \item Make the default value for the shortlabel argument of strata() more nuanced. If the argument is a single factor, assume that we don't need to prepend the variable name to its levels. \item Return the weights vector, if present, as part of the survreg object. \item For interval censored points and the symmetric distributions (Gaussian and logistic) response type residuals were incorrect. Silly error: needed (x-mean)/scale not x/scale - mean. \item Martingale residuals could be incorrect for the case of model with (start, stop] data and a pspline term. Refactor the code so that all of the possible code paths call the same C routine to do the residuals. Add a new test for this case, and further tests to verify that predict(type='expected') and residuals agree. \item Fix bug pointed out by D Dunker: if a model had both tt() and cluster() terms it would fail with a length error. \item Fix a rare bug in plot.survfit: if a multistate curve rose and then later fell to exactly the same value, the line would be incorrect. \item Add calls to the R_CheckUserInterrupt to several routines, so that long calculations can be interrupted by the user. \item The anova.coxph function would fail if the original call had a subset argument. Pointed out by R Fisher. 11May2014 }} \section{Changes in version 2.37-7}{ \itemize{ \item Remove a dependency on the survey package from the adjusted survival curves vignette, at the request of CRAN. (The base + required bundle needs to be capable of a stand-alone build.) \item Fix error in calcuation of the y-axis range for survival curve plots whenever the "fun" argument could produce infinite values, e.g., complimentary log-log plots transform 1 to -Inf. Pointed out by Eva Boj del Val. (Add finite=TRUE to range() call). }} \section{Changes in version 2.37-6}{ \itemize{ \item The plot for competing risk curves could have a spurious segment. (Found within 3 hours of submitting 2.37-5 to CRAN.) \item The lines method for survexp objects was defaulting to a step function, restore the documented default of a connected line. \item Add a levels method for tcut objects. 14Jan2014 } } \section{Changes in version 2.37-5}{ \itemize{ \item Add vignette on adjusted survival curves. \item Add vignette concerning "type 3" tests. \item Make the tt() function invisible outside of a coxph formula. There was a complaint about conflicts with another package, and there is not really a good reason to have it be a global name. An R-devel discussion just over 1 year ago showed how to accomplish this. \item The modeling routines are set in two parts, e.g., coxph sets up the model and coxph.fit does the work. Export more of the ".fit" routines to make it easier for other packages to build on top of this one. \item Updates to the model.matrix and model.frame logic for coxph. A note from F Harrell showed that I was not correctly dealing with the "assign" attribute when there are strata * factor interactions. This led to cleanup in other cases that I had missed but which never had proven fatal. Also added support for tt() terms to the stand alone model.matrix and model.frame functions. (Residuals for tt models are still not available, but this was a necessary first step to that end.) 26Dec13 \item The Surv function now remembers attributes of the input variables that were passed to it; they are saved as "inputAttributes". This allows the rms package, for instance, to retain labels and units through the call. \item Update summary.coxph.penal to produce an object, which in turn has a print method, i.e., make it a "standard" summary function. \item Add a logLik method for coxph and survfit objects. \item Allow for Inf as the end of the time interval, for interval censored data in the Surv function. \item The predict.coxph function would fail if it had both a newdata and a collapse argument. Pointed out by Julian Bothe. 25Sep13 \item Survexp can now produce expecteds based on a stratified Cox model. Add the 'individual.s' and 'individual.h' options to return indivudual survival and cumulative hazard estimates, respectively. The result of survfit now (sometimes) includes the cumulative hazard. This will be expanded. 29Jul13 \item Change code in the coxpenal.fit routine: the use of a vector of symbols as arguments to my .C calls was confusing to a new CRAN consistency check. Both the old and new are legal R; but the old was admittedly an unusual construction and it was simpler to change it. \item Fix a bug in survfit.coxph pointed out by Chris Andrews, whose root cause was incorrect curve labels when the id option is used. 27Jun13 \item Add rsurvreg routine. \item Change survfit.coxph routine so that it detects whether newdata contains or does not contain strata variables, and acts accordingly. If newdata does containe strata then the output will contain only those data-value and strata combinations specified by the user. Retain strata levels in the coxph routine for use in the survfit routine, to correctly reconstruct strata levels. Warn about curves with interactions. 18Ju13 \item Add a dim method for survival curves. \item For competing risks curves that use the istate option, the plotted curves now start with the correct (initial) prevalence of each state. 22May13 \item The survreg function failed with the "robust=T" option. Pointed out by Jon Peck. Test case added. 6May13 \item Kazuki Yoshida pointed out that rep() had no method for Surv objects. This caused the survSplit routine to fail if the data frame contained a Surv object. 3May13 \item Per a request from Milan Bouchet-Valet fix an issue in survfit that arose when the OutDec option is set to ',': it did not correctly convert times back from character to numeric. \item The plot.survfit function now obeys "cex" for the size of the marks used for censored observations. }} \section{Changes in version 2.37-4}{ \itemize{ \item Subscripting error in predict.coxph for type=expected, se=T, strata in the model, newdata, and multiple strata in the new data set. Pointed out by Chris Andrews. The test program has been tweaked to include multiple strata in newdata. }} \section{Changes in version 2.37-3}{ \itemize{ \item Minor flaw in [.survfit. If "fit" had multiple curves, and fit$surv was a matrix, and one of those curves had only a single observation time, fit[i,] would collapse columns when "i" selected that curve, though it shouldn't. \item Changed all of the .C and .Call statements to make use of "registered native routines", per R-core request. Add file src/init.c \item Error in plot.survfit pointed out by K Hoggart -- the "+" signs for censored observations were printing one survival time to the left of the proper spot. Eik Vettorazi found another error if mark.time is a vector of numerics. These are the results of merging the code for plot, lines and points due to some discrepancies between them, plus not having any graphical checks in the test suite. \item Repair an error in using double subscripts for the survfitms objects. \item Add the US population data set, with yearly totals by age and sex for 2000 onward. It is named uspop2, since there is already a "uspop" data set containing decennial totals from 1790 to 1970. \item Not all combinations of strata Y/N and CI Y/N worked in the quantile.survfit function, pointed out by Daniel Wallschlaeger (missing a function argument in one if-else combination). Added a new test routine that verifies all paths. \item The first example in predict.survreg help file needed to have \code{I(age^2)} instead of \code{age^2} in the model: R ignores the second form. (I'm almost sure this worked at one time, perhaps in Splus). It also needed different plot symbols to actually match the referenced figure. Pointed out by Evan Newell. \item Fix a long-standing problem with cch pointed out by Ornulf Borgan leading to incorrect standard errors. A check in the underlying coxph routines to deal with out of bounds exponents, added in version 2.36-6, interacted badly with the -100 offset used in cch. It only affected models using (start, stop) survival times. }} \section{Changes in version 2.37-2}{ \itemize{ \item Two bugs were turned up by running tests for all the packages that depend on survival (158 of them). }} \section{Changes in version 2.37-1}{ \itemize{ \item Add a new multi-state type to the Surv object. Update the survfit routine to work with it. The major change is addition of a proper variance for this case. More functionality is planned. \item Remove the fr_colon.R test program. It tests an ability that has been superseded by coxme, on a numerically touchy data set, and it was slow besides. For several other tests that produce warning messages and are supposed to produce said messages, add extra comments to that effect so testers will know it is expected. \item The code has had several "if.R" clauses to accomodate Splus vs R differences, which are mostly class vs oldClass. These are now being removed as I encounter them; since our institution no longer uses Splus I can no longer test the clauses' validity. \item The fast subsets routine coxexact.fit incorrectly returned the linear predictor vector in the (internal) sorted order rather than data set order. Pointed out by Tatsuki Koyama, affecting the result of a clogit call. 6Nov2012 \item Jason Law pointed out that the sample data set "rats" is from the paper by Mantel et.al, but the documentation was for a data set from Gail, Santner and Brown. Added the Gail data as rats2 and fixed the documentation for rats. \item For predict.coxph with type="terms", use "sample" as the default value for the reference option. For all others the default remains "strata", the current value. Type terms are nearly always passed forward for further manipulation and per strata centering can mess things up: termplot() for instance will no longer show a smooth function if the results are recentered within strata. \item Fix bug in summary.aareg, which was unhappy (without cause) if the maxtime option was used for a fit that did not include the dfbeta option. Pointed out by Asa Johannesen. \item The coxph fitting functions would report an error for a null model (no X variables) if init was specified as numeric(0) rather than NULL. \item Update the description and citation files to use the new "person" function described in the R Journal. Also add the ByteCompile directive per suggestion of R core. \item Allow an ordinary vector as the left hand side of survConcordance. \item Update anova.coxphlist to reject models with a robust variance. \item The survfit function had an undocumented backwards-compatability that allows the newdata argument to be a vector with no names. An example from Damon Krstajic showed that this does not work when the original model has a matrix in the formula. Removed the feature. (This is for survfit.coxph.) Also clarified the code and its documentation about what is found where -- environments, formulas, and the arguments of eval, which fixes a problem pointed out by xxx where the result of a Surv call is used in the coxph formula. \item Fix an issue in summary.survfit pointed out by Frank Harrell. The strata variable for the output always had its labels in sorted order, even when a factor creating the survival curves was otherwise. (This was due to a call to factor() in the code.) The print routine would then list curves in sorted order, which might well be contrary to the user's wishes. The curves were numerically correct. \item Add the anova.coxmelist function to the namespace so that it is visible. If someone has a list of models the first of which was a coxph fit and the list includes coxme fits, then anova.coxph will be the function called by R, and it will call anova.coxmelist. \item Fix a bug pointed out by Yi Zhang and Mickael Hartweg. If a coxph model used an offset, then a predicted survival curve that used newdata (and the offset variable of course) would be wrong, e.g. survival values > 1. A simple misplaced parenthesis was the cause. A recent paper by Langholz shows how to get absolute survival from case-control data using an offset, which seems to have suddenly made this feature popular. \item Per further interaction with Yi Zhang, a few items were missing from the S3methods in the NAMESPACE file: as.matrix.Surv, model.matrix.coxph, model.matrix.survreg, model.frame.survreg. }} \section{Changes in version 2.36-14}{ \itemize{ \item A supposedly cosmetic change to coxph in the last release caused formulas with a "." on the right hand side to fail. Fix this and add a case with "." to the test suite. } } \section{Changes in version 2.36-13}{ \itemize{ \item Add the anova.coxmelist function. This is in the survival package rather than in coxme since "anova(fit1, fit2)" is valid when fit1 is a coxph and fit2 a coxme object, a case which will cause this function to be called by way of anova.coxph. \item More work on "predvars" handling for the pspline function, when used in predict calls. Add a new test of this to the suite, and the makepredictcall method to the namespace. Fixes a bug pointed out by C Crowson. \item Deprecate the "robust" option of coxph. When there are multiple observations per subject it is almost surely the wrong thing to do, while adding a "cluster(id)" term does the correct thing. When there is only one obs per subject both methods work correctly. \item Add documentation of the output structure to the aareg help file. \item Change ratetableDate so that it still allows use of chron objects, but doesn't need the chron library. This eliminates a warning messge from the package checks, but is also a reasonable support strategy for a moribund package. (Some of the local users keep datasets for a long long time.) \item Fix a bug in summary.survfit for a multiple-strata survival object. If one of the curves had no data after application of the times argument, an output label was the wrong length. \item Fix a bug pointed out by Charles Berry: predict for a Cox model which has strata, and the strata is a factor with not all its levels represented in the data. I had a mistake in the subscripting logic: number of groups is not equal to max(as.integer(strata)). \item Changes to avoid overflow in the exponent made in 2.36-6 caused failure for one special usage: in case-cohort designs a dummy offset of -100 could be added to some observations. This was being rounded away. The solution is to 1: have coxsafe not truncate small exponents and 2: do not recenter user provided offset values. \item Fix bug in survfit.coxph. Due to an indexing error I would sometimes create a huge scratch vector midway through the calculations (size = max value of "id"); the final result was always correct however. Data set provided by Cindy Crowson which had a user id in the billions. \item Fix bug pointed out by Nicholas Horton: predictions of type expected, with newdata, from a Cox model without a strata statement would fail with "x not found". A misplaced parenthesis from an earlier update caused it to not recreate the X matrix even though it was needed later. Also add some further information to the predict manual page to clarify an issue with frailty terms. }} \section{Changes in version 2.36-12}{ \itemize{ \item Fix a bug in the new fast subsets code. The test suite had no examples of strata + lots of tied times, so of course that's the case where I had an indexing error. Add a test case using the clogit function, which exercises this. \item Further memory tuning for survexp. }} \section{Changes in version 2.36-11}{ \itemize{ \item Make survexp more efficient. The X matrix was being modified in several places, leading to multiple copies of the data. When the data set was large this would lead to a memory shortage. \item Cause anova.coxph to call anova.coxme when a list of models has both coxph and coxme objects. \item Add the quantile.survfit function. This allows a user to extract arbitrary quantiles from a fitted curve (and std err). \item Fix an error in predict.coxph. When the model had a strata and the newdata and reference="sample" arguments were used, it would (incorrectly) ask for a strata variable in the new data set. \item Incorporate the fast subsets algorithm of Gail et al, when using coxph with the "exact" option. The speed increase is profound though at the cost of some memory. Reflect this in the documentation for the clogit routine. Note that the fast computation is not yet implemented for (start,stop) coxph models. \item Change the C routine used by coxph.fit from .C to .Call semantics to improve memory efficiency, in particular fewer copies of the X matrix. \item Add scaling to the above routine. This was prompted by a user who had some variables with a 0-1 range and others that were 0 - 10^7, resulting in 0 digits of accuracy in the variance matrix. (Economics data). \item Comment out some code sections that are specific to Splus. This reduced the number of "function not found" warnings from R CMD check. }} \section{Changes in version 2.36-10}{ \itemize{ \item 30 Sept 2011: The na.action argument was being ignored in predict.coxph; pointed out by Cindy Crowson. \item The log-likelihood for survreg was incorrect when there are case weights in the model. The error is a fixed constant for any given data set, so had no impact on tests or inferences. The error and correction were pointed out by Robert Kusher. \item A variable name was incorrect in survpenal.fit. This was in a program path that had never been traversed until Carina Salt used survreg with a psline(..., method='aic') call, leading to a "variable not found" message. \item Punctuation error in psline made it impossible for a user to specify the boundary.knots argument. Pointed out by Brandon Stewart. \item Add an "id" variable to the output of survobrien. \item The survfitCI routine would fail for a curve with only one jump point (a matrix collapsed into a vector). \item Fix an error in survfit.coxph when the coxph model has both a strata by covariate interaction and a cluster statement. The cluster term was not dropped from the Terms object as it should have been, led to a spurious "variable not found" error. Pointed out by Eva Bouguen. \item If a coxph model with penalized terms (frailty, pspline) also had a redundant covariate, the linear predictor would be returned as NA. Pointed out by Pavel Krivitsky. }} \section{Changes in version 2.36-9}{ \itemize{ \item Due to a mistake in my script that submits to CRAN, the fix in 2.36-8 below was actually not propogated to the CRAN submission. \item Fix an error in the Cauchy example found in the survreg.distributions help page, pointed out by James Price. \item Update the coxph.getdata routine to use the model.frame.coxph and model.matrix.coxph methods. \item Add the concordance statistic to the printout for penalized models. }} \section{Changes in version 2.36-8}{ \itemize{ \item Unitialized variable in calcuation of the variance of the concordance. Found on platform cross-checking by Brian Ripley. \item Changed testci to use a fixed file of results from cmprsk rather than invoking that package on-the-fly. Suggested by the CRAN maintainers. } } \section{Changes in version 2.36-7}{ \itemize{ \item Due to changes in R 2.13 default printout, the results of many of the test programs change in trivial way (one more or fewer digits). Update the necessary test/___.Rout.save files. Per the core team's suggestion the dependency for the package is marked as >=2.13. }} \section{Changes in version 2.36-6}{ \itemize{ \item An example from A Drummond caused iteration failure in coxph: x=c(1,1,1,0,1, rep(0,35)), time=1:40, status=1. The first iteration overshoots the solution and lands on an almost perfectly linear part of the loglik surface, which made the second iteration go to a huge number and exp() overflows. A sanity check routine coxsafe is now invoked on all values of the linear predictor. \item 1 April: Fix minor bug in survfit. For left censored data where all the left censored are on the very left, it would give a spurious warning message when trying to create a 0 row matrix that it didn't need or use. Pointed out by Steve Su. \item 31 March 2011: One of the plots in the r_sas test was wrong (it's been a long time since I visually checked these). The error was in predict.survreg; it had not taken into account a change in R2.7.1: the intercept attribute is reset to 1 whenever one subscripts a terms object, leading to incorrect results for a model with "-1" in the formula and a strata(): the intercept returned when removing the strata. I used this opportunity to move most of the logic into model.frame.survreg and model.matrix.survreg functions. Small change to the model.frame.coxph and model.matrix.coxph functions due to a better understanding of xlevels processing. \item Round off error issue in survfit: it used both unique(time) and table(time), and the resulting number of unique values is not guarranteed to be the same for times that differ by a tiny amount. Now times are coverted to a factor first. Peter Savicky from the R core team provided a nice discussion of the issue and helped me clarify how best to deal with it. The prior fix of first rounding to 15 digits was good enough for almost every data set -- except the one found by a local user just last week. \item Round off error in print.survfit pointed out by Micheal Faye. If a survival value was .5 in truth, but .5- eps due to round off the printed median was wrong. But it was ok for .5+eps. Simple if-then logic error. \item Re-fix a bug in survfit. It uses both unique and table in various places, which do not round the same; I had added a pre-rounding step to the code. A data set from Fan Chun showed that I didn't round quite enough. But the prior rounding did work for a time of 2 vs (sqrt(2))^2: this bug is very hard to produce. I now use as.numeric(as.character(factor(x))), which induces exactly the same rounding as table, since it is the same compuation path. \item Further changes to pspline. The new Boundary.knots argument allows a user to set the boundary knots inside the range of data. Code for extrapolation outside that range was needed, essentially a copy of the code found in ns() for the same issue. Also added a psplineinverse function, which may be useful with certain tt() calls in coxph. \item 10 Mar 2011: Add the capablilty for time-dependent transformations to coxph, along with a small vignette describing use of the feature. This code is still incompletely incorporated in that the models work but other methods (residuals, predict, etc) are not yet defined. \item 8 Mar 2011: Expand the survConcordance function. The function now correctly handles strata and time dependent covariates, and computes a standard error for the estimate. All computation is based on a balanced binary tree strucure, which leads to computation in \eqn{O(n \log_2(n))}{O(n log(n))} time. The \code{coxph} function now adds concordance to its output, and \code{summary.coxph} displays the result. \item 8 Mar 2011: Add the "reference" option to predict.coxph, a feature and need pointed out by Stephen Bond. \item 4 Mar 2011: Add a makepredictcall method for pspline(), which in turn required addition of a Boundary.knots argument to the function. \item 25 Feb 2011: Bug in pyears pointed out by Norm Phillips. If a subject started out with "off table" time, their age was not incremented by that amount as they moved forward to the next "in table" cell of the result. This could lead to using the wrong expected rate from the rate table. } } \section{Changes in version 2.36-5}{ \itemize{ \item 20 Feb 2011: Update survConcordance to correctly handle case weights, time dependent covariates, and strata. \item 18 Feb 2011: Bug in predict.coxph found by a user (1 day after 36-4!). If the coxph call had a subset and predict used newdata, the subset clause was "remembered" in the newdata construction, which is not appropriate. }} \section{Changes in version 2.36-4}{ \itemize{ \item 17 Feb 2011: Fix to predict.coxph. A small typo that only was exercised if the coxph model had x=T. Discovered via induced error in the rankhazard package. Added lines to the test suite to test for this in the future. \item Removed some files from test and src that are no longer needed. \item Update the configure script per suggestion from Kurt H. }} \section{Changes in version 2.36-3}{ \itemize{ \item 13 Feb 2011: Add the rmap argument to pyears, as was done for survexp, and update the manual pages and examples. Fix one last bug in predict.coxph (na.action use). Passes all the tests for inclusion on the next R release. \item 8 Feb 2011: Change the name of the new survfit.coxph.fit routine to survfitcoxph.fit; R was mistaking it for a survfit method. Fix errors in predict.coxph when there is a newdata argument, including adding yet another test program. \item 1 Feb 2011: Fix bugs in coxph and survreg pointed out by Heinz Tuechler and dtdenes@cogpsyphy.hu, independently, that were the same wrong line in both programs. With interactions, a non-penalized term could be marked as penalized due to a mismatched vector length, leading to a spurious error message later in the code. \item 1 Feb 2011: Update survfit.coxph to handle the case of a strata by covariate interaction. All prior releases of the code did this wrong, but it is a very rare case (found by Frank Harrell). Added a new test routine coxsurv4. Also found a bug in [.survfit; for a curve with both strata and multiple columns, as produced by survfit.coxph, it could drop the n.censored item when subscripting. A minor issue was fixed in coxph: when iter=0 the output coefficient vector should be equal to the input even when the variance is singular. \item 30 Jan 2011: Move the noweb files to a top level directory, out of inst/. They don't need to be copied to binary installs. \item 22 Jan 2011: Convert the Changelog files to the new inst/NEWS.Rd format. \item 1 Jan 202011: The match.ratetable would fail when passed a data frame with a character variable. This was pointed out by Heinz Tuechler, who also did most of the legwork to find it. It was triggered by the first few lines of tests/jasa.R (expect <- ....) when options(stringsAsFactors=FALSE) is set. } } \section{Changes in version 2.36-2}{ \itemize{ \item 20 Dec 2010: Add more test cases for survfit.coxph, which led to significant updates in the code. \item 18 Nov 2010: Add nevent to the coxph output and printout in response to a long standing user request. \item 14 Dec 2010: Add an as.matrix method for Surv objects. \item 11 Nov 2010: The prior changes broke 5 packages: the dependencies form a bigger test suite than mine! 1. Survival curve for a coxph model with sparse frailty fit; fixed and added a new test case. 2. survexp could fail if called from within a function due to a scoping error. 3. "Tsiatis" was once a valid type (alias for 'aalen') for survfit.coxph; now removed from the documentation but the code needed to be backwards compatable. The other two conflicts were fixed in the packages that call survival. There are still issues with the rms package which I am working out with Frank H. } } \section{Changes in version 2.36-1}{ \itemize{ \item{27 Oct 2010: Finish corrections and test to the new code. It now passes the checks. The predict.coxph routine now does strata and standard errors correctly, factors propogate through to predictions, and numerous small errors are addressed. Predicted survival curves for a Cox model has been rewritten in noweb and expanded. Change the version number to 2.36-1.} \item{17 Oct 2010: Per a request from Frank Harrell (interaction with his library), survfit.coxph no longer reconstructs the model frame unless it really needs it: in some cases the 'x' and 'y' matrices may be sufficient, and may be saved in the result. Add an argument "mf" to model.matrix.coxph for more efficient interaction when a parent routine has already recovered the model frame. In general, we are trying to make use of model.matrix.coxph in many of the routines, so that the logic contained there (remove cluster() calls, pull out strata, how to handle intercepts) need not be replicated in multiple places.} \item{12 Oct 2010: Fix a bug in the modified lower limits for survfit (Dory & Korn). A logical vector was being inadvertently converted to numeric. Pointed out by Andy Mugglin. A new case was added to the test suite. } } } \section{Changes in version 2.35}{ \itemize{ \item{15 July 2010: Add a coxph method for the logLik function. This is used by the AIC function and was requested by a user.} \item{29 July 2010: Fix 2 bugs in pyears. The check for a US rate table was off (minor effect on calculations), and there was a call to julian which assumed that the origin argument could be a vector. } \item{21 July 2010: Fix a problem pointed out by a user: calling survfit with almost tied times, e.g., c(2, sqrt(2)^2), could lead to an inconsistent result. Some parts of the code saw these as 2 unique values per the unique() function, some as a single value using the results of table(). We now pre-round the input times to one less decimal digit than the max from .Machine$double.digits. Also added the noweb.R processing function from the coxme package, so that the noweb code can be extracted "on the fly" during installation using commands in the configure and cleanup scripts. } \item{11 July 2010: A rewrite of the majority of the survfit.coxph code. The primary benefits are 1: finally tracked down and eliminated the bug for standard errors of case weights + Cox survival + Efron method; 2: the individual=TRUE and FALSE options now use the same underlying code for curves, before there were some options valid only for one or the other; 3: code was rewritten using noweb with a considerable increase in documentation; 4: during the verification process some errors were found in the test suite and corrected, e.g., a typo in my book led to failure of an all.equal test in book4.R. Similar to the rewrite for survfit several years ago, the new code has far less use of .C to help transparency.} \item{21 May 2010: Fix bug in summary.survfit. For a survival curve from a Cox model with start,stop data, the 'times' argument would generate an error.} \item{24 May 2010: Fix an annoyance in summary.survfit. When the survival data had an event or censor at time 0 and summary is called with a times argument, then my constructed call to approx() would have duplicate x values. The answer was always right, but approx has begun to print a bothersome warning message. A small change to the constructed argument vector avoids it.} \item{7 April 2010: Minor bug pointed out by Fredrik Lundgren. In survfit if the method was KM (default) and error = Tsiatis an error message results. Simple fix: code went down the wrong branch.} \item{24 Feb 2010: Serious bug pointed out by Kevin Buhr. In Surv(time1, time2,stat) if there were i) missing values in time1 and/or time2, ii) illegal value sets with time1 >=time2, and iii) all the instances of ii do not preceed all the instances of i, then the wrong observation (not the illegal) will be thrown out. Repaired, and a new test added. Minor updates to 3 test files: survreg2, testci, ratetable.} \item{8 Feb 2010: Bug pointed out by Heinz Tuechler -- if a subscript was dropped from a rate table the 'type' attribute got dropped, e.g. survexp.usr[,1,,].} \item{26 Jan 2010: At the request of Alex Bokov, added the xmax, xscale, and fun arguments to points.survfit.} \item{26 Jan 2010: Fix bug pointed out by Thomas Lumley -- with case weights <1 a Cox model with (start, stop) input would inappropriately decide it needed to do step halving to find a solution, eventually failing to converge. It was treating a loglik >0 as an indication of failure, but such values arise for small case weights. Let L(w) be the loglik for a data set where everyone is given a weight of w, then L(w)= wL(1) - d log(w) where d=number of deaths in the data. For small enough w positivity of L(w) is certain.} \item{25 Jan 2010: Fix bug in summary.ratetable pointed out by Heinze Tuechler. Added a call to the function to the test suite as well.} \item{15 Dec 2009: Two users pointed out a bug that crept into survreg() with a cluster statement, when a t(x)%*%x was replaced with crossprod. A trivial fix, but in response I added another test that more formally checks the dfbeta residuals and found a major oversight for the case of multiple strata. } \item{14 Dec 2009: 1.Fix bug in frailty.xxx, if there is a missing value in the levels it gets counted by "length(unique(x))" (frailty is called before NA removal.) 2.SurvfitCI had an incorrect CI with case weights, and 3. in survreg a call to resid instead of residuals.survreg, before the class was attached.} \item{11 Nov 2009: The 'type' argument does not make sense for plot.survfit. (If type='p', should one plot the tops of the step function, the bottoms, or both?). Make it explicitly disallowed in response to an R-help query, rather than the confusing error message that currently arose.} \item{28 Oct 2009: The basehaz function would reorder the labels of the strata factor. Not a bug really, but a "why do this?" Unintended consequence of a character -> factor conversion.} \item{1 Oct 2009: Fix a bug pointed out by Ben Domingue. There was one if-then-else path into step-halving in the frailty.controldf routine that would refer to a non-existent variable. A very rarely followed path, obviously, and with the obvious fix. The mathematics of the update was fine.} \item{30 Sep 2009: For coxph and model.matrix.coxph, re-attach the attributues lost from the X matrix when the intercept is removed, i.e., X <- X[,1]. In particular, some downstream libraries depend on the assign attribute. For predict.coxph remove an earlier edit so that a single variable model + type='terms' returns a matrix, not a vector. This is expected by the termplot() function. It led to a whole lot of changes in the test suite results, though, due to more "matrix" printouts.} \item{4 Sep 2009: Added a model.matrix.coxph and model.frame.coxph methods. The model.matrix.default function ceased to work for coxph models sometime between R 2.9 and 2.9.2 (best guess). This wasn't picked up in the test suite but rather by failure of 3 packages that depend on survival. Also added a test. Update CRAN since this broke other's packages.} \item{20 Aug 2009: One more fix to predict.coxph. It needed to use delete.response(Terms) rather than Terms, so as to not look for (unnecessarily) the response variable when the newdata argment is used. Pointed out by Michael Conklin.} \item{17 Aug 2009: Small bug in survfit.coxph.null pointed out by Frank Harrell. The 'n' component would be missing if the input data included strata, i.e., the initial model had used x=TRUE. He also pointed out the fix.} \item{10 June 2009: Fix an error pointed out by Nick Reich, who was the first to use interval censored data + user defined distribution in survreg, jointly. There was no test case and creating one uncovered several errors (but only for this combination). All the error cases led to catastrophic failure, highlighting the extreme rarity of a user requesting this combination.} \item{2 June 2009: Surv(time1, time2, status, type='interval') would fail for an NA status code. Pointed out by Achim Zeilus.} \item{22 May 2009: Allow single subscripts to rate tables, e.g. survexp[1:10: . Returns a simple vector of values. The str() function does this to print out a short summary. Problem pointed out by Heinz Tuechler.} \item{21 May 2009: Create a test case for factor variables/newdata/predict for coxph and survreg. This led to a set of minor fixes; the code is now in line with the R standard for model functions. One consequence is that model.frame.coxph and model.frame.survreg are no longer needed, so have been removed.} \item{20 May 2009: The manual page for survfit was confusing, since it tries to document both the standard KM (formula method) and the coxph method. I've split them out so that now survfit documents only the basic method and points a user the appropriate specialized page.} \item{1 May 2009: The anova.coxph function was incorrect for models with a strata term. Fixed this, and made chisquare tests the default.} \item{22 April 2009: The coxph code had an override to iter and eps, making both of them more strict for a penalized model. However, the overall default values have changed over time, so that these lines actually decreased accuracy - the opposite of their intent. Removed the lines. Also removed the iter.miss and eps.miss components (on which this check depended) from coxph.control, which makes that function match its documentation.} } } \section{Changes in version 2.34 and earlier}{ \subsection{Merge of the TMT source code tree with the Lumley code tree}{ \itemize{ \item Issues/decisions in remerging the Mayo and R code: For most of routines, it was easier to start with the Lumley code and add the Therneau fixes. This is because Tom had expanded a lot of partial matches, e.g., fit$coef in the TT code vs fit$coefficients. Routines with substantial changes were, of course, a special case. The most common change is an is.R() construct to choose class vs oldClass. \item xtras.R: Move anova.coxph and anova.coxphlist to their own source files. The remainder of the code is R only. \item survsum: removed from package \item survreg.old: has been removed from the package \item survfit.s: Depreciate the "formula with no ~1" option Mayo code for [ allows for reordering curves Separate out the R "basehaz" function as a separate source file \item survfit.km.s: The major change of did not get copied into R, so lots of changes. R had "new.time" and Splus 'start.time' for the same argument. Allow them both as synonyms. The output structure also changed: adapt the new one. This is mostly some name changes in the components, removing unneeded redundancies created by a different programmer. \item survfit.coxph.s: TMT code finally fixed the "Can't (yet) to case weights" problem. There must have been 10 years been the intent and execution. \item survexp.s: Add "bareterms" function from R, which replaces a prior use of terms.inner (in Splus but not R). \item survdiff.s: R code had the old (incorrect) expected <- sum(1-offset), since corrected to sum(-log(offset)) . \item{summary.coxph.s: This was a mess, since Tom and I had independently made the addition of a print.summary.coxph function. Below, TMT means that it was the choice in the Splus code, TL means that it was the choice in R 1. Put the coef=T argument in the print function, not summary (TMT) 2. Change the output's name from coef to coefficients (suggestion of Peter Dalgaard). Also change one column name to Pr(>|z|) for R. 3. Remove last vestiges of a reference to the 'icc' component (TMT) 4. Do not include score, rscore, naive.var in the result (TL) 5. Do include loglik in the result (TMT) 6. Compute the test statistics (loglik, Wald, etc) in the summary function rather than in the print.summary function (TL) 7. Remove the digits option from summary, it belongs in print.summary. (neither)} \item{strata.s: R code added a sep argument, this is ok R changed the character string NA to as.character(NA). Not okay 1. won't work with Splus, 2. This is a label, designed for printing, and so it should be a character string. } \item{residuals.coxph.s: R had added type='partial'. (Which I'm not very partial to, from their statistical properties. But they are legal, and I assume that someone requested them).} \item{print.survfit.s: Rewritten as a part of the general survival rewrite. Created the function 'survmean' which does most of the work, and is shared by print and summary, so that the values from 'print' are now available. Fix the minmin function: min(NULL) gives NA in Splus, which is the right answer for a non-estimable median, but Inf in R. Explicitly deal with this case, and add a bunch of comments. R had the print.rmean option, this has been expanded to a more general rmean option that allows setting the cutoff point. R added a print.n option with 3 choices, my code includes all 3 in the output. } \item{lines.survfit.s: The S version has a new block of code for guessing "firstx" more intellegently when it is missing. (Or, one hopes is is more intellegent!)} \item{coxph.control.s: The R code had tighter tolerances (eps= 1e-9) than Splus (1e-4) and a higher iterationn count (20 vs 10). Set eps to 1e-8 and iter to 15, mostly bending to the world. The tighter iteration is defensible, but I still maintain that a Cox model that takes >10 iterations is not going to finish if you give it 100. The likelihood surface is almost perfectly quadratic near the minimum. (Not true for survreg by the way).} \item{: In Surv, the Mayo code creates NA's out of invalid status values or start,stop pairs, rather than a stop and error message. This is to allow for example coxph(Surv(time1,time2, status).... , subset=(goodlines)) succeed, when "goodlines" is the subset with correct values.} } } \subsection{Older changes}{ \itemize{ \item{25SepO7: How embarrassing -- someone pointed out that I had Dave Harrington's name spelled wrong in the options to survfit.coxph!} \item{9Jul07: In a model with offsets, survreg mistakenly omitted the offset from the returned linear.predictor component.} \item{10May07: Change summary.coxph so that it returns an object of class summary.coxph, and add a print method for that object.} \item{22Jun06: Update match.ratetable, so that more liberal matches are now allowed. For instance, 'F', 'f', 'female', 'fem', 'FEMA', etc are now all considered matches to the dimname "female" in survexp.us.} \item{26Apr06: Fix bug in summary.survfit, pointed out by Bob Treder. With the times option, the value of n.risk would be wrong for "in between" times; e.g., the data had events and/or censoring at times 10, 20,... and we asked for printout at time 15. It should give n.risk at time 20, it was returning the value at time 10. Interestingly, the code had a very careful treatment of this case, along with an example in the comments, and the "the right answer is" part of the comment was wrong! So the code correctly computed an incorrect answer. Added another test case to the test suite, survtest2.} \item{21Apr06: Fix problem in [.survfit, pointed out by Thomas Lumley. If fit <- survfit(Surv(time, status) ~ ph.ecog, lung), then fit[2:1] did not reorder the output correctly. I had never tested putting the subscripts in non-increasing order.} \item{7Feb06: Fix a problem in the coxph iteration (coxfit2.c, coxfit5, agfit3, agfit5, agexact). It will likely never catch anyone again, even if I didn't fix it. In a particular data set, beta overshot and step halving was invoked. During step halving, a loglik happened to occur that was within eps of the prior step's loglik --- and the routine decided, erroneously, that it had converged! (A nice quadratic curve, a first guess b1 to the left of the desired max of the curve. The next guess b2 overshot and ends up with a lower loglik, on the right side of the max. Back up to the midpoint of b1 and b2, and this guess, still to the right of the max (still too large) has EXACTLY the same value of y as b1 did, but on the other side of the max from b1. "Last two guesses give the same answer, I'm done" said the routine).} \item{27Sep05: Found and fixed a nasty bug in survfit. When method='fh2' and there were multiple groups I had a subscripting bug, leading to vectors that were supposed to be the same length, but weren't, passed into C. The resulting curves were obviously wrong -- survival precipitously drops to zero.} \item{5May05: Add the drop=F arg to one subscripting selection in survfit.coxph. temp <- (matrix(surv$y, ncol=3))[ntime,,drop=F] If you selected only 1 time point (1 row) in the final output, the code would fail. Pointed out by Cindy Crowson.} \item{18Apr05: Bug in survfit.turnbull. The strata variable was not being filled in (number of points per curve). So if multiple curves were generated at once, i.e., with something on the right hand side of ~ in the formula, all the downstream print/plot functions would not work with the result.} \item{8Feb05: Fix small typo in is.ratetable, introduced on 24Nov04: (Today was the first time I added to the standard library, and thus ended up using the non-verbose mode.)} \item{8Feb05: Add the data.frame argument to pyears. This causes the output to contain a dataframe rather than a set of arrays. It is useful for further processing of the data using Poisson regression.} \item{7Feb05: Modified print.ratetable to be more useful. It now tells about the ratetable, rather than printing all of its values.} \item{8Dec04: Fix a small bug in survfit.turnbull. If there are people left censored before the first time point of any other kind (interval, exact, or right censored), the the plotted height of the curve from "rightmost left censoring time" to "leftmost event time", that is the flat tail on the left, was at the wrong height. Added another test to testreg/reliability.s for this.} \item{24Nov04: Change is.ratetable to give longer messages} } } } survival/inst/NEWS.Rd0000644000175100001440000014667412545002416014204 0ustar hornikusers\name{NEWS} \title{NEWS file for the survival package} \section{Changes in version 2.38-3}{ \itemize{ \item Add more imports to the NAMESPACE file per a request from CRAN } } \section{Changes in version 2.38-2}{ \itemize{ \item Change print.coxph to use the printCoefmat routine, which l leads to nicer p-values. Other print routines will follow unless there is an outcry. (But I forced signif.stars=FALSE: my tolerance of bad practice has limits.) \item Make those parts of the competing risks vignette which depend on the cmprsk library conditional. Otherwise the build fails for those without the pacakge. \item The coxph function could fail converge for a set of very collinear predictors when using (start, stop) data; revealed in a test case sent by G Borstrom. This was due to deficiency in a check for near infinite coefficients, which had already been updated for some but not all cases. (2015/6/3) \item Update anova.coxph to use the model.frame.coxph function; the current code had scoping errors if embedded in a function. Add an anova.coxph.penal function to correctly handle models with pspline terms. \item Fix an error in the tmerge function. Using the options argument would generate a spurious error. \item Pyears could fail on very long formulas due to a deparse() issue. \item Add the number of observations used and deleted due to missing to summary.pyears. \item Allow the combination of a null coxph model (~1 on the right) and the exact calculation for tied times. No one had ever asked for this before. (2015/3/25) \item Shorten the default printout for survfit. The records, n.max and n.start columns are often the same: if so suppress duplicates. \item Move the anova.coxphlist function from the survival package to coxme. (2015/3/3) \item Change the logLik method for coxph models so that the nobs component is the number of events rather than the number of rows in the data. This is superior for follow on methods such as AIC. \item Add a test to the coxexact.c routine for too large a data set; too many tied times could lead to integer overflow. "Fixing" the error is not sensible: the computation for such a data set would take decades. Add some more explanation to the help pages as well. }} \section{Changes in version 2.38-1}{ \itemize{ \item Fix an error discovered by CRAN, which triggered a core dump for them on a particular manual page (but never for me). The linear predictors from a frailty model contained NA values (incorrect), leading to failure in survConcordance.fit. (2015/2/16). \item An error was found in the mgus data set (a progression after death). Now corrected, and added a little more follow-up time for some subjects. \item Add error check for infitinte weights or offsets. This in respose to a bug report where someone did this on purpose, trying to mimic cure fractions, and then found that survfit.coxph failed. \item Robust variance is not supported for a coxph model with the "exact" approximation. (Rarely requested and a lot of work to add.) Add an error message to clogit(), so users get a more useful notice of the issue rather than a late error from residuals.coxph. \item Update the rats data set: it now includes both female and male litters so as to match the documentation. \item The term frailty(x) would fail if x were a factor, and not all levels were present. Pointed out by Theodor Balan. \item Fix error of "abs" instead of "fabs" in the agfit4.c code; pointed out to me by CRAN. \item Replace all instances of the obsolete prmatrix function. \item Modify pyears to allow cbind(time, count) as the response, giving a cumulative sum of counts, when the counts per observation may be other than 0/1. \item The lines.survfit function was incorrect for data sets that used the start.time option and xscale (it neglected to rescale the start time.) \item An increasingly common error is for user to put the time variable on both sides of a coxph equation in the mistaken belief that this is a way to create time-dependent coefficients. Generate a warning message for this case. \item Update the basehaz function to a simple alias for "survfit". Prior versions called surfit but then only returned part of the object. Update 2/2015: reverted the change. It turns out that 6 different packages that depend on survival also depended on the old behavior. \item Make the default value for the shortlabel argument of strata() more nuanced. If the argument is a single factor, assume that we don't need to prepend the variable name to its levels. \item Return the weights vector, if present, as part of the survreg object. \item For interval censored points and the symmetric distributions (Gaussian and logistic) response type residuals were incorrect. Silly error: needed (x-mean)/scale not x/scale - mean. \item Martingale residuals could be incorrect for the case of model with (start, stop] data and a pspline term. Refactor the code so that all of the possible code paths call the same C routine to do the residuals. Add a new test for this case, and further tests to verify that predict(type='expected') and residuals agree. \item Fix bug pointed out by D Dunker: if a model had both tt() and cluster() terms it would fail with a length error. \item Fix a rare bug in plot.survfit: if a multistate curve rose and then later fell to exactly the same value, the line would be incorrect. \item Add calls to the R_CheckUserInterrupt to several routines, so that long calculations can be interrupted by the user. \item The anova.coxph function would fail if the original call had a subset argument. Pointed out by R Fisher. 11May2014 }} \section{Changes in version 2.37-7}{ \itemize{ \item Remove a dependency on the survey package from the adjusted survival curves vignette, at the request of CRAN. (The base + required bundle needs to be capable of a stand-alone build.) \item Fix error in calcuation of the y-axis range for survival curve plots whenever the "fun" argument could produce infinite values, e.g., complimentary log-log plots transform 1 to -Inf. Pointed out by Eva Boj del Val. (Add finite=TRUE to range() call). }} \section{Changes in version 2.37-6}{ \itemize{ \item The plot for competing risk curves could have a spurious segment. (Found within 3 hours of submitting 2.37-5 to CRAN.) \item The lines method for survexp objects was defaulting to a step function, restore the documented default of a connected line. \item Add a levels method for tcut objects. 14Jan2014 } } \section{Changes in version 2.37-5}{ \itemize{ \item Add vignette on adjusted survival curves. \item Add vignette concerning "type 3" tests. \item Make the tt() function invisible outside of a coxph formula. There was a complaint about conflicts with another package, and there is not really a good reason to have it be a global name. An R-devel discussion just over 1 year ago showed how to accomplish this. \item The modeling routines are set in two parts, e.g., coxph sets up the model and coxph.fit does the work. Export more of the ".fit" routines to make it easier for other packages to build on top of this one. \item Updates to the model.matrix and model.frame logic for coxph. A note from F Harrell showed that I was not correctly dealing with the "assign" attribute when there are strata * factor interactions. This led to cleanup in other cases that I had missed but which never had proven fatal. Also added support for tt() terms to the stand alone model.matrix and model.frame functions. (Residuals for tt models are still not available, but this was a necessary first step to that end.) 26Dec13 \item The Surv function now remembers attributes of the input variables that were passed to it; they are saved as "inputAttributes". This allows the rms package, for instance, to retain labels and units through the call. \item Update summary.coxph.penal to produce an object, which in turn has a print method, i.e., make it a "standard" summary function. \item Add a logLik method for coxph and survfit objects. \item Allow for Inf as the end of the time interval, for interval censored data in the Surv function. \item The predict.coxph function would fail if it had both a newdata and a collapse argument. Pointed out by Julian Bothe. 25Sep13 \item Survexp can now produce expecteds based on a stratified Cox model. Add the 'individual.s' and 'individual.h' options to return indivudual survival and cumulative hazard estimates, respectively. The result of survfit now (sometimes) includes the cumulative hazard. This will be expanded. 29Jul13 \item Change code in the coxpenal.fit routine: the use of a vector of symbols as arguments to my .C calls was confusing to a new CRAN consistency check. Both the old and new are legal R; but the old was admittedly an unusual construction and it was simpler to change it. \item Fix a bug in survfit.coxph pointed out by Chris Andrews, whose root cause was incorrect curve labels when the id option is used. 27Jun13 \item Add rsurvreg routine. \item Change survfit.coxph routine so that it detects whether newdata contains or does not contain strata variables, and acts accordingly. If newdata does containe strata then the output will contain only those data-value and strata combinations specified by the user. Retain strata levels in the coxph routine for use in the survfit routine, to correctly reconstruct strata levels. Warn about curves with interactions. 18Ju13 \item Add a dim method for survival curves. \item For competing risks curves that use the istate option, the plotted curves now start with the correct (initial) prevalence of each state. 22May13 \item The survreg function failed with the "robust=T" option. Pointed out by Jon Peck. Test case added. 6May13 \item Kazuki Yoshida pointed out that rep() had no method for Surv objects. This caused the survSplit routine to fail if the data frame contained a Surv object. 3May13 \item Per a request from Milan Bouchet-Valet fix an issue in survfit that arose when the OutDec option is set to ',': it did not correctly convert times back from character to numeric. \item The plot.survfit function now obeys "cex" for the size of the marks used for censored observations. }} \section{Changes in version 2.37-4}{ \itemize{ \item Subscripting error in predict.coxph for type=expected, se=T, strata in the model, newdata, and multiple strata in the new data set. Pointed out by Chris Andrews. The test program has been tweaked to include multiple strata in newdata. }} \section{Changes in version 2.37-3}{ \itemize{ \item Minor flaw in [.survfit. If "fit" had multiple curves, and fit$surv was a matrix, and one of those curves had only a single observation time, fit[i,] would collapse columns when "i" selected that curve, though it shouldn't. \item Changed all of the .C and .Call statements to make use of "registered native routines", per R-core request. Add file src/init.c \item Error in plot.survfit pointed out by K Hoggart -- the "+" signs for censored observations were printing one survival time to the left of the proper spot. Eik Vettorazi found another error if mark.time is a vector of numerics. These are the results of merging the code for plot, lines and points due to some discrepancies between them, plus not having any graphical checks in the test suite. \item Repair an error in using double subscripts for the survfitms objects. \item Add the US population data set, with yearly totals by age and sex for 2000 onward. It is named uspop2, since there is already a "uspop" data set containing decennial totals from 1790 to 1970. \item Not all combinations of strata Y/N and CI Y/N worked in the quantile.survfit function, pointed out by Daniel Wallschlaeger (missing a function argument in one if-else combination). Added a new test routine that verifies all paths. \item The first example in predict.survreg help file needed to have \code{I(age^2)} instead of \code{age^2} in the model: R ignores the second form. (I'm almost sure this worked at one time, perhaps in Splus). It also needed different plot symbols to actually match the referenced figure. Pointed out by Evan Newell. \item Fix a long-standing problem with cch pointed out by Ornulf Borgan leading to incorrect standard errors. A check in the underlying coxph routines to deal with out of bounds exponents, added in version 2.36-6, interacted badly with the -100 offset used in cch. It only affected models using (start, stop) survival times. }} \section{Changes in version 2.37-2}{ \itemize{ \item Two bugs were turned up by running tests for all the packages that depend on survival (158 of them). }} \section{Changes in version 2.37-1}{ \itemize{ \item Add a new multi-state type to the Surv object. Update the survfit routine to work with it. The major change is addition of a proper variance for this case. More functionality is planned. \item Remove the fr_colon.R test program. It tests an ability that has been superseded by coxme, on a numerically touchy data set, and it was slow besides. For several other tests that produce warning messages and are supposed to produce said messages, add extra comments to that effect so testers will know it is expected. \item The code has had several "if.R" clauses to accomodate Splus vs R differences, which are mostly class vs oldClass. These are now being removed as I encounter them; since our institution no longer uses Splus I can no longer test the clauses' validity. \item The fast subsets routine coxexact.fit incorrectly returned the linear predictor vector in the (internal) sorted order rather than data set order. Pointed out by Tatsuki Koyama, affecting the result of a clogit call. 6Nov2012 \item Jason Law pointed out that the sample data set "rats" is from the paper by Mantel et.al, but the documentation was for a data set from Gail, Santner and Brown. Added the Gail data as rats2 and fixed the documentation for rats. \item For predict.coxph with type="terms", use "sample" as the default value for the reference option. For all others the default remains "strata", the current value. Type terms are nearly always passed forward for further manipulation and per strata centering can mess things up: termplot() for instance will no longer show a smooth function if the results are recentered within strata. \item Fix bug in summary.aareg, which was unhappy (without cause) if the maxtime option was used for a fit that did not include the dfbeta option. Pointed out by Asa Johannesen. \item The coxph fitting functions would report an error for a null model (no X variables) if init was specified as numeric(0) rather than NULL. \item Update the description and citation files to use the new "person" function described in the R Journal. Also add the ByteCompile directive per suggestion of R core. \item Allow an ordinary vector as the left hand side of survConcordance. \item Update anova.coxphlist to reject models with a robust variance. \item The survfit function had an undocumented backwards-compatability that allows the newdata argument to be a vector with no names. An example from Damon Krstajic showed that this does not work when the original model has a matrix in the formula. Removed the feature. (This is for survfit.coxph.) Also clarified the code and its documentation about what is found where -- environments, formulas, and the arguments of eval, which fixes a problem pointed out by xxx where the result of a Surv call is used in the coxph formula. \item Fix an issue in summary.survfit pointed out by Frank Harrell. The strata variable for the output always had its labels in sorted order, even when a factor creating the survival curves was otherwise. (This was due to a call to factor() in the code.) The print routine would then list curves in sorted order, which might well be contrary to the user's wishes. The curves were numerically correct. \item Add the anova.coxmelist function to the namespace so that it is visible. If someone has a list of models the first of which was a coxph fit and the list includes coxme fits, then anova.coxph will be the function called by R, and it will call anova.coxmelist. \item Fix a bug pointed out by Yi Zhang and Mickael Hartweg. If a coxph model used an offset, then a predicted survival curve that used newdata (and the offset variable of course) would be wrong, e.g. survival values > 1. A simple misplaced parenthesis was the cause. A recent paper by Langholz shows how to get absolute survival from case-control data using an offset, which seems to have suddenly made this feature popular. \item Per further interaction with Yi Zhang, a few items were missing from the S3methods in the NAMESPACE file: as.matrix.Surv, model.matrix.coxph, model.matrix.survreg, model.frame.survreg. }} \section{Changes in version 2.36-14}{ \itemize{ \item A supposedly cosmetic change to coxph in the last release caused formulas with a "." on the right hand side to fail. Fix this and add a case with "." to the test suite. } } \section{Changes in version 2.36-13}{ \itemize{ \item Add the anova.coxmelist function. This is in the survival package rather than in coxme since "anova(fit1, fit2)" is valid when fit1 is a coxph and fit2 a coxme object, a case which will cause this function to be called by way of anova.coxph. \item More work on "predvars" handling for the pspline function, when used in predict calls. Add a new test of this to the suite, and the makepredictcall method to the namespace. Fixes a bug pointed out by C Crowson. \item Deprecate the "robust" option of coxph. When there are multiple observations per subject it is almost surely the wrong thing to do, while adding a "cluster(id)" term does the correct thing. When there is only one obs per subject both methods work correctly. \item Add documentation of the output structure to the aareg help file. \item Change ratetableDate so that it still allows use of chron objects, but doesn't need the chron library. This eliminates a warning messge from the package checks, but is also a reasonable support strategy for a moribund package. (Some of the local users keep datasets for a long long time.) \item Fix a bug in summary.survfit for a multiple-strata survival object. If one of the curves had no data after application of the times argument, an output label was the wrong length. \item Fix a bug pointed out by Charles Berry: predict for a Cox model which has strata, and the strata is a factor with not all its levels represented in the data. I had a mistake in the subscripting logic: number of groups is not equal to max(as.integer(strata)). \item Changes to avoid overflow in the exponent made in 2.36-6 caused failure for one special usage: in case-cohort designs a dummy offset of -100 could be added to some observations. This was being rounded away. The solution is to 1: have coxsafe not truncate small exponents and 2: do not recenter user provided offset values. \item Fix bug in survfit.coxph. Due to an indexing error I would sometimes create a huge scratch vector midway through the calculations (size = max value of "id"); the final result was always correct however. Data set provided by Cindy Crowson which had a user id in the billions. \item Fix bug pointed out by Nicholas Horton: predictions of type expected, with newdata, from a Cox model without a strata statement would fail with "x not found". A misplaced parenthesis from an earlier update caused it to not recreate the X matrix even though it was needed later. Also add some further information to the predict manual page to clarify an issue with frailty terms. }} \section{Changes in version 2.36-12}{ \itemize{ \item Fix a bug in the new fast subsets code. The test suite had no examples of strata + lots of tied times, so of course that's the case where I had an indexing error. Add a test case using the clogit function, which exercises this. \item Further memory tuning for survexp. }} \section{Changes in version 2.36-11}{ \itemize{ \item Make survexp more efficient. The X matrix was being modified in several places, leading to multiple copies of the data. When the data set was large this would lead to a memory shortage. \item Cause anova.coxph to call anova.coxme when a list of models has both coxph and coxme objects. \item Add the quantile.survfit function. This allows a user to extract arbitrary quantiles from a fitted curve (and std err). \item Fix an error in predict.coxph. When the model had a strata and the newdata and reference="sample" arguments were used, it would (incorrectly) ask for a strata variable in the new data set. \item Incorporate the fast subsets algorithm of Gail et al, when using coxph with the "exact" option. The speed increase is profound though at the cost of some memory. Reflect this in the documentation for the clogit routine. Note that the fast computation is not yet implemented for (start,stop) coxph models. \item Change the C routine used by coxph.fit from .C to .Call semantics to improve memory efficiency, in particular fewer copies of the X matrix. \item Add scaling to the above routine. This was prompted by a user who had some variables with a 0-1 range and others that were 0 - 10^7, resulting in 0 digits of accuracy in the variance matrix. (Economics data). \item Comment out some code sections that are specific to Splus. This reduced the number of "function not found" warnings from R CMD check. }} \section{Changes in version 2.36-10}{ \itemize{ \item 30 Sept 2011: The na.action argument was being ignored in predict.coxph; pointed out by Cindy Crowson. \item The log-likelihood for survreg was incorrect when there are case weights in the model. The error is a fixed constant for any given data set, so had no impact on tests or inferences. The error and correction were pointed out by Robert Kusher. \item A variable name was incorrect in survpenal.fit. This was in a program path that had never been traversed until Carina Salt used survreg with a psline(..., method='aic') call, leading to a "variable not found" message. \item Punctuation error in psline made it impossible for a user to specify the boundary.knots argument. Pointed out by Brandon Stewart. \item Add an "id" variable to the output of survobrien. \item The survfitCI routine would fail for a curve with only one jump point (a matrix collapsed into a vector). \item Fix an error in survfit.coxph when the coxph model has both a strata by covariate interaction and a cluster statement. The cluster term was not dropped from the Terms object as it should have been, led to a spurious "variable not found" error. Pointed out by Eva Bouguen. \item If a coxph model with penalized terms (frailty, pspline) also had a redundant covariate, the linear predictor would be returned as NA. Pointed out by Pavel Krivitsky. }} \section{Changes in version 2.36-9}{ \itemize{ \item Due to a mistake in my script that submits to CRAN, the fix in 2.36-8 below was actually not propogated to the CRAN submission. \item Fix an error in the Cauchy example found in the survreg.distributions help page, pointed out by James Price. \item Update the coxph.getdata routine to use the model.frame.coxph and model.matrix.coxph methods. \item Add the concordance statistic to the printout for penalized models. }} \section{Changes in version 2.36-8}{ \itemize{ \item Unitialized variable in calcuation of the variance of the concordance. Found on platform cross-checking by Brian Ripley. \item Changed testci to use a fixed file of results from cmprsk rather than invoking that package on-the-fly. Suggested by the CRAN maintainers. } } \section{Changes in version 2.36-7}{ \itemize{ \item Due to changes in R 2.13 default printout, the results of many of the test programs change in trivial way (one more or fewer digits). Update the necessary test/___.Rout.save files. Per the core team's suggestion the dependency for the package is marked as >=2.13. }} \section{Changes in version 2.36-6}{ \itemize{ \item An example from A Drummond caused iteration failure in coxph: x=c(1,1,1,0,1, rep(0,35)), time=1:40, status=1. The first iteration overshoots the solution and lands on an almost perfectly linear part of the loglik surface, which made the second iteration go to a huge number and exp() overflows. A sanity check routine coxsafe is now invoked on all values of the linear predictor. \item 1 April: Fix minor bug in survfit. For left censored data where all the left censored are on the very left, it would give a spurious warning message when trying to create a 0 row matrix that it didn't need or use. Pointed out by Steve Su. \item 31 March 2011: One of the plots in the r_sas test was wrong (it's been a long time since I visually checked these). The error was in predict.survreg; it had not taken into account a change in R2.7.1: the intercept attribute is reset to 1 whenever one subscripts a terms object, leading to incorrect results for a model with "-1" in the formula and a strata(): the intercept returned when removing the strata. I used this opportunity to move most of the logic into model.frame.survreg and model.matrix.survreg functions. Small change to the model.frame.coxph and model.matrix.coxph functions due to a better understanding of xlevels processing. \item Round off error issue in survfit: it used both unique(time) and table(time), and the resulting number of unique values is not guarranteed to be the same for times that differ by a tiny amount. Now times are coverted to a factor first. Peter Savicky from the R core team provided a nice discussion of the issue and helped me clarify how best to deal with it. The prior fix of first rounding to 15 digits was good enough for almost every data set -- except the one found by a local user just last week. \item Round off error in print.survfit pointed out by Micheal Faye. If a survival value was .5 in truth, but .5- eps due to round off the printed median was wrong. But it was ok for .5+eps. Simple if-then logic error. \item Re-fix a bug in survfit. It uses both unique and table in various places, which do not round the same; I had added a pre-rounding step to the code. A data set from Fan Chun showed that I didn't round quite enough. But the prior rounding did work for a time of 2 vs (sqrt(2))^2: this bug is very hard to produce. I now use as.numeric(as.character(factor(x))), which induces exactly the same rounding as table, since it is the same compuation path. \item Further changes to pspline. The new Boundary.knots argument allows a user to set the boundary knots inside the range of data. Code for extrapolation outside that range was needed, essentially a copy of the code found in ns() for the same issue. Also added a psplineinverse function, which may be useful with certain tt() calls in coxph. \item 10 Mar 2011: Add the capablilty for time-dependent transformations to coxph, along with a small vignette describing use of the feature. This code is still incompletely incorporated in that the models work but other methods (residuals, predict, etc) are not yet defined. \item 8 Mar 2011: Expand the survConcordance function. The function now correctly handles strata and time dependent covariates, and computes a standard error for the estimate. All computation is based on a balanced binary tree strucure, which leads to computation in \eqn{O(n \log_2(n))}{O(n log(n))} time. The \code{coxph} function now adds concordance to its output, and \code{summary.coxph} displays the result. \item 8 Mar 2011: Add the "reference" option to predict.coxph, a feature and need pointed out by Stephen Bond. \item 4 Mar 2011: Add a makepredictcall method for pspline(), which in turn required addition of a Boundary.knots argument to the function. \item 25 Feb 2011: Bug in pyears pointed out by Norm Phillips. If a subject started out with "off table" time, their age was not incremented by that amount as they moved forward to the next "in table" cell of the result. This could lead to using the wrong expected rate from the rate table. } } \section{Changes in version 2.36-5}{ \itemize{ \item 20 Feb 2011: Update survConcordance to correctly handle case weights, time dependent covariates, and strata. \item 18 Feb 2011: Bug in predict.coxph found by a user (1 day after 36-4!). If the coxph call had a subset and predict used newdata, the subset clause was "remembered" in the newdata construction, which is not appropriate. }} \section{Changes in version 2.36-4}{ \itemize{ \item 17 Feb 2011: Fix to predict.coxph. A small typo that only was exercised if the coxph model had x=T. Discovered via induced error in the rankhazard package. Added lines to the test suite to test for this in the future. \item Removed some files from test and src that are no longer needed. \item Update the configure script per suggestion from Kurt H. }} \section{Changes in version 2.36-3}{ \itemize{ \item 13 Feb 2011: Add the rmap argument to pyears, as was done for survexp, and update the manual pages and examples. Fix one last bug in predict.coxph (na.action use). Passes all the tests for inclusion on the next R release. \item 8 Feb 2011: Change the name of the new survfit.coxph.fit routine to survfitcoxph.fit; R was mistaking it for a survfit method. Fix errors in predict.coxph when there is a newdata argument, including adding yet another test program. \item 1 Feb 2011: Fix bugs in coxph and survreg pointed out by Heinz Tuechler and dtdenes@cogpsyphy.hu, independently, that were the same wrong line in both programs. With interactions, a non-penalized term could be marked as penalized due to a mismatched vector length, leading to a spurious error message later in the code. \item 1 Feb 2011: Update survfit.coxph to handle the case of a strata by covariate interaction. All prior releases of the code did this wrong, but it is a very rare case (found by Frank Harrell). Added a new test routine coxsurv4. Also found a bug in [.survfit; for a curve with both strata and multiple columns, as produced by survfit.coxph, it could drop the n.censored item when subscripting. A minor issue was fixed in coxph: when iter=0 the output coefficient vector should be equal to the input even when the variance is singular. \item 30 Jan 2011: Move the noweb files to a top level directory, out of inst/. They don't need to be copied to binary installs. \item 22 Jan 2011: Convert the Changelog files to the new inst/NEWS.Rd format. \item 1 Jan 202011: The match.ratetable would fail when passed a data frame with a character variable. This was pointed out by Heinz Tuechler, who also did most of the legwork to find it. It was triggered by the first few lines of tests/jasa.R (expect <- ....) when options(stringsAsFactors=FALSE) is set. } } \section{Changes in version 2.36-2}{ \itemize{ \item 20 Dec 2010: Add more test cases for survfit.coxph, which led to significant updates in the code. \item 18 Nov 2010: Add nevent to the coxph output and printout in response to a long standing user request. \item 14 Dec 2010: Add an as.matrix method for Surv objects. \item 11 Nov 2010: The prior changes broke 5 packages: the dependencies form a bigger test suite than mine! 1. Survival curve for a coxph model with sparse frailty fit; fixed and added a new test case. 2. survexp could fail if called from within a function due to a scoping error. 3. "Tsiatis" was once a valid type (alias for 'aalen') for survfit.coxph; now removed from the documentation but the code needed to be backwards compatable. The other two conflicts were fixed in the packages that call survival. There are still issues with the rms package which I am working out with Frank H. } } \section{Changes in version 2.36-1}{ \itemize{ \item{27 Oct 2010: Finish corrections and test to the new code. It now passes the checks. The predict.coxph routine now does strata and standard errors correctly, factors propogate through to predictions, and numerous small errors are addressed. Predicted survival curves for a Cox model has been rewritten in noweb and expanded. Change the version number to 2.36-1.} \item{17 Oct 2010: Per a request from Frank Harrell (interaction with his library), survfit.coxph no longer reconstructs the model frame unless it really needs it: in some cases the 'x' and 'y' matrices may be sufficient, and may be saved in the result. Add an argument "mf" to model.matrix.coxph for more efficient interaction when a parent routine has already recovered the model frame. In general, we are trying to make use of model.matrix.coxph in many of the routines, so that the logic contained there (remove cluster() calls, pull out strata, how to handle intercepts) need not be replicated in multiple places.} \item{12 Oct 2010: Fix a bug in the modified lower limits for survfit (Dory & Korn). A logical vector was being inadvertently converted to numeric. Pointed out by Andy Mugglin. A new case was added to the test suite. } } } \section{Changes in version 2.35}{ \itemize{ \item{15 July 2010: Add a coxph method for the logLik function. This is used by the AIC function and was requested by a user.} \item{29 July 2010: Fix 2 bugs in pyears. The check for a US rate table was off (minor effect on calculations), and there was a call to julian which assumed that the origin argument could be a vector. } \item{21 July 2010: Fix a problem pointed out by a user: calling survfit with almost tied times, e.g., c(2, sqrt(2)^2), could lead to an inconsistent result. Some parts of the code saw these as 2 unique values per the unique() function, some as a single value using the results of table(). We now pre-round the input times to one less decimal digit than the max from .Machine$double.digits. Also added the noweb.R processing function from the coxme package, so that the noweb code can be extracted "on the fly" during installation using commands in the configure and cleanup scripts. } \item{11 July 2010: A rewrite of the majority of the survfit.coxph code. The primary benefits are 1: finally tracked down and eliminated the bug for standard errors of case weights + Cox survival + Efron method; 2: the individual=TRUE and FALSE options now use the same underlying code for curves, before there were some options valid only for one or the other; 3: code was rewritten using noweb with a considerable increase in documentation; 4: during the verification process some errors were found in the test suite and corrected, e.g., a typo in my book led to failure of an all.equal test in book4.R. Similar to the rewrite for survfit several years ago, the new code has far less use of .C to help transparency.} \item{21 May 2010: Fix bug in summary.survfit. For a survival curve from a Cox model with start,stop data, the 'times' argument would generate an error.} \item{24 May 2010: Fix an annoyance in summary.survfit. When the survival data had an event or censor at time 0 and summary is called with a times argument, then my constructed call to approx() would have duplicate x values. The answer was always right, but approx has begun to print a bothersome warning message. A small change to the constructed argument vector avoids it.} \item{7 April 2010: Minor bug pointed out by Fredrik Lundgren. In survfit if the method was KM (default) and error = Tsiatis an error message results. Simple fix: code went down the wrong branch.} \item{24 Feb 2010: Serious bug pointed out by Kevin Buhr. In Surv(time1, time2,stat) if there were i) missing values in time1 and/or time2, ii) illegal value sets with time1 >=time2, and iii) all the instances of ii do not preceed all the instances of i, then the wrong observation (not the illegal) will be thrown out. Repaired, and a new test added. Minor updates to 3 test files: survreg2, testci, ratetable.} \item{8 Feb 2010: Bug pointed out by Heinz Tuechler -- if a subscript was dropped from a rate table the 'type' attribute got dropped, e.g. survexp.usr[,1,,].} \item{26 Jan 2010: At the request of Alex Bokov, added the xmax, xscale, and fun arguments to points.survfit.} \item{26 Jan 2010: Fix bug pointed out by Thomas Lumley -- with case weights <1 a Cox model with (start, stop) input would inappropriately decide it needed to do step halving to find a solution, eventually failing to converge. It was treating a loglik >0 as an indication of failure, but such values arise for small case weights. Let L(w) be the loglik for a data set where everyone is given a weight of w, then L(w)= wL(1) - d log(w) where d=number of deaths in the data. For small enough w positivity of L(w) is certain.} \item{25 Jan 2010: Fix bug in summary.ratetable pointed out by Heinze Tuechler. Added a call to the function to the test suite as well.} \item{15 Dec 2009: Two users pointed out a bug that crept into survreg() with a cluster statement, when a t(x)%*%x was replaced with crossprod. A trivial fix, but in response I added another test that more formally checks the dfbeta residuals and found a major oversight for the case of multiple strata. } \item{14 Dec 2009: 1.Fix bug in frailty.xxx, if there is a missing value in the levels it gets counted by "length(unique(x))" (frailty is called before NA removal.) 2.SurvfitCI had an incorrect CI with case weights, and 3. in survreg a call to resid instead of residuals.survreg, before the class was attached.} \item{11 Nov 2009: The 'type' argument does not make sense for plot.survfit. (If type='p', should one plot the tops of the step function, the bottoms, or both?). Make it explicitly disallowed in response to an R-help query, rather than the confusing error message that currently arose.} \item{28 Oct 2009: The basehaz function would reorder the labels of the strata factor. Not a bug really, but a "why do this?" Unintended consequence of a character -> factor conversion.} \item{1 Oct 2009: Fix a bug pointed out by Ben Domingue. There was one if-then-else path into step-halving in the frailty.controldf routine that would refer to a non-existent variable. A very rarely followed path, obviously, and with the obvious fix. The mathematics of the update was fine.} \item{30 Sep 2009: For coxph and model.matrix.coxph, re-attach the attributues lost from the X matrix when the intercept is removed, i.e., X <- X[,1]. In particular, some downstream libraries depend on the assign attribute. For predict.coxph remove an earlier edit so that a single variable model + type='terms' returns a matrix, not a vector. This is expected by the termplot() function. It led to a whole lot of changes in the test suite results, though, due to more "matrix" printouts.} \item{4 Sep 2009: Added a model.matrix.coxph and model.frame.coxph methods. The model.matrix.default function ceased to work for coxph models sometime between R 2.9 and 2.9.2 (best guess). This wasn't picked up in the test suite but rather by failure of 3 packages that depend on survival. Also added a test. Update CRAN since this broke other's packages.} \item{20 Aug 2009: One more fix to predict.coxph. It needed to use delete.response(Terms) rather than Terms, so as to not look for (unnecessarily) the response variable when the newdata argment is used. Pointed out by Michael Conklin.} \item{17 Aug 2009: Small bug in survfit.coxph.null pointed out by Frank Harrell. The 'n' component would be missing if the input data included strata, i.e., the initial model had used x=TRUE. He also pointed out the fix.} \item{10 June 2009: Fix an error pointed out by Nick Reich, who was the first to use interval censored data + user defined distribution in survreg, jointly. There was no test case and creating one uncovered several errors (but only for this combination). All the error cases led to catastrophic failure, highlighting the extreme rarity of a user requesting this combination.} \item{2 June 2009: Surv(time1, time2, status, type='interval') would fail for an NA status code. Pointed out by Achim Zeilus.} \item{22 May 2009: Allow single subscripts to rate tables, e.g. survexp[1:10: . Returns a simple vector of values. The str() function does this to print out a short summary. Problem pointed out by Heinz Tuechler.} \item{21 May 2009: Create a test case for factor variables/newdata/predict for coxph and survreg. This led to a set of minor fixes; the code is now in line with the R standard for model functions. One consequence is that model.frame.coxph and model.frame.survreg are no longer needed, so have been removed.} \item{20 May 2009: The manual page for survfit was confusing, since it tries to document both the standard KM (formula method) and the coxph method. I've split them out so that now survfit documents only the basic method and points a user the appropriate specialized page.} \item{1 May 2009: The anova.coxph function was incorrect for models with a strata term. Fixed this, and made chisquare tests the default.} \item{22 April 2009: The coxph code had an override to iter and eps, making both of them more strict for a penalized model. However, the overall default values have changed over time, so that these lines actually decreased accuracy - the opposite of their intent. Removed the lines. Also removed the iter.miss and eps.miss components (on which this check depended) from coxph.control, which makes that function match its documentation.} } } \section{Changes in version 2.34 and earlier}{ \subsection{Merge of the TMT source code tree with the Lumley code tree}{ \itemize{ \item Issues/decisions in remerging the Mayo and R code: For most of routines, it was easier to start with the Lumley code and add the Therneau fixes. This is because Tom had expanded a lot of partial matches, e.g., fit$coef in the TT code vs fit$coefficients. Routines with substantial changes were, of course, a special case. The most common change is an is.R() construct to choose class vs oldClass. \item xtras.R: Move anova.coxph and anova.coxphlist to their own source files. The remainder of the code is R only. \item survsum: removed from package \item survreg.old: has been removed from the package \item survfit.s: Depreciate the "formula with no ~1" option Mayo code for [ allows for reordering curves Separate out the R "basehaz" function as a separate source file \item survfit.km.s: The major change of did not get copied into R, so lots of changes. R had "new.time" and Splus 'start.time' for the same argument. Allow them both as synonyms. The output structure also changed: adapt the new one. This is mostly some name changes in the components, removing unneeded redundancies created by a different programmer. \item survfit.coxph.s: TMT code finally fixed the "Can't (yet) to case weights" problem. There must have been 10 years been the intent and execution. \item survexp.s: Add "bareterms" function from R, which replaces a prior use of terms.inner (in Splus but not R). \item survdiff.s: R code had the old (incorrect) expected <- sum(1-offset), since corrected to sum(-log(offset)) . \item{summary.coxph.s: This was a mess, since Tom and I had independently made the addition of a print.summary.coxph function. Below, TMT means that it was the choice in the Splus code, TL means that it was the choice in R 1. Put the coef=T argument in the print function, not summary (TMT) 2. Change the output's name from coef to coefficients (suggestion of Peter Dalgaard). Also change one column name to Pr(>|z|) for R. 3. Remove last vestiges of a reference to the 'icc' component (TMT) 4. Do not include score, rscore, naive.var in the result (TL) 5. Do include loglik in the result (TMT) 6. Compute the test statistics (loglik, Wald, etc) in the summary function rather than in the print.summary function (TL) 7. Remove the digits option from summary, it belongs in print.summary. (neither)} \item{strata.s: R code added a sep argument, this is ok R changed the character string NA to as.character(NA). Not okay 1. won't work with Splus, 2. This is a label, designed for printing, and so it should be a character string. } \item{residuals.coxph.s: R had added type='partial'. (Which I'm not very partial to, from their statistical properties. But they are legal, and I assume that someone requested them).} \item{print.survfit.s: Rewritten as a part of the general survival rewrite. Created the function 'survmean' which does most of the work, and is shared by print and summary, so that the values from 'print' are now available. Fix the minmin function: min(NULL) gives NA in Splus, which is the right answer for a non-estimable median, but Inf in R. Explicitly deal with this case, and add a bunch of comments. R had the print.rmean option, this has been expanded to a more general rmean option that allows setting the cutoff point. R added a print.n option with 3 choices, my code includes all 3 in the output. } \item{lines.survfit.s: The S version has a new block of code for guessing "firstx" more intellegently when it is missing. (Or, one hopes is is more intellegent!)} \item{coxph.control.s: The R code had tighter tolerances (eps= 1e-9) than Splus (1e-4) and a higher iterationn count (20 vs 10). Set eps to 1e-8 and iter to 15, mostly bending to the world. The tighter iteration is defensible, but I still maintain that a Cox model that takes >10 iterations is not going to finish if you give it 100. The likelihood surface is almost perfectly quadratic near the minimum. (Not true for survreg by the way).} \item{: In Surv, the Mayo code creates NA's out of invalid status values or start,stop pairs, rather than a stop and error message. This is to allow for example coxph(Surv(time1,time2, status).... , subset=(goodlines)) succeed, when "goodlines" is the subset with correct values.} } } \subsection{Older changes}{ \itemize{ \item{25SepO7: How embarrassing -- someone pointed out that I had Dave Harrington's name spelled wrong in the options to survfit.coxph!} \item{9Jul07: In a model with offsets, survreg mistakenly omitted the offset from the returned linear.predictor component.} \item{10May07: Change summary.coxph so that it returns an object of class summary.coxph, and add a print method for that object.} \item{22Jun06: Update match.ratetable, so that more liberal matches are now allowed. For instance, 'F', 'f', 'female', 'fem', 'FEMA', etc are now all considered matches to the dimname "female" in survexp.us.} \item{26Apr06: Fix bug in summary.survfit, pointed out by Bob Treder. With the times option, the value of n.risk would be wrong for "in between" times; e.g., the data had events and/or censoring at times 10, 20,... and we asked for printout at time 15. It should give n.risk at time 20, it was returning the value at time 10. Interestingly, the code had a very careful treatment of this case, along with an example in the comments, and the "the right answer is" part of the comment was wrong! So the code correctly computed an incorrect answer. Added another test case to the test suite, survtest2.} \item{21Apr06: Fix problem in [.survfit, pointed out by Thomas Lumley. If fit <- survfit(Surv(time, status) ~ ph.ecog, lung), then fit[2:1] did not reorder the output correctly. I had never tested putting the subscripts in non-increasing order.} \item{7Feb06: Fix a problem in the coxph iteration (coxfit2.c, coxfit5, agfit3, agfit5, agexact). It will likely never catch anyone again, even if I didn't fix it. In a particular data set, beta overshot and step halving was invoked. During step halving, a loglik happened to occur that was within eps of the prior step's loglik --- and the routine decided, erroneously, that it had converged! (A nice quadratic curve, a first guess b1 to the left of the desired max of the curve. The next guess b2 overshot and ends up with a lower loglik, on the right side of the max. Back up to the midpoint of b1 and b2, and this guess, still to the right of the max (still too large) has EXACTLY the same value of y as b1 did, but on the other side of the max from b1. "Last two guesses give the same answer, I'm done" said the routine).} \item{27Sep05: Found and fixed a nasty bug in survfit. When method='fh2' and there were multiple groups I had a subscripting bug, leading to vectors that were supposed to be the same length, but weren't, passed into C. The resulting curves were obviously wrong -- survival precipitously drops to zero.} \item{5May05: Add the drop=F arg to one subscripting selection in survfit.coxph. temp <- (matrix(surv$y, ncol=3))[ntime,,drop=F] If you selected only 1 time point (1 row) in the final output, the code would fail. Pointed out by Cindy Crowson.} \item{18Apr05: Bug in survfit.turnbull. The strata variable was not being filled in (number of points per curve). So if multiple curves were generated at once, i.e., with something on the right hand side of ~ in the formula, all the downstream print/plot functions would not work with the result.} \item{8Feb05: Fix small typo in is.ratetable, introduced on 24Nov04: (Today was the first time I added to the standard library, and thus ended up using the non-verbose mode.)} \item{8Feb05: Add the data.frame argument to pyears. This causes the output to contain a dataframe rather than a set of arrays. It is useful for further processing of the data using Poisson regression.} \item{7Feb05: Modified print.ratetable to be more useful. It now tells about the ratetable, rather than printing all of its values.} \item{8Dec04: Fix a small bug in survfit.turnbull. If there are people left censored before the first time point of any other kind (interval, exact, or right censored), the the plotted height of the curve from "rightmost left censoring time" to "leftmost event time", that is the flat tail on the left, was at the wrong height. Added another test to testreg/reliability.s for this.} \item{24Nov04: Change is.ratetable to give longer messages} } } } survival/inst/COPYRIGHTS0000644000175100001440000000024112257335007014537 0ustar hornikusersCopyright 2000 Mayo Foundation for Medical Education and Research. This software is accepted by users "as is" and without warranties or guarantees of any kind. survival/inst/doc/0000755000175100001440000000000012545056257013700 5ustar hornikuserssurvival/inst/doc/splines.pdf0000644000175100001440000035343112545056257016061 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 4 0 obj << /Length 860 /Filter /FlateDecode >> stream xÚ…UKÓ0¾ï¯è1›ll'i"Ò‚vyHHˆ '–ƒi“¶ÐlJãîãÂog^NÒjgì±g¾ovÞTg×&›©y¬U–Ϊf¦“yœÂ4WYlJ=«–³oÁM¨‚Œí>w0ê02‰ \ ó}ÛóŠ7aš–ÅÛ.Ôóà‘·[ØîÂHçÁ’<€»ð{õñâZ•3¥â2Ë4ÈæqQšY¤ Šñ«°‹=‚Áx: ªu(Jbdás@‡³2VELlæ;øx ¸ÁV³Ô J`¨Œ©`\æ:G"‘ÊU\h bò¸È=DÅH!hmôtjYì솈ÂnXé0xLáf1 rmŸD}×Ö=¯;–?‘xï¼3'™»c4íðÈj=Ò‘Í6ÔÆ‹óœwû_ìç÷ÃÄÕò|Ê«µéá³ jÞ£+J»o†e=Ri\¦)H¬eÎÙZSÍÁê!ŒÒ¤ \Çrɽª„¼8ÔàK\c6œ ·ì 4(V9ÝÝ-¸Ñ¡eñ„XÈÚ¢—8Œò¹.©vn4éGrc$tò”ukq!L˜HÍ‹æ¥n°õ¶¬qcd64i-—‚åo)½ƒÉK$mÒ® ºWhn=5ß¼ÆÐåAnnPjN^ã1k´\Ô¼Õ@viÒÖ”jx(Ôͤ9Ý= at\j©ž Ü’Ï‚O9b™‚Ë`øm‡÷ÛGKjy|xÇ÷ @4\¬¤`÷Ì ‹Ìß&œb”ENô´ÛŽÇÐý£=ÕRBv  F²ù æï`|½á5v u[ªÛ$Ãgào€)ÒÀ¤ï&­tºÛD™‰÷}e÷Ä~Íel?¢QYìaò”0áx¦œ¯`»›æhªIíÊIVUIu‹D’—×a”iÃ=ã7½|o½Nïe Õ÷r¥t~†§ `%1s¬z‰OOéeÊÉ_d=âÞœàŠºµŸ­°=g‡þÿaåÈÚ3¤Ý?,z9ôÈË,v¬OëøCÀ-U{P©c\‰ò•P[ Ýþÿ™òø^L‡òŠ[ápœÐÀ©5Ðã¶ÕŒ/Øï0®Æ#¾[j#õ¬GõïãS5‰öø†«"‹SSÂc’è8ÍSÿšÀÁ³«êì/"hô1 endstream endobj 3 0 obj << /Type /Page /Contents 4 0 R /Resources 2 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 2 0 obj << /Font << /F35 5 0 R /F19 6 0 R /F8 7 0 R /F44 8 0 R /F49 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 13 0 obj << /Length 1327 /Filter /FlateDecode >> stream xÚ…WYoÛF~ׯ Œ¡€ˆæ¼Š¦@ŠÆ@ ¿´QŸâ<Ðe1IG¤âÄ¿¾s-weÊ-ŒÉë›ÙÙ™±ŠRøSQU$E^F¦J“ÔÚhÓ.¾.’*ׯCðJ$%b²qùG«¢ßûÅ_ðçH+Q¹ tþ¶^\^•Q…êòh½‹”ÒI¡tdŠ2IË*Zo£Oñúi©‹¸_®¬2ñץРü #üÜÃê»IýŽŸHØÃjðÓÆðú€|#“ëÓtYÆ7ÊØ}ꟘH¦jQˆÚEhƒ&;$Ž$­âþl/XOÝmEž O¸«Y«ÀÙÀ~Ý1w–؉‡Ìù" ì÷ÝòóúÏh¥lRAìVJ%U–q|Èh7¶$«õ3¹¸åtÆØ*î17û¡"ŠöL¢†Ÿ„ãN”ÖϨö B¹€ Mèª4^;óCÛGÂíNÇ«ï™é@è$¤df|ÇÁÃy.8S¿»ZBÞÄï¯?ÂÇŠSz!ò}œàM¹‚Ï‘ŽVû8é{Ë$:J”Ùàéíy“N™ŽJÆmŒ?`ÚÖøU3AB~æÏ ÆQxÙ%ùQÛ²ð- 7NSà4¥á–Eok€XÆ_<’>%QƒOT>à FéòÊVá݃Wx[Á3³¸_—«L³e—÷-§nÿ¯+~nÅÄ!XEò&ÍR!í$¬j”Dú¬5¬¿aýC‡9±¼Tˆ¬pîñ{X×°> ;RçR@<©E|¬ä6HYª,ïÜñLb Ã~¢Ta„¨á<àïñ¬O@P°>3<¯ ,#ËÂ*`Uô-òÚ$•RÞh'FÚ ÿÿº€ÈÿÖ…ßFUwBº8çè«I¡N(ï¬þ,¾¼%¨SeIž«SÅß¡š*‹©V ×=Ǹ¤|Ì,¿®àUKí6†Ò“ØRaÉ„YsYšXÏP{kÙÜšñL¬eD›Y3/¢O¬ålÍxkùÜšñBTUsß\æh±è3éŒoÖ[+æÖ\ô*q‚Ã6³æ,•ÂZТ:¶øt°RŸ–™õTÚ¹(6\‘ƒÆ”ÜÓnsRÍ-TËFz[AYÍXú‰MÕܺXlê5Ðw\O¥âÌðäêÔÞaƒ2ûcê£ ¡7È2•¾á4k¡¢ƒàI±ê•l:ªªÿÙ훃«þ*…Š>ò õ4Å‘§“{)I.X` ;¶ÉãkÜù‚î4,„桉ºy„Û¡Ðd³kh àͶ§xcÈšp“*ë(\ã°û(Ñp2A!´†ËÈÄ1ÌàM4õè´i¯¤õ¹cS/܇Ïd<æRì¤Ó¦¥ ƒ×“YC•†fÄ oäÙMóàtt˨¨ŸLÍÄá½5·pš1¿Q¼ ¬A=“ñ¦ž@Ã׆‡¨Ö!*ùdá9î½bJÁÍ £÷b-ÍÔ§ "Ö"~¦Ýžò!£öyïîVs.bì7”f“¥ñò²;_qhHþŽÛ,QÖ€)ë³"P%ªTŽþm™å}‘ðÓ¿ &…ÊÅpèNeZmƒÜx/íÑ…‹pñÕZZžø˜êР>=øîšŸXá£ÁÑ¢ÞŒbIØ)ÅÎ%§–äTœèiÏ›î4t€Ý8PÛÁ°é4KLžÁPT%eYºV ¤Å‡õâ_ïV]à endstream endobj 12 0 obj << /Type /Page /Contents 13 0 R /Resources 11 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 1 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./splines-mplot.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 15 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 16 0 R>> /ExtGState << >>/ColorSpace << /sRGB 17 0 R >>>> /Length 1799 /Filter /FlateDecode >> stream xœÝ›Ko[7…÷ú\Æ‹°äð½MÐТ© t‘d¤I‘ÀN“¸@ÿ~Ïpæêa«Öë^IÎ"Š?SÔ½¼$gæÑÞ¼0Þ|2_g/ÍWš yx­d]5“-Ù|{oþ0Ÿg?Üþþó3óüræ¬sÎ,¿^>ÿuæ­'óïìÕãÌŸ3o^àß§™ç7˜_f)ØäM­ü¦›ºž¥fk0¬¯ œ­O¦eKTÈÆl¼s–cµ¥£ á>Öd½Vè>6o#ïÉ_§[—+@; Åx 6‡ÎÙæ.r“žŒoh ÉæÎr¹ƒ9zÛð$BµµíÅÕƇ_· §hÆ1Uá8;Û<8ñ¯§àb ë!^š{pÁb¾b“µ8>c—`>“çy®™÷£Od«;GF¼HØÝ¼LÂ(Œ‚õ²®ÿ LÑóŸ*ÿ÷°GxLàf[; F|F<ÊÞÖ²Sä…ìsà6g#ÖOF˜NgÁÙä„¥DS0±ÃzÉÓä90~Àz)^ʆƒ9áAbý àhù12>6Ž/Ñ·–+._¿ê)7’±žP²¹¸WäkäC¤©âÂÈïȇa)îà ùóSƒÔoÇçÆ…ŸGLå UäG<†8 £nG¢öµ—)S0ò;æ¿‘Ôï£3!ßb}´$ÂftXxÈO(SÈïÅ… +r¨ÿã6£õE/Ì ù,€±ßóZF>J†Ë„ì×1³+àÄÛ¹qà!LCãýJâ@HD’¿72K`g›êZ®ü zŽk8ºÂ’•¢ÊÚKÙ›4´ã!ý¯†+ ýÆè¯Íåf%T·ÒªzjhpÞ®õÓÐ>à¢]âã¼]qÞ®ó9´8o_¹ÝÅÝÏÛŸ]éÃyw»æáܾû<ûá'¤sõÁT«¿——ŽP ³ßm‹«ó$º sõiöãUÿà‡zz,4—º¦­»’¬ù]óö]‘c[êZ¶îð óòXëÖ]£#J‹®mÞU&&:Žæ7:M”‰5hŸÞÕæ>mB‹ÖÒwÈÐ*´hõ.ðŠš—ÚCâ€05ÛÜ}T[åP¿Ú®Û>j ‡Êjxa GÏÕ×Âöe±lÃjëFìx?«Í 5ëÝ^,¶/’ Œo«-›ÔE?9«›xMÂbóBm»¶«‹j5ºIXmÛÂ1õ<ذøÎŽÕÖÍ^÷ïÁ,¶/ªn³mâÁVÍqÕ&~´¬¶,â¯Ø0'gµy‹m´6nãm>«-[úä,6.ÂF¡)XmÙ"Õú°ÚºšæÁfÕøòèXmàêÕ¼ÇbÓÖîžœ€Õæ­Qm×]Ym[¤ùZ`rV›·Zßöaµi›“Böø,¶-Û’á<ظ-Ê×¶£³Ú¸(Ûã4Üm\r]=NÀbã’ ÝŸÅæ%W¤ÝÅö… ª´ «­ë±?ã¬60Û¸ëYl`HçäÖ±ÚÂЂQlb±…C°yÙf¹ë¶aµ¡çÄ¿Çb'/2ö.¶0ÂX¿ŸKÕ½ÒÎÍ "8U9b"¯"r».:Ï"y¬B»G"xkE³J¢~±[{6ñ‹’¶_}о®ê Õ¾>èÕ¶äõý›Xµ+/ÑQXµ,æ6×½X´mйÙȪ=CìÎåéYµ*¶` “°hY¬ý¾×væáÉŠŸE›"d„£°jÍ˜ÙØ:CVíõ+³ÃY´-Jâ–¶àù‘"ÿðp$)Ø΂Eˢį´«VM…Kè Xµ'²–89ë&/G(GçáHR°ä΂E»æ¨G‚eÕ’¨8ºwýèXµ.$§÷kY´(J•#RGãሓWm¹+G–w -Ë·ÑöaÕ¢|$,œ„E›–aþ'æù&'GlGçáÉøÆg9ÂT£±›õŸÝS°qªMþdwîÚÖC#ɱM,Ú•½‰Dc°h]ö"ú¢ûÜEŸt(a«öu$ßõ Ú×EñþmëôKÕ¬Z¶®gѺЋý÷]´¯ÓzñröröàÑ=× endstream endobj 15 0 obj << /CreationDate (D:20150701163330) /ModDate (D:20150701163330) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 16 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 18 0 R >> endobj 17 0 obj [/ICCBased 19 0 R] endobj 18 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 19 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 11 0 obj << /Font << /F8 7 0 R /F49 9 0 R /F11 14 0 R >> /XObject << /Im1 1 0 R >> /ProcSet [ /PDF /Text ] >> endobj 23 0 obj << /Length 1692 /Filter /FlateDecode >> stream xÚXm›Fþ~¿‚V‘ŠÛ3…,µ‘’4©R©­ÚXí‡$Æ>'6ÃåÎùõWX|¾êR¸}›yvf÷Y?[^<~™{‹`‘šÔ[®½(2A/µY`bë-WÞÿ÷ff2ÿf6­õ»ëYä·ôÍ¡·é ZñÐ Š¹õ¢ž™Üï±'ñû†G8aÿ°R8šû+èØÂ×q»©¹ÜAW³‘ÕKœ´sÖÄ =|WÜ.62`Cém,3Ì#+Öøj5ü+«K[øÛ€ä"?ÀŽ4<7—ìü8{·üÕ›GI°H(£`a3õ#‰üˆ‹usÀJ>S› LŒDŒpµâÀ º‡³y’æþkÔŲ€k4ŸÓöܲò}CÛ‚Ó«w9¦ÈÒ\nWôâÇg¶€†Poñ¥@Wݽ*‡†eQsg‰8¿‘È4—«x±Œ*/œÂVVä+ˆ¨<‚gãŠqNv$»%›ûoC4ò_­Yž0ª$ÊdÞͰtÅBuÃeSKG³fIºE$JÔƒëY€Ã $h÷ ä½q—¤:‚Uô7ºªw$F{Þ†Q"v"¡®©AïÊaGT%UÔ“æZl[C¦‚<yC “¯1Y&UÚ8(ÙÌTLr{G^ˆIM}ÉÊ(36{ƒ0ïÈ×xlqá +1~Ѷ¢|Wè]£u)9ÌË`öÁnFèÕã—ÉÂ=ž  µ¹G£œ}2›[û¥ÄK-º$vpèG¨Î¹z#«èNHԴΤ½´Ññ„¾pÿ%k8Bõ„ÇOòYøPÍ;ÇòÓýx2NŸ®tjâClyĉâ?pÑ@ÏõÄõ¯rîGËQC8öÎgx áÙù øÒqè´›ÀQÚG;ûÇI…)`DØ|wf«m`ÌB¾?£¢k¶ޤEåöìîì–^`â¼ãöW¢y;£§I;j¹' æÿË¢$¥Æm( ïdlØüÁïÁøÁ ._×ahO®Ã^|nÅ›³ö¦_n÷ÿìGsÕ:Ìe'ìëÈ7N9Ý‚R¶`'“‡x÷æIQ )/y+¢+ïeÊ·ð=b‡´rRu%ç•FW- vUb Îátn¿öß|ó‹®D¤ùvœ®J´i.]¼X^`†^ä-² KsØÃ(ˆÃÌ+÷Ÿ.ö8!§JC:M:¿ÚïçæâOøÓ¡¹¨œ;:ŸÝÇW $ul Ãü÷,‡q«þ~ƒž„þ?³”2¥Î¥; $aÊuâJÓ½`7$ú9 ¹vs4­Æ0?}s‡ ú1 ø÷ôÌJ,ÝÆuÅ=N"XÀ)i­ÿÕ=±SòŒN§=·ÿ öÞ"²ÂóæZ~SÑèC’ÜÿMh::ÊÚêÜø¯ ¾fÅþC}7>Õ?²vºÛ°d%îü £ëoÎ C6µž vAãÛÞðÖœd2™Çònô×ÌÒƒLs~U·z<˜¿©·¶øØ.õ7©ˆóÖ3¡ à¬Âfƒ$“t•Þü 8kJ endstream endobj 22 0 obj << /Type /Page /Contents 23 0 R /Resources 21 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 20 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./splines-mplot3.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 25 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 26 0 R>> /ExtGState << >>/ColorSpace << /sRGB 27 0 R >>>> /Length 1537 /Filter /FlateDecode >> stream xœ­WËn7 Ý߯ÐÒ^T©÷² ÚZ ±.‚,Œø6µa§Hl´¿ßCRš‘ð›.<×%ñq¨Cro¹k÷ùð›ûìb÷±Ìgcšã”}-îËÑýî>¾¿{ûó+÷úì|Á­Ï³×¿âmÍîŸÃ»÷.¸Ë¹7ø»>|à~9äè3¹VƒœõÈ”Øw¹à9û ë•êçI8­xFAÀúXF¡Š0`ý<×Cx°eD ˆÜqpx‹mHýAPÖÏÙ÷ ™¢õsõM왎 ¹ûj2âóòE왓d•JôEÖçd#*ð7ª\|E> endobj 26 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 28 0 R >> endobj 27 0 obj [/ICCBased 29 0 R] endobj 28 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 29 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 21 0 obj << /Font << /F8 7 0 R /F49 9 0 R /F53 24 0 R /F44 8 0 R >> /XObject << /Im2 20 0 R >> /ProcSet [ /PDF /Text ] >> endobj 33 0 obj << /Length 767 /Filter /FlateDecode >> stream xÚÕWKo1¾÷W¬à’(Ywí}W¤ÐF°œ€Ch6‘MC[üvüø¼‡-J%â0ò®=žùæ›Y{öYut”@«³2)ŒÞsس:i3ö ¢'Xº›»¦´ÿþÇt(DÄŠ‚ûdø¢Q”'.R&Ê 2 ܧ·Ö”xQ«¯aYˆyâ4E7Q™ê„ÎÈ×ùE{ y’1žÛP2ãØqÇ;¯º!’¼ì$8”ß !ÞA½UI0MUÄÁ*T2¨)•^㨥 ×7¶¿×œ€z‰×0[ƒP-‘ÅVI ˽Få¿"·‡³£ê#çMøQ *”oœ²<ö¡é[ÔãåÖ¢ÖZmÔcäNxÊ=×5òoëáA𲿒1ZE&_z*ò›/«°ðÈ¡ó‚\ÏWäHÛï.NŒÙÜõ!»½}‡½Æê 4­=E×ø²÷ȶY×®<¥"á§¹‚fí]‹^°¥RºúÚ¬œÁ¨kÈÖ6qÇ8ÕáÒ-¯]…P&–MµW> endobj 31 0 obj << /Font << /F49 9 0 R /F8 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 36 0 obj << /Length 1542 /Filter /FlateDecode >> stream xÚÅWKoÛF¾ûW½”l-𻤸hšmÒ) EôÐô@S”,˜®d'@ÐßÞy-¹´è¶HÅåì¼÷›™µò"øS^‘…Yš{q…Q’xU{ñî",R'Äà,iK‰˜®^µ±÷¢»øþìÖBT.?®.®^æ^êRoµñ”Òa¦´gyå…·Z{¿û¿yâ×Á"N”2òë5”ü2-к`¡3¿;Âòf?{x¶¸¯}94ðCtRù×Ä_¥*OðkÄ)õÖ½• €7;ä>:ÌèО¿÷Â\¡Þæ[2ø±íé·D‰Ú<æÏƺ]—=K–ûõ¨·ÃŸEðÇêgo¡’°€ .” ‹å’³ä™A Eî×o#ShÌ„Á"½+k8ö¬÷¡4|û˜ ·Œp²9Eæß![ÂÛ†-*º ‰Zú¥ˆVžÔÔÓíÈêìò¦•ìÆd€ppâv”2ø×öL@ÒXTÑP¾¢i¦ÚϽ5á ¥dÄ1)mDaCå£Cu!išKüOà¼GÍ„´¾f¢Íí-eŠi6lXnå½ARO(C$ `G™H˜šÃ¬ö°Êc-Ü[1Põ„[|>†·7]Ï‹!×°nËÆ¢Òµç²lð4>:ØX䡦"@=ŠM«Jg Ÿ+¯Ã8ï$#:Ëð«-o­ß¸ ç@ä“­Ä1MþP5'Žd ºw"Ȉa=ÇJš|8ÞIŒWxµ|°:N³ÕônÃï5•ëyÊÌ€¿;ÁDà v.<Ói²â$õMÙÒ*#g‘RŠ¢Q‘aU'y:V5ìlÛjK[™H•tuפªåá·íHƒ1|Yd;¤ùDΈàêeR8M\ç*Œ5„Eá1õ>X,uìï5n±CïηÑ2"ØÁ’bKî´€‰’èÌZS –p¨U–8Q²$›â’˵§ÚTþRбyŒ mLß‹£7L¤žŽ¯iÃvIXî;Laù¥Z€Ž²´]“k§í 3Ùni.Ýù fw A݃Ån¬[f)OÒüºv€j%ˆAä±`¶'9~ÈìéØc¦kë¹æ‰X ØÑTµJÇ`†*2ƒ¾á¡Ñjh—T™Pñ):¸ã *]‘=e‹bŒšÑ G¨Œ¼ ®µŽ_ Cé¤ÿ¢±¨ššrïèµõ*àR”°òj#ÎsjHÃýŸN6T¹f ­ç¼1c¿™É¡™ ¾¶³5ão;ÀêrmG¨ö땇3ÕV â¡ÚöRW35tÉ•‡JñŸÂ“ ÄZù\;óoõB²§Õ‰Q3oU×ÌόۙÆñúÌWR¡F—¿(Îiù”_ñ¤¼>´g üû<è/“‡Ë‡ÎaÕˆO[Gé­ ã¹œëN ¿Ã»„*Õõ›Ñö3x>ÂãJ}|ô†¦2 EÿÙ„i*-x‰|?­.þ+õŒÇ endstream endobj 35 0 obj << /Type /Page /Contents 36 0 R /Resources 34 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 30 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./splines-fit1.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 37 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 38 0 R>> /ExtGState << >>/ColorSpace << /sRGB 39 0 R >>>> /Length 2724 /Filter /FlateDecode >> stream xœíÝÏoæ¶™à»ÿ w}Ëߤ® v hvôPôP´Ý.‚$‹6‡ûׯd{Æ3‹4iÇ~þü=z?Ú’(‘Ô'Š,é穤/Ó_n~•þ’Úm¾~Ý5òNµX3ýõOé7é››Ÿ~ûëŸ}–>u“#çœÞ~}õù/oJ”šþvóÛߥœþxSÒÏÏŸ/oÊõô‹›Ñb”4zÌœ¾~G_½xÍ}¤±bÑG׊¶Ó,‘ }€VVÓìQ;Ñ‹Ð.Qgš+úADDDDDDDô¡ÚQsZùœˆˆˆˆˆˆˆèåéQZZ=ò """"""¢'£’kä•Öípˆˆˆˆˆˆˆˆø<®¡9wŽÕñcqô´{DDDDDDDħÊZcï´WŒˆˆˆˆˆˆŸØ5åz@ñ}Øf¬‘޳ """"""âËÞbéØQ+""""""">*GŽÙRÉåê~ÍÌÌÌÌÌü)yاGÌÌÌÌÌÌÌÌÏ׳ÇÈéšäænRaffffffæ§àU¢÷TJ»ïÉÌÌÌÌÌÌüÞÑÖé¥033333?'ï­¤RKäÁÌÌÌÌÌÌOÇG:NÈ™™™™™ùYùˆ²Ó5}ÞÌÌÌÌÌ/Ì5Ï(5•Ö¢ fffffæ§äÒ"ÏÓ+êdffffæOÜGG*½Ü¿ÌÌÌÌÌü¨®3ŽvzĨÌÌÌü2ÝZìuúˆ¹˜™™™ùé»çØ%•q.®ÌÌÌüB½bõÓ+ŽÍÌÌÌü1OÄžÊQbÌOÝ%GY§g”Îüѽ¢”TsŽÁüè®=ò8ݯaùÙ»åëÁ‰šûö?3¯8jªåÌöúÝ{ìyzÇ8ø¦Û†i­-Êóð¾nTÖºîžvžã§¶}þ3^5f>½®iú>†«â¯ýõõпê£E?Ïïó0(×öõ³aÑÎúfŒØãÖ+êy|Ì|u£úêæÕͯÒ_n~úí¯öYúüÕMŽœszûõÕç¿<—®‘þvóÛߥœþxSÒÏÏŸ/oÊõô‹›Ñb”Ôn¿ü:ÝÿÃ;~•^ý¿ð½zŒ»h©5ö~~͇øQ£Ž‡ø=ßÄïûË¾Ž¿æ›ø}yü:þšoâï¬îÃÚ¿‰öÅýÎù÷߱s¾ýÃ77?ýÏšJúâ¿ÒŽûåw/·ìÇ5;̹Qy§/¾Nÿ6ò¿§/¾¼ù/nÿð÷¥,g‰ÔÛ[Içû'ÝçÉ‘ßJºÞ;iëºzHºß;i«óöH|“ôxÿ¤çA;÷[IK~“ö.gÖùávåß­®›r×ÕÚ•¿ï†oóíNÑ’Ç5úëð=ߊ÷|•)oâw|+¾fäã!~LJx-%z¿ç[ñ>¯“óMüŽoÅï6çMüaëÞÞq»ë'ûð*{:Ë´¾nwÜOÊwíóïIZr» Š+íwf×÷%mû:—oóëC“®vÕºWÒúI¯–U½û¯íC“öÛý{%íšô,fÏfÓí)üî‘y_rÜôz›ë÷åÞkÖ=îËùûœ~‡¯?|Wÿ‹EÍUå_5ãU~^ëùû?ÿé½7rÇ<3¥·×»ç«ÿùsúïßÿïïÿúÇ×â\ÁôöJ·}{²Ö>bÍô×?¥ß¤oÞ·Iweø©ï«PÎê:©ÞÑWDDDDDDDDD/^×ï‘f•‰ˆˆˆˆˆˆˆˆèck]ýQ爼ˆˆˆˆˆˆˆˆˆè´z´šÎec½íõv„¼:ˆˆˆˆˆˆˆˆ>T;jN«ÝŽODDDDDDDôâtŒ(-­y7ý= •\¯!$Ö­ """""""â‡ðˆ\Ò®1ñ‡cqô´g䆈ˆˆˆˆˆˆøDYkìöc""""""â§Æ#vMG‹‚ˆˆˆˆˆˆø~l3ÖHǼ"""""""âÊÞb©ä£03333333 ³îQ+33333ó'åu=K[òŠc333333óóõì1r*¥ÄÎÌÌÌÌÌÌÌOÅ«DïéšÜenffffffæ¿ëm>®n°ÌÌÌÌÌÌÏÈ{D+©Ô³233333óÓñQ£ŽÓëþú™™™™™ùÙøˆ²Siå¾3333ó rÍ3J==b-fffffæ§äÒ"ÏÓG…™™™™™?qŸoŽTz¼™™™™™ÛuÆÑN廓™™™_¦[‹½R5údfffæ§ïžc—Ó3fgffæê«§2ÏÅ™™™ùcxôk`–2GäÁÌÌÌÏó\¯”yDËÌÌü#zÇ©¬c13óƒ×ˆ~œ^±'33ócy×è-•óWYÌÌüŒ}D›§gŒÊÌÌ>æõÅu9òõØ93?¦[nQûéó2µ2?u—|5¬kΑ3óG÷ŠRN÷˜“ù±]ûÕÑ®æã¾|äçív6Üvª¥G=øzÅQÓ5,kÝOѽ_7²êyXöÌ7m”XÇéãîùêOß;VKµûöÍsó1Ïö[ÏÑÇ?ãUcæÓ3rù(>bôTÏl›ë‡ðÑ®Š¥Îv×´Ÿ ÛvÖ7ó<ßÇ­×U0]·±oÏÿW7£Å(©Ÿ»u¤¯ßÑWDDDDDDDDD/^óö›ÄQcDDDDDDDDDô±µ¢í4fƒˆˆˆˆˆˆˆˆˆ>@«G«iæ82Ñ‹Ð.Qgšý²Œˆˆˆˆˆˆˆèµ£æ4÷Ý,DDDDDDDD/MLjÒÒ*×TŒDDDDDDDôTTr¼Ò×P­ˆˆˆˆˆˆˆˆø<"—´vì…ˆˆˆˆˆˆˆˆ?ˈ£§]£vDDDDDDDÄ'ÊZcï´GÌŒˆˆˆˆˆˆŸØ5í}?ä""""""â?d›±F:jô‚ˆˆˆˆˆˆˆ?.{‹y¤cÄQñQ9rÌ–Ž#FCDDDDÄÀc¦’[ÔÊÌÌÌÌÌÌÏ׳Çȧg“™™™™™™ù©x•è=•’¯Ç ™™™™™™™ÿžw´uú¼¬ÜÌÌÌÌÌÌÏÉ{D+§wtffffff~B>jÔ‘J­Ñ33333ó³òeŸž¯¯˜™™™™_ŽkžQj*-ÇèÌÌÌÌÌÌOÉ¥Ež§{ÌÂÌÌÌÌÌŸ¸8ŽÓ;æfffff~l×GK¥×X‹™™™_¦[‹½NÏØ›™™™™Ÿ¾{Ž]R%raffæê«ŸQ:333óÇðèWÇ“2ލ“™™™Ÿ‡g‰YS™íš6“™™<ïãôºï’™™ï¼Fô#•u›™™™ù±¼kôvzÄÌÌüŒ}D›§Ø33?ø˜Ñr*»GnÌü¨n¹Eí§w”ÅüÔ]òõævšëƒù£{E)§gôÊüØ®gE>RÍ9Æäçï–ãØ§ûÝxüܼ⨧w¬ùÝ{ì™j©±6ß´Qb§çõXÓsð™Ñ-Õšc>KÏsîÑ×?ãU¯¹µîë²äcø¸&¬­ÄQ~íjHÔvD¾Úý¬hÚYßô}Þz] ÝÚGì+þêæW7ÿe[Ðþ endstream endobj 37 0 obj << /CreationDate (D:20150701163330) /ModDate (D:20150701163330) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 38 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 40 0 R >> endobj 39 0 obj [/ICCBased 41 0 R] endobj 40 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 41 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 34 0 obj << /Font << /F8 7 0 R /F49 9 0 R >> /XObject << /Im3 30 0 R >> /ProcSet [ /PDF /Text ] >> endobj 45 0 obj << /Length 1995 /Filter /FlateDecode >> stream xÚÅYKÛ6¾çWI€ÊÈZ©—•6)’" ¤h.éöP¤9hmùÈ’a¯ãu~}çEв¥Íz“¢­HÎp8of(GÏ/£lùY¢“ÁÕl Tê'Y4HâÔ×a<¸š>zj8Šuè† †ÊKá‰åQÃOW¿‚ÁH)?‹…_÷óGL¡Ï-lrŠvH±¼ÃáHYìBCvgð$¨Áó˱cÄ(ò³±Œ4è¤Æ¬ÓÕx‹á( ”7Ï—0©pxKyßÀò,·<_ç˜á*2Ovð§„VIÆÆÈP_—$Xy+& ÷–‡ÒoXÜ|ƒ<9Ê»€•z×;’OÔýP§F»=YÒqD›¡”ͼàY¾ 9S8J¢Ôû{8½zÇ´ž=_ õ¸-—…a F \EípåÛÏÃQ˜FÞž¶d"*…i졦[­…Óe*„§]ÍqõIìï=6‡®úrš-É „ÑÀ««‚ûvh ͘¢JVM)§¼Lì›zc‚HÀÛNeJ÷Í<·ÎÜÕ$6I؉7ì°z·¥*5ŠãÔ{7“ͼï(BöüÀ ˜ÁÂK¹9×ᢥ@p(Є8É:#ô֨кðU²:ꉩquÅËÖ÷‘7!‹©d¼’óë»uRô¼ ®s4Yš3ÊY¢Ä{¡t¶È”ík×Ô"2 Ÿ4\œ8…acà‚qÌß–Ï)(• Šo .ÿÃÛT¾¶”"iÿ1~¯ÌäÐùì¹èÞýRÈ/èÓDzm… ÉWð|€ç/xÞ6,Çç"ë%<¯áùž?…ôQ]-²í;[†ý ñ×£~Ÿ¹{§”m¬Q~â»6•?ðfrZ.)/ÜŠyC"tÄ¡‹Ž8oûAñS€b_ëÌ0\vÈ@uo ]Z>Ôý»%„†þ¾G…{k@‘µxp¿¡fÇÃogþÆîD>«ûe¬ºeÄÝÈe#•ýHõ&—› ¾3Ÿ€å®t¾;É*Ùµ?$›ÔØVž$»M">–Œ[Ëø$³â®Ìš9þÌíîî*tãìY÷:縖<•\<å« nyƯZ>ÏÝÝâÎs„šÜ9( šUó³oÀ±¤ãåcu;Æ‹š¢{씽iÁ‚”ÅÃL»ei/)ì6[âo×Òo‡­ƒCQÚOVÞG§FÖê ñë'Ùþ¤éÆþ_ Mlv"áË=Ð'Ò¾°«võ_võ¿vÕ,EëÇ!vu>bÏ[dz¦¡´ÍÚº÷zÑJÞs Õ üo·}?êOÚ•Ózîrç‹suY7’º¼Ý‚ÍñsŠåé Ac¬’{_?à¢c¦ÿ¼”0\HËOµ¯³£ž_й‡Ó{„’HjçÝöÂiû=µ"Ë|5ßÛŒò{¯[N€ÌÅâÈ.í Ô¼­]ì)ùãäè7ã[ó«<×Î]㵯Ço=섵t½´%ßÞÔ—R ?7ÕuÖ$sál^ßs€Ë×AÔ[èÌFûû£í¬jç¤ÒêüT‚`aúÝ©tÞ'COº§—nÛñÄÙÍç§¡òÔtwÈÍÅ`î|+NñãvkU½÷±|µvÝÌãð}ùn[U¿u[›ãÂö¢ÍrÝ|Š‚0Œ|Ü¿(Wß]”M€Žþª’È#¬/Iì‡ò+ ýÛôÑÛ«GÿÆÈ»Ý endstream endobj 44 0 obj << /Type /Page /Contents 45 0 R /Resources 43 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 43 0 obj << /Font << /F49 9 0 R /F8 7 0 R /F53 24 0 R >> /ProcSet [ /PDF /Text ] >> endobj 48 0 obj << /Length 2038 /Filter /FlateDecode >> stream xÚuXIoÛF¾ûWèV ˆé™ápCÑC‹ÄEŠE=4=Ð"µ ’h“Tlç×çm³P’aÈœõÍ[¿÷fôBÁŸ^ÔeZÕ"«Uª¬]¬7Ï7i]˜ÌÒ‚¨ISZ¶ÉÀÝçƒ]|ìoþ†?7u+$o#š¿=ÜÜÝW‹É‹‡õB+hçÙ"+MZ*½xhÿ&:]ÞÚºN¶KtËÛÌØdÂæÿOðo¿~™,úôp~¿cÇ£»‘¿§7¶ÜÁE=ÓÛtSÛʦ¶Á^Ã+ž–·¦Hz"º4%-yjt†ËýO—N>à@ž¼ì„µ­0I˺W’K'+ät²aâîøˆÖ,4(Șŭ֠°œõ¸4Uòës“ ©ùÈaÛš¿Âl¿»·u¤öB¥&GC¯QaöøÝÃïWøý ¿/ðûÄTbÛåyZz Úfs: dGdk#æJ%N?ÀÏ€B'9nâA²TÎj9öbÈQ¦ƒ E¶£ëpeÈÃh¶’´;0ÁI”=àÒŸa¬€~³Â-[Ò,¨4¡ô\·L.Ï «6³‹{BwÆ5,[¡:½Û4ÎÄ#¯~…oË*sí—´¼d ´‰ÖAÜñØÍýQ'©áü£JU]° FéŸee‰ˆ¾mZn´xÂJøn;9ºÐ>J ÉOñç8Ù¡»&;4û0‚ÛNèßp« 6“|72Pä(wQ&iîì(r>€ÜÔÁw ´¦˜¼q¡ï¿*x•0Õ_ðû^ yî­V—Ü»A°µÉ2Ѳ«K²œôH388¸PŸÄëßxr-»úÕðÔ•Iu–»ØZ‹ÀHÂ\Æ¡±Ð¬Üb‡ xHÓ.ÙÎ;Á<'YU{àÂ6Y`³?wòöJ Ú%ØÍœaãYh7Ç–i|f?4ªLµÊæúep¢……­²I+_–ÎÏn…ýØ Ú Ó¿ñúG4Q<ÙŸö¢x…ØÈyÊ‚Ê{§  (qgîâ8q 0àqlÂc^‡¼³È‹äKÄü‘c‘E¤txÌ{#<³µškêIü¢s–[r8M—v·*-´wr¦“OƒN˜:¨»ŽSñAtŠ?Âù.ÌR:Øð®móí´#O¯1´°JköCˆ;TsTéš@€– DÕ1Mépß醸tYYU‚nTìUuæÜÓ˜IĨ8½%=€ý^xja£yrâM<ã›ÎÎñ@(>«‹çáà£$¦Kä€êv)èƒ×ª Dé¬(Ìõx઻ÂLRZbÕŽaEÏìnzRÙ™_9\-m¶œZ”dž—­w5ài˃»°¨bA5}U¹‚‚¯æU„×¼ bëZn$ñ¿³ÿ„ä¶bþ¯e/a9“ìõ;®Cõ©jFýgàêN˜Ÿ\­§Jv¦ÁççšØäŒÒhÛ2ª‰²aö‘¾á³_9üÜ x„ûâ§æèÀ]„¯¯W¢´‘ªÊWQÎ…¢îõ´/^Ûðmî+Ç%W ¤ÚÌ©=¯nFgð5*ÖF6;`߆Òæ¨b{œ™šU 2^>9½@ûÛÀ‘Ôr ƒ>[®páSäü}#Q4Âa½ 'ÙaúCY&Å×­YºcsˆNì…bB¨.[pÁK= ÁΚ±€à²»#»û™o³—16pàêCÚ|±âb£‘‘« h`býn5`”I+UÆÕzœ!Ò¨Å_¬†[Xæ6K¸™¼LM}oöâÚ¥Ba„ £Ä.ý4"°æÑ¼wÕ®á:¨?þäî?JÒ;èËG' >£¿¸*GÆÂÝJf4×é5ªÈ‘»LžùÝD–"‰iÝÍP'uþn™y*^öÑfyü…Ü'¸ŒX;Ü0]/þÀÞ ¤ñJ~L7ÀNz³lFïÉß„¡CI¿kSØäkVì´»3f³¸Ì L’IÐïFñt/,ñÊ"-ÊÆT3®"I0“gQÂå€U:¶çµ‚ÖRJ`­­L ¡} `ûYŒùK,î%YÓ5‹¸²Jþ’7<‡ާ^¢=>`¿ eòÈDŽý$) ÊrPÎÌ(Sζ®¢Ílé“­ i† Â–ÁÞЩVYÍзòe‡ŽÌ×ubT;O\Öm¡ï³‡­ÎCŽtK¨üe×^. =÷·”ü`CÃ}*‡™´Ïý0KáGX7 ÞTßòvhì®f}@ < œß?ËøN²Â(]èô}$¬ˆ(P,sƒ»—z^<°Ä0Š7'üz}A»•’t—ÖïA Me(N¡y¤ -!ˆÂRÏ‹RâÚYrœ«½œÛ¶sößUù‡†Nœ…¬qW)?{‹=}öþr•ºLî±<¥`…Ùg®iŽ\˜¥;=¡a¨ÎãŽx>ð¼S$¶©žnØÙô,äéFèØ– w«3€O?Ò´Ü+ÐúG¿w”´9öCÙÁR^VÅn°õâ¢ôŠúÎÞ+{afí¼run'±EÛ1DýŸE§cè–ò FL“FB7qýbMîëµ–_NäÕçÄHp¥¸½‹ð ä ‚1|qñv‘Úgz¹z˜=ßb!$$w\g KñOyNÃ׺ÞÝT…§©ø vžèèjG—„þÉ•áÍäËI”Û¨<Í ¸Zf&ÍÊŠ¥/qêæÓÃÍÛ²wK endstream endobj 47 0 obj << /Type /Page /Contents 48 0 R /Resources 46 0 R /MediaBox [0 0 612 792] /Parent 49 0 R >> endobj 42 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./splines-plot2.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 50 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 51 0 R>> /ExtGState << >>/ColorSpace << /sRGB 52 0 R >>>> /Length 2207 /Filter /FlateDecode >> stream xœ­YK•7 Ýß_‘åÌ‚4ïDz­J%$*Ñ© ÄÁôAgZñPû÷{Ûù|…¡eÁåúæó—Ä>Ç>ÉD÷ÈE÷ʽ>=q¯]ž>7ýɇáR©¾7÷æÆýäþ8}õöÇï¿qß^‚!8ûyõí§èkpŸž>sÁ½Ìíà=jfhî«’ÓR¿þåæÞû¾Qr q ¾O/~¼¹}þî·¿.QšÜÅ{séD|ûííïîÏŸÝË›çï~½|æ®ÑÛ¿„”Ô^ç¸ÑÀúX×+(Òm‹JüË•(¥=¢2±€aQÙ©´-*‹•…[½ˆJ‚[2¢2±0PQYE6rã$ÓŒ¨LÜõUTŠQ9‡hFÕ”‘5ÌÖ”…5ŸjJTÆ­¦ $mŒ¦ÄóÅhJH¥¬¦\½ÙhÊ*N4%Ö¼ñÖ”™ƒ¤ššgiœ­)#kXÕ”Hó<Ó”¤œÑ”…#©š’4f°š2³¶WMIi,VSÆUY·¦ÄÏ£XM8ž[SJõ¡)Ç`ͼ5eÈš2Â5žjÊÀZ5%ò1’Ñ”aÐ6¦L¾£)£€lkÊÀÕ” éÚ¿jÊ´¤á¡)³ÀR5%¦ ÍhJ„yilÕ”@U`MÉäÀ4³'®ÁÑeròk.&P͉«ðÙGÈÑyLÄ8&sâJœ%Gê9†œÇäÄWçÞä¨|$PrÈ `Ÿ¸+þÍŽÆŠ{³crvö‰+3š7;ºœØ„IÙ¥'®ÊÅb³¤h†uªöÄÕ©qì Å_퉫òú•„~ËBÿ;³SÙÑX'®Î'e¶5ÎN\MÖ'ì@fÎØô¯ê°Ù1åyai¾dO\ƒ‹“°ƒ%£WO\]Æ…QªÉ>q5Ž·²ƒâ ;–…NeGŽÌVeG–ý(;ßµedÒª.ÊŽ*øPvÐB1ìh“ŸWvô¦ìˆ÷¸Ã`õLðVá kFaÄ“D®l!öã`ÒfÐ |raõÓÜYÌ~Ö^èÞaË…BÔ‚áõ„•u!QDVR7,ÂkÖº”E8Õ ÛcèµÁ°(­×,Br»½·ÈÚ…EypOP•¤÷Ì"dY#,"Vtâ*÷"Ê",kFÃ"ÿ0,8J2,B–£í1] ’²Hs¦,Rƒ•E8ŒaY¤÷(Â"Ô¬4 ‹¨…Û{‹)ñkøAYÔ¤‡‹ðó8cQ‘{eÑY¨,ÊfâNíŒE³³('Ë•™zO1™UWÿ*´vïæ¶nr§äKŸ–Þ!7e‚üÀ=Q/·JÒ;†TdA¾Š#A~•1F>ð“Mï Ÿ-ð›4">^²ÊÕ~¢æg€ß˜ üÀåi_ïÿø4‹/~äMoàé|üƒüĉSàÓ\³íCÚ×~^ÇÖ£}t¹pSàÊ¿~q¦íc‘ø‘ÅÈn•ƒ¿Û‡w?ÉüÚ>¤\šöQßkÁ‰§³ö˜˜»}d¹ ;.ì¦WQÚå¾°“xlqÕ¸Ðmq5å}*®´]bЉuˆ©%Ta›hß pº*nàCŠŒ”öΙ€‹Âg€£<„ãä@ȼŠÞ€«ÒÒ.òy#¼rCÙŸüî]Ú³”rEx ÂSän„Wa„"ˆ¶´¯D„™_Nî¶´7r;Ži‹½’ðʰ¥}ðóŠpœÀ²H8ƒ#|žPfú°¥]K³ ùÈV щ:ÚÒ®WØ‚plƒ[‰–ö¦¥žOäÚjá$÷ÛYi—Ì+ ¡Ñ–öÁñT„#­«•) ‘Ý )³àC^Š”rA8â­ ¢xFƒð&è~¤¡ÇI„w¾> endobj 51 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 53 0 R >> endobj 52 0 obj [/ICCBased 54 0 R] endobj 53 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 54 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 46 0 obj << /Font << /F8 7 0 R /F49 9 0 R >> /XObject << /Im4 42 0 R >> /ProcSet [ /PDF /Text ] >> endobj 55 0 obj << /Length 104 /Filter /FlateDecode >> stream xÚ31Ö3µT0P04W0#S#…C®B. ‚‘)T&9—ËÉ“K?\Á’Kß(Ì¥ïé«PRTšÊ¥ïà¬`È¥ï¢m¨`Ëåé¢`ÇP„ÿþ7Ô3`‡v(P†ËÕ“+ L5* endstream endobj 24 0 obj << /Type /Font /Subtype /Type3 /Name /F53 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 17 27 27 52 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 39 /LastChar 39 /Widths 56 0 R /Encoding 57 0 R /CharProcs 58 0 R >> endobj 56 0 obj [43.59 ] endobj 57 0 obj << /Type /Encoding /Differences [39/a39] >> endobj 58 0 obj << /a39 55 0 R >> endobj 59 0 obj [571.5] endobj 60 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 61 0 obj [562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5] endobj 62 0 obj [583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4] endobj 63 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 64 0 obj [667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7 667.6 484.7] endobj 65 0 obj << /Length1 1645 /Length2 8811 /Length3 0 /Length 9870 /Filter /FlateDecode >> stream xÚµTl6 "µtƒ„KwwKKwH/Ë"KìÒÝ¥€t—”Hƒt§Št,€„”HH7>õ>ïûÿç|ßÙsؽf®™¹gîkniµt9e¬áVE8Ì“—‹G (§.kÄËäááçâáá02êAÝ ÛŒW(&ö/†œ äö`“¹=Õá0 Š»—È+$Æ+,ÆÃäãáý›wʃ< Ö@u.  q0ÊÁ¼] /mÝêüýÈfòŠŠ sü”q„¸@Á Päf q|¨9uá`(ÄÍû¿R°Hغ¹9‰qs{zzr]¹à./¥X9€žP7[ Äââ±þn¨r„üÕ€¨g uýÓ¡ ·qó¹@€(s}q‡YC\€ÕºÊj@M'ìO²ÚŸà_Ãòrñþ“î¯è߉ °?‚A`0ÜÑ ó†Â^m  ¦¢—›—³þM9¸ÂâA ¨ÈêðÇÑA@Em è¡Ã¿ús»@Ü\¹\¡¿{äþæaÌ 0k9¸£#ææ ø}>y¨ ü0woî¿.×÷„ùþl 0k›ßmX»;qëàÎîeù¿8&Àl/!n@Aa!Q ÄñÛrÿ. çíùÃÉûÛüЃ¿¯Ü hóÐÄjyøøº‚< @7wˆ¿ï¿ÿ¼¼@k(Ø hy …þ“ýÁ ±ù?Ü¿ Ô hÂó ?^ ÏïÏ?¿Ìf ‡9xÿ‡þÇs+êéʳÿÕò?NYY¸ЗSÈÉ'È äåå òýÿ;Ï?ø»û?¬Z è_§ãùOFe˜ (úgÓû»¿”Áò×Ú°ÿ»‚üAÏ ËäoÊ#È~øÃûÿ¼„üÿiÿw–ÿ«üÿ÷DŠîøYþ$üü G¨ƒ÷_Œ=»»=ì†:üaC`ÿK5„ü¹Ð²pëÿõ)»6DöÒáŸ1B]¡^k-¨ØöOý} É 0ˆÜúûÁròòðüïaçÀöŠëÃ]ýá‚<¬Ô—T€áÖ¿wOPrqyxÆ'(ôå}XRkˆ×ÚrsÁàn!À‡æü6pÀïárký6ý‰øÜºÿ Qa 7è?HÈ þñ>h”ò/øùò_PÈmû/(ä†þ йþ2;þò> ö/øPþxà><³ÿró¹þ »ü >výr»ýfÊíæùgîÿ*ØÝÅåáÁùCöÿÿñºA ^0`~³« k¿ø Céɹ1*9ɸa˜ÎÊé;ïÒá~…‹žÂZòÍåL&e q]åôÙÂÓ[ßÝ–zôW­IÚm×~7 :m€¹qÒ¾±w»2uŸ©1©8õžmúÝ:ûÛ£´ w©0æ9»‹àj]x~Ròªû\öu8bfC{³ZHë¦ Á£ÿÆ4¸xŠ1ß*gšœÍ“ƒðÐ oêôl’0wìþ©J;Àÿg ‘¯ñ _ìå´ÏR…Ÿk7…195Ê)áð“¯ìvª Ù¬oÉû8•ˆè÷6Ôæ3MRx gÚ®y‰²¡±Õ!ò‹`¹2¼õ‰ˆ8°ö£ìÀŽƒnn'9a^p1â–×¢àªÝ¹uKê]9•í*Ýgï¶§Å÷*ãc¢MØ)‹4Zçö«íÔY¡9Š‚¾Ï¢æúÄûÀg s|Vùn&=¾ /œ½ÎG¿üÇÆ”`oMNô^ìIé#Q ЃymisÅÑ,÷Þö÷UÞÛK!S–ÆLKªÉØ®ç̶` =]ÿγúþ×:ঞ!þûÇñ5ÓUC .¹"„鯿&U^Go3?.ꔌô‘ƒ¾kO©z„VÓyVg¬·®wÉêL—Õ ÁÈGKQwj§Ø¾3­µþøî.3¬j’î¶¥±Ó!ˆ~ùÑùüÖ­©hóbò¦æ§¨ÝÝS€ïœ%£>úl#_iø+¶ÎÖŸy’…úlh‹=š@o{Š;³:•x DÙÔ1B)"–‘üccû,‚Ç”âÖF ;ËåñÜÌã”ö`ÍO÷ K#{±Y•ÝxÈ(âÈa,ƒÌ ›OƒÓzljBqn¸‹À†i¡òé_n)JDC¯túXLñJÂZG‰ šGteÞ08%ÑŽÓQ/‹Ö½ÑJ ù—f/Æ›±ïú@ÂÉ ÐM21Ÿ 96pv²}[ï^xOÌ0Ÿl oÌYÚS•1e¸ šêÔP½_Ðܘ ç~^t8£¹[ÇK’PA{LÊbxº…úfgÑõC²•}¨Ì@‘ ¼Éÿ™ÿÃÅûÏX¶?-P÷ÐÚH¶d`ñß@™ !MgÝ_p¦ºÄó‹M2Fr­ÅÆ¡`’™ »[¹ku¾ï–Â.UjqÇb…äßëÞõ_­TvެÜógÕ-1KV"‰hmD-¤´~Tûc(¾+=«”BÛL'JiLGßÀHj§ñ<ÒÇ`W!ßC½¤°' À–âxPe^Þ¬/U@@¸ÀúX”,U)—“Ø"Œ}lêZßE{¼c&tór4‡É¥&¥ôçÉg)ãòýáe鱨‘5-cdÇíx4zGˆ¼ÖÉDfÉk¡VÊ„äKÑ÷’¶ ÝF,}Þñéè,¢OŠ·È¬g$ –ÊPÛ‰hã±ä²Ð>(ШœÅ–w6á›ÎM™å“·¦ÒÔWeÝÌ;°5O Sн jà”¢r>KiVk½®ÿ2güö‹\@݇ÓËÓŠ(Â6ë Þw#bÜÁWÀ—´Ä_‹þ1ù¸Ú£ì‡è¢ü¾ÀÎ퇙]¥ª†Æ5a$Zñ§½ÙW7Ëá TêSevQÑ07=Îí¸]Ƨ²C©µðY>ØÌ°k:Y¼v ¦šÄé3HØæÉ-{¦w<ýu- %C”ìikÕøü’ªàÒ`‚WäøÓç§EI-h^³ÈVê8ws2¹›W’ó¤iAEüP…ÝÔ0ú”J…âZˆ|°×ž‚(vÆ1X”Mh)è1é8nÛ yúÊà%$—«<ü³†ñ´äоa Ø¢ü\efÓ~Ú¨ö±ŸTòÒWÆ'éÂ.ß}^  ´ `‹0£~Û` =¤%¹Tf:¥;ªUë€6y¯½Ùt/‘qýÏ"¡JBž &d²UÇÚ |cmJsµé-:Ιg´_Ù„ÜÕ4ÛG%­Göo,–T0oåB”¥%¡béaû¸×ïù»«Y¦ªÏ•uÒ ¿UíœÄBÙ?ßj;ú±$â­ Œö=:Šç—çaÉJÓË%AEÅŸ?¿™¢ÊBA\¸™Ñäà¬ÍLE,¹~Ì‘'k[™Ô¹L*êÕÃi × —óýLe ¾ñ–7U¯'Ð}>w¨Ó>ãbs¢Tl~c*(ø*…~}ÞƒÄ$üÅX=+É©êwÝ"sªúXKF¥ú£k¦BaNŸ5Ã]ÚR –¹·!aðœ÷XW`×'òbŒã fmU_ÖyɤpCùvk©6øÈã,§ù”'Ð0NÞ3­Dµ{«8¿Š3z'T:¡.Õ\µ§ZFCß|¬pßçG€çþ÷Á{¬Œdjsª6š|xKUÁk+@©zýøjÎzÈ}‡K`î›#wlòï7zö¤R­Ì+µÏ÷&±—E'B8“Y/»®ÝkÝîìÄõÖ§ßvš}JÁj®¯’H/ôTÐLÄP+PlèĶëÉùýiœú½]÷zI.öåOÆxÅGda÷ ¯•Öú=šBФ-uÅVvÅT N:}Ãxk¢YåHoò†–$^œ›ÀúÓ+O¾,?ŠÝÅ ž\é\ ùÁpÕEOÁ—$œ¶§Æ¢þe³l–Ÿ÷&w«Š°4¤[Ë“R¡A´øtŒA£-lA¢à¶äþ5j›‘+Ê­ºË“¹ Õõ?s¦Õnè¶i²Ù³gÑ"ëËD&Œ >Âææ'ì^÷Ný%DOÙu¢=Ž’|½NßOt}‹52à^DmUÐÌôM©«n‘ø\ëÜ´ÔÔÒju}šáÌE²“uD:žûý“sÑX‰,8‘¬>þÝyá*®<ÁÒurÙÆue`F]‘(ˆŽÜ€–h‰Ú‹€÷åfñn…sº}ŠdjÄÏ7œ@cÞöˆ˜êö¥ç&U™ßA(­ºÏ…ÐFÏ-„7ý­uæý£Þ¡ƒ|F×ÁoY¦(ËÈ‹I3–؃€ØÆˆ¯õôÑâ¢oxÞÆºRîÌ[Œgµb}ùWŸ]A›"ÛóJ­ƒ(›¶‚öüÅÎÝ¢aed Ë–ýXNéýS3qÎã]R4%]öù*{¶ÑÞùÀ3 ÿ¶:Í;IñGzزÓÇz>â–¤OþÊlËõd²rcRÞÒ™šȨíi„7bk)5V [yfa¼ƒÃK+ŽPºWªpEk«xГ֟÷ùË6JæÃö‰,¾[GÄ wËfœ•¦9«')Cààii’šÜ›aÛsWvÂÑ‚À¨«>‹׬·}ºµd9ÛÔ ·o(rͯ‰a—»Æú#E~Š›—øE¹EËÑLêþs_Kúùsåíûm-cLs]Ü­¥¾ý?‡¶ÈPZtîG’àŠ`ûS¬©è‹§0Ë‚+Y’iV¾Ì”¯‰ßè÷p„.'|ï0nÊ·ÖªM=NõRcu9ì½PM^ÖOáï…Û ÊaM‘,Ò%&šÛ;ÈÕxãµ.2RÝ=eg„5N+fkr#«ãgxWOm;wì•.<…*„3¼Ï•ÒËbó -ûlv]¶•Êzòš‹Œì&`œ•j€4eæ¸l0‰”° yTuÚÎú z×8bä|¤¢ß—gbóAÏHxŠ›ëI8øit0†þã×ç¶Â"êc?¨0}¶EÅÖ‡ëVíî]ú¶A1¿¡ñv¡åüYç—Цì]~(‡!)Ä9žLýÊô¨Œ¬¶âè5Éô* ^íÇþ¬)}Å…J¬©›ùê)ÿ˜e‘Œ°\\€»Åfz§½ÒwƒOnfQòËÖ8õ´˜3~‚ gT&Uåxž-eõ)‰8Wrœ¼ê4òÐ"<Äøý¸x¡§•×qû.i(íMX¯!·cJï`WešÂJ;¯Û)±™‡€?Z}úÝ\è|nìÜöíçNÚzÝ@å5½æ÷d!W|Þ!œ¶DèeSÕ\u_›YUy¸ˆw¬ FžÔ¬µúÄÅS¶îh~ÏnhqÄoßåÏufaפæÃ*9y–1{º›eF|銓-˜þUÍÊ!8SÝÜĤ9h$1ü#è;içã§h…Ÿü¹7¿‘¡x^:¬sÓÉFVù£b±$7êz¸m¦}67q,Ûm0V˜U·prši„÷Ï4ÆãŠ:½üh樖Çt[Q¿‡”wPQ EßóÉ~¤”`³vI±í™Öâœðÿ‹ÀÁøötaÚ«ýñ/Òœ¸ÛüßVU[uûœŒ^ˆ9'rKÂ&¶À¹™ܗ¡Ò©Ô¢™vðƒ‹ D`ßÛ¥NŽlxVš§Âížö‘L¿øuè&”q ϤÜ ú…?föæÈäÎ$ õá`‹ I!Í%#¯ŸS¶Dýc¢¤9nÀÉÕâé¶LgŸ^†)]ò‰ã ûkŠ‘XÜæjd‹›r¥ï–`DÑÅæ¹›|ßëcݾ…n3FMøÉËמ~?ÉE#þ炱.ΖÚdY Z(ýµ'°â(¯h›\!Ø\D»ùcÈ(©)½„bqÈo(‹7M,v´­f¾°ZµE¿žÈÞ?“C ­;D˰³MÝ·}—ü-jÜN„,•œÌ±Èó.úgwÚ1æNMk ”hÙÑö&§ÅúéëŒTLw©P­H_ieš¹Uë]Î&=Oµö¬j"yçéŸÌ“ÏS }Å(vÝ–ñ ª…cððx¦¨éMˆ"—j‰à‹¢‡'ß4ÄD¥J˜){­ì$kXµºÑ¶PRbõ₹ü•òG4'o+ÊóH ¢ó•èVOBdñ7Nן§¬®ÐðµÁ™El¤µ\aÆy`v%Ž,ê‚Ï×ìÔz&çæ¢ºÔ…v®~„D’èÀ6L¯ËA¬+ÓµÚxÁ>ÖQg£™‰¸2€@0Wåq^‰æŒsãEÎÛ|tP"€Q8\s.³öÚï¹d{às°i„¿ñ ŽZ༽w y˜Á0ý±F¢è³zûdØÖJ•ßQHJˆút|.bˆmµ^Ìfî†ÀSü>Ùô™}>ó]0샎õ²-68G¶h×I×¶U\9[5ëqí^œšMLj¤ê}‚ ðžm mÈ©ñ'4ÎòäüPh\¦­nKœù–Ö.¼Ê<¬QΘAãó3Ÿ¿B”Üqù|å¤üp¨1a*b¯¥Bjè*¤‘“qôȨú9ï"läÓ4Êù!åK'ì`á²8³Ón¤Q1vó(ô÷‹nTÜÈ_œ²D¨z~}ïò7ý¾•TUY¡Ü·Ñðf$׎[fº(IªÞìNÜ*ê "¦<éaûw$çà~RŸ¾ñ¾Ì¾ËÖb™+›:‹WEs¼¿æÜƒÔ@ûåß<&ÍíÞ|q “1¡5²¸þðR3x¼!xåbÜÂò‡až$?~¦ìÅ,¥IvQþi4F\Z‹dÛ µkbm¨ÓE³‘—~Åœ‰KcGÉýZ”© Djô6¬®±­ŸJž ÇáÃø×–Ôú{ÖGX‡šŒ·æºð¢Þç_¾Îî ¡èî"'šåï\Þf댼äKè÷0÷™O@ÅVrºšüÞ;¦%›q§gJÌ@/vÝ~Œ\w×üÐ4pù$wp˜2ùnšf™´?³b¬ÉZ÷Œ`U§ßâæ–1žz“挰èÄ•Ht™Í?pŒ‹ßãÑýć\(ÀË;ª¥Çæy?JÞë·£C%H£¿ÀŽ´–[7š€æÒIö6Aòè~vaa?5ùíÓÙgQ¨ÙÌìrUËqÛ:ß°ýÔ8¬¼¾b¶žUYÜ¡ûg úˆï§Z™g#S1n{¥– gÙ/KIÚ·á_.YmZYZsŽÞYOåUͲÎWÌ)Wtff$ÄttÃmö¡rîz×Y}·íÀîZsÝÅ^?Wì½¥ÝåZHÚDz¥“ç Š%a3Þ:@…Ѹ‹Œ¹¯ðk“Ga¿PýÇ–ê§žDq÷žç¼~Áñ×._‡ ¤@/H½“ÑáÑb’µ®½ÔØÕ¸¨a›k¦õÇ`&w¾Ä_ï¯t)%Lí¦m‚@€ùê–·vgfdDí†3›Á;D«!4Šo^ÓeGX|WIÖ©· ^ $õþd{œ$·Áu‡p„¾s½êãÜ4`‚ⶤ6gž)ô8çvík6ÔÇÐdAË—¦¥)5¡0,¶ÖÀŽ:&àà£E (ö5ÚóøøEvÿ!Áâ›äz%y¾IÏS³ÊTÂA…[š°B¹VO‚Z~#¦qI¡÷êlëû4Á[§¾Rö€¢¹ìäíùÑ”]n^gG]ís Æ`ÙbÃfáû$iWƒ|ù¶ã:y™Órt‰wW\Àm_fìòÔtõL=—éúZa¬<˜1€ØésYO­ÁË|¶/§´¨±»lêÚ1§w©_O™&°Ì:üý~È SJ)»m~2ò[ÞošÌ,%½Kð»"›÷ •ßÔÎ ÒéèÉ_x‘c£ÜLµ¼×š©þ¥ËÉðW´RÃöÎxê;h5Ms¸ò/]•pûA-nð#šéióøÁµHE´ˆn§¨áŒ7” {Š‹5–@j‚4%‡ðÝ:®r†¬ðæŒ7› -¸ˆñó§©uäýÇ áÙÄûÅ^ž+ Ž+5Éñº"EÞþÚqܹ.jt’áD̃Mb2•ÊY%²4Â_“í‘}žšl׌q(YfpaXót€T¯¯˜©â¹ôåH85Ù¿ÅÝà9nȬøˆhµ{a£(±«´£b2^.^¾)s¥½Ï¹?hl@2U£¤Nuå $ð‘4Œ.îÎþj§YWL_2û è;ðÎæ1/amn›& n>ÿýš$P(bƒ¿4€µ©§²Çž]»%®*>H“²Œ:I‘üøÍMØù¸gkÇžÿ³™zœ1ž0.Ýõ(¯: ¨ü¾Þ¦µ-Q¦iFb5ÛËÖ¤A`w#íÇlRó »ks üÑ Øq ‘ Á§¬0’Öv³Iµl“0V¿ zw<ìnÿ á¡«¬”S¦MMŠ'#zÝ"ç+ÿ#Øï½àkåǸÙc ÛU;ÝÊßOê›Ï©õ"X „ïŸ =ç2¿&{¡ª<— ™J[°[˜¶;kc lùu\ï6ŽCié®ÍÒÁÙZÅ”ˆ-×éQ¸~>5Ñ…§…íÙo}øŠÏþ7$Ïr*ÿâˆAi¥¯ìB,åÙÅ–j®´Ä^ +ë¥/–rð…ú xÒlb„ýÛçåÔ1Ê£Z÷9â*[.{ƒXÉõˆæ•™ÇýbS{"^1D=IÌ—Äõ¼¼Þ þ¢•‹3êó§þ=es:dŠûfhfj ú:sÉk"º ¦à¢ìešO”*‘AZ-fMèÕ)tüÖ9¯e½åQobã^á]K;M)«ÿlÞQç sä³Y^³;ºU)?¬½ÛAèJ|„ˆsÂ8¾—_)I³{|ËS³K²zÝ"ëS—@ò¬µ§L._7÷ôåÿ«Š2’ endstream endobj 66 0 obj << /Type /FontDescriptor /FontName /FTSDXZ+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/P/S/a/c/e/g/h/i/l/m/n/o/one/p/r/s/t/two) /FontFile 65 0 R >> endobj 67 0 obj << /Length1 1416 /Length2 6056 /Length3 0 /Length 7016 /Filter /FlateDecode >> stream xÚwT“ÛÒ6Ò„€RE¨À¡“Ф÷Þ;"5„¡$!‰”P¤w”Þ›t‘ÞA”* J¥ƒ)*Jï|±œ{ï¹ÿ¿Ö÷­¬•ì=óÌÌ~ö<ó®7·ôøì‘v0U$ËI•tt4À $, 88ŒáXWØßv‡) #’ÿPBà X¼M‚Åu æW X“ß•€B Äß@$Z¨ ñ€Ûu€šH àPB¢¼ÑpG',¾ÎßK ”–¸Ë÷+¨àCáP‚u‚¹á+B!®@#$Ãzÿ#—´‹’ôôô€¸ahGYn> 'ë4„a`h˜=ð'e .Ä ö‡š€hìÇüv!°ž4 ˆ7¸Â¡0òaCñÕFÚ@= ñ¬ýÀüs9@°ø_éþDÿLGü †@¡H7á G8à®0 žª¶Ö Ë„ ì!®$>â»Bìð€_G‡U €<Ã?ü0P4…Å`à®?9 þLƒ¿f„½ÒÍ †Àb?ϧ Gàø{÷üÓ\Òáó÷ÎްwøIÃþJÐwÓPþƒÁ›ÿ¶9°@Qø]aqQ Ìó‚: þ,`ì‚ýr‚šñü|PHÐOæw€á>ˆ ˆE?€ùùü§ãŸ; ´‡C±@;˜#øwv¼æð{ï?î´áå‚~~þµ²Â+̉põþ7üW‹uÔõ´uõyÿPþ—SQéôáò ‰‚€`ˆ8ð.~á÷Ï<ÿº¿Ùÿ²êCàN÷5HàÏ?Yà¯ïo&¤Áõgn¸ÿ,¡‹Ä äú·þ-A¢ (þ üž‚_!ÿ?ñÿÌò¿êÿ¿O¤úÀÕõ—Ÿë7àÿñCÜà®ÞxA?Àâ‡C‰ÄCÍ`¿'Zfàöß^ ,?$ G¼ÐùÁ" ‘ßv8Fî³×‡c¡N¿Åôw7ð5\á˜>ÿùàÁG@ÿåÃÏÔÿpÁà{öÛÁàû«½?÷0ü¨ýó*(ÒþçL ‰Š!h4Ä€—~' ôã‡×æõKó@A‹â9ûhÀÏFƒ…@@A¯Ÿ6À?òB ÑøÂ¿/ú÷þ×àÃ`^0(`r • q® i;¬V`öä_–&ÙJ;¼'Ä?\hM†íQµYL0ÊΚÒ*Sì«Z;wè*ºæÌ}ØñY®c«÷ÙçgS]sd³‹›¸Ø¿4žèsÀr{‚ª™à‰YŠâMÉt>k4u'¹œ½c•ï_µ]=9éKUïj†¶é¶wV—fi3°H˜~ùdˆm±ë7^Ýeü+÷“Ûãà¥ËÈTÝŒÛZÒíG€Ï*ê…¾þšnÖ—ÑšA+ßÙð4*Rý,Ê"—«Lgò¯žæ–xlfÐúØ©ÑÉoÚ÷<›çN9¬ÕÝvZÞÞ´ºÅ•Ôy×É6É!üwÄJƒš¡WLCç¬FÙO焳5õÉ&`L»Òáê¯æ*>±ºh†‡jyOw,œ>ÐJÕ°xÛÂŧ“ s^¾üůãE¬‡Caë_Õ)4Ë*¼ YÜ~Áù½ì®Ù3û\¡ãfBâiNmƲÅ3!­÷h„…¯xÀÓ*M® 6|î»\d¾ïƒ9ÏßioWäœr¯ŠFÄÒÆàêf|3 Ÿo_×CÓkb”pžn‹û¾U^‹VÚ}nÈ"¿ \››è×…4b’@¶œ§Q!±Œ­‘·+¦—º#ön–ÆÙ”êæ=;Þ dxÓ‹l#x”ÆoâÎ6 3€ºT=ޤLÿš”éh‡£X_~¡ì$ßÅÙÖT½M7ÐЙ¡pãD }¿O;o>Cë2­Íb(üÁÊ[¤e=y*'?~ºi®GW‚è úƒ×Ç´),ì¶äFsÕ0¥l+• $xú9ksWØ‘ùÆN–ÌU6‹v¶ » ¤ÍµïŠ,¦…óǽAƒ—·Ín³}e"\nÌÎ|µÙ^GÞWÅèÅs‡$rlÙ m·xÆè>¯S—ÝSöÛûN6ßcž|µík} Øù~e°céUÒ°]{K–fÄ×¶Ý)¶Äî!÷KOóà-¨'È9é{í=‚»Þ‡Žº4'&£¦Ð|N’‡~´¨sþÑ ‡ k ädÂÐÃì*‚3]ßYCî,t°ÚÄÂä«êxq¾)ñÉfp&‰!îG}陊aFæ1Âo?vÉ㳎±KÈøŽl!ƒP †ŠØ(’.mБÐΜôƒ·xG¦hÆ.Q}1k¨@!³”O²žEµ{É8|P¯;¶`³2VÙéœaØå/È{B±¼Ì°ÌH£žu_ÆÝ›I@Õîæff«ß2=˜ ´o¼Õ e/’_`\D5)*É·tìAªõ˜gfBŠ÷öÚå{1U´¸+.Àë…cóž=²÷^;±p¬Æq˰5g¨ìÝûƒx™ˆ ‡–FéïÎZ§ÆØ]š]÷ç8lvÝØ­¨Ó%–ø»ƒóV®»@´Þ˜9J3~«öyHwƒ‘ùè}Žê”ñ‹¡6Ò°O_ó*|"µG79#:;÷²%–ŠŽ¾5ÌËÊÒM-×ç}ìG~3ôÌ[”„ßÌ„µl·É#[×XY^22xŸ|m¾ê’Ž]º—ÜzûnÀ¶;6dÁáØ¯“vu$‰»:LœØÈpì^²÷uZé,L½s²žuÛbp+5G'åc“ìÖÐÙñXÁ¼òª§JnÈùÐ0G¹ßÔôuûÏ>N "²Q´Q—­=•‹¿Ëà ví¯vm@àöæ#Ö ò rd(J¸°Km bO‹+ìääfßî%ÑÈí -èúª °+{úúÔ9"SÈÃB×üzðØ»0`òŠŽ”=Ýd˜RP-j t_X’1»‰¹_Ú\mÄ|x5Õ`ª&ã²AëÜ>ŒÄ†ÚÏ–én† YAáÅ}‘ ™-Ö7èôçÞÓUNÖÖE0¡’³JtòõhéQÃûä5üóÚæ¦÷Ë쾈Ü˱=rfUw© ¨½u¯§S;®½ñðI_¼Ijõ~æP'ÜõØX¥1zîÙ 3éí½8Å„*ÏdE!C…Ù†}­Ø€UVé˫̸–ÀYÝJÎäIjïMÀò a‚„R}#—6Þªâ’Ø‹ÛL> j¹~pór_Èc%÷ÔzúÕ©« ÖñpOÛ%»$Vš×x#½Þ5:W~Kõ*¨5z‘FdéÝÀ¸.ךcº¤ÙÈt󈔗­ŸÙšgÁ€›rH«  Mýà Á3káÀ$C¨CË <ÆÍÑöãi%u¢{Ye£I‡%ÿý[1L7s65»êÜ,hƒý¸VÑý‚‡ÆŒvÚ~‚²Ã¢.´_d <óõÎï2ˆžFÜa–LþÀ—obà­/ßÐ7p]W©rÕHgÿ=|ëý•)Ùä«4½o"V(W~ó|p¾cÖ•­¼!¾;=æ»s«I5j~`(ªùüÆú”v¢”ÍC€½AЦ{ckJ—y?×GF7C÷5ºƒM Ƈ j" ¾”gbuŸ®Û€ Æü=Ï7Ÿ²ñ‘ÈË“QìL(Œ÷Ì{»r-i_jy‹}žÇF´eGûíËy'Ó37ùQ†“8cz‰¸i•Â"3=Gþ©”‰—1TÕêë¨UÆ«uÒ›·É,q…g¯¨GÇn†¨8›’ødƒ#wÀCÁÆ5ɦÚg£¢ÞfZ¾¸€Ñ¥Ôï¾SØË…Ίö롦2•¸Iá©Nã Mÿ¶U‹Úö†ó÷¶gyç(:Ø„RÇÔ¬,s_à39?v-=m½º¢V—ýþG|HA8@ÚH npížµ}n—±EÛJ ±É0ÂjÃ}?}gÜF­rÕ{¥È*úåÒ¢|t²G”ºÏK[ùM‡BZTãóÝbÖw†d@íè¯Y½^«BÊ\)ACˆ¼Ú©‰¬iøLlÙÛ‘§1˜+]“m±ÁŸŽÐsdÕD7wT,žò•Ÿ5}Ù»º\ˆ{‘²ò]IBµ±0Ê^¤XF¡M(÷Y²Y‘a81÷3@9gtü@ d…± $ì&ûf´&rÕ0ˆäΗp­W¬=ã§qi{Þ’Ó›âR´¸{ò\ëf¥[ô¬üœM¡Â_«:ï·ŽuÒ’ñ±ÂÈÝÆ•-ˆ*b;cQþƒß+¿GRÞ›IT8wSÀMî}Î"¶oוº½œîÏH:’PÚŽšoÛO½Ö̤¤V|5 Q÷]nõ£îy_) ÖåobÔ óã}¶¶ËíÎ’‚„× î×=¸87)æà9c¶ü¸Ó¨¸º˜8/=°±¢ÃŽ"r®èì]÷ì†òÂKçà•…šeã9M§ãÀ&E=ƒß1kæ­ÙΡ~ùËWXu¥Ë[G7ÓÊ‹+Àçä>pX÷e:Q¹ô“*æö·g¾¯|R¡Õᮇ G\% S•ÈÒx¿½aif™¼ß|Žbrþ †”PÕ³«”%¨ªÇºpÏ_Ÿ+&)ÇήÀ3]ÙA^.·Î”Ò%$Ö?Û ó?÷sñבÿüµüÕ•A5™nëgN7¸]ª2[àäâvcñ‡òëë‡^jñadT›Œza÷£¿9W-©ÒÏ´½»ÂÇÙ ¾a8]}ŒqHª/(ühéu#kxW’…­Î¼TÙ·ÛŠæSªtßéâ ;»G·©QÙ;žFŒ~a9Ýø*èõŧCÒÖ#a±m%÷åíxeU–Ã{¥ák¤5O'ZÌHöQaã;.¤ùÁÃý ylÔ\“V:ËLh¨È©é÷gâ¬+FÞë%׸ξ}ç-ò[jU‰ d±Ÿ B¿Î]Œ¢P_(|'t*Ac(B_DØ’h¾Iö}xâçf–µÎÿ£AgQa‡±ÅØùÅåÄLžuašåém1m‚„©9Kª!Tz *¯ž¥º~‹žs²T³µ»´¹)^˜ÑÄÎ5·Ýom"LWåžQ?¾·µ´£DrûÙ…GGtóžó9 A¬\ZîñÌ ëXÓª¨@)F²ÝªYJ]¨2ôE­œkG}-”žvùœ@¿è`÷Œ¬`zƒ ~ôQð†`_¼tXrâ̙ˉàŠÏ<éúH2s8.äñ›^¢¥/Eâj4‘¨á¶×›½M:?$Ç’g¯Ø¡F›Ì¢·ÊrZAÉQŸ¦äCo¥4ú/+“Ý—¢|ød»þbCÌšk½Q-ã¨ô=?¾>pVîhmóî²T¬UbÚÿ_O=±*R{Jѵ³&¦ôô®šã£–´£×Ì’TGŠ2ì{—ôm2²gò4‰Åï÷;¡©ön“³ÁÊ¥Ö¯Œšõ”Úû¨ƒTÒ‡É çäÁËE±^ðê(Ó2wÊø¡æªgIԨώ…º—<½­æ',³f0BînßÎår>i5ïKùìí·ˆQ‡uæ¾z\¢›¿5N½@Úî²¹Ðm¿àõRÃIE¦/¸$¢È*Á 4Nèô»ÒúM#èäk¦Ð,Så+EáÄéç=äÕ©€÷ ÐgÞ-&†. Dqø7’†Hægo;ÂB|Ûîé百èáZÇÍl¸ùx¼xÔ¥‹åÉú‘¥&PN¯º‘@a#…uŸãÏÌÞaå()i$Q-=ÚJg¢<œýZ{ã‘‚^.WeæµC ×Ì jXyZý¶'I?*ŠBkË¿:ÒniWø–Ì•ƒõ¤T2®Mei©:çêî¦ãùI•#! =k~ãü‰B½c˜Zâ]£¾(h¯Ðøië4­@ØìÒÝ÷J¬•ó£×ªéefÑq#¶)ÛôY®î\BV³éº0æl­µ$Pë¹’Q8åà…ulÚöUúk4¶úþFl2Z³r*G«ZÌ™™œu¼;‘hKv!rv‹1‚^(áv‚ÍOæ&OÊ݃IxmóÐÀ§ ‰&â½}=È8†’åèŠJѾ«K‰<€wgHÊ0>‘“)äý6¬Š!d²Ñïæó5­õZ¦ÆÁwž‰ßXŒÉIF©H¶{dçŒ0¨@i)s»Ê»d4öÖ#K4ûl¿¿NºÇmèµ5—°GÔíE.:"0åTz‡¬UYi#wÔOmðÙ YD¹Ö¼X$éNÌÐÛïä­5[®*%m~‘û.ôÔ¿ËwûYÓ<ù8 Ý®± ºã¬˜­jK¡Ûn£­¯ þdÙµœÃ· Ê€øc>'*ë+éŠhY6În¯Ê©Üùaˆ=H¾½±èrg6w7AÞ(ïØW“‚ûbWÝ’˜™”Þ/iÔÃDNYš;`0e˪0ó¬ð[òóï åŸæÖÕÞõѼL$!hÊ0f_ÓÐÙ_ôŽ&ÈgÔaõ­n긓+c+v£—„${…nXÜeà&<TÚî(9+·å`)—1‚÷¦”Ì•Lbµé‡qâÔwÐ(ΰZöH—½å7òKd ÅiÑ€ˆW)×éï ÔÂ|OÒvíª>ÝL§ Äj‡‹Ô€R"|ߎm´ëÝäžw·B­šŒ%OÑË%,ñª-jí!‰žºò±ØfµGæ"‘úŒ7µjTOb%LÁoˆñ‰%Å¢·5ŸŒ—MÌËg&$ÏwÎ?Ó# wVœZ— • ù¼¢Lò<– žÅâÑ®·rÒdÇe7³b›; vÑ‹_On ÛªŠ.8â éżàÐÌ6FÝîWç\õ–ñ•áÓX )i(ag Rpÿ1vä…EòÎïÏÝÆ-õ·6‡Ò»Œ`Œ^@K°ñ¥ºi—÷¢©ö÷0Ç)êj‚í]ÌPwWܽ G¥ûn.zQÈìògyDŸª½¥÷«6où†ƒ¦ïV½åç†þp3˜ç*×ÎHY2sŒ~zé=,#XCÕÁ V5öS±úÑ&üd¾Äˆƒ…S§Ù;¶À\©0‰Üîh©¹8–—\ÁóšVàà¥">‹¼@El™ÕÕ .M*r~¨3`sÃÊ8"Ýÿ¯4*Ö9o€óB軜ÂŤ‡/)Ò|ù«¡ÚÔÄ!»õÆÄ4¹ßpþý44„YÌÚÇ7ÂîŽU¨ñÞïÌUŒ©™êØi&ãá 4¾ûLÜÿaxÜ´Ñ…Þ×äWâl×Ùj{^4;iž4¿ÌcÉçr޽°[áañÅÆ0óAGA´= צ­êú7iÓVO‚ò>ðhÒÁ6¶ÇªLÁÛ©ü¥]ÓuG4qHÑàþ¢™guöåDÔ*O8»èƒ>’ö–>ÔnÕÏò¿_š2²U¢$ºë3Î(d(‘¶+¢@¡°µMs£ÕXc3=Û0–Ñï"z^ì^ ^Ÿi~<:¼×Ì”W>è(®¥Foó Tw+p¹‹+6 þ>4—ì#5kãÝ&NÕªkë80&öüüãn\Þî$†žÚØÖNä”43ênK.9–6hÛËÆD†j]μ""üÞò›‘®¤ø“†¿‰×‹çÉL俏—Ç í8FN¦1Mik²QbW–ºûü¹ÓÈõÏžöÝó}#x{pÑ×%½” ÆÄ.È8§ù¬]¢y!ñfßÞËÎòŽjÄx£{{å¿a—lÏð1ùhé—~!ѧ¶µ­V_ Ñ/m ô &‚¸m>›Sü1K9üi¼¹-´Hy–© íGUÃþÊ,c¨¯Xn…Ü”íËr³Få EÆ{ÌåÝ¥ôƒ‹ùdÀŽùo{HÕß2§œC¸1_SQc¼CúÉ &¸ 3FÑí?¶#CžÇ”ÊÒÙÜ¥rï¿ªŽ“½ŸüD¬$ÝÞ;h‡+»f¨Í’I©íMù­ïB”RrËOžŽ¿[‡ßW‚<6«AïŸzIÄûf¾Ó1kÓ§·x~äa=.SQŒtõ¶}²~ÂøCÿ|ËTábŽþr½lÒóýa•Nºµ°N¤–6W´b#å:™…I~ˆö_Ò&ᬕ %2+ètš¬2‡…ïØ÷Z {ñ´­¤ùèëçU4ý Ü×k >àU™)6¾%ÃD“/–dª+Má!{í‘{﹌â¥Â8¯ª~é¬A _2ÿÁV3p¥¼C’LƒõíÄË©]ñb²³0ª¦L幺ø¦Æ½|…}*>â—ÿkD¢“ endstream endobj 68 0 obj << /Type /FontDescriptor /FontName /MHOLNP+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/x) /FontFile 67 0 R >> endobj 69 0 obj << /Length1 2509 /Length2 21327 /Length3 0 /Length 22763 /Filter /FlateDecode >> stream xÚŒ÷PÌÖŠwî0¸»»» ®ƒ»»[€@pww îHpww îäN¾#Éùß«º·¨‚YÛVoë(H”UDLí’öv. ,ŒÌ¼1…,Ìff6FffVx 5Kà¿Åð@'gK{;Þ¿ Äœ€F. ™¸‘ ÈNÁÞ ëj`a°pò²pñ23X™™yþchïÄ 7r³4(0díí€ÎðböžN–æ. šÿ|P›ÐXxx¸èÿqˆØ,MŒì F.@[£‰‘ @ÕÞÄèâù?!¨ù-\\x™˜ÜÝÝlíÌièî–.€@g “Ðð;a€¢‘-ð_™1ÂSÔ,,ÿ%Wµ7sq7r@K 3ÈÃÕÎè‘TeäJ@»ËÿË€ðïÚXYþîßÞ¿YÚýãldbboë`dçiig0³´”$å]<\èFv¦¿ lœíAþFnF–6FÆ ƒNnQüwzÎ&N–.ΌΖ6¿SdúTe ;S1{[[ ‹3üïó‰[:M@e÷dúWg­íìÝí¼ÿ Ì,íLÍ~'aêêÀ¤ngéè ”ÿ· HÿGftp033sñ°€Ž ‡‰ÓïðjžÀ”,¿Å  |½ìf $€¾–f@Ðxog#7 ÀÅÉèëý·â< ÀÔÒÄ` 4·´ƒÿ$šý ƒšïdéÐaÍ €ù÷Ï?éÆËÔÞÎÆóù?ýeÒÒPTP“¦ûWÆÿÕ‰ŠÚ{¼Ø™ ¬Ì–ßCÆúàû¿aþ[€ÿ$ÿTÙÈò߇û+¢Œ™=€ç_9€Š÷Ÿ<Üþ=Ôÿ^Àÿ2(Úƒf þ3úºÌÌ& _,ÿŸà—ÿsÿ;ÊÿÛèÿßIºÚØü£¦þGÿÿ£6²µ´ñü·h”]]@k¡`Z»ÿkª ü×*+M-]mÿ¯VÆÅ´"væ6ÿ-£¥³¤¥ÐTÙÒÅÄâ_3ôŸ.€ÂÛXÚ•í-_6PÃþ´p&Ö  ÅÔ«T@Ð>ý/¥„‰½éïÅcåà99yƒZBoІš=þm£½ ÈJÏ`fïÿ»£œ&‘ߢ!N“èÄ`ûƒ¸Lâ€I⿈‹À$ù±˜¤þ V“ôÄ`’ùƒ@|òˆOáñ)þA >¥ÿ"nŸòŠ©ú±˜Ôþ P~êˆOãñiþA ­ÿ"ÎèÅøYšü±ƒø@÷Ÿíëß}d2ý ‚êüS¹ßÈÑ4éÿ‘°Îj‘‘í_> ª™ý ³¿ ûohùÿ6þ9Ø~C·?”¿õö®N…™˜ÿA-þdª™…§ƒè9øc’ýEÈ *¤Õ_T-ë¿ ¨\6AP½þJ te1ý‰Ìrµú_zP²ör¶ÿ5(‡?jP0лfg4sù#eù·ô_·ÏÅ S;€îû¿šzÕ™ÿ4ç7r:ÿ³ÿí7ëo¡½ ÐÔØæhØØÿ(þ—‰çßšÿ³üîÜ_m`UÕùO)@NÎ@[K{›¿Ø9~ÛÝþj(ˆ3èÍøoZ :Û9[ü”éZеËäbáük"@esq·ÿËÃõ/ê Û_t2÷¿Æ äíñ…÷ü ‚Êåõçp H^@§QýÏÕeâêêË? è^ûþç è4_š·7á ±ªéx¨Áwgؘ¡ØÓL¥að^rêt}B†I¢©É ÚpºIúö~uG‚úVx™øÅû¸µ&¼-A¥ý—ϳA܇©½vøÅI¬ï…Ç"_áÔ„÷}^}4­!ZÁºe)r]¹‘•óÑÜû¥<¾”¯Œ†Íï©ì×pÊ!<—O3|VÖ ,™¥È3ΚÃ!…va „¥E»ð@™½½›AË™x#–£ƒ÷=ùÌVä­½Éó8çµV©Æê܃KŽ«Cq‹6:Eé-z˜,‹½à]Z´ú}Ñ£•¿ˆ8‰>e•á=ã!kFå‡(»æ¾:·Ñ¥.–ÝÜ$@þwôÝ„º²f 'R%Ìšvƒ(t—:6k áa—™èZÇF¯™U Á2Ïô×7À,’–Åq@K¯÷¯ǵ¡áï ÷a)mCÍÃ;½¾š%BBæî,´:‘«+¼„)På.cà=š@Œ÷ôËnIÜÏPF¡üÞA—¢Ox,Û…ÌÌÚÜP}`£g¹>—1¾UŠ]›³¨ 6îhzÁoÁœŠoÆÝ)…¸l†™aÔy½ŸòÖc©Tøê³ÁåL›UXŒâežÄòñÛ‹—?È‹Øe¤úöª‡"uƒ<­y Ï4ªÔÛ¿p¨ìžD1Œ6ЇöóÜ™nßUTˆêcÙk<^T¹GˆÅl©q…ä}êúÓ㬔&%^|¿3×gLvª&êFbÛ‹¢©&ð´3ÖSüìN_.#NAU´¤0 wÄ·(^Ô‡[ Œõé¶9€µ¾SéúbæûµŠÌý2ƒÎ¡fËWŽè4›8J’…€ôë êèÈW³È–uuw‘†e‚ºŒÙïÀ“ÁQHQ2˜N6f Áñi@aUæº|ÆÛ|B¶Õ¶W–ðql‹@µ³ný:Æ-h…€nù†¸P„úO¸ÉýgΤœi©<»øé‘¸»–5ó‰Ù$ Sðò¬ï°ët²ÓQovŠüiH¢ÍB?Áš¢FÁ¾Û៣¬M¾ü€®Ì&‘GÂÀ[‡J…ºù:)‚,‹RõÔ_/&;þNÖ¥üeŸ7ôÎ:x²@ɾɞã¹yé9ü„ë(¦²adÉswRIŽÄX÷KÕrŸš·@{þñƒñ„)C¿&þ,"æ2Ø´7þö#Ûó†æÇÅÌõ!‹‹õܪ£ØÆÆÚÚ ¡!\C¡k®äc,ƒÑ\(¤®kne~v–9Âk—‰éwË(H|œi­-¾Nr®,Gl¿ÒÕÅÞAY¦c5H»TÕOÍøÞyâÍ¿ú™Qèn8žè¶µí³+±»‘èdÜË´åמ­yÆü¾o…ǼAXq¥ û¨£Å\9ëæÍ¹“8¯ûÐwÊqJ'eø‡HÌR›Ö’c\•Ãù9DÓK-5,çm!¦-*•Ý ODè¹B)GÖǪÄ<£‘RÍ)£LNnóë該…!{S§çx86õðWnt’•Â<ѸÆ+ü^v7ÞÕÞO¡’µÇkÝ5>-7|–“6½¶iàRÅÔO÷xé†P\"– ÎÈÄý–î…yŽK¥PÙNÌDÿÍ/;hµc÷΄ýò\SE]±–ŒÂñe¥»ù á"MVâCnGyÓéê|ûc¿] ¦èAQẛ§Sæø^­<{¯GòÓÇÉûýUÝSC˜î[ë[¬zöå}€jƒ!‹×Ãï‡éã^—ñ‚ü-žZ\±.¿„DÒÙxãÁÞUîç¿›as²áIo(¹MyáªaÔæŒa|×®`!¢-ßýñ­&VC@Ÿ÷Zù(\`• Ö¡†ßb^ÇXÄ 2Zµ£Rö¦%}7wÅåAãƒÖÞBÕÒø;Íåɽ`Êž|æ@9À+.f:“%œåÒ´+œq6Þ71xîý^"ÒEs9l`ðžŒW:¹­Ûk.,"®UåYˆ"Œ:쪋¬Æî‰t&—²Â3E ŽøhýÂi ‘ßn—ƒn¡».šÖ¶¼þ‡DõØêa&AŽÁjk²TãýÖÅQ5‘¢M3úÔ‹á*·+ãµ+ÝÙÎéÓµ ß½¿×Pü}{ÍF®Ä sªsêü .š&‹¤ÇÄÙìaÊÀüÕ6îs£‡†çmã2êT©™—NJ‰ƒÛU/RaX\ÚÝâ@)£“(Ï'rL»-±*1–§~—rSa²ûRîw“ÖŠá³Qrµ`¹3|=?gl%Ÿ¨¡£¯#‹Áø6·ŠL_+¿ý:‹¯zÿ¼AÛÉÿ½8/ig=½¹§˜‡¿Q*ILÂ0Tã(ßb³s¨ŸS%a\²È^¹³{ 5ØŸ¼üK'é[ǽ2Ïωöggžp0*Ã*"¤§eÍG^ÅÄ¢“Aç™Û‰ÍÄÂÐfÁˆ™ú„šk¡”¬ìîhe6^û䳬šçèõîk¡ßO”xŒß öH•ŠsЇÇ4ûh%ÑôÖ˜Š =Ë£Uêà"†uU’¾áKÒ–¯Zªa]~>ÔWÜ™(†©R¨}Ul“㊭¥û6¢ãT”ñ<¨u;€~ªåôí¨[$@,Šº_Ô W~º—S—È?¹~¾˜‘ î™C‘ަªoÓVG‘YÚt¡(Ø}KÎìÑ)¥áê§ìzε·ËB¬Ã=Ò¢š]1úÖ¤ÖwIõÑÉv#ÉÎÀ. BV> –:Y¾þ wd2¾ªQ‘Ð[N²qcåBß:Ãxý›ÝO?¬SCùš=Ø£Bõ<°Öïç4·ísùÕ¶ý|ö„–Trt'[‘·3ç$Œo߯¯ì´7 hÓÚÊî‡o/ì30Œ…½ã<ç`IÄĈ£¼±M`5ýµ›NΉö™±í6BÍ•SMɦtïjÆg3(åùº!ÑZPlÐ}Zý2¸Ÿ‡ú“›èÌÆxy–ºoÔ‹zU‡ç„ݰ ¹ÍvÀBƒÿ¼î¸]ïûfT@Â;Wh Ÿ¼Žcî¾DûÀµ' §%쩪çwšžVõhc/:T™GœœåFÂüxØ)™„¸¶ñtV&±a¤dö ìÁ š‹éóFó¹oEŠhã½:)›è ÇÒõóþÖ´.éL…ôKòxÚ© ‡âO¹ƒ +—l»èËu£3ÄYd“9õM™¦ =²‹ Nõ6÷X‚”ó@K /ñr=N ʈ'[ÏjËFÜH·¡#«owš¥QòýDõ ?ˆÙt{|5È]ÐÊâ­çP(©ÜØ?[Á8„Ai¯BiÄ÷/ýÅdUº9°”Gâ»n˜xv,N©I0X¾;¯ò±•Zz£äeÔE¦°ªÎpŽqI‹àRxѳÂ)ù½Á͉Ýjà7X¤.¥çE(y$»ä¨!(ƒu,}ƒÐsŒøF¿l¬ï±|Ÿ ‰j'OKåDÃ×/8ýM½«Ä5q-ð‚‰‡6Ý*qaxùþ34=aMYEWLŸù`Tq©õ[oyw] e묄bxT| – f–æoÃ\Qq)û‘h­FÕz‘ýž/›ÿ‹#¾Ýâ®k ‹¼KÂÇ~Ç^ ¼cÜB¿ãtQ ÐnÔ¤‹˜*]A2KŸp\•u ­B6Yay8$Ɇúr¦‘–¨ 8¤²Çdë È!ǦùŒô¡é[Y\7Š”N®ÚÔn©awý¬îÛ=ÐÐwS ¡˜"IÖJ:ùùî‘dLvVk¯t&WŽâФH¢¦å„èW†€èÒ¨Œ~€*4Ë=€ìz¦ ÕyTG%LÞ¢\‹a/ÈOÕéÕÓˆ  ¬%U67ÉGÄù™@²l [ëVT¨0G@üÚÆ5üCfœ BV_pó¸´UxþûgL'Ì Õdø}Æo‚Ë“ó55yÖ)ÕË6ý|”É5düËéµWÀ/4fS»’^†Í?š…òwdLÞ «JS&ó¿AuÄŸAÝìÞq£fIèîhÔ•³ \U‰xzvìð^ÎÐad›x\@%O(¼ÎÚb2óeÛHÎ8)þ±C*U–ŒZu@e¦à<3KüQSýÅ÷ù®±¹E«¿ñ³>BÀì…ÕØé®ó(%é7Ø¡±bƒÇ¦(¾Ãðúñ+ÿ¸mÙ·<äú@½U2xò%˜r”*1 ¡@Ž4þUœÂöÓ‘ÚÞKá„*œ,¸t£Ê†Œ4“Ȱ#{M°ê«Xh§wÍ%d+† “vZÞV+†l>6àn„Õ†mT26¹/þ›u${ Ïy·3Ê߆O¼ó¡%jRÏÆŒõ-Ž•¬! ³šÏ2vž«¸]=WÉ®’×®“\áÍ|Mûè8z=ýH^Öô沞ðzNÕé«Àº\riö-y¥¯r»”BûW¨« €JG„’]¯Ž“ÐX‚•®‡›ý R¼[—ý^ÁÙ¡ÒÚº±ê§n‹=ïˆ&½vôq6GD?{R(ªm·4± àT<Þê°ÅÀb˜a ¹'´ÜY^ýh6übÝòñ8ãöÅ*à‡-#Õ˜êÎ7ñöÑ%Ïgb®¬|D­Ûy±¯U¶GúM5R®R0K,Å¿P\˜¢g^7?n_Ào²ªc©²‹0ø%.èëM·oVnÎcANøT:¦'|®ðh¸¨?Ý*%wÇÇQ]èþã¸rWÆÉñ\¤”ÍL$hC ÕÒµ2àÛÚnßUçÔùÉ4š–Óû#BñÛ;ÞØÏ˦D0Š Ö ".¨ˆÇ å_è1h¶ÛŒÊ ‡Ú,¶'xtfK\ß`UpxäâqCPûÕäÊŒ>’ﺛ%4)ëlƒ{Dá;þ.ð„”¡N~ë°zÏ!_R.ÓcÒèëcAÔCç%¾®>ñí‹{»ÖgºÎ—JmÞñóƒþ‰ÂL–î«Fÿ%`þ0C×ap<ÈÊbÌùՈÆôïúËœøgLÙžS¿ .òЊ‘"ujÆ#ßNwJd|sÇë…:1"ÂGÌóÚ·¼’7 ÁFóµ®=þU¹ç;àï²®Ù“¶gî“ÊlG‹ì\c#¤]æu•4Œ%ØrÀ}JìÀ ?E9Xð â\ÎP`ØPJñÍ09ÿl­{ȹ\¨Ím¤ý»¾îó'ÚxÌù•á­REc³Pº•Ž®õl{R¸ÏK%êp^S׿?ÍÇŒvc¾¸6HæN¥ñ÷¡|ò\k)¯œ¢í4¿{cúȪ÷í {©túX~ÆÂUº~藍<о>3UýôÚñ¾(>~’Kð„æ{áuš´Ý—lð¶Ò*.Ò#oñ°üL¬./ ?n÷Š~YóÈ|¼‰7+(Ëœ‚Ø~›xùE‡¾TüQâ¤$ùý ãÏ-’žÉ>RZÙU.äže-¢X‚C¨s«$BÊé%‘CYDºŠýˆîÖvX´e4Ûdż=rN·zU<5-ÔOd+8Õò)oìã%gBï&¨îNîûý%±M5¤äçéé°œ1+¥^ 7õ™ZÊú ‡îÀ,¢P*µ‰*½BÄË\„Ùîhé9‡ ‹Â‚V4üš†» ‚o ùCo~[ÀÍS qÙ9Ÿ”¶¶(¾S%bN‘ƒØßÕEÌp^ÀS â•}ù "Ig:äîfÈÆãÔr›‘%ïŸtð^îõg ñ øÒàz|~~ˆëö!I»Îr¾© ¼L¦Re©° óyµÄ³]«Žò–[£ž®P´¸ÈÆÁ0 ç)½e:i¬~ÃõЫm…î¤Ù>ÏémÔ °Éz„à°ïz¯s7)É÷ž^‘²” ݲ,R·NQ¥WæJzvìf’¸GH÷TxQòû@ñ Fõcïs%M»Å¦æÿžSA—8-ä`õNð™wHÃàøà\1­ ³/ÖL£.©÷"&ì9›ú²×÷Ë!×]ÉL3ˆÑ¥ÂÐ[åÒa‰/Ü(AÏêE™Oî4Õæè‘#˜Ž³¼›¦VdE­5«˜0Ž?Š2;œí‘Z…ãáå ¤¸bwH«|T&ÃÁ÷ËŸöÚW^Î>¡A¾ýòs ªè‘E¼fås¶%ßDØÎcç/TXûpãæu¾š6/&<Ëfê®~Å"ª"¸Â29YÆ x~Dñqé²b™úž!ràô‰YÝ3UýŠ;*ëö^ŽØ îª!t"IcýË—LøÔç •Õu~õÞ¼AË+B&tw›up§Ší«¢ª—g“öÍ÷g~5?T;êXÛÕ7.Ó€…^¡®§%D»CW>Ò`FF'ì„êc(’8Uõiyñ…Qb;¥ð‰˜Êˆá2EúÂ3XÀ8½tGl®PF_"3åØñhÇåpÖ4‹T÷”£ÆªEÊ¥¨‚û±úWŸ°¿NNJœ³.NJ+sõb×jíj1ÔK¡I(Ä-襊~ÚÎ!G{§wÉJÁ¯Òº¿í·a¡¿E´s4zRë_ðƒlù‡·i±ƒûTá;Ë'ËŸÄö°;‚[x\ÑÆpK¦Õ¢í~ñ:§Ù¿! Y-á#â{oç•R0¼*µPªæ(£*¦3¥4@ú…îxPšëv“@<`´ÍËÆHWê|Ḧò4Âøì<æÀ8JÄõPX;ÐÄ(2(ÇA§ð¡¤nUÀ,“ؔڛíçq þöDœÖ->¾ë³Ü³ÀTÔCþÝU³!S4ÈáøKƒAkǨ<ìÞ.àg¬ðL”2âÀ‚´fVñ,4ÎÀ¦gZzE"(Tóòg6T_áÎäžÙ–èCdÍÞÜ25œ¹/¥¢bøúç玟7dµ®¿´ðyšŽ6¹4`§aÄ6<¯êsÚãݨˬõZÊæË¸L|Ö?èèsê(¹A¤ÃŽÕ¢îf«S!T£´»\Î ÖG²ªÊ$©‹(çX 1ÅOð»%týÛŸñ"¶ü1Á`áž'Îé_«Áœ% e½]H¹ cÓ§€³•×%­²ªPïFÀQëÔŠÎQæêl¬°ím¤“¸3½#`puÅ=)ñâã”a>Åf—›þJ)Ðmc$uÏD_'Õ°ŽÕ¦öø*u´Ç¶Êmê½Ñ (•Ê/-Û¶Jßèg¡Q­#ŒÎ×ûFt\2ú[ÆéÞ˜’.L¢¯mŽ»ÚÅÐóžòÖ PºÁë$€Elû>Ž\˜¡„@H îPY¼Ô‰ªí<ÍB>”Y=¢ç5/ñž$¡…}0yº¶‘®?R¨@{…yéØâGLدµø9£˜ ĶúÁ§Š%„—^É+ã!ήzæÁQ1QÚƒî.m|ÉdYáÄùnõ'‰ëZìãŒ7m‰&Hæ²*Èxdà¹H5*-Xhím}‘´Prš¯ÓG60ò+pþ»"—yÍ™þÙ /G'³)E3ÂßÙ.Êó©è+¯Âüì%éCÌ-<Ð+_,“$s­æ ‰Ž˜è%‰šNBŠöV(ò3ïh ì>3càËu 2.(O.ß:¿ÉV\é7² »;gˆjVÅZÆhØŠåǶ~”gB'š§äŸG[^‚Jðoo&uŒk‘ ¡õ:øÙuÖhïSÀ@)øžòíWìÇ–Ìfl§ÄûhBÔÈ3„¥"WyÜ5±-„ÊC½gLÎúï×H%ÛF›6þ¸K?ÆånKaT§Â Om$M‹ÏðfÛê5,¨Šò%8Èœ(¸"O½LÛUqÔ?=Ýïκ›>?аUK}¥³ ZúR‹G`X«xÓ­˜±@ð_l¸„ HOË¿:ÇpâXS8x$X˜¯ý4Jí~ÊIò{–ªfV{y\ïšIG`Ù”W eN+ ÿÉ«–Í]0™(Ø ÎJÞk"ìr|Bq!ìþíæ|ĵÞ?’ˆ?imwØá¨™N†o×]ÉlÃÇ@L`1Å Úe•k“kÈv+v*>L^Ä%fÑ]±%ܦÏL~wë^à\¶Yð„v­áU6€|¾Œ÷HùI1®w{íÕlÈ·ìqE…ñ+ëÐ9œo#{²i^ºs¼aåRÔ/u”q¸ËºNã|®¼ãF:2ãÎI^{]tVpcµ ¬N(iC >t{ˆµ" Ú½àù4ƒ“M<9÷%r,Ma¸î®Ò˜PquŠíN‡ë0¦‚½…¤ä=-ÚJowŒvqásæ“X]TD‹|ßçkEÀîã[,zgÛ…+f[%2e« zǸù‰,`!CNÏ„­L’8n7eî W”OÅÛ_Æ'[­¦÷A¦?û‹Bü¬ZÅ´tÚâx‡‡48ÔÑ}-`Ö””âÀ&2æ+/•à –#±èçáÛs2°VÎÕZsé»-!Ò×¼E¸N$D͆ ò†*E¹F—=ȱéü(PJ‘ÚaÜ4¸JwRJ †™à[§Ç:ðU¢¥égâ±Øå{G <™ó©Ú|Z‰:j %›UsU>|…< ‹qÑ®‹wU‰¤yŽäµ©ÇÂhHà…Žº~\c/ ‡×³'Í ²²{vgþ[|õúÔ/(¤,9("‡ö ŠÇi®©þY¤ŠÍ®Hk³Êó¼ê¨ó…q½µÚ’I ˜bÍìêÓ ¨|«ÑA›3éà‘îµÐÑqTïº7]6 ƒqZï#²a›Ñ˜ã9§»øúa×´sÈ LìÃÊN„Ô†¿çQ›<~H‹\ £T„xÕ"À_÷ÕO§Eê^ͪÕSï¼Ô!)®8%õš€8úDȤnÇߨ²Ÿc@>)e+…ççN:¹®”T<¿Ù¦­ÑRª)vqw{%‡sãñ…ÞÂøîO¾Ÿ^8ÖufÔxԻɃIïq0Øf¤dý$}¦SÚþqÏÔ²=ŽÕIJީûS·‚¦ )[™×—Èʼ³áýÊíý_#/Ë[ß§àc×üHÿüôÀ—Fèø¼^žô±sÆûžzÏ4¿û!ŸFM`»JÌÚ5œbín½`zŒÚý,y Áw–"öÀL8€H‰J×­Ú–ÿ˜²ž£j ’ùãGr°ƒØ‹ûˆ¸e[N+"¤º5o·Èf?B‡@^NÐ’*"X;>/nìåW—XžÈºG'‰E“o ©ôàüòkD}£²ëÔ£ÿX¶þÇ‹KÔ6±©<–‡DÌÕÕüòÕ#̽۠ €¶¬CBU̘±ƒVY¶mDUÿ›¥îðÑiŠ0m[*©]³xZƒË’“ÊL&öæéD'y'–B]?}zfÍ>Ý¢ºÇÖ¼PŒøIÙ ¹ß‡!Ônó‰@¹üÎ7ó¼ @gŠayÃpI CD×À f;9O[¿æhr5GB$f¼˜QUú÷WhJmýç2K>Pþ‡®ËàÁf3µH'bš»×åù =ìîÄ“ô>Š_ŒÐ"‹“¯0Á;&p!ËX¦>¥S=ʼM”JÔ®Õåz(Û•ªt¯/[hIÝ©b@VvqÌÊ#h¯‰á»+Wp>Y›ßÚ#ï< LäPŒqÔsØYË@/È4—K«•è3‘SL­¤”OÁnL%¥k7p¹FZv™ Ìå}&‚q_²'Ø~a;Wv…îùsâ0‹>EœÌsп×q“ècZpZ‚%GažPë›ì9x²&ÕñRæ$tH¾ÔK@€…áØ^$Á±èh(ÿq*ô>¨ fQ㡟ÔÈ{;“° ” Çgû»àå’‚ùãcˆ•eNûãÀqn¾Ã‚0¡–"ËÁÂð}Þ+ϲaùCµúäë{¦Èª°KW/£ê/|['›Ý^Ï;Ól½_Ö)\±ª´±= yÆ ¡hHÔÚ¨vkî0ËÇïJçƒìý˜®!Õ›÷Å}q·xNŽÊÓoIž4uñû"‚È­Lã%l •˜~Ü¡-Ç«£Ø†$¯d7Áȯ‹–­¼ož- pÜ<Í XU¸¸Ê`tŸ!H¡cÚ ef»Õ|ðp¡i2½º …5Š»N-‘Ì ;°Ý9“&±1/ZÁ%´ÌÑ%çpLÙ\À¸¸ÃÞ2À˜Ï šœ±ÑÎàúÝ™6ù2X ¨ëÏ)fC©EZD–>©7¬UÊÛ|g;þÉ矙“¨_˜fÿÁ4Å«ù¹h§5÷»C…´›faësð¥çd¹Ç‡69_$8äÂÑEQ“AµËŸÀE²'8h¢ŸÒ §:ÇŒÚ`Ùº½ï ‚x4lU¥õb´ =\µc¡íNR=ã ’ªVÙÔ¿ŒÑж •ôg"—7<З¯Õ•kKu>ªeP¨Ÿ+^èZ,ÈÓ²?â7Ã$¡îÌ©"F•ĺÃ>Y¼GË¡¢œ¿Gª#µ.%öÐ-Û¿Ñ·œÒQ¸E6æŒwKehMÍIÕ5‹ÂöÝê,ÈB~^Åßþ¦i'Bÿ`-žý:ýµCõµj€1×½4ª;ëH´·=–è&Έ×n;Æ€áõ"xqM|JL·!ùŽ×ëücÿ9;ª ÝqÅÇpdJ  ãâhUÈqG’·êøÔÑíØƒZº&ßhæ@áxƇkü}‰9#h,1ŸŸ>¨ÏâÄ-iº1"QW{–>ÞÊÄù<|$ãW‰ÑUGcªuÞÌ9ûRçdü e ŽÃGed  ¿‡ÜðPüÌЊKS¸O\“whšbV $œ¡35,¯¯$ë€ú)øs(l¾1ÕÅL\ݧb†"h® Ir¨–¹ í©øüÈ줇ÒcDΨ?GSMÙÓ…'E £šõÆX¸p f} Ö¾·Nö`3ÐîëðüÈ÷1†…©¦1Ä0×àp‹AÍiô¸ñ}>g¤àRßïü^ývÛ´ÿ½OELpA™øgÓîÀã4JáC­tů+˜áµý§ªÚ1è’3¦ ñù–6hgÚSeCéºÎrÃîƒû  (Pm}—Ç:=ÐëànNBµý~ïJìüŒ/OöðRß ªÒ)p…ºPÅ7yt-gjÔE•_s]j{ÞR"ß³k¦?[Eš§r¯ãï‚el*kCåtŠq5¼¥²LÒ1ÃÖ Æ<¿u¢LŠÖ½¤ %a+¢Ê;¸ý².hBެ¾*‹4ô¤Í3ÃÎfJ8NcÎ|ŸV÷ò#8V–•wcÅѯÂg(µ®n°»I8€lRhõ ÌÊŸ¶ÑÆ>ºJ¸_=ìml¯~)jç{òÌÄ´_#&5Y'\Ÿ‰ã°‡À'J+ im$°†4 çnòM!Ñãʹ‚GS¤zäOQƒa®g¶ù U» Tëí¹Ç¸;±~Ýz©„8… Uc·ÀÆ [Âé|=UÚ«Á¦¢¾#b%”KüÉûâödX'‡ëF2ß늻k”L®â8äéC‡Еφ^@!„/@ÐÖ\‘tõõæ4è¨×Q0GIÐÌð+ÆJlß+¥û+Þ¶H)œÉr™÷`¬ôD~€¸Ó»‡Pƒ.%3 g²Ø¥¾fs/I玔›fŽÆ²cÃ|i$šÂ»LI|¯6²U¿Š¼ßÛámY÷TUXÚ=;Å,¯šÇ›‚¸u…ͤ,Áí­lÊD¾UŽ&îžœ–û€/6À¤k{ÎÏG~½«Ô”Ó¡$KÎr’‚£íÌjœ#÷ ³O)BœÌU%³Ár'ø•\3#r·=Lpõ¬ñÝÕ¤C€ª¼#ÐJŸ&ߎQ_Q'¦¯™yØÍ ‡ëž¾”¤Ãlï5#`Eå£Ê~×1‹¬¿æ;jõY³‚ý´0wÈg³”ïÑÙx:%-´ù ßô2bLúîl;“ž `®|VȨwëá\ß‘3£›4¿ñŸøFÔ2SXBÙ~%v$‰FY/—ö‘ïiCt(‡Ç÷¶¾"Ql‹o³7¡À}“l$˜éÂØsÝÛ¬˜}Ò\Üß’MoÙþ ùÌ15"9Sˆ9Ä0ÄÕÈmîˆù™oCµà"²ÿ¥Ãª¢Î§mbð—nS°ÁÙzßrÝû°ïX´ôZêõ½î:[¿ÌÙâ„×c+áÀí¢|µ‹;B d›+ÐGÙ wÔð¿ä+_ç!Ù~úÜ¢n˜3³R¥ß¤”"?Õ*m¾,,Š-9™ÀÁûMÄQ4GÌWž@Nƒ«ÞÞ÷n@ª‚¶$AÍ‘?y¡³þ)‹”ÙgéÇXæ;éÒîQ=Ñ;Ê7ŒŸO¾½ª`L·[ÎÁ²qÜF=à÷ò•ÞÒMÜTGDÔÓÜçÂËšÊF;ëf#Ç„¦ÌHï¡÷J-Šx=²‡3¡’¬ÿ2toÓ‚VFõ¦š_ ææ©’é/ÇØ2z¡µ‡¡\¬n0Âö­KÝ$dæÇQ~0¯^\çÕ.DyÞtQó ˆ¶Åí–ù±‹1f° wÝœÇd†“¬?:³YêŬC*uÒnÐ÷¥%/#ÌÆËÇ,ÒŽ.–Ô1_#ñÿxíýõ‡´Ú0†¥ö€c@M<䀩c˜\"à×Òê´œ2Ë˽~e”TïìOé§ •ibþʤjw«<®‹@ðú£r¢ï¬ÔÞtø’Ðæ¾w”*€+7®Ú©cTwö¢î+ÅMÐ×4<ß%b#è9ÿ¡<« ³œØ}ÃØÀ•^"òVk»i} ‘V„Æ&VP¡§µ&¬—HÏX‘:sq{UÂL6îåèQñè”ë~k(™ƒÜbrÖ_kÆõ"RK<¢A\ÑŸ!­“m3­ë[¦¢hµðÝß>ª–úi c€EÂEVÜ©7ÈLãIÜ¿þ}–¢ÖÚëZ$¶Ñ¹®ª#¬¥_ùœ&60“ªU¹ ›Émáé»eÿ=V=(³åµ' £ßÔô¶TŠÌhW(¡Q›VŠº ÚN'|aQ–ÎϪoÆ6<“Ošê½BanŽ?o–ÅÀå5·Jb¢á,—²îÔIÚ…*ð¨®ÁÇ™Ä&ÖÕiE/zúm"]dYh ÛN‡GxH/ŽúÈŸoc~ˆ)ôy™/´6Ș<6_Û,@¤^wY™õYâe:*!Fœuãp‚ü¢Ä»â¯Ôʈj…š¥ú-†¦ #Ý gI›²|áKÕ)JfX=¥ê”\7§´YF×±z®6߸¯´ Ì•åNå (Éïɇ"$ro¯ûfàq8Ã&¼/‡ýhÆ4)»–xïÀN®²ùô_2É€¹ÄÆš|½ó¼žµÊ Û!ï,ˆp•e/×#òL™Ú ¬9¬4ÓTÁ¨œl`²/uU³LmõÂ%""j âvu„­Ï¹'ÀÆVâE3bç0pNfí Ydi°Ûf–áØÂ±$Ý™èâ;RµÊÆKAFNc5xˆ‚eŽMÐ&ܯ„щüdɦ“Ùlަç·r°x|섞\ßG¦¿8{—¸Ï÷‹w§`µÔqt› 5Yå³Ô¿gFP•í¾3÷»žK¶‰c,™è¼Üq¬Ó¼zÀfDïg’²*{1¼°l}Úè}ÿ£Tè'š›Q¯·Ô A`Ü4y£K”ù °Ž‰Ð‚O㱄ÖÞÐbÿ®íÄ=Š ±~äwz%^Bíë6ÿAgi¿oó:ƒ´ nœ]”¢è ´ùóU ±|ÎE™üür*­LFå^KµüOÇø2ìÑ_aùCÁIa£vä½­m|ÆN+ém÷»‘q4›å|óÃÔ©ËvÁPöÙ³Ê|íý7…|E¨/Ì ²˜96²æÁ&µ"I·Úk¾ZOL%Í•…µØ“ó¬d¨›°ži¦K"ëdYGX[œì儵»²ŠWàBj ^FâýÖ]dà®À¶wIh&œO·dN{½”q³£F…^ñ_y^xºÅÖžÍm}uå¹ç_vÉœ´F;ßJÑ 47F†*HɹÌ`r3ý@½Óø²pvÔ'ç]c¯ž›CªT8gèˆee¢ BœÿN{âÒª]˜JZ¬b¡DÓ?æÿk¶B9£?–\F¦ .åył̂ÿÂN,=K†nEþs+‡’¹}Ç`P>Ý&*‡(4™Ë3­|²}PQ™Lwp9é¸ñúuÄ\CéŒY%üS±XÏ¢‹UŸmÀHEàÙù×lw„Lí%"Å"!LŠð}ƒ” áH…WHÝW¥®Õkê Ècbaº÷ŸË+›V›k&Z|I–ÆP×_¸­ bÜ”³(EòDJæèã½DPöq3Ö‚C·½ÃÙn?‘ ÆÍ›Áø5´f“ΡTÄ$dPx³’DZe¹‚E;“„Â;ÇÓF:Êm§vÂubËô5ªÔvÅߺI «Åp˜ DghMèÀtåg5•0;ä–dޱ[y¤Ÿ«Æ/ÊI”Sé׫¯m×Sê2XÒv9'påÄ´uG,Öp45)•e‚+JÚ‘’ó;-ŠŽ¿ Ô2‚$Á1Æ ]Æ _!¶‡KÎC"ЉìŽâ!†Êdæ½H§OûJÚP› û¦À–Ž•ùŠw]Îõ¬UÐÝ¿.–àh%ë ò]SHòî6¯~ÇfÜçië#qS¼×¿ ×y-³§I§“vµ0ÖñÙP’á’ÿ›Io4ZÝwü(7(.|,öPŸIŸnÝ |•¿–¶·Âå×8‚jiEg#‰#Œ×­˜ëcAÆIdvC”úûÑÍ`e¡«IÙ¼Ž÷?÷æ²ñœTý“h|"D ¤~É ý‚)Ô¹‡y‡¶cUG=O-a°ô3™’ü4áúÒß \د¢ÞÜ2„'Õ8µ¦Ò0?‡’&U{‰›\߬cÈv“Å”hØ~0F×oOäÍ‚ï…̘F† ”&EŽûö(‚­ä=ÎAs ‹&!†bïðÎPü1NJb(Ñ Õ–èçôz’*Õ{H·¾R'% @¹º42ö„½0Ïyh;Äs•„óbaƒ9QNB­jJqÍ;Þ4R´©vm]xQ×g!—ï6¡´‚xn’´¯û©­k‡+„Mæ®ÛrM$+ÑaT°Ž´Í—×íbãN[™TŠ%Î7¼Ò¾0Œäü¨ýdSj¢Z`Ô";tŒã*£º2ïWÂi?r±° A L¾ÃÇgùJç*V>²ÌŽ<_Úó1ä wí&uK<;FjWaCÔ8k7ç9PúÎS<{³5üi;ݦ¥û® ìQK\uTÉJHv)<F0YC_ÿ<ßaùË2[ÿ>fd„ÎTÎÈÕÇ|ÊÑ÷m¨˜q4ïMcQS…Vh‚ßÂS¿aÙFuB%i… ÜŠ¼\Õ\D/&ùÐ5¦9?uÁ"º4§²Vô¼»³,GpN[F†ÿÉFÔçr|lD~aîfFp?€¬¸t’ºu=äÆÿxG5†Å©ÑšÚéª:úl¤‹fß„n¡~òÇ[ û§+_åçDqB¼Pœ’Oȱ?η–E°;6óÕ§ç‰È];Lm~ÉzV¹1dÍ8$vJiD9¬Ç¹¦-òž‡ÌÄ ŸÞ‡Í}¨s@BxÑÃÈgÙ¯z]À]²|s ßÎÊG›ÍPΖ’E6…j²ê†u{¥6cˆÉsV·f 'ó¾RÚÓùÆÿr`èåx½6"Ò¤ê™v§2V›+w¢¯Ô‡S5)¸k“kTý û:µéšx±8‘y;f þ“Q³ç±ï½…´*Êó]}_u-ú–¿?°«µ¦GþóàÖì>¶ü^ÀýÎoDÚTí"ãP2}ñöfÛ÷q‰¼ ·±{®\šT¿QÁ<¢*åå?e´ó_ ©.ÇLÖ=QRõ‰X•sùâCKä8{3$nÄðWwÎ~2„×-Â7ëCLhx¾¡5ŽŒÜ‚¡^„FÆ×_å“À™J~qCÀ&¦. ¥X€´GPϘêÎ;åÂäQJ.R­¢òŸ áPš3žo)§½ðt–ê2W¬† SžÏžbkvÜEô¶Žögn„eøò&-cæëØðeTÂÛ™Û&;Ì©rë$‡-dG¸2‰~ÕhS>ŠYÙ½­\ׄå¹í¦=ý£¨$·fÓá¹²Õï¥çUÅHÔ‘"–>m…<"¯Ž¶ >¨CŽÞ×Ít¥Uüý퓚i¿í\QÖø 3Î_­S2Îä—|èëyÊ0~l¬’Ÿf7˜Ä:@‘ú­çJ“W½ø²tºL©uEj›'Ê ¿U })þ„õ•wƒ¨µ=§‹ … Tf"|Ø¿¼uÛp6rŒàQÐq®AÄÍYÔã߸/öóp,¿ßßðëé±Óòü¬±[²µ4öñ±å‰CËÖ1‚3š êe|Ù÷¿‡ÙÚ6ûÑb:ÚA¿L°d|S¶1O¤Ã µÃö …ËJ‡i¨ÓGd¶76=Ë÷:þø;z^ôI@ Èoñ•(ø¿¯z¤tß)>bO­ó·}þ!è‘‹0d±ÉuÇ ìÆØdvj ½šœáC0d¶åBçµ ×ö0ü7£EF<Šô¤¶Ž®r}Q÷ìÃL¬5ÔÐë¾jûòözÚ— Èb=ü…çÚØŠnѰc+ê—q ÅäQ5É&n©þESƒlç›Óe¿¹)MýOͪ° ö…K:bc<䙯SýÄ®è[}›Q™ã«FŽd"˜Ù!\˜½+–î5!ÃPœW ê¶²xˆ<ÃkbI2þup»1³C—Ü»ŸiçN­’ûõ ™ÒŪ€KŽœÊu¶Sÿfä~s7h_'ExóÖ‡Ù—ÑëΰïBí …˜W_û^µYŽô ÿÀFõ|²æšk‰äOâJašÀ!}ýXP<)›¥ÑÐΪ jTësôèXkA¤¿éªæö—Æì Çy&‡þ†þ ãÕ®m¹¹“j[æça¦qÍÿÕpSU·aŒ/…Ä™<_!§o]²<ò]¦#Q^ih’¦êmoØ‚ìêáryša²ÀéZíw#Q²ùyVjôJ+"Œ¿5 Z©eœW…öʧÓ©íX”žJ?¼ç¥×Â1Ä ¢Åí%…3dÀ}‘ ‘«u!Ľ„MKú ¿ú]-±/æF¦ ã —{;2ðMœÃ«Q‰I[=+f±ðïó7^VGK9š…ÝrCºX¹ªh°žW„oS"ö¦/“ÀŠ»dâ&7rj—¼©ŒaŸÉÆ…ã¨26Êù!˜'¯ˆ‰Ë¥ ~‹í^9RƒzŒ8 6 ŽÀ7ˆ-“¦·ù©‚;”ðcGþüA³÷uÛÜãa ²vê8q»Ë™ô»9ñ8Çq ÅCý'¬#Õöu×´Õ„Ž[¬C6 œñLyKw Èï×i«mQO³~­ë°´)߈é}}MW5Cj,+:k‚E‘•rØ™t°y÷ýÿLÛÇÙT|_Àz•ÊÄúuZÄ^Ù_Æ¿jÂhMÏ9ÕKµIîxèÏÕËzÉÅ ãËñöþó‹ßËšugµØ¼?b82(êEé‚+‚Xº(ù£d™¯®Óa}•þë³óHÑÞ9=m¯ä¶žbÃ_Âìgµç¿¦õ]åÙôuÈp1v»ÿóu j¯êP‘™•Ë.š•΄²–ÑÈÖ|Â3˜x0¢þIÖ´š”œ,Êê…ÿ×À"–@È“Þá¢ÒC#ª¦¥ÅWÊ&AøËþJ â³YÁ8óv|Øß=_õ:ÇRׯâêî¦ÿRN8ÀÍ*|}ȾÑËyõ¥%@Ìðà|O‡=Vµ4'M“ô«ÍWËã-Žñw>ý—ÃÖ=9­ ¯oÉóÍ'1zåo–Fç‰PKF³±a!%ñ©íÜáÎQwC€ô¤|˜ÞŠí;û›¡š Ã6ؾ£¢Š éÅ>%vöƒŒݽ§½`Lr‘šàGÒŠA§þµxÕœ`w_ݼòî’@ЈÀÏZýõcƒÒ›ý5äÙEJñ}8Ð<ïò°·O·-”79/¦|›+„2Á« ̆Ú!..=VUÖ·å+‚Зnzu5ˆäò›üHw‘YÊârëhÜa^×w-w†ºóï!É+,ÕÀeñAt™JÐHBLâÌHAÙŒ„¢í7¾Ò+“DˆwßÇh7-¥W¥•²ýܶ…­®J>úq@ãT øß«eÚõæzîÉwÏs ± UÜÏ©ÐlZæ. ’d꣫j®VG­×4||¡i¿Ö@L†ef)KÌt.€ Ñ«ö}=¬½«Gì¸mJO/î;f7g¿ÅûþÌã»èLú&ö$Ý%߇Z¤ù·ÅE)¼àÕpie©„ÑÙI­q’ ’ÐJëÊd@:KKÅ}¬î,{ ‰žèÂÍVf/v÷må©<šú6â³5Ç@ñÕ0ôóÝ·².‰ä§ D™g|GÝ,›NŸù^GæÂâŒþr2 †Lsµzlj^”߉žâ&¬43NtÐÐY̲/¾Úo«E³Æáñ˜*X&s!e]ÙÚrÅ:B+߯màÊ ª·IW‹ œ¾²XSÅ!þ×´º“œÕLáƒ!©pQ‡…üRrúâa5Ù¹€fލ>£j鋵a"ä0gV@Tïœ.Ûš(Õaqè>A÷@ü†—°—֓҇µC¯#u®ãœ°èk¾ô*uõxL¨ÿÌèþaT2]Ê炳kV/œþ!ŸZ DŽl1Õ[¿Ú ‹ ¢× µÿ {Zõ”P'Åzg€»ËÐã…pÿ¡Ðñ¶SÃ\1¸Ìæ—·m—šÐy=þ¢Y.öíQ´YÓUÿž”@ãû±À}‘l‚ç`^ÿtPш››>çK>íð?{,Çè¶–1̰Lƒð)¯ýaßZ—Dì«UZBÅËhŠ•MRSA²úˆ*,ˆW!6±õÑ-òKÍ¥õ­ãvŸ–v/¤lœ¸‰w¡µçÞ–ZgBñÎïÅÉ`åít*Ú¨°)ÓGâ„vÓ­T:ô„ôW¾ Zý ;\®.|šä:ݨYdýÖä˜$ÈNe² éÑAR˜l¡«uÉ &"x 6$?ç3_«¿¢¿ˆüÒ-FL ʲ«?A€_ÒSY¡„­=§nK^.sö WðYq¥«"t©iˆ\I¹ÿ¹¶üÏstáìTœLó_ÿ­ÂX¼ø}ð¬(/æȯËy×}.Ñt°Q`Åèâb|Åj ˆ@ð¶©Ûh‡¸K4ÏN^sYñàÔ‚7T˜ã†(iL:\iŠòbN^ÐÎÇœ'85ÿm)ÿoÚ£C±_§ÜE†§"àÔ^¬“½Ñ5©‚b'Λ éìGKŸÆoâža*˜Â{>NãZÕ â|KŸtX³‚o%¯~>,ÜYóù¾>pK1xŸ±ÚÌk ÁûîñÓҘϪ'd¸WbøZàú1?Í&Ë™Dpk×xn“ŸÏ^s«¥oÖô"&>¹w¡ÆãßñùIÌ(êsóH@?0¶‚~¸01áù…¸îfë‘´}Ö·ä²I4î¦ß¤Xñ«f›¦ÛÞ=6 ^üœHÌÜŠg(£2k¬¾Dá{ƒ¥Êu±Ùó ûâ뿵‹Î²®4B²pø6â6”Ÿ7‹W˜ã~="Ìe[Kõ?c¤ŒßM6^ƒ‘·!`”´nú™‚;‰n„º”ñòv’< óq½êq’û ŠÐŽõ>9-V\Ч¥ò®­zWóB&a}=ð¯yUÜ U:Z¡óƒúe¼4;•Sj‰•ûþo¹ˆÐVdE•±õ›>ÉH-b  ´à$__Ž´ßÜRïp_5‚K4Íá0ñW˜Ã;º/93'°öP´”k6ÕÔÇœ#ˆ{),’ð©@ÝÄÊK骘?~íª a&¯ÍÑÞè#ÂkScÕáL\‹% V¤Û|y:¥†@v®º¼ÏÊIÊÿc(¾ÿêyåD[ H9ÀáÛ:~hÖ@»äñ„Â_çÕ†ŽOQ!j•ÿ‚ Hooň1AL*áâ 4“¹ËEœû.zÀÓ˜z˜¨ÃÄ8`Äk{ߤ?$rÕk+],iÖÝ & Xsý­»å?¸ú AC"¾_D°£…ÍVYÍ-Ó».Y=–)¤(>;öµÕ/¢Î‹Ù¾KÛ@ëÉL+ig)èºÔ=Ò `Åfr{oI˜þ¼/‰0{fé‘ÖÒ²)†»¬.ÙQAŠ`ðvô›Î[[”Èì”æ>î`ŒyÓÜù^ÅšL{dh$¸–Vˆý_õà!n%@ƒSyþÕóžâìHütåèÁ°;WC‚P j~h“k›H,i1’E—Ù+dhNP£"X¦:ç”äîN eWä ˆ–d¦ÀíGf$R¥+¹é¦ßTÜu¯:¥ÔM­ù~Þ[4–úÇ’ñl¢ìý ì£²2耤A)ìá©nq¸òÚEKa%âÍrÍ_·Ø{5¸[L³µÆaÀå›ÍYÍb&¬ô¥Q%š›Ð)e:nc4WW¹TIOù}ß±P´èö÷íÒæþ±AÀ«+Ó»gq9À\K?+BÔBý€†Ð[M\ˆØ'øå¥’rÁ:–ø3Ñ^jŸ-7 ÒÜ0’Îý¯R~XÙiúÍîý­Û£ÌŽÕsÆaÌØ˜ÖÈŠèÞîtž‡­ò§©½¦tAµ»ŠmnÄJ I¿âçkMiî,åÕóÖ‰viì‚C·w&ÍÛ0¯w=u‹1<`IòvÑýîòýûe¡¼Ã%ëµ€îàba?¡ºc‡Ñš³×Ï©–ñI§¯p´çî:Š0Û™åÆ¬ƒÐ½@æ QçJ"> áEõï½Ppšg’îw/ ñ:…Ž„5¾PTk<‰9ú¦)Ù‡…éÁPÁŒ6x|=åy €X <êSãJÑ?RÐêži¤ªv3ÃóN;T0Ð LbñˆZ”4¡¾ÐÔ^ÙÆq~QI°&'‰–‡{V@4(b>ŠOŽh 6í`òÕ£˜rJúþÉï:º& †îCø7t÷ÊM£°+JîÖG(¸kÝɪ!i,^Ü<7¨¶‚™¡\wßG÷Ÿ‰ýýf“‹}LrhØá\ð¶ " *”€•ÒatQ1fBµnY‰#“²W‰6ÛËxKsc­ÌÂÚq•<ÏDIFâ0[+ƒ2üÄm}ܹ yì³¶Ýü*|îmˆØ©(´óUÕïâo°Ôü4cJ„ó,ÜU¨ §@n—ÖgKŽ/àÌõgž¬; d‚S=m h¾@÷â< ºê– Ñߣÿ…S¯éÔœ¹à–¿nù¸øA ìW.,x0C~0ØVaKüI4j ·k"ª›ªh‚/žwج9¬´!muÄJõ]u‚7ð´C:º¾6¼ŒGÜ«$õ¢P^ýr&[].ºòlÚgղɜ‡ Vªïxâ§ —oû¦a=Ã.©ÅUVb›mÔ°ŽªylBŠ;”KÛîÔEX}^aÉSªàƒÕ¢dpS̪ uªÎ3dò/ÙŸâõlÅêóÑOJN‘—ðx£~~m†þÿðÙJƒe×K.Qij³ÉyQGý šâõí¾U‹š-lø:cåØE3ÿå£GðÄnÏ-,{Çu)¦Ð®|ϱê›Jð C<¾Ý×"ŠdñÇ4Æ«]®7N+]Lí„L õ~Ày` C<àò²Ê—g‚o„·{ÁEä >‰vððŠÔîgÞnýdðÉ̽°e Ã5˜\k¥m0ë4¯,pݯC€GÙFëŸr,IfÌbÖnÉÝqȈ kɼ™ °yjL*®Ï‡PiÖ‡u⸾ãöP*<×ôɤÐ)׃*ÕÌU<™>O_µ¬IѲô¢‹Ð™x5'ª†â$•*Ðxq—8èËÞëÄ’“G…ä¬V ¿½,Ùÿ+‚£BÙ’ˆrMn‰n¤Y鯼I}3Ÿö€¿a [räM¯¡ ’úòÁEÒŠ'”J-^Ô}ÙÜ2|ŠzÔ šúM‹wúwí"jW?ôy+–iá¢õÌb@gURÓµ/´ÖS G¯ÛФå=ßœ!ÉÁ|§Üþ]2LÄÉÜõ˜l`Ð"IÄ]OŠºq,ÀôÍÊ92«jý2§òSt].D.á²²_zÓ¢WDè„:ú$T—ùDll€HŽæ`Ý· xo:ðg0k÷Vz_›ŠÙëz#£ ð¸Œ&Ik÷[®Ïd¡ ÿK4®¸”Ïïn¹åaFé/'5 7)CÃE,S™±!r“a²ä üõ‹9iÖG*ÖÁ¨bNÑ ì4qY™ 6J¶à* .¬xቆÝ4 e4N®*ìX6(ªª†¹¾öBi´ôg&è^Žô§êŒÊu\ÿ\ƒz­3Ã,¯ô‡) Iš=Rή ¢hÎu•ÑŠê gÂ¥1ÎÕ>”|“g¿ ·?¥¶Æ ·ähŠF)h5%ÏáïwF–ãÊ'D6–SàŒÃ¯aœåjÒ[³7ú=á"NZ|õ\hb’}\ÃàêD tÛÕ¶·d¡i¹s2iŽ$ð WR\G(‚WæJŸÝÉE%s'‘¸O¹.Ceÿ{6YÇÍJhPðµÎ°E•œsU÷Ž~š7ôWµÈ%ÀIFçlˆ”ªxñ2 ©€>ÔÁy®­ Üjª;@ó½}ACè/Z /„FÎ%OwB¿/,)ßÌ]w.?)žÖ÷üDÕ~z‹xïHa9_¤/À¡«9é¡»wIÝ¢f(ß'BTn˜nгg¶ôÿ kLI endstream endobj 70 0 obj << /Type /FontDescriptor /FontName /YVNMTH+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/L/M/N/O/P/S/T/U/V/W/Y/a/b/c/comma/d/e/equal/exclam/f/ff/ffi/fi/five/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/parenleft/parenright/period/q/question/quotedblleft/quotedblright/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 69 0 R >> endobj 71 0 obj << /Length1 1601 /Length2 8259 /Length3 0 /Length 9312 /Filter /FlateDecode >> stream xÚ¶T”]6LHHwÒ=4ƒÔˆ Ý‚Ä  ÝÒÝ Hw HÊCHHwI H úÔû¾ÿ¿Ö÷­{­¹ï½÷µ÷9×>×>kXÕµxÀæv¯¡ÏìN<ü¼@q€¬Š&¿äpXX´aNpè_n](Òf‡ÿ€, qº÷ÉAœîq*v€’3À/àç@ èo R q™TxJv¨#‹¬½;fiåt¿Ìߟv3?$Êý;¶…"af@âdµ½_Ñ hÙ™Á NîÿU‚]ÂÊÉÉ^œÏÕÕ•bëÈk‡´”âà¸Âœ¬šPG(ÒjøE  ±…þÉŒ‡ msüÓ¯egáä AB÷8Ì Šp¼ÏpF˜C‘€ûÅZŠÊ5{(âO°òŸnÀ_½ðóòÿSî¯ì_…`ˆßÉ33;[{†°XÀàP€Ú3e^'7'naþ ;ÚÝçC\ 08äõ=à÷Î!€g` äžà_ôÍ0{'G^GüE¾_eî»,0—µ³µ…"œq~íO†„šÝ·ÝïÏ“µAع"<ÿ2,`s‹_$Ìíùt0g¨¢Ü_{ο>K¨@Š‚„PÔÍÌŠïWymw{èï ÿ/÷=oO{;{€Å= ¨7ÌzÿÂñt„¸@NHg¨·çþÛÂáç˜ÃÌœ¯¡–0οÕïÝP‹?íûÃGÂ܆À{íñ€¿ž¾Œîåen‡€»ÿ ÿ}¾|*šZzšò\2þ'öô©À“GPÀ# ̉‰D…ïÿ®òÿ¿¹ÿöªC`í øoAE„…ô'…ûÞýMÃå/U°ÿ51€ÿ^AÕî^ÊPû¿ÊšÝÿðÿ?ëÿwÊÿŸìUù¿)ÿ7ôÌÿfÿÿÿ„!¶0¸û_€{%;;ÝO…ŠÝýl þªýs’U æ0gÛÿ*:Aî§Œ°„ÿÓF˜ã3˜Ô\ædfõ§„þ>…ûòpªnçûu×xøÀÿ‰ÝÏ›™Íý}âxV¿CÐûqúï%åfvæ¿æN@XA"!î8À{y  <ùïÔêö[Ù>^„Ó} àžž7À‰óëDE…|J¿\¿-±{Kû $ àƒüc ÝÇîÇÜö_ÿ½Jù ÿ˜Â‚> ˜ ô?â÷)VÿaŠøàÿšü@âßò ßýÕùáûdä¿ÅïÁN®vÿ¾ßœó¿¦ÀýVÜÿ-v¿’ù'ü¿úeæŒDÞß#¿}ßÌ¿íß—ê5Ù²3{d]Ôö£ Lãʳ9"9Á²©—ÊÁã9‹lw¾ÂÇJâ¨Ì XFžƒ“>}$\X—g?“™c¸ñÜk®Ã mIÐhýéum§9¶ÙŠ3ó™¼o´p\ÛK‡MË£-³åuãà¥ëoƒÞŒÚ¡Ä’ëà,†¯žOòõGÁ­¶·t~(djSc«Räîué8O”Nä+ÿâI–¼×Y_(c:ñÐ=ä$>r#˜<;Ÿ νcPŠãÂñÞ,ò4Xˆ¾üâ±ø^[À±“Š™Ê€’ýŒxhŒÕóév²Å´gIÑ2r(å yûðªm"?|›ÝcSUórç»1+˨8;%:©â›Èå–çñ¶ÅŒñ˜BYG JËnŠßE›_;r«ÅÃîÔࢵš¸ Í®«œZm,v²}ÎPM+²¤ZD·p(jN-+~XÁ¦eÿuÅÒ«~³ÅÁEÒú&¾©<]s»ßšef«ÿ΢6ã§·ÅÉc%ñ0—uN8éÑ©Y†]_ÇÖ°£¹-•1jK}}€ävUKRŽ.ÍòÚ,:çsP3 uSRfzn³`½:n¿‰QÍæ.GFåvk¯+ΰÞÝuÒRÒ;pey®X&Q!ös¬˜èËF´ÐÕagr…Ï«ârKÄ¡«JaÑ·ì!ù'zåO†?œð%xK-µžà©ÂI ²Õb•'ò¢…iŒAuݤ·âÅ/Z81¼¾=à ºóìÌ…ú•NÉs/SeÿthC£j·§¸í%×hôÞ²éwïøî‡·Á¨ð>=¡r7ø¤~æÔô`=\ê®zfÁþ+=¿¿–nªwL¿øÛÎ¥Rȉ9ЪgÚÓ×WÂó‹€ëÁPNòUY°¥pŸz@òÀ! Îe - H!ÀÊ/Ògôàd’CBPkŒòQ°yGšR7Ê…2½盳šàu W6þ\GÿcU9—‚†ÈlÂL^…ž:-D››Áá·Âxr‰ŒŸ¼YZoTN8Bæ%>5F~s¹QŸ°òc’àWdR3Ùt\,líxvaé\Õ½~@þzÅzžË 3äÇ]B|ÞÑÏG9¶êenlWë1¬"ž¦öSVá£×„XQ°7–Çt*e£¯ZU±ÞŸàÓWK릗ŒS¸gèC¿™ÑØÃÞ¾ÀëJ#ØI]ø(­‘ÊJž!„ÖcÐ3É‘r[ÈEn«ÑÃFYQÅ̇ˆm²Ée¢=fÐ6¥_r펚/=ÝI&$ÝoY qo8¿­3Þh,5ÅŒ9‘§ óîÀ’¸#?(a¸Uâ‰wl-Œ¾ôÖü:jÙ¿ ò©ü¼ã´×¹"ó`,5b EJÿ”æ´`Xþáûå|Œ=_þT¡«¯C£a“áã± =˜‹è-P;—ê^ZWÊŽ¬çpd¤–…ö`c¶¾'M§Ó ï–¼Áp?Œ¢>#³OÜA:W½”QAD4:虲HShˆÑº*ZÒùÝ>iÃ5/ÂlZ€ÚaS7 Üà%謞œ9wЬ¥ÊN½a¡î}©}ƦyÙOm=Ž’c7ÅO§RŠðV¾áB9JW»­ç2½ ü:‘3½I†BöI¨)º®L ?è)Þ›¡æ²—/UÏɰÁq¾–@rEDª‰¥'˜)?]¼$D°&f7dœ¢ôC"¥V?ÅØ+R”³T±Ó럺$ n‰zó×PáeàÊs~¨ "ED““º¹Ò¯Vªáq§&"d'lÉT{ŸLm8µñØ+ÞewˆkÎðv±Ë—Å1 œàüÝ…‚!< ƒÛÒyÇ:Õ±HÛ:³Ï¥Ç¼¼ó—Rx¡Ã&¾gJit„²6Ó$÷»Ulî 5{ Rñ;ÂBldKª,ôxf“nEÂí°8Sí¹xÏ-­á¹Hz+'š˜iAq"§ÙãYwVšgiŒ®ç ‡øU­ÏxU&ð=)ß­>IW¯ã+¤]zH^ªŽp)¥ô­W’ôаæ2ïgüŒ•Àu'Þe²'¸­BGPžZWñjo$ᢜ…Rn³–…ŠmNæxT÷f)Ø…:‡&S”_úi’¸…z0÷|p•‡L[TûÔ³½ËŒ¢á{]¡z·ÏFvŽþbþ£àf;Ññ÷jÿ25«Q%pZzm›”q11È%g"Ö“‹MÑÖ©i¢BÜ\•é0%  Qy‚f ¢é–ÛÍ|mÎØ®‰‹š˜á–^Ïyœ‹éß±ÚʥЪè•3'±™é¦ŽÍ\u¥{ÜRjì”,äm% [z®†ìy^®XÄ­Öˆ%¶RDµ:JϯeÉÀ º QUÛ¥œcrƒ `ªig?ÔÝäieodTªo4øà®¿ŒÝ^ÂhTé1½ 3ÍÛà÷î’ Oø e”'&AN¹NØ9û|s§ÜÁ2 ¸’ùµ8€pÛr¸R%áà½ùnGr§[î$øÆl¯7u°e4/ÿú¥ù"{§xö»(sÓ°ü|°Ê•úë ÈDa\èN¥Óƒx#¦ú¢ð>¸%ë§ hIJ9Ávòkas’1N5yláØGÕ:ÚVWT‹p—P+ êå$Õ³½lFzß!×{“J4ö—¶9õŸºoâÕìóûÌdM[¾ä®3¢{jü*Ą߿³÷²íćñ+¥ãràQW’$NH"¤åí££W!Dù¯*/wÙ8þ0š¥ûàýâe^%W½ù””±Ù’î‘>b ¼AŸ÷À ) Ç{Ø9:Ô ÆP[>‘#Œ zK{*|˹Քⶂh—aè’ÿ‰&§~'Õ’ e&U<›ß7nBÌvBp_–J‘ýÈË·aZ©p•î²ï^­»Nè¾YuªÏ"§›ïd4á´Mô˜#k<‹®Íñâ×¹Æe¾qÍ4ĵGÈÎx4Ÿ!ˆ<ì5÷u*:^GÒr³~ñŸÛ¡a¨zEÝW»°ñ9Ÿ_ùiUÙ(1“ÒF jýQçÉm÷q=l@Š2¢¬DL÷BÔ@Yî1¤ éÔbK‘¿„VzÒÚ©›í“˜8{¤b¶B^Î&Ê`½HDÿªÐÔG\y5ï瀙nÙ3ÖþH†ÕùV¶‚:6=º?0>xÛ×€$‚*¨vè«ÚƒÃ“…0u^?±ŒC¥âQÿyÁGö~®bÈž2“ü@|ý3¿Ûeâ÷ Ó8LÝΤcVyy´é3BJü©¤x²ó)ýi¢¢5<¦ý³Ì*ËU\–±vévɽjƒY{"Gð*Dô{,™棡† ®DAþö óÙè»I;†‹b“àýD;üu‰ÅÖRl´°jÚþG]hô¡n*±ŠÆ8í§å†ÌærÚ»S„—xãÝÖ=+ 0/5Þ¯ò ××Zh-޽O±-nîþîEÈiüÌ›šIOf+ºÉóúÐ'™Ú¶ÛÇú(13þá3ZÝ>Cø‹£^B¬Wl/óÏû¢O "rH¢èÂæ¼¶ˆA¬X™í½øKHQ‚Õ?6D#@Y8\ƒxó€Ï0­–Ò0!ÀOMÓZLq-µ©Ð3Š[ärUi­þG%·µÞWÿDr͹ gÑ:‡Gû­?¬má8!é‚9Œ GT d|Ðu {‡ÐÔ¹GŒ±F˜™ºÖìAÉ¥ŽNÚµÚàïù‚&¢æRAÔñ‡’/1X¶|ÀÂ@…ë\Û¹ÝDäpÕ>Ö`ñìa.ÑAAu–h$SÑTH9¬[ u±]Ú¨¥`9»ìnÍ©'5z/©.ët½`"°Ð?issFJºׇ†»óZ­ :Q¤˜ü#²¨aÑÅ×wŽñ“jù0#ʨµæ%ðì󓨩X™Ÿ ø™‚ulè‡MRúЬ øMñ¼¦”,mË²Š±÷•Qˆ¹òSç&v×Ô^VÀ%N‚ÃÉZQ@/ÒmY·>»Ž"˜j¤ß U™ÔÈ=ÒÔ ‹ÊžÅË@¥ªÂ°)džèF m>‰[ÄØ°P‘è›±?‚üªCò…m|¾ˆß¶gɹÚ?.{b}\~pUŽ=bª¥ùªáåüjÞqWÕŠÙê8›Mw’2Ÿ%ççL;ìbø£ª¢µ‚e>E z\Ë'Û @ jÏ™øŸÏé°¢µ\Ÿ#ì½åªXW{§îg‹Â@¿’L° ´üy4¯&aÎS|ÔeŒMÿ ³Q/µï%\Í:tÌuÈ1ÉÙÐðØ DìMÝWaÓ¡¤î•U¾É#cäaijÐàÌ”F¦ê´‚ +ðýòEfœeJˆ½¤gÈ„ Á@pòù°ûuÎzíE^ÚCu뺊*땜ËA"ɰ†Ì¾VÌK¤dj »ð¸VpnªÌÄ„]ž$óèæ¿Åƒ…Joä(=µÎ¥ÊÚ£Ó|‹†óÜxÎT'd…ËŒšKcÙëÓ†#ög*zz7¸}˜Tæ»÷~åuø¹v&¦6®JJfzyu€é¡ýdXž©G¤ç-[Ff˜^ø!Wz<Îß$œ¦ãAªé³]CËpÅçt¹‹ŽYÝ’‡ŽP‚ª¡µ¡ç¾…üJ#¥ï8£¦«:Þ)¦5öD9ršYžÉÌøØÚX¸”ÈŠ-ö“…j$7—ùÊ¡iÔug%=ghl/Í´ïVä~àº%mçQû=£qá¡âš³êÓçlû¨µGѽºz¥d‘Œú"“ÐtýÖ>_oÇîÃYéìóÞ þðƒv`ç†ãljXæ…3FøÖ)Å×*Çþ…Gîa¯ )d¿+¡ärø³.wÚª„³Ðßý.^œK «áßÙíÚfï0'ý-©Óàòû® M]SA\~‘¾ö§æsµúpÒ9}gÒülÜfú×*hÐòG›áœÐééË0Íy}7±fðvŠ¡5â¢çi/Ÿd¡ÊUOîžü´ùᘜo%Õn—‰z§^ÅîcÍ¢^Ò4ðIЧ^Až¤ì?&––ºQtùöüÁbÜ;Ö®+ç3Övz0j3ad@Ô$ËU4Èì•:ÞcöÇKx½k¥—7gõ!«ûŸ:»H“Æ Q<‚ÐFæ¤p=NØð£i"k ,Õ]{‘6GUů0—sñÜ-bí%AÛŽvÍðxÙ'TÇXžn{­H̲.ÙÚš\qœtÈ j†i¤9áé![Ú"í®Ö¢5Żŵ¢ÍdzmúxàÝg—i#ÎÚd¥ 4®YܦÒô“´ ßIæøai ^L`s }ýTÝ\¢r½Èx<š{‚‘ ;ÏåË¦í ®c\5§DE5ŸV ÚÀG4I•!·>ἦàêX€K5Ð|ïñš Ê3lù³xÃC^‚¼‘ UÕ9ºõ_¬(É" ¼ˆ^|•Ã+ÌþbõŠlBŠ +$K¨Õý‹L ÙFVôç§_¾¾dhjK<Ã~OÇczM™žw<ÇXTÌãõîLSW)‚mÎLduQæZ¥Îi):`gá5å=ÿþÈ©3Ç”òxÀ7,{û¼.Á¿i @xŒÍC5³Vü˜—¶¤KjUh·³!”®]Õœ9UûàÛ ¸™Ed¥ÝéLvÞ?ÿè V%ð©p”Hð8äû÷iß/§ù¼þZé˜\2¬íq¯zÆŒ:³^ë-ç1æ«.ÙÏ’Aº¿ã-bÚà1‹ˆÖ y¹(¼:{¬mY¤øqÕ Ñæä¾áÉ‚ ”¸mÒµ¡,½ztr1Oú8[Y™Æ'N·4ëY†Ë¥ÄDÖ}éu^/Ñ5„¶d%È  £3ß(ì66¯äü虹2âKÒIÁ˜uÏ:Ú¦pJÎj>D2NW³¯`ãí¿Y÷ì½ã.’haÓó,i9#  ‹½iê=Ȱ Øag!oôÃ-Ùo÷¥ç+‹Ò‘xÛô˜é3‰©ÈN„; ‘Ehy Y¦l¾8ð"áa¯$úêÆñtt6Ïuü‰hÇþ±¦ á°Z„OD°¶HžãÏšÉ>Ñ-¿·ÒT ‘ v¾’Q*°íûSÜŽ¡º÷g•ʼíÏð¸Ó]ÝûØ?Ú8#40ÉÁ\žf¦¼A‹è¦¾Ü1›¤uûr ‡ ˜õ2«»ˆs鈲×µ7ð¶‚8r£ÏÙY¸yMÜ2·åRËéÔ’¦ìfa²a võ?\¢r_ ~”é~B0É,‹GÖ¥†^˜”>Ñ&,CT4qšîo©þ½ ãqÒYƒ íÕÕ„y\’Ò¨9Óµˆxúxϲ †Gyâ™T°OF"ÝÄ •ªIZeÒçá9Ä`®uÛ™M˜.·+M¿#:…ÞïËd§k‹:ü†ú¶WÿFíPX–‡Lá+O9Úê)j¡Èó+llt>JÒ›§/ %êzz´†ZGd…9_ Ã?…ï é®ðØÑÏð,gñÕ©HîtþŒ5ùÔ™ÆË Ì7Û¨<Î ëðîeEoñ©ñx«Ï¤ø1CÔõ._T^d*uÀâ^™«Ó鄟ÊdfldE¯êyRéšMószR¹ÿqZsëyäx+à¶òyEzËþýóo(Y#b ”,‡ï€ò2È÷5Ï`ãk E›oJR¡Ó» ú^÷¹(¡@†ƒ4°ŽâήßFý%ö©Eï‹~"ڹʀÅÈN‘.#÷œ:‰wÙÍŸEÚ%WÕŽ ²–DUøc½íaÛR†רjП­ \'¤C{‚‘úi¾Kú­ÂãðС† Žç$–ßRývÇCk ¯ e>ú)¦µÖïí¼=ÄBéU´Pzf*_úÕ;ZGWµéaµœï=œhÀV~-8RAÛzo®lDF«îkh'ŽËò°G9³éU:º'°[¤ûðQŸôXb„++e—Ö›Ìt>¤×¹BW¬ÑdKb”ÿ µVÖs`Òz­7üsž÷ŽÉu÷î4Ñ…Äî}¢€×#XÃÛ(ոؼ' yá^ŠLK¡h¿âœm3I§T‹' _gèXFNb¸–Àìe{~ÊûÏwK°U 7D}ëdrKI"hê¢Ê5U³Of¦ŠfuÏ3^ðúxøb’ÞŽ4¨(ܪ¡Ä¾ùV§»æSã•qx=Ê"LjÜò&%< xaKàe»Åy‰QhïÏN§~qê'q"qW“&ÿ4Q1ÑÁ9Ç#Q0!Ùä'Á‚¥×»¶Ä–k ɸ“ ÃB€'_xmzÁÖåh•"ï+FÂõˆì‹‚KEÝ ”·¤@ ιÍü‚Ì ýpŒ@Ê{ÎIœ¨ŠšÉöÊÍtŠT£ö®¢¾~â¿s>‡>EÑGÀÅ@¼Þ´“R&Mƒ JϹôˆA5?óã e¬~*°çÒn • Ǩʥt‰$¯³2µ›`âæ–L(‰™Å¦ÈŠ_Ã6[•dH¿ý)ìz6Ð+¤ÚM.ÖRJÆ)Œ…­ú\ÕÆ\ØWÃ>™ŠùVNÖÄ cnïýÐ2øãO‡Œ!‡Œ|²dV]Aõ g5¾7À ÿX( q´4ëSÅ].pÔ«mspP©Ó­ßÀ==¡y kœ5º¾ô9/#Ò"åHÄ{¦ALÏÒÊ,pÝ*)ü¹I!OF„oô#æP¦?ƒÙK·Y÷L¸µåj¾µŸi²ôù;z/4.Êmg¬WŠÃ°-ãd®èæ_îi¿~ð ¶>áöy³Å#I&”¢'ê‚wÈÊ’&­á{;ðä~xúî‹m3|æC*“L´þd\]þ:¢ì´þí¢"Æénk$öðsÌFTzRÏæMVª<Îq a,ñµUs'}–”{,Y³-  ž·w‹óH÷¢- ¯h«¿m7;%¦®Dõ”æ Ú÷b2ßÒ ï"ù½Çnà¥blBÏr¹"&ÙliqÁpV“¹Í©r¹â%Td¹öOÝôhÑX­Ÿgì³BS‹¢«:9S• ßÜæ{«kxˆ[”¼¥ìr<%RM½f R W37Þí²–S¤Û½Ù]ÁžÌX¾©hfñdã€NËèå£ïrÜŒišÂ‰YCD%D`Q¡k¢ŽÛóJH½hÙ¬~¢ŸFŠjZäÄ\ƒõüâVvÈ¡N{LZ‰¸ñSž¸kK*Yx/ØúI+Qå³SÔï#"{#¸Ò“í»…í$ äŒëž›èÀƒÇx L6¥X'/,30ƒÀ›žê€áåza†¥¢º}ûÎBƒÏ#kß,¥ÁË=‡Ë8eUJ%4b<‚Åò ”‚|u Mß>OÆXW‘²)3aè©ëoÕco ×ñ.^°Q5.¢F ȉÇÄ2÷ÂK ­RíS¿e±ìe¾ŽjÔßyV¹ÍÂVwÒÚÕ†Àt4Å/ľËÍ@Àó…KJû¯*†ƒ¶r{*]]ý’Cœ Râ¼*;‹¹%~: 2Z|SŸÉxÐà`jaáüÜÚÉ%E ìÁÖLT©³ei„½ÎÙ…T‚0²Å̆|^Ÿs5– ²u UÑÚúLЉ…Ú^ø˜¼KK_£µ³ØgbÖ>-úàGŽMíq¤E¾ †0‘3ƒJCÆJvÑMý…ZEÃñ¬§Œ”Î{ÓBQcÚ ©8K?jÆ)ÁMÍ¥ÀïAů»êMúZË.Éì¯E9—j¨D‚o’¯ ›«‘ç$Åò€†¥©%cm¨ÕHš\¢R«ãc›»Lm¤îxžìèBÁüóDþ`© çöW©S'ª½µêY× _k^Úc²Üàû+6•òò}c)~@.|g5„Óà5(ñWîö¯ˆ¯:ó¦Ú¬ùö‚$5EèøÓè‚ÑØ¬7v\’’©áë˜{ƒ¡¨EËyfl)p3¦y”‡8lîq£ûÜž‰¢üžõŸëïºYR<äjß|Ãä«>P ¨‹€ŒWTbŸ–±>,Ž í*&:`Á’óª÷`ÎÐXÊ0ú ®ã‰ÓO¿¥‹dÚGpa²$˜={û9Ív35¶öû%«‰óñC>ÜÒ›£<þ?š3¾<=lw}i‘‰lSèóÓ_ˆÒ²†±};TÈ Ã]œ§}))Œ{¬ÿC—pîó[*¯À@By•©QgI×§ÿÁÜÞr endstream endobj 72 0 obj << /Type /FontDescriptor /FontName /MRSWRE+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/J/T/a/comma/e/five/h/l/n/one/r/two/u/y/zero) /FontFile 71 0 R >> endobj 73 0 obj << /Length1 1592 /Length2 8653 /Length3 0 /Length 9693 /Filter /FlateDecode >> stream xÚ¶P\Û-J‚» Ö¸»»»» 4ÐH7îNp î$@àîNÐà\4¸‡Grν÷ÜûÕ{µ«zï1m­1ט«š†B]‹EÂj’…B\X8XÙR*š|vv.VvvNm°‹èo3 .ÈÉ …þ#@Ê ty¶I]žãT €¢«€ƒ ÀÁ+ÈÁ'ÈÎàdgøW ÔI t[TXŠPÈ…F êàé¶²vy^æ_Ÿzs‡€óŸt€„=È l„T€.Ö ûçÍv-¨9äâù_%è…­]\ÙØÜÝÝYöάP'+Qf€;ØÅ  r9¹,¿ Tö ¿˜±¢Ð´­ÁÎÙµ –.î@'àÙ`6AœŸ3\! 'Àóâ-e€šòW°ò_Ì€¿{à`åøw¹¿³Cþ$ÍÍ¡ö@ˆ'b°Ûj²Ê¬..Ì Äâw ÐÎúœt‚í€fÏvÈJh€Ïÿ¦çlîvpqfuÛý¦Èö»Ìs—e RP{{ÄÅå÷þ¤ÁN óç¶{²ýu²¶¨;Äûo` †XXþ&aáêÀ¦;º‚¤ÿy6¡üÇfrð°³³ó ð@އ¹5ÛïòÚž ?NŽßæg¾ÞP€å3 /ØôüBñvº.N® _ï:þ¡pp,Àæ.3‚òŸêÏfå_øùðÀ€7ìÏÚã°ÿ~þýeü,/ (ÄÎó?áΗMCAQBJé/ÆÿöIJB=Þ,\\Nv€7€[àûßUþÍÿ_ÜÿXÕà¿÷ÆþŸ‚ K(@à/ Ͻû ·¿UAÿ÷Ä0þ{Uè³”Aúÿ(߈‡Ýüù‡ãÿYÿRþÿdÿ»ÊÿMùÿ»!YW;»?nú?þÿh¶óü;àYÉ®.ÏS¡}ž Èÿ†êþšdØÕþ½ .Àçé€XÙý»`gY°ÈBìbný—„þu ÏåíÀ:Ôüû®°p°³ÿïyÞÌmŸïçç³úã=Ó/)1‡Züž;N^ÐÉ è‰Âþ,/N€7Çó€Z€<þ(ÀÆ º<§žéù,¡N(¿O”—À&õÛôñsØ´þž}À£ß»c³øä°þylà@~›Ý? €Íþ?ã¹äð¹ôÀæðÈ `sú|^ÈùÀæòÈù\Ùãü¯6™»:9=_„üÜÃá?wä2GYœƒš ½³©y×vS%Aìβ3.2M³£—ÆÀâ½èÔîz‡˜ÌP™´æt%‘ü¥kyK†þR|‰üÑû ¹1¬%Q£õÞçám¼æäN+ÊÂWü‰Â‰Ú~Rdmñ]ŸGGÝ@[ØæŠ4]ù1ÔóqoÜûä”N±ÄèDÏÐä™eÏR"¸°"1âüôÀœ¹¼šÆÉx"WŒgBñ=Œáúèm¸Î{;ëµR®ÍéÜEDMdHH {‰3:Ië-ù=E‘`Þ»¤( «‰u„3ÿQ1Ûø0eA‘‚Þd' ×Ûf\©Qúðj=’ÖÚ>²{à‹€ôÓK¶Mª)jW/Ä@Σ[¤Å^ƒšô›Õü%–áU§Ïͬë‘âdq>Ø,}”½cµ92ö•S* ½ ®aNiœ/³«‡ÖBawÇO«|™þ§^þzIlŒý5õi±d¹¥üXðßÔÓù:0W2»wSEM£Þ"µµáÄaØ ßDÏÉj«lÞOè~,Q‘¹Aµ=?éi(מd~M¡]¥¿‘zíkeë(-CpˆŽé]ñÈ&á~#zªrt‹—ÒC»¥7»ÉøýŠÑcÓzªm;¬žTÁ=Å}8A8Gšî§—.Áw·ŒÁÉÏ d¬!Я+á:5Æ*²½ Ìkûƒ-zÑÁ§ÌÚµ³äfƒH¦«¬ÃÚ–ļwØôw$ š0â ¥s¿˜ß‘tiE'^Ô1åZÇ—4)…ÚËd3ø“LoÔb¾.ïy+ƒˆÛïQ)&5b•±:<›ª;ÞËŽPwÞ‚µ·ª®lË8^òÑzŸžtÑå›Om'nÙ¬ YÀÈÜš™ z²8¡U04m&ÝãÔ ›¾ÕCt>³áå¿l»ÙµÚ#kW?¤…:OD5ùžEŠx}2&¯€]™µ;7@}ü˜ï)7t=y‹)â¶È"ú]gðÇO¼ 1Ê+ÝxÌXñõáHl{27ʇÓwðÏy2yôâË N|Õî0ौ…åIXTd¾3aö¬ô*.Ö‹·§I²F ^Ò–= &y€{œ†Œÿ:š`û÷lŸ×‚]&˯.e½Þ*¸^° ¿ß]ð¯x N}¯W)ÛnªP5qÑO|Ë‹=5îìN1‘$Oˆ£Æ( 2b[g¯òÙ0«yL‡5^.§i»†7-u(õjh$ûÒë+è|„yïàGP‚0—Üýéq`ªÚzå°gൟްnT5 %ØäÚZ¼V×Ý Ê"×*yó½€£ƒlvKB¡Cé#®ÇiÕ=ÜjY«î%ìÈ Xsã¤~à—_ÁÈè6l´À1‚Qqã§Œ%T‡ñýòxEž²*+/X¾¸@+9ùŠð )úÀØÅéû¥ºV–´šãáº{µ³4=%ÑÁ0ƯÃèÂhvnrݸ4¤ûá £±Ý¿ú¤D¨R›eb6šÞÞ-Pý¢É51ÂYÛàÁZ¡`ÝŸÌ>ÄØb¥Gø37nÅ×ËÒË÷Q"¿ š¯r\›õÇE4¡ç3¦Åˆ{«ÙŸuog_4¥Æeõj0ã[Ìy¤³ÙS B`Ž ï@¿sßOnõt°.Mr2ˆëUЩê”m…Ð Gž/%„0±‹È0”22¨§.¾)°±«cœÓZ9¿toË2HÀO°ÓЛñd|12a8Á÷™Ê°~¦§ÊŠ÷FíP?)%룿ô–¬©¸u{‡aÙYõ SÕ¸Å* ¸ò@l¦CމŒG4?‰·`¶þ •Ô&@:¡± ŽR¼à§Íy1 4IgiA˜|Ö|û°°þVŸ´ƒMø¦Ý£aSÿI*#·Iå8L_y‹²KôSÚJ,›,xt¨QÅç­ÁdcaR;ù­9”`µ3ÄÆ¬Es²uÂôCj帯1a9v»HÂ.-ºENÏܼŸó´Q}%Η6ø€3b­¹££ºSšñÅj©™Þ•3_å®÷Zo:*2Ö¾'–pÓ+÷‘'íWD¥âæ²—Sð±çÚ¥ÚèN„·LÕ€î™=†»Uùü{¿é¦yüÀ\iJŠ ™ rZQŬ¶ÚÂI„7´Ô§“OľLßj²û¾YáËœÎ#ñ¯Ü0j®s÷%Ÿ»sMjØÐÙà”Ã…›öRØ#Ws¼4N^¬X6 A½˜̰{•Î÷‘÷-§Šˆ­’?µÏ×y:–Ö±ÅÒcQ˜,Íj¨ ?6òO0…<Óvc'+\‹puNÇk”öž‹pM‘^…¯<¥ošuì9·K®Ý¢Ùnãßè§F†þHWpÓHLÇaÉ7©ßcíbhK1²RÏ*,Oj<õd8¬Ç"@n`D {ÓaÕIó5ñ[eÄ`É-Ònü[…—2nÚ¸à½s±yû‰ ÂNªH¿'ŠŒá“ÓúçÑmF)P~Æ[ªQWß48eëOi89Ïwø$¸j“×| ˆÚ=k¦ùŸÆ|¸°u܆1xŵ´ßWõH‚¢ìÒ¿ù‘n8óÛ5vÂ|eInæÂÄk.q¶ŸƦ?º}Ó S :Z&H½ý…êɺÓýkÆòµnÁìk„¦uŽbØÀ üEš ²\½x‡pov‹Yjñ ÊXAÆ~µ÷ì?AK%‰A=ݪÙĶëeû¡ú¹Hv-WžbÙë%#‹*¥xDn,[K&•  S,Ĉ°©bbuºh1vero(³Ç¤„¨»üóJJ]àéë5¦>â $Lñ4„®j—wYNÜ6]d˜J‡ (´ÂI"Mç_¿a“Š¿¶ÁQ£¦È„âÖWvÁ9z3ÿZɱ+@õ¬Æ—þêÞÂOAÝÚ•oÉ9µójÉÐÌSˆ‚ÆSaPoS³)s`†+ê:øíÕì\Ç&g²èècðÏ%ÞÚìÌíË’—ØÞ£Aì!—­ÙÆ´µD)IºïüÊǧº ,–‰Tå“ÝÐÓÜDé'*tðhòp˜H’ºwzÞôwŽðGIDZì@t8är•QO#'†>É.?ªŠí~¡3#} ÛðfÖ ýÖÜ][hÌËfL?¼lÅ(‡O~üªñ;$ª!ÜÒö3ͶB§Nܰ1nœ„ÕÂÙ’p[éÐÚØèP¿\ž>cšO¶Xq 7©´µÑ$‰sžPAPé›ÞÍ8þh<×Çkrh†¿ú®_Uâ)pK•õbò»}qM¹ŒèmMñ©r@o.K%=Œ?äòj,64Û0&Å•jŽÏùÁE©·€ v(õv€AöˆAžì>%G|1Ë£ê]2MÚ—~wÛ1Ò½ÌWŸÖp’ŠÃ’Ìá>W4 \À"IÔ‰öͦ¢h¾ÊÎðh*¥ÕªÒUÝr¹'ðn ã;<¼¢<^*ÞФðk95n‹Š¶Z þ/Ò$;FêàîÆ2®MÁTõ†Ô!Õ<¾Ôë—œVe¥ýš5<ÕvŽ)¯[Èû.£OÞ‰ÍÎÌÏl¼ô†NmmþBl^¶£E^H ùêºÿÂá­XN·¹›~3ò˜ŸMxõùÇAá¢ð9\fÒãSŸô QÉŠ`¬#-™`÷µ ‘oðÔ6O«:Çtìž¿¶¥á{óìblLóŒIñ1ô=´PF[¯íºá^<Š‚¯ð‡éUEèÇ^¸gyÃDsL_‹ Š‚]ÕÁå~<ûÉÊqÔ>Ý·\÷›)ž"kCíïófÜ eÔ`Z"¾Xm¼P üiýã\’só3›£@ƒmÝzÍ!ˆ˜Þ½šQLÃ<´Lº)Hw¾Ù¦…NÀ~¡šU2‰ %Vß¼®ê‡H/CUo!+,< ¿ñýÒ§ •OÛuØÒL¯$ÒÂ|«ø©?0²øè[ä!ºzkÒðÌu劜ԇ>|»‚$>ff;vݰ”õ2l,úùúÇ´;<çcжh›]ÁÊç£ðóÒÍ€±r#uøåo/ñt‚áë¹ u±*0h©0Ùô€S] Êz8ä…çÌT³‘ÊšëlEÅa(6ñ¬+õ'ïîæPz3ß{¬mz”iÅØ»cŽÛ•lÑú3ÏbY××C“úD=ëŒ oÀ°ÊØÌëX ŸÐ0Çlƒ%Tz«ûá4l±CÐü”S¿<ìJÄæ6Š\)º]2¤»j§¨ßÎz4Çjy± lS»÷§ÊÄ”Fâl`ØÉÁ Üä§§úmZÇPFí%Ôtê~V ~3«£#tYªÎ˼Õ\Iä¶·>ÜÈ„HÞá@éªD±öa]fàš)䛞äIQ‰äoÝäŠf§àoi]—òЕàÿÌôÔ‘ggèfp-CÒ¦m«ÆN‘¬L/lœËŽÇ­·‰i”Õ4\âˆ4 P¯Ò¾AêëàH² ×]I¬Ô¹4“T¨ hü6©~ôG!½Šý*Ùå²u<¼ÙCz§­Ökæ¶—Ì›HJ« ÐÉ«¡Ð#…ì‘÷Co<É Dê¤o­ ÿmOÒI­Èµfý#¢=A[î:VÍyGä<¨¢Üìsð´on¤U6ô½‚óã®è‘ky5à…2¬dDoN] EÕm-»ÐÎÅó…¤ ¡ê‹÷â¶v©â!#l%ßð³¼…s¦wו×”/#ßœ­÷ùjo¶ëÎd‡çf›õz¸„ÅÞ}_é—÷U!JÐùÄlÎÀ¾ΔÛ/ýj bí‚(/ÃïÕzeú™p xìËŽ‡´i£†»e‰CJ»lé·¦ÛR*zÅq:X߯="Ç’d¦§ G–OjlPû"·ˆsL›ˆºˆàWМÓöKÇ /ßr½›Ž”ʬÇð܉#ҘlDÐ2Œ.ðÊÛLèÆô“åÑלÝåÛɆ‘3õèÚ >Œ±ðe4%u2È|É&ye*TžÙxcj¿Qâ&B tHß¾,–A­W ©‚^ǵ|¤Ö?ŠŠóÛ‡§BQ1ÃÕz'q|%KÈÏz³MËa.J•I@¹,)‚P}ñk`sјàq>Lp^ŽrW'혴y€ð R€[:£q¨h†5žñÙD;sº­×Æòì+hÕ€ú²¸ŽãX™ ` “ߘi—$QZÃr51–“=5¼Gä‘wO ßñ𾢇0’¡°~?÷†^ BI§„}[-ðžsF:ÁÁô $ÆxÒåöჩºË)¿ÎÏUÉsüÐj^ÌÜÙ;7Lû²ÕšÛ» k‹ ñÕÒ+ë+C@Ð ª¶6mø‡ÔO‚Ñ¡Åïâ¼Ó·¢$Ú´ìõjwSŒy(£dÜw©îèI–¬Oó²ú´»ì´±õZÒÛI#üRï¸%ÃO‡>1¥yÚX -1ùåT:Rh¼:qqäRoír=E½T\ñÛ¯L0§N±~:~{K§%"#DfЙ²ƒfó ü/>ó{ªÃwÝbR;lãZ· ¸$Rƒ¯tfø2‚ó±ìTfVÃŒå㨛í¥«¶’Pr¤ñ@™m¼&¼c1Ò„·ÒäJ…©ŸAu:-Pª—²%ˆ£jˆG'_ßð½¼³¶z±™öð³F Ý;и5ic ÇT`¨Aâ‚\Yœ¸ŒŠuüÝ!p¿%?Ëœ¢øE'\ÞOöÒyšTþxüì42‡ó‡Ê£GX-E\GÂ3û$½ª´,qªÉ…ø_†!^êx¥Hp¡¹;Æ uÑ&°–£˜3ì>å“wÍÁ1‹Þ&ª&L„kû7ód1&(³cáS4¡ÂšÆÅ–òâGΊcR}WäpÜ¿ýJtÍ»•™q½T˜$[˜[zE®ú å$ôh<¡ 5´ä]O!˜=áByÔåÛ›þ×:˜§Œ]NÏn¤¿a¦c?‡àŸ¿•ÉpUøf» ÌÂ7(2N{²Ö ßûªÊåi÷e¥"Q×SL«UÉ7aXß?Iu¥½\¾2 EM6QB¾XïÓÊQJÇF!–/]*ÚT'íQéÕ‹Ò°wÃj+žÖ»£ Ê—ƒŒeâŒØhÚŸÌXÓ~ÙU‘5‚MÖ˜¡ï×Â̬=¥EýÙ5˜â|òÙÄ™É0b°ï‚½9µ™Þ‹ŽléZóÜÇÚº—igÄ”'–MÅ{ËsÄ`­51‘ïqýUÑQGOqԔʽÎpói¾Ÿ4uAZn eü‰Áa-ß« 3S-€Ùg\˜¸$VÕ6O|%‰>®Ã¥{÷î @”§¶3–ÍÈÎòååçâ†ì{#×wA_Ãvoãé*ÀÍVF+Ðû- ›àÃÑ:‹7LŠ’>I¯ÿ~lMš•k+æÓž_é©^í£>Y‚r\—ëàÇàÈ#v¤šA3Ò¯â•F[Õ›t\ÆãäÇÂÅš’ÏO¡µv×ÂqO=Ba¹˜?¨uŒ¼3LjQ;ܧØüç·{c4µ.Ê>ßèN_ÃÂw댙DZà)ÏÝàÄU ³…ŒT£·µY¨R§/†ÃÎD=© ¾# j>H&0»VqÝL|P$ÒOurº·¾ßåÌß¾Ú{ÛrYX¢l±2¤”ˆ”ÜÚÊv¶þù"| |І‰ŠrÊVM~:h›+µKב¯;|š¬ìÝCLð “?D†­Œ€õ:KN_S‰U”·•ýØÞ\0Z·ÅÇqÁåóîô¡ŽS ;'¸RÔ5ª‹ è¹z"7¯7ã wqqÓÓçj×ueÓ&ó³ˆÂ«±*LÑ wâ0$¾·]šÉ"AÍøøc eö ì¶ãtøY¡…¹O¼Ý­¶!‰yp^YXçë,¯ú©¬å¶¢Ýô åËošãÚ·rÎb*N¦Y‰ ¼¼ ï¨0›¢R@½J!”™¢†&œ øp,z/ÃF£±õ8©Ýg¯å°4é6h‘ÔYÏöLÕJ/R?+™Ð„f5J(ŸVM¥æq iŠ«“op¢3K¦DPm.?#r1¤ÔÕhÍŒrl¯ŒÛý•¯ºõ:æ0Š·¿ E>^Ý0LŽ*™1wz;Á¬õ^Pί(:ωàÁÉ*É•#¬JN¥1ÙŽeéúٖì<ª Z’„t ‹ŸŸµ?ª‹V-XÅùlÊ~ò$X Ë*^Î}±´U䟥èñÈfdyÝy !wí+D.ê³üéô+EDHßYxci ù o™øÙšªÍt©ûb—ü/}|ÜG„Û”Nzö´º9û˜Dá2ç÷GTlY|ÅùößÏÂ%òNøzQÛÈÚ›šMºûø¿o/h¼’Ûñá§|ìÃ%Ö¥^LݘÀÚ„;4)iMUjšEÝSþeá%VZEõkQßRô6sœÛKX5[|Ž2Rûød°7]éûj¨À=_‘\ )»PY´ƒוªø¥ÑóÇ\Bn÷›46½7ä æP ­r€&À„h!AR?) ”Ù×1öpuWø‚îå’š¿CNç¦a(ë†ßQðþ“̲v¹”ªúg2 ‡ô¹ÆW»æÇÌñH·»fÐÖ É t)¦5™È±* f™À‰7þ›.Ä!õ8íSðœ< Ì¢÷¾4«ò¤qÐg&{ׯoIݼ†ó>E² žS§3Ï:t/Ñ÷><íYÍô‚á¾±!SçVçÖ\5V19s˜…_–lë#JdôàÕµzêøÄ)«~ØÃÞÂs¤( ÙÆ BÈ Ê AöŽIˆ êûð¶ÿn”d¥Œ†WÏë\…pp’ð,²°QúGåå@ ì°.ŠßÝ¡)2ï÷>Ã=CŽ”§»aV÷V¿ðçÑ´Qc{Àöƒ¿ ÇéÞo¬Ûi´ÞW"hq;ï•Z¥|Q£1öÇ?WÆBêMk™ª{(Ô»XÌþkÒ¶óÅ*ó¡XùuÃ/çúVµÕòO€ÞEÖŸwˆœ‘úIOáUŠw ú°a–„9kÉ´9Άbµ3±]ðÒev›b”ž¨Ã]Ûæ1 Ç0˜"õM/Ðvò”ŽF gÑ-ÅoX{­Dñ ¿4’èJ~j™3¨¯[é‰Èj¼¾ŸíÓZ©…6Êñâ2ŠØÔ²³¯l}•>‰ŸD‹´ùcMuë‘ëH½Ê)~)Û*ÍbkV#n½¼ç0礇ø#*¾xyCjæ×k&xS„òª€yrV»G5ôNÝÞñk¦Ò JZv\Ý©ø®#na∙µØC¢:Uœ¸fqc*´ÌäOµµà‰Ëfb.Éo()í×±Šî†ÔEE÷Ǧ\£:ŸòGÖpî‡>òT"~l=)þ’‚Å}žÕ1щ+°bP½ò#€zãÒÎË0…‚ëM**ïÕ×âTÝô¥lg7‹¶¢hgë_ñ‰0éÈs-úïÝÎÛ¢¿rxæ.BZbsá†1Ô¬=ÑÈdéh1 ²gRð]ZåV«Pt}QÇGǾÖ{½kV5ßyÃõ.MÚ+Z'Ç\ ¤¨â]¼ÆæoBö-Æ›ŠO(­Trîå©ö¥Ë 2«ë;£'(e­av¦?>`Pëó/?ìvpâe÷äÎßÝ Ð$Öuİš;ì!#.ˆoÃúè)N¾"âÝËrc"²¸0¸&%Hö詧 endstream endobj 74 0 obj << /Type /FontDescriptor /FontName /QIJACW+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/C/S/a/d/e/i/l/m/n/o/p/r/s/t/x) /FontFile 73 0 R >> endobj 75 0 obj << /Length1 2479 /Length2 17206 /Length3 0 /Length 18665 /Filter /FlateDecode >> stream xÚŒ÷Túú ÓHwI Hw7RÒ)Ý0ÀCww‡t7(Ò!- Ý Ò%Ý !wö>ûl=ÿï[ëÞÅZ0ÏÛý¨)TÔ˜ÅÌìM@Röv.Ìì,l Euuv6' 25µ:ØÅô_:2µ&ÈÉlo'ð‡„„è¡½º@íír®6vN;;¯€ƒÿ¿‚öN€w@7°@‘ gorF¦–°wðt[Xº@üü÷#€Î”ÀÎÏÏËô·:@Ìä6Ú.– [ˆGS  @ÍÞ rñütB–..¬¬îîî,@[g{'‹·ôLw°‹%@ä rr™þJ ´ý“ 25@Ýìü†š½¹‹;Ð €lÀ¦ ;gˆŠ«È ñP“U(;€ìþ#¬ð&À?Ű³°ÿkîí¿ íþVššÚÛ:í<Ávs°  ,¥ÀââáÂÚ™ý%´q¶‡èÝ€`  DàïÐ)±÷ $Ãòs6u;¸8³8ƒmþÊ‘õ/32KÚ™IØÛÚ‚ì\œ‘ÿŠïØ d ©»'ë?͵¶³w·óþ/2Û™™ÿ•†™««†ØÑ$ûî ù7Íäàfccããä€ SKÖ¿¨{:€þf²ÿE†äàëí`ï0‡¤ò›ƒ ½n €‹“+È×ûOÆÿ"dvv€ØÔ`²Û!ÿ¶!ƒÌÿƒ!ýw{ôØ ãÇ`ûëçßO 3³·³ñü-þw‹Yu$u”U%ÿIù_¦¸¸½À›™ ÀÌÁÉ àæäððó|ÿ×Ê¿ùÿ7÷¿©*@ð?±ýaOÖÎÜÀÿŸ µûonÿÌÝ?KCø_Jöiè~¿>7›)äûÿçø[åÿßäÿeåÿuøÿoDR®66óéþ#ðÿÃÚ‚m<ÿ‘€L³« d3í!ûa÷Eµ@ÿYgEØÕöÿre]€ ³³°ù·`g)°ÈLìbjù÷Äü· ë6`;н3ø¯{`fgcû?<ÈÊ™ZCnŠ3¤Y³@ú_’v¦öf­7èäôDfƒÌ77À›²£f ¿GÀÊbgïQ@²ó˜Û;!ÿÕRn«Ø_¤ÿ ^«ÄoÄ`}÷ñX%ÿE¼lV©ßˆÀ*ûñX~#ˆMÅßbSé_ı¢òqXU#ˆMµßˆ ÀªþA¢ÖøñC<ÿEì÷@gS0¤ª6f é\‘!;[ÿV…„cò‡*Ä‹ Ðé7›‚€¦Ö È«aîò›Îù/ý?þ/R$Ó7ĺ©½ ¤·ÿÁõÅÖöp!Mg5ûrB‚7³·±ù#vÈ5`ý¤Í¬ ÿqÊóßѲ<¿U ÉšÿV„kvûÃÆ_l{×?}@D,~[„ð-þzÞ@Š@b·ü ¤–ž– »?$ 4ðÒë? ¤¿Cä¤mó×dÿæCjgûBëoÛÜ[vø](ˆ/;W[“¿n‘Å1@N.«ýï(!6íÿÐbg‡dæð› ñáyíþ§»\ìÿPÿ··\Â;@ÆÇþw·¸ Us°qý# È÷VÇßABJæèjï23ù£9ìêue‡dòÛ÷_äöGY¹!âÎ'â·$ßaAŽ*«‹¥èîB²rq·ÿCR<×? ¤n@HÜÿ؈öÎ8 æ=ÿ€úyýÎbÉ äôWÿs—L] tùû逭ÿâ¿¿€@ SäÅ9{SÁ«Úo·_ĈݙwÆ…§©w´Ò陽Ú\ïÑS諳ƒÖœ®ÅR†º1W¶$é®D—ÈŸ¼šëÃ[’Þ·>ø<%¨Ní´"/Lâ÷O‰Õõ‘"‘0«‹îú<9úhZÃ6CwÈQç;ºò¡«|Ĺuï•ö¨ë+[ ›Ûy¿[Í#òXö9Fãƒ~`ñ uIÎ,!%‚ 3é+ì3Œ™«ëi켉r¹FdßãÎÏÞºë±w³^«êÎDTDº„¤°WØ£S4Þâû©róÞ%EÖ„çsÙèY ÖFûÈ}F:(Nôd± Ølœù¥G©|[ëXµ{P‘²wÔŒjŽÐç ÍÝÝ¥¤?*-ê5Ü Gº‰pb4St‹CTòôÞÝX„»Ò9º’?:©ÐóÖ3ßDBÞÄ[i_}÷îè‹«ˆ@r¹Ë×éSh•UJUãÝc5qì0í±ë¡ÖÖ_.ùÓ±0%§0³‘—z(᪇K©š¿Õ I¿‹^oˆhIJc™Åñ|÷™ë¤`T7OþÁËݤÁÓ&y¢H–Û»‰ÜÊ4áD¯=#Š)‰7¯pn¨L¾Fàö¨Ü²Y•èT«Ô1ô˜bøÞM—QYÁƒÖ ÆÚû#á¦ý·GQp3º9á@ÿ¯MiZ$QJW¸Q2Ü:¬K[Bºµý…û…á}IÐøÄ’cŸõFÆiòcdQº¹ögo®Ñ>l#¤bZ:X›7E(éUfSã‹? _TKu‹X¨fÈ—ÖDŽÅ ©M«}í|Áˆ«ø,|U¶©7JÒ“´hðå_ú¨ît—î‚ç°Ç`õ‡Úðš@-šwz ÓPY4Ø ;-³‚”ü$­¡*¾w¾í k=SÀ«9+£§ñ˜ü¥ò†U»ÝÃÃÛ~÷¸¡iŦiÛ¯³ø»:Ui¸£oWÐXœvp*â̽ÙÞuúðˆ–ïåä²V·[úLL²Ü«ŠÁýú"?k›msm¸B£µ@½:àMÖôñìÉVª÷õêi9iN}ZñÒlÆg¼”¼Ÿ,~Øg^Q|u^±Åeô(ÁØÕø2ÃŒà0íš\AßÚÄÛ\Ü]«>í+Éy×ròoŸ¸Z–ãÅe¿mD©ðKóyÞ·Åp±™BËdÌÌr¹ê£¹ë´õ¤fæîäp|Ós™ü>4œºôéC}¶ˆåeFÔW“ìg†ERªd³þÏÆQT•PÔЬH@LÊ üóRúØR©ÓþÒ£æwå8ï ]n1gÌÒº‹ó…+²²¹Ç~}óg$n,üiÅÛE}}IÃßÄÄ¢'‚žŠQg…štŒx­±2)uPsßÌžì"9ô½iõ£(ÚÔÇv¢žOµR†ª"©Ì9)9ÝEŽø!ˆ†Ù`Ø-Üoö8: ¸X‹òó¾rvˆ–oêáõZzN±“ÐG6±¬gœý Œ–ôÒ¡m,»9äŽ2õ¶³jçØl›E˜3ÅÝwgGy¼Ýç 0É“5ÏŽêòüZ+;ºf>WPºr´sx^ôPbnqLÝáPð >'W&££õ.›­Aw/VÛUeÑò½{9™"Þ¬ã˜ïÜõ%»‘ª Ÿ-±ŸŽiI8Á1„ îJû©[”‡€r´øB›_†­ì=%•ÙÜÈÒ |¸ä^Í.T±›dÌלöZþ”˜eŠONhÌi¨ŽŽœ§â†LóúË—­®ÃŠƒUÕ¬2‰Ïp†ÕÜOüo.ÅN6EĨöVèEkŠé´hw™`ážÅû¤ËP‰˜GÕ ÂRfñuÀp!âLÖ9°%{4˜þãå’”Y®ç­‹:ë÷¥Æ§o±¦ É êéR*§il3îIQäi1ä7§ÑhßæaeK®«(1 (î{G©qJs¨¢‹òèç­‹0ŸÍâkÇÐ,ï_ùKá³8Z¦÷Àßèç]<ØËJÜG•ª4ÆŠ++/73ùk¶på,¾›¸Sl¦¡Ž~sî Vÿž Uù®Ö$hH@Ë‘ÂÀÜ_àÞšdü²_X ‡øv§ÛxÝÒ¡U]=N¶ÜN&d³Ýø«m,ñ"ö¦'ß{õzn+0=àz~@[B汆N' W®ÃÞ¼{]™¯Ñ1UQÂz¬|¯ŽLmLsLâBG(ß'À"q “çÿØá!ÓÇŒã¶)ù8ÃåaÄM±—¿Øò•ÙËt™[l¶iÃMk2b²êÈ”*sÍPNú²Ï×·¤¤dw£Û+}èè:FÖ4P3§˜ Ó~Ùeñp0cÆ Ã‚hâÇÐÛ–Š½ÑÒŽ1ÜÂu¸i@fÿYëk uL—zê-özÒYq‹ÔX€…šÛ0âK„éƒ}‘ …]bø}©˜îa‰½Ã-OZûqAÌ*žBîW°5·ða?M–NÁ1˜lÑýW؈A aN˜›Du´ ÈkK ®æG̱Tð¦¢2Egù³WÁ@Ï'Iz¾ObÙ.ÆŒiãö‡‹‡É¯ª*Ñ‚½pw)|žý+ Ucf{(ñpˆô,.²$–Ž óÏd l°ùŽ?Zjªßî…(á h õqö“Æ­óbôð+»4êqúÄœq;±Áþjvè_x~XàC\KÑJ'q|§Ð×ö-Ö{åĸ7iÐÌ=™Øg>MZîÅ‹µ){_†¶]—*$r*cAq¼/¦p=6®_s4"ºð•ïšwµ‚ìžMzjì}ëáÎàã uûÂG ;_®·Mº/Nƒ_¼[º4t_ø%Ëòæf¤«6\¸ø—Þ®Û9¶±nø {l¦Æ©…úd›ó“ÏWˆtjŸ³¸Š{˜K›H>²¸k >„fØ3u"|W°¶–oðFä7eañÈmsàSNó8à;!ºâ áÓ Õu$i‡Û›•ƷǼǦþc=ˆåP´M0¸áv_•0&3ÆÙ?dîiýè/F,Y|Ñ\A-MWrE`äYÂàâÅŠö”Æ ›ê~^[C+ Dë¬ôÞD·Wî6Jö©0Õ}&ñ/Ü mÒ #_m¿[~y¤î›ˆÄï½¥ÇK~BÔé =ý™=Gy ^d!^o‡³ïñ$u(W-B‰­ç`в?ì0nùa¡+¼3±äd~G÷‰B¼D&b¾¡y¯?+hÏ(óì=°bÒ OÌœ·ÁÔwÔ'”µ@ãóY|…¦º’fJyU4öô•>,Ê&¡7M\ÿñ+5ó…wC…‚O|¸-ÖzˆeÆþŠl¿aÖ5 Y¾)DMdŒMW¨ŒôÛ…¢a®½$ÌÆÜi2šv„$fè7°ÞÀˆDÞ¾.d+ÝìãiÁÆÿ¶¸ž’Œý èD¿4^!ÛˆªæŒ5$.«ÞôavÆí{´ßP@ޤÚjífCåS ¾9fst8kC~ï5åd¿%sÿ>¢.ÃÍë|/«/ø ±xÎg\20 *5¹Io)פ­zœá‘¡v3lŒá# @·úE+îHH\~Œ¬J¾bûNé”Iìœ5®&Qµ #ÜýÝsõAoæaÆPˆ‚¹ˆê ·BòE,,Yæ·Ïø³É²?¡ØÃÍÞdcËñÿ¤Éƒ }S˜è³ˆ]k4ehÇ‚f,©Ä[‘Ò²3 SôPÕqÍ0Qm ªtËÜ31²P´æÖµïµ_ÅÝ&)É« þnžëG°v"Ýývæq¡áh,‚)3¿"}ËÂzÈF’§I6=oÁ¡Äjì€PŒ±7¾yÀÁ†-¹wA¢cÕëÝò”(Zƒ2:ñlÍj_¡…h}e‚ù&¥Œùþ8RǸdÏXÎH¼z̲xÆñü‡•Bpˆåbb€¸ƒ¡·CAñ¦bàt/ˤ#¹HQž( èmb/Ÿ%»k7¥bÿ•Ø}Ö‚ô‚3±k6Æ'AµÝ‘T묜ö¾cZí~Î^vNS¿qÙ<§n1£g´¦Û`tlFRõé9™»°º„V‰ÏC1‚‘QRÀírÔ0eì×é²ÓRÿ$`‘¾à„j¬q-êMÍx?(æâô´”ÒÄyìø–©÷µ®Ý0iOIçn÷Ðä(Ýç‹þû‡"¼oaªcÕ»¦—¼†ç•Š{øê×Ù‚°£RÃ=žrÃìËôj¦‘¹+‹à«¬DUœ7û`s˜XTúšn_Ý6b‘“rô+”×véØ<[ä¸ßâ>òƒ·¸×L°Þ=)ÝÔ’“Ѻ+t£¡Aï0a×ZQ¦@i¹Õy*Ž›˜»e5Þ‘Ð0ꠃ V…cN5~“Ö 2=éÅ£I*UéôøZO#‹M˜™ñrT¦ÃIËß÷ÒódRÞ¿ƒ›2¨/Ø“µq«òT¹\|4û0­Gä‚Õâ5å¾-8e eú™?µú@UïÛ;'}6G§BŽËϹùŒ–ʉ1Ë%ó)|——>„~Í–&ZAkôý Ú–o}eÅu@S`Øü¯„±„FJ–ˆÎo/æ€Lɬ'†MdŒ/ŠÃ×A¼$ÕWV¯m³ÒaQË(h‹à:p9›=p{‰…ôœ?·Ëá~FæÙůf8· ä¦)z÷ðK&õ5áN‘ÀŒòM­Šo¹-i½9ªÓr^FâLôíÆ Üõ3˜P{šò;|jí´~™šåtì‹ žfóýp¹¿ÀjØñ{÷^Ѝ)¿9×,IóÛe È?JcÃî·Á˜) ŸÉ™“íÆ;Ÿâ˜ä´MK[q°Eª¥ï•”ôƒéÌQQKïË!F™ìbÔ8¦h^¤•Œ—í̜ץÖü±ÔÞ¦è€ð¸{ö])špõS=Jš¿™ì'™$LŠ%!W³Í–þïÎklwè ÐÇê½ë½Œ¯ôÌç& ST*´(MFûy$ò| a¯Œ÷ží&Øãf±–áo=ÁÕØíê†\âþÆnè<ÇÉÆhÀÇÀ×|m›]Š•Kñœß!pH<±Q QéLCk‰+;±~úà»Î—r.”bAÄ÷ѹ%Ÿ"‡7´:Ha^7Æ‘§;û¾¬D² .9Þ¢Ù:1¤E)ßìý ]%ý¸DDœ—’ð¶NÖõ>s«#RÏx¸Øô¹±X•±~sƒÛe¾ÏB|¸×ôuÙ“±l–Oéå熶šY1eÁDó JÊ Ae¶ˆ½ÝüúPÙ¨lÚøXÇÀ¼Úñs|ÐöÑE•2Kw¶úÝê4L8™ãZñ‹ˆ¹SÈ ˜%A¶ïÏ$žE;Ô`x²GAáwÅï}QÇÄoâK²¢ÖÑ;BU/,¬;#Û‚®ê²ºèÈ3N“àH`ø4­Ò»VèBáð«»$"Gľ^DQ}Ìâ,™†±ó=¾=pyç*BÑÙuÐs1óý[½>&Ʀµ²Á8jûÊëS¼–ƽŸÇ6ùE?ݘøÇîáÏ·‹CCͬõOÞÛJ]Öïÿ<7…–óƒ·_Yò\‰‡Ì-Kµ¤kX»|oPä-q9—[áŠÆRÙRÂEÊ£©_ºø¢‰0| Ûú.¨],Zì:«“Ÿ_çB$d\…CÝäŠæs#Q ~úë´C6y‡ƒ#4µar¹|—ޏÃ޳ϱÒêMàVŠQ»´ªñãðÒ7 ˜œÕ:ÅÄ™+­:Z¨,\Š&–À ’â2]^’«È#î:“³§1ÑF˜ Þ…J 'p‘1J‹Ôák96ýûfêí'ÕÆQ"¡ÂU=Òã<$æ ·v¶OY¢x6Äî¦2K«[†î*ÁÝTk‹£ß£D7Õv ëûEïZ˜È3•„J[R+!FºŽø‘½é’Ìå £Ýô_X×Õ1¦þ¨LܽÎlnõ!%ªó›þ£r·™êjR >ýb;£9OÜù§¢7ÙËPPcÅJ„Z ȼäc³Äs.q:ɲ·Va'?¹ÝÖRÛH"¡ˆå ‚sèÁa'o·2ej—ŽExYiÉrçú8¥Y‹Úg¡…øÈ]VÉ¢®·Ëë÷ù&Ìü»67ßoÞÿ:—Þ(î Ÿ6‘‹‰§[ÜOIãWî§íš­¡iÕ¢qX¼Å»ˆ6 ÷%ïéQæô SGžìñ} ÿùÉ ½J™Õ=}[мokH„TÑqÄU¾ÙD­²å©æR!]t)Œª3 $lÁåâE N†âÈ¢ à–b ÂyžþðÝar°+ÅÉ<â×~ÿæëê]äâ…0Âæ¨ö[«¦Þ­KU$§§NÅÇ$Ðh#ªlo‚¡¸j ¢ç× f´@3Çe•Îá•|c¸Ê*Äà8µ/MLÓä3—Ÿe{¡üßW®ùµÔ‘G-³NIöê½Í]p+‡]Â5šæß6;xS³á1ecË$¦Ûå¤ð19ìö¢mÝà¡«À±˜=î²M!Š"c¿_lJøntA02¤ç•ãª6¢Ñ»%0W}+±`Dø‰q¼õíÜNw•éò©—êÓË>eSÛÐ!”—OûY¼×çC_×pºæö; X¡Ö@pTët¢8ºAÊÂ&Uoˆ8NÝvzƤ Ý͸c<}Ó¨ŸT«„wæûRuLœ~£Í‰Áy©l#( =|FÍÛø©øàZôSBR*NÈœ‚I‡¬ÚÓ’ÓoY ,:NÞ)—™õó—KOâmb¹ùïÃçî*i}‡Žƒa‘)t8{k'žôõÂÅ@ÎCýap¿g©è@ÄÍíy{ºb’Ò"y6n¶ZØ5þ°˜ÍÚ6,X¹Ë;ú&É·4½º2TYd$\m0ܼÞùb»\ˆosÍ…sF,]í§}·ÅäÓrâ*ÎiinÁ«æìcZ[D«¸´ÛÑBô+ëÎ*ŸÆÜºÔ|JaIæôž•lçŽÛÙÌ£~Cb¼ü<7ßÖ|^Ö8ÇåxtåœÉCâ̳;næþ´ú)h¼ïfjßÜâ³oBE¢é%ªÓ|ú°løA2Ù™¬ÞªKcé§Ö».U|e–‚X,¦AãBf£†ëØ6ÁvÝV¾*¡NÜåUÿšÄ±·H!ë‡SµzG!¥Š´mQå SF®9­Ñ+€ #xòíhúLÆ4Z"³Ô,"%†m/¦Nìv¬-=>®^–™ƒ% À¿ì îýyý„‚ʉ>tk–‘¹›®ØæÓòsP8f… :D\ZÉiuÿ1p€þIfIê×q¦‹ùî ¶F¡·L÷{‡ú® I1†Ó1ù8>á à©ÏýGŸñçÉÎxdg<EÊÎòŸòìý*5|œM1IkŽ’ ô|äo@~ÙM1N‡×IŽ}Ô†mL •Tîh›Z|§ßÎ1E/ã&©œÆŠ9¡K(vŠ“úIKwդÒ$OÜGšhšô¦îkN›šwŸ1ÈPÙmžQ#çf [œZ?>?y¡Ëîòñ²{–dZ;¿2Ùm…í-HŽÛ{¥"tÊ¥Ò£'ˆ@ymÓ®ëÒï¤ëŸX¢àFîþ*3'¿¬ÖÍ=°Hꌗ«'ßšAÌ ûªfð³šæ/9É“#o†–¨Ê¸Ärìc "(9>ÊÉw¹V‹}]vÞkqÁhš"'—ã8!Ê*–>.í2Õ=U‹Ü=H6“ßoTææ'xÄÎZ¸›6JÀ9GµPÄjÇ/lõP“% ‹+™"Ù=kµum\=»I–ì·cÅãç•ç+2@œ~ÃÎp÷H#Zƒ[˜åW¼³·4øA Ü–2Ýú†B£}÷ä®Ö^Ü^¬¡©Âõáãww[ã;±Ït¹IòÍ ºßä#QÔ•vÈbÙ*¾HŽdC7ï{2®1òtدªÜùÞ"˜ÞîOôžtÝDwÓ¡»äJ.µ~6 šÕz©y_5Iy¦üCå­vØŠUaÈ!‘~à Æn8,[ÑS:¥¶‰[ɨõ°€4£'àýNÔ&¾Ïd86%³q[iÉhѰL}0PY½ž³®4Á3Jd«ƒ¡†EK1–VVäœ'V‹YÅÁ ÞB™“S‘+zME>#¯ö:I‡E–„¾0ëmÖ$`xWbýlÀSÐS¶ÎÆ( Ô¡ {]ÂÚÎwÀ®8a‹õr¿‹kÕîR?ÿJ½Uìêc}H3æ9›“¬ÐÍÚS9åH­:÷}W-ǘ+àõòþúz¤æÊ\²Rú'p¯£T§»&ôXe ®U¯?àTøc[HÑÄ Z~šÃ÷§‹ó­òξà‚”›Òêf«¡Êÿ­TþÒælÁ’JÚÁÇlóê‹OZ—D2(.¹-åfµ¹ Ê¯øFÅ.èùñH°ÓÝ*$d³f4³×í,Òô Χƒˆ¯Láùܦ`×Ö¸7’Õ#»Þ`ˆ+Ñ4ÛŠle“ÈÐÃÚ¶ïÆöŠ™ãL­ÀÅh7NQîJïÒŠ½3ÎÁ±aß*ÛÉ,,k;9°öª†Æ¯HRëûÊ ÷–|Ÿì– yÀ,µqš êñª¦•RÉðKÏ“a1Z[þ4" NÄ7 { UrêWrvm]t½»Èü-j²¬NäÌ(Ô)èÙ¢mJçѦ3Š )Š &ðî Kýš2ôÔ@ŸÉßcQ§LÚ¯¡‹Rû„Š)o*Ûýõuó1p:þÉ­pø¡‚¾ƒÿh GòãÒר~k€Mø6’õðTbŸâïîqë¨ ÊÉôlÉž«ù–Ç÷Ù›&•ÜyÎßÊ‘ýšíFduFT¸‘“@¤OŽÂtR‰*jb‚®å3þ%%¯"¨bgè\‘MÉO/ŠëhD™Ÿ:#‘ÅSëxzßh»M°z-›ý|J”ƒ—Þ~à*À÷šq/`4Gf#¸Ï|t«ïñæs æ·Š‘%Åý•ì÷ÛÛ[Ч–4[Ï ¥®Ö!ù]C%z¼éîØÈjœMOÄVåíꘈȫþ~NPwù³•å‡Îk?Ξ>amw®WsÍWßóï*öà¾â“£;”ÈWà»FÚšE%üš>¤úÓJí"{b˾ZV1;²€,ÂcµMï<3ikslL%Ât$æ=NFª7 »0ýᤥˆÀªÙGÛsÖjuãm4ÜÒa9JU5µ#á—öìc$’ĈŒsÀJiºå8Œ<Êãú M‡5:ld™Úr¨ aX´Ì*•@®¬›&—?]²º ·û¾×¤N!!×u:ZM°AŒ}~TFöèÝÏnº Ý´Ýe±É¯~ó» ÿè|‡Œ1ù$eæN99Úà“™ü+øåy@e™¢Píã…þaF ëXZÖ¡ò»* °æ›DŠ&‰y?Etz½ ŤùI(‰ïT#¹Ýù÷p*¾S\P #•tS[ ƒÎêc¿ÒÎvãKðÏÆkWù{¬¯êÀE@㠵Ϟv4uYÓvíÍk”I±äÝnhÂ,TNÃ<‰ªÜ½ƒrmO/SD¨g˜­It 8çR²6€mkBÀq€Ââk’9ÑŸVÝµÓ µÆÃï`ÓÝn5AÉ.:Úêý´#qÀym/p½„é)×—ì”Eîì*eªÑ·Þ*Ç­¬˜7Tó!bïåsX£§LÅ)´Ô´PôÂæ—/“.YËmì3‡x¸*x Ikm9Y•ý[>(¦ÝöE¥?í«¿7€jØ·•c’/×=cëƒçžùÀ^Ì ‰;ý¶S#G?ŽßÚ_±½8HæB[lÇWœqý‚ÃÙ¤uyòÒ‰V„CÄY˜M(pÑ07èððhD2x\8}òTÛ¨!²AŒåV¡{ÅdaŽ«1g*”'ˆªÝª]¦èäUbÍØVc…vd¦Ñ¢¯ôžjå;­ù>´<û&ºÑsnÇSNÅÌ úeÿQUŸ˜h)æGhîæç!d¶—’%f¸µ[®\,Œ­óüµ‡0Í'ÛÈ°ÏØœˆè1‘O¦é§¶³®Ôèl*΋ óÉ•a¢¾Ë“³üYœ¹íb¤9£n!·èHŸ–Ù¢yù|0öFJßsžþp£©b£MLÁA®Â§—MÉÚ>SïÌnÎ~Æp&1¦Æ¦½ Rt|Õ‰d›Å‘ _'úýÖr]öçÍÙ‰dìQ†kÿ›ök¥B›ç‹ùZµ‰ƒ6òƒ“"â! ëñµcÖü݆KVnêHÇœ¹lv¦7´JùœŠqËqÅ$g’Y7kƒ)Õ¨Ã& üÎÔŠ'kÛQ9iL™*]Q ÷¶,ò¡¡u‡4!jC‚oß=I°Rfפ°„»h³R ÅÂ>Bž6eÊ YÚÒMœ ÄóL‰dΣзÔèžvш¦ˆ“¼_ùðÒ¯pÆ5)Ú`À„ H|oº$±w{»aÀƒ'š§.5„$"„I²öŠøÃ@Hâ¹ùÀ}ih.ôê{ðS¾ ³| ßE ³'"§âµjY’À¢Š S Ѳ¾H"Ziºyö¸‚]>ö Êk})>?÷5o )êPó3wï:½Sým4Êq?ïþ·Ò'¢l&ª]=-ÐM‰¿œÁúvW š”[Ãò–ŸFž rÑÕìHöòe×4úkïà˜…1½‡G)8I Ø&\ϧL¤Ã¬{y:L⓱s×ïwÔÞÁÆ”StÝ]¼WeŠâ¦Ôvß~âdEuQÞÄOº½)~5(”Å… />Ùó-`¬ƒ„ç¾¹†~ ßþS|˜•rÆ—¹¥"› 8¾:K(²áœ·šÐ^™7-<#j_Jxi|S…²Dásä1²Û> ’€LHSWâkX²^˜žFï·ÕöJ1QÙ‹„ðHŽ0mNQaªzà=»dsKžUC®{b²Â7„O5ß¹Ÿ“Âpz±]f|ð=æ–÷ø^I§´$‰2Éðœ\’­išÌäÞª6õ95–CäDeƒ{(†¡ÙÌY“\‚޾¸‰ Ž™êwžø1«J ÿb·10ÑëTþÕek’PŒøÜøÓäù¦™ë«zèŽÉËekâ$$®%Vͽãæ49d¦ñ‚òm³•,&øû¹˜›8W¿ Ô”Zòó<‹G€w†±°zA(áçÈ ùw»­üï…`Óê£de6qïë„D ©ï°$ÙTl²˜·¿$%µ*ƒhÉÝSTKªA`Whïé£ð¦M³6UX>½·}6á¡N*ÕB\ÜXmn0)ó„°¨óH«èm¤Ÿ±*ÃyuðiÆî:RÐ*)5‘ýáð¿­øáúí‰FµãxͶɲ”Ú]þtxz‰¿ô‚óð8Ãdd¦4|-d–)(޹¡ÐŸ 0'ëVo¶‰¼ƒ¿Eš×;*dÛåöµ%•NͰaÇ/>~ Öõ<ä¤òI¾ù‘›ó4Ò°­°òàxÎÙ­‡Ñ¢3ds°GæùùéfEõœý”˜;ó/ÑsõÙ}2Ž‹ôÌõš¿áÒ«ö>íŸ%Ș{©Á듞z7”ð?KÖyéÑ+ʼ‰É…E0òÄ÷¿Kòü ”ßB"‚¯Ö½Õ”©E-µCݦÂ7ÐGhÙO½Bê]¥>lòíâÌ¥‘.ê†*<Û†tsøã È’ž®*’á6žTsšI Åù#Üf(+JXbíwІýâ^SЩ#‘ú5§ºþ†®Qgc-qE[wÑ5×XÇûñº°„ô˹áéì„ñzd’Ò´ç¬Ã{ÿUhý0L ™Rs'ÏçâS­òë–!$–;œ+0°€ß=pèU1ìoÅ÷íPEEô*Âú rQáMl§9ßýbÇ><âÀ^l„ë‰Õt›¼PûÏ¥I³vÜ9Áû½û××ÂúŠ!:ºáï/–2(ò¬šÕ^ê¼F‰DŒ(\]‹qZ&Šc‘ã;Ì®kÚÂm e0|5G—uËOñ$Ç"Yj(>ÕQµ¦Ÿ÷V I6§ª˜óÌÄÐ!Ë%ó†¶%Ê}v;3Hèé™k5èM¯S Ìñ|ŽÅг·n6Ô—­äs`¯.ïÖÚïIÓ@kAïµÓjH^A8àìb0Ï.‘nŸ|ÏðâJf¸¯6»õ½J ×/ÙDß#ñÔrÔ"5s: æjwÜ£Ø<ÃeÃÚ)FÝJf”ZÏÉ q%µ«ÇôÚÞ>Å"Ò¶³Œ*WpÂø„ª2Zñ†¸4ÉU—'M^»¥ñ¨¾Ô¶ï×& Ée[‚„·<Ÿ-xvm9öüÝÇJe:¨øI~|ïMÏ~HÝVV¥åŠ‘¹•™[UáLX#«"w"UæëÉQ¡¥e·(96éÊØLr8%Sƒ¿°HëàùQê½ÒÏ›+¸dǶ6DÑù@ø¡ˆjbD¬·@iQüFÚͪ'V;“î¥ÏöÓIs‘…«âeísøÊîåp=]ÈÚ²‰&ý%’R±¼»ÂÎÎp>éPe¥Ú`'KêÍWÊŸÃÃIv®‚.T {Ù85hÁûe¾¡Õ# Ü#è!Q/Ùµ),Ý5æõ¨›º`‰J`à%4¼ß'·µ+g –8mUwÆ,³È.‘1ÐÇñõÉåëŸ8ö4ñx1¡¤–û#3nü ŒW±ZkÕ;Æ îÄ?p,r+Ô¬Gz÷³åŸT¬|ÌŒŽ¹~ÖªesP.ª¿ /}œ@‰KGÌ{V¢s;\¥”ÐÙÝl>‘ïãߨ!CëNP}ÏIâÃô8ŸޖMª/1Ì>d W“ò œµ1dY©ú€‰aóXðM7«{ ùLì‰ìW;ΓÃU[}Êé]ÎþÅ`ì`T}¤i™ç²ŽóÒ‚¡—}…ö•ªEðö=íÌ!³áEva„§=öódÏ­,šªÜ¾À‚Yª®~g¡\ïÖäŽñvð|£…w6°ÓÖÙKW<Æ©áN%x·KÖîª5šp9!'£ÆãÐÒ [Ť‚­é[®6ëv>\à=5êʰØ3‰ö=¼YàNk÷óÎðlîŽó¯´æ·FP!¸ü)è†5ËBj¼±êᆠº÷-y-äV¦$“– OD^;/øë¼³/ <¥¥¾04ŽJîá¸j†²t^W™À±·``È Ú©àPUëÏÏT)‰‡@—TPœz×+‚rZcbkáÚÄŸÒs©¦£LàÌŠÉpP:ô]2í‚ó86O°}fIu¦ÇæÈËägÏ´²?·Ô¢ {'¿I‹zõ›„­ï”?¼/ARÖ†{¹4¦µÔ•ù°á×_[64çç­$H¾.‡úŸš&Âé^ßàö´§ œl•õP‚`™¦ƒ6áå ß~¶–_^kImÅûÚWü˜Jc…lrcº&fcƈX5VŸ û^›¸XˆDþk/Ä¡J±±ßš‘Û°5¤-RNC4‹¬B䇴©Ñ¿û÷âv=hUø[â_…&Õ6éÄVþôòþ5Ëör¬z>ÇÆ;®€þ­¯Ñ Oñò¡Lè®]V1¨à&hU3š]) Ó×É=C>֜ږ4ßÌz­º7›Êã8ýa –“<¤'‹¿ÖC6ÍçT5‚zK7ã–ÂÃø¶:¾oâ¡J(K¬É å¹òƒÃÜÞ4øR*]EéFg»j~Ú•°HÿÒÙ3ÕÞû±†t…ÀFÍU¢+7(b5k»²ö›wÆ,Õ¡ïxšXŒZ¹Ç…ÔÂ#rÅ/=7#AžôàŸ6B¶xhÝ 1-\t´>™$}ÃdwÖF”æñ>î°|kñ¾m¢ºcF|«ê Ìt·É£r7sù)“\sö˽p’ðVô1Ë='¢PRñҪϔo²þþ ×(ªØ3#IemiÖéŽ m™ Î'‡Ÿ2̆ç²M;8É9è èèîD%—[•é#ÒDr|n„¢'t:ûôŸ‘ó¿8|cÓÕ@=›&Ö:Ù»‚ÝQoú•–ÁßòJ õ²†¦Gº*/ ÕÜð1¦HNg¿ÏKx4©yƒv‘o{ó„hÑèÝÑÊCä;`zD`¬þ[! Z®§í ô’î<µ¤'J µÑ¤*5±DÉcÂ×µ8(1¾;M¿Pª0cË"—}ÉÈ%‹%I”Úp,ª\¬¨åuMãÉ„óÕÊboŠùZó C·Ôpä XrX0k¥J!OŒyNØ3Ô>Ù’CÔÍ$tn\D÷}¸]¶Çë«öŒ'þâdkØ GÇÄÜÍñv>6åêú`òuWΆ¯o+N&Žý‰‹€«›ÊpŒF:a…¹ÇÓ£Û;ë’ßž#k6+‹í â_ü#ó0.îÅé§ïÈ^­Zåj_´B8aЗvÏuæ¾óŽØdöÖbÁ÷­•©pª5ŠñiI­c_6Ãxí·|h¸W¸ß¢+Yõ¢Ø±i™«Ê9P S«óíùN­¾§µŒ2ûP¸°aè÷³?óë#Â7kâ Ðñ˦р"ÏL·ÏÛ§¡º¼(ìŒÂr8˜ÖïeéLßvý³r¯‚VR t:j \ÛEܵÅFìç‚ü’£\ÆïÕqaÀt˜_V¬£ É®q0xu0ÄR×e0Ó1Eè·’²uÔêf“—oZ4ÇãÆ©ÂŠŽÇœ×¶k‘(©|‹Òy9ÞwÛòSšW½cmèëã\P ‰&m•?’„NÇ•¨© ?]§P£fõ>Ÿáq›>»íµˆ‹Ð”ò[§ñÂÉÛ§ØËQZ×èkòeœˆø,Î4”ˆ¢^:U ¨³æz³†¤w¾¸+óµò§·[¼%õqÓò‹_K4¬»¸-~rn=èg l÷D±á°Å–:…ær-6÷óßÿTá>cPÜÁ¿úˆåW¬°<´Cå’ªÂßq‰êÛ—kŒ[òiNa|cD¡E7Gˆ—_C¼,}4 vFµ`î(–ò)T6þ© ï›Ë”ꌻ.ó~'4ׂ>—úzdî Ù[à”…ŒÚ±ºc0iþ`ÉÉ‘+ù+o4UÒ• ;wà¦|XŽÔ/·}Á[¼! >â jÎÄÊ^â<²6ÎŽŒ$ë¶:« %‚©r³=l¥  Ñ#oô™Üõ>¦‘Ê •SZ®GÔãôðiAܦN öšžÚÙÁ­w¶\ÛüDw: ûÑUdš<ݬâ¿äòáD˶aÚ¨íÒKaðb¯t›ÙVy\oL+ÓCˆjâ + šh/ PÊü-ábõ†"B^«2 +¨Xì)ð«‹Ÿf\v„¼3àÛ~\ÇþâTiù÷é”y’“׬æ>Ùát¼wr4h©ÍêÐ’QyžD_c³:Ÿ¹X^{,o< Ñ?® ­w–&ÏíUX‡-¦¿Rä¥5—üö†«#ÇŒlõ°Á7ð ÆðúW]¼>I’÷sîëWuŒÖÅ»_ÕHæ°ºÃåDÀ\®uÔ£ñ¯œ9ÂbDµ®…ú¬§ z់ÙUjöõÄët[æòðgg«%œâž3Ú†éJï•:HUvùµkaØUÖ¯·VdCM Ðpb¶p—¢?t~lj”5|Î0-¨ÅÕTm΂Jú_TÇ@·Òá36†ÒœÚÆN˜þøù´«¯eI³¬N5,fEŒŽ7 `¸ô±Zë•„«bäÌ'²Ú©C‚RL$ò,of⢂ü½¿G¸Â«…ÿ› G·Ñ4â –)~Ê¡îzWùÑeNƒÈäT30”h”£r:†'>ÛÀûjKýcÌy•¡¿+qJfÙç@¿Øæ¨a™¦º¶‹jeGCú_‡°O¡ÙX?Îõ$Ѧã /80·Ÿ3á_•{¡\ð$‹ì)ÌfØ,Gq½9u5…F¥'/ø¢“k]šŽ6ÄcëAÇŽ8ºÖAóÒSAS= yi6®QÓ{C¢³Î¯$š8xâ“3‘‘U—$FŽsÌ7|Væuúâ%7‹Ñ}aÛC¾<úe\ ¿óµÅàÆ·›UýuÊ’>ê°ïfYñ1wìÆ þ톒!ï XÈCÅîŠìYš©=»{OFoªCÛµÖ!QzQOËr1˜lr{ìI¦‚béWh÷pۘΖ"zwe&×ÛD¸½¥i/͵E×{…¥O{‚l4ö"eÖ·†úvñ s¥òK~B KoÞpÁHfER“C<5²Ñ$I¦…³,m“9m6v>Åfû'¬ SlYþ1ºjŠ4ëq¯¡ÝZøa_KzS<|yÁªÜfq^¡#@Âyu­M3¾›ùèMÔø:y£düVx;ßf *žç£^®  µžj<¯Ýë)?-xî  rØZÌÈw·v£ûwHíR„ëAÒïò•=mH>öÕƒïcêpCd1°ÈQΟ6ô/Iý-?4~µK¯~'ïMÚÕOâ;ñën¾ô¥;Vø,øqýü%òœXõ:èZžéë‚÷ãh(sÞ ŽtÈYˆâ è›z°!õàay*Ì,î'Ü,ìÕ‰sOæ• ž-6±ï=¯K*ˆN,8)§&"OQb¢C\ )ÙèÒ°ç€F·Ÿ{ ?ÝîG‚Ù½¹î×7ŒBQuEÉ©«RŸÖ”*>÷ä_Hн= ø6˸|©j㜢˜"q¹ÿÞU¿Ü/ùæ¤N9Q¡üPŸóûkýU¿n!$öVEÅyï3ÕñïèM_h#_êÄ"x¶;Wg_¿½àÁ€K%RYrë’ú "%=Ö~Ïmc3–ñ7× iŒëOÒúïòæåϤ6N§ '»×x`•g;ù a¤÷Þ|B°fp²{+7c`…Nûÿ,ÓøúÄæo›øŸd*NÚ»·–1ÅaB3[”@¯“øF°ŠBÂzYaÿ/ÑÜW×Và{#~°åL *×ßGŸ**‡1„¹¬S¹o:œá „ÔÕ^y&˜m‘öE¡…?k˜s-d‹{•Œ‚…“F`G1"ÿ³>­ù¬}§1p<¿cU‰'ñJÚÎäÅÔ¨X2¹œºdŽ œ;Á~-ßd"Nœ û#)1&ÀR½ReOªüIõÅùk3âãCTQ°sŸØU!ƒDˆ7’¡X¾Ö‡—dmú^Ø’A´ZõipD{ÿa p]ÛK¸¹Ã®ðøB0Í ìsðõÖ2'z1?—°Þ3Ÿ‡, g>q^3€Ö÷Ö“­•ÙµÄ,¶¹K †à(¼,í#m›ÓÉ»ºÃÞÚ@Fâç©2†ûÀ쥶º€L·' øû ¾ÊKw’ »‚h½S·ìFÁ Ç÷‘Šå)±{$$ëè~íãQ»ÛÌõ3¬Ârm6‰=â¼cÔ˜ˆƒÄÎ#ŽNÌÎêÆ(|é]xåØÿR["0 %Ù7Á©F¸Ýi;ÅÑ#L®Û…ªN¼EÌDô‡½È¸ñr;›b­ñ*õ^ÛÌ´ Û"% Ë8%{à}ÑÑ@àa'uH9ë‚ æv’ç•+Yo¥+Ç–’sù/ÁC‚j£šfš"_å59w¶´ƒÒ"9B™²L{µ8U/âÞÛ䥇E+ÐX·”=í9ÍpL 4wNe»ñ…žý½æéÓH&8r`Ð䮸—ÀÔØÆ+ šÓǂˎÒ}ÙÊê)û ÷¦Qêø Ëxê~ï/¢—3µ® 9Úg×®nóæf%€LÚK]ReÔÞ¦ü°/Ь}uœçMöUÝ{ÅŒ^{{éÞá†óïÇ;hvÙhHâì÷F)S+Éç[ Î ¨˜¸éʶP:u¨_#B':àJ 6ydœ¯‘îü)9LÉ_{ÎÖrÛS0Wò1«oÿïzJ’PO*$M”XkvòÎ1Â\±Í3,cÃK¢yÀÇYna™{´Çôöµo$!çt`·:Ý‘]¢»VëBxRö¨òÜÇuϘd ¦”þÃΆç·TúsPtì§5Õ-¹S¥áa’rBí±ÇJðŒrc.Š#jxƒÆã…›7t]¶øn¾z àâIþƒ0¯ “s4‡kw ²_ëÎA§oʈ»X/e=ª)>W]¨®V­ÉÁ .8ƒF17 Íà2ÉlsKÚ* —E¨…qYs{UÔ­]O© úÄd)}o?a¥£Ñà~¢éOƒ°¹d-@ö?Èxš˜XBòØ#„éÛŒQ"¹¯WcnÍiN}Ñ-*:µaòsQ`Á‚È¢M ™ò9iXC ™ÿ‘ê+J(C2‰¤Ýr¦¶úÑй©ƒÚÑrKrq=( 2#.¨à–F µV!æ“úþÑ./6ݱ÷÷Ct&›M[õ ŠÎVžï>°NTm­¼|p[†P«ªƒ"«¶Ž2 âXXƒ®JÒ­žŠKͼ6ë*ùjBóíFKuÕàÍ®Zò/k¤àÁþúئ†ZåoÛcï£4.ñIòj‘œ_ý”#y8 6{÷¬e”ÛÍäE87o?mGʇ!Žpö–nµ‡Nrìw ûyg/UõÕãëËvkÈ`S³™Ð.T$/L7Üš=¥ª J| "|.Âe¥V~UÖCR" Ö8þ9‘åo¶¸!'"}õn±:Ssý“‹(,ÌÄÀ™Ðë?óyHo¦bxU}åQà=O´­Þc°0ä½ø’Wé…ï H_Zؘ‚l7X°ƒA¡Ä™Ÿ–ðÊõ¯– ú~š‡QÑT†Ü|T–]#ƒk£JÞ£†Lq! æùÍÚ£I(¨E*¢D¥Ê? ÚŒ ‘xZs.žt!<_Áëu-éÇ9dL;$^ÛõôK˜·rœ•·Ü»ÔÌ®¼R°×L‡qB‹c–’Ý´tDê¶dÛšgA˜:&T$ÝH@PO›~¯B»"\ÆYàP»†:¸òtQ¹Õxc2ʾUYÖ"À!(¶7X3ýj«²4‚‘mCoùMM4‹ÊÈײ¦qwdÕU|7I8³åëËU\æÁâï Ÿi¡¿ ¼\9iÏåÇQc_hLrÖU*C!¦½ ´ÐM¥9g'Àa÷>ù3ÏÊ> endobj 8 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FTSDXZ+CMBX12 /FontDescriptor 66 0 R /FirstChar 49 /LastChar 116 /Widths 61 0 R >> endobj 14 0 obj << /Type /Font /Subtype /Type1 /BaseFont /MHOLNP+CMMI10 /FontDescriptor 68 0 R /FirstChar 120 /LastChar 120 /Widths 59 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YVNMTH+CMR10 /FontDescriptor 70 0 R /FirstChar 11 /LastChar 122 /Widths 62 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /MRSWRE+CMR12 /FontDescriptor 72 0 R /FirstChar 44 /LastChar 121 /Widths 63 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /QIJACW+CMR17 /FontDescriptor 74 0 R /FirstChar 67 /LastChar 120 /Widths 64 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YEYORE+CMTT10 /FontDescriptor 76 0 R /FirstChar 34 /LastChar 126 /Widths 60 0 R >> endobj 10 0 obj << /Type /Pages /Count 6 /Parent 77 0 R /Kids [3 0 R 12 0 R 22 0 R 32 0 R 35 0 R 44 0 R] >> endobj 49 0 obj << /Type /Pages /Count 1 /Parent 77 0 R /Kids [47 0 R] >> endobj 77 0 obj << /Type /Pages /Count 7 /Kids [10 0 R 49 0 R] >> endobj 78 0 obj << /Type /Catalog /Pages 77 0 R >> endobj 79 0 obj << /Producer (pdfTeX-1.40.14) /Creator (TeX) /CreationDate (D:20150701163331-05'00') /ModDate (D:20150701163331-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 80 0000000000 65535 f 0000002685 00000 n 0000001059 00000 n 0000000954 00000 n 0000000015 00000 n 0000117999 00000 n 0000117860 00000 n 0000117721 00000 n 0000117439 00000 n 0000118138 00000 n 0000118278 00000 n 0000008021 00000 n 0000002577 00000 n 0000001170 00000 n 0000117579 00000 n 0000004806 00000 n 0000004949 00000 n 0000005047 00000 n 0000005082 00000 n 0000005324 00000 n 0000010018 00000 n 0000015094 00000 n 0000009910 00000 n 0000008138 00000 n 0000034889 00000 n 0000011879 00000 n 0000012022 00000 n 0000012120 00000 n 0000012155 00000 n 0000012397 00000 n 0000017987 00000 n 0000016178 00000 n 0000016070 00000 n 0000015223 00000 n 0000024248 00000 n 0000017879 00000 n 0000016257 00000 n 0000021033 00000 n 0000021176 00000 n 0000021274 00000 n 0000021309 00000 n 0000021551 00000 n 0000028854 00000 n 0000026537 00000 n 0000026429 00000 n 0000024354 00000 n 0000034599 00000 n 0000028746 00000 n 0000026628 00000 n 0000118386 00000 n 0000031384 00000 n 0000031527 00000 n 0000031625 00000 n 0000031660 00000 n 0000031902 00000 n 0000034705 00000 n 0000035134 00000 n 0000035159 00000 n 0000035219 00000 n 0000035253 00000 n 0000035277 00000 n 0000035667 00000 n 0000036069 00000 n 0000036691 00000 n 0000037120 00000 n 0000037458 00000 n 0000047447 00000 n 0000047704 00000 n 0000054839 00000 n 0000055059 00000 n 0000077942 00000 n 0000078420 00000 n 0000087851 00000 n 0000088109 00000 n 0000097921 00000 n 0000098165 00000 n 0000116950 00000 n 0000118460 00000 n 0000118526 00000 n 0000118577 00000 n trailer << /Size 80 /Root 78 0 R /Info 79 0 R /ID [ ] >> startxref 118843 %%EOF survival/inst/doc/adjcurve.Rnw0000644000175100001440000014753112545056257016206 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \SweaveOpts{keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{prefix.string=adjcurve,width=6,height=4} \setkeys{Gin}{width=\textwidth} %\VignetteIndexEntry{Adjusted Survival Curves} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text require(survival, quietly=TRUE) fdata <- flchain[flchain$futime > 7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) @ \title{Adjusted Survival Curves} \author{Terry M Therneau, Cynthia S Crowson, Elizabeth J Atkinson} \date{Jan 2015} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {adjcurve-#1.pdf}} \begin{document} \maketitle \section{Introduction} Suppose we want to investigate to what extent some factor influences survival, as an example we might compare the experience of diabetic patients who are using metformin versus those on injected insulin as their primary treatment modality. There is some evidence that metformin has a positive influence, particularly in cancers, but the ascertainment is confounded by the fact that it is a first line therapy: the patients on metformin will on average be younger and have had a diabetes diagnosis for a shorter amount of time than those using insulin. ``Young people live longer'' is not a particularly novel observation. The ideal way to test this is with a controlled clinical trial. This is of course not always possible, and assessments using available data that includes and adjusts for such confounders is also needed. There is extensive literature --- and debate --- on this topic in the areas of modeling and testing. The subtopic of how to create honest survival curve estimates in the presence of confounders is less well known, and is the focus of this note. Assume that we have an effect of interest, treatment say, and a set of possible confounding variables. Creation a pair of adjusted survival curves has two parts: definition of a reference population for the confounders, and then the computation of estimated curves for that population. There are important choices in both steps. The first, definition of a target, is often not explicitly stated but can be critical. If an outcome differs with age, myocardial infarction say, and two treatments also had age dependent efficacy, then the comparison will depend greatly on whether we are talking about a population of young, middle aged, or older subjects. The computational step has two main approaches. The first, sometimes known as \emph{marginal} analysis, first reweights the data such that each subgroup's weighted distribution matches that of our population target. An immediate consequence is that all subgroups will be balanced with respect to the confounding variables. We can then proceed with a simple analysis of survival using the reformulated data, ignoring the confounders. The second approach seeks to understand and model the effect of each confounder, with this we can then correct for them. From a comprehensive overall model we can obtain predicted survival curves for any configuration of variables, and from these get predicted overall curves for the reference population. This is often called the \emph{conditional} approach since we are using conditional survival curves given covariates $x$. A third but more minor choice is division of the covariates $x$ into effects of interest vs. confounders. For instance, we might want to see separate curves for two treatments, each adjusted for age and sex. The reference population will describe the age and sex distribution. For simplicity we will use $x$ to describe all the confounding variables and use $c$ for the control variable(s), e.g. treatment. The set $c$ might be empty, producing a single overall curve, but this is the uncommon case. As shown below, our two methods differ essentially in the \emph{order} in which the two necessary operations are done, balancing and survival curve creation. \begin{center} \begin{tabular}{rccc} Marginal: & balance data on $x$ & $\longrightarrow$ & form survival curves for each $c$\\ Conditional: & predicted curves for $\{x,c\}$ subset & $\longrightarrow$ & average the predictions for each $c$ \end{tabular} \end{center} We can think of them as ``balance and then model'' versus ``model then balance''. An analysis might use a combinations of these, of course, balancing on some factors and modeling others. All analyses are marginal analyses with respect to important predictors that are unknown to us, although in that case we have no assurance of balance on those factors. \begin{figure}[tb] \myfig{flc1} \caption{Survival of \Sexpr{nrow(flchain)} residents of Olmsted County, broken into three cohorts based on FLC value.} \label{flc1} \end{figure} \section{Free Light Chain} Our example data set for this comparison uses a particular assay of plasma immunoglobulins and is based on work of Dr Angela Dispenzieri and her colleagues at the Mayo Clinic \cite{Dispenzieri12}. In brief: plasma cells (PC) are responsible for the production of immunoglobulins, but PC comprise only a small portion ($<1$\%) of the total blood and marrow hematapoetic cell population in normal patients. The normal human repertoire is estimated to contain over $10^{8}$ unique immunoglobulins, conferring a broad range of immune protection. In multiple myeloma, the most common form of plasma cell malignancy, almost all of the circulating antigen will be identical, the product of a single malignant clone. An electrophoresis examination of circulating immunoglobulins will exhibit a ``spike'' corresponding to this unique molecule. This anomaly is used both as a diagnostic method and in monitoring the course of the disease under treatment. The presence of a similar, albeit much smaller, spike in normal patients has been a long term research interest of the Mayo Clinic hematology research group \cite{Kyle93}. In 1995 Dr Robert Kyle undertook a population based study of this, and collected serum samples on 19,261 of the 24,539 residents of Olmsted County, Minnesota, aged 50 years or more \cite{Kyle06}. In 2010 Dr. Angela Dispenzieri assayed a sub fraction of the immunoglobulins, the free light chain (FLC), on 15,748 of these subjects who had sufficient remaining sera from the original sample collection. All studies took place under the oversight of the appropriate Institutional Review Boards, which ensure rigorous safety and ethical standards in research. A subset of the Dispenzieri study is available in the survival package as data set \texttt{flchain}. Because the original study assayed nearly the entire population, there is concern that some portions of the anonymized data could be linked to actual subjects by a diligent searcher, and so only a subset of the study has been made available as a measure to strengthen anonymity. It was randomly selected from the whole within sex and age group strata so as to preserve the age/sex structure. The data set contains 3 subjects whose blood sample was obtained on the day of their death. It is rather odd to think of a sample obtained on the final day as ``predicting'' death, or indeed for any results obtained during a patient's final mortality cascade. There are also a few patients with no follow-up beyond the clinic visit at which the assay occurred. We have chosen in this analysis to exclude the handful of subjects with less than 7 days of follow-up, leaving \Sexpr{nrow(fdata)} observations. \begin{table} \centering \begin{tabular}{l|cccc} & 50--59 & 60--69 & 70--79 & 80+ \\ \hline <>= group3 <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) age1 <- cut(fdata$age, c(49,59,69,79, 110)) levels(age1) <- c(paste(c(50,60,70), c(59,69,79), sep='-'), '80+') temp1 <- table(group3, age1) temp2 <- round(100* temp1/rowSums(temp1)) pfun <- function(x,y) { paste(ifelse(x<1000, "\\phantom{0}", ""), x, " (", ifelse(y<10, "\\phantom{0}", ""), y, ") ", sep="") } cat(paste(c("FLC $<$ 3.38", pfun(temp1[1,], temp2[1,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC 3.38--4.71", pfun(temp1[2,], temp2[2,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC $>$ 4.71", pfun(temp1[3,], temp2[3,])), collapse=" & "), "\n") @ \end{tabular} \caption{Comparison of the age distributions (percents) for each of the three groups.} \label{tflc1} \end{table} Figure \ref{flc1} shows the survival curves for three subgroups of the patients: those whose total free light chain (FLC) is in the upper 10\% of all values found in the full study, those in the 70--89th percentile, and the remainder. There is a clear survival effect. Average free light chain amounts rise with age, however, at least in part because it is eliminated through the kidneys and renal function declines with age. Table \ref{tflc1} shows the age distribution for each of the three groups. In the highest decile of FLC (group 3) over half the subjects are age 70 or older compared to only 23\% in those below the 70th percentile. How much of the survival difference is truly associated with FLC and how much is simply an artifact of age? (The cut points are arbitrary, but we have chosen to mimic the original study and retain them. Division into three groups is a convenient number to illustrate the methods in this vignette, but we do not make any claim that such a categorization is optimal or even sensible statistical practice.) The R code for figure 1 is shown below. <>= fdata <- flchain[flchain$futime >=7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) fdata$group <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) sfit1 <- survfit(Surv(futime, death) ~ group, fdata) plot(sfit1, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") text(c(11.1, 10.5, 7.5), c(.88, .57, .4), c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71"), col=c(1,2,4)) @ \section{Reference populations} There are a few populations that are commonly used as the reference group. \begin{enumerate} \item Empirical. The overall distribution of confounders $x$ in the data set as a whole. For this study we would use the observed age/sex distribution, ignoring FLC group. This is also called the ``sample'' or ``data'' distribution. \item External reference. The distribution from some external study or standard. \item Internal reference. A particular subset of the data is chosen as the reference, and other subsets are then aligned with it. \end{enumerate} Method 2 is common in epidemiology, using a reference population based on a large external population such as the age/sex distribution of the 2000 United States census. Method 3 most often arises in the case-control setting, where one group is small and precious (a rare disease say) and the other group (the controls) from which we can sample is much larger. In each case the final result of the computation can be thought of as the expected answer we ``would obtain'' in a study that was perfectly balanced with respect to the list of confounders $x$. Population 1 is the most frequent. \section{Marginal approach} \begin{table} \centering \begin{tabular}{crrrrrrrr} \multicolumn{3}{c}{Females} \\ & \multicolumn{8}{c}{Age} \\ FLC group & 50--54& 55--59& 60--64 & 65--69 & 70--74 & 75--79 & 80--89& 90+ \\ \hline <>= tab1 <- with(fdata, table(group, age2, sex)) cat("Low&", paste(tab1[1,,1], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,1], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,1], collapse=" &"), "\\\\\n") @ \\ \multicolumn{3}{c}{Males} \\ % & 50--54& 55--59& 60--64 & 65--69 & 70--74 & 75--79 & 80--89& 90+ \\ \hline <>= cat("Low&", paste(tab1[1,,2], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,2], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,2], collapse=" &"), "\n") @ \end{tabular} \caption{Detailed age and sex distribution for the study population} \label{tab2} \end{table} \subsection{Selection} One approach for balancing is to select a subset of the data such that its distribution matches the referent for each level of $c$, i.e., for each curve that we wish to obtain. As an example we take a case-control like approach to the FLC data, with FLC high as the ``cases'' since it is the smallest group. Table \ref{tab2} shows a detailed distribution of the data with respect to age and sex. The balanced subset has all \Sexpr{tab1[3,1,1]} females aged 50--54 from the high FLC group, a random sample of \Sexpr{tab1[3,1,1]} out of the \Sexpr{tab1[1,1,1]} females in the age 50--54 low FLC group, and \Sexpr{tab1[3,1,1]} out of \Sexpr{tab1[2,1,1]} for the middle FLC. Continue this for all age/sex subsets. We cannot \emph{quite} compute a true case-control estimate for this data since there are not enough ``controls'' in the female 90+ category to be able to select one unique control for each case, and likewise in the male 80-89 and 90+ age groups. To get around this we will sample with replacement in these strata. \begin{figure}[tb] \myfig{flc2} \caption{Survival curves from a case-control sample are shown as solid lines, dashed lines are curves for the unweighted data set (as found in figure \ref{flc1}).} \label{flc2} \end{figure} <>= temp <- with(fdata, table(group, age2, sex)) dd <- dim(temp) # Select subjects set.seed(1978) select <- array(vector('list', length=prod(dd)), dim=dd) for (j in 1:dd[2]) { for (k in 1:dd[3]) { n <- temp[3,j,k] # how many to select for (i in 1:2) { indx <- which(as.numeric(fdata$group)==i & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[i,j,k] <- list(sample(indx, n, replace=(n> temp[i,j,k]))) } indx <- which(as.numeric(fdata$group)==3 & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[3,j,k] <- list(indx) #keep all the group 3 = high } } data2 <- fdata[unlist(select),] sfit2 <- survfit(Surv(futime, death) ~ group, data2) plot(sfit2, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) @ %\begin{table}[tb] \centering % \begin{tabular}{ccccccc} % &\multicolumn{2}{c}{FLC low} & \multicolumn{2}{c}{FLC med}& % \multicolumn{2}{c}{FLC high} \\ % & Total & Subset & Total & Subset & Total & Subset \\ \hline %<>= %tab3 <- with(fdata, table(age2, group)) %tab3 <- round(100*scale(tab3, center=F, scale=colSums(tab3))) %tab4 <- with(data2, table(age2, group)) %tab4 <- round(100*scale(tab4, center=F, scale=colSums(tab4))) %tab5 <- cbind(tab3[,1], tab4[,1], tab3[,2], tab4[,2], tab3[,3], tab4[,3]) %pfun <- function(x) paste(ifelse(x<10, paste("\\phantom{0}", x), x), % collapse=" &") %dtemp <- dimnames(tab5)[[1]] %for (j in 1:7) % cat(dtemp[j], " &", pfun(tab5[j,]), "\\\\\n") %cat(dtemp[8], " & ", pfun(tab5[8,]), "\n") %@ %\end{tabular} %\caption{Age distributions (\%) of the original data set along with that of % the subset, for the three FLC groups.} %\label{tflc2} %\end{table} The survival curves for the subset data are shown in figure \ref{flc2}. The curve for the high risk group is unchanged, since by definition all of those subjects were retained. We see that adjustment for age and sex has reduced the apparent survival difference between the groups by about half, but a clinically important effect for high FLC values remains. The curve for group 1 has moved more than that for group 2 since the age/sex adjustment is more severe for that group. <>= # I can't seem to put this all into an Sexpr z1 <- with(fdata,table(age, sex, group)) z2<- apply(z1, 1:2, min) ztemp <- 3*sum(z2) z1b <- with(fdata, table(age>64, sex, group)) ztemp2 <- sum(apply(z1b, 1:2, min)) @ In actual practice, case-control designs arise when matching and selection can occur \emph{before} data collection, leading to a substantial decrease in the amount of data that needs to be gathered and a consequent cost or time savings. When a data set is already in hand it has two major disadvantages. The first is that the approach wastes data; throwing away information in order to achieve balance is always a bad idea. Second is that though it returns an unbiased comparison, the result is for a very odd reference population. One advantage of matched subsets is that standard variance calculations for the curves are correct; the values provided by the usual Kaplan-Meier program need no further processing. We can also use the usual statistical tests to check for differences between the curves. <<>>= survdiff(Surv(futime, death) ~ group, data=data2) @ \subsection{Reweighting} \label{sect:logistic} A more natural way to adjust the data distribution is by weighting. Let $\pi(a,s)$, $a$ = age group, $s$ = sex be a target population age/sex distribution for our graph, and $p(a,s,i)$ the observed probability of each age/sex/group combination in the data. Both $\pi$ and $p$ sum to 1. Then if each observation in the data set is given a case weight of \begin{equation} w_{asi} = \frac{\pi(a,s)}{p(a,s,i)} \label{wt1} \end{equation} the weighted age/sex distribution for each of the groups will equal the target distribution $\pi$. An obvious advantage of this approach is that the resulting curves represent a tangible and well defined group. As an example, we will first adjust our curves to match the age/sex distribution of the 2000 US population, a common reference target in epidemiology studies. The \texttt{uspop2} data set is found in later releases of the survival package in R. It is an array of counts with dimensions of age, sex, and calendar year. We only want ages of 50 and over, and the population data set has collapsed ages of 100 and over into a single category. We create a table \texttt{tab100} of observed age/sex counts within group for our own data, using the same upper age threshold. New weights are the values $\pi/p$ = \texttt{pi.us/tab100}. <<>>= refpop <- uspop2[as.character(50:100),c("female", "male"), "2000"] pi.us <- refpop/sum(refpop) age100 <- factor(ifelse(fdata$age >100, 100, fdata$age), levels=50:100) tab100 <- with(fdata, table(age100, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab100 #new weights by age,sex, group range(us.wt) @ There are infinite weights! This is because the US population has coverage at all ages, but our data set does not have representatives in every age/sex/FLC group combination; there are for instance no 95 year old males in in the data set. Let us repeat the process, collapsing the US population from single years into the 8 age groups used previously in table \ref{tab2}. Merging the per age/sex/group weights found in the 3-dimensional array \texttt{us.wt} into the data set as per-subject weights uses matrix subscripts, a useful but less known feature of R. <<>>= temp <- as.numeric(cut(50:100, c(49, 54, 59, 64, 69, 74, 79, 89, 110)+.5)) pi.us<- tapply(refpop, list(temp[row(refpop)], col(refpop)), sum)/sum(refpop) tab2 <- with(fdata, table(age2, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab2 range(us.wt) index <- with(fdata, cbind(as.numeric(age2), as.numeric(sex), as.numeric(group))) fdata$uswt <- us.wt[index] sfit3a <-survfit(Surv(futime, death) ~ group, data=fdata, weight=uswt) @ \begin{figure}[tb] \myfig{flc3a} \caption{Population totals for the US reference (red) and for the observed data set (black).} \label{flc3a} \end{figure} A more common choice is to use the overall age/sex distribution of the sample itself as our target distribution $\pi$, i.e., the empirical distribution. Since FLC data set is population based and has excellent coverage of the county, this will not differ greatly from the US population in this case, as is displayed in figure \ref{flc3a}. <>= tab1 <- with(fdata, table(age2, sex))/ nrow(fdata) matplot(1:8, cbind(pi.us, tab1), pch="fmfm", col=c(2,2,1,1), xlab="Age group", ylab="Fraction of population", xaxt='n') axis(1, 1:8, levels(fdata$age2)) tab2 <- with(fdata, table(age2, sex, group))/nrow(fdata) tab3 <- with(fdata, table(group)) / nrow(fdata) rwt <- rep(tab1,3)/tab2 fdata$rwt <- rwt[index] # add per subject weights to the data set sfit3 <- survfit(Surv(futime, death) ~ group, data=fdata, weight=rwt) temp <- rwt[,1,] #show female data temp <- temp %*% diag(1/apply(temp,2,min)) round(temp, 1) #show female data @ \begin{figure}[tb] \myfig{flc3} \caption{Survival curves for the three groups using reweighted data are shown with solid lines, the original unweighted analysis as dashed lines. The heavier solid line adjusts to the Olmsted population and the lighter one to the US population.} \label{flc3} \end{figure} <>= plot(sfit3, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3a, mark.time=F, col=c(1,2,4), lty=1, lwd=1, xscale=365.25) lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) @ The calculation of weights is shown above, and finishes with a table of the weights for the females. The table was scaled so as to have a minimum weight of 1 in each column for simpler reading. We see that for the low FLC group there are larger weights for the older ages, whereas the high FLC group requires substantial weights for the youngest ages in order to achieve balance. The resulting survival curve is shown in figure \ref{flc3}. The distance between the adjusted curves is similar to the results from subset selection, which is as expected since both approaches are correcting for the same bias, but results are now for an overall population distribution that matches Olmsted County. The curves estimate what the results would have looked like, had each of the FLC groups contained the full distribution of ages. Estimation based on reweighted data is a common theme in survey sampling. Correct standard errors for the curves are readily computed using methods from that literature, and are available in some software packages. In R the \texttt{svykm} routine in the \texttt{survey} package handles both this simple case and more complex sampling schemes. Tests of the curves can be done using a weighted Cox model; the robust variance produced by \texttt{coxph} is identical to the standard Horvitz-Thompsen variance estimate used in survey sampling \cite{Binder92}. The robust score test from \texttt{coxph} corresponds to a log-rank test corrected for weighting. (In the example below the svykm function is only run if the survey package is already loaded, as the variance calculation is very slow for this large data set.) <<>>= id <- 1:nrow(fdata) cfit <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=rwt) summary(cfit)$robscore if (exists("svykm")) { #true if the survey package is loaded sdes <- svydesign(id = ~0, weights=~rwt, data=fdata) dfit <- svykm(Surv(futime, death) ~ group, design=sdes, se=TRUE) } @ Note: including the \texttt{cluster} term in the coxph call causes it to treat the weights as resampling values and thus use the proper survey sampling style variance. The default without that term would be to treat the case weights as replication counts. This same alternate variance estimate is also called for when there are correlated observations; many users will be more familiar with the cluster statement in that context. \paragraph{Inverse probability weighting} Notice that when using the overall population as the target distribution $\pi$ we can use Bayes rule to rewrite the weights as \begin{align*} \frac{1}{w_{asi}} &= \frac{{\rm Pr}({\rm age}=a, {\rm sex} =s, {\rm group}=i)} {{\rm Pr}({\rm age}=a, {\rm sex} =s)} \\ &= {\rm Pr}({\rm group}=i | {\rm age}=a, {\rm sex} =s) \end{align*} This last is precisely the probability estimated by a logistic regression model, leading to \emph{inverse probability weighting} as an alternate label for this approach. We can reproduce the weights calculated just above with three logistic regression models. <>= options(na.action="na.exclude") gg <- as.numeric(fdata$group) lfit1 <- glm(I(gg==1) ~ factor(age2) * sex, data=fdata, family="binomial") lfit2 <- glm(I(gg==2) ~ factor(age2) * sex, data=fdata, family="binomial") lfit3 <- glm(I(gg==3) ~ factor(age2) * sex, data=fdata, family="binomial") temp <- ifelse(gg==1, predict(lfit1, type='response'), ifelse(gg==2, predict(lfit2, type='response'), predict(lfit3, type='response'))) all.equal(1/temp, fdata$rwt) @ If there were only 2 groups then only a single regression model is needed since P(group 2) = 1 - P(group 1). Note the setting of na.action, which causes the predicted vector to have the same length as the original data even when there are missing values. This simplifies merging the derived weights with the original data set. An advantage of the regression framework is that one can easily accommodate more variables by using a model with additive terms and only a few selected interactions, and the model can contain continuous as well as categorical predictors. The disadvantage is that such models are often used without the necessary work to check their validity. For instance models with \texttt{age + sex} could have been used above. This makes the assumption that the odds of being a member of group 1 is linear in age and with the same slope for males and females; ditto for the models for group 2 and group 3. How well does this work? Since the goal of reweighting is to standardize the ages, a reasonable check is to compute and plot the reweighted age distribution for each flc group. \begin{figure}[tb] \myfig{flc4} \caption{The re-weighted age distribution using logistic regression with continuous age, for females, FLC groups 1--3. The target distribution is shown as a ``+''. The original unadjusted distribution is shown as dashed lines.} \label{flc4} \end{figure} Figure \ref{flc4} shows the result. The reweighted age distribution is not perfectly balanced, i.e., the `1', `2' and `3' symbols do no exactly overlay one another, but in this case the simple linear model has done an excellent job. We emphasize that whenever the reweighting is based on a simplified model then such a check is obligatory. It is quite common that a simple model is not sufficient and the resulting weight adjustment is inadequate. Like a duct tape auto repair, proceeding forward as though the underlying problem has been addressed is then most unwise. <>= lfit1b <-glm(I(gg==1) ~ age + sex, data=fdata, family="binomial") lfit2b <- glm(I(gg==2) ~ age +sex, data=fdata, family="binomial") lfit3b <- glm(I(gg==3) ~ age + sex, data=fdata, family="binomial") # weights for each group using simple logistic twt <- ifelse(gg==1, 1/predict(lfit1b, type="response"), ifelse(gg==2, 1/predict(lfit2b, type="response"), 1/predict(lfit3b, type="response"))) tdata <- data.frame(fdata, lwt=twt) #grouped plot for the females temp <- tdata[tdata$sex=='F',] temp$gg <- as.numeric(temp$group) c1 <- with(temp[temp$gg==1,], tapply(lwt, age2, sum)) c2 <- with(temp[temp$gg==2,], tapply(lwt, age2, sum)) c3 <- with(temp[temp$gg==3,], tapply(lwt, age2, sum)) xtemp <- outer(1:8, c(-.1, 0, .1), "+") #avoid overplotting ytemp <- 100* cbind(c1/sum(c1), c2/sum(c2), c3/sum(c3)) matplot(xtemp, ytemp, col=c(1,2,4), xlab="Age group", ylab="Weighted frequency (%)", xaxt='n') ztab <- table(fdata$age2) points(1:8, 100*ztab/sum(ztab), pch='+', cex=1.5, lty=2) # Add the unadjusted temp <- tab2[,1,] temp <- scale(temp, center=F, scale=colSums(temp)) matlines(1:8, 100*temp, pch='o', col=c(1,2,4), lty=2) axis(1, 1:8, levels(fdata$age2)) @ \paragraph{Rescaled weights} As the weights were defined in equation \ref{wt1}, the sum of weights for each of the groups is \Sexpr{nrow(fdata)}, the number of observations in the data set. Since the number of subjects in group 3 is one seventh of that in group 1, the average weight in group 3 is much larger. An alternative is to define weights in terms of the \emph{within} group distribution rather than the overall distribution, leading to the rescaled weights $w^*$ \begin{align} w^* &= \frac{\pi(a,s)}{p(a,s|i)} \label{wt2} \\ &= \frac{{\rm P}({\rm group}=i)} {{\rm P}({\rm group}=i | {\rm age}=a, {\rm sex}=s)} \label{wt2b} \end{align} Each group's weights are rescaled by the overall prevalence of the group. In its simplest form, the weights in each group are scaled to add up to the number of subjects in the group. <<>>= # compute new weights wtscale <- table(fdata$group)/ tapply(fdata$rwt, fdata$group, sum) wt2 <- c(fdata$rwt * wtscale[fdata$group]) c("rescaled cv"= sd(wt2)/mean(wt2), "rwt cv"=sd(fdata$rwt)/mean(fdata$rwt)) cfit2a <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight= rwt) cfit2b <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=wt2) round(c(cfit2a$rscore, cfit2b$rscore),1) @ The rescaling results in weights that are much less variable across groups. This operation has no impact on the individual survival curves or their standard errors, since within group we have multiplied all weights by a constant. When comparing curves across groups, however, the rescaled weights reduce the standard error of the test statistic. This results in increased power for the robust score test, although in this particular data set the improvement is not very large. \section{Conditional method} In the marginal approach we first balance the data set and then compute results on the adjusted data. In the conditional approach we first compute a predicted survival curve for each subject that accounts for flc group, age and sex, and then take a weighted average of the curves to get an overall estimate for each flc group. For both methods a central consideration is the population of interest, which drives the weights. Modeling has not removed the question of \emph{who} these curves should represent, it has simply changed the order of operation between the weighting step and the survival curves step. \subsection{Stratification} Our first approach is to subset the data into homogeneous age/sex strata, compute survival curves within each strata, and then combine results. We will use the same age/sex combinations as before. The interpretation of these groups is different, however. In the marginal approach it was important to find age/sex groups for which the probability of membership within each FLC group was constant within the strata (independent of age and sex, within strata), in this case it is important that the survival for each FLC group is constant in each age/sex stratum. Homogeneity of membership within each stratum and homogeneity of survival within each stratum may lead to different partitions for some data sets. Computing curves for all the combinations is easy. <>= allfit <- survfit(Surv(futime/365.25, death) ~ group + age2 + sex, fdata) temp <- summary(allfit)$table temp[1:6, c(1,4)] #abbrev printout to fit page @ The resultant survival object has 48 curves: 8 age groups * 2 sexes * 3 FLC groups. To get a single curve for the first FLC group we need to take a weighted average over the 16 age/sex combinations that apply to that group, and similarly for the second and third FLC subset. Combining the curves is a bit of a nuisance computationally because each of them is reported on a different set of time points. A solution is to use the \texttt{summary} function for survfit objects along with the \texttt{times} argument of that function. This feature was originally designed to allow printout of curves at selected time points (6 months, 1 year, \ldots), but can also be used to select a common set of time points for averaging. We will arbitrarily use 4 per year, which is sufficient to create a visually smooth plot over the time span of interest. By default \texttt{summary} does not return data for times beyond the end of a curve, i.e., when there are no subjects left at risk; the \texttt{extend} argument causes a full set of times to always be reported. As seen in the printout above, the computed curves are in sex within age within group order. The overall curve is a weighted average chosen to match the original age/sex distribution of the population. <>= xtime <- seq(0, 14, length=57) #four points/year for 14 years smat <- matrix(0, nrow=57, ncol=3) # survival curves serr <- smat #matrix of standard errors pi <- with(fdata, table(age2, sex))/nrow(fdata) #overall dist for (i in 1:3) { temp <- allfit[1:16 + (i-1)*16] #curves for group i for (j in 1:16) { stemp <- summary(temp[j], times=xtime, extend=T) smat[,i] <- smat[,i] + pi[j]*stemp$surv serr[,i] <- serr[,i] + pi[i]*stemp$std.err^2 } } serr <- sqrt(serr) matplot(xtime, smat, type='l', lwd=2, col=c(1,2,4), ylim=c(0,1), lty=1, xlab="Years from sample", ylab="Survival") lines(sfit1, mark.time=F, lty=2, col=c(1,2,4), xscale=365.25) @ \begin{figure}[tb] \myfig{flc5} \caption{Estimated curves from a stratified model, along with those from the uncorrected fit as dashed lines.} \label{flc5} \end{figure} Figure \ref{flc5} shows the resulting averaged curves. Overlaid are the curves for the unadjusted model. Very careful comparison of these curves with the weighted estimate shows that they have almost identical spread, with just a tiny amount of downward shift. There are two major disadvantages to the stratified curves. The first is that when the original data set is small or the number of confounders is large, it is not always feasible to stratify into a large enough set of groups that each will be homogeneous. The second is a technical aspect of the standard error estimate. Since the curves are formed from disjoint sets of observations they are independent and the variance of the weighted average is then a weighted sum of variances. However, when a Kaplan-Meier curve drops to zero the usual standard error estimate at that point involves 0/0 and becomes undefined, leading to the NaN (not a number) value in R. Thus the overall standard error becomes undefined if any of the component curves falls to zero. In the above example this happens at about the half way point of the graph. (Other software packages carry forward the se value from the last no-zero point on the curve, but the statistical validity of this is uncertain.) To test for overall difference between the curves we can use a stratified test statistic, which is a sum of the test statistics computed within each subgroup. The most common choice is the stratified log-rank statistic which is shown below. The score test from a stratified Cox model would give the same result. <<>>= survdiff(Surv(futime, death) ~ group + strata(age2, sex), fdata) @ \subsection{Modeling} The other approach for conditional estimation is to model the risks due to the confounders. Though we have left it till last, this is usually the first (and most often the only) approach used by most data analysts. Let's start with the very simplest method: a stratified Cox model. <>= cfit4a <- coxph(Surv(futime, death) ~ age + sex + strata(group), data=fdata) surv4a <- survfit(cfit4a) plot(surv4a, col=c(1,2,4), mark.time=F, xscale=365.25, xlab="Years post sample", ylab="Survival") @ This is a very fast and easy way to produce a set of curves, but it has three problems. First is the assumption that this simple model adequately accounts for the effects of age and sex on survival. That is, it assumes that the effect of age on mortality is linear, the sex difference is constant across all ages, and that the coefficients for both are identical for the three FLC groups. The second problem with this approach is that it produces the predicted curve for a single hypothetical subject of age \Sexpr{round(cfit4a[['means']][1], 1)} years and sex \Sexpr{round(cfit4a[['means']][2],2)}, the means of the covariates, under each of the 3 FLC scenarios. However, we are interested in the adjusted survival of a \emph{cohort} of subjects in each range of FLC, and the survival of an ``average'' subject is not the average survival of a cohort. The third and most serious issue is that it is not clear exactly what these ``adjusted'' curves represent --- exactly who \emph{is} this subject a sex of \Sexpr{round(cfit4a[['means']][2],2)}? Multiple authors have commented on this problem, see Thomsen et al \cite{Thomsen91}, Nieto and Coresh \cite{Nieto96} or chapter 10 of Therneau and Grambsh \cite{Therneau00} for examples. Even worse is a Cox model that treated the FLC group as a covariate, since that will impose a an additional constraint of proportional hazards across the 3 FLC groups. \begin{figure} \myfig{flc6} \caption{Curves for the three groups, adjusted for age and sex via a risk model. Dotted lines show the curves from marginal adjustment. Solid curves are for the simple risk model \texttt{cfit4a}.} \label{flc6} \end{figure} We can address this last problem problem by doing a proper average. A Cox model fit can produce the predicted curves for any age/sex combination. The key idea is to produce a predicted survival curve for every subject of some hypothetical population, and then take the average of these curves. The most straightforward approach is to retrieve the predicted individual curves for all \Sexpr{nrow(fdata)} subjects in the data set, assuming each of the three FLC strata one by one, and take a simple average for each strata. For this particular data set that is a bit slow since it involves \Sexpr{nrow(fdata)} curves. However there are only 98 unique age/sex pairs in the data, it is sufficient to obtain the 98 * 3 FLC groups unique curves and take a weighted average. We will make use of the survexp function, which is designed for just this purpose. Start by creating a data set which has one row for each age/sex combination along with its count. Then replicate it into 3 copies, assigning one copy to each of the three FLC strata. <>= tab4a <- with(fdata, table(age, sex)) uage <- as.numeric(dimnames(tab4a)[[1]]) tdata <- data.frame(age = uage[row(tab4a)], sex = c("F","M")[col(tab4a)], count= c(tab4a)) tdata3 <- tdata[rep(1:nrow(tdata), 3),] #three copies tdata3$group <- factor(rep(1:3, each=nrow(tdata)), labels=levels(fdata$group)) sfit4a <- survexp(~group, data=tdata3, weight = count, ratetable=cfit4a) plot(sfit4a, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, c("FLC low", "FLC med", "FLC high"), lty=1, col=c(1,2,4), bty='n', lwd=2) @ Figure \ref{flc6} shows the result. Comparing this to the prior 3 adjustments shown in figures \ref{flc3}, \ref{flc4}, and \ref{flc5} we see that this result is different. Why? Part of the reason is due to the fact that $E[f(X)] \ne f(E[X])$ for any non-linear operation $f$, so that averages of survival curves and survival curves of averages will never be exactly the same. This may explain the small difference between the stratified and the marginal approaches of figures \ref{flc3} and \ref{flc5}, which were based on the same subsets. The Cox based result is systematically higher than the stratified one, however, so something more is indicated. Aside: An alternate computational approach is to create the individual survival curves using the \texttt{survfit} function and then take averages. <<>>= tfit <- survfit(cfit4a, newdata=tdata, se.fit=FALSE) curves <- vector('list', 3) twt <- c(tab4a)/sum(tab4a) for (i in 1:3) { temp <- tfit[i,] curves[[i]] <- list(time=temp$time, surv= c(temp$surv %*% twt)) } @ The above code is a bit sneaky. I know that the result from the survfit function contains a matrix \texttt{tfit\$surv} of 104 columns, one for each row in the tdata data frame, each column containing the curves for the three strata one after the other. Sub setting \texttt{tfit} results in the matrix for a single flc group. Outside of R an approach like the above may be needed, however. \begin{figure} \myfig{flc6b} \caption{Left panel: comparison of Cox model based adjustment (solid) with the curves based on marginal adjustment (dashed). The Cox model curves without (black) and with (red) an age*sex interaction term overlay. Right panel: plot of the predicted relative risks from a Cox model \texttt{crate} versus population values from the Minnesota rate table.} \label{flc6b} \end{figure} So why are the modeling results so different than either reweighting or stratification? Suspicion first falls on the use of a simple linear model for age and sex, so start by fitting a slightly more refined model that allows for a different slope for the two sexes, but is still linear in age. In this particular data set an external check on the fit is also available via the Minnesota death rate tables, which are included with the survival package as \texttt{survexp.mn}. This is an array that contains daily death rates by age, sex, and calender year. <>= par(mfrow=c(1,2)) cfit4b <- coxph(Surv(futime, death) ~ age*sex + strata(group), fdata) sfit4b <- survexp(~group, data=tdata3, ratetable=cfit4b, weights=count) plot(sfit4b, fun='event', xscale=365.25, xlab="Years from sample", ylab="Deaths") lines(sfit3, mark.time=FALSE, fun='event', xscale=365.25, lty=2) lines(sfit4a, fun='event', xscale=365.25, col=2) temp <- median(fdata$sample.yr) mrate <- survexp.mn[as.character(uage),, as.character(temp)] crate <- predict(cfit4b, newdata=tdata, reference='sample', type='lp') crate <- matrix(crate, ncol=2)[,2:1] # mrate has males then females, match it # crate contains estimated log(hazards) relative to a baseline, # and mrate absolute hazards, make both relative to a 70 year old for (i in 1:2) { mrate[,i] <- log(mrate[,i]/ mrate[21,2]) crate[,i] <- crate[,i] - crate[21,2] } matplot(mrate, crate, col=2:1, type='l') abline(0, 1, lty=2, col=4) @ The resulting curves are shown in the left panel of figure \ref{flc6b} and reveal that addition of an interaction term did not change the predictions, and that the Cox model result for the highest risk group is distinctly different predicted survival for the highest FLC group is distinctly different when using model based prediction. The right hand panel of the figure shows that though there are slight differences with the Minnesota values, linearity of the age effect is very well supported. So where exactly does the model go wrong? Since this is such a large data set we have the luxury of looking at subsets. This would be a very large number of curves to plot --- age by sex by FLC = 48 --- so an overlay of the observed and expected curves by group would be too confusing. Instead we will summarize each of the groups according to their observed and predicted number of events. <>= obs <- with(fdata, tapply(death, list(age2, sex, group), sum)) pred<- with(fdata, tapply(predict(cfit4b, type='expected'), list(age2, sex, group), sum)) excess <- matrix(obs/pred, nrow=8) #collapse 3 way array to 2 dimnames(excess) <- list(dimnames(obs)[[1]], c("low F", "low M", "med F", "med M", "high F", "high M")) round(excess, 1) @ The excess risks, defined as the observed/expected number of deaths, are mostly modest ranging from .8 to 1.2. The primary exception exception is the high FLC group for ages 50--59 which has values of 1.6 to 2.5; the Cox model fit has greatly overestimated the survival for the age 50--54 and 55--59 groups. Since this is also the age category with the highest count in the data set, this overestimation will have a large impact on the overall curve for high FLC subset, which is exactly where the the deviation in figure \ref{flc6b} is observed to lie. There is also mild evidence for a linear trend in age for the low FLC females, in the other direction. Altogether this suggests that the model might need to have a different age coefficient for each of the three FLC groups. <<>>= cfit5a <- coxph(Surv(futime, death) ~ group:age +sex + strata(group), fdata) cfit5b <- coxph(Surv(futime, death) ~ group:(age +sex) + strata(group), fdata) cfit5c <- coxph(Surv(futime, death) ~ group:(age *sex) + strata(group), fdata) options(show.signif.stars=FALSE) # see footnote anova(cfit4a, cfit5a, cfit5b, cfit5c) temp <- coef(cfit5a) names(temp) <- c("sex", "ageL", "ageM", "ageH") round(temp,3) @ The model with separate age coefficients for each FLC group gives a major improvement in goodness of fit, but adding separate sex coefficients per group or further interactions does not add importantly beyond that. \footnote{There are certain TV shows that make one dumber just by watching them; adding stars to the output has the same effect on statisticians.} \begin{figure} \myfig{flc7} \caption{Adjusted survival for the 3 FLC groups based on the improved Cox model fit. Dashed lines show the predictions from the marginal model.} \label{flc7} \end{figure} A recheck of the observed/expected values now shows a much more random pattern, though some excess remains in the upper right corner. The updated survival curves are shown in figure \ref{flc7} and now are in closer concordance with the marginal fit. <>= pred5a <- with(fdata, tapply(predict(cfit5a, type='expected'), list(age2, sex, group), sum)) excess5a <- matrix(obs/pred5a, nrow=8, dimnames=dimnames(excess)) round(excess5a, 1) sfit5 <- survexp(~group, data=tdata3, ratetable=cfit5a, weights=count) plot(sfit3, fun='event', xscale=365.25, mark.time=FALSE, lty=2, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, col=c(1,2,4)) @ One problem with the model based estimate is that standard errors for the curves are complex. Standard errors of the individual curves for each age/sex/FLC combination are a standard output of the survfit function, but the collection of curves is correlated since they all depend on a common estimate of the model's coefficient vector $\beta$. Curves with disparate ages are anti-correlated (an increase in the age coefficient of the model would raise one and lower the other) whereas those for close ages are positively correlated. A proper variance for the unweighted average has been derived by Gail and Byar \cite{Gail86}, but this has not been implemented in any of the standard packages, nor extended to the weighted case. A bootstrap estimate would appear to be the most feasible. \section{Conclusions} When two populations need to be adjusted and one is much larger than the other, the balanced subset method has been popular. It is most often seen in the context of a case-control study, with cases as the rarer group and a set of matched controls selected from the larger one. This method has the advantage that the usual standard error estimates from a standard package are appropriate, so no further work is required. However, in the general situation it leads to a correct answer but for the wrong problem, i.e., not for a population in which we are interested. The population reweighted estimate is flexible, has a readily available variance in some statistical packages (but not all), and the result is directly interpretable. It is the method we recommend in general. The approach can be extended to a large number of balancing factors by using a regression model to derive the weights. Exploration and checking of said model for adequacy is an important step in this case. The biggest downside to the method arises when there is a subset which is rare in the data sample but frequent in the adjusting population. In this case subjects in that subset will be assigned large weights, and the resulting curves will have high variance. The stratified method is closely related to reweighting (not shown). It does not do well if the sample size is small, however. Risk set modeling is a very flexible method, but is also the one where it is easiest to go wrong by using an inadequate model, and variance estimation is also difficult. To the extent that the fitted model is relevant, it allows for interpolation and extrapolation to a reference population with a different distribution of covariates than the one in the training data. It may be applicable in cases such as rare subsets where population reweighting is problematic, with the understanding that one is depending heavily on extrapolation in this case, which is always dangerous. \section{A note on type 3 tests} One particular software package (not R) and its proponents are very fond of something called ``type 3'' tests. Said tests are closely tied to a particular reference population: \begin{itemize} \item For all continuous covariates in the model, the empirical distribution is used as the reference. \item For all categorical adjusters, a uniform distribution over the categories is used. \end{itemize} Figure \ref{flc8} shows the fit from such a model. Not surprisingly, the predicted death rate is very high: 1/4 of our population is over 80 years old! The authors do not find such a prediction particularly useful since we don't ever expect to see a population like this (it's sort of like planning for the zombie apocalypse), but for those enamored of type 3 tests this shows how to create the corresponding curves. <>= # there is a spurious warning from the model below: R creates 3 unneeded # columns in the X matrix cfit6 <- coxph(Surv(futime, death) ~ group:age2 + sex + strata(group), fdata) saspop <- with(fdata, expand.grid(age2= levels(age2), sex= levels(sex), group = levels(group))) sfit6 <- survexp(~group, data=saspop, ratetable=cfit6) plot(sfit6, fun='event', xscale=365.25, mark.time=FALSE, lty=1, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, lty=2, col=c(1,2,4)) @ \begin{figure} \myfig{flc8} \caption{Adjusted survival for the 3 FLC groups based on a fit with categorical age, and predicting for a uniform age/sex population. Dashed lines show the predictions from the marginal model.} \label{flc8} \end{figure} \bibliographystyle{plain} \bibliography{refer} \end{document} survival/inst/doc/compete.R0000644000175100001440000002526612545056257015472 0ustar hornikusers### R code from vignette source 'compete.Rnw' ################################################### ### code chunk number 1: compete.Rnw:24-30 ################################################### options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=10) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #ensure default require("survival") ################################################### ### code chunk number 2: check ################################################### cmplib <- require("cmprsk", quietly=TRUE) if (cmplib) cat("\\newcommand{\\CMPRSK}{}%\n") ################################################### ### code chunk number 3: sfig1 ################################################### getOption("SweaveHooks")[["fig"]]() par(mar=c(.1, .1, .1, .1)) frame() par(usr=c(0,100,0,100)) # first figure xx <- c(0, 10, 10, 0) yy <- c(0, 0, 10, 10) polygon(xx +10, yy+70) polygon(xx +30, yy+70) arrows( 22, 75, 28, 75, length=.1) text(c(15, 35), c(75,75), c("Alive", "Dead")) # second figure polygon(xx +60, yy+70) for (j in c(55, 70, 85)) { polygon(xx +80, yy+j) arrows(72, (5*75 +j+5)/6, 78, (100+j*5)/6, length=.1) } text(c(65, 85,85,85), c(70,55,70,85)+5, c("A", "D1", "D2", "D3")) # third figure polygon(xx+20, yy+25) for (j in c(15,35)) { polygon(xx +40, yy+j) arrows(32, (5*30 +j+4)/6, 38, (54+j*5)/6, length=.1) } arrows(38, 2+(55 + 35*5)/6, 32, 2+ (150 + 40)/6, length=.1) arrows(45, 33, 45, 27, length=.1) text(c(25, 45,45), c(30, 20, 40), c("Health", "Death", "Illness")) ################################################### ### code chunk number 4: mgus1 ################################################### getOption("SweaveHooks")[["fig"]]() oldpar <- par(mfrow=c(1,2)) hist(mgus2$age, nclass=30, main='', xlab="Age") with(mgus2, tapply(age, sex, mean)) mfit1 <- survfit(Surv(futime, death) ~ sex, data=mgus2) mfit1 plot(mfit1, col=c(1,2), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Survival") legend(6, .8, c("female", "male"), col=1:2, lwd=2, bty='n') par(oldpar) ################################################### ### code chunk number 5: mgus2 ################################################### getOption("SweaveHooks")[["fig"]]() etime <- with(mgus2, ifelse(pstat==0, futime, ptime)) event <- with(mgus2, ifelse(pstat==0, 2*death, 1)) event <- factor(event, 0:2, labels=c("censor", "pcm", "death")) table(event) mfit2 <- survfit(Surv(etime, event) ~ sex, data=mgus2) mfit2 plot(mfit2, col=c(1,1,2,2), lty=c(2,1,2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") legend(20, .6, c("death:female", "death:male", "pcm:female", "pcm:male"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') ################################################### ### code chunk number 6: mgus3 ################################################### getOption("SweaveHooks")[["fig"]]() pcmbad <- survfit(Surv(etime, pstat) ~ sex, data=mgus2) plot(pcmbad[2], mark.time=FALSE, lwd=2, fun="event", conf=FALSE, xscale=12, xlab="Years post diagnosis", ylab="Fraction with PCM") lines(mfit2[2,1], lty=2, lwd=2, mark.time=FALSE, conf=FALSE, xscale=12) legend(0, .28, c("Males, PCM, incorrect curve", "Males, PCM, competing risk"), col=1, lwd=2, lty=c(1,2), bty='n') ################################################### ### code chunk number 7: mgus4 ################################################### ptemp <- with(mgus2, ifelse(ptime==futime & pstat==1, ptime-.1, ptime)) newdata <- tmerge(mgus2, mgus2, id=id, death=event(futime, death)) newdata <- tmerge(newdata, mgus2, id, pcm = event(ptemp, pstat)) newdata <- tmerge(newdata, newdata, id, enum=cumtdc(tstart)) with(newdata, table(death, pcm)) ################################################### ### code chunk number 8: mgus4g ################################################### getOption("SweaveHooks")[["fig"]]() temp <- with(newdata, ifelse(death==1, 2, pcm)) newdata$event <- factor(temp, 0:2, labels=c("censor", "pcm", "death")) mfit3 <- survfit(Surv(tstart, tstop, event) ~ sex, data=newdata, id=id) plot(mfit3[,1], mark.time=FALSE, col=1:2, lty=1, lwd=2, xscale=12, xlab="Years post MGUS diagnosis", ylab="Prevalence of PCM") legend(4, .04, c("female", "male"), lty=1, col=1:2, lwd=2, bty='n') ################################################### ### code chunk number 9: mgus5 ################################################### getOption("SweaveHooks")[["fig"]]() d2 <- with(newdata, ifelse(enum==2, 4, as.numeric(event))) e2 <- factor(d2, labels=c("censor", "pcm", "death w/o pcm", "death after pcm")) mfit4 <- survfit(Surv(tstart, tstop, e2) ~ sex, data=newdata, id=id) plot(mfit2[2,], lty=c(2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") lines(mfit4[2,3], mark.time=FALSE, xscale=12, col=2, lty=2, lwd=2, conf=FALSE) legend(15, .5, c("male:death w/o pcm", "male: ever pcm", "male: death after pcm"), col=c(1,1,2), lty=c(1,2,2), lwd=2, bty='n') ################################################### ### code chunk number 10: cfit1 ################################################### mtemp <- mgus2 mtemp$age <- mtemp$age/10 #age in decades (easier coefficients) mtemp$etime <- etime mtemp$event <- event options(show.signif.stars = FALSE) # display intelligence cfit2 <- coxph(Surv(futime, death) ~ age + sex + mspike, data=mtemp) cfit2 ################################################### ### code chunk number 11: cfit2 ################################################### cfit1 <- coxph(Surv(ptime, pstat) ~ age + sex + mspike, mtemp) cfit1 quantile(mgus2$mspike, na.rm=TRUE) ################################################### ### code chunk number 12: mpyears ################################################### pfit1 <- pyears(Surv(ptime, pstat) ~ sex, mtemp, scale=12) round(100* pfit1$event/pfit1$pyears, 1) # PCM rate per year temp <- summary(mfit1, rmean="common") #print the mean survival time round(temp$table[,1:6], 1) ################################################### ### code chunk number 13: mprev ################################################### tdata <- expand.grid(mspike=c(.5, 1.5), age=c(6,8), sex=c("F", "M")) surv1 <- survfit(cfit1, newdata=tdata) # time to progression curves surv2 <- survfit(cfit2, newdata=tdata) # time to death curves ################################################### ### code chunk number 14: mprev2 ################################################### cifun <- function(surv1, surv2) { utime <- sort(unique(surv1$time, surv2$time)) jump1 <- diff(c(0, summary(surv1, times=utime, extend=TRUE)$cumhaz)) jump2 <- diff(c(0, summary(surv2, times=utime, extend=TRUE)$cumhaz)) dA <- diag(3) prev <- matrix(0., nrow= 1+length(utime), ncol=3) prev[1,1] <- 1 #initial prevalence at time 0: all are in the left box for (i in 1:length(utime)) { dA[1,2] <- jump1[i] #fill in the first row of dA(s) dA[1,3] <- jump2[i] dA[1,1] <- 1- (jump1[i] + jump2[i]) prev[i+1,] <- prev[i,] %*% dA } list(time=c(0, utime), P = prev) } # Get curves for the 8 cases, and save them in a matrix. # Since they all come from the same pair of Cox models, the time values # for all curves will be the same # The cifun function above is only designed to handle one of the 8 covariate # patterns at a time, but survival curves can be subscripted. temp <- cifun(surv1[1], surv2[1]) coxtime <- temp$time coxdeath <- coxpcm <- matrix(0., nrow=length(temp$time), ncol=8) coxdeath[,1] <- temp$P[,3] coxpcm[,1] <- temp$P[,2] for (i in 2:8){ temp <- cifun(surv1[i], surv2[i]) coxdeath[,i] <- temp$P[,3] coxpcm[,i] <- temp$P[,2] } # Print out a M/F results at 20 years indx <- match(20*12, coxtime) progmat <- matrix(coxpcm[indx,], nrow=4) dimnames(progmat) <- list(c("a=50/ms=0.5", "a=50/ms=1.5", "a=80/ms=0.5", "a=80/ms=1.5"), c("female", "male")) round(100*t(progmat), 1) #males and females at 20 years ################################################### ### code chunk number 15: mprev3 ################################################### getOption("SweaveHooks")[["fig"]]() par(mfrow=c(1,2)) matplot(coxtime/12, coxpcm[,c(1,3,5,7)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, ylim=range(coxpcm), xlab="Years", ylab="Progression to PCM") legend(1, .23, c("Female: 60", "Male: 60", "Female: 80", "Male: 80"), lty=c(1,1,2,2), col=c(1,2,1,2), lwd=2, bty='n') matplot(coxtime/12, coxpcm[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="Progression to PCM") ################################################### ### code chunk number 16: finegray ################################################### if (cmplib) { temp <- mtemp temp$fstat <- as.numeric(event) # 1=censor, 2=pcm, 3=death temp$msex <- with(temp, 1* (sex=='M')) fgfit1 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE)) fgfit2 <- with(temp, crr(etime, fstat, cov1=cbind(age, msex, mspike), failcode=3, cencode=1, variance=TRUE)) cmat <- rbind("FineGray: PCM" = fgfit1$coef, "Cox: PCM" = coef(cfit1), "FineGray: death" = fgfit2$coef, "Cox: death" = coef(cfit2)) round(cmat,2) } ################################################### ### code chunk number 17: compare ################################################### cox.f <- log(1- progmat) #log(1-P) round(cox.f[,1] / cox.f[,2], 2) ################################################### ### code chunk number 18: finegray2 ################################################### getOption("SweaveHooks")[["fig"]]() if (cmplib) { par(mfrow=c(1,2)) fdata <- model.matrix(~age + sex + mspike, data=tdata)[,-1] #remove intercept fpred <- predict(fgfit1, cov1=fdata) matplot(fpred[,1]/12, fpred[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), ylim=range(fpred[,-1]), type='l', lwd=2, xlab="Years", ylab="FG predicted") legend(0, .22, c("Female, 60", "Male, 60","Female: 80", "Male, 80"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') matplot(fpred[,1]/12, fpred[,c(3,5,7,9)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="FG predicted") } ################################################### ### code chunk number 19: timedep ################################################### if (cmplib) fgfit3 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE, cov2=msex, tf = function(x) log(x))) survival/inst/doc/tests.pdf0000644000175100001440000125654112545056257015553 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 3 0 obj << /Length 1164 /Filter /FlateDecode >> stream xÚíXÝoÛ6Ï_!ìI*VüŸ'Û€ËP †aíƒf«¶;ÎlukþûÝéH©¸Às$QÔñîø»ãG]6¯Ðu&½PÒš¬ùI„Q*sÒ TÖ̲ßò·…rù¦ù\áZµpë—pÛÜÂm÷ª(uUåWزŸø}½)Jeótw«\‹ƒfÅûæÍÀ¼FóªªE¨\V±áwR›Gbä¥um“PïÉ;´Ö¡ ™ÿH/xwp?`Êj!÷†¾9dEV"È3Π߱¤ ™”"X«P²”^ eeVªZÔAòˆ¦¨aâ[†×=ø]ãÃåÍ¢ˆß‹®%`I³±C͆ÚÖ`@‹`jÖ{ Âí=`«-`¾ZÞ.§‡œª„3àÚø†b÷à‹|ÅO]\ÒF<äÐÆ¥É9áÏJ£÷Àº®6ä»– ·"8`?ˆà”#7ÀpJFçeQ Py¿¥Œ€Q’ {Ó~ÉšASf*Ðt ¢Î2õŽô*ÖûŒ]2¤[Ð[«¼½‹‰»Ý ÉO(°F“!ˆj”ä´¤vOñÁ¶ŽiMcé}=r2»”µÖ&×äÞ£ùjöë;²Ø²‚ëÍP?1CÚÞ"ÒãÅÈ+‘&ÃA‘¿ÐãnÛÎi"4Ilй¹éY¹=ßz8Ó@ÝÑç£þ°]ë8wœõ 7 Ë‚$¥¬uÑ™˜V  ow˜Éß…òy‡Šç‹BÕ°êIu€O¢(MGç‘Ó%¿^e²¶’1ÆØ2 U{D¹n¾Á¥‡pNÛ#Lk3eFÝÈð$´ãÏííŒSàü€m‹©CQCŒÿµÙü÷ ^UŽâ¥#^¸p—ï*©_wqö ÈnO«A»vŸÿWj®LjHB©Ž¢j"ª7Hë˜Y›ødZüI¤%œçÝ]Ü,¬ÍÿÀïÓq ‚šÙϯȧ$%îæòø¶º_ ½hcþ-ÓúH¡9¹áîÉ2•DŽö1Ðß‚1}NYÉéxS7Ä”Y€[0ˆ]â6ü°fÚd ¨ÿÎ'Q@Ø;sKܨôÊ´«Ç…Ås5Ô.VCF(¨V«T+JfèqÕ ‚ÐzO7, òÖA‰¡ÚW-О2m@°ûXž·»(jd8?úæ ŽwC76U7‰a°XXMÏÑHÙÀÞ«ØU¨Î¯‹"ÄçÊâ7,0®hûš"jw‘m « ÁžZ§²§Ϥ(ƒ®âQdŽ'T2 äóÏOz¿zÛwéL˜H$î ¥´¯øÀlŸ9ÎÄXO"/y¯óŸoxixÁ¬;\$ æAª!Ë`ôÙFw´jîØ“HDˆä ¾i°´/ç¿/byUW3¸)Kó/h~Qâêê(¶†±½¼§¿ˆXpP9ÒÅòeÈú%‡«Ç'¢' µ é5—ãp ™‹<úé¼J¥ˆ<¿¤|– ÝS¶ÓKàØ-÷§Ìš~~H*ä÷[ëøŸª >à_QÜÿ>Û+Ue…v6+=”ü2Õû vñ}sñ¥. Ó endstream endobj 2 0 obj << /Type /Page /Contents 3 0 R /Resources 1 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 1 0 obj << /Font << /F38 4 0 R /F19 5 0 R /F45 6 0 R /F47 7 0 R /F52 8 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 13 0 obj << /Length 3844 /Filter /FlateDecode >> stream xڽˎã6òž¯hä²n`¬õ¢”œ²A˜=AÆÀ.Ìj[~$–åHöôt¾~ëIR–ÜÉ!›C·E²X,ëMþsõÅÛw™}0Y”fEò°Ú>T6²EùPä6JÒüaµyøia—Æ$ùâý£YœS³¸ÀGß=.Óx±¹®±uèhìãê_oß•UTIèâ‡eb¢2.Ój°ð7ÀäÔ.NNn¸1<2æºçN³ØðÀóAFöÜ®Oü{ 5“’Æšþ­»À5êÓN°v\,.{¿Ò˽¬p<ÿèq™åÅâ,W$ÇzYmø×z“,~6i†:\í™gm¨MÞOzjî¹ žè>žB‰ð$ø]´Ÿá‚ìDU¹œQG߬M™/ «D}ævÛV¤¤9¿/ñßÀ`™"™ÃøžzÜäù(gÔl‚§—q[ϺžqÑZx¨ð‘жmƃÄlú#˜ ¶ŽMvr %‹ß®´o9lDßàlò‚EoËS÷8õí Ç =ë'7sü¢½ãBÈ2€O*8Êϸ’[$é…{Ÿq[¸Gl°ÄI¿ì !/\» ‘È^{Y«Þ"&'C¬s?ëÄ@xøUÅ“y/"l0vä3ì~õ„9™ïxžªÊqš,Vpº¢„•^3F›éÔ–Éî4ÐÀ '=Š˜Q Lš“>|€Žo>ðà†InX>’Â8 Y‘NS¡¾áØ®'}å=ñ|œYoTOB*ÀŽh8 €9­jö‡Û, I‰ìþ]Õ#´´èAxß$ÁI^-vÄô½pc{•ãtçØóõ¬Ê:^%¥ÕïäYÕßà9‹'¤KÀTþà³ùDæÇúµ>‚ G5M¥— ø¬ù§ÅU®¡àÚøNI/¹âÕÕÀ]íu‘!ù×2b!á\+;d/Ïd Öˆkïøymá_KÓ¿³ YW0ë˜*0WsQM´Q*:R,4 Æä[È 4 Q÷M0•uÒ ®=CIþŽÜMB@âGËäò›é(:•€“ “Ö^¾J²BÏø¬šÌ,ìâ-ðgݵgAC.ï"fQYo•þYlkÄrV-A—¦ §8´‘ž;[§-´¿#yàäfÃ×…Ö;¼ã¼q03Hé4Nˆ(ª,»RDpXb‘ûAØ*–¬uýíCæYƒâ1{5¿çÙt p¢^§JGZaÐ0[§ PUžT8æ202öŸ„ ¿ðç–¢ V΋méE8”µQA¨¹M!Û‹å·ôŠdqîYAøT׈°0´|×3DÍ?´Óg™4B'b¨0Å‚1š5kµë¤XK,x{fk«ºîŒ6bgwÞ‰S]Aì·°–Ùäõó@ºôIhùô‘LtÜ£$êQT̽Y•plIÙ1ž\XyZ« WPW ]zwjTƒÒ,–ð±ôF\Ý™Ý*¡ÐÙà‘RJLJÝ&°­ü^5¼¬ÀêÓ99'}?Á°I?Âÿ¯‘ fn³îv'!õ÷Pn³x–€˜£GE›V7›~yÏE¹À‹ô#YÔjJ+Ñ“VƒÃî¬Ù>{ùûÃH>*vø;Š&ƒvÓÊZ>–¬DYÄ‘Ðz¬Ú@–& gn«ÌUÆy'üFnŸ6ê]‰ˆ‹ D$FÜqFmKÎö ÊLV -HËšz#~:Ô[ë¢tì~U•€4≅L•žÑ ™ÜÐùžÕ£mÄ*nˆPFV˜ %KlÅÖÀî8¬“ðOa’¾%6ªÒLA_ñk>°ƒŸU°Ù3n’²|Fqœ¥1eç>Á½t;ßµ&öÃ×&—ièÃjþù¤‰\ÿÂÎ.Ý$&¸õ<\Ÿ6ÞKžÀôçxÄŸüI­’¢‰gGö­ùc/˜å9[±u'~{¼6x(1ÑøyŸ%BüQ@4¶;¸žd¡.ÀGâÕ¶´ÏìÊ-GWI\°‹öN€g©‹Ë<ј¿ºÕI½ª+D ;$Pq5Zøþþú…ÕWÜâZ)| àÔÛFáÅÊ…âô¦ŒD›?—n›$Km“ßDÜÆø¹w5m“{²䦷õeA²%‘q…'Ѝ ,nä’)F:ÍQa²u­I4-y–ä³p€ã9ÕAJÚõæõ\{¯5ZªŠÆ®^UÅ\I«8qè¹rÆš´ç[[C\‰6æzc3§yʼnùûÜPœ9 îU¦“„.€^ÑÖ8²‰siÄ€_¥RI»ß«Zn]!·î¹gFv¨»v¡´·þ…Ç$i#·edÒêÿÂWס5YjÒÒŽr,6ê¬G×Ufñ­¯™É6Q`³Ô+'€Py§#ÃKÂäSŽ‚Úƒ™Ü¾¨³ó ‚ë œGVã7Ôˆƒ º ¯ uq hí2—Á˜Ñ)‡ÓÌF>ŠúnÊZÈ 7þSóùLÖvFÞLT©þgöŒ¼<¢õûÈt¥ªˆ2s!Ü_×Þ*ÿÓ˜£<²…ý ÿˆ”£#·éâÝ#VŽÑ–Ú‚/ËŠ@xì­é› aÓ(Î3¯£K›Üpc#›âÏ×£›2àZUÔ Ž(-Êtm˜²adðÜmÁšá]t€(äædÎôS_3}6*Ó4 @ËBoØ÷/z“sá6… >ãΓxkÉ!ű¬˜<²¹«µ/‰KE|Ë¿Ï^ wjj:®£ñÔ}í?ö"üf}vÜmü] ÷®ÎÄõ,î­[rKøù"÷ˆÔxÿàæ5櫟ƒòìfá¶ÖÌ-­’ð£(²mb‹"8xBúÆÕ]*ô…ë‰ñ‹Ë›KÿHêUqø„ ŠAH– §U¢†71ñlŠU8×ûÈ,+f©,³XeÂÁIq“eÁ0†>jž å/UŒ* ™%9È;…`žFÙQ 2ñ8° Õ‡¢ì§àNèÀhÍšÒ ”Ù…ÜÒèÚÕ!!Rר§>Þ­+øŠ!®ÓõRu¡@*%¶ùZÇN‚¤%¸ˆ¡ó–‹ÓAs˜at¥ó5…+ÀÖm^|R7ÖXKê5à½úˆãé}M|›ÉÅ1g ø{Ö¢QW»’SL™ÞLe<+÷ÎÛ:(%F5s@–˜Ü£ ³Úi ­åÛç*H‰5Q ^ÄÜvƒ,…Ç (‡Œæø¦×ÆZ7@P_P޽Mí±{ R8®R]ôjE.Ç’râIo¤'Äqq)b²à+Èz|áýòÂ#mÖšp&’q |Ó¹Ö;iq®6†hõ^fïý!Ê+x;*/#ÌÆ‰aëK##5/1ɲ ‹Ï™¯jÀ'"˜’ã—:¸\ZÊv†o Â-q~š” ì¦õD™u¨å!ùÂ3%­ð õü}Sy--OrH0Êâõ´¼*KÐ[Y\†oÍh]yì1"’›wo 8³¶ôÛlêÒíæŽSXâçŸö .Æø*ÎSRèÕÕü5epã–y_äu™«’V’ yÒÖè $•_ÒÒûÕ üï½ÌýM §‹¡„†.ôγ²’/Qìœ!¼sµ5ð“—Ú_€jb]„д7—L¨Á{´°òä%W¾ñcIbϤ¬}³Ä( —±Òá¹Ô:‡(ÈT1ˆîrQ6öݦhoî>âÍKóDjùyêbÊnh4pÎ%™r"Fu¡|\ìÍÓÑ›y56øQ:7J ©(wðö—_"á2Ó{ š*­÷?À¿“Î Ìy­‡Bëä©ó¤ðu¤>ì: žËáý¤þ¸Öx§‘ðؽ耨™ï6vtÿJqÁ4ºªö]û`xR2ø -ƒKÈãêžNM{ãšµÛ;/øvÍe⣿å\š<†D-¹S~„œÍÑ[$K{B¶óý=¢aܱ|à¶cD\Nie Ÿlµ7/’dÞlHXóeÄæ‘±WUÎ`ŽS…¥´Ži Ÿ‰5ò |2}áQ(¿ä…Äëi›dlèÛ1*¢Êõiš˜×ËEeúz”Ç.J¸¹¨&hJ÷Ô3k3~nÍïQNV«[¨áIˆ h~Òƒ¼¶)}ñºT¬ÃèQËÆ×jI ?Æ¢$˜@]JÉb;W{ÿdi³¨¨ÌLŒâ³þž£²SÛNÞ8¹½=|Б,ªZÙ×Üx/…†C²§©^oÁoê‹s0J¼Ì=Ç‘û‚ƒ¼™AÉ |ðX{›H=,Ê8Êòòa™ç‘Èn‚_|·úâêȆ9 endstream endobj 12 0 obj << /Type /Page /Contents 13 0 R /Resources 11 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 11 0 obj << /Font << /F47 7 0 R /F8 9 0 R /F61 14 0 R /F11 15 0 R /F60 16 0 R >> /ProcSet [ /PDF /Text ] >> endobj 20 0 obj << /Length 3882 /Filter /FlateDecode >> stream xÚ•ZÝã¶ß¿Â-j·ŠH}EKškS$Ú.‡Ü=pmyíœlù$;{Û¿¾óIR²÷‚<ìZ"‡äp8ó›ꛇ»¯ÞåÕÂäI–—vñ°]4UR•õ¢,ªÄfÅâa³øeiW÷ÆØbùÃÊ,÷ðwlüV÷Y•/ññtúUf—Ÿ±÷€-g$í‘v²ã êÙµüü-<Ã@³üÌD‡~Óå¦íVþõÕ»zÑ$MiKd,]Ü[“ÔiÍ<ýD| yºÜö|ÛÀ?wû#öÁï–i+[-_øy f¹nqGnÚ¹‘ÎÄ ÝÈïó^6´’<·Ü7àc7[Ø,G$ÙŸ˜þquk·4‰­—ÏÈI‹¯Âb§bÅ¹Ü üö4jƒtH02­C¢ ?¯{œé3ÊŸ˜3Ëņ’2&i 9½e·€Yzœ ¥uöB*Œß´<î„ú[Y^C¼aêGÞ>á|³ÍsïÚÉì¸ÿR9pЉÄ6â¿#Ї·&p2\G«ðé´üB¿á0ü‹Ì‹$ÏÔŠýO;d]ÇÉÔ —=ßß’ÔøI„ãˆKyMÂúG9‡öðšiEÍü¦Õòr¢ín\¼xªÇŽì#o¸g"?r¿ÿÌ-áˆaR÷¯ÈÒèe”ŠVœ¼Z‘æ«$‰à·UQÀ.DXŽ˜oeú3OÝ™’ xAã0wS@á˜Q6%¾ÝY´¯EÕ5•×çtÜæð§^þšø‘›¦ …Ê{Ÿg5«ë^Äd[`ÒоLEË8éfÕ¹e`Ép6V”•(ÞF–Æž‹œ(YÂI— h£RR:îFðKvG²ÛŒ¶çíyIÜkÌÎ2Öìè<àÕr/¸zå¹c­å—ýµ9茄vÖDŒÐëûÔä4žPµ4¬š&K²ªf‰ÉŽ6¶Ò¤#¸‚a›'uY)ù¯h0ž5s}ôÐæø‡¬ý±ý1>Ñk(C*˜…­Ièà$êÄæ¥È¾d>ÁJ˜•_D·šL¹jì-óšýFPƒÞÖ4C'ª ܰ—§eÚ@¢zÈÊ:p£jM“O z{F‘~˜ìÙuLBfð?Zl#Œx¾à2„;ñz¸=rΪè/·TôYã¬Êøc³HÈ~'!¢ØÀéÓù“ÿš²)h <Ò`šÃèýD!MžO9úÛælŸ6k¹oÐ[4àÖ¬]Üu@Õ±¹˜` ÎBܤ ¾QñŸ¯& d,zÜxYÓ‰!‚™lù †‹¤?¯ê-¬ ´¬ÌúF ëÈ’ý±å}´èäÝ!€iú-y‘eQ”Ï[d—dMM½`ž/âC@{Ñ”ÊtùLлÆÎ*ƒõæ3 ­z ï‹Mº¡‘ ZHG¢FKÁóº÷ˆêR®ÕŒŽêÊŽLŒ_™- vU<"5»Â¬&©ROÂØÂ+ûs¯ÞfÝðÔòdÕ¢Lš*«q²*KòÜ.²¤4"ss½bžä%@PDÄ*úaá°ìôÐÞ¢ÞXÁ– Ølôðñ´ÄC!±œ HøË:¦}Gî gú<¢!™o»Qr£„²¬L¹õfp˜ÛÆ]¼|½º/1Z¥ð%øˆ°PÈîq¢Ö3H"…ëÏ¢‡3øð¨<êÜŠa‡]@þZ$]ñÔ%Y]ÎÂ" "‹‚ß™q` R‘" +úËê³D2 þF&äCmx™‡Ú­Ý(”l»±¨ÄÉ‘bŸ”…1†pk£pÂ^eà{çk¢ãõþ³9òº¥5Ë·[„𔎶n¯møïCdÀ._ÔJ¬”‹'væßË,b |ŠÐ@Œ†€[bMëÍ)øg‡¹ãÂé°Â íú:LöA°;Ï9#=±×œzµë£„FÇ?׸¸y|;2m@c B h¸gԠ󤸚ñö~æ’‰e·¹G,^=«ú<$$Å[õ²ó|’Ù¤L%d’êÿØy;Ö?‰Þ~/+'w‡J–7œ u‹ä5È#$oh“gÁaèê/?l8r½GÛBß$á‚ù½BCß4¶Ön /¤ðët°ÇÅû•Nƒ¥³9LJ`xnÒûÆh2Íqtû¡¢L@Üêe¶ÂÊZôÕy ¹öi’Z£¹ˆáØ[d LZäS3ßRäЭUTΟáöš€çÕ$®„Wj}ÖèX4œ£ÁÏxpþÔJv¾UW‰v‡x`ÉÎ’Ó±Þg„ýð½g £%7ý10Æì×÷®—Fo§QŸÒêÒçWÓ-Ó4À„Ò'ݺ‡³˜Ìò§N£QŸi[,\BBÀ .™ÇjòG’,©·À {Ò Š”[zÁľÛxø2„?ù~¶†ÒP³‘ÿ `æ¢%Æ6€ë ¸!˜ZàèÏøq3STI­Â*mEÆ"Õe#àm¡”ZhÝE£!rB!¬žh‰s\É‹™,zR<´”œ&ápïË?RóSKÞ+‹¬øV¸1ps}4)Ð2\séWš’¹j"gÏ”ÍÒÖd|ˆš"d~%çŽlrh¯çq‡Xû¨I—#Aøvp<1í ä*ž:*[‘Þ«®ŽLíOFFž_Sš5eD²ïã¤b‡˜XVæh k©&O›?ÀOnR•ãÈÉ’>:MÃñ×ñ:DÇn™ŽŠs膆 Qpêw‹ꫵB¨£ ¬óC åfþÜ »¾?rÚwôŽÝu,ÓÁ.‚·Ã ¸]ÄrÔ|öذ}³hCÓ_ )| «HC¶¼ªj¶Å›¥â 6Î6k2ŸÛÑ–QÎEJÄxÖt ›²$ ë¶óUë*ÅU—M_塌†Î÷ÚUœùôHÀ‚–mkóØ+*V–Œ¶x­àLUÊnÚqÄ|å Ë:i2«j¨¥V°lg àev#€°¹ÅwŸ%"hÇ1Ô±+õƒÓ^¹gò„mœd|°\hêÄåÌ€ça!œ{Ÿé;¬Å~KÉva¸ _ô ÇÜ_¼æh3©`ê^øâøç@Nác#[I p°/ÌÒ>ž®½1E<ªtµy &²2ÊñÅ;í <ó»UÍî­•ñWF;xEW9v õ; ±£¤clÄ`1ò~˜‡ù£òdV½.2ñv@ËÞ mü$©€‡g ê@/`½GŸM¸7ÜñìYÂt6¨Ö‚E9ÜP;«Ç…4ãr`2Ò­k’hƒ¯—æÞRJ¡LáR«[?ð*Ë×äׄŒÐDUÞ6®~õÑl#?>ï´Ð´uÂÖ\¡ÐT&ª´\$º5hõZ_×sh2É€«r UEs­R„ð"…/¹Ø/oý-4l«§Ù[%5‘µè%,­Ì6IòÀJ]Ø¿¶;Jk- ø+UŒ¿.3”Éj©ùT¿ê&/tn[ æ„=þ}j2½¾À[¨‘ž <&Ã\L†¾#;$™¿ñ.;+”=WEàaZ’ÆvG›î·$3³­è˃"ºb,øþ‘‚öZàWƒ”LžÚpß›,UEÊèƒTõ—¶…ÜöÁ¯Ì“ÊgøI)²ëü­š4Þ<„ñkʆ†1¬Õ2žËtO¤..L⪠Ÿ·îúûPëµð©¦7*kð^2‰€¹4‚ªDú³ž4*ÄÍjrÃ>¾ûÑ{±çÞgg_¬•fÙ„öa^è‘gWH%L™—NôgqÊü¥k¸w y½=pÖàOÁߺKz¹#† 8«t )êÂcY]DAm]\GiØ8ÍV°eŽªµ\'Õˆ"¯ O-ùs]\)ÁxßaQ‡NY˜ëô+$Ó›öº”Üc”!ÁÙ+N: ƒë뛾惞‹ÑÏÛÊò@Ö!W–ð„ƒs—â8:ıgе~óàžH¤Ó[Éë) À¥‡XÊÜr‰¯p†qèÒµª$Þ¯´Üb¿„Û+>êµV´1t³Ø:§B#wMõ ‰bG¿·üJ²Ïüì“/º¤T‡Ê7Á²’šÞzðŸñI¥ùãÄí‘ÁùÌ6<³X‡äÚúN«_Ã+>Mï¥ÀfÉPz¹~wÌdc7_ÙWíÔ© ò¯1ì…|5XMf#¥öS@š~rá‘'üÌ£ÿâr¼ÝV!¹QD7ðh1…ŒÛV,†¿ï·¡’ñ¨)¿/œÀ¦b*Ù|ÄΤæ¹7ýÇVP<£Üþ;³;2&·ÜôW¼wãG½ñ8‡µ^¿A0Ë?É»Nø&Ì"£ñc‹&ôÓ×rúZM_ëé«!ŽÍ2ÕMJös÷ÝÃݧ;¼¸HF¿5y•”¦\¬w¿|Hè‰$YS/ž‰ò°0e™Øë)Ýâ¿wÿ¾û?BµfQ$MU±¤uR9L–&M3½e·é¢ÂÔð5»)÷6©M&µ8Q½2_žFÑ=}O óï;_l…¦þØa¸QRª„¨Cç0ŒÜ»—™Î;!\4VndË<úŒ‹¦LJÏ[ ´­y¥¿pÀÓ݃³ÏR9ZS~Ï-}Ô·§ê¾÷ž'þœë¸?㸗„‚ƒûg-à% Œà±èÃŽRªZ —Eb3°¯&ÉKEàÓˆoÌ43Áðt'‘æ–T¯» ò¾—XyÝSÔÁŸƒ´_¹ {‚nÿ‘Øèö;q~ü—{pü=Ü<önèë=‘5²ÒŽç£?²Å6KÞ+÷¸Çn9 £æÈœ;ºN¦¹„û½þ¼BFº(BgX„—½/¡ørIöW´'Xî¼cȓͥE’•¨ªMR×µ^µŠýbÍðò endstream endobj 19 0 obj << /Type /Page /Contents 20 0 R /Resources 18 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 18 0 obj << /Font << /F47 7 0 R /F8 9 0 R /F61 14 0 R /F11 15 0 R /F7 21 0 R /F69 22 0 R /F21 23 0 R /F20 24 0 R >> /ProcSet [ /PDF /Text ] >> endobj 27 0 obj << /Length 296 /Filter /FlateDecode >> stream xÚMPÁNÄ ½÷+8¶…¶Þ4º‰ÞL¸©‡n»ÝšlÛØ®îï;T7„ÉðÞã½`9.`u©JW1]je-°v̾2U;mŠ ¸iéZvÏ#°Ç9{ŵQ2YÊÏŸíö«ÉÎ1ßÿÅÚ\9ô÷{ã{ü÷é{ÁzÒXËáNÈ¢ªùýÐ?Óœ×G]Ò:~Þ ¡K~!°à-õ]#jŠÊÛy‰.csN6k$› û.E•+%¤à~øO}Ï¡ScYÒd+ñ3%_׈„ÔŽÏ—!›¤gÄKÄQ¿Q1þÿàwJU[?(ˆ(Áh>ÒÀÍ´ÆÓ•Þèí‘v$i Hã#jøÜGQ…[eœÅ¨¢VLÌ*ˆËž|ö y ~ç endstream endobj 26 0 obj << /Type /Page /Contents 27 0 R /Resources 25 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 17 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./tests-data.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 28 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 29 0 R>> /ExtGState << >>/ColorSpace << /sRGB 30 0 R >>>> /Length 1153 /Filter /FlateDecode >> stream xœ­VËŽ\E Ý߯ð˜²ë½LF)¤0#±@¬Bi”F ³à÷9vÕíº¢ÐMX¤3çº\å·ÐKz¤Û+ú@±s,ûoS4e®…þ|C¿ÐÛ·O?ÿœîî·À!:þÞßý„¯5Ó_Û¯¿Q ß7¡—ø÷¸‰ ·9 ¥Æ©Ði+S¦„·½ßJåØ(– TG¥,PÖBYí Æw…Ô3K¤œXª”³]m°sÊ…›K%sO”a€Tµï°Ã_íÜ”Špññ˜3%rtiŠ\:•ÌÁuai‰TüŒÁʹPiœÍ()‰s ¦nN‰ª²ºnm+ÕÝæ–9 Õ2¥Ç2Ù¿¹w;Öç ¨¡°(!5­„¡ÝvîZåpÔ7»JcäV©'‹ `Âw¡^95‡•k" ¶cÄG´Ô£d!õ`ª9›I$Y €+vànq†)’háÕæU%sãz±Ì[˜³ÝCdÅûI-’À¨©Àku\Y„${Â~à}3Ë܉¸¸Ã^˜%Ý1¢{ææú à}“ÛQ]ö¢œRrŒ‡`/ JývùÏò s—õgùó‡œ×OŸÎÓkÌ¢H/=¼¥Æóûøqˆ¹‚ÆPa'ú*‡¯éáqûîÁ/þœ¦H³Â]ªåzUt^ Õzµªb‚ Ü—j»Z5¢Í­ϪýzU-š|©J8ëŽÌ }Ч™'Å®2ò{)ö¼ ´¤‚Ï—»xƒÏêò—|>·Ë¯[î|³|TdÂG.9?’ÏhJôñ`ªéVÕ.,iTÝªŠµ’fÕ]æaöÉ–ÔwÉìò*6€øq»„ûá1r¾°±T£-Ÿfç³wo®v²qAxÆb1Ý?Üíº·±ùÏÓm†)½³ÌüÏìþ¥3ûÀ@ó-:Øx€¯¥Á>°t±°€ÞûÀjöÐt±”¤?9.¶¦ÔÅ>j{~²¬} ‹} &µ/ö,—û€iƒ¹ ö*ìa±˜ãb½0Ù‡™Þó¢‚m%þ!2×ì$ "e\7ˆ ÇM "6“Ââ ÖNƒ± "ØÎ).bŠhÖ¾[¿Óü׆ WRÊëNCj4`§!¶­Û†´båµhÜ®á@CදéICÜí¾hˆ»} !îv\4ÄÜneÑw;.2Ý>Ós{È 1·‡þ¤p³—í]t"úÅ‹+ °pAùöêeZBàõFM'D­·´Ïë5­á¬º¼znReì˜æõü±ê?Ê¿Î3TÔæ]wº>pɦÇh¼›4mö¢¬¹¿ÛTQàg‚i”n´7ÂÇ8vSM—ª¯¶¿$¨Òì endstream endobj 28 0 obj << /CreationDate (D:20150701163331) /ModDate (D:20150701163331) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 29 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 31 0 R >> endobj 30 0 obj [/ICCBased 32 0 R] endobj 31 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 32 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 25 0 obj << /Font << /F8 9 0 R >> /XObject << /Im1 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 35 0 obj << /Length 2704 /Filter /FlateDecode >> stream xÚ½YYãÆ~ß_1Ù, <âòh^IœÅfá ÄI Û\Š# IY¤v4ΟOÝM‰3–…8꣺ºªºŽ¯{þ|ÿêíǬ¼)ƒ2‹³›û‡›¨ ƒ¹{Æ©£«9ˆlOž¥Z ºƒ˜©óLõŒ†oDµ»¹ pÓbÕ;[¬£0™öäeœO"Vú÷ž (xz†ü#|‡ïƒhÓ$ R“Ìžæ¹§ NßvþDµïdëøº®ˆk'œkè/Gö^<Ís¸›ª?Lùzðâ~㹎 w2ˆg©—Ý_mš FQ©÷´§çåaØT¾;AÚÎmÛ"¯[JwªÚ8Ýáùs<_¸÷k.ìã, Ÿ+E‰Î?Ð~3‘€Ç/ŒüsF©îË˸€óˆÊÙì8C¥«ë­$2Lö"-Êæ¼Ó¯J—vOS…¥±8p&ÈÏ+Éx<–ÈXìÑ%a‰Œ7T¼„Ru‚Pæu{¨,Æ“9uv Dëý¤ðÜdAšË] ì?úQâ­X—9'¥jÄNÆ™h+­×oð¯öa² ïØÒyG NðOW;ibDiü·?ߦBWÎHÔ <óè2$õ«µõ$4‹8Èt}\,\;ùꇣqPÈ¢Ñ-¦Fþ/-Y¢Š½š‡çŽé’w¢•IäŒ]⌠¸ß.Mb÷ÖWglZ²¢ˆZÆ9éˆÓ(A/ŒÀr†4f¶¼d Ê½wÀ›æV+¯€lYLžj™?Î7¸â0%¶Ð2ÊŽhåÅâû(1|:1Ÿ>_U+ÿ¥^3ñÈ: ld“fÎp)­Ž5XÃÑ,NåÆrÐ:ÏŽ‚öƒ³xß!I¸hNšj†Gúþ} 5îP[oÐ#ÔØÞTâq8²j‡þ8Y J¥7è|x¼ÝM5ÈéÀÀžDmfjæp`@¥ø[«î}gU˜1r‹ö?Z¤˜†¡pCý¡Öf­q‡Ö‡“,^¬ŽnjÃýñÀ Rü@Àº£c€1DWpÒ°1^'×”óšíæÕòÆD_u²Ù#®éý©AAæ^m‚2©àëäèYDóŒ U’,~”²£.Þð°&£$ݪÖM0þyº- ä@Ê}Pe"$&‹1ûÁnILùvà ?\5ÿºfýþ¨ps´‡•§0“ŸñO£¨yÝèãÌ4¹†”V‹°dö6@žùñÈYMÂIùà ù3ünz”ý‘;ztØl8| “/úVüÓZ/ ñzkqR—q•§•¨¤à;¨PêS‰^ÊžWœJyùNÜÛÄt–”ô‰×̱÷êí@¥ì[¯¿±ø•†ª£V2GR,jÎO[qZNq+žsS,ö½ü X "C‰Ðç¶c’MˆËÕŠ°¡uË3n3öKÄ~vU,GH»z©à5ï¯' ô˜v25XsÎPh“àý¶§Pí È*ŸØ*K$t¡éÖÜ¡ºgæT`ý²Œ~•5 o­7ÆN“6¦ #5E«R’îòÜŠó›jpà`»•mDUO|»ýÞJØ^À«0}§"kÀ`Î@ÿn¸¸øU7…B½{ă€ô‘é¦ µSï´ÔüÓÂ5àÆiX¸°†E£m&™PkŠñ|id¡¸x+%sŽtå¢>Ý P™…ÐvÞŒOV>[|Ì$šªjgóì•LÅ¡‹×Ì/%!0Å“¡Z´ÆJ“e.©•R  ÑkÉ$4À•)Ë)ÇP}Yˆ`¸çXôœ.¾Á[­ÿDToµp—\œë~­ÞíÒ=j2[›IqÍ@¡`‰0äœÊýÈf½“K­™âà£!ª½Â"vŠ0t©Ú5r<&Ã+ÈïEWT<°"Éj@¤–RvÛÛ£!K¾\íÉOš%CñÝ•o­NdÀœ¸ó)rr"WãÜû!¢Š#îðă킢@B/E ed¬5À þ#açœEúD¿f8ä0%ÍN®©Yœç0v „ÏûŽsq†¾6g ;rä‘'ÚÊ®›ßâEìŒà÷§Ÿ÷¦g\L²´)½{w‰T Û·h)-;»ãçÂGÍzt9Øðöž`ƒWÌŒ»ÍYÇb@-ÊR›#¥·É°?L«—ź°€Ý|wT-Á¤E˜ÐœŸ¿¸]É=T˜ryâ> endobj 33 0 obj << /Font << /F69 22 0 R /F74 36 0 R /F8 9 0 R /F47 7 0 R /F52 8 0 R /F61 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 40 0 obj << /Length 2979 /Filter /FlateDecode >> stream xÚZKsÛȾûW¨r«D¼7‡”wË®uÊ[ÙŠå\âF$HÁK @®VùõÛÏ™ ɹˆƒyôôôôãëýx÷æí‡ú¦‰›Ò”7w»›45q•𛲍b“7wÛ›Gw«4jGü³Zg¹‰lÐ>AÇ©‹úàwšðë¸25wöÀøúÇÕÚ”Ñplá æT4gâØ„òƒß p†?ذ§§áÏ«u^BO‘ì®U–pÂF¨ÈÆÐufÈ“¡ ;lž“9ІöY¯þs÷÷›ušÇMžÃo7…ˆˆžqÒ–vÆóßÂÚ:!¶*82Þ÷$ºgìJ¢¾µ8Æ–g ;êBÑ´<hœ²´¯’-6C}Hï4?ÉÈ’÷ÃØÉ”p¯ÖEQGw«:#ÑWÌÒq«·ŽLžü~Iô0 kO$Žd.ˆƒìI—kQ”Eõ@hr×E!·­²!S&Yaq‡gþøš‰^Ìår5 ˜”rWû‡œ®U–¡Sõ ¶Õ[þDmQQ–bƒÅ8Ûþ®Î1(ü"™ÅBuM™x&SØÉ=–dÆÚ5uÄ-É|e†/ëÀíÓÌÄ cg7*½é¯ÜEß l¸£›ø—tçw4ÆÖºûÙÇÄÊ’2XÕ9h³°!’!“îî剋 ̈> ÛŸºAÛòωÕ´¯1AçëöÈPûVÎâ\Œî;w®#ÞÓ¥š‰Š¦Ÿs ~­hàœuœ†Ï™‚æçM½'³éDÂt’ú3ÓÓÔÈ 7¶Ú®Ä”Î9<NÃ3˜¦rî ûp•íñÜ¥zâY¤jdzhçIÁZ0‰³ÎZÖn\}Áâ–iåùy•M”Ñ:ñ}<²N€¬Œ(YVy)–ûœ{ROÜ_•¾ŸPøÝév~3~zë®–©n…!rr£*›:³n+v*qÓ¨ê*÷|a£"¾¡p6'Ý-#õ=9Œ‘ý+|ÊJœþ‡hŠ;~§YÍÇ7¢*Ÿß}fÅÈröì¹8KdQÔ\*\Ÿ…??Qž8gÉÊν\-v‘³’2Iý´7V„xõ ù¬¬‰ìö’Ò‹gG~+ñûèéû0—Ãödåé·²¾¾ÖñWŠO.j‚þM‘€Æµ3LQ†Ó’üÞ SSu¾Ezd[€˜Ô„aŸp¹G¸¨"«è¢ d~MÒïY¥ ]tïHRñÂQé©oZÀ(ŨÇ(vŒžà.a)«tü¦GóI¥ëd˜àóËg6Iš¼(yr»ÇI¯úæ"Â_`Ú¿¾¼јºŠzµ„Žôo¦ç‘ßW-qmDaÿ™.óžú)¶á)1%ÙÌÕµÐߪQg#1é¼(`Ý!‘³ÇMÉ}ô·íâ&Ù±iù§WC²'t–ïê±Xù„ƒh™N¬¥ÓÅÙ1ˆu¢T§ Ò@HMªy¨ùeê >H¢ R$÷¤…BAj¶r‚ tu²9ìŸ1dß 4yq¯ôj/Æ­Œö.w#¥µêÄ.QQÂîvt!øÙ÷JpÀêd‹Ô{·ÍÄ(Ø“‡Ò0ëÜăvâ^Š7-÷=u²÷ÇÈ·ËX€H6Ð,’J’žvÂ=;µk™PÆƒí¤¢èóv(ÔÃ/ÆÌ0+2ÍÒÄG·ÉòBSTüQ‘â !%/ ì#îY /8•À‹ËÊr!• è8£Ó–Ž.â$ÍÄLÁÐÀ³%îì‡1ÑÔÐm8´8ñ*ô´)7U´ÛÜâñïAÜÎìW±}V?µÓ(‹«¿B© œ0ŠLSiž <(Ráy(6D7åÈïÜk”³3„%›1DèÙmÒ²{IftcUC‚ÇÀY7rÄ1…ÛFÀN×ËòmN-Æ^l ¹n/ <õì¶AGç¢3»™3NZ÷öC^a®å¥Á:ˆ33qQ <*V ]SDŸ\ög.V9¥JÃö0q‡US¨õJóÖƒu—ÆA6˰­Iã:©½îgyãÌ_=‹€¼–G=ÂÑÐ\pà×µj ä<ó™þá‚–ïOóLúÔÌ€Ø^22q§‰Ôæ[zŸCiNýõ.æàz9@+~ø´?ýôC@Ù/pZðÈájÁAØPkS‘@"{%JžÛl¼H꬧‚Þ'W.Ðè$Ùf¬Pê-ÉÄ݃lÔkÆÒ®w·è[Õzé.æà'tž%Ï»å!+SBñ$‚aé ´y(OË"€¦–;úÖŠ!À¹’è.Èup¸“ß§Î`¹SÇ!(Ük=ÆÕE¶<²¦†ž¿-ÿLzHŸf×uì$aí9¶LšRà€žEï–ø±ù=%Ñ?¸k\ ]Ϥ'‹Mµ~^9hÆC³l0å|–xÛAZBœ&¤¬™l‡©÷û80r‘%Ec,(t¿¢12Z+dzz(Ð µ¨Ôx¯®xþð›Æ=åW6QµOQ)¼C }¹(5^ŸsQe3C(øM·¡ˆ‰ÝVé-}ü‚}d³eãu´ç‚œȃL|Ò2̼F²’VB)eHÏMÎóÁ•de"ßÌ;*?9hGÅÌѓ*}žG_Ó,ç ½¬£q ‹Ÿ˜]$t,De-ÒAxûç@>ð/@.©£G2†O-×è¡ÍŠD×¥G¼ÔÜ;ßtN= æDÜI»#nôä.¼¥†`§0^ìÉsÉu_é`_FN«wD7‚,кÙ1˜tJc°©â"“£ˆ]A ΣŸ¬ìÞr˜}Zel½Ýž6ƒ¯Ór|Ŭ¥N™äÏRD_ZGÛ&C[á }Xþ™X(Z%„ÒAH+1mtòe6—ÅéÆð>‰ã ¤uF:äÜ@R_†lêã V«Sã¾#Óùž5Éß°«‰~V€LüŽžE΃F^H®gø³T¬åþêkB™ÖÅr‰ñËñÒ÷Iâ ã%fzNLÉICŽüáJ©2`˜}òŽÒ•žY ¡ p’b:%WIz{Pk&J Ø0Þ[4êè›Îgës`¾gr¼˜N+7"«Ò9ˆÊU)`·å‚xÉîÅÉE§JSž¿oƒ¯ÙÜž³Ì°*T·!OBvÔÌŽr±“„K°¬O.Vxó'†ä RØë«ö¥žýÞîÅG-èV(}´+v9WMZ\H®ÊõV¿ˆb9“Þ¿Â …wücƒÇ‘™Ôè=Ñ aÆ¡¹jÌÛÄa ý[™ÌèÍÏÇð=DbÝãÌÀ€8á±)$mRtÏM➘ґçh1¼ cKocì¹DÏL¥HÖeÁ žù™d䪮Г ê&Ù™ « «àP ûSbxµÀ0éÛ=ÛÒ×ôdZ„¬óÚ0“:A5ͳ輠hqú²,eù2-æŒ?,ãsrx¢¿018ÓcZ¨ó“Zžƒ8¤¡ó@z¹„Læä ×k¥b-Õÿºz¹œ‚þ‹òY]]’xc^]‡©á½½¸Õ—½ŽúST=kÿ„íËsÝb1Je;Gc®jQTЃP31^3 ÀøbŘ4±×*=³§[Õ´p$aÉå±1ù\Úò6kùùõµ}ÎZ²»”1_0üˆÞ¢òÀ `KX ­r¶¦YõŠd#ŠÐaM«y[õS+’Ò\µ6hžˆdezXä?¬XK ¹ŠË&¿~Ÿ+fäC*¯ÓÐ^ ±¦!h}%´ù[¦ï`¬éôE€’N—K›zîµj~`€&IîÙ(#ä†&îüN„%6K“aR@šÔ©7ۓܽ"ó…¤R7 æFƒ#¤ër¥ZVêÉI)ß­I‹:.à¾L×Z»*qäÍû»7-€q¶ endstream endobj 39 0 obj << /Type /Page /Contents 40 0 R /Resources 38 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 38 0 obj << /Font << /F8 9 0 R /F47 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 43 0 obj << /Length 347 /Filter /FlateDecode >> stream xÚMQÉNÃ0½ç+|Ñ×K³qd©â‚HO‰4 -´I›Ðÿg‡FVF3oÞ{öLŒÐpŒÈS•&™°©UqlDu ÎÊëVD˜¥Ô2^æåÓÑŠ‡>x…3µ"oÍ<ïŠ`¹ÎDŽv‰(Úÿkc­ð/jñ.ס‘{ø¾.Ä&ŒVÚH{Fqfd±Œpß8…pìG€FTmWQYCþ}ÁÖ¯7íлvRžAÊ¡œ8=’F¶hûlî&kK|¹çVMÊ’‹±Áb…ɳ 1òxòî† V Øn1PÖU×ö]øQ<Ã:#cT'¼ -½¶ m&ø.çœä¹ÁfÀÊñì'+ÉwdÚ–(Ô Ð´ ç¼!R“¦^pQvWþlXnÞ˜bµÖŒÐßi1P†ÒªaÒ|åøÞùÊù"ÿÊOdÜ@P´«cå’Ö°Ê•3Ž÷b/x,‚?é_¢Ã endstream endobj 42 0 obj << /Type /Page /Contents 43 0 R /Resources 41 0 R /MediaBox [0 0 612 792] /Parent 44 0 R >> endobj 37 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./tests-pop.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 45 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 46 0 R>> /ExtGState << >>/ColorSpace << /sRGB 47 0 R >>>> /Length 1596 /Filter /FlateDecode >> stream xœ­XM]5 Ý¿_‘åÌ¢&±ó¹¤¥ U€Tf$Uh˜¢VŠv@ˆϱ“Ü{§”êEeÑ7õõ=¹‰íŸ$¸g.¸7îýé¹{螺äù[™|u•ì>ܺŸÝ»ÓW÷?}÷Ø=¹:yòÞ»ãïÕ“ñ´$÷×éÅKçݯ§àžáß›SÐܧ$”‚ãœ)Š»;eO©9.¤º·§\(gxñ§À,‘JÄ×峪ÀŠ(f¥ ldb}¹%jÀr¦a/ä–@EÔÆ ÃfjÅìBpN”¢ÚuœÐ"%Ã(v£ÄfWŠÀ¦dã!&ø tr!Á|I8µeà+>cïcÉ…1žP³ï¡êáÇkAíê©*>Söfj þHÞðXŸçÝÏž)Ÿ*ù`v£ ø¨Ë„21ðø¬×ù0 ðH@·ÅSžkK¡X.ËÛøHBÄDŸÅìJEñYÃ;£F€Ç´XãÁ…©&‡ðY:¸4j€#|E“ÇX¾|!¶é#ùºø ÿ©˜l2³êW˘›98D°ØËÌZ>0C4³QÌEd‰Éš1ÄÁâ.Q4àÉV3yÊÕ%«%5‹Ž™°(•äH5:<´ê´ëõ¡š•Zsˆ¨ØËÕB8E¹Ù|"ÞÑÒŽÞëJw o›y¥[ïÿØY}ßÞ¹1~7ߺ«ÜÃêµï¨þéžæîïÕ·ù‡¹ùGö§š›$aú§¹ùLwŸýæ|=‚ssÿ‰àÜ߀•¾Eq»ëW®ÒxÞÌŒM3ÃFf×wî"ùKwýæôôÚþ2„F•Ð|>´ye„ZΆrùÈZφ η´ ÕÍǵ¿a{fP®=f1¶(ÃòûÐmyëÖîmE÷Ãôvk÷*ñ4ÞÜÃ<øAà›ÍßÍÝ?&3ý‡¹—½/öÑ.ºŸ£eK—­?³Ï€Áâ’&8,‚µµ G 4¯¢A; ¥–E4KÒ8t|˜î±O‘Áiq’É4¹¦ÞjFšóåÎl_¸µy1tÕ^˜_ÿv{ö:«µa%Ú¾ÌonßÝ¿þãï‰_“<á $OS¡±Kž¬ ºIž"]LÉ“uï’§öÖ6%úÑAòh'oGÉ“µÓ$OÑ~xTåN´sÙòZ±íÀ¯J[yuÂH'86à€SkPéJY[Ëw± vªžiÑdêêWA‘ûñr5¯ ”¤ÒlÁçCA?Úa””V#ŒMšµÈTGçCÁ)Ú·ðU^0è¼ö%ûÕ ãì«ÝMt59ì³5N[J‹Pìí±)iz× ÖŽYé‹PšjõèW× ÁªM^ï)#¬WE¢WMwü"´™tÐm»X¬— X+ÄÇò„Q¼&HDìÚ@µ tƒ_¤5mŒ*s @òêWkœ—@«<ÌP…µ W‹ ¬¡…ã+Ø7P‚ª7“*A4®çe1D¦–ŠIÕEd±K° m ‰3±ªÇ²¼Ç%ÚÍ,êpY‡&]^g²+ÙTÖ‘¼ýn¯,#›öpPY^Ü-‚L»õ[e2Aó×{‡¶>[œ°š)ëUÝ!Ø_ÍDø*eGœc õ£ñ ÿ­ÝõL¬‡ÉXŒ1E¯u”°ëfÛâgqlgÇ×íïs>-UDÇzSm‘/.žÞýþúÃ¥{j¹x}óË[œë_]bû»‹Û·ïnn/_ºëgçÉ >6¹Ž±¿¿úÏÑžŸþ•ޱ endstream endobj 45 0 obj << /CreationDate (D:20150701163331) /ModDate (D:20150701163331) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 46 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 48 0 R >> endobj 47 0 obj [/ICCBased 49 0 R] endobj 48 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 49 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 41 0 obj << /Font << /F8 9 0 R >> /XObject << /Im2 37 0 R >> /ProcSet [ /PDF /Text ] >> endobj 52 0 obj << /Length 3615 /Filter /FlateDecode >> stream xÚ½ÙŽä¶ñÝ_1@D“¸e‰”D)ŽØÆ.`ý¶hº5=m÷åV÷Îî>äÛS'Is8ò nÅb±X‹üòæ£O_×WMÚT¦ºº¹»ÊM•ÚÒ\U¥K-¯nVWß'?dev'·ðA1¿^”Y–¼‚Úî?øNø³„¯Ý^/¬5Éñza\rÀþËÛÏyا׋¢°ÉÍ=Ô:„¶Éá€ß^›Y‹#"D ºP“":ì¹·ÇÚ 'ÁùÏR8uÜ{¸côØφ éi&žsÅà+Eߦ×?Þüý*Oó¬¼ZäeÚF,™ EÓ$oM d,L]&=T~A Z¤©Ã*®æÐÕ"TÌ\UÄ›€æ<»*ìÕÕ´2é‡îÅNð|€™ 4²—“†Pí~hBVшˆN¢ÿ¬ˆ‰J`qÆ¥¹­€5yÚ”"$wؽ£­Cž,´H,_Éœ?d¹íNÂó3¯E•Ö¹ ²h¥gšË“£lôFnŽ"€8M×rý–§»×1+lÎXHûŽ+(wv¯ïQp‰}@×3¶¸•HçQˆWGy°ˆà M\ïi3qèý¾•ñËn"²ñpÁÒª“%÷ƒá;YÀ-1yA,̆ûÑ"Že‡d|ÀMlRa|* 횥ÉD^±ÌjNhE@V ò@kÛs¥{‡ý$o{eðš»hY—IÉÞÉÚ{hž“9ÜML‡,#€>Ü“ì èð€?ë7*áÆº#MÌ“/ö¸ï‘™å±8‹”ÓB¡mTjý/¢~í©“½¨Àdôª³;Ö n&áÚsY¶ ›DEþDØÏ( èV· DLdÓù+$qÚq™”·ªë¡<?À(v”9%èd·Ðs?'G³Òn6 ª‘ïp䧯ó/¯4üÐ P&Lã=`Ïý·žŒ­W+ ØMò·} rPΛz0AÏ †[X}f]³Ñé–HÜ´™Ð“!(¦‡ÿ˜½ØüÈ V>ðÁî$¤ ì ð5ß|5ç9pFœ—N‰Í"^®vâÊчÆ#Gm£{zy1‰tyqÅ#Ã(É9ïÞú¢ëâîÿòÀUŽyIô(ÌÃÒûÐq9yMY ¡¬XôʤéצR6S£H<à^“(vÑÒ»9V^øìµ(2²€ôßò‰ñ. ë±q«JÙñ±ŠINa]õŽ­»Ï½‘8tÂ,îü¯‘‘ª£‚!‹›f"ªu yö²²&oéTJ‹CúßóXX9œˆ†%á¡a;owVªm©cž}£¡í²› jªÔÕ.,êâ@¤L+[’i§™ºŽ:-!ÆŠ`”YyA[”[:£ˆÂŸÚ^V“[o°|áÓ–ð¥Í‡æ„l6º¥±^ƒÕ!{Dú¼’S#IÜÚ«Jè²bb·¼w=·Þ‘”äƒà€ÚÉ=ó‰U@ÛýJN¦y Æc¸#"3ý&Œ·ª+Öðô.–ÿLþ?¡¥ªˆé­1Æ/Töþ0³¤B`P‰Ÿâ3YZ׿Åè~Ô„í“CBVÃîÝÉiAc®m…K¡;Û%÷þ‡ÿzñ\ï8U¬[I|±|Љ ý9Úxøþß·ð½ŠÉΞ¥×üVz©ú'þÛ…¡%¥d@ÜäBIû?¬q9\AØž<ùX&<iw"Ÿ(??i \z~˜'LfžŠð›¢fªâýý½kaGm,e:4ZPFêÌó8o#êTW‰\³ð5RÉCŸ2+"K˜õ„7Ï KC/+Þ’ Ê%Ö"¾žŸfñ»°Û¿‡ïòÿ¨pýåHåO·àÚÂñ»`Ôö"Tè1~ŠìSÈw<.B³*B úõ\1/çŠy1WÌÿÀ•X)¿;g{ë#*OâC´i/PôeÖ.^ÉŒ°ÆRjJWBH½0YSŠòäYZŒìóBÔ%xæŒêE-_)}¥¨Y!ÿy¤K¢j…¨a²9V=6w-œà­e>ýù¯#]6@P.nYç7:‰¶ð^¡0x¤rraàÍum)ö¶.ñ#Ÿ(æ’ 识¡…KÁóFò™÷ô©(¥96g>×JbõA=”—­ О¨ˆHXéá*ÆEP(\”Ô^Ž'&GÞÞ:“ÖÎA¼¦‰«‚4p&g„ Âî*­) dTQçpBÜ'() QÔ®u¸NÀÊ#Y[½ªr‹ïZ½§Ù·¢¨%'4Ž +“fuó2.ŠI vd$6á-‡´÷!/Ç-gŸÌ)\8Ôn&˜ÆV̈óqdº<Í ç‘,›ÍÍӉѦöi¬Ï÷‘EG0&M•Î4wsgšÊÚ«LHô™å„ðc÷X¸ÐÆÑQ€ s”V³fů%u7³yyZÚb†V˜ºNÂÍZÏ ÑÑvJ•³iUùCáQ%pø9î¡8 N„’œ9„fi–U¿j¾±@/¦pÅtô¼ìPbõrÔGi!Ø¡xã굄2ï«Þmu˜ŠÆc_ÔÃÝ©¤?‡nÉÊ%¸8"»¡¼„›]’’lF“ìä˜Ï™©z»*ÄʺaH5MóƒÄÆ Þóă‹^ ƒ?è˜JKcÌËæÁ°W28š™$íþ–®¶1 ¥—ÌKK¾ã3¼‘k”à .© ˆôRÎ%§X1ðdÆ|ú®ßŸ¸# °FTû‡×Ó=ŸÇ7¡8:©_4I9½7ð÷pz³Ýù$lðyáõAçsé÷zìÇû¬f }IyÖ;WW0MJ#„Sñ¥Ö{N>lÆÒ{’<ôOkÁrù%fË&בÀ®Bïž‹ïÕɵ§0$Ü,&@aÝ im ~†U0›îå[Fùl_²s/wCf»zÅwÔt"4.—8«®Êºÿì2|”"pþýÉç™}>¬ó›Ä&m£ë(l& $z‚¿Ý|û°¤ãr ´"ª sµÒEêwÜD!–ív.°%`ý}¡G $Â2âd0T…ÿ\™Û.žŒN.|Ï4#Wë8!mÁ´Ü{¶ Jh5^&B×ù ÿHݡטøÙ‡'ª· @&H& /”郼Ž8»åwWtIâ/Ÿ<1“çP4ç)Ì¿gôÉÄA 5u žOãÍ´VñEÀ$ ±NÁ6<%@mÔ­W§ [OÁVyêÊÑ)NäÁ:¶ úøÌ9v7´^GwùŽŸÔÄ?G¢Ø´”Pù¢gàÛ\âï¡üóÞÛs<,×CÄ-c\G7âBÝHL•hQŸilm³"uù8¸Æé®'Õ€’‚¤•Æp‚?ûð¶‚_(„[aˆ£…Éàw‹²‰4éåå¨g> endobj 50 0 obj << /Font << /F8 9 0 R /F11 15 0 R /F13 53 0 R /F7 21 0 R /F69 22 0 R /F74 36 0 R /F14 54 0 R /F10 55 0 R /F61 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 58 0 obj << /Length 3415 /Filter /FlateDecode >> stream xÚí]ÛÆñÝ¿BZT‡F4¹üHëI›)Ò‡ÖW…ãÅÓ)‘D…”l_’ßù\.O¼øâú±w—³³³óµ3³úâúÙó¯ÊE•T¹É×·‹ªHм\ä¾HŒõ‹ëõâÕr}•-·üœàé·ðsÏ{Øéð3$W«ÜÛåw©O¡ûùa}µr©YnüFà»úÄ:õÐbÇòx¤ÛBópeŠ¥¢ïÇoæŒKbkÛÉõÍÕÊäËî,k¼Eœ¸ )Šåw™uõú{¤Z÷±oa•’šŸ P¹ljAÕõ¼‚Þu¸ÔžõÕëë¿/ÒÅ*Ë’Ê köH.γÖµ7÷ˆûÜàú;f¹FÐïÒ,׽쎄}† =ÿ*¯"±dЄ֊޲ö_®VÞØ%±Oúôgh®¸‰\ìày ÏKä²@fŠÌÑQ@‚ô<ž žÏäÊûSzˆÞÂÅô–&)2ì"Rÿ0³%ŸS)À”øK|&MÊÒ<Ý눓‹•É\’z7•œpù^‹F_r-þº#= ܺ•¡FÔ·uEcøô7ã”lN{fh0Òp”½éЉ‘ ߊŠ]eD¸~ Ñ2eˆ”åbùd{ÁpŠ(í¤/¤y‘q¬Gé¸X#›QšÖãÌ© ²åsÙsDÛØtÿSN? IÄÄ€ YàMbÒ:uÑ……/æ49O²ôéš¼›AïA`õ»l|'»º'ªÔvÉ:\–e6ÝÈ;‘ž2áyIpë‘~|2²õþð¿ èAؼ¨ødb¢sÜü¿ûþí¡Mv‘u¦üymÎËáQz |jÇÈÍZ‹©’¼*ŸÌ˜óÿh.SOlQÌ2U‚M¤ß‘«Ê§ŒK¤k¦~N€Ñ¾Ö£g+œ©É¶‚Ž·rô´!2âå›K«Äæ7d¿iöÄ˹<©J;õöó¾$ì«gð"2©hë7“Ù3–•ºÄ¥Õ“%~ø0‰ƈ+ŽƒWY [¯„\רÚÔ/¿yÉï¦Þ…(ŽCN‰Jñãšáå²x¤æ×‘;”øyœÆÑ4Ãq¸ŠVÄáêÃw·¬)þîöa")ËŽ£ÂpK'·Ïš¦ù’ãäöNÁš!¢0»©ê†þ'Ôp;=ªÀï¼Z‘ª îà¦'•¦Ñî–ß;¥„béž÷qkÞ ¯|^,¿¦y\=²•ÁuÑû±ÛÉÚ7)[ ;átü¦•J/+¡.™Ò.»Ýš{RÄÿ*r* ÍUê ݉'ˆ®à„AóÞ¬/:†Í^©ò¨”°2ÂnîÄZ‘ A#¶zŒP ŠÊ‡EØ[ŒÆêX›pt¯hp™»¹°\µ¢öQ!’ê»ñLx¨ý`y.¯Øým†eëlÔ÷õ2,JˆÍšüÕ=wÖºZ¿3´çöÄßTÑêÓˆCÐQ¦üFm-JE0!}ïF„,àC êxÙ°‡Y¿`פ²f׈̀Ž1 {ú¸ë6÷üµ>e~W‹<`Üò¤y§ƒ~ÔŒ ¨<ÑrûmX®ç4ª07ÔHàÍ•ÏÙÍ Nr®ðöÒYÊï{öäo¹xã€cß^¡º©ò– Îp«îõ¤ ý+—wd#í­ê–”W ØÂs‚|ÇýPU ª‘J4j'u Êø à=¢MÛO†‡û ðñPLY˃O ¡B³4cŠT´@[ðò–gÀ½¬X¢AYÕC‹¾lH¡ÎG–xñ†0ÖìºÇ`ëU5å ¿bvuÄÜ {«„aU:ãµxcU칬¸ÎU¥Rä€e^û£°õî1{ëìlZÌ;£0ùQçE±AîŠåµì¡)žxÇûC8”õÐÂå6ž1å7ñ,1[@âνŠRci¡àóƒ`QçIn´¦ãã°o&D€ýœçðågqQpçü2øŽ—;æxÉ™©ƒ.¥âu~¬ËZçF³†ä=Q£ÄG_ü‰C’0³“æÝ8¯¥sz’ÅæÁ§Q²±•{_D b[©”¡ôˆ^‘À u*®(™~r7J×e›Ì‰ E‘±©ŠMõÎØ&´Ÿ:Ìø.(òZ@6Ë5À̶0s¡ "`J†j•Ï+˜lZ`Ž4Ä:ù†Ì´ç¸0ÑßÔZ7£pú„÷s ÝØÄ7š\JæÛ1¤ AÕ lßÔƒ‰= ûr)ómÂÁ¶áÏKlHÒ¡XÏq_‘Ó·dƒ'OV…Ð57c¨”ËEquG.óÞEÕ¥6œùgÖ3Þ‡†™Ðæ › ÌÀsã%C\¼Ïí¯¯ÅYŠõ+‰¡Á¬Ù!·1Iog};€!¿(6Ú8š¢øK§€Ó¤Ec(ø2œe7eq AŸõq¬sBá…†Œ;˜‹(éÆË‘ß~„ã ÚL‚x# –kC̪hxÃÜ¥TÝ ^É`²ëš") ^Ö'\GÏ2tþªõ/RÅN/°)«ØÂ³wmÉ?4§NCYÌö4= ¸i5VQO2\Ö ebËŒiùVOˆÃ˜˜ÞÍ'“|@ð'æQZwDH×O@4=¼ qFùqh7¬?›nêŽáË/ºKÂ{’ ]a;’ý¯éÚ2¥ÑcxÞK6#›9à°XûùKöð¿ðauŠ™F' ¨ ù­ŽÆ»4‡“²žîr²5Ý^ $\_'çÕ‘‹Õ”œ@Ä‹>¨ÌÆâ5™ëY2t2nãã2d*žRœ’SÏìi+…ú$¸(Èš‚EOW#Å™YÇÝG~ç$ì1¥ ¾SÒL:\÷šÌZÎæK3ÉæK6ðV²%ü,7™RgqÀ›¯7¤6šVeÌ/tà• §>ÎnZö]Œ ÇëÃHB8Càc-õ¸,®ò–&ÉaD*rZMÆ ÒF6MûEne©„&§p ³Cº0Å¢J|^(R ˜+„žµf‡Tõµ¹IŒ)§œWFW’_ㆬÅÚ1¡µãaƒãg=_4³vâ/ ô§vœûòúÙωé"[dU‘¸ra,pÁæ‹fÿìÕët±†o`š‰­ÊÅ[‚Ü/L–'¦ÄKŠÝâå³>ûâáŸ*Ljt¢”ðߣVµ’ Îpß•‰Ký%÷‹yæ‹®fÙ * ì)ý"{2:3‡.K¬·„ÎΡKo²BçfÐA$PøCçgïJè `š89³þÁé#r³IQ˜ß*7óqåf>®ÜÌÇ•›ù¸r3sr3UFYñý øÊ©'wÝÄà‹òrЛ¼å KèKÐFGXË“X‰EšÆ÷…œ®ˆâŽ¢ Ž¥çöž¦°¿ìé®ÖW»ò%ÄÇ— LÌL¬AZÒžäªËUºü¢s,¡]jâÚ&Ü&s|)…Û”“Y3ŸøjJÖ4Ýëåu‘¤y9 ÛL†27•œ$­ß4†(åILUVŠZ.ªã´¡üq’Ù\Êr!Þ¢GXc%%:GEy\¼¸‡åE=°áÛX]ó1òš3q—8[^J>¦΄+jú§¬…gHße ¼b¨U.¾«[YSÁÉõàÏÌ­©MSÏ…1äÛ½fP毀cqs’êè6Á<g¸ü3ð·3ÉÜvÐ`¾P–Gi §ýz¥$iEgrw£·Ðþφ¡*Ö!ÎI¨™UÖ1ä«Ì½žMf…> endobj 56 0 obj << /Font << /F8 9 0 R /F69 22 0 R /F74 36 0 R /F47 7 0 R /F61 14 0 R /F11 15 0 R /F10 55 0 R /F7 21 0 R >> /ProcSet [ /PDF /Text ] >> endobj 61 0 obj << /Length 3125 /Filter /FlateDecode >> stream xÚµko#·ñûý #-ÐusÚ,ûj{ÚÃ"äÒ¦È%ÀZZÉJôp´’}nþö΋\rEë|-úÁ^>†Ã™áp^ÔŸß½øâmsÕæm¥««wË+¥t^+}U•u®Myõnqõ}öîöZeýõÌ4m6`sÿö»Ž4Ù:ݼ?ro¿ä/‚tG^íðYÓÀ†'ï¯Ë*ëkº„xn-_\{Û Ön‡;Ò‚dtñôNð7àÔ¶‡»k]g² ñt #î=óãqr)Y3BÇÎïþr5S6o­…¯ÊÛRÄ2ßoq×ý@޲xËŒ#šßÑrªjJX ‚Ö²Zå×3Û¶Ù·»I#NP¸G¦—ñ -+¾F÷wÂú¦sÛ:B&bdª*ÁX1Æ÷EYàÀW1î7 ðwع3¬YÒ0ÉyN@$œ"/m ˜KØ!BŒg˸ˢÈÞ hî„X:üy€—,œÒt8f²¡jÉe2T®Š2EÅ|dï+\ÓÑá®á‘SG¸9ê§ð,"n¾Ý 'KÒï-NUÙŸ`£Bgw×3]Á!D»£CB€5é*4Uöp˲Рh"M•s<æ[AUƒÞ ÙMÐ)ºv¸2PgìòÝ$u?ðeÀÑ[B° g$C·—’Xf,Ý&×VÅ€ÄçOá‘…JÎîPŒsÂá JÅ'–³¬Ê›FÅJé­ Ptš™GïõS7€Dê(´Ç;þ |4dÃÖ¤|p˜‡ýfÜTŽ›¡u™¾FŒ¬=ÞÜõjçl  ìñ&1¥½Ówïúˆ ^"d çÏ6ÍÐ{½Üã™ÞżA>ðÞ¦+ؘïÙ#Ý VÆ—dÁ’¿ýûBÄHñ“,ˆ¹4b"%`yÍG#9ôì@àâukB4Êgˆœ¬Dé0eØÀ nOÉÖ‘ì¨ q‘$Ù»à]ñ —?žÇ:þÌÉ÷/ÉìÒ-_Kcì&Ìòz±„wÈ1Ñ2Yƒ‚BÐi¸2ñ 2¬Ãa{Šb¶JGø’~ˆxÈÑ‚íÉ }ã,†ÓIA:탉‘ Çê*[­ù$ÅéVd mJ½„¬”›Äqq“ ±8§‰±r¤à&«ŠYÕróÑ‘Ò (ª ¡ ûË”-ïeêæÜYÖi_)yNåì’ ÀjòƒÅÔvŘ´®P~îqU® 69Ö™pôÍSÖâ]Ó>Û J¥“®¾PŸ'¶£Ö§n§Û5¹…)ú,«'H¿”´ž .Xh8RºHí%«žªP\+ÁÕ+™Y¡~îOx1j[A˜é§3äÆXÎ{vÜÞßÈUî\,jl¼»å îŒÿ"°Rnrã²CÐîŸ2ÛàNØóäŸ0ðÈ× GéÔ>¹Ì$ØÖ%98¿ö.€oë<.¡˜léìädÆ:[ôdZ¶'y ²ƒÝ*yÉ™¼Œyïr¯œU¢j£˜Rä(Uþãõ¬Ô†r“î:Ì£‰’)!?P"OÍýRÕXÊ1ÍyößüqÂó¥ Ùb0¶A4èvUˆ¥#ñó°«'03ƒžø}Ærëœ<¢„QŒ¢v(§BØñ‡‘—¹°× 2|YÐB–gj'¸у0¥?Nü"HÌv²á6ØPh9ç1”oÀê3ñ9–OÁɧö0Y¶¸LáË© T™WV€ jµË9Cµá«æÀ F…#JJhĶ¢vËÃU0\Ãu0\ÃM0Üð°*l _ÿœ‰.ÈÐGÔ¼HM÷—÷Ò“±–´žÇŒŒéÎp– ²ÁP“:‡¿Nö@ªÍ3÷p´Ø€ær[.Lçjf ‰®À#j E5Nà2¸xb¹xb³»²}'[_³¢ Ò†G!þ´¢'èïE}`yz,¹º±+Ñô…IT‹ºu¹g/A#Ñ’ˆÎ`œbNøeåPhªær{zo«©Sr6¼wâòŽáöÀHåÁƒJŽ;a O–[\ ^ÖQ ݧêyŸC0ŽÓîPY¶>«t3FNu*_v¡Î|¿u•Uá"ðÔ§z û„³2E麋`æ³8΄“‚óè§Êòñ$Ú×Ó™+ÆÝWØáSÝI9¹Ôy£ëX¤7\ïåR)°ù™jm~ 8¦Š5lŠdÿ׉¨Rçj ?ÖKÔzkl”dúˆ±™O÷ìph0PœD(¬ÑÖøºÀÏçÔ´À²–žËá%J©¼%H07Vh8Š Ï¸ Åê+цíŽG¤€~Ð]º/>È„cmtœ:sµx—à[ƒÁoÊñfµBóè4ʼn½º¸Ên¡¹à¯EC(p!ÛäK÷:sUÊÝê)ñ7 d"åé7¡ž§~&V¼µ`û’Å"Ú§˽bì)-©ln+¯KfBukÆœƒê%?¦ëPXÒôÙ1ÉÔ±„$¦RU\‹‹HÐ,€Jܤ:oLPóúx©î»D†Úæ…‰ÈIaÑÀ~‘ó]ê*&`Ýçn&ª®aîàyFN¬>E>“ìtÎ[—‹*Â\‚µåØB¶%ì;±ás<þÀ/<%8øñ×Åèl°C· –ÎÆ¿ëãc|¿¹>¨Ÿx7LgRkxã'÷u7 fþyÇN®ŽPµLîà’„—~¸v/3X¬kuæ7/°yà>Â=·µQªŽ]b“âáŽI‘º”¦ÊUÙ>Ó–eP\ :KÏ[ºõÏ[òÄäß©&¥;ŒdÑûï~Ë~œî\~ж_·+â,~¿‘0ö ð}óƒ< Âæßì·#úîäÄ /Ó櫼V&–6Ÿ ËBïYöôr:Ô† êþ Ö± ¥WÛÜh;F#%¼;ðÛ=ƒ˜&ת cöMü$ܸçC¡!~Ém&ot@ñ3ϦkwÆäÆÖ—Â"o ~ŒW²gÌÿttHŸkGp\ä9‘QkÀì{¹Ü𾞠K*%$yœ“S€-¬&eÿˆJ§ÿ{.R}É¥‘nëJ:hD…ƒº+†àç C˜Åû0Qª¸üCocú \&í¡ÛJ+ }Ï~VÃÕ†±R”qo‚÷áO,áü"•„½p²=Kã—A’}¡ ss–pÓê{é¢*Á¯ÆÍO°âsÿ¾=R0¡ïFŸ"7pHN)~“D´eäôD,œs»sºêN.Ïé¾aÍÒÆ‚ ›óªÀ).qÝœ—‚~ËŸAXÚ[Ê¢Pb/ݶ‰rKPËÀŸ³­“u’Ë9ÐyØsÿ+ Ÿ9»ÕQIŠFÿE4¦H YÑMiºŠD÷qý0Ú,á/ÅË"–€òM ‚áBZ÷Š hÔüLÖ"©ŸE@ÎͯFô#¡†lÖ/Р²_ •Sü?º"ÕÔ «¼ÂTTC~ÖHDQôöâÍ»ÿ uœ endstream endobj 60 0 obj << /Type /Page /Contents 61 0 R /Resources 59 0 R /MediaBox [0 0 612 792] /Parent 44 0 R >> endobj 59 0 obj << /Font << /F8 9 0 R /F11 15 0 R /F10 55 0 R /F1 62 0 R /F7 21 0 R /F14 54 0 R /F13 53 0 R /F69 22 0 R /F61 14 0 R /F74 36 0 R >> /ProcSet [ /PDF /Text ] >> endobj 65 0 obj << /Length 2218 /Filter /FlateDecode >> stream xÚµkoÛÈñ»/ÉÔ5dv—»|MÞ!îÐ|(â´(Ò| %ÚR ‡#Jq\ ÿ½óZr)S–ÝsaÈûš÷ÌéHÁŸŽ2eÓÊè(w.56‹¦«³OŸU4ƒÅß"•fUÝè ÁuZ9-£g;ûùâìÍyUi•›<º¸Š2­Ò²ªU™–¦Š.fѧ¸i':ÞÁo¿U½f’˜,·1®Ì _Ù"þðaòùâ·³wg_;ÖtQ¦®,ˆ5eíC¬gSU8èáM—6u¹r›§¶È™·kà ž}ÿ{Ïf3»ÏH§#kÒ²¬~ŸŽ ÐM1äC¥@Øj’è²Êy¨¬vL¥b t9‹e©É3è0%BñÖÿþñÝ'I–™x £ÍzbŠNó4Øã4þk,‹ëëæa3•V™}a3P> ì…ͪl(¬RN„5•éjLØü˜°õ®ÿ×›-ŠKK^|XD]¦Uñ<ö4üË…"–‡"êÊ‹X’=Ýö|‡Ñrƒ–ÛŠ§^¤)§(rPìFšç°¹ÂÔC£O©s\~U¤Ê<‰!öó¡…ÍøFu¶4õ€ø? ¿+Å~ºEa®ðßö1b9ð •—Ï!–«Ê4w¡Y­²råÞs3•Ÿ” 3ÎD®ºbùÄ¢ÿœ”“”„0‘e Yf©ÖÙ³Yæiˆèªa&ªLœÚˆÆ¨T¹ ::µ›$Ä%-9ÕG±†Pu…‰ßu‡Áм—E"Þ\q‹só†û«z)žœõý˜g Ò‘Î ’u;•Õz¹Áá5Ó¿E’„9 ‚[8Àh®ÅÉzí#k;ãi¾õkÓI’gE|1ï…CžjŒvXRŽž0`¾õ¹w+ I>qxçÔ&ç%¥6äõß|BùÔ–"©7çyXÕi8ÇË(É4WÎ`±FË$ɨ» ¡y:_héO¨oîBfSHÇ/yH~ËÃa§Eþ"«ßy¸’áL$XòôTˆµ>†QS&K+-ÙãRàÉ÷8š VÊ$Jº ÙäãæÐ ‚#,²ëÙ…¼¤"vAÚýÞ™€!i:æz€CÖŸˆý5ïZ 4 nü~úIR®ùF¼à!ihßD†º>þÖâ*Þä‚ch«øxGÚ7ý¾"ŒLBwÌüU1 ì•AÇ\q òúõ^¶‚¡ÆÈáúàÜQ4÷Œ8 ¸¾9¦L‚üº?ÁïGŽª*Ñ î¾/aøM–§ÂD¶MoÑSb¶ŒÜò…ï@¯5A³ê…yX#gž–ù›¬nOBîže?1L_#á° B|÷xß>^ ­€½•¶í—BùßJ*üŸœçÙy<ü‡v8=2c†KfwzðÙ³•ñ÷§f>ÕkÌô]-+Ç}47:þ$93p\!ͳp[•ÖI[H[õ û<‹§ßЮĨÊQ _ ]/ó²ãë÷Ë`Fe0«•6—¶”v ÈC9¼¦Åx˜ý|Àþ©¬$|>FÊ×Çrïˆ×ñ«“QêìØ}åT|Ïì…Lû í¥»–ñ ;‡î_á÷Ëèùt.*~ß© +—nïžb™LôâÄ/ØÏ’M Ä ˆ–e-.[XH¢üYêSàyŸÉ—¿ïu'ðÓ,N/‡9eÙ4éU?!ÿˆÞN{ØHTØSö,%§¢½œ@ï_J^u‡Éiü:À• þ<<ÃÛõ+ñd>MϼÏÓ¥8¢ ”~É/‘ðy™ÀC^ ø8¬tyäuh½;¬2$ï¼{#YxmÖË%wðqˆAã°i ‹tÓÁ¥#;äù†|¤{N⊿wÒ‹rðœDn‘-œ×8ø61¥ŸárÝIfaýeÆËí†Û«z+¥ÐŒµC½/›5Ü‚uñí¢»lÓøðÁŠsÞ'ëu÷`5Ö2u~°rü%VÛøç†^{yâ~&AÝë¹JÔòLÍÍÊ— §ØfèÁ‹®›-ÓdõÐ…lŽXv=×›@!\©?ð6,Î1}âw3ã# ³@áA+23¼vãÝÜËs²¶¼¸IW*l!ï˜Üù?mìèîx,\Ñe&æüÒï&°m´^£eüóá9ƒu-Ê~ÈE¬¢‹!_çâ/ô(Â4aîê§ ¶]I¡)ÉJ…ÅÅY/üå»âY07H{¿ô¥™W‘ `¬D•dE%õý¡r¥& «w ÿúŸIÙp1‰ÃÀÚoÖ¯)K“ Á™]éâ_¥ØBŒc.À»âEC”7]UÈŠnF´Fü-ŸrR?‡nÎ9Ÿná²Fe+hÛ0¥ðÃêX;Ó,;ç9Î*zRi"E˜:Ú-ãÚi‡«VÄŒ)[LE£Â%ÕÐÁ·‰Ë9¹ì›å|`D† ÂÕ7+ÖþhvL(™ÿ2þÖÜuu=QB®¨[oýK¬ñšÈó>–3Wñ‰ÐÒ`¢Å7ê¾Þö™±‘Å·5Õvž]€&{špeZ¯¹s ^^øÙ[ï»]Wj …ZÕÜtFòé­ˆäC–÷+ÞLÊ͇>Ówc~60M_[üÆØA)H§Ƶ×MUr‚ O'œDM-Öb–Œ†‰“ê°pz;ç”"UT.Ü+Ðöºñy_X­[™ÒÁÖöõR9¨.<,§ŽèˆÒUaõeï ìUO•f¤ Ì+õlÞ%HbÉÆ» /yÛ,Z{7§#rAœ.é6ˈÖ=¹Ðª0”O0‰ÿ0q5RŠ"/W<à`áäE Ö¨lT!cµñ»õ•K—5Ñ0vû«‰”·É›`Isl<m^½§B^9»{ø/ù[¢žúO×=Ô¼gÊ1{¹üñCUzôÈÈôð AvÇLM¬"—”EP[F™´,\”dYê*Ë:ÓÝw”ÿó÷™ endstream endobj 64 0 obj << /Type /Page /Contents 65 0 R /Resources 63 0 R /MediaBox [0 0 612 792] /Parent 44 0 R >> endobj 63 0 obj << /Font << /F8 9 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 68 0 obj << /Length 2270 /Filter /FlateDecode >> stream xÚ½koÇñ»~Åùr¬ÍÓ>ïa4j4F$(Ѝ ÛN䉤Kò’Š"ýï×ÞíJ'¹jƒÂ n³óžÙ™õû«‹ËuÖMiÊìê&3•)*e²Ò×Emšìj™}̽zðÍl®Ry©Ê0®ÔCƵz¨q\Á¸QofŸ¯¾¿øîêâç )ø§3]šÂÙ0ûB9—-v?«l ›ßgª°MÝè.3uYØÊÃx›ýtñ׋÷O¸ÔªhŒËJW®*™Ë¿íg:o—_àïí þœá×-¥Æ˜\0SŠf.™Ùgg(J˜8ÙÜ™Âël®uÑxÏD?Ìj—w;¤¼íàŒnšüÇ L÷³¹µ&ÿeæË¼=ÂÄÕ[P 3…³ÆÅ³2(TžYˆ9±¿ÀÝIá‡ßE»ÍMY7B¡Lè•…”X­b0«4ÑS©´ŸÕ6oI±§ÙÜy+ÐÞLðÍé¼IÄ21 ŸÑLʽà,ª7ümœÅ7ua|õÿrk ã|j«ÉO”h6O]¥jÐ)mÉ"(u4"Øc ¶W56ˆd_׸¥sâ7CðFÄ–6Ùä–™=AüÄ@v¢ü±êCºE€\—Ê&Êð††ÓB£kõ ,(>5)9¸cÍâÖïa8ça+vè{ø}Rpçê\Ãï-C-…¼í¿%ûP† à"”¯ã->tþ›)–héN¸#uŠ:ÄYpûšÝ‡AŽ_G¶Z9$¸V"G/~!;¡zŽigÁñØ„ÏDŠ< ÊÝ.zåOŸ-"žöƒ|¢Ÿ‰dõ¢uþQ4ýV´ÿùYžü('Þ ÖN_òç(´)÷§H°H£­l¯Å-Ì”PþÕBÙI¡~#Ö4ˆúœIþ7¶Ý$Ûÿ-ÅoBæø^æ»)‡éyK;Æ|”kpºÔ-ìχ"mmƈ|ɸ7Ÿ4+mFûQ'ïâ\A«\¿é“¢ŸÆûåY¼f¯QE]›4ÐR%<—"c‹ï9wø¹Ä†\Bç¿{! Äg>Ç^úñÿÑ„±^ÇüTnyè—SþŽQm›XFxü×W¶Ò>g®}áÐiâkð,2ª¼ÿg\8U ïs(E¹ìl¯C1†?û>\‹qo¨T`Ì¥lwd4ý5iBŠìP.`­€»Oj©ƒÍMÜ$•vqûVB,8ìïH OÊø§²C‹çu`°Á$!WÂìŠ.Ä[¹ M åËš >.çf32e œöûN ­f­µUw±=–5 ʹPË(Ët»Ð't§3ƒ`$oà»Ð⾘Aaסhm÷‹t{/”B5ã›ù ±ë5©?VP='¤¦ñõ-Çüc¬·{T3Ò"¢ò»dN¨šJ –p*Ò$`Sc¥GÁÍFu"«Ž³Vh4ÖRŽ ›èYV©^ååÅRWfí„ï%Eg{y hÇ ÏþšJ(­2gñãÆj*bÐ3nÃ>NƒŽðl¹–PU0.8(U½K*XÐcƒhò 7 SÛ €kŸÖ¡—;s] kTw×#4£Wˆpš8¯%ÃGUø=:8“oTþgi5P½c áe›ðÐRnè©á;œ?jOsµ`Ü·<“LPë¡@ÇΦ/(¨^†Å ”C¹ÄUÛÁ`6ê¦*jy*òUcÀqùŸ6¡m:²¯YœÊ%Œ hì÷«PÕ¤I …ª ’ñ¾ïè †¦(öB¬N½C˜¢‡‘¡OÀDzåÝ“À`:Ùn|^I°‰aÝÚLU` !C¢"GRœ»ÒHøà jù”ô‡ÂûãŽÞ›ˆ}1@í§Bášï â€ážê¤ ogç`ÚÝéÅæ2Ñ(€ Eté ]Ö™)Êðòý¹t|ÏÂu¬+p‘ .%óÔÞ¤qáNÂìl~1“>’g—>¶ò@bž§CÃ.=¼]t´_g°À8Åuc”pìî©Rä«ÍÊÕT;˜òõ¯þbyúÕ¿ñ…·]U]Tµñ!ê]¹r•†[Þ.U]%ðëj]–ù\_zVÉû½Ç( Q È^¦n¿Øz³Æc/´-œôh©4–e¾1×t-x—ø¨÷é»"Ì©²BÓz;™ü6á1 êx%Ø­°J_âÑiiH.Õ.äA—¯‘ÔýD(€ºŠ²||Ï UVMek„ðXx® ׈Bß•…‹·/t\¡M„åÛ RøÌ_ê”Ø›‰RÑ•qaúóM¢º>zÓÄP ¶nBGÞ;º{ïíòã¿Åø&½ÅÆçcS7RœÂ@ŠS,:èU²âR©»,ÞðGªSL›ºÎOˆ‚îÚEH—‡³l9ÏÂxâ®»ñ9y¹Œr+—ì°ÜÇ·#)©˜¾Bî1ƒ£ i» ¢0£ Ü·PY[GÿWCúÒ&䓩σO endstream endobj 67 0 obj << /Type /Page /Contents 68 0 R /Resources 66 0 R /MediaBox [0 0 612 792] /Parent 44 0 R >> endobj 66 0 obj << /Font << /F8 9 0 R /F69 22 0 R /F14 54 0 R /F47 7 0 R /F11 15 0 R /F7 21 0 R /F10 55 0 R >> /ProcSet [ /PDF /Text ] >> endobj 71 0 obj << /Length 3366 /Filter /FlateDecode >> stream xÚ½ko#·ñ»…Ð/‘zÃ÷’i¯$HŠÐ&N“"—:yuV¢‡#Ù¹\èoï<È]•Cš¶Hîìpf8/Îì'×W}&QD¯ýäz5‰h|˜x×mÜäúfòÝt¹˜©én¦Ãôþ#1›[ÝL¯oÛÙ\;ýú+üuÓ;øñÓýBoàoq¿†ÿûC½™Á;-®¼¾%T0:ò‹ˆµå1.ß¶C„°ôÀá_ÓMW0>´ðï'ÁÁn ÿÞòÓýŠwnKÜû–g‹× ûëñu—9 Óëbï0þ×ןξ¿þÛDNæJ‰è:yì2¤êx?› ¯Òǹ =·MæÁ€,@úÑgJrWR ¯½ìä;gse”œzùÎçq#ß5yä»QÞÆL£|†¸®>½¾úé ¥#'j¢B“q¯%X,·Wß}/'7ð\‰0 ¦7ºh£†ãÍä««^}2vÇ*‚b:3ñÀ¢— SùÙ Ú½ÍĨTŒÓ¹0·JÂUçFšÁ\K7˜Ëfø\ÊH²‰BƒHX®ž7}AÛ±ŸVx/6즼Oy¯îoÕOÓNs%AÇp//là­¾ïn£ïBŽ›{Ú\î.GzÔLQ{4íéMK‘‡tÓÇw…°f:ûår –ÿ%ŸÌXá¥gü³@¿žèŠ;Ó"¨‘Ù““M–¬’:vÑ9ZsD`@3²}à ’µFÆî6™m°%g²ˆÅ¶Ýp˜CDèP¢ ŒŽ¤pέ€TþÇ?h¦ð‹–€ñàÌjº„ ôô ·åUòÙ‹r“ì‹—÷)î¿å‡û—R"ÂéÝ!½×I†"ñ=?…Z V )›Þ+‹ìNŸÛÓóónà­+`¤%¾šŸõ%y‚ó®ËHJHý–R”ûÓ>HÁž„ˆYV3Ò*û2M®spZ—Pþm0…¯ŽBBîf ¡ÈÇ^“¨hÐî ¨oO%€˜|)ÊnHŽ-w{)¥<3^hÝŸªåÓzú|*²iPrìÆox¹P,-íô¥2v¹ïíæø÷~ô‚lß9›šEÊ'÷?ôËor ¹íôVÑTZ X¬Êè2æÆHVh†Ùà£Uÿ΂î’E¥|6˜ƒZu‹”tÞuâß34SR5ÌžWKGÉžuð@0rÈ©÷¬ ¡éù’ ~‡!Q½¢„»MxÙԳ߽áÕÕž']²NdtL5ùå:½·+dŒp”•Þ­¸;~°MŽW7¼´H´’amròná:’$yäì)|í_ábýç™ó#C4´h¿¹I‡ |¨0l¦¯Û]6mÚõcØÑY:L¬^ ó|»H“uøˆã‚YZÌ:Ã7‡"kIoÒ9&’ êrÐÁ-ûX8PÊ0Ý.Ã[ž,îèâC‡½8Ôï8…ˆhYª Þí¯!fì;3Lª…ËdGÙ¥p°Öv9ØŽv ‡“tÝ;’6¼J§ÛælMD;2…ëRà]²Ö> DÕ ¶ûC¶¢è¹N¸ñw‘È<'9x‡9Ñ|ÞÌ€2©=/и¯k-.uŠy’žÌ”€1HCÌùÉiþmEv¥¯¨ùë *·¿á]S8UxuíO¡û çY ŽL>M…fŸÒó‚æÓœpfx—yz3[ÙÌ oÜÓwB«ô%»éb7WÝM¹ðäq(§)}^‹0[S9˜‚—‡«Œ1©@Ê¿éœ˜Ä E“0H£@¿ÊPð/¡i³kÛvF{ä÷7Žïdøl?²]T6íä¶o³ë‚ùŠl =Ã_1Z™éß[òS•ìÂI!u—¥´Uö&²ÝYíðÂ>°Ó{Œ±@B)žéªH:nh@!lÅZŒ‚tÈœTiô±{ž2èS]TØ„ Þ=fYàçŒ ËÂÁ½¾ô³5·6¼"¡ Lc¾†¥40gu‡äù9AE B´¼åm`%AàµLÔ$;ªrƒæz{aüú)e«jiÇ·ëóçöë¶Ó\Ý6\ ´ ¢–ì $‚€ƒÇ‹É¾Ã™ƒí$°þÓl®|ÓmZVVb –•ì:Ð@ €­…²þÄk×kß&ª÷u'Ç0Ôå³§`Þ{ÃúA˜³"cÀ0Q™HHSK~ÎÅ3 ;Iÿ^ñìÄt…´îRwèÞÛA 5 ý>‡záŽg5Tu€JCÀ&_ÞA=Ún)+¤–^Û|ÇÐýCƒñˆ"siy6²O±qú0®JÚ¨‡—XQ}µZs¦œ+Iô.º]õs›ïÏX¤ÃÇt§—lh¼,‹¦\ðAäw©Š²}Xlû=©MtÃc®©Sä°XwÍé>å¼Dî!õ”2µ\jáy¥ªðš$rXt([*RQ¹ëNÔ _÷=gzvìÒî^”©.õy.Á…\þ8ð;9jí É’*9Ç®´>©žYnœÑ஬ϴ7Eͯ¿Íà°+¹>äÎvûS"µèÏ7#¥¡•²$–»f£²Z.Á@bÏ%ÚÕ"wÿïy5¿ìWŽ<̬r•q[” ᱬW6Ć¡bÝM»æ>¼Ñ}Iߦ1¥MŒå˜oú‚膸2)1µàNV‹e:ÕýÁ‰øÿÐ÷«®G¸L»Ò¥©“ÿFÔ m_S“,À4M§kê øS­H2¡‡÷]'Â'BÚüIU–[Æ9®Í"úqmQp[¶áþS[ì±çåÿö=5‹¸¦)¸¡šH¡3q¦¼R.-QÒÅ€Y+KO4®äºJ^%Ãd‡C])7Òd‚àËrö\©™åX‡’Ä*é]Pâ£7ÿ(ûÔpZü¼€øÙj±KCëé¢JmF±ëÚô µ}™pjô~˜SE Ý1ÓÙâú–%Í¢öŽã®Œ“9ÿ¬’SÞæŽ¯õÙýæf£Ê…Q ùíû˜êÍ lYbµ®é]òH^ ¤ÜÑ=rd»XÈ˂ӈӟ/ÄX °f„tÔѲÆTúgÃÛ×Ãvmoñü‘¥œ/.j7YÇjªûö« >EJç=e˜o4¸Rk:àz¾ÙàøÈ-0'Ýàne[«DC0”ÂK=¶¶œÃ6¦°ºfMl†×͹6öÜ÷7Ò]üý;“¾¿éµæÑ<­V©Â˹O]•J6¿~?sf¿æÿ´Ÿýùs53µvœ¬bzmßÄq’†ÇhŽ/vß@@éîè°üsîfR;8EvºéÛX4ÝlßÎl'ëÍ:)ïÜ<ôpeI¦ÃöÀÉë¸û}r|o†æCÎp¯ÓgE]â¤pѹʇMr¯¹„»ëë\ŽmyA‰Ú×.0$_܀Ѥkü_øÃ–UÑZ4¼ôgÎy¸IÅý@W)“[æÏeègôÅÌk_ÌCËûå 8fåŒPuWÊqN™Hå†c\ ‰ê9(°€ ñôxo¯¹ÂiÇþ®ÁµéóÏûŒâ¬/9TÓ geOùóøí IŽÇñ \SÀ5‡´:¡G}Ò(Ä…Á¥?›þ4áëP‘ð‰´¦~ŸæšöÀwOacÛ¦ySÁ›yË4¨bÞÓ¯†š>f<ŒUQÀÅ÷l¬]о¤úx¦.‘ý¥¯Öc^áG%a˜B°6½Õ „§ X_ðÒœÿ†B–Å4…ÌÿÖ¦g™“æ–ðð_ŠF,\"Ê1)L6/y·?:$¡FAfÀ%Ȭ©.íÆL ™ IÀ¦€‰OC™×Bz.‡Ü=åœ?ËüœÚ¡˜ endstream endobj 70 0 obj << /Type /Page /Contents 71 0 R /Resources 69 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 69 0 obj << /Font << /F8 9 0 R /F11 15 0 R /F10 55 0 R /F1 62 0 R /F7 21 0 R /F69 22 0 R /F13 53 0 R /F14 54 0 R /F12 72 0 R >> /ProcSet [ /PDF /Text ] >> endobj 76 0 obj << /Length 2178 /Filter /FlateDecode >> stream xÚíËŽãÆñ¾_!l|à:!M²ùL2ÀF6Ø“g€8ðúÀ•Zy)qBR3ð·»^ýQ³^dS›ÕÝUÕõ®Ößo^|ñuÕ®Ú¤­òju³]eYTm±ªÊ:ÉU¹ºÙ¬¾‹þö*.s­_eÑÏžžžžI¾ß¦e ¯µâó'x2zZ~Ó’†¹€?4Bò- ÷Â2³4óøú·ð‹³šøþþ柫tg*Qªà³~f'y!¢;ͦ,iKÐwrÆïÝ‘ÁžžžZ®h©àÈUÒf™#<\]&yÈ!PÌGìðÞËÆ`ÞèPxhdÜ ®/¾n<ã‰3Ã0Κ¤)£¿¹%©Åªh£žºždRG#ë±Û“yå |Â|í'~_y-’íÉ^íÀÉ©M‚?ô=âG Ö'; \Gü±æ}ŽÚý«²Š:4ù}÷Ž6[Ý´H×j´HM#ôšµfì»q@–ïÈrT²è„m[©DÖó€Ý0^õÛ4«Öb¤šå6ñ¢ã·¬&°2ÏÉÓ2©TT…Ú™&«ÄÎ^!©TxU4Œ øT–¤ik6üpŽ®NŠâ ¾’ñ±Â‰eýŸ:ãtÃQ„«$œdìÏ$ œÝ³¹7;Š`u¨g†“ðOhã0³A „x& tC0,jÁAôfÙ”€8…ð¹8‰ï«0P¾¼° õ Ï—ò~ébÝs[îŽFú(Û$RÒkÁääxÈ›E%|þ†1{ ˆÂùVȬyö‡üc›{áùV$²Þ?’©ü¿eŠ>ÿÈ/äc'3ùï®ú$ì~þ騭>QZ?:µÇOÒøÇ|‡©þ\Èaº§×yÆ·Âè¤ôé,ÇÆáîÿ¥Âÿ°T(dã;Šcq^qêÜìÙ UGëAj`•441œyÍû]–Ühž5)g)ýÎ)O3/ ³yL©S{ãQ”-ì7Bz£Š'yD¥y’§™I|ç¡õ¢1ç ¹†[\r–sšyš—IªE¨¦M*ÈÞ¢¡ÄAÉXÝã&UÇe]-Ó.‹ŽlhµL/+’¬mÌ¢R©¤ûXö¤FU o”ïFKª¯k›Àehª-øä „š!ÐD–b‚ÀA[ë™°¶jêèí×¾eth>݉B°á…_z<àÊwDgèE¾kž#ÂH€WMR†²¥zS5Á”)…ñô6O³)Üênæï£•Q3@j`©Ššh‹¦œgPeª˜‘áGçQ¼ä5@~_âê‚«V<ä©·åŠe``VßùHÈ‘ùþ$ASo¼™Å³ÞÉ ¥Ü½Ÿ;+ª4zÊÞ3Š ÑÀÃq)4‡UiñyGt¤ŸÖÜÛ½°˜5Éæ-õÛ—ðêÊ[œ"}PAÝK×4& 8§±ÌéÖ³­¸hEÈ‘+ØÑV¼‚]$ h Â÷xN.ôýÄ)¡‰AóÍ€ôÖØö(+ù…}Ê~fÀêJ1±"öŠc”Èqg,{ºuÕ3Ù%5l}X©›B: }½-*)£s¦ø6ÏÒó¨€#µl¶É­q@&㵇h/3OIOˆC«çZ: ÏE^DÓÀ³o²+ˆØÀ’á«&ß.<Ýãxkœ©_jT•4ÐbÈÑ¿]˜*I0ÝìùÎLáŠC 4$Æ@Ë‘‰ ¯ê;Ï×T½ÉÙJ®ðUr,ê•‘?_D[Š£]ob…ØáÅ‹ –@JQA`âp._„[=Š`k·Ó<(ÓŸËv±‡Ö³­K–I5IÕ¶¿Ù2_3y›[õâxz¤`Ÿ×ý„·ê’áZ4(‚GþîÖìì4žx»Y3œv·BijÏÜŸOàš S ‰Ë8ÜRªZãìŸÙÖ½H únr<ZY9ZÞ(ÖNÞäÆ3öÔ3%Îá”wWA»:Ÿ ÖHÖÜ¿*˨3ìhÞáçHBŒ›)­ôŽš—ÙÃ]:`^Ù˜M…ܳhß›&fe„_6v‹SÀ?™ ¾kx*³pN3:²ÅÖ(nU6V¥$¾Ac©Q^§9{º «Ê³ìÅÛüÒQ6Pò™6ÌUå®Jâ²È£“ÖdQ÷Î\…€j›Š£5]Âtíç&X¢'Û+uç7±~ ê]ÚA¼#ò®}¸ŽÊÃÚ⢿áÊÑ]ÿu½©l™ê_`k¸5—2¹l¸í\ – '¦M¢ â=LKáƒÏ>ºË)âqCrd/£áâ¬þÍáë_†ªÊŠhÝ™+ÊO6¡w<ï$_ø=×h¯TNÉ߬é1«¦àðt¥]¥U% ß#¥;ÑcÐdŒç¤1‚ç/@M:(U‡ÍñùÍ’¹Qš¦KڳȪÏßf›Ëµ CÍ–ÜÁ_èºYösÐ{YM\‰9@ÖFß胩ˆExܦÐ5qÍ¥HÀŒQÀHï{ íZ:«QâÐgÜ’ýStkÈÙÚ"ÇÏfC˜¿»TöÃl#¡´vÚÀΈù!Ëãý|ØdÔrU­ÑåÐãSå…^àWûDuçõÿv½_*­ƒù'|Y,Æ«4?iIbŠå¼ôQ6 Q°FÊ¥ØÝh½;n³ŽÒ½MŒËd0v¼Í/úbºçêµ \ˆ 7ÍŽ_3.¾›ôlç¤1ÓGþ¤>ÈݼÑ>C/4<>.™ê0º¿zyN®LzÒ¹ÄE¡@{—ÖÈñzÜi÷?„ŸŠˆÃ÷'Ed‘/SI1üCý f¹›ò¬èŒkÑ”8³±­UV×`%Í ê¶¤ö$“8÷â«›¿ÛiäÊ endstream endobj 75 0 obj << /Type /Page /Contents 76 0 R /Resources 74 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 74 0 obj << /Font << /F69 22 0 R /F8 9 0 R /F11 15 0 R /F60 16 0 R >> /ProcSet [ /PDF /Text ] >> endobj 79 0 obj << /Length 2612 /Filter /FlateDecode >> stream xÚÕÛnãÆõÝ_!l^¤"špfx ’›´ l€¦èÚ@ìîMR6a‰Ô’ÒÚÉC¿½ç6äP¢ §h›öAæpæÌ¹ßfèïo®¾z“.2•Å&^ÜlY¢’8]ÄQ¢Œ7åâýòæ~¥—ÕêãÍ_½‰3ØÊšp0XP=üð«á·ƒ_.ï²Û§†* µÛÜ#TΠ°µjV&…k«SBÖà0#\Ì ½^Ãðõ5 ·«µI–eÅû:üŽˆÞJ^!4¹ Ï·ð¶åq»ñMÁX«Žaº@dŽ?”˜¨ßLìØÚŠz'Où%,^!²9‘@E‹u)G‹µÖ*‹Ä¹C˜÷$—mƒ[z~ùD’kQå/áJB6h»ŠÁö¨…ÄOÅ e§fz9òü ¯= X‹<‘3Ü#Nù€6®Ñôg¤Õ"•r¸¯E‡ˆ¨‘ÖlqúëœÐ¨ý¨·`ª1TAÛ¬}Å9º‰"Öé‘•¨— T¨ ‘F· §ØPí—è@Ò‰n¢²ÏE$~ŠZÇé¼8ò|+ Y´»½Ì*ž:ˆ3㮟W©•èy†\zPDGrx£‹¨ °‡qt<½—í]—£)ÿ¸Z‡àu_¥¡#üXSPà;…âƒ'OM8‰nÞ³›Î© É¢å]=z9L„c´áK#*À¶vAÐ~ÜF1ù`µk;±æ¯²¯h‡0ìò^˜„ù AîßQDË{‚[pkTÈž£y)áï{Í|•”É GMB»çEçv”N*ž#_ºØ#]?ñ˘U´¡xD7´“\¯dl6á kruÑ1ÿŠ^ûÀS>´=ã¦9ŸcœŒñzöÍ 9?ÍK4âPøÙJ’O‰3ã2*¶éÜ$ÅØ-êa!TBÐý-Ëà‰¸Fµ.üJ L\M9.ôÕÈÒጻž# 2\Øk.û&)þØ0-t©ºáç¤l@"`/y»ÁÕdér)B{šY<÷Gò6ÚÀ!ëÊ`—sœŽµb€Ùí·K€fÓ’NàB«z”¬#æ¸}èYÛ6°ËGJ¥Òº÷èt°g§o›’8™Iš.a'á4¹žT}(­‡ªÏúÓ×.K)éX/g•? T¤ÒC?d¸zLþÀŠ!CÙÄ.ß1g¹Ø§˜ásu`T¬c¿1Ùȯ"!«glÙHaò¶¸`}W1axýúZL/\È/Ä;‰s`[¬£4‡ù­Fõµ¨òŽ]ÙèT…éI;PT._‹ùC;8$1êhpPo]rt çGx —¸Y¢äÙ‡§%YI:¶ÿêеüN² H ‚yä F9%£ŒKà\ÚF³híÛÅX•™Èú+PÁ"VYbS„Š(ô¤ÂLÔTŸ›8Qaœ,<˜o™¹€\·ÌÛÐÑá *–”©ž ] 4ÊÑ|efl 3}RØqõBaGà¡Bd˜Ÿ8%Å&X¾å­~¶ !3i§’Ц³‚¤8^!”,TqΑÞ8c«ãóÃ’Om÷<çÞáðÓŠ”'|;Wþ­„Ñò“ßQ±#ʰ‚•Ì@òAÛph·pÁé7z·²+w<Ñ:8Ó/û¡‡©·$·ü}µÒ&¢ á_ztÞP \š(œ¼%ÄÎΓšwš¢Äƒu$ÌÉÄm<âמ¯ÜIf¨¦)¹Õàš\ÿj|Úx ¼¸ÒBâ/Cú7='ì¿^¾Y¾ôäàú§±r]O ÚNW>ŸD ‡<»7û0ºÚ[øýÀî>Gãïn¡ŽÀl©MíàbVü"bŠ©¸Cày¯½ä…";"Ñἦ³³°¼>¹˜ðOÈiωã¹ZÖŒö¢Æ!‰¦¤#!ŽÓ™ÿcP<+)ׄV(ˆC‘>•á‚ÑL¥…ž™Jo]áwø`¡ÝçmS¥!8ŠÝÕûÁ¢„5àUÙ,]<än¡ãX™ÔÀx»¸¾úÛÕ÷øŸF/"•%‰Ü§*B@¦áh<¹q0Á"¸@󕃆“¿Q©¶“ÿ7 /Ç? Cíö¨ˆ†/mâYF×Üïù¢ä)ôÞꮫø>ÝíÁDé;ï¸â¡-V&[>¬âï¡äKxÏëeÕ×wô‰­<#Œ—™ã’Âîe‡' oi寢4-Ax Ýç»ñ¥‘þ¾)c¡=ÈT‡¬²[/(#»ü™¼G±£pùf•"1X}’ïQ(ß'ô@÷+0ƒ¬¥Žµ^t·3¾¶(úh—wõF¾Âûø­ä—áª,â{¼ãNð×1¾ï¯dë@Ÿßgn"ðˆ$¢«¸4MYL=¸ì?gåäÒ endstream endobj 78 0 obj << /Type /Page /Contents 79 0 R /Resources 77 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 77 0 obj << /Font << /F8 9 0 R /F69 22 0 R /F11 15 0 R /F10 55 0 R /F7 21 0 R /F21 23 0 R /F20 24 0 R >> /ProcSet [ /PDF /Text ] >> endobj 82 0 obj << /Length 2922 /Filter /FlateDecode >> stream xÚÅËrÇñ®¯`)—E$¬wûJ%vÑ–uH9®”ÈŠ+å°KaA1ô!ßî~ÎìKÑJ\•¸óèéééîéîéæ·×/¾z[]Ôi]ØââzuQ—iYTE^¦Öå×Ë‹$og•OºÃlîr“g&Y·Ü^4½´ºüX²Uò0³eÒñÀVµŸp´Ý2d #ï3ãÚÅ ‘ÒXí.[ º^&ÚÃÌkìLâcÔ?^AûúZ(êö¸Éñ0£ÅóýÚKÜsâ4¾jpóm\òŽÑ/à@Ð&6ðÛßò즟ýóúÏÙÅܘ´Î…Sx|SšÝpE=PEÄ&S$»†ÒsïØñì Çv¸p‹õK˲ډæŽðÃÑv¨ Ö¤¶V@d¿Í’†XÉÛÝËÚÀÙs"'˜ ä]]^qcÈ^ÄtKlÛ¦YQŒ¹»‘õ®`®r³áψçÈâ9LÏ]ày\Hl‡/³}<[‚ö²ìŽk•¼¬bÅ™`«ÉÓ¢ôÊ­­¨ßO³"W0˜C!Í÷x†t6÷¹OÞ@ó^)wL›²xÉmI᱓' ÄÕ6G9åD#ðšŠI1Ã\·:GÖ⎠ù²8ÆÛ´"6ïÎ64µKÞçÉr=ÊòBIq/‚È¿á™í”P³â}•4Hî bܱ.f  yV¾TeØwÀu_¸Ö¬˜2왢<à>#Õž«YåRWm|Ÿå,[ =Ã4> q(,ïkb•¬;ܼݫ½µÄ’ã‘× ŽË«—h Xƒ%7þé[þ γÀ”1ۣܵÃ^õ¡e4Þ¥ÖajÁÄïöj»=)[O—„Ž Ìcþ[<4HÖiUòº<8’3@ÀÕýU ô| à‹X>Þ7d¹ÄÃ(öéLežÜ@3 “|`-â»ÔŸó• J]exßkUâ „Þõx“²,Yòô¸ëfÁ“ˆ6ª*1~3º´™xÖŒg\Zf„|½ D~MB&w0³d 3‘ÆqŒôŽ4t6Ïóä"a_Ì·V db[ Ï"mzw3vü°WÏî ½ºÜš®ì¢U±ÞšïøRßËîM$QÁï‘}?Å-·éØwQ|¤µèüÇa†bˆ\b öyIìkÕA·2Ça ø >˜÷µåQñ·GeMáKs@=Üá1íØZýŒ-yî†v ÁL!W\ìm;ŪI í—³q%´ïô,¨97$¦ƒcÛAÂÙ©ûpâlr ¸!Ô“á…a ÌvcÐþ£ˆ¬áEr!‹zp1 4-zqüÊ ¾u{ÑÍXý‚¾K(ÑS\HÖª âÔém(N`pl¯s¤?¢=ïô'þ¼E³¿à‡&ì{ÞÈðìïfñ:ªk½Ñ€A6²¥òOUSÉie~ÌY6¶crò^HÞ® ,TæËäО3Ȗر’<ûþ6ÑØŒL}b9ã„&=A¡ƒÓÔ—Ô aâC ø$KÊ1æbØpúHONÛ§§Ü)‰.­}Å$^ mM°PQ;û¨ã+nþm —iGe!k¾„A‡,ì*®å/ÂŒeÐÁ  8åöXW&˜ü2û²±œ=b÷â>\Bê¾úß¶ó_¸ÝïŸÚÎd©÷ã«ü.<Øô¼aI›¼44wE?Â÷&*Á•XÌÝ™zàÌGˆŽLmÀPÍmVçÉ_E]D῞²aAçCº„Fh(°šñ×E±ò³M°Òž¬Gà:®÷aˆ+4[^–·òMÖGÍå— H.ßJíMØ(D-™ F^Lì&®Æ{Yo…`#üµ²D÷ó|V‚©eM*F@aýO©Ÿ Ô)ÕòÈZ[˜:HR÷J)Ûç…—.²; ä0j%1£-΢عzkx¡Ö¦:‰ea=ŇEÇÐ¥—ÆcÜ×гæ FFIè¯4<‘¹µ æG04ZÌQ zäwpá ÙE… AÝA’gtH7€ÕMn8ÊFHÎDE°æ†#V>&ÅmÈðL?k€©ž/Bî$žŽ;ÀX J]9úbp&C+ØêÈŸM 0ú „5¯À´5^ñ<ò‘´Ãy“ŒÉ80Ä'¸M̾P^P=HÒƒ:îùÑOÀn_3î&bkuÙnMNA²¬œ²›Ê’á~ŠšÕr)¼ñk-‡y936§ŒHaYBÒ£‘“P%ƒ©E‡,¼Ó¤Î±¼#Pê.D8’Õàx·Š¹–zc×$Š\HØÈß×Àj|\§ø}Ñ2.MI¢ Q_‹Àöñìâ9:šˆ WôŽÂwx懲Ä.=üóÈòÓÜ:ämø7\ß¾«+ü:z0aÏÂ0’9Û>òtwC7F#¸v)75¯‚±ðY>’ bSÑb»á©/¢¤k25Ào%eÛ,?ˆÏdUYŽMéÖmK¼)BÏ‘û’Ãs[žHÛ» mïQ 4ÔìåúȃÑû;!ËÖÈ<ì SSšEô~|ßT$¢ ôú\´úšÓƒWuÔ»`#Ɔ<´4N]3XL“ËùÑ´i‹äL°Ž‹.¨¤ ¯ÝÒ¹ñû³á±…¼ÒùMJ¤N<!v6yHHýv‘½ÆC'uŒ:5U¥»¥0Õeêêj|ì ´x ý,N–¼—#É{g¢.;£0ûöÄœá ™A}&·jpKüŽ 0žNb†9B ñæCþòü°eêKÓôç(àØYùI‚>)À« QX£ßNäÓ,/ÿÏ4@ëÏ£ønŠ••sƒŠ‡5ëfŠü䳦¬iT}úíîâÛ 9„èõy¤ ùæÄ%Û»=åßÀ³jb,$f(¹(™L°Ÿ“.¦˜ÃHŽQólœˆß„„­lÅ9CÇvP|ÈNÉD»5¥š_ÊÔnqÃñûg³ä©¡9ìâ27á~á”Zx¯G<ñˆSZ`R[„ò”£¶Ì¿[M!h¶› 5¾6¦6í˜h@rÛI©—¯~ÏëH$5ȾA/Ä5–c'À:áà(³1Jš° …O3¿Þ*P^óýøò4sŸ»ÀÖÌ|¤ –󊻘Â>h .螘D°&ûör?àÉaéY ¥Ã€äÕ˜Q~  ^ðÕŽ²#ÏÛ,1PAÒÈànEá—íæÕä`x¢cºœx*´ôwtÁp’<†d!Ÿ½‘ Øs“jh=Oˆi?f¡–^jàCáÆ²Ðà_3©¥ÑÀ:è®Þ$?Ä$wÏãbb®þ‡w³Ê%\…âÐæbY9秃ZÇýR«wy²'ô÷d熕ÙK¶m¶T/ôX±¢'•Ô¦ó wKÃ7d4[Ù+Šùðìƒco}zý{²s¥Jò‰¢`;2ËVE"g†­©H!娅Ûö¨µ,;vwÑz*EV¹K­qm+l΀â†3ä´â3(¼“ ÎßQb\¨cÃL£K­ÿØD „8¬k& Òi™efì »'ë©áeÚ]P}­HñÑÀѬc@”õ“6_P0Ì¥`øÇö¥ç |¯.9«‡¥ÁãÌ­;ö 0й‹ož­þ„ÿO#ñ´ÖE±½&2\ô©1Ò~²PoÒU ÛGž[ª.¡¶1—Jš,]K›ÿ3#Ô|ú'‹>„’k›Yà¡{^¦™±’Ф,Ø‹ï¯_üŽ»hƒ endstream endobj 81 0 obj << /Type /Page /Contents 82 0 R /Resources 80 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 80 0 obj << /Font << /F8 9 0 R /F52 8 0 R /F61 14 0 R /F47 7 0 R /F69 22 0 R /F11 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 85 0 obj << /Length 3400 /Filter /FlateDecode >> stream xÚ­Ë’ãDòÎWtp²#°Q•Þl;IJÀa ÄpP[òƒ±­F²gèýúÍg=dy8ìºUªJeeå;³üõæ“Ï¿)’§z]¶xÚ잌±ëÒdOE^®mš?mÚ§_¿Z“,ÛüëóoªGæ)aÍÒ–‹×¥Y¼,W¶XtËUšÙ…ÁGº8ÂüÈ3WàOúŒqî…<¹Ã'Ïð«ïXÚjñÓÒ³øÊãjžáß©ƒŸád¶ø@Ø¿ÅUßÏ#£mYÁ¯.H7À߀ÿZ^<Òâ,͈¬%xžåØ%!Ùâ—¿&&cdÀ6àÑʘu ;ušF¨áõ Þ0þÓš¹IhÅ|_xÚÿI@°¡å‡ èž"b$ðv¥" x6ú Hæâïyb$ÉìÅ qJ¶Åƒ#ñ¥^›ºˆObÖËUV׋ð£–EG{M芥 ëÏ|>#©Ý{¤Š¸ø -L¾äÕãØ·8xMüŒ¡Ì—^$Íñ“È5Eà­m·ð“˜Ñ ÆÝlqˆ„ô— œÝyô(k§BùºÎ ³É ›¾YV*äÊVeÌ[$'\øqâ•w|P[,á ’èP ·ñõ;þôã~Ù2ÙÂ@:Îxe´;’ó™á~œE¡›ìNØù†¶Fœ§ é#@ºõ“Nö:Ë"œ ù‚îÊÚØ¾¢ã²b»ßæ-$MצHÕD~M2‘åY©0ïîÍ,_纊RI‹Å›f‹»xc²ðZ׃ÚRÇ Žt>:#<‹¡E<ïhwô~™ÌŽN¥„0W¶*<ê-9)6^6xR´–ål{KisêBjÙ¶ û¶Žèˆ¼%~ô VYµ¶&–À·‹mžOjÂpüƒø\oßÁ˜Ù»'±À?nÊD"£—i68q]ll]+ÞjÞÃTɳ]ü‚z¹£Úº‡´kÖE§f]•õÿÉG§üh.œ`œÉÔà©mXÕúà&rð÷È#}:õÈEì‘ߢ´ö´ã¾¹ªÂ¤^pfê•J¥ÑÐýè?ß¡÷Ân„ß"<ÙðãßË ÆWñ 8³í‡f”ЄóxÎ {‚}ð¼ÃYõfX†HíxS7øÈÁ0.€/ó²aò(•ÙbKÔÀН¿’ƬòÒ,6ªƒ°6RâЫ„a†Ýß^ÉW¶(á›æ#Ã9îË®±êS°}ѨVåÈXeø€‡.&Þh[7O¶â›q„Wà ìòhcÞ@„hÔ‚1„à{˺ŒÃKåï_è[=%e/ü qa1ª#è_gþ nà0ǃÞ9IW¶r ~‹¹0ŒƒûÊ)†ø‰¦x$SÄI†ü‘sb2Jj.¬Ù$á^¥p~FÞÜÞÄçàîÂ>û µ2#Z•ˆ~ìþŠÀ±cäíC‹ ̙۟}É¥„Q†HzçSó0AãÈx^Q ë®LìßÂúWoáKkYïFlj/L´"aL]A!<Ã^zß«&PJ@¾ÖØZ3•ñmÏKW= lˆHì&I íߪ“uÙ nÆ™¡ÐL¡•Í$šØ0½àhW‡ÑÎd µÙ$à ýÝ…<à¬I yâ+îÑXÉZâEð1|.ë)­7Í*))랣H×UjjNãLâc¼± K—¢o»‘ŸŽÜ™±ñ*İÿÞªÛݤ`M —¹=Ï=UÉ€ŠH”¬»a‰‡\’ Ó Õ¿ß´:ZZÈœ}6çš2wa$Bià¶?æ@ïì= ^¸ï£=¶þ½ ¢Þ6¨–´ˆš8ZÛ.% Òv§fâ7)˜ -?+ªË¤34øÚ² ƨªGbvh=²wÄ=ã ½„Õ-¹.™ÕdžÆ5ÇQýyP㪦v&Ïr-4ûб+A’ì:й˜­5Ül<Bl‘ÖÂC]Ñq§ϼ Į¤Ð¡naÉmk,RëšlOªÇÄ1'-K‘½‰aßJt,øú6†òu¶Ð¿¡“5ƒ|v—ô 죎TmQðô—Ç•.D‚l=ç³´ZËÊ„]î£H…]Ê‹tRdò.ŽÇלúáš ‰Ò Â9*Ò½ÿºm= oä»B7i€çIÎ97;WéH¿×ÁÀùHŸýìê'À‡kºÊçÄ1ˆè#¾OÓfV·µÍò˜[˜ûh.(GŒßDº–SjfWW5e>f…¿3 7‹=3CYâ H‚shäIªÔãš_é…Ùï¾£KKœ,æ™.{§©¯¼Lû=»Zr $é'×0ù­¬Ù«(y·{ÎuJ¡z"«Dc(‹À½a:i³Å[ª€:—IQË´ŒÃf9—æàlʰlŠôÑ4wÁÙŒ?’Wâ2û¨}ZDÊ"ÁñÌÞŽÆýN³é O¸pSzƒ„i¾À¹LÎ3¸†Ì4LAé¬)Èèj¦y7og²E˜ÜÝÄH\ôU«Ônr]Ò…MH×G?Å6Îù¥•ÁO+NÙ°3xVšIùAŽYo †¨\%ä]Eþhv–xòàs×ÄGè-‡"îIŒì?™’;Æy^Êøµï¬U\®ž´ÐÇ •:Bº@Y™Y7ó0`qÓ™3„U–TÁ²¤ŽÚ°ÒÏ~;tjI)ÊLåàLz*Xž´9à{ Ó®Y3º&,Æå3?h´ßÑÜðc6êâ zµó}ƒíÎ od£uÄf©ówµ¦Š¯*Ž^<WÁ,êš©©/·ä²u³á®\j°ÈŠ\À¿ ¾}8Gžíõ:Æ ?\=sÔÜr`\´êZbT{)a2Ü)·h×;ˆdg¸åâàqˆú{Ψïâ2íMÓJ£™7ƒdp¸æoµjn±7ÃÃö#ÂK"NÃirÚà¼-J‰¿™ƒ—¿<)Ÿd΋ªqcÎzðéi=Lî`µ½ïn>²ò‰¹m±'xm¹®JF›¯®XÂV³~>€å¤F{äþç0FûØ#7¾ŸQÜú÷û ž¢Zƒoáíþ¡üØÆ"þáÞ±e&jF(yDþBpíïÚ4îw ÁÏ|râ-ôD­´€ ]ûñâlþªn€$ãBD ?Nê6NG¬Üá\Ãì\‹l÷s J®®©$Ó;WLL££>*·4×´'ë•,Ëløã†›±xmÔ¬àš!-!áˆêY뼆]ZËËÞm£t|ZLïó)i–Æ­c®C›³¼Ý(X?;’^cb샼ï\Ë:#Tª"äò†¥±*!WyËYºÄ€˜‰²p×1¼]6Ö¬}›S}–9ߟ%.œ^ïÕ?+ÂNª¸ñ­úD/%Þ]¤ñŠ‚¦yn.¢J¾ãUwbHzº6t_Tý/%¨4aL„¶k½1™ÁÝHƒeŠr Ù]ÌëQJé‚‘VWÈm¹•¢:ÐÝçÂʨZådUÅÝ¶Š‹æ8•ªø>[«x·üP_Á6vÑ«³D[@ÿöÂ[ªò‘¨Ê‡,G”ä8%*ƒ<¬äˆæ»;£³-æ$‰ÏÓ0˜?´ÙcP›“°¥ä» ;¤cÐ’$ûõ$ãÁ™'Jôs‹Ä¥Ù4vù¾t¤÷+!Á¥½¯²v‡*á„$AD—+Ù;*N¡££~Cxªâá¹?Š`gâÙQÛ…Ñ/¶ªWuêò-®fƒ‹¬Â¦‹ï{rò¾w1ØRK©nm¦¹QÍNl&ÀH[ŸT®2-J Ã!úÈÄ$p£ÁläÜ­T-?PhKÐ"¬µ5w…ë̤?Šè긘„W=ohR8¯‰:‡Æ0ÃN“4¶¬®!Ô9åB´vqëBs±<êzá{ëºË»…€é7^Îô½k+På\çA®”/¤#1Ùƒ)êY Ò8!ð.êg~ú(>Œú› ’ä zÝ…q s'¶ú3â–Ãïo|Ó¾f÷øµ–MZJº^¦ 8ÕŸ³„Ó]i6’õjAþx¤õËNQ\ýÏEÎÍ;õ’þÇ#¬Ë½šNÇÕ¬M,$x9]•f¹dz¦ÄµOÞl>ù/Cˆ endstream endobj 84 0 obj << /Type /Page /Contents 85 0 R /Resources 83 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 83 0 obj << /Font << /F60 16 0 R /F8 9 0 R /F14 54 0 R /F11 15 0 R /F69 22 0 R /F47 7 0 R /F61 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 88 0 obj << /Length 3776 /Filter /FlateDecode >> stream xÚ…ZÝsãÆ Ï_qíK噳#~Jj2×$—&M“Λ™Nr<‰’S¤BJñy&|üìR¢§¶– vÀâ{ù÷ûÏ>»|µº[•iùê~û*IÒ»E’¾*‹Å]š¯î7¯~žÝ?Ü$³úæ6ͳÙogôïD{úëZ~“Ï©€Õ|‡‡ô³ ù4¯dÃãŒÏ‹Ô@y1C`ì#¿|°5Òåìã'F£@È«xѤ`"-¯Ë°4aÝÝÜæI6ûV_TÀ_wÞ¯jG,e9Ò œÕý͇ûï^Ý&ùÝ*Ïé7¹[*Þ«»¹ÍòB¶ÜƒŠŒV=ö"6¦µeòOưmÁH_#¹Æã¯H–i1‘c…øÖUf¢® ÌÌì”-“Õ*}ÍÃÅì_¼Üšñ*>ÕÝà Ó)"îgxLG Îh[]³Èñ+\nù_#&Š„k‘Ù|ZZYŽ™Ý,ø(ÌtæÄD=± jV~™¬Škȉm0&Æ>^¢õ=ò»ÌašD>4oc,¬™Ýaðs wGÙ0–:Α¶ÖôïÌ£¿ [d&ɲ ÞÈtRåMv_­ Ö–É&…Í’9ŽíQÈa@2:ì$Õ¿ñ n¿ºiþ›~dØO7 !¿ KTlBMX‡­ñEñ˜“CÅ£ç›e>ã#O ÞÁÀr6¡Ò–º-`mlìl§üòЉ¬Äˆ«¦•ÃL1Rð©òZ¤éX¥òñFZÐ> ݆´ŒL=X 2NÚØ/óbNOß¿»Yf³{zJ”ËœODWß²†`ð¦i007`â„!]›E’&åì—$Ë•T*”ÿ2O2Q‘ˆ˜edܬkÄOmN¬]‡7´Il4%¥J Û¿W­G˜óÏ8…³¦t¼Ã¹¹ò7óB-i?2÷‘· œÎéí Î©j£%™¿p×Öl:ü¶Ý N®ªÅ`X¹êÎ3€#;,|±OPÅZl~¹lE4âŸx±®íb‡û6%ŒŒ6U£ýÑÔUäË‚Hɽô5›Øçâ賈T¢Œ=ÍFd§¹ÐT­OãwUü¬[°×øé?_c–³_ùeS? ¼EPêŽgµ²œpšy"Ľ¬Âs­!MÚØÿ Âp–p ±ÐJ¼ñ'×âæXOñâò•W—ê`ì‚ÄÞ¤£DÆ>‹M#I‹I%PËT È·æót¶å%Ÿ0n‚;z„éËç#'ÎàuçŽ[bΠhëΩ$C(×{µaû€ù¾ÜAï ðay²"šˆUÍŠxí SÓ!XP¶d2Á‰kÑGS“Mn‡—µÀÔpEm±—çpôUtÈ£`*«»çCøJ^>Ý7×Óù%Y$к´ ˆs<ª´;R%çàcã0‹ÅòàqՈ岯óHzÄ_¸ËÙ â²ÙÙ<V™C[•&!¤Seê°m…ž,žd°Bœ±ÌŒè•äÙp•Dãâ=¿x81Å£ÅF-+„cy^w皥ªŸªHEh1=–”ßùb|,R0¯·yVZ´Ë³Ål'=Ëq JÕïΪztoЬAƒàí^! s”<þ/÷J­\ rÞÓ4yRÀ[Y–(+MS]Õär–êLfewVl¶¥‰ðìÄQJ§¦MÀØ5TÆ7éØb¹˜}#"° 9㟥©CxFw¨Ÿ$³@òìÕŸÔ¤$qZ“Â<¥<<(hd ð<•Uâ\i§ïM@ºÜCXÝ‘»=˜‰€Ì9SžÁ7;î£pOï%ÇwW_cŠ(¬š©,hã#¬Ì 9ÏFI?ƒ_ãm[hó킲¬o!³êdzEÀô\µøíŽî¾Û( óæG‚3Uжæ‘x¾ *Y-ËII1{¡bùT¢Ðç=b)œ)ÁmV ÍÈaüC°{Ö’Y µv2È?UøÙšÁsÑêi|žZþ¿¿È"ïÔá×Rcy¹îûwœÞ,æ ›«ééXÕT—•ôÝR,©™‡ÉÊö-Wß_b<*>rMîãí.æ^†Î ³ŸÊ±³2“´¶ò6Â`c)Ï)ê’b0€¶àÊÔƒ #ýW1‰%7E9L¡;O€OVÔV:ÿºDó"¼ÌÆÙX‰jH>›.lžü?9½2&Œ~‹9~Ù×Nků)ky6¹T=]#ª¦eTðü ]yQéŽ<\–wE¤n4WD+žæta£ÓÕž½éɃ¶tõ†–¾®º‘õUs£ÝùcêëÉGmõ. ±=`R壘ræÍ!ÕÔی겚û?Ö%<fíŸd9Û ÙDœX“š¼wIãuhüUV)s¶’gK2Ó˺!,ß[×™¼®yÑXÐÝ)o´×Õ=F(•'í¹“eC4^sÄ’ÑE‡*jë†KÍ‘Òl…ò3%.*<›éÖBà!ÖŒu)Hžb×ôúd¶Ò¯Öˆ8Åo¤SGØ^X ÐN÷•[„(:*|7MÜãæÿ<>s\r$I:ûJ ¿låù o¸Û["#¼„š”žŽ®¬ÏÑÀûÀâ Šl¢a'i’¶:­kÄÍ…"Á [„R‡ü€ WìG¶Ã^žœVHÅ{ŹˆöVIÅ<“žt‘éÕм[S„[!×?XïãÙÒÈt&ùCH|ÝÈ1 “Q@R ï 6_(m%û>ŒL»²vo:þ#¯ *òcE!77åL:òëç Ë–®¥&ÖS/ë$Ifà*¹ÙÕ"¬×X™‚û`söfŒ$J7õZÚ»µ°ÃÝé ép°z­w=óÚÍXHsK.‰´­^«î·¸»bã=¢;A*zÞbòIÙ]‘» ¾O‹ÍiïqQ±z½Ûð‹²Ê»VˆñK´y©»¥ñUV¨ ^ç)ã9fK¨ Ľä¹>)—H6C»køˆßËÑëÙ³ “äËõW™ËâºÈaàˆ«|™£Fß Ò™‹òæ!›¡ìC$ì½A$ËIÛX:‚ç+ä—* ;ǸWBT*5Ñ>»¿åjéb Î?[¤Ñõq–­î²4#!‹¾ø8£æâ®ÌW†a©l|çÆ{vgxGJphjR^:.QWAî›s½^©z+ ó&ÞjA^×ÖãH½]0œÂ«Œx3¿øÉã‹6ÌhÝ3¢ÄF[')­âMÓ9}óozL’òZª‘aODÉ/ªæYïû'ÖÊ’Å칳ﭽפq³IŸBš…JÀ’úÒ•CÖÂ¥}éúİk5—É.R1Ôd1ª÷’UŠVà ë™eQö?à|Ähy£s+E®¢õÙOV¨‡Ú¯Tìª?Or=øÂ38ÄVVhxSô}‹ßCÕº‚WßL0Ìo‹v¸%ä<‹s£ëH`ÕŒuÙÀúéÉe[Ž!Ø5³èá!_h%¸š“¿n ¤åè¶‚muiÿÈ;l=^[hIËè.¸ Sªß_l6®üž´Q³·=í*Éc˜}ÃcN)"Ëg'Œ¯´ñ òVsÁØ^r=uÕЕuÿ†Û«k¾”ûˆG»áy†Û­n™éí×`&ü‰T*åFl„²·¼FXÔf{ ÓzáS6~çßJÒxì¨"{É-·Ä½-ÜHºXp( ÄåViÔ­&uìÚz›ÏKcqvûÖ2•y9•v´Úü®%g1.× Þz¶ÏóỢ{vAZL Ÿ'Ä,ã…o2qÏÆewœ^#H‹™Í³ìd¾ðøZçôóBá3Ï]=A‰_ -ËëOjâ¯"×\¸ý+(;&K«ã5¢>lYnï)ç¡LZ#¤Â#ÒÁGÝZÔ~Üi•pÿ•Îӻ墠òlq·,–úý£$ÚŸ}}ÿÙÿ‡¦L endstream endobj 87 0 obj << /Type /Page /Contents 88 0 R /Resources 86 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 86 0 obj << /Font << /F8 9 0 R /F52 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 91 0 obj << /Length 3117 /Filter /FlateDecode >> stream xÚ•ÛŽÛ¸õ=_1H_d4ãHÔÕ@»‹4m€-º/E‘ä–hYr%;™ÙûíËs#)[ž´3¢ÈÃsãá¹ÉY¿xý®º[-W…*îÖÛ»$QË2QwE^.Ušß­›»ÑO‹{¥²èA/T}]¨*24s‚ñ7˜íiâ¸H¢aoÿô`ÿ?ÁlaÜoZûßÀÚHÀßðdÿ AþþÍ{ÚŽTžŽö¥µ”úˆ»® ؾ\Ügq­öcœd 5ž£îì`He#‹ÀÕp¤O†}ð=bû´þûÝ}’-WYfŸÉr•³¶`Çn)£ú Òšø? ØÝw‹û4#žLÝ,Á½#­19€õî†VBÂ{GˆxÝÑׯãî2{Ov´Ãó†te9v¼ˆn F«¾?>0Ä( 5é´A5ÅSµ sÞö<ÐѧE‰jØfÔà£=ÐÜÑ2RqBíP„‘–kvŒ†‰çûÅþë ¢2,ª]ï·ô ì^ÿµ¨R1 š©éÈ"ÇÈ‚µ´<+¬¥‰ôŽœ¡a"1øç6 ›ôÀA T[ÀÓŒÆ@(=Ô¢•‘ˆ]ܧñ7a€ï€,C`‡haG¹(ZK× Ï€žîè–¤f}§[¾Pö¤×¢#‹qDÃE抜è.,]\„ . œýa¢¤é=[ÿÚ¿Vïdä㜞ÜIÓQž¦ÆNïGz ¤Ùúµ'šwWY( ˆš8ô#êÏ-ŽˆÛˆ}9û؈~-`Ôÿ!Ag¨Ë"úiËÔBÓ±T‚ah•öµ¡ÒìÚËÜV °`¹ õU¾ð”ÆW–§<ñÇÄ1ß¾gª-½£{9Á¸Œ6,:¾$´>N¨{o°'_‚0t‚¸KÅLæ¡çÜ…þÜÒ%†™¸}xÑf‡ßÙÁЛÚ31ëúħ [ÁÍZ‚=ýn_ÓÚŒa=!¨wŠÝq)+_À /`h<©!¶»clÑ6^9Óª  ÌTB JÈê€ÙÔ›¬Ùú–ÀþëwYygcSš ¢º•&UË•âc/V8•GoQ}i=ZDe†a‚U>ˆ×ŽêBp©tYÅ ãZZÈfÑz±J™•‘Pjä*së†P~‹þD>”Yf2aꯖiÅDÞ€¢ôQÌ©Ct{P¯-z§ã7÷q ˜’¶·=Ü™GZ>ˆ§HȧޭYP—jP"sp–^èžrÀˆªŽYžGïœKä«òŠ€¿9ç™:G‰óψUc¯ÓŒ¡Ší¨r"&–ÌÇ$Í6dÙp kŠËÁ ’á üÆËXÈpQ¬»êÞ^o¸=ÃWúºÈ ¼K€ª‘íÚJ¡Vñ$›³ë80)d_@Ù,A†g¸£ŠÉ9ýŠ[˜ÝaÂ"AAÄêZÇX/1ä•ï%™@1‡9Õa /òZï…‚ÿnäîN\+ÀHºXåCY©CH.@N¹ ‹8žýø‚=«½ä­¡zìÍRK²3ìI/eªö¥–ÏDIo¨Ú!gŸìÞŸçeFcÀYV°Yl Š>ƒÑÑŽÖÊ«<ŽÞâ R5àÓ¸¬ðçÍ„)>Í„ã2Ÿ=ÐW/û M£ËîPÅ”v¶4uYqVÖ ËÏç0D|iOfp:?Pº‰Aq[W¤,íŒnÄ.3I 4mHâ÷$t›AfUçcp8ˆ‰D…õs×E`T8ž5ŸVÏöõˆG¬®ý1p†’Çeô`|N´C²ûœ»ÏS[ áA4O7„+wHÿ0‘YÓÎ71ȉŸ&lµßõòÆÎe06ŒÉìK:nÀúª#Ï·\@½ôI>Áèñ =¦´eê’\ÌF ¼`[éËÛ¿˜¦ÿd‡÷¢ç ùÈFÁªxÏDLïÝôÖåJ>î/us¡ ‡‰H¿Ñc¨ÄgSÏ)“ôÑ“Ò"iZM²LXݲ~jÉ?‚ÀŠ3$PÛ!£»RÛ!PºSÛãaì9]Ë)ÀÓCšÚ‚ÛßIõ‹©#Xo ÷óFð¼4ú^=Í”¹éæ‰8—K˜¥WöïÝ?8èï Ò×1šƒóEϲì-0»u] RüõáÁµ|w}«žŸzÐN£ú˜_^èÿ!PÏ·® ¼#ûèýÓÿ¦É‘ ïã~j‘&pßS*^ÒÌÕRÅàéSëñ+"œ³aCšã1RÓ…Ÿ.ƒéÒOWÁtEÓI^“æÿH‘ÆÑ§sY¡¹-ÜzÃ$ÎJTÌeÈÍ•<§RWl­Ë…Pt"¬x:Åé½ÿ|A>Cy‰TÊs•'•¤OJ±:D-]Üæà;JÌÌ1(s91x5—^Ì©™¹÷^}¸—¯,« ÛÊ{)İ^‰ƒ’ÑŽ}k?N®ÒÉØv‹©÷«[É @ rý¿Üħ¥$*Câi'¾NÛqðõÛ'ƒ¼Í‹|ňWýŽàTR$ã¯æÎZ’ñ2õå7Ã0g§ÎS:‘þæñò ŸÜGö›hÒÇ()²Ò‡’Öø6}u¤±ouæC»ãr1‰›X@–a®Í×iã8Ëkð3®ý’O*CçYûùï€Äºª*L¡+N{ù7&|[±ËCP×Â;µñ«(Á;/éä_'IsU^$Í€gïr|o¨’Õ§ åíú€aò³àsÇ;ݧ’J>«öõ1ÌHHè2¼‰Ð-Wâë©Y2“’åì𤂖S–]ývÅs€½–Ö} Ïóþ'*Ž¥ù´_,ŸØžÒo/u×í¤A×¾åsò-nð`…þMÇdë¤Ë¦ ý•ßñêeÉÄî2åâúPæ´3Ï}1À/LîLÊÉr-F¬Ç`IDL§¦/½Üà ê—ÆfÒÈÅïô‰é«ÇåíB6 ÆŸ½èÇ:Áwzª§/1õjÇ/áŸ!k}›—jzhOïÌ…˜J~ÏâüB·›òë4˜à"“m tIU.Wirg¯ä2Ëù£s²‚µ[¿øu¦)Ê endstream endobj 90 0 obj << /Type /Page /Contents 91 0 R /Resources 89 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 89 0 obj << /Font << /F8 9 0 R /F47 7 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 95 0 obj << /Length 2285 /Filter /FlateDecode >> stream xÚÍY[ã¶~ß_a$/2Z+"u/’“Em±H‹]'A̓Ʀ=ÞØ–#Ù™™ Èo/Ϥ4cƒ´E4¢ÈÃÃsýΡç‹å‹O^U³:® ]Ì–›™Jì8OgE^Æ:ÍgËõìûHÅóEV×Ñòn®"cǺˆöv¸³Ïs]EF¾îÚùB—ý]euv¾93AK»áÓôö ,÷R+ËzÚ ›cìÛm{‘IbÐÒúÙI•G½±ãŸ€ÎA09ÖrÀõ›¯þÓßÌ•RÑgÕÜÚ?{ãÏ>òÙCeÀ¾éæ?,ÿ9S:.µž-”²cCDqE&ùë|‘¦Új`Çk§Ûô4†“Yxøìá¬Õý–í3¤la݈Z+V HÞÛ1Hß%*[9Û#Ï%_8Ñë¸T,ºfƒQråeiBv†ô‚—è«ì}È! =P±öY•D÷;öóP&Q³5´e‹º^N<ÝÓl`„7wA@wSGâ°i@Ý3šbª.)b%–àHÿŒ!Ûó±,>ö©g 2ÏÖ€ýLÇ×im½º¦AL·-iCgéÌr^\r0VÖ@gÚ¶±;ˆÑÖ€ñ•ŽÞ~˜²ÀßvóîúüÌ'ž)ô|줠iYmN' ;k?+Þ!F6çÚŠÝ‹'cRêHbf‡ Û_¸tT¾Wi„›×4&‚7ðZ±‹9S–nˆ,§Øa „‚¯\pkÐ>w;(T³Úg~!™?ÌhwôeQÂâ{çg¨ ÉÒ“ƒKEKØ»óT4|N7š½G>]'‚|0´æÊœ¹tU;«qå’}ljøtSMGäø¥ëÊáìÛyæMÿq}â¸kÏA€ãªÆEÌ Ù•èèDÅ“ÈvÀû²—)Ž`Fù«èÀC) f;Êß’ÎO;gñ@í°å ãweú+À–)Ûµ@УüÄÔc.ö–DªÑeÍK†PnQØbó.Éûuƒà͉ªl{D)†¢ÞóA JòÉ«B…WUζâQ'A†Éu´E¨mÈØYš“ =V¼ º/ç6'ÍØKvÂÕb>qÐê©2.ªBN´Ò+¯ËÍÔN¤ÿ²*6”ši’ÆUY ­ Ž`'‚ÃÕ‰5‰'aq xhœÄ”PwN½mßå-ëe®Cî[Øÿ‡ Zj_é´UëzÆ®qÍ—†BçÊ[Ю½”öÖFu‚ØQºMWWm$Wte5TäVÉSÿáH¼Ç:v‚ÜØôÞîEZ0%u(®c_ôE &@Øû2oöœã ¡Á4 VYe©"Û)u$åBåq™~°6_¿Nbw×8oÓá DÎi„ƒÏë¢Ý6MÖrIa“q¥ó¡µÿA0‚j)¨²'4f¤[3ZR¹ì ­Ò,Ùµ¶[&tZTƒÜ6Úþìg}Ãxc5Ð5ûvlïo»þþÊ]3 AYqMh¹?IïðüÐÓZ»™ÂP¬JÔ| ²c½(3„ÿÊ\#ágމÀÝ`<Òiàw»î ÝvÐú@A-J_øKèmßß–&ÐY|@*•º|õøHt|”´o0ƺþÀ{¹Ÿhª¡ÜsN˜¨†ØVÓ,“¦¿¬¢×|¦†#ìÎe—»IÃkiJ¿T³ÌÇŽ4%Y>uEëi¥é„*ó¹bÇÇF.®;ÑåBN›%b+uäEO lw¤3p(ì”åÆwoì¾&Õ,)½/,‰Ó©N)@ý–ËJ²·Ö6ìì#ÃŒ³Ûúæà?ÞÐkÕÂÜ.T,mWÿwÃ2r´!~aOZ=KA~zz"äi7—[/N ò·A7\”[¸{0˜ä³”-®¼Q k»ñ?NLT±ð¶3*’Í.(.I, æut¶äJY¨ÃÞÃu¸›'Œ»ŸÛlÖ© É•‚~1¢òÒ#Œa‡µ ¦]ë:÷¿Àû3¦ã-=³-¶„~DÁ‚\ÛrâyR’¶q…¯ '±z+n†¤+šþ:OÑ"ŽÐ9w^‹·A6þì§7rëg¾ßwÃ56sÿë J‰«¿Ñ«g¢¿oà ®$‚+ãïÔRÿI´„Å-²Lùïè_\Õ?ýÓyy¬{˜Gg~óÁC£¹@ÿãQò1½¾fꆋ´ûÙLú,œžMÏô~Våqˆì gï.÷§/™…üˆñ·©£¯¸oÃK‹#ðß´ŸÑëÿ›ÊëgC9>È}*‰³l¶Ði\+Òÿˆ‹J;¬ìïÌy ¼rËuâ744¬¾ @¶÷¼5ûª°O¢¤Eœ[O‘,jè 3ä+¾U Ãfövƒ]%LU–F¿Ð†³ãxÃçöùÕ>¿ðÛÙ îu•C'tÏ+ⷰÄKKÂæÌXu */ýzÍó5®!Œ©$+û3^KÐ-î°È'X'RЕˇtA=?ƒfžêQ󣯋õÚýgQù£SËÏšX„‹ÿ»ÓR4¤€°ià3Ŷ$—à²òP¹!CÕOÜLŽ“D¹ñ„ß5d=%”½7Øûõ„`ú –0—S‚½…Ÿq¾HXÜ–«}æƒiàÿF5ñÛe¥ã:«mŠÕqUUü/LüYéÅ—Ëÿ÷ùé endstream endobj 94 0 obj << /Type /Page /Contents 95 0 R /Resources 93 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 93 0 obj << /Font << /F8 9 0 R /F61 14 0 R /F60 16 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 98 0 obj << /Length 1105 /Filter /FlateDecode >> stream xÚíXKo1¾÷WD‚C*ˆYÛ»ÞÝ  Ðªjƒ8‡h“’my ñÛñÚŸíñf7!pÃ(ëÇŒ¿ÏËy<Ü»w¤Ê^ÉJ%To8ëqž3U¦=•åLȬ7œô^÷î2!û·öߟõ’Þ€sVfÍ5ós¼Ïû'š†¶ŽôúBÓRÓ'MõøM’%úg¬i¦i®éJ“°KÜK“L)i¥=‚˜šý\ÓWMx+{ÒÒˆ3ŸOôçÎÁµÀ‰S»eˆéw85§fLˆXC`ÿbÙ.0œà„šõÀ.­0UúĶÆ8” =՟ךV€ ›Ì0}äz×r¹cGØrFM6XG?ÄþDUáR&D‹‰®0üShpEÌxî/À°ØsœS³zeÌ`>ÏÁ_AžG-㉆-$+9·°Ïa¾÷8¬ýa ’23ræo¾ë¯·ÿ@€í¡¦ïš(ã÷ÈnµÜs½ã¥¦cCFî@rMH™ß×úÔ8uÄ›ƒj_Oí’ÀЉefÉŠ×[x’æþ÷~`a`«qÒUw«_ ƒB©Ï(5%z-ËØ™9ÞÏÌØòñ]1ðýõœr ÿX/Ÿ³îJŨ›æù?ßüëùfcº1Þþ§ÓÚšn\.èL7]î¥!#“Ʒ–â÷Åw…‹¼`š¯7 ÂÆ´i?öcÃà,vî%ñÆkÈY´þHd˜Õ±ƒ#˜ï„ÙfŠ#i*®˜¨3h7š€/¨…yÂÒ”GÞ_¯CDŽ”Ã5pæ‚\ë;áÍ$è â*ÈÞj¬±HΤhT~w›ÓX®»\âDa5¼³øÆ¦ø¶1z¿­EmÊY®³#òxÿ(Dm‚БP7·UÔ›è+qÃÕ7!•7ĆæÀn¿ß-9n'+¶lldpôsì Jfvx ‘—]»:»±†¡7YRòTS¸>*µÖx$úÑXË1Ÿš5d꣇¤ØŽIŠ™¬U~ïüNvf±˜< Ã’rÝØœàN…/}¼-žøÊd£Ô%Ý–ãP —ª‘¦\3Q‘ž'´-kBKßKn„yì kèdæHUKßåøÎ†ö~ËPݧ¤¤?h¯iξöŠÌrç̦5£þ¥½·pÐÖ–ú)L“6p%ŠœfUêjg€IˆÓ†ûM£‰ö4+☢¸m &N˜»ò ¸ èmÕöØjãî§„sÚ÷Ðyò•‹¸»R‡o·$B·uL)´¢ŽF½ŽL¥Yl¿RýL‚ܵúy´Ï\ëÛ®TàáèJWJ"¬°ïÍ(çºT—ÙV5ª†ÜçvÃçDJ"V†¥fª«ÁÝñã­u›Bœ*ÖB’xH;´¾­kÁŸ)þߥEN´Èc-Ä–»h mÕ"‰µà$;hq蘒ÿF‚•i©‹mÉŠ¢°Š ã€{O‡{?§'^« endstream endobj 97 0 obj << /Type /Page /Contents 98 0 R /Resources 96 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 96 0 obj << /Font << /F69 22 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 101 0 obj << /Length 2280 /Filter /FlateDecode >> stream xÚíYIǾëW ä8ð°ÝÕ{‘Å‘ ‰(cø ëÐCöp:æ&6©!¶»ßòÕÒds<6”[«^½}«ê¿^?ûêMQ_ÔQ]$ÅÅõí…1eTÔÙE‘—Q’æ×ó‹w“æÒL4Z ŠFLcŠy}9Í“Tþ~s›&q 8¯§Ø+tÏä¥q{&ñ¥Éü–]Š1˜¹/Ýÿ÷׿ˆ/¦ÆDu)z=Ðxs9MM¦øCÖ C– ‘P÷YŽÁS*š˜š8+VæóýMµÜ|8¸ÉÁz?—††~ý‘,sƒø,¨íRØÑ”i똕î⺞y„®§a¤цU¿üëul‚d5 Lýu¼½>ø;§”ÎÒlÒ7båeˆ>Ím{j¯¤·k’lt-ý!ííÖ·¯|Ü^gETk×þ½Ù»[³À7ú£}W V§Â4¹&õœœÛd&4ÝÚÂüMGÛ@B¸•ȤE}L=÷V+ÂÈg ôgºr?×¥VÌfõ'}{¤%oø²…kœüŸQîÃþ b=Øû"_T7¼þÛ‡ƒ%Þ8îðB`áñ`å-†}H†±·œ\7ð½?>e]½àê=ü¹œ¼²Ï(Û%ÌܳŒBŠøÌ+ã%îÜãjgQë”l_Ýlöm~²z}q©Mæ[ßÑxí ݃vßi ðÖ ý ^ÜšÀ8Ö«Þ .ü*ñÞ›å]6t¾Lòï˜0gŒ÷)ÐfëŸéþoÄÏeÄì`ÄÇ^y¿ÐŸo ¬Ãð]sï§ëKw ×”0È>AÛfu¿Ñ§ˆ\ê&ò'¤¿”OfzxÍúÄýŽªÜU7¬ngÝúj(§Õð ¼Üywû¯p•š(=~ïœcþ˜Ø@ü[u̘WÞ—? Löç±Ú7ùÝŒW¹ÌÉó©Jó³nÑݲªÌÙÌàãüØ;øDÈäïÑo `õjÌêg&=ÿ£|úƒN“äSkåKÌè7ì§ósê²ÛÌé[¢©r>ò‘¤¦B'LäÑñÙëëg¿ÂôL endstream endobj 100 0 obj << /Type /Page /Contents 101 0 R /Resources 99 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 99 0 obj << /Font << /F69 22 0 R /F8 9 0 R /F11 15 0 R /F7 21 0 R >> /ProcSet [ /PDF /Text ] >> endobj 104 0 obj << /Length 1889 /Filter /FlateDecode >> stream xÚÍYKoÛF¾çWMJÅp¹|E€¦HEÓK…%QV"UÒò£‡þöîӶ湸`®4Y‘F Ow†ÆÖJ´sÿ…ç-+ª5½ &\k“ÂÎÊëil¤<† «‡›ØµlCë9]#`z–¶‰_¿3q=UæÕ¼MñÉ*ßL¦Q ± –#xù;ó:m%*,(8£k?ò'”ð‡wÛ’—Lv‡/çõ‚Úƒ>·MI;qÿ¦GÃB÷´ü–96#q‚Vã’Û{ŸMšÕ<0Í9þ—ôøÈÒv¿â¨vÄÞ3É]à…ûæ•]¸Jf×,JoàDfGXFÝK£ëcÜÀQcÕb`ÉÃO™byXäÔ­Û½4«ßDü•¢Í!×Ë¡s*æ{¬œX ‘°Õ# =®ƒ¨‹¥_ ¸>·Î×~¯\´™÷)9𵋴[×- ^_9›.E€§fLcNüRü±ŽÀV­]jö§õ¬Çc™ùEüK:º…ò[¢þ‚¬Ó“QkzÉiï%ÎÑ£¡ÖÁ¹`…ðq’\(."9®”&¶àÄñnc.Do[VÄþL¹u²6\`µp¯ýWl~j¼i³n ^„WsÔð­÷ᨬXïnôn:[Ö-ªd?>ÂP¤Bdž0;žúmÞSó¬¹78.Î->ÐÝ/Àš‹dãn°'‘¼Öy%ÎCÒIIå$Ÿìœ…È@ÉÀîÄ!sÂÍîY,BµÕÞQèz o…I(o…aì©DÿЯo.Ž‘™n™c½Zy:Ðg«½rÍ÷ÌÅÙ*žò(ð½4 þ½GOhp1÷{üäŒè‰/ðóýìNñ-¸ç¤(¤G€«Ò©ö/Õñ`±ÚétxºÄµ\KýqàÅøY#z|`ÒâÈŽ=ôDÐAO=%zàg^€€þŠ"t.ˆOœZ…Ç׿i¡`ÑHs,-X!:êXŠYÿÂc–âÑÐèè;ÇIÇ›ÐytäMäô:V v©n JLlÕuÔŽÆÈpô¼8BSøyõ…ó% CÙÅ!z:@Ï\´Ï;pçáãg¾^âï ¿ãç|ìˆ[”zôð¼gw¨œõ‰”üwŸHC¦9s ?ãSŸ2ç~'ù¢Ü~Æ\fd¯ü>•&ÐÌLIeŽÒ Ü‹w—/þAn endstream endobj 103 0 obj << /Type /Page /Contents 104 0 R /Resources 102 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 102 0 obj << /Font << /F8 9 0 R /F69 22 0 R /F74 36 0 R >> /ProcSet [ /PDF /Text ] >> endobj 107 0 obj << /Length 3579 /Filter /FlateDecode >> stream xÚ­ÉŽëÆñî¯L!#™lîNcâØ€Û?N`ûÀ¡¨Å–D™¤Þ{ãC¾=µö"qž ‡õR]]]]{óïë>ýª¨êU]˜âa½}H’rUÔÙC‘å+S=¬7?FZ$Ñ[økàoXü¼þçCü°L’Uç ð#L$ð÷ób™›4Š¡¹‚?üÍ᯿\ÆÄñéW•·í264—IµªË„Ñ~}^,SG¬ÙwÜn›zÔI¢~˃™2z·0UÔóüAßâ@w”å0Ò#ØÛEžGÍÝžlR¤q4^¡Ý"Ì^°Œ6~ÿ±o¢iï}÷f 56¸ï®3Ñ'žº,˜ªëx´7Ð~z#t €‚ ÒwFl¼üEèíNŽs×Ñß2@Ò!ÅÓvÇ3v<Üœi1´™¼ñL<·Eþ Ç DðSœ¤]?Àãú¶ãYœ”íFx/x‰ž¿ã;/ïÒMš¬bhÆL6ì‘ æót/"ƬL+p³“Óì†/ìòè™·Í^a!f2ý4Üög” ÞÆ´= ¸„¬]š<]ëo8Ì"„‚…1«Å²2yô†ù!ü)ãÄÍÝôŠÙ)…í«ûË4Wu­Z!ž€+_³¸rºתÀÔØý†,c—µà;Mvñ n¡“b@:ª¹9ˆ„ˆ5+ˆÊß·ÞZÞçE¤;ö׉ì0^Zç­$ã†(->ÿu¤‚†íšYwurÆ•8<’¿Ze­÷²Å?ñš1_;n«R&žR&l=›vòØÛøâc’ëiÄTØMУÝbÕÓlâ‡ñws%½dâËï„rãp1‡VñF=XOö)µ ( Ϻ5)ò¡'Ö—@»\z³¬k‘(Xù‚Ó{%+I®¼I`œñæ·üK÷ÓèõŽÌHõ7ÞÅ7"“­Þ§l=(á{¹o7þG¢føDò °yý‚Àí¤vËD̶WÆ'.a­7ÑOášæù(Á PXSø–ºd)¯¤^ÐxöÐ'v *>´qpá½ô·oZðIàÒœ‰ÖMwCË«4ÏpdV¾®thfÔ´%hpׄm>ÆpIl 3_¡5iØþö›/JÆÑOIš‰\^Ô4jãû3âÿ!kÉp¹e¢ô[´ai¾âÒ âN¼iKÞ C±÷p’âÌä©Žä ‹Å겑 Éɲ”Í”ÝD‰T[ĨDÐÈ«2š# Æg2 «Ü©æì®s QZ®·‹_¸O.Uȱ¾m5€üMiYç yYT¨ò"ÉVu–….íÚàq›Ê)Ø´ÎÚ¾€,³DU$êÙ/ÊÂøˆW±µ!ïà¹C·…U°­Jf⎊-:m;Xs¦VuÇËþ¹žejK<;ñ2Šs}²Fõé‡%ûÿ€W"©¬Óèi×Mhê<äjê!1ÀÝ <¼'m¬ó÷ÇfgijNîLø›b·ÃÛÇn놕ξʨ$‚4$ó[tÂæ¥ÕM]„tl¾ì*‚qx‰¤'8aS&7ÐLz·xŒ}#çÃP‡î ºSÄ«qYÈνPVM"ºëDiLY±xbƒ,U%BK6ä8ìûhl!GªØåŒÕ 9†+;žº: Eq/ÿÎiåÐh´ eªúN/Ew .6o6x’M‡ÜZ­Véâ0t†_6øKÓÜãMšÎ_q›~…¥@ØÉ¯F¤%ÉÔ~{?‹#‚qÝÜ µÐ…³‰Ð®ã?s#°mhGüã Y+ø…”"KÐÙä žÇ#ŸcêµìðÈÄ©øà å5BÕPÞgMñZôÍló ×E˜'guÅÅôâ†øgÆZÂè\>¼ ydT_|±[½¸m›üÞDaesYùÀWm0ŸƒÓ™r&Ÿ®XQÚ'6Yô†Ê9e]öìιçÉã¤y\˜$ÆT³ÏH¿“;^¦&wL7Aµ§N3Î)¥u´ŸÁ¦5—H>rcó\Š©p[ŠÊyösŒ;’è+áÕ ×D¥°"¤É”I6¨E/¡²%-.áIFnî2rµº8 ÎÍùgtМ ú¨¶ÓzµÌ… Ç&àŽ´˜Ê sQ!¬Ü¾ý(KǽÕDñ¶ èRl”kÀÀ<ÜÚsoRéC‡ƒ5Òö$à£øDÉœ¹¸œ‘Ž °‘ÛùՖе {{Þʱq¶zhï²€ kr¶–A™†e˜Ô‰6ŽÞƒ$ìÄ0WºJ½ê¶/jÒH`€üÁ¤ÎmâAŸ .×ÀïF6<Üz唪i¡aAÖ…d0˜“ EFå5uúõ k.ê—XgTßËÍÁu>ͲŒ¬o¯Ò…;˜Z4ìTœm;Ô­z(ÕȘêS¦õú¾¦A(çì( ·ÍØ©bººƒ‰Sʉ&ä׋G ­ùš:òÿc†ö4frØ¢Wž¢W¯«mu¯¶¶z4ãBƒá™á½$?Px¯l–IvŸb×·,u’·Üå6žD?òÈ wqÁI:ƒ/;½ãb‘¸}à9è6Þq›œÜøQ]ÅÃ\hnm¹«D)å-f¸ºD}_‡ ÊwZb «©ÊKP7‘IrýMlFª€ƒ3­¥õú>B¥ë³,_¤Ö‡KÞ—Üjw_ZTÚ!ñ¾*ùj ÓÒní¬'Ï@q=ÃMsW³¡i¿*ÝSO•˼Šn Ñî@г‘D´ÚDsRvµ…´Ü»%Ìxo8ÝÍÂ+Ùƒ%¾ âÆO´$ZLâéãì«ÞtV؇,·ÕË,w”B»£lä`ë¹2Ý»õöš¸¡)N¿s¯ZŒì½Ÿ˜j`3þùA¸Jp¡ßß´Åd@?²ŸGª‡¦-Cûhßm C6¬vSo¾hgÍ &*õÕ³¨±‡œã‹þDfa¸ÉÙKO¼4}.+ ëncŒ’ ši˜Ò`ׯ'¥TRÎ÷T^­iqâJ97&O˜+å†,ûÚ•R÷p–½Àÿ8ä[Pª‚ˆñ§þ–ÄÙôùN}m&)êê²S!çnB!I£׊ȄcxjÅÜG˜ƒÛ÷êW4Lð*EóöaçNÉïhb’Žr½÷z͆W>[ïF£x£ÞìèPŠ QpãÞž¤mÝ)ÌÛ8_nIl¾/e<ëkätipâ¹êøæ¹gí n†Æ mÛšAïëÚ#OO8ÀÁz·áž¿³õ¯‰z§Yf8¿þ(UjJ«ÊÄÿÆÚ^ªgôiWm{-±”ÉLmÑ¢œ'~îÞJZ8XQÜÊ«Hj‰<ˆYª(L—•´U«±|)ÁljZûÄL üïÈ0Ï,’´–ÓO'–÷Ÿdq¾2e¢¯ûŸà+‡¸:p²“϶âÞ4"ÿä¾"”Ö«Ø}g F0ÞÇL.‘ÉÁ&‘yG¦CÚº¨ 2qNlœ>~)¸óÌg29„u°e|~ØÙ0ü "ö¢O´0dš‰&°b[JU>ÏçÓæõ®KáJèc廑†Ó‰Ï\”˜±òaEz“ŒЉÝÇ%¹Ó9DþVë±ÃèÛ¾B ©Eh@¡ûÍ÷ȇõ#÷ÞéC›ýܬÈ?d3F‡³çmF­_4_.Ø ñwO’(ÆWBù$Û-Õy¯'­3§a}ú²’¤4ñ?éJâz•ç•Êé¿î91«4O€ ÂZîÕj™Ä@v•C£\å…|–ø7þÚð½~ß&Ê“ñð_ñ«#nrþ,¥*û®òI…¸ý˜'Sç;ÃJ¡enÉó=­ÐÃÂÍò!{¾Qk'…tÞÚÚ˜¼~¹D¨k¼Ygqö?¯ñ)¡7‰0ý€ÓâÜ £Zǽ$è¸Ð¨5ñ¾ýOŸÝ¦Èò„G®ÿI¿M±9¼ã)M[F¹èÞÖdì¦BmgåÿûòïU¯,W¥siêUU M†€}¹þ迺Évá endstream endobj 106 0 obj << /Type /Page /Contents 107 0 R /Resources 105 0 R /MediaBox [0 0 612 792] /Parent 92 0 R >> endobj 105 0 obj << /Font << /F69 22 0 R /F8 9 0 R /F61 14 0 R /F47 7 0 R /F11 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 110 0 obj << /Length 2843 /Filter /FlateDecode >> stream xÚ}kÛÆñ»…ÐOTa)äò%6@×mÄAœ4I=b‘”EòÎùíçîRÇ»—³»3³³óÖ?î^}ó]V¬Šm‘™lu·_EÆl‹(Yei¾5qºº«W¿?¬£ ‡§†§ç¼Þ¤&"þ‡òë+@ZxJxFxx~ç5Oo`ˆ›~]ÿz÷ïU¸ÚDѶH_&bždÁWþü+¿äAf áO·ÆÌIœ…ÄÐGÁ žÏëM&yðÆG!…/ŒöŸ0ÜóðG~ Ó^‡çxüMðtD´ç6IæÔ£¹rx2xv4^ˆyi=M!+[3N…Êó™õ=Ž¢à´Wù„Y@òûDëð§«ð»æ99>Ÿ¥-­Dx¶ìdYEJ Gša„×^ᆦCFoñE˜×U$ ã™àYÝX>:‘s˜‹ÍE|N¢9xüíû+R¾1AòÌ_gÚ±ØZ>Y»ˆ, ‹e3á)ŠX11#ˆ½H‚²BèÄ$ˆÞõIZPœ~OxŽ §:üv½É€ú»‘q¨ŽÃ°T(Ûjœ²ldÑã°-?Ó Wü:œûO–³—F¤±îRÕë8¾ÃÔz—¸à öH…UkuH€}׈æ¦Z^•.ªÃD ‡¤ÎH7ʃSÇIYZÞ¡Þ¯—9=qT¸{0¹à/ˆµñÕ礇fͲØÃ‘¹3r,qjRö)‡h€@p‹ªÅáÈÄéDà4$ç<½ì#ÀÑš‘*âÆ$¡ó7=Š $e⌇'K~ý6ù– Š‚¾µ*Ðw ]Id‚·%‡*ÚDfåXG¾EH5iÕŠ 4%tkᆺFD£¼!œ·#ÌŽ\väެ^¼gÔÛi­z²¡/ˆOTÀdà7êßÄøÐM›,µZ‰³}£§#®[žS7ÔйZOk j ûì|; }6¹ÜàÙ«wGoˆ;ÙGY*ðÓ™š¸,"=]Ý Ç| ×GÇBí%†:2ÕG1‡Ó àzID-›0ñBúŸf¬ž*abºªy¤9eI¢x¤’Hå[˜ÊèFãTLïb-˜vq€Á3‡S+{6w¾®#nuå#ËZdoÍV”¥b´.c"HÕ8°õH0®Õq˶ƒ%YÉ…ù±ÝË0H‡ˆ+ Ý«(Ù6’ÈñRÀI$P¾'º¸âFï ‹‡% ó±o¦_Zoç1È—Ê­ok‘+˜”ûþJ8v‰JN(v”â)¶9s'‡„(Èᔵ»pœžwL‰ˆ4¾Lðóz^–Ài§//‘xÕÛÓj&ôBk:0T=ï‘üýô— S·ŠŠ´Ð¿éaÃkˆëúäEº^ÏH`bÇŸ£ã§-üÚ7O 'Eùp|.9‹jÈ2 §Ë6Ò²0~óÇ”$ºƒJŠ9Ÿ×)˜ð¡qÇ–4Y‹“@Ý1ûõÁ;\E²m­l"×zA®5"5’i¡ê ©|–ÌŠÒ£‡3GЙ¦Äüá,»¬TOfZ ¢À'u‡¯Ã~ˆ† §ÊœBÍo‹òAò€Ð3W`øíÅÏçjKb;j`ºç G‘ft‘ ã¤F=š³î×»^°d‘ÓRR#DGk$D@%©¬xÁA"ï»Î‘? +Uß¶ª™@‚înh<ü.ÀßD+?»LM¦Nœ-Ñ…ç UYšúÅL.ÏØù²Ï:ßdÜyN<:WE'Äün—ÏÊ"@CY4yüIÒU“†Zuå7U¬G-¿X¥œ2^¥èC‰xQþ*¡÷ά y.{eâÍ:)å^4öü.uOk1©*%œ0±œÂ=Ÿ>HÞ±Öú¥c¸/©‚ýyÀ•òžöH!g¾lmU:yŽ6ÜXØÕMÍR9Ú³Òw/浡Ôé hXL {!8no‚à–åÎfœaðE+Óšò¾‹¬”Ùb|q’ÛÉØídá,ºÖ¼Äævw­Å<·„r}a³t«¬ÐûwwÚ¨âì' }÷˜Çç"£“°ä×MYó™w–“¶Âk²ÊD.aÉNVS´ÜöY?"žŒ]8¡NHÔ6!¡\³urv~\JZ–ͽ%ò1ï!DÜÑ4 ù"=ø3ù,•·±ª‘ïÄZÛûš«Tk'¨èª¯s’\A–ĽÖuä?¸_Äx]CÀxE³ f}¶<èÌã3r….¹ÂU’e\lÏÅR³€Øo}(i+C‹ØµÛD)åõÙüœ J¯§ó”«' «ó^‡/ÙÆI¦¾Øl #-¾7ÔÝK©[+Í%/°̼aûyM½à[¤²¹ƒÃ³KMAmw¡tNÿ·Þ%l² ÷H5ŸÊ¸‹dEˆ€“„lr›°GH¸¤¼×¦Açö“šL &.Néî”&î)­eÖ±ç7¹¶×àû»lZ,€ÆžV¶êŽ9„Ra.6‹†Þ£,Ië©¿áŽÄœqVò‚W†‹¾”%UA±_ö ñÙ.50 Œ£ÔÝ}[úÙ0Ÿâ‘¨Ç/b®­Íð&eh–R¤žšàe‹ ]€ù¶çÊÑp1tÕˆ‡¶±£í7 máõi8™P’XyýRìñÍ[ãðÙS>rPïC^ÀöGþÒÄH¶“ciE×Ç~5åIgBhXyÊnÐf#uoh% 2¿Òt¼©Våô±_)°ŽÒ®§…3wkI†ÆI G{Aµç‚¨æÚC½·Ü §)%“¦îÀÚ‡»h£Lú=KÔ¹CÃp’^š.6ò…NYQ*­A1gJ¸àrÑŠÀþ¯a] Ìk«©ã$¢à{®"ÕK¸ÿ"^/IHí3ÕqÑÆ´g|ŸE߈´Ô–]²³'j¸h÷!pì×5Ø|'`e,ú}Ìÿ uN{qh/N”üb'&”¸ú¥èª+‚Ý&µš ÉM;/óÒq¸qn¼ù¤7Ý[=íÅð¶wŠ˽û«…Ί9ûj” †ôì}?ºY—ƒå‘—>ÃŒXÉÅ6ûgYLnl8‡¡»y›{éÍbOšt,¦‚Ït«ÃYWbfY~Vê“áögCW…$IµAm÷\op<ò÷ƒUnÁÚ|U#=é¿R¨, Î@JrW~©p¯Þ]/ &rp’J›­Â¢Û3Üj(Œ­†.õ…5d꿚ªôOÔT»ÌfVw7ó. ×W†“º³a¡9*$¿w‰ìÿ|8ýNuÙ8pàÐùvà9M/ý½'™M#{Ã~ÿQòqž½1_ I¥ó°ÞX]E_? ¦+Ëe—o‹8ZmŠÝÖd’šç^ýëîÕÿöûì endstream endobj 109 0 obj << /Type /Page /Contents 110 0 R /Resources 108 0 R /MediaBox [0 0 612 792] /Parent 111 0 R >> endobj 108 0 obj << /Font << /F69 22 0 R /F8 9 0 R /F47 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 114 0 obj << /Length 2435 /Filter /FlateDecode >> stream xÚåÛŽÛÆõÝ_¡ E+¡Öxn$‡m7€Ç@Š (’ ÐÂÉWâîÊÖÅw½v¿¾çœ9CÎP£]íÆíK(çv®sn£¯ÏŸ½xí&µ¨K]NÎ/'u%ªÒMÊ¢Ú“óåäÍt5SÓí~°±oágq?kx>Íœ>ŸÍÑÓÝßfŠc×ð´þó",jÞÁÏr7ÓÕônë—\BÏîv‹ý¾£mö¸óŠ–Ã:¿!À~ƒö-î†`V;îþE¾¼¼Åì×ó¿¿x­TD™*¤¦žÌµÚyÊ>ñ<9)E]‡ó¬pÎNæJØ:ÐÿÖOKø$*˜M:Ë”¢PåDú¿H]nä„VU˜òçÌ&ÔöP&ƒs)L5ÂùR! c| ˜‰€Ù °B” 1° Ja•>ZLZ‘…¦ w‚8´QWVsH« iÌÝ1 ¯Z>VÔÖ>Bô:«jeUDÛcå–!MK#jYŽh³¶Ñ†s]hQÂ$”þN:þ—GÆœpdÌÿó‘Q%°ºLÈc(z}—cñÓx9œ+µ¨ë`T•Ÿö-Ìùm°ÑhãÁL[0äÉ‚M£Œü~vÐÄ®†\ù…¥!¯ßþÆönã7ÜÝî}ÝÓ ] úêÙà´ÙüÏ˺óݤýÈÐ.hÜÏÚ½#ä:8[8òHä~Ftéô0N=]~ÿëÖ·íšý4fþ>göô •÷Àa¢C×EtÆyÕÀžôŠ© Q– p°XÇÓkmi=FŠÒVÓ—‡¯ýëÿzå7†ðߎ–6á”ÐK'/•“ÈE~&eظðc&· œÄ€Ø@13ÿ˼ñ.óAš}M‚mˆç; múê„M£áKèh†Q Ë–$º±6=^K‚*œÑsxè•)DeÔɆ£9b8îÛßÇïðÒ÷à醹×­WO±gÉQÊ#!ú7ê$+§O¶r_D:±äsùAú2#[s ”h^RüV¼×¼`èÖ÷¨#xœ@yÃx“z™Ô=c³8Ýæfö Òj~F9‡ñâµ­&^@Ú{B 5!×6•ŽõáS* ™ô?ùhfœ]cÜ-Cî'|Èq'L¨ö™…-¨ê¢µßµ 0o×ãJ8ŽoB´wÍyvCJ×ùá ¸Cža]ã_toÞÖæ\t!{ˆŠè±^ Àw—~fŸíA›wÂ4l×q_œá ´›p-Cžº‘ºM¨Œ‡Ï´h)Åç^ãî VÓ¥¯Ã;s£‹Ý£«è¡‚ÆjÍ5öô…fÚíòÆ—–é+¹W3yì¡rý‚ñb|·>?«*¸sð];` ‰ýA1ÍTNHk&•°;¶dÊC|ÿE~°Ti™D%¢ºr-jW?­™Ö2d¡ãrÓÙ!þs‹%,G@xéÌã(x›- »bLÁ“ïÎ*¡T‘Þ-Ò½‚|~H džvüŒ‚¨Fdüž{ìÚØ¹B½£û;y¾÷éo³rwÙ’]š¯[[>…[Ç„^Vn™ÿ·V¹z> endobj 112 0 obj << /Font << /F8 9 0 R /F11 15 0 R /F10 55 0 R /F14 54 0 R /F12 72 0 R /F13 53 0 R /F1 62 0 R /F69 22 0 R /F74 36 0 R /F47 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 117 0 obj << /Length 1207 /Filter /FlateDecode >> stream xÚíWKÛ6¾ï¯0r¢ƒHIQƒ´@úºzéú–ä XrlÀ¶Ü•Ýí^òÛ;/ÒòZ»X m¶A{ IŽæMÎ|ôw³‹—o«‰O}aŠÉl1ñeZÕ¤pej¬›ÌšÉ;õqj*u;MlîÕaªU¿‚Ÿí'&ìa½l§f?¿|[ø&Ðûb’±Ž9°u(E®a¤¨ êܰ†¡+.KóÂÝ%Ñr·%Ù$÷VÍÐúŠô75ríPyWϧ¦TK¤WjY÷wa´7ML¡ÚejÔ<—OÝ‚gôpµAµk\!ÇŠ‚ ¤hõ <) uƒÖDtŠÈlÓ‰™-†7ItVBx:õNÒÛ¡éf%É©¯1Ó6'OÛ~Ï›0ÑâºC;7¸q`N«÷™¶í\²Ú3ý†\Š»[&Ö23 œß·rõ|/¬ðlÀo0Ó  kɶÓê'’hzÄÞ¯<ÍRL¡l×ÃJ¸!±ìm+ÖvâG·#ù3Þ6”±ì4U(‰™0Y¥Þk›ÇCSD}V]’ù}ÆÜtÿ6h逦ëcÄ/PȨ›®xm€c´P Õc&p®L(Z÷õFV|É…9dOi”}5çXòhùüH1$h‹ÔåV¤ –þò@ŒWñ~Ä :øò–šÇv i¥ s«ÞÀúÆ÷ÜOlˆ‡¯%Ì.†§SVÎÓáb¢ %[LäH ùŒlFEô‘,HÜä’Á~¡= aK3éÖ¢;}žF¢]©£H!<9‰Þcψ=+öÜ@iN0!/it'JµË„î…nâI‡ç*‚r&¹Œ;š)Ø|D'6‚_F؃Xð»:‡ë$O}eNÙL Öø€ÊÃß0õ&T×’÷Ÿ,pÏ€ŽµB€^>éÌQAÆøLŠ•èT x§šžYæèÈjÛ=Ys÷¢ÏØ×QüÐ}Ç·ÀÐ;!÷R ¸žÓ³¤MX`¶ä6Œj¼º\È;r•ç§'šŸ%¡¥!‚Œæ ã lî`®ÉTáõšYà… ‰‚æ†D¼Ã×l›€Œ5ï){Ö‚¡n›Ú¡,l7ôŒEôJm„Ÿº\8„‰¥¾²%”kùXîd™Ÿ±¾‘ù‹²ù Ùü×Ùýu@&z‹Æèæu³8" Ûs³„QSáax{ £M3°iã ›F°îŽ;õÀÆÐ§Åe7v韬s…u;(ÿìïíhö+èhöÿ¿GsOÛÑìxwê½Û]´T|¨|K.=²£ mºó.lê‘.ÄÊøïä_ÐÑNÿeèÊÀý÷ðoC§1¹d(9?Î.þ&+‡È endstream endobj 116 0 obj << /Type /Page /Contents 117 0 R /Resources 115 0 R /MediaBox [0 0 612 792] /Parent 111 0 R >> endobj 115 0 obj << /Font << /F8 9 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 120 0 obj << /Length 1208 /Filter /FlateDecode >> stream xÚWÍnãF ¾ç)Œö"‘"j¤‘Tì l94F±@ÓƒlË±ÑØrm+é¾}IG?ŽbK{ˆf†Cò#9$COg7w_’Qê¥:УÙj”Æ^¬“‘Žb/PÑh¶ýåL‹ÓœõØUAâ,Š-¶ÅNŽã Æ+pŠ ~ùÑ‹•åÞq!ùCvÄ/í˜'u^Hâ®sCȌ̒èÏ>¨œ¤è³#“¹Ív¤áņÄÖe•Ah‹ 8þy$P’Ñ·‚C*–†}>v‘µé`F2bIóW&àîÈÚþžý>òG.€—F¡¯ÈžÆ²¬$SŸ¯ès0â(šÌàl„7“õ´|Ü?'³yÊV5uFÞ}Çóž·Ôµ WnžýÈÇÓã~f3<‘?/Ù¶‘YSh‚ÄwžA…» o&OÖptë&˲“„û^騆—Âä‚„Q;V ÁDáÌù9I] PÊø† Äõ+±ØáŒ Ð‚²Ï)aGމ<žã߫ȒÎÿ1«® Þyc7T£º‘ŒEælOœ„\d Â[‹z.€òUÒ )¸ÿóás¢$ÓÖÚ’1ñéφóÝUJ9{Iàlq’·Zä6トñÈÈT·˜Aì%Ah´L(é ‘3§!f€k;eq⩌Ðl]gþ?¦.Ÿƒ«‚ЗtÎ̽-pŽt.•Cïð&•‰œ\à»):ŽyÔˆ.”¥“7‹vÈ^òº8;¨u2HR‹}Çڤ€.)n)_ŽâkíRþ6Ž´©†2;Yבþ/QÈ;¦CGÅKqÅl½‰† M#Šj›ܸð5m÷Ë÷†…#ˆô*WÍ‘û×–Už*LVò½uTÙ´ãä—K›hè§¡›7×iãÑ·¸sqåá»Q œofùÅ]³Ý2ps‰eÖ–ª»‘*’¨¬­%0dè ¢à–,}Ž]r!EöEòã[—^Ý¡—ë,ò´dR^’ªÖ ~˜“¨'w²ß ݤ©$'ÁTAó8¯äï¤}œ¶÷çªÀùEâw& Ò¬.\ŸKO¥ïZ‡QAt€­Wˆb¨Vÿ“3®JkæçÕ¯ÖŽ§ ®(ï9SjÁ:”‡=•7@:•wZ®*÷‡(Ï”Ÿ‡®(‡KÊÓ+ÊûÆÜïª!às£X«>™2<î`þC ð‡„Ÿø=ã ¾ðcoÐH{ {m«–\ö ÕxÄÏZÙôr+»¿Üï;:]77„ú9ž1ÕÎÒC³®:E3Ra†¾YCÚ­TæàJ\9úQ:A2ûŠakfØÈ8°ìœIÁ@æ´Tøwk¸õmÈÔö*#ã•¡(—9je§×úÇó™Q§î óÚ»Œž¥Ìjr]JùÜvùZv×´Ó?ŒŸk¤R`v†t” ÍNˆïRÈßkSe§m=V72úÒq/²ûjcò\ާ֔ö#‘#™ðøºýöèfà§%=¯Ê´–ŠõãÜ?lj›}i·}Hþ5†Ï0†¨K‡tAí\Ôú½¬ý ¶ýÓ’ÀKÕ¦^’$R” 1Þ<Ìnþ¦ø endstream endobj 119 0 obj << /Type /Page /Contents 120 0 R /Resources 118 0 R /MediaBox [0 0 612 792] /Parent 111 0 R >> endobj 118 0 obj << /Font << /F8 9 0 R /F47 7 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 123 0 obj << /Length 1777 /Filter /FlateDecode >> stream xÚÍÉnÛFôž¯Ò  “áÃ!Y´"7Ú³€ºHr %Jb¡Å%+þû¾m¸KNb'èæðÍÛçmO¦/^¿3é(õR˜Ñt1R*öLªG&н ŒFÓùè½ói¬œž7cWÇ*r|X{‹Æn„ƒ>Nÿù#W)/:l'߆íõó³Uðdðü OÃ)Qé%ž>=  bn3Àó¬ä¦sÙšÉ÷±fòLÖ_á×ï’ÆLá*XSÂ'^šÄ,gº¢´vC£œyvà|Ä2§<§õ‘ %¢g%¯U€ï€ªÃnÄΩ¤šÀÄÙm…+Ö¶Ýž×È3›!ꊻ¿QpCËù(u¨øwÌZÈ>zã§OÌv›q8·… “Å—¥‡.Oœ × 9”ÖX‹™©¡³=¡’d(—»½-†s±0H[:Ãç_i)•åé4ï4ýCðñÃ'/©¢º°I Ïá@DŒ¦þª®jø*çõ6—°h ë’RŒÐgÖpŽÑ‡?Ohn1Co¯·¨N'~ÙÕ 9¶5X˜FÎãÁø9–âÛ=Jß2ÿñ€kí¬ð0¸eõ‚ÔîæŽbêÀ›–²q6ˆˆ½kG˜ÇC!,SÆÕ¡ª2ñ3<­{´“º_¶Ì…ý‚ß­E€r5SJ%k4}# Fð£4íÌþ^&€J(çG„^‚‡©¤¡%Ìn± ±š÷ø'¯švÔü¶‘€ë¼´ŽÙd¶õÏq«å4ž=${9îXÔ¶M{\#øˆ½Y;ÿ líDâaÅ ©ðaÌeóA‚IHäÄÃãZZ…èuÜ ý‚rjãMä{I¨Ú¾ÆÌ-ªX+¤¶[1'•¿ÈP‚4c >±6qêÌòFc¡ÑAä¼Ù"ùbÄ@\¨ì˜d.² T„䯵ŸåÖ 0rR /Ò#ß*BŸÄE¹e]lj-jþ™°ÊmAŽûq9ÙZœ,8°·YÉ^ˆÝª®‡a7¿l%^òîÁVIXÿ=NB'kÌÒ¹È庘Q$z$Wi/Õº-ü¯q¢‘¥Ígh.ê„-$axÃ:ä¶¾…Ç0Õ©Žö:2¬g?/JÞ±T'渫Î%düêk¥M¶_J¿Eˆp!®;õŠn9Ê(Ö‡ÜCF¡aCaÍÞ GfY¶!K›‡eâº30.+eEQœ#êr5–Q2AÝßãv]‡YGÎÛö$`dh°Ø0|YTÇ*³qG¾[3€“ÅêóðbG½ÖCÊv; ñT+åË^3„8Íb˜ð!o³ª"ô4þÙ@Î’›,!k¥še>‚ull¹¦ñÁB^ôj»Š¼8 ,²d4ŠÌø5t@ùÜ¢ô»ÂÙž°>r€ÕcŽÚp\Wýƒ«„ð½E–ÂÃ;À›Ý‘C4탤×û\í›V2Â'‡ eð'„DÜRûþ3ÆKUå‘ÁGŽ¡}Fêùa4 =£ä¬‡Ø^œÂLÜÀºé{Ÿ9U'©i¡—ĺ)íƒï‹À¸f¼ ˆ-+5 ö4Œ õl3¶%Ó¢'Zc´3•Gç¶%6Ëíd{»à×¹Ã=iÑ´>s°³]ÕÛ Äͬàþg‹É6PMŒ°húçû†.ÔdwVŽ*¨|*Næv”Qtò gl?”°/y0ª~䙸š¬ú®N¼(Jí>ôSÑø˜ólÐòÜT{¡?Æb/JTëFçÄ sËÄžáJ~/MH¦ÍC½¼©†<¤y…¿“àyÅŸ7üã—C$MÊj´“«ÓÖ^ô9ÕwbÂp'EïªK$úœ †lE´èµm#­eLÙËòÂEØL~Š—Â|Ø Ã/ݾõ¡pUQ¬¬“Ú‚+¬\’j)yp¨µóöµN«šØ~î…‰€_ ˆÞ»?²âåwutŸüÅC-•Ü@„-¯Ü‰2E㦰™\ÞU£‡jþ4êßwrçù±ˆyãz3“K‘"WuvØ@i3 ½FµïBñ̇Ÿey_Ëûw©-–¤á·¯ùÆ™ÿi ±ÿÌÿN£g÷QgüI˜r±-B ´×Ù)â½x;}ñYžüi endstream endobj 122 0 obj << /Type /Page /Contents 123 0 R /Resources 121 0 R /MediaBox [0 0 612 792] /Parent 111 0 R >> endobj 121 0 obj << /Font << /F69 22 0 R /F8 9 0 R /F11 15 0 R /F13 53 0 R /F7 21 0 R >> /ProcSet [ /PDF /Text ] >> endobj 126 0 obj << /Length 3451 /Filter /FlateDecode >> stream xÚZÍ’Û¸¾û)¦öD%.‚ ™ÄÙZ'q•Sµ9Ä“JRë=p$ŽF¶$jEiíÉ!ÏžþAŠ#{s˜! 4€Fw£ûë†^ß½øö+oŒI뢰7w7u™–¾ºñE™Ú¼¸¹[Ýü˜|Ÿ.L’/n±Eòî>î?ÝýõÛ7ÕMÖÞz˜ÝÜš*Í+Ãcî¬]ÜæÖ'ïàõûwø^$§…­’'h8,nmÉEò–>ÂlÚµ' ÂY:j^éx“44Ô$=/ÐpßjMï3“c××ßãj'îmiîš%N€s¿Ä¾2¹ÇY{&Ët{-à¡æDÆ=N Ä8b¿„OÂts’5•Ï,%c"1åe™Vµq‘˜þ|)È*-l¡ýéâÖ9C¤O8àæ6¯|ꌙ“ê˜øže eò‰˜8 'Èì;ªD¥ÚÎ0ç=ðVéÚÿºäÍØ˜`º[Zø¡;â‹ONjÐx ±£”·(²·ž{ñÖl‘‚;»~üɤ€ÿØ@HÕ=“òNéµýÞÏH¦" Âï³"Cs†§'ØgeòfQ¹„9‡¥Ï´-tY§e=‘3˜XK 6l‡Ÿ›šôVø4žÙÞˆ‰Â÷§ m·ß šà/¨1±ì¶Âón¯ã@œÈú‘)x–eB_²A²q ÍÙZ·©“¾ý<ê¶ãîf-¬´óȸ bjiör@L¤]lç J˜¹æ¸GnŸH’ÙX†tØ­Ü¢ X6Ì–4»é t[™ðM¯ð‘'¦â' ‚凊Ø÷¬Yç\ò·yøÄ£èÀ‚úÜ^¸‡¡±Áç–< A's䱡7ûõY”Ùy®GG¦)3íÈü®:ƒ¬«(¯8Yå¬ôg2;ò]­G¶rê³hìm^øÔVךìrzÓeÞâѳ¹êø’oëÓÚòÍåtyêòÐÏLÂŒ.VÓž×€c~9Q¥Þºë~Èúà$—Lv?Ä”<ò€z\öÌDÓË“[õ‘­*TÝ-$4{ éDLñÁí¹É0 ÏÄúqVŽ>-J7¤q±(Rìã}–ͪ®¬¼’˜—ò4ó¹œ9+‚Â"YÏQìü’1g`ÒüŠ‚=Ø­vϪÖÍó¯·ìe|@³á<xsâKMéÉ4…ÝcM8®.½KóÁ’…M“Áþê2¯8[WSW;Çf.h×Y~½BS¬’,;˜Öç©éü_³Ì®Æp ø"ÿÂŅ̃GæœM_'½ s߸m¢‚­@è!ת V#ü¤ ˜«%©‘ƒ…Ön/‡´¶È‡vfÏ…ß¼P[ýEµÕ€µÜDm¹ËèüÍ~Îhr•û_i3fN@–º¬˜Y=ãÕIña†0‘Uv@+ÍØtžV&`´3>¥L]ð9~ÉgÊÎÌìYZÿµö•ò|–3‰CšxÍÂ< òÔ}Í‘Ø4›Z4yk˜Ó²Y£Áɸ¼wû¯xâô ·Š“û|b¢ë– Ëòx~+öïmy=®.ìe„aÒ!¼ÁÇ&fƒÏÒRlB8·ÉRÉÅžÁAšçDáÃ(Øß=g8øa(çÙô4ûqÁŸíŠé–6½Ì }œfœçDÖbÆŽ³-)Ì×,·ƒ"l™ËЬt"a­îó“ô_dj83þ¬2T´¸ TêÁãљ԰é²Qö vÖoä œé@ÊòÄïNv ä8’²¦¶ï'&«Ï»IÆ™é4›|d]ŽÅª®­6ƒèj^q`T¼W-ç©æµåzÈåÐ’7£¬h ‹mÇFÅ“à&Vö…j׺Û7ÛÁÕtB‘xÛxáàÇ6Ï&ë aÊ'âóÀ^Û8;G ìa§™PXÀOo×Ë®øØ»ìjz?òÀ.Ô›4a?ˆ™1ç˜H_mö÷œ ô]”Ò·’äkÛ¯gËdoº[øzÍ1çúoŸ´0¿ÝŽ T-ù¸ãIJJ%g2¾z0y-£KG—.yý´àØrj“#é¿x ñÏ`áÞ*ÆêÎúN‡\/„_ÔsЇ‘#è™V†´ŸE‰Gi8h~AÁEUÜ_r"׳âÁ0eaÝד¤6ó!Ä}¦;—Ë0çÒ¬¨‡ü‹!Ì)Ekfð²$íó±¥#”Á£Y‚Cò–ê¤'u¾Ejl>½n˜•#þw3»¢”ÏKê'ì‹ÛÂæt¤Õ;nú¼Þòë–î Œ'ný/?P8XÁþ„)ËÍ¿zs®oã'í[A-ü½t·M[.~;ÚA1¾Ý‚ Äg›+L’!Qs„QÕ=¶‹ƒÅWqÔÖóÕÅz¯¹j;ô?Fô¿, ÏoÃ|ìú¤u=âßO:Ã3ŒR™7o9»˜ÞÅ!Ñpïß µ TÚkÿ;pNY®pÐ’ý»¤rf€íFË»˜¸V襵°ŽD ­y*~¨åâ{Ï÷Cü‡Ÿâ%ñõn<´£Ô*;š€ô¶‘ƒc/A¦eë|” ÐZüÈø¡õÕ°ü>bUdŸ;JV‘B¡ñAWŽÙbæץà8¤}ÎfT Hçüë­HóÂËâe <Ÿ/åÐë—Öíöë]­ÙQ2ajÏÐÖDïË+5[6øš.èù`cwzG¯Yêð ( ¦¡û2¤øû6— ¢ŸÇïèî ÝvÂªŽ“)¤ÊAïRƸò~;lP7î•áÿé¹¢Aò>ïešÚåŽ[eC¸r"Aq15—€Î€)i•É£<ê‰ ‘¼”{ ÛùiZÏIyàA/`ËR²¡~£×`0j3LÀ2–ÎCÛ^Ð*¼ økCf›M TpT“»Zj|£dk|Ѿ‹lÀÅuTL¨BÜO2(˜m\S#²y=„|½€RŠZQÆÓöCjíÀÞP¹å±¢KøæKõpÃK ¤‚›ù€8‚æÚE§}Æ t§¾ŽrZ_QDÿ,3¯¸ÿr[aO0`Óósß åÿn¿V{ÆóA‰€O9¹¯Dnz~Æ`>ÿñŽŸ/¸èm¾vÆWÖeÅ™‡þš —I£,-âpf$$l3ÚÞª<²’SHx§YD™U¡¹ëF艃3/<ÄB™‡«ueT’¥Ûw;\ªCg«õšèîê2UàÊiÅï¤Õ¾rÿ³2‚ÿð’ ñS:x#?f’Õù9‰EYë€ ÓKþÊžGÝ›?“ÅÆJ|ƒ+À}j¹Ô-‰9vÑÆ±Ö‹­m€Vø…2Æ§Ê ßõÇ*ÃåréFI¬»ã&ªÀìdɇ#_-òJzn–”FE$œM•ˆ”Œù\ ŒOÝ-ùë’UöÐ ÌVlŽ EìVZ϶Áå+Nš)-ˆ¢1t0ܤ#{nÅuã o6¢S’1AA–$aÜX/VN¤Ýÿ*žüfRlÊ3ØunÉ-è‚âhØ<õÃÐwÈjü§Tȇ*ä#jÖ†ï"Êv¨%å¶H}îÇÑ¥?ˆ|·êXÉWžœ¡~‰tPrf7–$¸‚ ÍùÄS¼eòàԞ§_0¼øËÝ‹ÿ8Í endstream endobj 125 0 obj << /Type /Page /Contents 126 0 R /Resources 124 0 R /MediaBox [0 0 612 792] /Parent 111 0 R >> endobj 124 0 obj << /Font << /F47 7 0 R /F8 9 0 R /F11 15 0 R /F60 16 0 R /F14 54 0 R /F10 55 0 R /F61 14 0 R /F69 22 0 R >> /ProcSet [ /PDF /Text ] >> endobj 129 0 obj << /Length 1683 /Filter /FlateDecode >> stream xÚåXKÛ6¾çW¹Tb­H=,!Èaƒ&(zÈÅzhzP$®mÔ– ÉÞMúëËy‰”-A²( ôà9çù ¹ïÖ/î>dѬ‹Lg³õÃL).U2ËÒe¨ãt¶®g¿Ÿ´Šæ¬½û{¬0R³ˆXÖÛ¹ Ì|G*ø¸²ãõ&QPî7mgç;û;Ùßö@t˜÷Ä_¥FRÙ3£écCç¹^%J"ª 6¯aiÀ®–Ä5nxÀ££cÿÔ=‰;9ÆÏó…΄dá)5ëG«í5í½˜s.Ol\Û EÖMÖ' ¥Â"e÷=mÉÒ$Ž_ô4åë EbGëYø:⇉Ìã\çih†B£{RÈFá]PÐÊ~õ·G÷7œ¡@è<>E*9±‰ ÑOœ0n;XÙH Ê=4Æ:ݪYƒÀ½ýÙê"¦ìZÇžv̰§ùgÜ&â›áú±†S>Þ±¢£ŒU8³«H´¾²cÿs qt{9-d¸u+{1´,»©Xƒ,s¤ÙhdSÁFÃ'¦Ù ×&§V•ƒÝΉʃ{Zh0½@¹hI æ H†=»¶¡¥ö¾ìIßÏ­ á܃‘*äÏn¬YKC°ÖpBBŠU7=û%u•c_d™ðRþ*ãòhhžjÇóErÁ’)+PfK3_1ðÎ>à¼8~iakÁ8@ U8Ť AXǃƒRÒ¤Go±–R¢ñ®ɬxÆÇ²ó´¯ÎœÄù’±'("Ví]_IZ»y{´ÀÔI(ÏddM+%rP¥í Jý–´à×lÀ=g6,² ¾bÒ:ÉéŽd1:ÃyLæ0ñqžfœÞFéXÃ&P€º[I"ÕÖc²×pÒ D/ªWÀ‚Té&XGæ@fyN„;ˆ™—zOœƒ(ë¹æ¼ÍžÑl”‡t°ã¡AÛKJœ0eŸ ¥?Ì-FaŽF.j0^­¡!Ã8™€rDTPóöÞsš—~NtÉ“JJ 7좶^• lí˾‘Î/ÙƒV)噥ó"TI.†ý|m¸M”4ÕÂ0 Iäu Ô¥,Þ´ÒŠÄŽ3zz¸§ßû3å¶a&5÷ÌšÚåFZ ¡iÛl¸t€ïÔIßزÒ×)Ì~@ß Ä.U÷«×¼w¬Ü”‰”Óù÷äîWGU’³ÚÐ[zÆž÷,幓ÆçZ”±]¨Ê×±¢à5$@+޹bpâƒà˜Ÿ6²½åáÖÊÍ–sÒ?§äyÕ€Yî$^^gtnÛs÷+w¢Žÿ)J#Qk0Ïùýið€Ü1ÆWt5à ~ï|N–3•„q’ih„Ed#Ö¡9ÌïæÖ&«û8ràîxFÑòhi›þͬ;´ ó(¿xLY0Àm|©±Îå^»»Bí6`N–ìó…/Е܍†FÚˆû Ì\Z&¡Ä¾×cF iXÅX¿WµxŸpw –ïV©TÇ!H˜·ö·´?¨°7S…# öYÞàO²äôLzt¤—,ºäo-®çïKÿ(‡…r (¯²ýXÜŠ=ø…Hb aÜ…D?§\À-â8îÝð¾M/•{ë)ÿȧBx[5¥£0Ïõøw(NßÞ’¶é’=«;ã4̳ÿª;qZb%I?‹‡Ç§Î¼cZé½W•†lïíP¹¡vÃø‡ƒWyE];kd×k"ñí†ò‘wkÇž­&PâŸÞ*J–—)4¿_±Ž¾Ϙ$‚” •†%‹´ƒçsC½a™x½!ÉÂ"Oä¦þÓD÷HC­ aø:ÇÿQ»©³PEúŸDª(L’AæEÄØàÙþ îDöðû† N¹›«¢"Ì’ì[›×š/m;.ÛÅúÝê)2Q GV c±›g-Ôìß,ÔóEÒ³›_9‹ŒgÕ_†øzˆh¨nù”rwD–óŒwe2ì—΋ț‰rnÍ*b ¡ö®jë Í‹0¾x¿~ñ76þ‡v endstream endobj 128 0 obj << /Type /Page /Contents 129 0 R /Resources 127 0 R /MediaBox [0 0 612 792] /Parent 130 0 R >> endobj 127 0 obj << /Font << /F60 16 0 R /F8 9 0 R /F11 15 0 R /F47 7 0 R /F69 22 0 R /F74 36 0 R >> /ProcSet [ /PDF /Text ] >> endobj 133 0 obj << /Length 1354 /Filter /FlateDecode >> stream xÚÝWÝsÜ4ï_qà >æ¬èöl:y(cÊ4…!3 åÁq|ÉÍ\ìã>’–¿­ve¯¯NÓR该Õj÷·_ÒW‹''󬘢Èt6Y,'…6Ë'Yj…6édq5ù5ÚOU´r¾k7êiœj}æÈܸqcë§UtKK¶c†S;âx¿¥#7®úßkâ˜õSW4‚,øÞÿtúÛ⻉œÄÚˆB¡¶R§u£BQ›q-ArI’á{êÆŽèpŠpcIvW݉ZŠ<דX)Q¤„RE\%IÙ=hw0Ts ¸¨[RÿŠ×=ª†Ämï‹/z‹jfýJ¦Ò}$’ OßC¹XÉÄ2åNQÇ<°íÄÛ`”0:`’ #§&LN‰^25(X߸±!Ó«^²ë`B€-g2÷‡£8HŽvÌmÂ>ÉD‘'Î~÷çÈ1H‰Th]†_âF¼üWNPR$IwÄ‘SÔñ £~ð¾½OÞEvð)Yˆ,ÉÞŽã–²sO~*;P?À—NÌ… ?>xÿ.{{×HW®þšÇìMÃH6’Érì3ó88ý¤‚0"G¾CÜlPµ²ÇdŒŠšU¾Get¢ú 6R¤J å|Ò–åïÙþœZ±¨ªYdU¬<…Èût³“^Ã’|{Tú—D žŽ…Ó® vmìñŽ×ïô»]°èÀtãû¥ŠÜïb¡x2ÏyÝID‘$NU+²Ü ª‹ÔÏÆ¹g@ë¨=ì©•£«üä Ãxï¦:Z0îþ[ÁÑ­·ÀY“HKºi œ“‡"ÂjÕ[”_â˜ÓþQãÍæ7åµ÷†˜Æ™3ûÎîà¸KïØcýtT¿æÍ ò |…K—ÓXgQ½ ÞϨ x¤ódðx(aì´×¹ôybãŵÛÑmƒZÀšÞ’Kô¹§M~Ønûíe€ìûí-„íií­OÑ튂¦m8‰N£ùŠŒÞ1!7ýÉß¾8Gâ¶õ'³ä„Ùe»EbG åmmÖþn—'®þ„Dg(a‘yƘÎT 4H§¯¤2;,Vt òÍ@¸SCùÙ7ˆ7 WtUuâ²teŒ?m(<›Önd˜ ­U¯jíK0f–™ä.µ²RKRV%ˆ JhY×ïm:„ì%e$Ïe×]|V·>Ù=ù=­îXÿ¹c…‘÷ªï?²ÚéIT¡"ëè|ŠßdÌÿ˜6?1ö‡µ‰Çbx¤ð7$3¼5Öfš¾@þÌÚØŠèË®(«è˾#aC…£”ÍEjÜS»’®,xÑ!ÄPÁ²ع~ů0”’Òxåç0þNŒ%ë P@d‘FçÌôÎß«ÔJËí^[ÝÝKÀ¤u<»=+íZ«O*•à-¡{xþ zØAÎÞŸOg*ðh#Z‘E¡Õn±ƒ°Ñ„ŽöÒñôœNDg¤¬¥„¶¡uÃö¡†#š|Ýiл¥b—AºE.ˆmßaÓC}Í‘¾!>Òëˆó©_û°æmèTµ¬ºB›P…»g÷Zøß¹ö艾pö¥/”[×tÇJ]n„ÕjX¿‡å9~ÎxÌ_¯Fs ÔæËõs­¦°C/ÅnõcGð4¿j­H“Õ ª 1ñß2°ò¥BY3,gÞøŠã{+DU>|~‡ñó‘7LÅÊ'„H–Yx—ié.ÅŽ>ë(ï!¯‹è颥ÌúÝÜ¢±Ýší6ƒÝGŸçÂæN›Â]´É&ãäÉÙâÉŸøô›e endstream endobj 132 0 obj << /Type /Page /Contents 133 0 R /Resources 131 0 R /MediaBox [0 0 612 792] /Parent 130 0 R >> endobj 131 0 obj << /Font << /F69 22 0 R /F74 36 0 R /F8 9 0 R /F83 134 0 R >> /ProcSet [ /PDF /Text ] >> endobj 137 0 obj << /Length 617 /Filter /FlateDecode >> stream xÚíVMo1½çWìqCdãñ·9‰Ôœºâ‚8”6A‘ÒšV‚ÏØžM¼ËVl*µêÃ(³Î{ž{ž5ï›ÉÛ…W•çÁZ]5› ¸·ª²Æq©LÕ\U_ê‹)ÔßÑÖhrÊhU«)“B}†k})„­Åôkó©Áƒ ³uÁ6Ãlÿ8Û›áÚYZ‡ƒ¯ _>rÀ:[D‘ÀP”=Eø…¬uõ¢ˆ Ð8E1h ÐÞô¢álS=\Ü#P–å~º‡‹¿žÖu‹§ ‚I¾:ø~Q+*jLA³‘ÍF4YÐñZ%å57®bZq¯l.¬AÀo´©@f¤ª—è.³;G7–~¶E»F‹]ô mwd,Ð}@»A»,à·´¶OÑåÊw{…ÑýDû@„x̲-m¸¥õÚ|_ð²ÅÌNu›~^l·>äÓ}AXé]qÓ½½,rˆ ;ÚpM'xOí!¦Ìy†2$C(DŒkN)3þYJî=>§‘{oÏiä'>=I¨'½<ímü­nÿ/ugb=“AsLa¤ŒžGJ¸p/¨¥ðê´äþké•i)> endobj 135 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 140 0 obj << /Length 1118 /Filter /FlateDecode >> stream xÚíXKo7¾ûWè¸*º[¾@ÑCZHÑ"ôÒö Øk[€­M%¹H€üøräRòʆ`R ‡ÑJäóÍ|Ü™¡Þ,Î~¸zºèœ™-®gÑwÞ…™³¾SÚÎW³?šó¹l6,>[)ŒO?B‡ÏV‰h•¾ÉÒ%‰I`N°èykd„'ÍKWe¿¿¿ÎĬU²‹Ò’Ÿ ’>Ée’¿š·VéfÁj0¼Lr‡Ã#^Oö ^Ç’±wŒÅò¸Å9Æct´#<ØÐã¹D­”Bèæ—ôãbÞjiÑ—$ŸH@¾¥lñ±¼c7ózœþ›-,‹×­´^Àæ ò{ñ’Ôú½ÈB 7¼½Jë•õ6ùJveÅOdÿÇB17cædÄñÀs]ÅuŽ™Áõ°·ÁqË’u ·RtÑFB¸etŸ:-bL+Éšš°ù©Ùª9Y7à¸Üc›QiÕÃhy’aù-±œß>W±\ÇÆ”7€<²¿Ç1ÊÛ1²œß¢Ž×ÄÿËþ–!†GYþŸéo”iéCgS¾¦¼mG¾w…¨#W¸¬-ëjŸÞWzð¼gsýI•Ñz§±bny|Åzˢ׺ åTUEòvÇNÅÿ=›I*?Ñã+oüuꜼeG„ùtÓ!ÞqÎéÙ–òLÔÚêÉil¾áº&ªêí«L¯«SâYWWühÔƒqÕüXq–EN¨ƒ3›æCÕV‡ÄVeÞTeÝ—ÖäYBuÖÎfdÕ%xš{¹jÂUUI¶…Ÿ'ýp•’—è?Tå‡z ? äÍ}u®LÕæq+ø”á`¹OºíA†Ó¯áƒ™ðAT‰H–Ä3Ž=ÁE88Sn̨Ňüú…“xÇ}°@Ä³ÂøŸÎäÚ9_Zž–׿º&Ê>>xSñÉaU¬ÆNü1ǶŒŸÀñ!þw/ˆMº.Íbº-)·%iºhL²æº¨¹—XÜR¶ÖÂbÖ½]•‚cw\ÖØlhð~H1õX‡ú;,8ü§†“÷÷Z4—Ãz®/ÜeÅ þV8Ú9kôX§.sÖ[ô8•B“H ¦6{v6T³pè \Å’Ý4n Í;·XžäÀXå.,XA½»ï¯hÅv×ìû@P©ÂæàUTAIÚìù0< .»ßôÝÜf]ˆQ$_×>¹jƒ&ÆE=zYÊòÇ»qêú£¹3š‚†ò’å ¾vßmûÏ„³w{Ú5¶™pî*u#¤/´h¹áíÀøÀ ‡ë©ÓÌÑñcI‡í¦Z_è¥ÐvšÏ(ßjŽ_Gu.¹aí îJÖÕ|ÅÌ·@Š•îL”û ¼Î¯)ù?ˆÂ5ËŠa­x¦«Úûyµýø·µ{U‚Áÿöò º²”yœ >='þ ™H˘Ö~£2€ÈÄ)ù"ƒâqÂSnM6uÓ¢ †hŠg狳çbÂ" endstream endobj 139 0 obj << /Type /Page /Contents 140 0 R /Resources 138 0 R /MediaBox [0 0 612 792] /Parent 130 0 R >> endobj 138 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 143 0 obj << /Length 867 /Filter /FlateDecode >> stream xÚíYMoÛ0 ½÷Wø˜tkQ¢d]74@‡å–Û°Ã>Ú¡ÀÖ ëlÿ~²M%2#ÙNâŲa5¥¨§'êů}¹¹¸ZÕª¨KgŒ.6w…³¥5uaЖ °Ø|*Þ.Þ/åâ³[°Jjµôþ³ÊGIOíÃQ\úxÓÍ‘h»±^¾Û¼.ªBȪtèò @fåÃÒ3µö¨ó (zÆÅÁ‡ÙG•Ãvl¦¢×‰â 55-’*n§ÇDqI…e¢¸XîåvŒÑØDã-¥“Dß#øMÅ(Yi»X-BÛíš[8®šÖáØBž!ü–í¡Áü"jIcdõåUT˱<³—-FÍò’‡q|á‘<ܳÝ3á©–á(gäA%xã†[göŒ/ÎCè/Íö—âA&$æT4Û_ÜÇ8ÂCà«á!•‡™<Íøç€B±{ÌùÈåá@^N;€øL³:.\]ƒý—Îÿíß[ÚßøáM7¼öƲŸ>î}|%Ê>øø²›±òÃ_>||ŒÒ¿ÑgíòBÖ¥uª¿¾ Ì¯hÙæ@ï(î Ö=}þ@g^»‹v«Ö•µ…~ï]Gn·L ´FE}ÙêX_&ÌÂM´&é¼¥³ô…µtMgµõϸˆå·A(cÚ·‡ÖòPK·ÈªákÔM\G62Ln¶Q¥ Ý÷(OÆ©“Ubòt7ùƒÙM>Þ@Ú“ ¤Þ7L=óû¯GýiÏÜ8LòŒc•~é#ÝY™ÆŠzXý7{÷!î…çmÝlf1£‘gi´òlÍ¢ çdÝãü|¢ïb̈Q|"³ªÔžÖ,>'È¥´jšalœUìÀšuÞŸÊ€Ê41ì\åÕª.\é ˜æ¿’P5çíÏ]˲֪¡Z'uq½¹ø Ÿƒÿ endstream endobj 142 0 obj << /Type /Page /Contents 143 0 R /Resources 141 0 R /MediaBox [0 0 612 792] /Parent 130 0 R >> endobj 141 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 146 0 obj << /Length 850 /Filter /FlateDecode >> stream xÚÕXKoÔ0¾÷Wä˜PÅõøm q@t¥"VªÔÄaÅ. (´‹D¥þxügà}ds =Xq<Ÿ?ž;³ûzqv13¼2Ä*%ªÅçÊj¢•©”Ô„qY-VÕ‡úªú‡k×֮ݹö û?ÃxË•Ò5mZü$Xé:o+Zµ ˆÙîqæœ5kZ¦ÀÖïÜÛ2 €+m™aŽ ­K, ø.&²,ò 2¥€Li'žM8¡4”µ”XXEŽfá5šEXôh9`IžͶ/³¾Êún5+8”ƒ° %Ô/\,T„‘,Äd 5 æ8#Þ³ÉÌæãǤ–áÅöv<^;Q„ÌE@‡û ~¢5…ˆ…è¡6QˆQ¡?kN"Îß.b ôËXmÆ$pŤ$Õ(iÄz»Aé©·8•aù„;Áz!Ê&cd¨Æ ;ú+¼h#h§¨Vâ!àXÇEƒ­Õcò/ø¦˜ðTr¡X¹<Ò4‹òñXƒ¾ÜF˜?³4O5èGä ÷å΄§t2ôSŠ´Êî9ƒÑ‡w!”üxt&V›å¡À¦²Sv"X“©S¨œò0)TÏä.ˈâðï¸éÒêU–+«0­£‰å¬¿·ë çŸßq_Û?-Zc©®/]ÿÇ¿"nÙáZ¡A̶Û87”›‘ø±ôGÈCoe¯*þP?¦ûo–Æb(¤ÍkC³«+]oÛ†ù¬K?Àñ—Gj±¢mˆär|Ò¼‘ARÌ R‡É׫›áEº yó ]ÌLe‰ULùÃÀ®Vl#xÜWxv¹8û L Ú$ endstream endobj 145 0 obj << /Type /Page /Contents 146 0 R /Resources 144 0 R /MediaBox [0 0 612 792] /Parent 130 0 R >> endobj 144 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 149 0 obj << /Length 1252 /Filter /FlateDecode >> stream xÚÅXËnÜ6Ýû+´Ô+¾É -´vÛ .‚d€.Š.¦ñØj{ZÛiœ¿/‡ÒF{ÜzA’ø8çÞëK¾ž|uädå˜7FUóÓŠ·‚)c+£-RWó“ê·ú§¯¯B¹ eÊu(Ðþ+½o”åº6¡ÍBiQ$j…v¬m(nÖh!ë׳FîS†.%v(>¾ßÙL¤þï¿, Ÿý>SµUÃ[æµÏ\n€ûnÖHcl}j®dݤý$x5Þ ,ÌÓûQü Ÿ5Šv•‡¤¾ ¦ŠøÁß’¥9ÆÚ5ünÿ1êvÄãØGl³ßÏ0¿€]8±Ä2É—}[vvšô¯‡}Zb'Aìã`“á>Ì¿cá 0‘%Ø5@YÔ|Z£–ÈÚ‘ážp°à  ·½8È-8¢Q 2oš)eo â–pÐOí5²×$…QM·i’Q·qpÄå¹%±¢—¬€Þ6÷ZÛÙ‘ƒ~ÂýÖïe^¿e°oUŠOq<`w$N x?åö%H„£V÷ÝŽÏ}} Q›&Óëö¤$s[P2¤nI¸pÓ”8±B/¢uJ6PSAo½ý(É•–Ÿ ù™èiJŠˆÜu´2¥–Pr#ãe·Ä~”Ô%Nb½$&”Ó”Ô L™Îs›”øÈxÛ™bò¿x/%=þ‹-|<¾…ÿ§eÅó,+ŸgYõ<Ëî#©W>$àÂäœ ¥Ç´·Èü<­ÚH)S¾}~ gxŸ?¦$]Ø”¥¯>å´\JQŸ®®s#~8_æöâäcnQÿÇâëå*ð±u첼ȯ>—¾çyÍÅY™)ž NòËd±»dƒLŒ-Œb^¸3¹!@®2±hŸBó"Ò åxà"ú%4?á3^ýš_p$Yuhû>¢~YŒÏëÃP"†W¡|“üuÐHË´³U#8óÜAmœomýVˆ3Çlã}(Ø¥%†ïCû2y›¾%ßæŠÎ”g XëgtøP.ð|«ì‰8ß; (F]u©pŸrYˆU“P’Å} ãéH·s¾÷gÂUO;™¨$(‚œÊÁÉÚãüµÓC"í•ÜLi(;d€;.Æ[ɤ‘A ‚9KÔZ¶ìÂ\¬†:¾I“(Á„Óë2Z‡Ôƹ`Ö¬³‹ÂX‘ GŽr~˜ÎDm7Wyã—]ŽÜ`ÜÜMÞ¶l/‘_#«Þœ“£‚ÃŽx‘ë#o<ÅõÊh1¸îêãÎ&è–H@bŽò›+ä ™C=Ñ}ÊhN@ir‰j¶Xºí.ƒòœœç9ßÙ].[©lHø”ÚžÊÒ,3gqñí縳»DSÄDóòÊô¼–ç¦Ð´øP‚Ðêªäž!†Zæ%AÁ3ç W™®Tçÿú—» endstream endobj 148 0 obj << /Type /Page /Contents 149 0 R /Resources 147 0 R /MediaBox [0 0 612 792] /Parent 150 0 R >> endobj 147 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 153 0 obj << /Length 828 /Filter /FlateDecode >> stream xÚÕ—ÝOÔ@Àßù+úØ3ÙÚýÞ5h"x ”à%<N8ćøïm·GïÒÃhôaÒmw:;¿n·&Owœ,\åQÅä´¼®´p…ѶR““âc¹?âå%È È ä|Ä´å ¿‚ÜÄÛ8äå”®aê:ãý§ÉÛBÚJ;S0Á+ÏmôqÔãåùûIÆ­‘ óÁ ³¾r– BöÚG3Û0}•©Î²û`r>bÂpzáþ–N:Óì½ëµtCH5åäb0 &D\œ=¾ÙM·LcK ,I ăÔ(Lr%qN“NÐU¤WÓó:¹Ï×â%L¼Ù1#ÜãœyLa€³Pý­ßp¦H$éq‡BΤ€j,“fÕ[¼æ _âêî"£±hŸ©PÄì;Âê¼}ô†÷ ßéD`r8~ÕC’cCXáçÍŠ°>®»¡2çk‹@‘åÀHX»XlF„¼‚ñx€sÅ‹x‰–ZŠVz0Ü#WßšíïÓ6‹[í.ìßKª’Ȫ˜ppT1UfŠû0fÊr]n6Ȫv¨èlŽ)J—H’›!¡ ŽÏú¶Î–:ÞO6~pã¶ endstream endobj 152 0 obj << /Type /Page /Contents 153 0 R /Resources 151 0 R /MediaBox [0 0 612 792] /Parent 150 0 R >> endobj 151 0 obj << /Font << /F83 134 0 R /F8 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 156 0 obj << /Length 1885 /Filter /FlateDecode >> stream xÚ¥X[s7~çWøÑîeu_1Τ„@!\&1tàa‰—Ä%^§vÜ6ßsÓ®| —áAY¯ttôéœó}Òæ÷ñ{‡•TEôÞÆ1ÁWïB¡Œ'ƒ·ÃåH hÿöŒ÷axOeÍPö´Wq¸c%´š‚æ¡9hZÛ4®»~5¬ Eñ™Æ­ôjïÇOå`O•Et‘ÁÔ0p&€td ‡¥,êi1^pÀ=Ù‡‘ͱ‹Ù»ZVÝLïf2`VF¸ L (Kó؇û´ÁÙ…ï˜ÙLe;‘pK ÕŽñJú yºn“ß Ê~”g©¶nKcJ¥ï2°)ÝÕèP}¡«á/Ð6@ªÒRñ;ÍËèvr ­¡ºßÛX}WÿŒ5ç€Ïx~§þÎØqMuñÏÇS"¹ÂgʉÉöj§ùN'ñðÌ„-œI7t¦)*cšï˜ôs8íWp–S¹Å¯ÄS qò¼'©âº@œ!UЃ=£áìCk¡]A[à%½óŠðûP¢ý9´)4D{íoh+™· ²íKz§þÆ|6Ï”øiUD%žËà˜ÍÛ<3 ^VEu‹Îïó]6òµc½>e¼^ªå(~lÆ­RZuÍâŸK<¾Šë 8’p|’-^Èû¹$Û„QKŒ¯ÄdÎÝ'Ò•-³ß§ôN¦Å¡dw| í’·*J„})‰Z]‹Oö›Ñ¥/‚ö’ºªßÌ1ïU;S„2®‡úDÖ_I¥œòÂH .~ÍˉÝYzýÙBÔ?Tˆ²›w¥Ãäý)vWYVVLä »þÉžféh$tà-ɼéÎ:.Y/%SeçeØu+Èx TYš-Oùu v¢ºN“­VvBˆcåšx¸îàùö‰Ë4kW5FPdï.“}+’—®p.C²Û‘d·ÓêÇä·2•Ùj7¶]Ê9+@)ln㜒ͬ…-×dGà”Ehë”Ï Û½ûÊ™0{bUxïÖ7v$f:¥àjz¸Á÷ÙZññ&ŠûE/4÷†køuFõÕWü ôÜ@û²‹0o¶Ê~ûX3Ù•%]BË£UIJÒ‘·7ZºÝZ÷m€ÐÆzß!{Vºâ4D-B{Ü 6$íi“” >a¾`Ðì£ì°*e‡oq…÷õ‡Ï €¡ÃpŸÄá“D¿ÅÎ*i‡Š¡"â#P²Ÿ7lQ“¾vâÊÊJ#óü\‘;˜ó!YÖ-)æ„]R6—2 G.êeòõ®TÂÓº;-Úå´{M£îI’wt´êüÌÛ?Ç«,:¦ÒP§D‰¢ó<9 'ü™_b>õ|(T'…‚RqòxrUË Ã…áÉ|äð0€GÏV,\\¢I«(«îShÝžÖ¦àE¼+L@ÂÀ,íúdiIÖ>„ß5W,³Ÿ9”<-Ý}%,¦ä<¿Áý¥!ž.oQek ­;QÞiGw‚&…ΣúRyе!}#Lyk[i¬Ê"ŸÜÙÓ¨=ˆo(%ÊDì%¾Ï£•<¢Ršª$!±l$vQ“"âØ˜S©¶ëUor"çük1Mr‚ŸMk²…Æ"¾ø¸Â=“ù/“^3×>õ“EõÑè¢f¾+š˜ìËøA:/™yÄÈ›iŠótÓçmy}š¸z#‡Kª*Íu6጑'tð¾©ÇݤNY™±…(º,QÈV!}» P<¸°y&†Û¥½\ÝnxVÏfóË´ÒÉë|²gÅÅà5W]â÷Ú™ž%Q¥ë ˆfO›iw™€¸ƒíº¯\¡b'iÏÑì_†ñ°=K²ØN8(O²8¢ˆM:†m(¥‹ Broœ¥[Äü3ü‰‰ eá´…K#L´ri4ôo‡;ÇwþJî÷ endstream endobj 155 0 obj << /Type /Page /Contents 156 0 R /Resources 154 0 R /MediaBox [0 0 612 792] /Parent 150 0 R >> endobj 154 0 obj << /Font << /F83 134 0 R /F47 7 0 R /F8 9 0 R /F61 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 159 0 obj << /Length 1303 /Filter /FlateDecode >> stream xÚ…VÉnÛH½û+ä °ÚìfsËMdžq0ˆ5 ’h‰¶k‹HZ1òóS¯ºZ¤,k|ºØK­¯–£“³‹¬—«<1IotßËS•&Y/‰Se¢¸7šô¾ßú:ˆô6σsEke½¡³b¼Äó0í›,¨‰^-qjƒïaÒ§ÎsC¤&R¹ƒ¯DNé7£ß¸oRúàýо›…c½º—½ŸØ„„MIÿ•Û-6D—޾ëˆÖî-øŠ…¬ o‘‚wÅ̠󸜸ûÅ’7ŸÁbÖ !~Œ®{:V63½Ö*Å)Oý8!%f-+g]™`»Ý*úZ5³ÒAÍ”bcÉ;‹³ªâo0ì¸SLÇ剸ü>³Q𹄊ÈÙ`§ãÞ4í¸—†Xâ€]t‰ ð3¼0gmf>NÖ:WÓ ³%;îv›‹»—ÏÅ7Ä\„oÁ§æpò퉰®ÆÅv:I; É賋Dw@™LEyÔ ñ×p~|éÇ&X=S`øó¶.XÒÀ&ap»êÇp&Ÿ Ë.Œ‘6*×z?fÚ†ˆ‡Í~§ÉkÎOZç§âüÛá-Youpµtðræ`mj‡@cMpp]ð-Î1a˜‰Ý§t@ºŽ¦í#ŽL³qük@óyMI 0ÇDkÙ•¾(î$N £°Yz o¡ƒ3üúYD£¿‘{š•ÆÑW ®÷­øKI½™GÞéÑ„4&ܬÅA寥Ýþ¤gÞÙ’?_d…Â%'ú|'ûtR£â,õÈù“ó¾p€ÞˆRŠÁâð[b%Ñ*NsÏâÔ‰FšÍ%žá/9øŒ-è(øûœ[rJ©¨½±‘U:I­Í;ÖêÏAÿϾ¼d¨òpǽ¡™¾VOunhðË|5|Q=y<€2Íþn6xç' Ì¹Œ]S|èâ0> ôÖ\Ç&Œ'aTOKGüa¿}³B©%Û˜P¥Ú¶5›kADëM÷k´\ûò¸|àþ†pü´±/y|Ø©î“ÝàífÚJ/Ð~Mîµä.:k¶³[dÛÙ ·GÈýÒѯŒ¥Øæ¡€Öv¢ó(\wæ2ööŒ•ç"µÜÁ®0îsøn<Ö~6\ìrbSÀ,½›¼7ÂB*ÈÚÏ:DODpTúö%åáè8Áz³Îeuln´ÊÚÎÔˆ~?웘ÃôndŒâ4VRîdt<U)Æ/æ“së߉LŒi®Ò4À™¨LwÙ†8;ù4:ù€I endstream endobj 158 0 obj << /Type /Page /Contents 159 0 R /Resources 157 0 R /MediaBox [0 0 612 792] /Parent 150 0 R >> endobj 157 0 obj << /Font << /F8 9 0 R /F61 14 0 R >> /ProcSet [ /PDF /Text ] >> endobj 160 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 161 0 obj [441] endobj 162 0 obj [1055.6 944.5 472.2 833.3 833.3 833.3 833.3 833.3 1444.5] endobj 163 0 obj [557.1 557.3 668.8 404.2 472.7 607.3 361.3 1013.7] endobj 164 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6] endobj 165 0 obj [892.9 339.3 892.9 585.3 892.9 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 585.3 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 1138.9 892.9 892.9 1138.9 1138.9 585.3 585.3 1138.9 1138.9 1138.9 892.9 1138.9 1138.9 708.3 708.3 1138.9 1138.9 1138.9 892.9 329.4] endobj 166 0 obj << /Length 104 /Filter /FlateDecode >> stream xÚ31Ö3µT0P04W0#S#…C®B. ‚‘)T&9—ËÉ“K?\Á’Kß(Ì¥ïé«PRTšÊ¥ïà¬`È¥ï¢m¨`Ëåé¢`ÇP„ÿþ7Ô3`‡v(P†ËÕ“+ L5* endstream endobj 36 0 obj << /Type /Font /Subtype /Type3 /Name /F74 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 17 27 27 52 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 39 /LastChar 39 /Widths 167 0 R /Encoding 168 0 R /CharProcs 169 0 R >> endobj 167 0 obj [43.59 ] endobj 168 0 obj << /Type /Encoding /Differences [39/a39] >> endobj 169 0 obj << /a39 166 0 R >> endobj 170 0 obj [590.3 590.3 885.4 885.4 295.1 324.7 531.3 531.3 531.3 531.3 531.3 795.8 472.2 531.3 767.4 826.4 531.3 958.7 1076.8 826.4 295.1 295.1 531.3 885.4 531.3 885.4 826.4 295.1 413.2 413.2 531.3 826.4 295.1 354.2 295.1 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 295.1 295.1 295.1 826.4 501.7 501.7 826.4 795.8 752.1 767.4 811.1 722.6 693.1 833.5 795.8 382.6 545.5 825.4 663.6 972.9 795.8 826.4 722.6 826.4 781.6 590.3 767.4 795.8 795.8 1091 795.8 795.8 649.3 295.1 531.3 295.1 531.3 295.1 295.1 531.3 590.3 472.2 590.3 472.2 324.7 531.3 590.3 295.1 324.7 560.8 295.1 885.4 590.3 531.3 590.3 560.8 414.1 419.1 413.2 590.3 560.8 767.4 560.8 560.8] endobj 171 0 obj [611.1 611.1] endobj 172 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 173 0 obj [877 323.4 384.9 323.4 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 323.4 323.4 323.4 877] endobj 174 0 obj << /Length 149 /Filter /FlateDecode >> stream xÚ31Ô35R0P0Bc3cs…C®B.c46K$çr9yré‡+pé{E¹ô=}JŠJS¹ôœ ¹ô]¢  b¹<]ä00üÿÃÀøÿûÿÿ üÿÿÿÿÿýÿÿ@¸þÿÿ0üÿÿÿ?Ä`d=0s@f‚ÌÙ² d'Èn.WO®@.Æsud endstream endobj 16 0 obj << /Type /Font /Subtype /Type3 /Name /F60 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 5 5 36 37 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 136 /LastChar 136 /Widths 175 0 R /Encoding 176 0 R /CharProcs 177 0 R >> endobj 175 0 obj [41.52 ] endobj 176 0 obj << /Type /Encoding /Differences [136/a136] >> endobj 177 0 obj << /a136 174 0 R >> endobj 178 0 obj [639.7 565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3] endobj 179 0 obj [562.2 587.8 881.7 894.4 306.7 332.2 511.1 511.1 511.1 511.1 511.1 831.3 460 536.7 715.6 715.6 511.1 882.8 985 766.7 255.6 306.7 514.4 817.8 769.1 817.8 766.7 306.7 408.9 408.9 511.1 766.7 306.7 357.8 306.7 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 306.7 306.7 306.7 766.7 511.1 511.1 766.7 743.3 703.9 715.6 755 678.3 652.8 773.6 743.3 385.6 525 768.9 627.2 896.7 743.3 766.7 678.3 766.7 729.4 562.2 715.6 743.3 743.3 998.9 743.3 743.3 613.3 306.7 514.4 306.7 511.1 306.7 306.7 511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6] endobj 180 0 obj [583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000] endobj 181 0 obj [575 575 575 575 575 575 575 575 575 319.4 319.4 350 894.4 543.1 543.1 894.4 869.4 818.1 830.6 881.9 755.6 723.6 904.2 900 436.1 594.4 901.4 691.7 1091.7 900 863.9 786.1 863.9 862.5 638.9 800 884.7 869.4 1188.9 869.4 869.4 702.8 319.4 602.8 319.4 575 319.4 319.4 559 638.9 511.1 638.9 527.1 351.4 575 638.9 319.4 351.4 606.9 319.4 958.3 638.9 575 638.9 606.9 473.6 453.6 447.2 638.9 606.9 830.6 606.9 606.9] endobj 182 0 obj [656.2 625 625 937.5 937.5 312.5 343.7 562.5 562.5 562.5 562.5 562.5 849.5 500 574.1 812.5 875 562.5 1018.5 1143.5 875 312.5 342.6 581 937.5 562.5 937.5 875 312.5 437.5 437.5 562.5 875 312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7 593.7] endobj 183 0 obj [700 738.4 663.4 638.4 756.7 726.9 376.9 513.4 751.9 613.4 876.9 726.9 750 663.4 750 713.4 550 700 726.9 726.9 976.9 726.9 726.9 600 300 500 300 500 300 300 500 450 450 500 450 300 450 500 300 300 450 250 800 550 500 500 450 412.5 400 325 525 450 650 450 475] endobj 184 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 185 0 obj [458.6 772.1 458.6 772.1 719.8 249.6 354.1 354.1 458.6 719.8 249.6 301.9 249.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 249.6 249.6 249.6 719.8 432.5 432.5 719.8 693.3 654.3 667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7 667.6 484.7 484.7] endobj 186 0 obj << /Length1 1924 /Length2 14347 /Length3 0 /Length 15534 /Filter /FlateDecode >> stream xÚöp%èÖ ‡ÛÞaǶmÛÞ±m››Ž­Žm£c›t’Ž“›™3gfÎ÷ÿU÷Vªvö³Ö³¼Ö[›œXQ…NÈÔÞ(noçBÇDÏÈ ‘Ödb02²Ð322Ã’“«ZºØÿ+‡%W:9[ÚÛqÿ‹!â4rù‰¹|åìíÒ®6&;77##€™‘‘ë¿D{'n€¨‘›¥)@Ž mot†%±wðt²4·pùˆó߯J*íŸæ![ “¥‰‘@ÎÈÅhûÑÄÈ bob tñü”¼..Ü îîîôF¶ÎôöNæüT´wK €2Ðèä4üQ2@ÞÈøWiô°äU Kçÿ(TìÍ\Üœ€€¥ ÐÎùÃÄÕÎèøˆP‘’(8íþC–ýðWsLôL»ûËúG–v™˜ØÛ:ÙyZÚ™Ì,m€qYzZ€‘éD#gû{#7#K#㟩ą”FþUŸ³‰“¥ƒ‹3½³¥Í52üáæ£Íbv¦"ö¶¶@;gØ?òµtš|ôÝ“á¯áZÛÙ»Ûyÿ™YÚ™šýQ†©«ƒš¥£+PJô/·ö™9ÐÀÆÈÈÈÁÎ :€& Põtþ©dúCüQƒ¯·ƒ½Àì£  ¯¥ð㬷³‘àâä ôõþ·â,ÀÔÒÄ` 4·´ƒýÇû‡höü1'K€ãÇú1ÿøûû›Þdž™ÚÛÙxþCÿsÄ ²Jê r*4•ü·RXØÞàMÇÆ cfc01±³8>¾øþ¯Ÿ¿;ðßêÿ”*Yþ•Ý¿”ÿ¬¿.#£ÉÇÓÿç#øÓäÿßîÿáåÿuýÿoFâ®66ê)ÿCøÿÑÙZÚxþÅøØgW—Û³ÿ¸»ÿKÕþç …ímLÿ¯NÊÅèãB„ìÌmþn£¥³¸¥ÐTÑÒÅÄâ?Kôß)|8·±´*Ú;[þñàè˜ÿîãæL¬?çYý©~œÔÿ†³3±7ýãö˜?ænäädä û1úÄðfú8RS ÇŸ» ` ·³wù0|ç 0³w‚ýc¢ìl¡?DÿAìá€AäÄ `ýq°¤þAv²#NFƒâ?胩òâ0hý¸>"ýƒ>t&£?ºÁ`ú/È`þ Ù>¼šYºýKÀü!°wuú—ÁÅü_À`ñ/øQ¸å¿àGu6ÿ‚©Øþ?®ŒÁîŸPYÛ}Lç_úÜìÿ†¬Æöÿ£þÈÍáŸF~äõq€ÎýÎGzÿJžé#=ç"~(?ž‰Ô wùGýÞÅ ø¯n|$ìânÿ/ƒœ]ÿ?Êuû2ÐÿåùßçŸðvÍÄÕÉéãþó5øXÄÿâ?} Ðh»²hoÂbUÒñP+„çNw0Å7O~ ‘NEç½âÔéú„•BU“´åô[(e´y}OŒòNp•èÕû¬µ*¼-I©ýÙçÅ Ayö vyshºèL¨aŸNUðÐçÕÑG=м´[š<ßÑ•Q±íÁ}@£a°|m"lñ@é°†]î¥|Ž.V-F7°äyqÎ6É':hjÔ+¤w¿çQó¦ß‰¤h`}ÏcY¾yko3Ç=.xmTª2;÷àáhc€ß¡NÌRx §Jc-y—Çlñ-årDƬ† $.Ȩãø)b;…”ÑñeÈv£‘õä¹›d9Ù¨›LË(>5@{‡’—J˜%U“Ž–äà[ú?¾’¨$j‚šÒtööœ_¨uSû$Æcs3ÅMÙþæCÛíÊñcyüÄPa4+! ™b²Iº*¯€97©Rà+¦<êÛXä­3ø^@_|OgnîSìb«»É=çÁ‰¥8õl¬Øxœtùóõ[÷‹Q¦Zô;Œàdƒ.®ˆ½æ7–¤fŽëZ)f,aÄŸ9œ#Òé[Œ è¶Îøj`3JO¦šçÄø³ƒˆó¯9]?ËÖ”.PÏY•£%cáôæ£è3ü f&ñ²÷9/r¹íO¸¿«áº‰&iëØfè3Œ5ͯú/²¦A7ÙܲŽ$.`¸ÈòúQid†VºËq™£*B÷:Zì(dNåcöX¹UÏR…^ê9×9*“¼ÎÔ ‡’¾|^ô|Oy|…Ô–ñW”¼JpÏ⽓.¬ÎäÁÛž¥—÷;+ PŸ¯u„–•¦1˜…¶šoÜ…uyÓr“Y)$êêºBgÈ^)9Æñª'J -†âž>ru@l~¦8ýÎ%AªÞþíQÐÍ:jÔ¥…^ ßÇ«¢[»Ð2‹'W¾^A†ˆó ‡ä z¹LRh•£kQ€Ÿ¤"÷k–Ëyê¨;õìA„ \7œ¸8‚°ÆþS¨ !ñÛ X3uÓ9Þ§“†x1œÜš_ÚÙ¨UàBµ„HÞ è¸W€áîN˜ÀPh–àÌ–åumÖÂ;›JÅ‹Ó>c¶P;#Ìͪšrý6Ú.ŠF ]Õ‰›0J K5¯gj1C‚Ò=Nb<áøIh¤<­B™ù"‹e…B:Nlë¾ÉÜNÚn7î´Cæý>«ŠúS*9IoÆ»B³Äðý‰u^"Å­¹rÍGñYøÔ|½ÑýÆØºG¼6.P¾²ÄÌŠ'ÎT€ô‹ ûyAŸ9Øa¼÷ã õãæÅÏ4œY84Z¥}DæâöÂçJŠÏ¥vU8º†õzJÑ!ÁÜLj»­}ì68vŒÆ\a¶zW×´g «¥ßtµÈØquhÕÄ¢H‡Zbä Î)¦ëéE_vdkv˱uï¨w¼X>—S̤°‡\1§­Šê&‡¤¡L¹Í´0¬ÇÝzc¨K±¦DMŒ‹·NGv¨At1ü­"Üÿ޽»P§+èwª…fð»l<±ñ úÍ„¼ÉxVÚmÉÉ7}˜ËqÑýpfÛ{/Ÿ= µzÊy* E.$æB‰®ß!èès$ET†E=G#J+Jª‹öÖ3„®¿[Š\I ICuW R7Áùê)omÌTŠh\Lm»ì™MrÂùXWg÷8èï³¼Ñƾ{ RðŠ-WÁUßã’Å`ØQS©1jZ€U÷¡Þá }¯‹G+4FÌe¢µX¯–MaM±¯i8=Pî¨aøNäK ”1Q™Ìf“(‚•]‰'ºf8üºh³ÏEGADÿ5ÿ±eNGßÊ92:šGmÌ™q«kŸ^qØl]â€6ãR½Õ9Ç«'åå@ÈÁ¿A‚p¾pˆ×Ħlèrº™ÐKEDò»)xll);DNê„Ë`‘.غތfþˆùð±NëD=-ªE§•;HÎÏ—mü[Û’yù토qPþŸš©Pœ`Ûv›çâò$FkÀ¾‚kûe‰¥ÁF«ÑC ˆE‹´ï°äAú.ZíRJvÅÐ*‰E^t<&&=ªíúDI›HU-Ûè‘o-4–®-¨ƒd WGs W^ñÞ}OÌy¦ºê‘*“‹:cø%k*{Ïž/ËÖlí¸&˜2žèÂïòŸ1ÊtÁ@+"ôùXqÀA³ Êmxîz‡Òîh_}<|RFW¶ä¿ÙŸ9ûvy¾æ¡.¾ +da]gŸÞ\ ¹œÍ Aô ,âò’úS÷A¤¼®’[ +˜Wû{$ëÒ½m ÀŽ“Š@4çtËXox%ê÷rRÒüànùq“²ÈP9-ºõµ{͢’qnüŽ[U½Ö»£æÁ@òòK´~¸NÍFxxw!-Žà…5Fþ:°ßóÙÊÏßÍÌ+aQJš¡ËüÌè9»õ³ ;/ŽôNZÀ½µÇ¡ûŠêrÅV½ŽÃ›ÌG×?Õ³ÜUŠÔ·è££œ¹SÇÏ(‹RžãôýÞÙ¢Ú“xÏZ}\e¤š#\ îAÕbÄH5é}Ãþl4ð>$޲^ÃÔYMêïRïi>YÎg23ÁTúAW«\±BÂ3é;«)%ò–|ü~‹7 ¦Í£n@Ïz¤šS6ãôòJè’oúÁé ¾¯ïÿ›úÀÖãŽx_‰ÿ‹Owh¬–øñ-¿_¸9ט3`¦uVa1C®¬¶³1¦ u¹dÕÖ!n;=šjØé² >ßoƒé&®o)>Š—Çß”¶FžýÛY­bÃT©ÏUµu‡2gGÓ-!¿%\ħˆæ: @¿;¤!;G~®_°½Öõ8ǰFGX<¼k“V–ü‡­~vÇ1Z£00 ƒš¬–ÈÎÞc‡E€|×w£NÖ¥S¨Óóšn¦ÆƒÃú[Ÿ‹£ôɪõ&ñ)¤!!ê“y{8‘yÚ–ýæ:19.zO‡Eä·²öuÊ9æ»ãÖ¥¼†îºàÂZÀNãvF§¶ÎLDâ”®@œ†Ë8ïH(Ok,|É¡2~•íF_u^Uõï2ô,±8]§ &øÞùÙDûäo3h ž{‹‘+TÂQWº]‚«æ'1ÓÒ¶1GÇ•ä[NVýÅÙ±]¤îYѼf|£.ø½¢–žïªÑ’,Ô mâFC:§U›•xêËvGР,ìÜ}ü•¶å¨xYPó9Xn;ci˜ˆ'Ù¶ÇxÜäa§#+6ý.ðryeõi¤£* Æö¯/võäç;ÓµhZC)®#.‹þ©„kX¬C-Ñ{¼Ž$.º<B¥z/_¿BÖßá™aº{zúH¢xŸ†©ùaòÚtO½É]@¦Ãƒt \8¯-µ.o#Ìv Uç°c皀“ û5}K‹W8„D‚ÍE&„6ZQaÊü5 ö‰Ã¹"8ë[’`Ÿœ†z6º¡Mn öƒ¹ú–è·«<”hØA‹÷jEL³2R)c‹ Ä†™=&\n5þ"vm¬JuTƒ”OÖÀßä~²áöܰãC†EÏzšD-8ï,°kù1íÒh<¸«ëË…(TÑ¿­ÍûèFS>ö©FVd¿=eЦ”`,=ÿÑ3šÞCdne("¨[­4ÅwEWê7‰(gìi KÚ>§g‡ûpMT·l E™õ ¥—>õœ€-VSÞaí©³ÔŠË°šs)>zw`UU”B6hÂÇ`á~´2?ÑÍèt[å ‡C|š|O^XØðÂźÈF¬ï©)@7BRah’D<ãô§&rm¸Ïºçô¥«ßŠE…u·«:Ô|Q>(Æ ªŒFÛTçui² Zè3ʂ֋þrÝOùJßt_¸…$ú› Ë £¢¹<>P÷àü \¿‰lmpÖJÀÅEÓ¥<ò$¿Ï&â(l‰_D¶ C¾äñt aܹ°íI3Ç âqÈñ=fÏq`9ÎØ5áÍ 7Ï–»3³n"/…ö˜2_ˆúéPÕ+¢UÑó· Šò ŸôgRX¨eîöO^ׇ¾Sù΃WKŸÈ™“Ì‘±=q^t™Wˆ83‹4qÐÊää!-ä @OµOMSå¦D°nÂ?!®3‘‚öÙ·U¤¶~ÇÂ"Ò4ÉQ¼_ö‹üÎøzmô+h_ÿÛ ë+Æ(;Á|2•Bñª`ÀAßÞó5V?¿bŽ‚ m+-òúh¸‚¼°›—×}·”ÏZ y |íÍJ°RÑßú7¦?ëWñ·ó0g_jÍ×þe™b«’Œ=¿ NÞÁ½/½é,Í_­Z˜sƒÈS@¹8䧮ȼü¾è÷õFñÛÊöËÊïqÝ{Fž5ôâ„CƒŒ–¨>M©½P‰Ò L.‚,ruäÔ˜±*ÿf.¿Œ›Õlm@€v'%D2Pöƒ֞p¥rÜÙö<¡ýF¼ƒf±ØS9ßÊ:RÄØFLôøDÅ =>D¾˜ÖwðEI÷‡ýÀ„ÙL´ß¢YŸcvçwtéŠJQbíµ®Q¼Ñ9“ÞûSÊÞBMìá/Øô9ÇD3’'­,’˜à ¡eÒñw}àŠ‰)¤ã„ñÊ4-&j;ªª; ”¹TDÍ/b2ú¦kä=$¿ÞT¢éeËó(|8ÃÝ$^q–³žrɶÕ#gå º–ßllçØ™u¥Âü·ÓbÚÖA&»Ï‹H"X¨`óá@å9Ž¥9ˆ ~ì0q”,Ê4zQ‹ñ`o=ŽMb£Ú“;¢* PcbyöbE”kj¤É©¯ÔúŠ”M>MŽç˜w—™QÐŽè./^Þlóê2 œ¬5jbT °'ÚO_ôïH¼mä¿,Ú^%ü®k?—·!¦œ_ÉI½T~þ!ºÐ‘«Gz¿¯xcFõùÜí…5jº¢WLZÝJ‡«ÛöXĹÓD'0âXàùä9O£‡þÜEðº|púDÆIe‚‡Fñ)iVÃ\â§y¨ä¥±†Ñá6au JÿmË},}© !^Á–ã½,¾Zöß5h'æ (1dªceg¨Œòβlw+ S' ²D7®Qtq ì‹ænO`°²‡ößÄjZ† ý¤¨eˆY4¯§Ü;Qá@d•îP½ï{Ó]¦€Éôx¿M†té K„ò„Ò{?ƒ°È!¢ùlÔïm‹Ênëœ1Ðô9e˜ÒŸ=©Í’/JW9*OUð¼–è:V#É&¯n2ß`pFÖª<45‡HÁ¤ìC9¢áâz~¯£,ìÀ•J›ìrÒ¨v·¢¤;•lAœ8ëQ¤Õ¬íêe[ßüìSi”½ô<ß4x³zOW^jæ½Üüšh]€vbø^Ñ‹•eÕ»îI°·)ó•ãKŸ=:›Š$©L–üR%öá­É |>ä/Þ,ÓV"dˆþú­Ý!¡•Ï'4øœ‚žžJ*g®c(—ÉÓš^o¹¶'+,K·‘–2ıUήÑ3‰¦Ñ +|NØ"uš76Á)êþ¦žklÒH_f­×ÏÏfàTÚÚ-T6dZ÷c©Z0 –žÊq#SšlZŒwœ¯ðjß¼Aϸ¥¼ )*o ?àëA Óªµ6ÍKÀù¹Êîg&ÈëN“fÄÁÜ_—0Ü·í‰VyZS'‚BO20™æq}ò Ža§i¬ õœÈTñªß&aM@zP7JÓ•UZ§eÆ&¹Ï¬Y¿Þ ¯«@§ú®ð°ð¦®Ž5ñž¦TÔôÁ¦¢ãín‘ßÍHØ…¢a ¯ÑOAõðwi,R«+ÄÊU¹¬.Òª\™@è5å3†FUŸÒIŒi.ǸðÕ¦,ø…§ˆÆÒÊo¹mX~këþì ¾\Y¿Ý­—ä+Ëíh¿#½ôsq»r¹Ãº¿¡.",X&{ôì1ÉälPV‰C©Ë<¶§½ß¥M?¿+!p¿±‹x Ä͉itmnæHrðé4ºÀ‘uk-•ÌõÏg |eÂÚa )"e-Ž«·öõ9ùÓšڳÈËŒ=FÁ©†Õ yÕa[¬˜¢µ‹@£vé O¿VNœW‡¦B„6¶ËZœw2*çWySÞÒ¸š™íµ ¹”œ3„ÆbÝJg’¨Ÿ]èûÌØpÏ úÕÐ'6YÊqøŒ¾ê·ùû!c‹ìé½Þì—1Ïa‘ØAƒŠ”îDì‰Ksξö¿NŒ2{ßvë˜ £0Ç` ØjŽF¶‚¤ îxwWK­†¢ŠŽÀŒ•2Jäd÷Dì+|­“‡ãn1^h nÇaYÏ-Î5 äÌåæ6ôðT×&ÃËKxþ±e.þ£ùh=+èÚXRI§ }¦Œðš¸Þ>`# Ž„’Õö ƒgíæ€_ bM,(Tã˜âän†0׬*ׇ—7õ÷+µe¨íÔø1Y&z¿·ê ìbNµ.q02ݤ­_%§00gXÕáÙD÷ûg‚Hó=ÔÂ…ªßDf3ÄmÙx˜ä‘e€¦ôî3ÁLû1Þù볕‚‹Úȉ¸_|uú@S½W’¼Dð›zŠw3ª&Ç΃þäÁ7jœ‘ó¸îRÔíäOjp-e þ>[rc'!17*]5KJ'CeãxV[‘ž}ÃN 5¢ç –E9H™Ú&å²Dd^@ñ•§î>´šß iª`NU»8+ÄøÒ¹HòìÔ ÏW»}Û®Uš U::¸zb(%s“ÒÖßLxÝa9êÊhü¼Ó±ÏVR>™Ž9ʆ*^0ÿÈZµžd~µ­Ë!Ž×J!&iýUÃU¨üòéiË0¶™û-¦ÿ塪àTòû¹Q/-hoèhý^Ù’?á÷/pÊŸyš%*3ÒPR(t²<ѤõPÑ’ù¹d&_#,nÖ§7…G[ë3—®ØYžò0}çÀ Hn¥ÊÞ9êdž³ÛK¼ÅªNŒ÷·DÝ#Ù(Ú4{'nÍaK+F#VҨœ¨ðpL¶áoíÛòDò¤H1ëÎδS=܃°8u¿ Úc€(7óÐHÇžÿ>ßü[%Á\¦8Nƒ ,ܳOµ[ÉLè2­LM?¡¥§±–ä ~™ñK¬—„ÆRÞ÷IˆžÖln^ ã1Ö–µùËšêÑ„Ý*›úQYámÚ¯Eô½¯×„#-nAù.? +-æÆIÕ\› ÙÙ»„1”ÄöjȈ¤žÓnxxJಬ~äs•qkðŸwaÑ(½h0õE˜°~ÁÚÿ"_ÀLËåîò â¹2y>Ü‘¹Pz•ÿ-Ò•Ûã«Ð°þ€h)Îý³¶âÐj©ói‰Çsj MÛŽÁ!ílã9ÙG.†:òŠwÑtM˜)$^~%ÓJŒfÕg7'öž¶Âl«=&¦³S}Qcʱ¦}— ì9$W\«Ñ:F¦–ûËPÈ{™·s÷R‘ ²æ6Ï™ëjù¯ÑÁäèÆ`) Pñ;1ÈAÚàÄ É§i^ÁŠB¢ž'õ¹xeÃõ,ï$×I¬©/eÒ(PfÁÔàžïR>¯!õ*šã0:mZžÚ*”Ó¥éOä+ZUF…[šf]ÖÚ¤¿Ø|3ȼJu:¶ï."cPn×­w¢ŸÆ¢ÉpÖ¯¦­3.î·Þôžªuë^O£’áBA©åÈ—S+ar¹¦ƒËå1ÍA‰”¿åIsÅzÄQJK#I3Aq¥ã³½.º~½3›¢Ã8cJh» 2€›5N‡'å{ XHÈÄøfNüꯥ¹ºÐül¤ó›dü…ÿ²ÐJ¿(S$]EAùý·K7¨B_Ä›»£~B}|úÞOÛG=JågöªlTNw+P'Õik˜Î´×tëë_ØÂï® —lm­ùM!´z=Žo¥HâÂ@}_¤M}V¼½2^z¶o”¿_éÅsßâ„€aM÷˜Šíû„ï,·Íuˆ¢u1ÖÊüEv„îŠÁ÷åîÆ\¼mC’|´>»§…‡½Ï<_3å» f ܸöì¶”mæd…±†¦*Õ IÖÐÕ×÷h$¥„¾aâ±dRjì<È%^äþ]`F žAÕårŦtÒíE¼1ÎLÝŠÍS O¤KVvw1;3¿xäá>?-"#Ã|šgòV’ßMh8‚6„ñQv”ʱÃþ|ÞšâH´˜ê}ɦ¦@ÏÑ;&ˆJVÆ\YŸ­¶žé*¨}˜AØK†HžÁÁØøNV¤à×#ïHF$1ã$åBn6íÆØˆÌÐ/O§ Ý ¿G4LÜûPÙ\ï…¤»Z7o” ·ÿ:Z+µE‹?ošÎUZëÓ3ŽÃqJ 1SñR.“m+ܾŸÍîðc‰={Ø„â³Ð‰ñ\‡r¹‹ùëX€óDÜ1*RøfqÓpèò'ËU”¼›­ M°è¨ÆÀ7Q22#xóžQŸ†£`˜Èó9ˆ¤“̺*pâ·â ]ûñã™3YxÆ#Æ•`ð¨LýømBW ý6ûUëªK½‡¼À#í¿úñ<³{…ªv˜i¹Æ-ëY‘T~Ørí|ñ`RH:ÔPÍ|*ÊÊßðñå5«t\!Ž–¢¹ú–ÂÞ<(—\]>l_EåLyp'óúU“jÐqŠý%˜TZ‚sçu9qKæ6fNž+ÌœûOø–x,M*|¤#Üg¯Ádú¾Àƒ{QÚfÄçlf~£â˜øÄ °/Cœ!TÉï§ ð.p%ínT3;uA$pê¤ãŠÞëß#’úœÂ¡(<'9-´ŸC‘4/utÛÌ—ª€5±õy@³¿Sw–¹ïëÿØ:êË–$I~tœŸUMà Ř8\‘Ã:dûžü.|ùÂW¹WÆ\€P²2¯E–owó;èבp„~°dѼ»û% ˆº±àÚ$¬4×nédržK®s²±Vmá™oú-NÞÞ‚°ÕÔ4ÈŒ£€Wî\ŠŒWÎÝ´Ùþ‚r²°æ ¶<¥0áóÕoCÄ'B_ðV¼¢1Ã~dC_ÔÛˆg—¯ÍÞ©Ý0ôk‚†g¿Ϲú{OÐùÓcñ{BxƒQh¿bó«¤Òl¯].§’Dq?”™P`ÛÓ1:`d£E“ÁÌ—ä&Û©@0°Ž3¨PÃE·­&”ôö„Gº…ûñÕ"ª§YD¦ÿh±ÈRçQ,s±î9ÓŠ×Íö oÈDÖv죽ñ—Ó–Þ¥Mìíte°J—Î,åŒänq8žD<ðYzÅ´`&x„^2ùœzN®ZÄÔ¯þsZ îHˆš¾ûñÜI£R}bnظkÚóW:ñw¾»ÙZ$^¡ÃÏLð鿟o7.üf¦,‰[Öë¹J7î+,yµômŠýo†¾Ú7¨“[ß ‹¸í+LÄG^¯ Z {,Ãv@áÜõ§.n$c*0áœi-‘꼂Ñ?CAauÙ²}ü2ë%—´øÚVˆ¯ýê¾ì™&[ÙÕì)†~Sj ))pðû1÷£7àIjÞ\yF¯ÏûR2\l‰0O„ž^S,e¸3oEóБè=÷ Âð‘y¢óŽª+6õPÜO¡èÇ0º¨y²ùkÖ|:T_ԾƇI<^–hKB?åÏ´èŸ÷™ú»~óaYŸ.vêm´«²ŠÓ'M˜ÒwJˆÄyëì Gý Š;ã(žu°=ôÞ»§©4…)ÈÎÌT{ô«8þm’qˆÝ=gû&\/Ryi #Ú”J9  ¹ûyFœžI7.Ì/ug»7†ÛZÒZJ³ Ù~>/ Ð”â¾[ózP4=àáµÚÄmû¹ªÂ0^!}?µÌ4š{p§Yë›Ã„žXü`Q‰³Åë,GÅ¿cÔ^"§û}fE ²êó÷-q:<7¿Ë›aD%Fí§*á=WRW‚ÖŸ%Ib‹fÀÚŽRM‹HX(NŽÝäD®5ÀàïyÑìøªj|Ïzÿ«n¡bÕ U:¥ëÉÛN«z ÁÖ-/Æ›¡©¹Á$^²’ŒéYü}ð×´äD~çh,¤cÕèÚÎé*¦kìÈfm(Nw:%>™j·íÃr|¬ÐK·”üaÏ”£y?(þ5–ªy˜v§r›l>¿3æñ†i§YGæk›À)[E•H“±¨‡°ÜüFÄìè…Πã#Ç–V‚ThτȀxÞY__ )ýt¿eipe÷wÖ팷 y&ºb;ÕnÁ°ÏÖ@·Ú¡§§^ ïHîÔ§Ì Ïæ÷Gz:ÜïïaËf9NfÚÄõï™î©_wv›¹zlåÜÐ 7ðܤ‘,±3ïA›(^xÕ¸ñ:gN&;»b(B)L*Â<6Ô»¨cH«¬Ýre fÐCö‡¹·Óô‘óNóÈé]¨Š«]ƒÏ [³@ÉŸýy˜ƒödž_ÒMS‘½Ï9f¸í#ÃΟÛf¿~†eAÓ¾gKÝn¤)%Ç.‰hWdøÁ®*À7ß97Æv.£ÕCéB™è  E˃˜Þ­Æ¬¼Ü‘4¤{J¾Q¿µGŒ/âÏG—Zmšv gáZ“†• Þ©ü¼Sל7÷º•FÿÆë_p«óieTY\éÅÈIùŽWAIWöùûz\ÞV—÷úL¿f YŸ…rÄFSØÌòy·£PÀæ~5-%Í?•&ŽC!–½ŽsHvÝ`)æÝbÉï9´ÖMJ÷kWöª‰ÎDU…IP:Ùí•R÷Ñá¾Û¯¬–ï¸0cHZ@P0Ps1ڹο|¸#Aœ< ÁbR)*’@Ý¥~ü#Êxþ¡"»«Ri¡RèϳgÏÞj[9>[•ªR$“ ¸×Û{‘S}Oo>H Ûñe4&i½.ªnäVvÙŠtƒ¶í[Ìš5j”ÆÝé+<˜­³TpÒ«3§Soê±S.šåfkŸl4ÖÈúi¢õüÖ+…´§¥«çÇp๹Œ÷£ГŠï‘qïý½mä*`Ð0¼rʦäû“ÔúáW=i-Þ«#îFWË1k2‹Øå…â|*éÕÌóE”õå!v¿nH”ŽN•½Ž¾oEÆàQ¸SX6OŠÖq4»<)£FU½^ÁU¢¶/£bä÷šÎσ6òrAä©û? ª¤ÇÅ*ï-Fvìñ[¿äÕyÏ›ÆCÀ$ÈDåR1Ó\¶òÄ#ÓIjÕÎuÔÆ;£yø˜Å" :á›­‹ræÃb3!n}¹–ÆDë fã„( ƒQ£Ùc¡Uû#q@p+¹ÓZeön~©Íq±ÇH Ýa[°3‹[«Rl÷P “Ù)Oõ#JsÍ 2ÖÍÅ0 ÕeV¨[ËNš¡ÑŽ#Ž}¼9§aLC!!î Û˜f5ª',~¶X÷ŒYת:ßUÈo–{ç¿Ûq  r™J2Rì¿N}ó˜¢ôÖNfA~éEça\CpF øœÛ-8ʃ/_sçŸCƒ-¥ë¤wAëNB¿RÆ÷ãp­êõÍ«0ÅEþ1å¾Dµ"‘4ä{w¾®„˜Rç[ò%díJ€ïÏ­_þÉÒDœ×v¸ÛoQ¿`“{Û­&Q×µàFF‡Ð…k¨Í ‹ŠÙW“Pò*Ož$QÃÂs´¸¾ ¥µCocü0=*œ^cºÂíÈžÙa@–¨£¯Bv{4n¸–)ÈÌHÙ©\ѯÛ7!:ômÄø~­ÉQ=³|SÚ!3¸¼³a·½öI&è 5˜-MUë1´úJt}îÑc°—è=<‚C°ßs;^÷^ëtϯ-s•½Ñ“”âFOû5oT/£¤[,ˆ•¢×•GbwãN1ØçWRê ïÊý ¨Z2N׊e˜t §Æky„==ù­öú¯z·šÚb…¤zs—êÉÍ€'¦-ù Â}± ¡GcLYjáì}<§:ü÷AêÔ=àN õ uF…Dœ«'YZÑËSf IÔþ^ßw¨;1BÉßYaQ6S}ÃV?Ý'}rû}d UW5ü9¶‹-Y"…öòKNJk¹b¦KH Je‹ú¢Bû·v2ï<÷-•×·÷Ým·ÔÒ>7Á…øM×e¦bCϾ”învEÍ9)ü»õöʬ)õ zËÎüx;Š(‰ïoúgG̹ó+”{›(`/¼àØ0ÄjÌÙôê?™`Âã¸Y t¬-äþÄóÍ`à#Í ôÏýF{µçæJ¸u½ï1T”ËÃÞ$›J/f ^Û±ÔUŒG'AG,ÓEqÜóûç“oŽ—ü+F+"¡Ñïö–Šû¤ÇI+•,!»)2›ÄlEdWŠdI×|гÄJöî¹¢ :U`Ïi)ú«„VMôÂs\ OåVé#ŽsZXs·/ËQ³W—³yµe†*µO b03žñ$\¼‰â×^s{Q\Y‚áˈ7Ö~I`Wÿ5ÏA“tDbkÔ†¾ ù4<·©›5&TlvŸ‚VJz:oõê $õñÀøBh ®%;ó'•š"³“œ+Œâ¬lš®yˆú˜ ú”ƒ"ÃúdýŸ ~{¤ }¥•[û`„²z÷„ ï+•7zA?iÕ°ã›å õž+æ‰(y‹h'ŠØIwÜøTFφYFI3"÷¡#ˆYðB™Ý›¤Ú$9§®ÂЬ%ÝWïo¨ßÆFêÅ«7ïõ~CºŸÛsN;xµ°5¨ÈUpzgÅü¿à§œ#f¨³uùàªÖi½ù¶®²\Ì·zÆß¨Þý¥r1„Ê–7½ ‘sgG+×þsei¯}ÇD4« 2¢SãÔ…D;ÃE•ÊSÞRoKDpÜBÁÃß÷ǵ˜wºèçmôæm†)7Ì'#D™Ä¢­ãø;8Å4¨ˆ³.¯:+ìwÅèõXÿÈ€FÎOÖàäcy… Ap,iS~WhDan½­ý‚½~ Ø?—4n~ç`ÃÚ¤oÈÓ‘ÜCUQB*ã£bУ”5S°8Ï›w&¾w|)cÐ^­„óŸíÓ&¥éY™Iß5÷&$ì“þ6É …|¢°r墢¡®î¯™ŠÌ=ø†¹>‚Chƒz/Q‹‹fÖ ÇØæ.¦¸e®¬nA“j …‚OƒGX5=È iëÍïP³ÑW7+Fq •@ç÷Þ•JvÎn§åvîzÜð=Ñ-Èë—&nij:¤¤€50u ö:@ÃæLXžÉ0 …9›:|®2vnó܈úîlþH‡@Äi¢‹Só3ò![BG^Kš•TœWÄfÈså!0GZöØ…Á®6.ô´¶0ž‰ª3¸(¿r±æA$¦kWíIAÇ÷Ç„ #uœÈ9 æÌ òJÓb’¯ãdíóþf;æ¼/vËÚ«çõµµž`ŒŸÓÐÇö^íd°ØæÙª·Ëí:=(ªy¥xŸ_°ÈÍ#"­’œm¼3~jù;þüŦê‡Å1ß`=ƒÑí¨ê¤Z–k; §­þü¾ÄV)lÙFáâ{Óî> [9,Tk•£›!ÕÒè=N÷Ó|ä•.WÒ²y«0 †“Œ‰ÂéHñ Åa«çуøg˜,_Áô(NæêU$ʘü µŽ-O¼_­Â&OJöÒ¯›s Þ÷„ð¸x ú ZtÊð ÝJ¹ãJô|½¥v¨DÒ@OúÒÇQP‰ye*Ql)[Ì<)„{éK+‰CŽ’òà2á¸,¼Ûûô[bפ|×û ~o8pF÷gÊÂ6Ý€ŸrûÀ[_ôY#%©áÀéb#Öà!ìá5Û¥(W«|ÅH0!Ñ má v3à´èpM°Ú02jv}² Nt…es.öBÖ§‹n2‡×[H€qu+k~s6žhMû6ƒRázëyi•gîsGLÔÞÛ|È #I¾aNµ§44÷ªŠ¢2·’]›7–ÖU†îØêŽbQy5ü~êb7*hB¡/Îúç÷sà½TE§WØfÇ.61r­MÂT°¼÷ìF õC{çQtÝW¶Ãß8áÅ»š¨Á <Ó~Ã==“öìr8VótÛÙ8[¨ëó„Wj–Jr& sZ p)c#¦}VÞe屪ô›õÒX²k9!Èb͈Ðm Pâ>Úk¡Š~”¾ ™1”ø§Š HZÌ!X‘«c9”^•׊¬¯Ã kfR rJ·oß¿àmU^CEv%$Ê&wŽ;›{vÿ®± 0B ›J'à‚xЋ;¿a#±w“¤2jO¸ëÀ¦Ù^­†߯™eì¤A˸ñkš6¹åÉåÜ@Êlz¿ ǧÇåõDhÝþš—0¬¡íÞâ¯ä’ù… {—ªµû³¼ÂZŒ¢zÈOpŒän ‹Ê¡@ŠIÇÑ`}½ò‹X{:´¦KΚ6÷œµ¬.€:>ôÑ)‘âÁY¢oW¦àŽˆ5óèß&ÁõüMÓ@¥ËÊ<#Ç\ƒ–€ù ß^{ ñ“~š}A17¾+éÝT!#™±HíW•›<“G=I Ž*Ú6ƒÀxñÑUH2,[ìd§çñtÍ¥%V@â3LÒlî‘¥¾”‰L.TJeû %«Y<‹&ŒãºhÃ^«%†ªÅë`óM YTE™vØÀ ‡0¸¸‡1h“þù îº5F1ÑYb9¡,_ìë‚neSOèÚ0í¯{<_xîòª !³vñ£Q=el˜—Xvm¥ÃøÑÂSÄ5FFDåÌ‘ž‰&J¡ ý § $ѶÐ]dSJט˜ü]P4ïœ]óžÓÀl¤fø]AcÏø!¸“Ý[‰)Å6‘û 7M¨ë¢ ñ˜;qC9’?Ù |ñò%‚ .É4k ’LÙ²C\”±mp1{ª S<ºs~ÛÚÈM¾›ôÏd‚*3Yu=Þ}éZU(_¾ ½¼¡ÓPÅÕŠ bçòÒ¨ÜG¤ïKk|dX,‘š*¢§A«väÀ-"çÁ¦“šÙ/î¶ÝúÍñ”pa>:°‘HÈ< _kÞ'ŒîèÒ\Ãë,ïG÷g¥àñöÓöX"^ßz‚}IN!ÁÒûãÖÇ݇ј9C-ó{ùƒÁ߇ 1[3Jôz/Ÿ-±øØWLºÃÌGÌ‘]ṆŸ›¥ÛÌ)²»Â UÏEç`ëûß«ªºÒË"ƒÉÒp c’Ò‹×ád~L šµ|û`|º–¾Pç¡RJ× @·6³#ç71¦§™… ÂM%X?¤Ök&ÃÑ¡@f°xC ”®E¹ˆ¹\]/õº*ƶ²hë¸ph za±KÑÁµˆ¢ã(Ì ëA—–,b&Î".gB;k]Ï;¾<²Ï¼ êUý2Œ#Ó­–Ÿé÷‚XˆµžOURb%¢p³Æ,+&ú†¶uÏdGó2n¬ŸÍß½¸ãÔÞäx{MŠÚ[RZmäñÃâÔ.}_8þ:o•9_»ß‰Š‰{îjᜨB£=-‰ò\y».y †#CPdX¥‘T™^S7롇È_h(Nç!'½}‹ª…²&Þ›Ó¥zŸãvŠ"HÀßîz15Èç]‘„ÑšËü:"+Òt$üfMÁ7ØS ³·°Ò‡,moµ4ûé+.s£¼¸À»Eþ~±æšh¾/ý ‹Û:Å·4¤‘õ5Y_ê`06™dødîXo”CJ&&ľYæ¬_7ßðØ•Y§@LðªÛ®E-~£‘.I½kN<Œ* |}•.=]íˆ_~Æ"3ž™Ë埽’Ì^º)gÔðבnësðœ_×f")ù\ü•NŸU©xcŸ&1HjD.‚€‘Wd¤Tè½+ð«"3¶…-xØs8ôBƒPÐ/…]ðΧW‡A¨R*•ð»é‚Ÿ&ˆUÓëŠüÒÝpl¬9–í!¡ÄÚ= ÌUfܳü;ïQsº…-ù7Øß8ȵÙ+Òº÷µM,aéÔ8_ -zùM‘‹½ëeä¸BX ëŸé ih»ØobqŸ|å œò½eÏTQwÕd>[Š*Cª;¿j¦-†Vª[Ӓɵy1xi8uÓDQ Øí†\ŸŠYÛR(ª6ît*ùåÖÓÞÅ´Åb&×Ü;µZ±š$õTqs‚S¦ó“è"Øõv5e•Îé;™ºxœçýëq²Î6N7Ò éu©£é‚Ý~lºó“8^eoO“ù+$ý±mRç5ôx” je‘M5f<°ËÓH“mNr<ÀŸ[v¯ý*$Ïaì.ÓÁþ ê>¸O‰Ùë¢÷[~Á€O;VSöð•`%Q6Èï2òÖãý%»d‰ðÕôâÜ~™Äz¯½÷£‹¬­ƒ«â%áryd.+\ßñ3C®™¡1í¦Ð·…´Æþ•”%3°ŠY~O¢*‹U=›‘e?üª1rƒî†ŠøX¯%>fü謡ô©"t êAØ+áÙa­6:«û…áÝo'x“®mÁóó¿6ôrPu ö:Ú˜Ug°E–Ã$.µE271Þ-KÉ(ýIïÈÊÈ{VÞ¹B8³zK±2õtÚÁ-åýXºt&ø]t‘Ó…©ˆÚ²âûžÖ€ºõ44y[TUÆÙŒŠµÞ«çΪ`³Jz.šOö³<·6&;±ÅÕÙ#ŠA•š… pG“Cü5É}Ó€Á_ê|_Î~ËyÓöe¿@/܆UvŪ g¿_9¦hl[dý"mMÓéà^}òOׇbî©Â{ˆ¦´ t~6B‹©36é§©àoKK$(p$aA'i.ªl GÅân¡«»úÞ.ù÷®í}Æ/bÕó̼¬Föe8ŽR؈7Ña”“©x 0m’þA,ù11;€‹²1ZÅ-å°+È`„N¸§¢,”D·ÿqBœÑ7Z÷ñ Ð¸ß Ü8i„ÏaWñè$ lÌ—ŠÛÑëü1ck&ý7.J"FŸ+Ê(ÔrN ¾¢ìš L¥9¯PTbt©–îø™œì’õÿÝ (Ó endstream endobj 187 0 obj << /Type /FontDescriptor /FontName /LQVOMS+CMBX10 /Flags 4 /FontBBox [-56 -250 1164 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 114 /XHeight 444 /CharSet (/A/B/C/D/I/L/P/S/Y/a/c/d/e/five/four/g/h/i/l/m/n/nine/o/one/p/question/r/s/six/t/three/two/u/v/x/y) /FontFile 186 0 R >> endobj 188 0 obj << /Length1 2085 /Length2 14463 /Length3 0 /Length 15712 /Filter /FlateDecode >> stream xÚ÷tk× ÇVc6;¶mÛNcíØ¶4j㤱mÛ6šmûË9zž÷ÿÇø¾‘1’=¯5æºÖºï2"E:!;# ¸­3=#7@DNX“‰ÀÈÈBÏÈÈ GF¦jál üÏ9™:ÐÑÉÂΖû G ¡óÇ™¨¡óQÎÎ íb `b0±s3qp32˜¹þC´s䈺Z˜äèÒv¶@'82;{G 3sç<ÿù 4¦0qqqÐþí²:ZÚä Í6 ­*vÆ@gÿ AÉkîìlÏÍÀàææFohãDoçhÆOE p³p6(€Ž®@À_’ò†6ÀK£‡#¨š[8ýË bgêìfè|X[m>\\lM€Ž€ì)Y€‚=Ðö_dÙhÿn€‰žé¿áþíýW Û¿ ílì m=,lͦÖ@€‚¸,½³»3-ÀÐÖä/¢¡µ“݇¿¡«¡…µ¡ÑáïÒ âBJÃ…ÿÖçdìhaïìDïdaý—F†¿Â|´YÌÖDÄÎÆhëì÷W}¢Ž@ã¾{0üûr­líÜl½þƒL-lMLÿ’aâbÏ fkáà”ý7çãîÏ™ÐÀÆÈÈÈÁÎ:€îÆæ %Põ°þmdúëøCƒ—½=ÀôCÐÇÂøñÎËÉÐpvtúxýÓð¿މ `baì 0šYØÂý‰þq 4ýþ¸G w€6ãÇø1ÿúùï'Ý 3±³µöøCÿûŠdÅÅåeähþ-ù¿Faa;w€ €Ž™ ÀÄÄÂà`cøüoœÿvà?êÿ>U4´øwuŒ"JÙšÚ¸þ%â£{ÿâúïÉ ü÷ÚPþ7ƒ¼ÝÇ<”Æ_‡‘ÑøãÓÿç%øÛåÿßìÿåÿuüÿoEâ.ÖÖÛ)ÿEøÿ±ÚXX{ü›ñ1Ï.λ!g÷±!¶ÿ—ªü×B ÛY›ü_›”³ádžÙšYÿ·Nâî@E gcó Ñná#¸µ…-PÑÎÉâ¯€Ž‰‘ñÿØ>vÎØêã¡âôqW›€+õ¿)ÅlíLþÚ=f6v€¡££¡ãÇ€1³±¼˜>–Ôèþ÷lèmíœ?\â|¦vŽpÝ(;€A诣!vƒðÄ`ùƒ8 ¢ÿE,©?èÃOöúð“ûƒ>üäÿ‹8 Š3€Aùúˆ©ò±Tÿ ˜€áËׇÍðúÈgô}0ÿ‹þê0ƒÉ? €øøQŽéøa4ý?ÖŠÁÔÂõí/¶‹ã?ü?(fÿ€ Ìÿ?ZmñøÑ3ËÀ²­ÿ?ê¶ùG)uÛþ~”f÷_ÈúÁýxõüÃüQ™ýóG"ûM³û‡ð÷ƒÃŸ«ý ¹þÃ?œòÿ!Žé£|§?â?ŒN®?æ,ÎÌ:›;ÿÑ­Învÿpø¸5—Àý®ÿ€šÜþ@æï$cþïñ7üŸe0vqtüxQüý¸úØ”ÿà¿ßJ@ ;Ðîç¢1O°empûCµžÝî$ßÙ®F ×OÇ—'$èDªªŒÀ Ç;¡Ä‘^”Õm1Ê[ÁÂW¯ã–zè°ÖoJmÏÞ/úñÊ3»mpËÓ˜ƒSùÇBu°øtª‚{Þ¯ÞêVà- ]ÒdÙ.œHйhnýîu¥¿ÆCw•öªØeà_Jg龪EëÍ“åe.`C9ÓÀP£^¸#ÏßÞÍ¡fM½JÇÓÀùœ|e)ðÒúÍó¸à¹V®ÊìÔCŠ£…M~‹:>Cî%|$µäU\+UhJ ·ØÄßlèÀZCÔõ“·tljs ûP?@¤y{&4ÖXi‹LxäÐZ%«5»Ž£hö•I?ÎI©sÿÓ3ªb>Ã1H`ÐÊùT‘9,µ/Áúsð-¥TçAØAÒû2Nîà—ï,E}»Ÿ`ü2³Q޳v—¡;âi'ÇÃ¥ÏÔ”„-ÌíÕ/§üªÓ ø#$†£LæDY}Ý%MɾT9OÀ›¢Š#7rBÜÀìˆM‰S.8„ñ# a+ˆ$¿I˲çΊSO‰\¨bÓ})|gŒ‡©ŠGñæb—ÑïO}\.ö1ƒwŠ[ÃqÕò.Ž‹T²|.æ%1 ³Ã¢“?sZ{ç£ôŠ|°›š !»»çávèŠ'=Õ¨'~}f)ß÷½Û–,pà òÜUâ[n#DZÜœ ÿ>ƒaìÞZŠ™çŠ\ŠK™ˆÊ«ÅɾœÆ&ók>­tŽÆóå“DG#(U·Ÿã—òo‘ÜKUè#ƹ ˜Âˆ%9Rê ¯‹»Ùï-éʺˆ \ýZZé6ÞÆXvÁa5q#R:6¡ò ¡¾¶©ßvOî²qup^MeÒ!–!Ûúß}aVÖ&NcÒ+º‘AÁy@ƒ)G)Vö’{Lk‚_ Œ5’ƒDS†^qŠ9AƒFŸ”)u‹ƒ['Ñr!úÎv¥½Ú_ }#š&&X½æª‹Vðqh]ù”d¬ÆéÒ{C°Õ¡µSGaq¡lè:©7¶»³¢“þü®.ªEW•ÔS™:¯±"“dßP…p–Ûܘã«êr_p±¨p\Ç„_Nô€ïk˜NÎÁš´}¸:Û5MË×J3Vª.Že¼Ç2ÀR=†3Äi5o~¢y e݆±/d ·a˜¶ÛØt'ß=„8ßÅ}‘S¤J:‘eÂ=maŒ±˜jù*ò,Ǽc0ÎæX){͇½S—?üô»¢sÂ÷÷›q6ê’œlñï4"S.}~Å£*ñü’» ~Èå½´ÄÆè]Öo휞@05iìSÈG«O8¼ü´’¾•zeàjüU¾ŸP×}© ¸°’$²èÐõƒi¦æÃ ’™ÚãlÒ,ö'3™@kKNnøµÊÎÆ×¦b&8·µ@mâ Hl€¢Š73iÅáì­xñß¹ ]ùÌźõ5)=âBåR )¹p‹ö±LØys×5YK!ÛØÐˆb/4EÒ¡ªÅ>KßÅ”u6¡è,Ïëjæ c·&}®¯LùiMÝéãÈŠ "0³Ô‘ÞNBKÉ'¦2ƒ47LD_‹¾©Ø>ælû’|õÓÿDªKð£AÇâ‚ãQzÂ"cÔa,ÅF S0íÈ÷ VyEò¹‹°Pgw3lÅÚ$µ© uåEñÔRÂÓ“]Øg”?RvH»#úEÍ1â"«Ð>Égä7á{ö¢¿& û*(%ÀËvY ¯ŠàéTTêã"k$·ì_ÚIÌñC¦süì~¿íj'tÓCåuu—ïíj@õ¹—^4Ÿ%çç^¢ i"zÞlj;€–ů6-ivÍ©/S«í÷œò{Ä·‚^UÄFÿqãè2æ“ø ²€Q¹úO*\’ËJ_Û2Mo$Šô^tØØÂI¶ºbh‡|™ª§ÂØÇ”ÙQ)Ðï÷µmI­ÃÕ|¦å$Ïã0D4!}K^K„×sfÐ@ Xö˜êòëêÇ.‚¹.gmV’ñ¢úÉ7¢!ÚnÂßf‡ •F­Þ³Hø¨‚âìS¤­äÄÚv“ ëñƒÅÎf/RdÓ¥7Cø¯û¯Š?tºÉ)¨ÆU‰:I§Œ)áµÖ?õyAÃ>t¨®w6Ý0WªãÄWÊ;mÌÆ<_!ûu’ ­˜G_ý"Å_¢Ý¼æÏÚ Ã½¶;”I^!–ÌœªBšGÎ4—1O £«MpPH×P6ž· ¥Ï\U3)J­ÅmÞÓ[g+̓ZŽˆ“Yf+ÖÃ1µy1TÖhø¯ËÑØÈ4· ™Ò¯2°øæõWñ<ß%qä2pôTðÙÑí+“ùGîý¶M—nÄ´Ý}gb<ì5ïFWçªÎR]DðŠ£Ì%O¹ ݉óã»1ソ•{·ìÞ‡.ÎBXÆ;!‹ÃR~_ —Øvm „0Páþ}Ì- ´WoA~ðÛâJ/‰Î[ãýr¯m;œÜXq3´s ƒ>÷»ss1ˆóˆô©‹„3͇º`EOpiºnœ“5_‰ZX è†'ÖÀUt;E*߼›ûZüNl© Ù¦éþ*爻|$VU’¹ ûB|@tBmŠM“Nª&?5 †@+À‚Ú½íΑtÉN‚×õnHtɾM2Œöü ?’8âR€e”ÛÌðmC¢«n• ©x¯S¢c`´¹½@zçÈ×I5!ÇPˆ{ÏÛn‡&¬†rtü“£’>›­d[¸q[Y ZWÀeHŒ­N„¶MNàþ‰Él¯è¸Ü!Å*‘/)ô$š ÅÔúµª}MRû¼2mGܼUÅUp6Ïâ°ÑmÖƒäÕ$í¬wì<…ÌitÃ7]øñ0§Ks¾ǽ{eÈÜš/0ƒ¶ºê’½»-‘µøF°–{0ǺüiÀÉpóàg§Dë¬z¤iÛô§öœÕÎã‚q)P^ƒ–³:³[]ºëcL( ÒφÞ÷›vKV.çné(¯UÉ âG)Áëö}õÌŸÑ[¾õûHQ¯×c ‹Lñ{¤)è€B¶³&£¾po%Ö‰íg«ë35Ϊ_<чÉ؉3šÅâ¶žà¼g­›²Jè—[%Pzí÷Mp¯ÂÁ¤6¼­ëÒ¥›–pÇŒ0j²^ÆÍGîhP'sý"ŸõœÒ ªÔbeÐ:{X€Ö\ÎŒ;¾µÕ_‰³à¼˜¡d¬G‘çÊù,ÿ*fɵ67øª“åèb¿qÉsoÑ"„§ßyg'®?“jlu+Ÿ>õ@hkû亄±@Åœ–ø+5a‡”ä‘ýqÆëíæ¥l«²QGÜõVíDƒ¾Ëßú ŽÌRÛ¬~å4Ä’Ý·U~c•8!AÏÊZ¤öܹu• ÿjšÌ¶QcA¬ç÷öxݦ団ëmºâM "B6Œ‘¿ûUÛ¶þ}[á˜l_s¼GYhé· ê›)6äeA/Óc%‘KŸ2“Ñn˜á¥;ßåmǺC¢*pֶ׬™¸Û¶hSÅ<»¦Sÿ.¹9= Ú¤3v}†_^æË#ç# 3iGã–ûu*>³o§Òažð3N¡‘ ;‘›‹–X°Åëu¾žé²^B: ʳæ¤TQÜžò*Ò²kø–z?ú¿%4€Æ ⦚Uö–ÑÂxRd„Œ „’VÄô¥ä÷åÚ"ï+XwíÙÒ{'† ‚ì¤êÓdÛÇã3—Ÿ 4”GY®-!ûʹµè Å´jæ*°ïÏÊ^Ç yyLôvÝ óÚc™»¥ÀJPÜþ³Óhí}7¿å+¨sŽø•ÇÕ^Ýîí%ù¹×µ›’6V¤tB®C3M˦ üù¶-톥œ­˜©‚ |±²A•DEý¤*ÎÞ´ù¬â¼÷L”Hn<Ýþ|a¢Ò,yiŒÈó}Š%!ú;«n/^ÅáÞd]R£Üm2cåú€ÐMÛv"ê í«´.ttÎHÂUyd}ÕÌ£4¦FBò3jH`¾¼y‘‹”÷íc0Íû"cI1<7˜5ãË~Ŧñv^h7$ ƒµpöŒ/:¸gÈ=Ý^6|§a2-š !½Œ¾Љ!­²wòj”û@røs-·¦îû[10K5ZEPT¹ñŽ '¢Þyñ|ÙŠ‡¯ì²î÷Z.¥Å¸ËØ+à‚C£ å²>ÆÏÓrp¬6ÀŒÀF‰ª>£EbÇYñ¾RÃ9¨Õf*ñzhßÝóºÑ¯ÃünëÌvínÊÛ”U±òí¸ òz"zFçCªò‚8Ó‹Û…ùkŽD¨ëÛ49·]MÞ xïÄ®‡œæ‹u7lf’¶Qnï±évöõœ%VÉÇê‚y+ÇK_©`±˜››ùË4׊ˆ*#Ú^í4÷ÒÀ€Ía”/!‚M)¿ñe±šº)žYƸ „8Db}5iô÷R:•I$v„Ôûu#E×M;ë‰`ý·¨\6 yÞí`U¦Tã—9~™9-¬•¬Îj$»žÁ1ƒ¹ÆUÓ$7Ÿg[VèÜ¿~ú=¯1É\£·‚ýŠä)Quú°Ïž,{}õ0íÐWy'þÉÿLofp‡+Ì.±÷:|,œøùeØDó,®ý6¦ES̤&ã4$d¨Ú¦em­T‚Æþ eaW Î[‡Ë2Ÿú¢þ¶¶%O†¹®¥ç6‰½Õç•N.VÍkÅ@Ž8Ä»LÃ.®i’ò.H„Ëò]XI¯2•)±°r/ÄžyöØ44àçXž~ÖùÚœZÁ)çÑø5m;¶IÖ;%­D¬¼hí–—ÒCè¾õèbåò5b˜Sú¼ÓMµ%1J­WCY~Œjo„y" 0 ŠÎƒ¾¼µ%ј̄é\w±Û}M9­š_î>hÁëå‘XÐF™EC2­/òôçÐÔÎF@‰…>4AÕ `o“*~–¸•f8‡'õ!ˆ!ªßòÉFÜö‡“©ƒjÒâ¾ËÆéìAŒ·óA¯WÊYçŽR‰ÆüÖ¤h®9¹·âÚòš¦–gQÊY†½r^Ï®«wÿýBj¼¯½ü[ ruš2oÏÆÕV#ñ97=ø±ïdê yÅsèÏUØï ²xò\u¾|4†“†9 OK¡m”õpïp]à£sïŽ\¥æÍSÃ{[y¶³dð2¾¼«cî)[.†ò$—JÆXœÓ¶¤IÚú…½­on¡jÚ=r&û§ÇJ©oë¦õÐÓ nÜkk$cHäºr­<éÔMìñ·/[zr½³Ï&-·²øEìDµdŸeÖíÇÓ¦mÛÚ÷`~–õÕ¥ñv5Ñ`¯¬õ`‚6*C‡Öd…”‘(¨%é?ã˜ÙgY sÄDA‰g%¶EœJ’¾þ‰Š•¸IˆÁCï +*³:¶j¨U8Í“çfä~Ý~ŒDôÜ«Á`“Ø;Ú5ˆ—,ö»Éù]ËUŽ%Jná‚Fý~yêÞüÇ@'Q½ŠŸÔoYÕæB¬À'f@:s4èÒù*ú:ˆp]£JWGžŽM¶ˆ›š­VÏØ8¼ÖC…Œ†”öéc–,Jføâ›èàÔ¥ªãt]ôç'Ĥ(¶Pd*Y#ë€49=mkŒfÿ‰„>ÃÌNB¨¼~†½_šXÐ#nÖÛ Ä•>ð”ßU\÷¯ôÔ…lJ wÉ*Ïu«ZèètåC†ÑÏã5¸ìÍútmd³É_ËëOA fü òô²8Ô(2E"ìÃ$|ïÍýX˜ŽÆ6ç.È•½‘ˆÁyzÞ÷¼g/‹\ûЃ¡µ*H…äŠf=qE1âŽÉn~¾k©ƒé®Fæ’jf“%ÉE™†ê0¤a'TP×)Çø…໘¶`Ja›Gú9=õQýA”ÏÔH%³¤H8Â/‡ns<“âèÑŽ}¹UsÿØìEvq⢊b_»¡M€×ÐŽÁ–çkÀrÌÄo¿qjx\aùlb§:9t”J¾¬û» FZ{ø­zMûÀ_p¨‹œDœÓbV+½‘÷[4h°Û 4&‡®Èº©C8Ç$#ÝíœG¸:ŽÙ~ó³61æ5¶åüæ/xñ-•„ »â° ÃÌÙ×H®œ'b¦¦…Õ¬t©<‘Pb.ÞïB wÒîΊs?HîQ­Ç<|Gté3åºØÝq—³Î Y½¤+Á;Cf¶²Ü>BKð7÷ÖºKÒæ7†:{`Áý¾éØN[ï}p¹¶{xźd’lLŠ ý"wÂSk·iqüœe˜Å÷ë0‘,³hê1_›Y¨#£ïøsì‘ûך¡#¾0ÇÕ‚8ua šö-LýX©h^˜ÿns3åïäEúv3ÑÜ…QcÁ rñÓU4¤â¥høÝJX†|ZӾ݂‡†’‚-¯ÎÅŽÔÛáò6ÒLT¥W{ŠkãñnòQòÞvÜ=§ÃÈœêÛ¢Çò*r+ËéwÔIKÈöxî2ÀÈ4ü†kÒäÏxk†–¥¦]¿ð(oÐxå]ÔOÎÚŸ¦!bHjÖ ‡iˆ­ºÙÅ.ïZ]ØôЧüi6¦ÞþO£y“÷©B\úÎEñÈ»šõÁÁ#w³}g6H¢‰-ù·*Z œ½ôt-hªƒQ·€7Ú¼ÓŽN÷­ý§~Êè}7ÝœŽ£©tyäâ~WçœÍ&È íwE#Ð|5C½Î}R§±âÖ§*Û¼ÈçÐ`)y‰!ýÌ"¯Ì šóZ§˜åˆ÷(P¢BÆ:ËèP kÐí% ì#Û©õºw"Š­h•{ãÝ«°©WúMKR¨´4¼Ýn‹R õ„Ñ£ÏöɤdÀïëw¨Ó(7ö¢ ¡¤O6Õ Ý:¹» ®-5.ß“^Vr¼¹:ïôætlb˜ èÐÙŠY®›Ï£/-fá+ŸY1aT#<[4×…0 «*²:W9¿~]‡Té0<~rDÿB€ ÉÌìõÙc‘sìˆÃ,¦Yú›‹(¿.Uƒ ì+xþĆÌ@0~6XÈð®£oȼê\»lIÙ¡ÓLÊL-·2e¡^j‰Àí5¹zp ¥9–úÝõ+Ç©8Ü›ž”ðo"9´­UÓ5·þyi£œJÀkÅZ–¨òœœMK³Á¡ÝüÓ.^õ« E‹K§!·c!%Rg|½,ÿ^ Mô¤–ŒyŒ¶q]b˜)ñüÏÉ;)Š.‡(åÞ„ÏŠ-‚–##*9Æ_@Ae·OÛü(ŒEÔ5³€ˆ“¼/’œ¯4‚BºY`ßÌøŠÖuƒu>Í¢–hsèp¾pŽš¤µ= 桇Ӱn‘ï]2·]hwNHRå€Cù.ÙÁeŽ¿˜o`ÙáŠð¦¹Õ~Z"xuF&†×Eˆî¼1¥ðèJpjº‰„ ºx.*s'¿ 7~¥£P"»”Ñù‘ç3é g‚uϱç%ó¼Ú@M!:Li{ªøM|) 2V@’Sÿ]1ÜQÄÆn»WS•qǃåd¡÷%oó{W*¦H7QA ¥a ñÈ\ã7÷f6Qõà[èõ:Ya×ð ß |­!‹D&4»–;ŸT‹ªÞrÖ Ò´ij(8lJ0¤C”úwÍ$‘<#X¡`¦F¡€Â³P(ˆô‚¤Ä+ÝA?ñCÁ’quŸ% ”à¾Ê©ÊC~ò*"‹LM™¬å„žaŒvåóŽM9J‰’XMöÞ¹£Ó%ƒ.à ÆnÑùš`Ë Î²,¤Ê9‚–ÏU<2Õ|ñ-4>¼G–<šÇ^¡@&ËÛÞ7r¨“ûvA-ï@I`¹†4[Ò~j§¸ž$¼m6®·tò7¤Š" m< mí'2ù›€>Fàãî©ãçtà¤b:ªI4¿ƒ…j~í-êï«ê+Ô –ömKE þxÒª¦æfV¢F¼,S0Pxx6{·X½¼¡˜Æ@çøZD¾¬vÙ®º;„ìöÑòþ-™ÁUvžÀ¹=LXŒÓö  @p‘»ÊW½ýþzöÉD…[ªÆ ð>’þ Aõmþåq§ÊVÀ…‘(‚ÊVíR†9Ûý@3«c°¬ç‘µë¹=Ÿ£Tpñ`8éW‡O(>.6¤ {ÏðŠö°\îªx€_E⸑„MÓ‰ìþ]-üÅ`¸ïTÉÍ%?Ÿa®ˆy óUŽœ*Ü"ÈN1ÃGÎ.NE}\]*MØþCû\VMM7οÈÊÑ™çBÎ/y£1¬ƒ-D~2Š¥cFf{C¯&b81Uˆ!?«8" " ;!…bý¢gšVý~£à¨×Úúg¨ôX¬V³äAž®h÷ýÊSÍt~ÕD‚wÇÞº<§(¢8½¼LYö¢ëÏð ×hüð&€Å¹'Í0## VŠvñ®é+„NÌ·I³u·¯ç‰¬U¿æ0$žŽ¯obÞªÊËF¼ª>ï u’¥V=±Ž:§Ü——9™=pÊ|/.ºFŸðDMG*ÂÈ ¯òH# Þ4çׇZÈÕΈÇî-pÜï Ödã›fõ÷Qܺ¨TU[²’_ùw7É_ͪóD¹½am‚€%»‡ªtrRà¸ÚR}„ùzu'È´3¹•ÏÀiz`-ØTáx1¼õ¤rqx” Ȉá®Ã rÝ-ÍhpSlBw¹I) ¿ôf|‘æ¿È¯Á=íá>¯1 ;ŽYñÌÄv¯Vž¥józä]4"µ®LæXM.Ìáwéu:ê VÅwž6_°šYC#¥pYoÔÖÓvÂh{ òßžI×D^ @—ܳaéRÇÖ™0ô:±½V\¦[¥3(ôkÆ9òž#cžØÜöÉä‡W¡ (P\zÕ‰?L>¤#÷ŸÅëÑðÖÚV‚ÊÛ™G>Y>QÒ£ÿü2v„â½tß`c@\½æöw!…iëùª$~‘·²HþôÌs"~ÿ Î[8K_â|Ç ¨_zßè• ô‹†òµ?á[>M“ vdïC–…ãÑO¤>spUú¡‰Ó½ÑåÑ- ”˜S‚‘ 36#y—–]…áÜ_Ïhê¯æ2ÃÄ>Ŷ QóÏQ‘q¨óƃçLÌ Ú8(ž+u‡6Àøú Lšü²­„‡²r[0bWÞ­ìmöóˆétã‡j¿u)‰Ö¶´4rÝ‚õíô=MC‰ŠÔP Á;}âô뉖d¬$H»…Ùõ!ìï½³XˆÎ j ]éÌèÂ$øÈ…B¯KãjöûY4%‚›1M´!†Äo1;ò^ß ’ d`³i gcÒ4Ù‘´96©ß ÕË«•ÎpÙØÛö„¯‡îöÿŒ¯âSÞÁ̧-%kß"¼Q3ÍbÑ0ý–=Ç’xßM4mÚД¹‚;w(Ю;8QHFŒ…ñTYÇM2<ÊoÔG@Jßõ>B;ªJï'c„·¼⽚R2F'ô«ÇŒ9º]n‹_Òi–o #W>œr[ˆ/£ÑÀŒ2\ãÂPÞ$ØÖä-‰¯‚c®ø1½Ì”¡ö“ªÄ,NÅá/1ëŠ4 ¨}â}Œ‹ —pL¬Zå>[35ºè‚¦gÔ@|˜`ST[ýÖ·å!oë6¸+gGòôÐkU ¢dsF‹ÇoÔG¥Ðs]“0à‚äЬ¾pXõ«¡°ÃÄF 'ô°wb6é=P ™&‚Iq+Hƒ.÷& ¹d£áVç”Vê‘”+ã›ÅÛgk­³ïå›J&-/8×»"ð]Gç9¬²ä¦Óä0ä ûÖ®çFöúSÝØt½«rEjÎfæ(ÒÀ6Cæêeh¶yñ nNíçp–£¹½'f´7ÐOÍ:x@ˆFf¼jöÄrŒšïú…çÖá|qlÑ)_‡Ýûuê÷”½ërÌ ÌY òÉ é}ë–§ç´X“hG«ð÷Á`?•R§âªSmÓøk¢N|fÖ÷§S̨zìÃûVL{:gèÖæÑ«,Õ¼;懦|_t£°(kRó«– Û©©ZD–æQ­âèT¤Ì…‹ñÍ3§ ôB>y§qÄ * {\[9†Ç:ªÅ_Uk}ßõ mEÌË‹·çg{G.‰D†Ê[ãÑÃÂoZÕÓ2Y4QE§~u©ƒ¸o37DœNj)¾peýêΊºæ 6Öסã4ëù}˜u×Ó‘ÚÈEí6WÝ-uÃN’'ª¶jÍ]TõγO ¹Ÿ«¤Ì䞦¤zZ<º9FœíU¬/Ž˜œrÙRÂ¥6„ ¨WsªG£:¶òP¦P78úÀåÔ¦ÆÞ3w‹ªáÂy¯pêË[¹"(ã„Í:7CžË‡ë¿}I¦ž>„áÖÙBRMˆì•±Š³#æN†þN3÷l"EO.pdhÁ1ݧ6Nì&¤ª!ü’ÐãòÝìžlYÖ²À´ `Ђ‰<¢©oDp¾÷# ÌWËXµAzÿšù!™?5kŽ5¿’«`ª–ÊšnÄXã1 WtãeÑBDoÌüvR¬â:t^Êy¥ÉÿöÁA ;|Ä›íÁ¡Û×QÿË'ûLe³[ÝZœ|&rò´LP<œÁŒó±JcÜ7O $ù§ðÇ$;`pæק\ºãΫS |My¬_$ßX×s~ÃRš¬V§2i¡æøf›>¹¸ŸX:L7„¿¯hF2øÕÞ¢*Büp¼ý¹?ï» ŸêUçS“ݰÿ…E?!Þáþk×AÕ—z·*f³bÆ¢”mÒzåú69‡’Ê»eF.:ê¥ÔžÐ•Ò)9íïë m)ò"Ÿw ó[˜îÄãÂÕÌJ 5Þ$®Åõ0]jù4‹ÏeŽgwâg&õ‘jš0 L(1þPŒ•oß‘tÄ`3V3Uˆ€ôÃco–eþÜ?ÌÕNcÖÓ¦âψΰÝXÆ><–8çõsZ¤}ċƱÛ7Pá÷&`r4ºe?Th[Â8Ǩ[åÕB1¡íZ6ʰS^!$ÒŠÿ|¹_,‹p¤½Q˜qÞ°~¡èä {^/ê)umÜ=÷*ÑŒ?§ïò†Ì¨¬RÊq± Þq'êy%OTè²H5>˜'f–;9›A‚ØÕ#pãÆPJ%ÁD`,ε•cë©ÎIÚí°Ê #ÂS·“’&ØAZÄ4жd>þìe†ÞU;•¨è`Ÿ’¦°q”÷*ÇO›+^¶%çñQdÐÎÿÝØÝ%·ôÔjõê35¸#†+§µ[>Ó‹zB)«w¼Öi»&þ;~dŸÜêKEBÐÊ.ƒ{Á'j#‹4¬zÄqî!õ9cþ”¶Tò›‡-l¾îZ9_î*nï— ÜFÛÔ+×D&ÔÏõîß4÷Å\\ñˆÇo ÈÚPtU ü£ŠÏFgÐJg*£]Aí!n ŸpÝ¤ÍÆ§%!pÏ2‡FL+I‚²Æ®cü8ßþ¤Á.ß²3˘«tlf™&‹¶É¯`4p_ª«ñoúÔ|^0ë=?ʦ £3ÇÙáaåáküX‘³&R=><²~0¤þÀd„~·’›â|µ‰äüšm¦Î8Ö_©õ ‘¿¿Ÿ¢î·÷dˆô—¤ÆÛ$ÍSÓ§_ENUÄÐég’VÔ»6Ý]+lš_Іåé^]õ“K^!ßÄÆ(e¿¿¤ÌüP¯fÏÝŸŸaóð0‘ˆ"MÄÓºr‰ì7åì7ÿQÕR׳ûe¡a¡—|W;È©_$O¾—h²¬ÎKªœM.“ÏånŠQŸ‰æ©û§$êéH4E—¶¤Ok/qãeÅrØ` ’*IyˆäÓŽ>•j¬àƒáRË”®=}e*IÒ¡™ëûš · %h°e¡1sûoyÝŒ[) wžÄ?„Ih††žC~.äpV#œö®½µ©Ëçw–Ä7Ÿ~« ]шñIR‘E°1´†í~­f¸ý"Æ8 5°9ªÔ5ëš™š9ÊîM6Õªí9Ւ΢¯¥Äàúr°zÿ\I0ÂÀ]Žu&ÛRÓŽÎÂ!êïSût0"ƒ“á•Ôô>¤{ØD4< 'Ê?ñ]ʤ1Z;‹F«½»ø•âZˆþÁ¼j-é£³Žª1ÈW:ÛZör‰´ sªÑäÕLQÊ·vxæG?ˆ-#HÜÈNZVò(ôg3Béúd.šËìÏÍÛ ),Ô_ÀAUß{ Ü~gŦ'·"-oË"×ê9[Ò/ŽÄ¹”MåÀ$ÎÕ¼T6w6k4\{ÞHé’×l½;¹Gw-qåŠOL,âóGaGþfI‹À%´î&4ƒ4^F `“Ír1Á :ݵÌ2ùî͆ÿ,¥Úpà³:ï0ÓPºØ/¸‘ÓäÑ‹Ï>¬÷jýM|ì”ê9Ü)2“ ÂV@»Àû7çé $šã™1<™jeþf›ßµ9ɾ—Ð…‰”’ûM(À‰äŸ·3Zýëq& ¦Ï©˜`F«€îPqC°–¶<5\JqNù5º?1W•¨™­·ï^iu"äŠ/\¶cÜ@B¬ˆI}MÊùÁ±±jOUûTg“ú¬äªÞ¼C=Þˆîfšµž-û)F#µ+û©Û£Z·JÊȧI^{±É_¿&$~ZÔhNa»ÎˆAxY‡þL÷ùLöêY"9ƒ£Šê/%°™º5P9¾PV\è.–ޱ „ÓˆÖý11:k°²~ž€ÈöÊøøÒ5Ò{áNvîà&û«ñÁr¢•ƒ™°LöÈ9pKÃäð±=ãäø27v‘¬øþ2ãø [Tk¸›) =(>9ãQöµ„ýá-¢usgˆ}mqox§a9^%xp]$·t ä„‰íëòéV—:Ò¯\ÿ°g¨OîÑY_ôõczTA˵£+ïÌŠc¹Q.çÌ+»ž+gn­÷6Ô(r£X“0Jºäà9Îy(›Sl&`KÜÐ6¡©ÒeJÊX$gþÁ:æj×M)·{òÚpÿ¨†TâcK§AyÓ¸5~Þ[ÔvMH¨‘X Ûõ5CNW«Ì°°íµœàä×%¥·˜ë5:J¾$ƒrnO8¼Ø¬TZô"*!à;)ßRø*…iKlbÚ¾OE0òþú=²OªB¢&¯[`k€zÖí_î®É?'HØÍ^o&ç=Íë¾ÆJ¸Aå‹ ÕˆËè“Á¼{»ƒ1‹ ’ÛKfÓÏçØ{Û¢Ãz,õÊÍ_ÂÕ%y¾m \f¹>'ÓRökrÌiejˆ/ÅþMb`iö½úYÿê¡V/jÑmgÃpõP.¤ž…f_iÞ`¨v¬!Þ3Èlýv¿¦Kfb¥ËLfuœP¸®{æê߄ݖÑH£Æyó¥…htßÉÐ'Ï*%Ôd÷’‰êÍQD[¢Â¶¢ÌýC°8ÇâöZôynHmc4ª³Ëß"‹¶ÂdWÿ9»T@Bu+V~ô—“>dÑ÷GµT·iÐOôÓíº?3Ò¾>&¯gJ4@‹¦¼Rªàk–’>"Ýaü,M~S‡…”õÛÞ*Ãjx @ð›}7'ÍÜ»¾iï"*Dþ­ü.'nÜë³%²žÏ É–¶?—4Æk¹rr‡Åʧý9Gì;…ërõÈðÂ(|Á¸~;Š–Qß »e»uúøË“lî ‡glÅzbìÝÆS-Qå‚jà‰ˆÊ '£ü+¦'óÏF®ª%nÚîä7·µ÷Ä[“cÏ¢kŸ†7êcòì÷P»«ù]5¥Š=U&bš¨S“ßýaÉ"˜xÂÒ=0»¹EòŸzoÔqYÖÑ8ô+²á–&b"-zDb¿ X±Ò‘ôZyJfúYk–2×”µi½´æ¼éeNfš˜_+"±‚I¼ì<´Ò¹}5‘§>Nê‘‚¿YóÂÙ¼ µ4¯ôÜÜfrôæ¦.ó«„À:)Í¿…à¿Ár ‡ÒpÊØ Î ÿ޽ü î$Rï&ÞÅ=,Áº“rÝ2}ÅP%,ìUÓ¡wÙû²Ò•MÀ“G8ydO¹»2=€øäC_J­ÖH—#}‹$)?“2Ø2•õƒJ„õ»ÛÑrG¦FÚ|È[Þ$ùëÀ¤ä$-{ß8¶âNtˆ|hß0qÀÃàÃðÕY€\x‡Å&âUF^Wq>Ôil–ÅŒ6Upó‹ÿF|9?é§R³´(¯69©¡-b«ŠgåeËœ˜\.w[áìOg yâ¾_3£†F¡`N9 ç}ÿ-åø0}·—÷·ùGƒA 6Í5Áº¼;º¸Ñ¦ð ä$*šg`÷‰ß´Æ{°u”1þ¯BÉÛ¦|„÷ÄuÑKרöfËTkµ±Õã3ø¹üßX2Ø2¦BÁ|†ºì¦Ë(ŸäJð :^PÒi8ÏqÏH¢¡9ÜfÚ! ǯ>§|ñwuC8¤Öõ±ì“ÓïÙêv™3TÀJüNìr˜!A°#ÔI¦¦×.,‰ÆÂC±Šwô‹VÒðŠ 20ˆ6Ø‘v8ì]éµÔ7ðk^xGWôøRÓ¶3MÑä®ä;L9FØ×²‡D”5«-tÄQÂéÄ ãÓñ”yzcÌz­%ßQF(÷haë9ÞÝÆ¹Ì3*JeÌŒ {úª™I5龸›¬.Æœ“q-|6c^Ak®õ©E?v%x ÉË! »}¾¨Ê¾«{ õÜVo7\‘‘YÂy_„ˆì´4†*¶h1ôÑA–7sa§Ó5ŠÂï® D„‰˜CûaÃÏ 1«·жðíÈB\—øpö‹ŒØ goƒ8Q°ÉþõQ"v^ø`I¹rI l甘ßJš î[,†âÕ¥*œhèûL«©x#I‰vñè¢2¾ÔIQ¾=ñÿ¤[ÞbÑE×€WÁÖL\çK:|€®wÔfÝM¬R¬˜fêï>é M…ýóSã2s ¶nìä%ÅhÙš†jðŒNk˜yÔ®& ÂÄCâ"螞̹$ŒÒ˜?•ŸrMX=']Á€èøø.…üønê`›vϵk èõ¸@Á* ºjJõ&7;¨Ó±ó0¨\~¹Æ¡ð9l9´– ák4ŽÃ€ª0Œ" ;ú§Ð%QMÍÛ,wß×´d–Ï£ awçi£‡–ê¨/|äJYg²î³wËÙ8É _øtE?S_©ËP9PÇ6µ}—Q;‚ø{T»ûÿ÷$´8 endstream endobj 189 0 obj << /Type /FontDescriptor /FontName /LFFNKM+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/A/B/C/D/I/L/M/N/P/R/S/T/W/Y/a/b/c/d/e/f/ff/five/four/g/h/i/j/l/m/n/o/one/p/period/q/question/r/s/six/t/three/two/u/v/w/x/y) /FontFile 188 0 R >> endobj 190 0 obj << /Length1 1435 /Length2 6118 /Length3 0 /Length 7088 /Filter /FlateDecode >> stream xÚuT”]×6Ý RHçÐÝ!ÒÝ1  1ÃC·”ˆ”Ò¥(Š´”‚ ©¤„t(!Rßèãó¾ïóþÿZß·îµæ¾Ï¾®½÷¹ÎÞû '›¡‰ Š3Ü ¢ ‡!B"²5= K @DDLHDD”ˆ“ÓŠô„üm'â4‡ |¡p˜ì0ÔmS!ÑD=8 pÏÏ%eR²""Q™¿‰p„,@äuè îÁa_"N5¸wêê†DçùûÀæed¤~»T¼ (èn/tF0È`C!È „à‘wC"½e……„@^¾Bp„«"¯ ŠtC|!ˆ3à—d€>È òGš'ÀÔ êû`wA€Úà C`¾h?˜3@g˜hë ¼!°¿Èºþ+Üï_ °ßÎ 0îå ‚Aa®¨'` ©+„ D @0ç_D§/íòA=ANhÂï­ƒš*FZá}¾`Ôé+ä õü¥QøWô1kÀœÕà^^Ò—è×þÔ¡}îAŠëƒÀP¯\ 0g—_2œý¼…Í`P?ˆ¶úÚDôo›+ ‘•’”@|@°›ð¯¦AÞß ð—­!å ÷¸ e@B¡.ô‹å ò‡?H(ê?®ˆ€@€3Œ8A\¡0¢GG›!.­ÑõG@6"èöD~=ÿú²Cw˜3æôoúï k[êjóÿ‘ü/PU@ ŠŠEe$E@q q€””( ôŸþuËÿm5Aÿlï?BjÃ\à™¿T ïo%þZƒçÏÜðþ™AŽnh€çßýo+"!FÿÿÏSðÛåÿ×ü¿¢ü¯ýÿß;ÒôóôüóüEøpÔ3èÝÐ~HôpèÁÑ#ûoª䯉փ8Cý¼þÕF‚ÐC¢sõü×AB}5¡gC(ìöWý]txO( b÷…þºs‚@‘ÿÂÐcö@ß+¾èjý† è©úgJ îüküD%$ D„.>z%@Ñsê üÝÞa!‰v å…\à¢_5•–ûúyyý¾öœ¡¾Þž  _Œ¿@‘ÿ‘@äoäùÁ~z<÷zs¯ßH L47 ËŸ7Ä´ÿ¨Sa \UøÈ¹n‘Ã+ˆšCtøý$ÃÏä­Í‹ZBœªd¾{C>¿ªÁs¢ü‰õ µÓÚ„ÿú‘QÛEÈ¥ÃCãñõ6¢Ù1Úe;*ýÌ„L‚¦Ê!W>!æ‘Ø­˜]÷8‹|ü¤É K¨~ôi6ö¿øü>nzÝh£VR‡øòÅ„`ŠY²mä³IÎb§ü)zv<¤ 3åaà­É“Ó”…nXï=ä' ÝM+GY/‹¦žO/T™Šúv3Üa°¦gÆ>¡|?Î…RÝ̺G7ƒª(_,dKuñ¾ÑR²95DzøåqKЀßFl„jTß©fîªÂE2FHߛà µä"K¿î0žT©é9ëà3aÇqÙZ™¢8ü«H½ÞÚñ4$¤^k…-“ðG‰Ùô—Ç>‘Jëϧ•åòÓ/¶S]#¼¡ .½ö¿xÃCù%Kê†6™¦0®‡5¯ÖSÈGÁ•ù¤Ê-ýüžô‘&?›‹ûáÊ5Tev¹Oͯ^m¡RÞ¿¼4ÙȲǼ68e˜(WÊi‡È§P.[ÆÄ¿I-¤(ÖÂϬYŽØ[sùyO¯ º“]rKÒÄ8'$\è>"-–KŽòyE1:XñöFÇ-™3ýñ<ÁOßüK—y/MMKÕù˜Û´uC [|˜ê¾T|v œQ<;eö# íH«áÚ’ÝQð´›bÙ}<—}ù²)~»LãQR5ôó³=›*öY®NIm²…o­kȽ=2ÞJ‘Ë cäéިè¤Æ6¢(çêÆÛ9E¢j= ”=g]è–R¦Û8ŒËç¡b•ÿ„™’ËÂÇ:e80O-Êõ¼Þ™%ËÇ |¹Ä±ªMæÖвٻ®y ѾL0iÄa.˜#‰a…­}}‚*³&¸ù§ˆ‹Ô\¼Ù±ó7ˆW:ZH²¦üpHŸC{¿†ÓVf>àÝEo¬a ¶÷ƒï׸¯·; Õ»8-Ђi— ¶æP∈u…*j­¬˜Ôyz½OôÞiVÍšâ@Êø#ÁýQ*©äœœ™iã³t=¦äroë6w«ùmÄÓT¿óa0evøˆ‡ÉI5¢ÿ²±c-”°LÎÚ¸fÄ2+Ælt×÷Î+Oí¹QÉ—†¼æï›ÂŠUr"Öv̦)Sj¢Ô9åR‹°Ÿµ’D! …'S,ó‘EÍ?£ø‡Ú8k¿;©’޾–âá%Ž»ˆ ¤Ñ‘|œqéó¦×èêÊsA–®†ðé;GŸ½É háwƒÒN©-ê/z:l]u<¾Ûô.òC"mw­w]›žõÜÅ+§5²ê'ɰ­ËI+Zz,ñ¡Šk[ðz´ˆÈËÊC5f|j›Èš¬ù@†än…À¸‹ð»ßžCÈ|Éž%RÒ\UÄŸ§¤”6”uD˜o×$ÄBs'ïÅé.ÓüÉ™žÉ€š ZÈÁÒÄ`¸d—%ª™ÚÌ_÷äß²!…MûÆB VìÕ×…²¾åä‘\Yjýw VóµíÔgs*_p—<ˆy8zpª®o‚×gtg"“ñÅTL¾¤üÍ×uaø‡ìê[]®S@êeZø$Ž0¿ô÷šÈÅy«üŽ(¬Û–yü5ƶëªO'b;ïwöȺÐUZRèõÌï’GñP> Câ…&÷mƒ+ø_8Þ?\ó§fÏãÃfÛp]MÅfö$§—ãôØm%åˆJ~â¿ËNõŠó³Þó78*é’Óqú¨ðØ·œ¯3íRèÕ/å6µwÚ7JÖ][ØòŠÕL ›³½9p—±AWºDOz?Y¸XQP³ÙáBÉÌ¡-¨—¶Gw—ûÑò}—Jå•ÜÇþ³_ª¯­¯j݉õUì¢Ë©…‚^AO,Äz/%P Éoï°cPÉ1nÍX]ÚU¦:±AG¨Ë†zë˜:Ø­ïU1æ&þµ’óý_„A^wBV9©çÇ'¾ìñš†«u}Íšô`/}(ÆJ½ oPæ36›H–LX“gä#Nà˜¨>p؇ —Õ˜F~fAéÅHrÈÆgÇÕ<± øºXp’«¦ò¢ “ÛÝ®XáâBG<÷˜y_C¥W¡¬~TvXÏ`40ÁonùÎfõ©“À³hޝ5‘>}DR&Û:±În¨¬­=!:¤Ö‘eèƒ5mÒÿHs³$`v!Ê;0ÆÌzŸeZxMpÛÀ¿^»Ÿ¢G9(üyR” Àî5ÇØ æÌøÛóï˜aq(Œ6ÂáÒàwòÖ'ðT!¤Ó ªËëÛ‡F-ËaCoJê8'ªŒBlIT¡:|úœ‡a;Ìd•(‡ .§¹„<Ó+ê#ÈÅ8íÑ`ßφOD‰xÅ]2"bÖÔW­Çˆ1O‹Úö×" ¸WruJwSôé^¢–÷6°Ý ´³ç9<¡M›AèæîÜ ¨–ä› ¹‰PKLPoÑÞj<ÑBù«Æ¬tÈàç'Q#÷h¯ÅŽ“>_ž¡IN^·ÃcþD_×#€­ùýL¶înñlú´§©fÖ]“ð}(ÓÛ‡·»d3µ¨­V¯b^ÓjÄ Þø²°Æ0Ú¬õµZ²WPĘ~¢iþ)ëÜõªž ÒªW5³W;T믋cJü”2³Ë–b-"!Hckþû¶¹Ï4Ü"Øü%L"”mó±)íëÖFýfBô©SÄæÞÓvF½¾ ®® eƧ0®ÑiÄ7M&WˤMÑÖ‰¬¦²P† —°gÔ[Ìy:µ“”Rº3HçI3˜'+јêÙÊ™9ô¬ëf‡÷ï»ÛÁçb]™Kúxû&øÊB~tµ»>qÍ ‰”+„òó¹,Sí»Y«c`:G¶f½9dßÛh#ñ+Íwxèkø©Â…×u_9> KË@'ø$“(bm£»•ФÄþ0ãd"]€ÕxÊ%m·åZãmnÍM1³ŽÝÿ¹Yö°œìúÑ‘ñÇu™Y& çθó€îÌ!»¯Ðªi†/RL@ÃŽ«zç}çOÍÃÝJURQ¥uª!Ù—ÛY“ï¢ÚUüüÞ*¢LûÕí¤^ßì¼;t^ÙE=:¯&˜`Xµ@Ò|‹Öó¦Š+‚~Šä¡*>ã· #õ¶¶“~ÙÕ[ÒZötи½;{Ùýj4B ‹ƒñ|fòc}Ï›Êíê' vVR”|y©Óˆh£k›…ÑüÇDÿÉ‹SWÌeg=–î°Û|m† €¦N¦T¬¼åD§%±˜Ê-Õ×8· ä$§Ê@c7wܱ´±ç¶g9#ªAಜ`Ú|²!w9¤¹ùãÇ»3Jiý[”M„™f™aÔÞ[ï5¢˜nm®=¥ªòÔ~T³ w•ŸÞj£[û¬ºýúÜzǶ K•†¥2VÂ-*Ñw‹A£´(†ÆÙžª?àÅ›D¬9hóK/@‰þ‰( =»Rÿ6‘£C=ísk=¾´! ¦íÖèq@No$S´ §LÝ÷\çÉðI˥иdþÕÝwV÷>Ø)χ6ÉX;Õ-(–Ì^øÚ•ÑÛOep·¢²R3x<ÊA.뾿NÔ$E\Þ-Ù|‹Éj­{ÌÞØŸÑ´¥„ÛD§ÿcÚ|B{|.Øü4Ð>ßPektêf)¬jãGN>+®2Ýäæ¡£uô²–±#‹i¦«ØîÝñ•üîqˆŽ1©ã‹ùH|þ—•‹ +‹äî!+­Róðå¶»ú†x>”÷Ó(á‰O„ïm-—¢ÎÙ“¿béá/Ü;ø4+´uôûxyŒU nsDå_ÿPH;ߤ?mP†œÆ}Ô¬I×™?¡ù‘‹_Fûpy_˜—Qq½ K¯¯Ö‘Þ¨Óï¶±•èJŸbM¨ó}œWÑ£ mf·2pŠ!{;³—ßZ1¤ã”ùäù”š×6ØN¤ íÚº¥ÒéOUŽ©ÞûDôl ¬¨\÷êôR›Œ¿'‘Qn÷;X¾›]këÜAøq«,yûA¨ø‡o;Zá+úÄ,¾YjµK:$œuZ“§“vÑV’AHóÖ©âo¢2NŽ~æ™¶þ(y·wàûÆÝ>IpcœÛV ¢1°.•5X»žT2KÍxþ,­S,XNZyAŽó,4MØ4WpyL½ÞÀÆd°fÄ4T¯;zªœCò|VˆÍ,pì[Ø-ˆ±»‚™Âh¹ºÏF½ÎÓ¥t®?™­ œŸÜQuyÙ%@smÅŠ®S‡XûÑMx xí—óS ]åÌ}9œšêž>l’CºŽuÐ^KŽqP\!égëO²S)½žS;Oí#Ïa3ÁädlÈK³ä§XSO÷ÅyûÝ ^¦ëáë9÷·n¤‡_O“ìÅOš„-/î¿ôÕÑ‹°ïã¨TîT5ž@WZÒ=S±Uõo%! ¥óÇꬾ†!‚d’Ü’ú•¶ÁnsYl?cXÁDÔV8_“c ’É;>ÿ—QžÌ‡úÜê”À-Ìβmù™Åó:hŠöÁÕSû/s;Ø<ºÏZˆbsîþñÁÇn‰LS[2[/3j7²¥¦©¬OÔ¬Nñ0×S#Zßô,¿`{™~eF«v‡ï“›VäNLñ”Mƒ¸ãNþ² GY’ìcŠÆÆ‚²ï,u9»¬ hJÒÞ©ßw®¬°YÄ»¬ª©‰×,1Ý.*‹Lp¶GÕwuÈ–1Àeã3mQû1Õ‡7ínœþ2'ÜHZü3Uêò¦r£þL®èŸš7Þý“åŸ'¤-’Öw²[Nw4[?hFæ<ê4*Ó°hS×h—ð-à2ÅS’‹Ä,$VÕË^ÜYåFðmU[d„äžâF,Ö˜9uåÙÚiTøûâ‡ô*ézÀÊR jÊ&¯¡ƒš)÷3Gx¤ïWÎiœ£{Jà åß>tŠ`Ï«ãÔ!°‡JÄöÊô¨…œÙ1ßûº†E'^ìtïOì}¨–‘ó6>#2i„¨¯úö¤+dóþº0;€'’ã5á’2ˆF5Îé›F½ÈQuO.Ö¾™R­-`ÅɪìcÇNü™vt™×Pö‰å c Ã7þYª¨G¢>Ö²8î«õO·zôu:/ä¤kžªä‘øîrù\ÚÂ}*^§š»ã8¡5b‘éw`'ßO蹩{’58øz‰,ÌÍI:Š´— ¤wè#ŽW¯°'—á$SâôN^„o(ʼ áëO†:Ç… ,DFƹE¿Ç‰lLÏcŒú_x ÙA’<E…ï1Ÿ/0÷³ÕXegçN³_Ek.y¨>ÄRárJE©úv máÎ/ZÄ÷¾AÝÍ_  #oŸÚVý¡g^húÜ;­³åóôƒð£¾79¢ÌPA Ú‘ZÏn¯¦F’ÿ`èhó(üaòþKÉw攣ۧ„´›MR¡ÈÊõëg)R^ǧs!fꉋ#ŸN„AÃ|+÷›sïKƒìVÃ^ÊzÜ9Ìo JkçGÆÛFÑ‚ôA°INB¹µ™©»¶ºjþ_µ7S=âö£…;wávôwÈôa…jÓ^!ܺÜ_uy}’¸1Ö±;¾.ƒ) æoKµÍábÐÑ7»£^iïp(‡-RÀ)ã.th± *ŒÀå¢ïô­ã›’ z 3Ïab Kðà¾R˜Áƒ’lÅPòq{¸´#ëÍqˆgÁÛWçÅk޳cmƒýæJB&F»G‡-¬ñTušà'ò®qÛ÷¦¾T›Ó#~<Ÿ°s;û´²„\ò5;ƒ;>»º%Yº(öSpýEéFV‘ŽÙ¡øE)ñvÝk½ƒìe Ÿ"_¤ºV³DÄ]rË ®ßWéÞòýÈ.N3¼&Ǹ½æËSÂa¨:× pƒ 9ï=hë%ì¥÷Ç8%[’½<:ÎJøMzÃY]÷eñ­Õ(X½Ïä{|m¸ëŸÁWîaG qŠ¡~!¹‘¢T¸!¶ŒúÌùr¬Vã{E3Ôü¦»èÆ}àcEmë³ ÃÒhý@›'§ÄcÏaÇá†3Ÿ€Ìøcþå¹*Ò‹]’ÅÎÁ^åëî!µö$ /ó„ò;p¸¶áUÆ&_iÊÎuïcŠ_ÉúQ©šŸ)ƒ˜ Ö3E˃ŸÀØzˆ½Û\M@1ZóR× ¤N[ÁÛ¾“ceeYVÊ•)-¡Ø"‰(™KÕ}É;CÙ“ÏäÙS÷ZV—2ä–£éMÂÀù£*k –KÕZÝG#9¡æ8Ï^+P`xø+Ž%)VÑÂÄh9B*Ü®ÂÎ3Å¿ã9(ÔN¬¿'¤OaÃ]Á-¬³5a:ˆ×ßþ™3ÖÞ&äÔhý ¿8·ˆÙ•’öоéb`¦¾¼‚ߦۇûU‚CáT‹¾CFî†éÏÀ¤¨Ø-®mAH[þ9Rï*8,¦y"ðlxKW¦ByT‹1‰§m­Òõ³#æ¤í¶«, 5³‡of<_¼îø0^ì(¥ "¤ªs¡{èfxPÚý™æ˜®ô½ÄŸ øªs!n¶qš-À*‡Üïš8îrà‡”-†Ø‰5¿ŒBt:+›Ÿ,] ŒNr•yúHlÁá›ùÍÈ^“y˜C‡OÐ^„ñíÑ/¼¦Û wpTp7ut›yˆ_Nçðö£§X”Zä¤AA,ö†ÞKq×WÄU’ëÖÆR1ª/¿í3†.}ºch5K&OùhÆË²¼+®˜ûèlQSù²·=ÈŽXÃVŸÉ‚<60A³Ï…SÑQs³:pî•Èu/¨¶µŒ¥ a§¬ƒA!å´gª´Âô“ Îâá óuR€çiÿx¶õuLàhhØWâH*7CòU_|¯,Èz8 wªºÎ¬¾ô ³rùl¨UnÅE [+u½Š@~­-Á¡\ÖÆòÎxá1WBå• ´µÇ×(×Xúú?›° endstream endobj 191 0 obj << /Type /FontDescriptor /FontName /IXNPPI+CMEX10 /Flags 4 /FontBBox [-24 -2960 1454 772] /Ascent 40 /CapHeight 0 /Descent -600 /ItalicAngle 0 /StemV 47 /XHeight 431 /CharSet (/summationdisplay/summationtext) /FontFile 190 0 R >> endobj 192 0 obj << /Length1 1809 /Length2 11289 /Length3 0 /Length 12440 /Filter /FlateDecode >> stream xÚøPœÛÒ ,¸C€ ƒwwww‚0¸»»Kpww îAƒw‚»ëeË9{Ÿïÿ«î­©šyŸî~ZVw¯!#RP¦4²1‰ÙX;Ò2Ò1p„ee% Ìt Lpdd*fŽ– ÿÈáÈÔ@öf6Ö\ÿ²¶ße"@ÇwCYk€”“%€‘ÀÈÆÅÈÎÅÀ`b`àü¡=@èlf¥HÙXƒàÈ„mlÝìÍLLßãüç@aH `ääd§ù“´Ù›­²@GSÕ{DC %@ÙÆÐ äèö?.(xLm¹èé]\\è€Vt6ö&|”43GS€Èdï 2üQ2@hú»4:82€Š©™Ã_ ecG =ð.°43Y;¼Sœ¬@ö€÷èeI€¼-Èú/c™¿ h€‘Žñ¿îþfÿáÈÌúO2ÐÐÐÆÊhíffm06³äÅdè]i@k£? –6ï| 3ÐÌhðnðgê@€˜ "ø^áßõ9Ú›Ù::Ð9˜YþQ#ýnÞYÔÚHØÆÊ díè÷G~"fö Ã÷sw£ÿ»¹Ö6.ÖÿAÆfÖFÆ”aädK¯jmfç’ùÛæ]÷Ìä`eà`gæ`€ì WCSú?¨¸Ù‚þT2þ!~¯ÁËÃÖÆ`ü^ÈËÌôþçátí@^ÿVü/‚cd™: @&fÖpÿxƒŒÿÂïý·7sh1¼#€áןtÞ'ÌÈÆÚÒíó?[L¯¤(¤¦ NýwÉÿU Ù¸ªí †PŠ}Nþ^¸ÑrV!DÓ_üƒZ_ª™DïZå¾…TH´ÛÒlßÔÆ³“t²¤Öä` lïkÅαWYG”³q0xë—š$ÔQj¬ ÊC/〼ARËÌ•[Š Yu&-ŽÓv¾¢ÎÀV_ TffDg³”*U$ú†ßC 5o=ÜY^ó®:;…È—ìª#¬£Ð"Ý7¾¯x¦ä>ž1Òÿ¨7L­vw±Ú¼õ¬vÝÔ‘²²)ôªto·b=ÙX@ýià Ì*=ì ŒÂ¦k #®\Þî½!¨1+‘Ë)¼öôic†s.ñ‹|"íî†ÑÈäÜ)Úx¼;mXçã÷C_Ú©WÉù5lsªä’6³ÉBj9ƒ£¿„ kÝÇÇëÁÆr{:iFò Ì qÜ[ÒÈ h%f÷½´s Ü0~ôêFÛ›J ¬ æ€TVšÑHtû„Æ“áPÏdž(¯Û¾ЊBÖ­šÙ´áÁ7EŸSQ]*le³¾Ë0AVá•k»üÉè·‡) _8ZøG]‘¢sâµR?Ä!ÐÌHsJwC «ÆÁŒ™¯ÎOh EôôD0t}ƒ1E\)mx°§wÍ÷üãYÅfŠ·€ ½ùÇè£[ï"‰-:4iªê°È·Î7“U‘é+3vfÓçaMñ)Æû7C5Iœ6 ={Yâû0Ç #_߀Õ2¹£`&C3àæ-ËÕïþ¸}j›³Ûrµ©.U„ôb­§öõ/Í0ô\Ù£|fª†þpgV/º²B΢O5ÙÄlÀözT$æAšW [)“—téžZ ÖD[äÇå1úb¶nïÑ0†K ¡¸ÐÁWËS>—¸5Cž- ͨšAHWg´t/ß1ZÒtât "Šw E%GîïCHæÍý™ð;r8·,•3Ð,©×ëßWÞÌ©ÒõM¦CËjë`)Š~súÕ("ª Ûì pß÷ ƒq‚ÈÅ´("àòžƒbŠø'°£ìÁÂORÚ[õ®«‚>¢NËè¯ýß.É.qz`X½‹² «"¯f½Aø1]òÐ0÷ Ö¯–v]FSík™Á1‹F–ÖƒÓ˜9Ÿè¬Eo‘Æ@·LLgã}îP¬jrÍmú6@6ÄòQEoØzúÕ-½¤:8øöì$¹’nÄL8á©C¨ÖIÝÙûìÞâ·*WEžˆ› >§Ê(ÐLGÏ Î=4%ùápRÌ=´¨ÓWõ½48„å‚1›«ÝF>¬\‹W­¿„ýw*²ÇeƒD‹øêgÍZ7ןæU§É®ùuÊ)Ún ”ÖŸø[³Ôöý¥q © ‡?[ƒr´°ÜWÆŒSj}Ñ’Œ%©VµŒqxÁeѰ˜g(ÉÚ/J«Pâíʪå) º´i¿EâdIõÁIPâÙ+ÞÆ4à+²Þæû¨`ÈxÑóM°Z cßñ*ºäÉ¿²c±>‡’|æJ\ ÉSÕÃ’Qàmù$'\µ§,{;gv6‡¸Ä—ˆ„:8yº‹°{ÁLµ`N¢Þ—)rȱAz9ãy™EÔ$¾>2ÞüŠ{°$Ï­çg¤˜$%IÚØšÔ§9Lñ ËJÉný._J ;ÎÛx‚“&©œíûÖ'=ÆÇüo|ÛQ)ž2 ?c4 üÕ¼àìÀº›%…ÿ¶Ì‡–IǶBˆ3Õ´Óã}Ønœr+i*¬§ ΘeÑðBuyÚ¥¤ùžHä‰Û=l$ 41Œ¶{ÁK/ ýô A ¨¹”G&cØãX€Jm¢šÌË4«›º´§»ïôÕvò¹çR£ãÇs!£ƒ 5Þ*÷Eæ¥n•y)ïö=­ºÎ†×9fýçœW[kûUY5õ²ô[ºßèš´Žû©)5•uLî™s±…`!p<ÊlŽŒ –ý«úmi{á`Ù­`’GXSøz”™_S¯fõ@(¢HÊŽˆ…ª†¬Çã›þ6¬(C®2"ã]‚)á>/rEø?•`2'ƃ®{L"IþcÖ9uKóËf+Qe“S¥‘ˆ}‹íQ[tök05W¢Z¥´/MÇ7Hû%~îI»ÑöUPȇÓ¤…Beð2àôá1¿¹š…°nók ä¾âŸ³¦'Øtôýàâ®=ÓZã)j±X²×‹)Î…ê^hQ-ÚÛZ²“, 7¯ŸÚ¤ ª¬"VÖô ¿ì-tméFƒ¡!UÁZÍŠhATFuGÙzžW‡ÁCk¬Ä ¾pX º/Þü΀4ê”ã&ÞIõƆžŠ+é´]o¿MÆlÆ/Bò—û™]ó­Ý“ ýñ” ¥[pýnvH_§Óœ‹üØ×ïѾo¯ªEdTlŸµ]5 ímÆ/ÅòŒî*D’ÚB˜Wúv¸ô9Q›•¬™UH•Íf5=Ï:T FÏö5[3̓¼òà´{%„?ž<¤ÒÉlîÚF`=Ì@ýÛ©<¹ÕŸ;'_<{é<’ kB,ï¾å‡Rc-ùäPŸŽã5ã-~m~µ%_ ‹‘ëIEËâÄ$¢¬À(×?­0CU8®+횥[’2¹Z½§rùrüÖ› mó²ð–™ø}RÑ‹8*ÎÛ¯[núëÒ¢:½Å –Ã`&ö^ÀA÷Ó½«ø]l0 ò¶ü]pÚ׈Sóêm1 ••öŸˆ4¤À³\¥åšGã„úü‚_ÚV¾†Œ6µÔ{>0!c{ë¾=¢‡!g4ÂUvW}äÁ/èG’ù郳)öÚ˽þ?Þ¶ž”@6úÎq›­öFŸlù'g«ª3Œç„'0¡kKç[Ô¡nmƒg¯, ó&†sQ(udwpP‘aQ0nWbtý*§æäk-W'ºñâ]—躳ŠFúḓ¾úž¬1ŒZ(‡ÛFwžƒ›Û;Àu„êCIÇ¿ZD´·Œ‰‹šÎ’ïócòÑ7ª>ë'fn pF¥Øh ÈÄU’u”á‘Ý»ª¿àRõÜ©rÓŠNáîÅÛ´©l­v±îT7u=w0%tò•K»Ú1÷nüÐs,ƯL?ˆùé9¥¾@Ê?Æ»'+PÝš@wƤ»aÿ½+›_‚¯´¿¤0¶ú}–»úî×ÐBn —•zÆíEƒì¦à„»ƒ¾ƒWLVäâKŸC³†^¼C;=xrÖ¶XhW ²|œhoy²åiÄš©¶¸¾Á‡[ GAZ)ÅÚå°®*3zµ]ZýìÍÙö•0qù›sWDóù+*XJöãʨ þLÓ+]‰W§ÎÐgn ¦ª Ž:~Ë®ú:C ´W0…»k"Lþò!­#ã·_ô¸ôC±<Á‰ñ+/Oô»ëÐS 6é9èÑãƒÛÇ…âd¨a¶í?Ž›d/¸f­_zI •}>³•Åeµ2$†o- %5zïˆÀÎpp#øä^Ö¿]9@âgëÖðšŸçÅÖû­ò?ì-¸äŽÒÓ‰o·÷~ÿ§["¾:ŒÞñ¾ïjŸM¸”]Üä[KÊÃÏ\ÈB¼œ¤7/†Ž°IÓ©šØbo‡Mí‘oˆ` AÜû£H¾Óêt%FÞw⯠¢)›°`kŒ;…Q®f5ájev±cÍÕå •(ó¶¿M ä>¸¸é¬Ïkg¬80ÙY¾ògÑý*”n|¼åö¸¹QiañG îÎî.–Ë;›EÙ€î´8Úè7Úpí‘4å (-Ô‰Ó±¨Äu{!¶žJú?¤3­âT‘ †@¦¾hÁÖ$ÃÍÅ:½¸µ¨*ùYl@ļÿEÒ0ö¹|ܯ+8г]Ca6G­c€âÀ}mêÐÊùã[Ÿœ#UöØEš! ˜Hjín¼·£Åìmzæ! >Yqq#”XÉÃY*ÂýêIîAùlŠª`‡Ƶd+O¶"ËÄLäP)´¶áðÒgÞ5aÛ×ÌD¼N¢WŠ É0G"<ÜßÍkúGš×%w˜0ðóןëM‚ÅãÇ唇 ™fŸ[ìÑè‚W·Ùç„¿é ålFì×`ð®ÚÇLé']bdXÚUR0é¬v¥Ê>gJï'0|Ç?æ/ž¶eLº»€×J¹DÂÀDÕWðV~#ä•^}ã}Ø“þœžNþú*,ß^›” –Tk!’ÞuÞ–¹ìhKS5·Â.ª \?ÇPz'_¯Ê184`Ce뀀÷€(ZxkiQl `­H}5Æ­iÆä6\¸ØèEж/}ðKZM%€¤œw32+ÑV”«Óe$3k KÔ >>»¯¢Wòæ ¬XjHÿüG‚¥’ëÙZÜA D¿+,ëÝ’i L«ˆÎaö"Ã0¸aƒwO –T!Âvpó™_!—$Öà°©›ôjý«7æ[öÏ gï>ÏËò¦uØYTôkzû®—"Âê3Á~ƒÃö¡üú§Ë 2K÷k†4 ·Ãïxa9iíÆÕË=~Q’ %ëé»DâÃM ’Õìë8åœGO)xÊ·k mÈÏÐ^4 ÓΪü"<”¾£Ig:é/§‰mçq[kâ?‡P{â¡ÀšÒTH÷%eo7Ý"ÀÒ8°Xèeñ=kšºH²yõÙp¡À¡bÑ'8,FÌòö¡y 3²[>Àmgc‡Rpó"¦CŠ/ûø¸s °+ÛÛ’ב†YÜìŒ lÃ4¥De}C{“>aŒÔ<ŸR® ª·RáeB8¦j’GB='g;å (×í~%:Â&ÉgƒçóºÖ¶@E,!þ*ÒÛà}‹Gy¡N®ž–çÜ ôÃÎÕ†ßtÓE¦áuÕ‹ì)W…j‹î^Ïe“‡ 1ZÚæàâü½+Õ‹§õíÚM$a±ë£•z‘IœAð-ÉÓ…UR¹²Ï ô:tÐÒIeªØ3 ï¹ tœ`mÍøÓCCw'ÙÐßF;Ná»Â'^ÝPaQ±7žILêÝ+ò\c3:;Èû¶¨|ø¾l1Çšl¤áð˜$ñ­6ÀÈBÝ–}×ÎMišgˆ`Óž÷š6Ã9òY|ëF³åÔÝ0uöºz’–Ò üÂJm~¢B&-i[Ý$¢ôëý/}-r9£˜Š—¨ÎE;sîz±2¹l³[T¾¦pA¬ÁÃvsQ5¬  ¦t>Ü]¯266_”mßRÝ>„šåõ^vDWG%4ÕûKŠs5þ¿¢œùFÐϬ‚ÍŸøOÚCÈÀëzHÔìSwïaTTðŒ·Ï2¸Áì³…•âÔ_»G³…"k×ûy›a¨ÈÇüTØË9¼}BbHR¦7»,{91 ë:šM¥žšŠzrðò(Ì£Þ v©°x=#?ÓNÇ3  0X6ÕÔ;𨵺€U 1N'ÜåRF‹á\&Ó–ô-@õq·a .\)ÿnT"êÿ4BÞ‡áÿ z°¸àÞG¦…ŽŸ¡~•vn{IY_̟ݤ’V€UÌcPokëX×´6]㨢.¯Œ7}ÎB'ïJêçúõâÛ½F3NNŨ ‡´8†Þ)Ãw"¿y›k÷"õ€ó±Q3®!hõvêKÈdéïí]Ïó*.¿KìBT*:MýÙì—/eâɹÔ5ÃûÛScÛ&e 2F|‘a!àÒÓž˜ç´O•å¸8r󱾘@­Ã¨ò¯QîÌ*M=-ÙD6ù¦ìó…³!n÷yS¦ÀÊÇÁ•—i¸°ŽÓnzZ¤–P€pHé±×¤Ê;Åá›7â †nz²ðdolHLm§TÈú/w?Ð*åPðRM<ÂN—3±†š¶ïŸµzJÒ±:õè+*äßÍûSê_”&+*5P<¼c‰gÅNü®[°è!»^mŠÃ*N¥Ï?®$&[‚¸Ñçû:7f²ýíжÛxiÜ⬓uÙ¨¢ %öDºÿMKέÙ¶^¨»‘´Oà NèÛ‘úÑÜDÌ£(fã dËÛ=Í“¤JiHÔë„[úbÈiü‰$[³–½75CÓ,lKòåõ\±¾U®–^ù‚R’áp"®°T° 2Jð@ìítY×ø—á!b=™äG~¼^#4ªG®y†3&vWŠãV[ø*Žv*<:HgVü,‡Qž›oßudƒΨî¥O¨›ø@t´K6e3Ñû¯Â8U8ÈI¹ƒ¾–#(gƒSÚÞ8JÔ˜¤Î}6I"…©K&Áë0¯n~‡9û@ôeOëìxJ>ƤÜ‚Òüzž¦Î"¬uĘ Ž$_.ƒw;[Âè>ˆç‰žÂ(KMª&Ž•áÊ! B~Ýš²zØÚ&úasÒŒu\@œ ñ"6ÝG{I™Iª«‰´Yëi P*„d}ãE•79¶r¸SU0M‚^Ù V>êÌúdØ^~QÖ–xÖÑ,ÌØMcZË@"úÜGT±wÖJЕÓ­WÝ0)«¤iÍáÝicʠѳ° ;ƒ†ø¤Þ¿/µ¶­d¶ŠTùÂñ¹ÎÏbÑdá«•Ä ”dò¼ƒÈøU˜MѨß³§ÍS»ðOê…ÈÙ0xI -¨‡}Xe`³o·ªìÓྲ_/ÐbyšÒrÆÊ.|‡®÷ÓbŽašy é–scÍ/õ#•©ô$É)¤=nÚ)õ"_…Õüù;I4¡U~âc¦™ß£»Œï눺þmmu½æåžÍ+Êðä 7’vÒ´'Ý[‹HzxPË7>‹iuÀ'ë¼áÑh #Ôç§y0Öý‰À„ˆ¹¿¯ÁžVgŸY¥¡!žÞáT;?)žV£s ‡ÝÒv:xˆ?±ñçgPW-ºŽFSotËd%#V m®’õñ69A›Pc9A£xÚçÕ5[«}R±æ…:¿¦Ó/g´ù´zyûÒ …[¹{͈)cXôÞCm· °m_GàÁ&é¼ ¼.}e\Ñ (Nh›OR‡^‡Ît¼ð?@«®MãT9ETQ ž/—-¥}2æï¡–+l¶‹•«ª‘ß|ÊÈTé‡Ì1Dü3ƒÉ¿g6ùðt$du'­ðuü_"¹‘b€¸[êGØbÍ!JÕM.3kìA}í3¯‡zìÚµVM×z¬n¼ß}GDìÂ1Íp±1D(£Q­—¸ŒÂT•µßîv7ïLŒÌš°Åd= k-aD$œä„ÖW¡‡mþ«¶¦Kq8[­âhŽ<]±Â›6Ä t"¿KOª§§^r¤àqÊÌ^)|c×Ù Ôôõô_²—i¬yʺû=žyo¸¹¹ÂÐJœ¬ ÷4ÈfT¦Zp¿]˜ú°»¿l§¢“‰©œ ‰bÅ®c#¡Žð 8(A~\à"÷•è€QË«ø’o?eèÒ'ÃWb7!žÁé•]Ì¿j° ]÷ªæC&lAt0Õ/uôÕj˜;( ˜P?c:­*M;Ó73†¸Å>‚7ÔL¨>Öj¼ú’Žô¢ýø6tеEÙ bæË" ¾¥S¬Ã«âP)3Æ{Åÿ±S¹74oî½Ù šGJ}°â¼\°øÑÛ ìQ`ºi±šÅ=Kú;4^‹)ÔâAà A[á ÅÓ82É,VæuO®ùÏ®sOÊÒÖºöëô „·Ù¾µsI¦˜4êNŒ´­­«®­ð5ûÍ;]ÚXò± Q˘/? øQ¡]—ëug®¢ˆìû)z*ŠÖ3½Z¯[ð¦2¨iÓ=.î³{ñ\ì:Íu뵟“T —§“~¯ºõZ~ú¢™m£µ1ùÙŸ.Ñ÷3šº €áS(E²SÅõ5(ƒ¡… ;†û=œ¥×¾DQ&'ÔøÆKÊ@ŠÎ]ŽqbFŸyW ÔÛ£˜mÞŸ#× SÈ}ð GNÂC Ž¨ôð•ÅG9óUò1C‰)=‰ 5ÔŠGVF1œ[Í8EN°øje­žÑ¥L⊬\Fœœ”àÊrµ>pdv¸VÖ8'3®µZ`a«Üu2‰ªòæŠ`Ô/Yørž~¿ÿÅÐÂ>u]MñW¡Ò¾j)Õƒˆš?'¥‰•ÓÑ~uŸ /¹);ÀÊði¾ ¡cÔÚó£‰âä8z#¸ŠN¾ZŒJ&Õ¾Þ¥s‘ä=xôÅÙ—*h›i[¾@ÑY_ÃH™!Œ$ÝYl¼¤H-Pjæ&uø]Ô_ñ‰#ÉÀ$#màt*¢Mè#³üÒV6#»ïûý«*¦jä_éBdŸ«Ë¤Oe¤\µEëç'ÿùam)‹@é9—œì³ŸÁ|!ú e÷ô:úµŠ‘®%Ö¿Úk &Ãd¾Ìùt·Ûh.õIm¿r¶¼*+‹HL8ˆˆ]™ðdªÖ(+Z’*—Ë$Y¬ºì68l½Ñ!ó ùIÒnhû%£”ƒÒÓþÍ +CYµº‡>8)ÓQ¿¤-!K¡;ÎSuNyêV—M`ÚèžØ·¼Ñw|Ö+öV=«÷Bþu/bŸtÉ3±Gòù!é‡0¿.Ó¾.‚XV$¼(©Ã~ î&1„SîŠ*ÞnÛ8Á|\öâ’±É3£}¸ý@6û7š©‹/—+T8`7‚Š¨ç…°Û*Ìwø)óJß067†´ÅÌæ“?Å™,fÔ$à8éVÔòå~sÛ1¯‚‡]sû^ýevHk‘wübû‡pnþO –gðÃØ†mš½}ö6ùZw‰M¦Ÿ+˜ºNÂFݧúeNvøƒ¢Ç£’{ůÊÊcÖ» ëü^÷q#¤3ñ–1A´¿îìP3“¨Þ$RË¿ë¬`¤›a$¨C+’ã<ÜŽ¦ãSë0¨VÆÓ[ñ zD`Ÿ”²îk^I¾æˆõašs³_%ͧ£™\“ÉYè½Få:†£ØùuÀ.Eyrõr–¯XÌ%ñ>ÉS:Y¦ðÆ  ©oÿP‡¢:á;¶ò5ÂÀߤÝÂB0*=üd][0ôQj—fÿéÚL!ÛÂ5óšÀËc5šÞ{°@'-ЛA<´ ÔRF8³ÜtömRÔ± òþÖïÅ0™Ð|ÊέÆ>¡å 3Ó°®~¡·–3 ˆÒ½¢gǶRì\ÍFÁ@?¸gE’K ì-Б·6$-"ô1ɤ3>¨àTõ¢m„’¹/V›So×utÐÒñ?}„aÎà‹@ÛU¥?~qà“7CgާPÌ<{ ›¼öy}€ËxK‡¦ÕŠ&έb>ZÍÌ–T1`0÷| óp}6=&yýH°<ŒÇu.œ^»äU¬ºâO@Œßót³l†=+DWgÒaT»$‹‹5'O½v‚ºa¸QÈî‡_±HU(©]½ý¼¥Nͯ/>ÅŠ|Ž‹þIà ¿Wúì4,.èÛô‰2 ýäÔ­  á­ÅøD£þ¾aµôDÖ©h—´óõX@Öpç"s?k‰ÂA¨-Éîy´e˜¤/9ÐÛŽÔi'¡+~sê’ÎBÜ B˜ÚºŸ‰Ø1ô©£˜1cÜDÏgÚþ„"­RÞûeút;§Ÿ8«g õüS£÷nË]ŽÆm¾ð’*1­SÞ[;Z ´Û™(ßI*Ü×õï¥þ´öØÒ$É<óŽ>1®ÀN¶!9^Ô¬ßmà\TãÞw‹ì BctM„ÞÙP p§×báÇH±ôï®n99ÑÒ/´‹ÌÙÆ}ác1o™ƒ°°$ƒDmO²„ê=Q¼·Õ_e¬Ô%®<íãÆÂõvpöý’5†Zök”BfB;¯ƒ‰ÿüF«®²Ÿ ‰ºû¥‡ ÚŒê ³*×…S‡õõ ³³RVÜ;øc‰çD¢ök¼Òšrq›ºÁϺ.‰Âs¯®Jðö`QœäËMâ+mÒªŽ¡Ëc® `šbb§Ë».Uæ‡Ñ¡D¬á¨í$. QjêËÜí~9·Óñ¡q9öM¸õ %‡M¥ÊÅœõú þUp‘•æ•Üì"qZv³¨<¤B ®iÀì‡;Å ·ÌöF [#|~iS:=œ œ&TÃô­:©aÃbU¦˜F½ûÝ™ÕH_@ˆ|)ºvÁýÈg?œ•K´–EʘðÂ{õ7æ1ÈÃ]0ÃùGI.®0MSË ’‹3­¶õÎÕꕈ™P1}Î+èTÖøß¡<“¦Û_Tì Pg«ä2ŒåÉŸ±@_~:„FFÁˆÔ­AI6¸ÊŒ¡ö)h¦G4 ²ÿhLFø „© ÿèLT‡/mêBUÖåKÈŸØÀø,Ûúá±ì~Zh8¸æcñ¹è³ â*Š ÉPCÛ¬ZÅc„4qÅ''/ôúVº“ÉíþÔ’¦U®ôæR°}öcØd•Ÿ·÷ÞðÖ_¢xøÁÕfZgÕD7ÅH?¼ÁävSj=2Å¡I~´:O[²îÙÙ'‰½aZ+„Ôabqt Q…}.…´Æ‚ÀÿêhxáŸMøÅ i–ÎfaXHÖt fôQ»Ûg ÷AðÚ úîp÷[–Mnà<èyz b^^5ŒFÌb¶£1 ®]ŽÇ}ÄñÕo¦8`Éå—¡¶:ªŸùöÜÖuï®ßîaRÃÛàÂò´õPæ›·Þö&L°Ž¡E_b·ûµl‘*™PO‹¤ï•,5ûïYŒõäªÌhh]ÖE·³!+Q⤌h×4¹&·»›ª»¾ãFœù‰°wFVýú#ÉOæÂ³ÔçÀeLR§Œ*½„•áܲIý8it›e 3¤Np^t9j3¢ŸeÉùb*3NdÇnv[\["b ‚T|C`bŽÚfߎÁdzA™„ã »NˆÎÃä“n›fÉñüÔqì²Úà9K8ô˜@Oýº¯J:Ìà&´¬›¡G#äH¾¢Ð¬vkÃûB7'H3zVžP3¬2D#܉4$_¹´Ø-ä¤{k£b ÅŒ4Ï Ç—½NZÖÕ{é‰=N3Ø¡”;U‡ ¨›yZìQMñfâ¿8„RÕƒðÃ&âñg~IÑ9é×ui"¯¶áàö-fÖQCügHer®5V¦ƒà<ÔO‰¸ÝØŽÏ¶ÌæÁ€¶4]À20ËÕ\0øñ.ÂulÑ-w'GçYüµ1)èäÓ”ëùÕ­y\]8ÀÌ÷™65ЩŠL²ÏF—b7{Ç`*g¬ëq„é’f3$¡ÞRÁwÏ¡©åVœ‹k"tD2§ä^ÆqÀIž³?&yy‰¤©VêlÍøÙçM‡­{&Ï÷êçÙ"–1¨" ;˜9‰˜öy6×<¥ÄÓû7ráHÖõÅ—8ù,™)­¶Õרž.á*žþ{ðO©=S협œÑšvÔÞUâM1˜´kÆ\+¹i“ÓR2ü†QÖ +Y€1¸d{ê¾!Ñþ.h(èíPÒšôi°_å\Pn¼)õŒ!ooîÖÖÊר•·“®hÉ@Ÿ+#(ñyá­3 _)•ÔÚµPMc,šžINݾJÆÆäÙž iÑ©{äf ­†ûR%ê5­ƒ·+CÅnoŸpƒÇDõФ¢PV#í Â"ÞHû+qÚUp$oºš´—jy~•&Îé´6ãmL¡$ËßþFàÇ,#¸Þ-t?m<–ÛrÔ‰~}í+Q{vnDÙ—0VF~A4òÚ0hÓÏæ\j,@f(Kꖦ͆¨w™•ýy6~L~ÝÆÕì®>W™  öá×òèQUåÑéšY!¨øì d“E¬ÌzáH‚¿W<óœ}\ãÎÂ&à 1P%Α9~"=}£ùŽa^!ÁúášÌ=ňƒo°•YÁÈhÀP3 •½c(cƪÑú¤6z®EÏë²UãµÓ½ŽVo‚ãPíX#ÔNüá”»¬V!·ÜGÌ|lu‘\?¹z†ÿÓjîáÞ¯pñK¤pŒÂq¾="üž ;ž ÷‘Ùc “'ê8a hÔÓ<ÌþoŒÊ0úØöJ×sœÞ9~jÑvÖÉU3ص"`m­Ö‹ã£D_µ{øq¢‰©›Õˆ†ý–ö¥üˆÉVq&Ï©E¬ºX7Ç(GŸéïRÁÌ%wá& °Ö l:yVQ ëfbß ^;Ë 1£ÃË/ÖÒÝdÚOñþFp°Ä àmLtƒti›sš®ÒžŒù¼ ŸÛs"À–à×!êe€ŠZòDbEœŒåæQSýˆ¾6ôv~—­©SÆÌ›uª»™ôsQ‹ ë!ZMãK¨ 6]s=‚EÌ]þå´@‚rÓü8øŒ^;î}^ÄQº9_’è 0• xØ@ÏÁ”²Ð šn­ÏSM¸ØVO ¦Aq×ð÷)‡ß4µ._ªE—•˜$ŸŽV$ É„+û0On* òÙô¶Þd`-Ó Âè®ÎZÁâô})Ý1j…¤é£ëúÖ+> endobj 194 0 obj << /Length1 1467 /Length2 7111 /Length3 0 /Length 8101 /Filter /FlateDecode >> stream xÚtT”íÖ6)%%]24HÍÐ% ÒÒ1À3ÄÐÒ’Ò Ò%)HHHwJƒ€¤Hó¡¾ï9ç=ÿ¿Ö÷­Yë™{÷¾î}훉NC›SÊf –‡Aᜠ. @FUUQòr<ØLL:¸#ø/56“ØÕ ƒŠü‡ƒŒ+Ø~¯“µ€ßû©Â %wGˆ Š PøoG˜«@ÖÂb På(Á `7l&˜³·+ÄÖ~_æï#€ÕŠ äør»B¬, U ¸Øé¾¢•…#@fýÿ‘‚UÌwáæöôôä²prスڊ³qÖÃj6mjùÇ­¬…ÁØ[¨×l–ú>ôn×ÏO„«˜jb¼'¡dF¹Ý÷ >¿æ6KæAñ½¢æ] ,ÛþÜNÈÅÒmžŠÅ Ùï¸3f;Z{ý¢œ÷XM„o]益ۓqÉ%G#×ͪ±kg''¢$—I®´Â–ZýÅw×Ï4š³Ÿ¤»­.mÓ§ž¾Nº£üjü(ÎÖp7F-Çw‚ôL˜|‹'sy…NêÓâY=X1IPٓߎz˜]×\m…òªì(A챋ïüFßEàqI\Ð>wj$õ#´„Pâå«„P÷Ðî<ýbŸnÞ-{VÊ¢He[h@¾±Ò5aö‹A'Q±UîÅY[‹Ä½\˜]6v†ˆ¾ÇHýà2ýÝÛ:|6ûô¼)ÍÖ—×eÓž ™ytàæ¨ÑwóòÄ uª/ž=7¸Öô@±v¹Ý”1*ãÈ~­oBŸo™ ŠÎ¶R»ŽV¼áfâüþšzéJï]sóðÙûÖ–l£Ñä+T•’sÔzÃÑÁtø*\íxrk`³|™×®V'Šô‹ üÎŽáŒx­ ú×ݤˆ›Ó¹Þª|½'D”Ù¦œÏìL.n"U»­Ç™kéHq<”ÞÒ©‘—·LGþ_Å*+½Fù̇¤`ÝcyŽÕ)ZójærÈ(¾5t+ܰ“v2q'‡§}8ÓË™rÅ‹)ËÓb-ƒgÏT®mQ’ËÚ‡ViP³Ú®·äK¥O¡¸µ•Ð .Â2Õ–VdÌ”jÆ8%팲5»üµëjÜ>ë…þL”´³Ž(nF¢xZÚCò¶‡ãÂòüc=LTR›Wƒ¿Dt–´ ®N7Ö:ãmJæ­UDÀïpاãJâ¢*ü;Nð¾šòá~à K£eH‡Ò5´ý(ÖqQl—a|R+„'xãzN;ÜLÙH*@¦W(Ùä§zíCÐ쪔Á›,%J¿˜J%^ õTõ×Ò¼0’œ¾]}K—;v~<¯Ï(Þº¤ nâRÄ+ê\(–¤é¬UIºÛ”•çPóèêµf/Œó 8Ï„¤Y¢4aeÈýšØøžÍÁ‘*LöÕbk6¨SNpJŽô9QêCé`§CÅÏF¦ù¤€_£xŸgè¡ÎœKì=[¶&ó@]dýHm9íXwYoú’(Zø¼ ¬ºÇÉp»_ù¨kcúx—½ÂÉmK¤|+˜©«K†ÚÖFÍÕœýbIÊ´J†Ì}—8Üãª`±bØ—€íÜCW.œ ëÝcVF‚ ›5ÝÔ”-Æm´’Ót5>Ýïq¢U[ЮÄiÒG랉Æ]&EϪ¦¿¹‚°Hÿàm€È‹ùd¾ÚxíðDÂTÛòiJiþzÑ7Bª¢·lBùh?õM=Ü®÷4“iD§P‹™Ú2ôb² ±HŸáŸ„/nfÞ…œT hÕ|’ÈÞ›ñpqÇ!~!Õ8×ìÓÒ¯¶%[CÌOBeJ\zu6ªØŒòìMæö"x³»~5Ä<µz.}œ<§r,RDsÚ¶ä#€Íßñj+¶ s´|¢QìÅ7wN àPdÿCò7 %û§ac9ó2„N“TÖY# ˜Óï¬BŠõ~Þh~= … —Cì8z\ àÙ­K+b:Bl;£©ÈÕgnâoÆ’;ÏJý­¨c6Àç{ó ,3b.k])lyCí«¥=f "ëMÕÉY\ïI¦ yÍ9iP ¿ ˜•§‹ kGZÄ–ÀÎ7â¡íócçÔÀ¡½ú±³§²$,¿õ¸uÝFïsÜÏÂv¾¾Ác>€A@GB Û»‰#v [N´áòá…Ú'QNk~¬Iºx ÚÇ {uM§űÙ?)=–·N&gj]A3²-òtà›­œjÌR­ûf’³^©3³ ‘°Ë¡&LñûÍuƒ^*Z¼*‰EOü “‰¶òz-ðæE­Í÷º|¼kùæ+|-=ëDÓÓ´Ã*,䳑=eœ¯œç½žíß Và¿ÌÁ`ÁØŽ±íð™'u¬+MŽ8?Üåx³I5öŽ¡©8³Y꽌|eFª¡Óë0Sœ=ÉX.¢ây·Ù–„q*:êù9%Â7SŸdžÝÔ¸“·ìÌh’ÆMyËÄ·Vé»-~Åí®YgŽª RÔ¡"” TÚoŸà{ÉÌYfÔ{3îÀDV‡Õ³LºŸêáÈpÆ"p”…óNè<9Wa¥]*x^0«À¤2IÈ 6Ä\ôÙØŒäö¦°19ÄÞ?Gv1Ý4O:º¬ëB"Üògy—ã‡j~Â{‹Ö¹s<ͺµª˜]¢õïÚñ™Ö²—ò†Ñðƒþ±÷ÌO)°Æuê~.q oÆN>í*­È|$“Cjš¿n{µµ*¿n[wÔäk8táØžõàö\qî ÷ýq8†0ÿà÷!+·V¾³®›—uÃê$™J‘~¬B"·à©Å¡ärx½ußµ_ÖK(Í'ZO禭Þ^w..àv)FÌ?äLèÉ Ü?ј¡G‡ÓwŠÍå¦(5{§l÷q™ª¦c_˜/UŸNKÆa‡X¾£tùcãà W©Ú³t3¨Åφ™* °Ç=N ˜QÚˆßÏD¢óåÅåcGÏ¢Üð;QÒcwN¦Ñ‰\tRG*zM[GýÙGº·o‹¾ e/a>¼QÈŒn>Ñ)|MCüðpö{£ìÐúº²H/ûÓkÖÓùÈÝPƒŸ¸õµ¼_ç9!üôŽÅ5=sE±Õ®,‰}Ì ‡^c ›ê£4PϘ…ái9 ‰ó!ßø¬'ŸÃº£ përЉÓ^mQ‚‚9V”ºŽË"Õ/µêßž¼l,j} ,&œ9¦Æ#?Ák0/Üy`¬Çû5。Óƒ;Cs´Šùr£:r‰T¹ø)X›fTÊáNmˆlÿ}™Ú(é!y‹ƒ•ºËy,}t°U 3›âân’dÒÛ7±EØN¬OKøø\\ž?vÚj˜q'.” $éÝ“›÷ÃÞ·ŠJÆÖªÇfæ·r£S£è-XÕâîü¬õ˜!gTk”LO˜ø+ˆ 8¸6·bJ‡õ’`6gaV\¢…ß›¢q¨-.¸cª™¶KßÜ-a=q§'W±»HçØ¸.¦Š"èýR¤)nosªEXòƒR§ó»Êö~oÉ"^=Ñ´/¦m/J‰}<kçVYE2Ù{¦`o0E¢9ì¯Z0ÛøÐÝ–èý7Òødi–u²KKºP6±ÔÐ:É'€ËÂŽ4l|NþK1ŒjßP×)Ûf䡬ÌvÈž£ t±5[©åïý•ƒKBÌô“ß’Ÿì2œýÔ8t®´äTH6‹ ò:®o¿% Ž<ä`È2>~V52âÇŽØ\~ÊÀ‰·]ÜÍÆ³ð¬}§ × ’ )ø9*f¶äCiÕè¶‘’yËFŸ3Y\+*J²yF”ü‚÷¶\#”Ë>°~ë'È÷ûµâkáYœ3u2¾•8DcIÍP%•ÐÔ¹\UøöÁ¶´×T½¤7ßÓ'­†&_„/ʉmo¹y3dVô‘Nž*½™ —‚š®|ú|rÄÑxdOOdÊ2P-C…Û±ùÑWÍá½J—vâ9캃,ÏÓÉQS}„gÏõa¹¦¬ yV·¿¤?ë+úÛçS7º9$ü™ÊôÈSü²:·'Ø´twJWºpGyûC±êrM/™¯­)"8'ÙLŸS?¸ 7-ÜNîA󒱩vÛ›:,ŽÅ¬;zLÇ÷\1(ð[šì(ŒzãÒÓS^`^"‘6Å1æ[\+Bx? ØÙ©‹È ;ÚÚö—’³ÆŸm‰O—Ò”"/ÄgÉ2Å©PŸã^ [È&¤Ì×6LÇIúÌ£e¹LóLÌ_ÔÆ¦vL§ ûêýtèä1ç<zCÅÿP_†îóþìÜ .±t‰§ÖuÅ-Y'óU5G5ÿ.Óý7É`•ïJcµ N;¸ú¾)BlYÕŒºD>‡svs­¸¾ŸÞ!%šhpî;4·äH¾Î~­ý}?hÜ™>0¬Wá޶n³¦¦{nÍM„%Ž– ’ÛË îfVÀ¹ê`ÚÈ_ÏFîëwžÒÚßÉløƒ¡×ô\b$9“x$¹£RcÓìúz<{ì›q œ¦¦äÎÎq‰Ž“¤­f¬#¼rÀˆCDƒKFPuÜVQnM`ÆÓà†ØÔ#§«å1ÃÄ]Ü„åâ Eaô}Ï‘v¤¸eÒ\pÞ¦,CdŠŠÎ5?nƒ „Ïý ¾Êg‡ýkh„(WíY#0c¡ç8ÌÓk©™(m—Ìf¦`kXšcfâ[·ýÓçQV0ÿ|wËK¶½å˜ /kÖKq°ŽíÓØ[…Å*Á`V¿’$iÞÖ­M‚ºÉ×X ‰ÓÔ(ÚᲺӹ=} t_©qNzw¬. ôyãWú<ˆfIµúþÛqƒçüؘ£ Ë\³Hta¡^û-ÓÊ|(¢’™êß¾}Ãi«£éîì|@Õžé0y®³4Œ4Zdä½ÃßÎw,Ô‡nï~=¤¸NÎùÁšõÀhÃõSõš¤‚7 sœJs匾“K‚&|S„‡”jxtë/ãšWñ¨˜Ré‡@ NþVH3g&òõ9õÜgO® <¾—ÄH}ä% ‹yM«ç$õ‡¹íUµ•_Qʈ£LýkÀàñc_ ¤òËq±˜J¸Ÿ8ê?`4Ötx%@%D´€cþÄræ™6úü®z;3íj % 2¢’ z®ûº@çqÁ Ap V¹IŒ¾ã;))œÑç#ì*ü“:þ5[ÚÏt(V…åZ.s ¸Æ¢„&¿H~9¶—/Ó(ô­¯R¡‹çЇ;¡£ëÌ Ì™ÞÂAW# ˆWtu ™ùÏ^é¸'Ï úR3yÂ8Ë×7"¤~RÂW@Qì³ú2\C‰‰¥€«wÁÐîZjæ¹]£HyE`Ûı¦©FV;û‰¾!' *’l3ÏËeàä‡ñ˜Ú¸é¯z…,àÜÁ«NFóu°¿¥P<d}¬Ûº‚æIeq2­ºÐ>ª—!#ˈ>‡‡pzhÂ@¥Ê~œƒEå±€­Zè}­ã`È>Ò,ŒóÀÜH)TÁBã‘­Ñ%+xT˜øphnªú¹{ü±sS$ê™7ôådiÈÀ—̰AUQõWÁýâ¡)?/c„~p ˜[S Agªn%Ý œÙjŽ/ÀàgÃëÈ^\ÖÑ5Ö|œF˜ún¢4ÉŒ»n·1¯­¢W„–ÅIß” nì&Ãy!üüË#,³Â=žŠô™'âEªß¨¿F8bÆ¿ýô©÷æâ‹Œm`^}Y¹po\ÄPá›Ò«I¹Œ…1v"ßÆÃ¾SQ)1ÆÇÅs<>H>§ÓÀªý£ÝͶºÃϯì‚2¶±«ÆëÞµ•¯rÉ)Äf•}@»kýúÆý³mn”©¿½Õ«`ÛÒ§êuo*BˆœÔ¾çŒ"*¢Kw&C7/éÞ†­¦áì·-„ʯßÕ3æíé=’Édç,DrÀUùtwK‡ùyZþS1<ÏQVR…€±àñîÝy(Õ€R`ädÄã»êÃÃ}ÿnî­¹H’¾c'1h.ÏþݤÁ¨ã¢§x,K°0Õ/`FbZÖA„¨Éÿ€´¼G endstream endobj 195 0 obj << /Type /FontDescriptor /FontName /GYJBEK+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 81 /XHeight 431 /CharSet (/f/i/j/k/m) /FontFile 194 0 R >> endobj 196 0 obj << /Length1 2831 /Length2 24168 /Length3 0 /Length 25749 /Filter /FlateDecode >> stream xÚŒ÷P]YÓ   œƒ»»kpw ~°ÜÝ=¸·àÜÝ]ƒ» ®AîÉ̼“Ì÷ÿU÷U°Ÿ^Ýý´®½!'VT¡6±1ŠÛX;Ò330ñDå”™™LL¬ LL,ä䪎 à?bru ½ƒ…5Ï ¢ö@CG°LÌЬ'gc v˜YÌ<Ìœ}—‰5Èí·ú_ýe”Ò’Ó—¢ý;ãÏDDl\ôlLzv&ó¯!ã?xý×Í¿ø_òI -þ îRÖ¦6î¿sïy8ÿ3Tÿ¬ 5à¿ ò6àY¨~¾.;“1øóÿçøËäÿßÜÿòòÿ6úÿ7 q'è¯cª¿ÎÿŽ ­,@nÿ(€GÙɼr6àå°þ¿ªÀ¿WYhbádõO¥ Áë!lmú·Œâ®@E Gcó¿gè]»YXm,~]6zpÃþÏxáŒ?‚/p¯þ:‚÷é¿”ï­mL~- ;ÀÐÞÞÐ Üz0bx0ƒ7Ôèú×h¬mÁ&pz^S{„_å`0 ÿý8Œ"¿'€Qô7â0ŠýFÜÆ÷ÿ"N&£øoÄ `”øXŒ’¿+xz#6£ôoŽEæ7Ç"ûc‘ûÀ±ÈÿFàXþE\àX#0»òofWùÀ쪿˜]í7³«ÿF`vßÌ®ùÙµþEÜ`MØÝÐãßœƒ±…½±“•)Üéÿ‰ÙÁÆ6 ðTýÊ/Eð-ú;¸_ãÆhòGüí\àˆ™YÀ.€V&†æÈÀa‚‡òOÇ/?vNàUþŸ„õ—’«1ÈÐê6piLC°éí´øƒŒCvÖ_Ðùw°Ì¿¿ùØ©Û8ÙÿÁV0û‚ýÿ˜ Üys7[s õ`ÙüLàjXþÁ]þøWôwèLÁW4ãoÏì`Skðjÿ.˜ÉÚÉÊè×jöG¿Jbó;F°O›?¬˜•Äö÷1˜Ãüz·þÏ ±1ÿ#ýﱂC°Úƒ_ã¨rü%³°ù=làšÙ‚œþàKì~÷ùr:üuyý;,¿„6Ž@#Ðbaeû}ðšãŸ“ÿêsÿ#ý¯2ó¯Ñø£±ÌàêýŽlä´²øïè³ÿÒ:ÿÑ^v°‹ß뮈è?ƒÍ Žê7-øÅÅèhnüc?Àwt±ùÃìÃéž ç? 82—?ælíú»wû‚‹èþ;8°'w ýßTÿ¹üìÁÍuüëõ ^Âÿá¿>Á€@W 1Ââœ1o eu`ëÝWa<úÝqþò]djzEû6§ØDêÊtÿuûáÄ¡î·+Ûï©®…–ˆž<ŽšjaCšã•Z~z>êÇ*Oí¶ ,LbõOä ×ôÀãÓ« íy>Ùyªû}„j‚è&϶sâBQÌE¿sé•p­é+Y žÛUÚ«äA|,™¦Tû¤ëW8Kžc”ñ ›Æ‘žŽíÌõÍìõÍ ZÖÄ ‘t,-‚×q$k¾‡öKÔý7÷Õ2U‡N2ml¨k´Ñ) ‘ƒ$éwóEù+ý ®M|ùDYÈtŸWèß2°¤UZ(GX7ôT9.¶3ïd')ñúÑw⫊0ôíI0+[ô#ЫX? ÚMEV[oÔ»L-?ã/qO×¼f‘µÌ|»<~ÖÚ­ ÷Óßn¾jÞîòÒ(ìÄ7sa¦yo®¾²Ì“Fðùu‰ãd§P ã-Ý’s"×ãkà >ÿs‘\æ­<&&m®×=£ø?س=Ï£¼ÊëÎäÛ7fQçA.h^8ä_Œ:>Ÿåᘳ¾7H¦ÊI{ß–³M©Ä[ ¬› (a<Û(Çb+v#’Ûb_8 Ëg“’èÙ¯ ×õw³¼çþÂðC½\­%†ý‹ñÎqýØAß H@P/÷ÉÁÖMi©ˆæwõû³r—P‘¡¨MUÎÀœ°ve߬N… áÈ‚Ûío=ðF¤'ª"ÎÄV Ño4Tù¶Ç: ]èJ¤ÄÈ)óåúd¾ó…x.~XËïÁùê+„vÝàËRݦpGõ )ýíjyúÞ2©~ÛPƒúù3®žŠê].©ûÆ«žùé±mâ‘©m2P¹n![o#€—ýS (–‚x~Û7õrŸêSø³ixãšš‹pí~UÚl?pýxpZ„¶• C`üXWL™¾&›O‹ñ2Ÿi¹åž!ô@ÝÈ_á [½†qÍTʯ۩A¶.&ªv?¹÷È‘˜5í*‘£d=PŠ—޳cQI>—IËn(•„‡ÿ*!CuZK†aEûpˆº‹îİŒ‡a¨¨¶ˆDf!Û2J¸ ep‹ÓŒúç )E22ÐÏïeôñï6t:±fcCw¸p82£ó,äf F•î!·Ò£8)tN6Þre,½ØîkNq)Ù!³4©-4È}ÝÈ¥S/ÜãE낽_d°Œ©±G&n7uÏ̲Ë)‹·£&z¯~Zè¹´Åïµ’e›ÈëŠ6¦å/)ÜÌMåk°¸pÙËšLWlãÙyï0«3~᪺z8!fŠëÒʱq¿';¹Ÿ¼ð%ï-ï˜ÂtÙ\Ûdù`SÒ³¨Ð”2yãt#bõCæO3 ÜB’ \ñ Ç­çÒ€@¹([Öš]åÂ…¢˜þhæke¦„û]÷!6ÿ eÜÕèS«Ô£¹üCíœÇŽ%™JÖÙø›æk‹º‚NŠA¹ãC«^¡oÁåØyåV+FäL‹É¯i R`{sâ×ãM¯{2@mäŠïYrxÕ…©UDd÷øJ°~°£—ˆf Y¿ÍñVxŒt"R‘6/h¾¶‡‰#‹x•ü*ËCyÁ&µ«WõÕŽ¾Æ¤‘𳌧¢>úÜÌg ÔõgüØ6d·¹ƒßÒì*½¶[CÇL"8!æ£ÆŒWâçD õ¬n„¤)ÛÆoääæ'%ú´ŸVu ¬Ê^šÝ'>‰ûÀÀ£JîKIº¥*€Ð ê"Ñ"ø^ûêÔ1Ð+*ƒË8t9§~k…§ÊhXXùm-â‘P%mŠ m²ÐS¾®wÐØú:ãšê¡¨ï.¯”²+‘U_®²ï€Ãtq¶ÏK¸ö>æññNXç1Ÿ hA¸p·@¥Û¹}!Ów¬¸’늆ÎSî8jc˜ú_³Æ0úµK™ iJv"Ñ*£åhŒl}ƒy=VKFáýÊ•X±ÔÕyrZÇBgPÑ*27,èZ¹¸JçŽ{‚ƒÌ²øuù7„HÄ7dÄïTsŒ~&Ƕ«ûE¨@f]KFÓå‘«Ä%ß’¾èn>ݲ)1¿û:ê (åÞ²SÏo,ÍÄ9è}<Éu0‰J³Fq킃-ÿ²ŸÇ3‡/›éŽó! õ¬Ä1Óû˜f´Öm}èub [ùåܲažZDSŸÇc(šóÖ©ìúJó^.[üs2™º½•ðˇDŒîþË kíu(š”þâÛa‡ë3›4 #!X·9D8bQQ¢wÆp >Úõǧ„{Lï¬7ŒlƒÌ“MÈè§to*ÇgÓ(dùl; Ñ߀Б<›¼Ój¹†z“êiMÇÜÝŠœˆ6õ«EÜ+†C²‚¯Xg[á` /[¯çCz4! ¿sá|˳BHZÁ6sY$†E€»ã܇×rSùà}”™R~²™û^O””áLÌtÐ.žà´ºþð£øýº¡N¨¿é¢+›¯dP6¦ç!¬Æc-ï²áÆ[-tV‘+öÅËǽ»MI]’™R˜;æ¤ñ”cvù;î[Né‘§Ë:F¨áõfTWÅ®˜Jt(Ž:[\cñ}~ ¾Ö_ÃÒÓZ6È«óÎÄêæ4 ü:yR$#²vÆì­ïZ!ªŒmDH8u(ÙÆã–æúÌPw×gÜ×å_0FòªàP"é5]óì8BKWY b¸–Ö¼§b’æoÆ\V²ƒŒ*Hø¨JÙt–ù–7“/ÆÏzaÇÉÏ$+ʯ٠`ÃýŒ`;ßk7ï'¸1éÎ,ªBÛ‰ÿE‘¹Ç›8©Â2†ÈZ*$·4˜¢:Ÿ©#ƒ#l)í2ZÙʳËpBh<Ã=©{–Ö Ã%“Ê7´›f*Ùœ"Õ¼:úz÷{®Š`0ÔSÄ#IZ‰Ç‡¯î‰g!¤àfµv‹f²pdÈŒóßW6þLãY•ú€ç«Ã| ½œiF5GÕQ –5/Ñ¢ßõ÷V±v3dÈ)jI›ä%䆎Äÿ]<õNëVD»oÜêú.Â]z¬bFO@ؤeHîÛmL{Lו$„=†n¥É¹ÊÊœ‰Ÿ+–@½¼I•¤|ûK©_/€1Ô¦S;âîi  ‚¹ÛRƯ„T$)’ø^^·Æý€:[¿âB÷Ë÷Ç×ÝV¯*aí»4,&vskÝæ9Ÿ¡ÅÈ4&þâzö:ÉßhBîy.l“ÑÔ‹u=)íh¤``›D¢8 µ|ŸÒTÎa v–HSCíÉëñ¦®¡Q«·.RÑwvžÜrl€X²ý4BAònh,ŠHÿ¾>‚÷ ¤züÂ'vKºÑ9¥Úïà )Ù"lÉ›eJQsA?ö¾@ì'‚–“‘¯]çBñåØð©†eµ¾ô)Â>ÆáÁßA6µrTÑ0ö¯ I— æ'­µ<,— X=AÎÍ”R ì'Ÿ*âÝÚÇœëÅîác„\˜÷•ɧü㶆zˆ~ æG ¡ 2~¤mÛÞ= •s9¹­Ù_$­^&:!˜zÛõвw¹y?­~ø–ñ€w×yŒ¢FWÑî˜M½gÁ#y‘ݮԻLU¡TðûN ®Þþl7 ƒ%Pæt°ÑK/ÁºyÞë”&©­K­vâ¼ÐùŠyÒ}[{cD$ÒÜW^u«±ž•»BóhC°Õ C:î‹~†5ð–Àb{^Rh%ÆÏt"àéc£2ÃQÚõ“3¥ï€å˜Êv·XËè"Èí‘›3+IëzN´¦¼Ïê{^}¥„»¾Dì]Ý"sÁÏ7ŽŒŸfŸ7"G·Î6ØÂª˜Ë­CõŠ xÙÂÐî™–˜q›“<õA|:•»7XЛn’¹áe/…És8jÞ‘²·;.b5ö_×Bµp*óí^Ýé¹h›:=žFÓ²û@ìú†':rÉ„V9j/숊tD_C‡A½ÕlX’7Ôl¾5Á­3[èô§²Í-‡ˆÚ«*Sl¨I¶ãb6_¯ ³ég7ðÅ ZŠ*é¥Õò-»la‰T§q—§9a'Y¬»ØºÚDwŒK‹V$mÛS™6Ïøé~ïD^_:sÇEÈg؉7aàt'‰ë ]vL„9·zP›Ú¯·Ä÷ƒ1Ómê§áYZÁ r¸Nåxø[ƒéöÑ÷™_º]p»^â!å¸ïY\ȾCó²þº?$ôóKø®WŸë›½Sú×°)»Ö!QÉLÖ4((• Aíb÷ˈÄa´Æ}®"[zÙ)ŠÁ/?¡Ne µEäÝI¹?V;†J›GZúõtÃhâ0çž”†7‹äLƒh—[Û×3mHà# Õà[ݧ.ÍÍÆ w¢bœjų§RøzÐ5ÜVKÊ&!hZ!Ín^Ø5Y>tǰI¦Žå¦Í_¤êuMÚÉmªÓ“ÕN.½‘nóãâ&9Ž©ûó.S$­c2!›‹Ê…9I¾{ˆ+ó1²8>Éõ \Ý;ìøþüÈ-¥ygú¥8} jëeâé'-úbæû㤷ƒ ‡›Ä“=$4ÒË*œ(KZ„•pø¯O-—‘ (¦…¤‘hK÷B;šZàÐ–Ðø¯“äsvÉ8œ«UpUµPÃH—±+d?¿°þØL{5Ays|Ûsï#þÎD]BvŽŽv˳Lâ9oC±±Ø®¶zèÂüSß›²û{mÂ2÷@±bG!$Ö[ ZjZÖÜ‚€%5Ÿ†¹ÁŽ\k¨ÀË|îЋ÷&pãD]Lú›ççæfÍ7^S…¢öáƒïúU…M±Ÿ “¿Ä%(zñ¾'™i•¹™!UÍn@¿}ÐÝÇ}ºÕ›‡"À…BäMïô^Âö}¼ã 3âØnÉ<ÕŸ&¼oÿŽYÑ9UñŒ3*íÜM'Cä Q4‘¨¾“Žü˜¦X«³²Æ§Ö¥Éƒí¿´,hL{C¾Q"ß²"¢ÂÎ|ÐUY’IÒ3×›^cv ú½uuGϨX£f™ªš† ðɦìî€3üSüv†<±}y–{>_~0Å;ûϼÂ&R¢xŒ¡~^ô–wÝÐśѧðt6\š1?zœùU|•]ň± á Ê€^¬Þ•‡w5“Ó£ïOY&%9»ÞUžhíh ÑW‹¡½—‹ÿ:(¶•E†öêÃ9 9ŸRÓÞ–÷”¹Þ&áö÷Ñã¯>_H—Áò «S´*©Ú*åË+ë†ß*Øä=i(p”0˜ôo–ã ïÏ>Öíx·/ƒX±"ÚÀKî¸2:ÔÂøzO…¶Q£ Y§=€|„ôq\R„Ôh† È[âñ=oT á` ´üçiñVÌΫ›»{ó>±7ÑIV%ë­]„QߺQs,Q døú¾¬WYžÏÏ—À﵎ëa¤lô¤¡›„•¡[%s"‘ú?Ò¥[âýD’>‘áKZžæžý¨û6:uóff©& /Bu[Ȳ1Þ¤ôI‘K£ƒFxÀùSC2¤v Íh[[]Ëç¯?Ð+ú÷Eµ€ÜÐS‡xÚ-ÏWd‹u.¨Tþ ~BŸb[—_¡H7\»µ ÉäZ³ÓuddS¬¸ÏdF›³ooî#‹ÇvVEDlÏ"Žšô¸I‘ÂU'2:F“ê0u’çæQ1¦ 7YzñTÔx! ¼\“¦ ß"§Æ~ ‰>tga+6¾½e#;gY¯‚)=rá[0Ø*Úà¤F)+ÕïYDbÊÆš¼<]Ðh™6%W®²·Ÿnèœ8þðVçGn¥XÁ‹þòŒ±Uͪžö‚9§±íü~Ûø;§ëúOð—íé¾[ÜÛu¤Z¯ÌÞ`Û§¶Þò}й볇:`k»#îì ÙAÛ;äèM7.„pÐÍÉå§ÉHZ“øTÍí¥ë+z*†ç“í¼ûxÛÏV€7k”7¢-¥#¸èß(ämúIéòZ%WD Ò8mu\äBRª*pG¼£©LW¥ðlª\ûçïŠ*&®70‡>.7ܢǗmEmuìýpÛ¤„ø’k`:0FIîXµŒ©U›&lc'xÃ-lt…„Ÿ1 Üe3}:µ–|]»mfœK³®!7¦º0¥~%–q1*ÃQ­'õCáMŽÚO&­ EKéM±bç+™Ò8S…°ò;=þÅ “ƒ»æ¸t-Ïñî”´õ&¾Ú–-“1ý ÂdûW£íAéÈоSòXÂab| ´  teòí8A”“Ÿ[ŠSbëºñȲNïÂ]·s),ÏQ5”q8]s‰S/{1’å{zQý8ßf¸óy!V-îôïPˆjŸë^ŸŽž 5!A‚Ʊõ*U–*ñÆÏ¬œ£Rón`¼jr{–ÕÕ¢"X£Ýè¤zÔ–»ˆXÐrÜ©á1ý¢bµë¢³ç:w¢ `4yZu|*ƒwñ>ÍT9º´ó¼’C /f…]¾¯îå2™I À= My~fG ¸¡A¬ˆ!dÇ¥˜Í@»EñïWúä¬:@ÂÍ1éÚ)ÿˆ˜2àuØ”-äÅH©ID{m7Iq [³HܬØ&´ø†ÔÔ,ñ\g>&`÷qÕ;8%Hwfuîâ¼/ ÀÚBöÃÏE4–1úXíK|zŠîŨZÙÙ|·4׸ØÊ ¿änº¯ª¹ïéíH*!ç&yShk®ûU¡¥u®;Vwróœ&…|CÉ2® Á8•îæýhriöBÌú”÷®‘ÕÚ‡y‰ÔåQçÆ(Œ5_ u1Õ|+iU!•¢lÒãvgï$qå|ùëO¾é³@ øbC<<ˆ¶2£·Àº•{ À¼>‰^E}Ô[sôZ f–ò¯¼‚^æû؃}ª›„{%º¿†Øuå} бêPÉ•ðµheŸ¬þ¥þñ5’ ņ҃ª«ØÕ›é‰³ -P7.)Åœdà† ¸ÒÇPÿðf`ÃÊ rWô|vgÎ÷N&€ÝnòêÁåÀk¼¶¶Âª`l‚°»C£í×psy{¶w1ó*ûìv—?ñ[.c¬Plé³b’ÃZôY`¤÷¼ Ì«¨ŠIšÐ6øÛ*YLØœLÚÝr²ÄüØJ¹êjÚ†ªèe•¯ý²Ï¶ƒd¨â£¾ŸMOŒ]íª…½ùˆ ÍCØyxÔè#ù>¿HU vãƒ0&y'¶goÇH2}Ážñ­â›u)¡îþ¯¶66Ï3xåu`u²~Ã;—à'Ø;aú§ê¦º1‰dÕX¿²ÁÁˆX½‹‹J޽ròT¥ñ¬M¯ï˜Éñ߀{GN Qaãêy„µ)Djp]±Ö¯%®¨,ÎãrCG;…s]‘ÚØ‡´ñ1äs^“f¬€ÿᆼPjëq;€úN}PÑ;©D¾öóDTÐÆÑV£Q*õñÂ-ñõÎŒ_Dí$s€¹3­ŽÁÄØÆå*{<ÏJ÷“]–rˆžzb?†¯¨Ü\) õÈ„Ñ{ã)”Ho~ăê×ÝíÃrùìáªdÎýe2éb†¢Ü…®“‡27å1¶áLgØÌŸêámèY(¯üV}Y䇸Ô2ñÍlïñ—õ¿èÃ×3_>PUõWïN u0&·õ4/³•Ì`YS/ÊQŸjjÀߟ†rˆØ ¬s[¥¹•”û"¹•äÓ®å=ˆÚÇ &·ç|:.à†)ä#Ыr*CGð–ÎÍÌxòèù;«ö˜×Šè(yD Q­`ÒZ²È_~>O÷¶Tûˆv…]Ñ…›íÈu‚©w7d?çb»WêõÉÄ+“Åg3•øÎÍuV>²ÝìIï]Ý;”h0Èà ý™ë$ãÃÖ@ä9i~&ù¢øMh}lgïÏŸ‡ë5¬Ù”4l¤vhYZJÖóþî³´Ì'º}ÛÑ~H‚œCR}«_i<˜jâö` ^ÖnÌ9?wé¾­Ü–»\8'ï3µ]ÉL:W7:jžnÑS"MšJ×EsN=›‹W3÷DÝÏá–dÞÐ*öˆIGaÇR#ôXÇûÉ©>íûVÆa÷d°ÙB©¯_#2è Wü;zSÙ+âõýn—àƒEt±qëí@<·$r#ýèÙ@´ÄWÎÛè_fâ9eð}Þ·µ;}ªÛySN;iIâg<œ0ƒÀLOG­ ÒËBÖ¹Xš¶W¥ö%Ëf0ú„ý¦FÕwìƒäAøÙDN`$êÏÍ†î ¼´Ð~­F»aêõÊ»}1/Ð[ïSG¡`=·ôUÌ¥0›ìZutâ–O&S l½0™úÎð,dôøxéÔ0$_>Ì¡.KOjŽIÜÃJªqwb æ…ÿ SÝÆ¾]Q÷ð‹¸šÇtø˜¸öÚþ*f-ăQáÓƒ:ÔåTîë\¶]A!N!öÄ"Æ3R¢¬Ãì È;U¬Ü6”¹ÃŒoZ&;«÷…ºÚS=5]Q²ß ÞªžcS Zº¤& LÒQ˜Õn·tpût‹£ÏÓ'ð廘¢ˆ³†D~ú0’ç›#oüƤLúj;ÄÊí9¤Úv[EÛ¹°~ä£k„LimÍ”m·S“Ýh’Ów¹ŸqÚ©Šz—Ù{=²»Ý}%aÞNÃj—¾$¢žõaü¤~þ‘Båû¦NHÜ`]w›*-¿˜ÙÏ&G•=9!5½¡Ìp*Aƒ G –Ç ¾Ç¡c:[uz›µÛ­e£6A;J5Zòµ,RìWŽ'#qúMFº£ºÊ“#>mÒ×ïßø1ÜÚõ¯5Ú¶ŠÉ#©`ˆÒLAoî†æÕfŠL?¸63bÎâ§Í*ÅíT gù›ÁõO‘/ñ ¾ÌðJZ’º§T^ @·: {ê ¾;D"Œ‚DKøy;=(*~ÔÙ?‹‹#üÃ+ew|±bƒ<`€]v}¦Aà$Œ¥À Ô¢¤-;Ž@LEPÅ<(ǯêøÌgÅóݪ—ÝŒŸíååÝÎu#l²ÎB|ù¨hœ¡:¥YRo—vJ}EmE²?[˜$ñMT%+þ-ʶ֡ Ö"~\¶3>° ¹¦)¶d~Yd8ØPk·Á奲î}B^p¦Ít%„à–Qº#|ú krÄ®+ yœÇ 66áµÂ'$±jd<<‡HÊ‹Ÿ·f¼¥í2D‘»·(ÑEÇdXû©s¤Î¸Ó8*‡«óRT}óE4J›L õÛ>ïÎiî5…šÉ Im‘'ûí’œ³JÌùßÒÙ•zЯ2ƒl×ELÈ'p¾mÕ ÍA´Ñ}ÛÝí_‚¹ÁÀ”ôÒi’LŸÌv…@ï ›=> ŽØé~²¤LŒ3¨¥4ž3Í·FVl+—“ŸžVp:îëøÑ~º5¶tÔñôC@|USa8/²ÆàÅÁ멲 ë“¢Ï‡`¦ò2!Ÿ½˜J•×~_`[Åì³Ö³6Ήås}3vTU+”Šªê©Ä%CtÄ¢™üz6£‡Å6ßëÔÎ̦lÝ4 !!·át!7h;€ÅO7^RʙˌL»Im j©ìlÂ8õ„÷ß =õU(U#AVÉÀQq£G§åé>´N±ðü{ä7Á8\ ×ßèf–_MB¿b·ò ^ì¡¿.¤Á L~H~ORS—HñÁ8ÏEüB»B×u„÷DÏ}1±p§‚"g‚6·Ìœ,Ï _×õ ÁˆÌ~«Ÿd8>Þ†¬}„ÿ4Kä8€>˜&jù‰bgÈþèÙqœF!ûÆé[61Ÿ,žŸ“±ØÂÝ!†—ÉM“ £Ýæ¢IBÀ'Tæi!&Fj=b2.›x|¢¨A<ˆxi÷•xI<ËY–­ Õ0ŠöÊ›FÝÉ1$+õ¹¹“xo[a«ùöh¼cÚÂòÀÞõ°½–eŒ²BkÌ.:Ç ÓQ1ñzäJÁ’4óJþ\Žc;‘]Ó±N\bƒ¯Ø˜…¢yÐ^6Íž§<½íÓ­²usõ˜ã—o–ËÂRæ0Ń]Ì.’à˜µÉ Í¿€Î÷ìbë_ÕgZãs üy&ÃØU¤Þãh©$fF©&#Æ5ë™”ë¨F/’b¤x4§ÿ©yø±øk·¬Rùskj Ð™ áJ¿2‘eÍ0RhœÔ¡þˆ¿Ô½b&‚§yyÛ†¾‰ýAr[ІªèN;ܦºÐg›eíJOkíÀ5ɆGÄ–„¾1GÏ«-^A¦ŽE—ô·ê,<]–2JBC—Ì®ˆ·/Z×ßO¸aYKe°÷;.x3t´µªÌ+>^ô‘¿&3Z`r˜•ºo»³¤-€ˆ¤ÁÔ‘äûy ej`É0Ýä€BSÒ+ºqEïü&ÙÛâåiÖN­´€7âvÉ+²Þr™}7+£@l¿žoŒÜ ãžUL¶âXÏçzb¨D°8éÃ]‹‘̨0ã%qæÐgך•r÷€8`Z°„„¡s8qöµð'»/ÝÑÏ÷™TM ]ö *ÈI $‡0a—V8­«wŒ1þ>gL’äþ•”Œ8½P3!‰Ò/HÝgÚj;ë5 Òè4Íô× ¹†Djî[‰»®4,˯vDË~ÖvQРؽU²ÚæÓ2aæêå§ç¨DHGƒËÙ7±0ÇnaK‰uó%Ö·~-ŒA5L“ÂÕ^EŒí6µý›3 Q‚*{VçÈŠÐWA¹’&KUÌ%“§i´¦rš;»90Î ÐôF¯~ðñ˜CʰòQÙ×¥±•”FLã *ë·òù¢E£ÛÄH@CŽ®*À,QrÓW$Qp{vm6»ÝŠ›y™ˬ"lí»*ÍÍR(Daý*ø>„-XÙ­ HïBþ–è0 i„Fax}ãÑ{ m!¶Èu¬­V?¹[bk>ð‚ר$]Åç‘ØoŒ”­CÛªÐÁ[ÀþÔ"cÚRj©‘ —.­Ú¨€®{ÌÛ#5žÓ©úÚ?ú\¹iª"·ñ‡)Ÿa}çÈ–~3WGÆûÊløgþk ohë% îÓ’[ÍrM®âËÇ#|ô@Å—å68)¸+zò÷cUÚK¨‹7ï`?¿*´1o„ìô«nø"ÒÜù¬œ40õNaà­@tU ±w NüÓg!“³á»mx@AÆÃJB¡ÙVìȘ§IÃO/”ç„¯Ž»mƒºa×T4zwÇùžQâ—²‚N;xûÆ”ïMv€i=Ø1†ïø!÷‡;^вS²9gëÕûÏ~VÜú˜”¨HÓ\£™n—OWj'ž¡I™äj QÁáÍ}œ}s¡ahr‡»¼£o½ÐC¬®aP¾é®Ö`ÍFÉКµÇeŒ¿!<œa¦á5ßHrŸ(pO»±{õTLjöÌM¶iŒ×Ó^©Æ¹ˆP¡%¢«ãâKò'ql²…¸ôàE[ ÎÜf(A‚Ôè}é…:®îœ®+/,\”Óæqø”ßîäuä:¢'k>$bDW:–NV´,‘ZÂöJÙ}²ü–ø%)Ο_±(&Å{–NRAöS³¤$C_–çdÒ"Ù᳆ C=WêWÃjjòœ ¸léºph浦Ô@‡§Ñ iieÁ­ÅejÏŠ¯Î=Rï»~X}´²ü  ÕZŠŠýB b©µGÝÝ1Hÿ iq=Áä¶À²ÓÕ~ò:pqXk>|¶ÂJX.ÃrÀó‚Æ6ÝÄ7ê« ç†23KH¦…ª}KÃKt™\¿mKK¿ÿ•¾j ËFj@þò²b®\@‰’•¶Î„¥ÕªM;Ù Sц9­ö4p¤p\ 2¥ÌýSø DÝâGEíóx•“ȉ/þa“˜.JÙDFéë¼T噆Q…• —Ç$Ji³ŽíI:ò7Q ùqËz|!Ʀ¯ÜTiâdv‚âD´Çִ̈́ò?0÷7„ùW*7559o_½XJˆ¯ãXã§MY# óY$?™ß=5šÉ“–O¿Æ½SX ©]3i •øÈ‹²ák|Ïâ|;aƒµ29‹ðnC.Qhi$ÞFù-ɹýP…~ràÃXä$*r…8&\££á{Lʅܤr"1Uz~ jY;77¬|¼Hó“ÕmP¹/öÛ /6¾—ýpÑú´QàÒ$[º–V^×5yß’CóW¤îÆÅZ©§vK½×—›º ¶×=ItNI£ánn\‹|½£ÉÁ oŒŽëß“m‹Ý³Røi±ßÂ5V Äæµb<¯Çc¹›ÀÎ?m†d…ýx ƒüvž´¨˜‚[ó¬±Y¨wXIÞó|Úy„UÎzŽàÇ­‹¸PPlNä ¼“K§É¼GFèC¬ÙëÁžíÈC’ÜeÙ&ÛUáïr¶Ãi,quLlä ‚Óe|«öa.߉½/‰¥½uùÀŽ›ÛþK÷,[ •(g[-F®¨À×ÃêÂ-ôàj¿w3ç¯fÆaºç\“öäHâ9¹'‘ÐÊMµ+*ô!ß5¥Á|IÖšàzÂ|Ã_ŠJ(‘] —‰Ý} ­ðø5d'»ê}r¨vSdËÇPr½ðÍr鹺3+¨ÀÂA!r˜8šÀ`ˆÁÀDaÞõ†”ÏÅ×fÝ;üPlW)eÎéP£ML»E ¸vb=ƒOpÉR¾Eš?'x>Îͤ²Å±4ov3ÁšÇ¨o‹lãê–ãß~>æ‹&„²gZ¼ùÚþ4éÙz“h­Ð¯Wß]Â…îè]©ÉJ,þi¸ùSPÆš”.ãE>9žHžPeNñÐ×ʾûóë÷Ed9R‚Ѻ]þæE¼è†?KŒ 4…@¶„+”$ûú&õþÞY¡f2Š4IGr¢Êg°œEã* 9± ~̈H‰xH%,íÙä7l)§ F¶©‘$½ph÷Å~´m‰\¸ F]!ßC9%‹¨¢˜°Þ“ï®ßô â,¼8U×ÌÃz‹!ÝÒíaЇ 謗ž“šþxòíôEöÍ’AiÈë‰Ôb‘5ø"ä#g÷TíoEo_9@‰~>îRŒŸô-fï Y±¡Z÷±íp5МÂMEò:Ü‹²$%wî÷®4T÷2Þ¥;Q^7Òm@áïM‘\–¹ï~û¼#öZäÑ›]訹¦DYU2BHQà® —ö‡ÞË(±oe˜´Ÿ>­©Níd&­áü-Sebµböóââ´dHL‹ ÷°4ÅîØöK¨Blª%5h#ôuãm ë·Z}ï}8¸W‚£à¯Õ÷IÄ † œÖò3&<=×Y^9Þéï•X·2c¬è&ì2Šð-ÃøâJPQ;CýÍWtS ]‘WSÙuÝ‘#&ã2òÓ-G©Óº…þFÚ=—˜É~f~çq¤QãúOävw«³j«Õag]¡Põ:%q;M×]|Ëw\בðõA¸yD[ÞQy<ª®äŒäÙ‡©žzóûÎÏß><'k)1I@จb<›å&HÊà…Uj ¿y›'‹¢,Ýö5ûØ|í¶ñj©ßªþ= mŒD•Ãw/¬Kò¶1x,+hç(N‚Y=›H‚¤»çPŽ,CpZgØVK¹X=¤¾0ÓÎskj¥¨W2'+Tâ“âðßž®½¦ÌǾ ÎYFÀ<ßXÌNöÒ5úxu…û ©*ús¡I¬ÊÛgMÖÙL*Þ®=vãkx´Ý*Öi5¡ÃÁ°‡NհסÍXŠ"³&—\Â;>…úˆ;Ÿù¹™¦dM2Q©7nußY–­ÓF›]‡¬žò Ž?f} M½Ú£ŸGŽ 'Y‡N~ÐE·,jȉy‡tý¶¶S¶ÝËìpðÎÛˆ!Cì;7^.=Ú1/& ’Z„¤8QíUì˜bv‚â¢pö±ÞtQŽ3³Ô³rÞ¦îЇG¦[ ¾Rˆž1ð¢`G¥7OÁ£°R×™þR K‡Ý|oÑŠú?ø90{nävAlzÈeó–šúžQ’\ãÜI?LÀo[ÒªÎ;©ÇɹökÛ¹u\zÝüëH‘¸.—ŠÔ&þ벎áòT‰ù[5NÊÁŒ1ÚѶÔä>÷›SÏ”FþPwJ~¸btcdoäQ´àNÌ~}n4fåEpCÞuKhòè̘ UçJ± }ëJ*4 ãh ”4Œ!žÃâÀ"lÇoOÀÐ^k÷r­`™'š‹_“jæc«­Kú’ˆ¬m¨dJS_{0É ÿDK?Œ—zý“w-éšÞbènÍާÐJ{Âqn(m] cJ<¼ßÿÔ8§Èê8:Íç;º‹vNbe·¾^®î7Û„©Žåeâ«ót¿,”žLa}QŸhâÀÝå^÷®·SÙõðŸÍ¡Û¡ ;êc®ö6¾à~ÅGâkìýD& #x¥'¾áS5º l|«ëذC¦Óa¢<ÁN8ÍÇEÇnœ“€·Nõi§lÝï'>çѺ@ÊÚ%g†82Ä™v…ËPk÷7 ï“)Zˆ|¦°Ž ‹¤m»bõ´­¥ß©÷"³#Gl–Kg^¼_þÒú±ZoNný»sÇH®P 3*,Õiëf(ó’äé•Çü]ígUÑ®t&i…ÑFT|Ñó€9ãÐ*غ‹Î.ãÏçšMŠÜx‡&…«£…7ªjw¬1j0®Y³ÒOÚw¨Š3<¶ÅW°¶ ——™qg$}¾ eÏúM}uMOC¦]˨9%?ÚybÂ+8Rk öZ€™ˆï *lr#è(™dWØÛ‹+‰4ô¢OWfp4 òˆ—ݼЕ¢ÖŠÙ¸•³?¹Í÷¦>Y6?í=×øE bÀR„¸šeѰ /²Ç-T0!*µ‘†êÆ‹ JnjϤ—ó½níä‹X6‹Í¬oÚ‹?}¼Ý¾¹%Æ;@anÁå*zŸ ã(Sá!¥Ãz°:5 _ÍvØH³¢éõ@ª~ø9ü WÎ\ßsŒ'T™Û§Ç!%³Z¼ÛFy#It¥ N½Ræƒh¶ú´Ó}ÅËDÐ-±nGV,c‚prir¤Iƒw¹™ GkCÿÚ]ÃiãàãäÛg¤)¹±À~Õµ‰v|B&U÷6¤òû‚Nm©ÄNÂMM¤ÛfÇóÄô&x°éïò Ðvû wbý¸ÜŽà g¶™fT_ûØ´áx+Nd§îÑ÷ºèpA$Ÿ§(3¾‚ ®  ×"%Ž[wæÙ¶¾Û TñYÊîŠâÇr!^•UtM0ì¿[P2"ᄜJhå8÷‡$Q¯=§Õ¡zÆÍí¬™<„À»:•£j5‘ûÓFRI¨Ô\f?tÈ Hµ= Òy\ãL0ÇÅÊ;º~zöÔ{ÇPé¹t彯Ý=o‰_™/)–:wjN‘nZ‹èâÆËoÊé;Kšu0¼S‹<^EïÖÖÒ|—Žsy]ß þ@Ú'"£«Ôp-O­:'fÞ¬Êg.Öì~ÒŽÓ¢¾®ëYqŸ¡Ê-Á"dl».[ý]\1~]‚SßG¢âù©Xÿé<+Y\Ä¥<¬{zq8î1%"b³ç‰žö©cQ—VˆßÀvì9gQŸ¢é³ÇåËÄ©wUŸüš,ná0nñLky@óÍeìXŸ¨ø­Ä?ˆRœºÝÏÇh Ð?²rÂhØ?›Ò| ºûjRfäjað”Ýå”ò\¢&ìOŒÔÃmÚ«ÊŠJz>€aØÖ(E‰^°j€â$µb[•AÃÐëÔÛ1<"‚cMQ äÆ¿°ÓÌ/)§îZá‚gZ…矗ïÝ`ðO—çH²Zç)ב"Ù„gTŠí N€fQ©@Â6ŠšEö\ЀSfÝcƒ|´|µá©”á|örÁ}÷ÿ›"dÝ™þ'ä{oÏÚ2 ÛÆNÎìdi¯ÃQ”ÕòÔÜZ,¾66÷úðÝH]~Ù P줧“þç‚ôï`ê6´toV‡éV•A •JùÒÝ"Ò§tÚd(Ãf‡ãÃ=·¾šœ½ž“U’ÅÝ"½Xô¸´ß­ÂÆ;@§,ãnž )»Wå!t¾!Å’¿¢è ã3ˆIPFüºïÌ,u¯‘Y²,Ê£ŸK‰ ¸+yÚu}_ÊÝ +±Õ®Æ)Âä­„ÝãOß8¤¡•Q©Bàµ¿ÇØ¾e?XôžÙ•æhøUvñŠÿc ×¢¤»ëºp: T/÷Ñ i\¯’¦„éHÇZÔ‚ähPzœç.öap¯yqi´QfˆÅáÁÙ‹¤*©`²¹bóO Rÿ#°{{Õ|³¡î% €û !pöóöXl5ý…?áû «à“¼þŠ5Ð>¶_xZä¡÷WÝ" G•H$eÿJÔnľÛ¿ õIeŽÔ¢ñSV[5Çž8*ŠÀáÀ ·R)·æ÷¸ÕÕJƒ8½3ÎkOa[¯èª²6±ÿïs¹Ýg¶žâ¾¥AAü€D Ñ€‹±Â¾N÷œOÑ‘š ^arÖ _¸ä)l‡Î¡Á&iœïGê)€Õº´Õp¶È(ë˜^y2éü²ëu;–ÇåXõUŒœB>žð“—…†"·§ƒ‰ÏÙ=5a‚L~ôü¼/A@b¦  ìƒð^4£í·ç­0h¾o- èä̸Ûa[ ¦@¸ÛµÕp® ê(ìSÒŽ6,W%ë°½ô±C@´[ŽyçL8ß'ñü5Ûœt@g#Àjኬ ¤³?è䥓H+.ñ;Éò¦s$“j™Žû;y$DÚÌyˆ+È÷1CÂå:iNèiq¼³mÐJª¿î§UûH‰Ž„¨L°xÃÝÏ:ôÄÃ\üK6˜ôÅ“˜Ë¶e Hj9öÀÏÑS¹ü¸D0 <Á븅ŠF£quÁr™jÞ„Û:†Ý£û®ž|ôÁÁ®µ;çPâ÷½/vˆÀ¦/¯¥V N>óÙ›ëá´6(iÆI¤OƯ‘¡80 Î6ׂºÙ–maÚ–Ã ly•ŽÓ§ß fxó×öäKBηa„;#ݰô¼üû€Ÿ£+5ü€ÖEàæÑàoýO[º;uGS¢¡QVPegit†¥÷ì‚ë ‘!º„Û¨ï™*B^®rWz”Ι›Æ-Êäî9Ÿù³s6<½Ø ›Ëô=!ªfw]`ExØú&š@Ÿþ‹Ï~¥Dq­føm øôˆÄ¨Í+Ћ„ÚŠéãÓ‹”ûÄE ¶F¥8³(g¼bÐ;îõÏš·LÍK‰p×—fsý%¤zÄnK &:5ß(žü›@öäÏ0ºÆò¡¤äÕþç÷½õ­ | PpFk²4ž'Бtî“_ºÞ3Ú ^ t ·H‘gÝ×çMëâªëzjJ7²3íc2æÇyå|?)§ÐÉq-¼á,.þ\éIÓPU&sô£¬e×$/ÎOS7òí¸Æ8aã°Å·Ù¢E`Ûcˆ³ãùë«9CÙ V]S?“gÁ4J£l”‹÷OãìoY;¯Xå§ÖoÀ½X5Jõ ld1@Xô®:»‚¦8Æã—±¥‰›uXdQ˜™éòà[=€‚¿¯Wi¦W ä*ñ$t¾8ùÞ`&'¬¶×)8{£š9]HJßå_-{ Æ&¬eßÝÇ’ùåä R,¤KˆçXaG<ü¼DŠ'1™Sò…[#¼Î}YÝTÓÚ ñTÑàt"ë‡Ý΃hêNHmsœ©(jì/µOþ¡ MÔF=JRèªÊ‹¸^–8¸¥ŠêÐ ëÚ‹YSçQ³#ßšQÄî)õý+ c›BÜb¾ôYàEÙÁ`_ç§zô¹ Âûæ>Â,Ù2ôsÒ$¶EŽzü¡2×ú16éãþ׳\мÿú&ö<Œj>E¸w¾]ÿWÆm(º~€2ÐûÉ0à¿1 6LD]Ÿ¸ŠºotB¬ÑÓÃæNk¯®¼¤Ý»ëÚéN|*ËÔý6Æ“Ä0Iªk‹ (Ö„©CzʘŽŠbÛ0 »qGÇ’™k¯ä\g#´½ÿ[c‡¯A‚ÂE 2¦mŠM eû´°«m§\RgóúßUø}­˜|ñåSЊ`Û,>ש†t è›P/Tz‹²VÓ®S:Q±G•PÒvË!'§‚"ן²«m8 ?Ï£HÚ᪣V°Cu£-»1Ònd'ƒ^ÒÍ4·a¡˜ÚR4y&YâñÖŠ=Š‘\§MÁ^yZ»„Üøõ^¶&êF«0Š”tn¥Bl8²¾H<ÖèŠJØUÅŒn™ÏY$­ZARè$}ðDu”|˜£FÁ¢ ÞUÚ 3”ë4·OÙóäêôuSð9g‡FàAV g4¬«‚ΚÄù q”W&ÞZ‹+4ü[’)ë|„¼Óë3%rTÂÁÌïÁ0ÞàýÕi´‚ˆúØ×hVó_Þß›<ö,üø´èÔkÑÿ™>;I'—ßCf¸(6"ØçN´ïì7ÇÀ®PëDº ž•=Ì$lZ‚OÇo¡~&­á¹‹µ9¡ù1kÿ»–W¢ùäQ¹ÏVê‹2-àJµ¢i Cèî¡ ëJ÷¸m}û­‰=nJ,œí½`¯”’LV:=Þ'Õ‚GŽ¢Â[·—»Sn€DÝžµPØó¹ 9â+Ô©U ,z8áe¢°ý,J¢ÝïÄA'Jg 0É<Š¢ŒÙÓüAóh‹ç>÷iãQûé§ønòº×<IVð1@£Ù~a̸wYD®Æ¼Åÿîªì~/u,×NØ +ú¹Œ—ðíªÞäæÞùo`´ÒSA…!³QœKÓt/9އòÆå …©«à³­:35ÜàÇG ”wã‰Íú$ÜÔU­7ùHsIMþ™EÍy Ý&72ç@bÜ"å…Ì÷q·+ã]]¸¹MÃÒAZ¯‰EíUEÒs*=çã\z%&™$a@ Sižo¼) =ÝZ·tFÀÏQU³=øP5ÁÑ™~kò"Ó”57 ó°¢NIÄ„vîtº|H¨{±ËOÄZx72ÙQ­-€áÀeª$kß2«¸ßÇâäNµÀ•¾Í=ÒÀš•­ÛàqŽdÉJCUqÛ)K°Æ\5è&„<=E½B¸cKjÙ™7e;‹4—æ e”ÂÆö`s’áò2á㮊aß0DÅâéÓt‰'“:âÅ8’©-#;Bh‘lpÜ…’±"„\§ž¡ ÞX È:{'´6nñÌu„¾Fp™ …ßÃüÔFCn¢b‡Ú²^ÿÙíÃðø>“ûzÁ8ýϾXZWé×\(]cýë+à_eþºªB± ý˜×UÓÙŒ´eÏÇܯÆL&hî[<}d ¹”¸5fgñ•Jˆ®ƒ?~‰iJÈ&Vk@J´ÿ(ªàûÑI³¬ùú6ÛóW§ù}–÷yZ'"ô+'²Ë¬‡u„»Þ*Þ]‚ÒÛïÿÈ~'ia©zÐÒ+Åkâe¦Òæ7{”ª¬% üŸºY"(ö¤Þ@Ã[¹*îF²ç`½ÁæüÕ¿= èû25@|ô¢Nw÷ŠN¹¢s¸f¥p¤ZÏÙ IŽfrxâH×Ö?5ÄäägùÒÇ®wéìMù€eÚþ]=» .x— Ô“MtI{À~Åöé»8ü .h‡ Æ6[€žán±¬N$¦~§Ð7­ ïb›ÅÁòpŒ|¯±%#dFÑ%®>ðƒ«Ï5³dý÷*p8kšbâuë¤ÞÛ¸ï£ï„sêt4D 2D'ôì1 Ó {•Ç mP¶ ã,¯q ìST6,%ˆå[³+€;¯ƒööëªÆÈz™ç"Ô„B—8'xÓÊ\k íø*««'F’Ӱù™Çå#Æ—òâÍQo‰+/¬fŒÙkޤj}ø=&mÕ=$0ëRô ¾_ ç{eu’ h¨pC'-U2ñ^º‡ˆÄz”À­í·øM€Ó&]l#Nɼ^m~“œ;“é¹…œŸ¦ç¥G¥Â’ˆÆ(FC@OÂßPÇ^H®÷à øfP^ P½mS¢|³«™ÊÊÊ‘äŒZO`Œ+ÉÓé³±P0\æk¸³1iÃxck»I³ÑOŠÖSͺhº˜À"·ß±GoºlÂáÃÿŠÄõyÝwƒèN"\"€¹bî͇{D©7‰—uøúŸ~‹‘›VºÜ”-ûK~—ÿW ßÑ–G¶U]7%°ò߉̶àEþêî…±G>÷Ø»‹8,V•üÆ™S„ ˆŠ–1@Ü× °'—t£•¢ø £¦î®ÇbhÉá•nñ`Ïw1:Â/>ºÎçlãPõð‡/<õ^¡R”q&3¨„ô`Rf+*ÿ\/èaw a[›!_=w¬dž§ ÊÀ=ÁEœ­â$¹KN@7R XÜRkQx88H§ð…½+å {±7ϑÌ »f]gÞÔsºÛ.×'ü„×dþéKHV¥-`BcŠ(“û:Â6žÔμ:êÀûc¿:S78@^›z¶:$ÙP&Ök<¬«YÔ°18R›ý‹Ñjø…Ø+dã„öœ [ËX£mo«X[*yob®Ëà¥Y4ûéè3~BËãå&[yspê=mìåï„ðˆ}àð1Só/Jotçº ÍB’ùDb61NØÈ \|šÞŒg¶¢‹€\ƒ¨'µÖwV_PBcÝ‘vZg~ú: u÷µèkJyš¹ŠèVPÝAØ@B{ÔE-qOCÍ{Û|¤¡ü:é‰;Rm¨Iù;Æ}_3½›9˜ ¼RSë\$Ý(¼ßßzY§¶:¶ÃëÆ3ÀiáÑ¡Ä÷CG$á–±„!”¿¹I„ƒ…{{iOTD—m”—¡e©KÍ«^a]Õ¦«âýòЋ63G¡.T úÍZ#¹Ý¯pßÖj$wáÐUm zÞØ|™`‘ub&åB q¥cÌ N‘žƒÿµ‰Œ$£ÖØ@ÐHÝ*gà­Ï¢¯±Ð: ‘?±'‚J#_·m­xp*jÕÓl>€´h¨ÑšÝF×y¹¼ª?…0m÷× ?¨¾ìLÁ.×ß@«ʈÿ¥ßoWi ¬÷ùáY\¬dÑKê\×hJÍçŽ?H!Ò<|}EÀ¸Ö@®Ú>;!ˆÌ¾Š—,¸ÝïaC¥}øÞ[ìÜÓ:ó i0VðÆY3­Ï$- ºã9aÈp屢—Zµ!*¤ØÞÞkÜ÷jˆÏ·5ƒ{™ÆÔ:ô9Ù¶)Rµá{/oê gâÆŽJp2&yù_7'à–tÆò¾_4Ù·'§X©„œ.¨…•în {Ã<É›øööI¿OK›*©9—ñÂáÔ ÷¨”Lîç d›´ƒ€[`âUð¿­ÚÀ‡£Ë®ú±ò.œƒ“dÞp{°q™Œ.m °Æ„ѦFçWð(©s#Ó‡ú¿±+Ê ÅGÒaqŽ® Îüý31L2~ð€F¶VïíbÞfþwA½Fvå*ŸŽ•0~ Œ\C@¸“ü ŠlŸVxÎ%N9µÝ!§ÔþÊå”ÙIý®uüÞÿ?¨\ 7&ê<ùK_¸?°E±õþgÚø=_-Ìü-2†£Ÿk0ð“YÒ•øø8ðGëIåˬ6Áµð&;4ΣÙÚŸˆðlüïîúÒ†ª6ëO¥ì Ü+_`lDóøèžïû­ž]¾ïäX~,û‰¦–oU߯iÚf:uýÖsæ Ü(¯ÇûÜy³õ’9ŒÐ`¹¤5 OÛ⮆c•xÚ9å`¹Çnu>qÙÐ WϰsˆØJZ G&å4J, -…ìZ6ú%!ÏbQ¥h™gQ*ÈàhÀïãˆMðGÁ—"FðÂ÷âk Kú$BIz£•v¹A7<˜' °Y’/ÿ~ÛH…BúûŸ2ä£>ý5¦ça"ƒŠ‡*ÿŠí©ýªVж´UbKû»¢ËôôO}ïV¸fàIo×;‰‘Els¥x¥Î¢Œ…WÕ^%¾šš·‰‚Õ^hñ½s%@ñ7qQÐòµeLZ®îkÃÑÉPTʵ;½Ÿ¿ñRйŽ; ’Ft¦Á$d×:†±•¯IÑÑusÔŽ?Ï,ã æ8r/¼š¿çg‰Ÿ‡£–yþ_Ðn É„üfúÊq(˜dǯ!°K¤³“ýùÑŽ™zãÝ·•pg+õ O¯$Ä'‹èó¢^ÍcëñôüHöã®j^'›Á±<ñÚ‰žŠ.ê&@'¨¤?ÙHÒ ztf㢑8¢ìô. 8"¨O‡7»åhöÙïÏÂå«âç*¹Trãß^ÐûuJ=ÌýÄ#œq"¤ Òï"tm|†òè¹lÝM¼ç©éÛ-°;„­ñ’ÖÔZª¥.Âc¿êX0‘ýêe³ùKy³jÇj†ùe•7éñèo£ ™SXéïºOu#P(­å}É× æ~‹•=iºÜ*FïyýÅÏ[“½½çÏ®MÒa¹üYZOÊóv*FÏKzRŠ$‹Òe“æá6¡X$9¶ CãÁ¢E’‘_t&ÝZÍ léù2²Ñ"id=i¨çƒÇ³rßàÖîù¤_ŒÚ4üÖ(õâaîp] IGÉ¿ò9 MÀøƒìud>4Úª¼Lw3ùÉx(šØˆY•°Á67ÊI m+6àì!÷ÍYmB×v•ƒ¿°g\—oÛ÷Eö F¼e/†ˆ{qœþÞ*:i9¾«»‹sUõÁáÂÂÊLÝÔûÿo~`0Tuá¿ &4¼òËÐWpIôW›+ຠ¨vçGÒ+#ÊÈ;&_eÜ[Wq€&¦„pþóû¥€±ùøðÒ2ãÏ)~]dïÒXÙü3}5Ô²’ê1²±´üŽ2¾hØÂ‹À9çW÷w¼~8X>çÏ®„%g½W«†j–&à »Ñ|ûHeNüÔ —k>;â©0<ÂHv§JbÕ¯6B#¼´F@Ì®­L„ÌenS.Gû¬ìE‚ʇº||þÇósZú£â~/.NÉb¼É=ƒJ¿nT^” Ýj=‹_¯„¤{1?  añþg;|gJ³#Ú\–o×"»ù2’_ø 7J-˜„8é #½VÅ1-½Á柊’çç)'r"䎱ø6Zž`ã±Àioq×hB÷Úlý×g]EÞùñ£©HÔhþغ•v½v”½y sGµ=0ù¬&Î%N0‡dpdÖ —no&ÿçÉõVZ—ÿ %&i3ÖËOB¸Á`½àð=F·Oñ ¢­FÕT ((!/&8s–‘ØâLù¿‡jptõ­EºåȾ³C…“9É}v“ýÕÇ…^k!‡ut­ðvœHåArü V‘ðVZ©áIƒ P‚ŸŸÏL¹ÎK¢5Ç?q$ÛQY|ªßÅ›)ZþNTý˜€M½r`Áš“ ®3 ¥@2BØ÷¾]‰ÿŸ‹f .fÔˆÎUÈÄb‡+‘è£0¨¶ñý%§ ÿ¢s8gŠ u·²á£Ð/q6’i]‘Ò þðyÊ÷Á/<|È£Pl¬baˆCËz §X/s¿Iãm¬oh¦oóU¾PšËcïDTfïä(OÖ ˆ·œüˆí‘Iáû:'AGI S%-≶?‡—fÚ=<;˜‹Š‘Vb¦ÿö¥vÏ´žs½=ªOžãJ¹ ˆáˆ g&x=•ã VŒçL}Í\×µƒl€Ë4Ö½ jl»ö{ÙÅKht´ÁKv«:q¹ÒmzdÉ}›ô—v8BI†Wɽ<¥Í:T;&ÈB q^<¶’Êm?é¿çÉ»e—lÝ™Mds‹îYÀt–ŒÎ쌆ºmd(5 î‹é^°3­Ý‹ÉÛÂ6ޝæ‡Ë8õY¾Áì©íÄ®ŒŒ:jÄò¿A® Z6ñL\-²ã2 ÚUHvå›”57½ÌxØPÇKÅC†¼ ‚ˆ}R(½ï¸<úo/{ú±‹9¿–N•œYM-10iZ±ƒbUT3c,”Hv}±Ý+LL>W+ž¾J-)kC)5°•º m™…òb^ µ¶w„«YºÍØÙnvA4J”3ɹ< ”ò“—òï—ÈõC7ÂIñŽf¤ýÜŸem²ˆ>’óN?ŸãkÂLœ7JGJr_úÂ‰ÖØu÷m{}Ê•7õw&u´Ò){Í¡Œ²,îeúî4„tí€i‰¦ ¬©8 »þÛ`ðˆè"O+'*.Lh:K µSð`í&%,,“l–BáÜ:§·3…3%YV! úWHÒžuÛ§kÂ[U}MFo’i¥wáöÞ±ü¸Ãb¼ ZûŽ^H›ƒâÌ– žSÀ»¡°é“)dʪ7פWÒaàTdÔ¶›'ÛÃày¯T;›À\Ú/néB O‰a,?Û4úEÓ=·r ,:-H$n=¿?bm endstream endobj 197 0 obj << /Type /FontDescriptor /FontName /IYMWNI+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/R/S/T/U/V/W/X/Y/a/asterisk/b/bracketleft/bracketright/c/circumflex/colon/comma/d/e/eight/emdash/endash/equal/exclam/f/ff/ffi/fi/five/fl/four/g/h/hyphen/i/j/k/l/m/n/nine/numbersign/o/one/p/parenleft/parenright/percent/period/plus/q/question/quotedblleft/quotedblright/quoteleft/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 196 0 R >> endobj 198 0 obj << /Length1 1615 /Length2 8505 /Length3 0 /Length 9563 /Filter /FlateDecode >> stream xÚ¶T”]6LHHwÝÝÝ"Ý-ÈÀ 0Ä CƒtJ#!‚ Ý%Ò t Hˆðá“ïûþÿZß·îµæ¾÷¾öÞç\û\û¬a¦×Ñç–Á¬ÁOaP7?Ÿ@QS_ÀÇ'ÈÃÇ'€ÅÌlA8ÿrc1ánTâ?á` âÁ§D<Äi 5w'¿ €_D‚_T‚ ÀÇ'þw .Pz@@M€ vÃbV„¹xÃ!vöˆ‡eþþ°Ù°øÅÅE¹þHÈ;ƒá   D؃V´:ôa60Âû¿J°IÙ#.¼¼žžž<@g7ÜN† à AØôÀn`¸øM tÿÉŒ‹``qûÓ¯³Exá`ÀƒÃ b†º=d¸CA`8àaq€¾ª@Û ý3XãÏ.À_½ðóðÿSî¯ìß… Ð?’660g ÔµØBœÀí§</ý:¹Áò@ˆÐú!àOåuÀ‚Ñs³C\n`øŸáÿÕ=w8üáVùCß­ýÛþã ƒ½À6X‹s0ÉP‡ÚÐö«jy*OîqéæãtvnßEx‡û .F {UVðüR>e¸eK™íBn‰î—ïaK=FDk²nÛO¿Û‰zS;mX “¤E‡òuý4˜ÔÜr»~¿\ýŒ‚Q[;Õ˜ó\ÝÅpu ˆ®<ûT¼êúË–GÃçvtw«DÔ±o˦¹c cž•Ì2ç[g"g@GpÓ<æ <ñ›½¸œ!̸§SKäÄò?Š,ö5[ˆ»þä³úÞ@À­‹‚‰ÂŒœõ‚ptŠÅWa?UlÞ·´x >šö‰´clÃù5¿Ó>›ÏŽ–Þüà»% ó„9/*±ê«˜µÖgIÎ%ôIèBÙ'+jk^ªßE[¬Ý¸´“ ÷ÚN¢uÂzØâ;Ý7¹u>ûl$µ&zö$)utÐ^áäÜ:ÜÈ»ÁÛX²ÏÏmVGV‰šÅq­”iZ:7í²j=ƒV è‡W¼†TgOÕ|¦#=¶8œˆOÎm2a»cn g Kä憆`éýêÖ”\%lNªW¤uÙ4î—â-Th”±Œ)Yy-‚ :؃/,jw±Ù3«öÛú=±®"û¿l—–ÝËWUä‰ea>È»nB‰Øs'U™ÜPúL±¡wÇ^pf\Ñs6öáìØ¢›OЧÌN_GËé5‘YŽv‚ÆñL~œ0•¥x}/ñD‰z+šß·§XÁ¡÷¾½Q€¥ˆÀ²9e®5ŠœŸ®í(d2|–i^‡©µºýw¬&½_¢Ú!8Ã’ÞfÃ:ˆæG[(¤2÷5 +.{´üAúFÕÈþñƒo»>—Ï@|ö}ó¾RÎèŸ<¿Žæ¦Þ„“„Ù ¼D N:Fº¡¦W ¶Œy9ñõl–]JPŠüI¨³E­iŽLƒÖ•ãÕE'3UØ–”'+ž[Й¥–’GacL~J_½>´ÝËìø[Q©ÏLæOžlýWÉšçtìáËRÃM1ß<~éÌØ2Jñ«2j¿Øq[-jëzúÃνºwë+©õºÃ2'=üê>9)ÿäç“\grv/Ö›­x_+Û9{¼(²·ø±G¿ìNi4Ë'ž·ia¼?Ã=¥­‘5Ê(&óšÎ4³¡r¼QÇé~ƒw&ºÒ#«›ÎBš)„ÒgÖ7ËžvWÄIå¬ÛÇJ$.²± *Š^ÛaUÊB™b 3á+<iθæü2ZµP8H%ž~";˜dWôn<–[;e¹ÝTf…!Nš!ÌsIá ‰ù †æU…#Ñ]¸»2ql꯷7a7¸#þ²jòqàdxCâC_fÁ—ý)ËaK·vå—ðérŠ­@ù\¥{ S­ ‡ê˜ùÒgüu)º¿PûR¦÷ó–ZNL|.{Îxzy„y&zÛ{â è^é_hÞDZäÂ$.¯Ošõr 4'bäko-%JfMå^Ï|W+F”2…P½—÷â-dÚ _±£ÓC%¨D].©×m„ôÕp=ðH–Þém¬æâû);"íýeâ0ƒ>Ñdò›FωävÙÔbÅÇky©FÖÄ¢–›Ž™9ï.Ô$ßÛ<ÁÒêzÍqöd1V…ºëmÑú¢_fÒl/m¤aѽÍXåZÁ>îp©ÓlÀF²È´}†š9¦ÀâZÃdn-Ùº‡áyƦS»Ž3»Ñ`3®óIë|y6 !¡4K;kï=é¹Z\¯[zè šÂÑÚ¤7hÞq‹”~¾zÁ±"}õ9#i iñ¢ª?(1‚È"‡ÒÒ(Õ‰£nK˜qîÓåD…°åî‘H¾±%Q·óH#ŒÑYåLÛ³º¼ŸÞâôJñ3ñ¯•}5J 2'H D#‘lÙß?“›ÀÁÒ"tê²ßôнÝ£ÒæÅ[Úãe:Êê¿Û¡St²´Dõ_jäv—kZ߯”«i»{‹‘×äí© O¶!_Ó!aÛΑrkè’L)|Æ›\Ø$”¦·ôŒ¼MøCÅL úÜŽaÉ<×Sj¿1Ž•˜PT¢÷Ñ@º—(\ò-}œIsz– 4¢ê~¦¿O?œ™Qd :&«7ÀÝêMå4¡ÑMîãÆV̲”xºb4çž*‚v4w’íØ bôæ§æe‹À/œdó ÷.‘ÍtŹWÌ”ý¦¬zç1ƒ”ÓH¹°9~Í2¨¿Æ]<'ÒI²„ö]ï¨Õ¹½™Ü™¸$’a¡æ¸úr5,ÜPœW£CLå¦ÀOÕÏH0åìøHUu_SœI}–DOûéá'õUcVeq[ëj#³1ï¢JVÁ\ÍFkrî‘2²+êÏ_/ŽìT._¥¾TˆL'RLð85Õ"¹Ÿ3ãf½Æ)ñü…Ék*Åá¤àŠ•#‡N#[å;‹œ^<a½nón6åŠ#y5zA7 40‡|Á—•ÂQ4³»²!e·z­©çz›É²SžåkœˆžÈ™ïY2º,-T©ƒ^•;‡£->ÐtÜÎÈpGÅÒj[WcîÅ”_!"®Q0 ŽtNžK;'³<8­=‚*~\VPB ±xºèÍBõô ½§Ô¥+Ý1î×ê¶§<š3¸¾äï6$3äù 7ÖI•²}D¦Zãœji[UDý",yLG™?¤°½ ¿0ºàý˜¨F…R†œ;Tól§`#]Dï³”DˆíÌæúÔôg«À"Å»Fg,É*®õð‹ˆ¼"üÌè-ûñ>x*çmk¦¿ 7Aœï³b…„¨x­+µîX)ß\¢ª/÷îtœ~?¨ *×¶ŸP“|“Q×.cYB(î‘;“àÍA…ÁªêŒhž©”i1§¾Ö˜¡Úêyåõ݂҇è;ô°‘_gDf4pœæ¡un‹ïæ‘éWö+ˆz)“¬™ªoŒNÃË,©BþöÂàÄÏÏ´"á}Ï*T‹¹´›0DX׋)6&œ êèÂ3±Bïõ´¿/1z‘0G´ ríp·±5Ñ«54™}ð6YÃì(¥·¨ò™_‘›çi |wM‚#ü5 i‚;>YI£^ø|ñÙÎA…«]pI/-|¯$ßn¬J‡9ù•à=èKgj—WÞ¬ü/›Ãþô‘Ö‰ü‚[SÐ*[(‡DÎ,dYP ¯y£cÎ -JŒ8¨B ªM¼sǧWõý£·p°ŠV§‰–‹|Tªº¡›…µ¤]"2m,·†Ä^Ì^’÷K•£.DÀ…ÔÇx[“ü^׃¯¿m[%¢u¥œ²(+£äÉ6Úà“ãÎ]ã(.§ ¾­å¶\dzTU¡é±†ñ…æ ©_]‹[_Ì8N¥ˆ }Ÿ#4ažìI°e -Æ¥Áfat?J^„½† ànI­¶•a¢DÖP>éÆC¡ðÒLPí´Äê8¯0g)|1›Ã¿Æ™îuèkZN†øióì)‹{Z룴ºõ+`ø$lÌe ÕÓ7dØ ÍÝ‹”â¯/YcØÀÓéZ¾/'SòãÈ3¬AM FOùۄ ‰Aê4ÝڊܤZòï6¤ÅÉì„¥¿[z²¡yÕ÷Ý/å(ÐVw/!ÄÈðýµrÏçTã”FB_'¥Gu–gÓuÒ%£|F d°ðGn|JÜÙß•¸¨l@$0䬑Â|Û’ò öÓÓÚn& Eëg±“øQÍ!U¢/¿¸R†»ÞWÒÀ›öÆØ+èÉò‚p®w¢X/NëºQëVUÉùbIKï÷ àl|˧þ”tŒ²8:T»qÍÜè·Ç/S){_:œ¼ÎJzüÔƒÚh@¥~²ÀƒñœÕ´àr îÜ,8:—(–&*rÎo—Pœ#«£÷³-\o£1P·ý%´‰ !¥K¸ó•×<¥ƒšÜ<98P[Ï\Bv+³£Ò7]ìƒtSå ó±ŠËÁx/è5©ÞR¥»h½ë“£¶+g'¬ðè Á\ú• qxRè!e,óßʽOŒ¾V˜‰²ÎæQ鵡á›[íÉ—þË¥xÍW ™Ðè!ÊÇ#¸£©×OèìZ?` !;ÞÂ<ŠEçrÔ1è2ûH÷1•"!»K5‘¨‰è©¤]ŽÔ¯ŒFx8Þ®#c¾¸îmË­›&¶x/­£èz»òB`epÖñ×1Ëá ö$¸FEt> ­X5õ*¦ƒO×¶›·5ñ3ú'ÅÚqf˜^ÓO .ùé¤dì\”$PPâBÅ!!âûKÇ”ŒÑEMVùW%Ëz2ŠÔ­kš–þ7á ÷f6Ïô¡~À'5¼ãÙú±|~ÄûŠ^°ÎbˆVL ºœ–\zÌ!qúÛª¾íÕëÏÕE=9x¾èZ83u‘WôÔ˜P±è«¯b›¾†ÖD ;¾ü$qב­äéÂP.épZ-ÿè¦sÜJ_ïy£é#ÜžiO­u²_Š5‰Ž;ÞDå/?»?c<`ÃЭxà­ñª’ÑbÛIî'óiÚRú.$ý|*Hƒ™¢ïù êâ¯TÃ÷ãæðÜûbh[Xš%/(«|Ç£‡Ÿ«€‹¼†¶”†f3aè§ý½”³Å†©>%½•PIMøU¿'l5šRȵ¾Á;{b —–¥Y8Á“ÑÍÒšW1ƒ`„¼_û‘•h—î"í>óoè ßì³1¶õîÆ3Å>~™£õ[Fªš[U¶k…¡"=$£6{•K–RiYzÂcÜž•;šx ñ‘×g©üÁFoq ²ÛÁj ¶ayÙ‡4zoQ0 ¾ÛÏë…ì±ÙÂó¨ìú_¶c‰}àž‹›ÿÖ1F,wÅæ”I~µKj…n€­™–ªSQluì2™oåã{Çš™é…uL™‘„õ‘Lê›b>I\+c±{tÍIõMÞª[v¯ô±¯ztsôÙe@¿ÚxÙ;ŽØùêÎwæošúbÝ8lì.ä^:;Úz”*Š­Ž …†?y"nàb&oI2:€Zšui*hL ÖÕ(¯­-rézÒD ’Ûc׳辷—]2ws‘e<ò*Lããzä¹+ ó©ûžÙ´òXuÓ]KáërÝIY¿‘qI ½‰È,8äíŒ7ÀßíÊ{,;ã+Û²¿8Ô×¾®m·ÅôÈ>ôîhQ»çd{Õnƒ…*O¼#Ÿ‘)~WCÊcbYërÖŒb¦z´•7Ú‰‘ñü‘|g÷äèwQÅ÷y…%„Åj -|—™z/¯ ïE$Câ*0=t6Ù‰Tªn·vtÈ™ñjm-˜ãµAñ-;[äN¶íòŽ›ù„2Ãs¹a tþr‘/&šj·îÑ޳앲£1^ÏVgÄÌýäÌ'cNØ…¼xóN!ÛÇãѱµ…âÁ … ç~`ÐTLkh=GÈÈuCåŽ&_K{øüùçcCw谺峸3çV#¬w\Åæígè…‹sÀÐcÖ~xÜ҇ŦqiÎ<žÄ‹†i.ÖÀ·=:ìKg“%9pŽ‚ !ã0—y¦ZM|ÇÒmï~ÄÒA+΄;Ë|'nó\‡A…á3NÿfÙõ¯‹†ð#´á®nâ”i|$Ÿ°GÔ1¹iœ ÉÛTÍÄKÍ!eF›Ï¢ß$],Q4Äò«,åþxæ½iŠ×~`P;6]>Œì–ÀÝ뢃^Þ­XW›ç*Œ*.î¶A·Ð›ñõQ¬Ah•õÖnÕŸãÙå\7`´ºµbXa8&f9 ÈgÊì°P‡XÖ®îSèùLIÛD¤²_Ù9ȯ&³zEX+耤ª¶Š-§ã¸fèñzó=>í8/`«Ñ'ÖpHUÖðê×¢ õ HkŽŠû›Ð˜/ë nÜÈ pj…€6=žŠ`*_$™óàåϘii+QÑlõìÑÙ““ÄúÑž¨ïÉO¿âfS߸!™‘áÄájóþ$WJò-;nRáÓ7sSºæöטïi¸­n bR¢^'žŽ%Ò—pû½»Ð3Ò„‹`‚˜b”{V‚)s[+b¿²1óXñ\~‚èJƲ"? ˆÌÙ¿¬‡òïØð >SË`½ìõ@Fèµó6ñgíj”»ÅprÏOmw­Á+\G$/›èì7÷F…³ÝâïŸõ¸ŠU MžŒÿ>ð)ó¼ „'H?sˆLޝ#:ñyß”EW¶µñZ>}Öçb—E`ï7¾$Ûøv§øUhÛ –²R,N½ ƾ"\â´úŒ†`göHÿ‘tá6Râ>ñæh¶q*©˜/m¢³¢áB“d“.âŽj+Û¼2Úl­ŒÀa £ÞÏUWˆnhWQŠÄ <±ð ¶½s£HËÄ™™TšA,^ÿ´³}«ô¢öC ý| Û:&ÎÑ«-ßþ;1®b©VVcß2‘Ö <šÈ„_Íý_3ƒؘI›±K:hyËc ¥Þ630NY‰D{š˜…Ö¦àå Õ!õäÇ)4<Ò¨3h§óq9Ü·Ig¢G§z‚ø’òÚÑbÜÑa†}¡gt1«ôç@¯äðÇñ;y¹hK-N™àVÔ¹³2â9S½úSio˜Û#ÚeÈâNAÖŽ‡×"<Ì’Âĺ«ç ¥%öxè1 ’’ö$k1ü¨{yba«ï‚ ¥ä†ª›nm(ŒÙ­7Ã=_WOsb;»Ç‘›2D<úŒ&¼}YõÁ/Ÿùþ£µ[z˜½sKÌÐökŒÞÉÜÛ Ž)?y “!Ÿ`ÕIt/«1–6YÑh–euÖé™%{CßÛÌ”€ˆ „ÕÆ ·Ð¹T†J}h©B™ÄG’ ,¡#LUf3(µ*̬D¸DQúçç±çmgK¦ê •X‰8þ‘¼»+k˜Å¹ÌZº3Å ÷e´m¦ö9¿±ë»®ô„m²Ó¨›2o¬PÜqõkÃÞ¹ùÝ)_¦´¡ Ô OBøËGÄŒõî&«ŒzIñ6%ÉL²ÀîÄ$ ¸õ•B®‡'^Ú_—“ù“TÎ.Íp)ª9V~xÁLü>Ë7öiôBȹ3H)aÞ1­¼áÓóA1nÇñÖÓ[K´&j­ŸJýžŒ1*G@%L9?˰ɘÉk°¡ º¨ê‹Ò'P:`eñª¸à{sż»0.©¤l:V8ê(õ¶›×®uAv VËñ“‹k¶E=H0BNtŽº·¨rßc³l*üÍÙ}‡ž²+ló.ç4°,Õ ÌñØÐõcê,5¨mCʱÒÜI¾%žD Rÿ2˜÷å¤ë=6ÌiÓá~Ï£©f'ôv¼ n¶¤¿¬‘tßg:=ûž;î<Áî{ôslµ„mTšG+h6~mÑ/~c]ß¯Õ Û–ÔÚÛD{¦N¬WÀœ¢]¿h¯Ãäf«ŒW¾S†•â’ÞÕhçÔž¨†#YÄcè¬×­`i¬àœÑ+m¼Ùϼ2-#õ£zQôëBóÚPú°³é#%µ|œùÕ ö‚×j!J$x] ÕzÕv”—ì‡uÓp^Ûô7{½^úM«§¯;à<›é{þ‘Xâït ”¢qGL,wÉKÏÄè dõV}èŠB@çE?-^ã%×i­ûÑQ _{/µ´`Ê9’¿áÆ]^ø1rZ_º|c?â;ˆÝÈ»t?j¾ ªkxŽå8Áj×cwO9ú5‹å0{œúgãDÆMœ“úY·ÜÝ‘LCLsQÒ«¢¸ëH´‹ Ü‘"'ƒ«0^QZ"¥u\¿, a´J39ÒJ¥e’•ý­‰{æþõ, x/³ãÃU3ö‡‘ˆ.—aÐ}ÙZ"”!£kÒ±õAƒ8~ÛO).æ£Â;µ­Åêñ^‡`°‰O˜ ã§ÀÊ'|¢n‚°~LYÝØUÄK­ {`œò•U¥ä×¾rX÷Ô4Wñâ6N-¦6}ÔìFT£¸Ï*^åƒ_Í÷l6.›pœªî³#¥W`Ï»"É¢Óä÷UV¨ ‘Ó‚ÑðG‹5ì 2/ësqgãð±à„{Vb¦2&…„í§Ä¿Nß!áü<¦ÞS†z«e\ÿL£siƒƒXã»O:‹ '©½S›þ×+²˜½œU™ãØ¡vmXv~{6>¤èÄê°˘ò§euüØ~ÁåËàK’C®ZLå¦Aaôi‹Ò;ãÛˆâÊ…j’5¶`6ÄbÇγð’ö ²„‚î[w¦È(*:2Ï«`átÑIü¸}éQìJ1síXçÜ’‰gz¹Yh”¹~™¿Å"Ãvÿ©Q {kP&ÏôÉs~ÖUãEžÐ.¾\ŽöÎaû‹ü†Ú™±ˆb)ù=`Ôƒœ­ƒ• ³\Ø­‚ÉÒ`„9Îæú•$'/µÁj†Õ€úFöîÚ]oöËÈ ’bgœŠ«c/rtq NrË3=‹ªªÍã ȸJ÷xV&t–/æ3*•¾'zòàÐȯ4Mš—¡áX~ݾ§ÚÃÑfŸupL’ýTØðÍ ?\0á{r{Ö­RóS-g0ÐewÒ¢¸‰;½·Ð½ŸkRXɺ Æ•-zÑí‚EŠ56o£¡æÇýqu›Æ,ym‹hhéÉFèÈ@úÊËG m8èçßvŸ×¹2…ê ³(lú¾sÐùñ·³éêG¿‹Çô½¾r‡Dµ?€É—</ˆÄ@Q{ÉJI'“G¶W´}téîa"ªiÐ× 0|¡¼:Ÿvõº7ŠÅ$\ò½rÈœkí)ïÐø«ÃáN|¥VF--«ã©tÍ5wR®|Ê_Cp¨U¢&š{âóŸ-?yEë¡â­’s[±ä–—½<°×žQ,¾¨á+Œf£‡ZÞí¢¨Ëߘ#Š~rHß›Ô3¡Á×öps©œÓµ3û'wÕknØ£O6G‰¨_ÂÝçy‹u¡üxºWdsÂMøV¯¹zÞ®Z4K›4°yKwµSäw7ÐêsÚú÷‡§É¾z)hIEE«#ß@úgIò” a/m‡”NVÃF-æÃÌO^èÊ«âýÃs ÁÐÝó2`¬éU‘÷Ø/2sJvXNG Ú¤hGºûÞ³VÐdÅ'ëÕ6µ»’›ÊÙ©‘Êsø'\y‘f½öÜk‘vƒd¨27oWJ·‹'¦¶ Ÿï‹¼µíîþˆù!©D.ÂMäâ®GH°ìÖNËOgmE\¢Wò3ºD+cÄÎæX„¦ Ѻ>ïƒÓep¬ÖЭ!¼k]9r£"ãǃ›¢©sè\%ûÕ*Í8shóÿ‹ܿb «èHÎHW£µ J9ùEbL >f!ÒÐ÷äøÅ·”ÃL ýÒ;í˜è×!rŸbÝ"£’±P@„•éåvS·,…n&…ò³ßDÄ)çEoŸåNY¹ º(›ÞyMÞç᡽mªhÞ‹™Ö!-yDžŽ)2_ËΧ™ä¢>þ?N# endstream endobj 199 0 obj << /Type /FontDescriptor /FontName /CJOFPZ+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/J/M/T/a/comma/e/five/h/l/n/one/r/two/u/y/zero) /FontFile 198 0 R >> endobj 200 0 obj << /Length1 1690 /Length2 9381 /Length3 0 /Length 10466 /Filter /FlateDecode >> stream xÚ·TœÙ-LðÁƒK£ÁÝÝÝÝ¡kÜÝ-xp‡‚»»w nIÐà™™;3÷þÿZï­^«ûÛUuêÔ>µë¬þ¨ÈTÔEÍ!¦ )ˆ½ #+ @\Q•ÀÂÂÎĆDE¥v±ýeF¢Ò99ƒ!ö|ÿ w]^l@——8Eˆ=@ÎÕÀÊ`åâcåæca°±°ðþ'âĺ͊L9ˆ=ȉJâàé¶´ryÙæ?3Z+//7ÃË¢v '°Ð t±Ù½ìh´¨CÌÀ ÏÿJA#`åââÀÇÌìîîδsf‚8Y Ñ2ÜÁ.V53ÈÉ døM ´ýÉŒ ‰  avþÓ®±pq:/[°ÈÞùe…«½9È ð²9@]V ì²ÿ3XáÏÀ_g`ebý;Ý_«'Ûÿ±hf±sÚ{‚í-`[@YJÉÅÃ…´7ÿ´u†¼¬ºÁ¶@Ó—€?*¤DUÀ‚Ñs6s;¸839ƒmSdþæå”%íÍÅ!vv {g¤ßõI€@f/ÇîÉüggmì!îöÞ °½¹Åoæ®Ìšö`GW¬Ä_!/&¤l– ' 7/'äy˜Y1ÿN¯áéúÃÉúÛüÂÀ×Ûâ°x!ò[€^~¼n €‹“+È×ûߎÿFH¬¬s°™ Àd ¶Gú'û‹dñ'~i¾Ø Ïò¢=VËïÏßO†/ò2‡ØÛzþþG™uµdÅéÿdü·OL âðfdg0²q²x98ܼßÿÎò7ÿÿpÿêÿUË? eí- Þ?)¼œÝh¸ý¥ š¿&†ðß;(A^¤ Ðü£|N³—/Öÿgýÿ±äÿOö¿³üß”ÿ¿I¹ÚÚþá¦ùÃÿÿqíÀ¶ž¼(ÙÕåe*!/³aÿ¿¡Ú ?'Ydvµû_¯¬ ðe:Dí-mÿ>F°³Ød®v1³úSBÿéÂKz[°=Hâ þ}×YYXþÇ÷2of6/÷‰óK¯þp^Æé¿·”´7ƒ˜ÿž;6N.ÐÉ è‰Äò"/6NN€7ëË€šƒ<þP6€™Éâò²ðBÏ`qBúÝQ.n³øoÓˆ›À,û7âa0«üx_"#óËÐÛýcù]=³ù¿ +€ô/È `ÿ ò˜mÿyÌvÿ@Ö—Töÿ‚/© ÿ‚lf‡êzAŽ®¹©­-ÈÂåo;Ç?Ž?UøO†—bœÿ¹Ìÿö¾0uý²½ãñ/øRŒçð¿aæêäôrAý1*/]úþã6<@fHË 3þëÚöÛjQBwƽ ÁYª=í4ZFïe§×{T„´UYAN×¢¾ô¢¯îHÒ\‰¬>y¶Ô#„·&©¶=ø<'¨Mïµ!-Má NŠÖ ¿&bÔÙ÷yrôÑ ´iyÕ%G•çèʃªR€uëÞ/íQ7Púu,laOu¿ŠKù±t†ñ½fŒA`ñU¾iö<9¼ #1"æO´¹«ëYÌÜÉgR¹z$ߣ÷ì½õ6Ùbïæ½ÖÊ5Øœ»ñ)ñõðˆa®0Ǧ©½Å¾§Èá.z—|ŠFofe+x’Ë6<ŠDZ’#£1Ú èó¶žïA8ºÞŒ¢ö…uŒîßøÂ#þô’j¯GŒÞ×Õ•öèlµS%‚$Á`2ƒF×»îò¹w=–›.ÎÀ›¼cé\o‰Š}딊D祗SïËà꡾TÔÓùÓ²@r Ò©§ALCõ`CeVøƒôJA:ÜW•tîN´µÌžýT!“hcÄóö6|Ì÷ü¨vA·1 RŠÛ“ZKe ·(v§=õ¤;’ÍÎbÈ4ªu¶Ro|-my³e0qÞà$\yW<1‹zGÜ Ý†©…ßa§ôRïhÏ\q~¿¦óضšißào –uOqx3’(#Á³ðS—] ÷»[ÆÐtå“, Ócè#dj-B³ÖPQª¯‘áqã`¨U»3&øŒA£nžÔtÑdiDÃžëƒÆüžHVJD€¶tá±ðCQ·zLÒM@=}®%pbEí„—=yKÆš?9ÆÅ•HT)DÝ÷ín¨ÿç<£¢Wm˜Þ<©@ªÉrcùjãevQD \¡–áŽÑûþ\!›‚XÍ:ÁË¢|ˆqÜК«—×)ÂR>êv_®GÐØQ¼ 1Û'~•G‰&öÌ3™P8‚¢)Cl›M‘œn vE‰DÔIIRýg› ´ÒF½ »CqVB)e!dä w†5M'3,)£++ ¼j X(ÛÆ‹=Ñ óÒ¢×'¨h …¿´}(q/ÖÖÕè] Þ 8q!;÷™›°¦G c˜~ZâEÓ±òä,ó7c§úˆa•qL¸ÎmÍ ì1Þtµš¦áѼÒ)ß§ªúÁ÷Ý…Ž¬æG¨8?Ü’7‰ z>? δBÚ®V' Ý’ìKWç´ÃÃÍÛqêáçbÊWp·Ñ\=C†2ËýG“FHBcJd«‰Oªz¥løQnÔˆ9VdÊ¿øÄC`ïnÈû†öè ƒµ³XäG%Â4ÒËëVB¤§F/V5æƒOSwmná¹ûæó?ýê†ùÕ l‘)F[£9-n!ÉÃT÷FV›Å{}{HzîÓIhò÷× Xb57Qã¡bXjøVmkð¨]ÛòÆáW]Š—×Ë>šQëéeóÑtãÇtË?{fT<ì:sªHÞnEóL2ëŒhà¿åîP cvÅFq¿Õa ¼ñÓÑŠ®´ÝXI€Îë{º@Y¤ü%úß Y;IæwDe;å?b©²žU?À®™¶i]ÁŒÎÕ¶NùŽŽíÂÄð>S¼#„7(núœƒº‚ì0qPž ÇÉYV oéÃh)-S1)N»<û°RÊĘ/#°‰8ŽÉÓ&!·¾¬gºTç/=ç«LÝvúe@wpGpcðqQôà'B3è«–‘”Ýj¤`áIiàýý-|Lq†ÍÓõéÞÄ'–¼5Á€iv³ó¶ç5žÚÑú=DòÞ[K©…¼63l,w¯Ææçï` é6¡´ð Ôº¶Ùò®ôÞýp…}¯º‹1pö,§X—eŽd:–ÞÑÃ[óªÙ5)ÒYC÷ÑJ¶pÓŸÄ.ÔÐ|­W …«bêªô*.ZðWaËuŽk‹Î„ äbΤáÛzv¥ÖÝ<ÁGJ,f!¯FSî圧wÖßdñá")Á{ïÓ;½L+³clE´"Ú…ï”4ËvB©øG¢.VCéY%iKéhUR—õ -mëéÔ× .î†ÜÛ³tqmUµç<éž@´€L(6ðk¦ŒŸÉ™‚܃AÄOÜZÞêxÍ/½5k&~Ó×aDj^åÒDi0~¹Z¶<ƒþˆ5@´õ ÅOtÐ-Šy`R1µŽghÄ †%×&-üi}QŒIÖ\Y 7Û=*j¸Ó!î`e¸íðhÜCÓyÏÈmV< ×QØ!ïúœ¶–Ã,nDPtç4Ör,Jî ½3ƒà®w…Z›¶ªM·Mšä¥–‘OøâÅ“b$†Dyu«¿[fóÌ=Åþ¹HÝo¿Â¶çÒn p®Z¬¾p|\F5q*W#>×·v.ë«Ð©®ßYqo%Ÿ±ñ=©„ƒF¡Ÿø$ù ":+—©¸œ—›%×6ÕZk2¢u¦$ûÀà1Ò£Äíß÷U+ÍãÚZsr|èBà°ÓšZ5ÄV4¢±µ!t üeRàNÅW;s6ŸÈ¿jË ¥ÞTÜÿáÂ}Z#oKs‹M vÖtXö©²ã•nð‡åŠU“Pä[~Ôé [‚¬tî\ÆlŠ‚Ú¸Ôòþ”>S‹ïÛÆ—KO„ ²Ôj .<¯‚ɬàèw›º˜`[jr: T‘–Xz/#Ôûd§8Kõ[4íŽØvKnÜb˜ïôuR£Â~¤Ëº©&¥c25|cê¦mO4°TÉ**On:ó¤=j@Ç}ÝH‡!¬ßiÙE5•ô5š<ò°â‡e;ñµÂK+m‚ïÁ¹Ø¬ãT~/UpÀÉJïYŸ~³rl—NTaL‘#äê›Ë»Êkµõ™W5 3çåÿ[cD ;ÉM뀠ѻaRðy܇CÓíq­‘KD].®ºW m›>äL¼åÌcÛØ ÷Mt”"þ¹ €©½ÂÜ}æŸýŒçöU3\…3èx7õî²'Ó^4úœVá<|ó&k1L`PÁ2ÕI®v‚C¦7‹’ù<¥Hy,Ý€rË9ÐJIRSPoR¶*¡ÍfÙA˜N.b€mëµ®§pöfÉèr§bé#6¾ãΊQ•,èã #!L`Sª°p½Ê{Û2i}òìqq~ÊnÿüøR¸#šÕ™˜H‰3œaëeÝ“wÍ—fcá¼²m°bˆ³7úÌâ 7֘ʔd™¬†ªnXGo†_k9¶…Èž58Sn­1Ó­k¾Š¯$“ÜðÚíMš›Íd…&6U>&=QrÏ^¯þ@×(O´D<ãž½&«tc®òYÔäðg©Õ'%ს/ïÌ„‰aõçÝÞÜ™¹kð{YëD4‚-é¤qHO‚ß6}·n…·¶ÿL³©Ð¬Ñk $…Â'£·²µ&ÇV9´5594¬–§Ï™d&,W ÂNËïl5‹a^$VØÃó*öÏîgœ|4½ÐÏetdгú®S]âÉ{G‘õjú»]$Å•bÛùÑE¾¯f8õnVꘖŽä!%Gd9 Ý£:äUÚ—w›qâoµho?o`&‡'›ÁVV4Â_ ŠÖ õϧ"©½-Îðæíˆ WoSì®i½úƲ‚úNN;{xZ€@ZY“ÃÚ¼¢½Îœç‹ÑžÊ¸§©Œ}›/U¥1uX)Ÿ;õšÍ²¬t B­–³ÆÖ1… •´ÿ*æ4Dx~nqn Ú2³E²ý ‘¯eÕ.œúõRZè”ëAå2cáœ37†×ã~Ö‘àõ<ÒCŸ"°p‰G Ï4}Ò/…Ä*‚ÑÕ%ƒÝ7‚¿ÂQZ?¯kž¼cñü°+×—oûÞÚ$ßUÇC£i¬íÆö±öÕ³žøg„FIfü•{–7T ëì%ð“î§`Wp¹çÁ…xJŸž;ö‡íOÁËᎸüùG7½$IeøÖÈ/–[¯dñ~ YÄü¸Ð™fÛ®dväm´©ß¬=Ò˜»×Ð «š…•I4ß]l÷ ¨¿‰Àe¹TÊ*™†nhÙTòC ¤‘$Žn0—TLØú~åÓŽÌ­á:baª]en¶Sü<U|ü5êèJ3SòÈÜMÕš´x^?Žma2·j ƒ-‹¢Vø#R¤Ê( .:ÍbÃSÚ=¶ó Ùó+”íî`U…‹18óE‰@Žp¹ Üj¢ÇWhlMÄ`¸vH}¬"J*Tö! àÌGK”¼öõ¡ì« ŠËù(µMæOÅáHÖ Lk §!÷ È}™qÛeêïíÜÑ&¬Kv¨ýæÑ­ ÉýBžMõ†z·` †MôÆÏ(hã6Á¢Š½¯k–`Um0BQüyR¿<î‹Ææ6‰Œ^˹]1%ºëf(ç=ZbÕ½˜w)ÝR%ß—Faën¡ÚJC Þ¤§úm[½'þ–XÛ¥U)¬?¯©ÉUªÂÅúz§¥0 ßíÛæHG2i§¹«< ‚½•w ý*ËT¶Àä4_œ|X( `ç6W(;“pg\ýú¢”ó] ÎÄÏ @o=iv†Vûª}Ú¬Mõø¢¥É¥µsÙÉ„Õ.!•‚²ªK<"¾š.òuÚWû†zX¢ì¢MW"KxµdEŠB*¿mŠшЇÊ?ƒ Å>÷é´Æf ]vë#›l€IÄV®!§;ÜQÁ$$%ƒ‰q”ŽÃ’(ƒ3ð 9´¦üÀÙwWøx1µ<ö£vÑØÑ­Mt_žìP Ù­ã“\­Z%À™>¦wÙ¨2eîzÉB=3 ¦´¼‹LPSHž¸òʸˆæìÅOû7ºqŒ{“é‰êoÔžìpÛs7Ñk/Y£A妕Á³¾¹ÉömRaq²ÎOûBÇ®å5€W 0–ö£Ú *â8kJnÙE¶.ž¯Äøð”^Å‰ØØ¦Ú`ÈûFœç/]ЇxÜTÝØ#µzøæìÄ(ëïÖŸK,Ì·hÓör4 Äá_™ðSÄOÔüÌ`F˲AŸ; ñv0rã??ÃïífUú¹@ xüËž‡„I½£ª»E‰CJ‡Té׿»R ¹÷"ï`|§zOÄHÈLÏh ;-ž•™!vŸÜ"/Ь#ëkíù¦@?r>ÎÚ­œT4B³‡ÌFŠgVM zîE “ªS N7½P×å‹)ôÉXOj½ ɧ©=¿/°• '¥ïÕ²á{gäÎhNîÉã×Í‚f»6á¯Èlº5±Û*q$:¤ï^K"×kø•@ñ­yb”:ÇÑñ~pHЦXê!¢'7ÑRxyûJgE£"çÞÐgÍçÛ󨾪äñTë¿?C£tØ…C¶jt-L¢8_kÎqg$ Û*έ‡.ËÄS¶Ø:JTï$#æˆ#`ƒ2·Ú¹Œ¸Æ ßKàÝI69Èá¦VÒ"ë5[!ªÕÐR%cÊǧSúÜÐ÷V–¯¶ÓÖª¿ñ4lKÞÇ4án½$U!,£`š9B´d™‘¿ê‚ÍÿÉRºH•êÁ“€“Fâpñ˜¨H~ü£.‡åˆwn—¬]–%B1½”ðK/ÔK[´6,woа±ÀÛ+Ô VwvF…9 ö®=¤|GÌõKi†ôWåÓ ÞIc¢»¯? 1Œââk'Ïv“S@)"^Ó¼ß/®ó¥lÖ,}~ÞÕg†¼¼×fµ:ƒI˜°Á‚å>2G3BgAßb Äm‘ã[Æ"ª‘êý”cpªõ.Âu :èµ3f!Ó²;sÉçyÖ@;[-©¢å ’³…ô ù \¹ñomL0J+Ò³2F»Ÿmn¿e@&iÖV«Ùr1ÔN*§ºˆêNÏC'Õ?5ZõMÉ%iÎËà_ú…òH²jGêé‹jF³ c‡ëª½ê+=ŸiËïÓPp*.°=…UÌWíu~ma‰ N7¨ùHªýú9þ²×#$»èLÚQùîðaîóøeŽVe ² þÚ¸§ìWàSå£ô:­'}j‚½…/ê-ãû1 ÚÓ +RùZ+×Û‰=Fxe™‰s‘mBFQ×–©ïNËŒ.ÀÇ›ú/t±Ÿ»5ÞT',vë2)n+×~Ž¥þþmhÐ:.VuÕþvn MjVL~kÂÖ\ŠÀ‹+yïpLAäN¥Eðªz>G–’ñ§Üõ¶ì‚ÀN´µ]vü0'‹†ð²óðtiötž²J¥øV!Ÿ1›jÅè¤9®èª(“¹ˆÿ}üÜþÜ‚-•kqÈÔé•Lj}ÝSTSjµ|p]µÈÚ"ïtGI©Cy UÊÞ íš‘(Ë|°Bûf‰åuFHáX•lÒÎy§3õÙG‰° ®žG<î×ìÞ›@Él ”²ó{2sÓâjï_w¦²Yùv Ǜ¤b‰M™×bc+òõ÷%®Çn.º—²ä óó!gq÷ä¶;#Ú¶†óÁ9™LÅ馘w/d‹¶#ôp§ÏC.ZøâHøBØöφߗp²ÓêœäÙðáL·sãõ ÌLQœ>J,6†ÆÖWEŽd…ß”`®qÝ×;iˆF×;vî6‡ïn ïëäöë%/Ý- ­ñZáK€6Šl<|¦RºùŒÀ6ùqµSÕ° ¯Û.âZLÚÀZ_ÓëHKå¾imæs]ã+ItLe”îÒmžc –-¶$™-55öól÷T:Æó¤öLy/Ù’”a¡o=ý ßöV? e,O XÁ厴6Y:Ð~ƒ-èÞZ1h——ÔoNKŸêÜÜú„V×ñöa=˜AèÛœ˜(Nùäêöqš–v|9—rM7Ô`÷CºØýüù»qpaÍõÍØÞ‡ÎkQyïÄuM¿)¯Å¶¥j{ËÁ²¥CL’}¾Ê^n°÷`hš@‘ƒÕþ~aýtHL9þ=ˆÊÈИÖñ§-üÞ“ä¶ÐŸ–\sưh#“Cxé|¦…ËùgÜ(IBŸ©LkÔ"½%¯ŠÖ}º'­úëŸ;E=,GK)á±RÝšØß«ø¢ò‡ M³þ<¤öêu«X~.®ê„ËÊ,£¤ W~j¤sn»'íØ†˜3Ò°3Û3"Š}§,Y»¯Å¬†Û©¾]fÕŒg…ß‹ãp º Aµ<ͦD0ÿ3††¦æxTÍþ¨ŸäÅàSäV-êÀñÍj€Ž6¯t™wQçxøÑù­Wì®ú×né:bÀÍÝW£!S~\ƒlÀ÷àÚqÔ "'Ë{[Y/s¾ò- üö%KXü‹{Á ¾>@jK%ØY ž‹rä‰4Ë-Çø~ŒÄR£@b5jÍœ/ö¿êf5Tâ6gàÃïö´›L;=Ø«†ÓœkZ±â™Ý à¦'3´ÄÄÁ¢}¾ˆk­+NX›ŒÜ’7-PÃ0…±D–9îÊ>1g’ùÚdèQÆŠ–ú •qU݈MÂË`ÎJ…qj9ñb} ÷Vºâ FƒJò‡ºå±=F\ˆB>ßlè=¹‡$#O¶çtï;箓êŒ}çDf×2÷ûØÈ %(ª¶’ÏKj¸âвƘ1m›|0 Ë|Ôrm¦TÖ&ï0Qtc¯qÝ»KïlÝ8‘IiµVX€T„m×?rt Ée†ª}Óòra¤±ŽÏڌʉ|zp½rý^/í’ûW=c{óœËæR{xJ“ùQ ÆçÄ<îß–¢G¡SÜŒ¹2Æþ8À¨R£;³;e6ψöÙ§Â)ÿ©ç!ˆ•þ`#Ö?ƒu™Õ# tt¯ì(Ðû®õö6%ÞŸƒ ¾»¸yã˳¶š€9wð—Ïy‰›Ge ­þW¤iÉFh<= X•W&éJ~Ô9¯T…w“W§m¢µÕÕRä—÷ÉÃùdÈÈÂáÁg5G$"¡=„½~¯jhy³¤iZŸuóD’kÇQ{èãÝZV®2€TÔµ?gÔ´58V«Ë8Áÿnƒž×mk»Ò«œoètn¤íœ×‘Òÿ×é› ¤yúPK(8äÝä•Á†1×íã Òèl$©DÚbæ>„Äþê½€JÝÔ2ÊûZðN¬zÇ\¨Ý“uµMîhD6JMvÈåG•da‘ðð8–$‘Õ:™ ­õö™'a©G‘Í[!Ô{}”²NÝ_Q(óžuì‹öKݦÝ×ÉRï4ÂìK5³-rÌ©ª5¥7L–éûÖ8®cÝoeª(6¶,æaNÄ«.p*?0f©¬ò¡bèÁ¨6q®ÎÏΉ ´ ù¼sh#qމ”1]˜H_,è#jŽžó,O?Ö˜ 8ï³M7f2Ç:(+¹¿ Ë"ÿ漡黆AÍD×Ö{®0œgðe戧–1}Û‡è/”x.FOÿÍýÕy #.ÃGí ÌAŠ49WÌW¨ñG4Ÿ½­Ð錞ΗõW 6,Hͤúe¶( †úoèÐPÆÑÊ´â‹Û‚~ Ê™©‰Xä¿"aõÓÎÒ.‘Ý-èÎ/”ðã)÷W ˜©`©rƹ›+`¥È $NÄ»Ôn¦{¨®–F:?÷%?ΡÕ8ž¥„¿o)Åÿ$ôÓ÷ Æ“¸ïU?Ú© W #_åK—oÙµtKey(S¿JðÇ LAmd‰ J®ð0ìrcòL›r]B ‹×j_¨€îƒiUÖ¬ÌýÀHp˧÷¥0òøÉù£rFª^g]ÌB‘¦•«]õ/Íœ“ÅV`ÆF÷×ë ýùó(ß¡Gé’¢«¶U¼™Æ‹êlóÈ!c‰~i½ j3ΧXðŒŸð‡ó=O$Õ6  ÷éâÍ£ïâÀtãUý‚ñb“f° ÎFºôYfS‚göWÆžº¶ó%"à…ÃÞäÀ ‘ᡲ¹Fp»µT^Í­N§òfoÔíFÎr§K`üÙìž¿æçº>ìN•ìú›v¿œpx †•6EÕ‘C ´=D9æü5е endstream endobj 201 0 obj << /Type /FontDescriptor /FontName /YVMLIC+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/C/I/P/a/comma/d/e/i/l/m/n/o/p/quotedblleft/quotedblright/s/t/u/x/y) /FontFile 200 0 R >> endobj 202 0 obj << /Length1 1395 /Length2 6096 /Length3 0 /Length 7041 /Filter /FlateDecode >> stream xÚWTì»§SºQaÄèÝ]Ò JL`ƒ1`t—„4‚ !ˆtJ‰„J7H§Ò! !q§_Ýïï9÷ž³½O?¿÷ù=ï9²ó+9Àí êp’_H$ PÑ5€@" 0hEºBþÐÍ O(&ýßì*‰Ö©‚‘h7]8  åå ‰K IHƒ@aHê/G8B  ö†:tZpÄ“¨w÷E@œ‘è*\öÜ!)) ¾ßá%7j†tÁHgˆº¢=Ø` ·‡B¾ÿJÁ%ëŒDºK úøø€Ý<à'yn>€é 0‚xBÞÀ/¸=°ä70" ÀÄêù‡Úîˆô# ´Âjy¢¼`]`¬©Ðw‡ÀþpÖùÃðçÕ„„þN÷gô¯DPØï`°½=ÜÍ 󅜎PW@_]G‰BòÀ0‡_Ž`WO8:ì †º‚íпԕ `4¾?ÑyÚ# îHOO¨ë/„‚¿Ò /Y æ wsƒÀžD¿úS…" öè[÷ü=VÜæÿÇÙ spüÁÁË]Ðõð‚hªþéVý£s‚ b HBR ñ@Pö΂¿’›øºC~…~©Ñýú»ÃÝŽh@¨#ýCäï ö†/H ÿ7ü["8@í‘;ˆFôOv´âø‡Œž<ŠX‚ÐÄ€~}þ>Y£¹å‡¹úþãþ{¸‚÷M´,ÌLyþÛ¤¬ Güù…A~a1@HHJ >þ;ËßøÿÂþ[k†þÙ蟌š0G8@êè»û †÷Ÿœàús]¸ÿ® Góàú‡öV 1=úKèÿMþß!ÿçeù?hÿŸý¨{¹ºþ¶rý2ÿ+Ø êêû§Íb/$z#táè½€ý§«9ä%Ö…8@½ÜþÓª‰£7C æäú÷%B=Õ¡(ˆƒiïüþš:½+1€{B=3~!è?lè]³wA?%žèIý6AЫôï’j0{¸Ã¯€°/zðhI à/„^Nê7¯‚08@à 8ÂD¿æ)*D?V¿”¿etA¤ü·ü¯Bö^zù~ÝÅ_òïM‡@P{¢¹i¸½LÄ£ºˆ¶³¥Û>ük#r“À5ó§Üüþsˆ·^¤øéÜÕ9a+ˆJéýïÈ¿¨q(Î3_ùo·4àG·¦¾ùpi“l4¾ö†hvŒ¶g´h[©¾û.á~Åõ€+³PìÌ-`¾‡—$©A!Õ™Ï T}wÙÂPÔôšázµ¸6ñeÙ‚i¼UhÉG`]î=+’ÿ.åŠìãÉIʼÑf­d^¢À‘b‹UáÄó)¿¥ aÏNv ú»Ø'”CãþÊZt3þ¯ŠWC™S´o‡?¹¥ ¹npù­é!6¿?äŽJs1Ð bSkÆÅ¯´ÞKq+aIÁÍ=XÔZAi~—h±óäÓOÞè»JÔáTL¾dY·£¾³Gûp{e}¼a„KMžT¤)£/»&ãz(ýžð¬Xß8¼ð“1¾Â{©¶üÑ•¹O«™Ç5”ygñÌãwø¯£q–âù|è%1t< §›ÕÒˆ=îÕ¸ƒŠä£I{׈6ø›ÒëõF2QÈÊìYúœT^k§äš-*‹Ôý‘×$WpOŸ²Ô CÓPîœ&ɸª>ðãÏJÁ½Þò¨ž„ƒ÷ÂÚl`sA¡Yq ûÌ7‹ÁqÆÈìÜÇ·&´Ûè%b„Ûs+UžÏBA4nB^ aátS¢±ÖÛßæŸV; ߨ‡,º»mQá7›RV“97çOÖåu1°û[Uk9àKòvç«É•tÞ½5 š%'&ª§ó4,EÃãß«Žr/ìN<ÁKÜ —Mxͽt¿IRz \Æáß#îsqà%·÷6Úã<¦I’}ø±U?\F]»¨ ÉH%ë›ö?–ãÉ£¶“½ÌE¿ü7MãcL‹€œíKç/ñž¼«<© ¦ü9ÕÌ+dV®¸îtÉ©Z:p+^.q!'Ül¯æRbuç2•äÎÅôØ_Ñ fºT’9—¯  ¸Sü–Û¢Ï~‘f¦-N.0)}T¼ó!ÖO'Äë£#î½jw”Àq'>«'Ëù´ñ+÷Ù:o= ‚¾L_Éô ÀìçÑ‘c´Ô6»nU%{›c›5ýõÎq>æ^6]_¦ÄßYÒ6¦È5%šõZßyPÞ,¬l#ªlíÞâ~,¨m”Ö‘Û[Æ-Ã.ˆµÃ´‰¸ioD dˆµ>ŽšípëG­ËgÜšzÚ Ÿ{AASÓÉtŸÅòKØÜl>N¢íy¿7Žã÷€ÀOFI„¢¬q´xAÜx^4»Þ·VÔØM˳ڭâ^›ËlÇ?gœÐUÞÃ>]\²—Tià¢Ð=ìûÆ÷&F4²•J÷ÇÒòÁ`ЇœŠ;à€.£YÑa®«·)è¨_IÈP2L‰?`%ÖNÈÖ¯²ö !)cá奅y¡@/b¿ôV~Ð×^ÏŽ,Ö×v[tÖF¾YYYˆ>U¹³N¶ñ÷V…Ÿ¸îºÜçLªkVÎ}ýŒëôýú 󦳓OÖÈά¾Iù>`nû£îüã{Â~¡:9΃¤¾é%çÒ÷ßÃ]zžÌm«8ªå„Š]¨[?‹óÍ€9Ür  ÃÇäêâ¸þÖiPS¬,õZå/Et¾¼EøHÿé”O*“bTŸ@{grW‡þé"Ž÷ê{|ñÔò‰‘ÇñwøÄÒ°XyU{šÎ:²ÖøJ¼âëZï´^K}[[èHÚ£ >}$Ë"è*\þ"óÝ×­1ÆžôÚN§3Ì”^Bº5BêåIÍçÚ£öVÇ9;úÌ—¹%Q“I&[§4K?©¤rï¨NpmÚcMJí<õ„Ì0DæúìEKJ,±ÔÔÀ“=SZèíi·‚” õŠ ˜ÏºîÝwêÆgS®ŽÐ0%KžG¥C?祵¤í_:/h¤¥%‹§I\“ eʱN'p:Â*ù…sL ÜM½ë/”Ì{BEdOð=Oo´òàs²Þ›j{¯‘Ä7¥f¦µ™æ«•ª(+ã½=VÍÄ*¼uoÒà Í¡2Œ… ņØ6\g™¿%cކãÝ´‰L.ÛžóCÁ/mšuþ>¢Ð§ú`ÖŸ7ö­¨òV[€þüg¼¶å’KŸeÞ;X2ö1aÝ^¾Ê;2ç¤_=pRÂL^eÇöîôbQ¥ÊÑ*– |©m|JÆ6KÚCüW´¸w‰ç¬Œï\“ê`êf\¹%é²]En>xÜЖS>Î>–¿S²uŒs»'à¥xKVóZVRß'Ôš™M¶]5ôx¢óK~DÌêtµÃ. …ŸàVÑÅŠHÑê#šä"žN®»Z˜ϺÙ9«‹*•¾šœ56Ý,C±6vÛ= µ¾O:ÕºžÒ~z÷©T•”‰ìÛyÏ´·°_DöåIÑZjú?5¸g #5ò²çàÚ+£ý1 Z,óÎ¥¨;ÌÏ“9ðß¹Bá¾± ´ãØåAÌó÷(Ê …!ƨUÌ0I¦å·á;YŸüÞtn6mäàuøÀøÌ ò©«üžaeÄçÈLóÅ #͆ï~‘&¶šðzÙªœŽß½{dÊ¥â݃í£po†ðò&C5¯ˆ…ܽ˜ o½x¬fbÙ‡"Ã(ŸL¨‹(n ã®ç]mhÊ"Ë#¨íGE‰&5–’&‰kÞsª«MsFÎÂêìÖo°r­ÒIq¶ÇÚ¾´$-ÓtÙ2˾^V%ôÊ'gã]ؤ—‚ããV콆ÖÓRr9ùйÇF7ÿôoïÅ'¶x;ÂlKßÊ}ípâgîį“dWœç~ú“Á×jÁËɨDœ`²ß;ÒY¡ðúZƒî¥œ”©ðRÉú§pÚú•š«—LGIS÷ˆ?wÁÝœçó8êÆÛý¼$¦oûÙÏ{GZ©ˆ[œeæÇ41 ‚ðCG2ŠÂ¤™àªg}¦o+éÕÊÏÕ›}á2 ùë» Ë  ncòŒ¬ƒl½é=5$Ô‰&bZ,<ŒgÁktçÆtŠ%ºël’Ó;ü/KÎó}Ca¤ž-Ÿ§Nˆ5ÕhDÔš,O}ì0 Æ?¦$:-ÏÜå5Ǩ5¶dÔßr¡I6psˆÝä£(çM3£ïº*94Ù)ˆr9ó=!7’e'¦þºBøIøLÀ1ñ£™6/tV®XåcþýšOz#$R´leuŸƒztãNNrMj;”ýh¾Ï·äR×ù•—ÎXSRyÆ•™lO(Dd‘Fjð÷.Œ¤t;BÖü*׎BÖ¸EÈ’²÷tÛ~àß«|ŽºýîÆŒ?¥X€kÒ³æBfu0`ÅOß$oê9þýí )[& ܈[—ìæsEûᇠãrŽjÀEúŒ ‰îªÅKnÇ~}¸¥Ó¬?µÐDÕ½ž.ÆõëŒÖNãÅAiÒ˜¹ùà×Jë¶âç»Ïˆ^–=0kg+8ÿÜڜݢ{Yx`ýòŒ§·pæ…ÔŠä9½Âmþ3§Ú „Î>’]X’¾4Áë' ”S€=¯§>ÎU‰?óa‹6û¥~¨°ñ¦¬x:Ì>_ˆeM©Åï¶Ù˜Ñí ššYo¶fëxM\:átRUO¤ÉVúÍ­Îc¼“ȱ|C”ÄhÒÖ{Þ9}Çp´Vx$Î}ÓƒzVæ»À¢÷D¬0ΉoRZ87.;fO=¡øÍkJÍv¯jcRyìsJÀµTÝ•ÍÑ@ú0ûh8,7Œ:Nïå‰ÙèFíÂó¢Q¥>e^¬wŠýœ£ãO£<ß…GµFÙÈ´âgU4Q?ö8á²)½jΕÞm˜“æÞ £e²'dTÀ±»Fv è¼Ä èüh0çÖø®S0Æ«z•MB‰Õ´ü=wc„‰ÕŒjû ^¸Ÿ7þa,xî}ø6TUÄÅ4ýزsb†—‘o»¼nú===ß 6>ñ®(p%š—êçew­ … eÃuc\2έøË®©¾Bo¹–ýÀͨf•˜²œ_[ñ¤<ÍëÇO°SË’ø@j¸¨}šÐz«¾ñ¡r#‹ƒ>øv¥µ9ЩÓ~Í$ÿ{Îr^c©μnŒpWËC½GÁ{¤]–zSŠéù<¯Ü]—asv^Ô–øÆ&,S­²ÔXö³V˜ÄG…òIub2ë$seƒ¯j¹VÛ°]ð»X‹ˆ3 Šbå©öW™çq‰ci¾ƒÞ!éRâšÜš¬„ì,ɦ_¿–l?§púmu5P„›¿°%%“e>§fñgxœ«ñr¬ ÒÆ¾Î¥Vï¥Tó·aÌuÃÄÞË2ƒø,ýѴܵl¼Ã©<á3QààQCpE 0¯Âp¯u‚¶ZP¦+9?i>ùíq„±È"tmRk]UÆçž}# aìÇ¥™Ñ#c{Ù›”X/Jš&ïÍÇ«»|(´õ­ÓÝ=ÖæKuâN%èjr=œž¾ÖVqÔìTÜn!óD:$2hÊ•âÿø‹œ‹G¸ºâ@ SˆìkÎì2§´ ÷ô²µ(óß¶8Êþ›ÎœR-8|î!Ké±)F]l k¾ísægŸ‰ús‚äŠGUÆD",ÛB’W˜‹4Ï/Lôg Iˆ Ãoh˜Ÿ§ù.r«Ñðtþ”ï¨à\‰^ ‹!ar& Ž0 É\²åÛ8ñFˆ˜ÜM4™ô_¹y+B£ß¾pµŸkÓ˜(»=1©ºÃKXrƒ¹ìL³Ì['ÂD5F%ûi *°uàsQºfr„½@X•+´éê¥\yå^ù$“6¼"×…=·²¤ƒG&Vol8Àf®NÌGÍXzÊ6ñ³~ƒ‚[44°Ða¡ÐO×ÏvÄfx’Oy8úŒ¹°† ^[O‰w? ŠsË>öK~½£ÍUG¯9F9ίù8œ{djµç†Þ*…D§R]êÃUÞ…ì_ò ‘IP öøÀé0ÖV§úÏ4«ˆk£e³êÔ Êx˜gV8Öýœ×_5e«¶iÅ,Ñ$ý·g`Tíµ8üLÑrc%Ùª=]IívÎþ™o)ËdÒ© DÌ'˜p°†â*m½ý$dΣ±€$6Ûµb²Î{‘ž‹Ù^¬ð&W÷Î*8†üéÌkmHß+²J*–a>ÿëææeù÷ë Éñ- ΄G ? ÷®{pžÉ¹_“›±¥vyIË¿u´%eµQÎ…ÃÒX|ôÚ3‰G}UÄ^Œ"ìæ1žHüQ|¡Ài&7Å4º}{¶ª²€H9>R¼uàÇ.üÁûCà+ÅÒ,ÜM¾E庇Ÿ¥±õºµp¨;ž߀Ž$”F¯” ’§JÕDóûý”*7–IÍŸ †¾m§ï͹[´žºËõûçÁƒß túcã,;©÷Ò¥ØH4©]ýþ>½Ç2ÌéHÓÿ½ƒŽ’^JO0a öÈN” º‰R³µWî´wWhy‘ÇRj’¥ˆì¨¨®²I;¸i¡÷Ö²¹r©B%jµP„µQ¥ ïiú’­^AüäFîÕÂ|Ê g7¾C Õ¦—™Ä%Ïw±QZ~Ö8U^SèÉ&Ó6ïÓ÷c8S¸xœ€2áð&2î,`>G9µK2­ýÙ"kö÷Ý G™2&üx ÅÚ¢ÔŽ1ÜŸó§÷¼8Üæ²eªõf4°(yž1’ô™ìœÅËmY’ÓZÖwY"¶ýïÃöYAõ¯—gofÓø\¿Ñ-žæ¥¶éaU._Cö @¨¥¼¹ÑÖ£x-©ùÞ fúîQR7Ìûü7ß5ðÞ͆äÄR¥ÄôTn HV@#ˆ8·ú¦:嘨ø@fwI ð#Pd±X3»ä’×E.J’TÅ\á;Í>Ë Ê83‘îr÷Eí0ÄáJg øj?Ë„èUzF›¤Q[ gm[_™Ü. §ëì ÄMž›³Ùù®U ]„ }&Om»Å©¿Õ²6ôfzÀ]zoÓ§&•¬›þøš½eÓ]ªÿýÛlÙ쨵~ì(rz«Ñ– ‘dòˆòË’eRfÒJ=[×xZ#ŠÃô®Š{ FõS¡2µ”{ΕE-‹9!ïoÑI—W ÁÈ‘\[lŸo/'eNj“$ž¬ÜÇÆíæÓÀ&:Í›ê‚\3ŸÌüü&H÷_(å”Ê endstream endobj 203 0 obj << /Type /FontDescriptor /FontName /XTJZVU+CMR6 /Flags 4 /FontBBox [-20 -250 1193 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 83 /XHeight 431 /CharSet (/one/two) /FontFile 202 0 R >> endobj 204 0 obj << /Length1 1532 /Length2 7365 /Length3 0 /Length 8385 /Filter /FlateDecode >> stream xÚvTìÛ>%1:¤•I(½”îî”cÀˆîFé”RAB¤ Aº) é.i)i¾é[¿÷÷ÿŸó}gçlÏsÝõ\÷sÝÏÛ=M)+”%\…tãð‚E2jÚÂ@0˜Ÿ æ°±é"Üà 6}¸‹+…ý»Œ ê†Æd¡nh75¨ìî„ð!B¢aQ0È‹üåˆrÊB=V@5^ 2 w°É œ¼]6¶nè*-ì0 DDD˜ûw8PÊî‚€A‘@5¨›-Ü]uê `¸›÷¿R°?¶uss<==y¡Ž®¼(qn 'Âͨ w…»xÀ­€¿èÕ¡ŽðßÄxl@][„ë°ÊÚÍê¢ ŽtE¸#­à.@tm Ž’*Pà ŽüÃYõnàŸ­Bx!§û3úW"òw0C9:A‘Þ¤ ÐájÈ«òºy¹q¡H«_ŽPW:êE8@-Ñ¿ÊKi¡h~²s…¹ œÜ\y]¿‚~¥A7Yi%ƒrt„#Ý\¿Î'‹pÃÐ]÷ý¾V{$ÊéûÇÚ´²þEÁÊÝ ¤‡D8»Ã•dÿô@C€0¸P  ‹€pg Ü f ú•\×Û þÛù£Ïïïë„rZ£)ÀýÖpôÀ×꺹¸Ãý}ÿÓðïZ!`n@K¸  ø';†[ÿ±Gß¼  øŒþõù{eŠÖ– éàýûïËiêKëËÊqý&ü·IZåôåáòð ‚P½ðÿw–¿ùÿÅý7ª Eüy6ð?•Ö( Èнû‹†ÇŸš`ÿs\8€ÿ® ŽBëdÿGö&`A0 ýù?‹ÿwÈÿOó¿²ü/²ÿïóÈ»;8ü¶²ÿ2ÿ?V¨#ÂÁûO;ZÅînè‰PC¡çùß®ð?†X n…pwüo«’=RH‡¿›ˆp•GxÁ­4n0Û?ô× Ó; pM”+â×3ä€ÁÿeCÏÌý”¸¢oê· Ž¥—”CÂPV¿fŽOPuqzÐÞ }!èá´‚{ýÖ5Ä‹D¹¡C€hzþ@k” à×} A€ ¸³;ºëhø7"ÈY#ÐÒþàC(w—a ‰>Ý߀€„úÏ=:…“ƒ»ë?@+ZâïÑUÝl]àÿQŒFm<-—bðäYa[5xÁÁã;éÒä~N‚—Âñ>#dÞåD*¥·lfYŽýXrŠéÊw«î^x}’VÅߥy¢öðj`bˆúÓàÛ-©ª®;Œ<º’k~WÎ~úÁöØu˜-ÊlÙÎîI4ßPžzv*xUuO _ÕZ{/¤BxYü•'V/Æ$¸`”-Çòõ-3®Ï|NŠ}/ÒÑã“ЬÁ&åD.€ÿv,ž¯ñ_ÜÙ˜Ïl©.Ÿk++1íìcŠ/Ã÷}¥7R•i¾ùæÍ厚¸çGòÏ´b˜*Œ/?öF$>¬èÁ¾·f³øs»…°¶É­Q6•úv튻$°Ì Kpmý|œÕH¶ë;~§˜¢¦’¹åò­°Ì釗·—„"ƒ”YîÑ‹‰?HfÚw‰žò`ŸbR8¢K-»"fÒ"–'QûÐe r¼ËÀ™ƒˆfÏd¾‡(j˨ RŠuå^X%q+›•T?c ÃáÞ4 1{ðY§¡¡ÿÊÉ@pPÍao[4‚e§á¹Ñ4ã¼E°oŸ;iÍ\"èå—ëÞØ]æ$¤ÄIþÀ˜ADGfRêX4e+ýÀì|‚XûŽÔµ¨‹Xãw“;@Gñ:Ä„©œý)Õ¸0?¼•äUƒ.TsânâÕýç²JË6ÄC[¸ëis¹)âüio½‡|Nâ¥Gíç“WAšDú»«…Ô ­NûÄZ »½±ý‹Öôê6¹=þîfµÁ>¡´ºutÛtkm¨f¯nÌn}þW¨9ëÞw²wĽî²VTaüqZø'iÒtÆ:ÖÞîD™…^í ùõ‹óÃêg¯³b  _—r íS„iF”ÒëÄÙ.7üG¦Wu_ÞpÜfUb¸q^x®©ÄRÅ£F‘4¸¡ïe?@ÒîÕµ-9×8:Ai<¯,d´÷niJÌVjª|~à ®(Y KÊrÜ£~Ž o6¹b:"ÍÙÁlÏéºNêݸu Ó|/íãÛEÿÆClþ´Æ µ9«ðêoä[«¤Õ€l€·¡KsŠl™†qB¡\/ï`ÂŒØV;›YaˆO¿µ–†£¶™þõæ×®îöùýǽz:ëz:¼[ŒÚq|`Z‹¹ÈºªŽG Ì“-Ê8ÕçŸCˆ}T&Ôî8V0õçUäeì_Np™}¯÷ªj äú®¼˜Š=·w ?*yIaõ|Èz‹×s>&î5%Ó^ܲ%8Œš|´wÏ3gÄì{ÑÕDgÞ„G2;öѰ7On¼ýGãHC.¾H;ry;M"N¶ÊîÜ–ÚNþ:¶„¿Š''»ßKngÈz<ó"–:á '?ªq±ìEˆÑ(Ø—K÷üª:¨K|0"âXvðÇn%å Æ€µé—pfW'dHy6_„À—ÿ\LîrÊ®l KàxÔXý3ãý±J§gp½Ä5æõÛ]–ÔÙa8"ÅôŽp¤nq¼Òô&!ŽðŸ1«vtgß͘èæ;ÊI.-÷ pjÇ}¬o¦â€Í4à£ïý³ì-³\ÞCR²Óœµ®•ÎÍÒvM$ëQ:/,LÌHéo^]på™f²B» hÊØ%O)àIÁ¹"F¬t–Å5߆R^ì‘f}ÛbT\³ùf}IÚÁ%VLî½wì¶Ú÷%Ã4‰ÿû5áK=ÿ!ÔEŽd Áá…YS¯ÜA‘ÊW5þ§èÞEöióçà¥U¹¹iæ0<Å÷nŠyÄ´õ#½ùÀ~b¾ÐýVCú=—åÈŠ”Sñ•R&ûø ].y&~Bé…$/{T9àžíù´‰3/¤¹\F¥`ëÓ™ÙüÏÚ<éi ÷>”Š‚}ÝéTw;Þ IÝÅ›¶ó÷uæ70“Æ(zðz–ýŠ’Ö(£Ç 7èÔÞˆ×Y”Àp§8Øê$ºÂÚN3îb*¯I‰ˆG$"?,}›lªÊî~Á±•ä[íú^lÃÑ$$8$¥ÅÜÁLÁV½‡iWòbk1-Á•Ø §á\èvÒ$cŽå¹Zxð ÎMüšØÅù]B«+«[È©ì:ߌÌò—LÜ /!¾ér'Û¼ãÚ^ŽêtõöókìTž”éuü`µ3²bì–¬éµE9Œ(º*ÀÍÖ‡yü—ÇNW;ýn;Þ³ôó¨Rñ3¦oÁ5Ô<íýG“Ö¶bÀÙ—%¼©»nÃJä/håFšv1û#Mƒc­»Î×õ7ËÆ’Jæì„õ¡æÐÞÖÃcf_–‘šµžàóÛ€ÑuFK´é'?8Èwß5âJbMÉm5$wÞüÔ³Ézl_í\€¡J×Q%]²6þ™Ó®t¤žÜ–d\)æë­fÉ)QG^Þ0s[±“P:Ä@lbý}Õ]v7­‰D%kuÿ•mr¬‡—©#V² ÑÎ:'³ËîiêÇ–˜G&÷ÝvWŸˆ2q)ȉUÊs1«“†Fè·Â_8Ò¥I¹Iê úmŸ f9Ns‰•ëwR*âdk-ðø]`e;9¾37ä5ëY ¯×|k¦ëy¢o˜¹—›ïÖô^þs³¿ËXwQ@vîZë©ê¦¡1¶Û=¡§>LÇÏÔ´FÅâ4ûöd0VŽTÉuº>9)/ŸÑE˜Ãæ´^‹îÎcØ%}_.œ;”dfG’mô™ŠŠê í¢°´ÓçmXL÷° œ«3öŒk —‚^µõ!GF²e°_oìy3gFâoUÚ±pu/u÷˜¼÷MêOôNY/N1j"Õ8ͧŸö”ƒ?wÂ׿8¼ð¾÷»T25¶*]ÏWW£š×¬ô•Y%Ëüö¢ï6…ÑÄrnÒ7§{{îS_‘‘©d‘Þ;x3F.ÛØâ?ÅvöÀIã•/3¿c(•š#ï-h ím†Eæc >:Â磑⼆SPè\šÃÀœ |ví¢ÔÐ;ÃÄZ 7·sœ‹{ \n$YxVßS(x^´AöÑW©)nKœò‹÷­õJÔ̇ÙíP&)>±Vé3‰ÝÜÆ6fíòr@›y/w@ÍÅOg^9? õe0MWèBÂ0Êò‹@èì‘¿ÛOxfë56-ÖèÔÍGÕAHÀÖý$U+ã~ÿ#¾0ƒÃtrÞ-‚¸PÞÝ3¶~μ¡;UÔ}²“~{ÆU;O…ŸÆ@ö1ßqË>Ö\†ôÙ™žE.ú(7Sž¦dÏân>¿óƒJ–>KBÓüfÏm•Ó¨•’ õVuüñÆñÁ ã|xv´#-•®8UÏë˜ ¿Ä8¢GKO´!Ø«So¢S,ŽÕ’’¨šä©-ã~#ïh—þø¬? #™+Y,—H=±U}3aaòAêï00íð¤ŠÉÖŠMt«;a<ãcø¦;… ØåÝ) >mécHñ+6ïºK¹O;ŸP{m?ÈþNþÆ€\~È™yÉh„ÁÆÌkƒð_Gz¢ô^ÚA¿ëÜÃF®åÙ0âÔÄ‹ÚÉï²Âu½Ô²— ge‡,«ï§[Qrjšà°S²øK´œIpçdÙ×ÄáÔÏæ„ Z?ñ}Bµ;#‰i²(éçb~ÙÙHVl>ËÛúéÍLG{ÕTÚûîû¯Ì¶ëL?󱄽k ß/x S­~Ô‘£n,ÏMÑÈœÚÓû,*zçLTEf`BzýùöU¦,ç(¬|ÍyŒÏnoÛb.W·w÷/js{¼Ž&•³ÄéZ«7¤£?ïH?KÃ{¹ß5žAušËr»ÁV‡B[ªj[œ¯s7_cL”ÙyŽtXvS#,üØÝþ[ÀÜŸiº;“¥ˆ#ÏB½%â¬^‚BÖ`.³Z:)ýxI¬ÄfKEƒ:‡6ÝÛ¹>šÄn¡Ü÷îox·vn_¡7Ï/ÿ‡2Mµê§Å¹õÙ uPqí»ñ€þb÷†(å ;OÙAZˆÈ~Š™°ê &{õ×{Ň.]4Û2 ên\süg%߇—ïQ5= ƒØtØX?~µüq«Æ=• æbIß ¥[g^"ÇV¦P5–Ù ¶Ü$Y ˆO å(`q¿†çnÌ.àmp>~CæˆÛåÏ€{s×?2úq ñ‡’Ò£ÓSáš´e䯳«”µ12¡È“-XÀ]‹w8(5‹Z¥…•ç 2Qt°®¥XívvkrÌDR‰¥”ïâ\xF*x 2ÆTÏØŠ“]©8œ¾þœóÚ?¹QÔ ‘ÕþfûTõ+2&|š|©¨H Ͻy"9ý1áUw‡Ï› eµÁ‚ûú'U¬VÝ´ìÔahv¿ `u˜®ç:ÒRl u:G][‹Íˆ„^Ö\4œk>Þ¡l·f¸5Á¯ÏÏÔ²{ŽçN¶y¯š81ß-pÕH•›EnóÚ#´oüªö¡D©äÐãïä7ŽE{O_-µ×^´í¦Ñ´ ™éû¨Y4OrOöž °|ò_M«›†^Ï\à’Èèì¿q²ïié¹1éô8ë¢åã†ÌçêÞWÖÞ“q¿Op’e)‡7$ùõ]øßU¾xú}i„õš6ÝGcÀH7UÆt 1Ú :ÿ´yT¡3«Î×Â…öÈÜEäøÆà>Â\|[D÷Ô¢ |ç@²Ê@f¶ï§}$Ÿ‘âYr‚ë–òdçñhRšFµ¸ W01#—üô!/!›ÊÌΛ›oS¤$Sßç Œo¥¸ÃµÈÞ\ŒiÇQå'¬ë¾…dV¿Ίï¹l„ž¤§îDÅ<’º«0^ðãÁUóÑÊ®¢\«ÛÛÔ†áƒH?¥ÆcÓþÂ\/gÐ:-?ü²Þ‘¯Îjé”ôÕéãc=Ëc0Kob‘¢üÖþYÕƒîL#šø÷=CÜ}9¹ÔŸCÚuó—±4á‰ä_G"ï³Úw•@(ç\‡cד½8&î’;QÕ@Çî™ì´2 Ù6cÖõ§é¸PaÏÞF¸O«5ãxÈTš¿¹醡†¨y ø}ŸÞ¤.Õˆuµ{žðŸvµk/ò¤Þ‰ð†´•ÏÍZ?îãð(°*dJ˜ki“Ç™LÛÞ²³ßl1 Xàóh=y꛺Í@Àür 1ús6·ÕO«ho³øâÇ´¨oQ–eé‹ÁzßÕŒÓ<îl£ m;m2K½ÇJãxߨV=ï^U Ë)!i‰¦®nÅ ÜìÕý|ìË#®Hó¢íÁ4’I©&Ì6Åü¶]Ý@Ò Y§|•ÅAH}J™pʉ’än¬ÈG7nhlºÀoâyêXé›);ÏeÀå]^ò‹É~Á5Â/n‹oû§¶úÖÝõ anTQ< ®>iAWBF~îÔ ØÄÚ(ª±Ãš.ìJPjƒ1Þ‡–«œk*w›¨' Òƒ½p²'!#µoY}›ÀËðþEh)/hDDKu=ç–TSÞB6Î4Ï?óö’ñÏÄJŽÒø‡N!ùOÂI<¿×ös²—07<¹_âþF´—ûµË+÷]ž~‚(±…ÕÌ.2ÒHW›1†“§ûý6#§ƒÈÊ¢b9Ÿ©Ç–•%PÂCoŸÉð%.ŠJôœŸÁ×£Ó y¤„Ý0±:¡RTEa)ÍÁÙÛ1’²;Ö£ýæI F[m`CÅ"6r.µiEl7ÏZàPL[ô>»ÿ™CÐaçT#4@:÷ÊÌ –Èôå*ô}a9x˜î(|U ‹W¢„¬Øå/›P“Q`ÛÖè Ûä)böuÅ'+©†h_ÆæØrlª NS/«ˆ‚«ÝòÍ—‰9-4 ðE #Õ¶3_-û'ú®«7èuÿŒÛºÇ2O¬õµ‰T¦<5S©P°†bøIôËĉZbUP"»ý¼D2ê#vÆDÞ.âøB5ª-oѨЀ¬É‡qi¹Áñ—Ö"N…Žx„|Èôzhm°6É^ð®+X½Ù ‰b¢—R^¿Í&@ðxz( ÐadœÒœ¯o'Ù Ú®¸ÁJM££…i#jUgÚ^ÄÐÎÜÑ4O8(öƳX$) vhÝké÷,\bžh5¥`tG W²5Ë4ýø&Ÿmè›ÁV‡áé–$#@ï%öÁ$WkÒ#¤JƼH­jpeL'U‰E=O,„(ˆ«»í -}¶Ì®kèË´âÁ¡›IÌÏÝOC ­Ê¡ÖtÛ«Í[R/¬¬Å¡‚²2d¬EBuK6 ÍW)5œ{ Nʌˬj %ù׎KV¬xUI…¬àøÚ¸ƒ0:sYAÌΰëÉaòý¥‰aæ²TOKz @ºÅÓ÷;©~IaË¡S|ŽxÂÌ«gYVMB8¼Âòõö2k  È„² Uú-Ô9¤,µ.ˆiÃ#D—a¾u>S½{pÃOÒ¤‰ û^\Æ|[¼ƒ’Õðø!Lú öµ/Q^2Kën¥±ú'íªxdV¾¿vSdOÚ‡xíûK󡌘o¯ÖÛ×yHGðårííÖ·f‘öÌϾ Ø€øbJ—-£ãåÊž‰F°´É§:M­ø©Ë:½KÓeÔ2fûq1¤H#»ìr³+>ðáÏʘ2¹ äL&«ëóD´ïSLt¯ºõ9™u6K¾dq÷z,°lIÆ}ˑӿØé]û3õDßë£l¹ Ikõì’Þ;VÌ]Õ¡´…¥.„K«äô®.õ‚t1:<zT½d4æ^¯°n,¾È³Ì¯¶$ø`Ä?Ö›+ÔËs…r2 ¡*T:†§óÙí-ʃÎ+>:`—‹ž.‡‚»n9ÌûÆ%/öó ñ„S:”ŒJÐïO†šÆbav}8¼9¬ëWâÕQ{i¢‘¯Ø£q$D%Wѯìl°|ÖÎN—>ØÊºWé²VühØû§xêÝ b6Nö§©„ÌšˆW ò =ÜŒæ/¶¸Eéb?›¿á®ëe§½À.ˆ4¶XnŒk™”-^Å¿§È¸NžõYCÑ>©Ðx ÛËq¤S,“Ô¬Ûæ' Ñh&…­VbiºÉ|{RÔM›¦½šheO0R»€êºÂøáDmz1iŠ+Fòg6ÅÌ_Œßy =d˜ÑÝ]·“ƒˆðœØÊݪaøèиeiÎ_l“®Ù•‘hqÀõ$äÞ&GSù5ùï×®€ >9ÓЀåÛŽcÛTäp¿˜0v{`£|{§ *÷_y¿ÅÙ$.­‚6µè»I³S«<**[àj”ŸOW6,Ù}à­zÔM¥åÇ,¢¿È P_|Öa|¯Íœñr‡rvß+WÝ?  —&_>¨ô/Æ–WM]1ùø9G6Ä÷ΚÊé¼2„3Ë/–‡kôPgÊÒÒ¸I»Z\KAíMGH\=çüœ|½ö«Ö…lPS÷O%<²Ô3­¨cÈf[EÕ+µU Ò:wÓÞN›ã=€=a[ûÖE@ sÐôZxY ßÀ$ê+«X,í=¶&ÙÀäê>8÷åÀù»Ì†˜Ã֕ךoñÞ¹>Kéó“úy¸Î´ÓVób§b¶í|5ŸC}ñ &p²y-sR^œ?³”‘Sg”G>k³‚}üÝv£Ì6m€Þ!…ÒBÿ y4v¢4jÁ^ƒi=0§Š¥Õ(›P¹ºòzÝöžÏôøw1›wÑÙZœð%ƒ·0i—2:ýÚ4ë—SÏß–QgíšËQ²±v ,—ŸKÙf ¥°õq†FX6Æ;ñ<ñy«Mµ°šùš,…DzXª¥ÔiŠDè›Toé_;ÒûGkÛÙÝ}·0ó#ÿà ãÇî’½ÝLOžg߈]œhbêTPµ¾^´í“‰¿(«U]Þè[ÕLðY´¶<¹{7Úã‚°bŽ<ÏI4×o6¹†óÛ<&µ6ÝfËKãêÛ[Ç Òɇ y×D<âXTxLôU€ïXqâë?¢AID ÅÕfÃyŸV£Š›Áb_ãÛ>‹C.G×^õ ±3U6\S5áwíêÏ*ïÉ­ItbP dJò‹½qÑ2›+Tj’åÌçj GÌp)¯Ý<3ûš:3o„Ûç_E®m¾w„9ýüKG»áfpG±8íqå þÍÀ 2š¥Ñ1ÍýÖ6eß%‘±a{nM£jï0S“nȼóè…¤ I*ûRM³vWt uu|[ê‹P0çhÜÇ!#•Õ¶.¦zŠ"@ÑÉý² y¹Ùú‡‰¥ÀµÅ»1Å –=Û7¶e=õ úàìKaæ¢M;n¤Ú³/°øº¡Ù9KlÉFqÄÌŠØf£„:5#kY­B«i-7‰M#ËÅ £-ò@Öà­í8†ïÉŸ,ë[éN*¹® ôfNÇêãžå._Q$[Ü’^[—àÖxžÛìó¹`Ó`«d×Ê%Aq^–WÕçcµßæç‰Üÿ!aJQ endstream endobj 205 0 obj << /Type /FontDescriptor /FontName /PVBVDE+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/equal/five/four/nine/one/plus/six/three/two/zero) /FontFile 204 0 R >> endobj 206 0 obj << /Length1 1962 /Length2 13473 /Length3 0 /Length 14668 /Filter /FlateDecode >> stream xÚõPØÖ ã!¸»†ÁÝ îCãî îÁÝÝÝÝÝ=@ ¸?fæÞ™¹ßÿW½W]Õ}Ö¶³×–Ó¤ Ê BÆ C ÈÆ‘…‘™ ,«Ä `ffcdffE  P1w´þ%E PÚ;˜ƒlxÿ¥¶8¾ËD ßÍdA6)'+ €…“—…‹—™ÀÊÌÌó_C=/@ÄÀÙÜ ËÙ(„A¶nöæ¦fŽï·ü÷ 6¢°ððpÑÿé²Ú›Ød Í€Öï7X”AFæ@G·ÿ AýÉÌÑÑ–—‰ÉÅÅ…ÑÀÚdoú™†àbîhP:íÆ€?èä ¬cD ¨˜™;ü%V™8ºØï+s# Ã»ƒ“1Ðð~7@YR o ´ùËXæ/zÀJ`adù;ܼÿdn󧳑ÈÚÖÀÆÍÜÆ`bnÈ‹É0:º:Ò lŒÿ04°r½û8˜[¾ü™¸@LH`ðÎï?ìŒìÍmÌ­þ`ÈôG˜÷"‹Ú ƒ¬­6Žä'bn4z¯ºÓŸmµ´¹Øxüu61·16ùƒ‚±“-“ª¹PRä?ï"„d¦@G333 hº™1ý\ÅÍø§òOñ{þ^¶ [€É; —¹ ðýÁÃÁÀp´wzyü[ñ¿…`lnä0ššÛ üý] 4ù ¿wÞÞÜ Åü>x,æ?>ŸtÞgËdcåöùŸÍeS•Ó”’¤û“ðߪ/_@®6N+3€…™‹Àõ~ðúß(óÿ/÷?¥ æÿÉùŸˆ’6& Ï_Þk÷_Îÿ™ êÿ¬ ào½Ï1@ýÏØk3s0½±üþ?]þÿÍüQþ_Æþÿæ#ædeõ§–úõÿÖÀÚÜÊí?ú÷)vr|ßYÐû^Øü_Ó¯À¿–Xhlîdýµ’Žï›!dcjõwÍÄÌ]Æ æŽFf Ð{ðÞÊܨr0ÿã™0°03ÿÝû®Y¾?%ïúS|_¥ÿ½RÔÆdüÇαrp ìí ÜÞÿŽ8,ïËi týs®LŒ6 ÇwÀ;=/€ Èá~rr˜„þý…¸L£÷áaû±˜$þAl&É;€IêoÄýn©ôzשüƒÞoøú7âyGÿ n“á?ˆÀdô7âx׬ÞËô_ ;ûkëüÿ¨“ñ¿ € ø/øž–É?ðÝÝÄÄüü‡ö_ù é¿à»¹Ù¿à{ÙþmüÎÃò_ð=Y«Áw&Öÿ@–÷,mþß³ý ¾çaûÉw[Û÷ÛÆ hâø”å?Ò¿Vëo1ç»ø}¡@ÿªÁ4íÿßóvø|÷pü|§áô/øNÃù_ð†Ë¿êõžšë¿à{NnÂÿ™O#'û÷<ÿ|?Þ‡÷¿øÏ? Ðh„°¼2âûfQó­í¾JˆÐ…a‚–bÿk2 ƒÇ²}»Ó# lMeºÿ¦ý­PÂpÚÚ®(õà É‹ÇÏæ:Øà–8ÅÖ'Ïg½¥éýV„¥)œÉ¼ŸBµýÄðD *‚ž/vžj~–ÍàRÙvNÜ( 9˜÷.}⮵ý%«cA ûŠ•œÒˆÏ%3 ªáÚ~…s? 3æñÈ`ˆáh1.\Qçnng1²&ßH¤bè¼N#Øò=4·X#æÝ×ËTXºðÉñ5ñˆ!o0Ʀ)=¾%Já.zåoÚ%Íã´o[dzXQ»ïË)ÝÛÿÖ¥¤˜ä¥ÆÇc‚Ä’ü¾Ù"k]H Þq±&µé*ù›«ÙÐ^>ÖüMÞŠ«E¯ïÊtº6¹Ý“ï ¡G!É7#©®©ÜËÃµŽ¾KÒY*Œ­O,ót¤Tæ7ÂÿÚ™wII öa†Ä޽Ÿ<ïµkŒ“ÂðSü›L áïÁñå{I*à ÂÖÜEoŸ$BO…NHÛ9#y즊r¡ñ››ê/Uj»½d4*f1“ºÂ訮‰xÔJ»n ô]4t›ßùv­Ázhq?K¤·“QÈE´òqS^­À‰Ã`Fhö&tÅo!뚟’—7]´\G› oã²û6¶Äh¬x½>¼U²í©#Ä@#Ë:ßCQˆË]t,öãÒÏ´3yëO…Þ‘¥BJA£E˳²q¯D›ÃÂt$ÅýöÚM¸öæ| Æ~³.ŽĬäz©P^‹{¯à)]±±ˆq6k1¸Í¾a/{˜ÔqmÊøλPyCñË7_Ëà™Þ+ìTŸ8ÇðT‹¦¹?bS ù—Ü£z¾„È'1ÍžTâw,=œ¨Á—úI#ïÄê67õqÏ·~¢‘-ï Úýa½ QÄ7v˜C‹Zî%oÌ¢£©B%·E‡;¾äìê±p¨uö¹42UÞ†›ð·’éÇD>‘¶ÄQÂÌ)4Öj)º€Ñ+Ïåén8ñô¸Â£ù³ôŸX[–<+5 XW@·ô—"¦!:zC6—fž¯Ñ÷\^‹'«ô³ŸîóÛ¹ÍDMù¯ã³Åªn/*¶]Ž‹OM‘¡—zÏ ¬ìE‡œ3E±Á‡8¤0sòi€cæ‹áäµÚ[BW¶öäÑ.ñEÚ¨£¶V!Ù!6,Ny¶ÆZç£gV–ÝФVJ§{ο}¿³‡ë¹b´é} '€QÓúõ'¡ù’y']ùd¯író£-Öljâ\¯&U Y±’?w,æ+a.c(H8Ȥ=™‚aqµ±³VésÛN˜"$Ī£àõè`ß§æ`aûºg¼­s÷܃½8‘éÀdjÌ[ Wå /œè$˜€¨o`Ç)?úU÷ʱá-3YàT•Š<¸Ž Œ¶o–ÐXÚýtÌLŸìxÜldB ¨”$ËU¦¥ã{茷Óð€V½éé•Cdp,ž)ÆáÁÙÆ©„ß™@÷à™’ä„Z9kôAaùý’+ŸYS—‘h~f?·Wàvæ×õ{»w çì,𱎠ÄU¿S ‹¿•w• zD°³.øfFƒ¯>Üùƒh~*ø;wCÀ ñ÷•ãD]'Ø^äïF¹´û|E£@ÃýÍA0+µ²WÏ» tÁ¼¾iô"‹ŸB&;<õjφ×?já1¸!TžVk,ðØIÕÚZ’÷€57öašÎsšô@µ‚™ (fMˆõ³¯ýây0Îo«Vu˜ÞÇçË!UZö2Á^vp*ÎÖ û¥†²}¼¦™ÛgݽÁ·ð}jò?±Ì'ãVJP?”ôœv³¾c`g ½LG4 »uýúˆÑ^Ú^aëФNÁiÇõM•êóåGÔÊ&ËàýU‹ƒëp½¤rnTZ_ú=†Ú™î™×>[ºèˆ³LJ 5TŽŠî¹ô-CŽ¥@øåð¯ÔÙXÆm>cšùm[¸/f¸ptìeGQNªðˆ´5¥GÛ¼Öqù|Ÿˆ)ü«Iòw†¹ÕÓÆél)Ùo-R°÷ðnåuÛXªóUl#Du“¢±úú`IÂSW‘¯]É“ëÏèL¬®÷\ccðpuWòïà³(s~&¼°Ç~”% v°om2ë¼!'厫§ lŒ®‘`žtóâû@´è9¼q8‰”é>R ìn¿Xípî`»o ¡‘ªÚ¨1qáÚfã–EL(\{ލ­ÉàÀÁ¿ RÚÛiØá;ëL”J›.÷áåÙreÇå”óVË-+ aƒ·ÍÏk$\úýR´¤öÄžêÑ»óQÎe½ã6ˆS{¢ ­8“©²F’–Z›q( D]ý(¡9K0¥”úA~NRˆ…î;7J¿5¸j¡¬ƒM9Âr*f75F;tB»:¬žJe^bü[ÙD.3ÉÃN“¾’ RppËêÞÆŠ¸Éã…LºOä:ü­?~ÝqUE# —'õMÝ‹Ý ¨ˆdÕ­Ç%®I¤ÚWÜÁ”/dÔB;… V,Ô­ÍNÌ'7ëQCÂþvöãŠã¦l¯K¦²{™ŸTƒ±#ÃË\zû‰'ßÎO9’–Jé‘?Å—·cØQ \ÏËO^ïßwí%U,†ÿr'’˜V3«5âÛFwvš8(x‡ ¹(B/D~ ˜¼Jž{89C}oMo?þjOo*Jœ¤/Ï/Iv¥Ødh#…>½5Ni­yæ)ÚòÖŽ X’ìg««ñõƒèÜGÍù,˜S=G©ŒRl»Á”nX©ëÌñ–NC_Üj>Ò•T³HœÊü4‘‰Ä¤^Üì ~5$©Ò¡QŒ@_´0±µ5óCÛw,Oá›ÐŽ^ ¿åæ=¡÷©©á”B `ÕúZ!Òâ!ÒϵÖJÌÓ}KH%廑y7µÿô‘åz’Å"‰wÚ†îZ<‡~±’>{ÓÇÊá•­mÝ_LòMEb+J‚(P?ì6Õ9½#ágñ>?Å,£Á¸tPTh¹›ì—Òve'K7g2ŽêUw„…Â5B*v‚á»Í °oŠžÈ” E-C]³-JxFl–”³mËRÙÙHìÖÈWZK]ZqCfž:Äé˜{›cuÖf7m%öÊ5 l!Èf !!SUè50¤ªPBY k;Ûnß—p4T¦è-àK"£¿1: ÉâöD±WL; >å;¥æèklÎá µŽ½mû^ÙäöÙm—uEUˆýÊ㚤/³®Ø-¯tÌÜIh\õöú·˜1”œi œøDËÅ-d,Øø¢_Q|ýéêäÆ^ÿj‹ÛN-îÆ>L)VȘŸK»°ôÙAIYu昋X·¢I6 ÔVë÷í„_'‚=mi‡$+ûÔý!D ºá©D¢ÄPSµ/MÕÔH&Rî¡c»ºÓê¨jkà .‘•u¬ãQAßÚprÎ50P‚4›~\¿hÌN›÷«òÌ2ízþ¸žU›lWÑÃÀY§$ ×K+à ägIyrÇœ_ßvZ‘úzù_E—"n¿d6ƒ‡?Z,¥‡àRÑ•ÌöP#?˜añE˜ÁÈæƒ–ÊŒ*£P'oû]©I.Ì»ðÂ\›°äÆšœv–o?¨œßÀEó·pêÔR’_óÈk`Â/´ˆbM¢÷&ÎNµ,Á|îØ-¥Z/™*”åK&m4¤´É¬‡öÏÝ=X~ÔBFÄ׿B§ïÓBB‡H=KŸ/ÇzBQwäØt¸c.4†Ø‡)!,ô:ÚÅËÐ7+5"ËŸ µ.𥏠ð.¸hÉ(IÁ|'&2Ê0î#(É/=ÄÎ ¢ô­ÚѲzKhÂù(=ƒdQðZ`ï»z3µo3Q”•áDsQd@–GŒ/!}Ór™„”‚úàh3Ê©á<7ÀÜÑ›ºÄ˪ ¡éÉï8I Féi>ëæí™ˆƒ6H×aààíçdÄ´7¯EîXÔrÔ™½O%)úç©Ôè0­%3‰Ÿ…Ã’Xh…‡S‹6%12¢á° Ÿìƒ- Œ’¾ÚuÜ9ú hIš“|²3ÞžÍÕp9¿EK­«[Ãf ,Ù Ð÷*˜±¡³ìk†?•NCž¢{vhq £~-Ob¡ ’;󮢨§•F1z (Å"ÓÎPˆ”ãÓOÔv¨žìa,+ tXÀÄ‹ñŽÕ®G›@ÁkYþòÐø:ú¼Ø²LÈi@9ùçØÄñÕ™L Gû—'V,¤»·œ_Ç*ÿŸPYô;“kØ»ÏOÛ¯ÁáœÄ›¥>B+©’Rß©¢<„¬jÈÍø~Y,WlúÀAblÌLàÄFWB² läWr£9‹2ý}Γ¥óKŸßjÃv6²Jȇª–i¯Žïë~m'W´×ÑY ¤°òºÃÇÈ|]œ›s}’£1á)Ö ®‚W³«#Q3Z¬`m¡ðárDQˆ‰Ét¹æ>­i9tÃá÷¥+6èN^[YááçVmXññ"õüÃz™p^¡ËÍk´,al;<ÅÃö3•qŸ”Ýêµ U‡µ OÍ:4»8Ý#•³g$ýŽÌ¿o¶ç1_»VxF«ÜöáÏ~œZ _V!ªÐ8Iì ÿ.hâ´¾ÐÿÍá8 ŒzÄÔŽ"$m謺k{½nƒW xÃÆ—·* Ùï3Ì“”ªÄœZ§Y²ZX2 qXýBãò`Zø†8E”k“¾¦ði¡·Û*§üú;™òÑ­.k„´K°ó܃ŸäÆÌ2÷^ x‘×bß*|Àò©ÁŒLâ$%†@LÀ(ƒ#’ìÞK:#Lyr ÛÊÄùl'eû yt:ÔQg¹Vmyû?Çç=í‘üê !ð/9oD¡›¬ìÃd¿Ä>æmH§—˼‰Ò~9¶.&ZsAÌY7¢®2£ÏúôѤՕâ•îæ {v‚ÛÓq§\ÄÇȤ+e‡¦È@º)Øk‹Šö¡í…§Å£^{»ý,©ª ¿Îèÿ*Ôn¿=”‘;ü}Ø;‰Ó¤ ?ý5teÏ„*lR.­ÛMJ_Æ/ÎÒ<ìÞã•T¢ñ¸V{·›ë_x8M{Âl”Q)½ÎÆÄdÒ8ìY¼Ô2Zø&%+7‰äiEÙ)ôfÇ¡üÝB1~t,íà#^~7|ãÚÊÉVÒ¢‚e#¸0ýyŠds°{Wgj"WÅ¡eß#[Ò°ƒfšÝÎHCø˜q -ô‹IN¨xjÖvžáI ã™ù¢¢™´Àû¾_}þhÚ1ŽÔÉÿÜÃïÎ@SúVÚ`L|_`úÕÊÓÆNf¾þrW,JTb&MÛÛŠŠð eæ[ì(Vóö寉ýh§§]NÕb`Aÿ³Ï¹xÅ—›Æ¨iٱЌgyâ§ðTàõó:t¦ï¬_ÌÏ^53‡Û7õçì¾&ÍWAªºsº.k·dŽ=Á¼¶™3ØýÇpv­ªúnã¬[p?^½ñ)Çëäù³_ïù!'ª/¢Vn@?»;5èJ¤VÓü¢ÂÄzûÝû˜Ý•,}}tÜé3­zyØz(ëu#‘J±"ÞÜžú(›Ê´~×9Óå:«Ÿ`¶ütf}„”¬ÈF9…ñκ&0 H‰uà¿'£¶ÝWl+¼Í&SRì:.T¶cq + ?p¡ßƒõ¬ŸÏì—,#)Émöç&‚ mÊ—;õŸ°¾O¶K‹È™f“Tîùp2ø²l¥äŒ»©eçÙ¬›}½ßÂØU¯W}S®T à)ΤJ¥ ÿx¡:¹-¨~ÝÞmõ”£Ùp,LQ/C»Îìó 6ÜŽ¸9°©%ΘØŠk\%ñÄߦ’7|Ñ”ºÂ´%/i|Ùá aýŽîrÀÉŶ?¾ îãä÷±ø…´d|TÊ¿«)©Ì€ÝÆéAG®u”04mœVUL1|üô)Zûô™–»¿±x%&”÷«ƒWµ ülºvÈÕñ¢gTÎÙ€[ ªï&Ê*/½«Éð]—·ºóµ<¢`A:XòK­«Œ‘Iߟ< T\EÉ·©lBÈÜ–'c¬ÔlZ,•*”« >-9”°—êo{ØÃµÐË}êî} _‘ú.T)b Þ”7µ)ÉX*‡ésº[Ƚ@~QÈ~êK-àî+ïN¨ÔŽ!?ã6B¦OÆT Éé€hͶ}@«%Ç­MK ØG7³ÒÔB¶` :ìÏs¸ºœvÄÇ7Fëeð‚‘$h}Ø<¸áï+ø.÷›Ãäú|¸êèC0O³>¨ôé5áפ ¤)ÒŽ¸¯?É' ùéS@ÙE´›s¦~«¤ÿÇÆ‚=üÒgÕÐ8å»Èž ª(ÊúyÔôï탵óÓ"AiûYˆ»Ó}qn‰¤F9u_ûC'øjvšZŠý”fÇ<GH¬ Å‘U´š–ÉMçb–®3:Ôò?÷¡§íq@¼2(Ëtdj>]$!éQ€ÿ2ð=ÒͶ\¥éûÇhι¢UÞÞ&Z×#cqÁ˜Øªºæg/^yʽ“=¾÷>KYŒþiYk0ç-¨û¨ßSXfWEì·hUTÍ™ô ÏçohüdñĆkPpÄí„+@yj×1<³FÌV^‹£“Tt#ë‡ø')ÆÚ||lß×®`•³Œ™ºYù§Ãýò™‹¸$¡6¬¨=òl¾Ä”6•Hbú§¡HÀEÁTçyX‡rƒ[¹b÷;pp«³æ×/#œ®È¿ð&Ùõ±‰å‰¸Kú¸‘B}¢Fâ}& ™SNõ6?Ð’ÚÏG6hXЕČÅ(N^÷;·Ö46  t‚!ÑìÒYÝËŽ"LjmŸlÑ”ÎùT‚ou'gIYŒï†ý "àhá\ÙŒ>‰Ž>¨Òd¤ÄÒ÷ÒÁ“¹ÑÅ®k6b‘‹Ü2í#ÜGi 1Õ8|Ý¥Ïï&´jþ*cÝ^ú½.*•»úÈOÊøiò:X6Òo–ÀãØûy¤þeò¢í´´hv$‚M‚G¾t£;fnSüiÝnn*ÝÆHa x½ïAÙ§­K¡Ø}òëd)…eΔ£ëCá9Áf¿×q0õDs5öµøº!—K¾P5yŠ·þÝïœÇòOk´š´í«×°Xɹ†CjÓ/È{ôKWjKßì9Pɼ'~zÀ˜ÚÖÏÁóB^IµâáŒêÙ÷ùq]9¤ea§ þú–Íûâ»S‰ `9aR‘Ï5®1І"äÅÄ\ò1wuU'£ÅOƒ© ÿ[Wãab)áïǣ˩¦ûÐ|Æ„SÈ2!«µ¬¬Ê"¢[¥ôn½=E}Vfà$OY-á ö{ !}J¾’ë=îÔb-°%î5 VŸñ/I Ó|–wÜ-Þ¹íb™æî5§«:t>ŠËvçqy¼ê†3³ìƒ¸|c³e_4FUÀ9>‚ÅËú2†°«õq& .çå öঘ¹9 6U6 Þí—ÇÆÔ Æ>i[¼Æ0’6ë­MðïÔ‘oº¦‚/?Ü̼ÜTÆø]1=µ£eAÞÚ‘óÍöùq]èe[ÓN¼n´$L¯¤úÊãn¹õìC“·ï®Mó>,æ¡æÞ.>9Ô·0àÇ’Ñpa×ôaoÄ®¹Ü•Gru÷†’G†ÅèÊ)‹g„¸}ë,ÏRXrâÙjyÚÄK¨ej¾“¨j”f,©ÆrîÓé™S’èÑ*ý ê¼¶@®Lý6¾ž•ã"éª@ÑTá¹u5ÁŠ[î³ðý '˜ç9¨„QÍlJ‚b˜2Ö ž]pß1™àö@ÇÑÇ{azÍì) ‘‘©aÈÑâz£Ïdo”RÌSs«æãÆÿ!–‘oi–êPâ-¶j†½§$ñ+ºã´3oE|×i½¶±k§á8Z5ý׋LÓoPE®EÅqXßó ƒ@¸KÃÁ·£âWYžÀ½«fuô”¼÷½wVD^ÈÅYÁ“}^´ÜR{ƒJÓ©ø¥ûÕ‰§ol#²Ae8™n0M¡5‡ej6/4RÎãc½ãp(tü“½ß>%»wY´4kïcëÁy=ÚN}ί&h¿l2r›¼5dRí£Ï,#‰Ÿæ±>±K‹!‰VaýŸ±ÙêóYAAFSˆüÔB†e’‚u&ŸÓ¹ƒZU^ï)šgªæœ —mù‘‰aö M¢nцp¹[ÿƒn‹Ã¼¢Š>ŇÄê×5o” +ìÆLU&ÚªR¡»üÒ(àyT˜EG|øŽN@ÒŒ¬&ÉÖ©?&"“ÂK’‰¥NèÕpöÄ1„zAÍ£”•LàÈŠ¸u3dÚÎ+b“Xüîr¦óËʸð¸cñÂ3šÛ¦é›Tú#ÞXÞÁýÈ„ 7Xµ­2G'ï{°‡1Ò9Ïfäk+=¼T‚X¾tz\n¤uò›þ›"‘Gý•,D”êu`›µ¥* óve"uwbHÐ:Ûípz©bH,HTy^•ïŽI«¢v««Ÿ–ø6ý‘)F¸Ûyí·dÝ S2y×5’êã/“Z¡üÄ/nK\oÇÀu0HÇSb䂵W›ÍSü3â ó³—ÆÜŠbÖוIIº« ¢aô‚WßOá¹bMP˜¶£«ÔxTö…îPãÀ¤Ý0 ”–BqØ„™ÖÑóð/òéÚ"I*“?¬m\9wÖx!+©ÞÒÍ4šÃ1|É[Óø ËÞ‚ÎpËÖ(1k›ÚЂKå,•ËËM™Y׿ôQ3ͧR+¢vÎØòð%Óq »q{„1’#~{¨öÔõZwœé7p”ôvÒ¢j¿ÖÏQK 1ôº¥¤‘øW~œTœ`»R¢§WíûíSƒ)ï‡h¥S;a=Þ×V(Ö“)»åoªéK'6Ú—÷Ä¢Ë;.ú²–ƽ¬¾Q%@ºõp¨èÕ-Ñb.ÝÞˆðbt3œ~^â^UAéýø%õ LNÚÖÛyÔÐ ã%-Z›ÅüàgWàˆÒ”€¤Z/XNÅì÷ÉöTyk!Ã/NøÙŠÁŠî´tÀ[·Ë 5Bnǧ;ƒX…ÛLæåÝaÇ© S‰iª¸ÕHþžfÅj”S#Ë&¹±SÏ\Ž:,Ì“ÊX|F€»\ú­i=Ž.K«AaëÐ~azÏ2;ÍœG/Ú̽Lqí,©ÝSNEM d¥àÖåÍlÍì§š>ä0©çˆ]6˜ŸÇAÐ2Ô*v_¯ÌÖE3ì.'á >ú¼TÃhñJ Ò4ûç)s—)ž¯¯®Ópùûkãa‚LwBÊ2¡’²Ôˆ³fIµU­p`6Ô 'ü'ýÊĺí Iá‚?iû®\~’ܹ¶Z k#ŠÆn7{Ð ?Î’;Š¿m› ÚåÂß#“B;jä¤gv欶16Ò¹ƒ‘s ñº€L’Ž,-˜³G¢^§"Ô¨9É@AüPHÀFÍ7¬ªt°‚:ØO†›ÔÓ‹GoZÆ$ô•ïâ´ ­ñ\( NÊ5¿í¥1·«·¸ûÊ,µoz³©\¤ò—…i›¶Ek2·X: •sF EMýÕ’l RŸ‚Dˆ¹} çœ$! ¥6/œâÂT…Ççyê²öªH½”=/\=øò3ê=桦²æÝv´ xq¬)9æ)È*áFEH•©qÕÌfó cÀR´ä-¿¢·¹ØÐáÇû†FÈ#N<}(åêtoR(¯à¶LŠ>r”0oêQÂèàYÆË¼Í;à*»ô/ë¸+}©S?åò®«Í)N“&ë;oŠ(÷ª£&&<[A*NSùjS/à¸/äµy$õÚðGéŠÕ~ná(tLæX ÒçÖ—£ªK—1†ßÐë¸QZø®Ý¤èRŸ‹óY¶NBU×™eNß½ÃÜÑT5&²Ûdؾ7÷“‹EÍñ‘¯cú×Çt횯qã<¥­ÌqR5qPßýç©ù3Wìæ€âaŽj~òcR>_/ *Ý}CO&ÕݘةxbÂ’šMU9æ-5²ên3x´,8{ç’gï\4º¸sÈçÜkÊGðˆæÏòž€Â5©Â(Ø5z1¦æC9rØ»Þô(½ÞNå_÷"¹ƒ™Z:îé) C”v¡ëw‚n² ¶FV‰¦ó¡£¬1n°2ãM*¤^§(÷Ô ß>v¨SÀ¼"ò~Ž?×9çmøxÁ¿ÆmN޼øCmÄT0,1¼U‚ÞÀD}BE½‘†kV‹öëQÀ_Èt©i§Ü  ºk£Ù,êƒ.6ç'ۈߟ}ß OˆÛ\è.µ£Xë¶H6Ö£ðiw!ÀÌû nïblM7‹ÇÛXãªxý JRF¥QË–NykL°¤~Wnɧ³ŸüÝn…FºNKÅlŒ¶a ºÆ¨>svÓ_uL?,¯œÝÖ½GÉø@Z·ß^±(ÎÆ·þ"DdE~9äyä(wWõTù(5náúh×n© áyœ¹P‡úzY4½êÉ#(} i"SƆp%¦‚ä4º‰2¡<…ÎN@NÕÎî]”Ž97c"Ãeèƒñ£p†>R’÷¦Ù“Ð[Ê}½nþœÓº×úèRHI.Úf°ÔdYî‘„lÉnfíäÆ÷¶d›'¡”Cõ"‚¼ñå.$o½óWÞŽ0¶!-âHZ¬nŸô·ÏSÐøÑL_?érmòFN ÿލØRŽ…ÀÇì+rƒß;u -î¡—#UÈÕ¨‡§rmxJ*†\¾m3å³I[)2Ýt¤a»Þ@…nFZæ¢r«ÛùÈÉ÷$ü„õ çi5ù¤°dŒfˆ$žæƒ‘Ô¸ß*r>Ž6Od|dpDÅoµ¼u ŸéÈØ›3)ýq/>ŽähÕ}’N¸\¬Œ+s›ÄƒÎL™ôäµFLØÛÆRü#ˆ_M·ÒxÌ@‘4®nÃiù>Za sn;»{Jªž¶ókÕfîϦð Á²0ÉI¤äᩈô4.kòøÏUÒøœŽbz¥¼f‹©±C{ô7i‰üPZ*L |=s}ìØ™¿y€†ü¬Ñ­ÿ8´r”•éº\/Ÿ¡⺗R•Lí„¶Cµ?CõäxÖ?¸rxÛ‚ Y±×ªÿr>xùy‰1åµe|7¦½:_¦Ä¢Ôy*'5¾ô1ÚXåúšðåæã¾sÅÞÍ^X¯Jj„²iœˆÙ‡ôP‚ªH¢øQÜ5B•ü‘‚ÇÒ=ĄȠŒ8kÞßm֛Ųq­Œ‰sØpªü‰¾#* ‰Ãùm‚P¦â¿C²GÜžµ—à¥Z”3cÕ³“ïÅÔ‰ìÀÑM°ˆ6ÓXmóË-¡ûzÑÚ ’]ž], È?b|š—FM+©æ,-–TPÑßlЭ}Í3΋î¤fS|Yvƒ#%<Ò¨Ó@Ÿ(ÒSè“å+²Fn!ySÀ4œ(xkO ô°Þ9gÉÀx5s6ª+CîñžÙ°ÿ ¡ô‰ô9vŠ`þ`•µˆ^ ²ê¶VV9æ7eë¥vÿRc[ŸÁK—£µ·Ø}%3Ï5‡IbFšáKÕ&[lg+݆‰"ö¦ õÖìí{0úak2í¾šoÿªÄqØÛèp'jÈdM"ƒÒéSÒõ:¬¨yퟗ˜¶å/™EÛ3‚A ØÉæ}{XÀœ<~ÝÔ=Bvž·*ÁDP“þúÙäѧMVã1T@„Hœ̌al‹Bz)šãüJ¶L‘Ú¢ipÄä¢ÛËëzEŽôI¦¾«ê‘I„” {˜Ÿyž;j_?3°²ƒP Õ9Ï¢#Ó:Ä(õ[]'—FeÍ ܆v®E]ŽtQØkö±iÜ«–æ¹w2ºTõé0²×zå¶ì‰ö –]¦ “לS¿mãMŒÂÊêfâu.(ägcb—W{æT =ÖS?¡¢V~pÑîìˆIçb‹h1œ*èÛœg/83•§D©¨žUQ±Š 0f}ð'¬U²ïuÚy-îKD÷WOÓü™óÏF¸7ra†GN´IR)ãü;í»4wc,Âg¬WdDd!a7#r‘;Œ¹/WX"jw_¹÷éšó£lVœ,â¡ÜÂFíÙóÒtÛófLû•]Ú}‚¼•ÏÙv>Nq©˜¾ü²„ÎMP™CüŒ_øM‚¡†U[’~&)-d˜Ö¶Ï$1Y*8‘‡gøÅ'ÌüƒTã@½|~™&=3רvç­éÌÄi]‘Št¥)‚h8¿§K?îØÇ• ¸‰½YkñÏsÍ`BÁÃÔfГž±¨!©h2 -þr•ןÅÈÄ„Lö„$P¸Ñz¶h1²ý+Ä“VhûH7ôŸ°:ó ½}å~FŽë(¼$›ù**]#hÌìc÷X¥_€²nbøf\‘üµ^Ž=EÁ{0’ë ……WÇÕ)7x+U¤s]›%–Ävþ)`NÂgÀóAv¥§êjdqÛà£neÕ\Øtœœkµµž6ØãOY–~ÞX/ÔŸ_üŸÃj«ÏK‰•Ê gŒ°îøTÀ;½Â‚ÓxÍ4v=ørâO²”J_lgåSŸ½ìäUÚÔ27…ƒœœù¡ª²Ü2®àå_­Ê×>Ïþ^:á t+_b\s•£ŽY=+™äV· ¢ž1„ƒMØ?˜EËMÕ‰1º¿ÏÌHB¢5Ijî÷q·ÜÓšïÏ\ÏR œŠ.Z+æjùE"˜9+ij.:xpé°ôá}êãë*ŸEƒ×)2¤kAN­¨Bß‘ˆìNsBOÜ µF_¸ºôXpµïq^ ÆÈ ¢Êµ-QxYä!Läû˜._ÑAꯤ¢òe¢T²/‡£4/œx‹ã`¾§Èyð`Éz…¨² |íÐEp§7ÕúI‚K‘Ü8•ÎIuöQw[ÇHÝ"× ‘رhp‰Ø•4AÈJø5¢·Kˆ-ñÅ–ür¼º¶‹|R/R«*¢º_\õàM0‰>íy~ù}2ÒYtlí,=Ž7†ªŽ@6.yZçUÜù WÜgï­yÛ8ñ1ÂPpEpÓ7x„Hk¤m ŒÜ†òg+X)Åkë·² @¼¨ªŠ¡bÇ×–š©¸ð"éçÑcêK,ou±’¿I.{¾úaÑÔðÈRÎ=DÈ¡”Ö§‘írštaqûëö)Qpœ„°Îóˆ.Êî4p[X©-èR½ûÜNs2öÐO3vL FÙl³{©ÿ©ÀžmäðTãh›qó€EgÈ{?Û Nÿgø²ª|£.HÏ/¾ [²ÎDŒ>ôE¦JˢƂém¨Ä­/9Ÿºëo¹z+ø,´íê«™Œàð¢NMÁ¼Å6†€Ô²f-–âñ IÄø’0g³“¶+jÆx9û:@$²Ù"‚ñr¹×”„ë†C.{P*š]Pj`Ï-¸"lu#¯MViRЀ¶ªÆJTCß®¾Q?ö4oÊŸþ tÿ… üléþuøÜÜ’-Ý]v}¡™9 ÜëCóÛèåo!‚KIp½ åahQƒâSŠíJ> endobj 208 0 obj << /Length1 1474 /Length2 6427 /Length3 0 /Length 7427 /Filter /FlateDecode >> stream xÚxT“ÛÒ6ÒAªH5 ½.½÷Þk!A:H¯‚ôŽ HG¤w¥JoÒAš ]@@¤|Ñã¹÷žûÿk}ßÊZÉ»gž™ÙÏžgv²ÂʤkÀ+g°ƒ(#à(^> @AËÀL…ø€@ABVVC( ùÛNÈj qGBp‰ÿ@(¸C@(´M„Bµp€º ‘•‚@ øß@„»@ä µhñÔp’UáæãutB¡ëüýàsÄÅEy~‡ä\!îP0Сœ ®èŠ` `€C!(Ÿ¤àxä„B¹Iðó{yyñ\‘|wGiN€åЇ !îž{À/Êm+ä5>BV€¡ù—Ãá€ò¹Ch †À‘è¸=Ä€®0PÓè¸Aà5ÿðþ@€Oà_éþDÿJ…ÿÁW7Ü w8@a€Ž²&ÊÅÁíA0$òAa ;4à÷ÖAe9=Íð?$Øê†Bò!¡°_ù¥A³Ü^áê £„¿ö§u‡€ÑçîÃÿ§¹.p„Üïï•nïð‹†½‡¿úØ¢¦øƒ6þÛæA„â""ÂÈcÄìÄÿ«€¡ä·ó·Í!ÀÏ áp@Ó€@ èB?$È@¹{@üþÓñÏ¡€À Fì ŽP8ῳ£Í‡¿Öèþ»C½@´üÀ_¯=Y¡f€Ã|þ ÿÝb~s%SUEyî?”ÿå”—Gxüxżâ"@€€€€@TTðÏ<ÿ:¿Ùÿ¶ê‚ v÷Õà€ø_$Ч÷7Ï?Êàø36œ€VÐF õ pü[þ–@a ý&ð‚ß!ÿ?íÿÊò¿Êÿ¿w¤ìƒýösüøü W(Ìç­gz6´è ÿ7Ôò×@kAì¡®ÿíUCÐ3"wDëœWà!ðá_v(Rê ±×…¢ÀNiéïf kÀ pˆ. ýuï £€Àÿò¡G삾[è–ývAГõϺJp0Âþ× ‹@îî B´Ð+a€ŸzVí!Þ¿%àçƒ#PèšcÀáNø«±‚~Ç_ÄòØÝ´÷·àw…Â=ÿ2?ðÃ!Ž¿ïH$ „tú·GÀï‡þÇŸÿØ"ØÃÝ=Å¿µ„ÞÿßëßWâ ÎÏ À’aÎoÂÚΫåè¼x7Fp—W:¢Ìz£„Ql³/üœ4ñ²T&ËÛÚWÞJ×+‹˜˜ñ§ãÚ?õ¶|—¼”¦š‡ÂPê]“?äÍøqüÎÂh uM^#?HÔc µµÀ¶`O6™¹ íÚ|ý¥ðª¹i.¾pŸ’1¥Ñ¡âär;v0Ÿš1±¬ Í,ÉžÝq·Èð0T3»Â| }v&ßKóÃgÍâYæà§OïsúzÓ…Ž 4º«D*#(hîúª3å%õÓP&æ¨7"1Øä…OÊR˜iŠ\#Ø–°lE—~Nôµ&ö yÔ;j+ßÖy2ÜàQ‹ž‘ÊíZ:¢·YÏ6üÁO|¯))Ú‰§:EžµÊãt0ã J–×MÅÀ­íX¸ÑšB””ž™ib~lˆÅ{*dýeüÞ¹m\V¢€ UªL¬ÿóKy]£ÛƒÉÇËG© ZV ]5,m&ç¡<|ß_öB+Lï'ÃO¶šÞ· vE"Xö('ƒLÛ³vz¶Õ„O?s)‰ j4f6œñ¹ñSúy¹Øº­*’æ²âsãÌñÆB¤ù†ä³†š›ÚfárASd$.ûe ¯?¾~!Ò$•*|²0~xKQo7ž¬fí¶^×ë;÷C~A|k¼s'ÛmvuÔÃhÃNÂå–¼Ùdð¡èÌRžüµQ  EéË<‡¸ŸÏ¾ožÔä™ÊÕF™Ã¢S:'×l^3´›÷=£Ïx;¾ÛÝΞºé­ZqŠ}Ha7¨"ÑÏV«hˆnQøÔGmdÑÒl;iÂwïvh8ùA$`q&ùp=ØÍz«^ûÀígùcyEÀRö¹pq¡µJ­ 6q?l¢(ã@EâïbwˆAãõ!‚3p> '‡Âu˜ŒÆGª‚ –§M1äµ*n‡§ÝôŸ(ZjèùÕû}ÁgXE!Êb?=u…?ó›êˆ¼÷‘CtÖì?ÙÙ´2yL§¢ÕŒeöp¹ä8¢‰¼À,~z¸ÀTWéPóJpÍF°ùvesB5ÕP€Ú{’æîÀ6{g›+‚øüôÂDû>ƒ~–RÊQñ^fsüæÉO¶”µ Ø}å*Ë@MrL}'׌ʹvߊ+Ǹ#aÄ çV€â¢….ü0zO‡iD(·Ãؽ­½ûµg0z0×Ró~¥u þ«/Vå®ëd %aå:“ϘÔnǾ™zósóÞw}òkÔ”‰dfçð·Œ~YŽå(®áŠ©ÛF/rüü×¾× m(—°R;ŽÅ¼±_ÑáN¢’TMäJ»*W%k™>/_+´ Þ 5FÔ¦/x¶äî&ÉtզܗLÉ©‘Z”æ ë[+©è°d®øÑšE=¡¿‡ã©¯:W„£§èDHC2%kÓÛO5”CÚîr‡*Îζ)§v®kY^›é5ŸEQl¯6 êëÂ> õþ™À3¼k0h¤«Vëà§1Z6‚¤ 'Ú1«jËßãßõ|?f‘ßüsޱI“U'_v'S54"ùΣ%]ögê/‡=§yúXꘌJ´dNjwdV’ݸfÂD=If©¥tk*¤É²…ÚIÂÖ‡B©Ñ†µb-Z½âÈÐfΰ:âW@Ñ©/“¬$áa*D…Iú¾/)޾úEŽ[“+:-ž2áîž]4ŽÎmMY"ZC¬gYÙb7ެÈÅ`©ñëÈf¶°°wü ««ÖœYL†=A âcÌ•Wûóãµ+TÁíÌù’ñ[©—ÖÑ„rÂG™"òRä žð…LÃ5üÊÌøš±6æð‰³2 +~ÜÞÀ¿¨‰¾Ù¼©4î§ôYP’[We`Ш|`a­nêó´_9±vYÆ+òGù×Ã\FóìŠv!UÈúÆ+ãë£ý¡ ²,«“-™i"̽#;T>,9ÉL$ñF'ʳ˦Wÿëp­x‚ÀKwIÏíŒà]ÝY³»ÈYÛ;É¢Zó´­Î¾uÏ9Fq"ògTõªI>n’´Š³k€dgš=œq‰Cý0«WŸ²í|Ud˜YCü!ŸŠ¼˜¶jÒFD↊ôu·X¾46KÜ´àh¦ô`n«ÈžýU÷F'osÑ{A ~º^ë·g‘ót±s‹ÔD®CáUæ—´=š$bcoOæãÀCÇõcˆ»ó˜¯b²5U‹>´h,ž7ärf|Ç{œË=»6óIíi-Ê˾EHa£5{eË+[‹¼cŒ{EÇ~’3V3~Z†2(($µ—ùTcJ‰MÁjÞsæ¢mRoövOyîÑ–|@mÚW—Ú…n¼eý=ŽÓcÖ‚s¬d@L¹Ë«]×°VŒØQÒ¯0×PÄÔûݲ‡4z-”•aù>Ï)ü“ã?zÞ¿sÀý$¡0±'E©üìÛy¬$­¾YþÊSªÔ\8Ö¥*$+b^W¡î«§öþ&³ŠHQñ.ñ>^^®U½@¾uÂÛØnÿÍjŽõfãéoË1*¥¯`¢!t&w}—Üólö¤~ª0û±—?Š0ÕstÞëN­‰8 ñå_¬m˜÷Ôì6óoŽV¼ÅØ÷—ôÇ PiÏ­!†ÖÜà¢éX®%ò~Tíûš¹&¹õ”æwí(XGͨêU¾†a°7rT«€!\7š«îR_ìPÜáåò1ò㹪§(~i<”9n¼û „È0w\WÑm,í¼s'Ú÷xS¬¯×¿³HÚþÆÙÇ0+.¸»ÞصÿKú Íã’aG…AšÄãă¹·-Y·Ä}qZ7 £–ù{POZ{3"pJ \{éâ±}îcÓjOóÎUF{§ÞÎù-|<¸#çÜvÕ3o¾u«eÃ5ô̱„‚Ïtyሓ¶£vöGãfí˸Þ]þs’Ø<ÑàX.%áéö‘ß ÏÒü¬Ê¥Y®±” |™Æ£¸^"Z0¦pœc>ðjÿ Þ)©D‰Å÷jxé9~ŠsËá°¦ ᛉ‰z>ÚµSC¤iýääýÍå¹+ñ]h”ØŠõî.s±mBk™Ík½F(} íºÛÛÓÞgŒƒxxÈ$La>ÓÐx,úɑל߾ÐæïaŸÚ hä¸TT Ç8 ¬(À§ƒ}©ËU^3¬ÜÝ¿*3ÑûÊê6uOýMóœi!mA—JÍØÓñ’{êAt¯EòƒB†PÅJB’Üܪ”¹*J¿×,ex¸ëLÙÂEá· É'±ÄC¹ªèvM%¾wlJ÷·eIp,i28`ŽÄQv‹'±’œ%¹ü([‘œãå’»«5¸¶ƒM5º0ŒØ}ç͇å>rÀ÷OÇn×N¶¥wg&,hŽ.“¨ï}µV„–^YJ =_lìbúvwÉ(òÂ?#íîaã°‡ £˜íǺÁg—“þR7±Yèß3]Â:þºÝ¸[È,ño%ŒÏÖ7’¹¸Ýä‚xªyÞIù;W “+ „ˆfŸ&J:H«ªOŸ£' äûʯ\JG7Ð<,HCé笴Ä`ò1Ó{Aˆœhi:Itƒ^@Ög\ÝŽöÕ#u²Õ5«iÁ76‡Ûr“B%«z”_;åOž°Š+´P%8K>µ’Tl=÷ó©²!:+RÞwœê¬}ø\>sk={W®´âÎJûã7c‡C¤÷—ß.M<¬Ç‹aŒ,Lßx­ëpïÁ«cžƒÞ<‘“µ6úî…%‰ŸÞyÚ~&j«¯ùG}Òè}¦Âr“GV´=Æß¿Õn'fˆó{LmgÆ!Í·* "j§K Ü±á§”D9XÔýÔ+xžf%“œ…ÒxeÓ‰•sêëÂââÂbõp§ŠûìДQEÖª€m’&{4tüSèùiºÛ}˜^Õ‹d½/™ÑØv+¥”ünÇ·ê_Ð&?îPžö¹ž–#Oê„ïPè}À»)pÕxæÇæ+ùhº37ØõhÍ*:O¶†ªÁ3…å !mŒMsd­—¡®‡_"²­e¬ú\7÷)7ïóʲ|£§é5ÊßÎu™[5F#"¸+Ú&Í: ¢2À ŽÔS|¨=]–f¤£ žø»;fj¼ ±0Û×ß÷%m§)7¿Èã«JÔ½~±b¦Ê`'\¿ÚªÚï×äÚ¯om6‘ECÂCø‚E¹ôCšªâü‘1­ŒÛžÏDzó]Ý%Ve©Ö®þ>ùHÜÓùÓ|9]§Ú'ÚuMØjƒ6ÛQA ø‡†T¾K“›¦ÒRb/ƒoÙ>D++ ̨³·]­|b± Ð'À% ÁûhYD¾ª-èâP:Fx¨‹ =$Û®;ò\Ùî< ÔX0ñËÑšZ’™+Ó:²ú!aB#y—Üó~B8YŽéª$N˜üý$‚èSŠx>Ý5I™<&ÃÂ(ßíF$Õ.Ö©‘ÕúëÐ$© Ó˜ëÛWIÂuoc‚܃´PÒ>RU\kÁv XO¬‰ÛQ½trî5ù0ö†óÞâÔËã[–'gsyôšÁ˜;¡ùºÉl¢()æ©‡Š¯¤†ãÃÇqŒ’œÌ_k%/ RÓº<Û‡/¿+<©•éTmåORf1â«Joâbžp/jèM…i˜§*&¤v‚¿!¿.í…z#Ô­VfË™†½.Et>ñ<Í Bí  “¯•Õr¿"/j|Uùö’A4¦Ö’õz¦¬Ïgsc°'Ï0/VAøRÅÝqÕ·‡Lx˜%>2=–Í¿:¾¾3ÿIê'•òõ9±Üômâü*\-ÏegjTIèÚ2€øÛXüÛ^¶Õ¨U¯æ°É½p¹DÚ(W¾Ø—Ï* ô-:¨Ç3Úœ|Á¿fÙìi¨¤˜#y+pŸô0ÜR+ð9x†Cb_#s±FažùQ#Ê1»¬w«°«l“”B66ÇðU `štÊÌÚ§áo«§ð¹àýÚ‘‡bDNr4IÀ1[ÐU˜8ßÕLI릭§<˜;P©“[†h@¸8Š &~bÚ¬+=wWèˆJ»˜\¸”µc™V±‡¾Ùg(¦Ê!¥ìzdTòfV{ALaï¼€œ%"‹”&Ê|@Gû-mÅnšjO)έ$Zs¨½Ž‹Ÿi+æ×kuëÚüœ?’ævß7²ÜÖÙg”Eö´ž4OÍ0#pÍÒN¯ï{!Ü(N¾f2e*õè>U–AÜëqÖ%_=ªdÎ~ç˜,ó’í›<\ȵíÂû,o—mCxÅ)õø{Š6´j!A¹2sïú¶'«ëo1’í:­cYÅÛv\}mÝ®°Ù|4q>’ÖÇM¶‹¢ï*¯¶Ë©‰¢QÏ¿.Ó‘-dÕJT YúJŒ*®ŸóÄ ˜¶¤mæ·[›è>Ò2V¦ ߌr»¨mšˆK;¤”5ÈÆÏ{N„P‘ 7-ùÎåÅÝþÌ%u_Nöfóú,(§4cœò0Ÿ™Ý¡¬åœC‹ºP._'?` ÄSM²Äàyå=ꂬBzï©æ>RÖ _[œÃÙs*¹Ýl.2~/ÕgiÂÉ€×ÄüN< Ö¨ú=öÆê³A©ÒõE/íçµ X¹D4˹Žð²îz˜W·Íx´0†IcI(z|ˆÙzcâ‡ããkq`ƒR‡¤³Ôõ x–/ÛE¹7­0´¼C óãÛŽíAü·¤ ¿¦åR5j…4+,~¯êç®wñ* >°¼ÖûÉ8¡ yRpœØDåáE𹣘!\¹i)‡ø sغR²ôH„ƒ¤-6i/‘çOÖóû« -öÀ3¸V!Ìf•t#Zç¸å Ü0Ÿ˜C”§÷´„‡½^"•‡¹ÍÝ7†öÌÉå/í¸7ßéDXÍÁª¼?³‹x­‰˜¾Q ëTÔ`ÔÁ£yéÍá(ëeåPlœ9Ôe±ød'V@ùÇWFkR9!ÏWý«:Ùž“Uô ‚·<‰}/6 ›–.†€óŸîÚ=8;e‘íë3–4Í‚ xµà2-œn¯/·kËÑ.ë— …ãc"Ýyµ.}‘Xµé#û¥÷¦Ùhåÿ÷_ ݪ’ý`¿´ÃG>òv¢e0&6ïø–Cٲȩø¼öùØ íëÛ–]#˜_˜3ÏØ=öý)™¤„~ë⨻Û^ZO¿N¹Ç9Éëרß\­3s!H¦_ÚUGêý~kÿ~”^/þ‰—¼ãxâjÛ•ÁUkn6ól9,ÛîðÔW²¹&CÍgÚRiçÝÐv´›{©‹Q£,®l„íí ¡Ô X÷`4ÊàìK+Cðç¿›~ÿùÒüÛræ{‚õ„J$ásîžùjd‰ù;-þÝšæ’ïŸ7í¢ôÞОVÀîä9÷žhh}ñpÓ±38¥ÂŒíryqnnŒ"~w[jGôNŠÙ½ÌºÓó÷‡Û'#d`áÎãa²™h{¹¤îË.´W{4W½x­‘¸0ÆQß,R™¬´P Õ,ÄSlŽY˜‚5çÑ;ÆÑnÁà/÷Y^G®hw8ª8QôÜ6Ñ€’¸HÅq«Ã%1+BY Ö‡DØ nºN)¢P’šãßh.·OÔ'ÞñmÒs ¹ÓS2õÃè ºŸ¾‹mÝþY,²È¹ÙQ@†õ,ñ[ÞjlBq‰?¯÷^î2B±¦I×ß0tá‰xèS0ïì>¬ÆÚ`^õ±(¸àþó&P¼;Qs8àΡnœ›°?Ï{ uo®â28‹$ÉFÇøÇ”-Ge×kxïêëÓRh½wËåiHrRÜdâ:FíÀŽéÊK±oOpˆ‚†b‹z"|å¼–(þ쨥~  Vuýªñ"¥!¹¾‡º‹Õ޽Îp²ÛòöfëA)›‡SàP0ƒ’ÒÕóŒöãC†ôi'Ô¹oþ™œ± Ãlt¬˜P¾ækõV1Øÿ{9l endstream endobj 209 0 obj << /Type /FontDescriptor /FontName /ZEXHDB+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/greaterequal/minus/negationslash/universal) /FontFile 208 0 R >> endobj 210 0 obj << /Length1 1393 /Length2 5926 /Length3 0 /Length 6879 /Filter /FlateDecode >> stream xÚvTSkÓ.H ¤D6RŒ!ô"J©Aé%„¡$”Ð{SšT©" Ho¢ô*E¤HW.E@¤ƒòG=ç;ÿùî]ëÞ•µvö;óÌÌû¼óÌÞ›—K~]ÁkRÅbp×…¡0@I n,À`¢PLÄËk€Æ9¡þ2ƒxï¡ÜÜÑXŒÌÿ(¹¡8¼MÃã´°@Ãà „%d„%e`0@“þˆu“”žh@ h`1(w¯ÖÅÇ mg×ùûàG ÂÒÒ’ßဂ3Ê D`-Î匯ˆD8p,Âùü+¿¬=ç"#$äååE8»C±nv· €gè£ÜQnž(àa@áŒúà â ìÑîìp¬-Î á†ð'4…qÇGx`lPn¾8W¿踠0Àwþ À_gC…ÿ“î¯è_‰Ð˜ßÁ$ëì‚Àø 1v€-Ú è¨Þâ¼q±ùD8¹cññOÚ aüÞ9PUÐx‚ÑsGº¡]pîPw´Ó/ŠB¿ÒàOYc£„uvFapî _ûSF»¡øc÷úÓYG Ö ã÷×±±ýEÂÆÃEè.íêRWþ ‚7þ±Ù¡p€8LZBBL@¹(o¤½Ð¯ô>.¨ßNá_f<ƒ?¬ `‹' @Û¢ð ?w„' À¹y üþ·ãß+°0`ƒFâk”ú';ÞŒ²ý³Æ7ß í ˜ÂðÚ`¿~ÿ¹3ÇËË‹qòùþ»¿B*Fºê×þ0þOQë ø‰×¥ÅÄa11)@RZøw–ÿðÿ›ûo«.ý×Þ`ÿ$TÇØbé?ðg÷7 Ï¿TÁÿ×Äÿ® ÅKðÿ£|3˜8 ‰¿ÿëÿwÈÿMö¿²ü¿”ÿßRõprúíæÿíÿ?Üg´“Ï_¼’=pø©ÐÂâgóßPCÔŸIÖBÙ =œÿÛ«ŽCà§Cc‡Wøua1(Lìí®ŠöFÙè¢qHû?:ú»øNh JëŽþõÀÁGÁ`ÿåÃÒÿPqÇ7ì· …Ÿ©×UÁ ±6¿†OD\@¸¹!|@0¼ÆDÄÅ?aü”Ú ¼Ë‚b°8|€çØbÝ@¿ÚŠ×‰‹Ooý+5ÒÃÍ ?w¿€¯û÷ú÷£PÞ($èÃ$y#Ü¡&¼ù¨JÝëúÊ éübkT‚qw¤8Žo*×ÏþY¦Ú¨«¢•Mù¥·iºï‹ïLú³ ~=ò6ëÈ ™K½ƒ#Pé^VܾþôÍñ÷;Ôe,›Ý°dZÔ2Ý·D=0•)±éÕdÃɳÉNѵŠÕg?ZÝ×L¤7g ð€–+•÷P!»uóBŒ1Ï"®ßØŒ6ŽÞÍ4ÝÃ@ýîð 1è¸CÅæXÖcŸeÓ¸ŒþÙÙ¾¬žî4ÑÝ<Í®J‰òû ¬Œ¾\Ùo“zY™³4^¹ð)Š(§$\a-p¾Ï7Gd%9w:–×Ó”Ø+êQg§­B©8ðÒã ½~É—˜Îfl¼*-HÓÎÆOx#iIѼ—àª|v[”—UB)·Þ âC+!¬}¦PR@Ð[I£"Rb|”œWNû-Zê›—!\e1wiDzT3ÆS‰÷žRŽíýý­W[z<àJ“µ˜Ø6‹× þßèÒ¼iγ§A¥ ±«—êY°^)øXyì8tzÔ»³%m%†D¯;&FÔ»Féf¶‹3^uV¡räYæj<ÖTéáƒH½©êµ\Šû&—z&ÖoŸØ_@(ÏfÁ;Œ|t6Ñx7Ó`2ô¡n ôò¬™ »yݦ몭ÍÛËHÖÅp‘û Lde]¹áÇëß5‰ó¾¸d&R.…²ø×cNTP¤·[­‡ƒ9=‚ !¢.„‰Æ¥:Y‹OÒu2—S8øî¹jÁ`žj¢JwòT¯ ¾„®ü˜‰l¤©\¼wÓÊ_J¥ÓìñÅ‘^k¸m®ÂH”î“óVŽ>)­"+Jær+d¡p„§­ÔÐ6ä…¿]•i#㎯8ç÷ÔÊ–FQ‹YJ ?íííü7?-ü 0©èÞæ¬ ãzÝ}¼–ÿ>²ßYOõ2‹îÐ[‰Jyuæ­U1º“tO£ªõïK}ŒJfÛ·EvùmОåãk-_à‡ eˆÚÛåÄåŒ|$‹Ä*wºhü´Ì/–¿T9K_ÞüÞ4ÏO%X€z×ãAŸ\·©äóÞJ\ª7¤™ç\!Üã0l•í!–~ha Æ^ý¤œñøÈ>Ws¤hý•Tw¹`€®·½ünÅŽ ›FÿÜ$ ×ÝžYÏuÓþ™Ûß;&7¦]]×#¹WÌèºw(z¶uÙô¥šÀñm…Wû±»w˜–&UÜhA½a¨£‚_œ>àø —c¼ÃDÎhì‰e°!D2ûWÈuÔFIÄïtzL·ÉV|­6t¹òBʚ̛Ø#Ûè©?óás€Ì΀{²J»6ßÏíµÕ¹;c ž‹¨%¹WÔ•mé‹‹…Ï­>âhµ[ʼnÙ^Ÿ×ˆ®¦°ì®F÷qÍZ?ÆC®T›Ü®T ½Ñ«¡Fãà,Ùm‘z#ŽJ;3£Ü.qô舭mSÛ©¹m˜[Cý «ïlÙÇÐ"…éê(ùøvС¹˜HÍ®úƒ®%!PŒP’S> m~©¨?LÇê’%{w‚Ý‹z„­NºzÚ@éΛf÷H E'ÁL3NýÇ#/ôÜœæ5©è×2ùûܧcå­¤©îTP¨ª•9õ]Jz‘7fžýˆæü.áyÃðR!X€{Æ@œo|ZŽá1{ò‚2â4 qüµâ½ ªçKu’ öðAû'uÉ"‚Vw£øD>$ÏÅÀ† ÆRm±ÊŠ2Éß ‹úAFо©ªnL†m™IFÐ¥#ˆ„¯ñójÑ€~áUég›‚Îxî W§R¥hX‹~Ú’4·ý‡kï5{|˜§"Ù!"ZT)P‹cÁm0ù^ÿýþœVr ûׄtd Ls‡7$òÏEM¬jT³Ì”åL¤ÇdÐõÖb¢CV\U8üìíŸ÷V>I;^èïÛ}~¢X³¬ŸÝÉâ½²qÀ8Ðý|}4$Iá¹Í,IÙæ$”qÄÆ×ÅÖÓÓÐŒÁ3OYŽKN<¹¢`XÖáÿ¸[/ìãžÃzéR´Ü"zú³ÙêM 2êÑ˔ΊýÍõÝï Á_ouU(îr1Lr^SO> "=J#x4;¯`ø©M¥††¶Tÿ©BÄTOpÕa•ü­…Ùfùܘ÷õtî¹Ç˪ £³´HX&¥•ÐóNÊð‹£\*»úVLÀ†\±Ç‹SЉ3M‹ÕT°nàp{:ï®åO<¿þˆÛçíÕPÂ}Å7.ls_½ÃpüêìÑuï:ûV¨2ôȃS¸ŽÏ¹ä¤žç²ÊÆsb’Æ{Å{Æ÷ÐÌLþ¶Ò複ŸçN¢Ø%@¸— ¢º 1“5‘×ß%[éª?ÉÖ‹s¬ ŸŽ ‘Óß»g¢í~’š…Ë[ ±éÏzwξ1÷dçÐ"vÎ!3=ó¡Ý&ÿš‹ê°z˜ùÃ+‚âÐ;¡MÓóW4µ#ïUf ¯l `‹€r+²§Õ¥yU”œ íD¹ }е_´ ŠVäëQVÃת5æj´;ßÔ¶ÐE¸ÓövD¤‡° ÇGŸ^›SíïŒÓÖÕ2Øi\˜wäÑ#¬•tët~òê„ÏŒi¾Ót¼o"¹o7lŸ•ôêÙßÇsR‘ÊÈeÂqQ}[u3Æ®ú̇„/“^„X^:f~l.¢Ýðqv`TçhçR< &–ñå´ï£R0ñÎ2ÍТ@ÀQ±‰.ðÚ¸rM±ê^ÇÇŒWïÔéšÈ—š‘´œó˜Ü»õªŠ,Ï)ÖçI„ã²3BI†åN1}ò¾šEZš;³õሠk!H=Á1ZPbI.HSò™¶#Ò]ÏO³“ ÎNõùÃí¯zcôÅ/kÔ2MæŠ]–&GNCü“üÀ–OYò&É̶ rï0."ê›NEÀ-àuå‰ÎêÁ†È¨¶Âê×퉞jKMXãGï\>YÁ’Âö™‡¼È†ÎËd ð®ž!ÿŒâ ¥×z–ŸÃ…ió&ËuLè(/ÙmzõÎÙkçaÆïòĨs÷]ÛÝÀ=zöúÐyHd[‡%ˆÃ…rw7_½iHã<ø˜"¦BâFvÀÔvœËÓOgK¹br=W,0^ñdkïw\k[cð¶ÁŒ¼Ÿqs³íVóÔ)Ë7õ°Cµ1Bˆ÷w‚Á•R;\îõä 7Ù#ú’tï?iú¢¬lVr³ÉGù‡VØ58Á~O[>‡cûÄ Dó¶t2Ýji_Óók¥ÃÕ¾à|õ[·ü_oùÙZÝ¿³ûx SµJqRXmEyÝÍÎÄënËè®_‰g4àØ‹sÈc¿œ3 ¨ÝJDm-k®@tF&¢)-åÅp3©!Œ¤V Møø1j9çµ™Þ¨¨5ò4­(ã8ð<6tÜ[p(4°¾5!bù8ÈüŠ"z~é•¡Ra9Ê?DH©%*Ãf¨¨EºI/®¦µ]zO—<{LVµ$e-¤ÉX\:™àü¬¯ïØ´CÕJps‡rEePž¸ØViÓì€Êrã„J÷K!Özî²khâƒçtÁªÔ晈úVLµg o-Û0ràm_ñ¦ÿQý¤¢ÿît¥ý¢`é‘• y$'Üñ~”ê§ñŸ:w¸o­U\R=Óº\™êŽÝxlØŠ¦NÕRBÍ«2<1ôñô°lb´›[jmÙ½ðß¼ÈÈ6› Ù}¡ 5æT,Ip½<÷óºi²hôó":1ë+hBÖÉ7Jç÷Ó›'ò­sŒÅów3ß­~¿…z5DH—çj°QéL q Gèe˜’¹³:ièŽìö£î‹z'R±jír¹ï‚ÒìÝ—z“<¥®Å&®À.HXB£üÃ&Dȉ3]ºÍÎ>àI›Ò:…ÁbžÅ{±‹ƒãõìà ôݰÇþÝC› 2Kß>Ýy§BF°8MÓB¥kPïR¢R1h/a1ÉÏ`é@_»Þ´¡w話ÆU¯Ôú4àÚú™ë¢öÅè™óˆ•Ÿì9$¨×=€F#íd±æº:mDgWåN/k”÷¼™ùgúü+=SC`Ó–6çÀÚwnàÁÚÍo=<(=Nb$ÅICéÃæg»g¤Ý+ ÞðÞ:NƧ”n+Ù¾>-Vw†¬ÌfüÏ#—ûUýFuJéÈkŠÅ¨³K'<0,Ƨǻïìí4å¬JǦ"£ÌÀ0†Ÿs®Ro^ÄK“]o:˜Ü2ã’up•1ËŸð×5§:½¦éUf.è‘w"ë´êÆ«¸¼ºt¥½çËêC§ºÝDãä\&³t!3…„îmäz--¿ßýL~YiÔù šŽ³¼»à,“HÓ¢m»ÍW–ÑFRpªÐOÑgÄ­³eˆ =Íòöí’£¼Bu`úìÑ-û:Yµ–ævb·ò%” vgç\|ó¶ŸÁ¨ ¯Ð+?ìvuZ^þ¬tñäÐ:ѵd‚<Ó®óúhÄOP)½ú–°±ã•NÉ©þ2ûʸšÍ¹[ƒâ3Í1£;Õæ¶ÁÆû>#×ÄŸ 3L*NľӤŠúäf´qÔoÔ[¡$tïãE£lôfáàœV³^ðÃÓûU„] 1KÐ'»Waµ¢.á‘ïß¼«ëRý2âÕº|¿l‚4²r+äï0 ÐtФh.š ÂX\I =a’6zßÄÍ—KOÞ£Ž<râàëQ`Úª‹¤y{¡9WúOª†5"žÍô:àã:4.½X±Ä¦ùfêzЂáIrŽøÓ‹Žº %òÌú¦!´ Üí§ uéÇgž×}Luº*R<‡Î¶ÌŒŠmòo—½áÔÖǃÏÞ¼`]‹UyÍí)ÿ‘fxö%¸Ó¤ª0¨2¬¤œÒU÷)‹5O=c2eþ>x§l}µ¯ñŒx‚0«<´›¦¢:ž ±Qø)Ò¼ÊTG-Úv!”‡v@û%‹Ÿ¥Üs¾ôºi&«¡oF:Ž&½Øä¿Þ¯#9R«æS ïæ¦sW0¿áò›ŠÃlÝrÉ”K}O3ýÇ/V{®ÄøPîÚIKë ¬Þ-Kî.ÛÐ\…6ݘ€ J䌒ãÅèÏG¾O–{欵<TQ¼÷xv°w¢,9D¦ ’u—§Â‡èµÂ4‰Ršž)ÐÈ|ž˜8œi7+.Î÷T7P™¾Ý<ìâjÊÆäIj„óÕò–ú ö¬Ts³‹‘†ŸM|¥î¬ÏFkíç±:7¾6·ËX:éÎŒ[åÕ5Æåz±·^¾ìÉHÒ­âæLm¼¸Î_:SÔí«Î¶8P~íQò+-'Ÿ—.q”o¯r¶ÕGx¾)rè¿åÞô‘£ˆÜÂJ4*’iίÙhœÕ…†CY+ʎɧëù”ÈöŠÔMÅj)kQR=¯$nȦ׌,ó´§°; xã|ý -êïi¨Í¸û=C"³t9‡@½`JZ¾ #ÂCÉW¬;ÛM…Øe‡¨1}ÏKck­5•nN¶Ü5¡”mmö5Ý8Â.¸f|Ì¥]vD½˜–¿âgÙ˜×'w§™'2É›ô&}ïWÎÂÏŸj>í÷«šÔ‰-¿‹þN¨L"ðLwxep«Å/jfØ¥)š£îñŽ=úëŽ6R;«ù»hΪJÔl½Uf0gQJð9¯ýr‡|ÁŠ8ûÑ 4ƒ”¡Q¨oªÇû‰ˆØÕH¦ºF¯¸È§¸PzÍŽqÑ'c×¥Ìì‹‚öYŸF> endobj 212 0 obj << /Length1 1411 /Length2 5990 /Length3 0 /Length 6951 /Filter /FlateDecode >> stream xÚtT”kÛ.Ý(¤ƒÂ0´HHw#]Ã0À30 Ý”„t‰”JH "ÝÝÒ)Ý!‚ ÿ¨{ûßß9k³f­wÞçºë¹îûº_6-]ik„TGñ€xùIJêºF">>^>>~"66=ÊúLĦEºÂp±ÿå ‹„‚QhLŒBû©#à7GH‰ˆññøùøýíˆ@ŠäÀî0k€:/@‡º±É"œ½0[;ºÌ߯'ôè‘ÈÃßái'(Ãê`”Ô ]vè" 0(Êë_)8ÄíP(g1 ÐÃÃìäÊ‹@ÚJr>xÀPv¨+éµü" Ð;Aÿ0ã%bèÙÁ\ÿດ  G wEG¸Á­¡Hº8@WY  é …ÿqVûãððWo ^ÐÒýý+ þ; œœÁp/Ü`s„4ÔxQž¨‡0Üú—#ØÑŽ»ƒaŽ`+´Ã ÒÚ0šà_ô\!H˜3Ê•×æø‹"ðWt—åáÖ²''(åJôë~r0$‚n»ðÏdà¸Ï_ÜÚæ k7gàS8ÌÅ ª,÷— "ú³…¢B|„…EPÔbü•^ÏËúÛú£øù8#œ6hP?˜ ýGäã v‡PH7¨ŸÏÿ6üûD¬aÀ j ƒý“ CmþœÑÃGÂ<&|hí|¿~ÿy3CËËwôúÇý÷|ÆÚjz2ÜÿÇ&#ƒðøð€„<„@?ú!"Êðûwšÿ4àoò¿Q-0ì¯Ëñý“Qnƒ<úÃݼ¿y¸ÿ% Ž¿V†ðï ´–¡Ž¤oÊ'ÄA?@ÿß ð;äÿ¦û_Yþ_Òÿï )¸9:þ6sü¶ÿf°ÌÑë/´”ÝPèµPG —þß®Ð?«¬µ†¹9ý·UF¯‡4Ü-q /Ÿàæªó„ZkÁP»?Bú{èŽ08T á ûõÅAGññý— ½uôWÅ=°ß&(z©þ]WAXÿÚ>~!a‰{ñ¡EÆ/$ð¡×Ôêù[ß /B‡Ðý6$ѯ±ò€N0¸›ë/ô7€Ð C÷ãò¯Z7$½‰¿%¾Èßçßk…zB!D3SÈãPûªÐÆoÒ <_†ð–VšŸÅuE ¡Ø§_úØ©ág(Ž»ÈXZ—ÝHÕúü&|lÊ—ëàÛ°§i[VÐbŠR. C¾kMæˆ'¯ïò´ÍÊl ·I}¶F~˜ ÍtÛÒÇäA’ÁÔ0‘vÍò‚ŸÍ®›ÆvN€¢¯d,) ¨Xéœæ]¬h#ÖT¿‘)YÒ$ÍMOùéÈO¬hXÜ…Lc ݥךI\zÿÂBoVwWªÀÉ+ÕŽwÂeá”tTÞ*,9‰=tÔ Y*u®ì2B_eß$Çߣ+t g_Ävªš12²ÆuÒñ9Ší¤a‘˜‰HÍÛ*#5 cƒüì½â,?¼½ð^²I ÝjDê%k‚±À½"°2ȶWqBóFóñ3Åš/KÓÉîû`#Öì(Ýž£;ûí²¹‹†Ñ.GJ·44¯¢–7ÛÒã$`ÃfwµË…äÆŒguçÔªH|µ… 1ŸìN7LeéžSPÜ‘½çxl¸1¡x¼»‰4%ã ϸk‰ûî†ü¦ÉŠºÅ•’úIä@‡ðnåŠÕ¬sÎ’±U­ü(yK@6îá€9b^p½f{¸ÿ%IŒb…î2ß¼éµ9pþtôDþ•Z–?ˆþ˜ŸòQ2R„wc_ƪY'«ÒgñËÛ@yÌ*b]ªNÞšG´8ý ÀX|' ‘Šcâô÷k·s€&邟Ší½›¼]–ǘÒ(ã%ÖÏæÓxï¼ó\¹&Ä´Jãù®¡ôœ‚6‰¦ÿ\,E.ËUM¶€ñcê5 «f*ùáõƒ ·´“Ì©bŽÝ|/ Ê„ÕK—@Vç½°l×®yäHä]]ÙKûê‰î¡ ãæî8¼¨e•Æ5ýÂóÛœJ}ÕßÎ]ÀÉ iR­Øí§ ïÆÅÙÀªzUÆ"ú,¼½î‡7R’åOñKFdügNMY£dE? ·ä¨|ÀÐ]bª!{Å•!¨ëd&?îY]Á«i¾~e÷p©,,}Ä+¡›í¶Õ«Ëz•o£V@þ"iÔ±Pþ—u§ûšb*öx ‰ÙsãÑá€ÚÙ3VA¥F-÷|ž#}º-°HÏ¢ 4Em2v®²ßÂÓvnþZí“ˆ d¯)”·ÔùÀé2îÎdp×'Âr&¢¼j ÐJU«åÍœM?³´æ|‚¥´'¿÷Ä€#IÛ¼ðËùPO?ùÊžœÝÙÁˆ»É¦yøqïÉ=¬ËkŽ;äëâýô›Ì©ÃÇ`rV«¦óbßX ÁTÏåó¥&’›Œ·DÅ¡äo8ŠbÓ^€Ÿ@öuçàÂÎvŸ3Œél¼½O¤¢Tûˆ©fè®X#.kUÀ(—“þA²–‡á81C'¼§ ›ã–‹¡¤Ù‰*T2ä?¼’úIæ'wLg í5–ÐOMI.Zz»&R' õñ5ù‚v\zW5Ç´W‹Þ›¾ ÇKÊO«VìœÜâ¦ÄúTLͤ:ÃÔ…þ¼LégyßK3tˆU!EGéš‚¹¤=ÛO?:áÆ×âº%Bࢠì|û˜ø'‡œ#‡*\T@ SX\Ú™Þnt7çòg“X-Nõ®ü\%v·?¼ëQ}6{k.½Ì˜XŸ ‘^~}”AÜ>}7î›K|y!µeûÕ «ñ—æ¾k0)›±m¶,e ‰õoKFŽä hî&ÐIM®Þªz/×è Çÿ~ãÊá( °ŽïYgh#Ÿ]V ¹;äL5„ æôɪDg¢œ²—ZyMüÄ/ÖD~’Î6ÄÊáq©-Šøë¸ææ6H*_VôÅÆ¼ó \ìÏ¢ýúYîö³¾ýܯ›¸¨ †›õ ž·»saÛ-ƒ{,S†„åˆo3'ï´;°sÄ©4“ ézв ×ìí—® }d,YΩÆ© OÕ¯ßQIùâò0$îi™qGè‚I¦Ê§&mU:˜.µTÑw«aÕùtOøb+Rz©¿_ÏÎÙâûöV(TZê­|q8L¦|žÂŸ–=žÙ0ÛqTv?êî svß¶§U—eˆQÐÂ’¾8Â'gr‚¦{|áþÝ_æÀšÝKQÃÇ•À ûbå;Ž–ü¾ëílvÉûꃆÂeòƒ,ä JgBÛì2îo¼Œ® Cމô•‘:H´Œ_ë.)ÇŒIGã4(QrFøx=[¯Ê¸3k»öQZyC"cæÄë7ã[/ƒÓ¶Ž0&âõh(Ë\sqÙ!\³“ôÁ÷W¿ƒ©`dÓ=¬ñ×É´ØlDP—Öû\LS,ü× ‚%ºé_î•Rzꪟ)X3䟷qGN7ÞzñÁšž7–™©®GfÖNÛ{´ë±ÈùÆ>†á0Ýö{+Åg=PAÄ•çù‘«Ù"Ñ2èÔÓP!÷’\ÉSxF¤iuE‹„ð‘aÈ íJ¼Â>[À¬õÝ­ÌêR–‘'K‡ôø~~¯£\äin1$—zY„ƒ –Y`Wޮͪ_»e€ï8µh¯{i²8 =ª¯FÞ þLbJ1Dñøé³n Ô©yÌÚ}š0Ì¿Änš \˜A*n˜DÏ®ú÷O’Ðc X‹w•GÈåÃhQ¦á…7þ!-À7¹Q ƒEq·r‹2øÁ¹ø‡úü³B’rgj߬iÜ24DÂpÄüÙzžû ^Kè?-.r²bn)J]å!FÐ':øg³É“&ˆ}’&£x‚U´>ÂÆÂ¾—;ö.%,YÀF4wÊñjÛ:䇺N&ÏöC¾7 —p,&ê,ŒF_=ÆvÝ:ð,¸bo¬Ö™« qË=Ê$¤\$š>–ÉŸé$«‰'åÑrÝ»Ø!lK©k.ñrˆß|ihYŒÓeÈÔXUÀé“êbåf}‰È¼Á·¦§Yé ã4–eÙ5$]7óÕïwúŠkUê2å—¸¿UZÒî’0J_òälêΆ "\B¾ñ®³þ]¡y6ÙJƒŒ–ŽÑ[Çdzoà‘5§\.$ü‚ jì‰~éBúƒ€Ÿ¬X n¾Ì/Џ˜ÌÔgæ»Æ“)×kнâÝ.wž~-œ”U+üq«âˆé©áM˜£6µÇmÏÌSÝ1åÑ"¹©iyïAWä=çK‰\âXʵõz/o–—y˜$’Œ[‡¯zPé!SžƒÃ§q´x}ÔwÜÈe…Ó¬’«êWî:¶Þ_`âF8¤{GÚ²œJÌÄReužV ü«wØqišzÿ^8’·ÿ¡¥ÓåC/Eä»hÅ×ÁéŠÍBd¸âÒõç»ÓÊ%N‹þUkÁ¯LCî uî1Iç•_o˜‘Ò”šþ€/4ðCÉYØOz·("wkŸ±šÂä~žF|Íô™@²nÄžÜ.›‹Î Š”˜Ú¢W1J…K`K<çòbǯ‰YÁQ}dÅcô~}ôs¨çùz§’”FXÒÆ ¼Î¯+”&ÜÄx“CêИ°.w÷Døó« ¤³ÌÂFÄÞŠ4¬ÍþÛ¶¨I×UÔMíIm”í®žÏ¸PŒ³ÿs”"ÕœÇd$å ¿f‰á ]á–ÍDCº÷ïÕÏ—–¾ã%Èk`–’ǽ®&ŽXŒÑšÃM* ŽV´zü@ˆéÓÑù»”û“$BíÒê×÷;–F±ÞêbêÚµ¯¤åW%–Ü£\xäà#ŒÆù¿TÏÖ®}y«þ ‡9°ñ¥¡ÅO²*屪BMçŒB¡Ö<ðkÒp™]¶îÏ^Ï7Û1fY¤þ<ûàþ3àe9NEj´ –¾Õ8Ö( Û\Q—'¡kMêªK½Ò‹²sìè¼Ç!ö/ÜÇîôù\ì-ùÌmñD8p+i’âV,ý4*º‰:ÑSä§9ôŠ‘,w)åÀ]Xgÿ\Ç”;Ž4)ÿýÏŸ¬Þß-ÂÒZâËü(D–f]GT‹£'!ìS“p£9#ÜP^:çkN® ú±-{·;kÝ{¶§e«dJdä$ f¶ƒÏ}M‡ÞIJ¾†9æÇÅlg,wƒ˜ïÕó7pÆ Aƒkœ×6¨ØßÈW7'ï™Q/Ìr{Æ©Ðác¯(on ýV$ˆ£ÀÖ/^}|f’FŸ§ê×·KkªáJ“¨óÒÈéH²õF àŒ-mgéܽÄJX…¸Ñk8Ð|–Ár¨–Q.ôÍ.woû¡²gÄÕ絤ÄgÍ1š‡K¬ØqÒá†*‚z›F™oï±ó2†0ù|uA†Û7½“½do¡_½UöQ ±«\ÊÐ(§žu¤S÷0åù@„¹xSðŽWm[6ƒ!gßÀ&f.(5͘ØÇT F‚ôr 1^2¦ võC8j²%ôÙ4<äµHTlsôÞ‹ÛEÌœ4Þ™aw¾·I¾nªûᕨFÛ*¿§_P¸ñtÖ>O~X S×x}¼y y™Au2À`kP½F©š(Î|MvÜôêfçig™&’ië OÏ·ÃW(¢H7Ûó l)-mŠÆ/‚?ÓßëàFª_ 4Í0Ozc„Ý ©ž4Çáj[}¯A§£ô?üñ.nø^wã|iÝ7›àdšìu9ѹÿFÂùƒ\*ßåºd ¡{”‡þ$½~:êÄ÷'JAáÔ\3—ykq_€$þÀ«n‰v`½=HÒŸ OÎvŒcàÙAóBKè‘»+ 3 +êXB~<ûÕùÂÅchèÜ$o·éž~ŒJÅ¡c‡V.l!w²­Óª3-”Z]ÆUª?Ì=°¿Bp +2u,W—3Ô‹,DÑt†ƒì#aõö²Ñš°¶#P`¹7o{ûU´å¹ìUîu»ºƒö#:²z+?5vÉ‚`\¿˜Ã ÏsÌÆê « p\ßgì±t¾=}M%±è:ó{•¼¶(’‡1„”pýà]*9‰©Ë⻤˜íÊo}»†C;Éêú¦Ce:}/#>~ïÂ,|A–DÚqVô.º£2¿YÏM›½K'ìoœ§>~Z8[-"-ئ—¤•å\‘•£>üæ©NkEJ'~ Í^$þwCrBs èKrKWƒÇpIð€×­!i›û8ÅUÃáœ+$~zð´òV5õ­Ìf£$ìÑ»!|Nf£{Óæyñ²ÄïÙ‘â&U¸é{OH=N8;E;‚ŽY-ó䄳-œûJœ¶[m>K߯ÞçÖ›IÉûëÔ–rœ_ÏFR<¨‰²Ô»jY]éO}¯ƒ°Wú X4P—Uws…gíiAcÑî3ËzëU¾Vˆ¾Wª1ù6®ëe¢¹ÎJk”ßÏÃËJ&TüGº"©”H˜ÅÖ<˜®úoThñ'»ZÚ°9ŸÙúòs•[§U<Ÿ«÷,“<+õå~”²c¥[:3rŸNouV®ÒÛkòh•^âbêW*(6¯e7%·ì]'Gq¨Î}ïÌ0íqXç­Ø?']ar,0¬Ö…¬ù"¯Ÿ›˜‘ ¥|×f(f™¿ê5,G•e7› 3Tu‚ê-kgºtƉ}ò“X P,ÐðÒ†ø-MóÈôs‡v,¦ß¬½øÞ ›ýÄÐáÿ)¤^\]ä}àLx¯ã‹öýÉY×u W±A¿¿¦NAN/8Þs½·¢‚ ”Ø¥ëþ]zÂd¯‹‹o;~Íš¸Å¨­gÉå~GÔmáÃNÞQ§If›9ÔÕ6¶­QCxÍÈ@@’Ë,£_šÕöhÙ>‘æÖUR L„þ;³h U&Rf½3ÙxNÜ«¦¦a;"™B2ùÚnp z²Éõð+ЮºZË~£ÙgÝ’>„¸ENöÌÂҎöì°ÂYªTÎ’~ÛÇ ;jw¸ÿýÃSb3lË2vÄ#ŽA&I<ð¤cc nóÞ·1‰gÁª8MDKÀ'µÒ£¾ÕXŠÜ‹z2ÑÛÙ:GÓ!¤šSÇÒk2ÛÇ|6\ïQƒ.DkÃýãô Ðõ1Š›xípRíÌÀOt˜™D/RŽb¹`ÍÛuX!ó ¢{\UyTËlêe NXá­&ÅCyfX"uJÁ!Àí‰^¼Fè‹2`r$!I~ ‰e*ˆ’—œ¦Væo]¨Ö#µù0P\;7„ž¥ŒãÄøpì»=©Ç½^»·¦ÐVo¼ìõš+Ai—On<-ZŸobÍäô¹»¨Z¬f·^Šó,—Í9|ð ÞÌȳۉ]ôÆf6>Ü¢4˜Â91CÚ®Ã5^>—ZSbvwŒ0‚Û¦ˆ—»Ý£Xt`;§?¡—S¾`µ=÷íôB1»FÈ;(gªÐ*f ({G× ­¥yßH×È€hgJúì+æO±,}õÚÛd¾.“ðÏníðÚ> endobj 214 0 obj << /Length1 1915 /Length2 14701 /Length3 0 /Length 15879 /Filter /FlateDecode >> stream xÚ÷pæÛÒÀ Ƕ'¶mÛNfâä‰m';Ol[Û¶™h&6o6ÎÙû¼ßWuo¥*ùÿº{µV÷ª ‰²ƒˆ©½1PÒÞÎ……‘™ ¦ .à`ffcdff…£ P·t±þGG¡ tr¶´·ãý—…˜ÐÈåC&näòa¨`ouµ°°X8yY¸x™™¬ÌÌ<ÿ1´w∹Yš²öv@g8 1{O'Ks —8ÿùP›ÐXxx¸èÿ<±:YšÙŒ\,€¶MŒljö&–@ÏÿqAÍoáââÀËÄäîîÎhdëÌhïd.HCp·t±¨Nn@SÀ%l—ÆGP·°tþK¡foæânä|l,M€vÎG\íLN€è5y€’Ðî/cù¿ è7ÀÂÈò_wŸþѥݟ‡LLìmŒì<-íÌf–6@€’¤<£‹‡ =ÀÈÎôC#gûóFnF–6FƦnQ}Tøw}Î&N–.ΌΖ6ÔÈô‡›6KØ™ŠÙÛÚí\œáþÈOÜÒ hòÑwO¦¿/×ÚÎÞÝÎû?dfigjöG¦®Lv–Ž®@ñ¿m>DpÿÈÌ.fffnVnÐô0±`ú#€º§ðO%Ëâ|½ìfe}-Í€à¼Ü€'W ¯÷¿ÿKp,,SK€1ÐÜÒîïb Ù_üqÿN–€ÏÌãÇ`þãç¿_zfjogãùùŸW̤!)¦&%A÷wÉÿUŠŠÚ{¼Ø8 ¬ÌVv×LJïÿúùoþSýŸRe#Ë¿³û—G;3{Ï_E|tï?…¸ý=Ô¯ à#(ÚÌ3@ýÏøë2s0›|übùÿ¼ùÿ7ûxùÿÿ›‘¤«ÍŸzê¿ þôF¶–6ž[|̳«ËÇn(ØlˆÝÿ5ÕþµÐ @SKWÛÿ«•q1úØ;ó9g`agdfÿKné,ié4U¶t1±øk–þs1l,í€ÊöΖ¼;§˜™ÿîcõL¬?Þç+ûSüجÿ+agboúÇ ²rpŒœœŒ<á>&àƒ8Þ,»j ôøsÄLŒvö.G5úÌìàþ¸XN“È¢¿ˆ À$öq˜Äÿ!“ĉ‹À$ûq˜äÿ¡/ ÿЇÅÿ73€Iùb0©þCl&µÿχ£è#ºÉ‰ã牽ÍGÓþ#ù£wL¦ÿBð_øÈìüHÞÌÌòþCû/dþHÄü_øanñ/ühÙ¿?ò´þ~¤fó/üÈÛöüØZ¦%ýñN0Ùÿ ?òpø~Äuú~Äuþ~ôÜå_ø‘†ë¿ð# ·áGîÿª÷# áGžâÿÌ–‰«“ÓÇóûç#ð1xÿá?ßz Ðh·ºdoÂlUÜñP+‚ïÎðcŠ•qè*òFë@×íP%PŒÿñš­ jE¤gágÔãÚ)ké{³µ6³µþ¶OZ/Öjc¤™ñ>½Ï šÀÓ©³¼+ëO¹îîO†lv¦æXç-ª#î7Âó4ØôŽ—?3¥ø)[ªï«,Ρ©¾ÿbD›Š‰‚ÆáæfC îÉ쬭™_è’!w¾žúH(yü˜68Îʱ7´VNj¹›†r‚è w“y"øí¨OÎp0OqÐ'óü•:ð¦#ø µ±ó„š_–ÌÁÌ•ôÕAŒŒªy‹ôðø`*¸ØÎWÒ‹X‘cª=}>ÚnñÒ¶ÈÕdõ"/6ªò¾Š )tIXøXÉÙ_ÁõŒ9¨r>òu£³ úêÎÉYóiEôãèߕծ½—YfM`Ç£‚œb-éE; ÷ýUƒ@²c ­–m 6§]ê4 ;ú‘¥á)H#/Ÿ‰NÉ5 |8§Ï&¤x,ó0S'ž™:¨˜ZGИNAêÎp‡ûMŸrí—X®‡´Æ£ ¤Â–àú"²ëŒÜäIŒÀÕžRJKpòÍ¿‡ùº\µï {ãR&—ÍèÌßfFÈÀù¬ç;£ÎZšÍ¿¶îŒ‡é½7ðƒ…ãx½Ôý²|ÏH±ÉºÐJ}!Ô¨›0‹„¯ðÎsŒÑ7¡®Löoçìó;ߊ”Uë®/hp»%Oj¸±l %¿¬WІËO'ÎÑ\U‰:Ù<‚a¹ª¨dã,w--Ÿ2ôE«wõ+¨ïÙ#ªÚ/Há¹íþÑ_G6© P mÆhŒv5?R±çߎ(E4Ð|ZöëàKÞ©q>ÚÚSNyÀH&MW±¶ëèB%wý7-8߀; .ó'jò©ãEm7¼"ÆA#b.íá90o?h…4¯Cé!ønõ]ää[«âd¡ìgÂÝt(}¸Ey‹ªêâg,_Bà,ô>e×!½ˆò÷dÕÛ‹_Šžü€…Ýý™Ã#13”HRþèĵm?E‹ªH æsõ0ñª;ta±Žg—kv|ñL—wÚM¹¥™ m îŸLk-ÙÎû4£Ò~’Ò‘Ss,"ù0&©ÞŸÅ®H}®Àô˜^!Pàq¦ü>Ì…ãHC+°üý†"ñ5Æ+¡@–MR„âTø«unìñÎõœ)SX…^¶ùõÀ—@Öñ»áb~ÝçŠíD»i-ê W¬mʹ-ÓØB3­Ú— ½ÔG=È®6Vâ[0ã«ã²lHM$°¦Y5¿¥Û²ê&Ä#Ê$Ã×c‹mg)Ò†}ŠCäø¡ÉTÒY‹³vþ÷¦XM..´j ²Üd2ü4hõééÈ<y1P"ó¬…6UÕW¶üu__þ«”øÂ×/iRe5ºL³vNzK ]è}ÓÐCÐ@Âæö›˜:ØÁ¬_=°‡‘Bpò^cu·òY­ì˽Ñ=¬´Úg‘D–—Ðx&0êó£ÃÌžä¼/è›+Úhò^žKÃo.Cõ¦*a±Ñ6 þLDÐjGâpVàkÙÞmã(MU˜  Æút‹”Ò S¿&¢4m“ÌZ¨)$K|‚BøÜŽ?žîÜClŸ Òö,îrõ|lt G0åŸÊ뻊ÒÉä°ªgö»J{ÕÆìª€áIrÊgЯ€a¹Â UJF®eb}ã¨a§ÿh/;aÒO7ÕE€ùl4ŠZ(ëÕÐ"Äçšu€P7£Ö¼nͤcMeNÃ1V~Ì«[hœ£“ä$ÕI’Üê à©[tê·5É:j4j2÷*¯Þ¶ÇþF.®Ç Ÿ_™£zø]âFù¦ñRË¢qde²ÓKOSf%¯Ûk½áè§ü”©˜D­»¯èÏÌÁ΢ãjæj6J‡k·v^×a¹oî7tvb ~nÝwK.ö Ù ¼©F ôˆžÄºÐ¸:8Ügòçõ½ÓÓ5X~ئã=èAñ}ÅvcBPâÔ—ñ‚=þdz¶ «^yŠ/á°hx…Hs×ôãä¬L¤¿A1®ÍüKr8óˆ)hT)ÆŒ7¹W­Êrƒ8¡ˆmÏ13šaÆ%TĈ'SäšÞÅ¿T]5éV ¨²ab>™ ’è¬OÅžò´+„«ÔCM%3i.š6ìVƒÂlöѽ‰n€aÖ`tr'ZðžZZÂfŠÿ0!ã¶&`|8³ù]Ó‡ìcナPÖÀTñ°‘ÎG•Fͪ¬Ç»kw?_·ðg£; ö²°Ã[¼IåÃ0ïè‚¹Ä Ü/I"¢>KDLOÖ³­¥"ç$? 0çs$Û·•¸U ÆÑ’˜¡@_¾Ó±‚MøíœOw;ÒðÉòµ¦Ï†ïȲ6Ìyê1²<ÂìƒP󣩘AtKOÐ:]x ⾘÷l IÒÙž8%±-aÕ—Ñb¸¸hËỪ‚ËçB.¡œŽ— Š.°ƒþ€ˆP¶EɰˆÚ¼u®2ækÙzTT-s n‡¡`–ÃÛ9fÜïˆýeÌ= žŸ&æI¯Œéú›‚‚®k(ƒmr`¥ õF?PNUfÝ«µ4gõÀìq·Þù ¤›ª,MÒÒ]Š’u̸Wœ_;ùvFzê¾Üma>ð ^P;•=ô pr—ÔEÏ­KƒíjÔ1¯µï+ÐzçS‚1ë£ÿE«­´U‹¯ãk{…2+–Ϧ÷¯¢Q¿ðñ ëzåÝ;vûÇbƒÓÃ0”_{?P1c`¯ÓŠÜ2âîcP®+F·×'¹48>·Ì#À÷ÎÒß¡lÇvŸÆyß_[XbPŒ›ÁÅ‹y^±}Ç% ÃÀ—º«˜ïM›ìòU7‚zdîÂ=-Õ„&¼8‰ j¦ŽWïÀËÈwD•,YMÎØÝÂ4A{—JÖËËÓ|¬ŽæOZÓäoÝ *¼ÊÓ=ºg=SîÅwíL×nÂV-Sï;òÉ¡E÷åîýŠs-PÁ¾’¦Ú|0­ñc‘gîQÃ?ϧ(ÚŒJMúsýO™òÏ.‰üåè032X·*ÌÌPUcƒQz !!HáìŽ0ä‘—í™"C©Ücáa*õiµÐó“ŒêK‹2Ã#ì¿_>Gî"%'!Ûœ±âÒ4RÑŸ0ƒèVCî»Ìß 2Âá_߃OºHêͨÌú…ÜÛ#\–3åò?->»ÜßRB΂åÒó“¼Òìé|£÷úö8è5Ó ý³¸†‚…$¹d8÷{mC¨©ƒëô ÉöÆHÕ‰¥t{ßéå¥]X,Œ^?«Ñåû>sqhÔ¤œ Ú¹÷b7²¹hÃ3éâvÉý¹¸ÙÉζ€OTB™Ê2ø>ïËe"u7ìçÈ~¿R)Ôg7"+ÿ´±Dë exs‚}¦ œ¨%ä뻯z D¹ß{“½NˆÔÏéÉ5ËfyP­&ÏÖ?m7ÝĽ£¾û£sìum7—¸/Ž«8NTÿ^M›“²6™ºÛ=ó.Ì‚õ)Qm{{•E?ÂÒÃp±8"B¦æÈóää>áæ¶?Af(Dª-‰ˆÈSSAqŠJMÊÝn_Ûo×´v°q3Þš.ãõ©é×í4 ð¾˜+m¤¤ô"ÆÌ¼­EÒzÒìT{²Ô‰9×, Û™O ÷M@踕º}9Ê ˜Éû è~ÓìØ£DeéKÞ5—7oȸ% àŸq˜¡Êu4Ð9·ç‰{å²:_B–¾}?a§ë=öïêŠøÁ+(bçïð¤Î´´ÚB©ý»0Z´™·®K/¬Âç^ÿv_¶€›sV[î5Ò1_²ƒ5}â\¼‡ß´N0ð¾¸5:x”6\4$ÇN¹c@}B†›ŽÚA“^•ÒÏd¢rÑz×z{9œr%à 5‡ÆáyQÜi¤\¨ñ]k&Ë¢…Ù€[âz 8 còyj+ÒfÍ X×Âê!qIý…o=2J¥®À/¬6SÝi­Ìm†[Ñ*(Tqt]2ø¡Â ÇÔ´{Ø-×­‹(¸MCžº°:¹¼ò»N^M‘£OR`‹&bˆ„l™è3º2™¸`^— e X½¤Ö‚~»Á‹—WšU™™¸®«!òÒ iÊ/l€=Ö¸Y}·«̧»5Åáåò¦`{žÒd9ôÖ)¾¼R§wêniÛ{Ô/?^’w·kR¬Ð<›k-˜ -S/q&0¨©3x>ѹa*ÄØ¾ ž†1‹HÔó7@C Ù]âøJÇ#13‡Æª>=Ñ2òS¬ñ"V©õshÃ"zõüâÕˆù£@ÿ…ðN«<û~7x×b‰˜~ ÷[HÏEÅ·½»b~ÏfSÓ/êÍayÔ¯ÎNE«–cÓÝØªM¨eŸ°Edíì厴É Rïr`J—€äþIÆ œê[ˆ¾ßÎÓôYÇ3PEEžëô}µýJà°’§7S„Ÿ°Ôõ ÝÿëÙ_„âøW »K4» j¯¶Dàn­›ý¯´~ÚùÜ©¡˜SUû£"îǘšËGÖÆ¾µËíúùN?I£õC’|(‘^GÙQˆšµz»{üØŽ‚¼ß¸å>‡C{˜·Zý2XÚ(ó\Ó~U·½ªçT#†W`•®ÒCùÒçòCÓ¦ïkõÎ{`¿}ÛÙ•Óµ›!ݱý'û>()_6/òþéµø¨ºt õ§-E¯wœ4ýÉ!HpsrÁ 'h”@‘ÓÑ/µfPË“ÔÏÛ H %ØO¹RBY¶à³•òÃgï‡Ë—P6ʶ5_P¿C¯—;ÛÑ{ú÷°/aM:—€Žñz(f4eaÉi±’=ÁÁÄ‚»º¸SY¤q… ¢×7îO-ï! ¶­…¢?ÐKnÙ]{ê™I¼êó)"yJk$@2m’‘Ò-¾»´@3–”õLgXºGxd'ázûŸ xŠÝl"ˆûDÍtáTÁ£4·§ñÕí's¡[÷&&d½Ëª±\u~;5âiòÚÔêq.Oš6¯…„Ñ–~BËÔ5]ÏJ°êò§aóØ °{¥²€taúË)•<ÕØªbènÊp8Xb*ùqΘ ë¬s$ÓáÍöñ¼îµ,7'266@Xs #½DÔŒ"„ÜÕÒ#9¦ˆŒö%n€‰ŸF´³+Š–»Â\õì{º Ïxé9T¶Ž”$‘Ŧ0 õM$ÌÖ,0mØl|¶Ù¤Ì<ÄÀC‹Põ‹Ór%)¸I¸¶Ó⦓áp×Ñ“vCߨïäÓ þF|´­ˆ‹Ó»‰ÖzÁ~•:½I» 5p+àý)ë—gœ=<Åä¦Àkí ›²½Ám4ÕÕ·Âѹ6°§”^e¯V²FHµKÊÖ(²ûaÃrïG¶'‚çË©qÇŒ™¶ü×8-T•ûBaäP Îjàúñjk VÃP²Ë‰–®ûÐY³­ˆâDG€üîÍ~+øë§EYùu¥ËâÂ{OÙJ•D‚ePpVr~íá *1X%¢Zˆ\¤Ì3â÷,»ÏP³¡¼=ñq\€óBëdst»ŸvŽD§©áp¢œç úV­ÇkÑó‚ÆBõ2• J!ìö[7{J²¥÷…w o‡1N© óâ༡ÃÜDÖ­o¬Ba8êigôO*Å:‡°2­ÏÇÁ ¢Dz]”NŸŒa±Æj•Ô†øü%yLFÈŠ+LLRqÛ–ÒºLå‡ ™‡RÆ»Q£Ò·,?-_Ô«aÈ– i¤Êá%ãVq÷½ùĨ€62&Ú¶TúÌ•0Îü¥4ʽ| ÛÔk;\ˆFU©ý÷I‘P`}y0Èð`á1kš¡œ_jtÙ, QÜãkã3É#f~ì0"¼Â¤ÈmÚ_ÍýÊp¤/’ñû>bW>ëz\¸3 >zéÄå]YÊVT& ¦êàÊ-€zg0e¨z˜ .¾åˆØœx¶žÝ†©wP[œnÆGÆÌ8Ø;Ïb A"/†ï/飒·R÷ßÅײ¼¼fr±:øâñy/ü9°gíMôñ¢;‰7¢îöš0­n ÷ß!†¬"ƒ.Ù´~ö¨bC< Þ¼Iä"‡˜!êÒx¶7©›a‚núß}¡¶È†ÉžŽp Å<j8ížãÏ+‚~±nzúJ†¡VB´Îï rN›ÚsaæÓkI–aÑ›5ñÛÅB.Àek¬žÒúJþ#=f¸¥Òmt_û!ž“{%(í(Ãà*),_Ó‘6õM°Ðx½/빉;54÷¼éXæËi‚ÞK.ø7Ô“­VÔ˜©óöJ{¯HÛy)²{ïGÄ•P§ßxû}ÆùP ?æøÒü]Å='9SqiܰDÒçJЀ7¦«‚î .Šß ‚‘äXz#_n¦/ M©ñBR÷ tGª¨ù=1ÄÂ_»HÑéz7Áƒ‚:™¼—ËMWþÈ^Kšo§ ÎÀ¦Þ¥'ý¬ÃáÏë&P¯ãˆï­ý4- ™L'WÈCa’q.·ìD‰Õè>ñ¦$;ô4vŠ:÷ŠEFã,-ÈB¿Ök“kV!S°µe™£ŸÀ0tC/߯¶#áwdü€¾ð¢ 8²ÑÇCd \¬’•™“¦åp³ÇqÕßñ™ÉfªåÆMRjÆy|×F>ÂîNâ¶íÑ9¹…qˆ* Coü„±¢†%D0 Ÿ!Hk”„,Ò,±îÅLÙ}k‚aÀ[êtGÂì¢-Ú:ûô3ÞQ8ö«æeÍr$v×à3]áã×`°­Çò)ïekYrEÞa‰Å€–ˆ>þ*”bæt“®6ä};(O"O²‘]¹ìÕaHÜÚÌ5s.ûî©ENQ’mc«"ÇÌ®F¹ãT‹»éYn©ϡچ¢Š%·“é}„4AÈqš¥¬ÞÍOº½?,e® •Ý—bL¨z=N×ÄH͛ىÒ_ ìÐf jLŒ=KkÉ]1ÄÔ7 L¨£€ôþÚÍ‘Vro™)XÚÉcQÓËT2Gì*92qîÉ1Y? Q> ÎüÁt¹ù)Ø  :[z´q¼üÕûÑôF]°"ÓJÙÉÌZLXY˜n«!Ï*áοJkÇ}TW}VÛ~­ô0w§‹{BXX{:n ý±ì(a‡ óÀ]m—P¥éåJn¯t‡v .2¤ñÆ’äÓ¯Œ´Ä(±ØbȪH»ùŠš^ÝJàAŠ]¢u0)Š3S½Û1‘‡8<­[y—Ú„áQö5XõqoTÑe„AŒigŒ€#!yx.e}µð\g~r,Õß<󯯇 t‰³ÇWû®Š£A¢5–€VM¨JWoÖac*|¨xȨsXbK²NEv-òíšb¨Z0 Tû°é\°• iÓ¦¼Ñ3´`íBPX5g¬F`¯¬„±™@hÒG3Ò<‡ ?(eVq }W>·å¢O÷ª=×W”\99B!ETûÓe<Þð˜Mê(nÖ)íg©‡7|›"²Ymçû !=¢h©œ¨ÒWr°1L²C *9§ÍíËØ¦.,á4B*iÒžÒ‰Œ;Ùƒ÷9õà†NÐÝZ° ž¬*}Ó¬Ú|Y«'uã97œœWxŠwëÍcŸ¬“üx‡/rhÚè8¨a`°ªÈVi}T[y÷WŒošöKW× Þ?e ¿¬OpäînæKJ=ÀøÜñžR¤O”üÂ2¼˜¼^%¸iaJ áæAKŒZ̘©mAÑlü`%«ÌI¸ -›¯á# UæÐHØŽèêÎ/‰ØnGW!¯Q†› ;:Zš†%1' ÌEÖžÒ–™ÐïÉ´’̾ÝïÓî@S¿ØFÓqµ«"Ï¥`†kþR/¹Eê]¹–º± ·ä¥ÍÂL,™]_œ©ƒˆ(Küjèw½m|¦C+5õŠÐ`~†£µîU ½·¬>$‡ë•çwýñ|)èošë”¦+ÄÎÔ]$צ5 ¾áꃒ:â÷:ܘýäxÙÁ¤×-ü|Èh)“o¯”Õ¬²ñ¦,ªv J1UoíÏQTƾžCXá‹¶ Az7 ¹ß¦ÌÈlÌØ<Æ”ô¶II{q½_FÃVi¯a„ž˜-a‡°p¹I"¬„ÎôYŸW<íƒÒ¬I}]•sÄeZ2îÔ–×Êa/zöCo’bU—jÛ)á6s+½¶©®+9{s,ˆàHYåË)eý4á]*„ ´Í`atïõ$¡s„ÜŠ»ƒš©àÈôw Šÿb±ÃD ê`_Ès-e‰œ¢‹ EÌ;ÞÔ€°hRnÚÝðîÃ,×Ê6~küå¿LÓšˆ¿Øf&’½¥Ë€~iœûʱTçÿ]¤{ÝI^lú=¯Ÿ±Áж鿆0# Q¬+îâ•Sëâf¢‡{ §I\+d•ÙB®ÇâÎPiç›õŠu)°}ãáê ¯ý¦²»xSœä3áòI5‘¦"vF¢—Õye¦´pevê+¥1úP'G¢ùöF=±'„$¿Gº¨—ÊýWîryÊàIyÄËzmtaØ-Mºx'ÿ%>†_ .ë?â^Xª"æª&À¥+佟öq¼¥A%SU&˜ç=ÕT€ÌÏ´„ö•®¾¿’û¥lMñ}ðñÚ4³l²`c¥yêXGÌ¥Þ-£ÍtÛ=A ëu#º™t¬^âG¾l$=t 9.\С a9ÍìÔh |—Û¼Œ‹˜}´íejÃðÃhój.‘ÅÈ;ɡ¬x(`ºÎn¢/b¾á¶²©Nw«ÔñŽƒÇXVëÝ •¡¹$B+22p"PUR`—ÈãâeÏ1h‹‚à«°Þ£·˜£Œ=p’ƒ4½æ5nmkyŽœn³ËÚ—ÉddC:>Áfåø ~ÕibM²>·Ž1Z¶0ךtq/Ãv» ç”ÛÌ/QÈÅÑ}Ÿ 7ð»ÒQñF ‹ý+\ζYÙq$ª¢‰ÂmÑ0Þ2›rór)9ö>o*¶ ¶õ[‰TM´ôALaý‚åá3?’O%U#=EGî¾"è7^Š4)Ó$ÁÿÔþ!BwÙHªGظ ^¬¿ZuVƒ§Ú¨«9;{n‡â—V3ó]—iµñÊÆ|¯§÷7Ž•^.fváÍB2¼ºÜ/ÙY¾.šÍ’ôÒ®Ÿ'/å^Öéß³ø÷nÎß“ôD~NMnX›<$V"®ÐjO÷ô-°U¨¥›za÷‘¿Â¬Z &lV"'ËÂd¤LT|ýØ>¯ýìd#žpó\bdïÊ.ï[bÑÇjôÚЄiO „¡“ ikv.p¾bk­ú¹DG¨t Ä=˜sÈàÃÏYyë! EÞ#KÇ”]Í4H®ËÁãªÇTM«™!£TÝ­Àâž[T_“ÞÇÁÒu:>„ê^Çxv:?†×³ºôÒ}x Òf•IßAy8H°¢jªj}añKûn26S[rÍf”äÛh¸5º¯·½Ñ tœ„ùÔS‚£4ˆùÞÊùŽNB£\êm¡ol˜ëþÒ>kBEL†èìÞÕukóõæg‘ºç³cf€ÕL©ç‰k ,úñÍzªÉ”aÕžÎ0qgü§à¤ à¦ù{m]ÊCÙOìð–âa“šðA¦APS5 ¿Ó5ñ#…“81t ’;h׺ÖG^ON ü]IRœ„½@¬™«ô MÌ&ØÆ†¹Ìþ¯Ÿ¿Ù>5 môÆ_Ð*~ɶ* ã­zÑØÎâ™õ tÅ @o¢´*Ýí E,…¶<é]U ä)z’Vð¬Ÿ9RX®ÙÍ3…“óDº…²o);¾öä „=®Rа°ºênÍA’À+DjÁug6ñY¦[-Àùä$¬¹Î¿#ÉL8L |ºÒ™×•y.¢½·ÖÖë^7ÑC£–vO/ƒ!‹ÀŸù¼œ¥DÔcþKÜÐzŠª‰`”Žˆçñ³εözù§xÔ˜rÕ{5(õÝßÃñ|ùæG®9~€7¤£‚\Gº­ë¨Àì¥jž´-®½]¹[ËXKþ_þuÈs•~9¿hFAFa!‰)ÎWX!€t³=ë·5½€A0÷p÷oÖLÄI_A¢­5ã)._“§K­ÖxѽjTçÜE<.6¨Y€|û)>ãôË Tk‘ŠZ˜W6VE~Èkg?æ©È‡–̆¯F°µ¹G%ã\/mà"cm•„¸Ýì<]Â݇ËæöÈðù(ÞÃâF[¸Kµ¨»2âÚããðŸÔ»qáã3×1uZéÁ‰&£ÌÓ19à¿ ¹òAZ+u5x~»|C,ìPäôÍ›Ó÷lÒIs`4ŸÑ’5•¾˜ÿÖ#¼ª7Ý!xÉ£NRLuæÑ†bÚãŒÔ £Q(çámRÔVËÇd*6C~“‚¥¤öp³a#ûÐöuŒå¤µŒ‡ÀŠýgÎw;!õä}îÐk}. #ÁA¢Ö¥15U=²ú戆¡eék”Üs»l‚¢§C|™µ =3…Ë-ªQ¡bÝ[ÎWj—œžyM{¦­~yõ½ ã8àZÂì›M±ì!ŒU¹€áŸ}zÊhÕIçñ¾Yu©ÄS·6õOX¼«ƒ}­ ¤Ç I–ÆÂ=p¾àˆ=ÝLÎø0þ?H¨]jš(â¹k“Xš¢æöÁKª»«öŸbR)¶c‹§‘P‚» `] Ù?õ:І}¾¢ o„ÃFa¢+j;m)øè0°‡  €¯êµì8³Vßw8«Óýd ]ßÉofÏÖÞÅ—‡)ÉBn·øœþT|í0þËöt’õ;j—² má²D¨~R³M “]nZ×jù¢»Æ±unÔ­ùF59(˜œ!#Ë^¿Ü/øa?YÁ½™¼1iV¤§QJÑs«áØ[Há@v¾ÁlKx,I ®í17]îr.¹ÿ{ƒÑÈ'ãßfÃÆÙŒµ6™²#§£•ƒ®B)ÜJl}„ƒ?;½à ÚÚ2лA¡¨ž ‰u”2êE—Ü.¥¡‘ªÓM½r°¤lý\LË|ó;ð vZÌñ-ÉP)Q)ÕønºžazG%¶M@Ç ¯Ž~ó“•ŽˆÏ¬ëô¡« áÔrtžóô úef¯?£€y¥¨<®ž·§5²¼ý+€ô‚øÅùK`ÂU„ âò äSÀpÈ®¼DÇ›¶ÌˆØò Šo’}ÄÔpà•o,pcîoöÞcX‘Ôí;†cð7/¥Óš­*áêÎ)°kˆnÒ¥oŒñpص…_k3—‡1÷䎺Žï-ºAÜhbTËÓéŸÖ©SÃssÖ1Lƒ°ª› …µ{+è· êôÆÌx†sÏ>[µžï«¹@Wš!Ái3{:äTäÄzœŽFmd¢]¹ŠàŽƒ‰h!ìöŒúÖ;¾÷Q:ɼäá+$3iËÑ|C ¹~ ú…£DŸéºõÛß3߳ދÎK »UX žÌ7jkD0è´8¦ë¦¸ã¢*N¬£0Ûªp\aß60I5`×ô¡üÆ·€½‰½&8¦7 Çኬû…\¥LÛG’žà]ò«ål¦L":²—G{€„ÂqÝqOÜu4ˆ†$®ÝÕðFÀªÉ-] iÕY®¯VYfœJˆ–è—^ŽzŽLË]byÕÝM±\ÅÄ/cÞG¡¸‘‚õG¦Ï \­ûªtlŽ1ŸMGÊF¸„w á2•ò@çtÂñkA)Â);| E~ãëI8´YR2–Z†h…Qœ›yj h­†”R&xžmGxYŸ%>ßéÉÞ ¦u7 Íôoe„æä#ûrŸ^ÙŸ{pÞüe‹tüÌ9hVHˆ b8êu§ ÓÛ|l.Ƭ3|­˜ÿ[Jµén榟ߚþ©êp:ûo‹+,cFDå°Ê§Òyý„;Å/†_Úè[Ú°{È8a¶Ü¦FJ°ÂÎ 5{ÂúE£eøGÂCì1Ú ÆÌg÷0OTÉ/_¨¯ ‚úk˜Jmý0ÌSÊÓºËÜBÌ´P\µ e~:·BN=òk¾)WÝ$3X”àÂj«†!£åR§<ù<»gaj¾#É9OÞSi”°É;]Ý£Žœ~e¬¨t߃ǯ輎âp]d ÑüÍúH; cßFoµŽáNê óÛy¸<)Uùb9s(n„M8;=«Ý×´Wœ|Xe½=œ7ÒLÎxšoäÏ~£,}‹³§ûx9¯|Âvæfì'ÛJ‰Ì=—‹”9Šë×ísGÕ“¯ö-œßìýù“@¼@½Ò϶OQ·*éQ&JŠ!V!™}bdôºÈà’³\KÇ–Íö®g]|îuB–XÅmuÁÅÈ\¡ l!^µ+E…{•&UPj³ozˆÿ(ÛD°%jŠ/º& E•%Î^ØË5ý„Ï鯅ïúMX-æ%Ç,‚&ùö*šzj^8÷«&‰KD¼Ñ÷AE¾ "”¦ Kæ^ܼ7Ÿn›òÔ‡…/P|¿ZÔóhñJ¼–>u\˜2ã¦Ãlí•óÙ`hJ§é–n]$u¨C¨¾²~ª…AëÁK¦5_Ûͪú–±/SAûcZhÔÃÐllʈ”=.‘:0Râ“]ldhÔòí‹j.ü†qûD…º+xì·QOF¶û÷‹©ÏnjZ’7%5ÝŸ¾§×´4?£Jüzì‹’íñ¶bA¦Î°l›‹àÛ++ÒžÊ2£æ9ánhZQ(rHB´ö§˜i`•ˆ­S:EëHM·ÍÀ-MþÂ' ¬p#„«ˆ_G'R‰Íf$möƺ^^¦ÊÛ¬Šf ZC“»—²kb‹óÑý¹¾÷°gÝ €2¾ôÉJÓÍÚØÅà49¹B¸å¦N7V A­-+ä ó\mB­sšO¢#Ô•ÃäKãš«–9Iý©!,N­\_P¤éWsÃÈ/¦2®9 œÜaÕ’ªÇyšaÝŸ Âûá¡ÙÉS)~ñ“…e¿uº4¤BÿH,K¿h5†í÷’5Õ. ÎÄhwŽfͽ>së í¢DÀ—–¦‰.·qècÊl>Ñc9Þ>á=VúÓ¦ðË’ØÏGç->U;·gíãð†»O^ö8¶‡[† °ó(zµ˜¤ƒd çÊi#ªGðêÈÔ!òØÄØ7}¿®SuWðöX,:w©M£h§Yl ß´lvO™ÃJŽ,ÂàVZƒŽçpΪåʸ Õ¸#~ÂHz®Y)¯Šf FU¿a£¸˜µxOOJÂåi¹w[«ŸBÙus§ÄLù½6”þd §™âÿuù …÷K Y-¢()•™ÂW R#qJçh{U®/„ÂNœ\èc|£`Ƹôª……?ÓIÈ»”Ü «7¤ºrD 9æÜ:†To¼âÆã̱égµ~¸®Œ»ÚÆ¡³¯ŸÑ¾Ù7Ç‘êÄG{úío0‰àÙHdïssÖg{P#9ouV6¾ïŸòNú­¯ÍåXÞd Í‹5²Ô¢ Ó‚1Ø]fªz.Äg£Ž*ûŠº{pbåjµˆ²ÙY#Ù%Ž\!M–ß›8"Æ6‹ÌC.òí;d!y•+j0ŸT¹Dg5L`ù¥O›Ç¡¼ æ2Y!»ÌÜ:’ ”š`lÀ1ä{”îê0h¸ùv׌jƒu¶1—Ñœ<|_©:ªÞB¯œ“Ü<å÷f1vÄ´Û® »@@^£™Z}e­hÝLð·9Ô ÂÀužþ‚jÍ­Í߯ tÐ nĪòÔheè"ðD2óÃØ:½Î¤ÂQVTKwŒ+—Á{O(”ûŠÂƒ  ‘!3ßôüEgšÁg09+f™( àž‡£:~7UæÇü^r7iòƒÀ¹Ó‡™8Ú¤ó-RØqŠGa÷”¶lœ¬Õ‰D¡§šê†¿hÍó.h1vyr¨O"Ök7¼ŸÈ$”D*¥äál˜©ÖD ?ƒ¸K¤=‰S›KZörîi«Â%$Z†&û,ȃüÀÕ0PÍðá@üeâë†5ãMßÅàŨ§”?‚u¶V}r´æ5”ÅJ ÜÍ|Š#Ô)K–$[¨%ÊWFÿ¹¦Å>4½á,œ ±mÉþã„öÂÓ£‹êä¼Ê4»n¨z„_Ç;®4×rÇ23ò©™9À:ƒ¢<<Î}À.YR"æUÞ q9‰ñ± ü\”cÚ9”7¤({ù¶C‹¾TGX ¢‘‘!¯È/eãëò5Ýjb:‘‘¸ñPÂðÅjÎÒþîKÛD†(OÜ«›Å¥/™¥’èýÎÚ7´\P˜ÞáXÅž]RäGµåNUDCÝE äp2lÃ`¢¦a‹»ô´'I²U†WBTŸŸ Ä'˜#ž‹e¨JóÒ£íû8‘ósLâõg•?Ëæú9jÇ3iîXypÈ£UJWbZ‚‚À,™µì7§y×Ú«¿ÉÚs¾nôà` x€f»…å¬g%½éüêàçŽÆÆ d8‚n&bã÷K!Áû~·NòpÛ©ž Õj8fPÇ} ¼Ø¸‡Âç7 `†œ ûÜ -µX}B$¿zn“èíÊDsî1Nú]Œþ3¼R¿ržѶrƒq¬N´œG¤ži÷ˆ×œWïÀr˜b-û7+—žÚeña_Þ›"-|íä:8wí:›®ª2ì|Öê];&ôý1I át;oïf—´¹+¾ŒŠyF½ ÉÃ̹óY­CÎy„†µ<Ì+®•çAû&—¾NCj»üO%RÅÚÜ1††+°jóðö~3‡"ß÷8ñ–VzÁ@ %Ä&ö™z´}û¤Í‰#'rkº~†Ñ^â±—$é#ƒr+{ߊ+ÉNa‹ Çžá9¥´2k)›7YKÃÚ뚬¥–K·ærœ†DWç]'JŸ£LIf‡CgŠû¦jRúö4ÒÝÖH;×Àô¿‰`ËG þ†·š¦Ú¡áiÂ!›~k\”¼ºÏ ao¿w`ò+”·G°uóT}£’6D‹°Û"Šsú–&dòyMÔÍy<;Ð|H®Xrl±¿Hã¸ã'}ßóÛ\ª§âP«,5Úa,î/AÉþ[A)%}λÁf`¡¼='ÉÆëø²ÓÔt“DÑàú»r.J0Lùô"ÍYÙärêÆôDHƒ#AÞzJ vó$¬ß<ŒñVÍnr QÌ哤÷»›’r [¼i¶ ˆjÏRŽôNŸ)X/t®N¡{b¨£T â~¶Xâ…½c \x]ùÅS ^®¥–­ó%õ(úÉäÔ®7™‰ v6‹}m YoyˆbŸB‰ºó'‚‰ð¬h\¨ ñGà,={­¢ÑÚ vñš¨_dïŽA›]KdÙÛ#lúÒ˜ã1Ø–Ö­–¤Ž€ 00*œí,‰¿<ò+VÜ[Ë6¸4éiboØ““ù1uòs…~–Þo¢$(gòUwé•D»y*ùHöxY‘¸…ÐìÛô8XPè‘ð`eD‰®ZuT¤ˆ $aÝ6匡LŠ“Œu[+ÿ Ñ^K)ûÖ±þëb˜4ëò9Ëм ¬Âª,÷1À–ÉD §®:»wÛþÑDoeèUY<®r:DA\bµq£.S§Ï&àôùº|DóaK0øó·í5£f€-¡š ¨ë‘ —f%ŽU㶪´iXVLv1I¾_ì¶r³ž¾zSÖV²nt‘´Vš ¬ÌÐ$å’#*²Ü :ŠÕ^k©k%Ú%Ãw·«UƒÂúF:·Þs3£>´c±ÇvçÍ endstream endobj 215 0 obj << /Type /FontDescriptor /FontName /UFCSGE+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/A/C/D/E/J/L/M/N/P/R/S/a/c/colon/d/e/f/ffi/fi/g/h/i/k/l/m/n/o/p/r/s/t/u/v/w/x/y) /FontFile 214 0 R >> endobj 216 0 obj << /Length1 1515 /Length2 7537 /Length3 0 /Length 8542 /Filter /FlateDecode >> stream xÚuT”m»5 ]ÒHô3t#ÝÝ 1 tw+(Ò¢„€ t(!-ÝÝ(- }ë{¿ÿ_ëœ5kÍ<ûÊ{ß×¾žaz¬©Ã)e ·†ÈÃaœ`.@FMW Ìx¹@ l&&]¨‡3ä/;6“>Ä…ÃDþ#BÆbåqo“µò¸TƒÃÊžÎ0/, < ð_pw€¬•Ô ÆP†Ã l&¸«¯;ÔÞÁã¾Ï_V ,,,Èñ; åq‡ÚXÁjV—ûŽ6Vθ âáû¯¬b®"ÜÜÞÞÞ\V..¸»½à õphCw/ˆ-àe€º• äOj\ØL](â‡ÜÎÃÛʸ78Cm 0Ä}Š'Ìâ¸ïÐQRh¸B`«þÀøór`.ðßåþÌþU ûlecwqµ‚ùBaö;¨3 !¯ÊåáãÁ°‚Ùþ ´rFÀï󭼬 ÎVÖ÷¿n—ÒXÝ3ü“ÂÆêêàB@qäþUæþšå`¶2pÌýë|²PwˆÍý½ûrÿ9\'Üæÿ²ƒÂlí~ѰõtåÖƒAݼ®£ØLƒ59 ¯M­P±‹cÞ¸)©¶±¸‹%¢]žÒ»z'C2“ùâÓ¬8²éÚ&4;ëUާÃDâ—»UOž •ÖVcK^˜­=ÙþGíï“'£@r·Ãõñìm1ægåû,u{\¸(ƒ q„„xqQ›¥²›+¤ßK޵(1"Žž ~O£‘ß¾ÈúÒÇÿÒ5SÆ ]΢ÚÙÁsU9ÉÞ‘ÈÜê T±Œ2­uÈ=ºJf =i aµFôëˆ}ÔÖNâRM ²mÜñ\SJuUäbß|~˜zn6k%³2˜×¬äw=» bñM«þV`¬öxçø¥ýÒ[•‰*ô¬ìúÝÄ6GG1›âÌÇpu²—ÍágÜêŸ$'œ{Æ’DÚ¥î<ç÷ë]¶Y±ž¯Æìµ~ ±) û)û]GˆU)r(Ôjò*"P`×ú­äóãÙÁ›q/LDž¡·$Ì¢é:sä­?”Np+’_Z‹%tÊ”Á(T*×ù&é‰@Ízû,ã"dËap{F« VW±yOCŸ/vÎÔåÉã×S…ÊŒ;Rtzm YÑUä­z˜¯GÒ‚ˆ>УÓU‰<»Â°¡Òz§Fí&¢öMcXÿ”“2æåãøn)AOtÖîÆr¥ð…§J(1µÆ’öh’Ê:Ås0s „é—:¾ACçÀtçÍÒ éù×Ü4’‡>(© JB|ŠÛî[(¯ç_ÎùÈÇp!!'ëHÞÚCV êÏV­w†YÔY²„ÌáÁ ”ßñSrÃtçà éÏ,P§(kq<«xÉtJMJsÖ+y+Ìâý›_5cÄ=jJx}p W¹‡Ö»FJgðãý…S-ƒ’Öi¤pδ-Õ›95>K®¥½F¸ÊiÊ äÈDWEhu}?Y7w1õÔHƒ­Td']ëJˆ41ÄŸ2ØMoà·©®é‹ÝP(3À€Æ;§\mHŒze¡¬¥†EâQ”Ÿ8ÅÆTÍ®â¹á5ÉUw¨¯!å(YÇ mb¡ùS¶×ÂAœ£OÚE2Tž¯Y—$‡ç”žï½ÿ ‹n5œ fÊåKe¦íclúÓæÎ§#úB>-UHí`æ|óÎ~Žö ûà!Z°E‰éÁ˜d§Ee}ÊÛ̼àfg÷ßÉ^93dj°(Ø,•â9ñHd½šÃºyÞ×¢6Ê\ü¹¬Y¤†Í+”ZIJv*›Õ'áv ø(ë'‘ÛÔÚjmMX3A®Iëû o"”õÐ ®$« Boì„Iä!‚ÔLŒqXsM„(ÃßC{b`Ö8"ÏVÞÊîž U¨áÈÉ®ó!›ƒ®ùù0 ¨øêÝŠKV‹”h»ÃÍÊjEsMâÚ3ëŠæ7”\p«ý §ÇL¨›ƒ¾:»ˆ±)~Aº˜ ·«$U¶ñ¼³Ä{ùNˆ{s$$ûnôQh¯¹³kWéBìj#¹4Íú€ E6–5 L¼î9YJaâÝAP,@XÜ{%î“hþPŽu“kRä «üZ|Ÿæs\ e§Íªu& ÅR…#5:<¦U¾ÈÅÚ%JìÙx-Ç£ÙQp§xQÅ‹•è°¯%ŠÆÅI_Ðû_R{kô„Ê¢f­N‡ fɤ=Š3Û©³ç°y$5ÝA _’Lúñ-§±Å±ýBR4¤Véyž”BiòŠ‘f«Š ioÉôݧÎ-é#„ÅZðˆp«TT¥,’ÜâÞÁOõÜ÷l¼P)&ÐPü78¨8o?urÁ}p—;z˜‘ud‘èßòûGÀ(»â~–Æ0‰ús‰‹ÄLW˜I¥%…o_Ö—ëAqÏÅ£~Þq‰„¬Lx{¡íß à ˆ…ÄÂ.VI%õG‡Ùƒwz£ãŸ¿›J9©ÒÆYuG© qA¸CŸ%ñõ# Óì“•Wц(ë¯î2ãÓ… Þ)VÇËúyu¥†xmÄDè–¦×ô/»‚H±«é…#óÆQQ ¤ËüŒ=Ã%>˜u‰g8ÌÖ…~xEQÉÐ:7ù!µo_½ô¤[|PˆÂÞéÓ‘Žù–ñ`è€-a^µOêf0f¢ñ1íÁ.N{¾¦ô(7ÌþE­âF0‹<*Óc1¯æ`—>xõ‡šùÁ\Þ¥é\‹bg³iëmÒÚ@¬Š’GkÁU:t pÊ$[Ús/û´àÂë€v즉¹Öù×è–ý_wò&Ã÷°jñ7nwcžt4榋=×k­ºEq3% Õ‰©.‡¢„À†(íô¡‘Ï{oq Ë“ÃëL(i?|ózS0É}™®§ÄÁKŠû2ÚùcÈXœ¿Ó¶;klp¢ÔÁ¥v’PaË »Áâ3ßQž2t›õ?ýRê²[¢/éöþrNíÔÑ¡™˜‰8N<¾#UZøvü7ØïÉíÞ,E¤vݧb©Õò‰µf†Üô²~¢”¼ûË{„q|MäiU1ctH)Z§Xøòm™x—ôn†I»™–ÞÕ^!N³¨ìq8ŒOUnÈ —‘(‡&ÍìáÚÞ)ÈJ½P×Ò&C> ìŸR»ÀÏÁïSˆÖ‘祙]ªõõN«gÅ•K^Ä òsòÀŒGçÏïT\µúm>÷ò8‚îë»» G‘7ó9šL¸’2Ãý?I3©±„çƒÚ¥ôO\ýr—{ˆ:•‚™Œ m*”±wËÝi|ºˆ\Ù©ç_»ÛÕÚxS®>/o“.“Ä”Î/Egfö)ž‚™|›çúír·–¸Héá<³ÒXÿ‘Ï äòÓž é”tu{%â-ÿjXúØYVGB H†¸ò”åųCL¢È¶gŦ½òo'òÑ[n×ü»Lf´ß.ãÍò¡™EЊW„EbÌ·ä, 2ûr6J”+ŸÛÁDšyî4Sbèˆs¤ßޕޔö ñšžénò”?SPå$"sîW ΀²"F êà൯ª~ëݯž®è°ûX£…®"(Ù ¼y"ý“Hòâ äÕñ›»bôŠ ê"1\°kîN+IdX(öHÌ®çz™G-³b£ BqºõSqŽŽûÈÞ¹0/Ån×K¯Æ?É z9À{ñul~²U:‹Ë«’f®ý€ÞaÚËU£÷E۳φP39¡Ò×—±é§lì»¶-ËÁ!†äÚá}ÔJxƘG¡ñ%_yååKÔ·n]Cd€`¶‹=mý’jlGèŸ5J4~¼̢ƒ^ꔺ1„½ß ®´S¸n“dvQú)›Ö£:‰—ä,•êˆWÝö„ã%#Eú*]ëÈXDiŽíIÈH_:åÎg"¦2ó2<?rŽîtÿPI³c< ®kÌùYધÀÌ„ö£ÆÈûÙûÊœ§økxß᪚Lß©/È7«™Aä"îÍ•7 )¿ájãuº+$£ Æb"½´e\ùŒ›:‡â¥:ŽòSîrH+~C î¡ÒgðîcõøÝ¨x¾æ–GSl¼«jéèë,´NŸ¼B{éð!WBÜä'.x8$çÍ㳡¢¢©k(žÛëå N’ ê&-7ÞFIúî¾J„æi¯âÃÙEHn’µùåixàÄ[`Ö³\cÈìÅ3Ž¾Ô “$l_¶J4É¡ùšT¹o9¸-zÕ7ëlÓ^ËÙÐÇeoû¿iXFñäER†í쮥hSòǨì-Œ½¡»Ã l¦i‹ s“.HÌFŠŸ,¯-j§d•¡e¾ÄÖ¨ûzz3´—²Áu¥o˜öDÃfbÖ2¯ ”9¥ËB½œÒã:'N#É^û–ÂyhŠØr°÷úoé¹Ì„°dÂ^Ëc‚œ—8>W2?z§›û GS¥àÀ-+SU¯EÙYÀCÄ¡Á¬­)VJâ•ö'“‘Pr¡¾Ö¦9uÿýT1­¼bÌæ)¤A|lÖOíU@É Áçxï5kÙqÂÙÃ9"½5üõ7âSXGpgƒÑ§¨Vd0H’>Ë·£Gc)_RN À±Ìê,)d ‚®ˆF{¥k¨{¶øˆZ Ó† “÷D\ ~¹åzû“×^ ©ÝÓÑÓ2«‰$)v3ðT¯®u2ínsͲÔðnió~é®Ì¯×¼PKœz‚×moów=oÜ¿ÝLT)°¼®O;ÔÊff¸.o–ýìÙŠ©&¹Ò®5Xu>Ƥª{nS†~í#vl,© ½ÞŠqj¾Ð0Äù 5~h¤¸†¹­pÉùj:ªá¶†pñMа¹è léWñùØÝõHâ"ÝN¿wE NntȽsîýcê-Cwýþè¦÷ÞY°hbqâA75é«8†€–³‚Ÿ'8ËuÕ¤5×fŸqÏ­,As æK\o âBxo<ö=€ñÈ~Yó[ÌàÁKëç´pöñä[ùø$„çèê“S¾Ù»ruM ¦áfÂÏ*ÊÅù‚Ö Æ € ">¸ƒ<žpšŒ™!K´k0ßSÊž¾©#Ì2+iAÁؾjBØòÙNý®PMpO87g+jŽÝ?O}õàˆìY…¨‡-^ ›}ü¾ßà $G+5Ò[•÷@tiU†ܸþ賎Һ‰4¼ÕÂz‰Ù»å xÓ‹´­?Ò Y r=“Àu1è1ë)dÙ9T„ºcK>=¹ÕÙMÈÃ|§ÖcÉ¢fXß³E¦ÎW5\è…z{ÿËBüMUXóRfïór,:°XØŸJT‰HY0v:ö^ž-Y{K\!Dž ¼hOÕ™N?a–¡)a_˜Oã(P>Ôé¬ñçypx«H09c€ï× äj°Åå)HŒ1ì*ÜÀ£•‘5_Vœ¢¹JTýŒ§¹€æ!j“¹Ôà®·ûÀ@¡zþÁ‹±S*¹øÞ…$ß±ó©Éç4+³GönE.z}ÅŸÏÙpí¸TUЩPÌðxã$"Ôàz§shˆZB< åUAÁ®Ûk”ùÓ!BŽâãô TdƳ±Ë¹¾$KÛD8ì'x‘¡‘5”iIECÏš[’ÝüÆw<QeS2äñö=nµð“öRkíJ;?dmúÊ—¶.± æ2ïUîvÙ~¦´ÐEÕ&Ëv-hP:»ÆâW¢P8s„x3uèv·Žij(èåRèòàPÝ)÷!R•âåé j¿È¥®éI½< 5!l¹XèÁÅx×Rôn+~<þãè°î3*åpl uã¼hlì+¡wYô[üÈ0–ñìô°²ícÔ唥٭îkëxKñî±DZìÞ‹™£g*@_»ûâ‡Ùâ3M©‡¡ó³6,M:tz䜨)OŸ+Ô´ õ¤=åmlS£Q)OpjW ¤gQ¶¹C̓GüédÅæE]Ckë":ýn¥zR|JªDw¶ž–<ý (-åÜñýø‰ ôyÉ+Zß88’ôk»®•O¦;ÉBäÏtæð¦¶g—÷%ÅÇüT_ø`kÈQëOÅ÷$ŒYï ×›n0©3sŽf³ ÎìÄ;-—I–ë(›sÔÐJŠe&O¥;9±ÀPCÕñcÒÜ‚Ðq=/l®mÎ'q4‡Õr¶n fÈú¯Ÿ;h—sÙÙοO„Û»óúP8ÏÆ_Ç]Ö'!¾Ô-Í1Xùn#g\–A¿Oña¾wñV• -g6Bîã{²NìÜ÷fû¢q;÷­õ ‡x{Á\‹ç×=â¥é ³ÂX¯8Â@1[»"ElF%³–9Ì1u&„y7¾—ч.Åw=ô(tÐî¦L±bA—B’¶níëˆÔhJG`Ñ'§Ÿ #~c<=b_—=QíïãßJ{*,:ÛDÉñUw;ǺüÄP¤ž¯=ñÆ=³6žFƒài5ÁËFÓŸ¬Ã8Jð÷ãÎkÌ…^¹.L•‡ÇAÛþº¶(««n LÝ@²k½Oјä§"t ç•Á~’馬®Ä>±*†áO&3¯ÞÍD/ö›«tû+˜ìŸ¾0'b³í¯Ã gœ¯µP¾bsq=mÍíSòó3d½c~ª;ÚgšYÉøÆòMÈãÀÓI-ObHcÆD<—ÂÌ„òºÏ¢¤ŽíÆ\P·IH›O©ÈꟴEÖdÔ}'zË¥`$¡šæ$OÄÎ&ªßÒ °[Ùµ´—s²Ô%vbÚŒ¯æ­fKÍŽYz›”¢Lj¦d:Z/8Jª}m%?ÑÞx“ã˜Oý.5­™ÙÙ×/3¤ßšˆÙ{SrÜ,ºb¥ÅuÁóúTjÁÈcþ¢8½kký}­‘²ÕGòÔOÌé"mkË“¨´[ˆ4*OºR»›X>RL‚C}ï“™bPš©mÓW—ký*qåŸû.,*“#¥'i&×{×Äýh‡|wÔ%~uÎÙyój±†-@cqÓO…l$;{ªÊ+Éõfò²ë‹Ù5Îrï“»{çY±7¹xWzìÂìŠO~¸zQĨjc?øCFeøúÈøðWÐdRuè“ïP´ú†èêGÈ„B•>­ˆpbEâmÉ#šAáþ¥|7pœ"Þ²MèWì=ÂÎ%ú×6¾1çÌØ'¹ÆL£ŠÃYÅP åêËpèZYÅ€!ö÷“¨Îl{õƒ$ђƯt=8V¬š oqøò¡`=͹Rûµ­‡úx}],[Má;ÜuSéÄ»|¨~}GZ2ö“,ðàÚÚîöì¹³´;‚ i>¸ÌiöÑU!h€¯›©Vf}éA¾Ÿ°nøCÐøc,v‘A5Á —)ü¬Z²ñ€3ZƒÙ¤6iÝ4p’ƒâÒ[®v –½[A™éฟÓ<¹øVÆ'nšÆóž˜0Gz¯95_ögÈ4º›«¹‡À9t ±ø'+}îä…ÉÄ[ÄÀí 7f¬Ò°ã!…ÚŽ‹¡*ÛYO®Äw˜¢ô×õdBAÃ0›LñÉÅ-[„âä²6’»}Ÿ ÃoÆ85ØOçèž´SÛò-h±P™ØÙ¨…kíÇH.Pç?v¿Ã8iß.ÓªÚÄê¯ñÙ²|²Ÿ)aýð«aTaA4æNq%ZÙ½p“¸VšáÁ9ù!Y$eZ“ň¼²½ˆD÷ÃÐ(¿e§N<ýþÒãËÎmhoY ÏmŒHÝfaCÞP’ú™:–Zƒo{ÞÚƒQIœG›Ìèq»Ç?ª/™¿O4}ûøYó¦Jic­ÀÃl:Òx$7  –UìJs Ïn,ûnÁAõ“©7P†æS‘½“6*j2 "ù˜+qÛŸB<ÿ*îTq¸~¡u™ª+ù米чf¶nö›¸çôv«î¦ 㦬úQMÓÏq&TTLï^‘»ÃŽCg/y¶›2”pG†u˜"w•ƒ'˜‡GIÛÔ ¥„lÁ±gZ!$OH¸ü¯°ÈOHo·«’Y^,½ òQê40Œz͉Z> endobj 218 0 obj << /Length1 2773 /Length2 19686 /Length3 0 /Length 21266 /Filter /FlateDecode >> stream xÚŒ÷PÚÒ cÁ%xðÁÝÝÝ5¸ûàîn!¸»[pwww žàw‚àÍ9÷Ü“Üïÿ«Þ+ª`Vwïî^-{äÄJ*ô¦öÆ@ {;zf&€¨¼ª*3€‰‰•‰‰žœ\ÕÒÅø_9<¹:ÐÉÙÒÞŽç Q' ‘ H&fä2”··ȸژYÌ<Ìœldbboë`dçiig0³´%ä\<\èFv¦Ù8ۃιYÚƒ þNÝ !ü`bø?g'KggK›¿82þåTfq;SQ{[[ ‹3ü_ù‰Y:M@u÷dü§¹ÖvöîvÞÿEf–v¦fÑ0uu`T³³ttJ‹ýcÁÿ–™]ìLLL\¬, #èabÁøWUOàßJæ¿Ä ¾Þö3  ¯¥ôÞÛÙÈ pqrúzÿ©ø_ÏÌ 0µ4qÍ-íà{‰fÿÁ þ;Yzt˜@ãÇ `úëçßOz  3µ·³ñümþw‹ÅÅÄTÔ>ÒþCù_¥ˆˆ½À›ž @ÏÂÊ `gåpps|ÿ×Ë¿üÿËýo©’‘å?¹ýáOÚÎÌÀý  Úý—†Û?sAõÏÒPþ7‚‚=hšªßïËÄÎdúÅüÿyþ>òÿoòÿòòÿ:üÿ7# W›¿õTÿ1øÿÑÙZÚxþcšfWÐfÈÛƒöÃîÿšjÿ³Îò@SKWÛÿ«•v1mˆ°¹Í¿…´t–°ôš*Yº˜Xü=1ÿmÈ»¥PÉÞÙò¯û@ÏÌÄôt •3±Ý)Πfý­‚6ê#ŠÛ™Ø›þµz,ì#''#Ox&Ð|±°³¼™A;j ôø{´Œ vö. #;_€™½ü_-å`0 ÿ%úâ0ŠüFœFÑ߈ À(öqƒ¦ð_ÄÉ`”ø˜Œ’¿+€Qú7Eû@ä#P…ȧÒoÄ`Tþ@>U~#6£êob¤öâ©ÿF x¿(žæob¤õ/âYý‹XA–F¶ ¹ÿëJû××ÈÙÄÒÒÄÒÉÄÕö_93 Ç? KSà¿r6–¿Ä Ñ²t¶þíäÚø£ ŸÆFN@TŒL€6@3—?Äìÿˆÿ³`ÿzcþØèò?öܬÿÊÿÏk“;({Ð0þ›3Û_[ÛßµøkJÄÕÔÞÆæÏœA×ãoÚ ¹dþOPŽ¿ôŽ® mÿ× (?ДÚýQE&P¹Ì~{Y˜Yºýáö/µ½ëŸaA&濃€ôæ=ÑÀ?M@t,~“ÑÂÓÁh÷‡Hfù%oõ…õT¯ß$8@…±ùkYëAÕýƒè®füŠäË´ä¿‹ mçjkü×õjþGJ W„ÑþwÒ Ÿöœbfuø­Åp=õvÿÓ6æ¤ÿÛ}Ð ÁšjГþ‡)Çß2Kûß=fÖÁÆõb ¯CŒŽ¿Ý€ªêèjï45þ£¥Üÿÿ7$33ÈøŽ0ƒHÿvÌ:ä ´µüß!dÿËèöG›ØANœAÏæ¿‚ˆ8Û9[üáÄãwXÐÓÃèbáüc~@…rq·ÿãȇëÔ[·? (3÷?ötÚãrïùµÄëwr O^@§ÿ„úŸÛÛÄÕ Ô—¿XÐÕþ_ü÷·( Ðh¿ºdoÂlUÜùP+ŒçN¿7Í?O¾§‘FMï½êÔåú„ “L]“´ét'œ<Öÿ~mWœêVèÑ‹÷I[#Lh{âÇŽgŸ_ñÊs{ð+³XÃ3…' CpøôªBû>/Ž>êÖmà=2äyŽ®\ÈJè Cåß'C–ö>î×pÈ"ü*ÿJ­¥X²@žoœ½øÚ…ž–íÒeáön-wæH&žÞ÷4šµÈ[{‹%æqÑk½R•ʇ Gûä-Úä…·ÈaŠ ö²wiqÔ&ÿr5Øæä‘ÏDñŽœ46“Ó*·ä$™oGc æ"\ÖžŠ¡M½Ã òòZ3ww ÉÅÖâAýýÑP¸û0'ZSy·XOïýíU¨[­“[Ù“³ú¯q³8ø̵‘u±—PG_ yhü›}®^ŸB«Ì2²:ï«™S‡y}•®á ñsWdx˜Âät”÷è…Íœäcñ·uèlºp5ÜÆ`pº}Ô:c8ß$Ô¾38^í¿²åOj§qÉ>{¹»ã7yÚ$ÍK³£µY™ÄŸicVB §GÐ%ræ.•ËÖñ< É|7­šë8Ÿ’=¸ï3(Ï6E%5§­:áo98‰€ZÐÎ5òomIÕÀP¸Åˆb×bü¶Ë§]?\xX˜:”Ž…'>U¤31M‘-ÐÏv¸x‡õ:僅ƒµYK˜‚NU9–ȯ±ñë‰~AsåtÙ²½ð©8>•y¥ ÖÞ7”ØÊ"þÛòIüÄU½ZR¿´Iíù>íÏqÑš¨úк‘@ 1øù#°L ´¦½öE^nüŽÏJ¾¾Ýñ›=sF·KV /ÓÑyß&¤ˆšÖíö£õ†ÝcÇæå[æiŠJ>6¡Ê6놫¶6+qTwׯœä">ôµe½@i® ›gàK½„¾²` }vröu±#V0Œ~ñs ³¹ûýKo>cq^n+kPÅŽn-§×w#¯ã9fü#êð†­ôËŸ‘âAZ”í£å÷Çìa&þüvVÝ,;]‹ÐK&‡qî̈ѿ&YŒs¥õB³"ú<[ëyb& ™-å—­G&î_?—š™Ÿ’÷8ä?óY^)RQz—Å.ÖO »W¼ªì«3)¹ÄÞ®&æðvXf¦{÷} ï%ªDKíç£ÛãÏÐõ¡>õ¥žkçB#œ¬¾Q&ÇU… ƒ+zÏI¤…mgjŒ>?Ìð²¯g*Ù /ŒÒæ©/{­IÈ4Á" ý`Ȫ¯Š Ÿ«äO[[»Ûw\¯w¯¬œY.Z$ ¥_ÃþÂMz#|¶M+8"rBv°6B},”PÏSB¥A¹O ö*2$YŽˆCï”0©š’¼ˆ¥e…,Bg Yz@ñÞºBœ$ÓõªcUkë©ÌðBuN߬‘*¹jžÚ6ý‰A–Evg‰R 5K|KIŽGþÐ;B…U’EYˆC7wKþrQK3š2àûá­¿ƒ£EÚÀ»{-¼Üëg› £òR÷ÉU…j5‡©ꪛ yî:è2æ_݉wR';ÇU¿&U‰õBñh8ëù[ç»w$¾–AÁDQîõnYèC÷#ªªÆJWØI…m~4 vM%\ÇÜ䢭ß-í¦Ü-¯(K Â6‘©ø¡*´˜Ûöï*ÂóÔzæ*KOŸÞ©Â“Rœâ{S}âa=µœ’å.èñ¢GwÛÿµÀæaÀN¶›·Úþ+"cî&§ÄtdžÒxÂxÝ‘.Eꎠ˜X{È5´ § `w¯=(yýÆÑuаe¤nI> ¢û¦ÏüùhÁ”‚ÆþÌf°+m»½e¼ˆáð0ÑÏ|^ïk vJ•rá-Œ;ë,¿K`(@EÌišðÅyïƒv…†Vªÿõ[ Õó7æ·\ypÍ_+ÂVqÄêP??Y³óSdjåŸZ®ºÿ ™ðÁ.ýâ&Z©ôÚ•€ªÛˆ>2”ø´#£HÜ[ñê5œ?2ðEœšë‹p–‹!mê´ýñêqluRÀ'/ Ã}bŸWÿ*#šê)Ó5„8(ji| Çøå×0B&È<Çó€öº€ƒ`Ì5¸!Öa¢ÑØ-N”ÁnE—fVŸèKv'&ÈŸmÃ+¯Ï+\0›ÉiøŽbrC]1Þkg†õ‰£¦îIx>ËÁH’2o^Œ-Y‡R”ÝÚdÁá«`é+òÓ…\Ñ…[1±ÃÚï'ÃbÀ aù87½kä6PîÕ©ÉÑѶ­Ç×X?Ÿ¦«ÚþµóehÑ~s­õîpÂîSÓ~ã/Ï]Z¬Þvgãþ&°ehãØ=ÅT¸íw„챓«òÙ'ËŒ›h¹R°WóŠÁuTÄÃLÒXüƒ»ïóçt{º^8ì"9kkÙ&on3Xsó_tà6G>x_? ËÑ|Y » 'èq#]k8åT;59òŸh†©£lÀµkU@™MŸfŽÊ8(ÖØ®x…)]}S_C,KSp…¦å~û†ÂƉé)‰B5×ÿº¹‰TˆÔßARå5¾ƒl¯ØoäSi¢ýŠï%R¸ÿI¨E;@˜h½ûqKÈè Ð#eÏX0î@€d3üm F«÷óÅyÖvø‘j±¹H£ új4·Ú‹Ä±L šŽƒ^?·wãÓQ+}yÓ½ ¥W!j\3‹‰^ùÝr!‚ O‰Ø¿xe#y«oŸò5ÞÈ'5“‚o¡ÈVBy°Ç¸v%ôM3D¦ —d”©À9tèUiÎØˆ³¢ç<µÀ )å›^¨ ª`kÎÆåÑ+”„I‡ÍâTIÌð£H ÷©¯š% Ê݃â#'z¥,rÇ!ƒ`PˆW‚Ô\“˜Œ"FïÍ>Ý¥$c·3…5]º¼o‘犄Ôd÷êbÞ$ã Ãlñ}ÔàÛd¶Ú³Üáä8x!ëf«‚×66ÊäöèܦóU]?Ú·ñ¬_ó_ù4‰7N”RíªŒOL11$öÔÄ0oP#˜>cE~"ý9. =ÅÛ ¤‡Ö5ÖtÕ†¥ Þóšüë0¹]’.‰dT03;1ðn×X$wfzô×2—ÏÚwÉ™œõ ]Þv)Dx»ámh–°Í"ÍihŒm#ŒwqÖ¹L~Q=5’ºü!-ågßL¹Ì•–úÕD&Wãi/sß ´)©Œ3– º|㨠:ÉP¦qÉäÇ>dü‹î63“–Ùù\§Æ®>s¿è‹Ôxv­Ùþ@•½+½Sx¡°ÂöE)ç­ çÛ5ÌácÆëöÈIkuüu×}óþI/ù‘Kçï})Ùl’g“ö„kz ¥7Û8q2g½ãô¡¼<éM¦>x ™áåwM=¡qÊlÎ-]=/Ï÷Þ7ªd—íRæââyyã´7Í›X+Ãú ‚­À3å×õ<OyáíJÏW#©~jT.'›§+" *=eÌ;+<Ý ¤OƒæêkHù%ç®vŸuÊ–ºòm.k˲8TËp³¸¢§Eíx#_HÓ’í׎×ëµÖηᨶ/:+’ªõ%/WL¯}bÓÖ5;dšãÙWñGœAÕ¸Iu±Í/Õ€5B½ ñ'ºMùˆ$–«°~¶éør©Ì>D&5F£¢³ÍGC›ê1øxVJnLh?tšÓTqàËþx°£FdˆQü5NÍZv369Œ! ‰<ß$4ašçÑòŽ·‚´î»´aöh‹ÅuzÝĈÇr„Æ ˜}t5æŠv”\˜c¿FË«ª‰±qÅB(y ^?3cBwC²Æý)q~ilðKç+ï8Iò§T¿ ¥|ÄKrw˜¶ÊÄL\´Ðd¾oॼ-}ר_ìˆÂ7CuñÓÿ©]²!ÈõÞ ÎhZÈÎ4•Ö¦B•ěВ£ÖÀÍI]ñ;VA,xœtÊAnP¹Å§MsÆtŽüZ…¹³áy5ÁksÑsŒ¤ð¾ZÒ0²:ACatÝó÷ÂiÊJ“D0JÖ–|÷ªà{‰…qH['†a.ÓŠa JÃ6“±wªÞ4ºLªS„ä£"ßöòòëß1%¥:༽Ê]6&i6Ö!}ÛxK2R€Þ0 Dsêd[N\Ây4í.OíK:p«aÇh€8nÏbbŠ™)Ó¶˜ð+YB¼9W|׌ç"xÍX˜µµ ÈøÎ+^U#w<¨—%åS¥’Ø_[Ú ½íé)¹|˜ h,Ù;¨ííF)·9@¾­Ç{Lߌޗ^«p c=Fì_ ÏûzˆYÁ`E´TÍß§AAƒ®óÅRÍW5J› Ë­˜0ÃÌ\0Ï-îXx½Ô,6 uIbÎg¡^BöÞ,}¸Ø5ù´\›_Qê¥Dóq¢,UeYyÏ”†èÀFá {%ÖÝj,´ ½ËÕ`¨ò>ÓæÖ"ƒÛ~HÆÃMÿW7 ®f7;ðW”Òp P;‚cèŽVåWéÔðã1'óõ´9ÍùÂh#s³N;eR³I¢ÜK¶Nï|8êÑ~J>’ó„^Ë8½6ómPU˜Kê÷66øÕx(AëÈîl,à`÷mi=i}›H$yhM"Ã4ká$Lý–±é‹Ú– 9ñÃþ0ì€õWå­êg ç­_®Ç·DwÂôx" Ã~ò"-ÉrÖ ž[Á‡ á3wP‡«¥É¾óÔ™wxmŸæ>Œwd¢£:s„–ËÀËO_IîÁžë·º‡yqY}„ã½$¶µØî4`VTð¥ÆÓd¶KrލýÞ0ȳJŒuzP®†x1s•ÝC7ÛFīؿ’ähìùð.VKsœGC$iBPêšFÊ>Þ8û²Dáx¦vÇ•9vÑÜŠÁ´duRFc…PÍtd7ö!ô;yJà;byîåÄ1öý˜›Ž[$§¹ ÈqðÌøž‡hf᪠¢±§1¾”°óŠËÍ0S# ,’¾ßÈVK-1r÷Ÿ­d›[´§9·@I# !óQñ™npºRºQÅuLDZµ%jqÁímxØ”O§ÜaífCæSeyÊäèpÙÿÑlÎÉ~WêécXCº›×ÕAæÐ§K`Øê[¬aéÈ$°Ìø>­]°B²ÆÈ㓱Ÿf{ n$¼Ã/R~OTô¦ ¼:é–é+‰Sžsæ´Šhõ>ÿWÏqÄg…ç}>búbf²Zv¹¤ yTTizW¬Å$és0æPSÒ,4îsŠtéXÈϤu€™!ó˜ÍfšTpÚÒ*Ì5 ;Óyße-×tc%È– *·ŒAsykvmûAûuŒø¥¹ Ÿ¾šå…@úaožIö ,üZi:™ £Ë}ŸW™¶kn=¦!Î…Þ"™–»bh¦Àhè]‚r0½sÄÂb Yúä2GŨ3(´ëÃ+Z¼ ¦gäı»¨y‹¬ÑJÑ)¦ˆuþñ4\˰ôÀPÆ@¤f‹Ð¢dÁñê ‘˜wŒázf¯‡f°GNþ¾räâ Ó¸'©X^'ÕÈÛØ^6SzßnN-ؾÏ}Ñœàš5¡o1Ú'^¹Û‘@õ²‚ò© e\åiÉ^zI]w滞咪ùhd´ŽÁ¦v“Á©U~õ—×$öšRJ.wù0rZq·(øˆq’`ÈÖùò‹2ÿíD£b,Þ-ÄÃ[Äûºéa`ôõÅE‰±ó–¥£Ý ®¶Ý8Á@i&ë~pv’ª¨Î|xÅþ¹³ÀOvªüØò–ÛôºVùô®&—rG†ÇĪœµ8ÈŒ^!¡‰‡UT4ŽWñý¸ Ÿ?Öhƒ° 7ÏB¨QÂü‡QoŒ§'ù‰Žˆæ; _¹};×ìbÿElßsÕu&·h >—’¹ÜUÄ134ûŽ-Ì÷™ÌýªñþQÁÇ—évz¿Öº‰5éµtÊ÷„0v•2:ûeŒ{Ò zãCS I–qKsúx™çÑlùR¾ú)@O#åCÿÒ×/ë»L9¨·þDJ…" -4Ú¢cÙŠ9Ç*p'L G‹F±ØíAÓK½íÐ{*ÁuìÓÉ÷ô¾œáA° ÃX«ô}™4Ø­Ò¢Õ±ÞÌ£m!Âi›ÚC ?`ªqœã©$apíÒÐ8v‰0:c ¸-wÙ7QÅ^îë‰)Ýåú‘À÷èÐê­H’À4Ü<å§ÍÜ2Û ñ)èC¤`ªC>½Ÿkî”Ô 2û|6ˆI‘X¦ÔëÑÚH!ö!#ýí¤0Ö’Ÿ e"5ïЋ'HÇAvü$°5§×˜ mãVí©t³úË4j^‡2ȵÝkÎýIwJÍ‘&¢ ×)šwv<òb¶V¥ ›5¦sÛ%%™m¦KÆKè>'u0õ¦-E¤œÆäǤ]9ÎDêª󻀖ÀåŸñSñÍ$ a½of€°÷â™/4;ð(µØòãwA<œø5·V¸¶™iˆåÄM”ÅP=¬mƒx|:N€óPçáy8'vqëúéÎíc<¹ú©òÞÜâ‰C-sø“\së"».|ß4HO4œ¿Ã±&øö¿Ü÷Ó›Q~™ó»ˆ}–èèµ~›››du #Ÿgòn8 >]£6íù‰}”ÀiÉkË1MTï¬U$ý£¤75îþðé}2M}’ÝtïK,Œ¦IY:4Ú/ÁÉ'9ÀOTfˆ?É%ñeà‹fûhÕN‰ÛV)ÅSyóÅ4DŠ{ÃdQ¦îÓ™–2» Áï«UNZ œú¹ÌO°æxóUç.k2æÕ¥e°UN*Žœ ¥.^ËMÖY0"Ínº¬Ž+«d¿VE¨±Ô1qŽ$ óºôÛVàÃa(.f7–·ªX·"ÜxU)}›{º.ÝøØCÄÞ+qㄽÔP¤~ó‡Ij«I¬È`I<Ö¨åç{&ÅËÈ¢ý2¸&ï³Øþ´ª¸^5ª¶ö.»;Ð?î([rhšamÍ'æ‘ÕQoõ°úi7T ­r˜ üǪÛÍ£eЏõ*xÒbΡ×zèç3ÕͤÕÝÝö·–€8ôÄ×ÐJÒ\õIžP¼ G3=à`•)Äjɶ:u&ÙŸ>½¹â™R%…ç)ꬥåP³¦cã³r+–qûûàòï¶J±$ÁˆÅv¬ê¯›i)X7y¨@‰²Uã~ DG)SIf|12géø0¦k‘-S_a–ðÆr%S J¤»+ÙvžÅÉ×ÒÒU˜®#Q¢5/÷…íW>WOýKÙÇ\TÑx¬ (}x{Ç0‰Ø:§àÊ>ãÃO™/ü,PYF™®üÖŠñú±i¼ì³ã®üF%WCi™…Fó=FNÕt纮3[“ ÑGò‹€Ù· èug:Ú}³r X\ô”‘Xð2ã—{w¦¢©ÜDï¹,SêŒ-×*]œøûTŸœŠ­å5Þ“¥ÕUg–1§Z1Œ9WA³3žÌ¸œ ?xÍï;Î' ‹ä³ 8rœ¯’hÚ©aÝùk°v¼Df?„gpû=˜¨2sJ"ÎÛ¬šk;Ù'ƾ÷ˆ)©eà/X9n;âl1-Óº¥)}[¥ûdî$ýø´vB2|çŽ7.)Ž?Ÿý–f9Éð sr©ã]}+9D†K²ª/[x#ƒLTöe§ýƾ)߀•M\ô)±íǼëTÀ½‰£ˆM´÷—Ê{”ƒ4Üш¤ïšÞ±ºN;.ê¸C¯»ÖÖ[G1àÜ—t-I„~¬MLZOÜØx†§ÅQãi´àd -õ>$eFZÄ&~ßìÄ­C"ž8húÐÒÃqšC€mómœw0«j+â”jñ+D³¢´…ãøÜ½+—õÙ¨áÇ%%úáêZf¼bÊŦáÄ{ŽùÙ ¹…Ï–‹Jï*’vµ tF؉ÌVuÌ¿DŒE2 ,}|Šöæïè²z¶ÌHÅIÛÐt8¢É=$·|žñªÚÏM¼N>:»r<¯õzÈŸÐê1=ÍQq,C œ ƒP œõ²üï†_~–Rómã½NþÍtñµ€/„2[ËÀ:£f±¿¦¸üZ‹“_µ¾©‘‡ú±9ü6׊‹ôøqd⌔$ »¦†ê[øqèÔkRj“{ÑcÜUçŒØ¹Rßñ#AÉÁcÁŸmÊÇ’Iøm|=D—I,€=>ãP-†j”ŽTÁ,¾‡+éŒtÉžöÑ <ðöÇvY‘­žýr;¥­’g›¤f¨6±É*‰g›QÔ5t1JTgåD§²&Š“;â“ d’Ä=\MZïØ‚Ï$%¶û§¿u|þ*¦¼·ާž“fF6¬À½£äi›z¦(O£,BHúÌÿñÛT€ËZѤ-!jû+üˆÛêÔ ´ ‚Émôpñ„Ôò·§â•_Á‘PþëBxE¹Ç‚B~ð“j\µmÕa?júÞ‘4IÊmå¼Þ!c±°Ê²Å× +§Äˆå1ÊZKn‹W=Å#æÂœ¿¸¢‚‡&Œëc5m“ a~J¨š·ö8ŠÄãoºÃNâ9ÍÈþ`ü´áÎýHÒ¶Ø|Å÷†¢W-[3ˆ<˜*ýv¹/¸Ô%þ SY´Ä|—øfpymdâš“^`Í& 6][ét+tîÉÓ~}~rS¿ŽŠ­î¼—o™ì­MÀhº#Ùô9‰²ÀTOeå}Þn öĦ¼!žGä  Ž®|ÇíoÁëéw¨ª^íÒ‘9qLÔà•¯ûN¸¨ ó[HAâ”ÒÕÇpal¬ÁØé·9ÑY÷Òô¤ú÷÷½#ÄW?´ 2¾þhÇ«f(Wñ_pä÷á'øz&hê¬1ÿ±¥’hxU"`)Sf“Ù:qÁÆ—«ñM7´‚  /¢Þ_Lþ‡"‘þÍ2úË;“näzQû-¢öïú?¨k¾˜tÆ%E\Cd>ê *Àk²¬Öõ|·=u•ÙÉÑ·²ò7®Ôe“éOÄØ_Çå5Žçã0ƒ0ìË*m?‘ê—²´S«4§AMc5 Z®…ôko\˜Æ\B´7MÔ^,ëäÒ>îv_–Bê0zÆ„|âô›™Õ´aU΂`£ŠÙQ!ÄÆÌ|f¡šxGWؼªN+?2 L0äŠû‹=IU™Noª'/'úpL8{[5§£¶„:jž‡¡†®Í727o8–WYÿN>*Ãñ@_M£ Îñµ´óÒYX£ð—3b„  @| &LÍÞRe짬ËÔâ¾-¡@;ž~˜Õ<åš×¬9x—Zb1TÛÖáú@·—ˆàÔ©$g+;Vi¶U¯î®¥×8ìiœ¾ÏÑ•ûý¯èÕ±ëgzv'Çî;öè‡Æ4,¤‡jöŒõ·ÍPlp€ísF’’5}„ž['§µâÇp™e³×ºñqKÚlíÓ¿Í&ú…Ñ'¹mnUEb =™ÑàjìkAEù=Ÿ-x>88"oº ÀZó-3Ùâe ´ÇˆÄ—qÄ¢mmo2ÂbÍÙeÜÑÞí¥ÑÙZd²Gé)œè37¹ax”É·8ýæ)H‰(3qØoËx‹­_|(DãÆ'è"Å·|¦qqCº>Ô¬‡Ï`Óç›ÞŠÍaÞ=¨Ø¤›_}=p@ÙHÛ-iI¹äýym¡sßÍóQÕác‘‚²ÎQ®ùÎëÖ˜ŒsÕ[.RÓÞ‡$–hÂË"cšòB(?Ñð„®°H9 Råív'ûx›ÁOð´¾#×Â0“J[XŒï+‹±Lu­úËKÖRhB$î;þöÏ)À¬ŸóºÓmv´ø&'%Ù·nÒù•:¦Zì§`°¤C,¯†Î{ªf©üîÖHÓ¯C­²f†7ç¸( XÇטhÁ+²ÎI¤¼bF ÌJuQƒÈ• óìTÑÂ3Ö¶Ê;áæšw<§tD) 1’ ²«fõ•Jdà'lSˆxÄwz¤g$ÚŒ×^ƒÖEù<Œ¹|xgÉ4±ÎÑ÷3ò7‚”(³ëvñMWþŽ8 wꉉ:›]Ø,¿JŽÒOÊ ÞfYÛŸ¢ Åw)éò1‰ã ›©x›ýòFcì\eŠ¿tÓSy®Q¯ÕI’/Xf VÑÑŒÄ çQ‰º’k®~¨*:ƒ‡ª«_ÂU¬‚Þ³·ÞþŠûA|ñŒÂ­ÁþÍ®*šûy ­i8üûnö’fוÐë#˸Q¾ ‡x»4 ³;J¾³l7!~Ù¸2@„±‡Èµ›‚T…ÎÞÉÒ¾Bsÿ²¿b[7tN«ÙÜWá5E»$ì™?\39”4æ&"Á&¸hŸ 8CØTY¿p£9·>EBanÂÅ9Ýp˜{DwU9µIx§Á×g­ µ"¯·?ÁgPÚ«Ë¥TÞIàœj8Š&5w•'„1VþÓ‚tFßÌ"`x`ªÂ‹÷ÝgÅóÚY¼µ×¾x°Š^ó×§Gg'+¦!'è Á¬S™+{‹Eÿ°8Ú~öÌ›šòÝqP'a†Û¼ò1|îQÐ…z½Ñ„«÷´VîñxcÓ@ÓI‡ ³&Ø` *:âB€¸‹mæ1Ưó:öX9zÛßÄ{Fþ|‰Ì|ã½\³äú¡ù¯°•¨ˆ9/j0«vÄq»ðUk¤½ÏÏÖ݇Ÿ>Ã@;ME$(†F©!çk4S®wÂ÷¼ƒºª KÊÔŸxèa‰EÖDµºú²#¸ÎMÄŽÇ e8ÑÃð‹h¾4¼…lφΆß(ì$ä^lfª©„¡åÓŠqg®A&ræ’ÜžxWˆuax‘µWÅ06jG@úç£Æ£’þænwÎ:ì0´œM–þ•À4, û󵓺ڔ€ßñÒª9ÒjJ¹I0߇ ÷´jðãɺÄ~P(k“gÇ¢&$—dìysH9ô®æ¸þ1æ¶‘b?XD œ‹>äe­8æéMÕlïD"ÉÒƒ©v‚ûdguU½ªáŽŸÐqØVn³çÉ ÊÔ*±Ÿ7 ¼ÖæcÁcÀxÁ–Û_*C¤¸n Uß™GŒ”¿ Dà¶eÛÜM¥ÄŽ»1&#ù\/“^ãžÝ ièÛÞXQŽ‘Ñ4"\¿óüÚZµR’Ûÿ‹¬ãþ(96Y‡ ×,iѪ3¹|eÊtC¬´•µ,q`æWÝX<³gyÎ;ŠÀq Õ=cùÈq—“0ùÍÑÏ“¥‚ÝñóÏ—nPçM¤øÄâ­Ë« J²‡}†ö°ÌI3ec³Ï«Ê¨UF) RXc³Ñ––œLßBÈËùF•:.vŒrǦ½ÔLjCì”/øÙ/Twæ;¬Ðfe'<A?ŸÐä± É ÄšÞÀÇ:ZÕz` ¼bÛÌ&’ø®à—óÓ+ ¿ÀS®˜š}jÓ­me4„Õ+ÿúâ6B;þt»Ò NçÿVJ×mîR³ßñr§}-(ˆÔøË4|;úŽĊÖñûŠGÝp‚=áÞx•ù³»­:Âþ^ÊX×IJ°l/j[S™J&Û^0<ãºð²æ§‰å&íS…ƒx2.8q®SÚ5ø;pã(î¤Á,UÏ7B_Ù´…š ±̸§wî¶/4-gM‰èÝå*Ÿ3 »Ÿ\ Uû† ÷ „ʧÅ$ér…ä­„y] ÒÈÓÝP]Ú"L€Hc\ËÎAšMÒ¯ÇEŸ[Ïé!Û´¤9% uÆžý–EX*’Mˆks©%¹ßzb Z* û‰ÑŽ”È˺n¶pÈÉ¢¡´f}̀ѫÀZv!4Ž.N!ý¢„ž¼xÕ¢äýäq'=íÄ×I¬<£'AÛ*ýIQõHòH1h=X»’èçy6=úœªýbíé€b>¹h¬徆=$Ö‹) ªú;¡4y‹&ÆC)fuße°C9o0Nòˆæ„mkLQeÒ®nÓ(Õ_‡ xDiÛµŠ^’–FâCàԕ­h)‰ß&|…Imº:¹gsÊߟDºKRN¥ôA6mÚ¹"Kš .±²Ó„^c?=ªŒØSû_cž0±€›ã²‹Ã:k Ž(pï—Hݳz¯*Âyé¬g,¯l »^»Îü’tù3¥¯0$OÒWM¬=ÔÊÛl}òs~Ê•—¶X+‡Äô´ÙÇ‘I„"5Dß2ËŒ(¦e§8…ãò &Ú´î‘ü¶Ÿ¸nÅÕsß_ZdTº•Ä,Ö*ËOòßu¨)›•¸ÒiÈmlçy’LSë~óí»/þbZ4ÔÇxy„4WǬ>¡ÞôüÎØk |Z/_û£@¼h!'peÅñë]Éuî{Ý:Çù&ïXŒZü˜:E”}ì燑Ï)ú^ÇKO¼Âv5‡å=CÌÆ`¬×ù3‰X†GpDSíN)Øâ(sÇ U#71´ùïvØqUò”O·–£®¶`DZôŽ­(”Ÿ–c¢'êêÖø”;Õ k]©ÓOô©>Üò~_?›Ó:xu9Æï Zjˆ±" äì'Ý2æ”Ð ß#pLëÆC,Å óŸ94;Q¤@•ùÑâæ6©/åÀQþóu¼~Ä×Þ@´u<IÆéòg¾vw¨Ãù0-$«?â³3§Ámñ;Ê×¾ËVvÅÇ¡LHæ_üÓÛ;bÝ„¥ù»ëÄkþ'×®" NÖ½Éb;]è<Í?'ÈÙÎwÞ;¸®MëîÏjoV‹X³rDb&o¥”ÒcsJªÜúqÞ Lrpªûˆ‘ßòFo|Ä;þé! ³(v¦®³u·ß_§Å”}HJH<g›K¢s®<âœç‹”|$4€VÆStcꉞøa—ŸÔ˜îÓm>z÷äyë#qóáÙ®†1‚É£N) ¹¿ƒëxÞÂ4kÿ#{‘©È‰\&J‰"í„°*t\øQ¢øTçî^×ôrÔkœýÏqƒüÆ0÷ HªŒA­3:'ð´ìöi\¨ó¸Ø±+é;ÑSC?ËëøA—ȸèOO[%35ûÄ›¯=ƒ%ó)3CÞrƒ2ŸKî›B‘d AõÕ|ˆ5=Lݪ‡{5q+ i׌þ%ÞÑÏZ7¾ E ×bhyÃÊuUþCwò í•¬¾MoîIÔinÉÀ£Ñí ò(«R³BÛЧqѤX_ß „Ùutñºð%ÚgºG;DšªÐt·åŒ•EžFÈÏ" Ш÷¼Ì²¢ÙÌ ¨kh¡¹"àÜ™Iб»^×±^Þ¨ï쬌%­ËºIk~ÅÛ}ËÕ rˆ/‡¯PÆLêr¥Ží˺Ðh¨ ÆØÓca 2zþ޹¬Žá´#ˆ»é1%… UÞnv~°bäNø}÷'듈%Ä—aN–¼m¥¥¢GQöœÅÊs„êæSóMNä´¾V¶¤k§ V )[ÍîU¥G{•Ö°â:¸X©lZ>Þ- $Àœ›{ä-dÈ„8[¯5{Ÿ<ªƒ ÎkÍš…] “«ïl¸—Hö¿«ëËuýÂûÔþê.§CØDÜ ‹KžØŠG’³ºiÍ÷x‰PWÐâ Ž¤»šµ/íú³Or| VÒ @qHìñô°Þ¼Ê¸–kÓ¤ éû'ìGÿ¬Sú*#²®¹geï‚0î®ËOO}}4ö§++ù°l®aSŸ»œXM*e%“ÁœÙ¢6x7–ŒËi( ÈSÏÍw3ÔÅÃFsMFI¨§P%¸Ö~<.йVÕá,R¾îÄì7mÍ „ðp ‡9iŽ/lo®jz°D™AeŠà_6ì[#ÕžNh'´Oü»±õò>êÆN?X¯@…`Í |a±Îùé#žä5‹AAe‰³“­Ö&ÐQMûŽ=¥0󸆫Ã0Ù¯†’#Ž£(®m“ÖÕø½Mš}ˆ‚$‰OõËùç¡í74q'çò5»W犋¬è¹UUGúïÐæš™ÐB!fÀtã¨rÓèeÞ° ¨¨¸kAâ©Ö´STæRœê­IvQè‰/ò)þ‰ÏæV]z¹UaJaÓ•´¬óïqÕÍU0¾3eodb›ÞÌíp"›&ÉmÆ ¬›ÁÕgÞ‚-r°ìÔû=Ó{Ä,0ö·3 âžP’ÜŸäwó Ô’ÖŠ[J]©KPLÈa¥­ïdïaäU^—,x~Íe”„›¸+zäÚýY¤ô–óêA $8\â=Œº'ˆ¹/CJJѹŠÇtÍáåÐ~‹¾°´š¶¸ôÊVŸtñ=€€»Â…äá­·‚<³iH_ïì â4!®ä/Þj‰då®ÈO+g/ MTæhD,ÐõeÊkN#‘ w{!îc·¯|b‹õé_QÓÖjh3à\ƒ…Ÿ9öÕD`Sà{?X§3sd}¤Æº¬n¤ž0èûNÒ8¬ØÏTúŒ·ýµòu¿Ê:v7‰búcr³¼NLMÀ(y(z²ÇÇÛo«îyþøqí½I$ ‹¦L³îJháÅÑ/¬Š ª[y&qí[òÅÂNdÅÙJ\;ÐÎ… 5PKr6ݤŚ#÷ü k)¤©™´ÄGtç*S^o¬1_Žê¡¤®»‰vPi‹ÁáÏDÞí|â?ƒj•LdsMô@½ñÔó6ª£ÂJ>3íq„ÛÆB"Ĉ—èÛOô¢—%´wRjŸfŠëy4¬‰Ï‚®ô ñ@_¹‰.qós÷ìM0nºlòS’Ë?E¿Nr%¿§t×kÌR ~¬ú´ú‹ph‡êz²G]Rˆ§r°m(âYYtøPPÌ×¾Röb!j—Ü@Å×\¥äÁ¼7–V†ñ•¼‘ˆ%ÏöGö…¥ŽìžIe•‘š9F¸€¢á²yË”fRj:ˆ˜¿¶•7?G2rAóKï±Ö‘„½J‹¢p¤äûDþHÂÕéÎ˵%„TXMvé¯å¾| ¯V™UxéÁâý8oð.’­¤3f¾A=•;Ÿvc cXqG×…zJ;œ˜ƒÃÿ¬²-öS/Ä«î0ØMj+ÁÎÔUºE ¶•Áy²ážÉØ™-¾õ—Þ¼Ë}–ä]5Ê­ŽqèÃò®ml”°®ù¬³6©=Ú”:L‘7í›ŒÚ n/Á /Ý—å=0 ½2©ÉxbB1d³¶èÐÍûDÝ—Ô/6V5%KX£,vvYpeã7;àõˆ§Á\õw+T‘~Ïèy~Yøºå®|ƒpùå>¤eVDŒ[ñzGE‰¶·fá%˜ØôÇ»&¥ŽXÅ&·Ïí­~ç1^Œ,RÀv ¸1;–,HªÏ؇D`A‘Èø)þý²Y’»5(“ä{$œ‰Ñ E–d=eç[½§Ÿ’Ò±üC£,ÀäJP¥ÌRÏÜËá°o’©D.梴H².`Ò2m ý^¿î¬M) »r‰‹óДà<ÕiÜúŠÙjÓµ¦ì[ŸïªokIP™P$[\ªc³cvy³å»‹ð±Ôzqàa(™u±³&í6?_g ÃF¡€‘™v6I8u@å=~5‡/n¨Aé0Ñh)CqŸÈˆ:˜˜&¿IEVò/ýÔõ3­HHMÂóu›Xh†ö%áʸ²Á^5˜ôGO_hjå5‹%å’,‘#FèÊı7`w&& [*¤Ší%£W‹ö¼ºú%Úî­¦¬ŽÔ_ܹBÐËm×íL•¼âQS.s¼&z_Ö´k¦ Èë²JyýàrÝ;QVO^ŒϨP¸_ò”*V'Ø?Óe.“É¡Å0ç–s‘9cG „1ɲ…G@eLrûÎr+Õ 7Ž6™íAœ&qi˜cY•’º£ýP9è"@iRˆp#~¿Æ¼y<‹0c¶«M®3 [÷ ÔÚd¥+_Ø“–p½Û»g¹Q~-=Ú Zº•F³û•^ÕÂR¿À´¿Êv¶ß& ‹yòÝ'µ®±’µ¸hª%$x”ÐÎÿVࢊÚÐÞ)è4‰ÛŽ¹Çž½ºÞ´àªQóö³Y¨Et[ÄHåUo •´§@·éóhX„1tÍ|zqqcì™ ÓŒ·æ—YÓ…¡“3±Þ&’GºÊþ˜à ÕFÉ…@@µžÚÅVfnÝ«·©jônSÁ£‰FDbÎÖÏy  šu¿ OöhÓ=]’‡³y#EÚB©Á&¬¾qh#í><1…ÿÂ`PLõƒÄœMÝ,è)<¾àSQ^‡Ö`â/ìéPUáP “•Œ?vñ—e$-}3•ZýÖŸ)Ž ½à Ë#¿N[Œ|1ïKü 4äÚiÌ´-ÕÏI4ñáZéz Çlg‚^+ññ¦ÓkXUZ朤h½ósOÅÉë¤to}ßÏa+ØH(vlÙZÔoösÀ-…ÂO9A¦sy>0hñ…Îé– ºÑŒ†^A‚-‡M§Ï8ÊÏ?L"“ &EN‰dˆÇwõ]öÚ*8IÓT);ç51ÑÝ#Þxb¬³d@o¤¼r÷ÖmuìÆyY*Ž‘l¶P¦BØoy6êä ü"†:·r.u!nP“«Ë¯[–rA^R#QWrld}Fb0S‘‚¬r÷2ß &z²ÀnxRÆI—‚¯â™™^4|4¸Úç ™¹ÍœÈß²ª+P‚èxü^$Š(âЂÈç•vUfRHB>m@ûn…S@ó>¨·@ün8cþ½,ñöâø¨0­ÙÀºI»¥ž‘ŽìÕKæ·ü®îŸ?š…m•9˜QáìµØµ OÈh$VLÛU翚ñáä\v'ƒgè`/¥«/[ ŒëN tWº$6!a¬øO–¨¡:_è6Qe=ààÈhé ¨ÇãØ;s´ˆ³¯ŽUfkXM4?ZxâòKTéÊÁ<íiºÈ)#ÖzˆÚ€\Þùt®k:¾ETyŠ¥‡žšóS鸟”@œš("Ç%Ñ⃠;j¤äÚ7¤äÏf^jzo¡Ù &ÝOä½ë=}¡Fb~Î÷ä=üÁ£cû?e3BSšv¦‡m ýÃ< W¸D†71q…ŠBí5· Ê Ê:1¢?g$6ƒÇ7in>TÖb¨;'H€,~€!›) ù!ZäJÿ•ã-üˆ‹JQ…›púõ-t É´ˆÐßÏ6n±Þž—M0”³Cð¤ã¯§L›¾"Ù{Üü®¸{°\ O«ù¸i³6~ ólx¢Ä’qDPú+¾³¾w^1Î&ØóNŒVs\¤Á‡b—üÇ¿®¤[¹`J•%»‹Á‡¹ûpZâdäê< ¼÷Âç‹ÔÖ#Çjí¸¿Ä´—‹0T°Ç«×åKEî ùúf0Ëú𻸊öÈჶ%(wqÞñYÂ`@'_\ÕÌ"“`õ™!÷£#ñ¢WVÀáe¼pºüé-»X:[>°2Ô}9‚ÝÅdÁÁF|Ç·‹5Ç|-Z‚Û —"ñ Dsë^-Y9!ocJ£(åä´¨ž0/߬˗zª"ÊÍ¢{¢nI³ ±†Í†V­±œh:¯;v~Ü|¢%j’ôÉmY•ËGL±y‚ìÏä FRZ}Ý_7¬¦½ð$Áë #®¢z#ïä-_vDdCSF­‘^@ÆvHæó¥Ôåõ¬Z16wë‡qî{@7Ê~^„¾·l9Uüéöe·¾ýê¸5å>6¦ eÛ‚A¶Ä?ÖäÕ’G•:ê[GBÝ–nîÕޜÖl’B\-ÁE<"M‹0ÌØFÅJ¶ÏÇJŽ-¤0Oqš5óî¸Ó*Ÿví‰ß«`Ybz!` Eæ`®QpèµdÞÈÉoÖ(”8“ѱmóáµ-ÔüXùyz— ç{úA$‹å`…8¦’¦wïñImæŸ&o¹áø(ò‡±{ðÑ1¶&—nçe7Ó±cö騞Ժká‹Òþ“<£\‚bÔt»õÄìÞ3ÝF>›/$ÇaHq7c|µœ¥Š©³ï‹Oä¡ò<4‡‚V˜…¹ò_¶–mn R¼G_ã†ÿ+icŠ·i}ʬD‘  ïç¼(„ MžVd]îÚDÉÆ‰wO<$Yd¿½Æ¨’º±ù”ÅHR9yŸß|{°ºÎÛŤG6ÍZܺ„G nÊJ÷„FÍî*'Š¡~y§”°A D¼Öé…´cÀÊs&ªòÎiƒŠC(Aƒ§¡¶CƒwßQêæ¤eõŒ—âÊÑçhZèV€bü±ø¦«'¿L{î@‘:Ÿù]¬wÍýX᪰jI>š²^—õáå¾™ý{TÉæƒþ•¨¦ÿž¦í­èváú’§€¦…“UôŽæRÁç‹YòeOŒß¡öó’Ã~‡àèG¤É\AÖiµ369Ñ©C/«Ò”ÃþC#%s˜ï䨪ÖÓ‡Ù¨x‰¹¡Ä]Ýc†|hG¨Ë^*gšÿ=À EA4JŠÎy§‘€uq[—¤cwÙDؽI­ŸÉ_Í!ŸPG!⥬ƒ––K½¥x¥Þ¸°šhPCúw´ªi"R5Õõ1‚…$¢-(jÙî2é®|éîV2jUjgo0\êbªÓTÙÛfE?‰®§ã¤ ‡¦Tá–×[Ó'):GÐŒunálWv¬0(4ÞÎýãñü:G_bùh—êYŒÜ+Eš!²X Gq~0CæÐ“›4>G,ŽëìýMDͯ1ªcÕ¹Æ[Žt$’0X{4ËŒçµ (q.›¤Ì•jààP}äuà ¤·qŸí7 ±ôXkïÍØò«¸WOßjö<–YhvmåÃÅŠÂáÈOÈdñÚâº¦Îø.ƪË8‘GA墡ðNÓlÉ*"gÓÕÑžøख†WY¼P»¦ãt?Ê:…$ÎP#`IqWE“?{ª Íî$Diô»¢w˜«¡Q)š—уó|ð'èǧ£Us¯U®l™æ4>¾ÜE±{%|9^µ{òâö¢œÀˆü¹€¿%¡6Ä-Õd:g0瘨žg2£¸9•*¾Ë^LwQ&}ËcðD-:9ìüCëS¹\-u-xŠûm;l 5¤/Ë moÚÍß8ô»c_Éd"vß2®Ù­+‚.¤MGs ÞäH+Çjï(ªé\Þñ›§êÞU>H «ÛÌ©ctÍf›Jƒ3``ùUÉËdQ‰W¶y®ÓÔë·r‹9}kàUâßrsMÅ®9Î XꂲOßw¬Ð(JÒê(ÿÁÌ*/†úÞ…ýÚ:†n/Þðt}ªY᫱µÿ/Õó1#¢àÁdâê,aJ­Cm¶’YE"GTV;H•ÏŽ™YÒ_Ý'“TR¼ÜÐìÙXáÐãÕÏñ&2,å`ÈÆy×å(–¯ëÜφŠ¡ ÛŠlÇÞ"ïíiñ{ïNK­àÛf«´2LܵºÝŸE뺵œ”w³þïìÙÂ3 SÎñnÚÂi…È¢Ê L¬)‡¢°nˆdáæùì3Œ£gæ$¼"ˆ½VD zX4ªÑIÅd0ùN§jØÑ½¡-Ò}:òÇ$«ÎÙ¿HâÇÂ[ÜžOd¨ù×rÔt§[ª¤Q›fƒºqµ¯™ÎË.¤W–%~›¿í‘s[Ö~X7ƒ”<¦ý©ñܘõC…ª_õÊÂïhñU±Ðu¨HÚ‚[Ôùc¬aôŒ%ãÝZØÇþµÖPéNƒ ªJaÜžíÃÿkèž<«¸¶‚ûÂö‚èBÊ(XL}“r䌨ÿç0-;/©‘Ô";r¡l‚Õm47Œí˜™³X¦Ò×Q§)ùˆ6&_e‡*>ÉEdvö8Õ O«è‘½Öߎm˜æê†VÔÅóÔ¨¥Ã[Aîk3èÝ* µé•Ñ!ä¨Äæ´³RkrÔñfHÂÏiSøÍ®›‹ÿßýBKÖR½-"ó÷€ËZ¦ìhÙ d„ê]6 ¬}Ãé¤3yPÁDeÅ)¨ƒHGD Y'ÅrúhžM%¬7{°>wz|£!Uø•vPÆ â nÏú¡‹7’·®Ã…Xë(¯YE1xéXF‡eæªýVʌ裳oÿÓX¨DXyKÄEºS(ìÊØ¹GHg³ëB3Rç5ÿ Sˆïe–ƒv4Û&SIYhßÎòÁÏnòD§y‹Ø,µëöxt÷Y?,ÈÏ¥³$¿Zš•£C(PmׂˆÐx+Ó Ý†/²I'å0b#ÿ¡´3žÙ¢«ÒN8å/°F„mœÃ[¼Þ]T JK»ùJ¢Ïyм®j° £“»w?¶Ì`=ÀèÖtt¯-C<Õ|­•ƒÈ‘ˆ&ep¾qù^ìÕGlPü«Îr‚à©É|w0çÉrCíeéÒŽA ƒÔg‡—KéÞž ÄH±tœµqH4æOt“$ÁSIY,®A/òýŸk‚Ú𬽄J†@ ³þ†)œmE¨&(!L‚×Ù<Ççü)~qMv3ÃÖöÑízGõÎÞ¤ ŽÕÜH[¸4‹iw6¡4;º8ßx‰§6 ü%ó4Q¢jXeL=—ŠwÐgæ~KüQÓðQcQè>ÑX¥û ¤}vÌ™ÎV¨ö¤ˆD—›÷{U/¿„}1îéÖ5 œ ;‘ofÒ\ОÏz*øŸ»³ÝËiXFÄl£liÓì0õzŠÉ‹§ã12c 3ÎùÎ@)Ë5Р>Û JM²ÄÝ ±„LÃ/×0ò)'ËIîLÌ7TWÞ‹Ó?Ö±s8……qpdw° ·Îz8µ>ð94%‰[=¿Uhn+Rh>_¿z­ \°®Qí•ÙÜZ _8·R[ªjB?ünÄ«o¤/‘Ñ¢U<¿ij×Z`ƒß-btEÍb)Í8[ù&S¹©Üššª –%¶ØQ|i‡ñ°iž*Ö®è`&$%ƒ8Pá•’ŒÚ\˜ÚfU0«ó<'%,̧Àr·‚7&¸ið§ç8/1†®©ú' ˜zfÂdÑd endstream endobj 219 0 obj << /Type /FontDescriptor /FontName /EDDSUQ+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/I/L/M/N/P/R/S/T/U/V/W/X/Y/a/ampersand/asciicircum/asciitilde/asterisk/b/bar/braceleft/braceright/bracketleft/bracketright/c/colon/comma/d/dollar/e/eight/equal/exclam/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/nine/numbersign/o/one/p/parenleft/parenright/percent/period/plus/q/quotedbl/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 218 0 R >> endobj 220 0 obj << /Length1 2400 /Length2 15308 /Length3 0 /Length 16700 /Filter /FlateDecode >> stream xÚõTækû gÛ˜êζ5M¶mÞÙ¶›\“›l{&×d7Ù¶mμ÷Þf?ÿï[ë}WkUÇéãÄõ£$URe6s0J8Ø»2²2±ðDåÕÔx,,ìL,,l””jV®¶À‹(5€Î.Vö|ÿ0u»‚dbÆ® ;y{€Œ›-€•ÀÊÅÇÊÍÇÂ`caáý¡ƒ3@ÌØÝÊ Ïq°º PŠ:8z9[YXº‚Òüç_)-€•——›áow€°ÐÙÊÔØ oìj ´e45¶¨:˜Z]½þ'€¥««#3³‡‡“± “ƒ³ÅGZ€‡•«%@ètvšþ" P0¶þ‹%@ÍÒÊå_rUsWcg $°µ2Ú»€<ÜìÍ€ÎPr€ª´@Ñhÿ/c¹0þÝ+ëÃýÛû¯@Vö;›š:Ø9Û{YÙ[Ì­lE 9&WOW€±½Ù_†Æ¶. cwc+[cÁß•$„•Æ ‚ÿ¦çbêlåèêÂäbeûEæ¿Â€º,no&ê`g´wuAø«>1+g )¨í^Ìÿš¬½ƒ‡½Ï¿¹•½™ù_$ÌÜ™Õí­œÜ€Òbÿ6‰þÈ,€®NvVÐ ô4µdþ+¼š—#ðoåßb?GG€9ˆÐÏÊúƒàãbì¸:»ý|þ©ø_„ÀÊ 0³2u˜-¬ìþD‰æÿ á;[ytY@»Ç `ùëç¿ÿéƒÖËÌÁÞÖëùßóeVSW”¦ÿãÿêDD<>Œ\F6vv'€‹—à÷¿AþKÿ?Ôÿ–*[ý»4–?ñ¤íͼÿbjÝX¸ÿ{)hþ}0´€ÿÍ àÚd €æÏâë±p²˜‚~±þ^ÿ¿]þÿmý_Qþßÿÿ$áfkû·šæoýÿÚØÎÊÖëß Evs…¼è4ìÿ¯©&ð_‡,4³r³û¿ZiWcÐqÛ[Øþ·V.Vž@3%+WSË¿×å?CE·µ²*9¸XýõÒYYXþtm¦6 ×Ä4ª¿U@Ð1ýoFq{S³¿®Ž“ `ììlì…ÀZ.6NN€+è<Í€žï5€™ÉÞÁä±ó˜;8#ü5P.N³ð_¢!.³ÈÄ `ýƒxÌb/€Yü¿ˆ›À,ñ±˜%ÿ 6³ÔÄ`–þƒ@ùäþ P>ù?”OáåSü/âåSúƒ@Tþ PÕ?ˆt*ˆ­úÊ®ñ²kþA |ÚÿE¼ ñ誘]@+båbóÇT®É+(¯‰±ó5(žé'ÈØÔÁ4ôÿ:€fÏlöê!ð=¨ÒíÝûÿ—ÞÉ t.\@e™ÿqµÁÜÊý1þR;¸9ÿÃdbñ'"Hoñ×Ç øO Ë?ÄAí³ôr´þ³nÌêÔ)›@Ñ?%r8ÚþµÍô ¶Øý¬ ƒ?±9A±ìAWð=ˆµÃŸr@Îÿ£Qpü£s}ïìmæÇÁúo©óÿô“ÔaGÐDþÌ€ÔG[·T úü3;ý‚šó^±‚:ñǘó/tÿG«8Aæ. wþ(çŸ @O#³«¥3ðpõpø‡¨!nÿ€ æºÿcå@æÿˆÎŠçõê÷n Wo ó¿bÿσbêæ jŽëß/>èµùþû›zM–LùC­B;žê„ =÷'g)÷5Ói}–;Ý^P`Shk³‚7„SFzÑÖvÅiî?­¼ûœ¶~‡ oKRnõ}3LP™ÞoGXšÂœ,:þ6@ÿQíÓﻓ¯F d+x— ež“ŠRæ“G¿¤ç·ŠÕñÏ ûʵ\²ˆo3Œ±ê1zA¥s”ù&Ùóxd0®ŒDptWž¨s÷³¹“¿IdèüÎbÙ‹}t¶Øâžç½×«ÔØ\ºñ)ðuðˆ ï1Ƨ©|DŽRep}ÊJb6sXh™À–ì{ ®DÌ ˜‹IÙ%éýoJL¦Á¸¿ž¨lC~™¹@ܧoS:p6_±+´¼þœ‚Rª7å½PgHÖ>­fÁ†–kèz8¬¥Wî1hUʶp:? &µÒ˱Üô,â¡kÔj¥ZË¥ phøŠðZýÃXúÚ¼¸Dø’i ³'ÌÁp0\i–ýUù¬ü“K¤#ißêb×sýÛÖÀ´L>3€V@gWœÓ3“PÀAï@QA¢þ=DÜ4Yœ®úÍR!á{ÖŒ©ª]½p–wòéa¥˜<]ÿ€‡9MÕÙ8­´½§ò˜lød†ƒG\CùãWþ,‚¦YØœµý‡N]Þ1Â$Òb²"ÃèÙÏ`*Ó³&¸óéÓ·pH#nqže³ò–gsrìJé«çX3Ë^í(œYUX ]dSÜÔ‡mt0çâE¡Põ:à¿gKMÛ¯+~šðÄ:SItïÁm”2×g¿8?Üw„­?õê&ñ‰ÜßBîKëQø§ Î%|? œŒi.ˆKÇYÃzÎî"Ý;Imè©Wîþ®Je(7áµ§5jA×IJ.Â/a†åõ‹ì.ÁÔæÙK¥¹ˆ;y8c)«;sü£úÅú®!dí¹aS-‹ÑÒº%}ª;Žo¡ádÜÄt`¶À“ðÙʂ؊% àP'U¢ì;U ;»EÌ5ûÁ‰‡Û#Twœ£HL–ôx 4J$Ÿ¼S•ÊÖqp¤›ÀºØ•´¸à$Ý´ÌSµŽÇçÓ (ö ôÂ3íú&V4kYa—XþÖâ«þö¸G£ ƒ9/Qe8¯öˆ¸a”}¤¦!*|·ø+e „)übwÛ¥üË–òº‚ñžn„Qàa¿ ¤—­‚§¾{QzL!ÕKº¯yóŠ•r‡ã÷Þi/X;å9g+8ë/c«z?»Õa1DÀÜ*àù\Ž‰Â«¿†yqŸîÀvÄg«Pï«uÂF¶‡MALù˜9g1_™œEy›»{Ý·l Ââ1ʾ[®øç\p «Zf«?Ó¤y§¤F›{“’¿—„¹¯ë¸ul‚éiô©‹ä¾xT?½3ÞûÔ³ 8†ÀiZ¶ì½KÓ§#ªœ wÕ+¥þ.ùÁ¯Õ! àvževÓ_XGGu»¹¯q ¬Á vt/±Vð,"ãêx£Ñ"Ù«‘–úÂqÅÊ{Ÿ#"áM)‘ê[–¾®zÕʈ –{CŸ$Ÿã¢ÏèºÜ¿~M-ÿ¬q0À^f÷õw /nOðU¶ ¦cIý®íB|ýb8 »E^9÷ ߨr^ϧF%ÿÆ3·O³õb—z„.ªå‰ü¿<ùBdjª&‘k¨ãÓÀoÚ¼„ÆÅˆŽÜcÜS…É„]‰q>[_ûQ†Åò‹_0Ç–¿n¯/0O~Zú£@ô¸‘ Þnpã[|û]…tŠß(¶¢o—“Áδú¡Û-A–‡‰ÑÙ4oçdž¢§ 83™0¤ºþ+–ó»[­/%W#D޹Pº®·ë[˜ÄÐ#nG$”K%7FDÔ÷,jùHPÒ>ð$a\Ö‚ùø/ãQKð#Å]+‘pícÜŸñŸ†ÓßÀL®ža<¯ðù¼”`•£Þç¶… ÙJ}ïøJ{“vȯXc^T`§eÏõ8}l3 Y)¢ÓÊê×Òˆ¾Jʹȫ§A:'ÑP•ùñ5?Ä]·½xíǼòEˆú™»®ôð{1P¿F±Qò) >yR½û_ö—qHùs`\ÄÈp—‰oedNõVb‰Ö(Lgõø¹yõúf?Zô¡ƒ;3âd óä¹íßæãßméö‰éñ´î׌<fYâ‰#G¤-õg!êô½v)¾l«ÛÍo¿±·©ðDˆ~8’D¯‘ÏKʘž€ÁKd?kç8ÄúÁu §—L`¾În—Žўƿ[ˆ?a˜ª˜¥Y€µÞ>ç2J e:øbºíéVí`=Ì)K3šbS¸Ô¦nA©\ƾ–£%ýd:v?½iÕÈÎxž P-GÕ`Lg²Tò”@AHùàJq\T„9yt‘£r!ŸC°ËÂÆ~´aPhS#÷ÂWYŠ(GB³2G&×7‡ô›8X0ê1GÕÙú-åñÝ’Ïî·O.øò&®ƒ’ö¶ujp$À*Û€Ó:°¼jª=ñ+ù»d£Róújˆ‡®|ªÞyKÐz)ø}uD¡— WpQï—¡Ý©bOhËùtÄ)€5o“·Å0Ó×áÞæ¶ \’½Œ¢¨‚¢âÕM”daîWl(U¿/4x‹‡¾½-èTžêD€OßÜí¿ƒ™À¶&6§ƒÓÊsËÁ‰† kßwê"X‰f…'¿Íú&;µÌAeá½óq’\tJ€7ºÜÄ>aÊ\ ßm|'” #ù…âh„ Ž)ð–>¨"–ç[ðʱáb=t$MÐïÊ õ¨Sp= Š}ÈZlÿ·ûñpjõí+Âà©ñŒŠý°Xé8a |0ëdÏïî6„Ê(™CW}ÿVÌ1‰5·oàýùŸ1-e¤ÜÀý ú‡¡ÊÚΓJ’c÷f ”órηaë¨á'V²¦$óX¡+©`YH;¯€.g>|RŒšA‡¬õ÷mo¸À Éæ¶íhÈ9è eßÞpZq^^ ®²óâÒ!@¨N·¬ÎIa‘ûpnmõvLÈuøÈyµ+ -ðA®kKlé2ÃÈ ¼Ù=¢Bs€tR5Çü±&·… |¸yÂôs¬'"¯ùÛëü2{EuwÒ[C™Õ×í÷óíT×_ˆ®3_º›ƒ ­D qqðo넞<{VJ¥p|Û/ð»ˆÌ–µú¨LSp¨­U­¹ ÎåcÚ‰ë·Qv?^0s(ÖN“­ý6 Ê –½„Ä0·œ¡ ¥Ù¤"úõIÙC–‚'OøR0}‚È#݉>¡²ï«ñ&£5Æ©HM›K¼?àt}1‚°ôÚü4õ÷ÒÇòüù•¤¹`Ñ5ƒ‡Á:µ:¥(T‹…È­Þ×Uï,kÛ«œ!–¾«|ÅØÁÔ¥¾È²@Qê¨FX!aëˆEN¹Yy¹w~òtìñL›ˆ‚ˆ˜5(† GeÐeù½ñ¼Gy¥`-t½¹(Œ¬sü)À·L¶[éC˜9í&°žQs{";%ûø:ä}„vƒš¢ŠcyoÝlOæ~®aaÕט€`3-{*ÊŠûf”‘÷9I´_†5ƒ9wZ®Vórº¶©ÍHCQQä4ü}ŸŸi)Ï ¸œ¨oYý€Ee‡v% /SLÂM£Pæ-^x;0²aŠ>0x¥Oi&èáRšƒÁ<Ï L|)üëƒ>¥ÚøÇñ÷« ?†Uø"Òº€¦Û"$ø¤t÷t×ý)îÏ·D¸ºøŠÍ…½$õ*PO•ÚMᘟG`ÎîSU³Tç*8çslé?¢üŒë«o±EåªÝW¯Ñn»,§®.{6,p¤„‚½¬{Ó7à³a à™tA³Ï4¸»Aèý ‰O³!Íø\shŸýrÏ6òI9( Ó}ÚoŽÓ›Ýnk)ã†yð¾¬ö[U.þwCÔÈ[ÿ®ã)Ió+0\ɳЄ$Žä`†âêèÁÎ×'Êü•Vý=ñl©(sû‘Õ@ÏèA•ßo^rå˜I¨[ÜpLYÍ»×@èÇÉÎÒÐ/µáá„Æ3’~ûaxG½¨Î ùþ´H•ú)ÐWµlnE„ïî7‹ J˜¾ÂÃo’6EãˆL°JÚՃӳĊüeí"*Yh«ËªŽ5Nêú®Tô§ó(³Æ¸y³DÈöS©Œ¸ƒÛ[ð'øÔ¸ßÛT^Líù/EÚ„ÁFvó|Úh¯Ð~hžo¡Y"¨ñ¼ZN¡q"¡ÒÒ¾al¯æ¦«$¼a¹ôL¢¿píòà·0&cåì××5ÿfË¿u¤{P4.mgóeÞÅcK-–ª²¼Dg¼åV,Lªÿ³k©iœ« Ù¿—óq±I(‹`©&ïûäôëvO/ì¦MÄQØg&Ú¦Ûïù)Þ† ¸ÄBÊÕšÓìla°4¶e»§ÈDQÓ¾AŠ‹&Œ©‡ÓúL©jT²•İ“æêwÅ^·Ä [2õÅzkzå I¡˜÷¿t£Ü÷È^  ÝW×-lH„X(ŽJ#ì`CÛ¿:)Õ¶·5WÒ¤[,IÊRÙõ·¹¶x÷¿U‚W8 .q”zs1ÆrñÚ\š¡º¤|ñ(ÓóDÖÉMäüŽ|ì¥÷5C¥ ‚°`sæìjÒlüÓqàjfØíÚAoãw ¨²Ú«ŸäÑ#„½ÆÔÍ>sÁ_¬¾aúËÚZOïkˆ\\_$y«`ÖbERÍßslùeIÅív[ %y§Qv'l8×ÃÞ/ÿö¼³íßk–ézÁÞ!I%04ü:·˜â7ƒwõd¹Ñؽ÷Â:Rdº½*ðÙï˸DÛáú”&ÅPŽïÙk$Ü’p„´®;uº[‘Ó ù?–~ù)”„§ÆÌ!A¸EU¾¹M‡Û¡" DhåY¶”½§}tÆ©ïâ ᲪÞ)ˆžmQ˜,=Q"E¢ŽcÝSô–0¨åËðEMŠ*®¿iäsSž†T!¸·Ý ƒR ãCQ|<³SV(ÎøÈun¬~°Ž†WØf!z!÷5ø*¿æä×ÉMF˜vQáOIÙ'Ò®ÁÁO–>ûZñúðÖ‘»ý®Q2¥ŸÚä'ïà †R4ç¯èPéuªN7Ì6*¼ÌóËìÜç2õ]áñC­ 9|_Â×OŽÒ”ªÂ¨;ßÒ†\÷Ã¥VR0jSu1Ç¡`ȾœO™•{˜Û»m|ñ@µûÆbG¨‡6·Ÿjô1BMÑÞ¿kQžLk,d÷ÃCSǾXœ9˨ ¯QN†aFü½”Ñ®p˜ð"ºÝLF>ÂÂZu’qâeA*ØQÌ…µ¹ý¸úÒêט>ÑœÌVãB×f昵ÃþƒN°÷ÕµÅÍ%§â J†ËaÁfÏGîõ-¥d¨j¹`¢º€_]¾e(%3A¬¾Üwi"”³¿›ŽÌƒI q5«cÉ̺·¹à”a¥;g²?£ ˆ3m üPͶÊzîIÜÿJ‹9Oߘƒ8N95¥mÄŸUÃI.ë §Úž¦0Oœø`©d¹'vµÒ¤’ðñGhM+êÍñAtèa»|ÐMÉþ|‘*Ï+IiEÃè$bh¥û¤î‡]Ȁљ¬ñAb/çI/y¾™Gì±&¥¶mä\>æïÁktÊøs=> äy>3ç~£BÜ^O˜é–øªÍaH[F®:%dHtø8GÊ¢T².0n¶£s2;h\‹VÎâ Z?6©ÜG‡¥ ¥+Kh¿ ­Ë©àmÓÅÕ6C+÷ZŒhÕ‚¹eh0L%’UP~)'ú"7ZDÆá¾ÖZÜ»\îØ×®3tO °??~ñº©ýéI.*Ë'`Ü}³->6zBN'iŒÞÉJÿ î‚Ù@Žv—8ê{Q"¡×&+àmD¸ìI#fò×Z«FWŸû!êÏál f {™èSή ›ßF“;$ÖªK0¢Š`"$lñåà½"!Í$„ä ‚×éŒ{ÐScˆ [— ú{8çÛöf;SzcHÛ«" ÍJ‘ÙP¢w'2Q¬^Ê<˜É&ùÌ5oÈâÒ àjvY(Ð-B/k5oóÝOì˜Þ~ˆ¬v3›~Êñáë2# ~iÆsŸ|ÒWò…½¢£‚µñå¿Ë· äZû<ƒâ9AÙ6l Å6-c¡ŸŸøDx6®aWÛ\ ^"oéù¸—«DuI3å“KxplñZ0e°Pˆ;üß_4—8æ{^ßFsff3¼OÐuÆœ?±Ö+8— é4)CíQ%rQèZÓIX˜c¡¸n¤XŸW¼i:wך„T£—%KwËœþýõ;ZM Ü׋/ ¼:,î{" ÍwÄš²‚i{Ê úƼh“ÝŒÕò@ßÀB'ÂÂn²~ƒÉñE*[Ì<ßãx— «ßÓÊ‚M$ÝèÇ÷γPû;]0"ü>pg9šÙ–Û°Pc­Þ¬Œ>iÛÆJ`ÔÓì{ Ç`'Î4¤N‚]°¨zÀçï/KÈo ñ)ü¶Ý¦Pü‚¥Ì`£âAä§[aNªŽó„ð„jݰW¸$f&'YañŸˆÄ\©ñ¨×Õ%c'}F?•Åp…'+¦ˆ¨ìK¸ `ËÞ%gv)¶áÄ~&Þ‘è*·ò¸(ˆq¥[ðÕ*¤”ùÅóÄÿìtKò+4†ƒÛÓnÑ:5ýEšŸŠDrãÊûÜ…¾âÖII¦Ô׿†áÍ#²‡'S‡£~·_i—QÄžuc/üužê˜ÁAk‰pWG¿(\¼úóxr¦ƒcs«o%É>³±R§o¯ù]¥3­Fû—w{¯§T=¯†Ï”ÊÞT²i¾†UË4c&Ä-€ÑIèTk4*~l©¯¶½tx|ñ´—ÿ=odÇ‘5±“y‚óíìô5ópñT{Qí\ñ£œ¦0¼† jVªh*4“¦ÇÑà„4g¾¬½Èº3c÷…X{A¯Ê;£.!Œ—F°òú¨xá…C L•+èˆñR*¯ë •%!¤DCéÄ Î2Ú%´µ©ýû€Ôú"š“=ïöì·L4Æý2%~ÙʢΊ—ié©Í;zàu°ë®rO¶±Gj=³aIr½Î` 2?¿!9$COL­^ûe™u_µÒ3u¸ùHÿ†ö¼2WëóÙ2ÊÏu0yh1Îùätz­Žz|~ˆ”º{t$©ƒ9w2çÛ¦LmmîÞ=4Éݶ¥;/A* K~äùØ#±›©t=ý’Rø ÇðÌ[Ssí¿ÛÁ©,Ôvµy›¼ïZÞÝ nš9šì¿h×2]âiÒ×õîê22©ã¿_$Fyqß>’òhW_væÛÿ¬Ö™o˜qôؘ>ÓtÒ•IˆÍLÚü);íé£ëe{ü$uÐËCŽ8  ›½Üq„Cdx+(ŒÁc-â•!#õáøý:p•m~±ŒÃÈ¿åÞyÑÄA¯^lWXm;ëy¹Ì4‡(wÃp¬ÌĨËÓ)@•ÌÉÚß—W¨¹Ô㬭®ñŒÐdfMÛÍr ¢ýðªGÑO«kAל,¤Q6(éJ/ù`Ó®Ûðs—êÀc]ÇŸ=À¹%!lz1ûßűæµáD%I㼸ö¹>|ãflœâ#-½Ïe¸¢Ñù;y ¦<.õ$z}¡‚è<²©*/]ø§ÃTu%—×¶¡T(ÇfUÊ­YÔèœýécò=µŠYé;ìhTæmJ IRß”?àûFþÀk„ë+8ÍÁâEðA Q!!j¤³¿LÃ|ƒÿë…‡š7ŸWzÎC-NJ‚!18 Ø“Câ‘GæIszë÷ÁH>®NYŽÒ&R‹¬ö:E·tÖ‰|KkÓ„D)×±;§âYX]Ó;Åjà¸jI-[;›¼;4¢-ó¯´}L_j‡ wKÄi+Bµ]Šç@-[ª¯Wö”.ŠK~–jo»×PªƒHoÍÈÓ*Ú0E#â¼Ñ§*Åùpɰ »L¥\ƽ#>\QP~8[»ªÐÍ€Ëi¯wg’ÈŸÿ0ÄÊ¿‰5>23˜àæ]BH£›¯Â’ Õ·Ý1­ƒi‹&¤<'CÔ»ÁWìaZg#Ç‘ž‡i³FKžx«Î2ç€sqd SejÀ²*º1·†‹ÿ¾w¦©¢­šžm?;§ñ‚a’4‘· '³%wL);ªn¸O@—{ ‹ ¯°*ENƒñù÷m'{Í'@ÁÈó (E]ü5Û‚'­úí ZXø4hTÌ îÒíd‘ ó®$BÓ¹BÔÅÖü…Ô÷«M(Jša;ùy¤|cp€MñDH/§ùÄmÕ_’ò—í«óO÷4ºÔ—Lá_ÓÖäw–e2x„JoÕp“‡l}‘ê› ¢ÂS~"†àSP-=A4U6+SúôëaHEØv·JNUÍßÙGÀÚlâÁ©¯fxÚHFÓ›Ÿpõ"½Ëïf_a†ã7šR½³o†ñÈHh–Ç®qWWôÏ0YòUÂ@ÆûÆ»5ÅL ý:í9ôeÑO¡³wxÖ*ÃdÁK Îh¢0œUêßÝ©øœøJlÁ8Þ DçÝî—SAüò‰Â#Ò€þ€rtÆS›G-KRÙk„i˜8ùe~ÿD÷3^ÄÖÊûÖÚë0ÃÃòË9]f7@Ëa¸gzôõ4Þ»èè|Ylö&nÅëJ¾ŸɶêR¡çjU²qå,•Óœ{ÏkyâŸ*¼kã|?è&ÐôUik§‹Q˜ž¤X²¯®lÇŒ§ÓåÖ½o„¡ø‹FF‘è^üÄ‹¹8B÷©×Í_>è¬ü½åù•Î ¾cHòÄ(ã‹÷tšmHáí6Ƀãc(l‡?†QS”æà÷Vs„Çañ Ý\¿RÚR· <“޾-óŒ¤/&ÎÙφõW×½L07$šÂÉSLüßïš×µh¦™”ç Í’oÂÔ~^²šn±—C|#¿|/ˆˆ`ÑFןJhÊ{ªn?3šÒº\©‹ÀË/eîógpjõ¨R „1ã()æ•Ôß…õø “â—îõÀ]P~/9^™J³…*ÐÿXCìj‹`j§D¸b@6c¯¥ÓOÀ¶é;$}žC’iáÞh˜T;²—<eò"ï0U©S4Ukò£ÊÀAÛÓZ>Ç[Î\Pzr™“ôù,ì]?Ój[œaЃ¯ÑÅ&,³4(Bð÷£ÍÃQeq5^qh2 Tr(?Dpƒ|g+¶ÜZAk>™ú2ÿöƒxݱ.@¬9«qí…"AÇŽz3"¼RÚóHãìŽImnžX £øu¢® ÂX‡‰˜‘„âZªýãÒ˜j£ -ºg&tÍÍ‘-'DË¥ý9Ë fsªo+± e¤u Å@Yâíã7{LðüÁ¨ª¢³~9®9­Ëþ†žÄ"°Yr|ZF‡Þ8”ŽEÛÛô¼‘â/öVt—}cwË£¹Á8Û­B €ob°‘mj=Ÿ%Ž*Zž¦PÇŽ××–éÇ|¡*­gzc«M†<naãzšæ1óǦÖÉÝ\• l¡–7¯ÏB¡YyH"š88“Ú:¥­¨{ l·œ^xœÖF¼Ã¬ví©!lé>,ÙÌo]§× y0*ØD–è'¤‘ ž15=Évž¥íį±œ;0™~Â’¾Êû"¤jªÝIð–®Yf§Ÿ:¦33}‚|‰œÇöá‘»Õ(Ÿµ½Áâ‰7äQ÷ƒò­K_K„l«Å‘H‹Ô>æŽ 8ÕÊ"u¬s”ö¢ìgáH rŽÎáh<¶æ>Ú¢¡Õàùy-CP²Ãü(ßBBxç9A@êçî–æz–mFavÇúSNG F"ó}Ñ È˜ro&¿ø±Œ$ý¬‰è<Áöü1UˆŠ­³2ìÉQ4 A~Þ]Ô`¹5ñ7‹áÌ£˜……ü>'Ø›ÑÊŽ˜ —Á;çÅZgÂ'Ø‚êÉý/é0Ôé=Ÿ£õÀÕŠ€ëÎu½Ø1¹‘ì¹NòŠ}Tºª·ð-ž9ä™çE? ä¦x7ÊFªCÜs[›‹.÷í«¤w åK5S‘µWþi¤Sµ; Ûå*X·ñSF"Š`×¢N©ìïÃ[FlÙ5—°;Þßé ¦Ï0š+Ç­Ÿ Í*‰áÝôäÔaƒP6Ù›˜¾˜u|ÖQ™ ;Q°2“³,ñC%Pî/À;ý^e£^¥Â˜ÈWn¸ªMæG7$ìh|žŠ¨Ro“ç§…-,O†æ†Ë×õånÁ«E‹<¾Ño„#·oµŸ?ÉŽ4Ô®uÛü ÃUÞX­ V¦`ÁÝvøhã&-.¯D CÉÝÒÛ ¼é!´ƒKê÷e‘â†òž8xÝLam6t ø’V4¸yÀÊ#xU.î…9ð¾î­„¢¼¡!WHÛªÂÕj(nû²¸?; Æ9fI!>ÅîÏL5H?í,ôy›p¼ûùì°ÊÏIš­6'êüj_\~Šº›8|p|‹· Òsdœùe×£UkVÑ9¾P£¨“¢òõ‚)V0 ¨ùÜB-"/ZÈÂô»•/9Z t®Þò"„¦vˆ3×;÷ƒªqLy¶É¸«8{A7lή¬YÜãÑN»ŒÚ48¯ÄŒ³×CŠÑþ:o¨Ïƒ%ei£–“r2kÙþìBNÞ&iå ¬¯²…š{Tø“=@«S7œÿÆópôòQ ~5ó±…«ž8Ö½å¤ÈÊèËÖF¦C<ôÜù¤­òZ¯™ÍúâTýgê&°u¨xë_ ¯ÄV;Ú?ý›®Ccy©ŽpÙ>úßš ¡`aDR,«™86ŽºaçÝâ}3Ç"ÐrÒx—¢ðÞu„¥,ë~碰_]Ö—.¦‚F8š8qðyrË·eœÉà¢ÑåÊ?E7a(5&g^“47q „ÛmŸŸ)½\¨§‘a˜&ê×ÉNÎ6€®y°g¨ÖpN¿‹H•ãªG%-W «ùúà—TRÏÈ$ü“[E¡)sVkÑ º:ܤú§ßEG~†Q^ ï=a9i2Ä&üœöõ0ÑX¹G#¨öu*º‹É#úo–’ßêÅÔ6/û.¶j$HcŸ=2ÜŒµ/ÙzΞ˜â£ŠIt Œð£ÜȈ,½ùÙ.í±™™;ŒÎ»;’ÐzL´0’ Âke~èò9iw½ú š­ïàµRìF;NŠèH21×^]4Ãö—bñ)Hå‘eî5²} Æ$ˆÛ ÿhiž‹ší¤JÇtÞy\êÙ…€`IÚÊNò(ÈŽÑ£ƒ>[ùˆó‘1ÐänQ²3×Ï8EBL½2P£J¹÷KqŠ× çÂV8L×`¼„.¯0ÌψÇßÃði£˜.çW˜X5L(ÃçF,g5"ß—Çü$C´uÃ|çV‘ÓpDå1#›u›Õ< €Õ…×öR^ §•Ò’\Ï;Û®ßûÎvkÀ7G¨]ƒCû½>*îÅÀŠüîqÉÙºl&Ô Nºh´Šÿ‚ÄeÇѰ*ýf?˜è¤'‡ªœ©Ü({þ.Zíħ‘O'Ô¯.ƒeðsʨM Ò]Mrõ,ômŒ#†‹Çc‹«×ÜóêX•P£l*©52«¡±Y=¿yÁøhKG­²]É!ø¢¤dq‘ŠYqàä’çÍõÂÿµƒ¹Hý·DóèS¾¶×+øO’-UþÛÕbþ{‘íVyE:7Áèžòi¥²ÀÈŽ’/ <Ê„ LG©dÜ{1§¢ã²K«X¥viéUFÇGjÈ~‹Ô,ÙéwB¦±ÁiæfcM{¿ñ©)ýr«‹‰~ž¯ê(½±wxï\ùwެOR\<àeÖOåìiÙÒLtUªŒª‚IC¶%v!]Í¥&‹òîÖ™&Ž{þA<âª<¬'Q«¸±1°lT"ÌÓâ¬gd¢OP(Ãx‚\s®a˺!Ðt¼’(Þô007òø!ÊÅüWÔ($¸%´‡°‹ã¢,þÕßßçè«ãÖ! ˆ³ÍpLá(H‹ ÓU ¸©vVHɰúaLWHzT©Ñô+PeD´`÷F1¨~ ') 2{R,Ê(ÅU‡­`Ø.®`v/D¥5Ø o¯ ]-•èÙÉæ¤üi–…„ft”ù‹^qÝŸaÞOCo,ÜU|//ú{ñtÜxx´·ÙglWùÙÁ´£GÁï•×*½)ª…=™{ë¤"f×R…\¼EP§©¹ˆjÃÞÖÊ< \¹2¯’fSƒaÜR‚Ì3ö˜«ñÝ#Ì´wK6yÑãÉû3‘±6B6 œ/Ÿë½Ì¥ª†– ãóëjÚÝ 81º Ô)Ê1õ¢=çx¥·¼‹\T1â7žq™ÖÏ=?´û—˜Þ$×z9å%ŒÔ}FíçÈJÍÜzUŠqtAu=ÒÝV&«ÓΧF¤aOíc¼¨¦ ±ÿµ3+ ¡7Ê=ÿó ÏyM´qh0í£÷¯N êÍ´¶TÓ+¥uü4›p`\l á˩۸ùƒBwS\ÇKN¹¦(5³®g8S1Wiù+9ÀLÙZà´™f¢Ù“·,"K½èfÕì^ŠýÇ7 =¾=Úò#ö?È=lÍ3<ª:w˜€¨Fíö˜‡²½š´0à+1.DŸŸ+´KÍQ*ñp wÁ¡ð¥²Ýâ:ØRpgyŒê9ê‘#ºh7 Ÿ‰Y6ûlíœ]σ²5îV–ò¶s5qß–p}ÕYLUUí`#(e˜(òÁ–‘´}'l…‡¸‰¾?3hçK¥®9\;‹“ ö’ˆ9Ë(P%nï”­»OÆ?Ã×)‡ F‹7£—C«ÐGpÞ5¥¶¾ŒßœÖ~Â÷ƒf —+È¢´ÕA+‘µ*ê€×‹zH¥Øû`æ=²™ =¥­I$nßú.öÔK&¦KŠ4ß”ûšjdv»jÿqš 7ý·ʺÜÆK*»!—öÛ'Ý%6Qj¾´$ôêc/®xßåtRc™‚d=çSI=(Ó¹â#poÓÀ‚ί˯Ya†Oâç° ÐôU?øÀ”Z·ßgLù”‘F^b•"Bfô$Å{ºv–|ªÓìûdÄÍ6},BÈß×òGXM¸O?ÁZ%ÜO»/É– Ƹs#…'³õ7¹¦©zkªuÊ dÞ… ÈÁêEdEóЊ©GzÝé]åf2b†·¶&uŸ~¨QˆJû%È}`˜3†P1MºUM™8hÑEïYƒý|̦{“αBò n0Ìdô)·ˆpl½ù—cÇb°ŠäLÿÌ+g8^;®¼úTÜÑÃô;3àA8ÅÛk0l¬+IûµoiTrÁ'~Ð#i²«—|D&´ž+þ/¥[±†ÇgÞm4Ç¥è[ì2–ß½:Š&‘üJ†s0yb¶ªàªíöæ¾ÞŠ;L¤êÍ 9§ßrQ+é£|³‘Ð5Ódc !¾ BSk]ÇŠÍÌñ&+-fÄ"@>;9(OBH¿»ìÃñ^8h«}z÷ ¤&ÏáPÒ•,`$ü½bïg¸XøCv ý,Gɽ‹œ{/²NÀïí;MOU—À¿ù»¬Ôà¨,|¦z¯xí è°×ïÆÝeõfÁå(å–º±¬.oÒö2LvýØ5…ÏãÓ›O«¤ømx^/uQ’»­Ð6û¤š¦&ˆ“ؽE4ÄX’5úáu–È?wjA–Lôx‰£ØÝ[lÇ‰øÅv«J甚˜ßš ;­qñþ,eÒ2š¨uל1þ„ån!Û)^Èî&æî'ÆguÛ"ðEVîqäcçn›zSµ°0‡Ð‘á&òŸ1°<ó+CÔéH.®Êõ“Ä„‹J=±4·¯\Ÿx¹×À?B™Þ²— wk‡}ôDµC ^=¤³ô4á?FO!¢k>7©sÝ–çmãÕÈ\ÒÌ㞟ÏOüFœ‘-²oÑõü PYµ0‹ý­x”=æc©^µH¥oíp¿JÒMÅšþ:‚žÑòh3¡NT)"$ $¨}™mÎE@¹ÀÙ–qÁÜyukPŠi™xvnaåûìE\ˆ©JÙc¹¾2.¨˜XCeà{¢1­Ý@üÁиÝdV™bþ<¦¶a±.c3·2 0 7yA›“BM!²”u¢ì–þWØÜÝÁ®’PkýQ³ˆß,²ÏYXǘQÞ~d_ЙTòå7&’öWc¨žÅa’G'D½9À(”÷IìeÔŠÖ?Ò»~Òi8N°@µèÔXo…åÎÝ« ƒu(,Я[”+RôDghU&ëï›’Iá(¥àzÚ1€ ‚Hüv’ÆxŸí'°ù؃Š$t3ŒN6ÀÅû W™Bø•ÿ:a¾'.ìg9}‘…ŸÕÂ$ØïÒÍþ.R³ŠgêUŒTµ$ÿÚ3âFÈ1È~¼…°™-°é[H ?sÍ\É Š=¶ S‰r^Vô‘gV{6eYC’5jâÉãÊ& ØáS½Vá½-ïÌkÆú {ˆì^¢41°²Þ· ž1´÷·L2F–Dî66Õ¹g½Åà%Áîå­]Œ×“ݧ„ww=¹/ÅYîÞvüÍÎSeÊ–W1ç¿beoû#gFùm¡Æ·t÷ÙY“E² ÞC©m÷ò•–Ý¿âgî¢A c…§ÂŽñëkeþ¼’¸Þïp'ÐQùÅ)ýK”9í¢¦cÏ×emN>Ò%>ÒçKQqEMnÝ[ï‡ô¨òU$Ì-øûRÓ‚ÝÑ)Èg¨ITÞ™.úž¹çŠªøÎ'åߘyþÚ5°ýˆøž…_VÏïZs8Â#1pÍ]5‚ðï¿|q î;¶ Úš53@äÑiã"eZNq~¬,qh#1Ÿ®¬’)¦Øx:r.Û°GàÏ­1eåü¡jã®ß¨QðSW+c‹kÎÓEJâ ¨ Ž‚·”E1JûCå7ï6ÒÀÅnƒ(–m™NÚœÚ4„7ˆà #\ìÎ#º¹ ¶)/áDè0«Í šl¹Úù¦/Ã7wùÀÚ+ßq³ÔÈ[éx‘¤Ð>‹6^¦0TåZ©Õ¶—ØA¾ÞE Æe¾ô~Lz®íøí½+AÏ7¾Ù½‘Á\W=‚RFk¤yÒ%|u9¾&ôPØW¯ŠÔS…ÿ!ÕJ1Ù:W}4¬ò…v7ÿ›¼´«ùL4Ï“é±}røg Œà:m,ƒ.ßB¤V ¦}Áê £ÏÚ‘þ*_gÎŒQÑ88À ƒ; Ð~f~È<'kBÛ/ˆÇ òÑò³Š®Ÿé7â£[DÆ‹¡¾™U€±–ÑØ]dû—}Bó‹¬gw=ñw>^Ü›ˆJ^™gù¿óò¨ó`§çÇKäv +ø"D:ë4t+h© (”¢óÛ¶±g7òÝS#5 åÞ xPYAã|¨wó^m€|ÛW„þ§¢rÿç„/­8ûT8¸ÚºwäÖ¦ø\¸È*c6¢*8”41Û."=ŽÎ‡…Õp?r˾à,Ï@èåêw–TBz=,K=1GÌZ-_}»¢31N°{Ët wÅ÷ü-ÅhŸ šÙ*´n+´±#ovà"‚¸+v§Ò,¶£òËhF^eSߘ jsÄÚ 1°©Ä8r*Gð*øÎzÕJtuåk×ùÏìúôNWí°~¡”ýpL0*³QYgü3ò›{óœæqÕEÊu?ê!“Ã%ÂÆKÅìãe³-F© Ço!#Þù‡òÆï´\Š*pldÛ¦?(YØ\Ë)i:.ñ?Ÿ c±ì?( ŒáQÙ©~›¿¬ÁÀBhŽpŒÅИ13ýŠY^·\¤ ÂW¹å6ÊúñÀЬå~ëf!ñ@1¸h\¼é¼gq•PW—ý¤@«¡›?ÄõÈÑŸO0ÄtéI£G0Žwæ§iÛRnVç.ôeljfIqvˆ ´GÞ»®¹­=IGÓ".ø€…%dJzB™8†3UB{suAæèxKKéá-q¦h…öA›ÔaÆ×<™š$W=Ó±âõ¼jÍGê1çÓÊ×ã  Å yet+Šh×|íuþsÃ[Àd¢Cüªêºv–ØAH&qyr-~µÍˆßØOôíÍ¡c‹‡Ñ·‡¶«c’ZYÍËÄj.¤ñôÛ“da{=“ðS4…$†Ç¹}ð:û_jh|U]YB$¿¥>W{brGlصgÓË bfÐ/ªÄ^A,‡<áCÅ®ò[›|CˆýB}9Äê‹Jn1qþ¥k0`¦*‚ÕÏ]À­µ-¨^z~QÓ£êih„Ü_&„¹GTWÞ¨¹”æ¦]àrŸ—ï(ÿG—ÒÕ3ܸ'Gø¶™29+¶¤ìDÃü`s\Mã ™}C«µ—"­y(w’ÖC>xb’RùÕóAuP¤*<) ä7.õ)O¼$ó ÝÀª È3 ¤ß´­»mã‰Ë{BËù°6ÒûtwûÈÆ -–àB­¤$ËióÊk]º¾¸ôôFù:T@Ï0’عÇ2™Ö§¤&2»²—™Mw¨«&©0 ¼¬à¢¶¹UERljÉu˜1$ÜËXúÉ´œ÷#aY¥|r¯KýÿåHŠÓq§ÃÞq1ìgXÃÒæÆÄlcÚ‚…ëŒ ǽ=Ú%ßQð=qÕUßb™òóKÓ­ÊE|Sáé–O„Xi"Ž€UÝîsÚ¿ûYjU<û&˜Ÿ6ÁÿÒø`À×ìòèßô9T¶ºãsáDL³¾ÎÓƒ{Ó+4¯*íärŒËаwÔÀ!ç!iû½™áC rTèåÜ7œ˜úŽ!*pù†6k$͈Âý¬I,ö|Ƙ+uÅǨIén,ŸÓo޾ÛÒÎÁ’ûìóLž·z?Ò)‘=³€õ|aª×Õÿˆ§í¦\#ÌJÑþ·½žÊFcÏ;Ž:%º£¬8O&CÖîs/ÄIr&|£›ÔJyÖåŒ~Ï9[Ü•*[»–Ü,†ãlöfmȧ)\k¼Çó´‰zv*×ÒÀœ}ãÂ|)ïß©ìöáçVù6 ÎieNëè/DžRY…B7?Ûš¦noww b{ѨÀ Îõ·Û–!c7ü…R ÁÄû{ îוÞ)kõ«vB7¼Rœ+)}æÝïGiõ°z˜7*I˜3Éä#ðhVgƒí%_ü꯻œ›×Üel9 ÊšGö33iyÈͤ9ÚtÃήPI|é9ÍšhƒØŠÐÀЉÍz•u]Pä¢8ñ“š’Òºø´AgU{ä€Y×”•%2~_ý¹½twî@¿ªO»*‰$=`çìâi"zGªh‡Nçtãá­øy•£;Ý©Î"5çKÏ“{¦ñOb¼Ã³û×ÎuÑBSo¯Ø 9ÞÈa´2èVÈa%‰¸ nбx«ý(QüKÓhö‹Æ;_ú¨h_&ø“Å–)‰°ÖÍ|·µiÆG†”£Þg¢Ö,oÀ•c{Ä:{Ãü³o‹ÆõáU³fâ&C‘C`X 4éõ÷u8Êõty~5ÁÍž»îu•³"G¿:x´Ž„‰bˆ\ ­Äy 4}?aYËÀ°X­ÑeüT=0Å–£ o…*ñ7yÓ™^€¢9õ³Ï2e§Ä»zbš|ßt´'ɺ“Æ2=éUÄ„û3Ã_E×ï~Œ»OÁÝJiÜ¿•=8¹7 wiºò_ÐB†B¡EÍ~Âÿݯ„vqºM‘åäÏwð{}|[e̵0dÀq µBת1Áö8Ó¹ÌoË!k…û¬@û»â©Ë—Ë–8a–ÁpyʺŽo®œ¹Š·iÊ#›UA¼‡?KÀ¹çO˜ë>~¨†w¶N 0úýë×þYùÏgÒ¢ùÔN!`le1jJ˜ÛÎåJjÙϺfÌtBH“ý2„:¹PÌd›Y®Ž^“d+²`É öY]‘òÇ b‹å„×c׉å<ƒÒÞ÷{˜j÷[.¹As8g<®ª¬>(r}Z6ADð'æ?g1!¾³G¹³oŒ]õ+U;Ѽ†° ðzwUWâ|„’·/9˿汜n$ÓÇ_~=p—æ_²E;Ñ">ñòm@9ÞL`Œ ð2o¨:ÃH.@ÉH endstream endobj 221 0 obj << /Type /FontDescriptor /FontName /TOEOFI+CMTT9 /Flags 4 /FontBBox [-6 -233 542 698] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 74 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/L/M/N/O/P/R/S/T/U/V/W/Y/a/asterisk/b/bar/c/colon/d/e/eight/equal/f/five/four/g/greater/h/hyphen/i/k/l/less/m/n/nine/o/one/p/parenleft/parenright/period/plus/q/r/s/seven/six/t/three/two/u/v/x/y/z/zero) /FontFile 220 0 R >> endobj 8 0 obj << /Type /Font /Subtype /Type1 /BaseFont /LQVOMS+CMBX10 /FontDescriptor 187 0 R /FirstChar 49 /LastChar 121 /Widths 181 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /LFFNKM+CMBX12 /FontDescriptor 189 0 R /FirstChar 11 /LastChar 121 /Widths 182 0 R >> endobj 62 0 obj << /Type /Font /Subtype /Type1 /BaseFont /IXNPPI+CMEX10 /FontDescriptor 191 0 R /FirstChar 80 /LastChar 88 /Widths 162 0 R >> endobj 15 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RQBVPW+CMMI10 /FontDescriptor 193 0 R /FirstChar 11 /LastChar 121 /Widths 178 0 R >> endobj 55 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GYJBEK+CMMI7 /FontDescriptor 195 0 R /FirstChar 102 /LastChar 109 /Widths 163 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /IYMWNI+CMR10 /FontDescriptor 197 0 R /FirstChar 11 /LastChar 124 /Widths 180 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /CJOFPZ+CMR12 /FontDescriptor 199 0 R /FirstChar 44 /LastChar 121 /Widths 184 0 R >> endobj 4 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YVMLIC+CMR17 /FontDescriptor 201 0 R /FirstChar 34 /LastChar 121 /Widths 185 0 R >> endobj 23 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XTJZVU+CMR6 /FontDescriptor 203 0 R /FirstChar 49 /LastChar 50 /Widths 171 0 R >> endobj 21 0 obj << /Type /Font /Subtype /Type1 /BaseFont /PVBVDE+CMR7 /FontDescriptor 205 0 R /FirstChar 43 /LastChar 61 /Widths 173 0 R >> endobj 24 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FUNZJI+CMR8 /FontDescriptor 207 0 R /FirstChar 12 /LastChar 121 /Widths 170 0 R >> endobj 54 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZEXHDB+CMSY10 /FontDescriptor 209 0 R /FirstChar 0 /LastChar 56 /Widths 164 0 R >> endobj 72 0 obj << /Type /Font /Subtype /Type1 /BaseFont /EXPITJ+CMSY5 /FontDescriptor 211 0 R /FirstChar 48 /LastChar 48 /Widths 161 0 R >> endobj 53 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZQLTBN+CMSY7 /FontDescriptor 213 0 R /FirstChar 0 /LastChar 48 /Widths 165 0 R >> endobj 14 0 obj << /Type /Font /Subtype /Type1 /BaseFont /UFCSGE+CMTI10 /FontDescriptor 215 0 R /FirstChar 12 /LastChar 121 /Widths 179 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZNPDQI+CMTI12 /FontDescriptor 217 0 R /FirstChar 67 /LastChar 121 /Widths 183 0 R >> endobj 22 0 obj << /Type /Font /Subtype /Type1 /BaseFont /EDDSUQ+CMTT10 /FontDescriptor 219 0 R /FirstChar 33 /LastChar 126 /Widths 172 0 R >> endobj 134 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TOEOFI+CMTT9 /FontDescriptor 221 0 R /FirstChar 40 /LastChar 124 /Widths 160 0 R >> endobj 10 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [2 0 R 12 0 R 19 0 R 26 0 R 34 0 R 39 0 R] >> endobj 44 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [42 0 R 51 0 R 57 0 R 60 0 R 64 0 R 67 0 R] >> endobj 73 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [70 0 R 75 0 R 78 0 R 81 0 R 84 0 R 87 0 R] >> endobj 92 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [90 0 R 94 0 R 97 0 R 100 0 R 103 0 R 106 0 R] >> endobj 111 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [109 0 R 113 0 R 116 0 R 119 0 R 122 0 R 125 0 R] >> endobj 130 0 obj << /Type /Pages /Count 6 /Parent 222 0 R /Kids [128 0 R 132 0 R 136 0 R 139 0 R 142 0 R 145 0 R] >> endobj 150 0 obj << /Type /Pages /Count 4 /Parent 223 0 R /Kids [148 0 R 152 0 R 155 0 R 158 0 R] >> endobj 222 0 obj << /Type /Pages /Count 36 /Parent 224 0 R /Kids [10 0 R 44 0 R 73 0 R 92 0 R 111 0 R 130 0 R] >> endobj 223 0 obj << /Type /Pages /Count 4 /Parent 224 0 R /Kids [150 0 R] >> endobj 224 0 obj << /Type /Pages /Count 40 /Kids [222 0 R 223 0 R] >> endobj 225 0 obj << /Type /Catalog /Pages 224 0 R >> endobj 226 0 obj << /Producer (pdfTeX-1.40.14) /Creator (TeX) /CreationDate (D:20150701163333-05'00') /ModDate (D:20150701163333-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 227 0000000000 65535 f 0000001363 00000 n 0000001258 00000 n 0000000015 00000 n 0000343970 00000 n 0000343829 00000 n 0000345097 00000 n 0000343118 00000 n 0000342976 00000 n 0000343688 00000 n 0000345525 00000 n 0000005517 00000 n 0000005409 00000 n 0000001485 00000 n 0000344954 00000 n 0000343402 00000 n 0000112125 00000 n 0000010336 00000 n 0000009702 00000 n 0000009594 00000 n 0000005632 00000 n 0000344251 00000 n 0000345239 00000 n 0000344111 00000 n 0000344391 00000 n 0000015024 00000 n 0000010228 00000 n 0000009852 00000 n 0000011809 00000 n 0000011952 00000 n 0000012050 00000 n 0000012085 00000 n 0000012327 00000 n 0000018011 00000 n 0000017903 00000 n 0000015119 00000 n 0000110290 00000 n 0000021918 00000 n 0000021304 00000 n 0000021196 00000 n 0000018137 00000 n 0000027048 00000 n 0000021810 00000 n 0000021383 00000 n 0000345634 00000 n 0000023833 00000 n 0000023976 00000 n 0000024074 00000 n 0000024109 00000 n 0000024351 00000 n 0000030946 00000 n 0000030838 00000 n 0000027143 00000 n 0000344814 00000 n 0000344532 00000 n 0000343545 00000 n 0000034712 00000 n 0000034604 00000 n 0000031109 00000 n 0000038175 00000 n 0000038067 00000 n 0000034862 00000 n 0000343260 00000 n 0000040755 00000 n 0000040647 00000 n 0000038349 00000 n 0000043293 00000 n 0000043185 00000 n 0000040835 00000 n 0000046985 00000 n 0000046877 00000 n 0000043431 00000 n 0000344673 00000 n 0000345744 00000 n 0000049513 00000 n 0000049405 00000 n 0000047147 00000 n 0000052417 00000 n 0000052309 00000 n 0000049617 00000 n 0000055666 00000 n 0000055558 00000 n 0000052556 00000 n 0000059380 00000 n 0000059272 00000 n 0000055792 00000 n 0000063483 00000 n 0000063375 00000 n 0000059519 00000 n 0000066867 00000 n 0000066759 00000 n 0000063562 00000 n 0000345854 00000 n 0000069431 00000 n 0000069323 00000 n 0000066958 00000 n 0000070828 00000 n 0000070720 00000 n 0000069535 00000 n 0000073379 00000 n 0000073269 00000 n 0000070908 00000 n 0000075563 00000 n 0000075452 00000 n 0000073482 00000 n 0000079427 00000 n 0000079316 00000 n 0000075656 00000 n 0000082579 00000 n 0000082467 00000 n 0000079543 00000 n 0000345967 00000 n 0000085299 00000 n 0000085187 00000 n 0000082671 00000 n 0000086874 00000 n 0000086762 00000 n 0000085474 00000 n 0000088356 00000 n 0000088244 00000 n 0000086955 00000 n 0000090418 00000 n 0000090306 00000 n 0000088448 00000 n 0000094178 00000 n 0000094066 00000 n 0000090534 00000 n 0000096206 00000 n 0000096094 00000 n 0000094330 00000 n 0000346084 00000 n 0000097881 00000 n 0000097769 00000 n 0000096334 00000 n 0000345382 00000 n 0000098797 00000 n 0000098685 00000 n 0000097987 00000 n 0000100190 00000 n 0000100078 00000 n 0000098879 00000 n 0000101332 00000 n 0000101220 00000 n 0000100272 00000 n 0000102457 00000 n 0000102345 00000 n 0000101414 00000 n 0000103984 00000 n 0000103872 00000 n 0000102539 00000 n 0000346201 00000 n 0000105087 00000 n 0000104975 00000 n 0000104066 00000 n 0000107247 00000 n 0000107135 00000 n 0000105169 00000 n 0000108848 00000 n 0000108736 00000 n 0000107352 00000 n 0000108929 00000 n 0000109288 00000 n 0000109311 00000 n 0000109386 00000 n 0000109454 00000 n 0000109779 00000 n 0000110105 00000 n 0000110538 00000 n 0000110564 00000 n 0000110625 00000 n 0000110661 00000 n 0000111340 00000 n 0000111371 00000 n 0000111766 00000 n 0000111895 00000 n 0000112373 00000 n 0000112399 00000 n 0000112462 00000 n 0000112499 00000 n 0000113134 00000 n 0000113791 00000 n 0000114423 00000 n 0000114848 00000 n 0000115491 00000 n 0000115768 00000 n 0000116198 00000 n 0000116741 00000 n 0000132396 00000 n 0000132713 00000 n 0000148546 00000 n 0000148888 00000 n 0000156096 00000 n 0000156343 00000 n 0000168904 00000 n 0000169211 00000 n 0000177432 00000 n 0000177660 00000 n 0000203530 00000 n 0000204126 00000 n 0000213809 00000 n 0000214071 00000 n 0000224657 00000 n 0000224940 00000 n 0000232101 00000 n 0000232325 00000 n 0000240830 00000 n 0000241095 00000 n 0000255884 00000 n 0000256211 00000 n 0000263758 00000 n 0000264021 00000 n 0000271020 00000 n 0000271244 00000 n 0000278315 00000 n 0000278546 00000 n 0000294546 00000 n 0000294845 00000 n 0000303507 00000 n 0000303745 00000 n 0000325132 00000 n 0000325722 00000 n 0000342543 00000 n 0000346302 00000 n 0000346416 00000 n 0000346493 00000 n 0000346563 00000 n 0000346616 00000 n trailer << /Size 227 /Root 225 0 R /Info 226 0 R /ID [ ] >> startxref 346883 %%EOF survival/inst/doc/timedep.R0000644000175100001440000001355612545056257015464 0ustar hornikusers### R code from vignette source 'timedep.Rnw' ################################################### ### code chunk number 1: preamble ################################################### options(width=60, continue=" ") makefig <- function(file, top=1, right=1, left=4) { pdf(file, width=9.5, height=7, pointsize=18) par(mar=c(4, left, top, right) +.1) } library(survival) ################################################### ### code chunk number 2: fake ################################################### getOption("SweaveHooks")[["fig"]]() set.seed(1953) # a good year nvisit <- floor(pmin(lung$time/30.5, 12)) response <- rbinom(nrow(lung), nvisit, .05) > 0 badfit <- survfit(Surv(time/365.25, status) ~ response, data=lung) plot(badfit, mark.time=FALSE, lty=1:2, xlab="Years post diagnosis", ylab="Survival") legend(1.5, .85, c("Responders", "Non-responders"), lty=2:1, bty='n') ################################################### ### code chunk number 3: timedep.Rnw:152-154 (eval = FALSE) ################################################### ## fit <- coxph(Surv(time1, time2, status) ~ age + creatinine, ## data=mydata) ################################################### ### code chunk number 4: rep (eval = FALSE) ################################################### ## newd <- tmerge(data1=base, data2=timeline, id=repid, tstart=age1, ## tstop=age2, options(id="repid")) ## newd <- tmerge(newd, outcome, id=repid, mtype = cumevent(age)) ## newd <- with(subset(outcome, event='diabetes'), ## tmerge(newd, id=repid, diabetes= tdc(age))) ## newd <- with(subset(outcome, event='arthritis'), ## tmerge(newd, id=repid, event =tdc(age))) ################################################### ### code chunk number 5: cgd1 ################################################### newcgd <- tmerge(cgd0[, 1:13], cgd0, id=id, tstop=futime) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime1)) newcgd <- with(cgd0, tmerge(newcgd, id=id, infect = event(etime2))) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime3)) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime4), infect= event(etime5), infect=event(etime6), infect= event(etime7)) attr(newcgd, "tcount") newcgd <- tmerge(newcgd, newcgd, id, enum=cumtdc(tstart)) all.equal(newcgd[, c("id", "tstart", "tstop", "infect")], cgd [, c("id", "tstart", "tstop", "status")], check.attributes=FALSE) ################################################### ### code chunk number 6: stanford ################################################### tdata <- jasa[, -(1:4)] #leave off the dates, temporary data set tdata$futime <- pmax(.5, tdata$futime) # the death on day 0 indx <- with(tdata, which(wait.time == futime)) tdata$wait.time[indx] <- tdata$wait.time[indx] - .5 #the tied transplant sdata <- tmerge(tdata, tdata, id=1:nrow(tdata), death = event(futime, fustat), trans = tdc(wait.time)) attr(sdata, "tcount") coxph(Surv(tstart, tstop, death) ~ age + trans, sdata) ################################################### ### code chunk number 7: pbc ################################################### temp <- subset(pbc, id <= 312, select=c(id:sex, stage)) pbc2 <- tmerge(temp, temp, id=id, status = event(time, status)) pbc2 <- tmerge(pbc2, pbcseq, id=id, ascites = tdc(day, ascites), bili = tdc(day, bili), albumin = tdc(day, albumin), protime = tdc(day, protime), alkphos = tdc(day, alk.phos)) coef(coxph(Surv(time, status==2) ~ log(bili) + log(protime), pbc)) coef(coxph(Surv(tstart, tstop, status==2) ~ log(bili) + log(protime), pbc2)) ################################################### ### code chunk number 8: timedep.Rnw:467-468 ################################################### attr(pbc2, "tcount") ################################################### ### code chunk number 9: veteran1 ################################################### getOption("SweaveHooks")[["fig"]]() options(show.signif.stars = FALSE) # display intelligence vfit <- coxph(Surv(time, status) ~ trt + prior + karno, veteran) vfit quantile(veteran$karno) zp <- cox.zph(vfit, transform= function(time) log(time +20)) zp plot(zp[3]) abline(0,0, col=2) ################################################### ### code chunk number 10: vfit2 (eval = FALSE) ################################################### ## vfit2 <- coxph(Surv(time, status) ~ trt + prior + karno + ## I(karno * log(time + 20)), data=veteran) ################################################### ### code chunk number 11: vet3 ################################################### vfit3 <- coxph(Surv(time, status) ~ trt + prior + karno + tt(karno), data=veteran, tt = function(x, t, ...) x * log(t+20)) vfit3 ################################################### ### code chunk number 12: pbctime ################################################### pfit1 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + age, pbc) pfit2 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + tt(age), data=pbc, tt=function(x, t, ...) { age <- x + t/365.25 cbind(age=age, age2= (age-50)^2, age3= (age-50)^3) }) pfit2 anova(pfit2) # anova(pfit1, pfit2) #this fails 2*(pfit2$loglik - pfit1$loglik)[2] ################################################### ### code chunk number 13: timedep.Rnw:669-676 ################################################### function(x, t, riskset, weights){ obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } ################################################### ### code chunk number 14: timedep.Rnw:686-688 ################################################### function(x, t, riskset, weights) unlist(tapply(x, riskset, rank)) survival/inst/doc/adjcurve.pdf0000644000175100001440000171464312545056257016216 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 3 0 obj << /Length 3286 /Filter /FlateDecode >> stream xÚ…ZIÛʾ¿_!ä °h.Í-7çáxrzAœERK,‰ IÍx‚üøÔÚÝ”(û0£f/ÕÕÕµ|Uä__~ùø9-Vq&qfV/»Uíábìþ&Ó‘ÄØSޏŠÓ,Ìò •†i‘ñ¡¾Ô4Nt’ÚqÆò0f›059Ë#.³02«I²¬xm¼ÞÄ1¬ûH¤1m>ô(è ½5øt´Ûü\®ª°Ê“éEÀE–QÉ”è®ðï ‡+‚%Ð* ÞPl®]µr“–íØóàQeüгèf'ľžA]«ßµÐÁéß×|‚Îß`ìÏ·v(|:T?píø5‰÷HáÒàòQVÞDST sTBºTBŽÌN}pûúŒ28³xe &VJÓŸl“š8À™ÇýAY5QÐôga=ÈY çÐñ0mÀòE^‡#MS¦y~¿ãßVUtkç“,&u­U²Õê‘W¾D¼&™qB6ªJ¹çq<éñŽ„z¶jK+^éâ‘IrôNdõá,HŠè oè½Uð &ö ‰žÜ–8†w‚¿j¥Ý‘®ú®*³3Éø{Y™Iäܱ¢Òð¹'ù¡8ëÓ‘8C½¯K„ëM^”žãèÂèžëï^…×–ÅM7ç¶žøyI¬‹B¢E(WTGü1¤9ØßQo|­ÁÓž¨<:ª¨½¼uJ¸Nâ0£Y¡ß¦æ3Ћ¯‹ª`‹„'žbµ—Í…N âÆîZ¯ñ쬉Ëiò t~dâÒŠàœ-ÞÏûãjß ‚¢iä*ÄÏxM5cü‰ØõP÷ÉsÅO±Ê…óê+sYo²¬r£<ùªªuôTËnÚ_xÖ Fߎr'² Ô¬³$nö²­Ä~xG³÷<“ºéþZ&uPZåãÊ=Mù9‡½÷cÇÎsí¹ƒÚ¸éÍÞnÈ>ýkœš¬Ë4À­ÙÍÁW&Ð|îÚOÎØØÅ§ õõ"ó?°0•Œ^ú‰(†4eUŒï²F1_Õ‰{ûí¨úÈaŠ#7N9˜°<0cøE´Ô$[ò‚'îà@-†f$B€/4i¤·LF„s}H4:zÒ$åL¥f lÝj „›H½â2œrRÃ#y(kºäÈìâ7ˆïå9¨:D´×>ÊÁIžr"æ\ %ˆmA|êhI³¶rWäGÓÈ™¶GÚþŸœì{ÙÁ³¾¥Ö®ñ"ÉòkKžÇ[õ)¤8óø‰VsÛvªrw|ú¾)Ìfvd›‰ Ÿ99pGCú´óüñà­°îÌ‚„:b…yéØË?4: ?‹ 8Mn›‡éÄŠ5šá°Ýdà¤7,3þ‡?™=8vÑþ[•á|ºÖìŸDêñAê¥<ê܃#È©–U¨wø{fãA Ñ)háû^’˜uË©gjy!ìHŒµ^œEWµu,ÒÅÑùLø=ØÔŸCÝi™ãÉ‘;höÒqL<$Œä ÓPÃišÅøé¸Ÿk*•µ+özÄÂ8òÌüB$5$dpQõ¶š˜:›²ÉŒîùMC¥ÆRèû&reéhÎÎT•¤ßÐÞñ…"¿7&i§é½CÄ.ñ3ÀÏs‡üi¤˜dC[êÁˆB6wÈùŠüˆMáä9SOºqælp£˜ºxDÃ.þ‚ˆW}–Ùõ¥å!Šã†âx'³yÃô‰›ä±™GY dpv±èvYù±»c´‹2á,¼³éMO‰qhb ±6'`.ixî 5ÅÉ“$†¯Q’Žc¯brlOlrˆ¾¼ý‡I3 Æ0,i0âN u½=c=ogïOÒˆ7"‚¿Za©bæfQ˜IgˆÕÌ¡÷EÓ20¼vã|)Ò‹œï‘%@ÄÃdykÁqNW§‡Ç³õ˲ƒï`Faž9OæÏ$Z|¸*´«7Ï’f‚|i:@èyj¯M^L„IÕ•/Ðýq€bGMoF·îÂMã½|ÐckTCVGËFÇD½Dª œ¢a$ë'wýÅwõl‹9êaß9û6ú¬P¯$¬Ò“ø§iIÕA¢Vã©(aRÈ€G{»ZÁƒÎíMœ´òLØi³’‚ÞàŸ¤Q4—•qðûŽ—èÒþ6QX6ÏB@twFžú¦D<¯&)|àDçÌù;ª‰õßtÝ­àI"A—·Ó¡f&k 8ªûo¸¦¬‘D× ÞdOîXt³|70nA>ØÔ (¿‹v’—\\MAé(Ÿ’7uC©»xí(Ÿ›¬0£&Í‹P¤.½Ì{ÉœIU±Ûrå(ŒؽwØ ¢'Eª5u^>MaÔºÔ³DŒ¾$[ÝË ±.Õ7?Ÿºó³kôⵟࡳjÚ“c¤ß:€Ð »”º´šW§2¶X f¬ñç¹Xš¢ãW.£Ì¹…4Š÷ [Ê «kr°ã\«'ÂÐuU¸Õ×ÖuiøÌâÌc$pCšªƒZx§¦óaï&•ù,ö Ç@%¬ª TNΟu–@ÐGT\“Þ<”›Áì [º•ëÓ»ÄÓ£–¡²Ê1/j‘RYùà§b™™Ù<¶6Ë‚‡yBäM®…>´?à‚”ý@Ú…²ÿ³ìÄUñ9ê>3ÎQŽêXÅN·™gœXÔ“ãK ¶2W/ztBXP¹ ÜxŒÔSÈ ÷tí (%$|Ÿd”´€Ô€¨ä8Äðnìþ#‰¨«yæt8K ®Á’Ï鼞ä‡\%—GìÃpª)8¸Õš@͉D»èu¼ø`ªœ=ñ(ÎT03öO½üj¡ÚI°º"SiÊnkÛ“Ü)ø*Í‚¿£îxn#+·<\åŒÄ¦N±hU°ö¯¹{Ô‚˜§¨SLƒ<°$Î#,§UOpnZFwõ‰2¶Á¶Å£ŽÎ¨†ÎÙªBÃ<1¥’r]pð}yé½í(ãå ð ÒÌÊÒ¹ª’Ê9ªo²ß‚§ã™$Óobò²g¿¤&w»NkW1MÒ|Ö~Èï©×s&øxŸ–a@3:ä±D±÷iŠ ÒK ªø ìo„6…Ô4€-ѳC7Öƒ\T3ñjBר°@ár òõ—Dz^,\ ¯¼øE†%Pã®xƒ½:&N®5+7å‚@±×½=+í¹°Ýo½7 Ôcwo‚]JQ>-g˜â1Û(½T©”W—ïÂ]“`òýmxLpÊ{cûYKÊ|ÌÛ pz°7¨ïÏ oO)w\W½I>´o€qéK{P×pâ¡ÇDV7¦sûûòãrJCK@jQZÎê´¸ÂkJNb_x߉­¦ù&“ÒÙò4Œ M^*ÎhbôÍÄüÔ 8#‹CS]TÓí¹Ám­Cl:î°m…¬ d÷/M±OÜÔ]ÇÐ.±¯_íëÑNjœâêr¦¹w¶u‘©îEx ”¤Ç_LÌpHÒ=ú÷Ù„yeEÃXœyFQ9%ŸW&ŽÂqË}6-„6ùйÃj<:Ä^íW¦q%Îå~¯Š“®¯×Ï_Å!ÅÞÕÕ‡££–ä‘dUWñäQ…Qaåuô û=oè<¾BLË\ù¤GS^%Ê!¢amá:V›Ã¸œYÅR)g):lÍS2÷ÖkrøˆR™Ü˸rÎ|“Ò»6$ôܤÈÛym~…lëÉf–´vÊÚ¡ÊBn°lhð[;þ® †~ùíå—ÿÛ¥ endstream endobj 2 0 obj << /Type /Page /Contents 3 0 R /Resources 1 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 1 0 obj << /Font << /F37 4 0 R /F19 5 0 R /F44 6 0 R /F8 7 0 R /F51 8 0 R /F11 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 14 0 obj << /Length 1711 /Filter /FlateDecode >> stream xÚWÛnÛF}÷W¨ P@Dï•—¶/®S-¤E ô!ÎEÑ’QTE)ŽûõÛòb1IaÈ\îÎÎΞ™93Ô3z–§qšd3›«X97+ë«®â<1Ö‘À`HKZ¶ÉÄõoµž½n®þ„¿°´•‹Î_ï²YŽê’Ùýcwl’ÅÚ¤³ûÕì}t7×Ñ~ëóþWó…É}¤œ/œqÑ_0EóŸDêÓÜû¨Ø¡”‹šG~¦Yêxß±™W¨l?7Ytj‡Ò>z+;øÕ-ü;¡ØŠnaÜœyþ{žg.z5_Xe¢%Ñ|ÄiÐJúðNIGÃ*p¸Ù!w)œiŽrZ˳K- ¶‰Î»7·ó÷¿¢ ­ãÜ'ŒÜ:Á[¨82&³Lp¬b1ž/ÒLÑAdC£Z€ ­´~¿ãœóbPPßõÖWiíb›À!¤¼d‰¡/³8uyX¯Éƒ>…/áBI$§T5,F ºÔF´RÀK£Õ†%£ Êp'\ZÌì¦w¢²Aä?‘?P ³Ã%•¨èØ­ÉaË39€$ÈM[v* i©"€ÚØX«|ì T³/ÑêºF·YŸFeA8¸^gÑM˳-ùí|9e×Íct%¢îÈÑ.5@°u‰A*{ZçF*…Ázùu7‡–;´ê6õÑD|M§Xö]ÊŸ›‚ƒù‚¤° ×q¦2ÖüŽ¢…ê3ZWäs£bNVq|G'ñP q2\tÛsIIYs WÓ4}2$!½[~!¥\èŠ\ÄÑçO±÷Qè¤gÖNßF°r 1\­Å> W¢ j|›55¯Ë 2ä‰X)‰;ñ #W±©·¡ì¦üéƒOîøŽBà©,¼¦h¡áÍ~] œ;çu€CrÏÿwaËÛ::ñ&4Ì0&$z Žd}®z#)ÎS7ª~ðúVPƒÐnxævxŽ› š|o>ðÇ–P“.ƒïªGjö&2‚¥Ãßj¶±ÚÉ¥Ñ6 ÙåÕðz S¡Û…µ#G„ Ñ&u@T-–|Oêš1m½Ó˜¨§çÛs é¥H†ÜÏ-Æã7ƒ™ĹМO¼•lç!„uà»m¨SxÈž+ÝIá'“Š£Àí>÷=ŽÛ?ZB¼&ZÔÅÚv}æÏ—É­ »¶Lÿ pƒFžÃèH„æO¡lÀúRǘdI¶ J«)YÝg!Kp×Âd!ƒT§A°Áî²zÎ!OF`€Öîû,gV>Öl/B­¢/X3o[ÂÒÕ¯÷Wÿ@ZT endstream endobj 13 0 obj << /Type /Page /Contents 14 0 R /Resources 12 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 11 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc1.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 16 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 17 0 R>> /ExtGState << >>/ColorSpace << /sRGB 18 0 R >>>> /Length 12054 /Filter /FlateDecode >> stream xœ¥ÝM.K•àùù5ä(âûC²<5-!ÛR÷=RËjõaŒÛ\¹ý÷{¯琢½¸\˃¦y¨ìëçß>ûǽòm_üüÛ*ó}µ7;– õ]Œù·O«m™ÿVù·¹>{ªðËFûcÿx>»±ñËvû#ÿ¸”ö¾;ÿ±lº?ög²Ãï­èÛþúÔe~i%Û§czXžC›ŸCsÜgž¿÷îýÿ³ÊO·8ë[y_ÞÞÿõgÿõ;ë(Þ~öûßþõ‡·?üõüùíûßþùþé÷ßýÛÛ×ßü¤oÿôOÏnoþ?ø¯üþÿõ»·Ÿ×ç—þûÿ÷ïÞJûÙoÿôú}Ï—xûøÅêò“Ïúó9Þþúû·yûËß»ˆ=ÝSú{±Óü«¿šÿ‹Ý:*çÓÖ¹Ÿ_ñ¾ô^»d[Q|ö¿s&űÉ)éèi_]2ƒe(ö Z_)XÁ=›Œò^·dÛLà˜Š8ªÅæ¡XXª¤ÿ¡ö^¶¢ŸaO£³«rdZ è'6kNïV#ý  ”H?(Ïh +® –.YAßísƒ½JvpwIÿÏ()Ž Ú @ÐwÝ~Ÿ’Ýv]O6ØSl wìͶ¢Ø 8ªbõ?”­SQl 7ïHÿCå=/ÅÒÁ6$3ÈÝ~1Û®ëÏxeHÚíͯ ‘iƒÜ±7;¸«¤ 2b\â\ŠkƒÞ¼#è»=2Þ}Îz—étØn lKÒ?Æ|OM2#+vÿCÏõ¬*¶ –&YÁºë×”ìàn’~ȶ:Á2ÁÖ$+¸†b¶18’‰ì`n’ dCº˜èÜÈòpbÁ׸%ýe;q—ÿ¡‚ _dkUœ ìMÒvÝxlWô‹õþ%;8†¤ÐÑpá ô‹õ`WYÁÑýb=x²Gvp4ÉÎ¥èëÁ“.°,pÉ®*éhžƒr1°%I? Ë:WA¿”ž°‘ì52ãBÿü»­ì 7ÂH; ö+»¢_è'›Y¤} » E¿”ÛgK&°OÅa»îͶ­è뇫Iúš8éÛW‘ô³Ð]Ö ú)°l°Iþ*{¤é—ㇳIúÚ6ô†ô°Éî&i;g%ÛIO—g;get¹‘œIq-÷§7íc¬‚'2‘Sћ٪Ø9‘ôáe  ;·ÈJnžA¿¤FúÇð»bGSôF¸.|‘än¿èÃˇeI6°vÉò],þ1&:ŠÈzÇè ø¡z#}ÇòÄ ôáåâ‰ÙÉ-™Á1ÇÆðòáÓýcøÃWA4`œ8‚¶cwòA¯ íØýœ^SÑŸÛ»ÅÚ­G¤>Î)ÙÀÕ½o¿ (Ú®ÛÅ/š‘>øÜ͇¦‚´›¸H|>ìEÑ›ÙCîØ›þõñè42/°NÉÎ"™È­èȇ< 7}çà„½¹ÐÌ6N+Á ÚÐ4ÒoyöôË¢`íB/˜A;é"ç÷”ô‹S2 iy‡,˜@»¤FöAnÅæ‡lû°V°‚ö1ÒRNÙ{7åN·¢éѤýÖ&§â£Låêöûå¼é2´›»ùˆP9ÑuI{£4soºoá‰Ç9u{R"Ï3Ð3E¯MÛ3PåN{ÛŽžØŸl¡Â•öþ+zàï-wK'zi÷Aï¡ýµÑËE{cÍö,¸h7ÚÆmÊþyrFoöš}D+éѤÑ^Ï¥]é½¥Ñ^sAï#ÜéÚ´3݇4Úã㙄Ûcn>PnÇ[;Ñö¼LØoÖÍ+kãóòzöú¸gíz¼¤Ñ^sǵNû“ç“p¦G“îø<Óï…Ù^'úKáF·"]±¿x¾ —ã-Íö¸Ð WÚÆ˜Âl¼ wzíLï$ðy¶?W.ôZÂ|Í“ Ïwáv<µí×»h´×ÂþB¸Ðv.<=³v¥W—F{}lÏÑ”}—‚ë…p¦½?F{-¼^F£½>®U»Ñmj'º7i´çR1FÆþô÷.ÊŸ§ãz \隥ў÷­Ýé•´±?Æ´ÑhÏeúSåJï,ÜN{^¸GnÇ[û{û3a¶çíwúʾ¿«ßm*£=ׄýöZö—°ïÏZüÎ[9Ó3I£½š§4Úk­è…=ÿ†mO2…Ñ^kC'ŒýÕý­€0Úkþä_¹/i´×Çujwº5i´Çê>iü½éÏ>‚+Ûce{Æþ\Ï'ºwi´ÇÊö.\Ž·4Û#¯7Âø>ýY4ÛãÆ½”°·G››U¥Ñç¤]iï¢Ñ^Û{åF¯¥íû³eŒ—¢Ñ^[Á}µp9žÒh6; H£=¶‚ñ°p§Wׯþ¬þžPíõ±‡… Ý–pa{}ìíMûÓßÀJ'Úž² £½6︤ íí)ã]›É´+]³4ú[óÒî´÷÷ÂØß>0TÆýÚ㞥Ù^ÆãÂö'FÂØŸ˜n)Ìö:1ÞÎtOÒ?žKûsÙ;+e¶gŸ'Ýé±µñyöi/·ñ¼¡mŒ§nçÓ^ý)¥t£ýz$ìû³'ܯEãyBg!œi¿^D£=ÚL-öøx íDûýZ4ž'ô‚ñ¸p¡Ë–F{´I»Ñ3KãyXçxB¸ÓÞ_ gzgi´Çî3|¤±?;î—¢Ñ×­Ýè^µ½²4Ú£yicŒÇo'¶×ÎñŠp¥ýxGãy˜Íöøî´ŸÏ™öç1ÑèoûÂñ®´?o‰‹U»Ñsjcó|f{߸Ÿ.´w¢Û<žÒhÏø/Ò™SÚ_5dìXéBûý\´¿nÈhØÒî[;ѳJ£½ZÇšµ3]Jtç뮌 t¥­= £½Œ‡”½½b`#ho O|ŸæãYåBÛxGí7¶ÒØß߇ ÷M¯©ï3ý~F9ÓeK£½ÚƒÙ!Íö:í½†t£­¿f{þVJûË_,(³½ú‹éJ÷!:=“6ö‡ÿ zö¸ý~K¹Ð6öW²ÙÞú6íF{{NtÝÒhÓ;6iÿ>6·£Iûd³g•+Ý·4Ú#æÊHwz%m|Ÿê÷KÂmÐeizNéŠýí³G¥]’tÙt[Úø>ÝgÃ*gzmé<é=´±¿q½Nø>þbDºÑ­icó|»=1½ÆlÏC• m÷«ÂkÒ£iWÚûëè¹i»ŸUÆþÞ8^™®Mz Ú®×Êþ}Ï×hŸÀmÎY»ÑöGãyÙãֵ݋t.ô¬ÒxÞµ|VbôÀ¤A³ÝÏ*wÚ±?x=Æó‚ÇÞž… ݧ4žw-^Ï…½—4îÇVÇñî´÷™ÞCÏÃlæÕ”Æý˜Í®ÊÚø¼ ×{aïþ4Ï vòûmåB%û±í7FÒõxJã~lûœIéN¤èµ¤1žÝý©p¡k>ïÓvÅx7ãÕ]1žîtOÒ¸ŸÚl¯Â…öþ6ãÍí+o”q½Þl¯Â™nIãÉ=Ð ×ã%ëµÍ=úÆçõŽO:Ñ­Ic<º}à&ï³1žŒÆõúñ¨Ú•žI×ëÍóévãõvûì=iû<%ñzí×[³áz<¥½¿3÷¦ÝÜ×CáDÛ|á‘é^¥û¤GÒ®´Ý 7|Þê÷£ÊÞUŸ·a¼íϣ̣J—IóxÞÎø<÷ Âø<ãìÏÛ>ÕÌýÜ·v¦g®hof?ß„‹{b<½ð}ú3a¯ñ~ôì´ßï 'ÚÏÇè1èQµ ½š4ÛÛÆõ\Ø?oN¸ŒF{Ê ãáD¯*íσŠÍ·ZÒhO™çƒ°Ÿìk7”ÑÞl¾ÓÖnô¬Òho™ç“p§ûÒÆ÷içx|va{Ë÷kÂ…öó%í-ûÄi¯þ0znÚÇÓÂýxI£½å‰ñŒp¦û–F{{¼’4ú³<ñoÃxY8Ó»Hçy¼µñ};úóèÔèU…îwÍ~=ÆñÂzyåLûxqEÚ§ÍWkÒ¸ž6¤(º iŒ÷A„Ó¢ÇÔnôlÚ‰^KxòzióÝ´q½lž{!ïã7Îʸ^6ög•ööëeóüi|ߎ¸9át<¥q½líQû9Âèïš?Ø–.´··èŠïËþ*ý~±4¾/òH”=†tÎôJÒiOm|dlöwÍ_¼KWÚî§„q¿j¾­íß·³¿ŠFÈ” åN÷¡h{^%Œþ°c<¢ìûeô‡¸1’öýÑÙßD£?´ç¥ýáN”Ñßud–('z iôw½¢½ãû6„ºEã~‚©FÊ…f{¸þ°·Ó>»ó~¢³?ntYÒÏud})gÚÛK4ÆkL@“žIºãû Ÿ+Ü6Ý»v§í}®2Ž'ÒR„q½3Oip•öïËü$a\™ ¤|¶÷þ,ÚçSqâŸt¥m>0î˜Â¤Üh¿Ÿöýeóé’4ÆûÃS[”ý}©OÌ-ÒhOƒý¡0¾Òœ„ÑžûCa|ŸŠöö48Þöö0ú9žŸ}Þ Ž7„3ݵÙ:öw4·¿ø—nÇKšÇigʶ÷íÂï01I9Ó#Ióx#PÙÛ»¥³LiïÉöã=ÓÙ_Á¾?˜Ú$Œã=}b¾t£¹¿‚ý|˜ãµÛ…ýɬï ûþ²ÔÁ)þÆÒ“6¶ïG£=ÍŽûIáB{ö6Ù…qÛï-=±½/ WǶ^D¸Ûý€r>ÞÒ Û#ïG9ÓkKWlß|¼ªœéº¥ËÙ~$í³½÷WÑÛ³¿ÎôJÒél¿¦p?ía = gÚÛK4ÛòV”3íÇ#zží½¦ðÙÞÞwybÂl/ì„Ïö½I³½°?ŠF‘Ùß'z-iô…ý‰p¢óÎg{O Ÿíw“FRÛ¯­ídz"/JíykÊgû¾…9ßÜlï{•3mÏ;„×ÙÞ¯gÑþ¼°Vާ„ýxVäE £¿¨ìo¢Ñ_Tö7Âgû>¤q¿Ã<8åJï"]çñ–.0ò\”3]¦4Ƶ¢½ 'zUi\O*ò¤‚YÔ¢VöWÂØ_×Ãèµèš´±¿·"Œûå:±?…Ñ^‘§"ŒñjEÞ’0ú»q¯Ò¸^؃…¦ÝèZ¥q½`Þ›²ÞæÁ6Êhö`¿j{{`Þš0ÆÌ[Æýp+¸Ÿ®ôÌÂÌÛ±’›Þ…ñ}Æ{ÑèòD”½½6¯æ¦ŒñFë¸ÞFEûx^Ÿ‡í5÷³í1Ç»±=Fãx·…û)aü½ñH4Æ›ÌãÆó›˜<¤Ñßwäù)û÷±…3ºlžÏÌËÆùÜ‘w£œé>¥Ñß3ÏJÇ«WÔã.ôÎÒ8Ÿ±ðO÷‹íQØç«›÷ÔöýÍ<,aôÇÝ'ö(ã|ëÈËS>¿oeáó<³£Œ§0ž7uäÙ(WÚG4ÆkÌ»ç÷Y{WÆñX~½FÊ<+å³ýœÒ¸ŸÈ£Æxkø‰¥ŒóÁXʸGðXôy8¡ìŸw ¨©ðêtîÒ¸Þ ä5('º,iìïáõã”q>0ïI¸¿·—4ÆÃÃoL”ý}f<Ÿ¢q> XW9ß7Šv¦m>m0瓚ý|æþîhÑÜ߸ž£½äQ(ãó ¯F¸ãó OJÏK­°B‘Fż'aŒϧè¼éÕ¤±?Qý*údåŠHð´WÍÚ‹¨”ܰÛ/2L›kA"ýUzk§>èE¿oÕý1\{Õ½èQN æ@b5¬`»É€äæ…J}úWgDúƵ¡‘>4±%d‚œø…x‚¨ý¼OÇ‹ƒ{mà@?Ü^Q•–5\ýr~*4&nL_eÃ>³’{(¢1`¦}$ÃzÕ½úH¾FgµãH 0OAo*ˆ:D½Û|ê]¬ÜØ«3¢r9µ§.zS±‘w‹h*ën"½©ŒŽTè@o*Vœ§)zSAi"ÁÂwVô¦ÂÙÞ7¹Xf¬W­œÏô¦b ª+¢Ú1jNGúcÖU^‘•³ÊÅÂ=á;Ð’ô‚,o‚ÊÔ‚Þ,[³+nìÉÄ{.q 7¤É® ÐÒdÅÄ‹¬°þ R<ñ9Л +éFzSaEÛHTÃÅZ•Ho –ô¸" ÃÂ<üHÔŽeÍË@o «¢ra _€–_–QKµ¢úƒ³sofVem¾Àɽ²L êÎb^|¤?–Ç· úðƒ^#}øÁ®‘>üX< O_ÌÖìdΊ>üàLÝȽŽ[à©Ëî=m£º]$*â„ô[‡]‘ u‘ÁÅÛ¶Q‰³!õ)Ъ°ji$ê°ŽW¢ïgúßõ ‘~;¶Ò˜Y,2¿ôÙ\¬mö0h„Á¦ü “½Ìb‹å„ÅÞFk*s‰F˜`ª'Ìõ6ŠU¤Šïs›“¡N±Ma„ís²¡0ÂlŠy #¬1õ^y»ží=Œ%ºœí{–FØvÂdþàS,.Fya¾œì'ü*–iÓ„¶G±.a„Ar²0‹ña2®0Â’9™-˜3¼Xe‘žØ“e…¶Éb–ÂcϘ "ìCƒÌÉb Ûc2kð)6•;¾¢ÆÌbŒÂåd)a„[˜Ø’f1·rÂ?›anî)=Ïö3I³ØŠå #¬³4„!Þ>ÅbX,O˜Å 1™EaëöToH£ØDÁd5a„=㩸2Âôí©}—f1=L6 æËÊÌÉ›/xÙ&Ìbs aÓÑsçdaè˜L*œÏöuH§³}_ÑmŸbh(ö(ŒóÑfiã|d±:a»›'Lè6ÂâëÄb³h„ÛªåÅ0l3^g{_ŒóÅ΄¦ÛüÉ…2‹5áe©p:ÛÏ-|Š1ÛÓ2m„i³—0ÂÖŠá—ã©ó™/›ƒO1_sÆùÜsEã|Fj›2‹ÉpFã|f±(a B±¤`¾ŒÈ|Ù+Œóµc²0‹å´vp×ÓÞN˜Ám\Où2W×S¾Ì n ƒï>«XÅ2:Š¡ ³ÖD˜T4Âøù²W¸žíg–f±¨…°Œh„9w¶§ÛõcÂda„ÍtOßF1¾Ìf1—Šþ&aìƒí!ý)_F ³X çíS¬Åj„1ÞxÙ#Œþ’Å^„Ñ_òe0‹¡Dp>Å$¼ê•2Æ+|Ù'Œ0}KF±¾ F1›‰—ݸŸ±çèY8b‹‘£ö>y¼£q?3Q @÷3æ%ý9®7—í…‘9_‹/³˜Í<‹ånsãe¯0®W|(Ìb.(Æ#ŒýÍbÂ,Vá«n£O1ÃÆþ^‹ý¢Y¬¡àûDã~q¡€0îÂüƒç)–Q1™=û{ ,æˆf1Š‰ÅˆÑ,V²0.û{!ì^û{'L†‹Æþf±ˆ`†a™g—f±ƒ|»ÜÆþf1a´÷19šÅòkqÌe“@±“àÎã± Ú{ô:ÛÏ"Íb ‹ñ¢q¯ö†7¤Â0va@˜³0ÂΪ§®(3¬sb„Vž1™ý6ëyÇ]¥ÑŸVÌ0f8Š1צ^º4°°jOai³k‚O±’Šb Cï˜ì°ºŠ™]ÂK©ÃÆõ©Ê»±ƒ)|¶ìÆ"KãzP1LaVÅ„1ÞCj‡2Âð8ëQ80î…¹˜Æ½¤æÍþ(׫†0Ya†WL†N',Úžwß~ºQ†UcÖŸ0?¯W…SFXSC!ÌÏ‹bÂü¼ƒfX7Âìƒ×ù¼hÂü¼˜%ÌσþD˜Ÿañ ËEØhð ïk~…ÛõJŸÇfíUi|žŽ°Oát~߃Ŗ,_µJ3Ì–Ç;a´Ó_\*·[Š4î‡&ˆ…ÑŸLÌþ¼N±—é7öÊ CÅÌbaœÏsŒ„1žã,AaÜ/Û|ý!Ì0°Â™‚“cªð§°Sa†án¿^³¶ÙÎwaóZ Ÿ'šaž8Ÿ…Q k!ìP¸œ°ÑQ¥q<—ŸxÑçzÁ™‡Âó„}Úóaô§œ#(\O˜%ßmôW «fØ&÷÷í~Â0}œ2Î÷µñy¢q¾o„õ ã~œa”ÂݘÜNX#¦„¶ˆ0Iá~Â&íz/ŒûÇ6^Æþ܃ÆþÜ<Þ·ë ‹ôÿ¢Ì°G¶‡h„íoÌ?.'¬Ò®÷Â1¬1¸°½n„• c1ç c±aB0c%¯h†9",083LÆÆöþÀ¼´±XŒa{Âí„ÕÙx]ai¡?‹Æ²¯„°•àÄÅгO…6’1_Qaóe„F‡° átìûó³½2ìó¥”Ç {Œr;®C‹E&¦ŒÅx˜ï½¸p.ûƒ5åq<“4ÂV²ÏçRf˜ŽÏçQN°‡eEO~„e)ãóŸ§ÜŽýûF#LaWÊéø¹±Š>a~ÅÃ씆äauÊX,ŒùvÊXŒm«n«4Â0±ê7šÏóÌsKc±¹­ÊNÒíØ÷Wtù&¥Ì°³‰ýy»1,¤°½E3LÌæ”öµí|UÆbÙâ/ΣYìÁìçk4Â0ŸOùF”›t=aC~>D3 ªXÇÍëEEjš2³# Hí½²=E£?3oi´§êóÕ¢9_´"uUa_˜x¡Ì°ñ£Œ°æ©AѼ_aØŽ2c#,G™a1Ÿ7ša>¾êYý=ª‚/.¨¨:¢·Ç·°›²¥ÆÜýÁ†2ÎgL WÆõÌ&¶a.ɳ0˜ç~[ý]÷°e\l%ò ƒÄº#e„1 ŸÏ©Œ°œá7–Ñ\êV‡§ú(ãxMp®Œö:½êJtaXÂ$”ë±]/‚OØÙôù€Êü<Ãîç”_aÏ…#šÅÌ­K# pzUÂàÉùÚóÅ”±¿—?¸Œæb¤–|¾ÐËßù«ÿŸÃ&ù1Má3ÓÒˆ<©ÌZìdŠ>x@‡ÀŠGkÓ+ió†"é+×/<ôÆäÇ@>²C]A?a0±S°µDr±A{­‰¿XI{™HχøÌ±q‰ZB‚þ•Œ.˜J+XH_‘y‘Ùä|û™È:"Y·‘Ù‘íp*ú´ «½¶";:=›œ]ý‘Ë‘Ff2·È†î®1! °“­(²ÈÊô¼‹ò0"+i÷Ø‘‰ì=²0áƒkÓé«í/fÜœ¡‚¨ rV&Ö¶^LxðÐP40r­+V²E¿$¢®øÍÎjÞ,6ÙH_ ˜I/t~‘u(:Ëœú‹âŽçØ‘> ··l;’“hl1YQô[ÌŽTƒÈJö­x²4|u×ÅñJËàÚÊ‹´™K‘¾p¥û b oKYr3²“{+ú“oÄn6Üв˜f¤_.±¬UÐ;U¾K»ÉäŽ$†H¿ˆ°æMNcÄjaÁN–¦˜™Ãaÿ&'0tŸH&ØI{»égúÆÚÚ‹<­:R"}`„Uè7ÛÆk#+i™ýÄ9)/.%*ý%ÉxȽè§ß`Þä „S|2°“«+Ò O]l~ ¼yŽô‰sL4‰Lä^‘|©‘'NöÈAr¦ýE¿ÐÛˆ©˜È>"yÓ½ÖĶé˜?¤¡ÜäãÂ5O÷ÅN¶¬è7S,“x“U ×Ä:ª@¿1Áã&ÁJ®h+ǽ1¬|q0……aÀ+éÑ£‰),©F2Ñd¡¬[¤M—¿@ô×~kcVúE&Úòp¤wåÜn&©äfzÑoZ7R…"ýdß»îbÇe‚e #ýBÀ"„7n[6JèF"ߥŸ9À3Ó_Š 'èÚ‹¤¥èƒ“ͦ胓½ÎÜÝO,x4µQ¶2Ò,m̈D& hà+tÆ'EßÎ'T/…±(,eL:¿¸(•ù„±èo•}ŸP›Ûù„$T¼~…Þ”)¼ ’|v±²ïLóØÒXtÆPaîOL:~…ö "ž0 ³¢œ0BOB1‚ûÙËï=„ûËKºœÐ™š…ÛÙ_¨'Œý•Q1Rû+£â0öW®¯IŒŸÌ[+óÈÒØ_˜ý¨ü YY¸œP˜I¹Ñh¯¬Ø%Œöš'&ÍÞÎ<w„q<*#Œã‘ý¥et:¡7¨˜%ÌãIÂXTŸ÷™xǃ¡4·/f¬>PÆþ.¨è¼¸¿ñR\û›¡6ÂØßÅ濫O¨¢×•±h“¡4Â%*¨ØÌÄr ¥ñI-Ñã„Úì$]OèLßÒmªÍfr£¹$if˜Gn Õ¨<Ñí„ÆŒ%í·!¹¢âGpÅãœ\ý ¯ŒPVœFhJFó¥xÆênåŽý±0i-š¡+= æ¤( …Ú-chŒpí4'MÝfè B£‚C¡* û¤ óNÒXtßü¥ãmŸ”„Ð?^Ê ÝñÐ e,ºn¾:<š“ŠróÙÊèïšÏUF×üxE3{ÖÌI:·Ñßõ„ï]?„Ú(#4¤ûìƒh¾ª0û¤šh†²øêde†®øñŒþz“ºô+Ô¦%iœ“Fn7žÅQfèGã+ã|ìûëvåùˆô>e„"Yú`‘.“æ$™Ïæ¢q µÉU¡}¿&Í\Æxd$LŠÆõI¢Ïxa¦Ê …ñ·)Ê8ÞÑZ…9éÞBq|Ò]4BÉÌS¡Eƒ“v.Û¤”JÛù Œã=<4EÇÛBs’0+Ú0TGÇ1ÑÊ8Þƒ“nŸÐ„æ(#tbøìHeïIÑ8ß§‡fEŸó}&Lʉ®'´gi„ÚÍŒI·ÏùŽÐå|B{|ßæKx÷”F¨æôŠPÊM;“2nW†šLO¯RfhWtQF~&]Üæ­0C{”1žš“ž¢ë íiM˜¡ æ]¥ІÐe„ŽNýPÆõ|úì«èWèѶñœ2C<Hãi„Û¤„þør;¡=6©X׋åÏ#£Y‘Ê^ÓF{Z˜4ÌI½æ¡ö‚ÙYÊo_}= ¹;öG4Bã¶—®Q.'TÆ'ÙÜfÅg†Ê(3ôÆC9”1ÞÀìfeŒ÷1°‰nl6{:Kc¼¹Ñž„±hÐB‡“0ï·ÝSºžP›T+œNèŒM: fè)Cg”±(Õ<¥Šã‘îÑ/Q|uÊÆ¢¥ä¥Í•±h¡-ÑgQÜcë…šã¡BÊù„®Øxç6Rõÿä¡'v<„zQÏö·já¡Ñ‹‹³§û)·úaãuá|BGž ôd¨ )÷JbãáŒÐ‰fE0ðU¥êðxOi,ê³ç1]˜ÏÓ|õ{’Æ¢2„~(— b“ö‚ùŠÈžÛý„0UO—TFèB9¢9)ÓÓE¦4B PúL‹ª­ô\>‹šŠÜ”±hïJ”±¨¼úܶèÌEíÕg¨)# v´—h,:Dº\4Cƒýmë’Æ¢Öæó”ñûðv:¸¿B ¼cTþª½¸?Q²I¹ŸÐ»ßfhmÁ\:e´'̦S.'´!%áÁEå˜Ù¦ŒEãÍߣ*ã|khOÁçsó{Êè,ýw ó~Ölý­0Ý"}X‹¾Qè)º2” ¡ Êè»ÏPFЃhNÂcèr;¡ K;½B¦ð¹"t@¡%˜¥Œëy÷tÄètB–_/…±è¶o|žh,ª íõrÛgÑ¿ÏÜQFh ýD/†: ŸÝ®< m×KárBìzÌé>k5Kã|EÉårBÚ梳‚ ·2BãÆÆçÆõÚîG·0Cmf´Ý¯ c<3½a+c<6ýÆ0ºEù>¯C¡χ÷‚99Ö¥ÛýŠ0­ûeô·¸qŒf(pA‘eôËgø*# 35¢O¨Ðò9fÊø¾k£½Þæô#[Ômã}aô§»à÷]>¡¶\t­Œñ6V·)#´c{qªès¾ì…ßñÄöôØèÉE1¶ú¶J·³èÙ®GÂXô“|QpôY”™ª÷ÂåØÎ·`ŽmQuÖÆ¢Ýäó—”±h'ùêÒèÆEÁxß«\Žk®çóøûÓE™‹šy¼os2‚yvi, Ëþ>.:sÑgöse,zË ÇãöYô—½èŽ25eïØ•_‹Žm¼w»0Ô¹´á‚E«þ¾,úTä¶ñz—FE[½ˆ>§¹hU¹h5øTTÆŽQæ¢P/J¢ŒŠÂÏûƒ9!ÛÚxKíE#¢î7΢Qát…Úû¦àS±¶á}ŠpÆ¢Êæí1¸œí±hR8E‘Ö^ƒOè ¿ýùß»NEoŸxœ÷YDå3î_þþË/¿~ùÅÿü¿|ûÝŸ|á?øÝ_¾üâ×Ï ôöõoëÿ;þÃi)‰³ˆÛÛ×?¿ýì×ÿéWoÿáÍnK¾{ûúÇ/ÿðÕþD¬Óûw³çŸdTõì!ûÍ~«óó7Ë¿úð«ã*µ¿ÿ«³Ï0ô^:¿>ôüô›ÿéËÿHàdµ endstream endobj 16 0 obj << /CreationDate (D:20150701163317) /ModDate (D:20150701163317) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 17 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 19 0 R >> endobj 18 0 obj [/ICCBased 20 0 R] endobj 19 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 20 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 12 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F51 8 0 R /F14 15 0 R /F44 6 0 R >> /XObject << /Im1 11 0 R >> /ProcSet [ /PDF /Text ] >> endobj 23 0 obj << /Length 4012 /Filter /FlateDecode >> stream xÚ­ZmÛÆþî_! ÊCï.¹| ÒŽ£NSMè‡ÄxOR,‘ )ùrñŸï¼î.%Þ¹E ã,r_gggžyv–¿¼0‹þ™Eš•q‘Ö‹"ÏãÔf‹åþÅï’Å *¿[$qVW‹jºÇæ&®ó|±[üëÅ?_|sûâ‹7Õ¢Žë"-·÷‹470T CUqCÞ®?Fyò1¯¯nÒ*©£"ùXès™|,á93eUÉŸ®ÞÝ~÷âÛÛ¿8ÁL‘ÆIž‘`‰µÏ –Vell ÏOHfŠ*N2³(lÛ²`ÉÞ\™èûW8ñoŒ בÅi†£Q³¯¸Åd8'E© ²ʪË8ÕÚ4®ªúw©Öfq‘”Ó¤y•ƒ³,ú)É–ð`®nŒ©ëÈu2©Í©- ¶Æ*v-“$ÊŒ *o,´/¨–¸¸)ó8nDâP—ÔIôñÑâoižÑK–ÄufŸ^ª8¯*ÊÆyjXkÓÉÚÒ:X¸µÅ3•i6­¬­¤fªmc½Nª$ÎóRtò¿Ù×ן²¯OëÑTq]Öÿ=¦iœ²©&5ÓªLUNU••ijçMô8_Iz4i—¦^ܤ .@ Éq{UeQsMwð×rgó%¦Í¢Wýþ¥ÍÿmGø¯ï¸A¿it„²ï׬åa…àoă¼Üà¿£¼ôÝÈC  À­"j±áGë®ÒŠZ޼4lwßüТDË«´„™©€…ÉÎ…Ñסõekœ¢?á¢Æ˜µc³°ŒKÅUZôqKýLÔá\$ üÅ “G·2ì-ÔØ8ð¾Ùa òŸö8JÇ%$.³¤!®ßŠrZn¦ßV·¥1ñ©]quãY—¼"Z³%A©Q…PÐ(•˜„ý0Œ9«Ð*³$‰ ›/ûŒØdué-Ð>× á&v8Û/ðwRÑQÒ=­š´±FcêiÃñi‹…ã5+€>‘’ ÝØåƒîÏ~Ó“³®ÄH-8áÃÆ12¾‘‹‘© ž;¸­é‚”‘‹q{›{æ;/||ï‡á”íhÝRÄÅRZ7cÇ j4žž#&/ñLU«•#ûAG«*9ÏM=~Z ·˜G.PÒ o:¸S£‡΀´¾iÇsMM.…³è–ñ(;îwp™‰÷Wy!ÙèÕȼ+²ÆÕùÄÛàÈ&¶E®ª{‘p)’4þ/ëYç6ÓŽ’ß°&U§Stþ†hx#LÈX21pxíïÑj¯PL*ÏX6‡âXÞ9#ØùÆgÓ´“Ãäà+äì*=3BÖ<±ŠBÄ»¨—l2>r~d©fÑM§oŽ~=ý~6ª w&ãN–†-`Ñ…TCç22È”èLø›2¨_Ñ´ ¿,{¥DR{'¹!zq|Pˆ±k%p‡"4Kµp2?YË Ã„~v‡×Q‡7/9™RwØ ÉÖM¡•'k†%öc_sÚB@b¤Q²:cEíp²šC9”±s‹ÍS©°Î\$¡VŠ•Jþ Ly=>+¯ïxˆ}³õ×™sjð»­÷è ~Ôvðc‰úÀ)+„K¦ç* QúníìVænº¾cÊV{žÏ)ç-.„{@1Ix¡KG»~/>‚ŤûÛBÝÿ¼œ„a|á ^™ ì‚ÜÛð€Þq9mį<›C}YÐÀŸ$H \ß ^Øø1z~äõ¸ô$>ÂDBË)z¹gTB|á¬KŸD>)k ÆäAüüÖP%>&2Ödi˜eï°†)ÏÈ}2þyÒ{jeXtx‡îvžt¸DŒÐPÂ|'-ýfCu$j…êcqÇmŸ>^¬ü™ñ&­|Ò:H¾3tRÝÊ¥6BÔ³„l/­N0×¬ËÆc§ŒÊKYñ›òª`¤@R36 V.ùtð;Ù-^²&lj&Äé4SŠÅ@Umnh9áºQ‘¦((uPÏ\m%6ƒýlî<( Ü”A¥âWLúh­&gé¹L‡¥7 ÞaíìöBñJÍ5Ìh—9ƒ°.®MΉôãªVå³hOtûèòšŽÀA[¢j#ýOœÓÚÀ»9ó£0‡+žô½(…·ðñÄ´Ò$ÈÁLïi¸åƒ’Ý ¿w § +Té?±î‡›ç±¥ ;gEèH_ùC6X†iPI“`ù½š’4É-üñAø+†°Y:׌cà[XÂ^°Ô$å0ðV’* “FÿÆìú¬æ6:’â\Zžýxt\¤5•GoòË ‘X2žr}³žq'~UUœ$ú…CÉë¦ÑüÌ=›*O^ µî;¦ðAh! •‹¥|Pª6òpŠ øì®"%µ Óp)á$W1ºNç(p”/9‰Ç™ëѯËG-ŒE;/ ôd.4ÌȽ º":”K~@‰&?¾ÿË•-oGùõ-a†gZrp'Ê¥&ù÷9‘Û“ÚÝZa™ê‰o+ãß÷’"¿˜Müý¾‡D¾1à t§8ØãháPz®N𠛪沙`Þg_Üz‘ãtiê¹WšÂ&çÉ—h¥€^ŸzêÎ=øHWaÆá%À¨ü™£«©£ÂLF‰™£Ÿæ¥³é‰aš„VCï—bê™  ¶×¤¦šî ¸/ äö4 °ð^!7¼f¤WÕ;XŸÃ¹æ€CA ¬D £D( -å´>H¡Ä §`þmÐOcrÍ—¾[eþîâc¾Ä-öjÖ u:»~¯ bb :x%íR2t†fT…;Aé­å$©ž×œïZºC¤¦¸‹dªÎœ’ÅBR;ó)ô0ÜpÜ8ø–¡ÂåÀ«œê™O?4óz et ž[ÿ9G.ùö¼>Ÿàâsh²æ¤Öaô‹xÛÍéÊùVfùaír¬¸»YÎüp©w]-rêß_`åŠu:5fúÊYF‘'$À’¡¼$ùäÈá|oäºfðbÈg5P%sí‚ì&¶äO زݥJîÎe¸¤Ž)ÆÜ5Fö9ß©¸l^= *X'Žâ£9ûeáׇja6ù/ Á¦°Ñ_ÃáøSÇܬDƒ¬žL4JàuxÆ1›«"¬¹‹9Îù5Ž^öAóLråÄÌPØà6pð½M)SMaȓӫèåâ’ ™Öråzð93‡ÏoÍdÍïå2„Z0cHÄŸÿÂÇ>±T÷©²è“ô©ugú­»hGdÆ;] —)…O^†å½Ð/$ÃÕ_–RcMLøW.غWªŸ6–'პ±ø"”ë a.A5üXÊ JÑíhÑkgNÁÁ¢—Ö[wâ!/Q) P‰ûˆ¥å›¡“d"X7ÝÙajÆÀ:ùœ v§î1®ðéª2ò_ßÀê]îJ¯$´ñFž÷.½¯ Éu”Îù±GðYV}?¥¡ŽÂe¬I)u:rϹÐÇ^+/d×K=Q›÷šª ù¨èQ25š”ßîÑé;r&H ¡Æ»ý-M­9•ýÛôóœ¬p+Ë$Ïì?¹Û…0[à ï½¹¯&ÈWï\¸(ävä@ý̉ê5'Ö¸ˆMsð°nÅ,£øgé7I¦q¡j%;°>Ië WË Ž6zsp®¯ \g®& <¦øur§e®ŸØBÜÍH¨•\OH‹¿‚Çßbþ"ÃÐÝósõúð Ò=« ‡þÿŒw«DÿLônæ³ß*Ó.žâªªäërýî÷?}ùRí endstream endobj 22 0 obj << /Type /Page /Contents 23 0 R /Resources 21 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 21 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F7 24 0 R /F54 25 0 R >> /ProcSet [ /PDF /Text ] >> endobj 29 0 obj << /Length 2749 /Filter /FlateDecode >> stream xÚÕY[oã¸~Ÿ_ajwבu):l;@‹.ZtRE§²­8ÆÚVjÙ› ìoï¹R”¥Éh_úàˆ<¼Ëw.d~ûæíGŸÍª¤Êm>»½›S$y•Ír_$ÖùÙíföÏùûÅÒ[7¿[˜ù~5üÎòý•|·ðkàgyêï ¹äæšYñ9õ)|^³Ñ·ýjY•ÕÌ=ü²~v+ʇCy4T„!nWý´rØ5Ð4rmhä_·œ™4K²´œ-I*/ºÙ ß+áûoôn$ƒLîDð¦Šfù+a}$]µ}$Nµ•^N Q”‰Kó¡ WggÑÙcÍšW¨Õ†³IÈŒzx‡ìúò}Þ$®²³”ùúµL‰êk+°œØ'˜Wï0äïF¸FM|¿›^ú —.«’¢ôC~ÅEÕ'øµâ #7¹“ÙkYÙÊ ±‹!Ž>óugº0â†IÄ‚ý^ Frå¾}5r º¯Æþƒøià?ú·ðsa‡Tç/9/Ž@fü¿?Áïƒjž>N¸wÙ›!†ÃDse2R›W ¯N{ÿåE#xM)é}ï>¨÷ØÂŒPÔ ¾N¢Ñx¶(ãÓÕ”ÈΦqÅa‡7ÒUÜ ó Æ(ë§`dN'¥}ƒ¶CþÇJ‡dcäáGÑõµ<ïÄ:Q~iå +Ô eV¾Ù(œíeÿç¯ÑУÈü.l‚º<ñ™Jü$r­E€}Ĭ‹¢}";ùþ '™®nðNpõÈ`§>'ÝE&:0é“Ì:ˆÅôð›‘× ·Ï_8òY;ùªD7Ã0:¡±¼¦Æ§¡å¯lcD!f˜ºS!G:*"ÒЂі‰„€²Õ5Å”]ǵ(ù(§þ„ŸÿÊÔ¦o?fÙÌd‰Ër‹‰véó$Ë!”eibE-nZ²~þW41‚±9¡¥ëf±t…Çôᘇ‡Ëž‚¢¨=v|@åq@¥5I í|{O»t–ιy·ùC>bÛâI¶˜·ˆø ¡Y#E{dwÁê`„VôÚùº=hîžüŸh*o629Þ¡gäÔËð¨µl¹=µ4I—‰¡:£‚²Ó°”&Y,³ªšOÞºoÃïºÞÓÊ¥w¥(öÍ3P&ÈùÓ–|f½ß#Bú.ªDO;qæK¤^~Ç_ÂÆYo/G‰l G21‘m¬+_”Ze=­W%ÎzÇóä8Ò—p¾Q³ ùrÏåã™ÇH¿øåÏ# Üî›^%0ŠvEÁ:>Öe [3Œ>ÝYØFlZ:ˆá [ár§áö®é*O-ê½¥yG*¹è$röÜ`h髌‘°DÃŒ¨I€éÚ©Ï`ÚGxŽÄ³i9ÿl\ÖÕBšuÃdÔ·M î­'ƒ=à‚¨CÀ%Z%…µ„û<Ϙ!«¸Š®N§#ᙤ8rGdŠ(äY¸*s½;8®d¾¦ó>ZœÚoßµñÓ&°p’s”Ц?¨¤\´I‚ÒQF'±QdüÃsn„¿#nŸ¥ö ±ƒ,{7ÿgqÙD‡8ëÞN¼ Uˆ¨wᙉèã¸622’7ê|8&ÀAú®.ˆT˺ïKü¢3›‚>Ø‘pó6œ|ÒÐLjÝó"ˆéš³Fh ÊMÐF‰t¯Q?îBK³´+гYR¥…„\ËçþÀQ‡ùk)k ×&ÃÇü¨f  i+)DŽPÓÍ´ËŽ‚æAq¶o·Ï¥P!À Eõl\^ó'Bvœ3p“XIì¢%YŒaœ€úª»fÃL(U6Þ+T¶²Ý$´'ÂÃøØà7EņŒb5|ClGÆA<é‰ûŽÎ/f#˜FÙèj+›¦)F´tþ7dy§áaÃÄOzå º°B@<’3t,ÁÓ¦ÆM(çƒ7Oè¤E¤Ÿµ¸gÍ6AÇNÔœf­Ù@KTYËŽOagÏSÅK¥"lZrR["ŠæÍÌàÀËÿÊ6ìH[bƒxQ`§¡¦Ž'®k®&Þ2¢£Ï©ÉâÜ$²N§wf"y|c@—£ÐÃ%þŽ<¼vM2¬9qí@˜ÙÞ/¨O“õò`&Kô†Æ…4!笑—Pý(u'‘¸ƒÅ€£ ú}A…­”g‚ûͤ´X49¦ElH½0¤*¡âúdi:¿Èy(?Ñ£›Vò€ì%š#í†iœ9î¹C!¹“mÖb9µe¦b›ÀȾoÔ'ÍÂ7.«Q°ËjeuVå/”Õž^ïe¡•ð_HÍ“bÁ0Ã;=Ÿ8]b®Þ ‚é!‹Z•:äeÍ¿¥vn‚;&SwB°¥³Ii¤¬Ëä:øCÈD”ùÅ€.„Ø| ØÔòó¨Cµ†‰AÝÙºD.ƒ? ðæŸèi*rO]&M™¸R®YFS4|ë¨QsÂAˆ@¿ãbœÚ”f÷ ˜¨q(f):€Úése³>G™|XåD0ÝŒ¼ÜŒôZ4È·:S±ŸûPät|Ä++`˜yP¸„Нé¦\2°VVãªà@Cw¤.hDA±J£'` {žRñol uaÀ÷zlG@š ¯ÿßJåšÈ‰‚åàpÊ]AypðºÂÒ«_½@ç1 ¨ÐÝ÷³ZÞ¶]õQLJtÈ·ßu> endobj 27 0 obj << /Font << /F54 25 0 R /F56 30 0 R /F44 6 0 R /F8 7 0 R /F11 9 0 R /F51 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 33 0 obj << /Length 2708 /Filter /FlateDecode >> stream xÚÍ]ã¶ñ}…eô¬IQ”Ò&@rÍ¡MÓí-š‡\ hm­×Ézå³ìÛ;,ÐßÞù¤¨µ®IúTö4’óÍ™!ýÕõÕË×õ¢É›ÊV‹ëÛ…©M^.*_çµm×›ÅÙëe]fÝ~i²ö¾ƒÿ‡å×ß,LéóÚÕ‹•1yã=/ýrÛÑäÊÔMî]9} ›¿}µ\9ç²íýù°„M“ùâɗۢȼòÀUñT)¾òO•âCñüSP|]<Õ 7ÅïP˜«¯¯¯Þ]™EÿÌÂ8“Pµ‚¯qn±Þ_ýðc±ØÀä7‹"wM½x¤¥û…+CðýâÍÕ߯¾º°•·y0vQY›ûJ”ü¶_Ú=‚–¡¨²àêåÊ’ô*›B¡Ò äB%E;0dQÃ&„ ²j‘—ÁMmúW0c·nà²Ì¤P÷R€Z¹šBéSE\œµ^yåU;;åõ§0ÛÞ-We|欰pÊ´ô 8jF–NÁ~däÊ͹².ǫS{Šq¨AÀäΚ©0S;ƒöÂÀ{§ÖµA­ëL´iˆVPù1Àl‰ª_Z¹šµ²‰V‹Ôl´®+"‡è *'¬ª_id[©‘Õ¤®QZF]S©ýƒŽñÄ'äȯ,m/‡úzY»¬½^dh>”ö3`[ºìˆ:¡vº•w6k·²´}ì†á|à€wˆÀíGÜŽ<Î8ÂAÿÀËnû#8q7J0àøŒ¤?ò<$[eýá,‚´ %2a6l ÑØlȺõL°,,ÑÂáî&¼ÉÖ°ÈÖ,e¸‚…ÂIÌríãëNV#bè^ÀÈÐ}Ãè{‘åg$Õ=î(9FÖ;bM -q¸oïÇA]¬0eá5*¢1{@&G ¦Là’/WUY³{Þ¼íÈêh”bR-oL<¦ò°Ä„zDu;6ú£ºý~tp»?Lƒ…1!ÚtDþòÑÁ}GúÕ/{¤ü»!q½ ¹¸Í¥ÊäMYNãµXÒº*Ð¥¨Þ{ÔæýÒW˜9`ÊgëqŠ4#MqÏ-ZãÈpâ˜Hîfàð'*¹Çm©«q³¤!R ‘»|[˜ó9ÙcsŒ!žƒ½]*é{ ¢_Q݆)ÇGuãÏLl{ì‘ÜAêpj? qzYÔÅ„ºáðÛv¨2F:`Qíéxà ù‘a\׉ŽtÄÒCŽ Z !„û[Êh¬MOI-&_®¼Ï~BìZ€$l0,Á*Çqý±“”›p´ÁÓP5Ù÷Ô¨ÙN58‘…B±‘ÓYnßI*˜9:êWV’ûÀóïJL>0pו’%ÿк4o*ö¢h«ÜÑ@hß‹´-fø3j\ƒ—ÖZã)¾sûó Å –Z÷)¼ðÿêyÅv5dgr6Åà äcBìñx§ç §5ºÛIÞÚâT;XF©Ö­' L­ÉsÀ»§Ã²–è­_¾ö&¹¸àóàðÞ!GÅ[²¦8ÙcHЮôJJyü´âq ¸éïç䣴m¡# Jt¤ã–7âÊžA!…¡r¾Ñèh£}wTCCÁ)~Í!ÀŽ ôî¿z·ûþœD¢0ç•X@ƒ™ÄNR§”9\ˆ)©Û6&3ɲs­Nì 2,‹’:=Ã}§ùA¥Ä¾iÝqûÇgtÇÇ…ÆCË'\Í)y¤‚’õ=[@½È3ípO.~œ^7ã…ˆÞ^¶@rI„Cp%ùž“â§oßq ×úŠûÂx(‚ žÄL†ß½žëü*P%©´XœØÉwâ3ÙÀYw£R,ÝÇØ<ÕÂ{üÿüì&,,¢Y«iã‡{ç.C¤ö±ûôMaMÇ䘶åp<1£·áÚ<Ó{&uîŽb|oxÞ#NiQ«{@üáC#ËÅŸxñŽl¸ ¤6xÛÉSíüo?ñu9yë¢çè_䦽Â=P‚Iïöp§ßµ^¿çœnsg¢G—Êûº­ÿÉãôÅ_.hZ ygây\O)o “^£-~•× \°óº®™×ßÈÿWcc¿ endstream endobj 32 0 obj << /Type /Page /Contents 33 0 R /Resources 31 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 31 0 obj << /Font << /F8 7 0 R /F51 8 0 R /F54 25 0 R /F44 6 0 R /F11 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 36 0 obj << /Length 1779 /Filter /FlateDecode >> stream xÚXIoÛF¾ûW9‘€5ž…’iS A =h㢇$Z’eµ’¨ˆ’ÿû¾mÈ¡EE!HœåÍ›·|o¡L¢ác’ºT¥¯Wk¥ó<™o¯¾^©Ú[—A4¤-#ÇdáæÃÖ&?·W¿Ã'lÍ„å,âùîöêæ}•ÔÈÎ'·÷ýµ¾RÆ–Éí"ù”¾ÏLº†ïêt€ße6s¥IíëlVX“~„%ZªÇ¬(ÒfƒT:Ÿ`åð˜Ù*]v|ðWàÛnqnÓ†sXjº%üÎæíá†vÍvÓÍ CsA÷€t™-ÓóN¶ä¶®Ýˆ\ ^ éytðs ‹îá d²ŒÉà»C‰ºáº/·¿‚1gƨºðlÖ‘4½s¤i{À±%e–<>‘zg¢&»>uév X48k˜ ʹ<òÆg]èL´ 7œv¢ì¯wÎä#±Ã¦*(á¬Ò†u@ÃÚ;d}·Îõ(ÙS6˵MÛ{|:Ò°™£ô¼Ñ¬–7ä¹oð½YZ4Èž‰çìi ¾ %Ýpn¦ókyáX, ²™/\ú®¥=üæ½1^m](«-¨Cz|ÖÎ0U jx/zš†í5Á ÉŠ@¶¿äS+ 1!ûj¹e-?ù–Ð4Ì÷hvðD®œó¡‚y-ÑžÚE~ ƒBä>,ö2¥w0¡ø8šõ€É‡(Æ£0úyÓ œÎÉže*¬Úû ÙÒ)ë!ÃX¯<$Òé,„:ñª.]…„¥2΂ê*¯…ªéÖþÊ•ñ>‰ÈÞLÝZ7`^æßó½W¬ÝÃÂç’“SU™÷ðø!›_‚Á¦är>˜Ñ}W¿ÜF Øén¬[U€&‡IöÓ,`‚N¹ºJÎDºM u…xÛ$!OS.Ëx••*mý". ˆc÷¿t¤§O'|`­²öB×ÄøBySY+ç¢ ¤Â8@^Õu x(•e¢>ÿiÿræÓ%æL"’Nhi!8ìBAÀÉ݉²S@Ù'\]޳.`úÂÛcIB®êD°ú6¯H@ålt¶¤qƒâ¬(Ì`÷RP ä^PrÂ-U­Š²ü~þaòKîŠôíNòIqã|ìúH¶9qðç°Ð,¸>{ÌVTXaƵlŸ\¼ñ4‘sXöT‹‡Äˆ$œ‚\E[}.‡ñjâIl{¤"€vã““å7¨tíesIˆ©‰k3\L2ÛtD°[ð€ñÆ~¨ßh”¥L*÷ I•‡`¾=—Lÿ²ÛÈÒo)¯’ðÔék`Ô¨\ó¶$T"a”mx"÷:B ³BþFäD‹-[‰ÆsAUÔW'©A˜¤·ì¤ü;3.†‹6N¿ñqo'X=-ñšqéÑE›(dµÖ Æ) ýù1ƒº@ ¢l/ÂoâªF˜-=:içív+—ZoÜKngÀOsÖwLjÇa4$Qû‡Ç¶kî'ágõĬ§ºÔb„Y-œ£±Èã˜õtmB@r€ê¡ZyÚË vNÕ®?´.jX‰Že×è’+ní8ÐsHóÖŽ4­5Á8‚‘f¶ª9ì„ÜWwL‡Äg„œv‚¨¾‰÷”õpkß§Òhy%G¢ËÿP|ç‡#?I x6²Oþ94Èã‰w‚sBvßð‹çuKóÅšš-ꨀˆa+„w­0Úl]¥}3z=Š÷bÔÿAÊ®¡ëÙ¦‹†*¶»’‡HhJ·>ý+«r 8ÔîðÐSSd‡tJ+ G'„÷ü„úHOÊH´ä½y 1nØŽC¦ß‹"ˆÁäŒz?Ø{¯ ¨m»H4ûŽ‹îD>dùmíÄG65ß0È{ÓÑf¬­¬cG¶r‚·º>éoä°å$¸®Ú™ LÆÖ½±‰ˆBž)yEX›»þmð2F­Ç¢Ù×U6ç4|Òô½ŒÑ\Ù¼¯´¬þ³~õŒgQ’V¢¶G3‡®ÖÊVÏjôq‰#g|\sx¡oi`ÜžhPDoºÆGÎ'Õ\6ÖRa‘â°cx‰&§=_d=ù—(¸é`‹÷CÈ› rQ™þ¶<3ÝT3× ¯Ã3^2•ärâ|{Ù=‹Wê‰fƒo&Z_S+—WÁco&€Éº¨†·:%Á„¶ºGü€x /8ÅЊ]^7ƒw!c P»BþAù ÿ+qi¨^÷£jA[?â<œ®'&ý$"v"þ<´\Ò"5²v”[pMzóB4z=ÒΤòÀµœêW"âRrnÓÇï]³œ¯^Øî™öT6º2\ðe*5‹.43:/c#M›’}Ùɱí ÒKäá娪Båß_jUUËæÃ»Þ¿Ú¶–× endstream endobj 35 0 obj << /Type /Page /Contents 36 0 R /Resources 34 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 26 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc2.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 38 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 39 0 R>> /ExtGState << >>/ColorSpace << /sRGB 40 0 R >>>> /Length 18766 /Filter /FlateDecode >> stream xœ¥ÝK%ËuàyÿŠ3$*ÅûP¶é^À0 JL²MÂÒßwî½Vœ®Š½.u,ß}ºúTfdddä~ä×ß¼òë·¯ÿóíoŸÿ÷çü»¿þÕë/~ø–>RJ¯Ïÿ÷‡¿øoÏÿ:ûëß¾ýúï_éõßòëožÿ÷ÛoÙ>ðú¯ßêþ¨ã…ÿûûçÿß}“¿{ýþØÕ>:þt§UÞJ½ÿ4·òQÖû¿ÿù®öÕÞN¾ÿ¼´þQòûÏ¿ÿùc~ÿsòýçµ­\¿yòûŸþe¿ÿîï?ÿÕ<´¿ù£8´üÍ¿|ûó¿*ÏyøñŸ_ëƒÿ;þ³ÎÜ^¥|¤õúñ÷¯_¤_¾~üí·¿üÑîŸú‹k}Œôé/–Ÿûs{þ{|ú›ígÿÍ?zùô7ÇÏý›å9«{}ú›ëgÿÍýŒ‡ÏÿfþÙG¨¶ú1>ÿ£ùg£ºûGîŸÿêû a <£¤}ëeŽT>] ï?öa}ÿÓgïùþSèûŸæçÿkëýÇä§?_ݾâûÏÁï^júèùýçä§?Ç—}ÿù÷ïþù¨|?öý•ù±ÚËÎeÂ`ý'ãOüÝgÔ¶Ì¿+ÏÆŸø»¹>Gªð/ËAû§þò|cã_–ãöOüåRÚÇîüËrèþ©¿<“~Eß×—)ó[+Ù¾gÔÃòœÚüœšsâ¾ò|øŸÞÿ?§¡üL‹³¾ÊÇòñþë_ü_ÚDñúÅ?ýÃþøúç?ü¯ß¿~ø‡ßÿïßýÓ/ÿþõãßü¬ßþ™ŸžÃÞüðùÃÿýÃ/_VŸú¯ÿó_ù*ýõ‹øÝûç=¿Äëó/V—_|6ŸÏñúÃ?½þûë_þ½›Ø3=¥ï&v†ÿsí´œáÿæóÏ9ëG–lÎþQ{d³[›Ó/¬@»¿µçìŠÅùÜrä3òÉÒYwdù˜Îå“wdvnûEóÇ2öl'AÐ…þü¥™ð%»ßŸ‘àœS19×ÇnãùrÆ«,2“¹E®EÚ׈l|fšÀçzvæÚ»³~´®XÈçv{|Øl0 rtÅêÄLÎaw‘Àçp“«*6ç3$ó[Š|¿ñP6¼#»³ùeuwÄoÏåU¶âàó—“ó9À#²ØÀ}øÌJM±‘¶´ŠÌäl‘Ï¥ÑÛr`²û‡Ópà0®Œ!ø ¯‡Ï×™ŠÉYñ5¾²oûŸ´#—ÝŸÿ‡"{mr‹,äsñÎíœ~º#‡Ó—Ò‚•´ƒéÓõóñüÈÀéúY¨¦©èÓõ~~ýª˜Iÿ‡.vL替` O}{úÄé“ù^ö—&Æœ²}eŸsZö”}rÌvŸS®˜s.~ ~.‚ín~nÇÏ쯜—û¹ Nál‹ óðû¯p‡'~ßÛ —Î<^Ñãx'i_'å’|‰#œàìóËíçV˜;ýÜ9£.uóóʽÑv.pñ;xÚ4WŸÃ„ýò1Û¼,\a_Ä*'¸ûx æEd®YºÁÏo û­Ö<“ðs{y{J'OÕï%Ñþ1óÀ‚5Úo#þÜ]¥3\ΣòW.QÌ6ž„;\mç"šË”üüâ6û ãz³U켞žWª4®§Çω‹î¶acžöÅ”Ûñ\Ò~Ó²G‰±…n[ù¹0íŽ(ì7.³mû­+?ÍôÁÕ†±¹øž€pʹ͘»¯Ú£ çãgâž]ºúù²û†r¶_îU:à û·ŸËhÐ6Þ…;Ìã]ü|ÙÊdG—ÍùêY1Ù|/<`¿°”1_ÙJzKc¾²h^œ¯°æVn™öïs{òzzž|«.:ûùxžìz ¼ß=C×›=ó%éâçëYáæ%ܱ\4¯-]áé› ÁÍnƒæå×›0æs[­NiÌç+ùý(¸â¡1/žÏhÏÇ]ÇÓ֥ʅÇs矲?\äí «èŒÅ¸møñŒÆxÜ{AÑ>˳^«EÚ—¨fß±½p½—ô,í×{IÍnìÊ~?2Ûõz;o<ä”Ô}¾öù¶¤é;¿ÏsaTá…¥pIÛ÷x…ý|–g½9«´ŸÏ’ñûO<:šû’öù¨<ëM[ï ûzª<ëͦ<ðpbÿa÷aßN±¿˜“tÅï³üy<¸ãaÞl‘öX;qeJûõj·uiŸïì˜E¸aÃÇ71µ›ßgbè]Ú· lâIÊ•ãÉ&Æ*ñôL¬[ã陸W.˜/xãSÆxznÜ[ÛŸG|á‘„3îw¶ðÚ½Ó¶^öõ‚-¼|<ßæÆW©þà <üû6ìA —EÛýî¶]fþ}Ÿõ oa¿øÂ´ /l™Ÿ‰S9áçU¿ŸOû1æfŽÊ~¿°…»­g…3~ßáOËÁ›Rö`4´}=mnv= ûzÑ<«0åíÁÙ®WaŒ§Îß'Ú×Sþ\¿„žÍ6_ ûzÔvì~/\m×cpÅýÎöBl=\°ž0÷&]p¼–ÏgÁ™óaÇŽ±0æÃî&ʯæ)œ8^mWlKûýÖöÔüxDûzÛväl½qÙnÓþ^êsaú÷‰ö-eó³ð^|ýRŸ‰ã90ʾ^©ÏÄôÌ7Ê~½TL|Ñ û æç?”+~Ÿn»ÚÑë-s©Ò¾TÎGðÀó@}nL6ü¾8Ÿ߶^‰îدªÏ¤6i_oÔgáÅÏß.‰¶ñÌýú¬Gm>ö×:5óûDã­³=¸,a®÷ª=˜i¯õYOæ$ñdöE¸ð|ÛzpKûz»ÚÆ„2ïWæY¥ýyÁ¼¦´?ÔZm½°­¶_7¥Ûqׯx·7Á;úyÌÆx²ç"íëŠý¤hîwÔî¯ø•ý~_»ï·EO¼¨x·¤ì÷Û:ü…}´¿L„Ÿõ®²ßO+ös¢;Çëhvà•q½Ûýº 7ŽÇáû]ÑÜÏ3{ H´?OÕç?øù¯æ~\µ·—U8ã~R'""‚Ö#Wÿžíwۯ៷Ãæz¯Ù|_¿û‡oYQ\aí)ñ„Iž¸ˆä{Ì‘ˆ‹È¾"‹Ìäl¶ªôx {:¹y"*¢ùÐð+&Òöož¸¬Ê#ÛçhÀLÖɨ‰‡¶ém"þ$°kE&Ü*:ÃɈ‹‹•ÜKÙß1^<13‘~؋Į˜œÕwçn.L.—¤¿zä“ÏMF\t<7FúcŸzn2ââá*ŠƒÜCÑL|ŠLän‘'â" ÀÈ_0 þ¡‹Œ©xæÉÒ;ioén2¦ÂþAI‘-•[1‘;G¼<}ÿdԄͼKÑ>Ã#ù éQ1^:ØPÝŠ•\CÑ$Ï„‘S`ßX®?´·1‘S1“þ&ýâÂò™zm Šô‰‘‘-‘Åh‹‰9m÷(‘,éÓæòA¿½?ËÀ*铪-BK$-¶ªè?‹¿ÂÅŽ©O}‚¾©÷ÐÞùÜlØyè'%°“+GV,7ׯéôÁ¿3¢b}ð[ sŠäð6NEߦ~'GŠäÞíÄ0\ô€í·eAAñ<züG "j–ïŽ_D\àï¾å„€áŒàx‰^|ùóØf8a¼üOÿ^4^þ'?‡Ñ“/ÿ‚O‚}’‚ýûFû䓯²£;_î?¶Ñ!Œ—¯9ùË a¼|Í .¹ÝøòÕv©«t?¶Q"Œ—¯÷Á•/·rõ—ƒÂ>õYpϘ›£Ücóª°Ç6YðNUææwÆ[e_¤Xð‹½¼]¹yg¶Í{aœ/{‹Ù¤Œà–àÅà‹âoa¢'ÏG™x9ÝüãOQÊ8Åw飃wjF,t=îKÚ/W ^)Y˜ õ\=8Y×kõ(¢èv‚O’¿|îÇ~¼¢ñò¶!x'¸žà“ä׳0^~?+máÂà°Çv½ AÏ-Íà•Šà’è„à”ñr;cs/wúWn0¿ot>Á'ĽD¹œàî¹lÃìØžM„û¦Û”öe³Ùskn/ïî»ÀÊ8Þ@™¥q¼íî?¤q¼»GIDÏs¼'x¢[¥=¸ Á ¶ù‘…ïWçSçÃVÅ]çcðxÜî Î>±(ãzÀæ„2®‡gíêÁ#·Û I‰Æ|?˜L»]8LG #xôqÏÒ¸>kàV„¹ ò]¢%œ°¹—±¹¢Œà­Åïà- ¯nÒÞz÷ì+#¸tM/Ü^ ŽÃ.º2拜ùb#¸$xò~mžÒ˜¯wÆ÷Æ|½=J1ú7=¶ùV˜Á7<^·;ïwÛw©•q¼Ÿµ¬ÿ¾·×s¶š­Ò8öMçc3¸åvå|òØæá†à˜„à’h¿^ì-(ƒ#¾šÁ×'8H¸Íã)¼Sñ}oçœSýzî'øÈæaO<ëíÚ…<èoÁ‹4^n§ï]p<< 98ṁœ´¥üóØ_>ß^ ®HE­\p<}—p•1º¿åRÆxë]¾WÕ$Ýíã1šÁ[<Ÿ·ó þê8^Ѹްïx¿<^Ñ^žB§ŒàµÑÁCHÎTFðaõäÆèù¼ÖîÁ]Âþ„ ß†oYɼÙ{dÁܽØHG»˜ñpÒ|‘)èS#KXÝd¶NPÕÅA¶®XI° ô[^÷<Ø›ÜL‹l¤ÇÂfrµÈ…©¾ûN– oDw‹ŠôÛŒíÒìHn2÷‚‚'¾íˆû‹¬ä)®ó•¾|ìá8°¹Ú ¹6ÒÞœGúå¾nvܶ™ÙIK‹ôGæ>'u±á¾cFúr ÏvÑ'UƄެ˜T'è7&ÍÝäkÐîÛ™‚ôSèSPÇà¿™1õ’N´R‘ˆE‘ª›¼¬:¢ #}ãiøøM #Y2ç¢_8L໹×™ÑXH‹ê½ÉKÃRš¢? ?„‚•`#YQ°“^j(°{G6,?ìæ9ýÅc~#É(³/d…)–‚‹ô-Gìh &rH¾eDp¤o¶?ôÂFYwŠÁ‘¾Ñ7!{“9©ö8žé…=Dåá^Ï"Û_ðO ³ÈÌXã1#Y°m6Ä~ú´91" É"B_Èˆà‰¡é“ê§ÀÐÅJ¶­è“ê\k|s`RžÃ'è‹À‰Ó‰xak¼yâ…QT-²nED£xáÍMìyk‚¾ìaɽÈBÚÒåfÅ0[¸Ø#ir¤ßè×@Üd`"ûˆd†*‹ùE¶Ã©˜YꯌH†,”ŒìdËŠþ0µ|¸2Ô¸cÉçP¬¤-k/†˜âU‰àø-YÉ=cÉS\¸i²,b¤/M×F¡@íÆ¢‰7'#ÍÓ‰™»èSÁNˆ8»8p±oÏ’ô‡VËažŠ~±ïŠCw±ã6±Yµ;Ðoy7[,w<+úKr¼LôW {øÒô&7÷ô÷û‘¾8Ù*¾8±Ø¡Y³¿}™‰˜ý}âê.ú¤ºyBÓ Ê÷ ÉÛÓjÆK 匠yT æKÑŒ på¶è¹¤Ô—P!ñvÞ ªNÅ×"Âý¸LáÅ cVê¢K#¨,T@ŠFPuBEaOi=Ïñô så†ã‰¹K8áx.TعÝÏñðÝwåþö’FÐyò Ñí/þUÆñÊ^AAÇ+×S1è6ŽWö Ñ|´2,ã•Q¹U8}®Pü®p:P'ã/­”™´1}ÅœOÒ*˜ 3iƒç#šIÛWEÁÌö1÷*Íó±OPÝíZèÝ¥q>˜Tq;q{Ñ*Œö&ã]<º+šY/½”Ûç¤ aïâORÑŸ“2’4“2´!Ì ¬^*"šÂ,©#oiT FP1*(#¨•Fƒ´g.IA­(UÍ xKúðó œAý¨Àù¬„KfÐW@ÍҽРڻ¤¡Š ¢ÁåTxºŒFÒBE…Pár’J<èîvþž42´¿$£Âbó _9¡B)*¼'V`l¾‹¤Œ ŒöÒ1I#(¿µSé³=èI9~¾””ß:*ôD£ÂóãÝ…6‚­‚kšÒ˜ïšg_(c¾C…×èSù± Fc¾³—šMó’n”S¦-h#xœ ­~¢Y¡µxP0’ΰKÝy="iG×c÷R}ʸ-•µ 7^HÚQÆõh¥È‡4®ÇŽ %Áõ$SAé6’Vúô a$­t2‰.x¶·³VaCI-}ã÷fÙ„ ÆhÜï†ß£Ïz™N„±ÁÛeœoKä¯Â‰I¥£¢ÂV4*H›§4*”ϼøÒÜ\š4Î÷º‹Æù¶>Ix¤¥ :ÑLZòèe&-á¥~0·u™Ô¤Ì ÂÛþ¢2Ï··VRÆõ>‚&oŸë%5”q½Ït*@ÝFq‹Þ©ÂçzGae$õØ»é$Ì—ðî)¤4ëÆ‘¥YÁA ÁõT0nú‹Æ|þ؃þ¢1Ÿ[B棰ÙÇ[4ÖSÓ³'”Ñ!Ž£3“†§÷¥PF…sVèFRâô¤eÜÏï"Ì l]hAÛ¨@=7‚j£±ž¶ }ÂËŠ¤3°QÆ|±*‚ö¢q¿X¾Í ~xMã š£YÁ×£{µ1^–gW*c=nû;]x`»Ú£ç›ôx'ei¬qãŽfP±/ ´Y!z"h1ëí¿¸2“Úæ©hõÕãñYÙõ*ŒõæÆxN')m'a>o»§4*L&ÏþRFÒŽ-¼‹paPz*8ÑHÚ1Oét’Î<¨ñ6+ly)Õ!ÝNÒYÓÎ'),UaÞ={kIû–¸—­ÒIw^–%Uigœa$5äz>IÙ“¢¢YtÃìAÑHº°ý”) òì ¿èɤl )³¸W(WΟ’΢+èfPFP¼•ÚžÒ¬0>=i(˜ûi^;Iw$eŸO„‘´Xüm|4_yöôFPziÚŒFҒ΢+“"m?dJ#)²x)sår’Òl½Ìçm{¯Ð—4’ð®DØkAÐëí|*d{„š2*W$m $½Íóù¯NL ¨~cTöý{ƒlAÛÂøyx;l‹¶ù@¸œ¤µ¾…gówzÊ8žÍKÙF³ÃKA,2Æ¢é”ËIšKIx0)…IuÂHêmHºÆõÖ0ž‚;¯çæ{ÊLÚ[8ž·ùò·G+ç“”W[tå~¯%Ý%ív’òº6“ö¼•Bô¹^6*´ c=±Q:˜IÝ–tg×£pƒQ¡\8!)IoÁƒI[É[ )—c»Þ‚¹´ äY›I¿¤Œ¤{ž ï*Þ÷*—cK >IoÉßç)#©,{¤‹2ŽWæù¾Í`³%% —“hûùÁ™I7ÙÌ•+’ÎÇm¾ï4Û~·°Ïwf[ï û|dI¶Þ»}:U´šRFR$ÞgF¯SÝ;,*£B9ªWEÏ“è/V”QaºøûÞèñ½Â¹í{Œ%ÕÙýJÈ+öûƒmIw¶ÞÆx©Ø¯n§Â:’Ò„q>,×§ W&­6¼OÎHŠóVÑå|Þß7*#)—F}’â„ÓIjóïvbÿá'ÁyŸ$*¸û§Úž'û©ÿöí×/OB²k ÐnòÚ!¼³½n6r+úÀ|Î Û¡ßdÖU+Š~ÓkÞîLÐ_iŸl¯Hÿ‡PôD[,7~y((¸#XÁ=‘+„ÎS‚íõ»Oäo2ëjLE_Ûd‘,¯ß}êBÑ—/ÍÓONÖVD6ò›"³xB¿°aÛü{û›~RÞýíoúIy·»¿¸˜+Tºdý°"Û ÝÇ;¸»¤ dmDž,$/6é‡Î+â ú@G~„`y`o¦×;G)Ò_þ÷wåE#׿Ù@Þ‘þy½`aR’ìUÑØ™g$èù>Í£É"Q0¾{³ƒ»Jú×@ÒMAgß®HdÖpxG¶×;ƒI0Ì]½ˆ´Ž)7’L¹*¢ýðxTAÿž­¢˜^ï ¦Hß4Å3» /˜£$XÁº+3˜Ö”ì¯w|A?eþÈ!X˜£Ôšd}½3˜"ý!i°yBds“L ÒE”ÀO¸áF§û ùꓽÐýBa¤/ûÙM°¼ÞùM‘hDP<¢OùM­+úÍz ÇP°ƒcHzRÃ/M 8D~Ê«Šô›õàÅÙÁÑ$8—âi€Ð$ýåáð @Å®*ÉÌ©*‰¼9äU ú EŽadbnTi’ì5’¯ËgF€Hæ7ù ŒôŒäÎDú~r˜EzŽRò'™S ›ÈZîL$2¶C*ø)ƒ)ÒoÖs ò@¤ÿCȉô‰6#‚Ì~JMуV&ZDzv93§ù£FRôÛ±E34Iÿ‡%éi"OA°»I2c«§Hnr-ä p&Eädeœ”HÏ`*¸p"9}˜¡—¡b}y8N.X‘¬¯w¦X¤?r²í†  _°+fp4EdÓDèfyØ/"¿i`ÉÉL±Ú%Óëeé[\kb¢ˆü”Gé˜Ù^‚~`yᦓíU$ûë &˜Á1ÇÆòr±¹M¤ d{EbãÂôL"´äô¼ªŒò¾øÜ>%(2Û«E_|2ÛK°«)úÞm‚"=SÌ÷ç‘Ï…ìÁ ze‰ÀÊ<²^}˜ívìMÿõ½ô `fZ’íõnÉ"˜È­ˆ62ýœÐ›~ppÁÞä«.„,+VpuEäÙÓo‹‚Ì}ó&Q‘œCÑh,¾iJúÅ%‰„¬:Áz#¯@ßxÜþ«zØ‹…¿wIfäYáåÈú9%N¹Ó­hgz4i„Ä&ÏM’>)kyK#¤Ór+†6R¾º‰ 'šun£n:rG¤ Ý·0›´ZÊ\_Úø>HÓFÈq§öPp§W—f_ŸyºÇWÚç¯èqRörÑNôÒ~§ìí¡ã…daa¦¬ù»séFÛºM)u³Y4Æ+ ÄKgz4iŒ×Œ”PåJï-ñʾGÊ®M;Ó}H3%)oÁ¬Ën)…µh·ã­}RÛfÊ R•ñ}y?‰ÆxÍHrV®ÇKã5÷Ó¶0Çsœb^Á™MºŸ”É4¤9^Ñ7J¹Ñ­HדR™ÂåxKs<¢o“r¥m)ÌñÈûp§Ùa28Ó;I#e oßW>)k ó5½{ªI»OíDÏ*Í”KÎÂ…¶çsa„¬#7]ºÒ«Kc¼>¶}4e¤|¢/–r¦W“Æx-¼_F3e)ãÊnS;ѽIc<—Š5’0ŽgC‘¿h„´³ï™r¥k–Æx.÷káN¯¤ã9°¦ÆxFÝZéJï,ÜÎx^xFnÇ[Ç}ÿ„9ž7 0 #¥5¡¯m4Æ3S~…™Ò‹¾hÊHÉ-§0ap¦g’Æx5OévR†ÓÐnôü Ÿ”e/]OÊñ*Ú8^£1^™R¬üN9^Ò¯–r<µ;Ýš4Æ#b‡¤ñïyŸáh†¬”eeÏ…õŒp¢YÕñ6Æcåx.Ç[šã‘÷aü>óY4ÇãÆ³”0R’Ö«Ñ-á9]¸Òìg|»”î1´½–6R¢ÑwO˜}‘²­\ާ4Æ£EiöE,X wzumÏêï …1^ÑàSú¤”·%\8^[=eDƒq<ÑwP9ѶË*ŒñвtÒ…^[ë]‹dHÚ•®Yzœ”÷º´;íó½0Ž·/ •ñ¼†Ø\eŽWô5Tn´ï ãxNÜ¢ëI¹/[;Ó=I—“Ò?—6ާ Væx^^Ò@ùSÊ¿4¾Ï>ãå6öÚÆzêv>ãÕw)¥í÷#a¤´£ï£0öØ·Sù”`»òÛ() ŒñØQRBù”$ðçµhì' n»t¡Ë–ÆxD±RéFÏ,Í’\Ow:WíLï,Í’á#ãÙñ¼ñhØ·v£{ÕNôÊÒæ¥ã¾²Á‰ãµs½"\i?ßÑØëÏãÂöëY8Ó¾ù%¤+íû-ÑcÑ£j7zNmo^ïÑïÏSÂ…ÞI}½ûÆóV4Æ3þC:ÓcJ—A¯¬}JRì.’,ØÒî[;ѳJc¼Ž‚õ¶ð)‰QJtç뮌t¥G“fI ¬‡”Q¢ž²ßÁ‰¶õ¶0J. ôEV.´­w„1^ñ`+ã=PÆ9º*™!ßÇû‚HgºliŒWÛ˜Ò¯ò,Ýh›ï…9^½¤‚4Ž—¿XPæxõ#Ò•>¥Ù/£/´½-JÚ8þѧo5J†HšÛo³$Fòh åFç¬}JŠÔ-Í’ >±I£dDFñèqJŽØzVù”é[ã±2Ò^I¿—QFI ‹Ò[Ú…žSºž’$v¿QntIÒeÓmiã÷ñbÀÒ™^[š}­½v„4Ž7î×Â(4'J¹ 7º5mo^o·'KZÌ…‚ýÂ…nEz}*™"]iŸ¯£Qpú{dioô5WþT2E%Û&î×Êþû,^¯Ñ(Y´x½ 7º évJ¶øýD8Ó½I×I¬’,¼ž£±_¶2î7‰îE:zViìw-ô,¹óØžg•;mÏCÊ8¼ŸG¯SÒÆÇ³p¡û”fÉÞÏ…OIœ½¤ñ®ÇKºž’Kë'|J*Õ©}J&µ&õèö…›4~Ÿõd4î×Ù²&¸Ò3Iã~½y=Ýn§¤ÔFs!áS’ÉïÑ(É‘x= Ÿ’OmJ£äDJ§N0JeÜ…mñÂHÉÜ«4J"$^¯Â•^Y%‹’·~“nô®Úø¾è /\OI©Q¥Q"µs>o£/8j÷JãûŒs¥9ž¥®ìß§¤3‚]¦pæx²x¥¡i/ѸŸÞï„+ݵq?-Mü„OI9ˆ(éVN¸àN¡i¿_F³$2Е Ýš4Æcáõ.\éY¤ë¦×Ðn´ß¯£ ¾/Ú‰*gzi”CÉ=éSr¯wéÔèU…žwÍ~?ÆùB¾¼r¦}=Í’…œo„ íûÑ(i‰’Ò§$àZÒœO„íëuáwIÁ,Í’ˆ ç;ºáûnìw zfi”4-œo„íã%Ú÷ ­™T®ÚnK;Ñ«I£dWe{Ñhç°’‰©jWºîè¶Yr°"Ÿ_%ù*æeµùzOxÚÛ út¼<*½·4J6ÖŽ¶‘ÂU;Ñ3I£¤ŸÅ‹UíL¯"]q>‹¯\èZ¤}¿Î½¥Qò«zbt§gÓN´7OŒÆz­¢ ið)©W½ä¢4ΗÿÃʫѶ_"ŒõXõgÒþû4´ýUN´=Ï c½ÖØÐT¸Ð­H³$gò÷‰Ê•öñûeËhš)ÜèÞ´Í>©·±žk¸"íç³±uj4î§h²)]è6¤±Þk¨"Œ’¬£ÜèÙ´½–ðäý’Ì…q¿l^÷B¿?8+ã~ÉvæÊ•öñûeC;eeü¾h[®œŽ§4K¾vŒGaÔ)Æ|ÇFáÊ…fcÞÛ¿/ç«hÌgøÁÒø}QD9ÑcH£$#J¾*£¤iCUeü>¨±<8ß5ñ.]i6¾çÕ¶q¾…ý÷휯¢1²Jr§ûÐN´íW c>dóse?žXø+c>ă‘´Îù&ó¡=8/mß8QÆ|×Q³D9ÑcHc¾Cí_iü¾ „£ñ<ÁªFÊ…>M¢/c>d»õàÎç‰Îù@¸Ñ~?‹Æz®£Ö—r¦}¼Dc½Æ HÂ(ÙÛ¹žŠFÉl{ñ£’«öâ«kwÚÞç*ã|²)u4îwÝKî*c¿Íá• =–4öÛXOI¹Ò9I³¤3×[Âöû•p¢Wf¹¶ÂªL¸¢d³´ÿ¾¬Ÿ$Œû!+()ŸÏû|íñT ü“®´Åãy€U˜”ÍféÁ~¼NÓúh¬÷[„G£äõȸŸF³¤4çCaü>¨æ$Œñ48 ã÷a‹úhŒ§Áõް‡ÑÏùüêó¾`p½!œé®ÍñÀfõÑ<ßþâ_º/ižoT;Sî4ûÐ߯z‡“”3=’4Ï7ê*ûx·ê,Sç{r™ë}a?^³a>‰Æ|cÕ“6>ÐÂO³ãyR¸Ð>ŸDc¼MŽGaœÔ#Æý̪Hi–TGKvåDÏ)͒騣œi‹‡Æz†õ¤”3]º4Ö3sóy›%áÑ(=˜ùVòÝ÷“„íÏkÑhyÁêQÊöù^ØçB=a¬—YaJŸGe3a¬—¯'áLmìǯŠõ°p¡§6öC—×Þ”.tÍÒØ]¼Þ„+½›4Kî£j0ö«ª¬)7zNmo´ÔÆýj­s¼‚³ð)¹¿8Þ…3íã=zÏûý@¸OiÌW õ>” “ô8ŸÏS»Ò~?‹Æ|µPïFÙ?¿Ñ\ëõð<,ìç“u©„ë±Ï·ÑØOßþbQ:Ñ%Içóù±¥±°+ÖSÂ~¾Y…êve¾ƒÙæ+e?_ÛöÊxßÈšPÊø<ê> c¿i£6›0îg×›òù¼íÇ c½l5œ’4Ï7j<)ãó¨·!Œ÷õ6”Ïç÷ÐÆùFË{aŽT\S¶Ï[KŠ:¥}¼˜m¼+ŸÏ÷%Í–>1Hç×»åEôÂ|döó)|>oÏóÂë|ÞÞg)ã󸞄}>2ÛóŽr¡w’ø¼÷F’.´=o w|uÞ” Ý«´ï'™m¾U®ôZÒõ|~wm|ž×stYt©Òù|¾Ni´,H㌇àFwå‰ý¢ÊzXÊçó6ß /|Þ7^•ç±Ý•mÏsÂã|Þžç”q½q¾NôÖFK†ä‹t¢Ûnçó>ž„ÏçW–®çókiûçsÂx‰FËšì/†•óqMÚøyÙŸ—„Ù&ûó’r>Þƒóë…)ŸÏÛó®ð:Ÿ·ç]åóù½¥Ñ’%{b¸ò8¶|á~lÏÊùxK£%PF½åL¯-]ñùæëUåL×-]ÎçGÒ>Ÿ÷ù*:ã󜯄3½’t:Ÿ_S¸Ÿñ00ž„3íã%šãõV”3íç#zžÏÛó¶òù¼½ï0ê‰ s¼p>>ŸïMšã…óQ4æ‹ÌùF8ÑkIc¾(œO„‡t>Ÿ÷õ”ðùünÒ˜OÐ;G:Ñ<ß_ÍýêÊzaÊçó>_Dã~T ÖCÂöë=zœÏûx>ŸŸIºŸÏï¥Ï{`­rÛt.ÚçómIc½RPÏEÙ×õ … Œz+Âyѽhûz¬ðzöx&sÚçó_Íæf{Ÿ&Œù¢ ^Šr¦ý|GÏóy{ߥ|>¿š4Ö+¬g¦ŒñŠz)ÂX¯ÔƒæùF='a¶[xÆç=0IëÕÂùH¸ÒcJçóùµµý|VÔ‹Æx`½5åóù¾…on¶÷½Ê™¶ýáu>ï÷³hß/¬•ë)a?ŸhɦŒù¢r¾‰Æ|Q9ߟÏ÷!çÖƒS®ô.ÒhÙgÞÒF=åL—)õF­ï‰^Uš-éPO*˜M-jå|%Œã5p?Œ^‹®IÇ õV„ñ¼\'ާ0Æ+ê©c½ZQoIó=¸WiÜ/lc¡i7ºViÜ/XïMÙ¿oóÂ6ʶ±_µ}<°Þš0Ö¬·&Ì–~Ï•žY˜õvjãxÆïÓ°Þ‹Æ|ÐPODÙÇkónnÊXo´ŽûmôX´¯ç…ñ}8^£ñ<Û8£q¾Çc4Îw[xžÆ¿·±‰Æz“õ¸„±ßaÉCó}G=?eÿ}zÆ~Éå²y=³^–0®çŽz7Ê™îSó=ëY ã|uï$&]è¥q=#ñOÏ‹ãQØãÕÍ{jûñf=,aÌÇÝ{”q½uÔËS>?oe᳟ٽpµ2ö›:êÙ(WÚÏG4Ök¬w%<Îϳñ®Œó±ü~+Œù”õ¬”Ïçç”ÆóÜ@= a¬·†_Xʸ^PKÏã(<}öêE(û÷þbW-mQØO÷»z ʉ.KÇ{xÿ8e¶\E½'ávþ½½¤±þ`¢ìï3ëàõë…u•Óùy£hgÚâiƒOËßÁë5šÇ»cPBßõj„Ñ2˜õ¤„±_jŠ4æ+Ö{ÆzaðzŠÎ›^MÇݯ¢Ï~ÓD=áuì¿o4ž÷¦'–*ãxY<ÌîÇ{Kc?™õŽ„±ßcñ,KÇcza eÜO-¥³^{Kc¼M¬WÆz–õ‚„q=Îß'Ï£ õ8„=¥®êûÑÂX¬ŠùüöyÞ_Ý÷ã„ñ¼µx¼¢ù}|à+ãx²žI0Ÿë ÿð-ËÑ?·ôÂ?rz?_ôSÐÁ‰>ÊïÞÏýöõ½óE¿8„"˜ùá9"Xz:4ú²‰Y!‘èH_²2ã#Ò<íݳö":%7ö‹,¦Í\H•ÞÚézÑÄ[G÷ß@߆kï¾³}åô`ô%VCÛMHnÞ¨DÐÃ?˜é› ¹¡‘¾4±2A~!O½Ÿ÷éãxqðÃÞ8ÐOwg‡×@t¥e×@¿Ÿ͉ž32áQ¢û#£ zôzB± /O—å@ôoöÍ$AŸÂí c*ú `{¾ˆ,¯ieV²¢?|²s¤O}̈ôG[×7Eì`T¤}uäÝ\X’v ÑHthF¼~¤/gú|· ûÊJÁ€HûH †õî{õ™|ÎnÇ‘þÅ<}¨ Ô± úÝæÓÿèb函;C z!—Ó{ê¢[y·È¡2wéCetT…ô¡bÍyš¢´&,üðΊ>Tí}“É2c½{å|¥P]ÝŽÑs:Ò·Y²¼"+?Ì( ?ì¾} ÙE/Èö&èL-èÉjkvÅÁ{eâÀÆ{]â@H“SA ¤ÉމÙaýY¤xÅç@*ì¤éC…m#Ñ ¹*‘>¬ÒãŠ, qø‘èËž—>VEçÂ@¿-¿- ¢—jÅ ô£sofveo¾ÀÉ{g™@ôE\|¤/?–—oôå;¼Fúòƒ=\#}ù±xB/ž¾ˆÖìdΊ¾ü`¤ndᇽ[àéË*ÜŸyŠX3Æ6Ò—¦;Ÿ Æ=Àñ§‘ž¶ÑÝ.GqÁFú£Ã®¨ u‘…‹·ØD'ΆªO¾©Â®¥‘èÃ:Þ}¿Ò7|7ò"ýql/Tc d³Ètþ¥¯f²¶Ù‹D£lÊïb²—Ùl±œb±·QŒ5s‰F1ÁTO1×ÛhV‘*~ŸÛ †:Í6…QlŸÁ†Â(f›ÐÌKÅS?Å+o×óy/Æ]Îç{–F±í„`þàÓ,.SŒò6Šù2ØOøÝ,Ó6Ó„>f]Â(É`;a6ãC0®0Š%3˜-˜›Þ¬²HO|Á²Â(¶Éf–Â(Æž "ìKƒÌ`1á„Ï#˜5ø4›Êž¢QŒ™Í…QL”ÁRÂ(vlÅÄ–4›¹•S ð«YÌÍ=¥çùüLÒlv‡fyÂ(ÖYŠ!Þ>ÍbØ,O˜Í Ì"Œbë¶«7¤Ñl¢ XMÅž±+®Œbú¶kߥÙLÁfÁ|Y™Ì!Œbó/Û„Ùl.¡Øt4й3ØAÍ0*‚I…óù|Òé|¾¯è¶O344{ÆõhQGÚ¸Ù¬N˜Íîæ)&tÅâëD²Y4Š[Ö‹òb1l63^çóžŒëÍ΄QL·ùÎ…2›5áe©p:ŸŸ[ø4c¶Ý2mÓf3.a[oh†#\ާ6®g¾l>Í|ÙÌI×sOHæŠÆõŒªmÊl&Ããë™Í¢„Ù,Í’‚ù2"óe¯0®×Ž`a6Ëi§ØÁmÜO{;Å nã~Ê—¹Â¸Ÿòenpc1øîQÅÊh–ÑÑ M˜Í°&ŠIE£?_ö ×óù™¥Ù,j¡XF4Š9wާÛõ4cB0ƒ0ŠÍt’§o£_f ³™KÅ|bìƒã!ó)_F ³Y ÏçíÓ¬Íj„±ÞxÙ#Œù’Í^„1_òe0›¡Dp>Í$¼ë•2Ö+|Ù'Œbúl–"Œf |(Œf6/»…ñy¾£ñ<3Ñ @Ï3æ%ã9î7—í…‘Ÿ9ßÉ‹—ÙÌfžd¹Û<ÞxÙ+Œû_ ³™ šñãx³™†0›UxÖmôiæÃfÂ8Þ« Ù/šÍ ~Ÿhv;É(·1m¼ìæù'Yä6›]Ì“ òÕl6`ö`èhÌ÷Åà…ÙlbdÛ,ÆfÂ(&ÀfÂ(6ÁfÁïböù£ßF±ÒTŒb{¶”¤ßÍZ•F±»Äë9ÅÀŠ!Ÿbé©!Y%Å,3Xü6Šå°™0ŠE0ø@ÅDRG0g4’éÓÀñ¼Í`ko6¥Ù,`"Ø?Å Š £˜aB±ua6s@±ua6k˜HVŠF1À´p½Þ>ÅÒF0~ô»™ÁœÒH¦·dÏ.blF Ìbì(v.ÌbýhÖq»ðõ¯7XÒ(&Q¬\˜ÅìÑL@Å’r9Á¶·‘ŒŸë Ö½ýnF0·ð:Í<ªG×sn6ŒÆõœ‚ £Yl¿!X+×sö?e£Ì(VÌû¹Ù’„ßÍ z“F±†¼ bVŶ…Q,!£Ø¶0›Eì¬ùÕ§fñªÊ(ÖÄbùÂ,fïUq”ßÅæý÷‰F1­ÒÎ÷¹bÚÍ^‚O1Å‚bÖÂ,ößLùÈ¢ò–4ŠÉ•`Òh&Æ|ÄbïÂ(F‚ÍýjxCÅ@XŒ]˜ÍPÌYÅΪW]Qf±jÄijXyF0ûm&ëùÄ]¥1ŸVD˜ ³8š1×SL½tiÃBÖž2Š¥UDןf%Í„Y ½#Ø?Åê*"»„Q,¥¢¦0Œb7ö`0…O±-{°ÈÒ¸TD€ £˜UE3a¬÷PµCÅðõ(œN1î…™Lã^Ò,æÍù(÷«†b²Â,V\ N±hÛï¾ýL£,V¨?a~_ï §ŒbM ó…0¿/šIóû¢´0‹u£˜}ð:ßãA˜ßQQÂü>˜O„ù}P,^˜ÅrQl4øï(Ö.ü.Vl÷+a|‹Ú«Òø>Å>…Óùys6c`±dáyŠ%Ûû a~_4{ÆõÒQ =˜Í ýÅc’ž§X±Ý„1±˜±0Ö›ãù6ŸÝ[ó5º6*£ØºN*ã~ÂhÀàÊb[Í „Y,‘yÂ,fHTa<ë=šaIî-=N±U‹²f±[ÃF±¿áU£•ó)¦j÷ãàSìq ¥0Öƒ,Æ*Ìb¯7¢ü.ÖÚ«4šµ0âOÏÛŒù fH¯Ç«Vi³åùŽF1Úé/.•Û)[Š4ž‡&Š c>™ˆþ¼N³—éöÊ,†ŠÈba\Ï1FÂXÏ1JPÏ˯?„Y ¬0RPÅäX UøK±SaÃÝ~¿f3l³]ïÂhæµ¾O4‹yâzF3¬…b‡ÂåUçsù…}îŒ<ž§Ø§í—c>eŒ p=Å,ùónc¾b±JaÛäñ¾ÝO1LƒSÆõ¾6¾O4®÷b}Âxg1Ja݈ n§X#ŠM ³Ø"ŠI ÷SlÒî÷ÂxÞxlëuaÏb°Â8ž›çûv=Å"ý?”Yì‘ã!Åö7âO…Ë)Vi÷{a¬ÇX¬1¸p¼n+F2ã…‘l˜P HÉX‰ç+šÅQ,08³˜Œ Œ!íóyi#YŒÅö„Û)VgëuaCH óY4Ò¾Š­'&CgDŸ £ØHF¼¢0ŠdÄË³Š §c?ž_í=a—R§XØsb”ÛqÒHE11e$ã!Þ-z1q.ûÆšò8žIÅV²Çs)³˜ŽÇó('Ø‹eEO~ËRÆ÷)§ÜŽý÷F19»RNÇσUô)æW¼˜2‹!y±:e$ #ÞNÉØ–u[¥Q Y¿ÑÜÏ3Ï-dsËÊNÒíØWtùTLJ™ÅÎ&ŽçíÆb!…ã-šÅļؔ2‹}m»^•‘,[üÅy4›=˜ýzF1Äó)¿‹å&]O±!¿¢Y ªØÄÍûEEÕ4e$³£2Æ{åxŠÆ|fÞÒOÕãÕ¢/ZQuUžx¡Ìb+þâGÅšW Šæó ‹í(#År”Y,¦àûF³˜g=+c¾GWàÅÔ‰Š®#ÊãË©YŨئœñûz±¦èuŠùLŸO…1^šÇ›)c>j8ŸÂ?oÙƒoô)¶xEe+ñ«2Šw/¢Ìb/ɞǔq>zÁïs{|/vS¶4Š1wߨPÆõŒÀpeÜÏ,°}3%ÏŠÁ<ÏÛʘﺋPÆýÈ2i”O1Hä)£ÃðxNeËþ`ÍT·:¼ª2Î×ôseŒ×é]W¢ ‹E¡˜„r=¶ûEð)v6=P™ßgØóœò»XÄsãˆf³sëÒ(8½+aðd¼vE¼˜2Ž÷òËh&#µäñBoÿðísY…ü.®0ÉÏÕ¾3³šB‘§ªk-v²E_< ƒC`ÅÖCcnz`%-n(2‘ž¹~±`ÓÁܲC_A¿`Ø)XÈZ"™lÐÞ9ñ+i;Q‘‰ôú_96#ÑKHÐ7ÐÉ(páˆPZÁBzFæEÖ&çÛŸÈDÖɾ¬-Ù§¢‡MXïµÙ1éYpvUô-7¶#Ìdn‘ Ó]c…€ÀN¶¢XÈ>"+«Gà=XäøT#²’öŒ™ÈÞ# +|07=°‘žm1ãá DQge"·õbÂÆCCÓÀÈA¶®XÉ^ý–ˆ¾â7;»y³Ùhd#=,0“Þèü"ûPt¶9ôÅûØ‘¾ ··l;’A4–LVý³£ªAd%ûV<µ4<»ëâxWË`nåÅFZäR¤'®t@ äc)[nFvroE_ròØÍ†Z6ÓŒôÛ%ÒZ}Rå»´›¬ƒÜQ‰!Òo"l‚y“aŒÈìdiŠ™u8lðßdC÷@2ÁNÚÛHߜ鹵yYuTqˆô…²Ðo¶×FÖÒ2+ú…sR^\ØRfuÈB²YøòÒlňº#ldXIÏ4 LäÚ‘|}ZTúK’ñn{Ñ/ ¾Á¼É§ùd`'WW,¤7žºØ°üxóés¬h™È½"ùRy°am j‡°_`"ÙÌò nå|Ûé‹m¾ë¼É:Ǭhéúµ4n²‡Úd‹ÔÀAzƒ½@E}¨›•ÕUg9Í¡.æOMn²2-«—¢O›³Ÿ6: éqùY“-##}R+Ù¶¢OªÖ¬qELªx¬ôEàÄéŽôeíJ(ó{‘‹@6IŒl‡[õ?Ð ñfc…¼ÁŠôeÏj'ÿàb!ÙZà ¹9¹p±G’‘öýFoS1‘}Dò¡{ äĶé˜?UC¹ÉíÂ5O)î‹lYѦØ&ñ&»®‰<ª@0Áv“`%W ´Ìq 딾8X……Å€/VÒK&VaI5’MÚºEúÒtù $Aí·6¢Ò/²¢-ß GúT°ÑÎíæ`%•|Š™^ô‡ÖªB‘~±ïŠCw±ã6Á6…‘~#`› --t#QߥŸà‹™Õ_Š tíEÒRôÅÉæP ôÅÉ^'v÷ ¶¦6ÚVFúÆÒFÜ@$jÒð„¾‹ÎxPôí|ŠÊॳ0’ÂRFÐùíĤTväFÒ)Þ+û!>EmnçSô ¡ã•ð»èM™Â‹EA’G+ûÁ4-¤3ÅæñDÐAð»h:â #i˜å„Qô$¡(Fp?Çcù³‡p{I—St¦fávŽ:Æ ãxetŒÆñÊèø#Œã•ë;ˆñ‹ùheYÇ ÑÊï¢1+ —Sf (7ã•»„1^óDÐìíÌó‘ÑqGçƒEe„q>²¿´ŒN§è :f ó| hBIõyŸ ÀÛ8,Js;q{1#û@Ç» £[ðâñÆKqeoµÆñ.þ$}Šj¡ôº2’6Y”FE‰ :6³b¹¥ñ –èqŠÚì$]OÑ™¾¥Q´©¢hF0+7šK’öe†yáÆ¢•ç#º¢1cIûcH®èø\±“«ßá•QT„g„Q´ ­ £ùR<#»[¹ãx,­E³è Šž3(ÊŠÂ m-cÑáÚiMÝfÑ N, ÕÐ1H؃2Ì;I#é¾ùKÇÛ”„¢;~¾”YtÇ‹f(#éºyvx4ƒŠróèNeÌwÍ£G•1ß5?_Ѭ=kfÎmÌw=á÷‰®ŸŠÚ(£hH÷èƒh¾ª0{PM4‹²xv²2‹®øùŒþTô&uéwQ›–¤q=vÜn¼QG™E¼4¾2®Ç>p¼nW^¨Þ§Œ¢HV}°H—I3Hæ«™4nEmr•FQˆ¾ßA3—± ACѸߡ#IôY/£˜©2‹ÂøÛeœo+ÑZ…toEq<è.EÉÌSE‹ƒv.[PJ¥ízÆù^4EçÛŠæ$av´aQeœo”‰VÆù z¸}І hŽ2ŠN ŽTæùÞ ˆÆõ>½hVô¹ÞgBPNt=E{öFQ»™tpû\ï(Ú£œOÑ?Þ·ùÞ=¥QTszG(eM;A·+‹šL¯^¥Ì¢=ÞÑEóù º¸ÍGaíQÆzjN=E×S´§5am0ï*¢h(Ú£Œ¢£Ó‹~(ã~>=ú*ú]ôhÛzN™E¼:2ÖÓ( lA%(úã6Êíí± baÜ/–ïGF³#• ¼¦ñ´4Ì ^óÐÆxAt–2Öã۳BîŽã¢qÛ[×(—STƃln³ã3‹Ê(³èåPÆzÑÍÊXïcaÝ8-z:Kc½¹1ž„‘4hE‡“0Ÿ·ÝSºž¢4T+œNÑ : fÑSQFRªyJ³(Ž—tÎx‰âÙ)CIKÉ[›+#iE[¢ORÜc›…Y4Ç‹ )çStÅÖ;·QUÿw^ôÄ·0‹^ÔóùÛ,jáE%¢“³W÷Sn§è‡­×…ó):òÜø£'‹J`cH¹Ÿ¢$¶^Î(ºâEE¢ÙÌ7¾ª4Š:<ÞSI}¶Ó…¹ŸæÙïIIe(ú¡\NQ Ú æ+"Û¶ç a$U¯.©Œ¢(ÊÍ L¯.2¥Q´­Ï”‘Tm­çªðIj*¾pSFÒ6Þ•(#©¼zl[tfR{õ5e¨ã%I‡¨.Í¢Áþ¶uI#©µyÜ€2~ÞN÷wÑŸ•?Uˆ^<žhÙ¤ÜOÑ{ÞfÑÚ‚X:eŒ'DÓ)—S´!%áÁ¤rD¶)#i¼ù{Te\o ã)¸ózn¾±§ŒùÀªÿ.a>Ïšm¾FÒ-ª+#鞢+‹’¡hƒ2æãîñʘïQô šAx,z ÜNÑ„¥ÞE¦ð¹¢è€2Š– .J÷óîÕ£Ó)z°ü~)Œ¤Û¾ñ}¢‘T=Æëå¶OÒ¿Gî(£hýD/uÝ®< m÷KárŠØý0˜áµš¥q½¢år9EÚfÒYÁ[EãÆÆ÷ÆýÚžG·0‹:Zd´=¯ c=3}`+c=6ýÁ0º¤|ëPFQ‰çÁÃÖ{Á ޵¤t{^fÒºÇ(c¾Åƒc4‹4©QÆ|±<ÂWEÑ©}Š -1SÆï»6Æëm†YR·­÷…1Ÿî‚ŸwùµeÒµ2ÖÛÈnSFÑŽíÍ©¢Ïõ²~^4ÖÛ«ÇFO&ÅXöm•n'éÙîGÂHúIž}’2SõùB¸ÛõÌõ %Ugm$í&_RFÒNòìÒèÆ¤`¼ïU.ǵ ×ó}ü}ž2’º²Gº(3©™çû6ƒ̳K#),ûû¸è̤ÏìæÊHzË çãöIúËÞtGIMÙ'våwÒ±­÷nu®ãA¸ iÕß—EŸŽÜ¶^ïÒèh‹¦ѧã4“V…ÑÑ™I«Á§£2Œ2“B½)‰2: Wì÷3 Û’Fm½%Œñ‚¦Ñ Ï'iT8¤P{ß|:Ö6¼OÎHªl>ƒËù<’&…ÓIŠ´ñ|Š0éPøôçÿÞít:z{àIpÞ'‰Ê#îßþšäõ½ƒ®§ÆÿÛ·_ÿý+½þÑ‹l“ß*Ÿ®×ï_Öy§½ý»×Oµá=͸?ý5t{ýT‚ÿZöò×ßÿlíW?~ûó?þÝ_ÿêõ›?НÿÇßüË·?ÿ«ç/¼~üç×úàÿŽÿãô˜…âcüƒ?þþõ‹¿ú/ñúO/{’úåëÇß~ûËýŸø9?ÅÎìgøsØŸ½¬8×ô‡$/w¾Èþò3þöÛÿo”±| endstream endobj 38 0 obj << /CreationDate (D:20150701163317) /ModDate (D:20150701163317) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 39 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 41 0 R >> endobj 40 0 obj [/ICCBased 42 0 R] endobj 41 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 42 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 34 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F10 37 0 R /F54 25 0 R >> /XObject << /Im2 26 0 R >> /ProcSet [ /PDF /Text ] >> endobj 46 0 obj << /Length 2148 /Filter /FlateDecode >> stream xÚåZKsÛF¾ûW0Ù¤Üo$±«œÔº*©ìe­œl@¤˜‹ -é’ßžéfI)¶w+ÉÂ`Ý=Ý_÷tõÝͳëWi2*ƒ2‹²ÑÍ|¤Tde2ÊÒ<ˆâtt3½ñ^Œý4нj¬¼…~jý(ý„øàзºéSs®›0sªŸ½~ýìôó6LaÁR?s&²ÒOËm†¡SØóû ~odòÂAyWÔ=Ðu™ òVfÅŠ{ß÷¤|®Ÿ”‰ÝãÞÝü8 G¾RA™ºšÆ“³š»cõÀìÛó*¹:¦»r9l¬n¥¨ýÞt-ØP`°ƒ~¶F3üº¦‰kâÝy9/èåÀª ˜Ò~ì«0Ém¥ìXЭa´e=,hkö {BõÈæº½aúwÆ5Ó_°öȇÊñ7\Y@>«ÑӪرœ—¡µG£¥žbã L ¢ö†·ùލ†¼,ç'åahgØöã,˽˜óh^¿*¬ˆà+ÝÖM?ÒB+fsz©É6~œjÑ­6ènó6T ¼Ðj"Ÿ%òX«(^'¾ ŠÛFæ„§IP‰x  ,µd³ÎšŽ¥)k˦maé Q´=×ÉžHÌlG4£RÄúüu“‹§¿ ýÎàþ ¿ç¯-BÃØ›ždëïÎxSˆ{Ç”hn&„íø¦ÄžÎºxÝ›b!FŽÄ¥¥M¶¦ÈX¹XS4úÛéc­·µöü~ uµ®I©€2¨3×k©?á’àÖ „x’¼YØ¥}ºIù¶é¾å¶U¹‚Bª'QiII1{2LɉÙåkÛTO””4õ¸ój½íNTºz… ÆÎŠE·Êa_iõ$‰ ³ý˜o Pg%aL‰vkç’#Z<Ô»R–â M¹Ì$]~ÆÆÉ‚0î¦h“'‡üšÝû"»JÂÄ«1ƒXÚ>Y? +òXX]Ödº®z-®75”¥ÒFR’ÁÇdðÐf@óì=L˜T\Œ¡¶K¥7¹Sµá*Y[ö(VÂzi*Wœ«ÚÔÊ}`yï¨*ÛH±Q~0&VFE‚I–uS—-q¸ƒ}¬°Ýí‡Ö·¡Š‰1vciG‰þŠo2ôBSKƒ·AYxá6ˆýn3‘ x¥®*¦¤î+úª¸,Yš ¥Ã‡8 ˆ¬Ã™Y°!pS¶8X7'z8>W³Øfð‚ùÿ~â\omM%7îÑ£¸¾*•œÉû×Çnæ±åÃÿ–ŒÏyCkëýùp5ñÜÍ"¾E–·²ÞN’’FAöÂé}‡4TDyéþh2thZÒ=œ òJ<ßúQgÉd6´´¡#@T"µËÁ¢è®ÆF¾ç÷ÝALžÙ%Y Âî¢äŸ¾’QÔ]½lHÀ„ bs´HÊVê̹HQ)2KYj!Ñ*«{ <õ[Ó£ sòá¿j­¨>(xd'•Z%\Ú®ÿèÑq÷” ¸Þ®8Ù·Tºî}\År^žËUŠüT×9ÙŒNQ 6œØ·- ê/<45¬/ü¶Çù¶4oMàt·å yÿ¤*ª¯¿Cµ;uÙ1 Nß¼öáb““4Wø‹nË4æWݹE¸2uÌñNÿ¨l£_êæ¿ôó¥á³´"£u<\Ÿ¿ í_jÚiÂÚr›S‡Àñõ”¼—3êŽ-ƒ¥aëªË=,ü(,SÇÀü[ökï½®¸»ÏPa$½Ÿ3äÂÝçûë¤w¥&ÿñwäÞ‚ÎÕ™…Œ”é ÝòñtåwùlˆnfÉ› È«zt…OÊßÅq¬r¨•‹‘« àÒ!‡iÏþ}óìwJ¼ endstream endobj 45 0 obj << /Type /Page /Contents 46 0 R /Resources 44 0 R /MediaBox [0 0 612 792] /Parent 47 0 R >> endobj 44 0 obj << /Font << /F54 25 0 R /F8 7 0 R /F11 9 0 R /F56 30 0 R >> /ProcSet [ /PDF /Text ] >> endobj 51 0 obj << /Length 1009 /Filter /FlateDecode >> stream xÚÍWKÛ6¾ûW‹••V|ITöÐÇ zhEÑô µe{Ùr×ö:ù÷-ÉñnÒ¢‡…!QÃ~3ü8$Ç*Êá§¢ªÌÊÂG¦Ê³ÜÚh¶žü=ɪBKƒOR)&7?¯Môc7ù ~A• d:Àü~:¹¹õQ…pE4]D*·™÷.2…Ï”.£é<ú3¾MT|Ïòðï&I1±ù&I­5ñ¯‰öq·=€¦…§Þ‹q·a;”;|Õ¨Þqç;zƒUûö ¶:&Wh×Ð×^³†Uïr—ƒøÐÀk‚âîmæç4;¡ÝÝ.@>bàÍœûçw̓Éd?òuGs›á÷äQÅYò×ôõÍ­³ú´É´­€b•™Ê1{Ø:xRxð»JR`¹ŒCA;– ‘È%úL•Ê*' %(rÅo{¢Ïd#²ý¨€VO€9Ù<ê‘ú  å Aö—@+}•¤:¯L÷ËÀrƒ¾ ò?ÕUf5~—¤N›x+YÜÁƒ !« ±¼Þ*¾ækJEéú^¼„-°¦¬Vñ·ðÜöCfâ Õ¬w¤ÈJÑ|®iYI¥úÁ­à”Ágª#mV1NØ™³fLá™×L&Ђ5’ªî”Mü-Ž>„½ŽC†\Áó‡@>vl¾±#J¨ëX!CÛó«SÈg þñ —tüG¡ýq0£«@ ­ûˆ‡b´ò­ ÞH$»çW¿f6(¹^p ¨ŸVÁáÿ?¤@Oæ%ÄO¦zADêI$-¥ÝHH2Å0µL¦w=õ(m;^‚…€Ô2}l¿'aŸdc~qæý×…@¼ÏJ[Ž ¸¹â›¶Þ´9Ô)XÚõ×îb—iô›  ×Ÿ¯«`ðÌr\H¥Êä™öP @¤…÷ 3 …Ž÷œ,-¾ŸN¾Œ»Û]Ò2R ¶Bi/•XÝK»[Ñ€ò(ãë;¸KîãZ&asÕ×I`ô.Wv#ØÑÞ‘6èJ 8J\˜¶‡8,á8bðv1Üe¾ áÙf^ÉàâoMec³£,OmU‡èÆåN+~¡ƒ˜ªÑ““ýȃç¬Þu¬ T™ò状Á‰(±ãfN2>†ÀèÀ*yqÂT‘ rEÄ8ªí°aÒ4\D®X1ëZÙlkÑ3Am8õ¶-W¬nEªŒkLÆ{ p‰ ™*þ=ñ–&¡¦‘@x½gùT!ص§el÷D<=‘4g/ìaI·Ê5J:>öÀÒ"M:w™)Ux>ì=*$'?M'ÿ!Tìf endstream endobj 50 0 obj << /Type /Page /Contents 51 0 R /Resources 49 0 R /MediaBox [0 0 612 792] /Parent 47 0 R >> endobj 43 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc3a.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 52 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 53 0 R>> /ExtGState << >>/ColorSpace << /sRGB 54 0 R >>>> /Length 811 /Filter /FlateDecode >> stream xœµVMKA½Ï¯¨ãJز»úû%„’]ÈA<ˆQ‰¸®‘üýT÷tOÏšE¦GrØš·oú£ê½* § á»oð* ²åé …Ò…§+ø÷ÝѺ;|þþù.Ÿ;‰B£çóå}wxBüÁõ5xÌïûG ¡@N¡äÿl`q}ëÛîÓ:}ø-¦„AÉ€Æ5RBíAz£&* ‚q'QëFªÑh,Hkц6ªeÜ7' ñ¬Ê´šW hmU ‹Òƒõñ²v˜\9ëÇ«nœïþ¹:þÊo?ÝÙ9øÙI8åßm'Ó׿t}AiŠ[Ûäò"/Ð;¸ƒÕk8Ý[U”‹Q™í£ŠJ⫪pG¸•±ˆ¼+N‚\ñŽp-QºŠ÷áï3àõlcÁü{u;‚©©XÖüC¯ãg}ŸÎør_Fß »´§L¦F²”ɶne³ôÂÀ¶­ì ÐúÂölR UÈlÙziÑ£ÄÀnml›NSò…>¼+!y–/q\jc',^%Y5”Ç^“‘2¹"úþš>Þ\ÁÍÓöåaòi}² Rè)~álqòtÀf ‹‹Ëß¿¶÷°½†‡íÃËÝEŒÎa}šáÿ÷ íˆ,×_ÚÚb3½W°«Ø£ÓT#‘Ø b‡mÔh¾ï6F5RÙé§A'm¢*‘’/¥Ç ©±/2Î5$7¬ëšM: û—9£Žó\`)nhÖ\ ¨˜\ó\\ñ‰æ¹Às…ùys³¨©e.àÞZ©y.àÅ['¯<˜Pt8™™ÌÍÒºy &¡UüYÁ>D5ÎR°ãZ 3ÌÜxÖ¢`v+ag)8ÈèÁ³ÌEøÊ5Þ=ìEÿ“¥EòB)”®ÇÑ.œ£$‰ˆÊ(,3À%¬xTO¨xœ¢DÔ€—°âQ®â9pÅÅNu{%¬¸±ñðžÃß9n=ý€¿³ßsy+*5Û˜5#–fïL¶(6ùÙ,M˜,+­Ò03°­XÚÉKÇI.ˆ1Û,íäµIë4W l'–nòÚt´ËÛ,Ýäµ4f|k^,ýd¶f3ãsñaZ÷"s endstream endobj 52 0 obj << /CreationDate (D:20150701163318) /ModDate (D:20150701163318) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 53 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 55 0 R >> endobj 54 0 obj [/ICCBased 56 0 R] endobj 55 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 56 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 49 0 obj << /Font << /F8 7 0 R /F54 25 0 R /F56 30 0 R >> /XObject << /Im3 43 0 R >> /ProcSet [ /PDF /Text ] >> endobj 59 0 obj << /Length 1840 /Filter /FlateDecode >> stream xÚÛr›FôÝ_¡G4c0 ˆ´Ó‡¦ñ4ÎtÚ¨Oi K" ,Çê×÷\a18ÓñÈÚ={n{î+³áÏ,ò,ÈÒÕ"Îà ´vQžn¾ÞyÅ–œ%!ÀÝû“]üÒÜü zä Kßáùóúæî~µÈ‘]ºXo{±é*0Q¶XW‹ÞýÒxøì.-ü¯—~åž}³ô­¼"ø“`=-“Ô+ŽŒU^Ò>-£Ð¯cè¶iyñ°ýÀó×$C;Ü4—‡ôÒ‰˜óŽ-òý¶Œ2¯&÷( ùÖ#T°.PÀ>½B/kd¶oø²d‚oȇ5£}×E¦p<’|ä‚ô·ËOëßÀÀ¾1Až¤l2"G!ÉÊkZµÑ‘i|!& ì«ÊRbŠã• \:Ë7Ý­Û;ªê™Ì,ý$ ½µÚö5™D’Ãj²8Œï©¼Îiâ>£ÞÝ£x$au\ŽÝ Û?`‰5é/dytp‚‘fÛO]hŒSÛØrCè3Ö8ÿ`QÇóIÌ ¾ËýlP £¾ÛDÞV}}dÒJŒõ}'»jQåÖ†dlÃ7ºÍ»©z>–ƒ¼èÄLÐmO“aÄÁMRV¦ì¼-›ÓÉÕ»3ŽÐ¥ c¿…WFàüDÓöÖ.ÀÜ‚È|™ô=lŸ â,"퇴n1*¬ -'*Œ°}âºZóÑ´z#’Žˆ¡†+*­ÛWFs`}»8ýÁ4>õ×c'ªÊû>˜«±n}%·Q¯%”5ý D"Zu7º[¶NPŽæ3i ¹…-—.Rlä:5SŠ÷,z¯¡"J«-7F™bÃ`W047d…ÈU#×ÏÒÌ{f¤¿˜$ ¨uwŸXç9™(°á ”å)³Á•rœIÜL6O”€³µŸ3f¯óªð4 ¢4we»ó îT~d&U¢©X<¹ð\©‰y€˜¶Ájµ;Æíšúd’ÙÇâ†#MÀ’›lP‘žšÖAi8ÓòÌç]уt*# C9õ>ŒSoݡ֧bb•²úŽS9Ó "ÏŒªãA+Af#tÈÈêNø’(oµG?3îI¬–{þ`^!çt³‘‡Ñqbh‰–i—‘4C™ó…ž^ƒ 7Ò™ÐΠŠ"j#ý@…êLB+Êc_Ç_ÁO6Phx7”> endobj 48 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc3.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 60 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 61 0 R>> /ExtGState << >>/ColorSpace << /sRGB 62 0 R >>>> /Length 35139 /Filter /FlateDecode >> stream xœ¤ÝËŽ4Irà}?E-9 þô‹ùmKB@@¤@³ xDp(i"__évŽeU…6¡…D~ü£³2#<<<"ÜÕ¿þ¨ÿøñ~ú¯¯ÿ÷øoÿé/?þêçŸÊRÊÇ×ÿÿç¿ú/¯ÿëÿúÓo~ûQ>þî§úñׯÿ÷?Õ»ÁÇþ©Ÿ}~àÿÿÝëžãÿôñsúg—ýø×S~ìöþWêý¯ÕÚ¶ßÿüü÷ÓïW{ÿ;ùþ÷fãG«ï~þû™?æúüwòýïÝöÚ?¿<ùùï_ìçoÿû_þš»öoÿ víþöŸú‹ÿØ^Çá×ÿð±ðÿŽÿÏÙ×j­ý(ûã׿ûø³ò«_ÿãOÿá×þ¹ê?ÜûÇ,_þÃöKÿÃj¯ÿ}~ù/íÿ—§þíË9éÙ^Gõì/ÿåþÅÿåyµ‡¯³þâ=Ô­ÿ˜_ÿhýÅû¨Ÿñ£Ž¯ÿé{'¡!¼Z‰}¶õ¶æÒ¾œ ïöf}þë« ŸõþWèó_ëëØ~ÿ3ùåß÷¸_ñýïàç¿·^~ŒúþwòË¿ã˾ÿýó»Ý+ŸûâÏ?wP[?¶}ÜcYÐXȃñ'þÛW«µÊÿV?ñßÖþÚSÿ±l´ê?^¯Ýhüe»ýÿqköã þDzéþ©ÿx•{ø½}î¯o]æOÖêývìQƒíuhëëÐÄûÎØøgïÞÿ?»¡úêWÿh?¶·÷ßüÙÿøÕí(>þìïÿæ÷øø‡ßÿ¯ß}üü7¿ûßÿô÷¿úíǯÿúýúWÿôÚíæÿÿÈŸÿïïõñçýõ¡ÿò?ÿåWm|üÙßüÓûó^?âãëëÛO¾ÛŸ¯ùñû¿ÿøïÿüo]Ä^ÝSù·.bÑü/÷x7ÿ7Çí¨œ¯¶¦ÈÏQÜý¾ò£Åu@k’\EqNrIÖËWûŠ£m*Úo_)ØÁ³»ÿ„vÿ§¢6Ûç’—ýÅÖ©X+غbY—ö£Iÿ¨áWå'½púMÜ ôn5qùŽ8(™¾ñº_Vpn° Inì»=qtptE;à’þ_C¢"ÙÀ;(ÈìÎsŠ`»»}”{Ø;Èû`= 5É Î®è‡{Ô»û;èÍûA\§íGÝŠ{ƒÞ¼3è»=qùÆýžÖо±ù5!ÓÃ0ìØÌž®8|×ùY±[Ñ»‚Áæè'û‹³+6þ¡µ$ù‡NS¬·Žy‡éŠ´­Xük¬û? 8kfCW?öm‚8ÜûG3Éö­è‚÷’à1Iÿ’çG©Š~™M%³{*zc˜Éd°š¢_D^lU’û7±/Ðwb É»cg½ÍM°ÞÃ=.|‰å€½Kp˜ämf³ß2ñÕ}LpÉúÉžè i.|™ôšèMåÅ9½£˜<Ù38MaÜŽJÑ7æI—h œM±opwIß9 %±MЊ¤oüºx-źÀf’Ž.yðë–©X&ØMò6Q4¨Ä‚¦²ØÌ2ïO¸—¸©èÝÈÉŽ¥èÈ×иIŽ ÚQ4ãÞçAÿ¨…“.Ñ//î&é_r£»NôaÀ‹>dÊ4°7E,)Ñìà4EoH{ú…O°ƒÜíO°mÅBö‘ùºÀWÐYâ¾ê^ÞQd¢© c4ðz3§oŒG°ƒ÷ÄÉÜøÞzrã{‹—iÜ[Ò¿¤?|ôÏÍ'óîØS|ЛÙîGzOzA¿¼Øªd­)zWÿâZ’Ü–¹ÐÕ¿xª¢w#LJlŠ·1¿}ôÆð⽉䯣)úéüâ1Iÿ xtšé'û‹}IrãÕ;7^G²§+úƒ…Ã6s€K³€{(úEä,¿, °›báÆ~Òerã³2'.1‡§dâž -Eîí—TÁNEžãÃÚL?Ù_¼ÏïGÕRÑ»eûé~mMÚOékï•…ñyÍG™Â Ÿ×oÿ-]é;ö¿ó¡r¥û–.±=÷Vr§ÇÑ.î×m { ^OôLÂ…¾Ï@…½™\{ÛÎ^0[¨p£½ÿÊžø½û>ÂV‡ÞÄFŸ)mØþ —öï{Ÿ7íBû¸-»cûŠÞL¸Ð£H·Ø~šv§×–öaÇõ½ÓUöýQzŸì²énÂv?¾Ïo„w|ž÷ŒÂø½†‘@¶_HÜG;¶·)=ÞUzà÷ðz’máQµ-¼µ }Lšíç“p¥§Iû´ÖåwÂuÐÞ_ WÚšpãÉóMØÂG»ÐÞfïAûS8¶÷ëPöŠíwÓ®ô)ÒÛ&®ÜhOfûCŠÚx¾ —ð’F{yyumlÏþ"íååv¤[xUíFï!]cûÓ´}¶†ëEv™´÷çÂø{=ÚÃw7¶——½=dãzó²·áø¼ûìCx-ú,mlo¸×ÌF{i×áJ÷*=b{¿^ Çö»H¾ïĘVûkùSa¶—…;aüÞ{„l¶'¿ß“Æö~¿¯\Ãˤ1žéÅŸ·%W b¯}eïØÞ÷W¶ßÄÖÞpçãÙy½ÊÆx£7Üe¶ïè³{xý7ú>ÉÆx´ú;aüžáo„1èÓŸü+÷ð–Æùür_Ú•6ö›iø>]Þø= Ï>„ööœ½ð{6Æ3ƒC»ÑÞÞ³1ílïÂöëMöÀöýY6Æw–ÁÑöïcãÕlœÏ/×¢Û{”Ýbûû^A9¶ß[ºú÷µŠñ’°ïk¸¯~xðá¥{IÏMßñŠr£ïxXxÄçÝë²·§;G Kã|6S)]iÛÒþHâz mìó'tÂ8ßÍ–ÒFß·‘Â8ß_¾íI¿gx{.‡îU;¶ï[xãÕõí¿¦ÏÂ^›¾íIxžðÖô}b¤ì×KÃtKá1é;ÞRŽíïýœ0ÛËB{ÌÆó Û÷’t§­Jã~ÀÐ)ãû´—lo殌ë±ù‰—½x?ðò½)ûß>×MãýþBφ_˜•q¼ïL£môžÒ#|ïׄq¼îlŽ!í/"¯ÛÑî´·‡ìÛ¯ªïƒñ„r¡½¿ÊÆñ|ùTiœ¯Ão¼³'ï‡Ï´’.ô½_ÆýÞÿѵcû{¿.Ìã=üz£ìýýðÙzÊl¯(wÚw6®×h8Ò^[×óÛð6ŽçÆñ.ô}Þ"Œëýð7.Ò^KýÅàù.\éaÒ5¶¿ãå^Ò¸¼“#̉”/eŒïÀ°jÇö÷~Nד;pîÚF#öto,º6¶÷¹ÊàÆIzÒ÷y‘r¡½=f£¿š ûñ¼7æKã•û`âhcšg…ñ¼ò>xÒLŸ)Ï›þ<\Ï+ïûŒ¥ý½ü~&Ù0÷ÚûkáNûþÎFtßLuíØÞûûlŒG®ã¹1ÉÆû’ûörHãyÁd‘ëÕ}Û[´±¿ƣ…¾÷[ÂmÐóHã~áΞ0m£½=dãzµØ_<Ýy<ï\•¢ÏógéBûx6{ÇöãHã~æÎ”šÚ±ý.ÒèO0ëPÚè¶¥ñ¼ûÎ]Ú~>ÝÙA]׫;×¶H÷°mm|ÞðÙ°Âèîœø#ëÉQ?µ±=¯×ÙxŸvguíJ› s:g]<ß²ñ<òåûg£½ìŠëÍÓ•ãápÊñy>ÈÆóÊÝ|vž0ú‹—ý~VxÐsh{µy=ÏÆõæeoÏ•KýÅæõ<ïS·_X¥ñyÇ;ãÙÍþ@¸ÑgJc<»}`ªŒ÷wvUÕÆçm\ï³ÙúSa?ž§à~ûéÂöò²ßO ûñ8Ïc„kxIïØÞï·…cûY¤ÑŠû-aüž†þ4׫—{ÑÆöãÝì¶.þætŒ§„[xIãzu:ú[a?_î\¦*þä°½ šÇûi WÚϧì½è{=WöíkñûAaÏ^ßñ‚°¿o»Þ]z`ûŠþ0í¡ò|Æ÷ñÏÊ誌¥½ºtÃö<Ÿ²q¼«ß˜Iãó,ŽÇÓèêðû5åFÏ#ôO+[xK£?¨>Vynzm£w‘öû÷ÒÆñÞþ¼XØ}Ç3ʾãma¶ž¯Ù¸T¿0)×ðXÒ%¼‡¶Ÿï­ =<=x¾7žÂ±ý¯ Œãù²·alÏë]öšôОáû>I9>ïÞ¸ùýš°?½žS;¶÷ëevÇöX®Üh3i\/Ïwa£W“öç!×{jÇö§K£?hæãådÎ÷ô ‹¦ÝÃGã…;ßiHc¼ÐúsáFûõ$ýAÃzyåFßñ¸ðÀöìo„;}Ÿ Û¡G׎í÷–îØžý‰°Ñ>^Ïn±ý}_ª<èÓ¥+އ\¥+½ªtYôÚ±=ÛËws±ü £ñþH¸Ó÷y¢ðŽí}<’½Â÷~Mí¥óz/l´³-¶¿÷Óʾ¿;ûaßÝ0ÞËÆõ¢ó|nôªÒ8Þ磰÷g} ¿ÏFѱB_9¶_E×£>q¿%\éÝ„¯Gkñ•Ý›ôá#½âó|<(\èeÒ3¶¿ï…öׯõF¸Ð>^ÌÆõ¦ûÄånMšÇÛo¬¥}›¯mUF`X£/\ÞÞÚ¶&Ìùœî£Û{{Êö÷½÷±?b£}<‘½b{Ïúmßß÷‡™ô´ßÇö6¥q=1$‚(}߇ ÷M/“æñh¸=]bû‹åöûÙlô·èø¥ íýYöŒí½½ wúÎ÷hãílîÏö.Üh;ÒÜŸÈ)Pná#Ýb{ooÂØ쯲q?ˆô·†<’§Pßñ®r¡ïý”ðŽÏ»÷SÂ<ŸA'<è;^Q.ô½Ÿž±ý=ÞÊÞŸ ôWÂ虤\è1¥q=HÜRŽíïxDã¹ûâcK£=àÅ4~úaô×x1¨\ðyÝßw$oöÇ™%Êöö’½c{o/ÂØÞ|>†0ú ¦)ÇösHûûŸxq¤>ýr¡Û–î±}ÚSº…ïxJσÆSÊ8ÞÓßç&/>ÏþàCÚèû>W×Ó´áõö’žá{½U.´·Çlžì¯„ ]‹4î·0ñV÷[édÊÞEã)¦2 ã~¿¥ýû0?I¹ÐÕ¤1Þb†’r£§ Oާ§¿Ø“®ô½ŸÆṳ̀%aœ¯w>]‘žá;ŸY¹Ó÷z*Œç½“ýa6®çLsîá¹µ½½ÍŽö”þxb¼£ŒíÙŸdãyÞÄxC¹ÒC9ž×Ý…ÛGÏ[¦8åuÂ[ßigÊ…¾ïÛ…>‰Iʃ¾÷[Êø<ä ãyþ*±?žÆó¸Åöœç%«ÄþJöýÍÔ&a¯›v´ Íýõ4Þ×Üœ¼)ÏÛnÊ^ׯß3ô'Ù;ìýU6Ž÷M´¬ÚFßûIåB{’óu±=fãyÛBr§­Iãzûò˜Ú•^K×Û… a\o—'žHc"eC¸º íNóx~7ã"¯ïófaô—ëøó$åFßû5aŒw˜¥\iïï³1ÞÝÈãP®ô*Ò87’Í” íçS6Þ§ÝùNÚ8^»c<œ]Âëß×|>MrãxxÆËÂöó-ýñFVŽ0úÛÔe줬 ãyÔf{Æ÷Y¸žf³= MM¹ÑµJ[lïí]8¶]ºÇö³h—ð’Æxl#ïC¹ÒµHc¼v½´cûfÒ%¶_]Û·?÷3OWŽÇOÁýp6Þ·2—J¹ÐÞßfc<~š?oWn´×²ÑžoÙODî+ãóB%Œóù /G¿gxÞ›0žW3J÷»Ì}Jæzåk¿ßËÆx÷ð|nô}/ŒóõøÂ8e¼o`Æ“0®—yÊFûõ,ãჼeüÞƒþ,ç#ª_Hßí{A†°ÏǼöö.Üè±¥kl¿§6¶¯q<Ÿöça×<žßÜø<¬3Jx‡ïû,eü^œOÂkÑ÷~G9¶¿íAØÏ×~󪚶Ñ÷~[xàï!çM¹Ñ£KÛ o+Ü7½·¶ÑgH7ìOœÏÊ•n]!þy?É›1þç»r¥‡öô)úö·Âþ¾©_¸©<}¯ÇÊ…¾÷s£‡·´á÷¢¿P6úhóx­Üh›Ò<ßÚ“p§w•®3¼µýïÝ<¯#í×ãë:…ù¼¬3ïKÙ¿Oõà å½i¯V#lá£]èQ¤}<è^Ú±ý½ßöñ`g^˜°ß¯÷ÚP(í…yaÂ=lM»‡4ÚCEÞr¡÷‘®ø<óñªp9t?Ú…žExâ~ûÚû+alÏþ*{7ziôy.Âo¢=eóx±¿ÆöÈ[¶CÛÑîô½ßæñf½œlœÏÌ®›nS»Ðä Î'öGO8_ÙßdïE{"ìÇ«±?ÉÆõ¼±?ŽíïxJ×ó;ŸÎ´±½\•q=oì/„cûU¥-¶_[ÚŸ§ôÖ0ÊÆx­ñ|ÏÆñl<ß…ãóV‘Æñl¬#ö4Ÿ—õæO¤öñ’p¡mKóx#ÏE˜Çy…¸ž7ä­M¦ïËó=›Ç yƒÊEÚçut,Ê åê²ë¡­iÇö÷}—pÙ4ëm%ÚÇ;OwöÇ y)¯7äA ¯°·…{xIó|õ…#Ê<Ù 7z.i›ô>Ò~?Ö;ò¢”;]—´¿ÏèÌ[S.ô}ß+ŒñXGÞ¡p‰ï³—pãñêO ãï!/JãñÎþ&dz³¿6Ú¯gÙ3¶_K»Ò§Iãx_iŒ·;ò\„û¡ÛÒŽí­Jc<ż9eìOäI WìOöWÙOÝ<¹£mt/•ã펼åBûþÌÆývGžŠðÚt+Ò߉ ]Çó`<™=&Ý»vlïçk6î§11KºÐ§Kãx3oM¸ð’®á¡]Â>^xšëYoÉMoÂÞŸ™a¼—ñ–!OD¿×'Æ+ã|µëm6úc,ööbl¯Ùí1ãac{ÌFl÷SÙxÞaãáB{IÓlƒûõ‡ëáù4ç§ìß÷.¬^Òþ¾¢3/Kýé@Þr§Ç’ÆùÆ<+aÜÏ xCºÓ§Jã|Bpˆ2Æ3íQãÕ ³´ý|f–péô\Â\¯Ù<¤Œñ ó²”ñ÷PÆSçÃ@ž0·;_giúÞc¼2ÐÞ…q=»óqŽ4î˜g%ŒþˆyVÂ/NäQ(º7áÅñüD^Œ0ž·LôÇÂh¯†UÆx~¢¨©0Æsys¸ŸžÈkPntÛÒèï§¿xU®ñy~<²1^›¬#ýiÞ_%³ …ûH£?@0»ò Ï&çÓƒ™”q?4y¾fs´áBÛ”ÆóƉ< aôÇy5Ƀϛ™'%ŒëÝô…³ÊxÞǼ'aŒ7&ϧlÜÞJB&çí¨Œ¦Œö°‡’lìϯÙ8ßPýPyáïuüžl´‡åÁöÊè˜w$lø<ð¢Œã½üÁ“r ó÷>Íýáæ²™WÕo­³)ñó‚„q=\¿'ãÅ<aôÛkq*c<¿;úó§Ç Û s(ãxmî¯l ßçqÂè˜g’̈šÓ¬[™èÍÍ »=kA2ý ˜á <Ø1б꿉~›mﺳbDzs"v,V°eVò>Êôî«3žäà ÃÚÐL5d>E_Ðo,í`ç$¢3Vg6nìµQI÷]áõÁB²†ë7rºJThND½[V+OôG¡7Òh+ú@L‚ÞME•åDÈ8|2• ª;÷¨ú¼AUxñÌgçdF…fß±‰þj…+2}à1FÔË|ЇŜ՟é'Âs 7ö&ú“ z8_?Ó_G TâÌô‹éÀ\àLïÊ9Ó>Óªc£ÎCbãÆ¬ˆù ª;c Ñ“§Õ¬¨臌µ3ýöˆ³Ë3Q ·á&¢ÞmGe°D¿ñ™Xw“éÓ"où,I¿)3¾äƒ~Ë3±âìI¾p>x¦'8Û;•ƒY‹1ѧš` ú‹Š‰šÓ™FΡè©QùT°qã;BÊD ׂúK²fÞÂìþLo «£vP¢7†ÕQ·2qrc¯K•è%pñô‡º‹}ÊÓòô+é>90œdEÛL?  kU2ýVö&=nE¿¤nÌÃÏôîz³æe"ª´vT.LD¥Õî+h3 ¹YžŒ³s3}pÕaýÑgîf¢¶(æÅ ðÞµeú 5*¼>ÈH„¨ášèp¯w½­ïôÛeÎÖÍô›½¨ášèƒÎÔÍlÜØë¸%¢¢-f©?Ùð˜slïÆ‡ }8Íù§™þxê°º]"j‹ò„MôNõÎL-Š>(:^€LÐo롪W"j¸²j郜²•GQië2ý¤»U<‹"Â~K‰¿ô´…½$j6¾ ‹…e³Ø%‹f³Xe‹bFßa•¥G1œ§Y ²ã÷d#Œ?Šmf[˜ÅjžFØca1¯l„…–boÙ,æÉb‚ÙŸÅ8köˆ°Å‚ÉüÂs-((Œ°VNö~üÓ„_P¬Kaßœl'Œb“q…Yl“Ù„Ö]1Y6yG±AL–f1AœßÂ+¶÷°Ùl‹Äd1a쯊ɬÂþ2ç{ô°Ýlû`1Faî/L–FX(žºg/†i6SÊF1¤ÆbJÙ#¼Š4”Y,O˜ÅÜ Å|²YìçSòd˜6'³ï°‡¥g#l¶a²š0¬ï¬Ç%þîÎúÒ(ÆÓ0ÙL¸Á˜Ì!Œ0؆—mÂ(FР¦ŸŽb¼œì Ìbg5ŠE<0ð^£XÄÓ,FVq¾f³˜Š= £?èÅžFØ?‹Õ £?è˜ì#Œþ²/„™?Åfo*‡6Ú‹™ ϰ‡íg¿‹­yØq6ÂÈïËø#Íb[xY*\ÂëG±ë(–í·cïb\Ù(Fa,†“ó=Šae³˜^6 #,;Š9e—°ûyº±?,Æ’0ôÁý™½Â~=Íf±KÊÆñâË^a¯a6ÿ4Š ªFvßc诳ßż¿Ìæïa1¤lšÿ4ŠMŽañO³˜ÔB1«lSàË^aôç¸ðd—(µÑgO´¶§l[:†þ4Âäg‰0ó§ñ{ù2S¿÷VýíÒ¯N¶‡‡ßaÙ|)ÌbA8žÂèoX¬Faæ/{„Q¬‘Å^„Y,gGöÓ(v1Q "yG±’‚0Ma?ž|Ù'<Õ= C~×[û…½ƘÍß˰ûlüÞSü~MÅ’X,Bã‰ãUo³-ŠT„‰eãx³„ðŠí=l.›aþû+׫ƒb'ÂØ§EØéÓo¶‡l«==ÂHŸf1C˜örð²?¹G1¼ŒFyx> zui<¯9;Â@ŸFpv„}>ÅÒQÌ aQL ›aò<ÞO7.6/<ÞÙƒ* a¹Ù{) aÝÙXl\x>g³C0˧YLaÈÙc)†ëåÓFYì{X姦вW|ž3ÈFNñªvÊXŒ]æ;Üña„ ÞŦUšÇ{!ü/»†›I#¬ 0lýéƒ, [ÏFØñT1eîó5aåDØâÓc*,’°¥Z0^Ëf˜>ÃγK˜aŒßüC¬(Ö!Œ0Z#lïé¾ã)a,æg1a†Ñ7„…d3l¾c±}6÷G°Ä§ù{ ‹åŸÞñ{ adÙü½ayO³˜ƒEÞÓ ë·»{ýAEX±0‹1 „‰e£?¸³MýAÝþ¼7y1Œ©"l[xG1Ó¥|"Ìíiô­ lJØ¿Ãò…Ñ`U‚2ÃÀ{üž§F×ì>÷p GØÜ7óyž¯*ëÒ;Ìp³§ÑÜU‡[šûÃgU*#ŒûúH£?dØ»0ŽwC˜³0ÃÌ}Õy6'kF»ðаõ{?#Ì0é‚°•l„aõamOãx÷Š0¥lŸá©+]šaߘa–lÆb Âß§EØÕÓhï7•Jdzcv0ÂŽ;Š3 } ì.›¿3»’ù¼Íî’æïYø=Ùï°ð¹¤ÑŸ£cP¶ß[aYHUF{î ³Ìæïeoö;¬ûTá£2¶·l†1³?ÊF{5†Éf[xi„1a`ŸÍÅb³þ„1^µ0¬l†U³¿ÈfØ0ŠIðlÒü=³F˜•±=dWüÌŠJæûTð3¤ßa̾¿²Ñß †f†%i„)†%f3 µ"l*aÕƒaŸ7.Fò£SzÅçí.ã…ªŸÊo{Æù„ª·Ê¸_BUäìaº˜(Œó‰aÆÂ8ßÚs2ŸW¹4îWî‹§)°tT…Wf˜.f 3¬Å „q¿É™yÉF;0U˜a²³ÆñdX¬0Ãb ‹)³q½œaO3ìuD˜ÆÓ5ÂXù}¾›Ï›®÷Æx3Ü”ù{F)̰[̸F{™‹½³ù}w„a<ñgü%Ç{œó'ÌßsÖ”°3†½ ÷{-Uº††ð4Ã6FœÅff c¼ºÂh²q?·0³Xã•…9FÂsæ,Ád¾/óYõSý5g £=0 Ua} ;ÆýÂò‰#ʸŸ^;LŽëñ.ø>Ùèv‰ÅÞO¯åbå§1žzyviìï»ØEç3g&s1ó ëäbã§ßaž»H3ŒsÇç=ö̰Jáöý]£eWƒ½¾ßGßç”Xìûôˆ°ÊQ¥Ñ_ÌF{a¥0®— “LÞÆØ°˜/{†Ï’Fz,}aÀÇ[ûóø‰Ÿ½Øß´a<9˜*Œþü`>‘0÷'ÂØ„[„1úb¼l„­p> 0Ãò”<# Ç+›aw ÆbØâe† ,vËfX Âö„±˜ŽóÝ’#ì©øÀJ‹y ÂV„±˜°bö©0ò0_Qد·7,ÌÛSv‹°1›ÒXÌ\¹?Ÿæûœ6§ôŒ00_Lœ0ƒŠ° a„1LL‹å9ßM¸LzváÎ0¦ê©uÊ»ª˜Ï%ŒÅŒH…PÆbF†e ãû0,+¹±½ß ß–FXð„±XžaWŸçaÙ ÓB˜p‰°©Y„k„!a¾0¿ï@X@6“¶‰0‡lîï‰ÅºÙÜß¾*\ýMc{yšÅì#LJxE”ïÏìÛ{{˶ø{¾Ø9»aœ¯Ù +óTˆä[#yÐ÷|F˜çó £?¼>Lº…ïù¼#LÉS‘”vä©bÊãk0æО„±XU!”ÑŸuÌWK^æãÞ•q=¹/vŠ4úcTíRfØJõë¡0Â$¶“<#¬a&Â30„= 3 ¦ûý†°?oîxñœa†ùJÂ8_9_P˜a8 ›AXSr„íÝçY[‹ù óÍ„–d<žÙ8ÍS{•q½á|Åäa5ÅŸ× 3La Â8Ÿ†¿ØTF50_)9ÂÈ^¾óá…ßa7}H3,aÂ-ÂhîóHa´×á©›Ùq=‹Æù2>/»G˜Ê} Œ°€‰ùœÉ…㑹џec¼q×6iôË'N>]³Þ0ˆ×õFã „Id³ZÇ|1e„m/Ÿ¨Œöxóºp„ý._Ó§lñêø•6òº_ÝeÇxóÅ”q¼¶?˜ÊæxØŠÏzû矾Æ*Ôw¸ò;MáA¬¶ÇÐÌÆ4f-<ˆÌ”î{’/ŸPAR°qãµ2;×⣨_æâ—ô•ë‰HD¨‘ð ß’ÞØf"™ïÓ2±s|ZLbÃÃÍÜÜxE#="±G>DÍäKótÁÂôˆ:3 7ÆóõLŸ>À·?™ík¶D¢w ‘-‘ÉÇJ¼ëÚ'xg fz§ˆÊè‚8((–é/tPµ]°pc_ùàÆÅ–ïº2#cLÅÉ?´–¤oŒ·¤™øxG&ÈÇPô,ö^ªþ`åÆ\mÿÜÞÍfú¥ûŽ,ª"ò0P40Ó+JfúE˘[éÙ(§üäÄÍôå,5šu;™'ÊÀ›ÚLofƒ©‰¾Ûï[¶“ɇw1F“d†¯ÀLDzÞßf.n|ŸN<ÉuÏ…Í™´á‰þbyx:• ²Cþ’éƒÞ‚e™•ime2í•Å43‘ðÁF˜Y?>“6ýQï`Ã|Î"˜™+¸$ l¦hÜØ†äíïBΣèÞîd¢y¬­}°àÕè`ŠC"’'Pt<Ó§ÕßÿsU,Ìðh%q°^1ÓA xŸ7gúpi¢o¦ŸV³ÅZÜ+7î=“Uvg‹uªúcx–¨ÌDZ†}[£úÉÊd‘Q‘‚â“‚Ü#“§Y˜òIN ›xóœé—r&šdúc–¬Ìô‹æDÁZÁ®‘É< ¾/Íô®žo;Ÿ4 Nø®3ÓL4Éô)†zµP"5ÓwìB½Lº¬Šó`g›Y¢§™h’é]Ð2¬-KDîÒƒ /ö›J¢§Y22Ówݽ'êŠÈYA9®L?5n±Æ­Ø™†â«y3¹ñÞ™ìrwÁJöLOË@‘ÄLïdvÅ—ÌôQ 1Ó_q²¼a¦1³¤EŸ· «¯ý–gXû÷Ì]Þx3“é'ìžXwžYÀ1ý–gO4àDäÁr†™>°aJfaVJï‰Æ‡±/ÞÓ*Ó§:±Lbfü!_Gõ 3|_ôYå‰ÞQl˜)¸"gå("G9+™• .g*úä*¾[,à8™LÅÝ'Öµ=è=^<¦è·K|'üäd ʹeúá>žª.è§äAªP¦_pOÇ®Kô› –)|’aX„0Ó‡%tŸd…݃ AÿC(›éÝ×YX_šˆŸÏ¦’ˆŸ¿}nJ&~>ÊVfúeñ`ÞÀ“ŠÐD,j.¨x#ŒEc|é,ŒE¡Åcú”±è—ù„[l¿¶p‹Ð„Ús{T¼J®¸Õ¸nK¡ÅW7)AÏ#ÍP„âcQsÁ¤á†ïƒŠxÉ…‹þXQNxGh ÍSÆ¢p”ÞS~‡Ö¬%Eó(=¨ü­aEì§ß¡5§h{{eEaœ¯¡Âµé‹Ržæ`¾¢ô¸2?Ç#»FhÌÜÒ¡5ã[Åå·#4«s”±?Yq&¹pÑÿ=±ªôÆþZóïmTDÎÆß3„ž#´Ì*ÅdcÑ3CcžŽIñ×{J#T,*Âd[„ΰ"mr¡Y‘ö»w|žEEѧùy¬Ø’ë™!4CçëËgH#TÉ&…>ÍŠ˜×Ö¥;>ÇKíÕ6*Že#”é"ÙÌPa\ÏFEÅÀ§Y‘ðÚwöÄßk1éüi†Âðx>ÍŠ#ŒñÌððne‹í}Q@v‹íçîì/†aÒzöŽÐß_ÙóУJ[„Öø¤ðì¡8{H3ô‡¥ž~‡üÞlôHTfh &#”C„ª5*6f#´g6TŒz:B‹fGÅÅlôW³cÒx6B挊Uß]"g~«ØôÅh¯¡)ÂãšS¤{|žW”ËF¨Ý\8_²q=aÅŒ§+ëFhNòfq½„WìσŠ&ÙhÏ ¡Y ZïŠ4O#´óF•OiŒ‡t®ÌP„ö$ÏÏPŸT™ÝãóîþFhÑbE¨ìwhO)Ò5Y†IáOGèÜbE—§-öçÄïÍÆõ„¡=¯ÞPž"Ü9^]b­ŒñäòÒÑÙLÇÐaœÏ ¡Â-Y;*r|we³üÆJ™¡Kžþ)P„ ã~n·˜ôùÝ|YsCw|Rq6Úó]T±¤q=Ù^CýÙž˜ôþp‰ñÌF¨†0úƒPä½Ïw†4îwΈI“O3Ôg¼'M>Œöz<}2{EhBe„G„Êì"ñÊñpwe´§ƒÐá¡3÷|MŽPσö$ŒEÓÅÓá” â&²ÇgèŒOªÍöý}g·ù¤Óì^é]¤±(õzI3Ç0éîiã¢KÌ^TÆ¢Î×3m,ºbh‹ð;Ô¥oáû¡B¡*“г-BOüxdcQÔ=ߤk„Šô"E½µcÒìÓ|¡ÃÒHÊ#|Ÿ‘ ×-™]ºDèÈZ•‹v*"ŒP‰[z«K#”¦´×lþ^0”]ø{PÚQß—¡ßÓ ‹âZŤ½l,ŠCi^e,:Gélå†Ï˜Ô˜Íïã¥Õ²w„‚ÌÛß)£?¸«»4…ÞP-ÍP”ƒïû4'Uz:A‘ÆùØ}†š2õwŸ¥Œó¥¯ØþéŽÐŸ[Íнû๘ôŠP„r¤±( ³ä”KxáÏçá(£½˜—æV~‡2Œ"E·˜M—mñ÷&åf/¡Ê¡v¤q>ÛÓÓ|ÆôÈìÆövÓ)·4ÿÞÆ¤ã§+C="E‘(ô”Íû%†6(ûxÜßöVé¡Uç+B’7+Æú{yíö=XÒ¡>×#{óz„yQʽ^šR¹DèÀT‹.ÇÁ÷ÉFhÌôÒ¾ÊØ_˜¹“ÍI} …~”ÑŸMç¦<½HcÑéôÒŠÙïЀéÇ[x†6Χé³Ë•:‚9¤Ù 4¾o¶OÇòRZGÚ'¾ù å)]±(Ý_üeóu¤ÏÊÞÒèÏ1¯CçËë~g/é‹Öûn±¨½Þ £2õÞÿ¥I£¿A‘šì÷¢yOGRÆþÂL eœÛçµf3´¹áÄUÆõæxè‚2öÇiñyß¼¸¨‚‹®•1^?þ>G™‹šçX*#´ëìø¼ïŽñÆñPe,:z ¬îù(ÜcQô½ £"qñÇÙ‹ëïÀøHcQbñÙýÙœÖpUWåÁE-÷Æ©J£¢ó½Ÿ˜Ò5E¿¦ÙQq´øLBeT”ÅêÆì¨úºŸ¸ý0Tï싪׮ñyþ>N‹º«ß˜*sQóÆñxº°bmõ¢;ÊcÑgIcQRóùÉóà}ooh‹NýÆPÙïG¯ïþNf(BGÑ åŠE³þ¾7;*>cÑj6Cuú}qfÒ8_ðâNÇérÙœ”Ù‘n§Œã‰¢ÊÃè{½Æùd>°Ê¶X4é?e„*˜ßØgwžoX4©ŒE}x•Ý¢"¼¿xTÆ¢µÁ¿÷tÅ|V|Î.xmeâ÷Òßy}VÐõÇwÿúÓo~ûQ>þî/ùÂ!{¯ñz²’Gqs è>è“ +*2£øî}“!8>>×xeú D)fFiÞGáÞ 7¾o¸2;7f¥Þ'¹šªÅöuW&7žKÒ¿†¯RD![®ÄJô“Þ|r€¢ÿ!óe­Or@ÈUM‚\™ä4Ñ\·”‰`,Ü›éc½E&J/#ÊQ°‚ý(v®.]r€gHúGa­F&èŽâÊúÐ" ÷&ú…e`U„àøø,Í›ÉõCÖ2û»¬/Ë'?èÊ'ȵGg)úÍØð²Þ‚~«e}3 8º¢㸺(kq|¥âüø,Ü›YÀÓcýP«Š>ø%±rã1 7æn²€\±ú ƒ¶»Â¡IrU“¯£MÄú!O×Uô?„HËLuƺ¥L¿”âN]q|¼W& °oEïoŸ%Ù?Þå„3;W&•*9@3Éúñ^·”Ù¸2‰¥µ¬\ÕTM²‚¬Ãý V5\p3¹1÷7ò÷,¸à&b] Öfb%VÃ…/³~¼W5eúéüâ0IÿˆŠÍôîzbe¡`çTô“!‚4Sô‡‡\M•‰•I#Š£?ÙÁiŠ•ßY‚¾1OºDïÌ'bgŸd*,Ê úí"#'3×—ÕT‚þów”N²€~Â&b¢&}‚d3½£`L£ ¯Š™L}³ØÌ2}E”ÅzôW5ùÀ&±rU‘?X¸2©eN>T庥ÌÍ5OÛ$ý£X8eª?+¸?ÈUMűcY08Ó>Þë¥2ý¢¥¦3;¸LÑoÁ׆&úØu¢üƒ~û¸Î»Tüƒ¾6ÅŸí*úª—ú^Žÿ>èÝXi'ØA”oÜ\ÖPØ6sKÒ×uìœD_Nq{Iß[&Öia}˜ WyyìDü(žÙÁi’\VŠ¢7†=±Û3¹qÛŠ+Ϙ‘ð`åG­¢X¸N‹9ßÈÚ\=&ÈÕcwÐ+è_rGÁƒ(RÎ'чÄ\&È5^s)â€bÍ¢ ÿ|¬ñÊô“«¸2«¸N—ôõC5Bô@§bu~bã/k’ü¨µ±N‹']&ׇªˆU\þøDÑÿÖ|<ÉW>ǰþ>kÚ°Î2ÓÊ‹Ç$}ç Ð7spmY_’Üx5EGHPÈ䯧+ú+æ(ažià’ô‹æA„§ ÿ¡…Ëb¢wæg!Ç"“¯©X¸ñY’÷„=<%dn(×Òennì—ÔD2_ç)è¤}¹JÏËÌ'ޝ á”+mMú]í}šv,„»£Lá Ñê‘fµøŽ‘ˆ0ʱš{6&â^q„c{î­ï6VÓ-ñ#Éø¾þÆS¿‹3…YÍi©ÊÞC É [höŒ…zÛ´c¡]mÒïjð[) +c{,VÆB¸‚ÑD6&ãE4¶¯èͲY-i¬Ê±ý4i´·Š… Ê>GýYm^¹ÓÝ„c"pmôÉÆB…Š…nÊX¨h ·ð‘±Ѧ4&Âr¡¢2>oDÏÓ¨FZ±´YÙÂ[»ÐǤ1›iÄʱrš4—¿èÌn<ß™V¬\iï/³¹P’ç›pioT›W6Úǘ…¶%ÍãÍó]¸Òw¢§0ÎwĶKWÚǓ٠y¾gc!KCÚ¼r¥W—ÆùÜØ_wºiLdFYìÊ…H(K ÛßçhÂ\Úp½ÈÆB¯ÖП ûñj¼^f£¿oX(.KÚb!íÒXhß®ƒîUºÅöãh½‹6¾/Ò¨…¹Ð 9…±Ðä.Ä­Úø¾÷OŽ7*ãû<ÉÞ…ö;ýlŒPVIúëB_aœÏiéÂï1mZç[çõJ¸…Wöˆ…ª\h,ŒßÛ{Äm=½b¡ñîÚø{>ðRFÛ}E¥4ö £šwŸˆmÊÆxª{¬³t|ž™4’õ‰Ð(áXˆÜ¶4úS.TVÆþØvöt=ôÚƒžM»†4rãz£ŒíOd¤}7'r]MIûb‘˜?kUÆõšiáʱýí„glß+(í‘QÂXˆì«òøºP[ûÃb–ž^\È{g4m£ïxX¹Ð x{çÛ#Ð¥q=bš¹0®GæOéø¼1´ýx3í\˜ é±PY¹Ó÷m¤0™¡= ÇBõZ¤[,tïUÛÂ[ç  #(ÑFÙ¬Öríí)A<×[ÛèûÄHý¥ùŠ1éA·#jç¶'Üéµ¥-ê·¢]h«Ò=òÏ#çv"Jðiô‡æe™•y<ŽÏûIÔpp=ÊæBüâ÷kÂ8ßû áFÏ!ãÁ åJï)=Â÷~Mã=¦á+ºiìïÑЄcûU¥q¿88žn´÷WÙ%‚NæT¥Š‰vʸÞà‹+óx ¿_ÆýþðÄéØ~WiœOcD`e2¾ê+WÚÇ+ÙÛûñÎæñö@8éN¯-Ý}Ž6ö‚”íAƒÙ5‚f×ôZÒ¸<ß…;=L8‚„ÆÁxG8¶?KAUw`~¤W˜žO£¿ž¨f¡Üh4ÍÆxwV\o„;=Ž4ž7MToPÆö F{š^I:‚0îó"áÛOÓŽ Œµ¥Ñ^¥Av´$a¿úÝ\Hâ3އ6>oø[ aÒzÀÂRéNï-md5š6Ž×F–Í ­ãã}áö yáAûù(\i§y¼y> }¶4ƒª ‚æŸæ|6–5–Ž /xAHìSÆñ®<„ý÷ÞÄÁ.Í ®†"Â^]Ú"(Œ…žî‹[Ûû# 7ü^OžÉÞ4æû[¸Ò»H¯^ÒûcûóbåAûxF¸ÒsH£?¯<_³tò²5iïƒñJ6ÎÇŠUêÊÞÊ‹ #Ȩñ|nôšÂœovííE8‚ܼ?ÎFPƒá„q>·ŠBÙ8Ÿ§ŒþøV«Òbj e`„í×Ë쎿‡èÊ6“FÝx¾ z5iµŽb3ÂFûõ:çcCÑäÊ ¢fï  ÚSfÞˆB"O#¬!ˆPçc¸žcc½¼0Æ[mb<.ŒýÉþF¸Ð^d&AhúSÆõúå½µñ÷v>I®´×³$Ô6Jówúti¶‡ãÏ»•#ˆpUiô÷ýÍӅ㽆œ e²?n´—ÊÞ±½G²1žï(**Œñ^çõ>AbÙÂè¯;Öó c¼ÞÙûþ¾Aºv¡ý|ÌÆx¾ŽG6®÷ç£0þ‚B…„Õ±B_ù$Y²€rí¤}ûseü^¬ÅÞ“îMAy©Êñy³Hã~îVÌ1éŸç…0²q=é(>ªÜéÕ¥q=è…v²Ñßw0(ÝéiÒ¸ÞŠý*úÞÏ%oCB‡0îç˜f£ÜÃGzÅöÞž„±½— TÆõÂ0žP6úŽg” }Š4ît*]麥q?Àôa}"DAÆÌÇQŽ¿·LxÅïõ—2®‡,_.Œþμè»2ÆCø+£=³ˆ¹ò §iWÚ Ëd£?c±ra\¯pcš=y?Šwi|r „dÍòàʃöö–ÍöÎþ*{nÚ¦4ú#`r¼#ŒýÉþ$ÛÞ^Ú…ÚèY¢^×;e+ãy< Ñ+ãû í,¹ÆñB¹ya\Ϙ˜¤Üi¿ßÊF9‘¨ì¿‰µÊ¸_[lÏÙ¯/¶×lœoLmRöó‰¾Ê¸ÞÝÔ¸*ñ:òù¤öñ¾°·‡eèO²q={yáx³PøYÙh¿Ÿ.´÷'Ù8ž‹í1Çs!@¹ÓÖ¤y¼Pˆ]…NŠ­ £?\ÈÀQžô¯ìý ó¤„q>.¯ˆ$]èÑ¥K|Þ®ÙíÄþ8(T—ëÕϵµ}{¦G cnô÷Êþ}7ò8„ñ<” SÊØÉfÂxºq>)Wzjc<²=iRã&Q)ãûxðBöæõA÷ÒöBÙ¾^íúÎWÆñÜH­Æõj#eMÙèµ´ñ{PH[˜ÇijÊ…®UÚbûÛÞ•=ºtíg‘FÉ´(eü^ä}(ºi¼½^Ú±= Á>]bûÕµ}ûƒBàÉ+‚þ fã}s©„Wøö·ÂxŸy fãyÁñBʸŸ8- Í>óñtJÍÆùÈ*aŒWòr”ñy…T³ñ¼’™PÉ“×;æ> £ÐÒA6›0ÞWžoÙ82PÆýÀY(Dœù"ÌxRÆöÈÛP.ô½ž ÷ØþLiœO…î•;=Ž4‚ð ò0”;íí=»ÄöckzOa>a¡ éJûñÌf! äQ ûýzGÅ5e¿_¿…4ü|.4 ?=bûS´ñ}=8FÙÇ«×^È7»c çMyУK·E{›]ãóöÖŽÏ;C¿‡çs6Bú‹çÄg3¯êÚÛ[¶÷‡·Ðˆ·áBm¿žvæa £ÐB™èo…±ýBšm“öëqvß´ò.á-Ýðyì/„}´+Ž÷ñçÑÊ•¶)]bûѵ;½«0 m»·¶ß›çu¤ýzyí…ʳý~„…^¤ýxÕê÷KÊ…f¡î§ÑTäù(7zé±ÃK;¾ÏÚÒvès´ýxU_¦ŒB(Ì öëÁµ5iÂi¸^ c{äý£PLõÙQhäâ4íF÷#½Ã>ÞÌ^ñyÞ_ c{öWÙ3¼‹4úûŠ<a_x y{ÊF^Ù_ c{ä­·CÛÑ6Ú gר~›4Îgæ‰%s=agž˜ðûõLŸÇþ(›ç+û›ìyhïO„}ûÆþD¸ÐuJ£PVC‹r¥Iãxßùt[»Ò~¼³Q(ŠyaÊ%¼¥Yh©a<”]Â~¾?…¨ÏwáØÞŸdïJûõ,çcëxž Üèڤ߅¬lKãzÞ碌ö€¼Ba/ä­÷ðhÚø½<ß³q>6ä *ziô· yMÂo5ä¥<]}·²Ñ÷}—0 µ5ô¸3ÏLŸ‡¼aœÏ yPÊ…¾ãmáÑÃK×Óû`ªI÷>ÒmÓsiÇöûHãxuäE £óÖ”cûq„ó~}ß÷*ú>ïÞá{=ÆñèO)ûñÄ‹ eœýòíB)m±ýZÒèO»,¤køHã|ìÈsÆýÒ}ñ¶´c{«Ò<žÈ›K^<;ò¤”ñy评q¿ÜýÆ]ºÐ½H¿ áÝ÷ Â/÷…ý™Íã<e£[‘öç‘Ó”q¿Õ½b¡t§{—n±ý>Ú¾ýXÛ¤}>ÃõéÚØykÂ诙·–Œ0ÆÌ{ö÷‘,¤›Í¼çkïO³ñ¼|!oB…šòP„Gø˜4®·Ë–2ÆK¨©\ÃçHãx2ïèérø<Ç•7ìneïå•k”¹?|b¶2 QßZgSÚ°=ò‚„Ѭ¿ç»7æKö<aô‡Û'þ)ã~xûƒue\/¶W¬VÆù¸¹¿žŽûíÁ©Ê>¢3Ï$y²Ø.óÂ?ÿTeaè_Z ;é]ûùAï +Ø2Qß÷]ûùA¿¼|Ö`~ÐÈûì;³“kf+d³fm¢?ÖæªL”sMH¦wI\ñ‘é ÒX·2ÑßѼ 'ór-H¦G=˜á $ú‰b#ªÿ>è1í]wöA_$5˜ Ù{&_˜ß( z÷ÎÕ™ØuXšé·&æh½c³ƒ“è¯Å «ÆŸ¬¬WÍ̉¨gÌ ¯‰¨†[¢>胨wËæˆšµ¬Vžè]ùðKŽ *{ÁƒD.*ˉþð†=E3zT}5z;ªð&¢R²£¹š¢w2\9ð`Ô˼ãzSôÆÏYý™>èXS”‰ÊÐh¢™> å|ýL?aÇ»ç72 l`.p&fÚgÎàQôiÖ¬vœéãÀ¢L…5+* %V²›b!ïÓŸ'YOq6üÀDõ„B‚¨+Œu7™¨ !ÝÃmJ£¿ádaô7 /Û’;ÃL{A˜|6Âj9ÙA˜ÅÞj‹xšÅÜj‹x¿÷®zÜÒø½Å…ñ{»!Œ?¿7ŠÕ=Å:&ûó÷.\o²QŒÁJƒxšÅÆXÌ,Å` Å…q=ŒbgÙ,†äU³£Ø¦áe©0‹•±XQ6ú¤*³Ø‹qe[Ãòó-Ç+Šae³X^6'—G1§lô7£ ØO6‹±K¶?\¹Å˜|f³‹Ee#ì>Š%=lœŒVù²WØ_ÖVTõQF1½Ûq,i32„]f£˜"_æ #ì›/s…ö?f„Å÷ŽßƒbhÂc¾ªJÅÒø²WÅðÆBØX6ÂÒ‡¯ÊSf±,´'áfúÓÏÌaæßã¾Ìf1™áÔOã÷L¶‡lûÁËHa¯Éã™0n«f±¼ìIŽbx,ö"Ìb7û†ý0‹Ù „0‹•„if£˜_ö ³X‹)d#ì/“¹«.¼ìÆïY a‡Ùø=«#L5ýÉêvü4‹±°@6‹yLüžl\ÖD˜èÓÆëÁZÞŸ ãz·ÖžñÎÂË^a;ÁË@aŒ·×‰0à§q¿Å4²Yl¢ Œ5›Å"x<Ÿf˜ï-Váa¨Ù¸ŸØ ¿'׳ÍbÙ¯l†ùgóûzê‹2Ž×ž~=Jæb’ºW„µ>û¥½Ñfãz¾vŸãu<µQãµ(‘ÝÃkH£á©³Ìf˜?ò³ñ<àø[Ÿl¾Hº>CšÅXì${…vú4Ƴ‡í!›Åz„“>þêúÃlË:xÙ/Œö|f„½~wÃ><Ÿ³wxui´÷³# ôi¶‡aŸO#L0Šd#¬"Š d#L-Š dc1{áñ~¸³v+ a/ÙX }Ÿi„ œÏÂÅ fù4Ãð†,ŒÅäÅÖ•°·—=läi¾ÐŠbÂëk1a#ðªÊü½3Ÿf±€‰pl,¶/ aYÙ óG˜±0'ÂÖ“#Œ  l]a˜µ¡Œ°²æ“Íb a‹Ù«Â¬e†Ý{*­2Ãêvž<ÖWK„1>0©ŠbÂ(Fq‹ li„mT„• ã÷²˜€0ë׆ý™0Ì;«pI3Ì¿¿Ã¿™¯°î¬L†ã=ãùr5i†åÂò²vV-Ÿîñy«H# ²"¬X¸6Úª°1ì³Æ•¦QýÆDaSaÛÂ#¶?]Ú"¬¿š4ÃÀ ¦²f°|a†ýW„Me—£çïùîþæÎ°¹§ÖÔ,Âæžf;¬…¶nëÊFÚM-ØÒÃnýe6ús¤NdóyS„½ £˜Dc˜s6Ãè=FÇ›aìÂè¯ܳö]",ïiô×½DXÛÓkëýÝÓy*Y—ÆùŒÔ3e†¥£ƒðˆ0ô6¤Qì¤w„uf£=wÌ®FÕQŒ ù?v—óµcf—0Ã0³&†Tbeþž…ëu6ÏŠp¯§Ñ_uÌ{º†‘vf:æÔ#Lг…v°|aŒ íM˜a¼è„¶k“MŽ0­;—ðH£˜‹»f3̳þ„џ؈0¬§q¾ú a„É¢ª–2¼86„Ù #lUë”ÑŸfE 3Œý‰0ŠE ª¤2®a£Â(2Öž<±Øú>8.Sad÷Ár—vìû+a}s\…[|ÞîÒ‹ºö«ð;ÌÅ„ö„ªÝÊ#ÂŒ{‘FÀù ÓE˜q2߇ÝkÞž³ÑžÂÊ…æ4&ŽG6Æ7Lx ¿Ãv1P˜ûÅ „6™yÂü¾˜‰*̰S„Ù #ìa±Â[›†°Ä§£Ñ´ëxíyŽÓxša±ÃÆxn„eãxcΞ2úˉ0ÊäÊþ„a¬Â ÃÝ³Ëæ÷Ýï0Œ‡öÇÂ=¶?C׃yÖ”Í0Zï§öàse«ô;ì•aOã~j!ŒX¸GØk7éa¯F“ͰNÌ,~ºž«Å#aþ=ÌF{ÀŒüìýæê‹Å²6É0Ta\Ov*Œóu,fËfXî‰ÅþO3̳Ä÷yaîç³0Ãk,öÿîaœ‹™²¶Ùb1öÓ ÓÄÌCa‹0ÏfÒ¸žrŽ 0ÃFw|ÞwÇó†U 3¬•û;×›}°Ø*×Ë—w‘fØ"Âú„Ë×0Êä(fp0HãM†Q 3 a’Âxpóe3 ²a1ßÓQlò Vaü‡Ç;÷«ÇSå•¶ÈöçóO…1Þ?˜O”Ü#ìalÂèOÂJ…6ˆù€ÂXLV$Ü#,ñi,¶cX 0Ã܆ß/'s2ëõÖ^ø{Û¶³‹Å¤3 o£?Ë®ø{ [yšÅðnX™‡Ed3l ó…&T1_F‹±«ÅbÒ§{l?º4ø0_*9sâÄVF˜REX°?Ï‹01a~_ÌwîFs±éÓ-¶_%û.Æ|{I¯ËÚEšaLËÆþdX–p‹0ª¶¥Ñž†•‹-v%Ì0*_u­ü“*&Íï‹°:a†a¾0ÂT›wÌÙ‹ûû¦®Vi,Ælž ¢Œ0‚æ+G”F†ö"Ü"lêÞïóû"¬.™ó9#lJxF˜Ô}(Ìý}ü|æþöªlÊl/›K\LÌù|Âa1Þ1)[„ùùͰ æýe2ç#ܰ_¬›ãÉ0 aÏÎö”Í0ÅÎÙ“혯&ŒãÙ}bOv°š‚ÅÄÙCUFe„=ÜÏMí‘a;¸^0,'™Ï›¯ýûfãz|'.˜t‹0ï²æƒùJÉq=â|Aáöþ"a†°&áwŽ÷§Ù8^†ùfÉ…a›Æã™Í° ŸØ¦Œë=ç+ ×»i#¹ž›þb\íaø…QaÃç+)×p;¦†‰·Ê8߆‡q(#,`xª¯2®·ÃS»•…… Ù¼êwá…6ûOŸ/«Œö:}>göŒ°Œ}OeÏé+•±?ïšå.Ìùò7 âu½QFØÂ$²9_½c¾˜2Âx–ÏTnQºp„ß0‰%öv‹¡ i„żî×GްZÌSFXèö¿Ù 3±âkÞþù§¯± õß®€AÚgšÂwv¦)”©è0?³,ÌZ3ÓâÇ  ±ñ­•ÉTiLàôa3ߊdz§n5âô&mxc›é7¼˜¸%è·“¨9žÈÉh,4šéCóKŽ w‘‘X¹ñ}Jõ$+4¡R|"o1îT¼©ˆ´ <_ÏD"Þþ<8™,Êl‰LtÎl‰ÌÉK±sã;k0³‘¾¦.I (ö$œ½XMч+Æ„€D¿Xó]W&Êfú<߃eúmó02qZáÙ“|gL&HDVÊ{múƒ~{ù"WÛ?ˆ‹w³™þhÍÖ¶>È¿íXöý º” ÌôE}ÆÜ‚DcâÅ}"™é'‹>ɬ=–Í4&mxDbaJÇ©™œÈ2˜jèÈQ±ÛYžñè]Ð`ªA"/ðþ6³2iã>x’Ÿ ›eúЛe=3ýÅÎ]ÀV3n Xr3¹(X–‰ä ¼Ë,LÚ8‚]Ð`#LÜ_“6'³4fUÄÎÙ±¢úA 7ð–G°‚Í2Y¿å33ýÅð*Z‚Ø9x»“‰Üï>±ë˜âðƒK'ŠŽg"[¢øSËL¿,² 哜rÀtÌÍïóæLcþG1ÅNrmîƒ~«4ñ&"Ó;Õ»Úÿd.\žX¢2ix‡œé]ß`f"Å'3+¹G&ް0e&’TðæùI¾8`¢I&~JVf3K¬("ýo´2ý†ïKŸ4,gáÛÎ'ù°ï:3}¨ÍD“Lï̲42½©,”HÍl¤¯á}°áÔX;'ÑÛ‹Í,Ñ_¯3Ñ$Ó;ŠeX;šèSCÒƒ¶›J¢w#,™é]ßšXžˆ]‡r\™>aiùÄÁÊ4”Þý¤[<Ü vÁJöD$‹ Hb¦ß¦_E¿ cÄÌHCñN&ÑO«m±~ò+cù㋾:'ÑOº“=ÓÈ=}ÇîéÞÌFŽ©X™†âkbä+m–3ÌôêLCÉôÇ‘³)29+V ?j¶L>ß ë¨‘³ü2‘iäîŠ8 +týö9+O²6ÝÆÛÁÌÉ–ûn9Ó‡µeÝ2½ zqNE¿AØ~Ë%è”dÆáA9·'ùÂöT4•Dœà±„ w#§c×%æ»øJŸYe‰E3ù.÷-H¦7ïƒ Lü|”¿}’Õ±ÏÂúÒD¿Y>l*‰ÞÉœµÉ‰ÞÊVfúEó`Þ@fãÆÇk„Îø¢®l†®à¥sr,ú*Õïz…§Ñ}H[„Ö¬-\ ÀPa,Ú-¨x%ŒEå¡#O÷ƒKûõÒÝ(˜Ô ÌïƒPa‹ÐœÝ¤±¨ñ„±è–å’ÙiV¼ýSæïÝXTðôŠP„–cQaAEIa¶TŒK~‡® b¤0¶¯¨ø“ÌE-×ä±èýÎŽ¨Ò-ª¨X*Ü#4fWi„.ð¥{²qÑ9_ª cÑ9J;(÷Nû¢µ§;†€*#ŒPƒŠ—–ÂXÔÏŠYÂ8ß*&M—ØÞU<Ý>CmV‘áñÊf( *º #ˆ/Å…ZƒP›äXÔÚ&µdã|o˜´ 1)9¡}EÅí§±è¸oTDÎFh•!ôD¡÷¦ÌDìÆ¢`«¨ˆ›ë+Â×±*ÌŠÉ׬Xü4ú;³¨Hüt‰Ð?^Oø=ÍÆùtÓ!†4BWîhxI#tĨŸlñ{x¼²Ú³£âêÓ3Bqü÷d„ª ÔF¸OºNáÎþ|ðxgãxöžÿÍ-B{x<³ík(Žp=´ébô0áúš3·4ú»a1iýi†ÆÌ¨¸ø4힯O—ÏÐN úz³‡4®7¥„Úsð{³ýà†ÖÔš]C¢"I6ÚST ÉÆxñ>¹ÖÆù>*Fe·ÍYExózp½¤ªƒI; µ™˜ûôbhÑDhŠð~‡æiî„êc¼‚·MÊ=BsÆ~„æ$sÒ€{I£¿¿ÑÇ[¡i ¡YÉ¡8¬H“ß» &g£¿¹ê]¡! íIî¼^¿ìû;¡˜×KzF¨O©ÒŤ e†ò&…?ÝbûwE—§q½Á¤‹ìMch0úÛÊS¤*óò½If(ÇõéÒèÏÚót‰óu!ôC÷{/{EŽì¡?w<—¼£½L:ËFè C„1ÞߨØ"ÌP g—ÍŠ‚uûÛBeôÇÛ߀*ãxlŸ4“ÍE×S{„wŽûÅ30©.{-z˜4Æ#˜]¤<zãOÕ”q?ÍP™dV»öIÏÙ¥=Þq*[„Ò“n:Sª4B¿ÛS6CU¼´ov,*,~㡌E©¥bRíÓ̼¾ÿ‹O:Í‹öý‘E×KºEhŒi~„Z$Ç¢¼2Q±&{à÷"´E˜¡/ “V³±( ³¯³9ÞñÙßMz"”‚Ç#{D¨·š¡$•Æ¢®Ú1iö»o39 R–ô ·*EÛÕ0)/›Ïg|)3ÆCE²#´ç®nêÒXyK3.i†Ò,LÚ{šOYšY‹îú‘=¹¨©ULÚËÞúQ§4C+üÆGÙê10©1ûÊaKš¡$^ÚS¹DèÆèÂ#¾¯¿¿QfèÈÁ÷ÍÆ¢ßîsÛ²-B#|†š2CÚKöˆÐnÿ4B¹îƒá-í×C0n±èë¦Óiœ˜%§ÌPŸ…“Í— %›”ÑÜIå[‹²1—N‹1›.;úC̉Sf(…ÏlSÆ¢S˜Ý#TÁÛÓÓ׫ûŽÏ'Ý>¼Y‘ÍÓK·4µšÏÎÎf¨©¿,Ò%BîþL^já—2EŸ ŒP¼±Ïž¼ ô@yÅö[×sÌÔÊŽE¡Pæ¢ÿ…ï“ö4¼ôq¶1´hxike„Rƒï“ÍE÷ž>¬Œë)fîd3Ô²¡Ð2Æ#Ó?XýéôÐå¡÷z˜Ì k>÷°J£?DÉå¡¶¤q>LËœ]š†y±Ê¸žÜûÑ#ñÖòtMe„®`nt6ŸWû¼ë-ÍE÷>¯CyÄ¢õ½¤¹hÛK%/>Ͼ‹ÎïxDã¹»ã›4ÚŠÔdïøï»çç¢ê¢=ÂCýõ8hã|<;>ï»ûƒã¡þÊX´z¦]¥ïê¥"íƒzñEÁÙœ–Ћ_x•Gøžoɱ¨ìÞX)GÅÝâ/F•±(¨xú¡2eûyvTt½«K»4•oøÉÈòx6”Q1ïc•ýzرº=›Óâï¢äWÃWÆþ¼¥í›tÅ¢b˜gÇ¢§êEw”Ñð`4{²bróùÊ\ôŠö‹–ð>S™‹ž û;›Ÿ‡ã™l\tŠE«Ù±¨‹V•Qñº{ú­2%u/J’ݸ(½ûó~eO¼XW6lïÏë•ѱh4;õ›?øWFEeó÷)Ù…‹öÌг¹¨t¢½g£¢1ÞG%ÇtæXt(l±èïþ½äX4ˆ"Ù‹UW‹Ï{ûû"¯Ï º~ùüן~óáÏAî9øÛòñw¿xñî{µ×“FEX}–Ò}’«®¬)²L°¿ÓÈœ_W{eúB¨¢`YÂ÷ALa;x–¢q]U?’\“eS²€s)zpï ›dûø\“•èÃóZ0Š\ýÔŽ¢ ¸¾)³põè7òµ,W0 r­P’~P°ò"ÓoH ¡Ž‚ôÝž¸¸ÎèÿUv÷:Ö+Ûu†ó}Ú¤ú%‹€¡@†äĉ­ ŽŽ À”œÀ·ï¯æ;jínÎq)1ô<ß^MÉ"YsÌÙ-'|¦eü ª62/U!©Íò›±ëÔÂ7‘°ÔGªÎH;öÍòõ©QÊŒ‹ËüTM¾ØUg¤–Ôo¨–ÔoÆ(zð6U!˲Bíö«ªNKêUg´¿ÖdRW5ÎŽ}s§[ÆÏ  éÍÎm†ž[†KUH1¼3Ç×§‚ÉPõMª]}1–Oìú¦f”Õõ™—ê›Æ²T}S–åëSÁ”“RžÙ ©}£FÉP5J}9Ærœ]»s[ίOcaÃ8dO|éÈlªoò}*˜2i;¬&Û™ª`ªÃ²@ ¤ãõ5-*ÔáþAåS\…nfü‡¨"ÌŒiãÕ¸ñe¶¯O}Sfä_ÑÝ1êŒÍ<ÕOó±œðº,UUšcܬ/] 2û×§®*“Z0ì™^òÀ½~#3nÖ—NºÄ˜<^иº¥*§º%ÕOÔUÆ¥Æ03nå—NØÌgϬjÂL¤¬á„13UßT¦cÜèo ³Ì¨œL{ãV~O&6™îÚ™ÌxwSCjؾ>L™q³þÅ5,ã?DýKf¼X¼iÞn;‡ÖÁ™ñÐs\˜I •S†ú§"[ ±ª®ê–ñ¢J4“vÖÔ)ø ˨3Š·¼‰…:ÐE ‚¡*¶îâè«rP2£î¦qâdñvŒa¶:;'³B%-¼È¥‹[fÿúTŠeNÕ‘)‡áÍø1aw¬ðŽ1Í U)¦ÝþbL/±-Ëñõi²mø­Ê,3^ᨷaûúÔ‘eVÕ‘Ý—eìX8‰T™éÄÉTµW,+¼îÄKÉñ‹êEÃøT{e2€9q ÷Ž}ÊÉ»x3*‰lÍŒÉçS©ÓÏlp4ÇKµ`÷m9àŽ1€ŸJZBfÔUÅûyØ|>TvX‹cL>*.3c˜íÐÃRud»j$3>Š<4¦6_4$Ï,âãÈg~òL^Œs}âM¾SKEN+Ã×t\ª}k·å„}Xª2î¾cÖ^ßt[ÆŽå”Ìd QUgXàjŽSyëqŒ×–Ѿ†FÃùDJTç|JÞ“|Jî®aÝNIÜže:SRÖO&ÌÛ,/ànMI¹¹ÎEîËš%eœ½•ÜäùŸ%Ç{íÝòæ÷P¦i¬’9rS§¼¦5K‹F¨ñ)Ù‹ëWöuJövîs‘—·úÖS*ìÌþ¢XØX%i€n=äû0¦¤®r5Ëf¼VrY«| kÆk¥$Ô¹ËÏcÍxUßyç)÷á]åyY3+%oÉZÒµK {óÇ÷)YÜïËŒY²¯’Eg~¯î'ÙŒ×J‘³s?^ÖŒ×:¹×³?u>WùÖ,Y®ñÉÓùSRÙ§÷•ãô6%;Uç›q;~¬5é;ï|J6÷ÓXãQ÷!ã)¯æ]å§X~O¬-·>%Ÿkë3Om:ßÇñí]ä¸ße3^›®ÆMÞÏçÆônäž;wyMkÆënÀѼ))mÜ/Œ«¬p¯·¯M÷ËlÆ+µaÖC·w‘ç°f<·ÎÉø”Ô>Óš%ãmr?0îr¯ÖŒçFn°ó”Wñ>%»÷cÍxVn´s—OÚÚ3žÏÆãøñfG.gç'žôc÷BÄWöý½äטñÚÉMw¦$–Üiç*ßÅšñª\jãqJ†Ëå=äû/¸È˜Íxíãb÷6û+Ö&:3^URìÜ—5ãµ_\ÏŒOÉñÖŒÇ7>kþ{w¼ûHî*Yvf.æ3ÆEŽÇl•„k¼·ãÇZãQ÷cþž‡ëY¶Æãó”1%Í…ùj6ãQ¹áÎ]îÕšñ:Ê'ÊðíSò½–7%É•ùR6ãU%ÛÎíø¶f<îÕÍšñ8(užòšÞ§d¼vkÆ«rÍ›<–±–mÇx3fR²ì\äý–Õ˜ñ:ȵwnòz¬™ïŽÉx1îr¯Ö×)‰ïË{Êq½7f_D-fó¼6(‰6Öx½˜9ÞŸ’ú:¬5^oæ[ÆUžÅšùð¸ÆìÏX[ï\OIÿ¨ÞS¾o~ÏsÆËÛ¼o -»žño)­‡÷#cJÞ ÏkÙ¼O˜º^W9îÙ÷÷HcE)á\äx^Ëæ}¹øÖMnµ" ãÁxÈwµî'ò \ÞSŽë•q•ŸjÍxdÉ5ûsò¼”­H…Éóºñg÷.òªÖŒÇ9¹ß³¿)ÙOVߠ혯w9Žw¶"'nžÇ§ç³q•ŸÇšë­"œ»‘Ï´¦äm=äùxùî֌׋Èç*·–=ŸY7ë.ïñh¼N䯽¼‰Üˆ‰u‘#â3›H›kÄjÜä=ß1f¼ò`k}"7Æ´žß#3œù{èëâ\åöX3^¯ûìï·5^‰äqò,Ö¯D*8³¿âÂs=‘}zwy^ÖDì¯EÅûD‚”iü‰Ty>)Ôo7yÏGŒ‰LÛ_}‡÷c<ŸÈ‘þX39âL¤G4r¾nyÏg»<kÆ#ke¬¿G’8ó÷)bLäÐ^¥·¼›|ßÖŸH’Ú½‡¬Ðñ·Û‰,Ë›¿g\l\eå…¿]où¹¼Ùßܯˉ4iÝ{ÈcxŸÈ’ê¬.Q›V½›ÊÙD)2ŹËq½Î&2ã^'Ù<™ýMäs•û°¾NäÊs{9¢ó5›H·¥óÕxÈû}°1‘8»d±{Wyë~"_®êûsé|Îæ}Ù®íœÞEžÍšÈÆÅ|À˜÷]Ô¦fŸÈº]Ûº¼§|Moö‡îçÙ¼/XñlÜäy[ó¾ké~n<ägYóë©q“{1>ßÓ"YŒYÔ™OOykEi¼79®·ÙÌ7Ÿ¨¼qæ~ýh¼WYMÞf>ù\\ûñ²æ~ý\ã‰Ôoï"aÍ|ô!’Ë™¿ça>™Íýú—¯îÝå»Xs¿~t>½­¾ŸÑqxG¤HÑý"›H‡¢óѸßÖq½‹l¡áMäQå~h\ähÝ}UyvkEBé|5îò~6üÞÏ£ÎC~º7¿w0ßÉî'rêêÖô/ãÏ·‰L(“çc~ÏuöçÛD”ëìÏäyüxWùÆñ¦½ÖDzÝ̲‰Ä)‹ë™1Çëa¾ŸM¤AyxÞ3.rœÙDjÆM^ÃZãíá~n¿·ž³Oµ0_0.òêÖH-"ÕŒOUçƒ1‘ZQ»áÌxÛëï!ßÝšñVu>Oy.ï¦ãñÓMã­Nž×Œ›çK6ã­ÆÂkŽ×Åõ0›È–z1Ÿ6žÇËZ‘d7óã*ÏÇšñV‰ 4æzFöª5Ç+VZ¹uk®gu1ß6æ÷ê|ÍÖx{xÌ&rlgãÞÖOT©;Çïi匇ä"·Û¸j<íõJ—w•c¼ds?mºßwyzs?mÑdÎz?Öñ<ÙÐÓ{Ê×å]å¸_fOöèÎ'²n kÆcÓùnÜå»Y÷G^—÷ã~M$U§ÝUr•Ÿf]ïãÇ›¿wr=Ï&ªM®çoEºµÉýĘãu&SÉUŽùx6‘aM×ã&ÇûÀìû–g÷îòi¾õ²Æ£®'ÆCŽùºq‘÷÷Rã9å§[9×Þw7ù®ÖD¶5]oŒ‡¬öboÇûÂÝ{¡vï)å]ä5¬‰üê´5ŽõÑ;£{w¹?ÙãQ$S§žß˜HµÎõÀ9ÆãîÒ­ïKÞç£s“÷ñ0¾ºü<ÖDzõI‹ ã!_Ý»Èêö6‘p{½X÷®òjÖãA-¾s“{³Ž÷uáÇšHÉ…ÖS¾‡w‘÷÷Dcæk6¤Éú^²½ç‹Î¯ø;¯!ï÷%ÆÌÇz|8³Ž¿gÐö×¹ÈÑæ+›ùÚ ¡Ã¹É£Y3_Û~¼»¬¦zos¿‘½g=ä9¼‹- W¬‰,¥uª±"EIqnò¸¬™ï AŒË’¯Û{È÷ð.òZÆê¨HVgî—£Óp͘¿gÐb.›û¥Ú™;w9Æk6÷ËA;egþ^Ú–;—ãÛšû%‘¬Öìë´W|›ë…;7ù´X|¹ó÷Þ§âÛ\Ïø‡­ù{É#q.òuY×*¯bMäô •Ä™¿‡Œ­äK×»ñÐdиËûyÊX‘ÂÇÛ8þÞ©ëU6×C¥9Oy^ÞEŽÖÙ\ÕüÜ9ö'g®‡<YÇþ˜ºÞds=ÜÎË›ý/Nœ¹ÞM2Kœ‹|úk¾¬ÈÛÎx1æïHfgž'”jäÜd‡·¹ªÝzòÔóÄÔõÀxÈq?Ëf>7Éúr®rŒ—lækJ@2&²zj>•M$íþðã=yNï)ïï¹ÎOÒRŒ¹ßÍ›fÙ¼oSsxç&_Ëš÷mÊSrîr-Ö¼Qxç!Çýʸȫ+®­)•ÉX‘Ô1qµŽ¿WùIÆÜ• ä|¶ëYv¬§ÒÂ?ë.GËÑlž”Âä<äxž0Žý¥¦õÆÌ÷/Z„Ç÷ÒX˜Û¬O—®‡Æü=¤93ž®vºé&ó÷Тޘñti¾cãášçxþôù^pi¾a\åé­ñ@Ëqcïøðo=Ž—µŽ7igÎSŽVÍÙŠ'1ɹÊW±Öñ&Ð9Æû]Îþx›ã}k[ã…Ä5ç½ýnIÑoë/Û{¼;Ÿí粎ñÔé g]¿>-/²×£í8žÆgûýÇõh{?ï87ù)ÖÛ÷Èkqnr´œÏžlOΛs“g·¦eA‰”oë.¯eÝÏöÏôf{ÏÙ´(¼á\Ïöý¶Žùún9¢ñ<äé|ó¾¨+Ëùl¿¯·Æ‹íoZÒgßÇÑ¢ÞxÈûyÎø:Ûïç9gÎ7]/Œ‹üxӢąźÈã²gûOÆgûU­ûÙ~-ïØ¾ÆKv\¯¶ëe]{ñæß«ñ¼d\¦Üšw=~Œ/]”æ|¶ßÏ»Æël¿ŸwÏöÏc}³}†;_Ç»^ÄxïççzüX¶'ïǹÊ뱦eLÞ­ÖUîu;Û_Åûl׫lµÒõʸʫX—³ýºçãɸÊ1^²5È[q®rìûl¿Ÿ·Ïöû{‡ñ…É3ÖxÑõÈøl?‡µÆ‹®GÙ\/ª®7ÆE^ËšëEÓõĸÈõ²®gû˜OŸíŸaÍõ¤E‚´u‘u¼Zï«»òœÏöq½ÈVK¨Æ|ÈxÈq¾g_gûOÆgû»Xϳý³¼Ù>Ö:G®Íûl?–5ó•Fž‹sÌ7y…Æ´tjä­×%Ïæó±¦ó=›–n¼Aç³½ÆÃO«¹Çöþžf¬–[ä¥8W9Žwö}¶ßß»œÏökX3_Qž™3㕼cæ+<(coòœŒûs|{³},Lrf¾Út=2îòu[׳ýz¼ãxvò¢ŒÊ[s>ÛÏÇXëÍ·÷÷^ç*ï÷Æël÷³lZ†uͧŒãxÒ’Í™ëE×õ&›ëE×õÆøl?/kµl#ιËO³î÷ñcÝ0y.ÎUn·5óÞïÆE^ÝšûI'O*YM-z×õʘýuq?̦%`¿˜O³¿È[1æy¹ßìOcÆ+y*ÆÌW;yKÆ\öƒ{·æ~±_, ï!÷nÍýByoÎñ{GÛ83Fôž¶Žñ ¼5cæÊ[3æyx4ž'Œ»|Wcåíô¡ñhÌß3˜ïes=ä‰8ÇxÑÍÍ™ùƘÜo³i‰¹Vo~Æk6ϳCã1›ã=4³9Þcñ’Í|Sy\Æ¼ïØ “/k®÷“ÿÞªÆç}æŒàjgÞ7Mòlœ»Ç#›ùšò®Œ¯óïíñîÌñXq¿5æzª<+ç³ý}[óï/ò"œã÷^ña×yM¹Nkîwy ÎEnËšý}Eÿ8gÎå=óß{–5óá+Lœã{f¿t>es>¬ë\οw5ï*ïõ´ÉZOºçk¶ö÷dUdƃãž×ÇxìЪþÌXô5©)zs1% ÑL:4³^?3¦3óþ´ ûÉ.>—#ƒ•ö™ †õé{õúŒ®nÇ™ñ‚0OÃ*DÒï¶žþG/vmÝé…ÜNï©c¨ì™÷ȼ*u7™1T®I*tb •Ýœg8ÆP¡5‘aÓÆOuŒ¡¢ÕÞoªXæZŸ^9?Ce¨éH·czNgÆkÖ‹*¯Ì®ÕåŦ#á;1Ò>é ÕÞ„ÎÔ†1v¶æt¼´q$'m¹Ä‰1n] c Ýê˜ø¢:¬ÿš¤Dâsb uÒÍŒ¡¢Ž¶™tÃ¥V%3ÃNz\™Á°X‡ŸIïXõ¼LŒÁ°: ã´â¶lH/ÕÎ LŒ †Vç¾YÕ=T½ùomeé;˺øÌ˜~¬ˆo3Œé‡:¼fÆôC=\3cú±t@_<|Y­›9ÅZcú¡•º™MG·ÄÓV)ÜßyB¬µÆ63¦¦O= Æ/Æ7­?ÍŒåiÝí2é<Ê ›O'êE?°eH'ÎAêSb¼TQ×ÒLú°^ŸDߟŒ¾õ ™ñ8ö,Ò˜Õ,²œÿÒO«X{;Â@² ƒ-õ&û²š-¶û6a¬¥æ’M˜`é'ÌõmšU”Îßó¶Cf›Æ„ík±¡1a¶…f^Æ„5–yÂ+ßîgûcÉngûY­ Û.,æO>ÍâÊuÂ(ß&ÌW‹ýŒ?Í2÷Ë4ãÁö4ë2& R‹íŒÕŒÅ¸Æ„%k1[²^fD³Êf}³=‹e ÛT3KcÂØ+‹AŒcjPµX̸°=‹Y“O³©: ûÊ&ŒYÍ Õb)cÂŽw˜Ø²V3·vÂZanáÛú>ÛßÅZÍîh–gLXg„!¾}šÅ¨Yž±š²˜Å˜°õýVﲦÙDc±š1aϼw&L¿µŸÖj¦Çb³d}¬¬ZÌaLØ|ãc›±šÍ¦³ s×bcšat“׳}¿¬ËÙ~®ìñœfh4{4æ|ܫ޼9Õ¬ÎXÍîî&ô6añý¦Ø,›0â]õ⼆­ffÆëlÅXÙœojvfL˜îˆ7ÎjÖÄÇRãr¶¿ãÓŒy¿-ó&L[͸Œ [4Ã1nÇ·7ç³>6'Ÿf¾jædÌù< Å\ٜϤ¶9«™Œög6糚E«YÍ’’õ1¢êc¯1çëd±±šåŒvð6÷Ó9N˜ÁÛÜOõ1טû©>æ&…ÁÏXUìL³ŒI34c5ú “Ê&Œ_{ûÙþ®ÖjµËÈ&Ìyj<½ÝO3&36•S<ý6Íô1ÓXÍ\:×›lÂØ/‡l®§úi¬f):žoŸf}jVcÌ|ûâc1×K5{1æz©uÆjVD3ˆäzšID×+gæ+úØgL˜¾š¥ÓlAifsó±Û˜ç™ý½—ÓL£SŒœMØû­ãÍóÌM3cžg¶—5ûó¾¸ß¼¼?Åñ¸ïOñâËjfsŸb¹·µ¿ùØkÌýJÕÌ…f<Æìo5Ó0V³Š¨ºÍ>Í|Ô Ã˜ý½Å~ÙjÖÐø{²y^\40æyqæŸ|ŸfÅìÙìïuQÌ‘­f7ňÙjV²X —Íþ^„ݳ¿ŸÂb¸lö·šE$+ kûžÖjvPO±ËÛìo5ƒ0f¼?•ÅÈÙjP?Å1/«™ÍN’§ŽÇÓïÙël7k5kèães<žqŠQÞæzôð±ßXÇã:Å"o«ÙÅ}ŠA~ZͶc1t6×û‡0xc5›X§Xãm…ñÓÌÀ˜050&lBÍ’?aöõ,F›°ÒÒXŒ™MØÞ~T¬?ÍF·&ì®è|Î& ¬†œ|ÂÒË X%›0‹_Öbñ· ËQ3cÂ"´øÀ˜0‘2YÌ™M1}¹ØŸok±u4¨Öjp³Ø?›0ƒB˜±1a†…°uc5s lÝXÍnŠ•² ,‹óõíöPãgšÜ·5Åô»ØsZÆ¡fÆ c'ìÜXaý4ëx»éóo4XÖ„ TÂÊfO3cÂ’j;‹mߦ¿ö³X÷íO3‚û1^§Y@¬êqæ|®ƒÅ†ÙœÏu°Ø0[aûƒÅZÙœÏ5^ü9FY +NÖý|{;šÌaMXC],†Ê&̪¶mLXB%lÛXÍ"ž³Xó§Of‹Ô gš–o¬0ûHÅqþ„ÍÇß“M˜Vç÷¼M˜v£ÙKò Sl„Y+ì°˜8›ëÑ^•·¬ “k‹I³6~±˜4›ë‘ÂÞ #aàdë}u ¼Ëš0…±«aÎÆ„õH]qVX5kbŒV^YÌþ¶ŠõâÂÝ­¹žvV˜+ œf Æý„©·iMU{΄¥uV×$Ÿf%fÆ CŸ,öÏ&¬®³²Ë˜°”N¦1÷Rœ »Ù·ñ ÛÚÕšûAg˜1aVfÆÌ÷Híp& O«Ë ã~ª±ŠiÂËZaÞºes¿„É+¬¸³6»œ°èý¾ûí_—Q…U³êÏX¿7ºÂ9Ö4¸^ë÷ÒLÂX¿—0hc…ufŸ¼Îïe<ë÷²*ÊX¿‡ë‰±~añÆ Ë%l4ù„OÂÚ?aÅû~eÌïÙ«öº5¿göi\οw_Æ—š1(,Ùø>aÉû{…±~/ÍŒ9_&aèÉjV‹õ}Š÷ýÀ˜ë‘ÂŒ™oNç·õ¼~¬¹^ӵљ°/ºN:s?ÑjÀä®°­I3c…e³2ÏXaÖ¬D5æyCa¶Æ sÕxÉæ~~EÕ{¶–%…ëë„­îU¶Æ »% Ó˜°¿+R£ë SÝ÷ãäöxFiÌ|Pa¬Æ {u#Ο°ÖÙ­iÖ¢Æq¿NV3ìí}¾ÓÌk~O¶Â<9Ÿi†µ;4n'lôêÖÏ'^ö¹_hå¡ñ}Â>÷ûc®§Z#hÜO˜¥þ½·¹^)¬ÒXa›Úßoφëàœ9ß×ÃïÉæ|ë3æy\a”Æ„‰>¬J'¬‘°)c…-&i:Þo÷ÿ‡³Â5² ÛXjÜNXå¾ß3SXcrÓx}+5¦˜Në)6,„SŒUt¼²æHX`rU˜Ì—u\¶—7Åb Û3'¬nÏ× C(‹ëY6e_…°•ä¢bèÊêScÂF*ë ¨¬—1VaAÆå8öçOGd륜¯öëÀ8ã~YS,J˜˜3Åx¬wË^*œ«ñbÍù:¾‹5a+5Ös9+L'Öó8aYÙ·~aYÎüž ãœÇqü½Ù„Évå\Ž=XeŸ0¿avÎ Cа:gŠ…YoçL1ö®ºíÖ„aRõ›­÷yÛ÷cM±ù®Ê.Öã8öWvû&嬰³›ýùöPXHÓxËV˜X„M9+ìëÙç«3Ų->œg«ÙÃvœ¯Ù„)°žÏùFT‡u?aCq>d+ ªí C¶îÔ4gŠÙ rf¼w§l®gÛ5ã©Çzµl­í¤®:öÅ g…­Ä‡gÂF¤eëyEa;Îc–㬰˜ÆïÍV˜OT=;s½§+HòRéD§ëˆóuÂrzµ&ŒŠ…mΕ¿7š²× ó¹ãzjÌx±ÞÌ™ëÑàx7þ½µ|³OØëV7VgÂŽg„8+ì¥ìç1gŽÇlü=o_„Ý´Çš0æ/6œ9ŸYîÌýl/l¿ŒU’·Ã`~=o;s½›áÌýhWÒ8Ÿ0Hꎜ c¸b=§3a9Wagw¬tÖï¹öóœó',â×#[ͶǴ& ðŽ®„É·ÖkwÖ‹9³¿W¼¸ÌV1Ò(±^èãüí{¬BýO‡+Üâ÷4…?X•¦P®Ì“j ¬…Ä)ÎË1&tpHì¼zªMOìâ^7”YĨ\±ñқʼnzeG_Ã8aXØiØÄÞ2Ul0>5ñ/vq¿‰Ê,bäCüäõðI/!Ãx @'£ÄÅ-¥´†MŒŠÌ•M®¯?™EìW¦ú6*["sÞŽ±lb÷^[™“‹Þ^œÝã•›Ú‘fV±ŽÌÁån(! qŠ£96q^™]é|˼¾åadvq?cgqÎ̦„Õ¦'1ªí_¬<œÑAÔœ•›ÚÖ /M3/qLÇ.Îæ·DúŠ¿9ÕÍ[ÍF3‡Õ`‰UŒFç/ªÅT›óÄøP+S•/5¬M$;Díü‹¨f–?ظ•ëkgfL¶õ­óMå+Ñ$3ôo²4ÞTµ[-R/1ì%Æ#(ùPov¥«Þí4‡z±~K4ySÉ@{Z½ã²yÏÓFçÅ&ƺü•‰©–‘™qQ½/»8Ǹ¨îf+óâ¢ÊcaLowfLkW!æ÷EMÕ$1s>ŽäÐ ñÍ¡„¾`eÆ´gSðbÕZàõrrq²g^¢VÚ¿7úýâv,â¼2õн.jbÇáíX¿¥¡¼©×…ë>QÜ/NqTÇx˜R›Ä7ÕÅpÝÔQ%ƃ ¯› »¸zâ®Á°N|ð‹—RXüb#z4±(…¥ôL%š,ÚºeÆÔtÅ$Ãøì·V¥¿¨D[}ÎŒKÁC;·7/%©Ôfúb<´>¤ eÆÉþtvÝ‹“Û„ÚfÆ@Mß<¶<´ÐÍ$ßež5À/V¥¿4C-ÐÝ’–cLN •Ęœ<ë¬ÝýÁÆ«©‡¶•™ñbéaÝ@&™4: ‰ŸÐ™Xýv=¡2|t6¦(¬T¿]T”ªŽ|ÆòõØ9vñ µy»žÐƒBÇ+ãOèM»—BAJ¬.v޹}=Ö)ÇXû“EÉŸÐ:âS4¬ŽrÆ„žB1’çÙ+ž=ŒçÇ˺Й^ÇÙ_tŒ3fU:F³¿*ŒÙ_µ1þ°­¶¯jÍþbõ£ó'4fUãvBa.åf3^ձ˘ñZo;]u<*wŒ9 •1æxÔøh™]Nè ³Œu:¾‹’݉ãå¬ÐÍp¦èzDux¶Õ«;¹ÞX=êÌõnÄñÊVöì¶é¼Íõnþžìþ-ÔÆ™Ð«²õ©b;Õd+”%ª“ºÇ3û[èM™ÖŸP›Q¬9§¼=t>Šã¬ÐŸˆÆwæ|œûëí®ó‘ô>gB‘vú`³n·¬E2?­¢ñjS»5¡óù,šy™ùÈUX4”ÍýŽŽ$Ùg¾L˜©³BaâkŠ3Ç{G´vc-ºß¡8±è.›P²íۚТK‹v^Þ‹Rº¼ÏcŽ÷¡)ÎïšSŒÕÑF¡:Îob¢9Þ—=¼}BCÍq&tâŠÕ‘Î:Þ‹²9ßïÍÊ>çû]X”“ÝOhÏsYjwW¼}ÎwB{œë í‰ýý¶>‡okB5ïèåLhÚY”ñvW¨ÉéUΠ퉎.Î\ÏÏ¢‹·õ(¬ÐgæS÷Í¢§ì~B{Æ0VhÃöÓ­ E#´Ç™ÐÑ;B?œ¹Ÿß±ú*ûzôìùœ³B"È™ù4¡@É{Q ¡?ñ`ã-ªß‹5Ee„~8· ²í%ëÑ~#¼Ÿ'Œ)ªj‘.éLè¡ÙZ”é"·5¡´>s¦¨z·žëƧ¨©ÅÄÍ™¢m¾•8STÞcm[vUQ{j΄ôÉxɦèt¹l…Ç×ÖeMQëˆuÎü{|NžŸÐ‚¸0:UÈ^ÚŸ´lrž'ta?o$+´¶±–ΙñÄj:çvBJ1¾TTÎÊ6gŠÆG|Guæ|Œ§ä©óyÄ‹=g®;ýwëyv{_o)º%}Ø™¢o=ew…’ÚàÌõxÆúg®÷„dkžBœÇ MXÞåzpŸû!¡΄–°.Ê™ûùŒtÄìrBVÜ/)º¿'›¢ê«0^_Ï)ú•;΄&Ñè'{)ÔéŠÕíÎW“÷ýÒ¸Ð}?LÖr‰XµZ­9_iùãÜNhÀ¸UtÖ¸q;w=üÞlî×ûyô1V¨ã^½ŸW™ÏÜ1°™Ýñ`˜=NQ~¬ëp&Tâ׃Ǟï%kqì.JßÏ+Æ*Z5Î\oypÌV(p£I3׋+| Ec¥Fö Z±ÆÌ™¿w=Œ×·µühuïù¾1×Ó§ñï½|BmUtíÌ|›ê6gB;žhN•}Ηgñïe3Ÿx"=6ûVQÌ®¾íÖã=ïû‘1E?%Š‚³OQféq½0nÇû|KÖ|pUWoŠvK¬_r¦h§DuiöPQ0ß{ÛqïÆýüžøžçLQW•.Î*jÖñ~[‹¶ïiMQXïqÙUEŸ5Ì)z«‹ãñö)ú«ÑtÇ™¢¦vçOÑñžï½ÝêÜãÁ¸Q´ß˲OGî=_ŸÖt´¥éEöé8­¢Uc::«h5ùtTfÇ8«(4š’8ÓQ¸ó¾?Y ²wÑèžo3^h‘=xÞ8E£Æå…îïMɧcíà{Šq¥¨rÄxLng{Š&Ë)ŠÜã5ù„¨èÐøSôÿ½·Ëéè O’ësЍbÅýÇ?‹¼þè ¥ñÿï·ú?_åëŸ÷ò¢ý’'Ÿ®¯ûÚwÆÇÿúõ—ÚðžÿY,àþö?£ËØ×_*0Óÿ¬Füõÿ3¼ÿg÷ûoóçÿý?þîëO6?ÿÏú÷ßþæ~ý¾~ÿ—¯õ×úÿçÿ Æš…cü×üýß¾þË?üÏÿþõß¾ö“Ôýúýÿþö÷¿Çâ?òov¶ÿxû«¯ÎõŸýGJÁò·?þÿõÛÿj¸ endstream endobj 60 0 obj << /CreationDate (D:20150701163318) /ModDate (D:20150701163318) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 61 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 63 0 R >> endobj 62 0 obj [/ICCBased 64 0 R] endobj 63 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 64 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 57 0 obj << /Font << /F8 7 0 R /F54 25 0 R >> /XObject << /Im4 48 0 R >> /ProcSet [ /PDF /Text ] >> endobj 67 0 obj << /Length 2642 /Filter /FlateDecode >> stream xÚíZÝã¶߿¸¨Ýžu©Ï¦[ $Àõ¡H› úpwZKö*±-Dz×w(š¿½óEŠ”µ×\îáR ZKäp8üÍp83Ü—w7/¾I“Y•™ÎfwëYçQV&³,Í#mÒÙ]={=ÿób™j3_-âùžž7ý ^—CoÏ;xð<ÀóF¥ ~¾ƒç Ϟǡy-Í'ṃ§ç93¬å³asïÏü³¶ð:pó™¶ÒÓ ›FFˆ8qm9sÛyñ÷VÄ·?_¼½ûë¬0Q ð-ã8*SÁì"S!ÿHN8ýE¾i^d²œâ"È÷²†<•0y?¬#TŽ[Ï—>÷‚‚ÕÔ‘DĹµŠŠ"†©A„4ç©-VŒÛÒdYNË_/–qš³ŽE‹1s[:vÞJP”3‡'‚'“wC¿ÌRKŸ îS“}8¹³²–EÅWA¨ãl=cèŸÉç£àú£`ýls°¾ñÏÂæèÖ?žÛn*Ò&£®ƒ`¹’¹+1gϯ[Ñ_%ÙÐ/BaâÈhß{TýÕ®õW\{·f»ïuk÷ ëA6ÍS&ß‹Ùÿ<2ýÜkn§(Y (¼þ [šRéçpKWãÇ(ßJ›ßÿ|XC#$wðüžïáù:ð´ëFÊÿ7v½ø¦ð<üÒ¨63ÐQÓý­£íþÇÅ2IÔ¼EyVvS×­H¸/  ^tÃŒƒ³£("ÞC1˧}ï•H¦ŒŠXÛDx$ài>ž~˜š÷~²#Óùü8ИÏWÕ–fvÕ™ðë¹»=ñï‰Æ:®ä +¯ï¡á÷ òoHQ $>Uσ̦ٶ¢R‹”ë<*²2ÔÊã"Íæ$æYv¥Wík|Q²J˜ã`vGyYñ@èDŠmurçuwc4ly¸˜1\ÃWHUÌw°:€­2)ÙÊiˆÈ~iEÚ- ‡]牵&`-9ávôx°Lå6¸ÑƒŸëÅñP+!îV·kDÇÔg’1¡9#³ÎCóN$ˆØáe¾«yt¥‰¸ßWû ~Ä? ÊVätlmèVÝË·¼6C¡’]h`Ëg.¼èÎ …¾6N³(vNšý~ël-ãñr²Ð4à›¼PkÏ“[ >ɳËAÉ'†¡ìŸºƒÝÌl0æ=—yŠáwr@›æ$ΔeT‰Çek÷m„éìíT‚ Ž= Ò$ÒIf1x£L|S8ܔ燬½T{~q.ìàeÅ-+﹕ÂÂí@e•t!ååØº³‘¸žú ñ³N*wN È0„$ƒ,M¥€H%*·‘6ôÜ|}wóÓM P³x¦uåe1KàìO nZín^¿U³:aoDº.Dº›§Ôà¸íì»›¿ß¼Ä 1Ïñ‚ß( FYͲ¨ÌM„9t-J Ÿ$ªªo§ ‘xV@äZ0Ù-- ”œCPE¹ŒþöˆZµpt^0ÕH¶DCÒj•V}5¥ULn… ‡-ú$+EEY:ÒI^ `‚ͱ;žæfŠHÇŽx%F™N)¬_©ä)Ñ?²pO‹þt:é*ñ³|Í*‰²|Œî}€2Ià%‰Š$±CÓðœ*òÉy3é2©"NžÒÅׂé($û \üeyòYñÒ¹«ÉÃpßPÈëÉ!K¢Ÿd™ßu°ÀÊæÍÍ–“V$´þßù°Nª­ËÉ(º$š!’ÁS•Ïך»î=²j«Û´öl¦cŠ (úØÐß¾· vìØÅ‚³¤Àt TR–ûTµøZé;Id|¨Ñ¶J²±Qpµô HxH'…Âø;ÕùÀßÑÊq98ù +>‰מ6vš9W¤CAþÝÚ“â(9†±w[I˜´eìl„‡}¤%§Sà„ªâ´¡ZáÁòÀ±}ÑØ?Eby®ìÔ.>ðž%hp õ» M—3€#nÃe:ZÝê|TB÷ É$È¢•†Áµ{1_œ‘™È‘:mRîDÞØs¹ï]œœw²fký}4‘þâ£úea-ª“ € JD^õi/…È«Éo¥åÓÚ¢–ŸmÛ Â³©¢IXXÜp(7ª™T"[$“½*ÈÑ¡LÆÅš/ÅÚÇEÿ"ÏvTÅŒ¯dÛÈB½:Ϋáu#Ï­<ñTqf=¸ +ÂCá+’Ãøß‡å˜w¿¶^œ›Hi¸=¬P;äKÍÊ*ýÞ«-t-¿Ê^3|bý‰ëß>Äég†Ø|"Äæÿ!v™ùpçãÁÛŠà°µký€—x>Táîø\Ýi}í£Üè“,÷à ¶|LdAþ„1ŸË~7q’¤Y‚£W>Œf‚ySf~1{wïÅ'wi ÃÖ¡kúH4õ'¡©?Í8Ï)¿ü@S øW`b>•D‰ú­Y˜ÿÙI¨Äåäö®\ps~’0Âö XÈô…w¿° ¯[ž N&ï`UxúZØ¿eVáÕËõÅŠ ±MrêWkˆU&e)]&*—ÚŽ×Ð!äÛ÷L®¹ÂQÌf9tW^Ámï’„Lqæ¤ w¹´ßl‡¼û]„‹ãv~ÒÄMmÏ¿û†³4IÕ€‘㾘+‘Ù-c‡æ“9ÝN]ßaŸÖè¹;6”Þ96Ћý@š’«Ò^XñX¯ ‰¤}sò¢g©V"]·f  £*ˆ±÷˜)&f~yp{’#^#a‹K|5[åµÛË’½`/g(«“ÍÇ´=®éõ¡ò³óDŒÌ×7Xn(3¹ªA Kƣ˔RFüµÒá{g#ö n!Ï¢æÚæókø²ˆS´=Ó\¬¥y<ݵªÜ{:§2|»ê"ôð†Ä()@ˆ´½½Ùp7SoTœH•XÓ>ÙØÔÝ“f¥Ún´Ö¢kWN͇ êqEõ©ìQÇ qTm1D2WòŸMUþ²çÊÕ"œ½DJIÙ)y"àçSûûÖ¢˜j×géŽ?2/Iû}vXÝÇoòÂ{ÅwNôN·Ûý¾±ÀH±êv^Å–pw-ƒ´þ=Ö½½yêåLœÅ–~L.wžþ3ÎÍ?aMY¨4ø®êÚ;7[W€.wå¶ë…ë^LßËR‰ ÷¯©>Ïï½}{c‹jï|H+ïòóœ/iäz7 ¯o'W%ÊàÿÌÑQ‘C QcQH> endobj 65 0 obj << /Font << /F54 25 0 R /F8 7 0 R /F64 68 0 R /F11 9 0 R /F10 37 0 R /F14 15 0 R /F51 8 0 R /F56 30 0 R >> /ProcSet [ /PDF /Text ] >> endobj 72 0 obj << /Length 1751 /Filter /FlateDecode >> stream xÚ¥XYÛ6~ß_aä%+¢.K=AS䥨>4JKZY‰Dz³I ô·w.J”í[ YÔp8~~3\½ àO¯ò¿I³U”~Ç«¢»{wççiŤà4©KË0<Ù%«ï‡»ŸáÏv­ÅäÚ±ùíöîù‹l•£¹tµ½Ÿ¦M3_‡›Õ¶\ý¦^xZ5ðÔç#üVÞ:N"•|î­Ó,RÛ= Bu¬ sýà……­¦Þ{a¦N8¢d SËØûGøÁÞ#šÞÁs>ÉpÒÎÖ£›çá Í‘uÀÏgÐÊBuÚGï÷íO€àZk?ORÆ{ªÎ´4)üÀ€(ªWßA;†ó £DÿùÞ:É2F-Êu2¨UWä%)-ÁÁva8°›¨¹pA´é4Þ:LRõZGñgO<&jšŸw•š¤f,`LËbœ±7èÍœßz›é‘0ØÑ( ©®E‘Ñö[:ô -CÝ8ØÓÚ8 ¨}ë@˜ûY–±ÅÐÛp3‘æ··{Ck³;ŽjFÞSô¶ô,úД!jF ž‚ÀJ3u8²»">yëˆx'ZöK‚€`}ïÁÖ˜É9:¨B;›òZöæÄ#z[ îžÝ{»!°ªÖY Ñ[4`Èž*‹MD Œø]ò÷|Tš8dAÅ´…݇‘æú(ƒÑ5ZòÛyÜ€ÍÝÎІ,À*pbG0mãÉ @ÿ£—Å·Q ç ÚÑç´;|Ø*ÄÑŽ'G|h7GžW¸Gtž¿Hb‡ö¢,÷Ó Ì+CœC'$ #õá¹¶J XGYŶêr©Îü8M¬Í‚Pmï0CF >¾ç¨#á^©ªd›B9‡cEcev´ªA†eª`b±Åm’ã…£:ó–ºg‰àMmCP,‡i·FÖ²ž6³È=øZ}&Ô/–°é@&‹ ÂAõ`d |m=PÈR©±_r6…Žl–Î Ê|’1£Ý@jæ›TýmýTbX¿Ã§Â\Ì(¬‰pü‹íL…åÀFzK/¼–¦(d¨;89‡—ÛγÜÝÏTdÙ®ãìª@:ñ±ðíÄâTaFkjsIr#Ž£ÓÑÎ:¡W|oä²ìYÇAÂÉjävi½ª­HHH¶P¢ù¤X¤n,o‡F]ªÕ¯Üi*Gù‘ÍÏ„B¦Ý$‰ª{¶ ?œ–,wƒ†=†åPO›u¶LC“¸º[ð,aÅ8ݱ ¥ô%~ÑD>´¨²(5QÁà+ºA~ñ\ÒcsØ9óÖæ4p}ÈÕÛhîåéjÌ;¹Û4‘(cæìh¡‘.ôì ]óh{ÚyÜeât'»‘zKÀ©¬ J‹F’·ÔåÔ;åÒE=,TM•Tën_KµIGÌáp1V¾‘5owËÉäô¤ì7ß¿* _?¨Õ«†r.—Žd˜kßR” §€7s%æì"ðMË8Ø›ÓñÙ­ÀÂî#ƒK'©²YÖæfÍæesFqÂ(8W‹zïH*nŸ{©û‰hˆ®.̳ 6ÆàpÒ¸½5o+ç^f-Å%sªw+qÈæk=Ÿc;ªF¹ú Kvgxa}wÐЄ֚Þ××|q@GïêOS>¢®/ñ¿%rÍh‰Ø1“^/çf-ÏWòhîÒlão~]ÝVè5_Lˆ·QDõ‘¸bÄâ½ÀäÊ96‘d%ù½ht²(N+lé‰$[»mƒ£gD÷‰õŸRä­>\x‰5ÿxá£ÀÓÿ¹Ë[`…Ü£Ÿ¦Â=š¸ûa{÷®C·Õ endstream endobj 71 0 obj << /Type /Page /Contents 72 0 R /Resources 70 0 R /MediaBox [0 0 612 792] /Parent 47 0 R >> endobj 69 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc4.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 73 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 74 0 R>> /ExtGState << >>/ColorSpace << /sRGB 75 0 R >>>> /Length 1004 /Filter /FlateDecode >> stream xœ¥WËnG ¼ÏWð@B°í&û}Œ$€€p´€¶OòZ¶`)‘#Èß§ú13½’`OGí‚KÕ°šd9LgÄtE·Ókº%“”ñóg¥#‰u*xº;к™^î§÷üú’.î'­´ÖÔÞ_ÜL/~LLgø»š¸<ý·©6THʺní%È)šà ?tÛ\­j­^“¿fw3;¿Jwþj®~—°ú›Ùù+Å¿²hù5™»5Â>°‚²¶¤Õ=Uo ™ãRK=Šõ¢|k¡Ñ¸¢½2+£qÅ£`—¸½¸\OV8g¤š_fS¢S,°[UŽÍùŸÏKK>S‹„Y%DT±^ÏŸ.ty÷ç׿6ŸÚ`Ë-µ>?áíÉ›S2šNŸ/?ý}ø@ï·_7ÿÒ»“Þž¾§ýY»OcJËÿCi½œ:À¡Öa»Ò:ep®(Bó­ EóBƒŠ6¹SÝ Ô‰ Á=5šsùYÜ|ŶC1Ë´¥uf j1ËØÊíµ¯2o«µç\) FIͺÙ\0È 8LÕ Az£1‰âÁ¨¢£rðCj„bˆ{_x79Ý 5г(/iéÐå -a jq™0‡laýùÑb9~`µ]™ì’wƒš»Ÿâò4–µÍl8…\õQ°3ŠQpƒ…Qp^ùPk†,ãªAbÙ`K£`ƒËæu)Ûa0VÇ,-¨¢Œå1²6‹ûvã2…Ù„2XO¯U.æ]X ¾•u˜W›1%K²A&he²-8EÞHq…„³[ j!ÄÙ6è®ú¹ÇBˆf±yÛ\ÂHV”se?žJ™|%ÅÐ$L‚…[> endobj 74 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 76 0 R >> endobj 75 0 obj [/ICCBased 77 0 R] endobj 76 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 77 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 70 0 obj << /Font << /F8 7 0 R /F54 25 0 R >> /XObject << /Im5 69 0 R >> /ProcSet [ /PDF /Text ] >> endobj 80 0 obj << /Length 2235 /Filter /FlateDecode >> stream xÚÍ]oÛ6ð=¿Bh;ÌicF$%QÚ–-ÐÝÓ¶ØC–U–‰íÚNììa¿}Gòø¥ÛiÚ¡„hòx¼ï;’~wqtzž&QAŠŒeÑÅ8¢EL„(¢,„ñ4ºE—ƒñ1”Ðî M¡ÝB{„ví´O8>ƒ6÷àJ„•0Ç)=¾ºø- Nâ<†”’"Å~=¦Œ+à1.^Cã ¹šúºCÝ Ú;6†Ï×`;ÃÆÍæjí¿úS"P­¾ÑŸ•¡ƒ-´=4B6Öø=C ›ã'Š·.Öþ7á½ÔŸ 21E¯‘Ê•žã6Ký³Æ*iÅ»D¨{h =|¯p8Z'NfS${$×N›s¤bЫר¯;q'#f£`›šŸ¢ôkÜÈ(¬GõÔéQvO‘Ä%®YZ4]ˆ¥i„•tâ¨{DLµ§Gƒu…ss”“¡ÑêÑKÎI N§%i <‘AöUd߈AÊb‹PÅÏ “u2ýò.ƒNЬ×*C¯oZg3&d`éEO™ÝaÄz·Ïð»qÒ¡;œç%šHèÁFРĸéÜùX#*¬1nÔn¶FJëãR¤»¸¶q¡)ªuï—ØoŽ¿ò”·µÆ;CêÊüÔ•Ä@z?"ˆŸÝRÂXaÎ;pH€ƒœ!WÚphš‘ föKBó3qÔ“J‰œ4Ù{Ïh–ž“ õ!nª¾e1Y@_¥b\“–çy×ý{^î £Š¥¼.qÉÂKŒžû›?ifí ô­#öí Rþ«ÀÇOÞö[Üzk˘v¶e9pžpg=ûðœïJ=žDôœ¤þñªOµ‹êíþ’î~ý¶ÇR^,˜ÙãÛ®Ð:0mpÐë@M°Ðûö\ØSp9á"Ý#|næß<¿Þ²îæ×‚cqÖÏkïœÍÂ*Ÿ&$N“]‡è·¨ÈQg=|z)ê%» \é9µ±QÃä{”P_Šy… -Û–»#´ÕžšT$…zææ&b Mdüè™ßªã¾XžõÆò[ÏSê'{Jã{m_üôpÛ?;t=3ÙàT@2*h§a™P>m ØlCÖÆ7_-¡§ç™/¤!…¾¼GáŒp´’?•T¥"|8‰e•ÉÙ žê”ÝõJ#Ë=\EBòÜüíêXf ßùåÏÍ1:N®YnîÖ‚©¥‚Í#ía4™!!rp:ÓßÚ¤ÖRrs¦ <žòæ¶+í;ª?kP½ŸJäš”•Ï—¸EYÉñk»LéYshYm&w*ŠÁD÷ ä F¦øy’u Íê`f&)ÑBr&ËõRÏ)jå÷“:Å/•A¤²-÷š™}”ƒ¶â&þQ‰W.rN¡Y«¨?Ì>øh¼¿BhŸ›C(S²•‰>ÜHUë¾Ë ¤§¹üòh%•¤àç|VëEëƒTB=3Ör­§Ô€?õ£weYâxƒ±R²ô`NJh‚û¶Lé3è Ÿãí®œàn~¥»wFp•ì\ëE·¥Üe¢¶é§´¼u ª4¡Fé{fu­EÈ8µ›q 9×]jé36¤«á>·ãÔ2ÅÁlŒ îpN ™ÇFl:ÆR¿9ÔÂ8ýF†¯5ÊV|à1™-x<ÁêÓTË)&ð{gälYZÓ^z¤•3ï‡ãzŽε•To±Šá ISŠ}?*8sˆ·êêä]çsÎ"¸³4^ꀪHÀƒ‚Ù(ÊŒ•‰õ›ÓsJƒJ„ìrÝAx”‘Bð\‚A:âTSSªÇ¼ÕPPs¨ÛYA(„±]£„Pš4Ñù Âd4ôàκÈO /€[Xdœ¶‘e@˜ðNq*Nr@‚0åÏÇCš‰AG&‚-Áۛǖ£÷GŸ¤ãˆF,  ›e‚ÄIUwG—Wq4‚IðG‹<Ú(лˆs™4ÜÛèãÑGïä»_@—Å• Ъ`Ѧ+%1åÏà‘&½LÞt bàoV¤Ó‡$ ç­Ã£ 8Wf$ç‰G*”Þ) Í …݇LN!NûQ#÷ôý»\.&ø:y§‘$9°Í÷ÐÊÔ(/€¦|—F)礳ù\ÞÜ}äV< ø™“˜÷"°ñþI`õêÞ.K¬1û¡½'’†ÂÍisçC c?EOÀHÊÕ„jé,Ç!$eÞÍ¥ÊG1¤£Ðôb¼à”O^º0ü&úñ–¦Â]„Ê9Mö \w”™‚‘4.T5›jš2S™ü‘¸vt endstream endobj 79 0 obj << /Type /Page /Contents 80 0 R /Resources 78 0 R /MediaBox [0 0 612 792] /Parent 47 0 R >> endobj 78 0 obj << /Font << /F54 25 0 R /F56 30 0 R /F64 68 0 R /F8 7 0 R /F51 8 0 R /F11 9 0 R /F13 81 0 R /F14 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 84 0 obj << /Length 2535 /Filter /FlateDecode >> stream xÚíYKëHÞß_‰Y¸IÛåò  ®hÁˆ+Ó‹aî¤:ÉÄ!vnwoøíœgUÙqÃfÁÂr=O}uÞUõûÇ÷…]4«¦4åâñy‘eÕªlì¢,ª•É‹Åãfñmò»»eaòd}—%Ïðíáà3ðµÜõ(.è¾WøÎðíàû[Z¤ðû¾+|ø>‡ægi„ö>߯˜àFª­ ‚÷þ“[!Û ­37ÿ2`:HO/dœÌ¸ðF)ãÚß=~½hr`LµXfÙª)„› þ+;˜¶ úY éoý@Ó¨¼H-Ë.çÖý!<ý__*„À}óc%ïð#þõ¸xk,Yò•츗1Ô#^Ï‹ø‡§zŽeIo$…-䫲–-|+ýßñJXD°|+úûfüJiΑÔýCù‰ee(.Mº*™ö㎰,óšyäg{œ¶Øe’ Âí¯¢ CÏ­4‚‹/w¦J‰jwgjbAÏdUóÚ‡¶Ç…#ι_ °ãƃ‚ٟ žöIÖ¼íZU·@ÛKG*Ü–E“ó÷^úmÊÄ]÷u²‹]Û{D:t"¹-m•'{4±3­<`‹¡¹Ø£Ût2y³úh²û¨0—F È«4­Jï:ô;#¼Øe²ØžX@}=Ù)7ñkÃCÕZ/©!ªkZÑâ{»v<ûEw'RÅFå§ !Ï1ˆÄ‹½ŸyÈ2/Ro\‘{:«Z9À–Û6{àÂ{êdž°ëMÆc`î¹îW†ƒ¼mS&UΜx:ÙÜÝi{Ù‹ño…Ì ‹qkUמ ̾¼°´B‡À8¸°7Àþ  3Üb;[³*3ŠBÙ+o‡5φ‚'‘ ­Å ‘–‘²zç–´®½l‰SÙS­{¾8¨Å B$¶>’ßm*ËÆ6¥Ø‚wû/ËšFÃÚ>Ú2ÛbÄHÒÕ˜eâ¦Ð³ÒŒ‰ «äž®¬Ð’9y#‹te޲U;ëvÇtÔ™á”xÐÅÚã¹€B8Pwo”Y-×{'̼KAúì#n°hv€¤N("䀄"ë’xãq eëÈÈç[»Èì*·¥A¯¬Ì ä‡ ³²¸Κ"ù(Áè´ÙÓΨŒÞ'¯,ƒvàôòöwI€¦ÉVuZ3Í?ãL“yµÂò‘œÐv¯d±­=‹]:03ØÇŽÛEøP63«1sàî'5•ö$Ê#Öb5–ÖIÏ„{EÁ¨Þ+Á;\%•M£Ëivñ>ÃÒ-®ýýU}’£Åf Ÿ!¶hEyCLËKlÊìÙ6ûÈè: t2÷Hs<÷òÒ;ƒÜSÖÁ› %ü•‰*¢Ó8µÔ8aPïÎQp"ª! K?“BÁÐQ˜Ãd×&ß#uXcÆÀ‰ ˜!˜ªL°¯§(˜ªæpì²XsÛ–øAþìÛÔuÒnw¡'¤©Â+{S7Aª:ˆòêíßÙÆ˜ þ(BÕ!BUI´Œb[–0à1 ÃõËHÛmÜÑìÉ)#NsjÔMÖ:¢ÒL»ëc«:ó.H›ç9‰±\ÄÃ]mžWãy}5ó‹ZØI#ÅRï~Ä´53V¤xaã°•†q=?ܦdHiˆ»°‰g5RžÇ“*äú ßöÁÁÊ!Æ©ÉR؆ ´c/Ú[ðê²ç£\«!rGÙ Zyj“O*VÞã(±†©”9 Õ¹ø‚Cš;Fbf=ó+;ökˆÜ²ylæÍß?Yä°s ŒH-°„Xñ Ì_‡ƒAcý(¿Í^6ºÖc¨rƒŽ]yiVµ­ÇÌîwªMò')ïÊ»œ~¤Èÿ²’ÃöŒ)ùìÄ‘Ù÷ÆXJdÊ['KÄrjw½ð ç>Øz&qŒOe*ZöÄY¬Nò%/´ßÉÖⳌâÁ™g¡KšÑLO87ïº]蚺]Îr>J Ž-×gD_3S­Jhä `EGELlòÍ̃ǺõúD[™ ýY½Ê댉ý…ÙÒãàƒõöìSVÏxÓBFn%ùÄ£ìREæó'%ªXŽ#‹ÑÓ^mã[P²Sñöbþtwa-Ã\ÀÛ«­^c‰yÓ $T³¦xÕ†¿E¤ISôÑÕÝ8ôî½ÔhàCNpìÕË_/èÆŽ®ö´î¸ÜÇ÷ÕûèøPÍ‚²ókÐR¥r¸|"X=ÿgEX–~ß7$½®¿ .õ±„W8bªæúÇn9ÇŒ&Hܱ8TŸSÑg<¹³× iE¯œ?óˆYI¦ÙHŸ³‰\¯Gõþ^n׌Œ+ó`ÞK¾ÅKZ½”Ë寶ÝpE}$%tŠó£4ðÜ^¢£Ê žÛg4^E0#ꎒm¼áIÒã$ÕFŒ™¥Üuô£œ}u)MÍ`zª%³Ÿ¬Ë·“ùø'—wö_³ŽæìåÏ^ow`’›žîsm(£ª+3z±h}TަoFýäa(~‚ø²·£{z;àw„BÞŒ”ÿ+ïJ$«Ê¬L•O²yèq´bü ÕKëkÀ0÷p^‚fé c58P3 =Ê×Ê6Þçæ%ãŸs¾Š…‹+šÅ“ùçž_‹¢& F6,*oC¿ˆÖ}Š^š>s·FiUÛ!’ÏÐtőօöÖï'««U“òZe£ÛíÉc—Ü 'Zè"rŽ äâh~£FüH‘¾*Aúåþñ+KêÀ»±f1BžÑ–R¶ax¬n¼„À¬¡7¦ÌuCåÉþ@jÇ ÚRý§]D ›ÏérÂé‚ÈýLAŠ@D=ÔZõ(#õ(œN[Ù[ú?ãôúSÚðÓw¤.7WYm %hp6]Õµ¼Ud4ðÃ?ü ðb6° endstream endobj 83 0 obj << /Type /Page /Contents 84 0 R /Resources 82 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 82 0 obj << /Font << /F54 25 0 R /F8 7 0 R /F44 6 0 R /F51 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 89 0 obj << /Length 3540 /Filter /FlateDecode >> stream xÚ­kãÆí{~Å")Pù²«Õsl§íIÐCŸ Ð.Z—+ µe[{¶´‘äÛÝýïåsfôð]Zôƒ­Ñ ‡Ã!9$‡Ô7wŸÝ¾Y]­ÃµIÌÕÝîj½ —fueòe˜¤ùÕÝöêmpwXÄA¹¸Éâ,h±ÙÁßù=üŠz‘,¡Ãi@#ó~=ó<(Ž<ÚÜ/nò,x(a`#3wÑ1òlÅÏÍ:Û‹d h÷ÕâÆDy@£€m_òÔ}Û à#‚ðÄWüHøÑáäçr2œòã üé[nû¨BXme‚»Å (–á²_¼»ûÃUtuÇá:Æ‹›t™Ð¦«þöGfTºŒ‡{ ž]Ór¹vPÈ$ø!Š3„ŵ{†PÒ°mI£·'d¶ ¬qkåÖ!m!Iå½]9a:u.JeÀÑÞ›^àS‹„qjãú‰F^ëàöYc_Þ"#Êç9Nmšâ¸ÇÅ‘S®]5µÈfY¦¨f( dè wôÍ<àémÎ} êe¨‘q°eèNôðT ¶¢õp’H¹ Í" m<<§W"2·Ê$kžïqk¥œ‘pŽ'ß"O–Á½PX+oöÀÐU2Ðxݸ“…Âè°—'v Pp±¸çv³ãgÁ5Î<ó<·›!öæ„<<÷V<ØWG'¿¸ILPnŠ3o:Ëbƒ¸ÃeYMNJɆ°JÃþ›V˜ª8Ã%LÊyØÍ¨àФ¤‘%q-PºÂC/ ¦"t+§’ßeUæ¶œ:óQ|Í ]ƒÆ öÌî¶[ÈíiŠrOf°“Û7yæYÖL*h'ìöÕÉ 'ù¢O/<Õ·ÉY†tâ'ÖlAUH¼:ŽÁ-ˆù‰Õ¼ôÜ?2Á¸!dúM†µZ ïØ·#aG…4&xªj»èšs;5Ë0ÍS¥×ò_¤ÞM7˜¬Ã¥›P ög‚ Ñz$P]W ¾Ï1$Ä]¦à¹TfW’fŸYqxW¨;…€.’$õè¾aå#çWíkµBPÃMžAFó‰ |#3Ôi!¶&›jz»÷ð÷+4Ñe3Ra ô5Œ0j;•[š…I–}ÌÏ$k6½›¤Ž{ ~ìôÚsÄ÷„C l s`àê:×לËþÿ^˜!8b­!bkçc#Rñ,‚¯eQˆ’ÁI|2ìŠÁõ¨\µ§¦ÓLq0çÍkÑ[fÔ P´)„ÎgGw{ZÕq—ÕŽahÜtzŽÒÄú8=Ñ©·f\‰~h7­‹Fˆxž„¸ãà–/¸ ¹­ÔªÈïU<6FL%¾-"ºö(&éèEÿzw¿š šµe¸NÞþk`=¸ïgoQñ¶ØýkhÞp³“Cú#¹¨<‚G$—7Å("s¯GïDïîîüøåð[2Vâ([_ a!æŠye,Ò¥|ìåtK&”W(¸bh}’¨9Öè9ž”×nŸêHú ܈“Øó,GjhÈz ·îlä( ©å B|áh: >—¢aJŽ ²”“mæ“mêájg…í1Á i~û4¥!ÚÙ½@²Mw¬ß2ˆ[Zeöqri±1™OB‚ª—Haç­©tËõõ^X^ºY…è*v%n†2êYc·²=_Ê¡`¤ï'¨ÖÊÒÊSoÌÝ¥úõ¬ÒËúÂ*±½z¾šè׿s‡)DT“ð²‰?Nïˆå½ªÞÂOW§að/GÄ!V¾²ðqðNU?¾ ÔsûÞ{r8;’«¹ƒ>ϲ‡‹,SÂÆlK"¸{ØÖyáÌ<ë.¬V{†˜«Ä™¯çl>¦vÜN™¶xÚnoFѺѹK(y‰wŸ°³Ð_úGz¸»WVã‡ø…ô;e¸@¨oW>Mè' g­~&¡½07­òO2.3Äÿ[:3:%CƒSjþKcþ£ŒöNËÆ ¸$ÿ×Sϧ9Þfˆñ#Ê7V O_aIºÈQ‹ñ£–Ì„ñ2Öøÿ—3M&ÉZŽ38àçcð‚˜'ïxNaìµ7Ž ±èQ"ÏÌY¹k=6D21ö'’ÄÑ:4Ùè^ªÁä-ô¬Ésqx8ü9üþ1 ‡¬IUzrb+<õV—ùùt7ãEþúј…a¬Ê¦0‰²YÅ;z š:O…}÷ãíÝ7­ïå ŽuI}3 Z}Nþ¿äý,”n¼Ý—^¸g$2 Kîqfx¾Dð)³é&?Ïzi‹ :^|t"óš÷Ø´m&†î½ÇgÌGKÿn„¬äûË–'oœpåªÕÑM7 ¾··³Vu£¢I˜u(=2¹3ÈxÀæôF¤Ÿõjµ}=ë앆O6¹³-…†ÙbÎß°ZX²û¥›÷¦P¦ìÎ.e°á" Q2­©§)„ƒ¦™K—æ‚q_Œüâqg¸K«û–—'F›7#5Öte5’öi*zo±ÒmõàK—zŠ#ñ­›7qÅméç$*Õi¹ÞâÞ)sQ[ÉD¥É°þPž¼h¼àîÞÏx¼ÈЩ9»tûà"¼µ»­9k E»åAâEÅÉ*|œ…ëld`îü¤Xb2VOlôˆÄÑpÇ©à´ê"¾Ûðœ¶úa‘åžb0¶æÄX%Åm{ç·×v¬ø¨¨Í8+Ãi¤„#5Å¡%èNrCÐUéÂ×ñ éç| d›Ì çˆÃÓ„;ŸiÆåM‹žÒ,©–J©m“0Ð}b$œT‡:Ây:8öyÊÂæ`˜JªT£P’t.¦·Þi¬¶I£n²ì‘sƒšÍWA5K&z¡‘\Hí!ÈŽücG ™£#X3@¸Ç¡wRÈmº[JDËÆÑKo$ÚãþÀ=¶b3ÁƯ›wÜC<Õt¬WÒͨ¶a”ÃüBãÔì)ß\ÒÂlÁ—öÓ"bPTǞʭMqVJòaR·¶»G©@‹Dbª ^¨k•OÖ.‘@R/Z-˜­YZuØ¡°ÕÉBp7YB‘H¥¹JžEE„Ùóþl"4áÏ$£/T숂¿lõºþàW©í>ÊÞaårá˺XIj­ q)<´t¾Ìê‘Rg뺷.]Úóâ ²±ý&Û–ÙF) C@ éçîö’§´3qõÛ²®~eP+¼‡uæK#1ʉ·ª ÅB†|Ñ]òõ¿“ú®PîÒË×üµ+™¬WDÑzü±Ð¨·¨1eñgÝ3Y)€}/=[Ò =ŸëµµÐü‰Ök\¿ª ¼ž5£#ˆº^%±µš¿– ªmdýù€ñ‘EX¯ü0`†)£B3f«½Ä"ú¨Â`<º¸Èd5)^zÅ-®2(ðYÓ ¥x+úüi{Í•¢£¸>ëÌIÃKX¥ƒîïŠï¸S"ïZ •8Vð+ã;WBàô ³ÞÏjä–¶Æ(þ¾{ïêŸfÆ®b3)”¶Þ”^0bÍùFÌ/(KWhºÀD¬v½ièDôHgï…—Rþá²A=W?‡«|'äÍL c?ýÔ·~_BáYƒ5,]ñ§3í¹ òÔÛF+W?êøý@§QI­ý¢êx]aþi­Ì¢PŒk>œ´¤aß÷SÇä<@_³›ÅEž˜(: rÕËï± ¶[žÛ5;êÕ°ÒZn@DW rïI1÷s÷|;iõÓ–L¯G±‡tË#ÎZËÜÕGY—…uc£‰dø(P7\­hnœ ƒñéÇIU+EåÖž©^/ʨÀ¦sºÁwmÖÒmÈHšT÷SÉù¨”ɳîÒVÑ ç‹RÂÐY<J+«W mªbö"!¦9‡cŒÐ»½Áææ’•À‚åvúyœ†+¹ Üœþ”ê«òá]:Þ¥µ¢kܨ!ßh`Û~Ë’s4—sØÙ_ºšØ-êË„t-ný`/‹N¯ÞÉ͘”lÉ9ßw‘ô²áv³å˜Ì8s;¥GpÏ•Á³Á‡r±uÐ_¶Ì›é7^Žg•¹fœ•¿]¿qÂW.+W`eÆß¶ûrâ*‰’pµÌð?úÄ8ÆÆ>ûíÝgÿJ@Åx endstream endobj 88 0 obj << /Type /Page /Contents 89 0 R /Resources 87 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 87 0 obj << /Font << /F8 7 0 R /F54 25 0 R /F56 30 0 R >> /ProcSet [ /PDF /Text ] >> endobj 92 0 obj << /Length 862 /Filter /FlateDecode >> stream xÚÕV[oÓ0~ï¯Èc"×vœÛ{`t/ ¤¬mÚ‰fÙš–î‰ßŽÏŹ”!PäØ>×ï\ìDyÒ>ÊËS‘&™åRHc¼Y5z‰<Ñ‘AÞYŠÕ˜0~]%Þ«zôÁ>޲ɰgóåt4¾Ê¼Ì%Þ´lÝ&™P:õ¦sï³(ÿÖŽåncß‹ Œ¤ñ“ó 4:ó'¥mA¢²¯V‹9ˆDþlg7›ï•Z4D*ÁB]‘‰‚&g`ƒÊàèZ*Ó³SÕA¨S¤µ}9÷ X×wK²²¿e3+RÂeÝt€KpA®™» õîÎnf5 ÃøfƒsRd4[’.šàËôMh¨”Èã„RØŠfå@GÚ_°¼ Â)éLÈD“R³e—ƒÀÙ³‘ÒÇ—!Š€©oÖtž“^«Ó¸ÅŒ”ö«vo“¶"$~ƒ‰ÞþŽ˜76½ ¥–è"cü銲ç´(GLغжd¤­,0‹NééÊZ¥Kô÷H›a™ûIŽ)_{®¡©Ö\(ò—`–zŒÔÊmCQñfã@ïÖ 3¾ŠMï (»´«PÙ3@ž/‚0ÖdkÇ5ûŽ(ª’ǵŒ¥>ˆ1¹d²ËEÅmvFÆç¼-Ú>FMEÜ4-ÙlͶî‰ü¬Ã×ö¯‹,—ìEwŽ&=:‡-§äl©¶8‘09§è’ùkçǎɉ ,9 gÉ•¼ X/húÒ ™ÐRd™6ôÓ™f½õok’e"5©½e"áêòŽÔÞ[±¼È¥Kß6 2a$÷Lž±‡žGÆB;&î¯ìÉ€%T2‰Î†!ÿšYe#SøykÇ%9zNS„'TáœÖ¢IýØ®a$vH¸¹Tî7ú´è@Îð^ Mùé±kæ)ˆ`_H“aNФ¥Fr)µ¨uo‘ îUÏLzìøœ˜¼‹£¨TœÒ:&üèÒ´îh¯™o÷Zæ1BÊXF0/Gh¹m±$fî´öøDíqÌÀ_uÇ êó4ŽQå6_@3êæ¤’aDšÁÈ0ƒ®b¡5ŸõÞ%ì>ÿpÒКÖÜ|ŠQ¿'ÌÜŒò‚GCìïÖöêKÌ9Àª»Àî;Ï’>±ý¿L•ÚßLï¨H3®bM¦£ŸT¤dœ endstream endobj 91 0 obj << /Type /Page /Contents 92 0 R /Resources 90 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 86 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc5.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 93 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 94 0 R>> /ExtGState << >>/ColorSpace << /sRGB 95 0 R >>>> /Length 13310 /Filter /FlateDecode >> stream xœ­ÝK=GrðýÿSô’\¨ïÇV‚l`IÀ0-òЀ´­!,}gÄ9y»»â ‡‚¼r~ìêÛu«ò••™ß~÷–ßþøö/ßþîí_Þê~¯ãþs•÷´ÞJëïs¼ýéoÿíí~û¿ü×ÿô×oó÷ôžRzûüÏþæ¿|Ëï=½ýßoÿoéí|Ëo¿;ÿûã·l¼ýço½¾÷üVæxOåíço}¿¯zØÞk~ûéÛï¹Ö÷ºgyoã°¼çm\ïóœÌLï;®þž“3×Ãß[y+cÛÎ÷y>yL;óŸÎßoïé|ôù ÃŽÎç„ê<>²Üçû¸Ú —j'\FyßÝ}.ˆ}üù+v.¹ö÷ÑÞÊù»™[~ßÃüïµõ^Îç÷i—㸷÷~>¿Ÿ+YÌã|‹s!úý{çDË9ß^Þ‹Þ¹Ý>?½/?~ž+u>¿í÷â ;ñÒÎ÷³+—w±?TÚx/~ü^ï+ûKv|IÝ.Si;öq>_ä\VßópŸ:ç{¾Æ´¿_JÃOVŽkzoçó­\øï×a_¤T¿ÌÇçƒÒùüóŸ{vŸÏçŸË”üóÎçókyŸ~>çBîs=ê9å^ïõ|þ¹Ì~þV$Æùü2ñýËJv¡KïË_çzëQº]¦ã}®×ùüsÚ9¹Ïõ:Ÿ_Îý³ãk:æ\’߇ÝÏšÏõ:ç[’1ó¹^çóó¹vþõ|ð¹0vYüü*N´dÿó9±s¾¹Y±9>6ŸÏ?»ýóÎkÕ]“û\¯s=N±[výê)héœoÚ¸Þõ”ÛSæK:—Á®g§ œÏ?§ýx«‰çzœÛˆó[çÀs¾‰å§îf_¬¤‚ëÑÒ¹^öùÙþŒyZÁyù«ë©*';å¿P•ÑPüüÆ?þôöÃãÇÔ©ø)ëãýñåÇÏQ_^?'_?gy¾?¿üø9ÊÛëçäëç,÷ç—?Çýzýœ|ýüË×ýøö¯Ÿÿõ¼¸ÿø‹¸¸¿üãiFÿc9­îÿô¶ÞùßñgÏ~·¼õýñç·ïÒ÷o?þñÛßþèŸûk¿˜Ï ¶õõ›å7ÿæiz¬E|ýfû­¿i…ÎÚ²×oŽßü›§XX+ôúÍõ[³žï9ǧßÌ¿ùÕ^ÐÔ¼~õ7_£v¾©µ¿úºH( §Õo^^\ìæ¼<}ý±—èã§§ ;Uöþúøi>ÿ:-Ðý1ùéç§?<§øú9øñskÁO¼?'?ýœ}òýùǹ¾*×â¯>.5Òí­ë—½´¾Ë›ñ+¿{Z³–ù»ònüÊïæz®Tá/ËBûk¿|úÕÖøË²ÜþÊ/[os¢ü²,º¿öËÓ»/E×ësôõó·vº«svlT/‹E>Å}¿q_yF ÿïl‡ò £¬Û·N÷œéß÷ß¿Ëç?|÷‡ßÿé—·úÓÿúùí—ßÿü¿úÃ÷ÿðöãï~Ó·_7Yx”>ò‡ÿó§ïßþêÔÈïþõŸÿõûÓ&¼}÷ûŸ^Ÿ÷o Ióÿ¿´ c½!©‡?¯ôDk}~„¤ç ’ž ²¤O!iC tCÒb]Ù§ô¼õ)$='à!Ô Iû²Èó#$=]Qý’öŽø’V„¯´0½!iFHxCÒ©yÈrCR 1ó§´1º!©]ùù9$müü’„„¯4óónH𾆤u1ddHz"«ñ9$=_kåO!é©îoHšQüoHz. BF†¤å~CÒÙù½!éùX¡oHj­ûþ’æëCÒì‘×GHš»ýÚGHš«… !é =ä¿!éùš¢Ýô„€_BÒ"DdHšâßôüZ+ŸBÒSŒüûÝôôQe} IO$˜ÓGHj·ÁiÓä=PZ’fëxúGHjNé#$õÛš?BÒ¼“5J¯4ŸyˆÉ4ŸËâד!é'{Ìôò×(4¿j¯®Ÿ§ÊùS ëç©T~ÓY?ÏA~X?Ïö*ÃúiÑÕü¨Ÿ =é­Ÿ§¼£þ±~žòŒG:ÖOëx÷§úyʧ‡ø·~žGÔgÖÏSLGþT?˹þõSý<å³~®Ÿç6ûý½õÓ>ösý<,~ö·~žÓ@}gý<å±®ŸöHR>ÕÏSÞð÷X?Ï×ôGŒ[?OdÕê§úù2êg>ÅØyX?­ü¬ñQ?óî(ϬŸögüN±~æsÕû§úiåÇÿ>ëg^d¬ŸvÚéÓ#c^|„aýÌçû#8ëgž“õ3Ÿ‚ƒGBÔÏ<ï#)ê§]6¿ž¬ŸyfÜÖÏlã õ£~fT‹Wý̧yG÷‹ú™qY?³µçë£~æÓž§Oõ3[á\Ÿêç¹í¨¿¬Ÿˆ‘?êg¯¬_¬Ÿ§}ÇÏY?_fý첸õóeÖÏë[??Œúyý—žO”púnküÐÏ>ëmC샰òÅŽçsãl •ÄÈÕAk3Ø»¢wR‡«*z7nIÿCÝ ®bëRôÊiã@[ÒOcøà@¤ßˆîцbûLà”ô˜±Oï1 Xº¢XZk'èhyßéUçT‘¼½¢Xc¾#ºÝÑ$í¦œ6ÓbAoôÐAÐÔCëÏ+Ø’¢Wþá¼bw—ôÓðDpLЫU¤]ºÓFŒ¥èMÐiªJ•l`M’‰œŠÞY ÿHAoÚ†W Åî$y;xÉÒÁº$9žÜº¢w8ÄI¿¡ 8Ð;£Ã¾%;¸’¤ŸÆ¾…ð ñ¼oäøƒkƒ²6rK&pTE!?´h[°€{)z¨sºÈZ$¸–¤?Qyx!èAÂáêŠÍ¿ŠGh‚Ì[±np,ÉÎ&™A‹”"=°WÉ Ž¦ˆwï8";¸‹d"·¢‡Ö$dÉB®È‚ •¢ÿ!¾ˆ ô°Ð:|ÍBNŹÀ¢é—n¢Ã ôÎú°fÉN.É Îªxús“¬`íŠï°v’ôÓÀƒ`ù:ìAï¬O´›»¢“‡6 ,ØÀ¦™À½ýÍ‘…ã]²€++zWn\’véNüʯð…ýj¨8‘ôªi7tuþ@ï軘H?‰Î+лòŦ>Ò/ÝFñôøCÆ" ¸§¢wÖ›Mnd½úƒÖÆËÁ 6ÉÊ·®}(ú[ÊÃ9%ý£Ø\zw¼Ù\GVp6E¼ü-¨"ýâTŒwG&Eå ÉvC i7ew„jþÎòÐ;‚Èòíö“ \]ыٞ÷Uø“ô¦/ÐH㔬`kŠ>„i#&EÒOÃõ3¸§" á¾'ùd[Wôðò°oÉŽ!éƒ"‰CôÑeеkzTi ²$/uÒ•®[ƒ¶ÇýϸÓü>Á™Þ-zl„›9¡•U®ôjÒ^¨sòÈEÚç5ØhS×NtÙÒ¯y*z4i Ê[樓ó锿 ½?£Jgzi7\Ïi½¨t½žÒuÓcI³<.¼‚.´Õ$ãI>/ÝhTTNô®ÒÄLèó”3›ðºå‘ÿ•¶vDxaÐÖ[éN[[¢œé5¥ñR$g|áBÛ ¿0Ê[Fªœh‹…û¸žÒx‰Ã— Ê•^Iå-7o”;݆v¦Ç”Æ ¶^'mœÏ@{ï |Ú¸ž4WNtÒ| €P¹Ð} O læ¼|üQç³ý)F˜å=¡r§m$Qå±$Ìs®ôÒþ• ú+åDsÞÝÓ(%£¾z7i”×RPÞ…ýz•êc‡Ê8uQæKÕæÑ r¦Û–Fy=U»Ò«I£¼ö'Âýzkgºj£<–Žúþô`y,O¸]oíD{}‹^ƒ¶g-eœïò—œÂ,¯x)«\¯§4Ë«7¼Ò÷¥z®Ú™nY/ÝðöYåµ&ôÇÂízJûc½y íBï*òX3æµ ×ë)òXÙ_ wºwíL"îùì©ïã³'¢ùâÈ&¥¤Ý®§v¢}ÎF4Êk­ˆ7„ ½§ôÄù°¿®t¯Ò(¯Ç#iwzMíûº'i”×c{É,ÌòÚ/ ãz´ÂézK³<û}‚q=}ærY´Çû•ÞEšåyÝó îtKÚ¸žû5¯üa”g›ý2´+Ý–pcyníð‡^š6&ÄÓÑ(ÏÍG¥ íñB4ʳM(ÏÒ(¯­"^öëÙØ_ ûõlýy4Êë±½•Fymìï„q½|½tºÞÒ(¯ý0®×zÿ0Êk[xž®´××h”ׯiÂ^]šåq£< Wz,ázË#&9)ûõìÉç©+'Úëc4Ê£ykcÒLÆùF£ ßIIü>O£<Ï*òØ;ú[áNç¦éV¤Q^mÒPÕ.ôœÂ\'böx\×s`p¢Ç”fyè…q>óË|¶OF¼kó*†v¥û”F{{<ªv§½?Æõ^xžÆóÚñlÒ,¯˜d¦Üèݵq=÷ý¾O³¼n´‡Â™žYñ°-èÚÚX§…I§Â(Ï#áyF¸ÓkkcÒmÆxA4'f´Og–×7äÊöï#ì×Ónl‘Æx>éLÏ-òh {•Fy˜¦œ®§4ÆÐqJڟעQG»ß7¸Ñ«JcF×I—ª}î-ñ²íó[¥«t.tÕÆx×öÕ}ÑÓónø>Âöï#œéÑ¥1^°ã|±hKã]³•Ý‹4žÇŽ-^SÆ÷™ˆ7„óõ–ÆxØÆ|Ja¥G¦}ÅqtŸôjÚ8ßz¯×ÓÈ“”Ù_ã|ÙÞ 'º%i2{}F{f«Ý‡tÆù Ä£Â8ÌwNwQq[Úî];Ók ×[ÞÚ áB{¼¼Syû|ý©%øKÚízJ#©D­/î´w gzLé>èµ´q?±Þ@å±b½r¥û”®›öxZ¸]/iÄs–m¤jgz6iÄsȦ#}“èÔ)¤6« '&Á©ç+Üi/ÆùN<¯G#ÉHe.Œóe牊ù¢Ê•]šåqc¼]¸Ñ¹j'ÚŸG¢ñ«r¡y=žF¼g“ÆxL¤mÏCʉ®[x²¿Dö?eô—x0–.tÛÒè/;ÖK*ߤz6*Œþ²c>®r£m>Šr¢-ÞFi“]×c"©[4“ŽÍ{¾ÁåzK×J¯,ö̼´ñ}ÚSa|_ÌÇΙöö"ã½ñ†r¡glï01AÚ¿ï@<"ŒçÕõ,ʶçQaæ Oh/…;½«¶_/LlRF{h§’v¦W•F{h÷¦v¥ë’ftÌçUnô¨ÒhïFE{/œè=¥ÑÞ $TÎtÝÒxž˜ï«\èݤÑÆSOß$›6_nk7ÚÛóhÄsó•3½š4âµÁx+zÌë%Ýñ}Ú³è¶iæíÆùNÏWNôhÒè8VÆx›yiziŒ·…xS¸ÒµJ¿’®Ž¥Ýho„ñ}·¿ f8÷–F86Úaÿ¾“ñX4úC䕾ÇÛ|SaŸOå>Šv¥m¼KÏ“ñ˜p£½=NtÕF¼?ó½Oã}©å¶ÙÒ(O³ =®4¯ÇÓ(OÈË,ïSoF£¬Šòø4÷‰q/m¿^ óA…ÑÞ,ö—Â÷xï/£QžÖË(ÚûÃh”·…ì‹Ê¸õ!ýÙÂzaÄ+‹ýp¢‡6â™ÅþN8Ó£J#žYXÏ¢ŒãâýhÄ3k!Þ‰F¼²êÛÓ\aöx_×óU…1ŸÂ2{UíF¥í×sg$ufÒòìù^”ïñ6^+Œxyc=‰²_Ï]P¢1¿ ú;a/¯»Þóyã¡ëI”q<Ö‹cÞöj/Œ· ßã}Ÿªèq_M»Ò¯G£½ÚÈ.¬lÇ×”/F#©vÂzåFÏ,]a¯Ê>žîûÞtíD¯"q<6¿öñ€š*Æ„3=Jtåz³Å£ÊÅùŒÂØô ùÄ'éB÷.=+mñ °÷g–Ô=5íJmnÚ€ùŒÂ¼ß˜Ï¨|÷M-¢ë=~5mõ–ʉ¶ú,Ìò‚lÃʉæß{:ßã­¾(ûñÙÏ)£Á÷x~Ÿ§Y^¼ãRF{Q0WõÙe´e£½¾Ç{{ïñ­hûñ5!žŒF{RÚ;á{üH¯6¯?ã{¼½ïþØ”¨j7º'éqïSû?µ;ŽG–oåFÛx‹p»ÇÛûVå{|¯ÒˆWj¹×3Øã ›O¨]`ì=$œíñœ°ÇcY¿…±©“Ü©}÷xïinböúö¢b¾±r¦[–ž÷xׄïñKñJÅ|ee”WìÄ Œx¥b=0ï÷D¼ç) ìª6Žg¼xµb?åJóz<ïñ«jÚãÅh–ÆSÂ…æõøê»/uÅ|gåL{{½îñÞÞDÏNÛx‘²ßφõBÂh/l`­H£½hȇ®|oCÏ;-ß}ȃ+íÏ‹Ñ>ŸÍ=¥ \ÐßgÚæ #ÞhXo¤œè™¤ÑŸ4ß 'šÛd˜m>Œ2®vÊ^¸^Ü´S×Ë'&(ãy¹1Þ.ôÖF¼ÚÆÝ7þi´m¢¾E£¿ÀÄ&i|_äKFÑ0Xç‹];„Q:æÓ*gzViÄý}4ž‡m¾ÝÒ®t+ÂÌ·cöú(ìßÇ–Li´Xø"è­x£7ô÷ѯ³…gC»Ðþ<çÙ>îñOã~ÛÂã&ÍM±G{`#ÞìØäSãH,¡Œö¾c¯ åD{öpÙ¬ÏëQ„QŸ‡O<—öò8°^SíýÀzaܯQ° ¯p¡­< £>Ìׯóâ@{(ìóÕmÍ6´+mïã„Ñ#q¦2ê›å+Úõz ßñL$žUÆxÓÀz e|ž/ÜVF¼†ÄÎʈçÖ{(zi´§ë5” móe„ñ<70ßQñ6NPF}±|`[ÏãùºƒïxàÀ~Ê~¾ë)„}>™Ùâ-aôwû')'zi\ïé £”Q°/•rÃß+O #ÆÞcÊþ>ÓÌïó4êÃÄ~Âé~žÅƒÊÞžL¬×æ|RsÑæõÆ|Ha^oŸx£ŒòŽ$¥q>õ9ï§œ(¿6éÝKíÕÄ|?aÄ s =ŠÎ8_¿°Ê¼žÈG|Ç›&Ö+£½_Þ1)ÏkßÔ:×ka?áצâ6_EãÉ óý…1Þ³*Ê[4®Çb{þt!ßt0ó˜½?‰Fy[˜o.Œxv ”‡hÔÇ…ùäÂxݾ1•²ÏG©ùŒ…lì|Ÿ÷wE{ç-›/Ѥy>ØïE˜›Âc½L0ŸZF>Áë¾e¹éôoÝ\Ú_"ì&ýàäÁ¾«d`çÁÜMúAï¾>vu~Ð+§­MØŠ™¯ɉ¥wÏç@›,½OU¼ûEï©è!kŸw¯Ú=àéÜë2{//ì7û “i÷…ÝÿýUzÇL‹HïÜÍ2Їá:²xFú ÊÝÕ9ÐC,KE9"™ y ¿I$vHFvÍH¯ý*ôÐd óÔ“œø5ð\‰Ý¤+ö <Øwg ìÓkb+Aøä®Î‘Þôa 5Ap´_JŠþØ1ÓýCú¤¯‰1ÿ'BÒ‰‘>áob>C$vÁÍÈ6XIÏ5è…ab&C¤ìJÈ×èÜ?9Ò_BL䨊Ä~ÑÈð‰t±ÛpdåÁÜ}ëAì®<îÞWzQ™™_(*sâ4½¨L¼µŽô¢²r:zQYÈ(YxpŠ^TN4Uz$ËØ\æ¤èEe•»“ʃ^TòEú0ëÂ[ÚÈʃ¹Gʃ…·©˜xðî‘ÜÞdáíf¤¤…Üt‘^w^ ô‚´¼ÈzAZëîªò ¤…w’OrÏöŒ@‘^T¸7o$vXÅÛ²ÈB®¬è…aãMÖ“…a×›çÿÁ»­g¹ô°ö¨ ôhc-»³bþKd"o¶ûÏÌxøã~³‘~lìˉl¹ï^`ãÁÞMb—c¼‰ä–™¬±ÑH™XgŸNwËT¼òî/oinš±¥Mt¹Ç÷-Í-gËMÉþÅýÓ–°L‰ü4RN&¦<‰FJÞÔoJã§‘’íD.RÆ–ÙûŒ¾[R 9 #e3·lFÊûšò?)‘íÝ¢6úƒÚnÊ»¯fÊ7ó¬ÒÜ2õ5š[âaK=a´÷uÞã¿ún)Ã-õ„¹å!–L £½¯X,Œö¾nÔçh¤„ÆÚeô )Q„_[î-e¾Ò´-ó˜Òèi¤¤oX²$Ì-é°%­0R¾ÛÚË"-3Z½)žÎ÷ø5¥‘ò¿aIÓÓ–²hÐ7¥ÑèÈ] ŒúÈ-턹%¶LFJyKiS¥±%rF/¦Ìæ–gÂØR¹•Q߸%š0Ræw.éŒÎ÷x¦„y)é;S&<}·l¶†J#å6·ìF¼×¹¤>º\3åÉӨϷ&á»å/·|F}L‰úlOIš[.qIvt½Ç.Í-…¸äúi¾²ÈxúUF}µ'ë"Í-¥ÆMñ4úSì=¬ŒþtpI]4·ü–IÁ)ã—ÔEcKÁ%×ÑÜ2k#%M4RöL‰®÷øÙ¤±¥ÔÄ+alÙ`)ªp½[Þ`KEa¤¤Ÿ)o¢oÏŽ”OÑÜòK„ë=~Oi´§£ÂÜò‡KŠŸ¾[úqKaÄÛcÂh/¹%Œ0·äHw‰ñÓh/WF{ûtf¼²Ê]ü4â•…%GÂH¹Ï-U„±%ÃÂøp¾Çï"ç™…70Á‰[n¬Ž%ûÑØ²há-Œ0žgÖBÊ«h<Ï,lI&Ìë‰)ŸOÛLÇNsIßÓÜòfß%‡OózsÉK4·Ôà’•hôW;ß%uOãzsË aniQ0%üé»å·ÌÆõÞ¾7¬2·tà’Šh@˜)û9e9)¸å€0·dÀ–Á¯”÷S2£‘Ò4sJq4Ròå†òÍ- Ê{4Râå†%µÑH–¹¤êé›R=wLÙŒFyÊKÆ¢Qž¸å0RJpËa¤ÉíY4R’flYÌ)Ùæ=¤¹¥… *#åA^X2”‡¶¤³Jsˇ…)¤ÑÜÒÁV•Qò¾ßç«oJˆ‚”ï-8¥øi¤¬*lO£‘²ƒ[sËlé#Ì”þýíÃ…/‰ÍÏ #åe)˜òÍ”÷hï„ÑÞŒ× #¥rñНüÚ²`eáu·èwJèÓ¨Ïow„QŸ1·J™)ù}®Œ2êsᔸh¤¬Äܶhöç>w.Ili¥‘òÃ.\“FÊ«²ï”ǧ‘2±øÀ‰2R¶`nvôM™Y‘RW)˜R_˜)ï1ž+üJIŸ»4RnÕöšrø0RncmbôM¹X}o!en €÷9Âh*ÚCa¤œ«œRÍ”øXò!Œöˆ)á…‘Ò¹®;¥ñ«9^í¹#ª4Rb1e»0· ØwŠÝÓH‰ÖÒý¼§™ÒšSz£™Ò<ûÛÈ`.éóÜSMí©=.i¦ ç”ÞèzS®ï&”Y6pP¤‘R­ù¼´è»¥Ic{Í”éxƒ,Œx©aI†0¶\ÀÀ™2úƒ†÷IÂH™Ûð¾(ø¦ä²Å,þ ù‹Aå~?ϧÌF#ÞÃÀ¨2Ræ5.ùˆN7e··gOsÉ Þ•™ò)¹…Ñ_áʼn2S7LéŽN7¥´O©}8ï›ÒºûûJažïðñ2a¤têØ²B˜ç‹÷EÂ<_¤\fJo¼O ^÷|—?¯ó|±dA˜ç㹇”y>ÛãYa¦ÔMÞß”á[:3¥1ÞW ã|0kBçcó5¦tz¥LNƒ[60¥²ð¼)•­< ¿R2[¼+Œú2°åCðÝÒpøÚuåyS¯$öˆ)…oŽ~ÿÞWßçÅá{#*£½H¹(Œ”áÃ×&(£?aÊßàÊçÙ÷ Âó¦Üõú͔מûKÏLy+Ì”¯H)+Œþ|zî×hN^ò¹ÍYñÈôe¦Äí(/ÑO˜÷3Ï{Ó×ÖDß”)ã…2e«0SÂâ}ð+¥«õ§ÂØÒe.Üh/ÏM Œxn!%»0ž—Ú“`¦ ³”¬Öß c¼‹)S…¿¤DfÊ\Ÿ•Í-³ÍV^„±å×ñÞÒLùé{)cË,ÛòzK—›’ÔïG4îçö‰ Ñ·¿°y‘[z¾R‚&i´§L *Œxž)5…Ñ^1¥¥0Srnï î7e&Þç £¾o¬FJ„ú$Œ%•LY)Œ” õ-¸Ý”ŽHY)Ì”ŒH9)Ü_))§4R.$¤dFJ ÛR¹KcÉSb}}ºÞ”’xß/Ì”H)%Œ%}©¢¨øÊXr—ñ¾PK­áÒX²•‘H˜)Õð¾.83åLf}ˆööÀRòÙx‹0–”1%Ÿ0SÊ!å‹0R&dlI)Œ”()U‚—L¤ FJ’âÃÊHP°e¼0SNa½Žpº¶þðaßÉö åqSЧܮ{—FyDÊ1e,Ù+>q#z1ekñyöÊãúÏóÑë¦üñ÷5Ê(/æ%ö¨{îårSæô%|Sr ¥Ž2êS_^Ÿ…‘¹û||e¤„éÛ Ž2îÇð½_£oJÖáƒ2R6O¦Œú<*2“ö.ãIîV1Ú]iú ¿(ë= « ä$š1°Æ(Ð1ÇÄ ÖÀJúúÂÀ›q£ÈñÊ©Qºb#ÇPô…+órŸäcéØwÕヴbé!§å)‘ ´3Ýõ†VrwEoT™ãIfKži‹läΑœÆˆL2‚ÞäNd ˆÌÌÖQF$'0Ì‚õÁÙ:RQôÁ™‰¹ÃO²ZÍŠ•ƒM Ú²4²ME¯8õ“ CÊÌ!YÈž#Y5fÇj´À›Ä· DþdqмF|Õƒ|=ÇÝhøAI2¹&!ЫÆÜwÅÔòÂÜØ‚;°“½(rŒÈ†ðcbfs¤OœcÞ“ÈDŽÉ—Ê+ßm/D†dDˆL¤¯wz°0K²¢Dz°Í¬(O22óžDúƒþªwÍr§µ…U‘>ˆ»ú]ó ?‚.Ìp~°2ëêH±˜?å=y’ùƒÓz³i+W$ éóndæÌ…9rv»#+ÉMuôFuç{ð4ª;ß Yô p#f¤‡µ»ÝÍT¾AàÆ<ÛÈv¹½ Ù`h‹l(H›Iø=ì±è¦XH_½ð 'mû»¤èÅÌÞ,NEïè÷¾©ùLäš‘|趤')I·—§tþœ4%ø.:Oh/„ûµçMFR‡ã­Ì}ÝS‹FSÆ¢èzstÙ\ô™^y“ŸHÚ‚¤,Âõšy‡ŸN7iËÍCüů¤'X"ŒEë ‹B„±¨5Õ›·ø«™×¼²4•¦†¼¥O›t¥aÒ|4ͧIÎcѦ½½Wî\ôž&}F#iDš˜tútã¢Õ„EÂL ƒ¼ñÂù&©M¸Þ¤.X4!ŒEù™“Š£±h4gäÍ~º0éEf^àh”2[¯)\oRëÍ…_IkJÎ7) óòFcÑen(OO'.jÍ ‹6¢±h5sQD4“î }~:ߤ Ùo¼ò+iNQ^÷zbR³0¯'EcQ-“êózbß•àWÒŸ…IQÑXt\÷]˜ISÒkÒæs_*KJcq¢pyK—›´Æ'å=ÝnRäÆõ*õNz|׫pÒp4®0GóÑÌlÏf¼^h/„_IgFfØ™ 'ÝE£¼,:FyÅê¤èÌûQ&ÍFã~0)0îGEÞà`¾4˶ÁI“Æý¨žðMí…­¾Ò¸Ljótâð¤%µÉ[×Û¢Ã"¼nR$n7éMÒ¸ÞußÏûê›”«ùÛ?e&UARa$5j¨ÁÌ‹nžÚã&µiÚ¯¤3#éSó‘“hæ‡4û¤¤hÄ#Íß^E·›Tgx¼!ÜnÒ™\¤s£Ç®ÊmaRS4âÁ㙥6ÔÇ`¾TÏ õM¸#éKöÑ a&mA^û`Nª2sÓÓHzƤ3µ_/éWÒšš…“JuLJöI¹|^4íwL:Ø'5!io@ªÌ¤=“f¢óMz“·0'%åîyÇ•ÑÞõI5Ñhï†OJf†[Kª“´ÑÞ‚I-ÑõSReÄ«ÃßJEóU‡yWi&uñŠ¥Œ¤-£cR×ÓŸ’æä&ýJŠÓ¶4ê£%q®ÂõIu”™4hbÒD4êãðèÊú8&eE#©ÒðIëÊeÒk sѹ%ÅÉCI%ìí^“F<2 ÊK4ú»YP^Ÿ¾ñòl˜´xdzòne&íiwRÍWsÒ¾%ÍñIUÑxœœ4çÁ91iïáµy¿§¯þUÆýž>椌û=}va4÷ÍaReÜï‰ú.Œû½&-<}“Ž éŽ2“º =fÒOÊ¥Œú¾¼`Dßú¾ Î'ºÞ¤?6éOãÇ>iæé[ß‘ôG9ߤ?} ó%¾yti$å\“.¢‘tm¡½ ®LJ³ü½¸2“þ LÚ‰fR¥‰I!Oßça$ýQF<µ&]E×›ôÇËóÓLúÀÙ[ÊHª†¤?ÊHZº}›eôçÛßÑFߤI;aÒP4’"í„ïxI…‚í’aÒ p{%ý©Òè/¶¿­ˆæ¾WÞ1&i”§íûGsR°uì6iXåeûìe,jL˜´<˜t!yâxe$UHžýV¹Ü¤4Ö^s_i&¥QfÒ”7a$Hß7‹R’¿%Šn\”dãUEºÞ¤6µKcÑ!VÇEóyÛ\Št½ImlÒ©pºIk¬½fÒT>ø+cQköñHe&Õé>i.8ã%Œ¯íÒXô”ý͇2’*!éKô]T—}ßle&ÝÁ¤Yá|“¶øýxئÿ ©&­ 3i&• cQYñ—èÅE“N…ÛMb×[8ߤ%¥ O&¥(^ð•ûMjb“V…3’¶xR’hî;æÙ;Š4’BÔga, ,¾º;šãiž}G‹Ò4D¹Ü¤"Oó“g‡ÊÒXÔˆ æ•‘4I=¢9©ÓÞÎùõŠFÒƒêÊX”mû.+óy›Ùï”±è»ùƒ«2Úëæ›(Dg.Šo¾úKI&µ cÑ¢Í=hÂL:ìs"´±(¶û¾—Êø¼î[D÷WÒ_½­ü9)Côâõì˜Ô*ÜoÒ;Ÿ`&½5[{!ŒòÔ}õ»r¹IÚ\”ޤÊXtÞ=ûœ2êfBEwÖçîÊhl÷e>Ïú ¯*E»vá“4wLª®Lj†¤Êh‡'QF{¤ ÑœÄǤ Êí&M(Y:ݤ ¶¨"øö‡HZ Œ¤'ÃÇ[•ÑŸ_=nÒÏ.©ŒE»cã~FcQöô ˆƒÛ¾I|ö¸2â)&Ñ‹I¡)B[{.\nÒ‚¼„9ÝÂæiÛxŒ0êëôìÏʨoøVa.Z³¤6i_IçlK¤*þÚöÜÂ7)¤=xUiÄ3ËgÉ)#[˜ôÜî¢~Œ #^]þb!š“kmѺÍEïF{»1̤¾(}J£½Ø å-IÕ,)j¾I‰6ÆÓ„ñ}·g¯‰æô%[tnã-ÂØwÓ®{ôMŠËEÛÊoÛ¢jk/…±¯)^Û9ú.ÄÌÊX…ì†Ê¯EË=]˜ºVŸ£«\°èµz|÷ý¶_[ûæ"ûpôÝך‹^…±o4½ß}›ëÆ÷‰æ¢RŒ' £>7Ÿ/Í ÝÙé•Q^šW¬è†ç»èT8ÝE¥þ}ž¾ûâvo•ó]”é÷ãérÇó®0öÍî¾ûAôMšÀE‹Â\4è›ÑéîîÙ¥ƒ3¶ìï_þáÛß}û …  endstream endobj 93 0 obj << /CreationDate (D:20150701163320) /ModDate (D:20150701163320) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 94 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 96 0 R >> endobj 95 0 obj [/ICCBased 97 0 R] endobj 96 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 97 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 90 0 obj << /Font << /F8 7 0 R /F54 25 0 R >> /XObject << /Im6 86 0 R >> /ProcSet [ /PDF /Text ] >> endobj 101 0 obj << /Length 3317 /Filter /FlateDecode >> stream xÚ•]Û¸ñ=¿b±= r»v$‘’¥CÓ¢ .-ŠK_n¢Hò •¹¶[òIöî¦(î·—óER»wy°E‘ÃáÌp8_Ôßn_½~§õU’¬Ê,K¯nï¯ÊõjWy¶^¥*»ºÝ\}ˆ²Õ"‰ÒÅ2IR½oK•Ds°ûf»øtûÏ×rUæi(â«eR¬T‘ÐìÛ4‹¥Îâ¨=ÛöÎØ¿Ž:ªÓ _’¨­êEºŽvÔ= SC»ÙÀröwÞã«ý«4èú3+iY†vOMj¡iáìÓ.X àÑeÄ´!dÇHú/~6À^Ì“müÕHÖÆ0wý ä„¢AY¢i/Ûˆ´ˆAÚe#¼?\É.p¦ (®øsÐ æÆv¤ 1p=°d yæ…„JH¾z¼Ì ¼~ŒÝ‘¼‘‚qÃ,d–@ŽmïÉjï´®ifñµ ­•¦„pÎé„£ÑàB3²¼[¤­Tt„¹H¢JqÛP)*¬šJÖDáðÖ$zUÚ£0Àù£€ß³˜,.œPuÌŸEöâ$ž@tÚ@‘éX–0ôÄ6B P‹ í@'óhó=è£zqÍN˜Ø£¨mXààm BzòÈXµä°¢ó™éà€ZN‹Ôrj{RæôÏ‹e–*{~ [×Áž Ýé2ËW:/‡ç=Ûït†ÍŠd}`ůÐAZî_Ä`$òÞ3ú t<Üqƒ’"Ë÷›ÀNmAô†ñ[»ÏÚó äôVJi¡P2e'Mtz̆ØÎЙw=w(ù#Y略rÇòç$˜GØÌàLðÉ®zÔ{ÜÕ0zH öîÕ™_‹`-H5¼M74Š, «B[ü3,‚FܰƒÅ±ºæpÁ èìqKíÀi«! LÍGˆ§ Ðí³Ú2)U³¡FÏ&¡˜³Þm éb–“ó>RA1¨Œä ;™ip~ØWõA S¤rG¸Ì”X ÚYâA$P!ùv6Rj;Žm'öõàLy{$§§§Œ5¨Ð( œl{³)Š’5µ—;Õ±€Ó‘ççM"2kTE«;3òª$[+䇦ДöD‰J‚`Ô@?qJM:Ä×{Š ¼†h´ï0rh%U;nÏnŠQìŤ¦ˆF„ŒO!ôÁQûñ-µ·(ƒË‰Ob–ëèVB 5{g‰ßgÄDåNLé‘ñGÙdHt1ˆ[ὂi§QÒ  éër x2ÄMšè­îúhš @dõ™ÞphdØhºËZ€H꘯Ù<ÒWð ~ÏŒ8CÞ—Yáá²ûluú³k<6:æc\¯Ð?âË×ëQÕõ C¸9OÔÃÝ1 Å`ÝEOo²Þ¨n)[3"Ù»±¦è/03-jüî|Œð"¹(Ýœ\jÒl`-wFl™çiô$êQdð Ž•Œ å×É¢tüÝqÃÞ`ÀQ7€ï›r-ïÀàç‹XŒ`Ù[ïûÉÞ"µï¬'ûWq²J²ÒrÜÖ‹,Ú[ÄI° ’U‘8`©S,ú,'ò,;*Ü"$n è(°6¡X•ëbè-1¦n¶|Âa µÖ¼?7l{$\ë¡AÑ×A8)xT8{™Ú£ó1QZR`N¤¬Ž_Ó zl> ÈÒÙ›:¶M{ž±i:šâdÒå0>³›PGÀu‹¹{w~ÑcÊäœëX4ûh¸| {Ñ7ÔäKO ÈJŽ:K—úE¥Éã(hÔbe«Ž: ÄZ5ŽG ÷Ñã $×KÊ^жÌhù5 ×^ŠñÍÉÔIÜ zå6M DiÛù•¹Ø“;gè6GãLÛ\¼¡ÅðÀ&ilSÎüe'‹‡´ Uƒ=ÍK.Dy'° "yx¢þÛç;T|âb’þ¯½ø,(®ãàÐÇÑ[d7wç§çŸ¨±†}²õ·¼ŸD‹x$‰ :¡Ïw¦ c)Ûø;L­Ž ¦»<£u@ÂHXgAÈ“±žudlúðƒSâ†Ñ[оêqÖ&#»Åk$Ža%†ÀÂÀDÞQP•D!”!¤>âÊ|ÄE«W§Ýûaï„1úÌ8¡Äm&ò1©J) “8V¥’é`ÐÒsËÕÔø va³2{W©…aŠ]q¦ Ü,bÃíVÊçy'”Ÿ ÿ aÅô ÿ‹n{Ãõ9“OªqŠP¿C9Ú÷m×Jéç… à¿…Æ]È9dn¨MŒ: ëåÃüÞ•”Bí Åa¬ ˆF)-pÀÕ3íÔM+Û¶ååy²RŽAÀN½¨³ø+Íñ šOrUè‚â¢'º¶ “e™θ ù“(0¼ —ÓX:ÕzKS|ÖÓ4—k¹%àSª×=„¯ˆÍÜá­DX-m($ÊCÐ_P:Œj/eúŠÖÜûå¥ô! Ì”>tFóô3¹Âsqü$ï¶Ò–¤Bœ6g§À";{$5åü™8È¢ËÙ£dÛŽtŠRµ¤‹\}F²…8‰;«ç™]€³øy95„à\} ØíØí¹,ÇÛ‚¹©g°˜8ñ´üß:‹7 W ¶; 5\‰EŠ^³2—^B¿3öqèö:‘ŸDmÆíd Ï„áfãp÷\’´_Äÿ–ÈʘR»´×…ΨõB.Ü @ÓUNl;¼DIKÎåj)NÂÚžíLó^Z*™†ˆ]£èw‡än?xuTB; mê’°øUR Ú¸~¦úgÁ³8¨åq¸¦ô3Š ‰>žIMgÒ+ ç%äÜpY0=%”L¬‘9/aßO˜ºº#(ácÅà$úŠÑ¶Í‚µ¯}' z·÷ ½+÷c:ìæPxç˜Ñ"·‡—•Ž•«Ps'S[. /ät`a3õG»3„@’phãnºJ©ŠÊ‚žXGØKAÕ&»5¯}•“B<î]Oûf°$y­¾¡¸«™…/§[C.<äAc#[5Œ¢”†ª3ePBn%ßZôAÜÿ–pî¾9’’ò4Äd_»*˜3ܺ ©5’wQüq㫽=ÒA=®¥E$`ÜwÚ³Þö7³¥ø^Lû¶ Ãh¥Xl6¨m¤ã?ïn›C™)JèØƒWqc˜jHÖ¦ÔÔ-LoåñzϦ=)·µçàBqþ¦ø1¸Þ ïÝÇ÷Ó7Sœ’zÈ×î¦ÝAûkvw­û+7Ì—¢1½_¬®xC.Á%pçTßQ´ àÖ0›ÊÉQûÐþÄ¿ ùDêCṋvÅ"ïF$N䊳ßÌ Š¨”oŸÜµwbÓ´B—#›:üDâ¿ËgÔ×|ë~Íw÷ð|ÞIQáíÿ·5Ò:øð¢aSš^Àì?°˜e÷ÙÝÃ;‘ÑÎAÉÿeˆ|áð=ÓüÌþÌ~ âÎ {@@Ë$ÖëèwÁ±•eÍð{‹“¤®ÒÏIôÝì7/cYÜóŒÚÕõh3ø‚H”gÒHvüý÷Klð1N¢âU– ?±0¬“=/'Öìa4ö‚Müîå‚ iÉ'Ÿôüú·Sã/lLðøË3ÄÜÌ7ψ2Ø”G^hϘwþ“–7CM åÍL½HW%8¥V:ÕÄ’à«n_ýZ0íð endstream endobj 100 0 obj << /Type /Page /Contents 101 0 R /Resources 99 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 99 0 obj << /Font << /F44 6 0 R /F8 7 0 R /F54 25 0 R /F51 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 104 0 obj << /Length 391 /Filter /FlateDecode >> stream xÚuRMOÃ0 ½÷Wä˜H4kš¤i9¨7DoˆCµ~lÐh×ÁÏÇN2èP”Ö~¶Ÿ—Á$3Ü$)‰M̵d×Ï’X*›°2mHø2lî{C¶cðë =e¸â¼)‚Mž’ éR4?muÄà/*òLs&èv»Lð­Y(MÍ5 u”ÑÛ°éÄâ"‚ÎMhæ8¹Ì#Øûze[´èŒË;–^d$-+p^K&èY+—Üœ)ËÖ——ƒajýå쓟·t³øßäÑùÍeõ#%(6«;øp8t‹óà˜¸kÏŽñÀÀÌ^Š3‚g:qÍ{,bÀö ùRºcz­ÀÝiµs2ÍEìé{ç÷%z-Î8”bñZ äì‘s@’£›X)EŸ£ó§«þïlSí '$‘¢ÇËè~Á á æR8c“kµzBÒnŸ¦Õ'i<vSöblÕúÝIÁe,ÏEܪÆ‘æ&…+TÆ¥.( Fƒ»"ø ³¼4 endstream endobj 103 0 obj << /Type /Page /Contents 104 0 R /Resources 102 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 98 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc6.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 105 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 106 0 R>> /ExtGState << >>/ColorSpace << /sRGB 107 0 R >>>> /Length 23933 /Filter /FlateDecode >> stream xœ¤ÝM.KVàùù5¤ñý1Ù–lɾ-Yb€øŒhlwËð÷ýÆ^kgÕ{æ"Œy8yϩʌŒŒŒŒX;üåGþøÇÿó㿾þߟýá¿ý§?ÿø‹_~¤Ï”ÒÇ÷ÿùË_ü—×ÿuöýñWý‘>þîGþøË×ÿûÇùðñŸÔýYÇþçï>Z’ÿôñËõÇTûìøÓ]?ÓןRÏŸæ6Î?íì|þ¼¤ôÙ¾þÜùõç½}æoN>^Óü\íë‡#¿þ|äÏþíÏÉçÏß~Ù¯ßýùó?ÿ-OíßþAœÚ?üí?ÿø³ÿX^×á·ÿð±>ùÇÿ0Öù™ÛG)Ÿi}üöw’~óñÛüñ~kïûwú\õÛX~í˜Ûëgÿþ_¶_ý_îu~÷¯ÿrüÚÿ²ôü¹Ë·ÿrýÚÿ²¦úY÷·ÿ2ÿê3TûüLoÿé¯>GíÕðçÛúœ$4„¶>ÛW[/s|¦òíVxþØš ôõ§¯[`ÏçO¡¯?ͯÿ¯­çÉo¾úùŸ?¿þ¼ÔôÙóóçä·?Çûüù×Ïþý¬|‹?ý:AÅîV>wBcý”ãü·ëu*2ÿ[y5þÈ›k=ÿ±l´ì?ž¯ÓØøËvûGþãRÚçîüeÓýcÿñLçò[+ú:_o]æVòù騣:ËëÒæ×¥ñ ÷N?øëÞÿ?»¡¼_ÝîëÖþ\ÖÞÿêOþÇo^Eúø“¿ÿ›ßÿáã~ÿ¿~÷ñËßüîÿÓßÿæ¯?~û—¿ê·_ŸãuÚ›ýì¯üåÿþþ7Z_é¿üÏùÍ«Oøø“¿ù§çï{ýß±ºìæ+­Îññû¿ÿøïÿüo=Ä^]wú·bÞü×|šÿÃþ¹øºâŠù\ tþ#Á•ÀÓeEζ¡8*8%ûO§/x~…–Oûl¬M2ƒs+²AÚµL`/НK nżÁ•%íäÔϼÓklÖŸ¶$i'§^QpÙ¯ßqÉgGV \U±Û/8pÚ#íày7‚m}KVp6Å:Áµ%í×_~ê.–¶)™ÁóhÌçäôô™§dgWL ä©{ãë)°Á×99?òœœ^NSœìE²€vbÇ99ý5¼­’ÌY±ÛÉyÂ%i'§Ûh&²5°.źÁ‘$yð–,‹œŠÙNÎø,š¬M1 °mÉ ŽYNGl\]ÒNÎü,EquÐ:™Ài'g¡£ ´gz¤5¤}~Aë®_,Y²€³(Zw}¸%3h¯@ë®Çë”%38»¢u׃·Uä9±¯—¥Ô­»¼éÓç’äÁ¼ o|½FTpWÅuNÎxý8[òœØQϯ"h7û‹µHf°7E»ÙG;·¦bGW´›ýÅY$í`Þìv³¿Ø«¢Ýì/Î.ió– ´›ýE»(‘ÜEѺ‚1ÏiL¤Ýt‘vbºÍ‹ ƒ±{“äÁ£(ÚÍ>^¯š ´KhCµÁ›.pœ{n¾&ÙÀY­+8Ýö”¬à^Š6 8CÅ©húÉŽ1ò4þ‰Ù‡H{п¸–ä9±s‡¶ =Êçò_ábêàίjòóª¶%8³¢=Ê'.Jä<'g¥Ó ÇÏ0@°€g0Ùí`t}‚|BÁ3/p8“ä9u ÝWdíàî’v0:¨ÈÒÀ×CLІâ«Ù£\°µ+ÚP|¡ VpíÈušµqI;í'6кúÏÀUО§³œ<² ]²e¯x‘ÖÕ¿˜§dkQ´®þpKòà9­«_Û©‘ÖÕ/tn‚Á ¶®˜yp/’<]}¤=v¶®^°€sFN<¶ 4Ï%Û%p ð Šx^Ò#íp󯉴ƒÛéäíAð¢]²@{=|±$Eëê_<ÃA»(î‘u‘S²€»(f;9Ó&R3Ø—bâÁ+IÚÉY~QÞ80È?ÓÏM’Û¸h—,ÐùÛ:9ÅÖ¦8xð Vp E{ÐçÄgŒp¥íÞ‰¶×€ã1¤+ÌûG8Ó-IÛ#ß<¥3\ì G9Ñö\ˆ¶N¼„‹pÇÃÿØ®EôÚæno~ʶ§KôÄõ8_¸¿oôXôyÇS.î%Ý'ݪ6Žg;Žf{Ø~¾‚3]†tô\Òö:˜Ïw’ªè•¥3ŽÏ6j¶×¾ãÕ„††Çg,¬ŒãÙ£mZ(çjoRÊ•®Kí!7œ/áJŸY abÏ¥]hëy£Ñ²8éL·&mCÍã3ï©ìÇÏ*]½–4ÛÃ)Ú®WôpÛõˆîƒ¶û-×»d¼gzdi{Ý0oíDÏ!]üøµq|Am¯Çö<‰F{xymÏûùvÁËDZµáêÞÒh…ýp¥­=DÛôÓñy_WÆõîxŒFRx¿ ûñv¿G³½tŒÏ„Ñž7ÑlO×3šíe ¿Æõ~"«4ÛËĈXØMší…ýA4Û˲Ù}e¿ñþy;³?yÙÞ@…+Ý»4ÚKM= Ûñ5ã]2ýIåý&l׻̭D£½¼lýµ0ޝFãyôò˜ÚÖ^j³/ Âh/gB—®î9µíz×7ˆh´——y>‚ýøU¥ÑÎò¢Ýè5„ÛÃËöÎ&ŒãÆ£ÑhçËóO\ioFO?ÞúSáBwm´‡—m–$º»Ï—e»Þ-Û\µ0®÷ùݵq¼uLÊhç{òOœhžïÛh­ây,œh{ŸˆÎ~ü˜ÒhÍFÁSŸÇç}LÙ?ïcÂlÍžÇÊhÝÚ‹0ÛC·ö ìÇ¿þae¼ï4<” }®·pwï$÷™6mVR9ÓµI³=L+ûñkHã}¸qy›pr/i¶‡eßð•ýø=¤ñ¾Ü¶­ˆ^Þ^6Úƒp£­=D/?þ<¯•ýø½¥Ñ^z²ñ˜r¥×”p> ʘ/yùÌ)º.i´—n_¥ýø=¥1ŸÒ‹Í)gºi£½tô7Ê8¾Úûœ0æ[º}ï—¶örVÃ.i´—ÎþæöÄÛ±µaÿûÎ×5a´‡Îþ"Ú&ÚòYð7Úúƒh¶‡as­Âl¶šX¹¹WÑÆõž6žæõž¸ÞÂ~|ËÒÅU×›ýI4¯·Í“KûñkI'¿m¼£lÇä×ëÝœ‚=>ß”}Æ‹ÂË_?±oÏ“h´ ¼”m2öø|™PÆñ×3ºoz.íJï"ö„SéJ—-]ýøó>¯\èó>/Œöt&.º6Žçx :OÚÚƒ0Žïö>&Œö„‰¿h.«<>ïcÂ˽b´‡çiôìôùn¥ìÇwíáÇŸ÷ya¶ö®Uºùñg¼ªìÇÏ$]q<ûáJç&]Ý~b?~-éŒãÙ_Wz/iû¼—'ûakO“ÏÛÛ\Ô™±âIºÐÖŸF/?þ̧(gÚîçèùxKÛ‡à|V¯MiûØ{ܳv¢g•n~üîÚ8¾Ùû®pÙt-Ú[:ûñ;kãøŽñ^4Þ—Ï*è¬íÇÏ-\97y¿ Wú|O^~¼§…q¼=8•1ž¼Ÿ… Ý’4Þ—_>óÊ8~Ù|›0Þ§§½Hzdi¼OOÞÂÖž—­ÐQÆûÒâx^Çg¿·1Þ]o '÷–ÆûÒâý(Œã«Íwc¼ûr«Ú8ž÷ËíÂ÷¥Õð¾÷¥Õð¾+ìÇÛx5zúñg¾I¹Òö>÷íÕñ>,\i¯Ec¼¼x? ÚÞG¢ñ>µø¼öãë”®~|ëÒl/ÃæS”3=›4ÆË‹÷›p¦G‘f{àóøvæx÷ekѼžËÖN(7z%éáÇÛxYÇÛÄ…2¯ç¶UÊ•¶öñìY½3´íøÑßFãz¾Ü·¶µ‡]ð>ñì˵jûñ½Kãz¾<¶4Æ«»ùïŒã‡ÿ>ïæÒÔc{ž[{Ø¿O4Þ‡^¶ŸGØ?óÝÂïî…ù<áD[ö²mÝ™2Æ»›íMØ·ñz4Æ»{£¿Š¶÷³Û÷̧+W÷–.î2¥³ÿ}µIÛx´$[Ë|¦¹àbëA” }æ[„í}æøŒ—”³{KOo i߬¿ƒ®Y;™»‡…m¼ZÒ°ñªrroéæÇó|Þ¶÷™ãY¥mïCÂsÒ£kûñkJÛx³dÜOÊÖ^²­ÏW¶ïÁÇuhûñv=¢í}äxwé ó~ˆ¶ï1%Û®$éDó|ݶ÷‘ãѤí}¤äf+Ä•‡ðÄ|ûñÏ*ãx[X¦lï#%œÏèé¶ûA¸Òv>£ŽŸ¶ŽY¹Ð¥K÷IŸñ‚²?´›{/éŠö´¬¿.n»^Âho›ˆ£ó {ѶöR’½ï ÛûÂqÚ~|ïœ/-%Û|Œ2Ž/ö>!lï Çg¾EØÆ“¥Tô'¶û!zàø†ó%\é3Þ¶ñ¤yI7·ÝÂÖŠM\)Ûxñ˜çóvqªé9¥Ñ^¶ç­0Ž·‰[å4è•„»_ï‰ë½Üv? 'z,i^OÞOѼž ý½p£G“î~üÜÚ•>ß³„y½íÁ/]éÙ¤«oÏak/5áyöðò(Ú~¼µ‡h´‡šÐ„­=ÔŒööPíÅKÚŸU¸±½ÔŒç¹p¢­¿F{9ëyª4ÚK-h/Ñh/Õ&N¤mí%í¡V´‡h´‡ÊþF¸Ò¹KW?¾Nm\ÏŽçY4ÛCG{öã§vÆñã=áL·*Íö0Єýxïß®xß,g=Ò”Fò²]oáDï)=q<û áäÞÒlöa]ýI] ãøëÝ6]Š4ÚCc \é9¥q=Ïz¦¦mí¡ñ~Æû$²t¤‹{ ^Ïóoéån?q¦í|Fãz6ÞO‰¶ó=üøµ´q<ï§h¼O¶Žþ8׳u¼7÷–®°=ø•˦k×®tŸÒxŸlˆ Æûd› ûñ«gÎ'´‰ö½ÐžlwŸt¡ízGc>¡ñy-Œö²í{£0æïGa?¾éîÇ·¦mí7ʘOè ïûÂ~|oÒÕ·÷a?~5iÌGœ‰½%ùˆžñ¾ܵkWº'áÄù†žÑ…ýx{^D/¸`¾(ó gb~iz'iÌ7àÃ’t¡Ï÷pa´‡^ñ¼Îô.Ò Ç7´WáL[{‹f{`$œÜ[ºøñsHc>¢<£Ù†_ï`ÏþærÝÞlϪòr¯­]éó<f{Xgà&]èó>&<üø¹´ýøs=…Ù‰'œé¾¤›?«¶¿»4ÚÃYØ[µ=¶4ÚðÊè/ú åFŸùva´—aÞhŸ¯íA¸Òv½£Ñ^Ž” Ý´Ñ^†e0IãxܯÂh/Ã(£=œkC;ÓCÏ lìSæõnhÑx [Ø¢ÌëÙí{¾0Æ ÙcÁÓ¯W·ç2®úaŒÎz§!=ÝÖž„q¾lá 2ÆØC.íÇ÷"Ýýxk¯Â‰ÞCãa·Ò‰>óMÂ/Ž…ö]Ü»Hóz²¿®ô¬Ò¼ÞÛæ“”íø“ÎÒ„¿WMd*ú¬W^~¼µ7áLŸï¥Âh'á§K£=Lû°«Œ÷“y´µ½«4Þ&Æ+ÂøÞtÖƒmo W”ñ¾0®G4¾7N[x&ëÕýz¼»ûõBº 0Þ÷¦MüJgÚîçh¼ïMÞo‰^[ïsÓ2@”ñ¾>-õDïs“í]ØŽ_loÑx__ÉÏÇm[ïXÛ›p¡s‘ÆûþBr r¡Ï|¥0æs–-ì’ÎôLÂïû«¡¿¶ó½ø<‰Æ|ÏYÔ´ýx»ß£1¸(ŒùÀ…<@åFÛýïË:e|Xl¯Ñø°¬c”ÎôÒøžt’|²v¦Û”ÎE;Ñ; s?zYÈÆxêdúTíJ[ï…;a¼&lçs#%Vã©ÍþTØÎÇY¯S¤ñ¼ÞHƒUöãg“Æóz#½O9Ñ£H£¿ÞÈ­FýòÜÚ~ünÒèÏ7ïáJ$ù¾ÔË`ŸïÛH¶TÎôY!ŒçÁFº¥r¢Ï|†0ž›í5÷ûF¦r£í|Gã~ßlÏ•îIºùñö<.ônÒÖ_Ô„„8åìÞÒÖ_¤Ì¼Ÿ¢­¿8nYÚÞ×j*xß¹ÍýæÇö¾mýÅI×ÏI»¸§ôœ´½¯Fü¾È±TÆï‹$KáÞéÖ¥~ŸaóÝÊÞEºâ÷Eâ¡p™tû‰ mÏ‹è<è3ß-œðûØÂ#éD—$ÌýàÇ]{m÷Ô>÷KÍÈu¶çÁñHÚ…¶ñ^´=jF–¡rvOi{Ûõ¶ó‘ù<‰¶çÁ1K"ܶñüñø‰=«tÁù@>¬r¥Ï|²pžôªÚ8Ÿ|E§Aózã|vôW—ËÆ÷âãs=…Ñ^2ò•}ÆóÂl/Ãú+åê^Òl/xÞ)ú´a^oÜï¼ÞH§SÆïƒTSa^oäm)Wzdi^ï…*.ѼžÛÆËʉîÊ˯ç¶ïí¼žÛÆkÊöû»q•q=‹-œ–.î%ëY„ªœéó>(ŒçA±àéD¯&çAa]šh{_¬§Ì”Æó ØÀH¿¯ Ü”ñ<(ÈSÎtÍÒx^VǹÍõ†ÇÖÞ„mí-ã—Ïûœ0ž÷‰”ÂxÞ¤Æ*'úŒw„y=»­×Æó¾ØÄ…t¥Ï|Ÿ0ž÷ ±Ê…¶öñ\a"Œë…<@aŒç ò/…1ž+6ñ=0Ÿ[ 20• mµž¢®²,•3}¾· ÏN¦è]¥m<_+ò.•«{JÛ~¢ZÙ ÛÏ[Ù_DÛ÷–ã3^¶÷¹Z‘(œ}¾·(ºé4éž´ñóV›æzÁcë„}¾×£?¯V|JýyE§0úëŠ$˜û•Í[ÛÎgcëÙØ_'úìÆõluî¢~žŠç‰p¥k–ÆõÆÂ:éLÛó(š×YÚÊø÷Ø_Dóz"U9Óç}Q˜×yŠÂx7ä%*Wú¼OÞÏ yªÂËÿ=ëO¢'þ=ä+W:/iô×gá~ÑÆù@Þ°0úëÆþ(ýuc"ŒŸIÛ˜oÃÆBéLó|ܶõµó~Ž¶ï­æ)çqçû‚0þ=Žngާ;òH…1žÆÆqi»ß;ò…m=^Ep†t¢s“¶öûµ³= gÚÎw4úëŽ\ya<¬$]è>¥ñó¥Â/ä *gzi̇ öwÑèoGÇ|аµÇçK¹Ðg<*<3}¾' ã~È‹Æýrê£díDŸó-ŒþpXñWå:èó>#l߫괅Òöû¢:“²­«¨¥Œûa"Z9Ñss½X¶ðZïyyÊ™Mã‘i”1Ÿ<‘w'Œö>‘7+ŒñÆ´àyeôGÓ6.)ã}ñäÕlíLÛõˆÆxb"ï8˜y}Çv>£ÑMô·ÂOLäQ)£½ ÏR÷ô‰/å¶èš´}ž÷ÂÕÿ½Ù¤‹ÿ{kKc¾mò~Œf{F^fð`2mâYڮײ…»ÊøÞ²g)Œþþå3Þ¶ýÇ{J£=/äÝ £¿_È›Æóx!oRÏÓÕÑŸF'÷nÂï; yÂ-ä¹ c¾q!ïPx¸Ïx@˜çÃ*6)c¼²‘w(Œû}Û‡/eô§›×3ï‹ÛjçE7ž¢m>çø|¶ï%-!Ï'˜ûe~ù‘e¡è_] ¿ÔSûù?2*Úü@^ûù¢5·¯Ì­±5ìÕˆ´¦v¶ì E»_šßÈ2)­xõß‹öhU<í[ùQí¸¢j Mß6¬N‹DÝh¬Ž´GccÍË@똛Ø ¬xìµ§¢íE›ôôÌöHk *h/´'X *>U–³¢×Q¶Ê£¨uÍÚ“¹§aJ$jÖ²y jÖ²‰Z#ì¬GhðìZŠÖ½Bs u~iDíg¬Ü½Ék°´FèU–­vÖ´Fر£ •tŸ®­‰vÛö-hMÔk0ZíVò*AQÝØ‚ÖDOœpR´IžŽ5º‘¨9Í:¬¨*Ú„‘öøïO-Õ‹¨í½pêQ½+».¾xu¬ëŠ´×âþUÔóöÐ 5[í“ÒÀŠH{à¯âšï´íÕè#QïuX#íSÏ@õîHT´ýª—ùü ?P¥5Ò^‚ÖðE¢ê0VE¢®0Ö/DÚëåÀê¾HT‡MÖ¼#íÕ}~Uˆ|§ ü&ªJßœ˜öœXui“Úk "mH<3jõÚ ßd-Õ@›®šX-i/kÕ"m(4q³GÚ@dâ;T$êÝTº8XÑvà‡ ´WÜùT=¼hн’n ½^yEÛ@›]<±¨ðŠ Ü‘6dZø®y³ãñ´°ê ÒO k’"íñt¸íñtjÇE{­ñT³{§=€¼:l ½j/¬$ºÉ!Ób¿@{MY¬ÒˆÊ²¬Lh¯p^Ã5ÐO‹µ'mùêN^ýí¢½œy ×@Û:´ñeí&ƒ#¼Âk ½vl|U‹´ÐÎ^í¢}2ÜÕ+Š]´Ðn¨GhŸƒ^ìCuI-”FÐ@»ûùFòÚ?Ð@«D#Qá•UKíä•Gí|oÔ¿ Dy¬”ü§¼xæô”/»xö³<7I#¾2ñ ÞÎ_å3­dr4ËcòG#þ: :ñÅ•íå¶—O®ñåÑ çsyùŒÛh,O'ŒöÐx¾£Y¾‹ç;÷ë‰jÒ¸_½|Y4âÏ×m/ÿàåÍ¢Ÿ|–w$iÄ##‡Où)_fç3å¦W…Ñ_{ù­h–ÿAù½`.zÊ_E#>u”ï团Q´wÿ¿ò%ý)ßp›å‘XÞ%Ïk/u›ËgžòHшÏî–Ó«Ìòcýa´M–äÎòiш¯î,Ÿxóóà×FyA/Íò)Ùãóo³sðzG£?¼žÑ,Äû-ã‘ÉëÍò,OmË'Nù;_Ñ诼¼Ëåæå!'ÊK³¼ âë…1Þœ6o ŒñæDù#a–—AyaôWs"7ýÕDya–Y6 ^^Þg#>5ýÙÜ8Ñ,ûAýÙByaôW« þ)šåKªÇßN‹®Ê“åVõxçÛoåC„Q~f=ñÒ·~”ÏÆxm¡|0Ë¿ ü…0Æk‹ñäѯ­åñÔï/¯ß'zÚâÅ¢Y¾åk„ñ<Þ<ßÑhÏ»!;íywÄ/F£=ïx»h´çÏ—Áí™å!„Ÿò֣ў÷ôxÚÛhÏ,ÿ Œçõñ’fyÆÓG£<Ô^|å¿Î|Gf¼í±ÅËE#Þ %ÄÃE³üÊ‹#^4ÄûGÛxô”W¨]ñN/ŸñŠ0âCRCø‡ÔÏx»zùÆWG³|ÄôøÒÛˆ÷òшgðòш_IËãeo#^#¡¼Œ0Ë3lî‡`,(ÿ!Ìx|Ä £¿+(#Œx*ìÃSF_,ÇKñ1e<ñ—Y^`<ñ—ßVß<½<âñ…ŸX¶ÇãÝf<{ÆÏò•×#ñZµxœÜmÄ'•!ø›Z<žð6Êg (Úãç*¯W4Ê[T”ŸF|ãÝ…??ÿø Šò+ÂŒ‡Güº0âŸê|âÜ.#Þç4,åŽ÷uæ„*#žùxJ£AýYKßvý:NeÜÏ­ Þå6—7Úƒ3Kã~nŒFÛÿ/ã~[ˆÆón!Xý)ãO…ÑŸ2ÞT˜ñ´Ûÿ½wŽÇ7Ö„ ?ñ¢#I3þñºÂè/vFüL4ú‹]°½?ºx|¨o¿ŒçÅF|apæóbó|Gã}ƒñžÂ(ÃxLaŒ·O)\=>²fiÆbí³°õ‡?9…YþãÄKZ팟Æö³”Ñ_Dc{+ã'…ùûð|G#ž!Ußþ{ÛóÖ`ßÎÛ«£¥Œí¥/Ÿõ6ÂØ¾˜Úûöß/3~²Ûõ®Ý½¤ÿ9l>Sñ^Œg ^Ü^˜±.SññeÂŒ_Dü§0¶fÄ‹ cû%ã…o‡µÌÂØx•§o}·oOe¼ž0ã'7®G4¶_žèæñr-I/K]šñlÝ·¯¾›å=Oø¶†x¸àŒïñÇg<'Ìx¬çæÏ‹xEaþ¼ ç#º|æÏ‹õØÁ‰ç»Ù‹—2Î7ã„„õÎÂh/ ëE…Ï´ðûFc{qÇzÙÛÉã;:âO„OØ"Œûµã| 7·9ÏKaÄ»2GýMÇzäàåñ?ÃâÄoѱ^YñˆÛã…1ÞèöaCý=ão„ßÑï<=¾ñkˆßÀÂuel¯ˆf|Œå´+£?dÖë ?ñ7SñañHÁýŰOåéñ6ç}H˜ñ2VÇIÏËa•?Âóy›[•N|Êé/…ÿ5±W˜ñX+Œûya=\peüÕB|‹0Æ kà~Æý¸Ì÷UÆóð¼wiÄ#,¬' öçÁÆz\aô—ëA…1Þܶ1%:q¼¿-ÇLçcó|FÛ|ˉg8ãËy3žˆñ ùñ=X!ÿ»ã9ýä)\´G`³®^°½§-\¬øñíëb¤]üƽë™´Ù¸w= Öm4Šô‡sOE"×À*© Ú –zº¾–D|žõ‘Ϧh…'DþUç.¼Y0 ÙPò/'‡»ñ;>ßß"mzU¿mñœ§K\Ìèú=]"Ðƨ‡.h­†"J‘¸IðÍ(–"í…¥a¾.Òg Åýn¦'ÕÂpàâ_e 8Ð>4µ'}àbÿžˆhÁJ6EÚ#º¡ S¤-`o¶çæàò–6±5IH=ˆDS™Ø¡ˆl”iŠDŠ ÷½Z¦ r¯þYú]'ó2vRD:ŠÈÆÌ‹Ú‘µn$Ò&š;Òa"­éÌD¸8Ñt«$h·FGÂK¤Ý½a_t u×™68í¸­"mê¯cf5iHx¹9ðZÜÑ€#‘–‚c‘Ömv6Ñ@»»¥"ñ„Y ‰iç[ÒMnÅê(̹˜Ó±³¢-c>ÜŠHñ@I¦H{,ö‹h]PGîÌM.Ï(†i/ƒùöc &D"kƒ)öc 똹©ià›@¤Ï ´/ÅHOIU?FÅ^Ü‹t pL$~ÈŠ hp`–:Òå£ãÔZ(¨iMt0{"Ðn«ÁÆ‘µè»ë@ë  •EZWà™&öZ5¸£>Ð&ÍRi"OöŠL ”WŠ´×O< lL-±<…@OKÙS11—äŒÃ/vVݞȉ´EŒsô‘6Å21CiDßOn.0 ã~«øÀ(œ*=³0ëUœ€˜3žÆõ<;¹²tÅ¿‡z ÂÈh¼—ëfvò)cyK¾Áí6pb•àRQ9Ït Ê8_ õ‚¹ÁÍd„q>OÇ™¥±ÿÌÖip4|@Æý¶°ÁæödXÃõf€ÎÆšèî6µKãy€¤ eô§×#)CÁÌãÏHªQF€@¯ØÝ<'i—Iï&º>P¿÷6³ÈÇ’Æó¶£^ƒ0žWç{J#`‰9 ÂýÌ \ǽIÏï9ÂOÀÎ^ÒÍplƒS4&òâ…“ØÔ&\9¼_£ñ<=s}[ýB>•pƒ€$aÜOõ‚™K™Çôzð·0xæÆš4Æ[*¯œo IÏ{|#‰Î `:_ÁŠôô€œÙ¤6‹×s¿Í€°·èä:u{Ðlø}¢‡›”n£=N c<æÔª4n†×csÙìïç@½Ëèáæ¥Û BÀˆ0ú#ø{€áœ^ùöpÛ¨hÜsbD4ž N—1ZÉë]¿›˜Ž×F@ÝÊ^ø6ÆÃ ðfÀO}6,]În«—w›ŸžŽw—Æû|„»{éê@{Ig¸cƒÅm|¹iãyÍ€ aô'kû†ÛÕ=²4Æë ÁO 8„ñóì‚öþl#€I8»KfòY‰³´±A'!À@¸»­ÞZ4ÒóïÝÆNÌsúÌf–ôtÏ.ݽŠtuï$03°ñ¶o IÓ7ÀÜÆîã%Ý= ¦$éŠ FÑiÓ^ïï;Áõ¬$ÒÃ^ƶ÷=[é[¥beE”m¼ï.·3tYoîöð€Û€mãe[iߥ±A4/,¾½<€"as4F(GwT €G׫à| 3 #Ûó ˜µl§W“ƳÓp·t+ôiÂØ W*°ßïXaÜ&û­ØJKež… &ÑØ@Éal`+pö  g'x“ÆîŠal`e@‡06p"i!ºñ~¯ õš¢'(:6E7ØXÚ˜9I+SBgåÞæ ›øÒØàÛ0"Œëц׻º€“¶|CÄ»™,o÷Mp{ÆŽèä ¥ g¤1`A/÷*þI«Ñ‰ÂèzŦèêÇŸÕUÂHcÀÃí´Ùž;@„ùûàzcoÇìàåXW&Ìë1±€<ôûôö·“4dåÉ€8wPY @N0š0°­’ÀÆóx <˜0@ìØ¯G4Æ帄›ô"Í ¬Þ öþhb½±0°’NAka…³ØüÛœo`A3eÜ/(Ò¦Œö:6€Dc÷X`~›•ÃN€@ÒxMyÆxfb½X0çÎûÔ¥‡o?ï›Â¨YV6D |ƒY9ëlp[÷ËÆka´×uá—ÏyŒ‡¶­ßVæp[¦Ì ìÍàßF°mýp4˜–më3”àµíÅYõ–“­Ù‹fÀØñNÒÜÀlkY¢àÊ ØÊØÀ’¬$“26HžŽk 3 µ&koʨ‡™l'ƒ26XdKÚ‰fp6P×!¿/[’¦2~ÞlIñÑÕ7pózGã| °©26¸`'vtáf‹l3ÊØ ‚Â×ʸžÅÖHEgŒ·+ Ï+£^<’”Q·XñËh¯—^øûFcƒd± ªÊØ V¬€GðâbºZ­D–26ÐU[ÍŸZmÍ27ØÚij270wÿûÞÍgƒïœÒÜ k«§• Ðì{´×ÆVeÜØÀÝy½›=˜” Ðl=…2š­Å‹nø\±0Bõž»5Œh¨q©26ÈuÛiÍeºgƒçJÒØ€Ö!fiÔëÆÊèÌ ¨'+Ic/6 F{ǰïÕʸŸPï9xnnNâ†/÷û†¯¯zº¶:û_üÕ‡Ûž5cý‘>þî××Ù]ßw~ÝäÎ/Ûe¸¸{‹…u/zQ^+b‰]m6g9¸kÌvÊDroW.Š6úVß÷Ö5Û®h¿ÂSÁ÷¢½h·‚¢ž‘íãkoW`ñÝ[YÒ.9Ö¥ ¦¯ýY™»·ZRLÜEÕ³$÷Iñ’½±}ífÊŠo»™Q¢˜û•"¹o¨uÅÁ}C}KòàÙ;w3±¬òEìU\8u‘ýãÙ‘$X>ž=G‘ö úÚ'œŽ‚õã«do`æn&îs¼ÉýJ¹(bNA‘ÚÈ öYñZП݈mÐèå~# hû íƒ#÷ ÚÁ,2ˆÂÉ,èYÁ3ÓiÁþ” ¾™È©Ø¸_©HÚÍÞŸòÕ w3µ-™ÀQQÐehírcG$Êý>¥¯or¿RK‘Ü…‚ëÛ~%A߯”'÷+ÙvÁþñ”L =¼QvåãÙÍiað¶Š´=)åØQÀ˜7]$žK»™PÒ5²»JÚþNŽÌÜÍTº¢ï„ªE²ƒ½IÚn¦æeÑßÈýõ(¢ªXÀY­! Þì‘ÜEÕ«¢5´Žbð–Œìà\Šh ]}¤ÌbÞû³VVÄ»åEÓo°7Eös• Œ}‘öIil¿d78·äiÀ§!7E›ÌšÅ÷tßä>©6#Yp}"8NÐvoÙ[¦ u#“c¤ý(iȉ=¡‘öy˜»™"±·kù¯p3<¥ #+wBq×öÍάhÀɋÉ}R)+Ú0€%¨+ØzàØ,t}‘(ºÌ2Ñ œIqr÷ËÀßàî’¶«TäàÞ®óý]Ð~ ”)&5ã/zéë1íf_–®¡˜Á=mÑw 毲Ùö˜XØ÷'XÁÑ3w~¥"YÁ<%¹ã¬ÅäÜ‘LM䎳HßÛ•‹âäÞ®¦É½]ÜwÑ&·½ wdýø*¹Øyp/’<øtõ‘6)±±§C°}<;¿"Qr{»m¯SÅE ´É˜ð½H{¼Ø†$ÿ*»q±Ó­yrÂÍ ò’½‘5ÕvÞß@ëF0Ñ«È]cÌF¸hápJòà];÷”Õ!™Á¾±Î>*ÚÁ¼qm¸Yà:²€vãÚÄ‚—6´_pcÏ~dkS´‡ÈF,¤ ^CK(ž1ʉîExxÍïd£*á·-vÂXâœ)ª\ÝSK8SA@C4–È&l¡RÆßÇ»![ï‡èg ã2‚ñ÷!³R[\Ÿ.ÂØ"8ü|ÝÆc¯ù.œé’¥“_–v¥[ÕF{`;¾íK$Ÿ‰.Cší…­9[P2¶ (zeilɨ™-Ü} ßjÚÞKKxsñ‘`ü¼ØÒ«œèº¤¹…®ùù Nô¨ÒÅçù ®ô.ÒÜ¢×d!ÜèÖ¤QÓ3#“3¸±fkFf ðò¿Ïbc„ñûZ¸±2–Àb­›´Þƒ…±Å…[…±*/Œ<£›{mÿûÖÖÆÏÃû%šíac,/Œ-ŽØB-Œ%ÂÙ·Xòzûß·²påõ,بÜ葵“{Ks hÆxUØßYí¡ô‡Â¾E´véáÇÏ¢mýQáý%♛ʾ…tmi´—Âþ@8ÑÖ¢Ñ^ 2_…Ñ_ x…t¡[“Î~|ÏÚ™^I:ùØ6´ Íëùîâ홢ÂxÞ”éñAÁ….I5à ûaÏþ zøÝÙ´q>6Þ?£Ù^6Þ@…ýøÞ¥¶ b ¡²oÁe4Óm´—ÊûMØ·èž¹a´§ŠÌZe_1ÎnžÏÛÜ‚kS´oÑâ`áåÇÏ)-fuxÓm\ÏŠ-öÂo`¯¥2·Pc ²pu¯¡]è]¥¹{a<*ì[„“6ÆÜb¬ülA^ÒÉ·÷Ÿ8Ó£ 'ßbkÅO•—oQîU»Ñ³Kã}å|MÚ^?q¡w•æel±T®´½ODw?~Linq¶µVÒ…nI÷3Š?Kûñö<ŽÆýܰXÙ·€—%˜†-ÂʾE¼è¾ýzÙ‹Ÿòð-à;iãçÁaܯ'ˆ³i7zTiÜÏ ÷«2ÚûBN4"h"”} zßÒ¸ßîweü¼AIѼßmŽUÚ·¸-üøµýø½…&¬|‹¼r¦×”ÆÜž=æíöð-ô^&Ü麴}ÆsÂÜâoi-ÀÂh/ÏSaÌWtô7ʾÅ?winѯVŒŸ§¡=Dãzwô7Âxwô'Á̬;Eÿú áåUó ýAôð‚±´}žÂݽŠ4¯×DH]4#&®·p¥[–.~ü¨ÚøyØŸ'÷”æõ´½Ðʸ_ûF ÝíÁ÷Áñ€r¢-¶.ï{xðIûñK×{ ^Ù³ýn£=``&ˆ‚â!„·ñ¼VÜEºÐ»H7`ÈM»Ñek'zViF TÄo zwiôãåL³=ÜfÄ"”="¢$éäM˜ +3&.•1^D@±t¡K•ž~|ÿ‰½¶4ÛûáF×*Ýýø3^Uô™PF{`þŠÊ™nÚ¯ öÂ8ŸÛ1oc¾r "Eã‘Éþ@|ÞFc¾’ÊUØ#`ÎZ¬®]é]¤Ñom_ÑG3ÃJH'zVéîßîÒ˜/`Dˆ0žWX—)Ýè±µ½³tÁïƒa¼/ÌîÑ¥Áͽ¥ñ¼š¼ßoW¿žˆ QöO ã÷™OLée¼NÞÏÑÓ#DZÒöãÏ|†0úFŒ(ãxDàc¾{"ÂB¹Ð{IãyµbK£«»þĈÉ~=n£?XoGãyr¼µq<ïÇh|O[ˆ¨PÎt«Â\ÞyöjØýùHìåöãí}7ÏF¨(ûñ •½=üø´q>:Þ‡£ñ¼a„вÝO‹÷S4¾§­÷áF—ªé:¥«»uéâS:ûÏû=X÷»q<ï·h¶—‰÷™Û™ãáÅç±°ÿ}»Hc¾ò¬öÒè/ïwáîžÚˆ²•PÊxÞ¬váL[{‹F±a$Œï©;£¿ÎtßÒŒÀ)x.t­ÒÏ2bG8ûß7¶6þ¾æ¿ïm´‡=ü÷ ®ô‰íeOü>¶ŸG8Óþ÷­m«Ž¤3mýa4ú£mÒ•^SÏ«SSºiãøþ*§øW‘FÍí”ð¾-\è2¥1“æ…уˆaF°Nt^ÒÉ#–vÖ.îݸÌôì•?ß3•Ad{g”A“ᣌ¯#â75Ó‘¥ Ý[7'"zh'zViD¬¼¼²¶G4Õ"Í+ŒÇ”ýø3ß%ŒdYH£=à~SNtíÒl–Õ!]éÑ…—G€á~TFDî'a´‡l›òôˆ©ó¼R®´]hDøœõ:]÷ƒ0"4N–ÍÒ.tŸÒuÒ!-ìW)K#¢$7DH 7zi{ßµâÓKÇ?Ÿ·“ÛîáDó|¾› JOVÔYǬœéÒ¥òò/(ûñC)jÂÃ#¾,Â9g=O“f{؈ØÎt/ÒèJ²÷=åBŸõÂÅï]×» âJ"„Ñœõ,CÇWô'·#l ïak罹ü’^›¶ˆráæ^ÒÓÿ>»¢~žnã9åF÷.mï+Ç£jgzNio£¿hÍÏgp¢y>o3B²Ù÷"eü¾ÝÏ÷» ŸG­£?.´½D#â±ñyç jõH'ÚÆKÑ÷ñR4"ÚÎÀkk'ÚÆCÑͽŠtuï$ÍëFåLÏ%þ ñyñfÛ(1%\èÜ…=b¯ñ~öã­U4"»-<“nt-ÒÓïM»Óö¾"œèÕ¤‡¿—6ŽÏx_Æó¤#S¹ÑV‚*óY=£=FãzôŒçÅmn&÷ˆVáå¶öþ¶³?Æï‹Va¼ïuDø*Wz&iD˜3¢U˜ç³¡½ ÚÚ[4Ï'û#áâÞÒÅŸCçcày÷Á³ð$I3¢•ýÍåÊ*æ¶°ªk'zméåßy ó~²}¼Ò>ïcʉžKzøñçz*#Âv{ £ÛèûF‰áD{‰¥Ëxžwä‚)Ûñ(¢«ŒñÜ@"üDড鲤Ñ_D0 #‚ùl\H‹ýñ(hÂ…¶ëñ6ŽHºiO?~NmûU­£¡ÄL4"±1K:ÑC»úñÖ„+½»4"À É c>謧JÚ™YxzDqG‰"a´7ô'Âxžbã°òt[{ŠÆxzL”ˆNtÒŒˆFD²²oí5ï[Ø9¯Œ÷­a§Ò…¶>ÑŒˆ^h¯Ñxg„³2~žý”»èY¥1Þ:™!IÛ~Þ™P‚íöàxz"›P9ÓV‚-šЈðžn+AýDTŸùåJŸù^aF<ÛÀHÏó‰d(a\ï‰ñŠr£Ïx[ýñ´wÒ8¾ázDc>ÙÊ8ßݯǻ}¾“ʘoa„¶0#´y? ãçáý&œèµ¥aR–v§Ï|­2þ>¶÷hÌç/¶·hÌÇ­äçã6æKñ­œé\¤q½’•m%Ä¢ñ½†áÁ>ß¶0žUÆ¿×Ð_Dc<¶ø<‰Æõ^ýp£í~NôNÒ¸_ò…1ß¶ù§ìÇŸõ€ÂxÞ.¶Wa\¯ió…ÂxÞ.¤’ ãy»Æ+ÂnS:ùñ¼žÁ•ÞY˜ñ‘e!#PýåÉô©Ú…¶þ&ãS¤}jÛùÚÙKÞÆxw³?Æñ¶1T™òHƒUNôlÒøž¶‘Þ'Œëµy?Eãzìæ%ƒ3=·páxøÒv¾7ï—hôÇIœÂOD¾½ ã| ÙRóQÛRx¤ñó ÝR˜íù–Ê~üÎÒø^¶‘€©\h;ßÑÕoU;Ñ=I³’0•3ÍëuÛÆkVb kg÷–N~<¯g0Žçýt;3µ$”‘Ö%íDÛûF´ÇO „œ´‹{J#"?U¼¯FWü<ȱTÆß‡$KaD¸'û0'ßgx Î۹ѻH'ü¾H< æ~å㦽&mÏ áBŸùn቟w{‰ÍÛ£Ñg¿Œ0"ØÛ»°o%Z£m<\sò’žÁ•IÚîÇcï ãxd —éžÚ…¶ëû1óy"\h–P½¹øõzsá|ر•È‹f‰ äÃ*WúÌ' ÏI¯ªãñ<ÆýšJŒ Ûõ>%ö²ôW‰Ž¤ó…|EáÖé3žF‰‚—[Önî%]6=ªv¦ç”Nþóî"ì%02Òé”ñ÷!ÕTxuú<Ï•m%D£í{Ó±]ïèëaY•Ò‰îÚ½Òç{»0J@¼¼«¶]Ô’TÆõ,ÙKð÷’Æý^„ª\i–ð½Í,%p…ýß[M:ù¿ÇöðnΗ±Œt¥­p4J•гÂͽµm%7£mß[„§_‡¶ÆkÂxžW æ’.´õÑxž×ŠÊÂ~¼•ŽF $”€R¶ù”Љe–ˆj6_#Ì뉼Ceü}%š£y=;ÚËmΗ™§¶Æóʉö’Þ—y½‘w(Ìë=ð¼‰Æó.”1žÇ‡iü¼ %´£y½g¨\è=¥m½Q­È F «ºq½£ó¦÷OlÇ7Û8£Œ÷±–üz'š×ëÝ•ýqKèï£1^oÈžî3Ÿ§\éÞ¤q¿6ä ã~l¶qPÚ®GcrýE4ÞÇš×IWzn邟·ây"œèš¥1kÈSf µŠçÑíâ׫¾•Œÿnü{ì/¢1oÈCæõD°rsoéáÇŸù4åLÛûd4¯7òT…›ÿ{kIc>¥!X¹ÑyIcý¹0æ3&úsa¼MÞÑ<ŸÈËTNô(Ò˜o\ViJýñBžepç|ó²sÊxÞ-äa c¾o!ïVã ”ØUÆûàBÞ¤0æÛWGö°×ÜØŸ-ä= ã~[Èsžø÷w(Œö°0FtÖc i[¯P7ò…q½·+ãy¼y=£q>^>ßÛ‚«—Fž±0ÆCÛ^ì”ñ<ܶpDãżJaô›÷k4Æó{ùß÷î‚ñBKÈs¶ëÕÏGtsŸïQÂÖ±$t4ó¬[IxпüȲdô¯- Iü§ôÍd?*ÜÚ âW-è‹Öý+ÑüNëœönDÚÔL³ªö‚öbõU±ùܶy¶GmEÔ¯®¨êˆÅX‘‰ÔõQ­¹aÛ¥ .ÖRGÚEi¬y‘µÓ›M˜ ÚkvûªpûNœØ§JíEœXÛj.hg£U´î½}UšýNNnœˆ®hŸ†Úöâ¢íŲaÝJ¤uÛõÉmQ…×dDe]Ö'L¤Ø‹\®â›Qÿky#m*´c%o¤ ÔO US´nÊ«.Ú$cgð@T{ƃ›|Aí¬éˆ_¡[ïé›­&s }Zé6´ÇÙ×3mX|â…“¢Ý8kv##ppCOj^´ÏQ«#íaÚj1ZWÞ·WŒ¼ˆ:áXçYx𙥌´—¬‘PÃåâÂm…R‚vÉX+8Ò^FñZ…QuY#QÿÕ¼#íÅg°~f -‹¨Úi/¥¨´'h¯<¨ÈÀÔ,´áÄÀj¿H{Ù9 j(ÚR“ÉŠ‘´:<£ØH›¤žXci¿þÌOí¾wÚk²¶êE–Ï›X=i8Ak 7{¤ <'¾KEÚ´Äk Q†¢Mêž5ûIÑ–QkÔ6¡Z§º»ÿmP´‡×)ºh¯õ«F#QÓ•UL/rÉ®W" DåÕz˜vÓrŸ,tq¿‰åÌ¢›û¼N #î;ñ F³Ü%>7³\%Ïñm«LÏ›h”ÛIø\%Œ8~/·ÝÜÖGG#®8M/Gtq¡i¡f9Oö¶ÑO9κ¢ûöršø/Œ8ׄå6ˆkÍøÜ%ü”ÃÌÚˆƒÏø,Œ¸ï—-Î3å(2Ê ³ÜdC{4âº1/½Çñ¹Uxùñ,wrqÚÓ¥Â,‰rrÂݽ›´}Î9å Yæ6Ê}°£0ÎWIˆk¶Žó”+äÏóîéåñ9E˜å ñ¹Yq½e î=qÊ,—'Œ8ܲÇܳ Æi”Ë^nk/ш›­øœ+Ìrmø!Œþ®fÄÍF£O-ˆ›.pEœu4â¦+ÊA £ÁÙ—V…½/öý)#û•^;⤣Ñ_V¶—h–³ˆ3F€}ëʈûg¹:aôç;ýeãù¾íåf›ý/Êh,g&<Üv¾¢Yn劢GŽ\9e–ÛB¹La–Óâù¼íå ¦[…í…ì)ÇÍrI(Ç'ŒûÝËaE#Ž»cºXqÙ^Î):¹YàÝÅËE±œC4âÐ;˽Dãzz¹¨h”ËñrIѸ^}#®9׫oô‡Ñ(7ÒYNívößg{œøí§ÒÔÆïãå¢YnŸ „Qnrd<£ñ¼ÁxVåFñ8ûÛ,×Ãr)·“—ƒb9²hŒWËD³ÜRó8ìÛˆ“G]qeþ¾ ÏÛhþ¾Ûãºoc¼:x=/{\¶•³ÉÒ,ç‘=nû6ú–«F9…‰ó%Œr,÷"Ìr9(7!Œr˜Ó^^®dØû€r¢ÏóL˜åjP.E¸»-.7ÏÛ‰r Â(—vf;¶4ÞGæFœj4ú£ó9wOÆí¯âq¾·Yî儇Û⠢ўWEd4ú«U÷ßw¡0Ë¥ œˆ0ÆÛ‹qÓ·Ëi±œ†0ÚóB9aŒ§YCãåŸòh–‹YhÏÑ,×±=îö6Æ å‚;^¶O¹ŠÔ¥Yî€ç;×s7ÄeGã}awÄ1Fã÷ÝqwÑ,g1<îö6Ê%±\„0Æ刂Ë)¿¼·4®7ËA£œö^ˆ«‹ÆõÞŒ«f¹”3æùØ8ŸÑohûrÊ9X\\4Ë1 Üˆ0â’SA»z9ÆiG#N#Y])iü} ýi4â0RG\c4âÄ㬣‡àå ¢Y®‚×#q$^NàvaÜwZˆ Œf9”›FÜKÚx^D#®)ózE#Ž)'«¼¸¼S® IgwÏÂG™Ù¢§âå ¢§ÿ}»K#'£¼¸¢Ìö¸ÁÌ8äh\ïÌrÑÙmq¿Ñˆó=ën¶°ÇAæŽçy4ãþ;ž—Ñ<å$¢ç?8¦Ì8ãèânÚˆKôrшËÊV77ø‰C̈;Æýž7âÌ¢‡».iÜï,' Ì8úäqŠ·ŸrŒC¼óQ²Ç!Þf¹€ìqˆï^Ÿ=îî6~ß’w¸šR'͸~ÜÂè ʳâÈ…Ñ”ÃFPЃ'㘊Mü(£½—áqm·§Ûâ8¢ÑÄù*ãçA\¾0úìÛWfxöŸ÷6âè*¯Gtr{¼Ü›9Ÿg¹.Cz¹=>î2úƒŠ¸WaœÊë8îŠrÂè÷.Œë}þ—&Í8s”c ærMcž·Î8·Ûˆë9 [›qôÛãÆnãzW«;¯lß3¬ãÒ,€r ÁÍ㸷,ló5–Ë[¤§Ç§-ëÙlbFqÇq¡ÑŒCG\¯ðgÞ·0çÛlàÞ¥ùûtÄ!E?qáÇþ/>Êí‰OÒˆËB]e´çÆ8Íhþ¾Œ»Œ~âºÇö8*Ll(3ŽqÛÂh¯×3º¹[—Fœæ™x,ÂÜ.f³SãÕθÖhÆUw;ºÍ¸aöÑÝ=Š4”ƒFœUˆ«ŠÎø}¬.R4¿§Ú‡¥!ýÄ1×&þ¦o<¢Ÿ¸dÆçÜFœâùpÚ¥‡Êö¸êÁ¸ÆË…Û‘ìÃñ’f\0â’…q½F~ât.c¼9°®P÷ÓYi0¥ñ¾4ªÇÕ¼{yœnµùRaÜOŒ3®î¥Ìù*ó’ÆûÊèØŽ¸ô¸rá'N7eiÆU£½ ã}óÌ·Na£hoÂOœ,ã.nãz2.V˜q±XÝ)ŒçåäõˆfÜkÇv¾èìq¬¶ï6ç›lø’Æx`bý°0ÜïÂŒ»E«0ÚËDœ¥0^ĵ c¼7±"0¸q¼7w,Ìßq–Â(·Â¸Wáêq¯©HgwiÒŒÛd{¸íåf^¶í¨Ñ¯.ÄU ã}na%½0Æ+‹í-qÎ/3àÝü^vâ^×”F½,ŒöÀ8TaÄ!3îTï kû¿wïÓkBƒýy¼''Œþh#nWxziÒOí‚íþÑ8ß/sûòmÜÏq†ÁÜÎ|â>×”~â<¹÷6ã8±Sí™q•ˆS8åj³tz¼¢3ËÁZ唯vÅ„µyÂÝã*[–Æv¹”±]5Û‰G)Œ8ÆI/c¬¾ôöpï%í^©Ùx@qDÉöž(ó|6l§»=¹=ÙVel‡NØ Œ8§³I›qzX³)\<ޱuiÄ­dÄ™ 3.q ÁÃãð7&̸;Ä £=e¬efœ VÊ 3.«%…“Çå-e{*<ÑŒSËØŽöT‡$̸,¬ó®øû×!\£'ɸàáq5ˆËFœAÇzdaœ¯³pµI#.¼c½r°Çtl—ÆýÚmaŽ2ÿ=Ä%ãyÔ÷ìq{qlÂØÎ?G"Œ¸¤¸aÜÃv´*3®q9ÁÕãjªÍW 3NÆ^ì”q? ¬×F5x½o{Ù@\’ðw³µ›ÇÑœ÷!aĵ Ë¥UÎ7sžÁþ<˜<ŸÑ¸_&â „«Ç©œùaÄL¬Ç N,¬‡Æxc!ÎEýŸßß}–‰à|/»ñ•‡ÇA¼îçh–C;~ý¾ÊˆÛ^¶žPíqÛzÜhûݶT¹y\DÒFÜȶ…«Ñ>ÞÞv>•+ ÒDs<ܲ­ç{üËïÁ ù߯€‹üä)\Ä~ûdo ‘…y L[¸ˆÔºÙ<{{× ^5²r7~ò‹6lj(*‰LÌ)EÚ+i³© AdX] Aœ|-¹Y0ÙÒ0W¹xðlŠ\[±2?âÜ…7ù‘¥¡`db~„ïÆÿÎôÜ»¢-À21Áò=]"кO—ˆäÁ+ŽÍ´UŠ´NµÙ+¶cÈ3Á|]dâÁ¶7òâÂÖ™‘žˆqp¤miHÌO"F$~A”päÁç -Ò&XšåD Úð¹Mì_½ÈÁ]CêA¤=ºÏÈ´+ÚÀ¡¡lS¤M+´gßûE{h5Ìi Z6A²7â›X/hÛ ::ŠHüUH‡¹ÙqQ:º‘Hkfé0‘vÚ;3.r¡ÛË´bíä"?¢ù¾è‹“i–‰p‘;Ÿ;o«Hfm´¡ØI;±H¾kú¢ z;JŽEfæe¤ɼ׳1¯)"ãƒY ‘v0³6mª·£PÓMN¤wîóœÎ-iy((ÑiKhúÆE‰Là™?ˆ´éÇâH‘Ö¼ó#.&|Lˆ´Ç`ÊC -«?_m«bbŠGî‹™"˜@ÛÏhåܙH»­N“YŠ™×É:»£bL MÃöìb}§=Ž‘ç ˜™-2ª"ÒC0?+hÿÿMNNt×7¹4l pY¤=Ê™iiÓ.³–‘öÐH¥Lày£ºÉDŒrK‘HÓ@½Ù08aâI¤ ™ii)ßQ"­ƒšÈ‰´;1GiC—‰ú›•.ø~iÃifšDZ4‘‰‰“3±³ûbÁ‡ý‰²Z‘6œž(Ê9yð¬ŠHZA¢M¤ÝóÅ‘•y(g5ƒ`GŠd—»P(JÐFŒz¤u2‹7{$ÞIÑ>q.ËplL-IMÑ–Ç-|=ˆ´WžÅÓ~‘ÉË É ‘vÞ÷æ-™À‘í•g-4Â@›òX(˜iæ¡D&¦¥´Ø8»0iK¾TDú?4w$S|šY¤u sØ‘“I+½(ZWÀ¤•Hä’dß}wÑWmdC&ðÌ\ßd.îF,z¤M{œYI{]Ú(™ts0‘"íro”+Š´[rãk\¤=pϤES´—‹/q7¹f#ý'Ò†³à7Ycwã‹ ÓaÆRÄÑ„ÂvŽë”Æç”ñ{D—J¯& ÌFsP~|zalpN¼ºÑØ@Æ€alM¼ÂÑØœx£K¢í.¾]<¦=Œ.?Çß`ô8ãµÃp’48> ÷N·&Í€äcƒsBýa^Oô¼Á‰×+£¾‚ðòk/Ñ pAýµÛ•ƒ\ó–ÆÆŒúÂØ ›Qß$Ø7(gÔÆî<°õ6óïŽÏÓExùñ«I#€%O,ðŽfÀη06xfä»sà{¼º4ÚKÁaX”„ü·;¯7f„±¶ _]¸#ঢ>n4~_‚QN~¼-ð¾ÍúKP#Œ ò(§Ì€õÔÿ¾Œ‹‚üqåï7Á“lù¬tÎÒØ ^QHø °YèkÆ‚èhØlpŠ~lXÏ:ØÚkm¾Áá6î׊L ¸ÁÆ`ì3n eþ}ÈçÎ ³²tB€ÍöúÕïöÆëóÙú“Û‰ N 0Ԧ̯¢>s4ÿ½îõio#À¬¡>€06@3@æ¶/Ï )#`쌤†4ÚSÃDeü¼ õ·n/ÿûp½„ù÷m¯ÏyÏ3$y(3€&yýÖÛXê¸Á¬™;†„+þ>p£½¾<†4 :>p3qÔn„ñ<ë ìo³>á±/(¿<ðï¡~ƒ0ÏçB}µÛœÄð€aŒg:î7áæÇ÷&]üx[`{»²¿è<¢—àØãhû˜h8Iºy€Õ¿ŒÆý4p¿ #hTlP¸íGÃæ{”Ñ_ $ 3Àõ„Px¾`dil¹±&ÍŸå?ï»=Àhl¯ŸuýÕñ”FÀĄ̃Ït;}ä|¯×ôÍh¯³ >Xt÷«ßëÅ€aÜá[©2ž''Œ~FgNñz€Nðb1`ÏœZ•F{žÃøÞf€ÐÀÓhxN¸c<4Y?!ãyø¯€«Íó=±2šF æ¢àƒ€aœ,Ö—¸ít+ùàwsR>õS´ñYíy#DÏ“aôg» =^N>žÙ`F°-ð8š¡g%ÓÒ¶÷³Ò꼯cƒr²¤:el°IÏ¿÷îÉ€Ì÷'`fIÛxåxvilÆJ=åäh¾Û’%1)cõùÁ³4r&ÀÝî_4%Ió|o_àx»fÚ<^ÆÕ³RwJcƒg¶$æèÆ ˜ÂØàùºñg‘Æ,¸?/c ³¾­íhÒÃ\ì~ˆn€Rµ‹”Ì)öRÐFcƒ/ aFóã q*w÷ycZA@…pò’º…3Û+F„0qŽ[5gàÙ¤ùûZ’WtòßÇfç•ùó"äòYÀÈ€ pPƹjEQ”«p¤.]2݇4~$9D/ig¼¡Œþ Z!3el­¶ÓHýÝIz™Â\`iŸ[š£Œ þÍF”q¿œ‡)] ±°õ6øìûZ“žȾ`ó660w›XQNnk/·»ÿ{–Ô ÜÝ{J? -Kcn·‰«èæÿ^= C¹ xPnðК4îÇÎëq›óa–„]… Û[Ÿ8_Ñü÷làðÑmƒ¶2ž÷g]†2ß—à lãqKÒÏÒÅj—N 0ZôbýX®GR.€pîWá„€ ˆ^| \a`Ÿ'íyXÒy47`ÚjImŒÇ¦­7VÆùš¶’.š ülµ­6ú³iÇÊH™V(Pħm@Ž~†õwÂÃ}˜ ã~šö⨌’iEž¢= hÚz1e[šu6؟牰-‚;àÏ‚sá¼éQ„ù=òlOCýùiX[÷˶ ÖÊÉ7°¿.\tñ î¶jCÛV)£¿yuL§?~6ÐÛúaeœ¯m/Öʸ÷B{¾Íç³Aül0Ʀd Œ ÉÖ²Oß` ØÊØà–¬a(cƒ3vj)gü¼hoÁË7h[Y=el@ÊöaVõu³:RF}âl«ü£'Ïg¶¤ elP̸ÞÁ\âPÏNä)ܹÁ%/œïhÔwF¡gelÊ6pŒöú£ÅÖH)#  Xftñ Ø6ðVÆ¢b/nÊÙ7X)œÿ_egÌÓ0 Dá=¿âÆfª;¬E ±€¼UZ E ú÷ɽg§€ –JßpWUÖ9©ýÞ]΃ªÅÉà 'Å4yŒÈ*Ù¥ùµz‹;p¡¹±˜õà¡W(¸¿æ}¯€5ŸÁlá¡D±˜ < œ%ç ÷ý³^h`-9ϦµäÔ`G ¬ÓÆi1ë%@Oa1×3àÁ\rhªATßÇ æzv¸˜³ç5É@j1ë©ÃÅBÉ!(á´´˜ ¨-Ù§z£ÒâdðsÌ÷“Û<D‹¹Ÿö8x*¹¡ž#Í.Ùñ<:´0ówÃ×ežn£»æ¹ÚlÅÉ*$}ÊèŸÍ+DÇõ3Ÿä·©¼s†.‰kœm7Ëq ,\—8²Æ­bµŸîV²_0îÞ«åzZV‰/Á››?€MË׺®Ž8Èf±~¸‘ÓG-Ó¶¾8×[‰÷ÕmÄ÷ü%Qƒkš)ò ÏûZâÛ8tÍ Ç×CÎðX}¢ P” endstream endobj 105 0 obj << /CreationDate (D:20150701163320) /ModDate (D:20150701163320) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 106 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 108 0 R >> endobj 107 0 obj [/ICCBased 109 0 R] endobj 108 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 109 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 102 0 obj << /Font << /F8 7 0 R /F54 25 0 R >> /XObject << /Im7 98 0 R >> /ProcSet [ /PDF /Text ] >> endobj 113 0 obj << /Length 2837 /Filter /FlateDecode >> stream xÚÕkoÜÆñ»…`$(êh’ËgP;p]HÑ E# - îxÒÕ'R8R–…¢ùíçîòq’m}(òGrwfvvfv^Ç?ž=yþ6KOª°Ê“üäl{’˜$ÌRs’gE˜˜ìälsò.8<‹ƒ®®Fîø~×^Æ^Àµ†k ×N`Rû%ÊâgïÏþ|²Šã4,Ó Â*ò/Ÿ­²Ä×B­lÀŠàÖ!zÊXWòŠ<~€+0¿òx{ëPÖ²ÈÞc[‹ *¹§Ê½EÞ ý;AžLݵ‘)&‚û69 u´ëO²·µlÀ¤+‡+“ %ò, }pÕ¢<…ë_BBåÑ3øV^; ý,P(¡koñ§–åhÌíÝ‘%‘ÐÐÿ(bÿèíèéHýKrx鄇ȭpÒ·Ã’ˆ£´ø¿«?9®þØÊ2Ã"×mç_¯~'ÈŹ ‘Ð…Ü[aI¶¨[ e{žP⩈ë/p½v4;ÙÞS‡´©2ß<y)ZB>/}Kùœ£öµ:Dá•eX¤ÅXv磥 Übî»Å(,ËAÿN |Ç™»ÔùvÌ'à'•ÜãPÔž¿-=ìUœ"` ì*.™Ì[ñÍÔ²JM|ë/I„IÜö<2ˆ"ŽPúò@&³*L¼î®ÐeÔ°kIƒÇnGß»EÒ×a¬;ð€á[ý7.Ù#Æ›/°8Õ¤òùn‹½=þÅ)í=BãXIA i–Ù¢o«ùˆ„cG·[\ áç¾ÎáY7RãwÝ6 ñºNn4èÍox5ÍÁRÙ%ȸŒÒàD7)ƒ»xào@"'ŠIÐmǃD°î»–¡üuoÆÛè<|F'[¯eo«dnqìÙÛD‚oæY„•³çw $ð`…lç2 ¯Ót´ç$LXæŸsUŸFü^ˆøö‘'¥Bås*‘νX`!†m˜ôñÛ°$-Ê#rÈuþ=yUšç¤o:‚ð ‡Í _² íðH¬lLnø¼`wýl•ä™n­¡¶k—d“&a\~®pðˆVeÐw¼ ™3à§Fëÿˆ ò²z¶‘Ÿ-ß1˜ÞŒÓ ¶µçÙµ›E*½Ðm7OV•c«îmb3"—TY°¦)¦Å¬$³‚÷9»€Î|·'×@dH° Hç,\~ÁÙOõZÜïžâMx‡_ûš\eˆÎ.Î<‡„³WÄÎG“>JæÄu.‘¤Äj/|’ÏB‡…äbÐ>ÖÍ’7• ¸+v£ìàVb‡· ¾ö²µƒ5$ñàÍRR=¨›Âç+òƒ{P!x}­Y. V¥Uz^õ‚@6<ìd ùÁØ•òS¸U1 9šä Ü0q¨{Îxx­vİ/ Éü@[‚§lµH„h ½ª±`5 ÔkŽy9¨ÞÍdUÐ8Âqg`(±|ìï4¨"Ô•÷ºÞ;ËBB6+»lHž„Íg±õ^!;ˆÐGúóø"?Ò DcV Gn’¢==™ïi~Á¤úŽ$fƒ­$Æ@¾ÙYe¸ ›„aÈl×Ê^³!á"}¨B«4/òªß Jó=§¯pÇ{kM€Hñ­¥Ç³kN‡n†‰?ÛD|œ¿Ù&+»QªÙDy*ŽâÞ"Vè‚F»ÛyÎŽ¢».Ø*2븈ۙÄQJ¹”î…”ê 2rF=É^ã"Œ]ˆé'E¡+ÞfÀTa™»xäY<9’:åt&Ê8¶Æú[ x¾— ‡Špç•)ÂëJÆ5Ô0bš†þ€‘oÏ!ëû›Zäjr_{M•ò<?u‹7R¿ù+h‘ûJª,,Ìßø5ãÄ ½t…”¿—Ƶ ¼댆¢N0\B3ªuª â~ùÙÕÎÞÆ+k"_^|ÍŠ'3n<™(Ì’b±bDsM{e¥ßóõµâà¹ð~#=‡‡0<ÜKשaÙÒ«Úñ›k|ÄU¿—F§s8úî5Å¡RsvnÛî™îr˜˜Ò;yÆêùý·ËFÈŠüž®éRcUD5íØŒyŽƒoàN—ä‹EíM‰±ã[xü®o§†au}Lÿ]¨É³8LÒRŒ;êI®ŠS¥NB g²g„5cw­ÌÖHgÐÌrB«³±ý€\fÜ\>šª1µ &‘EXͦîO>’44ÆÆ©©áÛÍ«ò&üN t&¨ZšÞQFÙ¤Yë²­½Ë_áðdeéí‚ÎÜ͵”© þªy¦KU” œ$ø»ä@š5Í‹"ßï¨[î’¦QY6sLÌV’HÄVª8Ó² SôsÊ ‹ ÿáÞ,ügI„ÁÝ^JGÂÄ·¾~vHù?±AËîµe…öð¾Ñ>œWÖÒñqË ·üÞ¨cq¥QÅkÜêiáÊIÛ„NP޽k¯|ø‹ÀŠO"\ï(ˆÎ3s‰+k›-ç–Þk>ÚryÀµŠ4ƒGÎ#åìMzy&«ÊùK˜¡àK ̬E„ƒW^ÛóÛNæ0!çéºâs¯m 4œœ›@4>h£Y¶q®&áv9LE ¬Ö‚¾Ÿé„KÝ…î É£”ªWÄï<KÂ$²b ûipæý‘¤{•ǺU9ð7 ŽÒ|òwDÝ8‹Ip¥GÈ˜ÄÆ*_b ³V ãé«×èç"!êÔ³dõc~Û‚ûltÈ7lhøŽ"bmC⴫…è€ð>–¨ý*œEìÿÕ{ÿ£4^`ÚÈ8_¬“;Oã—nÉé2¼•á#ÝÆû’ap¥Ðý•oµ÷/÷w’ì³ÝÑôï]qh“R¯,F.<‰Ýû£¿‹+ú˜äÌÛ…þŒë5,a¼tÌ<,Îû“°ýëÖ=ù}NSÉ8”Ç|äsîÈÜ ¨ÿ×þ ÊQte¸õkëã6øyŒl½E^<¾IÕxºh5ªò0޾ø_þ¯ûˆ-­÷™Í¿+zÜÇAý½ȱÏþ´pnûû?ÊùЗ~6m~¦¹ÄeVîßäßœ½ýzÉ©%z8­²Ç~ž5éu?(ØVÊß¶`§ß#§,\m>oÍúŸSí„­vE§.]Û^K'8ÝŽ~H— —æOa)ŸÅ%Â=ysöäQ¢H endstream endobj 112 0 obj << /Type /Page /Contents 113 0 R /Resources 111 0 R /MediaBox [0 0 612 792] /Parent 85 0 R >> endobj 111 0 obj << /Font << /F54 25 0 R /F56 30 0 R /F8 7 0 R /F11 9 0 R /F14 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 116 0 obj << /Length 583 /Filter /FlateDecode >> stream xÚ}SË®Ó0Ýç+²t’ÖŽí8,yTÁ”°p[·½Ü&¹¤í½ð÷ÌÃnS ¡(‘çqfÎOd¾„GæmS5ÖåÊéÊ,›|Óg¿²ªµªÖ”0;RHFXt,>ô.7f_àI¡2–,g5ßtÙbåòËÙ¼Û]ÛZU׿Ý6ÿ&V…ðî/|CQÖZ ÷º(µ•âSßîŒN)žàìôᥠ%6p{ MXæ„öÀUÆߎ…jÄo6ú±(ÁÚb³#»Öˆ>aá-=F^°Ü>=æ…r"2ù¾4Kj„D0<’c/ÈQ¶éø¼Á’Ó3V §»ÞÛâG÷„,¥¬ZcXœCCnïQš=ü‘}ÿ£ˆq¤ˆù÷‡#q”¢*ʦ^ŠîÀrkÕÞ¢n8kÄZ“—©ÃûŒ©”&ì JÀóîBÃϬQ#¿AÔãU(š``Áñ<-ayï™ËdY&¿¯pÊ€7ÜjÚ¦(8'jŽ[F«9çÇéÑ6‚à‹™â‚ytý)œFÅ,Lú•Öôöª<Ñ !®£± í(šcÌ Ä~i E“m##¢–v˜¤±>qfZ·(­ù#ßÖÛMô°á9/Ýè?¶êþ€„ÅÊèÙOªÚÊ4- ({uòñ‚#Ü=@«”Óðt‰ëd)•#ªs!mïîž c…Ç¢B„íbs쫊µŸ±Â@W€ñL$)mJ æ×ñfBE¢H]W®± MÝTµqÌ^¶ÌÞwÙ_À«8u endstream endobj 115 0 obj << /Type /Page /Contents 116 0 R /Resources 114 0 R /MediaBox [0 0 612 792] /Parent 117 0 R >> endobj 110 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc6b.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 118 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 119 0 R>> /ExtGState << >>/ColorSpace << /sRGB 120 0 R >>>> /Length 44396 /Filter /FlateDecode >> stream xœ¬ÝK¯5Mræù÷+ÎèUå=s*ê°»0 ‚B¤`lÉfàßwe¬µwå Ÿ`ØÍ§Ï~÷µ*ï‘~þú'ýüãÏÿýÛÿzÿß¿ýóÿöŸþÝÏ_ýá·ë×u]?ßÿù‡¿úÏ÷ÿ:ÚÏ¿üö7ûsýüýoéç¯ïÿûÇßÒ~ÀÏÿò[m¿Zú)ëWé?úÉWþ5+ùO?8þLÕ_ íã×ZÏ_©ç¯ëúUûóWêùkJùתϟŸ¿—ö«~^Úùù{¿æçÅŸ¿ÏëWŸ¿“Ïß_õóÉŸ¿ÿ»?ò‹ý/¾Ø?ÿ—ÿöÛ¿ýùþþø_æ/þïøc-¿òúÉù×5þø§Ÿ¿¸þòçÿøÛø£=ï¿ö{û5ú×?Ì¿÷Îù+ׯXï?L÷µ0Ò׿ì¿û_–ò+}Êù»ÿe+¿ê÷»M¿û J£ýZùûŸþî¯(­ù«|¿ßô|I¸êÜéŸxUäž•jWÃñg»J Ï_Ó•~­òü™üú{»/áëówðó÷|Í_s<'¿þŽ·óüýóî¾?÷çÓþ›ÏWǾ¸ë}‹]¸…_÷¿òoWß·þmúŸü·©®ý_øÇáõ¯üã|•_%ñ—ÿÙÜú¯Áð×ñ[óο‚¶/E¶yÎ<Û¯”oó»Óü‡Ý]e9uû¯ioäŸÿáçÿùoÿ_[’¹öer·Nö9þæ/þ¿¼¿Ðëç/þáïþùÏ?ÿõŸÿûŸ~þüwú¿þéþòoþø×¿ë»™¿:~”†_ôßÿÃßý?ÿçŸýÛÙŸèûS¦Þ÷í’«5LüLÿ?ô9¼™¶Fñ›é£¶¨®™Zª5¡»39´~¥D •ýžµÿJ—ª.ª¨z†rSÝ_‚©TÕœP+‡Úý-%¨7UªÐ,ª<¨®*|×+P³w=ö7"ꪕ |J¿® õ®Êöê÷YT5C)PŸPΪ™ ZT»ÞZש»'èTW%{gëוU%SSU”»ªM¨Õ¸ ûîÍý®÷ÕPT«Cyž*¿®Ým¨('ªªê€ö7xª_Ð 4ì%û–NÍ¥¤²;u_ åTźÕU¹B#ÝÅû—»TvßÚWÏ)»‹÷¯šTvï_µ¨æ ú©¶¯`ÓÝ’‰2ßÙýJ¢2©©j UÕ+´[†Sƒ¯°–jÚ7Q웫ï_ÑT§ÊZ›}5tU½ ¨hf•µ6÷7yÕ˜TWYK´¯”ëÔØÿÚ´ïŽS©Cû;e­Ô­ÝBŸª|…ýkžjÚ¿æ©> u©_a•S­éVW%{–fwÕ©rA-µ´[UÕÔ‹ÊZá}5š´¿ùSÖ 7ëEÝ­hƒö¯r*ñö¯rª$¨ª|…û]ˆZ‡öp Ÿhìw!²^æVé‡ú…žd_ Ye=ɾ’ªðYî–ST+ÕUÍ~鹿]Q_TQÍ Ý#ѲWXüDߺ[ôÝw¿Ÿhí«Vd=e·‘ ÈzÊn³Q‘õ†÷çKeïz?[Q­uQÆ»ÞϨðYî‘™ïÌF "ëÑ»õ"¼ë¼[@Ñâ³ÔrªìQ…©*º[G‘¿³ý žòwv·"¼³²¯r‘<ºõ"¼këU¾kkõE6òØÿë¥Â'²–]d3†ý¿&>QÝ£Ñà+ܽ›hòV Yõ¶{¾CŸ¨Ù·t Ÿ¨íÖX„OÔì[:Uù ûŠ<Õø sª_á¾cEøDKMoáYë}¨óõ}7‹ð‰lf#*|…ûÝ‹*_a_­§_án¡Eƒ¯0—Ê?Ñ**|"k½ ~¢±gT"|¢±ï >ÑØ£`>‘ÍlDøDsÈEøDs@DøDh¡O-¾B½NM~"›‰2_aÿ*§ü]¡Â»^{6)»^Öã»^ö-½µøêè-N>r_å§ðÎ֞͋ìËz¼SöζºÊÞÙ­{V ²Öm«ºÿóâ+Ü# Q¾ {(*|…û*U¾Âý.DøD6ËYë=l–%Â'J»%:Ä}X/*Ê|–»•YOyÿ÷=š5>ËÝê‹lT2¬‡Y>l¶$²QɰÙÒ!öèÃz<‘J†õ¦"¼ë²¯d‘ºnÕ¬²‘ã°™­¨óVRÙÈq«ªð‰¬7]¼]²ÕŠÈiÒv?©ËE÷ºfzåÐÍ_oµÐ}¸WèïÃv£#¯ìž+å·ízPçIç+4ÛfÛ ˆÜŠ{„þ|ý ܼ û‡œm÷³šíðÜ+A‘ÙÛzd¶Å¶j™­±í\Df{l##u÷öxâzP³=^hOչѹ„.ƒn)4Ûã…ßKÍöØ:¥È³»Wh›tïH¡y`—RÂõ¤Î…®3´MηûÚ&èÛc…¶Iú¶µWj›¨o¯Ú&ë;íºBÛ„};åÀ«}=&Û*sd¬·d¬ÆzK¶†+2æëÑfkØ"c½%c¼xùëÙõpš‡O·÷x&0ûÃaým`Ì'2Ò4®ÅÝC³¿œø=ÕXÏf‰ñޏz9m=(0V¾³îˆŒùD¶†/rõÏk׃ºùëíñhàŽï{á÷Tc>‘mWDݰQ¸½çc±J~{·ç[v·Ð€šÊ…ûQþ¼\ø¾ÔèÏ Æ+âŽÀmû¾Ôè¯oÛý¤F]Ú[µÜÎ%4æCÅ¢’#£?/v‚0òôh¡Ñ–„ïóôàzRIh/Õ˜/¬gF¸½Bc>Tl§#2úÃb{‘Ñß•ŒûéôdW2ÚK5ú»Ûc„F{^°ýYɸ_ÔèÏ ö3£½/¿·í}Á~‡x±?»]rḣ ÆsÑŒç£?¸=b£?(v/2úƒRЫ§v=©Ñ”Šþøpa¬y…FP,ú/2Úûíí}ÁzO`Ì— ƃ±Þ^°Ÿ#f‹ížB'ÿ<{¾8ûóíõ´Àhï‹EjGÆ|©`¿'0Úû‚ñ^`´÷·÷õëIÛ-pfP°ýAÁx,0Ö£¶{hÌ Æk»?ßþ|ö{©Ù_`?G\8,ØÏ ü¼ßÝ^ÆzZ»ã‰Œù^ø=Ôì-³@dìG”ßCù`±ìêêý!öc§â^¡³¿ÞnO½›ý%ös³¿ÄzV`ö—¯æçø½Oó ͶÝjö—X ŒùpYø=Õü<ØÏ ŒùpÁx+0?ÏÂï­žþzµ…æxû=bßo*X/ ü|û½ÕÕŸo÷1Ÿ/ïÆzh±ŽA= a¡1^©ÒΩñ~ê…þBñH½Ð^¨ñ~*ÖËc½¶b¼)ž\Ø¡‹?ß®þ|{= 0Ö#jÂõ¦æç±ˆ˜ÈX¨Ø Ìσñ¬xqŒ¬Â‘1nß—óáVð}¨¾¯Šööpóó& ûU±ܰžë ëqÞ°áÕXOk׫û)mâûTc½°¡¿ Œój óùÀ¼_p^SìçIæëy¿`¾˜÷ËÂï¡æý‚þ80Ϋ·…ßCñdÃ|=0Æ[ým`Œ·:γˆ}¼Òqž%0Æ+ýÂï¥Æ|Ù#£¿ï~O5úû­°†FŒ…‘±ÞÔ1ß®·uœw Œó&ý}`¬7vô÷«?߯õØ-J72®ÇŽù²xòzì¿§×#RÒGÆõˆl‡‘q=vŒ|Ÿ¡§?~^¡1ßï8¯#f2ð„Ì‹‘ÑvìçÆx«c¼ã-dCŒÜýùì÷Vc=»ã¼M`¬×wœ§=Ý}<Ò±8ùóíö 0ö[º ŒŒxŽñN`œ§è–¡92úãn wäåÏ·Ç;b÷éïÆ|­c= pñçÛýEàêÏ·Û›ÀlpHœÙŸu‹4Žœüñûz Œþ¬c?'0ïw+’ý2KFîþ|½„žþ|»? ¼üùæ̤¥ Y(#c<µ³GÆÆxª[féÈÝ¿Ç ‡?~w³½ÆxJìû%û=Ù^c¿'0/œ· ŒñzÇ~OàîßýM`þ6yùóÍØ×3úÀï¥Æz²WFæý€ó6ÙŸb=&0ƃÈ>yúóÙýzš‰I¶í÷T'>»Õ8‹l”‘«?ßì¡Ùža½%0ÇC8˜ã!ŒžÞÑ-+YäìÏ·ÇÙ-´jœ×éXï ŒýVd錌ýòŽõ À8O5pZÕl¯-SVdŒg&â™c=rZU*õä|gb?&0Æcýy`\OÓæEÆ|y"^)0¾OdñT/öû)±33î5æcñJÑŸOögjܯ;Cd ë}b?ÿôôó–Èâ×2IFÆzËÎ&™B£½˜èoľ0ÑŸÎþ|û~ \üùöý¸ùóíû50ÖÓ'ÚóÀü¾°^ öþYB#c¼=±^˜¿7öCc½uâËl*äD!³S(„ŒBfo1Y³m‘ïpw‰P,×Z>ŠÉZzh²©µŠ"/f™Ù\Å>E6èÖÚZÃa¯]dãgì“‹ìì޶¼5˜7¢[Z5‘õ°(t)²…'±ív&Md½h·4 "°+,ö­Œõ‰në¬"ëa{e°·¬—é–(üלQØQäSUYO‚‚އ*z’n=D…Ï‚BXou>ËÝJ‰llˆ;èÏ…¡ø¢È ,Ö¢²Q^·åÖCŒ•GñEQã#gWáʲ-ôCLšŒ}L®:Â:…«Îâ¿E¸êlnvˆó½nyiD¸êìN᪳eÒCŒqÂ'²ùS_,ôŠVY·öº\ûñB‚"¢³-O‘ò†Å‹l”‡­²C ç̱M&²=¼aS0 §Y.B‘: :Äs³£°ˆÍ[¶ç4ìH”Èæ»¸»1VwX̪EÕPøç V¿yËÖ¯‡¿>ÄôÖEv‹_¡˜-é‰PPÅ}Þâ]ýx埀6XÌr+DÕ ä89™0Ñ\ˆpUÚPi!‰ÒÁ¢[O28t¢ÎÊ“þ ªá“þª™´íE`&MOH:~8û¡&õŒC] ‡v3é9“Þ©œ˜ôî´½&EF{„ì䑇¿ß™ÃyËžC3©6‚zãPBB{˜Ÿ‡bÄÅ?’žæç©Hº§Æ¦r¡Hqõσ¤ôù{¡=Ìß‹I]Õ,*€C‹âæ×#’†Ʀ,²OGæõ8<éæÛÝï·áI#Oó~á‘ÀÝÝFhlJ'´Wb`RúÀÍ=Zhm%õŠY}tßøžäñ0‹*0©ªúIZoI?O//À¤©ju`ÒT5“æ/$ýT³h“ª.4‡ìÒ‘«{ó~fR¬Ó~h Wdå0©|àî.#4“ú^–ô\ì‡Ú2“Z©™ÔI«Ôè/™^ìAéA«»¥ÐLJ V±meÚ \˜I}1žWöãáÀ8ÔY˜Ôætó¤Îhc퇊+ƒvÕL:‹CÕÑTÝž.\cRÒÀè/™ô30æËIAÄÕ“2)ˆã™Ê¤j!»£Úƒ&*Æs1Þ¬ÏfÒʆ öÓÔϤš™T×Öñ##i²+ª}<[~uñ×34æSA—™tI;ÅÔV1Þ Œõœ›ŽŒù|EФ؃*ŠÎfÒÉáAé§±R±ÇqÇT`?Œñ<Î7«}’æûEÒñø<ß¡›'U´ïSÍçCQmñô¤w ºV3)’îÆýÀ¤‚b eRÀÀXojŠ Ì¤q÷ÃaÄÒÂ%…FPuC{xùû鑽hEC{¸ûû™=4úû†¤4b/2é^`&íÄxD\¸žÓ°øI w­ÐÓÎåØ×›™ô-p}’ª]¡™ÔëEâöI:f¿‡ºzÒ5 jSc=­c?RìAƒ½xPàiŒ‡pN<2“Šá~{QÄŽùH`\ó‘Àzì :=íûMA£jv¬·ˆ=i}gШãu&… <=i”}߇×åI«†nž´é¡Ÿ¤L{=GìEí:Æó‡'-j-pö¤[X/Ìëiáý¨?I‹R`/ZˤE»'E² =õò¤D»½WþÞ÷k`&…@,nO]nž”h÷g™TIÄý“Ô'Åîž´‡Azoû|ŒIc=``½'0ö[˜4HìE³˜4(p÷¤5{?Aìû)ëÕ1žÇÀ+ò“Ôf·×‡óåEŸvv—ØÍ}·‘—'}¹Û õ“´Çæ#‘ÇãØç{Ã’GnWèåI_îûI]üýØy‡ÈÓ®Øë®Ÿ¤,ö}ª›'5i#ôr÷¸yR§÷£fR&[Vûxp¬Fè')JËŸûÁ^‘™ôÇöã"3©ÑWOOº³ð}¨‡'=¹ûkõ“ôkáûP7OzR’:ù~À´ó‘«'E™WèéIIîöDíýÅ´ö$ò“¥ÕÀžôzÚzoäîIEî OíAÕ{àVC7üH}¿nZlNdüÞÓ&V‘—'åX9pó¤/6žWw®‡LÜ—'µ¸J`/ªˆ‡Ú÷ÃÝëù˲£©—¥'¼ßÃÙçãH  NŸ¤í =•S<†\-¦Èî·jÛ¡‡2ú’jGeD²ß®SÜÃÇ”¨òÍëTÅÙá´¾A„ú¿ÕОb_JÔ= À<ŘjÛ¢ÎpþÜOq}¢ÚÑâ³ìOûÖDÏ…t‡˜à é[D6"ÛËâSJìÐæ—ÓË æ‰)3¬Õy ‹QN±w÷°Ù·SX´¢²Y·ú§›°e'Qc ‹û®Y+Õ=lö[=l·+L„oÞÖkU¦©°–HdwjDˆúa¬o5¤—é6C!†mi‹ð-ÙLãPÇ^Z·ÐtQgjˆœU‹Ï‚×o ´|ÝrDŠÓbÔujòÊêö žò´=«&ßuo§«‚3"¤ ±»R„«À–øßÚquÖ]tœ_RZãÞx0=95j‹XùBûdŽrò…ú¥ä²OÇ©Q%~"œy:Y˜Y'–Nò@Jg¤£Ð–K¸Û œž›£(™o³ã¤ÑÉŽ…޾<:ó`a†‹ëZ×…\9BæÁØáWÚÉ@‘f¥m/ [¦r¨;p¶H‰Œ qŸóŠ`ß_iîx¢*Z÷ÈÄûÓ…›wàD¦Òº—‘vÇ™_iMò ÉtøJ‹õˆ“:ÉìyÖ#Ú!¤ÁσžýáJk™_‰§«N6æ*±_»gIÊÁ‹Á–Ÿâb`ÔåAýŽ˜Rá“a¤DœdëJ7bÊueg.Æî}ÓCõ˜n]‰·aK'¹O±œL|vìänTOf~98ߨôÌ"ýŠhoùC„œ\MœQžJi—“[Ÿdm÷‰ÒÜJ›êO[Öfl)"?s@:OËq)äv>SB+miiÄÚô‹7¬Ò†ÞȺÙQp èdÃ!€‰3ÂJ› .[ rz…ü"­ Z(Qyr0ë:z¥uåË2î\ÌO²;‘“¦Å¨7¡u¸Ë¦ÌBf\èÊ•È÷€®ü`½˜I"[s­DNl.fFÉ]™pP}Ù«€6T[,µ|0c‰†¹O”v³/œ@>Y˜í·¤²3«ÊœäTj!7‚ÒšMæMQ.ϛҕ¬¶lBpð©,&ë`ÇMÇ´ÒÊά*sÐ/Q4ÊÆœ+kEDnD"Ÿd îBC¡ôü%ö&2wY°N@ü ?è›D¦ÊÎÌ(5+¹µ°› l|ðÌ̸™3Ã2.ÔŽŒmù a]bß½ç]ö9ÔØö½Òæòrõ´,H»xv(N ;-86*Ξ¶Áâw"ãXmi8–|ÚÃ’ŠÍ5#¯¯°µ×/¶³¬nþ}u{W·}§ýXk±ë]íi¡ŠýA=?¯gÇÒÕÃÃêS ¼üû°ñVäåaívìüpzÂÆm7^ͪ# ‹Œc÷Åî'ufØC±Ú`jKCXzd†õZǧöû¹ÚÄ!2ÚËšö«W›ï©=maµ°ŽÈHûQm¾¥fÆþ½¢bÇÎÕËÃÒGìãZüØóÛË?oÅûQã~®¶&Îç;aâäa÷¸ßïWìó}”¾R·kÇûQc¾RíÆQ{í°ji×ÔOØ=Â2óùlTíÇÚ÷‰¾ØÇ›ÍÒ~EÆx¿%|_§'…·„÷{zq<Òl×02Ã^ylüpñùtCX€ØÃ~š¥mQ3Ÿò^Ùa VÙìü¹ÚÓ5[•S7öË<üñ;ìEìa™(7¥–kãiµÏ7Âf#íE³_µ§Ým¿%®¯w”°Q{mõްqæ÷Ñ-m\d¬Ç <”ºxX©M,Õ•asÝnLuãzYÏx½ÓßWGØZ`´7agbf]¶…‘x²=ê6_T/û´•cqó´ƒ׋81¬¨ÛÀ>òô°ÔÝ^‹=-W·R jnýXþ‹¸r<Ðm`¯nþ{ØItuçzѰ"—j;Þ§lkàéaqèïÄ‹óI„5Šûåµé-¬12Ãúp½ˆ½¶ð°õBµ_ÏÃÆ«j_ï¶Þ öñü°XêöñîÄÝŸÏvÕ^Û{XÇ­žþyl>®^þ} ¬îôðñ4ŠT«³‡!,MìaÇ(r¬®¯Ú®êÎûa@î?üfù¶þå·¿ù±‰ÌüíÏõó÷¿/</=<è-Øï ²µkVÑ|‹‡‚«peÚÜc7QaPL*ªÊàÜTÖŽ>U;ß ÐiE5dԛʃ“f9ÔxȃA?§’ýÊ `¹¯ Qù Þ9õï\„ӻʫh"0ë-»¶sˆGMø"ú n5j]ªÎ œÕU6«V¹F„wf™nqý¢Úò˜ÈÖ&Љ2ƒ[ZRáP7ªvžòªyªpH•9O ›ÜW¤h1H¥–S¬ôÖ,]¦•­y°IJ*TEDEÏS8€ì<5¨!ó¾‹EÀ’ó©Êê”vDO”ø,µ¨PéÔK‹¼‚h[ªÁgéYåïúnåµ§ºè¨* ·¾PT(ù½e­p+¬FûV󠘩Â7oGõD“žà–|ŠUl[e¥Ú·’««2ƒ[ö/}ªxàËTU>KíªÆ`šVTƒÜ¿æ©ÉGö¬B°—d<4žO»Š*3@guªÊÚÒ™-ƒõ¾¢Î œ]÷ÔdP 1ßbàË=Ö‹š¬.Š [Ÿ%O•­4ÛšÕOŽ|–=…«A¯§&_®¦Z ÂAÝáo1`tïµgUá³ìoþTcÐÏ(ªÎg]e½¾IÑdÀ «¿´0“Û¡Ît ÍÖ E6†l¶-葉ìŠì6õy•WTD~Ë_ò¥B•I;è}ˆk‘ÝRk‰Ê'ðEÔø,»²ñ)TEðã©É àZP½N1õ3¦Ô¢Ì œÈZ)TŽ¡ºhfpî[:»Âí©É œ«©ìèN·áPa`ˆõ¢ÄGŽ®Â'ªvœBÊS)ú¥ÊG¢Rô[Ö;u;Ø&²D·åKÑä+¬@¨&lð‡<ô‰D™Ai¨ìhno Z~ËFQy ÑœªÎW` ë—¼Ríu©‚®rŠuÀ€$* ÞAô[•A?÷¯#j|…ÞTA?óRM¾Â\*¯î»Ê)®;uk½E¨ükSvQþ„‰P…ØB}Dá8WQ¡³`*Ñâ#S?å•ÑBŸJT½TÙUTÕÆº ÁWâ?Õ2Äô—&C‚îO´¨~bÞÑ[œB`Ëî¢ÆG®¡ê @JEek9ÃÂD0¹,Ñ‚ˆÏ‚ŠäßZ kðÊâo%>!ïoy…ây©ìîv$RÔŽte•ÝýÃfY¢Áð©ÝZ *åиÐ; ;ž!ò°ª]uüT¥FWu¾úœªÁशT“aMHð­„hX'²Ùõ°½$‘]uÃŽŠ<$©fUg0ÓlªÁgYIe½Ì°Ü"‘ ë eV!¶ll¢ÄGŽ¤Ê®¡ÂUg=ž¨1ð©Bø‚ñO >rÜŸZÔl§¸Ï7,XVäUWUe†Q•¡ÂcG¼Eš]…JÊ"²^tX ¾Èz',‹•Ë©Š5\T|%¾zYªÂ ±}7žòw¶¿ùS¨>m=‰h¹æ©ÆÖÍRE¨V>™Vâ-É K#j|–}EžêÔÝ*еºj2˜ló§Ðº¡ý[%쵦»¶ex‚@­W¡¥µãC¢ÉGî+ù¾yëÕ1µß´™”(1Tîþ_E(·œ*|dn*„ÙYX‘t¦eñ >ËìªÉW¿Gr‡&«‡[¨¯ÈfŸÓZqQæ#óRÙ>ëDÏuÊFØ"uùí_úÔ úR-¾ú§˜tZ ®¿Š‹)lªÊGæ©j|díªáÁ„K…°KË ,²™÷´™â[~fuZH€»mæ§ZP_*›ñMë EÍתS5Ð`ÈbË*¯Ô¾DTx4&Âz WUH0§¥Â»¶ÝR½m-Ô}çˆ&Ÿ¥ö@LO±‡E–aQb õþOá.LØõV¡JVÙÈá×¢Fµ¡B¸¸ÍpDÃÕU“ïzSL²‹`oZZ;ÿ/²Qó²ä·"|"•ˆÐ‹ÚQWÑ ZS-¾³žO±Ìزð3Qâ#çRe>ònYDhÁì„-˜fDÌEÕ†^š W«‚T|d«*Œl5õì¸õkD2íË6L•Œ9½˜Éí`óˆó+D<âÙ•ã‰v_—ÇÊ%çá–-¬'_[ÄåœQjî²ü\Êá-àòøÿ©\þµ[t¾²xìü¾å„Õ™ZÄEæˆÝ\K@Ä4]6ÌPNØß7Ø›vy€ð˜ ð¸x@½§Ñ;\=`Ÿ‰ôN#àìBú’À8°}!`1ðô€ý 0ê‚1 _œ>û×–/¤) ŒºY—­—FfB LfB›EÆá«âû>Í¥ûð¿Ó NžPÀ¾o5z®ö$<\<¡À(ÜFF@ÞÅ´šêé¯+°×U½&®uò„Öžªðt1ù¦š¿§­­Dnkh^¯Hhx> /Vh$¤Äö7 sãzMèœýñý €ÕÄtœêê¶öJÍ„ HÈ -2ú 5Æ ÿ§?>ÏÐËo×Ói¯[“,0úã„„>Ñ'öçêîÏ7{h&ü@úÏÀõ%$Ü{@i*ø½Ôü½È30Æo{BÕÃíIhr…îžP„9VOOOhbý±㉄ºª§ŸºR 3¡þæõÌt£j~^›ßGnže··ð”Ðß^þ|û~óh„¹…Ngh$´H¨ÛC ã…À¨NžžPe"ûãËÌ &Û;¡Q`&¤±€úÈü<°¹yÂû=/º¯ÐM6ùR—O–;»K ñlBÝËÀõq€Ùd Gd&ì±iZäá^-4?/êZfƒT°âÊ„TiáûR÷Œ]ÏÐèO“íµGÆx"#Åu`Œ—ó…ßK„ ùz2¿ÜØŸfÛ_Œß+_h¯ÔÍŸoÏw¾=^ Œö2#a^dO¨“K`†nlÛï©Îþ|ö{ªÑ_0!Oàæ sìûRw÷nï#/ºÇæçEx¹W<8Ì×»š l]?2Æ»óáÀÕ]c3¡QA¤žîVcûóíñ¢˜ÛCÛÖŸ¨™° ýy`Œg3êžÆ|:c¾í}FÓÀLHTÑž^~?a<÷ ϼ%WW£®b¶Ì"‘›'TÚó¹ÀL8Äñ†+\LØt:{@{Æx$0Æ#ã‰ÀO¨^Bc½ c=!0ÛÃæÉÉÅþyÖÍÏ‹„‡ùy‘ðMìÑãÀÙŸÏÒ…«ù{¢®u`~^$¼ Ì„d¨ûxúë­Ø×ƒ2’RÆ|:c½ 0Öû2ÒVfB1ŒWW¼]êæÞã™ÀÃßÏl±ñ}"Åm`~˜ï‹‹·7¿§:¹-¹šŸ—)îÕÕm©ÙÕÍ=RhþÞHÈx¸ízPc<”휹ºúç]ø=Õl¯PW<0æ‹Ù&N‘ù{"o`öwO‰ßOAÝñÈ‹)tz\Cg·}_j¬” ÷“š Ó,©qḋ Æ3Ñ_ÛÛŽŒõ°’€ÿ´¯$4 Ì„_ ß§óí‚´å™P ë1BæžÈLhfû‘‘Ђ ëÄžð¨dÜOêì¶öRý$ÄÛ ù¿,pzúó­zyB½4ÏOÂ=û½ÕhO ö;3! M¼"ózÄx.0®a<ýeABØÀL¸fçÏ#?ŸwEöºõãµÀhO™p00O–˜Póû¨{àóåRq=©y?>…&Nc=³`û½ÕìO±ß˜÷›-œG^[àá×ÛBùuö×[#4bZO·‚ ‘Ñ_u”” Ì„ï˜ïŠ'ÛƒŽó®±ŸÓÑ߯|• #WüÆ~i·‰›Ú×ó;æË±ž½óÎÐèO: RªÑ^ ×adœ§ê/FÑ1Ÿ <ýõæ¥Þé%›»„fÂœÇ Œó8ÈMë!ãÀLØó6™ð³`ÎÛF{ÞqžVü,À~P`ÞoÅÚƒÀÅßïî/ózÃx'0ö«û0r°ÞszøzÄÀ~JàìÞ÷k`œg6ñŽŒþy4#c=~à+è®nî}?Æxyâ<‚¸úõVñ{¨ÑžÎ†ßCõê‰xžÀh/wNÅØ×Ó§þEæçE˜¯gÏ"üîOžÞ=qž!0ÆÓç5?ïg×c¿bb¾˜÷£åyŒŒýæ‰ùºxp¼6q^30Æ+ç!w»{¼ÝD0îç…þ$0Æ+ÈS©nÏ-Ëë×ój¸Ôü>°Þ)îÜÏ]hïc>~ÿ?#òàý¸ï˜×Ö Å/²x=¿/Îç3ï<þÃï+ÄjmÍSzõ­EÍvŠc¸§ôê[öû<¥Pß²¾¯Úá­Æs%ÝNù#ï{÷Pz¹Kº²½´Š’n§ì>ÚÇÔŠ ß„ÍåçYîŸWdk¯eÛÞâº}µmq‘µIÕÊb¾ƒŠ"n§ðkN–¨|ËÚÑ}©Ÿb‰§jM¬ÈÆûQS¡ïbÉÈ·l^ƒ_õPÇ=‡’¶"”ñµã`‡ËW^,9øÊlZ9Ñä#÷·ôÖ|Jšî’Ÿ§üÚ¥BÙDK‹th± žõ-"û–XîôJò%–ü’ïS6[æU>r6JZY‘]‘(…zˆûzÍò›ˆ9Ò)î§ís1Y…oÐbnD¶Ž†;Aä¥Bk;Å;Ç»D(Ïi[§‡7Û Ë¨½•©ûêá›@YºSø&P–î-æñhöߢ¯Ò¢F•¢Ô̧:Z>[e>r—Ì:…Rš·!B¹L»ó1Îç-D¶GÞ,VCÔùÈÙT6÷m^Öì[<7cC"”Ù´m^Á¶ƒD“œ×)®c´É2\oá[²VQ„²{^.ìKƒgºµ6"”F´ÖFd½S·2c"[oBiÌCÌõ„£."”¬³4k"”t³%ŒC̯„2”"kÙ{ö’R/¡”Ÿ¥=TpµâH„ÈÖK»…›Šì×ì–6\„Òsovˆe•ºEÙøµ[zÎC¥õ¬Ç¡§ÝÍ"At”Åz«£%Âv¼¯`Kµ¢A•zj°D¥¬D(³iÿB4©yš,Gˆ’Y§PBµ³œÏ[¶Ÿƒ}×C,Ÿ×Që”íÓt+ë#·ksU‘åÁþ–—˜é‹ekÞ²s+Ýæ¯¢ÁG¢lÍ·Z¾nát"›ë¢¡eÔ¬eÙ&¶@e–C³s¾"ki±u&BÙ=”A8å¥õ¦¨ wÖ£‹l4lÎ(B¹7ëµ±´É°¹¤h𑳩ðMTûßbô2@§ld<¬g¡$_õâ"_ê¸s†-ŠlÔ5l½M„2”¶Ä'Bi/”úy‹åDPMdë¶ØŸ¡ô#Êš|dϧXN {&"”FD9ŸSÖ²[ j—9zk¡·ÀÖÈr¢Kû"²;|L–ðx ¯€ò:/-Î"Q¶Mdë ÃÒ ˆ¬›6›8”0£ÖcŠPVÉŽfŠž’`éTf™*Û²ÙzÔ´»ñcÙQ²I„²C('r ¥wl­[4(”œøV}J_¥¬B 9KÛ Â+Ø]%²x=¬KâZδQ³Èö‰öÞU6òŸ(©ñ{Øi=¬(óúR5¾Â„ò]–òàÐÀ sZo(Â+ &ëËË?±ÀK#4-DíËòÌÅdúoájµY¹¥à,íϡŲm‹IößBY³Åûo¡4†•#|ËÙ_ å Ú2èB–%j˜0G°Ð+ð\Y£ë›JTböÛƒLŠº%mkp!òU‰Õó¤´e»Å¬¶Y? ‹Ÿ­s]̰²6RóüÝQ1„Ùâ…¨ùÒðÅäQˆÕ<“÷AÔØažw¡]Ш4#ôº"8C ¬|°eP¢Öó³ ñÅbuõäÀ°a Ïm}Ð&zkàb8È¥¤5—Zˆ ˜YÍ…Öœ,ä´PÚ h!c×ÉÅ/gáG6Ò³-¿ÉdîØÌ?X¯ÑqQn$2‹! XAÌ›å;"#Xëb2P5‹Ed$‹<½˜“qªù|ÉÕ,v€dQbO&s!X>0‹1à0V`$Û¸°˜%YІˆ«'›¯žlö4+\H˜¯‡`vñ“ÌŸÉ®Õ,dÓð~Õý¯ÆÀ8¬qá°¸˜]°‹ÌbÉ8Õ¼P,A<üù°™›],v˜Å.pHì‡u‘õ=2‚].&k>íÁ:×|’—æûµ1xd~Lf¬f²môº§wŽãA[2>5‚ Ûf2l;u™ÉØ‘,DìÁ* íEàê·dfj\ÏLö˜Éîq˜Gì‡Y“á©ÑÞ%&ÃSw|©¡Ÿ×«+pù¼ÞŠÍb–Õ;2‹ Ø'0_í…¸z2x– \ýñ}†îþøÙÛ“„Ã’qØ…Éêã0“ÅÆf.“Å‹ý0cÂaÆÀfJH&&þ~Lí}ž|ð4# á‰Ìû‡IÄÓ“Ù3Y«ººí~R3Y;“¹žöÃ|LV˜Éþ‡'<Íû‡=ã°Eb²ÕÃÅë%sæë!˜#ðóz–ÌMÍb –Zí‡á‚9W|¡§?~æÀù“¬~ÕÐ8,›/O¦w‡å™l^ìÁê™É°ÔÅßsh&Sg²+õóz–ÌêtåõÂdñ™ŒÁùz†ÌdàɓٽÝ<Ù8sîþøÙBãzÈèOćå2ú‹ÀO²úk…f2dôâÁʳÓÀü¾pØ30“ùb<É’2ã‹}¼’q˜>0ÆC3àÀ8ì˜íÔbdŒ—˜ ]ìã™Ìd™ê'™úl¡‘L%Û#®ÓæîɾN|©¡›?žÉ¾N?¯7#³–¼%óž¡Y a ٚɲ‘ìAü• |ÏÉó÷HÆ£ÆxɶÅÅ¿/ÌT£=ËL†¤F{œ—'Û:Ídï(–%®~¿¡˜HàæÉª÷x(ðó|–lç´'‹Ì(6˜ÉZÑžf²ÒË»Š=³ -0“Å2™ÁvÁdbO¦Pì˜ÉJ™,Hñ`A1,ñôdÝÙ“1Æx¸0Ùž‡ “ ©ñ{&:½>ɤízR3)£Ÿ~’ •'YÐiÌ'˜L90“™Ú¾rdþÉ*N{²‚ñp`&ÛD1)±3*˜o~’ ïö00¿o¬Ùf²U ‰‹'#ïHv¤nžÌ·µÐO2_&»y»~’íZ²5“•c<-nžlãáÀï&»Q/÷;Ì÷£=-vÌdÍÓ“§¼íÉR˜ 70Æ3…ÉtÔÃoß÷é'Y9ÆÃ™L{y²“·=˜§`¼øI¦Ûªz]›ôœç“Éjã~«—'£xÛ“QW¬gîn&g8ÍdvŠ ƒI+[f²W㊟dŒXO\Ým†f2H& 9íÅ*“…¨‹'Ë´dj[U$+ûz\Åx.0“Åb<÷{mv?=<™#Ö#7|`ŠÝ±'ÛÙgž[h\¯Á˜ÑVcŠ—¿_$ó ÌdŠožÞÁâ]jh&Ûd0¥úy> 6>íãÍŠd«»¶À™ë‘É&c<ŠsÎêòIÖ¹¯·À|¿H!öb'LvøI†¹ï‡ÀLö†ñšØ“y4$#ŒþY#÷\Ÿd_Ø× \Ü%‡~’Iöx|%ƒ¡™LŽÁò§Ÿdšù N=üI–Ø{Àd„‡'#54“¥!™Îéé÷Cc0÷cC±mqâx…Éóó O`ô‡L6(ΞlÉ3ÙƒåNî§4;Ëó͆`m±'óoh/wwÍd«h/Äý²ÂEFѬFìÁ|LƘÉV§󞞬®÷À~½w$“ ÌdIH&˜É„p¿f2ìWˆ'׫;Ö‹c=œÉÔÄž ¥g~;ß çàÅû¨Å¤[l¬'t¬×ÆzzÇý öd¥ó‘ÀÅ“EY0ššÉÐŒãíŽûAœ=ÙÖ[ĸûë­Ø“iv;×㩎`R±ûÄyØÈÝ“Y°¡zy²"{¿§»'«ÃzyàáÉ„ìýœ~î$ü$J#ô“쨴À_ÉŠf =<е/þžLóÅÁ`Lõ“ìg÷g‡w0“]X²È¸^‡©W{0>ÎçFÆz²êD~’õÜß—:ÞOM¡ûW2¡ÈË“Õ\=pùJö³B3™’%“Vû~à°¬û‘ç“Ì&²'kG²µ—†mlFnž¬å¾_Ôž É^"c½gX2ãÈ8Ï‚³½êÉýÀaUi"wO³j`/¶5:^O]=Ù‹}ŸêåÉNÚP'_ÏÆÜÈË½ßØ‹[Vg®Wì¬%#ôòÇ·¸p¼>l¿0òòd'û~ûxÉN"O¼}§Û£aë ‘‡?Þ¾ÓO2‹ñ‰<<Ùɨ}½ÉPÔÓ“q = ŒñÀD{"^ܯœWyy²”ÕO±XdýˆŒý¤=°­}?cÚ|"òðd W œÙ_N;¸ëÓŠ}¨Ÿd6žWWî—Nܱ0m=TÝx½M»qÕû Óö"c¾¼¬ªŒzp<¹Þ¯ûiH ^ŸdíR†Èlß ³:}’Üã õ“ÌÀ:µïG/K¦ýñ²³éjOÖµ,.LÝ8Ÿ]h/ÅýÅî¸îùòãß<ÿ {ßáò/5fF±Qú„Ò‹ìýU»|%ô]ÕŽ•ÊùØËJå÷Ô°·$òWÈùTE(Sµ5‘c«íg‹*_} Õà³ÌëTÃܪÚO#²}ýj¡)‡:Î0äÿ-†¥V g=4Ÿü»Íx«óœkµmQ£r?ÅùOµÜ+‡X¸©D•ïì¾zD¯p-yªÛÎM>r‡¿ÅÐ SÙMµü¯¢É4)RDؤC ýBµð=Ñ 6ýÖü¤48ÄB¹;¥Á¥*|×;øÔà#wøú[ƒ¿˜±>Äð¨ŠàS™)Ö¥²>w^)*;PmÜvh=©î+L„0æk¯—ŠìWAý›·FzÒ$Ü¿€ÈÆ^;¥A9•1ïn9~«àh6dù«ßßÄ!žEnvTÔ˜Òà†‰y·Â‡Æ?[ý÷Ö˜OMj‡i¿Å„¾¨‡#²«¼Y8å!¶(¨s#Bè¾÷<Ä1:êß¼5Ó QßB„X¾Å¤ð(>"ª|–ûÎY:ŽfûV‡òó,÷»u¾úÛ<…oa›o¦@Øæ)[ûlvÆãÇeت7® ;;ôI¡p5î8[¯M>²\§Z›V%«Ÿ¥ÎSçš…°‰¬·ov¤þÐB/Šó­½hÏb)vDÖk7ëaEÖmu¾AëE1UN³­ûC\wkÖÊÖ'íÄ!ž÷k6=ÄB°ÍzQîÛ†ÙȪÙþ!žÏEJŠC“W+BôNUên1`g/i•Í÷P4C„t*¶•õ–Åï,pôˆƒ+ö*ôInb ÷„GL=& uµ{}¬²{Æ“q‘ö£¼Y¸1pBKiÃéa¹„ÌN3°;{’•‘‡-a +nÉÈÍ“œjޏ!²y ©W"Š =…¬ÇÆtîÊÉü"q] S¹+m³dL¼ƒ u(C¨Dþ•åEØ¿YY£‹)ØOrpàŒ¤²ñÁýR²ú5r‘¹½=yú\ØÉ–#Úѧi!RBNæåE¡_ä¶5&ˆÂΜVy#`'{‹hÓ§‰bJdQÁí|’™V™ˆ\‰ŒÕË ô<$<÷þâÄâݬˆ"V>¸¤ˆ/ÔsD|@Ë&dø/² ÌÎÑÛ“e¿ßlÌj•œÔ,Ûn ˆ,',å"´aÞBCq’™]ˬmÒº&ÎI¬®‰SÒBÜ’, r°!„m!BLi ãBg­ÄÏ)ÀInCàQ@¦f@]z±‡f\ ÌÔ%ÉëœÆÖéÅ:)§½®+S·ÆÑG¤DŠÌÔ;½Ë©xê•ÝKŠ=´ïBêqý¤váÑÕ·½nÅ…Þ?ð“:†GWßfvM[wZ¡›§~¹ràÁ£L #ö£Ó릨»?ß¾þ?©gj ¼>Ï7ghͽЎ^ÉS‡àèµ8»yôó´‡~35Œ˜È–ú¥…~RÓÌøù=z¸}§†Ï/п°.$^žšI¨ÛÑ@wo¡y?Û:Ld~_–ÚX>®Wàìß§¥/ŽÌÔG G O»ld¨æfïöl½®ú…£Qbn¦îÕª;»GÝ·Ð8:•*Ž&žöºXH]¢^žº¥âèçátñh`ª8š¥.Gh¦ž©8º¥žžÚ$÷À‰Gc’Í©#—Ç+páÑ›dG§Õ•í}êx¿êþ•Ú$2ŽÞ Ç’ºñh²Ó‘«§>±£‡jÔKú£öÐI¬á©ÛÃ4ñ~Õèo’ÍáÕ^'ñèàéå©lç5rñÔµ…ÆÑæluã#/÷hê±ÃScÌØë†eËA¢ÎžzGgO-‘rh½Gê 5—¶÷ÑÍÀ ½·™¿ºúãq´KìG»²%ET{ê/¤–P{èSn~tîmﯳ…þEžÚ¡Åfê;í¤öþ8¼ž¡yà󪧻~Žîeô'ù}Úiµ÷÷íy`\ÏÅrhDžžºÀŽæ.ºnã½È8z[¬€›ÚÇ ÅRDz_W`¯ T*Ž&žöºØÝü<Ÿ]<=ŽžŸþ{6„ˆS3 GIä'4¹×À~?7;«öÔ‹ÍVåÔÕŸÏÖ#Õ^—³a>¡I ãiqgj߆ñt`†î"ôFìýÙ}cÕ˜ù·÷x)0Cc#öоŽÐ˜ÀÓ¿öŸÞ[Fm´Nííe·})µ§–êvòS̓bÛöz§Y·ó¿Ýíéýÿ]_ÿù‡¿úÏ¿Y’òùíoþöçúùûÿQøŽî?Õú2µ^õùòÿ¸vêóáà²UªL؈ƒoõo¥ëÇS'!¨À·Fð¥û;ù–Õ Äk¯èG÷ËøêºÚõÒ¿RõÓWˆÁG;¨àKö,™ŸhÚ³T~¢U^²¿5~†FàBàÀ—ÚKó[÷·¼ûøþãÉm«µi óC| õ­oõ×ߺý;; Ãúƒ_zÿm}ëþ^¿´ë~©¾4>š B©<ŸL¶FÉJ…íð®O±RáG%}«Ú³¬ÏÑýoõo¡Ö£k¤—Æ·æëYæë–Ui¼üx¾…Π¢â²ß–YïðKó91þÉë~ÉžÅkûîMÛøÒ}Oû'+#~´¿ù/G¬w˜Ÿù©¿´¾•ÛKÓ*M^P±J“6kG8@ã¶ ®ýoYØPçaýv½ôþ[i}kÿbú­õzÎeωj}v8}wªÚ÷‘WëCÕÄÊc°¶BÝ*â—Ö·î¶ø[ã[£¼dÏb}'**>²…/% aB¸ƒ¶~ T8ŠY«ÓUìyøju¢Ò]SçP{ñ[k+B iš¥•nL>ˆÜ,ߪNŃ}¯|4-œjðÕ×õ’½žù¿ò·Rzi|+_/YˆäÛÆî—j{inyÈ6¾ÕËKö,Õ®—ÊK¨>ʆ~Ç­FøÖüÖn]¶€Ðü ¿eÞo¬Û´¯Åk–ÙZ®Ìo­oåüÒüñíTŒüRµ¦àÐì9+Ú=gc B·çìüw½}k–o-{Ÿl±ï8Öpk6’l<…‡oÍo%TI-P¹^*ߪù¥ù­fÏRü°ï͉ƒLÍf Õgp-~iæÿ·´·iõfY²ûæ÷Sìak \ùž9•° änð@h`Úò@µLSî\=Ó–t[ÒDpçü¨y¸¦…ðŠDå ZSèÁ¿ÕBƒ¦ý΂¤èG­dr瘙è¸Û"È"Óüȧɉªß‰Ð Ãeò¿a¾ýûK+ÃX¿ÐŠÉY,|CYje`÷›“ŽƒÓû¿Yù;]Çz6º^zo8ö…ðŒDÃ¯á ¾)`*FLg¥ÔÞ'¬•éx.> €ãK¾tŒp §úùÙ4QwŸ`yÀ™4Ñœ™–Såè/L¢¿°“ç=9ž€£i&Ï_gHm܆<㇖bµ ‰~Ë ÓÈ´ýÎ9Sè¸g±¿i?Ìk.§ æ°2£›|Q3-÷3fÈHm™ÚÌÔÝyð £gšMèdZ~m~a'™ö늌 “Lþ„Í ’çy=’á§j& ±(JçuLF¸I¢æÞʨ‚ ä—¦Ñø‚I ¡Y…BødpÈ32•’©>BMhfjþ<ü}¾9y¦‡ûM÷£®¼sžLË©ñNûy9þŸSZjš<-SqB@‡Ï0-õ™ã.ÃCWøX—/\%Ño-OwêÊ<§çÒ˜çþ\­—ŸtXÏû[öùÈgƒáj—DÅ}³ñ;}º[臛¨yH{e8Ëoý\…¿Å{< €ásfZNë ƒI´=—xúY½¾Ù¤V2!Øþ¥“i¸7öà +3J&-ïãw]Þ’ÈÊ3ÑÉtù<.Ó6j_èJ¦•©¶L þÞÔ=Oo…Ä’igš3Óö;Ëñ§o>áô× |yOi¿¨J:™~Gð‰ZꙺäÙá.>AãdZž þvŸ¹‡Ìt¾'] è&“_Û ª±2ãè¬F&ròÊ\ ´¦ÐÉ´=ÏÆçÏs0ÏãyÎ/H'QiB'“•îG r ÔýNxGûl)Ѭ™6ÄÐiùüÈdväcõ“'…Ö¢»áDœÉsé ÿ±¯ÿ#x°OæÙüÚbžÝŸpx'Ùáƒì+¬™N&+ÏD)ݦ¤DPi™ 0¿ÈUŽ­'¨yž†dõ3‘ç9™Ëò;}^µ¼A©:mOw˜n·LæpÏW+%*îPCïÇ2Í×+~y0Û¢*– ÷‘_L7=ì¹8i¢ã~ð…é¬)l°¼¯²òð ŸI-ZGl?N²¸‚À§DÕ…7|v yŒEQA%êž§¯ *ÓÉô;[\!@ÀS"«Éᇋž‹çB7ú* NîBOûÁÀ¨ §úÔý ›é¬¬XoŸíç ¨šP:. Â<K7“çÒUæëpyû+¨ºxÈææ NÑ?øQožC€V¢ßï)Ñ~„Üžð §5c=ž B¤Ü:Ó2‚û6PùoAÈŸUM4ýy`ó~å%oU7m4앨T¡ã²&ã'B»2Læ#Ÿ¨gêž‹ïÃé~S¤!_‰¬Ì>:~­38ìx.»‚ÌÈGVJéð¿NÔœƒÈL †kØŸL4›ÐÉ´ùliót<®Y}¡ ÂöéM˩톕g¥%š;Óò_¶˜Ëoë¶¹ßì—¼·HTz&«/§DŒ–hº0LaÛr ™Êà²å×:sÙChÿlŠümïB4Ai/ùÌfs]{;¿‹µ=4ès’DVC>úmm6g×÷ãÖíV›óbDl΄é8ùÈj»ï§sŸiì‰v{ë½);¸}åkSdr1!¹ iéDk¾Â7p Ïä9p]÷YH"+%œÀÏe3`ξ¿êÚ™šçy"”n™|‚§¼eß =ìñ¶|‡o³·å!é}|²Ùsß«ÅÛ¼LÇzBguýüÉYÉôû›B¼y™\¼§1h¢?ðâF«Ï>î çÚ6ø2ù5ïIŽï²nöc拨ïe:™Š“÷Çg/›ìÇóîÀ*ŸË샾!™<—Í\¬·8tÛö¹L¦þsh½vü¸Ä¡,âñþáð,ýñþáPÚ a‚‡ç´˜¨L&ôÐqÝM wã÷‡»êÇWïÏö ˜ð%o=NÁ¹O¸º'ú­Ÿ‰~o¦ž©ûó:Ççâ-B OAËw|½îTúyû1´SÑfogNEK{|¯o¡ˆÕÁgÞw¼ämÞÆ:ÄŸJ?qä9´è@Èb¢á2M9õ€ÅD¿íg¦)t2m¹Ód¨>2ñ¥ÆPLo­OòÅóô™Íñ¶ëpwùL. å­"ÄÏçcÇWèçcvL4j&{+œ!ð1ÑoMNdu—‘MpœÏt>‚ìU"—¯òYÜæ35#o“:™¨ù/Cð£÷q‡çâáJŸY‹yNˆg1àrzž>G€_}¢ãRZhÁB m«ÏsýÖÌ‘nw(≰ÌDöŽù|\ÇûLž Z0?H{&ÆÞÇWèO©HiEH¥÷†öNù[ާC›å­œ½·ç'‚>3-£Í°M+ùÉVÊëÏÊá '·½Mdß ÷o÷›‰–§ë ý´¯ã£-×öÈdï!ä}—¿?áæùø`A¿:Ñ$È­mïö2n½zjBïÝÎúb9KQ\‚ÕuÐp°ïñï„%ñã]ÜÙvèÄöN[à0œ”j%vJˆ1lŠÓ•Óâgü~áçÐÊïñCçuF÷pÔs(ýxDú9á{ïýbÆâiqT¡ª‚~uG hë‚ÝuÙèh¹šç‰°<_Ý3®æja–O8p0lUÂnt w+>å“a©Â[¹@ûa¼S}žp oË é8† ûþ˜×—E¶0V«09l5s¿®¢ì’i‰]$­ðêãm¬×¢<•ËPvi°Âc¨·™Æ#Â^;òƒcÁã£úò…U",6³¿ŸÄþ>x´ÌÅݺ«Ö!,à(ÏJG+†Éf6Iá¡la³™'õJðT^Èa«>Ê÷ã==3Âj3Ü?#,¶ †Â±Ìªa„½ZØíÓÇ0ÛÌ^þáhÿP¬-bv›ÙßO楿J6Á˜/l»™ÒÏ>¸Ç H<ÇòŸ±¼P|àÈûôh¿ §fÕ(î÷÷Û£A[ø>›ìÌÓ£…ƒåÓ£ƒLÚ3¢Ù‚ÀÑÃéž1EÙpAA€lº|ÿÌ_û$ûûË<”ëþ”]€¯¸þÝf¬Ü.†^àvìï+óu}?Ê÷×.x>ŽÉ?GÔ ¶,Ñ~ô‡…‡¢œ”= Ùß_f<E=eC_^KyåS/ÆïƒÃM¡hâ:•Û¸éÑ~8Ì—íú³Ìþ>2·‹çÅǹEØ¿}?%á $ý>FÿUZÈ

åƒüØÞÁÑðcH _×KW®S¹!=ÂðË ÍÖ—'_ÙžÂѾôOö`׋‡ë¶²½=¡ãŠñ[ñÕ<åWÈ@øÔ>ó<• òG{[!˜V"L­BÄ0s‡R,ÚOÈ,{{7Ù^Öï3±¿¯aö¾§ÄÖŸd>HßCV¡à~´oµ@Av²=«^Æï¡¬D¯…·²_‰'þ^„ÕÖ‚ï%ñoå]/¾îß×ý§dÆx@xº/ÚãZÑ>&öö1±¿O5£,„p”ÒSfŽ×™'®SF‚‘™ýû GèŠñ@æƒôè*æÓe±?¨ ßÃö(3!ìï“:ï.– F{_Þçf{_!;YvÈ6À¡»¼2Od^CÙß_8°CCÙêã ÙŽŽ÷•Ø¿¯Ãöµb~\Â1®Bü®„¬W…ƒ·ðt¦,IGû™x ¿ù/äwB†Ã%wÎï!òëŸy’áðϼ“]H7qEúžd9ü3?äŽôè ìŸq¤ŸIÒÙe; Óœe<„íûÈlßCõˆ: {ë u†ÎsYûj¬7TŒj¬TH— åÖ/Æó^Ùû^2È?cý²#þ²—fèC÷)9Hù~…ô§}V'×1߯˜gö÷•RÖX€Œ‰òQ¶öî·?ñü•iô'1ñÏî!ûûiÑ`~œ™*ÕdÌ—í3Á¼fƒùq¥î-eQ„)UÍþóaa]—ÈÏÚ£Ú?™•…ôý•UéÎ#~ßZÊþ~‘¹n|/‰m¼`ŸÊãë ¬•ác.Pý\Ü”­=Ëló£ÌãÊÏÕ©eØ*Æ™7ÒCæ¢büPc½²1C&FòÞèoúÿÌ}*{ùòP·sw†cbƒŒbŸ,µÿÂÃÙÈ™,×ïs‡µ1æÍO“úg´‚rʽ(ü>¬7ôß5d1~y—‹që ówaž·Íy qŽö»¡®±>ÐÐ? #Ê¢¸ƒ£p-WEG{Þ|¹[y:£½lsOìíSÈX6 ¶3ÌØÛ«˜ÿ7Ìï3¤G˜>d€„ òCûذž]Oȶ`=»ÆzAsk[ÿ j0ÄØÑ>¶WœícC.¼”µÝ7yCž²8°ù°Ï ~¯½?«öø½è¿•\3Ú¿_¶÷ñËXßl¾mz<‰1ÆÏYRî/势g… SÅïaûˆõ‡Æ­uck?[Ȇ6Œ2Ûü¨ñø³3| Zägã=aäñyÃx¢õhï1žhÜ27>+3Æ™K½ùa|Þ6Þwæ­ÜpýÆø þ„Ì”°¿ÿÄþ=†¬"d¥„îÇx¿aüðñÁû ‡ù†õ÷6B† ã‰6â÷¼¯Áõ æz†Â½\lÏŸñ{±‘Ù¿¿ýÖ'Úü~Ÿ¿Ïý™7¤^G0òŸ··ò)wwŒÀü£a½¢Åú{ð~£ÿ„Ì–p¥ãDÊ ùaþӱѢ„,—òu}Á¤¢E~þ>y ÚØ¿ç¡ìX¯h< FÙ.cþØÃ4£íÄëká Û‹é½ýÌL§‹ÈÏß÷‰çc}¢ïùs*¯¡¼‘ž2eÏ´óý>{Ç›)#&\`µÁ߇ñIÙÝŽñI¸ÿÛ±ÿ`¯!Òwxs츿#Œ`­ñºsOð›¿'…§3æã‘ó¿ŽñGýfÈš Û÷Òc¾×±> <•­=™±?Þc>ØilQãïaLf/Ÿíì/dv“‰ÌÈï}ÞÜÊ×OȨØŒð÷`¾ŸÙú‹Ìn–Òâ÷a¿ ³ÕG+¶Nî7#ÿùÍçâ”C^õâqñVvs“Ì=3æû=d°¡†$ܺr‡Ñ ÿ~Ìß;ý¹(óæ¯%î§ ÿ^ìdöúù2ö§{¬ÿv?;¦ŒüYaù•ÙmS/\ß!3·žïÛõû„ÏÍ=3úË>âïGÿ˜ÙËkÄûGØG”Öó3»­MÈC3Șß#æÓ}Ä߃þÑ$³ð=›Øûk+ä3›×+!{×]½¡@nêOÈè {û·B†í c´ÿ0÷×rÈ ¶6˜Ï šþ$îGÙú£Ìs+¯+½×§˜ÿ ˜‘e>HB–ﹸteoy8Ú-gêÅÃãõðΠ í›<›òoøí²×¿X_ìo^®6pb5wäñ´z”a³ƒñ(d…¸Fþ^¾”U£Ì °×¯8Š6ê×á| 9%dÒ(C(ì&9”J+0p/!ˆæ<œkÈZÿ9BÖu ý!Só2b¿v`þ8^™Dì/د5†±d:æ“#æ—ÐCòjÏ[7/eû^G ÙFô/™ÏÎ<œèãõy[ùZµYÁM¹#=æçÃϤÚkÃz$ŽÊˆóWP ñj€üa®’¹ÀáíÍÀ~qæzq+wåQ/Fz´Çc¢|/0ÚÛþAx^låÛB†ý‡UËBöòIlßç w¯ËJâ~Œ—!3)ìPó è½z5žd/_zÓhÂx5íÁóó„¢L¥°—Gïi£þõøž^_(~O˜}Œö?¤¹Œ½þ0ˆ‡²–Âþ÷Çú( ˜Jv{yÐåÕy*{y„l$d1Y1±jø#Ôωþ!sÿÖÓ'æ+¸Žñ4i„ýûžìO'æ+ƒ‡ª ”iXíþ„ §°×§Ì~Xþ÷B¯Æ«Y^Êîc5Ù_LÌ7Fœ·‚BŽWËFöòõO[µŒû½ü37xlàW­JöòÎ ×-¼ß‰õÑç™!*ìßëæû…ÒWËÜ.Æóñ~'æV-ãºÍ÷F¬_NÌ?]Þ(3*|êÅÃã—‰õÓ[؈½\‘Æ7ýOÈZ(âx5û½ü¿ß‰þ)óÄï=‘~Ò‰¬eë¿'=î·óˆëÖÏè&ú³W–úƯ[™óv2ì×LôgVíâ~{?“©ÎǹÇýé1^žèÏLê¨oeûfúŠþl†mÂÄ~¨U”ú#áöSBD©@cGØÆG™;,ÕоAaÇ=Õ"¿Q”'òß‘~΋²µO³Å÷0l¡Î«MäGã5ô×ý[æJŸ¶Œô3®{ùgnÎ;î÷÷‘xÔ‹‡3ÆÏý]fkfçzÇÄz§ðv¦¬/Ö7­šÌà~ñR.×ý÷Ïȯâù+®7<Ÿå‹ùWfûÞLn©‘ý}%¶þrÆùèaûkÇuWÓñ×>É6þ›1ß™˜o͘ïL¬OÎí-ú[aøÝ•HïåM)ãEó¼¸ßÚ§6ržÎü>Ê/±—ߌïû™+"/ç÷û÷°¢Âþ¥ðvî‘Þ¿˜/Môß3æ7Ó•ý„Ý”oEû…ùÜŒóó·—Ö …q?Æ× ýsæÚ•Ýšps¿j:^ÌìíOâU”7=wðtÆ÷±°ß9c?la¾.ƒfìçõ;q¹¸Â¤õ ;¾„OðPöò§äPîŽ1ÞÿÂ~è<|ÿ«„)âËÖ¬Ø[Ø]±ÞGêçCV ¯CŒ7iD-|” òÇüo¡·bÜd{_™í}­8 ]/¶<[äoíÕ¢¸AžÿÙñ{íû±?³/g̯úû¢]¢2-#¿³3c?Ó~V ÞÊö>ìgŽàîŒöx¡?ÇÏO<ÿDþó([Q!òpn‘Ÿ¿?Ê8/çéÝF°q¿taþjÿ{­}žîæˆö|¡^άÏè…çÅ'³/D Ó=íÝBÿ½(`lã•ç5úËÌé1Þƒ&ObšsÆúØÂþâñ|Ì_™ÿ…þ3<'éiÌòÄú ýŒBö¿7ñÙ™Ñe®¸ýéÂþ[f¯Ñ?Р{E<ÇÂ|oÅùIšpÛÏìÁ+3>gÔ?ô_™+Ò︿.åQ/ÞÊéOäçßcœ?\˜~L{ÐÅñ ôr„ýýS¨·@3Ç‹)®[ÿ²b¾¥áS.{#ìíUD[î?¥z[¹¸Ã6t×Åþ=lö×ýÁŠþc¿¶¡Ø?ƒ†p«Ê£_ ›Q”çÆzœÉåà÷á|iæ8OØŒV²¿ÿÌ[ÙÿÞèO¶7d%„tŒgW^EÙ¾‡¶~P­a±‚mþ´)>gþ¤6>ÜçÛÐG®SÙm> ¿Ú[1F~öý„øMŠMäxÑ65î?Of´»p¾Í/¦'ø8׸߾‡ëSëq;â ]âö$ÛøAx¸M+úŸ=Pq>ocþ’ÙÞÿn\…ÆLym6¶74–-Ö+7ÚÃÑî°-‚°Õ‡ãïø û3{ðqÆøncüoFäoíßÛ£ñÿŽólã}Ñ™d«ÿög ²¿ßÉñÀÆx~ÇþÁö…ÿOðJ6²ãûãa¤ïXÿØßï°½¢ úŽýêñüŽ ðön/ž¯Øˆ·ÛïµogÉÿ±Š"\a;‹þz nK[È6žÛas¸qþïe¨t0ð€ -ú{SãXÊ Œú|0þ<ž,PO¶öôÄ÷ùËV¾™í{:dªÕåkê(+3Æo!blí‡ðvÆxó`}?3ms?^Îø~ÖoNĬï e«O'Ö›¡SÂlÀ¶»X/1†ã<‚ýïñ;¥Œ 4Ê ehWÛ÷m¸ãCa\ó¿õç±øû¿fYã«C´¬9˜Ѳæ£]…Ü4F-aYãßðkY Ëš—f&ZÖÉEî„eÍK#-kÊO¶¬!uØÒ쟰¬É¤×ÜPf2,kÂÜ–5A°¬y© ÍLG®½æv/¾NIËšDn£âk0´¬ù¨¡• –5/ÍL-YÁ„e·2¯eÍK3Ór3Ø -Ú¡ø×OËšöÎtÜÈ$lpNZÁ²æ£¢äÖ%¾Z@ËšDn]K—Mã” ˜£øª-k¹AˆÏ¨hYóÑ*B+Óv+Ÿ¹Ó²æ£#×Îü–5‰üy°ºeÍGÕ>|m‹–55¹Öw¦Q2-Oç+X´¬a¼-k&M+ZfÀÔ–5<©EËšª›Vx+FËš`‹á=-k>êPË4üLNh¼4…ÎgoËž°¢eÍGçjÁ²æ#˜dUZtÒÌÔÜ(ÂG8´¬ùh¸ýì^`YóÑjB3ÓvK XžÀ²æ£òy.°qi“ÉèXÖ$j%Ì^š™†Û1ÀÁ%OÍ*´3-˜H4RZ™`ñt¡ó‘¯TOîGòfRÊ –5±¶­öLMhf‚YŒEm#‚Ì,¡b=–5“§aY“è·åË$w.¹Ó½ !7Yðñ ,k&[7XÖdr oÝ`Y“© LV'íLÍŸ““N«† X5¼t2Í*42mIw!½–ž7hãð’ÿNo“aYæ °¬Éä– 0QrIæL; ^Ꙧä²N¦íyú ,kÏÅçk°¬±M·°$øÈÞX"¹Vw¦>3 ¡-¹ÉåȰ-Ú?„ ì^Ú™`ñÒÊÔüé>¡eÍG½ ¹‚óiY“È-|NAËšVš™¶çé{!´¬ Ú4ƒ*J-S™ºß [!X ,¶ƒ°¬á^4-kiJËšÅv –5œuѲ&(ŒÐöÀ²æ#«Ÿ’i¶LKÒ­“é|éhYól¼u£e WshYshdËš0€eÍGçò;=&Gb–5‰šRËÔG¦Q„V¦)é–ŒžŸ°¬y þb`Y“¨?BMHÒ ¹ö['…aY“ie:SÈ-|”‡6=QmB~§ù°AŸhÀ> þ„eM¢%´w¦“ÒAˆ?‘›ÀæÇ{™DE¨¡ž ¦A0=H–5fÐ~²&‘•ÏóÎÞù©œL#Sq_†eÍêX%æ{¢æ"öéúÎYþÉ\¦§óµ$l'ZrmW¡ "ý/õ|úžÉåî}m–5™N¦Z35—Ð÷=Ql±¯1XÖd:™†Ð¬BChg‚4}åó M?xm»¤=Úßž_”W„eÍK‹BõAfS±ø/ÊÖ¿´2AÄþ¥ö Õc_8ÓÎÙú ®yo‘© Lef²÷@ÓjXÖ$j%S„šÐÌ¡ú0·™.b¶¦AúçÞ/¶ÚY)~Ó‡Bõ/¹À=¾M7–IdA¶>h>B»Ÿ?aY“hW¡ÉlHØPOT ~ß~²&‘‰µ'rI{_‡†eM¢~^)|XÖ$š5Ó~„šPÊ‚ЉF¦R„V¦ê¹x»ËšD}g%Óœ™–’ä¹%˜€ ýQsÑ|XC@æ¹ÐJÄ[êDChÖL«­L°ðÕóA'‘Ïãl_£þ„eM¢R„V&“Ûf| öõ*ärþ>c‡eͦA¶¹mHè—Ÿ°¬ÉÔ?‚ÑÀG¥½Âø°¬ÙT"€eM"ˆæû¬–5›{ذ¬‰l€'.pkÆÍ´2m¿ÓÇŸ°¬yÉgD™äZ¯BãæaY“© ¹Ü½÷9°¬Ét2í‘é¡þl>*Mâ÷õ',kv¼LJïõ“iÌLÓÅèaä}U¢ßÞis§–5™v¦Ò3UÔ÷d˜«drQy_3†eMìíÀ²&Qß™¬¬?šžoÚ·à"Ô?r1ŒÍ9 ,kYû™hgj%Sw¡z¼wï«YL´2mI·]&oÓW<_ò~,‘•îGõ’t]ÒQ^Ÿ4ü· ¾Ší¿™üç',k2u¡õ,k2ÍLe¾Âø°¬É42µ• bûACÒ%±}XÖ$‚Ø~Ж§ï”}ÕGV?É}g]he‚ؾ¨aY‹†°¬It šÏ;Mð:È-k2L”ÞÚ™º§ó5XÖ$Z=Óoò5¹ÃØ|ô ‰ i›Ÿs%‚ˆ}ÐpÑ|Xùøœë Úùœë ô†°¬9ÔžÁQL.[SïY¹|Ô›ÐÉ4%ÝêB+Äèaä½E¢Ò3Y}™4º‚œÿG¦qþѪBrm¡é¡þÄý?ª;S[™º ããëðƒV&Œß~²&Äå_šB'ÓIt(,Ô«ÐÎ4J&km}Òû°¬9<…öé0š–5gÓxÊg!‰¬”ÂbÅÛò³_³™R…N&¨ÃûN6,kA*þ¥•iÔLSÒYé~´[¦#éÎ:AÔ?‘?V)ôÿ¨¶L­ L]r™Ü¹¦_K–5‰Îú(¤ù_‚ü¼oùÁ³&c+Š]°CZßâá[#x]WüÃU»àÎYA aé‚UonŠi}– ›ŒóQ<‚k žžÊBMÌÏ+¤Ïè=ë2°²yeíáe“±CT~2«1gUŠ[p-ÁÝfuòs¼>¤ô¾7ÀçUÞ¤ð~àè‚“²ô¸ ï-Ìm¡KJê'„büÃ_åÜ8“ÐåµÁ:´äa*BiýN‡(*ë'œŠGp๾¤¯Á-蕟1Ž<2”ð@Q¾óGž“ç‡Lo¾—`…š=ÚNˆeœ‚­*Aÿp&›7êì8Å)¸!*Æ g’_¤ä~B¤E‹DAý„G°êÕÖ—`‡Ö»¯¿Jë8†à¬‚ÔÜšöä´NXŠâ¬M°iV½*jΣ z±óÀ\qñ›ÑžP²qR—Ÿ>ˆ³É:õ-ø(÷¦LÝyú*@÷ ó©™ ÚžÄ¥\¼”+tìé;sÔ™ÛŽ^<.ÞÊóÊB'çDa–#Lú—7çªú¼LŸ€ÌÔÁÈe(WêÐÏà¡Üº²ëÎ&Gy^ù­+=ué?¾ò;x>}!à ð1|s„¡Ow!øæS‡þ!»ôë3ƒ^^ø(»Žpæëz¿®<qiôÍ)¡cGßœé+÷Ó7…ºÀõóÁ©U™:õ/S§^|s2Ï‹WWö÷ñ2t„—r}”ò§â`KèºÒ7§Ä¹rúæC'çüé›#<.¦nû!Sýåö×WæiÊ¥^<”ëÅ ù±ý¢O@âþ\Ü.žÊ£^|”©kÿ²?gDßáqñÉ `á­\ ãŽsÊôÍF~t ƒîðR¦.:t*蛓™ºædtâe†¯zqá£\ ›N“3ú$®[¹eêÔ ]rÄåÐ7'3uÏékß³CXäý(SÇœŒ>‡²}ÂÐ]nÁ§^<2c‘;s=Êíº¿ß¼•Ç•ßlOèB/è0¿>=þþS÷¹½¼2Sw?tžé›“¹VåÖ•;Ò#®˜¾9ÂGyTåÕ•÷s±äݶ?_šÝç\Ÿ‹›rßÊc)Ïë~×QÏ|¥ßЙÞÁ§\¼2S7?óQ.U¹B—š¾K/´Ãõ úæ´×—ˆºú‰©ûŒñ }s„·ò~’®4}s2Ÿ‘¾9Â÷uä?ƒ t©·Oߜ̵]|”t£ëËKyå=.†®5}q°æÝCw†¾9ÂSÙuŽkøÂPÇ71u¥_öò¯ì/雓yåU.î/åÝ.Fþ+øìÌ/eêPÓW óé_nÁÔ¡†î}s~ùè2×`×ñNLeúXaü |]§Ž2˜¾9ÂG¹áùâ›óê(Ó7G¸+ï¦| «<^™©û›¸¶‹çÅG¹Q‡¹;t—éó]‘Ì^~‰7îï/åƒôô5 îïË˜Ï e¯ï‹ã1úæC÷øõµq]ú¾%Ô¥OÜ¡«|^žè OòØÊk){{•:ÅÓ7çcêÔ'.K™:ÍOeê6|å×/Ð1f}zuŒé“BûÄ¡k\Éçæ•ý·ðQ®åâ¡Ü®ë­+÷çâ+ÿy¥Ÿ#é.Ó7'ó‚î2tê蛓ÙuSŸz1ò£ úá~ñRvÝãÄ­^<”tœé£„õ…Ì:ÍôÁÁx`„Ï}s,2ïý¿Å·àyñQ®ÐEÆü†¾9™ûsqSÐa† }s2¯¢¼Ÿ‹¡‹ÌöëÂãâº2™ËÍÐUÆú}sFèRÒ7Gø(ûûJÜ‹ò@~ÐE¡oŽðQ¦.1}¨°¾yß¼•t“_ß™çI:ÆôÍÊÔ-fû¾C·˜¾\;t‹é#´C·øcèü²~b÷ÜâC ynåµ”Ïuÿ./Ö{蛓¹Ü¼”ëQnȾ]ÐÕS8|lty1¤oŽðQÞåâq1uv³oÎŒõjúæ¯ÄôÍÉì:£…õ‰¾9¸Žñ$}s2ëzèî¾|]wÒÌW~ÔÝEý¤oÎÇðÍÊõ¹¸)»fèòÒ7Gx+Ïçbêðò‚®-t¶è›“™º»¯Ì¿d.ÈïõÅñ÷‘ùºÞ®ëÔéýø(ü¾>;óâÝ.ž™1É\ÛÅG¹e×ie gøæXD~%¯zñÈŒýqá£\¡Ó‹ýúæS·w‘©£‹þ¾9™½|ïëú©™1ߟ#|Eпgöú™º»ŸÒ7gÆþ&}s·ùçõÍÉL\¬ŸÐ7'³—gæÞ ^Œwé›cñï‘ÿé™1ß·˜È‡ìõ1öé›óêØÒ7'3ur_žS7÷ãÛEöú¾ÇôÍÉL[¬Ò7'³•§ENòèÊ뺾¶2um±_Lß‹Á=3úËÌV?Wáú}s2»njfêÆv²ëf&^E™º³/[ÿø1|2—¢ì:™¯t˜3OèÜbüCߜ̻_|]?+1}s2—­ÜúÅС¥ï t˜#z<|s2[ýZíK¿–òng]Yúæ|ŒùóêÔí¦oNfk3»îæëSþ&ó(/å™ueé›#¼•O¿:°ü=ÔINÜÛÅÔ=ÁGye¯“ãWúæ| ÝÍÌõQî×õq¥ßEùÈý=tg?ЩäZ”[¿x)÷Gy ?þ½ýÕ© žå⮼¨CÛ‚¡+ûæ¿çÅе}}z¼þ½Œþòº^ߜ̕:±+¸Açµ{WÞÐ]EûDßa¹NÝãÄ^Þá;Aßœí-}s^Vúæì8oAßœ¾~ôÍÞÊ“º¬ñ¼u±}ψdûóúæ ^íÏë›ó1úÌ:«ô Âúmæù\<•WUÞWú#Œö^x^|”© ûò¼ò£.,æ#ôͱ¨8Ô´ÿ™ËRî×õQ.¾®{ydž?”ÓýóúæoåS.î‰é›³[ø" È\©KÛƒ·r»®÷Gy^××Í]yÏ‹OfôÂK¹A×ûïôÍÉÜ/åâ¥<ñ÷`>Mßá£| ›ûúôœ•룙ëÅþ½†Î.}s„ÛÅGynå}¥ßWz/ïÄ®‹ëÏôÍÞ™±^jòËìºÏ‰½¾gÞÊ^þ™© ÜÈÔþx^|”]‡7óVÞ=3ú³Ìõº^·r[Êý¹ø(+ý,ã:}OП /åýdFd±“#¸]<•]§9±ëgÞÊ×éã‚ù0òÇü…¾9;Öw蛓Ùuœ{{ô2u¬wøÀ7Gx){ù‡/}s„3ë7û¿ÌÙ>0ìïï£| £L_)¬o~ŒõMaè8Ó :Ú'ú[úæ ¶ðÏ뛓™ºÈ/[û&¼•û•ÞÞWæ}=o¯ÌðÍÉ\êÅC¹åŽô¯ÏN?Ê®+¾¼ôͱ˜ÄA¶öIxf†oÎ úæœðá£oŽðrÆü•¾9™]g:ö+雓¹—‹ûÅKy>7ekNè´Ó7G¸]|œéËßœÌÖÿföòá‹Ýñ3¸ŸDß‹Æ{È®Ëy*{}ϼ3Ã'áLîçÒ7Ç"çÙu¾3Ûï _#úæXÔZ ÎôíÁþç‰óôͱ¨¯x¾—2{ù'¶þ ó¾Ò[ÿü1ö?…§3}2°žxNøŽ`=Qx+ûû ]wúædvÝõÌSÙÚ«ÌëJ¿®ü]w=±O™¯ü÷Qþ}¿ÕÂÄÀ>PTÆõüû>”ûÅKù·¿0¦oŠ÷ßõ)Qß}}Ry(ïrñR>=³Ï_…+šñzÙ~O ß"—òï÷¢ŒôãåãÌ¿ÏBëÓâù¾?)\嵕÷•þ\éýïmñþ|þ¨Ü”[Qî×ý£*¯¥ìoú0Q_:û7Hþשìõ¡³=„dO}b}¢=Â^ûßûcðÍI¼ñ÷'ö÷ë˜È+/gÌgá›SŸ8ÏßcŒà›#¼Š²¿ÿ—Þwf¤§ÖÁûÏlåë¿ðÍ`Œ ÊS-ìewg”7Äw„ËPö¿?±Õ¾ÉðÍ^`ú¦xÿaŒóÅðÍþ­(¿í-á3 ßcútx{^KåùPÈéÛ÷ñ±Ÿ/UžÊöþKìÃ7§–>'oâßñ‰°½ïÒÂGÅÛ_ã×í{>ÊgdvI4eË¿s|™ak3ëþyÝoßOfû~2ŸÇëëÐî×õ~”í{È<»ò¾î?–ÿŸ™Žòxy <W{Ÿ¯OÌÀûOì_bë2[ý/OßeË?¾Gøæ[ |”ýïÉ,éþžÌK¹e¿™¯üÆsñ•ßlÊ«*ïë~/Ÿ—}øæØ¢öù ßœY06‚oN¢æŽyvwÕ˜Ìs¸ãÆažcfšîráã=l?ÏŠÕ6|Xb ßœLŸC|s°˜¾9/¹oN¦!´3Õ&$¹À=Âç=ðÍ ÷fœ^⛓èt¡õ|s•)ä® þýÁ7'ÓÎÔŠ?ÏÇŸðͱEćÔ2 ¹6à}ÐHî§®=æéñÑ™àvPèýsÜíž%Þ²¼ä‰Ì©ã£*wZÉÔ”Ž|¼}Ê´~ Äý‡¾9XûCßœðG†oÎK>ˆžœSÀ7'SËd_@¢a„'À7ç£îéâ }f‚ŸÂK'“éz?ïæÎ´ŠPr'øoø2H¢ß·‚¥ª?ôÍyÉP•"´2UÏ¥Ó^/¹‡·YðÍYõM×ÝaI¾@’h6¡™é·å[íÍÅtõÛwÍÝ|Ö ßœLÃUý™šûAÐÜÁÑ ÑÌT!¿OwåM[ú( hîãýùJ"Ózÿ¨»>Þ˜Ÿ¹Ít2Y-H4~ _ü‡¾9‹ÎðÍɤ׿GƒºúøVü`Ù ï‘Aø sã`Œ|su×Ç×—ª/À7'“kÄ>o¶LP…ß|Tá_rÝòÊç.´>ò±M¢òɵº3õ™‰:ðDåwÒ.û!MS‡çÓ¤n9hQ·ü¥!ä*æãóÍ1åÚIZ?áà ßL¢ÿÐ7'TÌ_™L‘;QZ®F~@« M#xÕ,êAïÜGHðÍٌӇoN¦–ÉT?2ãAŸ3ìñ|s-WÏö•(øædj™ ÈíkÎðÍIÔf¦¾3Y)1¶¾9›+ñðÍÙ‹þZ®I–踚µýðÍITG¦&w¿Ó¿~øæ$‚BvÐqµçÂtg~äío¢údjNþUaô•¨Cßyÿ„oÎá¼Ûþ±K7C§9h9ù qøæ$2­Ô ht~ÔJ¦þ5!è;ŸŸðÍ9+ÎPmZEÈõ–½.Á7'ÓùèUm&™®i"¹Ó”L?%Óô'øÚ3|smOçh87¨Á7'Sš™ŠZ^æ ži¸Â²÷93”™_™¶Üy¦PºæBš™\}5ˆeZ™¬<32EÃꉠá4%éÄr ßœD¦ ;ÐFÂ7'ÑqÕfŸ#`Cý%Ÿ¾gj™ ôÏOøæêxÀ7Ç&û¼³õLªÆL×]WØ×ƒ°Ý~اÂ7çÐ_¾9‡'çà›cûjý'|sl׬ü„oN¢ýM¡“é¸^/\_¼ÏIT \ëÓ:î³g„¾¬sŸÔã|èÎC=Î]ƒšDpÏqÙþö9.Ȭ6u^ýfŠh6´²ÜnÏxˬUqºT-G}s‡ j¡!Žë<&œ‚ãR5p­„IX ìê3AøéBÊoè„”i µK{ÐLõÙ6·Ö3Á¥H)S?ggŒT6uä¦ûà*˜ëNÁZ‡`(›ÖŸpØÉ8¦à¬‚ûQlŠPö„k …7BÔû¬UBÊôÅ%X`ƒ²©ïY®Ú¥£º€&Ë`¸#HùÎú–;½ØÏk¬CíLÇsk¼0€;ã!T&9:‘>Z”ÅüpEˆTú—²¨‘ÉÙ·ÀKØáP“ ‹‹‚˜ · ËÍ1~Q ³Ñ¥‰b˜ºVÛ‡VÛ3Ryqü„OFê6â‚®:—°'¤懥)ÁŠçzà CÁ.jŒM0´‰K³ZKpëÍ'£Ï¹LmË—Ó`Í“±WÅ!8(IX~žGp Z}.TÍ_”¿\Xf†EOÆû ì Ê[°éqÁ¨À#èÅþa[‚]oó„YOÆ­WOKŽŒ[°tÁ¬6ˆSö‚G°ME½ÚõAC45í„*¡ˆáÝãÚSÔ t„$GÂÂN¬Cq ¶"ØÅù‰b¯_p îG± žœeº8äÇ©€Œ¢…3p ö¡…À‡8– õ{àÁ4èCuNØ«âEp>ŠSp)nÍ™’€/B±ïˆÄó„þ_ åþ^„ ¿Ax¹æI#BÏw¬á"äŠ"‡¸),èuòC/öw<šö@sÎ5á&ô!&b·`Á¯òÙÓ¡Ü㇠ênp~€˜sBjÓÎGQoöwÄT eÜUð@Σ<E9¾¢¹P¢hMAÆ.èzŒÚHµ?8îBs¡„s.hÁ þªÕ÷LØBX.°tÁŠ›aŠs¡^°¼Cs¡„ÔŒ œâ\š•ëiQ_ƒæB§`‚Ô†óaÍ…Lz­]YéÃ1'tÖ¼§£¹PÆ-è"Yî&x4+«Þ4JX– åáÔÛÂóÇÅ®>ô‚íX+¤¹PÆù)µÑ\(á†îZcV2k‹iOI8CÄÌgô4Êxël]q vÍÙÕÊ"g1!Ö(á¼Jið’‚0CÂ=ÏP<Ž0­¢hbÂ%èï(aSL‚k4Êù542+ÔÌÇœâÜUðä´ÔYܯÐ3ËP<‚Ô0ƒ‹ÑÉ2x+¡;NH³à‚þÊ>\šóÖ´'§…òrB “á+;¡CH²@ªŽvˆ’MfåHk×4JH²@°Jx7®úÂÍ…G_4 s¡Ìe\|]wñ©Ä«(oˆ“A¼˜æB£†¹ ºÖÌ ÃaJš N ¹Ð+öEs¡Ìíb/×Ìãâ­L±°Ï›í⩼†ò¾Ø:ÁZ4žÊ{¹B< â4!vCs¡1).Is¡ÌÖg¶êžy]é7ÄÅøþ\4V¼¿baç׬Çßçf° Í…2‡xØËËŹhž‚õTáyñQn÷šÙ\h†ø Í…æå‡`äÌ.ž”ØÅ§øñ„¹ðT¦ø×Ëö¾2Sìëå~å7®û'˜f1=ľh„ŽýcŠ#e^ÊóB0Í…„‡r‡xÍ­Fˆw}<•W½âV4—AŸÙZ£Iû÷0úSÖâ±4ÊõbOŠ`Iš Íw¥¹p»xþ0æùÏk.„Ô?¯¹Pæ q)Öˆ)f>õâëþs2CLYx+ûûàJš åÑ/¾ÒÏ+ý¼Òï¸ð¼ø(—z1ijX¿1±ÎÜŠrÇóLCs!á£<êÅ×u?Ê^ÿ3‹¯çyÖh×k¾qàÅ‹0†¸ÍUÐfnýâ¥ìï/óTžW~!¾ÕÈ^ßw˜)¡¿CŒíŸ×\èãb[/ûûHìân™¯ôþ~2_ùÍçâæbZÏËb[-ø(¯¡L1/ò~ź‚Ûu½·‹çÅÇŸž`Šs½<ëÅãâëþñ/¯Ò\Hb]O°‹ýdÞ™±k¹K¤Ç7séÊb[4¿Ã´6sßÊÿzy]÷ïzñ}ýÊïô‹Wbš Oå2”ëVnW~ý¹ùà1”gU^ý⥼§ò‘ôÃJ\›r»îoCÙúÛê›ÁCûCs¡Ì'ûø(ûûãÆH˜ eÞëúÌ„ÆÅÇ*ä×c< 1+š‹`!;³µ{†G q§—Ä—hfQCÜéåQ/¦XÒCö÷™bK˜oÐ\Hø(Ÿz±¤Ç"wæ ±%ˆÿÑ\(s¿y+®ìâN™§òºòÛUù\ù•«ã™ËP®U¹uå~¥ïób”Äh.”yueÓÊ<3c§vŸ0¯Áø!s}.†øÖ;h.”ÙÅÍÏëþU/†¸TßcßGæS.^™1_>Ê¥*×›‡rë_Ïs1›ÄãJ?ð÷Ð\ óáfž4Ê|Ffl ß×·2ŵ>îÊbS4Ç¢ØÖÛbîÊ^ž‰)¶õñÈŒ½å̵]Êþ>2åv]÷þ$±·_™‹¯Ñ<â™'ÌOi.$¼•)–†þæBÛR`kJ| 6ãy.nõ2Å»*"h.dŒý'˜ ›øHæV“xÌ…„ĹÎËÓ žSy‹wfŠežåZ”ÛRv±°ÌÍŶJ°‰Ÿdž»XV˜/À\¨>±Ÿs!áS/.–õ>ßßÏŽü!¾“¹€Yþ'Ë ñ«ìï+óQnC¹åqå7Ÿ‹¯ü\Ü,ó•ÿ¾y+yÅÏä^ʵ\¼•Ûu½?7÷anäâY…ý)Ì…„]¬§|æFg\¼3ˆ)%.7/eŠ‹½Ü®ü\|(ö›a.¤¼•)Fö²½O᣼¯ü÷•ÿ¾ò?×ýçzÞY‰a.$\(VÖȵ\|]·ï9ó¸®O¤ÇøæB&æEsŠ¿e^Êû¹xf.xŸbp0®ÏÅM¹Õ‹¯ôm+ûû sE˜ ¯¡ìï'±¿Ì+³_L\ ë0®åâëz»®÷çâ+ÃJ_éû•~@¼n{ýJìõ)ÖWa.¤|”m<ñqÁ÷Ø¿ï×ü…b‰½>fÊ6^^ÊóÊoAæ,3Í|úÅ+³‹I {}ËÊ×u/ÌSÙË#óvÆ| :@Ê=1Ì…„ËV®åâëºÕ7›†Û÷y^××Í¿¿Çb5‚­>õø>a.”ýðrÆù0˜ )oå~±•¿ð•Ÿ•¿p»ø(Ÿ+¿#ùAÜ:³•goQž¾¿)ìå™ÙžßÙÀ\HØúãÌ6ß¾Ò{yš‡Â\Hy\lù‡8;Ì…„m¼Ùê»ð•ÞË?ó•Ÿ—wæ+/ÿ8ßs!å­líq_\†¹p½®×­lýYf?™ò¸Òû÷ùº¾®ç¯ëy[ž‡þH¸]låµÙÞÁ\¨¾æõ0RÞÊãºîï#ó¼ø([û”ÙÄ™3[{ô1ú·Ìõº^—²—ævñQ¶þφ¹3x([ùÛ0³’­ýÉl߇ðÎŒõMá~ñR.×ýåºßæû&éYÈÖ¾ og¶'è_3Ûû‚¤äŸ0R^™ý ˜°ÿ„‡r-ÊýJoåŸy\é½¼ï+½µO6¬”_b/¿Ìýâëþzåg߃ðVîWþýÊßÄÊM*2Øú á¦líÏË0ª#·CçHù(Û÷0â|5Ì…„½¼[û“y]÷[{"<•½¾g–ü`&‘¹\\Ÿ‹r«å^”ÇÍ×ó¼üã¼7Ì…„½þ'¶þùãŠ÷‘y^l¿?Öïa.¤¼•í}Ù0é!ÛûÊlí—ðT¶ö*óºÒ¯+ë/2Û÷$lùS2æBÂö~?†Xý|Ít`.”ÙÞ‡p¿x)[!|”­¿¾ž¿¯çí+ÿ#ÏÇü5³µOÂSÙúç׌žæB™í{™a@s!á“æB™­½Ê\e¯dö¿o|lãKaû{'ã¥h.$Ü”½þ%î×ý^ß{}Jìob¯//£ÿ qí0Êìõ!ñ¼®{}Hìõ!ñ¹ò·ïw†ù/Í…2ûûOìßcæ+ý¸î·ö3ó¼Òû÷˜Øß?åÃ\H¸+ûûÏ<•­ÉlågÍþ&[}î‰i.´â<Í…2ÛߟÙê‡pW^Û÷ð1úƒÌe)[ûŸÙÚÿÌÖ~¬0뢹Pfû>>†¹ðT¶÷/|¥÷¿7±O2Ï®¼®û÷õ<{ÿ™äs!á®lïßš½B¶ö7³—OâyÝoßOfû~2›¹ÈÇh3÷ëz?Êö=dž]y_÷Ÿ+ÿ#Ì…2Ûü%³¿ÿÄú÷Á\(ó‘߃öP¸)—­ÜçÅGYÿ´¯ëïÁx?s=Êú~1žÏ<ž‹¯üfSÖ÷ñ}f-Œç3ëû…¹Pfýûa.”Yÿ>˜ -ªÇ„¹Pfëß_¦¹Pfÿ=‰­?μ„1þÌ\›r¿î÷òMìßSâU•·¤Çø-óoåY”W¿øÊKþXß>Êm(Ïëú‘ôãJ?®¿—ßOâuñžÊGž¿®÷qäyÿ—ÿóùtèwúÝ\ݦԶ¥ÿòŸ~þןÿó_x×÷_ÿ—ÿñßüüÛ¿ÿËo'ôû«Ÿôï¿ÿ·ÿó_|ÏãÿùËø?ÏÏÿþ—òóï~ÿùÏ)>ûŸþR!ޏ¼Cù§¿Ôí~z‹º!Õþa²ú¾NP­Z?3Ü”F÷Y]=>˰w7!h½°F[Î ‡8q{Ê+ÀÛqÆÞBQçågú ÷Úã_ÚäÚÍb!>èsþV`à\3ÑŠ/ûQX{~«”êØSiÕ'³~ÈŸog¨!kãkŽgfÍ~Èoó9€oÍùóí Úq?ÓÚp†Ç—ºýù83áKuÕ1|“Rõ­ãŒõ*P§j¶§Ø=FÎ×8Ú` Ń‚†=ïšüù¶&_Sn³àŒn…ÔL›¾ùáUÓŸÏ5Ч Æ¥­É©¥¶:ÎøQß­Ù–ƒþüýÀÀŒ1mûäÀöô}±YÔ|OÔct~˜í¹ùœª÷X°=×Ui&A4|ÀÏT¶ƒ>£W´¡Ý4;°Fêk"ýi±¦äkšýA8ßsýÎA{ñàT›Ãø™‰nN‹ÍÇаH*l&bìü›ü¾¥çÿÏ·k&|ªÿôc¿à·jÿüüý}=‡íúñ?ù½N|¯·êf·q=ð»>\Ûò½Nü®o߯x¯ßëòƒ¿ßÿ^ÿ7ÿÀòùÇÿúÏ”ÏýÇߦèø­?ÿðüxÝ~Þazb»„¶ëkðþéçïþuýW?ÿðŸÿòßÿƒçü7“Ú€hä¤å_š´žˆ¤Ï¿8%¾Ð/å¿ü™P~ýRþ‹ÿP«Å~r"R¶HÉ~Àvª›U M‘,ÃjÏ}ÃCŸ6 iù®·Fy¯Óu?Ûw¾ÈùŽíj)ßàt÷ŽhÒøÝßßß‘‹è+˜ý•–o u÷Û[£ý­´¿üõŠô·ÒúaØõ׫ÒßL‹‰ã_­L+­&úíùþjuú›iaNðÏU(~íéSù¶Vöáù’K¼·‹ãvŽhÊüžŒÿ®¤äP忱iÕ7”«õ¿Ëø»ú/ÿêwÄðówÿÛÿýŸþÕüù‡÷/+“ê³7Wk:Èçÿ™|þ?¾þv‡ñã,mUï¯u%»MÚJØo/…c.¿…=}½ÆNxÙÛò®›ïùéåºü4¼)þøˆî¸˜«lDwÜÙê`ÌæûÁ(«nx È·U¨QšÊ&†p>6uÀæ F!ڈ΢Jž†UÑe¹þö4ÈWôÀíyUÉ1¡jJèE·ò@åyâ”dûzl¨·nqü&tuSîʲÿ‹ê—› {爥N¨ÏÑÎjE½ÑÿÖ\®ÈU0âëP±©8¥k#¼ UŒèL!0VÆo`Ć˜jœe^0ÂCTئø_3¹0DE¶Æ]EÔ×ÁˆïD‰‹17î:C¼É˜Q 1bµ]œŠö]³¶ZXþº*‡à:,a[ŒàüÔ£š­b4?eäÃy±5œ2q]øåN?µ°9bÛØ¯1bã®§ÈmÄÖŽïÁþó©±ëëΫ6"›±-¬rùˆO¿¡ò~CüíLf±aI´L=ú¥ÿ—ÿ4é2 endstream endobj 118 0 obj << /CreationDate (D:20150701163322) /ModDate (D:20150701163322) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 119 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 121 0 R >> endobj 120 0 obj [/ICCBased 122 0 R] endobj 121 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 122 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 114 0 obj << /Font << /F8 7 0 R /F54 25 0 R >> /XObject << /Im8 110 0 R >> /ProcSet [ /PDF /Text ] >> endobj 125 0 obj << /Length 2629 /Filter /FlateDecode >> stream xÚÅZÝoã¸ß¿"¸}¨ƒžµõ}ín ‡Û§ËÛÞ>(–,k[†e_’»ö/çƒäH¢ïE‰Cr83œÎßïß¼»K“›2(3•Ýܯo¢(²2¹ÉÒ…¤µ¢á +³Ô–ÂÉäþó±ööEk#Dl¹ÍÒ1Îܵ™´¦a;æÔº 0Öø]$’𹡥Ѷ|·6Zƒu&IáëA$B3ÙÈMûü¢E¢0ÉqþÁ)0³¥§•^¸ËÅ¿ë>=G›÷ÅÍz`†g·—ßdŠ\SBú|æù$’ŒÈõÀ ÙÀ k–„§ðîm±“Fj° gÌQjhï½'Mhm]H‰ˆö8ÑëL¢Å»¹¤¡ŠCØ„ôg#¿ÇGV_-ó•3^=•Ñ#Ü¿9kzlýaš$Ì‘Ò ÏÍý|Z\½Ü5Ζ?X™¯>p¢²â<}娈Ý3eGÉ75*J‚0õ5&ÆÇY„-:…º;6Ò³°ÈŒ±ápïî !ì2ÒßúsA^ä$Õ=ˤÈ'L¿f† ဂ½+,»U…9_€Zx2$ Mw«r¬‚DܺOL=<Öœ8ë3´³Å±B+ì¨Õ­iÔ¯aÊ´—“±àU5‹ ]V¦j'Ö‚ö™‡C†¬·"Ëu,•YK à‰_å®­V#&qºÞ½n§%­TÓ7(Öé{¬P®CÛÉÏÈ KÑÆk×êðˆüƤ¤»œ[E›ùþˆ;ñD½NâºU³d²?í62(ëîä•jÃú¶œâ­°&êñÔYª8^´@è.GjmÉaàÓTؽ±âÁ¨¹{†Št×›Ûzø|¦ùG[ª£‰De½=;©9ËÒ ºðä9Ñ·1|O´CêþþùÇkt‘õ é²ÃHõèÒ8òœÒUçÜýÞä‡héd#}°WX•Ô4é%÷ ´£Ç¹w~"µÙ¸@!çÜl`‚ÞÒ=tⴽػ Š:Ì=ϳÞ*˜tÂhNæn2v²ŠWP:Kô;»odQMM½¾£-Ñô4 „ÙrÃÏÇð£ÙCTK°sE]Æt5Öp$ª2u)sωò‹6æ31sÁlëlú»jù£1rÒ.!Íx®)íÄŽ„’a‡¹™ôiŽx‚St'sŽÖ¸ÙëýBåœÊcv5cÍœ ï'›ÜŒçÚÉç8ãÃP±ûÐÜ®‹ºaéô´Ö,„¡shÿ¦½OV¿^54™½u{ËoŒcΗH¨ˆV«À³ååkr©È Àp8ÓœÇ[”¿74ø¤pKó7ní“Fž}Ä -³ÅŽôí¾ˆ#¼…®½Þ¸;XÊ7lÓ0¢4—9 Ù]léŒÄýÊPPàI¯¡K€h 4„na^¹`Í¥,ƒõÁåÞ‰k|Xà¸ã³ èÿâµÛÆHÄŽ­ûz‚Œ&tΙ˽׺‡º6ã;TÀ^}Gïê@ÝoQn‘îÚ3w¯ém+  =ôÆ“I­šèT èæ‰CD[C°.¸ûÅMíiƃXÒe~h‰­ªÍXÞ'`w¶‘Àz­0ñ®‡©Üšÿ<˜±© û†™uÞ8Ô>²e÷ØQ£c¿·YçwN¼º«©LÔÌØ¡+¾à›u³ôDÁ¸_a©å‘F'Áƒœ÷D´Ž\|`~X—Î=ÀbkxT®\ŽÂ^<×§>ë9Q}žr­ÇFÕ´*Ô¹ÃëggoØã«Õ£€÷6®_ÏCw¦y÷Ÿg7³dÃT”ó[u𠙊 ¿ -“Z6kÇ)ë(AÖÊ»w=s×Ð~ôi›ÿr½æ¨Ò4(®¾¾I¼Aе‡{y_{»´»CD•I™ }üÿìË"R•y}Ã,´Paÿµi'Эw'x©Ñâ‘/µ…EM{;‚TwÂÛz‡_ÅÎ{+:6Íü$À‡ç Ø¥^ˆ’Z “©7k ÅÎ &¡¬åÙø+—’f˜ò'‹%Ó#àžøÀl©÷ŽÉ<Ø3â£6¢Øé;!e=Ëu4bÈ5œr´—4`z¸zFÎo—¢ÏBÞ‡W¶U 6v‹œÅA¦ó„ ñ¹H¤5±¦}Ôž˜g®çã<²—2µÔOŠIw¥9Ñ৺’Û·Ã×Û*Ô7%ÅíxÔN}ûòâFˆñ¢Ñ+BeÓ~\Ôô£væ"–È^°Dè˜^e‰Èj>l‡sB¤Bˆ¹Eç„òXîE!bŸ¹°Dî±Äœæ/9Xt,”òÕ¹°DîÑL½bþh: g|Â[(ÂÅBD¯X¦ ‘Ú¥Oˆ’…ø3TËQ)öîë2›Ù†Áâ3² uõ’dUÄtÙÂÜØ÷D1e@ÿÅeJUÒ1Ú0À{0ø_ñ< 2è¦çš§kˆ¹»Ì «Åþ>XãÄÜÖôF!*³¡:J€L:Ûv½ÄC 5Àà+Ô´AºîêäÙ6(eOÝ&”ÖãÊÐ^55 `ZbkÓ2´UÒ–®{(FÁ–Ž6xFá&ˆ³])Ê¡uK`Ú 5,ð©×WEêD„8UmüÒð´¤á[Ê3pÌAìÞg’1ªØ3_YSí—ì¤@3·Pý©0oý€éL Þ©D¼§H©&±³©ÅR!ËÖ¡•AnaDçð" –ÛŸ¥+{‹Õ#%š20gþ¹†QèQhlPhÏĪeUÀ¤‰ÿ<ð"[¤)Ú=ßèéͲá¾­ÁÚ`ÆYTH=‘ìgNA?> endobj 123 0 obj << /Font << /F54 25 0 R /F56 30 0 R /F8 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 129 0 obj << /Length 2116 /Filter /FlateDecode >> stream xÚÕÙnÛFðÝ_A¤•R‹áîòì mE ´ˆÑ—$´D­Eº¤d;@oï\K.EJI´éƒÌÝÙÙ¹vvvfüíåÙ£g©—ùY¬cïrée‰ŸÄ©G‰¯Mä].¼—“멚làWð'¯§3Å“] ” óUƒd²Xe7Àx]ð)VSLîïÌüŽ×–Ht›#JÑœOg&0.m—LED ””›ÃŸÈ]•þt5yr-ÐjUnµóMÃóf“ÕªhdKÃ’Z´|7}}ùƒx3¥ü,KYáŒÑ“m;©¸J³ZOu ˜!“ žd84“uŽÖ¹EÌ‚!9¢Ó«@Ü[Ó ²P$Ûã`Îü ÀŒgŽ»ŠÒá½Ä“Áëi¬P-™Yk£þ”x ÌžŽWhùjƒÖòÑ<žE¡ã[ †0šÑW,öx:‹´™àq-E5d¡H¼ô5 gV¿{øÝˆ0¯‚(€Ï øá‰¡·x)`ë [qÚs&¸i.(BPñê;þ¬„l%´ó—²iÅ–@´/ÈI˜Þ½¡d,ôÅzUí0Ï;±ÇX’\­ä¨Øb¸YÇÙË–¾úÄ--,NÜ%7b÷øS´ûüi÷‡Æîÿ¹¿WB®} (^²½»àÒü|mD²RÆKgÍÊPËüøa,|‚ñPNëiÏ8ŸñÇZ®èôªä·^v\Œ=1¢VîàÞöyèi¡,ŸŸ µ'—¯N/Ïݳ@qá4”Å}"r"|}ß8¶mºZòð{1έ`X-çÁ.|%ÙH‘¯ußNßµ7‰¶mejïÃ5¹8-ÙD¦Oq=D˜Š´*“úŽñÅ@V"¡:‘ßuNÐ ˆ=»@ÿÊ tå‹Oʧò}ðËù‘å>mWóÁr{‡Äï‡ÄI‹ðojñð#k1¼¡×‚º’눷æ7¸dQ¢èö®Û0¡&¿wñABÅOßÇð{;íozÛ{Tà‡aŸ»h6<%qÌ~£1³ëq|tž¬ó#%OŠ¥[Tº»Æˆ’ÊîX,>“mÉØ•3'¹ùÀaY“Y-N&8ñCQG4óŒ…nìÐSò³kæ€g ¼Bù&½ç"é½n;±É–¼èX’THØ=òîE‡iÂM…—}¶Î3 ûb¸â@Áp=èâÿƒÞãáÄòÅéåç²|,F<î^6{K GT;—“m)Ú+㎠é„ûú a#°çà )ÇaýÄÈ/`,غ—‚úöÀÖÙR.CÝÇÌ–¡˜+QíRÊÜ0ê w ®ºÛ´Ï6Í\»Ékn´;©öÆÁ©Ú›2 D¢ú½ú¶¨Æñª®8–òlCY¶„JŽÈo°ZŽ&¿JÃW( ¹ic2÷>u·k`q˱K²bMº^u,`nRÂTc7"ܱ»š,œ\q"DKù¶,Êo"7º¡V’5"6Å=øz·öÃ鯑—ì‡x7ŒC aé?°µH¼š1—6_“þL»ƒÕ–c)è8ç]×Hø,ª–‘¨M‰÷h(_Ð5£vˆØ¦cD^ËV N†©ÝrÕÒƒ«UiipKhMV’†JâÅ~–˜”ú)&ôMzÆUl_µ‡ |#â±MËgFi•‘çñ(Å'بŒPÜL+Ì»›¢nÝ~Tp½Šªrç­`Çí4Š'ÔÎÛSK¡¥ÓÄ9Õ` hX€œ¶Û“dk A/«Îm`«¶¼Lžµ³ñ­.Å-µnU©ö«µ„¡¢ÛAbÐT[«.\MTlÞº<@lµ°ÍÉy* J£U×°Í‚ú®*Tl3[w°sNnRZWĦe´¡©£g¯"ßžo¦~em 'ktÚ…'²¯ÝÀмîè:Ö·Š8:ñýFi÷õ±²4“.eÙùïˆq­îÚ†–6ö))l¬qŒ¼ï%ý­|ièhÜH(noˆm¹ Š‘e.=WݹBÇ{’ýÄÀ†ÎBl¼'yçdë~9xد8ï’›¼uŒ®,–‡l7’ÚìÞŸé8 Þ¥‚z¬wìè­凱‚c$•?1 fì™E(œ¦TáHÄRé‡ÊOâðƒÉ·Íº–: |ȇåÂÆé¾ ÊθχuÌùñz¨ßdjdeÛ­t5Djüˆ¢ºü¿ðã+žQ hxe·Î%‡ …žß¦&˜)Ê5eŒ×n¿Å혦ôŸQ„cM±oˆ‚%XÃã;õ9ŽÖm’léYÞ~9Œ2’oxâì.¯ +†îè‡AµßaÐÛ‹ÐôïÖF¼‚0d´±ýŸ+mƒ¨Émâ…T:Êiw ¨ÄX0ö(ô¦Ùm¨¸ÈKûïJ7oÖFgb?Q¶ ¤¬»ü.Ûñ• endstream endobj 128 0 obj << /Type /Page /Contents 129 0 R /Resources 127 0 R /MediaBox [0 0 612 792] /Parent 117 0 R >> endobj 127 0 obj << /Font << /F8 7 0 R /F54 25 0 R /F7 24 0 R /F56 30 0 R /F21 130 0 R /F20 131 0 R >> /ProcSet [ /PDF /Text ] >> endobj 134 0 obj << /Length 326 /Filter /FlateDecode >> stream xÚmPMSƒ0½çWä˜HIÂW¼ikgt¼8ÃM{ÀB‡[ýùî&iåàd›·»ïí[Ic8’š\äYAU®DšJº·ä““)¸‚EèR2´`õ` ÝŒäÎ%ÊhÁyW’Õ¶ é2Z®²i,2à/kú¶\²n{šàÛðHÇš™%ª`·õ;€'¸ófkL+6#2Cç™§«ßy`œ|Œ-Ý©ö?|Zû¸F¤ú@¼UóBh<.˜õ;pØ‘«ÔaÎKýÚA?¾ÉºX«À¯±Là…ŒLê„m tª–Õ|W>ÂJ#)…I3¿¤!X=6aR­ÙÜ]FøöÀÒ.<ÑÔä±s4{°t¥Àe9#ö_[a²uÂn¿Z1aÕà<à¸*V¢ÈS:1BKí§V “ä¾$¿€0 endstream endobj 133 0 obj << /Type /Page /Contents 134 0 R /Resources 132 0 R /MediaBox [0 0 612 792] /Parent 117 0 R >> endobj 126 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc7.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 135 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 136 0 R>> /ExtGState << >>/ColorSpace << /sRGB 137 0 R >>>> /Length 32743 /Filter /FlateDecode >> stream xœ¤½KË7Ërå7ߟâJ¿]yÏœvûìsÀ¡èVcÄVÛÖ1ô×wÅZU™ч-<Ðå·³Þç_×¼DF¬•~þî'ýüóÏÿóÛÿzÿÏ¿ûËÿö?ýûŸÿð§ß®_×uýìÿûOÿá¹ÿëh?ÿõ·¿ÿ‡Ÿëç?ý–~þîþŸþ-É?ÿóo­üj駬_¥ÿüËO½®_õRüýçOŸf¥ú«±5]å×5žf÷½å_=¿íŠO{¾ò¯4Ÿv÷½¥_£¾íŠO{¹Ò¯¼žv÷½]¿f{ÛŸöãrß«ÚÿýŸõæþÇ¿7÷/ÿñ¿üöïþÇ|?‰?ÿçŸùKÿ;ÿ°átrþuÍŸ?ÿËÏß\ûóçþíø3þî_û‡é>‘Y·™ÿð¿l¸„÷_Ö?ü/×úµúö/ûý—¹.y-Þ9ÿð¿\ý¼Îô‡oQ©ý¼Ðô‡ïQYí¼ÒôÜ$¾ uþªx_@y”_µò}:›ñžÞÖUöVÒÛšÊ}x{š·öYŽvâ۞󔓱vÅ­'û´¿ç¾ß•÷^üwï ÊCžEÍ¿ÖÅ·õWø0þÊ¿ãý·éßøoSÉ¿æÒ>Ê¿öÇxÿqù7þã|¿S?Ð_áçò×þqï?nçk¤ÝÊo5'yÚ©æÙ~¥|³>¸íà?¡‡ÿÿÙ¥5ð­è™þýßüû“îÿð7ÿôÿú—Ÿÿü¯ÿ׿üüåÿåÿþýŸþö~þüwèêç¯~ßvùð.þ¿ÿ§üÿÏ¿Øõßçü³_G™øÖrm¿Fÿù×úùßþ˶îÓÎwŸ}Ó_Àô›Z˾ɗZª¿®æ©7Ò}÷ÝOT»§•I÷_ûÐú•.Ò¼JÃSÆYßoBõT©ÔôÈû–9ê¸ÚþëÊžF"Õâi*µËÓR-¡ñ몺‡¦û%ÝWíˆW;~åê©àÎOù¯Ž*þÊ’™‡£6Hyzš…TZ•´ú—’ܹßë÷žð÷ûÿ&O¹äоT©Ì€i,OxÒ7ÝoŸ#<éž•â O³\í—޼¿Žò¥,_i4I©{ÊúWÊðTp—š¼ÑŽêEêµDšÝSÇ]ê¿RñÄ«í¿òŽxCNºg¶8râN8š¤ryJJ- {Ozäg}ÓÌê:ëZIÞZGøRoºß>Gx#oªÅzï›î_r„žh!²ãi’îÌF®{ýœŠ§ÙH£zÂltUµ=-RÍ_Jz'ª¼íŽ~¡á.} _ñêÒ£8ÂLî¦V!7µì©ë_‘;ñ¥QI³{›|¯öîÖ®hÊ•9Â~ïÄNYfÒ ûÍtÄ;1eôu„/õ¦Ñ=5\VFŽÐƒÝt÷RŽF#Ý#‚#Œè9K9@Œ]‚µ;,ò{„Ép”Ñ[¥+É”Çc™Š9ÂÚïÇá±ñ²¼TGY>yÔË/øx®©(·õƒ•ëÁa¾åóö¸€X\{¬—¢<‡­)Þ/®ÇÎÓèòšyE1ç§|§_l‹Ò…EªÇ<s„¥+Ö«|O-<¶¡xw`»ý©»›ò¨8qcÎKq”—á=è±se“®%or€Kñž zÌvðýi8W„X$ ë¬ysn”Ç!"‚cˆµ½à,bH Ëí/ÎaïE§Gôö)¹RåÁ fï‚5B¾Þ _G½À*Ó<“·®ávˆh†àZñþbBï1Uà ÷¥DˆÇ°œöX ç °òqc¹í‘c¢Úãà%,Ð#æÿ‚3ÈËÏ:=²¹Qºk‡ìFr‘÷Ü#»‘ŒEl€S±F¸ yëv—ݺ"a-zsª ©›áýŠzD<7å&“zK{œvð,.þP—÷‹I;ÆŒ¾Èc¶ƒïqÙ#‘W °òb ës)ÁûÉz솫DÈ“œò¾y\E±'‡ Î3OrI¯î±ÞN5®[Q¼'¤»ÜW€ì6Ë%ÝH€Kñž–z\vð}ÿÄ \’ôŸ_,:¡ºñþr=fÃûZ”sSâk„Íî!â¶K|ø ÓžŽXìÜ$›ä$P¢šÅ¡Ù$2™äâQ"“5@.%29¬†òÈòƒ•èbƒˆD[„äé|qp^ðž"zäç,1±‡â=j{䄹c–â‘K-‰MõõEBÐÉ£¾Hñ=²«—(P P?œ%3úá,<²NíÌ%¢ÒdèþÆûZœŠµÈ©Ú@ÀÊc5œ#@v÷HvÃÒäò± ÓpFÈðÝÀ×—NÒ“<&ÃÒd§:dóÈNõ^ù^!NÅÙd8KÖÅòU‘õé ;I®XNþ©ŽAÓ!'  \—n15Ãczp¨/‚]›¼R€œ¨ÄÒ=òåø¾<ê% ùW_Lv …yäÒRÖ>=@1²B¹dW0(óȉ:W õ"PæQ/kËþÔÝ÷~1Ûë½äeðˆlž$3ö+@v7Þçã‘—?QÁà‘—?±1ï‘‹ô‰@™ÇNÄ/zä”éÆ{Êä‘S≀•G¾ !©/PÉ|¯ÈÝ÷`í‘ÁLάØsá ;3ƒ'bŒÂRp6WÈ?„g‰Ï35 KFÌç‘q3ž#dö’Ï“Bfo‘‰˜ß‹ð ¸Úý¼ßžõ~ ™ÎDÌïEr†VÈ5+ã~{nƽ„<“ò!³ÿgÞ‘çÆÑRxÄŒ™VΈÎDŒ˜ƒpí!;~”9.æ$žˆ1ÇZY¼¯ž1«#äÉãñb†<”[̘‡ ¯pgbr–¼Ÿ˜Ù_d,X".v<Æ#Ïz?P1g7—ò(Êxß=ó}ÉØ\y)˜õ~àÂpæýî2±‹ë}áz…¬÷«óyyæ|ýæÕBÖ÷ Køˆ9_-ƒÏÓ³~²޳Ž×ç‡{Èœ—ÅçðRÆóð<Œk™ãiEenÄü^˜=åY£žÂ5‡Ìñ¤âÆGÌõZM¼ž 9c=p5–ù@À­Ï9Ÿ¯ó€GUî=d¾׈±›”+Â¥žm¾Z±q¶ãkÌÅŽo5äjÇË| `½ÞÆ÷Ù³^ocä™ýEm|^_n:ÞVl^GÌñ¤vÌwæzµ¢ 7b~Ì:‹¸?¼bÆ|¨"Hñxx…oÏœÖÉçí™ýEEˆ*b½_È¸Ž¸Ë|#`ŽÇ’!·Bæ|•YpOã¼Bf<¨RåÈñÐØZEêZÄÓŽÇûôe­TÉ 1Áˆy?%_n†Ìùsâ"fÔßÏÈö–ùfÀعn1soÏìÏ™K1ßW_GÌùªI€«î'°L:âlÇ÷²žoaà™ëŽ?1×R.ÝBÖëAJÄÓŽ—õDÀü;㹎›½(7‰X¯·!0׳µˆõ~ #b}_;ÖûOeÙϸ/åQBÆó ™ýYïìO¾Üu?AJxcÖû1¤cŠXïÇàûä¹Ë|;`}&ßÏÓÏÛ³~ÏŒ';:ž÷Åçé™ïøx¿0ßçÉxdÀŒ§L ¼óyKáA Y¯é³¿(Ðôlóf¶…lÇ÷²ÞolEÌõèd¼0à^•ûò°ãe>0ç‹“û•3Þ!‰ß-`ÌÌG‹˜ñ‰…gÄO椮g~’ÖbžÊ˜ó­‰/b¾¯‹ñ@ÇZU",ñÔ€¹^d¾VÄ|BÆùJÒç™ïëÂÄ5bîÇIzf ¹§r7®=äa,ñî€9/lŒ{Öa™DŒ÷¹c³?!g;>•¯`âdÄ̇Yø‡s?k5ªX{Æë ™ñòÅxžã®ßÃbU$1f†ŒïUã…çbŒñ"`û{¯¸ò~ c‹¸'eô‡ž¹"ð¼”g ñá•ÖJÞr10Å»™Xq.ʵ‡ŒïQx”ñ>&NDÜy?;Ç#Ïc)·˜WRž9f?Ø|¸^쯄åyœ¦²ÄÖ÷zø×KYÆ«€[U^+äÎû…Ü눇qî!Ï®\¯€UZXâ¹—¥,ó¹€Ñß–ÄxSÀ|ß#Æ|¦pc6b¬'…s ™Ï;!ÛsÖç™ï ˜ïsB`,bŒ×à2â¡ÂxÞž±ž. ÷®ÜjÈc*Ë|0à9ŒGÈËþÞª»ë¡€ù¾'î÷œ‡ñ ¹6e™Ï¬÷*Äc>,\bæx‘ x&åó2î5æ¥<#Öx]áÆXÄ9+ã}ò\‹²ÌwÖëí|Ÿ<;^æsOÞÏxcÀz=°ƒð¬ùz%M¾žõz‡1ÇËÄxPÀXË|+bŒˆGS)döwšè¸ëóÊÐÛŒ˜ó…|ñyŒû% %dö÷Ü8ˆµbÂ2Ÿ ù4Â}†¼ø÷ø÷<8_/4‰óqa™œy?X˜Ï+~žõ| ¿'Ï|_%0ÞBžvüÊ!ó{Í•ßÛ—5ßNXâõs¾“QÊqæñ¬˜óE͇ ¸fã²^oçóò¬Ï³s< x)§°Íç2ãu³¿f`7âb,ñÌ€ÙŸgî÷Ìþ<öçžû2!s¾—Yð2îÝsÓ*Þ’á¬q6–ùHÀì¯26n#æ|G{9dÎw4ß1`öÇ…ñÀ€o^WÀªS Ü:"æ÷\ 1ò-À3äš”å{ ù´Â2_ ˜ïs’rÄz½H‹xpÖùHa½CÀ|ŸKáóöÌù¬æ;ŒJnaùžîMYÖ#ª,û¥Oþë-^¯€UÕYXæÃ;^¾÷€õ}m|Þž‘$,ñÌ€ÇÃ+äiŒ÷Å3ûkͧt\¹ß%,ë€SSÎ=dŽ·šp3–õNÀÝÎGæÓëûŒÀIÄs*Ëz4àEÆÀáYãi¥0^p2®1s=* Ã2×ã²ðK!óyW¨\DÌõZe½EÀì¯dá¶îÚ_I¾ä2Ÿ'^g;^ÖãÏxdÀ\¯U<˜ˆÛP–xqÀƒÇ7Þ/Ïì¯4Ññ`¼PXò1fU¡Ï1ßg™Øµ9Ÿ”‰Ý ™ë Í÷ ˜ñ„ w‹ˆù>k¾_Àì¯$_o¬³âp‰çé™ãmƒJCļž¶ˆ¹žn‰ãgÆä[=«bаěæ|¡1>`=ß‚ýî€Ùß4¨{FÌçA7¡ˆù¾5¸ðEÌùbc>û—%ˆ¯ãzæx"F ™óÍ ˜ý¡¨§Í9ÞË‹[BfظåXM-ŠæsÌõJƒ¢BÄ\oË‹WCæ÷Þ0qŽ˜ý]ƒètÄz=(ìð¬ŠÃEnl̼ߚÏ0û󗔈9ßÐ|­€ù½vÆÛòGÌñ¦3_=`Î÷;J#f¼·#±>bÞ¯Îx–ãªë¹ÎxSÀü;\¼"fÙO ¸ee¼ží\DÌxSg<Éq³ûÉxPÀœ¯w8³DÌõ–¤9Ó|¥€9ßì—ð¬ùF¯˜ëá>0_ ˜ý]ÇÀ±¾ÌG ˜ýUg¼Éñ`¾´0ÞgÏì:{#æz¥3^0¿—½RÏ/ï§gƃ”"æó(L‰ûuÂ2:^ú½QÚ:bÎgD(z…ÌïMóf¼l0ßøË"ÉÖ”e<X¯wà|æxGi߈õ~Ày+bΧD¢·œt¼p¯Šù¼EòyrÈ\Lü°g[OЏe ™ñÊYíø/ó}ž˜˜{.oœ¬W˜óJ÷y®:ž\oÌx‹æ+8îœïÕ‹õtÆú-…æ·ÈÔ3€ÇÆö$\}C/ý¡ÅÌ?µ±ý¾êÇVö$ô( ÉŽp5 ZB'5Ý©£å¬£®4†'D()ùú¡ÄìÎÂ!3YfàÍf­´‘ûf,5ÜeGˆ6R¬Õ‘YÇ–ö¥Â¯´án9 [me¿„Ù-Ø>T™yOK4G´OEPÊzdά¡7îØõúPc$¶#㳸Žl%GK¬óKo]ÇÇíˆgÀŽ#ÌÔröK4XíÛÑY8¢¥.,Añ¬i{’*u@ñ¬Qœê;Ó¿äˆOzÊSuD+^ÚÊž´’Q-yÂWÕñ_ÑêôR úVÈÐoÅQÑ#ëå ‘»À‡#Ú®&µãÛI#¦#© ÝIM¤ßIÈôYíøNZJ÷ ù¡Ì½;aö7¥rdö©÷ÿu„Yõ€ÒÛ‡tÆOŸG´HÅ®•#TR MOÈS ãNâÝ…QʇÔGC’s‹'ôÐ+XGu'ÔaFN}ñ©šÉĪÌv:'U=²Ô•ÄöèKSÿÊ\_ÒÈ×D…±#dÝLu9â/ 3Í-D1ëþ*O,ÔáI‹"ðð„ᦾԉä“MŽ´u„U­Zz~‰›Ðu„ÙàÊæƒµ‘î„-̨á.,á®fÆTav¼ExÒP内}G´Oíjµ“fäÐ*Ófs QAGøú*Rñ¬­s„ Z^~H«¿hkéˆg•#<È›-‡Åœ /3:Q½'±¢ò8ìàÞVsÄn¼G*^ØÔòHm諨³Î©]xá¿{¤Wß…ééÍáæjê%óAu[lêŽóÁ>ç °Êp¾ØÍk+êj‰UúCS®Ô£zõ]êas i‹fÙ ux©Xæ‘Z«Ô+󨞈 ¾hZ Em©[šPræ‘*Ì 2ý_\v pô¨—€~ØAG)z êü úÕÉ»K¼Ç3ªšR»Ì#5‹R0<ªS#&fõÆBþæ‹æS’°9UFhÖ#ÆFÖFHEÔŒÝ < ju}Q+a¥äR<R©3#ÏÝ#_~úüyT?6L2<êiÀrõ‹ÅN£Ëd£:Ì!;Ü£Þ${¤²p†g»Gúýdl|Q£–(ÿXR…YŠ?šC Žàî@摾ÙÂ_ìö§  íQæºô«ÜS€ì® v|î•Õ_œ¯‹Øª_ä+=Òå¢"çÕ#Ø+"þ_\ÚÉÈ~k*Õ\<ªO&È©þ\QiúÁué‡S©ÿë.¼ éÎÕÃéÍÐàuýEÍŸQ—)ê«” Hîðñ¯º¯Å#;Š?ò/šÒ}ƒ ¡GN!èäQýœðxT*„¨¾Xì1ŠyäçLßüp:îðÍ1±Ã‚Ô#u¥;<ª }¾ØÌ7ù2tHuz¤[ÇlУú” {拪ݘXûóÓ£z¡‡ôÈ®c‚ñÅ¡{ÇbÂ#ÝÙeë¤È®žõâ9LtôT_Ô½Ù”¸r€ìT;*=ª“ÐRµëR­~`óÅͧd¥ÙÕ¤ª{To ô„|óR^1sþ9å=«ÕìH"Vó–AùÀ/ky«Ú{DÌ/lL“?þ2…Ïâsž³š¸Ì‹òÓžik2YÞ0íiš±š0ý±–ˆ5FY!ëù$Ê»zV« ¦_ÌQ‚ÎksÔœÍ~ïäÇ4‚åM7ãQB¦Íã+ž›“O¦OLëÙ)êY­Eå{=ëõb®âÙú$Ú1DÌÞp2}!`^/m"æý¯™°– ·²ZÀk9b5}È|¿0åm¨Cç9«<@FFÄür¦<Žg•—gyYÀüs¦ü×—Mž$Ç"bʳæJy Ï”·Î×ã™ï£–_9®*¯—;å~<«}Êy¦¼d†¹_Äúžù1¿Öiy6{‘ÖMçËü[?äo6Öóeú®ãG~™å“ï•ùzÀ”?o“ï‹g½–w|™)d)_ ¸Úñ½…Ly]æ›yNz¾ŒpG¬ò°Ös?ò½³œu=À:Ïå•×UyŠ/Sª«<É—Mž¬g“—ø2¿'•— ˜òRëÇÍäk¹ž˜òѽ™<×U>”ëÇÝþËf< /ÊxæûÔå<ÓÞf`àð¬åÂ(ï÷Lùë‘x¿<ózh¢íÙäßòå"VùKʃÜíøñÒxÏ ½TÀ*¿Y)ÿãù‘íWÈœ LL7µÜ6ùÑ€åuž)¯98_wlòÝÌS¸kùü—ù½Žaåë_¦|¢ÄÛ{Àšb yѰÉSÚ+ÌãçeåÝ_æ|Z,­GÀÕäA)ϰÊ/+Oþ2çóZžàØìÝhõ1×{SËÇ=3Þ0¡Ãñ¤|$çËŽ»®·'ísæx5‡•Û~™ýáœ,ÿ²î×C~2‡ÌxŽÊOÌ÷AåÛ|^åVùÅjå¨'Ûxºh¯0åiDZþ=Ú§|YRõ/åÒCæ÷² £±ÊRÑq²ó¥¼aÀ”]¨ÃŠ˜ñØ5¬¼òä¬ë Ùª¼BæxºØò\t½¶¸_0Ç“…\ÒˆÙß/¤„EÌòÊë²òÉ“­üPå¦|Î…ˆõïÑ^±ʫ‹<^ª!ãz„k ™åy*_0˧Õ~Õqg,òo©†Lù• Xå©`‚íÙä‰tn#¦üd‚XÄ*Ÿ¦å•_žZ^œ´|ѳÊwA‡2b•·›ü^<³|-aãÉó2ùµ…þ>`•Bžã¡ñJás1ù«²Bî&O%û1³<;#ãÅs2ù$ÔBDLyÏÌù’ãlòSœÌòl•o ˜÷Så›ëzYX¾ç€)ϨñÀ€Y©òKŽ­ü.S=`½Ÿ¨ûŠXå­´üñËMßG•7 ˜å­ÔYxÚñR¾ïXí›D~(¯›±ôW«| ×;Ž­ç9é÷<;ë9«¼ˆä륀‹ÉC y"æxB{ÏÕ~:Dž›ögcñz¾lò¦´‡ð¬òŸõB¼àá?ý¶ú§K¹ÿ[¨5OÈ­n>9Zz$‹ÿ7zŠñ¡áù¡Äˆ…Øa6ÚŒúî4K ½zÂW"ÑŽñ%ÍÀi(ÞrÄ3ƒi§'ÕC0ÖEPÈù!}â þ现ÚnØuøP{~a O¼ó0øî`4dò}Hk¦ÞGE€áÒ‡tW²aÒê378H¢éÇÄ =²•/iadƒØ„#¬pŒñj±ùîh©(Â5¾¤³¢† ž#ìPhÁÑP1…:=ñN,-ØÞIM‹)ÂàO¥#€è¨©Ã=¸9ÂÌ¿#°ø!ÍÕïÐ@q´”fý’ŠG3ÅßzŽGèå;zÓ™ôDatìðrTU„!õ/iž{GôôCš1Ú1aÿÐboÓ«ŸDÁ„†Âä/¡jµ#HzÒPCFBUv˜Ó"3»#Dåù®¨;ü½á«êØuø ²ÛQQñ)žüŸ4J -¥±¾TùÈU/5ö ŒN8ÂYSØÁVë‹i}Ù@¢²#¼‘=Ž X$´¾4ž¿².O<3ø„;¢`Ì¢áíè=>¤æqŒj8ÂNò@ÆÓ‡4 o ⤩;8™ðŽ}aÃFµÑ´tu§Ä¬¸ ³#žF0GK¼ïȇT˜i@7óC…³ )Ò¼H #Þ]xU;¢( »?T9«eÝð‡4£u,-®ÜI£-²Y”= %¹' ÞÁ‰hŸ£®b÷›åÑ· QÆ©ÑÎÄ ëCZUC“ÖÅ“"Ž0Q(ãC‰oÝD¢…#D1e5?=a´Ÿ0Žw„ÑiÂ+çCZù2áÃü!ÉQ(Ãv\'á ™°øPåhÁB GY¼ÿ…#dN”Ÿ:Bo3!JêˆbXø}H£ Uް#°° t„;¸ýPçh¿ Áà³®ó Gx[v:a–¾°‹ø¡¡¿€ÄG¸ƒ fŽ(”}ÁM½Zs8ÂNÏBÄÄÑÐ#ïÌf ÆÒ„Ðð”ôÈ»×w4õÈ‘>$%gP]4ÀðˆÉÀ‚¹C+w½¸ð2N9`uŸNZNpµãQnäYÿÝ5«’°„³v¼,ÿf¹Ì…¼?Ï›Dˆ„‹f¹è…9Ä,Ǻ×çYÕ5å‡%]/`ô º£cϸ Ÿ—ç¿×XèYeLËå¾¼ìyt„;f¹ÛEw‡/§Ëîòj#Vq¦SÌr³‹ê³0ö¢{«c JC-!³Ì8%–KxÆFþ”;yVa º#8ÎZn'ÿO™2”F‰˜Åè©X9Ñ—)Í uA«;'ê~rÀª>¯uAž›‰¤@?"b~/‰îŽu;W¸ô»I´´+d½^¦k;*ð“ß'Ï,ºNøp"¦äBb:³ãiB/,çp¬î¨û¸eHÌy.ú¾æjnô_æ÷š©n0ûÛLu{ÇÖÒf=âar'ò=Ì÷-3ıÍg¬k"n¼_XÛDÌç™'Ë]¾¬é´*‹âÙžw¡:}À,G.ÜΘÒåb:ý——–ó–ô¸[˜ãÅUË.+îW©æþ|râ46ª+äl,ý ™ýaQwGÏ”Ò(ƒîªžUàeÚùœœí|¸]0û-GX¯w=nËæx[XÎ0ÂÌš×ëYÓW[Cf]¡u1團wë¹i¹rE^]ÄÏ+Õɦ\@¥;ãné´³¿©T÷v<ì|©î0åh³åYÝ·`CÕC¦\N{Ü]¿ÜLEÿÞ—)#AÑÏ&§Õ8 xueu=¸êv;lnRÈ”c=Œ°¦‹"ï2üH³4ºóyfÙ˜ŽâØäšºy.ü=h„x®\ ¯q†LyJG¬×³°]é¸1Ýy…+d–»Sæ%b\á|̱†Â`Ã0ÖrÉûëWÈÝÄ`J XÕ£U,ÆqÓt ÈŬUl6O;¾õ€Õ} y}%d•oiHg˜Ï£wsß:ÙÖ[}ÐdzÞÏAwš/»Ÿƒî>žõ|Þ·€õ|—¹Gl멾˜Ží™ý9ã«« ËM7•Α8ër19u_:ÙÖkãN!s< ª£g›oJÌrlóÁr®€U®†é$ó}U9š€§ÉѤðÒ÷y ¦+¸›l ÊÍ8¶÷i`ßÞ³¦ûϘ9ߥäoįÝù«»ªäMår7¹uÿ8¹éóT¹›€)žÅ8¨g[ïLÎçæ÷4ÓM=SÞtBæ=bŽW¢ïWžö÷ ¿îÙÖó“îA_—ާ*wãXí%o©ÔóT¶tЃm½:;Ó-¿üÈQîÆqÕù°ä=]Û|j¦z¦œÓd¹­c[o0f±O9€ÙŸ­Äóý²ª%K^Q!³?ÐrÇSï÷¢;sÀ|n–ïáËó²ã™î0׋‹î6Žã-’·RKÀYË}®Ât5Ï”óÐxUÀx%ïe䀭VåX~äT4ðäªåLçK³ÜòÂþ_Ä,÷ž›œ‰Þ®Ynrq½0Ëñ®Åt¾/k˜[X¾7Ç“ó9‘cé%dþžÊµL9€D÷¨€§É¡È|6àer+²Þp¬b¸YÝÁ¿¼.~ï&w0Ë1ÝSëzW¸·YŽ-îÜ9àlr$Lw ˜åô6­³ÜIâ5WÈÃä>­FRæôÒ:Ž\Û‘CfSÍJ‹XætéÀjS¹-szIü]_GW#=½Äò¶¥…F¥ìT¶v´ƒæuÐØiÕ—P®ÔÔ°‰J"™ô–$mTÑV´ì¨µúѶP†WµðG²:oÇ~(¯è°˜Ò*C3Ï/ö¡‰ÞLTêã¨ëÛqàw™•ͪ¾Zµf‹'æ“K/òexÑ :•AÎ#N´2½—å;Ê`;µ¸ˆÅ;v`·ÒÁ¹#RÞvì×sG”õí˜ãAœst:Â>Í[vž­2?ˆ@ÂŽ©˜ôLŠ¢ÜÉ ï ßQÞØ ”Q˜žLð$Ê”fÇy`;qá4´fƒ¾´D"a¯AÖ†­L°¤÷úŽ£¸vÄ{~à<0Ÿ­™¢-†rÛ7œ8+v} f«ÿnîê;Þwr\æX펪cXѪ6îTLÙ°Ÿ­­ÝÊEBÅÕ‰ö+ZÙ1f¬þ‡YÀg„èÇ¥…È,T|‘º*¦v`9ñžÝ ³§åÔ°]–0¾ øEüDåÞÕ~Ç{\jÖ e‹÷['!ÃŒßÕZHL/v¬hí†wwºãÀI2:CÓa¨k ì¡.ÊI—âýÆŽ¬”~ñŽïÞìEdÀí¨b=†÷4iG‘—Qó-¸4A½§NœF³ÚËëÄ„Ó膿«OÑòñ˜Úc€Ùž |ê_ÄÒ{{ (âܱõ ùyð^ôŒbžy ¡(·nC‘öyªY±tÅ Vë/¢E½Œo¬”zêBúÈП0ÒBìhÕ‹²Ð ¶"FÕû\°úfNÍÑï!o¨€·Vˆîx/b >ª>–ƒîXq’Ó°á4øh½õ¤º3UÑ„+5’ 'e‘º¢¼„ªÆ¥vCCEûÓ3vLíÀŒÖ¾•¨î(·Nýä@µàw׃CÔ’8'”òÕ$˜„"{‚ûèÌõÐZÖ¥ƒÚð¾i;. ÖÙBfÓœ]»$Ö”ëTÄv\@NÕ*ò v,@NÕ*ºâ1­Öž­;®!)²c9qBëŠÓ<º‚½%Ÿ1µ(¥BŒl˜0k|‡ù˜³¤÷E,üw¬øÝñ ZgRmC,÷w,×­ËpÖ ±¬º¬…¼Cm!àoƒV-CÆZ~G¹ü¥Sñ 1î¡ÒtZÓ»£LÔ0Î5Ipl¾c½õ¾©XNøË,>¡ I‡¡øú¡!ùïV6C¹±º]¦e÷;N´òäþüÔ¼/x¹ÁexßXÙgÊŠí:°CL2ÞÿêE¬1—©‹ ôê;ޣ؎%(NœF1\kC(­K;UºàÉn˜Uõçt`öuà²"›ý;Þ7gÇ‚ÖaÈ›Ãîš›ãËâ*ô¾YLÃû°e±‘%üR#<­ôß1· ”29XH‰.aE92pÞ‹e€ßÈ2I*™f_V`2ÑÈ ïnm©j«Ê¼EÚ¥)R‰Å +k|`¿L ˆ–++Û—‚ÍôÑ:¬î_´7ÌåÀrXë ­Ó°£uÞãÔŽòPžzv^È%Ëÿ¥Âɪ°c>[ïNuÇQ6DŒÈ4Kµ‚z©ô·`OÎN‰Xà[©`Aë4ìøÝe8wD€eG¹“¶_F7ÁeÓN–M{(F°cYÖ~ \þƒXà/›2qÓ~ÇûZ*Rrv¬éDÜ G;pí8dJ³ªé- 5fÙtkÂcÙt‹j¦ÉªâvQW½ 'EZ ï¹è‹Ó$\ tZwÙ‚eó+U›ÝPDd-øÃbíeá 9Òe³/޲m˜q’œnQÿ`i¹žà}[–ͯ& .¬õVWí…"ÚeÓž…8ƒU'Vî,Û¨’Œ‚,²­—a^Š ¯íL-$-MSJìÖvìåÀyØò Tf¹cOµã=²ï(RÄC_†…³ÛÞŠíô N£ {þUU´–ouvì¸ëA(Õ^†â´äÊPµ5,ýÀv(7çA¤qí¸ð—9j,Lð^4Áܥߘ:¦,dZíH•\1 ± îîËu`Çip=¸n.Áp¥©ªK‰ÔB-ÍdÊܣ߱Yoy!°°c?QÆÁ¥Þ£âk©Õú…8Ë,ìØqV,¼¿«²Ô@pAß—®·eÏ®Ëd$…‚„ã|0êµÙ‰ëÒ2D•éØYdm.sá¥^.’ŒWúðÜY‹ €© ‹,Äe.Û*ã±³ÈνL—¾+²;‹‹×e®¢²/_Á™Üy¿vž'§vrÎþ/²L;‹ ÆË¨@fEQ.×Ƀ¯Ožo3^<_,TÀ<že«Ü‡—t –³Ì ²Ê/p6­K#3™[óY&SfdçôiÇý×`ÉŠì,2>Wæ÷¨õåH1õ•m†¬ðÆQ6×Õ„úçƒ{9Yd'î;S6î2—å„ðÂÁãÃëË”VÜHd¬˜l‰Èfî\ûÉx;Žï<¾Ož/?÷ý´ƒ»ÊR›ìÉÅ¿¿ŒEÆãª¯,ŠÈö\º}e2*;÷r²È˜\æèê¾³ÊVWc‘]y¹ñ}ßXdWvóË\§]áwîíäñiŸG;eYv.änÜÖɓ秲4ØÊ^Æ¥ž,®xW3™D„Ù¿Èfú0Û³±ÈN_êâž;#ŸÙDÆx^ZÖ'Ü>Œç¥®º&;³3ž‡ÉZ¤Å÷ãBÙðËÏccÜÿUfKGèP¦<ã}×Ôuaô¿_WÝ™²7“3¿×ãþïüiíd<×Éë8>s¼Ü™Ç÷¦\úɵœŸ¡²@7gÚrãJ®Æ6͸ӦCeã&m[3®D¦íâù,cß̆C÷Ûd»¨o Ìþ¡.ŽwS¿gêŒû¯²ò¢ß‚ñmcô·S×ã”88דñ½hžÊÀ<ÖΉý­–è©ìÛÉ<~Ëü"©3”ÊÀ 'ãN’l ›“‡?JËdÐ/Ú¹x´ Ÿ‡æ` ãyhv¦.ÒMF®]'OžÏ|dæx>»ìÜÆ•ýíÆ2ŸÈ—ÉÀÑæbgØ‚l¼†~3ž%Þq0/&#—y|5†mÊÆ­œ,÷g™dÈêú5_)³ZIÔ—râñ˸ΓåûÉÉdø:mA’~ Ɉ'¯×vFôu.r1Nü÷œ¶A[“eþ½ó¸>¬65Æ+íŒÐóÁ‰íݸð|ÇÃóäÞN^<Îw)³w°ôïYe`2ë8O`Îo™ç"Ìõm¿ÌV§g¶s~ËÌ—ƒñ<6Yü›ÛÃëd%yÝXm{TV/ñydfÊeÖF"}(Ë÷”->¥.í;÷zò mç«j‹-þ$5<žóOÊ\hãÃ÷­3þµs§mç›’)CÎÆ°AÚyœ\ɨï.ÜŒ;m†ø¾tÚ®˜Ê OcŸ³É0÷Êç¥òYm²: ãù˜Œ¨Ô¬]`Æ'¨Ï”¸Ì%Ëx¼óø°Äs³çÍøUVÇm0ÛwYŃÏË|=«Í@fù–¦ð“Ûõù~Ýž?ã]¹Ûó¼ßÝž7RbôOž¼¿'¶/cÜÏaÏŸñ¨<ìù2µsa{1† ÒÆ2y˜2§OûäßWYEµÍ²ùÄ  àθþ¡ïÇ@ÍãÁýöM,dú°Ì·% P•m£Ø0/åà4N.´‰ªÆµžÜÕ6ª)Ïã~M}Ÿã=;cüšú>Q–òàVNî×Ésœ¼êÎ÷Seå…{:yæßɶË6 ï$œ™ëñÁøN¶Í2f¶ŒïóaÆg²ÊU cüxš­V7Îýäúa\¯mÇ1ñDË@”ÇÉ/6.íd¹EeèòDÉâÁ2**Ó(\ÉÙXúû;Û_šŒÏH©‰qQ[0cµ cìŸ&÷'ŠV_ ÚœMãºNýä©¶f|ž“6[Z‚%Œ÷kãI3}¾Œo”¢ë…‰ÀåÉ ¬ÏÚCO2Ÿÿb|£]O3yãàÅßãóg‚†0Ǧ–IÁ¿ʲŒïÁæ+óùWý>Ô&£hÁ–pb;¿…ŠÔe‰—“)^…ï‡æíËøZ4q'3IÍxÐÆ­+'Ú¸%ãRO–ñgçNfÈŽƒçÁŒ?¼^¸¬6¥iÿ¸h˸ó¢û;Z*£HêRF ¢Fà æz›™5Wµ‘+Ê2¾ì<>,ã÷ÁcçÉïkcÜ/Û/Z´ÚYÆëW¦wòýze3'ÓãéZ|ßú+£[ÒÉ•¿7aS8¸¾TÙ݃qÿ« RQér©ÍÎ_Dv·òßwãÆ¿ù®Êö<ù÷ðýƒß¹¸þ.š$<>ŒïòûYßD[½Ü™ßƒ'mö(zU^ß4Ù^®¯Y¿F–ùÈÎjë××Ï#Œ¶Ky°}ãýÐý–ÂɃն/ee¼7¶çòóÈ<É…ÌøsY*ã+_ü}Êîr#ìàYO^üýÎû…œ'ÒŸG–ø`ô—^ü}•IF’TÏ]KYæ+UKÄ…e¼©û·¢ù;Ë÷Wu¼îl§Œ.7zδ1Äþ£ð  ¢Ês½úrÂ÷SM–U6V ˜ï;e”®lG|EXî_M*K›PÌ'Ì÷?e³Qäû®ù;—O{e;¿ÍÇØy•¡„’ª,ýIÕüaØf“m.¼šˆ,,ãù΋ÿžßGB]l§ñ<9³½-c¶Ï¤\iIYÞÔh‹XTæ6Ñ´ªíx€sSnäÒ•;m ù½&Ú„V-¤. 2$”é•|‹æ÷˜:ïÿÆ2¨š¸+Œû¯â5¸ÿª^#²ÔõúðצÜêÉãÓ¶”×C›Q³¥Ï“ ÏgLåš?¼N–õG­œ/ÚPR6:¡r6–I¹“a ªù Eó)ÌÖ²Ðÿá`ªÆ „s;¹|Ú m3Ù¿P¦{ãÄ÷Y³r…óu2l-uÿB¸ò|×2'·zrÿ/ó“Ú9ßRYpa¬GÀlG¼]eÁ–ùIÕñXeÁ7æz¹v;æ/ÖnçÏ|ÅÚí|¸>®–ñ©;¿Êû?ìü¸þÝYæ³uØùUÚœ;æ#Öa÷›ùuØùÑÖ±;µi}˜ùuÚùuöŸÓÎOm[§ŽG™ëã:í}áúxgYÿÕiï6~`cÚy¼žä„6ìO¦¾ï™ù‹u™L:óv–ñéeæ#VÍwNëdÜßq—~ÿ™ëçºôûf ààÖNjÃÚ•'ýU¹xu½^4¢i‘°Ù´ÃæsãN[VÄw„燧LxA¢ÁƼ£*¹+2ûÛÂõ¸‰H«ÌüÁ“¶²”)çB8™|³ÈÊ«+ço²Ð¼Àœ¿É²œÜ?íƒí¯ ó›æ?¨lýÆH¬æø¬ùÍÆç?WÔ;ey^-q}$¼hcKYÿÂüŦE@àæü TÚ–êúYeò†mkV‡%ã¹¾oYm T‰ü}Îç…ù÷9ÿ*Ì7lZ¥S4‚uÕdé_¦m-ßïHTZ7e<_Kçó±ñµÀÕÕü{ÌhEÇûÂxìÎ2_mUÇÚ [×jÏ{òþl<>íãh_´}Ý6·ÕÞ—ÅûYõýàÄÞ—ò:r@(ùÎÆóäœ^Û^0Û9ÿ­Ì×kªWj¦ímÓþ’-T‰e™·¦÷¿~—u2Þ‡‡¹žoZóR8‘8Xæ&#,¼h ÌùZeþ\ÓüË¢ûÿÍlh‹p0l”»¾ÿšð2û´©jë»ñ˜'¯´ó@ÿÜÔ–T8ñxŽg´aؘûùMmJ…S99³]m¸oZ#TÝe™ß4­îžód<ï‡Y?qðzm˜‹îg7µ}+­²¿Ÿj³£û×mjÿߘÿµól;Ó–W~³2®OãñE÷—_î|Ÿ7Æõj%°0žßÒñM÷—ÛÒþ´AÇ6µe÷«³?z˜ù[;WÚDó{hŒG6ÍG,ºÿÛu?<Àì?6R„9~êþïË‹6ź¿.<Ô†º(/þ{Îo¨z I‚¤,ýK×ø¶°ô§=é|¤_´ÞXƯ®ù~¥s½ÛuÿºèþëËPŸI&F+,Ï·Ûz²CŽÚø>;ן;ËøÑ³ý>jÁ–þ¨«@é\/vß:ôg…ù}v®»Æƒ…+m³/k—ùü¬µÃÖZe0Jçú¯›íPçú¯H=‚m6ã1º_Ic²ô¯]ë-…%Þѫ·:•UÙ€\iíçÏõ`×øjé\ïu[oõÉçÓ˜¿Q:óÛ»õçë±n¶Pƒù<;Ëú´«¼Ú¾¤®ù[Â2?èšß^×C/c!.ÜŒ¥¿â6-¹}Xú§®ªPeTþÞÐõû`þÊËÜ_ëƒù¶Â¸?¶s><°°ƒ¾îï`=ÓËÌïè¶‘ý²ôá ›pޝcð}}˜ûg/0×#còý˜ÚŒÅïoãòáÊßãüf^<_͇´)çx;ᛵ¶sÿg¨:P™¬?Ïš¬ÿf[¦û+;W2¿ŸÉùÕÐ2ô¢û "ø&ÏëO¿¥Ðöç¿aôcÖ>4ú¡…Ícôc´vÊ0»Á³Œ~h²bF?FýhG›ýhôC2££t´Ñèy…Ñ~sñý£M„Ý5»ŠF?íâÜF?•¾~ŒhôcÔ¶q´Ñè_ÛcôcD£D.£’ý¥£-m4ú1j0˜Áh¯F?‰k5úy)÷ Œp°êP£ÍÙV£Ÿ—ÆAæ,xÕèç¥U_¢ÑÏK4¦1¢1 ú 5úy©¡ ý¯ý¼Dë|{jôóR†‘ úu5úy©ÆuÐx-WhôóÒÚÚhôóR®;•ôÚ±Ðè祳ŒHjôóÒÚÚhôó’<£—& Q0n¨ÑOå(¨F?s5úQ…j5úQÛI5úy‰v%˜+¨ÑÏF0=A¿¬F?/ÑÊ䡱SƒEæSjôóR¯; ‰•¨ýh^°ý¼$÷ó¥ÚvêÍþ~TFH~^ª ÌëÔèç¥ «ZßÐèLjF?m4(¡E~ÆcØs•h“a†=„5®ý¼Ô޶q´ÍºÓ‚¡z"5úÙhîTòA°A/¥F?/µ¼Sï;÷ß©ÑÏFý1ðP£ŸÅ5¥ý¨‘¨ý¼ÔêN=½”Ôü#¥ý¼”aFA³d‡vÍI¡ÑÏF­ïÔ#ûq¤(r¿´®ƒðW0+¥ÑÍùiô³l$hZ†îÞ{£~´£m¶ÚÌÌ'·Êک僎6Ñõ7BâÀF´¦xèh+u§vPÇ™auI£Ÿžø…Óèg£„³¦‘Ò6ªk§v´õ²ÓÜÁŠ2l+0›¡ÑÏF &˜kÐè§ëµŒ6šm§•_j~aÔûNs¼„0ÅF}î4ÚNó8râHô4úy©¥£-m4±ÀF?¾ÓDæK4ú1¢ÑOWiýlT¶–v ô 4úyëàrß©äúµÓ¢M†¼g4ú镳 ýtÍè¥ÑOW[YýôÊy~zã܆F?˜3`ÆB£ŸJ:hîÔÛN+´^BjÿFu#ä»tÝu£fPWAýt͈¡ÑO×úh*èmTËNbêÒùmÒèÇìVÕèç¥rÒÜ©·ÆÚiå—hôóR>¨Ò°¡ýhíÔúNýh°  ÏT“3å¡GÀCc§JÀúcF?CóTiô³ÓÜi½¤F?;¦|âyp;e6iô3T¬E~†©ØÑèg¨·§ý¼˜ÔYàA: Pé—šBŽK=°BŸ –4úÙqµ ±c¿ã,b#ìÀq`9[k=°áw©úEé¡ÇÙºvDPeÇ”,pxl€hC@(ý¼ˆØÆ#øO£Ÿtø)G£Ÿ'DûyEؽß1çi4ð`/N Åiô³#Ìgk=±ÍåN6sçAHoǵ6Ć>ÃŽ¿›ÑÏŽýzÿ©&´ã.õxP^ ëpm­4úÙ±à$©·L£ŸÇ8pî­ˆîXæµØOœ;ÂègGÚ¨-vÚGW!>ýì˜?8¬'ÞãËh¥T5~^D aÇÔ_GêíXÎÖvØa@YBu¿ˆ˜ÁŽ îÔó¤ÑϪBÉ‚ìå%4É;ýx¶–³µÃXavEŠöSW“F?/B`èvœý ÍžÑ2æ1UPšF?;vˆçS.žF?/"À°c¹^á}5úyF?Ê[·a=[~Wm–*ü?8ÎÖ³¢¾ýìØ¬g«t}ö³U¾î“êÿ?HýŽt4ú™×kåÓNœùÀÕ7ÌjÀa‘F?;RÿŸª×4úyûý;æv`ÁÁK¤ØÎŽ÷p<“JVÒèçÑágIä‹HØ1•+Z)FM£ŸWÛ°©ÀƒfÔÕ¤ÑÏŽ}¨úÿ†³lˆÎgª8ˆV îHÁÿ©ðÿà:[)øÏw’F?;Ö~`?Qn݆«l8UáÿA¹uöôzÐègljV} ØÞ±æ×­ØÏV*üÓ€:;;¦³U,*6¤àÿƒ#8Û†X‚ï˜òµØÊsoEòŽ58O\ûï"Ž ìÔ¨¨uûXýì(/¡™NÒèçE¬Ö¶³µÍ|(°Lu›ûu ˆÉ7³ÔA΋Ȅ˜&„Ka›+–bkv˜ ûS£mˆ°õŽãÚpªß­Ohô³c;±Ã;€“Oýì8ÏÖgÎEiô³cB«º/-µ?x°•dž4ú™ÆW£Ÿ[;Zú§ÑÏ‹(hŸÃÜ[P(¸ã=…ÛqÕRлӴmiô3O£ÆJ~7£ŸåæLó²üõýÀQü±ðß±¦ÏÖÓàœF?;–ëÀ |ŠÓèçE,ëwlëÀ~¶Î~àÊ'îÿY;æëÀ™6ÄŠ~DŽߥ†0~æ2Ë$,Î_¤Ñφé:ñ<¸Œdžjô³am¶³uœÈ›Ãy¬ý˜~IæÆ®K¿*®ìXѪŽB0úy”öiôó"ÔVvLÖçL†F?;–³µž­¢þ ÖÝŽk=°Ÿ8Ò‰óÀy¼ÐÊ~v¤;ÀƒíÄÞœ;BsaÇR¬ãÀ†VZrÐègÇÕ6ìzŸ̰ çs!w¤;Àƒ£¸ é¯^EXwïXÎÖz¶¶y`?[WÚK’[9pî?„UÆŽ4VP¿!Ô€íHc…G;pÎiô³c…à?õÃiô³T\D~VQ+=ý¬b¦*XYïØÒ+ËO£ŸÇ—Ã4úÙ±¡•>’L‡ÛqõŽòNnXæ ¨Æ7MÿÕ# éð;γu­¢ñ®¥¾jô³c)'¢Uœº þ?¸¨ðþ™F?;RÿŸáJýñ|Nhô³#ðgë\NõP›§©Þ–~`»œ”ôçÅîúŽùl¥À0ïžvboŽuàÚ[û;&Èãstfˆ~ÇÚì×"5ÿ rñw¤Jÿƒ=Ÿ¸œgëÚ7«†ÿƒålm×”ôþj^C£Ÿ –bn–t`Ã9s F?Ò€³5úÙ·N‘F?ºcF?öuàÜ‘F?ÊÍٰР(¶G:q8ëk?x¨ÁÁƒœÉ¨ÑφíDÕéWK!¬á7žÔ‘ß8Óg€3{šýÜè0íøžNž/ó=x¸´“k>:ýÏòá½¶?Ød™ÊÐÞXuü®ÔíWÖ‘íÜU׿)«îÿc¤ºÿ«î¿r¢ÎÿÆ)Ÿœ?í…ºÿ\íÒ›7¸ß4:XuûîÔåß}€ ë_•WÛ¹P{ãòiïåäyp¥NùÆåÃkîÜÌ÷àá”OÖëWû™fºüjeC]«hx‡®@«î>‡kÕuÙ¹|Ú+uä9€ÓèàAÝzÃ4ÚxÐwbãœO®VÝþÞ”U·ÿå~òü¿Žã©“»³êø?¬:þ«N¿z[!öpòÜ™º¹;§urî'WêÞ«u vdN èt0žÏƪãÏP‚êμœ¨;¾q¦.¿úèP§ààO{Sü¤ÜËɪûþòØ™º»;g곿 ÐÁªÛÏïBØ%Äû¨:7;C‡|ãq0ëÜvÎéäž?C/S×fçBÝtõ¢ÏP2SÕµ9x¬:ëy÷JMuÎÔg(iN£ù ½Ì:ýU×üáÒO®Ÿãq?_ÖiÜ©ƒnÇÏr²ê¢«O}†’ꀘÏÐÎðQظQwü²ßÛɃíÙŽŸªKÎvê´î\>ܨK®¿Ç:³—‡é ?\ê¦[®>CÏ“;Û³ýû>OžùÃl|….Õʪ‹þp£n÷eÇ÷rò¸NV]neêº&ÓíTŸ¡+u¾›ßÛ¦®º8«ÏÐΕºÛü=õÊ—þ}õÚyÕé3t0ÿ½úªPg`gÕÙv|O¦.õeÇv²ên?¿gºÛÆ‹ÇóûSŸ¡lº—ê3´sé›nµú í<ÆÉ‹:×ü~ÔgHt§«rI'×vrû´·yòZ;Óg(góÍQÝÙs?Yu·•›él?œêÉ¥\?Ç·Oûø0ž‡æÕ›ÏP¶þ_}†vÆý-ªc¯>C;w¶s¼QŸ¡—é3´sù´7ÕÑžÆóäÞNÇñÓt´–ïegÕÑ~xQ'›ãµú åª:¸ê3tðœß¨ÏÐÎ+íLÛÓ<9·“ñ>7ó½¡ÏPÖõ ù åf¾5Ô½Ýyªî6¾Wõ:xœ<¨›Mhõz¹˜nöÃ2ß;øÓ^¨«Íùµú ÉkW×ÉýóìÌxJ6ŸõÚy¤_Ù9Q›º¤ê3ôÿv/׎óP®€çE…`ñÍ0îê4Nþãþ÷(œÜá·èªc˲””ÓãÁž!u¿ÆrÁq>Þ¡ÏÈÕQ3§¹ÛÌÍQ¯Ç½»=Cæå.å2r²q~Æž!õþ]¶qÌŸ¼9Ýì*§gŒ=Coî5{†ÊécÏЛ[Íž!uÙó…Ü]5sª_÷k|W1{†Ôƒ9Ö›Î\âß×#”¹±âVÜÌ~½`öº`þã3æ;Ôý¹¼Ü£¹'Ʊ?fÏP}¸ÿcÏ:s‰Å™#úœÞô }fn/çkOÏ:s™Åí6r°Ñ3Éž!õ@®3ÎOØ3T s»Ù3ô=CµœÞ"æüª™Ýèvyÿ.O5z†*s¬NϺ?îyïry«™ ¬¾ÆëåÁœèzŒiö!—HߟÎ>=CêÞÝc¹WQ3Xü w¹Æë5ÞîyyO1{†*$8=Cf¼ž=1èR÷åÝ=ñ~±ýaÏú)—¯××êî—ÇϽ‘kí{†Ô¹ÒìÅ)'×=쪃ÇÓ쪧'‹=CêRÝ9Òø=²gH½ŸËSž¡:x¼Æž!u_î¹Ý¹ËèEdÏú©—¯×—ëõq<§îø÷ìÍAÏP=½›ìR¯©Æü”úiîÜ~­Ó#ÅÜdñîU¾çÓ3ô=Cêg¸cÿXÏþ‘¹Q'—ùô ™§›9Ë\ŸÐ3T7{ë˜3Õ~§÷=Cfä#ן9Síwz·Ð3ôóUêÞÜ9Çè­`Ïz-1{†ÔåoÈU~{‚zwr¯GO{†Úé`ŽÕgô ©[s÷ßeæ Oš9Â8dÏy¹—½ž¹Îç|…=CfäòbýgÏÐgô ©Ëïòt·ËÌ~=˜#\hæÓíÍ>~º›9¯™#üzJ®ðéR9ÎâŽÜ]ì?Ø3„HÿÞž¡Ïãä ¾ÆãøQÝ‘#ŒãWö ©×Ͻm|¾¹ÃÇÌ~}r„OïÐìjæ6‹Ÿk¼÷žjÌG©ë5ž¹ÀƒÇ«ìjg~‘=Cêç_¿ËSž¡vzÙ3¤fN0z.Ù3ÔæéÂ|Ògô ©GuÏŸó=½§gH]wÃë±~±g¨-®_ìjÒC´‹ó=×ÃNϺ]îÈÆ÷Áž¡¶Ù#Þ¡ÆÜÁÓ3ô9h¹‚§g¨Ÿ^?ö ÜÝÓ3¤®—ûeæþ¾žÈõÅñ({†:sOÏ:sz™sz†Ô9¸èMdÏš¹»4ækÌȹÅñ{†úé¥áý0êUÄìê§gŽ÷·¨ÛåŽ\Zôä±gè3ægÔu¹ûå¹Ý±>}.'W—½EåäâÒõäâb{ž!uïîq/3s‡ÅõçÍ=1¾OOÑ^_.îéRWŒãø’=CêYÜ 9¾ìñaî°¸ ‡–½PÌË 9´˜ObÏÐg\ŸP×înÈÉÅü ï/Qg®çkæ‹ Æq=œ=CŸ™+,®ÈÁÅñ{†Ôc¸Œùö ©ÇãÞ]ýœ\Þ×Óþ?ô ©ût葉'gÛ{ö ©Ûp÷ry»7rr±}ãýæé~®×3×–=@˜¯@BÝßÛ3d^_îéúŒù‡Èº*ÇÛ]‘CËåž¡¾xý’=CŸ™[¼ÏûÅ|‚ÙÆ1Ÿ ÎåyziÙ3Ô÷ù<è:¹³§g)uoÏÐ8ר3¤n»#G–ßî/QÇú…xØ¿·ghœãyö ©çø|z†ÆÃ\Lö ©GwÏåÞEó{uæÈ¾FϺw÷xÜ«\Þjœ‡½0ì­ùoÏÐgô ©3'¶xÏzÛ8ίՙCZNO?¹¹¯ÛpÏk|ýÔ8?œ/;=Cê†\Üç¸wwìÔËŒùg5sx?O÷jîm=Cê^Ü™»[O¯z†>ãþ¿ËÓ½[\ŽwS·“‹üy¹Ë5^®ñVÔèR×ÛÓÝ/æ^fô ©ŸÇ]0Žß#{†ÔígŽ3Íiöèžž!uoîyy-õ›;ý9ÎÈ fÏ:—ç8=Oè'7Ÿ÷_¨cÿøyŸ\é×¹Íè`Ïy‰Ù34x=îô © Æ‘KÍž!3rÑ3ž¡ÏèRÇüËXÌ¥fÏ:·—ê)9Ðì§zz†Ô1¿0™yz†Ô~{…âüO¹ÛêëõË^ž!ót¼ô°°gè3îGTÇþjžÜ|ö ©cù~ÆùódïêéB®çßÛ3„ljþÞž!uï¼fÏ:ÎÔ¯go•½sìRçòz{WpÿálÜ~°g(¦Iò÷Ëž¡ÙO/z†fg¯{†ÔýõWËû5z†Ô±=UwŒ³g =Csœ÷ƒž!õºÆcÿ5ÇùÿÐ3¤Žý“:×q~_â8^øŒûÍÛ]êåk¼^ÿ_ÇûãçÅù½:öGsžÏ‡ó{óv/{=Î÷ç<=Qèšóô¡gH]†;¿q¿þ}O|Æýê\ßÙ3ÓTXÿp~¯Ží£:—xb<¯ž¡¹Ù Áž¡¹Ïïóê¦fÏz›Ñ3dÆÿãuö ™¯ñü}¿FϺN÷¶ÿ=Cëwz{Ð3Óh•Žõá3ÎçÕñý,Þß~z†â1¯BÇöý3îŸSÇñÅzNOz†Ôq>¢ŽíËâõ¢Ó3¤ŽßÇ:½–¼þ¿ÊémÂýoñXÆÑ3¤~–;æ7>ãz¾:Ö÷UOÎÇÕñû_§7‹=CŸÑ3´xÿõéZ c§gh5ö²ðúºº^Žý½:Ž·ÖÙÞ²gè3z†NbÌéZçx‘=Cêñ¸óóp~ëô aZïïíZƒÇGìRÇþá3z†ÌËŸgž^œo¯Éó5ö ©óó‰cþd-ö~²gh1Óüô ©óûÇö[Çwê‰qô²gHËC¼Ì¸þkžîÒÜqü±öé™BÏ:ögêØþ¼fÏ:¶/êU.ÿ÷ýí—¯Ï~Æù®:¶¯ûÇžö ©ãûUÇ÷ùçŸ1 ºé2ܱ¿ÇcooÏÐgœ/ªã|ç3ÎÕ­ºû5¿¯Ï8ÿS—î®×xlOÔ±}UÇ|Çgô ©ÛÏÝíýà|O½íßãþvõÔ׳giÆoÏy»§çCûÌG±gHÛ+u®•ûWö }FÏzÂè½åõµÍçMOÏÐîì­bÏ:¶WŸñ<Ó>½xìRçòáý­§gHëëéMçõ³spz†Ô¥«Ñ3´ùüÃéÚ¼¾xz†^³gHÇ+ŸÑ3ô¹]ãxþä3žÿøœ×Wâ±<œ¡gÈüßöà3®/¼þßÿц¡çÿ×3”÷D²ghc‰²gH´³¿‡*Ùµ“Ñ~èõª9–wÉ£g¨ÿÞ¶¢õ3Íh´Éýz†:# Ñ3ÔÙ¡ž!Q5ý·˜zÁ·Šž!ÑÈV—œaGÏ(zUŽÐ3ô©dOM °gˆóŠìú]ì¦fÏóyÙ3$Ê6˜^3x_z†ÆK0/J—YþcG·þž!š*r¶=C§S­"ƒ í¸‚#Š$y¥¢gèt7 ghT¬‘è-ô%ô§gh0ò=C¢h±øTm¬.Õ,¦­Ú9–['\µ§=(Oµ£€Ð3ô*OƒEµ«šM[Ù{€†™<9~•k»j«ž|eÎJ gh ÌI gHÔmŒÝëßé¼–‰ž!Ñ’±Ì §(Oªc×Üÿž!Ñ0ý·{•'Ë¢òS5Wv ä=CªÌ÷Ï+èè:½vèÜÒ¢gHKwcæ=C¢héøT]3”G HÍCT^<íçSö ©2a?·|èRmUd êùʼÛ=C¢™ö¹gFÏj¨ÊT5{eË1éLßÏs ô M6DãßÏ‚­7z†DÑ Úªž™ïç™ÆFÆÃ£+éa¢ýÑΠõóï¢]à(Ÿ¶ÕŸiªÈ~Ôÿûwz†&Ñ3„ëÓ3$Ú2–ÙFspy惨“ÌôƲÎ÷4'›šrû¤²±ÈkÿÔS¹ž¡gh²Å=C¯:3µsMFÏ(òÅùËAÏÐÜøÅ¡gH42N;·èz•“1÷òûwz†D5ã¡¥gHµTÓÆVWíòi2Eúh ÕÊ¿—[>ô ‰â¦Ë¿Ó3´é àHµT[^™·,6°£gh±/=C‹³Úóçz†Ds¨–íù =C"$gßç:ÉÆG=…Ñ’sbŽrœç׈3Z {_ô ‰"Øó2}þÛÏ£Åçô Ì^ô ‰‚tû¿Ó3¤ÚŸm̆lô ½Êù#Q}LÆ‹&£LÆY¼ÿï0Bzö¿Ó3ô*ßaêPM”·å‹ž¦êH’mÿNÏЫ|Ž]ÄÙýïô ‰¶(÷n¢šÊ}z†Dˆ—|eDÒqGøÏf3z†D%•ó¢èOèEŒß§iZÅ´#ÝM8ù„׿u$ô ‰šõLÍ;6q¦ý*{†Dõ­šåSe†éQù©»™Ç¦èz•—ë7gSÑ3´™]…ž!Qí*d—E$_Å:ˆž¡W™!êÈâüý;=C¢ˆ”;ʉQݪfc½˜2îmLy]D´Æ§y"7Ë¿Ó3$êM5ll­O9Aá›ëßé•̵̼=Cªýoóx=C¢acÓÆv*çkÑ3´9[‹ž¡_‰ž!ÑHmþ»XÖOù¡ghs;ž¡½N;N^ÁÇ›ÿ;=C›÷ϱghór {†”û(ì>:|tfþ Ër29Aùä(»d•¬œ£‡u å_¨t{Ü 1xh@×Pþ³ŒaBÙPÆÚ5š±qHpDÝÐóãiû†ÌõoͱYâqOŒ³e ±C?ÞöËÚ!sÆÐ‰3†N̺×ë§Æe}uÆ"ò±¶™bÌžßñRã²¾º7cÌØO…ËúêeF¬:—ÿ‰AC‘¹]îˆ1+§lh^Ži?uÆ~½ÆeuAì’bÑGdžÕ½¦—ýÕµ¹Þ»¡p€z!æŒ}Hˆzb"3cÐ^7Ƙ:.‹ª—?X¿yhÍz"sCÌYp² È¼`–ãÄtnÜfÅŽ¢|Û?z™ë¶è£`=Eævy0v«#æ ¹¸è* £6eEù¶6ý w¹Æ+b™è‹Â¢|ƒf¬Ó댱yÝ#&~…èp´™;b›XÍ“‘Èb\¶7_ãõÏå/Î+ñ†ÑÃú"÷r—k¼\ãù}ˆ×V#Ö@]înÆeû§3¬÷ĉÛ5~b«Ž3hœ&¯˜6õ£Î(Ë8³„í# ̱=UküÄ6åúˆR£ø˜9!ÅV#s½Ü/3–éõFìR‡q[z4wÆà‰÷Rã6uéîz½¾]ã}»g¹lã¸MAë·¸]ã ïŸ5P¸m!V‹ü¾ûólf¦£ëÈœ±'?nßÐvdžÃ½ªz –Cü ö Ǩ<2¯æÞ×ø^ê‰X1cªÖñ¸<µè]ÕÙ}dÎØ–4³ýÈX&lïÐ$Þ'6éu+—·»÷¸Æ+…ýÿ‰?×xeÌÓ [sOÄ"•ãŒA©Œ™E’ùa Òqî^ÝËÆÛ#n{]Þ]]¥Þîçz}¹Æ'ÞO;Þ?uCL”zºû5ž±Q¼MšýHbÆòœ¸4$åÏâx"&¨o3nãPçò·åîÝ=÷´ñ‰X³Áý)º’ĸMC]—»#Ö§Ä¡D…Iæ\¿ÎI*“Ĉ¥,'¥Iæv÷êž0*PœdÎõMÜ»{^Þf<¦¯®ˆýA ú“Ìóò6ã±u½ÌX¢×¹¼^WlŸÔñX¼8øœ˜ÞÆÂ*¥0bsOŒŽúÏX×9«?sFá¶“úX¡’9c@x0+•̱98žA©’y!†¿Ÿ/öæóVã6“zŽ'Ѭd^·™ÔŠÉV–+™ëå9ÔxL@ÍXŸÏˆ}A1.*–ÌS’%sëî~ÏËKÿÞ;óºü.O7cg°þ mÉ<›{-õ;£Ka6å`~¡2€Kîžý;¥Kæ9Ô˜0Owkîqy>_l »—ÌŒ­y½m<ë—b³‰íú—Ìf[3P/óKóº6wCL êSQÃdÎX+ñFl fÐÄ”›ÝF—ân—Çv3væuÆZ½ÆüCÝÜ ÉÜ®ñŒñá´6;™ÌŒ•¡£îÕ±?\a1Sû_43™3†D̙׳»×£ž)6ë,°ycb^gL†xO5b‹ÛÃãWt4½±1,i2îÎX—Âó'ô4™3&âuAŒ†¸ w»<;3‡ºbùª§»"fÛwÜaŽõ³^Kt6¹ã‚ã7´6™b[0¿òż &ÅMæ…XÔ‰¡ºÉ\š».w¿<º{÷ºÆ7Æó²-+œÄ˜ŸPÇö· IJÅÉ<¯ñÝÔßÏ8Ësž”׃¢^îÞÝs4ÏòŸ'&åuÌ_´y¾ÌO6—k<¿Oqÿ¹WSã1uÁø~èºÜýòèî‰Ø®ïßÇ)„Äý槺ëåv½>?âŒUyÇ Ôµ»G¹¼ÓRèü=‰3Vê5bÕã8ŸCÕSƶl:c6ÄûQ#ÖF± ⌑àc,|ÊX–Ig¬‚x ÷¶ñ†Ø ñø©;b3ıþvÞÈÂâ§<,éô2#v±ï û)3Í’gÑýr?˜§{5÷¶×㱌~¶w¸³Á\‹»_ž—3¦‚7ºœ(uìÿÔírìÿð´Âß[ÕÛ©ÙZ'ÆÇO'GÌØ’×;!ÛÍØ~x¬¯3FðÔ@©óû,óe TŸ\ž¬R÷k<×ïÓYË(õºÆóûzù«qj ú©!c Tßü|¬R3ƒF¬qßüü¬Rçò¯Ë»«1ß¡®×ø¨îXþƒ±{§jüXëÂuÿ~FL±š1Ø¿³J݆{2ÿóê§¹3D\ ùEÖ@™ç#qj Ô1 ¸¾Â(u)îvy fµV˜?•ûÆô¨WuoÆ$àßc¾ð±ÒS¥®|,Ò£»w¹¼Ôû}ÌÿøÙ7׿X¥žjÖ@©ks·Ç=̨R—ínõò5Þ‡{à±}üþYežjÄ"Áëã¬R×ån×x·ñzûÇþ‘5PêiÆü¾º V€µS¨2/÷B,ÖGÖ@©ó±ûs=€5P8lú{k >£J1%¯Qõ5PêÚÝóòjÔ@©ëÏ=š{^ãû¹÷º^¿+€˜RÖ@©+ÃÇö…5Pê~Ä  Æƒ5PêeãˆÑ]'”5Pê¾Ýkªq?†:‹·k<Î'Õã×ø*—m5Pêø¾Ôýr|_ꉘÔ¬°ê3®7¨ër·æÎǬÅS_Ï(uî1Ýóz}>–ý0—5Pëáþ„5Pê²Üíò@ÌktC¤žˆ=XÇù}¨÷cpj >c¾ÄŒ×¯ãü¾Ä ±ïëçt3æ€ïó'êø½¨c¹:c°Ye^_LÁ©úŒûÍ×xîüþ^3Æb0¦’5PêÚÝíz}¿Æçåü>^#I×ÇÔŒYxÇ«êqo{?¸_R]¯ñ^./÷ÀûG,?k >c>E]«»Mwþ&c]Y¥.·×?ž&þ½5P¯yÿ‡úîŒÑwÄB°ˆ±â9ÝûQ#fXý,wéîv¹÷¼½Ýq<øó%f¼_Ö¡ê<4qj >#¶X]Êåk¼vw»^Ÿ±â³fó+ûwj„0Ÿ¢ÎØ1c ~ø¾0¿¢~®ñZ.ow¿ÆÇpOÄBàø±Gê2ÝŒix½÷îêyb>¯ŒIÀñ'k ÌÛûCulOÍ6¾N,ÃëççfLÃëØ¾ª3öà5æKÔårÝî6Ü㨠b Ô®Œ…f ”:öoêeÆý&êXÕ£º×åmóêþ\^îŒ ©¬5c ”z5j Ôc¸×ïòT#VYý4wyÜíör÷ëßëõãz}ÆN°FþÔ@©sýUow»œË_=Õ¸¡Žã7u~? ó[§J½Ëå­Fl³ù¯—ÇïòP#6J]w»Ü»{ ¶µ¬úŒ˜fuìßÔå¯×xÿ¹36dðxŠ5PûÔì±Jý<—gÆœàü“5PñÐÏß‹x ¶ä9Θ—׸goÜǨ˜6bMSÖ@™ÿ;¾0w5¥¨2ÿw¼Q~çø÷ãä´ÒCÿ·¿3Ëÿ­ÿâ¼ßÒüß÷a.ÕÝîÕÜû¹±+¼•5Pæv¹Çû­¬Ñ@ TL#=g|›3¦ÚüÔôlÇñúÎÚ7Ô@™ÛLã|5Pæ\þŒ)d ”yMõÀ÷!~š»-wïî9Üc£Þê‰ïOü wîVÝ3>ß<µMYeÎïs~µN¿æ~®ñZ.¯4jÜÓV¨í@ ”xãû[§f*ïß1ç÷¡žîµÄ¨2—k¼mw~›ç{¨ro÷ªî=Ô¾qkîü=©§{\¯ŸÏåå^öú|Ò6§åº–ËÛ=†ºâ÷%þïx=¦éò¡MÖ@Å4k¥²Êߟú¿¶yÚx>©Æö5Pæø¾ÔÿOšçåø~Ôû§Î˜mN þ(ó¼¼—:ç¯Ü×xiÌ/ VÍ<«{ýÜÛÆ3ÉÅ\›»_Žíýç…å[O-ÕÂòçòÇöûsîè8mùwj Ì«»w£*§-:—_Çý}¬rÏ4k‰¬oâüüâˆS˹<^|~Þ?ʨ˜æûÇöñsÎ÷˜ŸùÓˆ§ÊÜš;¶êu9—×ëœï1ÇþKëש ÁÄ¿9¶—ê\žâ8žx#Ë(s.?qì>çü޹^Îõ‰±;¬2ÇþF¿Ï×9?cŽýÿç…õG\†;–Où±5P9Í™ÛÜïâžîØ_¨ÿ;ÐŒiFÔš£ê3j Ìqü¤n—c{¯6žó3æØŸªûå¹ÝÛÞ_Ö@™‡ýýœO1·âîÝ=0ŽùÔ@‰ÛY¯ëÏ=š{^ã{©s¾$¦a±þ£ÊÛ×óük Ìk¨s~ÄÛuû¥±ÿC^¬y=——1…ê\Äm»cÿU¯7âþqÆD›Ë5ûÿÏ9¿aŽý¿9þ>Ÿob TçG¨úŒ(sl Ÿe ”øÁççç{]°>ðþzÖ@™óû÷_˜Ÿk<ïâ^.owì_?7|>õt?xý[ãÛKuìÔãörçúÂç{XåÞîØ¾ªÛõúØ~~Îùó.—m<çÊ[[Ž(sì_Ôórì¿ëÙ^¢Jœ±Ó¥¾5M9Ÿ`Žãuì¯ÕÛœÙMîíŽõí˜5PæÖÜ£»cùE­ø¦cù}Æù½:Îç?g†’9¶ŸêX?Þÿwj bš1«¨ãü[=/ÇïÿsÖ@™c{XÖgÖ@™·ãüZÇ#êü¼â†×#v5Pæ\?^ãüX]»;×q¿ÆcÿS;žOb ”8çŸÍ¹üÔ±>ŽS»4±üx)k ÄYeŽíKåõyÖ@™ãxåóÆò—áŽýyºã÷§:Ž(s¿¨gwÇù*óü;5PæØ¿¨c{]UÊ(s.¿×ëŸz¹cÿÔ|Å(s»ÆcyÇ4ú¢cûý9ó¢Ì±=QÇþÛ<ݱ?Woû÷™+åŽ÷ûœZªœ¿v_ã±½nå|þ Ž1×ÛÓÝ/Çñ“z™³Ê߇º\ãq< n×x·÷Ÿ1ßîéÎå_Y»‚ ˆ0jp?…xaùª¯ñØ>¨syª·;Î?Ô±üœÏwšsùŠçsYßj ÌOq—k<Ž¿Ÿ§d ”9ŽßC±XeŽùu¨c{ižî~½~\¯ã“ÏY•—-žãåŽã\†ø;5Pæ ³v çËê\ߏ>‹cÿ¤ÞEݰüű}Tá^ö÷:–§8—Ÿ:>/5Y%X>êí~®×çò÷Ëqþ§Žý»ùz}œ´Í Ô@¹§;ŽoÔ±ê?îÿPe~®ñX¾ýaÍj Ä8V—˱ÿVÏËq¼ó5Pæg¸ËõúØ¿«Gu¯ËûzýÖ÷‹(s½½Ü1ˆÇ¢ÿN ”9—÷ë¼ÑÉÛSu¿ÆcÿÕù¼k ÌëýW¯¬AA ”9öOêXÿÕù}‰ãxásÃ÷£ÞîR/Ç8¯²ÊÛßÏ8¿ï1þ¨2Çñ¬y»—½çûêØ¾š·»\¯ÏïGܯÇŸóþóìîeïç÷ýÔ¢ʜˇ÷±Êmãù|¥9öGê¦F ”9ןÅí=j ÜñÿoÖê Ê}çïûu>ØcŽãõ¶ÿ¯â÷(®ÛëC´Åàýã|^ß:Ž?Ô±}ÿœ÷Ï™ãøBÛïÁü5Ö@™cûò9ó’ÌñûPÇñ…:>ÿç|0×ü,wÌoŒÂÚ$\Ï7Çú®.×xüþÕ¹|^g T\¶A &j Ìùyq}½ ÿïÔ@™c¯Žã-óRç…slï>g ”y<îµÜù}¿Î q™çóvÇþás>Xé^îyÇöès>_gÎÏ'^ö÷Ögq~^q{ܽ»ÇrOÏû·ò2V£syˆ—×ÍÓ]š;Ž?>o|~ñ¬î­ÿ5PæØ¾¨W¹¼Ý±ÿúŒó]u×÷(s_5Pæ2Üõö÷Úõy2˜TŒóCµ}?è™0[^ÿ/Ô@™ë5Þ·{÷²Ï—ù<æösw{?8ßSû÷“7bš§¾5Pæ^.ÇûŸ¬ÕC ”çCŸ³ÊœÛ+q¿œÛ§×íú{íú{¸¾¦Îå#ž×ø²ÿà÷ùz^ïobù¨—;××׸~fÞîb¯ßøý‰ëå¦5P機5Pâvw¬¯¶‡¯q}EÝ./ýû¼¾pü¿ÿóÿþçÿ?& endstream endobj 135 0 obj << /CreationDate (D:20150701163323) /ModDate (D:20150701163323) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 136 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 138 0 R >> endobj 137 0 obj [/ICCBased 139 0 R] endobj 138 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 139 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 132 0 obj << /Font << /F8 7 0 R >> /XObject << /Im9 126 0 R >> /ProcSet [ /PDF /Text ] >> endobj 142 0 obj << /Length 3276 /Filter /FlateDecode >> stream xÚÍZKãÆ¾ï¯øb Yiù~q€xã ŽŠ’è¥D…”vfrÈoO½»)QÙuN9d¿ª««ëñUQß>¾y÷!MÊu™EÙÃãö! óuV&Yš¯£8}xÜ<ü´È–á"…ß ~ø^.Waš‡‹Þ×ð+¤J;¸/¯ÆÃå* ÊÔÚÑU;\þüø§‡àa†ë2&ò%D&ð=ù•L·LM6 ¾”‰ÔcâúdÑçO>+ É+¦f˜(Ê”³ˆãà‘ù¿ßÃïGøý~ß9,y•©‘Òkê¼k¹‡4Kg£S“Ä&Þâ Å[$Ó»|bªRHô+øý]XÕ£ŒîU¹Î,*9ÜÉ“ÕWŽë×;›üÁÛä,J9ʘ)ß*ÎÖi ÛÙD;ìQ÷Õ0ýjX®Ãâsjÿ_káÿ¦'*ów 5p²é:ËAq"¸0gÿŒÇm–«8èÚ‡Þ<Ò{Ÿ[»Vh»–E‡Ü3¸Ä†–QR¨ÆfÃ-ºP3êì–¶#?•"ŒQ›fWx ›j ·Ê„ÕÚAn©1ËVM.þÓ2*D£h‘kød{0U¡ðݤè(Zõ™¹A¶à&Š¢:07L°Ñ¢ßò Ç v«n#ö“ˆ—ªãÈÌ…#r@ÖY‡hŽKbŸ&°4Hd QäÂЇøE´ߢ »ñàû²P§¬Á?ñâýTgq¯‰»¾Içñ¤'Ò+ÿ˜ ÄbšØ¼ažoxU÷jùMg$6¼xÊ%¼ ‹b ™È¹UmvL¸ŒòÈOâUú#xÉnDЬpd€Þ3Oæ[š¸ö†ÐõŒKúK'Ž z†ª• »õG÷Ž6H³“ÁZ"Ð ÞfÞJÙG^ŸwÜ‹VËy*Íìy×m4_ ò¯iïÑ]Ó̱î½VoŠ¡#ÌÄî™éO È ¦%xpª3(áäÓE2´jÙÝçÖ·’ iK ™ÚíU“Xaà 8ÆŒ’>îdåÞD•€‚ fÑð…ÏÈ ušV·ægÐþ³ VuôYºø#]sÇ@èýÛW\U Üú aÀÏâspêÿ²Ìn›Ü´•I|?ögÞçÉ„Öyˆ"-E6ì=xŠú̶Gã øyåùQokt„Y~/DÀĦ@²ØhöÛ¹8z´`sÎ¥ÜEF íÑó$³+èæ¥¬gz£0VW¤Ô RI“JÁb [c¯>‡êÄËœûâ’X-V¼ì†;«“8|PŒj¸aV/Áú•óC?žù u¶!^[ÃXkö³Iò&ë8É"ô´ ¸8Z—Q¦µc”.Þ <ÖÝ…¨Pc¼õÔA¼"Ôøoê"ÐáE·ÔDzîàSõ'=1ãrJ#Ï ÿ$šÃ$B]ì ¯òŒá$\o…ê?¾›»ƒ÷ÖmB åB(—N–à3PÕväþ I’­Ž^Cèô¦¯›REg‰¬VçGÛ&Ü‚÷36xoeÄöÃÞ“;wïÉ[ÀËÔþ°G|#a¦NÍô3‹’ÅBU^Jh¥¿ßš]ð()·Rn]¿8hÖ½M…Fއ+™Å¶‚u"oÁÐw<›7 `óº,’ÅÛ9Mô±}’£Ù7n…b'\Œ&Yµy èÞ Ž‡"?hò=í4öd¹¨)Fª$[<âl¾³™Ì©²ª1e e±Ym8´ÿˆt"¥B‘'F@©‘vbÒ€ƒã­CŽ5® êm_yº‘I8‰0hÅDz'ËA>çʯ€‚Ï.“û Ûóîêø¦Q`YÑŒŸš ž=À:e.ÇPÇD®¬5ÿÄe­¢Ä ›[6ÿ½¡\Ô—ýÅšŠ ßwÍQ ågÀÖè• .Ó| Y9󓔎|àÈËÎî€?¦ˆ¨žÍ¨È&F1ƒFÊÀ,”ôÃm°y&aw¼â&ŸgÇ#­”šµöEŸ•¾¯”’sh$T8OýFϽHÇܸÅo‰¤–„GAÔi ¬–¡ÆvMÇ!0–ÉU¹ëÑ _pÃÛävpvP0Ä á4Þ[ë{ÕYÈ«Ÿ†WÈRu¯Øý¤Næ-ÃFŽ0±’ ~oZ‘Ø«EÒ‘J(Ѹ@^Üü‰^ÉnÒÏæº’ügFX*s5ju2Ýp ׬J¡(v©zà{¥i¸&Íœº$S׊M’èŠ#4Z·ÛÆ ·–ëqé¼ x4[hèàO sÁà‹$§øŠgkÇY^ØSï]2ÍÝ¢3ÆÈ|õÒ¬ŠÒ®^'þÃ|; -…ŽhŒ,ð=¯«+!à04$z_cã¢4ÔU²†ÁÃÀÅN¶"sºSˆàêÀË)–ân ÒzÇŽ—o+½¬±ÍHãI2…8ŒGvÜLåKlgæ<š-ÂÍ9—‡.À’=Ý$UÔë‡ÍœüHáò€Ü¾}`”NKª¾¿Øaê òСÜÜi/Š\€SÎÔ«Ö-»9A¨Þx.n,8‘dQ§ƒÐt? ެë @|Š#ppÔÜâÉÍ–…–":š5Ö8p&%#—†G´$¸ƒ”žH‹HJ%`t#¶"¤{| i™Á;tòcã¸yÞßÍ£]ñBq˜_+~0"w¨0˜ 1™T“á9TWÔŽÜoe‚L*¹ü Ž·¨\!˜{,Úf\f&F9)eÎ/gSeEþ)ºø®W,‹,ë²$Á)+j3lðÃ}¡‰”¢"QüÍ É^².~1ø+í‰Ã˜Q_e<Øó¬©ãéâG¸QdÑ»£x&ì›ø!옫ÀŒˆºÊØ•B”YcãÒy™9::¦ûf«rÜçŠSBÜ®Â3'¿½pEøc¦¾ô%¸#*3¯  ¬J¡›„R¦sF‚Ýzu@¢VO5jÑ 'Ü:±W¡$,SôxSñ4 çI×ÈTƽU¼(ÇÅMœ±q•OáÒ_¿q\ØSI\¶uÌî--˜¢q`r;WvÑpò/ņ#?Çƒá ª{õ…êï_Û_Hæ±@”²oBçYÝuQÝ`’8Œ3+~¸Ý^™bBªfµ“ùL½Q9¸˜'HòÄ!+ AHžU'ùní;ÜÄUˆñýÙÊÒѺysí·u¨šzÕ§Ã뮿]"‘„‘U5ñý¢!]åÆÄBɼ˜çÕûqÚÌ' ¦„ÃÈ¿d’µ[=“Åjh§½G~’®Œ=76­WÀ7·‚±ˆ 0‚«ë™¸/nÙË`˜â{®Ýh’æÊ’ÔûäeÙ u_a\El§Þ× þ€ÜR#RÚ5cѹ¯B@SB|ŸÃÌl®â Ïfá§¡«Ê’,ñç« ¥>“TßÖ´RéŒÀîVùÓ‘~KØs»â‡^blHÁÿ¶eß—ˆqµÂË4 Œì“Z­rcOËò ˾G½v#tŽîcƒedÓJ£4ù+No5ê#» ;èûGÄÅ@TŒŠ-lqKìMBÍ$:Ëã|‡–Zè¹5úeUì¨dÀÀ’ÍR %ø­ü6¸]æ/ÑLÑ%Ü7H}â6,µfšR²²œÚYxú%kWxœÚ®à˜ð÷eË'¼¿L Xê•Ú`‘Ý,¼3–LØR¥è?G`ßÞð³|¦FOËæÌ¦ò3DZ†#“ Äö\æ[Énѯã˜Ü…fQ©;KBÀG.wR=väðѺÈÓÈŠ×E‘ðä(Ʊ7ß=¾ù{öè3 endstream endobj 141 0 obj << /Type /Page /Contents 142 0 R /Resources 140 0 R /MediaBox [0 0 612 792] /Parent 117 0 R >> endobj 140 0 obj << /Font << /F54 25 0 R /F56 30 0 R /F8 7 0 R /F11 9 0 R /F44 6 0 R >> /ProcSet [ /PDF /Text ] >> endobj 146 0 obj << /Length 2915 /Filter /FlateDecode >> stream xÚÍZKsÛȾûW(ÎÁ`E„13xºâTɲ½YÇöºde+©õ`"¹&@Ú’•Úßž~ ÊV’K"æ==Ý_wO÷èÙåƒÇ/ÃðD…¾ c}ryu’%~§'q”øÚD'—‹“_¼d6WJGÞÙln’ЫêýLy%V"¯†bÅÅýÌ(ïê;¨w OÚ—ÐÑâÄvöëå«Ç/Ó“ÌÏbã¦ÁÉ\+? RÞï'\§§ÆÛå ÔpÞþŠülàZ©»­¯¨['Þ—™N¡½›W`ýÓ,н|I¡÷!ˆ¢WypwmŠ‹yµàÂZvlyÖŠHD ç‚}jœ^V¸zo‘){Frð¼Í 7 4mÁƒh¡~‹³p™ž‡,yH‡ÜlhQœ<;™+_‡|”ŸE"›Ê„{¤EO;ÏÑžyµ€‰,…óÊógó04Þû|-äÀÈþj`.B¡@ê¶®ãqŒé¤QÚ%¸©æbΟIé¸à`å'_“ð§*„}4s®càv7Å.RWO;IÒŠ}•)àJæ›@ W´ ¦0Ã’°Ñ—3R!§i/*,,`¤Ý—d}¨ Scÿg ~ï…2`-j1’V%—·Ì&˜» žžbsäYȨr»“im†w-¤ƒ¤E½á)Û3‡ÜöB Qm%…Gm§dv¤%–æ¶`ˆ³±Y”„ © æÌZ”A¬Ì!Ýf# #™ãVËzȵø ÜŠŠÁí©ô1~ðÄr€Ô®ÙrçbÈ6ø#Î9lñd,œ°éнr-,(K!ô‡ûÖN‹ˆÈraYÆL2C ¾”é˃Ø0T𠾚ôuU³m£U±ÓÊ Ë#¨†ÇmÀ¢PçÈ®¸!ç5Çd“Ä™÷¶¦u¯E6Öê­[«K6 2àw§}¶à»Æž—4~Ï5Z’öÊy÷@ßÞ=kÈÇèû®×ŸÑ­…üSiâgfLMeB{Üo«o´ÿákÎ`÷x"Ñ^¨ LúlGèï‹G¨ùýŠNeµ± ŸÊa¦èé¸ßŒ¬úQxQÊ2SÅþ.nw=ó·Jo¼RO9¬Á{*®!îÇЉ 4q‘¯ufôåZÑnÓ%5špo¿ï’B赜£nõ9eäL‘¨’–òéТ6tUá!cˆË`q¯Å0¾pKldÊ í)SßÞ[± ßyÓƒßp¬=sdo:º§_ÛK¦`}ÿìYÝÆ9”#þÒp+À³¼z訾¹c“çGL{+}NïLìG¡9 ÁMÏ£—S#Ò‡aôÀ0óA׿ƒóÂ1„ôÿ !+Žqêq®¢È@UâID<.1QVEy¯â/HÀ¯™e™÷Üç óÌçø™òÂÞ 1þM‚.ß÷´¬Zò¤sº_Kh-ã¤`?KYW˜­ÀMWùWŠº6~‡\¢o<ÏÆ5÷‚würý…å†[ö®Û»°ê±ÀÄ¡Ÿ¦”žYZ¬¡Ø7k1#ù”aêq3¬,Yâ‡ÉÐG%^…”ÉnU˜HÚEe™– ‡Šü0ÕÃ< C‹0P:am¥;¸„)(¯k+ç5#L§Æ{ÅòÁ9³ÇG–nóªr£.hH@Cn6î·}ÏÝ ¯)ümëÞÌK™y¹"4òSü.ýçÒNÿGDÍ€ÐÆnñó.¾þä¦ÿ Óßк wû°|êÙ¡÷’U|º‚Óz˜m¼ö…añ‚V31RÆÆ Yò{Êm”ÔÆ-ÿ³L¿À¨'ò~jäièŒíý½•Ô P‚A%NØÓ8\I:û[+3ލ܅ÁÖ2ÖËMmó”Ö¿;"릴i!¥/WœŒ…ÆØã”V¾vÁ •°‹“,Më¥wš™Ë÷b§»>ËÎéJ׺â¯Ë¹ÅÞ’ÀjW¡&1GS]¾t¨¬*õ“¬SV”f~ƒˆÁÖˆA]S¢ÀFå½kf„ø3ÊOOÔ7IÁn&÷Pß4Aõ”¾ó-ê«õ=õ5¢¾o|~·ø«|@º×’:§ûÆaÈîšT{ïfP‘ñÏȯr±»Qz?ã€Üê}Þ½@+gK\öjß1µ•‡•ZÞ¬¼ ?…¼í¹tq鈻Cæ´ç4õ€ŒÊ"ï‹‹™îÒá1°³À-†@h'O%{› Ù÷_ŒqÉ(WÃñM—rÙÚG(cù!PèÉ9óv t:ˆàöÇC4~nˆÅüNðÑuoŒ£”°•&·Q–2Á*KcÁ–N5ñ‰Á:p….4âa–²G‰ÛÀÓ«Ò'iÖ±¥Æ!ïñêBö-K:ûöilHq¤x\oä°÷,ñŽüCCº^¹‰?Ñ=Fòû 7ûLúïãàî‘lcRðy}‡'yBPy©†‡îÆë©[ÀFçbïGRº´R2Jûª{á¡9üB©ÄîKîrÓÝm°k™o)¡µ“¼>½A&lš ìžj¶ÝûhFŠÐK›\ÃÜxálÂ1LG±šÔ"ñmIé3‚î‹jiÍq+J€QÔCLB”s"ݪ^9Ź À°Fjô¾'wê¬L¬oá'ëÌi`!¯ÒÀ5…|Ôƒ<Úœ3ßúˆ¤8¡Þg¥MKoða7Ê8X‰ll-âÑ$rI#ñï_b¯ÜÍÐ?=[cLÄ_~ÉgyYÆpo­7]ìŽiâ¹M4u{™ò“DÝ»«$öuš ñpÞ]Ä¢ü¾Je÷³;ûÑ!ÉüÀÝEZ1-ßš¸3YÛJúQMÝa,â{‰bÓ¬€9_0ä1°áŽ·k þZÞ¬ƒ”Yu[^qí¼¶ ^J£ðuè _;;¡1BçÅ"6ÜšQÝ‹ Znew©í7y§¤$]zQ£Çܤ±éÚjÖèùVŒTê óÔý—ƒ ~Žá#}1]l{6ÓÞ)ÙчP‡Ê½[ÛnkÝØ†±$š>;Ç~œ&ßVä ‚¨&ˆSÖˆ"›ÀW‰–A„%!¦±ëÁ‹ËÿäÎ!Œ endstream endobj 145 0 obj << /Type /Page /Contents 146 0 R /Resources 144 0 R /MediaBox [0 0 612 792] /Parent 117 0 R >> endobj 144 0 obj << /Font << /F44 6 0 R /F8 7 0 R /F77 147 0 R /F54 25 0 R /F56 30 0 R /F51 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 150 0 obj << /Length 382 /Filter /FlateDecode >> stream xÚmQMO„0½ó+z,‰t)ô¼©ë&/&ÜÔòµk€*,»þ|gÚ®r0„v˜÷æ½™“NrÍ´ÊH¢5“*%Õ|,WI*,aZˆû2ŸØ< <&[<ÃsÁ"¯­Do‹`³ËHŽzŠí¯¯”L IŠš¼Ð]ÈéÞn™àlÂHˆŒòø:Œ”ô¦þ€ì2ÃqD¸v8~[þÉWŸB)iÙ#ªik&GÃ’ýŸjŠWnŸî\ªC³|Â9»Ì{9¯ŒÌèîÒ]¯1Ggr>øžöªJÛ Nb&ĪK;e×\A¤-Ç:|+aaç,—~è>aemëPƒµÂ(Im±Ë c‰­àµŒ~tœ–ÓÁƒ]³±ýã·iø!Ô ÃÒÃa›Ä*ƒå Úâœn˜÷~l”é= IÍìr–`B;»„¾q±Á @eåWãŒ|}ë)fp3­~ÂC‰`g]asÿ¬i0vthzÛ=’’8a™–@ Kµ£&±à¾~pİõ endstream endobj 149 0 obj << /Type /Page /Contents 150 0 R /Resources 148 0 R /MediaBox [0 0 612 792] /Parent 151 0 R >> endobj 143 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./adjcurve-flc8.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 152 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 153 0 R>> /ExtGState << >>/ColorSpace << /sRGB 154 0 R >>>> /Length 32919 /Filter /FlateDecode >> stream xœ¬ÝMÏ>;röýùÏRZøïæ;¹µÉ ‚‚-ÃÎ8±Æ€¿~º®«ª›ì*+#$‹™s~ç¹_ûf³I6«*ýüÍOúùÇŸÿúÛÿvÿï_ÿùÿ÷ÿæçßþá·ë×u]?ûÿÿáßþ¯÷íç¿ÿö·÷sýüÇßÒÏßÜÿûÇß’<àçù­•_-ý”õ«ôŸ?ýÔÔ~Õ®üýçŸ?«ê¯Æ¿¦«ýÊùù³ñý{ë¿Vyÿ®|þž¯ù«¿7¾ïׯRß¿+Ÿ¿—”]íù»ñý{/¿Fÿ®|þ~|Ý÷Û?ÿ7ÔƒûþÜ?ÿ‡ÿòÛ¿þŸóýKüñ?ýÌ_úßù`K¿òúÉ÷‡š?üÓÏ_]ýóÇüíú#^÷Ÿ{bºò¯µ?3ÿÅÏlõ׸¶gÖ¿ô™ùê¿jÚžÙÿâg¶ù+åí™ó/}f¹Æ¯¾Úô¢Òîmÿ¸é/>F5¥_×þyÓsØê”fð'm¹÷_e±=F;¡Þ¿Îû˜Ï_©÷¯)§_)=VnoãW-ïßÉíï«Èéüü|ÿžËý=Çówåöw~™çïïwÛÚ{¬þÕ{óø5ëO½[æÅÖü+ü±þ™çŽ%ÝŸ›þ…ÏM÷¹Ùš>9ü©ÿ¹'ßgâúäò/}ò¼›÷Ò'‡§Ó?ó䜫t™|rû—>ùîïfÑ'÷³jŸõ[½[Pµ.òwcžMNÉßíW?iþ.ÿ;¹´î&zâ_'ÓßþÕÿù×?éþõÿOþùOÿôýéçÏÿ§ÿû÷øë¿ûùãßüEßþ¾ ܇½®_¿Ù¿û‡¿ÿoÿùÏöýïÏü³2q"çÚ¤sÿ§øù?~þËÿ×D=Í»{ÔËôû«û³QÓ«O*5¯¹¨vy-Óøê>s5»WQÝW§*¿{kò §V¨û¿:uÕl^£Šú¯T¼îG@­y-U/µûÈwên÷N r:åA•âU'u7G§†£{ÿªÃk^¢ûHf¯¥jå«Äs¬áˆ8eÑ}Bç@EŽn¿ÿyyUU­^­RmzuÕÈ^rÉø­gi‘NSu\œV§Fú*ó»X+{%ùUz•_À)«îâT&Õ³~±[ÒB¾êPûµ†~Íû²—¦×JÔ}Iü¨ÈYÝg³~é[÷‘tâ/=¤{tâ/}÷Ý«ªVõâ/½ð‹}ÕUµzñ—^Ò»9ñ×¼[ëúªê¯¹¤·q¯9Ðs:e•|²¯ðkÞ*Ý«ª¤e}Õ û š^ø¥oI+ø gê@ û¨É§…¤E~%Ã3‘´º¯Ð üâN%«–WUåî…^øVŸ^è…G•Oá4T¹yÍJ•åµTµÕe2uëþÖ—W2 /Œ.nIïöUUõË WÑqÿ3{uÓò“š—ׄz¾¯–jö¯Æ¯kQ÷UÔ)A½âW墤µ~UUÒ+~Õ ûL]^=Q+y ÑÄÓiª†Ó”tŸ NèÁ&®`N¹P÷§p*¦é…ìªÕî…ì–\E¿êª™½0ò¸Ç¶¥{áÊu«7/Œ<æÝ2[ù­#{: =ÃWhåsáH|UuöœÐÊ×…3à«®šÉ -y%œq_-U[Ý#´ä•åÓ;%U^Eu_IœÐZoÝG×i¨VõB‹¼§°÷¨Ùi©ø*»¯¾·î3Õ ­uU¹"8eÕl^è¡W“3Î =ôBè„úVZ^]5º×PÝ<§ MéiÐCߺ?áG÷¨Ë´¼’ê>㜲ê>ƒœŠœënµÍ«ŠÒuI÷áٌҀ»ñ~xcogWŽp‘ éÃÂÑ‹ðî‚<“ñîH=1‚‘Źûðy£4ÇjäK}Ø΀ħõÄõNxÏi\3à"«tÖ_b6û;y¥y9&ãý]<óRÞ] g1ΰ’]º3Ïf¼7žÝÈù!Æ}éh¢êÈ/]Ýšc2JuÌÆ{ˆàÙˆ ƒg7Þ¿»ç0JcpD¯™Ò…F踲2'ÇΫ¾ð>­=“ñîð<3™dÚïYŒÒ«ñ¾x¶‡# zÑ”2ºXÇa¬WÀi¼‡_ÞmÀx· ÏD43Çl”fæX«ò¾ä{6“'ÏN6Îzއ#à$;º>Çe”&úáäÕ#%¼£g2J÷å˜É‰Ô×ᬫñn«ž¼ˆ¤…®Ï‘‘|áÒì8ŒwŸà9³\$_.Î~RÆ¥Ó3K XŒ÷ϳᔋL$<»Q.ÇŽÃØr@^4¹<í¹Œc}9tu.å&Á3× ˜É.mƳï›g5Ö°ïŽÂ³g ¨gÈwñ\Æ»Wÿ2ÙןÒT<“q]óÃP¿>–=«ñîF<ùõåþO Øé 8ŒwCòœÆ{,áɃSðM¿ÔuAá}ÕöÌÆ»ûò,Æ»§òÔ/ˆ‘¤g7ÞCDÏñpœÆÙê,2ZøÒ†—7ïO뙌÷àÓ3ï ½§~ý*WOÏj”æíØŒ÷´ÊSN“‹µç0Þݦ'»Í›³d·Yº|—/+çÚÂûb홎€Ùx÷êžìTËq¦g5Þc'ÏþpäE¤LœVŽÓ˜®€ËxwÈ_6½à,Ïz¦‡3`6ʉãXÀŠ…XÏj”&êÈÁIMr =98©8¯=‡QºkGN*–H=WQæáØu¨V+옌÷¤Æ3¥©8²ÉìÆ³Ë È¡iÅ(Ôs`X[±Dè9Ç¡õ›Ò¹9r ^'~2Çl”NÆ‘uÞ(ôä@]n¥­€œÔÜì%`7Þƒ@Ïa¼g žœÔ´$}—“ë?BéF9M“ÛUW@NÓä¦T XŒr²;V²È@ݳs Ør)÷\Ê»EyãjYÑ|¸¸Þ'¼{uONáe Q ˜«,dCv¬F龉۞\îEÿpG ¸È!»`>œÜÞ-Ù“—cá ˜÷娳× È˱,­—€œÔ4´FÏnyâµç4ÞŸÖ“@Y„nŽI/@ óÏd¼°'/@²Ü\²S墲g5ΈìT;¾=Ù©v¼£ç4ò>\Æ»·ù2k'# °-`"±¬î™µ,ÆûbíÉN¦c‘ܳËÈN†ëx—2µ€ÓXzÀe¼ßñˢˆ£Ov2²‚tÌÆRrUVfÀf”ìÈ«¬ å€Ó8kÀe\ÓQï§Žƒž¸vÜþóÌÆÑã=œöäÀUÖ=R@Õd+@ØÒ€9T¸]äÉe4™Ë—€Ëx_€¾l:T؀㙌÷È“C—…}Ïb¼{HO]8‹ôäÐeà òìÆûã9Œ³Ä~“4Ð’=—±4Ç®ËÑ2›[9°‘¹ÏqGÀl”ÆïX®€¼Xl0ôäåXæWÀi¼[£'O+™GLÇ¡§ÕÀ¶O^¬Vž=yZ l?ð,ÆûêéY«äi%£î+`7–p[¨§F¿žzZá†ü—zw?ÉxoLÆûw÷ÌÆ»ôäåI®é)`5ö—§‰ :ž¼^¡y¾UÄGÆJr®™× ïaÎ+4V“Ÿxãî§x•Àº¢œkáùæ; âšCcß…ïç]̸ÞxWþ•ãïf#tjœoÞ£›gèÉߣÉÀ'ò2·xqCŒë•w*æºðx^o¼kRך#í:x¾xs¬]§½Þ×ì*v§E^§w²ñvE¼qdö7’`†.fI‘˜ýMC qdö7â«=¹!ö+2Ö˜Å+6û£†û¡—º¦Ð¸GÏÀº«X|¿qälF&oÜãÊ ³Èœ¯È¤ØìoËõ$0ç+ 7´#³?–8ð{©G œµ?n¸ó9™ñzÞì¯æ¿‘Ù_3;rmj/nU-ã¹ÀǾÈÃ,ã±À3«çÍùžì{JuMYœz쥖ëM`=žƒç³7Ç› ;#óz×Ï7o^ïb³#ózצ}Ÿ¯õxà&cäe–Ä3Εû{²ÄüÆN—Zæó‘—ZÆ›9–ýR94W¿Ø_xWs¾Bs¾Ü±°™ó厭þ‘9^`,jdŽ:N\oÏw4üÈtÜT‰Ìñ@/<Ÿ/õL¡õx`däFã^NäÞÕ-6×zåùèÍõ„Ž¥ÿÈœuÜÓñîv¼šüð‘“ç›7çC<˜Cs<ÕñÆ¡y<±é$2Ç[½ó|ñÖã9ìñ_ëñòÁ"ó¬¡9Þê“×ï5–OŒöî­Çkòõ¼õxa³@d®v줌Ìùd_lÏ—ç“7Ç«§È<l5COsí5ºOŒë™7Ç›ãâõÖ›ß÷öÌ¡ëãšóe qš¡9–P¤Z¿OæõÊ›ãíÁõçÅ;Ì™A‘sQšãí‘y½õÖï‹ûc‘yOh^O½»×ËÀK=rhŽ×%¨æ Íñºxx3€’n±¹^0°ð™ãu 6i¡¹^7°/rÍj9·¤–ó1°/®×f¹^F^êšBs>3°0á­1b9Ÿëño.fYo ÌùÌÀ-±ÈœÏ Ü7ŒÌùÌàzL`‰->‘‘U@\®ÐË<#gîÀË99"óûLl‰ÌïÃÍÔ‘ù}&vaFtáïíÍùØÄ^ËÈú} Ûã×…‘Y¶¿ŽÐœ¯Mä?ˆÌß[vGÕÐÕ,ó©Àœ¯ÍÎßÛ›ó5ngŒ¬ÇyË"s>+ñJì¥ÆïíÍù·½yk扼.öGÞH;&–õ¦Àœï.䦉Ìù®léÊ¡y<ï·nf´oÎçNœÈœï.ÞosÖLyeö—ý7çc²=¨†æ|W6ù\¡9ß]È5¹šgÍùØB®²Èœ-Î÷s¾*ÛnFà®ó±Åûi9[Ø?z©åz˜ó±Åùràb.=4çk°½ys>ÆM ‘9[“×KoÎ7ÖâõÌ›óµØ½9ßXœ¯:ëx¹\¼˜Ée/„AEÆ|¢Èߺšg޽`ÞO Œñ°xµÐXÏ-¶¸EæÚBc<)–ñb`Œ'Å2Þsž/‰e>ãÍrq¾8u¡KV׺še=60ÆKð Ýù{`ƒY襱‡YÖkOþ^㽯¯—’-\Öó禖ù^àRÕ²^¸Ò“ý©wçñZì/½‡YÖ Ϥ-4úû’8ûº\ìï vC/µ´çÀè¯Ä½‡Æz’XÚk`ôç… «‘ÑɆ³Bc½H\zèi–õ‚ÀèIJ^R.Ü‘“Y®·1ŸËý¤À…ÆÙÈ쯸°z©e>¸ÑÈ„¹Ou½B³¿’…“zšG½ø{"àÛ[ótˆe= pjê™CgþžXˆ\Ìúz_×¢–þ˜ý-“…G®<ž¼_¸uµŒ7ó~Aç~ÅÀ¼žw¤G‰¬ßwò|üzÙ÷åý˜À2td^$ÉóZO \#s½bp?``®wÄñ 7¯'#ñzñqÓý"bYo Ìö<ø™ëÕ’Ÿi…æ|i CDdŽg.Ü‘y½¼ß˜ãÙos…æxm Ûƒ·W%õìš×K&ŸÌö<¸_/0Çcƒûñ¤–õøÀ/ dMŒÌþj í½·î·Ëzxà|©QŒÓ›ýù@ÎìÈ<_ç+Ùžç+õû.ÄcÖï»0 Ìõðœ ÞÅ~/ܸ‰ÌõŽÉýx9ߘ˜¸Eæ|‹©"#óz$é Gh¶GIÛXë~f±Œ‡s<<ï˜óõÉùRàÊ×C`gdö·™‘ÙMlŒŽÌùSâyë~ ±Ìó~ôìü½¼¹ž>qaÌþb¶o·'ïgæ|n"˜Þ»Ûñ^ØO˜ëIu "ó~Óâý€À¼ž-$sÌûÝ‹÷œm¼²2·7¯G ›È¼ßº8 ¬ïÇõvç©ý Ó:Dæ|v!ãYdŽ÷Öbêþ®^\¯t^l/bÆz\½2¯‡÷‹ãá*ƒ+°fýÃo),fû•«ÅŽ™§@í.]Mk¨Ñã4õ‘R ö\…å(°€¡¾!Ï„4;lÃ,â£Ì™‹É:ñ!(Ö‰…XÑò>ÒÙxC0¬Fª +Âé,–é>Òê?mjaÆSX½nȹᄕӆ¼uöŠ GÙ‰…X/-¢¸kp4ÔµöÑäØŽ*N¸Jö¤Eñv-ùŽ+¹®¾Á§$io¢¤ÔËW¸*‰ÆW±Æ‚±NVÜ5__iæÝŽŽÂ ½‹É:¡éàüH³Ìvo~¤+21)^,ÄŠ¢°Nè1:ÒD~¤wÛ:Ò@:a&ÑÑ‘}¤»Ô:6Y:a„À‚±i¾øŽY€fÎ+X8ôBá©S“gÿ@Â'ÜM8Ó?ZÜI0XÜé+\eY%äÔÐÀÈZ‘åTWþ•f.|tÂU‘EZ¦Š¯²+s5y`ÕÅ©«úõ•Þå—Aúòâ»#?¿“1mé+˜¸:ñH 1àG+Æ>§ªâ#Oá̙ɪ6lêl“©ö¿b!O&Ë?¥» '&gNXùšá:±À)áŸÒˆrI‰¹¼Ð£LLÖœ0›LÊŠ|âS8±é‘ì©©ww&RÇ9afÊ,¬Nhu›Î>JqOœ%eÁ…ºEN¸Ê°8¨®†,Öù‘^0qb±N¬ò~¤uVXÐ303ÛGšz!àÁ ³ý…»ÆNØ µÐ#}ÔytVsœXPE0?ÒÑëÂ]L'ô´Ìlå„Ö³Dí#-–«©<™Ã”U*¿\o Ë^2Ëá…½:®Ksk2¥'sòJ yYÃòKÝC‡d=W@­p‰©®'ó_M3ÌšUÒ€\™UùÂ$æKy#EF¨µ±ù¥er¿°jì©9Ã<™ùšu¿lGmEOVËH¸œz>U gqìšgVBj{@­­Èœ¼ŽÌÇÍpÐ/­rŒì!\ùƒ&Ì`<µå°D~;§ÖÒH¸ûîÉÌ›¬&èÉ<³ [|¿Ôyðf@­­ˆšžÌ°fú!R¾e5R‚xkÁ9¬Mzë-o‰J)4Ë4IÙÀ—sÄ6À:•ÒØ™ÈZ\®2%Œ7ûœŒ±·ÞRšÉe3·tæ¹–;Sâ}måÀXÎ/2ÓÒgM‰äÍâyòxÝìõñ;2_¯ð–a`–åÞZïn¥ß¸e:0KyÞtæŽ[.õýxKÍyjq›‚l÷ÞË<™²Î[ßo1巾ߓâå°¥tÁÞ—Ø<ž²·%‡ÖYÈ3íô÷ªØ;™í­rKj`¶çú¤D9­¨XaJož/µ2¥7ÏG¹·Ü-!S5eˆ÷Szl¦ÐìOäÞl Íþªv¦Løºj_X;Sxkù²Éboö´,æ­[N±ößB³ŒscJÉÀFbív¶Âm\»¬e”PÑÎ{hjl–¤ŠÜìýFÍënëL9òµ.½`íµ„~Š@•še½›†ü½´àGgJÄÀ<YÞÈ9[âÿŽ5°ÈlÏ,Û™e¹úòý5Û[o )ûZ·d‰Ç­åŠ4$ûkKÏR9‘µà†4{sôÚ±Nç]¬PÆù‘ù{õÅ2o-}r=!µ‡­¸: ãD.æ9B³`Ó`J¸ÀËŒ«¯õ–GâúEän–öx™× ¬)G¤\ BZ½µÈ—ü3eþ`J5ç¡E¶—üWs¡‡y^­¬àX I÷æñšRì­LÛ»÷S¤¤çÀV2ÅD"³\ï¿.VÜNæã#4¯³YHØ×ìÏ'¯'ÎVÔ%:"³8æì ™ñfù5¹åV[µÉ["9™“[~½‡¹–Ð<ÇÛÎVzlam+2K„±þ†·úZX;ŒÌB« ¹É"s¼³²…€œÖ-iZè"2ÛãbÈg`nbñ‡Èœý1›¿w×K,âY¿Ï`H€·~ŸÁ‚¯‡^8Œ¬Ÿ—)ëïÁ”?Îr+°eÝ›)'®Ìo¦D¸˜²2°n¨ü>_k9M)epÅfH8 !D~ %ôâ]¯·ÜAhAç–=o¦¬½0pŠÌ”³Sš8[JÉZC3eÇÅùH`¦”¸· ~m)ÿ®É-ãÞLYÀ‰Id¦$Ü34SÂ¥d[¦OMI&ƒ+4Sb1·_d¦lJ ysÖ9qÏ¡™rˆ þ#3å SíGfÊ&Û÷n–ò¾sË7Sb$¦ ¬…÷×ÝÊ5ðz˜))ÒäïáÍ”€iږ˯™’.¡¬™·ŽÄ¥‡fÊ3Mù˜)c5å`¦œÊ—mÙ<­[‘’?‡fJ£Ì”|™Ò‰¹u¼uUXRö§š)e80ެ% 2ßÏ›)]2î½;s§õïHù-1Þl/¹Ú–»¯µärFfšò䬷D,e`¦lÉ[j¼™’$¿[æg+‰€Ü‘ÙßäÎ-5Þìo4å~`-ÙÀ#gK¹™™$0S²äÉ-yÞL)šuKÝ×ÕJB,nùóÖã‰Ø­Èz<{YKV`oŒ·•6*—m¹úš){ ×S?)ûuËÕךœ)ל»¶×‚½6‘Ù^ Crs<À_dMùÏõTgÝÒm)å3¥Sé¶¥êkM‰Îg]””èi†Ö”îÈ­™)µ{™×“²¸eöë¥)ã*ç»5e5S^æxD%´¦ÜfJô¯û¥En*bó#3%cåzl`¦ÜÑ”åÎÉR¢cßJd¦,•ØZS¾3åW`öG²ž—[T™2"p±×“þ>p·×“ùJ`þž•)#œ5¤JR†KXO¦tÌë©ìO¡™²¨%þ^_Wq¿YdM9Œý%‘5å2דkJjÔNñnšÒŽwdM9ŽÜq‘y=çÞ`ï®ãÁÆñ``¦tmÈ ™)Ybw¼‡¥4Æþ”ÈÅ/ë5¥õäùà­)¿1qòž–Ò|áz˜ó1Ù÷›)1ûÅãíÍùJgÈ‚³Áê Y¬)[9 ¬)9ŸÿZ¶XeµÜÌù`gÈx`^/;K&9'íÏ;#³?ï\/ÌùBç|ß9kJØŽÜ ‘Ÿ”Á2^ ¬)§‘Û 2ç[²g.¶ñަÌëqgÈw`¦´ì\op®ö{O´§Àú{bÏ·®Ç åï Íñ’¦üu¶’-ƒ!Õ¹Àýn‘y=˜x{[‰?Ù÷ÖBkÊT¦¼ ¬)wÑñz[ÉÚ˜‘‹¥œÅïéÍþBS¾ž–’u–ÀOJßÅïëÍõŽÁ-÷™’z^ü¾Þl“ëñ_O›ïLì½ü¤4m#4û«ÉõôÀÏhÊRç¤×»É’+y½›¸1YS~bÏkäe)MõýNëý±\¯s=DS¢fÊùÉñP`MŠý‰ÞÅRÊr½#0çÓ“ë¹3¹^á¬!Ó™ÕÕ#³¿ž  ¬)S’àÜt¾>2¸YÊSüžÞ\ïД¡ÎÝR†2åI`M‰Êùv`Ž÷öæEfJiÙ¾Ô[ÊõÅû 5¥&Š|GÖ”¢¼_XSªâÞz?VRjâxys<ºrXSJ2%f`†LkJIg-4o)kJŠõ¬À ‰º˜’!°¦¤DéggÉ”›ÔÒfÈ“¦l Ì”KfÊ*MÙè¬ëÁ–r10SÞ\Ø ç£XÆkκßI,ãµÀ )½˜Ò 0S]œfH×Å”þÎESæ] ) ¬)Y,0S¶iJÃÀš2‘÷œµ$IamˆÈ ‰JXøŒ¬)ó°ð™)ƒ4¥žsÓ”€™ëéÈÿ‘r®)é3åsW{ëýfIÉ&ër™¿¯·¦¼ÃNQo £”^3v¶”^r=ÌÇÌ’/5åÙÂõÞYÇS’KÖã—¥Æñðî–²k–К҈)·œ—¥¼Bí¡È 9,…ÇÓ›)+KáùtZj–3eKÁ…;²¦<Â|"2SÖ2¥TdMy…ù‚w²”gØ£™ý1SFEÖ”dH å­¡KHù”C³½0wUdörc/…fÈ$S.y?)o§·¦lÁzid†T˵zYÊ¥–[ÈpÅÂQdM)„Ü>‘y>T”<õn–²)™"3eS2EfJµŠÚH‘—¥\9p·”KXÏÌëIÅ~ÖÈóI ´7%O)¡ÙŸV\ï"s¼ 7fbó÷âo-Š.)o®ší±%oo¶Ç–x>~­ëÁ’Ò¦õÐL Ї·¦lÁãœ,%[Ãzeä'匴÷À¼2¥Lä•Ôm¶”à kd¦¬h(©yš¥=:ë~I)³Jh¶¦Œ‰<Ì+.–²k"—Ç#4WÇ¿Dæx¹#÷ª·¥„eJ™È<ŸzC˜):æÞÍRî %jdŽ:rßDÖ”8ƒÇûën¿6†D®–ò%ÕÐôÅßãëa)9™¿÷ÀóžoÊ´Go^ïŽ#kʬçyÛ|gbãTdM©€{ÎùÒÏ3¹[Ê…ûÄ÷Nz¼™r!r·” ÷ÄÓ;k{XÈ͹[ÊiÎE¯ vŠÌþ~¡¤wÕëÕ=ðÏ#°¥¼Xˆ%Œùº¤,óÕ¹[Ê \ÿá·=Aú¦$X?{J‚WMGw ºwjo‚‚[u⢾QÄù#½ÛÞ¯ð#-ÆÃ¤!N˜…4l‚þ¨²‡l(þ‘l¸‹ó‘îla'&(Àìä#ÝqÞpÃ÷£ÉÕš†Á„ÓÜ)œZ\eÍä(¾Fi*¹x¯¥ $ÿ”&Öi”|¤ LÀðQål‚ >Ò–Þ°ÑÛ WÑ65—&±kØIïÔLã+ Xn(çdÉ îAÃGZtÉ œ0Úmµ|´x'€w´Ú›(âÔÐIÒ0½šêž¥|”žGÎáÅG&9Ã?ÊRÿJWC:6«¤q–3§¦ºÏØÚóÈûÌsâ#1ËøHï¬t¬p85Õ_iÒ›ŽÈ1'&ƒÀ¬ã£ÉÒ‘(Ä©©$hþ”%å@ sBKî¨ï}jjñŽy§fš^KLH0ó)M˜Û‘¸Îi©jý*sQnru/œý;1>*Ý|ª&­ÀÀÕ £Æ×銓O8aµx ×GZìi`#ÀGšè|`4íĸÓñ‘ðaÂ'Œœ"À>Zš| œ°c` XÿÔJ-Ôãr•KήòUf”Ô@ÛGz¥Xàÿ¨rtÅ$NÍa8ibŠÁä¯p6 È?Ò„ÿ‰÷pvH.»æµ4ñF_ ŽfVòpUXÅúHï ÆŤH ñ‘%˜¸¤—:…pãD–?6óy8j5ø‰ÄBÕD@±£fò“¬5"'cü¾,šù‚Eœ¿¬šƒ„)=›>Xî}©õ)&ïxâ´ŸŒÜûRkO¦NürpÚ?Q. ý'Ë9õÐ F„¸xòLä ˆí“ ?Lº™i¹_G„B,&ü2q ±´°#³b\Œ2ud‹ÌɵJÏÊ,RëˆÆ½´Í{aD@l^XŒ üRo­.î ðÄ`{15÷—š™y!¼:`×< 88jV¾…­1á\‹ñ¸K&ÁŠª›%ÑpÕ¯®t{EÖŒ·È+0C3œVÖµ[`úkR‘ȳªåvá×Y;$x…f8äÅp g½$Á#4Ã/­ûµV¤B•B3„ä5˜µb«¦/‰¬¿Ã5ëïÁŠEÎÕ~ŠÅÞú{TV¤ûZ³†!ïA ÍpäK+Î~­ËEÒ¢"ã×΀޲ÖŠàÒ_"Ö›áXì½—†_®éÍpËá_ WB^‚Z“Óp»„³Vt“Þ±ÇÆídôq)pÖp֔޿ÍpÊÄpç¢áŒ‰á[WËñvÖŒ­ˆÛ_¡Žœ*+By³=q]Ó[3–j\¾·f÷[¥nɸ|ï¡çcÒpPïnG¸â×ZQégrèÅd.Z1æë¥é&Þ8²¦ÕÁí>çªÛ1¥ã̱nÏÒìÞIódV4 ܳº•Àº]Ó Íöžµ"Éך‰KK7GU½jà¦é²VüðfºˆÌŠ›Î]ûS–FÜ,]‹lWŒé!:–ØÂ½ób8œ7GaÆdç©Çƒ¥%#?é^z¬ËÙ(­XCó÷Ôt/_7K—RXñ!påçÁúHd¦Ð*ÜŽçœôzQ°v™ç»¦‹qÖŠŽˆ›½[V¸?0Ó1H¸IÍtšnÆÙÒm±tPd†ûJé›ØÂ¡5]L`¶÷šnóµ%ÖªÜ'ã>½µâNœØL‡PYq °¦[áõÔyZ:V$pÖpcÄ^¡5ÝÎbÆç»fìGÜ_ Íþš©õ½uÚ†™B3ÝACéÖÈL!Ö²…¿œ¶t<­XxÊ×lÏÏœµâüpÈÈìÍt>šÎÅY3ÅŠÛºY:˜^Cs¼(qY=pãäJŒpo¶÷6-Üä´†ƒ .ê ­éb—É{p;3~ØZÓµ°¢—³.{j\RäÊ×cŬÀL7"Ifà¥é’:·“æxNÓÍ|-ËŽ—ZÚs`^Ï:Ãåu™q1%4ûëΊLÙž5]³VÌE:š+4çS]·»]ìû0]M`ý> ×CgKÇ6Pú-2Ïבâ­ébþnú{ n‡Ìßkdþ_[<°Æ™ó½Q-Üâ´ÞÆAÜF Íñ4ã*"O¾+Ž:kEHÄUŒÐ<ßî«{/íoÓµf:­y1œæãyé|ëú‘™®CÖösh¦™‰Û¡¿¶tNšî&0¯'¥ã¼m¾>±Î¹Yº 8­—±/¿†f)ûò[èién$ÄY·«Á3´¦›iÜ.ÿµ¥üœLç¸íétœõ¶ƒü‹ôw™.i2ÏyèzÊdº·Àßwp»ù×–Np²bQ`ögs1œÄ›óMYY,­¿Öt:¹^±x½ÿzi¸ öu·Ð<âJ¼Ó™N'0Û‹ìËΡÙ^âê½³¶—…tÞ‘¹žµXQȹhº7M׸Y:9ßs>,û†W`Ý.ûr÷íᛎÊ}·ÞMÃ.ÝÎþµÞ’·ØXëöë¯nvévó¯µ*‚컫±î—X‘Ýyj8JÒíèÞšn„ë••°Î{Y:Î3œ3ñzö±f¦ÓIˆ‹ŒÌp5æ}ôÖLø²/mæÐ ‡¯Ð‹ï×¹]ôë¬á|²¯¬…f¸ÓÁx K˜ÏDf¸hÂzŸ·¦ÓCº–+4ís’…Èš®%ñx}Ý4)gn·õÖtÈûéÝõ÷ʺÝÝ»[º’9K7‚tg‘õû`=*²~t¼ÞÓ¾òÆDÖt(KNlï¥áv ‘»½ÞèÞéÒpQ¦ûˆÜ,ÝÇÌ¡5ÆeÛ}OkKævNo†«•Æí°_g §g:ŽÈ W,ØWåmáÊýmd†Ó±t´·Ž·5]FdMW‘¸]Ø›éjâvá¯õ~‚¤ËÀvNïié2z ¬ÛÝ%ÅÕB³¿eÞ$ïaé@&¿¯7Ñ¥ô^¬ëÁ’ïçÝ,‚ôg™  nçþÚÒ‘4¬§8gÝ.ŽtWè±Ô#N–.y"3·-nýÚÒ 5ì›õ.–Îaq»§7ÃÉ;ö"xW½^3ÝBäjéä|¬éØ:kÅ[Iw€í¸_[8jÇ·ÈÓ/Û½5K,Û½3Ü·s;´ó´t•Û½»¥;óÅYçkšŽ 2Ó­0³¤óäëa=(²þ^K&ÞI/¦ˆÜ,ÜÿîØ"s<3pa÷ÖêK’@·À —¨Èç]t<$û+´¦Àzïjé p¯>2¯Çq#ÞMÛƒLœsèÁp÷‹ÇãënáæìO ÷—íü΃÷ÅÒf9Ññ{O޷ؼÞN|0ïeé —.r7ÀõIw€¸È|¿… ›wÒÏ+q£#4û»…tµÞ–n€û»"3]Çb8†sÑt‹ýe`ŽWe?G \5\Höc\¡ît¡\ƒ·–pn ‡ ¬áä˜ïyÝîϸ®È §c8´÷d"^±‘0_ó^Zñ5aa%2õÇ _7 çLÈÛÛ‡$œY>¯sÒp°„t¡Þž—92ÃEvóx¶g w–p®À 7Ë<Ÿ«†CeÞ‰Ìï“q¿È[·¸‰e¼ì¬ói g.+4+ÊfÄ"xk:| 7–pØÀ ×Ío8[ÅÍÂp£À¬¨Ë¼ÍÞšîVÂ}å|øº_.$ÿ Íp«‚5ÞIà ¯·ÎZîAÂqå| Œù¦„ã®Èš.ZÂq%)ð²pT9Ÿ­"®†“:ëzŠ…“žN*íÉÙ*ÂÊx;Zq¼q>Xà ±ßÌ{j8\Cܰ·^O,œ30Ãñ¤´mö¶ —”ñг¦û”pGé/†ë1œÑYÓ…K8¢\O-ZëU4GœLdVLgÅbo ?±ð+ó~õV„EöÁÿþÛßþ`yóîÚßý\?ÿñ–Åp IJà§ûÂ\§†!ôj×Ú%µa5¡IÀ­jù †=íZ»îk\Õ® ¶ªzß™áU›îNrSÆkbÊz³uih«6õ²kÌ]w ªÃê³Po¶j(†^íš»^e¼ÁVuqnÁðªMß•w1íß5_!tdSʇôÆ &Ü k³.î«Õi]Ü‹÷Ò´©å]£ìšãÖøwÍ]©ï*øœ ‰Aê >ìˆ_o‰ý™V×}5P9khÒ5w%<óX†;5M*Á§M odp«ò¾ºû©]cë/c&É„˜»Æ®šá0ÂæNÛÄ0©fA>UßÏtŸ-óÎkí>jZ·Ù”ò¡µ«”]­îº]CÄZ¸ãËjCPÁkbȽM7¨wì}m™+¤¬Ñû÷}7•¹«œé‰Q#ë÷nj‡:^¥kxÕ(»¾íxë6%|ê©!TÄ(ª#¬iJVúÝ4Pó!†^qS ëþ"È•!*ÈT¾)¡¶6Ö–µ©àUp–aY›> 懬¼i­WÉ‚jU9Âó0K`צ1vM¼ÊÒðª«ïÊס!º BÛØ$gÀ¦)J–•Q<ýXÖ¦ûSlšõ*·´úi%<¯hXVÆóXó7¹7Õ±‹ÕÉMŸ £Ö'~„0«V92¸±ÝtçÀ6a¹"gªâyë õÚ´¶¿¡òyÓ ÚÉxk¼I KTJÏoà×&é…_±zÑ­+Z»ÊñH¾¦‰¯‰^‘Á]²}KC½&ž‡ŒÁ]›rÞUÆ.9s6Í]ÏÃH /›¦uX6hMk3c·hÓ”0KºMK0Ö4XdàvyÓâe Û4ë®…Wᙊ*“¦‰Ý¡VAžác›òØUꮞw ¼JÓ`±…õ8ÿXýù6áÉxGÇÒÚ•û®ÎjöõÇ‚Èdô£!e| !d›¤µ¾’ö©)¶&¶@7-Ø0‘oÊæ§ &Û$-òÕ8$#*CoJ}W»úö7,·4½»<‘·i*d˜=B@Ù¦Àí»Z}… ßM˰Òö¦‚ϲ4lî¾æt-DÆ º~é/ô™›$Ñîi`@Ý#&]xu_ºÞ]cÝíMw;Ý´ÞGJœÜõf±Ð°»o¢õ¯±idçÝ ºm’½xeÂô¥_º!‘y/1MèO8ÂívV| ­`»÷…Ä* h(ÝKì<98Þœ®Yn4 ngÍ'ñÊÉbêz=È¡e¸1 í@p!™COo<[¾NâSq7£åvÖtr\`·H¶«d²‘aqrklDXÛ΂ïËMßr÷¤%ø.äÂézKK«ï¼;Áž-`q/ÿÖŸpCÜNéVœò»[­:V0?86âz׳îœcÉòóÚˆú„×ÁÔJ{¶=“j u QcˆÝλùu½A§õÍwÞÀnÕk~·³ p)¥©è^Å{ÈÚè]ï´¥‰ZÅvIÀք܇ÊRéݶµ1loç=±x‰œ˜½Xép, —r Ý (&L'ä>ñP®´9x—øw ÷Û)™k´V—ÉjžØÀ»‹Ý¹ð`îŒMHlÙ­ÎgÂœì‰ `\àλ­Êì»*¥>¬h„V21;ûµ;Ë<ØÖÁ‰çN‹œx#ž8©£s«V*C×å:(­nãÈ'×Ai*šB£øv¦vð>¤;¥ûjaˆK½i@úvζÛ¥v¼¯ÖzŸrîš:AÔäÑÀV©n…Y’~gû:8ÎçÎó¹k{®”Ãgæˆæwò•¹A6cƒ­ÝÐró;¥Õu=¸SÒ iªp”¢O?½¿¡‹Òr6J–£/ÅÓ™eê_¢‹Ø™úÁ‚çrmÆÜþ„â4ï`¯Œ;W݈EïyløFÝ‚þÊØ Æ6¾DÛÙñF ´ÉX}Þ¹ðRÜf›1ÿ³dQ¨Ž;…¼þÞ¬ùä2D#cÕyç@*^Ž3æZVŽEâ-¥ê4@(—ãaѤp£$åzˆMS;S>X¯ƒ­œø¼àfÌ8˜Q‰ÏÌ ˆŒèÈÌ¥!Mv‹ð-Ø×µ^§0!aVÎÆqPšŠEW0ºs§Œ,-6ޱ;¥ê+T°ÅÇàFxzîœxå' t!q£Z –zwv|*ÆtImÚŒL^ø„Œè(ÈÜ˸ß*]Ÿí‡g¼è΂³G*¸ÛÞ—\kðïÈÚ‹2Ö÷%:½߈׊YCï `=oĶŽy¼ÏýQtXË߇Œ=šÎI#|GÑQ®ÜÍCŠ¿bñ¿wÏpp‚Ý{á•5ëŒ;3rrÆ×°~8ŠMù;Šòbq§ô'>3®Ü¾3ŠNñ†â–zPc†e9½*ï_Ç B!4æU£!’à‰wlX.´L…Âçr ÞLwhm9¹\™‘ƒm®tï”~Ã'¸¼säÂíLøÌË‚–™\ña-Ûup˶¨…-GÓ.Ù‡Ö+‡&W8•÷i>´t ‚ØñW^Rê¼ón`ÃñeUð²l<ëSî¼ÛÌÎë×GÃKäÐÍB¬š™e¤àRÚ°úæ #óÑ-û}"Q$¿~G&Û¡ ì<=ºÅú"Vúà<8ñ`^a;ÖB%{rUf¼2¿ ,¿àÁŒ¢c”õΉ7┥ f¬Ä¯Àërµú};º&¹gR•¹/5œ+¦Ã²k0:{ û ¸¸s¬UŸËFǹÉÒ”cZܶ4Âaß+§/±aDriZöÝùì”i£\€6Joö°£]M;íjjz†wi_ÛA†;|6{îÄs9ëØõ1¦}AÜBÓ¾ éΊ¡aߨ>–}Há4¾³ƒ’w|¹4cèÃt¥™m”l±™nT6ï\x_Æ8ÀKu ¬¸‹ ÔwJ/jQËk­C7 ž«1߈9?86bÅuX‚2 ï,×ÁŽÁÃΡÅÎU62'íÆÂt«Iy÷¢;%Y©rǾs +,£Ø§ zŠ›½Ä–i£Ü%²yÙE‚ïiÃÚÂ@/1ªŸ¶vÍpøik×?vÞíyÚXÈL°Y9ËFléØy_»g²/ˆM;ïæ=“}_lë˜É¾ íÍd_7Ryþ;Bì ^Y¿ ºÍdßc~¹Ó8”3\x°~#T5É~}ìk±ú3Û/ˆÐ¤’A7Û7ŠëλyËnþ¡,ã ³á>ä=¼ÛóKT„›º“EÈü¸/ñ¾z¬žÌº 'gŠÜc\k#sæf=yÀw¶~°ã 6‹ëŸøF<ÎUæ¦îªGø>Ì;±ãÉÇäbêjM°³ç“ëÇÊù¡î^ŠòÄ‹iÙ°øßgy“T¼/© Â;'ž;,ÓÀ±¨;5 ¿æ Ø)°èõhV})^p'H§Æ "ØüLݪ)N!/¸ ;ï+ûΙ®¶;+äþuV&¼TµTå:‰O¥ú(h>«ŽÍ˜W`VKÐõS=œ×ÁU7b=vç9’aꀩ[Q„LÜüPzÑ æÈUõ¦%Ç™XQ™íÍE‘ËYSG`¯ÅÎqþõ¾:OÇ.¬šNÇ.ôÓÆ±ÌRð‹¨×Á:fŸžÊ¹6b„UsÔtçÁŒ—ê–û •ƒ£\ø Ã\x0ìÂV Ët­i„S)ÔÎ%©¯/ËY ¡ë‘dŠƒ’Mü! eN03ŸÁ´óÂìNiWÝ]•+û´òÂìiv; ýd·ï‹_vÚ™‰¤eZž„ŠÉ–³p”^¢ZÜ´¤\ Ë³ÓÆÀ ëÒsا½ö`i„–0C2ŒƒÌÎÆ¿0–dàY)}ìFéc7.WË?ÄN‰ýüëÄ_ù+0ÉÇKl]IC”Y k%ÍXª¾”& Á é¶ áÝêV²C‡}+Y¶\…Óˆç2–;!(‡·Ô"×A-)À÷E‘wo³´"ªaáÁLJ!·’Qœ€%¬™k{çÀsV6vJÁ€£œ×Á…—ÒƒƒQýNiuÇ:8ñ¾ TMè"–®Ö ó:XúÁZ¶ë$>†ö%o±Šf¬ÉØpµSÚ¤æ{Þ!^*ãÎÂ*ší!c ¶´Þ†&GÙ)-g'^ k×HIÐN¼ãØ™Gå%¶Uì,å`ŧbŒsÆÏ±³Ïƒ|߇wÇõ»Mv¼/³åb/eda Í‚ Íe©˜³{çVÛB(°jÜrÆpwçÄK1+QFzK²„b]ùà}½XOŠ+›ÁŒ #ºïËxtæ‹y‰aØÒbwBiu¥WÍå“1”Ú) ø!ö`¬¦uÏ3Æ?«i‰Œ;¿K·† W>¹~¸áñwKZsp,íà}EXº \8ÆÁ•NÎØ’±S mlÌ×Éq°µƒ߈yc2öoìýeA…†¥õ¦4‰ÍÎzìøÌÌhÂŒ6;Ìxyæ³y‰n|i6D¤m(;ž;«UN.¾/6{,+ÏL8;YŸßg\ö~¯çi\?4"Q¬Å“½>®'K?sù.¬<¨¹ø•uxðñúù3ûóeŸ?³ýî§ß_¿_æù±ìógއ–þ^Ü›(Á:ú}´àeßNx©QIð²ïƒ­DËùðº¢}¦Ë¾jJ"VèR7õR£–Ýæ•wcJ(Ö±æDjUõ¸N£"a²ï‹{ˆNbîALᥖþ?%û¾Z³0Ù÷Áv£”’ý^¸% ÖÏ‹Ä2Ðô»åF<=à'wââßõóNVŒÌöyµîáæÁÊŽÉr'J{Û]Òéúy|Ë›§Ö>Ô\èšKQÌëåĶŸÃ“ïö|´§ÇZëp³–žäx‚¹7³v`Êz<&k¦¬Çc²öÄîñù»äêNEÇÛv6³v`*–‹±°½ìæã9~šÅÊ]>Fû)öýY pwʧõýKº»}ž¿ø~l_“ó·Ýõó÷É÷ÓãËùZ²ÅUæ’„-·dáãõøv¶§ÍýÚ=Ø^lýuþþ›QiÓV`':þÃhO›Íñëœl?6~Ý‹íg3jcnîí-8*^eóºØßØøw¡Hœ˜ æëâñóp¼¿.ö?U—ÁWb5Ôj¹.ÛKÕñþJìošŽ×Wb{iz…÷Zž"§bôZûJro^||±\œZ•wä&" lêžNO¾ç[\ O¼ÑCþó)Y³(ÊùåÂùf ܧqº¶Ó½ŸÆï«¹6ÄZŠUÍZ©¿¹AË8º¶›q½é:]µVs>”t»‚¸ðõyS–ëƒ|ê^O¯²{òú¥;Ä…ŸŸë%\Û¼xý°\Ç‹µ8’n·~ºóý¸þÁÜ¥%W)úÝ,F0¸^¤¹L7'žïºHZ¸0!fn¼‹µ “Ö2Ó\§‡µ6nãûcbŒøÍ©®ü;s…^؆¢!›ôäë­ªÖZ¸ÌíyqüŸ&¯7…Sx©+Ï\Wåïo¹+™{õ©¥+–ùên™¯¾nVk·µÖÚ}\úé‘>æó{Vëë¤F{yÜÙ&oÇJ®×ÂZ½WS£ÿ_šËñb-Æ´xcXsÃnf-Æ´,÷,k/î–ùåîz}Ï׌ÔˆÇj­…Íó1?µ°52×_w÷ü1ßoõè»q7ýpemnæÒͬ喵6²XúyÚe4?c{ÜœY{›¹c3×c³Þ—×\ʇ[yky‹;ßý7s+#ªœŸéÌÄóRw~ÍõŒ ;ÂΧz±ö÷Å÷Ñ85ufíï\Õr=ËZ;OÜøøv©ÑŸmFmn½™ ³v¸ævlߺ>X2v´^|m¬%ºµ×7£?|rSãbã³yœž¬UÎþ'sýôðÚ\.¶G­m)®|<¯?,Š€ðzæŽFšÃ™ïÏö\Û³Ž·Ä=Ì÷g{,\Oݽ®ùúl¯Ì}}x°;ÛoÉìß6ãùº>(¹¯åzš-W7‹ŠŸž§µÖûãÂ×/CÝùúÕro¾>sÅ3×öfL´P+¾ªk;=ùþìŸKcûÕ(.qaíx^o ×C³åž/ßK‚©ÖÚò?u®ï­E'Æùeë_ûÊÐþ™éŽãø?æþ…2´¿ëÜP4"¾0AÐfîGØ]ù÷nÆù¨Y–Äø}´v©Xæ÷er=³0!ÏkÖ¦8ÜùwÎ_Gâïiã§Áý¦Ð•ïÇÏ3X«·Lý<ƒãß×™ý£ŸFáõ~iû…ç×ÒÏ?¸þW4’¿¯XdC3èйœ.|<×SFµ×ãù0x¿þ5ï·ÝoWï¯×Ëjyp¼Yõ~¤X®Ÿ»¥?¬—Î7ï§¿FQ1Ïï":bŽ?ï×K¯gL fÿ=8Þ«6¾ïU[¿(µ!ÖãÅõ²j㇅/1û“Áñ`µñÀäýëª÷Å…çømr}­êý»Â¸qä ÂûM,,!YPW'þ×GFe#?>ïäx­f]ß•`Þ³}°H·dýáù8q£h3n¼ …ÐTËzYÕìö…Á¬b}¿fWw~»^NlþMUóÅ‹ë8ãië “ëK²Í²«e|ûš÷C9ìSÏÓƒÏçz„`ñïœ/Þ}ÍõªµiÅÒßìÆñÑû?bé^óþæá ³^™ÇÇÖ’1–þgwçë³VÎâõ¾v=>‹ë/U£; ÷-‹9Z¼ßV»¶ï…ÔÄ’æózqþ]õ~Xú·Ý2~¨ Y/D‘ RþyjéHÎ!Ì7þð[ kéüjæ`ŒùÔ̱J8ÿo5s´/Óš9oí­™£Òš9¦µ+ª™CYÍkæ˜rÚÅš9w?5sL¬aƒ>ü©™cÒš9*ÖÌÁÕ‘5sšÖåeù]s+Ó˜úñ¼1vͺká5šÄš9°µA2OÍ«™³kíbÍkߘÚñ*¬™ƒÝÉOÍÜ[zjæ<»ª³`”£5st¯ÖÌy%ÙÍM¬™“xuÒš9›Æ®’wµãyÏCŸ 5s^MÔqA 5sTZ3'kUK+›˜ê±à¼Òš9¯æx•´&É£ùV/aÍœWåÚÅz%¦v<¯³²Éüyjæ˜X3§pF¢5s^嵫×]¬BB­&Âú/¬™£W­™£‰Â´fŽ&rК9U«ùT­bbÅ\ë´fΫ†Ú˜÷hÍœWcìbÅÓ«àz¬5s^±šÈ#TœÀJ™ÖÌi¼©5s^­ëÐxÅš9¯ÊÚÕñ7¬jÍœW µ00Óš9¯ ž‡^Qk漚éÐöHÖÌѹˆÖÌyÕû®zØ•¡5s:{7­™ó*µ]õxd;ÙP©=ŸÖÌy5û#­™³ Õ60bÓš9¯rÞUú®víbÅGoÕ ­™£¹aµfÎà ·ÖÌÙ4w±Æ„i Êz>­™cbͽ«®5sæS ']¥îªéÐÜ%m~ÓÚ5Pɳl­™£ ´fÎ&<•…X3ç•´ùWuìjuWŸ»øîè1µfΫÜv•µ«Ïke×ÚþÆš9›Æ®\wÕ¶«á»£¿Öš9zï^kælB•Ö†aÍœWµíjùÐñ¼¹=5s^±ë=±fΫƪëç©™³8†Óš9¯X'ôÊ+ÖÌy5Ò®µ½ÊÒÊ&VÆ0±2†i¥C(½€Ý¬™³)•]yì*uWÏOý æÆÚ´®Cx?ì~aÍœ]¨T‘>_mêéÐÜÅwÀ š5s6å¶‹õ5L­ìšùUÑŠXdÍœž¸ÆÁš9]3q±fN×ý¦¬™c5˜¿Èj(°fÎ&–.À¸›5s6õ¶käW˜po’$Ù¯JÛÕ뮹ý ûzæ,Œ5sºÖ#gÍœ®1ú¬™Ó3çÿÌR²i]¯°u¥gŽ X3§ë>JÖÌÙ4ó¡õˆ5s6•¼«ö]ýÚ5Ó®…GbÌÀš9›Þ£ÖÌÙTÆ®Zwµ´«·] U 0«cÍœM9ïêm3ü³&ïðÏMd}„§oZøæcŒ ßT×®qümÖWÈ˲)á51ò`ÍœM5í} :PC‹?<(B¬fŽUéeÍœMéûq—™5sv­]µïêãIì¯5sG Z3ÇÄš9z]kæ¼j̼ý<5sô¸ÖÌQYÍœ¹,ã`­™Yÿa?_y0…ÿPjÎ~2YÎ~cj+>ÆSB§o™õµfÎε‘5svŽƒ{’~­™³‘IúµZKÖ$ýZS%ëÇ`q­™3´NˆÖÌyX4IÿÃŒ”öL¨¬5s6–zPÚçN¦´Ïʵ“5s6–rp¬™³±äƒõ|pOž«“X3gãê»æì˜ÛÁ’væ ך9ËÊï°fÎÆ¾åÎך9Y3Ç2’jÍœ ÏeŽs­™³±¥ƒÌÙÿp•¬™³43·ÖÌÙÈ$ýÛ8È$ý×ö`­™³¬LÊ¥Õ–vP’1ëòµÖÌÙ¹ð^ëí`3ÌÎ<Žëäùו6â†ðN&øX¯ƒšÑßÈâLˆÎš9/‹fôXòAI¨¼‘ý®¾óøeìûû¢fÎ¸Žš9CïûkÍœó:96¢fÎÎ\j­€©d­€‡#Ÿ\oqÖÌy94ýÿC–x˜ñ\æ¡gÍœ#D~-74´tÓÔqÿôÎÔòL›Êš9;WÞˆ3wgÁƒY ;ƒG¶I˜rìx.S­³fब™3ô^¿ÖÌÙY‘K×DÝr»s–I3Í3‡=kæ¼Ävˆ’AycI'‘ÿž Y3gg¿N"K½öAÍœ©,×ÁÚ¶|pž^u#u+eS5[<Óÿ³fÎÎqþu•Ø’Âs¿[Íœ9ìÙ³fÎKÄmìÌó`EÂ{­äƒ½&;;>†ÖŸÁ²ó¨šïŸ5s^¢fÎÎ|¬HZÏn“»ß^NÍ%ÿ0åƒyèµ®VŽwÎrp^c#6Ë­w«5svÎõ’5sv¦ë$^™½kæììç_ç܈=õ;S?ØÖÁ~þuâ•™4—5s^b©wt«‹‚9ÈNéúv®ƒ³oD’•9½™õY3g§T?Ø8ι{cvæë`Í'×ÁY®ý0³ÙÙ×Á¹È®% Öëä8ØX: )%[¼ kµfÎCÖÌÑm«™³q¦“û_Y3gcº6¤ðç)©5s.­Bð°0ýÿT¶t™õµdkæµfŽîµš93söW¥4†‡¬™³±´ƒ,À”ºZ3gcYI€9°µfÎÆÖv&éOÊ1NdåçEDkælÌã`Þóš¢5s6–r°áXIkæl\ÈŽ¯fX3gcY;óßwåD¦y¦ñfÍœyYm"̦n„К9;'^Š ÖÌ98övðx.Vw2…ÿC¦ð8˜Òž¿¶·ïÌó`ŃY¥5s²fÎNI"»q\%ïïι»rv¦|0÷ƒke¦—Þ˜ñRZÊ1§ÞhÔš9/±l¸“Ï}¨Iú­\ûƒ‘XcgïG9¸êFÜóšzÇSkæLMAª5sžìø¬™3­¼kæìHx¯…pif½±fÎNæÝX‘iž—ÖÌÙÙëÁ‘6"ÏôΚO®ƒ’ÂãÚÿŠ;S9˜¯ƒŸª™wÿḮ´{væ~°ç_×u_ÿ©{ÃDûË:Èœý{ÍœiÕYY3ç%óIoÌõ`aVþ®dfý‡Ì¬ÿ™õ•¨ó°“™õÖóÁslÄŠéÎrþµe²|­©Rõ¹S;XúÁvd²|%6ŒLÝœ«5sfÕ>–5sfµr4M“å³fkæ¼Ä½×'Ã=kæì¬ó`k{>8Ïç®ý¹ãȬϚ9;ëu9ûY¡‚5s^Nˬoì'™¤ŸeèX3çÉÏš9‘–~¯™³³ -=Ǭ™³³ƒ«½dÍœ¹¬x0Ë£±fδ Q¬™³“Yù2³þK| N‡¹¤°3çƒÒdzfÎΉÄó¬±Éš9/±Ic+Þš9+ÌzwZ3gç88ÒÁÙ2 ?+¼iÍœLáÿ°žžÈ%Ïu­™³sdf}.¤hÍœëà,Y3gcÊ'ñÜÚ•LÃÿ°Õƒ}\cc׬ük?8ÚÁy>xíšïÿa™[>ØÇAMÿO²fŽÕPÕš9 söW%ü?ä=œHZÏi‹ÖÌÙ˜‘Þ%û´fÎÆÙ_jÍ[Öš9×ÚÈš9k>ØúAÖ à²°ÖÌy˜µVÀÃTæëä8ØÎçvæì_Æup]±®Å{‘¿[Íœ’&vcG*}V³dÍœ Þkæì”¼°Y:à!_êåÚˆm;K:9v¼2 ²fÎΙ.¤´ç ÖÌÙ™ÿþ)¡SŸýkæì¬õ`›{?8Ï—Zx.»/ÖÌYz]kæ¼DiŽ’5ÖjO±fβ⥬™³så“k#“^o¬ç_Çù×5_²fÎÎÑÎíS±fÎÒX/­™³4õ’ÖÌYšiSkæìœLZ_5sv¦t9ÝYõ—5sv¶7Ó¼ÖÌÙÉðJ(ìÌõ`Å+?uZ>¹ö~p”ƒsnDâ’©lLKŸ”Lÿr\u#R”ìdzø‡ß9=d:óLÏÙ"kæœ ÌYkæì\k#2.Íû©5s®ƒÌÏ%VÖÌÙÉ|ð/çFfÈÞ˜ßôðZ3ç žÛŒL¯•|°8³“9ÝÎ÷ÁZ3gg9ÿ*pãL™ÖzcÛù×ÞN¼/rZ3guÍÅš9;ÒÃkudY–L“5s^")ÁêššŠ9Ow6¤i*ê0z2®¹{3vfdOg… ÖÌÙÉüèOUœ†çvãè'žË|·¬™³¦•îÁžÊƒó “š3¹2!úC&D8ð©´kæXBt­™³qÕ—Z3gc¹ÖóÁ-D®q&Óš9;Ï¿2ÇùʤæL”¥5sfÍq^Ljþ°´ƒ/Õã:9Îvp嬙£é}´fŽîbùÝŠæÖôæjMß¼¹ÕÓçëU¦Ý<ún¦Ù­éÇóãuºöÓóóøùyü:ß-Ýùã\Oëë3½;«è`k޹óùÓ<ênM÷¬+žZIç°¦?gº<ÖÒ9Œô¸'Ó‰Zzt–ÓÁ†³¦CŸæÙw3=ßî¬éÎÍEÓ“›k9Ý®™n¼šÓ‹óó°°Ò‡›{9½Ön¦Ù­éÐk:ôÇ鸯Çë4Ú×Sʇéþv£=kxºÖØAúps)§5]ùãÆ×ãï5óæ¡éÁ͋龙^¥v—uºóïÕ¼únMǼ¹î鳦cÞܯÓh¯Õ>oeºèÍ«íf:¹’™³¦óÎê–Oþ½šçõñØÝÙ>7£}6«rÄô$»;?͓馟Š;×uºÓCÓi›ÓWk¦'y=™¾y÷<]˜N»›[?=øþÓ¼ÒÇLçÌ‹‹ñNšÎÚœ¯™®ºš54¯”Ìcrzmé¢Y“ç5‹òÖôЗéÒí’_žtÌÑo^ŸÇk:huú »“kçu0W óŠâ†`ödoU'ÿqýüR'o¸—dÆ´:SÆ¿˜ñψ×D;{¨Ÿ¨ŽÏƒ¸-÷L§q'ê\Ü?ï6ñ4ÆSOºw÷B|ô­Û‰¸~qBÜt>®Åݦ{v5âL~§g…=æœO÷x¼²ñËêã$½=1ïxV÷JjÄ-ªûrOÛâKÔi¹ â­_[¯(îÉ=š{Ÿ/¨÷þö3ãOý j|¶G‹Ÿˆ¿>nÅÝwÝhnF—¹!¾šu4?ÄyŠçT3¾Q½Ü ñÙå8#›u* qÒâ–ܽ©3â:ã‹ÑëcnÃ½Šº ¾5z™râ¤q|F¹¹f7ã¦ëq¯îˆ—¾fÜó‰»f½z ÷J5.›õ>ê´Üqà8>±ÞG½~‡úÆE#ž™õ>©œúÄIÞ¸iÖû$Æ žzŸÍzuaô cüŠ[q÷¡F½:ýÜŒ“ÎÇ qиe½yºÇ³Þ—·ÞG]º{<ës{×ÓÖG½Oª§õ>éÖ ÞG½Ì¨÷IìÄ>õ>êÂxèB×ôxºWýy„±?a½za{ذÞ'u~þYXÎúÔû¨;â“QOÅzŸÏ¨÷Q—쮈‹fÝâhÔs©o|õ5·÷ùY¿wÿ¹gR£Þ' žï±Þg_–ºewGœ3ë”ú‰O¦g-NÝ]~î6ݽ¹ÇrOÛÞ|~âqÔí÷¿/>ï¬÷Q3NŸWÖû|FœŽ:'wA\3Î?Xñ«žî•Ŭ÷Q矻1úG÷ì^øýp>ÂzŸ4y7õ>i1Þœõ>¸ìý»õ>êÇ ç¿Œãë}Ô'~ùúY1nê}Ô%¹[qÄcü²ÞÇ<Ý«©ï­ÎÅ]°=ŒÖû U÷ïÖû¨÷|=îUÕ¨÷9ñ˧ÞGݦ;â|O3ë}Ô«¨?¤ÎX-›¬÷9qȧÞGÝ«{0î¸Ñ‹ñÆøýyZ–Yïc–xâSï“Ïñœõ>Ÿç‰¾füïuÄS^°ÞGÍí]¯©F½:â*Ë©B½º§ÇˆÆñ‘õ>¹²ŽŽõ>׬÷Qïó u+îŽxaœ²Þ'Ÿ:Öû¨g~¼Ô¨÷Éãõ>êÚ¾xÞSïóõ>êò{<Üm¹ûóú¸Û˜qê}>£ÞG]~îÞÜãY•/Î÷Ôû˜‡»=Ž8õó•ë}N|ï©÷Q—æ®Ïúã瞈F}ë}>÷G|îšÜŒ¾žˆ×E/ë}Ô¬îeËç‰û½Žñ+¶}Ôûd>ñxê}òàý*Öû¨[³ÞG]~îåøB‚õ>fÆÇÿƒõ>yžúÔû¨{S3î\̸ßë†íãxËz5ã~¯â†QwÂzŸ×ß{°=ÖíÔ'Œûk¬÷) *8õ>ê…øZÖ+áú­œúVÖû”Ï/Xïó×kêÜÜåY¿bûüàû„r®×Y¯Åñ…õ>ŸQïSÒùûQï£çï1âb9~7VòWß³¯÷ÕýñB|,ëVpÿ_§»"¾Ÿ_Öû¨—-O'>÷šq¹×ãçžUzŸrê¤Xïóñà¥ðþ3ë}ÔíYq›âw £ÞG]žå}¸¶ówÖûœ8ÚSñÁø€S÷W¼°>®GYï£.ˆ‹e]ê}Ô#»#òÔ²ÞG]º»ýÜñ®ø¾†õ>ŸQï£.Íã§ú!Ü/W×ånXŽúXÖû¨Wú|ê} ë@O½zýÔˆóVO{=îO«Ÿ{½ìõ¨÷QW,GÜ&ë}ԌϽf|.óuAm\zuK§{ ~qø¬÷)¬=õ>êÔ$~–õ>êøXÄ)³ÞG=ßš͸XÄݲÞç3ê} ›ãO½º5wì?®Qú¿[ïSæWÿÃx×zÖ·xWÖû”yþßóĻҨ÷Qïó岟ËzuìoLxê} w§Þçšõ>Ÿq¸°nòÔû¨Ët·gýþ¬ãíõ>ŸQï£.Ý]‹»W÷HjÆ•Þz!ÔûÔã2Yï£^Kz5ãd¯'{ÍøXõ>õwêxP?w_îaë£Þ§2fèÔû˜OËú!ÜUïñ f\-¸mµ¿õ>ê9Ý«©q¾j^©'®Ÿõ>êò{<Ü-=ÆëÿËzõXjÔûÔÄx\Öû¨WVã|Wñ®'¾Ÿõ>æåîEzuıŠ#îU=Ý=K<0ë}Ô£«Qïsã‚YŽžåKã†YïS3ãYï£.ùñr·âî?÷°í!N\͸ãëQÝËÖG½:w­îö¬Ï8eÖaþy¸WUϯŒ8pÖû|F½:â„Å¥»Û³>ã˜Y…zuŒçrþÔûÔÂßõ>êšÝÍ–£Þ§2màÔû¨KuÇx­_P_îÉ8æJÇøScû¬ƒA½:Æ‹8öo§¾‰õ>Ÿqþ®NIâ—YïSuê}ê‰óg½:Æ£8Æ£x.5ê}Ôµ¹çTc>º w{Öñ*ž?5ãÖû©ÇA½šÛÃùë}jgœ7ë}ÌÓñì'îŸõ>uœú(ÌŸ1/ÄOãç£ÞG]±>ëŽPñ)Žý­˜ñÖ×kª÷.Þç{užº!ƹ«‡˜õ>•õå§ÞGÝm9ê}êäùë}êd\7ë}*¿8õ>êר÷QG<¼8ŽïëÔ« ÞçŸ#ÇàÔû¨¹}œo²Þ§ýN æ»´ß©ËA½O{ê}ðxõß­÷Qïñ¥^¶õ>ê=>Õ{ÿ§ÞÇû–XÁzŸÆxÅSï£ÞûßϘÓ8ßàÔû¨kv·gýaë£ÞG½Ç³y¸ëóú=¾Õ­»÷ùhãý—Sïc^aÔÕ±Þç3æÇ¨÷øUïë•vêYïÓN]"ë}Ô­º÷õ†zf÷þ~ë3®_®Yﳟ‹¯Ç#Œú[Öû¨÷øUÇxºÆü–Æï'N½O»u=¨÷Ù_SW:Æ“xß?cþ‰:Ƈz¹ãý«o¬÷1O5âþÕ{¦^¶õ>ꎟëÖû´Æz0Öû¨÷þéó8˱¿b½îÿn½y¨QïÓNë}Ô%¹÷õAëÜ¿°ÞGïgg½ ë}Zçù8ë}Ú9Þ°ÞGŸ÷[?ƒùêÙÜ\ŸÖû¨ãý§.÷ÕûøÖNë}ðwë}Úäù4ë}ÔûzQŸ/q|Þ'÷¬÷ùŒzŸvêòXﳿ†ÿë}Ôñ~Š÷þ¿­S‡ƒûíÔͱާÿXïÌzŸÏ¸¿§ÞŸ/uoîý~õß©ãÁý=düÝzõ˜î}þÐÓùÿb>B?×{¬÷QïÏ£z¿?êù¼~¿ŸýÔ¹±ÞGÝm9ê}Ô{ª®¿ÇýÏ7:KN½ÏgÔû¨÷çKï¸ÛöPï£ÎØ>ê}ÔÜÞõþ<öÂûM¬÷ùŒï×Õû|S½Ÿ½œºÔû˜g˜õ<¨÷QážUzõ²íá|O–»?ëÏì^?1ë}Ô%»÷ý5u×ߟõ>êô,÷õq¾öõ>êZÜûü§×¯¾gŸ¯«÷ý2u¼?×í,gê}>£ÞG㹺Ôû|Æý%õ>¿í§—õ>½sÿÀzŸÎùY§ÞG=u9ë}ú8õ;¸¿£ŽýÍ8õ8¨÷QÇþg²Nˆõ>ýÔ‘²ÞGÿŸkÔûôuêxp>0Îù3ë}>ãx¯ÞÿßϨ÷ùŒïÛÔûów|ê}Ô{ÿ6þß§Þç3ž_ùŒzŸãÿý-öIÿ¿zŸøÎ’õ>‹U.¨÷ù”£þ&¾=¹õ>ñ¸ö­÷¹*Ôû „%ÎnÎ5ê}D+*gðºø¨q¶4ê}ZÂ\XÔûˆj6­¨y¡zQ(ZÉü ³©Vþ„zŸO5j^ð{¢Þ‡ß;²ÞçÓþy¬Ìd½Oáÿ õ>ŸVT² ®‰ª“øîõ>Ÿ¤ê„õ>Ÿví+²Yï󩡤dýwë}>Íò õ>w*Yïó©EMHŒ*ÖûtìóYïÓY·ƒzFç±ÞçÓ.Ø7°Þg`f=ë}žûb½ŸBa½ÏÄ>€õ>3îXïóiEyêSPïÃg/Xïâ<Öû|QÃWõ¬÷Y˜Óƒz¼úÔûˆJ ê§ÞK§ÞGôï IÔ›j'ÅW,¢ÔT9êPÚƒ|íÄÊ ÄkŸj£˜X/?ÓP­ü)&T¨FÔäÿN½jªZSõ(ˆ»„¨÷需zÊ ˜nßùtê}zÁw¨÷íØîO-‚ûc5ê}uê}®:ëŽÊRu[6š PŒè΀*ÔûˆjÄãÇõ>û¿Ã×ÍX3¾íB½¨EÌ}<‰‚zÑ¿ñ)ZõSÔûô†2¨÷ÕX3fáÞG5Uc©füô¸‹„zÕPáçåȰ£ê}D-bõ×É4Uk}Š)ú¢¯‹ã5ê}D;ê›U¤¨÷éü¦õ>Yš¨÷åŸi¨4 ˜Ì ùÆeÈ™ýê}T†ŽršxLT4마žQ-U秨FFz<™†zŸÁ·qR¿FÿSVsbPï#Bâ÷QíªQÚ(,Bî%ŸB½j¨rUÕÈœFuS“Ÿ~ÔûÌŒcê}D3‚Gcï†zŸ+DZ~jëR(?•¡BêrPcÌ£Þg;Pïs…°ÊO9¢Iã\õ>¢aËfÿ—U³©–,‹™É¤#ÔûˆFäfâÿ“[fÅñõ>¢”MKU{¹þ;õ>¢Y>!]’ר÷µú)å¦*Ù´T5+ã¼õ>¢QU3b1ã›FÔûˆjSµ¥š?ÕŠ5ÙœÁ$J„MÆÝ6Öû(k„\"F õ>ÊÞŒ#W¼Áب÷™ƒa¢¨÷¹‘¨÷Qî=Á8Í*‰ëi‹‘P…èy«c¢ÞgÎS¥7'/‘Xï39Ûšõ>ÆeìÕ8üµˆ DN6ê}”Hò[—ÃX"¬¨÷ÙS‡/§q‘h‹zŸÅ &Öû JíïÔû,¶¢±ÞgåS£³_”u{2íô×®~Caê}V9QgÛeÉÆ{B£¾v0í!j—3^Ë ¦xžÐ±aé0uc­Fdª]Žxm=\E¸˜šv‰Ô´K¤¦Ã6ŒcW¤—±Ñ%ê}”9wr’°f#‚Ð0&Qïó1&Ã(ó0ŽŸÓ—"S L¨÷Q"Dm"í²Uã,ÆGáÅ'vXï£,Ùøïµ8ý–õ>ÊdØ3 ÞG¹¹N”Yï³x½ÄzŸ½¯d™Æö¥y±ÞgïŸûáÆÓKÊéVø ¢Þg­Ó`'…kÿ$&¤ï=o:^îUÔ „ù1?±»-t^î^Ý£F’Ñó#f@L:;˜°¢žÕ× €—ê®ÓÝ5èÑÝüy×kŠÑùc®?·mµ?∬4—¬f@Kæ^Õ?æ’OwC "¹ñd¥{¨£hƒ @qtù¹ tÍ¥ë‰õñ9Ï t©Ì$E9/wKîñ,_L‘>3ÐEãQ<¦{-uGŒ8OwÍŸ×÷âÃ=mûãUœî:Ý Nh@7PKO£OÂïÇf!LXQGWÇì2¯ßã¡Æ„{uANžPô-Aæô{<ܵ¹;–cÿª 1&èŸ@'>ùenËݱŸô™c|±1ÈÜ›{äÇË=».êŒ$œ^¢8È\‹{`}VÐ$®O'ÃíAq:Ñéš#g(23€ˆnÌ÷åf ÿêú{<Ü­º{r¦@\ø¢ƒeBæ™Ûr²¨‡°?D£y!ÀûGt ¥túòP*´]Í@ ÏÓ=‡¸2 EÌÀ ëœÝ¥«Ó BH,Ú…ÌÓÌqF ;uò üÁþ3ßÍ =(8AÉ9XNÌreËuA8w®žÎ jH\Ïë¯#ÀBÌ€ŸëTGÀÄuC`¿Bdã¹ÿÜ3¹WSwVˆGqOÛPÔÃÝÄÿ&à§z~œ´³|b|ˆkq[ðÓ9/GQædš>×5/(!2wΠ@ßD›gÊñT§Àsà‚ó t™×O5øù@‘¹÷\jŽˆýõ˜ŸøK‰Ì«©ë ̹f`Îõ(îUÕíð\30çº>ë·ô˜5‹Yݘ#NÍeœÂŽ¢8Mtc`L¢;_peƒš"q¼Qæ_êåÓ½l9AN ÛŠÌùç.Õ݃¬~4™_ð÷b•:ww{–gùbÀK>b4™ Ypüî I¼Õ"žK Uê\ÝåY`9Î'Pad^M…)ŸzÕÎ@q:ˆ#€áº €A\°}Z¨2ÚfáR|ƒ,®çõש¹Kw·Ÿ{! Wñh42çäÞû7õ@@ Î'Ñjt`Xkd.Õ]ÃF¥Žñ#Žq<ˆ—½>f¶˜ dâÑÝÓÖÇù\ÙtŽ#ó œ‰ãZŽÌ 0¸Ÿ€ž#3hp½‚¦£\ItíîŽ@Q¢ìè3ڎ̹¸ë³~ËŸ×çõ ‰/£Xz$f ˆ8#Àè=2÷æžÝ½hr+‰" I\º»>ë·ßc Tš%× (ÁùÀ¸%4'Õ¥¹ë³þ¬êzN>ã÷a‰&ª{~¼Ü1^Ïx´!mãüuHîå®?w«îØ¿Š×PãuŒWñhî‰õqüA1’”ê2Ý )©Ñ+Ø£IŒ ”ê<—á®hÁ78¨HÚ—ÉØßˆB3Eœóã厀‚ãÉ@ñÀöðíª’â2=öèJ23 …u2 ™½â¨KŠ€“EŸÀ˜'âTÜù÷x¸Sðy70åór¯Ÿº 0A\h‚óT'™¨rÍ@•ë…€œ£>)!'ùïô'EÀÈ¢kwÏüx©°".ÁxD‹Ò6ÆÓdàŠx p%=(RŠÛ™^M@JuB ›—âî·{¸ØqÝp/P¨´ý;•â6Áùy«ªqÿ[=`\¡UÉãU\‹»=ë¯üضsuÅrìÑ®·)½¦ýJæÑÝ_§Rz¼Ü•…fÀÆõ\êŒñ(fÀ  ©Î´Àùº–ÌØ mIŒFÔ¹º+pK 9ΧrÉŒŠÎö& ðþ vÉœð@=®ÏP¼dæöŸ‡›ØÓQ§ææôhVDÿ’¹7u?àžîò¬Ïð¯ù€==ð€¼¸¦ÇÏú­»;~_”µ¢‹É=Õ<ÿg¿P¶cæÏÿ<Ý=«Æ'§‘³“É<óg¶2™ –Çù4{™Ì3©q»6¨£šÉܰ<ÎçXθú<ðg<\#@Q݆š1â^Ô â<Ü­¹ßׯ©îx]œ»»w}^ßž×çõÓ^?ðº8¥ÇÓ›»Ûú¸¬.ùñr×înÅ=ì÷Ã?j(ÓpŽö&sù¹9XÀÃ@™Á€·Ä@™ë˜½ï~–g º7èŒ@…Á@¿Ä@qin\ eNq[óGGàeS:ܰÏÉÜŸåóY¾À†+v«[us{Ÿ—{t5dÄ%?^îú¼~"€=Y(tj·€ …NŸQè„÷¿[褮ÝÝŠ»?¯öz|«+ @ÉB§ó@ÿ)tj™–,túŒëu›î±Ü³«ȨÎË]žõ+¸ËB§óhÒ)tR$f¡“º÷ZjȨKuÇñ¼>8…NæåÝ=«{çýÈŒB'uÁ÷g¡“:Èw>€<º{_ï|FÀ Sû»…Nêþ,_]numî–ÝýçÕ=Ÿí-ÛhÔ¹¸Û³|<Ëçï1p‘:w©îšO÷ø=îùlo=Û[¶= õÎA:©Û³<mij¨qÿ\]²;ö¿âñ¬ãóH°ÐIÍí}~ÖoÅûWñLjÜ?7/w{–GÀŠxÚïÇ€#qÂú8¿a¡“ºMwoî±ÜÓ¶7ÏÏ»Ž@¤SèÀB§~ X褎ý£x Ðç³,tê§P…Nꌀþ³ÐIÍ@è°ÐiüÎï‹B'õÊb:Ä€{:©÷øœ~ ëN¡Óg.©÷þR½ÏwÕóYÙú¸¯Þû[uO±> ™p?þ3æW©÷õÜ(¼ÃB'ä4ýÝB'õßæù¿&þ»…NƒO°œB'õ>þãkÞ¿[èdÞÛcpÇ)tRïχz_ß©÷ùîgÌ×2w›îýùPìÞçŸ8¥Îùñr7ü|Üßb¡“zL÷Êêq~Þu|>Ôx= Åp? Þÿa¡ÓgÌß§p…NêÿâÿbnïzÙë1?K½÷ÏêÏç~" ÌËŸñ>¹f¡“zŸ_¨[qÛÏc¡“z?ÔûúÒ¬¿ P<…NêÚÜÃÖÇü-uŒßÅÂa:Åý' Ôñùïó!õ´×ãûˆy®¯Xè¤ÞŸõxÖߟ‡Ï(t2wƒq~ÏB'õÊj\?ªKs×gý½ÿŸ ߇ŸB'õŸŸq}©Î?÷>_Rïñ¯Þãñ3 ÔÛCÁ6 ÔË–ãû õ¿3óú†…NŸ1K½÷§êŠõñý fá÷,tRÇû_x?……Nêù,ßãé3¾ŸPçìŽñÄØˆSè4+ß,tRÇxÇx9+,tRïë¯Ï(tRÇxQ?ëïó]uŒŸk:Ív —Pè¤æÏ»ÞûSuÃëãQáSèô×oêú,ŸYë5õ>?ž…´,tRÇxïó×Ïø¾@ï¿xï?£ÐI±þ:Ž×ó÷cþ’y¨1ŸI½¯÷' NN¡“z1 Ôyºãý·Ÿ{Ùr|? Žñ"?wŒ—kÌoRwÛ>áÕõçnÏò}¼š“Ÿ:}F¡“º<Ë÷õ´z>£ÐI½Ï¦ñw Ôûx3¿¯a¡“z55æ©Ëpïëóõãõ Ôûüî3®‡Ôûz\½ß_u^¿ÇÏâó¶§ÐI½7êšÕ¸_®ÞãCÝžåûzeÂ4:aýÝB'õÞ¨÷ñb‚ :©÷xøŒûÓêýþ/žÿŸB§Ï(tR×gù>ßPÏáŽ÷ççÿê2ÜõY?Þõþ} Yè¤Þ׫Ÿ1G½÷ê÷?O¡“yºç³~Œ—k<Ÿú…Nê:Ý{ÿ¯ÞûÅãË)tRÇxïëKuÅëQXÀB'uŒ—Žù<§ÐI½Ï—Õû~‡zïoÌÏúû|äš…NŸqXžåeºÖG`: Ô1Þ®ñ¼èg:©Kwïï70-êï:­á…NŸP«ŽÏ³8Þ?ñ>ÿøŒB§u]Yè¤ÞÇ õ>?ýŒB'uìÄñþ/|?q Ôû~ÃgÌ·Qïãƒy¹÷õ’ºãõ,ìB¡Óg¬«ùúëç{ÚU>þw>Ó°~ô¿ý‘8ÎWÝë¿=ͪ,úßþæ3 Ìå÷x¸ÿíoÜÏë{sýù(t2ÿ{cÌ+«ã|w{çé®ùñÞ>ÏßYè$Ž\s©îšïŸÇBI:¹—{ô0öÇ(t2ן{÷z–¯¡ŽB'snî’ïŸÏè':™c|‰ÇÞ^c Ì}¸Gu/[?Òk̹¸ku·gýÞÝó÷x¸—m/æ×˜—ýüxž4¦ýe:ïÿo?…\1߯Ü~‡{6wŒgõ£ÐÉ\³»ÙòȰ1çáŽñ,îùñrÏg{ _<ÿÏB'qÆø=…Œ(t2ÏìŽ×_Çù»9Ƴzºss—ìŽñÈ8?:™ç £…Næß‹… (t7Œ_qìÅíY?Æ«8ÆëuÇþOû?±oo`|Šc|ªÿ­i”§ÐɼǣzGóRÇó§æš?ëïñ©Þû[õÀöøÿŽûâx>ÕÜïñ˜Ò)üŠû—ŸQèd®{y>Hq¿R…Næ=¾Ô{ÿ¨Ïú{||ŽB'sî’Ü{ÿ¦ÞãEýnÙïó]Ì©»Ëp×ênØ>>Ï(tÇó©æZÜ{ÿ§îÍ=öïW0–…Næ½ÿýóaö4Ó|¼Ç·:ÆŸ¸=ë[?&n™c<«‡»>¯ñ]ñ} Ìc¨£IfO[í×Ëãõzbü‰cüŠKqïý­ùÙ^{¶×§{f÷êê¸~ùŒB'÷pÇøí,¬E¡“9ÆÓÀó9,t2ÇxÇxÇxïãûç‚ñ!ŽñÁïGXèdŽ÷ï¸!{ß=ÕãAû3ñ²åQèdŽ÷_ïÿuÜ_Œi¼ƒŽýÓõx–ìÄû|Í<Ôÿ_ñ¿ûžÆ‹B¤ê›÷õzÿ¿ÕË^óAÌûýRåÞŸ÷k:Å´âJÏææút:™srïã‹zßö´â8~ ÐÉ=Ô1ß"¦ g:w÷þ|©÷çݼÔQèd¶ý†ÿ¿x¿Ÿê½ÿW/{=ÎÇó¹?ƒ‰ ⸿gÞŸ/uoîx¿Äñÿ¿Žù æ1ÝË^óÌõY¾?êxÄóyýÒŸB§=í9òãXè$Nx?Å{ªÞûSóp/[ABâx>ÏŸ/qmînÛ‹‰æ\ÜåYß·Wñy¼ŽïÛÅñŹ9/wî‘Ou:™ÇpOû}âÄÚ¼l{ëy¢ÐÉÜŸõgvÛûƒB'sÉîÚÝ]:™Ó³Ü×ÇùÚç(t2×âž¶~¤B›Ë~}ãþ…Nâö,ïøû¯öWâ:Üñÿ¼ŽûKæ}~ûyaÿ¤žîQÜS—£ÐIŒû;êØßˆgw/[?c¯žnûÿ ×\—»Ûöq¼WÛÿ…Nâø¾ÍŸ?š…Næñs¯¤ÎØ?]|^éÿýŸÿ÷?ÿ…ý¹ endstream endobj 152 0 obj << /CreationDate (D:20150701163325) /ModDate (D:20150701163325) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 153 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 155 0 R >> endobj 154 0 obj [/ICCBased 156 0 R] endobj 155 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 156 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 148 0 obj << /Font << /F8 7 0 R >> /XObject << /Im10 143 0 R >> /ProcSet [ /PDF /Text ] >> endobj 159 0 obj << /Length 559 /Filter /FlateDecode >> stream xÚeTKoÛ0 ¾çWøH±jI–-õÖ®`[Š1 Ežã6Á'ˆ,ÃþüHJIõ@‰¦øøô‘òm5¹zp‰—¾ÐER½&¾”eá’–R›T‹äž…‚òE¤¹÷PI‘§a†¶¨V¨.[\ö.m} ³‚º[„óoÂåð˜‚Ç=êõFè~¢Ö7B;XJñR}¾z°j„IùL:¥’,À¡[a,¨èeÕ½!ºLÁü@0PŽd¥¥^Ó‰†;ÒQ†úZ¤6÷p¢/ÊÐ-V´Æ^…#>Q™S0ÎÆõà˜3­µTù_!I R™—Iª”ô629Ç;Â*2i).ß…3Ðîãuê·)Rd <‘Ãï ÿ —-¹üB‰:˲XNY™;;âңƹظ[¤]_ã^-·¢ÀtÚ:xŠg_¨qqÊeh*¹Ü…F’ú÷Â> lê.žz7¸åeK‡møÞv!ˆLËwsSS’åÀL »SØk#Ô§ê´M,ÛFh=Á1F…-€³N™¨ O£¶¬"ñkŠì‡8Cõ9©1G§óO`}„™>{B{ºPÒòàQ !¦}&³p×ß3q)¢§ßTÑgA;†±âZØz+Ò3½Š6Nºâ<‡óo@Ï ;Çm êŒóY <ø ? 6}˜ìÒÊ̸sÂi¥²kú!ó·4.¢ò^QÒf6I­Î1:R¬ :ÜW“«PË endstream endobj 158 0 obj << /Type /Page /Contents 159 0 R /Resources 157 0 R /MediaBox [0 0 612 792] /Parent 151 0 R >> endobj 157 0 obj << /Font << /F8 7 0 R /F51 8 0 R >> /ProcSet [ /PDF /Text ] >> endobj 160 0 obj << /Length 149 /Filter /FlateDecode >> stream xÚ31Ô35R0P0Bc3cs…C®B.c46K$çr9yré‡+pé{E¹ô=}JŠJS¹ôœ ¹ô]¢  b¹<]ä00üÿÃÀøÿûÿÿ üÿÿÿÿÿýÿÿ@¸þÿÿ0üÿÿÿ?Ä`d=0s@f‚ÌÙ² d'Èn.WO®@.Æsud endstream endobj 147 0 obj << /Type /Font /Subtype /Type3 /Name /F77 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 5 5 36 37 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 136 /LastChar 136 /Widths 161 0 R /Encoding 162 0 R /CharProcs 163 0 R >> endobj 161 0 obj [41.52 ] endobj 162 0 obj << /Type /Encoding /Differences [136/a136] >> endobj 163 0 obj << /a136 160 0 R >> endobj 164 0 obj [619.8 590.3 590.3 885.4 885.4 295.1 324.7 531.3 531.3 531.3 531.3 531.3 795.8 472.2 531.3 767.4 826.4 531.3 958.7 1076.8 826.4 295.1 295.1 531.3 885.4 531.3 885.4 826.4 295.1 413.2 413.2 531.3 826.4 295.1 354.2 295.1 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 295.1 295.1 295.1 826.4 501.7 501.7 826.4 795.8 752.1 767.4 811.1 722.6 693.1 833.5 795.8 382.6 545.5 825.4 663.6 972.9 795.8 826.4 722.6 826.4 781.6 590.3 767.4 795.8 795.8 1091 795.8 795.8 649.3 295.1 531.3 295.1 531.3 295.1 295.1 531.3 590.3 472.2 590.3 472.2 324.7 531.3 590.3 295.1 324.7 560.8 295.1 885.4 590.3 531.3 590.3 560.8 414.1 419.1 413.2 590.3 560.8 767.4 560.8 560.8] endobj 165 0 obj [611.1] endobj 166 0 obj [585.3] endobj 167 0 obj [436.1 594.4 901.4 691.7 1091.7 900 863.9 786.1 863.9 862.5 638.9 800 884.7 869.4 1188.9 869.4 869.4 702.8 319.4 602.8 319.4 575 319.4 319.4 559 638.9 511.1 638.9 527.1 351.4 575 638.9 319.4 351.4 606.9 319.4 958.3 638.9 575 638.9 606.9 473.6 453.6 447.2 638.9 606.9 830.6 606.9 606.9] endobj 168 0 obj [619.7 502.4 510.5 594.7 542 557.1 557.3 668.8 404.2 472.7 607.3 361.3 1013.7 706.2 563.9 588.9 523.6 530.4 539.2] endobj 169 0 obj << /Length 104 /Filter /FlateDecode >> stream xÚ31Ö3µT0P04W0#S#…C®B. ‚‘)T&9—ËÉ“K?\Á’Kß(Ì¥ïé«PRTšÊ¥ïà¬`È¥ï¢m¨`Ëåé¢`ÇP„ÿþ7Ô3`‡v(P†ËÕ“+ L5* endstream endobj 30 0 obj << /Type /Font /Subtype /Type3 /Name /F56 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 17 27 27 52 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 39 /LastChar 39 /Widths 170 0 R /Encoding 171 0 R /CharProcs 172 0 R >> endobj 170 0 obj [43.59 ] endobj 171 0 obj << /Type /Encoding /Differences [39/a39] >> endobj 172 0 obj << /a39 169 0 R >> endobj 173 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 174 0 obj [569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5] endobj 175 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1 695.6 816.7 847.5 605.6 544.6 625.8 612.8 987.8 713.3 668.3 724.7 666.7 666.7 666.7 666.7 666.7 611.1 611.1 444.4 444.4 444.4 444.4 500 500 388.9 388.9 277.8] endobj 176 0 obj [565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5] endobj 177 0 obj [306.7 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 306.7 306.7 306.7 766.7 511.1 511.1 766.7 743.3 703.9 715.6 755 678.3 652.8 773.6 743.3 385.6 525 768.9 627.2 896.7 743.3 766.7 678.3 766.7 729.4 562.2 715.6 743.3 743.3 998.9 743.3 743.3 613.3 306.7 514.4 306.7 511.1 306.7 306.7 511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6] endobj 178 0 obj [583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000] endobj 179 0 obj [625 625 937.5 937.5 312.5 343.7 562.5 562.5 562.5 562.5 562.5 849.5 500 574.1 812.5 875 562.5 1018.5 1143.5 875 312.5 342.6 581 937.5 562.5 937.5 875 312.5 437.5 437.5 562.5 875 312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7 593.7] endobj 180 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8 435.2] endobj 181 0 obj [693.3 654.3 667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7] endobj 182 0 obj << /Length1 1672 /Length2 11333 /Length3 0 /Length 12397 /Filter /FlateDecode >> stream xÚ´Pœ[-Š„àîÁ:¸»»;wk qw—ÁÝ]ƒwîîî\‚ûåÈÌœ™÷ªî­®ú{¯O÷úd“}VP¦2¶5‰ÛÚ8Ñ1Ñ3rDä„5˜ŒŒ,ôŒŒÌpdd*æNV ÉáÈÔ@Žæ¶6Üÿ°qÞe¢@§wC9[€´³€‰ÀÄÎÍÄÁÍÈ`fdäú—¡­7@èbn £HÛÚ€áÈDlíÜÌMÍœÞóüë 4¢0qqqÐþé²9˜mr@'3õ{F# @ÙÖÈääþ_!(yÍœœì¸\]]éÖŽô¶¦üT´Ws'3€Èäà2üA ´ýMŽ  bfîø—BÙÖÄÉè¼ ¬Ì@6Žï.Î6Æ À{v€²”,à‹Èæ/cÙ¿ hÀDÏôïp{ÿÈÜæOg ‘‘­µÐÆÝÜÆ`bn|—¥wrs¢mŒÿ0Z9Ú¾û]€æV@Ãwƒ?¯ˆ )€ï ÿæçhä`nçäHïhnõG†?¼—YÌÆXÄÖÚdãä÷ÇýDÍ@Fïuwgø»¹–6¶®6žÿB&æ6Æ&Ð0v¶cPµ1·wI‰þmó.‚ûÌä`cddä`ç€ì 7#3†?¨¸ÛþT2ý!~çàíigk0y§ò67½ÿÁy:]@'g·ç?ÿà˜˜ÆæFNC©¹ Ü¢¿‹A&á÷þ;˜»´ßÇ ÀøÇïß'Ý÷ 3¶µ±rÿùŸ-fД——T¢ù›ò¿•¶nO:6v3#€‰‰Àñ~ðþï8ÿ®À¿Øÿ)Ušÿ}»D”²1±pýEâ½zÿ"âò÷dPþ½6T€ÿÎ oû>Ï åÆ_‡‘ÑèýÃôÿ¼ºüÿÍþQþ¯ãÿ¿7w¶²úSOù—ÁÿG´6·rÿÛâ}žÞwCÎö}Clþ×Tô×B ÛZÿ¯NÊ ø¾!B6¦Vÿ.£¹£¸¹ÈXÁÜÉÈì¯!úWÞƒ[™Û€lÍÿxptLŒŒÿ£{ß9#Ë÷GÅñ½Wª@ï+õß)ÅlŒlÿØ=æ÷¾€îpï­GlO¦÷%5¹ý9Ûz[§wÀ;9o€‰­Üå`0Hý!úq2”þ¸8 Àÿ Nƒá€Áèßè›3ÿ2@ÿ€ïILÿY fÿ€lóÀ÷H’Æ^ð,.ŒØà[Èâ\웓QÃ%öQÀq,¡ãK÷“í@'íÌv5Jw°R3šSÕÂx‘K˜ÄW ëgâ›û>¼(Çi€Óüì"ì<9Sí öŠ‹ÆáfŠü5nMxˇ¾ÝžéÃþüð‘¡ 8%! •h¶N²,ÿkzL9×[LiÈ1²®ÀS»ï-—¾ðŽÎÒ"Ø«ÐÉZg»Þ[aüÉP¡î þüôå{GŒÙv}Æ fÀ$Š~;WÈΤó+K|Çe ­3¶0ÒiY&ç tÊㆵ#¾*„ú¤â£±ÆÉgü©>¤™—ÌöÓRÅ3´ÖSR¸d$¼îL}ª¯þä¾@†ÿ.çY·í!w£ê'çÑx-mëT=†A`ýÌê¾ïˆfêú¼³cxœÜ1-þ¿­TË>A7Jå] h»–A|Ú5‚%×úeP„;U’ݾ^…àh, /L¦Àj¢vÜ?@œžôW1æšn„£| ZÖŽÆ’&™Ö0µ>› Cë$B3Ârœ•&ôP)Óže±-PIF>[»®3·’´ÚŒ8l‘z…¿M© ÊB'ÄëNz–i¼=²öÑK$º¢7”ï®x)Ü3 ™®Ö¹^ZvŠWeÁÃûË—™XðD ÄXÓ°Ÿ”t›BìG{>Ü R?¬Ÿ&ãNÁ£Ó*+î"1¶æ=•“SÛüÀÕ1¨ÑU  à>@Ú^kîf·Âµa4äúf­ÛvqI{̰\ü]G“”ý“6¿ªXIS„\î—JĉzÑç-ÙÊíRê-Á{Š¢BòÉDöÂÀ æäÕadP4”‰×i–~58¢¯ 5ƒ‰–”hCqQÑ–)(v•HN·z‡Â=Á8ÛsÕ:’þ>Gšèú·%#qu¯˜0OèFdÈ>”Ÿ—á¤]‘½S¸ìç]÷'7=wrØ¿¡UŒ;ŽÇ£Ê¦ÂÞ“+ÒõØ}ýEJ–*G^ÃQ‡ÚŒšä¤µñôAÇÀ׋-Q®¨‰¸ö+u{%rÁÉòoUÄx¢hTD-m«ì•‰MBìÉpûÏŽðÛãì¡Lb7Æî;3Ȳ gmÁeï­ƒ%âù¸!c©X1jXZ€EǾî~Ÿ }—“[3 f ìyœ¥X—¦U^e¡·q0=HþWíÀHL"´o!Q‰ÌV½(¢…M¥±;†F0ªhƒ×=y[nHÏ%ÿyf[÷Ò º[Uıa³s·^á·©ê¸Amê¹Z³c¦Ggâóž¢o+¢á B^?¯‘UIÿùD¡‡²ˆd£1dµld1û‡Ì¤Q§¾N"ˆUÝIœA+ÓYÈ៖qºšTó–n 8)Î[ø769$³sZ À{£ Û|)z¨PàZ¶¦£j³%†æ+!¾çAjù¤‹üJ† W¥ÿ€(.ҺŒ å=o²Iê-ÚC-ÿüà/ò¬í6:æVaÓ-bORO¢bÞBrm~¯¾Àpi¾G c°<”•Û‹²ô‚÷æ}hÊ3Þ[ƒ\/Xð3‚^²Ò¯¼ëøéá¼dÅÚ†k”)õ‘.øn/ç ³D%¼ü´±á—“< KŸÜšû¶Gaà>í–€ÖE£àþ£†‚°9ÿÕîäñ÷ó“—5ñ9!3ËjÛ”† ¨M” ÎD!R?œŸSìè'•—ú¤èRÄÅ áÑúʺpgí °á¤"Í<Ú0ÔXÊ‚~ã=S¥4Ý»Y|X§,0øÂœÞüR€³âaÈŸ?rC‡£¢VåÙV {¯/y®¬ÝB·ØÞFLŽ"xf¿ ìv>†`5óówð²‹Xã'éÒ(ݧ6NKp²£Hn¤aÔŒ0k!x:OAhNl«¸,‘ |DQ=ãm¸‹-°‹‘»ç½´•Ò¶ªùÙü¥"ÑK3oÜPÌ*ÌÜ?ïXª(VpKÁß«˜ Uâo0Öl‡üï£(+ÔU¥îùÎugRQä¼f Òb¥ïu4K•Ê$ÜýY)Q6ä£wð›<‰±¬tº–ƒœjp©÷àçBç||÷7µ]=ß[ê=k·JÐ]9þo>þáªÏ¯9= Y†œ~“ÍSˆó©r%U?ë"~ -m£™ÃÙEm¦„S 8œ— äøì®1]¥Áw/D§AóòøÓVêͰß3ɤUè´/öº¨°lSâlᨿ&ä7‡ùÒP èuÖfdÊOçá ¶V™¡dTj ‹·¯ÓÊ’ÍZ«âg´ × ƒüÚ¨I«ˆllݶXd(7ÝWj¤íÚyÚ]/A)&ª<¸¬¯á‘5Y¸Š-š¯âkcÃ>š¶™&oØ®¯"~&#Pý„ÑÙfú½¤u•r6À|sм]ÛQWD تÛLýùG{2$n\_G JÝi„w0ˆ§9¡žx_ ÿ‡õZwEv:UÅm FºX”ŽÃ :B×ÌTœmÂ÷It0÷ùÐ%*áѰ vÁeSÿÈ ik€˜½ýRBà5'«ÞüÔð6rÇ”hv>ð™‹a§ ©sÒóÇPQ:Z™Öç:}:‡e«¥hêóV{ð¯é8Y»øK-‹a¹NdEÀM§á¸m ¥aCe[zí£?% 8ü²`Óëå‚,•WR«•Fþõƒ„¹ù;Ʀ†ìdk¢ ]³?ÑyÐiÞ7‰påC6kSø¯=±º“°O®Pñ=”îs~>aÍ ž Ϋ»»—$F®çÑ7U,^«ŽQà«Ü%T X»ê™ãÊBóâ&²þT[ïÓø¿øxxor˜Kœbðrö+ê'ˆ¸ÇŠLvL²,#?H’‚Ÿúøä°‘/ûPˆ ˆéÐØ` à’2SÚï P0`¿[ÞÑQÏœçð(ȽÎ5†6¡>ÀVäâ{œÏ¶RœA-Þ‹qÍÒ`¹Œe$¢"vhưT@©ÅȳطKCª3äJäâÈoµüõ®‡k.Oµ[^¤Øô¬GñÔ‚3Ž+pæ÷!v`u†}Û:°‘\HBe=›Z¼.4¥Ã+eEv[ÇiŠ V8!¢1’ðÜÃéÝD¦—ú³É¿v¨'z/éHí} q Ê~Æ–6ÅCÈìÜâÞ_ÀùT2 ˆƒ¦Lÿý µ‹>é„€-RC Á€aåñg±'4—Aç&&rtøvï²8š(¹ì×Q/ý¹Wú¡òœ8àѦò^2‡*Ù¨%äŽ,‚°°Á™“e7¬X÷r½ŸNˆ¤bÿ±xêÑ©JU°×ªûĹ³Ïž"ªY™e‡ «ôLA8æ!šŒzËx&çeq‚o7 zøêœæ³Þb,Ý©ü¥w 7ü\<ýU¬ºy&°`:Ü5 H¿ì“Ïh†$k9àì¬þ\eŒßgi®È'$È!Gò`¢Œ†0j”LØú°Žc+K9ùüåõeŸX€­=!OGÈ\ëÃ>jT û‘v1‚,ÊüŠmS^£ ÂïU˜’\ñЄ~¾6”ËE΋†0Ò˜¬lã½prF¢)Ûû'gíºűÑ'1ÕÏø8}#:Dë1Žè®boW÷å•K'ì-eê ‘ðßÐ9aOkyÈܬÈýñŠà毠œù›mómŠ×åm3ë4ëÒîúÎT :€(‹‘¬ÄQ÷…iràqÈñ=dLs`;MÚÔú:‰Ò0UêÊÌºŽ²ÔiÌ`x&ê£MUùmÊZ”Ërð²&AÆ`'ÐM†œH°Äó£Í)~<™<Çîàáã¡ÛÛU. •ÌöûD¹_>#½aIòŽÅg­(AÞð‚}į·¯îóV³To/Ä5€•‹™\ô‰€Î5UkÛéÔ3˜%æ0#ˆdo*¿Ì_ñ›Ü¸¼¨Õwç7+âD°9¶Ý}(%Y»²ÈîÃ*Ë¿yÅVη_;íCÕ}Dêò¬5q¯ø~vèÉöùß y6Ð;¢Iw/õÆ!røi53è ÄŸV ÀÓ²ÉÀ$z°[0,ÚiÉÁΰTÉÇ•ËK‘Rñ"ÊR”Œ™¶ëì4u?…¥¯M} 1ã¤Æø²kæG†èP’-…ýúê'³(Hêµ$ªþ#5|æšAA¾ö£Þd" 5£ÌÍ®ÿaïËj#•÷ d…ô¡œ)ñ4)Û#ç¥~»i™ˆ#³H=­Lf6ò\–ÌxëøUÎ~bë:Â#Ò* x·mKYRs#6¦‘†Q¦ÂÝ¢Oh# ÿøûë®Þ÷>ÖÌ!v‚™* ‚/…Ë@L@/QÏÖý%R/§lšœ }#9ôæ×@Y^/¯ëv1Ÿ¥@Â8äÊ«…`¹‚¾å-–/k¾øëÉ7Goj—žE™B‹¢Ô+#;]_ŸTNÞNÁ˜®–†|‹æ=æÌà¯d‰à\òã¤>1zÝ]aüÖ²=²ò;\wî¡Çµ]¸ßN`ÀÀªDªÏT¢ôcó`ó\m™•&¬J·Ì¥çQSÍÍ£ˆ0®$„ÈúJx0Z£ÎTö[›ž€Gô[¤³å“ô_ ˜›¨¢Ã}qn©8afNzaâÉæ“»÷buygm{GM&Ã}æMÒùì3~6bH÷–•‹~ÖZiŸ“Åš6êº;¢ÜãÍÓÀˆÁ¡Ï< š”°m| >tÔ%O®\ÙÖͰ¼:íV®“=w?Y×µ¸IQZhà=ßümžƒq`h,ëÂN7…îZu'ØY—ɧŌé¶Å`S–D$‘ÉFÆÖ&GX(ÇÙ¿6êCÈÁ…úÍÛnÜL„ò¡§fc»_h‰âŸSÐÝ]QùØyõ‘j +«¶€ÄQ5_uc÷µ6iä˜Ú©ËÕcÆã©{xå–V3å5™æÝHª&L(?B¿…ÇÒO¡‰õVM†[ŽxU¯žàÇÜRžuäe»×¹Y„ð¼”Ø ÍuÓ"H~®’»É1ʪØÉçîüL×M$¢eäæDÀ¡ Ð£ lš€iT÷ùXûoGÉXÁj8Q¨z£U¾Áöu¢­§()7OÈ q[²æ_a¬*Ã$y/ñ°ð&­ ×ó%”Uõ ÆÃ£mí®QêÞLˆÙ…Âaóö/ í1ŽÀuñoÛÕç©Õ¾DÊýpZž§U¾0Õ‡Ñ×­Ïa «8¢“Ö˜»÷¨= ‘ðŠƒÉ5ÑsГˆ]ÙÚ¯dD™ÆYÃÎéÖËÙä¿%øPV‰Ö0Ô]/Ë–¡ú†/oŸ/)ýàY‘HJ÷¨© o½ºë¬üxk4Í·öÙL3¨ÑbŽ8Øš ñ ‡šôÜ’6JVTº/@°·1¦\G^ÉpOýÝxk+:f R*-ýaõJ2ØC ¸z=0C ð\ó’¿€²p:ìÖŸ!"Øú/g7ΫÉ8!PüàÐR9ºŽ¦ *QÒγ¼7í›(Š|ÃÀ ˜ÂNBu—Ú#éUÎ_Tµo, ÂRï<û⾞8Ü,]zUps7¢¨AÉÇéRõ@Z•4¼JTŠJ"Už’²“²:ɶգTJ‚×õIÉKáˆÌgØ×éeRQiû ðû‰Ä¹ÃY%ÂpPdY›ÄÓ×Ù3õm•p훿N• î9Ô Zw­ä'üÀ³S!™xÂÂO'˜gÞZ”¹³Øx è,àü&¡»0¥×Ö®Oî‹å5›ºñÞ²Œ0ö¶[Ò §€Ôë¥óM\ÖÝ55aaÿÙ_OncLŽú%帔:ÌÃ;Z»íZô3«pwkÛxþG O™uÎ œ‡ˆÈ,›vIÏœ¨¢ö´ñýíä÷kÈÚ,r×[Ž{æÐ'J»šTx~Å=5 ‹ð`&芊V;3®A@ç#Ì(>kóyø½WÈ¥–Qµ³ÌKØ!™ÀøeÙ\H%{yú$¯DX5 .…HD¬¤ÉqQ¤þÚº: 6s4ß_uzžºÃH#8^;§²&¯2 pQ°ræl•.ss÷iæÄt¶«ÏóBl`;¯Â}#Õ¦r|‘7æÍ$‰ªœÜS½qj=—6ž¹ç¸»Ž§è3òÕ]-ºãVG{M:.’D›À²(ÔÛÉ¥ ¯ÏÝ?ÕQÒ+%ˆ¨9w›Šf±„mö[-šÿ™cp¿öá>æ›Í/¾öÍ &¹µ³tü–Û/nü[|:ô¤rƒªj·&§þ¡‚¡-Ž^O5Bj•‰gg&/ó ž'‘˜¬•nôÝ™ ¥­'¾O–9ÞSê÷†|®²Ûe›CCˆó§½°®ÚD…wkD÷-‚D|Í.ý™…ð¢/Â'ü0Œ ê¼aª_‚Lr¾Ì'Œ‚#4‚èÄ“R@dDëäÂ}€ã·kÔÉ×b+€ nàI-À+Bä4±\ˆ€~=mP´µ—Û—y@GcgO’3°dRsâ]YðéÏ€aÞª£5mcñ§IêíËF€ÕM²m+àŒè¹m}Û•÷C¹™]ÌtFŽÏ/þ4=fŒܴàì:cÉû¤÷€±a_?•’bá&h.êá;i°ŸZ¬níâ™È3¦ á”·…Þ‘¸}&ËJHŸƒVË:–µlÎMùe^kg0ùijPÔDíwEaÿŠ€)«çP|âJèW#Tã°¼ MùŠvNR•XE}Âè+ð†ܥΟ_ÑY¸ìPiJ3FW*”ÆN#Ò.:ÒDU Ù‚.èˆó]¯V¡g]F²ÝvŒÕÚ §Œõp oA ÚЋMW1Tùµ©«³Înu£9˜ÿÍ©`ÄmŸZK¯DáØá[­×o÷>ƒµ5y„Œ9BmúÌè-‘±¼pjc}-fGqùjÝÑm&!_¶>h.5èÕÊ«¶á¼&¤ÒÙ?èËx¥Þ܃<­ºÕÌ™OËÒí…üÓ6®ùg.±}Û8<'&šÈRÒà¶5M穪¬“i:™&2t&a]Ïôÿ6B¨Ž·è@¿Dø¾¸Õ«ïìÁæ€#õóS“Ó”´¢ü£Ÿù}7|í,jí"P-2L˜NÅLa&?` ¹QŸ¯Iw>ÕjÜ{ûßy»Ðo#ú7rZ4ruSÞïLÅ\_ÇÁí ¿Æ¦ûV/æõ´@Z(Së‘ÄŒ#:»p÷ÄQ»‹‰®îŸ¡²?Sj £ÆÖÖOË7­2hPâ}ƒ8!Ã#YèÓ}Ö#+ ñˆù2~ E¿á‰R}7FxIÊ'ç¿ÓZöÈ'„6lj’¸>ÔH×ç&L4“VLUØ qõ×ݵ+ÁºA«"662QŒÄÎè^X-r6sXSHÞöñ#örú“éð:¿ÇH¦Á™¢7•ç‹TIд~.úZ/#Õ½|YAÙˆUCò;Wlm{‹)ùÊ—ÌE¾iŒíÞ-tq…ú£6’ó‚rSçëµ/‰Î¬¸*1ØqáSŸTД*¸p}ã]rÛû^;06Dœ†õߊ” ¶¢ý\W{bƒp‹âÆ*û¯Š…©ôZ,Oá“3Ä£Ž0º5û[{R+¼kv–†WËŒ—áýQNb”4}Qݳz½}ò8O!_”s„pKÖM„»p&>ò-Ç@žÏ_D]/®5ËR–ÔWT ON€×skÏÌÀ¬‹‘:_{à¬]î AjØ.R°ÄY%`ð\©À;•c8d” ¯]OâöSf?© ßòòؼ8áðË(÷4¥þäµy CÓÍÑñÿYÛxsߤ֧pmØ€+íP£È˜å&šbP8Õ.îY¹ ™ñ‚”ü÷€7›ßDA(+ EZTE²uïZÝã*uÙ,#4f¤-BŒTÂTEg4j“¿@ÈGÅr‹ã GØÌøJ›:ôQü‚éW¹K1- TØþ`$Ð;Ëó MƒÎ®I¸x„»ìcíOÇÛvâ—;zLN;OtY;w'¨§œÑØü;Kë¼C+v¹õ,yi¨%\+Öµüh}ÜÃê±;K“¾¸Ð§p¢;cºßeZkÊÑU¿œ·çÓ¬,i3œ¦þ9|±B£{ç¤&ƒT[gÝ Å©ÆYÀLࢋ晄2æBßéþÕ£±N†ðÖü¡ŽŒ?ž*tñøiQiÑ™ZDöã.ˆ~dO®¥&çŽ8fªtº öÐoû0…‰b™|—Ö–DšüU^mÏE¾Üî:yħ/ÛßÝw—iT²ÊKúÒÐAafˆä'¢™æÌÑBpüóŸ©$”–œ!…°íýhÜ€ÖÕ‘ôŸÇOšu¸üùñlÅïàA®Œß¤ƒ|¾ühºô‹wU½À~™4pq×øù{ç&ìµ5S×›ÆkjŒñ±QLÍ–tœ’rõ.:l<Àꛇý°„NQÏ61ÈBxÚWàæˆá×ÄßÐìB¤üÌÇ£Ö}]—SÜúѳÖþÙO†1bŽDY!D=áéþGpãЬ×J´qyLù@pEQ"EÇ38Ï—)4߉Ð"¢¶ŒOÀ„_…&ª&·¹š(¨àS³ÕnoßùfeûW"榟ÒÙGu_'øRɱÑ“¨gl™â±™;Ø!,êq¦Ú¹2ÁŽ TƒÙ¯ãú‘8Ʀ çóá–Ø]X¦|‘©êÞLnGny8øì)îÖ>ˆy#ÅÀÜ£fàn;ø$Ï’c+³ýz·Ï#íßw=s×ÙÁ?ÁÁ#qæžÎ‘lª ò¯¸€>VÓÙ¿1ü¡Øù <Á “›)’(6|Lç·~óí}@p2¼éïÐßüîÌS‡ÜdÉ#pW¦h”/Ù(qô4“û°bâ±›|òªìj oä]#ge¢BŸ‰1·q ô:²¹h0;Ð~Háªj^v®KZ ˆg­¶E t¶¸A ˜©×Šý%ŽËüÆÜû}-Àl2î©Õ7àû)‰ó*7 æxæ“’[w OqŠ™Þ¨} G9¥³/rëi•e.[pì…Æ¼XZo‘ý ÿœ@#¢™‘fPA®¬©ZçM“±2ºp@ܵ£çdJe‚¾×MÓ¾UTÒ„ú•Z»ò(dÉÙ{INǯ±Dlv°CÃø Ú?ûÛÉéÃîxõÞ€àÕ la-oªy@õkþÃÌ“pÊ×È(˜*²e A Ó~—»ÑœHÙlü+«¼aü\k¼n#8‹¾Ñ!Q6ðN4;¤o„ܼ&×^ƒi,XJžzë=ÑW¨ð9t'"ϧó`õ"¥ùH Ú^odFj³¢´vÝí%¥pg.œ;„yEÐŽHr¾X @@Ðr|~¤üåªô–ƒ'  Lk†¼ÿµÕfýð|ц®½‰Ðaò›ƒ¼)9õ¥Wñ»x2FãöÀ[$âP%ÏÝ3Âêñ‚ ENOƒ)=Rœ#&'µqKõo‚HGQþ)ÞJZ$ ‚¯>·Yýve ãýmS=,ˆÍµÞº“¶¦»㹆>C¸ž‘è–ã²oWö”Dª}sò%³kmhüN?”òClÕR”¤ë˜×Ÿåðë;ív^32ìl #òȲCûRSéN&&Ê8kž wKÓ”-ÁžL@»Í©kŽN 8’¿#(¶¨¢Ê»0àES©mD7¸zoH"yb ©´?[îºMÍÆ GוßÁe~â®î¢Š¼Œ¡ò ñq5ætS¢œâ¶[Ô%Uª>²Wű,3,¾-ç€Íä#hþÎÉjY‹n[ÀÝ`OƒFíú½ÖLœX¨ËË[º&±C”_=ie&‰Fug݈×墋´õ5ƒ™‡yŸ¶|³P*2 ¸…ÈÊ0\ý!"¬jö™{–Á ®, Ý eÇl¦,²ŽõdŽ ΢FòdçSÌ49êMr!z¯ C|ee£ ˜N}FK(|@©ÛY[Ý5† ¨Ðü(dÑuHŽf‡^5é6’#bçŽJßôõsꨉLRiŲzEÞ»‘WT 'zŒæïŽCùCÛõçr<‚¼—©×>4.ÓEÿ‹Ë~u¾góÆVK‚@Ĺ•\ëK¢ë¹A(#.ÄÅ®œÃ¼ŒƒÝ‹ôoÑt×¥se=ç+m5VQ…[¯»Û?á?}‚ìGXl®…Wȶóªjzl×xfÖŠ0’÷}{¬±¥ùl‰-z¿N¬êÃùÐê†{ô¿½âr=4=„SŠ8v ÷ Š`Ú‡äÏ6a­…p^! ÓXê;³LÏH´Õ/&ª¦…FöXžgd7?Ä •ëÛóÒ gÝ©{©9Ø›¨1›uØ”¢IY5­WÈôäZeånƒ¤Àdñù`3½KêïÀoÛφëÐâÓá0& ‚mѵ‚#+ÈdUOQ¾õë[£Ú°OÂ+œÇ- ^23G¨™¨iqÖYþ²ãD/«V—ÓjÚˆûà@»æ¿¾F÷½>6æf ö‘Š£† sDý&ª5|×y{€nlü †¢‹°J´g&oLR•:ºÖ˜¡Û‘íA/×j Î1é{àu¬W#ó¾aÓ¦BD§ÜS’]½|'+ŽájhõxYrϳù-,žª~T¹0,ZÒzÂùSpk9xýÒD“Ø[Û€ ¥7Æ”R¬r—Òt½Ú+y$z×—sÎzæDÈ×’p8*XÉ%Òú$*-Õé8óÌ+>¾±–~SWCb ¹à¶`º&Ú;…³n´x*÷‡Æ)K#^;K¿Î($0³‹PØ p$RO} ‡ëîóÚm)ɱív£Q\šÐÅÐà¿Z.¶.àZ­Ù#s`ÆZš8õPî>áf®Ý8‚ Û¸±+`ÌÀ°P(AM4/Y€ŒC ·¦ÑÈ Œy¦ˆWÔÀº‚ûÂd0Ϭ®ÂÌuùîóSèv2¼éA¡_kLÓÁ­8ÌÞÅ‚8˜©1Zæ‡Ã7K®Cß@°Ò&K eå¶8Gžm §´aIdàºm¬bèÿùÑ:²â“N Ž0‡XÊz/ö3†ç,iL‘êºwÐÑu„ÙL¹ÿ¾©Öe£íåªÇ‘‘ÕòËTÅõáV›—=‹œ_&Pƒ„&ãÞú3ãäPŸÆÜÈÅT$c&4ÛÌ¿ä^‚é{þäF,VÑU¶zXkô슡vh}éübn§·î8Á|V‡óÅQ €ŽH®#ê£6YRÿ ó;¥Ê šAáðcøKí<‰4> ÓÞÆ@#yl=˜`–DVÈTrlÆÒ–v"&–¥ìˆiùÙl\¢,[#ïÛÿó¦ endstream endobj 183 0 obj << /Type /FontDescriptor /FontName /YHFNHR+CMBX10 /Flags 4 /FontBBox [-56 -250 1164 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 114 /XHeight 444 /CharSet (/I/R/a/b/c/d/e/g/h/i/l/n/o/p/r/s/t/v/w/y) /FontFile 182 0 R >> endobj 184 0 obj << /Length1 1924 /Length2 12321 /Length3 0 /Length 13511 /Filter /FlateDecode >> stream xÚõTØÒ ãîN°Fƒ»»\‚{ÝÐ8»C Á! \Ü%ww ,¸ÃcæÎ™ûýÿZï­^‹î]µ«NU]zj·lRVN y'G76.vNa€Œª´>7€““‡““ž^âfú¯^u…89 ÿ‹!Ý^l²@·¢ª“#@ÉÝÀÅàâææäpsr ý—èÈ= VUv€’“#È^ÆÉÙ ±¶q{9ç¿?Œ–L.!!Ö?ÃR (ÄèPºÙ€^N´ÚÞ:YB@nÞÿ“‚QÔÆÍÍY˜ƒÃÓÓ“èàÊîµgbxBÜlZ WÔdø£e€ÐôWkìhômˆëoÀnž@(ðb°‡X‚]_BÜ­@PÀË逷Š*ugãÈ*ÿ!°þ€‹ëïtEÿ‘âøg0ÐÒÒÉÁèè q´€!ö €º¼ »›—+èhõhïêôôBì/„?Kä¥4À—ÿêÏÕ qvsew…ØÿÑ#Çi^Æ,çh%ãäàrtsEû£>Ydù2woŽ¿.×ÎÑÉÓÑ÷¿ q´ÿц•»3‡Ž#ÄŤ(ûçÅ„öÍäàãääà€\ /KŽ?ÐövýéäúÃüÒƒ¿¯³“3üÒȽ|¡ùº=@7¨;Èß÷ߎÿEh\\+ˆ¥Àd qDû'û‹þ~¹(Ä `Äù"?.矿™¼(ÌÊÉÑÞûúŸWÌ!'ûF]O‰å¯–ÿvJK;y|ÙøxlÜ|\..!€'Àÿóü=ÿvÿ§Uù«:Î2*:‚BÿiâezÿmÄã/e0þµ6L€ÿ=AÍéEÏ ã?ò7æäã´|ùÃõÿy þ ùÿ§ý?²ü¿ÊÿÿV$ïnoÿ§Ÿñ?„ÿ?ÐbïýãEÏîn/»¡êô²!Žÿ—ªúÏBK;Ù[ý_Ÿ¢ðeC¤­íÿ#ÄUâ²Ò€¸YÚüGDÿ½…—äöG†“+äÀÆÅÉù|/;gi÷ò¨¸¾ÜÕŸ.ÐËJýï‘rŽ–NVì7?…½Ñ8_ÆÍÇðåzYR+ןÚp°;:¹½„^šó€ hÜ(?€CêÓ€Cæo$À àÿñ8ÿAü•ÐKœêßHÀ¡õz‰{û7zaÿABË¿ÑýsXý r8@ÿ‚/iÁÿÀ?äoø"úèñï¿“;ô_ñ/ëA^‡Í¿àË ÿ‚‚ûÁ—Bþ\/…:þ ¾êô7ä}á¾ü'ø—û¥çÜ/“s~¾Ó¿:åz©å_•r½ÔâúO' èŸù^è®/¯Æ?/9Ýþq¿”ãfýk/õºy:ý+àåÜÿ_Jöü×h_xÿ ÿGk–îPèË;üçkð"Äÿâ?}È d‰¶0ëd)fû-¬õ¦ZŠÌ“m{TlŠ~[/‰ÍwÚæ~‡…œÂTõ1d z%•2г¼%Çx)¹Hõè{ÐT‹Ùœ¤Ùrï÷`–¨5±Ý‚6?NÔ;öå@ªæ*9›¶äŽß£‹Ÿn°|l‡}Ž‹» –Fþg‚WÍÒ¥áˆÙmÍ*~eô‡ÒI¶÷:±ÆÁEÓô¹ŸfHhÜØ(P˜ñ~{aO_^Máe=S)%² ù¾ç)ð5\玻ñY)׿ví$¥#5$¡€¿Äž`ð•ÞKU"žó-.ŒWŠˆ)S˜Î6ˆ·a]x¿Rw,ˆ–m ‘ì›Ë”boMDÄ[jnÒKìÛ¿Ín'ÁË©(š|ä2ËGsÕlß}$ò‰©˜þ •èµs;ÒàŽÌøþÁž2ì’Q©}/r/uŽž4¯÷‡PÀäëÚÏ’‰óܹnF]¾@/Ì£v8›Sÿ±1G”ÏFÚGâÚã0ä´ÀA.êl$ó£Ïzý½•xÏvâ°d%ï'à”m‡¿®)uø`=Wû|`*±h*ºs+‰·>碽í6oUjšˆË´VêrDR¾¦å~›Wm”|ðã“´žÄp¹*lˆSØ7ªb ðKjTÞëW‘ÖE×£µ%“ïÐ÷­ª¤­>øM(–އ¿1õ|§Š]Y6¢••=áúTÆ£—”ßÅ><~Þ¡ÄNÄ?W9M!š®Mª:pHS'%a$-W é#üœŠ±J›%Þ`hÛue“7PÇ’®óäYõ|º…öPK—ðóûø¦ñ†ž»L£î¤ñ‰ÚÎyN[w#.ⴔę½ŽkW‰j„FÃu)z{œ7ø­wñÆl‡Å:îÈws!w ç¸Þ+Íþ„j5wè,“Š˜»MIÜLÈd¿ìèBns—ÆtŒi‘?IC#bgç4ÚO¶âQæ‘%JžòÝ€«­7žÄ¡>ÛšÂp3È-TX³#ìÉ„–^Í¥Dù$i¥¯3CðD E£ùç3ù”—¦3k`§X|pÚr9a™:¡åIÑÂsU–y°DÒ˜%¹Šº3Áˆ³Û9zÏMYZ&˜u…z ³(ü,‰Â"¿&„(;D¨ÁD8f$ÿÚ>¸Ê!å4&}«`dAægRZƒÕ{žPWFŽâ²*:±aáE`Ã_/îP§u¿†b>pXꥅʦ÷=’ †Þiõ2c‡5âç!ô,Lvd>Ð9ÿÆO¢§¡X>ª‰Õˆ žù—|\N0a9ð…“è"e ¢¾Br¨ckg^Ûê\ç-$ [HÖ•5d«JíªÌ˜Ö[TNu®«Â8Îk¬Ï Ðv¿.ø=«~PÃE˜XN}LÌbàýº‹»¿<Ùñ™U¬™e¨TWžØr‡çOõiŸ Ýt›C3Ä#$ûÂ])G”„5`æv]HÕZgæt‡ðïÜ"£ º‘l+áqˆ%ál†í£Ì½*÷Oóa>h¥Jü¹p>ÉÏš/ýwëí#ëO–9xsªæ¨ÅëQøÔ`!3q_Uþ(ò_J®*ÄçwÒñSêÓ‘·Qx“Z)}@p:(,J$Gˆ·¤v¸¤¢â¬o*MËúauÄ«pñV˜„ˆS²ÙÌÂXƦ#‘Ò¸oZ2!;·£Ÿ¸` _SJ/~ˆ–¯JŒÅnjÂ:ì% Ñ:€d5.&2‹ßñ7“%&ß zˆÙÈušé3öz'D¨Ý¤#3 ½*Ú%¶â›Í[Õç-EláçŽÿ¡/“…T-G©tWÖÞ€cjKp1ÐLq§.ÉÁ&©>{ÝÚ´kä’ÕP„ÂþiV†Ó\'ïšâ°¡t{‰UjÕ“†…·†nµ@|²éKºm%HR™Y®U.ªÊ Ál€M©-¹Ý‚b4?2ÿÜÐ*þw‰Mdj·˜øq/µ£’¼Uç­xWøÂOM *A ê6âçÙSNï/¦rŠ?½ê„YTù­Äbì€6JE®DzÏO2ØmGŠ7ë,Dƒ¤†0—ì2jû سá÷ô•¶[­ôðÏÆÖ€‘)XC™dŽös•šK;+÷²\TpQW¼/JzëxûåÇVíû@3\æŠÏuÆßÔ„·J#ƒ.CéJå¸-ç7óÛ[X,ÿ«(ˆ‚.¿'°–xƒÎá[„‡(Ž¡&™©ÊÌ.§Ú/-—Ì+ê%fwõÖQ1‹À‘€ã³%ÔG™E Q¾ÓDtm ×¢Ú#wóTÕù ÒvÏÊíSèðÇ×»-g?#V½Õ1¾gg _ür<Ì™(»Ùe#ÈÉyrvR”éRH ò÷P3Z¬<ß2S'W\¿’%nYŸÒzŽJ*èÖÆ¬¶üŠ\Æ}˜Jüà-k¬Z‹ûVèÍüoÍ÷­3Ÿ à …"Óc>¾ÈÚ­B£pƒ±Z&Â]"åŸo LÉk›2*T_éß³ 2ä 1{­èžÒVRÐMÝ8ôð‚ç½Ç:;zHŠPÎËy5•}™ÄF`Âõd[­Ä[œ^~-f9/¤¼‚„±°%5>¨<[Äû•_Ñ:#Òðw(¿ÇR¡ÒÐZó±À*Ì6Œ°œ¼ö>b¢'V™W«sc¯To®Õ¦kuªØjAÏmÐÀìØ3w ­²&ïXm;{"ñæ×ëßÞMa¬ M„°%3ÝvÜ»Eëtg!°¯µºüÉ[·ßèSb©â™Dt£­„d$ŒXî1úê!|'VB/Jí&F•¹³Z rr¬>ËOsI çà ';ጛG¨s–17¥¹™\ú„·g!–¯&kqd@y¿ì©Þ§ÍÀdÍÉ{öŸ€ø^Ç™œÖvjŒP:÷ÃÆÍÍÕB«$ú5€aØ£*I¯ÙàÉ~ Yâ–Æ,_ó(øhÃ’î¾Î\Ìwì¤í7‡‹ŸÌ™}ÖՈ´´,Ó®P%몔>¤IÖ\{×ꋌzܦ½ÚÞpÁ]©KšX©æº6w†ØN/ºh{¶DG>Ǻq.ž½ùêÜi_9m‘æÍ§±*¬iì@c÷¨4N±ÎˆÀk¥¯Œõ'-}Yg–\Š+ ×ìö9šÓ°¶òô¶Ùŵh\Ç->Ü•_õ‚VUY”šzûÀ¬4Ë<ù=U"÷W©y]bCNl»ZôAÑ;뎩äJÒõä’Ž;°ÏüRœûÝ6ZòBÕÙÚ³¾ Œy é¿éÖÂ…‡.l:^]?_Æ«>Ûvî"gcÌ“Ò'ÈÃë†=/¾SØì÷hA’0+¼~ ¬rÖê®À¾ ÜÊ*ƒ‰ÍZ5¸6rìO«¯¸è[…‹;@!˜Zoߘ üEw×AKBÛ±£@§8¯‚ÃrSãc³‡Ð+ñJBj14<Éäꄊ.ÇèÔZÂEó‹ŸßÑØê#¶è»Â?ªB_Íÿ’«ª=ü4£ò@³G}È &aÉ¢†Õ‘ŠJ„Ã`•àÁëÜòH=å§%ëxRŸG‹½Û¢íÇ¿DHp/ ޱÈkäHZSè¨Y¦¼Ñ¸6.16·ØØš¡»‚е3H$p¾ºŠÍr—ÖÁùu° PÉžÃW²E"]¿¥¥Ȩ)ÒèRão1PxárYï”»¤Û¥ˆ¥FƲ ¹Z#ÞWµ®¼1:©Ìü)„o~ë!9™Ù¯÷Üô¦}´Ê|†ë:‰À¡w\ È2†ÝFÝL™0Æ,¢ oŸ¹›¿HLà//û†çlo)dÏ~F‘ü&Ü›k_~÷ø#¸·Ð®Ð<© n¹PÂmÍ]n?(V„5o:Žc“8¾4a;? BRðKG ôøÞp˜«·s}â™…óX•æ #ÿ+=lÕù{-7%ASR¿"ój-±´Ì˜¸·D¦ºÙ,b+oÞƒðfÊW ¹Ý]“0®úIÝß·¬¡4‘ÊNòœV ÀW͇¤ÏÃÙ«`^Óár»Œ¾»ßG„—ÑP2êžVMØ@J ó¯R†,ƒg$¿f? Û \»²àæFßõšÕ¹f}î}ûøÓ+ЭØûõtbúÔD_{&ÏCú`S]°çê?¿TÜÏ“-k×ocþÞ8ên%¾v*r i’"3k?Ž"Ds’7›È°´»TËšŽ¹¡r4Ï»ó˜#œaâÎLYÊøð“Žö“ÿvÂ÷éå¡lw³²ÞXÞãRçP½#ÈþÞ ÑȺvç(Ü–? O}šp™æÃS;{™o'ÞØÍËôäOTLãôŽõz3òÕ9`Tq2¼+¦wÎ ]ÚŽJ© ŠyáÁt…Ù2CÚYÌ^¡¥?LîKwS™.¨ Ù°°n¼†Y©º0Ó&ÛQÁºS0‰‹ê÷IÁA•—­LÖ0§&´3z68%Þ#pµ¶¾À4û«pËH¢˜`„w¬×6'‚ªy¼c¿ÈQ}ö„„·†k6lŸ —Yq´˜ârP,‚¼ÎZ¶Ì®[ ‡T¾7öd£{—E”&…V_ŒñaOhIrø‚4eNýˬ;Qúç®æ· 5ûÔUð¼-Ù —-±`|§†£ Õ)Ó׬éXÇüfKz¾6ó#'Q™¿†mwkÞúO>)Eú O¸FDÏÔÿŒÞ˜µ%F-^­ ðÉRñu‘2†Ñš´¡cŠöQ{«¤²âMnkú¹'¹ïElB/9%Ó¤öxÀ¸²!¼á'b|’’Q£x€Ð&Ã÷‚U®MÓg”PÄv¦¶H}¾]21xµÉ Îü·ú÷\ˆ›³H\ËÖ»A¼ü5Áo®‰¨ŸÎëÔ.»ló6Õy)ŠÛÚ-Vž·¿4½‡uË•?ó>Û©Ù¾SyAø÷åÌô¹@ Rœý„c¦ªç¶¾èGt¿Ìޛܯ߫ž$Ü´-ƒÂ~Cã­ü«¹s¼on« ¦í §Ša0rqÓ§™QU¬ÝF™^¥!Áý8á’ éëPrâ†Î×wT­7ÊýÕt‘ä&ËG¡êèQË™-{|ô·™½}¦Y¿g·6¿z%ͯ°‹öØ BùÐo¥œPGpפ/çóõãQxøH¯Ål/ÛÞ^³Ÿÿ âTüA^´&áÏtƃ!üÓʦ=Ѫ&’©ç(ÉJhÌwß,¹ŠXá<;1vÔ>Ùi&¹tñ×E¿‚´ì€™Å¬*°r ÄΫìÓåP×@Ÿâ¹[¨ p8²CÓÊg+lØ‚¶¼ùzàšºµø`X´]^ÊŽ`òÔeBœ'J]OhÃ4 øî—6zCXj¿Á³ý¨jz¾s¯é•\· ŒÂŒÎ$>¸¶È[2H@ß('yß Oï5‹’Tñ½Â¥Ç :]€ Euí¦æV‚ªr ºTƒ¡ðUi{fb¢“?Áp¥¼fîªpÌÛX¢¤ ›·ÈD£;‹M™:ùXÁ2’Å“Z~ÓëäߊÃß[Ë“BݱÙ3µD»ÖÎ6ëiN„ÙáF3nP*î#ê´Q“aTÈÔ„jÄX€£mÀ\Ž»¹ˆÆZ´g´øÁ©g¨P©MãXÿÅf¾cÛ$=ºr€èòWú¦»7PöTÓ’XpÜñ–.ÕȬ0¤»¢ùÉ3"JǨKÕj÷è@³=ãi\…<^ç)¼²B;„Å`²§Ú,’ÅÜÀŸxù°iªÚ=é}¯`Õt©B^ÄOýžRyÕy=‘5sËѹ‹ˆR%@O–ÅÏuÐJO¿ò›7²E²>/¢²BÁ7Úžc‰]ž6(N"/#æ¨TªÒùRÊ¡{l¼¬ò ìвSˆaá¸H¾§…×yëQ(õCX·‡CJ÷`G/YšÜz+—Û  ±‡ OŒêÌoSþëù±k›Ï?Ú©kß*®«h7‡Üq{‡°Ù à#—NW±× ¼3±¨ô€Š´mðE]|Ýlö‰O kÞWÿù±®É§uü€'Û…‘E‚½ø"6,cî‚é Ë„à>Ê35†/›IÅÂ>8SÕÔÈž°1häCøwàO¢v*¤üŽ%}bäÏ[û-é¨JDtÆäú·n;)g?Lu¥J÷€Ûô•'&UMll&jáý³'õ ˜zBÎÖßMTrËk` &‚~™ç›f è¼þ$å©p}kÈÃõkhcê7ƒ– ¼H×óŽßÕi‘;U A²a”ZpѤ)T–0á€þbáÙPÎd9¦ $—ŠN?‡>U5aÌ‘Iò¡‚¹F+.0œÜÜL¤Hbfptï¯ý™ú W=“òœ;怸*ÃÁD:Ô ¡•ätóëõmW˜/ýïCw3Òço3F8hu­„|ùæ<\âÎkGÀƒ Û{+'íѾ_ ó&ë(JFýOÕ+F{A’} ´ò‚-TS¡¬zѤXØ0Ï dK.[‘}C›t®ÕÇ,üñŸÒ%È ÜΔF4ÄDç$¶sÀÞªôo‹c W»Ð'^N$ü ‰ì‹ ó¹9§Ïùx‰*¦3 ¨ ÜtõÐñ`UŸõ.œÒ²æ¹¢«‘ýÅå;oÅwT>c«^= åê†D[jrÄî·ÖHRH±ëni&ä3»˜â'¿iÄšÊ ‘[EÂ>ªHÞV‘pžZµ!ƒCVZƒIÀŽßR°ÈnðÈ9H¥èu's ¿(ŸNšÍÕût h`ŠYl Ø?(tù#4C›Dú›7 E>‚éíVÍK+mnÛ,-rv ³LŸá%i×òàÞ¸Ö2ÖM—9Nò^ñ­"¶‘r¹ºovëê¹Öi3¯ŸâƒŸ^õ£Rà,Êy$ˆëªÕð°O^†x °^ørCN/Êë¶Ê+‚Œñüž³¬ Âe·T2ñï÷´Ú#d¥“à™æïV$"ûõwÓ¸±â»õ‚ipCÆz'ÊýfżÎ!>˜ÃÃ)Vž†žÕ¾xŒÖ×®T;ŸÏ §êRruhÂä‚nÒ²Ci+ïZSò“íö2ø×ú~R‡ÒIjô¤é-TíBºgÙ›ºú|\µ½ùfÞ‚%õ¨×ü¼l¡*Nï¯`âj| †¾È?G[ ˆ…OŒ|,ˆ$Óhêò*ŸLb]f\âªQËM¤,Ù.n0»¾ñ\‹ï‰A¿DÝÇÞÿ–‹“Ý“Å$ÿ® Ýðî!ÁÔ;S#½9†Ÿ‰ôcPµY‡Ü›¦o”–x×ðd»ë€ñX›Ûów›D<#o“ÎafŸs‘åö¾èsI·º)À'ÏÜypÔkÚlpÿJ/Žešg_0:.CÜ*ÊZ’õŠùû~¿;ù¢ZG_ mÔÍáŠ"ƒÚåPù4818çwʵݰØÒ›ÐKmÝ5Þ¬=ᜠŠqáM†d¬÷:ù*‚•üò‹Çq;µtì)ê]uÇ®ÝÑvcÒ¾øû¯Ë2à¤ÃÕg$]ì¯}ÞgìÖÀt jy ˆèº¦{¨~ÇØéAœfßIs¼)?æU—¤y‰°í ¸O±õlÖ n0.Åg9ÅȲ÷!kñÏ ¸…' Iºö×øM|»Ñp?«M‰¹Ü§ :*üq×Ô&WIæ²ÂO¹wÎsÕ¦.»µôó U†Ê#×âh’Eu#P•öíîÊ–ê)Å0wô³¿X,eƒ`Oùé´®+–Pn3RŸ <§­Ú}0syX†æ¿Ó¤•Ìõ›Ž|rdžeŠp.‰{3ý³'AOMŒ±"ÅáóÙï< ^Ö‹±ÚéÕYY:{Qsuu¸~®ù<+¼»u¥åÓÊ}úÑë¦Þvsô›Úu ?}¾†øMu“hû»[®I(+el!á:5 O%T¬‹ëèÇÓ'PÆVÝöX¯Ê`åv’||SOU0³:)áåË(o¨ÉÐZKs@ðÀ…^ùÇ%ë¶Êž–/Þ¡¹4¦c¨‰íO$õñØ¢ù†"l¥®Oq<;f»¾º#;ãÁ’Á9bfHð­wáS7Þ£¸\s,ø\’¥æ- ÎJÞ݈Nõh¯…=øÕçnfH9`¸ú©ZûAe]d‰w9Š/Lê;HA~*ó§†/²’Žwñ­ñï÷¤Ò¾µ§h]S·tÊë«Úc·d“ZsáåŠÓæ·)b÷¿eó?Yñ$= ®¡Ó§#Ès Õ}ìg3,|zSÑöt¹ÁaÚ€:{ØWÊYÞ '^XžY(šFÏLiaf뾦Û7Û»ÏÛÌV'#-Ùi)S°XP`@ g§ý¦ö‘ÐtI+d·?“£ôáëækvüìê'NùyÈ)G[m?Fß…ÅèT‘c°m9HÀ·ôËË ñÕÄÜ\ò„T?¾‹™§}©!-ifØ2…‚E±•].I·¼gb.Mñ¸xQÖÝ~lþe&ƒ¬JŸ%1#ã.¸ÜõLoÑ´q}ú4tU×ý_Þ& §JËi¡Q- ÂÒ0߆ñê êec3mã›N3½ËOºBlü+Ý:ÏËûËß“Õ9è%ñବC™p ²MœNzÄ.2ßt?¢~ñE.êÌÊ$l韉Ýeiò áóé`iXB¸¾Š³{P (¨Oà ÐÕf+¨Xì‘r9hLQN8öíƒæ3ìŒñlñmé*ƒFŒÔýkheÎÑÊ,T ŒÖO]®ñõ׿A%5òüj û²®ïñÓ^±\`öB§zlÖŠzyXeÛí}h=÷Á轌'¯T½áÁA>ë‘ðä¤ÕÒÇÇÞŽ±| )\;m%Î…æ}bÊ E]&ƒó­%¼’=¡¢vŸÌPеûQŒ)zõ MMš õ+JUW›ßÜÛl9ëÖ°bV4âAþéÇÓ¡ BÆ+Â8v¸z4ÓØ‘ ã¸3”ïÇŒ•[& Ã;#,æzy‘“˜;QÕ³ ¹¿ï黉£á™1#\Å›M›ßÝ•Æ÷H‰‚¹$cÉxÑ4ž7»¶3nZ–ášÈv«¯z„q÷ÞN+‘Öùû3Xüžso¯UËíl[çr´5n27ÞÂ~,[ZÚ¿ •‘7ãp Ë–àN ß‹ßϾÐêIÙ1þ9½|üÛø–|¸˜\âL7R¦¹j±Š9:ãΪÌe›œÁå6´—uöf2»ñ tvï«x™Z•’§ûAäFÈrù}ų§"ó>g]’OwÍ¥‚1Ó³D$¸o!aV•Á4e! Û&ÁbKñ5yŒI^²ÿÁbèÊ‚;½üÓ¡n`ê,UG’bOs¦èûÅÅìµ, âXÁå">XF}¾ŒÇ·WÅø|b_¦Gð(SÓïñlu-,²JJ}GѨ}‘Ü›4”q$ó ºth~OTã¢Ûûžs–»öGÇ’7?(n;+® çn'¾D4G”÷êc‰Láó"Jµ$AlÒ2!”TJ[J±üͤüiü=2#b•h9PùA+lHù$bÿa·¨Œ+asFñú¹CËÖào[ÔêL™ùÑ1Zm20} ¦•r…v¹þÎ!î§n½F4j\l°Ž=ý†ÓÁCx [éW¨‡Êy²´‚*Ÿfù ÒØp#‚ uîQhÇ9ý¸üm!Ä[Éæ\$e†Ò½>/ß>´°½¢ºÕ–4,yÐèÕ1 ¶LÃhëž‚¸Œ¥•¤M_*QÓ%©Úåä'oëß'!åh¹–§1»r9ÿ•`+¢ýY ›péÖ2¸Žh:Â?2]Äô’Ÿ#–žÁ¸êhØÖ¾ád®ý¾oøç½ºîWæ÷I÷Þ5…ßu×y"ÜkËGñV!§aSeZ¿p«’Õ‘ñm•ÃcG4£ì´ÒÓ]†w`¤·4am9êlåóŽ”!¶+f¢ϋͷ¯¤–«ZÆ/Ÿ¶óu´öX3(Nô‘ÍK%YæäÖT ÇÇTâ…À{|R˜9ä Z¤U‰™ÜÖWº•`©Æ‚žÏe”d@Â=%ÐÝéž×ëI5®çƒ“3ãŒ0¯ÉéryÊ`7Ü~Ôò4·‚Ž ™&ðŠ…tÄôÈ7‹…ÙEZ¢hÜ­¥ì)_ÊV œêý ,Ó‡tòaÀ땜UÁ¯o\?V’©È]ö¯§¸ÒþÚlÞòh¡¦O¬«*»ëÚ¿d™Y«2›]cÙ'O¬NÁS°ÀÈN@Ó« &çáL+dDö_°×þJO v?á ^,´bÿZ}ÁçÀ`*4¯ˆà¾9½OÚt´<–k¨HNM¼ÒåTp:8õ*ÕEö‹L=È›žòØ5‚0rSuáfm¶øÊé¤7{±«—÷(wZp6~,ÐÛ‘D0Ù%Ð<ëý0GÉʈþ–«¢2vGÉðî€_Ü€Bç ÔdçÉÓ3ÍV£`™(–ÇJd07æ¬˹ŽiVÿÌÒÍâû4¿§†Ûø1’j+Ok†|jÎì&ºK!{éD•õÔ@Á@–æXFj†õ—ÎÒ'ÝÎ:žÛê"G×¼ü¨ bþÂÕÈHÕ•§µ$£Jã 9瀊·n÷Ò#Y1®$ßÏ…>bV.wé/Ewù;}O£ ‹£^³¼Ç@<×H‰ýy¡Ë^Vžrc/ÀI¥dp×a¾`‘ŽwUøùÎ%”üÅ1Z©mzÂ,Ep&Ë~v~ã3ÌT2ˆRä_”„iñ-å© œƒzY£’Ð!NØÉã“ö%¦k¿L’°©àñ½ô,µIá÷;'œRÂÇÙ\bÛ¤XÀûy÷W[IŸý>|á>sò…L¶ù앚$Ì -}~.ët1 L¯·@YsxÆÝWŠh®Gµª ¸øVfÏùy2‹dÅͺ¡QBÁ±åÚU ÅûLÐãÀ}~A!×X#ÔM«ÛЗGÓ]ö²ñé„¢GåÓÝmR`+›LšP„ùƒà*:WKgáFŠWsX\¯•Dm^"1êQ–÷§ïãw¡É+‹Õ5ÑWÓT¯×¨®èx3ÆìèŒ&¼2(˧•ò~o…Iò6¾²gôšö ®©Ö{pøÌ37™lìu7ìV!^B´±åÍgôÜÛí^O)›þ&´Ø010'€ë±-”…ëÂ]™ºg{+±CEg:wÖA¥“Òb™*ûšÔf9sÍ*{_,Ï?–6Ö²’¡ñò¿¢ðÉAýT"ãÛMáD'’åm”í²Ž":DG:ø|á0ÐÍVþÅÃH1Â$HïTŒá͸ô%MWúIYÿ°è´Ž`w¼¯ß¥yæÕ󙪸1:Ff_µð©o—h@{{µLÚé÷DP !ãL+—9ÐÞÇ·k(Ô¡ë¾[¢Øùæ‚ãÎNÅõÅ#SI¯wZoùŸ“¢VÞ„‰íMËŽö #>uê÷‚ws½‰àõ,þÇÓæ¦)®6Q¬i:2Ñ%Ηç&ÉDáhAO,ÐežfŒ¦ƒ¹ê [$Ÿ ˆdsWÝ“ÒQŽÀk×;1íÈòQ-¿/ˆM´t,ñá@¿Cèu›ÌbVäâ‡: aÙ´7y˜É¶~iðwÝ‹w$+ùgXÌS…¡[ÆW2cTɰ€ÐÚüy¤Ååœà*°u’²õݨpŸßýަT}¬’^É4ÝÐ8ãý íÞS‘“bsYh§VÄRÃC—?õ`^ßÅmóDLcüÒB¾?´Q—ÉhšÚ€çQÌÿÙmœü“¼&Ó$8ç#%5Ýì¦F‰4Ÿª˜‰±ÝN}Ä3Õô÷ÐM¯Ÿ}#òu=‡ëX‚(Ÿfë`’E}GFÏN¶)•š_O'‚F|[Â!J‡þŸ:÷8úç:i-9` GüXTzûÆ´›J„Qsé6¸OcmòŒãœWùã ¸†ö…¯´õN"Ô›³È²>ÕË`‰ð¿14Œ8³¿®S¹”r1¨¨ª*ù*ó†ðZ~; ȹ÷n÷«ÐÀm\&Zc¦,tþÊÊÑÀ¢1bš¦Ìå¼;í¨V¿®îc«G~'~›IEŒÊÂ;@“óó§SjMŒä½ãŠõÇßäQu”€©…~1Ó¯­Ê¦îÕ­?çû½‘'9ûñY&dÎ/oxÉu«Ö§úÄ@ç{*MîϨƒîDËzÍzÍÞC”VË¢LõwJeYÔ;r"†› H¨Ö}Oÿæ“x÷ܱw=‚•âA38Áµ4éG¶¾Ú¦ÈMOB€Ílï0«+DˤBqƒ,V@lM'ü–íÞ—‹^±ò í4dÁK"Íäðˆ8?Ic¼ø ²œÎ.†Þ7_ì€ìudH˜¬=»«›Å{v\n,å³S8íqáëÛvjkëI{,NsÛ,qã:‰9‚Ö¶ÏsÄo3Z‰‰æOH âÜ·¹½—¢×O÷ äE€ã©ÖÑŒê;α%KpXpñ|-T|Íè3â¿JL¿­UŒp!3­{ïÓçæP *$Ð×óÙÞ\¢h‘ÀÖeÆÅ×íUzJ™…ÖµûÅ(•aÊrÿ sÆÝœ%,]ÌÜÁ5†/sT;égmî™|M´¤Æb¤O~%ü±s°pc‚ Ç©ƒlwk”|bý¨zÕôúÞi$7_Ùc†¨°ÌuÄúªFø½ž—%³A0ÿã¥p !ñ‡äÑ¢­üvŒ¥! ¯I«Ì¦_± =ï+q…«³Ñã¿]P9õûª¢HÖA›¿›a-¯ç;¶_còfjöI$Ñ'|í"ï3±TçœKI2ƒ ý{Èg3—ì«ØÊó¶+ñþ&ÜöoWNm’±L)Œy“:À ééDAôÉao„÷I%2äç†ÓÆQrÎ+F<›"6‡La ;„ŒÄÒ[_¢gfÌ!a)§-ºJù›0&Iß²ç 72·¿æY—ôiªG-} Öc5ŸÌ¯¸™1c}2a(Ú HŸºsÚ™Ïí[aúu²mËnÆSm¤&P»XõM¦Š,sõ{e꺟Ë5)a½:så®j^sI\‹l½ß;@ÙÅ[Äëãx„uŒÙŵV—Ò_úÿþdA endstream endobj 185 0 obj << /Type /FontDescriptor /FontName /EDHOWJ+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/A/C/F/I/L/M/R/S/a/c/d/e/f/fi/five/four/g/h/i/l/m/n/o/one/p/period/r/s/seven/six/t/three/two/u/w/y) /FontFile 184 0 R >> endobj 186 0 obj << /Length1 1652 /Length2 9100 /Length3 0 /Length 10181 /Filter /FlateDecode >> stream xÚ·T”í6Œt‡Ò )!ÝÝÝ ’ 0ÄÌÀŒtw*Ý%-"ÝÒ- " „”t#ðá›çœÿ_ëûÖ¬5ó\{_{ï{×½æa¢ÓÖã±…Zƒ¡'·(@NCC…‡ÀÍÍÇÉÍÍ‹ÍĤF8ƒþ–c3‚Üà`(Dô?rn âA&D<5 €êKg€GP”GH”›ÀËÍ-ò7ê& ºƒmœU(Çf’ƒÂ¼ÜÀöˆ‡8?XlX<""BÏÿ0ȸ€ÜÀ6@@ˆp¹0( `÷ÈlzøÁöÝA„ÛKŸÏ*þaóðlÁ6€5È Áþ×ûƒd÷'~è¿Ø`Êý0~<îߟžÌ&Ì qöú—þG‹¹”uµT•ÕØÿJù¥¬,ÔàÃÁÇ àààðpó „üþÛÏ?ø;û?¤Ú@ð_§û*;(àwˆßY<”ïïLÜÿ –¿ö†ðß!4¡ °ü;ÿfÜÜ6_<ÿÏ[ð‡Éÿßðÿöòÿÿ=‘âKgç?ô,þ?z  ØÙë/ÆÃ@¿D<,‡ôaE ÿK5ý¹Ñ [ðK—ÿÕª €K"±t~Nnþ?å`¸"Ød« FØ8ü9Lwã!†3Ò†ÂÁ¿/ž+nîÿÑ=ìžÓÃåèÙŸ* üa´÷7=¬ÚŸCbµý½“¼‚ ›Ð ûa$À‡çaymAžÌ<€‹E<˜röØAݰ7ZPÀ¥ð[ôpÿƒD„\À/€Ë„øW ò`jóøø? ¸,ìþ‚Èþ÷]úPð).ð¿n—óïìÿÑÿ ûò>°aàÿÐ>àÉ‚<ÈwøÂá<þ#ƒÿžÀÿª¡ÍK7·‡"ÿ1üþÿqÉ@ž ìoÓP±PÇÚжËjJŽ1q´ƒôKc^ޱ" LD¿Â¤åj¢^NöœZ¹â·>E ÇNMY×ËÜ¥™Ÿõ:Úz/þsZÅ-{ZëøéûóGSI>TôÓÍHoŒReiDKÝú‘´©c»°¤líûõ˜|ŸÕöÎ&y0—) ©†µivtU—e«“R‰î}×E´Xéoœ’=Ëûîò:d-Ö,*M3“^M¼ã ;Ñ£Špepxh¨º;&HÕ=xã1A¤ 6D%ÏQe¸P€ÿ+¯Ô}7ó±µÒé]Ûþwˬ©—µšÇëÇ»æt,É]BVÉv±<{ÞWÔD°bÙt·üÂÍhÛùÜZ”¦Aɤ ¸aoX¹géýwj'Õˆ05¯¹©Þ‰•_/ÕÒTLÇBYžk$JÜU¬ïù50µÇ¹Û…#·>«N%ZW`'ÍfÝñ³ )`tÎY8g ›2â5NwhÓ—,Y m 4&âãÃs§Wàs5üÄ(69÷ñæ¿+8éèežs­ŠÄ=Žõ^©[ðÍ,zs}ÀÃõ±Þ&£&VÎÛÃeõÜ·ÊsÕ\}ÐZì÷Þ»ÍEàçÊ Ñ«X¨(€6÷íNgthgkýûùµ¾È3š äx[¤2Íüw×§A¤Ÿ mH|ØîeA±×»Þ·»È™MlX½7Z¹6‰˜ád§½à`A?²ä·ç÷G’­iZªü.2Ü£·º|Ó-lÄC>줭ëEˆVÃÀ/7Æ5’×µ¤ ü8žÌ³¼Â &Ö ÙN|Û&³rÚ’Ãú¥†rê'n[™D&§ÈÛÙ»§|ö”'Ùø´¦–´÷Ö÷PK’CY*âåëàŒÆc#zÚŸäÈë9Y=+Àãmè E¸VkhrÓ-ïq‹G¬æ‡÷_”%ÏäýÎ1çsúMRðÛ~Ö×ð8¾¨ ±/ÃGïä±îhÉV-úZu8Ä•Z°v{˜„ؘ2s;&Ö# y]Úc?Iw ×k +`F ð{ »ã˜Ü±›¶o¦-±û)ù½ ·O¶ º'+Ô–q éª:6doßÔ„#0g“È(ë«ÁŒ,Ùp½WËp¾ñׯòŸ[ÄZ'fz´`qFÂø-eÑ×vž =^¸»æ9`Ÿ˜#úòˆ`Ϩ¡–wÍ–¿É~…Òá)9l7£\wmJk®¯pÞ혩;íÏÅ~ƒ³¾NºÈL'\t^÷6^H„Õžæeå(Ó^,Ù6Ò5ŠÙòê|}¾víNÌ`£öšmaZŒÝ{œÃ8¶ê±7žài‘Ηe~I ë€µàOœÊu˘!s˜èó׋‰È»–FñCGµ_¯¾0:5;Ÿ/1Yžº0šfˆ¬qô…äoQꓻ̑]]ØÃÆ9"kÃ%+×û Ë’RAOæÖëõóg‡ ûº™_VE±#h²@-Ø\ÇXQ­[ÔTÝd¤^7?›ñýBó#›sÆ!­ôBÇ®ˆÐ»k¿®Ç›ɬÕá¨zº_ŒS¼ž>φ×ó:¦hY´­†´2uá¾6Èi [ñ‘œŠãʯ¨z+ç]Þàcªð{›jûÃÇ”_2úq4†…‡|É¡4HçÔ¿e¶5™°X‘îl„á‚ùĨjƒdû[d¨A774ƒ§g¸(ôïÕl¶7-±±O%}ü¥(b£zÒüqäÚ«(ðì :c†Á°$°60G@‡÷Ÿ(YN帉҄Ï彡 y5Š ‹¥›ýe"Ì600d±\s7œ×Ü \=ç?¹—8 þä–ñÁÝk¾ÊÁ‚­[Yí[¯Ùé¬I”Ç“ ÝB¾çl }Ñî~œåÅ"%O«óèmõDø “íL¿xÁ·¼~j!Üû–†Õ¤cb1æG¹J»üG¢¸d#îœ÷%=—ŒwÄ¿+›ÄU zº?n⸮ÈÁ˜EŸeMLÇÒàí$¯£k«Äßv,jîËqÂY×$?çÿÿ*ãt¹náÞQ†-ËÊ~2B§¼¦‹¥ä‡ØHPµžm¨Ìzë“´Øe¯’h›YÑ”%»àWØUüU‰ÔYœRôÏô¶VËÓÊ0 ¢I;-L£¥ÏÈkeye/¯¸/✸+™«Îp?N–]5|Å ªáXV71|Qn½Çoœk:ÊýrÔ‘TRaÊ©6°–θ¿K=¾£ñòÍ`‚AZõyÖ@#ÂùZ_™j2yç•Õ2¸ÿ“A]ÈŽ††ñÑŽùšÓªÌ3Ö7)½[‚5+™S(’•¾ðH7„sr‰ ‹ N¨<ÚWôNf,é4˜åRCFqž±ôt]ɧÆò®õð« VVæªccºž·NqJªúeRãõüÜèX¹ŸæYX«×žŽbæÕÀ y*Õšk¸¬ÚHNs…ÎN;Dé Ë7%õ^µK¯ |œ6c§Â¶hjG.¬ñ˜”ï +SÛÑÛJÂ$×òÊF-´N3Žt±ä4¹»ª½ØÊ¬Tn:çñ Ô:ç…údÖê~\’cN÷È.$t< ´î„H~E2PŠ¦Ì®+æ?rXy…ƒn¼$s+ì"ãýíìG6ªm‡¦ýz†?úDbYl¹í<¤™\N©?0Iós^õ«¾e_1€ƾ a—ÌòÅÔ •Õz‡£(òS¤u¯ïï J˜Ø®)ÍfOe7W“æćw6´ca(Žï»¶=zaìà²%pe‘jùTnÓ¯)@»AqÿÈ!|ˤ5Ç1̯۬¡,·ƒq0|E“Á…-¿ºø0Œåõ­áf Tˆ¿©¢ìØ¿õíáôI³©Žp¾ä~UÉRJ: ‡™Î¾ÿ‰ª™êÛ‹æ;#€ùˆXJPϨPž¨¨ç‚ĺüt©ˆ­±¬»Îrfäfòt¢»•Ë Ùþa9ÆñÁÏÉ_}BúÇÏŠ¼%‰>‹w³;¬NUY-`,aë/ —Òp‹§—žJ÷œ ᘻdZá™/bö«Ö‰õÚ>ã=g„PèÎW_Ãí’ë ‹fÍ\mœx 5웘£›Ë¨Ý ;Ö1Ïå*]Oz™ÃoŸìªf L¥£ºµ›Í7ö¼ÿ~£ ‚Z¹'®¶ºÙ>…IíOUVeÛ}•#A¯y;Ýb„v Ÿ:qB/’ɧ%dùf®±N>CD€EH|¾oô~â«VJóâøg/ ªÓ2 o…Ø roÆ»à›ŸKÜ#NzѰ×í2‡È¥npœ°öH+4ÕOIwN1m-£J SÌ[R$!a€Â'+•”V)9kÒ"1téP3iõ$ ~š¹.gÀ»³ê[÷…X8‡ÂÅFXôƒ¾õ÷ÅNõª¦Î_í¼É’ïöÔ<]ùz‹W>Z"J©ßgmÇö0šaŠä.•ØTH®jMæ<àµXq«+Íc¡V–¬mÍi.ÖM‰U]Ìμ õs1ÊÞæ8jÐX•ó†[Á­ýâsc¿Ýö›¥‰-“àm\Èi¹kŠ‘¡Z‰ =ïÒœ÷c–Ì”¬¬i„‹2›Y–އ, øžT¹fÖO¬È¡Ñ¿»wïŒi>s¼#BŠ“JÏ»^‘¡þÒ´)ÀYí0¤Sæ­ k¯•rµ!~¼~‡¤]|qJÂ,œßá@ð¼šå¢àL°OIZ¸uºáÚðYFßžH†få? }ýiem¯XX‰‰( 6Ööqw IãHôKÊ5䶇ÑF/p—R€®<1·•;%úûœt]j£ÿº<Öa1Ü€7Çõ÷'pTê<‹j {¹Ã‚„ú E©«­Ýo ±8Kó¤67ÿ‡÷Z太(.Ä/œÞ“-A¹·BJö¯ZÒ¯>RŠ\ÉJˆ0ža î’1f±5 &œ]9¸œÑ…`Ñ‚*ĶFð'ú9Ëlý/”î¸ÒWÇ°Š–¤yÖ‹ã<ÁÕц宸 £ÍUï’ßNÃ~Øi>òð2_ž6Ë^€óººìßIårΫ5^Ÿ‹ùœé·ð†wåõ¼.Õ,8˜"\AïpÚ]é³]ñìVqP ),6O4wè'vùáµî«ßüÌâ]$¯dÀ+Ž@͸Ûé7ŪNÃþšhóòÖ«Å@7Èi%þáIÃpå»OA᡾mÆÚSù†íý,ÛÞK;.>îí¯z5ly£G™6„Hò5£‰8–b§©%ެœ~j¦ÒÒF4Ų«ƒ rÜËÅŸµPd´òX*Ãáíj˜Xn`còóÊô¼Ìhé°hµÿê(ëµS>:‰— ':ÛÉi˜,»òâbuŽÕ}ÃM×ËßT6®xáÄÔË72õöáJIŸ4õ£mx§~µÎ¸=æ _\ú*÷ÊBN65f«šXbÑ-~Â*õ˜8ÛÙõ= ¯ùbg†&ˆ2Gm+™»ŽzOªtÆ“zq„c—~ŒOLBd¥í¯wO+¡¶x/¥pµ©F™•Å\Ç~UèfÆÈ‹Åh:Ëå9 ãë@Úýnbð¦Â5­ Ù*ß ðöBç}’ðÀ`?4ž Ç¥ºÂS(>wv*µ–†è°ŸŒŠé&ÄBr£x½ÆÃ«HC¿5úÑ|ØRÛžU3Ôax'L±››SíðÎÉ U°yŒ“”×[Ñ+¡r¶Uª:huø1Ù˜U×ó`)q;¥ÏK`‚sΡŒ³U^™|'ï÷²Mƒw ©j1îZxóAP±¨+*éÀƒ—Úbý¢˜¸É}Þç°_þ½¾Çïš–±¦ˆžœês¹uÞ–ÐVÈôYï´ Ö߬;W09{Ÿrgýá?’äµå4ùÕÌ7¥Žt!“)ô;«N ‹y§‰Òzù×¾ª8¬÷§Êf¨”èÄ~Ï“'Ý ¤äÅYGRÌ‹²n‹öS>&V|_ÚVúù['†‹Q‰%C¢u ð>Õ׃•HÅj…à}/ôÀÛ98Usôaô|C2ÒÌ›^Òí?I¿ æBGCîJ…r¿FLPàx⤜œ±’² 5([^ꈟ’ñ¸Ãµ[BôÕÍ;}H³5†_§*¿ª ±u2‚ m¸zéNŠÒ¬zâHœrd»G¢þR'ö«6iÙ÷¶É˜:­ç`#¹N/³T¨g¦®ÙǼ}tç6&ÁUCÐÉÌ£¨ï§`~ÔÆ÷f¹T‰ŠY£Ù+®ÐD®(Ëúj­¹$ŽKƃD­û¢@õ>!«PApÚÙÈ-BÿîRcØ’Â\?2ÃÿYº{õ”޶ãJØçÜ¢Õä€nœt_ŽjuBÔÐÓz}T¢¼}oÿ!""äì{JõkŠp¡©â÷Jì/ºFòdck–£;w$š1Ù˜Gƒô…Þ ûDÄ3¤O® t:§ôÓ’4ÒÖö·7;¨Þ4•tçS°8ÆÝ[o°‘Jø"b)ŸÛL&q?îçvn:¨®ÿ7lõ@ªä™L¾xCËFûZ‘<ä8£¬w¾îŠ(À*2T¼ð®Î¶…P! éf˜¹—8x} ´è2@½…SŠ»~‘ãëÚœž•.R°+H?³ˆ´TܺCÔ6-MV#ô´¬Â©&ù9µ<==½º4n&ϯ±VS"¶Ü箣 š†žz—…ŽŽ€EÑÚØQÓÔêÚ:Më£züx wЯèpt›Û9VOb52‰îkËH¬Ù—5¨ÛJÆFE «adáIßð}5Û×Ó~SHúÌu»dÓ@j‰u}JךiâfÛ˜¾%-ˆ·Ö7èÏšŽ¥•ˆýîí ±‡À'.ú‘U_§Œ29#Ù’ê»%œæ•$šÁ³î\*n3(ƒlB›©ïxã‡n>‹d1ÛØ5ÖdÙÞ¯@¤CÛÖA«¯ §ÀËn+í‹é`V«£·i::ƒÛ,¿0.ø“ÈRê,JC–Bƒ‰5–« ‘Ô˜JlVRŸt5ÊðVzホگ½ÚuÙÍ-#CPî:xl+ŠZ~ªõí>6UÇöjvŪ—íjdì•~F9,¸Mg´»‚=­@Ò8 Òr xùüFE¿›1"înÌ+ë[Ä~Ò#åT¸eóÞÐÆ!ל̙y~ùbï¢hK¡VÑ6#ÃP4&¼hR R+21l-KÃ3évhÞR=íZŠªÇö1ÛµhÄ4÷¯¹'Ë^+? §R[¸ŠÕ]€:>b-~öªÎ\]'œû€í²QígÊ {“´$ˆ“cZþåõÖy%9jê›@çaƒ 3r]vF÷^hª|qÆœ}ø2æWÐÕNþðɼ/$/‰U’'õM5ÍÛB Ë+_÷yZzmóãðŽ–(ÎGJ û˜zªãF#)£åšx2¨/¾O¸…]}_£ûýÙLºWDŸb|«8ÙËqyÌšÊha‚¿Zã ~ Ð}; ƒ*p/A¤e¿óIöf«àÝ -`˜OÓdÃNlžé™©|kŸÏ|q$¨£â\ÆT2)¿Ñ÷¦Žîü1ýÔȵaž–F]€Ì9Ò½Ý"}zŠÓc‘óþuj«ž,=Ì.NËɲ`þùÔ#„È=ôã\< ã½CŠ—ì¼é·ÂIJg7·ÌH=Nþ:¡ù¦J]6I„éî\ʶÍe`©ö+ 2¶ÍŒ¯p0eE‹¯›í˜Jù3á%\÷pdaøðèWmÁ¥ÉQ¡.t”‘}Õ^à\»‰;øós;ÐsFGöërh{|†8/©(»ä銽p“û@1P¸³j†Ï%Ÿ®‡öÃîg5WöO¤OÙd°KäÖyµõÉ' qW¦Mͬc YhÂÒ]„Ù¼³÷±|Xc©Þ‡a$¸÷>÷ç4ÇTa““—yk·œ‚¯Ú-¸zGHÄLó*²ƒû4*sÏ”át–(ã]/&}$.ˆH3V ÜÛ<“Ô#Ч ¦nû+Àñ¢¤‰ ‹iL=å$ŠVÜŸ³Þ½Ê±—'¯zVàÑèþóÄÚsµf"ö²_ânùÖî5|,½/Yy¾ÝÝ(=Gì±} ZO{ü­FÛ‹Ó¡ ‚“D®wM}w½rŸ$ÊÏŸÑVF<…>íÎèá3c¹$ã hIl·ú*ôŽºá9{‰iðŽÔa‡Æñ¡ÚÂ~}^Ï¥ŒÖ‡à92Y:Dzÿ p»‘D°JóÛSI¯ëÊÙ~ñÂsjý-:¶ÑÜÀ6B:!ô3õÝ)\RDÏúSMUn`wòÅ%D_A6Å.‰&׿E§BKAMvªü;“½þóë=áº9åµ{»Kò7…yOÃD‹––ƒ_Ì¿ÊT¡g#mAhÞ<3m*;ÒæZ L4Åæ‘[JL” ”vhFÏs)û%±…C1#Ç=Å}îo­SOŸôözx®ßíµcÙÌ8˜ýeÛ;ë±Òp‚/ºHõ"J£‡km5îQ÷¯ˆ–-ü¥êþ>/'«pMa·†ÚÕJδsÙ¡G}¯°­Ð%k¢X1Ií—Ü/ètÉŠ!£byà°Ò×â¿ýj+=ýã`¤t_¶e‰U¨÷B”¸ÄžÿíV\o0SÚö3v:öÇK‘ýQF.I_ë¤ÄŒ²®ôû²¹JÂHÊ*¿›‘]¿ðåðfÈÔŸŽ¨XÜa›ö2œe'•e#þ1çø`vVâÊu¤—~=»z£ûÆÃààEô¬µ8ï8ú«này5ï0‡¹óåÒH˜“AU*ôªjµ‹W‡SÙZsTb›vÖRïÑö7Öw1ʧb²ÛNü¯°ôlTE=uíÐ4Ôü‹N—wÂÕÕ {¥ÒX#¿PSŸû ã„S$í—Qo²¬Ä{VlGÒª!ŽãJ²òršÂ᧸9»†¥ã«›?æå— ƒÆNüý'4 Q‚)Ù^œBJãêµæk. 5'&ßhòyèz0“ÏÔrÖ¶Y<«Èé­å›ªÙ,ýOþLJ7bÎÖO½~ST€Fšù¡mÞí£áNÑhAѲZÍíº7R*÷{·€Cñ¦5›Áùø€,Y>RÊz]è/p×bBKíÖ¸yˆ<ò¸Œ~Ds É¿å‘À#%›’(Ÿ­äÃ3\‹%(‚X³¸ugý(G‹„ÄŒQ£Ë£’›R«fTÎâhNh£ƒ}ò±Œ_çÞs,[7}ÙÑÆÞዪ‘epÄx„}èÕzNyÏ(ù”õ5%ê•• n®ÁpÏ«ÓáÒíÚ>ŒZ‘î([$aë…ët“w ŽeTwMª®#Gê·€—`®H­|²£›!{©õ²¬í=>£·Ä fÆõªí"—h2Yî)P8 9ôþ9µçU<¡]d,¤÷ëçÆ€Õ—?±Ÿx­9—œÀi¾ëX—ÃÙò¨{à/Áñ.ˆÀèJ™C´-q²-A9ãCh¡Zͨð@Òú–{–rÔ\9væw9grj\r&1”2›ëöœ±yË…-Ý]©F‡ äu‚sNÄ»;ËG1ݽx¡CÄûl×g±Lê¢ß*ÞdG¿î9!žñë22ަ±t‹ªÑcºGÀpJ^%ÎçŽçU$mɼG8Éà|©v›'RT©Ä8·³â…ãíE¨ÄâhÒ•ó~’ü0òQà×órË!ËL‰÷ ºM-µÊI“dŽ, ¤h:üÖÁÑ Èœàê™uG7²’f ¡cá³ÉtÃåÙŠY¤ ·#Ñ8‚E«' §uG|>qty§MßÝøß^$l2Ñ‘JT,›ØÏÕ™=; nýÔÚž‰8C>× =±'ÍŽ“ŽDääî £æÀÝ2ë¾1°µ.XµŽˆ¾Vt+¬æ•"„+Ì1.ÑC’ Ðó¡ Ž¹V=a½Šîﺞx:ë4í `¶1Î:ã[²|N–r¥ÔÛ0¤Lê™Vß–0îžIM«J°D#/b9(|ËÍ©,+Ÿ»ó#B/ ŒÍèøúã6ÿ ϖ†}¯U~«ðøÓy'oHêÆ'ôg웃ì@¡£ï38Õø®K¨™Êãqn H²y?·<Òyšt‹¦øm-Ú•À ÿZŸœ7¦ø½Öp™=ý×Ù2D\h³d¿ð3ñÙ vg|£ÅàO–åfZs#vÅò~:Žgøé,¯†Ü½÷ {Ã?p‚#.1ÇZQÈÏLkC™s®\ä$«æm oÅÒº–ZŸ,ïD7%l4FbÉH.:¼cÑŠUŠ{×Ð܃ßÍÖoÉzõBÅ›ŒçS±ÆHõ³— çª-QÀøLf€hïæ&‡è.Ó¬X¯ÀÀ3§Ý·ÕÃ/J“r§Gžêš!²5<>Ý?RÕ23,x;”j{ª‹ß*úÊ¡A­ëqOÕuq‘+ÂpiÞ8{•=žô„aÅ-¿öCóíx[ÓÍ‚°Ž¾40qÇâõÙ îüð¼ò§ŠCî ={Ã3ë®C»ü'%lmiŠ×N5…¦œ³¾+u4 éwgy M­â}*²g²§$÷I§­L–‰Iô#@I÷1&ùáëúÔèŒÇû¹9øn¸+ká¾Ô\$'ѩνí"âmçcëúëÍw¨:kåïr™Ý}ä1úçñ‹ÎÓŽÚ¼ãHw»˜,ûЧfctíÖ½—¤èUôЪ9ævûìBôYµž¨ªu`ó‹rnµÉò¨°ÃlžÒ1Íøq¨YrZzƒÁ,ãÙÈ%kaÖ;%P}»kIÛ×kÞÆøÏfÕ³ºiS%W(“ü}ŽZYžYáJ䩸šf}.ùõce¯šÎ/Û1&£ù#¡qNž\N\_Ë&ˆÌ4L£BïMBB—ÿjolí¸Vü»þ2è—e–< ÑÒw$í_Ó±GôË/æ½lZý­ž{]ŒI§+ÄÕ– –0:mnÚMŒó2Ž˜È£Þ? ,H¥Ï:IïÜz OoÏÔÙ$+T0J­?«—‘DI§óOƒhXXÙšiçQ:>rQÙ3.ƒ6ÌM»t—@³¶ê~l~õIº•ž˜åh#Ë!£ÏÍ4¼o ÐÂflácOÖ KøNÏõ¹4ySbF#O{ÚçmÆð2ŒOë|¦~£×ôÖk3rëÚÀPYð}ûÓÁåë`šª0si}Ì§ÒÆï²È ž~DÒ˜ýp÷¢ZCÝ‹ÚæW]ªO±¿÷óº–úÉ&Òƒ£®È§2(¿ÿž#G9ÔÞx»‘áøx\¼@(êÆoÐøÙó z¢¥_õ™ Á¼i„¦_(ïGΈ›Ø»*vct÷ߌ>eî³RáŒoVAм®•—B/7ñ ¦_GámEî÷µ’S…œ¯,ÁðtMØyv„’”"Y:u“^~¾z†wÙðfMUJ1WpÏê=}¹çkBãÁÅ4Ýð…/5‡pƒ&ØÇU š“e÷û=j 2S–0ân³Î‚ˆœñ²{ä‡WÝÞ ÇÓ!äæâß?¢Œ£«.Œµ[oLUN•Òq€&roÇê[ËSûË eë+¼Q ð_O}"gqy!ÈBwû±U ÌÖ=QR¬ìŇ³¶¯WABÀ5’¯²ÍÚHJN–0º×¦o¥sÜÕ)Plɟʘ¸±”9“ÿŒƒfõÝÈ6V†…-ä—]ÇuØêbŽaŒÐ|”œìŒno¨æzÖ@NÑ :Ø—§$ò}™—Q¼{tã ͯeW:†¨Y‚OO ˜xvI¿EÕçôÜ ý€à‡K?ž=ÞQÍœW}˜¸ú¹ )Ó®•¨ÂÓ>_I'ÂhY¢\\ H‡¿ä¢Ò‰ÜÛ·ÕÙÚø?ÏßÂþ endstream endobj 187 0 obj << /Type /FontDescriptor /FontName /HROJHK+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/E/X/a/beta/c/comma/f/greater/i/less/p/pi/s/slash/w/x) /FontFile 186 0 R >> endobj 188 0 obj << /Length1 1436 /Length2 6534 /Length3 0 /Length 7510 /Filter /FlateDecode >> stream xÚt4ÜkׯND‰èAŒ2¢›Ñ[t¢÷‚Á`”et¢÷Þ{D'¢“è¢E'D'Z"ºo’œóžï¼÷®uïšµþóì¾ÏþíÈ ©Ã%m…°€* àH.07H «¦¦$ø¸A ^| P†t€þ¥ÆêC]\a¸èÿruB(‰òSCÀÊn0,( ¼ ÈߎQ€ÄfPã(#àPW| ,ÂÉËfc‹D•ùû`µd€ED„8‡¤¡.0K AÚBQ-!„% ŠôúW Vq[$ÒI”‡ÇÃÃâèÊp±‘`ãxÀ¶m¨+ÔÅjø q„þAÆèÚÂ\ÿèuÖHˆ €R8À,¡pWT„Ü ê@è(©4œ ð?Ϊ8Ý Ì þOº¿¢%‚ÁC,-ŽN¸ n°†9@ ªÜHO$'·úåqpE â!î˜Äåð»s@AZ Aü ž«¥ Ì éÊí sø‘çWÔ-ËídŽŽP8ÒÿWr0¨%êÚ½xþLÖŽð€ûü%XÃàVÖ¿@X¹9ñèÁaÎnP%¹¿\P*üt6P$@$,Ä'Ì €: ž–¶<¿Òëz9AÁ¿Ô(~>N'€5 Ôf Eýáû¸BÜ¡¤‹ÔÏçþ-áƒÁ+˜%`µÁñÿÉŽRC­ÿȨá»À<Æ ÷ÀЯßNÏPô²BÀ¼þqÿ=_]%EEŽ?ˆÿc“‘Ax|¸À.^ „P¿gùþ¿±ÿÖjB`õú'¡Üƒþ`@]Þß8Üÿ¢ë_+Ãøw uŠËPë?Ô7 €,Qðÿ÷üù¿ñþW–ÿõÿ»!7‡ßfÖßöÿà q„9xý倢²µjÔrÀÿÛÕúg•Õ V07Çÿ¶*!!¨õ†Û (ÎæçñÿÑÃ\`žP+MÒÒö‘þžª† ÕD¸Â~½8¨(è¿l¨­³´G½*®¨‰ý1A\Q+ˆü=Ü_2µdÿîCn‰°úµ¼‚ˆ‹ Ä E”$ð£ÖÖ êù›ïn8‰  0û¬.ø¿Æ,"àüRý–À ìŒ]‹ÿ*jéæâ‚êê7WPý-ÿ~ PO¨%þÜ4ÂR,Ä®.¤í¬FšÆƒkcLû ãÌ—k¬ÈÙ'?a¶’¤“›3¯òJa®¬`jש.ã|öâË̱Ïz=}ƒÿ)½ÂWz‹„éÛSôÉdŸŸ´ŒÓÄMhùi2EK]úÐ4é¢ï½»#ieÓ§$ö}T×3›ìÁBV¦ ¤Ú¦Þñ®¦,G•‚VDoUÙl1 »!pBõ(oÕ1>x-Æ$2]=‹QE¼ã?É£úÞrÿÐÀ"I/]Wt ²{ÐÆ!s’D^l€VŽ·Z±€è*¯Ô}7ë¾ÅRE6㠶䲄ì>ƒ],ÿðê!ÇÍØM·B±î»ÝWW9ÒŸCè‰r̵FÞo|«¯=\T>¡¿¢Â Äi‹¯â;”zñp¯è»bˆ¶Ú¡·‚ ·JûXç=ɺ3Cï­ÇÇðzf¨c¤™®½xDpñ€eßÁ*ÅøÕóè›äÏ5çÅQé6»Oot­(%üØ©"e®ì?• %>Ê[†ŽpEH¨©Í¶ ã˜Å’5óˆˆhbçj¿¥¿ëpd]XËubõ,ˆèÎëë¥Jy½>šë^³”Ï­?Óçî²_Øý$«ÑUê++^V‚‘⛯å>Ûåï\'gËŠòDê¿üO×z”ì.2¹;^Þù¨òþ¦ó ]ò;“ª?d¡5¸°#òƒv1Î÷Eªâ½šfŠÉÇÄQ«ÃFÏ_€D`á¾}å_» …è5µ‹D}ê%Ý#_êŽLzœÖI¤`qD´òš?/[½ßGéÍ oL1ôMÛ„€ŸÒ(¦Ji1š£–Ì—p À –‰ÖÒfA¼n ñ·zýÊ$3¸qÉG®ÉÕ€^X@øÅ²àÁuëvõ‚ïm¬EA…Ã"x›˜WlÞŒ®WOEª„X~já¾!À‚q³9îN{vÅc–Ì‹ásIË7P¾?üùmd ÷£ó$„pø(að~ç­ ›øÑÊý(ßW|ßå°DùWõ/kz<˜œs5ó\-ßvoo熗æåË*lh5|¿yóº|¦Óš^ÿ$Ócya“1ñøEÒ-µÔšÉý8SÀí­<ÿ1ÚªMÞ¬yô%éOó§ P¥44!=t[Úa=sõ¥—åG â}f×ߟ|/ ÜçI‹ ½OŽ•B–ªry½+Üë5´3Ë8ÿ^/ÿæGvê¼he{ˆ1™$ÚIŽï £˜ø’Ïü´ $qï€a›ƒŸ)jà>Ò0¸ÈxûºŽ˜Í.#B«m;öª|Ò£1+ŸÚ-V;« I&¨X§†ëøä©á­ˆ–;†•óÍ–¬q9ÿ@Î ƒgŒÉà¨Kõ«(¥k ×·´Kh—DõÍÙ,çoÚZrŒG“/1UKÏ0ŒFß1t«rwɯ@Í d_ ¸X+1Î3 89„1µ­w…3\õP„ï^ŸÌôU賓>È1åzªhû,áü:B­Çjœ¥ÆÃ¼Þø´ÞÌ3óFAã§³'ýä lFØÛü™xX[©Î2eÿ¡«Ý8·Š¶QÌÎ+r&·°o©ÍÝk)%épí±©é”ˆ^UèÏÝΩ­¼¼ü{ô4í¼êÝw›)n>«ÁF¼.;#N :節K 0~L²•V— Xð‡Šñ¶ø1ÑRRu“àcÊ’;S›~2G›ÂÖoð4ãZ ÂZJw2é›/¾»ýph-80eß„EüI¾6Ö`e~£´<æ¹w?‡gaöÚE’HÖXÊ,¼^y™ñ£å“ê_Ý#ŽªmEÉÏ噢¡âËý [Y18š)Èm÷3T†õêú0_ÂÁæËÍœGMÉADè¤o˜€¬ [ALzPqàÅ÷ˆÑÁLw¿²~^™ "Œ¾²‰‘cç×À£çkâ••ž£üæCÒˆž±|‡êÎTíYuœ™\Jê¯=Š×]@®’ä°ôÖSýÜ ¢èò|v+,x92gªreó•œ](x™3»}×å†j¡ì1œ°¶2þ›¤\­¥ /µš9NY'³KŽ&T\êšÊ.|V›‰ áƒáÍ{¬oDðxQ[?;Äß‘8Cw6Ý0OúxQ×F²ùüQ}®¦ù1ß V×ö÷IÖÍupÅôýóîmïIm;i3·*ˆ0R}¨Æ9ÿû#QiPó=ô/©Ád‰:94È v¤+;£Ö½Hè|Àû }#¸i=ÁlQ¦ *~ÊÜg¤è†ø½‰—æÁ)kr|ÌöÔœ Ãò®ìF%È´0V~uÔy5-mb˜“æg-—FÙnØëÌYûgÎ2ÏéYŠƒÉ"ª®_¤˜¼ÕÔ50ñ{Ô"@ô~ç ^6Ï;7*~ÀÚü™ü6­«}AÑWÄnƒCÈÙ¬$™gÈÇôUùÆ”Æ{•Æè=«I/N›”%X–#e‚ŽîPéXfÍDP?òó Üp”ÓKk³ÜVñ«±Ù¿?œÅcÓ<Årñ8ø"w±›2n,Û/)ý“Wt Ìlß-ÙÖ¢ÅÚOïl[L¤{>(ÛXë8Y†À°XìÚ2Ó÷âjÜ"ö",C³|§õdU¼³†ýÂ.W£þÛÏ·Àìò”i˜…á†ö!‰ ýV8æ\®k ‰æÛX&ÑÞßlï·3¬*4L}Ì´®–iš¶ž í],eeC«¯ÎdÔŠ]Z|”¢êsæH2Õ|+(ýaÝþU¯ß2Vˆ¦èV,l»àyEve?£¹öü¤Y^#’h°¯å Ú mN Þ®Ú’0½{6çÙ:«â¦ŠD{úË´¿Ùx¦wŸÖˆýÆ–3Τ›>RÐaºE±ó“°ÎÄA%>/£2)Äôl}·àÕ}OõTi}9M;d¸V£tM<‰îœð$ä0"k ð• •üts膙e¶vܽl^wõûï¾dëÑ?jZ÷òÝ’G+<%㣹¢«5R@¼4ï~æäŠE™ l;Üð^#ìD/úôz0co9å)Æ;Ÿº–]9.'Añ¸-¡¡Ÿý2?¨ÐÞh±è­ Pàð4O‰y¥^÷~„>ðVPÍð¦œlŨœÛT©Ïç¶S^±©È±JÚU¨N™v:*ÎO¦Í:EI#ˆž˜Ö²a¯êÜhzÊTÞ mËV»#þ Êdš2šzÓ¤%i>MǽåôH£’®µµú°Yè_‹{ý€_$–¹}*–\OndäeËÆ:´%vd,–V–Ñ\óñj’dË‹S~ùè|yJÿ²ïÀ²$ÑèϾcÜ‹u ®E‹ò! Ï<[7vz‡†*Ÿb’3Âoó¨N|ùÂx|Ù‹®?o¾œ#š5 ÛÓ¤òc=p»÷zñãL;²2ò%2;9¡¤ñ2„G-0¹rúþ°Gëüs|è ‡­ý󂬂ãý’s–fcvøÏÚõ§ls±^™õ›¾Š¡5÷•,¹•éÔ»ÄÁeÉoAÇÙX<9¼¹3'¢Øë¬Çûýœk œnYÐfu(÷ænµìÆ;í3w•Mù ˜þ†‹¦­óæ†Z1çT⤇Äx›È„+  wòqú¾ÝóÚî=#ç¶ý­ø8.Ú´†¥CƒQ2që©-˜FÛȾD±¡Jœê/cŸ„H5sd.èçÃJ±5œL€y‚=¾¼¡BM:-•ŒC6‰¯\àÌ…ø+·†•yšê=ÆÇWèÂÂÜ|”×­-¦³$Ü»ü~RšçPo¦Èz'=ÿ¹Šó jòìõ_7¦ó ©^Ëó˲LØÆ7?ÓhØAÿØ–ÇîŒëÖý\àÄÞ:Œùøx´»¬"ë¾l.…iÁªáåæ²ÂªMÝQ“ÑйC#tÚÇcÉ©/ÌçÇáØ®ùx«_kv?_#½BW]ë$§åq2ãX…d^ácÈ¡ÔbXƒÕû!yZ¿ìXÍH-vmvæ™IË×W]ós„$ÝJá³\ ½ÙûÇšG±‘Œ]â3y©ÊÍ^©[ï¹MÕ2ðÏͪO&¥â¿pÀ,Šê8·F[Û{!U«ö,\ k‰óA¡fEŠäGq¨2ËÞ÷ÉU£¿ø^ÄO¹ÿ¦\}”âªÅÞRÃù,†1*H˜¦”…Mi~'I*éõ˘b|GÖÇeÆüüÎoýŸ>tÚišñ$ΕîJÎ3º%7i/“ŠiTMÍnæE¥E2Z²ªÇÝúYé³ÀNiVÙd…œÜ[þÑeÃúIëÓPKn±¢®ßš¢îÒBÎy¢«€[þe/oî°»1R©Øž fp®_•ÐD’ ö})Ö’°³>Ñ&)ýñ@·ë›êÖ~_éYæÑ*å…C›xZH8ƒÔ»à¢hŽ3Ÿ˜KàB·Ú'ÄÛå]ê–U+Ë€-z€Ç([|ÅFzñ[åà‚p«™¾PòkŠ!(ûÓÉÑOÍC§J .Åd³˜@Ïïà 7dÂA‡œLÙ&ߟTÌ‚ðÃ73qmUõ°ñÎ=éÅÞn"4ƒå‚ ŽŠ›-x?ˆ°|뺞šuÃÆ˜û±¤VLŒ|ã”4Ù—ïæ•f·]@ÃæO°Ï·+¥’Ó8'&ÚdbK ˜æ‚–‘j‰©S´ªè5ΖŒçDƒ”'ÿ;{›¦ïW«¹™Uc§^«6³º'gÌgÆ•i÷ “}äÀÛHòCI•*÷k@í°ÏMÁL˄ݗœÌ›mòpÆA~ ì®ƱMß+MªfØ´Ø}Ý|5F3q nEu=…ŒÙäð£ñˆÞÞˆÔEz¦*Cóý5­á½JçC²üºƒl“£¦»½W‡¯´äžQe÷<—~ΚÂxótâZ/—\ +K…}B@N÷æ–žáöÀ…!ÌAÓîP¼ÚßJËÚSv­-ÕPôîqð“C@ê`«ópÓÜÍÇ=xþ½d|šŽ¦NÈwq«Î^Óñ=gf\jâ–&[j㾸Œ ¡T_¼ 4Ц8 QÁ×¹€•â]/‚c«þ}RŒ±ïAàý9Ìg"~…—áØGþwÃxO)&+t¥I+ ‹ÛƆΠæaÁ½%UÓpµ$žG¼ÓQÛ^í„|;;î£íc´Ï´‘UŒÇÿ4{ÉX6žü2þá¸SÎN®E²L€š]»aà;ÿ¾‰ÊùšÓ]¸Û -¤+G\HLSfIÐ`„=%¼†È%dÌÖ6NÆIyÏbe;Oò~˜=¯IëœLñÑÿißÅkÎu:ô&šF€À@–áóþôÌ/¡@‰L¹‡öUÅ eËe5g5P`¸ÇøU*Hõ›òXm‚ã6¡A¦Oªðy[v5³ž$©÷áŒíŒa¡Ï§z´Ägš\ûöÍ-¹’$/r^è|Ûwb íàSÄÁ½«£Nج¥å–WsnqWÛÍ5öO3+àLm°œ ëlD¬ïG'¿¶Í“¯¼rÊ|‹n5òœœIÌ„%g‘$wVjn˜]¥`Çsļ§¾ÛÔ”ÜÕ5.Ùyœ´9ÅBƒËÌYH/ÚZÎw)‚á{¸6ügÒ#l±Œ{¥æIø+LJv'ÅW hñª,FŒ ³a±ŠwhêI›ž’pgJ@²Eúœ´ïSÑ}“½{¾¢zþ%@'û“‰ž2í…ÇÑÑO¢pLS-Å‚YœÂûÖÑ™2ÁÎt©-¶’,âž.Ø´ý£è|/ŸR‡f\Xi Ï~ôJxhès8©]“K’ÁIMì½ix²”}¯(ÄVlmÉølB=î¦Ø¬Û j½û’¤¾s¶Ûvcä¡}¤=·–0eíòµ¤o! ˆÕíÅì–%\”,Ûœú‡Ò'ì»x‘7{­>r•Ãz)O‚× Ó·[_NújŒ—^ÆÆN¿û&”„fç lœ±É9–;?¸F·½Ž“ðu¥ñ”öÇ2d,à|q•iCMƒ†#&G?ežhƈ w!? ó^c‰·Â_i›¼ËkWM ‘Û² b¬ ?ä3×m ]õðZ(Ñ›yöó<Þ‰½“6ïÎÙñ¬)¢­p¤?à¢Íô2ÕFL—çaÁ·µÈëý€ç7;ß~¦æ›CsêÞ=$¶ÁÄÑû2A]å½+`ð>ÇðÃ-»ÙjËQ0ZÝÙóÖEÒä£NcLF`¡ä²QbvOW0m3z· ~ÇÒšöå´76ÃÌâá4@ЦUÍ…ôæÓlY°;ªkÕVÖ팇‡´µ±ù&–JÝ‹!MkX×ù%,„[TÔ]E–Á°M55òå|Ëx2ÖµÛ¡ô¼éP="‹,cGñ‹åGø'Khn´N6:øck讪—‰ÜôÂõ]`GxÏ*B`qéN”r¯! ¤tÙ3»þg qÍwvL¢{W¸Ÿæ|lGöÐòi[0¨]×°,W•A9ÅcµGŽ_¯SM•[˜ìÈרFÎâe`g€tò»}‹Æp×;\}ËO'q¦äU8I¤;ªnV³Õ´³µª‰~BxdCJsž6AmŽ tš~˜k ´dçÿ(y'¶‚׊!žÕ'»ÿðÉ\Ikb}«ÿ¿öŒz endstream endobj 189 0 obj << /Type /FontDescriptor /FontName /WTIHHF+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 81 /XHeight 431 /CharSet (/a/i/s) /FontFile 188 0 R >> endobj 190 0 obj << /Length1 2771 /Length2 23484 /Length3 0 /Length 25027 /Filter /FlateDecode >> stream xÚŒ÷P\ ׊âÜ]—àîîîׯ‚»» BpN€ÜÝÝÝ=¸;ÜÎÌ|“Ìÿ^Õ½EôÚ¶¶žÓP’*«1ˆ˜Ù›%íí\X™yb ª,Ìff6FffVxJJu+à?bxJM “³•½ïbN@cLÜØd§`ouµ°°X8yY¸x™™¬ÌÌ<ÿ3´w∻Y™²öv@gxJ1{O'+ KÍÿ>hLi,<<\ô¹DlNV¦ÆvcK -ˆÑÔØ fojtñüO~K^&&wwwFc[gF{' AZz€»•‹%@è trš~ P4¶þ]#<%@ÝÒÊùo¹š½¹‹»±ØX™íœA®vf@'ˆ &#PrÚým,ÿ·=àŸÞXYþ ÷÷¯@Vv9›šÚÛ:ÛyZÙYÌ­l€%IyFz€±Ù/Ccg{¿±›±•± Èà¯Ì’"*cPÿ”çlêdåàâÌèleó«D¦_a@]–°3³·µÚ¹8ÃÿÊOÜÊ h j»'Óß“}ogïnçý0·²33ÿU„™«“†•£+PFüþ·Ìèà`ffæâa@SK¦_áÕ=€)Y~‰Aøz;Ø;ÌAE}­Ì ?ðÞÎÆn@€‹“+Ð×ûOÅ< ÀÌÊÔ`´°²ƒÿ$šÿAÃw²òè2ƒvÀüëçßOú õ2³·³ñümþ×|™Dä5ÄTßþ]ñ¿:QQ{€7;3€•ƒÀòkɸ@|ÿæßü¯ø¿¤ÊÆVÿ$÷GD;s{Ïß5€š÷¿:ÜþY šN†ð_E{Ð.4¿W_™ƒÙô‹åÿóüåòÿoïEù[ýÿ›¤«Í_jš¿ôÿ?jc[+Ï @«ìê: {ÐqØý_S-àß§¬4³rµý¿ZcÐyˆØYØüÛF+gI+ ™²•‹©åß;ô¿)€ÂÛXÙ•í­~=l  ýèàL߃(ΠYý¥‚îé¿”v¦öf¿•ƒ`ìädì =q¼Y@jôøkµLŒvö. ¨<_€¹½ü¯‰rr€¶é—èoÄ `ý¸Lb¿7€Iü7â0Iü‹¸˜L’¿ €Iê7b0IÿFl&™ßˆÀ$ûr‘û@¹ÈÿF \~#P.Š¿(¥7(åßÄ®úØÕ~#»úob×ø@ìš¿ˆ]ë7ñiÿ‹x@:ã;ˆÏØ´?VÎÒ5ù@í1q26}½†Ì]~ËÙþ•ÿ}4ÿ*@|¦ÿ"P0S{ÐÂüËÉþKbkû;‹_›ÄdöQGÕü +(ÐÖÌØÙò(оý)ãüÇÑt¥ÿ“°ý2ò0µ1¶ýƒ Ôóßäcþdÿ­þ á߃ítû,Ë/Áo>Ž_æö®N° ,þ€ ø¿f ÕÒÓÁh÷‡Hö?3¨Ö@Ð8ßÿAݶù‚FñG¥ §/ÓïÈ W;ÐÕþ¡Õnÿ;³ýÔ b~«AÁ@¯h»ÿl;Ë?ÒÿîˆËèzÿaÊù—ÌÊþ÷äÙAÍq°quþƒ$qü=Ð_Èèü×èß}cý%´wš™Øü'6öߊÿ³¢œÿhþkÏóô¿Æ,¿và ²€ò;C“3ÐÖê¿;ÎñËèöÇ9@AœAoÎKuÄÙæ?ÌÊê7-èåÃäbéüã@wq·ÿÃÃõ¾Û”™û‹ òöø‚Â{þAMôú(’Ðéoªÿ<ÀM]@Ãuùë º¶ÿ῾F@Sø…Y{S¾ë!ÍwßEÜvF¦(w´Rh¼œZ\a?ÑV¤­9݈|èD]Þ’ ¹^$yö>l¨† oLTizôy2ŒWØi‚ŸÇîË;©ê!zCÈ .¼ëóìè£ø²¼M–2ÛÑ•Y9ãν[Ê£ª§di8lvGe·‚Sá©d’!F#Z/°pš2Ç$c— Æ…ŽýÌeúúf =kì•D6þ-¼ïQ [¾·Î:kìýŒ×Ê7uVçv< <\"Èkôá *oÑýdYœ9ï¢üåÞyþ|’,$úÏË ¨Œû¬iVªQvu]•nà ­,ÛÙŸ!Ô½Û‰•Åu˜†NdJXM†Q.•lïDû­æ¢+Í7šæÖŸ y&«^ÓHÚ–‡õÞÕŽ+ƒ½ ·aŸïê·:|µ …z„-ÜYè$,u#——xÓˆ>C—¸Œ@´ke0QéÝ>q?A‡ò{‹>à³læ13ëpCwžpdûœÇú–Õœ)¶®O£ÍÙ¸£ë¿s*¾š´}>;Ëód“0J£ÉI“èø³G­Â÷X3\Ât¶^†Í(^ìI"Ÿ°É1ÞG‘Ï.#Õµ3W>©äi}Ïó•ñD³L£é#ÇWÒí£(†‘ýšÀ~ÑàÐnž³ýÍ›ÒRQ¬{Íû³2÷ÑØ u®œ­ªYíÎJ_¤D¢ n·fºÞ˜«‹º‘ÚÖÅ¡h© ©-Ǹ,‡ûs€¶ƒGjPÎÛ8BLGT*³ž˜Ès‰9RŽ:¬‹U‰yJós9§Œ2…ÍãÁ÷z†DœuÝvÜÃiàÈ@Ô]>žÜð+¥E’I…Wø­ì0ÞzW{?…o¬Ý¸^«® _²Ã§9éR›è z:‘J'^yÚF‹Ö„º¿Ê%a›ÒâÝnèYd¹”Q oÅŽu_=ÚÁ¨º·$î6Sd›)ê‰Õ§å.*ÝÌŽ çk±zìs;’Ê›M–oØúm³h2E÷‹ W^=“2'thçØ{ÝSßßöPv—µM `¹o¬n°êÛ—t­Ê Ÿ…d,‡P\÷sDmO$apxkXáæ“ÍàŠû=9oƒÚWD’HE!¸òvjîTÅ ‡Sß+2¥¼îŠx~â ôS'lÓ ?7ËTÏzo+¡SX åÝLgl± ­j Ú©úÉŽ~Úv ×iÍ„•áæ•Ù.›P2/¤@Ó=¥Âvç$®%2™_weØ´P*K°æðkŠÐª‰Ê3íò—`Ÿp`”ˆeŠ|]»Í°á+÷jG !oœm' “@–Âí­:oŸ€ÖÑ­ŽC³m¨ua"òâ/çc¯€`ˆ1;õY mí…0¾És[aÿ•3 4»Ò åÖØ%“NX‰å0‡)Lòœ$©–Í“˜üË–)Š‚ÂܸTÎóŠ.¤í·×F¯±hI˜7è6%÷¥d²‚å@¨yMÑ8-QBß=MÚPËjýKVxô9§«…§ªèØõØù-M’1}%-ÊŒû-òPwP¸†º£ï4#Ðpú.¯T²+Ô•_¯Ý‚³ïáƒô /‹øÁNþ–‰õ‰®ØçC¢ÞÚxãÃÝUng? ›ãp²áK¯)»MxáiŒ`~ÏÁìÕ)e!¦+Ùî‹A¯ˆSxÛgâÆç½R2ü&°L… {_S“o>§y$b  ½\‰Q)sÝŠ¾™›»t–8ø¨+,Ô"K@O`]˜LrýANòN=Çä1%¾U30Ö†ÆÆ¢¼#cÑd²,f…´~¿d&yfHoãùÞ˜]…çû° ”gÓQ3¿¾p8àlðþ8×Ù,6ÍÙ+¾¶ìë^ï,¡|¦ž~,ÚY‰Bd~nI°›¢xk”‚…Œ¼Õ U¢È¼Á† §'x¸ì›8I¥š ~¹7;‡Ê#yJ±""¸ƒna‰þ«“6¹T{5wÚ;sÕÔ £`Z‹ã;ÁTí¹Ìr€<¬T&«7V “®oL2ñ;Åà¹ûº½D¤óg²ØÀá=/t³=6W\XD\ËJ2DÆuÙ¿ªÍ³š¸'½5¸ç(ñq- Ҭ ô}!%öÛ.å{Ëå —ç½)o°Â!Q>²¼Ÿˆ…N˜e¸Ü,Us»qv@C,‚hS‡1qÂb´ÌíÊxéúödkìáRïŒÞßk z‚Þ¾©b-Û²Ÿ9Å9m¶]‹EÒcìdzÿóÀâÅ6>¦ÆCÓóºfm"ƒLŸÜË÷s¡ƒÛERQXü—›ùž"F'QžXvõÚbeb,Ý.%fÂä·EÜ`ãïç£ä¾ƒgOñ}mÿ9h(ù@ƒ0­xYη¾‘oöò­óñ<¡ õi®9’¿· çÓVFF ¤s{T’˜„Q¨æA®åzË@7%–Jâ¨d¾½rK÷æZ°?EÉDz 6¶æ[ež9ž#gžppj£­2b¤#§E­{^Ťü£~ç©ë¾¤:aó`Ät"­•P*Vvwôb/„]ŠiV­SŒî+¡½GJ<&¯_wÈ” ²0G´ºè$ÑõW˜ ŒÙ7Pܘ~U0 Î S£T¯Rl”㊫€{«úeöH×)?í©_ûºãXÛèÛ\9Oˆ˜u;¯®üp?* ®!‘{tù}6%Ü>ƒ"Mý£F\GExEfaÝ…òëökrz»n-W7UëS¶å¨]b%Þ6õô’qg­z×9õ;'ÿéµ/Lv†vŸ eås°àh’…áìe§Â—Õ(yËIÖ¬-¼O3Yí´ûé‡}l$_ñUnÉ8O#¼¡Çû)ÝmóT~m¹q7—=ñŠ%…ÂÃÉVäõÄùfgïå…Î$Ý—&âÛAçë3û4LaïxÏY8R11’(oS8F-Ú£Sâ]f;„u‡P gå3 † ½›ŠÑé4*y~‡6(ôz DŸ¿´jîýÆîäÚ·æ£}^žEnÁ$†?D½Êó®X…ܦ›á` ~^6_Ï…w½7þJÊ7“g Ÿ¼Žká¾@ wǵ#ùF[ØSMßï<4 âKÙ½½èìÁ·Ä7$Én¤Ì÷ûu‘’IAˆ+k'ÅkƺAæwÊìÒ¡ÙX>?aµžªù–¤ˆg1Qµ1ÈØD¯8.Ÿvï6¤õȦJuaîX’G¿›r(Þñ”8²rÉ6‰>_Ö83AžDÖZÐ\{`©Ð#»ê–or$J9÷Ôò’¨RðèsbRE<Øz–ÓZÕàEº XwÞhEÉw«Š©ŠÛ´zTfÏigðþàP(¬€ZÛ=YÂ<„ňm*C©!ð/zd²Î ]ïYÈ!õ]5J:9§Ò" è/ÙžUy×@#½V¸‡2ì"“WÖüÆ¿†qA›ð\xÞ«Ô)ÕðêÈn9°©UéiËEDšAÉ.9ê³´á*¶aè’fB_&vo_Œñ÷ñã"9ÑðÕ3N`wmÇ2IE|=¼`Ò¾M›ŠC|~i®ÿm{XmÆ}MþS ¬5¢AÃ5ï¶ ”l¥µðOLÒª` f©A¾ÑF¬%GˆØâ¾¤÷êÔ g™¨|™ü ìæ·]]läÅXßu8Fð?Ã;ÆÏu;Næ mG{±ˆ©½m'üªÌÒåG €PcA`+•MVX ùdCs>UCGÜPÙa²uPäã×zFúÐv-ͯGJ'—­ë4LU°»Æhø¶õtïu]Á`j&H‡’µ?ý»'—›ÖÞ)šÊ“£Ü7Í—¨¨?"~L]–Ñ'Pƒa¹_N5¢Y"몄É[–h3ìù©9½xs”µ¥ŠgÆùˆy bõ¡Š'p´Á°£B…9VÖ&hðáïÒã2º‚ëÄ¥­ÃsQ·œ±œ°‚<–“áw;Çg+*rÆÞ._´éæ£J® çß[Lý~üHk>±-é•fT×W'”»%c &¬&M•Ìÿ Ýœpt³ãÆÌ "ÔÛÒ¬,aë¹4.!õôlÞâ=Ÿz‹™iJúÕã :9ÈdLáeöÓ¹/ÛZrÚáPAß™Tq2ZÙµ¹‚óì4É;-gß§›šºzíîš„€é9Jë‘>RéÖÓ(%éW¸‘XÃûÚ(¾ýð£þñ›²õn9È?õ—Éá)`KP–¨Å,…9¾ð‡¨á>5}ï8N,ÃÍx“jü­:€󋈿idؽxµÍEŒX]!ù’Ñܸ¶·õ’› „Q¹Q#µŒMö³ÿz%éNõSÎõ”rçà‘wR.ŒDEÊ©À¨ƒñ¾Bà¼å¡Ò{H£Œº“´-‡»Ä2nWÏe §‹ä•ËO®ð澦Ž]o9:<ýHŸWôg2îÚ5èËÀ[]²iw­x¥/²[•B»—hÊ JD’š­/Žã0Ø‚ß\÷×»¤x#6뽂3CÓ¤uô(ã4ŽÝæÛÁˆ‘ƽ¶ pׇDc<)Õ7ëkÙpËß® 5ÛbÊ&Üa3L±…ÜYmÍI / 4 ~~_¯Êx˜výìFÐgˈN=¢¶Õ)Þ4¼`ãùD‚Ë•ˆ‹¨}=+VUÖc{PcP[!åÊa(•{W³ÀRðˆâÂ=ð²3¼y¿Îþ¡’¥Ì.ÂðQ\Ð׿í®y‰%ÑCQ¸¾ð©Â½Ñ¼Ádƒ”Ü G)Lž{ßaÔ¶Œ“ã©H›¹HК6š•ë·€Î•í®‹–‰Ó£Itm'Ô"ñëÞ¸˜E3bXÕ|È@¢ˆ â!CÉGzLÚÍFã’¼FËÍ1ÝéB×W8µ=\¹¼´nu¹bãwÛJÁº›QŽ}Á_=¡dh’_›­Q9ä KdÚMk|},‰Û)â½Ä714Æ:?º7iǼmyþ¦Ã;zº×=–דÎÒvQcã¿l'è7r½†À÷úvD‚5»±_Úk°ÈIp”é9ñhd|–ƒ^Щ[1‰j4Ù:,‘ùµÓ¿úȘ˜1Çk×êBÞ8Ý×zðûÞ€ðã×Èß”ÝS­ZhØ/;váα)ÌvtÈÞÈ6B:Å^—QŸÑ£ë÷¸Iä'¨ú¿>BžÊ Œª‹(;’sOVÚœK„݆šz ôž>Ð%`Í>« n)š˜‡¾]jn] É´'{³P¨ñ¦ÙkâÒâ§ÅˆñvìG×jÉì‰/ü]ïÞEF'»¤Æ…ܾ¨M\G¸}j½„DD5¹ ²/‹ø¶t7¢­¡ }]à:Y1g‡‚Ó퇾º6Úò%Ürùϯ죅'{Bi`cÔ7G·]÷þ’8fšRò³ôo÷°±¾I½ä­0Õ;VwA Ü€[F÷ |»¿×!þæ"^ì"ŒÈv DOMËÚW˜´¦åײ4ÚVhŽ|ËxõÛ®kŠËÎø|nl|‡â;Q(æÙÓ«Þ/Ò¿`Žû ‘ò5!IÙ—:’lªYîfŠ|4^=»YòöAoÿùÖ`:’ïË›vŸŸªq½î $i×iÎW•žçñê ¶~>òúX¶KµaÞk´ã%ÊzÙxX&ÁÁ¥×“t"'ÍÕÀN<ýï 0Í‚´›§Y5ú6÷ö­¨º{9ã’|¨ôŠTEÔVÅ‘z•Š*2ÒÓ#Wã$íBê¼ÇÂó’§8{Н°jï:V˜+ù¿Ñ6Y®kͶ°›7HŠ×YÆ-1ÎW©CШAO0‡Ú¹û>Ó#Dè(ÖK`‹_UÜ"h#ĨBk¼:§oÎØš'õàkç]Ž_^´78Œâp$ý Ó<:b‚!52}¢¨ß%Ôô…‡H £hÌ¢rÊ è‘| Ù[¾|âÐ4|„œ)à¡“aöÅžªÑ#óžÇ‚;eÓXôê=pÝ–L7‡^È ½V^+¤•øÁô¤‘ŸþàN[~f9„å8Í»nfMžßP±Œa ëØ‡ŸÞìlÄÐ œ/)Å·EVæ£2±[ò°йðrö òýÚ-?S­†™Ï`^2cÛ\Ø)Âv7{¦†ÀÚ-€?«[eV7Ÿø$›®·\…M\Fxmz´ˆðtâ3äÒjÍ2Ñ›&²çô‰UÞ>Qþ‚7,ëÖI/Gâþæ¢:tì“æêÇéð)Oi*!պ˫üïx ¡‚—„LßÞP®W¾9VlZUãà » î¨)É$ëšíN¯²ØW?h^Ù60)ÖêƒY¢ùAGˆv‡ùv·O‹¸j€©HêTÖ¥í•ÏŸF…ãô™OÄLFŒÀˆ)"ОÁÎé¥7ds2ü™.ÇŽO7(Ç€7·¢•¯¶£5R.R"EÜݽü€S5>9,qÊ:?.­ÌÕSA|¬½­5ÀðCr]B!~N?µ_ôÃf:˜þ9+%¿JÃî¦ß.¤¥ÁñÖÁðÑwÿ¯}ä‹}Þfî!D`VV?IìáxœG4¼í¼†|9Õô/” ·Ž±?Ô¨šUÔTòUu #o•ìóžµ”8KÍz7Êð…I÷æ>ï‹à—ãìÉ!”/‹ÕñQº,φ61A諾­×ê@Ò­ÇÈ "¾•%7™b†'ñ“ºGà‹-!îÿ¥øyR²«ýjÛÌþîÞ²G|Þ\¬NšMÅns~8`›~ÙR[Œ éMmO˜U–ÏËË%ð Úe-ø9;CyyÄqEÄfɬhŒáIºlSb Ø>Qr4¡´õiîÙIÍÌðÄ ÊÔbL2A”ú–°u}ƒ‘J´2·VŒHŸ[t] „NÝpKKMëçï'å½{bÚ@¨‰Ÿ:M/W Z5Tîh4AuÂÑñÍK`Ȳuמ­Âr¹vômÙTË^Ó¹áÆ,¤Û›û˜â‘íQQ‡3çÁÃrÄHŸÔÇ2Ú†“k°ô\SfçÐ0'Œ7X» Ô4ø êÀ}ûÜC’' Q‘Rãõéâ~z±²›ÞÞôɳSœ³.IV”ºóÏm­sÑ"+5ìZ@dÎÆ”¿<× n²1oH©Xáh=]×Û?v9ñÓ@l¦Z&ˆûú‚¹ùƒM3íkV"jËMb—Çô€ËcíôÍötÏ3u ±ZÐ7³»ØÝÒ]¶G5»lwöPlakuA¤ÂŸ>¦Øoy怾éÄ ½9¾Âã2Jkœ(°¿½ôc Ñ­{9ÞÊ»~ãðÙ€²J}#ÖT:„2C¥hßKNŸ×,½,Z ˜Æåà¢ë®þ¥²È/Ž6Ø|E†À¾Ò£w¶þ®¨|ìz“xàýRÝ-úhâ·ÍØÍ¶½ïú· *¯¹Fæ}#Ô”.•KXÚÕñi"öNpB7<"&Wˆ„ÓÀvóçS;Y!èê-£4Ó\º5-…íˆù Í+éÃrœ? Dâ°O4Q’ñ4™µ3”u/e7ċݮäJÌ•>”Ý,Œ¹›m0¹¸›ª¨*±$Õ·Ï·*g«î•§MñÝ5&¤kû¼Oô¢~[k cÝ4þ±·N„b¯þj¸584 *`B[äƒ8TÒÛu› w U"Ê­!äãðÇMå ñõ5½D$yWœH­\*ës4-U<.`¦\åÔËnÌÅ®n´@.Ô /þC_ãÊ…íÞm*1í°s½ëÓácŒÐD›Q\ƒ õùÅ ‚Ñ3[·ØÔ¼XA¿¾ªÜ®%MØ(¶8Oz™.¥Ÿ$¬Ÿ”r¼hሱcãujMâ²gÛ·ãˆ`Þñ6ëúW„íDOUº¸·ò‚) F³Á.Ýÿèæ6›J Æ?ŒH²ñùÌðAð†¡ü#1>Õ\Dú-rP¯J´›zç¸G»`þ&`ˆÏyC¾óKÕ'thÇqªSتàFù±Õ bC£ÔK删Üû•@Üxà„ý™í~¤»Ûž l[ʼn»XçÃ㱘½‹¬V½LêÓ+¸g Â5Ú…€¨úqfK{Z;£™©½ûUöhž­^´c–j¸£æ§^Ì1…ÙjF¶`d!ƒÕ¯°*pkdZ8Ü.~5|SËrù@SÙûcgb )¥¥«Žt‰½d ÛŽvAöôÖ›ûÓNI€3‡‘]n³,ŠjO ŠbÚµ¢7Ië¨ÑøÖ¬ÛÜ •bFeNExÿÁâ¹…oƒ@;Bå.ËÑj=5£Ø<š-Lú£5«âåçóÔQ?k÷èW¸åðon³]¸ïl`j½Œ9ιYGîUºý3 ¾É²›Kð0Åž• ­Å5ú0øÕÆu|2êçeƒúÌ}œ¡¿ÙWqNžŽE¹ yQßÞýøøs­Š-›šŽÜ=KûIÞn.Èkú-˱^àV\ bŸ×Å€ UÏØÊw:o檄]È×µ¤K®Ïz¨BB[ ÷ýË‚ç”=FòË™)¡çª‘&g“M*äÉézèn©g³‰–>h{9<Ò,ëÚ…ƒÞqCéÈØÄÞk\š“Þø¨rÎ;ÇýV*=½Z1¡û…|’ `¤k{îaû â£v[!žÉ”‚‡î†qÓ!èŸÀܶ0¾N%rÉúK´´’„¹F×l£”½·‡ 4Lš‚ga g¸ˆSƒ²1ÈBÒ½XœqR±Ií IQüÌh‹R¥0¢/½y6–ƒö¸‘ÄÞ™AÑ«]ï8H»VñÅßqOÜ×ÕøûÄûaHXŸÍ>C5KÙ}¬Çf]Ý„¥S%Æñ/J¾ uXÌ=gVrü|ô˜ÒÐ!³hK²ãïFd÷ïáN¥5xÚq…òÂÃN(Ô·p¯‡—5½£®æ°œßZ…vºú¸îÍ€ ý ‹y¹ù ÿ™m.Û± ¯wlóñ“¼ót¿(Ž¿:vn òìqÆŒ¶ÙöÊ}{a›žÎDAWUgl›üªú1%.M¿µ{j²à8=•EõºHSp§$†é]Xr¾»9²äŒÒ³þP^@Ž¢)ŠÙ7Ù«­p[Ï—ð[j:n…µCï]Ù¢äJ««&:]‡“]tRõ‹º—غ½³;½¤aP'auJQ„H…i§ý™ú£5ÏßSÉ!lè†'ô×t¶¨¿·xlpñVÛUÖ0ÈŒd¤2 }Èh}¤p1¢»Yc°Q½ÕümØ>´`[¥J[‘¨šU†ãÊåx(ÁЪÁDo¸\Ou|È¿EöZ%ñ†Ü±wµÁØ¡Y\Q SŒnjc'"¯:SlüèÁ󌸑 ûçtaÚTXˆJâáv9RV\—Èå"/Ú럴5¹×—ŠkA¨f×A]œŸˆÄ±èI·“ýbŠQ‘‡=½Óø„EN|¿ìŒ.”¯S÷qȯ­ÁÔ `-°µiRÑ‘…'¥!ªd™QPwyá·å=°íæ°`=¡¾[Å»®‡MÑHJ,«ƒ3Ö¤¶HîîÐùR[Þõ¶<%ˆ„ýƒ‘4éMlá-ò–öO3ì„l7B`AJUC|ÉÜ’è`˜±öNòkE›cë˜îuÀ™è•0¼gFéRÔG‘Ó¯ØØË˜CŽFy]aã“ •¢Å 8ÇÀP_<ÞZðµ•t¾!ˆÞ¡ "ÇQ`ï¥>IÍ’»á Nâ©Qýä\™“¡y/ªUÚ`^hØ¢´3ûn·!ÂâLAX¦àIê8ã\‹åÕE^Ï0[JûlS÷0¹*)ñ8jÄ-Æ<:±9ÏÙ1;28 ´îͧ¤ð5ªùèC7Ý]ŵu›Iíÿ_¤†,£kùy¸"žø¡œNɭч[®Š^!™qݬ‰ÕU=à°LÁôÆ@éFŠ|;ÄQïuÉÿ®8*ÃwÅZe9ÔÄÞ†Qn>µ4Dö·‹Aݨ™öØŽ'¬4m‡ëßg”@vî€o„)Asƒg`F’r‘Þ !tÀfÿTÞºéŠEê†L©» –rAO_s‰¹¢?UѰ£'=v? l¾l°±ÆúÒl5o/¨`aék6HØJÊ Æñ®DÀ(šÛsP9¢èó™DÊ2ª›C&}ð‚T„fUêQì+,.ZôÌr@f¼^TåÃñ3V ±¾ˆ%:Ò‰®¿Ì¥%Á'¬$EíÂö}á*™áПøa”\{»¥6pÍVЪ$MBoáóÉo(ˆTÎã•ï¹Üx›ˆíû·=¨UIÞÖ§.æõôp@uÚL9¾ìw6go0þ(à‚4Ä›0z_ƒ5'¶ËáÌ·ž¾ôü”XVK^‘ç—¦÷îRÂ0Üu6pÜÅa“±^À÷¡›Çö ‡ûKT™´ZÚZKW©Êö7°tʳ>ÍhÇ\ ^p’èMÊ”À½Ÿdçl2C³A·ô&ŽeÃÞ +,6k¢f”ýcóx3[ß è³à-ô3;;½‹07˜XÒ¾º ÒéãÙ0‚ms¢ÃgÀ´!G½hkêO FÕÔ¦³æùvHÊ-e Š““JÎ2G=m'­§›#‹‡mÏ'‚’+ï”óbªŒ^}Ÿ+ :¢ý•ýõØ˾ Ãûïž`©T\~%‚mw\ÈZËZ?'UÌ ÈØVW/W)úÁXK#þS:\W<)``?ü³Øþ FãÌbÂÁS b Nbým°øùÆWF5“b‰‰y'¹Åæ‡Lv6qœf’ÄŒ‹±¡µzŒÍCm pXÒäÉui²}½]<2ÿÕ—ÞÒ²Ô©Ô9LŠ~V*ÇÛòË ¯FuÚJY&Öºó’ïÉܲlå¥HUBxûá3ÿŠ9»aUÌÞ­ÿP¤A®ÙÑcÞ×O7” «R×0æêjþ&w ÛPºq<ý6ó“;ŠÚ†hnˆÛ‡J©è¾2šIÕç„;Û‡ùNþ'ñy&[ÙÏuK}¢_{Qá'!#­Úàtg6‰*še)ߦuºŽûymª ,Sl*l tDBû³Hê@L8|uá1|–Í“ÕcèÃúÔäÚÁ3ËàWŠ“vmYa † ÆÒ¸)4ÌSƒ •ºùb&ôŒõ!Ó[qõ>¿²ˆ BEÇ%š‚ÝÎ[­w/6¼™Î<»j@«²%a8fgGxB‘ò‹A-ÛÊÛýÛ0QÓÈ%¿sÚ¿|DjÕ[˜QÁ7Ïû[ÆsðÚMÉNÊYOIû Í çáÉRÐ Ö•š¼ØJ’ðXG!\pƇõi>Ú}fS—uÕù”rLa›³öÐÌJG'd‚nZ .I²„GuÖçÓX,Uð¼Âr>NùÄ^Ë|è’"þ}»•eS)zߊˆô“sÖk‘ëa4i¬ÌG€À4B¥ Úc.ôÃW‚õn.‹Ï÷=ÕvTþ{‚òìS4´³¡«†ê‹Ï÷é‹I,Ø`åŽ;â*Iö‰zQ¹2éÜŸ›¾ì--šãHö?^5,½ûH­8uLz°JvT&Ý®xMŠ“Ï’Rbˆã»*½Þ¥¼–à¶>w¥C4­là™Mï¼ì ¯P¸ûÓmÄ^WW5~iú’÷wªÚA‚êåÞ!øþ“àXá2µgø±,& ʺI&BE{‹á‚LJâüX64ãNQ­#qwU’zn>TžÕ8Ÿô¹nï¬f@DÈóöl’·1_KzÂL0ª%ŽN#eõ°s”åy…¯ Âj±Ž§lìJ*úãw¶-D!Â;—Á˳ìaî`ü¾>:Åʽ¸;×í! »¼˜6ßâGÚevƒ>ÜaˆG›J^áà ÏQ Wû\¶Û®qb?«—yw>6êÚÚÒ·m­ÔJ´pÈq”%zE_äÁ Á±ü^]Þv¶äÚAè̬´¶Cå0 ÷ˆSP‡ò²ˆLÃI•)q‡~q¢”ÑA–ŠÊn ü*ŸÑã,Üñ£‹þѽ³˜÷ÃÏVŠ& ıÔHÁki*c°{ï\Þ”^»7ëŠ/RÚš6çavV:ÝÌ0}™Ó¦}²ÕÁigsn¸ã­Gª0ý½=DäKmÓCÕ–Ù’ÆT’6*è’…‡ÉÄÒÉ·~æÛ’¥Õ‰;Lúñq¦ê6|¾¬!¶3ä 3¥µ|;¢( Í™”ã"­eOLר—Žï[Ûrie䣃k@ UƉâÆDM=£85€ïÂÁyüØÞŶ~9+5šÆeГ±êœ8&S·˜ÑsVblQåžøyz]j½µè²C*þÅ2Ü”'º–·© y­z.Ü,‰Æ¹´š f¯×:v/øëðqx¤eÊ^>öšs@§J¹ÅEÝ‘‹üb«ÅJÏ;¢¾,E®?“ ø”>˜?DsG+¨\ôa!Æ#)Ó$÷¥´Ônbâ¦9Ä1~%Q¯’‚W'·vü®œ§¡?'J›–§æÜÃ09ùÙlˆ׋¢d$&#½¯½[”a—ŸºHeËß:Úâmb“«$zIÜ…`ÛLT\\· Ÿ¢¨ú†‚\#Ê7“ÐËÃh—\±Þ]X£Ci†¢÷ªÔÌr1»ôÒæ–d”¿LØ 9ûa¬pÆüð€nFCÙò‹mâ¢|+“ÛEi]Ï1-\035ËE8Ó7¤ž Jr¶d°ü¹8âpìj)*üENÎ ÿ-ÚjZf»@{¢ß„î–\ý²Yþñ”,E‡Ùo›@9ß(ï¾TXqylc¿ííº[*¹ºqU?Cìü,äOVõÀ´ÒU&x¨çaq€nj,€Ä8¹©&zÎ'“íTÊp(øæéÇ ïvè±L̵;C%.·³ ;ëRÔœ u'††Ý¶ÅRÕ‡ ðY?[ožISLl©á(ÕO„fiîT£.§åKÙdû£Ä Á-•MÃZ¥º3$Jç9 ù%ÝÏ<Þ6ê„N´WZÿZÚƒóŸbXQñ2­ ?,,C;Â\Xv ÷.6„¬×‹0ÓòÛ7˜_7Àkè#ÐÔö“˜ Øï)LøóyiÛtÞŽºŠ(#cöΰ=»Õ.¥7EEýœ_a>¡»6ü¨ž‚÷ž,T\;²°‹ æSQiä»ðþfn]•ZÏDvPÔ \‰%ŸRÌ©8›|«®,pð?gºCmÜEã"¤û-š¿"ZÍ`ËZ©£u;WLŒT5s_UyñºP“8\‚sid&'½T95AaÓ%r+€§(a^6¨±,T”jb_:¯áŠú?fgX«RySåvjÉÏ Ÿ%/uo`ß.^ÙÅÆ:xËÔstò¬yzC:p_½,GìU¸%5ŸH.§C›Œ,…?õ4ŽA~‡™¾z¹U_Œ-{¹L«!(Ðø.ÃǰQdÐ8…ÏɃûq öœŸF¶ ´y±¥Ã·|ÿMP.;Õ-\ÛovX”U ×¥înRyG²GµåáäÇ¡¥áÆÕÍå°E‡V:´ŸøåˆTðUû~$.X팛y×Ç/ ;Š—\üä½wYž1©](Y´â Ûôóâº9OR6JΑh²nõ›€÷ vÛ=sNk¾‘ȧß%%G Kæá‰umN#[9(dñOÚ\ì ,ZBÀˆ­F&“©»ÉÀÛÈ­Ãq-d¼ßÙ"Á`*èñ}[8ÌF„(3Ná`¶2)ä‹ÃÊn!B.Wñ¥ö2þØøî!¡?4ž¿ðÝcîŠBü– ¿ÂŸpÌUÏ^?p’OÏË[ø\èCáç]xøIpVZEJc€~µÞ¡{Þj‘ ½ñ°Örªó¤‹ä„. |]r¦kx;ÀáK’¯{§p¡ô<‚¿ý‚Þ2ÜH_ùAÁ¢Ü§ 3©0Løpí¹ü>¨2+îs‰¡“jfFú² ˆpV›ÒD FQ:1ÁGõá “fJÃpu‡¶J… ÃÔî‹ f4ˆ¼ÜC)³»ñn€Ê›ìþ¡õ¤Ów‚cOÔÉ'iÐ~–XT¤`Zp°A NÅb4_… °#['õ×Q™‘?ÝÆ¿Á{µÈÉߘ„Á€ªì;ã xçrÄMûÜi·&jB!âãúcŸò?1f]Ì? é%‘½‹iÃ+u”C-¿kljŒ3«{›f¸Å^&j¿NÜ2¼/*V@Üs:Ã>èô½NîzaJ…–ú¿ß݆bÛ„øŒu<¼S>ÂvtTÅ*½€2ÜÁöÝ…Ø}>0ÛúhyìÙ- @˜-ío@YzRFšpþC®¾[vâ1ý‹ƒïw=”Br‹×ü:Æ·Võ›íI0síTYýêo€íƒ]¦ŽjWøm‰€)Q _3c5JûˆAk.kÅèÞ2Õ[K©ïðüÒhäúF“$È?#8ÖCØ!*U”44¿•ë¶ë$ /léîº ESnM|¸fº Ï }ª*…®yÖ;×ß{iuàLý™no殞ùüós“òÆ ¬¯ò'ÜíŸ%/PàŠÃr_LŠ’å÷[ÃâRÂÌ*½ sêÀtº+˜*P²ÁI»{{;¸zú[Ä‘L1åý’ù ±lƒp ‚[îiÆ·5Þæ ëâì»®5Ü,©8Ï3&‘@Î4>œ’Ù¬B+yå.€<ËÞr¥\ì[Ã4cÂÅ6¯âØNÌäqNºÀïþÜÖ!C_FÖkÐ5{Â(ÑØn«‡jY®?`œËÿ14ø†û#-ìÒž:/æ… ò=²Ü®Òúˆf°óåAʉûLÝtÐeko^ÛDcÞ:Š 5üN© Éή;æEƒø»¶…0¥dÕrfÏ(Mô×\Ùði«³¨º\Ô@Vˆ¯E’´F`êZ¹¢øÂŸ¿“q}KvŸæ C+”ê–ý ê+†ý §IçDy:¸ºw'ÿ¬Z!åÎS'®;Þþî \¸ÌKsµ~*%#PHéìó’[€v¢}x™ÇeîQc[ ^í0¶ï亂òMÐܦäeÐ@¦BÇ€`1E¹¿Ý?$íù*Ðz|!<àü)Ì­ ã¢KÊú£{t ’áhöó¨Ör°ïâD‰¬ë,÷l­Ž Á*g%CK)F„™°(Cß¾e3ò*a»Ý ÀF‡lv>HŸ¶¥³¯ 4È B2š+(œK¬cVâV}3~)ê[ÖTqÆáVÖÔñ\A&ÜFبW†ï‰£/Wͽ¼Rxߘ«Y8;l•—ŒË»imÁI¤TñrÒ˜$"zê)ä)ÀoXîdëûò¬O%Îþq~‡«é²Û|=ºÜ÷Žfḑuù뭃蹚0$‰ìšÜG~§tqCª¦Lï;¯ý}÷ðÉ<aü옅c}â'„,(ð$õdêæ?5Ц°aßu²Sò¬Rà©á•ËÜÔÀ«Tæí*½ CBŠ!¥ÍõECÝ[0áH^Ì·žÿ4êœ}T©ˆàL »dh*lÚ1dt©¯ÿ"mb,å8¡ž+mlq³û„g›€}'xTžÎÉõ,Õ‹ÁÔ±îZN+ =#žæ]¼dëܾM—&)gceåófÆìøfÚfÎ œGô›‡õÂá ›¢›¬%"Œf⢴d žøIñ©C×»¹^8•òÈ ‡)Ø/‘œRÒlѯcb–qеš;ÚI¡z®INÌ1ü»7ITföm<ã¬ÍWKþéÀl.˶ -ÔÐo5çR$µi|Ǩù}ê‰J·t"nb·”—8W Mï3ñ1hüÎsl4IR¨tàkXI¹'!ñÄðÛ—×:vj·%²Rák ïNYžÈ”·E\YTÕèNvðã†õ57Ʊqr§¼˜´Igº×ìÞjiñÐ<#æ`5ñNZ\Œ_´™Ë€¶[jXH]½º&¥š?^© ŒkÇݽo²œÆÅÿ6klYÕÛÐ }Ú,Vå<¶%8(¼L¾h\OÑÄ£¬nT•ºŽ=„À[Ã-Å´¾¡MþgÑezXàüuÇÀªÝFæ]TCNA¨ÜNt¾5Ãö¾bP@°k Ô Y‰ØP‚è“·×óÄdoo”Ù#¾vÿEèÌ¥*ÜË™ÎÝäow°4º p0É«$–{LTÄÒŸ ±Ò Q€€RÓÞ©áÓ&¿ûºƒmDDx…+c‘96u®'jt^‡ãÙ– Æ@âpù«‘›ôÿÝ"àf£‘Ü6£ º›ó©¦Å5»Á.¡kµÈÏ>ÏÓ¯nÜ⬻†²¨‰Q¬2­ÜcÚ“d¦©þVH+¶Ô#+¤ªeßÇæËi#¯¸Ûw>–€ö”«¾‰·ñ­n“ Ò*DAuÒÉ-ÁæMN²TåŽËÀƶ¥ÅT ø0ÖÅ:yÙÈ´Ù»<¤û ú˜uo —ª=#4s`u4ÞFw¬¨ºmôù¬[5œÄϯ÷TÄGÀúËc€Ã¸C§XÍ“.µ¨˜¦¢f*”Ž^“± &®½0éîß4#/ÕSV#öV.Äæö€±…Q!µí„|BXæ‘Uš#Ž/ùÅ›ð@J6ÛJ ˆãÑm‘7ˆ/ŒeÙ¨·E'ÄÊ&Í%¯|ÓÉV&¯& ¸Õ­h¾‹|s—Jqÿ"îGH¡º’²6Þ§6,Œ2ÀrÁv}Th/Ç…?†=…‡£«'©‘C 7RÒNûÜCŒýŠÿ9¬·ÿf$Ò„ÌX¿ÐUIÀàf>2R ƒHXNO–‚ß\ ùÅ'aFñØ‘«‚E¤âœËÑØ0+9ÑüÄÇÅCb?t5€þHÜ G)µU¾OšÊo´è’÷¿,³é‹Y:»ÆVØ,z{½Êh5@P‹¥ 0bŒK>P±˜Oã/§£×Ê‘Ò_è0 Ö¢;ìûå(ÚG^ØL„|«f+ ó‹Çð(ćûMr€½p×`Ã|Ǻ=+˜¸Â÷¡¥Ha–’ˆJÌÌFqÅ2¯Ù®fcœLYIqí0\¹ü\IE!ÇUÓKIeë.œ*8ƒ|_ H{}Í£Ö ¼ï¬K¸~ÜüOÛÞB°„¬w YÀKx¸9zƒçyߺkKèÖhd•UÍ©i2’ç—¯ËG¤ûÇb{€J8üŒ¸¶]Ç1RT8§*—çQ« ø#žág\*Toˆ]‰GñGŽMQϬ­› ¨>-›¢šâü¬y0Àu•»,@?ß½J¼Æ‹ò1sÓe¦­)}ªXvYó˜À•O¥ "—†¸aÂæ†'Aı1þv!ŽþY˦3ÑÏ/¨™ëd™ ãLº…w>A ÌÇP?µ‡µ…E !¸.¢èb>pa6> Üuä Ud>W¤ÅôëÆÿ¥¬-ØÈïL/Ê ª˜JÊ¡dVh“Õ°u˜f›½Š8u°®Ãæ…aY¶4ÀªÁyþý(]1.­÷†ÑI|ì"^[xmJ¿7o ½DDbα§D0\ø¤¶Í1PÀ`±_š…ËâÓ4¢|WÞAëz‘ü½Ã% 8-•´†!*‘b`ºHƒ‘n1à«êDwgµFntOïhK›¢|q'aOCˤMòš<§QçEf  3¿òàágÄB.,íaâÉ'BÙ:#_ÿÆÛ´tæ»/žÜ{5Ò5`oÿMv>¥ú1ÜÛª²¼70·Â)qÏ€ù'"Àœ²ºŠ°Ø?M-¼TÖÓâs,)=ö̤Fè_–¹]Dx7ûÙóÑÃóT¿-^Â%þÿè5lUXróq4µ ë3o'.6ÎSHCY³(Ê:Å7>Lëãu)\´šÆn"KúuÇóbcÞ¹ðíóžð$ò|‰,ð6«¾æmÈ*€€—µË‘;)d;2•QbtØŒíÉ1Rh8lÊ “˜Ù_q È‹NImÁÁ7sþX²qØÍúŸƒ‚~ô@ú:®ÍQí=™Ï\þª–#ºàŽ”j›^·P´ ì½éäøQa‹¿ÉSIK?`w²o2 öÉÀMÑAªÛ­`6Xçqmrz0õÞ£÷Û‰°?H¯Ž3ØÕŒÆn§ò†"RNÔ›jÝÍd÷š|R³ =¨5$Rãg`:Ù{RTïb5[4„EM3-3)r›E bY4D²oŽýO¢ ¡K$Þ²µË¥{ Õ¤Ú[{eÏ1qó6Í/ú´ÔÐa²õ/­Vý]ô—ƒCE‘rÄm”½AQœ+yÒxÍ3?¼0¼…ÇÆåŽa[‹û¸ñr˜‰©ºÅã"{JŒq’¥çY`øÊ á… Êõ"œ½Q+µµ‚`@‹©*FôWÁZ‘ªù9Xð›÷'Á»>Óƒ˜V¢sßäÙ!#ÖqlØÎ•#ØŸ›}Ú‹G¹>B^•àÿƬDbAâ¡u騱ÞÛ˜Èô¹¶fôøqò´ú >$Û=`N¥)(Ží·=óкã¶0;ò÷R¨Æj¿=š„@?Ür~E÷ÄW”gƒ-JF¦ÍĶÐ=´X™òBLç Ã[æmÀ˜Ï_}ªD‹XɳüH‡Õi—ð|ºÈüð\ð÷˓Ȝp=L‡ Áu}G^mµòÍj“9È[^d¿“>ìc/4rýÚY¿ôŒæ‡Unoæk+¹ô좣Œ¬Ù•ñ\*þÇQ/0qá‹G €JK-¾–ºìzg…&›ž=$²vCA®øñ$ÝÃßÑMD„³ñî¨0ÛáWFÞ^R8óŠ‘Hѱ™ý‹cýrê=:t²Â³%/Æ=qvñ+ÉlDüíòéýi¸ Å÷#´Ú µýÐQ *#ï^¥C‰Öuõ}nÓ|&Z¶ÙdkÂi"×9þA•^¸¬'òÑ·Ê’E6Þœ%ØbÚmØãD4ÖKÛ'5ã-Va>ŠÂÞêxÍWšªå¤¡+R×á`óEšƒ°3uß).±OË8ÌwâÐÝ”![ØmQ`ÔÛ²ä&çÍ%qd5~Í dÀ¼4üð#àãߢz¹­¶ãYÆ?à@¢~ø»§ÀnB$i4ÝU+}ÎÝ2OÊí§”† ¢ë[\ô\-rG:Vîm Çî9„ágᔋ-Òä 3Óý‚Æws‰]À»Ðœ” Ê`U^À9Ær+${êNñ¬Ì)’æCGOÞÐÕ%`€'#ĤJµ×º ¡KOɹ¸™4}'mÂù”Q€ä?!sZCì¸!ûk TNY(Aìùˆžx”bœÊ.ÁÒX”xžÔ•Ü—™ÓfT“7jóé±;²|8b1ª‡À£¢J ¢ ‚fì¸9VIj÷”d±xÉ–2£¼÷» Ó£gkI‹àIÑ !V3 Ξ!Yvå‚̲Có¼Z²+~À©$7ãÄ›g™¼Á gùòÝD~Ê2víÆõëž™Z*´Jª ªçhÝ2Ž ÝÎ8œG¤ÉÊ)×`½¬×åŸæWgÛ¿õðùÁun¤}„ÿâá0/ÿfê:5íé%¼w…×C» rŸâôñÝ>•phø)óÜ8¸Zô`t·êwË^âÅÄÊ´L„«¦›¯D*D ;€Ã¨‰“¨*»+yô"Ä[BD¿S#°Xpçe±ç?Ã’¿aÆl¥¿®ÌE#&0úŸùëØ¤8Ý8Dm¼FªF±wRôòáu}6¢êc\x0ØØ´,ÂËþ1^‡ãæsƒÌ‹îv¶Эv3ò—{%~>AA|ˤ¾üd6ï«ý"·/æ' “ã®x÷›½L!/éç´ákè úQÀ/çE7°‚Ù ŽÞäk¼<¨µlP¹Zj¿ü˜o† ÀûÚ}Ž?ŒÈø)ú¤!Ô`ÐjÞ8c^9«u=ªIÚQñý[ÿ÷sÔlèü´Âø®ä¼L;Å9\¾ò&--Í&E[à–¶9PåvR¬D'$É@S*¾>cì!=Kc¯bo|ô䢾,füŽÂbÒ#•qŸ.1úYRñfGÚfTtªxFvet» èÙÃQ Gñ'X³/¨éŒ¾Ýd¶¥Š«&ŽS¾®ËN, ¼â2"u ö»Ñ ñ„æx4V¿Äsü×$ @¨”s5lÁU6_W‹&¢":ê6qÕK |»f%º/nðy±‹~þê-óËf6áÍí¿Áo*š»§˜å÷{‹WjÅOZ‹ŒÆU»ñ-@èm$eõóËÏ=3‡™|â-7X‹ÚÒ,ëy.ÞFµ¤î¦IÅ–² ñ®´U ÃnÀÉŒ¥¥ÈŒîu‘áÌÎöýÓŽQÌ“ðú(Æöܤ­%£ Ê݉:Ÿ‚~rQò.[°jÍrAU"Ëͦ¤1(´V±zŠlù±t¯“3&« c©½X%¶YüdIK ÃÂk=¤ÇSzF-glÄpÎÖüt#)#^ÒMLà%õbÌtqQ“€z…÷´½ð¬|rÃè#K½­{`VMƒ²LA“@oøÃÖoWm‰ÙR³âÕö4 ¯Ã.T¯‰ö~iÏ’0… Ÿ-ýF!â³ÍÅ"|lJ›LÓGsT[W‹òéX$àΪ4îpXÔ_d}¶¢e^ôˆ|uëlÙîݹ‚¸×¹ÒÝx~ñKvàr#P‡mâf}˜É1sõð/­e#Fì¨Àüo3ezÌÚ"Ï}å}HŸr’ÓŠß³@—ÏZtàY߃Bfïfáïo‡±i¡:wØ÷/àÿ6ÿ¼(¬D%†ÖÂâòi‰ZêN½2œ¤)™y^×¢äL$K~þÔÚ¢‹8Š¿gç¾'ú†˜OU¢bgœì,ìL s…˜J_‚~ŸÙ+NÊXAŸW ÈhöžU(=Œ±c%äle€õêQ…ßÞ9ÇPù^0\gÎS ÎÏÊî`&a%ƒô⊔¹âˆÜ¨¬¼ÜØ5“d¨•™øoj»/”›x¸Xo u*ý-Â.dG´ÄðÓÐ$ƒ¸ÎŒ3S*èwÍDæEwr}ÞæIö#hS¨>ZØB&ö´ “J¬ÉÈûâ[¼ú°Šl”Ö<£ƒþÎ\Hkæ~–ŒWƒPh½U:@R$åÛ“ Â¥`ìK«e é€åàKaò}‘ͱÙC“¬Ž2á©ÓjZužXö{^ ü-›çŽ >TF8ØG8ÿ ãmª¸Ÿ˜°N†-WÜY]JEmuãf[§ÝX׊ž,j=k=Ý ü½±¬‘áÛ‘t §Œ¸,‰ø)U£Ò"ýžlˆío`ôõx kFZ ¢çLìa)¯u–yimÿ6MÙý/ÁüŒOøtU ¹Öé-gé÷R?Ü”¯òj½»èø!þ­lˆq’ 6nïºOœnEµà.ħÝ$ Âô€®5Ú˜ôQ=a)Ha1ó` kõc¸ÁõŒüÍ ¹¹‘ÿ>’JeD»¿tåÏż8hÎÆË)1}½¿(°Îãi) yY¥COö²£¥ùaˆ÷8ºà÷¼é‚ÚQ#©V$G‹é³,°©ªmi”»<_VäOëã½»æìZ}³'£}biÃ3rª°¾Yx«Rò³.X´åYÛ!îþ'¦>†Ek¹~²Ç©»M^›ê§ÛÔÌ0Ö•½!–‚§m=¼®C£zý"M¶Se8%’1ÇÏ|î¡T°˜‰•‰‰gÂ)— lV‰´© 3жÚòÎMa’^Pn»€l*•E0s´QÁ5Ü9ë"¼«§Ë›\^ô(§ÊJ’`˜] @cX<”ÌÊùi=¦è¨­T6ò{Õ j[\÷=.¿‡ng,c–#¬1¤¤ƒT`ñ"XÕÇÁÑ¿‹ø&E;5?é+‚ h…||5aüŠ‚âEÝÖ/e`ü3²3«ߦ¬yqÛ”º`¾b` ±±-ŽGc¹œzxxÎûÓÃÉköP`ˆjeã‰ì¯á œŠNFÖGŠ*p.Z÷Ö2#%ºxƒ¬ivYÕy:l»šÌ±ÅMð‚7»zôò¥aµ³×S­EÔÁº©Ìb»¬9ûMÅÚ^1Ûtȸ]¯ËN6yÅH–6Fêê[jP°bùè±?ì—.Æx™AÍ*¸^™Û ‡ïá@}ýü ¡°E Ñ/qøîò¾òZ@â -û2"p¢yq±Œå8;lMâÙÓj*çòfl_¬+ª_mÔš?©±!6sòÚs${p½Vs;Ƴ® i ëç¾>΄ÃQÕê°¸µ.íˆ9·ÊƒdrjÉæŒÁ<°uÁ)H6íp¼w†À.>¿ˆr¸°n,ijKÎgâøÛ— ^€`‹ñ /t¯\‰€¾³ ú¶ÒAÇüÍpæï*À¤Çâ}ë|$x¨cˆ=ÇŸ‘ˆRÎäûÓªU¤R@™Ÿµw%súƒC!`>œSè5. §ÛRÎèúøˆHy*wÿˆ¢ÛõJ®ý-&898O@Eù_ÒDü ⇶«ÄÔXlp NÈ«AÅ]Ò¥j$ÃÛa² ¡’8>ݨ0Ñ…hRªÕYÌiIqÜÅ  ß»7w}÷VgëÆ~õÉ|‚ Šãk;cÃ. j ©-ˆiUÉ€ëAСÜóƒòå ’ÓSîÉ{åXÁ\­(Æ×$¿ä¶ÁòV"ØH'Nõ—RŠ'ÔødR…¼sþ{Úgp,¶Œ„VÁȾŒòѬôPbU‚¸n" ü«_cÄ4_˜ò½GÒËþH˰³B¸­9Â:-qŽŽJdºh+7Ì:Áô9*'`¸¡Ò.PŽÌ#jUoûHiúßц£¬E·3:!PÚI˜¤Ô=®/w`uMŽÎ£”­¸¾É¢Ëm|¶—Á§ +‰V wÚŸg YäÝ^Éer[iõÝ«÷²Ûèæå`óxö.™!@;€ ñWÜÅ| Óp}Ä#üUO¬¾m½þ¤ºS+üIt"ä.ùó3–ÀëÑAØQ»ìOë ¬þÜxZ£u¤¬;ç§á,Rƒ­vgM¾±„b½š 㽞ðTØ›í â`Ó"`W Þ=RÔqÃ6 °[`c­v´Pä×»ºŸ,ì5ÄSíäÜ z°õÖÂ_îs¢í“ü–4cÓ[fÏÕ?þ· Kyrˆ.§NRè”<³ ~lM.’E~5míoô{R5JõÈŠ¤šiuhç´¯)-ÓV&=³"¿=úÐ[ô*5Œ26YÚ ­^ÖSx¤©¬Ï,MçÎF×(êkL´Û/±ÖgYHÛÔÓ‡gs¯ Ô ªòj0³Œèa–¨ƒˆ…h Èü±ë^Âóÿ/–aë^À€ ݶåpâ1Á·`P•ªU³4$NEüÉþôTXv²ˆl.R=ˆöëœ8Ù’¥ãöÅ"mÕ©Œq“y '€ØIj¢–ûÿ{" X>„ª—â6är¾ÑŸXÈ;|Í~3 çÁìädiDt@ŸDRýì„gñ Éßd¼·¥jr1S¾p¦Ïƒš^ú{Pà*¥Õü÷ j–£Ø¡ÿn5¶Sš×¹O^V¹jgTÁ,í$VÀ{gæ1«T½+?س;Ø\Il48w˜‚®\—Þ ¦d^ŸâïáaKÂÞ³Wz«ÐÞ]c’ä±ë„z›]ùÆXŠ»](ÛÆ£˜bE á§G„J4Jž*E ô ý©T`g©xpŒ3ð5ÕzÒA@ÖƒÁÍØÙ"ø†Š« ;íÓO£Z,nƒµA!a]_‰êRnô¼Û7ɦn=ðô©ËÈ)À8îôð{¸Ü ¶¶¤‹=‹~²TZÀ"µ|¶×‹¦­¨Å(| =/müop¼œúij¨b‰Žëz¯ƒM"®=|£Õ—Ÿh†y¹ŽG˜·_¤ ™’2Ó)+Év¹õ¨ f#* dÃù¶¼§ ±êAì±vð f~í›¶NP^wˆ4sÿãRa…3_ L ì$cG Ù.zíÒ²bT";)ª ÂêIÙ/kК2Jt}íQÌå  Uµ(® ×‘Мð³•òœxÂðÉÖkåÛòŸcTµ,Á85,Z-˜^¼_vêþeÚTö‹’±ueHO[09'¯UCÛCÍ2è4šD  bd‰hb€ÿÿd;ñ<}ãa¥ÊŠ¡RÚs±[ašå]# 7Â58e‹ä±ËøÛÊ!÷M¹M>l6[MŸÌ`^$ÙÚXæ«ÕãæU€}Ðàæd²µÐNߨq, ÏäðÌ_v¨ãtÿÖû#ÿÆÈE¦æ{Œ—;záþ[›z‚‰]&¨1O5k‰".e_¤'é"‹Šë;£U=`s7F;ÜÔaH[ [t0 €ÕÚ. uµ£¬~sç0¿³sÝ…#Oì—$OaÆÕ*]« xf3–™,ÁÊ—¯Ï{ÍÜS)ÐæÜõ6k8±Lr0‰×‡ál“Þb‚ï˜k Ý!é°¼Ë ¢—š=%¤wPM~û4èÙDƒò»±=Òµ4X0O¾¬;(÷ÍÔì°"'ÙH0Ÿ~÷—Úȶô;Œ€?È%o"Á<>;K}ÄÓª&_ò{Šï¯1àc— »D–-…UN5ã„Y/œue®ûåúI$8ÏwYT£’/^ú'ó„Îßxhe$;²U*r ²ñÈz¥ùu™ÿta|ÛwÓHjæ’[©Ÿ¿¤°y!€Ï/–²°Áù›Ï^—±}²/ÀF,`Š_ƒÓ©•'-]L;ÃgØäGßÛ§Â (®ÿÄk vÔÕü –‰€B¥&g˜»ƒ÷_#ïí*“;-È8ª0g®äs^MŒ‹Lœ¢ì¯Í–$sùÙùéYâÒžû8MR1Ø2iãÀ´R]ò[±*<ÈÍÑ\˜Ù˜ ^zŽ~úWLhJHÀêÆé%lû½!ý0¬œÅªNehnYc7p ÚŠ©^áùçîL£:ÖË*›¹ë2£Cµõf2ËôèFñtØæÁÀRHOç‚åõnEoÏßE“[¾š| 6z@;"tú ²V‘nax8‡N„èAZ^ÎÖôä"y#Ðaïå @¬ûA7®·7(ïuì.‚‡0üòêò HÆv }0ÏݾÒÇþjÌï†T¦{¢!Wr/ž1ÐÔ°õÒ74Ó¤ÛÚzÝ_¼ÖÑ8Ü7=¢º£Ù}S=o¹Éµà¡Á‘ 3 zì: endstream endobj 191 0 obj << /Type /FontDescriptor /FontName /ALUMCR+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/R/S/T/U/V/W/Y/a/asterisk/b/bracketleft/bracketright/c/colon/comma/d/e/eight/emdash/endash/equal/exclam/f/ff/ffi/fi/five/fl/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/parenleft/parenright/percent/period/plus/q/question/quotedblleft/quotedblright/quoteleft/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 190 0 R >> endobj 192 0 obj << /Length1 1790 /Length2 10642 /Length3 0 /Length 11771 /Filter /FlateDecode >> stream xÚ·PØ-Š<¸CÜÝÝ%¸Gh¤Æ îÁ5xî 8àî$„àÌÌ™{ÿ¯z¯ºŠîµ÷Ú²ÎÙûTAO£©Ã.má`Rp€¸°ssp‰dÕ´¹y\\¼\\<èôôº`;Ð_ftz}ÔìùA º<Ùä€.O<5@ÅÕÀÍ àááâðpq ÿ‡èÈÝÀ5€ŠäŒN/ëàè[Y»<•ùÏO“93€[XXíp€´= 6Bj@kýSEs @ÇÁ rñø¯LbÖ..Ž"œœîîî@{g¨•3Àìb Ð9ƒ n ÀoÁu =èOeèô]k°óŸvKw x2ØÍAç§Wˆ x*ÐQVh8‚ ’Uÿ$°þ:7÷ßéþŠþ ù#hnî`ï„x€!VK° ¡ Êása!¿‰@;g‡§x l4{"üÑ9  ­> üKž³9ìèâÌá ¶û-‘ówš§S–‡XÈ:ØÛƒ .Îè¿û“CAæOÇîÁùçÍÚBÜ!^K0ÄÂò· WGN=ØÉ¤,÷åÉ„þÍ äàçââæ€œ ˜¹5çïôºŽ ?œÜ¿ÍO |¼–O"@>`KÐÓº—3Ð pº‚|¼þíøo„ÎÍ °›»Ì@V`ú?ÙŸÌ Ë?ñÓåCÁ0À®§ÙãpýþüýËèi¼, vÿÐÿ¸_N¥WJÒJ¬*þÛ'#ãx±óòØyø¹ÂBBA~.€Ïgù[ÿ´ÿaÕ‚ÿê럄ÊK€ðŸžÎî?2Üþš ¦¿6†ðßÔžF`úgòßrñs™?ýáþžÿ?BþÿÆþw–ÿÛäÿoC ®vv¸™þðÿÜ@{°Ç_„§IvuyÚ 5‡§Ý€ü/Õôç&«,À®öÿëUv>m‡4ÄÊîïc;+€a M°‹¹õŸ#ôŸ[xJo†€4œÁ¿ß;7×ÿøžöÍÜöé=q~º«?\ §uúï’òs‹ß{ÇÃ/B¡@t®§ñâáçxq?-¨öÇd89 .O!€'y>K(úïàpJÿ6ý‰œ²ÿ a§üßHÀ©òzbªý„xœ:ÿ '¦îßHø‰ ü 8ÍþF|O̧§Ãþ?÷Óäs‚þ†üO‰-Án ùŸB¬ÿŸúÿ >³ý|ªf÷äæpBþŸJ9üÓ˓ڧ·û_î§JÐÁ§JÎÿ‚N—ú|Êìâîð/÷S#®ÿ‚OÉÝÿüºÅʱ×üÄ c¼ÙŒeàð2u´œ±Æ '¾½{Ž Fº·:¦Tû4ú¶Uµìû˜ªZR?½d‚6‘aúiNî.z‰Ù•†³—"¸ôER+•(ƒ¡çuÏsÊC+Y¸½V#¾°Àúœ² JPh›Q.aüµ!WþñPSú5ëþ¨Å²‘Ì^2Á–d‚U1УáHº¨fÜx«±Ô%ö\˜(ŸcœÄõYV‰)Ò•¿³4zôÊGûû¨Uÿ¶ð»Ê±=—=;½BOšR#ÆP¨ä­$‹%õê•wÜôÅ8Sžü™bW_‡r+&.Ä6ãç.èp!¢7_ãB¢{eS%;ªV:‡9û[ê§Ð7=h(­eé”záÝâ÷ÈGÑ$ü焎ï5içè]6à3ôsµµ”àÇ0&³¯e|¬ÂOwQ~”†± çÓ­Ó,YQk#’(ùéÀQ|ÕîÒ‡;Ô[óû!IcW­6_6Îc…¯±»ÈA¬Û#˜HrÓà>šèË,ÙŒX(ûl57Yߌ@ÐxÃ6#û㹊h™9ºzç{–S¬ùhEŠÎ¢‚µy©nª0½‚Gó‘ŠÕ¼]ìÁ»)ßõD ëtä7h<ó«õc9qTÔ¯×ÜôÎÒ7ìÚ4½é™õû›°íµÌ~Ê¢ÄÃgh¢/#:S‰évN 2¡Ì®IHCöˆñ'óöÒÈ‹è©ËJ‘I‰´Q¾$Cõ'õHåI!»¯„ßÓc>¦J òaH=Âîó1%P´qˆÃÿŒP›fOX3:–Ml²Â’ÜF }Bj$#‡qgxqå(ERŒbë?LæÄ±04ó;ÎA7`Ñwëü DªÜXckœ [©c“á˜mjY;cc”aŸ…v'a©–ñUÐã šTuëCjn£w„»Z4Ñœ@xU“i+[ÌV¨µR4©T°_‰31¿‘/E{A‰¤•ÿsù¤Ê̶^=á,›…÷ËR`T"~YzÛ1›$P]sì C92~Åv/¯÷ô¥==‚„nû¨mÕú³=¨kpÕ+)5HD£ë7SzI2-!Ê3wE^+J¿Ñ6 ‹B”¦%»!´²fž{ÌD½õÓs×NTÙ™Hz²ÞWºçŒÚgQýd6p93Ü”j¥Õ‡XV¸_‰"õœÃ¦—Rß's&c¶ áùšbê>© cÉ`FÐ}zœ®R"D“Ž÷µâ"RÖzOz*¶"Š’rëæ-v(€:¥8¿%åíd%±>ë¨L\N_ÅDexæ–4´#èÃ]' o÷IºòåB…<µ@!î³äd£Ä®ç¬é7kÕvñg&„§ŒÉ¤c¼K¦¶,º˜ L²»qøµçØßt1É—ÿV¡áuA±Hçí/åc"¿~(w®S²¯3+=æàX¼–À ý6y’)¡ÕÊÐLžÜ«Ø>nö„¤bw„…ØÊ–TY:°Ï'Ý 8…; ²¤:²r\XÙØ½Î…RY»Ç~“äá)t™?ž÷` WH£q»p¢>Â>¬jUàP›Äö"ù¸.š.͹1K¨tëÁ¥þU%¥o³¿W€!—îGÆmœ†Þ>­#Îåh"„,ð̦ŠCw+ î<”d—án¨ÐöTŽguo–¢C˜pçðÔ câòk?íçø°Pï×4ƽ8ŸÝå³–Õ?Áö™Ñ||äœfê?ÉÒ._.~áÝnÇ=>Ù«öÿ¤a=ª"ÝŸ–^Û&a\Œ'ì–3çÁBŽÊ¨lïÒ4Y!b¡N{”ÿ^u’|¨ Ëí¦ ¾³ iׯ€Ÿá–^Ïrœ‹âß±%¼“K¬SÑ+go;×M–`FWu£ÜRjì’ÌçcÍŠ_QRƒö(•+²i4¢ 0®’®âÚåÕR‡d =†¨kì“,ЀphCºÙÏô·Ù[™iTê_ö0\Ek/¡1ªôœ]’šåhðûxMˆÉ7Ê›(§ZÇoWÚÞ+w² (®ä¤‚~/x¾k5R©IŸ ”Yìw$wÂr§¤ïÍzS‡ZF?äݽ²Xf bɾrˆ¶0 ËË“V»Ñ4 Š a‡ćîUº %ÑÖ†÷ÙY1 N‚JRÊqv“Íø-ðÇY4äÑøã°ªõt­oH—íÜ:„[ÉM˜á¯§H‡ée3Òû(Ÿ³35©Å M·-hÞêG&h8~êw1—5m™ÎÝàʈé©ñ«â/{%ÍÔ˸—Æ­’ŽÁŒIV‰ÿ~R|")ï³tÂÍ‹x[y½ÏÈüÕ„ožò³ÏËW*Yë-f$ŒÍWô#úð¸ä_÷¹° Üà“2—/A¾ êÚòÉ~n9XH†ÿe§/(¶i—¢î’¿EÓ|”hÉÑ(Ÿ/þ0n‚Ìw1^•J^}ȳ¥]«p—ìrì^¯{Iì¾_w©Ï"¢\ì¤1á´MöX@k¼ ï,06Ù'¤NŒXçâÛ#dq<›Ï!¸^H{\Ú?ô*:^GQ°1Lû/ì‘SW½%ë+[ÚËãV•©ú4ŠG«²U_ÿ«óô¡û¸‰Ò‘˜˜‹p@÷Í0X7þ¥àkU¹Àü¤3Ëeî É)—nÆA!N¦(ålÅ9 ¨(pCõýH¤¡©Xb¬j>B/W™°ì9ÿ¯âau¾•­ÂÛžÝÊŸiŠú ¸ EõCuGéðd>=g#3Q«xxªhvU‘ïQ—œ„e ÃŽøÀ¹äg8"›cܰëþ÷'[¦ñ(úIÇ òò9þ’ æÏI°g®0eSúÓkØMûçé*ËÕÜVQ÷)÷‰¼kƒœ{¢¾aVÒôXÑB,FCß$ºãnùC˜‚,æcR¦¨/‹M‚¼wàÁÞ[n-EC«¦èÇêÂA  …©Å)w£·Ÿ•¿¡³ÓÝ=óüs¢Û¦§q- ì­Áñ]^ØÝL¡Å¹WÕ3n}&} Ž¦>ÇêBÁÌ@ÉóµÅ3Tsh*uóÉb¼7¶¯ô‹UˆaÞð1w=|+_0$Þÿ%e—†,;‘ºôÇuqab+~ñCcØÈ02Ì¿®çq![Þ—ªªkÁEˆø¹zÎÙŒrŒÜPÐû„ÔðŽOé…Ô4þq$ƒ+È~/Œ|à:z;ãçåuñyìµÁbhE_ ¼@"®þƒ^*Ë„þ]äÖLB/}ò¸ ”E“G¬R쵞âzΔ4st¬mîÄ9þx¡ZrNÌ–ý–ÍÊâ³ùâæîÿ€¬ç,Æ >dÔ´’˜šä;1M~ì(wGï’Éì»ßÙüzŸ™ðLÁB¿@þò×ÇsÔ·Œ¯ò.úbÎ^DäàGS†‡Íxïà 3 f¶÷b¯XBqÖü´ÚÞAÙªüáZxÛ‡œo"Ú)HÞ$øih»‚Љï$¶{F1 =án*m4¿V²Ù|÷O¤½Pá*Xç„õ£õÊÆÞ=$"7‡fé©04!è€,Œ¶÷ÜÔµGˆ¦†ŸŽ¬Ö©äZO/íNcHôÏb Nî•…DPÄÙ³!ìáäk,j«–ϨÈðvzwn…÷9¬¸µ/´è=Å{èJôàà]Å U´S.†ê–†CÝìW¶j‰éϯ»[sj'ŒÊÄ5e}‚î–Lx–ú§lïÏ ðú0öDAÕŠ‚³ È…ÊÉWQí\Z–]œ-ñh·Ô¤«GÙ  ož˜D…1Ñè™8¡¯È¹¢M\è%ó;Û¤ôáy5FéÈâEm YŠ–U5cŸ£ U×&&÷Ô^À´¢ ÎÑÔgh.o‚]YXŸCG!X=Ê凜ºTjÔAê¥eeÏÖòu ÏJUÁð ôƒà8r=E>,b|„¯P0òPèkãa_uè@¿í»i‘‡ö,9wÇŸDmŽ«¤‘nÊѾ™êh¿mx…„]Í1ᮾF|/[o»íÿéÝŠ«í“ö¨º`-À g•S™˜ ÃJt7‘KÍ’Ìk.áV—-1:PÇ] âè#WÍuˆzspæq¾´,ô+ɔ敔¿ˆáÐ~ž#ƒ ¿Š¼íŸ‚l>ªç­qRÂÚ¬GIWŸ « ‡ÄÝ×}ç7NÊg[[çœúe å—$šû…#¡•©>«øŒX¶z™o•â(î2i‚3Æ5¥4"ÁôµÎÕ }²ÐÓ;Cw¸nS_Ym³Ò’e5?Hà !ê°ù÷Šc±”Lmþ7v÷ –m5œ¹Ø°ëÓdîý¼"Lp¨äV^€ŠŒep.iÖ¥v*ØkK‰¶ŽÏƒ)>—ܪ÷]ºÐgö™˜Ùýàö©+&Ÿ$wá;Dº™(ºj)™AšåÕ¦GŽSaL=£¼32Ã`˜áG„¬é è_‰Å~йÌ&«§Ïw ¯Ú)+Qæ.;gu‹9ƒpª†7†‘ˆ/| ¸U¾•~d‰ž­êø¨ú&­±'Ú™ÅÜê\jî½­¥[‰¬ÐòP?~P–°®ãk©XcÜÀá>Äâ \° cã'ž†¤-ÕOµ‚5ŽXDRß™µº­%ÞØ;JÒþ€å§p±!¹ïH:xÖžd4.=SÞpU—QÂaü_û+†¸Wß ”0ŠÆP` ”nØzÊéëã|å1’•~È´è#Ì~ØÎÕ¹å<ŸÖƒr銾sFü½Ê¹?_Ë#ìm±ì‰ \.³?Ãj§½Z8=9Òfîpjú[$鎮±áAGÙ²Ü|Õb¼B™f®ó Üó íw×ùö¢1åhnšÌøŠ•w›Ûš$ô85–Fô±.þ¾±ÍÛ›$v–mÒ¶¨ð$gr‚×ë©}¤ÂLFk6‘?ž0|/aF¦½þ6ëW•5sËú˜n0Žgl’öQ&ËÓ éÈ”ÂHv¯ßÈœu×Ï<ª%¨™%p躾b[œ«¹-äòÃìÛ‘‡>êÅÓ˜Sû}ôl…ºoÚNQòçg€AGŒ½Ð˜…ÏóßÄYs9ö\Lê'Øýоh2/œŽgCY&q×ù ‚gér1(ô¼(Ù£JÆdÂÉO9Q!dN ¹iyè@Sao§Íp‹Tä+ëßx÷È'»[èÖ÷9O½'Ô´oÙM£È×*1?Š ì¨Ýx÷æt[½;ñ»|p-}SmwÂäÐ6ÿˆ2åoE–f'óÃ=±±©k&ˆÕ/Ê×ñÌb¡ÖÐŽ VÎЕš /[£™ÊŒG T޵Κ½Zk/„š¥wSä!:xIoµ ? äz¤ÈO®ãøg|ALûHn3z§+ÌMr¤™X ­l˜?3ÒõîwåJó\3P‹Ðù.^—ˆx !‡_& œ°Í¸Îxè‰~šŠfÛ¢–ΟQÈš³­ã4{{¾¹[㦼|¹3ˆÐˆøA oï4 ^®zjÿôÖöÊ99ÏZ¢!Ý48>ýQ³Šé ½fEHÓÀ z½¢]’ªÿ¸$BZVèVý¯ë¢ ¤åø ]Ö®çp íTÒð!tÏ£¢§èob„Íßjb¾Pdz±‚Ù»Qz}^²þy°³‹ iâ9œg0ETN ë‹Ä-?ò&‚…¦ÀRý ¥ˆ´„óÒúhnÅ…œK%XÄÆ+œ¶=Ýš‘‰OƒðÎqìÝŽ:Q(ŸºdkkrDÐ……a s#íI/OÙjÄÐIŽÖ5]ZÓ;Ós´>ÛøL[é ‰mŠ@ãšå]RmÏqqKÜð½dæ++éåDFX¨™Œ¦…Xåf¡ñD Û$ N÷·émû9 šøj±ŠjN„/âjÃÂ>†”oµy×o¤yXÕ-^l¸Á) ÉŸ'¼9âÀùðmòµºz€9åæ—ïÔÖ$„QùÞÔx¿^~—àŸˆäàgz¹~C8)ÁŠÈÊ×ê1-UB¸‚œ3&3ýóÍ+ꦶ÷çhe”ì¦w¸QIáïãGâi ‹Ù½?žkë«AÐ,èpm¢/?¹W€ÈrZÊ£™è9L9.N°\:ÑMIŽ|òw/êb!ÜÛÖˆçø"H;kÍrm¿E°¢Q…ð0BâþÅ]ÕE½ÿ Ûf‘•ö¨Ÿ?Õ%\¦ôÅI¨Šg°`—÷8ïÛÉɬïtÆY^ ‡¿N: ë±N{DüÛžq£Î,3ƒÕ4yê+…Žó„ÀîŸ\ –±mv±ËÖIty¹hÌ:GÔ]Y¨ÈqÕ)%îöÔ­çHâù[pñ»ÃYõäˆDB^Tñö²zs¢Z.ä›Yo*"^¯–âáÚô¥×y¿BÔâ£Ø‘#| ûIì°µ}#çGEÇš‘P’N R§ÐÑ6ƒ^r^ó9Šf¶ši óGä¦Wïƒ[¡X £W©@Ë9eXÜ}Sïa†}À=Q£FÉv_*ÎOÑzbEM/hÇðMö"<ô|«ãÐOªË/Ÿ%Qrˆ#®O"ÏÆd³ß%œ vü8Öæ}.*­!Ä‚F”¨Ҵ¼‚o·3º8“죤m…o>…Ä Çkt7)ù¶›CÝÑ¢É¼Ë ÔsýÙ:ÖÁÍ³Ç ¼L9ûiÈ7Äx¤EO¸‘ .)dd\à¤ÁÏÜ<Þ@‰@ê~Ô®±xƒê‡´Df±Ë³îvÂRÞbo#Jâ¡ôo’¿ŽTÕgmÍäµÉðÍ ·¥q×áKŒèçßÐ:´{·þê@ü¬y¿þÊrè%PÛ=G®¬óû3ƒû©f5û—ÞŸ*ßæûDo¾j}i 'è𒯥?67ù˜Â2F Ò“’m•cDuw§PùSJ=áXþ‰ lÉçâ]ºÞÃÉælƒ€ÁŒï- ù&ÑÛJ‹­$.OœÒÈ$rµívBÈú‘ÿö¶çÎÙàñxì~ªßi úd¶á ¶§øgŠé*"Iïò6)5©”­° [. 3õSi3擦e†ÀƒèB´£Ò#ŠÖÛ·ÍÚ*¶ÜNþÎOxDc[M§VÊ;xº˜LîéJ›ßX´¿’¨ ½¿ŒEÄéƒS¦=f?¡†ˆT²Ô£'KusZ¨ «%ë8u Å!CíPµ$3µM¬”°p²2cUÚ;šµ·rXÂ$pïÙzyMÜ»€Ž4wyX×À4$Y ßÓŸ`Øôô¨qâ“»$/ÌŠ¸Ø\aòã܃xÜí”ÛÞ¼ m€i¶¬Ò‚—Œ1/H¿+H}W¤Ù¢V,iyðO®¬ªôk!pàÃ'upÚÙÎ¥¦]Ó‡¿Ét£ß âbºþæñ5Õ2—ÉWTz†êCsW=çn¶°Äÿý«üQ7 ѪzœC- Iß]ä£m;'w_XMc\óù6*‘E¦k9ç²Av&­”Ë"4¦#¤öÝŠ»­JTõ¸}îý{^èöv‚Ó—ÚÆ÷‰ù½ðùýÈ,ûZÀ€t×î>µ¹³ct[²îûûÔúa/³ÒË ®ÜH¢¨™ZY~GýW¾ç¶ÍÆÔ!”ìä‰[ñuHsê[w ‰hξ—ªavN9&x® ‘QÓéyp D˜D×È ¡+±3rh!ÌP´ØMþ¨F õ¡kyi ê$ÉrŸ3¡™=m¾êu·16ž˜ ÁŽÂ„I££tA>T2ãc`_Òl§­AáìFÏEF|=ÿ‹»û†õbhHç›Ìz¸q•ƒ ®“áŽjyçË7ı "ìî/ŒjT ¼Ã“t5n|}~°¨§‹|œ…-ì3¥™ß8zëíßâí*LTRWΗˆô~|éÔŽ‚—lÀ?`¦<ü0ðƒµý|üPbšr)ÝÙñ›Ò1ª’AQî³2Qm… —G‘ðœfiPÞ½(‰îêØ¨I[p‹Tàdù»ýzûšé׊@µÇí¼;VO]¨¿ÏÓç†ÇtKªÚœ·€ÐÈÅËù\mé#ÝØú§ÖB¥á`²}‰–\Cs«/û”6«¤»IUˆE·¤p´?SI[™ÃB—Ò}dœå@ð¼ çjÊó×5lØ ÖãјÛ–= (¶³÷ ]”³J2ÆÍ®*DˆÑR¹Óe4©èfѶ3Ö5ßÎJ%§{Å <¾oÊM:@Y& oO v>Öý\ð…ðOA Ÿo°”p„úûýdvoÕÎÂàî^¡°‡9r+-ÛóŠÓ«³öo„®9”z,÷ß7~uË:ö¶šn³–^#{±ªìQÖ'Aéq?ÉÑÄøhcqšqàæ¹‹žÏË!”b ‡eÚæMÈýYOŒ³çZû꫽ã­oÜ/„)§s÷ñ± >„ê;ô 'qñ×ÁR¤)ï®CûÒ&1ûå‘pè=[×8|¡ëǤôåÖê´B¸“Rg´dâ®´ âˆÚ·i=¤2)ž½o|î+\ˆëæ¶4Úï‚\”­Íq\‚!2ÁKÓŠrÀžâïÍÔµ&ïÑme—MÃ|d­Œî m¨ª©WH 1§;”{Ïà%Ly~Â笮*³ n¿—§ó^‚õ¹•jɺG%¶¿¶®®wÕÞ9м¨ ‡ŠŠÆÆ wYêk{D} É~¸—L¸e')”8G™ZîûvìµyÁ.2òÙÁéw÷'Â=?„u_B¨/N’D3Ø€‚ä?Í[Øæè é£Ù:“\üûMþ8bˆ îRý]ú¦öŒ¢Öç[Ÿ-`)eŽdI~»}Öã3Šx4±oƒ)í@¯úÞÏÒá“Ê£w_¦S£?<²)$8ϼuË~†¡–‰4w?qE)Wwyÿ¶“-‰ùQ*CT«äëñYI+IŽqNb×Ui†Ù"ú«ts×.µÒÖ³‡Ùv¤³`Þ:Y\Dbàö4†4UPÑ è`r=#ÒCöb:7°–ÃÆñ<ÔÞÌ7CÆÑLÚ6ä·b•º©1d{‹Ð™ð2q¬ý–T?E».ÝqšåÆ3Î_à™À¡ ¼—â3ù–Zæµ ébçb|Ks‰ÛBü—ÍŠwŠÆvÚ%¤•©•ƒ©âH_ê n¢#Å^§iéŒ?Ö¸T'ët· Ý!wÖZ\}.x®œÌırÁvÓ×Á=ÃÈh±©±ùZ¥¶êe TA½`fÕß_)5f!¦ÝË„ö•Ü-Ÿø%V±·ÅÑ3R5d…xñÆL¬– o¾ž²e‚2‰=Í]“ó`ÁÇRO3ÉLòÁ”ÄMª„ÖQ@Ð3¸ëþ‡îôB]ZJ}“ ¯-kúÜæÆN„¶•ýdI¦†Ï7 - gXe·%ÔPJƒN(!+)SÅjõËXstwJ{vÊykÄ3òYá=Íbƒµ¦¤ó¾æ÷:´™b¼Xî×Ä_ixF¢á&û)…&ÞTSwTöûùmò¶é:]ÐлW6‘³i„Oz¨ç²éÄ [°¼šM"ˆ[§i`XÓf(­#jŠ˜/¾¡¤«Zú‚**¦Çµß+,(D#»È~Á¥ç’¦Àg+ Î½¡³‘X¨ûmû@üÃ9.®5‘Y(ãæ±[öç–.½5¾ïW õ!æÄ…çî}ɵ©J!i¡ÊÁ`øX”$Ã5 d‹ˆÊ{…èU ·[%¡´¿é¾ÕàCÐò0М—\”Q(ôi5I ÉÒUÜ&]¼KWW¼÷+B¤Õúh©›@7‚7«ÅWW_eU?o_¤aòÛ°í3“s¾}X»Ü^Ä1ãsƇþÒÜ­ºxæ–o›Íþ]òiòàN9ŠÌ ÑaîËøûšfþàjÂÏ6¼ûÔmEùõoZ¨—Á'@¹´ƒ×õ³©NÀñ×…%§«¢òm­¬¸; …TÇÔ®%Щ•à·s›öļ! …Ó¡Ö€@H$%AŠÃ³°ÔŒœ¯ewo˰RÒGÚ«²dKïxÀPå…ݧN¡ÏÜ•„e}Ú©¹XVÙCn}”œN!D£4ù}¯²õ‰á~¾S" Ù7UØ?ÊkøÄì²{ýc‘ch¥:"— ¾ªš“ øÌ+eðšzzÝ‚®°ÝÅrAPñã‰Ï#šòO"pÚѿՔ,*+ÀèrPÞsË×^VŽ_€N`¹f5uvÃC®·J"SŠW@ŽÆpåçÊ)En‰œùœãç…dqØÜ^~fTš^Pç ík2ù+Ãóý±Œ·¸}E1Øêî7ñ í£Ø„]v=­r_¹­Å=©suÞD(|D»Úq\öÔ€êw|Ïo£¸{iEB@컹¬Ó‡ûþK?  j®˜ö<'x¾%»” ;~««sï\8C¨ð’³ ¡–Æ6Û%)n.׿'3/Sö¿Ç`?¹Û6qP§0Sc®Î¥¶n¸Xà+$ ´é.}.oô…÷#Qžkò»ëÔNÓ‹’Ê:€ñ®,G!“—åQÑÆ^ˆ™£ oã \—Ì’ÜuãZsh—u\ ÿ½›f+’´x§Ç †—¦8 ‚š/zwàI ÏÊ™þ\óµ1lØ5aåë>%[´vŒ{‡46 %^÷ôOC6|tm ØùrÔ ˜Lãkk1Qÿ8{gò½©AÃæ‡Ü¯±8nÚ½+7ûª˜±òŸënq]- ägá·túîíÈÁ”Ï€El„³F´sæz³_±k|¿è'IÓY|çŽÄ IršIBæ@‡‘åääVz^á”TT\~µ–ϧÁul4lU÷µÒ Œ#7Ê金«î‚¿ò”¿[l¯šÉ@3ÞÚä9Úw@Éði~Åž£î­Z—,]æ6”|‡î\M³ç[Äô,é›8Ô+‘³¼á)¢ØnØ«€x27ÖÍbG‚ÞÈH§²ÄÏŽ¤ì¤"¡¤Ù¢‹àÛ·ÉÉœ¿ÍAäçñÃà¸k —âeÙÜðåsÉÎäá¯DßU¨ðì†{ÕÔ©ÇP*¬Ê§ˆ ¦1Z‘7ŸvtMbÃÚ®ú¹™^—żŒÁ—Á¼{àñ³9¡´-âPõEQ/ŠêÞfHM£ù‹×‘´ð\hH HÕý×ARnZ/Ð<SODZbò]°sÂr©Tµf§Y—R© !ÝpÒ$<´½rJn3ò9>OÜ`lfÇS|úúüM¿e€ñ¼G-;[ç³±¸çÚ{ |ÝZBré4ð† -ÞVü’œñÛJ؄ٴ Ú”Wø¡>úpŠ+C™¿ÊË.ùè}rÕ|(,²r¸Z{”U¢êéÖÚ³–óTÚº,\àð_)Ò¶™_ô½Œ:ú?Ý`Ÿ:•D±$ sxÊå‡â—™ ø¬~ý>wã¬osQjÂKÂÄ뿚}^³¼î7:Ä‘/2ÜàÑ…¹u[ÿÁÞÍR>ÌÉšÀ PâPêC¦8…Æ<õEã’üªºÞ endstream endobj 193 0 obj << /Type /FontDescriptor /FontName /BHYHAH+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/A/C/E/J/M/S/T/a/b/comma/e/five/h/i/k/l/n/o/one/r/s/t/two/u/w/y/z/zero) /FontFile 192 0 R >> endobj 194 0 obj << /Length1 1576 /Length2 8124 /Length3 0 /Length 9150 /Filter /FlateDecode >> stream xÚ¶P[ß-ŒCq+.Á‚Kqww ÅB‚»»¶Xq—"ÅÝ‹CÑâR\*-N¡<ÚŸÿ¿oæ½¹3¹wm;gí³ö™0ÐhêpHYÃ,Aò0¨7'P £¦Í-y9@t]°ô§Aäâ †AEþ ã²p{´ÉZ¸=Æ©Á ew€›À- Â-(x€@á¿a."Y °5@  ƒ‚\Ñd`NÞ.`[;·Çeþú0[±¸……Ù§¤A.`+ (@ÍÂÍäø¸¢• ³ƒÜ¼ÿS‚YÔÎÍÍI„‹ËÓÓ“ÓÂÑ•æb+ÎÂð»Ù´A® 5àa€º…#èfœè ];°ëv˜›§… ðh€€­@P×Ç w¨5Èð¸8@GI á‚þ¬úG;àÏÞ¸9¹ÿ.÷gö¯B`èïd ++˜£“Ô µØ€! €†¼*§›—;Àjý+Ðâ {Ì·ð°C,,~ïÜ /¥°x$ø'=W+°“›+§+ò‹"ׯ2]–ƒZËÀAP7Wô_û“»€¬ÛîÍõÇÉ:@ažPß? jmó‹„µ»—ììR’ý3äÑ„þÍäà‚Âü3äeeÇõ«¼®·è·“û—ù‘¿¯Ì `óHä¶=¾Ð}]-<@7w¿ï¿ÿEèÜÜk°•Àd †¢ÿSýÑ ²ù?¾ Ø `|Ô7øëùûËôQ^Ö0(ÄûŸðßçË%«¡¥§/Ïöã¿}ÒÒ0/€///€ƒ‡æãò üÿ[åoþqÿmÕ´ÿ¹7à?• 60€ð{÷ ?UÁüçİþ»‚:ìQÊ ó?ÊäZ=þpÿ?ëÿwÊÿŸìUù¿)ÿ7$ïüv3ÿöÿÜŽ`ˆ÷ŸJvw{œ 5Øãl@ÿ7ÔôÇ$«¬ÁîŽÿëUr³xœ)¨-äï6‚]åÁ^ kM°›•ÝúëËCÀP&Ìüë®ppÿã{œ7+‡ÇûÄõñ¬~»@ãôß%å V0ë_sÇÃ/°pq±ðF>Ê‹‡ŸàËý8 Ö ¯ßÊpqBan)€Gzþ˜ ú¯àpIý2ý\2#!^—ÎßHøÑgñ7úµW.ëAnè_ð±,ø_PÀeÿ/(à‚ü¹ù\.ÿ‚¹®ÿ‚¹nÿ‚Ûpÿ|,åñþ§-Vî..×Åoá>öì/üûn¼@VèË 0«gaöõa×µRäž{“b³ {é,¾Ë.]î·Ø¨¯Yj²C6\.¥^¿ïÃ]Ý‘c¾\¡¾÷ýÖÖˆÙž¬ÕñÃïîE¢öô^úÒ¢¡©âoR ƒ”O(8t%?ùÝ;ûé; ¶Á¿SfÈwvÂÖ,$¸öPðj¬ø8±°§õ©F@ã®b†#^/îypéCeÎ< -Š%+þ‰ÎÜÅå,~ÞÔµr"ºÿAuI#DgÑ7έHðgw÷ÒY*îí>±-”Œvéj’6Â3ÂÞßМ•x­°R˜‹üQ3C°g-«÷Sš¸yì ´ï¤øñϰCD¯ãäuÕ¶Lé—”©)"lÑí.N{+t¥XÆÑèÖn¥]ùÛ:8 ç(â`%^øVÝsIùF]‹_Gh‡Þ¦ö1îÌï_ð¹dõÚ¶›éÜà‹l¢TòLuÂMÍ•Z81âÕ'þâ‘9<]}¯DÅy~û°¥Woª&ßßÌ~·±?ÜnÐzÊ®Û0Om9Œf¾Î9ªkƒB.p‹Çl}K¡¤- ')ÊR±pWô“=Œ¢G'.ù*¨‘-ÏÖbrEûˆ6YôÄQ!(‡%bv«‡¬²ï…*Á W˜ò9§œíÁ÷™Æ£ÏY£Ñš®ŸŸ!:ÚÖÕtf­xÉé¼ÊH9ïñ/¤·M´iS‡.agm[¼!´±õeóÀjƒá:Íz'é•¶ýëFòõª¹ :¯?Ù~¦êÒ<`„¹NŶó)7¥®B\›‡œ%>¸/INôQ¹¹ÁóXæ%Lˆ?|‡ ÙŠk…›a:aªL¶A2ö²«Qì zþé’'ød‘J«ô"“‡H½7ÒâBÎÚæ826¦Ð•$g^vwŠÃ×ÛìµV;aÊŽ#ug‚¶\à&¦H×—t?2‘³Õ§ò>/”<óG—l‚$|Ú ¬ºcM{eP#ße¡T;u>H~#€7;éêI3•¢H‚¯Á*¬6:C‹PWà!MwQ=¸%ÎbSš 5õ·ž~-½Ò@þÙÍãMY P´–çæN!¥¦ãÿt7<°¼Ä¬¾ßW(Í‚aªò…Jí‹e^)4 ä"ÙgÙ¦;fñQÄâ…/õüÄ΋ `/^ ¯¹ûøÓ .u‘boçŠäv…Œ$˜SPâñ¼y¦y'Û›u*é禎L6CñÁ͔СWO27WІ×鹚ìö”µÉƒCNE‡ñ¤ã7EnIry q Œ3]&Óº–£`êØÊJ“°öæö‹ÅÞØÄy©®Ë#lœÄ¢’÷¯ËìÒ“ær'Äô¼6ä ¾ 0Õ­´‡š]ÙÉ‚÷¿7ö¾eSë>+3ùRÄÝM5¿#¥Ô­RB Å}Zûi=IJCÿql¬½uÜ4ô3 hl|1NøŽ‰åyiKy.ö †Óä~e¢2?ÿÛZ[DÁ„`[Ū¨)æà—˳?VÃ99 E7IÑ&Hq¢…:d•×—-—漈5;ŽCÂß íˆmâ Ý-Jícy£¶R€>ê›É;®F‹…QßÞ^£Ä•f:Œs^ÿØ“}à ,#Yã“4 šæµúÞñ°&T?Ö¸‡æ@Ûwm+¿ßaEHàéÓÅÑúðú9bkâ&œ> œö»mÞùÆL_Ý‘âµvñ÷#THÔ²­Ñ-Ç3ºz…ëà[Ý“£]uî씊6©ÃM­×úD«1ùñ ª>\T\¼ŠûYÔv™ëÞf8)¦ ;›3/Eý¼žS­3OVbNOÀ%îÓl)¸œ{ÏdÿY‰…=š¼ûÂ÷cz§¯›sevœ§˜EÒ ªˆI]ïíN8óј³•¤p6 ˜K+‹fÚ²I‘­¤‘uAg îìfس3Û(‰( ¢e0çÍzbdÁñ€pg©"˜Ÿª*ÿxÞ ±W±;\ ÈhÏžIØt$v•Ÿ×<7WJX®•Bª Æc;à’*Â¾Ç ò…ãœRKkd˜šq‘h ¨‹NìÏJÑa)z+K¢ÔóV»ÅM7†”]Ü\¢×]^Í{8†2™y­jG‘†ª;´=âåékˆ\òàñ‘fT5OþFÓ!ÎÅ)]Ô7V0âõwáö–íÚÓSæùioi'ýMI¨ñ’Âb¼Š{t˜–y¼óŽ Oc +<{nöÈAßµJuO&•ëdæú×¾+ù«ö¼ŠÖ1鮺µSÉÜø’\ÆÇ¬:@y”²_›FÇYZIK,̃¤ÙëOEµÏÔƒ”~°{öª öÔO÷úŠ³Öš’¾<ⲦŽSgs@’ŠjnoÊ žH¼Ÿ½Ñú›¬ fÍPÖl=okô•¼>óäÖÍßÒÛâQ @š QúL­á|aúz¹jÕ<ãúöt&„,;C°Dàš˜1£J ½ß‡E&ŽŽ‰åŠ#q¸lí:˜›Þ“0½2ÛnË;N¤vѺÜn2­oèKÀ¾ó(m±~¥ü&mzŽ<»eWq\7‰&†i1_3”<´’3ð9 Íš>sö°t&‹=·ÕÌ.®Li9õf9hÂ%~ÒÌŠ%aÒmûŽáCòÇXÚèÌÈäÇ*U‚ôI‘®¥V]Çr({ibƒÞè~pƤl›Õ㻬2 ÂÌt¹âîþéH«Âv[åÂZéø¹wøW4¤:32£)A'Tݾ óÂò ?^<=»QœfIÖWµ}Ò XHƇÀCÊ-W!H3`/Ò?ÉYžòd;€"Y¿û <1[NJâñQ/Ró~Èá*qÚÍO oν^Ü92ý¢y2”ÖMîRtÄàÂe†-ª<ƒD§(|_ ºõ<½d&íKÖAWÀ¯Ê •²ä–¾^õ-r‡Í·û†yhAöK#o‰œÍ²±ånµŠ;BR޳%PÉ 9*bpKš„D£>f<ä­‚ m΄Ì3úžÀ‚2” 7äæ&­™|ô¤þæˆuÝ·g=6S7­ç™V㲑ÂJHÒh³…W&\2‰Wöøô4Y0‚¦š$g_öŸk¹" ï:"ÙQíB4ô=…6<3±{OWŒ-½ŸÑἕ† Vq´[ãq†æøqc¯B_\Î/toó¼¿=YÁhÈÉÚ½(CÀó†_tä˜26¦¦è‡TNÎôY¯’ª+¦´=ñÀJ÷gžªÒ#d(Àg£HéÝë3|7&+›À±ÕãVhÎSÅ8™)—_½W—Ø~Ïd%Aù±ÙdÞëÆÊS÷Ù„ý„aT3Ø–Uˆú(ôiËhl;J¸½ó$Ý¡J¯QÒ¸%˜iœ‚ÛÎÓžtø²Æ©£¥Å©iµ2cÎ<+‘j¹ê›Ò´ÊÎV«4þYREXm`öSæQ‰õ“…³K¢=ÐÃÚ2oáºløé/ŽÑ(äõ•r* õ¥ù2•€þ<Žf¸@&éÕõ—x°ãøTwºA×;7•þ"6Ä‘´›!ùCdª©¹’Ë1¸^µa¯Òßz:LP~®ÇyZ¾ŸR™b…T]ÕŒrŽˆ&Õ(>0Ÿ†®ý´043`D¸ëU„ŒN‡ZO]ûÅgá°ì/ÈÈÊŠ„i„#Ó¢d z|öÖU ÖBïe)öžkn{[Þòn‹¤i6§¨¦]!ðØ¾­¬Ò®ç¯ƒ8§’µS\ćIÌÏ-Îm!øÂf¶¨¶¢‰´­:F2>YJÿà¾_EãôB"·×ÊÃ0ˆýÉD€}4x=ŸºdXôMÔ;1å˜é©ž_ƹ¸tU(\¨çFˆØGdzû‡u½#& ÷Ï ]YäþH<нy)%¶¡—n3úxÇä® þÁX|I4ʬ.Æ<ï™í Ç={nñÆèM¨»&¸2€ÿµj½_ï ïíTo±ó‘®WówÆÉrˆ(íÑïm·à•H¾âØÄ}=3œæÙ®ærnvhܬ?‘3[{Ö±JhYE¼•m ²`:ÛîÅÔÁŠ"ž«g—M£ÈH4µmª`3ËQÆ6YË‹Š©%n}¹ðëÄÔuµ±4(‹±¶Ú)} Ž)=üs€¥ÙÊ™2:wU³¦ “?@)JÔjc‡Õô#ïУ5Ǩˆñp™›îÓo ]hÖ¸à1·{BµTÏÆ‘­eÛ¹•Ï5‘W“¼>"ê¡…"7ñÂ_ªÁa¦Áå|cœúéKÑ6!=ù¦ÆNw>£ª½Éõ¦4Ý>‘s­é8ìvé?ë•ׯ¶×[xGOœIûòÆ@öy\»¦&XÊ€¸wK£©ñ5Q}·¹gÂ!TJ­ïIÝRð Ò–^8f °jÚû»OR/óZ$Ç.•=.Øw²= 3ô/æ½Ú^êøp íÒ{¦ÉÅWÄmaCà†¾]f¤lÛÅÓÆ~Nª§_­j2¯§÷ì¢BS€ûÉN[Q ©ÇçÍѾTên'ZwfT¨o0÷"“ÀR©Ðü¸@†vD<¨pç:O<'Ÿ|gBçò¬‚Ÿ©Œhò$Ð×H“©ŸÉ» MŸu¨8E³5?·w}{4i·KΠª¡å–€Fªm„q™þÚÔˆD‘S¼éNa‹¢É÷\;E®ˆ!`›îë`,Â7“Ã"éò~c²ö—™fÃç=&æ`*é•ÄKØñŽ`L(•.UÙPÒÆ+zç9Ì¡”ì¹,–Ï,f™.HIâê…º¡cޱ„±í-¬ï¾îìÓ¡;®“R]¬Ú%"[Þe¼sÐùÆ™µë£÷À%–ÚafáâÓ\lŽJu/ÿV ‘b*s<°ÑCô¢/eˆ¢AìJ»éÕ‘¸3o·þ,‰;fTUiY:럗íx¥äzÿIüн²¯Šh 3XД!ZS÷ØÈ)†¸yÃK‹¨Ã¿’t€¤AKGžã©øG}/X:c 󺪹‚¢·û<÷ÏÝyU¨a²Ûø]~ta¾Í€¥¯YT"èé—µAÅÉÀ5Ò$½rv+àn[Þ ìÓ¡èsÒ‚Ì€§›5ßEƒÁï÷¼d͵¶ÞTÐ1+ÇK2!úè;’¦¢1¿;e9,ê¶yÖà‚9¾ñˆ>ñn¬‡Š|}Í-™u\9ªjFxÁ6,“U3‰í½+A=¤C‡=Ýò$HÇH$®(ü­XÑ~J?~ª€¹þûm!D>’š­OßAän‚C0³5¥7ÿ™QÎ0—ô¥ù³ú¨¬–ksÇ­21r §ŒÝËáR9ŒpÝ3uYB{¾4½áalBÀ>2ºš%N˜ÔÑU¬<‰çõZ#·•D]1M̪´JÝùÏ¡íeSâûÅH‘Qd!èm£¬sÊö7”;µ"‚Š9­eKÜÉÌj3ݬÙ&Ì~{›ï@ëFô¥]LÜGªtS¸Â–,HŠTE=ÇålôDnÎÌègʯ‚ÄŠÝdìÈU}$1,ÅÍÌûy&ZÁJe魯,~ðÌÉ&9™?€$X{<ò?²˜kºò é¬KŸEÔq ÷ãäÍßzâøù9¾]¯¿¹²³>—\¯¸¾º4„ÌaèÊà1Få§•‹ÄE” ‡%~köÍØ‰•êDÑq4hø”dÊO{ï"ç±¼EwËL±bwZ= ÛÑÅU6lnÎè¢TŽH»å“Ž:u)gK÷v(²!Ya 9 È­q¦Ñzzìæ,¢ÙÑã~Šq¡¼æ=±_“4`Ï8˜ á,?º[ôWÌJÅD‹}›ÃbË ™/€ iÁW yk"÷NÜžâÐ;í"cص¹%Ó/õæ3“B q!js둦ˊ ôm!gÙÚŒ¡\TBPÖV§€™ÀDQ¼,É,u‹“J1qZ5+(ºQ¯¦U‹ _†:®zxüÁDáÖÎ~;ýî¤^Ë7Ø´#ekß\x¤YêœZU’ü-çdØšÅ~{a¶M)ü;¤‚`Å"Cš—P"QN:•ÓÙ]’íá=¢Ž23ÉwǃÚôlIºé¥ÄŸÆá>š„RhHy{C¦Íá>ñV°Ž³„+6â>íƒoý·#ƒ|6ºV9”+G“Eªx3ôù‰¨†»QmÓREÉCWå©w2—層‘øJ>þLv/¸‘›s¿Pš¦ZZXyJ­ñý8âp2é-FDÑXX_1˜+é\uÜíã —чF#|¸‡ÌO<Þez|hk!ã,çAn‘ß+_Ó ÕY²á› Ë¹|–·•|¨­TdÜ——]ŠÁØL5¯S§Þ†ã|õ ӓްziñDüµ™Ê“óÍ\• |QTjåÊUhù›mõEÊ>µqƒX-GÜÎÒYƒ[ºÐBèDþ˜½¶ãñ…ãûOjòÏáÆ_kÍ1êàd5#2ˆcœô §º?Ä´ Bç2­F8£{¯Oíg?ÇÅ´÷lxïãîüëb%ºW$—O#|ÁiÍÙ`f¦Øçþ³.³»‘™æ°µƒoSå ¿|qˆ2mIVi}òÅi£Ð§;W'Œ3àxTœ¼"YÛ¹H~©ƒ5©Ç g û$Îßðî%—‘ã=Bui8KÎçîð·!"?]{;_gÌS©˜; vÑÁŽãx©ƒ6Õz1²&fV§É;á\,O>¶ò _]ìúÆÒ¯ëù µÈ×FãRº)ÛEåô‚oò¸Ôîµ×8ìcĤ)³=´tpch—Ìä9zø°§}¡ 1Ó/ämwšZ¢œ%ZáHM+ÏA¸e0ç¦õ6ÒžÍ5lµ;ž}ÅÊ Ã¥æ”/’^óÅñVMFÅH«Ÿ­¡·®”$S¼—yûû‹ ãEþ‹Æ(wƒ~ð{¸ó¨ã¤h<Œ6O˜)C±–÷–u)èÛïtŽ‹Y Òƒ—)&‚xrbv}c’ذbÞªK¯“Ó_¯±«ó&F}âãq p®rëgw«›ñc°S}–vÄ·ùíÉÃÉŠ YrŸ­ÅnV¾Pû5¿ m›#¼ôÕœ×ó3ª*LzœTõ‰Ê¥b‚±²23¸—y—õ¡òN©8Ž˜±®?©"m˜EÄg‹dtÒÞå·¿ÄQ¦´6 hÉá2›_8ŠÅfŽ÷¯Tto†,RŽ È•lt¯ÖÏ‚>²GoE(È@²ò|Fƒkθì'.*H«!‚·pom1{Å ó‰ß2«T‘j-§ÆIRúÄ™–b’¦/Ÿ˜Ö\çQnø¬š•`=Qs&Ž%þ F’k¦ÐÆ`pCþ´vàY}|¬ì%c¼Nò©á†Ó{m‹uÒÙü­™/gÏ6u`Áò‡®lIÙbŒ1[iH§±Òïf°dy­—$¦¤ \KfQ«‹ºÖïBhjݸ‚¦°mõ|]…!{¬º+ÄK7K"ñö]äu&Ò¤ùSëè'Çè_k©ÂÜŠ®¶»­b¬çè[4_U{6íÅÈȯÍOX9LÊ«åôÐ\¸—ñ¿27JöœÈUØÉ~ŸÃù5V+§­ðŒj%˜“97¾ªEÁGÜoˆeÃGøetGö+âÁoFVûq¦±å„ÈE½Ôž3nÞ¡[gcÜìih['…ÒcpE¼y`SC­yA)ýö {¶4‘GÀ ßȽ˜Øö ‡ÆÉÓØVYî/KòšZN~üt¨¤ë›ÙJ¸á( )Js¡~_…\¯ÒQ!îP笫wfusëÎqjÐðÆŒ3¾ùŸªÀ˜›òÊÈd{fLf»Ì»êšµ¼íË Ã@9õê9Sm™ š*u%y_ê´(úAÚN1èžO.ÒÜá¾6ȰQŸ2L´ qq·3­F†}m_žÖ"ÊC^b™Ý$Ô)ˆòp0ú›3âãÃ*m¹hʱz㢪OøÍ™.ú­¸ë X$`ã†þ¹,tø=Î@ Š„±°uM¼L‰qäíqÖb65n]x‰Y|ðÕù¦éNÏZPjÀPF=q•*©‘qü<ÿÃÓÅàµÅ»(õøb¦ŸB—öâ›cnß:¤ª½$¹Æ¼Æ²[+c$óÙ¯†ïñtÁ šýXMþOAÕ&Rh·$ãmb‚¤Ô†ÁãYá<43Á/ˆ£ëŽ2nÅæÞ+½JVB­ó™¾í—Åã ;ýÊsC¦ÞðqX\~ig†Ù!Ð*ÉP|ôÔºd…f›Úâ0æÉjÔK$Ö}i+Z¯ 9ãýjÀ÷”e ïgüyöÇí:ØÇdl¼ïcõŸøwè˜Þk±æ„™¹0ƒ˜\ àÜUîŽöÉsvéîÒ•šmª¾hYírù¯êþ|7¶yþ?š$W…Ø¥ƒ4ÔPY`¤â(õÂסêêÌM EƒEÑß-.=Z¥Ä^Õ¡1ͬÈ5©¬Î­,8B+XD+ó¸T9ï®9æÄ&Ÿ¬z+MÎqñÙ³0=Š?¯ñpA“ÜU™AÉïu·¬_ì¹-T*ÞõÆ™ —Çø™¤¦Î¸éš‚% ë¹G¹&¾$ó R*ødå¶èU.ªcµ!”¯7äÖøÏ½[™¸9TùCÉ(æs+Â<æå¢?Úç쨞9$Nd‰ÝmޤzpfkömQù¨LGæbæk6éÞÓà×ûuò1çºÒï/!å~Ž’ð™Kªf\ƒYÆÇ)ЬÏ>ç´Z-ú"pB~Öì´›Éh-=£;4šz3ù0slwUBß;<|PÑ[úðó¤“ÂmàÀIeBˆñÍ,ñæpÔ¦i«Åÿn§ÐÙ endstream endobj 195 0 obj << /Type /FontDescriptor /FontName /DOQUVF+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/A/C/S/a/d/e/i/j/l/r/s/t/u/v) /FontFile 194 0 R >> endobj 196 0 obj << /Length1 1379 /Length2 5945 /Length3 0 /Length 6881 /Filter /FlateDecode >> stream xÚWTì»§S@¤QaÄèØèFº»”Æ0 Æ€nAB@B–‘ ƒé–4èP@âN¿ºßÿÞsî=;g{ß§Ïó{Þsä06VqB:Â4‘´0X$P30•€@â" hG»ÃþR-a(o8!÷ßôj(“©CÐ83$ ëã‹ÀRr`i9 ÉþeˆDÉÔ!¾p'€@‰€ySÕžXÜÅËò×Àå€ee¥…~»T<`(8‚@Ю0\F(Ä`†„Âahì¿Bð)¸¢Ñžr¢¢~~~"o$ÊE‰_àG»LaÞ0”/Ì ð .Àâû L„0w…{ÿ!6C:£ý ('p‡Caoœƒ †àrÌtôFž0ÄÆúþl ,þ;ÜŸÞ¿Á¿!P(ÒÂÀÂ.g¸; `¤©/‚Æ …„Ó/Cˆ»7çñ…ÀÝ!Ž8ƒß…Cš*&ߟ輡(¸'Ú[Äîþ ¡è¯0¸&k œÔ0Ú›âW}êp Šë:Vô÷XÝH?DÀgg8ÂÉù'OQ Ü˦£þ§NDñ̆H‚@ iYÌ Ã@]E7ÇzÂ~+Á¿Ä¸úƒ<‘žgX܆û¡ð†øÂh”,(à¿+þ}£ƒNp(às#(þ‰ŽÜÿ¸ã&‚c6 ñÀЯÏß';·œwì?濇+j¥cª®¦/øðß*UU$ ,‹I‚`°¬8@wúw”¿ñÿ…ý·Ôÿ³6Ð?uÎH€ìp½û †ïŸœàûs]øÿÎ`ˆÄñàû‡ö¶ I÷þ“ÿ·ËÿÆù_QþÚÿg=š>î|¿ÔÿC ñ€»cÿÔãXìƒÆm„·ˆÿ4µ‚ý±Ä0'¸ÇjuÐÜf¨ \Üÿn"Ü[Ž9ÃÑP×?ô× páÝá˜1Òþ뙃A ÿÐáv ê†{J¼q“ú­‚áVéß)5P¤Ó¯“”@P(–7xÜMÆ-§ ó›×Qsààœ‘(Š_ó”ˆâ«_BІú P¸eû=x\Ö¿î¿7ÃÀ 3“H¨|ä½úȶãZ•ë~Â˃ŠcÀe«ÇüÂ3¨W>§Ô¤iü5Ùየï*i}¯¯Î}Ñà;ºõ‘ý<`³¥ô~ë#“—?Ïì“MG–_RL3vmª<w“ü†°ù­•Às¯@Ë07Âü]`¾— µq!ݱß[-Ìówå³¢'—MVj¤ô(ÏÊG…,âmÃJÇŽ9Ìœ$há›d×ö04ãGßÇ®å ]²ë& Rm%ˆX/‰%žLøÏWš‹yw²p³X3ß$<ºöa„'@u-]—i*àYñ"êCÆã«O©`÷5>ÿeCÓcÔú·»<À!9>fQBz¸øÅVíRމœ½9ÝEŒÎ7éGo!£ø¥‘»t=QåX ÇŠ#ýÆ»›‹+# ƒÂJÔâ¥hy-Òx…ey÷}¹7äÇÅFfiW ?™ )¿‘mËZœù´”qX{-ï8ž}ä†ðÅ}âŸe$Ç~¬ ³ ž¾÷ôd³F*¥—v­'¨H©ÿ>uÏ2Åš×ÒÆìƒ–T¯¤ÚhzòJZku;é÷­-FnÂZöó¶ŸÍúéj±ZÙGú¹‡ç©YÑI†¬2§.O¶ÄLH‚äÖß]] âó4K/Õ‰Éh&ÔzÊ$Ù^d.{èGX°¸1{hW\Ñôš5w~oͽv¸W5˜9·Ò,öãþ±Úè¾Ï¦=ÅEkM4ŸDn‹Ç²Ç­Û§a²²‹=[]ˆ¸<W*óÄò~eqy>/2{6ÆÚSr›¨_6øÁ—ÁG¥ÏJŽ£¼f¹ Ÿ±æ‚À)zu3` Vñ"ÚÁaf~øqnm¹üYLô÷•ªgIÚBÑ«i‡£ô„{k…}:—œ}ëŽóÍk åa$‹yy!!yá^i1¢Á$÷xm4Àóä×ûü\„zõªBS©‘“J2 òµ¨2ÀµÛô„7Ù’î[Òº1лL ‹FVn˜ÞeŸ tßwÄS mò¯Õ+`³'*¯&€ ñN‚‡Lö3¥ï>oÕh9+ßÝ ‰»)ËMØiýn kçŽfKD]ͧUÌcÎÊz¯OG¯ ü0 ¹ ~e»Ã‡¦ùÂ(¢ÁæÿЊ/†+ÜÀðCr£ÔÇ`) ùÉŽËaµn–ÂÈ]ð¬dé|¯k Ö™¢Î›öêG™°¡ZÉêN(œV‹[²ß´«gî~Aèy6ukFÉŽåÒùó Õèj䬴ÉC-ËÍÔ¥JDÑÕàþ8·¸“ƒyoÑé0ί*½SbbÇêð"¹Œ¨ÛÀc˜äãQ×§Û~ër{+ÏMå£ÑUYÓÌÙí\’k7è¬Çéû¢.¨Î‘Þ~åÖL,Âø³›dâª{!>«Ü…ôø*aºÞök‡·Ù#ÞÏã…eƱì²_Î…Ä™¡³Þkå<®¼2ª×Æ,#ÖîœS¥–; i1x€}”Ã#˜&$byÒí6W?>®q¸Ô óôØ #m¶¸VCãÚœ?VŸ×ÅÂíWl[£ëD*#Øi–¯¡XÚy;öÑÔo™ ›U­™ÌÓzo#ÿF}ˆv{ô!IrÐF„B þ- çe’:Ð å6‚üy›‡$¹½§Jô€!Iáîx«Q„¼¦aQA“©ZæªÞ©0óѽ¶£Œ9ÿAÒ—M#Ãls€ìÍ­3×/ñÞ‚KX,„³kØilßžóÝèRT·qâ¿u¶$B‡¥ós=›I‰5˜ÉPQ<‘4ä~Æ 6Ép«¢q­Xy{-ðFMÈ+~ë^èîÃT[œbPRÚTç]‚~E¼0Ÿqgbí*OŒÈa')§7ÇɤÙ3Ïéz_C<²Þ ¬LZðw`Vîý¨aFzûmqu™žæÔzmßs×8?+û®/R¯mS)-{ìnÜ©hS¶QTµ¾Ûà ªgšÚSÜ™ .',ˆuÄ·¼l™­E¾O—l}%žîðèì(¥_™xÜ`”sJËPÛÉv›ÃæKøÌt>Q¢ÃIŸ/‘ó·À O¦§IäœqŒ$Áü´$> Û¾W5¸-*2Ûmã^XÉoÆç²Žêª®H½äž,.]‰Kª2vS~7€}‰½Œ‘ˆj¥S þ>¿°×Ÿâ¥¨¦Å ØcJo¾å4ÓUGÜ|Ч6‘ W\Œu;øWÕÝÃ4$¥Ï–œYÛ^-éAí–]ÉþÚãÝ‘ÉùÂqƒÉΛ™™ÉƒêU]uÕÏ2›àߨô—2XQüœAwÁÉÛbd´–~Ñë?y€˜`÷ձàZ…©¥—)ßÞ[9|¯?CÞ®WäÝKÊšœw-{ó-­ûá̦š³Fv˜ä©¦Ýƒá8l:Â5ðÆ'2&R’aQ¾Ž`ž‹ÕNãÚbUÙj—˜ÐxYŠ“… ò‡@æO?„dóHñi‡ŒÈô¶Æ¶õ™ÏùÞ ×ЧWMŒ:Œ¿!$ÕJÀ)¨ÞÝ|€rÕW°#U”ZÑ}­ûBvuy¶#i‡!DìÇ=Qw±Š§¯¿ntH²v§Õuº\Zâ§ô3- Ó/dËèäê Am³·ŒØÏrJ£Ç’‚‚ÌÓ7~0Ìÿ¤“͹¡>Ê·Þ%“Ý-xì ›b‰ÊñÛ6½/#=ÏQ?X‹LöNiavD§^ V-4,2f?îÚÓ¾íÒ;@Ê¥ú8ÈÀ–,s]”ÿœ—Ú’ºuxæRÀoOþØÂùóòm®-]ÞR 0@øX°àÔ¡BfþÓ-ù7NÖ!û¬½l×®cò¬x™—ö\Tð“—¸ };½†toUPŒ±ôö~¥Ãë¥í¡*[<»¤²G¶.¨õ ÒÏ=’ ¸Î)L)¢Öï< ihË®áÎߪ Ý>$ºÞX"Õ’Ⱦœ™ÔûPë!½NF“Ã[g7¯ë [èfbkžÞä׊_ÉÄ‹#ljtJ]nXäJ’EÀB«z»ÏËT^Õàͺ&tVú•QÎBÖ Ýsµ–sïX¯ ¿¢ ™%=N$*(sPÓa[°Ùe)ò<û!ÒÖív) ¿-ÿM5‘൧ݬR~Ūpµ.O£'”u×h4ßwebÔíòؘ^ ’ß±Mc3oönÝ·±ž O-ýÕžé©||àØ$%›¦úCó™³V¢4d‡©Ë’äÎä\ÊÏMô ª×„1ß B|¡JbŒV\ÏÄ+A«›(î*ù%•£‡ï“áo§FíœÏ„(è"äøòj›­ »eð¦[»¶G96u$‡sG ±è´Ñ%ŵ ƒ}„ª=C¯Xº=aõ-„ÏceýQ!u+ úñã˜f<EŸ 4.Æ4©›cª›˜»Ä&Ö?ôä7c˜†L4¡¦Úä6w°­ exOâH|~ÝXº¬;ÙëéËæVjþñÔ¹© ð4ÞÎ{ýGlÝGfçãM€‹ÖÂ^Žqgœ˜².°ÑXûåâ<£pËÝ-§Ä¬aôþŽ®î®.ñeQŠuâ£òõKUma¼`lÛ(‹ç㣕ø_ò#c–&kœ¶©hýE7ŠNÅ‹–î1$ tòÝÔÅ7~òŽ›·¦¨Jå«ùqcÓõþr gã»CÇÇÁv·©'ZWRÚÜ|,[-k®ðj<ïý½ §Q½y²Œ6:O•"X#$©M} <|;åŒßÇ@så¾9´õûùyò{ëâ碠—ö@‘v"Ç<˜Uþmy¡Ì ³¥Ž.Ãö´â:r+síSÀìËÎõ¦µì 2’?„•ñ^>}µÿ‚ôølùIAÓt”åÀíý/Š”¶£>%­ªi¤ï¶,øÔ|» ý”µ§Èï¡/ƒñÔóŠ˜1èíÓ±ˆÖÓÆ)6½¼Š1Ñ„úȱ¦pþêÁ¥†¦Lš<ê‘÷u}˜¨IM?ʨ“¤t´]j‘“¼÷ §õŽ+×ßÚºW}ÐOq…lž9ŠR—ë¹mXf],¨“ûäSKq ή3Ë"I‰+w^ÀŸ3^ãsÁJº›ÅíßoþÐÞCJiýjÝ$–*¾ ”óÂéÈ3ÎÊEX?ɱ8ÏóÇO¬í¬? Ÿ‹i©Ùl·g°³R/áÅ…S‰¢¬…Ø|éʧÆç-jµç%lIÚ”Ÿ»ž‰(Þ“Dšf›}‚T¯úúç¸{[éX([\ågȇuð ‚IÃÓ‹ÂåØêǽ¯ª˜5*N4›±HyÐNjЛ{Ê ³£³ÆÃJ¬œý\=iݧµTô‰æ’º¬Ç!ËL'fL·J V¤¹dx&·„KJOò±aˆÓ·²OOJ6ÕjEÖ™/LŒ÷EZ†^£øQ‘±-h…WgfÃj´á!Îlìá».D[!˜j:ÌÜu^ºo¾UívŒ=ºjªÀMÉ0ðu>”ü“Ø%„Œgô{3c^Ø´b±Úxþ÷ÝÚO†ƒT²ŒÒ\åõŸƒßwÄeÏ™×u*7¨ú3|›2hÉ/4¡¯÷¯(›²»FçWn¾¹w8f|Ë82“š;ÕÛxcªk˜¿d|ó5}ÑY< Ôw³ÂsÂéã KŽ,‡Öêfs‹†TzU©Üy^ßêãyíý:"º5Ú^¾•4³²‰þן}ÙysŽÜv›ÈŒÿN8#›Í9«2‘ãúˆá^ I`ç¸ñŒGwÐëNÑŸš%.iN‹Š7ü‘æ¶SÖ˜Mm*2)6{;oäm?DôÄwÿU˜º¸›EÚ¡Mçè” «ÐfEýä6ff¡~B2#Êm àâ}nb:mÞ³wuj$´¢×.ã’‰®ÄŸvM ú*¶ì­Gg³«Å”ggc¤îP 4¯>$|„Zè”!Ò % ©à•V#³}ÕF'R2©Í*;+ K'tÙ<ÿ[öB^c™Ñ¥’AŒXWË]Ã{!;Ô]6†·Òò¬_|¿¹"Ïåê:§'½Ê%&_£6ßXþ³NŒÊOíÚÚÄdÎ1ön›¬z…n¯×€cÈëXëÈKi†vñ±ÞWùܸÄáTl¿ohš¬”¿'97GòÛ Ñg'/dÚ)|Æ…XÓ®Ú ³§Ô4 ì'ô,r´J†»`m‚á+|jáÏ}Tj_{± »¯™C}lÒ)ÓîM*>ÔÈ"ÙŸÈc;V“ö4„TÖ2ˆ :¡LvZG éEå»’ó“>&Ÿ©ýþz+$ð’ù®c7üS§—à†QP+¾ä§ªp”ý£Ú}wµ‘™'«4¡¬}Ä S†4\%=I‰Ï%¨S•|… öÁ«žž±ö_j·ª@{pãЋ´´Ýøð¶ÊýûIÍ.ÅíÖà˜‡r¡QÁ+p¾÷ø_ݼª!5•{"`š¯ÙÓ ¼ròü“ vì{B›RL>èªþŒJ$bæ.GÙ¡>5^}l g¾c.û“Ï}ÙÁŠÅCjÃâ‘6í°Ì9þÜ=“Ss£i* “ˆKöÜTì_±ƒ@çO¥ŽJÞÅû‹’1Tl®T!‘–¡óBkG¾(q󛉿c‹—¯ÄŒÚgÏwsì6#ÇÔ·ÉK/ñ\XcëÅÙèÞ‡Ó)|Z†‹lìùi³ÊÕŽ r·‘‰©ó…5—(VTíTŒ±é!+sܸsz¯•vÈÇ<pYiRòг–ýàýù|–™P":¨Ðe­ÜÇÔÇuÀÝ?Ê¿¶?ô„½°L Y^I‰÷üç‘uèŸübK¯žYvŒvž1[ös:ñ ÌÐmÏ »R»ÿˆîÌ©ö:t÷Ld‚N‚ãqÇM†srº¼5z¢SM!Zw_!³^ø\€}j‘gÅßuå)MS–z›nÌ<ƒHÒ÷÷PVõëýÏ‘-—Öð«ƒu?½sÍú™o£ÀfÞ «D~L0çä #VÙx5ý l% µˆ¦´Ü¶e³Ë{š–ƒß^¬ü2ÇàÆ$æê Û±Ïò#Ÿ¨jÙX–ù_××Ï*¾]¬ÉŒlèó&Ü›ý>[¸sÑMôD©ÈÓì<äª%×£.9¥WÎCÔœöª9HD*‡Ÿa{åVTMòéÊQù#ÞC韃·ž* FYJÆM° m^Ÿ®®* P’j}ÿ}yç;WÈÛ»U’©Öž€&”Мj ÓÝÏr„†ït‰è;_‚¤U†,–¢’'Jƒ4$òûüUªÖ¨­¢>|:LjÏxZ·þðTì ÈC†¼é Xåâ-?zîc@»–h^·ôíMÚ;"›p—€7Nú*†)Ý!äAŽ„7`[Ñ¢šÊT*ÍvŽ>9“¾}]aE^ó’lĆ$ TÍ‹)¦å˜)¨™±»´ù) q¿ú‰Yg3Þ Ïï­ ®Ç#§M9óè» 0¹ãôëÎæ¦@ž[«àŒÊ3Š3Îò”sè!ëV” #íP§[†ðíÕû/-LÒžuU\Açjiq_¥!³`zTùbjpBº\²/açt80nÏ>¼B«„šÚAyIDó „AÀË—ó|´„1&«ª&¼h‘.÷ÊI²µ˜I]ÒH“B €‚^BÔõ„Krž½ÖX¡íˆÈ™ˆMLhù¼¢ég0/L|°›¢X1n®¨aîÉmý0.õñœXSþšV§z5íÉÛ* u¬ÒG‘õ+[w{šfØoþ–Ó?¿ endstream endobj 197 0 obj << /Type /FontDescriptor /FontName /WIRDCL+CMR6 /Flags 4 /FontBBox [-20 -250 1193 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 83 /XHeight 431 /CharSet (/one) /FontFile 196 0 R >> endobj 198 0 obj << /Length1 1397 /Length2 6127 /Length3 0 /Length 7073 /Filter /FlateDecode >> stream xÚvTlû7%1º[&¡ô6Rî–”cÀ€m0FŽ$EiARZ¥ADJBBI)I”ÿÔçyÞÿó~ß9ßwvÎv_}ýîëwÝgWŒÍÄTÑNp-4 +ËÕ Me`°¤8,0G`=á´[pŒ’ÿ_vu ŠÅë4 X¼›!ÔóõB$yˆ¬< ”ƒåþvDcäP?„3ÐP¨‡FÁ}êh¯@ ÂÕ ‹¯ò÷(BäädE‡U‘p E ¡X78_õš¡a86ð_)ܰX/yÈßß_ŠôGc\•„Dþ¬ÐîÇøÁ¿à Høo`â ¹ÂçÚ í‚õ‡bà@¼Âƒ£|ð¾(g8ˆ¯ 4Ó5Þô‚£þ8üqþu5@ˆ8äŸtEÿJ„@ý†Â`h¤ˆ@¹]žpàM-qlVE9ÿr„zú ññP?(Âê„wøÝ8¨¥j„âñý…ΆAxa}Ä}ž¿‚~¥Á_²&ÊYDÂQXÀ¯þ48 ë ßcõ@¡ýQ¸?gÊÙåg_/ áí ×ÕøË¯üGç Ç¥Á`°¬÷Â`n _Éͽà¿_j|ÿ!8/´Ð‚pã8¨ˆÅøÂCpÿÛðo 0,Ð îŠ@þ“¯†»ü‘ñ“Ç €·ÁxâA€à_ŸNvxn9£Qžÿqÿ=\¾¶‰ÈoÀÿ˜ÔÔÐ@œ˜„,PLB „@$$€²øCÈ¿³üƒÿo쿵ÆPÄ_½ÿ“Qå‚Êý€¿»¿aøýÅ Á¿ÖEøï Fh<á@ÁÿÐÞ, †á¿ ÿßäÿòãü¯,ÿÚÿw?Z¾žž¿­‚¿Ìÿ‡ŠDxþedzØ‹ßC4~/Pÿíj ÿ³Ä†pg„/ò¿­ºX(~3TQ®žÿ\"ÂG w6F`anô÷ ðé=(¸1Úñ뙊AÀàÿ²áw æJ|ð“úm‚ãWéß%5Q0´ó¯“–B1h ?x¼$ ÄAðËé øÍk H…ÆâC€xx!@4ðkžø ÐHxõo”„¾~Ëÿ*óÅ`ðë÷› ø>þ–ï:‡¦'ѰwÝ_Ý}}T¥Êé/¶2¤8&°b™!$†›Æ´øžÐ¥ ½ÈŽ˜Çª¦öwÒ}\Ö÷¾uǃ¸‘°MO ÏÛ÷:qã‘·v@uOùì»èÉ“Õ2ú”gåïÅî[$ØÞ)Èwz2ÁÆKŠã&fØ  ?8cȾàÑK„lÝ—,ÂÙ,H<8žúôÌ\§ŸÝ†›ø€áÝèUœÚzšë\iÑ\ḭoqœäÇv;íÐÉ% Y…@DÒõ—Ê}ÄWV]lµQ6  Dµc+gÒž®ž‹V„UzI¯®rûÂèvp“Üå u¯xÛΞʪÕd2/ÉÄ…ëñ]áPT<¸–³‡¹7ãÇ£4ã½ÏžVy%DÍcB­EcXÚãB^æÎGÜÌὂ(ëÈè‡K3*QõžQ8'‹êÙWÔFq²“ˆî ÛEØ‚ÖXY9Þ sR|«‰õp+#r¿é¿ÞòÑ{‹ˆbÏãªhòªƒrx滟ý±°Ë¼É(åÃâ¡ ËØd¡œä´‰{<ŒíCÜÄÅ÷=ºÒVãS·ïÛ<¯Oéí?4c ¦LTN­% ƒ“}NY±ìqE·&í$Op¼­|åÔ¸î÷9j­2¡EœÜío’ÒÊO:fôkx{ø8ܘêÖÎJ)ËTb»Â#©ÁÑ}wbï³c{Vm‡f……š|çK­å¥»š‘˽ŽÙö†Hã~ó„)Ò¦â÷PþÝ ºçÔý¾ÎLÑS’LÈÓÕØmÌ\}¥¨ÖsJ:Š›çGŽ~æ&@¦8ß/å[-z¤Ê²Žéf5* pž­‡Œ…¦®Í®˜g^•…2óë ]x/<2Öå«3dH^¿à1DÓг¥2÷z|6›Ñf^OÆ-|÷ùÒŒ¢›êLÕüÐ;Ryº6¾D´Ó¤_ÓÞj{γO›¿MØ™ßà Nî_¿t3~¡„ëá(p!Pœ?ªs†Äù¤ýø® ò¡]ÅyH#4Ð ÓšªQyÓ&±T³_|8ñ£âf² €}i HâV{«PC+Ç“/ï{z? Îïm#w›Ø]šØÉ.1p™>³9ÎÅÕ› ÷A„‡ŸMfÜgI,–D¿^¬Ìˆ°ãDŒÄOÎkÃ{”†cc¯)æÝ©ïÕÕ;åê¸Ø½‹QçõñBETåID@(p’'Ššg3î•›DRã6Fo%/ôC…ýï4)ÿ$Ügyº¾Ã—V“M"WÎqcPC¢zIè±q ¥llÈGûZ¤/§ÖNöT¯ÄI|~ʳª€P¤L'©Â-{}Obž¡ [T>åÕ,ºN«Mw”¿Úó¹»¾»U-Ö½…fíu|—Y†c¢­=-ÇÅãS‘"»~hÕ5cu÷”m2U¸@ΚŸÝé´¼îÃÐHjÆ. mî‡M.U×.g´]"Šåô»w°+ï²í’%7~RfZ„Œ OóU(¾ß¦ÑþdPØ'-÷êq]«ºOvšwÊû¶ 8ÃèÜCAï`—3 Oا:Äκ qq8 í%Ë\m·ãØÅ,ǽJ=Rú\AË#u•\»GSd,“UJ“Ùg ÷ï,fKúØ•¡&‚WK„­ÍæäIFuø²Ò®Ô<“ã|Ù —» FT/“ͺ‡à¼%-íÕÊ®=ù$xÎÈf]DÐç†OZv›®?4[T&ðe8&˜é¦:'ÚJ·éᩪK}ˆHBÕ,O¼Uœni˜ÉëÍÚLÆÕú¼P\GÚF܉Hmó¥÷´×v3êã™"U9Ý\äJOô¡ i>‘aNžæzIâtbsçƒðòºkÄåÅ=2+ÊŸW6Qmªy¸ìœªLÑæLH,.Kó0vK|Ò4i$ÅÞä1¿*ÈDàϘÕè) 6<¦+'nË]]´Õ$ˆg¯Æ.Ö‰jæÉ3¼Î·±ÛŸ(8æÑÛ´JÇB\ÉO¨Cú‹z³ÃöŽrÀmÌ ñ´ì¨.}›æXËáeœÝû.='k·¾TN$W̹ËÞ‚:@ûÛ¿ðŠáø¶Ç¶YWK8Åîœ0å׸œn¦ÏÞþ*D¿£óü5Ù¸ªÑŒæfsJ÷EÉ ×\Zïö®jµŠÕÉ·ÂîÏÆšèÝh&u®Qñ_jU™‘GŠ‹Gƒ¼Óʽ0ØÕö$ø¼o@oîÕkýÆKoùŠ=Ö6gòD~gžÀ=yc¹t^~â8T‘“G·>`+-/ge1 t'2Íšwå³Û%²ô®ÍÞµi(MZnÜ1€ËS'~²¾È›G¾ùÊO¤w©·Ï6s—<˜˜ºVžj7ÒB{ó¨˜cÖ_þÈ‹üÖé÷Ó`à‹à3];çgkÙàÚZtëªó-=~•ÊàÝ{—[¢Y(…¿”r´fúﱜÓÑSéçÒ_ùV0A®\ß”<"öö#Iתtà¶RMË× 9¼'m©hOD’`©4>&t3Õ{•¤¤$Ì»*²0“7õ£ÛÚÿ‘‡¿DÉg•lnû ô¸ÊZ¥ô¸©¯Tú¤l®§Ûò`ŠH‰ñ]ॵˆWèõšO[‘<ªŠíjÇÊ;…¯;xMcé«®ý¢¡u§?¼Å5ƒ(ô»áts™e«x§wR‘ŸöC°?à9í?‰ÙˆÆg.ê †!¡›W“ œmCö%¢-¿gÑ‹oJ•<ˆßi¶.á®fИÞµQ®Þ¾+{÷%dð9•0µ†‚ñ2dÀÝî8n2H¯•ñ(5ïé—GJÜ_™48r•.v±#èüÁïúñ©‰ñ×`åñõÜs¹æcòî!Ù˜ÌH•˜:Äž$Lñ'= º6´tÛB¼2Sp/ÕqP«¢‚®B¹qí½ï¹5ùIKdÀ_’ñSnúg!S+Z>î†âã¹"Ÿ ©¯Vq¼¯øœ–hý‰ p90‚h­p£ç‹îäײÞ$ä™Ä'a¹Ž%8?˜Ï BwÄ·+"wÛ8SƒËz.3î±Í'6üô|®u –¯ çœqYH󊻄nHš©M=»2˜þmÐgîúk ‘¡åOÑÔi©ˆŒ†é ÙÆ~³æãÊï|+/fÛÑš§†Æà1°WªüÝ=MÛ;ÝÓ•§"Sܲio(›M~EšvÇQ³!4Ðj²ƒóòPüÄNÌ·º²Z-yöwk™L÷|÷Ûo5ÚÕÈJðE?o‹Ù{eZ©7Ódi–®*L½™35²kñV^ýÜ›ªšÎÒ–öç[æ¯+<¹ÞñDŬ¬Æóo± Ìn„˵½ƒ‹¦¢~/dïÑj:Dô¬6ÙH3)r(ue§“eîõLf3ò17»™1˜ªVo)ItÏ‘F’ßœPö¤åõž¤Õør3:æÀ×ã€XÊ!„g¶7‡¯ŒŠ ÈѨ-ölTAA×ì ¾òlZMaI±ÂuSÿ& tm¹²}%fXÎÐÒÌlÌYÚ\¢Jõ…q™ÐœN˜>`*N\3 É©} ü¤´‹Y<ÌJÛŽO¸¡5rY{²äëµóVnùW=e…ÎÌ[,V1è`Ý×vƒ¥…Þ 56yÌY“¢ØˆßÉ+-ü½W½‚…Ó˜¯?©LGksï¸úZïGBkÖ‡/úFDKò »YÞFtš_»ÞÜï•úÔè“Z"¸#×­HªzN"G>~mRÝÉD&Wü3u•ߣ§Â8ç3z-9z1@hê2½StâŠív;ψ[+aãþ`º†‰ø3šš‚Â×fÖ°•ÄOý•CÁOy4z„EÌRw÷"m±¬N±§cÒÿP|$Á§§ÓtQ,;6Ò>R5÷ÉEa@HʯĹ”'q®­C‹4a:}kïöö^«ã4`Ar¯ý(áF$.m‹“-”7s1þc6·9ȦãẘñuV4°¨AvÂð_õ±vxp¼^ ºç6kû‰e—ŸyÕÆÍèÙ ²ê:¾Um5å:žž^ÃÜþ ÓÕbâ³}‘8‡²­átšiÕÂâŽó0Úo@ºn­jÇoM©•^À/FšË÷åê±¢Ð2ØlI ÙÔ£´‰g3îþË€³Ëâô§ÓƒÒ«”ï°‹ÈÁ™-Ï5_‹R–)¾K Ô3 &íCÉ%)š6ôR6µ:Ή~Ý…ea>—V%àzÄ&R%²Õd¦êK†ô/î*««î9Õ2SX]ƒÀßÉ,…É™¬å_ò„k9j<°‹% Éa^²ù‘ôJèÙÃë^Å·chü7…+x›o_­ð-ï&~‚yì»#6H¯˜QZË‹QWCù¸NpÞÝËt;F½*+× šQpz5T¥ü4³Ä)Âð ÿç"høÉø,HK*9q}'Z±Qæúeé3ÖoÇO'h*¹]Æ’­7;ÀV:eRU"†³:ÄXÿm6àHBǽ rÁcÏðïÝ3¯¡¡j…çö°$IN¤—á4ü=ÜUú¸ýP—òÙ½xÙ–…ŽØ­‘ÂbÔ¢H‡p »ìaŠ®A„ÉäÙý|7¡õ0£¡¡ÒY–år}%Jp-¶Øa™ZØñf ¹¼ôkÊ8íœÇË!IÁG+6Fͽ?l^á›§6yßB«^•–£Ûh ]Esþ çÄ/ñëCQ½ÁrÙM±ÛrOÉ¾á Æ‰µN—XõYA.ôÈQR6QðCå39¯RäõD1™ :‹¡>6W¢/tâkÚÎë4÷Ây8TõÖ˜¤(fGBK=Ç&$¶S,kÝË› ovµñ|&ˆ_1›EÂË8;y»Zæ)‡ „årÉñ°ï.ýNÁQ1ÊóT+©uknR•öH"»¬ƒvc ·"ÐËÍ!«¯~³mÉÖ€þ3âoÓ"íÉ7PúÙór w:*…™*›ÄîC¨ÂE^w|KÏÚe«tïy7«óí»%Ö6áÇΛ‘oÚíz‘.žì[+­›ª™ß\ö]” Òêtüe2Kž®‰­ç©uÂR» |aÆ ®e~ÃΊâŸÈ%g~²êäR~p |uÒS ýxVK-ȹã/k»‘ik•³¬Ú×–&•åøÎîÅvZp²_ôr䌤Mè!eÎyT®s‹ ‰¸¬V“‡ú*h2¥'휵ÊÏÖîzwÂéQ™ÖγHàaÎ9ß>ŸcÔ;¼¬bÛ"@ü†,®î°¥ÔÅÈoup¦ö­áš Žª(•krù•Ñ …iõCTnqˆiK"dŽL-ˆzu#“Õa$;áÃãµÎ51Ú1rÍÂ÷ÕáÍH”ï#ÎÒ%î‡ ‰„ÐgËN÷hVFÉ}åëÐJCZÏ–­…ix=O7ç2—±øz:¢C£;ëÁº— ’ÌÍž±e §ÐUƒÀM\Er¦Ü·³ÞHvM€`RÓ“åÛIúsg/ì«Ã(%vÔ8TøË ë¶ø¯ nú$ß3~k +¶¾Ø“—H5ºúÍ›ždð£TÆA"> ô–µ_uç:ÒFV3`»6†RDùaE™Lÿg¥%ÜÞÕúÊ2j« zªÏ§¯ø>0èèˆh$øÙ„ZÞ¼¯|‹\²G’؈R+jìN¶Ü!Yà6/os-ù ž›¾™1€›ÿ)iÀÃ!ûpÍ5ûIouúëMSGýë/ylövfRÃ]Àny/á •×™ò£¨#§Þäbl_ÏoiD7Ú?ï…ÌAÞݵâ' «RoÕrÚýîihBt_œ§i`)OwöAÕõ?ª ±é¡"™œ¢›wÆÔÌo="aÕ)›µfi÷Ò 1Ì÷¿Bzµê±KF—ö[GC |¨>åv’Ô ›™X4©;܎б$Qcè6quY:îµr…rÒ—>غðn–ø!ˆ„(Ðkä—‡ lPÊ bxË0Õ»â;2àeßݪ’Éçë?¢ µìD'z )Rîõ¼3Ê( ×첟§«âb”·àJ“/ÎíèÌÛSÉÀVú C¹´B?Åáƒi’’fO0ë²`ê§cuxL´páøN‹f~¥-7DðPÁb®Ô¢È&Ì3‚~©ß5:[,èáX~ò²J\%´ î¹ä9{²8()#ÃèY1®Ìzgo:Òî>ÕaOÍ÷«ñ±ïƒºâf†™¶7“à/wY‘”è”jŽÏÛë|ßÞϚ楄”y‰ßþ¼cñ’Ï=”Vd-(Ol791/Íò1Žo`§Ê/sck/ôËÝ:Ïû"¬ôöä¶©,W.Þ£){ ­R•bðóHCìe´ìDâò\–NXkÛÀ•!› †Üïû3ÉùûG$±ÙnéLË…1,-¬~+$-a§Ù{—ÀRâÐjHÄ2Är•øb.í»¥»DDª‹˜lSö´?Ø1K¡´úÖýEx„/¤­D‘ýn;øÑA|Ò®qïƒ7V~$–fÂÖ÷Ã’±!Ë•¶ÎéäR Ga,ðùÆE¾SÊYâ7’åÖŒ=µ_hj0 Ï' {kU¼˜"TïºÔ&îm,ÖŽð÷HÉ/‰u½y¯ÓÀñôYmÍ|Àâ£.æ» O.8µî—¾|iŒ¶‹râÚÈ»1rÖüC53Ò"*×x#D.™$ž·HèýiP=áFü<ÿsV‹Ãtw‘ÉŽ _Ñ.U¬ÿf²N‡!sÝÛ¬:q;¤Âs ¢qX[ Šã̇ã¡åm¥z“J¥Ã}„JXÁ“œ€ÒÏ1ªœ‰a² uÞ§-”)_ë«ì…ï¼<‘þè» å»÷4°gœtdâÒç(¦>!zºÚC¯0Sºd¦$«¹4Ÿ6¬}E¢+Œ±Ý¢ªïJüº› endstream endobj 199 0 obj << /Type /FontDescriptor /FontName /ZKZGQL+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/eight/one) /FontFile 198 0 R >> endobj 200 0 obj << /Length1 1744 /Length2 10582 /Length3 0 /Length 11687 /Filter /FlateDecode >> stream xÚ·PÚ- A‚»KÜÝÝÝ5 n înÁÝÝÝ‚w'¸  ¸Ë#9çž“{ÿ¯z¯¦ŠÙ«uõîî=%©Š:£¨©­1PÊÖÆ‰‘•‰… ®¨Æ`aagbaaC ¤Ô9I(µ€Ž [¾?ôâ@#§™„‘Ó‹™¢­ @Î `e°rñ±ró±°ØXXxÿchëÀ0r™™r¶6@GJq[;w¹…ÓK–ÿ4&´V^^n†ßîQk ÈÄÈ häd´~Éhb¨Ûš€€Nîÿ‚FÀÂÉÉŽ™ÙÕÕ•ÉÈÚ‘ÉÖÁ\ˆ–à r²¨.@SÀ¯rJFÖÀß…1!P4,@މÕmÍœ\€€d´q|qp¶1:^rÔeÊv@›¿Œþ2`ü}5V&ÖÂýíý+Èæ·³‘‰‰­µ‘;ÈÆ`ÊR LNnN #Ó_†F`GÛ##ØÈøÅà7q#€”¨*À襾¿«s4qÙ9929‚À¿*dþæå’%mLÅm­­6NŽ¿øI€€&/·îÎü»­V6¶®6žÍ@6¦f¿J0u¶cÖ´Ù;e%þ¶x!ü+3:8YXX¸yY@{ÐÍÄ‚ùWp w;àoåoñ oO;[;€ÙK @oðå ÁÓÑÈprpz{þ©øo„ÀÊ 0™8Œæ „£¿ˆfá—Î;€Ü:,/ƒÇ `ùõùç¤÷2[¦¶6`÷Í7—Y^V]CQ‚þwÁÿ¨ÄÄlÝžŒì\F6N+ 7 €ûåàýßQþ©ÿ?µÿ–ªþæÆòoDY3[ï_%¼ÜÝÊpù{&hþ^ZÀgP²}™c €æß±×eád1yùÃúÿ<ü¿]þÿfþW”ÿËØÿ/)g0ø·–æ—úÿ£5²ÝÿÖ¿L±³ÓËF(Ú¾ì…Íÿš¾þµÄŠ@S³õÿjeŒ^6CÔÆüÏ%‚¥@n@S“‰Å_ôŸ¼„ƒl€*¶Ž _Ï €‘•…åt/»fbõò”8¾tê· ø²JÿRÒÆÄÖô×αqrŒŒÜ^ÿ‚8ž¬/Ëi tû=×f&[§ÀKyÞ3[„_ýäá0küý…¸ÌZÿ ^n³Ñ¿ˆÀlü/â0›üƒ~1g6ý²˜ÿÂdfö‡–Àlþ|¡`ñä0ƒþ€/”,ÿ€/œ¬þ€/4¬ÿHôBÃæø’ØöÈ`¶ûr¼¶{?Û?x³¾Pqø¾Pqür¾ärZƒLlÁ¶&y‰ãô|!èü|qrý²½0rÿ ÿ«‹&Î//Ûï-{iñðïgtš ,-Øšð°¬ûÐ~S#Jèʸ;!8K¹û6•–ÑsÉ¡ÃùåumufÀºÃ•hÒH/Ú×mIšK‘e’GÏ- ¯CZTÛî½ âÔ¦wÛ¾Lá Nþ­ †'bÔùæõhï¥åoÕÙ%G™kï̃¢’yãÚ/íV?P¾2¼°«ú­šKñ¡|†1J3R׿dŽ2Ï8k Ö‰‘ŽãÄ uîòj#gò™D.ŽÁû0нÈS{ƒ-úvÞcµRƒÍ±Ÿ_êclšÊSl/Ywѳ´hÝa,e§c|Ó:‘¼G㱫¤vã°¦OE9ÉGƒÇ …%¹Þ*o]BË‘uòUnÝMöŒ»ÅØ‘A9ô¬ æ®C1è?7Ÿ®Oípäâÿ–Ô«’â—•ÒÐ\åí„áÖÀÐ-ë"ÎÞÿ"ûðs…ì½à×#ŸòF’4ìïY2[`~x¾1·8f•‘ûÄg…4³¡ñ(õYjGã5¶‚EßBX/•.(»$%eìæU¢ã%——µb5ZƒÛ}d´qÅ“úâè¨n¯’ñhÔ¶Ý“ºiéeÖ#ø·­!zép…¿ Édv…ŒQ*EµñóPŸ/ÃIÃbFi÷%u'™n ëƒ)ªšOZ/bAFhÀ«DÇܦþµÍ@)Z0Ÿ÷›çjöwq0ÈÄŠ.7P¥”ÒJ'‹¸ 3ÅÌ>†Sa×d‡Pr0h±Êlì<˱ X@$£ž¬t€ÉN‡÷Μ_ñØÛâ8AÜr·åéû rYŸÈ9<•61Îz=Âì%Os£N¢u¿¡yWjhAå– <ó•­Ú—8ÞZ©Ñ ;lJ´€òô@¯ÇPåæYããÀjüÎ/÷ZðþòÈ[ñú-ÍýÁÏ …ñ.C»“Èâðj•3þ'7ä†O Þë=ílˆWDeËI“GŨ…ŒšA=ÓÖ¯Ø'Ë>PÃÞâ‡v´l•bj¡MÒ«-úJ?Úe»KàÉLº¸yÚùa˜(à@ç_³º¶ïõv®Î`ôc‡Á7Kí+*™Ó¨C†oB„çh¡[…ùÛE1Ç ŽÄ”Ö+¥;ß]ÆH€Pl‘¡lU>.©·÷ª¿[‘€Ñ±úqc>¤sì”ÙÓwŽY«©/wòydÚŒ-¼9{ô”ûú*ŒÜÏíHb¯¯Ãº5ºãÒвÈdĪ™¯V­k)($ù]³­m>o-Gà°}«qžIÿ=UmêJ0Q¬í!¯£P!òíh_[g\ë.B5e§¾ììʾx ¥mÛìCEtº² ᙚ9y2¿D{ò(aö[}àè¹×ÒtœtfBÉÞüQæ¬ +Þ庅@¬s {>úc)ó0/=£1»k ïÛØîWÞY‹+ô£ó›-DÍEOã³ešîvwÝN‹÷ÍÑa§Il¥ß¸fKãC¾ãÂÎ)gÈÇ@‹‘kõº¢çv±®œ‰¥º¨£vàÐÜÐ÷6¬Î…v¦:Ç£G`«R°ÚáŽË™_G¤[FûžÁ X-Ÿ÷° 2ÚÙ×òÕ÷º®—yíñ¾Î”ÇuéRÈúˆ­P‚;q{XIsYÃÁâÁf©”Œ‹+M]õjBí[áªP¯&PD.F‡úZBÄð6Ž= ¾í$HL¥Ò`^½«øðà$'!„%ý‚:/ÍqÑÏ{–÷ˆ¯XÈ‚¦jì5”!õ`uý:°DÇ2n¦+agúÇ.£“‚l+HrÜZ;#Âf|œGÞè¶LO/G†Äâb)™mz“N1Äl{9˜-KA¨“ÿõ=ƒLp8FQç€ì²[úí܇ֲ¸]E >n‘\]Å¾Ö A8¯ß]Ii…'^‰K*»Éß!XJY° Å7Ò•G4?ÁÓ8D±¼Ÿì¹íüº9"Ȥ€n—¿t(¼l¼»>Öª|òºž@)ìŸfDÏ'²ü!j¶Åû©SëÁø"¯ƒç•ÆýŠi%é¾V{kê°îÒ!\ÛeN›¨U< -Ì‚ ýjõèí€t!ì­Ëó ¸Ógÿxp)´F‡ÓA!ÄÛNãÄÅÚR|a·ÂX±ƒÏ<{ó¨§/ä ¾_KùhâÛ÷„årÔ7Uä^ÓîÖ׌㌠£§™ˆfARWnoïP¢:*:>EÙ96¿#â²gŒù ˆJ}Œù˜7³¼NÆ:4ÈpÞêè6òIV½ :£?óCë™LÿÈ’{—=Sò³‹lZ t¾†þ±üc¾•päéÈÏôÙx¦M~SOÚùM;81 \8zŽÊ½gMxDººŠ½W›|wÖ EüĔ޵¤ŒE[#<ï2Æéí¨8®,Ó°wð¯”õÛYk‹4ì\¢$‚ôSb±úû_“D¦¯ _¸Q¤~:¢7_ì¸ÅÇáç}ãî©ÜÂgUç"ñAÕ ™j¡´u¸{¶©U τ݊j¶}I.7‰Ãzù\çK·N°…ׄq&&çÎú¯\ãpTË›áËù…Y–*”ùíó³ü"W›Í.ÏývUÑ&¨’²ióY™äH½×’È·5(çq¶¾ZYÏM»~í|4!p£ºÛ…ïI^; tÕç6o$÷)µo (æ #!ïS×|6hdL× ‚¶×u±Û$¾)çl¬N3XÀ—EFfrUÅíáø8í$ò°W䜞ÿÍðýúÎpÛgÑçM¿s›‚~ûÍÊî˜Òc? ¹'*­ë^mK–4sñ %Í>”rˈ¦Ìãu§/ª<·§>æF kÍŽ*0Ø©\i$¬'rÍ̳ùeŸ:Ö{’„óÉÒVÄff%ܧê…$W8LŠz¼ö¼ãPBQ`vï¾×`5Ž»a˳=¿j¿×ß=Ö{ ­)ªÅýÑMÕˆM~,)7fÆÌ ©ÆV‚d™±6„Ò0g#òJ u.]– ÛÔ«5‰þåA3ÖiU*¥ÉR,[¡_¦û‚%°ˆþ!¹Ò¯5Pi?Í ƒÖ„¿Æ.¼3˜îÓõf³Ð1¶&«?FÝ×[+“›>þåîÜßeŽg'dÕjókÙ)ép^Îç#o7€ ·ö™w ÆPfÃ~¶žIrñÒKŸò(ao*,4ãw…©Íö‹ Év\(L·8ëlj˜Û{m®ùbÉ îüzñ]z>ù¸¹ƒ°„ÚÍÛU)Sh%órV8é‰Öo”WPñã‹þ¥‰ŸW&×vVYÝ·êq×va+°BÇü];ÄÁ0GßÊ+k³ÇìY¥zTÍrY¡7Ú"’xu“~ˆô¶g|'yµ¼K3JÔ¨™N$I =UÿHÙ\Kƒd&ç6¶­ÿ1£:©¾îÕI² m<&ØèC;ÎP`þñ{ ”…`íæ¼‹Gm¢ÀÙió Ó>xMÞYæm¯¼‹Y­É  œU* ƒŒJœ AÖ´{Ì éEéM·¡=$æWaduÅÕR×—¶0¥õ tqQÕ¸_®P—ÈߟÚ.ò­QÅZНâùW9á]Êw–׌:÷FÃ,O ˜¯2ï\$¢²Ûä`R?ï€óØ ÄÍÇìÈȇ Ë/!Á™¡8%Ôt@5‹]FÔ(¦7X¥üQ°ŠE¶ƒ_*MªcP'¯ÜîhHN@Ýxá®ˆÍØ·JcÍÎ[I_¯Þh_ÂÅ –B¶ré´ÒRŸ )ê`#OtˆâÍbw&Ž“ïDt¬ |¯9¬äÚND˜?ª+—OÚ¼—Ó%³&Ø=öðdÍ«‡!ŒJüô“¹K*÷ ¼ïMÓ™oÓé*Ž¹Ð æ¡zeiÐÙ!]‰¾ÆTý>ºêÁXç¤EîOIÞ 7)„ßÄDV%ÆMÅ©§ôíqqŒ°!¸-§¯œ60’ŸÊË%X¯õõMw_¶îU6Šº:œdŠ‚í‰ÕÓchÿ´R6a•ˆ!$ÚŒzz$ï%°`ô²!ù´&xxz2'¥1É$3ÃwÔ‘8dƒtÙ ÙÙqLFLwùT åE£D“=Ñw_•fxœNƒÛV>c™,$žÂJ'>’^º.‹‘ ‡eìtïbùÊ$å­}絓Ÿ°Ž,ˆDÀÞts¶à½ëñZzCÃWlÖÁ@z‘ò@Cïâz«þøÃYù ä)úÇVGzš{Hɪ¤aVú`Å„#ŸÊ]jÛŠ˜1&ÏAµxdºJ‰*|†‰úN̓Œ%•ÁžÂbKØD)¾±úÕX3hx«Ÿž†oG[w‘ ¹Œ¨&#pqöÍœž\ȤpTx~zaÅÓCyø(ùw®Ð Šù†)â _»š]¼¾â¼ß| ‰ä"^¯ð]N—•‹ Žñd×QXðÿ´&¯Rm~ÃIbjÊBàÌN_N² nâ_~©=‹21çÅÚ%Ö¯Çÿ µq3Y#ôMMë´wgĪûÁ¹'ÝElN#ékeýÎ}dþân®õÁ¹~ÙѸÈ4kÆA7‘óW³+ŸcgtºØ ÚÃà#•ˆbÄ—Œ“Sé @ß|Û2òéG"o*–mн7²r"?žD[‚Û±%> Ž@eÃy‡-µ|¥c gÞâ-qàœ9©NPwÿ¤k¬9¢uRéÔîVáòˆnRÏ‘ùXï(d¹pûè«qÕ?K>1¼¹¤ú™¨ZØà,W¼-~VÜÌe}bxÆé4 Œ¹ÃÔ!$m쪹nºh‡W |ÆÆWW††îöÊ*WcN­Ò~±ZZ±ˆr€¢q{2/|@œ"JJxm“ùUE`¡¯œ_uA¦¾w¥Ï%ïâ2w+Ê/»6³$Øø¡àPºÔ{±>péÐhF!y’ C8.p”Ñ Iqç1“ ¶*5Ð6óB>ÚJۣȂ ÕƒÉÄ€ÞëªÒ©¯êØäÚÑý\T< )–ÿL¶ÎÆ1BöSмpM>³JáY’NlߺŒè«+bþª ÃHC޹Y›åý›uNŽÜ$÷ûý.5¸(òè”suÇæè ú©×–唃Û ÷‹{}ö»9r5Yþ]±u5¨=þ;(Ÿ¯ñw__˦|ô7ü ‹®î•tIm„MÊ­sµ MåÇ$æ"ÏÈáé3^M-™ˆ Þ¹*^_½ãEâ2ï ·QG¥ò>“RÈàt`õ~·—Õ&Ì?)3T½N¬¢L'ÉAi0;í¯ê†Éð§gí€ <ùì½æÉ?>¡«ž –—©üL‰ ;P¨J6÷zgíüˆRKâ¼,¬2"º5;xæ[‹ûi(? N‰¥!SISÒÍ-oÝ×­SxÒ’Dþ˜˜dfÈþˆs!r{Û-Óh½,e¡nXAFÚŠçŠFSâ›bó·`/?8…ù:øÓm©I™™ ]05á3Ê̇øQ¬–ÍÓŸ»±Î÷Û\šeÀâßcéb6M1½òŠcaYÊÄ÷‘éÀ‹‡U˜l¿Yÿ¸}Z¶jœî„WÓ\rû›µŸD¨Žé»­ÝS9wDÞðÙeÏ`œìÃÙ·iº#°mÀå=ùàS‰Œ7(9ñFæ>ÝBMÔ žÄ,_ÚþèézO_.?¸’á.Õ7àäÑÏÜä¡fåç{«çÁ îãeï¥j@ÔFªÀŠzv¿ï§j®ÔEÈ ip¡/pyw€ÙúÃYœíJöc.Ê!¬OÎy`Z¼£à Ý®j{ÉU.™šj÷~‰º=«[ØX1Ã÷}V³‘Pé€N=7*’ñ$t_«œÔVNɇRj¤h)ë”»4½ª°“¯Ö}ž“i¬¾ýÚáq©4ƒoª1]2”» ÞðåYMåãágpøÚâŸô¨<ûÌå#UìÕæ€ù·"®WëWŸFY‡ ë/ÜeEÒãIÍNš‚/J$¹ Ñ'èW·]3sÓ™ ü‹èß)kM(ìwkÈ ¿þ–‰ïlÅE2è°a^Új”ãd/x‹ö]Q°úIøÎù;‡·‘¥ôs™+äBrpáhÁ|ZÍ£1Ï/îý>ß½Ì$Œ~Hѳ¡OlŽB"G_¬¢©(lvç&Bmª–º ï±"&ÛMå%”Ì‚rIªw|¹Xý8ÈÛ¨¸.ë9x×?ŽÍ>]Žo`l¿“þ¨¬ù¬^­È[–MN%I~¢9¹)¢yÑÑã} T¢]s*Iÿü®|| í"» “Øx3êò›M=qÖìàFâXÓ ‰þ¦8µ²ñ£þ¤Ü9¦EyÓã_([º·­#N¶Ã~ððq§¨ŸýÀ?´5‹\­èº®ü“HÖëMœ> t$à×ÎrÆâæµÃšSŒä±º‡Œ¯yšÊ¶QâÂøÞ:z׺Âϣ놞ï/zËå º—£ú­£¬ð1¸™\wû¼óe¹PF)΄H}¬wS01ëÏûÁ«bIÍmTšz•Î. ¿p^á 壒©WX¦A®ýöä2­Î =,ÌFŽ˜X†;¬3)X˜»£¡ÄÞõYÎ_1@iç³zàà°Iƒ$Of†ÙŒk+飣qmŠm©ÕÒæÆÖuÖúÄ<Õ{u×°*óµ4£@÷ lýëèQ°Ð«3^"å»*qŽnÎ¥…–--ûÛÊ÷é"}ž¼Ý¶ë+"Ù|p\ÕʵÀ'¸Sc¬«å‘ ^ÓøS\|š™6$¾LLÑ^GqMN ÖÇâ ÁN0 34ú§Šk3sÒ³ˆðŒØ]oè€j~©bÛètü€B0†ô»ü±V3h¥mÐkûÖ4œm¹Ëãcè.²Õ‚ûn—>æU‹,fPpS¬ !‹¶±øì+\¥ˆ·FÄ’|Þ–2¥MB}·ýß’¾<Ípd©r&NXÝ`cH›eekïÓ&_¯ù‰ ª2¾*Pû(¸»h?8¦®¹¤@v¡_ÿÉÉ‹ÿÚyغðó-ó¢;o@5|]ÍšÔGbJE)žWôM‘ï[8ÿ* SaXDCó\ðYüFkþMSNhcT 2Ñ?xP%(Äây†ßëÔÀ50è@È­´#m¼C–Š¿ŒÊ*E.d’Jnqlò5Ñü+‘p˜)¼|3• šÁ˜$î'‰’‹ ¯KZ#N1`ŸšžÈ†1ƒy'ú›t[Œf‡Ô’ Ý›M˜4®»Óeí».q¯Ô®‚W°¼:[p1=µÕÏýòÑý÷0.¸ö¢¨8S„Ü糜X›A3Çñú'¹¥áê];*$.œfnÂMâÌ ŸêÙtˆ–‹Àí{çªÔ@ˆîló/_kfût ?ûÁNtP«=ôÑ“2.µÒ© ®ÓI•}c«RN˜®ë®*ú‚7a¯û†sï–‹Eu9ï|“;@³hÛ0…œD¦¡Ÿ™gÚeŒo\ßÇ Ú¥,ÿ£:ÔVAÿörÿQ)+@Š©ëþ4Ö©‰Dd>Õ™úäMò¬¿ef[šS¾%êe¨mˆ¤÷ðË¿C9'Qx¡ÞƉ-­WÌÁ¼s›†…ChÊ3ktUÏÇNî%Ðê®Ék›RáÍù‘– ¨JJF4ë,NjsçÅ‚,+Þ»ý¤áª*mÊ#œð!ƒÙñfrÉÌüq¹V‹Càƒ23îŠ7^{±µ?²ßǰºÆ¬Hð‰XÊ{˜¹ãb•çðW¸Š­7=•Qƒ5;!:tÐ: S=!vÊÑêïÜ t"±CÅdzO|³’<êî}Œ³:’Ë :܆ÍUZúa%vjð¨¶7'‚Ï&˜•Uî<Áh¯¯ù´wCîoMºö‘ëÜC½p£P.•E®/r«¢oõÞ–x—M,Còt¾œ^ÒÌâ<+|ÖtJ°ë¹i»Nš_W¾’OðìÜ–„°w™d°ê˜YG¼ÕA?ˆO@tH‰»k>ß³o'~ÓÑâ*¤ˆ‘á%Yž' 4”@ñ㢱à 1!*ig~©ë6Þ»)"?ð"!þQz–~Zéà¬ÿÄšúb:´»¼…Ç!R5àòÛ°HÒm{,·ÝBß—wg'ܾŽOSV„6Ð*eQñe§œê5æû]±ßÊüÀ_ËbŸ5Íc‹¸â.ÙÑi2µo<:¾Uã,5¸ê7¨$|®œlÃc[G•ø ¹x|¨nȼnÒ"ï'/‚qI,²úcM_¥¦;ÞÕ : ÝÊFë”3nÛ[1÷ĵ×U ·zÏ_«h¬Ïëƒ^¼± ér÷ÖËŠ¢r%³’‰Mö!ÓûÝ×H‰œ´yß i²»lÇìÏ"Hã1¼kÁiªHŸÀ­Wè;+|-ew¡Ýç²00:Ƹ*£¸Ö-¢´¨mœäø„cPê;-|9ÛÃÑŠ…·N¤Â#y¸ã_¤gö[ÂéÈû^«:û|šfn\“ävªÌïM?Ê^qG—Ê^9Ü5BX-¸òh¡®(zc≄}Õr¤Œv9 ¿8?"äs.9mŠÎ €ÂÒÞ ÖY5:Чc˱V\¦1퉫w­›¹ m·†N”‘ª½îô> »d*ts|3xJp™@óŽdüìÊ'f¶[•«ÍÁ´k‰ñ k–0{S¯?Râ[ú{ܹÔáô”—Ÿ0E Àôt1÷Ï´¡3êZàð†ÔƒÆiÝÀ jž˜03‚î¡L£.R²ð=ÏFö䤼Ju“™‘••TC÷ÏäÃŸÑøOx´‚l`öJó‚š²¥„Úྵ÷Üsï¹ÏA)ÓdïÒ‡˜ë:<›Â‚oìN^”Xõ¶C&|å0êCž34•ýbe#så4½·…ä¯ŸŽ¥en¦›©êA€þȈË`L僗Q kGíÞ`xC.Âq5~Fsb7s¯!…­ŸÜ¤|ŽŒ.]C±(ÀíUvwCKM^㲦zkªäIyåÒ›wQ˜èl(}Ïi…ð´íUË=• ·&´£âz{Å#\£×ê'R õþGâ†ðõêçKs-†]kµ¤\AsÈæÁË”;Qä5Ôãôö nj¬ÆÍö)oÇj=—w²~õ¤gƒ^þ¸ü™"•й1C…“Ûd?Ž77õŽV¾4$]îv6àMœ,gÝÔžÈñ÷ð‹r|Gu÷lġٮ$B©¤Vöóo\LáÓcÏé¦l³{i â55ÇôñW¹5²=°u(dhÂŽG…-ڋݸĻB¥Dùº”õ`Øg‹ÐøPH_ Ó»§;|jò]f5$%divÄÍÖ¨pvB&OÜG5à-¿]¾Úë“ú* äcˆ†4Ø€ e ÷Ü.W¯)ül–A9ùÿÉ÷#Q”=ÌYÎ4 ‰˜’LjY9Ær‹çãÂfÞíq†ÊÀÎ{L« #!×BÌ~±Fç, àeç½Vûû3ÚùY•Éo:¨ËÛ„Óí²†t.!6P\:/&~èaû÷ÃÓé#–™˜)vpH`ŒIiéOQø„Ç _snX¯Þ±B–“@m‘™ 5 ¡î?§ÕewWV»Ã(‰:h³ƒ°9Db4G£3Ž®?CJŒ]°IWÙ<$±U‹šfr2¡ú0D%^P()Ä^ìHÅ?Žç>}ºΚLxíwÃh+ ÚFñ|˜•c/kØj ±üÈ}Å“:ÇÖÞì„‹ü…y…ë ±ÜÓÐ]iÌE@k`D~ÃÒ±Ÿ²b‘¹wÈØ#3_S=šã?ÞsWµ´jH;eÅ¿wo2¹ëvŒÜ©ãõƒWW¥WËn]ÐZÕj“|ˆéŒd*áøÔÕ9„šô€¨}ð‘â(?ŧâbqý^w.4—«Cë²OIVÚ gìm"š¾0Á¥+!ÿYin:÷ÛÀðýå,U*¹ÙpGɳ]œ—D¥™«—ƒøÛsËÚ\t±¦w³ŽÃÚ‘©b¸¦Yi†dO"JmÚ¥ *càlü~Š ?_„ y{reýg׉ڠÞÐÑã FãA8 E&)ï`9ǰÃãv?ª¶é]!;‹U•^åáítg!ûš]S98ëuòÒ<û«t¬•…•ûÍTA›ÈÃÌBOÑžÅ-0LWJS[×=%ú²&* *¼PE)’4î>æù›î•86 &[Nr&ÜC? `ƒìÞí˜.–šƒ¥ÁѾ6›@)¿Y†B!U['ÇHô~~k!£dôÔÙœ¶NÀT‘ù̘vÚ[ß„±.!ÏÖgµæ‹?®–f>.K¤?‹N2ν6ôq ËZ—Õï ŒÑ©-²ñõy.ï]É](…)Ë—Pô©š+¸œ6ˆ`*†ÏÙW6VʧȮT’÷Ò"‚Œ|å0¨»Ïx®A&u+óp_G;¬_Olí­‰̵p[eJ¥uðëx¤hýÈ>ýL<6×IèY$—äm~>©dd뀮ô)Þ6È®ÀL‰uàÙWyw1ºÐ"âH2Ù®žák‚àxaD´_ƒïÒ’[:@«@Ï®¡!Eÿu>ø“%@ZJ"r2 ò{](³¹ø I€ËSt“!ºw^ÊÚ|T×[&y5yÕ{:¥S[ 7[;\DÄAp)ÏWù÷ˆ¼#W’¡ÓV­dJ³Š©%ñžvƺܻ†­d¥ÝuO÷*ìÚOß92Ó, yÌ`PB{…²0ÙÁN³%Âó{äI/*­.'W8æÌ¦¼pÄz#k/)|X¹ZÝû~V£B´}Ä´ÖuU¦j«úéVp±w#4L4ŠêQG<ý9Ãué.Ãsw;k(Qžþ„¦’oBŒ„óLbþ[:“q׸Ähö7¯nf.ûG2¦û—¬æÇodºì$†=qu²¨´"ïK~ ¨>Ù\'>{ãÍæ@¹5ßß”×'›Žß¬9ÇÅåkŠTqŒéAè·ÊçîmB¬¤Væ|™™I•÷ [>s©¹>:{®FWÞô8§¥:Æ\&ðÐGOå}4Ý7  5Mú4nè”5Ô²YÓ3JiQKÉJz_úaZ+ª­‰†–åZ^BñzÞþÓ€NÈ«)ø`²„œ»Ó¸–¶ªï¯D¯¾28—Ezô^‡^ôAH€ð9poE¥‘{,ôXH²=Ygïõr8®ÖÆNȼn¡°cÑü[bÎáýûK‚ܦ¼‘ÇJß"O\¬ÑnÉ"¶V6!½û Ÿ¦ Gù:烇Â~ÍÄŒì·ÍBv½b¡$ßDËý˜ÖoE©´±o0q9HN¤–%®únR¢ÁŒî¦dò™CKŸ¶ù,ïzÀq5}"¾:$zÉ+¼&û1ªõä¤@lq°IQÔRþIU-é„x5l$x¶Íº»Mž‰;WÇÛ3ž ÷ŽßÌ)[üJ­÷:©@ëéhë¹àÃ'͇ï…)¸b'Ž· ¢æþºó'cvç IH^'4ãT¸¾=êL&²#vSaÕaD;Œ,ÞcÛ­YäG±@ue҇זÒÖņ\ü2™A–‡E3ÔkÞ-±”zˆd‘]•ö’M¼)"êcdF¸ü]+.øD›Eޱ™‡o&ÆœaS˜ê«D¡T!³ùŒ˜Æw(#ê"‘ôÂ?ký̽…hÁ€6¿2Õ6)å×-ì}=«@úƒ„aèÖ?À6&¸mܼ1l‚”K×åàÈJjM.)QF”HYÀÒËé7ÿ°î9hô#.¶g>M±hVö{Zfüì!š¹ãa1&µuqL•¡÷ØžQ÷Beho «ÜJXhØÚTÆØ,²»~nxÔX¶Ô’¯#"8™°–X)=vq6‘&rXý0á(‰k×=‚.ŒìhRtÇ—4¢Ã] ˆ2ÞøÁIXix,½þ¶lÖ%Ï9¬¡ôNu)ºš©%ärÝ=Ýw5æR36I m‚w³dV²5WÚG˜«$t©ï¯7¶<"³|ñ?RDF »‰!D²±TÞR(# 7tÕÅŸ¢i}R]®eÖfr Ç(nñK…ðÕpãÞœ½ä|u7x(BÄ×÷t³³=Ó§i»üPª~J-ñ¼—Œm;,ý¼õê•Ôéq¢oÐ"ÃóåféŹÓ`÷Ìõ`WŽ™”dåsÄÄ«»R3‘E•Ó@Øù»1K~ÏzÎaÏíYX½Sd7´è[kzKÇ4ŽWZÖâ輋›bG‰Q£J;ر·›‡vgƒV­PMWIÁ¯BfBg8ì Ý«½¯Ä[¬„¶R¢p NÊš²Q~‚ÑÓìç8€NVr•qZ݈%!²çà½MFï7´©äÖƒ¬Ôšw¡ï±²‰@]fM”æ6`¦ê–E÷&ôAËH†©.©{?ÍæÔR^ä%uñ¤•e$êÔ5ëBÔ›³bŒû5I÷Ѳ3ç(ñ¼fþjéØsœö¬”Â'Á¬O‚×d&prɉO4<ªö“ÜñHqZ=<Ð+<ùÒ{ï#mçiéC©…÷d;–>d^±­Bïtª> endobj 202 0 obj << /Length1 1514 /Length2 6588 /Length3 0 /Length 7599 /Filter /FlateDecode >> stream xÚxXÓmÛ> ")‚’¤»»é‘cƒÁØîé’FP¤[Ié’R)éF@ù¦Ïû¾ÏûÿÇ÷;ŽíwŸWÝç}×½caÔ3äQ°CØ‚Up/¿$@IÛÐ\€ÀÏ/ÄËÏ/HÈÂbEÁÀã„,&`W$—ü%W0…Æ”(´£6xäD%Ä$ùù‚üü;"\%Ê@w¨@›ð# Y”.^®P{ºÎßv@@BBŒûw8@Áì ám Ê쌮†ŒòúG viÊE’ÏÃÃèŒäE¸ÚËrp< (€ vuÛ~QèÁ¨ñ²Œ È¿ †Êè  †#Ñ!np;°+]`¨¡ÐuÃÿrÖúËðçp¼ÿJ÷'úW"(üw0B8»á^P¸=…ºªZ¼(O7·ûå„!èx ; Ú¢~oPUÐÑ ÿðC‚\¡.($/ ûÅ‘ïWô1«Àí”ÎÎ`8 IøkÊPW0}î^|šëGxÀ}þ^A p;È/vn.|ÆpèS7°†ò4DøoÌŒˆðKˆŠŠÀO`O߯F^.à߯ß0šƒŸ ÂAÓûA!`ô¡è \ÝÀ~>ÿiøçŠP@`¡¶`{(œðßÙÑ0ò×ÝW¨'À‚-?ÿ¯×¿ž,Ñ ³CÀa^ÿvÿÝb>E}ucC®?”ÿeTTDx|x%<¢üQ€˜˜ÀïŸyþu³ÿê¡v÷5à@â/èÓû›ˆûe°ÿÀ?+è ÐzØÿ-ÿ'ü"ü ô›Àÿy~‡üÿ´ÿ+Ëÿ*ÿÿÞ‘ª öÛÎþ—Ãÿc:Ca^<ÐzvC¡gCžø»š‚ÿhm°ÔÍù¿­( zFàöhóóò ÿ…C‘ªPO°røKK7]…ƒõHè¯{ÅÏÿ_6ôèœÐw ݲß&0z²þYWBØýAAQÐÕèEˆVz%ð@ϪØó·Ä|¼p @sô@®„¿+$àCG!<þjÚöàðÙ]ÿD®@†ü§ŸÐøŸñü>g(Ü ù/@DÀÛÿ¾a‘0 Òá·åT@n®®èiÿ­94Ͽ׿¯0Ø "œŸA€¤Bß„¶]T+Ðzð¬à--wD%˜÷FŠ Xg_ø8hágªM>U´±«¼?ô\o®4|bÆ—–sÿbÔóÉ»ì Å4õ<†Jïªâ!OþÀ÷“wö`;ý˜Øó¨UòƒD}zR  ¶dÓ™«0±.¡×__]w 7KìœEgdŒi´¨X…ÜŽ¬gæŒÌ˨Aó'dqw\-ÒÝŒ4ŒÇ®±žAãÎÛb¨¿{­ZÄe .,¼Ïîë}.t\ Ù]%ZNA}×ûcîPR?5ebö£F$«¢È™Riju¡s8ë"¶ØâåTA_kb¿[½½ŽRÐm]ÿá·Z„ÈŒLn×â1õZ–Ñw>âûMIQÜÕÏ™Yªì0ÎÓª„`yÝ8T ¼ÚŽO7ÚSˆâ’ss-¬ 1øþ!BV_Çï_ØÄf& Z¦ÊÀÄû¿¼TÕƒ1¸ AL?^I§(hX(ô4°u—݆ró~{Ù -7{ ?Ýlzß6Øqˆ`Þ§§œ J0kÏÜéÙNlÔ9û@n©"ŠªÑœEtZsÄçÆOäåâèµv¨I=–—˜gŠ7"Í7"Ÿ5ÒÚÀH×¹e.¦8EFBá´_Jÿúc…èð ™à&™Tñ@~ÿO㇘Êú»ñd5«·õ»¾ãZ%غò XiæuÊ(c‹> VÛik×0¤áô;‘€Å¹”ðZ‹Õf½ÎËeÙSEeÀR¶¹0 ¡ÕJí@V ¢H“eâoâwˆãõÁ‚3p^ HáLNó#UÁŽó³¦hòÚj.‡gÝt -5t|ú½AçØ…Áªâ—îzBŸ¿ð™éо÷R@tÖìûïlXš>¥UÖnÆ‹4^*> o"/0Ÿ.0ÓS9Ôº\µl¾]Ù\ŠPOõ!¸ç9I}w`›­³ÍÚ·A|qöÃTç½A¦JÊqÑ^FsüÖé%kÊj:ìê Õ“-r,× ª¹¶GE•c~\0b¥ K@Qá§.v‚P:wÈ4"„ 2v?[{gïA­ß9LŽĹÉ´_i`PñÕ²Ì9|ì´8T³Lw`2ŽQãvÌ›©7—÷¿ÿ´FM™Jet¥÷»É³?,CqSLÝ6~‘íã»ú훾Ph]µ†Úq"î‰SA‹ÿj•¤nªPÒU¹"UËøÑdqü'±RËàÝDíóOî-¹»Ir]µ)¤R²kd†>Ër‡ö­—w•^+ðÉÈXQLÚžñûñ¶o>)aS:ñZ\©¶î=õZÞÄ ž\§{ÊBýS3òJØIWæÞÞÏrùc@RÉg÷‚BåÙ⽯NX28IÄžb[|dÄÛ=ÿÑ8:·9õÑl5ËÜȳ~rlI.K_kD6³††¾ãKXY±âÈd4Zï Ÿ`-WìÏ×.Sµ2åKÅo¦^¹YaF*ˆgˆ*Ê*p‡Ê0Z%¨Ìˆ¯kc ›8/¥°äÃá ðó‰œèÏ›JãzF— %Á¼.G]àŸV릾Lû”ë”c8° ¿—mæ0šg—u^Q¯­W˜|\íé–fZžnÊMsaíÛ¢òaÉéd¢‰7º‘î]ÖÅx[õ †+l½=F}—ó[ïhÏ›]EÏÛÞIÖ>NÛìì[sŸcp"ò"£ªWOòr‘¢Q˜]$;Rï dK„ZV,dÙz«)É1±û‚ =7kÒFDc‡ ô6™¿66KÞ´àhSy8·YhÇVѽÞIËÓÜ@ô^ƒ¶Àrç1ßI3÷ù‘óPXÕã+šž-ñ±·§ó± ¡“ú1ÄÝy¬Šè,-õÂ-šŸOOr9Ò¿á?Íåšý2÷Jíi-ÌËÂ$¤°Ö޹¶á‘¯Åç¿c‚wMËvš=V3~VŠ2,xEj'·PcF‰CÁbÖsî¤cZoþvOuNzSѯ6mË©öS7þ’ÁûÇé1+Á92~,…«ë]çÐVŒ˜QÒ-˜¯sbêýn©0µþz eeh¾W…orüG÷w¸ü^å1ô¤¨Ôƒâ>¼™ÇNÒîÛ™å«<£§ÚAÍ…a_©ƒ3Ãçõ”ê¶Üuö7˜ÔDýˆŠv‰÷ñór-ëòî%¼Iáòݨf_k6™>ZŠV+©€‰kBhMïz/ºæYïÉ\ª1ù°• …›iŠØ;ïw§Ö„P{ó}®m˜÷žÔê~êÛ¥ŒÉÐ'ÌCú}™_eÏ¥!šæ±á¦…þåˆ ‚ÈÚ÷ËÔsM k)ÍïÚQ°ŽšQõë| £ Oä¨v}˜^gÝ 0¸4¾RÔááô1â>Ã…º‡AI<”)v¼û ˜™Ÿ~îgP]y·‰´óÎ(ï“ ñ¾^ßÎBY»G/£ÌØ îzçþ¯Ï—©ŸÛ+ R'ž$̽mÉÄ”ðÆm]7ŠTZ àëAù·$ö¦‡ã#è9÷žKÄô¹ŽMk<ËG8VïižQx:æ·ðÞâÆ¹à²¨žysDÔ­‘SÖ:ÐŒ-tÞ ÇGœ¶·³I›·/áyvø.HbòÄ‚b8UD>·‹8z…ˆKó±T*“eþ‰­zËÛžq<’ó%¢c ×1úÎeÐÞHáH%âX<¾WÓCß~!Ö%›SÀŠ‚$˜w&:2g´k§†HË$ÊÿôýÍÕ…3ñ]p”Ø’åî./S‘MBk©õkýF(]ÍšËÛ³Þ8†A||d–ˆ¯YH<6ÝäÈkŽ£¯tFù{EgvšyNååÃÑŽ å´p~ï{ej¯é—‰ ‘ᬢÝ5“üÖ±!Ûß–)É9²¨EÁ‰¥ì–Hb&9JqÑûP¶òGÜâÿ¨dß]­É¹d¦Ù…õɤÍ{þñɰýGŽ[ðý³±ÛµÓFm)äÝ Ÿ´F—HíoY)CK®-Š%‘Ò9Ÿ›»î.GüðÍÂH{ˆ÷]Ä$t2a½ýT/ÈïüjÒWæ&&ý{¦KD×W¯o™)qTÌç·¶žÌIHÏeï¢È]ÍýNÆ×±\P„\i X,ëÛ4QÒAZUýó9:²Þ->ÕZÚæ!þ‚4”AörK4/˜È†º“D/ðxmÆÙåxÿQ¢N¾ºf%-èÆúp[a2@¨xEŸr«S‘ÁäÔŸùPB¥ …*ÁQ†ð™¥ |3ÇÇ·ÊšTDè¼Puß~ª³V8G1cs-kW¡¤üÎrûÓ7c‡C¤–Þ.N×ãG3D¼Jßx­¹ÿáVÅ ÷Œaožèéj.Ý ×Wʼn ïÜm¾µÕ×\ÆÞ;í ð~Vä’gó)dC±œ«bÚõÙÊÒj⣽®&]¬þKæÌodîÒ| žÍ–Û¶“qY •Ѥ‘lOÓMc°pjû©ÓuÓdlæJ)Ù½\NÇpui§®¸ëx§JNCÖ CÒ£R÷ÔØÎW”ž¤µ¸pàTrœáý̶~I^ÀQ3ÍÞü#«°þ¶¶v a}m(ñÓ>wÖ†ªÁs¥¥u!Œ­ÇÈZ#=7ŸD*d[ËXõ…^î3.žœÊÒ|ãgÏkT¿_è1µjކ‡s•·Mšw€Å(äøËÙSÏ v´´™ZuRöNøïî°›kð$ÄÀl^Û—²™¦Üøª0N .Y÷úŲ¹:½­@rýJ«z¿O“GH¿•ùD&5 7á fÕ’iêÊóÇ&4r.Çø^Ï›ïê-²¨Ê´wõ÷)FàÍŸå«iÎé9ÔúëÔY4áh ZoG*Qy/NBnmÜ’žJK‰¹ ´F+«̶TûØò»¤)µÔ]r÷ adé¸f+R¸¡Š’nE›QÄóê-kHÉå1½ŠðÞnDRíbŸ[®½I’™0‹þyû:I¤îmt k 6JÖK¦*œs5ÈV »Ãߊ¸ÕK«àZ“ck¸è-J½:Á|rz>—7@§„µ’¯—Ì*†B‘b¹©yKiÚ <™ˆejtã`Úª•úñÙ05­Ë½}èðê›’­\ç£V¾$U&QcÞj~Ù <¬S®ÏšúS¡üÔLƒSå2;AGȭŽOÄ#ËåÙ2Æa+QÝ^¿g¹¨¡òÓ•¡ÒZ® òÂÆŠÊ·WôbѵÆÈàÌ×3¥}^ëƒ=ÉøFy1J"Wj®ö+Æx=d"ÃÌñÏcX}«ãë;£ üSÔÊÖæÄsŸoçWái»¯/±;ÞC‡¬.ˆÆ àG{Y–£–½úXæ÷Ãtçi"Ixc^ÆUÞ2°è¸7žŽÐ‘v|ø•à'[ðFO›@%ÅÉ[)¿¤‡aO´r@3ì’sàø¹³°ðUŠÇ š‘ö8¥½›¯ºJ7H)äc²*RÓlÀ3&–>M_}M¿/ïWݔó“£HüNX¯{Ä%x¯gŠ[7liÜA\*\rD"E‘¬0‰S³f=Ùq¸ýé- Õ I¹.TÊ\ž¡±î(¨ÄQPÉÙ¢ðK¹®˜l“é§Â‚Òv¬!³*¶7ûôETÙ¤”]ÒÆÅofu>‰k!ì?!gI„È"d‰2ÒÒ¥M ØÌRí(%¸TÄjuÖð2ìbÄ}z-cœ›sø"¨o÷‘å¶ÎÆQÚѸS?3Ç Ç3O;ûùÀáBqº•Á˜¡Ò£—#¦*‡¸?Öã¨G*±r\É”õÎ>Y1:'f´w^o?îTJÑJ?^·4Yè{'§âm¡Ú7x+6sã’FÉöMžµ.ø§ÍÂéÌo—l‚ýx$(õùz ×µkÁ¹rsïú¶'«ë1ÈvÖ°-£‹l;o]oµn—[o JO\Œ¤õqáí¢èºÊª-Ʋk"©åÿ,Õ•Å¢¨¼¸EŒ*ªŸsÇ Ÿ~BÓÌg»:Ñ}¬m¢J9È/r3Ê失.h*! Ió+m_ô4œ ¡"ÊoZòË‹º}™Šë¾žî'ÌæõXPNiE;äařߡ¬å˜Cˆˆ9Q.ýL~È€¯žôƒ»ÂsÔ÷Y…ôÜ+TÏ•VÕ$_]‹˜ÃÙs(¾ÝüXtü~†3÷℃!éã;ñ$Ø£î³5VŸÊ”¬}öÐYÏ©ÂÊ$£˜/tE–ôÖB=º­Ç£Tø£5…¢Æ‡˜¬Ö'¾Û?ý)Á_Qéc—ra’ù9Íòf9©!÷¦•†v‚V£oé~?Ú±¹3HPì’”îÓ´T¢!zO)MóZ ›Ï£:Çù.~ùçAÃ̯õ/&”´N N›¨<Ó=n}é(¢SmZÌÂe%ög ]SI– ÷m²Á!í%r¿d¹xð±ÊÈb4ƒgy ÆŠfR{nLã»TŽê}ˆr÷œ–t³ÓO¤r{ìe}÷‘SrÙK[®wºá–s°*Ï/l¢Ç«¢fo”È:•5tñ©ßE@²Øíå=,!E&C]Ûþ;1ªß·¬H„Ü+nõ¯èf¹OVÑ5bºzÿX7jZü1Ä?¿p×öáY¦E–·×XÒ43X â“Ó´Ès;…]öÞ0yŸ\/SÙÎÃÈ5Ù‰U^ò_{oš—üþK¡[]ªä“v(í¥h+V cdõŒo9Ä•/øpJÀc—ó©Ýmmû F×ÖW¦ô€s6·ýA_JFÙOB—üu±÷ºÛ^ZM¿N¹Ï1ÁãÓ¨ß\­;óCÌ ¤«ŽÔóýæÁƒHý^‚SEûñÄ•¶kÃëÖÜ,¦Ù2X–íá™·TsMº†×ô•YÜwCÛQ.®%NÚDòxò]`Ö·/„RÓa݃Q(Ãó¯=,ôA{›|.ý‹óo˘î Ö»¦«€“D.¸zæ«‘ÅÙhîÖ4Kÿ²a©ÿ†æô°Úv'ϱ÷ @MãïNA¶¸âÏÂK¡éýDçn~tèÆÕê9•Ê6>÷*ØDp’\Fû‚‚ã.OÐ#P¿É—W’“4Æó‚M²‘³ï“™H;õM[^„è^›HæÔõ¾lá58‚RÊÍY¯–>ÏÍQÄïnËìˆÝI1¿ŸQwvñþp›[A>¨±¤Ê—-w5±Ll DòÖépn¼1j©LH$ÜŒ4ÈÉ1LÒ|aÚt^0TØc¾p-;v·[ÐñDõ¬±;p+/ædÝí².éÚt'ñ:¢£XN·áÚš7•CàØI–¿¦k`ÍÅx_®£ÑmãD^sq ÷òðÌ¢Þéméì‰-šâù0£À®öÊù™þƒèÂð.õãï[IKc´¥JY²ÂT@åê-÷ow´oÞBgŽp®YÃçÕyÊÆ‰'­1§^ »êh‘"Er½×QX†Ì='â.OºÞeJÞÆ-”¡S©ÉÛW’éÆÛ_:ñš>˜7(X§â¾Œ0{mýAðQ¨>vÙ7Ãd°‹ý 9RÒüÿNßÖ¡øŸª§§Iy‡é÷17²pçbý|·—7m Š„ö¯ŽûcÇ !šä’=Y5R—Cë¯!–^â]“ŽÕ|Û|‹+l˳jðº.r vä¶È‰9sÍ\%u½‹Uã»ãºHƒgõMÜ*§î;Óª„wæ(G·;FÊófƒ½ 7¶G¼ýõn7ÛM߯ [½¿éI§bÓ]Ÿ³$ÚÅuËãdšjÙl YQ§ªÛ™úFo²G*m+ó‡Ç¼ÈsjYYâ‰WêÇ1ú[ùÛg@Û#5œ"j»÷mß½bX ¼…òsýêVÁáÌÁùCKÏg#¶ñ÷ù’äSƒÌÛi.M<Ø/}™ž^¼ZD†¿m¦ñÚòx‘œƒšf6Ð5JL÷½^1e„µO:7Ê­9ßÜèE5š'VpRë–³rÃv¹Y ×LX|Z8xÝžÔóÑ…Ð*£‘yô™ðC©„"Џô9í'ņã¹ÜµáÓ¦x…Hˆ¤Hß" &‚Sp+%ñt²¦-MqšeʼnèhM]›7!C7$‘Fô3É`¯©b$;b/ÅS)cÃ`c4Ý2ç§&=>C›s±ðJäè*t›€eÛˆFšD±…ÏatÈ»`Q±eøyºVèŒÂ Ÿõm˜j§ÝiggHQ¸à(*[¢7ʃ–1l^urPlQ/þhqFë^é-xDÊì[ß¶¨“]ò~ŠÌ*¦¬(–¢«@n3`_7ÃY»íÊÆÐÂøÞ^ov µ†Tg…‚ƒ3ˆv|È|rn°¿ Ó8—xTÎ"²‚‚Ǩ°ôT{ª8ØU.è½ÓDÜgB{m*¥ž*bØ×›Æ½Ò;„–å—¯gKÉ\?½K4ñ_Kï"6‰#¡-}}Cö™ôö‰‚FÕ[t hI¡‡·m^Vê’üîZhüºÉ}´iÆÀlùà-™zÝ3ˆFrœR¬Rø6ã!ƒÆJ8cÄp~®Rn£ž­¿ü]ªÞ ò»ÿlûÒ‘ endstream endobj 203 0 obj << /Type /FontDescriptor /FontName /BVQHUS+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/arrowright/bar/braceleft/braceright/minus/negationslash) /FontFile 202 0 R >> endobj 204 0 obj << /Length1 1400 /Length2 6059 /Length3 0 /Length 7015 /Filter /FlateDecode >> stream xÚtT”kÛ.Ý(Ý1( !ÌÐ!!ÝH ÃCÌÀ0tw#!%%’*(%ˆ¤´¤„ ÝÝ-(ðº÷·ÿý³Ö9kÖzç}î|®û¾®—ƒM×€OÎa UFÀQ|ü I€‚–‰â‰88 a('è_f"#(Ò †€Kþ¯$ŒBÛÁ(tœPwwD%Ä$A € $ñw ) P{ÀlZüuêFÄ¡€pñFÂììQè6¿¸ Ü ±û¿ÓrÎP$ †´À({¨3º#ì0@@`P”÷¿JpIÙ£P.’@ §§'?ØÙ´“á¾ð„¡ìúP7(Òjø  v†þAÆOÄ0´‡¹ý± lQž`$€68Á P¸:ÃnEÐÍjš(üO°æŸ€û€¿fàøO¹¿²‚Á'ƒ!„³ î ƒÛlaNP€Ž²&?Ê u†Ûü ;¹!Ðù`0Ì lø}s0@YNFü ž sA¹ñ»Áœ~Aþ*ƒž²ÜFáì …£Üˆ~ÝO†„BÐc÷þÙ¬#á ÷ýë` ƒÛØþaãî|‡¹ºCÕÿ A›ˆþ±ÙAQ„¨¨°8ê €zAì¿Êz»@;~™Ñü}].[4¨?ÌŠþ#òu{@(¤;Ôß÷;þ}"ØÀ (€5Ô'ú§:Ú µýsF/ ó˜ÐÜ€~ýþóöM/ÜÉûŸðßû()ÈòþAüŸ¼< àË' à“¢bâ‚ÿ—ùÏþÿÛª †ýu9Ð?Õà¶€Ä èáýÃã/Zpý%nÀ¿;h#Ð\†¸þ¡¾9HA?þ¿ð;åÿÆû_Uþ_Ôÿï )»;9ývsýöÿn°3ÌÉû¯4•ÝQhYh!Ðâ€ÿwècè)kAm`îÎÿíUCÑòƒÛ¡)Î' Ìþc‡¹)ü 6º0Äþ‘þ^º‡ ÕE¸Á~}qÐY ÐùЪƒ8¢¿*nè…ývAÑ¢úw_%8aóK}‚"¢0 ö&¡I&("ð@ËÔêõ›ß ?B§Ðý¶$ѯµ €`74f˜›#z ö¿œDÿjqG"ÑòûÍt÷¿Ï¿µ…zA!D“ãȃ0‡ª°ÆïrŒž|«xó‹Íщ&Q"(Ή¾öšøÏUF\å­lÊéúÒu¿½Š÷cäÙû>èeþ);x.M5…¡Ô¹,À—ÿùâø“”ÅÁ`7$–½L¾Ÿ¤Ç|ÛÊ Çì^ÊãñŸábmBëo× ¯šÝÖM%¶΀â32¶4FT¼\nó6V¬ û"ª×Äœ,i–án¨öhè +–p.ßGá½l–Ù;;Û“ÝÕ™.tôR£ýhy%=•:[n_r7=uR¶z§¼È™Â«ÔÄ;ôEΜsØÎUcLLì ô g@ “CŸ, 3©s[}¨¦¡q¸_³GŠí§7Þ é[HÃT0¸G VÎÙô.éMj^k>ŒV©YŸHõ Ú›°çÄtÐõí¶q)äÍǺ¨ÞÒÖ¹ŒYXÿ”™èÒ |ª÷VDqØtÊ`Z³ŠÄOOtWøÕþxÍ\þ)§C㵉ç*‡Ûë˜Áãò^ògøLÛ³pßÝÃPZ7[Ô²¼TÕ:Šêk}Å ¡\´žrÉ7µ®UúBÞ˜ƒ»ßg˜^©Ùì}A§Ra°š1¿¶Î9Rfdx©™€âG<d„|”‰ã_Û•·nÖêô[}¤„YEl@ÕÁ_#A‹HÑ; ŒÇ÷üpÄ¥2,Åp·v3h6š)ü­¡ÄÁ§ÉÇua˜å>^jQm_ÈTÏÖ;¯ÅkBLë ¾ÚªO)hShz÷)À%²ä ͈¾˜ óòמ©ìÎÝžêOƒx˜­[D†óÝÁº¸†àx@>°ÏÝͼÉßðLÎ.tÙtZâO!œîµp:ßDrÓ‡ñš¨$ŒüWq|Æ3ðCÈ®~ÿ4\ÔÅþÛsSz[Ÿ#Ù^FbªIúKö(“‹Zu0Æõ¨·Ÿ¬åÙ~NÜÚÁÑÿ±€íaËùÇ0Òœdu*yòQÏ{ÂÞË^‘ù+39IÂzL¥ÒÓR‹—^/‹Õ)JA}ýÌ> @Û/=KA:ÃzKsÅïÍ_GEà¥dT«ŠwL oðRâ}+Æ'Ó]`š€¢~æÌ“üïNeÏõ‰5 Å™:Â\y¤Ý›>:ã&Öå¹%Fàªìxÿ€ø'—œ+‰*B\HSTJ Ö‘Ùfš{qÕ$YE‹S½­?UFIÞÅí蔨ÏálÍc–üìIBdXPó8a—¡ ÷Õ¾’ˆ˜æ‚ÃÒsë‘~Ë0YÛáMŽglå£+ßçMî(Ò°&ÑËŽ-ݪy¯Ø‹èúÿãÆËIH/&pß«ÎØV >µ 2Ä:àB5€ ˜6"«ŸŒqΙoåŸ3ó—:_»"jˆWÄãÑœ 0*tËËkT¾¨ø÷Î?h®7›öì›âíèÏ»yg븨sÆ›•àî×ÛÓá›-ý;lãÆ¢J÷¤6YR·Úœ 8¹÷Ô›É 2½YÈ­ÄéªH7ªßRO[u½šð¨Ü´=lÖ,Kýk“ž=a\.fºø»¥ðêÑú‡ øŠ´ê×SÓvø~=Ê•V†‹«Žû©”OÓ3rF²&â»"ÊïÆ°ö³ä|Þô²î´ 5 ~Høæs.á›;6JÓµ>÷øa K¬ wdÏ顨€áãJãßSx¶øGWi×ív'ÌݵçZýÆ¢åJýläIª'"›œòÞïüLnû\“*£ô%HtMK æÕâ†å¢ˆqT©÷¹#}½ï¯T=§›¹µY+‘ñ¶!™)k´ôÕÈÆ‹ŒŒÑ„ ÃYÊr·< {„+SNŠøîÒ0Œl¢›=ñ:•›ƒêÚz—‡yœMðzVø †Aæê2J/­£b•>Æ‚“O¼Q·ž}°aà!³´Œèä§ìõ|¾4c=;]ÛÅ0¤ßÁ~o­ÝF\z¸³?™Ã‰»W&Bâñ&Fò.ø<ʼº¢EZôÀ8´Ÿv1Qy—#pʆu#«ºL™mèáü>¾¿iŒ«ÍíÆÔ2oëÂðã6Ø¥[³ÆY—<ð·.íuSOM6´[ãåлþ«Œæ4cŸ¿û†P¦wá”]ñ×Qã‚ ì¦QjáÙñ!¤ÊšYìÔþòqÀãÃÞ1L!©v²·‘Š0Z”¹eDEÑM@h ðU^LRqÂÞ­¼âç‚`®<ü}#Á©FmÅÍï6´}îϵÅÂq$8ºŸú _Kè¿ÎÏq³cn¨}M_â#FК&;äp(‘&Iþ#£xˆU¼2ÄÁƹ“7ü.-±Ø6K»¿RSâè~Þ˜¼õè¬hLAY±h5òVÄ Ó/JÛ‡0Ws|‡%Ö…¯º}ܳEf]ÇêνΨ;.ÒyÄñ”Ë+õÞ>l/Fó1IdÚ™6ŽgކP¹s¾½ýG ´xŸ©éÜÉD3¬S«êY÷ú[éÀ瘸‘.éß‘¶,¤3³U=‰9ͨþ„Õ;n¹¶=J¿\)ÊßýÐÒáú¡‡"ê]¬JiH¦J³®”\ýéö„Úmç¹€ÇUË!/ÍCéúw˜åòß^¯=!¥)3ÿ Ÿm„’³qõlPDm×Fs|1‡)6\ÇöeùŽ"Ù×ân—OÇf‹GÉöo0¨›¤Ãå ±e žòŠxsâ×Ä-⨊ |dÇcòy{ ô‘t¬ç;£«$¥•±} P[4É mbºÉ%ul JZQd=ýöò¹4 RI@õ¹mktj©‰}[8“‡Œ÷šY/RNoo Ä(ˆí7–ÐÀøºö‰ñbÈô›õ¯±$sÙ<„Ï:-qôç£ L8°j´E>/EŠîŒZî4Û¦b—E"Íè] Tf¶yû‡ ÜÞ½ ê(l=/ß ÑvzC]BÇ¢xG¦3‰"ñ Óº…±7G¼bjL_*‘ñ¨©dË$Œ®š<Ÿ#´ö[‡L¥Ú$å¹)”‹Ã¤íŽc ³,eê7L•›Ä$švL Ú:WZ[æ‚Õ9ô„Kv,}¼RÉþ²Ë~Å–²u|X™Zk,žÙwv-µÆf% }k`ÍQ°“µ¤?»ô=9u·ÿæÅ1"œ´äÔ‚Ï"?»y‚·(ûäða”COܬó2æ¦v@Œ¤6ÆnÛÐwD$Î%à©#J…jÚs,ŠòHPçá}ц-]²1ýû÷Z§óó?ð’”t 1Ë‚ÈJ«‰#çât§qSބĪbZ?¸'ÂüõàôÝÇQÊÝ1‘69­ë»íó_°Þ}1À4°o[Ì(¨J.xÄ6¸ò)‡˜L ¬VOÕ-z­õ‡%¨ñ…±åY•ÚpU‘ŽËó"‘Ö|p)i„ü6G×7‰ÒO¶LÙæ¨¿M=x\¾È%ǩֈ¦×uÆ·ÁúBÃ1]Üé‹ÇÍLèV“¾äZ¯ú¬ü;6ÿA¨Ã3aºÏ¾{ç;ó¾Ó|‘޼ª:dû¸óWŠobŽtú%Uiö½ãdÞº–qáÎÎH0ãáŽ0s96ãÊ_ÆXÔ!ôGÎvçúÍ€•L ‚³$UGØ»Ž„Zµ¡ ¾ µ.¡ß0è˜Âʾ®‹<„zÖÅ1É^07žgMêø}¨z°ÜœÖÚ±à­<½³%pðšô@8Áxd_lì²ídï™Â`¨U»T×fAÍœ“ÙkÑ}|ä ÅR}Ñšµ ÿöÐéÑ!6Óo=Žía2|4ÒŒø/Ô>Sª¯ñéËñè>¬cso‰>¨Ýåîn÷ÀyM)xuÅîóÃ2<£%ñíƒÇ?‹eÙבÕRß©û”H;%5¥Ýiö ×ÔæOAÍ©•Á?7X»²W|¦º[6ÞŒ‹ ÆMµƒ<–õ̬êkXâ~žOurÇó6Hú]>}g NCT1º%xa ”¸ùäæG›ôÀ¬6'}Êí¾*JUc_‹sÚù'jLŽÑñU½ãù~aãJ5Xi¶Ón1D5E¶Òˆœô¦¥íH"¾g˜\ «2)…-¦­j™Ã^móö´í«yE^~[NIŽnŽãÒÙŸgÇN‹0V6\7Éz}‡“Ÿ)”ÙÄ÷ÌáÐônXaê.’³…aéVIøGMĶZc£¢Vö ^ËÓœïæÜ)LÙ'QãÓ‚ 9ù6{âŠÒÔ ‹@%ÀD°•ùVZ’ŸŒùœSkŽ…k ‹ž€‡–ŠÅÄ7Çþä?¿]ÌÂAãc™N÷ã“LiSµäOïdMrØTí=ìò—‹ÞiêýZ˜–véáú5äÅsª£>F»ÇÕË”ÉR,×d‡M/o¶u”ë ™7Vùbø¾ï¿DåBº8ž?ÆÖ—ÓÕ£m\¾Ê|¯K¨u!RØ4É2æƒNZ=fÃ%ûié½6½:ŒÒ3bÿ组à{ƒµÓù[Ü‚£ ²Ò·D;äkI§÷ò¨üêRµEîPîôé©“ß©GPóL^ä/'¬I€7Ö]ÒmÀb™*m‘² \L£Je•æåœ¦Ô–ë’Ôž@é‘æÝŽ+ü»§¤‹ÌN…ÆÕe?äõS³'d{?ô8#ËbÙ¦À/{Œß¢Ê³O›ÍÏ5˜ôƒë-…k';õGˆ} RØX¿pQÌÒðÓ†úÏMðÉ÷ò†õ·ÏeÞ,?ûÑ›º f løZ/¥%ö>h2¢ÇéYÛîØ‚Û †›d¿®S‡P?·7ï©ákqáJ첕¡NCQ²Ò’’ÛNgÙ£·µõly¼ïˆº,]ã8ÉÛëtÈì²:? oêFÖ^31ä±È•e’XpH¦¹u™cxµ+Â"D•…”_éH5–ò®©iØŒL¥I½¶îi­óÜ?ÚWWë:¬5û®X1„·(*œxZRÚÓž|V¸AUjÙr¯?sÃÚïþøðˆø ¶U9'‚ßQ¨I<æÔ‡›Á²ó}X::D§‰h¸'ô°Vî‹_5– |ìfŽþÚÁD(©Îø¡Ü²ü§¶aß57§;”ƒ‡çb€åAÇÞË$èÊ0ÅM¢^„v™²^VÐWzÌ,¢g{iñ<°æéÍ:¬Ðeñžª|ªi­rFg¬ˆV³’ü'Xbuª!¡ÀÍѼ7LÐgåÀÔ:HRŠÒ2Ë\¥$3A­&Ø >[mHjûmü8R“ø\!åî°ýs"Oùc‚Ûol2‘Cßü¾×øæ`ô®Šäë!f›w›·sduKÖ¡Q_Âr§ùùú†ÊVŸ8Uuµ|>*Ï¿~ŽAMu-5¨ŽµÍ§R2ÜñЇ å°q°¯ìàÔ"É}¸‚!7=ª©©™ßñFn~Ú°„N.´•Y²æ.¡Ì-+ÒkŸÎaÑgˆ]Ý?îQ´Q0É"ªQîÜRõxEU¸·@0P°ŒxjUÉ¡«®G^6Ê”ò”­Ût3ú‚”ÄÐé3#gÉpH#ô<Ƹý™è1¹ÜÝ|›±õJó ™Bˆd{c’ÁÉA™>׿õ!ÎîfÜ{ÊÿPŠ…H÷U<ˆç]g"cËŸvÐM–|øJ¿{)¼Œê#«Ã+Bÿ×Ô¼d‰ºo”ø ‚Š«Kƾ QÖg'®¡Be¹ªf Gœø?ýÉ#üÄ«y§ãõ:æ§jÂíI!çÃ&Ñ—çä†C~ý@ƺ»Çk'ÍÇžELºÚ®9«*sñî‚'„×Ïön\Ì&út¨UÛJ卑ªW3ÍëÿÌnÍ^3Æ¥å­e¤ÁÈyËO®Ü1®/pâSF„G0¢Ø#‰E»dW{ä[arä/rC!s¹'º?»NBÜ¥‘~ gêDRò­ó&¥˜%Ǭ¤¡#‚>z#ßÂâOÍÄïVò¦ˆyÁb<[EæFö¼òz¯ï±zÿò©1ô„x8m'ΗçhÞåE=â³WêÑšF[9¸öR¸Ç{ú J¸ptãeµ×â(ütk2÷³‡«†åRNë…ßBù´ãßµ ‚A|ÛØÅ¯l§2' »ÄÿS*žS]'ERŠ¿˜øèƒÄ”K &}QŠÛ>áTPDÍá7ï IÚ;“²s½ïWSèÈ}É™S[ øˆˆY¨lš 8‹,óJòLCí‘IèxDàôLœ4$F¨ùV.±f™Ó»Mönwz«¥&œíõ4¢oc÷(näß¶oñÙíÝ>ß/_é ‘X—p)º¹åc Ø’„[¨x•&IŸdfK„¨Éfµxpéåר[É‚Î7Cýe²Rº£…€"J|Øœ8,K™µ<ŒRP ¡¼'úY•Éý"M‰®§x†ã¢(Xþq áôwÀh¸`|Uë¶dú{¸äµ&i Væ-‚8)w¥tC÷®Ì¢½¨¸;lú,¥Öd%¼ŒÏêôƒ£µ™P8ÊëcŠ(ÉÙQù¬´s¶H” ¹µ ¹ÎšÞé¥8²?Á¹²1.ù‰.Çuÿx+œD§¢0Káïošê endstream endobj 205 0 obj << /Type /FontDescriptor /FontName /SDVCAT+CMSY7 /Flags 4 /FontBBox [-15 -951 1251 782] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 49 /XHeight 431 /CharSet (/asteriskmath) /FontFile 204 0 R >> endobj 206 0 obj << /Length1 1918 /Length2 14135 /Length3 0 /Length 15323 /Filter /FlateDecode >> stream xÚ÷TÛÛÒÀ ãîî„âÅÝ¡¸»k±Á‚w).-wwoq)”Bqww+îy9rï9÷ù¾µÞwe­äÿÙ3³÷ÌNBK¥¦É"n 6Ê€!,¬ì‚Ie-yv;;+;;' -­bü…Vèâ ; þËBÒhy•I™A^ •ÁŽ7{€ƒWƒOÀÉÎ.ðC°‹ @ÊÌd Pf(€®(´’`'/µ ä5Î Œ>æ?Ýâ@…™#@Ù btxhafÐ[€€¯ÿY‚AØqdcóðð`5spe»X‹22<@€Ðèâ´üQ2@ÅÌøwi¬(´-ë_ M°ÄÃÌx؃,€Ž®¯.nŽ–@Àkt€¦¼@Õ èø—±Ò_Ì€¿7ÀÁÊñßåþöþc!ãŸÎf`'3G/£5À d¨Ê(±BW Ä•ÕdÿGl,óºÍÒŽ–’` #Äåü¤@.@‹×}÷bûûpíÁŽ>ÿ!+£¥ÕeXº9±i;‚œÝ€òRÛ¼ŠPþ‘Y!vvv~N~Ðô´°aû#€–—ðO%Çâ×ü|œÀN«×2€~ +à늫™;qqúùü[ñ¿„ÂÁ°Y@æ@k#Ê?«¿ŠVñëù»€<†ì¯íÇ`ÿãõß'ã׳;Ú{ýcþç³ij)ëHj1ý]ò•`O€ €…“‡ÀÁÁÉ à{}ðûßuþ»ÿ©þO©šèïìþµ¢¼£ ðW¯»÷ŸBÜÿî †¿Ç†ð¿TÀ¯ý 0üÓþFì<ì¯oÿŸ‡àO—ÿ½ÿÇ*ÿ¯íÿ3’q³·ÿSÏð—ÁÿÞÌdïõ·Åk?»A^gCü:!Žÿ×Tø×@+-AnÿW+1{qGë×>gáàfeçþKr•y-Õ@ ›¿zé?‡ñÃäT»‚þ¸w^½ØÙÿîuô,ì^ï××#ûS|¬ÿ+íh¶üc9yxf..f^(¯ðJ<Ž×YµzþÙâ6VG0äÕðZ£À ì‚òÇÁòòØÄÿýE¼6‰ˆÀ&ùñؤþ!›ô‰À¦ð½ú)ÿC¯~*ÿ%~v›Ú?ÄõÚÓÿ%W?³èÕÏüzgñ_âyÕY€í_7í?’?öŽÍò_È`þ 9lVÿÂ׸ÖÿÂ×ômþ…¯{ú¾¦e÷/|mÿ/|MÌá|K¶eõz°ÿ…¯i8ý¹_wÛéuÀÿÊûõû†Íù_øš™Ë¿ð53×áëák¢nÿÂ×DÝÿ…¯‰züƒœ¯‰zþ _õúÿ§½,Ü\\^oà?ï×ÞûÿyÝž@ ”…Y°…P¨m}hû]­8©ËÎ('7zßEÔ’î¶4±ûžz°¤ðý%WAô¼ø·éÝèûuœ#ÎÒ—&;=v;“5ß´hA‚…/íðVæ[̾ã8"G®Jnœ»Š]]¦\Ž–Ö§Í?=®ÞM12;ŸïÌdÈ Ó5WßVÙœ"Ò7þbEƒF$âççBƒ íÏ쨖¨Ÿšî”§q½x7zœH.spŸöcˆ“g³o±œ´‘Fzxˆî¤x•y(šºßÃK¢h:˜¢Ýî‘&úý‘!øª=4ˆÁÜuXSŒZ(K~{üBîb;V^úY®öG ¬äúÇ7gëÈâ9žæô›&Bo»¤J[£’´Š¼¹xèË{**dqEdQ{ç‘e&~…Ö³æ`+ú*Õ L$˜h¹&eM¥1eà6ªi^úÌqLX E‡¸Ä˜%:È·4B ²cmçì߯Aê´ Û¿cÊ¡ÒRaFÿ¦8¢Ö)$óå;–åá§ ÿfk§gkù‰“ÀvUwBÜÿÝò!<ËqÙ׫;ý^6¢w¥'2»Î Í]铬æC2Ki Q¾õ^c¸ä¢mã1]RÙ¸þÔu–i¤ëL7åå¹vݰէ(Ô ¿±<çX æï*“|"†x{üO×Ä¢4»êË{ܯi>7\ÞМ׫Ơä§Sæè,¨;‚ZÙG℺¨¨äâ-w+„.5õéwó/¨o‡Ú¤¨Ú*H¸îÚù^G=¢,RŠhÅjŽs1õ³b‹$  ]–b ­é¨ì×vPÞôEÊRœÅ‡œò4€™|š‘JÃÛÎý3õܥ㷰xB½²Ü$Xx8Äì 4£3zî$ouc\cÔàž«‘môžµ…HŠX˜QòéõOòÀøø#*§yïÉõ¡vim`&]Û'‰öþ&ßHG0A™Q²©ª.þ-!ðÃ"N náÑuî1‹«5Þaj\ŸýRñ (olïíKZaE©{¾±ŽI\\óW±©Š¢e?5] —ªÚYg s>9_tDúÄv~£÷5·4Sª ÍÃÀ²ÄuÚ£¶û†‰†g×õ³ÆíIœé¼¬a¾çØ<™²€+]c?‘3-Ò[‘¹Æ+ÚħL$ï„.qÚ£wírã&)×/'-Ùa*Œ³­/{ƒ‚9‡nú‹…~GÒ®%:Žé2„àñ¸É¬ÑM®ZÈ#ZéÖ¶C*ŒSîá;¹8)¯aÌ/ʲáu0`¾Nhúª^—UEß§ûlút`³®ç*+þ¦a‹vóSßHÊ› ›“6á—¯q:||8Õ6Ô¹IÔ¤iˆZccQy48%Ih ë¬éV '®ü%?e ?Óž‹äO…OAi²e5Fl6Ž.Ƴž ÔÐY B[¿Oã?hîs¶{QÑû»n,ËÆu6Œ«Ë ôKEÏÞ–Ä$pfÀ Ù˜ Z“ª[§î™!BTÖ“ ¼i¡.‚–z{NÓ_Þž‰³éåe”(P…*”Æòñzó1K]ÜãŠí|¢”KÂ<›Òš^ÕÈËË#'¦ˆåBÍÙt ý;ŽéÒø¸óÇ ;AeX}ÌÚÑ|o÷³Žˆ1š6!¿ÿðèöF,`éâŽ;ü | Ïžù%AW‹íü–BÈ3ÓÝ| 8‹*z¯¼„Ñ+ÕDiâ0çeÚÄZÅ^Gðü0Tuù¡ïr–†õãý'\É%ÂÑÉ;K—mìoúyÞ=\ŸG¤HLÞP2ü^)÷J(+¯Yzp­€ s• %‚TsSð!*’ÝÝM?:[å—è¥6ù­ö÷ª- ÉÓ?kHÛ,·öÃþ¦ k „v¤r:)ðµuÅØ;ËÑ&ˆFÁp>\c$w!Õ/Š«ŽÙ'†rVj ©Ð ÛH‡Ò]¿Q‚óÑ¡ZKA.~˜m+’ޤú-`u°)/[·!è-Ø[]°<ÈŒúξëPÅ9.ˆÂÔû’0kÙ8Ÿùšøé;me‡l&)S–:hH†fØ… Þ ÍrŸÙãݲ¶ñëÆ5›–À1lú¦ˆôÖŒ¯Ë{©aWiÔ÷¤t­‡ŸÄn探ƒjöj.:§K÷Üê‚nýŠ©WLŽ’Óþî]7³0TvC°`ŠY3ºW0¥"±>ÿI‡Àu§¾ûll¬†ÀŸÐrènHû§§bÇA1)†s´OÛ"¢ß&"¡¼óTx#qH Ñ0&/™‡h8ÙÞCaâ]Z”äðæQÒ2jК¯ð/Ø–å†ð"&P:œâg|uF’Vg‘¤DqMVlš~‘ ªºøjT,¢ÁE8€ÿ`)J¥¿4w$Ц¡^0šÄ¦3cÙ°Q ´ÒÃô,±· ƒ_s¯KL"~¼L&í”%~]‚kó¶¥ ðY¾öû>6éÅÍ{ÌÂZlm»R]ÎYYºXˆ]1“Ž0MÙ{-륣ùF·'¸+ o±S“/S^u B•yœuËLJWÃ,ówþ|U„?Œ±®ƒF]ªJèÔ“æØ ;3&Ä•³­fôvÅ=)¢A-wi»gŽÚ¢§AùË´ƒÍO ß´Ä>‡¤°â¡›q(Á±~ñh™…}X•«ªx>¸¥–Œe1µ!!D¬¶ZaDÑJ¬—– øNÌw“Æ ‹ÉZ£Nñ¤u» ”Šÿ‚Ax5.«ïò=˜f6opšs©´ýíÇ^¢à‘-99SjÇ‚,œßŽŒõîÄþ¸¦Ó×ÞûZY[ÅÝrº}§š±0ò†ãíTÝôнQ¯äã|àº`AÒFL&½/Ë”k°×üYhàÖTÉgJ öoéȱ‘z®ÅŒ¢â|²¤»^ö|ž$GÊÖ²@÷*Ñø·T–tèG§F&N˜aÿõÓ±.—F!¡–ô‰ˆuί‡“^Ƭ÷H[PôÀüzö½=(£ÒCåˆ Y|õ·µ¾¾}&‡àC—Ï\³õeoñ =ER7 X%Ê\øY¬£à¡Ò¢ÓÜÐ;pqÃjXY½6‘@ûíç.Ã*sÁæÕ{2—Ð6$ZvE’õÓ~Ö­ö¸_æü½Zùi’>QÌjønǨp´oP:û28pFzç[±¯L‚µ±å—¼[Js¶­î7ê]÷ ºèËÒd@²tœƒæEÄRÂzñ˜×ãr£·]°îÓSÁ¨¢z)ÜaD¤?¦é6Þ@ŒÝ;µ¹.œóZz>íÖ Ì)8ïÎZäl#9üœŸÚ*Þ«qø®˜ñÿ*ð¶«WÛ¸áß¿?Ú ÇúE¶)"²ƒ‹|™Väž1ý)ü2ë²b`mi„ÿN›Ç°åýÆ!j÷ãÁAÄapãa0Xðå©™c:˼ ]ªXà‰ÐoHÆ4v¶«ŠýÖb¯iÛ1_c9ä›üM„Ó/¦Ø°730±AÓÒùâxŠzAWÏRÐáÛ(LCyè8ÏÏò Ú› tÇhž»”ÕÕÆÚ‡¿|õ(¾ic»tgÛ,2ú²þ^)é]XÑm¹Çw•Éf„P?™eK=!¤–OƒQ'Ñý»§£´­fÊ¥ßsŽØòOÈÎÍÍ)™ð328W+¬¬°ÕƒãB±ºMááÞ 8îã)aÎÙ¢Âè=b÷P‘*MÞêâæ6«/-ÊŒˆ7žÿŽÚ ÇHúŒiÂIÌø…ˆû€Â4ÿåÃm§õsAFêÓKèa'U½õô@z«ïbÎ m‘¹LÅ|ƒ™ßÛk:ø ˜\faj¨'ÆMýTfïÔûÞÑã]±ðˆ»Å5´TI…Pý¹µ aºoœÜÆŸ®Ì4\8J×¶\kqßI†3 šd5¡C7^4L؋âG!8§>3]˜Ö ¿ß̬•ÜžJYÝQ­¯‰øF'”©Ï¡Àn >ž'2t!F}÷/•ÅŽuu§° HL´«PCµ&Ûb;#jA˜Å¼¼ùøÞXŠ"·±;Éû2Më”™F§lBÛvädÉ`íëUü ùK.ÏfçZS $tfHÝy¸úx!mRÖÎbôfãħ0 Ù·D£õæI  uŸ°ÏÏ]RÜfŸ¦“'ÏK@”ÿHœŸ|ˆÉ1tWˆQ[þ322OsPõ'š¸»˜XtÊçܵ¶˜­–O:v®"öîæ«cÐ0ü¾õ?¿u¼§õ9›,m¥£ó&“GÄϼ®ÅÐ}ÐéÐ|i‹Çžê0­@M%Ðü”òK@k¿–½~ÜÏŠX)ùbö˜Ž×ÁXÑY&Ò¦7MåMËòî 8Û¤'|§VØŠí L®myRÞ¹œ®ç?àKŸ¹™º:;#wEÅœ´Øfšéô^⦊Vò–´â˜ßi…|ò¨@}¾-›&Î9©-÷þÙ>U²N0vè œ¹E‚]±K°ñî?»6Û¾—31¥!L¾áîÅ~ÀD‹^Ç‘[5Éd£‡ é¾è>M >ñ¼5ÿ‹H$ð¨²9û…nºÆo±‰:ë-Ò2Ê,ßC˜È^8›ïÃWÉ"=ÎÌÀé!]‚oTâ)¿Hí~Ðk) ¿Óœ­î°Sã·"®h«Ø¿¬ú0 ØÑ‚â Tdûº±×¥Øe„.ºFK“2½057\§¤©‚ÎÓ##²ÊÙGE=Ç ô˜Iœ¶®KBx¯R=«ÓŒ{½,H’WšUŸ™¸d¤-þØo)üî=áà—•ê3” P!åÜÕQoÈ3Œ²ÃiòWPßs‡BÄÜ|ñ‘È¡{ÿ»ÒPIÞ͆E±rÓD®h‚œ|¶Ä•ì}MÝû߇úgPl…xËrg´rcHV‘‰ÆïqÐÞ|Ø€žåùÈ$ =>‰Ã©Õ‹8ÜüsWòËYœj‹aXà nõÔúÅ$”õ½H¸ðÙ»õN%î>“.ØÎ™G“Xþu<®ßÎ*R7oŠ…½š<,wµšÂó<Ÿ\]*¶Í–qwô+s¾á3˜zÙsíi#‡@Õî¹àäN1™­uªŒy>ç"(ÿõ‡gÄ“öß@uu%¾£SÜ…¶ €ÓTžñxæ§„ÙÎ;¹ï¿~ˆÓüjæ†Äp[`w‹éIoÔºƒ¥}·;•;Ú{¤ޤ࿭y”ÇÝs~éY<_«ÿ0Éë/c– µG• Þí¬0W³¸ÍìxK×^wL\î»·Ý·‰­û]ž@kŠoLÛ¿£êú}7¼Æ)ýOÓ ˜eéJ7¹¾|¹S¥¾1ËøE€VGøÜzrárÐjÊt6÷!@ÉúqyÓ|[ü]—.­õ°ªâýB”f2òS ^?šFtÙ+XüèN⃷f4èp9åƒ×u¼¶*²Ao®¬X–ìD¥RÿÉËÞÜ9‚½šCMv#ÂéR¹Ó;Gf¯€2îY‚×èAÁž*_³BõÌ8©Pâ`Ý  ó€4¾ð¸õ_¶†‹öç61DÛúÃpï˜eV/½Œ­äÞ?™©`xÉi'À³­P¿a¤˜y4#œ²– –}£ËyDzf&&÷ 8Dó’¼ZA“4÷ï$ªBÅjjKªÛJâõëNü\¸õ¢x¤ÉqÑ‘º­`&ðÕ{E÷›kÁÐç1K©Zx$=¹œL#Ë¥¬ÛÎF.ÏeZÇ'z{T'~€¢jÉCƒžÑŠ<_ÕŸGtÜúþÏvÉDÀÓfó\S"ûÀ—8;þw‘5hnj™1œ“Åz—a¤Ž"[¸U$ÊÝ.¾myA0ǽ*[~–|~S—Ì*Ö3œ0Q3Ͷl¿lh¿B—¹‡G‚©á¯ë!AUp•pûÛѸŸIžÇCßXÎw¹¾CHûM*åUÈzŒ 1Q÷2¦m_µ¬ð¡¬Ü1ØÃe#~È‹„„HàŸˆ5öKtM ç!>Úµ=<|¤ÆSN+¡Xvˆ¹¦nB8c„Œ›“ä9Œ'Þÿôß À, fùR="f4ú‘f';¶¿¹Ò}`K‡tíî/ÿ|HÚ9PžÅMæ’/>iM{Úhªh¡ùROÖïR þ”°ÜÓ¯òAG ÆÏ¹8°©Ø‡ßœt[°cEFOÛ*ÁÞQSÚ£Ô·>÷èóa.Ç$[=æùÊ»“Bi,nR^#¼)ÄŒîâéǹҌd°_Ò5 ×±g¤n~ÀPå€|0/ WÒçÞíj Â3|Ã-èŠR×ô`‰C½„È2ù|E…†v±x) U(«ÜÉ^ü<ÕFK™AȰÁüÆX‡÷A8¯‹L«>žçSw­Á˜ˆL“||¡­EÆ©"Éœ ÁágU…¾‡Áë0ìÉ'jFW9QæÅnû\« ù‚ÕUSø•OY,}WÌJß5×¥ý÷Íïp§5ûö&$èœÁ3U ò“ƒrÌ<î`"7“ußñl`¦fnüN¼g*©æ>1pc„¸u“Éuß=œGBµqyõÑ–™Í4bœ,­ ú‡ |‘N1t¬]÷~òÆóW$¼ÁÙ üNFè¢ÕƒÝlTçwquÎç¢Ð;üfa*¼ÿ ³z_>ê3g'¢@£"Ø/=Øy'$\¥Ÿ\ÌžnÑÙŠ¹Åâäˆð^Û@üA-ª3—»:ƒ_½¦`²Åã‘Rä-Ó:ȳ ~Àîf–;D?³‘žEæžÂó;LÏTB¥äz$½‡œ1Ó)^§”Ó§éÁ¨{$I®æ1×kAßmîy´(ùƺ‰›"ýé½#ÎxA% Šú—GPVUhQá K“=ä`wcð¨gÚÑ•ë#iíîÉŒŽå)%X0þÝc)ÍUìíˆPqÜßx#á¬ØâÕÒso¦â,CA¥‘÷ wû:®t7ʆ±ÿDÔŠenIéi £øØ .‘ ×òWÍÖMð¬ÜYTeíYwU±1kš|~È'å bsD9V>”Š:zÏlEÙK+ÚÎNo¹JrÍ1Ò8sà¸e’3W!gÍ¢”tàñ¬èöˆ|,“”ïÐ:énè­Zg.N­Ä»Dã`8^)LGa 3B•;›x¬šg|i©@G†'ë ®-ÔžB†ó¤ó$jÁDż+ü8BŸí¼Ùtþæô»š™¶K:ò”?ø§ÜÝ5~£®í¾c+›ŠjcuíC2“&ª^Å ’­\U°1“iePV„ ðóMôeõ æ˜(¹ :Áà·¾Ù†™) -,má5Á6õ=ˆ±ºŽ’å~Á[wûRšŒÕ…ÒèhŸgEî´3$Šà_në7ÉöY晫|–=GÃÖ›•ÜNVï¾:Æ4;¤xððQîä_ËÆó¹DAR)f‚Žj.%·¨L¦/kŠQ^¤%”‚ø×Ã2ÐVB\Ý 1(ã‰mÏÁdyºfüªñ¢'`ÈKǬ^Õ^ìU¹¤pÉçÀ†Çû¾qwNgî›fXç½ïCßì8V·ŸœPäj•ËŠ®¬-ÃêiÅíMº¨ï¸&¾ÑxÙÌØ2´F¹àá~ÍœvN |Ö‡°ô4Š)/Iã±æ]ÑÙ#°Ì‹¹'t®eÁ¿Øc¥ädEÉ`t „¡¾ŸÄù<ò8 _È5ÀO Í3S&”ƒ>—+5à‘¶üò3Ô݇1Å|ؼ§ÃÐf®Õ<Äø–Ï“M›¦Ñfî(2’ µ€£¹×þìn!™4¼Óèw‹‰ßJdÎ÷  /²ž}R&²(ÍY]¯Í·¢°GyÎPiEîàÆBûž—"VÌu).Å ÖŒ§Z’(aÞ+Fnâ?³î‡j¹*ì„$òkw“Ìï¼#È4h ²o€"z1¶Å7 Y[X='{Gµ- ,Ô[|I/\)«ÔE#öHs~¡'«Ë|³bwpÕ§Mõ¾½ZÛŠþˆÃp’“'Žv.ìèl`QEÞ¤Šv…)ÙɈßì…Ø¡§ØÂHŸ!¹deCó…sza—yUqç0ƒÙDæÃÜ»¥Œ’›.…M_â0¼Òdówv¡ö‹~QÕϲÓ9ñÍ»Ò/ï dEñkƒ&q¶Ö"JyÐâ±f/öŽ™íDñí¬äØ4uÌØޱiŽZ÷-'a=ƵUh7¿<#ƒú•²ê=é‘ú¸~¡8OÂP!.Sc-ˆ—›FAÑÚm©wNéŒßJJÒ/þRõRÃM9 {æ… w oâ²G:¸±Ïû+k$¡æò×$T¿ùXì'äÎäÃï$AÔIìúƒ™ÒWÞCEl©fU¹ÒÕ êäÄÈÑ>*æÞNü\ ÷•ú¤KÓÕ& ™/CD²TAWˆ®ò|§ÖÒƒ^oñ¸È)©ÂÁdÐtî‡ ê1lºPzô‘ äï¦tö[ÝÏ~Lº9%ÆUÖûû ?áIPyºH†tY´b.’eýmÔ,â©ãĪæ¥S¹>GÕW>ðgVêêÔb&õÖ8jžDë“¡%*@D(ßÅ?ÐÉ•°EÂdc€ k‹ô-uF.ãµÆŠ½çD¬NõÍ\P«ÍÊ‹ùÛ‚‰ÓøZ]º#“@W“ã‹Ä%0ü…²Ü¹=A£·(G? ŒšÐ·¬hÇ„À>“F&ï ºÈ½¥g¡¬iÝc0šW_$õ>|‚!\¨3ÌI¿¼À‰®ÆeõªxB"–¿k!Eæ<¡WC‡<•Þ{ì%‡å— s*»Uè zé ¹ÆØøßí¬rZy¼oÍøj7—¢ëi¥e8 uý*vcôYw†ÂÃú·g™ðÀ‚Mü·‰Óæ»=š¤ïÜNÛ²çTÑÀ(7ÿ~÷ÚònßÁY±íiàÕ a4·­MÙ‚n¨£ƒJמï,>a,8=tdWÅ&s÷}€f䌭k‰2ÆOÙF2$»Cƒ‚û$}ä9®Mq*¢HÒG9AVÎ +í2Í"¤Äc•·¿Ã‚ƒ3’ÕŽÜÍçJîæç…´Ÿöíòv#ʧœæëÞùôú¯\i{_"}Ä'eùÜ´8üþ0ŠsQ+å¨TRÐo”1¸Û“ƒDÖjWÌJ9CŒ@ ñÔƒ ~ëSl¢ˆxtÑüdUìÝ¡êŒ$âß> r&ì˜}@z@là‚¶_Áª#ªXçA&=™W§#g†ÿÌ’)±#nÏ'¬Ï4¿–vþ³µÆž¿€À[éhBÁ(PQZø6#aamǬY<ï2"ª ˤ‡‘³°B¦ÈÑ–§Ww´™ýkêè·ÒuÔ‹ÃÙf·VÀ½¬±-»0¸ülú¦l çßxòsÕÎ8_Þß««‡#àú̬øSHñhà¡\Ųt³T‡E¯\I@BOÄ™\Ö V¢V2:\¹­4tÖ=*óˆMUÉc| s.Ä…(ê÷·¾ {>{ë&5lâ#úœ) Úè•tXÍ®%zK«~F ß¹ćs=$nð›/˜­>ü)Œâ°í×€]dƒˆñsçÍV[h÷ö,å™ów+5ã]eETwik¨Zsâæn{‰P#航©«¯¸¯íZfÑ­‘Xx´Q©ÖW(kµvÆ®x/;Òš, æÆÑÇÊŠ×a¯Äâ€s`½¨çÞÄšäv€Ì%ž<Øa©ÕÛ5¾ê|&IåoÙšÈ E¡æ†Z¹‡³+ Y™$7 Wìxrñ¿uÒ Þ`Ad<®÷i½’®­ƒèDÃÁwï~x?CSAåôƒ9ý²_o¬=I:Á²õLâÉD ¶Ú² ã8¡@ÌF˜¡ Ò^mÀ^?寃ù’Ã$°×=|ËýÉ’ ª!qKK©:QŸ¨Òl§¶C¬O‘‘Ä%u@êÖ˜ó˜ !s€6˜kö`Fôõ¯ÿ>‡Ú/O¦$¾N’“ù·ýÈÇ Çó!çüS3þç*ØÎ¦Ò”E½‚‰®[p–Þ¼,¤Ië}ND¥+Ƹô—û¾¨MWôLˆ°žº×êÒº m œêjQsÂ7“-ß¼Ç-S|ôüøÔXLücß<{²–ËÔÿ UŸ’ß e…5¤âºÑæTÙ?¦£ÐÈËÄ;¤R—Ìk_Jê)pu;¨YÃâ.¸,ô|$ÛdÒ*&çxÀ³ƒÂË›* /@ÐDy@«¸ÆµŽ¼=5bæú ý­‹´x¼O¶®‡àòíýÁO%•+ã½Â3?éÕ–cÉsx¢g=‰x‰Ù3ÙÞIYmÈDM/CÄn”X8o©„»¯Ð2‚ki#ħÐuIWN„²-aåõz® Ùãý~&,´¦ J *d¾»4z‘dy¯ç~‡4­r×™eXüÑ'6åó–ëTÑ›”©¸ÓõëoaʶÕú¡?Ì5CIh*ùZä+õÑÁ–d–»%hÛ‘“ã–ø7x‡Ž=iÿ*mwMèyñ™o[©4<~ Ný!Ç[ÅÈw•$†Zg¼u1Å‚„"q ü«SSÝ&y´.2¯”šÓIwYÂÍWG²çÐʱ!ø˜}±Æq0 Á¥õtÖ–€o¿º{qJÕ0Ë…ë^ OÚú”½Ë…ªãMÑã]3›AØŠ¼cûõŒ^6+¯×~u­£háa%ˆ»Àÿ±|âóá¢EDM ;&BîÚ‚üWž¥Mëü>D%LvWÔ/M·æseúr'‡¾ â©O«¾ƒ Îýª:1.ó¹Ç$_|HZ׸·7¬¹Kª¯d€Y¬ÅBlñ?ÇÒžsÁ1¡ˆÙlRÂInù´i–~”¹¤E’‡Šû4Þ‚$p˜¢Œá§ò±¿w{)xzÏJ˜^i1C7D:è-žP ‹€º‚ V$¦Ù5uãº/_ô·–4 Ôi®J‘‘÷7éH§{QñRÙÄ]rŸz”ƒ†&WËþ÷>Úr‚©y´-MžBH8äF)סøp{­}ÊT°nÍDéB¡;´¤V8ûèvÆ¢­ûv=MT¡<~—Û€ÑæfÒyUÕ]Í”&Ä~d¾ÑL$qO6ï³Z¿1‰‚™±;<C¸\xHèÕÈm‘ Ms%Ä~îú*艠xvÓLç{ˆ‘~´Ï]Æ®r¥!±—{ÃÄIo:N6´ÅÞDXq—ò†á¹†Zbª[¥(p´ËÝá¶Übs¸5˜‚a—|S´Õs6)+C«‘l ðv{.ŸÝ˜#3¦ž…q$ÿÑþûVºÀú<#d¥)ºü™%c»ÄXTÎúòt*ExŠCk‡+Gª_3ûÁ“WZ²¹¦ bìÓ…í]©u1þ5S›o{Ãù2Ø:ÖÊ-ÚÎzª7¿ ¥¢,=ù†òᄃ°ðb¡4¾ÊáÁ„³<´º\S[íq¡Æ} ŸÈ3:™¯‰Ðñp¬øº¼÷õʇi¼¹èb€ƒEV(ô+>ù-Í23”·:ìÐxJM f3Ç[v¡cK÷i~º‡F¨õ_R24‰²‰ÉçoÙêF̹º]EÞjw\hˆ¤øôQš¦ì]¸äÝgvG.¯26”òJ–ü,¯-*ÒÀÇÞj-Xi=ÊÔ/g0+äc"²¿‘e³#ÆË‹sm]ôV KžÊ?òÇoõwW4‰(Kó°ÉÍ‹SÕ¼ë÷WäÙNzÜH• å.µ_½Á½–º6×ÔÖXÄì ¿ÎÕÝ•Ð!…0#Ê™Á¸öµjxútÊ6¿ñ¼±7$ƒå,uDO½ÚOS½†S# ‹P*ç‰&ÎÝš´×9Z\Ì bJ%üúlï?ËøòõÀ½Ñææâçàp&÷Ì÷c >–˜b·^.š¯^õ§úäM;<ã—§+Ídº:ç¥'‡Ò…6Wµ †±˜o ޙ̈́¥W È_”M'N#¾_O»}BÄT=—­¨<Ж&FŽ &à‰È™é_ªì6 Ã'3ÈaÈÜxp™Ù˜ Hß­’Õö…’áûýœMu¯$(ì$cJRÍXQï©'¢`U^Ô®ÄÎáb̳¹±Âùjöí÷rÌx¶=™šöü¥Ó0M«¿qéh‡"@£§J„¯8lŽÝˆ ÷ùÇÔYèJóÑlõ¬¨”i&çb°ÒÛBõ(¿›””ë€Rת­ÉwÍ(8´œQ1É©ÇÃòwù[ä £(7Æ4+*¿ðóûe®@kþ• Áz0YÂLEóŸ¿óù†¦Ïv Â{eº­—ë¦\†[#óÄmL±ô‚[‰ƒõÂ̓œ§ Òy¡gþ›oÚI­Þ7Móϼìv¯…a;“Î+å‹|Ú3ºQ×êqŽÆ^`ÙŸWoߺ´ÒQìø焹«B‘˱±Â÷§8&[ª×ª–~ËÔoç€\­„æ–”³»b´³ªî¹*gWþÎA“¿Ïy­Á~6ø±x»#b ñs…ˆ€݇åGÀR¥Æ¢µÿ¼>œÑ§n꼃·Ñ¤•ºQËŒðÙR'kF%4œËMoñµªuÏZ~õ°Ã¤%uö”}=”³Ôˆ/@?´†ê5¤rNª„í»MΞHrÏúï?â®dh¢è³ UMïÄñ«“Ì‹gI <œH9p’žŸªoäkæ§¼—vÆVœâRŸàú2!¶µºÐÓ[ØÒú¶8yƒ}ßY$cw3ý7¶ &©0l¹íË“<Œ7à³¹"Å îÚ‚“×ÅI+"‡L­‘W׆íÕdêyºjšÙ4ÙWÆ%.š~SÓ܉ù}œÂOÔ:{Ç Š¬¢øáÌ¢Ö3¸ÚBË(ÇpœQŸûzʵ²Þ>Ðæ ü@Ž²Þ§ȘwƒV¼êW¬Íä2ÞàoJùÜ@ Lk Ú¡-GŽeÛã¿޵H9âûÚbÎØ¹>È 3¹4>Ì×(ßi·ÐÛñPà:ûÐM¥tw+™XNEÞX^#. “¶XÇŸåaÂw|8¸Y‘#„»güµB!þÕŠü,M’Äþh˜ ¥8Óˆ£Uêq%y7̽S¦¥AÎ%*¨óW˜ÚÔ$“B¡#tfk ó+MP;E8DlýP´/]Ž£ö$€þði÷ ͇( ¦lÙiÓ‚H™Kí-êå¡ö'dÅ$$<2¤± AÚƒ'{Ãn7¾‡|¢!òúÏñôKšõë¶ ɬ \ýâ˜×Ô_Ý=2Vðʯ\Q•oŠ•®-¿o°NèphkÏëhÂ(æþÜ g\©ÈUÐ øþ«‘qö šB6P%9;¢·•x\êvÞO¢£Z JŠÙ}ZZ0]ñ3š9w ?X&4¸4ñ‚Ì;ùrž¥Doé7)u¸TpãÿpM‘l¼Áèš—P7Q%<[úÈöë³îÓøCs >&;’ß©±ˆ¨QNÌB°Ðç3 £ã3¥m‘/É<¿y}9§O2hèmrz‚¥qÖímä•hÎ G(t)]28Þ=[*4ojBèÊ $Ìd#8*]ZêÊõD¡äò/Ýr€ÓŸ0èù¾>ß’ò)Àðt[e¡cž/Åç‘'>ïÈp_R÷ÑÀÕ1­*DŠª^ ‰tÌ“.9Šð;òLôV.unXSíQ]ˆú•P ošÜÜñ¶„|2p°ÛD pã›–fjò­{MÂÑá2 g«úéíjt•Ûè'kïoHPLZt ‚¯Ü¤"?žV¬=Pš)$é‰ÐÑ{ù®é¯j¢f´Ǿ¤F¨á¸›(%gàÉœ<éspÛVÕë=b§Žçœ€UÞßúŸaÂÅЋn³÷[+Ó·g¡ÒÙ®y}îö™Ð7 ƒª‰ŠªgÊ»­ ˆÅÐÜö©Bp¸ó1 PŽØ±_'pPûeÄÄ—´Â‘È¥Ï æZ‡vàŠé(ãÞæªšÃâxÃ÷Þœ{b»z]ì`W:›BŽU ¥9»èȵ/†î³ èûŽå–Öèå ëÇï@:°) Š•¹z7½fnO¬Tz™, —¼–í%¿ð›ä¸óY«6†‘ŽÛ@¸íò[ñhú)^3>¥­lSÏ·‡ua×ï’iÍc|ðx-Š2àÿ“Ûà endstream endobj 207 0 obj << /Type /FontDescriptor /FontName /STMVCT+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/A/B/C/D/E/J/M/N/P/S/a/b/c/colon/d/e/f/g/h/i/k/l/m/n/o/p/period/q/r/s/t/u/v/w/x/y) /FontFile 206 0 R >> endobj 208 0 obj << /Length1 2687 /Length2 19049 /Length3 0 /Length 20590 /Filter /FlateDecode >> stream xÚŒ÷PØÖ cÁ݆àîNpw œÆÝÝÝ5¸;÷ÜÝ ®ÁÝy=sçNr¿ÿ¯z¯¨‚^Û×>{ŸÓP*©2›Ø%ìlX™y¢òjj,Ìff6FffVx 5 gkàåðŽNv¶¼Xˆ: A21Cg¡¼-@ÆÅÀÂ`áäeáâef°23óü×ÐΑ fèjagÈØÙà)Díì=-ÌÌAyþû@mL`ááá¢ÿÛ lt´06´È:›m@ ­ªvÆ@gÿ AÍoîìlÏËÄäææÆhhãÄhçhö†àfálP:]&€¿( m€ÿPc„§¨™[8ýG¡jgêìfè€ÖÆ@['‹‹­ ÐÊP•–(Úmÿc,÷zÀ?Ͱ0²üîï¿YØþílhllgcohëaak0µ°%äÝ醶&Z;Ùü ] -¬ @—nV‚þÃÏÉØÑÂÞÙ‰ÑÉÂú/ŽL…µYÜÖDÔÎÆhëìÿW}bŽ@cPß=˜þ9\+[;7[¯ÿ"S [Ó¿h˜¸Ø3©ÛZ8¸¥Åþ±‰àËÌ€Îfffn6VÐt76gú+š‡=ðo%Ë_b/{;{€)ˆÐÇÂúïådè 8;º}¼þTü/‚ga˜X;Œ€f¶ð¿£ƒÄ@Óÿ`Ðù;Z¸t˜AãÇ`þëçßOº  3±³µöømþ÷3‰h‰‰ËHÓýCù_¥ˆˆ;À‹ÀÀÊÆà`ãpòp|þ7Ê¿üÿËýo©’¡Å?µýOÚÖÔÀó  Þý—†ë?sAýÏÒÐþ7ƒ‚hšêßÃÿ‰™ƒÙô‹åÿó üíòÿoòÿŠòÿ:üÿ·" kë¿õÔÿ1øÿÑÚXX{ücšfgÐfÈÛöÃöÿšjÿ³Îò@ ›ÿ«•v6mˆ°­™õ¿´p’°pš(Y8›ÿ=1ÿ=Ptk [ ’“Å_÷ €…™ùÿè@+glºSœ@‡õ· Ú¨ÿÍ(nklgò×ê±rp  =à™AóÅÊÁðbí¨ ÐýïÑ01ÚÚ9ƒ\ v>S;Gø¿Ž”“À$ü—è?ˆ À$úq˜Ä~#“ø¿ˆ‹À$ñ±˜¤~#6“ôoÄ `’û@ä#P…ß”Añ_Ä Ê ô2¨üF  ª¿;€Ií71Rÿ@Ù?þF ì¿(»æoÊ®õ/âYþF  †NÆÆŽÆ.6ÿÊYX9ÿQ8[X›ÿ•³³þ% ‘…“Õï  tF¸‚b:þ!‘2r44ZMÿsü#þÏ*ýå?b+ óÿØó°ý+ÿ? ~Æÿ"P1ÆvÖ ±û·fö¿$66¿Yÿ5L&ÿB6W;kë?k]TL¿iƒ& ø?I9ÿÒ;¸€öú· ¨7¦¿]@åšZ¸þã/µËŸ9@&f¿#‚ôf½¼À?M@µ›ÿf꘹‡½9Ðö ÌâªÔò:m«? ¨9¿+æuÁú¯ü­µò!]ÁL¿Sq€bÙ‚v÷wß@©m]lŒþº5Íþ( ô80Ùý.Óî/QûßjP{Ð nû?‡ÍÎòôtñ3ÙA/õ¦œË,ì~(;¨±öÖ.}ËarøÔU;g ‰ÑçÇ’þÑz»ß8þB@×?:Ï2w=pÿ&ÕædmèdþGPi¿ =LÎæŽÀ?FÄÝÙÍîP —? è¸\ÿ€ Nºý±? o÷? (¼ÇÔeÏßÅ"yÿ“êîYcGP›ÿ~ A—ðñßßw€@w 1üò¢1_°e}ð·ûZa7†ÝI9Š]4¯eÇ—Gd˜šš¬ÀuÇ[ᔑ^ÔÕmqê¡’¯_m0a_“”ÛŸ¼ŸõTfvÛá—¦±§Š~ 7 Á2¨ íy¿8x °‚lï’¡ÈspáFV*À¸wë—to¨ø1º¸«¼WÃ)‹ð\1Ë£ý) tž"ß({— Ú™–ýÜeþæv=wêD&Þç8†­ØK{ƒ5öaÁsí‹«S79ž6.ä úø ¥—ÈÁgœï^e%Ñëßs˜iÁÖÇH¼ÇºHOxuä¤Ñq˜­—y$ÇÉ}Ú˜4ûá²vU ìiëí!Ç¥3us“,Pl-é×Ûƒ» w¤3‘wƒQððÚÛ\†ºÑúu#ûë¤~JÇsTÇt ~ kµEdMì%ÌÁSšðz»Û»È2³œ¼Î«ÏrêØ~Î}Ï]µc°RüÔ¦(%£(£™‹b$ᦃý\ O»ܧ¼ õŒ{¯ì'ùãÚiܲOžnn„MÖÉS%ÒFè-$–Æ 'ÚX_ Ó#铸r‹G*dëxïß Éü0©ši—8ŸÛ¿ëѯÈ2A{oFWÿøK åàïH¨yíì0C¿Ö–T ÂH…ÌH)-¦•m~íúÁ¢ƒ¢ü°$plñ‰b±Iʼi„^öƒ…»[¤èèÏ(¸æöV¦-á :UYØ"Ï#£—5½‚f*é²å ºñüªsJ­Ýo(q_Šn*¶tÆ û’–ukßû¦kÏõh/yŒº×D×‡Õ hPŠé$Ì‚eR¢7í~]à#ã!lQòyðéLXïê›1¼Y´dÔ™ŒÉ[“"iZ³Ý‹Ñ;ºt‹™“o™£u·i]ÀÞÓªNÅÿp¼ŠÄ踋ñ%ÞÞÔ‹Y¬Û›S¨r?÷0›O÷ˆ©¦ÓÜ{jšñ¾TEêçH­ì<¤M–õ­Þ*¥Æª?NÔÚqKÁù‹D?þÚY%QvcjÁÊBú[¼„¬¯4vh1—¶—ðªü“¢r$#_ìZ|´Ô(E¨f]ŸO}â}æžå€–æøw—J’o…Cíö?âE¤¿•læGª°ŸËó¸ì•;bØ™Á¥ÒçØÏ]tޑܴ:ú>gäìf³~Óqžžý¼Rݘ%h~Ùj”µùJ»LAîo2XlIP A ÉŒà @% Ì»(§‰-÷VŸ9,ÿmŠýP‰!ÖTä|:o’Ú[š'ð%3‹câê›AsÑ©%WÅ/ ¡·Tì-Œa4<ð™ð15&°i‡p|õÕi‰Ãº»ø6–dgñ©ÀÙ–µ!¤™¼¾Âz =ÁÏ Ù)Ù—½%ØÁ0zY¤OÑ´,fnwû¬ÝùìL%y¹­l]B•[Ÿj¹<zͰÒD´0n¤Ÿ_D‰»jQ}®¸;â—0ö°µìdÝêX€^4Æ<ˆwcAŒy¾?`ó7Ê•Ö ËŠìñh­çßA2]Ì/_=ŠJÚ=º(y*35;¦è²Ïâ·¸ Q$¦¦ò*[¨Cv«|UÝSfRq‹½]ŒÍl5°NMvïùßITù‹,”ÙÍÅ|M8ÁЃ êI=Õ.É…FøµüF•_Š ®è5#‘Ñg¸™©1ðt?ÅDZ–U¤d# Íèƒ5R§TÍQAÛ¤?!ÈR¡ÈnÍ!Q}ÈEËßPR ç•?ðŠTe“dUAâü”»!Èp¾ ‡­CåÿãàÆO›ÑÁ<­ïÝAîå“u aE™Ûø²BµºýD©AUÕõV†¦ŒÙ¬‘éV*âø7§þ€QµÙd°*±nH£À^ R]?‹Á|·ö$ƒ·ƒ¢r(˜hªÝ^ƒ s=è^D5µ8éJ[© õNsZÇDâeì]_žúÚíâv@šÿí÷5U±}ø:2µT¥KÛÞmeDžz×Ì—2¦cÅÇwjð”Ç„^ÁÔ¸²¼Œ¢Ç²<]îR ®[âÏóìîúÄá»qyË_Ÿ#3úé¯sJM¶¬9¨ŒÆŒÖè?KÝ2“j¸6álï´û%/ßÁ8¸L· Õ-Ê'Ct^÷˜=Λ0AB0ÂØøÒöw|FßüÚ‰2ÚÄ´¿k‚g9­÷±Cñ—;¦þ|æ%Œ?í$¿Md @CÌióÁCõF¿ÌBC/Ó›])¥~ZaérÍ•×|^¶Œ'ýudÅ!p4H™©•lA¼ìv:æS†›ê*Z¥ôÜ–€ªûsh ´%£HÚ]ùê9˜?ÔW(ÎHÃ](œål@—:iw´|” []…ää‰i°GêýêWeH[=a²¯ŽCÃè,Mhîðý5œ˜–2ÏáÔÿk͇ý`¬!u¸¶A’Ḡ.”~Eçf6ï˜sGfÈ«6ûÁ¥×§%n˜õ4B1¹Žo±^«'õ˜IÃ&nÉÞ߃‘$eÞ<™Z²¤¨:µÉƒ#–ÁÒ—ä'‹¸cŠ6bãµQÇÃcÁ‹`ù¹Ö½jäì¢õÝ}¤¡À@ß<°]e >NW³+zµõaÿТýæ8\ëÕîˆÓ£®ýÆ#^‘»8/Y½é’ÀγòaÃÀÏ¡s‚)°hÓ÷Ù}ësœjˆw–)É÷/‚ÝšŒ.Ã"FâÏŒn|O!évôÝp8Å"rVV²M^0<¦°ffÏôàÖ‡Þ•”ßügq‘å‡ i çÉo#ˆº\߯68æR?6>ô›èk†©£jÀ ³mU@™NŸd‰ÎØ/Ñø9Xù S¶üöq±™lX03;)àvÛH$wjrøù;·÷êÉ©œŸkgŸø¾J!ÂÛnB³†¯×¸“hNBcæ± òn`m¢µ}ÞEÅv~OÜ·-HÑ6s—b©ÚÔñÌÈN€…øF×ÍiRÅŽ!-C`ƒûƒ9]o<]Šâ…Œ‚à¢à,Ü×  S:vï"!KA£ÇáÅ­´é@ƒ @bQ=R5yw †»«E~b@SÔ[u3Óû÷]$öiW š.½ ºî´„tµš•ûy¦¥™QŽT’ ,#5/|ÏÖYe!cô4D*E‘½CWÅõl¿/ô=ÝéW¼ô'Œu3·8íª[B›Únò‡~q–JR[}™úHQä›vØ Ðå~ Çôhm ·“sDò÷0ínþBN¸𡽠¢x BF¬xt6q*ÀÎWð4nnÏ(} äôð[êmÍŽX£ˆæM* ÑÁ#·uœ<©º†yw§ 5„+Ò?°ãÒÁ¿@”p¸¬ È«]>V%I…cs—im¶`Û¶bLOy=×1 |»„Íq~›õQz>NvîÇ*(•qoHáþ><§Bøc“ç“רËÚ›V8ISé5ÅaxÜsWï§GÛÀP’¦£×¨©Ëô—<‰*«/õ ï.;-»R;ð«×Wöèa4€ø ÓNåò S£h*îÓ>_õlÀTÈ\h}¼Ëän<îféy6y/#Ç‚-ˆ!ß8,ˆA6iT:®ÜƒLxÖÙfjÜ2=—ëØØQƒÀofís–šÀ¡U#;% º{¡ûË^(¼È_†cAÊi#ðéfkðˆérëkÔ¸ÕGÂ5—=³¿ä—ü¨ÅST*vë”éä]¡‡š.}Eéõ6.¼Ì)&¯x=(OÇ \w$S‚üŽq’Gt.™õ™Å‹ûàïsÝwªÙåÛT¹øž^xm¿ŒšŠçŒ­”Œ`}À–à™óëzûîK&< ¶¥çªHƒ‘Ô‚Z£T*Èç苉ËŽ™òDÀŠŽ·(ÅÓ ¹{Ò‡žåáÜÔﲎÙS—Vúé³6,J´ ÖÛhKºZt1×ò E´-Ù¾_ º=W¿òÿôPß MÕ* %ÈÓý:¶nã’:Éùä£xˆ'ΨfÔ¤¶Ð曦Ïù± ñ/º‘€]å,Ž»¨~ºéè|©Ü.T&5V£‡² «Í[C›ú!øhZJnDh/l’ËD±¯po4ØA#*Ô0á¯fµ '‡× ‡Æ•˜–Lžš8Íãðû–%‚ô§wiƒ1æ k( Ÿ’"väˆæ±zèkÌm©¸±Fž‡+ªªIqðÅçC©øˆ^CX°aÛ¡Y£~TxW†Úƒë}²¹/·\dù“Fþj³BŸÉ” ’Ýì'-3±’Ì5Yîø¨nʦß5ê•8 ðOQŸ]ù=öpHv#¹Qõã 'õHôáLRé¬+Õȼˆ-8kõ]?*þÀ.…7…þ¼ŸXa´nÆ”îÞž_«0s28§Î*xi&zŠ©P‰ùÞÞGKFV'ða œ¾sîN8íLEiœFÉÊ‚ÿNC |7©(iã—A¸ó¤bx¢Ò`†õxÜ­ší'f5„ bŠa‘•Ý<…<ÄzÁwÌÉ©öxo'¯rçÉšuH+?Ä’  w? ;~cÿž´ „÷`ÒY‘Ú“¼ïZÃÙqô5‹ˆÕ/fªB×b, dñæTùC3›è5c~ÚÆ&Ê?ãŸxUÜQ¿n–”w•jRom]X àýÍpLWéùýQcéî~mwû J…õ>òM=ÁCú6`ø®ì2X•û!rïXtÚÓEÊî+¢¥f†š½]热–¯f˜6ž[9aŠ•9o–[Ò>ÿz®Y¢OçœÌ’ÏJ#¼ˆìµ^v¶mô½6/¡²ÌS‰Vy¬™DÈ>\;ù°Fãy) Ó¸áË9ÃhnÄbY°¹D« ³A¨e:p$yûþzLä?d}êæÂ3ö ÈMÇ/–ÓœWäÜbB¥‚0à%™š¿è€hìjL(#þ ŽuÁm- N~#IŸò6.oàá#yT’…3*Hcœ¸úîçÕçŒ7®MYRn‘ÑŽXkðÆUD»â)즗pWxÙ–`UÁ fmþ ¢o—òz9v⛋gÉHüšá&†@Y$=ß¡:8RäÎ ~É6×3fž¥dÄ,‡%'Ÿª€“_¤›UÐFD¤ÕZ¢æ]/Ñ—Mù$p*íV®ÖäÞUwÇÌöçðÊî`3ŽvÛRÊá 鮞û™AçÀðå ö8ƒ²¡q`¹Ñ]ÚWÁÊT5†îçXĈ½´›ØpCùàí¾Qò»¢¢×ÕÉ7̳dŽN™“ª¢Õ{½³£ˆO:óOózü¤ %,äµrÉ?åÑФ>¼b/$KŸ‚±„™¼ÏB—á9¥L—Žƒ y_˜0‹]o6¦mA§+«ÂZ•а5 •÷™RÑrI7R‚l ¬rÍØ7Ô7“·âжë·[ÃÜ!,Ëmš5Í …ôÅY?‘ìý0ÿ¼Ôôk"œ>5ïKÚ¶™Õˆ.„87F‹ dZî’a¸“=t)ÊþäÖ!+«dÙ£ó5“N¿Ð¶7ŸhÉ:˜®¡#çö‚æ R°F+=Ä7ѨH00EìÛåã-ƒ²}}‘š bóÒy‡‹C nDR¾ÆË©!‚.Úþ.9ù»/Cgû™F]É%ò²x©†^Fv²™Ò{¶3êÁv­n fD—l‰= 1Þ *D‚hç•Tå({ઋvÒ‹?õÃüеXT3ŽŠÑÑ_×nÒ?¶ %¬.|Mæ(ª)£RàvC‘§€¡çu†% †l«8+÷ÛL2,qÇæ›ÒBŒ5h°A¼«›Æ\ž•“9mX8| ïÇ×¶%ê+ËdÛëNS×™ .Ù=•`}«<´¼å6½®~y|W“K¹%ÃklYÁVhÊ ¯ØÄË&*ϧˆ:êJÆï‡=Ü ,È#Ð5f˜8‡;ì…ùø(?ÖÙ|Ké#·gë’]â‡[Ìþ#÷£Îøc~ˆÑQ¹óme<Ëc³ÏÈâÐ\ñÌs×N%?¦ëñÝjë:ö¸çâuÿ#ÂØÅçáéÂ50žqG蟸MdYFÎÍé£iä‡Ó‹ùºŸq{gç?®m3Cæ Ýø‘(‰$¶hÐi‹Ž fF*æ©øÃýbn8ô_0ŒÃù8¹ØýzW5¸Žc2åŽÁ‡+"v~{™¡'“§UZ´:΋e¸-T82m]{ a¦ÏQ1Zß6 s›ó[\Å6ǺšØ‹Â]= 1•›\/ø.=z½%YÊ!˜†kƒ‡ü¤‘©kf[Ñ!%ýB¨LuhêLó7I@Ó“~,ʤr¥n÷ÖFJitÜŒô·_EqÔ(c©yž¼:>€ ò£Ç‹P3ºùûÒÖ®ÕJ×ËÏ&Ñs:ÄÎh_=gÜvøfÌÁŒ‹y>×j"ªr£{e' /dk}‘a·Ârj;§"w¤ËtÎx Û㢠¦Y·¡Œ’ÓWžGÚ–ãJ¢©j0»õo ý~•0‘ÐLÆÞýíÍŽ*žùB»R‹#?zÈËEXsc‰o“™‰XAÚDUÕ…ÉÖæŽÙOÀ¯ã8Ý©„:ÈÃûe¿¦—îôu„7W/UÞ«G½¨{¾ºS Â/3¾g±ÃOíÝVo33ãlásÌ^ÀŸN‚—hM»¾bÊx-ym9&I¿Õ*’‚þQÒu»BM¡-&aH¶ì~‰£—Ñ4.oÇ€F¬‘|TS Õ ¢6E¼¢$8/F˜î¡S?&m[¦—îÎø®Mµ0÷ñçJg‡¼=táÞuè¬BîZa#BªŸ‰t¡T*#é ¿‹ÉV0ãà¬Ó:+ó²xÀ±ZÿF?¬ŽéâtQ `Š\‰ ¡Eÿ%wŠP–[.Ö@~ÿÕvʽÝX¢¯DƒJèâ0”¥'ª<;¬“N+€¯‡úbj´1-A^ï\œV[11ë–?]øÉiƒ_­QÂOšd¸ÏÕl54ÝÏjн¹×ž̭Âvä"ÛÎZÈ[×Çt}‡E‚‰]þ‚Ço)`D$¦YFá\]c…=+±æÝÜá¬Vm%»7B)ŒrTNYaÄè¦æaÄá›Ãæ$œü\÷@íj3ßR~ Ãð=Ò/GÚÔHÅGˆõŸTkD+Ĺ¹Á :©¥m¿a3´;P+ôM† Ï]W|jkr½Îó^Šóœ¡¨x±5p„Ì×ð&º.nê¨[ðVäK4"#»äSdßßËk ‘ŽC§ŠõvÈ­Žß\Tç·‹*©’Zy°ùÔ«FiO‘1©¿ “3?µ¤‹Zh£|.ú*Ô¥ Á™5 {(UöAœ¹ˆ¬H YEíŠòW\š/YñuGtÞ4döP“¤Ÿ%ABp´LëY¥ÂZ®é…‹“n½Œ$/Èd«œƒ°õ9¾?ts$íî9컜ŸýÖø eËJQw±sÿ kókóþé±u^É©+=ÏÄ㻋Òã«O'Ðï l$®N/,ZB*y,v`Í9oD‚H|Mû˜oå<Û$ÏUf?v!³Ê…¦´­€ —KÙ¸rYûzô²],¯S8Jø6³›…GëR0xR‰UÍ膢¸ /; ?õÈYVëéð’ꨵ5‰VŒLžsWÜQÿyq¬¤Z‹E;é8îCê<ùäqXy‚«DöZƒ|âüFX&&i cÀ&ai +±6áMÄ/Ž|j£ó— ¡fˆ/z}KU¤Ž% (ŒGø2ÌŸï(5_TšÇµðø‹ÖtˆŽsá.];™ 3…°¬ ÜŒq¤VÖ¶õÜ”‚zQÈן—/Çg#…¶®ê»F?ùFí™I•Ò—…HšS(@E¸LŒù¿ïÏá ¥Úò[ÚÐÖ2 (P$èÝ`0µŒN‰ìþöéY±×Då1ið+6Ír'=_%gÜEaÉû¬``¥ ¸Mð\ÏÝ$ ‹ÎqZêÉÒ÷–¡¢'§®ëŸ;#ÀdÀ‚²›` ‘Ó7ƒBO>lgHÕ¯ r1Qç,° K2•t·ós“8¯GÞîT6pO™òìYßÍÞ)_]Hn–v‘Ì™ÊÄ~‹§^>HIåQ¤êY¨£l× ´_¾ÇºŒÒó!éëSdó UƒŸïóÁWü¢›V­Èä–¶#k±o£‡÷¥ë{åÁuS …Ò¶‡ªóÉ’kDÍìiCbÀ6Tv+LP’k&™?‡s Æë\ô¬ýôpOŠ£iøÕÁà~Í|ÁªÜ¡ôÖ¸æË–l®ík8Ç—nùç$àx3)¢tÿh=•2Öt¤Í‡ÎJÝ£«yöPUÕ0Aqªµ-œë™yéø_‰Þ®Û‘,Ú-+5Úõ(õ"‹sy›Í‚ß1&â«ä‚³3m„YC‡±æ¤6ÙfÆ2QŽÖ¬t~–‹>§Êfüx•îóS®Z÷éªã'‰üÁ4#Þ¯óa!g}ȵrCHŽgÇäÜêYch¡ö}·¤j` Ý™k©ežSиÆä\…ý¨§¼ Sg’i9_|¨ª!’f»\¶:ñ|Ì@øZéè|S+»ÁM <ªeÄn@nš!Ù­ÉåCDÉw†µÏÄòœÁmÎ;¹û:ã:áb¾4ÈŸ2Ê,,ØÍão™¾01’М•±¢]c»£’6#íaó6 »qû– «/g:ˆ(Y‡HBq60¡?@rÆTÅ1«nH)¶VNb›~‡ ܱ2«)þ¨¢M¿R^"w?“¨>˜Š¿²ÖJÏ Ð§¸›Ÿå/“˹Qâú¡ãàÖ›M~;FŒ æ]È3¯Êè¹§}Œòâ:ã'…5ŠI"ëS«ô”É2ôH—"ç­@ÉbðA|)½Zv‹9jb®{æGLdQØ+kž‡i]aзޗoW|‰È¿š¿?!ÎB“ ¼¥Àþé‘«7ãùH’ð¥6L)Ù%fç({Ûj!íy"½öºC.¸ewEÞýŠÔN&Ât‚lü€sûsí)Ò¥B7º]jÓ2ª¶ý»¡¬óaé ðÁ¬Šïõ ^Ëÿ©:ÿúñšb°¨qÆ¥þrŸUì×2VçVµ¤ýØz?_Ktqˇ^F|X<ºp5¼2¯k»á)éPÎËmd³Qi¥Û“)rôÚrîÂe‘”ý<2ï(9\d_ÏÚ¸dö﯉s¢—“ 5MÄÁb× #ЬlqF8–1êÎ`ÏT_ל¦æ(SFÌ!:ˆ7?7 á~aqç8jª©€­á4$¿­)û±&™±„©t~x—è_–ÒÚíôVöÚßBÓž¿Û'×T|]ÂÇNBÛb•³³ÞøÅw¾ í«Œ5½‰®{,ÌZž{Q ‡>xØàƒ­³n@ÜjºÌw±+V“[Yo™E°â¦1yX5W|$­£Íë{$rÂomwiüèw6&Â[`öFÙÊ?f„[±¯f;Ä‚ÅN¢›ýœS`9;‹^°ù{y‘·/‚Fßø’i2"`£¶µQ ­œáÊ*w»²~%(ò‡¦Ç⨆·»4<ÚÐÅÔݯÚ9uNÀÙ´ÿ/ܱ’þ-ÛKÔ¨PQKoî¾ó†ÜƒÞGËÇ€cÕus-þmËÜ¡æFÎ8u)ÌÄ·¨,`§Øüék¬Ð9žà—Yæó‹`¢x¨Rä*Ííà†|m¬&X ±bÆÌÄ'9ßÖ×½–uÎFY#°¥/ÄÇ ÙWãÜ «Yð«$ ÄW÷%ù`ãû¬o²Øñþ¹àå.Ìdé­Beè·(Eƒ ωF”ßé™÷S½s!džÄ)g?ºuAû(38¹KXÿxŸ¥™ŠóNäKéëÅcÅ¥GL¬Èû?xic ®q3–±ÅuÂX€yÚ[ªX†çµMßøI\R‹Ê|âÕµ<ï}P|15 ’ý7±û<“ŒÖU\IžÑB¤Ó„ÈîXBTG® ¢I™æ}3’}5$ŸDû3Ïvª3#MGÇä²”m Mæ êÜtãÝ~XžUú ˜Ò°s~”^-YÜá{Qì|Ðܸ¶ï&F5·ûè)Q„A“níqâT\·Ê:¾\ÊÁ9dGûã¾ÛÛ9úx¶daï,ÝÍ6O‘“|ûpùŠ Ó¡_ŠÓ½li¨NŸ`g¬ppólzCZfNÊ_nÏ,€Á/4JùŠkxg„=.|³ÚïM¹wFO¡YG)h¾{þøÅzõÇ·ö©4 ›dhb½,ÇYïÉuk좬¶©Œc®!½ÐXùçPÇtEUÓ[kUí7½èxüúó„z»Ÿäj‹:R´í²î“.Äú×úÅKoîÍ{w¡¬º5±â*JÈVb¨!Ka‰‚2k¦Ô€ì¢ÁÉþÝ‚ ½Tï¶.ꧢ˓Ó;Úåpú¸Ï¸M%k·Ìˆ6iéaÞO,A£Û{åqb1‰Å5©Z!¶âú’7e3¬›F[vF'CÜÊú`éÁYrÔÀ ¶O=¢4µ)µ)‹öæ‰XãÍx)ƒéÓXÝèk/N9Òa3-Á°÷¢ZÒ{¨ÌfVïÉCãrº[ ''£ù˜”œìò‰>†¡eL¸Å„Jh)ª ¬¥@éÝË®d<¡ åg±]¢SxÇqjL“ÿ0™Úœ‡«ÌþÿÎruÀhÎ0`Ô”aŸÖ×™öU¸p£Ö Ç¥ nÚN^´ø…× üQ¾!jñT¤Á_±n¥ÄO¼w8¢I2ƒQÝê`¦rPŸO­wš]ðˆqaägßX„~õ´6^åÖôTªÅnõΪ!Žkð$u÷n–”|ño‹8Ä´6³1,Ìí!J˜wŸ Mm½Xô‡Qx?¾W· úÍ‚‘¥èÕ¥S3FGhw±Ç/ &ï°O×{Úï@ îrUdšXi½r‚/nfÈþ(÷ÕäýU÷¼|Ðëº:ÒàóWÑ»ÈE»0övY sü+ŒŸL³5X¾V…pÀe}.î·«iÛ»o¦1xL#žrªïºú~ÀšFÏ£­§'a÷I²bKfAÓ:ùNqIæ*8ûwº (/¿gÔUŠZBÆ)³"ÍÝ[ªÁ{ù·"cMM†àžÅNìw6¼ç0Ü?mŒrâTŠ·D°ªïߦãò ÉSSÁÎ=¯ˆŸX:züuE¸8hWÏNóÍTý*ÃójÎ(]ÐB3¬S™™$ÇÐþ§îè +ݶu¨ÜájI1_ïôש(ã®d/+û"]S3 åw/x£’¤‘þ´’½Tqt -‘ØœàZèç¶xv1.š¶˜V$'t.鯨’.E8Õ49„²αû ¾¯œÞJsêÌ_º …FxcÍ8{c•":)2rN~ÆáZÜ:]ù’šQ:±x†tQ¤cV ìñ7¥r¦I$»Ò’##=[TÛH´ ;/ŠK>=ШdÿrS7$Ýw€lÔü¦¡àCëúf,þ«]*ê±á4%w{JYv¤Hõü—hüó bêxŽ'ü|ɱ {-›àÀ·]ýY­šlºž  mº ó|Ó’Ì0:ËZM‘¼w«H:Åþ¾%’h\+ÀW÷=_N;¦“)6Fœ§+£#³ü9JBÊ þËIŸZ+‘­Ïkfxñ–ˆèÆ&r™‰lÆûÁ¨í’sãjäÅ„ÞMîÜ1M#%-:ÕÄÀ¥BuîÓ*ívó¯&ʯÍø_cž_Á´€Ìî.ƒ$å=×î ·~xêOKÝzÝúDô3¸¿--ðQ õN긤?\f0;XÍq¤™(姈V)¯MyjM¾!™ÑT‡Ü6,y ‰°À0SguÜÖ6C$Ù$»ik@",‡á쉱_Pe8‹»h·. Ñ•ZóÃ!-F×TAO[ž:Ú<òeûpð9Ì—2_ø hˆ”³þSd¹%|O¸ÐLCQê@W@ÜÁŽ “´FJ ¾Üù1ýì»Ãq¹ŽýqÎ-‹ãœíˆ) ßÁ_$eëï$ñÒhÓ3òóÚj8Äö^`1ø÷±¾¦m"< ”;Ö/Ï*¹sS‡‰ÀÁË$„ðÁ¢UEÇ[Þxê³_^VÃZBÕUD5êp)+Ñ“¡¸ ®ëØsŠF?Êžš¨œ+9tõTÊÐ#^)\ÚQ]ðgWïKIñ àw=·ï4¨bÃÑáÅ¢ÍæŠxË2…oŒPõ¦½Î /ÖÓ¼Éâ’DÃo÷éMZ»q%Ê:»}*;Å€×[¯Qa/iR‚÷_ Ôýf·ßÕ²j¨¹aUâß-̧ÕÀ,þž‚²X+&48Gc­È3œ›„¬"¦6Ýÿ®cLEÛÅèin+n›À–TªžOÞˆRˆø9æ»T›kÚl÷ãºîqéóB5‡ƒ=n½ÕÏÛ–qä‚òžÊtã:e,) >¦˜¼æQ†+%…-àŠ8ƒ—‰ñµãÆÜnóT>âXïhwVdÀtl݃Ëw?)<·â:kÈÄ…ª¨â6¤ïF5>u¼ ©{ò|}#gB£Zûu’1c\"§ÃÝâçaœÎì›[¯ >þ†¾"ø,§Vgüõ·u⪿´ëŸã÷|Csç-OµÛ{§tûýðR?L¼…ɸ8(ÎÅwÚà Æ*ûÏ~pŠI”ôýE`In ²Ä3v„I³d#xéZ¬ |BPˆ0«D¹o&<§nö¡ld³oòVãôXμ¢vùh2n¼„!•¤G±eÅSs¹!i¹‰8m…ÞÊL¦¦™bäwï÷í¾•”7Å%b<5½õ:&ÒÌ·5¼°Æ{üË­¯1Ø¢Ÿ,äGÅåd²ù cCÙñ‰{öÅbY»c‚¶Ô¹ÝïÏ&0ô3aõK2„û½¹ë®lŸ?öp”¯]•w˜äÊóóLìXñ!žµoãç2ë«ê´ŠÙ*Bˆ™z. §¨vÅ…¶b•à ê£ú5ã;'™×á™ê‰Éh»]]Rç£+'±þֆᅣ–—,9Å%J¸[Ö¯¯]d:[æØ‘Ø*Y=ø|B_s^t\±ýHÅÕ¯ôBß’$ÑF(éBi¦®ü)ͦ [·°HÖxôÙÜ.7 á›Î4ÅŠ@MþbýÑ'@„É›éXd#Éh—a½©q¥îŠcZ§,qþ¡åÃ!ŠÚÚóå#!J>o¤ÇÃðjK¡‡¹ ÈT§G~ç•õ9<É­§_ıfóˆ/ÂÏ_NoŸk…’4†D¶™!m¨ì!UË¥â²ô>¯u8,›y±Ê¨îŸùûí"ZãØ•"zÿ”3©/NZ§©¥ Øù†±:éjë¢-© ã8Ýaû«pE{žÚï#6&-“˜»7=<ÏúœŠhjjSÊÚ;ÿƒ¾ñÑÎáM‘õHTF]ë¯PBŠpÙëG§ Ø>¦1®aÎý댲×]¦:ܺ{ˆV‡ H ¸¸ ËáE9Goæ¦1e˜°Ò¡éÐa8°Ê†o ²0Ñ0·v:Ž¿(Ôyçk@ ñkLÑwœß97¼WŽ”mè-5ËKùÜM¡Oö#÷q8'Ñ¥¾\M‹S轈ÉG@´¿{Ë—7~hãe|^ >ÂŶH,ØtFC1%¥†ÍWpëç‹8–¯KוÐaw_JB†Þgj/Á£½™vqçNMÂ+Eâ¥@lŸšùQø“% ežô¡ÇX˾¾_Õ@Ô¾gÃݰÝ×8íD¹;ûtzHr†6Àí:,}†UO/ŽËq¯Öõ £®§FÙÇjVØòb{½p3¨Äl{inø’|bàƒÝàþ>rh<'×`0#_+ß|?í1È&ö@,Å=¸×–½vr–4¥·©"ƳÜX°¦HIáX,úÛ'«)É&Û!$¢Áì‹Àü5áœØ8Âç®Ä'¨5·t*†ÎÊÄÀ”¼´"Ü /MÓ˜dmV6_ÑOº…ÃX1«]rªj¦ÔgfÉ<™>êü„€<'â…aB=æBtp<F½&ÈÖ‚ #?áwþ~¢Ñ^áŸS|[pÄ+iôþ&»æáÜ®ÙØÁXæ„Æ½Š£oÆ0DJmù3ì Òô«[k?!FY1í=]¥^(=Æ¢}%Ï*³ŸÎ¹ã+©»ËìB]!1&£ ¡•¢Jmï‘Ro“øªtHò/¿$$R™HêÉSS„ò¯Ce4ÂaCصhkHÒùÂóußa}ût¿óEçôœ}mì(SÆ„r»ÍŸ_˜c…Bzq³”sÑ™Kƒµc|å'ì¬\ÆtŸ¨·?ÿ– Ì[éדږ¥êtŠ‘"Ïù9Âo‡½Ñ)+Ú(I’ЉÕiH§ßǺå\Ëö‹ÜD“pÔí? ÑÁ ”Ç: ½t «-}Ÿ¹Ð±±~À¿ìéõ›+,Ãļß÷%#¼F:j¾”äÖ볊Þ WK¦cleá”"~l&¾曢°T~3O­ø"Í…qb`ÊwÜüö|g°«[@‰]8`cÂ|Ý/CþYÏ«d¨“0ôé \,˜å”uè1·3·ÄûçäRþb)‡óô¸ã@\8\oi‰RAÛÇðêôqT—áöºÍ‰p™h=?O!Í­õ^my©B¬P“ݹ¸Cõ¼üO5iÇiÇû/!÷–˜¨9ÃŒ˜Þu\&/}å«ô[­¾ýЖQPÝÌ­]Ïr€X8©¼P«=Ïô8—geDw?9 ¿/Û+î±³m‰è•ÙªO;·¯Øér |¹o~­²FdB·]·ìâ%`2+{Ž)±eüA '´2/NRPVY9ýE¬ÈuÉÒ[±]ú›€â)г‰ÊUÆ0Z¸ð¦o€óãÇÐ0ÇíòYŸgò¤¯é™~e{ñ¬LÏh 5¾í÷/ž­ qa "“NwË_eO 8Ý3=g†âÍ1ÉÌÇÀßï»3í´KbëöÒî•úm|…úD/ÃØW6Ì5V;‚1ˆVü儺¶³•iPpAÀµðp¹/Îý6ØZ%á"’¿ñõÈ/‚Wg¤!x'EPKl×nŽÞ·”@Cz¿Æ1ä@MzK¼Ó³Myaƒ®7X›i[0Ä;Ô7.Öv¡¹wyç !è²5Õþµÿ%øÁçNͯwg3*}7' Y årÙ^©›Kü!×lèGBE3£xÏRUfó«Q-ª²HäâWá Q#Š6tRGÇMaûGfN9ô¾vû_‚Õ[œþ¼n‹tþÍùòŒRÖÇ—7í8õc1qÓ쬇®÷¸6Lç¨ò[hðA{ÑWBtø©Rí‰-åÀÔž.)qL+àú†H?W„Õ:=D™ç*i_[G]ý˜Á^"ZþÌEf œëý¾T ©¢F?ÆØ¹íºÏ•¸ì†iR „99ê}e“<ÄXý”îÓ´y×c]GkSàõgÓרtô=W…ø–'Ú@š-b3…&gÒ"_„ ÍQ®fšF!¦”Mýô ÑÑÊ4÷óaßTæyR½~ßÜYÅkLÂÊ€2Âø+*œà¥MV”¾ÒÍð~2š— „sbÌ+篊õÞi‚ÕåDN*ô'iþVúq Ñ•¬ãò{fÚ‹IêÏ5C®šÕl§örª›¯0Bsªƒ³~˜µî¡9pîÇxÌÊÅjK)ù-ŒŠmÐ-ªG’‹D¸,ka¯‰Ä×Àˆ¨ñ‡ ãϽIDÓ¹ÌZ0{;eß“/]Û8e hÓ>˜àfxT‡i j·gB6LU ¼ñ4%~:ðÄN…öת»¯ï ©oÆQ{ˆÆ4ðííPk°®4¹?ýtÿ>G‹%ÔPÆ;×û<›ûXßøk^9h2ÞŸøLìì˜K ½ßQ8F¶Dä×ôiá¡@ CT(X›»ƒÍª±¥;Êa„ަuG ÙýÎ2î$ûyõÝרöÊŸÙ‚XÏŸ:)MÍ…­P ëHaÏ?Åñ j¸|ÚT§¬!|m’‹‡èóè˜Á<‘SwØÜ6›¹Û¤¥P>¢¯VÎ\²/Ðú#pòê"aV¾“r°6}É‹D¨Ž€*Ç/»€°(ÜÕrÉáã‹í¨´×”+œs!ˆFŽ©« ‘ª+@‚LS¼ñÅÌxrbcé·‹ˆ¦QH¿»š6ñ$Ðj™¥>·W(6/!*ÀIZËø)”mKÄP€(t{ƒVw漣(í¼v6‚¸'&½Ò0Ÿ+·e”Ü…ýÂJ€ÁËÐýKÜEÄQéså¿W8t‡ZJlž¹µÅµb±ˆaÓ¤{-9µØì÷ê^÷ì N¯Tw(áöEÔa4Bõ{‡Þçß#W6±L-›‚!ûƒ9P†ñwWÓ“ÀŸ¨ -Áð=`Îêw¹Æ—³ã‚‹”ã’³’ãw½$“Å/íÙX‰¡‚*²Ùs <„Þ:¯:ëÁ=ê7‡˜™˜˜LkšbXFtK`è¼(Ä`““¿„B°%d~ƒ’úÉïˆzjä°i…ÀFht¢ÊNÓ¨%ž³©Æ K9!k6•Ôíö0ac™T›o3¬õ̲¤÷îÝë÷ÕÖ´E‡´&8xä–÷,_EG¸4G.ý£ªêw"µ­çnóŸÚ[¸ZŒtg§ø~šÛX;fÉľqs²y“ èò€Ñ°¿ŽÊµúP¨­Ì‘ ]&8Gõ.’-M@!4æíj`í™åm?ª\qævŽáõ)غò’šÓ†~оfÁ|¿»>–Dåêv5dáS_Tõ–¥âøA’ëŠÔ-ÕÍã%R´¿É ×gÿÊð¼úo,×ÀÜ¢ŠÓ™çßã#î7& ú-±œ 3ÇíZ*ÕËÞ±Ln<†¬ò¨„U“IÐt ±²Û (kZ¿ÏÎYê¸r¦¿ž§FOOn2׳àYcò‘ ½óßî[’jŸ9 Þ·-ôsàŽ‘_8Ÿ³¶€Ë¿éì¦!º,’S`ã¢ÊmUâŽ÷GXû%Èa¡WY¦]c?¬R"÷t S=}tBº…CïÓ$÷<Ü07\&ê°OÌÓéÆ}åB'”œjØ6écxržvl¨âµ[2—„¡ÑLwdòòN´f‡Ðê¾&»$ð kOfíÅdÕßzÀFK?L_tO>b€–Ždù!Þj±Ã:ûÜZ9'ñCÚlõ@þ²9XT®vq:&Œ<0ç(yú¨ðyÓykZ 8ár·\‹G¸ÿý·qÙQƒùgSL—Jñ õ¤)Nw8ìßPÆX… —&e¬,.ÞLÕ–"uNo¬–’Ý—²ƒÂÄDº]µˆW’·ZÒ¶"2ÆJKlŽ'ÿ™¸üšÐFçžÜ1ž½kªÞà\%qGË-š„ZoÌë}Ô8¾MK€yR5)q(6|â>•’G>À6?ÑçâçjÎcÂV@¨ÉؾKìªã!3ù÷ÖG· xB¤CQ|18Žu†êÞÙRUù`“/p´!âñ‰iû²m=‡mW b¿&Yø¢m (ІCÌ­àöù2æé¤–ZiÍåâ™›)5» ƒ»1kl ‚Ÿ^XgkïÔFHcLÑmá§„mXá8v©Š__/ç|j#†xfÍnF¼Ê’ “;ý7h7CÂ7šäŒø‹6¦WcQþü¼°|ŠùU ©ß¼Y×ÁŠ™å;fѦ½LOÙ¬ª¬+¬þõsš¥†n¦ß(§¡¬cP…{dÅøqâÃ(_;ŠÉåæg%®bôÊ„OVæh rÆcÞºˆJ©áÙOù{;ñvêDÄ(5èmÛoc§+Èhy ¯6áž‚&Œ^Á´?ô'ä?L\4Wz’­wç.䈛›*9A_s|î .Å_oÚÛ”}·¨çä_Ò}{ƒ0F´•Pôuf–_âKßwÈÑRê]ôsË ¦˜8sO|NÔø÷X¡ïø3MD=dc¿›…Ö¦xá¬]]ägÌ©BÛå|(’ßè…„{˜6j…mVå$? vÏ›ÄâJኋ½[=HÖä##©éo¬â³¨›þµs>þiªÂô GÌ#Ω]Þrn(¼b"8ÉÕñÝ· ¯+Ρ‰‡ƒ‚+RB|vK$ÝßÝö T­â:ôRÛ¤ZI5Ö˜­w$åó¿•0ËD›Ìºµ»²k£Vû.—•_U¬Öe6½û hT'±-ܦdÖ/QbÇetâÃMø”kñ)ëÒ­6&<(Qê™çŸUÉ$Òû¥›{;´ýà|xÝF¿á”äaÐ…Y^~ÏÀK‘ AvO&ºàM¾i{.Ñç4¥’@ÉQ#X 520øg{# Š5ì™Ï1Æp¬ûü6f$²Û»¸ïg¥‰Z9 sí¶!Ö†êP–Yî›ä)E*¤B–E]:·Ó!'̽ñ"Åöñ,ø1>‡IÎ5PM†v ¿â1¯ÐÇ È€:.øè³uU“G]Ð¸Ì "‘×£™W.Eiênv®U+)’KM §PßLŠ.™Y Þ Aç×­â[õë¢Á§tæÎýÞ_‚ÈvOªªHTu"ƒ½nFO´¾-÷L.Ò-ÿ£.9½u ­’R&ö²SÅúßj[ÃYÞÿ9nôÐ ¼ÚÃÍ¿ëŽaärð0ŽÛŽmØ ¶Â,YØ©–Å ålÈÛV§—æwÎmk·‘ë¶™³õ½e\Çž8NMHÌ*û¶ÄŒ"‚æ4¾ € fâRÒO8Ìx-¢¿Gû£6¯¥Ö¾)6!áÒØ–«b?…È>M<éÐÉ$.{ë½–gK®Z£«¼-W™oqÀÎ7è «x¦â>ˆHÿfÕÍómWÒ &_è¦Þ/€2IîNÞž¤»‘á*k÷@çrº¿å-x!!wÝCÞSl]™ Q„ËiÓždÙ›¯4›ž²)äO*·-”BEÏ€:üÅäŽ>›Z™æ’œ®¡ ÇûÖ}*(ú3íïy¤ëa¿„ý#uðMiämdí¡Î;š›t©J'»tþÓhqUöb¼L%Ñ ÍÈ3Iª?5¬” pÓ{ÜWÚ@X ¾kÉs6šEžÏ…eÁôà‹eühÜî§üëw­DÛ+W±UDSà¡¶ÎÑúψè1—OŽŒ°«¶ú¾]·ü¶Ÿ1µ"Œzw‡'|¨\Õñ¢ÅÙýn¥þ³iëê^Õ! ®÷ûÞŠK+ ê^p>N];[Lì&¸D—E…¢.’¡ÐW)…˜Òªr•ÚÕJ2ÇMÁç·¹Fv”½vÆ»ý<ÖCD¡|O`òš1Ÿ¥P!Åö´ÜV©ÅúRé21‘÷¸„Q¾É—Œøšî¼ƒ‚âtÍc83JQ,R‹+²Üq\¼¯‹ †qƒ6ã^RˆóÈÚRðù¯«ÜUnFièÑÒŒ\ð!õ{vq›Qçþø8Ð$h±Œëëð[°àþÙÇ"!|,O†p%´Ü…wQçJ¬qøéZ爐÷‹-ô÷Ä7=ð5nx…yÎ(Ÿ#§7QùÖéÛ ]bÞ¨k‚2]ˆI‚Wy«ØªdåŒiÉÁ¤æ»ÐY%‡ð¥ï”äÊ_¡W ¡“™µ\âbÿᨱ¡dçõ!‘F‘Ì(‡XÄ`>Ž ýõÉržÕ0sN¹k†Ä”åÅÊÉ8'›-DîÜdòô‰t@{ú-ä®3p冱>ò¿D=ÁeŸe¯ Ñ„"è—…'7ÒÃ|ðV3 »èqûÛþn€+Q“ÒU4õÜ"‘sQÔPî¤ë>²´¸¸Ÿ)Çm1_£œ ¬!žÄJ>âësy£¾EîBfç±–Eôózš´¨ï#Á]Òb'Ž`$Ï}ô­Ÿ \5<ìÀˆ®’Gù8)[@h¼ -Ê’äç{ði/öq/WпMzVr#&Pí³©†å´ÛSgãF“TI±Gü¥dˆ«¸²$·(ÝÅËK¨¥Á2KÚBAÀbýS£vÕACAÅ9ÓŒ"ã"Z¥ÆKsûÏ·½”LõË%O«H6®ÎŒPcÿ-CÞ²JÇ[g[$GÏGÏR5ÏgµÛa8ùë”PW7¸¸ÞÂNnJžG©3Ó¿pʬ­¢…‰ÐîLàŽSö$Q8ÁŸª®W.ö¹u&AŽ8U!d”»¦jà màÅp&Z,ÅPðÒt8_ò[áZ¦ÌTtÿ™ÞïsßëGÖ Q‚²c‰S×ÛTaái‚!^®kÍ?!a û).º'Ùb>þ©ˆª!~1ß6²&¯"ý»0âÍ>–ù¿amíÓWà®T‡Ú°‹ÇTR+A›ñȯ"#„ìt_ˆ®ðqÅ">Ðö ©ìLNÅB=Ü%Oð>7Ø~k¸õŽj¶hXÀ?úMQ©‘@($öDœÕw.¾“óäÐi`_/…Ç#ÍÆÔú=¸ëýå;>Âr¢œÑS0NÈÀ¬ßoâÆ© í?Íen]ŒÕvñ@™¹;Ð/"=Ï2òKnê{- v°ù×™@$£Œ #û,!h8»itâ‰$àO4DíÏ0é¯ù£÷€ûÈcˆ‘Œ’X>š€Û”Û·ÜnÅÙ-¶Ã.„fæxG÷žàaÔ"ïcù¤º*rû á ¥fÓåA¨¬_?`ú›¬ôaè»åQ(Xàl[Ü<ç»Q?øå²»¨y`$4ÞíLqŸÌ\›ááªâ‘CšV‰AàÞ]–áÞn¼pèQ€PÈá»x‡âðÃKÚý|ÏÓ}àÕò¤¥àQír5cêwÕÍŽTá_:rä£Ë÷vþ§7tl(iºOy¹ÇÂKó7²ëjÿD5™&Ï~é;Ú|z^Š_6°W6ùû1Xm’Ÿ£¡´F8‘ÎågqÖ®ôô^WˆÚiË:½y¡8„q!ëýPÜZ–3Ï‚á…T,VþTÙÍØÎ… t4#¤PC? 8º°\õš+[Ú·P·ÍÇáÕ§§ÿ)'Oµ>à–~¡omÆQ"×Â{öïÛéLà~?`¹Çi½~ðÜ(ºð^/[ƒ¿u-Es—A8“Ñ4=þþá/ùÃùcϰÃܯ紶 Δ_ÎYHYK°I:-‰×Êtz†â|Öþ“M¤¯þßèôø ZA=ˆSúk1ݨËRÞÝh ïá§q–¿ ‘¬3Þœ.1…gR—ßÃüé"Kw£=¤­RmæÈ­^ÆÙ¼ÙtºTœÑU•ð-cÃOÞq´5TÛ0Ú½Ÿx«øÁÿƒ´Ž7ÜLãÿ{ ŽI¦r­ 0ü…Ŧ¿‹lâçèûÀ«T”¡^æè‘ iÂq™Ë€ê«üuÍü!3ùDÿv'gˆ››ÚÅ$§sçr€ËÇïh°¦èº ïg8(Q/t΃”Íñ¹á£ZÍë]KypU¬x›òšF­YÇy\'²úÜ‚~¸L×Ìw•¨ñ(çûGñx¡jÆ_˜’Ã7Z`ó»Ÿî\µšl½÷N}úµ°r`—Òd´ÿiò| +¶Õ˜4u Ë !Эûc¨²Ïà“BÈH0oue€ny¥s1%ÀÈ[¢X†`Q1o ph=Dµ¨èð`'ùF¥M«ÔêÂH@IøòÒvØdOBP Ï$ _áÀIÄ‹ß 5= k¹•[~Uw*Kbºuzg¬çÔ9Ìt®C«ÞiÀ"=Ñï¶á»§CÌækáˆÒ¸Æ´$‹¦ŸÀÑh—žÏý–p<÷‹7Ã¥®(à—+K†‹#˜&B±sSf¥RA¹Ë>êúã©7¹<ô«b3‡§•ÃÃÓ•ÀÚË,K2cåMú@ƒžb ÷,[™º™|– µýŠð°¤4îÐ={ßÿ~T¤W;„0Ó0 `žeo|´0¸Q¾Tq¥Ç„ÖéørÉŽ…Á|¹KT¦êÅÉ ÏôÕš endstream endobj 209 0 obj << /Type /FontDescriptor /FontName /BYDEJI+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/C/D/E/F/H/I/L/M/N/O/P/R/S/T/U/V/W/X/Y/a/asciicircum/asciitilde/asterisk/b/bar/braceleft/braceright/bracketleft/bracketright/c/colon/comma/d/dollar/e/eight/equal/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/nine/numbersign/o/one/p/parenleft/parenright/percent/period/plus/q/quotedbl/r/s/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 208 0 R >> endobj 68 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YHFNHR+CMBX10 /FontDescriptor 183 0 R /FirstChar 73 /LastChar 121 /Widths 167 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /EDHOWJ+CMBX12 /FontDescriptor 185 0 R /FirstChar 12 /LastChar 121 /Widths 179 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /HROJHK+CMMI10 /FontDescriptor 187 0 R /FirstChar 12 /LastChar 120 /Widths 176 0 R >> endobj 37 0 obj << /Type /Font /Subtype /Type1 /BaseFont /WTIHHF+CMMI7 /FontDescriptor 189 0 R /FirstChar 97 /LastChar 115 /Widths 168 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ALUMCR+CMR10 /FontDescriptor 191 0 R /FirstChar 11 /LastChar 124 /Widths 178 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /BHYHAH+CMR12 /FontDescriptor 193 0 R /FirstChar 44 /LastChar 122 /Widths 180 0 R >> endobj 4 0 obj << /Type /Font /Subtype /Type1 /BaseFont /DOQUVF+CMR17 /FontDescriptor 195 0 R /FirstChar 65 /LastChar 118 /Widths 181 0 R >> endobj 130 0 obj << /Type /Font /Subtype /Type1 /BaseFont /WIRDCL+CMR6 /FontDescriptor 197 0 R /FirstChar 49 /LastChar 49 /Widths 165 0 R >> endobj 24 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZKZGQL+CMR7 /FontDescriptor 199 0 R /FirstChar 49 /LastChar 56 /Widths 174 0 R >> endobj 131 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KISTMD+CMR8 /FontDescriptor 201 0 R /FirstChar 11 /LastChar 121 /Widths 164 0 R >> endobj 15 0 obj << /Type /Font /Subtype /Type1 /BaseFont /BVQHUS+CMSY10 /FontDescriptor 203 0 R /FirstChar 0 /LastChar 106 /Widths 175 0 R >> endobj 81 0 obj << /Type /Font /Subtype /Type1 /BaseFont /SDVCAT+CMSY7 /FontDescriptor 205 0 R /FirstChar 3 /LastChar 3 /Widths 166 0 R >> endobj 8 0 obj << /Type /Font /Subtype /Type1 /BaseFont /STMVCT+CMTI10 /FontDescriptor 207 0 R /FirstChar 46 /LastChar 121 /Widths 177 0 R >> endobj 25 0 obj << /Type /Font /Subtype /Type1 /BaseFont /BYDEJI+CMTT10 /FontDescriptor 209 0 R /FirstChar 34 /LastChar 126 /Widths 173 0 R >> endobj 10 0 obj << /Type /Pages /Count 6 /Parent 210 0 R /Kids [2 0 R 13 0 R 22 0 R 28 0 R 32 0 R 35 0 R] >> endobj 47 0 obj << /Type /Pages /Count 6 /Parent 210 0 R /Kids [45 0 R 50 0 R 58 0 R 66 0 R 71 0 R 79 0 R] >> endobj 85 0 obj << /Type /Pages /Count 6 /Parent 210 0 R /Kids [83 0 R 88 0 R 91 0 R 100 0 R 103 0 R 112 0 R] >> endobj 117 0 obj << /Type /Pages /Count 6 /Parent 210 0 R /Kids [115 0 R 124 0 R 128 0 R 133 0 R 141 0 R 145 0 R] >> endobj 151 0 obj << /Type /Pages /Count 2 /Parent 210 0 R /Kids [149 0 R 158 0 R] >> endobj 210 0 obj << /Type /Pages /Count 26 /Kids [10 0 R 47 0 R 85 0 R 117 0 R 151 0 R] >> endobj 211 0 obj << /Type /Catalog /Pages 210 0 R >> endobj 212 0 obj << /Producer (pdfTeX-1.40.14) /Creator (TeX) /CreationDate (D:20150701163326-05'00') /ModDate (D:20150701163326-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 213 0000000000 65535 f 0000003485 00000 n 0000003380 00000 n 0000000015 00000 n 0000491586 00000 n 0000491445 00000 n 0000490878 00000 n 0000491304 00000 n 0000492431 00000 n 0000491020 00000 n 0000492716 00000 n 0000005506 00000 n 0000021099 00000 n 0000005398 00000 n 0000003607 00000 n 0000492150 00000 n 0000017884 00000 n 0000018027 00000 n 0000018125 00000 n 0000018160 00000 n 0000018402 00000 n 0000025439 00000 n 0000025331 00000 n 0000021239 00000 n 0000491868 00000 n 0000492573 00000 n 0000033579 00000 n 0000028478 00000 n 0000028370 00000 n 0000025541 00000 n 0000314500 00000 n 0000031499 00000 n 0000031391 00000 n 0000028603 00000 n 0000055884 00000 n 0000033471 00000 n 0000031612 00000 n 0000491162 00000 n 0000052669 00000 n 0000052812 00000 n 0000052910 00000 n 0000052945 00000 n 0000053187 00000 n 0000059650 00000 n 0000058350 00000 n 0000058242 00000 n 0000056014 00000 n 0000492825 00000 n 0000066146 00000 n 0000063999 00000 n 0000059542 00000 n 0000058453 00000 n 0000060784 00000 n 0000060927 00000 n 0000061025 00000 n 0000061060 00000 n 0000061302 00000 n 0000104824 00000 n 0000066038 00000 n 0000064118 00000 n 0000101609 00000 n 0000101752 00000 n 0000101850 00000 n 0000101885 00000 n 0000102127 00000 n 0000107761 00000 n 0000107653 00000 n 0000104931 00000 n 0000490735 00000 n 0000109850 00000 n 0000114392 00000 n 0000109742 00000 n 0000107911 00000 n 0000111177 00000 n 0000111320 00000 n 0000111418 00000 n 0000111453 00000 n 0000111695 00000 n 0000116922 00000 n 0000116814 00000 n 0000114499 00000 n 0000492292 00000 n 0000119795 00000 n 0000119687 00000 n 0000117072 00000 n 0000492935 00000 n 0000124767 00000 n 0000123625 00000 n 0000123517 00000 n 0000119897 00000 n 0000141616 00000 n 0000124659 00000 n 0000123717 00000 n 0000138401 00000 n 0000138544 00000 n 0000138642 00000 n 0000138677 00000 n 0000138919 00000 n 0000145916 00000 n 0000145231 00000 n 0000145121 00000 n 0000141723 00000 n 0000173398 00000 n 0000145805 00000 n 0000145333 00000 n 0000170176 00000 n 0000170320 00000 n 0000170420 00000 n 0000170457 00000 n 0000170700 00000 n 0000177427 00000 n 0000176535 00000 n 0000176424 00000 n 0000173506 00000 n 0000225374 00000 n 0000177315 00000 n 0000176651 00000 n 0000493048 00000 n 0000222152 00000 n 0000222296 00000 n 0000222396 00000 n 0000222433 00000 n 0000222676 00000 n 0000228305 00000 n 0000228193 00000 n 0000225483 00000 n 0000231356 00000 n 0000230707 00000 n 0000230595 00000 n 0000228398 00000 n 0000491727 00000 n 0000492008 00000 n 0000267649 00000 n 0000231244 00000 n 0000230837 00000 n 0000264427 00000 n 0000264571 00000 n 0000264671 00000 n 0000264708 00000 n 0000264951 00000 n 0000271215 00000 n 0000271103 00000 n 0000267746 00000 n 0000275141 00000 n 0000274438 00000 n 0000274326 00000 n 0000271330 00000 n 0000312770 00000 n 0000311610 00000 n 0000275029 00000 n 0000274566 00000 n 0000493165 00000 n 0000308388 00000 n 0000308532 00000 n 0000308632 00000 n 0000308669 00000 n 0000308912 00000 n 0000312460 00000 n 0000312348 00000 n 0000311708 00000 n 0000312540 00000 n 0000313019 00000 n 0000313045 00000 n 0000313108 00000 n 0000313145 00000 n 0000313830 00000 n 0000313855 00000 n 0000313880 00000 n 0000314183 00000 n 0000314315 00000 n 0000314748 00000 n 0000314774 00000 n 0000314835 00000 n 0000314871 00000 n 0000315262 00000 n 0000315329 00000 n 0000315947 00000 n 0000316570 00000 n 0000317027 00000 n 0000317659 00000 n 0000318296 00000 n 0000318732 00000 n 0000319071 00000 n 0000331589 00000 n 0000331848 00000 n 0000345480 00000 n 0000345797 00000 n 0000356098 00000 n 0000356371 00000 n 0000364001 00000 n 0000364225 00000 n 0000389373 00000 n 0000389945 00000 n 0000401837 00000 n 0000402123 00000 n 0000411393 00000 n 0000411637 00000 n 0000418638 00000 n 0000418858 00000 n 0000426051 00000 n 0000426277 00000 n 0000438085 00000 n 0000438365 00000 n 0000446084 00000 n 0000446360 00000 n 0000453495 00000 n 0000453727 00000 n 0000469171 00000 n 0000469472 00000 n 0000490183 00000 n 0000493250 00000 n 0000493341 00000 n 0000493394 00000 n trailer << /Size 213 /Root 211 0 R /Info 212 0 R /ID [<0838B38052C29D6C05EAB6A23F3EDF63> <0838B38052C29D6C05EAB6A23F3EDF63>] >> startxref 493661 %%EOF survival/inst/doc/compete.Rnw0000644000175100001440000010003012545056257016016 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} %\VignetteIndexEntry{Multi-state models and competing risks} %\VignetteDepends{cmprsk} \SweaveOpts{keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} % I had been putting figures in the figures/ directory, but the standard % R build script does not copy it and then R CMD check fails \SweaveOpts{prefix.string=compete,width=6,height=4} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {compete-#1.pdf}} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=10) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #ensure default require("survival") @ \title{Multi-state models and competing risks} \author{Terry M Therneau \\ \emph{Mayo Clinic}} \newcommand{\code}[1]{\texttt{#1}} <>= cmplib <- require("cmprsk", quietly=TRUE) if (cmplib) cat("\\newcommand{\\CMPRSK}{}%\n") @ \begin{document} \maketitle \section{Multi-state survival curves} \begin{figure} \myfig{sfig1} \caption{Three multi-state models. In the upper left is simple survival, in the upper right an example of competing risks, with the multi-state illness-death model below them.} \label{sfig1} \end{figure} <>= par(mar=c(.1, .1, .1, .1)) frame() par(usr=c(0,100,0,100)) # first figure xx <- c(0, 10, 10, 0) yy <- c(0, 0, 10, 10) polygon(xx +10, yy+70) polygon(xx +30, yy+70) arrows( 22, 75, 28, 75, length=.1) text(c(15, 35), c(75,75), c("Alive", "Dead")) # second figure polygon(xx +60, yy+70) for (j in c(55, 70, 85)) { polygon(xx +80, yy+j) arrows(72, (5*75 +j+5)/6, 78, (100+j*5)/6, length=.1) } text(c(65, 85,85,85), c(70,55,70,85)+5, c("A", "D1", "D2", "D3")) # third figure polygon(xx+20, yy+25) for (j in c(15,35)) { polygon(xx +40, yy+j) arrows(32, (5*30 +j+4)/6, 38, (54+j*5)/6, length=.1) } arrows(38, 2+(55 + 35*5)/6, 32, 2+ (150 + 40)/6, length=.1) arrows(45, 33, 45, 27, length=.1) text(c(25, 45,45), c(30, 20, 40), c("Health", "Death", "Illness")) @ Consider the three simple models in figure \ref{sfig1}. Each box is a patient state and each arrow a possible transition. The top left figure is simple survival: all patients start in the alive state and can make a single transition to death. The top right depicts classic competing risks: all subjects start on the left, and each can make a single transition to one of 3 terminal states. The bottom figure shows a simple multi-state situation known as the illness-death model. Traditionally the first case is handled by the Kaplan-Meier esimate and the second by the ``cumulative incidence'', the third case requires use of the the Aalen-Johansen estimate, which includes each of the first two as a special case. The AJ estimate is very flexible: subjects can appear in more than one state during the course of a study, subjects can start after time 0 (delayed entry), and they can start in any of the states. The \code{survfit} function implements the AJ estimate and can handle all these cases. Let $A(t)$ be a matrix of cumulative hazard functions, whose $ij$ element is the estimated cumulative hazard for transitions from state $i$ to state $j$. $$ A_{ij}(t) = \int_0^t dN_{ij}(t)/r_i(t) $$ where $dN$ counts the transitions and $r$ is the number of subjects still at risk in a state. The diagonal elements of $A$ are filled in last such that row sums of $A$ are equal to zero. Then the Aalen-Johansen transition matrix is \begin{equation} P(t) = \prod_{s \le t} [I + dA(s)] \label{AJest} \end{equation} The product is over all time points $s \le t$ at which a transition occured, and $dA$ is the change in the $A$ matrix at that time point. For the two state model it is fairly easy to show that this reduces to the Kaplan-Meier. The $i$th row of $P(t)$ estimates the fraction of subjects in each state at time $t$, given that subjects started in state $i$. The solution obeys the obvious constraint that the row sums at any time are equal to 1: each person has to be somewhere. I will refer to the resulting values as \emph{prevalence} estimates. If there is no censoring then prevalence is particularly easy: at a given time just count the fraction of subjects in each state. In the simple two state model the prevalence in the alive state is the usual KM survival estimate, and we have P(alive) = 1 - P(dead). For simple survival we have gotten used to the idea of using P(dead) and 1- P(dead) interchangeably, but that habit needs to be left behind for multi-state models, for them the values $1-P_k$ = probability(any other state than $k$) are not very useful. Plots for the 2 state case sometimes choose to show P(alive) and sometimes P(dead). Which one is used often depends on a historical whim of the disease specialty; cardiology journals for instance quite often use P(event) resulting in curves that rise starting from zero, but oncology journals invariably use P(alive) giving curves that fall downhill from 1. The survfit routine's historical default for the 2 state case is to print and plot P(alive), which reflects that the author of the routine was working primarily in cancer trials at the time said default was chosen. In the multi-state case, however, the curve for the initial state (leftmost in my diagrams) is rarely included in the final plot and curves start at 0. Here is an example using a simple competing risks problem. The \code{mgus2} data set contains the time to plasma cell malignancy (PCM) and/or death for 1384 subjects diagnosed with monoclonal gammopathy of undetermined significance (MGUS). Survival and progression time are in months. The curve below shows ordinary Kaplan-Meier survival for these subjects, the mean age at diagnosis is just over 70 years. <>= oldpar <- par(mfrow=c(1,2)) hist(mgus2$age, nclass=30, main='', xlab="Age") with(mgus2, tapply(age, sex, mean)) mfit1 <- survfit(Surv(futime, death) ~ sex, data=mgus2) mfit1 plot(mfit1, col=c(1,2), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Survival") legend(6, .8, c("female", "male"), col=1:2, lwd=2, bty='n') par(oldpar) @ A second model for these subjects is competing risks, which corresponds to our second figure above. For this model we are only interested in the first event for each subject. Formally we are treating progression to a plasma cell malignancy (PCM) as an \emph{absorbing state}, i.e., one that subjects never exit. We create a variable \code{etime} containing the time of the first progression, death, or last follow-up along with an event variable that contains the outcome. The starting data set \code{mgus2} has two pairs of variables \code{(ptime, pstat)} that contain the time to progression and \code{(futime, status)} that contain the time to death or last known alive, ignoring progression. <>= etime <- with(mgus2, ifelse(pstat==0, futime, ptime)) event <- with(mgus2, ifelse(pstat==0, 2*death, 1)) event <- factor(event, 0:2, labels=c("censor", "pcm", "death")) table(event) mfit2 <- survfit(Surv(etime, event) ~ sex, data=mgus2) mfit2 plot(mfit2, col=c(1,1,2,2), lty=c(2,1,2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") legend(20, .6, c("death:female", "death:male", "pcm:female", "pcm:male"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') @ The \code{mfit2} call is nearly identical to that for an ordinary Kaplan-Meier, with the exception of the \code{event} variable. \begin{enumerate} \item The event variable was created as a \emph{factor}; whereas for ordinary single state survival the status is either 0/1 or TRUE/FALSE. The first level of the factor must be censoring, which is the status code for those whose follow-up terminated without reaching either endpoint. Codes for the remaining states can be in any order. The labels for the states are unrestricted. \item A simple print of the \code{mfit1} object shows the order in which the curves will be displayed. This information was used to choose the line types and colors for the curves. \item Since these are prevalence estimates, the curves start at 0. \end{enumerate} A common mistake with competing risks is to use the Kaplan-Meier separately on each event type while treating other event types as censored. The next plot is an example of this for the PCM endpoint. <>= pcmbad <- survfit(Surv(etime, pstat) ~ sex, data=mgus2) plot(pcmbad[2], mark.time=FALSE, lwd=2, fun="event", conf=FALSE, xscale=12, xlab="Years post diagnosis", ylab="Fraction with PCM") lines(mfit2[2,1], lty=2, lwd=2, mark.time=FALSE, conf=FALSE, xscale=12) legend(0, .28, c("Males, PCM, incorrect curve", "Males, PCM, competing risk"), col=1, lwd=2, lty=c(1,2), bty='n') @ There are two problems with the \code{pcmbad} fit. The first is that it attempts to estimate the expected rate of plasma cell malignancy if all other causes of death were disallowed. In this hypothetical world it is indeed true that many more subjects would progress to PCM, but it is not a world that any of us will ever inhabit and so is of questionable interest. The second problem is that the computation for this hypothetical case is only correct if all of the competing endpoints are independent, a situation which is almost never true. The competing risks curve estimates the fraction of MGUS subjects who will actually experience PCM, sometimes known as the lifetime risk. The above code chose to plot only a subset of the curves, something that is often desirable in competing risks problems to avoid a ``tangle of yarn'' plot that simply has too many elements. This is done by subscripting the survfit object. For subscripting, multistate curves appear as a matrix with the outcomes as the second subscript. They are in order of the levels of \code{event}, i.e., as displayed by our earlier call to \code{table(event)}. The first subscript indexes the groups formed by the right hand side of the model formula, and will be in the same order as simple survival curves. Thus \code{mfit[2,1]} corresponds to males and the pcm endpoint. A third example using the MGUS data treats it as a multi-state model. In this version a subject can have multiple transitions and thus multiple rows in the data set, and it is necessary to identify which data rows go with which subject via the \code{id} argument of \code{survfit} (valid estimates standard errors both depend on this). Our model looks like the illness-death model of figure \ref{sfig1} but with ``plasma cell malignancy'' as the upper state and no arrow for a return from that state to health. The necessary data set will have two rows for any subject who has further follow-up after a PCM and one row for all others. The data set is created below using the \code{tmerge} function, which is discussed in detail in another vignette. We need to decide what to do with the 9 subjects who have PCM and death declared at the same time. They slipped through without comment in the earlier competing risks analysis, only when setting up this data set did I notice the ties. Looking back at the code, the prior example counted these subjects as a progression. In retrospect this is defensible: even though undetected before autopsy, the disease must have been present for some amount of time previous and so progression did occur first. For the multi-state model we need to be explicit in how this is coded since a sojourn time of 0 within a state is not allowed. Below we push the progression time back by .1 month when there is a tie, but that amount is entirely arbitrary. <>= ptemp <- with(mgus2, ifelse(ptime==futime & pstat==1, ptime-.1, ptime)) newdata <- tmerge(mgus2, mgus2, id=id, death=event(futime, death)) newdata <- tmerge(newdata, mgus2, id, pcm = event(ptemp, pstat)) newdata <- tmerge(newdata, newdata, id, enum=cumtdc(tstart)) with(newdata, table(death, pcm)) @ The table above shows that there are no observations in \code{newdata} that have both a pcm and death, i.e., the ties have been resolved. The last tmerge line above creates a variable \code{enum} which simply counts rows for each person; it will be used later. <>= temp <- with(newdata, ifelse(death==1, 2, pcm)) newdata$event <- factor(temp, 0:2, labels=c("censor", "pcm", "death")) mfit3 <- survfit(Surv(tstart, tstop, event) ~ sex, data=newdata, id=id) plot(mfit3[,1], mark.time=FALSE, col=1:2, lty=1, lwd=2, xscale=12, xlab="Years post MGUS diagnosis", ylab="Prevalence of PCM") legend(4, .04, c("female", "male"), lty=1, col=1:2, lwd=2, bty='n') @ This plot is quite different in that it shows the fraction of subjects \emph{currently} in the PCM state. Looking at the lower scenario in figure \ref{sfig1}, this is the fraction of subjects in the upper right box. The curve goes up whenever someone enters the box and down when they leave. Myeloma survival was quite short during the era of this study and the proportion in the PCM state rarely rises above 2 percent. I have often found the three curve display below useful in these cases. It combines the results from competing risk model used above along with a second fit that treats death after PCM as a separate state from death before progression. Only males are shown in the plot to minimize overlap. <>= d2 <- with(newdata, ifelse(enum==2, 4, as.numeric(event))) e2 <- factor(d2, labels=c("censor", "pcm", "death w/o pcm", "death after pcm")) mfit4 <- survfit(Surv(tstart, tstop, e2) ~ sex, data=newdata, id=id) plot(mfit2[2,], lty=c(2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") lines(mfit4[2,3], mark.time=FALSE, xscale=12, col=2, lty=2, lwd=2, conf=FALSE) legend(15, .5, c("male:death w/o pcm", "male: ever pcm", "male: death after pcm"), col=c(1,1,2), lty=c(1,2,2), lwd=2, bty='n') @ When using multi-state data to create Aalen-Johansen estimates individuals are not allowed to have gaps in the middle of their time line. For example a data with (0, 30, pcm) and (50,70, death) as the two observations for a subject; the time from 30-70 is not accounted for. The method also does not account for what is known as panel data, where a subject's state is recorded at some a priori time such as a physician visit but the actual times of state transitions are unknown. Such data requires further assumptions about the transition process in order to model the outcomes, see for instance the msm package. \section{Models} For simple two-state survival the Cox model leads to three relationships \begin{align} \lambda(t) &= \lambda_0(t) e^{X\beta} \label{hazard} \\ \Lambda(t) &= \Lambda_0(t) e^{X\beta} \label{cumhaz}\\ S(t) &= \exp(-\Lambda(t)) \label{surv} \end{align} where $\lambda$, $\Lambda$ and $S$ are the hazard, cumulative hazard and survival functions, respectively. There is a single linear predictor which governs both the rate $\lambda$ (the arrow in figure \ref{sfig1}) and the prevalence value of the left hand box $S$. For multi-state models this simplicity no longer holds: proportional hazards does not lead to proportional prevalence curves. \subsection{Competing risks, Cox model} The Cox model approach starts by fitting separate models to each of the transitions. We will illustrate using the MGUS example. <>= mtemp <- mgus2 mtemp$age <- mtemp$age/10 #age in decades (easier coefficients) mtemp$etime <- etime mtemp$event <- event options(show.signif.stars = FALSE) # display intelligence cfit2 <- coxph(Surv(futime, death) ~ age + sex + mspike, data=mtemp) cfit2 @ The effect of age and sex on non-PCM mortality is profound, which is not a surprise given the median starting age of \Sexpr{median(mgus2$age)}. %$ Risk rises \Sexpr{round(exp(coef(cfit2)[1]),1)} fold per decade of age and the death rate for males is \Sexpr{round(exp(coef(cfit2)[2]),1)} times as great as that for females. The size of the serum monoclonal spike is of no consequence for this endpoint either statistically or clinically. <>= cfit1 <- coxph(Surv(ptime, pstat) ~ age + sex + mspike, mtemp) cfit1 quantile(mgus2$mspike, na.rm=TRUE) @ The mspike size has a major impact on progression, however; each 1 gram change increases risk by \Sexpr{round(exp(coef(cfit1)[3]) ,1)} fold. The interquartile range of \code{mspike} is 0.9 gram so this risk increase is clinically important. The effect of age on the progression rate is much less pronounced, with a coefficient only 1/4 that for mortality, while the effect of sex on progression is negligible. Notice that we did not do anything special to the data set or event codes for the Cox model. The focus of coxph is on the event rates, for which the correct denominator is the set of all subjects still at risk. This is exactly what is encoded by the (futime, death) and (ptime, pstat) pairs. The effect of sex on the \emph{lifetime} probability of PCM is not zero, however. Because of a longer lifetime, an average female with MGUS will spend more total years at risk for PCM than the average male, and so has a larger lifetime risk of PCM. The average rate of progression is about 1\% per year, as shown below, while the average post diagnosis lifetime is 18 months longer for females. <>= pfit1 <- pyears(Surv(ptime, pstat) ~ sex, mtemp, scale=12) round(100* pfit1$event/pfit1$pyears, 1) # PCM rate per year temp <- summary(mfit1, rmean="common") #print the mean survival time round(temp$table[,1:6], 1) @ Prevalence estimates from the multi-state model involve the matrix $A(t; x)$ of cumulative hazard estimates. The $i,j$ off diagonal element of $A(t;x)$ is the the cumulative hazard $\lambda_{ij}(t;x)$ for the $i \rightarrow j$ transition, obtained from the fitted Cox model for that transition. These predicted hazards are formed for a chosen set of covariates $x$, e.g. in the model above we could for instance choose predicted transitions for a 72 year old male with an mspike value of 1.1. Predicted curves from a Cox model are \emph{always} with respect to a particular hypothetical subject. The notion of a baseline hazard, i.e. the hazard for a subject with all covariates equal to zero, is sometimes of mathematical convenience but only rarely corresponds to any patient of interest. The diagonal elements of $A$ are filled in last and are chosen such that row sums of are 0. The obvious analog to the univariate survival curve in equation \eqref{surv} is the matrix exponential. \begin{equation*} P(t;x) = e^{A(t;x)} \label{matexp} \end{equation*} However, this computational approach is valid only if the $A$ matrix is separable, i.e., $A(t;x) = A(x) g(t)$, something that holds true if there are no time dependent covariates in the model and if all the transitions share the same baseline hazard: a very unusual case. The matrix exponential formulation is fundamental to multi-state models with constant hazard however, see for instance the vignette for the \code{msm} package. For the Cox model we use the Aalen-Johansen estimator --- the same approach used by \code{survfit} for non-parametric estimates. \begin{equation} P(t;x) = \prod_{s\le t} (I + dA(s;x)) \label{ajest} \end{equation} where the term $dA$ is the increment in $A$ at time $s$, and there is an increment at each event time. As with survival curves from an ordinary Cox model, any such curve is computed for a prespecified set of covariate values $x$ which must be chosen by the user. For illustration we will compute the probabilities of PCM from the model for males and females under 4 cases: age of 60 vs 80 and a serum mspike of 0.5 vs 1.5; these last are the approximately the quartiles of age and mspike. Each of surv1 and surv2 below will contain 8 curves, for the 8 combinations of sex, age and mspike. <>= tdata <- expand.grid(mspike=c(.5, 1.5), age=c(6,8), sex=c("F", "M")) surv1 <- survfit(cfit1, newdata=tdata) # time to progression curves surv2 <- survfit(cfit2, newdata=tdata) # time to death curves @ The individual survival curves are not actually of interest, since each is a Cox model analog of the `pcmbad' curve we criticised earlier. Instead, the cumulative hazard portion of the results are used to build an Aalen-Johansen estimate. The $A$ matrix is particularly easy in the competing risk case: all rows but the first will be 0, since only the $1\rightarrow 2$ and $1 \rightarrow 3$ transitions are possible. Elements of the resulting 3 by 3 matrix $P(t)$ are the probability of going from state $i$ to state $j$, since everyone starts in state 1 we are only interested in the first row of $P$. A computational nuisance is that the \code{surv1} and \code{surv2} curves do not necessarily jump at the same time. We use the summary function to select values on a common time scale. (The \code{summary.survfit} function was original written to provide printed values at specified times, but turns out to be an easy way to pluck off values.) <>= cifun <- function(surv1, surv2) { utime <- sort(unique(surv1$time, surv2$time)) jump1 <- diff(c(0, summary(surv1, times=utime, extend=TRUE)$cumhaz)) jump2 <- diff(c(0, summary(surv2, times=utime, extend=TRUE)$cumhaz)) dA <- diag(3) prev <- matrix(0., nrow= 1+length(utime), ncol=3) prev[1,1] <- 1 #initial prevalence at time 0: all are in the left box for (i in 1:length(utime)) { dA[1,2] <- jump1[i] #fill in the first row of dA(s) dA[1,3] <- jump2[i] dA[1,1] <- 1- (jump1[i] + jump2[i]) prev[i+1,] <- prev[i,] %*% dA } list(time=c(0, utime), P = prev) } # Get curves for the 8 cases, and save them in a matrix. # Since they all come from the same pair of Cox models, the time values # for all curves will be the same # The cifun function above is only designed to handle one of the 8 covariate # patterns at a time, but survival curves can be subscripted. temp <- cifun(surv1[1], surv2[1]) coxtime <- temp$time coxdeath <- coxpcm <- matrix(0., nrow=length(temp$time), ncol=8) coxdeath[,1] <- temp$P[,3] coxpcm[,1] <- temp$P[,2] for (i in 2:8){ temp <- cifun(surv1[i], surv2[i]) coxdeath[,i] <- temp$P[,3] coxpcm[,i] <- temp$P[,2] } # Print out a M/F results at 20 years indx <- match(20*12, coxtime) progmat <- matrix(coxpcm[indx,], nrow=4) dimnames(progmat) <- list(c("a=50/ms=0.5", "a=50/ms=1.5", "a=80/ms=0.5", "a=80/ms=1.5"), c("female", "male")) round(100*t(progmat), 1) #males and females at 20 years @ The above table shows that females are modeled to have a higher risk of 20 year progression, even though their hazard at any given moment is nearly identical to males. The difference at 20 years is on the order of our ``back of the napkin'' person-years estimate of 1\% progression per year * 1.5 more years of life for the females, but the progression fraction varies substantially by group. Eighty year olds have a lower cumulative rate of PCM than 50 year olds due to a higher death rate, even though the hazard function for PCM rises with age. A plot of the calculated progression curves is shown below. The left hand panel has predicted curves for those with a serum mspike of 0.5 and the right for mspike=1.5, and in all cases females are predicted to have a higher level of observed progression than males. Although the Cox model hazards are assumed to be proportional, the prevalence curves are not, however. For those diagnosed at an older age the prevalence curves flatten out after 10 years, simply because so few living subjects remain who are available to have a PCM event. <>= par(mfrow=c(1,2)) matplot(coxtime/12, coxpcm[,c(1,3,5,7)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, ylim=range(coxpcm), xlab="Years", ylab="Progression to PCM") legend(1, .23, c("Female: 60", "Male: 60", "Female: 80", "Male: 80"), lty=c(1,1,2,2), col=c(1,2,1,2), lwd=2, bty='n') matplot(coxtime/12, coxpcm[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="Progression to PCM") @ In the competing risks case the prevalence function has an alternate form known as the \emph{cumulative incidence} function \begin{equation} CI_k(t) = \int_0^t \lambda_k(u) S(u-) du \label{cuminc} \end{equation} where $\lambda_k$ is the incidence function for outcome $k$ and $S$ is the overall survival curve for ``time to any endpoint''. Proving that $P_{1k}$ as computed by Aalen-Johansen estimate is equivalent to $CI(t)$ is straightforward. (The label ``cumulative incidence'' is one of the more unfortunate ones in the survival lexicon, since we normally use `incidence' and `hazard' as interchangeable synonyms but the CI is \emph{not} a cumulative hazard.) For the general multi state case it is simplest to use the \code{mstate} package; it was designed for this task and will also compute appropriate confidence intervals. The latter are complex since they must account for the uncertainty in the underlying Cox models. \subsection{Fine-Gray model} \ifdefined\CMPRSK For the competing risk case the Fine-Gray model provides an alternate way of looking at the data. As we saw above, the impact of a particular covariate on the final prevalence values $P$ can be complex, even if the models for the hazards are relatively simple. Start with the functions $F_k(t) = P_{1k}(t)$, which can be thought of as the distribution function for the improper random variable $T^*= I(\mbox{endpoint}=k)T + I(\mbox{endpoint}\ne k)\infty$. Fine and Gray refer to $F_k$ as a subdistribution function. In an analog to the survival probability in the two state model define \begin{equation} \gamma_k(t) = - d \log[1-F_k(t)]/dt \label{FG}I \end{equation} and assume that $\gamma_k(t;x) = \gamma_{k0}(t) \exp(X\beta)$. In a 2 state model $\gamma$ is the usual hazard function. In the same way that our multivariate Cox model \code{cfit2} made the simplifying assumption that the impact of male sex is to increase the hazard for death by a factor of \Sexpr{round(exp(coef(cfit2)['sexM']), 2)} independent of the subject's age or serum mspike value, the Fine-Gray model assumes that each covariate's effect on $\log(1-F)$ is a constant, independent of other variables. Both model's assumptions are wonderfully simplifying with respect to understanding a covariate --- assuming of course that either assumption is correct. (In a multi-state model at least one of the two must be false.) Let us look at the effect of sex on PCM using the Fine-Gray model, which can be computed using the \code{cmprsk} package. It does not use model formulas so variables need to be vectors or matrices. <>= if (cmplib) { temp <- mtemp temp$fstat <- as.numeric(event) # 1=censor, 2=pcm, 3=death temp$msex <- with(temp, 1* (sex=='M')) fgfit1 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE)) fgfit2 <- with(temp, crr(etime, fstat, cov1=cbind(age, msex, mspike), failcode=3, cencode=1, variance=TRUE)) cmat <- rbind("FineGray: PCM" = fgfit1$coef, "Cox: PCM" = coef(cfit1), "FineGray: death" = fgfit2$coef, "Cox: death" = coef(cfit2)) round(cmat,2) } @ The program has determined that female sex increases the PCM outcome by exp(.21) = 1.24 fold, for all values of age and mspike. The Cox model shows no effect of sex on the instantaneous hazard, but as shown in the last section Cox models do predict higher female prevalence. We had also seen that older subjects are less likely to experience PCM due to the competing risk of death; this is reflected in the FG model as a negative coefficient for age. The primary strength of the Fine-Gray model with respect to the Cox model approach is that if ``lifetime risk'' is a primary question then the model has given us a simple and digestible answer to that question: females have a 1.2 fold higher risk. A primary problem of the model is that we can't go backwards: there is not a simple analog to the Aalen-Johansen estimator to carry one from $F$ back to $\Lambda$. If one fits a set of Cox models to the arrows (hazards) then the boxes (prevalence) of figure \ref{sfig1} can be examined post fit. With the Fine-Gray approach we have information only on the boxes. To compare the two fits we can look at what the female/male ratios for each of our four chosen age/mspike combinations, when $P$ is computed from the Cox models. <>= cox.f <- log(1- progmat) #log(1-P) round(cox.f[,1] / cox.f[,2], 2) @ The Cox models, which assume proportional hazards, show a larger subdistribution hazard for those who are older, those with higher mspike values, and at longer follow-up times. The overall average, however, is similar to the single value that results from a Fine-Gray model. The predicted curves are however nuch different from those shown before for a Cox model; the Fine-Gray curves are displayed below with predictions for mspike=0.5 on the left and 1.5 on the right. <>= if (cmplib) { par(mfrow=c(1,2)) fdata <- model.matrix(~age + sex + mspike, data=tdata)[,-1] #remove intercept fpred <- predict(fgfit1, cov1=fdata) matplot(fpred[,1]/12, fpred[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), ylim=range(fpred[,-1]), type='l', lwd=2, xlab="Years", ylab="FG predicted") legend(0, .22, c("Female, 60", "Male, 60","Female: 80", "Male, 80"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') matplot(fpred[,1]/12, fpred[,c(3,5,7,9)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="FG predicted") } @ This tells a very different story than the Cox model prevalence curves. Which is correct? Individual non-parametric prevalence curves are not as helpful as one would hope: there are simply too few progression events when separated into 8 groups. <>= if (cmplib) fgfit3 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE, cov2=msex, tf = function(x) log(x))) @ A deeper analysis is called for, but will have to be left for another day. \else This section requires the cmprsk library, so was not created. \fi \section{Conclusions} When working with acute disease such as advanced cancer or end-stage liver disease there is often a single dominating endpoint. Ordinary single event Kaplan-Meier curves and Cox models are then efficient and sufficient tools for much of the analysis. Such data was the primary use case for survival analysis earlier in the author's career. Data with multiple important endpoints is now common, and multi-state methods are an important addition to the statistical toolbox. As shown above, they are now readily available and easy to use. It is sometimes assumed that the presence of competing risks \emph{requires} the use of a Fine-Gray model (I have seen it in referee reports), but this is not correct. The model may often be useful, but is one available option among many. Grasping the big picture for a multi-state data set is always a challenge and we should make use of as many tools as possible. We are often minded of the story of a centerian on his 100th birthday proclaiming that he was looking forward to many more years because ``I read the obituaries every day, and you almost never see someone over 100 there''. It is not always easy to reason correctly from cumulative deaths back to hazard rates. An advantage of the Cox model is that it has better diagnostic tools available, e.g., evaluation of the proportional hazards assumption via \code{cox.zph} or the martingale residuals, which can help to further refine our understanding. It is also easier to link hazard rates to a biologic rationale (perhaps incorrectly) which can help in explaining a data set. \end{document} survival/inst/doc/compete.pdf0000644000175100001440000100037512545056257016035 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 4 0 obj << /Length 2920 /Filter /FlateDecode >> stream xÚÅZI“Û¸¾ûW¨r’*&‡X€©šƒ35®²“IåЗd2Zb·d·ÄŽ(µÝóëó6¤Ä’ÊA‰åááÃÛ¡?Þ¼ùî½iÊ—ZÕvqs»P®)UåNÕ¥iôâf³øyùÓy¥–÷ð9í øð©=u«ÂTÕr߯ í–hìpØÀÍí^6ü¼†Ç~ÿãjsÚaçwwHò Ñýåæãwï0¤Ê¦®52BÙ˜E¡CÅìܬ¬u„øy2@÷'üqË›íJúp‰®…¯3“µnL¶ê°(”)d›0¸}‚혺Zþp¿;ìÖsU¥³À)·<ñcÄGxQoùWCSU a¡li¬#:…ò®ôÎ/ «Ë¦¥V…R´àüþß»‚Ç/¹0¾¦C8±>+§—-q½v¹Æ.£—`MÙ8íˆ`^•¡’]ÿÐ#RÃŽŽ‹°+¬­h©-/—^ÖNpýݾîó(’ÏtîéD©yGÇAÿ¨”ų?ó,U® ïÜòGĽ[îøD’ÕcË·LIˆ¶øË㤓ÀÐV:¯Ìá)vÉ:Q$ñ¹Ã…B\¨%a¢…¾Fê€BÅÏP=0?õIΜԠ&€J›ÐŒLõØŸ5ä0¤á–:û~&*·4ƒÞ/1Â6Ù8®4Ý=ný^úYH¢DÔõ²EÒ€ÕuÃÏ÷Ú!¿–V6inÙEf’UcQȆŽÀ§#Àå‡3ÉP§ô×]˦†9Ách""(ƒ¸m 8>xI’Z’ ÙC–z|[³¸Òóè Qþ¹>Y’º'œzFVE¦ô·Üg˜£“¸Šã>ÒoïyÀH„9>«³Þ)1 'Ùc¿ŸƒëJŒ¡½lYÃñT¦MbGÖ¥ö~B§;Bd åJ.u*ÊØ5ªÝ'EŸÍ‡ ™e‹Jž"½á^·BÇüuY%¶p=ï@t0`üH¹pÄéÿ<ËYY¾i¶ÅQ¨&vjig^ß‘!®ŠÅ6)cw`B]4â#T…í¯ÛdÄÈÏ’•ÚÖSMÖl-7°ÑêÆNH¹ƒÎçm—òDÝä\e4Á²FÀØ–±¥káL´ Ž` j8âdÉ¡çÝG¦Ì»%-¦Ývܼúr²ÀÍws5¾Õ’][ñ£h–­˜åè"+‡9ƒÓ¢ùx ýìØR¶Ö)><µN'Úø=ÚJ|žF-0ss$cМ(È€uO#†Lƒ zÚI³yZð( AMvÚÎX˜›D5: ëlcö²|5o“ë*Z6R;V\A µÙ†¿Ids”82\—tÞú¸y¶d0‹¢ |È¬Š§„69|lÑJÁ ‰­ k…$®=PSE±£P6”ci é’76Ë~Z€ŽØ­ sºŽÂmU‚’ĉ8ð|ˆ§\‰ðŸ½Ó¾£íˆ.·@¦ÏTÀS­¦ð“²=kØÅ¤8Á˜À3 êfêS0it>ŽAãí¼>ï¤þÜ "J QM©+!yw™/­×±ŸꚈ)ƒ¯ã˜ÜMY7~Dò-ë•Ä¿ ÞxÈfŠ¿q )ü¾êE`̶ý5Šâ†ÉÞŠta=%¯<€|×`ØÉ4÷C7³)mM鵊lï>_ïKéÒ7 ¼î^b¬}N‰|NŸ¼šxÏc‡¤Èm Ñ6Q¤êÒ© ‘z …†_þ˜#¡QÅ’!á!·2¢<ÕF¶qN|jÐŤR»kŒ ݶM– Î^¥k\Y[§Í@oKçS9C¡hª2(KU ÌD¾Ul5Þ„,à…*mS?Ô,‡_Œ½  öUSUÐ>,¿Rch}©jXfH€ð÷îq<–jJëÌdQ?TØÒ꣪RyáŽù˜sð]y©Slþ2³$p¤šÿ#`¯Ðøþ8 “õnÊóÜRæÇ2 _Pd5í²BMµ³Y'¢“e§11A¡4&-ÆaIrèï†i0ëŸÏêîÏ%„KeGù©’Ö¯¹ e0&áÀwÏÜɆ‰À¦rЇëp™BÉ“Š/$~O<Ÿ‹šðùÂí1’­ÌÎ6‚ ç|Ls9}ÁÒ`s‘¿ Þµw}ʘC˜1ô®R„3£V Ù‘ÌøXXÐØ$QmÌ Cƒú˜®v^&XðxßxÁ 8r¢¾•Ç”#3}š°‰y0e¨ë׸¯ç¸§ Œ’õûÌNÏ}¿F¹ïñ(Æ{]ÞÕ8­Ô.+W_ÄV/gd1Fz®z%éý$ØÀ9»QI|‚‚S`:Uq%Öñ¯sVχÿ~°ô¼‡Ð 6X'Eñ¿Í˜¾”ÖâRÀ3ó83ÑŽËËj;CÊ•ZûY¯*.ÛË%ÆÏs6 ÖI}˜ƒÐd|~?ë¥ÐåE›øn6,sµÿ Ç`s1`ŒOá—U¡ -Ñ1©R#Ãn=]”5x© ÅÎîá˜KI¨kÑXR•ÇÖËǘ¹1ý>Lr>"Ëu¹ÝÈÍH.x±O‹Æ6[å-¼šª¸~QháÐ\ŠòZÙÛ×X¢IÉs™Ó7/V4 ‹BŠ¥ñR-gó–-Çóþ©ve•=ï¬h8Š„¦*b»\…])žÝÅ€¬ðæèªf³œŸf„×~Üjý² U:(I`òéªÎMM:¾Ç*LÇã¸$ÒïÆVr¬[¾ÇRå‘Ny×x\8RVÊ„§MDc'‚ŒÿÞ¶±ÜëžN 4ÃS^8.³û£âd§É"C[•õeŽ}d¯(JÕI™ùÔÏV~¹,§—EåNX~¡ÄûìÑB¼¨Í )­ÑãhËwCcWŒï³ÞׂDÛ$¯ÿ¹Ãù7Sä§Ë©©”ø8œÓMçà9æ‹í¤Š"¢ßçÊ{Øc4ÝL‹œ ¤U9¼M;ååÅΜ¨imó~±0W›å_ßéÀõ\Q”DÖ}gJm/D/Ý&]óŒqr]L×ÇRåqÉ`Ä1íŒ1«NxE°F)0í#]ÛýžjÉÓÐ\†ÞO™á‰¦ÐÀ€î5)¥QkÉ”W##3¡Blêlc oz {g =a`+b¹r@;ø Ô'…O¥R¼i!„ìR°‰obgpÂJss)c0&ש–K¢«°Ž4\Ñø$C©}è©*8ÊÞ0Š÷öç(0þ/ )Ù™NÚÇ é¯ñŽóŽG>®jÇÁ9j#" ^BžÉ&·èÇU­iŸ”Ètrsºææëè§p7W¶†3ûI¥Ö¹fùáv’åÉe g^»œ†dËðHW;$Gýqt‰›’Õ.§nÑýËæ»XÌ»^à!—¡Õ5ý™#6‰3B6;ôDtE_7F²GÏ)bH†€…â Ž‡Ñ˜xþI0¾å³éî ²ãT|ÎÍNì¡29^v>k`L¶#z|Á,w®§d°Ù¡éª†ø 2(cË êø§èzóãÍ›`e£ä endstream endobj 3 0 obj << /Type /Page /Contents 4 0 R /Resources 2 0 R /MediaBox [0 0 612 792] /Parent 18 0 R >> endobj 2 0 obj << /Font << /F39 5 0 R /F19 6 0 R /F46 7 0 R /F48 8 0 R /F8 9 0 R /F55 10 0 R /F11 11 0 R /F10 12 0 R /F1 13 0 R /F7 14 0 R /F13 15 0 R /F14 16 0 R /F57 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 21 0 obj << /Length 363 /Filter /FlateDecode >> stream xÚ}QMoƒ0 ½ó+rL¤’’вã´Uên“¸M;°–4RºÒýüÙNú±&0Ï~Ïö‹b<йB¶dºÐÒÅ–>ùN¤³:ŸQÁMH)i˜.¼bOCò Ï9•FÉôFó±J¦ó’9”³¬Z_ÚšLZЯVìÏ…âœÏÃÞHó"ãêA¤F•¼j#<&¼Ð?ÐÃÙGj:bTï¯|?„¯ÁÊŽIcùÂM¨BB¨¹¥ƒÄ†ÚRéÍšºÑGrޱ¿ßö×ÖÒ*ǘ= cx¬pâÖñ}{%ÒØÞòfÒ;2¦º> endobj 1 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-sfig1.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 22 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 23 0 R>> /ExtGState << >>/ColorSpace << /sRGB 24 0 R >>>> /Length 693 /Filter /FlateDecode >> stream xœ•UËnÛ@ ¼ë+xŒ/ìrß{Ì£¯-Ú@AAâ¢5ä©‹ö÷KjEi­ir°äÑ·\Š"¸‚ÿÌOS€?Ýõ ¸ï.ù·ëHð©ó)‚5„daßå$÷ö %jf8’PÉ뎫æY’Cãg†bPþÏŠ%`"°ÐØYB_÷À!ù¯ ü„0 y̘׋ä+’g›±bw‡TìpÇ}g¹ú›oP‹4_*‹« löp}rÚÿø½s²]ÝÀæ²{»Â<¯CŽÓ¦Fèäb{{¿‚ÍNöÛ±4'ãøOiŽF±ÖNùK<×Òñyå Ä 6‹"ŸnöŠû ç(·~â/±òY‘ œá'‘ǹˆ¶¥‚©c;^4³ó˜ÁØcª1’o]ÎyWWsÞ#Ö<•¿ÄMÞ£Km¢ÑCÓDNú®i¢c^ŸšH£x'Ež]V<»Lƒm\-±òg—–¾q8u9ÔÊòV—#Ckm}imõç{Ð&~ÃǤ¡Oµÿ¿Ù…‚%Å,·¡éÕ»›Ðöõ»ùeäñ6ìvº»N-.u&B*N3†¤?â4âPÉkYRs>•ÎR”èö 'íJ^@%Ë|ÌðËåˆ)zôb¥Ë0 IN4:ô$ÎÖ†±È— ñ6zŸÌM¹(æ¾HÍø}‚•?,UP —ÜŽ«2n,î1C3 SЕ*©Œ(OXs ‘Ð…*éY2 ’% çÕx9`vPòä±xy‡øËÀ²¬WaÊè¨ê#³&{QÝwÇë ³p]}Ù; ;tµÉ?loû_ß_Þª2ü(ËgCû|{ûºýFJAü¢ðwG>öýÃöpP‰«î/h Ÿ; endstream endobj 22 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 23 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 25 0 R >> endobj 24 0 obj [/ICCBased 26 0 R] endobj 25 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 26 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 19 0 obj << /Font << /F8 9 0 R >> /XObject << /Im1 1 0 R >> /ProcSet [ /PDF /Text ] >> endobj 30 0 obj << /Length 2435 /Filter /FlateDecode >> stream xÚÅY[oÛÈ~ϯ0Œ¡ÃáÛ:@»Xw·»Ä‹¢Øô)Y‰Dª¤G}èoï¹Î eÊÍ[è¹™9×ÿrÿêÍ]yU…UçW÷«+câ°0ñUžaœdW÷ÍÕïÁÏÝÍ"M‹àðxc‚ûe0Bwßn¶­ÜÄeðtA/TX­nÛ`¹Á™­n2w0ž7´ðçËM–õ»ÝÒ­ã¥wÏvÖ[áé ²Ñ^à ÆçòqÿÔÂÖ/ï¸EAª§gRýóþoW “†UšBkÂ*uµ¸oÜììÝßÝ,’¢ê®N.Jj±_‘5Ž•oœ|ÿ1Ê"fÇ f ŸpËt†›…ìBÙH*®ž0áÍ"ËLpwS¦A?0ñHöƒo¿e=Ðä%q¿íuáÛŽ·ñĶáe\è]—õŠŒ¦*Ä‹›–.X$°¼Â6a+±˜`Ík¾Ì´¥%E±è6š·/o:„xG,Q6’.këôôj­•ÁÑd¥ƒUœZÆÔØ8<¤Å+›Ñíè¹û@AÑò€Œ±rq1ÚG±‡J„`Éåíqdm«~¸@µùY^¢ÙÚ¢@°¸Ìõl"³vÚá8åqË}õ C×ßÓ¬A¦ÞÜ™ÔÇ•¢£8%?FQ$DÆ#*Ã2Ê•æ½DWyXI‰y˜T9¦•Èú™©|+Â<¯<š[fkFí¬]4°´NâuÇ#ÚÑ‹¬H¬ l_ÖðÝŒD&‹Ã(.T¦n³0sZAOÄ‹ëÁ~Œˆ¾ÉM€ŸÃO0Ffj ò rýhÂI”䬥¿£ÈŽóX^ÄðÛ‰â¡/‰»¬Qæ€ÈVÈŦ(‚öÅ£vo/$5éJŠÅ~q—ˆ®Hò¤1¾³×Pëå96B5¶jœ÷ŒKY­9ú#ôí?b`ãUxiƒë=®÷ë_ý©?’›ï´qœ yÊPÄ>‚:[¶<ý/ñº*§³3…h”ú&·œK,KHÑÀ††Ó-ª¹4€WH]€´K•Å:ÒÈ ¤HDè8O‚ ly´Á=Ìœ¾’ûÿ­Y¢GÈ, ˆà¾‰iiu<£ÌO¸tô5¾HM%÷‚øŒ­ê#š|€FS*Ó¿X' ñzóe*NGÁÒey¤–»ý†ã»ås Y8GÝT—VªÜ„¥Fùé^!Åœ#“x#í;Zu£¥^_r>ºTmÔ¹\®%QŸª]Ú£YõXìkµ‡}À¸ÄDÓœˆóãp°e£$jذ7ÐRÖ›°O¥T/ žc³y¤R²À„u„I]Ü zÒfrxÐU†íRp[Ùq•Fe±›X9Š‚çKD”Ô¸$Žý¡?ú%³TéÕ£7 ¼úì;’ÉX!YÖUQÒïÛtÜ’¡¸ˆƒ…)¹†ýnÊCy¾íSo$À¸Ñj©`P†Ká>óraxGa‘—>lf”dš&3µ|‹©ßøM+öLóÀKpˆ?Öê­gÎq'a•-¡afE›&>ì'ijC˜ u£®tçU ‹Ûyµ$˜¨ÑeRªtt'îkV©×hÈz7Ja K’ ±;HiB`4“U7úZ£¢°ÑÒ?‰Ÿ“d* ‰r/Û)Œ-â¡3ѳÛÙƒ=ÒKqÞŸ{¢ýÔj…U›è+*â¨m¿’”!<ꣷ[ 96±÷öˆ—„‰’”g³Ïîn©O·¢ß]ˆP]÷-WÀ»+!™‡Ub‹åE$g <6~^hÆU¹êºQQZÉ]Êy‡®qÐ`ë›\~ÆYX–ùóð!m "$c5²¨>ëqWóâ²ÝÒ ÀBÉu§®å‰©Ä›OxçÊŒqÂÞ7T•ÀK?ö/š°% ôMR¦Ü£‡þ>‰) SÍtÝ]¯OWœÒ’FNÞQŠç2w¹e•ý]à QÖõŽ^[{aMž‰”x CL§RûÝÙ´Ø0Ùhõ³‘LêA+“ˆ®ÞýþüöÁ/‚iùÃÅ÷}œä6•!%ûãÚ^ôî¦j6)¦F†±}±$…­Ä ‹JŠÝ5c8§{ý Ô âŒ<·þ£¦½WÄÈ3½-m;„¯ü‚cŠM’o.Ôn4•ûÃài$çŸ`¯}%•ÆáÖ(¼Q]ƒ~”Š$þè33ýªDe!¾`¯^Ë8ÄŒFÈk²s?*ú"†JûéÈôÞo0ôREP,"nO4OÊõ%t%º1Æ0·ÄÀ[à?NàhŽ×†^7t.ý É]·bËôv[2Çeã­TlBÂõB•uQjfÔ!Üh•¤ùλkŠ}&øƒ°´0ýŽOè„…­,ò!k |‘#Ý ‰úÓ­h®:ûQ#­RÅÔׯç´…iV*‰þÕãáAïÏË×¾B¦,@ñ¬^ž¼÷ÖãËz‘ûr÷^>.Ù\±ú\s£ ¿N5Ô u7cÄ$³4ŽÈ¿%s|Gï@'ÉJâ Uˆo„b“XÆ^Ô_jÂ"—ŒA1†õÒñ¶VØÖ7„<ÉüP°µg@\ŠÄO+òMüŠ5 µ³Ë¿Ùe åÆ7©/+ºÃ Š„:Žª,HE²DÆF‚¤Ä|ŠïtGÞ8ž¸Æ w¾ã; áZïÀ~êYò}ðÆeÚJúíå@µ/€çH?õbkí¥Po¿)Ñx0<êï¶ÔçÌ1w^Žû,6;wX6£ø¯‚?º#¶’!åÓ'HéiüRŠø‡ƒóÀ½ÈϹÑÀÆCóÎ#Ñ4zíØ;]¸ð> endobj 28 0 obj << /Font << /F8 9 0 R /F14 16 0 R /F11 11 0 R /F10 12 0 R /F55 10 0 R /F59 31 0 R >> /ProcSet [ /PDF /Text ] >> endobj 34 0 obj << /Length 1392 /Filter /FlateDecode >> stream xÚÍWKãD¾çWXs²ÑØãn¿Ú+iY D–ƒÇq&Äqv~;õê¸c{f÷€ŠbwWUWW}õè¶òbø)¯,¢"7^RÆQœ¦^½_ü±ˆÊ\') 8Cb)Y&„»¯öÚû¢[ü?Ë Eeèèü|¹¸»7^‰êro¹ö”ÒQ¡´—&ŠMé-WÞ/þ› Ôeé÷M üºká¹bʆ …¼S×H=âØø'o‹Y Œß?a–ú¿Á°©Q¢gá­¼ën€QîãÂÓ¸ç#+8niòÇ÷m&qæŸ70AF€1–¬ÙåÉVÙµ+»aIÖu<ìÞ£à`cC«Û[ô.Vé#K+øuùµª4*ÅP©¨Ì2FªzàMЊ'|àÎQæiìß&%\‹ {Bó½ÀˆêwL:ózW¼/»v“¿xÂàhVØTOCÒ Â»n5èÒQ <1©yBMè#’%±«9žõ2“@’ž±·˜!ÕÎ1øL[‚ñ;ë£fEµ G[ ¢µ tìl/ïÚaeÇ¢S¸{Õ“%̨eJX”`#QË#îTµµØ ¼wqÃì{ø¿ý6 ¹bNÕË»Eoîî³Â©#5–ƒ‹6-hJÇ×¥4çTøhX“[yå±²Š0ÝuBGè>„Öµv†Žò4É@E®®Á¾DÊÊZÊ0MŒß"SáÈ$ü‰VK€‹¬ôL‚áB~}.!Vø*AQ–slÑêêag+ËÜΈ)ÀPoH?hl¤D·²o;tˆ²3Uùßàç½gÙ?]Ø”ÄÖ7œn‹¸ºÙqÅ•\ • —H*Þb åªô¿ƒñ¸Å¨4™Ï¤Ò MzÆ$¿d‚­b-UŒ&áœ+}˜3mr—QpÄà |—%Qn”í ·ør¹øªN½ endstream endobj 33 0 obj << /Type /Page /Contents 34 0 R /Resources 32 0 R /MediaBox [0 0 612 792] /Parent 18 0 R >> endobj 27 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mgus1.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 35 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 36 0 R>> /ExtGState << >>/ColorSpace << /sRGB 37 0 R >>>> /Length 3636 /Filter /FlateDecode >> stream xœµ›I&ņï߯¨ãÌ"÷åh±„dKöŒd!Äa } fÚ`ùß;ÖÚ"h¦è釨¬ÊÊ%–7«ãòé—ï–o_~\ÂÿH±Ñ?cà?ï–.oo½¾}øø¿|´|ùxëøZýùøåÛÛ‡Ÿ$¸Ý믗¸ßŠ~0ÇØ×—¼¶°¼þ~yñ§o^.¯¿»ýù5Ý[Z|O÷ð³ä5jþÉ»‡zxûåõ&ðÒÃ_9=|õñßàÿöºüçöùKX¾ºÅåSøï»?寷ÖÖ^–2×–ï—8êššà}yu1 eµŽ¶–¹Y…6k mMu¿·ànO4â›]p·—Š]ßì‚»½ÕµÆÝ.¸Û/³¿ÛfÿÝ“Û"ÎKê:¹9xsë·aùж<£m amóи>§q k)‡Æí9KXãñÉý9ë\g=4ÏiÜçÚáñÜó×¾ÎÓÌ”J_Ó i¾˜óN»uÂÒš›•i·Æ û·mfÁƒ½ÇµæÝθÛS˜kÙ.x°sg7ûÞ÷÷ð¹ ¨u™ÆÅÓ'Ûΰ&n›žß8¦±Æ§–ðÓ­aÏÎþÄ2|²u yõ‰¥ôtë<ÖY¨5l¨ƒC•%$s[Éw§\‘$,ü.[áéSV`Æ ‘qµÁ_ÝêXGñM-á¯SÌèÍØFnÙmÖÉSø¦²¶S+X•]Lc­§VumƒM#­µü‚‰üÿ±‹} …m“œÎÁ^hd±•µ„“­¡#gÛXs=Ù`ÛbHk>7¸±ÄHÑæ`„ÍUeÀ"ÌE:µ„YzÛXÖxj ûyn-i¹ŒU´%t.œZÂkÏ F oÇáþëÆD¾ñhMÐ%™*ô8óܶÀ£²Z;nƒ“µ®C;U"9ˆƒ^6ë9¨ž& ¦Jï\ÈéGæ´Ë|ÓÖ8·mW‡XÉU­³á?l…Ø ³qjЈ•Bùi4Ê:µÏ Vê¹mœ¸ØÚ3.‚Ó87œ±ö5ŸÚv6ޏæSSØ«SÇ‘Ó ÇÎâÔúÛ´å„ü­ù›ÝMjõWüÍïÎòN€8aôÃ!ͻطT)ÁÞŽ»àfÏ¡S9ÅÝq{ì‚»½ \¤›]p·7Zˆ›]p·ÃFk»àf?½ðþþ\*˜À?—tHOž‘Ú¤ûñض¾Ûâ9›‹Ïxp•OŸód½cc?yø…Æ |ì©ñsž ;1”ߘw—P1êoO–U¯™¢`Ñæq—lÔCfy°Gœ“¾_ |¼¼.„›ý æãP†Öq¸‚ùpEÊ S¾í áãÜóýŠýMÞ#IJ•ŠOø ±VôúœKšCl¯Iš§ç7£….íËoh½I^¼¶ç·O©jé üÚC@ì’cîÃwö¶àœÎﮘ¥E÷m6/¬—K aõ‚ŸýQ2Dï47%ãóŸ½ÄrpyñðæÝãò¯ÿ½|õí›oÞþðøíãË/–ן¾ïHA^D"ÇD¢;¿úéÝËåè/~þöç—0–Ë‹7÷ížï2%\’ÿÿµp©ˆ¸ÖmmŒ³«<ƒÃp}'nä+ RfH¸-'Ì-™Ñi†ÔŽï×±¡å€^–¹ËàF†ìÊ2$KƒÒðìpÓñXxø~³UË s.·(ܳÃÓ3`ÈYÃp8­cct6±ÞbîÝåBï@L÷˜K· iLOÂ[î8LÈ໻à ÓZfȇ-WàÑβ 'NÑe*ˆks8éûfŒq–á5+qÁÜØr ¡ƒwè•óÔñ(”3:_Ü2-SæVnØ döˆ†‹ŽGÃ2ÓrÆ ™*YËIö;VÓeŽaÀãÈó9°Ö³u¼ey†ƒŽ×ÀªÌãÀ<170 E§Œ'TTÃán‚yd‡;ÖØÀ#j¸á2c®Ùá"ã ï1û+ãàø,'¬z˜kw™C&ðLG޵[†²²ea¨æ,“N‚œ±p´Lesk.Ëx”5D‡ŽÌ¹9 e¾^ߣËì0c®ÃeM¸—y½q½g9cÆ\ªÃIÖË o9¢ŽÜ©00L3ƒã1&f‡Ìà¸-wtȃdkÃ’jâZ&‚yL‡‹ìw,ÿ³ËaÃ-'z‰ñ4ÀpÄn£D;²_°Þ͆㜸Œ˜çpxÈófÄ ÔrÃ0Å ‰‹å*þf&Ü8–‹øàÜÎ2ž“T=y¿Ã/¡;¬ûy’æc9êûd\Ø–ƒìGèXŒ.×$\šå1%†g¹Sá\qãzÌ븇ºæQ.˜¦ 7.-ç)<}æ|l’ã¶œ%ßÇf9ÒA1䣖†äù¨Ça ·b˜ÌÇÀÀî1û3Xع8Üt¾(°Y®¸Ìï· òQË…¯GÎÃezrÏ‹?D•& ‡#ÇRq²ÃýIÇù¦aHƒg†…e¹ó|#äpã|/‡Œ/f¹p=€ ù¤åÌþ:ƒcƒÀåq µ¹šðŒSŒLÚ«åÈþ¹E‡ïGd‘˜O\§ö·a`ð¸ðõ]0Ï<0Í`ÎÕa:˜aîÁáÊþ:‡c¹p¾ˆ -g¯‰Çrb‰\ŠÃ‘ó=ä>˜Cà…|Ñp‘zÆcÞO‘ND-wη‘Çp™ç+Rák¹af†·\±ÌcžÉá‚Û 9c>g9á01ÏîpÐë :fÃPÏL~^Å|ÍòعG‡׫¿¦0€Ü1ó8gáá±îg<ÕªgΑ!Ÿò8ñý&&&––ÌÃãÈù`NõÙŸ& Ü–¦iÌÓa)Ç–+¢{Ìë‘•cÃÜ»r™Q†X9öXÞ‡”cËM”?VŽ-WÎ2D9¶\¸råØrÖù&åØcVY9¶¬ñ˜•c«*Ã-[†mÎñP”cÃCæC”c‡ÛQ96Üe=ˆrlXÇK”cÃEö³(dž3gÖª;,Ê&+džõdB”cÃAâ¹(ÇóÉ‘(ÇWnS•WVŽ Ów°÷M9v8daTŽ oãÉʱá†nà¾)LJ*\«Ãe¿•eÃYòaQ––“ V– ‹2 Ê²aúàï¾)ËËÉ+ËW®SòYQ– w®ÜUY6ÜTYgeÙpá:B•e‡YYeÙpÞ•g¬l «2%ʲáÈùŒ*ËU¦±’3¬ÊŸ(ÏW.’ïªòlx¨rÌʳÃÒߌù®å.Ê•(Ó†›('¢L®ªÜ²2m¸ˆr Ê´aU^D™vXæ‹•iÃIOX™6¼Í+Ó†e=£2ƒe¨7ÆQY6d¸©rÊÊ´áªÊz¦|Èá¨Êtè}+׆³¬Q¦¦“]Už 'ÙO¢<;¸?¬<Žr²¨Êó•%“¿<ð¸p:VÃàÆød%Qbë1ë“à0žÒ_p,a8¬'[™„)Ë’©rí0ÇkQ® Wþ²¹6‡åã✌e©wUù6¬'x‚ÇQ´_`Œ'†ƒ*ëT˜]yNI— ýÙ‡‹BUõ»^L¢ùEèÉñÀF¼¢|PŠ‚7Ìç“'¢Zš 9ì„-xÁ¡ìÉ |H‡B:l¿+VIþXf¿bVžFéŠQºÑh]Ë:š6È¿aÙ]±³(Q"©”¿ƒ¸ýÒ'ט ãGK)ðß- ëÞ™þ`ûÉvaœÛ1ÿ!½ƒ^&±*V ƒþõË?@øþÍýÁýœýé;aÚÈßýÓ ôKoÿJ›¢¸ endstream endobj 35 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 36 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 38 0 R >> endobj 37 0 obj [/ICCBased 39 0 R] endobj 38 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 39 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 32 0 obj << /Font << /F8 9 0 R /F57 17 0 R /F55 10 0 R >> /XObject << /Im2 27 0 R >> /ProcSet [ /PDF /Text ] >> endobj 43 0 obj << /Length 1631 /Filter /FlateDecode >> stream xÚ•XYoÛF~÷¯ òR ˆhî.Ï-¤ Ð6)ÚF}(š>Ðe©¦D—¤|üû廤$'Œ‡Ë™Ù9¾™Yùõââò]šeTf: ë@©<ÊÊ$ÈÒ<Ò& «àïðv¦Â%¬¬—³yªMØYÃz€õ=¬³¹Jsæ@§°Ìl®ã2%:•Á{¦ÊðW _ÍæF%æ ýÏâç æJEe*g¯ä  ÖkóYpûK&”34“itÌßÈŒ‚x­&Ò±˜ñëÆH4°Z±æSœÆ˜5¬­lkoäR¸1ré¥qù§=Ey%è}Âz;õâÖJXµ³->5«‘ó®„ù¬¿FXèÈt—”V<xk%–!ç5¬ýˆe+ÏÞ²Ç'üMÂCïŽb´—¸ÕÂKÁFgæ“HgüXáë‘’Õ4™±7+"dN $Œ/Dî¸2¾ Ö’•±Á#?'þ±q/øŠÓŽÅαOàšdQYÁõµt®ž]K_W§P¦OWýà tØrÔaµN¢$/óäÓ7Â2nÂi¤uiögt ƒz¶‹Ì‹·‹ ”Š”y”gE á™%:Xî.þ»@i“ȤOVL6.Ú™àÇöâwø³Ÿæ¢r>ÒùçK1/:Ê•t‘FY‘³}‹ àÔ GYîüzjˆ><.¡ÜAxåÙÖÎRÑÐöþ#ÙöcÖÆÝAÏýü@3ÐIŽv-Ť ­·6¶Š…ÜB±¢s?Š5%XzæBUÛ{Æ-¹Õ²iuÞL¥á›‘Ëˆï¢ ×n F!3eÌÆì*k=¯ù“Gw- 6©—î™vIÅÖ@GÀw8ý‘_¸˜¬d]âHÑMMô4þWÖàšiŸ`ðëLpªNTCØwÌÊŽt[é»’#ŒW"L2]iš´t9üQ?&¨öw·î–C›·¶»¹¾›Ô­mÍ>íÖк"ež…êl»N]»n!Z0¼þň.>«ß8tölÅ8Ëh”OoØl‰ ÝŒkld=ÑKéhØá%E(¶•`4üîÁø;{Ïc’á ò+ô$*áf3)øÅÆÝê)‰d§ PÙŸ)ힿߣ֪÷ïi‚í½1Ö1zÇ’ékϲº± èPñ£”YÆedèj€ÕG"Kj mç­u°5Ó ¾ð:î©CòiðßGkÔrj1º)ö8Àk†cm'ŽâNa%ëÞý“iJ^Ž´ŽL=È8pí_Üs ¡X°’gÌ®€#i9½,@ù肇|»ÛaÖt‘q·äZ¾aXÀv>¹3áû²ÝIôWhÔ*Q¥­½þ† D­% ›~Ü=HwGÚ¦i¾Ç5ÜÐNîqÌÓÓ¢Ìa“›&~co8B•-»ábG½pTâ!Ç“–5ZN9È |¶×äì¿&m%&‡ëÒ×1ëñeŒ¤ᜂ4Õp, -<“¥Jçi¡|³ºl?ˆJ÷ïyߎtîżçV“®‰ö®½G'UȄʯIGÆ¿ùÀx¤ð®\ÚQ˜Ë($HÆid²ÉDEQ°ŽÔþ‚úô{d# endstream endobj 42 0 obj << /Type /Page /Contents 43 0 R /Resources 41 0 R /MediaBox [0 0 612 792] /Parent 18 0 R >> endobj 40 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mgus2.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 44 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 45 0 R>> /ExtGState << >>/ColorSpace << /sRGB 46 0 R >>>> /Length 3597 /Filter /FlateDecode >> stream xœ­›K·…÷ó+z)-Òn¾šd–6’Ä‘#0¼äqbc$;–‘üýTÕ9¼#gâ,4šoúv7›,Öëô Ûç[ؾßþñð…üûäÃ_þðéöÙ«‡c?Žc»ýùê³?É_kÙþõðÕ×Û±}ó¶Ïåß÷A?°ýñáŒ{ë[î{<¶w[>ÂÞñi{5&¥q4„°§x9<ðz¼ô½æëqâåx<ÚêåøÀëñrîçq=N¼OGÙt9>ðz¼¤=—ëqâåøÝã^ŸþrüÓלܷ“ûáíû‡O~e%^»a>¯?À籇°ÅªÃ~ýn{q¼Ü^ÿð»×våŸd2J¾9·<ãÜ’÷ÐnΠϸq<â~Þ:<ãα„ý¸u|Î{ßs½=ùwN¹ªq^ONϸsê§ZîÍÉ—;Ã8JÝ{P1Š%îù„ ÝNWº ǹ—r9L¼9^Š.×å8ðz<ioñrœxsù¿ŽîöÉo÷77³òÞdÈÇÞO˜è¾œ·ž-ë6 |Ï>;ä¶ÇÄÓóóO—µk•§ŸÏ>=æ²—ÀÓÛýªÓ1<äÐô£t‹cË{6Æ*Üãøð+óѿړ„&“ԷĉþêÅ__ÊÄÛ‹Ç7?}Ø~üáÃÏÛ7ß½ùÛû>|÷áå×ÛëÏÿËI_m â~T\øÏ?=¾Wº½øçK1¥íÅ›§Ç÷o/×”§ÙnŸ0¶ø1¥Ÿ·/·÷¿’Â^ I[Ö_ªüøXpâæz8íÓÜ\ƒÚ~6¥sq¦®Ï$T÷š<éy¨â ËDâ´ÅdO8:õî5ìµÎtÔBiOÍSêJ☻§ª#«§^m¢¸Yß[ö¤nä¡É(Ч®£nA#ÞDi¯:ê&>ï˜)¬§½œž²ÎnÏ{ª3ÉêëÝ{Ó=àèÌJb/a¦S#ö“¬¸Üª,0b;Øö÷ÃaÕ¹}9ŠÇŒj*{3Lš°ÌØtE».£Çj%™;‡%ÈÆb1«ÑMXXTˆâ ê[6ìšÄÌö¨–’˜G_ài7JQ`F›Eñ”ifw!‰åbêrÖµ›1éþÔˆÕše)f]P (Ñ4:,ãÃ’R¦6†n¨êðÄî bQâÒf¬º‡e–Zqظ(-©»ðˆÕÍÖv û¡KŒ „!!ÜM(»>ÛÆÑ-׆=è0b,ºv3FMÙó©®uÆ„Kix“œ±À§ÕÌØ5´ ¦º'‡2r3þ$.¨—5K²šäaúqåW—ð 1 Ù¢7AÂH]N¼„Œ‰:\œü5…‰d™«B8™(ê º£¤¦?BÍDE7«e÷ŽNÝ•2ŠTf/s Qu5ê¢õã¢îI&6Z’=» ¥_Ø<Â×D‘!±¨åO”à)u¿-è&ìMTÔe UÝ „Ä :’Ý.%ªLÔõ)G¸¼§n6Â¥#óQòdâA'J´„’¦Ž4gWjŽ2¼¢l)™å‰Ê^õ‰ä©e#Otª¡{¢ª?…ªîõ‰$ôfì’{ ÇÇ"ÓU£Ç°w‹ÞÇž’ÇÇ"Óò sÚþvhN¸GíW8̺0#qXÔF.²À¦F(“H·"(Óî°b£ÉäÖî±a;u«gÔ,Ö>lYÀ­<ÈadR!ó/Vë9©Ã·5w$RrÝ´à‚ì-H±Ÿó‚O8;]Iå<בl‰—ª ‡†ëZÒ{ÊuÍs<Ôd̉G^qIÛ%Ù[q·$åè<uð—tÐs„é2!ôœà,•ebWŒ )£çÌ”C¥äKžmš”‹zÏ Æ¤ #%çŽ \¹WÏ’÷!m‘…«+p Lj=G~^fʼn÷‹V­Ø‚ÓbÏEÍà’¯ø'Ýëž+‚SgÏiª¤@’íxé•p[p>t‹+W-=&`HÎ=G&àÑÊÒóy,}÷œ´±xIà=gú Y¸¼bË’/)þŠÍ±ëÂÈÆñÜh)[\XðiöŠ2Àsç~’… kÑaæPµêÂIDôœXH&‹ì+Æ~’…K+Îêà•ƒfáž ŸG¸÷²@®®¸é6V¶ŽŽgå‹f2+ÆóÊ –¤Ø’å––œñ<Íü½ã¤ÿ)›ãð\8ÞbQÍóI* £ñÀq£?@©·âj÷+æX=wtetaŽ5×HÖx0³$ǘ/Y8ùÅsÐ0ü4ÊIÏq\Ï‹ç¤Ëôd§lœWŒ×$ω^8‹ŽG³C[Iç‚+ç[[?uÁ ‰².¬86Ï¥¸VIݳ&Ãø¼åKž#J½ UÏœPd*k¾²`ä ²0bøž3÷[5½Ásá|£,_qÅç£Ï•þºšcðܘoieÓÜÏ´ºéž5éÅõNó׎#ŠxÍ7'æcZÑ­¸ðz²020Ï…®NœÆû#ÞÈÄ”7Feë«yîôZUxŽ’¦žO7±Éqà~FócÅ 5{ŽôçRˆãõÌÒD'FJ¦¬˜½~NŒç‚|L, >Q¢èÀkXpE<×—¸à†ý¤“ÊËs‡½ëÀB^²Ùƒr_p8P`éÀ¤Âòð¼‘ç¨ aX±•Jú‹8VÏ ûYà9c=£ä«Re{>G‹+¨a{67ti­Øò(ù”ìÏM›+Ú$3áıÖ‰,ùšçˆ†›lž¯'ñPçÛ1ë5Jâž+Qýq_pG>ÅßÔæ9%ä+I&NíÉq…¿Oò`½yÖǼiÞ]X›wÏQ„¦æžµ×.Í=Pƒg@›Ì‘UUh…MD}í.GVM¢¥åÈ<ÚV÷¤rÿv3“eXÔnfB;Ì=T¾hÝ8²Y×öUôd+Z3Ex4f™§F[ÆQ½¶?&JWqæ( ¬CØQéfÂÌêÕ½G4à)ì8¬7•ûŒM8Öå)ÝXU<ã ÅÍ x@j5‘¯Iá~‡• ªE”n¬Vœ±ÁR)¿x,7u’ÇŽ*ê˜û¨A¬†ñˆ ÒÍ„õ`ýŒúÄcÕA«šjÌmgŒÌܹzÅœÙ塘3c†{Dþà° ÀQê™ñDè¤Ô3c…Pê™±áñ)õÌØ)1Aê™0²/M©gƈÌRÏŒ 5#Åœ ‚2ÅœY¯QÌñX¯ï¸8¬ìQË™Y{íÜÖŒ^,µÇ#QËq\˜2PËq|²–Fêæ¹Â¹2yó<^y¡4s>˜:P rF‚-Èq„ZãÑë¥ä8£ÖZã‚ZÍöVXðШ9½6jEŽ“jEŽû¸´¢™Ëxñ‡Z‘ãÀÄZÑ‚1_ÔŠ'®µ"Ç™½0jEŽ Ó,$ÊžOjyÔ’Œ^4µ$ÇZµ$Ç/U -ifÉ+©AKZ0z‰Ô’öÒ©%9ŽW­èÈ ÎW­©­˜/¨°Xð|Ò¨59®LU©59nc>¡59f¯hhM F/–ZÓÌÚ‚…Ö-iÁè•%ËÇ=^'µ¦ãU/jMޤt–DžËК E-Z8µ(Ç'µ jQŽ+×›Z”ãFM­ÉñØïÔšf–äÅ^æZ“ãÈ^.µ&ljûZ“ãÌùFi·â†å¸ÜkQŽOîWjQŽëxÞªŽÕsc¼¡Vµ`h…Ôªw–ùÔª ­å©cIy°©e9öN-kÁЩe9ŽŒ÷Ô²'CËZ0ž—Z–ãB-‘Z–ã…áв -/SzöN-Êqc¼¤å¸ó~Ô¢L- ZÔ‚Ñ^ Ö4±i •Üâ’a¯Ðš<³ˆ Ö´bôКJ¾¥ßÚsÌzAY ß3ßeЯ +æ~PQ)¦Ÿh ­ÌqG£.êK# –4‰lÃs"W+ <—ÁAãç:8êƒ{îxÁ7Š¿Êɳ¤ V°_êXƺÿéÛUñ´Žr@#ðÝ[ºaû^îGÏÓÆÚíy`=ïyßæ×CX½^üËã(åþ<ðÿå{¹±G¤ºã›‡_½øæñÍÏÿí·øêÛã»7O˯Ñ}üzZ à›Œ¸œ]æßÅUÎ|ÕoßýÚ1‰ùpLz±Û}ñðoÚ'Æ endstream endobj 44 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 45 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 47 0 R >> endobj 46 0 obj [/ICCBased 48 0 R] endobj 47 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 48 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 41 0 obj << /Font << /F55 10 0 R /F59 31 0 R /F8 9 0 R /F57 17 0 R >> /XObject << /Im3 40 0 R >> /ProcSet [ /PDF /Text ] >> endobj 52 0 obj << /Length 2142 /Filter /FlateDecode >> stream xÚµXYãD~Ÿ_í icÜvû’`¥åX ´Ã‚}ð$Î$lÛs½ðÛ©³»dY@‘•>ª«ª«¾®ªî/n®>}“ç‹:®‹´XÜlÆ”qQÛE‘—qšå‹›õâ×èÕõ2O³èxm¢||·ð5ð­yê3h.¹9Bó¾¾ø6ðíà›àû-Éø{{B"íPíDö_zÑ£L7ž™áÙ?½d\óä×­züÿ\X߉¸$U^ïn¾[$‹¥1qŸï|_?ßÇeƒ˜èWáúÎ+ÒÉ4nù=|ñ…½¢roà{ ß÷b¦¯= TàQ$|.dj#Û9ÈÔ áø ÿ‘öÂ/YÉvâ¢c­dû@gã”A fEœÛlnÆ'Yш¡TÇ_„‹ZfôæîÇ#wBy'j+ÉNþƒÝ=@àÔÈVÔl ZúàuËC?BóKø~.-ËK›}åݵM[ÑPpÓ‹4ÀÌK1軹ã‘ìùÜñƒ‰ÿqÿF>rÐtÉ]ܵ7\â…Űš«+„/Ä[¡£§ *Cê*Ýó _ëâd„¡«ãîßËUyÄ“0ÈD68ÂP-âºR$lT彜ðqàœÂ-0«€¦¡tZt;[ A^©Ã¼’qa2p;iö‰„©'Ó´V‚ÃHð(Ю¾¾¹20˜,Ì¢.㲨™IârÛª»úã Wg–‚&Mé2øôÛÎ.¾ê¯~‚ŸN-…å2àù&Õ*ÐKhÒºŠóе»Ù¢«_ËÔQ3HcºN+ðLZ‚çhàˆTý­¢©yø1ˆQ¼ší«kâÜ”j—Ë9‹–…:gY\VÞ–ÆŠ´øziM.ú³dœ¥à=a?˜B³ý@ÈĤ¬3oxŽ-!ÒE¼“ž×·ÌV‚-igÛ¥vûDgg™BSƒz»æmJ@E]΃ÍÀ*³,«£~ƒÿ•KóÍÈÁ“&W­Œî™¨k4²ßŠÏ8QF» ÿ7û=7úi+Nx骡𣹫bñ ¶øŠÔA³ßØì]%!˜ ÿ5º¥H¢oÌ@í.®À¡-Ò>;C‰nj[ ÙûK™¡8ˆX(÷²¢T?js”¦Fë¶%:ðÒ á²e Ò«™x²ë‘NÜëO7Þ:—yýŽ£êU$öéï…ÐsC§{½¼ž×P蕸‹Sˆý{Ñd§ÿ^ղņÿN-qÁX~eáö*ÚtÈ×¥•¢“št‚X@Ò>ˆÛ©§ñÛܺJ–5Ù6=í-q 9R\:Ó$8ü22ÏáˆõÝQ€B…úÄêò䦸A+œ‹Dtå§MäyP7{áßŒŠ­$ 8Ðvö”µ¨Oùr<âh5køoÔW¸rèÔ«;‰C.…ßñ$3küvlw¯KÃØ -%Y|§K[á펙㘶IF@­®‰'1hãݳU=R`XáÒ-¨E*Ü#F¾CyVÁ)àK„D™i&Á ¡  iÝC¥oÎÀu êæê˜÷^èÊT$‰/…þÑ+°=³–à ëÔ¡qµ=í:•ó—Zª¼¾ïç·Ü§Û*FýN¥žºoÛž‰f'&š•³î^áÃaÒ}’³V-ó™Å ¼°ö+÷:Ý Ð½w òlD+9,ÕvR™;SwBÚŒî"€ÆÆµµs2çÅ" VЩB80ñøª§ñµ§\!É–`##.ìæÑqïi~rÞr¬…ú€‹ÛVi7žW cE‚/%ÒÙküÇ'/sŽE#(w©<°fóóÒ!”\E\0{™%h“Î…!çP}RœƒÖpÆrr:?3õüß8Ó÷; »4º4Tß&³+IŒÆtEvK,˜VSP9ÐuV¹wŒIDõöG½ÕK0H£_Ð 0H\Ç:H¹-o‡/oi)!NQ—\œ†[…Í5…,Ü”Åd- íÑ‚‡|ØÉ½ipm5à tÇù{“±sëOÏ>VRùÞ\W6¢Ìƒ‹ç"QÒ\$¢0G˳î÷ÁÔ8…¥&°¢û£ÄºŠÉÜÑCp. UÓIg~@`€à¦zwÇ l¯4–=Ȧ®óâw·Ó üF~¦z C² )uœW…Bãôµòü¥ò [yg‰Q«Þ=¦¾\'ïÐå4uC–Éõº• ¶²PP8ƒ63XÉ¥ÖÌŠë’Ó¥«9ñ,kLá'/µ¯Ñ•Ϩ¸æ1ªåšÐò8z?™Ý-•I*-±®´  ¨áCÊTÈ»ÕL0â`­W:$nF¦kxf†7Í$ËY*Aê9^¹/jzP¡z]Âþƒ{Tõ:À2©¬áÄ¥Ž^W• endstream endobj 51 0 obj << /Type /Page /Contents 52 0 R /Resources 50 0 R /MediaBox [0 0 612 792] /Parent 18 0 R >> endobj 49 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mgus3.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 53 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 54 0 R>> /ExtGState << >>/ColorSpace << /sRGB 55 0 R >>>> /Length 1371 /Filter /FlateDecode >> stream xœ­˜Mo7 †ïó+t´D‘¨ïkŒ¦€mm (‚Œ›®ëÔë6¿¤Ä‘f¤‰“E{ðÂÏr4’(’/µZœ -nÄ_Ó/ø÷êðë¯ÅÙŤ¤RJ,?/Î~Âoƒ_¦÷„'-ÎñïfÒô€x;y1 ›$(q'¬Ò2FÆ[qÑ™™ÌlÕZKÕc³;/½jvÆj7ÊIeª}ÆfwFZ×ìŒÕ¾ÚnÛ}µ¿¾dçîÎ=ìî§WoOâòwQüÙ> {%µhÙ—wâDŠË›é‡Ëüæç‡jt†³‹±îˆ±ÎJcõƒé—‹ÖGÌ NKµ\53sJÒ†åà#f66Pp¶Á戙Mò¹‹Áuæ.Ȥ)D22èBk³iÔ¬ørªµP³jc%˜jf\ØH“š½`³nÁ¹jg\ØËb«½­}é—…3^.|„‹‰˜q‘Â)0~¹åÖg‡Gt®‡oé³Ã5Xé8äv$??Ë–÷óøãç¥d¿ÏÏ7Q¦Ù}ÐמÉê˜=]*´™ËQ®q~ø"ËÀ.V:F IÎ÷'¿ Ó\_=Äç‡Ã“ø¸¿útÿpØN?ˆËóïôþ›Oã^—¿yÄ7㋯vOû‡{ñeÿô‡øùìm})nG,·h|¢EqD×â¸ÿ–ìiŠ™oȧíä½L0§íLI* JÒ¸Ž¼–‘‚’É÷ÒZ".ô„¾¥·,Ï©'Œš=8òVGÙuHž*sGAB¦$£í)J‹¥d¢T=¡Óì1`NutÙmLÒAO@É|;%%é wˆ´Ô¶'WV”\O¸¢ü¤—!õ”¨'¸Å“³Xàˉ“t³z!ö÷»‡ÇÇëÝ“Øý·¯—XÚOþ9ÅAx½Û¼Ì=;fÝ×§Ú=Ü}¾~Úß4ö‡?ÛÝnú¨ªÿ endstream endobj 53 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 54 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 56 0 R >> endobj 55 0 obj [/ICCBased 57 0 R] endobj 56 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 57 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 50 0 obj << /Font << /F55 10 0 R /F59 31 0 R /F8 9 0 R >> /XObject << /Im4 49 0 R >> /ProcSet [ /PDF /Text ] >> endobj 61 0 obj << /Length 2579 /Filter /FlateDecode >> stream xÚÍ]oãÆñý~…a- œ~SlëM‘ ZäuQ¹<Ðe)–HA¤ìóK{çs?(ú¬ô¡èÅåìÌì|íÌìêÛ»w_X^UaU$ÅÕÝúª*òX^y&i~u·ºú9N7qp³È³à׆Íx³Hrxøîx¼©o’2xºI–AË=rÏÏž~ÂŽ@ÃÖAé ‚(I@³+‹¾Aþ§áµò! {Äx¶ì}çr3¨«5 ¨äø>ŠÒÅ4C‘£#›àÏ0 “ã‹]¢¿ùåîoWÑÕ"ŽÃ*{"íª%Y@>Õ{ RV‰¼AXƒ¦ ðV)I´xö¡R5åF¾7ÂÞeeš¿¢ Ñ ¦ž]V°6%¾þâÛ8IÒeXU¨fUb47šâ4LKÅ©€ôpÚ#Ç9ò2ýz†ž‡Ëe¬¤hW”9<‰­Xѳ%Ó*ÌÊD ?EyD4yÔÛëÙ4{µ+€ÐY‹4ƒ=—¾Ï†‘ƒCåÈ‘ÀÇÇþÈÑš5)‚^\€t\{ pÛ)i†qN,FuöÂÆðÁ1iüht&ô=EÓœH¡³cèÎ@ûG+ÃNÌó(Ûa´ŒŒ·²OðéLà.HÐÚ‘|~Mð þ)гõŠpŽçâý5g§IåÇg”Ÿâ4£¤P{Ží4`7íd= ÙתÏCGNÀp}¹FúÈɨãæÉÝŠÊEVXDw6Ž;Ö”ÀìRNÁhë…‹xlEÓ‘s έqÜïC7O=ž/?c£QÈrT#æ`%YYw¤]–yé†d}a¸IaÈýÙŽ<ñlÝMß~ªFìÇÏHM/0&úšÒ–¬:WhÑ£9 €kŠ“V„ñÞíÌ’ 4ëÁ3R!QGñiRf” Ë<ø ù( Í×%n±–V&Puˆ–†A {V¤ŠÉ|h÷$»K•2/—ÖÊÈF  ›£ÙH­ “Kbhwž(V3îîñ¾”uÓ,…²l2*¢îeJ®–ÌK‰Yå•¡î§Ž]åTÜ÷’üJ(üUáçYIS(¦¿W[1Xs¢˜luW¥Zo K7L½•Hôçë®÷â@Of×+-Y4$Aã,¬²ÌóŸ7ËŒB:á ݪ(Idö WZò¶+I¤’K{5®¦È2tS—²ÜX¿¨TɆ …FÉËÀ÷yÓó`²LCgMbRaMN†I2dC“Âl%4ŽèÎ0‹9 õ^DŽSÓzì[ ö´`Ç­'Á i×O§+K»¡L‰nØ0µG†6ýÞ«úÛvü6‰Æ-©¤ËR0µTO•·£X'1Ž}c×¥Îm+Šm˜*9Îyr&S·”ÊÍζ])îNœ€Äo(§º%†Ûì‹Ô­èžpÎØ®xüW~Q±!ƆOî°Ü,ÚjÊâà§Ì»Ý× Zõ‘¿êÑáf™7RÅ‚Æczùø¨ùàÈTígTio»ê«5ägv¬&»ªPm¨—&ÀLà| s5½KnlAÊÖG5ÀÀ9ŠB4rˆ¦ûÁ Q: TVGÍͰÆÖ.KÖX³Óe•ûdÓ?`ò ðí|výéaÓ#g7£W«Ýð´€I àò–É0`O¢E†^0‰½G]×GˆI¡D‘YjŽAÐa`zCY–9EÌÁlJ$]£ëq0ô{šÀ“ÅJ¨D$›>„)º™>‘ÇNljb" ÇÃ왈hÏŽf^²â\KiI›Fºca/*LVÈü€ÆìÇ=òÁ'ÛÍ=œ.ôL ]bRU3=0S³$ŒÜR£ëôL,†g8m§ƒf·F“”¿7¦1PqM¸.gènçÕœ5Í’ nïˆS”PZëùÜ×kçÞ1†9 uùðŽ˜Ò)~1çnåO:ëÈæE “\Çèzæ5µ ˆ³¾õz#$P+ãø@á´±Ânìz^ÆlW3«˜I˜³'˜yáæ$’˜‡û^Â?ÖæçyÓjïbKa«!im¤Aª-Ö¶•jà{-ºÑkù¨÷NNõ{­Öȱս¶{ª#%X˜űr8X¢*ç­d C-P?^²úß@h$)íEMc{N¬þn>›ó7k.Çì½ô '©» k‹zÃÑJ 2ʃ³ µ£½•g-ýyâú;+°Ù»òVêØÊ0·È‚’Å›hr4§×L܈á:Á~¦d`e©Ï 8ß¹_bÈW¦â(+%G²Þ:²•,P;^»Ø“¼;™!æíþEŽÿc3½Îè-siH®|·7´}Þòë ö9ß&¯i@þÿÙæ”‰Ô 'YûVl¦ß£ 7VŽqbŠãæ(E‚SÁ|OjÖ:»Xd°bTúàÎt›\ejiˆ PËÕ¦t—¥Â:î9W홞CåúK¹¹5’:ã®çw?(_ÚRöÑÿƒ[›™é¼óÜÜü¾"g¦K”qâ^ª¸’Óxª³ËKÊð‹!’G2{˜çð¨Ê0)²™øà>vlºPd ùkŒ[mÞ¹t%\j·é÷üHFíÂÐïÈ´ˆ¸²òœ>oc¥5¹¿—#¼µ‡í„b.2Þh“¢÷ëö˜·÷Òù½¾w˜Ó;¯¢‹j©îñSЙ/“4\fyzqÅW\UæY1Ó1›£®ôTæp[£í缿•°³3—£)gÈÚ½&sî¡©ûîþ(-íþ¼ Ú”kÜ÷ÂÝÒ-|ìÌ¥âñ7ötÿU?waøB_÷ZÛ1éÌ’Ë2ðo-™qðÕl‹3U-ØžÄ%Þ6Sä#º'ðZŠSt\‹Üú5ñZ>U¢ÁYóÚr»–õÔ“©9Û^_lµ½÷7•ЉU^ÿÓÍ(ò÷ Ê5ß)̓cêƒß`Ì´[RÖþmC„Ï~3êúýöòè6ÌoØí`ΰž%ÎÍ?ÓBè¿ø-©ZåQŽsgŸðüžÄÐßY°s¶ÒL(Žr:Ùm;Ç(·† UN¡„f©¯÷g½Þ±wŽˆ±G=iá> ²nˆ[‰Ï9q+œªa{úÃïáùYÀ8z+Äâd¥Òë^gŸ¼¼"ÃOæoKŽ7W±ÎìNbÑ“giH—ôð|tw×”Ã}c­Ý:ýygã '³Âb L@“¤±v2‘+ò$3ÌMKVx;:.Œ­ó¢©{1—¨Ê­æi¦‰ùë÷sý‡IRÙ6îœ"\Ì@}4é~㢠ó›€,³¢bä1ß}w÷î?ÓO endstream endobj 60 0 obj << /Type /Page /Contents 61 0 R /Resources 59 0 R /MediaBox [0 0 612 792] /Parent 62 0 R >> endobj 59 0 obj << /Font << /F8 9 0 R /F55 10 0 R /F59 31 0 R >> /ProcSet [ /PDF /Text ] >> endobj 66 0 obj << /Length 1493 /Filter /FlateDecode >> stream xÚ­XIoÛF¾ûW¾”,š."‹:@—p )ªK‘æ@“ÔR‹¢MR–ÝC{ß6Ã!%;)4gÞ¼õ{ Çö'üø“tîÎãd¤žë…á$¯ÎÎÜ4VAH Ö’Ž|ÂåMM~©Ï~ƒ}4•3KçO‹³Ëëd’¢ºx²X³óÄõ’t²(&ŸœÅzê;xÚé,ç–[xê÷sûè–{áîà)™\éOÏʆ辳›ª¹c©ØñÅÐ`fɲEzb‡Öb.y½„u“åbÅꛯ—"ÞÝNgQèü…Näyþyñîò:š[(¨4pç~°¹DÕ4ÓH±ïðÌ¢6~¡r“$Ö‚ã°ÄÓ°þùƒøÔq¬àŽ‹ê =AìFi:™ù¾›Fkz_Ogw}'¡íVÓY„R+ƒZÉä­ ¡Ç óµ9¹Ÿ5"æG‘r®(Ú’ þ¼ÓȘtk±Ör@$–¨=Ëí4 Ó@> vLC$iÀ”‰Ë‘ÎÄ£œ=ÖÜ=A`Ó ÷j=U ˜Gì¼!h`CŌÓ "AʵŒiPŠÓŠ®>¢!®jc¥eÊžMãòÀA‚‹%‹wüuUR¸%ïË{«¦Wge ·¶,–í ^&‰;&Ø6ožgÞl¥§2d7~aÌ¡ç| RÅ©,ªìV”cC¡E±“mA¹çq e-oÆííùÔW뺑F¦B+Ûì0CÌeî &¯©AAŽXô8=5Ǿ ÿ‰¡¾â¤>¬¦¾g8=h€ îÄc9ÓŒÚ^f"ó`žÐ´2Í>FbÄðKô°În¹O%\V@WøâŽ9ôÜšœªxîÜ0÷:Ë×ËÎŒMÒ„Ó®Þk8B¯0z­çlÉÇÕúf0·4ÓÉê3«àº,­ABä=‹Æ÷[¦hd{@ø‘gŒ E;7Ã.eºÖztËŠJ(?¬Òº—O'4ñAš6 FÐÕsG×^4ïótÇÌwyÁÉ%GVð&³ÚRÚ‰é„ÇN´â8Øp9ñ)“IQ^ëÜÀÎÔΊJÞdOÑeæc4.2ò2Óv’ÄÉ0ðNO ˜ÆÃVÞøJÙÌn֘†S$w6a)Kp‚Ü‘Q)ª»FË¥ÜvH[qŽZé:*éØ÷œ_%[,¬dîT„a©½l,ŸÖƒÎâ Hú¯©rîû»O×—SéÀ7mþ‰z0%”ÏîÉc¾ DÖ݇%¬fôëoàÛ¥‚HñòXÎxy°æãšî<‘§+žÃ”».|_°$J-…m;å±ZöJÑ‚]PÁs%ê5„ý2îHªÔ“yÊ#ÁRþhéøÈ½N¤BÀ(O‚±”õ­¬ëbµº¿îÛWC_Ïe«}m-Åç½¶ó)r‰|tT˜O¦Î•Iß%n ~»þÜαW”f’ÙÎ$à¤ò1Ò³¡!à•èÔµßêk«äÖæ(±¹TuêÏ©Àgׂò…]QþéÁçžwÜW_ß&…°ëõŠÓüÅ2Œ~ !:þÉ”¤ï|–fG·€£rÔÜV¯p•à=> †= t¹Ä¤/nW¢Àê†ÊÿN:YÚ*Kìžáy/ù|;ôú`Á¥ŒoÞ±[v÷]IEþaÕ³Üa5žµ)“Ìp®$‘še#o«Ÿ_0øQ éydc´³úþ|Ð'§~ÓC°± «}9ýá(ýÁ°¾A>þCösÁo{<åǵ¨¾&á¡Ø£›—ØØ /yß=Ðr t)É×Ù*z°1¼¨wÔnGÓ½áóýÿ˜×fà+5ëñõ¹Ì¦©ë'É0öWô~“OÁ Åq½°ÞêHø•Ñe  S•ç&‰}ºÈÈÖíÀߦRë6ÍÝP…úß#߸oE®2ÿwÙÒà*õy•j]¸ÃÿÑ¡ëA G{Ò r½]œý ´¹ey endstream endobj 65 0 obj << /Type /Page /Contents 66 0 R /Resources 64 0 R /MediaBox [0 0 612 792] /Parent 62 0 R >> endobj 58 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mgus4g.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 67 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 68 0 R>> /ExtGState << >>/ColorSpace << /sRGB 69 0 R >>>> /Length 1923 /Filter /FlateDecode >> stream xœ­™Ën%·†÷ç)z)-B“ÅûÖƒÄÀØ‘#0¼Œ5sIFAòúù«ŠÝ§»Ø‘s,t¤O,6ouùÙ',¯—°|Xþ~ù?ß<ÿé»o—Wï¼÷ËþóáÕñßš—]~úyñË/—°¼ÆÏ‡K`ƒåÍ¥k}IÝ‘_>-É×ÚÀ˃i×Ö‚‹´5¯xmÏÝÕtm¸µ“o.Ô­}Åk{.®økûÀ­=úì|ÜÚW¼¶çèR¾¶ÜÚ˽®~kÿöqlî»ç“Í}~÷ùòÍ'ñø~Ñý¼~(ïBX¨ò´?-wþ~yüpùý£<ùå®›‘Ó®o¾¡oN.´]ßpÃÀäÉ•ý¤Ã #SÎïgM·ŒÜ»Kußù†‘cªìœ×Îñ†‘c/ì¹»ÎÛÈ깺ØE„¨t—£ºÐ±9^éÚ0H£­y஽DWwíŠ×v ðû¸µܵët¶öëìö+ß-÷w»]ˆÉ5ÄTc‡Å?Ï6îÅî‘ãûÚ?ÜÞ‘Ù·þtsòÅu¿ö·÷Gfj[ÿt<û‘.)4Ù*MŽ+RK. ëYq5~Lý?ç“P¯37ýéî/÷KÀpwOo¿>/ûòüåÍw~X~ùõí_?yþõùþçåñõ¹ì–ø¤„Ô§ÿõéYu¹ûç=¶h¹{ûñéó»§åËûåûWo¶GceË~µþ‡ùQÌL_Ÿ–—Ï¿U¤‚Ëþ·ŠÔ²KIl<‚lPM® ÈP‹œ„AÕÕh¨ËÀ/‰:â ö2!¢° Wë„lÌÃÔèb›##òtŸ² ‰1˱dcÁÂiÚ"gF8tš›w%OÈÆ•1p™´˜‡(0:_'ŒÑåÆ˜]êò£ o¡E6æGuü.&âe“£>aö.ÅXO‘å„`Þv” ÂAˆ76xNI3¯æõ2¥ƒ€Ê'ÜøH”[9aØëóçÖ‰y#h0rÂİÏb¯ªlæî‚2q”LÌ’÷83æ+ÇBä†3öM¸±/Lá’U¸»ØOØó/0y‰ Ë„eÈ|ñG §ìÕžXHÍÜ]Oƒ °_žO¸ó†7ûtÂ~œ'.Ã8†–…›&Ã<ÞÊTOX”sgÕ:1ö‹ 3þ(ý„¡Ä_"iN«?GɘGIÊfÆ|²rfÇ›ö]ÇCH3'Ïn †c×>s .ÈzR☟ ªŠ¿¥ÂŸËLjxÏ3ÛŒ—ÑÐN8s3'‰7ËpÝ/ÜZ2ÍGòåÀqfÖ³2^c—™cÏCu uf,ÃËzJ–x²‡1³Tɉóo`ò'4ÅÔCJ3c¿¥|°fiyfH> ™ÃÄ8õg®Øyf¤‰ÒK=6Ìç)ûÉU¦Íœ²–óÐ%1OŒcˆ²_¸ø2s¤ÜêÌ|ŒjߤZ®¤‚PO d&æç)V3#ÌK®â–‘Oøq¡P¥ÞFu•p .Æd1„¤§EpË4sÏZg Á„Éd_¯2—,IË2t–¤Bâ5'‹«²c) G1(yR5žóˆáÕ<%¹(×5êö¼™#šb°H®Çg±ÔË›9np=[$8váŠÐ[È?@þž‰b#{ çÇMö AüJM‚âFI8_xÏU®j*ÞT¬Z-Q&'Z-E¨ a%þ¯2õH›ef<ÒβyC›¥ˆÛ#m–"mo–\—AN8Òj©âóHl)‚¸ò¹ItŸ(^)”V•OÈ!×ð«‡ôr´œ èaâa‘âBÖáªhEWŠþP¼ÙXR椖1Õå*I-³äR.üâfbØKp­’Õ2ìÓ^²NÜøp•¬'¬%sHR˱Œ’‡‹Në3§¨÷=æ^OYSö¬“ã*I-ÇqÏY%¥eª£$­’Ò0ÛëóT2ZÆx*™¹c™%G¯ C2ZæJ®óér‰³œåäU2N,o ˜ƒ–8دJ¶UòÆ~&}Þ|†I^é1W-#†aFú¼ÎÙib,S% kÇ3ößX›´Cèe÷Œ»JÊ$QnÛ$ï!6‰j9 IíAtÊý Q SÔ÷²ñõ„spå„ïz%’Öp_oųt7ÈA ï%VAjÙ÷á-CNìWÁIm–ƒg¦‚‘/T–¡ƒŠØWIK#IèxµrXOŒÓ*²~ÁQffÁÙó…Ñ2…±›-«·fÁ)Ñ8¥e$õFdryWc˜ïQ*H£+'Ì÷tµWAj™_䉤C¶æ*cŒ†‘d’ NùZfb‘”GÂÁŽ Ú9ÚTp&>˜‰íQdnN)ÍŒñHž‡lL'Œ L*YE9Oœ»ž!›…23öS²aâé„ù a¾Ðõ™9ÛË|ᯔgæeŠ=tY©·u9,±’Eö–&§õAuâžÕY΄Y\UbòR1Vüo_Z8‚|ÀGueù&ïÅ~)û)ÿ_¾AãT/S‰¼+òVú½¾“~úôöãÓéûí—Ÿ”ÂxÒ<`¼êÿáòoÚ¼ endstream endobj 67 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 68 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 70 0 R >> endobj 69 0 obj [/ICCBased 71 0 R] endobj 70 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 71 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 64 0 obj << /Font << /F8 9 0 R /F57 17 0 R /F55 10 0 R /F59 31 0 R >> /XObject << /Im5 58 0 R >> /ProcSet [ /PDF /Text ] >> endobj 74 0 obj << /Length 1638 /Filter /FlateDecode >> stream xÚÕÉnÛFôî¯à­Nfå-zè z(b šh’¶ÔH¢BÚNܯï[fÈa¥,(ÚCaXšåíûHeþTV{áË*3µÒÚ¬=\½½u©%€dIW* …ƒç?Êìûáêgø‹WE Y$4¿½¾zþ¢Êj$Wf×·™RZx¥3ã+!«:»î²ßò_6*ßöðqÜFùü–ÓŽö*¿ã³ÃFÃÅîá/‹)lšûaʼkh˸xÙŽ°îé2@~Óì‰_/pÛ ¯)—Säu`䉱wÛáÕc¸ÇÍCƒÒMŒÝŒ‘™'-†{Fmö{’ ”y‡=!ÿ~ý2+”5X­PJÔαe‚¶¶ù–T”GƃC“ß5'æ‰l1:¿g‹Òú@¢„{– a‡Ûh\Îv‡v”D=z±)j]æ/6 ý{”ípZ³hø«#±_pÿnܶeä×ÒIØÉg›Â›^˜ü´)€O‹rˆbàæÈÆ’k3pUî‘‚© qf¯og"Æë¼AùD÷¸Õ;eàƒá#¬GO>n\ÉZ ôÃ1P¸ÅͨòÑ„Á{*¸üÄnƒ¶_Áu}gˆ¯~!ˆü†ï,¼äå.,ç€B¶m;.ìán–³Ž™'†y·mÂ鬂Îßç48®½‚QÔ{Š%ºYMNEjÑ'AhwÌ@´» vG¤/†Ó}ŒÃ>µçëí–¬×E{j™“ð=‘{ˆæZg§1FƸ㓕c -Ô¢¾Û@n ßLèDq×O$ ·BsdˆXj&ÎB¾A¥—mÌz¢KÀËfŸˆtˆe °l+ÓhŽ×‘YÏü–ÈOMV²Éæjç]þ*ÅÊÈ…+«üÕÊ~](½ ¢SQꣅžsL¥A“‘© ‘¸Ã1"Âý …ïG«L„í‡ækò0ç@K¡æâr»D"®ÜÂëó `Ñ3F¯Dʤļ:¯g FÝ)ÀÌ fÌ\›Ù¸jû@8¡ŠžŸ¼¦#/¼¡Âv×ÇRñü…­2èDÆ–û5o´¨up´Þ€.Úå?Åæšwý>&ÒºÓ¦V¢’cÎ ùO1;¨eDa—úû·ÖŽº?„˜|ÜÍå8µ^ûŽî=o.XšÚSÓ…À0fžÌRšÇÄÌ#ãªuí8\žÅ•Jgœ²eíIÿZU±Eiwn$'*†dÒÅÎÉQyaîÏiáê*¡¡Pë*ÿú’d^(çØ(’ÏJQ{S-"J@³f8yÎÓ [ÖYóÙÝ'd÷+Ù?A£2‘ØŠÒÔð©Tæ×Mx-•=稼ž0f%ða|ÐQðœqÂÊÌP §gJó=•¾6ÿÄIðå‰ò¹'´Ò—ÿSW˜ÏqE)låÙfå óÈ1ÊÛyZ&¤Wç´KQVö¿J,òT•¿?ñdÉÛ½Ú™_^p¼˜lÂâ_Œ/–W9gËhT›Ä·•† u‰JG«s´'HùÙåìð†c–ÜšºKŽ0`)ñ.x³Ê”ñ>H2Ö~\ÓéÏ&Îm}KC=Ãó„Kù#v›K˜©¨´¦É’Ü û8ê>Ú†‘„zn22pÏFö#Ï[Šž3üÜ›Yé ›#LHV¹ü:>†Ç Ý2¨0Jùõ<s“®Ó÷2öS]Ë|?¿Ù@3:¡!¦ï{’cWä{˜9A¢-¼Âc³â˜Ó^…a*<â}ö¬Ç0J^!ˆ!xó4„*úy@®ªAJ½ÇåMÒ„ —Tg@ô‡"p2p£Wk•Î\¸eÓ$Îåݱª`é¢"õÚúq @Jž_Öz=_â–(Þòø‰{й# o‘`Pž].d¿ÒB:û‘Y•; $m—‰ ÙÐ~>‰c?ÃÄ·ëC® Ãaìâ‘åx›ŸüO<7Ÿ'9FzŽ`(/îúqÑÀãåøc?NÙüòÆUBùWÍfå…4'žôºaN¸0êͯfœñ¼éÎfÀS|sºÒ#p#Nœ§8%ï'‘BTÚ¥b`öLó;\K'Lé²Â)!«0S×xuõÃõÕ_ú—$ endstream endobj 73 0 obj << /Type /Page /Contents 74 0 R /Resources 72 0 R /MediaBox [0 0 612 792] /Parent 62 0 R >> endobj 63 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mgus5.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 75 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 76 0 R>> /ExtGState << >>/ColorSpace << /sRGB 77 0 R >>>> /Length 2558 /Filter /FlateDecode >> stream xœ­šÉŽÇ†ïýuœÌ}ÑQ‚m€€ Ë3€a: È‘-‚‹Ì!¬×wDüQÕÝ)Ú ë@‚³²–ÌXþˆì°½ÚÂönû×é{úóòù¯üvûîþä÷~»üûþ»?Óÿ¶²ýrzýÃæ··§°½¢?ïN/ØþtªÑõ±åá¢ß>lÙ×»âûí~VJûhÁ¥x ïx/õ|W<Æ£ï.´c|Çóx©®úó¸â1ž|q>ã;žÇKr¹œÇñ«Ï=ý1þíƒ.î›çÅâ>¿ùxzù‡H;ñðã†õ<ÿ®Þ…°ÅƯýða»ó/¶‡w§ß?È¿>5Ðb”|1·Ü0·dúÅÜpã®^¾t¸áɱç/ß:Þòä1\n—“oxrÊó<9Ýðä4*[îÅäãÉ0ŽÒÜl"B±D—+Lèz8é<|u¥Êã¥ðvãÀóxôÉõxŒ+^ŒãuŽñóÛ]~ùÅçþîbRv^Ù»Qa¢n¹n_Mû¶¸‹7Ϲ»˜tz¾}:í]o:½Þ<=æâJÐéýz×50œrè|©†ÅcÏ. c®q¿ø^bôÿIB§E[Ò…~}÷÷´p~»{züü¼ýüéùËöö§Ç|üôüÓ󋶇Wÿã"Ð?e ¢ó 7þËç§J·»¿ SÚîß?}|ótÜ“¾f»üÂTeóc*LŸŸ¶¿mÿ[J ®xNI[æ4úëkÉIëT«qw®Ätˆ†KÅEó÷§&¶=‘îN‹®4K…ïÒ(ˆK½0ñš+â`ߘ*ÇOCy0ÑŒl‰ÝùÔ)ù™‚ ™©¹° vóS§7Œ–?ax—ÒL‘“‘ÜÍPΠQ,5~ùÝm"ʬ|—à³óe-)öj®©‚Ýù±À4‡_`“AzÌXøéŒÑ´ÀÔÉ”šÁÊÁHy¸/3›œEY›ió‹AÚ½(ØxË,ö,HÆÜ vXj w#³´Xä%S“ 38äósàü2£ø pŒ6yœùK'$Ÿ*¸˜‚@\ DI†õŒ´S²8UfÍùŒ‘7Åb•[‘sF‹ –Z“0c‹Ó묳šJ/üÿ‹¼$9TŽ‹šÊˆ¼w«l9ÙùŒÕ~Pô×Á¢’ å\#†ìj2HÞÔ«‹yM0’[{ƒSäV¤+É&¤“HiïF6˜\âeÏ$Îiýw¼?’Bÿê?öõBì½¼÷U’Æ5Ñ£d؇ÃL™gKÂq¦Æ`O&×Ô=<<¢ÕD4åJÑè%.¸ö$4Qu“C$>ã‘ ®‰ +òJЗµ6S@Â@òš(!€“uSl›Hå Û§ŸI#%ÛD+ÄfhHbkì56C¢Î>lhT$D*à&ðL$Ëk Þ# Ðö뤑œ¸î[`çO¦í‹Õbæ—&Lì¾  a_`Eð$EX»ÅƱ‹°° 08øñ{þŸ1hh…08÷ ’ãŒP¤$[´˜¡ $8(ƒ©Â`ƒ@Eìø|ŠŽmƒcaâ;c »I4‹ÊÌ7,¶JPô‹Á‚¤<¤"4Xa £³Ó,P’#™*¹©ÁÆËÍƳçNu>¸¯Y– øÀdyOd^ě墒ªH’÷–+B¥›å†àªâͲ¦ö‰Ú,g¯Ò¼"¯8쯹ØÀUZN4h™>ÒtðDË…Mš…žçºÐ2:y_HLË Žý+¬¸«Hµ<öç%^ÃEj#æÌ¾o9¨p#MkÆz‘—R4´œt? ƒ-S Äû4Î>–‹Ê,eË5 û2mìŠ%,…è¹à²Ü993´ºf¬ýÀ=X&]™p}ä »âŠë{µå€„§RßrT{§˜á󂳊câ¾â‚įłåªö€rÁrS©Š‚Árß×sH&33<´ÌçÀY—,½-O “ä?Rà¬+Þí»°}ÉåoâÄn9j¾§ÀêWœPl0s>[0¾·ÈÂ[.IXYÙ®( U_iÍÏëü`Ë»½W©ä-wÍ—(á-}W}}ÉEì‰çX2Ú ø‚áH2 ù€;+qɰWª¼ha,k!mÍ(è‰{YpÖVw+ê‚‹6‹(pñaˆáªùMËMã1Wr}Á_Ìç% †¿sE·æ÷é,l,Õ¸B\2âÕhTIf%×£³â.ëE-­8j<¤ÀFaÅ}gzqËi¿>³ž±œUÏ¡´b}©‚,èØ(^[ÞóyÇ!ƒá¦ 2 L|L·`äC4£,+zlb˜dV{àZ§/8h>"V\Ñ ¸,ëzGªg¨*µœ´mEŽCñhÅÏØq(qZΨwØ1¨F´ŒÃµ½¥¶â†çISͲ{‘ë±à†ö5olÎK–†lL]°ê±Hz¹®xàŒCÞ0É0©g"jÖ÷¡LÛ²žðÛ»áˆæ}$½Åg¶†µ^`&÷œµiIz¬®Xý!’žŠiÁMÁ˜å–u‘„m[0É$eI–“r“ÂÀrÙ9p¾³ÜvŽüá–)Þ‰½Q¼ÊÉ2É©G´_j˜ƼÜ@½ålmj°â„lo°^ž¡­hHZÜhÂÉÎÄ®IOÖÐú2„†Œ´‰ ÉÉšD×´Ÿ¬¡EdH²Z1éé™v,fðàèÈ T ªì£áèÑ‘A(<=:2¨õ¡Ô·3N-fIíf±]Tf3f}Irð¼Â~QÍxÔDÒµˆŽ*š+âí~ìd°DE>v2Øq1N¡&lªdQ‰XDˆ:ÀâØ«€Þ ví¡@ãÏ8´‚„·˜/ô¯Edg¶™6#KÎ mi™ÊmÆ ÇNz eU¿YMy§‹ó*θa… ùV §³^L²+†B è‘Õ„‘P3©EY:Σ¾¬p:Í‚36= #Ç!1o±!Ãxî°ÌØ¡Ï"mÒŒƒ›–Œ™MeB’¼Èݤý¢Å K7$ŒÍøkçj_ÿÉ’¯J@s¨È/ëR¹`ùéÞ¯¤‘}&U“W3Á˜¹zþ>ÒéÕ<ðoòc»Ha@>"qVãô|x|ÿôÍÛ§Ç/ÿÜ~yùiûù͇~Œ¦w#ÇÝ^ãvÛõ¯cž>ËmW¿¸ùú})V^¾å†×|üñ‹ÞQ_ôûÓ½×7^ endstream endobj 75 0 obj << /CreationDate (D:20150701163327) /ModDate (D:20150701163327) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 76 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 78 0 R >> endobj 77 0 obj [/ICCBased 79 0 R] endobj 78 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 79 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 72 0 obj << /Font << /F8 9 0 R /F48 8 0 R /F11 11 0 R /F7 14 0 R /F10 12 0 R /F14 16 0 R >> /XObject << /Im6 63 0 R >> /ProcSet [ /PDF /Text ] >> endobj 82 0 obj << /Length 1632 /Filter /FlateDecode >> stream xÚÝ]ÓFð_©ErÔÚx×^T¥R‹©©‚T}(}09'±.±ƒ“;îxèoïΗ½vœƒ"Ji6Þ¯™ÉO‹{žÄÙL© 7FÏ«Yži’Í“:2³ÅÅìwOs婹¯”޽G]ìös?R^y´óª^ÛE{meW;.áç[ZôH{7°0ÞΤ¼(ít;ÿcñóƒ'Ù,òD' >œù* ¢L‘äÅÆ¢•–"K“N‘S–' `NˆRìíFÛ˹μ ÁAPµh¨"²x ìniþ*T1ãБìÏšiÿ¾hüØks¢óM‘ACÓr E³"a€àœ –Ƚ5+”,@‡C0÷“Ðx¿Í³XðßV,͇C÷‘źÍ+ ïx‹æQäá É!- UÂ}í=j¿¾¤EyÔ»=s-r™•ÖûLÙ©ãâ<%5~˜ûFGh%­p¾'ð÷vê÷k;®8vôÔ¡ÎrSÞ× ó('¹ßM¡¼ÜÊ m€›Ty_r$káô‚©–Œ%ëm¿ MÈ‚OTñº%Œ%†V<*f(¨5« ä–£ú›”Ê ?µÍ)†+%ù`)×¥§¤ 1Ο¥aÎÝ`6CÈÓ o½µ#pŒ¼fôŠ­*{GvFÛ»é!}žXÈv<³ã¥ÅÞ¾ ãÔFBçrImr€ß­½}¶<*'ÆjvpyÇé—N0ñ&Œm)¡sÊlz»¼ä+Ô²¹¼bðÐϘœÝ`.eÓvÿ¤Ïè.|CŸ“Þ  ;ÇL ïòT^áxä!«4­÷Äü©¥7 LÎOÇ#–!îønŠá]ñVËz]užw#è¿hwWÿ;­'AžE3?2A–$C‹Qúê®øMOr¬»I®?œè'¢NTB!Ž,[ÐÅe\÷9&Òan¼o}‚AACñžâ½ÌÙÏ9å |#;bü¢¾ŠI~%L+5ëíóV2a½¿”÷|$.B~ôMFjÆ, æÒk¬f´9ç*ãðL–9 †xŽ §C96‹š0§»—LèI:‚lÒIö e¸‘ÛX?rŸ"Z2«< ¾Á¨nëD9=>v/4m=dsr2jÕô|Ô_™UƒöLN=’5nváB. /“לÀد‡õ@3¼H×%óÎÙl)ê L$jèK¾Tì'áþšÙ”Nž*&ŸÙArÚ:UDÙ[ù‚RöÖmú\"祿ɵó.—Ξä˜aõ™Æd© €0PQ4êŒáÌ¡¢rIÅÀ°¬6±W¬©¨I[ÊšØr¿5ÖÕþ/öçÑsíšVrã–ž[³ßÒ^Åu½eK}> W5W~Ðä$¡÷vÓq¶!Ø~G B¬#mäÉØ·b³’ÖÕ5H§@ܾ[˜º¼è˜ »H›•üE;jdt’“}`ÔÔöu~¬µ÷5¸$h¯Ñ* 4¤«fK¶…=¶?ʰ‹‘ƒ…ñEI($ª“(˜‡40ÇFª8nz=¤mAÑ`x´C_qÙ €ªŸ*ÊžÆêc8YYoà×xë–$ÓJ bx‚Çöì-Oð±,ð~0uù± mœ–D(±ÿŽ× ‘ô}ŒVÂ)¯vÀ–²¦¦¬¸ÅP-¶=:æèK JŒ9Rjn?ítI=$¨û¯,–¡“$ÖíšÁî.#¨”€'Ÿ7tµu†wÚD)Å6@ÔÕÞhøƒ“q–…Te·„J¢5»rêx3‹¼»ÚO7gœ)Õ'(©÷ç˺ý¨á8þÃeÝ'(ÕW:AöýȲ٤A¬ã³lN’TÊfQæ\Ù $¡Sß&N]ª¹žSÎ~ʃêÏaMr­,PÌûBŸ3ÍT¨>´x–bVˆ…AÈåœË<çŸ)B¥¸Ï‡‹û±ˆ)¶)~?¼žÈue†¬Уâ?dAfBï¤s5ã&€Ï"½Kz®ø,µvßñüÍz[BGú­ð¤ßú¼µ¸êÚÁÿO-nõ—s8?oXTáØ£r4ããžþK¾'ÓÖÌ9p^ pÿÂŽðw´ûçZ< â˜%=Aú}/Íwã¾s/Üý´ßg† ¥Í@ë¡?û„0¼‹É(ÉéÑZînÔÑŸ´=*MƒTçö.æA–e$O!â½Ç‹{c endstream endobj 81 0 obj << /Type /Page /Contents 82 0 R /Resources 80 0 R /MediaBox [0 0 612 792] /Parent 62 0 R >> endobj 80 0 obj << /Font << /F48 8 0 R /F8 9 0 R /F55 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 85 0 obj << /Length 3608 /Filter /FlateDecode >> stream xÚÕkã¶ñ{~Å6MozÖ‰ÔƒRÒH½¶®í‡"É­-?.–½±ììî¡èoïxùººª“º´åÕÍêªv‰+««²p‰ÍŠ«›åÕ7³›Íµ™µ×ó¼Èg]í»-ü}m+íÅÎí»…šžÛL„gQÌÞÂÈáÈ]ˆ¥CtÍþOÜyØã3›ÝA×ñ°>"RDß#4 šÙ ¨xõu³{ük$rüSXÈä@ аÀ± c4ø(f„²éxµÎÚ4ˆvݪð}AK7=ðÝÍ—WéÕܘ¤.„-8¾í‰ó¬0³[ÄõÈm›@Žm;[!¡;ø[&°B]Ïn ½iW"OÄ=ØóðwFÄÙ1cœh߯åí°Bº^¾ŠÂ šÚ$iæ€Z¢LgvÍ‹}OÈhZ|îYžT¶ÐI[šDk¤IÍ Ï7ÜSO̧&’¸ä=wõ•8“ζŸ$ÑÞe n“ ±Ñì¤ë‘x?·¶HªÒ€$èîznKªñ[Èë<÷²›eÙ¬ý65¾Š¸a°žv֬基-é~Ji8s Àº1hvHϙŌ{v…`>ª`ÎÄ'äÑò“u¿!É ?À{z€=•¸Ÿm$iBXiÝ#c5/óá&›¬ðjb£;¨ôÑ9lO,Ûpí”0=ò]Ë'jh£‘ž…àU.÷Øù3Ú>ÇY‹÷‹ž{œ¸&RÖ8~+ÂÑ&´Y“'užwü·ÃIe É2f¸[“Šâà±¥À.ù•Žã†ÜlˆÁ×f õû5ö,‰$^[f_÷0$AW%©id:ö it†ï k6lê]0B eÙŠ AïJTëÄ:¯HQ>0îN¦âðŽn ]°Ë‘æÏ FO3g^{ñdñ¹PmüpÇÒŠÛÕz–{1Á¾ËÍao¸L½(û2eÁÄù,wzŸ>Ú*B’{Œ®9öÒféP;Õï´£P:ÂG"{Š7 &m·›ºcý™$sFä­H?N’¯ÂÈjúHóºd½å,s^õHT €·lÙ€,ÖÔ{d3FÃ×Aq¯#iYòÐ-K2µci×oÓ"…·ÕY jÙ/>(Tâd£3ÔÈfö7"¸›šŒÄ B@“'øˆP͹Ò?}½U4mu©Ø±ÏÊV¨r¸ˆa³l;]l;S moI߬ÄlÔ]™²ž`›rorY£¡ŠjnÕ~³g!&ï«W0ð†û䌱©úAÞ‘WsÚÚÎ6êóT±ÏƒRÝ_ºŸ­l{Å~ìH¹®ÉÓ{_d‰ƒ ?à®’¼jõiçxS:ëyVÕ€Ö D° …~bY§ÆC:¦ Þß@óÏðûû×ѸގªòÚTå @:º±-Óõ=5ýHTÏÔ q'8#ðÎ6_Aç«7 ìMÃ~øÞNÉçä¶M1 ;Fq7U¸àŽöR×Û ?vêí­Û#?Á~l€É¸Â“E)áïcšÜÓËÀäçb ëi£lr¯‹p·ì|‡N‡˜ó;€*Ñ÷¤“ÄÍ.Ž ùT:æÀõÁ§ß3ì­÷sv~H&ÝoT`Zžk±ÒLŸ ¼#”‡^|À2Øþf½§nÝ_iž9ˆrg*~vÌ&ñç7CDþÞ(‡ÂÅw§ñ^b2éâ'µÅ‹O™÷G°;–ýÔ•×7ƒìþ4çâQ]q‘‡žµ1jý¯%ü8’qöÝu.³_„½z‹òd»€£ÿâG/s¼N`[ ÃB/ÝN@>£Ý@hõŒå&¾”}ðôû@ ©ü>~Šeföѵú^üÜËØK¡s þ)ÆÊžŒg‹Is7û-÷’b ”7ž/ž<|®@Ö1MÈ8f`$+æÃkX¤pˆ•U¼@Dœ2$Z^>ô5õ\° W‡g5«^øÝÉO9ðxßM0îEØwíl/þáµz»óAÆ?3”@Õ#™§|8:\h@½JýVž,€ÉÀxáKûGÑ­¹Dýb 3èø•ðûnB’ˆÌ$™µÃÓH[9w2zž¿—ÃÖczvÇwù˜8uáÇ9ÁØa¸0uÖ~;s¿ŸØ±Ž”JÅk´'¦&¾d,Ä$îÑûxܦu1ËE}d"2†’4Ü' ›ÜÔŒ‡/Ç”—?"æ,æ ]r$&z1…,žFÄX!wBLቻp;çàWÀ+ ä9{T(G>«¢PL¢ iòŠÞ}<Òi¬Cn¶"éí.Ž„à•S»È̃/ßòüalI]>ÃE¶˜a?jêpb¦wø÷lLìk§.)ÕÕþü’).øIÇIä&p$•Ëæ4áÏ'E]éø§òÜmx¸DP$eGDÞ¹DÐ2»æäU1†/BûLë¡ð‘÷ø®9Š_YàÉN’Ã&N4Fq΀|“ °öé¾OAÁ¥z{¹S&iæÙ} @‹Ö]²ÇDy%r I'ãÒQšrD‚Mœ« Y½ßAb¾2JÆng1Œ¥a4:éÞ•è¤ fÓ¼£`9Å& -³"P` J¯Ê¤vYÅTVuf’±|¤o§ššò*z†©ÅO0Õý²L%‰ÏAÒ«Ì]úñÙ:ÓäÑÅ&)*²× ‘Geb‚¼ÿfŠÑ Ž™L2/’$ÕQ;çì(ª–'²Ú·jFIJæwául´%jÃåÊeÝd÷É; ÿ»(“³|caä‹°D%Å ý6ÒÚ‡½$ ²4D«¹‚Z’ÜDéV¼¡ƒŽ§¨óœ>¢Ô…D²’|Ò0´äQOle8%¹ªÑ@Ͼ ö-¥?RŸ)ñF0l%¬²ò„—ÇiÁçͲŸ­¢)¹BédˆÂ)™RŠá"bN>ÙdÆ&Ž6"Qpˆ8™dM0#ÌâpO9¡lPgµ—%¬¹æAÚ× ƒêÀ. déðRïû¶KÎÆB”Ž%VóY$ =ƒ0AÐhøá,3ÅÇë@m \0Æù»×›°,û×SõI+†ØË‡ˆÓÏÒCç OCžA¼–H5pBÂ"¸ì’%èŸ ©xá2ËÉvk£Œ¸Õ¨"¾WyœJÖ. bC¸ou呞´‰-¼í俬l+.œ¸_«@¤Ðzç«z-Ï’2¢ J+JÌ›ŸSTH#0 qÆÙiRu¨’Áj’z„ٔﱆ“BÄ:Ž&³”é(axË…½]|Q–¤öÖM«ZŽ]¤åå‹‘ÍÄC'§¬Ci£¬‰)ð˜ØôWC6#¸¦alëMú¨û‹Àˆè‡³z4eåÃ¥3*ok&–vÑ3LèÚA²[úÙí cçsçmç½…w¢DÏí#UÜÜjùbÑj­v¯î•/íiø¨¥éŒj""p4gy0RÌÕõßy‡%ª-ªG÷DõÚÇx´i$ ‡<ÛaßLVPZ WºÖr{_¨½Ôò…ýX<ãC‚p•·ûjšŸÛ—0YÒ¤t+µGºc½T0›ýR.CQΰ•ITÒ¯|]Të«ÇðÍ! 讋¤Ù`…TŠs TÕW£³”Pââ Å9ìnöª£kî9©¦«ƒçá ªÛ{·td[–.Û07Âê›FaT|VùÊÃe_Ô•Qò<Ÿ@U&6>¹wpgQà¬ùù<üëÔ1dÁbÿ~¶¼¼-§¼@›”…{Ÿ£œÈJAðË&寧9p$¦+BR\[“Ÿž—pPú]}ð§›þ T> endobj 83 0 obj << /Font << /F8 9 0 R /F55 10 0 R /F57 17 0 R /F11 11 0 R /F10 12 0 R /F14 16 0 R /F7 14 0 R /F1 13 0 R /F13 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 88 0 obj << /Length 2872 /Filter /FlateDecode >> stream xÚÕkÛÆñ»ŵIP©±rùvëWÃ. €Ñªè'@)‰º“-‰Š(Ý#Eþ{çµ’«‡Ñ @?è¸Ü÷Î ï/Óß¼+nÊ ÌTv3]Þ”ygÅM–æŠÓ›éâæÃèñ~ê=þO’¨`xß}¥åÍøÇéwß¼‹"a®•e7!£ZÜ2ˆ{d”…J5Ä ð´ÞcV[Ï™ 8«ÞŽÂÀZŽkã‹"ˆã\#öY\j€Jpᱫ 3ÄgAª ½¥âL‚ÌùÆå¨BÂ=žq:,¨™²ìnjzU¹fW“ZWsÂ=oªðÅân&*Aé͸-SQ)ñHR Æ“,ÌG·@@\ä£Gœe"é½…á‰xÀã4Uk^šÛ%¸Ól¥Ú)JœÍV'-WªLF3v~ÚÐYøÖânŽÎ¢ï¤2Ôס'åÖí³ˆ€ÄÄ·ê‰Ò á®ÈN7r©(ù$} <öw{§Ò¿¾m¢%²<ÛAekW¸„=tÊiSÝ#T§Ñ2ºzÅÕÿââ’êFÓ£sšSÓöªîF0;•¶&7þ¤tžÔ/g%ß~¡ÛòÑqԷͲpt±ìö d©òùýe•ôio¥Ñp<-Ý‚Òå¢nŸá¶)üþ¿Âï­Å—B½¦ó^èüùó¥¥~si©ÿ#i!æ[iyD¤ÛŽÂi|ÓÎiž= 1nœ’­c;ºm-Ý•{$æMôkÝb7eGd²!Ç‹Oø˜ƒ.ø×"Ü뙋FŒSðóǑٛ-¢ÝfKÅe¬Û‚ÔÈ«‡sÁX‚§zvÈû+ ¸¶GTæz]ÙxíäÌøªÏ]Úƒf"§'ß7“¥‰£+ X މˆ´ÿEsý¸¯Â (”Ïš»êQ^õ c*ïZ1´QÝÒù¨³¾$< ¼OZC¦×†Ð7»í’óŽÚe-¾Š5ÕaíJÔ§ŒÚ …ä rDȯ¯!ëó\o%q@S;¤ôÔ.ú+þ~_¹ºà.—kZbè¿øhÓ퀶›Eô-øõÉÛår¤zÏ×~¶ŒÐ&>wøEêš0èý•NÚÿŒ‡s}cᎻ÷ì¿°û+‘оôœÏXÒ(Ó—keN顤oÄ=‡«Î_&Á™/1_ˆWÿà Èsÿ¹ÏÞªï‹M×á÷fÚ'Í¡³IE\Y9:!ቶ†ÓÆ”2LOÛÍ/î»Lôî÷B9^ñÁJ‹©§i,N˜Èã0bÎ|böˆÇs¿t uÚÝ}¢9_wvô½öн[«§å~Ó½Ø;©­A;_Íî{‹×ö˜Ælìkÿ´C5Nr°wrˆÃ©¯=}îxZ:ui/¹¨üéëK«Ï£îç¿«NsÉ^æFN}[Ñxg"ó¹ÃúÎacqÁó&œÐ³ÞM14¢³eÐs9ž-ó\° _ÑçÆã/­]F|µäu§œø ÛJãü×büÕ“ú‹ùàeޝÉì‹ÏSÂP<¬ÏÓy‘Ì÷ήX§^½O>… ʤ€Ë»„Ë»`‚"j¥¾x;}ñ_Ò~ endstream endobj 87 0 obj << /Type /Page /Contents 88 0 R /Resources 86 0 R /MediaBox [0 0 612 792] /Parent 62 0 R >> endobj 86 0 obj << /Font << /F8 9 0 R /F11 11 0 R /F55 10 0 R /F14 16 0 R >> /ProcSet [ /PDF /Text ] >> endobj 92 0 obj << /Length 2128 /Filter /FlateDecode >> stream xÚÝËnÛFðž¯Ò• ‘æ.ßA] âC9øR$>02õ¨%Q ¥ØJÑï¼v¹”([nQ íwvvvÞ3ý|óêâ*޹Ÿ':ÜLJ¥~’Gƒ$N}ƃ›»ÁÇáO#/Öáp2Rà žGx6ðàzÏGxÆð(xnGžŠS5üÞ=>¶…×RPñØ·ð|pŽi:v{óË xJùyܽv*×Ö¼üĬ¼ZÀëš_‘Îx2x ¹ùȆʵîÒî2Eç–'t¢«wtƒ2÷6ªáù"Bô[’ˆ(¢é4b°GfWÇwÂcò;w”Ætù~NÕá)U?gÚÅß5­§?ËŒ¾õÞRE½¦ÿ†>ˆ"bŠ-ƒ+ѱ, þ¹È„Í‹¯tá/I;ø»Fj{^ÌÔg sk®ª¯,‡¢Gx]³¨³}»I=QÞ.¤FË–·Š_¤ê†üÄKTä¸@`ëØ§@…eÍTiùïÑ )X¥¨`PQQ ŸY´¯ ¦’áÖ\†°Úp]ó&™ 7¸ý‡ÕÃO*ŒÐ÷Š ê⾋huÎKª¼›{§Ê¿F"ØY c—µ$Po{{cÖ¢n˜tÙØ®¥ØÒíê;þµézV“ƒ7ì<Œ¼!¿. ¥GaVQ^b€=T~Ì/uÉ›ŽÓ6 !FÍtSÁœVBtb€”¿Äâc†¡2wÛcÜN8 HV"¸m*â“­ Þù2Šì…‰t€RËŠ7‘6Ùßé$ñb—¸Ñè9=Z"NªÝ†5Š‚á{‰y¢%¤£¡cCTKñ«†×sÊ­nÑœGàgiÓØÈšŒ60|v+<°CœÂÈ-‘ 8\é­“hI9Øô¿»æãŸÅšâàL–ñuW¶DªÓä×£:Š©‚/-w¥6a% “aˆ“ͶœÏ˜4¡¸áK’äkÁQËnA=ÎÄ´ú®·°;ò©à6E—â& û}HÈŬ$sSAkuKò[ ¬˜‚oYmqÁJÇ_ãËø>!7#ZãauÁ½“~N)ó°ZIÃ{¢'ø\1ËÖ‹ÐUu›SñEß”›+<Ãa¸)0v«i…¬™oКÑ1¬¤u´c¾„Û¯vñ廂ÈÚ#2jÂuC½L(6˜ó¢`¤†aGÍV¨©Í‚ª+ÇP$ªÇS4^°]HÉp“ì×|¦aüRMC—1êÎ6‚uo$ ^ͱõ 2V”Ù ž“ŒBàIÑ”§¾y¹%1ÕÏ ñ¨éGŒ[”¦þML° £§±A´Q7•¤•€3·š4’r~‘æaÉJøKé±lýÍÜv²3iY±9õÖ)èq’ ß.Ù¢n4·Ò£šw什`ëÁ]™¸%‚Ú6FêyЪ߉ÇÝÊ´ppb[ñ–Ä/È£+®ÅTûMÒPlDLLª d{„UˆefÙéH`ŸB¼¶™¬‡9Ä®¶-u á~fLSK‘ “áÕ(‹»5cUcò}n»¥b¶fðéÏÓ™¤ýŒû¥ÛìàæL¨¸Y–3ö‘cÙdYAq³”©9öl1%;ã»k†W2h"Φ&*à ·y3ˆãjãÔkŠA©¹Ü±»š’<ÝTœÏ)æ¤aYÈ<‘bwÆÒ'xq4ü q'Û¶‡¨¹y‘P§r1ºÖÀ”pD]-„WþX‘}ºXÀº¡R¾;,Rn«¯¹Åôùcª;mv?¦Ò£9–ù\–¯û•ÌGK“îlÀ ç ýC'wD`ÌTŠNMã.žÞõÍR{Ø å7–ß´eÿö˜ðò´Ü®üzàO²Xû:8hÌHtÿ´>Çýú·£à½HÌ5‡Í»ær?ÑX‚îý¾Ç#b_‡fÙCöõ3çunÆíüïAÇe×V{Q§1é¥iÑåÓgÖ]Ÿš‰ŒÑtì‡AÚu·GV#þz¸”©Ø¯G£¡Î0mâØ‡ƒI—;æ¶Å®ýoDâ× ÓßÁsíŽåˆïŽot£Ä$Й͊G#9a؇]¿uFˆW'&ho;‘aÓÁDñúEØÏÜ‘½è»cé(ñó,ì*ìŒ`z2HÏôç#ò ·ÿÜaò8Ê´Žü(Íδõ‰HUgSh½Pg©¯¢_®6 ä7‘ßìÿœ«ó³ øÏ&ëÿp^íÎUøi£ ±'bUE“æWïo^ý Ø‘ « endstream endobj 91 0 obj << /Type /Page /Contents 92 0 R /Resources 90 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 90 0 obj << /Font << /F55 10 0 R /F8 9 0 R /F59 31 0 R >> /ProcSet [ /PDF /Text ] >> endobj 96 0 obj << /Length 2069 /Filter /FlateDecode >> stream xÚµXYo$5~ϯñB`¼íÛ-ÄWPöiy‚E¢wf’2G˜cÃòë©ËÝî™N²P”¤}•«Ê_}U¶žÔð£'MT1¤‰mjU;7™¯/þ¸PM0ÖфⓆ´,“ŽWWë8ù~{ñüä¡™ˆœ2¿½¾xu™& Š “뛉ÖFEm&6&U§fr½˜üZ]m¦3Mu˜êê=ü.¹9ß®¡ñ0™]š†ïàwsËã;lïáÏ=üîeM»ï<ù~€o\°ü0õ¡jWø¹?s¿ï# ŸãN¸Í–´ ­Èn¥ƒ–"pÓúQÊ{×ܾÇ-¶S0à1/Þtûíúõ«K Í´S xo¦µj¼gÍE»5þÁí[RñïlÁ¢;2=² N²Ä›J¶(Ï qÊêgF¢ÉpZœý †Ó­‹EÞª€˜™ ¬ ¼ô»é ,¹’Ùõ$¨&ÚDÇl”SžiåÇ“ïÏCk3)æ¼­}=²·U)º¬ïá\ŽU¾Iydh„Lª¾Q%ü¢Ò>M4¬šçÿ2¢>-[• v°k,&Íœrànj¥£œÓ˜ö hàocÖÐø‘M½Jÿ“ËŽçr¼Šµ¸l\†Ïs~>—TH¥ŒOÕC»E\/¤®Ï÷ZqéŸèº1X» ´ñuu²88ÍÚ¨¦iáï>oŒljÄ„éE¾t¬þÉc2 ͤ˜sÇ”fu*Y›Ä€®ø±XfÊ’Ñq³:V7ÀKô±= —¿ŽY¥ƒU¶?ØQmCè¤Eecreêø ~•¼ÎãOLôùaj±ïŠMÙ£)»Â[™Ùih.C¸¬p€nLõV[—¹n-ƒ‡-£À8à9c†ì &‚˜È*³Æ³R¨¶| Êûl «ð»qÕÜ^TïÓ óÚØnjS1u.ysÎ9@¾M€I@Η~æü¢ r !˜”P5@ăˆ’Ú‚»ßæÓ§ÎÙk„×{Àž€(Úçd½¦LÅGiḃL“þää¼ääÞ¡³¶c¡çPx΃L4{ j÷OHê“ò 3`$Ù(X Cøw¡ÅÎÛ÷'\$<¢SÛfk„Œm2']cÉ‚ë)Ï¿# -W]OÐ%d¯QØq•ýËN¬Á[_ H¡ƒ1±Tr»‘HÃïþŸë&ü^“²Ò8 ÷ÜPsÈ•C>â^Þ¾P£™y|90Žc´aÇÎrÂòO1qNÒ¿Ä ÊUû»‚ûpêã” EøŽ¤%bÂŒ³Ž¹@„Y¿çÐáÓÏyºÐ}ÿN‘ûZLç–§ìùÿÑÄ!gŒ9¶8:n—pެ…ìŒJ|Ü0=cp­éDF\ƒëŽ ÷b„æwWø?ÉižW’Á@Véx`³A¶N*5])ز<*6רÖƒ9¨FûW»#ç ‚µ^ÁÆêršlEéæ\é[cfrêb,‹x3ê§øé°]qŸ­&¿¼ð‚Ì9+¡ÌˆB#´îÈlEœjpt:|P˜ûaíhú|ÅÆú¶™/G¨Ýw4…äÚÎÑÜ{Âøíò+PÊ51Î1ŠRØ è% ßn„±[®ôMú怖v»¿Ç†ô.²TpMΜЦ¨Ø‹Wœ§3òï˜_f=äkÓöa'(h‹qŽÇ·µv%Ík=€¯Cײå’ós¾Œí‘=²Ÿˆ·¡êÐ'1&vy‚ÏùJ8‚ºùb‡ržÿØîж?ˆÀ91˱È<ØÝù8Ëèõê ¬e¦Á¶0s%.Ëô×kAßÇD -'jÊ$t; €|Í¢2‚ì3ÕzK÷aZžÏP1î HfoD1Ë0¡ØjŒêònƒ™ý¸k§ µ>Eɶ \Ã5JPÊ Ü‘s hÜÚ¡)Hâ´5¸²×reG­ï±­‹ëz]\×aà2ãr~$º@/|äᡨ‹:(¸ˆÌ÷,›îî5rN‡½2sá¼bL‡øÅ›mï‹lAöX\OSžÍn Ú¨€§PWßÈ@—­"(IÛ>Ê3§ü\öòÛä<_‡Á‹•¯Hvàw’ÎøÌ ¯í®x8¡ÊaÕîxÊ\vÀ< n5þ*ƒÑNÏ+C ±ÊË>W J°—½­älG¤s:OŸ*)¸1¤.RƒO>˜¸dÅU`ôYZæ.uW¨æjϳ2t+d·2ÊÔÌM(m'ŽXäÏ\ŠÀô“|Å•`‚béçL+;9Y˜üx7Í8+U!²eß <8ú¬3î{5²ë®±—#Š qù³wØþiÂÿç¯9Éw¯9'Ï9¶6/ܗ¿¾/ýo–}‰œž(9S §Òà='‚¸…jªƒ862d·G¼O$NV‘+wË¥h7M–,2çæë6Þ ^ØÊN7¹¾ŸŸtÉpXÐA“Ì5kºPÖZÅpK"òÍ‚r¬攀ޭžxl‰N¹¾’º–¶8F¨Ï¥p×Ý1Õvìö›àcVÌ…”†}e|5ö‚¦ûâ±Ìi_¬óÛŸ‘°Êó ô¼òµþ„·/wâ˜á;KÐŒ/ÆmÕÞ½`«ùD[ÏŸøœ#»‡ 0V]Ô/„ÞE#rO=ê15½y@ÁÞ4P#I,,¦ùh”#H¹RÀâú›\¤ýä³…†Xð/1m|ùµjÓ÷ܶ•Èk<=ÒCžPQKÍY/~¸¾øJ£ endstream endobj 95 0 obj << /Type /Page /Contents 96 0 R /Resources 94 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 89 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-mprev3.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 97 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 98 0 R>> /ExtGState << >>/ColorSpace << /sRGB 99 0 R >>>> /Length 9047 /Filter /FlateDecode >> stream xœµOï$·q†ïû)úhòs7ÿ3GqìH@>ŽÄXÙ° _?¬z«È·Fkiw­´»ºfº§§›ÍzXä<×/¯çúÓõ×w¿¹þz•ù–nûó©ù­ÝWÊUè›?^ÿvýùÝO¿ý×þÙõó/ßÝo÷}_üç—?ÿ—wÏ[½¯ÿ}÷Ûß]÷õïžë—ë¿?½{$àúÕ»ZÞÆ¼j›Ïõõ»ºÞ7½ªíªã­Tмµ·”…zÕ~Õ)ï©$‘눇PK… ‘£‚ÖžÚýVô]†D.šTÇÕ{ÝÐÈG^½h½[F™ìÈæÛL›Ö»¥&Ôê¼Z~Ë4ó¡G#­M„È©”ÞR¿Z‘cP’ÈuhÅlÊéÔ¦YùœDm}†úV&H"ëÛoeý&"æú ííÉBM#›‰öÖ²P ‰ìo|cëï|3µ"T2h®Ï0Þn=–ñ–&‰<4‡P×Ï0-Ò©­Ï0-rjäÄúý–"ÖÕ°é ‘ë;j‡’\}+þ) œ…ô:[Ôú¡l‘Në;&’Èu-Èùìåí~„ô{_$‘‡Ú8T5ríg‚r½d?´þï"ýz{»“ÐȠܘ êowf’H½2”ú}hhd•?•r*F鴾”‡Ð( ¾>C“ûàý»±ZˆÊ$‘ë y@ë¾ßôhä¡<…jIäÀþÆúŽÖgXûé ‰<ÔóµÞ9id–ÈEz>•›©—k½ó£{(r–‰ÊÃÄ‘ëüŒkíÇY52I‹ ´ZˆMM#³ü©´Ú'"‰têrE­Ï¹ïÒåÚ_Ç€ý ¹»ˆ$òÐ:ëµES#ÛÛ¸A¥0!RiÞòz"Š\´îôMë$¦u-®ãËÔ硤‘z×+•Æ„HP–vfh;­$‘‡ÖžÖÑê}4‹F*]Hï¸E«íÚTåþ%*ãšöýMmå65ù{ÑH ‰Lh1­ÿ»©kä¡õ7‘Df‹µþ‚ëarp}[3`ÍCp‘ÆBmܼq}´ƒÚYgRŸ,‚µ”à–ê¹›\ë•n¸Ú Æ1»ôGCp—¾‹œ”„ãxõ_k|ÂôÜÛ‰W¿åð£m‘œ½¹…WïC>½6³Â«OB¬;’¬Ý áu/ÖøÍIî÷À1>IO†8¿Äk;8Æi»·;ò”Ïç_úS¥·¢'½k<ñ,Ìí%¾Iÿ&pŒïÒÇ ã»ôsäkÒ[óy†ô_7ù|NáÙ˜'â‰[ŽâÓ-½žÀ!>=º}}mÚH ¯ÞNàÕ'éñèרŒWO&ðêõç—íYú:ÄEz;W'pŒ¯Òç ã›|ñs‰Ü:s—>MàߥD<^xJøññæ[nœÀ¹Eî|¼ùy‰¤Ÿ#—±>,„5~s’þ q~á"ý–À]Îw—Žp•þ q“>IàrGî™yÈ?÷Ê<¥_q¸ÜÒ—üº½óþËóÿHŸ"ðê+Èevº$éW^ýâü^M8q•^±6Ä\å¶ò¼>\oiX¯;ðz?ò4<øýW2\ÃûWyaSLrYh7ë‘Üða~ôyp8½pÑöûðÐö•YÚ»uZµ û¬ÌDÚÇÍ+;öióÐ\Þ¶èù“¾d%žÚ£Þü¥è€Êöo¹Ä~ Û‡KøúzVÏa] À÷×—/›²oEŠã[öVtÜ|«ÑÞjÏ–½gó÷ÚÞnx¶‹‹Hg»áÙ.9ÖÙ½ãÙÎõ|ò½ýg_Ù‰ý÷8±ßþáÏï~ú‹ui^_ýç…syþWÍ*¤E¼¯¯¾¾~rq}õ§wÿô•¾ó÷¿´ëõp^Z?þ¥CÍÇyéó »Õçòó ;–¶|ðA§OÙsÑ1½øSö¼2õ›?sþ”=O½åéÅ{ϸ4L„Ù…’Væ8q¼l·ÎVé9ö³Ù¶¯²Òv m_II¾ÏvàÙž² ßnHÛq¸{û9z>3t:þÎÒzø­„m5Ù« Õ xýÏØï}ùÔ޽üƒ_ê÷¾üÉš4àõ¾–¿ÿõ’¿wý§ï?IÇ_ÿá+úû_¿någ¿þåÒ²¶ç]’¼Ìñ½cq!ì_e@ÖG]ÕéÑ.]ZY÷½}ðßÝ–I"±ZÏl7Éoòï_¬‹õ¾~òÇßóí¿»¾úåGž•õOùVV¿÷ÆYùõ7ù¯oþøí·ÿý—?_ÿó—ë×?ÿ•Ÿ£Ót_Eþ±n®ûïpÞ­\ä¼ApÞÇrÃdçv‘ó†wçí´"År“ó~äJÙÎwª;ï„„ÈœwBælÎ;™‡óÖívÞFæ¼ ì£9ïócÎÛé±H'3ÙF©Úí¼«¤çÛy7t¬Ìy;Áyç}HL¶œ÷!1Ù]r˜í¼Õrnç}‘Nb²‡ôÍ·óQ0ç=‚óž0Äæ¼‰ÉVãλ›‹„ó&j…iµxb©Õ±O‹tZù]×,Ò7œµ;o"D:­ïxÓ"õïMÉ:¹_Ûyg8vsÞNpÞ‡ÄdBd1¯-&»@‚˜ó>$ÎÛ©Z¤^æ¼+d‹9o5Ûyë7µ÷!D‚º™l'‰lfÕá¼à¼µÏ¼·¥>æ¼à¼Ádë5aÎ{à*0ç}H"ÆdÌy;=é“Òµ÷ÄUç-ñírçM$ÎûÁÕjÎ[mìvÞ‡Äd'Ɉ¶ó>“íđYn‰Ìæô«™l;omm¶óvB$È·œ÷ Î[[Åí¼à¼ÄdWŒnlç‚óv2;œ·Enç‚óvçÝ0‚hÎÛ)YäÌ×vÞ‡ ‚óÍyçÝånÚÎû8ï!WâvÞNpއʸpvÜyoRç=¬Í‚ó–à¹Üyoêé´2m"qÞ7îTsÞ‡Äyçíç}Hœ÷!ÑØ·«ÖÛB‘y›ó&D°!œ÷{Þ7¡8ï)Ãy?Òî¾ßÎ{#œ7¡hl çmM«;oBÑØἓ©TsÞ„œ0zìÎ;Iãù~;oÂ1á¼ C°:ïu¥™òWåÍX[@Ûk§mÝ8ÝS[ÿ÷.»g±×}PU÷B½¸Ít3"ØP cVË}0Ç`uÜŒ!X 7ãzÜ0®{vnÝ­ÎAFK’¡®NÌÁƒµ“‚Ul3†`ÕÚ FÔj3®ftáì†ëÂ>8-xcË9>›‘ƒa³W+¡m•ÉlÆÙ Ue/ÄAÂd3®®ÌÁ·ê£à ZlÆÖ†`UØŒ!X³Æ\¶N¨öš1«»>8"ª¹fltðÖŒ¹ìtÖŒ«Ÿ²ZàÛv¿ ÕÌÕW3®n†4íxgUfUV3–;àêxTSÍØ+¡zêÐÔŒ/[;펚±ä€]ÏÆ€0VAÍX a~Åõ¬?¨rú ºéƒª¦åQ§æfz#Ä4£< åq¹Q­4ã w†“>¨JšqðŽTHË#WïPøèƒª£¦ˆ*£ª‹fœúÎh a¢7BDo„‡žÕ?¦¡7ÂB;~¹²ì®2û¸ª²aʰ‡eµÈ°‡e¼»ªLsã]UfdUe–o£ª,[å˜U••‹«ÊòÅUe–E£ª¬S†m¹Õ®*Ë–S£ªL;»ª ¡W•¼ªÌ UeNˆ”a[EÈ®*ë”a;yU™ªÊœPUÖ)Ã>„Z±l„ª2íÅïª2'D:IÞ\ÑÞUeÚÓÛUeÕjÌP†*6d؇PUöŒkW•ÙýiUe‡鄪2½BvU™“ÔŠ3T•!«õª2'ª?ÛUe 'DzU™gÑ·Z‚nud¨*Ó|mW•¼ªÌ UeNˆÔãÜUežo£ªÌ Ue%T•y¾-v‚£° ;áìîª2Ô­yU™ªÊ@^Uæ„HÍU-ÃvB†]б »ÀºX†í„ ûªÊnʰ«dJ;Ã>$‘v-Y†íôxý™‘dØ &Î2쎳dUeÖœYUÙ¡õ “ãÓú¬lõgz‘a­¼¹ë}î6Ñz>q$2ì>Á2l«´ Û öƒ®€e؇ B†}Hòf¯[C†¹Z†}‘N’7;×Órqm-Ã>„H%˰Q¤eØNȰI†]ñp· Û v Ue5T•ÕPUÖ¬>+[.î„ [ nv†í$v·ú3dØNȰY†}_œa+y†,ßF†=,ßF† ò Û ¶ÓЬTs)˰•É4ò!d؇êÍ$I³‘gØ—U!Ã&D°!2ìäÙ92lÂQ S F†`ù<ÃÞˆ ›P’fÂŒ ;ãaá6áè„Ȱ‹ –[†M84=8ȰM7z†M8!2lÂŒ »ûyF†MX[@wœ«i[7Îû eØC®©÷;Ã6ièöFdØϰ lˆ ›0#ÃÞ˜c02lÂŒ ›°Ýµ‚ÌŠ§wÙãWVAF¬a›ÛK¼UÇx« #ŽñVA–ð˜ØdÄZA–Ñ@î ²ÍVAFÜräïdÄ!Þ*Èîb×–U1ÏÎŒ ²uq£gndÌs0ç—í¨ ;Œ 2æÖ"ÇxT1ÇxT1çY*È£‚Œ9Æ£‚ìðxaT17>^« cÎ-rçãµ 2ælŒÓ¸ß̨ ;œ_dÌŠ7åäĨ c.wäž™QAÆ,d‡QA¶Ù*Ș_·wÞ¿U1—¹£Â*Ë*ȘKaÎßa© ;Œ ²Ã¨ ;Œ ²Ç˜¼‚l³U1K³TFóà÷· ²Ã¨ ca¨ {¬Þ+È£‚ìpzaTF³´wr›«â± ²ÍVA¶Ù*Èä2ÔëÏ*È6[™ó—sˆüã†ÄÓÅCâ6ñ »ÿ·mHØ[HØÛIØ%ECâ·¥èï6) iø†Äµ8b‰÷|Ñx Câ…†Ä3žÄ{HÜ·aHäCâNHÃ0$Žu×cÙCâ7Msò!q' ‰;aHü¦i`‡$ o;}Ǹvw÷¸†Ä0¹+[RŽHmó,aïò Ù û°!j$ì‡0¹ ÃóHØ'‰aðœ†Ä›Má°„ý t߸ŸlHüF‚bCâ‡é$CâNOˆÄ¸†ÄíZ²„ý‘ϼv§l‘NHé)ž%ì6eÄöC’°;!a·{XÓÀZHØ1}Ìv'D‚¤ ;Ñzª­#{¸õ¶ÐáXŸ€6Ô„]®ig\­éÁƒ5a_¸ó÷Ñ5ag¬9`Ö„}ØãÏvÆÑ «WšƈàŽì^öAù{­Ç Ô„1k¾îÜD ;cmœq®¦mݸº,‘°/¬4$>ª• a?¨ »Ô³Ðä/FšúÅ‚5a?˜c°&ìŒ!XvÆvœú‰&Íø’fÎÓ÷vœ…°Å`ÍÕC°fêŒ!Xóô…Lˆ¦éŒM«q0$ýà´à-ä`$茌ô|Ø3ʲsÆÙ 57—Š¡bØjÀ9sܪyùAMË[ ‚5'g Áš‘3æ°uBMÇC°&ãGDMÅ$qÆÜv:HdáŒY²ÑøAMÁ戚€3v=±¨é÷AíV3–;`Ï„šz3öJ¨‰÷FäÝŒ/[;íI7cÉm‚7ÍÙb´)Þ4c‹Ñ¦|Ó|­ƒÍÄú ÙZ‚š*"ÕÞˆL›±Ì€#jšÍ8è‘dÔ›qðŽ4Ãöj;K°j~}0EÔ'ÂAM®Ñ@MLÖšÖš‘YoDb-ˆòn—iµãÇ-¼Ò4ÇMëܬvâkm‚7ê ¥¾‡½*Ïð* ¼êÓ áíÝRïüÁc§½ ø£Ì‚ÅOë \w«N8øÅW–ù_ÿþýÿñj÷'|ï›­»oö“_ù{|ô,!{ i¥?x<ã3ŽçIñxÆÍ3R«?0õáã¢þ®¥€’¬r$P’EPŽ dk8Z90ÙÜ]ó@~G›xÁÜŒm$|.%ØÖÞÇ“‘+¦®Ñf‹VÇ©2¥«!«/¬Z;0š¡8¡º¶‰ÔüwêŽaNÙ u–:c»JpÂ2'i¥šI+êŸfØÎb-‚³jKÄØž€«IÚ(료€ ¸ãÔ\³ˆ„<Á§fá:Èœ†¶€ÝfN6Ía5Wœ:rŽ/e%*ù(=§ ¡àÔÌ.ã­Šöœ[ 85eÓ0ihy cÓ¾~›†S;ÊØ´#EØjÀ9 { îÚ‘Ú¨S¨ÖÿÄW¶0ëÚ&£¶N¨s€s Øt}óÊÀošÕ\V0—€mêø®¤ÔÝ0k5¬¦v‚M«aqÚ¥ÞX2‡cnÛ$ÌZˆH˜1Tù¢ÑP×´ ±j˜u¨²Ý†ý&¬",óÐAÁÚŒ;éªÞ³aЄXã3ê0„{b†ÔfÖA•Q!¬ñVËŸæðA;ÿÃa6ë4}¿Z×ë´þ9’p¹#w,{ m[^ìî‘Ë9Æ?bw$›º5ÞÒa]`sB|G–°-°Ù–0ξ Àf]ÀÆ™…uÙ€ œ×‰ÓAb'Öe Îg^™».`9¤°.@¬ñ››:0Çø†e6wi½ë2špŸÌñVë(\Zdß<±,)ha'Ö¹§TP^†Ä3¯7ÒDä6ó#®*ð:ñ’|è ³°ÆoÖr.™ëú8kü8ÛeëpF<± z1K|±“,£U-²Ä3Ë Øá*MàzG…¹!ÞÖÆ–A4f߬%’hÍ„°ÆË Ûáñ?äF—ôE‡y„GcÖgTàš#‡øtcöõ@û#¬ñÄ£3?èoY=©° 2K|µerJ:(X *f…k¼ÀÄY—-bŽñY—-ª¹I^7¦ÆË #³ÆO äT±X–9:¬cx’[éSEX–9’ù£·ñ|˜u$ï‘UeÙ"fßëØ^`):a–"•6ð¼Èrcw]GŸ?ë_`„=œ^â“0Çø¬E*Ì1^Güžn…N9Ôe–"H–¢“ÃñÄRtÂãUwŽñÛ­FRXŠT˜¥Hå°Žý=^(,E(ÌR¤rx¾l×ê¡ÃEÇýKÑ sŒ´H…9Æë``):a–"•ÃY‹T˜c¼|âòššnáx›0繇ãí/ñº(Á#Фküf-—'ž‘«Žà–þ’ ðh{Uu¤ŽXGç—;²ôŸë@`韮Ú9Ü´¿Âüº½‡ý÷—x] ,ýi5›¹ê[`éžßaéln:*F¬;"Öñ®äË’å¦#WÄ:"xý#ðz^kÎx„÷×i›‡»Ž ¼¿®/)Ù²w¹ëP q—çñˆÿ·ÚQZÓÏ]îHréÏ]ïHrÏ^ðH:>Ÿ½à‘´êŸ½àQFAÎç-x”Q…ðy Œ³~hÁ#³‰¾¤‘!-ôPi $Ú¾W=²ZöÈ#|Ý#8 y„¯|ägé#?._ûÈ"hñ£xä'â“–?Jõþ»Ö?²×öHöúÏ_Éßà³—@²7øü5ü >¼R×_¯&jð*H†{Ù£ýFŽë ÙêGØùÿËJH’kü(+!¥Ô>i)¤MõÿÐZH+ý[»XújÊË–_<þƒz—þ ø»ô'¬:‰ñÎבþë\¸ôG;±¥ÿ@ÆíÒ¢CbÒ¿ßè/›ô—^‘o‰P¥7YãÒ¿˜õ6é¿ÒŸ£73ºôïH“]úÛ/$¸ôßéOØž€³4éO(õ· 6@ú{M—IÆ–N=ö@UcØ3Û1CúK’oä»JƦ/•¥‡qéOØJÀ©S°p™™ô'l:µÊ†TúÏÛü.¤?c«ç$ì1X¥ÿAm-fò£Ré?³¨ô?¨ÒŸ1—€M'$u×úX} —ëHB,7`é? ¼\úøI—þud&ý­~l[ÿÀ:GbsÆœâl«^÷k›â‚x¿&àþµ2,][þ«ý¬:¿@OnýoS>·þ'ÎXú ús÷9›UçwÔæný?}¸úÿI¨¢týÏܱа „˜þÏ><`úŸXu~ÁðÃÖÿÄå‰ãMÿÛ*Û¤ÿ¡ƒŽþ7Þú_ BHÿo6ýo¼õÿfÓÿ¦ï·þ‡®<ú³éÿÍsšë}ÕÿVI¿õ?±Æo6ýOãMÿo6ýO¬ú¿`dyëÿÍÇ  #]ÿ÷É<}Ub­ŠÞúŸXŸÌ …Š®ÿ™Eÿ§Ž¢@×ÿ‡¡ÿ™Eç§aß§éÿÃÐÿiºN7ýÛõâú³ébÕÿÄŸ0"îúŸYã‰Eÿ†þgýÏ< sóøñ×'²Å×û¢ó³ëMÓÿÌ¢ÿ—xèÿ\ÐqpýúŸYt>sˆ7ýŸ †î]ÿ3‹Î? ý¿:_Þ1ýϬñ =U×ÿ¹cÎëfÑÿ‡¡ÿ™c<ô¶%\ÿ3ë*ÕÄ?Mšþgýú?Ûϰ¸þ/¶tëÿÃÐÿå1lúŸYã7Cÿ3ÇxèÿÃó%úŸ9Ä›þgnwäÉÃ]¦ÿÅ Wc'¶U¼ÓK<ô?sŒ‡þgŽñÐÿ%aŽëfÑÿ%[ûkúÿ0ô?³è|æýÏã¡ÿ‹Í¬výÏ,úÿ0ô±e]ÿ3‹þ?<_¶Cÿo6ýÏ,:Ÿ9ÆCÿ3Çxèæ\"·Î ýÏã¡ÿ—†þgnáx¡ÿ™s‹ÜÃñö—xèÿâÃǦÿCÿž‘Mÿ3w §bxÖôÿaèfé1÷Ì ýÏ,ý“ÃÐÿ‡¡ÿ™_·÷°ÿþýÏ,ýbÕ˜®ÿ™¥ÿqx~‡¥?²ÙôÿaèÿÃÜm8úŸXóÁÀ:üN¬Ãõ›;†ë‰Gxÿ‰árgèÿÀƒ÷ý/Ã×Z3dúŸXõ?ñˆ ýOŒ_‘ažxœèbÌ¹ß ý¯õÅrþLÿ†þßüq L}t­â}±¶0ܵŠõ:ÚÂó£]«ø°¶ ´ZE/]D­â`mQ½°Ú¢úQy­¢Õ9z­bê×ÖÍK®v­¢¡×*f·¨Uì¬-6z­bgmAˆZÅÇÅ„˜ˆ„Éj§VÑÜŠ×*z­âFÔ*nD­"p×*nlº°é¯U´zC¯U܈ZŨU´ê<Ô*v{­âÀÜÍS«˜‘ä{­âFÔ*ZÉ$´Åã•Є­D­âƒy­âÆV÷.¯Uô<Øj ­°ÑqNƒQ«¸µŠÍ.3«UìHh¼Vqã´ÂƹDacåZÅÛîÓ„m>Vظ=E>+•º¶ÈGKܺr'Êx­V‘…†¨U$Ä ‹7‹ƒÅVXœåÚ¾ÂWþóZŨU$̺>Ÿåúê*¦Mã4UÁ˜±úHž™ŠÍÝ 7g¬>’ƒ©(Ö–x¡bwï3ÜllÖÕ lu7+Áù:¦‚Ỹ$´Èla*˜m5‰Í1ÞLÅs ½°±÷Êl¦"{¼™ bßl¦‚¸`5†™ŒÕTØBPÛTk<±ý¾QÂñ›©xp§oSA¬ñ›ÍTÇx3›ÍT«©°‡Á6›‡›6ŒÕ<küf3ÅÏ™ b3÷‹©Øì¦"»©Sᬦ–Mܦb³*>–‰º©Hn:’›c3Äf6dò(4sSA\fä‘™ÍT×;²š‡Íf*šƺ© ÖøÍf*:Zƒm*ˆGe/ñf*l©Ìm*6›© ®9rˆ÷BE³Õ»P‘xtf˜Šâ¦ÆLó@áf…)€©(fÔMóÌ0Ì1¦¢ ?~˜ fëinÖx3Ûn*˜çÍ SQ½0×LEõöÈLÅa˜Šê&ÖLó|˜a*˜cë¦â0LÅáÙLÅa˜ æ‰BϦâ0LÅf+T|,#÷BÅÍV¨èü·WÓú袊|±0ô¢ŠÂ3)íG„NQÅãe(ªØ(ÁÙ’»]Ta³½¨âö‰•"úqSWVFúfvbbÑ)ª¸Ý] ¨âö¹’I•™¯Ù‰ìvb#ìa»Nü”½û‡¤‹ —Mع¼Ù‰Ó*06Š œõ Ù †Ÿ÷a;1}r$ìaK§®ðS¼Š»Æ¢éú8¦’U`4÷ùØtd¾f'ì'¹ÜN¶kP$/›À‚±$Ť¢ŠÑ£èÑNôh'z´=Ú‰íÄôšUÜH¼¨b#Š*!66¬‡‹v"Å¢ Â6}&e‹3)s¾Ž°A¾3“Òj|&åF̤4ô™”a', w;a;ѽ:vbØ´Z³a'a'¶¬€€Ûvb£M»´Ã2;±Ùìq¦_OßvÂç0»°ÞÔ¶Äj'ŠÙ‰ê3YÍNw¬ý9y%sy"ÇxØ ¯q;qw£¹8 ;å€Þo;ÁÜ+3ì³®…i?½ävb%áû‡`¶µ37ûÚ™l'°Ðûm'˜-~rsŒo¾¶æä: æ[S¸Žâ0ìÄ“ÜþÀN0[üð: ]›3[6ov‚y nf`ÚÝíñ› êf`kÌN†`.˜6›|Z¥Æo†xhZeAÙ؉ð̺v(±Å[ì³ÅoÖµD7ÃN0‹`Ûpvâ±UÝN0 ;!u>ÙØâ7Ê<^âa'žíÄaØ æš#‡x³Ï®;¸=~³Ø†ÃÇwžFɬñ».vbqñi•b˜ÅN†`Žñ°ë°n®£`–Þ%³Åc‹Ù f±‡a'Ò'¹Û ùutžFyv"Ùº`ž3ìsŒ‡8<_âa'˜C¼Ù ævG;‘ìÇ ÝN$ûE4·̳0§—xØ æ;Áãa'ä²ÑöÂì³Ø‰´§ÂN®¾–îf± Ì1v‚9Æwßž¹Ž‚yvfØ \Öï·`;qx¾l‡Ølv‚¹µÈ1v‚9ÆÃN0ç¹ufØ æ;q¸¼0ìs Ç ;Áœ[䎷¿ÄÃN,vâ0ìÄáÙìsÇùžl'ÃN0—;rḬ̈̽2ÃN†`~ÝÞÃþûK<ì³ÕqÞ\GÁluœ7×Q0[]çÍu‡a'gg-Ÿw;qv‚¹ÌÈ#1ÃN0ðþ°›ÍN0ÞŸÙ > endobj 98 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 100 0 R >> endobj 99 0 obj [/ICCBased 101 0 R] endobj 100 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 101 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 94 0 obj << /Font << /F8 9 0 R /F57 17 0 R /F11 11 0 R /F10 12 0 R /F1 13 0 R /F7 14 0 R /F14 16 0 R /F55 10 0 R /F48 8 0 R /F13 15 0 R >> /XObject << /Im7 89 0 R >> /ProcSet [ /PDF /Text ] >> endobj 104 0 obj << /Length 3013 /Filter /FlateDecode >> stream xÚÍk#·í{~…‘¸Ùb=iÞM·@sÈ.èA±E äòaÖ?rkë±o÷úøïåS#Ù³K4v-Q)‘IQóÍÍ_]W“:® [Ln“ºŒË¢šyÛ4ŸÜÌ'?DýñÂD·S[Dó54ûüÛÃvnqð n{1MÓ4Z@áÛ™ˆ‰â‹i–¥Ñ[Ájè×⯉š»nÉ`Äï†æ þZîöH~/òf2™¬ª»mneünMüm}bìoGiÖý…-•w/KoÒÆcŽáJí÷‰Ér{ñãÍw_]ãIÔdY\dédj“¸¶†E rÁM&E\—i…¸ylòj25qV‹ì?0–¯¢,®;ñpÞ'y2Â8«2›$Œs8§“Æy]é8Ð0¸·*ºR™¿‡’æp“1~`9™Ã™Ÿó+â*+tœ´m²è3ÂÎÚ¸JÍ3ì°•*Îõˆ0‹8KŠ_R˜ù3Â,}aþ8B¤C°Št5?¼=§cò$®ò£”€a•¤@žL³ÄÆu]‹9YF£4ƒ,ê¨AÃ¥#»A«-ªè°b3QY§õÀë9›Ìž£™x8¿„¿!Å™uFöpN ‹: :M£f§yþsw_zH°¨Ä™b2&¬|¡p>óÀš¢ŒÚ‡Ý#-x·®¿_Lát“§±ºØÔ¾ÑpÕynØUƒ 5l[–×(úÅ‘³ ²2#¢"OœVNõf˜™ï§¡{Të&·¼É¸ÿÙì‡F‰K}èÙ:‰ Ș¸ÎåÔsH ™A·o6C‡‚Cƒÿ>…Èp´ÈÈ:ˆMÔØ ÒñÎ Œ¨Š¨Ù¯ƒxRG¯;Ä|YçB„åù^1‰ qk !Ží¹@mg•s7›f>𥕷̶G›FK$úIo—ŒÞôyQ»0ÔB`ñf°`ˆWùeI9vÍìÀn¿(6Ò*­û} í™Õó¯$8¼F¥Îö„{™å©»¡y ‰E·çƼUu¬÷V2l´h4…áY•¬¶Œàˆ˜(³ÞZHÀ;Ê“Zê·”mX\µne”<"œÌp†©VžE?áì¾âÀãÍ’-¯F&´{Ö6Ù¡?àn„®&J|0ÚK€¡¨€æµªAÓ7’N©Åñ0ÿ¡yáFéŸ ì· C[‘ËŒ4TÎÎïsDB˜_¥ˆ“&9_R¸,ýßX"Q@¬¬žÍ#²´:É#ª0­©OŠXqÏ‹h2#'C¹1„cËÒA§F¤ú¤yЮL©szÙ3‘”¨§¹¥3ÒKBæÑ7(R±cÄW5!Ä}¥+tµ \£Žï[>ÄUÛ¤%¼wÎÝÂ÷êe‘ÁÚ‡áœyMnâNKîDj‰XYt¯NkÅý½ngGË»'‚zÜëèd¸÷ä̧níQoøÇ7³Üžx_Dù·L¤¢NÁtÎa ³î('< .Êów§–:lMÔ‡÷›GÝgêìÐf2Ù»c/·%IßêeIn"bPôJqú G+KzkžD€nÛJcñ²ÒGG!•;b;ÜÉãGgé’Xéb–…’ûs‹D,ë%bsZ"o ûÀƒ ¹X-‘6©Ú ŽÓ`ŒÜãƒÀ¶üûýëw'çÛß0t?DÊgþ16n|Ö.|¿Âiêô2ck¥ö %è6;½D·s>€c –›%~²rž,ॳ®ýlÌO-™ÂÃy‚™C"ëî]£q·Èÿ,[°È¢´ÑÛ½yçÎkÏ+Úv2DVÑ2ð4h`T%Sl˜â7c÷ÿúѯów>$Õ·xO."$Ë0üÍ$t»¹ÐVgBùn¿®1#~ñˆh 4-^,ëØ@Isÿ!Û¦âñ¨)ÇÖ—¼úÄ[= „ø/ZOjâÔŽdM­›OÈ@KäæFMÆa<’[´Þ~+.Ùù >¼'¤ÁˆÅo…Bëˇ7‘ø£ün…°\O’¬Œ~ÃD @¯d¢bö¤;&zÉhVÐvžðd(•¡¹¦|Î×¾pça)y%­ÎÛö½—O¯†íRºöó»@×i^§\:ó:¨¤”t”3öjľòØZwrßÐ@ób ¤ Ïù‚ݦ±)ÓP\  áœÆçIh&šÝ‡¦¢K-J°ÏÍ2 Ô‰‘±9ð­—³Î>lÇ£¿9ÑÐ¥€ÂwBéƒàˆÜ.9b%xǯÎeÖh¥Q̵óìóŠîeÁ.ÔðÇPÍ€úQ(»¬Å›¦è7ð÷øû+ü};¬×Ós°æâq=Û_¯žÍÿ®düŒŽë¸8- ¼DÇéÿSÇ#kV·9îÜ÷ ôKø .ð7¤=\ø=Sùš¯áïM#ÐÕ ßsOÂXEÒÊ8 ¾Lã¤*Â=|)ô;Qä³lOI{ñ÷t%Ö“s¾/Áyz‰(ì“¢H^$…ç9¿Pv̦uˆÛØ ½ã¹Ñ„Ö†ñèrI(‰«ÓŠÐÔ~cJsSz; 3Dž™ÕÝ3¸Žº uKóA:GN?±+ù¡ºµÐë=%/ÃéÊ$Õ#~K„ ^e¨µÁA†¡NòŸÇ=#é& ˜k}ä3½ÃÔä%º%’P¢ðJFóÁ–eJè «à¯SÀ“žaš•&Øçì’GØUcì~¶C`±+ýœ¶ÂR–ÁN¥”¿pë#|Ó¢xãÏøæ'ûO~v+œê…Çqºq%Ö„o•Ý’$²aÐÊÝè^ÞA 7'H¬Ö×µ«¦PgAéÅPN†<ŒžU|{„à“¼cxw<¨§ÝÈðPâM\’°OFfh¤ÆÓ¯ ½6SúÒînÎAÝÖì\±4k+)~Üa;‘ÛìÝ‘+]¡aI—òˆ‡m:–t&¥ŠËå ¸ÝÞ¬ZžïÞ?(úe·"´{vMÜvüëÊ&\à&¯)q•w‚m¦%„©Oª UJºã¨Óô ñXN8 9¢æ«’¬%­ª“5o¸š~º\:½«¹ ¼ÖJà÷OeÕód¬fàðN‹ps½æx`%t—+¾ÞòÃEËtvdŸ¢plnµš€å“¤ŽþvQeâcÇD„ÇÂÔR:ë¸Ck·Ü žL%HÅFèéÇyýD%sq ½Ý‹ì ÍeäžçÝ OwäÖªßbXï´¸¢gt{„Î¥5hÌ¥a¼w'hKå‹K6‡0Ô!?Í‹û²Íðñ¤Ðï °ŠIt¿æê¥ŠÇUVS¿¹—Ï5r_6䊞ü:äó†±ÂêºTxå‹þ!÷¶l†Ç?~ˆÁ¡ÙP3ƒU¸‚uðÔÔ‰e¬º=V3uN؈ÙÑu‹Tü‰¡îSRÓÒÝåÑ ~k-¥·ŠÞ¯?¦q2 «ù%Ÿ^ô*LJi.|w‚®ód™9}» ™ IÚ_ÇõHá´vÉÏh½”¸Ð!É­ÌYàðÞ¤•â˜Öú±Df|³£¤gàã#Ž5üs.j„þCS²>p@Þ{X{Úçîi>3.|b{)&TÊÅ’€Z$Ϭ.Ê{ÆH7Û97æê»Z½z¯oêˆ'¢XÔß+W4Æ2s'Yš,Ü2å­·xë”›äA@ïyêJÍÊ‹’^mÒ2—XË ®ÄbÅžW]™…ŠÂÏ ãù(H°ñã¥ç„K/¥÷¢P¦ç/ñ§âq¥<ù ŒDfÐ.¢‰ñÕ{ËŽ1n½ºº‘¯ (*öœÒe¹õŸÄ•>Sª³#͆ûçÊ |V‡›^òS¿û²® ýnUÓïkÅ &—J™ß–æàœ”í9Í­ ¡ª\Þ˜ ± œq3ö‘–Õí²À HÝg-¾4‰²n²*ðÅ7¥ k¢·‹“È'{+–% ôôU‰e<ÌòëpÓß|°¥J^÷œÐôÊšÒM?!ê]ÉÜ›/9ëS°66?ÿªáR­Wfº)|ñíÍÿ¤)ŸÃ endstream endobj 103 0 obj << /Type /Page /Contents 104 0 R /Resources 102 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 102 0 obj << /Font << /F8 9 0 R /F11 11 0 R /F10 12 0 R /F14 16 0 R /F7 14 0 R /F55 10 0 R /F59 31 0 R >> /ProcSet [ /PDF /Text ] >> endobj 108 0 obj << /Length 1645 /Filter /FlateDecode >> stream xÚíÉnÛFôž¯’C)Ô¤8ý­{hQ(P E‘ú ‹Ô‚¢@I‘“ ýö¾mÈ!EÙÊ¡@=Ð3óæÍÛ7ù§ûW³»t’yY¬ãÉýr’%^§“8J<D“û|òÎyœº:vê©Nœ§bªœýÔ ÂÔùË|8íàkà+>L£Ø™—¸ÝŸE QáíÒ‚ä¤<ØÍ1ê†:sè!Ü”5›ÿ=A/ZOHÚ4AÂp|Ì*^ù²áGb™}÷ŒF{T¹u’Æx‹šÉzÀg´;f¡\¦{¾ÂìÚïD†÷ìOºÄ¸3»GrfOÖ-†ß #žÐ²LøÏÒJ%—êƒäøQìÄ1ý¬EØHšµ hGä[ä)¨n‚ÑáÀ¼iÿ3‡ *öSŽË6pfw`޵«b°U–€ZlõãÔ´°‡ï‰2K_ÿ€YÃÛR0VðIMPÝm›C‚AA ÆBûY休<%¤À`‚mçXÜÔ09Š—òŽÊ¹ÊyGžcLavIã[6ùQõP¬b.a±PÉ[ø"øPˆ™(¿¸/ä#&õºGO¿Œzú"uÔ$ˆ½( ú©c0pM ãT¦z°t .ÂÎBßb/Œ³‰«ŸJ¥Ž™ê‘¥]Hã¡"÷PT4{tG ågC—ZøsÔqQQ ¹6$¥‡4ŽœÒP™ Kû' Õ&· ©5 sâ¼”„Ie¤ab¹lç14òhððh’Ù´|A¥óÕê|¬•´ÑïÚ–1hм曶Öþ.% ІC6Ôþð)×-”ã°fRdÃÕšŸ¤BáwvaÄw]»GÛh"p6ôðœ_Q€•²zÛÚ^’&.Å”ãGš$Zå6ì.*Xn&FÒ¿î:pc º0 Óœê ×y”Z¤q£\vX%4E)íI ÄÛ‰B’´î¶&bá¸7ÓÅÊô¹ˆ,Yë?$sRS˜p-­> ˆmÁgÍFC »‹ä[{‚A¬aõ÷ÆâÊ$Zôv wÉ¢¶ßÀVÅcÖSïj ÷—-J×[|w\àß5C(1`à ±në„Æ9~Ùpç%T² {Ä,²3qË(24¶Ã–Hµ¬…ÿœ—®=Âah ïG–3Gì528¶S#„ûˆ0ŒD߯¼4n”ËÄ\9¿x$ÉŠ¶-rÍÁ÷§Æš1wÆzyë4©+wyd²<ˆFÛNVõìó6Rc€ÿÁÄå`(¡Q$î÷éô›n ±|%¿EJÑÿ±0ñ3̾7 w24ݸPIÓ7ÅIšä¢?—˜y€˜ŒŒ'VÝέ¡g~6@UÂ'—Ò.íQ©­ž:þ‘ë; )}ËË^=õ ¦ßwS.cÉ,Ó—‘5>œÉÞªk&"·?C½1¿Eþš2¡ÐŒÐAiĬ…Èu¸dÁE7?³`ÿÖÌíé‹CW²šRö KV%Ú//i?Ò…l_í¬Yv ÁPÐþÊs™’¸Éž{fE¤P(k,kÚ¹ìá\ÝòrhÛ!>õ›nð?Ðo×góãf<_nÈŒ¡ò’x’­,®„v#öݶ•]]˶=¾näi­ú|.;!t+e)³Ê’Ž=åkó3훑„s_ŽP@Æ/¼×™A°,}ÕnûáñÔ6.yxý¾?E SÛö—g/<Ãîööå¤2°×vJ¸IêzàLÓ7ŒÓ?ÍüN$OÓý0D#\aƲ_Áb¡g©‰Ûß®Á~žÁwü$ý"6v/öÆ~^]‘ÏæÒµ y]V_ˆµÇý‘äС&éÕѽ½êj ]ìé$ö’¯¬&ŠdMd;îš<Ÿÿׯÿ 6þ=òoÀD{‘ÿ 2°f ÿ|‰óÕ/÷¯þD.C0 endstream endobj 107 0 obj << /Type /Page /Contents 108 0 R /Resources 106 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 106 0 obj << /Font << /F8 9 0 R /F11 11 0 R /F55 10 0 R /F59 31 0 R >> /ProcSet [ /PDF /Text ] >> endobj 111 0 obj << /Length 1637 /Filter /FlateDecode >> stream xÚ}XÝoÛ6Ï_á·É@¥ˆúÖ°nèÚfh‡aØæ¡ë‰ŽÉ–+ÉIóßï¾(RŽ7Ѝãñx÷ã}1jÃO­ê2*‹j•ÖqgÙª9Ü|¹‰ê"I3bð†4¥d™n?ªÕ»þæ7øÙ©PD†žÌ77·wÕªFqÅj³])•D¥JViYEqU¯6íê¯`³[«`ϸ3• <>DËͯÇuRf€‰g&´¸ôs¬R\ƒtsD–‰gG L½ÇŽ¢q7}\~Þù- ûuR_yúЯCøÂMLǤŒiŸÇu^ºã-UОoÎÂÀªŽÑ:,ª2ødmlPüŽyg›aiI‚TžÖo>®B•E5 *ÕyΈ}ÀÛ=íÂb[ÜV£Ži{œQU"õÁL"U®ôÿ-)l‰˜Á$’fx9²ö“ÐG¦í(E²·g:Ã)&òŸ‰þŒì-Svý ð.ó-€–¥ît·J†p‰—ˆŒhØá$^§ïDlj'Ø#! ¶èVO<>¡ÐþD£ŸÐ&Í|€âNtP8ó´cœhƒùNŒ1ymËtuDóD\Xñ‹6ìšî¡1·w`†$0„Q˜€}ªbû¾_‡y’’à-?ÇyL‡ v âð=<0­(G—î³…ù’Äü¨eÊR¿CÇáá“7»c‰±|oÓWÌ݈7 ŽÓ÷^¸ãÞ <­åí êų<¯ù^„‘÷»}´˜ãÉ?ˆ|$}e²Š³r¦ŸDÒ?ÂCh!!–äQR¿„LË¢N`ïE CZª YaDÓk¬Ê±>ŠdÚ[fÙ7ðüÏŸð¼Ÿõ¼¸D^Á÷œœ½vröv¶çÖ M þj±¢E0<\aX¾8“¥yT¦ø†Œ&~­Œ3íé;2¹3¬+Zÿkyºų¬w>°y£ç9KÜ»Y²Q4ú+„ô[%Ë`|áLj$e ùúÈyrÐèªr{C²´ëœ!”›PóÏ¿ïE¯IòJë8£âFTÅ0Å<â³Hx¼æÖq”J)•6/½R›eç2¬s@ÿ|A-÷¼DºÚ¥ETe…]<5l,™fÖÑÑœ¿ï¬Ž(4üIJ œQÈS;¡ r«%Æ7î4®lgù¿ê`§ß0… ,e}`ÞOò¶E³f‡ÙÎ5×0Ã@W7öéA|~œ¯ œt`"¬¶¹ ®oh/Â8nlGâ›¹ÈæüOÑø…‘55«ÒuÑ7—~Ï0×ÏÁìÙö[›ùFÄIT•95fe&ªçnÞonþÅ9} endstream endobj 110 0 obj << /Type /Page /Contents 111 0 R /Resources 109 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 105 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-finegray2.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 112 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 113 0 R>> /ExtGState << >>/ColorSpace << /sRGB 114 0 R >>>> /Length 4631 /Filter /FlateDecode >> stream xœµ›M¯Çq…÷÷WÌR’7]ýÝYÚˆ p‡‚ÀðÂdÀÄR€üýTÕ95Ý/CS$Å,t¯Š=ÝÓŸOŸªš+×·—\?^{ùÃõ·«®GNü)­ú2«³Ö#käG±6µFÖ'Ë#7Xk™Õm´Z£è“o°U«¡Mý×ÔÍZ6óú[gyt{Þ­¥OŽGvKû©o˜6ÿn51«WXºšÃgU­i5Ôj^oÙÌdÛáÍ‹ýf­KO‘õef«¯ÖJ°tUôg±™˜ÅæNÒ×J­™a }Rú“ÍÖhvÌ„Zº*j ë§þ.ùšÃÖF­a»A¢lÚéÏáõtvõ \1ý­+†Ÿo^–Χ\Kg|ÁÒYVËû²òcÖKúøV±UÑŸÍæei_´^µuS«Û:¨ÕmDkX}}Â÷îš¶j ò5Öç}&$%kN+´î¦Ølk ?-fêQZ†¤bë¥-,<Üs\ÞÄr[§ _ÞFv{Z«fOë£èq­ö|ÆI›$kÕû)¢Ë³Üž‹ö°ç+ö¯HµÉ±×.´×lz$uì}‘a[ͺQíaïŸÆ³µ?Ùmô?g[,I çJr±MiÝì(ï6\ë&Ƨ¶V´n5_I¶íE ö¦yȼ¼›nW±ŠðHm6õÖ ŒW9¨OrÂÉÃ’öG»1¼\—‚BòĪŠmT­_&N†èªéÿˆÒOl~^i ¤É–òC Õ¦ß^¢;O Í7׫wŠi•(Å™ŽRZw©m“³m𻳗ÓÜåµïÎh9Í]®;Ëiîòs0{lwùo^sê¾ûù=S÷ówzE}£›ãzý— ³µÀn~qäawÛë·×Wéëëõ/ÿòÚ[þpUåæSÕöñUõÈëÛUå^kä¹ò'¼XªÃwWΟòæ^Þ©ü)oVð>U.÷›±ºzuë¹z˵κSôV²5~§¸lk—:nÆ]Ló(×uVjÜå0r¥åJ»æ.gw¢üèÝ9òc¸ÿxÌBQ!¡2JG>±Çôß7q¬®[F±Éêï÷V—ší6Cý÷o·××›OïTÖÿô÷çÒíÞFýüÎÒóx¿d½df˜oÂÌFP³¹Ïf<ì=v]V€Û/½ÍÓ­W5.ìÚP¹S¸‰ÿøÕ|mgêúê‡?ÿôó׺^û‘³¢ÿk³Z†í3›•o~wý×O?|ÿ×ïþû‡ïcv¾¤¿ªý¢#ý‚×ËñPãPÕPã“:jœ÷1Ôø¡ÍM'êo¨ñd;÷VãBM5.TñPãýF5ž±ÎTãžÕxA쬮 ´Õx…*£¯ÐµTãªnÛVã ‡j¼ãF†W«¥[«åÚjÜžØj¼S¹@뿹5ÞéÏ@ÛõVã}Q·Cû:Üj¸/5n=ê·7¯!Ýj\ûà#‚×>¤­ÆÕrO j\­Ño5n>D¾Õ¸õöPãÜYTãô‘ ÆÃ£€Ÿô¡Æµï®N¡Æ­ïóVãjõy«qí»¯Ôø¬ØPã“{ j\Gâo 7w#m5®&ôÔ¸ ­o5>é#RëPË!ÆÕœukql‘-Åmìk+qsBæâjޱuø,e¸ÎntRæ!ÂO5øº%½Kp½jÝ \'Ï”šOMègèosUê–ßËÇu«o5ý S|Ûìέ½WH{JïDd„ôÖkaµCz뢜Ò[gmœÒ[Çêà é­ÄM•—¶¸Á¨U0º~ƒqæëSÈ8Á8ç Æ67 ºÁx†)2ÁHŸ`,p Æbo½ÁX‰&€1BcÃ%Û¹] Æn¸ºÁ¨›£m08:ãľ&'û0N¬2Á¸p”¦Hp7¦à¼0L‘¬7‹`Ì8)c†K0êLÌ FEZÙ`¬vsÞ`¬¸0ƆñŒºw˜B-o“`Ôk o0F`˜A‚qâò"'ÎÁ¸0Z†)x`¦ì†)„O"L!5o0Nž]‚±<±0¤0V¬ÁX­·7;f‰`ì˜k‚‘×Á8p)Œ\w‚q"8@0z6=†)¸‹0;ÂòFí$˜0FP…`\\p‚ÑÂ*GˆbÝ€±a4ÆFÿ`Ô•-6r€qFË#c^ÆEÆŒiad"½™Ðq£mF&g27m"ð2F&ÔÆ”02¡3Sò‰Çèzàqb“u6ØyÄÎȨֹµè󺑨oõ$ªÕ·VìܵDb"<Ä@(è5‘('D¢Üʱzœ¸3ÞN$f¬‘Xe…V¬T€ÐŠõI+VôšZ±žH´7Õ=hXß‘ÛÀ:¨ÖØH4¥Zo$ªå§H´h÷ÖŠ:;®j€Dµüg—HLh3˜¨P¨uFçFbÃC+fžOjÅŒPrhÅ Z‘Z±Pú‰v´D[‰0‡KßÙ6g(: qò^$-œshE5×ÜHœ”XDâ-wD]­âPüö@]Q× ‘fjÅDuG­˜(Y©åI+Ú‘6 õ @(ÔUsF‘„º4˜v‚0U¤„: áǤ·º-·lC}ëùmz(ó½m°VMOµ`Z­OsêÙZžO­ÁüÅ>Hªó‹YW¨–AÝÃ&ß|}‹š¼ýó›¾¾t}¿ú%ú{(l²Œ»Éßÿʦr{oïþYë§7&å½ý:š²ØN†"ÿ@pçãžúUiجt‡ƒ3í6ÏÃd~‹œ¬œNõ¾m3 ÞIN¨N^›ÙüZË•æê÷”0·K¨Ðî¸ÍŠk^ fEþu»_!Ü4Sov“ãxx2İÍæšËCΑ¬û[»§gÞïe3Œ ")O °ÃÍÈ Pqï½6šÊ“»‰¦^ñ¬îfe°ûlh›É%¬S%›ÔvU:;ÍåªtáaÍÎ!L—EF(™à »”ÉËïY“ŽîΛ½t«(§ôì¸3òve¸ÍpíÝSö¾-Ÿ Oªù,h·¦xÒÌ/¸¼¼!‹~JØžT«ðâòš &C"f×åI4ÌÛò,ª]~)½MªÊ‚çl¶ÞÀv[$”»$z&ý®3{Y’­£ÿE®ZÀŸhøØOÀ>ŸXžÒÚOÀ>ž`¿î'Ž~~D’/7wVª_–Ÿ‘ùe}»œçç¤~Yß®­ØóŸ–ûÌCËŸ“üe–ýÍÿ7ûû¤lß*5ç™þ¥yç{cIÞ±ŸÀLûþ?¦€K^_&œõÒý¸ðó~) ¼}äãé 0ÕK_@OÂȇ/J _@5 _€¾…/Ð((é 4dè tÆè °Óö:Ó,ô:£èá Âá dŠpøúö2·/`a£´}‹•í Œé ¨¹Bü‹Gö=€H_ $Jø œÑ°ÈNÚ¾ÀäGhô"C_Àâ/²}3sûs@{јü¾Œ®€>ƒCOÀ¢êm;‹º—~€åùÒv,n>·`ÊØN€JýÚ·°â tÓºôcÏtƒ»¡ÿUÆ—tè }ŸúßôÖ8ôêˆf…þWY.éÐÿjC_Sÿ'fœBÿ ?9 ý/BýGý¯2zú?ü›Ðÿ‚;åÖÿ&»ë¡ÿ-ØuêaŽ+ôNðBÿgFöBÿ«¿ãþZè•éÔóÐÿ“˜‡þÏ=Ê¡ÿõ‚}Òÿ™¾uèÿBç:ô¡wú_ý¡^ýo²¼ú߾ݙ‡þ/þõ¿¢KNý¯¢þõM@Eèÿ*ôϨÿkÿú¿2¬ú¿zÐuëÿ:à_†þW ý£þoL„þo™ý£þWÉcèÿÆ {èÿˆzPÿ›?ä‘IêéáŸAÿ›¿’ýoŸ':ÀBÿ ób—iÞô­é)“úö±– ï:é˨zÐ7â¤oG ,èÛÕ ïˆh èœc$&EØ‘殃¾%è{Çi@߆¬@Ðw eô]ŒÓ¾ )‹ˆÄ$lɈÄðã—ˆÄägú–ˆâT¦: rЗ»)èËÃô½ƒ: /¿Î‰HŒ> endobj 113 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 115 0 R >> endobj 114 0 obj [/ICCBased 116 0 R] endobj 115 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 116 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 109 0 obj << /Font << /F8 9 0 R /F55 10 0 R /F48 8 0 R /F57 17 0 R >> /XObject << /Im8 105 0 R >> /ProcSet [ /PDF /Text ] >> endobj 119 0 obj << /Length 1063 /Filter /FlateDecode >> stream xÚ}VK›H¾Ï¯°ö„¥˜Ð¼ Ç$JVÙ³¥vrhxÚfy`/-‰-Z¯ÏƒþFû‘~ú‘£8uâAL:¸]çe[¤A¸Ýe© þ&éC„ËÊÑš%ƒãÅ«²à†ý 'Nê®Fç~ðùòÒá¸ISáµ£ŸCh;.n¨«¹ÞÃÃm¿þÙD›1a™i¿,þ@t+Án—)¢`9AŒôÙ’;ŵÕju­°I$† *î}º•¹m³ULN‡²?k8'ºÎ7˜¿PTOÛ}з«u½ÑBƒsN§vä$i,iß0"Ÿëðê@™3U#'•Š!‰VD&o4Œ§‰v›½ô¿GÔ ŒÄ<ZN?.ä$^ìUMumd4lo¢HÀÂÊ i6Vzím˜›‚cZ ën =#Ç6vÔ“'$—;‘·Kw¹ ‹ÿ™‹ëAF GbÊW—ánu½µ‰È‚:Qs '’#®FÖÃÁYpo’ô«˜°¿­äc}•è³?êÈ!Ÿ80¾jü? ऑ/"_µ’ÀñBr•’%—Mô]ïFÑ1‘ß4&¿×õÛ—Hµ´,©èƒŸ¨‘áŠÂâ(‹}¶Ù¥û8$hH}¦€òîóáîͳ“ endstream endobj 118 0 obj << /Type /Page /Contents 119 0 R /Resources 117 0 R /MediaBox [0 0 612 792] /Parent 93 0 R >> endobj 117 0 obj << /Font << /F8 9 0 R /F55 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 120 0 obj << /Length 104 /Filter /FlateDecode >> stream xÚ31Ö3µT0P04W0#S#…C®B. ‚‘)T&9—ËÉ“K?\Á’Kß(Ì¥ïé«PRTšÊ¥ïà¬`È¥ï¢m¨`Ëåé¢`ÇP„ÿþ7Ô3`‡v(P†ËÕ“+ L5* endstream endobj 31 0 obj << /Type /Font /Subtype /Type3 /Name /F59 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 17 27 27 52 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 39 /LastChar 39 /Widths 121 0 R /Encoding 122 0 R /CharProcs 123 0 R >> endobj 121 0 obj [43.59 ] endobj 122 0 obj << /Type /Encoding /Differences [39/a39] >> endobj 123 0 obj << /a39 120 0 R >> endobj 124 0 obj [511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6] endobj 125 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0] endobj 126 0 obj [585.3 892.9 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 585.3 585.3 892.9 892.9 892.9 892.9 892.9] endobj 127 0 obj [446.4 446.4 569.5 877 323.4 384.9 323.4 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 323.4 323.4] endobj 128 0 obj [1277.8 555.6] endobj 129 0 obj [647.8 600.1 519.3 476.1 519.8 588.6 544.1 422.8 668.8 677.6 694.6 572.8 519.8 668 592.7 662 526.8 632.9 686.9 713.8 756 719.7 539.7 689.9 950 592.7 439.2 751.4 1138.9 1138.9 1138.9 1138.9 339.3 339.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 339.3 339.3 892.9 585.3 892.9 585.3 610.1 859.1 863.2 819.4 934.1 838.7 724.5 889.4 935.6 506.3 632 959.9 783.7 1089.4 904.9 868.9 727.3 899.7 860.6 701.5 674.8 778.2 674.6 1074.4 936.9 671.5 778.4 462.3 462.3 462.3 1138.9 1138.9 478.2 619.7 502.4 510.5 594.7 542 557.1 557.3 668.8 404.2 472.7 607.3 361.3 1013.7 706.2 563.9 588.9 523.6 530.4 539.2 431.6 675.4 571.4 826.4 647.8] endobj 130 0 obj [565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5] endobj 131 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 132 0 obj [694.4 666.7 750 722.2 777.8 722.2 777.8 722.2 583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000] endobj 133 0 obj [312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7 593.7] endobj 134 0 obj [700 738.4 663.4 638.4 756.7 726.9 376.9 513.4 751.9 613.4 876.9 726.9 750 663.4 750 713.4 550 700 726.9 726.9 976.9 726.9 726.9 600 300 500 300 500 300 300 500 450 450 500 450 300 450 500 300 300 450 250 800 550 500 500 450 412.5 400 325 525 450 650 450 475] endobj 135 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 136 0 obj [301.9 249.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 249.6 249.6 249.6 719.8 432.5 432.5 719.8 693.3 654.3 667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9] endobj 137 0 obj << /Length1 1822 /Length2 10941 /Length3 0 /Length 12084 /Filter /FlateDecode >> stream xÚ·PØ- A‚[pm‚4î<¸»6ЄFkÜ=¸$îîžàî<@àÎcæÎ½3÷þ_õ^uÝkëY笽« §VÓd“°t0Ë:@alœìB)eI=N.7;=½fþ·^ììq€ ý#BÊ ‚=Û¤A°ç@e(@ÁÕÀÉ àäâäâàpqpþ;ÐÁY rƒX”Ù P° ½”ƒ£§3ä5ì¹Ï¿˜,˜œ‚‚ü¬¦$ìÁÎ   ‚YƒíŸ;Z€ìš0Ìó¿J0‰XÃ`ŽB@ »»;;ÈÞ…ÝÁù3+À³h€]ÀÎn`KÀ”* {ð_ÔØÑèZÖ—94¬`î g0àÙ`±C]žS\¡–`gÀsw€¦¼@Õ ýW°Ò¿X]€“ó?åþÊþ£úg2ÈÂÂÁÞõ„@߬ v`€ª¬;ÌÆ A-ÿÙ¹8<çƒÜ@;ùsÀŸGd%Ô g†ñs±p†8Â\Ø] vpþQæùše –Röö`(ÌíóICœÁÏ÷î üëqm¡îPï#+ÔÒê–®Ž@m(ÄÉ,/ýW̳ íoÛ;0 ÀËÁÁÁÏ';ÀÖÀ?hy:‚ÿtrþa~æàëíèà°z¦ö…XŸ¿Ð¼]@n`ÌÙìëýOÇ#4NN€%Ä0¿ƒ@Ñþ®þl[ý ?¿¿3Ä`Èñ,?NÇŸÿü2~V˜¥ÔÎóïð?ŸøVBUNWŽå/ÊÿqJJ:x¼Ùx¹l\¼œNNnA?/À÷¿ëüçþÍþO«ò×é8þ®(µrþ‹Äóíý›ˆÛ_Ê`úkl˜ÿÝAÅáYÏ`Óßò7âàå°xþÃùÿ<¦üÿiÿ*ÿWùÿï‰d]íìþô3ý+àÿãÙCì<ÿŠxÖ³+ìy6”ž'ú¿¡ºà ´¤ƒåÿúäa ç ‘€¾³ûÏ5B\d!`K5ÌÂú_"ú÷+<·ƒ@Áj.?€“ƒã|Ï3gaû¼T\žßêOøy¤þ»¥ ÔÂÁòÙãâ倜AžhÏãâåxs>©%ØãOm€ìPØs à™œ/ÀÊÁíåã¥þ0ý‰ø9@Ù¿'(÷7zŽTþ|F ¿‘ hñÄÃóŒžgöoÿÌ€–ÿ€Ï•Áÿ€Ü໿Óy@kOGëç…ôwij òøÜÝöP´û|>ýßó¹÷?J=%ÐáïfϱÏkûn.Ðño7ß3zV©Ã?ÏùLÏùðùh.ÿ€Ï°ÿÀgÁaÖÎà¿KäßøÏu {€-Жæ,„CljCÚ¯«%ÈÜÙ¶ÇEgè·uÓ˜Ù¼—œ;\o±^¦0W} úî|)‘2Ô‹³²)Ãtñf™êÁû ¥þexk’zÛϽi‚ÆÔvÚâ$á—‰‚‰º Tr6­7;>N>:¶-ð] ô9N®Xjy¯®Ýûå<êJ¿†Ío«ïTñ)¢ß—N³ÅhGÍÒçšgÍÓ ÃØ(P^ãx`Ï^\ÎàeOš‹zçî¡WTÅìGç`ñ/¶°C5®ðô¾D;Ê &qäΟá?SøIò¾ úM3Ö'>ù¿IXä2Ï…öxƒ<0;_ð_ÿö˜ƒ¢|2<×Ò?Óš„#¢ sZSg #›~ÒüR‰÷d+OV3õBÑf´æ»B—ÖSŸÊ§‹`ñe‘›7/~,8im»†,Z–‚§b3l„z ˆ#Šw‘´\šyÕ†ÉY’æxâ£åÊðA!µüUÅ PAjD#isø»ó¢+5 ¡ñú©’é÷è{–U’–‰>Sò¥“G¡oMÜß+cW–M€i¥¥9³Ê¸u“ò»#ØG'ϺØ ù*g)DÒ´H”‡~Ñ4ˆ…83‘”«}%ø”бL›)Öd`Ósi7ÔÀ’¦ýè^õô{í¾ž.~«orÃh]W…]ªIgÚèXeg‡<§£·™ iVBüÔÎÇ¥§D9L­éª½3ÖÓJÓ³x}¾ËüîXŸ™ +…cì—KµÁøjWçyf%QWë’Ø¹ éAéñ¥ÜÖµÙ(“"_â¦f*¤îîY´-¶âq/í×cß(¹Ëwý.7ߺ{m« ½˜zÙF…5¿>Æžy›¹ãÿeJ³Rë]ý×EƒO_¥üê¾Ú]Ü\”GâµY¦sŒ oï¨Yñ¿¢u¬øÆäb©³œ¼äÖñt¦¨ÇØÍÈU64n`òÃQ S½€÷tbYß)û9­PŸâ/q€„Œºãvetžœ[þl@¡†ÃÌSm5Ï¢ê¨gúæàM5ŒÓ0 ±Î‘ZsOë ú}Ç£&ñ*ÙÝÚ¼Ëÿí 7DÆ7$¨Á»Ðþ—×UŠÁ5ÅÌ=‚¡L÷«Ð~G/  îÍj[`1ÈTi§!ÉÞ:©luÊØÞ°kè”ÙT„ž5/ÅáJ®wEñ«ƒ¡tû«Äª; OÝZ¡ØtKAš(m%øâkÝí2U0DÓ!6…Žä¯èæãùáùg–q'%Öá©ñµQqýb*‡%ykŽ›q.Ÿ³Zä”–T ­ÅβgbÎgrгH»á–„OXŒìÑÆ©ÈHîøˆ‡{mI^ÅÎÛ Ò «`̤§{ ;N?­ùܽ%°-DD×*Ý|3G±Õठ¿£L¿Žôq‘XøðËJñ Ëy'u`ELQ’&ô¦``Ó6Æß÷u˜1ŧ#È 5ÁÂäã°ÓHš½@±`Õ‘óÞóµæ <–ïeDN‡ÏÔO´Ng_æ&‚ãe Nf¢4·Ë¡²¯á”qIýí5ÿ«’jû¸¨¹ÿ˜ßѽéªêƒT¼¸ïïB$t- /—¢úC_W%³TåÅ€ÒNþOŠ3èGW»m§[a«žª€(ïÓÓøŸ73fÊ^vé0rrîÜ¥EºüüŸ¨éíVnÿÚŒÔéU—¾,i¢¶3OI…½Z˜£5/˸~%ÓÞ{J)×ãj ¾],°OwÛEKF™Å›³¥@§ømµf’›=‚^‰WT¡æN&Ó Xt1A§Ò²,’÷PüôžÆF©MÏáAÙ™tq_¦ªþWÖœÒ=ÍOê_¯­ˆY2©áµ%"&†_`°Šsãuozð§þæ£%ëzQŸEоߤ|u÷€>”2äZHež× Lú.×U·B x«veTbdf¾¾9Gwé,ÚÉ<&üLz%+’éðJRgÿ`‰¿’=‡·d“X²qSC^W(¢!Ö¡~µÉ@áËùn§è Ü)Í6E45ìW4À€³=,¦ª}õ­áqeÆ–,¡UÓíÍt>d¯Ñ}wÚ“öÁ2ãéEïÈq½Ëðw¿L#ðMÄõŒ1Sìñ2šÐö)ˆ«µ@|êÕÊŠwhÎö¦\öü'”7µB_ríÊo\@;?—:åZ§u"­ÚÎpÛsW: GåáEÌZŽbÙÄ.Œ…ÙΑå|5é(A>Wë ¶®ŒÇî™8U<ád÷ÓBÖûê¹(ñ[’ú}å_¯ÕIJMˆyŠg¨šÁ#µó|À»ÚH©1—ÙÍÑ1álœÖ9¹1g ¦ Wtå°4‘¶þ"yÍ^³â‘3-·MdòÞíZACIox\3f+4-š“¦ŒXΉÔdßZ]¹°àçùGÞ~1mpÉüôE³–(ë'+Vì ¯ù=5êüØB_*ËMrÿ§0»p-Š!OÙwñ[ñ w¶´í µYŒQ¶³«¥Ø÷ßÂW 2ÓΣ4YÓ©t Û •ÌÙ¨k*¨YÞ­ÛÁ3WFÊ·ôÄ-:ÚCL¾›)ïÇc”û²ÝÊF#Y· í_ºè]vGhôH†ïêgqCmøü ñ¤Ðg VhMlí¤j=±[WèÉ©˜'顺s²UpøÊ8éžþCø³»~gN‡%ËTù¼Ð@ºÏÙR#Z™¯=‚KŒïJwS™Ï© Ùðð0ƒÌT¸Ycûíˆ@¸„eÕ»¤À€Ê‹væwpÇ´Sz¶ Ú_r ­ªµôøg줡á„Q(ÚˆïY¯¬ù”óx&öÉQ½~ mŽÖ­Û<Ú;»]dªÇÒbŠ%æ ˜xœ¶mš^µ}QêkîÏF÷, +M ®>ŸàÅžÒxô¶:P—úí[f9Ü2¸pé·¸é\·G]…ÀÓö=Ñm¥–ïÐtÐ¥<ÃbÂÈš†uÄg ü¦ëm½86t<Œ˜Êص٭Óô~þ 9%ó´Ö¤ß¤¢‚Á­R\²‚¡“˜ŸàC_áš[φÉ J0R'sG¸ï.™¨5‚ÊtHg±¶„>†icÞ ™s囸n_àÛËÂêdz•󇛼 UŠâü¥NóÕ§í‚–xX®ì©çéNÝöÅo†cï3wuCƒH…Ä<§f4––õ7èÇ›öPÖï6ÊP+y {ôb 9Š*mÄŠz}¬*ÞŒk•P¢âü§,EœH!2ãþIêÒ"ñ‚Èã]Æ öKÛnoµÁÞâÛt@ó¼Mz úÀ°uh;5âëàÍÓŒ\A||HªlWS­¬ý ÎF *†£×Hà†"WyŸ‹›–§yŽ’t!tøxK"Í—'e½|•Úîª42 wþ“'§¢²q}"_¨r°§ iB=8ðáp7ÐÓ/ ¼½ƒbúÎרØCˆ1F¾26 Ö÷Œ™‚èßYbúµùÇ›oþTÔ-;ÐÎñ ;…¿Ì‹VÉ Ë»d•L‚ ¨»/ìÄïw+Ö-6óú‘X€v’9S~ø^¡Wl;9è*Hu ^éZ´9ÇM¾Ú Š÷ïÕaµ €7¢þ\#×Òè¦ÇWè‘þåß<©”í¢ï!]O˃~,í *1²n¢ƒâXg3Ž)Ê`¤2:&mc€f(/WL¥p‰à9ç-±°î‡ïA22X+<4]ŸE ÷‡Ê}‘øøé[Ówl\ò¶ŒEîæ"£’)x—>_…õ7Ð+»ìºe½KY .?î ¢½bd ¶; Ï7przSmFAŒ/U}ÄŠÅG²ìsªMò³Öëk‰÷uóèöŽ÷¹iy䳸í²Ú »™.ÒVdÊ7Cã?2Ô+R³sÀö˜Ë ãq¬N.æfÏøScí¦ ÊîÛz"Ñ}†0»®s›OÖ܉¹hÛ†…|F&ÛùÖrxÞÞTÎÚ:ÿö““‰-´?ŸýáVQeÎÚk˜áQ¸>ˆ£ú¦)í‡3¹QS7ã-÷ˆY¡¿TœŸ‹éNZ§­Ü–„N?Ì8Rzͳ³žuž#`ãu ëw¤o^çæ[DÕAiÕdeÎúS‡ŸkU"«³éϦHÞ¡À’jë1X½ç‹_TíÜ=»ý1«;ÎUc²L|å%Wux€¨äL—=Ñ,‰Ý±¿3»$•ä“ÛE{N¼ÕçÉ S-”g+ ÀŽØ#ûA.Y\Ó$ïS˜‚–í7·œYe¥X±õ(ÄyõØ74Ðg¸o—êüìmÑ4òÙ ý›6ÛÞÖ¸¤.G,ßm——²#?öå‰P7X3ú|2­ ߔڭ„q 6üîõk@.:Ö(jKeýi9ntÖÅç“vX5»Øý³…T¦WNnÎgú–U}‘ç›~=à œ(ø—{–xºŒ|m Åwr Àct:?)ŠXêú ßÌÍDeÅ:t‰&¡Ë’ÎÌ„_üÑJYõÜ5¡(Íh¤&5kÍ—„ã;Ën-Úù:x2âåãz>c“«äùѾöò¤`WìÐ×"=ßO7iŽ…ØüÆÓ¯Q*î–´P“á”ÈTëäüDY@ã \àíBXS=ÚZÂðÌ“³`©uóÄàÅF>´cš]ÑOdeÄ#mÃÕ¤Bû[Ý‚H`zC—jhú9¨·¢õÑ=,BÛ°GÙr÷ð@½3ýqͪ>âådƒ»Ðê*íƒñOåVáÌ×M| ÷&ʽӞwr–-JäE|Ôµô”ŠkŽ?X36¡Ž=„”J~ºÒ,>.Öºr|•µžœ/ÍËð_êñ )ÊÖÒöñOírϱ9âà$ðÐ@1Ç%RΖðÈSÖ©„Ùýàã¤çàGV‚ >O ç»›{œµS߇ôêíSz‡»¾}ùÑÎ ;&Ä—"r3FãŽRž;a1á»Zœ¸²þ4ÐI]¯é/ÿCI«ù3QÐ-—g›5Ê«—¥³UìuˆïÍ+Ýœ…;Öy#Îk6Z½ââÉZ÷T·>6´Øã´Opg;1±¨Rp¡ŸG‡¤/œ3dãßE¸`¦Fñ¦!Âc3+™Ûf(›Ú4Œ%†ö¶;©óû};ßôˆ^¹ßØmi$#*}‘Й’5Ý`;)§&:ö¥?AÛô•ÇÆU-llÆ*¡ƒøóÇñ˜º‚ŽïúŒí•rÊëá §öÍòM²ùµ³¤"Ãåý®n¬ý¹9÷GÖgN4|°h„{žv|¦¨~¹Rõ;ᇼ4¨ “P ,šö"q–&ˆ? ?_z2Ðya¼U”KE§—CŸªefÌ N¬x]§ëJîjÕH(Oléb,(ÛÞ-=yÖÃüÉÞáÔ½¯Ë-×1"lñ1!5-"X›€b÷¢7}†˜¶3¼¹® aªóZiƇÌëMå5жԜÏè¹G’šô —}“ë#vdË•-’XïÛŒ•ïüµ•èY4pȸ/E.-‰Ûö;ŠÝ ÔËé_“d…yòHÆ>ø7I³C´7ë@ðð{¶f°æ<Æ ‹Xòc:dw]r°%‚‚ûí`ÿ;ðy£„`J’òIJÒb! A½Ó©ëSüè)«vž”Û$ ÑÝ{v7‰þx+8z¤²/6ºB*‹¯uï¡sâE}h®ŒAÃý} ·o]T!>N˜¹;œˆ÷tg‘·ðf^Ë3Dæû”áGvÎ×Ëz‹C—ö1Á¢À%ÓH49º½HøØ²Î#Å“óÅõÓ Þ€-ÜV+ÏùÆDÀ£dÁ$ïôèeL3sy“³çJ^Z?©Ã»4Ü#›Ë^>c©GÈðGdŒfpí–6ß\#´ØQµ ×0fWüZ¤5£0e5Ì‚¨àNõ—ÿÉJQ}ÉEUc°OŸ$áùYô€lk Ý€f»•l ÚÔ03 BøQî%ì“ó1âzÀ=¢l‹ÛΫ»^×­L“ƒ"q½¶Àaõíj¾gÑwµ&£y”Q—¡ŠÂ’áöT–ÔxLL–ÜZ sað!ç)öšu”þ6—8ÓóWtÒ¹ò[Æ~ŒÚúšÒO¡ßñ”·Ä~_&÷ÆcGgòg}˜¦ŒàFÝË¡ËOÀ ˆØé i«SùÁ“²%zFã^ωÖEàÐ]ßuå-BÙÄŠøj³ ˜o)šK}²©Qïß4r”µœíÝŸ5½I÷³É°ª¤êÜèS8Ö±ù÷[5«bÉÀýÚŽ ä’N¢Wâ>2o¼ï´#[sö„ãDàVö’Xȇ_ÇŸð>°FÕG»ÖÛâ’ìF" &—EãE. ³‘hÁ½8Ì™X`_fBNZÝÊU¡1Ä'óm s¢±¤“÷BXËÖUæ|Õ8X|%:Gà¾Sý⨗{:çK߱Äådu­+/ñóg[åö•"癟f geK4Ií¬½šÔzêºO¢×*Æ Aø·DNQT|Õcfî/›F'¥>&¿YpÌœ¡Ð˜¶Èo§¶ˆD4Qêì&8ÁHrãôº»§$µ§T‘œœÓh4øŒòPës¢W¨‹Í kÌ®&@˜´j€äÆX¨‡ùnúŸåE")`Ò°ïj§a½“ßÓPŧ‡=Èú=ÔÉ=*-¶Ç Ý…æäE!/Ô’YK%*ﯳæÐÊòôÎ æHrsCd°o>«úßïV¯òÑ¡Äÿ\ëoE7À’Åh`?°”ÆX¼†½Hþ›\Mb~]a$Ö{ËëòùÅP¦ÚÝh¬e6ÓÍ%a“…ëãµr_tG¬šÑ?Vª–aÛ6íã&ÍôÖ’„G~qåmÖò¼K$ówx)gèêNÊ["m©ï=óÆ¢®³iAž†?¸êÈP÷G°üX(¹2é ŸÙ‡ÂkWdU}¹÷g{\†ã?Žæ¬¦WÞßò”»ßóqß(q|DT€³cùÑˈHÔSý"@ü–k¯ Oì¡ü§êΉx_´@›¾û¢üâÙùǸr í~Á{.îÈÐøê\™hEäHËü¥$¦´1 ¦#`'ž~Pª¾¨Â¼YŠ+an€¨øº×ìI†úNŒdÆgÍʪK¾Ã4ýv•?.cυ©RçWÃOì·šÐwá"Ð8Øe^ì–²ðt|‘(CºÚZ´8ùš”4(¤<âÃGGO²¬ðš XÁ/-‹Œ|:EŒ%RQ]›‰³9xèSn39q­ë‡Þñ9ë{+t<ÙÌ 91Ö¯h,t Û½d‹'f¢Ú1Ö¼#Ë6àYÕsE‰™‘Ùý¡“”¢ éP^‡qš­–à‹[¬wTÌZ·tæºÐLiÞøé̉׋w’Žñ‘ ß>kÒ[¢½3ä´j˜½“5'Ü…£ÎѦߥ2_…û™µ½G(ióľ mtÛ ÿ•‡­KΞ`(‘3ö‚ªûÉùq°B¨Ã[øÓŽR¡H°15? güCæ¦Ê—YÕÝK8{Âp  ¢›V"ï”Ü俦³ "]g˜ôЩáJN^p ¡6HoÄY8'§ùø…ŽÆ! Ï•e|šî2\­èRÏK0 P•–1“$|‹Ž¯]òf|sP•¥1ÿ¶þ¾?*3&‡ ³cÏ“†^j5Øž7Ä’”„BÚ ‚ÜÁ~˜Î~SŸZ¯—YÝ„¡È˜{d+«}¸=›ç\èdmÆDÜ ƒæí€׉±–Ž7ÎDºŲr˺' Ç­³ªL_ã©}¥0©Ù¾èM}Ú ¨À+ ”Rþ:}Ti“”¢çCÖ¶qHŸ/‚Um¾ÍóÆûiY+m•@™ æIz+ø‹˜Z¿xqüIG"¤ø%Õ†H™ DGã‹n›åjr¥aåò%r×åˆziýµÙ öÙw|!?Û \_jnAsu®d¬ ;ËW¹8lßîVc‰xÂïe ß»™ ®£réÚV~y#0ª ¶Òíž•{õ“ÿ•Á9ÃDöÝeÑ{ºé˜#²½RÉFÛ2Êݞȉ­¼ùóÏoµ6¶ûsûb"y^\¼Ÿ÷›ÐðÓûYèè}¥Ž3ÁÖÄhÉ>ÇŒðñ{q NÀH°âÚƒû^y¶â78¡ÝÒZ©ì~ßO!^k-úÜ«(×J¯<âŠg­=PÚàà—^* °h–ãgzbú®îÌ`Åãrªb?åb,%H%î¤'Çž÷ÂæØÁ=â"¿ªa¾´èÔºe¥ÞD͉º‚—#ú”à´R*¼ÔvÒ‹þÃÕ/ÍÜO)˜ã6›v8%>/£ö¤$cØ^*3µ¤bº%½Ö©p"\ä!Ž M§ýbùo«}@²NÂ>}(0ªºR¼B ›•ØÞ;û¥[b¸¿<q¬³K–èÕá¦gs+ˆ>K±¥ËÅ9àž®ûaAc3ÑbOÕ·¿6»‰ÙAÇL|1‰ÛèQºàÚ´øó¾™õ‡ÀÂ'ê. V—år÷w61Ü×UòêHHÍ(_¸öéÿú½ù6ÙÖ?BêH§’<*gxvȯ°«R¬`›äNcå÷/€Ö‚í]dB§dR“”áfemîr˜FøeòaJšhâ»bX/ §Õ·ÕvY½š®˜f¸$@í$ö‡M³qú÷ûÛ;ØLA[ÛX"\¸ ê#âAöo}â 6¦0ßÙ4ð¾~ÏTr&Z £?†'Îûù”=æ“ú1›<]l&ÁH½é](•/±dL4—ôoý¢iÒ€ÝbÞqW¿³O„&ãmæ¦N@’˜_Vtè ÑrZÒI>£YÍ‹5`ʤ X‚€ W°ñH‰ù¡’áùNÖ5 øhÿ#ÝR~|C"r0A_íãÑñl|ŒšP+CWJñ­Îtob—÷¢Ï¹ñϯ³íǹ;{ÍçBWØêbªÚ6¤uëMË@ýöøOSçµ·ƒŸð¸rºT/ò“CLFÃi‹¬ïÛÅL¸ž*ê…TCïͤ¦Ï.¸M°õ¤x>ŠXļ[ ´¬@£åîî»ÿ•I ,ýÎ`Ýšn5¤Ñ×ççÌîz½l¯÷-Ð ¶KäþÇVÚÌÒˆ 祈YÝÕÛé¡[$QýÏjñ¥ØÂZ˜ØèyÊOR"Ö¶ —4Oïo ŽO?›g'þðÿüô«s—{ ¤kxXs´˜òëA_Fû…ê’F7à‘WÛ4rÇYRÙÜC½ÞhZÿ`Q+ZiÎ:lŠ›@RkµÀ),¿óÞ=(V¤Ûè*i±„­“Çt>ÛŒ²÷É'g—$%u°ò$<•Ú­–vR±]Pl+§ ²6.ró̯£(?[JQ;{?táÉo7¿†„®wW¨7“†¿€™8°ÑÝ\`Z ¿ôz*³ùë·$áÔ7˜ýÄ ÿy݉|“_';¨á×ùòoÏUµ¤Äí1ãû°äDàJlƒ ë8áß+p£t“x=ð—u™ÞÁnñ~é¼€ÅbÇ1§¡Ÿ…OèPøÓ¥3¾Á)ªûumJúº;Â[}a¯[§R—ž¯¡rQúÕ>ÍB-Q!üwç§­ÁG˃("7?YšӼРKåãô¯íå4ó_ɘrêÀyk·òô~ñ4òe¤:”Í™î§U TæŸj]»o“þ(4»_2 Ö9Ü} ¾÷;´'Rœð÷Và_Hbð™y#(Q¡Îtb©qº´ÝNÐÁ1oò©„Å8Ý"Ï­I\ùøÎµó\‰¢jKN³LVpä°KÇÖÄóPzÓ!#|Ù5™ñ2º·"g‡ôl±ñJý(Vu¯Ï“'¬¦ø~È™¢ëǼ‡«{ÿêfíÑ®0ӼŊOH¤Y4þÆ Þø« Üæ¡ŒhÞ˜–NË‹¨ô*7˧lƒ×{ù1Ť«ÙA3Tíή‡J2ê#f€‰Qüth»Z8Õ½¥ ãP®%âM­û+.a=Ú5*æùÿžª3ôf#—P—`¨•ýÖïõræíUêGÌ(tOßÔ"«nˆ:ßœž qÊ¡‚jßÞ‰çâÝÔ¨Ù:¿ž6a(Þïû@äJM%ä{°þà"B`Ý ˜”bq Ø6’ŽmVPó«J€ì5X¸YÄ5ý¸¡ µ%Žd5V”6u)]ii»}æ7»2kzУ ßÑŒ’½«˜äÅò!|¹gmÖÛÖúv{I}N@!¡ÆééÊŸÎ¥£P–S­ üjí·‚Q¦wTq£i®,–ë8Ñäwh½j¸&ÛL0oôéB•\ Q+2íÊF¢Ûl Ì™$;Sd%ûoŸÀg”W 5]x¨ò)ŒIÝ)GJÝ™Üv<;©ˆî ?ôºdz;@hò´ÇÂç$e x-»1Í(’ü95£ïÇ>©uhgÚÈõÛçB ¹GU¶boNy¦©ÏP®*Î.‚ëFµ?8•`²Åë™g`øtcx¿ýâ½#6n!4V¦¸˜¨l­žÕ‹Ê™l¡_±È)Q©õNnÊ:Þò®K›5¡6TͶ“Ýx iw䦆«ù;êý«¡W͆çWÂÈîô­úÊ¥èM¾GœŠ<¾dè(œõ±ÈÇÙi܆¸Á™„džúU¡&L¡HÂ}áuFBTö»5ñȃÑ]·}Ûî]¼9ƒQÔ<þª!f/LÝ7ըɢG-pèé7/àYJîw®(SJÚ#~Rcw#„[ùb®=V`Ÿù6W¼òÅô_üÔ~ž¼ endstream endobj 138 0 obj << /Type /FontDescriptor /FontName /HAOGWG+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/C/F/G/M/a/c/comma/d/e/g/hyphen/i/k/l/m/n/o/one/p/period/r/s/t/three/two/u/v/x/y) /FontFile 137 0 R >> endobj 139 0 obj << /Length1 1435 /Length2 6144 /Length3 0 /Length 7122 /Filter /FlateDecode >> stream xÚuT“ë²6A@zQJé½÷Þ{o!%j(Ò‘®Ò”^Eš EéM@PªT¥J‘.* ôÝîsÎ>ÿ¿Ö½+k%ß;óÌÌûÌÌó…ƒÕÐD@ÉáUGÀQ  4@EOÍ$ sp˜ÂPžÐ¿íÄæP¤/ —þ„  Falª`¨‡€´ý< H\$!-$’úˆ@JTÁþ0g€ €C}‰9TÞAH˜« SçïG7„’’’àÿPò‚"a0 F¹A½0!`O€ ƒ¢‚þ‘‚[Ö …ò– ‚½|¤«<? †rC}¡H¨3àe€>Ø ú‡˜`êóýËa‚pA€‘PÆà ƒ@ᾘ?¸3 ÀT˜hé ¼¡ð¿ÀºøšAÿJ÷'úW"üw0AxyƒáA0¸+Àæ ¨ëQ(~îü öôE`âÁþ`˜'Ø ø}u0@]ÉÆ0üÃÏ‚„y£|¾0Ï_¥Á´Y ðò‚ÂQ¾Ä¿î§ CB!˜¾ þ®Gÿ}rÁ]~Ñpöó4ƒÃ|ü Zª0ñ¿m®P@LHJXB\õ@!n‚¿ ˜yC;A¿Ì¡ho„7ÀC sb~ˆÑ¾`(…ôƒ†¢ÿÓñÏ1p†AP'¨+ Nüïì3Ôå¯3fþHX ÀF³~ €Ð¯Ï¿žì0挀{ýþ{Ä‚Æj–&z|(ÿË©¬Œ „EÂRâB¨˜(@BBúÏDÿjÁßô[ Á°?×û”Zp@ê/˜öýÍÄÿÏjpÿÑ àŸô˜…†¸ÿ½ÿ¶BbBÌèÿ¬‚ß!ÿ¿åÿ•åÝÿÿ¾‘ºŸ§ço?÷_€ÿÇö‚yýA`Ú…‡#øC- )Zê óóúo¯ Œ‰ÜÕó_„ùªÃ¡Î†0Äí¯=ú{˜ôž08Ôá ûõ΀„„þˇ‘Äó^ñÅLë· ŠQÕ?KªÁ!ç_ò€‘Hp1fø˜“ ÂèÔø{½‚@8… `è…\Hâ_3•bFuE‚=a¾Þžà _€ß>I)€ 7áìAý§ëõ!~H$Fž¿ws¹¿Ï¿ßPh B×¹yQ9)l–dñtŠ£È)ošž%Àtƒ—ò0ðÖÔñÉ{Ê‚±kí|Ä¡{É"ehëá”ÓéàÅjSaßn†» ÖôL¸Ç”ï&8ÑÊ[™Út³èв¥’)Öïk ›Ó”‡_.—­‰!ômÄX«<¤œ±§ŒJ!}gޏq¨!Qò- `÷ÎqµŠž³!#n,§­•)šÝ¿šÔkÐN &&Ѫ¶\DÂ)bÓ_óD"I¤?VšÓO¿ÈNyHJq³tôÊÿ¼—›òS<Ž„6–™º ¾‡5 N¹GÄ£àªÇ¼e–~þO^“&=ýéÊ9“Xmvq@ͧZc¡TÖ¿²<ÕÀ¼Ï´>8m˜(SÂa‡Ì£P,]Á&¼N) (Ò Ì¨]¹·¿ær¦-½+9wŠöOW×;*þþóÓK¶œ°êÏøwžøWœŠ»¢5í^½ÞÕÂÂ"ñÒ¶ô¸Mä—B=½Ý/¢QÍOŽ_¢°ykðö\³çÞÿ"C3¹*kqi`¹@NTAë}Ïê šà#m@Ñàìªâ#\HÖ3.-š‰ýô³.Êâ —õýŒuŸ*>:ÎÊÿ8aò# íH«åÜ–Ý•ó´›fÞ{<~ù²1n§TíQb ìãÓ}›T%û̱ËÒøA›,Á[jÆ#2o¾o'ˤ“± ˆs÷?kй£ Â:"/ãêÆÓ9M¢nùB$7zƲØ-©H·y›ÇMÅ"û;>9çN_,Oè´áÀÛ8jaÎgõÎÌ™>^‹eö5-2·Í[}ê'P­sx“<¦üy’høúçw!è2K¼K²¨Píy﮿AœÂ×ÅDkʱûÃúìZµ\ضR Cç}1†µ¸Þ©ß¯ð[w:`5{xÜëÂoõñ°ƒ]ÊŠ¿=ORj~`£ð ütø:4+ü½„þGž~Ô1O©}•š·ýÂC=›[ˀɎW»¦.6»]wÞ/n£ n"ÉUf„k»¬¾C>©ô—â—õW3×àœºl\Òú¿…ט¸š*»ÎÛ»´FP—ZTߌ!ïÑYtýÜó;Üš{•¿srªß“6›ñ“²ºœå‡OhÎøv¦˜/µ]èÍýpÙÕOíšßùç×ÎxZÿ8(+"ÉyÎr.×aaîfu‹Óø ]Ã*þq¡}‹kW)Î4TŸê”\èQêEÿáÙÄÚíº+Úºîg·GVw¹ºã@4«"Mh²C=Å´{'çó¯5À‹¦]J¸NØÃ #BÖÊèõÒ"Rç™aí“ÌÚuùs€„ñû÷Guª%’Vñ²gSÓnLÌÑõ˜’˼©ÛÚ«á³MSþ΋ŘÑuÃG4LF¢°ó—c¡€cò£sV$£bbé«îÆþiÕÉ¡=:éÂÇü]cX‘2HÈÚŽÉôgNy°•<a¡^C¼Pä56›åO™_—½Ã.Ú+zÃ1)Puàð56BZmõ‘­-Î.—[ûÄ.àóRþqŽŠRe~—»]‘Üù¹Žh<þÓšRŸ\iý¨ô[=ƒ)ðÀ$Ÿ¹åÍZ¹ÿ"“xž·%þሸT–uBÝpi[{|TÈsGæá1jÚD”ÿWõ­â€¹Q¤0ÏÀ8Ë}æÁu5\ÿz1Ü~ŠÅ ðg‰‘R»VöñkìÙ‰7§ß±ÃbÑX_X‰Þ–ÉZc#R€(§Ae—ÖÛ‡FÍ+aùýÅu“ÕÆñ!¶$Ê0^}ŽÃ°]&²*´Ã$§Ó|¾ R–q‰y÷ØÅ8íÑà볈Šò»¤Ä7ÙYR^µ!Ç=-vïѶ·Z1à_ÊÜ«;ÏoPÐLÖ§{‰^ÜßÄu»S0 •µÀî kü²ÔðˆnþÆÝk$Õ²l#Š!'f‰ î+Ü£&.½lÈ|üø$òÞÈýT­õ˜ Òg+³4IIvLèëzøqÕ¿ÿ®Ó,š{8ãiEªž©i~@c|óàvB7ƒt†µÕÚÑetë7ìà}¬ûh k,Íúëf"K¶ ŠhÓÃ4MgÒÎ]¯êY¡-zù‘³ûχŸûëâ™Þ,g Ìè²¥X¿—Æ$ˆµ½ð}ÇÜgal ù&ʺõØ”¶µ¥A¿‰è†ÚÃÊ”é›æÞÓvG½¾ ¬­¾-̈K¾³N§×8•T#•6M[o$´–ÂLÊ/XÌ–^o±0îéÔNRZ@ébÌ ™7,É`ž¤@cªg+cæÐ³¡›Þ¿wÓÝ1ãÊTüIp06É[ò³«Ýè‰k:I„LŒ;Èe…êÀÍZ‹ëÐ9¢%³Uèm³Ä¯$Ï™ÿ¯á‡ ןH àã â{ë›Ý-D&Åö‡éÇ“ùYŒ§]Òöš¯Ôzmsj¯‹`ØulþÏ̲ÞÊHo|±1újü~CjŽÑ˹3ö4 ;cØî3,€j†á“#Ȱ㲞Ý9[ßùCÓÛn…j‰È’:嬋Ì/SC‘íJ~~oäѦýªv­×»C‡Î«{èG§57&Ö,P4ߢô¼©b a"¸©Š~ðÙòÓ‘z[ÛI¾ìê+n© €x:¨ÝÞ›»è~5zO‡ýÎéìÔûúžÞfÊš'rvV”¼¹)3È(£+«…ÑÂûhdÿqå‰+öгžs÷{ømÞ6ÀÆNƜܕ§e‘èªmåÖ@¼[Pr’E±›;þÏÚ1§¶?²G”ƒ dÙÁ´ydÃîrhSÓû÷š³ iýÛ”Df7¨½·ß©E2ÞÚZ/§V{„j=ª]ˆºÊÎl·Ñ­TÞi=µÞµíÂQ¦a®Šs‹ÌC6ÇÞbP+)Œ¦q¶§ê¨ìMÀ™‡5½ôË “г)ôïÛ9:ÔÓ>³ÐJ2cÛndçòD0FIqHRÐ}ÏqŽ6Ÿb°LVMâ[ËײªÐ³Ro‘±v¹ÝÐf/n_ŒC_0xUíó« :¯g3%L1±iJÜVèu'É5<É&Þô$6hôtâV·’Ñ 8[®ß­ö“BMÈn`½°«­ñÚIýn¨T‰hO:ä"Ÿ„è>Ñ-(Ï\üÜ•Þ×Oe Y UZ-Ã<e'—v?Ø n”¸YÖ-Þt‹Ñj½{ÜÞØŸ Ù¸­€ßH§ÿsÆ|Rkb>ÔThŸg¨´=:}½V½ù3_'ßõÐäú£uÔŠ†±#³i†«ÈžfèjÞ,×TǘԱr!Â#ïeÕÃê9…{Èj‹Äb¥MSßÀ‡ò~…¶ë8èdøþöJ ú”-é3Žþ¯„‹Ú_>Ì·¿~ÇÇ=ZgÑ@ØœQù×?jå™´ó¥ J‘Ó¸š5ê:óÅ7=rñKo[ö:Ì˨¨Þ…¹ÏWã«Þ¨Ó8ÏŽ±•ðêkùÚ açûrx¯¢FÚÌne|Å+‚îïÎ]|kÁ’ŒUä•åUhZßd=NMîØº¥ÐéÏŒWM€¨ÞùÜëû²‰´¢r=xQ§—Òh<øI0‘Œr§‡øÈÁrhnt½­séÇ¥´ìí¥â{{ÛÑŠPÞ'`ñÍR£]ÜùKüNkò‡¤]_i«È ¤¹Tq×7"Ó¿žåš¶ü,ÚÿâÓënŸ(°9Áe/…ÑX—H¬_M)˜¥¤W†?Më –QG‚W+Éñž†¦ :欌«p!^°2¬q%×뎞(f“<›²šŽ »5v—3“-SõÙ¬×)_~È™ï'µ}Ž7â“3ª*+½ hzžH±ªëÔiÿz Y¿ÂçøhW5{_¯¶¦ç5.É!]Çx¿9Û8(¶€ô£õééä>Ï)'ö§ðÙ`r2Ö·‚È ³¤rœéòQž~7ƒ—õõœ_·l> ¿š!Ù‰7 Zžßé ®£b;ÀSªÚ­n<†"¯6?ôLÁUÖ¿•Pˆ,Ì›X¬³ú† ’JtKìWظÍçg²žE³@ˆ©­ð>'Å$‘w|<ƒ”R/„úÜêÃ/ÈÊ´m>Ëän šf¥M½,·ÿ4¿‹Ë­û´ù&HLdÞÝ?.øÈ-a’qz[jûezmÂfÖ‰*ëc«ìä†{-ï zV*¥ùY_>¼4£ŽQ¹ËûÁÍF#¢—[4yË ö¨“¯4èQ¦8Û¸¼±±…4û¥.G—U>MqÚê}çª~›%‚‹êÚÚ8õbÓÂÒˆx×·–Èñú®éR„Ô`\†-ú ºæðºÝÃ_ê˜ EK¸hƬL]ÖXfÔŸÁu¦~íÝ?UöqŠ_Ò"qc7«ùdW½eL½ "ûQ·€Q©šEû¸ªÚ‹à 1ß|v )*ˆÜX(z1¡º^úüîb(_’w»Æ"=©•vLÚŸ×qèD‹¼Qîý }TÒ³ßÄ¥G$Ž¿®¹=¥Áݺ¿!È`Æ‹`ÿFM´¬È ¦Ñc‰uú¦V/ôµ¦'g’ÐLá¹-`Õɪô}ÇnÜ­¨R¯á¬c å c Ã7¾9ªÈGÂ>ÖÒxîkõåÛ( =úŽ:J™ÉçåJ¹$¾{œ>¶ŸŠÖsw|#¬VÄ#â!Ö]øñ÷cz.ê>¹$5vÞ>b ss’ŽB­C#É]ú{Gk—¸òS+á0’iQz'/¢^ŠR/ØyøÆ“áÎ  ¡ã ¡‘ .áï±B›3 X£þçž@;h¢çãñÈð}¦ÓE¦~ÖZ«¬¬àœ¶·—QêËÆ#›§G7økž8z¬r$fkù9/à¶WÆâüè–ðIG¤Q /ëö¬ÊMRÎ΋-,=i ¶SÉjízÐîiÄÝõ.Yñ÷½CÒ7vä¬ r¹¿ðU^:îj·Š‘†=É`Ž“pv»ñ§k¦ÊÎü#¨qz*z/W¸˜b\•{Ô¦O¥AlHçn"Ý .,Ðù‰÷|&y«‡¢RÄš5‘Iþ(E÷iâ²Å,˜Lý©Lmd@±ÐÒ:¤«a‹I)³.}!rodA—Ѧ–Y«#‘EΙØã‘=Æ{F¼0æwlHùÿš";VªÔv¯¹·’j3çöNaK¢œø¨À7®mUurW'9O†7猛DÃÁ2©²:ûå–öÄãadŸ8S­4O™«X¯®üÔÇö*×z±¿ßznÃ2ÒÏsS+yL9UG‰Ó)­ìÛ ÜÆ_X²ˆëëEkæ- ‡‘·O<­ëYΜz§u6_cŸŒ >zÝ›-Ìs©|UéÙëSWKô m‡?H:x)>dN9ºsBD»Õ(jªÚ¸zš,áut2b¦š°ô8¢|2 æ[utДs_l·æöRÚãîa^\[PZ;*Î6’¬†OqÉä«„HMojÚêªøÖÚJñˆ=ˆvî% ìèï’éà Tf¼B¸t¹>ëòø$ramàv|^:‚Pä/Ü3j›ÇÇ¢£orGW© î²+N„-P (cÏuhq Jw@+…ßé[Ç5&Ä£ôÀ?<ßÞIJ„\ (ô@<(ÉV Å·‡K:²\‰B°8òß¼:-Zwœoì7Wší}=lf‰£ÂªØL¯S‡<‘uÝÑžþTcNü™:1içöãÃê2jÙ×PäÂñéå-ñ’%‘TlÊÒŸ ,BsÃqK¢íºWz_²V@¼ò¼ªMb÷4É-'9|_= xÃûõgVÑÃÞ·ë2wvÖ}¹k’Ù •ç;nG‚vj[Q½ßÖ ÙPúâëQfüǨ(ÒkNèÚ†/³ïsµÒ ð» ÞÇW†;9þé¼ev7ÄN°TÏÅ7“ª7EVÐ9^Ž?Wû^Ñ3¿î.¼vx_ñ¼åiºaI³~ Í““›_ÆŸÁÂ) g?€˜Ø-ÇýËr”$—ºÄ‹œƒ½ÊÓÝÿŠÒ8p3¼È àæu¤²»¶TÅ$]«KÏwàŠ^JûQ)›ÿP3ld—?³öÜtôns5 Gk,H\¥/Ê‘:mïøN—v”fZ)V%7‡â % ¥.”ÄïgM=•eMHÙo^[”K—Y‰¢7 ƒä(­/Z.לktÉ5Ç{Ú*Gåá/?ž(_5L •£e©p» ;]ÊýfLà ÷|rã}2+þ*~A­ ã—8ý³ìñö6 Sƒõ'aQ0~áTWrBØûÆóÙú²ªWsöq‰6RuSt°±17‰JßÚ€6뻟Í;—2ƒÌÃ)"ʯ :¨¢=f|ÞHù¹ìBšÏ´yXÛ—åIãýãmW㉄/…,.»4 ÄiH*FØ]rs[˜·ýg)ü#oï°–Û‹û0d÷¶¥D…^ã”(Á@ÓƒwHé’qåqc·´9VÝ+äG m} XʆPg ÇÀ¨d6¼(²6’È’ÞqÙ5Îyš=œf€rø,ÜÜZ‘³ñClïæé3HŒMëuCM—Ëž¬’¢Ðû—ÄgC/ Å.ëï<¡òMèÜy;y¶¬Š.?>¼–—Nš{¯`þüqs³èYªLØFH†R¤UÒ‘!íø¬Ø-zÚ)2Þ‰ »ZmÍ=ë'üT‹}o¸Æš©€,Kž6ŸÆ ? |Þ’žOfž;:-šýÙ™hÎ'ÿ삇šì]êûB«¼®øÂÑ©ø‹cÔHÐPºßþÅÐMwÝL—CÅ·íÿÙ¿I endstream endobj 140 0 obj << /Type /FontDescriptor /FontName /OREXSM+CMEX10 /Flags 4 /FontBBox [-24 -2960 1454 772] /Ascent 40 /CapHeight 0 /Descent -600 /ItalicAngle 0 /StemV 47 /XHeight 431 /CharSet (/integraldisplay/productdisplay) /FontFile 139 0 R >> endobj 141 0 obj << /Length1 1782 /Length2 11107 /Length3 0 /Length 12244 /Filter /FlateDecode >> stream xÚ·P›é6LoRŠ“bŃCqwwŠÜâ»[q(…âPÜÝ |twÏîžóÿ3ß7™IÞëÖçºå™¼”ïU„Œ!†`qˆ Œ™‘‰ "''ÅÌ`bbedbbA¥¤T5‡Yÿ#G¥TÛCÍ!6Üÿ²±ƒ`O2QìÉPbv°0³˜9¸™9¹™˜,LLþc±çˆ‚ÍrŒiˆ ŠJ)±u±775ƒ=åùÏ#€ÚˆÀüá'ýî!k°½¹È ‚™­Ÿ2¬*#s0Ìå¿BPóšÁ`¶Ü@ ““#Èʱ7å§¡8™ÃÌÊ`(ØÞl øM ²ÿE• jfýS¡19ìÁ€'•¹Øúäâ`c ¶…ƒþS¯'¨jö/ƒ§°Á§ÿâûDÆùø_]4r°·jóë÷Ôâÿà?®Y0Øl„:; 1âñ·¨òo¹®"tbØå}~”t­ÉÂ0𫇠ë›Ô_Qùœ6'S,>ÛÃ,®gÑ./lw¾4sæ¶^MZãÂvÉ@*¾mJj5ýxùl*Ö튈lúU\–F‚0 w}/œ"q(fŠ€±i¯ å+÷÷UÝ?b¨p Å9¥ZäÛ:* Ódq‰>¨ïÿT†5¨n°Ÿã½Ïøié·¦ó)Q>…L†·í5Æ©s¥p`«‡¸3ÔGÚÑwã˜"†%Æ3@$ʆT®¾ýòWFã^ ¶›¡ÄkÁ=ãÞ/Ë4 ×Uò§fë§{ºï¨ã:8Í âL˜÷]oˆ±lp„“ì3s6CŒçÓƒHêûùc¯ êÝÉæ’]K¥?‰-¥ƒd\榺'V~9È$JiõùSÓËÅð=”¬ï{ÔR¶†;šÂ7½¯HÀZ£ÃM£Ùõ0ñËú¼pI0¥Á¢Í•d֢ʟ¿ðÁ¿É[‹•ÃÑ<©Lí%°v«)OëÒÍ•í!û¬­M˜jή<Ô&;Ìu¥zÁ=%7ëöˆøµÆ(¹2LÄÕÉzõÒ½ÜyUW¶ß’çQêÚbÍ~°2ƒá¦í¶‡ø‡ã16}"+_ë ¾ )2†+”Ïür{îƒ;ÒicEu,ô »Ýs½ßƒOÈ£ÚÖ¸>/V|ã>ÙnÊÑŸÝ‹Ï?KÿxÂß”¨à-Íf-Ä4|¯ÌZ7ÝH‹3àÖF—Û´Žýk5ÀÜacL.n]Að•™Ãëyê _¥¿˜È°¡•óÆŒPšo•„Sè-…|>ÈŒжÒöÎYM ÎÒø^’j·é“>>Bôß ©ç.ßöù!Õj‘àï×}NíZî@>ŠC¢iüãr)ŒYNÂä›K¿Iò_ˆz\#ÏŽîÕŠÙrPSÉlñ±ÌÏ´ðå‹vföÆ4éb»A›Yx¡M§“–ß‘‘6“EÜq•&§Ëµ)êë$3|•ú€lªç^ض “»&ÓzÐÝ´>j/¡›«úÎÙ¶òÙJ;±~8X°¼šÞÕ=!:^ܱþÃ0MDrªp JÄ2”u,2Ò2“^/Ì0&åØu¨„6À¦/übm÷õsì…»¾Ñ[æ#º‰9¬oÏ^íkÔV± ÚBÒDïÒ¾|Bhsæ4™‘¬¾Õ&ÕU»ì´HQñÒÝ¡­¯ãîR’0­¸Ö]5bl«Î3R?K¾S¿Zð1®{WçÍcÌ–£š÷j–›~íÖ‡ÜH&’vaš‡Îuì ’fX9¶+†%àm®Ò·e§^~]Ÿ!Çs¸Ôm㨠&U€xüûU4_ð®Icﱅ̯ˆo– V—K”úçÖº˜ÉÖzü27ÞZ‚d‡¼n’TÇ™ÝÆ}^àØd~Šøž.>çf¨Ú:Übò"ðçAf©Û'ÙÉ=ªàŽÔAÍÏÖóWn¯k—ù|^ϭרfþ€*;¥|[åF "I7¡OQ>5muâáºÜ4¼ôðÏ ‚mÎifû5‘qzŸÚÁüWLn=:°7'âh*¹U”¿iÆ»¼ÅæMƒÖ°XÄ+赬ú5aRv Gª}n Xq㟠f–”‰XC–7˜F)K<²lçßo¹™á²ñ‡`‡ é9‰æ ‚•Î_ö ì‚̵&ôVÛkmÑÍYyˆª|„{…ˆÁww$ýçèd¥2F;›ú¨¨çü¿¾þR…LðåÒ½nø:të’ë=KnûúE²Ú((ß»Êû3»ËGVn¼Ïõ„¼ZnÌ×FêRøÔúördן`ÆÞÞ~‹Åò{,ºFæ ÕK¶³G¾#âûäfG—ùr3=ÚNI™ÙJwóZŸœ^×Èíå°ÒÓÖö„8²{0ç}È[‘AÆj©Áz ÖÚIñˆBà(bññc:ÔW¯ÀmCšptx}b:µŽ î{°:äwŠYÒÜåý)©^чèìˆÝúÚÃ{]œ"•,Õçu­«¥¨’²!Œ§ºÿåŽEnCÏgK´uyüK¶Òo/Øn’Ï—«-„hS L'ƒ”Š+«P¨ó·x†|*”0ÔHýEl ·Gy®»Eà>€w¨Äµ©CQ³ÜDÝä Œâ…ÛÃ…$µ4韗 aMš( k,ïf:Ü0-^E‡[cʨÊÙ/пNÞÔ~ÿâSɰ,«¥þ±ØpŸM3]{˜ÉaØ—_lʲʻêfo‡lT[ÝuV´ZbÅeê@.ÈêVU’hÀfòÁ%µ¬Ö×÷ø  „¸ `%u×%”ë&¯oº6ú,Ê—QÅÄI|Wc¬ d~€çéŸz¶;&îG‘ßæ­ö(`5cÂál·’IŒâ‚ZåäQé+) ¬¦¥¼7Œ·Œ-ÿ¥Uéq¯³(;LtΩRiMBÐq©¥±y+Д®¾í+]‡Oró‚Žt€Ð©ëº0l’Té8c"E»¨m‚Ï/‡Ëú†²å¤¨ 3Ö®¸¬Núy»ÃÇwaø$é{Òݨ’4DöJ—QµÄJì—9^ªx†²@þQvKì}¼+>%§l…N\ö_Áä„Üñ3ôÙjú¸²ŠѵýƒoåEÊ6Uä.¿›}ǘã‰Õ7v¼¾qÂJ;cA®ÑýYt—k;”âô›ûiú»zñåÁᆂ9ÙX}/Tc¥i)Šº¦„n­êx¸ÖÊvÛ¯¯r¤µñbÐ'|á¨þùVŸù6ç›§­y¯ˆH…þ¹s¤€2ÚÙ´ÐTﲋµïšì³Æ1Xs&)‘Zöáþ6JþkÁIZÜ»(UœQóbðy ¦ s Óa¯*$wl7ñ^‚$_!ë¸æÞwa'¿‘ø‹Y¨?wûÌüéŒyØOµ2^]ö~’ÝECÆÝÕ{òl-ñØ}®†”k!l¼ ÎWæ:Ë:ס:-íÙ²©]ÕVûðÕàÞ1óÁÖÆÞO ]N]£8õ’qëµl;9i§¢´ŠÅõó÷“hÿ<¸ T^s­UÏ¢A³aÊf\FœÔî±>ÍçÉgSú`L±—*0Œ<5#öý‘UÁP_H/;f¿³¬èH»PRH×}¯|>ñ¸22@6ô ͤÏy“E”:ÁwØ&³jn:mÞ|!¼xl¢( ŠÑ=Ûî÷ó–Ñ~ ¹äLL»ˆ¡ä¾~ÿâå¶_¡kkÂF¤}ÙóW»+“yÂÅh²ðÀ¨-îaÜ ÀôÒ •·aúä(³“.ÔÀ5æ<Þ=¥)–º—-c¹€zçX¸êž˲¥¹1½/Á2oõü®Y†¤Ì:tQpI?WèýæLÇǦoØÈôÊ`ë)Qm„ÒðŽp[Ï¡ã²ãOh/4b……\g/¶ÒÛäyÈÖ“=ñ^LĶÙ.·\&¾iÀ‘Èé+?žQѳìÎA:äÀìZ¾šê70Xo³à¢Á¿…ûXéýø –OI{ËA¨óã¬Nxs5v.šwpwC1ŒÂÁ¢Ô»£oÇ©@g^¸d^–+]<•^ÿk Ъ–×;t ÝÖjúlà‘ªÓ“ÙE:¼!I¢Š®nšQÜÌÁ=kèÉ%¼Yå„m»c÷î]Œn‰FAV×L9ÁÔ¸s^"ÈIt‡#D D³l)T'˜ ‚W5bÅ1â’áÖp4Ëo—rYŸ—À–•7ÌS­(˜(-ßÝ‹$s{ØÙÒehö°ô”Ü:(éÂ’àëÑûböc—Ʋ<µÑ…Ëð[ôµ Tïíµ³Ä#ct ò«=<…«À”¡‡åkâ8ª -ãô #?åùŠ[¨I\MNîko#KfH%ݦrÐðæ2¢w§ØnáQ(½H™ÝY7Uà½æë=©œÔ¾©$DûVùº.߯?ï”ÁǘÕ&{ã·¶ûcSeåi&ßEFß¼¨,šnÔx~i8ufù"Ûot@(““zVWnë &ÎåB”žOéÄw…øJ«Å±q>¢óB=Wv±0|WŠß!£ƒ%¦!K•ÛÈV¡cx³{(Z@k°Ás鑨ËЖÆa 1³)ªm7~€œ×+e„ñŸ×?„¾K‚èÊÆ‚Tu•ahû¯\ÛËßÐv^©ñ2ˆMlÄBšU .¶³¯—×·ÿje‰kãÿ²/ãlÇÚ·òUV@\šº5î>¡1CÌ”3Ì·)-XÞÇxÄ¢·b_ÇAM,É_ÞžS^¬žâ)¿úÑ?“åïa­‘¶ÃpR+·*4ê 5€zD¥‡ÍÞwCqôc¡-@øÄô5ñàvÿW 1b]_­C—Ìt$ Ÿ]‚H¸rSê)¨—NôÔX_—Û¥ÔL]­‰<'ûòèØÚpañ€.”q»0$Dü­~“±ÊݦÛOÈ#ÉRÐZ%`Õ^Se„ƒ½þ§˜wuþŒœ3¿ËcŽø$öGòÆÇ.Ü[Þ7Ü–_ìLÄARG3_ûGŽô!¬íçqIPb}²mùº×W/wÂý-þÖæ¾‹ÂHÅ{ý]qLzS|ÈÏ9Á€w užë¢(ßȹxн²NkÏ ˆÄz»|¦"ÇÙÑ5>‹7Û{3®H<áúº±-öžOoÖ’±åŸ€°_hÝgÛ"Eœ¦I7_ ¹_Ýó} ¸@:éßãH¥­çˆ¾¸0³uñÎ…\³=ôÒ{R£—±ÐØóJâI,iu.wIy½./ÜÙ¼"D½Ø=z¸¡üK\)æ´í–i®ü3'Ýåi´(‹õáƒ@:ã<™ºÛK·‹ ÕF6_ÌÀŽŒ®Èùì£)Ì•m–{+=Æ+ÎRfb|ý~Áyº1ºöÕ˜Œ¦C)ß»ƒT–Eü2rŒ¼ Ää‡Ý^m”ŠDÔï1F÷.jÊ>–+QOÿHj?~ñiôwoÑTœÊToí¥Þq]šØµvslè–‡Ñf Ÿ¤a‰&WÄ éóÀ,§–R?ï²SÔ=/¼9JÆG¿^<¨"x† ¤A]ýjRII½pg;*:/IÆBõ<‰Á6MæÈ³â“áÚ9ë;>±3¥¸Ddê=Q^žj‹ŠžÁúÛåY©(qvÝòPi D숼JˆQËÔ¯¦{lÆÀÅ5Îï"z"™«¡Û8|‹öQ §8iVv¥Ô,º‹íÉò`ÂÏ2ÛqLÕÄû“¶Ì W'hzáI§/qÞ`(zª<’òÉ,> ˆÝlʦ¦RUÓ}ʱס`A¡Ðž :OÛ²¶ÁíýÔRË*±ó{^K oi(ºB+Uãêëï…DÑÚBщn0Äò.­, m”èΆy”£ÃlÒc>±¸Œ–ãúÏÖy4oËìüQWõ#ÿÂE°–o+ÆÝæ4ø9}WÌ-6£»¤›OêbçSt¿Áñ×8Me磥˜(„gö Æ9³Brä&QIüÝŒY¦x£ZÏÎF\é<ôµÀ†#Ÿ‡Ï¹*¯©÷yèÆ|ýÍ^¾ÇXÌ{ºÄòI…BÃxY:h«.z¯èùœõÃ:¿¨=oŽìXÎâP@²ž[ãðßÚ}ÞM¤qî"7ÛŽd­‘¯Oc zLpw¢Á’2XyUÚm¹ƒú ­ ŒÒŸUmÉX„©6][p~ó¾xß‘^FÂ&МÑâÏ.è†qi9ëŽ$Çô¯•xÏñ˜¡ Þ~ªÏªç-¿³'kBo$#*ýŒ-5l97ì\”'yûIVÑøÎÒƒIŒáxTh5º%O—1ИßX«O/S—Ȧ$¬i˜†={°åV¾j§bWõÓ=iaÍZ.P¡$¢’kp ÏÑÉC1¼YkȧCrz#“ƒzÕ *NCÉãX'¶Ò°â}Áöp-7¨O «œìù>ɱœøN@ÉÕb%`<=w5Ϋ-É¡ÂHÑÿ¼F+ãÐÕs >í‘Pö– s*¯T‚îcÇP†pXårHû._2-Õ°*ç.O¯ (ò¤É•¾v«ø..Ò7u¤U½­ fÒwõù™DÙÔᆴ¸|î°0Bz£ÉX&ì^&«ú£Šš(¯z“\I?ódÜU)-i¤8¾ßi"Ca÷|õ –—+„Ýo oáKµq ¦˜WìÝ U7Žï}¹×^²ŒL5‹ ß׿T DÐá|9íÀª)¹¸¼†yh¶¶°ªú¥É ˜ª†‚A Ñä1£‚3…sŒóÖIĵf~fÉ)—ŒŽþ!Sõ;Ÿiȹk¾†ßñð9wÿ ºSÄD™ê–ö_ÓªˆN[EP» Õ’63_û¹ÝSÙX*n ­ž–ähÀšiqn­¬1ا x¤T Á;Öï:‡*òÜ\Y9¸ïMwò—‘Õ–hÖ§” )'îæ aõÞIÛü!k=ýž4I( 1¨_Šú5ØG€dC«î–É…Ô`| Þ’ô—6 ´†•X’þ‹Ît"&¹ ¹™í„êeÏéÆ3åLjþ<ÚÑ[”ÉÂýö`³–í£&w)Fv‡N%Åœ«i_ƒ“¢D%¥þºê_HWl±ìxñÕz~Kþ¾8rËå`bdv Zeq¹BŒ•nï‡ïæk»¶&sËð6pmì}̯ŒKr£dzö°µeQ]ìPj„;ê(ºßãrì¼ÓØ[Aº{S×'ÕxÄqM'¥ÚIþ0ê’:tûL2ÁÖ¼qóQÝÈ,]ÏŠj~9K¼{‘»± S!w‡‚| ù(w!\(\% <o-UÎ9ö~ ŸL_6ñV€¨Ë›ö–;hšéˆE—Ó™z¿‰Í­L‘«…–ˆÑ‘8:dÈ{Q­+«ÈtD{]'s?CW/Èfd˜ƒ‹Ü~Á/Ã…˜Õçm5ˆyÔ7¡ã‰¯L÷†Â±’ š—¥ÉD¤ÐÓz¹Zén P.êBdäÃR0ݱq¹Ò–°Œ8rt§|IÒlόޫèHÜ«èg¾ÙM¾±‘EÄðº-£”ÌRuå"ä iÿ4ýVîvš˜Hœ Ï8Ô¹ Ù~c;ò­KÏFÄõ×¥¿N˜Ü‹8Á1_:^RFêÝñs'?,Gÿ¯sQÔfñ.ÂóÚ³91…ï“ïÚî©àº,=•œó9^”eÊ´ü0í½oè„9-›Ë Ô=<7ÚÍä—s_Â’Fw{ãQ©ü÷\KèŽäð\ÐÁáïŠK“Ü^s(;ÈBH8¢5Ûªj½ÿ`¼ä<£$üŒfÙ¿5*™—Wœ–ÿvÅ”«ÒÒ±/ÄÕ^>Ãjù®‹ôÉä ‚8T9F¯ ºÓþ3IA|VGà>QFáÚº_{Y?p¯FGnê&ˆú®–!™qÌÂbx$M—€ÜA†\ÃWƒðÝÓùU« ZY\?Ñ&%Vyó§¼ëöñ|ήۀqž7èŠá•ÄÕÑÒmk3Û«èuêgŽ! f Öi ó­†} }.ð|>FýxÊgíþæ¼×  T·ÛCµIü;2£öФ*šRóêp>‘Gê)ßAÔLîŸt|újJ-‚Ÿž5œœ1 ~ eå¥Ú—ÙÞ;#vJ§ÑïÂÔÄ:TRfÿFjÖ£ÜAíQç'\oÚqÍTyZ ÏHÌë’ÿpû)²Çì4%Y@ÄmWŒË05',nŸÈá†.ÖÒûum|©ËÞc§ÆóΡįlq´Ç­åÂÛ Dû°s]°[çi¶!|=Ë:ö¨Ë~Ö;ïTØ£µ+¢Ƙ%Öàl¡t(é&?œ>ø˜8-_“¨cmðN©œû«ÜËÓ6lKÓ AŠ=Œ|ÿÏǽ­†BrÒl½6Úõ§äáÏês_ZYwOÚ²šAT˜Ö½~F¥Èù p7o×KŒ/åœV€M™Í9Œ÷QÍê>(†¯Âr¢ëpÓ~·~ÀÆüÜ*©#í+ 1 àŸ¸%„@·.×ZÊ’´«¢×‘aÄœö‘Oó“ü¨ÉØê»r³˜©Qmè2ðºd+´$‘±+» c…®‡7¸jÉ+w »@yÑ·€·,…©,ÞbÀ!lô“el÷‡!Œ7ó}Uv´c‡âaL7žNqÖã~4Gªï"åüô|:7¢%µÈ>òÜÅ<¶š ¡ßœ™Ç„ þâÄdÅ‹Ÿ2~Qªï¬›’êŸn¿`úÜ 7jT²«ýôäÇ\ 4«všßèa-¸ªãŸÜNžËá,Á,å‹ÖV0õ«ÅEnt "¥FžÊô|ö F„+&Â£ÝŸÖÆÄRžTýd¢¹¸½ˆñžÙýa}èÍîš²+IQæ]MA¯ÃKàbŒŒ¹ìœOm¸¾}|lÃ0 ZÏ9?ŽmÕ WüyúÜÐßPs²`ÆOÝŽæ°OD¿Ú81žzª”Œ›Úœ¥µüÕŽ@ÉUд€]Fì۱ëÒ°fD“ ñë£6™aì5”ŽQe8»ÛBAjz´]þþN’âç•dgDs=Ì÷!X{ÊûÞÎH“x,qø¸8² Dä‹ã²üõýHr; êxRà‹p*CgºõŸwð§}rìy´>¼kâqöwþ‘ÙÀ†ñ”;åw”¢Â“ßZÍ"ƒžÓYÔ_zÒv %·8\jAn#+ê•Î"Ù×\V9”ê;øö“O^þŒM$0¸¹½èg€Ü8 oæÔ=àa÷,jÆ*¹X6·ÄTl¸è¨à¼«™9åCªPj½MRºH3²õ¼éw«šøúÖ «‚ÂdŽÒïcøhG<ñ›6:X«)5vÖú”V˜¿×‘ÄW3QGˆ¾VìÙå7Çs>üê$Ö™i›«w6sYp'Û­ì2"›f=Çœsõ«âm#gÔ‹‡ÓÖ½ ‡®÷É‹ ô¯XlÚŸìžEP·Gð^e?ø¿z+ÿ…‡H@ðb÷ósw3“­„_î×{²ðr$s§ªæ·Îì/éæû?øÔûD8)ï¾wØ’žB™aj©¾§èËÙDkè9œ%­¡ßbp@6å¾¼ â8ð«jºZNã_{›†.fÂVœŠlRñÀ ’±ç·Úµ5 ÇuÄÛÕY¦©9i†Ú± ‚Dl í¬»BŽŒ…ï-_Ä BOw÷¸Ö%q¶Yroй¤) Òd ¶ŸÔ« õhSj“´5RsêéX(é,‘בWVPa{,G°êôÒùTŽ`7Vá`Î$+%øÅ´K~âõFB9'cnC:ïÀ}Û‡ ¹ˆ*ê—Ÿ˜:˱ZÙ¾­zˆJÓ'Þÿºh’ ªe8lÈÙShôÏw¿?ËÜZ~ƒ!5g¸"Av.¶[\ðPMWÜ]P‹ÌžÊýqÆžÉÌ efV×lÎ-¼ó:Ž(¦5­{QDû'~aeëÕøñw!ÅOߢƒ]ÃÝ¢R<üâqʪÖÌC}FMDÛ5¤„›$D³'›–¥eüc[=ß;>k­¢T—Cü8®¸C·(fä­ÏM j•¼kå‹éD‰ÂÜ?µÉ·ªŽ¥pUY|dñáŒSW£ãýð¹Bü>yY1K'”¹S+ò±†\Îh…)×A·Í¡}þÁý>#'Ô€G'7}­¾-Eà™SèÐù,ù^¨Ý¥¥*Ÿ-|Áö#ýËó>!yÖÓ½“×}‹f}Õ¢;-¥ð†µ^µb̨Ûñ÷ºæU?~•¥äç¿Ð±±)`/¤EXV–£ÿ&Ó5Í®¡ E"yÐ}Mð\y?H–|ñ«!–ot*u¯¼²gºD,íž´õ}j(/“–Nýút͵’8åú4f z_ªÕ¶Çû˜ºì-AÊÐgP‰ªçöY§QºÝw­KhÎÐo¼»»ü…»! „; ‹™ÒŸPÚËÓ‹ÛÅ©¡žÒ¥)Ãæ­nb§{Ê¦à•ªªŸï@áµ×:gæúJìò]®5Щ ðzÍ?fL‘¬tÊľuÂQÇnÈ‚LÂU:™©ÉɾÛeº»*ˆýrcQy;`âÑSPÁIV>Æ|XÀMÅiu»žÙyæ\Öñ1Sq¬:‚°ýmMþu­³b œŒºׯH“·íÁñíгJ︶ÝúQÝ(ºêš!r‰<0'R` ºêÔ³T¿Œ\U˜·šËØñm"^*qh ×~ã­¯šjzœ¬î™KöºYË–È4"[5Êq±g aÛ –m'Þª{¹‚º÷âÚ#BF$hkDŠÛjî`}ÒY?ñø^7—ºïEmŸ×ÀF^åúZGßm‰Ê®lZÖ#GVmxàôа¢F­PnléX¤¶?¤Í’$ºOˆƒàOk‚OÜ¶ÙÆËŒ|ÖÊVvUÔ•#4 »YÆöà‹áçViÙ³/£¡!ܓվ BâZ8=ytûÒ$ ­ PÐ`×@9j¸Ð;`FÔÿ^àòñ…5·ˆÅ|Bã©©›_„þ¼ë\; «ëè ”P. ‘ET³“¤¯@kG¾+¼ÙÜG—Aœj5ðl"ö®ÃÉ M= \>2ɽ”â"¤‹Á"ß5µŠËoÕTA9ÍqE7|UZ¼ê52$o²JÍ.A¬™«úŠû˦¿Qžþ§ƒÇãQ{~°7X²ñRóð»mÝV³î¶f¦(KØy]Xº¾IâN¯_Y<ÊóMê&©£_%8âÙís±í ÇV5x¡€ýsü)›àÃŒø\‰"›}Ûô•onbøqo¸ìxŒìUƒ=êš—ïâý>ˆ÷3~vŽ+m’GÇú°vš2ûG¨À@;˜û:§Dä@…7è£8]édqõ%4ÑozÙ=w{#ré¸f™·¾œåv˜Kí–€ÏÈ3dU¾–‘d­ÇûÀkò¥× ²÷UG!Zâú€ËkP‘˜†õÈêóóÄÓM>ÿëÓ«Š; åàý®)üeƒˆ_ɪÜ×>Ä~Y¡<;=–vÖxkã[l“Í_MW:â½bƒ¬ý–Øý÷¯ Ñ0 ƒƒV–‡8%´' gÞ·Ž³£S}à°ëwLKâŒtõšI@Hê ?L~o¸|Æ–|”òÐ3ÕÝ7ìüÇã…\¨îD?…Fb¼Ê¼ÈkwKj‡TàU9äµPM^Ðnltò솺­IVËfa^M@k§†C4^뢜p]Ÿ°tÛ!¦=ÈxL…°¨i=ØÏ\å«Üì ]Ñ}|I-ŒÑ4UˆVÄNÑvd'°ûkgÚò°Q:å¬ìÜ­ÊbܯӑPHÒð,K"”C£Ru˜ïD¦,辡*Ú†CˆL°zÏP 5“sU­Ãeyv)7£ÆZ|¼{>x點a7®î÷ÄMù2ËÒ¡¹Ç^ªT#?7wðKBzÃÖE]¤ëx²'šët‘Ozº™V*Ĺ„cR§ ìv‘L¯gWŠq|—û5LÞ°dMÙÿcO\…Ïôõè»Øœ–¤c )­Þ‡”#s`)ã—äS…²ðKøßõ+ì χ3ð¤îÏÕQ¯nîÜ·£¤|JÃßÓ¼W¼åÛ¶îxÌü}}ê?~ªNJÁ/OÝ‹-Íu~dWUØâéþP¦/¬³¸¨w¿´Ñ4{¼ R HÍ+û@Ç8GiÈgá»G¯˜oAö(¥‹9’£‚Ĭ&àÁì¹u_*:Q»ðaC>¾_Õvñ†×ÍÊRyá¿Ábˆ ³ÚD+#˜íRÏç™içOôáå£÷ƒzZò®[’ˆHœ¦ËUâš³`ND û÷Í=ò/~büXå/­‡;ŸÛƒ‚uæ±ó~ WuTdu!¾òø±þPòþèå/’–[y/:csŒ©°7±f9;+Š y£äiVr#ê°“.›¢…¡L{”ÇϤ½ß€ˆÇpGtQ8œG0ëÒˆ÷„„õuÞ`Ìa½á¦Å­RÛ—½ÕŒ`KçŽSÛœÃ[E€êÕ-ÓGõ6}?—A>É~dNH;Ø4v /ˆøëélhŽŠ 6ÙÞ–iËdHøv¤Ò6çtð£Ø)I,¡í”½?¨å ‘‡¡´`¢=Òk’ÃzWzÃ7åN_ØÞÓ<ÃÅG”ƒ·Ù:=õ¥˜ÈíL¦™Œàz²¸_w‰j_ Å>L/ss»l\%[{éÿŠ%ØS©ÔS ‡RÍ܈© Þâ²ý¯o®ò«-=cÕ~ZñDe§v]1ùÖ@p¿w˜;Ð ƒ™Rj)d—•4¥ä-0Ô\º¶Ô£äíc+ÛËÞä‘Tî(^"'‚Dv#öÏv¸Åû ïõ–ÆÊê´j ¿•>#§1ú±ö}ø }깇Ãêg»O¶L:_¥iáÔÄMÇÊ™ó|„{úC Ë 6†´*žë_+ýå£úÈItØ(÷ª“B±†a3ÄïFß©´œ¶7bŠ»÷굋úVÐÈ6|1¿#ãʽ 9eôõº½9›Û>£23¬šUåp<{é’œìP[±O^ÿî“zº¨ YCé}‚äX~ä önåoÛnç½ÛõˆŒ8¡d1„»&P• ƒ!š­[ QãlSÒ¯±pb;òúƒJ{¦Œ(EÛ(ÝÏ2­âb’—N]JVŒî uö3kš‚Ö‰q5+Ù^¾_©4òÆûô¾[’š…¶9“[{»V\ka˜Z-¨aj¡I§ûZ>[Û¾ôÕÜ~¥X²WAOëƒâÛ:Ó`¯@…UîÞµg؈t?Eƒ¤dˆ“÷P]>Tû¢‰#Má.®”É—ýa(6B’3mшbDkwEPºßðÌ@G7%™ðšù‡ÆÍû ³îæ+nÀ‹¢à‚ýQÁä¯ãc ~|~áëíçšý¢÷ý[ê.™)‘YÑúœöòF°'>bɆv‹Õv”ݯ‡ú„/e)à€dùRÎÂå¤9wßzb‚—Ú#w•[çÓð4µu7vöKª" WÖkCºêg¹»PM>4ºUøŒ{… Q­W‡~’ÅF'È&íœ>,ß¶yÅžDËÌ µÚíºg‹¹ÿž‘هƻ7¦ÑÍê×£ïº7;¥(ÈòŒšÄòéew"%¹.fÛp1ôìS–)÷T¤þ~Õ ÂxÍ-m·s,&œ{‰²ÝÛý¤Á!Ë™iµ+5.ñ#á={颬W?¼»=»¼ÐéØlØ^ éae-kXè­ä–Iͳ«¾þ2€ä2ÒûÓ% p ÓÜÑÔï&3qukìCü·l´+ô§Ž8´7ôIŠòin…rìLÈZäEÈÿÁ8G endstream endobj 142 0 obj << /Type /FontDescriptor /FontName /GZDKMT+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/A/C/F/I/N/P/S/T/X/beta/comma/d/e/g/gamma/i/j/k/lambda/r/s/slash/t/u/x) /FontFile 141 0 R >> endobj 143 0 obj << /Length1 1527 /Length2 8003 /Length3 0 /Length 9023 /Filter /FlateDecode >> stream xÚ´TÔÝ6N—¤tÝ1tJ#H7HÃC 1´¤ RÒ-ÝÝ]Ò©tƒ€ Ò|èûÞûÞ{ÿÿµ¾oÍZ¿9{ïgïsž³Ÿ}˜è4u¸d¬`–E˜œ ÈÍ+SSSðòòsóòòa11éBá¿ÝXLúW7(ÌIì?r®üÑ'‚?âÔ`N€î ?($ãåðñòŠþ sȃ< V5nÀ ˜Ä ‹Iæìí µ±…?nó¯%€ÌŠŠ sþIÈ8B\¡`@ ·…8>î9t``(îý_%X%lápg1OOOn£7ÌÕæ'À ·hCÜ ®+ÀoÂu#ä/fÜXL][¨Û_~˜5Üä <: `ˆ“Ûc†»“Äð¸9@GY á qú ¬ú€ð÷Ý€ÜÀ—û;ûw!¨ÓŸd st9yClÖP@CQ•î瀜¬~An°Ç|ê²|ü99 (£=ü›žØê wãvƒ:ü¦Èó»Ìã-+8YÉÁ!Np7¬ß瓇ºBÀ×îÍóWgí`žN¾ÖP'+ëß$¬Üyôœ .îeù¿!.¬|68@WD˜_„q@¼À¶<¿Ëëz;Cþ¿Ý ü|aÎëG?¨5äñË× äÀ]Ý!~¾ÿøo XAÁp€%Äê„õOõG7Äú/û±ù®P/€1ï£ö€Þß¿¯Låesrðþþ§¿<ÏtdÕd9þbü,Ì àËpñ ò€@a @øqá÷ßUþÍÿ_Üÿx5AÐ¿ÏÆûOAe'kÈû‡ÇËû¿eÁú÷Ȱþ{ uØ£–!Ö¤oÂ+È ~üÿŸàOÊÿŸîWù¿Iÿ¤èîàð'Ìú'þÿ ƒ¡Þ¥ì 5Øãp8ý/Ôò×(«A¬ îŽÿU†ƒÇCÆÉæQâ\@n^¿üP7E¨ÄJ Ûþ%¤õâq¨Dæýýâ’ý—ýç©@¼ `¬…YX<Ä®*¤å²B†Ò“kkLõ8ùÒk,× Þ«0i¾§“ž¶¨R¬¸ÐT4³kW—u¹üômîÌw³š¶Æ[à‚‹Vq׆Ö2vöáq:Þ÷ý,^B–A¢,Xk/‚&u$~¦”•M¯Þ+–ªîùxOf¢BEá¡-êm…iª$T¢ú‡ëÚðFËÝ-Ás2–Œuǘ×Q&IêéU$Ú®°â<ËñWû‡– z¨;#ƒ^xo0¾(ˆPÉ  —ë/gãÞfx||êkùœPV‰Íx’-¾0v+µ×à% ¬|Èqûݶ{ŽxW ÝAñmšÌ×j!<ç4 ­‘¾­ïÕ•gAË/Îi†ÈƒÐZbÊøO¤?ÑæþPjá¥*wè)Áè­SIê\;ÅÎ?bÎÑúèñEÓ}ž*ÇÚi¤îGÌ;$Z>f>2¤³ú`\ìÙmÿµbò*ïm’ÍÁË{]+R©?v²Ù[û©Â¦grÃàЮp™l5µù¦\4ówD<¢¢š¨éÚõ´ON­s*¹Î­Lƒq1?ß­”*èõRÞõ˜Oú[¥M?`¿¶ûETzõ'3Âߢ‰Õþ½›ô.+ÆÇX#¥ºÝ¢Úð6ß½¢gnÓó…¹¢Õß~'i±|`êÊ5‹z2^TÁRút Ùå“.1æ´êOÙ÷L¡èáÐSâãVqξ\U‰MsÉ´$Þ[šò7£WŸxE¡a¯z‹v; …i5µsÅX¦z‰%Vº"â$Ûq-Oq×> d`¤ª÷û*ß €úÆ$CßµM°H¢Ê”—#9*‰^á¬X²‘ZÚÌ°Ï !VŸ‹MR^×®øÊ7¸ùs²q¾³1±ª[·ªgÿhaÍ ÎÅØF¾e³4ð¡w»}y&Z> ËüK ½Ž˜‚é~{܃êò–Ç<žÉ÷†Š1°èhøk}Än–n‹8˜HØ(Îë£ö6‰Óµ§o_óÿGX׿©èöd\rI×Ìp×wíí¥‡d0)–ØPi¼úîÃçú•ZkvsJ¶|m“<)ù)î\zÃäi´àaŒJAà á¹(Ù6ßÇEÄ:™©Å‹ˆr"‚°¢§ -Õ0‡ž…ú ÅMÑi¬‹ïüfßUà‚è #žÄpª§Ä(±!D *7w"=ÞCûóô‹}zY÷˜©ý ‹b¥­!ÙÆDR·i¯Å%VÔxgm@ïq`¶iX)b#5ƒËôŸ«ðØì’³&µZöÞÝM{Ö~Ì¢ƒ4FDŠW£Ï+j )U©¡¿s|þÒðATËÉÊå~GXθH` í“)}¶e<ðmXýö­ò×÷OT+7Ïp«S™‡/êZšÒŒGãoU .‘kŒF;èÚñT¹ÛpÖ æÙrŸ\ÁgÊô‹ ‚Îoq[6;à n»IÂîÎçz˲õÙ )Ò̸^*ÙšÆ^Ý…«u[3WxZÃ^oÖ¾¬6÷J©ÓÈ®º|Ó_c‚ΉøX|"’èÓò†Pg•´ÿÄ—Õnœ[EÛ(êgû-1ƒû›ï ]ò“œ´Çff?„÷¨Bê¼Úgv2î1B²ð h)[ùÔ»ž,78’<|×_ñ¹î8+êt”[`#ýÊžfÅ-(/´çkòc<¥"%ë"ÀB–#v!!6›²@˜AÕ¯ñ2çZFYIr6é]Ì{x}úåÄZh`ƾoJ¡òÁÚâRApÌëðÜéÊ)åÙ°?Ç.‚@ª,»ôy-3ùgÓ12Ù‰€ºg4YËš0\€Ë냆Ê+nŠ®JIL£)BÜv¿B• (Åw'YÏl¾3ß/xV䇇N¿z#(g@É–•œtý#|tN(ÅÃïœ(_€O6ü.RoáäÈ™Ëg¦Sÿ ‰ÒR¯Q‹!X÷X–Cy{‚ö¼:Ú\:)ùnm·ÒI'W~ü›¤æ ýôIWÜÈ¢,v+§"xÚLéÚ6™¼](p•9µõÀõžl©PÒ §²4Åé 7A‘ZS "FB9cô ”¢XuÛìµÛrœ>«…þHI†rœ<Œ„1´´'öd­Øã¢Š‚c=L”2[«ÁßÂÚ„ šW§ë+q·¤³ÖJÂàO8¦£ ¢#JüÛÏp7Ìpš9I“ÇèC’èj[æ ëº(·É1²WŠà ß¹^Ò7RÔ“‘…êçJ7ø©Ý:òJPïkT [,ÐR¿ÈcŠÔüš‰ŸdùaÄé}û–.‚¸^_‘¼õHܞɭhp#YÖ5DÈÒY©w·ª¨Ì!gÑÕhÏ^™dãs]ˆ¤ŠS‡!ökaáy6‡«2Ù•K¬Y#O:Â)8“çÄ©Ndƒ5%‡òŸL Èð~#ˆàW~™¢Gˆ+»Ñ®˜’J;œñpqBôJ¦~®Ñ§©_}[¾‚H˜ÒŒ¨ðæbT¹1éyæÇEÈ FwÍêk‹Äò¹äq²ô>èÃÛ`±<êóÖ%!ä·‚í¶£Š0F‹¿ÔKœ!ÙãdÌiã œˆ5Çï‚^ØIևޥÏË8NPZ¥Ž(ibL×9‚_çëÿºÓÚ8 ‡"Cm9{\ñái-K+º"l{£‰%ˆånÏ2Çâ;/ ýÁT‘›¼/çš-S"¯+]ÉmøCìhÔ“hX‚ˆ$ÚÅuÓ×{â)_âš0̇ßÌ*ÒE…¶!,bIagóÑöùqpi>¡½ù¹w¨º$ª¸MÓ‡¼n­ÿ5úWn›@ßà€a@{l[õ—St.”á²á…JöÇ5?Ö8=ÜX´¶q£ž÷zf“ƒÏ°8:ò^Ð(ZÅ“3µU¿-¡Ù“صQP‹ÜD¨tЊsÖ"qf6uY"Ђ)¿»­ÕOD‰Qã3õÄ\:YhKo×ï^UZ/ð¬+Ƹ_a|ÀÓÖ÷¸}ov~†rR†‰x1òª§ˆëƒó¼×ó£»cä¼wéèR,è;‘6í>ó$U…ña—'ûœ™[”cÕ ùeêäKS?…š=9”Žâ&ÌŸw›mŠ=~FiMgJ5?÷‚ srJÿù]…;YÓÞŒIô¤·\LK™ÛÉNwÅ:sDU².%âhà‹£6ÈðÞî;f®*£~æ¸=ifÏ2ÉQ¢‡Ã‹Ðiê“Èj‘ËøH¥•6™àyáÔÓ è\8?ÄcÑkxds+œÇ›ÜÚôkÿÙ%¢‹Ù–EÜ ëª.‚m–êt?d‹3þ{”νÓ¬Û›À’Ù%Zÿ®=Ÿim;ÏK÷2\}¨Â%ëu· 8—ñ[Âk¢÷:i_4ˆ³÷¥ß—¶GÖ6SnzPû{KFл_³½5ST ô{Á}ùEšzHÀoeØöaCžŸÉžœ3ˆ÷$Ÿ¸3µV4+‚’UBþöªœŠêýgͯZ®µr]ÐÏ)óö¦.²þ´Ìy¯‰ÂËî>}0©×Ô50ñciÄíÛÇô/Êeø¤Ç@$ŽY¿w R¹ÙgçîÂNÑkDéB.祈¼B&’Öj?Ôâ—·#v¯Ç}ºhxñŒy•>B6ø“Lüqîí99Í_€ |Ж£¼^b‹íôžJP;ïdƒM󃉋Çàüµ!ùÜzC¿¦‹mnÛ±b[‰xúÎþGRX{Ëò{꺰A¹ÚJÀù*ŠÂlÏÂÛzœ\›tˆs\áa³ešç¿¨&*ä›7ìq½ Ø3Ú²+"%"ç„Ú‡`¿_ê·B³àr3بyo±‡b2ÉÒw¿7|ÔJ·®X33‘b])"Û0k½Ú³\ÀʆP]žB¯õn•ÉrBš¬×…#ÎL³^HæË¦;mqß*Jˆ¦ØsVT»×‹Jì/ü@ôÚ‹ÓæµpÜÁÞ¦ççTiù¾j+"´òìUÚœ—›L\%÷eÚ³ßfÌdz5}zµFì·VP\Ð¦ÕØôá"€6³’ý_8U&*1ïJƒá ¦›ÙÅO½Ôdôå5íàaZµ21º "Ó “ðc8Ųd ³¡[æà2TíhüT>õ§}ŠßRõhY6½_m) ä\.ðSÞRW)Â2-ºLݰÞâD$0µœnù,Éã´ !†åN}L>\ýð©ÃÄ·ªé@žËYcP"z¢)4ô«_ʪ‡-f=ìÜÍ4ìÆqïÜ„œ»ž …¢j²7Ó‡ó(Õã+›2õÅôæ<Ò'T2b”üV² †ý¶’«óYóv1ÂpÜçf•l¨ë:OEZ†^2ÕÈØ²UîKP”™Ì’F’o›ä0Ç-&êx4]Ü3Qª$ií¿]glçT¢ßQˆ¾clyG¬„!?ÈkämËÆ:´#~j,žX˜ÜX1q;M°ãÍ©°zzµ:£ÓûeàE@îè¯Þ3ôëM ®e(³œ aœ¯»Ť$Ë"™æ ÷Û>­’X½6ßdô¦îÏX,âˆdcCõráUù¹´×s·<1× /È„§ÆÇæ·ÐãóÀCxÔ‚âKgŸ{6/þ¼Â‚ R sX¡PÐúgÿ´dï—Z›Ùe_c™.µêÏØJre³Þš÷›GÑŠYøŽJå?ȶëÝ Ï¡3g5!¢m-Ÿ_Lßc.ˆ*õ¸èñý¸âÚC&ó[¬Näëžä’Ëmuh_ z¨l+\Cõ¿Ñ]7ì\5ÖTŠ»$àÅÑàalÃco5˜„ |âÏ’Žìü+»ŽõlŒ\ZB¶Žvb¢u:©kVN F‰$¬gv -#GÏò U¢Uùè{Ÿ‰–3F¤S‹}=)ß@KXÄÚc)â(Vt"R‘ÉÚ9¤¼’š»–(v¯iSY¤<&?¤—¼Eáàæ'½kn2›'à>øò‹Ô"|ûƒœOœÿDÄ:ÅÛéËÏKü]ȾŒƒ„zMþ7…)Эn†Ñ7ÇýcuÌ’ޱ˜ãºU¿–8QwN¢&$G» K>>•K'1Ë^·Ã¹Ù^U\·©:mð5ºr¨…Ìzðx®8÷¾ñýy2v`1Þìלš,À_K«ØYÕÍÏ: >‹§T¦+‘ÊÈ‘H/¿©±êR òK}§¡Å®ÍÎ87 þ|Û¹¸€CÐ¥6ÍÛ“xt¦91B §ï”˜ËHxÑè°ÓÇm¦–Œue±T~>-ój™[MáÒimï W-;´t3¬ÄËâ 5ÏU">àˆ¦‰%ÃN)¬Çëg"VüöêšÆÅÁ3/ã̓8Éw.N¦Ñ/¨$”ôZ0¶ Κ‹ºÏŸó¾ç ¤-a`ß)}Œn<ÓËüDM„½Ë;û½^þp}]E¬Cò–õ|>|?Äð>NIM%ÿÆ<— TÞ!¿¢4—eñQëJ•:¸²ï5áuS{úŠÚÉ3raxZAJê²VÄ7&•ýkh÷[Cœªô|¢¤ÛÀ`Ε]?ŠÂ5®µk>Ÿ½«ÏkùÌ›OðEì.Ù1ܶL€E>ÿFV‘ã)ƒ;Cã[U‹åz Ä™âgç1,êQûõ!Ò£º"õQ’²&{°†ËeýÛ`Êf6åÅý8é¸Ï™QyXެ’…Æ.õ./i܇¶x5ÍyÞ/H-Ò»Ç7…ÖµæŠKGÕÊÇfæ·3Þ&FÐ[ ±ªG?øYé3C/(×(˜Ø™Kˆr8¹·v" ‡õã`Ö¡`nñÜÎ/ßÞ>¡]ñD–;3 ì8f>,a²»Ó“©Ø^ %snÞæSF õ~ËÓzfg}®MPð“B·ó»êÎQoÁ"n á´/†.m/R] kçvQI+9¿d¼4.M|jï+܃ŸH©”yÜ*‡}ï³çš&í¾¥¥Üï‡Ñ 0ùÊø"Ùôk’5BgÅŸêf©Ñ›K q+©ë)&ÏÇ‡Ž‡÷ô„',󪧨rð84>ÝÐ>,ui3$šÃª:Nõ<Ÿ538ðì¹=)Ö’7%Kíö—ñgý@ÿròN/X°ó£ =⤠¼îý4)Ù•îƒ"§Ý‰Dy€!™–µ—ÜFK‚¡Ø“³4¦)‡À„Áf—ᆅû‰C§,üx,Êý¶†vÐ «ö³ñCFtr¼¦[rãÞèädá„W×$ ÑLFÙ» kyÞØg·ýú„Hc?‚-'K€G Ȧ¢~97a¨§OÞð]L—ÄêÊ–æäµŒ ]aÓd 8ùH©&¢k={ƒýòux‡ŽÚÞz;èûå•H/U/½}Šœêt ÖÔü }áx|f ͸CÚ~6º(I¼l š]«aPGÀ>ÿdébÅÀÅ“yS€…î[¡qmáèbÆ*Œ*õò•™$ˆÝ:·üîðS÷œ'|Z!c òŒs;°N²CaÍÌØôˆ¾óäÀ/®6¨û=ªØAè1û Q­7/ÎA»ß÷äƒ#´û"ySÎÇÌ–Õ$3±#5GCaª~=æ2*­À‚TäC¡ZÛÍË.ë×±W­Ò޽ÕîãŠnÜ`·a¹ÍV¸ôIæÎn¡½–Ò¤:¡~Ë¡Öu{HwÈRqK­¡ô…Ÿ3µs@?~óU)tOg³]ÇîZZt6Åä« Ñ™ðRÒ‹ðëãg³¤ŸQ"½y‰s3 ’M>ž¯¬Ž–ö™GIu™æû2U•Ø>$ê«ÿ˾“Ï‚ëb¨.’RÛ@ŽîëÑìÜ Ž`¾l4±§ömÉ=i'óM9g9“à>Ó!ý®t°ê÷c•±Ž{8)¾ "Wl©åŒzR„>'s¶s†-8¾SÕïM5¹Žì›Ò¥>¥}Òù~4îLÚÆ¯„†þDG§QKË=£â.Ìò‰¶ ‚Û»/<¬€KµÁ"” ž¾µÂÆOÖ‹µ ³Øp³‘×ôÜûhüG¢‘øöRÍ-óÛ¨1Q™ãäOâ;;Ç¥ÚÏâ¶g˜1Oq‹=#öaµ.)AåÑÛyy¼lB3ž†wDfé]M4 _è¢G ,WÇMJ£u=§:á" ¦ ƒÁy뢱IJ:×ìèM6 ZM™ÏÇFH˜8wåE=oÊBÏP#F ¯¥F¤}Rë™*€Qaº¹©oÕÎS Ÿëƒ¢œù—ûÛ^òmM?ðûRg½”¿U±M}VZ,fõÛ¬ Ž“åoö°jiÖ‹ïX¤(ÉÒ"£Œó~¥&¡ê*¦‰Äv;¾3+Egßfþ|ÿÓg„ñÙó{%’'«Ë.ÌÕKþE-ÄYzjíØìôDª6ûwHÊ#ÅÏGâg2»‡¾àYåæUW6ì-jb^ÒêŽh·bDj|Ýõ¯¦ ¯dÍ÷gÈsÝHµîjv#ä-ÃF2¹¥]Ò·Æï›œ–=3ÒäÞ¦éV\N÷ajɬë+¥Á¬ÿyK= cœo9kÌÚ5/@æ™FDGš9lCìß»«ñ±32r}Î)êaT.yýô†xç…?°ƒXwpü´[唚R!ÂÂ¥5]¶.âT¦y7o±hPÌôõ¼™#‘“ÔVé³Pkž7É(ªí¢t½–LÍíŒh|é8TKhÍ©[?!M³þ IÞ:Œlÿ=%Ô¬ÝÓ­‹PXáà¶e(F öü±ü ~ cˆ“/¬q›G+Џ9¾­è/|€GP’½ë­—#Ç®Í亪Q²RžÏôÌ®å+wË:^L É+Ƨ;ÌåÊê6žlQç<éLÖj¬³S hto8¹jöÍ {PÂÂ~‡.Øé_¹d.|6i¸I"ŒÈl@Ç"Ò¾4Eêz²5KÀ‡¡TÖŸ#SاÓ<¤CÅ‚±”ù…´U¡aA%Åc8›íÚécê²PÔLãË`9_K]%63— ºN…ÍŸVÓªéÙ9N&¶›!jW=d`ÔB“rºbÖ:"?Yä‚ZýÀ½åð1NUbæiî~â§›Žú0p¾ñƒ×Hºµ|FL¼°¾­û’f…ÊqvIGPW«Ë˜P¸¥'ü,Ãî&»zÊdl{$7 …~>f ɨFð:;|-¡nÖ†`Ý2Û#÷óþeóv¹ŽgZsUQø5î•€)°ï[ ‹âÔ;¨_¶øÆäšCâö¶äØêÅ*6Ç‚f˜õkJ¾kS¨ÉŒ: sª?Ûo{ôÁ³ï“ÕAý¡rõf’Ž{ùÂÙ´¡ÄJ×ÃUо[ÁÑ G2’sØ„#‹ò.¹ýËj1»ÄÙh!-?‹ÄŽîõèñ»Wfꮿ^í †É´M=|»FÃíÿ@)|7†R;”W‡8MƒÃèOó|y7 %ü€neƒ_V‰‹´¦ðv?Õù/nÛäÀ^©RÕɉÞ*Ÿùµg¤§‡jPÕùrýÑêà:º©ý=’h¹DÍér ÕËô´úË‹n¦´Í]t¯Õt­Å«,û8E‡’™'H ´Ó„ã¥óv £Ò(ÝJ …£WÜñ»Çª³$|ÄñöÞ•ßüªeâD‚÷"†Ð¯d'– JdÖ_]Ucâ^u}’2L:aLq8¾îħ›%*q¦«O(ŽÄÕa¢ºyQ•Œ’U .åNÆq@‘2H†§xé:9'pn,[å‹Â)³y%IiBomM ±”†j¿JàãR&°Ü/˜·xïÆãÏñ¥q àH·œ|gýE¤£pêý>´$I…ï·\&{›Ý9C;îê½°p×ÐJnýšËìi"ghdïÿª¼Œcˆ1Ì-|þGö8ç}Äþôk,•)t¤IÜÃF í¦Æ’ûHJ’šNòë°ì¶'Y{yëê Âî¤=ãDä&h^: ¢­2„Dt¶}8gUE]kY.dA±±”Р `öÊSM³Œ u3d€ñµÔ¬ºRbÜ& ¹ç:­àÒ(ž< –¡ë'fRìÝQÙ™ÌV]V ÆØ;Â2îhT—ß_‘xóË•÷ ¯17wÄܕšh.Æ/ j±l.P ~æ‰è†´$†—¨û+¹˜RÓŒ÷Ñ…¬-ìÐRWZvìb~ßX §m†6ѳ`æ_CКp¾:èæØ)G³v»Æ3ú—Y¹WŒÉR?¼Yâ‘Xª„RwkJX &ä½SNïÉ1g'¡›âu‚9ÃÀ×~šF/å…‰À.¹Õm”3Ôh‰_Y¼£åÍqrü žKp©”m¡5×÷ð€Eó•@‹ùîÂwØ óo§ìüݽÂî–ëÐsË&T‘Bò¦qCD5r°ðM¶ç3c«Î÷ÕìÊI­C“ÒøÞùySŠ`Š|…åºeê*­œ }»^²ECÉÔÇáAóµó/PPTàˆƒòߨ?µ_po!Õ÷À§QH÷âÊ¥C®Œ£:HT¸Om@¦E&>»…nÊ^4ÆS|ýFÉLÚ<¬¯ÊHš˜ºŠR锈mÙ?ž 2~uíÕ"Pî™mùK —ÏÖѺ.ƒRìcìÆæ0ݧ’ZòÞQ÷b‘°îî†jlðÀ”»*{9) óÇÍ…¢s2!± \Û…*hø»¾×X#ÔÏß%Õ;šˆ ¥¼UZàªD[Ê,Àõõ´_ëÍ]Að­ü¹‚XÿŽÐ ÉÁ+\ê:ºÜçÿ\n~6 endstream endobj 144 0 obj << /Type /FontDescriptor /FontName /GESBMB+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 81 /XHeight 431 /CharSet (/A/X/beta/i/j/k/s/t/x) /FontFile 143 0 R >> endobj 145 0 obj << /Length1 2722 /Length2 23089 /Length3 0 /Length 24614 /Filter /FlateDecode >> stream xÚŒöP\ ׊"Á‚4îîîî\w—àîÁ-¸×àîînÁÝ·33ß$ó¿WuoQÕôÚ²¶ŸÓdDŠ*tB&vF@q;[g:&zFn€ˆœ2#€‘‘…ž‘‘ŽŒLÕÂÙøŽLèèdagËý‡ˆ#ÐÐ$5tÙÉÙÙ¤]¬L,&vn&nFF3##×ÿ í¹¢†®&9z€´-Ð ŽLÄÎÞÃÑÂÌÜæ_”ÆT&..Ú¿ÜB6@G cC[€œ¡³9ÐÑØÐ bgltöø%¯¹³³=7ƒ››½¡½£?-ÀÍÂÙ  t:ºM¿ ÈÚÿ®ŒŽ  jnáô·\ÅÎÔÙÍÐ ¬-Œ¶N [ # "% P°Úþm,û·-àŸÞ˜è™þ¥ûÇû‘…í_Î†ÆÆv6ö†¶¶fS k @A\–ÞÙÝ™`hkòËÐÐÚÉäoèjhamh2ø+sC€¸ÀTà?å9;ZØ;;Ñ;YXÿ*‘á ¨Ëb¶&"v66@[g'¸_ù‰Z8Am÷`ø{²V¶vn¶^ÿS [Ó_E˜¸Ø3¨ÙZ8¸¥Dÿ1‰à~ËÌ€Î6FFF.Ðt76gøE¯êaüKÉôK ªÀÛËÞÎ` *èma ýƒór2tœ]€Þ^*þ‹à˜˜&ÆÎ# ™…-Üovhú7 ßÑ ÍÚ=&㯿¿é‚ÖËÄÎÖÚã·ù_óe‘”Õ’V¢ù»âuÂÂvî/:VF3#€é×’q€¾xÿ—æßü¯ø¿¤Š†ÿ$÷£”­©€ëï@Íû_®ÿ¬å?'Cøoy;Ð.”¿W_‡‘ÑôÁôÿùþrùÿ·÷¿XþßVÿÿ&$îbmý—šò/ýÿÚÐÆÂÚãÐ*»8ƒÎBÎt¶ÿ×Tø÷)ËM,\lþ¯VÊÙtB¶fÖÿ¶ÑÂIÜÂh¢hállþ÷ýo zk [ ¢“ů‡ €4°ÿ£œ±èâšÕ_* èžþRÌÖØÎä×á1³±  =à@£!6€èBM€î­6€ÞÖÎä•ç 0µs„û5Qv6ƒÐ/Ñ߈À üqD~#NƒèoÄ`ûq0Ä#&ƒÄoÄ `üX R¿+€Aú7å"ór‘ýÜd mŒL ëAÙÉýF ìä#Pv ÿ"NPvŠ¿(åßD¬òòQý@ù¨ýF xÿ".ú +ˆÓÐ ´5NV¿M@)ýF ¦9[A/Sçßr–åŸÊ¿ P Æÿ"6™±5hMþÉúKbcó;‹_ûÃ`ò…þfõøŸLÌ  ‰¡“ùï¹þòqpÝáoPq¦¿!ÈÀôÈú ZüÁ ¿!Ë/èú; ¦_‚ßäl¿Ìí\ÿˆ20û‚øgÇ šˆ¹‡½9Ðö ìøŒ 2-ÿ€ 9YýAmü³4Pmþ( Ô¿ßÌl W[Ðþ¡Õn÷;³ÝÔ bì«Adö 7®íÆÍÊôô¿ÃfŲ:‚Þ¬˜²ÿ%³°û=RVPsì­]œþˆ ’8üžÞ/ätúëyòï"1ÿÚ9MŒ¬ÿ“ ëoÅÿÙ=ö4ÿµçúGú_c¦_;ðÇ™@ù!ÈÉ hcñßåeûetýcŽl 'ЋðßRAq²þs5™˜@Yý z—08›;ÿØpPÇÝìþpq¸üAÃwý‚2sûcqAÞî@½ÇÔDÏßɘ<އúÏóØØÅ4\ç¿Þ˜ ‡õÿð_¿Š€@w 1Üâœ1Oå· Öû*¡÷nt»c|Ód»Ÿ©è¼Û\ß~¢ªLXw¼ú4Ø…¼²-Fy#¸DøÓ먩ömhs¢RËÓÇgýxåÉݸ… ̾ñü#¡š^|X<:UÁ½?>ªû[A6—&ËvpáDTÌE»wë‘p¯é-Y ™ÛUÚ«d—y÷\2E­¥ã_8C–c”1‹M íL‡CzîŽ4ss;š5þJ(Oç}ÍRॵÁó0ë¹Z¦ÊìÔCŠ£…yƒ:2Iî%|,5ïUT°Ò·àÞÄ[@˜…@›²B‡LÀœVi¡iÛÐ]í:²ØÎ´“ý Dñ¾m'±º¸]ß‘X£²E?͹šÅ ˆÐn*¼Úz«Þij™‚·Ä5Uó ˜AÐ4?òkìôzªuXꣻ Ii¾lÚîôÖ(èÀ3sc¢3׎XYæNÃO*q…èÐJ¡#Ó.¹~â|†2 æõ ¸~ÄeÚÊgdÔâ„êÁ;eËþxã]^w.ß¾1ƒ2o톪øÈ.ÿjô=åü<ÇœEÌ =„2'M¬3Ñ=Ñq4;'9û>ñ*’¡xQ|ÞÌXœ-)ÖíB%Œôcå«¿ïØþXÊ»w“>õ¼LŒÅaR_y[~ØœÕ6<,¤¯ šŸ¯e­—ÿðƒ©««ªJÄ1¸âH>Â|§?’ ¥ý©ýŠÓ]‘—•iÿŠÏyü^l ‡=µ©ÑÛQÆ…éËÓ50(‹/˜µ’Îåß&§½o=° æ^|Œ ÈtÖÎütš›÷XX]‰´Óô駱s«Îj*Ÿ1úöŒ#q×ñK/å¤ä²4ËFb]½Ø×±“ætYƒ_ˆYpøî “2T§4eèWd°©:i±$†d¼ ÃD´„%2Ûàð=–)#d(Bº™§ÕS*Ø¥II­Ÿ~<–5Ò%bmhw`ÍG#ïûqapdFæÉ™ÉÌ’Œ*=Cï¤GpRi]ì|äʘ{°=×\R³CgØ©¿´PEèíB(|åú>V´.Г'“„iL…=<~·©sn–å\N.P¼3Þsýd ­zäÖ–¸×Jšm"¯#Ò˜–?¶¤p;7.X ÁìNxÀé@$k2U±ýÞîÈg‡I!j@˜O°úúñ„ˆ1¡S3ÇÎóôäaâ®ß¬§üûä †ÛæÚ&³®]I÷: Bÿ§€”ù0’ËAŽ°Í©84w"ÌB² Lñ€û]@ÇÒ_¡([Ö–MåÒ¼˜îhºª2SÂó¾ˆë›o€"a‡räg«Ô³¹ücíœ×Ž*ñ3¤¬«ñ¬ ”EÝ×òb¹ãC›ÁÙrìür›#2ÆÅÏPÔ_ŽÞöä$®'2˜ÞtgX·‘)Š1çp¨ Q©Ë2ìñ–`ž²¡•ˆd å­ßeXóTx w¼£$i^øe+@  ö,ˆÓKyÁ.¥³G‹rG_ãÂHèÅWæ£Ü;}´¹é ”õ¼ø6¹ƒWv¿Òìj½¶;CçLBA¦£† 0ñ ¤z’Ômc$9¹ù ‰^­Ÿ«Ú6e¯ÍžãQâ¾Ð°¨Ö%¥Ä]ÒüÀ7 ê±ÂxÞûêTqoVT–-phsÎü× Ï”Q11 ÚZÄ£û!KÚ3èÚdßLú¹ß¿ùŠ­¯=öA= «ÿêZ)»AUñõÆ50û8D›`ÿ²„èèkžøØ˜è‚ye@cí… sTº›‹Ó4ÅbgÁ•\W4tôÄQÅЯÊEïÓ*e" .ÙéF­Œ•£é7²÷ áñZ-õ/WbÁ©àxÀiú¦?©PÚ4>]_c5˜œ™Õ µØÚóÝé ³ÆÚ7·Õà¾c.£×¼]b…¯YhC£ÝÔ⨺« _ u½§.Y±Pżˆ>ГÜšèçÉ懨©ÖÈ7ËpÄVÂÐ(§Îk;¤=hÞô¢h:½[«ðà "ï4C¯áÆÄÁÕÄr¯^¢Î§e;f‘$£(¾Õñƒk) Ž!I-n8“åí¼&§whQqô·?g›ÙfÀWãüФ˜Y6ìªWí¾ øàè;³žÊ`«oû RZ6ó e² Ü·ýìá‰/påuòø^2âuëËçzViFk]¶‡>˜'²•yº0ˆùj9àM½^Ïa¨®[g²ë+Í{¹¬‰×LŸI}Ñm„^O>¡wõ]]Új­CR§¶ðß 9ݜۥ¡ zÅ{̽ƒ!!ŒôÂ2†¡×ðÕª?>#ØcIJ}·adlæ¤øÙ„”nRç¶rl&\–×þûÔF$k4øM>iµœôƒ=Éõ4¦cýžE®„›úß„=+†B³B®™\gZa !¯ZoæC»?€æqâÌæÛÀ%/‡b›¹-½…ƒ¹ç؇ÕôPÑõ¹΂H-°¶žûQ–K˜œáJÄøpÐ!ž¿ºþxZ,¶n¨`z¯èÎê'œññð­Æs-ϲÁ:²&1‹ð5ÛâÕóÞý¦¤ñt©6ô=SòXê‰1›ü=W‰=?3‡t‹ðÏ«:'ÈÓˆz3Êëbw %ZDg~íŠ-ÎÑD §ÞFnBeR.]vtò°G *‹:œ×Á–]·E‘²=ÊÂzøÊ†ÖßÜkô³ç53¸¿±ÉV¾Yß;]F?‹x‹ÓRŽT÷Þ·è‰Á2?x£w1‡È{Í éôH”\Ïo dgNéC)´äzá>Òˆ³T~y+¬ïÀ,ý¢&Þ…à‚-F©c2²þõ±íŠ B»Âó¶³<,‚mrdŠ”þ¦ž~ð²zBO&f_,O´AÕÄI‘ŒpèÚ9;°§¾s…°2¾Ž?éÀú»’}|ni®ï4UGH}ÆC]Á%C4Ï[ x½¦îç7ÒÕ–‚‡èî¥5b”ŒCYS^LבÂ4•vlìåÙd8À5~#>Ru//¬FH&—oh5MW²ºD«yïíÙï¾.‚FWL 'k~:>{ š—‚™ÑÜ-šÎ‘!;0.«l<&xJã^‘Ò}ï§Ít ¹šnF1GÑV ‘5/Ñ¤Û ðQq|ñ0dÈ)jJÏNðp½‰ÆÓ}S<‰¥ †,Èæ—°º>I‰ wŸÏÿ.£;°aPTÒ24yÛ Ã#À}%n¾‹ib®²2gÜ*¥bɺ‡‡<¹’„wéKÕ%0ŽÊtrGÜ3Í ¡¿A w[ÊLPE’<™÷ª5áô ÐÕŒÍ?C<Og[½º„¥÷ʰ\ˆÈãu›ûbš=Ó˜(Ïý*9Àh\îe.|“ÁÔ›e=9íhøkÿ6±Dq2Jù>…©œÓäÛÂj?½Ÿoë5{ê¢õÞùÍÌ“YŽöI¶ŸE*H¾Â ŽÆê?ÔGò„~»ôß’ntÍAü毻BGºø¶i™BÄ\ÀŸ-•7Hû'~ËÉpUç…`b9vìòZ?:ôT!_ãˆÖvàµr”—±ÐŽ` …$Ëó¶š^–Ë,­!\ñ+ š)¤¬³únTíÖ>çÜL+v {%åB‹U~>ã³7<Ð{ç¿`~¤`iÑpš¶mÿ_Îéâ±Bêx™¼zõÉÎÔÛØ¡›†­ÓÇèçªîlÆãûûŽcD5Úrðvçlª= nÉËìv…àžeÊ } ‚ÿ|qõö‡ hLþ2—ƒ: î°Í‹ÏÀÌà4I-²Xµ×…0„ Ïm=ìaáh2?yÕ­Æz>ìŠG­6èÒ ÷˜tÓ,AwøÛó’‚+qþ¦ã?­•éÒn~ºRøõÛУRŒªlw‰¶Œ,Z{<bs$bæÂkÞ̉Ԕ÷Úü¨Ó«¯”paÓ—H{{_·Èôõ É™!jæÝËFôÈÖ9Ükx5S¹m˜þ“(¿·=4Íži‰—9)þcQ/¸®à™ÜƒÁ‚ÞT“„Ì-[)t¾[ÿQË›)G‡3¡"S¡€uM —2¿®Õî˶ɳã)TMGäø¢7·Ü±ÑK&o• W‰BÎ(ðGt%q´èT[͆%ùƒÍæ[ã\Ú3….¯0*ûØ\2 8A(=ª2ņHwÜÌëµ· Ü#ß;ôæy¼‘¢L~mµDf“-,‘ê0®óþhNÐAï)º…¦6ÞçÖ¢MÓö³L‹{ìl¿g<¿7éûeµï"°ã}?¸Ë½ îÇ7eÇ„s+aµ_úô–ØßŸ2dzL>žç ~@ˆÐ®‹@6˜jËÌërÃí„:6$xŸã¹gq)k„…êm9Tµ?(ø”±ëÝë뎴w¦Qõ6u×6Ô)æ3£-5¢b¥µ€V±çUä§!ÔHþÆ}ÎB{:ÙIò¼'È3~€AmY—Arîéê÷A§f×á–>=çpꌹŸJC›EòF¦Á4Ë­íkA™vİы…j°­ž“Wf‡f£†;1q.µâÙ“©¼Ýh=VKÊ& ©[!Ìn_Y>0ëvűI~ÍM›¿ü¢×9á ´û–þYíäÊþ® !a‚ƒÿ˜ª/ÿ*UÒ6.¢¹¨\ˆƒø‡—hˆ2/³óO¹Þþë§Ý¯~OV\R®LóŠÓ'!·^Ç>Ñ -~ý v\˜Œ<@¸IÔ1ÑML-½¬Âر¤IP ƒwuf¹Œ€O>µ(t OSºö½©u •ï&Y>g—”Ýõ› ®ª&J8É2v…lÊ+ëXáé¾@Ø8Åíñ]÷ƒ¯8–‰º„ì-Í>¦F™ÄKþ†Cc±Cm÷›Á[pó¨^¤²‡-‚2Ï ÑbgAx–; ê—´¬¹A~K*^ sƒ¹Ö0þ×ùÜÁWŸMàÆ‰º¨ôìÇ”ææHÞ“…"ŽX}ªBïL±B|ÎKHRôæŠ žn•¹&‹WÍn@¿{ÔÙÇýy§7“‰ ùŽ'¶ãã¡2@T§'AÒe†ýU©÷çÄgŠ %–#w’ÆB–+B•îK#”“e²Fgéø· üC9 ¯§éøŽêkþ]8îºUMЭüT[gYuº~ÖlvíÈÚû9â<È´òäEhÅ:ÕòJR—’3£×„ªÜ'‚ âgXûò¯ïÞª|è\e¬æ-£j1ßИƒöoc5mm0]¦_¨Q… ÒSš¤#¶uóþIû.LK¾Q cBüºòîæ»he(`g×Ìí9Kë”\ý‚óÉË‹æ&›A,–¸~š{gt ¤ZæÇHŠ uý¡A‚HjsÈìRïtSƒöWnùŸ¹ÕõŸ ùg¿rQK1zcN×é{-`Àœ±¨-yö] ºìˆ§›BŽ,æß(® Q‰Å…q"<«¤?ºQUœ›¡E c8Ìpo˜X’4U®`„™¿uèWÞêd‡@×$˜')Á»M\þQi"b¯äq uééô18À;¯Gv¶V-¢€ÛÏ´dÖ¦µ°Kˆå,vî\åsNüœvIÃBâ³tºÎJ &A9Þ%¦ññ¶ßóÒGèaçvK¦É¾4¡}Ç—ŒŠŽÉŠœi×.ZÂà°—µÁãŸÔ×ââÒá>?§)Õj¯¬ñªu~àÆ{°´,`LsK¶Q {"ß²"¬ÂÆrØY]’IÜ=ד^cv ú£uuGϨX£z™ò5) Ê ºìþ€ #"*q;X]žÈ±¼[Ó³€· „Ë1…GÈDJä½C˜¿7%œÝSgØúiägDº +.õ¨¿ ÎüªFÊ®bäh…P‰E`fÏÊ#VÍÄÔˆØó„¤"G'V%Á‰æŽÆ Ý7ñAT1¹øyÝ/Âá[Y¤¨`ºÌd¼JM{[>{æz›Û?FŽ«|óúI–ú½L¾Ú»MáƒYµR”"§Æw|h¡~ר†ÏZ5Ô#mmu_™SªNÑ*úöE4\o&ßghµ¼\“.jÔ‘»¡P4ø FÅ·.ƒ!J7Üx´ ÊäÚ²Ñ~ÏÈ&_ñœÈŒ4g!ÜÝ>Dî¬ ÛŸ;Eó5éq‘À‡#|RÏø>’\‡¡ãòyn}Òp“¹ç½ŠD¸w¿[Pòd!2—x]êØCOfÖbã»Û~YVÒ æeñjèÒ#7Þƒ­¢ *IJRýîExÆlt©‰«³­ÀkӦϕ«líg:'Χ>ê|C­ä+ïcó^з¾±¨§½bÌiBDn»Šíqÿàp_ý²=Û÷H@^‡¯å÷Îìi¶Eµõ”ï“Ï­Øž?ÖÛX½š÷O‚óp5'§³QßÝ-F¶Yä±SÏ嘉|Wµäd2œEê’(Ô®ŠÇ^WBQ~xœßÒ xïSÃ!À±"£ÐȸŽV²‡íº˜8YuSÁ\¤ñNj’w1šKoÉ¿Ä4Ø(Ö ÂÏŒÅI&º)DUC^†|@€|$ÿZfíÇr˜r‡>=ÓmúW¢º <ÉjçÇ´y‚úUJ.³:®Lì.˃ƊLÆYL‘ÑÞ%‰ò“Z*sÖÓëbÝRÖ“w`„•ç|fŒÀWmE[Áä8Rk»³yò`L•ë÷dƇ ±yï;¥L8Ól°Y)7ªGm¤GyâºìaI¾K§ö\]i~´U wå+Ôï–`õ’l_ù·ºYñ/PW Ý«w#/˜ÈHYŽtðÙ$Ûo·é¦ÅìÄÛ¤ '¥>ͯϢͥðy7K¾vìMcÞËAæƒ-â¨Ä=¼Š›`×x=< öŒIe×JÀürða:…X¹M€1m¡wÚ¢+*=Žœp™ló,Í„¿M_oZô®ÉÅC9 S–_¦±q•àH©'&dÿ€q[’pŸ#CGšÖe¨™•}‡<áP€šÍZ¡>ÆuE%ä“M gÛäp+Ì'[%_ >¨™÷2XCBnp²E&êžájH×çE•59w#<»®blšñµJ\ „ïCˆ &³égN ¸qªb9HÝ85ôZ T8F½S«8‹4NÐkIìhs4r>¿’|ôªF\3„Ì•ä(õ®Ú“uU\·[zÓÎ¥±oµËfc5åF”|ÃI‹…M=À°Ã«4l¤w°WM²è8m.D# "aäYÐû!]ˆš\â{ å؇øsª÷ŸD V5PxFS”tÀhßRà©2N È]„iÕnr–=·(©î‹¿ã…•ÃuÔ?×E+¿‡Z‹­¼¨4Ê<î*_È~½}g± ¬‡&þö¸ QÔÞªkPžrµ¨€¬(ÜÁ¢ÌÃÎN±tÃHy ñ£x¥ÑìS¤… 9î¡b'ÁÔÒÛÀóÏ[µö8#|íæBôŠ&}Oµº¶Žšƒˆp?`,?$¥¬[•§½/’|øyÜgÀø¼|&f+®éS8»Â+Y–)ÄÀ r¸ñyÅ­—®+;D‚“f³!7¨É‡øÙðTC þ2ȧb_¥úÄšÕ™»O_Nl Ni¬5ëþÃX:W]/o—Ï;7m¬U6ýeøÇôxM€º'Ág? u}fMàNmÝ2U§Ý–i…Ûc7‹J¡RöBˆ±ÐýݬҚ»­qzøl²É+‘¥ßó=_o°Êøø(`ÛœzÇÚ~ú|lÕºÊãCðd![Åp+ïn| X}¸ê‹ôgsû ã»tUšÁ¹Dx•‚ä¯}}¥nC–i$•]³Ë:ýc±?ï«‘ÕV¦9 /@?øÞê¶a1е ó6]ú¦ù$iÓ§dÞÀZIâݡ芹¤ÉÏ`†Ûôžª v¨dHŸƒ5×ÄÛÞQãÁxU /c¾Eƒ®èÖÞÉÈ¢1XÃ*Ô©£Ã¯²ìÌ&"o•Êê»Lƒ°, Rü®S&á0B'Í•7êP¬h_ü2°Y~ûupI~gÀ Uç+Ð ¯ÎEâý¶ˆª–”·Dáå|¸bÑ#€Bº;€xOnª¦[8G=ûÒUáf! Ö†Æì6_ùhCw"ròÄZ|¶Œq…mðj{€Êz8õº{s/½›ð!q+ý YJÀÞö‘z¬øð1T åJgiõ7 žúN79âûDx,V8}Ë☔A¢5ü|òdÖ,}w±0ÏÊeh¼îÒbV¸õ×2¥Í£ù/©[ÝÛÓ³„†•45›#ôÂÁnÞŸö+]q¿w2è…Ó5šËÉ™ùJph¢GËòOõ'ÏZ&†5S‰~W4Ÿò&n´†DQæZž¼³4V16¼qìp“ZûîzD]ê‡zN÷µK‚K(ÇÔ«kvòûè ]!»ý—Dýfô3ã!^¤O_Ûå¼(¨ÓôB†‰ñ[nÅ6sËsèù¬¸c6íº–l"¼òì®—Ǻ§k»°ìç8æ­ùùoMð«,×P?®+ê—¨Yô¤V3²/R:-hŽê·ysTŽð»52âIwíöÁŽÒæX© Ly$+E ìífšy©B“ÞTa8~h‰>„±5ö uÙå4wÝâËSÈÚ¬›ÜŒéf˜v/T€£ƒç5¦äG¦·"be‰ÆVTù]ö'&®‰—Àu­\ez) i±¨“¤Ò¾Sãñq!šTLeòärÎÔ}ÜzX^¿&%Ö2'^¼,Ä) ú6åWˆK9™¶ éqÄRMœúö¬tÖóÞ&à:ÓðUäç”oOÏa^Wkç©n#ûXÓYÖ!›QŸd„. iž¨O§%y:(Ó"8VP1]ìq1¹M%sKÕlÁTÒ\îã´KÞ|)\Œ1s+T}k-Ëç™÷µ0ÇÛ—|»õ]ìC:™O(g_3iŽn_ü(ä>yõ!>Mþ,Ú"å+)Ó©|dEŸ-ºòFù"Ø+¶h´ÄOœg ¯Æ"÷¾\’5«z0§˜„ðøðZ¡¨Ã´NÄFµŠw®…Y6àžíÂb‰ß˜hXÂ}[éò)N’—^½=¾?¤[A¸R\ÎyEúÉtÓ#û2×eÖ±éúú­à!,Uõ¼k:@h¥«“ DTŽA1þòß fs_ß X 4< ®lëÃÑøTÄú.ÿƒSïÑWë B=4#T{ÜŒ8'XŒ Ìš¨ãHq4˜.`]*Že£jÒâ"Ƥœ?}ÇRåàVcQaÑmæ¾åMI*“%–U€)ÑAÄ1áœḵ{ÛÌ/C-ÆÝÅR;K=ù‰%¤AÅë/:ä6Qm(ÎàëLv­]ÔùÐÈ£¯ºû#: ¬ê,;¶.mbÞQ ¸BœÈazH8!’«`|ùØÃux^†W%ÐNäxi'_D6/Ï­ 5Ö¨‰íc—*Òþ>Ù»\@³þy»̲)fÒ}ɇz^N& Ü3=KòE)2Kاú„u·ŽAÏÙ+1ˆ7gšlŠÇ‚ ö×]µïü{fõ¾ð+2‘|®ŒÍ9Dó/2x‹7„â…¨g‘šI3¸+Œ¼¹Ø=ͺŒë/ß§ø¶%›Mi–Wu‚éð54 pü´àSð2ñ'W¬Iï¿^4¹„ÍÔŸZÒ&qâ‘¥òIºˆí‰a*8eí} Ò·Údá—½W©7XŠ¥¯ÏSªbAŽ`6+_\¨S2)1µù›\S€,Â^Î5øƒ€ªñÕÆ!qƒÍ€F¡“h¥3Å»œ˜\¬\BËk‹Å“ºå|0+‰É aAƒ›Òxµ¯¿‚ÒDH¢ËçiU‰±ÑNTô@Æ{› ñ aþÝ‚Oš4 óy}¥äŒË º¨Ï.51,V¹Çv ŽÜJÏù#Ö8g›Sš܇#:q¡'ì õÓü>=qîîä*Öñ‹`:ßÊ®ŸÄ°ìâÜãe`fšìÄC*¢4î/ã2¬›[®„ÈY5º¤i°êUU4Éo3Ø ¦È¡@åÈ)%å]tªhïâÄÂ}U1êQ)ýaˆ 9<~…û½ï¨xrÌ,É”:4n”m²ÆØ‘}Ÿ-¿;~,m·|Z¹æ¹Îï±;ÑŠ…ʈ=$Û>WMch ï¤½®}Àüâsë§CÞgÀ§¶è‚O…Ÿ÷#X¶9“EÐ$æO»c U`=oyù9Û´œèôl®õ;©$wÑ«¡Í8wW_ž?F©SàêÕàwà}ݰ!s@JîÊ ÷€F"Ìoò¯¢¬Úüˆ1þ^[³åcc²Ü×Ù¨+)‘—ð“ùAËcó|?ƒš§žŒXí=iöGêZs;‘Þ'·‘ÍàŠKØåý÷‡›® ‰"–ûYQGBø:HÄ`èÜå÷eMwBY+0™úlÑ‚™¦e.pƒÕ½ØRboê‰n¿b‰KïÔ¥?¼™„½eåî¡‚µk pŠjQZl(óZg^US‰Ï7­c]–DšçC }iàÓ—Ã^1ŠÓÝ ÷°Rÿè²gÒ± ø€Ô›öä:#e~Ga"ÓƒM,Å’×÷"zIoÒ§ÛÒØ‰Ý¦¤‰(TŒ#ÄJ6'ÿ<õí‰ÚÁåç}FaÉ•L‚ÔÙþÀÐübSeZlW1>-TQ*½`¤â!¹l‘¨Ãþ Žkí€N$OÍ‘èy>F[+IWƒè’šsfŽæF¦$o5ÛT¿Ó"¬À¡É|J1‘ÞÊBlÞK+-yrf¼édÎQA£>t“õÏåQaÄaß?àÇû‡Û4žáfé]ø)”­Ó0Í5û~¿™jï#k͹£Qjb è-å¾Ê4h9´ÈëÅ6Å—1|ƒGkz…¼$9Š}ëÛ×FÿT ¡µ=ð%7MÒ© D)ó`m¸"ÿaîF(%³G±zú±wÆ€nu%4ÆÜføzÍ’§ßéÊ{ž“L8}cÃ+ɘÏLêH?Ô 03]:2/¯ÂqŽìU&•yuL¥w «1º/“Ù«‰dOŠÞò@túöU¼µo¦ÿJIsýº³3'Ó ɲéaLÐ7tT7¢O啽TlÚr ÿC„ßB}J´x~ïñàeß½í8iØuNÑmßOIžá%p[zUcÁƒ õc?7A* z=iew±ó;Æç^_ä‰v/+ɠƀM ÝûÖìôÖ›ƒYFT]L°¯7:T^Š õÜ3¢´ xœYµ]P$•Oè]Kî´!›Š©3Â|PeQªYó=ñGÓ[Œ…¹Ñ&ýv!ÂT§(òH2Èå@Ÿkß&)äà€Ì\J¨vT¾—¡~q’ ×9 è43Å7ƒºbà­yîéo‡K>îj…öV6Ä»%¶ÑêØúq]SµŸì Ÿ{8âðç:[¿jœnÛú>t˜4!Öå›RøÌ³^ô°xȨÜu®¯EÜc¾°†ûõ÷­œˆF¯²%ˆV.Ù}“†75ay-ŒG‹P´S£-¡Âþ`fÑþcχìM®p&sÚø†¬mñ/ "Þé° äïä}v¥iÞ”s£Úñí>>ël Ãž³" ûšJA8 wmìÖ‘=gò³ÐcÄRÖk˜2É?ŸoDI@Üg™ÂM´˜LSÌ90˜LWò朮îe¤—º/H8^h ;Xû|"ßIµ5?Áe˜KyÈμv&¡ihYQ'ÏÛÏ`3úâ–ÊâE«E!õüœÇ›Ú7Æ·œ]Ú6ª\¢ÊÞÑ̈Q×ί%ŸŸ£~$„q®VØA1ƒp‹DK„<>¦;ö¾9ÍooïJŒ„;Q.$ÈDwQDÛ·Rxn¬7Ù0¦ÿb.R%]A—Ä“s¨LMÉÜÒ@LJ¸àâbKjï¡K¼µ·$i¾CígÀ:'aro«'”††‰ñ' È¶¶ù̲¬^ ¼´#ËR®úòQ¯ølG?0.¥‚ž#RÁÞiegJúé•b‡µY¼ùªN.=°Ùz9E›†Ül:Z=;ÿóBA³­¬€Ÿ}Û‡VµåÁcŒÃ‰˜k3ù”AÏqÌ4 ·¥ÌÆj{Áxö(¿Ò&Òu!i3¦N²XÜp°W>gB¬VÞvð)AOßÍj4«›®ÕñäµT¿]è1–bÈJC«lÚ0qÊëb8Å()Ø9yh=÷žüh¢â‡µnìQ2ós`YÍ·/{,!ã3g ›TËL v:_/l¤-Z˜Õ­wû´ –ë¢uw{,*Ó·uºÒ$þLgý¶@^ [³Kh£€k,#ÑZ\ˆrÊØ{ß,] Mu¾X“ýÏ“Îñ“¯wݘrø¸“9¾xbÏŸ®%NüÖÐàÛÞÞlwzìžßÑ„cÊDŽÕ£Í3¿?r9~—\û ).V–l¨.n…°$o„| ½(%a_m<ßZŒL¦Á™KÃÜúÖ¬éi9 ZlÃ1ÐvLÿ‘ìÆ€‘íþ…|gíÖÛF¿é2%²;H©X©í[ˆŠH7Êì¼Ð­ üOi÷óέDª~¸&•·‘fÅ?ðùàuùÉG…$62õ#›Ÿ;Çöë+gí ­jÏéI‡S¶©F¶”0M|¸âê© ê6¦ƒÄr¸Mì±¢}çÆ+GêµÅ25ÙÉ™øëTÜZ±¢BM÷LÑÀx©–x|bò‹!8œœ2tøE~l g”LÌHwL;^½Ýæõ·ŸãEªZ›•F—‰x ¯•ß_µð®:ûï©®Œ`½ÚÊ]ȽlA~Et±… FýíÇF“ƒöE±Ñµ+ùB«å{xe[Ó` Qz!Ókã0!;yçZ_|.£ØÌcaÑ„Œ÷¡[ßÊ©0«-Ê! äo©ëvLódØ´0”®ô·ûømžYìLÊü#LRN“ H6×fRI$“ôª õˆ‘i™>†ú,æà}‚”^`ëY­ˆàæ9k€ cp„>ËõŒÜ¯+Ÿ/<Ñ"§æ<#á zóFk‡qÊöÀ÷#gÜ®[7@d:µ8è$¾K±¼/‰¶}ê>}š«Ÿ0 vPÁžõ8­!²à» œ+A•O ¶¶' ë Ò‰?˜ë\f¿zÛIJJ Œ9AÛ;|t$‹,ájNUú AÏÜÍþþòÂ`œ ©éåH¹ÞUn>£PÑ r]<2Ì™=լѕ~h'tŠ{vÚ`1Rˆ>\81Lïmàtëh™aÅ*>ñv ÔS'¤ª©qótÕ»Ëyi¤yî&Rœ¶æŽ”è“ôXÖTâØ1Œ•BAOäcëp)6ÓQ¶Üº~Hùa/t˜Å7ý]ÇCú-Í“]ý ˜jø½žº ¤¬‹Ø›^dÚðpMíèöíÁÇ`×®ÄtÅjšÐð¢\³NòÔŽj§‰”oiœüsæS^$º…7ilå0»Qø—k¨ÈB'mÕŠ Íä*5…TŒÙ?·Ÿª1h ÌÈZÑ ø‘Ú ýKebN¬÷’Ý(¼f áiÑx”ÌXã[2‚kðÐbˆ:Á>ƒÁ¡>ÀÛMW"<ÜnªGÚ0$W^w­:Zc–K6‚z£SŽ ‰ã= ŸW'ž=Cv\ûç•*oŸîJܰ’0æçñRI·—PýÓ´UŽôcVÑ-É3³5mZGñžßïªjTò©nd8¨ ªWÁ»®S†ºé/Þ±ÜA3º³¡§ ²¤Ú æ5T y>}:Iãš]hÜ))¶ç!†FÑáp« ¾f­¤ðÛ"Ý]ç‰~…¨õ•5VžÎ¿'8Ï\]k¶¸׺LO¬Ç5ù<¡ƒEaÛD»é)¤`Ó€ž½HGh0pÁ¸ÍÓuŽ…0E‡$\?Œú§øÀ)Øê¼Åp¬HJQdDM>‹†“¹6‰“ˆ¤¤®é~ôSC“:©ç|Ñ£k^ÒêHVÇ_3©Oú-–Ý‘·ÂÓ˜Ò{`ÿ¢[{;Ãbó%Ô¹FÕŽJ bQýã§¾„×…ÙV¦´ÛzÉ­Â8éðV&¨W′ØKC·µÏ"ÙÕ†Ÿ^…­+Âi)BZ©jjÂs‡Ó¾2`™ø 8eÙyX ÁB\š«ŠFsA~nܵÌÓ#p-‡%U#†ýÙ˜ûñŒþìe¥ú–r"ƲF˜kæ‡bq†ê!lÇÖóÁçúÊŸòžÝ/µ˜¥ÛˆÔô¢øîaÓ^ ED~ÇõFÕ-nãy²Ñ¥IcP^]QlÖ žµ$Ǹè\3܈T(žyâ†}5˜14X‡Ìõk/ÏêßqäÛ›'-¤Û/ô>çôÙB¨ëH ð2a_ʱÜí¾.w|¨oMóÝŸ+"5,=£"Ï[›TÒg%óƒÿ¾ +É×ê(aRØí3†¹óš‘ÑÈ ¸—í[b×´¿H¯pX³p6ƒÍ„åßõÆž%u þÄ‹:/g¤ö*·„<ñÛÏÙÒeÝïIÊå/¢üI_Wf™¯æNb[º>‹§-Ö2"N﩯‰W©s•M¾]ì÷y¾ŸN èHw½©„µ©u“Ïtu5+¡3Ç’½}(œRÇé^zþ&mqköím|¼ÙV§Á•…±I©MWkŽ—äÁé»6ö.t_±5½ KB~‡$³{½)bðÍ…ÚðW’ªŠôrlU eër8.uŠCê‘¿í€ Óï;mð4Äô»û¦7Ƕs`.jmíšPèE°Ái4›0E¯/œMnˆ]ÛÆr9þ‚­‡Np¯ÙÉoÙºyhSÖ;Kâú¾÷÷#RsVhž§’*ÐãPéç0¾mdÉÔA)É)E¤—Lâ @ž‡Ó‚±\§Ãf¦ 6£œ Xç1\ž^ÙŸx GNKô uV2õš²Œ øa\,7 ×ëïýÀHº7â»_ð¿J#¡€5£l%üȉv)›–ÖßÌO· Ñv9´¹Žùá?U…CF†&ž!©f¦ì10Ĥ)j€Ÿ×óÕj‡ùº–»G&«¨D•4M ߨškF(©t¾‚ŒBG88••—ÊßH¼žF‚ô»µÈû¶ðä0¥\’[—/óŽ2ßöæèà7ážßXH0H<å)”‚ êØMú,×ߔ‰¹íV¢ûCDs¸V†9%]Þó;ºB#Ï„¶Ó¹5h iGµ\Ór¨ðü&]Åa«D|ÈOÎÇfß³äÌã³ÁÕdf(ʸpß-üÔ&3pœ$€C µLÀ‘¤ê8÷h«±ôÄuŒ­ìò..³»§ WáÜý¼ëEá¬Ë(¸ÿÄß+L’•F¡‘Ÿò Ú-®|Š¯Ð„9îÏŒ2½˜Óu¯r¶)ÂЀfÓ+^…›t' 2Y|DäF›a2‡f+Ì’PØÌ û*òd›øV³~ãf²²•Ù÷i¤-È”ü ³O•÷“¸‹W+S©Dm¾1Là7/;ù,¥~ˆp•G9ÆNç«e)b—bÈ4³oTýŠêPÆnO´3uCÿOG¸á––Õ†J޶6´j8":µ¢ÆÌòój>Ý—|®¡/5÷ Ü9Ô&þ’Üܶ:{ç<ì_0¼ës„nkaSúŒ?zoÖ錔^Ød,WØZ¿ÇsA/Jê3jáäšæÚ´¥¿¬³ôøŠ›©9{WÍ|¬2¿ªÒiUµvsšÏ©mê^¬b,¢r,>OŸZ×°¿žÃ@q;c1QÿOêdÞ}˜’j‚|úŸ‰7³YHß¾¹ýq¾5 IŸ€¥´Æ°p6”×òZÐûÿÄBš\ô\Aæ©°Óz»¶Ük¾Ù*È?¨¤f#aÒÔ?ÔÕïË.Qs²Ž¶ÁcwB[ò¹ÿÆÍ¾‰­œŒÅY8ºëÒÌ|ŒHÍØ‰Ã6îâè™9ï6êo ‹ŸüÖ­r Áú¥† ?åÅx9$ÀhõàªõVS—YÓ;)±ØØŠ“8QöÍXsðGÏלϑÁŸ\FЛ|‰`†´$BC ˆšj³ Öt½Õ ¶Š¬F1Í9LÕý¤jèó¹ËŽîÉ¥4“›„Ö*µç ”Xü"E ýì[è-6¦¦ œŽMÏ‚mˆõuW|qÙ÷þoIÄC”š!• ž ðe´ÕlîzmŠÂâÒŠ Ãuÿ.ÍùÉ}±²EÀès÷ŸÙEZ'η¯×²Mü:p™¦ÀæÅ&µ/š=b“<ƒ“ΊšnÛLNðrY%x­ßA×_î¨8F‚Ø%ˆè訂öu; ¦~Ëá-¡~ñ`_§˜`ÁÂÅÆ²¾¥E³F¾§üq È©„./òf9Ï€ç»Ô„¢³PxñªÕ ÉÆ@6àŠJ9¨üéÔ>úÇhßP–2»H«F•?EON–;ÁD}b‡EdOBO½w² )z¡]˜¿M)‰f–¹%`»²p«ã·`Žðv²¶¡mÊÖAÅÜÞÁ;ãŒ~ ©>waƒ-ª“ÚÝ+REÝ+i#Çù±(¹±-SâÇò úÀ¹=¿Â6„0âü`ih“õ^·.æu5s5Yº>èI—ôîNBí¤6¢8YPJƒ§GLH ¦(€ÞÌOQ:7K¦TY,w‡’|Å:¿[ç(ÿJä/„pçc}ßš_Ÿg·æVжhY"´ë’5“>¯jyg.>s;F¦¡WiúŽrÄbãñ:Y´ê ÐŽ£Rs(‚ô™§Á0o sÒó4XŽQKLån9RV/%r˜Ö ’’/ O$—)³—P†QXhA£i ’Ž"^T-|­ñÓL¸#Y¸;e7XÚ„æä2z æÉÆéÛ-`¢´«æ†ÚCkÍF‚I#Ü>i--IÙwBÁõ]4×ÚÔÞ¶*m‰äÙè÷¯ ëìúaýT*CÚ¿§R÷÷Æ™€n«|„ð™°À›tº‹ÍõíD:áMMBÊûzðQí©2m¹£{jÂv|öôCKm­ë›øZÖsIíÛZž ôi&Q‰lÅÊÖÈ8fÞÊÁF?a-$N^¿ÚgÉ_Yef׉ÌxBœÔ…[_õê£ÐazYQøÚ(ªŠ|,“‡ÅÇU]²ZM¯Ñü+¢—Rji×ÑX>©±;Æ2zÐâ e@üF} ×YІ‰”4pÒШO„”r\¤g ·8ý^†½!yêÅŒ_rÁ¸“PË.aAÌ\ÏüW¶m kJ´ô]Õ¤†uËVÛŒáçSñ…“èvŸ•íá°óDKX¦á¬3Âá(X¹$}ø”!¦f ï!c‰µMpƒqÃÝŽëP¡¼JH2‘-¶j'ájjì($›<,—½"–dÅþ¿>Loˆ =ãðüë.šX%$€’û®F#˜àÁ¼ÿ÷gzþ`Þ&Ó#/Y«24%´“§×ÖžO'í:R¦˜Ô'¡ÛÏ`f‘x”pß—Å—1 À­Ôp§¼Ó¯Kœ+Ë”óÇs'2lTª ¥ÿÀ`"‘Ïïâ¤#‘¥!D”Ä=ñÛJ—ÀÌÄ×€ŠJ ÅwÚ¼+‰4cºŠªh¤ê)ËÝg&o­ÑÄ+ÞóÆÑÌö?«F0‹æGÀâ]Y”bo/WžZüRHAE„'ì—2N¦ÿŠ\†NÈ©9¶Q!MŽÅݧ”“B™ é 8œ[Èd0özͧ”¥—ìê0Ó¿S4ð{ íó>úÖÜÏ[NJ0Ÿ™rüaà÷¡H 'ÞŽ{Ôw‰>æLÖK@–cOÝÁu9þîb*‰ñ~œ­ïH\v>q^…ešª’{>Ô˜…^.¬¡ø‚Jö]‹(pÈ·Ëã†çº{À;ˆÆÙ‹¤*©`²¶Kd}Âémmä&1À¥X –]\#§”dh€Tô‰D À„å;È­Í~8 «ÏÃÍ:4¬»ö _<ÌÖ äçÿ2„S|Ù…õüt¾úPjX†£œÈØV~wC™ÿ¤'±DÂ\cî˜X¡Ý­¾ÞÁì-5Mæ1:_#éû"ÒÒ¸w€a—oüXk4±.2ÖÓÙg^#ëî4Þ•LbÜNjv!>ò—ŸW–¾Í’tCŸÈmk‡¢¸ &B¢ðþØ”ËDà6µÀÖ¾W¨d`´çpÐ ]ì¬{Ùä×ß:¹¡;¡ /jixA!vEF}ú„ùº(ñ´#-^¥¹ ‘ Ú¢3Û8äs^Æ“ˆ*P1ÆÞѲ¿¶=רJé„Y+½b<ú冋á¹Pž±øEU4À küŽÖqYI—‘9·ØƒqŒkgoÒÜÒý½ß¿’Å–¡žð[ðnÂc]TùKmÎ÷1»eàC´\dö—œmçL8öq¿Æ Í3¿é&þÙ7õgB›™ê9d#Λ@×NNBèØÞ}öZcÏ*’~_Þ-$¨ºj\T_V p~^JçÓ²e{ÏIèÁ0’öÃPN¨ë_oZ.¿‡Ýøð{ðZöþæL,•/¿ÜbÉÍ^)çà^Â…l4Ûÿ­ÿKõ”C¼–Œ])ÇúAYõ› ’7}t1­`yгƒÏÇKýBKÓ†ÍóB7Y§ÛrgZÉjBt»jp³•è^Íd½À‘b·”š‘Ãi«mÔZ_ôƒhÊWª ±˜%ü*j‚ Í/{¿r¾ƒ{¸üô:ïfüû­Ç(ùCÈUýᄚU^ŠY¨¤õ¥lTÕ¥Õþ &2©„±a B |£È|›Ôtâ(¬_¿³8ýÔ!`ã°÷ßÎùn ÷.½Ç£1l”Ç™ï?Å®[±‚ó<¦Ň¥¿lþ¦T¦Ur2RVùÆ;§ºvr ÓŒ‹‚5„ùª`YÐù³Å"C–²)D‰1û\*jg }÷²óE7lIäQÛHº üŽäÆâ‘Dú{…p¯⦩Óéïkñ‚)ÉCMAŽ1Ä2‹äQÚ =‚’ÁÎ(~””î܋bwĪäí7Æw ”VG×Ûø–ÅÔø ÈHj”ÅKþäÉ2Ø€ú" ªRdEš ª}0–KnÑðåF¥•ÙÄ(0ì‹ Â †ÚÁaèV‘œ€Ê[M.m±6Ù—'˜{³¢ùÉýS ÏÖ솂Œu±wh vÄ™šÞfx£(8§IœÔ:>Øs–®A JìqÄkÕ³‘< s°³ Í4ųǷýÙÛm;-—ˆü[ÉB­dðôíN•è;SÁÀÀÌE“¼güè30H…àå¾ È7éÅSÙ˜«Ô÷J -ü‡Um÷éÒ„Töñ‘¦ä(T!1´´É³³cþ¹¥j§Ê$[cs“­KŒ¿ÔЗî3¶Ó##n³Û.€B¨”ú—ç0^tè1k¿ ¾Ló9îi‘`Ídka¥ðÊ-a×ã•õágZŠÚú™(jIÔzn±(Ô³^ZüÌ–¤É™¿ôÜÃÔc¸,p*í „‹ÂÔrý"Œ`{VWÎç—ê¡Nü¹ö ÓWÂÞzôÇkºKû&®ÉNô{7<9\‚—d ÑÍÙb¯Ë¯Œõ™ÆQâzô¡²y ú4M…™â\²&D½¿ ºp ™] x X†›¼ºi«¨Ü«Ït*ŒÆÖ^Heáý_fûðl‡Óô…$Âë8W3 }99ÜÄ*猲"°ÍŠ)(-^–!ŠÎG‹ò‡o4tP¾%=r ÿ>¾¯ƒ#ÜA/\:ÔÏÃõ¾(< Ÿn¢®¦úÔy~^ñßП‡¢¢\ï½7[¹ CÕ‰WrÌqñ€8=Ñ7 jWx–Ö‰ÃÏýòàtôG¶³€ÑZ ³¶Òs?gvF¢}›¡×/á†xTÎb{¶HP–b*4èƘr¨ ©%0W#Ìn)x÷øÇ>ÂÉãuá´7Ô7#0“ªúTÿ¥­ýM\*´Š™òÁʺž½Ç˜nÑt0£³]nh–—Ž“‰/ÎíéòõŽz×Ów–°ÚájÞ¯,±àMŽ~Ò.É<”/!Z›T™îâ2 6´ÕJ;vÊÜîškÁ¬–{7òîíwwÛÕ3oèD`§Ïagçò!%9?ÃÅȃÉõï1iËüþ²ËÚ}zŠùž¶}>Þ¬,ÖtҴ䇆õA¤hiÐÅÖÀÍP“5T R±ÞŸ´JNF‚›•“½6ª@ ’pV<]l+ºÁE‰h´[kàâ/·Ñ)yœÌÙìŠtoâfö½­ÒèüýÍÉØËäUõlÈy}*$L‰:,ÃÙ&® Æm›¶ -hS÷^-A@ ]a©? ¢î’<Ã=…5œã¸>bFAHáäã<=¬ÎÖkŠºîöaètcsƒ1¹>Èù.ç×*}°µG5Ò ¶W½äu5ýéµÑqƶ`'¶fdÕƒÀ‚K^ñ!w„HÅb% í( LˆT$(•|( GOwøg8ø>XªsŸ/Š3Ö³yÿþ)¿tê¦;£¢d¡)!„K„(ØÐÑüԾ˜?£CeÄ0œ?VÃÆFØb[©ï§'Š„D‹LZaßäàF^dÑÀÓÀ8™”=yER±¡b}^êí/|@I‘4–¢xûuö Õ ÛŸ/ªaLÅÒ2÷m²ÞÞéõܼ™ÐgÜ6 x5pÞ°¨öàCà3ø+9«´@£DÉÿõÚ?œô› ØÊj§Dxž[Õ*Ϲx¼zjúÅ)ÛׯÏ+ ¢¦¬îÚƒP„&Pu±·ýHI] ^ñZ·KÑ秘å¨(‚Ýéy*>2¸¨ÁuŠâ^I#(x‚ÊíÉ”õ(‹_²´ÙÜ;ŸÚØgh $|ç@øÜGŒNsÍØÌêžÃŽ2¹]y#kËy?p\÷sgÀÉ!œÔÞ€XÞtT#À÷ÝŽì‘•› Ÿ‘¤î-«Èý“µâ„‡$8¡¸´µ[ëá´¤Ã&fÖãÒ° .Ø,ÄØóç£k†±Þ$í/dßT *!kޱb§×ßÍŠˆ]âÉ;^)s—é$=47R ¯ê¼vnpxpØîß oʪëwÛš‰>”H¤ònu TsÍâᫎ™J¤kbrçnbK®vÇrLîŠÝ¤^ÄU‚y4–>ÊObž°õü]õ *Ó×&* ¥ÑhT¼çv+—Š\¸ÿ½ò¼[¨ÝÌñBƒO-”V¹dÏê–Ñ&Ša2aQ¬õmyŽ€ßŒ èLjc×>©—L[ãÆ Î+µJºÜÆK‡„ãB3Q õ5rŸßsdçÒâ,`@5.Öè;Y{ºØìæ”Ø•ïF†¤ÐÖlªÂf @ÕèÔ‹ÄôÌ0ß ÷Ü÷ø5[ ¢¸ýúÐD*©žË@±CÌ–ã|a÷Ñä;‘eABÞ¯ç–C\’¥‘ æF ýz”¨©ìYi!§¦Ãe_s^=’À,üJ–ö•UuÅ¢å7ŠSä%¨[’³ç†‡†ñÖ…ÂãDÕ¬‡²Ý;ñt–G¿‘ [VO<ˆØãÿNQFG21ScÀøÑ°‚mð)?në&ý ÚjVÅÀ/ò¾YÆ®`¯å«²iƒ…¯ñ~LÚ]} Ÿ¼¿~nš´^«rOåÜI¯öö»7-ÍÖcƈUÔž^®2ÙWÐŽ¿ÝÏ« “!/`ûž1ÊvæË ‡°ýää;ˆ¾Âèù±„³/sÜW!Pä€BŸTŒ˜ñ×BM+X3kù8LÊØý,¶˜ÚzûLÆ$Ú'毠ômK’”À/6áïíì?øM ¡„„¤£@å‡éÄ:†nöÏí=¥ƒ?¥‰™ªÙ«ªIg}r„G9è‚5ž—ÃVÒßô-±¼^ÍTÀÎQ90º‰`É稽ö•¼+4a#)|+Õ`¡ª¶¶ï…€ËB`J4à'A°hMm.¼EÔN¬{4ˆÿfËÇÄ^îÕ³Å\½·ðž0ëÀŽ!v <Ì+dÄè¹Æ3óÞ‹±¼=Ø~_={s<Œ´ÍW §ü¬9øJj7—Þñ•µÌtQñÔ‰ƒ\kºø+Ò|”íi’ Q¡Úßå  lÂ).-?0úïi•&« êùfÈ‹Ø8¼^±_¡$2OòbòK‹‰RiîlÀiGqJÆ_‡•¥€ÁsúS\8D '@Ai3ÃcÜ…°<p -Æý¶n‡Ãá=}ÒÚØáJq/w²?š´­å é"×x{Ñ|ÕTÁ °RIw5y÷ÕëY*šê®”§B2rs>,©þG_ãt”0*æ7ª8dä€?Xß[ѱ¹¢æ¤é—°Æ@KM†å ~4AÀž¡ ¦ß…5R¤< nð:Ú¬ 1Ý z kl°“š õ>ãCÜñå¦'nb<¾Í… ]¼èù 2Ò47ö·‘»/·GýŸ³d9k¥¹Ôn6°Ýð>÷tÌë3Ê BiÚÀÄ]‚fy’¹À¢Ñ/2¤‰OK±ÎY¾ì³¯l,¯Ãäò—7t¯1AþáÝ ‰§wòækX 2êÄ›G& Êß/&VE€†˜¶’Fw/×FàíBÙÂ×$åwºùëÛb?„þQŠ•]ð1ÒnÀDÍg¢Ö7·“gîŠI&ä³ ¿ÆÏë¸J¿és-Q¾FÀkÔ…`Ü£Â)ø­~!ù”Wþ¶¨Ÿ¬”wv‘O¨­S A&ÄFD1 ]ôd6^÷I‚“€vQëCIZ‘Š®4²äÂÖE…ä@ÄP0ßÙFÒ æÜ¢x(Œ~rŒ¢õC„/;¼-ºçÓ«Šzv0$³F}c€oAíGå¾ë§aßÂMÖ¦3ì]¦­í«Å^ü¸’òðõ#€¿€2ZuRzg7qv2@—¯Ã4U3ÚKê3¨VÓQõ-íÞ‰ âã­–ÆÊ¢îæu<ç[¼‡Ù<O‘€]7‡ª±±ÀóóÆï½Oi…Õ.\‚­°ðôô$ê$S"SÀûÏ¥\ëgrÄfô.Þ9XqîiϦ¡ãZ„^¥³‡#øæÈò§Ú9~7òl´>,!Ï ñáÿœ*ÔÎêû0Žã¥»èz¶VÀȆ"kz²™²†Ø o> • B’óÚ–3“òŽäááLÔ£ <¬¾çèßÒž¸,¼,ò¬%QYÁ“†íRJ¬Ãð ß8)BÚ•y…[>oyÊ»ÔçЦQÁNïòË,4Ì0…dgÐkuI?u3‘±Y¼}4ØD{m޹ׄ”}_)¾± ùãv‰¬ u‚S‹¨óË_XÆ‘ÖøÏÌOQ.ø­ý{ zReŠ·äòCVnÒ6–+2p•~ÕFGgDab'hqöÑÍ~‡\Ž“ˆ†}¦Ñ‘ëÔÚýšXÓU7ÛÁ<ß}ÿ‰ŸÛƒFϦzР‚³™Ì[Æ _޲÷ÄŸsÎv:'\†­úJŸÙÜòzP „ó†Ê¼NQ@…äÖ ÀG+AªÆòÀÙÿ+<#iMq}ñ³…béhñbVî¹H0÷üsâM~¦U²vAÛÙú¦\@ÿãD7ǃBÞGöÌN oÕ yŠägò0Žw)Õs6F).£†ö‡óðTŒÄt´EÝî;ÚkË<ì&ÀÃeiÞvºV é ­þ_§¬xje@>Š~¹ò—£@f > endobj 147 0 obj << /Length1 1615 /Length2 8505 /Length3 0 /Length 9563 /Filter /FlateDecode >> stream xÚ¶T”]6LHHwÝÝÝ"Ý-ÈÀ 0Ä CƒtJ#!‚ Ý%Ò t Hˆðá“ïûþÿZß·îµæ¾÷¾öÞç\û\û¬a¦×Ñç–Á¬ÁOaP7?Ÿ@QS_ÀÇ'ÈÃÇ'€ÅÌlA8ÿrc1ánTâ?á` âÁ§D<Äi 5w'¿ €_D‚_T‚ ÀÇ'þw .Pz@@M€ vÃbV„¹xÃ!vöˆ‡eþþ°Ù°øÅÅE¹þHÈ;ƒá   D؃V´:ôa60Âû¿J°IÙ#.¼¼žžž<@g7ÜN† à AØôÀn`¸øM tÿÉŒ‹``qûÓ¯³Exá`ÀƒÃ b†º=d¸CA`8àaq€¾ª@Û ý3XãÏ.À_½ðóðÿSî¯ìß… Ð?’660g ÔµØBœÀí§</ý:¹Áò@ˆÐú!àOåuÀ‚Ñs³C\n`øŸáÿÕ=w8üáVùCß­ýÛþã ƒ½À6X‹s0ÉP‡ÚÐö«jy*OîqéæãtvnßEx‡û .F {UVðüR>e¸eK™íBn‰î—ïaK=FDk²nÛO¿Û‰zS;mX “¤E‡òuý4˜ÔÜr»~¿\ýŒ‚Q[;Õ˜ó\ÝÅpu ˆ®<ûT¼êúË–GÃçvtw«DÔ±o˦¹c cž•Ì2ç[g"g@GpÓ<æ <ñ›½¸œ!̸§SKäÄò?Š,ö5[ˆ»þä³úÞ@À­‹‚‰ÂŒœõ‚ptŠÅWa?UlÞ·´x >šö‰´clÃù5¿Ó>›ÏŽ–Þüà»% ó„9/*±ê«˜µÖgIÎ%ôIèBÙ'+jk^ªßE[¬Ý¸´“ ÷ÚN¢uÂzØâ;Ý7¹u>ûl$µ&zö$)utÐ^áäÜ:ÜÈ»ÁÛX²ÏÏmVGV‰šÅq­”iZ:7í²j=ƒV è‡W¼†TgOÕ|¦#=¶8œˆOÎm2a»cn g Kä憆`éýêÖ”\%lNªW¤uÙ4î—â-Th”±Œ)Yy-‚ :؃/,jw±Ù3«öÛú=±®"û¿l—–ÝËWUä‰ea>È»nB‰Øs'U™ÜPúL±¡wÇ^pf\Ñs6öáìØ¢›OЧÌN_GËé5‘YŽv‚ÆñL~œ0•¥x}/ñD‰z+šß·§XÁ¡÷¾½Q€¥ˆÀ²9e®5ŠœŸ®í(d2|–i^‡©µºýw¬&½_¢Ú!8Ã’ÞfÃ:ˆæG[(¤2÷5 +.{´üAúFÕÈþñƒo»>—Ï@|ö}ó¾RÎèŸ<¿Žæ¦Þ„“„Ù ¼D N:Fº¡¦W ¶Œy9ñõl–]JPŠüI¨³E­iŽLƒÖ•ãÕE'3UØ–”'+ž[Й¥–’GacL~J_½>´ÝËìø[Q©ÏLæOžlýWÉšçtìáËRÃM1ß<~éÌØ2Jñ«2j¿Øq[-jëzúÃνºwë+©õºÃ2'=üê>9)ÿäç“\grv/Ö›­x_+Û9{¼(²·ø±G¿ìNi4Ë'ž·ia¼?Ã=¥­‘5Ê(&óšÎ4³¡r¼QÇé~ƒw&ºÒ#«›ÎBš)„ÒgÖ7ËžvWÄIå¬ÛÇJ$.²± *Š^ÛaUÊB™b 3á+<iθæü2ZµP8H%ž~";˜dWôn<–[;e¹ÝTf…!Nš!ÌsIá ‰ù †æU…#Ñ]¸»2ql꯷7a7¸#þ²jòqàdxCâC_fÁ—ý)ËaK·vå—ðérŠ­@ù\¥{ S­ ‡ê˜ùÒgüu)º¿PûR¦÷ó–ZNL|.{Îxzy„y&zÛ{â è^é_hÞDZäÂ$.¯Ošõr 4'bäko-%JfMå^Ï|W+F”2…P½—÷â-dÚ _±£ÓC%¨D].©×m„ôÕp=ðH–Þém¬æâû);"íýeâ0ƒ>Ñdò›FωävÙÔbÅÇky©FÖÄ¢–›Ž™9ï.Ô$ßÛ<ÁÒêzÍqöd1V…ºëmÑú¢_fÒl/m¤aѽÍXåZÁ>îp©ÓlÀF²È´}†š9¦ÀâZÃdn-Ùº‡áyƦS»Ž3»Ñ`3®óIë|y6 !¡4K;kï=é¹Z\¯[zè šÂÑÚ¤7hÞq‹”~¾zÁ±"}õ9#i iñ¢ª?(1‚È"‡ÒÒ(Õ‰£nK˜qîÓåD…°åî‘H¾±%Q·óH#ŒÑYåLÛ³º¼ŸÞâôJñ3ñ¯•}5J 2'H D#‘lÙß?“›ÀÁÒ"tê²ßôнÝ£ÒæÅ[Úãe:Êê¿Û¡St²´Dõ_jäv—kZ߯”«i»{‹‘×äí© O¶!_Ó!aÛΑrkè’L)|Æ›\Ø$”¦·ôŒ¼MøCÅL úÜŽaÉ<×Sj¿1Ž•˜PT¢÷Ñ@º—(\ò-}œIsz– 4¢ê~¦¿O?œ™Qd :&«7ÀÝêMå4¡ÑMîãÆV̲”xºb4çž*‚v4w’íØ bôæ§æe‹À/œdó ÷.‘ÍtŹWÌ”ý¦¬zç1ƒ”ÓH¹°9~Í2¨¿Æ]<'ÒI²„ö]ï¨Õ¹½™Ü™¸$’a¡æ¸úr5,ÜPœW£CLå¦ÀOÕÏH0åìøHUu_SœI}–DOûéá'õUcVeq[ëj#³1ï¢JVÁ\ÍFkrî‘2²+êÏ_/ŽìT._¥¾TˆL'RLð85Õ"¹Ÿ3ãf½Æ)ñü…Ék*Åá¤àŠ•#‡N#[å;‹œ^<a½nón6åŠ#y5zA7 40‡|Á—•ÂQ4³»²!e·z­©çz›É²SžåkœˆžÈ™ïY2º,-T©ƒ^•;‡£->ÐtÜÎÈpGÅÒj[WcîÅ”_!"®Q0 ŽtNžK;'³<8­=‚*~\VPB ±xºèÍBõô ½§Ô¥+Ý1î×ê¶§<š3¸¾äï6$3äù 7ÖI•²}D¦Zãœji[UDý",yLG™?¤°½ ¿0ºàý˜¨F…R†œ;Tól§`#]Dï³”DˆíÌæúÔôg«À"Å»Fg,É*®õð‹ˆ¼"üÌè-ûñ>x*çmk¦¿ 7Aœï³b…„¨x­+µîX)ß\¢ª/÷îtœ~?¨ *×¶ŸP“|“Q×.cYB(î‘;“àÍA…ÁªêŒhž©”i1§¾Ö˜¡Úêyåõ݂҇è;ô°‘_gDf4pœæ¡un‹ïæ‘éWö+ˆz)“¬™ªoŒNÃË,©BþöÂàÄÏÏ´"á}Ï*T‹¹´›0DX׋)6&œ êèÂ3±Bïõ´¿/1z‘0G´ ríp·±5Ñ«54™}ð6YÃì(¥·¨ò™_‘›çi |wM‚#ü5 i‚;>YI£^ø|ñÙÎA…«]pI/-|¯$ßn¬J‡9ù•à=èKgj—WÞ¬ü/›Ãþô‘Ö‰ü‚[SÐ*[(‡DÎ,dYP ¯y£cÎ -JŒ8¨B ªM¼sǧWõý£·p°ŠV§‰–‹|Tªº¡›…µ¤]"2m,·†Ä^Ì^’÷K•£.DÀ…ÔÇx[“ü^׃¯¿m[%¢u¥œ²(+£äÉ6Úà“ãÎ]ã(.§ ¾­å¶\dzTU¡é±†ñ…æ ©_]‹[_Ì8N¥ˆ }Ÿ#4ažìI°e -Æ¥Áfat?J^„½† ànI­¶•a¢DÖP>éÆC¡ðÒLPí´Äê8¯0g)|1›Ã¿Æ™îuèkZN†øióì)‹{Z룴ºõ+`ø$lÌe ÕÓ7dØ ÍÝ‹”â¯/YcØÀÓéZ¾/'SòãÈ3¬AM FOùۄ ‰Aê4ÝڊܤZòï6¤ÅÉì„¥¿[z²¡yÕ÷Ý/å(ÐVw/!ÄÈðýµrÏçTã”FB_'¥Gu–gÓuÒ%£|F d°ðGn|JÜÙß•¸¨l@$0䬑Â|Û’ò öÓÓÚn& Eëg±“øQÍ!U¢/¿¸R†»ÞWÒÀ›öÆØ+èÉò‚p®w¢X/NëºQëVUÉùbIKï÷ àl|˧þ”tŒ²8:T»qÍÜè·Ç/S){_:œ¼ÎJzüÔƒÚh@¥~²ÀƒñœÕ´àr îÜ,8:—(–&*rÎo—Pœ#«£÷³-\o£1P·ý%´‰ !¥K¸ó•×<¥ƒšÜ<98P[Ï\Bv+³£Ò7]ìƒtSå ó±ŠËÁx/è5©ÞR¥»h½ë“£¶+g'¬ðè Á\ú• qxRè!e,óßʽOŒ¾V˜‰²ÎæQ鵡á›[íÉ—þË¥xÍW ™Ðè!ÊÇ#¸£©×OèìZ?` !;ÞÂ<ŠEçrÔ1è2ûH÷1•"!»K5‘¨‰è©¤]ŽÔ¯ŒFx8Þ®#c¾¸îmË­›&¶x/­£èz»òB`epÖñ×1Ëá ö$¸FEt> ­X5õ*¦ƒO×¶›·5ñ3ú'ÅÚqf˜^ÓO .ùé¤dì\”$PPâBÅ!!âûKÇ”ŒÑEMVùW%Ëz2ŠÔ­kš–þ7á ÷f6Ïô¡~À'5¼ãÙú±|~ÄûŠ^°ÎbˆVL ºœ–\zÌ!qúÛª¾íÕëÏÕE=9x¾èZ83u‘WôÔ˜P±è«¯b›¾†ÖD ;¾ü$qב­äéÂP.épZ-ÿè¦sÜJ_ïy£é#ÜžiO­u²_Š5‰Ž;ÞDå/?»?c<`ÃЭxà­ñª’ÑbÛIî'óiÚRú.$ý|*Hƒ™¢ïù êâ¯TÃ÷ãæðÜûbh[Xš%/(«|Ç£‡Ÿ«€‹¼†¶”†f3aè§ý½”³Å†©>%½•PIMøU¿'l5šRȵ¾Á;{b —–¥Y8Á“ÑÍÒšW1ƒ`„¼_û‘•h—î"í>óoè ßì³1¶õîÆ3Å>~™£õ[Fªš[U¶k…¡"=$£6{•K–RiYzÂcÜž•;šx ñ‘×g©üÁFoq ²ÛÁj ¶ayÙ‡4zoQ0 ¾ÛÏë…ì±ÙÂó¨ìú_¶c‰}àž‹›ÿÖ1F,wÅæ”I~µKj…n€­™–ªSQluì2™oåã{Çš™é…uL™‘„õ‘Lê›b>I\+c±{tÍIõMÞª[v¯ô±¯ztsôÙe@¿ÚxÙ;ŽØùêÎwæošúbÝ8lì.ä^:;Úz”*Š­Ž …†?y"nàb&oI2:€Zšui*hL ÖÕ(¯­-rézÒD ’Ûc׳辷—]2ws‘e<ò*Lããzä¹+ ó©ûžÙ´òXuÓ]KáërÝIY¿‘qI ½‰È,8äíŒ7ÀßíÊ{,;ã+Û²¿8Ô×¾®m·ÅôÈ>ôîhQ»çd{Õnƒ…*O¼#Ÿ‘)~WCÊcbYërÖŒb¦z´•7Ú‰‘ñü‘|g÷äèwQÅ÷y…%„Åj -|—™z/¯ ïE$Câ*0=t6Ù‰Tªn·vtÈ™ñjm-˜ãµAñ-;[äN¶íòŽ›ù„2Ãs¹a tþr‘/&šj·îÑ޳앲£1^ÏVgÄÌýäÌ'cNØ…¼xóN!ÛÇãѱµ…âÁ … ç~`ÐTLkh=GÈÈuCåŽ&_K{øüùçcCw谺峸3çV#¬w\Åæígè…‹sÀÐcÖ~xÜ҇ŦqiÎ<žÄ‹†i.ÖÀ·=:ìKg“%9pŽ‚ !ã0—y¦ZM|ÇÒmï~ÄÒA+΄;Ë|'nó\‡A…á3NÿfÙõ¯‹†ð#´á®nâ”i|$Ÿ°GÔ1¹iœ ÉÛTÍÄKÍ!eF›Ï¢ß$],Q4Äò«,åþxæ½iŠ×~`P;6]>Œì–ÀÝ뢃^Þ­XW›ç*Œ*.î¶A·Ð›ñõQ¬Ah•õÖnÕŸãÙå\7`´ºµbXa8&f9 ÈgÊì°P‡XÖ®îSèùLIÛD¤²_Ù9ȯ&³zEX+耤ª¶Š-§ã¸fèñzó=>í8/`«Ñ'ÖpHUÖðê×¢ õ HkŽŠû›Ð˜/ë nÜÈ pj…€6=žŠ`*_$™óàåϘii+QÑlõìÑÙ““ÄúÑž¨ïÉO¿âfS߸!™‘áÄájóþ$WJò-;nRáÓ7sSºæöטïi¸­n bR¢^'žŽ%Ò—pû½»Ð3Ò„‹`‚˜b”{V‚)s[+b¿²1óXñ\~‚èJƲ"? ˆÌÙ¿¬‡òïØð >SË`½ìõ@Fèµó6ñgíj”»ÅprÏOmw­Á+\G$/›èì7÷F…³ÝâïŸõ¸ŠU MžŒÿ>ð)ó¼ „'H?sˆLޝ#:ñyß”EW¶µñZ>}Öçb—E`ï7¾$Ûøv§øUhÛ –²R,N½ ƾ"\â´úŒ†`göHÿ‘tá6Râ>ñæh¶q*©˜/m¢³¢áB“d“.âŽj+Û¼2Úl­ŒÀa £ÞÏUWˆnhWQŠÄ <±ð ¶½s£HËÄ™™TšA,^ÿ´³}«ô¢öC ý| Û:&ÎÑ«-ßþ;1®b©VVcß2‘Ö <šÈ„_Íý_3ƒؘI›±K:hyËc ¥Þ630NY‰D{š˜…Ö¦àå Õ!õäÇ)4<Ò¨3h§óq9Ü·Ig¢G§z‚ø’òÚÑbÜÑa†}¡gt1«ôç@¯äðÇñ;y¹hK-N™àVÔ¹³2â9S½úSio˜Û#ÚeÈâNAÖŽ‡×"<Ì’Âĺ«ç ¥%öxè1 ’’ö$k1ü¨{yba«ï‚ ¥ä†ª›nm(ŒÙ­7Ã=_WOsb;»Ç‘›2D<úŒ&¼}YõÁ/Ÿùþ£µ[z˜½sKÌÐökŒÞÉÜÛ Ž)?y “!Ÿ`ÕIt/«1–6YÑh–euÖé™%{CßÛÌ”€ˆ „ÕÆ ·Ð¹T†J}h©B™ÄG’ ,¡#LUf3(µ*̬D¸DQúçç±çmgK¦ê •X‰8þ‘¼»+k˜Å¹ÌZº3Å ÷e´m¦ö9¿±ë»®ô„m²Ó¨›2o¬PÜqõkÃÞ¹ùÝ)_¦´¡ Ô OBøËGÄŒõî&«ŒzIñ6%ÉL²ÀîÄ$ ¸õ•B®‡'^Ú_—“ù“TÎ.Íp)ª9V~xÁLü>Ë7öiôBȹ3H)aÞ1­¼áÓóA1nÇñÖÓ[K´&j­ŸJýžŒ1*G@%L9?˰ɘÉk°¡ º¨ê‹Ò'P:`eñª¸à{sż»0.©¤l:V8ê(õ¶›×®uAv VËñ“‹k¶E=H0BNtŽº·¨rßc³l*üÍÙ}‡ž²+ló.ç4°,Õ ÌñØÐõcê,5¨mCʱÒÜI¾%žD Rÿ2˜÷å¤ë=6ÌiÓá~Ï£©f'ôv¼ n¶¤¿¬‘tßg:=ûž;î<Áî{ôslµ„mTšG+h6~mÑ/~c]ß¯Õ Û–ÔÚÛD{¦N¬WÀœ¢]¿h¯Ãäf«ŒW¾S†•â’ÞÕhçÔž¨†#YÄcè¬×­`i¬àœÑ+m¼Ùϼ2-#õ£zQôëBóÚPú°³é#%µ|œùÕ ö‚×j!J$x] ÕzÕv”—ì‡uÓp^Ûô7{½^úM«§¯;à<›é{þ‘Xâït ”¢qGL,wÉKÏÄè dõV}èŠB@çE?-^ã%×i­ûÑQ _{/µ´`Ê9’¿áÆ]^ø1rZ_º|c?â;ˆÝÈ»t?j¾ ªkxŽå8Áj×cwO9ú5‹å0{œúgãDÆMœ“úY·ÜÝ‘LCLsQÒ«¢¸ëH´‹ Ü‘"'ƒ«0^QZ"¥u\¿, a´J39ÒJ¥e’•ý­‰{æþõ, x/³ãÃU3ö‡‘ˆ.—aÐ}ÙZ"”!£kÒ±õAƒ8~ÛO).æ£Â;µ­Åêñ^‡`°‰O˜ ã§ÀÊ'|¢n‚°~LYÝØUÄK­ {`œò•U¥ä×¾rX÷Ô4Wñâ6N-¦6}ÔìFT£¸Ï*^åƒ_Í÷l6.›pœªî³#¥W`Ï»"É¢Óä÷UV¨ ‘Ó‚ÑðG‹5ì 2/ësqgãð±à„{Vb¦2&…„í§Ä¿Nß!áü<¦ÞS†z«e\ÿL£siƒƒXã»O:‹ '©½S›þ×+²˜½œU™ãØ¡vmXv~{6>¤èÄê°˘ò§euüØ~ÁåËàK’C®ZLå¦Aaôi‹Ò;ãÛˆâÊ…j’5¶`6ÄbÇγð’ö ²„‚î[w¦È(*:2Ï«`átÑIü¸}éQìJ1síXçÜ’‰gz¹Yh”¹~™¿Å"Ãvÿ©Q {kP&ÏôÉs~ÖUãEžÐ.¾\ŽöÎaû‹ü†Ú™±ˆb)ù=`Ôƒœ­ƒ• ³\Ø­‚ÉÒ`„9Îæú•$'/µÁj†Õ€úFöîÚ]oöËÈ ’bgœŠ«c/rtq NrË3=‹ªªÍã ȸJ÷xV&t–/æ3*•¾'zòàÐȯ4Mš—¡áX~ݾ§ÚÃÑfŸupL’ýTØðÍ ?\0á{r{Ö­RóS-g0ÐewÒ¢¸‰;½·Ð½ŸkRXɺ Æ•-zÑí‚EŠ56o£¡æÇýqu›Æ,ym‹hhéÉFèÈ@úÊËG m8èçßvŸ×¹2…ê ³(lú¾sÐùñ·³éêG¿‹Çô½¾r‡Dµ?€É—</ˆÄ@Q{ÉJI'“G¶W´}téîa"ªiÐ× 0|¡¼:Ÿvõº7ŠÅ$\ò½rÈœkí)ïÐø«ÃáN|¥VF--«ã©tÍ5wR®|Ê_Cp¨U¢&š{âóŸ-?yEë¡â­’s[±ä–—½<°×žQ,¾¨á+Œf£‡ZÞí¢¨Ëߘ#Š~rHß›Ô3¡Á×öps©œÓµ3û'wÕknØ£O6G‰¨_ÂÝçy‹u¡üxºWdsÂMøV¯¹zÞ®Z4K›4°yKwµSäw7ÐêsÚú÷‡§É¾z)hIEE«#ß@úgIò” a/m‡”NVÃF-æÃÌO^èÊ«âýÃs ÁÐÝó2`¬éU‘÷Ø/2sJvXNG Ú¤hGºûÞ³VÐdÅ'ëÕ6µ»’›ÊÙ©‘Êsø'\y‘f½öÜk‘vƒd¨27oWJ·‹'¦¶ Ÿï‹¼µíîþˆù!©D.ÂMäâ®GH°ìÖNËOgmE\¢Wò3ºD+cÄÎæX„¦ Ѻ>ïƒÓep¬ÖЭ!¼k]9r£"ãǃ›¢©sè\%ûÕ*Í8shóÿ‹ܿb «èHÎHW£µ J9ùEbL >f!ÒÐ÷äøÅ·”ÃL ýÒ;í˜è×!rŸbÝ"£’±P@„•éåvS·,…n&…ò³ßDÄ)çEoŸåNY¹ º(›ÞyMÞç᡽mªhÞ‹™Ö!-yDžŽ)2_ËΧ™ä¢>þ?N# endstream endobj 148 0 obj << /Type /FontDescriptor /FontName /CJOFPZ+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/J/M/T/a/comma/e/five/h/l/n/one/r/two/u/y/zero) /FontFile 147 0 R >> endobj 149 0 obj << /Length1 1641 /Length2 9041 /Length3 0 /Length 10100 /Filter /FlateDecode >> stream xÚ¶PZ-Š»CлCp‚»»{tãî®Á$ÜÁ݃[.Áí‘Ü;3wæÿª÷ª«è^[ÏÚgíSÐP¨j°ˆYBÌÒ°+ +»@BIƒÀÎÎÅÊÎΉBC£ rµþmF¡Ñ:»€ `H8Í\_l’f®/qJ0@ÞÍÀÁàx#ÀÁ+ÀÎàdgçÿW ÄY iæ²(±ä!`  ÄÑËdmãúÒæ_?ô ~~^æ?é1 3È P2sµ:¼t´0³h@,@@W¯ÿ*A/dãêê(ÀÆæááÁjæà q¶a`x€\mê@ ³;Ðð›0@ÙÌø3V€¦ Èå/»ÄÊÕÃÌx1؃,€`—— 7°%ÐðÒ !§Pq‚ÿ Vü+€ð÷l¬ÿ.÷wöïB ðŸd3 ˆƒ£Ø ¶XìiEVWOWf€Øòw ™½ ä%ßÌÝ dofþðçäfi15€Ù Á¿é¹X8ƒ]]X]@ö¿)²ý.ó2e)°¥ÄÁvuAù}>I3Ðâeì^lݬâöùXÀ–V¿IXº9²iANn@9É¿C^L(ÿ±Y]<ììì¼ü< èiaÃö»¼¦—#ð“ã·ù…Ÿ#Ä`õBè²¾|¡ø¸˜¹®În@?Ÿ:þ¡pp,A®s 5ŒòŸê/f Õ_øåòAžöíqØþýËèE^–°½×ÂÿÜ/›žœ˜ºŽ$Ó_Œÿí‡x|X¸¸,œ<ì~n/7?Àï¿«ü›ÿ¿¸ÿ±ªšþ>û Ê­ þ¿(¼Ìî_4ÜÿVýßÃøïÊ)ôÿQ¾!;»ÅËŽÿgýÿIùÿ“ýï*ÿ7åÿï¤Ýìíÿ¸éÿøÿ?n3½×ß/Jvs}Ù %ÈËn€ÿ7Tø×&+-Anÿë•s5{Ù1°µý¿Çr‘y-UA®6Iè_·ðRÞªB\@¿ß ;ûÿø^öÍÂîå=qy¹«?.àË:ýwK)°Äò÷Þqò¼˜9;›y¡°¿È‹“‡àÃñ² –@Ï?ʰ±‚!®/)€z~+ˆ3Êïåå°)ý6ýAü/Èì?ˆÀfñoôûtl–ÿ€6à? €ÍúßðEyl6^Ž6/Ê"^l À—^vÿ€|6ûÀ—Þÿ/½ÿQêe­Ø ÿ€œ6Ç@n›ó?àK_—À76×À—c¸ýÿ5W 7gç—÷æò_†þ/üçq=(‹s ÁPۚж›*1–1áiš4ŸEçv·; Ä •YÁkÎWb{°–·¤è/E—È}šë#Z’ÔZï}LÕ'wZQ&ðûÇ?ˆÕö‘"¿fÑÝõ}tòÕ²ƒm†î”§ÉsrãÃP-À½ñè•ñ¬í+û>>·£¶[ùFõ¡lŠ%N+Ö0¨x†&ß<{–Á•…‰çÔsæòj'wü™\>‘ Åï0Ž«ÈGóÝí¬÷J¹&§K5‘>!)ì%ÎÈ$­ø^Š<Á¼Oɧ¬&Öa΂Gùl£Ã(”y zãÀ¯>¶c ݈’‡WëÑ´žÀðöáÝë@?¤Soé6‰:¤˜]0=Ïnáµ×äAlV ,ý«Nß›Y·#ùÉâ |y ú{Çjsô»WΩx(ôÚú9e ~Ìnž »;N­ ¤ú¢œ¿òÕ‹ëaëa쯩N¿ý ³T‡ÿ]5·s%³{7UÄ4Æé¼­•'NÃ!Xè&vNZSió~\»¨DIfƒj{~ÒK_¦=Ùâ,–B³Jw#õÚÏÚΉ?[‡à?ñÒ§â‘MÌ'òFä&\=ìè/¥‡vKgvÿ’;hïŠÑsÓfªm;¢žTÎ#Å}è½PŽ$ßÜ©—6Áž{Æ·É/rd¬a‰•H­#%é¯ ÌkûßZt:bCΘ5kgÉÍ¿!™®²iZ!¼¹Ã¦·¼{-§Î%*ÄP6÷PøÄúºK#6é:°Ž)×ÚllIý˜2IèÔA&0›!àõôF-&qy‰"nŸg¥ˆÔUÊúð|ªîøGæP”ªËAXëêʶ¬þã%O)øôä_]~ÔÖ ¢VÍÊàŒÌM³O„€ff½ž,NHUM›q÷µÜ¦_õ@žÖ6¼˜¶›]ëdíª‡´—ñ˜&ÿ óhaïR#ò Ø•Yû‹DÃ~ÔÇ¢¤DŸh™ë¡è[La÷E‘At~<œo8þ¢Mˆ1XéF£FòÄk„ÃïÚë„‘¹QòÎ<@§ód²èÅ—œøÊÝf—R–V'1Ñ.„Ù³’«H¸Xã,>^ÆÔZð’·^c’z$È`H¬£ ´ïe£ø t/¿º”ö6‘óÈZ° ½ß] ¨x Jש”n7 —«ÿÕGrû{2fÌŃbJ|©–ôÔÎåJžÃ•µLŽuu{-¦ªá÷j;, ´€Ðø#t+Ž_ªÍx¹‰B•É"—’|¡¤`–Ñ–q\o$ˆ]Á;­Z_á_ó!&ñßVܼ½O(Ú‰06srÚƒGL_“bs~T}läÊ<õu• ƒ'žz=áj„g×­%š“ÃkÅåh¨$yŒ„1.EnðlŒYVÍ¢?æp !Ë!J"­"‚Šz¡IgTÝx<Åž2¼´Tϯ>‡‹¶i2ß“8+#Ö~uŒ™XX4Øú¡Ä£XGO³gµ?d+ðØ•âÜwfÌf€ ‘‘yòqSׯ‹çs€M¶MÆÉ*¯-³ŠçtDãårš¦[dÓR‡ÂW5µ~tºrZEPñŽþ%èïç>t—>öOµ@Z¯—ÇI”öƒÜ“ïÁeËý³‡:–mNøuð,3±åK›˜n^¡ß2ˆ“†™ÉÂbKäªH+{¤íÑ®ÕIyæ–dË÷ }Ž p·ˆ×û”_¿î0ô‡èd±+ K†k¦F­] •™¾XÖœ 9Iݶ»Aàÿ:Ã’ÿé© ö©^Å*Sœ¡VkRÂJе]N‡Ýguó›Ì̧ã°ä=äz\ñêëèÑ0q\u"G*›ÖlŒýÎM“µˆËN¥_W‹¾Z’Ñ«éŸgcGûO»§T=:rªœÉö_mÄð³éi½âmWgsÃCó¸Ñå ºö×ÒŽ©´„_ÛH‚öÏëº;Yäš‚%{…d³[br E¸jgU÷p«Áæ­Ú—°Ã3 õ“úþ'ÿÂá‘mØXþg*:ÃâÆÒŒ%TDZýòDyžÏUÖÞ°¼ AÖ2²‘ãôAï§ï—êÂXYòe…Ö‰F‰0£øZ%åWõÍjdf< Th[aƒÃûôú·„×±ûæÅöѽ›^¿kK;,G ‡’ÝÝÝ Ägذ^ÜïH~¢g/!\áÕœä²8o}^á«®ÛA²£ì¹±–žËkµÀÃõðngizþ`H[Ș¸¥M¥Þ¹ÉuŸw©O÷Ó .Nm»o?üYP©6ËÅ|$½½›¿ºÉ-)ÊESïÁF®p=€Ì!ÌÈr¥Gè 7nÅÄeÙe|ŒðSaóUŽ[³î˜°:äbÆ´ñÇjöíÛYâ"Sj\6ïsÞÅœG:ÛrDÌQÔ È÷ýäVOëÒôçGQŠB:e­Ï[a4‚CÑKïØ؅¥ÊTS ­…íëç4V .n¿y´eé½Ço¯¦3ãÅøddBq‚î92aýMÏåï Û!þ¶ 6G+þé-YS ëŽCÒ³ª¿L•û«Äàʃ°™9Å 1ÑüÅúÝC ØúÆ•R›é„FÆl 8JòÂSÛ‹bH²ÖÒ‚ù¬ÅöáÇú[]Òv6¡›vφLÝg‰ŒÜ&¥ã]Å-Ê.‘Ò´•|X6iÐÈ@¢’‰Þd°ÓÇävò[ Ájg˜­y‹údë¸i^êgÊ1?#Ârì÷¡Ñž»4è9½rOðNçiczÁKœ;®m¶ðçjÅsGGug4c'òÕ3_WÎåü»â£4 :*îl2Öö’J¸é{I“÷+bRqsY‹Ë) xÙsíSmµÇ#[¦j€r÷ÌžCÝʼ_¿k§yþÄ\iJN› p^QƬ¶…ØÁ‰E6´Ô§“OÞŽ ݪ³û¬ðfN翨Ü0l®ó÷~¸ðàšÔÌÛÐÚà”Á…›ûA®ât©òa±bÙ4 õFc2Þ8+·è §’°­BµïÄ<KëèbÙ±T–z5Ä•ùDaëÏ´ÝØÉ ×"TÓA¬v€²ÀÞó+R]ø«ÜO™A³–Ã!çvɵ{,Ûm¢njtøÏt9wµ¤t–ãú¬] mI†֪YË“ϼë±Ñ"ßtXwÒL$}¡Œ:,ù£EÛ}¯ðVÄM¸w)¶h?‘BØIîóBñ…ÒV!bZÿ2²Í(,È0¡ÊqóKƒã_æ·Ù(åWKÃÉyyÃ"ÁUëó28"jö¬™”Žúrak¹? a6¼ÕÐ`Œ¯êÆØ§O‘n¸ðÙ7v"üÞ;I“žnæBDk.q¶ŸùG§K‰Ý¿kE¨ â-¤Þ>¡z±îtÃ`ÍXkÎ#4­s£À,Òlåê$:Fâø°;¢XÎR‹fP¾`ìS‰gÿ)\*Ij îéVÎV#±[ÿ¼®›‹hßr¥çõ6{½dx±C©ìÈekɸRX4ÅB‚Ô˜úöm6ZœýgÊìQ Aꮀü„2WøCúzµ©"”÷S< ᫚Ÿ/º¬Æo›~eXŒHFð˵‰#M\°I$^Û⨼¥¦È„àÖWvÁ9ù0?­äØ¢zUãKN¸G¶ðQP·vXqNÅì¼ZÒ7÷¤`‡ò’û¦³Œ©Þ‡Ù?Ãsbr5;×±ÉùAdä1ät çMmvæöe ¶ÏH0{Øek¶m-QJ²v¨ùØT—žå2‘²lrc ²;zš»ýx…M>ÓëäÎa¾É–°‡LC®"êYôø@©ôò£òÛýoƒtoIM` fÝÑo-<4G½mGu#@ÖŒ2øäÇ!¯÷À1-‘ –¶Ó4» ­:QýÆ r¸o8 d¬Ζ÷Gï*[ë—ËÓgL3É+„á&¶6šÄq.ÞW€ø•z§w3Ž‹,‘çzßšãï÷t«J¼øo©² '÷¢HjÊ¥€&5Åy倯¹,•ôPtâË«ï°!Ùúq)nTs¼.® _ ™`Roû¤àÉîSrD£±<«B?Ф öyØ’þ¨Á|Uº†“\‘l÷¥¢á,’XHïl*Šú«‚ ÿþöøp V¥®ê–Ëü¡K{ððò²x©x“BÄ2*Zܶ–mµ–|ƒ’¯w U7@ÝŸ¹6RUR”óyS¯a8­?—õU¨×ðTÛ;¥·÷^Æž„¾™ŸÙ€ñLmm>! 4/;DÐ"/¤…M¸íA§p4y›Ómá®ÈŒ<êoZÍ#/ú&ô)r—™€tØèLË7ý—ˆxEÖ‘†TˆÇZ°ðwxjÛçU­c:v¯§ÀmIø¯ùöq¶¦ùF¤øºžš (#­×öÝpÐÏú" +ü!zeaúQh,¨XŽé_fŸô>…¸©‚Êýyö?(&PûvßrÝo¦x ¯ýhÏŸ}p×O’REh‰´Þ€–#ü‰iûóBw’só ›ƒ]ÝzÍ!„ÞÒ£šñ­šEøgɦ@3º‹Ín4 ôHö_ÊY%“oë›×•ýÑ‚è¥Hcê-¥…„ú•7ö.}ÛPy5݆¬ÌuJ¢--¶ŠŸû‚¢‹¾G¢«6±&Í\W®ÈHäõâÛ&óª53Û³+iG< D©“`cÑÏ×?¦Ýá¹S¬°A£mv…¨)^ŒÀ[ÎK6rÞ–ªÂ/¿÷üƒ§…_Ï©{§…– •}@8óÕ£¬‡C>ƒ¾`¦ú5­¨¾Îö©8Å6‘u¥þ$ônÕñkf¼çÚ¦çg8Ì1Û£’-ÚæY,›úzHr¯ˆWc‘þ V›y«¡ sÔ.DL©¹z!¨îAÍ; -€_1uðaWì]n£èð•¼û%3àA²«vŠÚdÖ³ù†7[ÿ6µG_ªT\Y4žÞ†½ TÿÁMAzªÿ¦MeÌ÷5Ú_B fµ´/ËTßp o5F¹ÿXjäNF$ïp¤tS GÛø±.3pÍå LOò%(D ¶nrE²SpH¶F5®.ÊxèJðÇN3=uäÙÚ\Ëà´i»ªÑ3$kÓ_¶.ŸÇl¶IhUÔ\ˆÔõP¯Ò¾ƒëëà^g\w{m Êm¨ž¬DUHã¿Iõ³/ æ@åô(¸P¼ô«>qË» ão¿º PM×@dâK‰W“-Þè2M²’þ÷kñÔNRhýS÷Ì9 æ‚fÓt—D„±5|àa‡¼˜–FÆÁçŸ[ûT(«Dd—Ë6‰ðæév¬™ÛÞrPÏlÂ)­ÆfÎÞ M1RÈßä}~“øz,qÒ»Ö…oÒ“ÜÏôºVøZ½þÑ -w«æâ=Gô<°¢ÜüKÈ´_n2¸U:<^ÎåqWäÈ­¼­k Ö™S•À_Qv_Ëþhïê-.@¨ /jgŸ .0ÄVð‹<Ï_¸` õ¼®¼£´xúålŨl×KÍÍ6ë0ôp7½ Œ'Ú[é“ W"z¯UÊlÁÀ¾É”Û'ùª?jíQ~†ÿ«õÊôs¡ ÐèàŽ§¤i“š‡U‰cJ»tÙ÷¦Û2*zù8Q:X¿‰ácq2 Ó‡3†£Â«ç 6ˆÃ'÷¨ LÛ¨º°ÀðgNÑ´ÃÒqEŒ WètTDfå†×NÌ[ò~ *ŒÉFä@ =ØBÐÁGYÛqí¸>²|úšó»{ér¦m;‡QÞŒ¦äîÖ›•XZ‹·áT™ÑËâÂÕ¿žú7ç#†àeøÀwu’NÉ›J…¸e3j‡òæXc_Œ53§ëѾÚZOWõ¨/‹Ûé8Ž©FP™öÉbe5,WÓQ£9ÙSC?Hƒ=óï d;ˆ™á+z£>6Ðïç¨éÉ•tŠ9Ô±ÕšÝsÎH¾w4}¾e<érÏûÎ`ªêzÆÃ§uº*~^­ÇÁÿ3wöÎ=Ó××áójÍíݸå/ÑÕ²+þë+}@𠪦6md^j©@lx1hâAƒOúVŒX‚†ƒNínj eÌ£³”ûâN$Õýë%›³ü¬^Í.{M,y݆þôvRù(ÿÔ;nñÈ3§R¦4/»B+¢%Æà#ÿœJ' µW'®N|Áª­]ng¨—ò+^£û•ïÌé‡SÁ¬¥Ç ³¯eÓbÑQÂ3èLÙÁ³ù`>5è/|^ªðÝ£wg˜ÔŽÛð¨6-ýn…ITÇ +­ÞŒ÷!XöJ3«F‹² ÔÍöýN’U[ɨý9ˆxÀ̶7ÆoF ã$ o%É>¤~aFÕiµ@Ôª`¤KGTN& xaîl¬¡7ÓNk4Ð}‚ŒZ“7FqLùÄ~‘+Š’|¦b =D2Ûo)Ȳ (†î„Ë?e/›§IõäKÄÏN#s¼xx¯Dyô«!ëDxî¬S•–%J5¹ø¤æ­Š'V†ž»ÓoÔPàïii8½uÁ€Ý§|ö©98fÑÉc¢j’ÂD¸v0˜'‹3F™œ¢yR7*¶¢=r‘ï•è½ê$ï€ã.úþ”ä–+5ãv)7I¶0·ôŠ\!ü å$ühìýgÔðVüáО ô÷¿G\¿O±©àOÔéá@=gìrz•hq#­øô3ú8þü[ù® ß9ç“4"¶¬ô“y˜€H'yº‹’ Je銞ÐyQ3ˆîSø«êF²¢æäKöˆÀo±âþ¨ÉurUë×"¼{Q‰ Sæx"¾ØoU»Ý)nòkK±D#xž‘äV+$UìäÒøÐÖŸÅÐ:¼«ö< ÜGÁ¾âÊ -ãØŠ©LÉ #•Š%Å¿¸d+æÄJC†G“µÖ:ÐKšÍrœ×Šû©÷#5Ù L""ÈØB”˜lÒîÕ÷æ0šÎ –SQ ³CMXámœ’uŠPÙµ£3ûó%× ²Á·'…‹o§ã¿/€ÞØ™M{ÏÇÒãÀ®hðûÀù@`Â`‹y’vÝ숱œôרÇJ©ùq{20÷‚ü:~yY·Ôº‹/yÖFNy;bm‘–OÏ&MoZJJ^käX—tiŽÇßAbïí|æ§Ýxà] W•üΫ‹•“EèKå”jHŸdKêlŸ”ÈR¦$èš®Œ ¨?ë.åïUà3(—½yV 'ªsS‹ÎíyV^%< \+3ïf*ÛWM·5ûA2úÝѸåcÏZ-)Á'è7Öæ5;‘^÷qŸ&ØÕ±Žè2FW¬F¹º a¼Í1ÉœƒƒÓo Ý2湯æXÓIê“ÈQ™Š^ûHŠ;î¶]– ðÁé$<ôHnHGÍ×í Êd%*M¡åÄ[OdÅYïU£ïYWnÇI–:ðü8:y“Jìù,¶Öo'óšùŠÞ(Þwx€cò!KâçE{½È+_ñºbë%6ÅRœq x_•M¼û7è¼iQfQ "WycQF'pHµÅ+“VÑ„HÆ'ÙOæw+| ÑëãÀe’”~'‘~Íøö«ÿŽsf!¢Ã­8F›ÿ¥tx©v;à®ÿ'±™~Ü¡Ý÷îCnNiÁÊ>ÒŽŒ,ìò­¤ºÊ7áSâ&]‘ Jƒòöú¾œWk©>äãWÐ?ø ,–bWlóô{n±gم›8†emR*<Ï÷µ#UŒ¹Ù=J¢T')ðê‹LÑ,üæ¾ôPL­g˜ªì­é:B)alÓ½R¤c)ýù£n$Ú®©íÂÒ=eÁ¦BÖ.Õ£¡0|]2.öˆ?Ð>uÃ(ÎëôdúÁÌêqxC:;ÞÚσ?&s†õ¢Å“Yõ,©¤“qèáZ~ÍÁcGÉ­Iñí¹aËò2©“=Ï]Ò,FÖ´ í»/2´SqáÂb£˜ð&ŽÊOØÒkMyz_{Ù’q~qÑÈp®“Çâ˜pÑÚæ³™¦×3|‰t$ ×až‰®7½ÉK=3åû¤¾É¡K‘üÕ! &[@p•¢º'Rk AN$æÒÌǾ˜rî%‰DW ;ØàÅí… ±ƒmD:ªÌË Fmïc’³“˜™ÛS·;kŸ™ˆŒv”ŠäMÉM ÷e;8ú`^O‰?¬¢NAÿìŽY” Ž/YØ^€ºµ(Jâx–’§"m2ãC5¿Bzw¿_ y‹ ƒä@D*ß{ŒõÉFd*áD¶™û=’_-iè«q§)dûZW=^÷•úh>KY«¬¡õ½}‹¦ºOJoºï”I¹hÃuš%Ÿi´ó@’›‡úF“êèýsªÂÚ‘kuVý³FÙŒdÌÂnø´0 Cꊺ.î ~/—¢Üï Íbdó–éßmXkèô(KßpØ ø!×ä¿U¼\–K%]jh{}  Û¸®ëì äCÃ$žqØ2’m%­V´¶¬xnt§‡Npy4æËÌ8¨P›]JеpÐV¾]‚ ¡ºSö»¯Û«2‘þ´„‹4°fŒý£‘oïjO‰:åaâöB¼ò‰®¼]zWÆ¥ê¤(Tg´!š[M§|ªÈ{ŸˆQo*aÝAôþ³½iƒ2Ã*€åˆÖ¾ ÕÝ'z. yÃúÒ€hä<+†j¶Äí>TJ…Ð>›²Á׊6èŠiQ€Âmû¦Œ5ç÷.åÁ÷òPƒÖÄg‚Éü÷Ÿ}ìŠÑ„·«æUxµ«dg²Z;)'ü^U¦ó6ÚŽã®Ç¦hÙüpà®0:dº ø„z|ßÀ/ýcž7ÂÕª7ÉÇ%û[ž_MÆù´·ÓëêÃ:µQ*ý\_z ÕwË4C¿}¦úAüÀ ïžÎØ/Bzê¡fÚ ûœ¡?|Ò=^Jí]ò–)rhè“  I•ÃB=ËílÞä©%Çš6Î…u§æ}p_ÞB™?—K÷Rp-ؼí²~Çlü“é|o²á6žöóí œˆ^L}OÊ;ˆ.Áî#JnË/àÆ `ûsÉÊ´”ûðö–NÛ…Ïøc)ÚÚ³Œ7eЙf¯Aˆ£2lÖO $›ë6â;:µ«f—AQ±ÔÄÄ´p“Òˆð=¥YÙ1jZd½8ô°\vü ´H üÁd\ÁÒí¦Ù﹄ò>¾\­Ò—Ôœ£«ì)¤‡éY^HèR»YÈW5Ý”ÁVØ™"jñÄÎMY£bX䉟Rîz¡òEºyÃòw›Òçé߆×*VðR¸nšÖ‡Oq’™øѤÕ¾N5#É|–Mï”ñæ[vmæ\é˜ìøˆ‰(Ùcœ¿  ©”Ã9ï´¿ÁEœoBè;Þ…}$»m›gÅçB­»†¤X–:…/`Â’ĺ§¥93Ë·{ξ3©ÿù¥ZÔ– Ë0uuQ‚VÔ]‰öÝìóÕ–Õ ¼dbj+š£#)¸´o»ZáÜöîä€ÿg£ˆ×¢·Ê-IZßTÎ8ÎÊ®êîÌ2N˜ÿö ¾ôVù t#I;Jyìmd´ fÅ^^­UvÛº,÷R=ƒ[ã{ÇàÚ! .a=ñ›²Ì`ÈAçîC ûòR*Ë?ÆcSU…n¤V˜m’ÇM‰vJH”Þ»!ÛYgå²' º#i:Ÿ)<°[’r Ťu\:+.4úp[jh«S–&zaÝ5?â¢äàlkUºz˜‰ÌÖ¡§æ‹žw«h^}<&6,͘™­ùÚå‹(·MÖÇì)='ª¹ ¯ã·/SÕÈu‡Á\%«EwNº)lu€µª¬ £(Uöâ`ªXÈñ0‘aVÝéôßÃ8“`¿™yݼÈýÅ¥¨– GÆk¼ž· =ÖõkÏù°1aÉÑÕ ùëJ¶“ìÝúÇf5p<œöÚ&#œ‚ðo\ ¤ÉB¡Ãp¶Þ[ A%êú=²] Oü¬5`KŸêÉ:nÐ õÂ/_úa`¾êµ¢oÇLtD3g• ­¢£Æû£Jé%hÿ*¿Æ˜Zý‘® ݆a—zCMÔ…QìOÓU’á7PdÇJÛ6_ÉñzÊ[)íåPUÓi¿Ê®€:÷aÛ}”„.:YWN›w¢Ëb–%‘\ÈV-=zhÛêá+?œ¿+« :вlO\÷“¹ÉÜæò,Óª‰ð|ZÌ¥A5t¤O•Ðd>ª~dý`¢Øç}*ùü.òš\¶óî|®1¢ŸË¥-ZÉoR(€ééÓÊØÛvšöGY?Òe£;ÃËP8W™"XB‡Oñ€OdðNo‚Œ •Qk•©tÙ`þJ³5éÎã–|_\G½•iÍáâZ=q¶ó1¢Œç.Ô„ç„ÿ ñw9U:>3¨™º!ͼ”z½¨hR{,íBLF^‡KC^ãs N® hk×YRÈVËÄF9Ô<öqß¼2ͼõ1;ë¦Ð‡Pl’°¸“™3‰G¸Ý,§¬\Tž?¶òJª ÒÜy{Ʋ-bj?ŒÀ9«ñ{ äÛÝ«…Ñx–J”ãHéÉk ïöe?Gv§Âò=½Tîñį“ƒ#0Í¿l2ôJ6|\í|”^ÆÈ]m#j¹ß¡Ôã”HÕ8Æ}~Þ&ø,§FüS‹œX3‰WKç$¤ò8,®˜BC™;S…]ƒº‰2FiU¼‰¦âý\eöz5Ñž%üdoÏδy)OÏ›ý½­¾Û ™,±;oÇïÕùrŸÖ9'1iÉ@íú~þ+èrÇkDâ@µf›CÙú ±o*N|¼£'ÃÙŸ·Zî½2p$NÇ Ñ©mO5ú9kQ@ôqÃíÚÊD—LÂçÕKô¡æC¾¸Óô2ßmйY ¾ ËÛ#‹‹gœMÌËð‰6äž‹€oV×Y-i)4;'êã |eIäŽø·&â2Û>6ß=_•|D:7ªºÂvöËiÁŒ«Þ¤ N«Vô']&b3T‘ešÚïØŸ9b„k 2f~H¸»šÕG—H´bNϽcŸ¤‚Eœf;ëήŸ¸ÿÒÙ·> endstream endobj 150 0 obj << /Type /FontDescriptor /FontName /YIARWD+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/M/a/c/d/e/g/hyphen/i/k/l/m/n/o/p/r/s/t/u) /FontFile 149 0 R >> endobj 151 0 obj << /Length1 1463 /Length2 6698 /Length3 0 /Length 7686 /Filter /FlateDecode >> stream xÚvTnû6%1:¤…IH³¥;¥KjÀ#6#H( ‚¤ ¤´"‚ ! "¡„„´”¤ HóŸú‹ÿû~ß9ßwvÎö<×Ï}=÷u?g¼× MD”QŽ0 #ËUõ¥`°¸(,àå5E`<`P¯9 íƒ@!åþ—] ƒbp˜ƒsÓG!:¾@ˆ8"%‘–ƒb`°ìߎ(´P ê‡pê‹uPH˜€WåˆF¸¸bp§ü½ò; !²²Ò¿ÃÊž04 ŠêC1®0O܉NP  Ê ÃþG þ[®Œ—äïï/ õôE¡]„þŒ+ÐæCûÁœ¿è  ž°ßÄD¼@SW„ÏØÇøCÑ0 ð@8Á>¸_¤3 Ä 4ÑÖÞö‚!ÿ8ëýqþu5@ˆ(äŸtEÿJ„@þ†:9¡<½ È@ÒGxÀ€·5ôD1a éüËêáƒÂÅCý ¨#ÎáwáP †²Šã÷;'4 ã#êƒðøÅô+ î’ՑΪ(OOãøUŸ sÂÝz èw[Ý‘($öÏŽ@:ÃQpöõ™!Þ¾0mµ¿0O„ÊWÿ?Î2@P ú üG±N¾h\Ìo1á˜ü½ÿýZÀ`0'ÀÔÊéæ=·š{oŽª•ÙüEV†äGyW,²D°Sèß*’t#æÑ‡Êéý4Ÿ—Õù”¦9ϱ›¯H¢›RšOƒÏìSŒ?¬4&Gß?ÝT®í¹FÆ.bª´|îlîN؈ߦÛïí+CeXHäß­PÛS1ó>jbÅhõ…”.ùYÅG‘D³›ðÒ1ÞÇ'ãÌ\Ä‘k¤‚t{Ôc‡£tyל:)B€­Dñb¬õ‚XÒñxÐì3S1Ÿvkæk„tï?ÜÀª¬gè0}–ÏÙø–ÄŠnÇ³Õ X"“¾ˆH‘y©ØGx}ÕeñçVyÃRX3¦j:ãéê¹påÝ*/ÉÕµ3ðA^ß]šìĵ ºú®¶³§ÒªG¯]]’Š Óá¾Î*/À—ƹ‡ŽŸöãT˜æÔÜgɨº¢ä4¢Ô ÒÚyr°  âùs¹®#Ê;òºa’ô =EgdΩÂ:v•u÷ÙXˆ„÷†m#ì AkLL¬ÙÈ¾× Å¸»–¸Ýö_oùì½E@¶ç~C8uÕ^1ìÑû‹þ'®T¤âaÉиELª@njÆx<'};ëÐ5²D÷®ŒÕ¸t‡íDëç¯Óº@ûLè(É3Óëˆï:“|I[±èqAµ¦ì¤œ³¾«ªqì@ȸÂ|ŽZ«¨'÷ú›$4 RŽéýÞf‡R˜ï¬”1N&·{!ÜSÜvÇ÷¾8´çÔu¨Wš  Hw¾ÖYì‘»©Àã;fÚ" ûM&‰›J>Bíyv7hžSöûª93DMŠ'‘fª°X›À}%(ÖsË:Jšç?ߺÈK€L²}\*°\tO—fÕÎiTàe;[½ ˜ä›Y1}t)Pz•G›wèÒ{á¡¡6w­ˆ>]êðºy€ûUg@Ï–ÒÜ›±™ÇôÖó:R®a»Ï—¦å]•§«ç‡ÞËÑ´q'£'üšæÀ°V›sÎ}ê‚mü΂žkàÔþõ+N†/T‚°=¬…p<ùù£zgHÜ×YÍl>läS»’óZh %º5]­ê¶ur™z¿èpògùÍT~^»²˜y{“@C+듯{z¿ðÏïm{î6±À›XH®Ð±'‰™æb_›êyîƒð¿Ñ'é=|ÇA¤;©Íó%ç`ñËâÇ{g“dBv ~ʵo(õ ÍÎ ?ö­mË×ÚÊý?,^å=%é;W˜Ví¬hÛ”ü@™¶¿{Ý¿`Ôn#•â˜a²»xÒ/Ÿpÿƒ„°HQ¬óÛ›Ñæ"REqìòV¦b’Z­7Ì»#£“öIb%;i­HA~i€úöçàsV"cr¡ 8êÍbUV„Õ˜+d zr^Ö£0Ã'ŸþºW[甽 ·}­Êåã…Œ¨Î‹€aÅOäÕϦݪ6 $Ƭ Þ‰ð^è† ú‡7)^àï3>]ßáÎx•E$[ÁzsðŠHùŠ@¶a ¹tLÈg»:O_6Ç“½b'qiϪB=¥:‰o™Ûézr™?üYõ”K½X†Z“æ¨`µçK÷ëîV•·ªµ7q]&YÉ6vÔ¬—Ù§BŶ¹<ÐK>CU·´iMe˜@ÖŠ‡Åñ´¢þÓÐHzÖ.uÞ§Mv­U—Oð3ê.!ù ÚÀÝËpÌÊÀûǶ©âäÌBFP§J d?îPiÎÚõIÊÖdׇ ¼J¡û$§ùG¡\ïÚ€ÓôÎ=d´ö¶¹`éØBÓtû˜ø;«ýÐ^J‰ÔvKÖ]ôrìË¡ô#…/•Ôœ7H5{ÑÅ6ÁR9eTúô`þ%Ì)Ÿ»²T„Ðc*ÉNk3¹ùâ÷;ü9™ ×_=“c}Yô]…#Ê$3n!Xoq ;¼r¾'³üçôÌVÅx}®Hð„E·ñú“EE<_ºc¼énŠs‚­LëÎêúô˜ˆä«åñwòS- Óù½Y›©Ø:Ÿòëž6áém¾´vš®}œ“ÄJ§› üì™É>”DÍ'RWS§Ø_9žèG‡óÑ ~%­ç#¬(é‘ZQü²²‰lSÎoÄ>έ~Ä)ÜüƒÍQ?ŒÙ0ð4`irŸ_ågÀó§ÏiôëÓT¶åͬ.Ú¨ãűԢ—ë¯æIxob¶gÉXçQÛÔ Ç|léÔ>óÅk“ÃöŽ ÀµG•¢;˜Ú´YÌê£-;øä±¶á‰ðž“5ó¯Uã©•snÒæP{hû.,÷öè6Ój)›HøÉUÀ¡Ü»ãíÌ™;ßhw´ž¿!S¶$˜VßlNë¾,ýiæ’w˽λO¥«V¥ruâ Û³Ñ&ZWª í> ž+­JÓrž¢¢Q ïLÞ / JˆÒ¾±â«ãÄ‘ÙB¥›·²—íš–è`ëI&I±½&È…Þ9ÊxÝ–pÓFfWÒ7tg厧¦j˜Ü~†— `d”õîf´u–'K¦2FÉd8xëp8ÏsFH¾ºÂ¼›^‹Ø"ßpdAü{ð)A¾—ç#0Wsq«™³ôZý§V–¾;æ–¹»é¢%˜–ïZCÐãõV¡ùE¯˜áÓu-#ã¼Wû"‚8îkê/…ŽÉ'ìª {à}Ù×£5õêµz륳|HÁcï4gôDngÏ-uc¹l^nü8T™›K³>`#)'kiöèG`œ3ïÂm»K`á]÷x׺¡,e¹9,»c9:š¯Jød}7+7–t³Æ[¨w©·ÏæÑ6u0ín`úZEºíH õí£ÖuØC/RóÓ§ÁÀÁgÚ¶ÖÎÏÖƒëêP­«Îæ:¹áÂ}˜ÿ¹Æ¥'¹,#˜{Æn…'~Ì%ꂺ!n¢2ùìú`æ÷AŸ9™7dBC˳Q”鈬†© 5éÆ~Fµ³æãªÜ+/fÚQê§ú†àQ°W9²â}¼ºMx÷TÕ©Ðä5éŒwöäÍF?Iƒ"»c)‡™j(•‡åƒóó‘<„bŽWÍ»rZ-8÷wëŒ÷|÷²í¶m_I‹qG=o‹Þ{fXym¢ÎØ,Y]”~;wrd×ìœ"ê¹7E-… õÅ»«ßV8ó¼ãJ˜˜ çñÞan\uÅ_®ëì\4ö{!O­îˆѳÚd-É Ïz Ð•sœIòh¯gâ1ÃQ÷ÕfW:cåÚ-±î9âHÒÛãŠÔ\ÞsüÔÔ¾ÞŽŠ>ðu? ”°áœéÍå.§À+v0h‹9T’DÐ4Û«®<›R¹µ$_鲩{› :¶\ß¾=–Êï 3ðü‹¬ht¯qÑ@Y „Ĥä›2S¯ö§Ã üx˜1¬"‚ùùDè`…osœÎK7µafˆì^Ÿ®…´Þ%>ÝÇ6w­utÏ]¦-UPc/±=éýÊË×é)ZîEA\º\€¯?:~»Rï›AVºdîŒ2m´¯Tç‰*~ Ó³VÝ wüJµ@3ô ™Y ”Û÷V´>»@².x«Æ“¸'„ø’#$6þÖ]ëW¯!TÏöÒm¥ë3—‘ë÷ÏÓWÛ¨äi¤b7B9¸ž¡ô´¾<ÔTcqêYÐN4îä‡Óâ¿¢€P+.¥o(‘Xé’°©Z3Üç­Hóaðú8üs.`ïðRë6“šñ'×{z§± Ñ3´KååŠ á¯‡J3¯“³{»-§#¹çP½} Tk¡:;ðÓ=VÌJk&-Y“gSgªû`,5óv‚®P8%»ÆÌQr^ÝÏÛ…—Ÿ,¦©©¦7æ,¬¯P¤û:±ÑžŽ'1”$¯™>…äÖ=Î*ì¢×#s2¶ãnjŒphN”~ã;o½&WÓS^ä|u‹Ñ2z¬ýæÀv°¬(À´Æ !>kòA”ð8ze„}ôz}ËÌñÌÝŸR®¥±¹w\Ë×ûߊéÁ‹¾áÒ‚¢nÆw¦%|2Íý^éOý€>é¥ ¼pÙnybås"YÒ1¾ UGk‰lÉE,ò{O%„~ÎçCâZjÔb€À$­C=tüºÍv;çˆk+~ãþ`¦ špö*Š’ŒÌ×zF¿•ÈOµÆ¾ðd…a±ÇÞ»ÌX,¯—ïé˜ð?Iðéé4^ɸi©ž›…ßð+u.ãLžkëÐ N˜ÊÜÚ»³½×ê0X÷k?J¸‰ÍØbcåz´‰û9å4·9ȬåõmF4°¨F‰rÄâñÜð±´O:^¯Å»ÎØÌ2îò0yÞ°v5'¹IR[ÏýJYSJE±ž³§Dw ³»Îp£„ðl_(Ö¾|k8“jJ¹¿C«¤cÇô.õw M·F­Ã÷ˆ¦ô*/à´=G¢ìkŒ0´Üi¦4dòaÆø³Âi7ÿeÀ‡(íéÔ ä*ù{Ìb¡çàô–ÇÀš¯Y™†!®KÔ3 "éCÎ..œ1ôRÓiruŒ õ¦ ÃÈu.©üýA³PµÐ"Tý*E_*¤pápWQUyϱî*Y€%ö^j©8lTÖHo­àŠÔPÃA-É6†$$÷ê’õÏ”gd¼"JîDSùo4 òWr5ß¹Qé[(×/Høí»#2H'ŸUVÇ…VUAú¸Œ³ÞÛ{<è2z4Œ¬)¯Pš¾åX3T %ÿ4½Ä&DWƒûs4üdl¤!‘š¼¾%ß(õ õ²ìÓ÷ã§ãTU×àcƒö©ÉV›`K­r^‰j!ý-BŒÿ63p$¡#~œ”?äØ#ìG÷ôh¨Jѹ]€SŠ8+T†Írþ‘î*Ë®‚F=Ц‡|q+Y¶a¤¡#tm$3ûàFV¬…?Ð]þ M[/Âhâ,±ÀU`ý®ÁÐPÙ ãr…®9¸Sb¿L)èp»”TNò y¬þVnörHJðÑŠµA³YïϤÍëÜó”F[¨U«3rµÛ¡«(¶ŸäÀà\¢¸%]=²78@öqSÌö¸ìÓE’ïX¨arã&]&œÖó1³0øâ\Ö«ÌS&YD*ˆÆl¨Ù…à+M–蚦sá:U|'«²ÎÚU^ ²[3#¡e£ôöbÛiunÍÎú·»Ú8¿àÅ­˜ÌxÂÊÙ:¹ºZæÉ‡å =ñ+dSãœ~ÀûƒïG+ÎS¬¤×¬®+µGØæ¼b1s-}²ØŒ°üæ7Ó–jè?#ü>%Ôžz©ûx^¶A/¼ƒ>¡J¡Ò¡I$B&ô’°ã{fÎ.s•[ÏÈû­ï?,06 ?wÞŽ|×l׉„{°l­´n*?ú߇+@%ÕTixÊ¥—<\’[ÏÓë%vA‚ôYìË<ú l•%žKÎ<$µ©e[vŒp ^u_öG w‡F†§ÕLùZh°š×óLSvS)kÞo§#ZdN4g=·ŠïAây§m®ÓhjAà&öbYãkwrÞŠwƒœ6$¦&*¶SðtçÎ^ØÕÞ%_wÜåUaUâ©Ðw®ßâ‘áßôI7|gáTbu¹''–þÁàÆwoZ¢ÁÏYÉ<€2s+¿ÚηZõÄL&ÀM4¹âƒÊr©þ/ K>ؽ¯«Ê•(-ƒh)¾œÖp  ÓÒRãM𳵸¨hN*Þ#Nh@®q4ü±ì>I>àWsÚóìqLæfÖvrìB\“UúÁšËã'½µ™o6‰Œte^rÚííL§‡ÁÁ®ù/a·ª.e îSFN¾ÈÆ ؼ1œ?ÞR‹j´{Þ Ñ›ƒ¼¿gsŸ¦ZµUÃq÷‡‡¾~P¢(gÓÀRÁžöÌ ¼j™¸E¥ÈÔÐïtnñíðQSó‡DLZå3VLBí^:!&€ùþO¯Vñ¨²~«ûçÐBŠÙ¼N¢:A#+½&Cû;ZD*tÝF.ðU Ã^+{(mYÒÖ€÷3„@D^Ë€¿|yƒBNKÀ;ºÉÞß‘/»îV¥Gܾþ#·$–i„¯x †Tx=ï¼o‚m†ï竇Å(Á ò\¨ D¯98sõÔEÒ1—=C“/­ÐN²ú ›$$YLDºÌúi˜ì² .ÞkpSͯ´%c‡ðÜ2›+3+öÝäA½Ôíú0S"£2³ª–(kà'Hº!þi¸ˆz]¬ûÔ_Ñžg¸9“GI¢àwÞÿVíý=Ýæm¸Ä‹…šNí íÝMÑAý/ÐGðîwŽwXkUÜcÉ>ÅwOk%ö¨À/¤ìØ—91‹oýÉZ ´#Ì̦¹ˆ·–mª»ìý"Áçñ*JüjQLô¶éžP†xt¤«âÍZÛ=@¡½pvÑWUäG᱂’㬵^z”!KØ¥Ä^+­4¾÷󊬋ZKyVÆV–€PŠšEù¤§ÃèMµÏ%Þ’*øÉ£|_›š‡YO{÷…ƒR¾¢3ÈvEgÝŸ¦iÃ¥ä‡õÌ”•xãÃ'¯æDMì„…g!ºú‚Ÿ O41Ô2âRóJ³·8K!ËGU¢Ž:³–QoUáä]I®TcਢbbÃ$R‰ív´5&[ê¤}v}‚¸{º q»S×íÅšÀ‘oÄx¥&Á” Î¤p”>÷ÕÉ͵ï²£x~wë³ìÏÇ×Cx¼w_Œrèe†Ó™Ž)­‰nz}(©yX<•±ÅuŒ)JÔ§9—§ƒ}S´ÉåºîòH`ésþ¼yE1m­®Öy/Œn=Ò±(*$»Q{7à6ã“OÌÌa|Áõ.ЈÚsû¤¾¤äYÎ|s».¨ùÜrÕ÷b¯Ê•Q¡Ÿ¸‰â‰e…ÚÌ’@zq‡x‰…yÓJ0Ü%¸9 ;™oV>ç®q´ç¼È]•Ã-—ÅŠƒËžÈ~KŒÜKNÍÙPIAÞ¶ _2°ahx{s‚†6_a¦õR?(ù§`'IHˆÐ¨—MAs€;låôÁ”À†{knÂj‹»BOêÓ­t8éÁÛ§£{¾zJo…8šyplVí­5ß$\°*È‘ó˜?USÏ^ˆ"5Î#2 Ý,*JZm:Ñ™ZŽ~‰tÐߊ«è?•ÖúØö>Ç>,¿~½‚‹³óLærÕt¦˜•:ŸþÆ OÛ³!€´çµ4˜Ð3÷‡³„£vвdä9ßrðEφo­|Á¦r~‹w|9RÿáÙ,IìhiÙû«êôc5ݲŒõi\Ùj?$¢»_¶œ—Ý£ùöøÍʱ˜U‰ Áýw'ü73¡69²Ó€e|ÿ#EæòCsÆû”}¯+'»á5PÕéÜØòì-ÕÛ“³«2 K»ýp5Ñ–ð v;«›ÖŸümÔêsŒ–|2Jè+\R¬`¨Ÿ¶Ú€jÛ¥©§†s©_HOP‘Ù|«Æ‡ŽYæE—™a}$§†Î<²Ïsðz…{Æk‰Ì"×¢¦Ÿl<3ª\1Íó¼šU‚Ö&¡­]‰âõÄÆÎOT¦0°î×)˜4ç>!Ü "LŠ Ì”ÊYùrd•ö)ÿ½,Çeñ7ËÐJ­ð×?w5yÇóû;f&B1/–ž}-æ0Ô¹Ñxà‘‰_5V¼!Šü²j~OPãÔ@~lh¯¸ÃÒ%ÑêæI¥äs†,fÅìßÒó3·ê Ë>èꩦÄTôÌ+«Ü¸Ò[Hn¨:óŒðÌ‚Ö endstream endobj 152 0 obj << /Type /FontDescriptor /FontName /QBHYSA+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/one/parenleft/parenright/semicolon/zero) /FontFile 151 0 R >> endobj 153 0 obj << /Length1 1493 /Length2 6633 /Length3 0 /Length 7642 /Filter /FlateDecode >> stream xÚxTÚÚ6Ò")ÝÒ9tww£€€0 00Ì 34Hww ‚Ò !´€”H—„Ò! ßèñÜ{Ïýÿµ¾oÍZ3³Ÿ·ö³ßçݬLúF¼ vp[°*†äàJ”tŒÌ€ PˆÄce5† ¡à¿q¤’`³ûåhEÀQñ66¨-Êá÷Ömª Ã?ü 7ˆ+Á‡€@qäÿ•uÌ*0;%¸‹ †DàýÚŸ2Ä B»7ÿŸæ:Ãàž0ß¿Wö˜ý/vî®ü&0ÈSw°†ò„÷oÌŒˆ%DEEà§°È‘ÿWcoWðoãoÅÁß×î °GÑûCìÁ¨<_„€tsûûþ§áŸ+<€„Ø‚ 0¼gGÁ`û¿Ö¨þ»A¼@”üÀ_½³D)̃zÿÛýw‹ùÍ5 M¹ÿPþ—QQîðå”ðJˆ¢11€ÿ?óüëþfÿÕ·üÙÝdÔ€ÙÑ@ÞßD<þ(ƒãÏØpþYAŽÒ3Àñoù?ŠA¨'ÿóüùÿiÿW–ÿUþÿ½#Uw(ô·ã/‡ÿÇnãzÿñ@éÙ‰š 8jB`ÿíúü×@ë€í î.ÿmÕ@Ú fDæ€Ò9¯€0Pø/‚P…xíô!Hã_Zú»¨P ¬G@~Ý;¨( ð¿l¨Ñ9£îªe¿M`Ôdý³® ·û5‚‚"¢77o<”P+€¯jVíÀ^¿%àçƒÁ‘¨Š£?Àî†÷«±BB~TÜó¯v¡l¿aa ?†š)ª!ƒ‚@?µðSwT»þFQ  æŽø " à‡~_¦¨ Âñ·å»¹»¹¡û·¼P”þ^ÿ¾EÀ`/0on’ uzÚvQ£@ëÉûe{eµ#*Ѭ/RÉ6óÂ×Q'[íãSÅ'vUTÃú³eáÓ~´\£^»rƒ–ÓÕó‘h*}ëŠG¼ƒ?NºÀ vp‡1±Œ¨u’Ã$z¢'˜ì)§¯Âĺ…6_}uÝØ4—ظŠÏˆ™Òi‘q y»è1fL,«È!³ÇÄñ÷Ý,2Ý5LÆ®Ñc ñçŠm±Ô?¼×-Ⳇ––Þçö÷ejõT‹V…“R“ùh2å 'P“'åj¾E ±)Šœ)•¥%2S¹„³-c<[þ9YØßš4 äÞà «tOïÙ‡F÷:¸È´L^÷ò1õFŽñ~ª¦ä(Gžš EÖj;´³¡ÌÃj!h~&»®cáVg^Rzn¦>ß‹ó,DÈêë8ÕÅ“¸ì$AË4¨øÀç—Š¢úP×aû‡óWÒiRVR} ]&§U÷á<|ß_öA*1¦ÀN·šÞ· uGÁYèÉ?%>jÏÞíÝIz«!rö‰Ä”KEY«5ï´æLÈK˜ 0ÌÏÃÔoíP“2——˜gN0"*0&™1ÖÞDËÔ½k&¦8ILHê|PFÿz¾Rôà ™à&™4ñ@à³…ñ£;Ê{ ĵë÷ º`Y%Úºñ âZiåw˜;>)Ñe×D 'Bw¯¶Í>‰M/ç+Þ˜X“–>{œo÷3þûæiÃaþ#…ºHshTjçÇ€uë—èÛÍûãé2ßïõ´³§mz©7\ža‘Ú©I0‡A+DZ”–úÉ£L,úm´ÛN›pݺ?Ø7žþÀ°8—ÞrµÚjÐ=týYþTQpŸˆ}6LBh½J'MÂ?Ò4@™à»ø}›ñ†`Áiš£}#ÞTNkž¢pׂ%¦)š¤îRÍõ謇n‰´¥–Ž_sÀtŽQ¬*þÓC_hñ3ÿ#=Ñ÷Þ ðÎÚƒg»›–ŸÒ*ë4cG𠝔œ„7‘š%L}(|¤¯r¤}-¸n-Ø|¯ª¹ ®žæ‹'@éõ‘šlp‡½³ÍÚ¯NpqvùP—‘Þ0[%õ¸x?«9aýô'[êz&”Qõ–âq€6 º¡£ÀkÕ<ÛoÅUcþÜP¥ K@qÑB7n(‡ý<„Û~Œ*WgwŸ±Îÿ*GâZŽd>¨² 0¬üjYî¾Aü RªU®7ø1žIã^ì›É7?7©¾’ÜX#'Jeu~ø–9à.Ïñ Éõ¡„tòžÉ‹\_¿õïß üm„¾¨–B1Ò:NĽ0+iq^}D&«?T(í®Z“ªcš7]¿!Pj" 1…×e,x´äí%Ëu×¥2J¥æÖÊ /Êò„ö¯—Tt"7\d¨¿Åòõm³ämû'ÎÁöx•³[Evì•=_:iy›ñß ¢ñÓöX¿ÅûGÌuÒÆÎ.Râ» ‡U›_Ñôjн;‹ Ÿ4ŒÁÉæÐ+£s´Õ‹>µh-žž4æqf~ÇyšÇ=³2óNëi-ÊϹƒGj­ó!öú ¯|ð¾)ö5-ûiîXíøYÒ¨tð‘ÜRí#rLR޳°Þsg݇ fïöUg¥·ýëÒ·ëzpV ÷9æ§Æ¬gY‰è W×{.¡­h±£DÛP?—øäû½2ajƒ/-äU¡ÞÏIýRæ=ïr?K|•ÏЛªÒŠÿôf#Y§w†¿êŒžb9†q¥ÎŸÓWªßöÐ=ØdVõÇ/Þ#8ÀÉϳl(P L|›Êí·YñÑl:õm%Z­´*¬eOûÌgÙ-ßz_æ§³/{¹ðpø#@1G'UOZmø!µÿb]ãœÏGíóP¿æ(å; ý¼D?V*û®Ñ4æF—M' «•¸‘uïW©g›6R›»Ú‘ÐŽÚQõë 4ã /ĨN!}˜~Wý 0¸,¡Ø¾¸ÍÓy>‚ŠáB]€S ·4Â7Þs̤Ÿ½ ª¯è1•ƒtÞ¿ås²)Þßç×Y$kwëämœÔÓ`ê2ð5c•úiÉ¥!ꤓ¤ÃÙw-Ùw$|°Z¿G*-ð÷"Ÿµ$õe†c•Àé¹ö3$bûÝÆ¦4b àNÕ&ûZg¤^N-|wy°G.¸mj¦ß|ÃïÑÈ*k˜c7êóÂà§mÇíìÒãfí+Ø`Þ=þ ÂØ|± X.‘vß^ÁãÓ}-•ÊeYn0Tïú80Gr½„· Mb9EâÕý´?R4R?OèÓò4pXŠsÍå°"% 曎Ž|>Ú½[‹¯mõìôýíÕ… Ôf”À’•l¹øIbk™õkƒ·ºš ×wg}ñ C88ˆdto¾G! tG^s~ûJWh\°V|f' •_è\Qñ!Úi) ¢—ô¡,W{M¿Êø4êm®MÙCƒmVŸÐI*Í7ͳ^Ñv !5¼½/ù0'DõY¤L¯nȘ¥#àÛæW-¥¥l¦# sW[¢Ñù˜é<ÁøŽ4Ô„ú/ÀÓ.®ÇšðzùšÚµô [룅B%käÛŠ ¦§ÏXŽ$T [(dðb,å[Ï}=±ª­‰D„΋T&;넟+fmmäì)”VÜ_múfìh˜ˆqåÝò„pN4CÄ«Äñ×úöTŸîVžðLõ勞®·aѽp{Ub‘´Ôåñä3~[CíÏ8ÊÓίs5–Û|ÚТ1>ÿ‚æÑ{Rß–R.Žu¨Žd¥‚jj³“‚e¨ªºì™(ï0r ·—óaóozb÷FŠ·pÈÏÍÅdû€šÚå{tÙ4\2ò–^Ê>bzï’þ6$? ¥ |xËüÑTI}ëâ¹ø'gÊ©$,iï‹\]w›Tœ¢Òö>¾âþCt¤n}Qo$Tð q“³‚†e'¥ÎQ@£Qá×[³ô9XMp6vVÂ[)—U<ý@ûHhêŠíí ŸôÊhß«Œó† šýšû·;såÜ© ªÀFé‚gÞì.{é,“í W&ëôƒ¶ùïìź9jõ‡«X™²„è}¥#ïœ,ëïÃN„ÉjøÈÍ)ïÊè^„Ýi·…ÐÇù>¥´5ãežçÚŠéq.õº€¯E_ -O¢Ô×» ”=ך/,_â·Ó&’ïÚ½ðUI&*ê‰ñ š£^Í"ác!7]Ýt¤Aæžù8³8;³X ïV“cì› .«‰[•0¦ËŸüz~–áÊ5¨~‘bð5+ Óvµ”œßõäNÃÉ :ã”§ªâ˜Ö7S $ɰ]RƒO8·….Zñ¾l>RÒSyØA.Çë±cõ»¹/¿±sQ\?}h|vŒ”¦Šu>æÇ•+t¡æÌ³?m®ÞO[]+¼ð~À±^ý„uÞ·œ–¸„©Í€!—ȆXñj>µßT¬à®œr‹9TY9´@NÌ;èiÑżd¹dy#CFó9x&WnÇN^ÄIl)TN“N¸3E7…ÆÊ¥ã¯N×C“µ]”'¥d÷r5kÐ͵ºF€Ìé:_•„†¸3† G¦í«±Ÿ¯)=NoqåĬâ<þɵfÌ&)ä¬âh¾Ì)j¸§£Ó‚×PJð´ßƒ­±zè\iå‹.Z£¶9¢ÎÓXßÝ7‰ÑÖ2Vs¡ŸÃÍû¼ª¬À$&£Võ»É…>s«Öhx8wEÛG³°©°‚#í bGK›­]/åàŒ#ÑuŸÃLƒ71úäõ÷©'Sä›_ÆqÕ%ë_¿X5S§·HiXkUðmò 0´2›È¦&äÁ{Á¢Zú)]]yîØ”FÎõÇ{"£™L™UU¦5¸{ _1ûlî¬@MkVß±î™n½E¦ÆõNd î‘1…ÏòGû»›w¥'ÓSc¯‚î<F}*«Lk²·]¯.±<ñ7¼‹MŒ3ð¸4dMWÐÙ¾Ðæî®)&ô<€x§þØcu§ó4@ká¡o®Îä²ÜLB\™Î±åɇÔRd$Œ‰aÄ™XÖ¤°B“ïF•<"MàÓ_Õ’Ëg2~ à»÷A±‡qfb¹ñ:$YfâQôͽëd‘úwÑn:HYo™êp®õ [%ŒŽgVíÈ>Z·Ú({ãE_qÚÕÉǧç³ùƒtÚAè»!ú)lbH$ú™»š”–ƒðüÉDƒÔ[wNæí:©ËE£´ônöᣫïJÏêä:5CZù“U™EMøj€²›Øè§Ü‹Z“¡@jæ¡ÉŠ ™Ý oˆíåý/¸¦åêL9ÓÏ+Q½%>ÿ˜¼@äîp€ÉQ¡éÚpYw%IÑÛʪwWôbÑu&ˆàì×ÓeýÞ›_†zSpŒóc•D®ÔÜÖL°{‰E>°$DdIJùÕ$4tFã>K[R+ߘÏËØ!(¨ÆÖñø²ÂáD‰, Y_|+„}Ûϱµì3@ÿð*Lo6‰&Ò…/öe|ÕÑ]C‹ÊñL¸®´Óƒ¯¸7ìÁ›½mU¤³„ï¤ü‰ŽÂë<MsH΂jåÎÂÂ×Iͳ„´""0Ëú¶^u—m‘ÊÇæW¦¦ØmΘYûµüžhù.|¿~쮞›EèÂxÝ+.Áw=]ÒºiKã¡âPéä–Ã)ŽdƒJœ>jÖ—‡9œÞÕPÝ””ëF¦Î&b™è‚ú •\,Š>Wè‰É9?¼t.*,kG~TÍò怾˜"—ˆ¼[Ú¤ä͌6ÜÎi1C(D!‹Ÿõ€–æ[ú’ýQš¹·ŠXí‘î6n–]¬¸oŸåq¬Kósþê{ý߈óZgâÉ‹ìh<¨cÌÐñÍÒÏn=ᮤ§ÛYLY*½úÏÅTåàTc½NúDkÇUÌ9])ŠÑÏcGûæôâO¥­ ôÊR„~tr)Þª q‡µb°¼]Ñè'ܹͷÖÃß<ù„ǘÉònåI°?¯¹oÑ:p`žÜl×ÀàÎÇš†; Ä{Ž–ÑŶGw¯·[w*¬w¥'.FÒû¹1‰÷tÝå5c¹µ‘Ôš7ezò¯Xu’4‚—· Å ³Ø!áSišùm×'zŽuLUɇ€"·£ÜÎ_JÈÚ§û—½_ô6ž !#+n[ œÊË{ü˜K꿞$Nãå ZOjG;æ£Ç›Ý'¯ãœ…‹ˆ9“¯Ü¤<` ÀQO~ŒÆSé5êwЍFxí©çI«j‘¬oDÌbì;–Ük6§ CsáYžp4â}h~?cT“ŠýmÍùL鯢§î—çu6ÐrÉ(– =‘ýPÏëñ(`4“Ö²PÔø0³Õ—‰Oo$€µö*ýR®Ì27Cö ¾g5Äþ”ÒðnÐzô]½ßvŸÜÂ-qMÎômZ)Õ¥TJ׺VÂà÷¬yîB†S±8dô‰åµÁO† %íÓ“¤& ¯LÏ»Ÿ;ŠéÃT›–s°Øž1‡n¨¤ÈŽ„û5J=Á$êÃ÷øÉzÁ8_ml±šÆ¶< £G3«e˜Ð8Å­T`‡zG!=¼¦$Ýí ’(Üͽ­ÉÞÛ1§”¿´åÞìÒ ·œ…V{}fõ<^}ôF‰¸SY‹A‡º+Â>‡ÃAÞÓÒ¾Ø4k¸Ûb'àÙn¬€êm+"!Ê»kz9«éïxàãù\~1nZ¾Î-‘Ù>8˹c‘ãã=–<Å‚ ¨\pžɰ3TØ{ÂÑ&ï›(ƒåýP¶ó(rCö2©zÓ[þkßm³Éê%îï¿zÔ¥@¾éGÒÞŠ¶beP&6¯„–#,ù²ˆO‡i¸¼v˜ íî;ѺGп2gœ³» ù‘3É.ýÖÇQö´½´šzJÅ9•ÅëûV/¡¹FoúRذ´»žÈëýÖ..c¤Aî©§¢ÃxÒZÛµÑuk^óL94ÇöèÌGª¹6SÃ{ê±Ê V×ðN”«[©³þ[ylùn0Û»Bi™Ðž¡(¤Ñù×^Vú }Î-~WËå¹wåÌT‚ îx™*àd‘ îÞ¹D‰ù1; .YmsÉ÷ÌÏ›¶‘ohNjŒ ÷óúΉÔ4>8ÔhÄËkÏXùHµ|ë’D‡n^­ŸS¨ìàð¬ƒM?’ÈèPÂI9Éxƒ4A¦Ÿ_I~¤1™œh’ˆœyŸÂLÔið°åEˆÞõ÷‰$a.=ŸŸ-| ¦'Pj…ÛÕÊâììiÂÞŽÌ®ØýT3ª¬ú³‹÷G;< "°!Õè4þ\¹«ÑY«–6¾z]þ¥©VtB$åâµz˜šÐËLo+›aDÍö#·w;øÅ­ãs…f ƒz`VÐ)ºnÕãN9+aöƒÜ-Û²ÁÖ1Æwì’ZŸxº‚ÆŒIÓ˜¬oñÏÔ3ik.¢›ã–¢Ã“Nnè}Óû(æýoÔ;]Ì‘ðÔωË;bC+Iùyo;æl!²‹¹æz¯J^„F§ØµØ©Nu‚@x,Ì™÷üŽH4 áR‰jn gÔ-ÝöT<ÅXÄ{°>®  „7øÉ«Ð:ï–DÖ^DÉgmË«»æœ—ñ‡ú÷L7ƒ`v_“p2öPH\P(õVÈ7룽T‘ç·³»^NW> endobj 155 0 obj << /Length1 1422 /Length2 6374 /Length3 0 /Length 7348 /Filter /FlateDecode >> stream xÚtTTmÛ.]‚ÒƒÂ0´HHw#-0  1ÃÐÝ€tŠ”( RJ7ÒÝH« À?êû~ïÿ~ç¬uÎÚkí½Ÿë®çzîë~ØYµõxe¬VPEÅ âãÈiè‹øùùøùˆØÙõa(Gè_0»é CÀÅÿ—ƒ F¡1y0 í§€TÝ AHD$*ÎÏàçø·#)»Ã¬|UêJÄ.‡pöBÂlíPè2ÿ8!\ÐÇ¢~‡dœ H h€QvP'tEØ ‡€À (¯¥à”°C¡œÅ@>°“+i+ÅõàCÙt¡®P¤;Ôð‹0@ìýÃŒˆ osýƒë!lP`$€a(Üá·†"èâ=u€–3þÇYýÃÀ_gñþ“î¯è_‰`ðßÁ`áä †{Á඘# ¥¨Î‡òD=€áÖ¿ÁŽ®t<Ø s[¡~ï P”Ñ€Ñÿ¢ç AœQ®|®0Ç_¿Ò OYn-‡pr‚ÂQ®D¿ö'CB!èc÷þé¬á÷ùkaƒ[Ûü"aíæ |‡¹¸AUäÿrACDÿ`¶P@˜ÿ¡ˆˆê€zB쀿Òë{9CA¿`4?g„3ÀMê³¢?D>®`w(…tƒúùüoÿWD ÀA¬ ¶08Ñ?ÙÑ0ÔæÏÝ|$Ì`ÊÖÀÿëùÏßS´¼¬pG¯Ü÷¨¢£® ¦Íó‡ñl²²O€/HÀûP  _¢b¿§ùÏüMþ7ª †ýµ9þ2ªÀm€‡8 ïoîÉ‚ó¯‘áü»‚&­e(€óé›ñ óCÐ/Ðÿ÷üù¿éþW–ÿ—ôÿ{CŠnŽŽ¿Íœ¿íÿ‡ìsôúË-e7z,4èá€ÿ·«!ôÏ(k@­anNÿmUAÑã!·EKœ$ÄÇ/ô‡¹*Â<¡ÖÚ0Äîþnº†# ÕF¸Â~Ý8è(~þÿ²¡§‾U\Ñ ûm‚¢‡êßuà„õ¯é€‘H°?ZdÂÂzL­¡ž¿õ òÁ(tÍÑ`ƒ@ýj« vEs†¹: »`÷ËøàÑ…¡.nèÆüBÿUâ†D¢‡ò·:Ð{ú{ýû€B=¡¢ÏSÈ£PûšÐ–ïU2 ¼ÃxK+mQ ÆÝ‘Â(Žé>vêø™Jã.²–Öå´ƒiÚ3¥ácS¾ Ü_¾xš}ÈZLUÎCa(t¯Éñæ÷_œ~°…2[#èmGEÓ¢ÖÈu˜îXšâ˜Þn8õ3L´Cp«b³ðªÍuËäáÀ Pü•”5•'“Û¶‡cÌʶ‚06#'Gš¦»é«<½ÂŠÅ“m‰¥»ðZ3ÏXXèËîéNd$87KÂFž²èTË™ÌêÍ©×ûêˆ,@c ?ÙnšÉÑ=#'§•»ëxl´9©t¼·…4%ë KÆgÜ3Ì­¼¡°eº¢aq©¬q9Ø)RÊ¡X±šuÎ]2±ªWøHÖƒ{8hŽ˜Z¯ÛxA«T¥·Ì?ovmœ?ýx¢È@ÿR=ÛÄ$Ày')Ê·y kÕ¦‹Uí³¸ñ:P³æ–e_ÝCœò`¾Gã C¤Ò˜ý½ú\ éD†ÐLs‰½w«·Ëòój¼”¢úÁàÙjïÝJÏ•kBL«tÞšÊÏÈižS’ƒK¤É下bH0~N½‚aÕM¥<¸¾¿[å–~’u#]¹WàEN‘ØX»4qdsÞËqížGŽF²èÉ] 8©&z†«dM™{âñ¢—U[Ö ŠÎïp)÷×~?w§0¤K¿Çîx4MÈŸPl߈Uó²œU,*ƒ¬£á>§7RŠåOñKVtü*·®¼Eªj…ûæ¨bÐÈ]rª9gÅ!hèb&;î]]Á«k»~i÷`©<,cÔ+±‡ýŽÕË‹&•ï­ D2¨cá‚u§{Zâªöx I9]s“#õ³glBÊ-Úî¼G¦œtÛ`ÑÞEahªúdÜ\u³€…§í$Üü•ú'ßE BQYÈ~k(©ë¾ÓE<ídp÷' &C¢üZ ÐJM»½tÎf€YFk>ÑRÆ“GÀ{bБøÃ¼È‹ùPO?ÍùêÞܽ١Tˆ»é–yøqßÉ]¬‹kŽ;¤‘mñ^ÆMÖÔá£L0›àeëy‰o¹PšçòùR+ñÍ Æk¢’P²RÎâ¸ôdðcÈîÐ\ÄÙn&Ó„ÎÆÛûD:Zm€áågºK¶Hã‹zU0Úåd`ˆ´=ù0'vs˜àä ß)Èæ¸ýÛ»P’œ$UJY² ûBoƒ¤¯HýäO€Ä¡}&’i©)ūï×D$ >¾¦ï@ÐÎK@ßj Ö˜Îêbñ[³×‘áxÏ Òk•……ºf…¶y(q>USŸÓœaê€">¦Œ³üïŽe™º·Ô ÅUGZBœy$½;OÞ9á&4}â¾-Jà¢ìz÷èÖN.g"/e¸˜ 8¦ˆ„0´+£Ã˜%÷âªU¼†§vO~®ˆ¿‡;Þý°)‡ã}½ì˜x¿1‘~AS´aü}n龂°¨ú²ýj¦Õø sß5˜´ÍØ{2k9hbýû’ñý#ùBj–D:éÉÕÛõÂoå=Ãñܸr: êD¬ã{6ÙˆÂg—•‚GY†)‡‘¡ МiØçh§œ¥÷|‹¦~ßÖD¯Hf›ãäñ¸ÕEý ]óòš¥Õ/ªúãb+ý²i¾ÎÈ߉ê?Èûº…‹úÆp³ÔûzoŽ l§«}hŸuÊ衈Â}‰æ”Ý'®/ñªm¤ÃzÔB¬‚#uûe+Ãïß,çÖŽà Ö†…§4íª¦n¸<‰RnÒº`š¥ú©UGŽ06—[=M¬r5¬v¢€î1\UjÕëÙ9[|ß¾*ÅjKý• ‡ÃŠg©é9ãYÍÓq=Gå÷¢Y†˜súw<­º-CŒƒ¾éÏ…#|r''¨{ÖÁßÜÁªÃØrúÈ«`ø¸’ø÷å’W~àh+¸ÞÉá º·™©1d$R®0ÄJ–¨|&L°Ã! áùÎÇè :ä|—D_©ûXÛä•Þ’Jì˜Lä-œfeªC®¯cë5™´ó·wê¦W4'1fM¼*ß~œ¾}„1¨¿@MQóÐá˜óÜ|oõ˜F:ÝË–pBƒÍNuy›iŠUàzAè †^ÆÆÝ2 O=“b¥Ak†‚³à<‘Ó-·“­éùâø‘Y´²³v:ÞÛ°‰žo`Ðíc¿µRŠê „ !.=ÏÜØž.âÄÞ/‡N= &v“#~ÈŒ4뤪j—92 ¢YIP<`˜µfÙΪ-Sd}¼tHïç÷*ÚEúÎ0CJ™—Ua8Øp™véíÚ¦öµGXÉ¥MsÝÌØHÍ íU{9Z9t•>Ì”j„âõ3`ÛlP÷*œµ-þ4aTpÝ:A%´05ŠTÚ4™=\;õ7<˜$¦Ç´–è$­ˆ/€Ñ Ì,«ŠnüCÚ¥yщCÅñ_nçg€9óð f[4 „¥äÏÔ¿[Ó ºejІáˆû³÷>óº–$Ð~Z2ZäbÃÜVù”¶Ê{ Ac’äàŸÃ®@’(þcX†”ü1Vñú(;+Ç~ÞXejXŠ XÞ”ãåŽuÈO Ý,Þü¥Šp,&ªlŒûÆ_•<ÆöܾtâYpÇÝX­3×áVx”KÊ»H]™=’-øÜ=DZ—@Â=ªíº%þm—ðCjCÛ/‡„­F–%8=iÑFL-5…\>i.VnÖˆ¬|[ªÀqê•î0.ã)‘¬{F$ëO} œ¾âãZ•yÌ$ï$l—½épIüHÿæñÙí¦,"\R¡…ÅÙ€Exž]®Ú0³½óãícRýRxdÝ)· ±€P³:G’_a†°ÁàŠ Ôßµ¡„‹ÉLuf¾g2™ú`½®Ä+Áí[KÒE“rŠòE·›!Ž˜¾‘šÞ„¹êSûÌ1ö̼µSíR[Z–wïwGÞu¾Ì»G±¶ÞäåÍúb"“Xª“qáðÕA*3lÆûåð y< ^@?­™œHºUJMÓ Ë¡ãÐ{Zð7LÜçwt•$íËi·˜XkžFŸ§Wšv]:ž¤Ý¬æ4¶w¹4ö‘GVÆ(½ ÎPj&Å•i:ß›Vy£é´èoX³üÒ,„©{—I&¿âzó) u™ÙOøB³”Œ•ã¤o›Šý£L¾ù*ð4æxðk–Ï’m3îäNù\L¶pP¤ôàÔ6½ªq\¦[šà°~]ì Ž²è£÷˘K ·¸CïWÚjj)C€Êº€žpn˜H+ãM.‰CK`âº<ˉÈÌËLIj¤HÕß±;Ín¡ŽuG(“›”çšI'BFçË.?ƒ¶ï\j|3Ãëú§*ÔF+Áso¶>Å/fs&w[àè.E³cÕi ÷¯FˆìOXì·Û¤`—E Méœõ”Aó{<#Cc„®•o» ß+ߦÖv{AC&#y&&ó Âqü& ËÁg¢Ë®xÅT˜> á©ÈÙ$b‹DgŒžº<ïÂF»ÝcÆW]šÄå¹Ï)VÆH:&åÒ˜ËTo«wno«Û2.kk]iìš Ô¶äЮÚ2òH$ùuI¯Ùû&£(qXp¦™M$],ZÑßVÚ‚Jìë¥ÏÙ¹›ÝŒ'û¸Pâªo`;>³ìÂfÄÞŠ ìƒ}㻎˜i÷eôMý°(q}´íž¾Ï¸p¬³ÿ3”åœÇd$ʼn€ÖAº¢mÚ$#º·o5Η–~à%*hb–’Å¿ª½±«=‡ûüMpŒ2¦Õ£ûÂLŸŽÎ+ßMPL wÈh\ßë\úˆUùQSÏ®c%½ &© àÓìÂ+e4)x´Q;[¸ö}ôµÆ'æÀ–FW¤5*c5EZΙEÂïóÁ¯HÂe÷Ø{f¾úQðtׄu‘jföÀý*àE.NUZŒ¶¾Õ8ÖGjö¹ân<.&B׺´U—&åäòsì˜üG!öÉîc´ý>_¾í/ùÌmóF8ð(k‘âV-]ßDŸh ‰+ PzÅJU¸”qâ.,=¤ÂÃgâthÕ=½Œ6o@èŽ=X2úW3Ê,לlc°;jÔ‡,ûÄ×;‡Ì`Ð2†–} Ø~ õhˆe”¾`jù–õYË·±æÑzP[êûÞðe/Źý]ÐÑk’#Mà4É]±‘}ðž£Çsz}úÕ†s*¦œŒóÞÓ'(–ò‹÷éXۘఙǎOšÌô9t†JñRK2à¿Pé§PÝäÕ•áÖ~ÜÀêÖuTÀ50×9åŽóú¼àíÕ›÷‹°ôö„ŠG†?‹eÙ׵ߩˆ »ÅÕ%ݨ¿Ènª,ó·¥TýÜ‘céÉ^÷žímß~3%:z;ÛÉᆭKojÙTÇûóÛl7WO³¸ïå³R8cP*Š †Á5ÞkTâïFä«—›eÜG³ÜùìTìô±WR07TŽy-ÄYhë— 1þy’Ú€·¦’ûû…µ åHµént§ù(Iät$éz ð³ MW"ÉÜ}ý¤jX•„ñ+8Ð|–Ár¸žQ>´t§¯ãPÅ3ârfíyRT[,§Öáv¼L¸‘ªþ–qÖë»|Œ!LÆ>_]áö­•cr³÷íô«·KÂÞ©#öTÊZä5²Ï€tf¼D˜‹ç0EïµËæ`0äì;Øô© J]+$6î%ˆ‘`7£BRœ”釯! 5Ù5 y%×ó“ïÛbf. jo‹¬0Ú¤^µÖŠÿôJR'ƒíBUÞÒ/(Þx:뜧<¨‡ih¾:Þº†¼È¤<d°5¬]£PK’`¾&=n}y³û¤«\ É´½ÁÍûýð%Š(ÒÃþ,È[WZT[‡<¤eCè*ã­.n ¡Æ…HpaëgæIoŒ0ÚÚIsNé«o5éTaá‡?+ãÇoõ6Ï—ÖmpsN¦I_Uí“ùo&žßÏ£ô]nHѾKqèOÜ £Jz{¢NÅýù"-~ƒHì¼±ê‘ìÓÛƒ¤ü)ñämÇ8£h‡Ì‹,¡Gî®ÄÌÔl¨cI…ñœ—ç ßACç&ùzÌ~ò`T+ ;¼gåÆv'Ý>Í% |÷¹B³°P»Û¤FÍá§¹öW.paEöKëåÅgªEH6"›h:ÓAFGùǺ°G À o:¾ŽŽËËs¹Ë¼ë ‡t¤MV~êR…Á¸~±‡Až!æ˜-µAVUàøþì± þ}-eñ˜óX{ÕüÑÄí¢c‰©áÁ{”ò’S%,$˜ß:T_ûv„v‘6ôO‡Êvù^G¼ûÑY”Lúœ¤ó¬¸2 º«:¿ÕÄC“³G'ì˜`’¯1~Z4[+*#ôAÿ¹v¶sUv®ÆHéÝ÷U©]ø6û‘øs<Í)‰m± ”ö^îf‘7Áƒ^·‡=¦mîá”ÔŒ„s­ûéÃÓ+rÙÔÖzµ0ÛŒŸc犱„ð;=ýÈ Ãæ¢Fè Ä"J²šŒ_f¹™úÞæÇ±7k›—(Oú‘)aZƒ‹‘±ÿ˜Äã„«K¬3è‘Ý>OH8ÛÎÉu Ìe+¹ýÁgéûØÛ¼¦§Ò þúõeœç×ÇŸ"Éï×E[ê_¶¯® ¤½*¸ÛIØ'ó˜,¨Ç¦·µÂ»ö¤°¥“h/ ƺþþ²@;ÄÀ+Í„’’l×uŠ"ÉLwå}´ßÕáÅgN &TÂ;ºÝbéÔH˜Åö<˜®¶Ò7:´üÖ'»zš°9ŸÙ¦ŠsÕÛ§5¼3µû–Ï=« ä–q`eX:Þbä9œÞîª^¥·×âÕ.»ÀÅ4¨VTj[ËiMiß¿.I‰æP›ûÑUaÖë°ÎWupN²ÂäXhT«YóE^?3}J:œþå‡GDY ë,øeŸQª<û¼õØ™©Æ¨Ôd!Tÿ¹[wü–OÁsV–ì‚ä Ô|4!~KDÓ¼²<¡C‹7kÉ?z`³WAÌþŸBš$4Dß~ïsLî8˜•t]ÇpòËõkíâò‚ã=Ó-&ÔL]¶î/Ø­/Búª¤äŽã×ì‰Ûˆú&Ö<žJ¢ —X²Î-RÛ¬áîc;Úu„׌ Äy̲eÙ.Û'Qß¾|¥/=f ¤ÌBÊ®w¥˜ÌIxÕÕ5ïD¤K¥\Û ¾¨lq?ø ´«­Õ¶ßlóY·¤¹Õ./wæaAaGsÖ¬r„*U²e^÷sÁŽ:îýh|rë)¶e9‚öˆsÈA°U <騋›Î¼ÿ}L2*X §•h øEðq½ÌGßZ,%žE}Ù˜ÝÍ£éA­©c™5ÙFÇŽ1ŸMWÇ»#Ç o¢€µ‡qz‹Dèú/ùM‚N¸f™¢NVà':Ì,¢ä/©Gqܰ¶.ɬyE±}îš|ÊevIr'¬ð÷¦%ÃùO±D”ƒC€;}xo¡ÉåÀ”Hâs…5$–™JAjšJE Yl¡VŸÄffê4BýÖ7¹ç÷Æì2‰^ F“™œÔÕÕÍÃïz!w>lÛŽ@?¯t”YM°õs• ä–é´ŠÍå0ëÒÇlžöÉ[ËgÕ)vï*{—R~Y&.XÃ8·‰¬f×VÕ!+› a|þŒµ×d3ê‚„žŸ¾Û{^Æ‚þJèqŠq§ŸžÈLæ^¾õäVµÙº¹T!D¼³%Qïl’ L‹sçúç`'ö 9[<ÊïX‚™H÷Y<‚çÕ`,e×zÔKšt\ªÛ»VFùŽÅ¾”Ðï5i‚vű½_A`qmÉäw£Fºeƒ¶bjJ”_økYƒ¦gÞņ!?¹…žzv0íw½ÞÂüPK¸÷YÐé¸Uäå7rÛ>¿i8}ËÍýµ”£º¡G£¶¦KΆÒbœ›Àáuò—çG Þ]Ä*µvE…’û*ïUko1-œêþÌ~Ÿ½i„KÃSÏ@‘SÁG¦Ø5¥ :ó.#Â#—ïÇ¢Yµ­‰9ñ©2>ñ ¾!—ºü"r¸°E|«G-íHÇîܑϋHœ÷u:׫Ô¦]-JÔžV–‹1—|òèöÐÞ٫ٺ'Â)Õ”¶–¼µö½#¾ÓáÅZõJIûr©Y?óCç’ý^ÄBÈÜAÊwÑ,@9[üˆg©ËYYÖ‡ˆEä\‹k¹„ÇÖÇ}Fv¡ÁLÖE]Gð~_þ˜êÙ+Ða¼ í['§| å®¥n†’#–ðé0oWiM<ØÙ"=tí¾M õŸ2?ÁâËŸ#_¤i3Ö]©¼d€Ñkùï!õêÝ™63 ýzãÆ¡1£Ñy£«Ÿ”ް I¯*°Õñc+ž¢6¥(s‹ì“™5öŽEhØ=Kö߳ߴÙX;ÉdˆèëÑDÇo¤JO·z.߾͞Û.b‚ ~àe:öÎL&êÁcÞ½·W-I›~Â4¶¢í/_=Ö3ÀP°c·it+ºyûã°"ð¢ô¥« e3×Y|‹mûLÒšùŽÅÖiÓ<s:OXkXšSfáRÅ"ýŠxe ÿö*ÆmZÁ–(ªM—:Á…ùûZ*Cf‚ýR¸!rLž©òw¿ŽæÝå¯íF:$N†ÉÂuý™÷7ªß %”…¯E¤Kòó#‰éT J¤ºá]Í,S`5@[ÐÝ—¯È–¦b7w+_¬ÍØžEÍ<4ÍkoŒ³/šk¤…¯†;}¶ =%±ÕRÊ’#%þ$Â×äåœlÇ 5"4.P%©\EŠߨT¼³^RNË>áµf¿M¹j¥q˜±B7¬öd>GÞ@’ZÁ•-_b3€ÃZ‰Ä``U½øþ©%{á²{¸ uÐúôÆf8̹†nïL¡øðñFg»] ÃHiñã±b–ø³ ÛÅ éÌ4Oª·øáÍ´ŸX m†ÈÞ, K e¹È~J'½Òª©-ÿÅu?øíÓ€Æëá¡ÍeCE÷»ýgy×)²ëquºÖýkveáX€œO»ZªÙϹèGËŠ*ÛúLøªU5ùÛ£* å“êÊâZy1×çeò)(OU-)C3'NÝÈÒØªÍ kmZ»’F†SÿZKC{ endstream endobj 156 0 obj << /Type /FontDescriptor /FontName /IQLEKP+CMSY7 /Flags 4 /FontBBox [-15 -951 1251 782] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 49 /XHeight 431 /CharSet (/asteriskmath/lessequal) /FontFile 155 0 R >> endobj 157 0 obj << /Length1 1696 /Length2 10946 /Length3 0 /Length 12025 /Filter /FlateDecode >> stream xÚ·P›ë.Œ§¸k°E‚—âîÖâ H×âîNiq§‡â^ÜÝ +Rܹl9çìsþæÞÉLò=ËŸw­õÎ5M6 3ŒÆÆÉÎ!TÖ’çäppp³spp¡30hav ÉÑt@ÎP°Dð’Î SسLÊöl¨ì(¸Ø8¹œ|‚œo98\ÿ2tpH™º‚-ÊìŠÎ éàèá ¶²†=çù×#€ÉœÀ) ð†õOw€„=Èln (›Â¬AöÏÍMíšæ`Ìã¿B0 [Ã`Ž‚@ ››»©=”ÝÁÙJ”™à†Y4@P³+Èðe€Š©=èojìè -k0ô/…¦ƒ%ÌÍÔxØÍA賋 Ää xÎДW¨:‚ +ýeÀ øûpœìœÿ÷·÷À?MÍÍìM!`ˆÀl¨Ê(±ÃÜa¬SˆÅ†¦vP‡gSWS°©Ù³ÁŸ¥›d$Ô¦Ï ÿæ5w; ìP°Ý„y>fiˆ…¤ƒ½=ƒ¢ÿQŸØdþ|îÀ¿›k qpƒxý Y‚!–аpqjCÀN. y©¿mžEèÿ‘Y`^~.~È r7·þ‘@ËÃô§’óñ3/GG€å3 Øôüƒî5u`Î. ¯*þ¡sr,Àæ0€È AÿOôg1Èò/üÜg°;à=Çóøq8þøüûÉðyÂ, vÿ1ÿ³Å@9 I9i–¿)ÿ[ùö­ƒ;À‹›ÀÆÅËàääâ¼y~ðùï8ÿ>±ÿSªf þ»ºD”‡X:þ"ñ|zÿ"âú÷d0ý½6Ì€ÿΠâð<Ï ÓÆß€ƒ—Ãüù‹óÿy þtùÿ›ý?¢ü_Çÿ+’q±³ûSÏô—ÁÿGoj¶óøÛâyž]`Ï»¡ìð¼!ÿ5ÕýµÐÊ °‹ýÿjåa¦Ï;"±zžs6Nvž¿ä`¨ Ød¡†™[ÿ5KÿjÆs;0¤æÿqï<{qpüîyõÌmŸïèsËþTž7ë¿óJCÌ,þXA.^>€©³³©úó<#^€çó®Z€Üÿqâ{väLÓ*Rç‘ã-Þ~.M˜Ê§¨Ô¼nÛ=QÏ^:|H Ô#ËC†ŒC€GÊñމ~toFÏ•ìµn4ú@íÅãä šµÞ£¶Y{¯)õ½þI^/_åtϹ>Œv­ ìä ›Âd1âž;Ê d#ô%ëŠ/…wò0"JÐÊ&Á«¶V µúkl‹“_*ÂåÍñ]IKœuwZsá¨ÄÅ5_늎c“ÅP©Šíu–Ч£ß‹L¡xàïK½ºìâÊpßà0ÝÞY|swéD¦ÿ¤e¡gâÁòfOÒ¸:Š1™—}_Fè>6O¡,}Ußÿ†Ä‰õµÈ\ý9CâÃglTÏ„<n †ñ8Ûì˜Iêõ³I  ?B™á'«³ž\C—ý…Âwá k‰1]¦@^y¢µW“«yòhù–º_[`e†©7†HmûÜ\Ôf§{%Ÿt°ê&4}ýU/J¾Ô½Ü}•dò°g½®•• ­ÙdØÁŽïI¥°>j~ª‹Ñyóï‹5]v2y:ŠÖØXD ó…’$<•UÖô7 îÜ%es“®Ó”øü‡é²%•@ë ˆ³á¬{|˜ØÆ'~¼—þæóVÆËþöK‹’q Ã/õ%yúÅ¢'¯‹¢xF3†¬M­ÈÕ­2òwLQ`*ë)ßh›è à¥^ÿ¦ï/mùˆ÷ÃÃà P,  —/ãåA—¥.ávü=QÎ&cMý–‘WÏÇÇ+#¦éãLÇÕpŒ†ÃrfxØÖ;èJTRµv0ßÓù¨#bˆ©EÌï;<ºµMX:½æ ¥EòçÝ¢<#jo²™ßT|d¹žorÈ¢‰Ü)-böH`7ÆGØÿü4md¥b§#ø{?Hcù–)Ÿà–z†ž=î&_r‰x@tòÚÂy ·C?dz‹;iDŠÌˆ–šén¥Ô#¡D¬´réZ• "}á¡â¥@Q¤¸¾ž¾w2/È-Ò)Hk8ðYíïQ[’g|Ô¶^þÖxGEÑì¡q<Êó7·GÙ9É1æ'ˆF pÝ^`¥´£V/J¨ŽÙ%q•*óéo‘1ß7“e@;¨r1Ñà¾ÝIÁNïöL·)ýGýR}pZ€Ê †–‡ÍÈz v–§yl·2£Þó€n}†ç)Iˆz_2öW ×#?@“p/c»¹dh”:e¡óõ½én¾‚gM£\G¬KÖ¡Õ¸fÃ’CPß…ÑŠ™—`°tOÐàk€5ögý$gÙVšä&€@ÕŒs·5¼yÖn­i•gé—5Ò±4zñ0C~ŒÜòÏ;õÒçj絿u­k{àæGZ= epåçsÚ=—jv*zÓJJuóyûócokÁÞŽ/ܯÏ\ƒyÔ]úÓÜZÏY ’Ó¾®í—³0¸O5‚©¦5¬/=¨ PHõIøZ•ÐíêΓ±±J"_b‹¡üÀ–ø.ÇBÈ ²ÓoÌø-Ñ¿½ ûpøÏÞû04<²|L¬É3Ö!z. í!6Á™¥_Ñg¾jf †A³þ›’ì@>”jûcÂÌ:'Ô!iu6Iê—#Ðņé'©§uå"ÜÄ„·¢4úK£1ÍÊaêÕÈ£É@‹š/ð¨+],ow–+otI_Êÿ>X¦¿pÌ’¸(·~ÝTèfÕEúÚ½‹K~ziŒÿWÛ¶X—kVöÕv4̶ÅZG˜¾ÄX˪Hé`¾ÞåáŹ¹§€Ø±Qí”Ç}“PyWÕòáþù0Ûüµï› â^Cœ‹#ûÎE¯Ô“ç8Þ·eNH(~²œÑû)VïnH nºNÿyѽ=þ»Ì0ØøÀ4Ñ¡%O˜>ÂNðÒ”Sé{­[ý0Ø4$xU®¢ìqïŠN2šÍÄJ„Œå‹å 3ºVbµ´lˆß](1(SÞEš ÐAL†Ü cŠ7½*P[‹Eã>.«ïÜ@?›ópüùLi«£÷š •?ÑZðÀ†ŒöQjÛœ"”ß–‚ýúÈî°² ÛÛÎ7³¤Xv½œ!ĘÎÄ¥f(Œ¶¹šªš¾F·3葼Ÿ÷_ÌKÞˆúÈèÍ6 ð&œå…mI%Qƒpï¤ÃÇFª¹S± s)’¯{8ry“!ÔßJü]+Dc_ÓX^"Ãß;Ö³p! û®µ;0 )5eL„­+pÕíOz²sÞ nÂ1‚r£9vvà Š÷ñ”[a‚±¬:ÖúúveXìö“¸g‰ªK^À`zŠä.ˆJÔÙH³8CŽo§yà·_Ä «ÙãdõX‡ƒì¶ÛßW˜ 6®Þ¨h”85£2p(’­÷³o¶Äü2ãïÑÊM—ôŠ`U#t9ÀxÁ@‹Þ֗ɉ7Ò3ÿ ÷Ü(@QW~ɳ©øó–åÍF5t‡®±$]ì&ûŠkЬ€TJX/ûb\nôªÑuz*àCT/•'䔄¼wúÕ-ÌеM›ûtÀ)§©+d»þ.ÁŒŠëÆï¤É^Î&œÓÇ项ÌX‹È{Å”ÿWÁ€oØÐ°mµÚÆ%ÃM¡ñÁN(Î/Š""Û¸„Ñhgé®™Óñ¡gÑ8gekK#ü×ڼ§°ö1:'˜÷ÎAÈ"öƒ·ƒ‚OœÓÑ8f /¥ ˆ}†dLBgÛ+8®Ìw¶ ¹Ëò—a޵`M±aO^'Pb¦…ÓéèñôR=KA‡/f#?]ÔvÏûŠë÷ïƒ\¢–†wºcôíÊê‚jc-ÃG£n…—ÍÀ3Wq›F‘ѧuc¥dñ‚«R·n•ÉFä ™e =!Ô¦øÁˆ#·ÈþŸÇ£ ßL•‹Í»³ý€¹G¿Į́üY33¹œVË,-qÕb‚p:M^ТCv ”°ç€!ŒnÑ;¨åF¯uñs“L«‹ >†…;Ôÿ¾‹Ø ÄJN¶;â"e®}¿% d™¯ ¾j³zÌË Ãxx Úo£©¶ª}éÏhÙ-æTãÖ›û¨˜ûnævuñ i!›U˜îù‡~«gÚM¯gäx{4ÊÏÂJNšä|¸þìú¯5!º´Ž.c÷€$ûsS gÎâµMçûû¯øâ’¡LFY /aõOF…!‘#Š0¼c¯™vl«·5w´3kEWÇR–×4ëk"Þ‘ %ês舛‚÷¿™ÚÑÞGtûËâFC]©lüÒmËÔ0¬(6'$5"ȳØg—qƆÚpTÙõÉžûÔ‰˜ZǬô:%¸6#GKïÖêÎcŸp(Ÿüðy´­5Á‚f†Ô†¿.¤OÊÚš^nyåg¡yi|»AxP‚ÇÁØ%î3$p•”°Þ¥ïÃfâÍñå?àçwØÇæºÎÇúZú=<a¨ºOºƒ÷;£.¼¯ÍzQHîU~ F×¾š®ôYl ËzºŒ>ûæ6Dd'è}['Y ÇõÑzH—¨ƒÆM"õ¹í÷F-eaq͉€ è/­¶jü–¤eM¢be»gÁÀm-8.E`ÝÆN»b»ÁKÑ5ÊhúÔé…ˆ¹ùÃ*%M•—¼]2"«Ìá}4tsüð ïù‘Äi«ªddc•/³êøË‚d9ÅYEH— ´%îÛc,„ʼnkW¾œ oè )g¯Žòz”íSêÀ}­ asóU†n`ûÎÝn¥¡¢œË óB冉l[Ñ9ù b”¸²Êøn_ÿ˜O°,w 7†jžhègŒ‡I¼?ËÇ" =>‰Ç¥Õƒ2Üøý§díIŒjÓûšü/S×/O'á¬nDB…OÄ×k¸”xúŒÚÛfŠ FÑüëÜÁ'ei?. …=Ü,¶ïµBs˜Ü Î›Æ=‹˜kÆsä9ïÐl½Os-é#û Õι€”61™ÍušÌù7èpF¾ë·(G-w uu¥7Çø ͧoŽKp9†ãØñ ³m×rÝ¿îü$ö~5òÀ¢xÌq;Åô¤6¾º:üJï¶{=•=Ú} á°NÅ]y/¾áªíZü½V<Îç+cš µC“‹,Ñé¤0ð¢rq‹rEÓ’—sHZê½³Õ÷ƒðB·[žHgê͘¶okÅ…q'’Æ1ãw“S„eér¹¾\¹c¥¾1‹'¤E€Vk‚¨ÛáÛÑ©ó™ß7–=‡w]$Èp²>ÜžôÝc‹ñ‘UÒZ·«*žO$éF#ßÅôìèE—Qp$®ß{j6Àƒ÷—Sƒ=.‘´UÑÞõdËŠeÙ#N”+õ=íÌýF¶S³¯ü€[|¼Tê(aõðë¢à™%Á öº«dÖeï*ê™rÑÝ¢£Æ ºÀܘêïçéoB{ñ«k7‡ vç~`‰6÷-†à_³Ê¬BÎ< -匌„T°<ä´€+t´ÌT3O°Fäcö’!ÿ’š±L°[¸û§$RJ/¿}LÉóLI3ïÈñ6’ œ†æt¡ªÍä7ø¶‰Iù[YOŠšœ§­i[ ¦už+ºм¡¤1 ©¯H¨zr·x ,–²lÚü˜¹Ý— Œv¸6B?EÕ¢ÛJ{ ƒy^G0¡ª/¯è¸•¸þïV2 ÙD—ÀÃÆ¹†DŽÚš¶üâX÷ᕘÁ—_Y±œR$º—¤›öxTÞ–º žvlzÀ°Ç= Ë›¾%ÑĤ°‹u 'LTN—í–ßÛ­¼ú¸C@†®á«ëö–&ï<áêbØÇÏ"Ïë¦o(犿\Ý*¤MÛŸF}¸¥BJÒ¹Œ„×tÂsš2¶òz¹çBÄë]Ö/m¾X%†‘‘‡¯×€O¯šk\q]|Êœ _:ŠYÕ|•©Óh“µ7ýˆÔ~½ ^'Ûþ$æéq`è>n©M¬«‹«~•/Ž¢½ÌÆ÷´´·ÐôEž¨¦/vB£kIêÖwÔ`/¡2Ü⯴q¾Ù„øðnFBiIõwaþ•‡B¹z"3E‡<"½°^‹5£$š*Õ×ã¤X¨Ÿjˆ$ ï‘'B;âcߎóm“­ð!?!NT©HaeèÖoùŽkŒlšö)Ü£¦DÍĪåËTƒy0WϨ*Ÿ\å_Ö<îD;§NOI!† †ôóSÙ6=r‰…’h¥¿8b½U/ÔßA“oºÛ ªy+½‡¤ÐöʹåÑàWUÍ>¡i_'óït×…eææ©¤ßfÓÛ,”úóÍ8úR†Úq#3V­{ßÍTk(õD_OË–bÈÄ.nz I2‚ìäÍõ츬çCùrgÓ_ýÈ•¶O=ƒ¾ˆf×Ý|‘U—Áõ÷æïq¥ëRz£+ ¥V¡”L¾¼Et«½£¹!ÌéÇ|‰¡<"q‘>-Ty5ßáƒeö´Éý²-ȵef6Â/^~q ¾,IYüÈI¨Ñ;`#éMéûÒO߸KmÅC=QÍcììÕÓ@d÷–·4 òʱŔΠ0ä |aÍôæ7¤ª”Ìf%ny;¯3@ûZ% àÝtCÿ@ªURåêk“Z±û¥8”¬µI É æO§gÌJ3ÒR‹8]Ë“Çdšš£pd0U³p´Rƒ -àÂÝù?ôûDïEouäe=¬LL²°FÍR†¸è2cm™Ýô0…I’f~ ^ʬó•b8YºÄlÁ†á°qBÿ°·ëXQ÷!ó43z5e'¨o oËfðã=½qîVÀa°™k"úy?¦øpaÂUºhÇ›÷tWÒ10£Èuž • îKÚË¥“Ö×qK5—ç’wÙ'>a1Ý!MKyõ,V‘~ 1š€Ûúä‡+Š%7µ/ÓfÞ³¨"E(y²:‹¦Þƒ&NœlÕ'“T‡!ŒU¾Ãã—àKJR|§lMžb/}%ë¨÷i:ûr3ÛX¨™lË…’3~ÕŽÑRýÅ`3 5;KôãSIfâ£e©¹™¡¨·ÉÓ=EÆð€ tGöª^jq:ËkËs<›¨EØ1v†\ðÄ—xÂÁ܈Ì1ª•z9òòG¡ æ…‰ÙGʧ}6¿¹uvh¿¸=”ÎÆ2 ¶Ä0`ö2i®Ó²$„ï]ñ»üÀd÷âJð§±pXà[Â#±šýþ·í“Â9(÷¶Í··qtÊéETKÂP€4}ôŒ+JȰ1YžÓpÂø»ïv)Bí—1ƒÑ8úížèþÆr×Mòµëx>þùÀôß y6q%oBòÊ–ôÑ4Ñ|³¥®¬»b,þÔìãº=ù †œÙxˆi¸ûŽºM¸Ñ÷"£ÇÍåžöSÚ£tW^7/çCœÉ6»Ìr‘•úO ¥³ù¹HyŒð¥’2»IdfK3ïQ ÖfhÀ¯ãÎH]ö"Ð|{aŸæ¯d̉ïŽj "1uàçµG¨kúÞ²Å`œÁdY ¼ê0àáÍŸŠ‚ ÆÊ·?-&M53Pg3m°Ò¾U ç´ShUÇòÆw~}7&"“Ì"›/À`žy¬ø‚lÎùQ­Ûð£ªBßíàEîä3TN”“u±Ó.Û²L>ouÕé³J|–?[ß9«R·æº´ï.Mn9á¿^¾>VaaU5¨¸Ù£Ã]uîÓ#YP¸¤Í¦m“² iô´%‰Æ› lˆ•³rõçìÐY6À¦-TÌe,KkôQ_L'õØá(Ztþ4!¢ÚÅ›#X£îu·?Eºh8üµv#FEP‰Z!=;8òËfï`×J 7#vÆABqœ¹m$õ>>;׸?ù½¤\Åš÷N}ÇaU¬Ùƒa®D^GWclUú¼âÐ nóÝœ(Í–Zá@½‰z”ª©T¢Úˆý>’§•!6i]ÃE ï8à#¥(ˆ6!1žQÔ*‚u»À/LÎþšAp‚¡ñ¬2M¬‹®À¸ÝµÚÈ¢ï}eïÝÂíÒ•™|ôlÍš¨C rÞqæhUNÊ8¦¬!å (²¢¯P§Kw^3iü›îåoWAsØa"3x Ô½ÜÇœlúê®ø‡ÂRça±Úy1IçÜöâ…Xb¶ÜçÓ´}H0ôðê)Ù÷¹LÁüÊny%®9 w5Mî²Hœb(/>=¸Mdž\cॗ¢—‹P8)Zy]¡L„pô¶{F¦ÜšNYŸy#1NÆMKûë æ(øwZ¶­ÈsïÕ¾/åýnŒѥSEë~hÇH,¼+À<øÑÆô€ôxßæ5þ™ÜVÛï‘fû³ ­æÁê¹K«gtFp§O&Ÿ˜¢[³ƒz­ªü0úO¿žDVëf’ÖŽ°33_Âå³ Cë^?n¡Àú‰±ü¤íêAZ95+ÅLëšÒ˜„.,7R(fê…²aÉô`™ºò²e_ÊO;6[P¨œdØvêü&'ÛÜ¥KeD/Ѣϣ\»ý; M¹è÷qÁñëÜ–óÇx”‘y‰·œ¦cèã^ŸUàtç¡mK\¯¬üÔÛAVœt²aCЭÉ UsE'1œÂAТՉæp´ NU#ÔBa ç´C$Š!wȧÚ»‹NºNœËò*Êl2›<é£S:—Ô ’AôøùNž×çpiJW–!›ä–ós»‹…þ _ 7øwó¾+_Œµ‘y!7O‘"ر( £à’‹¡&þ–Xp;EyP ÜŠ@%ÕMqBº…˼Z>,Z±#±±ìÁµ;X‹£Ðñdt¨=)] ìØžhôµ»¥Y@Xÿ|?Å–LÀ ·kš8%R<£‚£ÿˆjÙ)/,"ô€’}”ÚÙøÀzÜKµ¾. O>o¹ÙQë÷xé)Ø\õŽÀOM‘$ˆ ™ÜAJVÓ¥'Ô͘¶ Á¸0 …9a"»]¡cêÏûÞÞ½ ÿ¥ëµÕJ¹$&ºlœÃ+ûyà–-À¦lÉ”£VeAöVÈòžñ*О«ý»®B_#ª¦¶È“ËYö‹d'9¸äÖ÷>¨º¾tVÕý‰=ÝçÌ¿;V¡kãò¯œ´÷ô5maÛ7b–ñW5™?KXÑ +ÖúIºAãçCŠžÚ6/䘪bKº\w‹é–ÓL¨çÓÈÄt;ñO+§ÖS‰òl’Ìit+ rßaò}DP Zµˆ? A9vÍê¸ÃH––1œ4V™°ÇÅä”ÀÎ:bùÊöV1Ê.„ë§ìõúçâ:Üfsl-)ïBÚš ³Éeâ|/PU=T¯è“?BÉŸ«•¯¹I˜,÷z ‰¹Ô·õ›¢/ógì7¾Ù²ÙbÎò—T ú: Ô+Ѭ¦ÅçbÓËr½à‹ÐÄnùHn˜ê½‡–h£føÂÛIµSŒ½UÜš²·… !Q\4[jÜpU¯V¹”Ì•Ë+O”èÿÝÀ¯Ù>’^wW§¡ò;¡¿G_}ú¢ž7†wýà]JÁúáÐw³ßߎü›Gt¨ÿÄéÌ7Uv­U–’ÐèmÛ€A»'^Ê2 ªî†e%ªs ÆéL÷*ŠÜÜý›ÜA}Iö»È´sRþlÄ£OˆÞ9‚¯>xywsº´4‘ƒ1ú…šÆ‘9ÎÉÊopbx"ïC’ãxË×ó^lÖ›ú&nQ".¡E5mϲÐMbXæÕ²FOp0V1:©ûޝˆö¢„_Ò¼´îØ4 5uÜHŸœ §÷I÷ŽWJ7ŠQ ±ŒrËuk"Ûx¤6¨Œåx°ž}Jˆ]1³bth<Ø·° ùZ‰ð ‚ky¥±|øâ‡Y!£T[×<…/pv ÀWQÁàöFàͦLà -ö16fÛ º÷1Žh‘v êY(çã7Ýù)»9Ü”Óv‚Ï"釳š½Pˆ²p3²¤že¸¸ôÒ•ŸäÅ‹tSÊ%‘i:dîo+i?§HCGÕ]@K°Ÿ'6µ„-âÙú3+↡ú›ÞŠð¨E„[ÙÅï5’Häspíhc€25n!AO4Ÿw;B4°œï¦ ¤KêK Î ®¼ 4c“vÊ÷ðÆåÀȤÌoåì ¹]²TFnÙ°E‚:•·oÆ4»[¥<)¤÷r»ìO¥ó›Vô»²¯»[Ϻäx[] [UÂ0ºÕbÇî”êDÏïCB¹EMãIu<²¿S4×¢ñÛÍй\§”ÀC®?¡žB ð"Êb}r•ÿV‡•æ3—PÛæ|ürz³É† ]‹€áÛ€º7‡0¨Ð@›?2t¥à§¸œ#£¼› øØºy”ÌÒÀšäUjÿ|SÁY“ŒÍòa úf¹è}ÉÞ¶ÈAq¦Â›Ò5Ö}7‘U9OD©®ELº#¼&AÏømÖwÔ²Ús´„ÝI=œÔRBÎÌšp½ñ N¼"Å›mF%S œˆ¼ÐWúaë.rö½Á$ÉÓTw ^ÂÛ.Àæ:¡UŸEäWC)ÝiA-Uª¶­–ÆW=³…¶ŒÈðûTepõ£—îkiÕXçU|fÞ¶¯ôWb}YQ`¡Ón£vÔ8Ïp×›¥ào"&Qí,·C”rTøñÄð‰ðAj˜”§‰ ×g¤ëg¼W‰Þ–eßô9Q1Ëåƒ y5ô¸æxó#.Ó˜‘)ÿé«\ô¶Ï8ñ“€þüNÀs¹!Ê\²ø-´N9·"³Q솔Mš]U Åeƒ¼ñ…´q]ªz¾FîK¨ŸÛ«ì÷‰ÐOÒ:Œ@& ckêf@,“Þì¯Ã¹M*­Ì|ú›ÉìVœ_$ŽèvÛ5v“IÍÑ–{ÃS•´ÕwÆkÃØÐHg~Ò¢á5¯`d2j Û‹…¼g³ÂR.Mûù Dní_Fµu¸m392Ï®Æ1VìØ)Ò² P]yËVœŠõù©±ßš »žSìôh`¨dûU÷.¬[Ò{9ã[¢RbSZ’{½ÔÄ©®¢ ðä°oðWžšÅåè(‹4Ÿ³f¿³!=™˜$_DÓÄxg¼³EÓ‹0P㺄úy¥+Õ_¬U§ÏaF}˜&ws§¹ 7½£òci7)K:ƒ\‚ m¡«™•ìÄ›«2DbµòÍ=EtÞ²t!^@^ y¾_ð™"Qû½‰À8”IÎl×{xÿc°» |›÷RT ‚'ÞHµMÇyê]ûÞ=uuã”õ#ë\ÝšfþM•R*=ºw5|Ä‹%ÝeóJ5óŸ*Ò‚_?¾Ì_,’¡Ðñ¸=üÌÄÍÿ&9‡Û{`R L€×½æ}øf v¬ÝÃÔWb‘¾âù.n™×º¼Œ<**, t™6úÌ@ƒ\§ôx5ú©›#áC¿ºÍ„-wV¥úaWL3^²ga\˜4ûJäFèBÖëÇlWß¶2 ó]™W´ÜlF0ã=ʪ,ü Vý±­O·I±Ð. m‰×Q€© nœªÀT‹÷]Ao*‹=uÇÛ®½ `.Ù÷çXëI‡²ñ—M7@Ø 4õF‡ÁRj~ÅK®g'A¾2¤>¬Ó6•=3¸øq QPÝàŽ*“#cþÍ—¸•c3â°i–'^Ï™]“Œ¯Z8¡¼1 VÀ³òþž ¾’xK§X£³^ûVcZ¬£è½yŽêxF!¨…¶äùxÊ¢ ï瑽ÀÕÍÃöƤvšû]6iÅ¥°Ü[’îð™dm2öï7…qpÕö ÿÇŸÀ·Lý=Á÷ œ÷œöDr‘#ü”wXÕ^ßéŽ N Tj¥ÿd dX+Ñmêz|ÁÞð¢"ªQ„0Ÿ(¾µYÉ6…L™Cìdd­Áw“»¬âwì0£hrŠ.b3“ùûTPMÉ¢ï×Ó:ŸÌ8Õ„ È—{~òQuˆðëpFM´¾öfîmeEçü %^¯ÉèèíŒñ8Í–&#”k¤ð›ÉÙoïûM›º±N#ÉXý,Ò‹+ž6šú¦Ãa §Œ-Áß~ÄzÔ1{M2pâ/¬×rvÙ¼Ó- Á}2 ÌxB7þ[#ó¶¯ X4R_ðgu_êääýÆÜ¡÷γKùçÒXÊ4#œµòÍœ¬`Ü/b«´Š o¤a'Ž:+î`@¹uÿÈwzlØõ…ë†ñ^®Ãµ-&#p.àý×;<‚ƒy`"(pzµ²×ï¢×(#+m@}àE£Æ^ûo> u„?¤_#xa0úÒV1±­<_BÖ(œbIõÝñ–\€ó¥óIÓ`~1ê£G{¥8ßîT#´Ë|š‡†¡I´=%uÓ¢’´/½KóòHºõ`8:Ë›AÙ‘‹œeù®êdó¸ô›½ô^á:!R¨P`f‰º'‡F†c>ŠËã<Ÿ Iðq%wQ”7|8¯ZØ#r+™«Ïï\½SôªÄJiu‹‹’ÅLœšò/‰Ú"jTÓº¥üë §OY49»Ôs ‡6ªß¥Øîäx¥’Z&¸«û+ªX“¦²é-×µ=ùßV:=ɪ³ƒ­Y¨ì¢ÕFI†¸„áqóóáûcÕÄ3Œñ…‰wËÉŽ¥a *qË~3Sö1­Öñ‰ˆŠù¾¿˜~¢AuåÊ•8þþ£ÞHò5Ù|É»¤! b5Îù#ØÝ0œÌËÒ¢úŸ:Zâí‘ ˆ,_4#7ììF?+þÆ5ª´†ÔÞ&¶MØÄÌc.„R²Ê¹T„%üJîC·@Å.¿ÕÙxxñˆÂä‚…3û²ò66N’Zƒô[†$еtfÓNÓ&}ÕÝÊ.!v–<2rÅMSˆ2ÌqäâÏc¦ŽÈ Æ Þv•5 Ä,óF O¼Åût­¯bõS¶I¼È‹e— 9­I\zKÒ€ÂD°ÜÍ"z­6{ñ¦5ÏtñCÏk¾@IQ9Š·JhƒÕ¨Õ[QúKpåÁ‹´«ÈT=¿®ixf _)˜»L&™…©!˲·Å¥Ñî4Ŧ\»1âÓŠwjý®fåq&« ÷%s˜á²ñX„Ï V$à{ãLÛ5ë¼¶Œ‹›íÝU«c]k§Ï2ã“MÙ`Ÿ$9$eiµØPWÎF5¤ãk©øŽONì &>´áweÌ Ü ø±•ó½çÌ2 ¤xÿæø…ÔcrØ2"¦rÒ²špS½ó2åÍKæÅy¦/89”Õ·tººA<íÕ›V`ó3Ÿû‘‡÷=¹^Z£-4ÃîØ ¶‘JN"ȆÚ. s^`½¡Lbäµ»t…:ñVhóB\@B;Ô¿´ôâÆaáþb‘ZðAÞCù¬ßICb­¨n—Gü`’êÆz²ÌëúÌ‹&ðË"“ÍÓä2ÿÈ¶×Æ£Kv{Ž£_uFø>òÅKÜËî]¸ÔFÉèž6µæm åiÍ‹È%…1!ë(½J¸fY¿à¡¼S£Ò'Ü Ïºþ(­ÊUôÌõŽØ¤)üã놧;”¤¼h8†–=ßXf›†÷À_֋ˇd( 6LðþJ¨Õ—sS±'ïØ&> ú¢™%Díkh˜5ŠÊ[ ÇÕ÷É« Ü¢ 6ä–Ôïd•?EE¦¤ÍÊØàòø™5êÐtf²ú¦ O½ )/«a†¯ýXARÄ4„:7ÁGü?¤üzu‹!OÕL€ºt¤†_ ¸:¼Ž¼úªÉollÿ,L³#~õƒ™¦½¹ÙË7 [§xåq#¡¿˜o$Œ¢,!Áþ?ðA'|†5¹­?ä (SUŒbbV¼Ud^ö7¬¸ˆ±¡‹’£/užø£¼ØÚïØ}r¿ ëQËc©;älêóœÞðÎ4|XÉ=“£‹Ö­Ù3¾™Í3\8¼úüÅsBÿ*ŽNtìùÁå1qºûökÛæ!øî?.é°æDBç®vžô@î±²op)ñ‡û Â-%ä>Vý‹Ác endstream endobj 158 0 obj << /Type /FontDescriptor /FontName /VHACHE+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/a/b/c/d/e/f/g/i/l/m/n/o/p/q/r/s/t/u/v/w/y) /FontFile 157 0 R >> endobj 159 0 obj << /Length1 1515 /Length2 7537 /Length3 0 /Length 8542 /Filter /FlateDecode >> stream xÚuT”m»5 ]ÒHô3t#ÝÝ 1 tw+(Ò¢„€ t(!-ÝÝ(- }ë{¿ÿ_ëœ5kÍ<ûÊ{ß×¾žaz¬©Ã)e ·†ÈÃaœ`.@FMW Ìx¹@ l&&]¨‡3ä/;6“>Ä…ÃDþ#BÆbåqo“µò¸TƒÃÊžÎ0/, < ð_pw€¬•Ô ÆP†Ã l&¸«¯;ÔÞÁã¾Ï_V ,,,Èñ; åq‡ÚXÁjV—ûŽ6Vθ âáû¯¬b®"ÜÜÞÞÞ\V..¸»½à õphCw/ˆ-àe€º• äOj\ØL](â‡ÜÎÃÛʸ78Cm 0Ä}Š'Ìâ¸ïÐQRh¸B`«þÀøór`.ðßåþÌþU ûlecwqµ‚ùBaö;¨3 !¯ÊåáãÁ°‚Ùþ ´rFÀï󭼬 ÎVÖ÷¿n—ÒXÝ3ü“ÂÆêêàB@qäþUæþšå`¶2pÌýë|²PwˆÍý½ûrÿ9\'Üæÿ²ƒÂlí~ѰõtåÖƒAݼ®£ØLƒ59 ¯M­P±‹cÞ¸)©¶±¸‹%¢]žÒ»z'C2“ùâÓ¬8²éÚ&4;ëUާÃDâ—»UOž •ÖVcK^˜­=ÙþGíï“'£@r·Ãõñìm1ægåû,u{\¸(ƒ q„„xqQ›¥²›+¤ßK޵(1"Žž ~O£‘ß¾ÈúÒÇÿÒ5SÆ ]΢ÚÙÁsU9ÉÞ‘ÈÜê T±Œ2­uÈ=ºJf =i aµFôëˆ}ÔÖNâRM ²mÜñ\SJuUäbß|~˜zn6k%³2˜×¬äw=» bñM«þV`¬öxçø¥ýÒ[•‰*ô¬ìúÝÄ6GG1›âÌÇpu²—ÍágÜêŸ$'œ{Æ’DÚ¥î<ç÷ë]¶Y±ž¯Æìµ~ ±) û)û]GˆU)r(Ôjò*"P`×ú­äóãÙÁ›q/LDž¡·$Ì¢é:sä­?”Np+’_Z‹%tÊ”Á(T*×ù&é‰@Ízû,ã"dËap{F« VW±yOCŸ/vÎÔåÉã×S…ÊŒ;Rtzm YÑUä­z˜¯GÒ‚ˆ>УÓU‰<»Â°¡Òz§Fí&¢öMcXÿ”“2æåãøn)AOtÖîÆr¥ð…§J(1µÆ’öh’Ê:Ås0s „é—:¾ACçÀtçÍÒ éù×Ü4’‡>(© JB|ŠÛî[(¯ç_ÎùÈÇp!!'ëHÞÚCV êÏV­w†YÔY²„ÌáÁ ”ßñSrÃtçà éÏ,P§(kq<«xÉtJMJsÖ+y+Ìâý›_5cÄ=jJx}p W¹‡Ö»FJgðãý…S-ƒ’Öi¤pδ-Õ›95>K®¥½F¸ÊiÊ äÈDWEhu}?Y7w1õÔHƒ­Td']ëJˆ41ÄŸ2ØMoà·©®é‹ÝP(3À€Æ;§\mHŒze¡¬¥†EâQ”Ÿ8ÅÆTÍ®â¹á5ÉUw¨¯!å(YÇ mb¡ùS¶×ÂAœ£OÚE2Tž¯Y—$‡ç”žï½ÿ ‹n5œ fÊåKe¦íclúÓæÎ§#úB>-UHí`æ|óÎ~Žö ûà!Z°E‰éÁ˜d§Ee}ÊÛ̼àfg÷ßÉ^93dj°(Ø,•â9ñHd½šÃºyÞ×¢6Ê\ü¹¬Y¤†Í+”ZIJv*›Õ'áv ø(ë'‘ÛÔÚjmMX3A®Iëû o"”õÐ ®$« Boì„Iä!‚ÔLŒqXsM„(ÃßC{b`Ö8"ÏVÞÊîž U¨áÈÉ®ó!›ƒ®ùù0 ¨øêÝŠKV‹”h»ÃÍÊjEsMâÚ3ëŠæ7”\p«ý §ÇL¨›ƒ¾:»ˆ±)~Aº˜ ·«$U¶ñ¼³Ä{ùNˆ{s$$ûnôQh¯¹³kWéBìj#¹4Íú€ E6–5 L¼î9YJaâÝAP,@XÜ{%î“hþPŽu“kRä «üZ|Ÿæs\ e§Íªu& ÅR…#5:<¦U¾ÈÅÚ%JìÙx-Ç£ÙQp§xQÅ‹•è°¯%ŠÆÅI_Ðû_R{kô„Ê¢f­N‡ fɤ=Š3Û©³ç°y$5ÝA _’Lúñ-§±Å±ýBR4¤Véyž”BiòŠ‘f«Š ioÉôݧÎ-é#„ÅZðˆp«TT¥,’ÜâÞÁOõÜ÷l¼P)&ÐPü78¨8o?urÁ}p—;z˜‘ud‘èßòûGÀ(»â~–Æ0‰ús‰‹ÄLW˜I¥%…o_Ö—ëAqÏÅ£~Þq‰„¬Lx{¡íß à ˆ…ÄÂ.VI%õG‡Ùƒwz£ãŸ¿›J9©ÒÆYuG© qA¸CŸ%ñõ# Óì“•Wц(ë¯î2ãÓ… Þ)VÇËúyu¥†xmÄDè–¦×ô/»‚H±«é…#óÆQQ ¤ËüŒ=Ã%>˜u‰g8ÌÖ…~xEQÉÐ:7ù!µo_½ô¤[|PˆÂÞéÓ‘Žù–ñ`è€-a^µOêf0f¢ñ1íÁ.N{¾¦ô(7ÌþE­âF0‹<*Óc1¯æ`—>xõ‡šùÁ\Þ¥é\‹bg³iëmÒÚ@¬Š’GkÁU:t pÊ$[Ús/û´àÂë€v즉¹Öù×è–ý_wò&Ã÷°jñ7nwcžt4榋=×k­ºEq3% Õ‰©.‡¢„À†(íô¡‘Ï{oq Ë“ÃëL(i?|ózS0É}™®§ÄÁKŠû2ÚùcÈXœ¿Ó¶;klp¢ÔÁ¥v’PaË »Áâ3ßQž2t›õ?ýRê²[¢/éöþrNíÔÑ¡™˜‰8N<¾#UZøvü7ØïÉíÞ,E¤vݧb©Õò‰µf†Üô²~¢”¼ûË{„q|MäiU1ctH)Z§Xøòm™x—ôn†I»™–ÞÕ^!N³¨ìq8ŒOUnÈ —‘(‡&ÍìáÚÞ)ÈJ½P×Ò&C> ìŸR»ÀÏÁïSˆÖ‘祙]ªõõN«gÅ•K^Ä òsòÀŒGçÏïT\µúm>÷ò8‚îë»» G‘7ó9šL¸’2Ãý?I3©±„çƒÚ¥ôO\ýr—{ˆ:•‚™Œ m*”±wËÝi|ºˆ\Ù©ç_»ÛÕÚxS®>/o“.“Ä”Î/Egfö)ž‚™|›çúír·–¸Héá<³ÒXÿ‘Ï äòÓž é”tu{%â-ÿjXúØYVGB H†¸ò”åųCL¢È¶gŦ½òo'òÑ[n×ü»Lf´ß.ãÍò¡™EЊW„EbÌ·ä, 2ûr6J”+ŸÛÁDšyî4Sbèˆs¤ßޕޔö ñšžénò”?SPå$"sîW ΀²"F êà൯ª~ëݯž®è°ûX£…®"(Ù ¼y"ý“Hòâ äÕñ›»bôŠ ê"1\°kîN+IdX(öHÌ®çz™G-³b£ BqºõSqŽŽûÈÞ¹0/Ån×K¯Æ?É z9À{ñul~²U:‹Ë«’f®ý€ÞaÚËU£÷E۳φP39¡Ò×—±é§lì»¶-ËÁ!†äÚá}ÔJxƘG¡ñ%_yååKÔ·n]Cd€`¶‹=mý’jlGèŸ5J4~¼̢ƒ^ꔺ1„½ß ®´S¸n“dvQú)›Ö£:‰—ä,•êˆWÝö„ã%#Eú*]ëÈXDiŽíIÈH_:åÎg"¦2ó2<?rŽîtÿPI³c< ®kÌùYધÀÌ„ö£ÆÈûÙûÊœ§økxß᪚Lß©/È7«™Aä"îÍ•7 )¿ájãuº+$£ Æb"½´e\ùŒ›:‡â¥:ŽòSîrH+~C î¡ÒgðîcõøÝ¨x¾æ–GSl¼«jéèë,´NŸ¼B{éð!WBÜä'.x8$çÍ㳡¢¢©k(žÛëå N’ ê&-7ÞFIúî¾J„æi¯âÃÙEHn’µùåixàÄ[`Ö³\cÈìÅ3Ž¾Ô “$l_¶J4É¡ùšT¹o9¸-zÕ7ëlÓ^ËÙÐÇeoû¿iXFñäER†í쮥hSòǨì-Œ½¡»Ã l¦i‹ s“.HÌFŠŸ,¯-j§d•¡e¾ÄÖ¨ûzz3´—²Áu¥o˜öDÃfbÖ2¯ ”9¥ËB½œÒã:'N#É^û–ÂyhŠØr°÷úoé¹Ì„°dÂ^Ëc‚œ—8>W2?z§›û GS¥àÀ-+SU¯EÙYÀCÄ¡Á¬­)VJâ•ö'“‘Pr¡¾Ö¦9uÿýT1­¼bÌæ)¤A|lÖOíU@É Áçxï5kÙqÂÙÃ9"½5üõ7âSXGpgƒÑ§¨Vd0H’>Ë·£Gc)_RN À±Ìê,)d ‚®ˆF{¥k¨{¶øˆZ Ó† “÷D\ ~¹åzû“×^ ©ÝÓÑÓ2«‰$)v3ðT¯®u2ínsͲÔðnió~é®Ì¯×¼PKœz‚×moów=oÜ¿ÝLT)°¼®O;ÔÊff¸.o–ýìÙŠ©&¹Ò®5Xu>Ƥª{nS†~í#vl,© ½ÞŠqj¾Ð0Äù 5~h¤¸†¹­pÉùj:ªá¶†pñMа¹è léWñùØÝõHâ"ÝN¿wE NntȽsîýcê-Cwýþè¦÷ÞY°hbqâA75é«8†€–³‚Ÿ'8ËuÕ¤5×fŸqÏ­,As æK\o âBxo<ö=€ñÈ~Yó[ÌàÁKëç´pöñä[ùø$„çèê“S¾Ù»ruM ¦áfÂÏ*ÊÅù‚Ö Æ € ">¸ƒ<žpšŒ™!K´k0ßSÊž¾©#Ì2+iAÁؾjBØòÙNý®PMpO87g+jŽÝ?O}õàˆìY…¨‡-^ ›}ü¾ßà $G+5Ò[•÷@tiU†ܸþ賎Һ‰4¼ÕÂz‰Ù»å xÓ‹´­?Ò Y r=“Àu1è1ë)dÙ9T„ºcK>=¹ÕÙMÈÃ|§ÖcÉ¢fXß³E¦ÎW5\è…z{ÿËBüMUXóRfïór,:°XØŸJT‰HY0v:ö^ž-Y{K\!Dž ¼hOÕ™N?a–¡)a_˜Oã(P>Ôé¬ñçypx«H09c€ï× äj°Åå)HŒ1ì*ÜÀ£•‘5_Vœ¢¹JTýŒ§¹€æ!j“¹Ôà®·ûÀ@¡zþÁ‹±S*¹øÞ…$ß±ó©Éç4+³GönE.z}ÅŸÏÙpí¸TUЩPÌðxã$"Ôàz§shˆZB< åUAÁ®Ûk”ùÓ!BŽâãô TdƳ±Ë¹¾$KÛD8ì'x‘¡‘5”iIECÏš[’ÝüÆw<QeS2äñö=nµð“öRkíJ;?dmúÊ—¶.± æ2ïUîvÙ~¦´ÐEÕ&Ëv-hP:»ÆâW¢P8s„x3uèv·Žij(èåRèòàPÝ)÷!R•âåé j¿È¥®éI½< 5!l¹XèÁÅx×Rôn+~<þãè°î3*åpl uã¼hlì+¡wYô[üÈ0–ñìô°²ícÔ唥٭îkëxKñî±DZìÞ‹™£g*@_»ûâ‡Ùâ3M©‡¡ó³6,M:tz䜨)OŸ+Ô´ õ¤=åmlS£Q)OpjW ¤gQ¶¹C̓GüédÅæE]Ckë":ýn¥zR|JªDw¶ž–<ý (-åÜñýø‰ ôyÉ+Zß88’ôk»®•O¦;ÉBäÏtæð¦¶g—÷%ÅÇüT_ø`kÈQëOÅ÷$ŒYï ×›n0©3sŽf³ ÎìÄ;-—I–ë(›sÔÐJŠe&O¥;9±ÀPCÕñcÒÜ‚Ðq=/l®mÎ'q4‡Õr¶n fÈú¯Ÿ;h—sÙÙοO„Û»óúP8ÏÆ_Ç]Ö'!¾Ô-Í1Xùn#g\–A¿Oña¾wñV• -g6Bîã{²NìÜ÷fû¢q;÷­õ ‡x{Á\‹ç×=â¥é ³ÂX¯8Â@1[»"ElF%³–9Ì1u&„y7¾—ч.Åw=ô(tÐî¦L±bA—B’¶níëˆÔhJG`Ñ'§Ÿ #~c<=b_—=QíïãßJ{*,:ÛDÉñUw;ǺüÄP¤ž¯=ñÆ=³6žFƒài5ÁËFÓŸ¬Ã8Jð÷ãÎkÌ…^¹.L•‡ÇAÛþº¶(««n LÝ@²k½Oјä§"t ç•Á~’馬®Ä>±*†áO&3¯ÞÍD/ö›«tû+˜ìŸ¾0'b³í¯Ã gœ¯µP¾bsq=mÍíSòó3d½c~ª;ÚgšYÉøÆòMÈãÀÓI-ObHcÆD<—ÂÌ„òºÏ¢¤ŽíÆ\P·IH›O©ÈꟴEÖdÔ}'zË¥`$¡šæ$OÄÎ&ªßÒ °[Ùµ´—s²Ô%vbÚŒ¯æ­fKÍŽYz›”¢Lj¦d:Z/8Jª}m%?ÑÞx“ã˜Oý.5­™ÙÙ×/3¤ßšˆÙ{SrÜ,ºb¥ÅuÁóúTjÁÈcþ¢8½kký}­‘²ÕGòÔOÌé"mkË“¨´[ˆ4*OºR»›X>RL‚C}ï“™bPš©mÓW—ký*qåŸû.,*“#¥'i&×{×Äýh‡|wÔ%~uÎÙyój±†-@cqÓO…l$;{ªÊ+Éõfò²ë‹Ù5Îrï“»{çY±7¹xWzìÂìŠO~¸zQĨjc?øCFeøúÈøðWÐdRuè“ïP´ú†èêGÈ„B•>­ˆpbEâmÉ#šAáþ¥|7pœ"Þ²MèWì=ÂÎ%ú×6¾1çÌØ'¹ÆL£ŠÃYÅP åêËpèZYÅ€!ö÷“¨Îl{õƒ$ђƯt=8V¬š oqøò¡`=͹Rûµ­‡úx}],[Má;ÜuSéÄ»|¨~}GZ2ö“,ðàÚÚîöì¹³´;‚ i>¸ÌiöÑU!h€¯›©Vf}éA¾Ÿ°nøCÐøc,v‘A5Á —)ü¬Z²ñ€3ZƒÙ¤6iÝ4p’ƒâÒ[®v –½[A™éฟÓ<¹øVÆ'nšÆóž˜0Gz¯95_ögÈ4º›«¹‡À9t ±ø'+}îä…ÉÄ[ÄÀí 7f¬Ò°ã!…ÚŽ‹¡*ÛYO®Äw˜¢ô×õdBAÃ0›LñÉÅ-[„âä²6’»}Ÿ ÃoÆ85ØOçèž´SÛò-h±P™ØÙ¨…kíÇH.Pç?v¿Ã8iß.ÓªÚÄê¯ñÙ²|²Ÿ)aýð«aTaA4æNq%ZÙ½p“¸VšáÁ9ù!Y$eZ“ň¼²½ˆD÷ÃÐ(¿e§N<ýþÒãËÎmhoY ÏmŒHÝfaCÞP’ú™:–Zƒo{ÞÚƒQIœG›Ìèq»Ç?ª/™¿O4}ûøYó¦Jic­ÀÃl:Òx$7  –UìJs Ïn,ûnÁAõ“©7P†æS‘½“6*j2 "ù˜+qÛŸB<ÿ*îTq¸~¡u™ª+ù米чf¶nö›¸çôv«î¦ 㦬úQMÓÏq&TTLï^‘»ÃŽCg/y¶›2”pG†u˜"w•ƒ'˜‡GIÛÔ ¥„lÁ±gZ!$OH¸ü¯°ÈOHo·«’Y^,½ òQê40Œz͉Z> endobj 161 0 obj << /Length1 2584 /Length2 18287 /Length3 0 /Length 19797 /Filter /FlateDecode >> stream xÚŒúPÛëö cÅ¥P´X ¸»»Cqw ¸;-îVÜÝâRÜ]‹Š»´P\oö>ûìöü¿oæÞÉ É³üYïZï/™’LYQÔÌÞ$eoçÂÈÊÄÂWPWge°°°3±°°!SRªƒ]l@ÿ•#Sj‚œœÁöv|Xˆ;€.™Ðb¨`osµ°²X¹øX¹ùXXl,,¼ÿ5´wâHÝÀf&€œ½È™RÜÞÁÓ laéÉóßSZ+//7ÃßîQ[ØhPºX‚l!M65{S0ÈÅóBÐXº¸8ð13»»»3m™ì,„hî`K€*Èää2üE ´ýC ™ n vþBÍÞÜÅè@6`S3ÄÅÕÎ ä€d¨ÉÊ”@vÿ1–ÿàŸæX™Xÿ ÷÷_Àv;MMím€vž`; €9ØP’’grñpaíÌþ2Ú8ÛCün@° Ðbðwé@€”¨ aø?gS'°ƒ‹3“3Øæ/ŽÌ…´YÒÎLÜÞÖdçâŒüW}`')¤ïžÌÿ®µ½»÷‘9ØÎÌü/f®Ìv`GW¬Ä?6òo™ÈÀÉÂÂÂÃÎ9@¦–Ì%P÷tý­dýK áàëí`ï0‡Ðù‚ÍA7dog àâä òõþSñ¿™•`6u˜€,ÀvÈ¿£CÄ óÿ`Èù;=z,ñc°üõú÷“dÂÌìíl<›ÿ}ÄÌï5Õ´Äéÿ¡ü¯RLÌÞàÍÈ`dcgp²s¸x¹¾ÿå_þÿåþ·Tþ§¶?âÉÚ™ÛxÿCÒ»ÿÒpûg.hþYZÀÿfP´‡L3@ó{øõY8YL!Xÿ?¯Àß.ÿÿ&ÿ¯(ÿ¯Ãÿ+’rµ±ù[Oóƒÿ=Ðlãùdš]] ›¡`Ù»ÿkªúÏ:+€ÌÀ®¶ÿW+ë„lˆ¨…Í¿;K=@fÊ`SË¿'æ¿Ç‰n¶)Û;ƒÿºoŒ¬,,ÿGY9SkÈâ 9¬¿U ÈFýoFI;S{³¿V“ trz"³@æ‹“àÍ ÙQ3Çߣ `f²³w¸ ì|æöNÈ)'€Yô/Ñ7€Yü7â0Kþ‹¸YÌR¿+€Yú7â0ËÿF( ¿€Yñ_ĉ¢ü±˜U#v³ÚoÄ`Vÿ ujüFÊtþE¼|À;$ÐÖ2£]?ÿ•²²AJ:›‚!½¶1ý+ç`ûK 9n°³õï€&¸Bê2qš‚l@æ.ˆ9ÿÿgÆÿufýØäò?ö¼ìÿÊÿ„Œé¿ˆ’ÛÔÞ2ÿ–Èñ—ÄÖö7Å¿…ù77v53{ Ó*~³„Œ3è’rý¥wt…,ÜoH+Ì»@Ê5»ýã/µ½ëŸ9 &¿#Bô=Aš@j·üÍÒ1KOKÝø©Ôê9Zë? ¤9¿+æ‚tÁæ¯åø­‡´Òö7„ÜÌ¿SqBbÙA–êwß ©í\mMþºÎ,þ( rk3Ûÿ.Óþ/VVQ‡ßjHÈ£Õõéÿ5äFf†L&äú‡)×ß2°ýïå€4ÖÁÆõb¯ÌŽ¿Ã@ºêèjï23ùãüX!Ò?ZÏ a÷;ç_äöGç9!æÎ'Ï¿I!µ9Û-ÿ)íw¡Û›ÙÅÒ ôÇH@¸»¸Ûÿá‰áú„—ÛÒI÷?öâíñ„„÷üBºìõ»8H$/ÓRýÏhêêi³ËßÏ(Èíø_ü÷ÈdмüÕÞ”?ÈêsPÇm­(¡;ãî¤àå®V*-£÷²Ó×{t„$ښ̀u§kѤ‘Þ׫ے4W"+¤OÞÇ­ ¡m *í>FŸTgvÛ‘—¦q§ E눑ˆÕEö|ž}4ý­a[¡»ä(s]yЕóßܺ÷K{Ô”ùº«²WÃõå±|–1Z#Jß¿dž2Ï$kŸÞ…‘‘ë§ÆüÕõVÎÔ ©Ü'zdß“hö"oÝ ¶˜»¯µJu6çn ]|bØ+¬ñ*o±ƒd9¼EïÒâ¨uÁÅlZ&¨õñRŸ±.²S>=yY,<§e^éq ßözfí>T¤Ì]5cºÏÇè‹øôæîîRÒùJ-Åý†{áH7aNôf n±ŠžÞ{›ËpW:ÇWïO?Oéyê™o!!oá¬6‹­I<…:úb+À]îñtûZe”QÔy÷YM8Ìyìy¨}¬\"û9‡YüGÌ5–>SÓL·%ª¢pßGjßÏÃÚfÚ\®Ri­úáE® y“4çÿ|²•ê»ö£‚8«!¥‹pe!m„=NêýYÜ"n\unÑU…%eô&þ˜µ¸(™Qzpˆv]6¿ïçøÛlì=«í+ÉE× ÒŽΡv‡oqb²Å›yjÜDz\n•/Ñ,¦Ð2ió ?]õNÐÜu¾ô%§gïf±uè¹LÏŽŒ&¯D5d [^¦E´˜dn>Ó-ÃPø!› GP”ƒQƒ3Âùü_“äž—ÑÆ”ùhÌü,;Ž2ǃ¿«x#ÑXèrûzÞ,¥·$W°2#“sâWÇGz¦Â3+îÊc°ÈK îÖ›aLZè™°1uf¨iǰ·«ÓR‡u7q­¬‰.’S³Íkù"h3ù~x¬}Ÿ¥ U…“³’².z‹qƒ 3É¢`èX-ÜoöÙºó8˜‹ssZØ»D*¶ôk¹½¾½ŽfX‰iÛ™6Ò~þò”ôСn.¿9â “2ý(hgÕɶõeþ«)öAœ;+jôãíø»ŸIެAhfDgËg¾˜ñ4ó¯ye«G‘ »GçÅ¥æ'”]yàsR%jï²Ø…ÏcèîÏj{j  j‰—ó±­z¶©Éî=_’©*?±…Rû¹è¶O§o á{RÎt‹sàQŽ—_¨“âªBС•¼g¤RÃû€›ZÃþ·Süœk™…ʶ¢óÃô¹š‹£^k£ñ¦¸¤¡øÆì†êèÈ9ÊnÈTokk·{Ž*?ܨªf”‹‰ÃÖp>ñ¾»=ݤ;¦Ø_¢=‰ÿÌWB£E½Ç õ,6 ]ŽJÀè?®ž’´€«öƒÃc°Î‚-ݧzýq²B’<Ãõ¼}Ygã¾Ìø‡挡1 TMRÕ5¼mÚ=1Ê{jŒ÷[shÔB9˜™’ÊŠ | ÞjìÒlªè"\ú9ÂŒ? qµ£©ý¾\}”Âer´Lí{u£C˜sñ`,/u_V¬Öp˜(1®ªºÜJWà­ƒß1–³˜5q'ÛJAïpî÷UŸM„ª’è†5 áÓr$3øÌsoO0~9(,ƒCˆ¢Þí5Þ°4„ïEUW•­°“ Ùì4åc~™ˆ¿ˆ¹éË5ÆZ»þºíŸêw½¸ .%q[G§„«ÐamÝ»®ÏÕèš©,e>Qº¥ŽLiLuBäDƒÿ~€Iü<ñž7¿ËCf€ñÛ–äã<‡‡'IØnlîrÛcDz?Ãev‰Ù– 'µÉ˜Éš#C²Ì5@)¡ö€g`IQÑîF·_úâ‚£ëIóPÝW…D˜ÎË‹‡Ãy3fX&ûÓtý_’±6Û:1F»@Ø·cȬgŸ}í1üÞCÐ$ÿð};í¬°Ml, ÀDÍnó%xíƒu‘‰…Uj8»RBó°ÂÚå–£­ý¸$jG¦ ÷+КSðh*C'ïL²ìþ+d̯?+ÄM¼&Räµ-W÷=úÐX*pKA‰¬»âÙk0o¨¯@’‰–§@4ÓŘ>eÒþhù(±º Í/Ð ÛxÌçùc®zÂl_%–ÉE–ÈÒñÓâs  l®ã™_[_Ð~"ÎÒû épì7F¿¯’K“»OôON'Ø_­ƒKÏK<ëIZ©DŽò_:b¼WO?c' ›¹'ú,¡I˽x17gÈPwêR…/C¥-)LòDnÄÄ꾋.Dà^÷®‘wøŽÙw£IKùkóÀzt•=0è$MݾðQÜΗC¨Y÷Åi¸Ö»Ý ¯GC÷…W²<çë¼tõ¦ë'Þ¡ c{ÇÎ æ€Â͇è[ɱjÁ>™æ¼¤‹•ÂÝÚçL®ÃbæÒ&’LîZüÁiö ÝHxEbòÖÖï½xÍ-, m}*¨:ýfñцtó×áÄ]nïV›„N¸5NL?Nô5!T@Q7Ã`‡Úµ(bL§M²F¥ïk}¬xF(]~Ñ\E-KUt…§ç}YÁààÆŒô”Æ ™é}^_G+õGëíŽ ¯òÝB·Wê5Jô©4Õ}&ò+Ü iÖõ%]ë¼Û€<’w M„ãö…È'q’ _Ftºƒœe~Å ?T/¶k°{³Í«ñ$u$W#LŽ¥ç`Ћ²òjÔaÒ2j©'w²;¾ô<“^‡í)C³t±I’¯rª4Ð\zŒêt§d™Ñ3ž:PÆÔ O4°(Uˆ5§£ o*”Eß šòS'7'Š"‡ݤqêÔ¼ŸôWÛ=786C +ô3Bà\qÞx¸ZÅ´v{­¶ |÷ôרŸ IÑ)!Ì‘0h[·uÍ ™äzðU:$dR7iÔÅ\hýŽiÄ¡™ßpF ß@È¡zHKŒ[øyºñèç&::Z™}ˆ\JŒVUN«–.Í]ÐÑ´Œ<ÒˆÈ^è$·™R_ÁÞh£VdðÓAÍj^6Ÿ3%‚ ¹‚À8Ë:>{‘Q—mŸ ,óÁúG•óËzWóË­|äsjY$E¸üö-á‘7ŽVåçi´È£1ÇésŸé³›òD±†fˆ§ò·Ú˸¨d¦Åy¿Ú:½òéãvúŒõKúŽ’/ôBÎé¹Ià;M……´aoCý‡åCxC¿÷[ßç ðz6tîu‚%ÆÈo[)°…Í›¹IRVÒ×}1[ãµÇvìÂöÙ3©·èÕƒA>ÜWrM=د(ćõfHÀ‹ Ÿ(³ä7PŸ7:ùß²ûˆ~ò’ÚÔá¸ÖBXR#’M•Û,É>¤ýð‚M™Yb¢×…©$Üw>À“£êºÞ:$YÅY Lž­µëãÇ¿P-Ëu “¨ C­oqðþîÒÙ—-ŠÀ[E”Öý­Ü‘ ‚öF ˜Ý5J‹F=Ñó“‰ɇãûxC¶‡nnQß`ÿœ´·EòÚóJ\û̯©aŒùH§æÏ¿À4t5|*%iƒÆ9籈’…¦¸’fHzW0öô•>*ÎĦ5_}õýWrú ÷æ{2±Ñ/16Ð#L«¨0öW$S¸ ®anÈԄ'Xtʉ;. ³í%a6¿þHDÐ3‚Þ£~Úhö£G"Cïܰ•nu‹ö´`á*i 'a=,>Õ¯MVÊ6¡ª9cŽˆÉª7G-Ì»]`-ƒóH‘TÛ­Ýl(|ªÀ7',Ž?¿ «x@Í8ÙoËÜ«„Õ§¹yïg þ…-ŸsÄ—ƒÊLnRÛ„+4©k€?qHP{é6'p‘†ò Û?D*슋_æ‡W'^±Ì’;¥:gLª‰WïÁöÎzŽ¢>èÍ?Ì 1³RÔrÊ'~WÀÄ”ezÆ]H”=ƒb 5{—‰%Ç{F•& ü®05`³ÞdJ×.† M_Z…³*¥eg¢à;¥ªãšf¢ ÛPå–¾o*ld¡`Í©kßo¿†½CTšS8kžûoýTºWhþq©ñx"Œ!çuneê¶…õˆŒ$Ï›faØÔœ%c`˜"³±| ÆþäÖ!¶ôÞe ‰†Y¯_dÛ‡_¼xÊèĵ½ }…¤ÕÂÓ!¥„{®r®c\ºo,g$V³AbY2ïx~ø†•Œ„ébjˆ°‹®¿K^á¦rèÇ~†IWb±Â{‚ ·‰ýû Ù=» ûB÷ â öøž…hŸOªŽÄ˜?+¨ïË0ö Õî¿ÚË~ÕÔïGøfþªn1­g´®ÛhtbBT]ðœÈYXSJ­È㎡F‰@/Éç…1JÛ2Wþ£ìãf°Ø—J5ƸÞõ¦nr}ñãG¹‰óØQˆ¡ÿ­®Ý(q_iû^ohzœ¦¨ÎbpÉþ¡§ Hq¢z×ü’Óø¼Zyÿª&‡vKŽÏÔªœ½8Àœ Y1¾‘]\<Ž_éõ¨¹ÀGÜázQa^Á®ùP`üþ°7öý½ÂX{DÓ5•¯üžkVñGü"Žo9šzãôLùFÁ&G¥.×q¬GLM¾#_‡æzLgk¼w*2ÜNnV[Öqǽ¾^† Ü£L@'O¬AñŽ;ÁÇoô'Ï4 piJM¥ð<œ.ÿš§y0ÐJÆïý:;/T°¶Í›yõ‘T¹P,¾Y‹ŽXW|d0#B)ûHÍß阥þÐo‹×0ùµ» ~W-¨Žs2é†Ñ—;<q~w™±'ƒ¯EV¼:Ö›u¸5D4"u]w ~¡šÀIé4"Ê[»T,®mRìŽØ|^ð6çº ¦Ä“âÍgRjwù^44è]¬ÏVäI‡PZnõž “&æn­…wDT !2Õ!¯gš:¤õ̃Oûq¨Ê”»=Z¨d±ðÓÓ^Ž cÁ‚4c)¹^|z¾€tŠ£{¡¯p3 yû²6n՞ʗËfQsz$°.˜m^3î;ü3–P¦E¼É5‡Ú¨j<'XÞYŸÐ²t*å8¬qœ[RS8Ñg¸¤?…îqÓÑ®ÛREÊk«Ì£mËs'ÐVÕ[\û5û‡,þú4ñ©‰œ)¬»ãÅöZ2ã‰n £Oaô:€›¨æÊê­mF*,j9Y#u1\6{«v?¡€žàl§î,<—àØ.nÍ0͹m„/Ç0EÁ»W2a {†fœgfMlÛE`EëÝq½–ó7$öxßÞ×þ{ ¦EŸf>üˆ~jï¶~™™gw£œcñ}w¾¿ÀlÜý ¡"EМۚm– ÙQ«Dù¡d01ê~ø:‰®ˆ”1Ñn²û)–ANÛ´¬ý <Ö£pô½¢¼"ž 9ê/JiÂ9ä"”éz²ÖejÉ8ÙîôE]ê…9Íï+_àó{—!³Š‘øk ()Íd d^¿a"[p5Û bœu^gc¹C—‡ö?Qïßè§GÔ3ÿ:]˜˜¢P¦Fi6:ÀÈ™"zÏ#‚h¬°ÿl7åÑnª‚ ÓW¬E-ò uÎÊëµ¢³^ €¿‡þfj´)õ“‚ ÁOI:]¥xÔDœk4Ñç 2Ô·ê R0eI‡ûÜ,VCÒ>ZOq4õºº!—ºçÃØœg™"Ù ùøš¯ï°J1s(œó~\ò FhRCPü©¡µÂ‘_ójîpV§¶‚Ã¥„A%2»´ |tSû°‹æðÅqsIa®{ v5œ…)/ÀqøíØÖ‰.%Béfý;õqþ IaNNÐ'¡NY[Ô\ÆvGžÉPѹscÑjcýÖF·Ë\Ÿ¥¸P¯¹Êò';c'Ø<-Ÿ âË¢Æ/u E•øãͧÈÉ/ø•XÂö÷r‚e#ư¨ãb|ýs>GÅm~Õà¶,®’Y¹³ÕïU§r`x“>©·Œ=ƒ<µdðº4ÀVå§ø³H— Wæ8(ô®DÅuBì ¶"+b¹+.Pa¹dÍßþ%àª>£‡†4íG ¦UjÏ*M0ÎrM8Rø˜¼hËEE~{…ðŒïÉí¡‹„«0YwÏaßÅülGƒþkŒ-k%ƒIÔÎÕ·?p6ÛšöÏNlr‹ÏÜx'î_›YëŸÂ¿3¶•ºl88;7Wð‚w-¹®Ä‚¾~“jKÕ̳v™mTà.u;—[åˆÄTÞVÄFÊ¡jX¹¨Õ„½„m—È!è½ÎèÆÇeåÕ¹šTfS7¹¢…)j‚#HÂM}›rô‘å½ÎÃá1šÚ¨ ©N´\®KWìQÿÏ¢iõfp;Ù8þ]Ê<ÅäIhÙ'7 ˜¬µz…øù+­zj¨ l²f&ÿM¢’$6]n¢«ðcÎ<“ŸO"M0•†}KUdNàbc”z &©£·r,ú÷7Œ”‡ÚOªMã:…kzÄ'9HŒn,"86„î¦x2+kÛ†îʽëËã³"[¿>wêˆÜ³0‘)a( –¶¤T„/ wû@ò®G2›;„zëãÒ†®Ž1e¾aUTRDw‡þ£R¯™êjÂ`.ír'ýÇ ®Øó‚âw™ß  &Jñµ‘¹»I'¿ºÄêh$úËÞZ…ˆŸžqº­'! ‡"”ƒ ÌjD¢§m†œ m§Ë|^9æf¦&Éþ:À>,Í\Ü9 -ÀCê²Fyx½SÑpÀ3eäݳ¹™½Qùu.½YÒE:g&,³8G³|”«4HݳPGÕ®Eå°|‹siêKÚ×§Ä<%ÙçûöUÑ“Ajµ³{êŽ>x?ÎÖ©²ë˜£âà²1€RyÛSÍ¥RºøRU;kHØ†ËÆmAL„bË ÷ã”b xó<5ë0=Ü“ädöë`pëmÍráªü!~k\[Ȫ9‹{ûRÉé©[á14ÞD†*Û_ o(¦ZŠàْƈ¶éoæøÍE¹{t5×Á®ª!0V­¶™k=#W“áHà™øå²í¸ÝªB«ÝÊ0¢(‡¯QÉ%èSü[ÕhæPøP¶ì$²†![²ŒfªÑš•Îdù¨C€iÊûôoϲýPUªÖ}G:‡êH#¾1ÏHöë -d¯¹UÀ®¼1šãÝ1;¤°~ÔZ¨}×-­`ÿíÄ*×9p\kr®ÜaÔKA˜œ¹3Á¼ž‚?.D ˆ„¦Ý®ˆ‚”¥Ar£=¶V2:ߨÂa|•„üL¾ëŸ“j $ß §3»¸ /^d\K&Qà juÙñÏáÜ×× “ø@‹®Ÿ^ ¶ºã°Œ»f®df"¥ýQʆy‰ëzòš¬ mWà° ÙçCé‰hÆr†£˜²M°4W=s Ö,WtU,‹Ú†ŒRKÅ$®ù"lÀŽˆE]é[ù]Ú/•% RÕSq¿ltÒ2ý(o ægŠB峯”¹¿é9º‰É¢¸‰#A‡ò)àˆžWcòÚÓ=Áxr›ù(ƒ3ŠAúÞ·+Fù!ƒ&xèž>6IÞÇ耊ÕXHr)­úýKÜÄ\÷Ì·èˆÂÐg¶\Oóº‚ÀŽÞ§Ž_üRš$(m–ïNI21¥¨®)q¿{æ!ÍxÝ“~ª¬ UgNtÞ9ÊÚ¶^H}œH«½ü¢…ÔŠ´ÿEÑýŒÖN.Æ|Šnz‡}ý}í!嵆 ª]fÛ*²¶}èèrXr º³¨â~AÖñ{¨Î»¼¿¤,l˜qý|±Ï&q¼ŒÓ¹U-íp ±ÞÏÈïÏÈUÔ,ÔËô‘€>L TÏûÒÞ_´GF6„ëbÝbTVùútŠ«¶Œ§`Y,i?—Ü'Ržýƒ×Fí R"Çâsüœ¸àÅýäBM#IÄ%ʧ%6öX<«h ¨Gê¶5ç©ùoZ#»´ÎÌi«]>®v+æpîÝsæôïHëYÞ€×§vë³p\úf¿Eíˆ-ËÆvȵv©‰Lcíìôçò~02~OR%=a'j2ò؃~Þ™±Í¯Ð¤8 ùöÄð8wlLQvqKÓÚ\M^X±ñµ¸Ü?NŠ[›ô¬ü¹Ì;Í»«©"Kó*Kq'’Þ˜Ÿ´LWÞÝ’¨Âx¹HŒðú•Žöz\÷ #3ÌG@±]7–°½Î;…Ê3·w [€Ë¼õ ÔG, Bbå<`Ha±TÖÏü,LD#bu>8Ô†ðÒùq÷‡™ù "ï¦×¢ ¹Ú¢ˆ`Z†ä7¡ÐÓda| ›Òj1⣣¦Kñ ¸ º˜€Mî>Æj…¾öŠQ-‹-áÛϬݷÑîK×@éœÕÝðKy$Ôiíe»v³Äïk«røÕ”ßj³d·oó²~¡ ~×T¸îº®êLG‘;_f—%|D_&”þ²r|Ó• iå˜k²}•Ú $ à,¦‡OKŒÇùa¼ä-îY)oË( Æh¤ÚJ>B)t«~ެ¿Š ø 홉îÂî/Õ¡¡ÆL4”wà1Þ Íb„+|•ä/rÕÌxVU³q3,6´:B5tóš»Fú¼:è,><·@0BO\›;,ì«ù³û9º<-µã#…Óö«qB;°ážÓàŒ’ÛZšã䩸Ún}œ¸¸ófHZ9Ju4SÃÓcT› o:Ñl whñ·©oëÁ§^8²¡Bë $­LLQîìÝÀ¦Ó!åJü€d$VÞŽ7x$ẋð$,ðˆï´f×voH%'çÈC\ŸMŠ‚ê•xoð:Å̲‰NUú¼ƒZ˜¶Sã¥Áet5ÎŒd)*‡ŸPªÑÆ`k?– o¾^]LÁÜæºàÅÜ¥ÈØóùŒ*€µPkž7/¨íɱëźl`¥Š{¨O:z$•Õ-«L_V¯óº¶G¯½Å:mù¨W)<…~Ž4)锃Çù>¹Œc-W»P{ê ÷þÒ˜­ ²š‰ y-2'@­ø6’5ÿ§83g7CÒ6gb}!“1:ðA +M^ *Kj- ›éƒ“9-0õgEYè°æ V:wð‹¢³ROW”ËhògG›—’ŠÜÓ…,"\· JÏKÞVÇ+*NP‡ä×M:§0#¢Íî}Æy΢OHàPfkÝœYs×Åtät§tó_¹MZôÓZHúF®Ù:GQ!ÒF!z{;â¾o’hO€ãÜÎîÇÎ,"d‹3ššÈl}t» à—aŠ_2•2^¾È#þšSÚÜë>æ°›OŸù{¬..ˆ÷ßñ ¹ÙÞ˜T{ëó#õÑþ°”N·aJÊÕUÚbäB¿ðŠ\ä4Ž;B-e9˜.øá£·ÔÁ«VO»Q–úd·ßÅúÍz¿nW»sw£JKZP®ðE~øê{^dœbÒ0lË‘­>¯ÇÀ5­¼õþÁÉNR¬£ЏŒÆCÿÞg ã5,ï­S¿šÀ½…,ÃõG¡UÉ‹ ‡cL§ Ie‚L 8+½£ºÎ÷ŒÑ^}¼£YgHŒô&± å­EW. ÜáJhõFî6oe…h¸?m$¯†èácwÌ0dñ>•y½`6ù¶Ö -sag°½gK¸ã_H[vÑŸl·z—'õÍOOƒ³„Kêôý³ ´ÔÑ3í`´sC͉KºOzS P2tõ‡Å´¨F€/˜S"Úlß‹8glêÄ?à íú$ÃFèÓ¢*L<¹‰©þéá5tÛÜÖÀ¬ Üô*ºÀ¨®)û…vøŠ9³6‡Ûê\«`=¥‡³©wzÏ­ðïygýÔ,9X’ˆ¾,ãa9Gh52ÒÞ†¶è“!6nEÕÑ–z­¶H&àQš-´CˆˆsóÎŽÖ‰ýDáM)º(û]x‚ÀzÈ#ágÎÏUâEgdöX¡?ì2³ÌÜݶŬ]æ‘pq:òï9ÖšœMóéfÂÜñ‹QÚÏ EÒò~Z]«d¡îìŸeP¸ºÁå9LÜΡÍV®iV'a±])}S´àîåvƒk¬¡OîPg¤X,pG¥x—8üÕhÈN¡ªCKU'‰ðé{ʑ۳±4Ûš}Dæ÷çïÜŽã×Ë&soÏ~!~4Þpá ™÷;¶/ç°úÜUÐXÞöV•›õkk;Ìoà[`Æ»‰«Eÿ gÊ» xZÞÀ¬Æ°‰_ÅûÕ–m.ßéÜ´¡p(ÅáW‡êPìeÝÏØd(m¨;GôÎXz|vxí4ü¼.o ˆ¬6ñ üÖJ {•) €=J؉qEËÉ®1ÀéÁÑÇê¨&„ëY[ÐTf3ì_ݱ»`Yä QÁ¯òbFí”o´ÅAôº#[¡@•+KÀñ³Ê×õ#Ä s$ñÉa– ÑÏDgön ©Q|ÎT±¹Pï ’dš‹ûA1Ýë»êê>¨dJ˜Dí a¹– ðßIºTú©E“-_Tƒ¢H««9¬)œúú›ßv¥^]Ssm,.ë<Œ}çI2šµè¹Ýê<ØÈpä2g†_¦‰Ü1ÜᯠÉ.5}ê†_Iä]t¹âTóh¸¢åáaôË6 Lù–Nt ¿½h¿9ˆâ“ß­ ´‘ %Xap-€Â‘iß}LbA"ÿb¢A·e‘²nÖå¿vØ×-F®^’Ä–f;­<ˆèõôèÌâ=7ìÓáYPwÈá¹a/Ÿ²4FÃXžF£"pz‚pÉóðž{„z£„[grxAJê}6NìrÝæp!ë$8"îo2¿‚¥ƒ‘~x¿Zž6FÒ»0ç'è£6 |öBl¾ÆÂŸ·Ç‡ëÍû@ÌúcªðNÔa'†ðL˜uJyø'fntpãCq¯ÜC\9wår¶ºÍ9üâçý~déTÌ«é Š‹9KNja¥}“ ns²ïP ž&4lÆcÖ*h*¨âQ×û‡GÙ·}׿eÅ ûRÎY6š–ô$S¯Dª”ç ÎÏ–¿5ûPIð€Ú{xÜì4÷ŒþÐ0l`¬‡¬gÆöÃ—ÕØüsÆÉb,3¶‘çsA r{gÈј+¯ъ묔&örÙNýM`…"óÃsTëT/ž£û’²HϬ%vØf›ã"£vÿ ÅdK«*¸ãÛáx@m2V6¡'a î¼r«$Å,Tèm': óÂR^®`âu¼ÿ(Òóikòª=B…ùüKƒhj ¶‘^T@Õ¸ê¤ê›àãlO…ÜË`£« ˜& û½¿ý bWåúǸ’×Ô¬¢1RúŸfi)38[ÑNl&+¨PStöùÆòùtX%eÞáh?¿TëøÊ^#Í/ç\ '1ÖEøZ¬ztûíxÖï¤帩㾤dðúž$¯ô×NƽIËÏX£6ÐãÞ¶×$SխÓ… ºQÈmTÞ‰å:»Ä%[šñ^hÆŽ[.›ãºíÏ)ÁÌAéºUƘ¾”èæ¢äò©î@}Wã‡r=ƒ³+ýŽži€^Ûº¼\\ÒÁøâæøpS†GDâ¹…u†˜*5—hŒ¬±•Ѝ{Â~éÏiÖN²Ð_^¯%Åæ†`%º%4~Ri~×fÕ /x˜§o¤MŰ g¼MAÓocP»~^ãrÁ.9N¶æìEß…­T/¡U†ïöJð.¸¤-ÞÌIß7/f¿í.îk™‘¨‰•_5b±n}³Ç— ÐݦÚ–¤1¾.Å\ˆxÇŽJ«ù¤ÄÌ ‘“,Ãm¥½Ö·”Jýˆù÷Œ?µ¨Þqè4ÈÿX ŒùÀ"•grM¡÷¡åáÁÁ·¥ &r@õcÚW, íØ™z…,ú:ÅVØãø²/tÊ!_ÿÛÄâöå·$d$”e @Ö¸¡=¦ž—¢tÃ@9²ã=ÛÐ+¶DqîÌœ³Ôvç¶Ú‹oö¹9CHY¾9ŠÕ ùžJ-xl´ñ|Ô'«ÞžòÈä·¾~aù%@.l«„އ-|.‹úC+Ž5É7HÀØ=úc=mmMÓ†Š!eq”8@åØîïÿ-aêFҢ̒ÇeÉóÂü@éI&J2ɃǩªXVá;*«ðºc$pc™Œ'r ±6ÌW ïúŠýG™L¡çt¬&™Æ;P'iÖ©Ö}ìpkˆ¾Ws´ëR `BÑå˜nC/Õ‰Á‰•r°ÆÄÿEî1~ô‘~Ì@yÀ£”Äzn=¨çÅ}”|ßúFLl‰×»Ô9?P)eîõ<µóœO/ÅØLùÃÁÀÿn]öØ‘’¾iüZCIú®ŸþâãOðEý’cãŽXD ˆ„W¬`Ó¬yyŒÿ^ZLî(4çY½#†8ñR[ÿeísÌ7 îREÎi»Á S²2£î¥RU÷÷=»kÜK$Ý¥§}—̢ЕâîD΢1ñO%‘^jåøËšzë­LË‹m}»¿r¯¥^<ÇÈë^¥­Ëà)}ØË…ÂmÅì Æ®®ì—Ü…·ú~Ç-dôƒÔ¢mQcíû-=rM±Ÿõ·(>së:»†³Þ…R#uœèöoàm¹¥ý ñ·ôõr­MO\¢M†Ô¤cdµ¾˜‘ŠVofHÒŽ£³Ûk;÷ð·-“¸‘%š'¶¿ÐÓ‘µÉFÈÔRšÅ½/ÊÃï_¹{,[@W42@ô!G\*Xß1·ùôˆQ·¥\ßd–L2¸L猀×7øl\ô«çúDd™2å|×y£hZªš¾á‹ô]Äê+MdÖa”ܾù T“ ¸&ÑÔ Ù¤Ñr޹y4}´Ã ôÕÂô2P–Ÿuëä‘0§ßªÒƒî©+^7Pç¸÷ýY7ÊʘÔ\œ?jD>¼’èW¹Ñé0‚Ówpá0;Ÿ'i#KZ¬0^!¤°ˆ\OÑwŠ#è~¥à»PV¡ ‘T6’€¬+$Èdu‡0Yè]M*»¾‰…É‚â1pàrC?‹WjmƒY¯}¤ŽTÍL¨"³\ä%0ËÐÄï­+Ø! Ån~-•2j@x:%M„i*â(éB÷SEW‹ËÖ»âEtÓÆ9ú~·]¹ĕ´ˆWªù´IK€úg¹yª§Ú–c_M15^ðSNtt×s„{“õ¤kN`‡¥O¿°ô!Ìç½Î{$õƒ¼åçÂÓª¤Ÿ> k ·ü}›J)¼ÅÈQ´³ÛÚjI¤+.=âY½š$Ó³åðMÃiP~ùWÞ(Ø–lY aC•û’)ö1>iõoÕùš:R=,agxtœL ôâLgš‘ÔîÓÄ×ÅŒ…Ñ)õð5Ï!ïiJ[ëí;Ëà`y‰¢"…ujQç÷ïiA’ä\½Tšñì÷¨Í¦zÀ²tx ß@-ê+ [*=á„ocõŸlºªx–—ÌŸì7}|.€ÙÙDø¿@ýðÞšqòc«Àk ™€½Âó}RJú‘Ë §}Û‹`ö½àëHÊgeIž@¼èƒZtÝsžâ¡rM¸œ‰á»µ&aNÿ%%ØO°`bηÇ-4D#únp KÒO¦¶A­Ä¬;Q[]¸ŽMƒ<©ÇóÅé:S3ï>¤°oxÒ× ¢ ÉUÖ1бlvY`Œ¼RòY$ÚZ5æÆïƒ^¥ýÄ%³R(uºHïÖ=Ѳø†jX?&¥õ£‚aé6QJÉÙùŠØ\¬†Z ƒ‘3bòº°S=W¨Ùlüõ‡9[ïšæ|»DDÌx{„eÃÊ#ve9ƒÏ!·â9ññ³Â-¶Og²gh’ÚÁ0t‚âRÃ#r§U@Ç·hMg¿|ïv³Y`óžŸõÂÈ{*å4µÑµy'=Ó“ëÄI—ú …S"ñÿÒg¢„'ÛT& ¢xi;Y4ï?›Róœbé^aÆÄ™z¤³<ãê-žÖd 7}H%þQÕÁ jÀIÞW,Ü&¨Ý~¼èˆx½³tìMã"öe¡¼‹®<:5!FíG×Êê½\åƒù•u|^;É<Ž"³÷òw‹u#[ýÎ1ò;Üse¬=¸1 igkŽ‚Šçµ·ºo>*ÅN%Ð{Â=UÙÆ31èUr¬S\î×¼c\‘IIÜz7ÉG:q˜:ý™_NRÉoý–NRbQ¤Ú…¨+/¥?iXQ/ã`ä2¢æ·ói&·œªÝò“3Ì üÁB*µ&ùp‹FCÚ¬#Íøz’×à‰¶å¥S vM8ÄÌðýzLë•XNG_›OޏN[i¼ßŠuJè™Ú$GžÖ»Wiýã¶ôƒ&LìÄx«c_11ïšÛ£,™¯”J ñ;)bF‡5Adiß·:p¦(±°ÛšDeOã‰^-&2ÿô’›ñJ˜ôŠúB Ùk'‘D㫎ö—û`:ЉoÜ…êšÕ‹}Ãöq¥GPÞÔ=DˆÒdWØÆPT `(›˜g.´“ñ€Ã,¬ÌÂB("©¬"núÁlÄæ„%&e°2CÚBp081Ç6 \.Í ¬È—°²œÝßâ$†h. *úòç¸2i'T çÍÞ¾ïa«ªapÕþpE®TÁdád'’/f(ÆpK×óK —Œ‹æÛÿ¼j61viïJ¶³Gsªd\QòœLCRÐ3Ù?:Ƴ“§1‰û>&lwKv˜nßyˆÅc¶KÔY/pOÃó“ søV©Ø=á;<¹Q¢-hÕ*±_²1•çºõÞ½ *cücé§ËK\³¬ÆÕÝèïÌRjÀêïÕp«$ ‹ÐdI>hĬ~†Ä8Özu¼4`2K™Ÿ$]Q1ða©£¯¥NÔ)=Ç@Pò8ë(#Œ1}ˆ"N›Â,Êâ xÕ¹æùŒfÉ/Ñ­½€·¾çHSOêCàAñHxÕÈïá˜4Q3%´¹i/‡EX1x‹"º¢Ò­AÆ&Ä7¥ ;º¾Ö€‡Ž0|„XRþŠè0<ÿð¢8g¡ê[/˜ÊD÷ì´‘ êÐǬó 8¹d*bç'•UªWþºRO-% Ãm^{Qäkl̼s¼úÉŒ©Šß̶ §wUß]ME´™(ÞÆŸaÕÈ*$­*îçs8y8ˆ.Œã4Ó•¼³0ìó tÛƒ3ÆÊTÐ]—³õ] Ôò¯Šä½z;fŸÚó¸]­b/ìñRÃÙm4gérkRéþ’ñ–QÄéZ,¹‰´+\ŸÄ/¼ÊöìÓ4âÊ&Qv6’à%#nž ¯Tˆ•N?$¦áéCÃÒ‚}7×c¢O™[ºúäN ,ˆáòò~Yî;Lx '‡‘g ѱU<øf ò]—¶¦ºÆWYUILP&01se­utg«½S§/þF¾½o&O3¿»¯ ƒò– C˜4Ç0i7™֛ŒˆË¡[„æLáFLÏÀq1•·Ã¬r݇xŒ~û„âáîH;AE˜d9G×Dmçuç:‡êDá­v YR=Ù|Þ¬ÿÚL{ð: Õò­Ôx¶ª„`K°—Å/¾ðˆ¢†Í*æhÆ2YÀù€[iËÍ bõ·“æãýN‰*3ý4­³«J·Ë” ½P9÷éT2 ¶·“þ0n»ˆŒ²\œ; 5ïäÄýòk)Ú sqaçzñ0qcF†³Eì lO$£´Ã•…²<ò$Ûo¢ðÉoên~!4k [ ªõ”Z /¨Æª½E›0¡VM©Ö¦y[¥ÓjH.1ò%c*¦ý¡$y`hüUÚV([ ]q‰z´¦rKŽD oóÉ|{Bi«NÕÂ\îC¹ÊÊÚ®@ÓÇ~csfáw+µñ>Ã@‘¹+ð¼êZú=Î’@?kÞ’Sã9ú×´±Ê©P„d•OW¯:[ö˜Òš¾kg³ìÐúÄ8?³îS¯e¿á=od­Ðü=§D1¼¥ÉM)WiËôpQî$õŒÌ¾¶G@jòËm¥ýëӪˌ¶©GƒÞ×è²J? ~íº‹;ŠCñx‘IZé‘¶%æxId- ‘ټ9¤5³¬}<ã½Òס€Ú³X÷ ¡{‹÷=u½fiÚr–Q'ZÜÉÂ6˜š€c¿ƒLÜy5ô¾G[ûÆCà†ÀMP–s‰G·%ü™€W–Füè W~±Dæ×ç3’@,øÜ{ö ùs‘XHÌ…<»¡ÞÕJ™ÇÀLߥÛ P…­Ómu$¯¾Q §ÅÓnȼm*Þ<;rücòt ÂàSïM¶ÁE_doUäz‰ÝVÉæ-æ¬QAs"м £áÀ~AÔë[m›þ¥aû Uæ‹õíÒ+"ñ¨/£¥ vÅßÏ«n±ì˜ä¹€>IV+÷Õ¦ø¸ʰügXHwæ³§…"mDXgö­š¯MÜìtÑN‰™~~«êÚ«À1Ÿ2bÃ…žœø30ÝC!_n•^ƒpãUâFÇ^Ý!eÊà Ý÷ðõI§·ÖWH ~l¾$IeçzÑF‡Ð_¤K§ºa0¤Ý)w™/í®ÚI’„8 o%¿š Š‚–nlPéò…^ð3¤È&ÒãW€Ä¨MªVßX?ÝYö(ÀuàN?y2p¸§üzþ¨ðê: g'Ò±Ú÷ÍüŠˆK–!Å8nNòòYšª/ÿvµAFZb•N ¼¿)Vê<Éͯ[çÔLþ¶Ç-FŽÅËh“µ«7ë<®“Ÿ–PéGv³M8YƒtãÒHȶϕB| –h¢96÷¼Õ—HÄ' Û§fX©¡Y"þêÿJ¡h¬÷I0¹Lêÿ  €ôf8TöÄ¿×xÁäI…?©¶C^Rú¦Àð*ô#!¸Ýñ5]®õz¨}ÞCßô`F8÷ê™_ ¾U©X‘¬Xs?¼9 EsÎßùWcÅßp2«“Èe]ñ«ß9¹ ÆRñÅó}¬ïíDjgnÕ]|FS%G_(c°“R¤tÔ¿$Âå«jR2é®ÝF…ž¡‚E°‡O  ¶¡TÒa^{Á!ÈTáx•ÅŸÖw/$Íæ~ùõ£óR…r Ø„äžò²¸äùÒ V,UmyuaYþî¸|´»¿óx÷Pò§³;’&ÖÖJGJòT[-µÆcQ©ºW͏÷§ºxi6{`†{Ù·˜›?‰ßŸŽ®ÏÑ÷+ÿ«œV‰Á íy—”ÅÜé§PŸÄ1^Œ•îó*ñ×½4Þ«ƒ{¸”íw-#èæôzÜîþ«~ë†Ðô ¥•í0ºNËhø—¼ºB3cR{yÌc|\U9>wrU’ ¸é#è*‘ê´à7Óe4žëÝg8ÀJ«ö;…*ÎׇahN#’;ú&¦Ë'/y®È8u(yf`Þ§5Øÿk€ÉÏÖh<,y)¸À/ Þ#ÄÎÍ.-DR?û­…|XaÈ9Ü¥æÏ!f²ÑäV–TÞs,ì"ÇÓÖ¬R»} öi|±„ÏÑ#=tãØ»ñ.[þÕ¹ŸÌ`÷ä<>“v)%vŠu ¡äzF 0·ÁûL鯋Á ¡[·ã~UÚ«z8L)ÌNÚ,ûkøëZ,qÊÿ¨›€=+š„ŸÁw³Q-yqÑk ªQ34{v?\Ô7*RK38®öõä-y”Kë`Øys…¿åê·kÆvä[‘EÆaAʉaØ6ÓZ¶( …ŸÓØÛûudmš-“MÊ脵­cio4Äïݸóì4HåQR$ô„ -jNEeüìÍb/½Ä±`œr7kÒæ.‹°øÒ=…(@>É+—Y’R6ÂŒ 2ãP3ãòsJ  d² r8ÀbŸíÄ3"Ê9ÀtÅàYøåôµ71ýväf‡`d›n^´ôê[zJŸ%㣠ó^?÷ —hy"ÔJqŽŠNp…x—¹?s8Âvóø&X ˆéºcu¨ÜvÒsJHžž"Zç[i»JÖëh‚rJH—@Å«1zƒ½ïF“U²NWû°Ï Úm¹ögöôêéL gæÎ5ìœÚZ·8ÄU¼OÔå©* 1˜…P$äÕéWU× ›„[!˜Óñ_C|{À u¹új¹ã`f6öàá}»ÕþœÒ“œ'(§É~w ƒ=]GÄ:%ü³¿riè§EnÅ’,¤Ï}·KÇ#¼Žsûªš«ÝB×>ˆ¯Â§Mµ‚¤³¹cÃàû‰>ƒ/È\^ý‹íÌls«6$‰…ô*þñ„Š TM»=D·Þqö“yÖøìt¬ÿÓXé£ù[Omâ½V|2–ϼñOþ¦)Î_Qîò:;¤Â»£žÓS1¼™ 7FK[“¯%«WŠºc¾Ï`š<Í]&Ž«D*Y·&nW”»° Ëß¼Aê]KÞ]·DåmøXsÛ,ˆuHÄÄ=›Ò–lv¹Û *ð‰.¶J¹‘çÝm EoM³[sÊg!CÙM3AL‘´îaÊÿt­³ïþ Êë„8©kÉG5å…Ì(.|!%rÛ¦ZA¿¶žN|mb×Ãï*”Æü³%Qž/MX½öqä¶?-)žµÜ¼¨LØ$ÎLø°.*Ú†Æ:0ª„¼ÆšÓPT<Wÿ;Ñ»A7x¸jd zZ‰šo*â¹Z]ðçΡÿ¾Y.é–ð…ì‘ôRÁ)ÈÃþ}«n°¶ž"%ÖåçÀ!œuå{¯ÝÂåù¡¥$¼J<|šY€D  ‹]²Í;ç_£Ž{ ³uëTè ×–´H;S:ª‡t»¿'°‡9Ž|ÍÔÊORö¸@'BL©¥Eœ«Ñ¡I¢K”HKý/›•Ó$ˆ'‘ÑÎß-)»C( ÞÇ…êU+tRF¤öˆ=z¬IR·V= |BAZpG`˜õ:ñÊE<¸,Ш~+= øyâ:ÜçšûÎo`rÕH_oVJƒhûO×™Ð.û)€å?¡U¸+S…c«KÌ­ÚÁ£jz»•cÌ%? d¢Éf¸À( †îýWuñJi­k!vJ' czÔt@í=b‘ÛšncÍ´`CiÒéš'è3?„w"ºå±Â;L()3ò.<¶VX!G¯‰³&Ç]Ø–¬@}Œðƒ]ŠYö£…ŽT|À‘­Ø‚2ª3 Y ÷›ò`SÍßç#.¨Rè?Ê{xËŒÀ€û_`ÇMP¬O´¶kÓße_«_²ÜMßlB0ÛäLÚVÞ–2.§µS þ9j4ÂYç1ª¤Ê!7n®Ý­è+7;E|ftU}»Y¤Û{a…¿¦m Ù…)¸qd—û8@î»2Ý?×Å•©H¼JÊŸyZÏpà ³kÂRãÞùÌëJƒÿþ´Tó÷™É$'®{AƶGÑØûOZÞ‡rìHã1 ÏÏÝß¹0õ{hwy]X£7q ùêÐâH>Þ¹¢¦+ÂóÒg%ÕGʱ ˜ú„½âÖ®{?£WQEN~ÝîûžìVÞ ©Xu$J·ë;d'ØŒÀÔDõ«"áy}*9ĺЦŽ÷K«š|nT¾~ y;~@ZÝØ5j9#ý×ê7P20$Œ ßUp»˜sh u«‡ö{í§^iÐtà–³ûÍÔ~%WÜd5/¢|#Q00W¤Ž»½Ò‹+ƒu°®Ý¿½l;œh®’P¬P± ·Û;Ñ#çï}E`'ñ²Æø‰1^é,g\ “’S½â¼ðî+U÷„Å©5zm#Û*ÅÃr¶[Œ?~ÊPeÉ÷Š[iÒn2ÌkG)ÎÚ%O·?/0ÃSãTÇ6Üõ8iíél77ÅëÒãbÔ»×dÙÞ—óÖó³pÉÖäØãÂQúïkpÍ‹"òyZæåCøÿ{$ž11†:ÓÅy79*ïYgsµ¥/–šjÞlNN;»ö^£Ã»XÍ“ÁÓªpx†JƵ“ppMùfiæÊW(€Ý©v%êñ1“øM@!“IJo¡‡Á endstream endobj 162 0 obj << /Type /FontDescriptor /FontName /UKVSWC+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/C/E/F/G/L/M/N/P/R/S/T/U/Y/a/ampersand/asciitilde/asterisk/b/braceleft/braceright/bracketleft/bracketright/c/colon/comma/d/dollar/e/eight/equal/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/nine/numbersign/o/one/p/parenleft/parenright/percent/period/plus/q/quotedbl/r/s/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 161 0 R >> endobj 8 0 obj << /Type /Font /Subtype /Type1 /BaseFont /HAOGWG+CMBX12 /FontDescriptor 138 0 R /FirstChar 44 /LastChar 121 /Widths 133 0 R >> endobj 13 0 obj << /Type /Font /Subtype /Type1 /BaseFont /OREXSM+CMEX10 /FontDescriptor 140 0 R /FirstChar 89 /LastChar 90 /Widths 128 0 R >> endobj 11 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GZDKMT+CMMI10 /FontDescriptor 142 0 R /FirstChar 12 /LastChar 120 /Widths 130 0 R >> endobj 12 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GESBMB+CMMI7 /FontDescriptor 144 0 R /FirstChar 12 /LastChar 120 /Widths 129 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KHLZJQ+CMR10 /FontDescriptor 146 0 R /FirstChar 3 /LastChar 124 /Widths 132 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /CJOFPZ+CMR12 /FontDescriptor 148 0 R /FirstChar 44 /LastChar 121 /Widths 135 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YIARWD+CMR17 /FontDescriptor 150 0 R /FirstChar 45 /LastChar 117 /Widths 136 0 R >> endobj 14 0 obj << /Type /Font /Subtype /Type1 /BaseFont /QBHYSA+CMR7 /FontDescriptor 152 0 R /FirstChar 40 /LastChar 59 /Widths 127 0 R >> endobj 16 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZJRRRV+CMSY10 /FontDescriptor 154 0 R /FirstChar 0 /LastChar 54 /Widths 125 0 R >> endobj 15 0 obj << /Type /Font /Subtype /Type1 /BaseFont /IQLEKP+CMSY7 /FontDescriptor 156 0 R /FirstChar 3 /LastChar 20 /Widths 126 0 R >> endobj 17 0 obj << /Type /Font /Subtype /Type1 /BaseFont /VHACHE+CMTI10 /FontDescriptor 158 0 R /FirstChar 97 /LastChar 121 /Widths 124 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ZNPDQI+CMTI12 /FontDescriptor 160 0 R /FirstChar 67 /LastChar 121 /Widths 134 0 R >> endobj 10 0 obj << /Type /Font /Subtype /Type1 /BaseFont /UKVSWC+CMTT10 /FontDescriptor 162 0 R /FirstChar 34 /LastChar 126 /Widths 131 0 R >> endobj 18 0 obj << /Type /Pages /Count 6 /Parent 163 0 R /Kids [3 0 R 20 0 R 29 0 R 33 0 R 42 0 R 51 0 R] >> endobj 62 0 obj << /Type /Pages /Count 6 /Parent 163 0 R /Kids [60 0 R 65 0 R 73 0 R 81 0 R 84 0 R 87 0 R] >> endobj 93 0 obj << /Type /Pages /Count 6 /Parent 163 0 R /Kids [91 0 R 95 0 R 103 0 R 107 0 R 110 0 R 118 0 R] >> endobj 163 0 obj << /Type /Pages /Count 18 /Kids [18 0 R 62 0 R 93 0 R] >> endobj 164 0 obj << /Type /Catalog /Pages 163 0 R >> endobj 165 0 obj << /Producer (pdfTeX-1.40.14) /Creator (TeX) /CreationDate (D:20150701163329-05'00') /ModDate (D:20150701163329-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 166 0000000000 65535 f 0000003875 00000 n 0000003119 00000 n 0000003014 00000 n 0000000015 00000 n 0000257197 00000 n 0000257056 00000 n 0000257902 00000 n 0000256347 00000 n 0000256916 00000 n 0000258044 00000 n 0000256631 00000 n 0000256774 00000 n 0000256489 00000 n 0000257338 00000 n 0000257619 00000 n 0000257478 00000 n 0000257759 00000 n 0000258187 00000 n 0000008104 00000 n 0000003767 00000 n 0000003324 00000 n 0000004889 00000 n 0000005032 00000 n 0000005130 00000 n 0000005165 00000 n 0000005407 00000 n 0000012529 00000 n 0000010821 00000 n 0000010713 00000 n 0000008198 00000 n 0000097914 00000 n 0000019703 00000 n 0000012421 00000 n 0000010949 00000 n 0000016488 00000 n 0000016631 00000 n 0000016729 00000 n 0000016764 00000 n 0000017006 00000 n 0000021641 00000 n 0000028776 00000 n 0000021533 00000 n 0000019822 00000 n 0000025561 00000 n 0000025704 00000 n 0000025802 00000 n 0000025837 00000 n 0000026079 00000 n 0000031237 00000 n 0000036146 00000 n 0000031129 00000 n 0000028907 00000 n 0000032931 00000 n 0000033074 00000 n 0000033172 00000 n 0000033207 00000 n 0000033449 00000 n 0000040805 00000 n 0000039032 00000 n 0000038924 00000 n 0000036265 00000 n 0000258296 00000 n 0000048224 00000 n 0000046267 00000 n 0000040697 00000 n 0000039124 00000 n 0000043052 00000 n 0000043195 00000 n 0000043293 00000 n 0000043328 00000 n 0000043570 00000 n 0000054320 00000 n 0000048116 00000 n 0000046398 00000 n 0000051105 00000 n 0000051248 00000 n 0000051346 00000 n 0000051381 00000 n 0000051623 00000 n 0000056293 00000 n 0000056185 00000 n 0000054473 00000 n 0000060180 00000 n 0000060072 00000 n 0000056384 00000 n 0000063402 00000 n 0000063294 00000 n 0000060342 00000 n 0000068171 00000 n 0000065822 00000 n 0000065714 00000 n 0000063506 00000 n 0000258406 00000 n 0000080761 00000 n 0000068063 00000 n 0000065914 00000 n 0000077542 00000 n 0000077685 00000 n 0000077784 00000 n 0000077820 00000 n 0000078063 00000 n 0000084166 00000 n 0000084055 00000 n 0000080961 00000 n 0000088077 00000 n 0000086143 00000 n 0000086032 00000 n 0000084306 00000 n 0000096261 00000 n 0000087966 00000 n 0000086248 00000 n 0000093039 00000 n 0000093183 00000 n 0000093283 00000 n 0000093320 00000 n 0000093563 00000 n 0000097648 00000 n 0000097537 00000 n 0000096393 00000 n 0000097729 00000 n 0000098162 00000 n 0000098188 00000 n 0000098249 00000 n 0000098285 00000 n 0000098440 00000 n 0000098757 00000 n 0000098885 00000 n 0000099022 00000 n 0000099054 00000 n 0000099724 00000 n 0000100347 00000 n 0000100738 00000 n 0000101416 00000 n 0000101875 00000 n 0000102152 00000 n 0000102582 00000 n 0000103035 00000 n 0000115240 00000 n 0000115539 00000 n 0000122781 00000 n 0000123028 00000 n 0000135393 00000 n 0000135683 00000 n 0000144826 00000 n 0000145065 00000 n 0000169800 00000 n 0000170361 00000 n 0000180044 00000 n 0000180306 00000 n 0000190526 00000 n 0000190783 00000 n 0000198589 00000 n 0000198845 00000 n 0000206607 00000 n 0000206877 00000 n 0000214345 00000 n 0000214587 00000 n 0000226733 00000 n 0000226995 00000 n 0000235657 00000 n 0000235895 00000 n 0000255813 00000 n 0000258520 00000 n 0000258595 00000 n 0000258648 00000 n trailer << /Size 166 /Root 164 0 R /Info 165 0 R /ID [<49279E25BEAF1996502BCA1E3DACF43A> <49279E25BEAF1996502BCA1E3DACF43A>] >> startxref 258915 %%EOF survival/inst/doc/validate.pdf0000644000175100001440000067153211741354426016176 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 3 0 obj << /Length 1857 /Filter /FlateDecode >> stream xÚÍYmoÛ6þž_¡ V³|ÙPèÐV C}i;@µ•X«l¥’¼¤ûõ»#)[regI3¤_¢y<ÞŸ{xr^ÌÏž¾â&ã– ®U6¿È“Ôq›f©"›/³wäsëH^•˼+ëÍùL2A.ê& ݪ@“ŸësaÈM}SŸÏàmYTçæ¿À.6ãœz­î¢eÊe I%O›ÌÏ EÓ|Â’7Ÿ†ÌWE³)ò-šÉ¨g3.©W6®z“ŸƒòüSǯÐN½Ý$gq™àÔ•Í„¢Bû¸ð·Eõã,ù¨3Ψgž£°§’Á2騑..{ÝÅøÊ6>·mq±­R&Ð&É |räoüSÄ™<>Ú¢‹õEZQ´Éx›”®!àb”ZAš¶ãd{ÎIÕµÇvù)‡—"Ð5NÕÍ'ÌM± ‰˜qC½´!…Z¥LÔÛÏŠ‘rƒO‡Öåeõ^¬F!·Ý*ÎuuÒ­ªíºÜä]߃«¨°¨×WÛ.ä¾Sùf™tê¨ÐY§©4ÃKn,ò6EŒ¡Ûlìp[_t[0É2’\5õe“¯[À–qÇSfA‚ô·QŠAU¹¹Æ1±7ôy[vIlËõUU@&¤eä#æ G2÷NC`u3åó'àæYJ9+á@ÛøÒ—y³¤‰ª°®L¨0%¡*ަ8äŠ ã‰‚sJa‚ħÐÌú¥NÅö]P]h£”Öƒ„(^¸¡k)‹ëo¦ö‘‰^£·›Œ²§<íØy·†$A¾º8´HÉÓ2RTq‚y׌\—¾ ‡¸ÑTƒ˜6ÜL¹ä©º×x†uÁˆIanÚ¿ŠE‡ QN²)b¥#€Íø1(á©7Eº²XFi a­úÁ5bFø‘â:I!œ å_­Œ¥²ŸZ›¶n/œ\ZÜÛ©7Õ—)Ì…Ô'ûOÒià8Ì ÍÄ¡êK”ÒÂ~÷KÌŽµä=Ólž£HÿÌÓ³]çU5êHÐR‰Ñt>Y%‰Á`;.¢¯ŠCv%ZƱ*Q‹“&öAá¢Þ6q²HKÁeÀá×iTsÏ)W;ˆ4Ó 2^@¤8#ÅͦcÊd¨Õþ{ÆÕ”QC­2{%td@îÃjÊöS”Q.êfoÊ@žP«8U*|â‹ÐE!iáæÓöhaI¥8YéžZæ†eÅáH¼ð„{\’ücUÄ…’ j´Ëöóç“4¢©Õ»²nWáȯRÚ¥+„*m¨4v(õæ2Š ÕÚ|“ô6ým¨4åRÔøfÑ‚B¢®#qG[o€€¬£œ'¥"_à’U½ çX—ÑÌ€Ö÷MR²ƒDö´N{b×=\M’‡« H¾\çxKavD¨R.bB}Y•ŸB¥T媎þ„¿Ë¨°Â2EÉÆCõø·³ç\(ìC oý5‡‘ý„œªGØÖô&Ê´E7ð* ØÝ~¦ÁžaH )/¨¨d_ã”6Á_ö𗼟 ìY=µ8}ñãô*_ÆÄ‘KúÓÆ¡Å4‡[o×I'9|iãdêVÎ^ÎÏ>Ÿ!~ù™–:€˜†h‚ë³wX¶„I”Jï²ë ºFuªìíÙïg/Bs>:j {¨LŸ Ũq¼ça¦ªxƆ´u3ÍABêMÒôÌjÇÝ”lÇGLöU¤r”Ad÷Ì¿z Us…厅–E¤îýÏT÷”ÁÅ>ϘL\f©·,¶}p×|8çmÆ÷/t4K™ endstream endobj 2 0 obj << /Type /Page /Contents 3 0 R /Resources 1 0 R /MediaBox [0 0 612 792] /Parent 11 0 R >> endobj 1 0 obj << /Font << /F16 4 0 R /F17 5 0 R /F15 6 0 R /F27 7 0 R /F28 8 0 R /F30 9 0 R /F18 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 14 0 obj << /Length 2749 /Filter /FlateDecode >> stream xÚí[[o#·~÷¯ÐãÙá’<¼‚è 6;/MS`V[JmÉ‘FuQÿù^f43æŒ.Ö¢n±~ð\D’ßwxn¤¾»¹zÿÉ £ÄRË&7·Æ%ÑÖL”d„K3¹™O~ͪÅz[NsÈžkw#²ùrÞ¬Ö™þvóÃûÜt$¥Õ$ç”å|üv»ä‚Æ&4´ù&ʲY†aê&·©î$RÔ-þF™Hɱ„i¶oDij<á¶nu¿¾›æL`kI!Õ1¢M3¶MJ¢&j?ú¯¦9ç<ÈRcDÀÕuë»´<i#/Xgƒ€Éc;¯¤Z‚ês.{\ê \öt¸ºB:úESB:Ý\,øßKËÓa%9ŸТQ0“œ)"ÁÔV&· 2>° -åÿ¦pæ'’ã•>¨™jK“ÎÒL3¬™_»&Șaè,r( ÇOýg¿$ç`ˆêy†YÅUË0A¸5œ§GíV\ÞnÊNXí½Y1"¬u2M ¨«?ß\ýqåx¡6aÖ*'ç &³‡«_£“9~öÄ@_üä[>8GI™³÷“ë«¿\}ç¼wÏ^DYš&áDnX7M¤´nZé7HI´DKÓÅí«„À kÿ"Ì8Qã‚_†‰FØYTÀ›§‚~V*@"0J¸°·B…—ë +4Ú§ÙlZ[U ·C ðÆeÌHL׬m›†CÛidßÒ;­¦Ñ s.'è5”èC’à2©Ñ”ï›uîY2å‰?Õ¿¨„}EÚ- —Òe$rL—” ÖÊ®.õp¯… Ôb%kàÒ>Ip²7gŽÌA·‹¹Ò~Òp»è`1ÖÑH1´\o_WrF9‘hÎr0„Ù8ï©ˆ¦æ_¬ ×0 óê(=·ÄXv …£ÊK8 öJZ £€ã;σöÂeˆà¶ n-êcíq¨„h¾ÃþŒkBù…pi„½—ãmhÛºhŸ/æ-©êš—–8A4 ³%Zì×$0œØ$ok_$Oª*3ü‚ª ¸Œ˜d—¡¤v%ð¦T„Er)\ja¯Àåm«ê犴!Òºt “1Æ»±CbÌþ„:×DW) 'F™1N§Ǹ]¡9QÒ|6·›à4–ãÀ/µáèÆ»ୈۥɠÛA–:XŠKZrtûhˆ.e#ërPŽD/¡T (ûhX\¶-(¿NÖuŠ“š»—hæÍÂשMV̪]qïîu¶]ßïªåzžfÓœeëÍ”e›ÒÝnqÔ:[¯æÛðÕjHÑ´«†¥_ÒJ  UÔ:Ó}u©+ §$TKÃ1éÊVîBß…‘>-–³)7Ù"<ÞnÖaŽU ûb¾)*lV8UD ÉÈÑÎÊh~n×›'dw_„‚þr›d…ê!|¯Ân˜Úx†¢NÎx?F® ò:$† `ɲ$ ÙE½±„Ò¨]‰•$Áúj_$–T³’39¼”ja€ª¡£wƒÁj\Œ&R!mõ]ZÖÌ\4ÉÅÑv 97ì]àvíÉÃô…À}¾î[ý=ÕQŽcñ¹ä¾ÝÂb¢«Uÿƒ5D„MŒ—;CjIû޲_vpò$\ÙëqZâZ$Ó\áìÎ*‰ ÇÎJaÿT탱Nnç}|\ùˆe:…ï¤æCþÞ5T(©“4mèRš ;É D[}œ>Ê£ô1e@‡`Ä…Ù*»ÁØ€¯‡ÑpÔ šž¦DšP7r5š‡!ÁTÒÎò.Ô¨ Œ 1ûcÊ^ab:öSùT­WÏ?‹­‹…¦Ù²ÂPÐG‚(áí²þ`Qlc‹ÕlS> Ç(©rï”ÍÖ·IzÜé *Ú¦$rÄE+fÒé­JhËæ´f…V@º¬L¶‹ª‡×2º,#!»®ŠMµ\Ý…9>-«E˜™‹c)™ƒß÷Á¢ˆâw[V qâjY-ë‡rë¬–àƒ»+ª2|°N>p+˜ÛCQ0ãxëdH.óÁ¯¤&ĸB‹ì§çŸã€ª),·á©ØÄ„/Ø,©J C!³/4spë_Æî“t0ÛvG„\s@‰oÀL_èâ}©@Œ7‡´§I6J$aÇØ_TëÉRš]#˜°ÌÖ«ír^n¶á1P‡7ÅýÝzƒš÷à•˼üëEá–Ú?Ý¿²0Å$-¼ÚÜ•ó®qcl±âÕ`êÐÄ~6¥K„˜‰ºk{ë"aŽ#ûÓuhÔ´Çj½‹ßtvs·õ«#Šbæåm±C¿¯Â³—Ùk5‹ g×ã²ðã¤Ý.W!/‹™›Ô"<=>—ûÇT".w¥3©þ\•«¨[Мͧ¹ÎÕ<„5\ÜÇçM‰8ã ݆Çbl›¿—íòá±n}»[ÍBbìžp¥úkU ŸíêİZÖd¸÷‹âßÅfþn tÙ¶žï±%k›@LÞ1*E‡£ çeÚÍÆ&­VÉzT7¸Gs¢]†È3ÑcÿH‰¡²ÝãоX·Ž¶Iž£ëá~pïuݨb¡P¸ß|ƒÁÚ²;јï*15´ÿG`  ëάÖ1_Dq…N ÐXw¹Þ}Â.³ßËY•H¥Ñ‹I!&\qÌkGk¶Ô…ýŧÒZ‰!W™ØŸ®²Jw°°5D‰Z ¿_…œ×Õ€Ù@Àc˜p”Õ»y?dºcÖ‘äËê.#–ùra-ïõýòd4g>oºô²ž«xT·Ãl ­ÑkI˜£]c”&] ÐvvZ& ÆL‖£3RûDD8Ot:êaƒ'º,&ÄúÌjnŒV™|¯Ð£J‹žUs£}èùP‚2§ÀGÐÈYy!ôóeñöÑDï™yø(Â÷L¸fhŽ_âWÏFĹ3¸€˜%ð·_¨³âeà”$Xœ½8m”—ÞbÈAi™±÷ÐP,¤\G'‹ñ´;ŠÐáXŒpLÑÀ¡»¾ Çh¹U½Íñ…åK±ìí rc$Ëa’Ñ-a"£/C²;nÉè—¥|’’üÌ[küßÍj„fC17¢Ù`0YüBó«hn ’¹Í!ÉGHWî<¼«ó ùm'õ5Ùv¶ÞÄÛ:ëMž^4h/n>&’'M?'«Ðý'¿Ç2+îg.c.çá9lʺ»&¯.}-£Ú,˘¦/£ ›©¬øt_†þvOÈ#{v¾ýv _hm©¹Í)®²NlÜùËÜÏ!z5“­ÏôTö{] UñÇ]¾àËn–®È>€wág`·ëøb¹ÚVÅjVÆÆ§i€ÄÝwÊ9ém#žËAÝn_›¨^â'¸ì[L õüeÍ u |]Ü·úk¢K÷ðiZeÇøQàXœ9r?>r$Úb‚7ÁTGûkvþ¯epÆíÓ?> endobj 12 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F29 15 0 R /F1 16 0 R /F18 10 0 R /F24 17 0 R /F21 18 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 21 0 obj << /Length 3260 /Filter /FlateDecode >> stream xÚÍ[K“ÛÆ¾ï¯à‘[Gó~D¥CœŠ«ìr¹’Ò^Û©Â’X-R$¸&@kmåǧ{fbÀÒ*ÎA‹Á Ù˜îéÇ×Ñ×w7¯¿ajÆ(qÔ±ÙÝÃŒ9J„P3- QBÎîV³çoo¾û(gŽ8͵§³DJ;c’p©~¢œG„¥†r¶’²@Ç]JgÑŽ¥‘ȉ8#Ò9äiá©nþzwóË ƒÓ›q¡ ›iE ×l¶ÜÜüø3­àÑw3ÎÙÙO¸™qI(0\ÏÞÝüýækTÇèu•¤„*6x᱌Zºní_Ý.8ç‰Ö@©Ê¡F›^k"+('R»ÙBÃŒœÕ0¢•ºHX M*l*DÏl(nvqŠH £W*ij #Î ð7>>V”çôâæå.2/{Ú¼„ä°Nû2Ö3û$—›Ø‚Y1©C¦a7ô§+qÄÍcØ@‰|ʆdbC©š%• ±z5÷̾´š3‚:ÂOuû§Ü›‚ ©Hl¢î¾/ÛSð"%u·Î"ÇvÁõoAca*k÷L£ìÕÍ€þ£YÁA AÅõa™Þ.À«çE}Ëæ«o™„ û–uÐŒÂÔ®lªÕ¾X7á¶Ø•QEqâa»^oo¹hˆ³ ÆÑ¼š ÂÀöý ÞíïoJÍÿ].Û\æÄX>S€¨4§|ŽbpVÌÈÆÎ%0“„±èsßÃÛ9¼ÿû¬Š9aÆ‹¢ô àÑþKØÖ‘æ¹Éå™W l4HÁ}óOþ•c¿`àÞ Tw û‰2™{&æàˆÝrSµ2 o…PÖ{Z­\€Û2w2MSã7GYJ˜Fë‰Í4šPÀL§7S@S`áÀØgJ BõKÉ ÂšŒ½ô‰']BÂwêåcž2SåØÈ·Bá²Ïro³©ˆ›a¶6¯¹ìw‹qÍ¥ç”Ë? —\Jò››pà¾5Ø-a™¶ìØ·ºÍR'6 °¢úÏÛ,â Ì@QóiŽ“ÙP-F¾˜Õàq­¡ÖkLJZΘ°î„)ëÚ¡˜_Ä i‡AñÌìÿ¯v8¥ 5 }° Mc|‡PòÿüÝqÜãÖ¿¥¬Ã6_¯ð¹BA$÷PçW%b”º\Åù6L·ßöqWFÒ}]ý²Ó«²h#iµ)›W0V°÷*n¥&N˜$‡<¸Â_ñÂPÆa¬ô¼XïË&ÜoðÊ&ªHÄLì‚"¹é<»‹ rû„éa aP/„@âK‡XAäS "nµF6Ú ÊÝS‹Ÿ¶ÈWÙD˜ƒöŠ~˜ZWo¯5¿pî’ÍîÕ°ƒð›eFjâG7 úe;ê÷M$7Pk.ë |8»Ã4¯Iù¢F™"¯7¨Óš4'5‰ÐK&ž.EfÒ\0 Þ v3Î.Âö*|ÁÎZÚÛ³VíóÄWY| —_‰)_ÅŠâ­¯èi¼oc/crxÒ‘êô'bî^È O¦æPÝWža°þíÖÊ9¹]H©æßàx» ÄËm}ˆâuUÖËØj¨ê°mP±p*Óm{ZoÛ¶ªß¥bq0ØCŽj0G•¸¨ïC/Âóîò…°¹'б±?}Úmßã|±A ”TQ¤Ýïê@°­#e—÷Â݃—eòe¼‹òà_ßáU'ì Û•øSIaEFá%–ôVëu¹Ç)wÿÜT¯V!ƒ{¾¨z/ü6L@zT}z”|˜©ñ.ˆ‚¿^«Î-=6}D¾)¶§õÿ( ]ÞµsG]»³10W~Ѽ?!M×G°¦|wü€âN³ùåºÙÖÿ\¬ËhIe»X´ñ¹ÇAÞÒâÄcñ{±‹û^5an¹Þ6è¨þfW®áב¢ÝF#ë,ëë]éýlqŒÍ:+‰m@0IáæÅøÞ?ã¢*py˜¶Áâðy[• X¯Ñ4È„s÷¬¦ªË@Ú-Ÿà‚ñÚD÷ŽÜŠ8Ýv–Ûõ~SçLö\d¦PÔ0Àw$Ðt«œOäªâÞ›ÇöÈ@­D†ŠÃò)l > :€A.Ͱ5êg>TŒq´Œo@»«EÎ-~\è-ò9Ò°œQ…Jä¬&Z×P; ©š°Žƒú¶ÌáKÑ3|°|È÷"¯ió¬\ÚY'ú„/Ã;’­/ŸŸ¦^Ç“nÁUŠ}rÀ =pºk;ì4³ ÃdŒX˜8k™¡•¡fÅ9Í 7RíôŸšýî×*^áÑ ö`o²ÈIŠáÞe•­­»ÌšØ“yPs¥–“d‘ôš§¾HŽû9:AŸ/ôu.4È9Üb­H($k#š€8 Ìø4cm´ ,÷Ä û5"« †G}‚±Ï@‡ßAd‡ß¥{· m¹Û`ÊÐÔÍïn­˜Ãýt2‚”O}ü‚k.uðVòØÛxn˺ 鉇¬‰t!¡À M³8MÝÇp¤ÜŽ~±„$Fî•Ã\sxëçÄ`%{ÀZ½ôÏgF®çß¶,Haƒf¾Û×µ‡ÂxŠd+Ƭlëø»z¹+7=jô,"ËxŻЊɤÍ FÕä➨ÞgžÕY¸¤ÀhÅUxé²Èÿ2eÕñ7öÀr!Zø[6žñpd@÷lL ÇDþ8÷¾&Ë–Äqþdžâ‰oÜ!¦^ÿ‰|J¥¨¥(u.éˆÀp&øŽz¦N 0&!jý" ¥Pµ>VÂJ0†QH°x^ËþQ´Þàz÷ë¶zZWe¬Óï1 ÿÆ?žÀCîÇC?gQ:7ü¢-£-…¸ªÌå4gçM¹Üz„ã\§ag' ŒÃ³‰}ðÈVšë‡ ¿2Q ‘‰ù°ƒMèÔáÁ„M|•y_XЀ( T81N%¼0©0uHbd¦òm6ð3‚#z(%Pµ!僶ûªÐÄÇ‘›€ÔjwÕsÜGÔ¿%z„êЩšÿ%”áf³õm•U¹ö*Ý}]8r`êmí¸"•^P„K¨— |ÇËEû‚'½}á»/*Þ̰c±\n÷5:g¡%3”ç¡'f@l(·ðŒF &|¿Žvzp4Âažn ºëŒs£¯)0SVÃmx“?®(!Þ‰3ê¢R&êÂe€ºu®Ê]¬"8‘Ú ¤áÔàvT=5Ó™½ªËtqD}cƒA°ÃžHŒ¾\@Œ>5ÆN¨7†ØGÝfO!Ò´~ÈGGo•ýjû¶ë²ˆ§´î»¦a4M±^Ç0ÿaWµm8Í%°Q‚K–†h !®9»¨:î þyrœ8ËŽÇQiB5~E5.~:þg&µ ¹ßLÉt{†qq! àX ‡ƒ¼4o ÎJ@ V²E,´Ä“ú,(zÎg+.ÔUÖä1(c"Ñ\ú®AWõ²ãg¶¹È™+N¬&RºÀ“nx°ñpÒ —@ñPàaþDO.]1N÷éŠÏ#æÉý8 ™½ ÓSçh¹ÔP<£Y"°D R I Ásõ<ôËp³ƒñ¾®~pÍDw=þ@…ýuUÏ…„2.|åÝ7ø–îr»ë×Ñè€;làâ"—eÓ„uÅäp³0iç2¿PG_Ü<–Î…} f–çm×úÿ92 ò}a%±{œ÷õ6\“Ï+Í  OÕ†‡›²hö»î·ëH»+Þ—aÆ7 k¸<»¶Zî×Å.¨DR0ÔQ–ØÞ7å®Ë}[ÿ½ÉÍ»küBå†]5|ü0Uy$ÇM/8,Š•g¾»€Œ½}â5ˆ¿¯|:Æé—SDÂ"\Ì2qm†½í;Pº·ƒˆTþƒ3tp¸‚  –Z ˰%€xwÔúŽôþä13ó5$¸u&«$Wûø-Ù×Y“Sb N¤]æzŠ ¡ªjº÷EiâʽJBM "ÇݨbµêJ±[ÅS×Ý7תy•G§jxŽï$Zè…{ìNh÷zŒõÔáPü`æÃRìÜnëþxÚØ]w~Cq…8BÔ ÄX7:éù_ñ,z endstream endobj 20 0 obj << /Type /Page /Contents 21 0 R /Resources 19 0 R /MediaBox [0 0 612 792] /Parent 11 0 R >> endobj 19 0 obj << /Font << /F15 6 0 R /F1 16 0 R /F29 15 0 R /F28 8 0 R /F18 10 0 R /F21 18 0 R /F24 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 24 0 obj << /Length 2236 /Filter /FlateDecode >> stream xÚíZKÛȾϯ râ VO¿ A ¬ì-È`/^h‰c1I%Å럯º›”(SÙ‘³Ù`lÀl‹õüººŠò›û›»·Â‚³Àƒ(î a4³ÞÖHæ”,îgÅ»ò¾YÔ·ïïºùñþæ_7¢àø+ á 3BV{f­/¦‹›wïy1Ãß ÎTðŗȺ vÅŒÅcñ›¿ß¼ÑÃó¡Ú[¯Êº]ÜN”R¥±šÑ}Ú%$óJc}Ü«™òª° R½I6ˆ~+Íœ¼ŽÛߨ¼{+ý>£aÚõ‘ãuâŠ"Óqü Wcr³Áu\í˜Ç,,ʾH)KyÙ,áÁ¼’3Ëm¡X0>½ ÇDNœaù™Ç ÏœöDd¥`‰ë„V&Õ:´áDhÆ“=¶,WŽ™(,SV=Í×19šI-ÿÛ®N„rLë!ŠÂ q!¯„"!bY|AÑEG]×pṵ„Œóèö‚kaš`&^pòÏ™‘ê©?p1!Üx͌ϻjvü¬%Jê+µÆ!kúìQ‹¸0ãÌU¶ò¾Ê ‘9‘~<÷XâXû.0óÔeØA:&Ú)îJÉ¡Xïü2¿-Zæ÷/Z?jº’Žj|Lâ¹ÁŸ´×„q,R^ñ¿ˆ¸Œu4„h˜Q0ß„¨+'D^)!ßá¬-&RYÆ•îÐÝ—Qi箳C5ñÉ—ú²C_vè¹|ð1_µ`^øK| ð¡Å¥ÚR‰þ0ä=ø–¾Ì¬ÚÑŽ9 S³;MBi Ìí˜^ÃpçKþjLžT̪žõ×qaÒšCa?$óq¶ ªZ’šk²©v8T…Š)ñÌÂ%À¥$„!z>GígŠZÕ6Õr:¦Ÿrä£z!NË’UQì¯á·ÁÔ¡½bZh°;§]À žY­1BX+”Þ){;qÒ*hÿ!ïØ"{üÿ0–g´|ývµc(à÷"Ófg«ãÑMÝüB™J…Á¼)ÏO¥¨r8–¾NÑrwÂó|[èÔ4É]RÁïå•rjÐiö¢óJÆ´êRÞ ›Š‘âšÿ'{¯àùF"/“¹ûp`OHœ‚4ªÒßHÞ…ð 'àc$Hq%üË”²gð³GP©›ºì šžŠ¦L~[ hQáìç{a‡\œü[Bôàò<žÂtÐZ®¿c›pì;Ý.>g ‘^‡Ñ~Ckæw]ÖåýF ƒöŒ‡pI»!¥gª5ñ•g¶}´`ÝÇ­3íFLHtvœÒ.Ñmp~²Ûè|P¡öh·Ñù¬B:Ÿç³ÒXÛ¡RN¦"~X7@Õ\R11 fð¯ó®›oÑ(òƒâ(O¸à çêJ.`f¶Öõ.hnµÏÆö8®­êêaŽ» O¸`Vs-œŒmkJý\{.`/:”s”€#~ÇæÜSyA‰ï¥£b !MùãC»ZBŒ3eõùs»ºU¢üµYT›ä‘Ë!ˆcŠÉmÌý¼ÆËÚõr°Lr0ªör"ù]§ø èÀ*ß×ñ¿ŒWå ûRñ>Éš5(wªn×éºiµéìyÓÖëÇhÅ—DX-¿fû6YHE3ѾÂÕðò˼nëüA•~œ¦z¼†M‰VxÁCÊ­(¡S@ìËu¢®¨¤•Óé¶e·€¼ü+Ñm¹]×ÛÇ´ŽRz@جòuÞ,?¥eõ!ý«í¦{Tç#A¹Y÷"Öi|ãCû«õz» .€)†«æ¡£$oh=«6UZÍ«YZ$cêz™nuµÞ¶u|ˆ 4x;ñ7Å´®(Õôë¡FrdļY0’Vç(ogév¹Ú¤Å<ÆìßôO(ûvávÓÔ3J§B›,u˜@¦ÕºÎÏfM•­²¨Fpc³¶˜ cðáë¾yOÌÂy¶˜‹òsDCÝ6ëy"D£À‘҉źŠIÁ£f¹ÞTKÒšy¢)‘·Êâ:£p6]à>¢6²Um„l€k»Îog·#4£ÆúsÕV›Ñ,Ô)¸É Ð(d†–{¡ ýN8êR³m›¨kÚ[ y1¶üg=MBrJ£©$²zlº,ÒÃj9Ë‹Mº"^ŸÒêa•EǀтÚvW»§¸#(lFx„­­)ZÆ—ëíÀX"팥»|\“Q´ŠFÑ"e‘„ÔÓ9AëRB£5e-R·û ;BõjÌz+“HáQÚGâµë rm“k¢Œ¯É‘×óƒcpFTÔé!U³UZ&ôbñi¹«£¸Ì›)ÝÎ zÞ—[f3V‹,c]/×£y€óËWXs车TtéP€{îSmM*¥ßÿÑj¢üB¥Úü)½‡ºø¡úÐ<6›¨"#s¥h–ðóc"E´ášÑ&:´t€§¼¿»,k!†YN¤*_7Ô³ÆM'Nã ëºeš:w]¼;5Ó²ÿöHGŠs1òQQ.Ú´ž®Ÿ·›x ÄHó>»Z ´uoÝ<¿1¨‡DÈ%8˜šׇGåYÑ›nWIAèK îtô?sÒœ7,@ôzÝ.Ö‰µYfR:½Uù¸úøØ|¢‘‹Àªmw¸’ƒ ’JR2Â!‡.äŒÀ1ã¨Y*àýt‰»gBºKEZW‘CúïRÜÉDH• äòD‹¾`]'¨&ŒPîpwß1â ]f½«kì š86õNöø^JÍ_¯IéC4©É=‰ŠÆÅ–® VY™ØÈ 冿“\2Ñ<`:óÅŽq8±~pÇÐd­}ú÷ßöã‚~ªe}7F4ÿh 0uЈ'ÿ¿‡Ð¬v8–|5Ñ’íc™RÜ<+ɣɑô›ú­Ùé÷š}…óÌ98i¯RcØ‚CôÛÂÞïEÃX*§™ÇdvA,á&’S¡ì… CyáT&dÑÖÐ7JÉ”Ñóéþõû¬ endstream endobj 23 0 obj << /Type /Page /Contents 24 0 R /Resources 22 0 R /MediaBox [0 0 612 792] /Parent 11 0 R >> endobj 22 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F18 10 0 R /F29 15 0 R /F27 7 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 27 0 obj << /Length 2989 /Filter /FlateDecode >> stream xÚíÙŽã¸ñ½¿ÂÈ“Œmsy Á.°“ì‘A‚é<3@c«§•Ør¯$ϵ?Ÿ*R’%7嫽ӓ /U¢‹Åº«Èïn®¾~ÆÔ„Qâ¨c“›Û ãŠg'Z1•Ü,&/“´XLgBˆ¤¾Ëp “UšôË&-¦Ü&u^çY`i™M_ßüøõ3nû¸édÆáV¬?ÿ& à’XÆ`®ŸóMƒÇ h´DJÛN¹-¥ˆT²ñŠ2Ãã3l;‰Ò=aØÎZ®ßNgLÂlEEla.ˆ±me £!Zv¿šÎ8牄,F#§DSÝÎ~ÇgM_Œaƒ=<Ãbë‚,Acþø5D‚üRí C‚ËØß€YrˬØÝ7#hõv<ð¬aŠª­$\ò«Û³c>ÆG´3Ÿ‡ýò ûÝiìâl;º¡™šXeÀ»i¢„my8sLïãNÂQþ?áž¡â·Uò/•Dq¤ÚË›¡Œ™áÑŒƒN*ë@9áàýô¿Gi´D﮺ç4×½`Ëä6p¿¢œÇÅ‚‚›õ§²<ã+‘Î!NÛÓ„«ïo®~¹B« 6aNaäDYNà|uõò5,àãJä1ïýÔr–2ÙròâêoWßaæ³#Ô™áÄHs¢ú±ç QÊáFŒ6çD £#r­羊 JúEŠ‚kG”Ö—E‡ì,Qˆ/^ô7…Üïá"¢è% uyQx¼è˜é…œ=m}(8MÍï:ãÆX»§œ&{rxO«vƒ™…?:Cιš@BLñ“Dœ¹LipæÛi '¸QËÎó4âc …BO”ìrŸ69ÊɇڴÃù™9µ;‹² O¯ Šs %—òÌôg/Nw2N5ºã„IÐBA‰RØÿ뇨"+¢™<"x ßÏmLûÌëøxÄ5P;€¥©QŒ¸-2 ,Ôò¼ˆ»S:ÀÞ­€ùrë–;'p¬ÿÞ¥’8.&3 9ÿ¸"B^ˆ1²Ç0f×éêàsíŸÛ÷FÆ'¶3ôù¦ÍZ`“İaˆªÓ)6âhaʬ¯~{béÅ””;pUP_]F-²³d!F•T>’ î׈Ë0¦CöÆŒe‡µT^LKgœ‚›„RI1ð»zŸž H.¼žŠãõTŒ;Sk¡Fci™W±Ç9Ó„XçØy‰Ú¨3Ÿ×™BѪՅÓ!{ cFœ©9Ö™êsÔt¸Ð%L úê÷û¨:30Lêí²Æfæ 6î¹Iµ^nê|]à›Læë²Ìª{x1ɺXTZ¯ÃÜ:üI§æß÷¾Õ±®ò:‡Mÿ\®¼¯ëÛxž$Aº†V´f#Ì6}?"/Ü6φ˜40T÷01Ü#viüƒ^ãÓ$ïïò9îã.>O‹¦ÅSžÛžMžúÆo@¨HÞ—y]gExI«Z©{4/r”Øfpžº´œÏ×UÌ"A²½öêáÆ4²PF;M˜b'4|{ýµ1<t¼ÑßûX@æD+¤D±ì­!ºÊe´†è CŒSƒ¢éᙆ:¡™&7£ßdÔuTˆ¤ûKÈ„û´z=ˆx ŽаýìPÆ*ØÅEØÝ!ë³;^uYuX§ÄtÆ´'TÖVš0Îú&o³2¨‹;9Žv‚¡®2¼×bläGí,uð°,ꄱ ÑÑ ©¼áì:¸Á9$vèZÖ^'0‚:?ºðþ¯!QRXºycnÐcôŽZênsw°  M ú `í4´0Ÿ?‰Ûì r» !yT–fkëy<à5„Â"j£"±ÀòÖcûŽ¨Ù â÷½¨~„‰µ¥m ±¾”õA3I>žÜ€ŸÔÑ .´ÛÀ±›ô,ÅÛæ’:¨ãÿ½Ó«}lìù˜G³Q(댋oSKHºCR1w ³±£*ã*š1 [ïT¥ ÒŒí±øqž*où8yû„,ÚÔ1ð™š.ó=ý°]ܘl+Áän“Ñ[( …ëlæ8?³U…ôwôìQïg§ÆD™‰aÖìE4ÃbÄj†µ§“¾F‚Í “òÍ2Ý–¿K?¥%ÞD’:)Öøá}û¡ ƒ4|ü×fuPŠxH•j°ø±8÷yÕÑV¶¯6ázÁ=É`‘)X“±øasÚŠ§"24¬›ë6—)5Œ¯u¾Â²ÐêD“éLr—|[ú‚ÓšäeXMÐÁj”täü1¾+Ç;Û{/òE@Yø²ÓCªù¦ª4«€Ž´©tá#*>CMkãJ†ð dX:•¼ÙÔš7;6˜îÔ…·ëå2h£¯ŸRfózù1¼ÁæWaÔÔÓ.©ÒU3JË·›U.Ð5ªœ6Í”%dI†;Ö,wBJ·Ìÿíw±ÌïšjÜÿ.šמ~(o²ø÷­mq™”ÙmV†!–ÿø„=û§7.„wÍôWLH/Öß…÷†ÝÍWouøvz`*ªƒdÉE¾Àvòª?«Y¼Ú¢èQÞò|‘ݦ›%hïÑeA†e€„…aШÓûûÒ‹éC§öÌôÿ ×aä×¶ÆÿHgj‹··ßð‡A10pÊMònªt’æËôÍ2 `¼A‰JöâÛ$@žM­H¼pá¥ÊW÷íÔjS¾-U*I—2ˆßq ­ç$#ר‘àfÀ¼b=›¯·ë—¹'ÚcH« Mú:F{Ø(l”¹À„~¸_®KÔG|{ƒø?†ñ³e¶Ê‹·á%Ü…ÁŸÓ²hí™ïQ7ÀÕàZÔQžÀ¯RâÒyº¬³²èÇÖ릳£ ºá:¶[!’ŸÒûeZüú—,ÏÊPB`Á-ÌÀlBœ¢4 –óÊÔz•uû§7Àöe^ø7°'P±ôêXZÂp ·›^æ®sy¶ܳpdê¬XxöÛ¿‰Û,á÷ا¹Ýbp9Â(´z ”@åÇ©• * øÅ *øì×0­ªà(=Ä[µ ªÖ™*0-/æeæ˜Iê^—¨ß Å«!¤ô°£—%B7´ÎÝÐW—HÔ+64æYCõ¾Þ5ߪø…(aˆ0;u÷bìÚA—¾TÙ}Zý7=edoÛ ]X°Ò:ø6ø¸ #y­K„Ag <~½oŽ"Ht!7 ¸‚±D‡G°RÖË‘ç=†Æ>SÚôèe ÒP1¦,‚ìAUÑD¶{%ÑÊ‘~§5æ©vªµS “†—?@#ï|¬)C¢P„ikpÆŽ/msãÒ"~è 0…¦qÁ¨ê}#ÜöîµÙõðë¨R@%E;‘/â¼7Ûf‚ÏȬêR\…‡Ügã-5M‰ÐîÔ £çë &ʱ¾(¸ê.±Ä/Éb¿wòBE­ÝvcÙZîç+ÛèÃÐCô(~¡¥öä‹ô—%W]X ü‹ˆß¡/­Çüj2Ÿ§K¨ Ó½noÝ"…äX>åP—T>Ï—*Ôp BæëÕý¦n²yü>(̉0vBÎ!åEhs‚W‡°ˆÃ¬ˆºõžgËj]üúmº ]âÝĸ¶ávùÌI¶)þTfÁ#ù5vª1áÜNâˆÙTYó ³¹¶~Nzeˆî[âŒxù‘‘ô,€â´LccIù6›kŽÏÇ.ž q’Ÿªã %'Üî9ÂîoN(ì‰ J¸Û›•3"¼?¿€ØbƒDms|ÿòá9[8¿1ÄІ›‰þ"JoÞ?šy¬ìt¸aI4­®<Î aa¿ÛYGãñ|=ƦtƒÀûQtócé>¨}WÄ^¿½†óÙ<±lbQKû“yzÄe¡£Ä§âk]Ñ€૤¼´ÏbÄ8–1½“_6ÙcjÇŸÖ\^\=àV÷¯ªÜ‚‰þ™D·vª:ÿ;nâ; endstream endobj 26 0 obj << /Type /Page /Contents 27 0 R /Resources 25 0 R /MediaBox [0 0 612 792] /Parent 11 0 R >> endobj 25 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F29 15 0 R /F1 16 0 R /F18 10 0 R /F30 9 0 R /F21 18 0 R >> /ProcSet [ /PDF /Text ] >> endobj 30 0 obj << /Length 2639 /Filter /FlateDecode >> stream xÚíZKsãÈ ¾ûW°öD׌Z ô;©9d«2[Im©qå2;©¢%y­D–6’<³›Éú%‘r“”cy“C."Õlh |@óÛ›«é{Ppæ¸ƒêæ®TÌ8[i •­næÕÇúæ~q=RÕóÅzó°\7ûÍÖÈú.Þ¨zŸ§|·],Ö×hë/þgs=9üί<\£©W×P7ñ¥åî„Êîñ!ÎÞø©Ÿ=©Eb¹¿ßìÒ¬f¯Ûåîïo¯?Ýü±âÕSÒEÁ÷ˇ…'-½f hÓÜé{í•ŬqD!¼ùeI|÷qbWEÂ1…y^”˜èÏÍþ~Ç®'’úxŸ“ ‘±Î’,“L«f¿Ë¢‡y #K´«ÉŒ2™é»’XŠI¥òŒ?pÅK” C#ò¬m‰ŽaZÔð†,‡X "žd¸ùÄüoÐ÷©Ûé¨v_zmE½ÙBº3:Sÿƒ,‰à˜9NzG"[ó·%z(˜‡©?—‰¡VO‰E,ö%ªdaád~çs‰ªeNÌ®³R` |ÿ¦HR¯3<ÌÐÌHU æ”ËêàÉf`;ó”=˜Ê&3ZT“©´N²¨5C®nÏ÷&+X´6HÅ+²£4&Eïàôøê÷7Wÿ¸òZáTÈ%íC $œ3 ¦š=\}üÄ«9=¤mJúµÕ—0õÁO{vU}¸úóÕ·O£’àH÷šˆI¦-F¾ñnÖl—ÍzVàÚýzþ’ô9ÀýîÒ~K÷ ¤ ®RŽ3#dËåU+«™Öîe«FºW¨O˜N…àk|![&…"Z†U9”v33Ü'ŠìuÜîŠC‰ŽcڷÆÌñçd§€b@¡¹µ°D 8å8òÿ“ c) ¸):k“»;`–t×É)CÚW>¼Á…¬®cl;ªÎ[>„­71eÀÚ¢–STY5=JÍ8^ÊAgÜêÿE‰i+e¯"I<æ!zU_ˆæQNû2GDͨS?öžÎé©FTŽìÜD QÆ 9²  &Ìs9>J0âÖ`™Cq!·FJ6Öþß­ÿs·îÐä=N.Èmåx°¦DJð§àã– Öòq$ŸU„ &Þ¸yE(ZTÉyÔÑoŠ»•ñòÖF[eÞŽƒÏó!m‚hÒ´`) !4”TI Œ€Ýeš´Š!œЬeÀ‘Þ *ÃìQHÆ¥hÈYÊTÒPݪ± åE<÷BXŠ‚3kE—)åDŽFim:1ŽF8ai¿w:;±Ÿ­œâÙ‰I*®»qV¡#H{¡U(BÌîó*U²Bð“U§Ñ"PíE€ÜÖУX‚$VÎ]h R2Ði ðœ%@g Þ5¥M¡$œï–± ±¦”N…¿Rõì1µ3š}šÇï›6Û¹/45ÒìkˆtýÐl÷ËõÍ*ýoÖóøÆn¶Ù¦±íb·œ?6«]ü{GooV«Ð ù’R:ÅJªO™/·‹Ù~õ ½dTý¸#.þVÖDê%y7¡ó²³f›‡Ÿ÷$ûfí»J»úÃã-ݨúoD,N~V¥v ,×»½ßÖi|ñóO¡Å³Ø.4º‹“švK&yÂ~³oVôØeù{]oîâX1´R|ŠåÙ™ù'4*Fóí¶\Ö‡vóCËØüˆâ…’ËÆ§Ÿ6¶»4Z@qÜ7}H7 '[Hëóót_?G2”xÙ%ÑFHËú&ohÚ¢pêkq±´gêÍ:^çwÿ_âí—*z·èEA@d‡óuÑ*ÿNz.Èt[eÙ N"áà+Åv[Ç›^Ù6V¶mc Ž\P™·¿Û’mv‹Ùf=ÿ&©<[A÷X!^þ•c† ðصãz³_ÎBƒr3êYZ¬‚û.®©ËïMˆ›•žÁÓÀúäý`BÐÙÇ©ûÅö!ôFe'ºìât/1ÒŽ/D•bƒ×s$0Tß=®Vñ.‡ï%õ×rX±Liû¼°Ògæá–ªöZ?ZÙê¼)tÆÿE7$ÔãÐü~øTÀä…A‰ëx=8–ÿ£½5 Üßd|¾ö©lõèƒ[õÉB­´2aôGE`‘»®IÛ©Oxü˜ºçM¸‹‰ÐÖ_vòUËÖ'k*-xê’uRØSìa ,ÒŠU|ްö0öÌLmcJ$!t¬€)n‰rLÈcV²Â„JpŠ;¿S2üS±à¡lwlÑ{›¶ ¨kIÂSTnÑ{9_<õ¥*a8 OþÚÃD‘#NZóú**ßön•]Ų^WÊÞ>fØ> ¨ÎQƒ•"¤d‰˜¯Úú+ƒìÜR±¡_H#"Øeª¦:NbD¦KG(×0Žù“|è$“/®\ü± mt>¦Êå+Lsuò•¡å]”PPMÉMljCAg/$ Õ“0+0Àö( VHx*`V¡Š?/Îe$¤âOæš#Im,€[Ó/ ê:+Ý…¤ºÎA Œ_ÅTT(-IÑ/¡Ê6q!ù¨h94õÊÎs\·»ó}ùøÎwvŽ!¦é9¾súôüÎ>9¿Ó£'¥öä¤úŽ)©úÏSo{ÎðÇ%RÏ?»}›òçš(:aÖ#zéÔoó¡x(¬…<ž¨çr˜Æ¶±z†ÚC-#:D÷ ´ádÒJ< ør\ ¹O^0Š 93¹§Ô´/¸bRËàCXA©%UkïËÊ’({PÀóI£ÄT1µ›_9µƒ–LS´ôJjð„!äTÃDzŠR¬üù†Ö¶?·'Sƒ%Åsù¸EKÂt™Ž×‘·ïÊg‘ Äã‹Q*M¡‘NvÔf´‘n¨Ê6ºÓHog+ªÂÑäsOc™¶g@’¬V#WÏî w‹wÔ¾z¯ ·vïzÊ<žÙá¨Ð·¡eد"ô@tÐç =+ mp\h¤@Ê»ŸD( òÆ€£CÞcÇwv`I5˜zvMu¢(Ńö ÏgLÛ²b7Ý]âôõ²~Þ}O]¦/ò]ÐÕ嘫ã«Éý_÷v`”³:‚íz»äÆÊ'Þ®)l›þê-{»ˆ‹ööH«®Ûë@c ÕYè «è,˜ÒÊ)ìîj„¿ ›Á˜7hK´m«Æˆ;…P¤B’£;%âÈtÎ¥ˆžÌ‡XäDû< Î<ëÍ¡» ÝOe¸RcìL›8“]‰gŸÅLžÅ¬§–@˜Ðêƒ!‡·¼ß£´—¢wæo(ŘG5Å5ÊÐDí†ü2­ñ0ë¶@J0âù¤feRº$Õ NT U@Q"ûžKû> `Dn•-$9ÕßÉ» /mjëi€ÿ SÇf×øïÇ•VØ/«ÿ>BPöFÆFœˆúoH’•P endstream endobj 29 0 obj << /Type /Page /Contents 30 0 R /Resources 28 0 R /MediaBox [0 0 612 792] /Parent 11 0 R >> endobj 28 0 obj << /Font << /F15 6 0 R /F32 31 0 R /F28 8 0 R /F29 15 0 R /F24 17 0 R /F18 10 0 R /F1 16 0 R /F21 18 0 R /F19 32 0 R >> /ProcSet [ /PDF /Text ] >> endobj 35 0 obj << /Length 3213 /Filter /FlateDecode >> stream xÚÕZYä¶~Ÿ_ÑÈ“vsyé ?xã5bÃìäÉvM·zZ^u«-©=»ÎŸO¤ŽαñÚI^¦©b‘,V«¾"çõÍÕ«¯UºRR8éÔêf·RV‹Ü¤«,UB§Åêf»ú>ùúº0IÛ]¯1I¾½^§iòSµbó9÷ ûʳl®× ¨¤b–®êëí¹l¸ÿþZIÙsצ=œÎCµå. þxó-HµrÂe:C™l!¬Ó«µ²Âæ)‹ôƒÔÒs.äO…K„wžS1›vs¶\¹[É0—ôsi=.—¬Š\¤0¹¦êëêÍÍÕÏW Ë•ZiàRÙ*Ó…ÐÀ¼9\}ÿ£\m¡ïÛ•Æ«{â<€faÍšÍêíÕ߯^£îëS)‘ãÆÇaÅŒÏ ™æAöÏÔC})-$L¤Rb êR`0ãcŠSFä°ÌBuò#T·Q9‘Z3WЍ.Ë„3öEºsB)µÔÝbÁi®™òLĨ x·|Z»Ù¨]óríÆt Þ¨ÌR§ŸEf„ÍJ³šsý¯¸·1 ñIÜ{œê£ÜÛüçî½pH –°«uþ¨7šöxÁVчìÎ8M5Ûª~ÄN©]ÁÉËôÓç)&zú<ËbèþÆçú½¬qîµ’’˜I;¡uñÿ|``n]èOr`@)¹3/5Aú‡˜ç·úâ3ßêx`>…±Â\¿›µ"Î Y¶È–ÞýgäÃ$…rЧ3Ь?¿ŒÏœMúúx×øö¶*‡=6MR÷LÚ´ç#"4e@(}‡z¥¹Qý‚ýs1iGøûÊ ’÷üÕîøw\{ÀN€­ Äü‘EæïíIÂFHý>¿<¶0%.©³¤úù\ÿrfIÙL2½<ºvG§¼EèZϱ•6¸öpðt&ôÕ¦=n™´-‡’[}5øÙŽL¸ß×~‘™ Ö±¤Œ—³ä'FÃ*Ù „Îàú)qlXøÉeRzÜŒíI™ÈÐU'ҸŜƒÄ[¯>l忍\$´·}Õ±j†º=ö€ÔÁÓ“öèç/?°>TÜÊ&¡"›!3‚³%^ûÚe<ü@9”ÍBä†@>ÇDÚÙ,Öª<æÚõaœgÂúBš¹2`‘ûzØóÂ%6eï…# TõÝÞû ¼c.ðaq½¶F%7{Ð$ï€Y¸!Eñ)R2…(/UpÆÓƒJR^Ðxví‚û¦e;P9!~¿¯<7‘ïë¦Á–Nîjöö0gÝs«ß´]Å,SùD3öË/÷[mãnxà ‹N¤±ójÖÁfXÇ—f»Pó¼®4ž^ œúsµm×ÁÏRö3ÐDšà“«›ç‹ã{CÊ,yË&EwÅ¿«š-wÌEOE‚ßï}U mpÛõsÐ1ÂFSÒF`œŸ¨ìüÊõÖÇŠzCu©ôQA:v€™äA‘¯A²ÑÀŽJÕhÙ —ÉaÂ=žì©IEO‚É ä‹‘/âðŦv:+©ŒMYŠ­‹š&]¦M'§¤º†Ó©`:År²ê !IÞ•b‹lï#-,ý•7E¤ÉJð‰VbÅ¥Vå.‚k4ü![`#ŠÉ2IÓF@ÙR‹J`âô\›xš´fÔí׌·FŠ¢0Ÿ`M °YÍâ—MáÜ@Ȭ¢”"3ú™=¤ÖN®²¶V“?D=K!¥}Ö±F×Óä u8‘¹Ñov±¥Ú‘ÏD׃fV¼Ä‘õ“Ž\“esG.ŠÄ°#c3šqà˜I÷ß1}LÁP^=úÛݳ–òžäƒ 5@Â:³ãÃ|WŒ›ªÇ²i ˆÂF’…ÈÓl‘}ú.– è|˜ït–'«š¾=þëK„¶ÞÑ#äþjw¼3 ºV9ì,Oç;íH¬&:D6CXPø™šÅÙvåfàVÀ"9[Š(~É[N{v²:Ä}-ThbU›Zû–2´NŒéÞt¸«ÀkÖ@½Ê¡4s¨…ÔSï˜à“ó«¯!¦ÏO¯ƒò|ª¬†ºê¿¨È)"Ë´Ðj [퉽„—ã_KlÏW…¦Näp ;¤Âœ*•ä˜ /*“"•ctøî»h‚ZÓΰolˆ ãVÿB™Yàge¦ƒ Vždúæ‘Шƨç ¨œZ&ýžý÷È=õ1pPÁ… ˆ$Ú£— ¦fœÄ7öå¯e·õþª±`¾ˆŒRL>Ú€= íñj†g’Ëd!&å— ãàË©«ïfAZ,R›|ÙÀNÏw{fó{2¾¥h q{Õí`2¬0t@9Ù×ýà!)ÓH@( ÁÑó¡ê‘b_!ÇÐC 2“ÎGÌàÒÉy|¨mçSňwê!RFo=ð&¯µ´h¤C±@„’«cCcð¡Q ¢î«ºãŽóP7õÀµ(EkÒä­ïìà ÞÜPl•ÍŒAU)w‘AhÒÚS¼bÊ U/&žmð‡:ù–aW Àµ½=‡*…ªyLfƒººÄüB/XwpíƒJþþK‹Ö{ÏÖà¶j˜‡o€ÙâHW(8´5«„ž0îWžÇÎî‹]+NÝòQf9‘ÈñžsÆöUôc‚3ãQo ª@ž@Á1= QÌ«y©ãE¿(¦ä½ü5©*»°×vç ݽ¤:¤×Õï£1à‰Í/TùàKc#KAh¿-)?áìxÀLêÝtª¡‰:X A²z3]1!ßX¦ÆgæÖ‚ŸÙó²NUÝxóƒbS V]¹ 1Žû‘ülEn€}Ô©ä'Ÿtßë]…qBfÓÀ-ÄŸpcô ],ð 2àXiÍðÉ=Ÿ,Ã'÷4|‚rJª1Yñ½Jâ«Ð<áxŠBð[²H¬öòx7îÙçh‚~]T3Ößá…çå=ÒÆ{ÌÅ;w´|‡׎F¬c;3„¸<xK¦S¿8?߶‚eÝ5bw§¼Û€Eª JûŽI>F:®£Xú Õºó"£ ¡ \ÄüŽb ¨àgi…Qô ò1Ó}ä]%Â),٪ȣf\Ÿ=R™§Ú~Ôåâc¥Y.¿à&`¶ÞoS’ùíJ‚ʦX(i¶¹ÇòEuêk¼[”I.[ ~|‘6b…¶vª wõ)ß^Q¾ªÀד›Ï ~#$À߇)`ÐlS/àÖ áK@ÞA±C% «Åž)<© «JÊçæs¤R>Ï9Ÿã7‡#dóüjY@ÇeAÚŽ· a<J ŠnÜ“¡¤˜ÿãÂïìFøÊVÌ݈.«9ébÉŠôŒ­ùÛZ“(üñwÇø}hÈÂ/~EA†¾ztO‹yž2#oª%òäÞòu*Ž`‰ŠýW‹²ûÙ.5h¿DÝÇ cSõ½/ˆ6›sGU´Qˆ;Ë#Ó9U@ã-ÿlÆ#T[$mÇwÅÜQaS@õDö³»žœ‘µÂmWÇPŽ ýÕûòpjªPù}~1Ê1™ 8PºÀ ®jz™‰8‡3BfæiÐÙz ”U¨Ð`jÄ̽_nË›*Žø ®/B¶ÅËmñRɧÿ/©›²» Z;ú2È”\•›L•‡ 5mÑb‘3O]¹ —qÕÏç 6O ÐÉä-Ô?Íç!;WÌwQÑ!©<–ÍÒb¡¹N†ßñ+X¦ÚU]ÏD’GŒ¾ÔÖѧr[ƒ·+0¸o6þ,¯Á„WoŽQØ¥©F½Iþg å_¬•0àY `Ê›÷t dàhžÊn¨I~øhêw×pSïIƒ ÿÝÆŒ‚“K‘:ëßlÌÊ| kð«&𠤱Œãiê;.f×xNH, MbÅ¢gŠÚ¿%ù0νü؆ƒæ&JÉ,ã›8®VuÓ4—FÖ.¾±«äT>3ŸPÅõmÈ;Øèêþ¥2£„m|@®èåØŒI‰V;ò ‹óÏ#•>ßž28§„pvlõÑï´ƒ‚R&þð‹aÁ “ #:…¯ùé x¨(çˌӄé/nŽqAŸ]Tž‰ á¬÷™<üÛÆ¿ ñJ endstream endobj 34 0 obj << /Type /Page /Contents 35 0 R /Resources 33 0 R /MediaBox [0 0 612 792] /Parent 37 0 R >> endobj 33 0 obj << /Font << /F15 6 0 R /F1 16 0 R /F29 15 0 R /F21 18 0 R /F18 10 0 R /F28 8 0 R /F33 36 0 R /F24 17 0 R /F27 7 0 R >> /ProcSet [ /PDF /Text ] >> endobj 40 0 obj << /Length 2693 /Filter /FlateDecode >> stream xÚåZÝã¶ß¿ÂyÓ"k†Ÿ¢ÔbÜ-pE Ým_ÒÐÙÚµYr%9»—üóáŒdÉ+Û·×K /&E†Ãß|p8ôÛû«¯þ¬ÜBI‘ÊT-îJ;áÓd;%´K÷ëÅ÷Ñîz©“¨nÛâ}™_/N£Õ5Œlêb•·0`dT?Ћ_<áOM/Úýûë¥sÑùªcÚ‡¦Þ2õ†ù=Ôû†zOþ00É&È:j›¢ý‰Þ÷#®Øæâú‡û¿,äb©Œp6%ùÿ!|‡òÔ¥‰q OJ2A‘¯©·Î³nÓRŸEºŒú-éUÍsÄÌ*Ãü|í࣢̻OXbø Ýo‰’Û—<óûÀ6Öâ„U¬‹šxh€8QQV–Hš\¤"uŒz4©p4 ZLã-™0Yx‘z4î…²n±Œ…ú@Ë~âDÄÒР’/'IEÛH'{K9ÓMÍJ %cXÉH0Œ6ïZÄ$5VXL×ìËÔ}¨›m±ÎÈäàyUowû.늺ú4ªÀª½«è-Ù’emÎ<»M“ˆËÖçIÏH¢¶¼fm@‡mzd³ñ`³0¬ Z¶¿Ñšõbj6éˆI¿mWuÃïz1òçÚ)ì“1|F:¡Œ•ÉŸç Ž…6®§ |ŒÕÑ-6&R,Hµ&< hÂÄÉÔOê*Ø%)ÃØÁ×Ì /tAÞ “T$>í§oætBëAÀž¨ZÓTÝæZ´ Ï^aÈ–GŠgéòfÛòüéÄÀL,Ì¡Y]0ñADPϯ—*ösýùu",¸ ¯.Î0ËÄŠÄ&¿ô$ÇyeXg{ 1'ÇR;Ž ­N²óÍ,2 ‘¹Õ€wú¤zìÁ€•ã“ åÕ¨w—Ѻ=UY?‚jmØÌÜÄÚŸ$gm׋ø ý—µŽLp²µ„°éÏÂïÁ[Fü曬á²=ŸìƒWú*¼â‹x¥¯Çë¥ÍxÉ9&µ|ZòÓü¼JÂdp,œIz×Y¦ÊDz%]ýÿkg–úÚÓ0wu 5°ÕHÎ2þ6»>HlŽÂÕ8¯á`¥¬Ð‰=¿.Üu–cRuʲҗ°ᨄM17ëí!@yõ§û«]aГ µ@»QF/œ‚w±Ú^}ÿƒ\¬á$lÂÀ×Op‹AY*ÜËÅÝÕwWo1×?š®g%!YL_©<5›‡œ3ÅUøØ°™9Ž©ð.™Âöå C2Ãߥ $NFÛÏ¡‡ÕïDï8=ŽCÒ¼2FöŽ3% XCüƒÅh3rA̯R­G¨ëëWˆæÂ¦¢˜ÿÍŸy% óŸû”#ï…MÌ(2#¸ŒU ƒÄ\¤†a7 c,¼L&‘ja´HUr Ó…\÷$Î mí9 Páx”ý1+ëÁ0ÌáÄ='Ø™5¯Î´Ôa£˜]9Ñ2B9}‹T{˜Ú5ÑÊÑŽ¬Àýàˆh•–7ûp4´>jërÏ9ö1lÌé°Ìî7°Ý˜Q^qýr÷>>åyëNyR…Sž”‘$iá°Öä-W(ªuK£]=k³RƒÚÎoÞ Üùôc=:“¼ j|üs2z*àÜ„˨T£,~ u™²ÀBô¨Æ²æElàØ"‰6ux?=Áfí‡í®«»P•Òæ°æé& aAé¥I@2¬ZÓép’ !üâzi­bC ¿ÍŸººúõ¯ÙnÓÖpj e |´žtÃÁúEµjòm^ᡳc™é€½Xzˆ‡©}YÉú4¿‚,ópx¿RÊýêÈ(R‡¸ÓœÐg|ØJA¡:Uà'Ê`:l¶áý?g˜*©–#ºSv®éG)X„Š$Oª¹…8À=€#ôtäDÿ6”t"QÆðy]¡òÀÀÞÐóàãŠÍ ƒ›þmWâ“H9¡Óh©»ÉšuùM·;OMa]À &oØxHê‚¥N” N”œßR1{¡ö:Íž{Ûì¹ØâÇûm@NFÀBs‰2çÊã1¦Ûºe vMýØdT¤9®Ã>e‰ ÖQÛÕ;ê±sTübS7Vµö9q¼É³J±AˆzânS´ÔãXWA¯–œ2ÌQa-‘] ßÏ7c¹öù´)¨´ „qmó¬Ý7Mð‰Ð‚0°Ú7?_;F²oxËxšÓq ÉöÕ å¹!ÔšlW¬Ã²áÝ#QÒ¬!n‡á_ò¦žC5kg]rÊÔ^ÚN k’!p<6¡ºüÔ .šŠä(þ¼­CÀ£¸{s‡õÖ”Kx8B;ͩCuUè|›— }“•yE#yÛ ƒœ8„ (ÚYö =õYõØEÙ»`!”AÆ4‡}äëM”?g«Žº[}—¸×$&¸=¾àr¯ õå`çeÆ#})öþèCá9@fäôpª…5C(ýúëY\!çPÃæ‘ív [à)°ª¬£7è8!XV±ÞgeKs,-þ¼áºö` ¡&l©tY(З‚3´k6Ü€QKOÅšÃM±ÊJ&¨{ºåoYêdÔ¼…u–dZ<¾Û‘©=ä–ŒODbP`¤‰S„ä#AH.€ Àžõù< HÔù<Æ È ÐšEøTÚI0Ùf Úm†7Ö©±ºñ‘0†4¼-Ç(ÊöOë×RåkÄ'õÑÝô. I5A аv;iY†$+L8>V|£„á¬)È)9ža‹[ªñ¯'(*5E^­˜(£fp\ún.%${{ÈPÕíå´âdÉ)qê/$’ÔÖqÅ)\)ÜR3›þXH‰2Ü^¬ŸF££–/„fyࣨÙWE7“©‚ÌHi|%éú=:eÖògnC‚~ÅÊrÓ}Ù4yϧþø4_«u–êçëAJN—¯½>Îå” ^ Íì±_ÇBÅê"ËaR¥É|EN4öP’þï-ó†w'±l?1Ö×Ô7ÏUçüçQžcnç¡{ÕIÛ^¼üHUšS+_Y½dfÔÚþÎÚ /†Z ìʹ¥{ü>¤Cáöz“ÝgOþhÀY7l•26S]ÿ=d¤å>l²g)û?/H¼¡ÇHÃt.û»uèÜÑɉâ¼zÈK~1v{x\…° þï'áa ;>-Š¿j‹mQfMÇ8M{:/çC‚o:†oö¸â’™ ÆTo@qÐð|p>uQ°'úSÎü‡R¿³²€EEí]¾:Ôù¦‰¿ÒþBÒ¯Æó¬M,°Âiaéc÷µÞ¤¯Çþ¢œÛZ endstream endobj 39 0 obj << /Type /Page /Contents 40 0 R /Resources 38 0 R /MediaBox [0 0 612 792] /Parent 37 0 R >> endobj 38 0 obj << /Font << /F15 6 0 R /F1 16 0 R /F18 10 0 R /F28 8 0 R /F29 15 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 43 0 obj << /Length 2174 /Filter /FlateDecode >> stream xÚíZmoÛ6þž_¡2q¼ãû†b@Ø0 Øš}j;@MÔÆXg¶³lûõ;Š”,+’,Ûr—CŠV¨»ãóðŽw'ý~ § ÖŒ£I´´Lk›\Þž½yÇ“+úã÷ gÂÙ䱜zë§ ¦¤Kn’×g?½¼8ûò¨8sÜArñ!AƒŒs•h…ÌL.®’7é·3´éûYFÿËÙ»‹ï“ÌY†ˆI²TQN¼˜ß³ À¸ôõ:_?¬üÔ/_¡mª°œq’ÌÃ#ú)gß^œýþtAB2ÔnÇ‚‚úIRG – !ƒÒ|=Ë„éÏóÕo4RÂ¥o9ê.k ¡õÆÚL:„ô*LÝVEfkeNE<~ Pã’¬ñ‡·œ‹§ ’`]X87CëF£™$a¼á`Ó„5•^Åáß“ÒÚèŠFеŸ$ÍÄóq»ÊZ¦< M[° náH´ªà^–;îEØ4ÔÆUýººÄÑ’b#­CŽaä7ÕŒ/<˜‰#hÀJ y"V3põSö)âG1wšç:Í+wÒ®–´ÂÌ3 6Êð\´9æsÍŒqL÷¬ˆ‹gË1>oޱ‡cšXkSçz’•°'"Y9ÃhlQ¤\LNv§<Ç4EÈÃàÌ€»¶ cz'Ûò¤öº%00OÙÆs³GØV† 1r"¶­ d Ú"Ÿ?ÛðÌÙ†ÝlùÝÇ·é ¨&bÛpŠ|ÿ“ýéÈ6çnצÔ\=ÙŠ’V Ÿe ÇÏ'7R3±Ù»¿EvU (t47¶È-|„-HÈ™£mAfé`R ˜sbc‹=é¡t§6f(ä0‡v"/hƒTãYɽJ€ F4Gû2®í½d¸ÙÃpI…³0n"Ã9Rn«Zè6\ å}´™ŽeúÌŠ4S„*¿¢¥ ·‹U¬¯òuîG²ö3SÛR£·€`àLt±VÖÐC$ÉÉ(ɨJ°Ë¸w&TÜ—×ó¡HÂs~´*Öa߬atÇYÕRÎgÞ}ôëB§ÒË…ïZü1S:Í—ó|]œ“Ó÷QÚü.\×µÊË|Ul‰õ¯ÖùrŸ]­÷ï¢Æõ Mú× é,J¤ùöÞX| X¹×rïJ.R¯Æß*nïoJãþò"Š+6Ë´R ^J˜ßrÉ)å®#Ò7ßtÆ^ÍPÕÅÐêºTòµ®¯‹0ÀzƒJÔKZÂÚ7nÒ«pû}°mû©âî*Þ)»<~-1ø"R·–_>'hº—>/ù^‹Gá΋_K¢FyJiù’öÙEõÄåâæáönf ó2¡Mo‡4bt[ ¥6.>îÕ‡BîâE_ççË« 3_áfmwØ87Å*ÎX·&øua5c(æ«­ ïáZ}œÊ?bªGVáÞ¢OÈ„*Û"ͼBvB(hˆ›öT³Ü6‘`6s8ïiÄøÄ7Q‹ˆï}Oþý¨8¼½Õ¶ÖÁ< Å û™(6~RYÚÝ. /}îM`*žÖÔ&óO‚¯«­¯b¢|Q_ºòxþÒ¹[öªZ ¼¦÷ù×afT¦cÊWþ[{'¸t¸ûÜvJͤs^¦8µ³–‚Ñ¥‡òcrø(Œ®²ŠlÿfÿT ß$=ûSÁOK…5Œ“¨i¨¨„Dþ§¨h§–-*F½íŒ€23Š« AÕ¤jÂìS™!M€òë*ð¤QîP>ÕŸUè§Øð‡%cŠ—(Ë*¦ª“Ÿ.£þ7¼SŒôN7ŠM{4›©ŽÓ°YÉŦ|löº½e`ñPRá8R¹†ò(…“ZË:©ÇÅÜ %ýpM%ÕV;k°íd[3zg¶-˜ ÂsÃî\²Öùlü"¼O:Ã#&@Ù™¦ê¯.:qßlL«ÀáÝY¸¡Swé•M½b/½`åDz±O¯îÒ+mOJèûŸMµ­ Æ5ó¯6üG¶\ö3*"0Æ*iJ1+bCÓí ®Cp@õTÂQO«Š[Ý ‡ÝK­ØK­4#YèÙìµ<Ütmµ(ǯ¶%Q‰ö7 ²ï]‘òÑUZѨ`Û!"* \‰dNÄ©ßõ¼”5›ªüEg ®(ê©á‚Æ+4Z•»øà—‚|P ~^É¢ZÝ:{Ò¯”[ v<µ0°lùE€¤s›yT¢ NxÈÑçÏã_–L€d-ëtHâi¤ì…›æéuXjH ¥# ”È´•ùaÖÆòhÌÃR2…Œ[˜ÌZØ Á<ÒÇ3ŠBÚ¿ŸñÈÊDWáò€X—ö endstream endobj 42 0 obj << /Type /Page /Contents 43 0 R /Resources 41 0 R /MediaBox [0 0 612 792] /Parent 37 0 R >> endobj 41 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F27 7 0 R /F30 9 0 R /F1 16 0 R /F29 15 0 R /F18 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 46 0 obj << /Length 2627 /Filter /FlateDecode >> stream xÚÕZY¹~Ÿ_¡ÇÄ¢y1 ð"Yl$ž<ë ÐÖhÆZèpÔÒÚ›ùóùxt‹-±%ÍZãã©Ùìj²øUÕÇâñòæêùL%Ž:6º¹q*ˆÑf¤•"ÌêÑÍíèçJ\ÿróãó¸Í%‘Jh”X{‰«W7Wÿ½b¨£#6bÖ %…–(QBŽ&‹«Ÿ¡£[|üqD‰pvô!ˆ.F’m%ŠóÑë«\½tr=VœU?MCv?‚ðs’Pf“>•sUX·ÀºZöP@ÛˆJ/¥jVÀCLJi/J#-és&!øùLÍû~©fDLU™Ô[pgà¡\µ k*e«fö1ÖÌ–“õtÜB›&I­zÒ®*ÐTtma÷>rx ÝÌgÍfÊ=ź›k$ÞõÛù4ª/hßg@Ö»uê÷ß©ˆåY"\nìÐQPÛ÷Ô6ÕöšUIs2QÉ&ÊÕˤnÆàq|°¢kǧRìíXDQÝšQ ,Ô4[@®TõëtÅ·@ýþ[*îVs¯ÔÊkô¡‰ë¤;~KËnpqìëuÖTa3©¬,µ"Ç#[ ÔçÕ†ò#[ XÏ ‰•ºvðµ‹í‚#ô¦{ŽXbdnBòš§i0“ÜÀ‚Æ‚0ºýÛ€Éá‡YÒAÇ~'‚Y{ŸÞ¯ø‰ÏáNœ€+¼½ÅQ{#eD8ŠS;qÞÐ d29–} »Vü¤“ùL7ës/]i,ïs`I¯Á†'á͸Ó>’;YyƒÂü—‹=ú\Sð¼rº¢sc.‘ƒX±˜áç›`e…& ‰Å#< SOR‚%¾ô¼¢ì‹bž&’ªQ&Çx­Œd¬N›)‹at8Úe—B«XÕK\.‰Ž8ÆstŒuƘ#èÈat¨Á¤(/} 4¨ôˆ µHy»|Æž™‰ÙGfb¼¼)Í”ÎC2îÇcþ÷§TG|]꧃<5-"ö Mî€\"§·É)Éw©AÿŽcÎW—‰NáíE'2šÉcóÀh_\è³\ús«#¾.uäc%žf=¥³8`vM´µ}bw‰Ø)âê!„q'Ž0»Ž,c Uê2ÌŽ…=aüÈ¡Qï€/Ëî÷ð÷Ü4Ìþ—C_žFÿËù¦>N·:§[…ÔŧŠ.§[¡]êk4¶°2ÝÛ°7ÃN¡±ÚD—q ,Ñ5eßœW|Ú$üT›EòëRG'Pó¹Ôœˆq˜¢0¤ˆ­{TJµ5Z'*ŊɆ³Ñ]Žb‡ƒF:B™¹PЀũeß*•~+Óü—QG}]Ylj ÑACûAƃ¦jÜpÔ¬[¹Pf/‘’}Nœ ënúÀeS$§Rv oMøþLÌè0>ÌA%{!|¸&ŽóÏI¾9>æ|œ çZc¿CE~‹Ûª™¬Ö©X/¯Yu˯'þHàÝ*4,ï¦óô!ß*禚ÔËX¿;’Àc^Ïnã÷Õv«fI®Žõ›úív^¯cÝ]ݼ›­–¤=€ N˜Þþü«:ªqæªt°±þ=¾†–±6Ž›õ¨Ø„ƒˆP.–1·ûöÛµÒU=ߦ×Õ]ùvˆÀ”ÑmÝÞ¼ ñ±xæÈ…êyûÀa¨¿wÑc…ò зÙn9×ަݾ^ûÛÒ^1úˆÎMaxFr½áýZ&½öhR¿ÌÕëðþ$â·‡§¬÷)‡½ááeeÉ|CGÆú4 ü òâ|9žZÁéãa•îN˜öUÑœPÑ!9+O|'áÒâ%8•Ó¸þÑà4ƘôÛTˆ¶ÙbZ´ ‚Óð yUo*%×c£iº¶ Ö+¯ÎÿâñX4±áj6­\¢ _¯r·û˜¨Ñ̯}ˆuÛP€Fþžåsd:;€VÑ«&›ø¦ÓLet²šoËxo1ªékãÑ"ª:¹2C"§cÿç&VOw„‰Úi<%mM䫼‰"ßÒ¾Ò/§ó;¦<ƒîM,¦“cÓñ+ŠwåK,Œ"hFVƒ—G9cùíQfÒíQ‰'ïFïƒqe Û]HXŸ[×ëÅÆ;ÐþbÓðš0Møãìô¬ã£™ÏîßÐæ¿G-q²ôß6ëÙò¾Å4N\~Éûpn¦M@œ¾0U†Ò4–Éxˆ¸Ó@jÙÒgþ>2üj¶œ¤ÃT‰Þ6qÄ6œ¡o¦ëXžÔM’«ããýzu¿®éëjŽ…¹Þt÷âÙv>]/f :……‰¾ñsüd åUqFu‚hÆZËA­{Äs|?xÑXô§ºó/u£ÝÔ> endobj 44 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F18 10 0 R /F29 15 0 R /F21 18 0 R /F1 16 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 49 0 obj << /Length 2686 /Filter /FlateDecode >> stream xÚÍ[Moã8½çW訠Ç\’ÅÏã6Ð Ì`.»“9MÏÞÄéx‘Äݶ³=ûï÷Q¢¾l‰’c;ht¬Hr±êUÕcÉ|»Ç?Qï™ò¶0J0穸¹ùíw^<àáOgä]ñ½zõÿ%nHU<¿ÜüýæãÝÍ_þ&t!8óÜ‹âî±Æ0¥„Yæ}q÷PüV~úÏ­tåæ5ü<Þ®ˆTy·}ÙÜþ~÷Óͧ»›o­&ä<3"“&x›9«šür¿ÛWã+)4sŸµ˜êñç:µèHÁ´ñW¢##'!Ì2cL=ª¼…2Ö—„OiUið©Œ+->µ±¥‹Ÿ>Xå` ”ûÇæ°CÅfˆªÉ¦Æ…êÌ;ƒëɅӌȆãU}!®àœ9e/¶ÖˆË£' “¤ Hò“c¶ïq mEjLU½Èç‡äÄ‚ 2åÔ f*AÌ*š5³}ïz3;Q§fžúᎈãºXmd““Ž×Ž3-/æé9HšY„!„㘟9çQþ)Àùƒ}Ì ¥Axâ<#å® V”dŸÕn:@²À¡¸bÜÐR8fÐhœŸNÔ)3q$&µkãÈp†ë+ãÈóRBÒB!§cŒ1%fxçÇÀ å`-ä%©%™¦8&ÉD$åDqúÖ‹sˆ(–‘&"¨3DF¢ ú;7d%•ˆ&ò(4L&VxK—'´¤³YHII†ò¥‡Æ© "8AÈÚ/,p›1 ‰á9'Cßaª*ω§™:P䲘Њ:3a‚ŠÞ—L§T¤™ö!´53ÊÏOâI ;FFšI´ù @­?RtÔÈê¡"P;L ÀÄ£û’™&$åCþº+1 ؓЕ¤¾p¢•‰¨â!(KTñ°Aý¨*Pßh@›æ.²šqAKt™ã®NTO—D3õNçŒW™¡‹Žñ@”˜@çÍ\>…öí$/—O¡6±ÖáË*O£¤<ÂÅÒˆ5³¥¡r–YésL+Ê Ö¬Öµá™’L.–—™Nº8…d0¡Õ™0;]ê…oä0¡ujÂe´±@•¥´ÑW%Å9¤©Ð'È ºD¢B_Èb‚é#rQ‰îü¢ÓNd⣘±._멌­–•rð‹¾Â,$yie” ˜ëmèDõl¸²Ü½¨2ÉaC+jÞÍ”º¼dOÓ¼‹‰©oAŠ™r¤VËL RkdIO$˜i‘åÌÔ‡ä"jò j"Ǥ¾n÷¥õ™r]¿Cca"sïYÆÓ>› ôVG«%û„3'ìl–êçìlêã vv¢Îìq½‡N¶™Oø^` ¢<¾}iÓ§¥|/ ú:›Å÷2,θ¾Ï`gëû;[‡]og'jÆÎf—:ØiI]²M}ämµF™c9Qq‹²€fw7Œ“αœ¨\«þÎÝÙú}xN|K÷rXE ,£S7¯¦¼w¬¶šYá³E Pu{#Y?×Þ„eŸËFÔ@›uårŒo”`Zë,DQç Æ«Âda«˜A•(ê<.ÇÙö2¾Éª818ó²<õ:ÁËó€,åå .±sJàç°Îº‘T¯Ñê¡MµvvµÃ1(DÈ`"щÅ+àl0¤ÑŒ\r¹Q(æ¼¥èE¡ÒwÍ ·ýH¨*#xý³] ŠI7ØXêBj0$pª&Én¬ë(tO*y¬%öµJ}cöaäzܾ¨`_ ÐÂ$p¿ü”€FÖe– Ìæ3{…Þ0%–E£gB¤6OZY}m—A8¯a¿Q¬ÓýÆ ½6OœUYìme-±WÏÙkNí•({š#y{‰Òvõó(µ÷\×Çíḽ¹Á: r]‹”_ÇÚT´Îäš7>sÍÇ|EÝKbt­ –2=I‚Ý®™ò¯‡úè3‘bB)ôÄy:ºúa{¸;6@CIY>¾íaå>üBåú_€Í–»[ü¨Ïrÿ€J—_÷»/û[Q®(õW¿oŸŸë«ýæø¶­eÃ÷¾‡»úaðˆ­=b[„7{è‡÷ÖÇJk>T÷Ã*íõ‘òÝk¼ñ¸Û×›u=DýÛá Vh]þ{s¡ƒ‡GOë‡úbýZ?ÙœžX§ Fõèø´Žw‚ ,:ɆM6È?T '5DŒFÉ$ÐAœ–Ÿ÷»0‚Õåú+°»%Qþ±}Aààöˆ#yu<]7åÛÝÓöâ——›?Ö/_Ÿ7õ/Oëxw÷úüßæªz& cph¸õ°Y‡¤—Aq€&=.åîôëU|W›ýKÅCˆ¢úV…îà¥,dȘ¯ú€Õ úúF“ë×/VÇÓÝSôáËzûZ_}{[×nØÂ–CýõVàÓ¹è¯U†ï¶ßí£Øuð¸{~ÞUy`#³¢DWÆ%J: ~²Wù'0£1†7Ñ“ãQÄ$è6†ÆH¥ VUȸôéÄœ>iX†ì!Ñ<(k :©ØÖýüó4‹j v§Ž=úƒ†Œ0¶|Þ}‰qfŇޡî<ùñpbÝ·çäX5*ª4+;]Ža63ÜBGÎÒžœ¯o¢(áÁž.ïü8¾æ^ÒýrUŠ1† xÕÁçX£ B¾“Tuå“ÔFÔŸ ÔpH 1ýþ6GNÄ+Pè’ZʰÐFQ}håõåé¥ÐhM ÎY‘èŽ!üš` +ª²«Zs­˜B%Æ´Óf"öD0Äq´‹ê:¶hdYbV輑]Xй¥®°“ Xr˜ÓÊÊbÎy4 I îR¢NåÃØr½\>*Dy¤} Y} òeÔpo%´GàŸìWþ˜H”ùAêÕÅŠ•¸¥-š&Á·LAáÓeßKK”‚a8Ìéâ/ œLå[# ÕCw¼Ió± µLK{AÎUÕoý–/Â)zkbœ iIã§¡Ò¡?•hÊxÃÂ:“ªNÔŸªAü!úò`iIÓ‰€wŽú±§%á3d´u¨œÔ]@—z:ß:BñmªרÛÃĪ¡Ñ¨Woý³^-­˜ðwÒ݃ϜS‚ˆˆC’Ì3c+Á„Tó36…Óé&ÏŒM‹íÿuÆÎ`N'+‡9Ëgl¹hÆÖó3v ZY= 2ÖÀÃc&ý7*Õ•âP«)IZ ÿ`¹— endstream endobj 48 0 obj << /Type /Page /Contents 49 0 R /Resources 47 0 R /MediaBox [0 0 612 792] /Parent 37 0 R >> endobj 47 0 obj << /Font << /F15 6 0 R /F18 10 0 R /F29 15 0 R /F21 18 0 R /F24 17 0 R /F28 8 0 R /F27 7 0 R /F1 16 0 R /F19 32 0 R >> /ProcSet [ /PDF /Text ] >> endobj 52 0 obj << /Length 2647 /Filter /FlateDecode >> stream xÚÅZëoܸÿî¿Bý&£Y†ïG‹àz ‡¢œ‹>.W@Þ•½Êí®|’|Nú×w†¤´’Lí:ŽƒûbQ93þæ¹~{uñúS£ÄQDz«›ŒIC ™VŒX+²«MöC~UíËË£NçßwEwß^þxõÝëwÜŽwGl¤a˿’)q4¥ë—ü³Ã5¼ºøù‚Á$ÍXÆ *&Ó †e.[ï/~ø‘føø]F‰p6{ðK÷¸\ú‰]öýÅß/Þ>>‹`ŠXç¦gyO¹N‰¿¢D»Aü)ñ5áBõ+ÞSESt$ár Ó¥Èâ„‘aImZÝô«6iq ÐáÄ(VýWe+F¨fÙjôá=¥"R°™!ÎÐÀƒ(£q½:,¤)>øðl´ê…>E“Œ©=ôi*Ð0\DS†Hžiaµ&°e—+Á¸Æ'—’çm”ÉÙ 4 A(ç_†Fn4aTM…AUò”. q¸þ&qyŠp®P¡Ê…E<¥pC˜‚+-û-›óœ%Q‡ \fl'ß3èE$ð'‚×,×<¼æxÁ›i‘V½ë—lÑ 0FSs70f¼À˜1K1fšh¸åÇ˦ü‘j`÷&­p©ägy%¸È'Áż(\Àˆ?3=làž­$` NA”tS ¥ÑBi´P~ÂB™#Òˆ3xœõÔgm¯¦©·`QqBªˆe‘,Â)·3aÒ·†Ç[¥¦‘CmÇHZ §ì³oUù[]qcÏÙ˜˜08ccú\¹%ÆŠqhbô¼†­L}eÝÄE”•SÄ/¡L.£L9N„‘/‚xe9diQèyÀ+LV (¼ˆ(ÚÀÎT:SË G 4x: /"‹RÄ8d NéDÚ  ï4ð|Y$÷1ÕË"àrJ/Âfì:%‹¹ ™ ך|ŽKzz2tÖèØWuH"™ß¢¤“\ù¼vžæŽ^\/Ñ:ÄÔ©gK¢¹ûBp)Mµ™¢šP'cåyiE^\ï üBäâw—+)e˜.Û.ÌnŠ®À‘ì/DÂl\jî+;&($uÖ€Hœ æÌW@ÉÉHɨžŒ’×KAc@WEä_m«ä¡2\ÐËæSxk«}µ+šðÒÕøyç™áL`†#ö žœå×÷ñÓCÕmã¶mÅfSuU}Tê›0».ÚøýÁ³®n·—Üä]KüÓéÕ^mË–smò¢¸ð¨¯Û²ùåR©¼@.í«dôƒ„ORs²Hv„‡zñ܃¾f¯y®k”j¥*ºò: š‡MXð€ßÃy,œ'L6Åá6Ê}ÓÔû0bá Þ6=xr&A‹ü[¤ch~¼¼©›À—绲[*Ù8ib–ðqïýMùñn)SÜù(Ÿgr¡<’fReÁ—¨‡ ”áã âgÆš€«*‹nûj¶¿öìã‘aNs&JRÔ¯£CJ¦- ªXH*VrÚØô‹ ç?®ˆ©qãŠ8¶,9$[°Èè`‡Nû¸ÏYTf• ÷¤1ê‹z³ž…KÍ|×C@±JÝɼÐBżµ5RÈ‘M¹þ-!ÃЇvF‰Ô1ö}Hi 2Ø}—Ýþ Ì„˜¤AÒCaŸÝ1œi†c†Uº¯(,›1dg>$bÛ9uÀ9™ÙOLõ=°Ù¨ÈF‹Fqc„ 4…]i"$~“>¢5“ÛCR0jL ùSÍCF …Ô—‚bÚ¶+‹MÕ;ßà@eð#à„êÛ•wDÖ;¢„:fQc§]ÉÏ÷>AIE”;„Š>œÌúüÌl‰ pñbviLM"0=Ûç<Ëõ,–›¡š\ ¦L‚»ƒÄ>¬?<¾å&Û/ÐÙ™ëÓ‡"%¬bÄ1½ÁqeÃýO#}g)ß |8¤ç0Ùý'²“#:¸A`ëõIKpÙ%Ç8鈴|Ú·ó„"žOH&é¸Ìô -P¢/„“žLÃ¥€z?¡pÁÂ>ϳ|Íß}'™æØ†QºPZš^`uI5ãȸ_Æ\Íð“–f@ƒÂðeµÂÅ©ê>,µä(,䙿jÀyæ-}0Ò$HIâ8aðs*50N 1Æ KáæÛnÆó"µX¥„µµLLùIÊ®’iá\A›¿¤ò%pPRüÊ~aÅ%Õ¡ =êÕ8C8D·¹b*÷ -L³¥´yÊ–ò¶Œ£žÖÐØe0P|„¶|¿­B¨õ¯XPB†_†q[ì˰*û]wíˆr}¿Û„ס.Å—úº+ªC¹‰•d8˜ŒªñkÜŒÑs ~·«ÖÒ}¶ÀY¨Œ¯àmÖ–“¡X€Íí]à»®ÐS–›ðÙWX÷{dåj"±›° *§Ò+\æÛj¨Mà“?_ªCª‰rSøâÇòXèÀ`_wUheD!a.?0ÒÂ`S¢”ßÅ:Y9+³Ù·v‰@¬îqîmÝÕM s3‰ÿ‘ ‚h5$8¾–šÂ™±‹çÀ 7ÕÇTb Â@âOI.P¦»mµ¿Û} tñâ‹u^B— ¹ù’GƒDáö›ª8ô«û®œR*°C7=îõˆ iUÔ”ÓyÓâáXÞ†¾•Ï  úp?Ö¹üÏ鮳á·ú飘û•ùïAH­{O7Ó'¶|íÔY ÌG´òxú=?N²qÑö<„UÅn= u?«©6#úô`Šå¥ØG >ǽ;9„ Ex¬ëý¾ŽS¡]‰£i»2r¸IþÈŒ¿Âp>‹¶ó£ƒ"A’õFªU”´,ÚOaÆ$ð¼k|›#"#¨@ ÿSÖØ]uÛ"‰.A.ý" àŸ‚YÄÀ®ø›¤”Øo<*›P»KUF> endobj 50 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F18 10 0 R /F27 7 0 R /F30 9 0 R /F1 16 0 R /F21 18 0 R /F22 53 0 R /F24 17 0 R /F29 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 56 0 obj << /Length 2317 /Filter /FlateDecode >> stream xÚí[K“Û¸¾Ï¯Ð‘ªµ0h¼‘ÊÔVœ¬§6‡ÔN*¯S%k43¬ÒH^=Öq~}JŠÔŠãرO¤HèþúÝ€^Þ^]¿9J,µ0º½“D[3R“ft{7zSlçã c¬ø0fº˜—cfŠí&<œ®çã··½~ÅL}&Í3¢aŠß‡!Å€H|‡À œOÈâÃc9só?†é7«§ù¶|šûÕx±Â‹)f³Ý:._.ã¸ÝúW÷’ ÅÇølúô~Q."¡ïñÇlº-WË qôŒ&†ÍùhœHa?SIoKœžs^̦Kw#Šw8‰v³»Óðò]ùnpñ÷ërßz¼ÜãŸ)ˆõf~8&ÂûÕ2¾Ÿ/g«ÝÒƒ9_ooÊíÎK"ÈA ‚‹Ô3A$²úñÇÒ8ÂTHßÄÉl2™!BèjÈ}nI„Ü‹‹å(JF8îs+Yú0†Ò¬jøÛjÔb…8ƒð¢Éa"V^Ç ÍH«i$Œ19r8‰„ütUM@êѤ6ì» :ÅŸ¢jšË6fTbÏFœL×ä™Ã(QTU£òóÝœgÅ©W¼!ÎO.¬Ü,a‚Ÿ›¼· Kðs4uE$Æò]–etVt"'ú Ïû¤Ð§r¢ÓƒìGÐÜ$Éj_´ÂQF,2›*û|.ëÐHs¾‹ƒKj‰U änÆ ¼h•/—ú[Üù­q‡ÊÏC(£tÛ€` hzçÓÊŽIÍê:B\¤ë&u4?¦s„j¾úG–`CÔ!;óš^'3A€S"U·‹44vÅT–D4G»‡óßV„qägã:Š>·œÀ<šj£<(Ÿ*d8†².ŸKTèņ!‘0ÄÚâ'^ç/dèY$Ä.P;Cì‹cˆ !¡gQ9Þ_å|Ê ½T§C¢m83mÎÌ|˜›(Ñʪ¨'Ÿx‹ïj£*Wzzù4î] :n}•YŸ\>Íoû-ÏûknNµ€cU®M*÷×YIJ¢@4bÈ'QÑë@Ëa25hæÖ¥§Iñ¦³WÑ–¿‰\1õš¾ ꜛˆàM˜Ò‘"0\ jýÛÄYít?<éžeÿÍ"³á> ÔX/ÔòQî´ßÄE4ÿMŽë7õÙáÆûãæƒ)gX™Ž‹ ¿ˆiù­ïq[[lV‹ëº†_³Õz=ß¼÷MçÕònžnWÙø€Y…8”˜ÙDßxìz¸%~\ͦ3)¢…Jüä„iã<¿»Ðã §U3]ûf:^V ä¹|–¡%ž­îŸxø@cEgBÏ\†åãw‹ËÕS9]¸_¢PY™+¤›ÛnŸË‰•±!ßÂ>5ey†æLXpemU‰Ê³ÖÍšÀñºm6`‰ûLkèN)zQŤ ˜Ô½#TsEi:+‚d1n䑊2+Š•Ÿ$j•A‹uý?ÿÁ¿ò)jºªÚ¸ö~Ö5[ãнuf?ìw¹ëј+XhÔÍ3¸ Šþ TºñD;›ôëi¥Ô‚Œ'BçÓ–-[s`† ýÕ©m#(À¤<  VŽrÌWìÒ 7h¯õ< ´»…þ|+­³fϵ ô<.ƒB±ã"}ìé±ó»q5=;…£ìÆñPáuIP™çSSŒ– o0ƒíˆq¦Uiü%ßC÷Å»I=x/ª…Ì© ?ÓÈ瓹n‰Ò:¿‚Q Ó–Яû(ÔÐú" «ƒlØ‘Ë=ŠýhüĶå/ù2,I³ºQPJ*Û15ü¹7æ¹~À?[¼-V-†ë®tC%5µòI—Õ‘w³žÇÛݲüeïKYL»1„Ó›ðø#ë„[ÓÞÐ’ØfÞ¤k?`ºÍ§½„šƒÛ|9K)4˜ƒÈóWUL—wáw¬Äß¿û"/ÜV™Hã&À5(Çi5‰E½¿æ8¯VìÓ÷ùl[†"‹cËø>~¢"Êøm¹œ­çOóeuÈÅ=ÃŒ¿>H³Ý“Ë«w‹i5kxþ8ýÏt}WU)u2=ChIx~0Bstažt÷ô&\䵉ïpª51FùtéЃC@€¢¼oÝ©võÃíÕ/WNÉQ(ƒ*/1 hötõæ-ÝáKÔ ‚F9úà‡>¹áÁÏ/F?]ýýê¥;“ÕÜÝ”Xàd–úm`OÀßZ¤ª5;ŽÙ¼@`ͰÛ>ÿ_É#@Á…ôðP8‰'³HµÓœž ¶°×ÐXÄÐXÿÿUŒª*Z„‰y³À„ù´0ƒÅ·.îÌÆšÆâ0ÐIè¬ n:Oe€MC§ÉÆ´¤ní7©Ì”)ylij(ºÉ—NÆžºƒ®gA(wNC{$eXAäê¬F»GuM²N“1`Š LÚ˜¡5ù“ò^,¤ -‚’pÌÁ†¤°DTÇ ‡²¨ú¢Iã“:kè\­g+[N±(«úÏiŽò\a³Æšlvö U²ñDk­¸ŠBVË– ùOÞmÂI!»­H+22cª×7!?‹A™TÈ?àÂ@ùI!ã­y !c¦Åᛌ‡”±’°»(cZ5cæþʉôIcÆèØFÄÔêDýÚDœ‹çÉVt¯~Zd÷–B4Ã) 1Ý0ü6j‰¢´L=ÁŸ»—¶Fòj˜ƒHMKÔËozò™é FÙƒžÓØÊøK¬¦Ä =Á²Z"ÜÃè‰SVž¢‘híožqH†?ƒêå·ÏU=×åÝwb@]£ÖŸØMT﵃Dñs‹ê)A¨0©ž2€E.ê‹P=|xÕ³†š¨zÒík·àψúÁ?ùî¨ïQ&PÑÒ 5:ÿ¬³;¡Z¡Y`e&]BWGûý_uøÑM endstream endobj 55 0 obj << /Type /Page /Contents 56 0 R /Resources 54 0 R /MediaBox [0 0 612 792] /Parent 57 0 R >> endobj 54 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F29 15 0 R /F18 10 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 60 0 obj << /Length 2597 /Filter /FlateDecode >> stream xÚí[mÛÆþ~¿‚È' ±6ûþÒÂì6n E‹;´â %žÅT']HÉ—ä×wfg©#u”Nò)ö!Œ3©ÝÙ×gf–¢^_]|õF˜LpxÙÕu&¬g‰ÌÉœ’ÙÕ4û>ÿv: l~UÝ”£®¾»øæêâ§ ‘qø'2)a½‘™ÕžYë³ÉÍÅ÷?ðl “ßeœ©à³»HzƒäвyvyñÏ‹×ÐÂ2l_€Ëɲ`,¬d"¨ÈXïg,½aÞ¸ßÉYØÀ¸Ï¬òÌyCŒ_Æ’[Ÿ‹=Z+Íœ|šÒ xjmú¼ßrÃ%òýê =AAk‰b$2ΉjKxÛR •vò»i^mc˜6f³K¡ &†“ŽI0Åq‚ ÉÕcºS,Ý ɲ±4ŽY)³1Ú¾ƒ ®öbÀsf¤:¼fÆûCÊùb€Ÿ1p<œÓ} ¼A ·ÂÖN0®Â¹ü¾ëÆ­[ø öBFäj ¬“æD€3‰>Càw^ž3×½M§ô¨À¼³'B€V PxFÀ¹|rìoˆL%ŸÞJã& °‚+1þrwOø>’G µ¿SÌ "©ô‘Rµ)J0áU¿J‹oÃ' ìp݉r”°ÌkyNRç$õ\’”IéO”¤¸b2¨N’âÏ2Iñ皤|` çg°P²ãå«ËÑX©×åj]/ú°š•t³^Œ¤Ïïð¿²z?I—¯Êi»¢©¦ëbžÖéú¾úÉiø`ÊçK£‰?„–yµ"‚j›çr!n>r‹ÒCr ÊõšÀf}3kaóå5];i’ÎF"/VéÃr•&lþkY/_$Óe u4~j™46sÌñ”Ãÿ1ˆ]É”‚~´Cw—è,ŽG2y™Ué‹”ê!KÃD0™fž§â;b8Æà†Š ›tÿÛCXabzȲ °B±Cõ’ÌÂ_ÀU¹|^ý/:ê®jJš¹^Öt„ælâÓq¬X$Iî™ô²_/'¸×l£ \\—smúÎ2àð#,æë"R€ëp_üf^z›IhçÓ¦ÿÝa%¡1îнåBYÀ2§7)‚Æ^¨ü’ä–­€ ø¡ „ÑXäðÜÒ†¢ÕÁñÛ¹¾)ÞáŠ_^åÓòºXÏWhq/ó» b5Ž º.oWÕ2ÞK]#èž7lnÄQ iu{ z\åÿž•Qú:m<‹ï€!zrnm‰ p[DqMWâ.òC+ÌAÐÜ9BGn(f`dÝ”×ë9ÝOËÛÓD¿g,(ó,¦)lI\ÜÞΫI&‹*góE &-A\e¥¸çÆ·Biˆàºœ¬¾ ¹Õ Ms”ˆ`h‰;€ùŒ¶87Å8†`mÒÇ "Lµl"JÚ:R&ÈN¯)–«Ƞͨq%ðJ ]š´×Î*}$¹ä&vh“¨Î.éÚ Î=Ȥ‘¢ZÙ€nZ‹üÍÈ«œ˜îŒ‰Õ¬ªcÞùò]SÖ°‰É ŠQ¬â—S>•Ѝ'Œ5%Æ9Þ‘j‘¼Y‹Iù¢]UÒÍ®x¹¾ýp“†&ÀqÃ1QǪósu“•ÊF;Ieˆ%‰‹Åû’&*ÚMèÀüi³N`Jl²ÄÛŠÞUó9‰BIr2Y×I”ë$ó,É6)š$ußD4†öˆTèv g‘¿O襤„¯è­¤t]/¥*h7àB s“y„Isã~¹˜V‹÷}ºi9/SÀ',Åñº(‰¬ Ç›ºLsˆÂf ^…Xú Ø|‘¶Ég)û?ŒCô:äÌ6 í& =$¶ò6¡ªŒ‘´Š  *©¸ŒäÒpÀ(%-$ìÇO™¸DÖ š‡@',´€U:¹ ‡ú%ÀOë–´L½ Ï+Yºí¢m難4ì‹¥N¼·ÌÄRI½o¦ZË^Å ÄŸÔåMÚª¡1ä„WŠDðâßK¨“‹ñ«bŽN•Í*Šæ·h0Ö¯fœ,ê4ØÌ¢…ï4ž 7t„Ì^·.»Y6hA¿œ¯o4mc«âݼ$*ÞëÞð¹ûùë¯q踴ïtÆ ²òhuA>÷jã %8.e,ùTß°œ¬ JÍ[%á Ä·žlþc[©'¨bh;¼›¤V=ZWÈy¨E…{¨ú­ÿI4ºÓ-Bz :SpNK¼2„gFì<7dÉ\ 5¡ ïoBñ ÑmBñè4¤ŠfRožW¬†· úþd…gy4ýË䟟owî NVúclô@B+{6:¨ï¤óbkDèÈ;F¤Ób׈ƒçEß=Јê1#ª°eÅ SÞ mÞ„›f]¨¨üÏi!0NGíâù´Ô XК!áô÷v¶Ë ²Ì²†;(¤Üo` é9Æ”ta+¦zç|ÀÅó!yÓ!Rƒ§GãÌa:ËCu–Ç錱Eô)õaÛ´éÏR‘Z¥…Èoˆ‚Š’¡"žVm²UÛòQ—;”¸vIúé‰ËÅmº |I>ä/Hð‚^¬É™æ[¯Ú\aO\Ö7‚ ë—ƒ&µLÝgžé€¨†Y磨¾‡âþ>’¹`HÔDö/z9,Mši»I" ŒbÞj<‚3eR$ ¿³.œd<¨LjÐÚ?ñe9á<ãNÀfmßtd¹§ƒB8Ї(‚Ü/‚<Ùø`s· îAw©83!t ïP\Z¨iúDŠ?`úÄçuÁ·(`Riº¡&‡Q!Œí…Zzä=ü”:>µ Ûl·vÄgм·™k:nɘÍ>&äXp(l[#È‚¼KºÃ¥B²à͉\ L¹~^. ‡¹4|V—~œ|§þ8ïøžÅgq´ºæiꎅ—m¾Ázw‹˜VöDÌð¿çø}’ºnHÝÁ¯¤à¸®¬>ü×îï»|ÿ™ îh]ű* Yàã‡5OxUêèïí·[]‚«púOÙé"‡Û%d*”?Q»„?!DŸ<Ö. ãáˆN“™zL ƒíÄ¢?>ˆ÷$¿7ú(¡uO…¥Åg‘]Ok„ß »Ž…‘é·‹‡5|ÎþD-<> Üœcò“¶%Ã%¿÷NŸø¾ÜñއuáÀÊ/?yç*µˆ¿{;¸s…ª)ƒ8¡Ô)cÏX> endobj 58 0 obj << /Font << /F15 6 0 R /F29 15 0 R /F28 8 0 R /F1 16 0 R /F21 18 0 R /F30 9 0 R /F24 17 0 R /F18 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 63 0 obj << /Length 2487 /Filter /FlateDecode >> stream xÚí]o7îÝ¿bÞnõÊúI-î!½K€‡‡I Œweï\öÙ뤨™]Ëv®—´z/–†¢HФ(’ëï/Ï.^ SÎ<÷¢¸¼.„–Ì*STF0i\q¹,Þ•¯fN•»vöáò‡‹WÒá{æ¬*8!¾çBÖ UΤÕÖ_gs%U¹ÞÝÌæBÃ&Ã%ü®DYo—9F•aÚU‰Š.,ó–G ÄëB1oœ:cTˆ1Ü~†U­ƒÆhUý„U-\]®¿U­c®òOZ•"\É iÈÌ‚ûêÏdæ#².GT¡°òó‰&M>tA„žu¥Àz`÷¹T$Ô›æS³½!kmgÒ•ûMh›Y‘ÞØõ~x·à…ˆœq.¤åÕùø ¾æœIߘ’Ä Æx®ÚÐÝ‚J\}ó Ù…õ=>=,iȽã€ÄDu ª Q“àGBšòåu»ÛEkÊúö¶ÝÍ”(i6uß8st$ÎðÆé^&J(¤ãåË™(‰.~]p¤{Œ½ØmpõªÙNVQ?¸Ø7a™ðá9ïÅm‚-ê.¹‹hnV8ö]L 8eé]^ÔxN©Ë+R}DáêîZ!'„5 ±îi\‡ºKÓ#¯⼞I )Ÿ(A>˜kÄÿØ&1o‘1U¡± eJÁ`B NB’hH‰Úp»nƒÒ€åb·ßÛH¡Ê˘Tº„¬$"ðM[§¥¤g€,AºUGs”“V7i§$@³ ¥]u_'&¨_ ú?ä—ç9áɱK:0vtE€e\ÁM<]$º[íŽVð8.È<˜ô€a»Û w¡£"§>´›Ôc„+ß„>RN¥ÍÝÁÖå²²´Ù`j!ZŒñ©ÍÑrŒ ;``¬ƒp¡ðª{›4ätÙí74‰Ñƶé>¦¥Å®Ê€ùu<§Ó‡*m»êBK‘•šÐïš~E³xC` ¤Û-š°OK=åæjñù:NÍ£w(£Ñ;px¤Ž2qÞž«GÔàÌT¥B–ú•*­žT)–nªRU•UjüàtƒJMR©QƒJÍD¥fP©Ñ‡;˜µ pycÊC(Nøñ‚›d ¨8ɆöG)"A(Oúƒ /cìЇ°—#Ëò'÷ѧÁëüJÜuó-ÖÍ bÓ.†=ú»$œ¦Ë© -+Ìz¡¸HqþÍ›œ†%$Dâ4I8.KÓ£®Ÿš³Ž3Í4¥åi|¸@ï9生ø\'¬i ëcÅ«>6Í “@jHÍé‚…¤®Ì-%3Bdõ/@«î×®Å~"f9ͱbb><¶/nòô°réeˆ@¶à¡`Ål¹bP«®=íÖžu>þ/Nág™ bšUNNRì¬U*VU_õ|ßæ*ÆGÿ¬i¾®ŠÄ¯TÑS̾U–†jPÁID:*~Í‹_¯ùßÄ75“‡·õрដÑJUŸ0ÌïýdùBygÝŠÅ£¡xê_Ðõô1÷Z<´†~ÒÙ²ò%/€µ‡’.`5…3È^nÚzƒ‚ê3„Rý‹¥XšÄÈ]îiUãmÝÆ‚?!SÛ¯1ûÂMPHl¯XŸøÓú¦ª%Sœt!J(£Ö=MO“gjÆà vq$f²¤*,ÇRŽIÜð '˜1ž$¸š W–ù2öÇ5÷±ÞÔ"cšhËAúúcÊð#æ W\zûâ-YâÙQ¾]1yˆÆ·«6dƒRL\|ßEE–«£«ó~t¦ë6üœM4!Û·£Sv=¨o“J<Ș+«Ê×ý_ºôë3Ìê“òmI†£Š¦¾A;UT™ãدb¡[¥ô'Qg·mœ@FóX D¬õš&±]1ÝŸ1>@Ñi÷}*¾2FŒ¦ƒj+uvÚk¨(Ö÷[`M¿u ™¾‘ÉÛpaÙt“ó¹t>íÌHŽÎ‡¨€È}Z&ôpƒ…Ld»¸ô  Àš~œ:é«tû[w×Â9ÑÉ…_ sUàÖ‹Öú8ÏØÁä±Õ 'Ì^Ó°lâFSnwiaB¢…·GŠ8·HX¤Ê,×rXß„«¶¦–œ*—¡m7‹`×CѵIW"þÆB»}Ó6„Î^6¯NŠáXø‰ñÔ8?-Ó€å¾ i16›`|{ž×ëØù¸ëè;UžT°Y 'vÂòÑÛx‡±Ž~›çD]Õ5ˆ úWúCb–|áðØ„ Ý9õ“®ö=-¼¦ã·òŽL„€ëzÓ¬›ºM´©»@mŒ&Á»b’lRå,FÁ[6Mjô€V»,Gê³ÔTƒñÐÔƒ•ÖšzL§±;þÞª¦=(8w 9<¸óX9»X9Êf[&l*î-Õí/„‹kòž¡"§VL¨÷K]ø|*Èé¨G…í€ËUê`êÓΈÇGV=„çp,ÆnFdúÅT3…¿iM_hõ.÷ÞC ïþLåÚº22!ëóWýä9SÕ‰‚0ZyˆêÏ)AUŒ›¯šÿv•Ó‡!ËäáR¾ËçÆ >ª“›· éÍß.)§Õ¶¤¶~ü·j6"0uÜݤIÝ¢¼D?ˆwüÛ9Qø{ÂI_Òg½è÷ðöÜpUßò„e.2"$_h™UËêñ1Á…$Ò*…(;6=aÖÅ̦^§ H›]›ÖÂ/5f@¨îïð阀TˆT…ñ¹†ü¬ä:=þªøÿAF3ç’{ 3ü¾û¾Ÿ¯Ó endstream endobj 62 0 obj << /Type /Page /Contents 63 0 R /Resources 61 0 R /MediaBox [0 0 612 792] /Parent 57 0 R >> endobj 61 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F24 17 0 R /F29 15 0 R /F18 10 0 R /F27 7 0 R /F33 36 0 R >> /ProcSet [ /PDF /Text ] >> endobj 66 0 obj << /Length 2242 /Filter /FlateDecode >> stream xÚíZKÛF¾ûWèHÁV»ßA°ñØEŽ“ÓlàH‹‰Fœ8»¿~«¤ØTS¤ÆòÀ|"Ù,«¾ª®ª®îŸo^½ýHÄ‚`d°!‹›û¡)£RD…^Ül·Ùá~¹bŒeì½òŒûÇC†ýhYûÇf[ø›u^þÕç%ÕYQ~ÚÚkƒ–+ÎEvÓÒÕÅú°ß„‹ê!fwØ2/Ϥ|\®€Ù¡®Ë»]Ù”E½üíæ_ ?FXêÅŠ‚ü„zùù­}÷ö#Õ}U5âZ,°'ù7&<™>‘ADi0ö4ÌÜmKµ;|Z®jyê¿”!¥uK^¥8*$yÇð5(J©EÙ^^'Y$M'ç]Š%GRÓ£&‚ü–bD¢¬åÝr¥hö~¬²)¦+B’0¶" Ü…}‡û àþ°\  §á†i†pã”@Â$>˜lLõÍ‚©5iÉÞ»ð`!%£¾ûÌ‹ÀôžùîoÔ~ù@`n}w$0‚Àmó±!l$ÝQ ~Ç}wŸïÒnNw©Y(/ˆÿÌ2Ü`jU7þamkí¡\‡—ëCUu¨Zö›Ú6Í÷þZüéJ¥½+”üСÚU¸½×»º¨þ\ ™åMyŸ¾ ·Ep®º· ø¶Pàúé4•ÙïK’t$[Vu¦U¶ÍkÿìK9‰Ä‚*cœ}.›­ûÞS箈ӶVpϵeÝÀ?ÊÝοÈ­Êúk“À©hÖPJ̳KêJËÏÇÅc±Iiõá¾r0@òþøX–`Ŀʇ lz沪xÜåÎ6ðPäÞZžÎ× vüÎŽþ×ß—M~`=Uþ©°:P“ýþä8Ãçy +ƒ@ÄžÈÂKüä°ì馜nPÚbW8»"™µErϷ샕°ö·Î~{PyãHöôè_äuûáØ¼„‰ÉY71L9¼@\tÓƒY“¥8E³q<²`cf„ÍzTœœ·A.¢¥ŠâÕHlÑÇ`”§XC$%½p•’/Òc^€zãí׺ wÙ÷m²e>Ù›Ó<ÄŒh“0;Æ/m¤Õƒýêi×”»²¨âöJ×¾ñ’ÀMøwÿÝàß©.k[G¿eÈwhÀ©˜ŠåéZAÖ÷ùÎ ¬våŽë®Ü\]vpÔÆSä.ñùû ¬¶¡Ô@Ö¯›r¨ª"`Á÷‰~ù%emÊ‘&Ã\3Xš†w³é~Ò¿“*¢ø:õyrnj˜ Ës 5¹Q8F©§XDâkx"”õ¶Ž,ò$wM._¨^<‹€¡¸ >.R•RñªéuŠ[´F¶§³'ŸcOr™A¯\ÝN9È qŽD“õPOƒyAµ]ÿ¹Ëdzç¤ß¡™™ F˜0)zkÖQyÉ,y“ Š> /$nr=lâ)G;?‰¨Ô®@Ž'M†E† Ì¢—™#“îò̦¹(ü B5RØÊa(âé¯Ie ¦Â½‰ìÚ:}]¢3 l§—Vxö<€Ê¤ ‡é<à¯D6®c²n,¨@Òþ¦]ô¨N¦u,©PÇÎ*Äf)Ä/Sˆ)ÄNr–Iå`F¨kõ‘'â=½p}w»\I> ¡HQ=à:Sô¨Z«Ë9Õ€­ñ~Fªš="hæÏo:pS¨\M2paÄ0{–Ài†ÏØt¡Ü7'ÅÐCð é‰õ~FùqÖ…8¢œ^!‚ŒyÐø| áKׯ“êØ$vlåÌÓgþŒÐrüÇìg§¢-j£…éQñ~‹;Y€»p8²¹ždCHÆ%[§ƒ.‘ò|9k6ÌÕÞþM\§¾î:­NO*BxT OE…é¹9V7GlÆ@à³@WA¹;w±jÝ’¼dÅv ‰Ý€rÓ‰"Ž-¨k:ªi S!Žé â9˜Ìui¡2š%:ö0/ŠÌÃUÏ 2¡³,r'Q‹ù,l“y…òʰ}<È+/³6™‡ë¤.'™¾øü7¼/x±Aæ¤ìÀÆÃ†FäXõ^T|qcæ3LEbAÛ–PHtäëÃî)lCôeæMòd>måÄbç’ƒVº@%HWübү鴢s*Þ¹ö›ôÞVW®Ûþv¦ÛOxû×ú3]étóŸ4í’ñãV,›±K6ªzç¹ÔP¢“¥btªe®žÎìé_~¢Up„ÿ¤[YHXéÑY\‚OÉA›t ɸyÉãNŠù$[N¤Òi=Fà^p“ŽžéØizÔkÜ%½+­.§mN8¤äiB¡;â2Oȯgr•29\T–<_¹"Ôvð8sfÕ4ô§ ÕÝjˆN‹Â øWK7OÆžZî’^ÈØõ›‚4œ'7·w%Æ9F+Ê¥ƒ ò4¢&N,"ÛæÿË+·w+³r¿®Š‡îÔ›}í÷±áÝC‘ïÃPãGšòÁí¬Ãóðìw‚ñ&0-×~WY¸wŽËvI²ƒ;£6ÊS;Ó÷v£Þ:•?­¸þ\õîào¹7ù£;m¦»Ófö ‡öçÀìûzë΢}ÞûÇ2\›ünö¥ŽK Bs‡íO?Tšw^h‹Q™ý#àçØH!î8yôQ‡È“;pç—EåPÀÔ&Iƒ´åq•Ê«7¯þ~'U endstream endobj 65 0 obj << /Type /Page /Contents 66 0 R /Resources 64 0 R /MediaBox [0 0 612 792] /Parent 57 0 R >> endobj 64 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F29 15 0 R /F18 10 0 R /F21 18 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 69 0 obj << /Length 2911 /Filter /FlateDecode >> stream xÚí[ݓ۶¿¿‚ÔÄ‹LÜÄm2éC'7mf?Ð:žO©î”J<;iþùî DR %ÅOÆÓ'BàÀ.öóèåíÕõ+¡ Á™ç^·÷…ͬw…Ñ‚vÅí]ñº¬šÙ\]n–ÛSK•÷ëMìjêØø‰ µÙ6/ð—…r[/ÖOwñMÕ6š‡™(—›ôë—Ù\¹Þ4ËõÓ6Žº¾ úPý·Šôª\>-6õcý4[6±«ÚÔ³7·ß]¿×ås!™V>®þ)ÒWXæ-$_;$cFÚH%U_œ)E‡ìf6× J!r£ì¸“8 ÝäFtÌ h)¾@6€—úEv0ƒƒÙ–ô4&à,&~âšCvjÉœíÔ7¹A5SZµG4šÍ&<“^•ŠH% ŠMáÛ•]ÓN-sˆ"’€( “^SJœ=¼€°úœl- Î-Z”&XQf8i˜yž–ÈQ-é+ Rs©96Z”ŽoHZYÞ<3ÞeM_PmÌž‚$/Ü^m\Û3ª6àÏÔFØ 6ø‚ð#j#åÙ£ûQ¥Á}Ñâ ɲÙÜY]þðüWjÊŸëEƒ>VY[¾úýÛØ¨¸Þ,ë§E;ƒç¥Æýój@8•—}ŸºsÌ^Çè€ÏfùXÇÐÃDwîuÏî±PŒs÷<8®279k€ÊhfÂó[f¸):T¸E`d™Ÿ×0·÷_§Î cóê³æ=_èÍ›5Ŭ„î¼$ht=Jˆr´H·ZD[úò÷¯cc¨EAeSRõUfB 4þ”J ô`¸€‘ÍÀœä Íú.ŸG¼ù´ m3+£+—Ÿ~壊(‡ŠˆîL)UÞ>Pò÷¼¥$SJr<¹… °;_IiinϬ5{–ç`]Éã°1ûÄÆc…ÉéÓ»j~«rSo—wÏÕj›¦ß$uÞ2ç<5‚›c~‹ÞQpô‚·ä‘îê›Û«ÿ\ÑÂxÙ˜Ó ¬-´•L*],¯^¿ážü®à¨Á®øH‰…‚F¼*~¸úÇÕË´#Þ€ ÁÁ¡p…vœi›Bô"'”½C)Ì%&Ñý=› ŽÜKPsÞ¦˜£jÜdbB(3ńԖ êrÑŸVJ¦ÐùiƒOqö¯pk¸q¥˜¡&‹·GD=ÑøäžêÁä"nô9»‘ÅK`€ß]fñ2:Ќ䃯 PçIöüZ¹ø>õ¿Æ„¢C¤ˆÈ©˜*®xy£¬¼ܨߨÇ÷×AãÄ$ßs~}!¶‚U~b¶ÖwÝœþäZp9É5Ç'úò˰Í=óÉÙ–ÐßìWÄ6fòl+“ï‹°­¼ÆjÑä,ôD¶±Gbëq ®Ë5æcB^cá ‰oǸX÷_IËÑÈ Æ­cÃÎeÇ*Bû‰Ǹ¬û|ÿ-Å5Á·QhòB|,ô„û=á»÷\§²7d7èç•ïÊ• ð-%„XXè A`âŹº pÜøË+À¤ ”W¢#å°êyšSÉoCÅëaWÒR»Þb1[5éÍýL”„wR»JÄ3¬y~K&&ŒÍrQ­â«nµ{>,›‡ØZ¬é»÷3|]m–4AÒÁtÏí+‰êäÒ¸~,q$ó: ô·|ŽkÀRº·#[n[PG +„aÆÛÌ^t沘Uš^Þ¾ÌM†i¬ƒ¼=[í)j—\6ù¡¼ôôSÄô½Ë£…7CŸ^vž$ŠM†5ôf »›ù­éz›Ž>EH7†õSsFåè‚Ú4–Ým‚Y°;Dýô`H”OR8ˆ[W"Æ}.X5‚nxí†èúû¼,&æb Ù6¡‰ê ƒÄ]lÃP¦²ÃQŒ}YVº£vA0 D±VcÀ[¬äðDºª&²±\êØÑ~SÅŸ«eÓ¬êÔ®6ïêM‚B(´×׆æ¥3Ç=A+Øx¹©q@Q®B@ü;~~ü%¶"rl˼PqÆ;Z}7b¹s Ä<îZŒö˜ÀD‡.ÊͶ‰¿›zóØŸ»iI1\›®JN¯>,W«Øz¨Hx ¢ƒÓaMIü+®û>’º"HN¯^§¸6æåÇšSÜÜiÁÍÜ4“nZèîÌ0Ž1÷gþHLX_Ö Ât˜üøäQõ0kaot–CöûSÝ?¾’±ÃwÖBÌ)‘az!§ •´UZ<-¢gò0h@¶ix°Íx|`±^iH‚–CÙ³ോçGzý¼ªšå{²ËÔ·Ü6Éxñ×ýfý˜Ø}˜Æ­7ùS}‹åÓîê.ŸjËýALr8ðÛx6’¾è`¶è&0É·Xœ¢¸ÕNH ò“nḇ,§Ž™?Ö®\õÍß÷Úà¶Ï;d˜f3=LyÒÙ–¯yE Í^Dê—“`RØB*Zë$Ô)óS ñn0iƒŽœ·°ŒÁ(ß•± îï¦Ú}ÖpôÖ¾r°€F‹Î‚*O@›²Ÿ-‰ÏÏ–ÀÓ])w[Ç´øðÿmi`KòL[-1UÃ'FDcÚ#' ¾Í/ÊìkÄ1‘GVÒxÓƒŽ«t&þ~ìîžB†÷WÇo*8ãÛ0¼EÅ ï@}’¦5ù-#!ï•\Æõ}ˆõ+¶]ù®nbç ›N~õ ‰ãÕ^Ù´SšíÚÔ7ø,ù:𨬃‹Aÿ¢ J™8ÕÓ®SOˆ e|ìN h}L\ZìHê ä­ÑN|Ñ«Så¤e”íU´'ØI¸íKÛêâ‘ p×àç@—ÿ$™´ò<< AÍ¡¤‚íOÐ7,ñ£žH¨x€ïŠºúf䊨–ê8V~<ï^ àØp…iì†$ìç&û¶sZÄ—€…žÌ“»ö²ƒ„1gý @Žq#}OâYA¡fÊi(î¨âOÅŒ?£­;\# •®nfäf)y’ávÏ=Å ÇÈaÍI‡£R¡*4¿Û™WtHfÐå]ÕT±•uÐÇÂq5´Š+EØ>ú§ù8QþÚÌDlb@Ðå:,?hÂÔÔŠSSk[§.º®œ.ß>§Žt¼ˆßËØqpʸ¥ëÀÜ—ß×ÙsR¢RÃpz1ÖåGŽÓÅÇoÆ&Žñ W¸ÿ˜YŽGS uªµv:z¢‡¦œºp;Üú "D„?"d]ûìÚ.Ö›ô6þۅΫ‰ö!QtNz_ä æX€Ç¦ÛßE|~Ú¦µ‹%}QßáæiÇQ')$¼E¥Y%g!y/O +绤úË/³Sé7$šíC Îí¡0F¯°ûšäI!ü·§‰jµ~zGM™t:w<Ö¤÷޲j @¼@¾Ž„ÿÕ±ÝfáÛD·—<ýJïË8RL.•Qý?*|'Ê'·VÛS.‰Úü%Ñ}ʤŽ&-渃d-a¦t[$%4ФK’ÒäöœÍ¶¡êw÷ endstream endobj 68 0 obj << /Type /Page /Contents 69 0 R /Resources 67 0 R /MediaBox [0 0 612 792] /Parent 57 0 R >> endobj 67 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F18 10 0 R /F1 16 0 R /F24 17 0 R /F21 18 0 R /F29 15 0 R /F27 7 0 R /F30 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 72 0 obj << /Length 852 /Filter /FlateDecode >> stream xÚíXMOÛ@½çWì1Våaggö«Ô"•[E¤€ChR ‰€ ©ÚŸßql·Ø¬qh‘g½óv>Þ›ÉÉbvtŠV¡†¨#ªÅ7…„à½Sì4ëÔb¥.æŸVYŽ:ºùâf³®nÏ·ËíÇìjqvtjBc  }Pº|ùW¹ƒò½Þ-q`ˆUŽà¨2ɸ Oþ.ëÚÌ663›ùÃ7£j•íGöe[¬›}\̾ÏPή*2Ú’bëåů›ÙÅ•V+ùñLi ÔÏÝÒM±œÀrT·ê|öyvò<2d C32Öw÷››»åöþ!Ëcˆz~©K?‚Õî°¨˜çgBcÁ p¶b4}gâÀM”ûÎC!1h95³Ò¡Î…#Óu–›à}ñÀ°œM¾£Ó(°Ê«•kоLž׳Ã|¨ë›¾#ASä*á@ Nò>‘ÖÍC{ðŽžæÎq2`VÞ«âõ0Wx× 1‡’.OV½wÓY%äø0«f Õ$W·µ¹a>í4ç_Å\wý7ì]j«“࢟(Œ>Fê #f•Iž*yøMRÖ½‰Õ²ÊàÉ´2)©†,DfjƒÇ©äÊ@¢ý©”³òvQV"D¤º®Z´«[´û”~ Ú5=´k‚¤=S¼?†ÖIZ‚4mmÀ \}¥ $õ€D+ŠjGjƒ4L6Pä({¦ ¹ñ¡³Ê÷S +rP‘ÓDEþª„f•[aÕï<ôuþˆøŸNÿy:5i:µ24qhÒ)e*W1÷0•ŒT6º‘„“#¡S-\ä°gT² Ýý@^ñ{2 ¬p?¸)Ûúj#ߟp~ªÐ÷ÔrU öUØMM:ao«(ï( AÐO£¢äXgêpµP¶µ¾-£=ZOžÒ4(¥3Ñu ù %µQbË—¸$›#O…R ž…®**QòЖ¤Ï—,)ãÈ–Äðž›(G ;]=I‘åx"É9Ì\LNtÅ6û‘—scR“ÛòÓÒ©¡”æ"Ö ©)A±ÊL;EýA°ãÊÇØò’Æ[>%ÈEh¾¼¾]g¹Ø˜óû,gæòñúq[>]-·ËâŽKg*¦];Xô»¾¡œuB ÿ7âèx© endstream endobj 71 0 obj << /Type /Page /Contents 72 0 R /Resources 70 0 R /MediaBox [0 0 612 792] /Parent 57 0 R >> endobj 70 0 obj << /Font << /F15 6 0 R /F28 8 0 R /F18 10 0 R >> /ProcSet [ /PDF /Text ] >> endobj 73 0 obj [445.6 511.6] endobj 74 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 75 0 obj [611.1 611.1 611.1] endobj 76 0 obj [511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4] endobj 77 0 obj [596.2 547.3 470.1 429.5 467 533.2 495.7 376.2 612.3 619.8 639.2 522.3 467 610.1 544.1 607.2 471.5 576.4 631.6 659.7 694.5 660.7 490.6 632.1 882.1 544.1 388.9 692.4 1062.5 1062.5 1062.5 1062.5 295.1 295.1 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 295.1 295.1 826.4 531.3 826.4 531.3 559.7 795.8 801.4 757.3 871.7 778.7 672.4 827.9 872.8 460.7 580.4 896 722.6 1020.4 843.3 806.2 673.6 835.7 800.2 646.2 618.6 718.8 618.8 1002.4 873.9 615.8 720 413.2 413.2 413.2 1062.5 1062.5 434 564.4 454.5 460.2 546.7 492.9 510.4 505.6 612.3 361.7 429.7 553.2 317.1 939.8 644.7 513.5 534.8 474.4 479.5 491.3 383.7 615.2 517.4 762.5] endobj 78 0 obj [826.4 295.1 826.4 531.3 826.4 531.3 826.4 826.4 826.4 826.4 826.4 826.4 826.4 1062.5 531.3 531.3 826.4 826.4 826.4 826.4 826.4 826.4 826.4 826.4 826.4 826.4 826.4 826.4 1062.5 1062.5 826.4 826.4 1062.5 1062.5 531.3 531.3 1062.5 1062.5 1062.5 826.4 1062.5 1062.5 649.3 649.3 1062.5 1062.5 1062.5 826.4 288.2 1062.5 708.3 708.3 944.5 944.5 0 0 590.3 590.3 708.3 531.3 767.4 767.4 826.4 826.4 649.3 849.5 694.7 562.6 821.7 560.8 758.3 631 904.2 585.5 720.1 807.4 730.7 1264.5 869.1 841.6 743.3 867.7 906.9 643.4 586.3 662.8 656.2 1054.6 756.4 705.8 763.6 708.3 708.3 708.3 708.3 708.3 649.3 649.3 472.2 472.2 472.2 472.2 531.3 531.3 413.2 413.2 295.1 531.3 531.3 649.3 531.3 295.1 885.4 795.8 885.4 443.6 708.3 708.3 826.4 826.4 472.2 472.2] endobj 79 0 obj [597.2 597.2 736.1 736.1 527.8 527.8 583.3 583.3 583.3 583.3 750 750 750 750 1044.4 1044.4 791.7 791.7 583.3 583.3 638.9 638.9 638.9 638.9 805.6 805.6 805.6 805.6 1277.8 1277.8 811.1 811.1 875 875 666.7 666.7 666.7 666.7 666.7 666.7 888.9 888.9 888.9 888.9 888.9 888.9 888.9 666.7 875 875 875 875 611.1 611.1 833.3 1111.1 472.2 555.6 1111.1 1511.1 1111.1 1511.1 1111.1 1511.1 1055.6 944.5 472.2 833.3 833.3 833.3 833.3 833.3 1444.5 1277.8 555.6 1111.1 1111.1 1111.1 1111.1 1111.1 944.5 1277.8 555.6 1000 1444.5 555.6 1000 1444.5 472.2 472.2 527.8 527.8 527.8 527.8 666.7 666.7 1000] endobj 80 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1 695.6 816.7 847.5 605.6 544.6 625.8 612.8 987.8 713.3 668.3 724.7 666.7 666.7 666.7 666.7 666.7 611.1 611.1 444.4 444.4 444.4 444.4 500 500 388.9 388.9 277.8 500 500 611.1 500 277.8 833.3] endobj 81 0 obj [413.2 413.2 531.3 826.4 295.1 354.2 295.1 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 531.3 295.1 295.1 295.1 826.4] endobj 82 0 obj [543.1] endobj 83 0 obj [565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.5 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.2 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.7 361.1 572.5 484.7 715.9 571.5] endobj 84 0 obj [312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7] endobj 85 0 obj [694.4 666.7 750 722.2 777.8 722.2 777.8 722.2 583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000] endobj 86 0 obj [272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 87 0 obj [667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7 667.6 484.7] endobj 88 0 obj << /Length1 1398 /Length2 7685 /Length3 0 /Length 8636 /Filter /FlateDecode >> stream xÚwPœÛÒ-Ü=Øàîw÷à|€˜ÜCNp Ü%X°àÁÝÝ!H‚[~’sνÿ½ïU½WS5óíîÕÝ{u÷úª†FS‡CÚb R€8Ã9x8¹E²ê2†<Ünn>Nnn^t]0ÜôA…!Î"ÿ ! á69 ü¨q¨¸9xø<"<‚"ÜÜ^nnဨ@趨sT Î :ƒ,ÄÅ ¶µƒ?ÖùçÀlÅàdÿvAÁV@g€:nrz¬htè@¬À ¸×¤`³ƒÃ]D¸¸<<<8N0NÔV‚…à†Û´A0Ôd øM týM k†ýåÐØÀ=€PàÑà¶9ÃCÜœ­APÀcu€Ž²ॠÈù/°Ú_vÀßÍðpòü+ÝßÑ¿ÿ­¬ N.@g/°³-Àì¼TPã„{ÂÙ@gëß@ # òt‚–€?W¤µÀG†óƒYAÁ.p' ìø›#×ï4m–w¶–…89œá0ôß÷“CAV}÷âú{¸ÎgŸN6`gk›ß4¬Ý\¸ôœÁ®n e¹¿1&ôÛlApÀ nnnA!Èò´²ãú]@×ËôÇÉóÛüÈÁÏÇâ°y¤òÛ€Ð}`@wuùùüoÇžÐyxÖ`+8Àd vFÿwöG3Èæ¯óãü¡`O€1÷ãúñ¸þõdú¸aÖgG¯ÃÿŒ˜Ë𥎖žÛß”ÿ唑x|8^8x_pxxø‚~ÿ™ç_ø‡ý«&ü÷íþWFeg@ø/Ýû‡ˆûß›Áü·lXÿYAò¸Ï ó¿×ß„û·ÕãÏÿ·þ„üßvÿw–ÿçúÿ÷Üÿø™ÿü~ ØÑëoÄã>»Áµ¡yTˆóC @ Zâhýß>e8ðQ!ÒζŽÿj#¦öYk‚áVv-Ñ?SxLîviB`àß/7÷ù5gåðøR=Îê ô(©ÿ,)ïl±þ­=Þǹ¡P úãèO/><"µyþÙm§3þx$ç°@ÑOT€Àõ¸ °?7xô ÿGv+7(ôQyæÿXúŸó™ƒ@ž +ô¹iˆ•èûê7-W•Òä[Ãâ [©,>sÐ/n7Ø(I,ï_¯@/¤“ú;q7ä™Ï¥æ©ï}kQB›´šo}ïÌã´Ç¶šÑgG‰{Fò¤kº)Ñ(8t¥¶}ï]}õƒž4"¶©0为 akæ\yt)zÖt/ ¾ÞÒÚ®PŸ+çˆÖ‹2 *˜dȵ̜"¥}ç DeÅ?ñÄ™<¿˜ÀÏùE­džîwÍ÷ÑÇh•7æzÊ{©T—ÖNFOfDJùäpŒÑGf7Y…dƧðSÔŠøL–`ø»¾ù7]ñSªúd´þš¤Ð7EâjmôíÙVPG}« #u ªOC¡¢MB9]¡y&8àúžŽR'ÞÑšíKUûá‘^«oü;Ržèa'ª q‚õÖL»ëg\%À1E5ä$„eºy—ÄãC:¹~òÚý°èÚ|ãî_¹œŸ.9ìŸØ‡ú~‚;™,‹tax ‘hßZjÖî&¿ÿØk×o\WIxN<ŠkÞ*¶1 ç}àK¨<}îÌK"ƒý½$S¨O%u…{ŠÐ F¡‡d0ªucmxHC1Ö-‰=qŸÙú½XrAëÿÿ;H;R)Ãt"‚3Í%À|tˆBò}ЦÐQ–dOä³Þs·|¹#c§43®>`ÝÄâvÀ4 jã_üñ\MÌŸ‘Ú =¤ÔC (l‹¯‰Úáj·¦˜>œCÜnï^>Ærl «rÕ¦½­è ¡»‹µS~Nº¾G6R ÐT:‰ƒ*dˆ]Ê«äUߦ‹’¯ŽqjøêOTº¢ª©°™¡ÚOÔ®£Ã^¹«ÎåQ™¸¡¦©h¹ÆˆéÇ+KO‡<ß¿¶ ˜/31î+È}V¤Óoþx-åîÑoà”§ðõ.i3ÊgˆfiTa¾T¥æº!£=AÃ-R’žl–” -ÉúI?Lî÷eÛ s|ц¡ €%c°yâCEóДÏËZwHþlrÈ‚ nÂ9áä ?èòM s9¦5LOùJüØôf" WÝw9=ÎZåÊäU±6v‰¢W Îg~kfÜw >´ÄŽ×&S‡¾r!}ô´+Äãécqñ‘+è9@Ó,à‚uËÉóœtYJñCܤg ’æú!ç«L}–¥Pàhã˜!ÖtšzQå—Ú¨2üÙ‚u|0ºKÌjj$K/ô¸3Çs‰çg:FçÌ»t1Ñkö ³ [6™ìší.Ã…¾'å-ÚBM‚ugT`Œ°gaõU €YÛ›š÷™ãyRÍ•vø»™Æ2 ¡­Ëìj “Nzï[v je@-¬ô•ÔÎÏ5>UîyçO}úVã<㎨ûT=Q2þ‡Èèê,2­gö?ãoÞÔÄE<³m¥¶MY,/bÑ0Pê='lo± ÿXÔ¼È<™ à=ßmœÉ®i« Î+`¬Õ®¦}yIj<?ln"cÿ&ÖÂ%ÚYG»­MQæ´ÔYžÁR~QD˜!cí3àÁ옋‡$~%@ðÚ˜Ÿc‘Œ81i•š· Ú‹Q‘tÈ»ºÎ%¾9â7›XÇi“Ë®§Þ bnä7´ú”õdà—ÑÔš[p@ççÞ±7»"¾Î ÍÚ¤˜kšÈ…c>U“E´Gq¶TA »Qkêr}÷<±ºcÿ¬KøI±†¶~ ÎN™ ˆhõG¬s5ÃáÚH%Á«ž$·>øt@2ÕÂÓ0þž†È 1WZ¸Œ‰h®tá²é݇ÈTÕçä6¤ý^^¾J„¹>ûoõü‰ÅÛ꧔ȩ˜­xG°…™ÆÙUó±–®Ûá) …áÈ®”÷x­Ð~{œì¼çHñ7å™D YVOèò{ýëb1R"¾½ÜFÆzB eÎéð¤ÿ…¶;0ܰ`f`ZÒ½ý¯ZXöĺ`4Ù“ñþëÚ=ñ-ìä÷ö´Áls}¥ªÑXZ/HÂß(Û»“{j©Ër„S“Cý¶F¢ÎcoÉý¶fÍ—ž„“?Uj&¹€¾ qA¨µì^7¢E cK—|]5»vg+xV¡&»Ùœ4ÌVH¹ „ôŽ0™âÚ+’ÓSv|®'›ñu›^a’ßœ‰òÖó0(uæÀ͉Š-9ffûšÈö‚œ$éó¢@< sÆçxœÉ‡”/¢ µ1-¸n¾ú [” ­Ἃ\ïšWÀ—cT{=èk>õÀÙ_šïÜ_ÕÙJÔct}²¡†)#cqwÈ÷C“ïÜ© 4 SÒÆê¢UHÛÿnˆ[ê»è5rìæ¿¥…gWâÐæÎ¯2‘ŸƒH´‡¯jÐ4œ)tZ˜ÐÉGƒx‹7õêÎl6Žã»Æ³_*¥ÆTçÏ8p 0ü ¹8¢Gp$EÉsÿÁ÷FOøKGGuǸCž¢ËØýèþaï­¸r”vGJبbdœöê9G1‰³tÒyïß²žvË'öB\©DÛ¦šG¸¶ñbšö 1Õp§p! äŸu1,¢å*„'VKˆ·àžÎ ÔGÐÇfe[oE2ÒqSA®ánƵZCƒ·òz½4¤ÝßL¨—caò¿~^•VÌ 4•ÌDc¼%ÐBÃú^#†ÉàéÈD^€>}V†{ä/c·.¾ªP›·BËkÒhʾ0Q%1b;*ì°vG¼¼›‹žŸµÈǨrÑ‘dhû~‹¼†û"£^[Š]“ ª‹_¿$õ:×==Š[?VìÁË¿Œ;ÒnÍey$çoÌRqbË_Ë·{¿¤ÈˆÀå¢ÜÉ€“D9'ZÖOØÆBLAŒ"uÙÝÇÜ'¼îôóPO'ÂãBðAû8Rø›Ó²(eÞÆh#FJ,2ëõŃ·×´ã$˯{ÚjÀÂɹj.ÁÈFpû’žd<íuŽ7Á )Å%0scÇ;aU„L ¯êòø|áÂ55Z¤ó¤;·(ksWQ Çk@w^ð‡˜üÂñúC»k¸¿lmÞµÓ+²Ÿâ;  Nk†Ìmn‰_ŸDWÚÞ}þL¿yj6Œ”#Áþê='¥Âíbb>bz6b¢âW’&BûVvF„#b=ÆaÒbnœý4ò¨’T²!ÛV“†ÎÛˆ´ƒ¥±§lvBœ¡„4ýµ’q¿øýÙºûÏrS­e§sE/¨»Z+?æ÷tÑIׯ“æÕã‚\å%pÚ´ÕŠÑ^?ªÎJÑûÎÉ”¥ENœqijÔ<3MâcåV=ß Úëº_ìùÌâ7ñ¤\eOÝ–vœþÅЩy«m‰,ŒW¶N]53g*Ku¸yx„%g;)Œó{‘‡±ÒT’Üø™„H™ÚÐ*SórÖ?ü3àÃ*ðÇëM³Ýü÷Dý”‰,ö”/?͉]@,3ˆ×}´YNÉ8£0ÁJJø9êNo C^›˜˜Çz¡¸ƒdâð“…{©RM_ ‡ âþ ‡oa~¬†÷_gU?Ù¤ý´r1 ðOk—ÚˆíHå«ÿ`߸śúš! QXPcø„ÞÛ?Ö¬³#IÂIí«šÆ†ð¥WøAMÙÛCTp†åÙµÞ‹§äÐ4´pKf… ¿öoñq̘acã ª޹¶79ªÑ ‹ëÚªà†àû•ÏnCçp%cG“hOn ;Þó‹êÄajÃtJçV¬–©Ø$¤kÐf4ÒÚ&CÜõý—Ï„*]%¥r4F ­SªräýãV—ûÌ[by†¤½±¤œ™»Ô£J{X|JÄa½!E*ïÎ;Ÿh ^Å'Ñ}£z§ÍÖ`¥·VB©«»†ÈœÅB]'¯jf½ÀPÛNûãA'’Sµ»8›ÑWÐ5ÔcJñžl6ã&‹~U?|L#·uöÁÑi\À•×DùmÀjJTÓ"ÂPÛa>m˜¥ =QCpWE°†úéä`ÑÖ´j­7«¼(éÊõÀ)ž„FEЯ-ÉÈ@LâÕAVlh¢®?Wé'[4tM=ô-Ó¶m·È†‘½Ïdvúøçª˜ W×aàB­aQ™¤@]FÍxü³ ”Ä™šþÞ&}Eæ€Uð&‰ñ¦r´˜TÃîF†x¥À…dÁž-^=–î€bÑ~n·LíÅùÂð^œõO·ŽDi[s÷$tµmÈGùІž<e–ZU>ÃÓa/øjZ—[ø>—©ðaP"'ù…U jt¶tjŸ:N?ïRõÆ(¿Èå¬ FD`&¤Ê“zëÅjË•xÜU8/Xº¯ÀÛ÷%Zuݸ«~D#ë[(ó44ì¡CR†„FbÄ´_,âM­aªX­‹ô³éPµ7?c¤ìº VžÒöÁLS*ÖMß;üüÞ©S«vìu¸lþ`õ-U{¦^ìxðc°q Êõ¹¤ƒ$Ã¥cÑ‹rcYõ;Ql'„ð…Žj6‰1#æL)éö™U7fòQ± ëFjܧ_«WÖ{¤ç˜öØ(„¤¼¼´tÜðŽG ½²œöæøˆ"Ê>µŒ—‡©IÌjļԸòW˶O$„‹.G‡¸‹Ð!š`‘3D«ÎØÔó¢8I€=)éU´tIۘΧq®o÷Sžò#T á²t½Óý8„Öm…ÐŽ¿T˜ª­Ó8¢:0$ràÀÿá'á¢j²ßœ(ŸXòÂþ@Ø~ @GÏ i8òÄå ·ö— í )éH´¼íSKWÂ}DSŠ‹VƒiVý—ï¨ÔËàóÓì:'¶æ¨æ¦u9Ü!åûІSWÞ5»lоñ¨¹6æñPSyŸ… ëÂûÛØ'´)Ó:uç¿RPÝ™+媹jÏæÕ,Kðö"g¬×ç´ËD3$%*_õuqÖ¼ÌvÊ~}¶ýà俚nQmĵ»6æ‰<ê³)­b•¢¥}•éêlͼŒ³ðÞ+íÇç ˆÜ…n±Šì»§‹?•B½µ‹g½”šwÕ÷fqgöÃ<{ÞËJ5?D%¨» ‹¿‹æ‘,¼õŠ"éc )Ò6Îs¸²í)ˆþEH©l]™ëCcݸ {8„^Îú–‹ˆ|ÆÕGÖˆ•yÙ¡çÝ¿¨dyT¬S®øN/@ÏWk©Ã­P±D¬CoQ>ÒP&•~ïZäTMÃcï)Ô”÷BwLBdÊr²IÀ`¬… ðt¯‰„lƒÆCŠð¶Ü(«±œ:–@#ˆÙiHR<¢™ÛC¢#?#æÜIrM¾vD ›ðMT‹â3'[¯'³¥Õ«m¦ ~jܨ®5•™ï½€´³¹ãU2þÍ%}Y™ "µ[Ï!˜yQ)³ ïÀ†Ñf«çÄ"º¢äåÒ:yÐ>èyfT­[}½ÐߪKòžœ+{BO+ãÕŽ1Ÿ:æÙìY|š‹`¤¸£AStç(þ¶GV”(ÑTN®Ê{ åæT¾Ö±¸wàÃÀGÍ\VUY=×x‡„Äà]69F–Sëëp¦ØþB}C­ŠuþŒ¹„ƒN7LGBk"¤zPT?ÒØ7 ŠÅÄŸë4ɹ*êëïƒùRCÛ®óNZé‰\;+DXÊórêÞD9ÈUF„Ï+ ìܧm7V—T̳vPƒïwž&üªušùªýstZ@YÊcš·Œ+@ýn¿ÏuÒÞóÝ®UÈÈ–3O¾/+áØ³sÝX¾&¿¯e”Ü¥u?òÞb*'hŽMóúçõÃÔ_LXeµÓÕ·×›à'?™„P™Yã\‘fF2œ£Ùà ·¢§èÇ`y;© Y¶Jï]—_cÙâÉA™¹mÓC1§§–J¡7îiuäæzŠWWriÃØb]îìÙ¢‘KÆÍqp«G¤î­5¹i&t;”îßþÅèªGšü£mþ›ÉôØôW:og’š]F¥ ~Aè—ÓÝÏ«¾tôá6BCs÷v¨KeÄ©«0GžP‘þ9ç¹ìKø(~K M“÷ ÕÝLŸü¸oœ5&4ìNîWÖx’ªwæùˆ“’,Ư< ›Þ!&†PØŒâòx}4-œ_†ìkE¹AmH›ÕºíW¼®(VžѶYßÕ Ò°P:WÁî\w@¬¶®k¾é!¡4·rH>[ÇÓ˜È}C_㪲#c=R¹Hþ¾¥¶ღÈs>V‰#&½ÍTµê~ŠÒ,a*ñV$o !LÛª «ÖK. ›ÑÁÌXÈ‹å/©¯>ÈN Tcðì^¬»ƒü%€§HÙ÷ü× Biù™%”¯ŽzM¹ñ(¿I‘«~6WµìeyÁ¹š2K†Øx*3IÆ!ËIDh##Õt§„̾©ÃnªÈú ö^¹³§ÑaR_“içÍu"i=dåâV\—ýû¸ÛÁ2ÑW¦DåaJaL뵌ƒHÂÜ¥ñg%ÐéY°Òñ43êáÈK)™ŽtÐà)’SÌK½§JfÛ“¸i5&Û ‚N¶ ذŒE0±/¡q8%  -i¼ËSŒ­Ð'ñä`ÿ“ÔÝl&ÎM•bغ'pÓæØ³„è¸ ¾}Òÿ·~,B1sH™„Šd`ç›=U¾1•‚U¶3_} Ã9ŸÕP²W¸0²¯%“œ•?}ï{SèÝWÕÚôÀEMð˃ZÂ^i·l»Î~6Í쯗éˆ_Šù'äÓ=B)Ô—î´o-,“ãx*K?«ùðýÄÍ0ûõ,~ȘãçŽÖûÊ·Lw–NójêÁO¥~´£{ø¦‹Œ%øtã˜T‰ª‚êaÁ®åKuDÀ×@ö½áƒllR,Ò½Ž*– /ÁÊMQÒÝ£8øu¥ 5bñÑ{¹~NßpìXž ŤY’Ó¡Õd{U¥ïÃÄO¦“ ²7~¯Üoüä&¯úzµ+9Âæ¾J†RÚL"§lþÀ,Ãì¤dÁ{Å 8_Ú|üOºk¯¡Ü¥˜÷ÔO xšrr1pFN.âp`DLFSÛifjQC 1 ë–"ÓÄZôŠ4¨Ÿ°H(ßJ¶TöñLJN2?_LÑØ½"ÖKZnœ@.›¸Rž,´m¢¯ ÞRO¨£á7ö:–Å+$GÜ c¦m ühöëÞ¯·ÛáÄf3…¦œ¯Þ»)Ðno䓸(÷R1“Å“gÛ™íç+ضÁ¼0ŠîH–-ɧ{V¯ŒŽh´ MÏ,xüøë[H mé®êM?êá"êãÝíë#%V[çÁ‡`Ùöéá€Îƒô(*£†CAK;5ц\Oë¯j,WgA]7û©\HV8!Ÿ(è1-ñ\–S–S›´ÓÉdð¹?Mís;TdTE£‹þòºÜ-Ÿ0kȸ:XÃâûŽuDÚŒ1¦³>h…Œ—ïVï¡&Â=Ô.†.ÖºM‰µ÷La§†m¼‡¬ìë;:Bb¬Cv9rBtî;ygË>†@ÿ“ç:…à`Råm%gWB6 m¦Š¢×;S"¶ç½~6NcѪG+¦@–AÉ‘ˆë^Õ£ëÿ´ÁÏ´ÿ|ƒö¬´€(é)ujEwÑ–éGé”ñàkéïÖoús™Žå¨ÓmZ+E«¢ù/Ñ¡CŸèE{O‡:¼K{𚄘i—Ãè¿\Hgï¨gûù1?[θº*Õ¸àƒdÝLK*® ²@gî¿L¯*±ÖäSOÚ>¸³Šr8¥ò…™öRD<5ÆÏfp }FUR|OÆ1êÃÄ)åY@ßß§Uß)òê׉½­çi¶¨Jd>±G»Ì¥÷|ñ]œP[cN0ºÚ7 Í\Ü&2Šþã—~ùrÆBX”ÅJ·sÝ'øüë´7šWR 1ë3t)ˆ MbG…˜Œ×l¶ ˜/i 7dªÍN¯xýyÉeßô³²Ù8L·¡íïƒÇè·m Ó›(<ãÔGYÇfN¹>fø9­Ú“ÏSüWCÛ endstream endobj 89 0 obj << /Type /FontDescriptor /FontName /XOSQUH+CMBX10 /Flags 4 /FontBBox [-56 -250 1164 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 114 /XHeight 444 /CharSet (/question) /FontFile 88 0 R >> endobj 90 0 obj << /Length1 1773 /Length2 10432 /Length3 0 /Length 11548 /Filter /FlateDecode >> stream xÚ´Pè. ¥¸×Üݽ¸» H€ÜÝ¥¥P¤xq-îR¤¸{‘-Z¤¸séîžÝ=çÿgîÌ$y^^ùè¨4´Ù$­œ,@rN;§0@ZUÊ€‹ÀÉÉÃÎÉÉFG§†9€þ#G£ÓA]ÁNáYHCA@Ø“L{2Tu‚”Ü\<.~a.aNN7'§Ð  Â ;Ø  ÊPr‚€\Ñ褜½ `[ØSžÿü0Z2¸„„XÿpH:‚ `K   „Ù‚Ÿ2ZÚN–`Ìë¿B0ŠÚÂ`ÎÂì@GWv'¨8+À ³h\APwà7e€Ðô5v4:€Ž-ØõO…¶“5Ìž`KÄõÉÅ b‚ž²´UêÎ ÈŸÆ*°þj€‹ëïpyÿ†üá ´´trtB¼À€5ØP—Sa‡yÂX@ˆÕoC ƒ«Ó“?ÐvZ<üQ: '© >1ü‹Ÿ«%ì sew;üæÈñ;ÌS›e!VÒNŽŽ Ìíw}2`(Èò©ï^ ×âäñù²C¬¬Ó°rsæÐ…€]Ü@Š2Ù<‰Ðþ‘Ù€`>NNN~!Èò´´åø@ÇËô‡’ë·ø‰ƒŸ³“3Àú‰Èl zúAóqºƒ0¨ÈÏçߊÿFh\\+°% `²CÐþ‰þ$Yÿ‰Ÿæ{Œ8ŸÖ Àùûó÷?“§ ³r‚8xýcþLj9´”´U4”Yþ¢ü·RJÊÉàÃÆÇ`ãæãpqñø8~ÿçïü‡ýR ø¯ê8ÿ‰¨±výIâ©{ÿ!âþ×f0þu6L€ÿΠæô´Ï ã?ëoÌÉÇiùôÅõÿ|¸üÿíþï(ÿ×õÿߊäÜþÐ3þiðÿÑÁ^Y<í³ìé6Tž.ò¿¦ú ?ZÊÉÁêuŠ0àÓ…HBlþn#ØUì ²ÒÃ,mÿ\¢ÿLá)¸Òprÿ~pl\œœÿ£{º9Kû§GÅõiV¨@O'õß)e!–NV¿o›„B^hœO ÆÍÇðáz:R+ç» à`‡8Áž\OäüÖNP´ßåçpHýý‰„²#A^‡ÎßHHÀü=YZþ~WÎaõ/Èàý r8¬ÿ†|¿‘“ô_ú§D¶ÿ‚|ð¿àSbûAA‡Ã¿àS!Žÿ@®§B ÿ‚O…8ý yŸlŸÞ詟JqþGýÔ ç§•tú®§ÒþU)×Si®ÿ‚O°ˆ=%ƒÙBAÿ$xšÌÃé_Oxü¹ŸôžÀÿª¥úôàýqvOÿþãu¯A/$“?ã,oÊ2ž¿Z¢¼÷Ùo®CŽhIÒl½õ½3{£5ù£ma‚°üã¾dm*9›Î«-ß{_½ {„føN%º7A,<¼+^yÏÚ¾Ò¯#ás?4·ªø•ÑïJ§ØâtcƒŠfèr-2g‰©‘`l(̸ǞØ3çÓ¸Ùã”JoXÐü~Æñø~㎿žõ^)×áví"¡%1$¦@8Ç™¤÷‘ÚIQ"š÷).LP )´¦0koǺð~¢ê\-_&Þ5 ’.ÅÞœ O°ÔÜ “ÜuÐÎî ÆÍ©(šºç2ËGsÕìØ¾'ôŽ©˜ù ‘è·‡hpG¤õ¼uxzÎ(Ô±±“2Ï¿@’×ß'ä?ÅP÷ö1àÕ›n‹\˜Q·Ðó ã™ÀÕ/¿ñqyJ–Ñ™Îëq 8òAà—-U¶’ùA–þ@%8tÛŸfˆH­Û’”Ä5`úúàÆ µ®/˜3ÂǹEFi´£ÙVÂ`K‚¹4»{é[Uîïæ#|ÐJ•„Sá|âïµn¾UtŒú{°ÌÁW5G-þ…Ge-d&®±Wå‡"÷±ä¢Bqa+/¹!ù oR… 7è™. ‹ñâ5‰ý  QqVÿJÓ²x]ñ*ÿ¸«þLÏ…ˆRä³ÙðÍBYÆg"Þs_µ%:¦ƒ·®Ç2¹à Ÿ’K~žõ‰–ެJŒÇ nhÂ;î$"Ñ8‚d4Î&Ó‹#ù[ÈÞ¼»*t³•í23`ì÷J W»JEf"-Ú&²â›Í[5à-ElåãJ8Æ0Î@ª–}©t_Öшc¼0cb‹OÜ’ò²®2ãnѹif„Lø5y=›8¹ó ÀEr“JËmÝ—ì/Òþµ_ίÏË£q[­Ò¸>Ž sÝl¨Xñ¿ µ/ûÅåbiޱ# ñèù:’5ãæöå+ë60à¨D(ŸÁ{¹°¬o•ýœRªKÜGDBÝr¿´ã€NÌ.}1¤ÐÀaâ­¶žcQw60‹„ùøRŒ´ iEms¤W=RÛ)ÝòjHâ½ó°µè P¸æËú…×a8þô¾LÖ"¸¢¥˜¾C0’ÍáõÁ ëuöÖâпÞÅbCm *š)"nÕ'9Ú&•‚¬OÚš·\2‹PØ3ç¤9ÝÈuó.)~¶Ó—þøÊ*9‰êÁFÍÂ[K»Z >Õü1UŒ¦ôJ™YÿY›lTì¹Ù ›Rû»/ècùù§†V Ç%¶)‰51 ½âžj%y«Î› ®…™Íò*‹ê¶â§ÙÓNqgÓû9Å™¤]pKJ"ÇJ,ÆŽhc”äJ$·üÄCŸíIðâ烄¨‘Ô0æ2SÝFíÇ€[~IlKQ±ÕJw¿Œl 8é‚5”)æh_WÉù÷?­•ûYÎ*¸¨‚*⊒´!×û6ýiâÌ^0‡›PdÕƒ©®•FÁ†\†S•ʵëöœH/fíkx,¿‹(°¼¿°žhÖ±&Ü]ÇÛP“ÌTev›SmOË%ý‚ê+³À ›ŠzÛ˜˜EÀ¨ÿáÙŠê½t°¢„(߯Dt o×¢º?7óÕ…ÀÒj,åŽit„ÃËíÖ“ïá+‚^ꀟ““ľ9îæL/?³Ë„““óä.n%+Ó&“àçï ¦µXyjÒS¦V\{2eˆZ¿Mk=F%|ÖÁl±ü„\ÆýóM]ЗŒ±jÝ m!……c͸¶ÙL¨õ™|‘é1_D2Íæ¢;QØëñ:&‚mBåïÚ¦äuþæ´ URƒ[VAú| f¿íÃû•dtS‡>nЂ×xg@g/qÊi9o“¦²Ó¢Ø(\˜¾L›•x«)hoÉ0Ãy1™ÊÂ~¥ñVåÑ"Á·ü‚Æ‘š¿S9K…RÃ`xÍÛ«0Û0ÜrêÒï1耉ŽHeAÙZ{¥2hã€Qm¦N7±Š­ôØ ÈŽ=qÃÐ*köŠÕ±w oaøV£p0±*4ÌöŽéºóÖíZ— ¾CÕùwÞúÝ&ïK׈$Â+%$#aÄr÷1Ò»°­X ý(µ«UæÌj5DðÑ¡Šõ©µÜ G‘z>ÜÈ;'œ±3à(UÎ2æ†ÔÌäÜ;ظ(±|õG”÷ãþÎ>¡êMq á,\ƼœGÀÀˆ!Þä$ Î°Kcô¥ó|üü|C!´Jb@#:ˆ=¦’ÄÀ†@Ö‡6nIð¢4fù’GÞ¯@žd›!})7ÒE3`*þ,a*gŽ!‡-£éyàjà²ÜU‡‡ªºNb•Œ«Rê°&YKÝMwˆ2êmP»ÎjGãw¥É›J5×µ©øÛì€:©%ÛØ“¯´äó¬ë§âÙ[¤§N»Êï—¨2Ç«°f°sMeÜcR8ź£ JŸŽZ)¾dLžTXri(®$®_²;ähÎÀÛ ÊÑÙekÔ¡q¶z[pV~Ò\UeqTn mîÿbÍJ½ÌÁãŸß[%r{‘"Ç!Ð]!6ìĶ­E½õ ’B®$Õ@þ ²ÿÈ/ɹûÙ¶ _KN¨:[gÃ'‘1>õ˜v-,1@xøÌ¶“ôòñWà–àÎb‡|Ë”^´uë™Ò‹¶ÜåŽý‚ExQóæÃx6‰Ãs¶Ó}B$y?mÚ—@_ŒËu'Œù{7†# œûª÷^Apr{©¡«Î=uÜ/ñ›“zý™W눤¤ÇŽ$ÒÕÍáÛxßãÞ o$²ÝÎÑ3 åj˜Ò;¾¶` ¡ŽPv’ã´²H’¶ü$yÉ^µæ•7)·Ëè³Ý3*¼Œ†’Vÿ°jÂRj\° M¶ š• ø”}7b;xéÊ‚;–}ÓoVկ]C”¹CÁ „{á?ýš>4ÓÕÈñÜÙàd¬ÆÐç©ú-|-àÉ–±°53ΆºY‰¯ý¹7K’™uF 9É™M¦YÚŸ«eÌÄ\QBÌónÜç f™¸Ó“¿¦½ýNKs€É=éóp„rW¶½QÙ`,ç~®ûSŸ½3ÐáÎÑȦnç ÌŽß¿W}†`™úí[S{éš#/ì–e:òJ¦ :Hƒþ¬Üu8|Uœ4¯Š€Aü™mÿSC—öƒ’%J°b^Xma¶ô°N³gHiŸÉmév Ó%8Æk˜‘¢7câø#*HoîÍ’úmRP`åy“ ø¡턎홒nŽ‘uµŽÀ ;i˜eaLŠîóHÖK[#AÕ<Þñ=rTï!áÍ‘Úu»G¨ûy†ÆY< ¦øÛ‹@Ï“ÖM³ËÖÂa•ž¦Þlt¯²ðÒ¤ê³q>ìI­W>ÖûšÒ¿üʬ†ºPæ/ü6¡µ»TU¼­÷Ù“‰ç­±ÖùNªÓ,¦ ¬©X‡üf_õ}lF9¢&Ó÷Fì¶kµý¦”"¼Ñ']ãg¾G¯ÏÙ¡¯Öú{g¨ø¸HÃiMÙÒ2Å{«i+©¬x‘Û™fõ¾ûò–¢>±Ÿü%Ӕ΄ÿ„²!‚$Á'bB0’’¡?£¸¿Ð}OÁª{÷†é#5JbS{„ß6™˜-‚ÚT0P g¡¦„.Ž qcΉkÙÆ]b;—¿V Háâ á>ÕÃi½ÚÙ}·]Þ†:/Eq;þb‡ÅÊãÍqð°\¹¯“­Úç¿è|N=4X £•Þæ¹4¡±4¯¿B?Út„°®Ù©Bd­1Ñ‹µä)ªtŸWԽƪüœ~¥FTœÿ˜©Œ-LfÒûQŠª´HM"P‰ úh›a^’ý¾ËGTcàó{‰´y›t–ZuAáë6ªçÌ!›'é¹Bøø‚Ñ”ËjØnf:™{é\ ”ô‡Ìˆ j¦ôEnоçס,sœ%õÔèÂèð‰VDÚÈÇeŸù+u=ÎÕ©eço(&ŽOöÅäzDû)s°'¨Ã<9ñá^lÞ£§úøÇõœ­R²‡cah0D­ë5ÒÙXaú·$Zlî(ëÏ”íëæx‡ŸÀ_äÅ ©e†ç]°*&@‡`Ôígâ wÛë–›ùá]ˆ,R9“þøÞa—l[9èjˆô ðô-[¡ x»aQD±ŸõX­áiE Ëaä:Z]tøª‚Ý2?š&T²]_ûDÉÔÑð¢_…ÈøKŒíDi!8VCÙ ££Ê²) ÎI?0€Ó//–ͤ_ÁsÍYaaÝ Ýee±–y©; EöËýèZÒ¶í9¸ì£*Š<,DG¤’q/|¿ˆ¼Þxƒ^ÙéÐ%ç³PÊJpña«@ U<ÄaŸ<ßÐÅåUµ9a1¾tLõ!+?É’ï‰.ÉN÷—Ÿ«¦‘[>gfµäÑOËíÙfÔÅxžº,[¾–ø¾N™Š¶ËT^ˆc}|>;s*Œï0 IWõøa úÝw³ó*·éxÕƒ˜›¦uHØwx¢5wžWẺ`ÆúË_1N6¾ÀñlæWº{ET•ëg£tÏÒà õœ×a¯S¿AÉUˆ»n(y†…Í $¤ü X̶R;´hä¿KêõÂL¢eV­0;ê¨Pç87˜ƒÜÖ¿zŸY|'ªN­&Û/ƒ¾ž<(¬Q‹®În ;$±A; Õ5 ·äO\PïØ>½ù6£?ÆýÉt‰ø*Ë[¾êàVË™+»¿÷²n7Ó~¤^¿e·1¿ •â—ßF»Ø—J~y7`ep˜Xßl ë÷éÃAXØH·Õl'ÛÁA³ÿ ìTôVN´6‘”g&íÎáaeCŽpUÉÔcŒx%$¦Ç'ƒ^¶"V8Ï^ €µKö+\ªøS£¢oA2Z¶ÿìRF•µr ØÞ³pèÝíXß@Ÿæ¹Y¬õw<°GÓÊg+hÜ„¶*|ÚwMYŠZº3,úQ^ÊþÜä¡Û„(O”ªÀ–iÐã›am¤@Pê°Îó5hÈh-»«€Ttòú¾¨5…uÇj2Âø´“ß).õ jf¡k§™Tö³ œü¬Î–u]‘׫@£ œxä]+\}þV%Éâ[ùs%Ž#tZiŠxªº ¿ÌÍÀçªÊµè’†Â9$ݘoÞ8ùáTÊiæ® ÇhÇ&5jØj#Žm-¹7ß§ëæƒkáˈ—ŽêøML/ß+Žô´•'…¸a‡1§k‰v¯l4P ³#ìû¥]¡ÐW܆/Öë ¾ƒS!Sª•÷c޵s9næÃ[ëÐÑ:†¦¡B¥¶Mã[ùö):teÑåaÏÔ 7/ Í/MK"Á È5mŠ‘Yað犖ð(]£nU«íƒ}ÍŽ´‡Uëº(ä‰zᕚa,z“Õ‘ æFþ7çw¦ªŸ§¼nå­šÏUÈ‹ø©jè^*¯:{Ú¾ qî&|©â¯/Ãâë:d¥/Ï_YãÅ…lQ†lÀ‹¨,_PCÓ{(0¹Í3Ë%ÄÁyÃK Á“LQ:]Ä%CJ^§$a÷‡OQž…^v 6,œÉ÷°ð‡ÇfR±pJW55r h }ÖüNØñœ)¿×cë«ò ÇµÃ&µTT¥":ã»mwØVòIŸ©ž¤céð]å‘IU3›‰ZØþÜQC"¦¾³M‰£Jý}yÝ\Ádàžy¾i¶€.C¦t”s„|ÿåµm×Þðúô1½–/5‚H÷ã–ï$å¯"7Ê^üPdà ZIµ ¢)o¨ Aâ>ÝÙ⣡Þ3“嘂à\JZƒºÕëpcŽtâ·̵ZñaänÖ-A„ŠÄ¶Ñ¡&Brm]2§õqLYŽN'=%°œ±‘r=cÂzQ_SR‡¶e«óC#‡øØùu¦ý4ßÃx®}¸úœNª ÅÓzcù'HkJN!:xöAäSZ°ëžéÕ!;’àÅòw’xŸ›¾ô嵚Ê÷t,Z8dN<¢VÄ­{íÅ šåtxgÄ$™á^¼Rñ÷2ì`ÝÍZ <ü®½9¬)!Ø2žüˆÉCŸd…à‰ Ì~3Ðkã :kJNR=NNZ(`!¨s9q{L¹ðú¨R×’R”Êú:Ô's–¾*K‚=†§é`eJ%ÿqe¯°|ÉZ'‹½:£­½ažvŸ÷š‘ø4+¡ã`rR¯‹Š6?KëÊ &Æ/šìc¶?ïm¿Q¯yûâY¨ãõ ¡Ih)‰R–~æFg£ ™U^쬢¿êžÖq¬FãÁ=nü˜k¯Ý²Ê~¼­É3CO®9^D™!žZY¾f3"¤-•ÊFƶѾ“½¢ ÖÒ[ª*u% À祻xZò Ãk8ú–‚ˆƒáÔaë1~òÝ„c²â}Úá ¡|ò79½óó½sž¸ð7Áç,×d‰\œ¾¦"h1gÛÙ[õ"±úrb$;)L…aÏZ¦›½Ú?”Óð›H?诿ªUlliy?ÓÍz¥ŒôÆS†LN™À"±ðÌùõ<®îI»¢b[–‘¡¡j°Ž¦­ñTÎ:îíO o×TI·Žç'§è»v‘ïLÒB8I·l²}2Ÿo§+‡W;œ«ÈÑEAßV/_OrŒ_…yo2! k*(”è3¬ãò:» ãCo‹Ã߬/ž.Å"£n†öœrô‰M±æ~1„޼‰ Ÿ¿i#ã/¬ú@ÉÝGO—ÝßöæÔðq³pë½ûët.ßœuj醸µ4A±2˜Kô­ÂÓñ©hx4Ì’w‡Là³8çÛÔžƒÕ·¬\[IòÜ€ö«oè¶ ‘ìJÚ}å¼Û.5¶°ÓŽ&yî êæŸ-îdæíµ—¡Áúël7dJ´Óܨ—¬œÝЧ’:ù3\åý¨V1%˜ýÑn°f6b$&jht*>¹Nò©éxÛ—µÃu"›ùýÃÏ»Rÿ3vØ4ä«_÷eÍ\•*v*JþžŠ²ãlj²ü…ê<Õ×È »_ÖáEBµï„Qªûˆr;HÚ^d¡¯² &JVpù Å$ÂËG¯¥¦óˆ‡¥× h…¶7’ž\ëz‚Uº0„ý‡Ž°Õ2Ç‹R`Ý‚-ôú|h+à@–ÌjØ;üÞ·Õ8®"UǬ_yÅtÎÒ5Š¢³"+X꫸F$f ÉQ h‘\]³¥íwêÑó·¦ubìµ÷³þαüϘfõ¢Æ,&hRÛ;ìññëûêÞ½Ý]#µöö†e$ðw⤄"±ÞS³(pyYt:Ì:Áà!ш¥¬IÌ^Õ aíå &õ Ð (ö¤‘?~ÚøaÌötÿþÙs«˜p8¢¼Ëßø×™Â—¸lÛ©{êÏoLHí‰vA_`Z’Iß[„öDX¨â°¬ßY‰Éc«–cFÆÙ°!2:ÆLd骮ÊpéßnƒBéÂÑ–‰ÑÈ:²GÙð82ôžsDt2Y“žvý…ñ1±<7²íæP4›ÿ™“H¶ 0^lzO¿9´Q¿K@KŠtL¨Íuf`·$ËøøUȯƒåÅzÂBÃíÎ[ ¹’Åê{»œ$<•„Úk²ÔizÝ yÊŽ¬<#’h%ß]Ùc7Ì Òq¸Ýæy޽ˆ__’95£¯hTž/±@±h§!Õ&þ}‹”ŸwBy?V]3Diã™Z¸`´dí ´íºÜ”ÌZ>å/§ä2f•…mϳ7 ?çDêxx­‡»¥0a÷ÑjÓʬ æ¸vþË IüÄ ·<'Ââ hH”°¬°H”ñL;‚Œu^!Ú‹¯ÿ,ÂÒZ¸F,=Îø(afP¸êCÎÏ` Ø|s×2ˆ…8t¥»Þ}ùÖTÓ¡÷y…+`ªÛCŸå…ýø$ªb“öÒÎ’:-jd$F!éˆ/°© øÍßo¥“¿DMò¦±h"|4|tûþ˜¿ŸˆºxxL^â:†^Å`µýNÏ:þµŠ~:_&òÛDúÛ;yPá%ûG„£Þ*θjÜSžO2M‘#3)”Øà"Œâ_¾TcÜò,r\Çè“­ì¾ iT;õŒßˆæŠÙ©Ù‘"FóiÆà{ ¦é«gX7C{Žp­­6ÔSÐεęÓÔÍO–Ñe6Ð’iÝ $U“fێ^©jq™îh¬Ï N²vŠLREc_ßüz7ø-¸UÏäg|‘¤Š+ªÿîý‹ˆÁV;Z}tÏYtw8Õ–@ý‘J¢³ëòãyö¯œp†ñôÂ?Õljærfe$\ºÔaè¶Þ‰V–µ]“¼T6(À÷em_sàY3À•z"qƒ6.å®9ýlCÍÓŒ{"£ÊZâo–v¦ÎÄ ÑüB—¾ ÕÙ‡’“ÒI¼x_0*œ¦Ÿˆ;YXTgþ·OL¦ZwŽpŸÃ«Ž\€lÁ±RiGgW´X2pÞ@¹ùé|¿÷©îàrÙ˜©|ÌŸˆÿî!è 'Z@U!wQªžÓkÔåòÄȾ›ÅÃ>ô3e†kÕo5H›¸o¿6®æ]¼ž'+@cí_%Rź: ]`~8û™³/`ú¿êðy0 ùU4µóÇéÌbº=è©>CÂÝÎψ1Í`âÆ \õŠÈP|ú‘í)Gjª8 –œ"¢ xßg‰úX¯]6°Û¥2 .|'4ú5_ùO r=lD{F0$ceïÌJÏ}û¡ò¬é{dŒî– ÕÄ•Üvÿ‡Ù3ljR.ZÓ+yw±Ñ±Еð{;橆ú´ãÏDl%$si7ïé%ùV†Ý"”?9¿„U-œMEä;w]‡M×¼"mè"ÅXKÖ¿ölVH™=.ߤvî­ŸmGÎ$”däõ/*yYWOi1!¬}Nb¢ò]ÚÁ¤dÃ3IAÊÕüüc¥XeùB«#ÛèÔ×—íd|—¼3A+Z^vƒ­¬©Û—õ£2‚þîMv“61ifåL©¯é$c!ï°âc<®ÐÃú03Þ9è…w1»¤x$t‡··«°ÞZ¿~îPQ’W-™o`_£gþž=ëèÓ29 N¹cœbÀ—“7DÏ=ô¶ÝPõÊ<±÷‚67ºó¯B°Â¤;è“òg}!’¬½nqK¡°a¦q£¸7â…û¨Žöb’éíÞâ!,bNgÞÜŒöYÜ“äãdé[õ7c¦ËÅ£¸,Sø{ãð²áJ^¾j Ú Áç—½1Æ Sž™0/\MyìkúììjäzŸÀR‡.¶›Æb—ž›ž* ^ßéýK yWåê©RæÞÖäfTZÙsÉ•Cæ¼à¾ ©—‹ $óápòa¦Ïý /ÙyÞÇQ"â,žp¤&¼ÃŽ'Þÿ‘ι˺ Á‰|îÒsAõ•€¯–Z+ כ竴—c7ý0üi4Ÿ­-±ö+"žÅá&]~¾Ê씲€·MTŽÒÆÍ‚×Göݶ3 1büÒÝp4À˜¦»6ŸW̼¥sßvkv"Ò‚7ÁÐÝñ†Žõô4ߦËÑXu²‡ ©Zjä"±6t¢Ø‚Š ±D-,Óß ã8¯Ï— ­ a½H>CþõLm€ôr‡ñEÓj¿\žÀˆ@ETí-¯6ë§¹¦’Ý·6÷®£XVþø++:ØÔ— «,‘*Czâd(­Eì¶! ˆÖ±8_°È±7¿d¿$&%7-è“Õ*Í^œ ò <ÐÌIÅ.õKþ¼'Ý €jš8´Þ?z¦C£×Àxþ<‡^Ô¡³§~a`‚/›­Œ%#ýÒ|vÔ&ȱϲø|®6Hj!SžŒ%ð·©£‘ìplýüÜÂbÌåñ¯¡i¾iPî•që”n8ЂðM–Y4®G†g¿ap¡+NWËú¬FTЫ»Žn^T`~½­øâªË‰ácØÌÄQ !j’a‘ðC%:7§ûÞ—¡”êåêpø¼éqä–{Bpé#¬ýæ¿bWzË^EP©.ú=)¾þ\³)ÈàêBT3Î"Ýr¾'f üvò¢‹ï1ÕáÚ ›e¡§»L²‹¤Ó¦>ÅKÛ£6Žl+Ò7ø³‡qËVQGÿ@Þ3ªJ·ÙÓññg’~öš‰»­BæØée^¸.Q¦ÕàÅ\(²ŸGÊ.£M€ìa¬l8.,{¥w]3kÉÙûdFkˆ ÌS‰úè¦iê¦øÑsǸ ï}Å ZBúµÕö‰ã¦ô=g¥ÊX¿òƒïÑÜ~t?AO*!Gö Án/V+íj7\9é³´°ÁœÊ7Ë6"ð›Ã¸Á¸jŒú2oiÍDpˆN¶Òh'«CÈ™‡eƒO%éíx÷¿Æ£àÕkÎÖ ï1(† ’°ê¥9W•oÔÁ¹…TϸhÛŒ=ˆ¹Žu^§6JĉÖr4¤Ÿ…nûZ™¯Òš¤,®Zù†r·KjïÏâiOÀøq2'.{¡úSlWšL 5ùTsã¨5†c´¸˜6Ý ¢€É ¢|Û<§y +pm³ï"쾤IîŽÉ?’šÎñ¬±Ò­ˆ,©Í–mÝ çPuÀ~d¡'MÉ cH’äˆËíHÊÆå.-`Š’·:æ 0¾Ú©•±'Ö¬œB©êz‰{†566ç©çDb¸3Ôö@öîÿ§sll endstream endobj 91 0 obj << /Type /FontDescriptor /FontName /RJSLPK+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/B/E/T/a/c/d/e/f/four/h/i/k/l/m/n/o/one/p/period/r/s/t/three/two/w/x) /FontFile 90 0 R >> endobj 92 0 obj << /Length1 2004 /Length2 9123 /Length3 0 /Length 10288 /Filter /FlateDecode >> stream xÚ¶T”k.L§t£ÀÒ9twJwÇÀ 0ÔÀ0´t·ÒŠtJƒ€Aƒ€€ ÝH7( ôn÷Þîïük³f­™÷¹¯»®;žw˜èµõ¸dÁ0ˆÌÁäæÈk(y¼¼üܼ¼|8LLúP„3äo9“!î…¹Šý¡!‡€w2âNQæ xäé ò€Bb@a1^^/¯èߊ0¸@ä4¸`®&y˜›/j‹ó÷#€Õ– æüeuÀ¡¶ W€áq¹‹h rèÁl¡„ï\°J8 nb<<ÞÞÞÜ nÜ^Šà E8t!¸ øI  rü¦ÆÃÐw€züèÁìÞ 8p'p†ÚB\=îL<]Á8à.:@OU åqýKYý/NÀïâ€ÜÀÜý¶þéêúËdk sq¹úB]ívPg@KIáƒà€\Á?Aΰ;{ê ²¹Sø•: $«Ý1üÍÏÃuCxp{@räùéæ®ÌŠ®`y˜‹ Äáó3?(b{Ww_žßÍur…y»úÿ}²ƒº‚í~Ò{ºñ¸BÝ=!ª ¿uîD8ÿÊì!€ ¯(Ÿ°â€øØ:ðü  ïëùŠï8ø»ÁÜvw4 P;ÈÝŽ¿È @À=!þÿ=á0Ô°ØC]qþõ~'†Øýu¾ë?ê0ã½? €÷ççŸ'‹» Ã\}ÿUÿÕbe5c Žß”ÿåä`>.>Ÿ¨/( (æü×Ñ?%ø›þ/©6ú;½?\ªºÚÁ¢±¸+ßßL¼~ëï½aü7‚&ìn !ÖçßœW×öî øÿ¼¿Lþÿ†ÿ§—ÿëüÿoFJžÎοpÖ¿þäuöý­q7Оˆ»åЀݭˆëÿªAþÚh êéò¿¨*t·$²®öÎÿê¡õ€µ¡[‡¿æèï>ܹw†ºB´aПw€ ÈËû?ØÝÚÙ:ÝÝ+wÝúAî¶ê¿!]maàŸëÇ'(Áá _œ»æßþÀ»=C|~7€‡Û†¸3ÜÑ ØÁà8?{*À ృl!Î;„ÔÞþ'üüüêá ¾ßFNÄO3›?1A¾ÿbˆ‘ÿZ!Üþõüùì¿ùÿüÃãÿØýáRTÀãBxCÞsÿï2¸›sˆ=ä †z¸9ƒ|ÿÁDøþÅŸƒ…¼åG­ì}o®õaÉÏLëF™l\þ3ðVÏ |Ì4¶š¡Kð3Ù´þ.ÂùUEÖS™YºkÿÝ· ˜QïRtš/_Y=Õ]oÆ™!ïýT´+ûª‡û—¾ÌÆãk÷dž!N¨o‘Û1å¹{Šàküð~¯ìóª§|îcääºÎFîUùW‚A¼yHé8S¾Mö%‚‹‹øÈ‡`üôì3qî§[ºGO9pöø‹ýM—ùÏ'ü*õù<:¨R™RÒ žeö—ÛJD1å_V¼X8NŸhçv«,mv¦‚pò|Á"H®uOü:dÆ]-×/—¶'ãMÂûhÃ:R)üê»{ÿ´R^¬†ù5’ÙÜDߟѫÏ¥Ђ¡((üN9p)ÿG(¿YOqÄsáxþžlr1fOÍ| ¹5ìG"Ü¢á¯Ë.Vâ/Ñ(ÂTñ•´xÐLÙäQJœBRü*ž± {zy?_:ùÞy2®ÒàꀔC¡ÊH¶¸gyiüí>ÍZß„v¶x!“<›H¦hó61—(_3­z9x=Ðîâ‘@ø®Èô¹¿Wª’ÆIÁ _^3dVn¢'ïx?÷ªüÒ7uClܽy¿«Š„tÏ呱5¶Z"éÄv L!'!z‘²×lÐææ¼jɺ(N6æº"at­eŽC"üY„˜§s+Ra~RÔ­¢ß5G_xqÍ~õ*\f»Ò×/T`§iVUÈmrPû¥lÎÆgJêû'~@ë“jæm±á]Ig‹ Ú½g£0±×¯¢vŠS⪠s¥ûfI²–韮Ïð¢ûÌ2xÖu‡Ä?ën'ˆ§â3ô ±ö¼|¥v_Zž~HJÜÞ­mâž2èí!¶¨«ÎKº…‰PŠ£ÈlV:‰Yä脬û9"Ù&´{£Hù˜_ÖiÓÝ]l¯–WUñê›¶º×•Î ª—\Ñz¯Ðhrfî…Ó¹®m~|ì_lKmçë• À[}Ùµká¥%}¼gJü)f@“Qõ šÙ\tÞ»ÿ²;B»Õ-éÛ ú;ëVhÕëß &:²Ÿ]qÁךxÙ–§fÒOƒÎn2‚> kαõp!NÙŠ,+UY-ç“5Ì–c­ovõíÊÌÚvÛïú-lûçÍã å'ùªÛM¾Ù~‘ï)êC/­B9·[¿¦8÷t­Ò?°×—³Ÿ±´{®EZdT‰AØÆ¯¶`¿ù êy´5‘ºv®ù6âÉTÚâʺïîeÀi=îÁÉޤÞ1A+ßK-š>zåTOº`š~?(Îë¿—UCw)Ùjdè`BÀ¬{ã_EÏ47߲Ų« ¢é¯K²IÈs*² £œ}9ºŠµ]{CžKÑñ’zhe—¥C&š ØK¶Âß褡¶ð$ØêìÒjæ½2hA¿]Õy vˆ×´Lέ(<¹ÞŠÁ÷è,½zMê ¬û+fX­R8~-s*é Öè4E§>¡ø‡Ú­½*3'rߨ‘¤µc¹ Š ç½ºû¯‡a$¢÷½™yŠ?­ltñØL}}ÿ¼âìÈ’Å?þJ›ÍðcC`¾g×Ô‚Fÿ2aBž".)óYxÑõ¢Fç1Uîéí<§"QH^ãE(Ç@3SÍ7+|9¼áw¬l¸‘—>djBÏR¯Ü»ºu®HœÄ(ª±Kú­Ý÷Ç}¡¹ß¬´ ÛЄ·I¿h¨Ñ· Ö²zìPÚI (òb5«x/šîJÙ7”¾RÁ(&×ÒiƉү‹‹àÊÌ.= þTɼÕĪÆËûú¨ìÈÎIž“Ô,$·:}Þ‡*~EÒ'²Ó(HåëK\h Ž~i,1Ùu¥wÔyBBá+­¢Ö`ÃêèhÖø£HõeÊÏLÉiTþ¾ ™(JHTW b8ÕÓ@³ù›îÃl3<×IHΊ—Â:wú×̶f«tùž‡9«Ùª Ó™½å,IáO‡Ï”£ô0Þë< vC¾_>ÞŽ)$q»ù£6óˆAa»Ý~"HºLGãáùV²"5o’ÝŠBmü‚£úÓÈNmÕÙXD[L[§˜E…1‘Fçü~~(+qi #Ð'þýÖ‰mG¹µŠíÓ5/R†#¨ôö«‰¨4΄”âLN[ÃoñCãŸ{í1`‡¾ašÓxÙ…&›,4©éñé]šE¥Âe½µø†”ênË«ÂÞuû&úùòzЭéîL˜ÝHŸ=E¬3¥§Œ?Ï7Æàh¡`êÈ”‹NÞ)Ž¾Ç’²cW!³’õÌkúKeìéu#®¦\½EX1é·ïè©÷%¶ tü‡‡ H$â÷·§L®,*­óè¡Àcÿ«ú:ÓðZW‚Çß*à õ8½j„æ{¾p÷\>^e"‘û²Ï¦$ß¾™>îÄPø”ŸŽtV/îk0Å™8)½&qŸQ KË:ÞG¡÷è=2LLq1Gë¯.Ä(•YýÜÂ{s1ç4K^¶<'ÅÑ"_òòRM ý„æ@Q¶[²¨nXlPCkÔ;ÆahÜo¶Zbù@ãšaK™<$$ša[k1PÔÜö¸Æšvà“4)yÂëXi«À{zÎÇÖ;BCC;É³ÆµŠªåU'ˆÚCÔ)ãô2.T`ñŽqäyjôÃù7dŸÀH¤CzìÁB¿~ S¤SX"7¦OÎîõ‘NÓr`Î@WA-ÓX¥nôcó{rP5vM¦£À]ü «1f›™;¸Äƒ58JèÃû ;Ý')}ï/êgqb1òsÚE8¸Œt‰oÞžÀGœvƒÉ[ÞñR¡_‹×^æ¼’VIФxí¿Ü·¿êp?·W5cžÑÚp¸åû*…bëá-œdI¢A• 5Fuçí"ÅàË•¸~•ž é›{<“¤º1Š÷ryŠ,>~݃f–²¶“UéÛw±Z•üéäIg<¥t½ 2èƒO©c;¨ÄÒ”IMVO®Ãß}U~…ì·ãodФ4[{ß„mÌPF®?;@Öx!nSGy«‘:µ_3P㥎¦[BEœÖnN´ü„Ê—i{þÛŽ¡û$ÌÈÏÔöK `ýÖ3}òwo_i6bc)&—'Nàº={²;ìòí„k {e0/-*áþ…bTÃx|•è“ ò:ÞÕDZâNž†Ô:£ùg›–{E¹ÄvºT"Ù"T†ñÒdúæâVëêA={¸Ž°™{š‚÷pŒƒOcìE´·8b?·O½"ž å`õµ[&9p0U@º‡ty›þŽ÷ˆa£ùžga6˜ó©‡öl™›ýLT(в–šßiNðÚFÇ[l½Ë£ÔÓ±dN:Ý »'{M7Š]æYÕ·ùPäZ¯—ƒâbë‡f:ǺŸ×E§8!Û"Ͻ;Ò,6¡Þ$“T_„µ[¯ëÁ™šàÙÆÁéJáÐÂZ¹ÇW;é‡ãý¡Þ-²žž¤üõ{,„ßÝîöWöüSΫ°Æ¨Vd_Ã4ÜH"ó ³!¬$ùß9Ì9)ðÜL-D^·w¼­ò¶u¶R¤Þ›¾êx3¬ŒÂxÿ|jüs]gWñNÕsI abö‰“ð0½‘ÎüçpxÏiù™=ò2XC”¶ã³+5{³v= ¡íA"Ê‹åX›%þðŠm¹w>hÂ{g2@]-GôäXçæß3‡ä|m½ñ3ýȳñ{„ÆÆÏŸU¦¤Ÿôl7`§¤a’ºmT }@°µVBÆ]éÔ§šR½l+`/1¹ÝL±6'·óîÜt×¼EŽŒ¶"BÐ!4ÞI@¥X˜N¶$éñ.ïŠE™6¾vˆcžòÝ£dîÙÁ±°¶ª#i  ÄѦE6ß>ñÎ|Áò L”I4ˆâ[8\O(hœÊ8A: 2žc5G½ß¤ìÑ'K_àùÀ>`—ÕÁß žújROåR±Ï©Ä^ˤ‰§aP¦–îr¼÷Â0 ô,gÃG«ÁÙ†UÉDdH™èb©n_ –»¢‡…ToQ&QWŸLj#ãÿ±¿H–™[/¯ßN*Öq¦^Ø›/”Á³°ÙžÚÝC¢¥R ‘])ŠBë;f$s#_ ºŽ’’ÃÅ îuj%KÉÌ9ž¢Û—hCîYà bK€Æš8¢u›Vnþ–ãÛ pÛµt¦Y‹Š©q´êªÎ÷¨÷Ž(Z×AûM™º¾‘¹xs¦³b ÝÎOÛÎ,CÎ]§üñéyàWñ%(%l=Z¯“505Àï‡ßn$ÝLÞÛ œ6ð_ƼöÕRò2 ÉVìV6öBàÏVš’Qå4 bóà¹"Ù£ µ&›p_Ñ8‡¸é[‡™œtú‹p:[R´Íø­xÂÖ¹ Û"âÓùw‚6AôÜŒtó¦‹tÖw¾ôäI×%–_fvQYÕK›p‚ü3Ž^Q~'±c&¶E·_§VÇndÜž$1=•79Ã@^Oxüö£Vçr¹'ýëäkÒù‡ì³fÊ!]LÈ [Z‘'mŒE¾)éB #Rºº\DbŒýÆêLí&9dOúbÀUœf‹W•ÕÕQJú;yE!ÑöƒÆ¶¡#uí­bET0Ѿ¨4sÿƒðª£Û&/ÑS9æ‚­iqC±NOsØ…Ò­[ÏxñÜ8§ˆQÜúnFÓÙ®ÒÛOJ¹!™)\:EŠF-# Šõ~£‚9ŒÜú @B]Þð…ØÊ:±Ë‡ apöí*£Ô œ¬3ôàÅj›öæŠe^˜»¥Õ`õE‰ZÕEã7Ð>¥„˜XŸ´!V‘?àÊ98¬ûBÄ›;›úÈ&˜áE-“–%T0¢[´Sþñw šG›k(ùnÇžØî§ò©™¢RCâ†pÞWQ+ÓA¶bÖy´h!Œ_I±—dèAdt‘6_ëx«:³PÆ0 ¤kÌ+6&EŸ[w£¾«†¹ dœr÷YûP}å˜& Más7Cs\­+ÙFiP¶Öª•‹÷ŠÔô–Ⱦ¸ç±Çì~es/{—h舮`‚Vó;…$#=t=ývJÉBÚ-¯ÈÈÞcdhx¯5OuY[Gd—2ødõUj|9zoB€ÒÆ»‹¨Èz´þ| m”[ÛºÞˆwh”…ï[$ïÆä<Ò°×¥·3·$ÎùÙHhÐ>ÍùM}µIF†_Ö$Ãàu˜Ò’“î㦉áuΪçÖN+Lq™ªž6 +WóÝÙ(OJ0Õ":Þb^úuížI‰^âÅe>‘±3¹Öv~µE§™ãœŽ:»Œ´h«·Ç©ë;’¦ZY}¬‡å×Ö»*Ø8ïñŸ¿¥;`}@iŸ¬°0œ6LŒ‡ï½à+ Q`62ëæ!E(ãRYñöüòrÕ~ ÕL&lu•ó›ÒÇÑH$ª—Æ-Mð•JÅ«C½ x×lÛ_mÑȦצ·î ͫن4+‚ÄWkïIÀ§õ,Ñ🡽ęŸqZ¿z‚óž(3R´È|¯©«œd#‹z%ïmœ¤WÐ0×W–m%B{Igª—Z6°½üÄ“$ÔöKŒ-qFñ¿0'™¨œóô= ­ ¿‰½éõTú´W¾Ú…ü >ÈŒn¨‡ W5á“\Š,³M¢¿œG÷6úü¢QTw—¿Jö¢/_ aËÄAiÕ§ÎyîÉs·'mM·ÈgŸxRÞweòÑ@­¸€òÇò{ÝJŠq^}ý  §ñ¯…ú ‰‡wΰɷ„Œë7¥ Â.'g3 bŸ…”ŒB=*N³bD@«Ö¯ÅœeG5û>iá@D™‡’ƒ4A®ãLØâ9òE'6TÌÕå½6U·"¸}Úöba”ñ5]så']³¨³lª³¹Ç± ­£¶n.žØåÌS ê4Ï #QP6:ú—û*òì2ÊŒ.äÁˆ#/ÕÈQ ²÷Ëyß(ß>‹jˆËFh€¾;â"ÛÞrHwÚR9ã¯h =k ±¦»ý$ÑAbÊùðæ<Ízz¤¹¯ÇPš[Ogïø¨‰.Š©l#µVÉö¹„}äΣ‰/U†”ðI£cßgW–KÚüßaÖ¥×B…‹üIÈ\ëå­x?^Ññ¶ND- ´¨ßhf,SÙ¥ØC”ƒUǘ{=Þ<¥ôþÀ~ü##?¹kpMüþΚkU£¶ÜLÀ¦…`Š}”ÔÜÝMé9Št†¿È-vu|’=†wË Y]÷ õ¨Q,ê]ç}Lcv£½“å•Ê^ìd%x†¤p)´‘ ]é³Á¿ì?Çôz¤Fñ[Y#Ôð¶#ïÖ±÷sYÍÛÒTíÂ0ZM³çg¸‡#/]O‚ˆ´§f4ŒF˜#^ÅY²"‹íBù`¿”7ÙêûÇåAí«loNîìÖ$FûfŒŠˆø›%±™ŽÔ#k1O9Ãï2 Zëi|Å~Ï]é;q­Ýšíõ½AáÊóÂ7© ’x6Û~;ã#E­Eé&2 M¨¼±þ¢WrB2ÆK%èc÷›V—û$SÅ—Ã(õm³‡µd׌—ª.•µ:އ2 ÑJßI!9yIÄIU » H’3>.s¸<_Løª‹a%Y3¶þ›J“OÌþhØÝŸ½à”(ˆW±½Q°ï\ çÕØ¡Îzò8ãSc`q{´Nž¼Zg©ågeþýÍ%£æÀ 9®ÓZVoOUéÞݵæÀÀÒ#1s¡üklº ™¯„šq\1z‹Ü`œ×Bz›ö½rÅî+Ñm´J™ Ö£š£,ü@™e‰µFí­V2Z¶×ò˜â(J<)$&ï’ž–“ŠŸa{ýÑõÇ:Ú^.ÿ~åÞžUÝ{$Ó{ÇEù§Æ´«ï¶,˜¾÷çtáq²ì{ Íó9ç})ȸ@(Ð>º»)µ0.ÔV_Ô&‘5±ÛØU–‰8gy!h’RϩĔ |þáÍÇV~:]•‡Æ‰ò—ÛxiŸÄ9ÑHÆÌØøôT¸‘ØJGKRŒ3¯pÂ)á1sD¾LŲ¹˜äeÀê§/'>–O;᳉;¿“¢ÀÕöZÅ::‘tœˆyÔfb/Jö´×’Aû*`Ùç5?ÜÊSÞø6ÎÌûbu·zañ×D´|¤<—;ó¹o~˜òXª=þœ7åÑÝ>ˆc¨d†Ú•Üž5Œ¤’ïß–'Ý)ù¡¡@ .%ßË—•Ëd]O"^ìq‚D¦¥râò œþî0ªð¢ø“ ÂàŠ{šÇÊÔÛçFkçÏfv‘g ‚•²\'YÝbBäN^~vò#¨Žª¯ZWüzßL\n 5 ©³èB8¡¤“JyhŸj6âãÍjmˆ‡õÁ”E„ÝœS¼ YÊDÛoŒ¦2•Õläìó×¶|‰Ç¯5`›G7 Á“³¬5s™™Ó¶¾Zîæ¸qW Ýò)'.Ï~áqØÈû‰ØFNDôY‘4«a¥‚MãçDÕ¢¤Øò U²7𫝸°í®~0‹âÊ»4|%ëžo“Õf,‚?¦|ê_ëbÐÜŸ“Ì÷aÇ^`ÙWè§ë´~yÍÿ`W{2ö"=Š•šA;‚Ú¾×¶Ùj†dã°Wy )ˆä$oÔ×™›ÄKR#¶€û¹«\õÛ²~jŽ) Çí6´û…©˜lzº•9^*¥ êPÐó^[¦yržÙ"ëT­XyÔPµ0pÜðükZsÎ)Ê—`÷ÌŠü=Ý@5³ Iw ¬ÈR,;–¢0ó~Á@2KªË µGí°ÆõxÔœoÎfŽ˜•®v«¬6j?“ų̯^„"våà¿i±•"«lÙÓºÌ#ÕR…é ƒÓÛ-(Ú˜K¹ïíLËqk-”ЭpËÄÎÙÅŒ0 “Djl–2WwÄÇ‚Hd4Ä×ò0s3ÍúñÈ1)«ÌFb¾eÄæfI'b)N¼9¹{¾VW3úÛJqž¤¦ÅTᢆ%òòŸú=­ µf©â?2ÖÖõ'+³2¦®»ì;CNg¨G8 R›mj„×~^žƒGY¬ù¸¾±H+›åkg%e)¾PÌŽ]½5æ¯cH™…m$Z> l Õ|©?²A-§‘µÕC9T`Xä_<]Eã/ yáì0;ÝñlÑ6òˆ@YÂ<÷ø¼)ìÝ“2ÀÙk¾z*{ž‡À6†–ˆfžÇöùV ³ânD[  íB•(¾JÙ{WEYº;gUóXôr˜AøÍ2ø~hŸº¾3èŒäW½ï÷=/÷í' ^OzÈk¯"»ÅZr÷e8¿ÈÎÎZè"€yz• ïj5n¤–/”øLÃÞºÔ5©Ø{FuÖ0 Î/D23‡w”ú߈>ë™çZñÇ$Ýü0ìä‡v…BË­ èýôjy2— ±ñ¾žÏÈpÁé ?7N£æ QI¹a­Žî5^Á~L?>Ù[ª ðþ-¤ûs’Ìá,©ß‘ô›ævGèÌì’îíÌÙ ŠX% + õC途Ëe‚Aµ^@Aâæ"ò> ðJJrPb ‹^)j¿l×vµ°¨í9]A•-Å"ÅTbi †¾EµÜ»üEÐŽQ#ð.Úˆ;è|•¾Wãc}Žé›oW̤‘³/°ëÌŒ­ ~þF“É-ð‹7ͯä«_Þð4S¨Rå2u#ahfìâ¡§ÚžU s½ü‘qÛ¾7’"OÙð6zG"J:›#3vж<#½p;’FV͹zl“ÃHg Ž.S™ó¤ÉÑ—Kàq£d6¦Vð𨔙®˜#â~‚{¸£4óûÒ ~ãA1;&™9,B#²¯ÄT,:-•aÇ+3àkæÐ»Ú¸Ë¢„à7rx A…&¥ú×ášØ³á#×Äâ™;½Ê¸†jZmÌé^3 †=õׇG}a"2í(1îvjþAAó–±zZÔ5à<âJñ˹ž Í+«©G2ëoA[,ï©Dªßâ }x8w˜ÃÍŸÇÂô¥øù{Ôž¨z£'õòãl}cNBÁÆ5U£ý‰Tá“ í­‰½kVMUÅ®/MÔ<:mÅœÄjžMØg* Èì2}M,'.-Ñf5Á•ØÓ*e¦:€ÆÓ’¬\.v¶Ó¦!Þ† P±4$å`…¿&QÞ×Á¢¸×¼øœ£ž¼n ý¡a ÍÃ7…~Ü  ð QŽ~íJ0€Œ2DŸ7™²O-W‚_§;©ØÐª©wÈ$©ÏûHoÅæì{‡a2^A&ýhd{S}a‚³ÖáåGü–€^–-Çí¾Íï>6[¾i›ç¯%_fŸq³×¼¡:›€ªæå4‹”Ÿã‘tÿ˜J´®òR¯®õpvðÃZjÃÛ'Ù㥹>…ᕦʎP‚i#ôˆ/GÉŸ¿ ÓVŸŸéèSÞžÓ õöÍ1?é·Õ2¡4‘¸ k¾•›z\Ž-EwEQ*L>ÅäÙú¦¢±°4¾ÉØ®DÁ>˜1•|²~4ô¢YFÈ-ŸtX!½K°2G(jœœJ¡nW´¨£à›lm\ ‚ã!b¼[º3ÇáË(]¶o«1¬QÕäý˜Üj71¯{wäÝ™…¬ã Jžå&²ˆc@—ö¶]oOG*Œ:6W®¹žS’¯]x”Q‘עƋá?[Ù<&ƒ{è)ÅæºÏ½Z·¸ÃfΠ!º"é—ÛhªX¬¨ONP[“ ?´-}løÅfÀLÍ[ör³LTºq;?³ ÃñGówäóóg]•º‹¼/Kö£Öh>Hj—¼áજÅGF»NIlKÂq•³¿Av8 ‚B bJ&Ì8‚é8Ž/{ES`M[†buÎ#[çáÅŸP¿´óq(ëß:œ“£’ÖBÉõ~jì¸\þT é3›º¬ÝüHÒlÒUPiÐç9þгóaÊI'ú–=½üg¶/+Ë5ȧ½&ë EÁzª"ýÏÝ„*D.U8@Ã=M’d}„Œž –î·iýp ) “$öe­ÇåÙdÁ†»l`¥_ÎÕøä†ºÀhF3¦  rP;ú7.~Ê#ŒieÂZí€Þ<ë±#ã—¼ïû˜ ‹—øôéà[xl{#‡ìj‘Ñûá4Jȃ!KÛ¾ ù¦óâÔÁ“>1í‡.t~s\^…÷‰Ìu%êø,9L£_ùÉÂf²qù)%½ ’df°$7”×`üËqcúS>Dß—*WdB,™¯|pâu=ÀO„±Ê­nCÈZ'FŽvB™Ñ£ ´U§I„§TùH'¦‡>iG’õæ·&Èk’&šŒË>èhb5[‚Ÿ’ž¬éXÅ7$„¡â‡®cã{×~ˆ×Ê- ª,´àm5·ÏòK„œÖÚe!„ç=b‚ÈÅl3q†BqòˆÆä<àôúéï<½Í^ß÷ø³Ö' U+ˆ Åg¸w?§äYNûœ ééBwÆ]âˋ̊üœ6U$ñ%Õï•-Ú}¤QqY]õ£ï÷B{¾"A»Ìn?°LéŸË‹x m–o…7m¶$Ro]YzèùÜ6æŽ8ÅÛ‚ŸìoÖ`º›²†¸‰…ž,K^HødÏÑF®.»ŽUÆ$&séÊqü­çD‚?¦²,fÇ0·¿Œ%ë‰l¹-à³,áv š<ç×sÏô–ÔU]s‰~Iå^ÂÁ7Yr§–÷êýNõ†·bö­p ®w!¿ +XOó[W·eÌÔè£è5­mPe‡Í®›e©h='öÀ]ßåì§Þ”[j™B9ñlT猥¾¬˜uªÎÆ:ô\øËPJBÈJËÖÃôÕŠÏF‰×={X¬êaLÑTeÕxÃçÓª\`ðSw$—:éÃBž‹||šNÖ*‡zÓÙ…˜³8ßÄAcÐõfÁ[“yv*ö®¬7ªmç$(yšeˆ¥]ª–Ø;Y’\43 g©Âò÷©'h„1T÷—ü*éŒK¿J•ÝNíÝ7—Y|þÐ ‡üÕðn7!&ý ÎD®€×ÐÂîÁ&¹u“Õ3†Ú"¼9ÎÕ²yL)EUÃS§½Z÷‹”˜ÆÅ!{Ó/Õ7’ÍT|¦Ò÷ï­äŠžèèsÒÓ~B".7ˆZ§´ÒÏE%‘;hÆ£÷Uw, ª{nCëå|Ÿß ½U"4õ6'“mIx“âÓ&Éæ)"#>óŽ]ùÿz-ñ endstream endobj 93 0 obj << /Type /FontDescriptor /FontName /DGKQXM+CMEX10 /Flags 4 /FontBBox [-24 -2960 1454 772] /Ascent 40 /CapHeight 0 /Descent -600 /ItalicAngle 0 /StemV 47 /XHeight 431 /CharSet (/braceleftBigg/bracerightBigg/bracketleftbigg/bracketleftbt/bracketlefttp/bracketrightbigg/bracketrightbt/bracketrighttp/hatwider/integraldisplay/integraltext/parenleftBig/parenleftBigg/parenleftbigg/parenleftbt/parenlefttp/parenrightBig/parenrightBigg/parenrightbigg/parenrightbt/parenrighttp/radicalbig/summationdisplay/summationtext) /FontFile 92 0 R >> endobj 94 0 obj << /Length1 2028 /Length2 13934 /Length3 0 /Length 15174 /Filter /FlateDecode >> stream xÚ÷PÛëö S\ŠKq‚wwwwŠKp'8www(ZÜÝÝ­@q+Öân—î³Ïî>ÿw2“üžåë]Ïz“+ªÐ ™ØÅílAtLôŒÜ99)&F## =##3<9¹ªÈø_9<¹:ÐÑÉÂΖû_"Ž@CЛLÔôf(gg v¶0±˜Ø¹™8¸ÌŒŒ\ÿ5´s䈺X˜äèÒv¶@'xr;{wG 3sÐ[žÿ>(©L\\´¹„l€ŽƆ¶9C9Ðæ-£±¡5@ÅÎØrÿŸ”¼æ =7ƒ««+½¡½£?-ÀÕdP:]€&€ß-ä m€·FOP5·púBÅÎäjè¼ ¬-Œ¶No.ζ&@GÀ[v€Š”,@ÁhûcÙÿÐþ>=Ó?áþöþÈÂö/gCcc;{C[w [3€©…5  .KrÑ mM~Z;Ù½ùºZX½üUº!@\H `øÖáßý9;Z؃œè,¬÷Èð;ÌÛ1‹ÙšˆØÙØmANð¿ëµp¿»;Ãßõ²µsµõü/2µ°51ý݆‰³=ƒš­…ƒ3PJôo›7ü™`cää`ád@7cs†ß TÝí)™~‹ßzðö´·³˜¾µô¶0¾}À{:º Gg ·ç¿ÿ‹à™˜&Æ €ÐÌÂþOô71Ðô?ømþŽnmÆ7ú1¿þyÒ}c˜‰­µûó¿FÌ )¤¨!­FówËÿ(……íÜžt,Ì:f6F#+'€ãíÁûãüsÿíþ/©¢¡ÅßÕý+¢”­©àwŠß]¼ß;qù›”ï àSÈÛ½ üÃF6Fã·7¦ÿÏ[ð—Ëÿ?òÿŽòÿÊÿÿ[‘¸³µõ_zÊÿüÿè m,¬Ýÿ¶x#´3èm9äìÞVÄöÿšjÿ³Ñr@ g›ÿ«•¾-‰­ÙÑé˜XéYÿ#·p·pš(Z€ŒÍÿC¦ÿNã-‡µ…-PÑÎÉâ÷ÅóæÅÈøto»glõv¹8½Íì?*C§·Eý5Þßø¶jÿ[‡˜­±Éïdfc::ºÃ¿Qâ ±<™Þ–×èöç ô¶v 7À[ÏÞS;Gøßƒfç0ˆüýqDÿ .ƒØ?ˆƒÀ õ±dÿ ·(rÐ[ù?è-ŠÂ?ˆó-ŠÊÄ`PûƒÞbªÿAoQ4ÿ ·(Zÿ ®·|†Л¥Ñ?ˆ‰ù AÿR¿¹ÿƒØ~£·‹æþ÷Lþ@Ö7´ýÛ€ ÀüߊÚ¿]_v¶Lÿ²yKkú/øfdñ/øV°Õ¿à[ÅÖÿ@æ·ðÖ†6F&R²¿•dý›ÿ¸¼m'ƒí¿à›Ý¿à[vû?=¾…·[9»?]±¼¹Û›ÿ©ˆù­>ûø»kÇÁ7õŸäìoÉœ¬ Ìÿeð6+пà[F—Á·Cvý×DÞ’»ýÿ‡»ÆÎŽŽoäþëÒy#öñ__.@ Ð~yÑΘ'в6°ý®ZÏ•no’ê4õN“™n²P4 6«¿¯’•¹"S&¾ÜÏ$®gÙ%/ìp—½þíÒs·Ž¨Þõ†ŽHüÀŒÈ(vñõæÝ|‚ç->É"r3XžF²0!w±ã˜"AJ7œ€‰Ù€ 9²×ÇÚ¾¥W Œqé vùÎîê’LY,|.õ“meP‹Ñ°êÛöÇœm›˜€H°ùtÞÎ{øx×*”Í¡ÑáUÔ~‚ž?iÿ½3²`ªÏ0¾(+L•új>ÒSN±Ëq:š§‘ºà±É@ùUò]­ü…ùîű.1eb7‡¹A¢i$Ó‰Ç=ªý*†pªcnÁ~¸É÷ì"ˆ¦!þ„['õ¾4 ÉÞõ¯ÛVÒ!A2î+ó}3›OÎ2)RÚSƒ”´rñ|/»'Þ äQ.¦Áà­«“QwÅh°2©Ž¼Mòɬ³Vo(ƒæ5˜µ9SÍÛUù¿¬r¶új¢²° ºX¤Vª!14ü‚)Òºñô`}É¿ìì¦Xq¨Š°B‹ôج[õJ/Ì{8eb©7N«‰ñpµÙºñªrÛÒ•²±+òþêÑnÃösójt¤7€(»ô¨+<0 ›¾5Œäë÷þÐk ðX°ùÜò‡+?¬‰A»v0x—¿È‡cçcðt€<¼}½T™"&úl—ûPþ8ÿ2íë9kŠ‚¯4«ãø³2Kãb 5ưg'M>Vë.êV…óÞ”\⮂ ²¹7úwÊhD ¥á€Ãø˜Èv¡Í«–œª¹¼ùŸïÛKøü`é¹Jd_±˜áá^fò!iwê½½Úécž ã«n< úÁ4^hýÄßmÌÊèÝ4¼8´û$î¤G˜XH~jÂ|Ñâ)ßöuN’ÿZÔûú ö{VÜ€VRûÏú&ËO•f%HÐ]LF-™Òe†^æQÅF>T=®Z§ÆÚŒ–‰gµš`îwfðè©æ8*MAùP>Þhö/t³G¦‹zÃÇE}ø¸«÷©ªSŠLžÓ~踶¹aÑ Ùâ+žFªãí¦ÐÁÛ?s¿z†ÉÎS„vwÃjfqíŒ~Ù|¸ýÕ°Á/à‡¾²[¯š»4l÷KÙ5}n‹>„0Ør ÏpÖz@€ßƒåþø³É;07´¿¢™ÐJÂá{á Ü4}ðîFÛŸI¤ªæ„TQžÓLrÿ€Æ›éTÏl™¤ ×¾ЊBÞý>F-«5hÓ“>Š!·¢ªTÄÆncq’¼Â;Ïþû“žæX¬üáhá0z®¢_ÎJW&HýG†&Z3z›‚] °öï-Xxðký„Z„€„CW×ï!fH¾ÊîëÃÃ_ñ?<©ÚÍðRã£7Œ=¸ú.“Ú£C“¥©Mâ~ñ­õÍbSbþÄÂÕ„7Ì«%1ãÉt÷j¬.…S ÁN©ï(Gr 2ñõ X+“?fÖ5¶0ܺa½|å;%˜pLksqÿ^e®GÝ#)³\ã¥sµ¤æŠÞ€+w\ÀBKÝÐîÂæM_VÄõåCu »a{=*Pë0Ý;‚½”Ù[&€ñ—¾z5Öd[$Ì÷q†böîÏcaŒVÂñ¡ƒ/Ö¿ø]ã×5x·%µ¢ª!Ý\Ð:н}wÅéÈ2H2Œ0ˆ)<¬D•”@ ¾S’Sâ|ϸveÅ_IQÅvý~d¶ä¾aa ί†nCVKýS™Ñ «f¶ö8£ó¸%¿Ø¼U­o-±æ@·llgã]ÞPœZJõMÆ@.ÄúAUØvöÅ=£¨68øúä$¹šnÄB4é¥K¨ÒMÛÝÇóhñ[“¯¤HÂM”XPcl¦gàçš‘zw4%î‘Hö¥ÓWm‰AÂú›)»›Ãf.œ;|­«w¿¤$C5ùÃw£$«„ª'­w;·éFËÊ_)nµ*©:î T¶Z³Õü¥qï¡iˆ†ñ¬ ƒrµ±–¤ÂwTº‰m Pb»)ðQÅ6’õfàŸd³B;Á¾åSrÍWxáÀb{ %ÅãNúF›¯¦%«(×04úA^¤r_EîfÁâtq…? upê6tïýÞ9 õ7KR¾,Ñ#΃²‹9¯‹lâ&ñðÑñðæÜÃÙ}x¥di)²ÆÖä>­aÊ%l,e‡ôÛimìøÏ¦‹(\´ÉåìuÛô™ æ>»ÚŽKñUh¡˜b`.…æ6Ü­)ýwdßµLÚr‰ NÕbÑ~Àuã”ÛÎRc=ƪbpÅ~k/ÒP0£[I^ì‰D®–<´ßÇF2”„&$Õñ(|îEa˜# ³T‡òÌb »dP­IR—}žes×ñòð½ÜI9óZiÁZ ›©óUz,³¬t«.Jnß×®ílxY`1xvÉ}±·u P{/§®Q–qCÿ]‹t–zXýµ–Ù#ká<.°,žW…ÄÔ`Ý¿fÐf”¾–Ó &uŒ5C O•õ)ír^ˆ"†¤B,R3f;™ÙŒð·`Cr“èJ oôyÿB0­ ø™i:è¶Ï,J™ì?n›[»²˜ùÝb5ªlj¦4Ò ±o¹=*`ûÞq¶‚ðRL»”¢â¹éäé  ÄÏ£#y/Ʊ ùhs–¬H¸ Aœ!<öw³0Ö5`qý…¯^öì$“3;£®“|üU’Wzke kÎF1åá™pí3ªU{[Kö`²UÑÖÕc› a¥MÄšàº~¡ÐÇýoÝŸZçºÑ`iÉ”p6ó¢Ú_£º£ì?Už…!@k®&=sÚy,_ÿÈ„4é”ç!ÙMûŒ =_Òi¿Ñ~“‚ÙŒ#"ñÉ7A~:§:ºË ó‹¥[hãv~ÈÀ`·Ó’›üاºß×µ/äÔìx:K—Âû[ +q¼£G{Š‘dö–_}»]û@A4%ë•…ÒeóÙMOó€µ¢±3§­Ö,Ë ï|x^¤ÈÌéè=a¼èÖ mÎÓØ¿ó> ¢‚7¯ ¯óhêÙ«—Þ3Ÿ:ÄúŽ1º ”²kÅG6•æ×~3þò§æ{2Å9Ša1r=™XY¼¸d” ÕÆ‡õB¨ ІòžE†5#¹›ñ³H·/×áýIº6o«Ï²3‚?~Vô"ŽIðõë•›/QYUe´XÀqÍÅÝ :é}¸s“x¥ †E>ÆV¸ Nÿñ˲jGCuµ}‘–Ìð4Wù{õƒ“ib}Aá’Ž¯±“] ;lÈøþ¤oØQˆÑi­H¥ÃeEð³&ú±TAÆà|*¤c‡Î÷Æ^ÿ‘×íGe KüV«£É{“©ùʪLÓ‘ILèšÒÅ ¨ûàùK+èü€Éa¡\"Êe]¹]œo¨Èp(7«±z~_g’j¬×¦¦Ýùð¯Jô<ØÄ"ýp<È^üÇŒ®3ŽY©„ÛÇt›;:!u„@IO$¼XE´·ŒKˆ™ÏS` Ð7+ñ ’²v¹"ˆSíte USt•A'È]Uq©{nÕx‚éÄfp÷‚ìÚT¶×ºØv«šºž:˜;ùËOdÜXúŠ6GôAÅ_3c§½f4¾‘‡2Œóí‹Å Vµ&ÒŸ2ëm:ÖqçPHòWVÇ×êæyªn—†¾åzÛhdÒ7Èm Mz88yÇfG.?÷95 bè'8µ3€§dv"+Ä‹õ–§XÿŠX7ב00zwcHÈY˜ÞDF¹~1¬§Æ‚^å^?}ºs)ERþêÒÑ|mù‚ %šó°:&D0×´ÏF_âÄÝ©;„Ç#É\ÔQ+`ÝU_kŒ¶û¦Xt{E „-ø~DbŠ^bÀeŠ3â NJX}¶zdØóÜ€>œI´Ë˜ÌEŒ™„Ø9)â” G ³Ÿl9l’;çžKz°}î%3Vñ=Æc#.‹ÏneL ß^ "Nnü¼+ 7GÊÉóÞ'ï¢þõÒ ’ G祥ÏLä,?®ÞoMàþàø› O”¾nB»ãg )@2¡*Œô„ÐwyÀ.RÊ!aÝ’z?‚Ç|/ÌÇEv s>tŒM–AÝÄwý0lîˆ|MG¬à9CòÕ /1ù|+ñÂ(–º5 V¸.È´ÛXåfQ®^æð>n¼¹ª<ñ+Ê¢ý³Bùw®îº‹:™«NÌ6¿^²é—Šdnx<¯¯U[XýQ‚»szcŠåóOçQ6¡;­Ž7ûM6Ýz¤ÌÅø†ŠC‹tãuªñÝÞˆ­¿¤üf0¯áT’"…@¦½ hÃU§À/Ä;?»·¨)ûYmBľý"i Ã+Ÿðë ôj×TœÏUï <ôXŸ9²ñtéˆî“Q猟§£€‰¦ÕŒãÆ#è󀬿×é2²ŽX È‹‹¡ÄKîOÓpÞß­ý¬Å}!¤CYì4bZCN¹úh?)ú]’„™*•Î>Aæôsu˜ÑÎ 1Ÿ³Ø¥Òab ,å±(/OeuÿhÓÃÆ²ÔÞ=³A~ãÆ£P½Y°D„¼ÊP¸ñ óüSë7G4úൎ‘h=áÜ­ˆƒj ¾5ÇØƒä ŒLk‡¯”̺k]iò@¼,™ƒDÆ:‚âY{¦äÛs½¨Ô $ LTÅÏ*¯D|2k¯b÷û2xu4—aŽ:dÌpdÚß"ÜíY:ÁŽ·µÔò* ¨Á r¥·B_Ô8‡ìb©íÞãß#ŠÝX[ Ú*Ñ\Žó(ÇEÚfLJ1»OWa.7z¶È.ɨ«–sânEf'Ù‹qwºŽfeÏ`‰£!$äôUôñI]†Kœ$jR)»®ÇÆBô»Á±ÍЯ˜—¶ŠJâå,3ƒ7|îiÁ’.z¿Ü|êWÄí‰58lî.³Vÿ"Îëõš3ôô¹Ï뢼inýJ•Á±ëù QÕ©P¿ÑQûPAýã®u¹µÇcºág§ ¢Š"ò¬2:kûb¤çʶ³·I$G[V¤k9Wñ‚*¹^ÒT¯W’:xÐÞ´‰³.j¢¼T¾cɧº…Ï…¿’ÚÎâ+¶×%¦‡P{ ÀšÒUɤän¶Ü#ÀÒ9±X伪›ºHsø Øq¡À¡â@ˆè“œV£„ùмF÷I™9-ïàwr°Ciyø3 %¾ûøxp¢p¨8ÚS×’…Y]ïNîÀ6|I0”ó íMþ€A:Z ôzL½2ªÚ&LCx’ ᜩaL õšš;êT ¤Úp˜ÞCOÁ%+ä€ð¹Õµ@E¬ .}Ñßà{M@y¦I©šUàÚ òÇÎÓAØr×C¦åsÓì)Wƒj‹éÞÈcW€ ±^Ùáü±' Õ‡¯}å.š¸Üc£ñEŸ4Þ(Âð5ÙË• UÊ`ùkŸ…¡¡—á S½t–ª= ˰ ž»`ÇO¬í9hè ðîd;†›Ð .‚BÒå5Í85Gã©ä”þï6“‹“‚oK€ê»ºïV l)&šNÉ’Ñ5&Vö{îʳ¼C„[n|Wt™.¡OSÞÕZ-¿<ŒÓ毪¦è¨,ÀÏmÔ7(+dÓ“w4Ì"Jß½8Nò1Ô wQ0‰«z‹éž·³äm«ãSÈ5»Gh‰&ÂÝï4‰¢rÅ”)€¿íàË„+cß%°Öp á…f}¹“ÕÇÕU Mûü1Õ¥ŠàQ@ÉÞr3h:»p+ѧ!Õ‹®ÚX2ðª^5ç—ÇçaTTðÌW<ÙÜ`Žù¢¯4ŸºÇr„#k6Â»Žøša©)ÆýT9Ê9?û„Ä’¦ÎnvY'õra6Õt4›K?6}éÉÅϧ´Œz5Ú£ÆâóEâÑÏ&0¢ 0Z7V×O8ñª·º‚U 1Í&ÞæQňã\¤Ð•ô}¯»Gõñ°c .Z-¯3©€@óIx¥èÃð_‚,.¼ó‘m¡`¬_£[ØYQ1yæÏáTM/Ä*æ5*B°·Õ6­ÏVƒT5 ‚ñgÏXéÜÈÜâÝ~ žGßi6ãäVŒ™qÊH`èÿb¬#ö[´»òø¢p6>fÁ=­ÑNs™"S×Þõ´¨ éú£ÔÉ!Dµ¢ÓÜŸÝqåèB6‚[C+¼¿=-°cVVØ kÂ.“(øÈ² óKEž›3¯ë£äá— X5uªÝye#ò™Ç»È&ßÔþpvÄþ¡ÏT©p ñðå¥Cš®l $c[^Vi%”@2ìuéòN „æÍ¡ëžl|F¹k;RsûÕCòþ‹½wtʹ”üEÔ“p³%'ÃÌl¡æí§­^RôlÎ=JŠ·‹þTç¥)JJC‡”O0·¬ lØIuzÅëþrUv¡°lÔÊâr%ˆ›}¾/ ãf;ÑGöݦ+à¶/`lƒLÈ&…-q?eúÑ´eáÝ›àê…»Éú?ℎ²koÂOÆ>ˆa6Jµœ²ßÑ>J©ö…D½Lºg,‡üJx'™loѲÿªnlž­‹mMñ}#O¼o»¥E¡ðŒt8–X”»,,ˆœ Ã{'CÎ-áyxˆD_6åA¿×ú;d‘ñ”Y—Ãò¤•Õ¡R‘³ŸÒ… Ûï÷:ºNWV)˜ñ”ú®QægÒ7š&A~ =ÝŠ]Ù\ÌÁ‹N%2drÞ ¯õ(ÊéàŒÎgeL2—>»dÑ¢´³à Øw¿û£ÜCôï^¶9 TüLÉyÕ„¥0ú÷^æ.¢lµ$˜Š ÒÁ{-aôï$òÅ~ÁªHOiŒ%—áÊ# A~Úžq ºßÞ!±ûÙŒuRH’ ù,>ÛGwwA• I¦§…´UãeP.‚d{åCU0;š±5äô ®`ž¾°®ÂèÎûfÚ_S$E¾“’¶åë`È—ñÀœy4‚¯gYüè/øä©>ªLaf‡Ô$eX–Êóýyw¦oiäǶ­T§þLzZÂû ¯÷dn¦~ùD½‚ ³yUË&WÇ‹×n–j7œUï•Eæº …’|à4¹z£žÖ:Ia;Ú”ÞXÖ7÷Í«±{±Z'R>æÛ{&ݾBؘG_Öy,å4I²P«Ÿ t~NÝn¢²+%Z¦ÎIɨŠWß.@|,dÅ^…Bª :l…nÊÞ7hÅb…uJ6O…¾lâøp1âijNaþÖwõœ›¶ŽÕ]\Ñ︸SçÖZØU$ÎÜÓ p «õwÅÒ F2yyÅÍ!+[JH1µÜž¾+qþf,<æ>iò•ÑK»ÌkLˆdÎÎ:²tP÷yg+&…}?•3 û/øEmƶ5axPD€'ë¤ÍLÅ$µ°:WmÍ6‹s@»lÆsEïKŽ·OžLõM„â¦è¢"v<âÁ%¢Áüœ¹’ŠÊ¶W¤&cZºðu‹ë äÌŒ`ÚöýÅ$ ›iKõgJ¸^JRVÑ猦ºð©s›–.ó4,kî¿Øäãõ8æ7L¼>å(x]OW\úà¤_S‘Þ-8pãl’»´ t{ðaóÅù±ñ]Ç‘¾|9æ€I/Q †´Ã%@7dΟ¼æŸ˜!ä|iÏ•‘¦ãòF`ÌÂô~ª£,ó¯v¯Pê²g(ÁÜûÄûÞõ¼'EϧŽê ìµjEÀ×[Ï„[4pÒôÛ—7@t¾`({€­ôÄ6‚VI¾éé|5¤¢¼óëˆ'ûzâ(µGVqfùú0`YBÈ0ã°Íé¡F¼³êpÖœ‚î8k™¡Ý¦S Ÿ¤·'rld¾KÜ6]>C ×ñ¤—o¢òNÂÒ®Ùi-3²YÃ/¿-§¥¾³§mÜ>†m#Þ4ÙK| i³KÐP#å]ò¨‹›škWc™ ?\­ò­>€Âþ¸ö|îâ1iÊÖÉC”œ öšiÙ9uG‡©¿Yj¯Ýg{¬ç«¿€¶W‹IinQºÿˆŽ|YL€¨§Ÿï¡_”x4'OøHƒk¦ÃÌÄbì³|SøÎ²ýû'C%XâM Ê ¬Š¯"ˆ#¥%¹< O–ˆCÁlVV›¸á°\n2(Xs”9|Þß˪O«¼PdþsŠmÔðc¸ä)’Vè<Æ’÷¤±\ܲúVÐGîaª©,½®ã”àW¿EÜ‹“ÄS½x!Ó8-%V¿: ùÔ[”TTÛð vOB;Ž¢ôûMâV<¾bç÷¡V>N)"…/ÀY] ­7Ü/÷¨¯ÎŠëÝž‘Zç( ¢róv"=®ÃÛ9¤0Lúc!£~¸ã}›<ÈX9ø”‘MKŹ]'yîÍhRÆÑ·S>ß³!¥žD¶¦l ”™ˆ™¾ìàÉýšÕÂ77ãD{U•‘Rj@Ð>(¸1A³¹5ÜsŒaÝÍ/øñš3s<ú}`'®hªŽ …ßx"4û^qæÂœMl>‰Nf0õ,§ £žGŸÂ2Z‚Rˆ®,˱ -ˆËý8ªTYTu? Dìc稶m;èñs„mÆC•zØo5Þo‡³£]ýì8Éù[”I6î47}ü­ñ…Dñ)c&š#èǦñt Ž-€1Hvp¤#M7f$=£¯'¶L¹NÄŠº.‚a¹»©Plÿ•ŽÐÕk ‘/˜Œ¡Ò¬ýzåÆJ†Ç‚ïå§½'øÖ $#Ît6.ÕŽ¶µ‡®I<Úá&9uÇ{-–ó*šÌ‚ÐSÙhD9»P:yܘv§ pšî}ø%¢’ûKÏy½{˜uÑ!¬í„Ñú·ÉÔYï–‘.‰"Ðïôv§J`oï¬à¢„NÃI¥xAJuCâ#t¤ Æi+i¾—q]Êmx°¾ÈÁn2ž‚e¡}I9ù»!l©_…œPö߈¬ű6r;ÝÙ/ð˜¶_§§øµ/o0M–ßáÎW+d|•“b†Îû¾™ Õ+ÉÁ,ö„µ‰E“?tdò­Hð‹áƒ>mççñ“IÐã¶Ä+>ec¦É;ï¨üz]céÇ80¸hl «á²næ¼xÁ„F9ú8¼RfÿE-÷&³—fv']ámsvð½£öÜ€ Ñf™î§8ˆÙ@—gÐTÃáý^M B7E©Ëåzáþ•™\D?§­×þ¼ DÖ‡/3‘å‘2€åý±?‰Ü^atís +o›ùõx˜¢hthЧõ…çv…³!($|[G”hÑ}2lÅ(–%žËT‘+ÈLÝÀ×Hé‘ÙÍ“¼P@Ó‹!èHÊCöìü‹jØ&2<ÂTc=ÈêCú)8 ÃvHEºÞ7;eBs‹ä€o›öØ4NÚ¼_úsgjß¹­ô,œ)“ðrÕ?²¥>’OÒ°%]Ø;0Ë †A‡SàAeœv“–7ËŽ °‘}a„¾ˆßEŒ38ž'“~6ƒF‡rò ݱàò?pH4iŸ,˜¾_ÇÏKàØÏµßs…ÈgÞßå ô~þµ¤žv“B>¶Pó² õ‘³Ê˜êƒÐØl‹>îÉŒªK*¦š¥WV`Q­C:CCºÃ«ALÎć¦ï_uGö´¬‹é\Ù´bÎ8Ø@ë(€›ÖI3KÎp³Åj·œ|38ô¾2-–XµßUÏO™»âödƒ^3X°m¶(E Äü?ô²—½£˜C~2YÃmc§·…[m_¨Ã@c4oa™T@ýˆ~½yºÞg:§KÉuó5WÓ~`èY¯:OØîÊ?Ùÿö3„¾¢úV£–¬Ö‘ßýFUö:^¼µóÞ]Bù‘#EJ:…G}1³†Ï—ñdÏ+˜öC6š›ôVDiƒªâ”ñ1‚›®)ŽÓ g˜ÙÏ–H>yûÍçõ…Æèôy…èRhG5ëÝ©q%iÈõ/–ÊÔ b??oûîëcÔÕMÇîQMüb½(>ù!<pØ’=yPxA§Àõ#΂ÔsL* ’~ [Õ9€hG‹G›Ë‹¢ÒÝßß¹õÝ’MæŽêÖ:•â$’):IµÕ7ü9³«òNªœå†b½[eçå$½×é“©mö¯šH*S¼×ÕBµ…Ts-"¼Hký47ÇåÖÛ’,’tX›tšQ4F*DC!Ràzm]Žk¤È}Vµˆî›Ab `Ë Oš::)HÌËrv´Ê΄±º4áÎú5*EO‘¿"oÆ£¿]*+„¢Øl':)!eÙç}[.€>òÌâ½NåIF˜)&:Ÿ(™'.:ˆ´¤{5Orþ®½Џ£E•µOÅ]ö]!O×­HAÕ|LÐÔ™..Öß³b q`dœžÜNÆ †©Žý‡Á W ¬'ªæ» tÓÛ‰@êÇ>L0Í–vIȱ+º©é–1ÿƒÃ[öháu^ôª¦ûœåƒm~¢<¡MŒ=“5WþÇ* PaUN¯Ù3$4 TT£–gœ"þ—æ)mú6yè\¦ ,dI„ýÁùÒ6ùøSÙJr¥TQæ*ÅIæòí, ÌB2è2ŠP‰ûs_?ù)ËöçÐ!!ÜU>S¶ë^“`Ö”gÿ®Y }ÿ0 /|³"ÃN“ÃËÖ{òÏ#l*(ˆ'sûkKJ·òÒwBGïºù÷HË¿…Ø]š”N啕(dâ뿸[å|ή„Ε=1‡Q§’@¹µ°1lÌË”¢YÀ½S×…é$àÊX«·¦K²£®­_OG[š4–`è%F“:ùCKÏc{„ËgåÚŒÒNûXu&™Å/2+ô RúŠ2ñ×å?{¨’¬'óÍüT`©ð½mä ØØŒ8E–?Têˆ`‘gLÏiÄúéR‘]yˆãFÝyÇ¢NeÜwë0+4žm0`+ §ñ­QHãÀ?Íóoºüð{zdæ Ý…GÜåÀpûÕ”xA«F[pÚÔj6ðisOF5Ýõ©¯ò%>2³ÑØIwNˆâ0«7‚ÿ'¥{d‚à ¤ßXÇi<nD¥ØòˆÍ;¥)M@…ñ©ŒáªËHŽr™šóVÕîüŽc—¯lŸk®9gf¿T¸*gjg2öÍÀòC~-ñe+¨CÅ•’,@aÄ@w+ÿª²f?8ºU=[½EJ‰CÒFí÷£5ËY¸ó¼Îe¿è(¦ŸkNö€ïZ¢Ùz‡¾K†GpêæºŠŸB²²±h¼¥JI:í/Ä wÜP}~W$5Œ-`þ…ÆmYFß§ØN¬z³¿Ù™Ù~tUIÂeŒ`¥Y:xá]^üúÞÁÞ A6­DÌE P‚«Ø£a$»›šS©'9˜´seÈ͈¶dšöBz\J5Ú¢‡¨ú²xhàºM3Š·+ضL!‡§1ÒûU&r(7i1,ëÜëKÔtƒðxþEèË'¼—·]ˆÔYèoYñ‰çOžf¥†*eŽ¥’Æ$Ç.³Ÿž˜“ ég°–õÏäF ‘• Bü ¸³%YzSå•KÙ(Å lfgur×ËÄž%+.p¸„©‡°ø-œ”Î ]ÎÉ,Vý¸–Ó ûÖòßÛ°Þý‚ê%Úö_ÓQ¶U³µît©P?[èü¨îønÝS$À‡jêÈŽPõõ5ó©HW4‚9ìE(‚3ã¤`ŠZ‡›½O›¤¿¥‘_¯§ýî>]¢™ÄKk¨$åí */J¤I!èêõw {ôìE„Í£šÚD·’Y¶1W‡;L £syi4;.¨æE¹eê™"®z3.>%‘ÍRNˆrQÍ?ÖÞž|Ñu'R@ à £ÊÙ=å¢ðm*’«ÚÙ_~ê,åßÑßôjËÕ(ݤ~‡ò³${Q¢'êXH¶ËØÆëbÊ(Q¦G V„0?¤¯†0)H(èB3Øh"Šñz85éh+Ž®|gøâcs"3‰ÊÌ:¤ê|–ºVÊPR@ØmŽ”‚ö¢íÅH{O.šñÄ÷âêàö§:Â÷ò¿6\)¤’·dq}8”?ì}*ÎÖ¬T‡ü# ‹'&<¸ç J ËŒv ÑvÊæi çgᇊµ†ï—(Ê«ãdÖr*vÖn NƒÒ\ünJKðˆ6Äï0?r鎢¢Ï•ZŸù ‡®º6&¡ú¿g®Ÿ@«O}¸J˜9òî—Æc¹BÛÍSú,Ó8»®È'KšP3vÀW•ûô%>)ýá Ø ÿEú×§Y Ÿi(ñ]m‹B»ùh1¶‹ï¤ƒLúáOìížûòæ9úЛÑ#ƒn 1Y¾`ÍUµvÛ°º²B^ŠË›ÙO¸Ñ¥ÂtÝIüsG¨6 ÏÍc…´‹®¡\¥Eð{àÑ;ôç5õ» Z°ÊœÂ|`¸bYRl¦~§jÝÏ¢…÷êE Sª]äÍ¥±{ô'Ø-Ž̼ìkny#OŸ¦Mj§ý×Yà O4"({u¹ÜÕgÝ©¤MçÕòÈ Ú”‹WPÍâ0¨ç°,Ñ´XË¥¿5ºm~ÃÍšÉ×%¯è˜%ãP¦~ÅjÌɳ‰Ùwêg{Г-U\¿æá~»³ŸÅWJ~ÙïÎsC†z u›nÆè¢d@|RÇç0 ‰ʸx‡„èSÙ‰ü2¤/e6TÚ™œ°6ÎGm¨KºbA8çOf`'Æ:A³¹4Hㆾ¼B ,A¶¾z@/vãN+^²ßctâ2EÌ]”Mâ/ÔË6ÿ(÷PH÷Íf·{ùHDÄùMgü‹Þ@Õ‡é‹ÈˆÎ;h¥D,Ìgï†ÀA›dÒŽAíÍ[ÆbÝ"È ¨É_L¢ +xÜ=(K„0—T <¬õTvšÆÈMÍP¦ë´ÅÀ6Ñ®q¿ª Ü,€XQÝÈÅñ:‰õ&t€¢*Ƕø0\s‚±™íó”rû·UE›»µ bØ&l ‚fûa&bOïÙüªt \ O=ëdv\0G´EIð5ê-¨eš;M_™Tü¤®¯Ù~O{ß?ôPÒL' ÃGŸ%i/æ)œÍñ Eˆ3‹³‚’jD£¶!FÖjRå‚I†>FÐ×õëòm‹ÅúõŒèyóà‹ïaŃêæ-L˜”¼b {ü2ùvÙRO+åi!:Ë^Ñ?ð¢m¨ºÃžX:ª—þð§çÞ2ÕyT (ϹȪÊD”ãâ¨Sò±ª=ÖóÔ‹i˜ûÂv«Æ¡ê/Ah¡#9é ÒCä¦ûÉEK¬ swV2ÇŸìLýÄra”îßüÔ©ÚК5‰ˆƒbzú.Åéu=ÒN* dÇòåÈFÌ;Ì/—ˆOA!¤ÓGoÌ;1rðI¡BZ÷Ø‹…³î { Àm¸ª #]æ[ßÓƒSïê¤%6Ó2Ç}Ž·—ršûÒüÕyã‰(¯iúÑ3¹!üéâçXÛ¯JO ¹Î/.Xam»²B”'̓0u>Û”1í,PÈ|h‘•'‰ÈÞ™!]J¢DÑ({3h]ä‡Y8dS®´ Œá1hˆñsÒ(iŠ·Ë€å‘^få‹ñË©Ë …Ô [¿>d´^•ª…åeþ¤°Þ²©|Zxi½[db+Ò˜"(lê ¨¬mÌÒ"[q² µ¢"ÎFË ×Š–6<õ²Ô›LÿœAûù¬œîi ‰…&š­ÊaÈ«fƒŸÊIš°å† tÀýÛ#`èÐØ%¨öÝØfJGï½cL,ô¡ÕÜ@ðS‡ªÈHJkq~€èCXeUÒîZ9Ýé”ðYÔ‡ûK¿»\©ù*qz·òê$ÓFöOÒœ`ì'*èIáþ-YIÓUŸ>Q3ç£gö5¬à 0b°½“äš!xºeÐ5ÊfÊfÌ7t:SU&•ÛþÉŸÁ“Ü|7`ò@–PT)Në]ÆžØã‰Øº:®ÏÁ7væâʨ怾”÷®¤qáÌÅs’?bóçØÎcž–6¦†µÙ ¶nxM•Vy™+©À¯Ö:¢«´ïÑ÷M{}`ÃÆb©œÊ±3¸øªÛÔ˜½ž³» f@øªcUEeØhŸ-®Žé’ßûÔ‘*$âCö†ã‰IævÜ sEW÷“‘o|6 €æôƒaÌ¥³gçÒ€ÉUý¸ö=¦Ý´ŠcY o" »¯³Ê…Öº×ës[‹å„F¼óe`®±6ž|ðW´Û¡ŠÜÎn(“TODÜþ¨²­® TåÉ’ÚÕt—€2ï‘GänѶ©¼Ö)[9 µ[l´µ‘I×|E‚@?çŒ|»V²Î9Üd³p«æå®3«OÄ{(߇6™öNªMxý„eFœqó5åž ŸIFI×ó# Á;7,¼vßEYUöz´‰’qÌóŸ ºòÓ žé¾W&÷ ðêÞÜpýÔ¹¥—›:O„ò<¨Æû„§¬€á1Dˆ±Ý¬=‡´ZÔœ>!®oøR !”Žo½ÈÚ Œè劉S”P¢Éâ%ß0½ê·]´¦gc»ÛÄrÞlÉo¥„ÃÖÏûA†a‚1έEMu‘2Z’°¶\অ44°­Xù‡æBFP ~dñ­\\Ðg  ëò%ÔĆåŽäÇÄÌÈ Gc`̧ᤢOh—ÜOæà`Ü'á;SUÌ6³2B³‡Ð n²_¶Ç–çЫ€‹ê4ó9PÚ¨…Ù•T5_ŒCK@n7nõ3°Ô¸Úhbq7Š$¿ü~SÇ9—@ o^Ô¾‰gcIsš­õ^[&9?¿¡Û©— )·Ö”øÜbÂP¦29ÿ1æÁÇmó¨ó¦ÖU+š¥¼0ä"¥x} rÀc4ÿ¹¢@sjhƒnÐ6c¥š ,Ú.[wÞÐIæ(LxZþ<µÚ#¹“žî·9<šíJ^¼Íƒ‚—M€§)A`yQi‡Là1úÆÿ ¼7 endstream endobj 95 0 obj << /Type /FontDescriptor /FontName /XAPWJU+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/C/D/E/I/L/M/N/O/S/U/V/X/Y/a/b/beta/c/comma/d/delta/e/epsilon1/f/i/k/l/lambda/less/n/o/p/period/phi/pi/r/s/slash/t/v/w/x) /FontFile 94 0 R >> endobj 96 0 obj << /Length1 1422 /Length2 6329 /Length3 0 /Length 7290 /Filter /FlateDecode >> stream xÚtTÔk×/¥¤‚t3HÍ’’ƒ¤ ) 0Ì 0Ä C‡R*ÒÝ-4H#Ò%)¡” (Ý~£žóžï¼÷®uïšµþóìÞ¿gÿöÃÅ®£' CYÁÁ($F$”(ii©‰€@A P˜„‹Kq€ÿ¥&á2„£](¤ÔÿrPBÃ!¬N‚Áúi¡uWH—IHa PòoGZ   qCÀZ‚uîBÂ¥„ròD#ll1Ø2Éøøaßûs5[§è‘xÓ†Í*júçîd¬÷1óíiòzœl£EV©t7ŽK(Å;âû0›n=.rîªÎ™X÷;Ô…` õàfíÖw…iš´Ì’†;«0 V}úkbôÜ™«Ž‘ŸÂÌBµSnkÈ´ž’ĸ—S,÷ô-ÜêbiõWw XûÎùsKEºYY”°Üp!çæEfÛv ¥·•*•­žV6û#z’>¥œ¸†]z³ Bc»´ÛX’mzò–œ”<[Ô³ÄS±ŽÀ{‰™“·_ؼò†É,‚=kmøðb`œîxöÚòÀgÒw£M¶¥…{OýÞ+è…Ú%•AM²àÓTÙ]”wÝ¡@ÒãUP¢ŒGF÷뽤Ö`)[n³Ÿ™Ål Ô,,s(+²¨)SžÀžÇi4a ¾ò”H"ư¹¿øië4$ï’áfë^=«=+¾<ô›ü=‹}‘²ëg4Î#™¤¦£›xO—ÐÛ-ʼEwñ¶t÷Fšæ‰8U~ÐË[Ì+ övxÜ&öFž¤8u> ©Nµú˜Úa^2N‹ÒÝH!Lª ä0©+>©¹=ÅJ¨€Ñ*ƹKÅ#ñÌž,ç¹ÛÖÔ]76½P/ÃŒ’÷„¥ï¥sïû‚ÌæBï‘(<è£hMôä!¨¸HÞ06g`vÂH 1–µX*W~ßå"^à „öí?1Ëe;ò³±Ý.§‘ðž>œçÉaÖ/Ú:“…DZȼÉk\Ì%{C HRIø©‘IX¢¯;ái syÒéÍŒÑä]¼þ±öá2fü£7ŒiaG6W*\Д3Øœ÷y¢×K^¾Æ8üH"µá2Ž”kFºòÂýñLª¸òŒs¸s"{V{îêÛ½o/ÓÓ…î@Zʼð—n5‡‡·Ùe%4ž ĸñ†Ç°·šºÝ¦] J£ìñî=®a@®UN (¼èØìênD¶EÞÿn9»a“Ú„™ù… ÷•‹V~“L| ÌÁÌ6z?"Š8¼*6üÅÔ c™:s{ºèѳC­a†oã».ʲs 1oôàGܲ ØFë#L¤ƒiwûy9¨„}=¥%éºvœº®] O•pÍnçõb5½É¢ìà,~O+«ÔäÐfŬ!­OÌõ%7Gû?ø¤‹9%Ȭ(Q#°^©ê›=v׎«]—|€ €ÎàRÓ$]H†ö”q~_³,[s„4?ÒîA¹hu-…r.LêÆEëÖß¾'•fqâÖF"í?Èž} ?ómLØ`µ‚æ­Â‰íÏ—:îšÙ^›aÏÛ¯º8“†~¹XÁ“7˜¡*ß!p6¬:9 …Ú¹ }Ë ÀÕ11¾/¶FpûKÈe³^Ðe›Ğ;Z7õˆ  ±ÌÔûw+ƒþåè­Ë—àp:eçIŸ¸ ÿˆÐƒ÷Y)ìÖ-€UµÚ‹¬á¨³SÐj®1¤ú8®¨Ìó…5óW.lŒenèJþ°­Mª'“ ˜!HkAšUe†Œ¦i¼³Uk“ŠÔVEµÉ<<Ü—I¯?üílϘ_88}?¯vàáu¬S ˜ŸYKƒk„oPɽ0ÙÖŠÀ•þ¸¡‹nõ©Ï¼e/?øæ²™að.—§&S¡¶šÄëÏruFÓštë˜ÜC»õ¤ç?UÇÈx'?e0:ˆ‘u¥Ðˆno‰gÛXëü¤k®?<×MUŽÑ”jßU'9á­Öç-p6ܹȳî~° 7Xû±P¬:PÌr¯6r!ô˜éçWAýèü5~þ-Fö›=ÄÑÌ-AÕo "1{YZI,e{ξ,â-sö÷ò­/tIÓÜQ¿P9íÒ³´‡¡CVO‡ëׯíÌèK¯h–EíAS¶{šðw§¢ð>áï¢ÂÎ’‚¸ÒiC_Ê\gKÖô¿düÚÀäú´*¼îJ9cpµÍm¯±7\¬.ðx™0xƒµi)™Úk¸˜,q6%¦ŽÂ­‹§Iå´8=øÊ˜7lT¹§eâo{ÎU¥*Ÿ§é'¤*ÜŸöÖh#–Áx¨íŠtÌöR±3ܰ¼HÜÖÉ6XüH§0O{‘Ùcß(ù½ê<ú'øF§°|Ÿ»ýOœ`X0ƒòÊL-x™'ô<`áÚ$àÛ³Åüñ—w`YI»wΨŒ3‹'~Y'¹h¯]Š`¹Ó=XÖë½C¿?¦`%è?ó‚ã ry$®ñrLN5¨Xo„¾¶„4uµQW¾üñnW%~’í\‹ç Tó£}ªÏµŸWò˜ l`1ÔŒúóÒSM ÞOГՇ?UßþT>eÕ-eMWb|K«½‡æ@Eî"vVÛÂ_R(1¤‡§œ:ñáòÞÛÉZ߈Ÿ!kÕÆö¥™rîIò›¬dÜxU:×LNsC!?2@9êu˜2†[-¥µSªv1Å_z<Þ7¸ü±-næuÊjawõÃFf¿xüf#¶¥ËØvÅâqÊb¾úQE³Þl<ƒ†¸N)3óµéÖä{ÉL§SŒÞGo\ßËT¿] E³8Ñú¯³yB+{3ˆ”_ynTÑÍ12NL~Í{ËöLè~ìPàB´.ÌË5£ªsCsPu—ÿšLÔ._ˆAaáÑ W’˜žã³¾2/h’¢Ÿ]¼ˆ;eËàíºÁÝÒjTY³ü^•ïL{9“£<õèìÐ>,טd°fØU%æÐÚûÅ5=.%ÝRo§±}ÜajXöþ¢v¦É’2U(¾’¨(&yuã%ícVÝÕ…ãSܰê6Ó¨‡;àˆNÁ0ÕõÑž9¡¼š0‘NÔ-J„I'l¤A¯ãCd…Ø>¯8FåOýÈrä–` Ú´ÊÞ€né‰ !²³IIÞ†½èéþ€ÛÐOuj—ŽØªPÖ°jÆ8`xŒïÞDN&ùóçÛÅñýU¢ýt„e¥T±…ŒR¨…Œ,Ъ ¶½[Ã^Âx‘¦} ɲ_ZŠs©w<ôŠ6W™•¤ 1S¦ŸX·é>9²éµ"|Ï¿ö½ÂT—X3–4v¿"\kg-št´˜÷œ÷ﬦ¹Zì¬Ó¯ˆBd‰ÐRÅÄ‹H„è%q.Ý|~N£5+Y»k¡ÉÁBøò£ØÏñjË<»¥·5-^о6æF"O鶇U=óŽ-¤‹—ûø…B«÷‰äiÀg ž»Zè‘»¼b%“*¨‘"îþî5•ZòÄMö]œâu·—Ö¥½¥ÒÕîÖ:!‰’[®‚_ .ðì–‹sÖ¼qlþùÚ1ï ¶A—ÑÄ<Xl_kÃðݯ@$¬ æÃ¿š÷ø8ŒqŸjýM|ÊËÑ3ávË„”OK8œ¸'ÍF Z\Ž‘Dj²ÌÈ×Ãö‡Bôèp3¾+ «(£%vÒŸ9—LF|ZinoööÙí˜Iôæ3 ’áÔKòkƒTþ\9!Z³0z¿ë}$}‚n ò§ƒGñÚL²øÏ8Ó=ŽW84Om§¶Uè_Õt˜Šjˆ%nt±6'µfxâEÒ`‘˳Ý5éÌËØpj>.6ej˜h±÷{¿QùønÀðõ7¤Íßgǵ Š#h߇¬#}×s8b<îïU|7 —{ßóT ?™l÷Ù"aц6À”–3 ™ª Ñxú¸ÝA»Ñ¾ï§-”P/+Ã3>þ q± ÒÏÙFC)ŸèÕµ³Ki‚¯x¼‡i¿ÍÚè|qÊ΀}—ãí¹Žž 2Hs ¯¾ÀÉàù:cB¼ªó(>-ÇýBX“ \DÑ––;Û;ôÇ„h„Š^N‘•$e(ÄO#_‰îöõÐÀe]Þx©ó„Xµ¾ð ¨CÉ,h²Q:®jÞ^ïšÞ„Óy’ä(çήÃÞè~^üB°ƒ³Ë-Ç’ô´R’UpÓvhXb¥òåôÝcJÏ;Þ›ÞÝ’LÙgôÜŽ?ÉÒJÄ‘¹wô“ zÀ9ã<4ï½Éhk¢ÞÁi×(«}:ÒI¼ÃQ/–b­[zŠãã–'÷4’F&=^—ãÇžà ºñyžƒ å¬þJ›d0íºÀó‹8–EõºcªïV±¡’ÈÓê¥iºfX&ÎÜsÄAŽ.H“ÒQŠ&ëõ…ð¶üg>RO¿9)çÉU®7ë"Ühµ¡:œ/é¬p%î„ ‘ïãÑT‰|*všÐ…´½É3ö_ìØ ̺]—ʘêáéë%|úþ†— žÓ7ŽÍ¯t\m!dvÚÙÁ#¶õèˆhqñ¥Õ’A ZBzVñð}.‹WìÎ÷ªï¨Ö–à&]ž_kîîm×lÊÛ XM¶´Ü5´eþ¤—¦ñm¦;V$3àôùÜ Ž°ÜYuÚ–kjÙ¦}²ÇËlѬ»¨’[èS”õc.uò¬¾×L ÊÿÍÌ¢©½¨ož®Ú®@œ¿ë°x­Èt³cY`c®W"™EäS¤8OüüŽ«#mg 7ŸÀøIصè[{vÜÌú6ô9ùkZ¶õ½Á0²—ŸHJt§¥q<ʹPKç?¹÷rgæ‚¥Iû÷H:Ä¿üˆµŽ¨Zµqð”®‹ì\S»ßB«âŽ™„PòIˆˆáÄ«y9åÎð.Ä3;—¸æoeÛ:hJö…ÜÃ×pDÉ}\ÞüøIlt ¥î¡Ai¯xH”ÍB]ûAû njB×5«Gøo³U¯|á¤f) 7@£SV ,Äû¬«ƒqnþ WYÞ6Ö¬®äGõ]á§ù¦EH;_hëØbÅ卸£ûGÍc,&öDMËø+!ÉCÓ—q™Þ¦{3¥'‡á³8fÙŒƒRrZ&$n‹I‘¯ãkf‡ðÀªŸ›ß“àòUIš" \1«˜î¨¹­‚D•¬øû¦ê¡îÚž¹NYK¯PO›oã=*³Ü`°»’úxgÔпù ¾øÄÚõ¨ YÔlLJÆ?Π-:yÞÆÇ”Ÿ¥ ·;l+`EhnâÍÑ×U, kÙù^<¼«_›Ç†g5ž£l¨šÚÝg³?ƒ|4Àíb.@úrKÁDSùZÛªQSñò¼¯^DÔ)Ýöžêø4ݳåuµL=Ï'#@ˆ¯é‡+z­¬§ë’Õ¸ú}?°x÷à"ÁM¨£Æé´jå eaã\ÛÄÃrýS°±}DŒš—[gê ßîQ 4=ê"7Iìg4¯^JCäÞK#TGÓ­‰÷i[åÓiá•mR3Çåz‡$ŽUŸ´¬ž”:n^¼V°žö…¯¬«vS¬X~öa+¿ZÚ–QUg{¶5S£äšt.†P|dß¶Þ–‡yz°¶50ÍoVhnÇ.n/ ˆè„HwŸÀ)=üîñøDçsá&. ¹e’ˈÊsŠŠo†»[FØC*ênæD_¤n™³©ø®DJ ªÉ ËÔß 'ÆÌHsγªèk³Ò5 jšA+ïehÀj£0ùZ6´”6½Þ¥¼R£¶BAîåÏ•ÉÒ¯grilšV<(²ÈK›N¬WªwÑ¿¢¿û¨ÜÔ=y©œìÃm>”Òr8¡öYðû "“~›)°¥µ µYuÁ”Í6¤ŒëÄez.o}ÉûvP‹X™Koòjl¡%˜!êJ¢nèU:ŒˆÓ³„Ʀ<–DNË&Ÿ<]OˆVjÄÅ×ç-¸™7ô"_—ÕHúž€iÕ f÷Þ´èpß[.ŠÙ´xçHœêÍ‹oÉÕ‡*þ-ªÝ{/@*ÊÇ\÷ßÔ+ˆ*çRž]LäÈ©O¿¾eíÆ@‰µçÂ,ƒ Ž’ŠÅ‘G H’!pǸY¢Êò)õq»- “íÛþY‘ý²* äûŠý‹EÐ{vðrZMäÀnY¢ +Ãï­Ý»dŒƒ[ŒžhÁh‚¯hÅÁ¶”Ðí<ŸåûLo ŒËõá™w|-8Â?)‡lLÉ)Ú̽ÎÍØRPÊÒÖmö5JÙªJ Ô ×(8S¾ÆÏójš²ñ $†®¾fåà ˜‹O¸Š–±œeWËíL}?¸C ;/ú!˜ÿþC‘²>+¸ìAø#zÊvJ~“OqûÒ¢È9~ùÝW·z„y¢:'ýé£ûˆ¾²kN’åÒTΙ=´ñw|^Þ ÉÃM¼AõÄÃX쫳½ÂbC¢Þ4€LDh§¬d²jf$5Nò:„?ß—PTÿÛó¹ú¬Ø§L.tÊ{mʦõî^viáYx1{û-ð›j‚‰_áñ‹4ˆQ»Þ¼ÈøŒ0Î j*YX‘M½7Òô¤麲 0’püX¼ÚûñÉ;š=õ~NÙR_ª­²V÷*ýDvá5Ÿ•®£öÀÐË«Æôù‹Y<±fšûd+µ¦B¦øæˆè6û“ËNxcs:+PËﺮŽtMŒÙ5¬Ô&ë/ÝèÊ‚ô—Ýïå|J·ëWÌd‹ÍœfZë~1sÕp×jê˜ÿXÎtŸÚÖò»R­äócm‰–wïÜBDŸ·¶pÝo±ïåTFˆlÒ¦s;ÐÈ`âI {(p“¶æ,t-ã+°º‚^?±:c‚<÷µÄmUb† .Vv"?°ÃMëããí¶ÏÄ䯴ãªË׉ÑÙóꌜKòm´Iªª`¡·º£/m;=½áIÃðƒ'>2ÍÄNKD‡@dœ‡ö\A´¨®ÏR÷êÛ6ISÁ±·Å…S}‚wéâëÆ¾ì‰¡Ë,ЬGÜ9ç<ÙŒŽ.É7„Èùù†NåÞ%Ì7ö窬¼NÕ>Ùµú;ÿ¾-5•}¥rËkä)QFeëÙQ‘\ˆ[†£ªyü;aétç—Þí83þºtð2Y:qå€ÆÌ£[ÆÁê• ðôv  bðõ$HÛî3~"H‘\÷þáòÍ/±€re¦ò1MïÎp>L¦òûQ+¹àÔÁT=ŸÇ¾4$³khvKþ!~›ýüqÙ5Gb¢eàUrβ¥þ-é‹™óŠB³‹QõÃÆÙx9÷U9WóÀg’•¶®w´bÞµp«Áp}~hÔÉ"1Òwm*æ[gdw.å@uH“ˆ\´Ò™·º8ÌŽúqÎܧ¹Ÿ"³´·çXl3êVvy™+¦au· Ït»4òóo°U¿Æßyÿ™Q)ok'üg·ü¢Í%¨à'ÔýðkO9i™¡/Ù úïo?MzªXÝ!'”nu>>Ö ½•xja˜Ïû‚˜ÌZ²Hhì·J-š endstream endobj 97 0 obj << /Type /FontDescriptor /FontName /VPXKNP+CMMI6 /Flags 4 /FontBBox [11 -250 1241 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 85 /XHeight 431 /CharSet (/i/j) /FontFile 96 0 R >> endobj 98 0 obj << /Length1 1571 /Length2 8489 /Length3 0 /Length 9528 /Filter /FlateDecode >> stream xÚ´T”í6Lˆ”RJ×ÀÐ9C—Hwwç Í 1twww·„   Ý! ÒŠ”4ȇ¾ï9ï9çÿ×ú¾5k=ó\»îûÚûÚ=š&»8n•Ãì`@RYY^qs€@\XôôZ¶èßf,z¨‹«-&ô’.PsăMÊñ§ ‡Ü`n˜OÌ/¸@ ÁÂ]„Ræî¶€2@ƒºbÑK¼\l­mÇüëÀdÉ ò³ýIˆ;B]l-Íaes„ ÔñáDKs€&ÜÒŠðú¯L"6„“'§‡‡‡¹£+ÜÅZ”™ àa‹°h@]¡.îPà7a€Š¹#ô/fXô-[׿ìšp+„‡¹ ð`p°µ„Â\2Ü`¨ àáp€¦¼@Õ û+Xé¯6Àß½€9Àÿ.÷wöïB¶°?Éæ––pG's˜—-Ì`eë¨Ê(q <lsäw ¹ƒ+ü!ßÜÝÜÖÁÜâ!àÏÍÍ2âêó‚Ósµt±uB¸r¸Ú:ü¦Èù»ÌC—¥aI¸£#†pÅú}?)[¨åCÛ½8ÿš¬= îóùXÙ V¿I@Üœ8µa¶ÎnPy©¿CLXÿج¡/H€Ÿ[€u@=-m8—×òr‚þq‚›øù8ÁV$ ~¶VЇ?,Wsw(áâõóùOÇ#,0±µD, Ö¶0¬ª?˜¡Vá‡á»Øz AÚ@¿ÿ~3~sðú'üÏ|9e4%5XÿbüoŸ„ÜàÃÎÅ`çâÀ`0Àÿðâ÷ßeþÝ€‘ÿcU3·ýûr *ÊÃ¬à€‡*Hì âÏpcèÖý÷=¤a–pÈïuäâ嘻¸˜{a=âñ|À{ zþ<€“G<¤8û¬à.X¿Ç, àÔûmúƒùœæÿ ŸÅ¿˜ëAÿápZþãñ8mÿò8íþÒâ„ýr=Ä:ýG0˜Àéòð!ñðá ?ð¿¨[º¹¸<ôæbúò/üç³…zB-±çà–¡v¡.ÄÉ=Ø·ÇEÐ3/õ¸ØÇKM0ýÒS¦ëÉšy¹KŠÕ2‹}`»N çËüÕùSŸ­&ê7^<çìÔ2»ÖÔ‰s÷çÈ3)>´s¸ïŠtÓ%¨„*\ú‘Ô(cðº0_B¬û5éq}{R<žWÊð+„}Péèj¨ÌU"¢Ôù¾¡xo1¤µÍû“„±`Ã1!d3Ö(*C%›VQ¤ã +Ù£ompdh¿²;&HÁ=xûˆ./-KöÅ{Yp†ˆg~éIfG˜¶ àBà…ú ŸRƉ—•Eåã\vr7yû|q÷‚î8·8ÍØõ¶áE|j•:61„›?« ¨¢æ6I…¤ççÄ™*"–*}êU¢“üûŸÌ ŸÕR=¤vÅy!—¡tö±Ø£Þ$Ð,@º@×R¤1.©éX¼a³ì¯zÛ\Øñ`ÀrQ•ù°¼Ý(×HÆA hz§m”Hø$?o?£44žs·ef¼ãìì4{-‰ŸÈÍÿSÇv$pi/ñüÊ­z–XªF)¦Ø†®c™¯* e# Xýíæ¬¸ÐÉ{³0[é0G&/Àxv+óÃGCrýÛó¢Ý«Ù7ƽ©€ÓóhÓ³yyðÂùÖÝÇÅêr‘>§÷—OŒ(Úd;3Œ/÷örE¥˜¤ˆØÚåÀ_‡‰CØê–n íÅ»²ìæ9°ä=pWpc¶YÂ?£:uW*FÞã×ÙcSl>¶øÂË1p¶I•3yå\»DÓ[w8Ò6ý¾MQź-2à}¯žÈê@õ+‡P¶yHhæè9~ëêaòó 4óä÷7jB9on,ÄòQHŒ—Ä›§ËVobji¡[g¼4Ì'¶$ÏÒÉ„r¸=èðV)¯Ëp\#Òvvö¿âÐû–þÌÿô6k¶šþþiŽ*mýwFBÖ¯<# Ó¥éëäÍ?¬NBpÕð1÷€#êç®7QK+݈eÐÁ,pŒ[ƒÕÆÙ ™›÷uÝÝ—bhUsÈ›sÇíý]oòbëm»8ñGdE¾Óîñ‚PüGÎç£+^2†Î=‹¸=(vaX{çÛH)¹; ÌvSY宨¢ ä•é\ÕÖ_?e¢9iítŠa(sú¶YüP«èæßµS’_›Ì#žJUPÓ}>^ï‹ec» «Ýo–ÐWçåÎ|ì×¥Îßáªhõå¡4ï{sCUùB`¾ œV¨üÀx"ì0Ç ¤tRŒ"ËAýD?‹JpQút« ¼œ[¶¿¹‹{‹§èÖ§M²DXÐÆ+f:!ažÉrˆ6ã£ü!!'„€}ðÝ Šlqx ¸[‰<ŠW»K´w3*º¯¬AüywÌ qο0g™¦ÖHO›ê‘¨ÐX4-jL/ly?s .Ö©Ö%¼f1Ö«zyóôå-’¢Á»Èp$;1uÆèjbC¯ímÐî;‘e"œ…«Öoiö;x>çîì(EŽÎÎa.}òŽ+3ÎHNAó‚p•ç1­- ‰fd"¬ê(¥|*÷•õ¯º%ì.¬k¯Ãr*üÌê‚7sl舳wè²´äúÙ Ez­Meô¼…·.Æ.n: †y¤B-þX/Œ*¥Ï·Î¦k,ÑÑpÑ¿-V¯{®¼‰÷:-º£ŸÌ¢ñ±ªöbò ÓöSå»Åx&ABY*>øÖ¿ÚÀP UnI#y ¿ò”‹7%¡ç|Ù2×cÉûhØóDáõÓGùI­;aÿTa|£ë$&LåÇÞËöšL‘ÈWï¢J>M´í¶,v¡!|\Á”¬õB/ß>­âf¸ßb?’n×;á4üpWâÞãÝ"¦xÜkÜo» @ÿ¹Ìê*“(<æªTT/¨7"é\.6ÀË%§“ôßå¡¿üüåbMïþF¶öžÊ-×>™vtP²`°|âŒÄ×@J^¾àæÍ«U Á*/-cøùR½~X·õ+ï_ýq‚Ö›Ú6äP™ý4§CæfȽXŠ”"ݚÓC¾Ìfý²ëz WÃ[ç'o 8üº9}ÈŒKçHêš1‰?UàÓÑ©»KOMë´S¬ èˆô |õ£`¥¼š¾ÓÆ|Y½W¶ÄìÝèk m›ÜÕ?™ÇŸÔPè“ø2†¤\…ÔˆÜþÔÇ/ ’ÆS-g§{Ô˜È8¼Û§jDTyÕ!!/ùh+8 Á×#V`½I;gº«ð´TC쵇D_売X¾â `®‘­PÕÊþ_äÁ¤-ÇOw’uTÅ·ägƸ4À§7 ØqK+b•]裤´óß%“Mƒs‰­ WÜ$Cu“õläú£_)qlÐkYºøS¸JfËkŒ‘´D¿aœÓ[€ŒóÕ1¦›œˆÆ $'i¬u¹éÌ{OrE½cZçrÖŸy¶>.8ŒIEKÌ¥ú‰áUnÕÜq?cï)7ÅÏmóm!jö­ ¨È 8Qs‚~%}©æE†ÝcO^=]}©Üâ]tšÈ³w"LAWÐ9›^¾¸¨’·6-…ÍóÍ7b脤…;\ªûÄL4{$ŠÏY–ë©âï2ߥ"zGL¾¢/àÐ Ó€XÈG°#!d›n°:î‰êôq}“—¨±ZkxcB(g –OsÑõ§«Ç`Ï2C¢ì/™›i‡*·ê¤>¦˜f)ó°{:ólÔúrt)4=ž«´ÝM¯ïb £Þ~i~RÛWE”>ieÔy©õmÄÙÜh$Ó-R:7µ½A%6ffäjÎ~OcÁ¤­© уWI¤ëˆƒøÌH7K䊘§=›Нg×Ö2gaèâ§é‘ôyfÕ9ñ°R|½Ÿ:³:±„ÂÂŒízx4ì ®q=ƒœóKöîðXçCTÉ?\¿”IÕïB1×ýøCÎ1$o¿¨ÑÈ!“_Ô²ŠåF *]Få‹iuÛPˆ,ëŽÃ÷Ì-4b§[[àm{_Þ$•#ø,À|tgóš Þb%94ËŠ’{XœÈ»ö}t½¾×«Ò‘îAðÑ$Æ>÷ƒÐ³/²OoD45rEî|ƒX£ƒè­oàŒ ìØÃq²ØÇÐàÇàolyêÓ±¤_3õ|Ýh$ýTwºó(Oó;,Â{ s·ìk®†üF„½%ÀºAF\­5-Ü]P]?so®°âˆªéA‡}·±|5z_Y ;•æ3àIÒ¡ßúq3v•·Ôö)(÷?)A±ž9·ÌËTß­|³gȱž(ÄAKú*´­©Ì‹"ÚÞ ãçëµzz¤g¦2N j/ªQß9ÔàF,­}c\ …$á³~1l «¸«>Eê}©ÚêÍ»óüOu’ÙUŽdà‰ú¤Ñöã_&}-°q¥&¤²"I·7…H*7uOü‚%ÈQÇWz,IMèãÅ©Ô7Ì“˜.þ®W´ÞZ‚¾v¯•F›¿:_k,½ž )’UZ}Í#0éhrdz!ÞâÇî½|ò4a r(,ìù°òYNƒ'Âÿ¹L‘©‰ïôá.ž ‚å¬+ZA5÷•ßË—|ÄxH/¼ã˜ ,ÓÌË•vôB­¨l—•HpöùL4yØ0µ—­¾3lŽöãËäåe ¹Ôä\Úe›²ú¬·ˆû;ŽøôëÜs±Ü´$?õ£_ƒ¬Ž2Ô‚³R³E£UGu[˜OãЊ;‰hª–¶õ5›îÂa*¨e[Á±;é@°ßyÏ|Ô1 êÀ1¡Q©á…O¡U/>ì3Þ«|·Ä˜cÑ2µàú‚ØoX‚6ÐûØÉ…–Òz%AAlI¼Ÿt+ž^pŒž^iå]'äFˆÄá”+òìöËŸçŽë+íº¨-‰”t£ïñW ìjfRߥõ¦¹¶Z?ß({ï Õ%ßA:•¶—¥sâW/ŽL»ÎJbAÏ€ï¯<§ ÈV•'7;R8B„[RXý1íŸàyÕ{o«öãŽv Ú[]̲ËLšª­)–E +)Ä‚é´{˾W YÉWFíºˆÛyRŠSA¼VqæÆØ:îX}ƒ¼m¢ÿ†‚cÏm¸D™ÌÒ ¶ü{™ÕƒÊÍòH¨lӥШÞòK©›cÿ’è¶SÅmmú0nçGï‚Ç ×hôBH á•xÜšþÐÖˆ­m}zfü ·P§0bÊESúõÇ'¬Hulo @é_•§Kéþ’i¤ ¼ÿ(f»¼c8“)†zà/”Jvî‰pO̯æˆ~f¥ÝD÷ÛijÏÇq©É,<>úš¸dðy›íãñúÍ/ÆÂÚ‰¥æGO£#qëyôAT¡šC– õ6Ùâ⎥éáÏ/.~¤î7œ»z}ëlf<50Æ÷lñ0˜Û·#—}›¯*?Î]Î6Æ» ß9¯úÒùºGdÊ^¬œÞUÞÈÌL\>ãµ {šÜ7”Ígf¯Ö¿Á?] €:ÊŽé?«¿3qÊ™ès v5NmÒäÉál"'^Lpë”Dßk:åayû\Ó ÐùÈÉìSiÉQP´ƒTÝZêTˆ‡]Ãc™”&tG°°ßC8õ 8á}ÔÍF"«ôQ Ãývª([ÑdNe2þ¾Úâ8y­3aW=ž6ÂÛøsS; ?xgóo4’eB\¶’§ì’è¿wÿôàcî ³ÉÏ!ò"§x,:µ/˜FŠ/¬—¬Û^ü{/a/'é¥Ä]ËdkW3…¹ó ×ñ¢š°Ø åÊ}»ÒŽ£#Á¨ú93ñxcœ°v+ñ4€3‹ Ô³´ œ±b©ªÏžËf§HP}¸ºœ¢#ÎXôCó¬âÖy?í‰ûöews•¢÷"ùv%ö© Æ^ÓPV×·¡Éóz±§yƧe|.®LÿÈ„¸Wõ‡Mh™tÕÛØ‚=#|R Ï(ɹÉ,`Õ“ÿðÅ@7ÙDÒÓ³”Þ=£Š¡9¯5õñfÞ½ïÕ¦wN ¼=Qš s&€ìñ­™—Ÿiz;nó¤ÑÅñ«Os“7óŠ™ó RÁ¯Ø$*øàʆ¾V•Ä•lÛwŒ5©† Uœè“ÆŠ]ÚÀ2Ó„XŽ+vÞD‹©¿ÌN y«&M"¨yjTs]µXY(Ü*(k­—l"¹> ;U¶QWO)˜òTžÈÈ’r_0Þ‹m+œ?¿®›*¯YW¨fy¯µQ \¦=²ùL6l«kÊûl$+ŒÄ\ñ£Ve5Ù)Ý<0)â=Úþ@Ð@ƶ>ª9šã×ÀÞ(‰V'ºÀ>¢g {ˆ;o‡0ù5=1RDNPÿ—wkÍ©dA ,ׇ?ÚQEÍ\z8õ¨(³ñÞê0 ºÂ«QÃdC{4ÊÂŽ»0f¡Œoz6® KÙò3ùâ¡BØD¼k7¿>©ýhïÃaHãºâ$¬Ë%mÌ>ÿŽ%âÉ¢ŸC€~qXI·°ÝÓ¿RÞµ7]\^xú# :üœÊ{?D’k$JX>}nÓÜ’‹3Ê7¯µGÔ|i÷Nn#}AˆUó‰T‹1«\™V?/ñóž+Â$#!£ëÚ3ÐÐa]Õ U~ªá‡Õû.Õk{âz/Ê:3LDÕ'µp'°|ÍûÅ/Ía=ö X“Mì°Øˆ£û\¯pueh_hoB˜ 2»ù#öªÓ;CòòÅîž‹žnNñ€š[66Wß䚺¿ªp{£®¯¤Xë-FÚºe}ãiå8MEi¾Ë³5ö:oæÿñˆIè¿°PÓ×Z0™qò iáþ0ÛàÚ(6pµêè¾9æÍ¶ïìD%§Òo¡±äAÞbËEIÿðäñ¯¶ÎÈú¾1M–išètzn†—£ÍåFXB9GÚÉx–Û`îàúÊñ)¾h^R ¼}v’TÉQ+¹+îpLä)~ýi]÷ȱ.2Úué*GŠ?*B–kóôv‡C™ím÷%³AZ¡N‰M7Y¶¼´&?‡æ uýYŒH–b|‚yæ~“‹Âò-‹{‘›î{ËS‰;­–§‹”Áê¦OÜô d³P6ïo­ˆ·Nó`þú v_ ½å¾W~ùke·ËÇÜoá¯Ê ­St/êè½.R‚SmÎMÜÆ•u˜<».µ(GšsíKâêM›ï73‰µÆU“!=ؔ֎ò¤À§6¢x6Zž•ÕDu9ZP• V*Ã@¡óaj‰Ï’22ÄÂ'-<™ ïÉÐïI<±J­ö-qm¯2{¢,'ëØ`U×d^å’|9h. i<9¦aJ8.qØ96tMãK¶ r®ïÞ~NX+””¬‡ amAãõ‹âñ;‡Ž¬f=ûLc¨¸d¿Sv¥6ò·†çmCòK^)'òW6Ÿë¡â£f±!XlXÒéè{Ís‘+.‹+§2¶nZ4Ý5ž7èØ˜UØÙÀL¤3Ÿ&M·Û>·’ø®ñªêg­¤xºpø;ã° ó•x8hÂï+Ç¡q‚-õ½Ü¯‡#r®BúB¿»v¨'ľÆÑ>«JÄz«?jIEí;Ez]lA»á¢'« ¢{ëÄââúqÐ )(É â5T 1 ˜uqjäÌ}\Ò7Ü· ‹Î*r: KlÅUþe´<ŸâŽõÚ´%lbA{›þµÂ-®U¿âsø+}=ú„›{ôv†%Ì+=Ú€…Y…]Õ«šï£2´~œ©´váÉq3%›œ–»¦é8{-ký³¿¶úªµCªê ­+ꃌMN ÓIÅw…$möÃC­¸\›bƒ_ÂE-â•~‘ê9ºäVó¥’ÞZ pHѸ <0#ðKMz³D¯úùTá¨`5”¬žbÓ}åTxðhXxCÔ3u–òÄÀ7pŒ>Ÿ ®n?rì±jkŽõì.Û˜ÜêÎ9» #Ÿ²…ÞÅWX¡¡—¼|_2Ž%î´Fs6Äòy'4Åé`Eó;ƒk{NÞO†¬ÝHð2£}¡Ë6Ƙ³®a8¸QR”ÒV.@Â_4Ô{3½ºƒ{PB$ôÆväÀï7¾¥®Úskù¾è2Ï×t\I'…búG^d°Œõ:ׇyYq;Pÿ•ͳ¨‹*Ëy¬°ÑE6­5¥1êŠú ½ªmg°ÀŒ¾dÞ§vѶ­„~A8™ •=˜Ïø64rëãŸï?¹?:.Âgø«‹Òd<³Ç±¡¿ÄäçÎS•ãu_±<†ÁO…å_û5°{¡+H2XÓãr­°0læ„_½µ¼á¿³’zQ~&ûÌøøgìPXfæ£çÎç½¥+ YúuC9“›ˆóErüõü˜’ûÈqá¤:÷   Õ:äLÍÿ…äÚW'r¶¥µÑRÆo…;ެs"×€¹K19‰‹üóÑ7ΊC¿A̱DŽg´×b3/9etˆ.Rsî}z_°é™åÙC^ÙHP¿c•JÎÆ6[ãúô¾i¿õ5ÔÝÖP¶e#¶g%‘Üó¶’ƒœovc¯š`’íó\9ô2 _djCU|Î0¦‚¸:¼ß³X¸`Ú-³§ë›ñÑ :­°Œ+¬‹”µ¼T‘D]w>~…†Cƒá¶÷©B=³;g¾ÃÒ-ãðÍÓ†þÍÝΤ“_i°·X×ë;3x$æK›‘óÐ/u_ïZoq¦/‚ŒE–lùee™¬]d[~">Ü©B‡Gh²O í(Ð7ë¶§%¨h *Óq™­nЉ²6ÃVðcŽÙ­¬QwŠ‹6.k×v9„†îؤíMË8J(©M0ȹša ™$£“AÖ²&„Àº$3_ÞŽŠÇŒ±áZÑ ÈN5E¤Ç\¤òèÙ÷ŽxðƒP:,n>·½·¥iÑ,×¥ÃÆi.–æt pÀ’UaGæüRúJ7•7¨»_–Ón:´rÎZôaÊšÛÖ•84ò8ÛŸ -¾¦”^㛊jJ¤ ŒÐ•°Ö¢ÀI`§b!Z*q æKþòxr<]²ZpÚ†¿µ8Ûg¬ýªé…•»ÝÈ RÄ®íìTÐháq¢J»xKÛÈ…Ìb‘zåø_&óò¢Á°»g*.8šéóTùÌ+À¥^s𩇕ŒÏÑà7ìÕ4EÒŸ[¶–EN¦0¥ j¦„VðÜ7—ní6Úƒó»_ïQj¨¾ÄÌ+F†ŸGîéð4i—#y$®¿³Ä,œ|¬FÙèoÂw9Gú¬-§^…r°èx­Øhl¡Sþ '«®-¹çÃÓÀëò%î½2;ä°vÕ*'¥­"Ë’Pò¸íN²Ïk#`®f^ÏAÓob²rV5­RÉ~S[²ý˜œðžKd µcPVtÍæÖ²Î€³ßkÕk'à› É ¯-€Ñ°d©=OÙ ¸ ®õÒÚgÇmãƒÑi‚b”¥3ñÉ-ú­GHì¢ñoß{ BJ‡î4#¨žqàýàèòÃQsW_¼(ZHOá5½VÍ¢Îî¡¶-lŠ?ŒÄ<'¬yÚ$Í9ÐÓ7¢Ú/hdÝDb+[v’øE§üzQÈmcÉY?m|‘?zé¡åÈ“×ÜîÞh»“I„jKzHL:Öë)Q>§{, ÚþþrÎÄ7mƒ0ø³°)L;PÜòúDœw÷‹ÈA)|„RÛÖ@ƒê\Œ+q“} 4f$C§<ܧ]9s §Kª‹héé—¼‰> endobj 100 0 obj << /Length1 2691 /Length2 22303 /Length3 0 /Length 23830 /Filter /FlateDecode >> stream xÚŒötœë÷ ÇlÐØ™Ø¶mÛÎĶ;McÛv“6hØØVc5F£ïôœó;íùßZï»f­™çÚ{ßûÚ¼Ÿ¡ QVc1w4J::¸1°02óÄTY˜ÌÌlŒÌ̬êÖnvÀÄš@WkGÞ? Ä\€&n ™¸‰ÈNÁÑ ën`a°pò²pñ23X™™yþgèè 7ñ°6(0d€®bŽNÞ.Ö–Vn šÿ=¨Íh,<<\ôˆØ]¬ÍL &nV@{£™‰@ÍÑÌèæýÔüVnnN¼LLžžžŒ&ö®ŒŽ.–‚4ôOk7+€*Ðèâ4üJ hbü;3F €º•µëßr5G 7O $°³6:¸‚N¸;˜] r€šŒ<@É èð·±üßô€j`adù×Ý?§9²vøë°‰™™£½“‰ƒ·µƒ%ÀÂÚP’”gtór£˜8˜ÿ24±su7ñ0±¶31ü¹ @RD`JðŸô\Í\¬Ü\]­í~¥ÈôË ¨ÊæbŽöö@7W„_ñ‰[»Í@e÷fú»³¶Žž¾ÿ ks‹_I˜»;1i8X;»eÄÿ1‰~Ë,nfff.6Ðô2³búå^ÝÛ ø—’å—”¿¯“£À”ÐßÚúAðu5ñÜ\Üþ¾*þ‹XXæÖfnS ¥µÂoï 1Ðâo j¾‹µ@4{,æ_ŸŸ @ãeîè`çýÛü¯þ2éHH¨jèÒýñ¿:QQG/€/;3€•ƒÀòkȸ@þÿuóoþ—ü_Reë‚û㌃…#€çï@Åû_ÿŒõ?+Cø/ƒ¢#h–êߣ¯ÏÌÁlúbùÿ¼ùÿ7÷¿¼ü¿þÿ HÒÝÎî/5õ_úÿµ‰½µ÷? Qvw­…‚#h9þ¯©ðïUVš[»Ûÿ_­Œ› h=D,íþ-£µ«¤µÐ\ÙÚÍÌêïú_@îí¬€ÊŽ®Ö¿.¨aÿGZ83[Ð…â êÕ_* hŸþK)á`æhþkñX98&..&Þ ÖƒÀ—´¡æ@¯¿FÀÄèàè:¥ç°ptAøÕQN“È/Ñ߈À$úq˜Ä~#n“øoÄ`’øq1˜$#“ÔoÄ `’þØL2¿ˆ]î7±Ëÿ‹@†ò&ö¦æ&¿õ x~#P<Š¿(¥7ˆSõ7¹RûØL꿈Só71hý‹x@è7;ˆÏô7åhêbbf ½K,Ü~ËÙþ•ÿ=ùÿ*@šýF  ̬]ÌÜí-ì@ úŸ˜Äaæh†ÿIØ‚.¿ßqüš&ó? (ào |€ÿ!fa¹Ú››¸Zý!… š¥?eœ¿ü8»ƒ6ðßü2ò2³3±ÿƒ TY‹ßtÆâÈþ ZÿA¿!Û/èñ;X–_‚ß|¿ÌÝ]þ`XþAþÌ+o'+ Ã ÙüÌ jØüA µý‚ªm÷uèw¦¬ `ìMÌ\þè è²eú 9@Þ@Kú‡TÇßñü9þG réô[ ræz#;üg~ØYþ‘þwzØA‰8nKÇßgÕÆÉÎÝõÿ ‰óï±9»;ºÍMíþCÁÆþ[ñ_6ž4ÿ³üêë]aù7-è+ÐÞú¿sËñËèñGo8@N\­Ï:;¨|®vÿ™JP–¿iA/ &7+àà *™›§ã@>Üÿ€ †züA‘yþ1Œ Ó^@{ï? ¨\>¿ƒyòºüMõŸ ×ÌÝÔ·¿^‰  úþëoè4CX^p4ã µi í¸¯Á÷dØ›˜¥ØÓJ§að]vétD†M¡©ËÞp¹IîE]Û‘ þ!¼Büì{ü©6âs’JûO¿'£Õé½v„¥)¬¯“ÅÇ"M„ð êÂû~ÏÎ~šA¶ŸÀ¿ÈRä;»s#+¢ß{öKy5 T®Ž…/ì©ì×qÊ!>UÎ0ÄiÄê•ÍQ˜æÌã¸1ÂÑ¢]x¡Ìý¸EË›|%–M Cð?‰c+ñÕÝd}ÿ0ï³^­ÎêÚKŽ«‹CùmlšÒWô0U{Ñ·¼díë’×'þâ<$ú´5TÆCÖ¬:kÕ‡Ö¾±å.–Ýü@(þWôݤ†ŠV #R%̺v£t·6[ áa—…èzÇ­f…MÁ ÏLÓ+`IÇê8°­Ç÷g³óúðÈW†»ð´Ï÷í#;=þZeBB–ž,´VzÑk«¼Y„iЕnãÝZ@ Túî'h“0~ßàKÑG<–íbff]nè>ð1‚3Ž|¿Ë÷þ5-Š]›soí<Ñ B^C8_M¿¤]\ãZ±Ig‡SdIôD|‹§Rák¶Ì…T2]lÖ`1ŠWxË'ns,]’—°ËHõí-ÖGë{Û<ð1žiÖh´à(R!Ù=‰a?l  ëç¹5?ܾ­ª5ÄõR]™M¢€$”·á-ÕÛÍ—)dY”:FèåþF1Ù 0Y·J0Ê>_.˜œ dAšS}SÝÇó ÒóøI×1L#È’‹–ž¤’Éñž—j‘ä~u¯AŽ*üiˆˆÓÆùsˆX˜+üà2^ùÛìÏ›[–²¿ []|Ëݨ9 „ki©¯ÏÆ5ºæJ=ÆB4ˇÖKéºæöRægg™'¼p›¼—[AAâãÌøÔæï"çÎrÄö3SC Ú:«YÚ­¦qzÖÿÖ‡há%ÀŒˆBÃùËÙ'¡¶¦è ˜w@>××àýÛ‹JÄè’ÂÊEñ®).ryëÊ$‘%Kü-;NoðYøxI¥–+~9ø½cåñb¥÷""8#áIï¾ÍØR´Üëì-6S-O€i­Lí…Pv2É^p13™¬á­—gÜáMsñzŸû}D¤KæóØÀ¼¯ôò½>m¯»±ˆ¸×Tæ¼q™dÔc/R[b5õL¦3¸UŸ)ñp¬ÄÇ×H3Hˆv«øè¸œô‹=½pÑt¶¥ ×9$jÇד0ÑòŒÖ>¥JµÜm]R‰¼±kEŸ>c1^ãvg¼v§;ÛY&}¼ä» ç3Üù«Ö9z£çzØ×%Ó×¢=R¥Ò<ô‘q­>ZI4ƒu¦RcïÊX•øÈád}•”^|IÚÊ5ku¬Ë¸CCEÀ­¹Q‘‚qHq¸…z“âg9®ø:8:ÕŒ…=—’¬§!è§:Î@ÿކ%‚7%1wK:Ê7’à…'×/±³r!Ýó(Ò±T-‚àº*Â(2Ë›nE»¯©ÙÝzå4\ý”]OùV9opt¨æVMz?ª÷]Ri»¼›ÛÈ`r0rH”•/ÀB„£NFh<ÈÊD¨iQ$ô•“lÙX½0´Í2ýÖëð=ëÔX¾®ÈnÕ¤X£üÓ€ïS$šÇö¹üÆÚçýBö¤–tòwè.ö"¯g®)½_¯¯t7 i3Ú*îF\\8fa˜ û&x/ ‘ˆ‰Çøb›Á1j½ÓýxrN´ÏŒí€¸iêf骜nNÎ0­[71—E)Ïïô ­ Åýß§€¬fîÃGÆáþÔtƒ>Þå!Ä[F¢>µ#yá7¬Bsp0߯;~,Fôiƒ›‘pãÎÛ#¤®FàXz.“À"ÀÝsíIÂë{«\†åAdÔ<Ø9Š.U'Á§æx0?¶FK&¿Yßx<«Ø0Ñ‹ ¶¸Wöb”ËÇôû«õÔÌ·*E´€ªƒNÊ&zñ|ý´¿%­O:[¥sÏ’:‘qjÆ¡xÏSé$ÈÊ%Û.ú|ÝâÊyýÑ’ú¦Â S…ÙMP¯v›{AÖ ¯ïr˧ٷ敾ÊïR ë_¥®5*Jjv½8OÁ` V»nö3HñFn]öû„ä†eIëêSÄkœz,uƒ!MùìâlŽŠÆyS*ªo·}dÀ©Õ>Þê°ÇM¼Çb˜e ½#´ÞY”^ûd1òlÛ¦ÊxœõãÙƒ*pОj\m§W¼}lÙÎû‰‡+ «ðα¦šû£ÃuRîFRY°÷-Ë,¥?Qܘbç_6ãÆ¶/6Ù£Xj"~Š ú;ÁÐí[TZòX‘>–€Ÿ+</Î|’’»å㨂)ö)g³ ÞÐykí^Ø»¾ÛwÕ9}~2ƒ¦ã‚zD(þã–7>nÅœVµr$âööÍ1Cåz šíÏ&•Åß­¶'yôæÊÜ_áÔpxäqCßö«ËU˜h“ïzZ'}T ÑÛ†ðŠÁw )ò†’¡N}í°Aå/«”é6kñ÷³"ê&OðßFטìýàÙ®G×ù\­Ë;q~Ð?Y<ÍòåªÅîÝ2°ÜØý^Ϫú„sa-ò°9ó«á 'þS®÷ôOc“‹´Ò!¤h½º‰hTã™®1‰Ü¢^O¼è"ü7>ûÖWò&¡Øhþ6#õÃÂ?‹¢÷üÞy¡ìŸk}„†ÍØsˆp}ŸÎì@‹ì‹\g'¤[ás“2‚#ØvÀ}JìÄ ?M9Tôò\ÎX`Ü\NÑkœZx¶þeصRè³ÇhûWCý§(ÚDÌ…g•‘­rES‹0ºÕŽ®o¡¹Ž¤ðqËeð>Ó×–ß-ÇMvßpo–ÌŸÎàïC×~ô^o«¬ž‚x¤í€²¼}eÒf5èýÀ^.9^˜µx•iØ3å,tlÌN×8½xsW’˜8Å%xBóµø:CÚáC.Äçò.Ò#_ñpU~&V·g…Á›×½ÒÀŸ¶<2Ú7‰EÙÓÛ¯“Ï?éЗKµ%NÊRQ‡¿o‘tOõ‘ÒÊ®ªq!w¯èÕÁBŸÛ¬"RÎ,‹ʾ¡«Úüò©mMàGªbÁ9§G£žºÎÛ(²UœZù´Wö‰²³¡,°IªÛ“»¾‡w’ØæšRò ôtX®˜ÕR/Å›†LmÎÍ}P÷àV±(ÕºDÕ>¡ânÂoØî€h™Yy‡ K‚64üZVÆ» ‘‚¯‹…ï[ÀÍSMqÙy¿´ÏŸµQü§ËÄ\¢‡°¿ª‰ "ZàÂÇ%·Ó—-ãxlÉ£,¯ž¸H\¿*íÄ–ÁˆP)aTND}âŒÐƒµ[Á'ÞaM£ŸÂó¥<´2ÌþX³-ú¤¾K˜pçl+>_/‡Ýw%³- Ç–‹Ã~(o”ÐH|ˆäF ~Ò(É~ô¤©½°DÅtžãÝ4·!+ùT·†ië<ˆX’ÝáêˆÄðI8ARŠ+~‡´ÆOe*b¿òq {åãêì_Ô/?߬†]ÂhQ9oßQÖ+Âv¿p¡†ÈÚ/€›° ×dÞº”ô$›­¿Ö„ETCp…ev²‚øô€â3êÖeÃ2ý5KäÀå%³¶{ºöwLÖ£—^ŽØþª9l2EóÛ‡ÙéOY*¡Ízkßø5z´y  ‚WV…Ìèn)6àOÛ×DÕ8xÂïCzšÀã*sIûú³›,Õ:Öw M+´à`V©iɱž0Õ÷‡4˜Ñ±I;a†Š$.5}:>%ü%á”Ø.i|"æ2bøÆL‘Aþ 6pNýQ»+”±çèl9v<Úñ 9ÜÅu­µ=å˜ñZ‘J)ª~¬þµG즩™1‰sÖ¥)ie®ì:¢S]­a†FÉa4 …„EƒÌ!Ѩís¹±ÏyHw·qã»ë¢¢N®q‚ÇŸ yÈÞD!¥¨Oæ|KmÁÔwO_X|‹1m²ÅÚ¯¦ÁÑ î?èš:]†Š”™`@ÿ݇•½ÂìîvPžü’uU²¦êØ“Éx»|“‹¹ºÊ¨où s>†ÌÔõù’nH»Å§ôºuŽ®óMýÃS·³M¤‘Ê5üø¢ŒíF6ͬẄ˜‰}³#.¯Ÿ ¶çÞ‰¨ošýsû[±ý5” k-ÀN¶.·7”xs§ä‡Ïc·½xà"a·§×£¸\æ£YŸ$§K﮽À¨õZ_NwŠBàÒì(ߨnÅÚ«FñÐCç)¿’ÑwH¯‰– fq9¹éy*Dd4ÔâÄÓ„X¬Ëà;6x}]h»/¯ü±‰A4l»Úz‡6‘T½ý~ûËA½ÁÝ'Ä×Bc‹Áq* ·†ULæ„,G8¡[Ó›79sÀ=v‹çsY!èFbÕ¨ ,$†3Æ1a ;Ž©,A'wv%óš_sóùZ1!›oÆB±5äôžÑÃç"Ö8–­Ü$­cc>¤l'h´÷!7¹ÜüÄ4n0P\h#²o'¼<´‚ȠƼv/sÒ1,76t>è=§¾<}¡IƒÊ«,‹ÿˆ8ºTô˜Dññ¥r¸4Ä9Y™¤‹VIa—‚{so ‘K<݇{ÊÛfu$ï'ðY»‹7ž-7ÜùC$ÿ§Ë•>ž=ÒÂÎImÁŸ[¯Üša»þä²…P[1æaÚ @gHïÇqv©_R†Hª×~¦ëðÃeªý´yߪÄo Yh õqÉxÃÍQ~hC~2øH´ôZ4 Ö“ÖDÝ ÕöHTêJeµþÎñë0>Ñù«Ú^.öeŽš@5Mo]®kõ4sÒcB¥0 Ù–{ö" @Ùç†ÎJžáíXO‡GæJÉ£ûâ[;ÙÜz÷¢zRl>;qg#Íìê/Ž-e?,U^Î>f@…‰HȪL¬—ÂY¬î>׫™#k.:_ãj !1? ¯‘ ?‡í†äçw)HˆXT2Ÿ+‘;’Ô°¯*ÕŸ —_ªC<ó’ÚÜ!±˜õL,¹Nv$YFÜËÄ+‚9ã»jèÈa ‘¶³®GNn힣&¾‡ýlU+oˆX²ò¬0b¥Ìhàïìæw¤šÒ”–¶Åä] asÉv—©;}¦ HÝaNÑ7ÅØÈ¶‹ÞèÎI°òÄYÒõ³?©°–‡ç¡f5ã‚·¦W4åëýjTèø›”-†Ø…Ò¯E»*ïéÆ/2Vie qÛÊï ƒk,ü§+¨,ôn&mqÀT9?ØÚÊTlÏEròæËì°ùiÞÉ„8>mm‹(ÿ*•yÒYÃ¥ö3²x CvR‡¿EP;• 9ÙÓcQ©õ¸Ë9ƒí›-¡8탃uÿR# Êó 7&Ô Irõø²?£¸Ÿ1=5 ÏþÄUÉ㢇݀WÏ!Äö¯µ¶žd½èA.2VÕ ¿vG·Úì'z˜µ ­û—nö§Ý ÃÚ‘!¢p½éE *’º:ϵq;9¶ óØY="È`7Ç$Y¡_ø?E ÒÃGc`Ìõ v¼+ì”Dι^å®8§¶’ªg“u˜‹«2çËËL¤H"‚Cy/”ʇM¿ñeòu¡¡ ?WƒÝÉ¡Èu}øØ•As;ìÞ€S8àÈQ¥X´ðˆ=èü&³4Ûu7¼7GýõXY¼ð)‹®PRÁ2ÇêFqmŠÖ•ŠqxÚñ²«‹‡,sUê]ˆ’Rã΂º4»{dĈ'÷¢IÅwœù*XŽyG-Ÿ58)ÓîÆRòG šn¡·:aJ+’ÃP‹I9;ª]òÝårñ´NXê×Õ9Îõ1ä=íÚ=¶ü¬°’ÑIþÐÞ‡ŒBC;Tš˜w%þog6¦Qªè—ßj9NR‹ëlx4°ÀŒ*˜¨]ß6â0Dú% è3³¾»–Ï?TáBýШPlœ~ûzÊ&ú¦¢SäK"ö@A.bS?%“V» ™>ó‘Wç~–¤\N thcº3A+¶Çê©ùc’d¶“Pà\€j§kÊI´ög¦þ•ÀbŠg°¯ ˆ$öCÕOÎö_ÄÞ”ÆBöà–¾õ·dß+Óèj;À Šè€i€g Ê’§@b׃IwMá¯[©e/iÛçx~5Ê6fl~—Rž×ÁÔ›ÃA;XÀ—J®¯šyoÛ®ÎÍí²–Û}@#¤òpA¢Ëº? å{¦Êc±Ý¢eô„GsaYª«i]ÞÆ0Xé [0ìõîZ\—ïPH&÷Òï]‡ŒÑÍ„ÌnáçYÙø>ÙÄ©¨]$¤´£öü9ÏNýo|b^7!Õ“òšDÉYfUúŒ‡êUá…Ž·D$;µIéÚÁOjZ¨;w 8ÚØÞŠð–:>ý˜œ²nˆl¥VžÂut+fc8v@d×2Ò;öVê¤jßšƒÓ¤¾np†aR[ß-Þ(¥œ‡IƒØÇÿ‡ë'"õ å! ùÊŽöÚ@öFZ:Cò„ÁJ™-˜ f§¸9¿ ‘a܃mÞQÀ6[–ˆAlÏ› exH!”¼Ô~¿Ò™˜ú êh6 ›Ü@ȵItŠ[4¡”&…ÎÑZ1œúˆƒ¹ñîÆšXüç¶4bÛ%^¶^g ³Ül”Õ'0"›•ªpYor¾'bèŒô…~"H‘3æoìr½ŠD¾ˆ”¤{|íK.Øñ¤ÈÎ$dî8t‹g€;rv/ˆ“Ô í™›ØÇ ¹%gCìÝf†tbymù6ìÅ×2—>¢v7c®9Åï¾l)³aøOáÃ&A“‰$S¦´Üö’ŠrŶža _A·‰Mo.:ĨF¨ênä¤ÌÏÈËÙR¢²Ÿ-–¾È «Ü}òœ®…þ¾‘œáùã9¤|)ð*XôVÒ¸¡¨G1 Ï©g„|ìæ¹©çÎÇráôýÕ·húÛ½f]l¢k‰>]ø ´·˜µ™Ì´t€«3 -Þî|†â€8w­ªç¼—ñpè*ä„LÖ)÷EsñÔGZΔT»;ñYÆì‰ VbϨDù[¼(ãb°«œâ9(Ë{¸¦øå;òýŸ¹'ƒÊµoì8zžëʺ²4Tšåg‡äiÅql6“l7ƒ©Ó¡lälÝÑÒëÒC|‹§Ä’Šù o)ûÉÊÉÊc\‰Í#i7Řr´±=Ïë…<øó^£Kñ?LišÜFéfè4 /Âl2L¡õ XâŒÑØä!û¿ž fë(<’™àûrÈîPrR†˜ÒŸyæþ¢(W¹fïP—üþ>cÏÈSÛ‹C ƒiG}Î:Æ ǪR=†\FêG0ìE’%÷þki“˜.KüxÜóDÒ„?Ü–WÀ°±R¯ès JéBí~‚Ûä÷´˳>FꮊGCÕhî¸H€ˆQZVôP“^ão†ÏúßÏeû˜ÑäJ_u‹&â¯Æ¿WÐ:/L0mLUÞ“¶'(êq1@õe6̘fÀ¯7µoC§?!Ïl0>Ë©"ýG×OŠÔº#¼fütâ>n‰°ÅÍìnZpÁ].?<×|æ«Ãø¦DüÌSw½zÀæ)jÆ|éRÊhÐÉœ\È  #á,]sú}6§‡rÊOÊewº§ë7¶“VK–£ïÆÛ#€Buy´ñ6ÐV6n´¸ˆk-‘© Ú(H1œ‹oð˜ Ò`Áds1{[B« í ¶–åíÏmòØœQ…5F4´ Ö%“u•Í©j&j¾÷ô˜“ø`Y’HÆ6 ±ÞtOÛÉ{#Θ(óuÙ­Â_÷{¤ºê[Éb¥!ð»8"Ûß>„Ñ?­Îô õ—–/˜”𡬘K:òB %¸¨Õ`ïçt£à´á»:I„{·øï !îíEѾ/sPQTßqûé¬-÷5§|½*À*Ä’“²˜u”¼ý@×FD1‡5«Ç¤pP·(ýya»5ä™™š~"í{‚¦¸¹€äw§BêAؾ;ÊÄ…ùBÞôw¼·6¶£|#‡jc~ð.Îàpeó¼Â$JvV| ¾õö)²Â‘ïæt}SHu¦Î*ú6‹1B{ÖpìÕº4Í»¨»Bæ…ˆdÆÌw5W˜TJ¸yš&§} ù%à€50Úïàû7¢ÊªOûÐ4Q÷Þx:"_¨¼™éTŽU¶'̽Á-lsþ¦éZó.˜“dhôdê6>‚–Å›à,±ŠïE¶ñ«ŽÍgÇ#×êÉcó˜%ßÈh$§/,Á•?%þl⋬[Uþ*öc3?f½©Ô´ÖJ‚†Eµ’1·0ø ÙQìÔ­v-$VØô“ü$<ÛBß:zäÃ@ÎÑðw×&tÄKÞîÄ&³8¡€Ä÷;ù­‚'ò˜JE(&H*çìáhÐ\ËÅò#x;[W©E]—…‘êõds“ñVÙué*•rÅzž8E%b¤«ËLjàä)Maˆ…¤z°§<ÙJŽeNª‡ÊG<×КXo¶ k%,W¸ÂWp’†fªÚÃecöæ'Œ $æÚá{'™âÃn"€G¥m}Ž}6”x¶f¹¹L¥¡^^û+mk„‰ípÀ%Ô†ë5é>ª¸Øêk­{·ÄÄ`F%`[rBΧ©·ˆÕ[‹œC° $"rÒ«Ÿ›ìWX1g÷vR–„¥Pç‡Æ 9â†ÆÜN´45çû状J¼C„ñ¨¢ƒq.Qò WÏQEù¸ic;~›VÄlˆÜíQ]¸Õª{×íl`§¶°„(áY<¡¬13í\â­†ö^é¹ë8FâmnTÌv× „Åïiî43¬¥Åü'±êþ3"v+‹i½Þ£>mª ¡àñß 8ÈTú—kÏ1šnY?¹\³ÅóÒ…´y$/…ElNà›ß…ºð1M{«L¬ê:alhÊåAùéׇÚ#b¨=rf%‰«;9¨À`‰ÐùšÉ”ùñ!à˜ÒÔ.¢TÜ^Cp}^ÙÕɆÿ<ùL;‘ª˜Ü ‰áò½!Îð 0w3Á«²Q–·p¼s™5D»;õ¡s7A3ØfSÌÔÅØ¥ˆ$Ž1$×Zš½ñ{çøGˆŒ:ƒ”¹œ#.ÿ)[.tL犒™éÊ­½žWX”~p„ûo²wÞÜfAO;ûÂÍR–YC»I:…bK/.^c­#ÖEê“u“`ïn Û"’¡ Vôn.Wß  :튃ýÈT·r»TØqÊ!uÁ3ÍWU¦È™³ñC1X°ö…¢çŸç…Rib ˜`5À>›,®§g:¤²xÇÌD›ñ\IÍ7¯È/ågÐ5‘Ÿ® ÓLS ã”R¥"kͽUìÖO Í(ðÖ"†Rz='ôôKÅ÷Ü´\SL‡-üž‘§ãã’âšæñ=ÜÂsâ’¾hÏíE|'&2¥«xúÔo•’ -˜{ºëôæøÇÌ;lRìlËó¦o¼¯Ò‡d=98ú‚-ôÔ¤&xZŽ`‡<Ãü¤éŒ/ò…ÊcåÙÊsNð˜äøÈ&ëopâ:wÒQׂâ“Qûa–ùãUFåk[Æ£ 쌒°4Hë’Íx╊ñ¹ø}Á¡G4]žˆöèA5bïB1R¨—íÍyÄGk¿¹Ÿ@Ö3{¬÷Û"h®î— XÑ.;ÝÝ'ŸæÃª¾uS`>E$ÑHŸgzÙº§Ë}Ù³ý.“©ž Ñ dZæN£“и #WHC«HTÌ–ãHFÚƒ´ y>³á|ï_›,›gøþù›s† ܘ3ù˜M—OüU™ÍmõÆlåäÀëK1 T­NSñHÛçKÛHWJÃÌC+½2‰T—;éhq®—Eãy÷Ô¹]—|B#®‡]d¤ê©Ü¥ Ü+ÁJÍœgWŽ+@ ñ‹†šßšruÁÝ8îÒE–>˜"¼aÚìUŽV"ÂÎ]gbQÖò x¤eÏæðzì~¹äö’aæèm¶rŽfs’É÷‚T$ ¼dœo ÖŠ8à”‡jt~¬Öf± r÷ÝŸ:ÅüáÕÁaµ¾×ûIœ B6™K¾ÙSÆã®õ®î,gß?£Ú>É::rVkKí2ÿsüPÓd H$Ò^¶4ýõ‹~DÅðéƒkuLsÍòø‘ÓèOŸÎ;— š{~6¥Y8Q…Ö÷Cc&› Zf2Ë[}¹ð#APgá°SÌÉ;QmßÝ2aVEñ±n–Î]YÈÊl«ÂròìT±ö…ãûÉ7…’()%®éŒõñÅbÎÛ\8,L%å«ìÖÄÙoÐ~|6{ ³LOˆ T;ìÆ 5k€žÒrÀÜ †[^ÅÙ@$jñ>ЍxüQ„A5Û ³¡çå9šÑF‡yªïwÏdxVmw”c¾Ð6üUýƒLuƈ¿„"Ç©5<š:OJc:-[ù-üh ELTÖ©VRh¬ØZŠñ®-üÞz«UÑ Y½Ö¦jýõš]*^X½…R'¤ %ùБ…™†šOà¶"ωy>còå¼4VtJNjíÐú@Ò¨Žcí¸ Y8g™Éó3Ž.Ê[ižk±:‚ÛBl—i«íÙ'voÊÎíêµ(¦Ù±‚~+[l•*‘v—še˜Çhg¾èŸ¸ƒÊÙðZChò0ì ‹pLöŸL£W­&ðöÙá%gnñ­:`CÎø0My4(U¯*(0 ïW¿!ÙÃy"d±±™ ­pIAŠkÅ<¦G¶¼P鿉[n¹¤ ô—¼'>ÝȹÞ’kØ¥zR*7gÿ™¶öPä^0íáL瓹äß뀡槩xj.µIBàS`0¢ë5LrÄ>9&‰—cÜÛ;“¦èÚû@u¡Lm›0)ÒÐòÁsÙ†ÆéxúùMë+ó'‡[{èëû/®c{Wœóf{Ô?íåóü`‹:ÄV YÏŒÁÛD:Z¿Ü…gÆÄÚ¾K4@-ˆÕœ$ÍjwÓ%$.â͉kɺXDüÔgiYLéT;! Ååb(_ëˆZ£(wl¬çÆWæ(£z‹3¸4Úq³Sӣߒ¾u—«AÕ ×‡¦¯€ÞÆ+U›GˇZâ mÛ¨AÂ(§ÿ™á-Rn朂v€ßC/-÷M—ЀPó‡;ü—kWNòŠÑG˜9?S†‹}´çÄX:g7’·w½¶£[d¼Ež¦GñRz® ­—!í¾ô“Ô°QË‘¹jlò3]JàdÍzŒÉèxYØÃtyOGå™7eN‚’#~ÐäQ ÷’<)ùÛ’¢ªh‚ œL7|Çêý*<=‹@!ûjZp 8à€¤°+ȆÍ#I¬‹m¸Kï=–‹ßrj \ð^ä‹Wœôè6©HV”ÿì„wn)´42á yn@Ö¯KˆŽMÂ…R¸fº>g}Ðò™ëýNfsSÒ5*c0±µ¬ jŽ¥-ÿINÈB{4ó„Û3ò Z厖"²ª¦—8_$ôí¬nzrÚá ?µ>“ËsªXY{Ø5Y< Ü="¶Xñ„¿g-`ûEZÝžð*öEvãT4ÅϤŒø 9ò1IiüÊh™Wë£yN÷ÞaMj]Ô‚nþ‰ˆmC…Á‘|½ÓÑ " ¢Œ¥š°#Ú!÷°Yo>C*™™ÉfZIabÍj5ÏÆÞJS·“?7*%%‚Þ™E£ÏB<~$8ŸÂ¡°óiï !@A]¥V¸6ç> al^3Î…wüö êÙšÁ×8õâËx=• äO¨·Q3lvò×´—¥äBû”4CZäcørfÞ•Ñ_ÆäÉÚß_@TéNžeÚ³}6dò=—@…û˜«3¸}SBÁýpˆ9Èï~£»sÎÓG¿úS€òc ªÐ?)J6wʇÑFÏÄÙÕßûûŇâÓ“ð)ö¢öLÙMuŽÅk"ŽLEÔ³íùp# Œ®bÓ¢¥€”¨YY/3ýÇÛlKAxA+µQRŸ]„úQ£Ÿ0E ŒGPòÐXÀ‘¢Š¹7Nk\£Q‚} ó@šòŒCœ²z6ÄZ¨µS‹N¹3&RñÌ|–ZózçN‘+0[a®^åññ—j.mZ›3¼þ˜(­Âªu°|ë @©KåIÔsR*ìüØ)¿ #þ÷;þaýh8* Ë áFq =îÔ2Î…'¢Ž9HqÁ9 hR0i©1”°å¨0!JòAáØòó8Ø’®0°é½-S%9ñ•“Qz`¡B¨ž°¨ÑšyfC[p‰æí_T-L!ìÞ¹|ô‹^àr3ž²Z;îIVæÔ*#ëÿXŒ]j›KЛóuúˆJ;q¥½V‰U·tmicå´ózö2Pæ¸w«·ØxXÑØm£‚áÂà ¯õjÆæµ·ÈÅ C÷s{↢h÷ÂÁ8˜PžÜáù²+·øÉzo÷û\RBcíòÃež)ý³åQ½ÂjŒU©ÁçR_Þõ“Ìhj o ¶k',n?«3¹Æ;;‰ãí×qÕbÏWŠê-úcñ2‰çʹ›T´½4i„M½k*Gyu­ÙÆø´lmf;3¦84 'Ø:Ïö#ÑQµW¸’nßÀ)H’£&\Šå%Jh«ê»eÖ1¶Òú2ôý—Ôˆ‘é]ޤ*Aú× VÉ‚ì.:Ï«þ¤p5µ·² Ø0Ûía-bVf@ð1‚Ïñ:Íç@Óv­Z7Â.Î|OŸüÉ4Ògc)(ì2£pH9¢Œ£ùs—牊N, ÃJðÀItËõݳc§²ç\žo° —¨œî>гâJ1ß¾É-ƒ#WÕâGN\èÓ÷ ,©.»#ã&W½ïJÊR—=xˆ-ëFZ-—)ü°Ý›P ·QÈ–ý$uù¦b©Å½ÙæÈZDÇ m$ß\©’à~:"DJì¾åæh¨ƒ_i~˜ŒãÜÁ‰O‚?<#BËmb×ÅÇb’%ú~ȯ_ó`÷™¦˜sPïzY7¾Ã§ƒ•@§c©Î—šq冲!X ^0 Ýhü¼S™hFo· #TÎÚ‚P[•øe;0ü»˜„8à¹;u÷уa$ã¸pŠ^_Ðtñíå§œgƒ$˜µŠX ¼³|i”ߤœ~3ÐÆ^Cݧù¼ÄŽ6´¦aÛ¾L¿Ðݯ ÙKðÐÎ7»¥œ=­1à žLŠä³ùörHkÐ9¶tjz çøz.3ìsøm,†':ç«Ìà®fе\ÆéLq¿OŸîU ÐÖ›V–ýŽEÞJ-w«Úñ'z‰6å¸'¼Ðùñ1˜_Sk?±mË 3/ý°ì^Ž÷}¤[F_+ô»YPÀÈrüCh)ÂÊê_•.Ÿ¡¡mêŸå`ˆµ³,8iðô–°+1ôq¹ì\…öF¥¢ Iø›e§Þ|‚T©Æš›@yË V¿Ç7f=û$!xZlJ6Ú†ÕÒŸþSè·»Žz挪®œúËäOíò q¶çõƒ Ae! .=/¡êoê9 òù.e¹û3«ß¼aYçki>oªL£YVĽ{@zˆZ˜(Þ|ÿ2ü«üÑAгÃoZHìãÁg†ŽéÞHö±r´ïOn¯‰Â/… zlÓcH.×IË^ (,²Â n mó¢]ȶ£Bæ.÷Åë:ðmßán¾¼šÍdÌuÆ´ÖmÀ}‹r!Ù»yzò—ÿAÚbŸþ^Ê.ckûµ—_:(oA¾âY,srÀìí¢m~Œ~ÜU±€^v!Uv _ÝœyÓÒ3Ñ›žAÓAñG$5ùC<$÷¯—1)GínÃ÷t.‘üêÒkö6„» ö$èµ$àAŠLç½DŸXƒ=:_›JªÊcBLÃEZä­¿8¡®aÄ™lŒŸ¾‡I_ wÇm²ñà™?ç7È Ó¦‚÷T÷lä}ûž:Û9ì[Êk Û•–x¬¹±V(míŠë1 ¬mv’€ï[âQÿ ¿9ž²õHö÷äçÁÕDñÇê[\±ïèÛ¹V¾$¨Ï~ŸQȾ@±è >ÎëÅ‹WaÓÏüàDƒ®w5rÒÌAÒ¯–~•Ç”¤'“ùœÉÑj6ãÃÍ—d»å´3'5ÉÃûQ7Ê–qÉë±Á†îŠêgÙ¶oìÇi¤NnEq®GÀHÚAnÍ­ÈnÓŠ}‰àó{Ö<h :²POmtf·˜MykäŠsŽfîd¡ò¼ ïsá4E¶µ ÉS‡8±ö‰ä\6ʵîK_vEž€S{Å)ôSŸù½ðV† ,ª~ÍÒÊ/ᄄža¦‡/³wE¦¼2Fýй¾:äer”ÏKˆMÌ*mJvµ\$p9:49¤pcr!I$hÙ=`U´þ¨R–‚kÝf‹t7ˆëÙ/c¶¾ÃŸ¢7w6ãO^\ äGøñ–‡ËÚc3 Ié¿ó*2:oHÖ.Ω.Ù*•¯°ÂËkžKa£nV0¾j%)ž$¶D»ï› «c¾ÚÙ/4xí´ûr:öN¢ØÔu¶–¯I{‹¸as !Ø/*È¥nSŸ ìs–K÷aoùg¶s“Ë_Ü&pš~‚UW¦ñ ±ñoÖž{B$Òä”é¿xµåÌ„÷.œ¾«?bUM“@ƒYðDê—ѤaÄökG mÿÆñPÒÜ)YR@ ˆmÿq`‹ÔžÞ´¯ÉÕÔìSŸ{ÕÀéJ¶ävƒrm z¶9Áž•ù Uò'Ø)@œ méêQ ¡$Zã,e‡¬§ÄL||9"£•ìÇgÔœƒ›ÀIŒhcB×A¡39*Íêžg+樗è`gïvÙÀ‘ƒemWT{¨·2Jmí³G ­=j‰fI‚»CÚŠHŸN‘ò'ɰšNÁšúš&/ÿŸ<Ãä¢! öƒúáÊæ G†ò&…ä§cƒ'žè6Éj3Ö¶‚‘åäëDÍ?Üvu“Õ)~—°]ÜÑâN´ü/Ñs…IÃy1°M.½F"Crèìñ™âÑú‘ÆÂ/¹ÅEư°sYwÞÉ6p!D]ã°–Ú-©ñ/4̘óÇmÌ-†ñì<:©ùŸ­{škùsŸãuxa¨¥AójèD ´%¬Ë†ì­†ÂOyj°ˆ[“÷™°a'ýQc /99ãão²”‡î‘)ýù‹UŒ%»Ñôu•á÷ç×páÙ¾íâ&œ¶}ˆ›\Ö,(¶'é]±ˆ¥ós7(eZš 3,™I†|ß• ¯Gïql–>ª¤Ôj}»Â¯>WhISKx{½]ðËAƒÑî_â“1•œþþÚ\÷Ñm ºÓS ðZ‰æTæÕÄé¤pâ/8sï7½±ÑDœä«"ýâB’Ý¢]rõ (üªB¥°ØF4˜NgŸh˜’áÙ›Tõ«lžtL¯äÇ2#«6Ë*F·©'jýSFV£[ˆš¶‹GöÉøì«SéKyŽ¡tnÉZ­lé&l‡™ ý°÷‰»a„™Ñ3{[§ï¥ìº Z_³+ gFº­pŠ %èži”Á]#8¡ ~YŽYì1¼Žiä'5üi^”ø¶å cü'ŸyÌ¥E9'ìÂë›æ>"×Þ’?¸ _úÎÁ@uÔFÔà üŽ˜–mWõú$öx!ÍÀ7õGm³ DZ¢ & ÝgäS V$ñ[~§;7ß Þ+’‰jˆ¬½£ò,±ªæwËûz ¦XJ D1ã/yS›ûïâéÕe#œ¿@`×èu&Ƴ؅R`êaß{Æ÷ w ‹Ÿ²ø$V`U-Ó£,;¤v§ JªœîÚü¹YÞŒ»rÓW`ÿ 6~‚jäGÝš®6·%\ÆÒÜÒ#–[°B“¦Š=øhDÉ%% [Crep¨h} …j|8Öÿt‚â@5«øFªƒ­8÷ÂzɃ–ú ©qÁ<=§Þœ+æ ^tÆY(]VÚ7lÊKA/·â ùÙ~¹{ÞHÒ¬²õà Žmbûü  › ÉÌw½4K/…Åò…«¬'ÄÓ‹™nYw­ì¢¢£8§w+°ä@t—1‰8sD“ï6!Ö²2°-ÝG+‰ètíh²¼ç&Xv±óOpO“;™DŠSbªð$Ÿ>ÔÑJñtV¯'Ñâ }¼Ñ’ÀÐçtüá%.—ß~š„Ž×•iXæãaûoQJ·¬‡ÙÅZMâjðM‰Þ¥lë…å±EÎ~m`UI[¸*Jë7™V?¿»jr°›i†žÆ±.O~òm=‡¨ E±Š°œïcúŒY_3&½I“Ê'p#¯·zýê$Lº\B¡·÷2@æþ U›Y'xnÍ3^¨ ž½íhëÏóBý²~äZZÀFÆT³ÜEùòƒBô}÷òpÀÊñÛô}Ü„UãœB(¾Ý×Îâ ”¬õîs émcAqËžØê8‡N0Q_8¹Ó›¬à“ṿp‚¢ã».áÓš¨(.×Rr$µÏH2ºdʽ‡&-¢0?@Ogƒ¸Å5àñU“Êq=„?4‡YK% rÝ>{â‹å§šµ’ìÜU7öÞÑgxg`Îìþˆ‚AŒ¸;æÖ{–£buy«‡T3â¿ë$ˆ-¥zߎ'Ø­8 †"#jò&KÎSÜk„š7ï­úò,Ðú¨GÐÆUë\.Ç«hMž™!Ò«…mZ7Õbñ4/TVª4Pô¤VFΕËþáU¾½¤¾Ž±ÊZÉ^mãm™?Q˜ÞØ—n4 / ›\0³ÈðÁÿ$Üÿžd L-Œµ>€åˆ¿•9Ò~ãnxíA¥²ÌL9á-7)›Æ«uÕ:à³]Àýઋ¸õYônWo|ìx©Û© D oÜèB÷àÛ³úïòqeß i4þ¦ˆä5mºëg~›j[\¶ëM.KÞOV zÜCÐU›U™PèŸ[³5þP¹Œ®nɼ¸þ’6úú$)÷-™wNŒ¶“ ¬úáGòükò|‡ÛcA¢“Ôª^ 0?ªX­0@w!²˜Ì˜»×¶Šsl–è¼$DÅú‹à‡IC_Aœ‰8þ„qOy èQºLtƒ€R ž\ml’¡±"̘=íõŸøÕ}ÑüÃá¯+Êêh‘4eÙî»/ @(sä:Í›ŽÆnµV¹a–ZcY ª`·ÎÉÈÐw£w9uÀ°pì*ߣx0xžØ¥šló\]î=kHvAß¹Kì–~dö†’À„˜<ŸAÁ—[½—®Ù˜ÛœxV‡2Uãžag²ú{$rÄ}øZ‡Hsß «&¡šK ·¬TZDò50ãaÁTxW’(›z_/mÙÁ/$Þ ÖCµ¦ Èæ8‹¼nù³í“ÙÆ©nniz³·¢ïráyYâAAxvÕê­ÓPLíÜãIwUomÂ à ˜ˆ»Âí.O‘7ž€h‚#n-‹V®?Œ¸¤©šˆPîöRÅ|öx1†e2{!ê”—BYAì:+ékÌ$ fL ^FKîg"©¿<]'T*`3 tÒÞÐ 1CÑ*°ýipO™ù—âdÔ;MÔŠŒ„€Ù]‚Ú v@Ošwzí7=ÈÐøÚ=Ü;Ž\º¤%Ê'ʘVÝòéoÊ…Ÿ¦áRö"ö8ö›\@‘]¥§bKˆÔ>ßïíÔäR¸çêš>¹uZ<­­_4ÙÐ[ƒxb›Ð™^ìeý]׬Ä(ÒýÆ%W1öãj£‘±MÉ,»”ßJ–+L„ŠºUÝsúìù ðBÙ~–LÁðD4¾&º•20¡7Þfâ¶òhÁüáÈ5Nágúålˆ „Ýa´Ó®¿» ô•u—ž˜G˜ÀÇwlŠ'QeÙv½dIa÷¶ ]ÆZÃiø›Ô7š×.M·{‹=öniÚ»¢»Î\=ž¨Ûµ~â ÐóªKÈT 8s±Íû ®t͏ ¢é€•0šÂ˜…r‡‡uÝGËKˆÓ’!-á¬_zÌò£Èö @ëY€‘"BüúÍÇ/¤S¬¦¨ã,öÐýxü“ýß³ïÃU‡Á¥ÃSïXˆ&¡€…˜Í ‚ã¤Ñ+Ïrü•äxÇ87ƒ!ýqËÉpñâÜðE6Ù~=òì×øTàLiãe€j‡s*¡œñ"ki—_bvýËK–†³U½«ï‘¦Í“R¤Vª(E(¿Ÿ”3½Û¦4èÏþã^pvì3ËK-w±OZïÇ`KvPŸ{Ûe~üŽõÊ-RÂîH¢ ]Ž~õíóÉ/é_ Ve‰åñs¹‰Ô˜:ìÔ‚½Ã …p1Lïbá=5(óIo4)˜Ñ9/CùpÌ>ow c•m?ÔÇxS<Vc /¡b88”‚G9ÒµGБc‘¼ö‰z÷3®$ë¿q­±)…À>ìü˜rY® À]I»}3[[<ár‘£;‰Â6¾olS5{¿T…È&K/ë1xk27ÄUâÉêæûБ­öý3cIfXp!ˆHy½Wz!Yc” €^ þÝ0+4dÄíé†Ð ?h`°tÏÂý‰®q‹/ÕìØ±a‰1ã—cÿ ÝÝ n~@jhn¾w$1ÊI¶c)ÊRZÑÍ'†ì@µó&a²‰.:^‚†¯óq¶/3>ZØO„cœI{\;§Ö‘ˆ{¨õAÒ dœba‘ñ7Îò%·«#xïÌw3"q\(Í=Bü•G©1 b¾]R)0o09öþ4aÁ£ ¹Bÿ¼×[ù_Â:‚…#û ól£·¥ä«{h¹¸èP‘£”@žD½…<½rƒ¡Š ²[˜À<-ô¤ŸdZ“tã$§UX»wâ¾5³»¦ØQ6²'õ\qô Å9Ó g¸w v–µêÍZ-¥ÕÊÚÁ/–D{†´+ØÕÍ]œÕóñ•„6{ ä<…µ\¶°®é}r™„æuË GÀš;"ÁR­ 3 vë>í1E'¨'ÐŒ˜¬4opxíD+þÑJÅÇæ/hž4`&¸ë¡-òÂa‡S¸Ö`Ö«Ðq¯ÉîqM·¢Ä†ëèÝ‘·`áÙ,VcnÄu2Î*"G9û[–»Zî¶ØÐ¢ÒË%8‚[2r¦W3Tüv‰ø¶recñ"`us×ÚÜrß²ÿ®¸…Œ³þ=kEt´ìO#•—·9"eƒ0g—ÁÄmcÖø©·1ÑtÞ±¬«CeÊ_SèÝxáor0I)S’Ô¼¤3…ÕyT\7µf` þ\T|-Á/kŸŠÙF¾±3â(¿ô§b£"°7mA @¡Ðí%*8õÉbuz _±ËStÐU3hG‹}hòç8b^^}÷æêþÕ|2bß³;§Ì±ßÁ'‰ñ±K¨òlH?ªÔåF¶Ü"å®Áî\üŽâǯ:Ã+r;Äó×$ô62-?ˆ/žEßíê±kUB?dû›nŸs\±8ŠÄaÐÞXΚçNê¥k9CÖ·­ïOÞ(I2¦ŽcT¨·É»,/µ¯RÁ1u`¦ÎÁ]'0©†4~sÅÇ(¬Zä †’0GdGüŽ(¬ë‚G‡!Y#³Î_ o(˨ƒ0I9sôAEáE¬!P1å™oÒS®×ôMœkàG;è´yêå/™H4º½ºRÏÞÀašN›s?»®~Õb0ûÆ#B ¬±ðª/$É~®“ÜýO=PcÒʤ¦. = ³¶̦<à#9eê*0QÍ£®¾rжî:—Ϊ†>J -â"óFÉw윑NŠEÈÊ¡Ù&°‰Ö aÙÙ sç!ä‘ß%tŠkŠÄ5¹1z—s­¨•çŸ%«íO=ãTqUPØ7Ç &rÊD›üøŠëHž_$u„¹=ð;Ø®†6šS³¼rËžªYA®Âí<4›‚½û‚úÔ­(YÚÙ:°®à–µc¦àw˜¤§}¡Žmêr±¬gFýÏ;÷‘ØoÈûù…tø i¦ ×HkG‚‰û ·U|¤Ò7h/Üú‹8üÛ}‹˜‚¬ü®—Å'¼ŠŠ©f…}å²°¤È ÐÓ^ÜÿFFz÷À_€3G‚Uôsƒ1ZEIx­¯ç¾’  ˜”|cBjìÜý”`Ôè 7fy±íŸ4*_Pv¸4ab<¶æ'Z^VØ`žýj‰¼w(xï'5èèbÌò„I9MØ“Æôî(òZÖj¤ËÎG “mõëFyxzÉ^0`µT&®#TÝKŒV}ÚY×F›-î† —BÊÑ$Œ¨} ÿ†À‹A`ͧJŸZ¯Þ½~.Ô¸}»K@Ä LTR–YÿÈjÍç'Wº¤ å§ñåßû`vo ²E±´FqÒÈ@y8cçÃ^Ñ/Á)vaûšqþM#Ò l^—Y®½‰ü­ìƒÓ®ŠÂ¥ ?üh–œÁ]m™c#t’åÛ´ÃW È­!/RmÌú;äÇu÷ˆWHà ŠÅþöÄ}Å®—ŠúïpY(P‚?{íM¶Úwô’H°£ÂPM ëBc.„ÎXgºF×¼§!Øjöãšú­€-ôïlfÈ—£R…p× H?غ¹ÈŒöÀËÖ0¹ð3…–`~ÓÜáKËmжåšV{Ü·$ÞwÊ©òò<Ëš¬ó :û®&Ê_Âs(§¨Ìô¬rXõáF•ež°éPЩdBå‚9T¬ -ËU¨¨{Ìmጴ¿¥À·M‚{Î/rÛŠC ;£³³Öôñô  K{¯¦ÿ+otŸE{µ$¾³W¡n¿=(÷…y7êȈ@ìe|Š€-fÅú; € ÑÕû¹?vu·=F”åbkþÍOÓá åæÌÐx×î ¼¼Î "#¬X¨ÃndIúþ7†:+ò¸lùöÐcy—Ÿ5ºåa—k¶Àù°x‘‘çiAp8Ö¿½­Åãw!ÉIë‰Sø’"Òù4ÉdªX&SíH-ÑfDÆáû‹¼~”øÓʧ;f3–·fgÈþ»_TÊÏô‘)N?5È•ºgW¢…˜TcѦo§ò=®k‰Š»ÇØôÅoLÓ‡>NL âDM~îNºÉçöÖ-¥Ì¶{‹È?x­ƒáÌó_¤ñíLîá‘ÜZrNS69‰6 ºa€g»7lÜÜ‚LêÜ›œ:,KâþÔ£…ÀÖ"Gñ¥Î†ÊCy—h𨰼´|tU¿Œä36U™BPåä§)B ïc ù.àÒd1Ø>Ÿ·’Ÿ¦üîj.D5Ü`úš¤õ„ö“GXq°ßd±DT}÷Ù·ùêˆq‡¡Ú…ŒJØ(0g 5¥*.ž¶²3yÑ!NÈO…#nvÆc \Nh]×ÇKÔ§1&\ï/ãËWž;h‚ózÁîs¸ó(¬2>UûÖž¾ííeîÃÅ|) :š{?ᣮ©Vü¹+hÿ¦“f­ä ^?þƒ:d¦,G¢ÖB°Ý´õñ‰ø&;MïïB!ž•©Tƒ3‰ðj8ÍI@ ¬ÌwG÷¢SÖ¹1B´Ž¶ çÐ俉ò£RÜÓ¿B)Ç’¯Àé¹›uåÂÝO±ï¯”ÑúªØošÌ”ÀÈòà*e¾ßè¾%X­QžÀ¹ l»«ÞŒNòt#Û& þÃÙeäj ŽvBQ|Ý.Ý‚TɧrÉMü¨ñÖ”éGÃÂ.(»10e‘xÆR% uÀɀɔš‘ä êk–$ÔQM 5Ñ?ؤ¿?FÓZ©Ä½/¯ëmB‡øÒ89ÍÞ,#|Yœ]ÌYQ%‘/Ô­7-Xjþá æ,ð¹Ýâè~t*LÖP/iÈ,z %h„»ú-JÚ—WÍ»µYIsd5"ÓuÊ­çž|›šç=F +ØÒN;ƒ®!y}Ðó {hS¼ßÊ“Îj[¨«9BÙ&î/k$U÷ükh:vM퓚…j\­uþê•Tãf†ét#Åì(D4]½Ë .Q.¦É‰Û 8Œ[µÑ¦£ÖB endstream endobj 101 0 obj << /Type /FontDescriptor /FontName /YEERUZ+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/K/L/Lambda/M/N/O/R/S/T/V/W/a/b/bracketleft/bracketright/c/circumflex/colon/comma/d/e/eight/emdash/endash/equal/exclam/f/ff/ffi/fi/five/fl/four/g/h/hyphen/i/j/k/l/m/macron/n/nine/o/one/p/parenleft/parenright/period/plus/q/quotedblleft/quotedblright/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 100 0 R >> endobj 102 0 obj << /Length1 1672 /Length2 9095 /Length3 0 /Length 10176 /Filter /FlateDecode >> stream xÚ´TÚ.\¼¸»…RÜ‚»»» !@ Á¡Å­8EŠK)w§”â”–âVŠ—+PäÑsÎ=çÜûÿk½·²V’oæ›™ýÍžÙ,Lz†<òönv78’‡Ÿ(PÔ6à‚¼@  ‹ ƒüeÆa1 <¡np‰òÞ¦BÞó´Ýà /€_À/"Á/*€@ñÿÝ%7Ô Í ÐpƒCP¤Àâ AxCì¿t@®?•ñâ°Œœ žÚ Ý> po€AÁ¸ç}„܂ܪktÝ!ð?ÉZ¸õÀÏËÿwº¿¢'‚ÂÿÁn®î ¸îp€Â ]-^¤/’‚Ûÿ&‚`žn÷ñ o²»'üqr@E^ºø—ê °ÞÏ?øûó÷?«ûñ²wƒÃüþ¡ÿq¿|êÊ*Êšæ\*þÛ§ àæ àðóÄÅÄ¢Â@@Ðgù[ÿ´ÿaÕAÿ:ðŸ„êp7€øŸî{÷ÞMû_Ãøï :n÷£ °ÿ3ùO€Â@ðýÿÿóüÿòÿ7ö¿³üß&ÿ¤âƒýáfÿÃÿÿqƒ\¡0¿¿÷“ì…¼ß m·ûÝ€ÿ/Õòç&kCì¡^®ÿëUG‚î·Cîû»PO¨/Ä^Š;ý9Bÿ¹…ûô0(¢çæ ýýÖxøÀÿñÝïØåþ=ñ¼¿«?\ûuúï’Êp°›ýï½€ð~¼„…ü÷ jñýc²|¼p7ä}à^^ÀÁ óûFE>•ߦ?‘(€Oû$àÓý‰ øŒþFâ÷LÐ?èž þýÖÁgÿ/Èàƒü Þ'rúðAÿü÷±ðÁûX·¿¡Ð}û÷õ,às¿"·Uã¿Oø¼§ ÿ†÷mäCú¸ýË}¯Ãë(p_ÎïŸôb>âOúµì…@Ü?G,ÆýüÿñöA ¾0ÎüŒX2ܹ>¼ó¢VžÖ‡gë“ô$Ë–i&OÀ<¢ËëŠ+£&'tq.Ÿ6üžèˆ2û™ÜãMÀ~[#VT{ª~ǯÀk›dƒñ­œ¹Ïc¯öåúé±éxŒä¾Þxš„¸ µ¡tk°xx‰è‘^øô©ú6ô—/ŽFÎlé­ÑĽ.Ÿà‰7Ž{R:ÅRh—;MõÉCÿ“ä»/áÔÙù$IþØ£F2NÐA¼`I€Åª@Âå´ÿR¥‘€ç;êÇÔTôhg$£ã¬ ;锳e%+ˆÑŒiŠ®k®/øa;ìþ[:ˆÝckV–1 vj*>42õçq+íj)®¥L)˜B¹ß¿h¬øª‹¶Ùyrë¦@ïta¢ ¸â[=Wù þ;XìäõfœQNäi Œð^á(z”üV‚èâ-ÇÁëxÊå'ƒà¥‘%²¦Vq[eú¶®àuÇœzŸÝ%#¦á/¾CêS?4ü'¢½w78adßOÁÙnÝ_?zÚ»R[£¤±45…JïÔ¶§å+árÑ>§hÈ¥÷:o£Å ‰gNËÉ*hlÒô±ªßÚÇåÈ®Ùéè÷Á¹ˆîßÛ +7.¿“¯©*Ë!~…­†•˜pÙ‚µöÑ‹Bõóš„Ò2IÔšFtÂ-{dщiÕû“oOެŠy€R¼åކ „x:°¤yºIZG“… ´Öâ½d·¥šíœßTpBÃïzc QÁå3ÊÜ+Ôy¿<:Q)eºœÖ¾ûéõúý·lf½»‡Ïv#b:¡ÔxÃ’Ô~ÃzgÈVô"T ™»º¹/îÛ ü!†&µ(A‰ƒ¯ß-—ƒNìN}³ÏžI¹bN øŽæ§_E’G8 ’]?n½ÙRn‹™x&N‘%Ì» Mã ‹{«á[ƒ'ÑSüõËØ‘yÁö˜ãà–øÓšÏ»È]˜ñ¹?S¹[Bö—,§ãÊE`Òôù8{‘ò©jÏ@·F1Ü%û©ÿäp!¶¿X÷\¦wyC#/®A>Ÿ#ïSfE”e6fG%Y½qL¯ô †ßQ<•ð¹û‹ï­ùÆ?[IYíe¤I¡lé<«ÙoêÅHÓÆ‘êwò¾œ£Åט¾82 Q©> «4è1yph¼| Ë{¯=_9.àHª»³ÈKaÔ'šJuÕì3–úŒC¶ ­DñáJAº‰™¨õºKvÞ›3 ÉJ0>ŽÎ»œ'øóñªtï^¿ZÌN™êeˆ6~uþX½R´C0\›z¶–*2á”%‚a‰-0¿Òô9?‰ÑbÕÛø4kÖ©ÈÂa2ØJàú½}¶"—ž„D𵄳]Iqª‘Ðë™nC_[2±Va&Ñmõ?~8ýÓ•UƨkÌ¥víá.Â+¢Ö\NÛâõÉÔ–E–†P_ŒþÔGUБ>øV²×¾³õ lÄ û#M›À ^ªñÚÉ™×;‘õLÅ™ç,4ýæFgl§qƒ4ÎòÝføéµËáAZ·‰\¾§JèÞ6ñÚþ” ۞̟LØ"@>,ÔšÐX¡C®€÷|tèq…9hºV[>ù™#B]ÿõ‰Ô²$fÆ/ï@©C¬)ÕùM9d¼‡YdœÌÚp¢»:eK-;ƒÙ©wÚÈWÑ þFqX…|æBu0 £H ñÃôt«Ô"®¬«Õ:Xò©ù [:õþgÁ/¶.œFxÍìÕo¬òz E„ æx,{Ø•«ä5˜=%0 œòE{_ŠGñ0,nˇ”=uÆã\ÁŸËðò.^ÊàE½ž<ΑÑïŽbm£Mô­ÞÚmó‡gtGGº(–Õ:x˜ò̧݄‰xĸaqfºsñž;:Ã,  NHÚÄO²‚%Èùó~¬´*/™|¤Î=k;Txµ' ¨Þ¬IfÉáWv)Õ²Þ}¤æ:Ÿ¸426jHûEX dÿJ’Âõ#Ùcv'ü9V‹§ ;u®å5ÚLÃ}pEµÃZA%¶5•ï_ן«ê-þntê‘5eÕe°Ñ+Rߨ@ &ë~·>Ê Y‡º‰oÃ-P×»œx!!Z>»j»6š—çhš‹ï·ºˆïÖ…Tè:iȾÌjè”±.%÷ΟLòã¤ÅbSwE¶NVKØë0e„¿Ðš¤ý 2ð-èeЏ¶gê2ÀEy‘ý,:«‰óGfH÷¦ø×JÃê~%{R—¹^š»ÇµW&?#Ë­‘éBANÂäe5hDŸZ•z ·n –Ûj õÚ1¬¨12'ü.RGwjÙô{D×(ï¡ÉO{ “FS‹Å[?³ì®2&«ÿÙ/r³¼ÍÁo.Éñ„cŒñ$¦*i5 Û æÕ¶v«<CKkøÛ¥¡D;ŽkôXRŸû*í÷ºÓßùLÉ߀÷û3GÚÇ ‹®Íí—ØÃ9%ò.Üâím£‹Šäµ¯ôìÂã"y௒£vkè)VÌM%10GÖáIHYFáNº°=é8§®2¶p~±‘ÓõÌ»[¼ƒÖ†årŠzô±»bvÖÀ>=×({«vötç‚Þ/“ç)ºî…#ƒH°¢mûtÁ:0;¡¯>¸ZL¸Ò\ž½Ÿm7%š_# —¦†ôŤô\VÙ=!a äA\û¤ærãƒÐ<ýÛ Mó®&ûkð²És‘ ²Å’{膌”š£zŽyÐel¨šÌ†¿¦…Š)ßr~Ïð]…wÉ1ö(ÿBUÒ»“i/€<&S?[<°n…ϿᚗË_¹0¯VûÈö¸÷®5Þ}Lí½YC6åRÐ/¾c²átNöÙ#êJ®íñR6¸S'䎭¸æš“»b ýÛÎàÄè»@ƒãênÍqtܬÓ! »´ŒµOhª’¾l~.â×R¨­#aÖØ¬Giúþîä¶÷G+½;%%ü#€ìÁ'³ âdMQ -¥G â´S‡¯êüet²SÎÈ^¶a1 >ö8õ<ÕÂü,Ì#M"±ƒèÔQ™øRa\…õo瀋ß¼9çÒÑÏj:Ä»·ü{Õß2¡¿hFCTuºÍtÜåcÒ…0=­ì$“Qây´$¶ã~ò‘W.Tº“‚æÒJl|æ÷½|q<´i›Œiò.í«²2j~ˆl3˜ˆŠ`æ‚ Oq1cð¥¨h=íàücôš*mï¬=ú=ŠÀ†VϾ¸OxÕ"fL}ŽÌpû±(ËTâÍð8{¸ý|B†Û”ãÏR›ˆƒnRK娍Ñutƒø=„¨ Q¾ÚIêÝÖ8]§U–핌ö,fˆ.ñ&zûZVCS¡º¼ÛÊâ>v†¨ížý XþIk3YCLMYŽßQ_Í\“‰”­.~´Ã#2ÛŽ“Miø žÉ?Z›þàoB銀'‡hÒ÷è*òPèÈ¿Y“§t–>4µöý8ÊŽáÒØw·§üŒ¡¶g)FI¤“:£ž 4þ"%3¦»"«„±›)$‰ju#ø‘UJ³ Ywÿ»äye#R!W¨öë÷App¿Ÿ‰Áa=ep£#Ö0‡Õ”<À„væA5æ¢Í~9Áj_i_ÆKS0Nþêœp¾_²X-^û ¦IûFMéé|i[ïqHh.§µJ #³,žíׄÖ`Ì룧é4®½O¿¿ÈIy¨âMg2`ŒÑü>ÇK„õ„ͼè| áÔ"46Ÿ4ž>&z&ð+‰8+VNW?Á²B”p­9X¿ó)¼… %£O²uÈg‹ÚEGe™¬kà)¥¼–ÙRíÃ-ñpUã¬÷¡†ÛÙt;ä…ÁBµ—h£þAÇ…³+ '26K0ŸéËwjqDJø>M<{—ÈÖ«OŒ©^ø1M½ìÒØøåµîˆäÓ Å2ÂVâ {™ðØ!š‡#£é—øŒŽío±0P`Æ×nÞ%7±ù\Ä ôYü¥û—?@ñ’j!×1PÍ8iü2åíº¼Ù@ÉrvÙÛ‘ß0AfU)­§~ýÅFàËà”ËÍ)ëþ*î®$¤NUt6£D=ý"® ¨ïÐÃמ|Žû‹‘zå(;Âd¸(ªòY2~& N,(q¦êœõ“ã©KZÖè¼6›üóÒEEºömë +«H{-¯VvŸÌ¡~VÀ´ª'áÑÔ[Ãx` ÙŽ¢ï€[w T'.XSNG.3nŸ,ó§CMßæÒe˜Àrí«÷ÐD¡è8F$ ]©oìøG¡Ñç‡bZÃƒë¢†Š„]žNKÜvå*ù¸?ªtþQ+~U…ýÉÖÐàI³9:Aï„Î*åb]²Ë–iÅÓe/5æ]v1‚1Ñ€7á Ÿ:%®£äN*PÛ&`.å—Š =vj|˜¡Ü=H©xˆuµêw¶ rx\–#/(«|žÀk@”¯@€²‚±’3Ô=.ãj3¦܈—žŠIª¦…'Ý4n ÛŽ¦s¯®ñM}·FH ËÒˆÌ}'”ÑÏÑ™Uµ€b…U®üÌIv̈t—ˆœ´!z œRû(Ãþ¡ÑË´k²Ä?0Ûh´qÃD]{£Æs¥8\ä=9Ö(x»zÁZ*#Ç@ø£7O5ç–6á\bôåI:¨IÑk½ÁkT,hÀ¦s£.{J­cÿÓN±·<3 ³{]Éä.؃²©®c¾Rå`ájgä„ëUÕ…Ú¹OEÚúÇܲeçDûâÅ‘se¥à| ”ú&…œM×Éšï]©«Ñ,yæöJyBkG×GÑ)ÏŸ½â×øTþ†3~¶¶û–åË–¾xON°ã™ÜÜSWï2E±¥‘AÒðH||q#w ¹Dkâ°Ñ´Òð¨¨{K…@sZ¨¾VEƒhý+÷wø-örÛV=wN² –®î²Ì¾Å@ntŸ¯²nþ ÇÙ-_ª¯{é(¨² 4|O ì71-'c2™‚d™uœð= ò¼ðû˜›uȾ$ÎsØ|·é9Ÿ݇ùÓ #æë)åv­ç`±*¾_ô“W”ŠÇ 8BXWÞ¹jǰТoŒvce=A—ïîùË6xzYìz'"–P…í­·ÎAªZs½±¥GÅBXï`Å’¨‹ y–ضµAsè”wÁZ/$‘žÉXkb j–‹¶k©ß¸ÃxsbJƺ]ÆÁ|ùi6¸6+næW!åãá$¯P›—orý½Ñlé¬w­À\;Á!ãqíáœa#—MÕ[Ú´À¶ÎÈŸ…³O>Ýbº 0.ž$œ¸¶›à¼á.1²ì<Á,žŸ…±õ#Þη|’æ*àÝEÚ4Mp³¿~¯Ç±pò¹4Á9I¼&dá>û¸—Î8€ž'®¹l©-­=3á\ÁqÏÜOíÌ`d‰Jf^ŒdEyª¸m vÚ¨*ü­NÈììÅ Ô`ÑÌW¬M~'CnH’öÄ€¼³À/Cyr0$û=ÚË÷a´Îc×FâüTGz©¥ˆšæùS+£À¾q/àKÿUSíXÃméÚ¤T´8-ÊMØ„Köeöm_üCÈT<÷N|ßô£|ñŒJîœK#ŸÝ“³zoõ¥Ÿ_‡Q[pÑ õI–INrQ”ê¦öN~¹\x¦9É4gÙF$çƒâßèÕ²?ufÑ«Žl–!ÄiR5…¥i…ŒË¢¾ÌÚlú~ùú})ù kO¬“×ÙÖ.y”ÈÇDq¡ñS,W âà'zxTÙ-ãõ¯—_Þœ5E®` ¿ë!K› zàN—ŸÁõ(u3˜¶•l¡5¬Üd]-öeÊÙuS<¿êBþO5ߨusÂÎ]£úÃ(žI<½î†q˜=Š õ8¡hââ¾0¦•Ád€¿bZT»¬Ÿn»á ïW®U#fÛkÛGs€ì—äùl™-Vº0ëú¥jÿqiâ˜ÝtŽ Ggù¥T6ß(;={©šë‰îI&ÂÞBïé-×9\ ¦ä:N©ê:>ÃzÔ¡÷¨ÒÚ£âAfô–‹‚kWò\:aöûÖ½¨ˆ`+Ÿ¥Xñ~š´ÐÑ U¢¥ßx¿ÍèDEWÈHò]s[Fxâ9¯0»æÚù¤ Vd—P‡ß´\ù2FnÂg…éo–挭/ΰ+éyl¯‰ãÒb^$ÿø˜ÌTRÊøæÌÀD!‚mÿ˜Ø9þg…O5„&¿½*þ…×–÷üù.Ç–êÇгè¼óÆD8ÿ–€î?¾1È] f†_ºn’-ëÖ¢ÞÎGRù¼÷ÑõâÔ¼ pyà ŽÍ}ygR<Õ#^©öÞC¬V`øÕ±à¢OÇdzϦ³O‹ÂxC ³0¹†(åX »b“Ÿô[½Ëµ3])d*ÒY.qŸ'õ~¦8$v—à“8ÊJñxîX;Љµ'ôÄ[SúDèÒÅ›’wÈÖGsM›hÑ(Ä’]çZ$[ô‘·´¹–Õ±+å$ÄÎYæhúBŒC_¥È- csß(Ý6·®”"¡9æb“ûk¥w µ&ôªa?hm¨b¨£ßhì^ŸQâzÓ&ŒoS©=§,¤–0÷qøŒ@QZÓåüJBÐ5›â9|‘¸ÂŠX±;Gkñ2»·}htB¡-θ}ýj£m2.Ëi]D¤Ð¦•{x擲p<%ýrwœîZPK»Ÿ Z/í•ǸÆ9D·u5”Ò=MÈ•¨ÌÈ,}ˆ‹+UPûóuöFNotNìk¹N¦)­©dj¾ftߨöL™¹‰íxfÖ§SGnTzíVgÓ›ƒ–Lúæ+YJE<\Ô@{òOz±mHÑÖ]/;æã;Kó-ëï:¥LTL{ûÃðýÜBP-±“\' ¯„Ì&ŸÅMìè¿ð[5ì%©Ç8ëã¦VžÀù‚¢;œö¡ÇÆ/ž)5Ì(0®!‚¥*ào$ømï4÷€B-%ÙÕgŽÕ 5Š&qq3‡R‚5YÞç` Ñì)œöÕ"x|ýM¹;C\(%CRŒ9¸õY Ìo9š6Ž­¶ÑÔe#TÄ\@IÒ~ÉÇ£;k®}š¨e̼ÇîL¯xéÉ™z……ÆY¸ÆÞóD6ò]2zD'ޱ6ÚY‡oèêgœS“o‘®TzDâÆŽ~^´Ì; Ñ‹ù U"¶â÷’§Ý—&éñy@í…¥Ì1GE`j‰;$ÊŒ|h‰ü ã:¼lõCRŒòU õòÕ»ƒuæì`*¶·„·Š·K2w¬…m²xÁS:O<Þ?·‰ë­>-«<“/™VÛsµ{se'„çƒû f¼¦ 5J¨6üËÀ@!¿wî{(GÝ‘›{¾‚S—«5Ëy¾³lAýKf´&x¶ôî!¶‘­CFEÞ‰¾æ/ζ¹ÁWIOèVsgŸþ´ÎÏ•î%ÓÂÛt³Ú•ìƒ Ùy›Ú[vDi¿˜Þî¬8«g/_L$Ü¿(Ðîÿù­Un¨Dû2!$GíYúó ²)'ŠS}&–€·ŠÐkƒ×EŠ`’Üë ª¤tý®dºÉÜQDuQÛlŽéT#,-óT¦™@0[>Ÿt]ô”ʘµ½õÔãáCäcüîòõ ôqáØ~åDߎFŒÊª0"œÊ¤^S´á "ŠWà¦l¨¨ã>}ºªwC~g8oFÁ¶†mªíÖ¾pd¼“[´ŸË) ‰¨‰¦$•Â_¿ ô?/œ±®©=Ϙs®6ˆ,Ì! Æ¿GD—Ic¥œY}Y=SK š¹Ò|µSç|[CpÎÝìÜ)NÔ¬Òð͘·e+i_cF­Þ7¸’‡ B C¾±K¯â±°Ìþ\×òxfÉœPKSµ , o ñÅK¢™¤äÇP:í˜6¤ûš÷ðtw°-Œ£7,,é 9a ÍÇEá§È½a+èZ<@ט—àyB¼O’¯7ºÒÇB“©˜x›Ï5w‹Ó^»à¡`ÍÿJ*ô*ùÂ>oãõ;©_ÀœÄÓO~ôõ,šÊùG5’ª6¶~Ÿi­ÛKQ¢³¥˜L4n1ët<Á8u¬eP2/´Ô¯úµüóYÏ#Åïm2¶ðCB–Àïí›4ãèG<´3×МÄ85åY 9Á¡*Ýï½{>Ör,Þü¦4.ʶñ©š5º±_¯z°»J¨™ù5Ÿê]9±P:ˆ/D™òtlŒ¢S›Ø:m0'SÏàŠÏU ³Þ-"çÓ/Åúd­Õº2‚{:ZÙ™Ê~)ú²NóFò j’æ^V´ý¦okÝÄÔXýl Œã©ZÞkÏ»“ìÓÅuW­U‚ǢկÈÉß\«Òb&¥]¾P{¨px—œÝáº"å¢öÐüyǧÞ"¾-s_Âݪ8ó9SØ{<<¶F’è€G°Ö`ÖÄÍU%<\V¸õ„¼x*†>fq†÷ô ¿’{%Ÿ÷2ÎXʨYIÇí@Ëf±ÓÅ`ß:N¼ÚôgÉ…Ý›Ö ¸QéÁ;(%“2s¡Æ7RC´w‰T¦ÖšJÁoAÃÊñªWÕŽxÍ{tëmÓÑN9–={D3zrsþf;šŽI¶Žv¬G›£€L¥”Ö›ñÅwâù‰Õ†Ø¹/³ZC’Ð1ËššÛëøÆum?ÚÚ¢7˜eÉTÞ)Í8àdaÙµU‡v$ íYµ²â•qNqùá£ÀY@%ÍÑ´ _Ó3šŠ,4;æ-oõUÆ•q "™aYðS(ç€ ¶úcœŠ€ S’1:F¥ t3²Œ¶b¯0Š”"¿|Lg^æ‘®eÁÎÃ1ĶŸ&¶_ßîùöÉésy4w»šø°ÒÒ‡pv­•šTm™×¸Ã$©Ÿ)ÞÖQÕ3XOÎ*¼ÉÈ`Ë޵ÀW¬m·a}ôDž$?ê]IÏЩVWeb((?ŽüC»Ôžãi›|‰Š`ÑÖ"Fî¤Ô²Ë׬—hm¿RÐlžå&Ó6qÆÍÉBŸ‹™*?¥»µ`PÕk¾µÊ~êz4ÝEëu5•¼ó¥×`ˆ>¿ô$,µÁ§Ã=”*Ô˜/±ìÖðC¾T ƒÿeþÎsôR¡%a/.v™BV]»A%£¯—Ò'—1]SðežpI³·9I¸Ž?~@L¡ÚÿÅ Û Ü9ÝîØâú‚ÞÅPÝŒ62ó4µ€ßÍ&c¡É<Î Käø¹Õ3´"7üXEÂ%ôgÇÛm _Ÿºwžiy-H½µûý¬¤¹–s¨D·[9aÝÚºåmÖ&ùn)AWù†a¾±vH\r›¬~ ŒŠÊˆ‹3ï±Tžþ†À¯_P’k þ:mâSÙtƒþ#ØŒÂoÿƒä`OÝžzù“¼k–ƒK©OèÑ‚€”™>›ÍtŒ]¸H {a+‹"¬ÈÇvQò«§œg”˜•ýÛÈ}*tøµVŽ`Ë”«§ó¨¨›û™Ô× ˆÑXf>ý˜¦sQàpKkœËµq#‚P0ì½D›´›K8ýß¶¥[eV¡Ñ’(k‚/ý:öÓ/@ðü“t±G$‰”ˆžcaÂV]¿Ú"ã ›ÒTE¦5ñÛoÄu|±Ÿ_Ѭ--ù¯Åý(ªê/Í¥­£=tÒÃP%M5È<•Ž·•KÜ _š^ÄìmQ¦L^Y$´W‘Í”¤´Þ&û˜‹ñ\PÊÉÄ,?õ"yõ9èqÛ ÝhT³¦CÛâD‹òM…ÚeÍì÷Õ.Ÿ».!^ÇØùEËÅ׌ïgʆ“CB„-õÔòdSßvc™qÐÈö Ëþ4ûueñ°w÷!®ÍûCJöx톣7èTìz” †âiR‡Ðƒþ=K‚œLƒ툚¤Òü ³ÛrZža ƒϬý¤ªMvãøÙ¯q—ÔïÓššpžWîëcK×AwÎÄÕÏvÑV°xï¡1zâ2¦+nœœWœ[ŸÓfퟦf¼Ùâa}JéØ0úVÄ‹{=ÄÿŽZDfÍQ6ŒÛ[òéÍó¬3Å̯¹+,ùz<wÇ=ººûÕ œŸŒ" Ì3MxJ•Vì­ù7ê`†ü/—“ Ÿ…`‰Ìz<(R3ªÙh´ ßǀƘf€]¨_ÀË€qè€_¿–À.{ËRë(½ ß =ÖÄÀÞÇjЪ¬3§Å¶eÞ¢á2$1>i¬Òcòñ +Ú¼µÐ–¨Ý"/#ÖÛÌý %·Úd7•ÕñÙ>2 /À¶In*úH§šŒ•Í:¬JÀ‘v­ÿÛ^,37kÉ—NÔ ²nB$¿×u—³ø%cQb%ó7ŠçôeXÅÖe™kÞS}v‹ghޱΚ&ƒŸg%|E–"cƒWš•0½‡cê$?`@4ÖÔ§Ï ¾Ï¦óy›RSI®ß¤èب’WD÷˜PìyêAx$´.ü_«}žuÔršødjý“9'}+9¸6]çIèdX0–°K)ŒWÜk ë=+Ë#9Ád/>m‹î}ŠŸ—,6fu:&o›ö“¶$鎩MØÛì‡C#ŽÆÊ;ùB`îOWÈÍb.:IJ]ZϔݼNÞM/ì ƒB69/Ç_ŽZÝ€þT endstream endobj 103 0 obj << /Type /FontDescriptor /FontName /IEFEKY+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/F/M/O/T/a/c/d/e/h/i/n/o/one/period/r/t/two/u/y/zero) /FontFile 102 0 R >> endobj 104 0 obj << /Length1 1591 /Length2 8568 /Length3 0 /Length 9602 /Filter /FlateDecode >> stream xÚ¶TTß.Lwƒ„ÔÒ5”„tww×Ì Ò!¢t£´twª¤tH§ÒÝqÑ_ÿ¿o­{׬uæüÞ¹@VB`ñHðOz®V.`'7WW°ã/Šœ¿Ê_NM-mym Ö?ÿí“”„y|ØyxìÜ|@€ /€ŸWà÷ß*óÿ‹ûo«ºøÏ½ÿ)¨µÿ ðØ»¿hxü© ¦?'†ðßTaR˜þQ¾1hõøàúÖÿï”ÿ?ÙÿªòSþÿnHÖÝÑñ·›é·ÿÿã¶€€½ÿ xT²»ÛãT¨Àgú¿¡z ?&Yd v‡ü¯WÁÍâq:$ ¶Ž·ì* öY«ƒÝ¬ìþÐ_§ðXÞ ©Ã\Á¿î;ø?¾Çy³rx¼O\Ïê· ô8Nÿ]Rj³þ5wÜ|Ï..ÞÀGyqóñ|¸Ôäõ[ÙN(Ìí1ðHÏ`sÁøu¢ÏùœR¿L¿ÿ#Rù <pêþ}£_{å´þäp‚þ¹œ6ÿ‚¼N»A>'ø_PÀéøäz¬ ý|¬ û|,åò/ø¸G· ÷c®×oøŸ.Y¹»¸<Þ¿uüØÂ¿ðï« òYaÌNÁ¬^„ÚW…¶\VH{²o ‹ŒÓoè%3³û̺´º_ã ¾c.O^r9—x÷µ o~M†éL|ŽúÎg§±5¼é­Fóï­Y¼æèF3ÆÌ7⾑ü‰ê^Jt vmñMß;g_Ý ÄFøvEúlgwõ\ÂKÏ9¯êÞâïƒaS›åÏ•0o‹ÇØcubŒƒ &ès,3&IiQÜØ)ÑX½p'ÎÎÇ ²F¨ãY1üvcy>ø.s¿¾š|¹PªÍíÚAöŒÌ”ñŒ`p”ÁGòÇ{E’iŸÂÑx ܹwŠ&»‘3Š4L¦Ý>öÃJ¨Ò»çËQ ^ °ÖÍ‹@?´Ã—²-R5hÑ›z¯ ä¼:Eš °Ä¯øVx†çí¾—“î{Š£©Ä`Ë@ìA`ÛbcÔë'.IDLBº†™Åq~lî^Z3ùm‡¶¹2½‘.ݵ’ø8ÛKêãbïäær³ñ¿«§ð·á.¤un&‰šG›¡·4“ľÀ _ÆLÉj«¬ÞŒè~(T‘GX¡[Ÿõ6”kM´:Š¡Ñ®Ð_Iºð³upÌ' ÙÅ&Ž?ó)»ã”ð‰¸½ Ó|µwEô¾‹aMorûŒ7èÇ9‹×ªÝXËox-¥‚ç{'ìþáLi©C]’©ŸG?Ý)Pqܾº…}[ˆÐ©2Q‘í®c»]Úþܤ×rĦ]=ImùÍ|‘£_Û…üù5>“õ5…‚¦ œ¸0sñÔmÞ=[(E‡VÌÛ‹ÀÖ,[‹á9Í}ڷ‡¹À æŠñ•jܧ¥]fp¨„½^å2`JcÛÝ㱚ý­´þHu×­ˆÛÊò–ô¾ý9/­7)‰§~¹ÏlãÄmU¡38i«Il]éܰР8úÓÎág «~•_²u>qå"´\nÚnQµªï2À\G¢üƒŽ£D^™P—!.L:žÄ÷aÞ}xï%÷å¢?ê WÄc–]ô+¶ Áç[ñÔh+¼“!ŧK¤¯[kDÐy1²f<Á‡ÓTòØg©ÜĪág2Ö6áÑQ¹®¤“Ò‹h„x#ì>Þ¦ï4šˆ× ¸Ôžqr82ËXB­?20|Ÿ u˜Î?9“}i¦à™Ý?c(~³¹PvK/œôF¯\¶Õ$L¡bä´—üê9þhô°«'ÍH¢<)‹ Jÿ-Beއ$aÌÙ§ÞQf›‚÷P?ëÑ7‘’sÕOÁ÷­Ð,žÄ)Byû‰1nZ~OÖ_å’šæÃ7ø$YДš)U˜ÍòH Y çI¿H7Y3í‰ uÈ}­Sí+ršŸ3{óyÁýåËÔ™V2œÕL¡¬€ÖàÁÝ8s J|î/Ôo¥ëyÒŽÄ}Ýä_!?£øæfBäЩ#ž™ÉoÃãd¬"½†3"CmôàŒ[\Én8 Ù‘’ç'—UÅÄ<Ñf4©¬ß¾˜›«Ôœ"ÄZ5›îŒŽŸ”“h=ßÇÁÏûðµù]¡gžv×b_ÈZà¾ͱïİÝVT¶Ñ»A\};o¾’+úøv©{ä‹üölj^ãáõgóÉÚî smJÝïüOôt>À½qò')ÄL˜z×Yt×7Ök>wš!WÙòH¼Ï÷Mîê…‡[·8× ³OĔΑ¬âº{‡~N+ xÛÈFõ*¦P¡‚r¿¼KÖáÖ…&%ßÔœ|éc߸=ÒêÅ6m÷çÀ –ܾ½t Ò€t˜vJ fõ\¨Ü·“yíɃ¤u‡Kþéî öœ÷ˆ÷µj6i’Ì•ˆ:£R62ÕØ z@ŸÅÕÏr÷_%þ@¯%”¬¼ˆz%I¨IæDg×¼€‚³Ý¾ªd¶~Ö®rz>ë«#µ˜R2Í2´ÇÒ;{Ø9¦îiˬp¡Ú~²-0©߯Mö„¿U3T’ÓËóRŸ;èÂ_¯_7ºÐbza' Þ>®él¥Sk¿(4ú‘ÇÕF5¹&¡Ð¦ôPƒë¨âi1زY÷ q`¬¹rPÛwïŸ70¸Ž#ø@ÇHŽb\P_”‰3‡é4¼]¯ÈÇWRbû‘?.ÈVN¾,bDŠ)èõìøÍ\Í+öyáe2´!2ÜHfiÅÅYCË™ê¹ /5†æ}ÄàW_{ úÖD–ñûn§%¶±½Q(@ßuMe!ó‘"yû¡ÔA××—(1©ƒç7Ò™€…¤ ¼âú"£e+§ðe«WÝ®þƒTjVƒÊ~¸¾òm‡hQòB"§,xðKªŠ'Ÿ™Áh°s~b+õ•Œd±ý•½e“æhóˆyvR í°Ÿ i5~Bh”W~‡ã,·wÖÑá4CttŽ{íÅ9ðX£@kjo¯æˆ~ø@±Rj¢{áXÁO¹ãM¤–Q[ÙµRêÒ·…¼LÊ=”û‰ÛeÑI„Y¥´$üÀ,Ç${Ý‘ˆ¦±* ›W§*@÷wÝd¯Ÿ¸ ‰q¯¦‚¾¸,¨âVÚÃ$"êšjS¨Çb_G„¯4~F üiã9å+Æ5ž ‚žw'ž<£ÚÙ+:+Ür„Hã_¶¨ÕœÏ BÞÍ–Í›¿Â¼|3šêø4=…ÿÃs3n=¥€g¾ß¦Ù›‡f‹÷EáÒ5+anøè‡`{'dÖõúv¤&áÊ̶§;3À®ÓM‘n…o|ÅF:]îõ Ϋx#ý¤¨°Ÿ) oSØsMk·8:˜[ÞŠÛª§ç—&Öy3ïÖâ‘ ×±`EˆµÙ¶Ó{û=š6ò0çå8ü½ì¥2aò°ÐkUë ÊF’H¯7†/œáƒëò§Áu)Pnª]¦¨»_2’༠ÝJ‘ F2Aæãþ ©Òô©Á?³ªv×’ynÑ/¾ŽÇm?nÝsq--–7]’ hÇ”oÁ{”+®Žu€p¿gYÊÃÕ,8ŠxÕÁúƒàÐx©ÇwpõÁ{ó$IW÷˜Þx6Ouó&Ÿ¢4,s` çÎÒ¯PeéÅ;Eø0¬'Ÿ‰§Ò¾béU{ü©š+|[ÜÕ©š¡Aî°\²¦Ÿ…èØtnà-–±\80Û¦R|KDæÁ¾6gZ®ú0ÆNŽŠTŸ$&V£‹ëX"gD›1$õâYG@N!J±ò.S­ÆØŒ„1¾º°Eí’“›‘«†ÓT«AépA…f$I´ñÜ #N©ø {5±g4i0ÂÚò$g¶û…LÇ€ñš¸›ö^—;5××;ÕΗ¦L˜§ÅSÍ–íˆ *­­4Hœ$”AQUzÆ7S÷?X£Oõ<7ݵ$ÞýЯ(ô¼¢K‡ý‰D!¯*•Q™UdK•º³ØË™à%ç_ãÃ2 cß»ÓMñ»Þº)uç±"~Iºêc–ÝcB¦ºyŸ)>…çUúŽ>ùk¯§ÃåVî“¢%‚Ä‚ðD+¤Oeu(§ˆh5¢=“IšOrCRý¿¶¾ “ÒjVé¨l:Û ÃùŒ¬(O”DôeTø©œš¯½uYKµµÀWiŠ cõpg} ϪP’z]ÒÕþ¤ nÛ’âÞ2Í*¾JGç÷O›¨{ÎbBÅ&'¦'V|`c+T«÷hBópô™äWßÜ·?aÓ8™‰evZy財ùÛG‚³©?|þ1EÈFB9`r¤ã›r**Y‚·§%â¹,òù™ýâÎ>#Ðû>p]¹;Ç1ÅÞ<Ç„’GßK»c°ùÂñ¶ þÁP|NÜϤ*Â4ï™îÃ5~jñÑàcˆ»:¸ÔŸoûrÜ3ßÎ+ž›Õ÷Þ"K§_ZßäLÞz¾•QCDiŠüj»¯@ú×&æç‰þ(÷ê'NgÁ:‡šåª]9“µg%‹˜†UX‰tC ãÉj'–v ðT5½pEJ¬¶qYÕ+ˆI†2ºÖZVX¸O%~åÇ™o &¿¶{¿¥^a”µÕZÁCoPTÁÞ÷¨]lõŽÄþ‰‹ò9©ìbǼD~F6G Šnø-F¤ú >Ótí]ò5‘ë>Í'<ÖjGˆ†òÉ ²õ´t# S¬ÔXy>Áë;‘Zr-¬æµ V\ÆàÈWW‚¶ }Gþ„ît2JYs™ócA8†}<ÇBíAèõ¦SwÚ¯¥U¯­Xˆ'î°ý^áCÛ$ž]m-,±GÔ»¾ÆÄðŒ¨ŒÏ¶ŒWW„…;ä"¡Ò…^9Ô‹t«á€ÿ +@P9éëí¦Äë¬zñsE36À­tGõØ3³I¯Æ×Z/9ûÖŸyö&ÉÄG¬à8ÊÁõí\æ¦$ù¯ÚÅÒFo%Tµë~R 1šÔÑyqV¬þœ }­1/ŠÌck¹¿ž7•ºÍ‰Ö]‰ jçÂ;K%´TÈ5?È‘¢ý"˜»v™%šñž€|mHëü¤˜±xø0ÐUC‘ª›Ê3Mw¨:B³5?µw-Ù¶['§WVÓp‹C#Ó4À_,>¼87O`jKá3Dd' Å„†ÆïÔù¤¬EK´ hAôª7“Møh£ï\dÂ}æ´ŽŒi×Ôçž÷–n|®3ÁŸš’‹ç¨2±n2+÷¬Ñ±ÏYºb-³/S •”¶ÒòÜôùP^¬4é•4u½“R>IÒ'PdNL£A¶uP uïà›?µ-üjòía•¶OIsâʹà—:‰Sjeqò:ŽáÐ]4‹í¦Üt+šøv¤œC`ñ4}’—@“º€`—hXËYÌq›öÁ§jgŸ]/›•®Aåb4MkŠ191F&¬iR`C#/¾çª8Ò>$ÕsÞN݆Äûáûý[÷œ+™ ÷3…Qª™©¹'ÔJa÷a{à %˜aÍÄ¡]ù`>ì„SåA·ïcœjÄßj àR7¹½ uxцiν\B¿?+ßÑ U[mZƒpó0d\¶dí@~ôT”Ê3lËJÏDa.¿7¯T¥^…ãxó Õ‘Œ0n†.úÎT ýt¹G+S)…@„Z:wRôqUuš²K¥GT/Zâ×R0®wMœ+J#°×„LPØ1|ÝT‘5†|§1ÁÔ«…›V}€H/ŠyØñù½ëÁCT?t"Õê G$¤ñòÈ~|+&ª©cÉ{oíF¦•…øNž\6‰ÈŒï܇£ÚÔT¾Ëý¾2µ­†‰f¯!“wYå‚ hº—˜2iFZicøÙi)÷e s¢R·²ŸÿvN&¼¢ešü\. {X‡ ˺(ÊWÝþšÓÀÈþáSÁ+æŒcwøëàoá›—Þ—Ÿ²”Š'Ž‚\´pbx{^k¡4œ}Y1­Tçr.’%˜{C *ryå@éÛjÜKm#´Ë¹T3“lÂzV::åóvPb]å.Û ™^âx-œÚÚ9y†ìIWˆ\Ô¨™¬-ÿºCm}„³X™Ié)¯æ\»ØHÖ±œ¨‚ìr˽-_:¤?S—øCÜk2?ÖªÖ i~ËúzíJI:ÂÓ|žµ½=Ý25¨—ç7ÍnˆrÛëï ¿7‰º;H†ÆÍ`ƒÎD‚=»1«Mñ¬éVë Ÿ%.'Y>h–b(;#jÝ'*;ªDUr‘üÙåÜ> ÉN¥710b“—ÛáYs™[³Ø+6·Ê1_z;åI'¼Ë;è‡èK~Òä/W¦ÛXxCì|—4möñ“ç3ÞMŽ)+90ŠépPyTõÄ+ˆòGKKáQœgW…È:½Ç…`E»ÞS…Û„2 ‰"ŽçIi%µg7½Æ•7ƤÙKôi`Èö ¡)µýIdËå ëÌŸÓ lµ]Z<À+5.HX FÈŸ¢øùm†.zÖ»ÉÐNŠóªÔ uoÛ“…5:²€ä­‘F⃹×ÛßÑéûêuDÇ´Óv$›ÑpGuKaÜVpQ×Ô}Ô×Ù#XÑÃ0òö9§;PÌY¯ËÍ9“£­´ÀÇ«‡<â‚ù˜Ï{ߎðæú’ÕRë•LŒNhÑŸT¡…_½™/‹ñ âEþ õÉ¢Ôf¦ ‡%(ñªØÖBê¡ÁÄ+ÐДßj{ù ¶÷Vˆ¬ù€ƒ QÔéêw¶}&HžoÄg ×éúZø?/ì]C§;ìôlÆ·Y[›œ¡8ÄGTÉaÕëS5u^ê.½¬4Ý™²6.÷ëÜ$M¸ú"ì‰ãàd9$ë6Ò¬ôw Uz‹B 8:zGñåÜ~)´kåCÇ!•;ûžÐ[ø¤)¡ºÇO–4óçÊ SüµŒžN‹1SjÙwrQu[à 6»¬XŒI½5tþì&t¸¹n,ýO-ÓBDÅ É)Ÿn&‡m³ †þ‡Êr| 07M+QPŽeqfß3ýÆüuö¡h§Ž©rÑS.°[‡ÃÅæÝ4æ9ž,à3<õLv£‚Ž amWÆAâÑ€jŒîŽÓp\äÜð—¨äÂÚH]ÆÜäT†\×\O;¶ÀUnƒ¢6Û€òÉö­‡Ã¢ “umšPc‘ù“:¸3ø|Ü3S$˜ò³¨¦A &Ý´¦Ÿ¯¼¨…(Òp“È¥·®wÍïÁ'wI3!ìl8ë±ÕOðúœ$DÎá­õÃÞ¦¤·3îäz ¢™>e…cãwXt*!N¶Û¿¶. ÒŸ\ —¬î¦dœÁmèøàó³ß.³´’òdgzd¹ÄļømNèËëÈÍ&jÜ}CGVQ÷¦‘o¥íKŸ[¨6„E¡‹E©áíóÞ®Í5µ?-­Ç=Ä_Os¿¨iííJ½'÷HÜšŽ%’¬»Ïù2Þ$V¼×O…%~¿åMrûÁu‘χ˜(a¦ž#¢/=ÙžrV>æ0ºò£Çðóc°$å‚Ú‡´”5Âù\±Ü¨±nøt¾üp&ÏÊÂ=#ïS¦µ|rŠ2õÒcܼSºKÎW‰!B¡åQø@aã¡Mv9ÄU»g@“0_È0ÈZ9ü3¹Uô‘¸¸üI¦l Á‹&W™bŽÞ[Á N¼”óù%Œ4B:KýñMA+gh. #­~¡‰H:Š/_pGýK‚›Ó¯=ï?„ åÑGFJíäʵ¢Ö¡p{AéÀM×Ä+¡íýÖunEÊBÂe$q±ä‚U+ â¶±#¤Œóâ`äÒÚ'*u?RÎKi‡w®ÞC³¹ÂIæ>û[ŸH y‹°–ŸíšâÔ’DÜV¢n¾T>ì{úñ~àÛ³ñLˆ¸yÞ *Ç\‡#*XM;·ppŒY8·” ?¤ñûoZ@öV914{ɘÂEm´* Í ¨Ï‚iÇ$îN0‡ÙTtÐFÅùIÃ׉^r¾çœ]6®}ì²?¸YX͘|ºÃY”g¨ºlùvJùɤ°J!„·µp¬¢ô­Z'æÀƒºÃúsñQy$¾wO0ÝöL¿üÿz<ÓÛrÁ LSòäéâ®;ËóX |„îÁq¶ï¯…¦åIk7Dµ´ÃÕÍ?*MâøIn‚îqfb?¼PhOìÄ2³t˜÷¯Yœ—›n0«3ås¤š!"]¼µ:LÓíûæó3+l¸*Å…ŸŸzç³+áÔv‡Y4/H°ú3àYÙäûÖ»2£}ǘ;„M.sçn9q¶bǶRzþ˜šUl)ËOÝÛõeáq 2„飰Þ8&‹-{ÙG(@(OðÄ•`üþ·e¼Ê0O6äôéSEþÉË–‹ƒC<œnhwú8Fbo󧻣l-ßWUýp¬xSB:ª‡=F´µ¶Ï9¿%} s© «nXtA’}ÍJ,-àøÉúæè,äø®§Jz,»eVœ"€·&T£” é.:Ô®^ùkFê>'\ã*b÷Èq{YµN6AØ€-&Þ•›¯ÎùIC'™á—z2æÇäåÑÆ6¨“!±VIîÛcƒ8Âße'{7‚ôFâ>ߨ‹Jî¼[9¯½íŽ ©PŽ¿™`uÞV{M ´¥‚/©t)û³±Ú endstream endobj 105 0 obj << /Type /FontDescriptor /FontName /RSTHTQ+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/C/M/V/a/d/e/f/h/i/l/n/o/r/t/x) /FontFile 104 0 R >> endobj 106 0 obj << /Length1 1414 /Length2 6307 /Length3 0 /Length 7269 /Filter /FlateDecode >> stream xÚw4›íÿ·­¨½µ%Fì±÷¦ö¦5JD"!‚ÄÞµ‹R£vÚ£jUVÚU³ƒ*j5jüÓöy~Ïÿù½ï9ï{îsr_×ç;?ßqŸ §‰¹ˆš3Ò ¦D EÄEAò C3i$! É@ 8Úö%ZÁP>p$BþÉ5P0‡iBÐ85C$ çë—ˆKË‹Ëȃ@0$÷·"%ЄøÁ†¢=$æCÔ@zaQpW74.ÊßG?T .''#üÛ æ CÁ¡À‚vƒyâ"B!s$Ccÿå‚_Ñ ö’ó÷÷…xúˆ"Q®ÊÂ8Ú `ó¡ü`΀_tFOØob¢ä@€…ÜçlŽtAûCP0ð€Caœ/†àbÌu Æ^0Äeƒ? €¿Jÿ»¿¬9‚#~C P¤§…#\.pÀXÛ@A  ç_Š$Îâ{@œp ¿‡´ÕL¿¿Øù@Qp/´¨ÜãC±_npEÖB8k ==a´ù¯ü4á(Wu¬Øï¶º#þˆÀ?g8ÂÙåg_/1KÜÛ¦«ù—"ÿs…¡R HFVóÀ0P7±_Î-°^°ßBñ_0.ÿà@/¤ÀG wá^ä>?ò…þoÁ¿oäââg8 p‚¹ÂäÿxÇÁ0—?w\çQp À„šÑ×cur òÎÓsbRŒØdO]mMI‚忾*¡Y :Õ»,,,hfûùv0ÕãêăÞ$ÝE°cwX²¯ÄÈ"õÊs&ÍQâ+‚ÔÓýRì/&²W}ÞNDZ²º³xÞ¬ºªí=oáòƆgý`·f2({a­<Œñx_k|Ïwә檳.„O"¿Åk„ÝåÑÔ2]ÛÁž}YŠ”¼Ü«\á…åûÌ.‘yy¾(:6ÅÖ_v‹hH.$þÓ興Âw-ÏIA»R–ßDû6AÐ,ƒ¦90 «tãŠà4 ·yxo<+¿¾Òþ$6æÇZÍ“”›Â1_ 2އ“Öl÷è]óölºÎ7GFèš*³`$Ë¡¡ÓÞ±b!$wùlµÄÉ® ú» èׄ¥P!=g” e›êQ€x[ „7šÙSîYѺ32¸¾‹ª>Ø0Sä˜ òØsÂÓë ˆ¨×/bw ª¬%€ ó͈™îeËÜy Ñ®ÕvDpVKa¦ër\£µ¸rGü+U¾ØÅçÖ†u!oð¡¥y/*>V/UÛ …ÓêðH ¹‚vô-<.½ÎfUç”íY/]>ÎXO~‰š—1½¯cµ™¾R(¡ *P²'þj“â˜H[¼ý Fp D†b+ŸºðÊý÷‰ÐµwGþ7@­Eåñç–·ó¢o0º]&…5å·ËLíÙŸ¥VõqŽ“<`:ìùpËÿ«üîÚS3…tMÎ;–ÜBö®©õô6o£/(Ï‘>þ•ÖM-Ãr[dj ñÕî@úý”1}I½C7#:oæBð³‚Yw8.BÌÑ9otòâeª¯Nêw°ÈÄ‚;]òj4òßÁA:Œžâ¾*‘ÌÓ’q¼í7¿¼Ïªs¹Ô[ðòÜ 'mµ¤«£vk-œj,èaåñ/µ«Ós&•ê6/ÔR*ï¾w‰h²JNÍ«LÖÍè¼±•ŒL|¥9&0ÿmò>IjðF¤bÒ3- ×eŠ&Ðå>üu‹Ÿ—$µ³¿JϘ¢xçm»q¤‚¶.aIQ‹™FöýS—û‡Û™ £¤Ï[Š&ÆÙ¹›[gnŸ}„V°ZŠäÖq,SÛõžó_ïQÒ´uP=[¡Ã p?™K‹3œËTS:‘2âyÂ"nšé^CíVµÖKt½.ô…€Ítgq¶#A)8%cLºûÁ° Ì÷­ ñÍ:J/NŒèA7)—çÉŒù¯w~FxW2±²!?€9ù÷¢Ç™¾y¾Õ”íoE}­|ê–àoíëÐóiZú¥-SsšR3y²U¿ýõÛU­``yMûë x1}³ô¶Ð ˜Òöüq%aQœ¾CÔe3È|=êÍC©öøh ñw]žƒ˜5å‡W§³šŒóNië»ÙoqÚ~Š˜{WH”ìx2èGäò=(øƒÙi ™$WIˆ-‰/ã7¿«ËZ<–UÙv Ϭ6óÙ&õA5W¥ŸóÌ”–¯%¤Ô˜¸«¼Á>Ç^ÆJF·Ó+ÿX\ÚJóVÒÐáì2?lUužëi n ÙT7•P ZŽsw ¨i¸‹hJy8_vfcGS,ÚÚ©¸Zò¹ß§+›ë™Ó³½6;;›5  ¢q3È1Ÿب6\Sú˜IÁÅ×fl¼þð¢W²@°PÍÔAp½âìÊó´ïo¬4ž¼}E6(ިķ›’+<³èVñê{¤{ßý¹M ­Üp©SmûøñìC„[Ðõaç+̤$ãbü]!¼_ºMêKÕåži\bÂåÈO–6ÈîY> ËâÓŽ_ÑßšúfÀ’µ@äwá¦p"ƒòdrôAâuaé¦t.!;Ö}”›¢=©šôšÞK½gr_Vç»R¶CÁGw9Å<ÀU3_~Þè’bëËhèv½´ÂOë'c^$cXÊ•ÕÍ׃Úänsœå•ÇL¥[<Ü8b\üI/—w]s’ÿë”`Jn§(Ë6ËçÿÍ재Ì"gãh=2Õ'­Å ~5D½Ø¨Ä„ã¸g÷æ-×Rnõ§ÀQFöTÙ“˜’ øÇ‚ô¶ô­ƒ3×ÒyJôôTét™ ñLE3®™$>D8×&ÐÒ¿öXͺ/(\Bñè¶p~Fó‰¦Ó¹m\èX8tiµ½0 l°©Í¡PºÁV}¡;Ý_ä~ªqömG¶>BUÙ4˜Ý)Z»È–hƒ~¿"`ŸÏ«×üž eûÓwÙã%G@Í&t•¦ó¿ù¥;p¼³cŸ·Î«8R_i?lp !Õ\þa›½·ÌûÝrã¼ù¼f÷cªÑó:¨ïhj×N7ÓÞJ÷ÉrmŒùö6—nr-ÉšUhIš÷m¯\’1ûK© K ”ñû8ü0|(ô«:ˆq?Iß*h¶p3Fòh ¤¹ëõ±lÓgPV­oýy^æªO#Ï.‹nHŠlMm4­~¹út¤®w—N±³¸­mÕìÃ^Ó"v¸Z“+è8n`üÊrkŸ¼?]¿«FçÍà Ÿ ÐJO½Ç/R¬‹uic;eÎj~ôЬwWj;#‰E¬‰ò¿É/S +1ªßÞ¨¾ð,ª\ÉÚ‚q¡¤ÐT5ÔµÅaƒßN_—7]ßR^ñŸÃè:aO}âZh? Ïaš¡ñŠ›JMaqÄn —[¶\—é“QøüŽE‚'{bŸ:týw%áYÆ®Ÿ—½I;vô+mÀ@‘c¡¢SÇ*ÙŪ ¯œmŠB÷ØØ ®-]Ãì[ó±¬ìºªá§®ðúu{ è(1©VQL° þåã_Ë;Ãã«Û¼z¤s—'¶.¨ Š ž{¦rŸ“›‘G½ý(´©#·j‚g¼p«&l'æ€èZ_P™t[vÇjvÊÀ}û º™-޽.îÞ× ÁªèVbÞ<¾Ô—JŸ¯H”F½u"Ò-w½n™/u%V3Ðw^¡ö¢ÿsa0ðYùw/LòpÖ+}Ð]7y®¯u“WµsdÞ‰ Éî×uÙmöX‰>ͽ´s¿UB Ãï(|UK$D÷¸o„MÚ¿T½®Ñãeüˆ¢ŽZûMO¦FÓ~¤@‘ù¥ÙmÐ;Ìhö ¢{¶6!3é埣;³½TŽ÷Z¤åRÂÕ´Ÿ³è$Ë@¶™{¬HnÏ,¤ýÜD_§|IûÃ0Ôª f²æ~"Q ú²‰2ä©]Rê2yú=ÿ~j¼ÑÉõD˜œ>Rž¿¨±Ù¾´SaoQݱ;̳m 9 _8LŠCgL®0*­'Yî!lÕÀ“¨T¾Y*“ÂŒšÖ– Éo1­x¬J¾ÁÜžÍÒ7¦0j”70wˆMm~ ÈîoÆ 2ÅO· f7Éloc3ÛA«ðþĉøüš‰LÅ v¦3ÎËݽÜâý𬅙ðµ ÞöÙŽÆθ†÷,.ÇMÚÔW¥¼<“î°Š9>¨Ù Xÿéâ<³xˈÝ'¯Ì¢eüæ¶žÞŽñeIšMòƒÊ¯—ê7E8ñB°“¬^ÓÃñ“ÕøŸ £bWfꜿQÒˆm”œ.K”¬ÜeL-ìæ¿¡‡oòè5_]IÚg‹ãæ–kC•®æ×NY’!’ö·¨¦Û×Ò:ndÉÕÊY(¾x[ðæ‘þáãè9&[ÝÀ,“SåH¶H)*3_(/ÿv%Ó)ÐB¥_mã^aÂnàW‰s Qðs h'‘S̺p›¶² 3ÇliâGȲ?®º†ÜÊ^ÿ8ÿ¼ûkËznð’.„°µÉn!CmÀ#‚‡‰¹ 3Bf±èh«‘[{Ÿ”F(ì&}ËÚÕ3H_Û·ä×ðë#ôW¹9Kv}‚§YP‚A;Šl?×2I³ÀPãUM‰%5F["j'„VšZ²© ¨&Þ4 b¢‡%SZŽ*¨R¤uoºÖ#µføî¿C4:­]ëµó¨6Hsƒlž9‰QUê»oXå\,i’ùRIs Íe‘C’Wo¿?e¢ãwÅJy˜'ìÝkýØÙOJaób”Ã4Ž2±”÷ÌùÐ+ÁÚUÄ Å©´Àëè'+Ön>€„ßÕ¬\úÊ l§´»Z?éÙ…s™’œ%x±|íC$ÓÓ6úó2öý”é›{^É(¾“÷DÚæ›ƒB”–/‡xŽ›ûGÛéY)ÚÜæÈÆuñ‹BHÃG–Dȳ#5,_Ô°hUh·b‘  ÷a7vU–æ'ç=Mƕٸ†¸û3úNë)’-¤ô8ÙŽCW™OÌ™UË ×d¸eyg¶DÊÊO ±áˆÓ^¹GK'&¥Zêu¢,–¦ßFY…’БUe~²Æk0·e3Þð”`L5ñtŽû*L[%”n6ÎÒs^¾g±Uã~Œ=¤1Sä¡`ù¼Fö| ¹Â;ù£•© üR©ÆÛÂ;õŒF)å˜dš¹+?†¼é3L8<Ì]°hèViR`ü>k61ÒVXlÊÐPU1kOGï“Pi±¹{0e¢j•ME­#Ò¿0“3ì [ ¨YÝ[ NÉÙ6ìøAZz³&Ÿsíå¥iZ©(ÿ”Oý©ÂÊP≱EÁt>é­ÍI9Gv⨫g<&Ös%°Û;‘{óJ.ZÀ·%Æl¬É^Κ¥‹žÄ3½öÚcŸÍ”tßÓO"Âè2Z¹“û“6ȶ“zNÌJUÞ¶êä.:ùØÞšÓfxV¼k_v,Ø_<ûXîm,ú„EŠåùþ#ŸÖ,Œ É3šS\ž±8) ‹SRAÜxÊ’?Á_C:Û;Ä/Ì”Sfö§øª(]¤€Šs®ªµ\³7»vÊØ«›ý|ã]£…{7’Y®6k^žzyÐÚî$Ö/E”†¦&Ř¾Ö÷Òõ;ž‹½Ê]iË>Ì£Jì<§Ñ}©âWá)yp^BNì¶vR-ð‹ï´–ãví7¼)õuL¨ÿÑþźÁ˜Þ&ª;9û#y Fe‡Vcë óù%cjê”îB¼/UùÆ&²b|^FÆ´Ç8(´“fW·0Ä{ò;Tœ·æÉë“ØŽ`b· cS!rº@¿5Ú-# ê~k2çÙü²[,Ö·n…[F˲ê•@s”…ݬ fó&% ŒÏß*˜è‚ˆøí½×”p·Ì8°ížœbÞ¬jœyÅÎÂ"y™%{IŽ]áMif_ ô‹ÁœZ$rîgÅ^c\W¡/S>ǣ䃹!J¥cãQ¶QqÙsü…»º'§ÆïL)ÉM#/9òÓ± ü¥ZŒ‚Ý?•»ªù–ï-GJÅR²»Q†FY…e.: ¯ú¡$,n$[L._¾`4îœ?ßÉshNJVÜšÒÜ"+¿Ä_rc\b‹k”`§A¯øa.º±ë“M¾~j”§ã X“?¼å¼L©ªf»jŠ]YçΓ7@WÞ%¨g4>òÀm­MÁKÂÀVqÄ=ùóé:- ¡dUpqÊFey{ŸºwXH·7öˆc°J Y]KKô: IðÌ9H}¶¥ÏßÈ¢;Ƹ̙¯ú;Ÿøeêuæ…_­€Ý{@fŒÔx¶s& 2E§Àñxƒg"¸‚¸\{éÖ’‹5ÜSÌnÔ2©ä˜]æ] p[{LÝ’£Ù¡»È(šòãM$”•M³ßfï#yTÛ¥ œft£áhùÁk·œŸ…¶Šì}°Æ @Ôû$ Q®pbµï>ˆ[ ꬡ)¬¾Ù±Û<ÎÈÃï,Uyžgx}KsHÂ~ì»:lì]+Çú¾ðóׯgUß/Öe'6 Bø’îÎÿ˜/Þ¾è#z¤\âe~JcÅý ÇW^ù…‹ã—ƒz‘ÎéoÔ™I1†Óz<†rRyw_æç¨êc¡h+©„iö±ÍkïjkŠÈÕ£¥Ûßüø†¼ýjÏ;´÷NTº %¼ žÆ|ç£<¡Ñk="†î¦¬ÒKоŒÚXür9*uºjø3Ðqææœ—Mû‘—Ò``2ô•hwàn¾Ê羆´ëÉ +ß_e¼&²pÝ× |ål f”ÖJìDx¶#¦­B©Öjïä›7ã7Ø^Uâ½ø ÅVBqLÒPÝ¢„Ñ Ò²‚2za[Š^>3£—´›/!X¯‹QGöµÊÑ®¦¸ÿ„/ï|~W9mˆï5©s-Óa”™Â §)twP[zÔ<ý4²¾Ø‡[¡ã½ÿÀ‘T||pé M|7ó“¼´Ærû^$ô˼bFGç{Øô»zaÔ‹²·¦}¶³U×Jç§7Íå‹»Ÿ€N箹ï{߉«§! ¦6gbûÜ@`Ñ層ßÓZ@Ý Þ½ ’‚C?›’ˆö¾ªe½šžh*¦l*AdÄOú²ÙÚi™Òü^d™Îî¤f 5ñqT‹ú¾sÃÄÛ‚åÀ¨{Ü®“ÎIZYæ0Ý3©Á¦Ü=ËÉ?›mü`‚Zè%[¢UÅ£ÍD»³à "Är¤Ø"Ü“Õ:;jÊL>„R(têŸï0ó\örnl ~Ìy(D+5>úÖ;°atý¢@5Çɽ¯Ùˆø€<5)üÁÜ‘ðÜû¢$lóA.éÙIýÈîÁTA['–xŒ zÒá pr§«ñ1“òðúÖ”Œ„D;G‰ ;h¨Òƒõ®¦>p&˜¼skú‰¥c–`®/5ÙúrÕ¹W?¯ÔñÆ©fô÷1V›€#sG çsv}b¡ØGBn~fØg–×+yô|ËŸ€3Ž[ÕyÓókékWÒ‘×íÔ샓òßÔøºØ ‘Þûä5áôé8í‰È¾upÖÔHÀÕA8Ø–án¨ÖÃçX"(“ßg6€4•éí’‡PnÕ®n˜}9&è„®fÔ/J9óKeÔ”/FÌNJ|%U2NÕÍòyÍ Þ>œs‡J_7è##Ã_T…ÝÚ ·¾óóç×+qÌ)Ôwéžß_V¨Juü"2½z'u°U­¦¤ßÛØ?š¿€¾cJâ±ÄoÝšM wŽÒ?Õrä„&GÚJê/‡¿h&ùÌÜù¾ endstream endobj 107 0 obj << /Type /FontDescriptor /FontName /HAJAQX+CMR6 /Flags 4 /FontBBox [-20 -250 1193 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 83 /XHeight 431 /CharSet (/four/three/two) /FontFile 106 0 R >> endobj 108 0 obj << /Length1 1613 /Length2 7937 /Length3 0 /Length 8998 /Filter /FlateDecode >> stream xÚwP\Û6]Ò‚¬tÃ"t#]"] ,,»°,ÝRÒ-”€tI§„t "¢tÉ·úÔû¼ÿ?ó}³3»ç\w^wœ™e~¤¥Ë-c³) n Ÿ@N]GÀÇ÷˜‡™YŒ€€þ@ñ˜ @pW0 *ör98ÈÄä-H5u ê€Bb@a1>>?Ÿè_Š0¸@ÞÒlPç¨Â  W²;¯T©æ|JÞ,ÃGR?R¶®:%!;lÞ›:ðÝf,Ìãbl4Ô¼èä*QÑË-ʉNű²>«.{ªün¶råÒLßiB„k Í{í&ëÒÚ]…Ä·RÞk¥f§Ö7Uø!H=빺TÜU#÷>GÉÙ*3Åúw-ñùÀ¿´>b;[y WÆaóLÐâÕ¼N¾SKðãÃhŒî… ««Õ—­sþ©°|µ:ÑS(5¬5)š*+dF‹OOkd« ú×{ØõìŠÆÍäHˆ<Ñ^Q³é¬{¥pu¹²s*/G‰¯;¡¼ç ’Ú–TÎjgaÖˆiaµ:^ÀQÂ"‹1êIéJ²ú`³B`Þgªh:j9‰ß·$%»æ6ö~Y Vd‡ˆù=¼«z¼aˆ—€I@§î~^¬¤qÔ1×GÅ5UÔÎëo1qι®ŠI¯ÉÿXd!Œˆæ6UQza½Ñn+¼1X4òƒnô~ÂBŸ€ðçv?ÁÛéc\O r:Êå:R{¼Šy;ËzAÚeµÀ=Xý1$4›OÔ¨­í.´ÖjöÜDk‡3½¯(˜‰_”^„€|o_j¦òN[WÑ„w|ºÞÛ3À- zJ°–hÖÜÔ&2Þú„]½¢+4:lý šÓ2FL‰øÈö«ÚD¢ ?M ©‘îùöC'­`ÛQê™vº4?õˆL­¢Á;þ‰ZÐ.fdMñtl7ƒðÛ—|g‚T”F¿Ñ> ÅOŒI(hÁªšÏK%G? ¢“F’uÐ2íŸ} &~F]ño¦ÆÞ1ÊAA£}»2~PDTÈ­*ß=é$…öx¬~ÿm+Îþ%ÍËŽ‚æµ2âqNùJ·Î9^òT´Êãîž>ΤÖj”ùç@¶÷KS¤÷{)"p›wÕ5ìØtQ#×áR3ìû-RâmÒêd#²ËåJ†§qò`¢=:Ÿ@«r^·í½®á¢<¦±ã׳è8»Ƈˆ¬îžc²}3©„£¡ÁI[þȦœáïÂçgŒžú(;=íN-±í§o¬ë™ÜÄš°tJÉçC#Á Ûú-Ä­óƈ_?Ùo[køÈcöœ¥&m1Œ6ö娫d!ÁåpOkGBË&^•ˆC‡7Êôâ®\*3¬uú¦,6C*òà‡Žã+qù¶WÃr&ˆùØT9ƒ‡}ç'»q”²’Šw>d}%_q]¨ &?yåß'¹-áåäâ¶zìÑ,ú,þBÍ/{no@HrðÕûãjA3mÓ›Ÿ£Óoïë{Ýê9_u!æ®›b#¾›ß¤ð ”l[ M¹–$†oS>šÑÌ0Ž€ç¢™¾Ô™¬È;Ùâ=“KLˆ†!/s_>‡Ý mŒ‡ ŽÝ˜ :ûî?£¢Í=CIÛvÌ2ÝÇ20þv•¤lt›sþ´ênâqš×–àÆ|h^›¡H`v¯]¢`%a‡štª› +£í©ÖÒ1åï6øÐ4¸Õ|rra›•\t‚g°Tzºñaƃ¨¼° Ôþ¦ÆùŸŸs)‡E’¾éèSYäϘÇgÿš{Óò¶’ªóM}Ÿ¿w”g´PgQ€SR(%¶á™¢Adò™œ‚¦§jØžƒ¢SQˆ=;á`;zíljð(‘†àtQ »¯|Öݰ{¢B­ 86ÅK†AR V›ËP å?}ÏÇH¤ {'¹Iòi¾ÊØ®‰¾ë0¸±:É«Ã%AÓ»^´©u x´kÐÖ’¶ª=…G¹Ïq ЦÂ0¤øðŒÐ–žõ)b]ºß-BêÉüwûç_V ÂÕÂý\pôŽÜäf7ˬÔ{IÅìrVº{ÂÏp{ 4¿’ƒÇ¶¶“J‰V0úNz9s Œrs ϺgªxæùìŠ0¦½¬ý]Œ³k“á£!î¸u"ÖC²Û¼¾¸…eà‡~®ãWÏÁw*º©™½Y¤w f›³†Ü °ÒÓÙ0òõÌŸžqç;JEü–1ȳ*nãÃþqÕGÖž ‡S |'ÎM÷GmÙڪؕSÒñ'tÌ®/jqã½Y1Ìåtf8sH§Ø i8Ó4kÖ¼Ñsv‘5K'ïíŦÎX$8ñdJ{wÀi 9ÙðLL ÉÛî®’X£ê I>8‚S\©?÷¬,ó™Þ¤Œ¦‡Ê}þLO¶×ËO€Ùßbîz'è&_nvÅ*µ¾Œ{ër¡¼ØÁî¾:-ÆÄ¯nc#£Âi¿œNš×Vp³eö|¿ÃæÄ€ƒŠ*{Í îòÜ…Æ Õt¬L™ÎîÃ|/ u¡³pnR~…XƼ¶ … jÛAØÇÏS¾}Óvdó¥˜xo~ [À;0|.b™dp9ÀÌ [pgÐç}T¾_ìHg/¨-wœWʾ™bÇ¢šÇtH_LN83̵òaÑA×”‚e¸¯è21Â1°Ç±8h˜Á.…„ÿеÕH&Kõq1¢Å­bE—þ°bƒq…Ð5ùÜ2<í•?ÉG½ëËÀ@,ê=#¢͸3ô°iãCºÍ…¥“IŒhõ{ßv𾑱nᘠCy+Óm"@'®1íSE/pwwƒ!xZ¾Ñ¥VÕ=/ñÈ@ç^G¶õéê5å—ºéè¬:W9„bƒETs¿ MnêCšÀyâÛ Å‰ïk±tâF¯[cB¾tçF+Á¢Açx ±sò\2Lû€ñ㴙˽"dkzúhÖˆÞ÷d㨊~£Ï­Rh¼ÏËlÁåDYVÎÙ'ÞÒ]Ys¢¤ªKBŽG—¶â ®ŒÖ>¾¦ÄÚ7G¨f—Q¸|HïÆV=ÉÉKvtõªeä,­~ TÞW³üj5¤ HÞŒHÏø|ºÙï™}µ)»v Éœƒ-;nØÙ>ˆxQ‘.>fÿ9úLDlËçem­Ä-º*‘B0¿ñ³Jùù>áÏ­t¢ÝgXU¿äœOl^3OÆ©b“PÎSL¥|®¹¾q®ŠM²ˆëÏÇmK/Uîô”Wâ”iC-QÎ2ܳ:R¾¾Ý”`žæ±ÜòB}Qá¥.{ }Ô®ëæèåÎ X³(ï7[üù«ÀýÁóå1”m_–Ù’–®éÆjëÇŽ,Óm󪹹øNcÇÆŸºŒ“ì}ÇLé²È7–GðDUbšœ~œÆã« ü¸þÚîÇM·9­tµÓg·M[¦Š^ü¶YÉ.IEùU´)¶Áe5áq, Ï û®q3°M@Œ-gØd]¯ð9=¶ë“κܦ•W=ÍR+xê¯ðvY«‡Ì¸3ôX0œäLÜWé.JªÒÍgiTHîxÜd4†©ÞÇ TN"¤ovÞ¸eäoY<_ž¡h’¹[ <†ôº¬–wÅU*|+ç*wªE[W(Þk•L™¾)V»/“ú‘º4yRá³>Q™#e0=¬QÀå¬u¦—ô2öúž­]ޏÊÏöåî©|†ôEÙéÙiy¯‰Z|µ}uI<…%®>ë¢Ùnß]¼û~ „/ 17%®¶«ÉF=)¬žòߢµE^›mš=;ÇЗ1®ôÔöQ0*¨ŽØò ¢êðã;d~ù@X?•FS$šÉPa¡°ñmIáD>hVfaƒ-%lÙº€G’ÓŒ­OäOæh"C‚ Ê[2å.“¼˜Ä”‰_KÍÅõgøwbcÎ’‘>p¯ÎîÓ ð3ÈæL¼&¼:rÇÙ_r\ýüöû£Nê×C~ž a£~Á¤²êwsL­_}}4ëà_#åÇBNleX˜Öä2VˆzL…Næ™àt;wym$ô±HáD8¿Nn“SìiÂÌ^DRÍêå’¿¢ ††])Gi¬e‹ù =et.¨$ùÝþâø—¾Å` ×ZÕ—M¬2ò—#Aír̃­Òòšœ b·¶m.c¥¯5*EÔ$åÛžôû¶Ìmz´…M¶¾—´ fÑ´ tu·ÌM5lø¶ªÞ#ëfïeÖ³¦ÔÕâ 5èêùGãÂ,CÚ(?ç>'%œ 3jÊ;¹5¢ žž´ µéÁÕæ]÷Í;™6o×3'¥\b¹ÏdžYN*L¿ö&SšSZõü°ƒÏ‹ÁPÛPT¥x~ Ã,©{ÂÅ‘SEõéŒhžqîùw؜ز%K¼ƒÜuÐ@Uß·êÓ§í¥ÕÃnïc1&è¼gYnW– ˆ˜œ6#äQ ß‹ãø²{«·9ͨÑ7¡ŸÂò^R³r€tì7¹‰bxÚ“—ˆÇØc©¿õ*·®Š#?ëó¼b£?wQGzÜk¢¸Ôir[KùDpöPïð'^¢µEÈ `žö³©+úȘ6Ñ6~cì`vÿÕ•´±ë¬#JÀ¹€£jë‘4o¥®fé8ô¹ª ƒÓÀýÍCo`^惘äw?1³69Ð1_ªÞ<=œOôÅ`ëȇvxÈ‘Í6YÐÌ;Ú•ÊI¾ðT=­¸±2>jVÝ)N¡>æ``y„86–]NzÃÂôÝGéò°(NÊÒNüº§”=8ZœÅ×=Lºû¢«'Çä,‡PWG¡€P vä¸Ãsû²wR#çA ‹´*ñ”nF´è)¨`ø´þÕ÷ê°Éñ(ÊÔ†ë¬Ì€%p{νPü“HÔÔŽöC:ŽÓŸ%èÞäll9c=×¥èé‡l$X­¥S¯$å"Srƒ%Ë*¤Ùñ8äVˆkx¸šuê3—ŽsD ”± ˜þ‰‹ÍêtÁsÃ3âŒúúÏÀþ`NéÒ` ¿¢)(§co3îþôÓL‚ Î×WN¶kT…Š” g˜zÒ5ó&+¬,n„ǧ_'‘€cŠY¾‚†k¬®Coƒt^«¿;üA‘V²¢ØHÝR¼í} \cÇo>ÁφoæZ6 Y²ŒGQQîÚ"~º3(Rj‰|ó%OäD÷ö×êXä” ˆP§$9÷°=Á>—¼^ý-D·\ ³¡¢ÅçÃ-©e²ÿæÄX¡ÝôPÞÆ†ï¾ÛcÎRúY9ë ÒS£iÂɨ_`§l¯©øQÃj.ÞˇÕ-“~QKAm{Ç>'ñ¯akš îˆu -÷Ïôª 'D§;q÷{J£M/ŹNwò£´EàFkÐÆÉÍ[½Jã,o´fæsF_”-@IÜüV²_GWE:@ÚÈ““åßI š£çàøEÌ7æF>–Z};œ:ªJz¢ëõÎÄJÐ}¬ÕÑoÔ¥%äÛ¨›;¥¼´ŽZn/ä;ñ¬ô×;륙fXŒž×¢­’2¸©­Ëý(jr:²ø!ˆ˜Å]‘™Ä=xÔÐY}Þöó¤ W;øŽ‚FRþòåf¯U¡Šæ@ÙÄû'Gƒ#ŸŒð}È7baÞÙ{´)IØÐ¬ÏZOf{º!ù'Q º;gfü1O=ÂÝg.eÄU¾LÍK4„ôÜG)•øÍõ.âÏï[N©½g!•JæFà«oÜfñ`U¤åº!¶¾k髲LÙ(즘Y¤;Æuí;‚’&CojÂPeóïh#–ù¾)2~yšU¡v§À!»ëô–ö³Ç½ü%k®Áj{®×Om[=™r>\VÈMñºÞíÔAÁ‰aŒM=ÖumŠ åœÀ>q(enïX½žÛé»l¾V­Îꌯ}QMÔ´A8tN³‰}®¼ŸZdñ‹D×7å”Õ’â‘°ñÙ¬K ¬ûSÑ`ÿÑ*V…d*ÈÆYÑòÒ¥¬(¾ÝûH¨.‹ßÁˆ¢Z¦ èg¸“Ý*%>®ü¡j™NK“CA€Ù|z#HÛ+‚ŒÄ lG >òût*>:f¢›yª ]>ÄL…ÕW¨Í0ƒ½ñåø€Ù@þømDyTlK&EØÔV³×Á£—â|”Å*Næ‹{ÅkŒÚu,Éè[³‰5ÍÖ”ìÂ’ðæf/»+k°¡»(²{ñ…â¨}¬Åý¾®§ <•iâa}pG8’8LÞ¼úýÛØf¼Ûõºþ[PQßMÀ¡R¥l´1îýSõ‘ˆìMºëè ÐÉÍfNàtPÂ×){˜Ž WˆÔRº{no“ÑOiÖúCÎ.'¯4Á é‡bÎ9S}G»8.­ú^£xü+8y?ýiX¤Gë5âÁѹ?/$ÐǪïÅ-œÂ¾vw>ç,}Ú¿˜©ØÓ‡õîåmôÖq ¸4õæÊôˆ>~ÏRÏ,¿Œ<æÎ뺗¥©Ü//¼Þ³ÀÝp¬å«›ÿºJe.á>–ÿë“ûvÁ鉮 lΛÚmÅg¹ :Ú]»Åº.@ψ‘"®í] íàKc¥¦ž,ô£)$ޝÓZnJXño¼cR¯ÒM+"Ž>;éõøŒ¿°OôE&¤±õ?¬"½uÏBhvÝ(–ç8‘OM“)£9êÖºóŸÉÇ/Û©y _‹:_ª[ú9 °¨;ñR‘ΈŒ}d{Ô¨vR¬ T0Gò$xŒsiÝ#+7ƒ×<è §¡Æq˜Á˜Ún—žŠÔó­,7G!ú~øŠ]I‹er8.%ÂEâ’x[]¢ê§Ô•Û¶€¿½°¥ƒÒÝ[ÔÙWa…ÃÓ«o­D>ù ïöì왓¥ö¢ÊüøðŽßURúþÉ1–… º93þZ1ßiqM5ÞfóT^Ã64—¾j#@ˆ¯?P€±•E(é´N@t¹rdúçéè 麡R¥¦þn•A°èÛÖ ÅhÆ#ýñUiƒè“ön8Æ%Hƒý ¢8cȰtt„ø$§ŇÎj5æt ZG—=Ý¿’<Ò¸HïK³*Ǫiuk6®zLæÌTÚx»&ö’?ŠÄæJY@ß ÝÞÛ¤|Óûx/èeK6Ë·èBŒºt÷ýº^ò¢—xÅÁ‚ãUÍ«e"%\Šo°[ÎiS‰‡è€š†ªgí©R G#„µ Ø`4×éÍV&h…š,”èkv6Ä_ÑõüÑxD¶c¶·]ú¤á;èq`2ímJNYäµöIÙ/½qâàg–ŸrPÃÍÌðjçX°lÙÓûÝz¥eûýÒ;ãg*šÍs¸Üº˜i®Î²»’mÎKOÉE–½{ ö¨»#ÓÎÒ^¸½ZÉÒ û­G ´§+ lÞÑ =y™ß™N#Ç©õ"ûÌ¿ÌJR[òâ©Vå¥O|‚±^»3 q?Ængá¹x,Ê3[¿´Ðž2]L‡×ÏÛ{^IÉnˆ<’ЦB1÷KÒÎR/-qͦä*Á­§ùr[LH‡Tpø½Çß a›æJ‘gGt¤ šÍrk2šå_¿pØe:åüð¤­.¯¾µO‹aËOO ¥õq‡2–×­ØÝNËšCn»yòãþ‡!'íjdEO¶ÊWR¢âˆ=Ý*#Ê_å®*w€€õªq ~-œ |ëÁVÔmöŽ¢†½g(7¯º•*'Çë—hÙJÕ ¾Ÿ˜³/A‹ø¯¢ð#ôª³Nsõ•^˜~k)iþŒ²æ¾Qœ¹«ñÞE(üUÃ[†8nŸÁ8¥z°ÝD¶g^£M×2 ´tpÝz¨ß¬ %D38ê©ÉÇüTS®°ÌR»;Wbv=‰µ 9äŸC æÜŒ@ÿzåD<Šî-¢‡+mÍ÷ÍÚ`Lþ"¢©öÕܾ™+v«Ðt {à€y÷^½‚ëbˆqu½÷#"Q~ù'ÂF"G;ó¤8M‡³Á¶‰roÛÍãæêªAúbxòecðtb†õÔ—.Ÿp?f[§7aQ2B™ëÙ´f•í ˜¶Ò‰@CàAÔ·5áádÆìý}<þg¦Jèt5=(?\³^̯eÆÈ—b¶=¢²ßL«3x|ƾÚßyj¥á0XÂ?A9ž,½«¹_M6“ø©º|ÅBXÈ uy¾ý®ä´{¬‹R ßø0ÓKl´‘\úˆùp#ªUAn¸½µ íNª ½ì@MAþXÖuP a  ‚ÿ‹ÚºëRÜD¼ò°ÆÎ©³°ä¡‘d¬Ñ¼öú‚Ùæ@LÓǼ!J8Xì–$†v/³üŒÇS”ª»©›¥ë|ølÎÅÑ6µ8D³B1fçwìåQžëA®‚Ùs9žøÂû<E°i§ ¬ô¤˜®QÖ'Rh–‘qBð¬”E0Ù«†>à²ÞLä¥|Øø×ÎÚGtr]šxdÚ£ƒ^ŸûúÅHf¾ß*¿#!YÚ ÛXüéŽñ'Ïè^˜‡Fë#ÝûTvyÊeœäxñ5ËÎS¶ÍŒWÝgl^†ùGÓÑ—¸(«yñìª|Äj/Ãò´ð¬œ÷8Ôë±H×oWÙsw<-È|ä¯ÇÚ \¢iZ$Ã?e}7=…ÃtµèQñ¾^´MÊdÒ€Z¾ ±®ÓsT¢wš‚Ƚ©»û~Ù¯ú¶y×Và¡6‹ïõœ0¡ÔÚ„M‘œË\nÈØWö…á':F©Ü<&ZCyŸõŸÎ]_ÃBí^(¬E*zÛ×÷0æp½t£OÝ%6<ôܾ¸£ŠÂ\p+YÆ;$TJÓ©''cg[¼Œ°¤ïðå‘ßmebß§|½ø®ñž7gUžÚZÇ?Œ¡_A’»4‹)¶Ö¨dDN~†*FÌ v’>f˃+ ‡ÎíÿZåÿn#àZôG4k3g€Â{nJÌ —½ð“BÂWLzaÅ4——š}CÚ2·²ÔNýW¯³ŠM÷*X§îm tæ’|*`^2ìm¯áÑ$aSìÙÛwѨ…œ.{©q?{–›®F ßùÖª9K—‰ü‡¿üÎQ3O¹R‹ œRP¿1ë=I®P¶ÏŠïí¹Öù¤ŒäLJy™XaÂÈ ƒÚH^4à„šL/Á-"½²6FœõJò6í"N!‡÷Ÿ­¿±^—.|þ‚ƒ.Ò:2‘÷ÆÿΤg”7˜ÁT?Ȉ£ß •£òÙ;ñ€Û0ï=rs n©-—dŒÒ“Öw ’•ç°ÓšJ2CPoYlõ;})¯Ø€¹lŒžæ/Ÿð?TE»‘S®E7²öŸ6~š¾5ë Î3ú1µ2(0\Œ¦&¤-Œ:íxéø8Êv† «[­'Œ;à ð1×õ|Bõ^ß·Ï;Œ…Ι v=¹Ã .©?>ÐõÉ¥ÆM²¡=´]ç±)+Á<K-”4D¶_T™SÙü¨\0¬?xSõ)¶ÇÐÊjHú=ß¾fºòƒîq£=ù×D‹a’¹—ze|Ó Òêà^í]Í}î:Éïl´a[/ Ð5’ÍxÊ=_SRü ÇIPR Vz™ågœ÷p²Ðy¥äkX˜À˜ có2¾°È°Ôr»øSC‡ËC·W¯Œ¼!ÅQs“ó²ãýÖ8"T< çÛVKw˜ZO¯ë•DoÐÉsòNŸe>Y‡vJaß§ð„D§–KîäÁ‡–c-Éø·*Q1b ¨àœ{µšÑðyÒ?«ËÎûŠÆ4mµe‘ Ç·Jv÷°+.ÅZuøº´¢eÕ4?ë"z¥²¡³Ótw("ûyoÔåžê*¿¿;È-µl»ú-Õ¹ò€eì×Ùç Ò.œ¨Û†±Ú,šo˜@å«y›ÄÇeo Òó¾°Ñl¿m‘×ú€mHçw#O†«¹Omé#ìxÕÔ­RkÙ—“éLüÐ}°‡?K@Ü Œq˜‰¿úÔß¡÷êµW [­<õ[Û+>“¼¡áˆÇ†׫:N¦iAˆBç5ætn¢ËF7ëpuýó.tÍi‰1»ßcˆÌª혻ˆüðyÏ’ƒ¾~¡ ?U/ÑÆ2²ÝìÍÁ™Rthö3ðì(ö؇py7/›Ì 7äì‡â?¡VzÌTzzlqBéÈ,ÅÈxx.:,¨õ5N©½Z}¨¯n2Ú¥ .l6ds™•g†u¡c4³L”I —¼XX×míÓ£L7Qãq.AUB†6ÀNð¨êßàÆÂýÚË+?¥—|úùÝKœe endstream endobj 109 0 obj << /Type /FontDescriptor /FontName /FFGBUJ+CMR8 /Flags 4 /FontBBox [-36 -250 1070 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 76 /XHeight 431 /CharSet (/eight/equal/five/four/nine/one/parenleft/parenright/plus/seven/six/three/two/zero) /FontFile 108 0 R >> endobj 110 0 obj << /Length1 1619 /Length2 7680 /Length3 0 /Length 8764 /Filter /FlateDecode >> stream xÚ¸T”k6LI H#= ´€twwI Ãà 14HKK7‚ ݈¤„´t Ò"ñÏyßóþÿZß·f­™g_»î½÷µï™5Ì Úz\26pk°"†àâãæÈiè™ðñxy¸yyùq™™õ!(øo—ÙìêÃDÿËBÎ D 1y i¨‡TÝ¡>Ÿ(Ÿ°(//€Ÿ—WäoC¸«(@è±hpTá0°.³ÜÙÛbg@æùûÀbð‰ˆ?þíq»B@@@ˆ°;!3‚€P€#¼ÿ‚MÜpåáñôôä:¹qÃ]í$Ù>>!€°° ÀÿßqþéÀßÕÿFµ?§û¯ˆ*0[8@ä¯"Ýû»?Ì`û³6ì€gЄ#ù °ý‡þÏxyAÈ7¾ÿç%øíòÿÇý_Qþ¯ôÿß)ºC¡¿õlüô@'Ôû’ÏîänhÀ‘û_S#ð_ ­¶¸;ý¯VDîˆ ÌÉs.¾'ܼOþÂ!nŠ/°6²ÿ‹K™ µán_÷Ò‹—÷tÈÕ9"ï7äÈ~«ÀÈÍúw^nókù…@WW 7.’HIàˇÜU°×oŠx¸apÒ€¬Ñ` wÅý5XaÊ/è·Ä/à:;»Â½À.îÈ‘ü ­±áž õ‰º!qsDÑþœ—Àcí ¡`[ÄÁà‡áð óA<€P0 IŸ¿ñ'" ¹Ø$+þyJÓž-}11ãGÍqx9êõ¬3;h)U9¢Ð»&{Ì•7ðã´ÓLg§²N‹\#:JС}`e†aÆšd4s&Ü%°UµYpÓî¶e*²p2¤R#berÚ÷ТM˜Vƒ&Ï_»š¥»ë«ŒÝ EC^^ȶÆPþð^3{™1¸¸ø1»¯7Mà$_­»Z¨ò %©*CÎPb?%YB¶j£ ‹¬à¹\iJ<#e¡Ó –%t+ᥟSù}ïúÜëí4å‚îk=np¯ƒ ÎHät-ÐX®géÿàÁØ”iÿ¸&I–¹Úå|0ý¨ZšÛANÀ¬k_¸Ó˜‚—\˜¨£Í7Ä`=°Øxi›™ÀÇož"}Úÿõ¬6”ÎyÈÖhþZ<@BÅL¢­‚®Éà°â>ôè1÷÷7½rcú$ØÙvÓÇÖÁ®ðc8Ó!-ÙdP¼q[æ^ÏnB£Šàù'"C!D­Úgx‡%{\NÜT€nn†öûv%1Si‘ÙqÆ8CyúDŸõÕ·PÒ5qLDÃd§ HKi«æ+„†_K7I¤< ä}¾0~Œ*¯³GX»v_§ëÇ=‹xk×c~l µÜvS{«bMVUÄ“xHÃ^üõ¶¬ÉdбðÌR®ì­A€%IÉóg¹¶±?_~ß:«?Ê5–©‹0…F&wL¬Y¾¦m3í{I“þa|¿»5eËK¹þêã˜ÄzPI´Ÿ1´ZÞ'Ø"·ØGi`ÖTo=kÂví¶m8ûÇgv!öd=ÈÙb»^óÈùg™‹¬<€øël˜ˆÀZ¥F ‹ˆ/^„a€<þ÷§ÄøÀñú`þ7нmî:TJmž<ÏŒ)º)ЍîJÉùø¼›f‘¤¥–†Gµßt^¬øô§‡¶À—¯<ÆZB½eേÏ÷¶Ì\¨åƒ5š1#Lž,Ÿ¾h"Ê7‰›Î7ÖV8V¿á_³äo¾_Ù\ WNñÅå£ðš¤$Øeíhu§ô{Ç¿<¿2Ò¤§ÕÍTH>):ÈhŽ›B;ûÉ’¼–¥W¼# N„¦kÏWE§˜cý­¨rÌŸ3Š/wi(*\èbÃ¥ñ°†‡pÚŽ=ÌÖØ; ¯ó¿€JÑ€8–"+-t+6ÍËœ^¬~‚‡ª•i L¾dP¹óvêíÏ­‡ßu‰n-SFbÃßÒûÝ¥Ù•!8†‹I¦î¼Îöõ[ûþ]Ç(°¡XEOi?}ê…QAU0‰HT6’)éª\«c˜7\¿Å—k$ 1„×¥-x´äì'JuÕ%Ó‹%g×J }‘|Ú·V\ÞþŒ±üGCJ&Å„îÁ=]¹”ÙBŒly{\J‚)iËÇ9ýäCÙÚ‰Éc­m¡[2*—Úheužs™$E6*Ê>Žl  ¯ŸñtQAûzƒÚ*u¶¾j£¥#nTax‡#&Õ­y<ûqcfyÍ?géšÔ™µò¤÷2”C^$‹/i³¾T}3´à‘0ý¸é£½A±†ÔYÝžÔJ’3ÇL¨°Ág ‰m>7ÿ5¥IÒš‰‚Ç)‘úuOm[^¨¼uŠÂCšµØCßáWð OmN2„á„*á\%êú¼!9=µÝñ · ’·ÿr&¾žï›‘±*›2´;ãõ¡o.Çß–UîÔ{iµÆ²¿jªJÚD÷4©F%ìÂB¹‰ªqýÄQK‚âà¶\zÆXòúxQ®½D±éˆö)ƒƒ@ÛÝlžsÿâªqtv{êü}°Åg¦F–˜Ós¢§Ð”¸õF·f–ÐÐNžøÕU öLýž@ þ)ÚJÅáÜxÝ yP›cžXÜvʵ»Š^$®ŒàI†¬ÑB¾l!C »2#®v¬•1l⢔Ĝs„+Àß7b¢o@:w*•3š&B€zS Ž:ÃÖÞM}ö-Ã×, F9²gvûQ¶sœ@iþ¼¢Y@¼¾Qa8¿>ÚÒ,Í4?Û–š~Œ‡vpbȃ&¥Û %ÜiExtYcêî ׉Äë,еtœ/p:ã©/š]….Z;Å ëLS·;úÖ=féDñ¼‰^’×+'z;‹QÉ|^$9Pðe‹놚W,fYû(ÉI12û =¶kSG„b‡ uµ·™6›EïÚ>±´½Tx4»]hÃZѽÑAÍÕÜ€÷‘…‡ºÀüí¥ø\uÌì <§¡°jÓkªž|u‚§cÎæbAC§õcpÒ9´Š¨,uåÂO-j_ÎNrØÓ¿c¹äp~^™x§ô޼/ÌÍBÅ%±Ô޹±â’®Ãâ%6ļ¡f=Ë«?/Eè• <°‘Z¬5&à a 1 ë¹pÔ4ª7ùp 8+¾-ë_—ºãX·Ðµ¬{À6?=fÁ?ËLÈ‹&s}³ïú%fôÁÔÏ)>õq¿ô ¥ÎF Yehž÷+¿¤¸yzâ#Îçñ¹t=É õ —ŸÞΡ'jôí}æ©<§%ßC̆¡_+ƒ3_Ìi˽ÛñÐ<ÜbTòÇ+ÚÇ?ÄÊÍ1¯çË“¡ˆÿ“Ìé·UöÞl8ým9J©¤*¬fKmDê³äšky ñS‰Ñ—µìÉÐ c @[ÇÃî”ÚG”><_êæ|&Õ»MCýš#åQéúžp=ø±Â«pàÜEeªwÕt*Ó¿^Q÷q…r¶If=¹¹³ m¯U¾É3CÑòrÕȧ ÓŽäx÷ .+²-jGñtœHw©ÌÇ.Œ]aŒï> f⥽ zWÞm(é &Žô9ÝzÚ×ë×Q(isçà­ŸÔ]oèÔ¿™¶BéRÏ%AL®pP ‡‚à£4›ðoð—©¾ære’L·èŠ8>v ãoà-(S÷¢>qiþ :)©„Ÿ<ëUóÔ±[ŒuÎæà³ !枉Šx5ÚµW‹§nùüìãÝõ¥>)8ŠoÎLºÏÍXdÿ¾Ô²J§B@µîüá¼÷%Ý –["š 7·qH:ÍäHû·Mš|ý¼”¢s>µÜ|Çòòá(‡Å€ò|lj¯E™Rí ½KdcܰÔHg‡Ù'tê¡êÛæYãªü.„Š7oOûnŒ©G‘½fI hCä)| Ntv®Vä(/ù^»”îqäª5e3 †Ý/ šD· ᨦÞ7ýÞ¾%Ùßš)Ê1²¤Nk‹6KÖ-’ÈLtã¤õ%{ÏŽÃ>>/“h×]£Æ±d¬Ö…¶`ØÀê3gz:,C;ÏŽ;<»_7­ßšLÔ¿ >ºL ºwt²c!)¹1+:uõ¥‰¿‹áé’Aø•_Jê#Ì‚†¡{¶Q».ÚAþד~w1™Èß3]‚Z~ÚݘÛn™"ߊé^ú¯o$qàÒrÚ9Ë>®yÜ)áçPÎ/H$7,œõ}/ñ(µº>m–†0€{‡G±„šz yˆ7?¡›½Ò…ÆÍHã Æ³§¢ì Ð| ^Ÿqr>9T ‡¿“®©]M º³<Þ•™ (^Õ!Ûé¥3<{Ît,¢ßBï m.(ß~åëy¯Úò ÀE¡â¡ÝTGÝ“W²ÛëYû2%åÄ+m.oÇŽ‡Ð/XšxRE^?~ÈW¥mûðNÅéã½Þ\¡³µÖ{4¯] ŠÍ;=¬¾âµÖ×þŒ¥8ëðºPbºË¥-ÜãöÏk½?œÜXBñe¬]q$#TS›™,ñ°²Ó–•žc¨­Œ“g˳'R÷i ›ôÜ\t¦>¨©Mº[“EÅ)-gñ¤1ÃG§ð·AéQ(9Ñ㤡œòöå«§Ÿ)¦î‰{_fkºã“±v KÚz¹‹úŽÐšï {" üo ›eTÌÃÙ)4.4èåoÞ™¤ÎÁê©‚c113â¦YJ8,^Òö'°^ƒ¤Q¬>¼fið_í-H»h§› :¨!¾Û+ãLT6ˆç=÷fuÚwÏ0Ø qf°lG+Öiÿ¾ÎZ¤™¥ôîhå^º$ZoÉÈówİScIÕ<©9ùƒ@ ÍË0Ô6k>m¬¯ …µ ›$ãÜSŽíèöÇoKøjÔ%Ê@ài¤òZ~œ¼çjó¥yA ^u<ÙžÍk_…D²ÁÂîhÏ 9Ê• "n&2Õ-{*Dö¹#“£#“ù“½j2ŒcmSÞU5á{9 £Té“¡ÓŸ¯ÎÓœé¡:Õ¯“t63"1¬WJÈxœOQëO_Óè'¹´+N<Ű¼–!Jì€í‘è|ºËwR{éËâ#&>Ý‘ƒät²f#·—ýæ+ù‹q >*ì!N=ãxƒ-UÊïDÉž34f{Ö\}²²šéýˆm­ÚŠyÞ·Œš°˜¡U‡.›‡—ÅmÙ«ùÌvK¶œ³bÚ5úHaùÈ 11o§¥F«ó†éŠé­)Õ×àÏÙR»6Ò vBs2ªT‚Ýišif ešnªŒÂ19›7+é÷\Û(køHnr‰¨;"@an´ˆ”%Ö‹U¹g©-Îì•ì瘷ٖô™DùìµÓlÍWY…õ÷54ZpëëBñ]újUõýPÌjšlkSf[Yô]ÕëeZk¾¤úÕ÷Êý¾Mž!ýº&™”q_3)–|JU–Ÿ;1¤’r>ÁòžHk&Õ^bV”xÜÕß'Žy>wž§¤6«m_÷\óY†Ê ånD ö±>¹ÏÒ¤-ÎŽøTjrÌuªÕä·²ߌ*këÍÊ"“•¿.æƒ`¬ù€g%D«šü޶ùÀS¸»ª°À«ÂÝw'+»gj F¾ÙSKRŸãbK5N̈QŠ‘yÐLJ¦ß3^»*KŸˆYlLÇ­½¢"&•Ë _à¾ßèF¾~n`¾^’(1au{ÿ&Qð݇¨@×@ „¤·Dõ ޵ k9ôöçømˆ^j×Ú<(kÃeoQÊõ)ê³³‹ÙÜõ ´½<í$aâÚ¹»’˜šÝ“ùÓ‰X:±FwvÆ:±«/z)©]mCÇ×ßåž×Iu¨†¼çITd2à®á•ÜÂD;ãü¢¦3ÊKÉ88U>!±ôÍmgé Ä ®j¾ò¹ŒaØóZHk‘Û?:'±7 Htœo¸:TZÇYATØXQùášV8ªÎÀ-8³j¦´Ï{kc°' K?7FNðZÉÕnÕ³‡Pp˜).<-†Å¯&®¾# ûyÊ¢RÙúìÓœ´]ü¼jL e6 DqÈÚ2ÿÛX>ìÛA–ù¨y¯Ú°ÑÃ0­Ùª'î˜7/+qtÍÚ)ÆÓášâ6±oYƒ·zZù*If >ˆùÓ?8{¦ð 4Ã&: Ž«•:{±Fbšñ„N-Â.£´w» «të‰tL¶~E2`šxÎÈܧæg¥£æÿ5ÿãÚ‰»ü‹ì¤HÿS–À›ž§"Ü73Åï·¬©†Kâe<¢¦ú–:`5N±!áT®=Ö\ÇÄΰ‰yêÛk~ãÔüŠ'œò~ß7œ÷Ÿ_’ÚPyPF› ½À4I=¿¥÷„;“œíd0d(ôh¿V”‚?ëqÐ~ ²zRɘÕi—$õ*f´wNûð噘¬…NœVi’ÀÙûu!î°÷èLË*}»w¹–Z˜à[«O¸ôéL–­‚ý¹DÈtxz 74êÀ9R³ý»“5õ¨t„ûöëèæQEÖÇ87;ïwË-wùÅ'.GRû81÷4]e5fcÙµ”ªy·¥ZÒÌ *ÁK;øˆ¢úYÌÓϨšy¬×&ºO4 ÉyïF9U6øD$mSýK„ã—= gˆð†ò»–<‡²EÝ~ŒÅï6ÏãgpsûÌȦԣìsÑ^š“Õ±ÏÂ…É–o“1`)'>Cy\á5êwæVíæuP¨œ#®¨F´¶>{oäÀ¾ø~³©ÐøÃ0§ÇKöz\F¦Äq裪Yk.%JÖ¿xjn¼ªBËD#™.µ—µ×C=»-Ç#x£Ô–"LJ-6&~ع܊ðÖÚ*ô±‰93JÜÚ‚>sg9*¹LË í­Eáhýø¶gE<ˆ]윘îÛ´\¢"D!—ªv#‡ÎãYóʉ«üË Þ'¦*Ÿtrêgù§ Mä^éž8_Û‹hÛ–²î±à?g ]WH’yá× f…ñ Ïã'ó%ý|µ¾ÙhÓü(ŒŨ”f@å»\ŽêuŒððšu·ÑI w7õ¶$}«oØTöÆšs«Së…ù,´Úë+«çÉšñ[9Ây5:-,ÊÎpÛ,6;iOsÛ"ÃŒ¡.³Ý€ç{1|Š?vè,ÈxTàô¯jeyLVÓ4ð£zàáú\mè7-] ñÎ-’Z?:ÏB5ËòñKœfÂóU,8N ¦ÙèÊì[±õ†IûÆKÜó6’ì8ŽX—¼J¨Þò–Þì½k6X¹Âþý—B·²X?È7õXÜ[ÖZ¸ÊÀâ×r|Oº4üÓQ 6—MÆB›ûúî3”®´MÆô€ V÷ÃA?2ÉŸ¼ïb)º[ßXLW%?dŸÎàòmÔŠk®Ñš¹â'Ô-éz÷Àëãö6}„N/ö™§¬ÝxÂjëÞÍûœ,ÆÏeÐ,ëãs±æÚtïég ŸïuíF:»–8jà5JcJwY>¼HI‡vF"ô.6{˜iƒØ·yœuû/—æ>”1>ä¯wÇMW' ^röÌÕ¸›ž°Ra“Ö6Oÿºe¡ó–êì¸FJœëÐ{ˆ§¤òÁò D!\ \}ÎÌM¢æóL“4/*tëzí‚\aëñØ’HBƒNÂNʤ ê7üZ :Ie0Ç?Ñ$9Àñùcャ–×!Z7ß'žphùülá5:€’ËMX®—¿ÌÎŽ‘ÄíïJì '›<Ìxw~ùñx÷±Œ lPeY1*…'[êz8*cßèÆýN“cë­~K`B$þn¤AJBŸn’ê+ã¶Ó2€®ÂõµkÙ‰‡Í¢¦¢gÍþ±üÒ«,Ò.Ë’®mÜïo4$+éVœ;sFRp ±Àò*š–”å2¯Tº­‰j/Ï`Þž^Y”‡!½­‚=±ES\Ÿ& $è5Þ¯nÓ¯ð.õˆïQšþØIœãø¾§Q+ŸW½³$-^üÖ+Ã0ûœFà…ðL$”ï-@r®#—tÕ>ûDIý“ÂÃeôPöçg§Á_|N+·éwz:7ä¡\‰M‰ô}‘”4ÏÏbà°§ö‘ç²Áp²}Û2¢-½q}OO²wýøq-ÞÇžž·Ši[Ò¶,ˆ•‰·åÓtÂô×^·HÁqüJü¹õ~š* ¼~—Ü,z·ý& Íª¥´Ûkªó ä숚{ÅòSÞÉß‚÷£×þ%Bø€’íCÅz¶^óu ¶­JO‡RãQ¦$ÇJÌÛ”`á^{™P0]O~Ÿ¿«6ÀY´»&+m³/Yã­uû#| <¦‰ÏXÍ7rÿ¥íŸsWÀEuhÀ¹feŠ&u²uÃBZ+Y¨Ñd¨óŒ@‘çÒ¥ SÓÓ*ï¾ê´7²‡0TÚož^ÁËW»æä«ùW4 T¾ŠÚûœ¯såoxw³¦ï›ž8ݧ¬Ô¸òÿ \PŸÏˆ”¹w®^ÆŸhÒ)Cˆ¨U.Hœv¨î>f‹à56‰µ]ÛQ‘¸l+Ûa”\ä”a)×Uén{ÞŒtæt~%}6<+ª[Í#o¥¯¤dlWFýX:V@w6UŽò!¸à£¥W— ßX·ØR20 OH¬Æ=&Gáñu0%ø\O›,å–íK:£¼¾zU&ÿe,ô%3çî*¦K]ðͬ“¨§AµvWú5Ö}3’jÖÜñŸ‹Mƒ.\ãYá´Þ´mŽ–8?<^™‘O1¶~9OMȺPÅ—…pÚq’?Æ/æÔjÓ¾øˆÉ6îE/ÓAø%"ÞiZºÌKÐÒ–tU0ß@ÓblÿmÆ$ÙòÇ õñæ§PñøS2¢}€»„@;÷%Æ”6‹í™N‰Ó½K¨å¿.Ñ,ÈŽ1åÀüL¥…¸Õtr™é ‚ ¤ê›ˆhžO>7~ÚVÓ¢ñüþÅ $(ô#­tw;mò{±¯Qñù¦÷L¾sÏŽr¿¿Ä ÁP™AçÄ”„HÄÞ Œz@z•11{U·Ë1Þe× o뀛”;"÷ÛîÐO÷&¢lS¦¾VŽ|ôÚiÎä£qt’°¥Ã;ÍCÏIŠçMÜ(ÁÄ>•Ò…³2‹ 8Ö™ÞÄ=„SËœ™ýy)ÏÄq`—¤ Ä&8JÃsU‹‡€ª«F^ 9šN ¦r5æ¼6ö1ýÒ%"}™éÊzqŽÑny޾k9×–î+­úáѱÀ¦Dhcâ÷öiéŠX¾¦Ð½ðm¶wûÐïÆp’¾ó Ì”¾ä±ò©\ßcJPÅc¯èN½ødzST~Ë7/HÛú§ãöy¾™£Òl¶0]¼¢±x©Ë]òÓË_g|ä“üüDLî=¦¾½>*PÌG:›bñ™ÜHÚŒþ?× ”&:á+íƒÉkrE±ju…‡ÕÔ¼ë±òÚù‹ ¨CÊ8¯ÃÄ-”\!Î_ï%>£°[IODu"à \¨aá°:. }ÄÝ¿@xá`Ø·þÚR<Úì*Ó% ¥ís.e_Ô^^ÆK¸õõÛ΄œ ˜ £ ž©“-íNõJ ýc+Ì6‘*Ù.* OésŸœ¤Ê}l>Þ´OúôÔ"åÚxf:4HÏÖ œ¹zàNŸ …Ù½Ú%{Æ1öÆzÜÀ´ S™Í2ƇV›CDhWÀ1ëoú‰Þß<ímóp>Ãmj0£^‹aµ_‡–"ºcœ^ÿ’Í¢µÀö´¿v>Ùí\ŽiÆÅßÜc²K)ãm«ê· * ەŨÙQÞ”W‹ùlœ¿j/á`ÕÑi‹¦ºxÝt¤½ ",…Ú³6Yct¢‡øyEˆîö¶eƒ”Wyh¹œ¹®fŽÏ6Ü@\B6³A¢˜‚ØÝÑ"aÙø1oŽû2ZE+Ž¢,°÷þ#Þ~ endstream endobj 111 0 obj << /Type /FontDescriptor /FontName /UHBOYQ+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/I/approxequal/arrowright/asteriskmath/braceleft/braceright/equivalence/infinity/minus/multiply/radical) /FontFile 110 0 R >> endobj 112 0 obj << /Length1 1452 /Length2 6412 /Length3 0 /Length 7402 /Filter /FlateDecode >> stream xÚuP“[Û-] ¤(¡w½ŠôŽéMC!@HBï‘¢é ˆH“Ò‹D@ŠT¥Wÿ¨ç|ç?ß½3÷Nf’w¯§ì½ö³Önv£»Â*P”L…Ä ‹‰€äjw­d „$àæ6…c]aÁnsG!åÿW‚Æâ0u0—g€Bu=\b@1iy1y(ÉýˆBËÕÁžp(Ð@¨‹BÂ0n5”›îè„Åmó÷#“““ú]TAÀÐp 4c`ÜŽ°+ð. ‡a}þÕ‚OÑ ‹u“õòò#0"(´£¿Ð ŽušÀ00´' üExŒ€ýa&àš:Á1ð»¨X/0Ä®p ‰ÁUx ¡04·9ð®Ž>ÐÐ †ü“¬ÿ'Aø×ÝÅDÄþÓî¯ê_àÈßÅ`…p#}àHGà¸+ h¨©/‚õÆ ÁHè¯D°+…«{‚á®`\Âš*Æ@0Žà_ô04Ü ‹ÁÀ]QýÕwËH¨ €!±À¯ó©ÃÑ0îÚ}DÿLÖ‰òBúýµxGBü"õp5CÂÝ=`:ê¥à À?˜# ”ÉIKKÊaî@˜7ÄIôW{S7Øï Ø/Ç ÀÏ å|€# €?€á~~°' ˆE{Àüþwàß+€˜ ‡`0G8ðOw {ðg>î ´á´'ýúüçÉ'/( éêóOúïùŠêš››š› þaüŸ˜ª*Êè', ËIIÅÄd¥€22rÀ€·ùÏüMþ7j†ÿu8Ð?uP@¹?p—÷7Ï¿dÁ÷—eøÿÞá §eïéÛ‚¤@Ü—Øÿ·~—üßtÿ«ËÿKúÿ} MW×ßa¾ßñÿ# FÀ]}þJÀIÙ‹³… gä§ZÀþXÙ…{ þ;ªƒãì¡‚tÄI\XLR$ù‡c4áÞ0¨ qú#¤¿GÛÃŽ„¡0ð_o\ô_1œë .¸· 7°ß!ÎTÿÞW AA¹O\JF£Á>Ndâ8ø‰ál …yÿÖ7PT‰ÂâJ€8ŽÀ(4à×XÅÄÅ€¢P°£#Ž6.ðÃéBŽÄ¹ Ž›Åß (Š€#=0ÿdÉEÝÐpÜÅýBþu(ˆ³ìoíàNü÷ú÷ûó†A3“(ˆB¸sMxÓq• ‹—ð×A’Ï_Zb­º¢¥° Ö"wö½ëJ©SÍ£ôÝܬ¯¸µ˜1‘ÇbLÇiŸÄÆ,ÞJ<‘a€’ón°m»CüîKWuç~f±°Ã¢²¶¤îµfžç_»‹loÅ5xÎ9g½ˆ¯ÿu™êßî'K½Õ-úeœÃº–?/Eê2áæÇ&€FûÁ…á Ûß|G:¦ÉŸü@ûVnì¸ãß…Ì*R4L οˆ£¼ç ¾T¶Š]ø(.´nö&˜¸£Ú'ȱ]…¦¨%ŠO•$ûÖ÷AÍã¢Ã‚nÐ(Ðw1¨uš³©}¶Š‰8¡¯;ëHÑ¢\UƒŸ¥öÒû)CŸt"&2—ýóË¢eÖ{ÝÁ~>PØøˆ½Ó‘ÔR ©å4-z-ò„©Ðφ)Õhn/lµÑ\k©Ÿ&4O÷#uµ¢×jÈw¤\¶€³A0†ÈFñÚõëvvØuL>a®±Ç‰ßSÝ€’'¤¹4e3OÙ‹WÝJWà JH¤Ì=„ð”bàø6Ž!fÊNR0Û†Àc½<ã"i4™õ§‡œç=ý¾Ý§¦ûìŠQ> vªÓ¹¯½ÏúAÅ–7›ݹ¬[Ùr û¤³Eŵ _^z^ÕÎH hKD½ˆšç ýBhß,=9lÑ?ÔšB³Eak…ønerû…Oc&»P Ê›D½©7{à0ù³è ®Æ†P›@~H‹¹6hwÞIÜ„W®Ï”I¶_ɹ3^·JaRà;B.fs¤ ¹âf¿VwgÇí¬Ä]UHCÉ;s" ë#2P—e=[.`ö©Be¼‘ωbŠ£æP_*Ð"8ìe¶óâ5fX¨m}0<7F\¬K‚T´ö–bœ?„,o./D¦™†°UVì¾Aìò$ÔÔtÛmD"-ÂB¯÷@ò—_Hñ4]“æ¸ø9õxýP}ªâð Tx4=Zmõý W÷tßlEÎu%¯“óž`ë(ðÛº¯²KmF˜âYyþã8ŽçÒƒÒœo¼Ú‹›s¸f¥Ïu0,ň.YÓwzá\$üI cía|—dV»Ä^Ê‹-N6fÉÝ ÍW9#[÷ÃÉ õþv‘…ÌkF-"êïyH "éKS™oå"ô*º}}»*†A…Ïuú²,žºÜ°¤Ny¢4–¨!~mvR'ËîµÉrO«`ob‡~úù«yb˜«,†-Rp  þ\áÝEkÊäÓLÍd"X¶v$ï¡1Õug9Ûv/÷©gÏæ¨Ÿ.ºÃ¼’"XWÀ;qª"347—8ÈùM:¸ÄÞå,I~‰yVñ¢"M¾ÝÒãbl¹´jåÝß:MÖÙF†¼£æÂßÜàwZ……2F.…¾¿² eIä["Ró4ºÉ#îZÀúð‡ƒ¸à ½‹\UÞY/­°næ¨ûõèd”•}º„›³í’=}ç|kÛ>$ô°ŠT£öQD7ßNMcš¬é©V6{…ÝŒ+j=º ŒRcð? CGÞv–jvBˆËzŒÔAKžÕ÷Ò¾3΋u¿~ÖȨ¡s˜"«ÆÕAd’#}¦˜i§3À– -Æ©†ÏønÆö_Ræzƺ´¹CYŒeN žßÅ÷IíšüÞ$•+® Søßp\cÚг¹‰*v‹Øšz˜_4ñÍšñ£é° !·‘jRÄž¢›·ê&pƒˆåAžªvšëš³g‘š–·¯¸†uFX½.¤Ò! >‹v}ýÍG FüêøëQÝO°HhÓ†´ÚÈ<_R…õÑé}¹ñ{ÊÔLVîJσ˜¼;‰nˆæ¤¬å¥~’Í®,¹I¡”¹xº\Vr÷pa0/íuF«,“ó÷P‹óŠWì,9›#ʇ¢Û4ìÑÝîœÅ–]Â-ZOa??Û'jƒóà¡,kå¾ ‚•0¶b‡Æ‘À¸ÜÄñ;s[ß$;#†`) íºÅclVïÂflRë}ºUmQ¡BåðýŸÙåe‘ö³4ª²Ï¥¨7fé 8+re™3‡-Úе­ïh=<Ô4 ÃÛ¦|#Jé§"ÒËNWâæ%æÊôQœ6¼ tKmn­áßU*q›ÕX*xx“Z2éãDÓ=N…§‹«»ÚƒÒsçŽåç…¦§7–G·¢|’\XQ±/’cë›ñtògoe£††¢^úC‹5NÛ7yã–’Êìï÷(?7'—˜†]!O)/–uL¼ Ä,×MÖ«ÏÉäŒ TÏèñ^cÿ6yÿGpë‰AZ”íj}¾,j=ü³hÛeÓ­§©¥–èÄêãàÜx š"úËO¾Jçý¯‡Âµ[“2 m¡$O -^õÍ–çbÔDÆ2”«ÌN›ª \h>>wÜûuèüjÅÑO ¯Î i=eû¹Ø·ÊE0\/5ªhÖ²È8Z>ªx“ÂÄ>Ï ÷‡wÑß+§ÐÓ:©K~ï—ͱn3¢q•lïD ´u®Ëã¡aG2r…\ì™+Ùyí¶\±ç(_âéÈS÷Í{8íZ2yÁ…puòº?‡”µ|¬Tn¹G¿´$Í‘Ñ3¾ý…ß~¦ÔßðˆoYRœm ié"çö]£æ–!|A´ÎGJPv5[ZÏÂ>ZÚ³dW³¸Â<9’*ZÚÑÐ[”^ËbâÝa䢼­.øA˜õP:G?63‘@iL‘«êÛÇ¥ýÒ=@Ëìašã²÷ð¥þlíೆé–âN¿çÙ-ÕgÈé}tQ€ rÅH‘ôí|Æ—£lÛás}{üFô‡výk…Yo*š‘_¢¿2Þx?v€Ïòyª««dŸÁTÕm| ‚Âø˜ÈH1©SèÊžVpoØ:tϯ5}ìjʬ?†­‘±{x–bRýJh1Gl©±3V–Ým/µV‚ñò©ÔÓ†¸ë^©hâ%=çø ºQG—MÖr®”Q0uLj²º¬Hé5Öô̟óÏäÂtkpù3qÙx…QóJÚ³†Á†Jm–óÉ£¬WhÔÕ?½êºý&½¨ö¹'[ãa'ãMÑôògí1«f[_£Ÿ‘"ðú°$¾ú”³Ð§‰à¦LºjË®ÍÅÏ?Ï$ªs©Ð˜N‚¯UÚû¤É?@ù˜ z½W¥LX)$‹lCGš†Úœ6fHÒѲÞc䄟^xZ©Þ°T˜Ü¼ãœãÙåB.aº4Û_@pí'ÃCycÛ^ÿYŠ¥èfa•yÆp|ŠÈÝè¼% ž:0Á¼caÄ(˜¿•š†{Ctoå«úµ·†­‰¯Ö*u¥zäÞ m’~ã(tu©¡ÁÀÌ­«ŸY¸ZäcQ"_‘_s).'Þö¯. Ú÷Þ±ÒέZˆ¨M OÍM”à”jÔGѹ˜qZÄCh“é›C• s Ÿ‘|ª~ZÙÚ'Ìé¶Ó=°Y‘¦[B¤YÕ`'}u&8ýSÒ*ýã0¯°¤­ýmü¹©qëGΤœ“]½üR^ ÓZ ê|³sßÂx”»…ò)8Ú/bÈ©h†Ñ[fè[´Z¯ñ+Sr÷w ѳ¡ÇÅ6¼qV¸pRnw<æ tŽD3®·æ½8}~qh‹±´3{/?u-A˜.C%ò0²£*uãKŽ s ¥QÒâ»°!ÃuÝÓPåôõ°2þ…€ZO{¿ÌëD€ü®öÖœ§õ÷Á£Þ«”J0˜UÄFv$…_·¡¹ÆÑRdâšÛŠ)ݯÔrm—*€˜¨”Þ~Ͻ…ÚÍüõ…+€¦çj|Oï]‘~|ô;ÒíowQ„ä‘R™º]ûüø©$zÇ2± r:ÂÝu®ç\y¦/ã.Á¯†±.œß÷©¸÷¤ªPº¡™”Ó¹2iošùº‚ðÏò‰ž”a8ƒ¢}ûª•¡vOUèsbþ›žá»+ÅÖ‹ÑÌÊH—hþ Í·;ÏY$ñZR¦rä¿“'*ÅPoߪºÌÜD,ä±åô'm—Gô K—Ê0Aç«ÏxêõÀÙ6]†ÜRk‘NC‘­“&[‚ºò"³vG ±P’…èÐÞÚ$ jÑl5ÇC?æ„Û.Øã²ÇsI±;$_±mûOš>rI¿ùxd½«Ú,gŸ˜Ú#Óº†á™WѲ¬å3ñ>ÎŽ•é[$ïU™ƒ^ašwžn†WŒ8¶`ŸÒC[=vkOTËÒ4c.˜T—«ZÞÞ³øð0OÒžkcþúKŽê­¬rÕ&Uöqc:$tuF˽?÷È3?ŒñŠÝDà«Î'Úr€ÔV»µ˜®É<á´O7'^š 9Û|¸îâH_Æž,ÊA¼>‘tŒ/íÆBw±ót·Ïœÿu)àdªåž¡ÉòÁH[@ý1ébÑþ{iJ_NŒ³ÀÈ,;_ïX@y\p§vÓ™g|èžo@|%1Íx~•öÓóÁýËàU¼bl×÷£B"ì‚®ÓÙ&£½Ÿ€"(QSÙà´õ(ãŸg0%_vÛ'÷Ó—d]LÕ “‚7í_ˆt6½Ê=Ü<*B‹TA¢¸ÖRb–tr¹{ÏŸN;_¨tÄeší¦4rå³+ ˜p„PVøµÎÑû ¹tGèwˆ7ÖîèšXˆSàë.«$<ð¿!_¨ÓI£J? p}šX«CœË¤>6'DÈf(ò#DnJýy¥M'-S*ä5×;VbA“=žI` ×€é=}ëw•øc:ãB¹cê= %ôÙ›MÑáŠó€°©¬ºvâœø˜eaîÈJTe»¨ÝÉB¿ÿ”1f«ž?•±œk(øÉËŽªž‘v†Ì£²ìyq­h@Q9„Àº¬wnKEuxGs°vƒe+XU eèýÖ’,W d%ãùÖm7éïÞº$}¸í•Ll4·³~ÅA¥i¯®ˆ/ƒL9̲™åW][óî¬[⮑ÅÂôœ¹sû•Í‚Æ>kú+ÊŠ °ì;«ÓŽ'FÉ ¯ËÒÕé*~÷k?½Þ-‘{sÁô‚Cj3B5½«òöÁ5‹ºÎ+#Ÿ¼¤¢6‡9ê‘üqUG~JÖ$½!‘PyÕQ;㸠Þ°¯˜1Ž|Ϊ —]ïRÅ×Ïëf§ ï}ü 0çùZ»µò~{°§`=d±g;†I€K $Cx'@ãþ‰*ž2—ô¤Åáø›ð׋+ßÑš Ö_åM4¨ËozBÅJótÞE’zû÷æ9ž`5íë ]=&fŸÌÞÕ½óÑ'-¬<¢ŸÈ•²V}rýÁÁŠæ¾¨iÔ·: .N½9èUU·¸øÂWuC°´-s×Â1—×rܧ pŠ+Lóéyª©)¤L=•åqƒ?sL*ªsn)ÃåtRè¡NEoìŠã ™Ï<ìGpÝpï'£'ä!'ᚣÚú g¶*Ì”¥»‰¤Š ¶»ÂYç¾üÐòhß±*Òr-~oï9ãÌl:„è}bCFÂ+XxmšÏ–]ý ³U„ÕþŸ-4¢üŠPÆÞaï’zæ¬Õnç­þ ÿqKÚñuÀÍa<Í@;4×;¢Êÿ‡n¹sãñµø­9«Äk6•'eä{¥¬öcì«” Ÿå)î€ö/­©vEoô’ Ò§lÝ0|„JëTt6 #Fw+4› µlT’ [ %÷~zsYÄ –5#KnSuL]ÙãØç5\bÄF'À[ØH:älз'}UŒ×3{Lû!á%_6ÇÛúÃX¥û‘£iËîçF©‘JVú O-A‘‚㥓µF§¸—µ­Æþ!¤³~xm»!Ê+Ÿß?T7 ÑÅ~D÷Ó 6oÉÜ*-û&RÍ'Ê„ãSsv^ ZóŒÈŒ‘„½a8Qóm?Ä[ïÑžÚ¸ÎÚ#=Fu3Ì~Ú–c9¿'ªôl_-·æ v. A\äAo“±²vÕyyZKÖëT•­!8©!‰z­ÙAüÈY5]¾' °Râk/÷åÖ¸Ã+Dõ…ùÞ·ÞÕAùwH¬PfÙw“>ä3»-[dlècMoçñÍXzv†ù:O¬]¬Áâõ„7µl*»%ã¼R»‚zX´Õ£Ò½,¦r:n)hAó)O i¨SBtÅý Ý•j]„àmdàÏœy UÑ`æš]F<y÷—{•3È.¬Y;åÐëUÉ\ZDš¡³àª¤é°HMê½DC8ÓÚÙ‘FP&õ´ÂiìN Âã+jæÆ†tÌO“ÇY<ÍIŠQ[¯yèÚY8Ÿ7Zi05-mNq[[qHõΚ£Ë2½ÊßO*(Ëûi¥ú\Æ *ž+Î Æ®¿ ÂÓ8ÁH>ÿàà9ijý4Kõù³–~}â™èi"Ë“âRoÂß ±¨³Ûl%zí\›×Üo8\Ö~“^Z×ëê¡ØÔËæ¹#•…×lÚ ÅÎ¾Ž®~ŒžTæývENWJ‘û‚hÚžm¡ªéÚUé ¥‚Ô Wy}OÙªÔë\T £(üqÞ`Ë]%\ŽWt¥÷{¥‘àNò—¦iêóæÇÐ¥œÀë·ÂXSǾj^!zëñq¬ÉzešS[]ÊÅ”è¬Þ¨a‹ש1ð²Í-Ö >fõÂVNjÓ‡O>…%æ»ïãõŠtôHN9aÉŠ‘³¾¥\µÊ¶ev[“—'Rž¾vAh’Å|zÌ_iI¿`>ÿèŽ"ÆGÕæ’‹Ê¦* Âþ¦?]½ endstream endobj 113 0 obj << /Type /FontDescriptor /FontName /JVVTVV+CMSY8 /Flags 4 /FontBBox [-30 -955 1185 779] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 46 /XHeight 431 /CharSet (/dagger/infinity/minus/prime) /FontFile 112 0 R >> endobj 114 0 obj << /Length1 1474 /Length2 7815 /Length3 0 /Length 8798 /Filter /FlateDecode >> stream xÚ´4œk6¬† º¨aˆN £Dï½w"Ú`0Ê fô½„(AD ¢Kô½½ у¢÷þINÎûžóþÿZß·f­gîÝ÷uïkßLôš:O¤¬ày8 ù„‡‹[ £¦«ÄÃ àææåâæá01éB‘¿õ8Lúþ‡‡Œ Œ¼ÓÉ‚‘wŽjp@ÙÕÀà àæy*ÌÍ qs ýíwȂݠV5.€2Aà0ÉÀ<] 6¶È»:¬–l!¡§œ¿ÃRލ%P#m!Žw-Á¸%‚ôüW VQ[$ÒItwwç;"¸à.6âlœw(Ò  A@\Ü V€_ê`GÈh\8L][(â/ƒÜévîPK qâ ³‚¸îªt”TNØ_Ϊ9pþ\€‡‹ç?éþDÿJ…ý[ZÂÀ0O(Ì` u€4äU¹HNfõË쀀ßŃÝÀP°ÅÃïÖÁy)-øá|K¨Á…€:üÂü•æîšå`V2pGG ‰ÀùÕŸ,ÔbywïžÀ?õ‡ÁÝaÞKÖP˜•õ/V®N@=ÔÙ¢$ûÇçN…ó_  àçææ  Έ‡¥-ðW]O'Èo#Ï/õ_o'¸ÀúÄj ¹ûÃñF€Ý ¤‹+Ä×ûŸ†K8<<+¨%`±Âpþ›ýN ±þK¾›¿ ÔðœûŽ~<î_¿ÿœLîf‡9xþ×ý÷ˆ*êúrÒª ÿÇ(- ÷x?áå<ñsxx@|€§wßçùÏ üþ·V ýÓÝ?2*Á¬á¡¿@ÜÝÞß@Üþ0ƒõÏÚ°þ]A~Çg€õ¿ô7ææç¶¼ûðü?/Áïÿ?îÿÊò¥ÿÿv$ïêàðÛÎú—ÃÿÇv„:xþñ¸ã³+òn7Ôàwû_WÈ_ ­±‚º:þ¯U ¾Û)˜ÍÏŸððqqóý¥‡"ä¡+M(ÒÒö/.ý=Œ»PDŽ€þzw¹ÿÇv·z–öwo ând¿M»Íúw]9˜%Üê× ‚ø`°'Îî$~€7ÏÝ®ZA<~Sä‚Á‘w!€;Œ¾k¸ ίÁòÜõ ´ý¥û#ò€ÐÿŠwDÂÿ! €ÈˆO@׈B ûoñ_-Zºº¸Ümño.Ýõÿ·üûÉ€@< –83SpK‘`»ŠàƳ2)j÷'kC ¾Ý‘GØ«r”nëZ2¢ç‡¼ï£¦¥Z'¾G/o nkí rÛ›.ú¤D ?œ©jÄ´¶Xáô!»ØB¨º‚¾«´´™ó¬lî~Òþì~$9ÎFÎ鼿6™¶¡ Êü©äô£í.KÍO.<´¡è(, AA^<´à&©´¦ér©ñ‰f%FÄäÐvÂ#ùó”®~ÿ·î¯Eœ¨ÐåêÍÍN*Gi›âo~´ P©˜Æ™VÛå(.cYƒ_°Z t$‹¤+­Ž(¬F+iÛ|RìYBïJF—YŠeØ[º/õÎÂ囩{‹lA}äL¢n®/?KQ{q±‰˜ü}ÜÎéûò£?ƒ+¸Þ©ø¨–÷ŽÆ›ê"ÓÇSr9ûSIj4u½¿ðŒZÞï r‰rJ7=Zñ×BɈƲûâ`68Œ,×Ëiì PÄe¢'ˆÜ¿¤Ýz¨ŸCã#0¼3  Í"K¸6²ë>Ç·PÊw({:¬.2áS<‡ÝýQf áS8íå`<7U'0ºÎEÒ“‚|Šl›õš0_äAò¤7%sbáȳñãtó%œ ŸÙlÑ(Ôzüºn…¦?Ìä6¸R-œÂëªü'ô65É!}Ï ù6ä‘^ù€u\$n±ïp–s´©%gˆä‡DïðÊ~v¿ÝE‰He–Š¢ÎJ·cÆ×•GÐJÆý —8ÙoéÞéÏhÅIl¡ÌÿÀ¡à(rÍA-è2÷%®põ{_шòöãÊû$¡ã–µŽòǃjbXÖ\Ä㟋W¨ü(ÐŽãÕnþ\}‘5Á B÷žšøëÅ»¢X)ÅX½’½ùÇžVæì6;:©H§Õ=BRbJn#VÆ¡IC7*vƒ—ˆè-MŒ›XÉ}[Ã=ª\®.0ÝSÞ1~4o?,µ¯uÅnÜÝe‚Äc»¼D‰òÎËGËoï™âLªÚ~,É»TB>ô%|ÅZ!en^ç”Ò¬9#Ð>Þû©î) R[^Yÿ!cM©åÁ`)ú2á뢟ºíÇH&î]ó¯a²×–8Âbœwö¿ÂðDâ€û'†Õ™ij(õ(xîFVePÞÝvý¨”ï Œ¬ü“ø>\¯µOwb̧ž“y OÓ¨ !˜kzžR83a³‹}©9bJ¸~G€íÿ^™W^ŠiK2Ö>3fŒnépÌ €Vl’asØù"ÔÒ“'j|Á´˜6` "åwUz¸È<¶`õ^é~޵AY#²Ø$ùܳy“DwŒfq°Q˜©V=ªã q\XRýàókóë Û%C„‚Cå Ó:A\÷`2èíNƒèmuŒþÓ§Ä%¶3S§`éGf!ñ0TePimÒ'êµµ¯y³g}Õ,}ÍÛ’âr®_¤(–mGa.&S•BQÓ¡äv¾q#]ŒçïV;qô´œXŽè/›”Ô¾V ¾Çžÿ2žo(­ßÖ\؆ZËæmÎ: ©¾”ôYˆ¡îq.ª,û>cOQcñ7OOã,@JŽ¡·'Clº–”ûp4šIÅ9•\ÿö=J€¿pŒ„N¡ËcPí1 jÈ}¢ŽC“íæ®>·‡B+^.nMw¶Ýè‹™àéá“ ú ­.G“fÎøÂ0ø×E>l©±›^Qºá8›®ƒ§ÓG­å³yÆs™Š’`n¾»0µQwÐÞß VÄ»`Í!½ µcäŠ=#‘™%ï;³rY%j}–åÕÎûzP–Ê”Žõr¾È3¾P¢¨töQŒŠ ¦@–ðÑòÓ`©ÐœM\9[æfçë義Ýò]èéÔœíWb¹Ñ–³«ïA¿¤¥i@…Ñ;í¼÷…6´¼tpVdɉD]ã'µ`W|•ÒvHJs'\ÜÃ{Þ@ÝÿÑJÏÆ»R)‹<¸Ü¯ª< òOö!lªÍ˜Xo7Ü3œq°>xÿäB~ÈgÐñ €+Ås@ªÕHPÝtÈ6Þ®5nM“Ç­ô¥ñ°Ÿƒ{‰r”½*?)¾æ~åš¾JP>¢S;  |fŽÅbÃÆOÚW´!l\hKðîÙk…&úÄ:O€Pù¤K‡-ªeú*pµWQÉ"œrø £d¸‰ nö‡wÚë5rGšGUû¶U(ÓƒM^&ò„dJÓY-^³•?¾«ù0ÈÈ•¹ûJOëjÊÝ(á.áev:t áÓvíQyãÞtÄ“™ðsk9™BÂQ2*…“Á•œ<錱(ŸQî4©km{ÃÃ¥ýÈ­ú[I‚ãÚòÄ`}÷dY÷ñâVÅÄ[ý7F#Pg¼²Ôù¯ÂïSåàárö3‚€ Û(¤‡Öþùï²è˜Ø´™ú,ægì 3ƒ°âéwÉR«±ûå´žÈÐ=D$©ÔNÜʾøxPmü!PL›—¼—ìÂJœþÙìPÌ–PƒZ¸VŽ¡D þ¤Uår *ö|;ÇôúYé¹%sÒþnàÒ±SºÔq>‰-{]! ƒÓ³¬ãõÁ‰AN‘ž}hJy-iŸÇa«/ÊXh¦k“¯º5]ãzqdé%$±kZ5îyUÎ*ò! T>·½y4ðdúÌïéGò.Âホ.󙵿p?oN•R3ɰž4ü.QãaB­;Kù¾Ó¿Ø}‘=ÇÔ÷éšu´UW"Ž2(e‹ô˜Gƒ«Ê½f  YPüX|³qúX&ú‰¹9V‰õ<ŽnB…œB¨ÿe9$UÉUŽ4.!Omƒ;Ο҆*Š{U…Op4¢ðÌ¥#q*ë;`÷Ý¡êjk×i“`‚­ð–Ý#èý4Ù5Kš0A{®³‡íÒv"¢8¼ÂJ`ñÙÜ[–V¦‰èýeØéxùÄŽƒq§ÌÕtÀ’ðûÄå—i,>OÆž!dSü¨•~Ùü×t¢K¹ˆáÁ Þ¯Éø¹yÙ4‰gÜÙü‰0ºúÂ·â¯Øé­î¡^9Õp€Ðü–v‡[\zÙD”EêÞŽ†/)ƒª7Çჟç™m­‡þ¤ù&&¶FD}ÿ0%×-u".ì0šð°¸wqvPðLÿyÙ8þ&nÛ(ÛÆ䞘cßòE_ ðíõ'ž‰hB‹Z´²yB×ä¾ýòæaèS-¹O-×kWaÙÚsA­J'áNUžP‰/~gHB¥Ž•óÁ-äyû@+]Y_ f9'E޼âgíïoe?l¬52f¼iQÓÖnh5ÞirÏ;iºIÚ}º]2SM” Í=-rïPût/ØW~ÎÊP».®/rÇ=ªçûîS=X­À²#Ó ˜½C³oaAë¯ÂA–š r^(¶¶&Ò Œ &l3ÇÄ`Àý U%øF†²¸G¯ãb0e7 É~ ®(ÈM €×ì_F.?ÂO|Mà°¢d«b†\qLW…œ6ÛܼO ǽ¾ Þl¦¯°©zÀbÝ!á\éÞü’¦’m4y‰<=fÆEËä}ŒrÍöíÙN¯7ç]^Q#-јXßóJ™xèsPz2kÊ*C œ\‡¯¯ÀÚ.<‹+.WWe$’2a¬¸¦éµ5Ë·Ú¦Üy¡Qƒ*Hâ]ïÉéÊK†ÉÅüÓ]Yë3ú¥E1Ÿ¨øB­/8è+ÂWû ¬-÷ŸGvø(E#ÜhíüSúì‹5qmhV€{•b÷¦ObÍLôPh3kÚ½6éðtw9õ G…ˆìwf«^Ý>ºõ'áÿÖ¼X› žì×r(ÙžIS°·:YÞñÎI¿ï“¯]Žv­ŠJˆûƒ¼Û„ÔMFÊöc7+–§¸à–”  |“€§ÿ,¿¬èsDD–NŸÆg<)7 ‰¨ä×™‹ “h+uqúö17‹…á÷¨h‚½>Ÿ;Œ›Ì˜¼÷Æ *Å™™½h”°ÈÒŽËð .ô›t. zRÑ»úï9æQÆã?ËúÆã5+_ýH‡ˆY«út8¶Ù–à„Qé¦ræ'µEµsJnñÄ«Ô;Ow¬‰T+9 Y²^™ Ä~fÁMÍ&GÛ†ssÄš°¸ÌßéB85ó‰Ùð6f¢7w>kV7†SR7(DĽ÷æ´p‚òÝNY‘×çÆñü¥‡Ã›^ÉSlôyûxK1¯ž½cðê¹¢ù¤9#yÒ _'ÑÎpԱ⌂i‰mpkp=Úwµ>äJÃÏnÑ…ŽE!t¥þ­=ºŠy¢Ô÷kíãtvì9œ©§¡bëa@Ÿ‹j™\CPZÀD¿ÁÃVzw©äŸÔöŸ{YtÕD%uFƒ£Kšì5­)‹ëÄ%Š~ ¬é¢€T€ÕËë-*-ÆÄE3&OÌôF~™Þ.WÕQÀß./¶ÀÑMÿø‹ *ħwz0a¦<ñž™zÉ”Ú'’ã9aª¬‚ô|Ì´„Yc=©«–L+QI3ò¾ªù’=œeý`µÌ…!~/ä ššãnR5´û¦I9üËt¹É–;Ô±íG‡j~ÖɲežZíh¦½x¼¢ÒGz>‚Ƭ´ÜìróÙ 0‡tNqIqÛ:"ÁÄߌ!duŠ?–CHndŒ¤Û‰5ðéów™ª½ºç¡•“$%ãgÆPlÎÅÂD÷$—*Aª|ݦ-èÍ“ù04ÓhÁ%RÞÖ½â7ßNòD=k=¶¬Ö®tkòX=®.u»OVË1g,ó–÷¾ø„Mf|iLÜ„h´} Lj“_Y¢O~Š£}“‹bê·tqƒµÓx ÑÒR}ºµK2Óp pšEÉ2É%ˆ‹Ÿj>Sìøyé/Å´ñóò%Ÿ%Q›„¡\àr™ügJ‡ûxæPwô–6üG­àyté•!ôTÕþu±"d›µajï²|6?@›“ûàªØ­¸«Ú=lu‹ù Û‡tÀëw\ýëÍ96àFðvŠ{( ¾¼^ŒÃ_ã¢ÊßÊé^,¨{ÝR¤˜~–À|nÃ(>ç‚E(µu&â¥S‹ ÝœKñ<ÂÔÓ¸oÔ™© ‘îˆ>úAµgçvýËþ=MÇÒD5÷vg‹œ$aœžþí4|Sù¨}Â]ê©ÕéäA*†`Ðã ìtW¤;kÍÕ4 åiXIEÕÊ@î/ßðź¿†’œqÊ/À=M¬Í®MEÔñ=õâ1óØh'o‘Ÿîírö¶Ò§BÝ#<2^S>òößÄó”9šÇ“±ð‰i¦øˆKXÛ"R¾’ø”ľ-áuÎjú­Ê–ÏAÓ›Ue°Pµ×¼A+â}ÿëa+Ù2LlCÅ â4c«Ùôx»f6^9&Ø5‹¥™Ì_E#ÿ¢ÔQ›Ôx^‰ß J¦áÇ/>b#ùl–5‘ƒj´]èúÛ§/µ ܽU•ö‚’øW¥½x!'eœøÎIR½í sh²[u|êÒE®Ø­+žH‚/§uŸó_3äÆ$qI´Ä–Nçæž;Ì3§­“RGhû½2p—¦z 3éäPâwf¢èF2WÑ$¢ÇÐó†î(hé¥:%EÛ&qÝßAjèð<ûü½Îc1o£ôŸz¯àªLƒób×eg€ æ†J·Þ "WßbgDY`k§Ö·2ùR ífGpfËÙt‰j-CÂÏójË"ÈcÄZ\5ö•‘ÖiŽ$A¨ÞÜÈìÆL]=‡ÒÃÊîDä½5¥{÷N­£”ú@c€êòÑJúµÑä³QÕYý¼œSOåZ l4­š¨è FQÞF[™û´e#”øi;t·•¥`Ïï† ·Æ½z ØÍ±O´!}‡9Ón%c†ãØJ ìVšÚÕm|¥ñx9.n!X¡ô!^#„Ž·pôMC¹`ï4ç¤òf=Ú%yb\=B8´GÖ¾î$F¡›‚±Ãy¡•÷lý¾RÝåFp¥´Ü¦r3³K£‘Åý‡}e:Ý"~ÎòB–ŸŸå[Z&SÖO¥4[©öäXpw'õ·E½]°í2ú²W¡DªœßgŸP(•5C¹â-"ÃqP²4tq¦M‡ dO¥0Ë–sL>„QbDsi+¬ÜJ…B*Š‚Qzºr6@)|pTD’˱ G%ÄÑ=b«.éÏɲczðઠJ§Lˆ”žN÷Dúâ[Ü®ð>h΂f_…#ØØLI æ0N '…I QivwµÒEÔ ‹%dò*5­åÉóOÞöߟão'±ÊÐO/¢ÊZ´cÓ ¤Þ“+ô š±cÚþ4úl»«}WÅà:žç 5z!i= øÌ)ÙœÉ%äÏpî.Ú¡@Á}b©ƒîïs“è ÂC)uÕ3{5`YÐÙñ‘ò1–bÛ¡ÈáÍ]GÅÊšj^Ù†ÙÔíŽ ·øÃ¬—˜²Þ_Ë)_P˜Ž¢¬fø‰ã+ŽoÊ7[v~U—%“Ù[˜BƒÒm F÷V‚k)*ÂцKˆ¥^¿V1R³¥Nr”;Up2̘ȌµßY•â)§æ ]ÉïPï÷j#±µ=Ê•Ý3ÙÅ;É!ßñ3P er;ÞT2ÓZ®,0ÝÅ¥t Þ }©ÆÀn¼ç “;m.³“s2A´¾%í²EÌ­"¥ªH’ñgTûÜ}ÆËüã¶?è½e `ÃV›v˜«ŠŸ¼È˳­043¿g bØó.ßs²2´Ùä“?º ‡“R»^h¤ë*¼êДܢY¼®C=ÅP «L,…Šz©î¯yQüÌjõ—îã-03!×a&¨ç7Lý°PtÓ½tHaÁ'ŸÙbf,h)Ä{®§]ë¼ðám”¡u©çï*~SŒ;±¶-”«ŒQNQJè&\¡Ù¶PÁà®_Œ;d÷®.³vé›\BŒâó<}vS fhXbƧòªÌûåª>ÿJ[}…Þ)²’üƒÈLÄuƒ«XÅGüeºwEE¿’¢ƒ¡tQ®dŠÏ?d–D®¶ö"Ü¢N‘ÞH²ÀçN“íª†ÑüP$#IÃÆ¢é  gˆ°.ƒ¶ÔCòÝTÊž°ü2«ÀsÒ²é©U¹ALDÏ Ièÿ˜W}ýœe`Ýw³y«)@/¿X„˜”‚ËWÄÃå9p,%‹¢¤÷ Ø&Ä›¬¼²Û2ƒ¤Ô5*™”ü¥§Z:ìt9{F¦Íf^õ†Cñ‰J®!eäl‡2‰éåxœ ^)ý¥x b8št ª6å¼Ï€ªÎÜMÚ› êjPŸüåY½:ºP ‰y•Äùf*cÞ&/%3ë ôúc…‡ó:Qíu$šä‡·¾äo0sØ÷žÔº¬ùn-á­ž')Ìóú™SÙ9T»½/¶J‰”ûÄ$oÓŸ_“`1øidá8”þé¬ßzÙQ'=ŠG.£ò>ré%·Ûá#CóÀ¢lÙRÏ‘$»QD#w°µïlèúV¥kǧnK²S ¯Qïm4g’¾´‚”ø6hT1Y,ž¼ŽÀÏê~Í£#€úq"%ŠiÙfPLÄÂ4ÁâQòýûsjþgËé±Y%ƒ¼ä«™A#æ5œ¦kè‹÷¥mÀ)f“ÔÙc¥.=Êc½ A!]ž8ëý¬2‘·«-gƒ6,¡¼r'DÏH¾0«±cV·±)(WàL1¾NН¨*¹~•ï5í•õ¹‡]’¼XçU1 Á+-›JË·®½õå¥pŸÇF±ˆ E¤ÃøøÍŒ€TµO¤]^/£š`ç7ÞåÑ ˆ”´< ®$Æ»L”§tæ‹j»#Œï—‘oE§˜ÞäGT‚ÍU Hô44jZ/7ó³t¶»hÅTÇ))Þ+¸ ˆE|¯”q! =ê»Ö[Ÿh2USÎÞ°~㙼Šêìñ#ƒ™5"ˆxÁ£×c ¢È‚ “&Eß‚gƒNz쿵·sꞆq= #u¦Žy@cFê–jÏÒ«<÷Ũø½ º4/ =•1KnIê 5# ®H+&EŽJüóœEOçû´|·[¼¦áƒÆy­ú6ö9³Ï.±oÞ y¾(¦RP K0±ƒûÑYzÞÃÐiGƒ|{Kk§_†W %mÉŸj‰Iš&ËpjâÕŸØ dŸ?QÙ!sWê ¶®œ³YÀÓµJÌ´â2‰âz¡¦ì¯u„A3ùëôBÎjŽÎDź²#‚¢ #›K˜&òªº!jÇ¡Ne•-$œ‹Ê#á*l쓇,>Ç´u5pc¹XœMñ‹ŒÛ³ƒñ-b«7X£m`•YåÞVQÝê<¾WžŽgë³JÛjàø6´æcNÊ[OÚR¦*dª¦å1à„!ö%ÿѧ͢ˆ±Z¶VáIõ:…Xœ"tµôƒ¾ËÀˆ×ò1§1 ÊŒžý󕊱Ctö2ì52Á7¥ëžìœÝ‘å#ô«^+ñób½P´·ºªÄ„,€´1ÑÅÓæ"œx¹±ú¥XòÛØGœ¯Õ˜Œ%UáÀ—Faªdæñé:[Æ‚«V›ÉlÚ™†Uíä+ß …Uyj Š„7¦j¤^/1æxáææi¦+ÝúSŒœr3| ¶ûŸ¾º.úF÷·k·]™Ð·æšÙÝùÒÞ÷¼©;ÖÎÁ¯ßL¬tmíz« ñtŒÀ³Ž¯è=‘Ï6‰<Ê7XÍ“úL Š™H!ÙcÛátKÆã…Tïû’~®> endobj 116 0 obj << /Length1 1574 /Length2 9226 /Length3 0 /Length 10263 /Filter /FlateDecode >> stream xÚ¶Pœ[-Œ\‚»;Á]ƒ;4Ð@c4î îî4Hp îÁ\‚; ô‘̽3wæÿ«Þ«®êþÖ–uÎÚgïó5 ¥š‹¸Ä $q€²p°² $•559Øìì\¬ì율44š`¨èo;*6ÈÙ qüG„¤3}¶I¡ÏÊ€‚«€ƒ ÀÁ+ÈÁ'ÈÎàdgø;â,º-ʬˆÈ•Fâèé ¶²†>¯ó÷#€ÞœÀ! ÀÇü' nr›Ê@¨5ÈþyEs @bA=ÿ‹‚^Ø udcswwgÚ»°Bœ­Þ00ÜÁPk€:Èäì²ü– PÚƒþ’ÆŠJд»üË¡±„ºA€gƒØäàòœâê`r<¯ÐW¨:‚þ¬ô¯fÀ_Åp°rü›î¯ìßD`‡?É@ssˆ½#ÐÁì`°Ûª2J¬P(3è`ñ;hçyκÁv@³ç€?[dÄß€Ï ÿÒçbî v„º°º€í~kdûMó\fi Iˆ½=Èê‚ú{R`gùsÝ=Ùþ:\[ˆ»ƒ÷ßÈì`aù[†…«#›–ØÉ$/õW̳ õ?6+ÀÃÎÎÎÏÅ 9@æÖl¿Ðôtýqrü6?kðõv„8,Ÿe€|Á– çTo uvùzÿÓñß•ƒ`6‡Ì@V`Ôÿ°?›A–ÿÂÏçï ö°?·€ý÷çßOFÏfq°óüOøŸ#fWTÔ}+Çô—ä;%$ on '€‡‹À+À ðýo–ëÿ[û«ü×ÞþÁ'ï` üKÂsíþ–áöW_Ðÿ54 €ÿ^AòÜÍ ýšß‡Ýüù‹ãÿyþ¤üÿuþo–ÿkóÿïŽd\íìþøéÿðÿñíÁvžE§žÕù,!Ψ¿”ã¹ÕØ@¿m ïoääú|[8Ø9l–ÿ€\6«@n›õ? € üøÜœlÿ€ÏìÀgfÇÀgf§ÀgfçÀgf—@^ôü¯r˜»:;?ß:ö¹Vã?×ä2Gý61 ¶© þ|S#Nâβ=!2C³­“ÆÀâý͹Ýõ9™¡:+hÍùJ!ü%ÎØ4­·Ä^ŠÁ‚wiqÔšÈB;+ÌÚX?…Ïh'å± ’<»ó7Ù1jß¶:¶@Ý^4”¬m SGÆZÇCŒB&KwwÙÕæâ>ã¡0”ëpg& e·XdOïo—z‡—Š‡Çµ“^#–›(¨›xËM+RaN¾¸ÊH¤?wø»|Šl2˨?y÷ÚL9Îxìxh´THŸ¸b "%§cb½*Êh䣎¿üôŠÛ¥Z Í Ű›Á`˜hQFk‰Õé|ç‘û8L?_ñÎËÝ´ÁÓ.i²XžÇ §‰ÂÆ<þXï#Fzs"_nÑüp¹â'Á›×b KUbÓm2G°ãÊa»×Ý&åùwد­˜joEšöÞF Ìêg‡ý››RuH#T.q#äxôØ·„õkŠöŠòÃúañI¤Ç?ŒNÐæEË¿ìáÞ›»¾BúŽ”‚IhíhkÙ®bP™Eƒ/q?¸pžQnšaô°ožÃßÑ«JÅ{s´ŒÎê¼ýêcœ£¥7»T—¯XÅnî~¶Ñ[u‡µÏäëM‰º8Âêpâ,¼}–Ý•ñ2­ÎrAäÊ 7ySÁÙƒ½LñÊiYv}j'Éâ\ú0WœŒ¢Ÿ<~è>1|M>ñeå;U5ŒV¡˜•¸(¹&p¨î§!ßÚ„›Ü›~=ÝKé× ŠÏ…<ƒmŽKqòŸ‹7ò#4ø÷ey|ŽoÛ£¹ÙÍaåÒgç¸Ï\ ŽÐÝõÚ{S2r¶³9?@§¾ ¤,FÕg‰ZÿLh6ËÚxdüFG€j1ðÁ4‚ ‚’ù^0‹*(,!¦ÌGkút ì0Ê’éWÅ+©†"è Ö¬EjOIžÈÇÌ,žñŸý™H‹NløºiÁbO©ø›¯†°ˆ`§ÃG5Ù`¦œÂ‰µ–§dö?]ǵp$A¥'ƒ¾4­ˆ¡OpõÖÊ«‹¦°d'g_ô;á#gQÞEÁ1rX¹_ïrvås³çå6suŠUlÖðy-½¦9H÷Þ7±®§ŸýŒ”öÒ£k*¿>à —1÷q°éàÜlŸCš7ÇÝ‹sç@‹¾¿Ûã 0Ë•7 ËŠèöl®ŒûŽn9Ÿ_¶|™¸}p^|WjiuDÓé˜' >§P%§§ó.‹«Åp¯xÔØÑeÒñK=N“lÖqNNtíø’_ËTHÌ•Bf¢[ã_#¼ëN=Ñ/ÎEzyøí‰.9®2VÕ{Z&í}/p#Sg(ðîfRˆg%«HÍ^|vˆ)O{aÄke¤0!ÓŸ"ŒÐ”ËX5WÍ •–¸¦f«ûàc¼Ñµºzf¹äIãjž×?Å7˜D%©w—ÄjKèuèv˜áÃ`%úeËшXœÆ4óC“çðõÀ˜Á̶Ù𥻴XþÒT™®çmßôÖoËLOß`O›’ÃÔÓ'WÎÐ!Ù§ß’½T¤ÃTÜœA§{“‹%½®¦Â,¨¼ç¡Á%Ë©Ž!Æk˜».Êr6gŒ¯M°´wé/ƒÏêdÖ‹x­G’{qg,/uû¦R¥å8^bZYùs3CYàÒw\«/fî”›©hcŸ]úG4¿$ÁTJuÁ› ê8QùƒòÝÛMŸöŠÊ£è¶{L×­‘zÐ45cå+äÞƒì¾7`·'\Ä\÷æ™â¬\Ío¦\-¬èJÉÃ×0èE*ô8Zv®*ÞçiuN,e;R½EÔD¥1¥="õ¦'Tìd•<+ tzÈõ³¼rÛ”¾Ÿåö0á!ߎÍûÖz‘ÑÇü3§ÄbÓŽ‡ÎlÔlʼn9EšX³Çß¿¢¢âp­ß'{ˆìä:NÞ4øi^9 ®ãg·ÕÝþ¬<+2䨱¯=g£µs¤„ëx3Ú€ÊqRë Á P„9¢O9õ'žrQÞ"3`£å4Œúaùà\dáÀà”Y,¡¿[äètËU†Õ½ÿ*nG©ðã-ÈÁm¦^þ˜ü›ûÐQ‚RÂìP7ÉêH5×– §Õè}S™w› Ȫ”]^ùƒ½…Ò¬ ü…âYPS¦Ô ÈÁ·ƒ¤U•èï¼pMw(}ý+ŒUã»Z/ãX¡ò¤ÖNñ áäŒìðyN'­Ÿªßì«à j¡ôs P Å®óaö¨B ¸|¢ÏxœÙá´8|}¼ûʼ–¬“Fê$¥Ôßþ9Æ{ùØ´7qÈÂ=‰Äg!]VáÉ‹­)kOŽ®CŸ:øý7˜ô¯ÊEüÑEë1±úXcá1°E/„ùÖ¼«•W±{¯µh^álìÙŽ,s½ >JׄÝK:ør¿iÒrªñns&èÖÒ.ÏŸ•­ÚpçX|³n !rêg *ÚðÛÇðØL‰Õñɲ Xø(Ú¥{Îê:$áa)k&}Ïê®#t’aîB!ø ¡dk«Øà,`ùÂÊêžÖnß§‚ö3IÀB å! cá,õÕ{²N·×ËoŽø´ŽÌ÷ýÇ{‘+`èšàpÚU0§Ò'8¢2v‹uV*‘K¿=i/£•¥©¸"1 <-bróaGzÊbÒO÷<®­¡—¢÷tEPUzlb@T{L’|>šë?’zIí¼kÒ§Xéøµ.,y¤l“˜‰Æí¾¡šÀK†}FÖë 9=Éš'x¿¯Yl%Qo‡ðê[´€ÖƒÌBµ(Ž£QÏËEÄÇ ë¨¯Ýy] ¥ç¡Zpü“s‰^ù»«J¡¢u·p‰÷‚䊑BU—·ù:O4cºIÁ—Ô_à ̆·êeŒ-2$&ÈçÔéay X4¹)³¢§=õ`Lé”èá_V¾¨>Q~U¡&þzÀ2>@“Ê’4Š Atú¨[B› ÞÑ'=xøÒ¨”S)à ´†í\ÔšŸv Mˆeùî*%‘’ •=¼á ŠÕ¤,Ü ¯ËëÕαFÍK×ä{¨#¼Æa³ ÞäãÝ} îàþ‚­©¯ì–)m3»Mäkº¾Ýƒ´®\ˆœû’5H×–ÒoiŒŒŽ³³&vVÇp¬2¼´¸Ã|÷úLj<Ò¸P[€šNûpÃy ¾:lçcòý^r«<ô5yï:¾(MËôu²Up„F'ÈÀ+#4â Y³fH½•:Bƒ·a¹ yÜw¤ÌëvéåZ'fö%²ìQo>°b꼄CëܵKŠŽ!t€„K»ØbòûE :‡Æ†âOÂm=ê“S†¢žÊËé>ÂAÄü“Îdl£—7ê8FS2òUš¶g™6Vf9r‰/XéQ»7O+²HY£¦àRiŠ:˾Á÷ÿøÆxß˃yÑÛ8aÜÜÈ9s“RÕÚÐr™?´ú*—¨¹,ù0¹HVšrÚ 4âÿEëÙÜ$QrFð±ìjüHM¹é÷e¼|Gó‹1ç¶üjáR€Ÿ) «âeÞØ–QrÄ#ë7îÿµ§98àŒ$›µ¨ ÅØ22sñ¢ëÜžêˆ×—§¿á&dB}€+æq]Q ×¼¸­L” ǃå/ÕÛhÂwh~55é}ÿ‰%óóÍí‹èÓ_µû£$h^A‰‚G]2ÿÂÐŒ:éÒ+¡÷ˆ©ëÊ“^8 ECÉOšý®îðØû¾Ã9g‡ PІƒÇÈÉ‹ô‡<™JÛµ/ë/:Tm:SÛ‰«Öw˜ê‘u@ÄsS.eÊsÚ©‘ å7i)?º×‘ËÎõ´ÏG3ùëº8º,^+(qà‹¾R®}EÕŸiV2ö¶ƒô´£ÅÒ¼ij&×¹¾½ú¥°•_ôij<^µâ¤H Æö¹Ñ¡ªXxQ€ 2ÏœœËzÐÝå2ÞÀÛÅfk䘭6éŠëŽÕwÿ¤‡üÈù,_:n»ä©¤m±_Õ&ªòk-|D™“lÞqÆ^ž,Îã„è–$ùíc·8| kÓóç7Á 3]×õÙe[t¹Ä$^ÞD-‡f fÌmÕÌ^ø Â|EeÏÿÔÓ{S<îE²%?SIŒ®ù®9R½œz†ùyPé[ÞžLÑÑf ­tw]úà½2Š»ÖuÖwê×Å>æ¬upq˜žéZ #è«‘S´ÓOåø"Ʀl¿V’.¯åVáUÏ@­É)±ÐT½ÂP’\)£ÖÑ5{×ìÐ Þ;_Õ}"iVM³}칿 l®í‚ú¢<¤znõ}ªX2ü¢Ú©†ƒ³ ô2H¨BjŒN7m'^‹Ž>ý¯àƒ)9%”a±° > ÕÞ‘`'ÈP`üQõr%AŽ   ²9#•²ðyšçþ¦€Š¼!bÚO´õÜ &‹abįïJäf³xÝÌÕVªtüxÃ÷Cå•U”ÄÒ³¡tBd!ø€°­Ð¬:¢@ý€Ý.iéÌÇ+¾_Tùfš_ÄR¨Þ’$¹;NØdâ%ÎYër\× Ñ]–N!Ö;a OÒŸþð¿íæ‘íz âÇ2‰N¿$¡0A±He²«Ð¤ò&óÖ˜¸9k«.ᄾ€³DJÙÍ *·~·fÅ–îÑ–_£2}<0£Å)za%y‚û6°÷µ#ª¯ž<²¢AЯþp掙kñ´Suµ1 d5[°ðµŽ&ìvbQúú¡i8tB5£,9ŸJÄžšOaM€×—CÈ%g7£dõ%Û»5]m˜åv»—µ$¿Ò·C×¥Á¼ø¿"vÎ@E'Ý”\°/$ô4­°Ò‘>ùâkækÓ&ƒÂs+&á,ñ2g­r‹ÛfÏt‹M˜ Iùœ âóÞk¥7§[æïjòâ+J½Ôߎ–¥j,¨o[0RìÚ©|5{e3V† çÞîjÒ_OHʾFÈõÁ¤Ó¾WžmoÍÿíùÏ:WËŸ›¨çtò(*m›¢Ã¯œlÊÏÓPGb3fj™róÅq§É¦œ 6ÛÊxi妒$æí}zùœkq~Èþ¢ »Pp~l^¥¯°’5î髯ó«öPÛG2PÜ\%"è^×»Õ ‘M¤ºiN¢Æµlâ#O]ÌXóÅnIÐýÙ ßåºÏ¢ÛdÒ ùð]úåzÜÈt$Ì#$’q]QÇNÓ’À[/¬aPT ØDd !Am®`ó—ñï î¨æú>ݸˆÔ’CÓäàˆÐÙ»—_¨®aîj×;„ˆ¹|Äã½d6ô¸¯t¿jʤ)l”äì3ø=áÒd•˜tb«Šöž÷ á媻‡­µ JWòŠRåèlûÍUÉóžDÃ%éÂÑZDì)þúéâËEäýVœÁXኬ»ƒærRÇá‚ÓÌtâ‰7ó!÷;¼MÞç¼ëâ#2÷ ÌM'þ ¤;«Ê»{džEg*H19{ÞWßY_JÞ ‹wÎo%K})Ëœü4¦lêé+{PœE0"Ê`ž°Œ¸ú#%ã‰oCñ%%¿ÄH{Œì0ë2ä’|¿aÎ5Ü U±)XCtœ]_¸œìó…²qDncþ4 ]X7ÜRD7ö\o `B¡ÄèX¶—mq‹ö´bxSROEα_|lX šø(߈¦á‚=,!¯Ù57ëvó ´×O¢ÞfëfGíS ¾>bwrjtæ¬xNG‡|ß¶)UX:ÿñ…ÇkSwÓæY_°H冒ˆÈ´§î—½i§—^#n>Æä¤&ÊÌ2Ù§[7†!ɺ0_“Qß66ú0_e¤x»hc’ŸÇOMÌÀõðd’eÊZïG™µ"÷nÿ…ƒ‰ÈÍŒ¥ýÚš÷Í@ëãíZmH©î€m)mwv­êärjK «vQ}rµêVçq]äý¯ˆ½h—Å`SWÁûÅ.’íU—ûȮԡƒ÷¤³¤‚cÅÌ®<€,MûXZâÄ‹†Æº„|~â”k™ž8”ïå¥Eª³5O,©kÅs«‡¥ö‡s¬½:&ï µÕ!ÎP­ë”‘à]”Gi…» ëȰɗ 3„´˜Ÿ9äC©Dsú– 0.NõˆONˆ?omsF‹bÑë˜CŸ Ùm1ýê²™äoU…6 ] ÙXþ?ºÜ/y¾øj.´“$E’Œ©Øœo0%Û´«7èp " )H Œ§å“^nY|žÈ_1 ,H?ÐB Žå2€EÅåoe¦áÏPxŸd`×­JÝê%¶MÃŒÝ0¶ND- }ƒ¤ ñ!dè¡ã/ä²×Q–YEèì)ð@©âuF†$wÇM5š©™[TãgBkEÖ}©>q½/!êV…ñr†å!¥za‚ÍM•®€[›GL&-§jÑûÍ%ܲžì$冤ôxäyÿë¡´¬·,y–óKí ;H}P½çÿJu܇îß›Mzša0õgœÇU0‚8G%/™lƒû×Z9¥¢>¦ °|ž¤¶7˜*`Æ’~ðÃÔµëdIñ“‰dqãÆ›¹Ô½P‘TèrÛ,ÓÊÌ S?ŒÑéGÅŒ% ßfåŸ3ƘtšŽ8ͤh¶.e¯iÍ;„ ]I}åy24)S£éË„‘Ѐ“üMgiž“èൿÙÿÚždklÏâ-ÒVFŸ°K4kÝÁ߸5gD½#$9 ö*¥ê®uïsÈ Œ!Ò€#îÙ~›êü­ãx)¯ÑÙV¼H7¦Tfð‘Æ^¯1}‘êÏÚúS}úƒŠÚÐñÒG£ôLQ{³ ¡iÈÜþ½ù‡!7iÜzcÜÏž—aXw9U?(LÓ0‘`‰y}‚œðÛY,·®Çôoáó8«ž˱AGO—% ?‘`=º¾Ø:ûòfÐå1ÄÜÌPWÃH=7^();8J¥Ã6låÆz·a”^îQÝf4”ª%6S»+BÒ‘üª*©L|§üõUxz¯ç:¶c{#¾rVÓYþ·dÞ/L¨‘÷„'ñ²Ð(}¢]«+CQë\TKìÚ›tÊÎàÿìÖ*‘®¶¼±o$a° ¬Ö6¿DÖÑD˜”᯿ÇÞ¹l¾Ž¯$C?ʞѷk—æ¯À!È•A¤=DÑ á«›vQ£6h4ã©/´yÅ̹üF%¹u>šËn.«+¨Ñ›P©Îj=¢iI…¼Þ‘ åž=¤¾N‘h)NÐT˜ÚùÕúÀDZÁ¿+6\‚‹ o[è.Åý…'— Ѩ’Ê7âï}¨.Y0zˆÁ ó:QŒW¤†¹áx'öú¤‘¾ElÓÐ"fµOÚ'º¥ÎTìou]³EC8H]4‰0²ãRœRô‡‚3… |Ë+—h“zõyC7¤>ê˜'Œ~`Ûð$A«¼Õˆ†{íz¶fGùeˆd©;ÚW=?ric/µåͬ­èd7C@ϾÄrØ-’àq²¤g·‚ ÿ¾û 2>ÃC*ã%Ó­è–-V,×ËÃûÄO5¾fyrG¿Àï°ûç5Pˆ‰Óyq„m7:n¥_‡ êì‹DfbîxJ+â7Ìéójò½ÐJ?ñÉ~;[ú>ü¡Ó®¾ ýò.ý,;kŽ|-c}Èèå,©WŽŽ"±úº¥›ã£tä"¢¯G✭ö¥ ñùY!®Òg?Ȇø—'33Ȧ7Â] ÎÇZ¾s_žRð½qƒñ!_gý‘T¬|¬µ)ÅzfÞ;có­ž«IXE0«TAÕñ 5ÇjÛ]_ÕçwŸê’£B; SxGüüü¯‘ñÝæf/ 5hA&Oè¶#}攊…ÖÙ7¸uk¸‰Enâ³#~•ĺþ¬ø©<í~rhÜÊ–‚— JªÀËS³;8Ç•#êÉ­5bÒ)Üm—ðPZ¸kù`««š˜ZèmÖÕqnÔ/X"›|@PÇý¾¢±ö0—!rˆÐ}%þ3?ÄŒVâz}¾Ì¤ù)¢ÞÏÏFrjd\¥ÚCW„‚0þá ç²5ÒÓ‰BÁtü™U§þIžk‹ ªóž›´_R¤|êHÈ>Bªºß‹<øôËE¢!4€RlÊØ-h•“ç'¤¬löÆ“ƒl{úÁY­#y~øµÏ6DG›Q·²ˆÊ}Ä™ª¹ÝÝ4ùN±ŽG&-¨&qèáã‰%n#Ø)­­ƒ”×1Ê©üºÊZ|W¼9‘x+w“Ö·ã‰vø"Âè³Ý 'ºM°`QÕÙS‘€»Æ’¾ãÛ;ê ÌåtŒ¤÷XP¦ŸÔÎøÊQù¨WoeCœlÖÉËUÑeñcüTu˜›¿8Gç$=H¸Æ4¨'ÎýÞ†Ë`"º§üšV¢ÎpŸá)¢–V .ÁgìbÕ£°|¼Ók²ڞηç}£vÙEϽƒ&y¨cˆ.Âzš•@œÙ+Ëà1{œ÷ᇿ›RãcÞ4>LD•*ºØ«zOö§¡ãP仞Ëx¥]š´Äâ¥챂#«8 þáÄú\h­êh8Œ*fÚÏ­ :8ë,7¯„m ±VyØ2#é ãÊî á`ñL'ÊÇrø½¡±<ǹAKê¬JVœGÙ£™ã5øL‹'Y.ü‹[ v.ÅA>uRèô§&ލ*ú@–­ö:ÛF2ŠÜÕ¯𡤤ê] n;îˆ)ý\ÎT¾š­Ã¼éV6{=´IÖÖnžy¢fsü%f€É¤òS}•,„ÙÄ]ÀêîçRßÞþëãp¢üŸA©L<ýUê2z¸“ñØ\&5°·Dl«y<Ó¸oøúHŠ</|ôøÆÎ@¥”±–YîÚ³0Ú” 4ÂÊ/ø™æ¿])VvºøP‰&-–ºë¯ÝJ=5ô¾öVØò}ÑñG=N·ùì'rǯ³Î'™½R¡euCX~ãä·Wéorw2e¹8 ³ÂWoæhÔ+­>¹@ì0õGo2¥ͤÃC˜NmPa˜Q•à÷IU5à{Q¦µ š*¤YûFâNŠÔ+ Gõþ°^Ó"´×TŒ@ Ä‹Þ>)*E@––z=}3Ô·¥ƒÉžÝ¦-*'t6>!àeGmÖH‡ |rî± fëÿO„ó` endstream endobj 117 0 obj << /Type /FontDescriptor /FontName /AKKXQH+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/e/equal/f/g/h/i/n/o/p/q/r/s/t) /FontFile 116 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XOSQUH+CMBX10 /FontDescriptor 89 0 R /FirstChar 63 /LastChar 63 /Widths 82 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RJSLPK+CMBX12 /FontDescriptor 91 0 R /FirstChar 46 /LastChar 120 /Widths 84 0 R >> endobj 16 0 obj << /Type /Font /Subtype /Type1 /BaseFont /DGKQXM+CMEX10 /FontDescriptor 93 0 R /FirstChar 16 /LastChar 112 /Widths 79 0 R >> endobj 8 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XAPWJU+CMMI10 /FontDescriptor 95 0 R /FirstChar 12 /LastChar 120 /Widths 83 0 R >> endobj 53 0 obj << /Type /Font /Subtype /Type1 /BaseFont /VPXKNP+CMMI6 /FontDescriptor 97 0 R /FirstChar 105 /LastChar 106 /Widths 73 0 R >> endobj 18 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FJSCRJ+CMMI8 /FontDescriptor 99 0 R /FirstChar 12 /LastChar 119 /Widths 77 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /YEERUZ+CMR10 /FontDescriptor 101 0 R /FirstChar 3 /LastChar 124 /Widths 85 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /IEFEKY+CMR12 /FontDescriptor 103 0 R /FirstChar 46 /LastChar 121 /Widths 86 0 R >> endobj 4 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RSTHTQ+CMR17 /FontDescriptor 105 0 R /FirstChar 67 /LastChar 120 /Widths 87 0 R >> endobj 32 0 obj << /Type /Font /Subtype /Type1 /BaseFont /HAJAQX+CMR6 /FontDescriptor 107 0 R /FirstChar 50 /LastChar 52 /Widths 75 0 R >> endobj 10 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FFGBUJ+CMR8 /FontDescriptor 109 0 R /FirstChar 40 /LastChar 61 /Widths 81 0 R >> endobj 15 0 obj << /Type /Font /Subtype /Type1 /BaseFont /UHBOYQ+CMSY10 /FontDescriptor 111 0 R /FirstChar 0 /LastChar 112 /Widths 80 0 R >> endobj 17 0 obj << /Type /Font /Subtype /Type1 /BaseFont /JVVTVV+CMSY8 /FontDescriptor 113 0 R /FirstChar 0 /LastChar 121 /Widths 78 0 R >> endobj 31 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KNVEBL+CMTI10 /FontDescriptor 115 0 R /FirstChar 104 /LastChar 119 /Widths 76 0 R >> endobj 36 0 obj << /Type /Font /Subtype /Type1 /BaseFont /AKKXQH+CMTT10 /FontDescriptor 117 0 R /FirstChar 61 /LastChar 116 /Widths 74 0 R >> endobj 11 0 obj << /Type /Pages /Count 6 /Parent 118 0 R /Kids [2 0 R 13 0 R 20 0 R 23 0 R 26 0 R 29 0 R] >> endobj 37 0 obj << /Type /Pages /Count 6 /Parent 118 0 R /Kids [34 0 R 39 0 R 42 0 R 45 0 R 48 0 R 51 0 R] >> endobj 57 0 obj << /Type /Pages /Count 6 /Parent 118 0 R /Kids [55 0 R 59 0 R 62 0 R 65 0 R 68 0 R 71 0 R] >> endobj 118 0 obj << /Type /Pages /Count 18 /Kids [11 0 R 37 0 R 57 0 R] >> endobj 119 0 obj << /Type /Catalog /Pages 118 0 R >> endobj 120 0 obj << /Producer (pdfTeX-1.40.10) /Creator (TeX) /CreationDate (D:20120411141846-05'00') /ModDate (D:20120411141846-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-1.40.10-2.2 (TeX Live 2009/Debian) kpathsea version 5.0.0) >> endobj xref 0 121 0000000000 65535 f 0000002056 00000 n 0000001951 00000 n 0000000015 00000 n 0000221848 00000 n 0000221708 00000 n 0000221569 00000 n 0000220867 00000 n 0000221148 00000 n 0000220728 00000 n 0000222127 00000 n 0000222832 00000 n 0000005128 00000 n 0000005020 00000 n 0000002191 00000 n 0000222266 00000 n 0000221007 00000 n 0000222407 00000 n 0000221429 00000 n 0000008726 00000 n 0000008618 00000 n 0000005278 00000 n 0000011289 00000 n 0000011181 00000 n 0000008865 00000 n 0000014592 00000 n 0000014484 00000 n 0000011415 00000 n 0000017557 00000 n 0000017449 00000 n 0000014730 00000 n 0000222547 00000 n 0000221988 00000 n 0000021121 00000 n 0000021013 00000 n 0000017720 00000 n 0000222690 00000 n 0000222941 00000 n 0000024164 00000 n 0000024056 00000 n 0000021283 00000 n 0000026652 00000 n 0000026544 00000 n 0000024290 00000 n 0000029604 00000 n 0000029496 00000 n 0000026789 00000 n 0000032616 00000 n 0000032508 00000 n 0000029742 00000 n 0000035613 00000 n 0000035505 00000 n 0000032778 00000 n 0000221288 00000 n 0000038291 00000 n 0000038183 00000 n 0000035786 00000 n 0000223051 00000 n 0000041191 00000 n 0000041083 00000 n 0000038406 00000 n 0000044016 00000 n 0000043908 00000 n 0000041341 00000 n 0000046585 00000 n 0000046477 00000 n 0000044155 00000 n 0000049811 00000 n 0000049703 00000 n 0000046712 00000 n 0000051012 00000 n 0000050904 00000 n 0000049972 00000 n 0000051104 00000 n 0000051134 00000 n 0000051376 00000 n 0000051412 00000 n 0000051520 00000 n 0000052184 00000 n 0000052940 00000 n 0000053539 00000 n 0000054186 00000 n 0000054336 00000 n 0000054360 00000 n 0000054982 00000 n 0000055424 00000 n 0000056101 00000 n 0000056520 00000 n 0000056858 00000 n 0000065613 00000 n 0000065839 00000 n 0000077507 00000 n 0000077792 00000 n 0000088199 00000 n 0000088748 00000 n 0000104042 00000 n 0000104380 00000 n 0000111789 00000 n 0000112009 00000 n 0000121656 00000 n 0000121901 00000 n 0000145852 00000 n 0000146405 00000 n 0000156701 00000 n 0000156969 00000 n 0000166691 00000 n 0000166937 00000 n 0000174326 00000 n 0000174557 00000 n 0000183675 00000 n 0000183973 00000 n 0000192857 00000 n 0000193180 00000 n 0000200702 00000 n 0000200949 00000 n 0000209867 00000 n 0000210099 00000 n 0000220482 00000 n 0000223161 00000 n 0000223236 00000 n 0000223289 00000 n trailer << /Size 121 /Root 119 0 R /Info 120 0 R /ID [ ] >> startxref 223556 %%EOF survival/inst/doc/tests.Rnw0000644000175100001440000025653512545056257015552 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} %\VignetteIndexEntry{Cox models and ``type 3'' Tests} \SweaveOpts{prefix.string=tests,width=6,height=4, keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{width=6,height=4} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default @ \title{Populations, Cox models, and ``type III'' tests} \author{Terry M Therneau \\ \emph{Mayo Clinic}} \newcommand{\code}[1]{\texttt{#1}} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {tests-#1.pdf}} \newcommand{\ybar}{\overline{y}} \begin{document} \maketitle \tableofcontents \section{Introduction} This note started with an interchange on the R-help. A user asked ``how do I do a type III test using the Cox model'', and I replied that this was not a well defined question. If he/she could define exactly what it was that they were after, then I would look into it. To which the response was that ``SAS does it''. A grant deadline was looming so the discussion did not get any further at that point, but it eventually led to a much longer investigation on my part, which is summarized in this note. There are three central ideas as it turns out: populations, computation, and the mapping linear models ideas onto the Cox model. The first idea, and perhaps the central one, is using the model fit from a current data set to predict for a new population. This plays an important role in predicted survival curves, see for instance the vignette on that topic or chapter 10 of our book \cite{Therneau00}; recognizing that ``type 3'' tests are simply another variant on that theme was a pivotal step in my understanding. This immediately leads to the important subtopic of ``prediction for \emph{which} population''. The SAS type 3 computations corresponds to a very particular and inflexible choice. The second theme is computational: given some summary measure and a population for which you wish to predict it, the result will be some sort of weighted average. There are two primary ways to set up this computation. In a linear model one of them can be reduced to a particular contrast $C \hat\beta$ in the fitted coefficients $\hat\beta$, which is an appealing choice since follow-up computations such as the variance of the estimate become particularly simple. A common, simple, but unreliable algorithm for creating $C$ has been a major source of confusion (hereafter referred to as the NSTT: not safe type three). The last theme is how the linear models formulae map to the Cox model case. In particular, there is a strong temptation to use $C \hat\beta$ with $C$ taken from linear models machinery and $\hat\beta$ from a fitted Cox model. The problem is that this implicitly requires a replacement of $E[\exp(X)]$ with $\exp(E[X])$. For a Cox model $C \beta$ is certainly a valid statistic for any $C$, we just have no clear idea of what it is testing. For the impatient readers among you I'll list the main conclusions of this report at the start. \begin{itemize} \item SAS type 3 predicts for a population with a uniform distribution across all categorical predictors. Scholarly papers discussing fundamental issues with using such an approach as a default analysis method have appeared almost biannually in the statistics literature, with little apparent effect on the usage of the method. SAS documentation of type 3 is almost entirely focused on the algorithm they use for computing $C$ and ignores the population issue. \item Population predictions very often make sense, including the question the type 3 approach is attempting to address. There are valid ways to compute these estimates for a Cox model, they are closely related the inverse probability weight (IPW) methods used in propensity scores and marginal structural models. \item The algorithm used to compute $C$ by the SAS glm procedure is sophisticated and reliable. The SAS phreg procedure uses the linear models approach of $C \hat\beta$ to compute a ``type 3'' contrast, with $C$ computed via the NSTT. The combination is a statistical disaster. (This is true for SAS version 9.4; I will update this note if things change.) \end{itemize} \section{Linear approximations and the Cox model} \label{sect:transfer} One foundation of my concern has to do with the relationship between linear models and coxph. The solution to the Cox model equations can be represented as an iteratively reweighted least-squares problem, with an updated weight matrix and adjusted dependent variable at each iteration, rather like a GLM model. This fact has been rediscovered multiple times, and leads to the notion that since the last iteration of the fit \emph{looks} just like a set of least-squares equations, then various least squares ideas could be carried over to the proportional hazards model by simply writing them out using these final terms. In practice, sometimes this works and sometimes it doesn't. The Wald statistic is one example of the former type, which is completely reliable as long as the coefficients $\beta$ are not too large\footnote{ In practice failure only occurs in the rare case that one of the coefficients is tending to infinity. However, in that case the failure is complete: the likelihood ratio and score tests behave perfectly well but the Wald test is worthless.}. A counter example is found in two ideas used to examine model adequacy: adjusted variable plots and constructed variable plots, each of which was carried over to the Cox model case by reprising the linear-model equations. After a fair bit of exploring I found neither is worth doing \cite{Therneau00}. Copying over a linear models formula simply did not work in this case. \begin{figure} \myfig{data} \caption{Average free light chain for males and females. The figure shows both a smooth and the means within deciles of age.} \label{fig:data} \end{figure} \section{Data set} We will motivate our discussion with the simple case of a two-way analysis. The \code{flchain} data frame contains the results of a small number of laboratory tests done on a large fraction of the 1995 population of Olmsted County, Minnesota aged 50 or older \cite{Kyle06, Dispenzieri12}. The R data set contains a 50\% random sample of this larger study and is included as a part of the survival package. The primary purpose of the study was to measure the amount of plasma immunoglobulins and its components. Intact immunoglobulins are composed of a heavy chain and light chain portion. In normal subjects there is overproduction of the light chain component by the immune cells leading to a small amount of \emph{free light chain} in the circulation. Excessive amounts of free light chain (FLC) are thought to be a marker of disregulation in the immune system. Free light chains have two major forms denoted as kappa and lambda, we will use the sum of the two. An important medical question is whether high levels of FLC have an impact on survival, which will be explored using a Cox model. To explore linear models we will compare FLC values between males and females. A confounding factor is that free light chain values rise with age, in part because it is eliminated by the kidneys and renal function declines with age. The age distribution of males and females differs, so we will need to adjust our simple comparison between the sexes for age effects. The impact of age on mortality is of course even greater and so correction for the age imbalance is is critical when exploring the impact of FLC on survival. Figure \ref{fig:data} shows the trend in free light chain values as a function of age. For illustration of linear models using factors, we have also created a categorical age value using deciles of age. The table of counts shows that the sex distribution becomes increasingly unbalanced at the older ages, from about 1/2 females in the youngest group to a 4:1 ratio in the oldest. <>= library(survival) library(splines) age2 <- cut(flchain$age, c(49, 59, 69, 79, 89, 120), labels=c("50-59", "60-69", "70-79", "80-89", "90+")) counts <- with(flchain, table(sex, age2)) counts # flchain$flc <- flchain$kappa + flchain$lambda male <- (flchain$sex=='M') mlow <- with(flchain[male,], smooth.spline(age, flc)) flow <- with(flchain[!male,], smooth.spline(age, flc)) plot(flow, type='l', ylim=range(flow$y, mlow$y), xlab="Age", ylab="FLC") lines(mlow, col=2) cellmean <- with(flchain, tapply(flc, list(sex, age2), mean, na.rm=T)) matpoints(c(55,65,75, 85, 95), t(cellmean), pch='fm', col=1:2) round(cellmean, 2) @ Notice that the male/female difference in FLC varies with age, \Sexpr{round(cellmean[1,1],1)} versus \Sexpr{round(cellmean[2,1],1)} at age 50--59 and \Sexpr{round(cellmean[1,5],1)} versus \Sexpr{round(cellmean[2,5],1)} at age 90. The data does not fit a simple additive model; there are ``interactions'' to use statistical parlance. An excess of free light chain is thought to be at least partly a reflection of immune senescence, and due to our hormonal backgrounds men and women simply do not age in quite the same way. \section{Population averages} The question of how to test for a main effect in the presence of interaction is an old one. At one time this author considered the phrase ``main effect in the presence of interaction'' to be an oxymoron, but long experience with clinical data sets has led me to the opposite conclusion. Real data always has interactions. The treatment effect of a drug will not be exactly the same for old and young, thin and obese, physically active and sedentary, etc. Explicit recognition of this is an underlying rationale of the current drive towards ``personalized medicine'', though that buzzword often focuses only on genetic differences. Any given data set may often be too small to explore these variations and our statistical models will of necessity smooth over the complexity, but interactions are nevertheless still present. Consider the data shown in figure \ref{fig1} below, which shows a particular laboratory test value by age and sex. We see that the sex effect varies by age. Given this, what could be meant by a ``main effect'' of sex? One sensible approach is to select a fixed \emph{population} for the ages, and then compute the average sex effect over that population. Indeed this is precisely what many computations do behind the scenes, e.g. the ``type 3'' estimates found in linear models. There are three essential components to the calculation: a reference population for the confounders, a summary measure of interest, and a computational algorithm. To understand how linear models methods may (or may not) extend to the proportional hazards model it is useful consider all three facets; each is revealing. Four possible choices for a target population of ages are given below. \begin{enumerate} \item Empirical: the age distribution of the sample at hand, also called the data distribution. In our sample this would be the age distribution of all \Sexpr{nrow(flchain)} subjects, ignoring sex. \item SAS: a uniform distribution is assumed over all categorical adjusters, and the data distribution for continuous ones. \item External reference: a fixed external population, e.g. the age distribution of the US 2010 census. \item MVUE: minimum variance unbiased; the implicit population corresponding to a multivariate least squares fit. \end{enumerate} Method 3 is common in epidemiology, method 1 is found in traditional survey sampling and in other common cases as we will see below. The type 3 estimates of SAS correspond to population 2. If there an interaction between two categorical variables x1 and x2, then the uniform distribution is taken to be over all combinations formed by the pair, and similarly for higher order interactions. \section{Linear models and populations} If we ignore the age effect, then everyone agrees on the best estimate of mean FLC: the simple average of FLC values within each sex. The male-female difference is estimated as the difference of these means. This is what is obtained from a simple linear regression of FLC on sex. Once we step beyond this and adjust for age, the relevant linear models can be looked at in several ways; we will explore three of them below: contrasts, case weights, and nesting. This ``all roads lead to Rome'' property of linear models is one of their fascinating aspects, at least mathematically. \subsection{Case weights} \begin{figure} \myfig{pop} \caption{Three possible adjusting populations for the FLC data set, a empirical reference in black, least squares based one in red, and the US 2000 reference population as `u'.} \label{fig:pop} \end{figure} How do we form a single number summary of ``the effect of sex on FLC''? Here are four common choices. \begin{enumerate} \item Unadjusted. The mean for males minus the mean for females. The major problem with this is that a difference in age distributions will bias the result. Looking at figure \ref{fig:data} imagine that this were two treatments A and B rather than male/female, and that the upper one had been given to predominantly 50-65 year olds and the lower predominantly to subjects over 80. An unadjusted difference would actually reverse the true ordering of the curves. \item Population adjusted. An average difference between the curves, weighted by age. Three common weightings are \begin{enumerate} \item External reference. It is common practice in epidemiology to use an external population as the reference age distribution, for instance the US 2000 census distribution. This aids in comparing results between studies. \item Empirical population. The overall population structure of the observed data. \item Least squares. The population structure that minimizes the variance of the estimated female-male difference. \end{enumerate} \end{enumerate} The principle idea behind case weights is to reweight the data such that confounders become balanced, i.e., ages are balanced when examining the sex effect and sex is balanced when examining age. Any fitted least squares estimate can be rewritten as a weighted sum of the data points with weight matrix $W= (X'X)^{-1}X'$. $W$ has $p$ rows, one per coefficient, each row is the weight vector for the corresponding element of $\hat\beta$. So we can backtrack and see what population assumption was underneath any given fit by looking at the weights for the relevant coefficient(s). Consider the two fits below. In both the second coefficient is an estimate of the overall difference in FLC values between the sexes. (The relationship in figure \ref{fig:data} is clearly curved so we have foregone the use of a simple linear term for age; there is no point in fitting an obviously incorrect model.) Since $\beta_2$ is a contrast the underlying weight vectors have negative values for the females and positive for the males. <<>>= us2000 <- rowSums(uspop2[51:101,,'2000']) fit1 <- lm(flc ~ sex, flchain, x=TRUE) fit2 <- lm(flc ~ sex + ns(age,4), flchain, x=TRUE) c(fit1$coef[2], fit2$coef[2]) wt1 <- solve(t(fit1$x)%*%fit1$x, t(fit1$x))[2,] # unadjusted wt2 <- solve(t(fit2$x)%*%fit2$x, t(fit2$x))[2,] # age-adjusted table(wt1, flchain$sex) @ To reconstruct the implied population density, one can use the density function with \code{wt1} or \code{wt2} as the case weights. Examination of \code{wt1} immediately shows that the values are $-1/n_f$ for females and $1/n_m$ for males where $n_f$ and $n_m$ are number of males and females, respectively. The linear model \code{fit1} is the simple difference in male and female means; the implied population structure for males and females is the unweighted density of each. Because this data set is very large and age is coded in years we can get a density estimate for fit2 by simple counting. The result is coded below and shown in figure \ref{fig:pop}. The empirical reference and least squares reference are nearly identical. This is not a surprise. Least squares fits produce minimum variance unbiased estimates (MVUE), and the variance of a weighted average is minimized by using weights proportional to the sample size, thus the MVUE estimate will give highest weights to those ages with a lot of people. The weights are not \emph{exactly} proportional to sample size for each age. As we all know, for a given sample size $n$ a study comparing two groups will have the most power with equal allocation between the groups. Because the M/F ratio is more unbalanced at the right edge of the age distribution the MVUE estimate gives just a little less weight there, but the difference between it and the overall data set population will be slight for all but those pathological cases where there is minimal overlap between M/F age distributions. (And in that case the entire discussion about what ``adjustment'' can or should mean is much more difficult.) <>= us2000 <- rowSums(uspop2[51:101,,'2000']) tab0 <- table(flchain$age) tab2 <- tapply(abs(wt2), flchain$age, sum) matplot(50:100, cbind(tab0/sum(tab0), tab2/sum(tab2)), type='l', lty=1, xlab="Age", ylab="Density") us2000 <- rowSums(uspop2[51:101,,'2000']) matpoints(50:100, us2000/sum(us2000), pch='u') legend(60, .02, c("Empirical reference", "LS reference"), lty=1, col=1:2, bty='n') @ The LS calculation does a population adjustment automatically for us behind the scenes via the matrix algebra of linear models. If we try to apply population reference adjustment directly a problem immediately arises: in the US reference \Sexpr{round(100*us2000[46]/sum(us2000),2)}\% of the population is aged 95 years, and our sample has no 95 year old males; it is not possible to re weight the sample so as to exactly match the US population reference. This occurs in any data set that is divided into small strata. The traditional epidemiology approach to this is to use wider age intervals of 5 or 10 years. Weights are chosen for each age/sex strata such that the sum of weights for females = sum of weights for males within each age group (balance), and the total sum of weights in an age group is equal to the reference population. The next section goes into this further. An increasingly popular approach for producing results that are standardized to the empirical reference population (i.e. the data distribution) is to use a smoothed age effect, obtained through inverse probability weights which are based on logistic regression, e.g. in the causal models literature and propensity score literature. This approach is illustrated in a vignette on adjusted survival curves which is also in the survival package. \subsection{Categorical predictors and contrasts} When the adjusting variable or variables are categorical --- a factor in R or a class variable in SAS --- then two more aspects come into play. The first is that any estimate of interest can be written in terms of the cell means. Formally, the cell means are a \emph{sufficient statistic} for the data. For our data set and using the categorized variable \code{age2} let $\theta_{ij}$ parameterize these means. $$ \begin{tabular}{cccccc} &50--59 & 60--69 & 70-79 & 80-89 & 90+ \\ \hline Female & $\theta_{11}$ & $\theta_{12}$ & $\theta_{13}$& $\theta_{14}$& $\theta_{15}$ \\ Male & $\theta_{21}$ & $\theta_{22}$ & $\theta_{23}$& $\theta_{24}$ & $\theta_{25}$ \\ \end{tabular} $$ For a design with three factors we will have $\theta_{ijk}$, etc. Because it is a sufficient statistic, any estimate or contrast of interest can be written as a weighted sum of the $\theta$s. Formulas for the resulting estimates along with their variances and tests were worked out by Yates in 1934 \cite{Yates34} and are often referred to as a Yates weighted means estimates. For higher order designs the computations can be rearranged in a form that is manageable on a desk calculator, and this is in fact the primary point of that paper. (Interestingly, his computational process turns out to be closely related to the fast Fourier transform.) The second facet of categorical variables is that another adjustment is added to the list of common estimates: \begin{enumerate} \item Unadjusted \item Population adjusted \begin{enumerate} \item External reference \item Empirical (data set) reference \item Least squares \item Uniform. A population in which each combination of the factors has the same frequency of occurrence. \end{enumerate} \end{enumerate} The uniform population plays a special role in the case of designed experiments, where equal allocation corresponds to the optimal study design. The Yates estimates are particularly simple in this case. For a hypothetical population with equal numbers in each age category the estimated average FLC for females turns out to be $\mu_f = \sum_j \theta_{1j} /5$ and the male - female contrast is $\sum_j(\theta_{2j}-\theta_{1j})/5$. We will refer to these as the ``Yates'' estimates and contrast for an effect. Conversely, the estimated age effects, treating sex as a confounding effect and assuming an equal distribution of females and males as the reference population, gives an estimated average FLC for the 60-69 year olds of $\mu_{60-69}= (\theta_{12} + \theta_{22})/2$, and etc for the other age groups. We can obtain the building blocks for Yates estimates by using the interaction function and omitting the intercept. <>= yatesfit <- lm(flc ~ interaction(sex, age2) -1, data=flchain) theta <- matrix(coef(yatesfit), nrow=2) dimnames(theta) <- dimnames(counts) round(theta,2) @ For a linear model fit, any particular weighted average of the coefficients along with its variance and the corresponding sums of squares can be computed using the \code{contrast} function given below. Let $C$ be a contrast matrix with $k$ rows, each containing one column per coefficient. Then $C\theta$ is a vector of length $k$ containing the weighted averages and $V = \hat\sigma^2 C (X'X)^{-1}C'$ is its variance matrix. The sums of squares is the increase in the sum of squared residuals if the fit were restricted to the subspace $C\theta =0$. Formulas are from chapter 5 of Searle \cite{Searle71}. Some authors reserve the word \emph{contrast} for the case where each row of $C$ sums to zero and use \emph{estimate} for all others; I am being less restrictive since the same computation serves for both. <<>>= qform <- function(beta, var) # quadratic form b' (V-inverse) b sum(beta * solve(var, beta)) contrast <- function(cmat, fit) { varmat <- vcov(fit) if (class(fit) == "lm") sigma2 <- summary(fit)$sigma^2 else sigma2 <- 1 # for the Cox model case beta <- coef(fit) if (!is.matrix(cmat)) cmat <- matrix(cmat, nrow=1) if (ncol(cmat) != length(beta)) stop("wrong dimension for contrast") estimate <- drop(cmat %*% beta) #vector of contrasts ss <- qform(estimate, cmat %*% varmat %*% t(cmat)) *sigma2 list(estimate=estimate, ss=ss, var=drop(cmat %*% varmat %*% t(cmat))) } yates.sex <- matrix(0, 2, 10) yates.sex[1, c(1,3,5,7,9)] <- 1/5 #females yates.sex[2, c(2,4,6,8,10)] <- 1/5 #males contrast(yates.sex, yatesfit)$estimate # the estimated "average" FLC for F/M contrast(yates.sex[2,]-yates.sex[,1], yatesfit) # male - female contrast @ <>= # Create the estimates table -- lots of fits emat <- matrix(0., 6, 3) dimnames(emat) <- list(c("Unadjusted", "MVUE: continuous age", "MVUE: categorical age", "Empirical (data) reference", "US200 reference", "Uniform (Yates)"), c("est", "se", "SS")) #unadjusted emat[1,] <- c(summary(fit1)$coef[2,1:2], anova(fit1)["sex", "Sum Sq"]) # MVUE -- do the two fits fit2 <- lm(flc ~ ns(age,4) + sex, flchain) emat[2,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) fit2 <- lm(flc ~ age2 + sex, flchain) emat[3,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) #Remainder, use contrasts tfun <- function(wt) { cvec <- c(matrix(c(-wt, wt), nrow=2, byrow=TRUE)) temp <- contrast(cvec, yatesfit) c(temp$est, sqrt(temp$var), temp$ss) } emat[4,] <- tfun(colSums(counts)/sum(counts)) usgroup <- tapply(us2000, rep(1:5, c(10,10,10,10,11)), sum)/sum(us2000) emat[5,]<- tfun(usgroup) emat[6,] <- tfun(rep(1/5,5)) @ \begin{table} \centering \begin{tabular}{l|ccc} & estimate & sd & SS \\ \hline <>= temp <- dimnames(emat)[[1]] for (i in 1:nrow(emat)) cat(temp[i], sprintf(" &%5.3f", emat[i,1]),sprintf(" &%6.5f", emat[i,2]), sprintf(" & %6.1f", emat[i,3]), "\\\\ \n") @ \end{tabular} \caption{Estimates of the male-female difference along with their standard errors. The last 4 rows are based on categorized age.} \label{tab:allest} \end{table} Table \ref{tab:est} shows all of the estimates of the male/female difference we have considered so far along with their standard errors. Because it gives a much larger weight to the 90+ age group than any of the other estimates, and that group has the largest M-F difference, the projected difference for a uniform population (Yates estimate) yields the largest contrast. It pays a large price for this in terms of standard error, however, and is over twice the value of the other approaches. As stated earlier, any least squares parameter estimate can be written as a weighted sum of the y values. Weighted averages have minimal variance when all of the weights are close to 1. The unadjusted estimate adheres to this precisely and the data-reference and MVUE stay as close as possible to constant weights, subject to balancing the population. The Yates estimate, by treating every cell equally, implicitly gives much larger weights to the oldest ages. Table \ref{tab:est} shows the effective observation weights used for each of the age categories. <>= casewt <- array(1, dim=c(2,5,4)) # case weights by sex, age group, estimator csum <- colSums(counts) casewt[,,2] <- counts[2:1,] / rep(csum, each=2) casewt[,,3] <- rep(csum, each=2)/counts casewt[,,4] <- 1/counts #renorm each so that the mean weight is 1 for (i in 1:4) { for (j in 1:2) { meanwt <- sum(casewt[j,,i]*counts[j,])/ sum(counts[j,]) casewt[j,,i] <- casewt[j,,i]/ meanwt } } @ \begin{table} \centering \begin{tabular}{rlrrrrr} &&50--59& 60--69 & 70--79 & 80--89 & 90+ \\ \hline <>= tname <- c("Unadjusted", "Min var", "Empirical", "Yates") for (i in 1:2) { for (j in 1:4) { cat("&",tname[j], " & ", paste(sprintf("%4.2f", casewt[i,,j]), collapse= " & "), "\\\\\n") if (j==1) cat(c("Female", "Male")[i]) } if (i==1) cat("\\hline ") } @ \end{tabular} \caption{Observation weights for each data point corresponding to four basic approaches. All weights are normed so as to have an average value of 1.} \label{tab:est} \end{table} Looking at table \ref{tab:est} notice the per observation weights for the $\ge 90$ age group, which is the one with the greatest female/male imbalance in the population. For all but the unbalanced estimate (which ignores age) the males are given a weight that is approximately 3 times that for females in order to re balance the shortage of males in that category. However, the absolute values of the weights differ considerably. \subsection{Different codings} Because the cell means are a sufficient statistic, all of the estimates based on categorical age can be written in terms of the cell means $\hat\theta$. The Yates contrast is the simplest to write down: $$ \begin{tabular} {rrrrrr} & 50--59 & 60--69 & 70--79 & 80--89 & 90+ \\ \hline Female & -1/5 & -1/5 & -1/5 & -1/5 & -1/5 \\ Male & 1/5 & 1/5 & 1/5 & 1/5 & 1/5 \end{tabular} $$ %(Note that for calculating a sum of squares we will get the exact same %result from a matrix using $\pm 1$ rather than $\pm 1/5$; %the Yates contrast is often written this way.) For the data set weighting the values of 1/5 are replaced by $n_{+j}/n_{++}$, the overall frequency of each age group, where a $+$ in the subscript stands for addition over that subscript in the table of counts. The US population weights use the population frequency of each age group. The MVUE contrast has weights of $w_j/\sum w_j$ where $w_j = 1/(1/n_{1j} + 1/n_{2j})$, which are admittedly not very intuitive. $$ \begin{tabular}{rrrrrr} & 50--59 & 60--69 & 70--79 & 80--89 & 90+ \\ \hline <>= temp <- 1/colSums(1/counts) temp <- temp/sum(temp) cat("Female", sprintf(" & %5.3f", -temp), "\\\\ \n") cat("Male", sprintf(" & %5.3f", temp), "\\\\ \n") @ \end{tabular} $$ In the alternate model \code{y \textasciitilde sex + age2} the MVUE contrast is much simpler, namely (0, 1, 0,0,0,0,0), and can be read directly off the printout as $\beta/se(\beta)$. The computer's calculation of $(X'X)^{-1}$ has derived the ``complex'' MVUE weights for us without needing to lift a pencil. The Yates contrast, however, cannot be created from the coefficients of the simpler model at all. This observation holds in general: a contrast that is simple to write down in one coding may appear complicated in another, or not even be possible. The usual and more familiar coding for a two way model is \begin{equation} y_{ij} = \mu + \alpha_i + \beta_j + \gamma_{ij} \label{std} \end{equation} What do the Yates' estimates look like in this form? Let $e_i$ be the Yates estimate for row $i$ and $k$ the number of columns in the two way table of $\theta$ values. Then \begin{align*} e_i &= (1/k)\sum_{j=1}^k \theta_{ij} \\ &= \mu + \alpha_i + \sum_j \left(\beta_j + \gamma_{ij}\right)/k \end{align*} and the Yates test for row effect is \begin{align} 0 &= e_i - e_{i'} \quad \forall i,i' \nonumber \\ &= (\alpha_i - \alpha_{i'}) + (1/k)\sum_j(\gamma_{ij} - \gamma_{i'j}) \label{ycont} \end{align} Equation \eqref{std} is over determined and all computer programs add constraints in order to guarantee a unique solution. However those constraints are applied, however, equation \eqref{ycont} holds. The default in R is treatment contrasts, which use the first level of any factor as a reference level. Under this constraint the reference coefficients are set to zero, i.e., all coefficients of equations \eqref{std} and \eqref{ycont} above where $i=1$ or $j=1$. We have been computing the male - female contrast, corresponding to $i=2$ and $i'=1$ in equation \eqref{ycont}, and the Yates contrast for sex becomes $\alpha_2 + 1/5(\gamma_{22} +\gamma_{23} +\gamma_{24} +\gamma_{25})$. The code below verifies that this contrast plus the usual R fit replicates the results in table \ref{tab:allest}. <>= fit3 <- lm(flc ~ sex * age2, flchain) coef(fit3) contrast(c(0,1, 0,0,0,0, .2,.2,.2,.2), fit3) #Yates @ The usual constraint is SAS is to use the last level of any class variable as the reference group, i.e., all coefficients with $i=2$ or $j=5$ in equations \eqref{std} and \eqref{ycont} are set to zero. <>= options(contrasts=c("contr.SAS", "contr.poly")) sfit1 <- lm(flc ~ sex, flchain) sfit2 <- lm(flc ~ sex + age2, flchain) sfit3 <- lm(flc ~ sex * age2, flchain) contrast(c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2), sfit3) # Yates for SAS coding @ The appendix contains SAS code and output for the three models \code{sfit1, sfit2} and \code{sfit3} above. The \code{E3} option was added to the SAS model statements, which causes a symbolic form of the contrasts that were used for ``type III'' results to be included in the printout. Look down the column labeled ``SEX'' and you will see exactly the coefficients used just above, after a bit of SAS to English translation. \begin{itemize} \item The SAS printout is labeled per equation \eqref{std}, so L1= column 1 of the full $X$ matrix = intercept. L2 = column 2 = females, L3 = column 3 = males, L4= column 4 = age 50--59, etc. \item In the symbolic printout they act as though sum constraints were in force: the last column of age is labeled with a symbolic value that would cause the age coefficients to sum to zero. However, in actuality these coefficients are set to zero. The table of parameter estimates at the end of the printout reveals this; forced zeros have a blank for their standard error. \item When calculating the contrast one can of course skip over the zero coefficients, and the R functions do not include them in the coefficient vector. Remove all of these aliased rows from the SAS symbolic printout to get the actual contrast that is used; this will agree with my notation. \item The SAS printout corresponds to a female-male contrast and I have been using male-female for illustration. This changes the signs of the contrast coefficients but not the result. \end{itemize} The \code{estimate} statement in the SAS code required that all of the coefficients be listed, even the aliased ones (someone more proficient in SAS may know a way to avoid this and enter only the non-aliased values.) %A general principle is that a given hypothesis may be represented as %a simple contrast in one coding but be complex in another. %The unadjusted test is a trivial contrast in the sfit1 coding, but a %complex one in the sfit3 coding. %The Yates test cannot be expressed as a contrast using the sfit1 or sfit2 %coding, is simple and obvious in the cell means coding, and has %simple but non obvious coefficients in the sfit3 coding. %Que sera sera. So, how do we actually compute the Yates contrast in a computer program? We will take it as a give that no one wants to memorize contrast formulas. Appendix \ref{sect:coding} describes three algorithms for the computation. One of these three (NSTT) is completely unreliable, but is included because it is so often found in code. If one uses the sum constraints commonly found in textbooks, which corresponds to the \code{contr.sum} constraint in R and to \code{effect} constraints in SAS, and there are no missing cells, then the last term in equation \eqref{ycont} is zero and the simple contrast $\alpha_i =0$ will be equal to the Yates contrast for sex. I often see this method recommended on R help in response to the question of ``how to obtain type III'', computed either by use of the \code{drop1} command or the \code{Anova} function found within the car package, but said advice almost never mentions the need for this particular non-default setting of the contrasts option\footnote{The Companion to Applied Regression (car) package is designed to be used with the book of the same name by John Fox, and the book does clarify the need for sum constraints.}. When applied to other codings the results of this procedure can be surprising. <>= options(contrasts = c("contr.treatment", "contr.poly")) #R default fit3a <- lm(flc ~ sex * age2, flchain) options(contrasts = c("contr.SAS", "contr.poly")) fit3b <- lm(flc~ sex * age2, flchain) options(contrasts=c("contr.sum", "contr.poly")) fit3c <- lm(flc ~ sex * age2, flchain) # nstt <- c(0,1, rep(0,8)) #test only the sex coef = the NSTT method temp <- rbind(unlist(contrast(nstt, fit3a)), unlist(contrast(nstt, fit3b)), unlist(contrast(nstt, fit3c)))[,1:2] dimnames(temp) <- list(c("R", "SAS", "sum"), c("effect", "SS")) print(temp) # drop1(fit3a, .~.) @ For the case of a two level effect such as sex, the NSTT contrast under the default R coding is a comparison of males to females in the first age group \textbf{only}, and under the default SAS coding it is a comparison of males to females within the \textbf{last} age group. Due to this easy creation of a test statistic which has no relation to the global comparison one expects from the ``type 3'' label the acronym \emph{not safe type three}(NSTT) was chosen, ``not SAS'' and ``nonsense'' are alternate mnemonics. \subsection{Sums of squares and projections} \label{sect:anova} The most classic exposition of least squares is as a set of projections, each on to a smaller space. Computationally we represent this as a series of model fits, each fit summarized by the change from the prior fit in terms of residual sum of squares. <>= options(show.signif.stars = FALSE) #exhibit intelligence sfit0 <- lm(flc ~ 1, flchain) sfit1b <- lm(flc ~ age2, flchain) anova(sfit0, sfit1b, sfit2, sfit3) @ The second row is a test for the age effect. The third row of the above table summarizes the improvement in fit for the model with sex + age2 over the model with just age2, a test of ``sex, adjusted for age''. This test is completely identical to the minimum variance contrast, and is in fact the way in which that SS is normally obtained. The test for a sex effect, unadjusted for age, is identical to an anova table that compares the intercept-only fit to one with sex, i.e., the second line from a call to \code{anova(sfit0, sfit1)}. The anova table for a nested sequence of models $A$, $A+B$, $A + B +C$, \ldots has a simple interpretation, outside of contrasts or populations, as an improvement in fit. Did the variable(s) $B$ add significantly to the goodness of fit for a model with just $A$, was $C$ an important addition to a model that already includes $A$ and $B$? The assessment of improvement is based on the likelihood ratio test (LRT), and extends naturally to all other models based on likelihoods. The tests based on a target population (external, data population, or Yates) do not fit naturally into this approach, however. %Obtaining the Yates contrast using a sequential sums of squares approach %is possible but a bit contrived. %Our final fit in the table will be \code{sfit3}, but %the one prior to it needs to be from a constrained version of \code{sfit3}, %whose solution lies in the space spanned by the Yates contrast %$\beta_2 + \beta_7/5 + \beta_8/5 + \beta_9/5 + \beta_{10}/5 = 0$. %There is no simple way to write down an ordinary LS model equation that %will do this, and instead one must use one a program for constrained %linear regression; these are far less familiar. %There are many algorithms to fit a constrained linear regression, one is %to transform the problem as $X\beta = (XQ)(Q'\beta) = Z \phi$ %where $Q$ is an orthogonal transformation matrix. %If the first column of $Q$ is chosen as a scaled version of the Yates %contrast, then setting that contrast equal to zero is the same as %the constraint $\phi_1 =0$; it suffices to fit a model using all but the %first column of $Z$. \subsection{What is SAS type 3?} We are now in a position to fully describe the SAS sums of squares. \begin{itemize} \item Type 1 is the output of the ANOVA table, where terms are entered in the order specified in the model. \item Type 2 is the result of a two stage process \begin{enumerate} \item Order the terms by level: 0= intercept, 1= main effects, 2= 2 way interactions, \ldots. \item For terms of level k, print the MVUE contrast from a model that includes all terms of levels $0-k$. Each of these will be equivalent to the corresponding line of a sequential ANOVA table where the term in question was entered as the last one of its level. \end{enumerate} \item Type 3 and 4 are also a 2 stage process \begin{enumerate} \item Segregate the terms into those for which a Yates contrast can be formed versus those for which it can not. The second group includes the intercept, any continuous variables, and any factor (class) variables that do not participate in interactions with other class variables. \item For variables in the first group compute Yates contrasts. For those in the second group compute the type 2 results. \end{enumerate} \end{itemize} SAS has two different algorithms for computing the Yates contrast, which correspond to the \code{ATT} and \code{STT} options of the \code{yates} function. SAS describes the two contrast algorithms in their document ``The four types of estimable functions'' \cite{SASguide}, one of which defines type 3 and the other type 4. I found it very challenging to recreate their algorithm from this document. Historical knowledge of the underlying linear model algorithms used by SAS is a useful and almost necessary adjunct, as many of the steps in the document are side effects of their calculation. When there are missing cells, then it is not possible to compute a contrast that corresponds to a uniform distribution over the cells, and thus the standard Yates contrast is also not defined. The SAS type 3 and 4 algorithms still produce a value, however. What exactly this result ``means'' and whether it is a good idea has been the subject of lengthy debates which I will not explore here. Sometimes the type 3 and type 4 algorithms will agree but often do not when there are missing cells, which further muddies the waters. Thus we have 3 different tests: the MVUE comparison which will be close but not exactly equal to the data set population, Yates comparisons which correspond to a uniform reference population, and the SAS type 3 (STT) which prints out a chimeric blend of uniform population weighting for those factor variables that participate in interactions and the MVUE weighting for all the other terms. \subsection{Which estimate is best?} Deciding which estimate is the best is complicated. Unfortunately a lot of statistical textbooks emphasize the peculiar situation of balanced data with exactly the same number of subjects in each cell. Such data is \emph{extremely} peculiar if you work in medicine; in 30 years work and several hundred studies I have seen 2 instances. In this peculiar case the unadjusted, MVUE, empirical reference and Yates populations are all correspond to a uniform population and so give identical results. No thinking about which estimate is best is required. This has led many to avoid the above question, instead pining for that distant Eden where the meaning of ``row effect'' is perfectly unambiguous. But we are faced with real data and need to make a choice. The question has long been debated in depth by wiser heads than mine. In a companion paper to his presentation at the joint statistical meetings in 1992, Macnaughton \cite{Macnaughton92} lists 54 references to the topic between 1952 and 1991. Several discussion points recur: \begin{enumerate} \item Many take the sequential ANOVA table as primary, i.e., a set of nested models along with likelihood ratio tests (LRT), and decry all comparisons of ``main effects in the presence of interaction.'' Population weightings other than the LS one do not fit nicely into the nested framework. \item Others are distressed by the fact that the MVUE adjusting population is data dependent, so that one is ``never sure exactly what hypothesis being tested''. \item A few look at the contrast coefficients themselves, with a preference for simple patterns since they ``are interpretable''. \item No one approach works for all problems. Any author who proposes a uniform rule is quickly presented with counterexamples. \end{enumerate} Those in group 1 argue strongly against the Yates weighting and those in group 2 argue for the Yates contrast. Group 3 is somewhat inexplicable to me since any change in the choice of constraint type will change all the patterns. I fear that an opening phrase from the 1986 overview/review of Herr \cite{Herr86} is still apropos, ``In an attempt to understand how we have arrived at our present state of ignorance \ldots''. There are some cases where the Yates approach is clearly sensible, for instance a designed experiment which has become unbalanced due to a failed assay or other misadventure that has caused a few data points to be missing. There are cases such as the FLC data where the Yates contrast makes little sense at all --- the hypothetical population with equal numbers of 50 and 90 year olds is one that will never be seen--- so it is rather like speculating on the the potential covariate effect in dryads and centaurs. The most raucous debate has circled around the case of testing for a treatment effect in the presence of multiple enrolling centers. Do we give each patient equal weight (MVUE) or each center equal weight (Yates). A tongue-in-cheek but nevertheless excellent commentary on the subject is given by the old curmudgeon, aka Guernsey McPearson \cite{Senn1, Senn2}. A modern summary with focus on the clinical trials arena is found in chapter 14 of the textbook by Senn \cite{Senn07} I have found two papers particularly useful in thinking about this. Senn \ref{Senn00} points out the strong parallels between tests for main effects when there may be interactions and meta analyses, cross connecting these two approaches is illuminating. A classic reference is the 1978 paper by Aitkin \cite{Aitkin78}. This was read before the Royal Statistical Society and includes remarks by 10 discussants forming a who's who of statistical theory (F Yates, J Nelder, DR Cox, DF Andrews, KR Gabriel, \ldots). The summary of the paper states that ``It is shown that a standard method of analysis used in many ANOVA programs, equivalent to Yates method of weighted squares of means, may lead to inappropriate models''; the paper goes on to carefully show why no one method can work in all cases. Despite the long tradition among RSS discussants of first congratulating the speaker and then skewering every one their conclusions, not one defense of the always-Yates approach is raised! This includes the discussion by Yates himself, who protests that his original paper advocated the proposed approach with reservations, it's primary advantage being that the computations could be performed on a desk calculator. I have two primary problems with the SAS type 3 approach. The first and greatest is that their documentation recommends the method with no reference to this substantial and sophisticated literature discussing strengths and weaknesses of the Yates contrast. This represents a level of narcissism which is completely unprofessional. %Recommending the type III approach as best for all cases, as they do, has %caused actual harm. The second is that their documentation explains the method is a way that is almost impenetrably opaque. If this is the only documentation one has, there will not be 1 statistician in 20 who would be able to explain the actual biological hypothesis which is being addressed by a type 3 test. \section{Cox models} \subsection{Tests and contrasts} Adapting the Yates test to a Cox model is problematic from the start. First, what do we mean by a ``balanced population''? In survival data, the variance of the hazard ratio for each particular sex/age combination is proportional to the number of deaths in that cell rather than the number of subjects. Carrying this forward to the canonical problem of adjusting a treatment effect for enrolling center, does this lead to equal numbers of subjects or equal numbers of events? Two centers might have equal numbers of patients but different number of events because one initiated the study at a later time (less follow up per subject), or it might have the same follow up time but a lower death rate. Should we reweight in one case (which one), both, or neither? The second issue is that the per-cell hazard ratio estimates are no longer a minimally sufficient statistic, so underlying arguments about a reference population no longer directly translate into a contrast of the parameters. A third but more minor issue is that the three common forms of the test statistic --- Wald, score, and LRT --- are identical in a linear model but not for the Cox model, so which should we choose? To start, take a look at the overall data and compute the relative death rates for each age/sex cell. <>= options(contrasts= c("contr.treatment", "contr.poly")) # R default cfit0 <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) cmean <- matrix(c(0, coef(cfit0)), nrow=2) cmean <- rbind(cmean, cmean[2,] - cmean[1,]) dimnames(cmean) <- list(c("F", "M", "M/F ratio"), dimnames(counts)[[2]]) signif(exp(cmean),3) @ Since the Cox model is a relative risk model all of the death rates are relative to one of the cells, in this case the 50--59 year old females has been arbitrarily chosen as the reference cell and so has a defined rate of 1.00. Death rates rise dramatically with age for both males and females (no surprise), with males always slightly ahead in the race to a coffin. The size of the disadvantage for males decreases in the last 2 decades, however. The possible ways to adjust for age in comparing the two sexes are \begin{enumerate} \item The likelihood ratio test. This is analogous to the sequential ANOVA table in a linear model, and has the strongest theoretical justification. \item A stratified Cox model, with age group as the stratification factor. This gives a more general and rigorous adjustment for age. Stratification on institution is a common approach in clinical trials. \item The Wald or score test for the sex coefficient, in a model that adjusts for age. This is analogous to Wald tests in the linear model, and is asymptotically equivalent the the LRT. \item The test from a reweighted model, using case weights. Results using this approach have been central to causal model literature, particularly adjustment for covariate imbalances in observational studies. (Also known as \emph{marginal structural models}). Adjustment to a uniform population is also possible. \item A Yates-like contrast in the Cox model coefficients. \begin{itemize} \item A reliable algorithm such as cell means coding. \item Unreliable approach such as the NSTT \end{itemize} \end{enumerate} I have listed these in order from the most to the least available justification, both in terms of practical experience and available theory. The two standard models are for sex alone, and sex after age. Likelihood ratio tests for these models are the natural analog to anova tables for the linear model, and are produced by the same R command. Here are results for the first three, along with the unadjusted model that contains sex only. <>= options(contrasts=c("contr.SAS", "contr.poly")) cfit1 <- coxph(Surv(futime, death) ~ sex, flchain) cfit2 <- coxph(Surv(futime, death) ~ age2 + sex, flchain) cfit3 <- coxph(Surv(futime, death) ~ sex + strata(age2), flchain) # Unadjusted summary(cfit1) # # LRT anova(cfit2) # # Stratified anova(cfit3) summary(cfit3) # # Wald test signif(summary(cfit2)$coefficients, 3) # anova(cfit1, cfit2) @ Without adjustment for age the LRT for sex is only \Sexpr{round(2*diff(cfit1$loglik),1)}, and after adjustment for %$ a it increases to \Sexpr{round(anova(cfit2)[3,2],2)}. Since females are older, not adjusting for age almost completely erases the evidence of their actual survival advantage. Results of the LRT are unchanged if we change to any of the other possible codings for the factor variables (not shown). Adjusting for age group using a stratified model gives almost identical results to the sequential LRT, in this case. The Wald tests for sex are equal to $[\beta/ se(\beta)]^2$ using the sex coefficient from the fits, \Sexpr{round(summary(cfit1)$coef[1,4]^2,2)} and \Sexpr{round(summary(cfit2)$coef[5,4]^2,2)} for the unadjusted and adjusted models, respectively. Unlike a linear model they are not exactly equal to the anova table results based on the log-likelihood, but tell the same story. Now consider weighted models, with both empirical and uniform distributions as the target age distribution. The fits require use of a robust variance, since we are approaching it via a survey sampling computation. The tapply function creates a per-subject index into the case weight table created earlier. <>= wtindx <- with(flchain, tapply(death, list(sex, age2))) cfitpop <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,3])[wtindx]) cfityates <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,4])[wtindx]) # # Glue it into a table for viewing # tfun <- function(fit, indx=1) { c(fit$coef[indx], sqrt(fit$var[indx,indx])) } coxp <- rbind(tfun(cfit1), tfun(cfit2,5), tfun(cfitpop), tfun(cfityates)) dimnames(coxp) <- list(c("Unadjusted", "Additive", "Empirical Population", "Uniform Population"), c("Effect", "se(effect)")) signif(coxp,3) @ The population estimates based on reweighting lie somewhere between the unadjusted and the sequential results. We expect that balancing to the empirical population will give a solution that is similar to the age + sex model, in the same way that the close but not identical to the MVUE estimate in a linear model. Balancing to a hypothetical population with equal numbers in each age group yields a substantially smaller estimate of effect. since it gives large weights to the oldest age group, where in this data set the male/female difference is smallest. Last, look at constructed contrasts from a cell means model. We can either fit this using the interaction, or apply the previous contrast matrix to the coefficients found above. Since the ``intercept'' of a Cox model is absorbed into the baseline hazard our contrast matrix will have one less column. <<>>= cfit4 <- coxph(Surv(futime, death) ~ sex * age2, flchain) # Uniform population contrast ysex <- c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2) #Yates for sex, SAS coding contrast(ysex[-1], cfit4) # Verify using cell means coding cfit4b <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) temp <- matrix(c(0, coef(cfit4b)),2) # the female 50-59 is reference diff(rowMeans(temp)) #direct estimate of the Yates # temp2 <- rbind(temp, temp[2,] - temp[1,]) dimnames(temp2) <- list(c('female', 'male', 'difference'), levels(age2)) round(temp2, 3) # # # NSTT contrast contrast(c(1,0,0,0,0,0,0,0,0), cfit4) @ In the case of a two level covariate such as sex, the NSTT algorithm plus the SAS coding yields an estimate and test for a difference in sex for the \emph{first} age group; the proper contrast is an average. Since it gives more weight to the larger ages, where the sex effect is smallest, the Yates-like contrast is smaller than the result from an additive model \code{cfit2}. Nevertheless, this contrast and the sequential test are more similar for the survival outcome than for the linear models. This is due to the fact that the variances of the individual hazards for each sex/age combination are proportional to the number of deaths in that cell rather than the number of subjects per cell. A table of the number of deaths is not as imbalanced as the table of subject counts, and so the Yates and MLE ``populations'' are not as far apart as they were for the linear regression. There are fewer subjects at the higher ages but they die more frequently. Why is the Yates-like contrast so different than the result of creating a uniform age distribution using case weights followed by an MLE estimate? Again, the MLE estimate has death counts as the effective weights; the case-weighted uniform population has smaller weights for the youngest age group and that group also has the lowest death rate, resulting in lower influence for that group and an estimate shrunken towards the 90+ difference of \Sexpr{round(temp2[3,5], 3)}. All told, for survival models adjustment to a uniform population is a slippery target. \subsection{SAS phreg results} Now for the main event: what does SAS do? First, for the simple case of an additive model the SAS results are identical to those shown above. The coefficients, variances and log-likelihoods for cfit2 are identical to the phreg output for an additive model, as found in the appendix. As would be expected from the linear models case, the ``type III'' results for the additive model are simply the Wald tests for the fit, repackaged with a new label. Now look at the model that contains interactions. We originally surmised that a contrast calculation would be the most likely way in which the phreg code would implement type 3, as it is the easiest to integrate with existing code. Results are shown in the last SAS fit of the appendix. Comparing these results of the SAS printout labeled as ``Type III Wald'' to the contrasts calculated above shows that phreg is using the NSTT method. This is a bit of a shock. All of the SAS literature on type III emphasizes the care with which they form the calculation so as to always produce a Yates contrast (or in the case of missing cells a Yates-like one), and there was no hint in the documentation that phreg does anything different. As a double check direct contrast statements corresponding to the Yates and NSTT contrasts were added to the SAS code, and give confirmatory results. A further run which forced sum constraints by adding \code{'/ effect'} to the SAS class statement (not shown) restored the correct Yates contrast, as expected. As a final check, look at the NSTT version of the LRT, which corresponds to simply dropping the sex column from the $X$ matrix. <>= xmat4 <- model.matrix(cfit4) cfit4b <- coxph(Surv(futime, death) ~ xmat4[,-1], flchain) anova(cfit4b, cfit4) @ This agrees with the LR ``type 3'' test of the phreg printout. \subsection{Conclusion} Overall, both rebalanced estimates and coefficient contrasts are interesting exercises for the Cox model, but their actual utility is unclear. It is difficult to make a global optimality argument for either one, particularly in comparison to the sequential tests which have the entire weight of likelihood theory as a justification. Case reweighted estimates do play a key role when attempting to adjust for non-random treatment assignment, as found in the literature for causal analysis and marginal structural models; a topic and literature far too extensive and nuanced for discussion in this note. No special role is apparent, at least to this author, for regular or even sporadic use of a Yates contrast in survival models. The addition of such a feature and label to the SAS phreg package is a statistical calamity, one that knowledgeable and conscientious statistical practitioners will likely have to fight for the rest of their careers. In the common case of a treatment comparison, adjusted for enrolling center, the default ``type III'' printout from phreg corresponds to a comparison of treatments within the last center; the only contribution of the remainder of the data set is to help define the baseline hazard function and the effect of any continuous adjusters that happen to be in the model. The quadruple whammy of a third rate implementation (the NSTT), defaults that lead to a useless and misleading result, no documentation of the actual computation that is being done, and irrational reverence for the type III label conspire to make this a particularly unfortunate event. \appendix \section{Computing the Yates estimate} \label{sect:coding} We will take it as a given that no one wants to memorize contrast formulas, and so we need a way to compute Yates contrasts automatically in a computer program. The most direct method is to encode the original fit in terms of the cell means, as has been done throughout this report. The Yates contrast is then simply an average of estimates across the appropriate margin. However, we normally will want to solve the linear or Cox model fit in a more standard coding and then compute the Yates contrast after the fact. Note that any population re norming requires estimates of the cell means, whether they were explicit parameters or not, i.e., the model fit must include interaction terms. Here are three algorithms for this post-hoc computation. All of them depend, directly or indirectly, on the breakdown found earlier in equation \eqref{std}. \begin{align} y_{ij} &= \mu + \alpha_i + \beta_j + \gamma_{ij} + \epsilon \label{a1} \\ &= \theta_{ij} + \epsilon \label{a2}\\ \theta_{ij} &= \mu + \alpha_i + \beta_j + \gamma_{ij} \label{a3} \\ \end{align} Equation \eqref{a1} is the standard form from our linear models textbooks, equation \eqref{a2} is the cell means form, and \eqref{a3} is the result of matching them together. Using this equivalence a Yates test for row effects will be \begin{align} 0 &= e_i - e_{i'} \quad \forall i,i' \nonumber \\ &= (\alpha_i - \alpha_{i'}) + (1/k)\sum_j(\gamma_{ij} - \gamma_{i'j}) \label{ycont2} \end{align} where the subscripts $i$ and $i'$ range over the rows and $k$ is the number of columns. To illustrate the methods we will use 3 small data sets defined below. All are unbalanced. The second data set removes the aD observation and so has a zero cell, the third removes the diagonal and has 3 missing cells. <>= data1 <- data.frame(y = rep(1:6, length=20), x1 = factor(letters[rep(1:3, length=20)]), x2 = factor(LETTERS[rep(1:4, length=10)]), x3 = 1:20) data1$x1[19] <- 'c' data1 <- data1[order(data1$x1, data1$x2),] row.names(data1) <- NULL with(data1, table(x1,x2)) # data2 -- single missing cell indx <- with(data1, x1=='a' & x2=='D') data2 <- data1[!indx,] #data3 -- missing the diagonal data3 <- data1[as.numeric(data1$x1) != as.numeric(data1$x2),] @ \subsection{NSTT method} The first calculation method is based on a simple observation. If we impose the standard sums constraint on equation \eqref{a1} which is often found in textbooks (but nowhere else) of $\sum_i \alpha_i = \sum_j \beta_j = 0$, $\sum_i\gamma_{ij} =0 \; \forall j$ and $\sum_j \gamma_{ij} = 0 \; \forall i$, then the last term in equation \eqref{ycont2} is identically 0. Thus the Yates contrast corresponds exactly to a test of $\alpha=0$. In R we can choose this coding by using the \code{contr.sum} option. This approach has the appearance of simplicity: we can do an ordinary test for row effects within an interaction model. Here is R code that is often proposed for ``type III'' computation, which is based on the same process. <<>>= options(contrasts=c("contr.sum", "contr.poly")) fit1 <- lm(y ~ x1*x2, data1) drop1(fit1, .~.) @ The problem with this approach is that it depends critically on use of the sum constraints. If we apply the same code after fitting the data set under the more usual constraints a completely different value ensues. <<>>= options(contrasts=c("contr.SAS", "contr.poly")) fit2 <- lm(y ~ x1*x2, data1) drop1(fit2, .~.) options(contrasts=c("contr.treatment", "contr.poly")) fit3 <- lm(y ~ x1*x2, data1) drop1(fit3, .~.) @ Both common choices of contrasts give a different answer than contr.sum, and both are useless. I thus refer to this as the Not Safe Type Three (NSTT) algorithm, ``not SAS type three'' and ``nonsense type three'' are two other sensible expansions. This approach should NEVER be used in practice. \subsection{ATT} The key idea of the averaging approach (Averaged Type Three) is to directly evaluate equation \eqref{ycont2}. The first step of the computation is shown below <>= X <- model.matrix(fit2) ux <- unique(X) ux indx <- rep(1:3, c(4,4,4)) effects <- t(rowsum(ux, indx)/4) # turn sideways to fit the paper better effects yates <- effects[,-1] - effects[,1] yates @ The data set ux has 12 rows, one for each of the 12 unique x*x2 combinations. Because data1 was sorted, the first 4 rows correspond to x=1, the next 4 to x=2 and the next to x=3 which is useful for illustration but has no impact on the computation. The average of rows 1-4 (column 1 of \code{effects} above) is the estimated average response for subjects with x1=a, assuming a uniform distribution over the 12 cells. Any two differences between the three effects is an equivalent basis for computing the Yates contrast. We can verify that the resulting estimates correspond to a uniform target population by directly examining the case weights for the estimate. Each of them gives a total weight of 1/4 to each level of x2. Each element of $\beta\beta$ is a weighted average of the data, revealed by the rows of the matrix $(X'X)^{-1}X'$. The estimate are a weighted sum of the coefficients, so are also a weighted average of the $y$ values. <<>>= wt <- solve(t(X) %*% X, t(X)) # twelve rows (one per coef), n columns casewt <- t(effects) %*% wt # case weights for the three "row efffects" for (i in 1:3) print(tapply(casewt[i,], data1$x2, sum)) @ \subsection{STT} The SAS type III method takes a different approach, based on a a dependency matrix $D$. Start by writing the $X$ matrix for the problem using all of the parameters in equation \eqref{a1}. For our flc example this will have columns for intercept (1), sex (2), age group (5) and the age group by sex interaction (10) = 18 columns. Now define the lower triangular square matrix $D$ such that \begin{itemize} \item If the $i$th column of $X$ can be written as a linear combination of columns 1 through $i-1$, then row $i$ of $D$ contains that linear combination and $D_{ii}=0$. \item If the $i$th column is not linearly dependent on earlier ones then $D_{ii}=1$ and $D_{ij}=0$ for all $j \ne i$. \end{itemize} Columns of $D$ that correspond to linearly dependent columns of $X$ will be identically zero and can be discarded (or not) at this point. The result of this operation replicates table 12.2 in the SAS reference \cite{SASguide} labeled ``the form of estimable functions''. To obtain the Yates contrasts for an effect replace the appropriate columns of $D$ with the residuals from a regression on all columns to the right of it. Simple inspection shows that the columns of $D$ corresponding to any given effect will already be orthogonal to other effects in $D$ \emph{except} those for interactions that contain it; so the regression does not have to include all columns to the right. It is easy to demonstrate that this gives the uniform population contrast (Yates) for a large number of data sets, but I have not yet constructed a proof. (I suspect it could be approached using the Rao-Blackwell theorem.) \subsection{Bystanders} What about a model that has a extra predictor, such as \code{x3} in our example data and in the fit below? <<>>= fit4 <- lm(y ~ x1*x2 + x3, data=data1) @ The standard approach is to ignore this variable when setting up ``type III'' tests: the contrast for \code{x1} will be the same as it was in the prior model, with a 0 row in the middle for the x3 coefficient. \subsection{Missing cells} When there are combinations of factors with 0 subjects in that group, it is not possible to create a uniform population via reweighting of either subjects or parameters. There is thus no Yates contrast corresponding to the hypothetical population of interest. For that matter, adjustment to any fixed population is no longer possible, such as the US 2000 reference, unless groups are pooled so as to remove any counts of zero, and even then the estimate could be problematic due to extreme weights. This fact does not stop each of the above 3 algorithms from executing and producing a number. This raises two further issues. First, what does that number \emph{mean}? Much ink has been spilled on this subject, but I personally have never been able to come to grips with a satisfactory explanation and so have nothing to offer on the topic. I am reluctant to use such estimates. The second issue is that the computational algorithms become more fragile. \begin{itemize} \item The NSTT algorithm is a disaster in waiting, so no more needs to be said about situations where its behavior may be even worse. \item When fitting the original model, there will be one or more NA coefficients due to the linear dependencies that arise. A natural extension of the ATT method is to leave these out of the sums when computing each average. However, there are data sets for which the particular set of coefficients returned as missing will depend on the order in which variables were listed in the model statement, which in turn will change the ATT result. \item For the STT method, our statement that certain other columns in $D$ will be orthogonal to the chosen effect is no longer true. To match SAS, the orthogonalization step above should include only those effects further to the right that contain the chosen effect (the one we are constructing a contrast vector for). As a side effect, this makes the STT result invariant to the order of the variables in the model statement. \end{itemize} \section{SAS computations} The following code was executed in version 9.3 of SAS. \begin{verbatim} options ls=70; libname save "sasdata"; title "Sex only"; proc glm data=save.flc; class sex; model flc = sex; title "Sex only"; proc glm data=save.flc; class sex age2; model flc = age2 sex /solution E1 E2 E3; title "Second fit, no interaction"; proc glm data=save.flc; class sex age2; model flc = sex age2 sex*age2/solution E1 E2 E3; estimate 'yates' sex 1 -1 sex*age2 .2 .2 .2 .2 .2 -.2 -.2 -.2 -.2 -.2; title "Third fit, interaction"; proc phreg data=save.flc; class sex age2; model futime * death(0) = sex age2/ ties=efron; title "Phreg fit, sex and age, additive"; proc phreg data=save.flc; class sex age2; model futime * death(0) = sex age2 sex*age2 / ties=efron type3(all); estimate 'Yates sex' sex 1 sex*age2 .2 .2 .2 .2; contrast 'NSTT sex ' sex 1 ; contrast 'NSTT age' age2 1 0 0 0 , age2 0 1 0 0 , age2 0 0 1 0 , age2 0 0 0 1; title "Phreg fit, sex and age with interaction"; proc phreg data=save.flc; class sex age2/ param=effect; model futime * death(0) = sex age2 sex*age2 / ties=efron; title "Phreg, using effect coding"; \end{verbatim} The SAS output is voluminous, covering over a dozen pages. A subset is extracted below, leaving out portions that are unimportant to our comparison. First the GLM model for sex only. There are no differences between type 1 and type 3 output for this model. \small \begin{verbatim} ... Number of Observations Read 7874 Number of Observations Used 7874 ... Dependent Variable: flc Sum of Source DF Squares Mean Square F Value Model 1 142.19306 142.19306 42.27 Error 7872 26481.86345 3.36406 Corrected Total 7873 26624.05652 \end{verbatim} \normalsize The second fit with sex and then age. \small \begin{verbatim} Type I Estimable Functions -----------------Coefficients------------------ Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F -0.2571*L2-0.2576*L3-0.1941*L4-0.0844*L5 L7 sex M 0.2571*L2+0.2576*L3+0.1941*L4+0.0844*L5 -L7 Type II Estimable Functions ---Coefficients---- Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F 0 L7 sex M 0 -L7 Type III Estimable Functions ---Coefficients---- Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F 0 L7 sex M 0 -L7 Dependent Variable: flc Sum of Source DF Squares Mean Square F Value Model 5 2212.13649 442.42730 142.60 Error 7868 24411.92003 3.10268 Corrected Total 7873 26624.05652 Source DF Type I SS Mean Square F Value age2 4 1929.642183 482.410546 155.48 sex 1 282.494304 282.494304 91.05 Source DF Type II SS Mean Square F Value age2 4 2069.943424 517.485856 166.79 sex 1 282.494304 282.494304 91.05 Source DF Type III SS Mean Square F Value age2 4 2069.943424 517.485856 166.79 sex 1 282.494304 282.494304 91.05 Standard Parameter Estimate Error t Value Pr > |t| Intercept 5.503757546 B 0.17553667 31.35 <.0001 age2 1 -2.587424744 B 0.17584961 -14.71 <.0001 age2 2 -2.249164537 B 0.17684133 -12.72 <.0001 age2 3 -1.770342603 B 0.17834253 -9.93 <.0001 age2 4 -1.082104827 B 0.18584656 -5.82 <.0001 age2 5 0.000000000 B sex F -0.383454133 B 0.04018624 -9.54 <.0001 sex M 0.000000000 B \end{verbatim} \normalsize The third linear models fit, containing interactions. For first portion I have trimmed off long printout on the right, i.e. the estimable functions for the age2*sex effect since they are not of interest. \small \begin{verbatim} Type I Estimable Functions --------------------Coefficients-------- Effect sex age2 Intercept 0 0 sex F L2 0 sex M -L2 0 age2 1 -0.0499*L2 L4 age2 2 -0.0373*L2 L5 age2 3 0.0269*L2 L6 age2 4 0.0482*L2 L7 age2 5 0.0121*L2 -L4-L5-L6-L7 sex*age2 F 1 0.3786*L2 0.6271*L4+0.1056*L5+0.0796*L6+0.0346*L7 sex*age2 F 2 0.2791*L2 0.0778*L4+0.5992*L5+0.0587*L6+0.0255*L7 sex*age2 F 3 0.2182*L2 0.0527*L4+0.0528*L5+0.6245*L6+0.0173*L7 sex*age2 F 4 0.1055*L2 0.0188*L4+0.0188*L5+0.0142*L6+0.7006*L7 sex*age2 F 5 0.0186*L2 -0.7764*L4-0.7764*L5-0.777*L6-0.7781*L7 sex*age2 M 1 -0.4285*L2 0.3729*L4-0.1056*L5-0.0796*L6-0.0346*L7 sex*age2 M 2 -0.3164*L2 -0.0778*L4+0.4008*L5-0.0587*L6-0.0255*L7 sex*age2 M 3 -0.1913*L2 -0.0527*L4-0.0528*L5+0.3755*L6-0.0173*L7 sex*age2 M 4 -0.0573*L2 -0.0188*L4-0.0188*L5-0.0142*L6+0.2994*L7 sex*age2 M 5 -0.0065*L2 -0.2236*L4-0.2236*L5-0.223*L6-0.2219*L7 Type II Estimable Functions --------------------Coefficients--------------------- Effect sex age2 Intercept 0 0 sex F L2 0 sex M -L2 0 age2 1 0 L4 age2 2 0 L5 age2 3 0 L6 age2 4 0 L7 age2 5 0 -L4-L5-L6-L7 sex*age2 F 1 0.41*L2 0.6271*L4+0.1056*L5+0.0796*L6+0.0346*L7 sex*age2 F 2 0.3025*L2 0.0778*L4+0.5992*L5+0.0587*L6+0.0255*L7 sex*age2 F 3 0.2051*L2 0.0527*L4+0.0528*L5+0.6245*L6+0.0173*L7 sex*age2 F 4 0.073*L2 0.0188*L4+0.0188*L5+0.0142*L6+0.7006*L7 sex*age2 F 5 0.0093*L2 -0.7764*L4-0.7764*L5-0.777*L6-0.7781*L7 sex*age2 M 1 -0.41*L2 0.3729*L4-0.1056*L5-0.0796*L6-0.0346*L7 sex*age2 M 2 -0.3025*L2 -0.0778*L4+0.4008*L5-0.0587*L6-0.0255*L7 sex*age2 M 3 -0.2051*L2 -0.0527*L4-0.0528*L5+0.3755*L6-0.0173*L7 sex*age2 M 4 -0.073*L2 -0.0188*L4-0.0188*L5-0.0142*L6+0.2994*L7 sex*age2 M 5 -0.0093*L2 -0.2236*L4-0.2236*L5-0.223*L6-0.2219*L7 Type III Estimable Functions ---------------------Coefficients--------------------- Effect sex age2 sex*age2 Intercept 0 0 0 sex F L2 0 0 sex M -L2 0 0 age2 1 0 L4 0 age2 2 0 L5 0 age2 3 0 L6 0 age2 4 0 L7 0 age2 5 0 -L4-L5-L6-L7 0 sex*age2 F 1 0.2*L2 0.5*L4 L9 sex*age2 F 2 0.2*L2 0.5*L5 L10 sex*age2 F 3 0.2*L2 0.5*L6 L11 sex*age2 F 4 0.2*L2 0.5*L7 L12 sex*age2 F 5 0.2*L2 -0.5*L4-0.5*L5-0.5*L6-0.5*L7 -L9-L10-L11-L12 sex*age2 M 1 -0.2*L2 0.5*L4 -L9 sex*age2 M 2 -0.2*L2 0.5*L5 -L10 sex*age2 M 3 -0.2*L2 0.5*L6 -L11 sex*age2 M 4 -0.2*L2 0.5*L7 -L12 sex*age2 M 5 -0.2*L2 -0.5*L4-0.5*L5-0.5*L6-0.5*L7 L9+L10+L11+L12 Source DF Type I SS Mean Square F Value sex 1 142.193063 142.193063 45.97 age2 4 2069.943424 517.485856 167.30 sex*age2 4 87.218363 21.804591 7.05 Source DF Type II SS Mean Square F Value sex 1 282.494304 282.494304 91.33 age2 4 2069.943424 517.485856 167.30 sex*age2 4 87.218363 21.804591 7.05 Source DF Type III SS Mean Square F Value sex 1 126.961986 126.961986 41.05 age2 4 1999.446491 499.861623 161.60 sex*age2 4 87.218363 21.804591 7.05 Standard Parameter Estimate Error t Value Pr > |t| yates -0.58972607 0.09204824 -6.41 <.0001 Standard Parameter Estimate Error t Value Pr > |t| Intercept 6.003043478 B 0.36672295 16.37 <.0001 sex F -1.024512614 B 0.41553944 -2.47 0.0137 sex M 0.000000000 B age2 1 -3.176876326 B 0.36950532 -8.60 <.0001 age2 2 -2.787597918 B 0.37048599 -7.52 <.0001 age2 3 -2.088127335 B 0.37292760 -5.60 <.0001 age2 4 -1.353746449 B 0.38703805 -3.50 0.0005 age2 5 0.000000000 B sex*age2 F 1 0.813889663 B 0.42023749 1.94 0.0528 sex*age2 F 2 0.716160958 B 0.42189464 1.70 0.0896 sex*age2 F 3 0.330651265 B 0.42487846 0.78 0.4365 sex*age2 F 4 0.313230835 B 0.44127621 0.71 0.4778 sex*age2 F 5 0.000000000 B sex*age2 M 1 0.000000000 B sex*age2 M 2 0.000000000 B sex*age2 M 3 0.000000000 B sex*age2 M 4 0.000000000 B sex*age2 M 5 0.000000000 B \end{verbatim} \normalsize The phreg printout for the additive model with age and sex. \small \begin{verbatim} Testing Global Null Hypothesis: BETA=0 Test Chi-Square DF Pr > ChiSq Likelihood Ratio 2357.5239 5 <.0001 Score 3823.3905 5 <.0001 Wald 2374.5250 5 <.0001 Type 3 Tests Wald Effect DF Chi-Square Pr > ChiSq sex 1 69.9646 <.0001 age2 4 2374.5211 <.0001 Analysis of Maximum Likelihood Estimates Parameter Standard Parameter DF Estimate Error Chi-Square Pr > ChiSq sex F 1 -0.36617 0.04378 69.9646 <.0001 age2 1 1 -4.18209 0.12180 1179.0289 <.0001 age2 2 1 -3.23859 0.11418 804.5068 <.0001 age2 3 1 -2.17521 0.10963 393.6524 <.0001 age2 4 1 -1.15226 0.11072 108.3077 <.0001 \end{verbatim} \normalsize The model with age*sex interaction. \small \begin{verbatim} Model Fit Statistics Without With Criterion Covariates Covariates -2 LOG L 37736.900 35374.050 AIC 37736.900 35392.050 SBC 37736.900 35443.188 Testing Global Null Hypothesis: BETA=0 Test Chi-Square DF Pr > ChiSq Likelihood Ratio 2362.8497 9 <.0001 Score 3873.5113 9 <.0001 Wald 2357.9498 9 <.0001 Type 3 Tests LR Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4607 0.4973 age2 4 932.1371 <.0001 sex*age2 4 5.3258 0.2555 Score Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4757 0.4904 age2 4 1506.8699 <.0001 sex*age2 4 5.2516 0.2624 Wald Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4833 0.4869 age2 4 964.6007 <.0001 sex*age2 4 5.2322 0.2643 Analysis of Maximum Likelihood Estimates Parameter Standard Parameter DF Estimate Error Chi-Square sex F 1 -0.16537 0.23789 0.4833 age2 1 1 -4.02699 0.22585 317.9171 age2 2 1 -3.04796 0.21843 194.7187 age2 3 1 -1.99577 0.21577 85.5504 age2 4 1 -1.10659 0.22256 24.7216 sex*age2 F 1 1 -0.21121 0.26896 0.6167 sex*age2 F 2 1 -0.29334 0.25518 1.3214 sex*age2 F 3 1 -0.25663 0.24829 1.0684 sex*age2 F 4 1 -0.04339 0.25527 0.0289 Contrast DF Chi-Square Pr > ChiSq NSTT sex 1 0.4833 0.4869 NSTT age 4 964.6007 <.0001 Likelihood Ratio Statistics for Type 1 Analysis LR Source -2 Log L DF Chi-Square Pr > ChiSq (Without Covariates) 37736.8997 sex 37733.0932 1 3.8066 0.0511 age2 35379.3758 4 2353.7173 <.0001 sex*age2 35374.0501 4 5.3258 0.2555 Standard Label Estimate Error z Value Pr > |z| Yates -0.3263 0.06149 -5.31 <.0001 \end{verbatim} \normalsize \begin{thebibliography}{9} \bibitem{Aitkin78} M. Aitkin (1978). The analysis of unbalanced cross classifications (with discussion). \emph{J Royal Stat Soc A} 141:195-223. \bibitem{Dispenzieri12} A. Dispenzieri, J. Katzmann, R. Kyle, D. Larson, T. Therneau, C. Colby, R. Clark, .G Mead, S. Kumar, L..J Melton III and S.V. Rajkumar (2012). Use of monoclonal serum immunoglobulin free light chains to predict overall survival in the general population, \emph{Mayo Clinic Proc} 87:512--523. \bibitem{Herr86} D. G. Herr (1986). On the History of ANOVA in Unbalanced, Factorial Designs: The First 30 Years. \emph{Amer Statistician} 40:265-270. \bibitem{Kyle06} R. Kyle, T. Therneau, S.V. Rajkumar, D. Larson, M. Plevak, J. Offord, A. Dispenzieri, J. Katzmann, and L.J. Melton, III (2006), Prevalence of monoclonal gammopathy of undetermined significance, \emph{New England J Medicine} 354:1362--1369. \bibitem{Macnaughton92} D. B. Macnaughton (1992). Which sum of squares are best in an unbalanced analysis of variance. www.matstat.com/ss. \bibitem{Nelder77} J. Nelder (1977). A reformulation of linear models (with discussion). \emph{J Royal Stat Soc A} 140:48--76. \bibitem{SASguide} SAS Institute Inc. (2008), The four types of estimable functions. SAS/STAT 9.2 User's Guide, chapter 15. \bibitem{Searle71} S. R. Searle, \emph{Linear Models}, Wiley, New York, 1971. \bibitem{Senn1} S. Senn. Multi-centre trials and the finally decisive argument. www.senns.demon.co.uk/wprose.html\#FDA. \bibitem{Senn2} S. Senn. Good mixed centre practice. www.senns.demon.co.uk/wprose.html\#Mixed. \bibitem{Senn07} S. Senn. Statistical Issues in Drug Development, Wiley, New York, 2007. \bibitem{Senn00} S. Senn. The many modes of meta. Drug Information J 34:535-549, 2000. \bibitem{Therneau00} T. M. Therneau and P. M. Grambsch, \emph{Modeling Survival Data: Extending the Cox Model}, Springer-Verlag, New York, 2000. \bibitem{Yates34} F. Yates (1934). The analysis of multiple classifications with unequal numbers in the different classes. \emph{J Am Stat Assoc}, 29:51--66. \end{thebibliography} \end{document} survival/inst/doc/adjcurve.R0000644000175100001440000004176112545056257015637 0ustar hornikusers### R code from vignette source 'adjcurve.Rnw' ################################################### ### code chunk number 1: init ################################################### options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text require(survival, quietly=TRUE) fdata <- flchain[flchain$futime > 7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) ################################################### ### code chunk number 2: adjcurve.Rnw:181-195 ################################################### group3 <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) age1 <- cut(fdata$age, c(49,59,69,79, 110)) levels(age1) <- c(paste(c(50,60,70), c(59,69,79), sep='-'), '80+') temp1 <- table(group3, age1) temp2 <- round(100* temp1/rowSums(temp1)) pfun <- function(x,y) { paste(ifelse(x<1000, "\\phantom{0}", ""), x, " (", ifelse(y<10, "\\phantom{0}", ""), y, ") ", sep="") } cat(paste(c("FLC $<$ 3.38", pfun(temp1[1,], temp2[1,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC 3.38--4.71", pfun(temp1[2,], temp2[2,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC $>$ 4.71", pfun(temp1[3,], temp2[3,])), collapse=" & "), "\n") ################################################### ### code chunk number 3: flc1 ################################################### getOption("SweaveHooks")[["fig"]]() fdata <- flchain[flchain$futime >=7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) fdata$group <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) sfit1 <- survfit(Surv(futime, death) ~ group, fdata) plot(sfit1, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") text(c(11.1, 10.5, 7.5), c(.88, .57, .4), c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71"), col=c(1,2,4)) ################################################### ### code chunk number 4: adjcurve.Rnw:271-276 ################################################### tab1 <- with(fdata, table(group, age2, sex)) cat("Low&", paste(tab1[1,,1], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,1], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,1], collapse=" &"), "\\\\\n") ################################################### ### code chunk number 5: adjcurve.Rnw:281-284 ################################################### cat("Low&", paste(tab1[1,,2], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,2], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,2], collapse=" &"), "\n") ################################################### ### code chunk number 6: flc2 ################################################### getOption("SweaveHooks")[["fig"]]() temp <- with(fdata, table(group, age2, sex)) dd <- dim(temp) # Select subjects set.seed(1978) select <- array(vector('list', length=prod(dd)), dim=dd) for (j in 1:dd[2]) { for (k in 1:dd[3]) { n <- temp[3,j,k] # how many to select for (i in 1:2) { indx <- which(as.numeric(fdata$group)==i & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[i,j,k] <- list(sample(indx, n, replace=(n> temp[i,j,k]))) } indx <- which(as.numeric(fdata$group)==3 & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[3,j,k] <- list(indx) #keep all the group 3 = high } } data2 <- fdata[unlist(select),] sfit2 <- survfit(Surv(futime, death) ~ group, data2) plot(sfit2, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) ################################################### ### code chunk number 7: adjcurve.Rnw:390-396 ################################################### # I can't seem to put this all into an Sexpr z1 <- with(fdata,table(age, sex, group)) z2<- apply(z1, 1:2, min) ztemp <- 3*sum(z2) z1b <- with(fdata, table(age>64, sex, group)) ztemp2 <- sum(apply(z1b, 1:2, min)) ################################################### ### code chunk number 8: adjcurve.Rnw:414-415 ################################################### survdiff(Surv(futime, death) ~ group, data=data2) ################################################### ### code chunk number 9: adjcurve.Rnw:443-449 ################################################### refpop <- uspop2[as.character(50:100),c("female", "male"), "2000"] pi.us <- refpop/sum(refpop) age100 <- factor(ifelse(fdata$age >100, 100, fdata$age), levels=50:100) tab100 <- with(fdata, table(age100, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab100 #new weights by age,sex, group range(us.wt) ################################################### ### code chunk number 10: adjcurve.Rnw:460-469 ################################################### temp <- as.numeric(cut(50:100, c(49, 54, 59, 64, 69, 74, 79, 89, 110)+.5)) pi.us<- tapply(refpop, list(temp[row(refpop)], col(refpop)), sum)/sum(refpop) tab2 <- with(fdata, table(age2, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab2 range(us.wt) index <- with(fdata, cbind(as.numeric(age2), as.numeric(sex), as.numeric(group))) fdata$uswt <- us.wt[index] sfit3a <-survfit(Surv(futime, death) ~ group, data=fdata, weight=uswt) ################################################### ### code chunk number 11: flc3a ################################################### getOption("SweaveHooks")[["fig"]]() tab1 <- with(fdata, table(age2, sex))/ nrow(fdata) matplot(1:8, cbind(pi.us, tab1), pch="fmfm", col=c(2,2,1,1), xlab="Age group", ylab="Fraction of population", xaxt='n') axis(1, 1:8, levels(fdata$age2)) tab2 <- with(fdata, table(age2, sex, group))/nrow(fdata) tab3 <- with(fdata, table(group)) / nrow(fdata) rwt <- rep(tab1,3)/tab2 fdata$rwt <- rwt[index] # add per subject weights to the data set sfit3 <- survfit(Surv(futime, death) ~ group, data=fdata, weight=rwt) temp <- rwt[,1,] #show female data temp <- temp %*% diag(1/apply(temp,2,min)) round(temp, 1) #show female data ################################################### ### code chunk number 12: flc3 ################################################### getOption("SweaveHooks")[["fig"]]() plot(sfit3, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3a, mark.time=F, col=c(1,2,4), lty=1, lwd=1, xscale=365.25) lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) ################################################### ### code chunk number 13: adjcurve.Rnw:553-562 ################################################### id <- 1:nrow(fdata) cfit <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=rwt) summary(cfit)$robscore if (exists("svykm")) { #true if the survey package is loaded sdes <- svydesign(id = ~0, weights=~rwt, data=fdata) dfit <- svykm(Surv(futime, death) ~ group, design=sdes, se=TRUE) } ################################################### ### code chunk number 14: ipw ################################################### options(na.action="na.exclude") gg <- as.numeric(fdata$group) lfit1 <- glm(I(gg==1) ~ factor(age2) * sex, data=fdata, family="binomial") lfit2 <- glm(I(gg==2) ~ factor(age2) * sex, data=fdata, family="binomial") lfit3 <- glm(I(gg==3) ~ factor(age2) * sex, data=fdata, family="binomial") temp <- ifelse(gg==1, predict(lfit1, type='response'), ifelse(gg==2, predict(lfit2, type='response'), predict(lfit3, type='response'))) all.equal(1/temp, fdata$rwt) ################################################### ### code chunk number 15: flc4 ################################################### getOption("SweaveHooks")[["fig"]]() lfit1b <-glm(I(gg==1) ~ age + sex, data=fdata, family="binomial") lfit2b <- glm(I(gg==2) ~ age +sex, data=fdata, family="binomial") lfit3b <- glm(I(gg==3) ~ age + sex, data=fdata, family="binomial") # weights for each group using simple logistic twt <- ifelse(gg==1, 1/predict(lfit1b, type="response"), ifelse(gg==2, 1/predict(lfit2b, type="response"), 1/predict(lfit3b, type="response"))) tdata <- data.frame(fdata, lwt=twt) #grouped plot for the females temp <- tdata[tdata$sex=='F',] temp$gg <- as.numeric(temp$group) c1 <- with(temp[temp$gg==1,], tapply(lwt, age2, sum)) c2 <- with(temp[temp$gg==2,], tapply(lwt, age2, sum)) c3 <- with(temp[temp$gg==3,], tapply(lwt, age2, sum)) xtemp <- outer(1:8, c(-.1, 0, .1), "+") #avoid overplotting ytemp <- 100* cbind(c1/sum(c1), c2/sum(c2), c3/sum(c3)) matplot(xtemp, ytemp, col=c(1,2,4), xlab="Age group", ylab="Weighted frequency (%)", xaxt='n') ztab <- table(fdata$age2) points(1:8, 100*ztab/sum(ztab), pch='+', cex=1.5, lty=2) # Add the unadjusted temp <- tab2[,1,] temp <- scale(temp, center=F, scale=colSums(temp)) matlines(1:8, 100*temp, pch='o', col=c(1,2,4), lty=2) axis(1, 1:8, levels(fdata$age2)) ################################################### ### code chunk number 16: adjcurve.Rnw:694-704 ################################################### # compute new weights wtscale <- table(fdata$group)/ tapply(fdata$rwt, fdata$group, sum) wt2 <- c(fdata$rwt * wtscale[fdata$group]) c("rescaled cv"= sd(wt2)/mean(wt2), "rwt cv"=sd(fdata$rwt)/mean(fdata$rwt)) cfit2a <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight= rwt) cfit2b <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=wt2) round(c(cfit2a$rscore, cfit2b$rscore),1) ################################################### ### code chunk number 17: strata ################################################### allfit <- survfit(Surv(futime/365.25, death) ~ group + age2 + sex, fdata) temp <- summary(allfit)$table temp[1:6, c(1,4)] #abbrev printout to fit page ################################################### ### code chunk number 18: flc5 ################################################### getOption("SweaveHooks")[["fig"]]() xtime <- seq(0, 14, length=57) #four points/year for 14 years smat <- matrix(0, nrow=57, ncol=3) # survival curves serr <- smat #matrix of standard errors pi <- with(fdata, table(age2, sex))/nrow(fdata) #overall dist for (i in 1:3) { temp <- allfit[1:16 + (i-1)*16] #curves for group i for (j in 1:16) { stemp <- summary(temp[j], times=xtime, extend=T) smat[,i] <- smat[,i] + pi[j]*stemp$surv serr[,i] <- serr[,i] + pi[i]*stemp$std.err^2 } } serr <- sqrt(serr) matplot(xtime, smat, type='l', lwd=2, col=c(1,2,4), ylim=c(0,1), lty=1, xlab="Years from sample", ylab="Survival") lines(sfit1, mark.time=F, lty=2, col=c(1,2,4), xscale=365.25) ################################################### ### code chunk number 19: adjcurve.Rnw:829-830 ################################################### survdiff(Surv(futime, death) ~ group + strata(age2, sex), fdata) ################################################### ### code chunk number 20: flc8 ################################################### getOption("SweaveHooks")[["fig"]]() cfit4a <- coxph(Surv(futime, death) ~ age + sex + strata(group), data=fdata) surv4a <- survfit(cfit4a) plot(surv4a, col=c(1,2,4), mark.time=F, xscale=365.25, xlab="Years post sample", ylab="Survival") ################################################### ### code chunk number 21: flc6 ################################################### getOption("SweaveHooks")[["fig"]]() tab4a <- with(fdata, table(age, sex)) uage <- as.numeric(dimnames(tab4a)[[1]]) tdata <- data.frame(age = uage[row(tab4a)], sex = c("F","M")[col(tab4a)], count= c(tab4a)) tdata3 <- tdata[rep(1:nrow(tdata), 3),] #three copies tdata3$group <- factor(rep(1:3, each=nrow(tdata)), labels=levels(fdata$group)) sfit4a <- survexp(~group, data=tdata3, weight = count, ratetable=cfit4a) plot(sfit4a, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, c("FLC low", "FLC med", "FLC high"), lty=1, col=c(1,2,4), bty='n', lwd=2) ################################################### ### code chunk number 22: adjcurve.Rnw:941-948 ################################################### tfit <- survfit(cfit4a, newdata=tdata, se.fit=FALSE) curves <- vector('list', 3) twt <- c(tab4a)/sum(tab4a) for (i in 1:3) { temp <- tfit[i,] curves[[i]] <- list(time=temp$time, surv= c(temp$surv %*% twt)) } ################################################### ### code chunk number 23: flc6b ################################################### getOption("SweaveHooks")[["fig"]]() par(mfrow=c(1,2)) cfit4b <- coxph(Surv(futime, death) ~ age*sex + strata(group), fdata) sfit4b <- survexp(~group, data=tdata3, ratetable=cfit4b, weights=count) plot(sfit4b, fun='event', xscale=365.25, xlab="Years from sample", ylab="Deaths") lines(sfit3, mark.time=FALSE, fun='event', xscale=365.25, lty=2) lines(sfit4a, fun='event', xscale=365.25, col=2) temp <- median(fdata$sample.yr) mrate <- survexp.mn[as.character(uage),, as.character(temp)] crate <- predict(cfit4b, newdata=tdata, reference='sample', type='lp') crate <- matrix(crate, ncol=2)[,2:1] # mrate has males then females, match it # crate contains estimated log(hazards) relative to a baseline, # and mrate absolute hazards, make both relative to a 70 year old for (i in 1:2) { mrate[,i] <- log(mrate[,i]/ mrate[21,2]) crate[,i] <- crate[,i] - crate[21,2] } matplot(mrate, crate, col=2:1, type='l') abline(0, 1, lty=2, col=4) ################################################### ### code chunk number 24: adjcurve.Rnw:1019-1027 ################################################### getOption("SweaveHooks")[["fig"]]() obs <- with(fdata, tapply(death, list(age2, sex, group), sum)) pred<- with(fdata, tapply(predict(cfit4b, type='expected'), list(age2, sex, group), sum)) excess <- matrix(obs/pred, nrow=8) #collapse 3 way array to 2 dimnames(excess) <- list(dimnames(obs)[[1]], c("low F", "low M", "med F", "med M", "high F", "high M")) round(excess, 1) ################################################### ### code chunk number 25: adjcurve.Rnw:1043-1055 ################################################### cfit5a <- coxph(Surv(futime, death) ~ group:age +sex + strata(group), fdata) cfit5b <- coxph(Surv(futime, death) ~ group:(age +sex) + strata(group), fdata) cfit5c <- coxph(Surv(futime, death) ~ group:(age *sex) + strata(group), fdata) options(show.signif.stars=FALSE) # see footnote anova(cfit4a, cfit5a, cfit5b, cfit5c) temp <- coef(cfit5a) names(temp) <- c("sex", "ageL", "ageM", "ageH") round(temp,3) ################################################### ### code chunk number 26: flc7 ################################################### getOption("SweaveHooks")[["fig"]]() pred5a <- with(fdata, tapply(predict(cfit5a, type='expected'), list(age2, sex, group), sum)) excess5a <- matrix(obs/pred5a, nrow=8, dimnames=dimnames(excess)) round(excess5a, 1) sfit5 <- survexp(~group, data=tdata3, ratetable=cfit5a, weights=count) plot(sfit3, fun='event', xscale=365.25, mark.time=FALSE, lty=2, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, col=c(1,2,4)) ################################################### ### code chunk number 27: flc8 ################################################### getOption("SweaveHooks")[["fig"]]() # there is a spurious warning from the model below: R creates 3 unneeded # columns in the X matrix cfit6 <- coxph(Surv(futime, death) ~ group:age2 + sex + strata(group), fdata) saspop <- with(fdata, expand.grid(age2= levels(age2), sex= levels(sex), group = levels(group))) sfit6 <- survexp(~group, data=saspop, ratetable=cfit6) plot(sfit6, fun='event', xscale=365.25, mark.time=FALSE, lty=1, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, lty=2, col=c(1,2,4)) survival/inst/doc/timedep.Rnw0000644000175100001440000010571612545056257016031 0ustar hornikusers\documentclass{article} \usepackage{amsmath} \usepackage{Sweave} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \newcommand{\code}[1]{\texttt{#1}} %\VignetteIndexEntry{Using Time Dependent Covariates} \title{Using Time Dependent Covariates and Time Dependent Coefficients in the Cox Model} \author{Terry Therneau \and Cindy Crowson\\ Mayo Clinic} \begin{document} \maketitle \SweaveOpts{prefix.string=compete,width=6,height=4} \setkeys{Gin}{width=\textwidth} \SweaveOpts{keep.source=TRUE} <>= options(width=60, continue=" ") makefig <- function(file, top=1, right=1, left=4) { pdf(file, width=9.5, height=7, pointsize=18) par(mar=c(4, left, top, right) +.1) } library(survival) @ \section{Introduction} This vignette covers 3 different but interrelated concepts: \begin{itemize} \item An introduction to time dependent covariates, along with some of the most common mistakes. \item Tools for creating time-dependent covariates, or rather the data sets used to encode them. \item Time dependent coefficients. \end{itemize} \section{Time dependent covariates} One of the strengths of the Cox model is its ability to encompass covariates that change over time. The practical reason that time-dependent covariates work is based on the underlying way in which the Cox model works: at each event time the program compares the current covariate values of the subject who had the event to the current values of all others who were at risk at that time. One can think of it as a lottery model, where at each death time there is a drawing to decide which subject ``wins'' the event. Each subject's risk score $\exp(X\beta)$ determines how likely they are to win, e.g., how many ``tickets'' they have purchased for the drawing. The model tries to assign a risk score to each subject that best predicts the outcome of each drawing based on \begin{itemize} \item The risk set: which subjects are present for each event; the set of those able to ``win the prize''. \item The covariate values of each subject just prior to the event time. \end{itemize} The model has a theoretical foundation in martingale theory, a mathematical construct which arose out of the study of games of chance. A key underlying condition for a martingale like game is that present actions depend only on the past. The decision of whether to play (is one in the risk set or not) and the size of a bet (covariates) can depend in any way on prior bets and patterns of won/lost, but cannot look into the future. If this holds then multiple properties can be proven about the resulting process. The key rule for time dependent covariates in a Cox model is simple and essentially the same as that for gambling: \emph{you cannot look the future}. A covariate may change in any way based on past data or outcomes, but it may not reach forward in time. A good example of this is found in a recent analysis from the Mayo Clinic study of aging (MCSA), a study which enrolled a stratified random sample from the population of Olmsted County and then has followed them forward in time. The occurence of mild cognitive impairment (MCI), dementia, and death are all of interest. The paper starts out with a table comparing baseline covariates for those who never progress to MCI versus those who ever did, there is also a table of baseline covariates versus survival. Both of these are fine: if you think in terms of an R formula they could be written as \code{future ~ past}. A model that predicts survival as a function of ever versus never MCI is not correct, however; that is a model with a future occurence on both sides of the equation. One of the more well known examples of this error is analysis by treatment response: at the end of a trial a survival curve is made comparing those who had an early response to treatment (shrinkage of tumor, lowering of cholesterol, or whatever) to those who did not, and it discovered that responders have a better curve. A Cox model fit to the same data will demonstrate a strong ``significant'' effect. The problem arises because any early deaths, those that occur before response can be assessed, will all be assigned to the non-responder group, even deaths that have nothing to do with the condition under study. Below is a simple example based on the advanced lung cancer data set. Assume that subjects came in every monthly for 12 cycles of treatment, and randomly declare a ``response'' for 5\% of the subjects at each visit. <>= set.seed(1953) # a good year nvisit <- floor(pmin(lung$time/30.5, 12)) response <- rbinom(nrow(lung), nvisit, .05) > 0 badfit <- survfit(Surv(time/365.25, status) ~ response, data=lung) plot(badfit, mark.time=FALSE, lty=1:2, xlab="Years post diagnosis", ylab="Survival") legend(1.5, .85, c("Responders", "Non-responders"), lty=2:1, bty='n') @ What is most surprising about this error is the \emph{size} of the false effect that is produced. A Cox model using the above data reports a hazard ratio of 1.9 fold with a p-value of less than 1 in 1000. The alarm about this incorrect approach has been sounded often \cite{Anderson83, Buyse96, Suissa08} but the analysis is routinely re-discovered. A slightly subtler form of the error is discussed in Redmond et al \cite{Redmond83}. Breast cancer chemotherapy patients were divided into three groups based on whether the patient eventually received $>85$\%, 65--85\% or $<65$\% of the dose planned at the start of their treatment. The chemotherapy regiment spans 12 weeks of treatment and the early deaths, not surprisingly, do not get all their dose. If response is instead coded as a time-dependent covariate whose values depend only on the past, then the problem disappears. For treatment response this will be a variable that starts at 0 for all subjects and is recoded to 1 only when the response occurs. For dose it would measure cumulative dose to date. There are many variations on the error: interpolation of the values of a laboratory test linearly between observation times, removing subjects who do not finish the treatment plan, imputing the date of an adverse event as midway between observation times, etc. Using future data will often generate large positive or negative bias in the coefficients, but sometimes it generates little bias at all. It is nearly impossible to predict a priori which of these will occur in any given data set. Using such a covariate is similar to jogging across the Los Angeles freeway: disaster is not guaranteed --- but it is likely. The most common way to encode time-dependent covariates is to use the (start, stop] form of the model. <>= fit <- coxph(Surv(time1, time2, status) ~ age + creatinine, data=mydata) @ In data set \code{mydata} a patient might have the following observations \begin{center} \begin{tabular}{ccccccc} subject & time1 & time2 & status & age & creatinine & \ldots \\ \hline 1 & 0 & 15 & 0 & 25 & 1.3 \\ 1 & 15& 46 & 0 & 25 & 1.5 \\ 1 & 46& 73 & 0 & 25 & 1.4 \\ 1 & 73& 100& 1 & 25 & 1.6 \\ \end{tabular} \end{center} In this case the variable \code{age} = age at entry to the study stays the same from line to line, while the value of creatinine varies and is treated as 1.3 over the interval $(0, 15]$, 1.5 over $(15, 46]$, etc. The intervals are open on the left and closed on the right, which means that the creatinine is taken to be 1.3 on day 15. The status variable describes whether or not each interval ends in an event. One common question with this data setup is whether we need to worry about correlated data, since a given subject has multiple observations. The answer is no, we do not. The reason is that this representation is simply a programming trick. The likelihood equations at any time point use only one copy of any subject, the program picks out the correct row of data at each time. There two exceptions to this rule: \begin{itemize} \item When subjects have multiple events, then the rows for events are correlated and a cluster variance is needed. \item When a subject appears in overlapping intervals. This however is almost always a data error, since it corresponds to two copies of the subject being present in the same strata at the same time, e.g., they could meet themselves on the sidewalk. \end{itemize} A subject can be at risk in multiple strata at the same time, however. This corresponds to being simultaneously at risk for two distinct outcomes. \section{Building time-dependent sets with tmerge} \subsection{The function} A useful function for building data sets is \code{tmerge}, which is part of the survival library. The motivating case for \code{tmerge} came from a particular problem: the Rochester Epidemiology Project has tracked all subjects living in Olmsted County, Minnesota, from 1965 to the present. For an investigation of cumulative comorbidity we had three data sets \begin{itemize} \item base: demographic data such as sex and birth date \item timeline: one or more rows for each subject containing age intervals during which they were a resident of the county. The important variables are id, age1 and age2; each (age1, age2) pair marks an interval of residence. \item outcome: one row for each age/outcome pair of interest. The outcomes were 20 comorbid conditions as defined by NIH. \end{itemize} The structure for building the data is shown below. (The data for this example unfortunately cannot be included in the survival library so the code is shown but not executed.) <>= newd <- tmerge(data1=base, data2=timeline, id=repid, tstart=age1, tstop=age2, options(id="repid")) newd <- tmerge(newd, outcome, id=repid, mtype = cumevent(age)) newd <- with(subset(outcome, event='diabetes'), tmerge(newd, id=repid, diabetes= tdc(age))) newd <- with(subset(outcome, event='arthritis'), tmerge(newd, id=repid, event =tdc(age))) @ The first call to tmerge adds the timeline for each observation to the baseline data. The \code{tstart} and \code{tstop} arguments refer to the starting and ending times for each subject and are taken from data set 2 (\code{data2}). The \code{options} argument tells the routine that the identifier variable in data set 1 is called `repid', and will cause the identifier variable in the final output data set \code{newd} to have that name. By default, the names of the three key variables are ``id'', ``tstart'', and ``tstop''; these uniquely identify each row of the final data set. Each subsequent call adds a new variable to the data set. The second line creates an event variable which is a cumulative count of the number of comorbidities thus far, for each subject. The third line creates a time dependent covariate (tdc) which will be 0 until the age of diabetes and is 1 thereafter, the fourth line creates a time dependent variable for the presence of arthritis. This is the basic working approach for \code{tmerge}: first establish baseline covariates and the range over which the subject is at risk, and then add events and/or time dependent covariates to the data set one by one. These additions will often increase the number of rows in the data set. Say at some stage subject Smith has time intervals of (0,15), (15,40) and (40,100) and we add a time dependent covariate \code{sbp} (systolic blood pressure) which is evaluated at months 6, 12, and 24 with values of 134, 126, and 140, respectively. In the resulting data set Smith will have intervals of \begin{center} \begin{tabular}{rrrrrr} (0,6) &(6,12) & (12,15) & (15,24) & (24,40) & (40,100)\\ & 134 & 126 & 126 & 140 & 140 \end{tabular} \end{center} The value over the interval (0,6) will be the value of the variable \code{sbp} in data set 1 --- if the variable existed there it is assumed to contain the baseline value --- or NA otherwise. \subsubsection{CGD data set} Chronic granulomatous disease (CGD) is a heterogeneous group of uncommon inherited disorders characterized by recurrent pyogenic infections that usually begin early in life and may lead to death in childhood. In 1986, Genentech, Inc. conducted a randomized, double-blind, placebo-controlled trial in 128 CGD patients who received Genentech's humanized interferon gamma (rIFN-g) or placebo three %' times daily for a year. Data were collected on all serious infections until the end of followup, which occurred before day 400 for most patients. One patient was taken off on the day of his last infection; all others have some followup after their last episode. Below are the first 10 observations, see the help page for \texttt{cgd0} for the full list of variable names. The last few columns contain the duration of follow-up for the subject followed by infection times. Subject 1 was followed for 414 days and had 2 infections on days 219 and 373, subject 2 had 7 infections, and subject 3 had none. \small \begin{verbatim} 1 204 082888 1 2 12 147.0 62.0 2 2 2 2 414 219 373 2 204 082888 0 1 15 159.0 47.5 2 2 1 2 439 8 26 152 241 249 322 350 3 204 082988 1 1 19 171.0 72.7 1 2 1 2 382 4 204 091388 1 1 12 142.0 34.0 1 2 1 2 388 5 238 092888 0 1 17 162.5 52.7 1 2 1 1 383 246 253 6 245 093088 1 2 44 153.3 45.0 2 2 2 2 364 7 245 093088 0 1 22 175.0 59.7 1 2 1 2 364 292 8 245 093088 1 1 7 111.0 17.4 1 2 1 2 363 9 238 100488 0 1 27 176.0 82.8 2 2 1 1 349 294 10 238 100488 1 1 5 113.0 19.5 1 2 1 1 371 \end{verbatim} \normalsize The data set above is included as \code{cgd0} in the survival library. We want to turn this into a data set that has survival in a counting process form. \begin{itemize} \item Each row of the resulting data set represents a time interval (time1, time2] which is open on the left and closed on the right. Covariate values for that row are the covariate values that apply over that interval. \item The event variable for each row $i$ is 1 if that time interval ends with an event and 0 otherwise. \end{itemize} We don't want the variables etime1--etime7 in the final data set, so they are left out of the data1 argument in the first call. <>= newcgd <- tmerge(cgd0[, 1:13], cgd0, id=id, tstop=futime) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime1)) newcgd <- with(cgd0, tmerge(newcgd, id=id, infect = event(etime2))) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime3)) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime4), infect= event(etime5), infect=event(etime6), infect= event(etime7)) attr(newcgd, "tcount") newcgd <- tmerge(newcgd, newcgd, id, enum=cumtdc(tstart)) all.equal(newcgd[, c("id", "tstart", "tstop", "infect")], cgd [, c("id", "tstart", "tstop", "status")], check.attributes=FALSE) @ \begin{itemize} \item A standard way to build data sets is one addition at a time, as shown by the \code{event(etime1)} and \code{event(etime2)} lines above. When multiple additions are done using the same d the same data set, however, they can also be done using as single call as shown by the line that adds etime4--etime7. When there are multiple arguments they are processed sequentially. \item Additions with a missing time value are skipped. \item The result of \code{tmerge} is a data frame with a few extra attributes. One of these, tcount, is designed to help visualize the process, and was printed out after the last step above. (Printing after every step may often be useful.) Assume that a subject already had 3 intervals of (2,5), (5,10) and (14,40). A new event added at time 1 would be ``early'' while one at time 50 is after any interval and would be recorded as ``late''. An event at time 3 is within an interval, one at 5 is on the border of two intervals, one at 14 is at the leading edge of an interval and one at time 10 in on the trailing edge. In this data set all new additions fell strictly within prior intervals. We also see that etime6 and etime7 each added only a single event to the data. \item If two observations in data2 for a single person share exactly the same time, the created value will be the sum of the contributions. The ``tied'' column tells how often this happened; in some data sets this behavior might not be desired and one would need to break the ties before calling tmerge. \item Sometimes the ``where'' form of the call may be more convenient than using the data2 argument. An example is shown in the addition of etime2. \item The last addition above, after printing the tcount attribute, adds a simple time-dependent variable \code{enum} which is a running observation count for each subject. This can often be a useful variable in later models or processing, e.g. \code{enum==1} select off the first row for each subject. \item The extra attributes of the data frame are ephemeral: they will be lost as soon as any further manipulation is done. This is intentional. \end{itemize} The last line above shows that the created data set is identical to \code{cgd}, a (start, stop] version of the CGD data, also part of the survival library, which had been created by hand several years earlier. \subsection{Stanford heart transplant} The \code{jasa} data set contains information from the Stanford heart transplant study, in the form that it appeared in the paper of Crowley and Hu \cite{Crowley77}. Each row also contain also contains the age, time to transplant and time to last follow-up calculated from these dates. Patients were on medical treatment from their entry to the study until a matching heart became available, at which time they transferred to surgical treatment. As is often the case with real data, this data set contains a few anomalies that need to be dealt with when setting up an analysis data set. \begin{enumerate} \item The coefficients in table 6.1 of the definitive analysis found in Kalbfliesch and Prentice \cite{Kalbfleisch02} will only be obtained if covariates are defined in precisely the same way. For age this is (age in days)/ 365.25 - 40 years, and for year of enrollment it is the number of years since the start of the study: (entry date - 1967/10/1)/365.25. \item One subject died on the day of entry. However (0,0) is an illegal time interval for the program. It suffices to have them die on day 0.5. \item A subject transplanted on day 10 is considered to have been on medical treatment for days 1--10 and as transplanted starting on day 11. That is, except for patient 38 who died during the procedure on day 5. They should be treated as a transplant death; the problem is resolved by moving the transplant day to 4.5. \end{enumerate} Since time is in days the fractional time of 0.5 could be any chosen value $t$ with $0 < t < 1$, it will not affect the results. <>= tdata <- jasa[, -(1:4)] #leave off the dates, temporary data set tdata$futime <- pmax(.5, tdata$futime) # the death on day 0 indx <- with(tdata, which(wait.time == futime)) tdata$wait.time[indx] <- tdata$wait.time[indx] - .5 #the tied transplant sdata <- tmerge(tdata, tdata, id=1:nrow(tdata), death = event(futime, fustat), trans = tdc(wait.time)) attr(sdata, "tcount") coxph(Surv(tstart, tstop, death) ~ age + trans, sdata) @ This example shows one special case for the \code{tmerge} function that is moderately common: when the data1 and data2 arguments are the same, and the first created variable is an event code, then the range for each subject is inferred to be from 0 to the event time: an explicit \code{tstop} argument is not required. It also makes use of a two argument form of \code{event}. Each of the \code{event}, \code{cumevent}, \code{tdc} and \code{cumtdc} functions may have a second argument, which will be used as the value or increment to the event code or time dependent covariate. If not present a value of 1 is used. Also note that if the variable being created is already a part of data1, then our updates make changes to that variable. Be careful of this. This feature is what allowed for the \code{infection} indicator to be build up incrementally in the cgd example given earlier, but quite surprising results can occur when you think a new variable is being created de novo but its name already is in use. (As an example change ``trans'' to ``transplant'' in the code just above). For a variable that is not in data1, the starting point is either a vector of NA values or a vector of zeros; the first is used for a \code{tdc} (or \code{cumtdc}) call that has two arguments, and the zero vector for all other cases. The \code{tcount} table for the above fit shows all the deaths at the trailing edge of their interval, not surprising since last follow-up was used to define the interval of risk. Three of the transplants happened on day 0, one of which we moved to 0.5, the other 2 listed as occurring on the leading edge of the follow-up interval. The other 67 transplants were strictly within the (0, last follow up) interval of each subject. As a further example of time dependent covariates consider the PBC data. The \code{pbc} data set contains baseline data and follow-up status for a set of subjects with primary biliary cirrhosis, while the \code{pbcseq} data set contains repeated laboratory values. The first data set contains data on 312 subjects in a clinical trial plus 106 that agreed to be followed off protocol, the second data set has data only on the protocol subjects. <>= temp <- subset(pbc, id <= 312, select=c(id:sex, stage)) pbc2 <- tmerge(temp, temp, id=id, status = event(time, status)) pbc2 <- tmerge(pbc2, pbcseq, id=id, ascites = tdc(day, ascites), bili = tdc(day, bili), albumin = tdc(day, albumin), protime = tdc(day, protime), alkphos = tdc(day, alk.phos)) coef(coxph(Surv(time, status==2) ~ log(bili) + log(protime), pbc)) coef(coxph(Surv(tstart, tstop, status==2) ~ log(bili) + log(protime), pbc2)) @ The coefficients of bilirubin and prothrombin time are somewhat larger in the time-dependent analysis than using only baseline values. In this autoimmune disease there is steady progression of liver damage, along with a steady rise in these two markers of dysfunction. The baseline analysis captures patients' disease status at the start, the time-dependent also is able to account for those who progress more quickly. <<>>= attr(pbc2, "tcount") @ The tcount results are interesting. For the first addition of ascites we have 312 observations on a leading edge of follow up (all of the lab values at time 0) and 1495 within the subjects' follow-up interval. This causes a new break point to be added at each laboratory date, and for subsequent additions these 1495 are all on a boundary of two intervals. Another 138 lab values occurred after the last follow-up date of the pbc data set and are ignored. The data for the pbcseq data set was gathered at a later calendar time. Since having lab tests done is a certain marker of survival, would a better analysis have first used this information to extend the last follow-up date for these 138 subjects? Not necessarily. Odd things happen in survival analysis when risk sets are extended piecemeal. A basic tenet of the Cox model (or a survival curve) is that if someone is marked as being ``at risk'' over some interval $(s, t)$, this means that ``if they had had an event over that interval, we would have recorded it.'' Say someone ended their initial follow-up time at 3000 days and then had a lab test at 3350 days (subjects returned about once a year). If we only extend the time of those who had a test, then saying that this subject was at risk during the interval (3000, 3350) is false: if they had died in that interval, they would not have had the lab test and not obtained the extension. We need to extend the followup time of all subjects. In the study all subjects were actively followed up and the results of that endeavor are in the futime and status variables of the pbcseq data set. Extension needs to use those variables and not just the presence of another laboratory result. \section{Time dependent coefficients} Time dependent covariates and time dependent coefficients are two different extensions of a Cox model, as shown in the two equations below. \begin{align} \lambda(t) &= \lambda_0(t) e^{\beta X(t)} \label{tdcovar} \\ \lambda(t) &= \lambda_0(t) e^{\beta(t) X} \label{tdbeta} \end{align} Equation \eqref{tdcovar} is a time dependent covariate, a commonly used and well understood usage. Equation \eqref{tdbeta} has a time dependent coefficient. These models are much less common, but represent one way to deal with non-proportional hazards -- the proportional hazard assumption is precisely that the coefficient does not change over time: $\beta(t) = c$. The \code{cox.zph} function will plot an estimate of $\beta(t)$ for a study and is used to diagnose and understand non-proportional hazards. Here for example is a test case using the veterans cancer data. <>= options(show.signif.stars = FALSE) # display intelligence vfit <- coxph(Surv(time, status) ~ trt + prior + karno, veteran) vfit quantile(veteran$karno) zp <- cox.zph(vfit, transform= function(time) log(time +20)) zp plot(zp[3]) abline(0,0, col=2) @ Karnofsky score is a very important predictor, but its effect is not constant over time as shown by both the test and the plot. Early on it has a large negative effect: the risk of someone at the first quartile is approximately exp(35*.03377) = 3.2 fold times that of someone at the third quartile, but by 200 days this has waned. The cox.zph function does not produce a formal fit, however, only a graph and a linear test of whether the graph is ``flat''. What if we want to actually fit the model? If $\beta(t)$ is assumed to have a simple functional form we can fool an ordinary Cox model program in to doing the fit. The particular form $\beta(t) = a + b\log(t)$ has for instance often been assumed. Then $\beta(t) x = ax + b \log(t) x = ax + b z$ for the special time dependent covariate $z = \log(t) x$. The time scale for the plot above of $\log(t + 20)$ was chosen to make the first part of the plot roughly linear. The simple linear model does not fit over the entire range, but we will forge ahead. (After all, most who fit the log(t) form have not bothered to even look at a plot.) An obvious but incorrect approach is <>= vfit2 <- coxph(Surv(time, status) ~ trt + prior + karno + I(karno * log(time + 20)), data=veteran) @ This mistake has been made often enough the the \code{coxph} routine has been updated to print an error message for such attempts. The issue is that the above code does not actually create a time dependent covariate, rather it creates a time-static value for each subject based on their value for the covariate \code{time}; no differently than if we had constructed the variable outside of a \code{coxph} call. This variable most definitely breaks the rule about not looking into the future, and one would quickly find the circularity: large values of \code{time} predict long survival, because long survival leads to large values for \code{time}; the resulting model coefficient is large. A true time-dependent covariate can be constructed using the \emph{time-transform} functionality of coxph. <>= vfit3 <- coxph(Surv(time, status) ~ trt + prior + karno + tt(karno), data=veteran, tt = function(x, t, ...) x * log(t+20)) vfit3 @ The time dependent coefficient is estimated to be $\beta(t) =$ \Sexpr{round(coef(vfit3)[3], 3)} + \Sexpr{round(coef(vfit3)[4], 3)} * log(t + 20). We can add said line to the above plot via \code{abline(coef(vfit3)[3:4])}. Not surprisingly, the result is rather too high for time $>$ 200. (The same dichotomy exists in SAS phreg, by the way. A new covariate based on \code{time} will be fixed, while a programming statement within phreg that uses \code{time} as a variable will generate time-dependent objects. The error is less likely there because phreg's model statement has no equivalent to the \code{I()} function, i.e., you cannot create new variables on-the-fly.) \section{Predictable time-dependent covariates} Occasionally one has a time-dependent covariate whose values in the future are predictable. The most obvious of these is patient age, occasionally this may also be true for the cumulative dose of a drug. If age is entered as a linear term in the model, then the effect of changing age can be ignored in a Cox model, due to the structure of the partial likelihood. Assume that subject $i$ has an event at time $t_i$, with other subject $j \in R_i$ at risk at that time, with $a$ denoting age. The partial likelihood term is \begin{equation*} \frac{e^{\beta * a_i}}{\sum_{j \in R_i} e^{\beta* a_j}} = \frac{e^{\beta * (a_i + t_i)}}{\sum_{j \in R_i} e^{\beta* (a_j + t_i)}} \end{equation*} We see that using time-dependent age (the right hand version) or age at baseline (left hand), the partial likelihood term is identical since $\exp(\beta t_i)$ cancels out of the fraction. However, if the effect of age on risk is \emph{non-linear}, this cancellation does not occur. Since age changes continuously, we would in theory need a very large data set to completely capture the effect, an interval per day to match the usual resolution for death times. In practice this level of resolution is not necessary; though we all grow older, risk does not increase so rapidly that we need to know our age to the day! One method to create a time-changing covariate is to use the \emph{time-transform} feature of coxph. Below is an example using the pbc data set. The longest follow-up time in that data set is over 13 years, follow-up time is in days, and we might worry that the intermediate data set would be huge. The program only needs the value of the time dependent covariate(s) for each subject at the times of events, however, so the maximum number of rows in the intermediate data set is the number of subjects times the number of unique event times. <>= pfit1 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + age, pbc) pfit2 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + tt(age), data=pbc, tt=function(x, t, ...) { age <- x + t/365.25 cbind(age=age, age2= (age-50)^2, age3= (age-50)^3) }) pfit2 anova(pfit2) # anova(pfit1, pfit2) #this fails 2*(pfit2$loglik - pfit1$loglik)[2] @ Since initial age is in years and time is in days, it was important to scale within the pspline function. The likelihood ratio of 10.8 on 2 degrees of freedom shows that the additional terms are mildly significant. When there are one or more terms on the right hand side of the equation marked with the tt() operator, the program will pre-compute the values of that variable for each unique event time. A user-defined function is called with arguments of \begin{itemize} \item the covariate: whatever is inside the tt() call \item the event time \item the event number: if there are multiple strata and the same event time occurs in two of them, they can be treated separately \item the weight for the observation, if the call used weights \end{itemize} There is a single call to the function with a large $x$ vector, it contains an element for each subject at risk at each event time. If there are multiple tt() terms in the formula, then the tt argument should be a list of functions with the requisite number of elements. There are other interesting uses for the time-transform capability. One example is O'Brien's logit-rank test procedure \cite{obrien78}. He proposed replacing the covariate at each event time with a logit transform of its ranks. This removes the influence of any outliers in the predictor $x$. For this case we ignore the event time argument and concentrate on the groupings. <<>>= function(x, t, riskset, weights){ obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } @ This relies on the fact that the input arguments to tt() are ordered by the event number or riskset. This function is used as a default if no tt argument is present in the coxph call, but there are tt terms in the model formula. (Doing so allowed me to depreciate the survobrien function). Another interesting usage is to replace the data by simple ranks, not rescaled to 0--1. <<>>= function(x, t, riskset, weights) unlist(tapply(x, riskset, rank)) @ The score statistic for this model is $(C-D)/2$, where $C$ and $D$ are the number of concordant and discordant pairs, see the survConcordance function. The score statistic from this fit is then a test for significance of the concordance statistics, and is in fact the basis for the standard error reported by survConcordance. The O'Brien test can be viewed as concordance statistic that gives equal %' weight to each event time, whereas the standard concordance weights each event proportionally to the size of the risk set. (The Cox score statistic depends on the mean $x$ at each event time; since ranks go from 1 to number at risk the mean also scales.) Although handy, the computational impact of the tt argument should be considered before using it. The Cox model requires computation of a weighted mean and variance of the covariates at each event time, a process that is inherently $O(ndp^2)$ where $n$ = the sample size, $d$ = the number of events and $p$= the number of covariates. Much of the algorithmic effort in coxph() is to use updating methods for the mean and variance matrices, reducing the compute time to $O((n+d) p^2)$. When a tt term appears updating is not possible; for even moderate size data sets the impact of $nd$ versus $n+d$ can be surprising. The time-transform is a new addition and still has some rough edges. At this moment the $x=TRUE$ argument is needed to get proper residuals and predicted values, and termplot is unable to properly reconstruct the data to plot a fit. Please communicate any concerns or interesting examples to the author. \begin{thebibliography}{9} \bibitem{Anderson83} Anderson JR, Cain KC, and Gelber RD. Analysis of survival by tumor response. J Clinical Oncology 1:710--719, 1983. \bibitem{Buyse96} M Buyse and P Piedbois. The relationship between response to treatment and survival time. Stat in Med 15:2797--2812, 1996. \bibitem{Crowley77} J Crowley and M Hu. Covariance analysis of heart transplant survival data. J American Statistical Assoc, 72:27--36, 1977. \bibitem{Kalbfleisch02} J Kalbfleisch and R Prentice. The statistical analysis of failure time data, second edition. Wiley, 2002. \bibitem{obrien78} O'Brien, Peter. A non-parametric test for association with censored data, Biometrics 34:243--250, 1978. \bibitem{Redmond83} Redmond C, Fisher B, Wieand HS. The methodologic dilemma in retrospectively correlating the amount of chemotherapy received in adjuvant therapy protocols with disease free survival: a commentary. Cancer Treatment Reports 67:519--526, 1983. \bibitem{Suissa08} S Suissa. Immortal time bias in pharmacoepidemiology. Am J Epi, 167:492-499, 2008. \end{thebibliography} \end{document} survival/inst/doc/splines.Rnw0000644000175100001440000002114712545056257016052 0ustar hornikusers\documentclass{article} \usepackage{amsmath} \usepackage{Sweave} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \SweaveOpts{keep.source=TRUE, fig=FALSE} %\VignetteIndexEntry{Splines, plots, and interactions} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{prefix.string=splines,width=6,height=4} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default @ \title{Spline terms in a Cox model} \author{Terry Therneau} \begin{document} \maketitle This is a pair of topics that comes up just often enough in my work that I end up re-discovering how to do it correctly about once a year. A note showing how may be useful to others, it is certainly a useful reference for me. \section{Plotting smooth terms} Here is a simple example using the MGUS data. (And turning off the garish color defaults of termplot). <>= require(survival) mfit <- coxph(Surv(futime, death) ~ sex + pspline(age), data=mgus) termplot(mfit, term=2, se=TRUE, col.term=1, col.se=1) @ Two questions of the plot are ``how was it centered'' and whether we can easily plot it on the hazard as opposed to the log hazard scale. The solution to both is to use the plot=FALSE option of termplot, which returns the data points that would be plotted back to the user. <>= ptemp <- termplot(mfit, se=TRUE, plot=FALSE) attributes(ptemp) ptemp$age[1:4,] @ The termplot function depends on a call to predict with type='terms', which returns a centered set of predictions. Like a simple linear model fit, the intercept is a separate term, which is found in the ``constant'' attribute above, and each column of the result is centered at zero. Since any given $x$ value may appear multiple times in the data and thus in the result of predict, and the termplot function removes duplicates, the plot may not be exactly centered at zero however. Now suppose we want to redraw this on log scale with age 50 as the reference, i.e., the risk is 1 for a 50 year old. Since the Cox model is a relative hazards model we can choose whatever center we like. (If there were no one of exactly age 50 in the data set the first line below would need to do an interpolation, e.g. by using the approx function.) <>= center <- with(ptemp$age, y[x==50]) ytemp <- ptemp$age$y + outer(ptemp$age$se, c(0, -1.96, 1.96), '*') matplot(ptemp$age$x, exp(ytemp - center), log='y', type='l', lty=c(1,2,2), col=1, xlab="Age at diagnosis", ylab="Relative death rate") @ Voila! We now have a plot that is more interpretable. The approach is appropriate for any term, not just psplines. The above plot uses log scale for the y axis which is appropriate for the question of whether a non-linear age effect was even necessary for this data set (it is not), one could remove the log argument to emphasize the Gomperzian effect of age on mortality. \section{Splines in an interaction} As an example we will use the effect of age on survival in the \texttt{flchain} data set, a population based sample of subjects from Olmsted County, Minnesota. If we look at a simple model using age and sex we see that both are very significant. <>= options(show.signif.stars=FALSE) # display intelligence fit1 <- coxph(Surv(futime, death) ~ sex + pspline(age, 3), data=flchain) fit1 termplot(fit1, term=2, se=TRUE, col.term=1, col.se=1, ylab="log hazard") @ We used a smoothing spline because the printout then nicely segregates the linear and non-linear effects. The non-linearity is not very large, as compared to the linear portion, but still may be important. We would like to go forward and fit separate age curves for the males and the females, since the above fit makes an unwarranted assumption that the male/female ratio of death rates will be the same at all ages. The primary problem is that a formula of \texttt{sex * pspline(age)} does not work; the coxph routine is not clever enough to do the right thing automatically. (Perhaps some future version will be sufficiently bright, but don't hold your breath). If we were using regression splines instead, e.g. \texttt{ns(age, df=4)}, the coxph routine would succeed but the plotting would fail; the solution below works for both cases. We need to create our own dummy variables to handle the interaction. <>= agem <- with(flchain, ifelse(sex=="M", age, 60)) agef <- with(flchain, ifelse(sex=="F", age, 60)) fit2 <- coxph(Surv(futime, death) ~ sex + pspline(agef, df=3) + pspline(agem, df=3), data=flchain) anova(fit2, fit1) @ The gain in this particular problem is not great, but we will forge ahead. You might well ask why we used 60 as a dummy value of \texttt{agem} for the females instead of 0? There is nothing special about the choice, and any value within the range of ages would do as well, though I try to pick one where the standard errors of the curves are not outrageous. If a value of 0 is used it forces the pspline function to create a basis set that includes all the empty space between 0 and 50, and do predictions at 0; these last can become numerically unstable leading to errors or incorrect values. The Cox model deals with relative hazards, when doing a plot we will usually want to specify who our reference is. By default the termplot function uses an average reference, that is, any plot will be centered to have an average log hazard of 0. In this case, we decided to use 65 year old females as our reference, with all of the hazards relative to them. <>= # predictions pterm <- termplot(fit2, term=2:3, se=TRUE, plot=FALSE) # reference refdata <- data.frame(sex=c('F', 'M'), agef=c(65, 60), agem=c(60,65)) pred.ref <- predict(fit2, newdata=refdata, type="lp") # females tempf <- pterm$agef$y + outer(pterm$agef$se, c(0, -1.96, 1.96)) frow <- which(pterm$agef$x == 65) tempf <- tempf - tempf[frow,1] # shift curves # males tempm <- pterm$agem$y + outer(pterm$agem$se, c(0, -1.96, 1.96)) mrow <- which(pterm$agem$x == 65) tempm <- tempm + diff(pred.ref) - tempm[mrow,1] # plot matplot(pterm$agef$x, exp(tempf), log='y', col=1, lty=c(1,2,2), type='l', lwd=c(2,1,1), xlab="Age", ylab="Relative risk of death") matlines(pterm$agem$x, exp(tempm), log='y', col=2, lwd=c(2,1,1), lty=c(1,2,2)) legend(80, 1, c("Female", "Male"), lty=1, lwd=2, col=1:2, bty='n') @ \begin{enumerate} \item The termplot routine is used to get the data points for the plot, without executing a plot, by use of the \texttt{plot=FALSE} argument. The result is a list with one element per term; each element of the list contains x, y, and se components. \item We had decided to center the female curve at age 65, risk =1. The relative offset for the male curve can be derived directly from \texttt{fit2} by adding up the right coefficients, and I used to do it that way but would get it wrong one time out of two. So instead use the \texttt{predict} routine to get predicted log hazards for males and females at a particular age. This tells me how far apart the curves should be at that point. We force the females to go through 0, which is exp(0) =1 on the hazard scale. \item Get the predicted curve and confidence bands for the females as a matrix \texttt{tempf}, and then shift them by subtracting the value for a 65 year old female. Do the same for males, plus adding in the curve separation at age 65 from \texttt{pred.ref}. \item The male and female portions don't have quite the same set of age values, there are no 95 year old males in the data set for example, so the plot needs to be done in two steps. \end{enumerate} The final curves for males and female are not quite parallel. One thing the plot does not display is that the spacing between the male and female points also has a standard error. This moves the entire bundle of three red curves up and down. It is not clear how best to add this information into the plot. For questions of parallelism and shape, as here, it seemed best to ignore it, which is what the termplot function also does. If someone were reading individual male/female differences off the plot a different choice would be appropriate. \end{document} survival/inst/doc/timedep.pdf0000644000175100001440000077163312545056257016043 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 3 0 obj << /Length 2202 /Filter /FlateDecode >> stream xÚ­XIãD¾Ï¯ˆ¸`KOª¼Ã @Bsi$$†ƒ;vw›Žã–íÌÐüzÞZ®JÒŒæ¸Ö·|õ¶ªïoß¼û)­7¦L¬É³ÍíýÆ“äy±)Lž¤µÝܶ›?¢ßæØDýþâmºÛE·=´‡Ž;?tÐyŽ·¶ˆ°…ËZøuÇØ–Ñ‚kLtc +>Æy5Лzü_º™‰4¼©#·Iðæ°‰ r$Rÿóö—Mm’Ê–›­5Im*™ªLv¦€Ï¾_wÎL»'‰‰íò¸²yËèoîÿ:{â|@¦ï~2€0Ìs‹¸ÙÄäD¨’ª6,Âm\Áê u…ß  ·Ädœ:Äà3LÝôz²üfRA>áŒGRØØ:©ëj³5iRg¢ñ¯Ml+Ø#n.¢›C¬ òwO»·Åv#^i’–ïþå󇕱yË_ C;ø™œµÎŠÉ’4+Hë­)‹¤,€Vf“ªª™–‰·ÆÀÞŸ‰u §Œ P»¨=í±×4G$« hSØ îè«èDh¡äxZi ÖÃ݇#žêB„36ÚHQ{Bwæá”÷µ=AŠÛ&¶R1NXtG‡`¸› b@h‰½ó뜆†Ze‰¸îÉúq|þ†õISO!«̶†lÎZ}°fw©:¶ @K¾C›L«P„¤nE\’­‚/lóÚt.ýV]¨ 1(~y5“Ðm\5¿eÍaÔ½O½À#÷çÑã6Þ¯R<®ÃÃ8+ÊØ"; ,{ tÊ[2P0ò¬‚±@*Í,îO® ¼eðþ/Êà¡iÄ Ž1c¢{mâöžÎ¾YšÐmiAù[VùÑ8æNâ©lYhi/±±DCÒ×D×% p–s|4†Wê„ûÑMb´æP½ˆÊ¹³? B|l1(µ Ÿ•ä’2wr¤È¾í$<ÁT†@çc\h‚"x)Ä ÷ÏŪ÷(,r©SvøúGÝyÑ`BvD³óî¸qI:Ãêÿ`ÈF½lWo”nƒ!­×ð¿pbpôGn’ûqxF=gÙ¹×|fšû¡™Î¡ŒÍ¢{`= Z$zy™´~ųÙH’x›e;ÉŽœyž«ÆÅ¹}ƒZÛŒ!lf }0B"‘<Òñ<ó?C,÷U=sæŸprD¦O<  ¤¼#ðÔÙ`Ž%ʃƒ„á“ct/gñ6—ÿ^¬Äá´PÉlôéÑåsž}~Ð] '3Îp:ÊñLƒê„é‰óÖ6+ä83ŽO>Æ.² r™ñ¾&Ã3/tˆ©¹M ïå®=†¯i ÝŸØ]ä(‰ŸWÐâEè§N«(ÙÉŠ"’ÈJíÙÝÅÛ<‹þê8«Ò8<ò>²í–Û¾´E©ØT^0Cã%£Wô*ŠÏKnkN˜/Å$ ê&®à…ïÔKü|rãW0s>d«2ô!ôNPîý‘IÛŠÓ7¸;µ2Ñ'ÞÒã·zà­üå-ÒDÓßô“Wìl§m- ªR¸U% Ìh[#5‰Ê¥æI:…¤JŠžžP-Y*Ûžo*¥ñf\(WµkäºôODøObg´ Ô0u}0iæØÍ_?¼ŠøÁn=…ñy€?ü8*g˹ðŒ¿–@ØÈ´'ÑßÏv¹$vc¼dgë:É*—Û·e…‰:»Ì‹ps4u¡+žaâ­ÖìtƒêÜI$}t7îSözÒüqà y;Kв¼Ì#ä/\Á7ªRš­Eq*§ú–;ràxñOþ°qhŽ’7í®¢cZüceÉ´0¡s«/yTƒ’ðIƒOÓj~é{*Ó°yZ͇«>!äkÚD±ý⎇К´¤š_ó~Á ‹ÐôáÈ•YãZ ÌO<”ƒsNwy¹«qR;tU6Γåœô²Öø¯ 8-ÇJÌDwG~ ä: vG~ƒ7föÒ°)µÞu‡æLŽBÞYÜÝéÄYˆŸgZB†5 Í‹$sµ•hµH› •‹)86‹6'y‰ó×=4C7‡CÞýgÏ'åR}Ç«8Q¼ðÒ“{¿¼v °•eë$Ä{SˆQr@B ÷yGÃ!_™ ÒˆþZ½,7š,ËØ1*cG/aa\Á™Æ~’Íùå;/¦% -t(º2#A\ù4³Ô6%”jËŽè^ÓÂAtïùËE¢3µie02‰gyÝ”ûMS¥c|ÕÇ£ˆÓëëäÕŒorÙ:.Læ;8Œ©È•ýîØrc¥R‘­QÜ“>ø6ü‘´¸ðFùõë*Š£ÜÖ ÛWއ52#;-v°ã_E‰ï(«ÔzUñÎ=ÚhV"Zík!¾Y\ͼg~¨&õ L¶òî@þºHyfK~¤]x«†ëF9È+ž¶½ŸÔä5<¨© ŠnT:œÔÄ¥ªþùÞ»À»›~Á Åj¥`• Í‚ê3:9‡Z°$ôF9?bj ¼oŽ×žIÄÖêqòÞV´HãEã鬼êÖâ|øt"µGÓÕô¢=Ïœgì.OÒ"ÇÂ;É3}߇™7?Þ¾ù¹îCg endstream endobj 2 0 obj << /Type /Page /Contents 3 0 R /Resources 1 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 1 0 obj << /Font << /F39 4 0 R /F19 5 0 R /F46 6 0 R /F8 7 0 R /F33 8 0 R /F11 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 14 0 obj << /Length 2522 /Filter /FlateDecode >> stream xÚ]Û¸ñ=¿b‘¶¨ tQßN{$‹ pE¯îö¥hú µd[%¹’½¿ô·w>IÊ+yðŠœç‹3CîÇÇ7ï>•wëpÇùÝãöΘ8,L|—gE'ÙÝc}÷¯àq¿2A³º×Iðe—AsÁIŒg@r;ŒŒ8¬í¢n`|„qਯÕ#«/Ø +À=¯²<¨Æ°Õ‰V›`bmÏ„~Òà0´ä+£»˜ßS¶nñ„ãî8¶)ˆ°™&O¼:¥¨yòôGÒªkVÿ~üëݽIÃušÂׄëLŒUÁžIºæE M¶(îÈã]Õ¡àO¸ÊÕÃo÷¾ûLœ/Öy˜•뻈ù^ÅX`­U–èû`¸C‘¸Ã*ƒ?_ð}Ãĸ7úé„F$ix³¹ß“p ~—ÍÂÕ} Vþ²¶c¿öiÔUˆ¼ðdƒcÒk˜+JwÔ/8ö—£Yª _3`èÙÖ±‰Â,2sc‰M‚Õ*V…ó‚ Ž_Ò~›B]ø'@å ív#‡Ö N0鎔•hI@Œz#šŒ"82ëùK!…ñ‡ Á°™ÉذE°ã¨ÕØ%òæ+E˜¬¤ÃÖ1ÛKdÓ&ò¥;÷´;š+šÛIÌŸ&k4NšF¬É†# ~Bd ¾©8î Ÿ™‰×l‘ŒG¡áñ/dm2ëÀdÑö¢ö†ÓIL]_€‘<;ndzÏQÁô<ß¿}€™agy’OâÍsÍ'Q/{»ûMO’Óy–˜bN 6" G!dö92©Fb™0ŽLŒ¦@Èä\…~*cg)D“¥Ä‡0Åü—ÃQ“&ï3ô¼’< Tÿl§z6²ýƒ¸˜²ü¹0˜%b’¨ø¨ŽŽáÄãí YŽ2…«§œ•µ›Ù&gÛpþš…x\NÊ…ñBGœ#ˆäèÞlÎb^rˆ¬5u­W3œRήoWŽù3UFÛ|^µ£-cd§a¿‡Ÿ]!ŠJQçe{䂸2µ¶Ä-¨”TlZX¢”‡…"šDŽ›lÐŒštNh$·õI‡ECÖ!"™ù$€ƒ¦)¿ Ø"Yb¨&&&%êêÉË@Ê)ïHÜôœ!Ü¥Y[ƒì /Ó«}N¬ØZf\Ô SãDÛ<&¦ìµqÉZè¿qØYó`²50“@šdÿ1øYé§³Ògs’’ÁÀc»?Íjѧ¦Œ’º ç¼N4­“¤:LN–ÊIW==å1øŠEjhGñÍÞf®îwž³“Ÿg> ÈìÀ%ÍËàã ñ’Â05N=ò („q€™$|˳uÐn~AkQ—¡ËU—/ ³uhá7`P>4yÑP6óË0Q·+Y¬˜Çƒ×}­®¡³W+¨PWâW5W+iInd-d|–ã˜ppô#¿öÕö Ðärô²<>P¸œ»†i½†”_,Î\}þCý›mI¡"VºD»ü"½Jðr’Yy ‡VÀBq`bû‹ÍÁÝyȽkr«¿À \°Œ}ìHô9¦¦÷Ý>‘{;íVy}!º4*)±- rÝ[¦ÙrP²­aö‡o¶„ÔŒˆE¿ÉÇ]¥Ç?àY+YËÏ|…Y_]a`‹â?Â1»¯±‰ÖØy#!.Í>w¬á—Á/±uÙDiüŽUüÙ‘®ü«tÑ'Nжˆ(½ö_"†¼"ê/˜x¸µÇŒ£3Oçu#>èu‡âß¿÷ž€:‘íè‰2Nb¸!ª«ÊòYÈG¢Ähƒ’ådc©WÊŒäi'ñ "‰äö¹‘^ƒo*䞣n›Ñ¢CÑ2³^D¨H}Ã=OâBŒŠíM÷LÞøyFiÅÿíŠDÀ·\’‹°¡øÁs6%•wŸfzýï{¼bùÕ/üþpËØ·­tôÂÓÓyÙxvãÎCW´pÁ(Ì'ø}€ßßÄŒ?9YrRŒˆ÷b2Î|‰ á®0—ú«¾‹Œ?PÖ2Á?gg–Òj7Ø· k³V(w^O^¾ub^nlx­|+¡ë›þ~I•‘¥Ñ>OdW§å´q pã_o„O-ðñµž8ü»Gzo_‰¾=Ó¨ojÂ"O_—tßß±øÛËO3®éõST¦OQ\¨YÇöÿ…ý $0ßÍA8+£0ÊŠ»û8)Bc “ÆH÷æ§Ç7ÿj” º endstream endobj 13 0 obj << /Type /Page /Contents 14 0 R /Resources 12 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 12 0 obj << /Font << /F8 7 0 R /F55 15 0 R /F57 16 0 R /F59 17 0 R /F54 18 0 R >> /ProcSet [ /PDF /Text ] >> endobj 21 0 obj << /Length 2018 /Filter /FlateDecode >> stream xÚXKÛ6¾ï¯ð¥€ ÄŽ¨·Š¢@Ò6@Š^ZlÑC’ƒlË^%–äJr¶›öÇwž$åµÓb±&9‡3g†2‹þÌ¢Ì×yV,â2\‡I²Ø¶wÞ­Ë,Šbðº4ed™^¾mÍâÇþîWøÓ©•ˆ\y2_ßß½|S,J—-î÷vÛ¼X‡E¹¸ß-Þ,MðPMËUœFA3bm?yâÁx†þÿ'üiàÄŸîÀÓÕf¹Šò ?»ÊD^W"¡˜@B°›XÞzùáþç—oÒÔÓ7 Ëu–àIÕQ¶ÿ"¼ÅŒuF±rö{O‘šû{èWÇ‘¶¢kè¾MŒíöBójz¦è ÏÐÓYwÐÅãnqén½\¥a¼b¶ú%pü….VQÈ’Åʘu G#åZ'¤>¢’…9£–‘g’³€D^F…Òw¤jÅ4p}bÆœÁºãznèd_*œÜ¹5Õ$ví™Fƃ֬Vrßu‘ ‘þ0“ŽZ}^¦˜ T3·ŠCZ=:¬©;ÙŽ²Y/œ›Í„a¸¦ “¬Ëä¨÷ÖZ lOGletÏÒaN½ ¤¦Cö Ôz븫“b¿¯¶è¦“ò²œ÷@8ÔSð¤ý¥îj5]ƘHO<#ÞA×¼à~$mþA…22|­õˆ¡~ž:#CîšqäœI”ð-<«Ç;Ô„¥ØÙz…êÒ’-ÂbŽfr€øÜäˆø$Jé¤(£9< Û4‰³çæX“}ibÏñ€ü„cÄ ¶zL컎$Š)¢àÄN½‡*ž U ³°q‘ë7Ò¶í;v€Jœ¸ æóa–±­Ð!Ù>d›à5ªP“«Ñ „@‹¿õ BZÔuzà9ƒÐ‹<1×É^4Àˆ á“EÁãRÐ3Š3¾ÛÍgY±cÓ“œÆ®î™SáLjÊê;úäöØÐ‹@è;–÷HêÚˆ#gšÅåË»x¢°ÑX(ƒöhœ< j†ŠUñL8< )I³‘H† Q<7Æ è"z‘gÑ¿òA!“Z†"ý¯OYú7 p7ÂÍsÑy²ŽÊ\~÷_’3+lÏ­œa¸ëG789¤ÉÝÚ1µëØ+œ›`”;Q 2ùL|C€àDbòuR¤s/L V빂ÑÓQZàîx#> endobj 11 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-fake.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 22 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 23 0 R>> /ExtGState << >>/ColorSpace << /sRGB 24 0 R >>>> /Length 1707 /Filter /FlateDecode >> stream xœ­˜K7„ïý+xÔDóý¸Úð. ` Ø3€a>–Ö ¼Ó»ößwD’UÝÍ,#¬šÑ7,V±’™‘Áòæ•ñæ½ùÏé{üûêüÃ?¿6ßÜŸœuΙëŸ÷ß|‡¿Ölþ<ýü‹qæÍÉ›Wø÷þäyù×)v‹??šXª-~âs¿J6Qï²u~Þð2^-—éîã!/ó7ÜÇ£ó¶_=~âeüz¹—Õïã_?Ìàüz>Îù×ÇÓWÿˆäÿM³óïã‡`Œ¶9‚uÍ<|4/0rgÞŸ¾};njï6囩ù¹S}ÉÖ_?Ö?ÿ±!Ên_Ï}ösC/6]?7<ÿ¹±àâ|3wîØÔlº¤P¨Åºp•aû°ìÝ Ëh¶×}tÐeÔãWjûðÄ«ñ–¹Æ}|àeœAË~Ÿx5>»_Ö~–K0^^"ªmɤ`»û\}fnC(üœ¾p®G÷0'§/\Æ4'—/œB²=ÏÉíK'C&°ýKêß(Ñ)ÏÕM¡Ú0`k=¶fÛ¸[Ü.¾Õü?µÁ# [4Á6IøŸ_ütg<ÊçÅÛ×Ogóû§óÍ›w¯{üt~w¾ûÅ<¼zVš-ˆ|’?È]ïÿ÷tg^FÜ÷wܙ͋×öûá=Ìõ»Å&R¶µ˜§·æGóøw´‡­NÉ[©€Q)ß ‘¸KUXšm²ƒÉº¨±H©)k+âA•Ølõ£õ ˜+ô¼H0;X$o•¥úVÌ•¿kÕ˜m‡üœ0) ™ö¨ÑÛÂà`9’B,£ ,Y!^ ¶F™­˜˜± ± ¹¸2qF6$`£°*Ä2Xllï #–ÁàÔHQUXmd`+ ?hLÒNµQ?Fš§‘(ý,°6n¼B‹6…¡1 ÀÂÅ*DqñÙ5¦‘ÞÝIª¬ƦtOaRèmlؽ„lpë(Éž¤`WÌ£4z™\1Ж+E{E×ldp`ib׈ 08Þá ëgš r` i–5¸Ä6mÌ|R,…CŽRõ {±q=¥0^ÒONñ€·u02Gs›Dï ]Ñìm«Â…¹¥Ù¦ä]e\·YÆŠìÛב`äî8S*ÉžÞX3TÞ×Gª—æ0’ßûd}<`”[ŸŒW ‘Çz%ç4K’‘+uJs’à}g¯Ñœ¸Lp€Kq ±’÷A5•vÀžFŒÖ™4—Îm&gæ«æ2Äßû0_c=nrí,i Ž™š!ý/œ kŠsg™“^s͘ õÒœGãó0©­pœù–ÄukÆ‘, 'ö*ÍŽœ\Å|¯L¯3®ïb«Ï–O¦—W<'Þ¸9Í00£ž‹tvÍu¾O‘ÂÕ¼ÕK‘î£Yl 9K=*ö3ÞHÖÃÊHƒ(Ï£/òœ)ïdéãšgÛ“Â*ì˜æà&ýX1ÚH”÷Aa<繟¶Ü8ÎzîMòA±<8ùb´“â……_sõ¨·Gmb<ô§Ù ½eŸîšû”› usÅ<’1£û3µ`ëc#ÄŒ;…™Ý ˆzÅk¬ˆÞ(Ëj³ üOãÒJc‹VÌüI”]1Ìða±„£/ð.§óhM¡‹y[ÑŸ½XŽá‹8wìî¹áý~ü0òŽè9yD–ƒÇt“ÛÁcÃÀhµvÅ2[}•Õ+,ãÕ§]q:3´#~‹Y0Ï&Gȯ<+–!Ѱ<Ã,Ý*׆E1œ˜ëò,Åa¤ÑÀGñeƒQ€šó4hh 9ðL6*¬nx¬ (…ß4Œ|[Ùñ1{RìËÆÈÍzÀaãfCÓL*ˆ£ÕæWo8 ¯ÏN(ó–i»02˜§xzg |NìÇù‡NÇ3Å(æ½LCuÃ(g9Õm¯8N®rPÌ4/S i0'öE²|¬Õ _™§€cãà E¾òc…bÇo8dI<ÅÈgß/"ªÙbâa}e¤Y‘û±“çöS´xLmš¡ìã~È6Å~4`~0@'WŒ("Pí1°›’ 'N¹2ò_åÛ6 ÜÂl›Ï+ÆqøŠ6²nA”þðöüû§Ç7oŸÎÏþô=ï lãß}z|ù¤îóýé/r˜] endstream endobj 22 0 obj << /CreationDate (D:20150701163333) /ModDate (D:20150701163333) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 23 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 25 0 R >> endobj 24 0 obj [/ICCBased 26 0 R] endobj 25 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 26 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 19 0 obj << /Font << /F8 7 0 R /F55 15 0 R /F11 9 0 R >> /XObject << /Im1 11 0 R >> /ProcSet [ /PDF /Text ] >> endobj 29 0 obj << /Length 2799 /Filter /FlateDecode >> stream xÚ­YIãÆ¾÷¯ÐQ´8dqÇ2Ø€ã$n ÏØ$%1-‘2)Mg.ùíykUQb7b`—zõªê-ß[øáñîÝ÷i¹*ƒ23Ùêq»*ó ÏŠU–æ‰ÓÕc³úuý§ûMjâõö>Zwð;óãáv÷5ÜðûüNðÛÃïc˜†pù~øðûì^Ÿ…×~-ü"ø=0³ë!ã†&™YÉOÌ4bŠÿò v4Ÿÿà¶9 ×ÊÛDï][ZíÓã«<Â"[m¢((S‘DãÍÄë·²Ï/ð»£M£wßž|7–m¤IÂlèï7qlÖMÅ“áÛž™Cš{,Ò0ÈÒUÈS—W¿Y6N‚,Ñ9²jŠìZ:¼ÉQµ¸dÚíö÷¦Ð7û ‡?ã›–ßàĽÜo‡<È €ì…„¹ãU†'TRkM MeQ¤îI(íM$&&±DYÆûœP¿O »dýodPÃf¢¨,‰´ öÈ„¡÷†,†_Ng=! ¿­v­Ü‘=´V ¸igLÀ5Kݶz÷ÝãÝow3\E«(OAà0¡ ò"YÕÇ»_?…«\…A\«"=®L úˆ ÜV¿Üýãî:Ÿ¯¤¨L€G¶Jáj"œ0NÊhÞoLQšu”ÂMY„ô")ÌÚÐ >ÀÆc’f87[äa’ˆçf‰Y'Ù›LÒ·˜Ð\d’Ço2IÞbBsc[Â}M¡‚‚Bó:õR D9Àé’гÒöLMéBÕ»-«ƒTgµ;Å[†‰™–à‹ñ™T‡ŽBëME*“Ñü(rÖE|8 mœÓs”B\IâÌi"â|æì-¤ä°y [Ì”ü†BYwo7!ó^öœDÈÉG~;ȵÎ|C>e… /:—°cŽÎ=„¾ot{š<Ñh`An­dW¶Ì( JÈ‚fÄ? X%À®P¯G”` ý¦ £0ÙñHÈ9”žÏj#Ïl6*~žÔE.'žàÑÞJV`8¡û¾Jœ ÐL¤POã¬X!Ãå,BâQ³ÐC¥nläÁ[¯Ðµ“ìƒä]+ø—”ÐÁЮ#×Uë„$¾H…äµø# ‘kL|Äm£\ê'ˆ`xšæÐpvéÚpfœ ã~z$g¡­ÅÆa´ä(*Q|ÙhT*É,Ñ>nØ2öOÃÜÄÍìéÜóãdf 8®Ò?jJ­…?{2šéM"¹£1ž™ΤS’Ö8ìFŠÁ”<÷”0ØÝçָƳ-)áh-s¡ý¬BAyÏâ@k¢ÿ†yá8ùÀu M£zÞŠNóÅ-/™ñ‚cžØ:þ7ó¬/“ˆ<‡ì½×ÓÂÀ@)ÜÔd¤ÙòÕ- $6wO}Ë#‹xÆ0•4¡”Wð ²I–¿Èm ­ŒšÔ2<Ú:d¾?W]áf…¸­hýkhÏ¢#ÅbŒ )—¢e(¨SÖb \“qÝj(v¹W|e”°ë‹ºÜ{óqìg™ˆ WUãG…·ÞEšü‹%Åè f¡¢’ d7%^æÁ‚W,Ÿ,,dK°>=ðÐùje—ãfž‚dé-¢ðd ˯Øò MCà^4/³œé4ÙÀ{!§Uã(qAá^u@jQsç4H)BÛðZÅ®tµa ®rm¡üN]%!a ^–‰)N„ÔšŒUãÄï;ÇÄË5QX$(š¥éÎéÒ×óá‰ñ J\¶~ “À{?Xëom:›„1b´¿ÅîÉÛñAã‚_Îx§U]Ê!TݤwJt•­ª ¨êÌWÏ:&+¥¡—N • Î Æ8‡o£@Mb¢d XrI3Ç.lJ-+‡$ódooýÖ$ÊxgüÂÎk€Ê AÙB<¼®«£{pÀ&uG° D®~u)%h=(4Ì‚¸µç+ê#õz6 lãp/Õ¬¸ JaÙ*pê³-0/Ìç˜÷gì „³¼F¡ß@ÎO¥Ï46ÇJÆNZ†ÏüÜõ|}ê`Ò0®’Ç%”«ê‹UÐH0kK=P»DB;{Nî<‡ÓSÎÜ yÔjòä&‰Øš~ë*ÃüÐ6Yp¤‰»«W®(§!m†#q26¶›ž]‹!;¾òxÑ(±ßa­…YÒ™NEb™C!¼A”‹!éCh„=Ç&(ÙsŒ?ð÷¨xjP4¶É˜K’Crß4í +©QÀÃbvÇ<‘œâ,";)ˆM½.nˆÄ¦2qPÈvêYãžÉã€ÊäÒךAÞÀ=ê¡ ·¶œD¥¤[4÷Ào¶Råôõ¼Ê)ÖÛad’'!é‚h* l\¡SºCãC§ýÖëvWÆ6±Î^÷}|½óG€•NÂÚsú½ÍÛöþšt{¢Øzæ„S¼z @û½p¹ÉG²e'i'ë‰ü{iE‚~“‡šœ t¤`H3»?ç׳'6ñÍö Mš«}ߊªH‚°üÝ¢Ë)JT[ˆ€U·ÕŽü\ñú**Þb­ž oy˜¦ péã{㈀äŸ\9HYïÕÌ­ðøNóÊfŽdµÔ˜Ù}a‚¿ëÖ8¸ ÐråAYÄs»’2³U·lq#uLÅß ¸°z5'MYãŸ=Ù1w•@ò³Hæ8¹RšþB@#$&`+<ö“ò#ЛÈ0*Ûª6ð!*³ÔhðêËÖ+ýHºWÝM’†ëïï¡Öa˜L®[#^ÝÚϲ4‹š»Ê:?çVbñe†ß•³jMÝ%{9RØxê<ˆ8kû‰š½Z~ã÷fÊ)Ø´Îg_¦Ú³Ã‘¯Q¦<±ç½ç*«ÑL BJVR–?‘]k(ZIÂ1AÆý÷Þ¥,O"r¬½°³•Ã×Kä]Ð?øátY,å bwŠàš‚ c˳ê©4Ôâ -Wiä•öÖƒkit½ `>O —öÏ?ÐIçiýz«= œ÷¢9Š6 ‘Ò˲ŽÛG«JrnZ.EïšYF dðIîü îë%¯Fo+ @ñèH’+œßxœ½ÏÕ“ÖÝ“ìkôX4 )}4Œx¿Ö¢ð~GŸ¿Ašôª‹ ä[7Î}X þt¦HŸc;ù\!v5>»A%Yôb­–rsÖƒ ¬Å÷^_±dåTs*‡œFÒ'ÂBß¶a°È·~å\6¾ÃæN´Ö¬FæÉä`KFÛNsS¯c˜Ï6¨*NÚ™´t&äë-l.¨­¶ed·ÔãQ ¢ªæc%½mLÀ ‡¿ûá¯Ò/ I H[øsÌÑoÍÿCÒó$ endstream endobj 28 0 obj << /Type /Page /Contents 29 0 R /Resources 27 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 27 0 obj << /Font << /F59 17 0 R /F8 7 0 R /F57 16 0 R /F11 9 0 R /F33 8 0 R /F46 6 0 R >> /ProcSet [ /PDF /Text ] >> endobj 32 0 obj << /Length 3542 /Filter /FlateDecode >> stream xÚåËŽãÆñî¯ø²0â’Íæ+ÎÈ:¶±>8‡[ ‘É+‘RÚñþøÔ³»IqÖ‹EK3êGuuw½«šoï¾zý}ySEUnò›»ÝM’˜¨HÌMž‘I³›»ææ«»ý:Yµë)’Õx†ö—-üÃÎeð³;hö·ï/Ð9Àßþltð÷€“†X›jîFÙ aúµ)VO´˜±â,c´<­7Ö¤«Ÿã,†±»—ðîød1oŽ’Ú_èô(HeíË¢kânþ¼þçÝ7›ÄF•µð›DU&ÛÖÜ·?¯7išÒ¹sD¦fuÀ™íQ°6ˆ°a0&Ñ)eÁˆpÃG!èÇu–¯ê#Oqä^˜Rϲ g|±»%š!= “'ǽ£¨Xxæ¶¿Œa‚µˆìBià_À‰òúû¬ „*¾Ù˜$²©eâüe½É€]ò‰oC†æ†›ˆ÷$É ·…É ±•ø{#„¨×,;~ËØæàFÀÏBÑSÀV•Ô`ùAv|#‡Á©Gm<ØY6Ö]i¿‘¾Þ!¡%(1¢ÑÚã8œAÞ$AlUälrðêåDvHGrf|’T¬^9)Ž™ÒX ‘Ó™ gŒ•Ç:cì-)ãõš©‚âA­‰Êj–[;Qt!1v^·,ÒbùJ-iÓ”%³vVŸMÊPve¤÷iüÀ#“2_ù ZÓô1qø‰ÃY.ü&…{ _Úž>2ë¹vìäôó"ÊB}´Ê#k‹Ï3*¥{Ï‚;ùJJ”¡-×û8ÜÍïDœHAàLQ|^.˜rÝþ׃³àÝá<ðÁ?v¯âTµ&•<ÍÖËû¶w>¨vÏ÷W ùf•vœ lü]ÔmàóüôWéûÜâIƒ­6bäó'÷,**!QŠn:‘WzR[®¾ý:ck15-\Ÿ¸b+Ú2JË„‘~K%" rÄC !3abÝŸui-‚h¢5¶µ”hpTDžOÅ"ol¬¬@Ìx¬aª:µéZN•‚ ¸¼úÈ(G±òA'âèLhŒæAM(úžÉA)Óh4+X %~–ÊU½=‡HSÑ)ƒ±A•ˆapÏKÀ–èU -ßR=1Ìôó”v<ÀˆïÊ'eC¿_vU[ÙßgºT/…¡–ÎÀè„F»Vkº 7N.ꨶ–Ö’Jª\S o»ðšŽîÛ¡ý<æ@ èßÉŠ¤*Ù-'«œ0xãê“ù*ƒ+aPÐHÁꢕ®hW¬ ò] Åc=ùØßô‹ÁE3gx6aÂ×púyQ½m×úS¿Y%oQHG¹æ}ÖO¤Žcje¡™˜’ªà`´iw.pÍ3-œ}Ú÷¼dÐ÷ Ö. Ãüi_ ®½+‡ÕÁq˜¯vN;y¨O§Å|Cì¾ûþýflµAÑõQ&ÙK G¯hÍS®HѶ~À=ƒJdÙÔ‡@ `@‹#Aÿ¬ik=DXˆÔZ$Å€w-š5´AïX¹ c\¼;Ƹ5´*Œì”¿T”0Yk;‡ Ðú…Õe^Ò¦¨ÁN+É3·9>°|_2î¼Ë£¼(¤vÍÓB¢ñv«Ñÿ ×Ã9IA…†nßFí†MRÝblxJ#ÕL©ž/ÁÓV$~Y¬þÞyÌ‹ÂNÛWêsÿ¢IµýŸã$Ç»ðQž;¿0î»B\D8jT‘zþ][ío$ØÐ¼1NbAÆí¥ cò‘ÃΉsîÂy%½žøRä,F9蘑Cë‡~ÎJ¶§ñ†/èÅ@ŠIø¶ $†B¹zh¹áÎjZ)ƒ~‚¬ÇB,¿oˆôpü:•r?+ß¿´¼hR)Œ¹yä;#îG*6`k¹h2ˆé+÷¡ÉVŠ~xÑxñÝ.Î]µÐ#qu œº¸@Ý¿ƒè•YϪyeJê±\Ñ«05¯¦–׿Q^¸óÔç!ðN¥ ;$eȆB*abž:}6Óȵã±É“3Ç…žL¿ò(Xài›þ8ùw#ÇsÃVóâ èÈŠ6<'‘“ê.’¥ rŒG嚊¿ÔO#ˆn¶¬¨\µ|Œ¼ú[áöåŠjO€f®Yjà›`zàå`„Aã‹b¬ÃÖ—p±Mú,mÃ?Ÿ´ò¡÷ZBn’êz—´Ho'diZx8"µ7±‘„§˜™°ÙáF‰sôÖ}´¼@^˜«º¯›àãâ®u¯&΢4ÏP ¢²&g8õÕww_ýEVô{ endstream endobj 31 0 obj << /Type /Page /Contents 32 0 R /Resources 30 0 R /MediaBox [0 0 612 792] /Parent 10 0 R >> endobj 30 0 obj << /Font << /F8 7 0 R /F59 17 0 R /F54 18 0 R /F57 16 0 R /F69 33 0 R >> /ProcSet [ /PDF /Text ] >> endobj 36 0 obj << /Length 2070 /Filter /FlateDecode >> stream xÚíYKoÛF¾çW¾”"†ûâ£h ¤Er(zj]äàø@‹´-T–\QŽaôÏwfgö%Q¶ì¦é¡=PÚç<¿™]8}õæC¥&uÞ”¥žœ^NDQå¢4“ÒT¹TfrÚMÎ21©29Y¦.6kx$ÿÓc§ü©á©àÉ-¹™(t••L‡Ç¢Í£HBI$wê*f ¦ç§?MŠ °ÈÓ.ò¹º‰JÈÆÄMäšêëgƤ{º(^˜Ê²µdP”2åÉN€®ôÝH{+Þ[$ƨÉêIc4£ŽŒÑ„fÅnˆQEŽ­ÆÁ÷”g=â:}PÚ†9«§¤Ý¡Ü‘VñxÃc¤­Ç¤5AZ£‰¤= hUhÆbXZs¤mE*­JQT†®94eºÅ¤º(vÈÁ ƒ·„gCr«(hÌñ @±Iô˜´Õs¥Ýuu‚“*•Ì[¾yªË4]5‡@^¿ÐÔœ—\ŒÄ£ˆ’“>^òQT4£,bÁ§ ¨®âŒ%iyÈõeÖ­‘Ç7Z%âìÙ:)p$9_è ¤èÛóÀµ/#1z60¦xÚ—åcHýêκßÛ ‰õRi‡.êŧîÀ‹ç6Ä¢0ƒ«T@f‘:ì{ºÓ¬8/ßÛÌê® vê;hΨ¹uæfŽWÜæüÞ1DvFuW¯o£ë×y˜ÚÝÅà zËhèÂ7‚ï””ÞÚˆ¢|æC(ˆ'Æ2Ü—Ô}œÈßPrÁD/™ðœ°…Soé²Í®Xm–¦±‚p–øB¹gú@ðúK˜ïkÚIîØéüP_=bò¿h òÚZ£Qeë#ã\Å Yš˜å#Z‘žI¼ÜÕ§ø'U©Fà4³ ƒªe[†Ëq09ás>*î"áN"Ö;/Ϫ̆«„WÊŠÁÝÛ×vbއëqXò0ß®q芇nÇ%æ8u±#Vñˆè;Ößmß4]mxzÁ’,Y0•Žììô{Ú½ð2%¬”È¢1¶b‚pøWBc)d¦ECýnKEèóú£qô8#ù,Fåˉ¯¥ÑQŒöïn1"¿bº=ò|~RÁŠa}ÃÉhõ·¼|øÇw:Û£à–‘¿ä* ð3tsO(z–ž-¼ú$:GNöÓʘÄ,s—Õ%‡ëIÐÿÜ'exIÓÒ¤Vˆ¼n1ö/êSÙ²†Cºûj\G6ø½¹›úQ‰‹è®ß3„ÙxÞÁó3<¿Âóþp’Ÿ9SÆ1uÆ7åsRéš¿Àó›%u¸j“8¢hñ‚\UÁX+›ø7äBœ¡:)¾/?à@éê8gUw¹¾£éÎU•Ùž‹r8¹pÍ*[üípÛuÑÉ´Àê–Þò`LïÜàe‰e#¦:\û ï¼Ø‘—|9rºJ™7ªqÆxé{Ìž‰gŽp‚*®tŒTțܔú…RȃRT"ºùã¸ç¥ÑQ¡ß•ϰP^˜ì£5¾çŒ`æ..ÎÞrá gw|‡žX <µá5åŽC_â5*1¡=´®V âuѼs ÜÌð“ÌNíÂo“µN>m`?®©ÈFe×\MC}]ÑÐOÆUYCÖÂÊ6l)cXSçÂÔ{êtÊй¢> endobj 34 0 obj << /Font << /F73 37 0 R /F8 7 0 R /F57 16 0 R /F33 8 0 R /F11 9 0 R /F59 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 40 0 obj << /Length 3272 /Filter /FlateDecode >> stream xÚµËrã¸ñ>_ášUeq|%•Ãd³“ä©ZWå°3Z¢,ïJ¢Š”Çq¾>ý@ŠžT%›x4F¿¡?ß½y÷!Ïoš´)mys·»1Ʀ•q7eQ¥6/nî¶7¿$Ÿ­ÉV_îþöîCbËÜd ò~»2Éöþ]à¿ýi\­óÜ&Ï:ºÇ~ž´ü9"ÜÿFœ?ÁßOœí¸ÿuU”I{€§Ž1¶ƒ4Æßd³óyµ¶eÒ!)“:9Öš‰]øÚï?ÕÝ^Èh\{­Oðï@ç¡Á~ÇŠ*BQÚÔXç1É@§ÄöÕ¾y‘Ú"×Ex¦‘·@v5E‚'k—ôw‚®= }6‡u×=óP÷/©cˆâþIº‘˜¹.mü¯¨óg¦/ï%pÝíjí²†7ýݨ­¨{‹‡…¨Ò¼rp&mŠbrLgj:_G¢ð€²7 ÃŒ ÇfÃ{!ïÏ<õ•WƒXp`ïß#ó¤q÷ŒÇ,@ÐF ‘à¼IZÜmËpÏHr;2ú³2Ĥ =3‰ÚíN6š÷·í G”´÷DS|]›;’]‚ —>gE«>á)„ >â£hC™PQIÇqÑ L¶‡þ±Åe²÷ëOŸÌÅqwá&JY-ŠmÜ,X‘JÛ Šä^À#û·À)ºüËØJ6£¿¸C0Ëú-v´ä>, ¡Óåw‡†IÝ\Á‘Þ«¯Q—‹o_•¾Ð›D@Ëshpû'[ Q²2%êÅÆ^!×öä: ¨¯¥ŠéŽ´Vq)|Ñh8êdAûê Çi ·9 Ù­kØŠŠÎ`ß~ìhÀ„m %¨–ÄÏ3ØÝôÞÖQ<„ š²¥Ý>~7`çÑ}Éà[ÜôD$÷öêBìÈc{-¸bðð¶È¦±RLàc´¦=s¨ŠÇ”ˆÖ$|=ŽYç†C´žnÂÌB~Sªýa ‚ùØXåÆùïŸCJaÉÀð”q=ì™w´âÔ_¦K©CZJò6H$Ž«½#‡¶" 9õ@ó¼ÒyÍ„#Ü{áùMæö“^a·è‡EbX5$ÿÛ´zSÄÃ(<út«ûæƒ?óõD^‰ó—"ýU‚¨=ïÕs¨B”‰^£„EìÍ}¼Ü1"Ô¯ND!z. ÁŽý ñ6¯éO!8)'EQ…ˆüÄý§Yîʼn€lÕ*-D’db8ê÷ú¨Qñ‰Á¦çxö)qá1Äç£ìÚU’£;S°ëÊØv»UgHŽO@4Ê.Æþ?ª™£chJ—gyL‘X–ØM®„µí)¤ÈíÎûg¾Î4ãŠHæ ²ü:Áæý^ÏâuWò¿£€3^ºýÉíd±ý‡ë­ä‹jŶ‘Û}Õðí}œ‡ÌJ$y•–Y¥ îä¸d‡¯/ú„Q}޳A*°ØQÔ^¤Š†ÎŸAðN±u€‰ßñÏ„nÆ`ñÚ!°A|qR‹§E]¼€6Éñ„ÒV°OÜë –Ë%¥M³rV.á,|''Ã˯°œ¦T‹ '<¨è‘ë![uw#ÃÑ]áƒK©š¡ ÆeÊqŽ›b?]¸bkª´Ê_¹c“üIþÌB) šµ/…ÑUÑ¡°\€¤÷Ÿ3“3™ql]˜pƒµ"ßý]/ ð·Hì¶Pk§ì¦bß´\ñÕÅò¿¶+Κ«ÒÆêS¥¾æIiŽV‘ÿ°fb¶¡»Õ¨;;_%Äž†´çY—±·$gú,øeã ÃR¢9Ì_dùvúñÂ3­Ð6²èPzfÖÚÁ‚ñ Á4ÛV výØž¨È‹ó^½¾À£”š·˜Ê™¿‘³™5C j„“øJÕ.,ñ2M=súòuÔÒÈäÌ Ô¼È²Ó™”`ÎN®ÜL®»›IÈMCÓðÐúô4Çœ{d˜ÇÃÉÙl.ÎÈ&ý’îÖej •ÓT­·×òlŠ´ }+‘"Ó#u3­:¶ÃE| Ä¹—/ô SÂüQÝEÏN쇘1­@>¦¹"ºû,›°º?üÚávà ¹?M‚{ž=·,ekÛ4Sl‘æ?‡z€é%ß“Z ”PA’$Û<ïÅ3±{ø=–0gˆÎÉìÃÔU&%pña{ïƒn%õ>)A/Zml‡Å¸9T µþ"6Ì•7 hñ*a™­ÒºâU9ÚuÈF fI?_Zk½n9Üw8<\$odËpxƒç5W¹á@a*6Hbæµ™hÔ‚fi“•*^¿Ê#}—„ÒyYœhGUí¨âpù¢E"Ÿ˜W¬Â!RŸ˜'ƒÚ¹3IŽ+à؆üŒpªb:]6(èHX«±z^•\zÚÆÒUסdpµïN² RW¤…«¦R@ðd[j8|[}6k}¢W[¿KÅÞÐF"Û³âGI{ê<ùŸXЦ$º}aªïaã§'þþsùNŠ&O~¤ò„s0qÙ€;è.LFÉû(d~ €¯5œe)“y ×+;}ztYé_Ÿ¤9D×Å—Ç¢ËBz¬°¾j9EÃùƒ¬Ùõš¾òq×xégžãÔv£yµ7ˆDßþP?ùËQ6ôÂïßðÊ<ùD¯[>ÉŽó‘—=¯üåΟB™ü§Ôù4óÙ)ÊÙ„RºÛr î%æ‚Öä‘s{T'?#hù¤}” j<‹è =MÞÇ„®Vˆ&²DÒâ’™)hžEÖzÓÖRWA@ÑPˆ¸o`1-ºÎY¬™=»Øp‹/2 ÄN-·×SœïyݨaÓC¸ Ç®õõÏÞ”ß>ߌA#kg¹‰5¡@“:ŒOLÞ›­•½dómp¾¢V>ŠâÓN‘““Šî¢&‡ø—m, œú#ùDí1ªëK¥Â \T1ïPR/îÚC dú èï»PªðU¹Ë¬ õÄõ?&/zVl¹ +Å—q^£¬O6nŒMM]L˯"Œ£6>_ë *Q'áíÈpjàÕÿÅY#L—)¥e4OæÝFÕ-M1ÓB4ñÛ†:(:éEáV2iRÓ-ƒFäü]Ýþ"”cÅ©¢'Å*ù4\ð!ýŒû¢¬¬Zý¸"7R8Öyc"‘ÀÉ{b%ÀgÞqÇÍr.*h)F 1ÁøÂ|c¦ŒTÅnü>ÚQiˆJ ÄÁÎÓâUlKõ¾Á<ÍÖ‰ Áu¢é7UòÛ”øpû °³ªµn=~ã7¼Ö†ÒD‡[ÙŽ ²<¿c°¼ÄŸµ0Üš‡\Æ_Ž[‰ôD‰Ò'yòå$ˆz,Ž+QäO‡™móOÒQSž±ƒ%+ØEp)–Ýë3J´K1 ¬iÆ@6‹Îqöì4^4C¸Äeʫ穙£ÓWdù²S´bA¥*º–4eõÎdïŒ^2£øBrY¯‘±bGô>ÎeKuWOaˆ“—W7ÉÃð%žL™–­]ÆE ø-õYl^•Uò“D{ÈoMph‰H]†¿¼_gà°þlÈÉ#³ 4ò| ·e7{>F¨åGršÓ Ÿ?Qwm:„iü{¢.¬H±á£\2yc4¿ÕúéÛ¤>ÍH9 ‡£Üíc -pÄ<Åñ /VîÕÔYꊒå*uú“½ §Þüx÷æ?Kãu½ endstream endobj 39 0 obj << /Type /Page /Contents 40 0 R /Resources 38 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 38 0 obj << /Font << /F33 8 0 R /F8 7 0 R /F57 16 0 R /F46 6 0 R >> /ProcSet [ /PDF /Text ] >> endobj 44 0 obj << /Length 2825 /Filter /FlateDecode >> stream xÚÕÉnÉõî¯<9´±§«zwìž` LÌetÏ¡E6%Ú$›aS–åùö¼µªz¡$È!²ö·¿WU¯ú§ëW?~¨.ê¸.lqq½¾0 Ôóô¢ÈËØ¦ùÅõêâ÷(/Y]Gï¡Èʨ¿4ÑýÍå"Ï¢OPm—'ìÏ£4ŽÍþûümá-[Ò@ ¿Î+¢nÏåªÁÁGn˜„Ë .gˆô;Ok °ƒŒàª¯^Bì^ŽC„ª='­6xçE¼JC„“m|°»‘µšîᦴÞÇn³Êf¦xgã]øŸ¯G–f 8•ÃÅÃ&uŽ¥ÉKÑE$ÃȼÈLÍí¢(]?jµ’ºu/3¥sȨ¿€9å¡õmhzž f;Rkø*1&ô¿‰6ŽâY¢W© @èp'@¯ÎíxºMâ追aßzúóùxu.ÚµådhÆ“Ù[ù½™SÏZ™këã­ƒÿ1ÎlïžðEnc›dàFp”/Š¡ÈZRl І³Búán4?‘|á›”öe%&Ô_ q ¾ì—\–ð+ÄÍÌhÛ–ʼ…‡ë sjó¸•q^'·!Èì;v”…¬Óõ ‘ª¥F¯å—JYŒHVvŒ°œÏ£Qªj›Æµa¢ÿ!ñô³¨GCó‡šc`A¾‡µõw"ÎØÑÉ<Çbë víí÷ ‹CNJ6xáombá*ăۋע_߸“Rp :{èè=\Æ?¹uš,®³lޝu;ÄôœIØašæ0ë–Ô’‡Á»öCÏ=Ý^‡|9_ ʽAÿ²ée=@Õ„Pèhg¢¤M’87J¼aNÏŠ¦ dI®‹èLE ûàÑb-K*Ì QçŽUÙÈPÈ,ähA\ „ÝŽRo$‘’'q•vnÊ¡µ„ªP©NìÔnQê0è¶Ò}¼½çœˆËбBh¬BíIC-ØLfó¨ÙKþt4ëcb²£³qêZúth++¾\æˆEÕwãÔîÒ?L9—íLÜ;ÓêQº¸ ÷µ#c‡+ÕPb¿»ŸTÇ·´Eš$ˆƒS0ÔóÐc09¶Ø¨É]ou†š’Tù½0ËU,M²ãs[•F?)<’U»v×Iê%~œe%ŸõÚÉ/xîKýË/VAd[>OO+EÞdg6Äù°\ÄEí¶Aõµµ(k98øãØ4’Büó‘”øw䓜[måÕVuðI ÙLõÁ æÞ¸$pKÝew>@5[ŸÂB úº&xî6.½]q]¯²ÍNOæ8ûvIx^s nDtº’g´{y`‚utظ¤Å¿Ÿ½FÛ^õyP_êƒÜ–­ ÿØI/vrrµUâÛsE^ƒQ¢bêØëŽÌØxD&À¨AP,Ÿ¹¯adØÓ>pWàâÞ¬yHß-«bàížh§wÆe+aß¹í§S@šÍu(f÷Í®snRôÑ1 t ¯v¬ºîUKIˆ÷=¯hdN+ú÷̦d$ò*Ñ΢&Í‚o7^s¯ñtÂôãYÐnPˆx¸AÏ'ââ$T³è½ [I4 ‡.«lìî.yûJ.O¹>ˆ#‰- ¶Üi«îY®›=—ƒ€œø†z¡4XøUâ@Üq¯îvTm˜TnÜéˆG–5u2HÊ_ß;ý.éàvP|ø^zPî;Ý7º/w=È%ï ªd|+Ä.g‘I°sÛ}JRÖ¼Íã¯Íá ™¾ì ™e•N”«ÓsµyéY“óé¼a¢—j©á·EhùÛÃ;K5w¬Ônj#_•µ·#¨kåË\>Q>0kgå(Â: §iähMmÔ¹kó‘;– )¦ç:l’Çi‘_,ð"XU «Â¡W?_¿úéº1 endstream endobj 43 0 obj << /Type /Page /Contents 44 0 R /Resources 42 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 42 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F59 17 0 R /F57 16 0 R >> /ProcSet [ /PDF /Text ] >> endobj 47 0 obj << /Length 2123 /Filter /FlateDecode >> stream xÚÕKãDú>¿"Ú Žvb\vùµb`µHpZiûªwƉ;¡§/üv¾g¹œ8MÏ+qpìªïý¨úê«|}÷æóoªE×EZ,îƤqiÒE‘—qšå‹»Íâûèn·4Q³üñî»Ï¿ÉË;Mb›ä‹„ÑN€µ†§ƒç ÏžS…"2çU¹{øi›å*K‹è¡ëù!;™ŒUZgøùe™V͗щ±$…°†I‡ù§á®Eέ ù ~:’"Ÿ®ãwïöÄÅDø>n»ïÍVøuRÆìFD¨ó©¡Iä9«öݳXàÌ,VÆÄu.NGïu QZØh@wöB¾UÒ"üxÍârwƒÐ? ¯VŒ`­å#ã>¡vnàÎdODBŸä5qù±aÖêC+Æ–4Oö‚±KØè'|{3ÞÃ/W65œehš9¿q4‡Ð)­cÿR(Ñ¡ÉÔ•HìÉØ¨AªlËLuGx'u´q¨ò3’·ñh"a=ðû õØS’öޱŸ–” „pòsÃàSÇ $º\8ûCæÞºž)RœÓm8©¾Ã­†5†ûÜkV aÛÀ.x‡‚Z ˆ[® Er/Kt;ç²1(»‹‚0 NYÅ)_šUóž­ óBžÛ4ºSý€yç×NÏôEÉï™xóH²l`zD?ª7ÐÆ´{Ù—pÐ> ö^&vÞ^X`âIžÀˆR¢ÌÆE…,®ÌŸq#¹˜ ÉÒ™Õx…0ÈÕðnÜ]‡I–edÊù}ý„dëù×ÖŠÔ‚¥~….Ir 5¾h÷˜æ™šhÌAýÙ0.ÉO8Ô¨íAæ7l¼ñi+軵ä>.vò¸SE™uGñCÐF£œLs †ÿý¾ÿß%)qq¸CdövÊl[_‡Ð¤{ŠùUí1U\&VG,ÝÜœxKÎ*'u5 'Ú M5áH¹W’4G›f««ªaÈÆó‡;nøã…ýÁ´æ‰ð¬ù]1M/Œ¯9… ´ä:y’G?¡šõ¢fô4~ì5Ê¿žGcöºý„€õ^2Ÿ]§Ugx;—ùOº°¨®ga¡¸ŽŸÍbcËëø ô@¯'¶ý‘¡t¿±a¶ãYR/źŒCüI‹Žô½| Ý=õäã0×R#·¥æ¿›€*¸#R–äé5øÁ•YØðÏ”%R޵ϒÑ{º¨*ÇžàÈã³^øËª*¢{=[ßL·AAl2Þ!Ÿf»Ã¬Š¾=òöÒÛÁØu%’§È‡g¹½D¸ÞƒQ·ˆüDk›døW˜ò•œÛŒ*ûpqË3 À®Ga.XË×q{¹Ê¦Ëk¥ZÜV*ÅV¿íÄWˆvðÈÅñô ‡$·yf@?Fc 8"{ëé˜h¡Ö«ã‰õþï½~³Qí7hº[:jW°']!&%h®—{ˆïH˜ióÊJk޸ݣœ+ÓÚ’ÛÝiºäúŽÁub*Η+"×¾Ã{fH»;^ÿ%ãRQd^}+½Ñºu=F2Ðô¡ãZÆÿ{¨Dë¯Ú´^wçñuÆ[þ>(5ï(ƒ¡ß|ÚuM©ÏL‚º~l\ñŒ–{í*zÏK±²ÑëÛS-hz_ðrŸñÁ¼þ¯ÈìF¥±&. ;•ß5X¶­)Î÷4µ•)鞟‚îb7=Æß_¨µ düUôô*ÝzOþ:ò·í+5f’ÿ·»(ª]Ére[R=˨~­L^šÇ·‘ZVÓ) DÔóuë¡ÎϤÔE÷¤ n ͌ͼ€ á¯úBSõ{VÿAŠßí¨þb.<0Ó\ýE ÌeŸßG)§?)I“:Ube”‡N†kî¯àM•^rº'šÓ$³"¿dq¥ÿ׈÷æ?wo~í'& endstream endobj 46 0 obj << /Type /Page /Contents 47 0 R /Resources 45 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 45 0 obj << /Font << /F8 7 0 R /F57 16 0 R /F59 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 50 0 obj << /Length 3192 /Filter /FlateDecode >> stream xÚÕ]Û¸ñ=¿ÂhªEc(Q””öZ¤‡wEÑ>dÐôAkk×N¼ÒžegoS ¿½óIR¶œ¢½¯$r83œÎ÷O×Ͼy]/š´q¹[\ß.š*­\½pe•æE¹¸^/þ‘\o®LÒ]-óÆ$x] GøÛ_å|Âh–ì`„?ÇÝÃŒ ÞîÃÒmX„c&Ù㟿xÊ$wéÕÒYòúª¶É°gä°o3cq%/¤év½ „ièyx¸­•NwÂ×ãU^+_›™ú&ç—á uHîÃUé’V™zA¤äÚ«^ÿy‘-–ƤM)‚Cat­²×ß!¨M:¸#R¥°i“[DŠðø¡G@Q?0ÌÛ¬Ìà«e0ÒåAF–0´7üÎ\ãȱS­”¸ x‰îæ^Vf@Áðk‹êXó»±MÉ UŽhÛ>¢­j¶Éˆù²LÞáèJ-â7wºäMÎHðÜfÂŽÈVr¶Î-ì¬È²‰µÇQ4#-?p;Ý#¿ßìYûžP¾.`C`$›†8M›éNð¡öÖK‡ü$„+äuÃv-/Yΰâ©u«ûyŽ‹aC½`#éìù[ üùI•(ò˜‘óæ‚Ú©ÍŒá¢VªÍrFé]l !дñÙò”læÈ†GûIàn•Hï™x…(¯få…C„4FО+³ä%¢ë(¢åœŒÈ³»"˜¼Gšþ™?îÙû¢r»I’‚ÆO<Îr(ãÀ{Q!ÐêÈ‘MSër>½J{&œÎïpß ½„0Òeþ\×ì«™ºñe£%5É[SXòf¡Vº1ɯx~çq?ë†{±â<›}%‡iXJ€ã›×ÆDÙ·³i™[@Ì;vë§iºÉS×8D™a"‘ç§Ya³èG¦ì¥ ;5˜T®Qšñ\÷Äß·f,þÝ ióXGÛò‡ƒëˆD%á6Ï&”èS¹ìFÓt\ΙÅQüMÌUDŒL¹‚Ú%ŸF5ŠX+ò€k ^$Àš*¥¬å2Ž7Šñ‰AD«ja&ðì/·­ìVP÷q*Fª‡ v›‘ÏÄ£ñiBŽß$*xB6˜1íu`åêžõ00çÕÖÐY,B}0cÁ!šJíDÏ<) °Vz;%crâ%[=q׌h߉ ÈôÖy¢9ò1Pú•V9òŒ'R/hrââ²ä‡[_ÐUZ¾ =;^ú8qžŒnR²ä'u‘ª*€ £L<ÒלèTÎ7?Ô¬©ÂÅ AL]ÕcÅç{yú$'›S›±ÙhT‡ïù*‹¦BÖ†¼È`äÐè{­Ê˜#ÚñçåüP ÞÑ0ekh$â¼a2bõ–~ÄñâRw+é‹zŸI^CEÏs\âó·r*)Ìw.•Ï9(z™È¯1)IαJ±À™›È >CrZNŽ~·ý á&JH5­.•ËqM z”žˆÒºäï˜ö û€ÀA5]p•¯iáõ,Õ+œ eÌdwZÿ&ï¨à›êç|ýˆc‚çô‡þœê¨~bÍ bœ;~™Oyá£Äe“Ì ­]|â€â ŠÚA2zÒvA=¯yŽür­åyñZWóM/œ a °vðr·`N”#z©Àq±&—u.Jî¬g‹\ eöÇÀ[>¥˜-¦€»nÂâLíç BŒKWß ™P׋䌋ä¼O­S %·O¥ìȾÅRhã‡7ƒV0)Ñ8z³SÇÊÓ>ü! * 2“¼;J•„­™q_Úü;-5q˜t† Ó~Ä…¶^˲‰{;bRPÖpc“\ g”¿Y‡…Hé+fp€³ÈÓÊ N Ù˜ÉËä:VqU²Å ¢D×ÀrÁ-ƒÊ&«‡¡,tÚã€ñ~Ï7¨xiïUÃ]¨*¤:hµ ðÝFZ·j£;µ£Ö=þº#\ê °oâû¶ß)“¹³]Ñ­³Ô!p§uf”ü 9¶y/&áe¼2qÿx#SrTôVôNÀ¹Uæ¸!W¢Lé[~¼†‘—ðû ü°¥ÿJ;ÇK“Ù*ù5C­ýM‰ñg¨e£Âi•˜ÊVEÊR§÷Õ—ÚN²ûÂ6Çú=š¿ÆþáA„ 2y#áQoÎdØgGáŸHQ,‘Ð Ò}#Ä¿ùq”ÂËoù¡qKõ³ŸÌ¾DÝ Äó°½.Ó^@{¥ý¥’9÷¤'Öô`VE¼˜ }Ÿç­ß ïèuÛÎ/Cêëˆæ”ÿ/ÒG™§yVBˆ€J›P*¹Ž;òˆ«‹)r˜BÅeô“€¹3 Ä@~òÿ ,•™ÓŒXÄ•gM™drîñò§†¥Çô\š²âqoàWÁÿ]¢”#º¾ùêd¾¡ùéXA¿3¾ ½¥ð—Ê3#>ôÇüÆôáî„vØ'?‹sùÌ6*ÚÆY”Í¡VÔ ª”´®¥h(^?{uýì?®šØ¶ endstream endobj 49 0 obj << /Type /Page /Contents 50 0 R /Resources 48 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 48 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F46 6 0 R /F7 51 0 R /F10 52 0 R /F57 16 0 R /F59 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 56 0 obj << /Length 917 /Filter /FlateDecode >> stream xÚVYÓ0~ï¯ÈH© Ávœ +q„Ô7à!tÛÝŠm²Û„cùõŒg>'N)­FµçžñÌ—=[-¼È˨NëÂÑjÕeZUTäej²ïâ/²¹–ã#>®ÑˆŸ˜Ñ’a úºEˇi†£÷PëÀßOO¾Å:µõlÜ…‡@ž½ŸÞÛxå‹¿)³Ö=L–ò¶Óû7|Ö“?ߟ0Èš|ÞÆÃ¸ÂòÆkÜw¨ýf™dEQŠ¿(ñ.¨¾eÃ21ªÎÿÀ±xUa”°Íe²+˜”ŒS'*½FÌÝø4â;Œ­;È9y™<›ÇVˆ™ç±OÀ÷ ›MP€BÜ:Àí|Âãc¸ÖÀ哯û’…ß=å{’i›Åïø>C¢t,ñd™ ¡jnùÿ‚Ö5ªÃ¬blý~}D¼ÏÿÒhë¿#íö¯¤ÿ-õUððAÏW M‘T¤ý'—Ú’ê"‹ÖûÅÍ••YVŽ,òf`õUˆ®Ú¤¥6‘Îtjàõ›•O‰ƒ‹ÞÍÎí2±Y÷\ÏAÝZ…•cY#œïKSÅ›ÃdÃÐ@pg ±tOдKSºås ~/6çhëzðQ¨‹¶(¸çßX»ŠwÃróIi÷t²1\/©p `sâm*~çÎ’¶l¥c2¨í§B›^ÀDÛ´¶v>½€yùѺÙV”0]ný™K.å:¬6îŠO¤¤Å÷¦•ÿI¼,Pçxà• 4‹øysѺV”wÞ¥Š/)w>4îǰ6›]À1ñEãq­`ѼÃ]Psœ—Ç4Ia+|WU·ßt²%®{F™´*š×iUaÜô¸ ¿%t@W endstream endobj 55 0 obj << /Type /Page /Contents 56 0 R /Resources 54 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 53 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./compete-veteran1.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 57 0 R /BBox [0 0 432 288] /Resources << /ProcSet [ /PDF /Text ] /Font << /F1 58 0 R/F2 59 0 R>> /ExtGState << >>/ColorSpace << /sRGB 60 0 R >>>> /Length 2244 /Filter /FlateDecode >> stream xœ¥ZMo]5Ý¿_q—É¢Ïø{[H•@‚DbAY (¥-"EâïsÎØ÷£(‚¼É¢I¿ó<¶ÏÏŒ-Ë«E–·ËŸ—oñïåýå³ß}õrùùã%¬!„åüóãÏ.Ÿ}©øüý¯K[çßÇkªÊšeѵµåþýrsÿÛû7·ËýÛË÷öÕö@¼h§o(i‘×Xþáæå›¿~z}ó×ëÛå×Û%†åæ‡å÷Ÿn—š—›Üþ¸Ü¿âWÓòaöçw˜}÷ù7økÍËß—~\ÂòËE–Wø÷ö"6ð×—š×–Ø9ôû%64Ól¾[îþÕ=[iÍ£W¤­íèÞšGÁõèŸÍ½_C[SÝû·æÑŸÓ*íèŸÍ£¿çµå£6÷þ˜Âyü­yôŸ§{Ì~ï&'jX%kœ(k{ŒAEÒó «ñÉÐ: Yž ÕÖ 'hÏO†¦ºæp‚J OƶÎ}=M¶=á5í¼PééãÆÚ×POØ&;vp¢¶µ3¬¥YÖs>í6ÆŒÖÑ+„÷îÙ<õÃz8úGóè×P×ÞöþÙ<õsöþúÿ­kKK1ÝáÄ_€©nõ Ed­²Ã]õÿ‚ÙkÛàáÚÑ5`ÌnpþõZtŠ«Ö ?Ýô)—„ K»nM…Hˆ¢=7áÓæöá;*órþ[lF3M™,yx³|¿|xžtÛ\eÌ?ð§ã“aütÈiéf4'ú޳DçýþPÿ–®kŽtæÖ $ö_îôÁU‡å%’•Rw§¾ ]“í1È¥Žy´„¸Š ùnâ@ÛbÓ«ºcͺ×U=‹f`…ýÚ݆·ºÉÁU膋Z9xÑ9®Ù±h09ÄE#dÔ3v_;”ªÉªÁÎ:cÆëÐðë Ë;æp±H©xÐãšC«6ÑZ\L8vGqtÁóš›ý ßà Y›¶Ã«K”…ñ aŠª#ºÇxx¨rå»oô æ°¸2;$Ñ!QÞÊZ]¶'¦MúZ*ìªó,ÆÊ9Î… žìD¿®ÁäF8/š»ç¢ì€«æ-­¾n‡!³P‰6;™bñÁ#& ã k&œYƒC'%êšfZö 8âÀäÙ÷˜™ËP3|ÆwÊ»$_0C7ïLcƒÏcrà‘¬A™‚;àB?Ä5Év¹á‰y¸ 'ô¯’ÌÈ’ Oxºsú•i®Ge&ÙäéGÚ<î6¯ ®‡3H;0¬â p9û€K®`ðÙðÌd‰ðèºRÖHw«®´gƒ‹úD¾˜JstOÎDRƒ°Ò“&K¥œÍ.«Ååáò×Ãá)R,…pE5È_„”>ã{d%ƒEðP~±í.±è™ÃQW$T7&«8GÇ`Ã{4ÇñUηqñ„랥ð, 0—ADl5—Î;åëáÝn™¸ÄUÔ÷žy?,E\2Y§b8ìÛq>GÞÀŒj—Gm⨑JÛò•ðΫlV|<£¨b/'À]\DB¹pÎðÊrcf€ä€«µÁí›Ãß#Ž8»ÅŠ.YÉVfÉc|nŒˆyé)ŽKLÏÕ\¤q6rß*Và ½¿°DÔl0ÄÏï.uÔù² Þ» ³ª5‘—¢Ùù)ëíh2è®ã«šv›’F¹­±ÐÞD»3ÂG›†3kãe®:ciTÕÐÈ×T{Ååû~Áú°íjïXÞ²÷ ”wN­v*ÛÛykŸ‡½eÜÊ íããÎþiã«ã,¶=Z?KôÅä$l#¹œ—Íúa/æ§ÁîÐfÉ£YX׸< {%›|$û<ìÅüØÎöyØ«# ìöù6ŸÈÈl³€íâ>Ûk ØËùq´7Â^ίÖtц½œ‹öyØ;çW¹‘±×(‘dî—Ѽر¸~‘*;NÉhŸ¯SAʰ7¶0JjJUd»ýkóó½ýKc}ìÕY@ïG¶·‹¾°&W ““04ž8Ù2ë;'yx׃“,3¥ƒ“5ò…ÁÎIVÀ≓նâà$Ë‚íÄIpÈæ¸q²6æF'ÛäÈÆÉ6‡Û89®ÑN¶ó““¼^Î'NbφÏLN‚S¶§““Æ™~p’Nõऊ0&Ø9ÉD†ïûvNršùÄIä¹åÌIœ-Æ©“²?9)}¬×ÆIà58Éï¯'U23à“´§œ8ɸo´'­0zâ$۶ד“œ_—ƒ“l›ý““lç~pÒúåà$Ûæ“““lÛx““Ïö÷nã³Ù÷bø„{|ÖîáxIáà¯!ËÁ=L©õ9‹œ¸—òð‡{ˆNJ:qOŒÊ‰{ˆêY±Dõ̽\†q÷¨¯éĽb/—îÕÀ²ÁÁ=„võÄ=rßüyÓC¾æè'=lqèǦ‡ðS ÷è ¦Ç“{¦·áàÛ)Üc{Œ7¸G|‹÷d¼`ݹÇñùfp×Ã&ãû6=„ýCϦRÂIË<Ÿ6=,ªzˆõÔ÷¸þæK›&»¿é!ß8ד2æI'=”2|yÓC^ îÉ^ÈýëÝäöbïx8ùíåì7„• endstream endobj 57 0 obj << /CreationDate (D:20150701163334) /ModDate (D:20150701163334) /Title (R Graphics Output) /Producer (R 3.3.0) /Creator (R) >> endobj 58 0 obj << /Type /Font /Subtype /Type1 /Name /F1 /BaseFont /ZapfDingbats >> endobj 59 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 61 0 R >> endobj 60 0 obj [/ICCBased 62 0 R] endobj 61 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 62 0 obj << /Alternate /DeviceRGB /N 3 /Length 2596 /Filter /FlateDecode >> stream xœ–wTSهϽ7½P’Š”ÐkhRH ½H‘.*1 JÀ"6DTpDQ‘¦2(à€£C‘±"Š…Q±ëDÔqp–Id­ß¼yïÍ›ß÷~kŸ½ÏÝgï}ÖºüƒÂLX € ¡Xáçň‹g` ðlàp³³BøF™|ØŒl™ø½º ùû*Ó?ŒÁÿŸ”¹Y"1P˜ŒçòøÙ\É8=Wœ%·Oɘ¶4MÎ0JÎ"Y‚2V“sò,[|ö™e9ó2„<ËsÎâeðäÜ'ã9¾Œ‘`çø¹2¾&cƒtI†@Æoä±|N6(’Ü.æsSdl-c’(2‚-ãyàHÉ_ðÒ/XÌÏËÅÎÌZ.$§ˆ&\S†“‹áÏÏMç‹ÅÌ07#â1Ø™YárfÏüYym²";Ø8980m-m¾(Ô]ü›’÷v–^„îDøÃöW~™ °¦eµÙú‡mi]ëP»ý‡Í`/в¾u}qº|^RÄâ,g+«ÜÜ\KŸk)/èïúŸC_|ÏR¾Ýïåaxó“8’t1C^7nfz¦DÄÈÎâpù 柇øþuü$¾ˆ/”ED˦L L–µ[Ȉ™B†@øŸšøÃþ¤Ù¹–‰ÚøЖX¥!@~(* {d+Ðï} ÆGù͋љ˜ûÏ‚þ}W¸LþÈ$ŽcGD2¸QÎìšüZ4 E@ê@èÀ¶À¸àA(ˆq`1à‚D €µ ”‚­`'¨u 4ƒ6ptcà48.Ë`ÜR0ž€)ð Ì@„…ÈR‡t CȲ…XäCP”%CBH@ë R¨ª†ê¡fè[è(tº C· Qhúz#0 ¦ÁZ°l³`O8Ž„ÁÉð28.‚·À•p|î„O×àX ?§€:¢‹0ÂFB‘x$ !«¤i@Ú¤¹ŠH‘§È[EE1PL” Ê…⢖¡V¡6£ªQP¨>ÔUÔ(j õMFk¢ÍÑÎèt,:‹.FW ›Ðè³èô8úƒ¡cŒ1ŽL&³³³ÓŽ9…ÆŒa¦±X¬:ÖëŠ År°bl1¶ {{{;Ž}ƒ#âtp¶8_\¡8áú"ãEy‹.,ÖXœ¾øøÅ%œ%Gщ1‰-‰ï9¡œÎôÒ€¥µK§¸lî.îžoo’ïÊ/çO$¹&•'=JvMÞž<™âžR‘òTÀT ž§ú§Ö¥¾N MÛŸö)=&½=—‘˜qTH¦ û2µ3ó2‡³Ì³Š³¤Ëœ—í\6% 5eCÙ‹²»Å4ÙÏÔ€ÄD²^2šã–S“ó&7:÷Hžrž0o`¹ÙòMË'ò}ó¿^ZÁ]Ñ[ [°¶`t¥çÊúUЪ¥«zWë¯.Z=¾Æo͵„µik(´.,/|¹.f]O‘VÑš¢±õ~ë[‹ŠEÅ76¸l¨ÛˆÚ(Ø8¸iMKx%K­K+Jßoæn¾ø•ÍW•_}Ú’´e°Ì¡lÏVÌVáÖëÛÜ·(W.Ï/Û²½scGÉŽ—;—ì¼PaWQ·‹°K²KZ\Ù]ePµµê}uJõHWM{­fí¦Ú×»y»¯ìñØÓV§UWZ÷n¯`ïÍz¿úΣ†Š}˜}9û6F7öÍúº¹I£©´éÃ~á~éˆ}ÍŽÍÍ-š-e­p«¤uò`ÂÁËßxÓÝÆl«o§·—‡$‡›øíõÃA‡{°Ž´}gø]mµ£¤ê\Þ9Õ•Ò%íŽë>x´·Ç¥§ã{Ëï÷Ó=Vs\åx٠‰¢ŸN柜>•uêééäÓc½Kz=s­/¼oðlÐÙóç|Ïé÷ì?yÞõü± ÎŽ^d]ìºäp©sÀ~ ãû:;‡‡º/;]îž7|âŠû•ÓW½¯ž»píÒÈü‘áëQ×oÞH¸!½É»ùèVú­ç·snÏÜYs}·äžÒ½Šûš÷~4ý±]ê =>ê=:ð`Áƒ;cܱ'?eÿô~¼è!ùaÅ„ÎDó#ÛGÇ&}'/?^øxüIÖ“™§Å?+ÿ\ûÌäÙw¿xü20;5þ\ôüÓ¯›_¨¿ØÿÒîeïtØôýW¯f^—¼Qsà-ëmÿ»˜w3¹ï±ï+?˜~èùôñî§ŒOŸ~÷„óû endstream endobj 54 0 obj << /Font << /F57 16 0 R /F59 17 0 R /F8 7 0 R >> /XObject << /Im2 53 0 R >> /ProcSet [ /PDF /Text ] >> endobj 65 0 obj << /Length 2984 /Filter /FlateDecode >> stream xÚÝZIsÛȾϯPåÙ@wcqâI9S™ª¤R¹DU9Œç‘ Ä˜/‡üö¼µ»A‚’'•Ä•(4z}{ïA¿»ýæÛê«&mÊ¢¼ºÝ\5UZ•õU骴0îêv}õcÒN×76/“éá:O:lWÉ»,·ðváÏÄ]‡æ~ía‚¿[øíÂ||¹Ùî÷¸~ÃuQ%qò#®Ãe®ùÄ;h~Ĺï2—Áø)<2cª ºržõ†‰38T`»N6¸7n´æ)JÏc7†Ž>’Þ‡ o2xèÐ Õžsœß]ÿtûÇ«ìê&ÏÓÆ‰lhd+üÀq&ÏYí$#‚À]_áh™Ü‰˜&ž|‡2øÄí"˰QÀF@ذOiÞ†.bAÚpzÛ£qu „›2¹•ጕ—8Šê3üö|.ÊëˆKWJðÐóÈz¸¾)XrN„ö=+gà™ÇU8¬åig=¶»%ñ‰M(›ÊCD"qÓý|]ÔtöAƇ^¬ VVt <î힤ý¨"LØ!O$P ô€F‰¥‰_Pý8íIKݸ׿ëa(õ½&Ú(l‘¼Ë~œšò¯®ó¡"œ­’¿F§ëôTæRÛÄ@Q%BÛD"Y]KÊ"‡ÛER1Æ ß µ]j?e‘£ý–6ÎmÚX;ßýúö‡<‚L0Î!Aw4) eš—y˜Ž{¾IëÊéœé|“º¦Šö@O¯¬!•«ÇÀs|ds§.5ÈwÄzx>ï0j¬Ùkh‚±Eë'©ò0Û/¿°]r{zò°`Y£#´t­ÆG!á|/±^ƒu¤JÔÊÑJɲ À9°"E¥K¶=?•}0Tï@‡Þ˘nh£Ö&µÍ}€ÀuûYðFg>î˜lZɾü¸ Ï¢¨Á‚¾Ža¸,ƒ+`¦±yЉ†xRÀ š ÝŽr¿´ý*¨pØHo׋eMZº|îÜw`keÒuä]†ûƒcé‚Ô»£©Ä4úf º¬¾†ºŸÙãã’š+çåú悚s¯™va‹Ü¤Ecž1•¬.þ¦Rü"SY‹}F,ÍE,Þî>/ìQ€y^-€Qæ³[ltܳͮ¶š±3à:~§è¹÷ˆ¥_³Ã50ñ@%Ü ®äx¼õÈs¸×³Ô fî³ÀG•Ö&'PP?§ævPÁÈú]Z6Þ ºäVAVÍ%ï V­\ŽØA1Çde€й¼A8Þ[(C|ëb?á*x>gðužæOÅF—Ffv“Wˆ”9DâF~Yá ÁÈ‘L °9`³¦»¶}/×81ö¸=Éj”WÁòÊÖÂX’íy6@Æhb„œål#ìýƒ5˜2ƒ¥w]†q·ãE£*gÏ)uî¤3 Yè<Ô»ö½˜÷`O‘¡Dx‹u‡É uv,;ãR&à ÎÉܵý½&9®IîŽako±­iáb6Àþs­KZRL«7RíóÀ·tÉ)8S0žnAǃ*?<ˆòmólÝ,ÄÀ+X©0qΉöem€8i-±ÇËÚâ"7 ¯“TÄz€†d`ïÏ×¥äô"ß W²i˵'”FîãÁ©ßöŒô‡;¦Síqän¯h“QD<°. ÓP”¢·âQ~ÉÈþ ¡+8(o…ÀÔïÀ C§o„‚‰³sìþ 4o¸Éñ8§À^´!:ø‹$7’•ïöK(~Åy0$Ï#ªZÄÿàÇ$[Nüú’ÊíVè9ÌFßËÆq»AGQþ5ܤ¶:Iž½Ö¾àÇNλ¿Ä`LFµåÉ?Dëˆy|¾±uŠ…†Vè÷Iho<7¨Máæ6®:Àµù¸åԙ㨸v@ªØV€)æ‹ptMS!ßD¥·®Á‘'Ïîþ¸úfWÍò€VÓ‚aÏ1yŠãé2'ì§Jî‘‘ÈÚã†u rà¦7§€=æ„|q¸"=äÝAÝ ~f©RBn1âûFF®³À¶Øá˜Õðã~LjYBöé]æD…‚ôg}ÔœÕÄ2+ΙßñœôUTRàžyÍ×jˆÄÑÓ²Å'ÙF£¥`1>óRÅÍC–†Ñ :äÊO ú`‚¯{1è‘Ìß+,ò‰{„ÐÍ‹·/—¬4J¶“ÖÃÚªï!“¦)|<2}ìxGWÐèZU)N$ÝaÝ(ù›Ê0ëŽ5²·Æq‰®‰œ…/åXbŽO† ˆ¶sO³Õ—àã™æY ¨Ð-Ž]çÞ)EÈ’~ͨ×k±âB \ÑFƒ•¨QSDoÛó-£Á–f°ýЮ¹±¢RΨQŸ*”Þim„j­ç™¾ ÕkªLÇÌõq^à ™9L¸C+LTŒ‰8gÎ'©öÉ×þ*5῾[Džf¥ŸPd§ÄµMm¶c ÑLƒ9Ýkßüþö› V¼þ endstream endobj 64 0 obj << /Type /Page /Contents 65 0 R /Resources 63 0 R /MediaBox [0 0 612 792] /Parent 41 0 R >> endobj 63 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F59 17 0 R /F57 16 0 R /F55 15 0 R >> /ProcSet [ /PDF /Text ] >> endobj 68 0 obj << /Length 3185 /Filter /FlateDecode >> stream xÚåÛŽã¶õ}¾Â}ªkERÔ¥íØ dû’´;@’Ðø6Þµ-¯dïÌ h¿½çFŠ´å½$Ù @fD‘G‡ç~ýòöêÙ7Õ¤NëB“ÛåD)–JO [¦ÚØÉí|òCòcf³k•ÜÞÿÅõÔäuÒ7[Íar=»ÖUrß`ÜnqüÄ‹‹G\…¿—zžÄ÷_ÃðÅkï}·€«˜±&¹ 0á÷÷²åõ.“ÿ=]Wy’^OóÚ$/xq‡A–Ì„{m‹¤épçæ0pq×ôø2ç·vwýÓí_Ÿ}cË@ ¦ÈÒÚ¨IÆ¢8?[BBà¡øtžšÂ:à„ÜÐ@N¦y–¥:ŸL•Jk+²½»žêéÑu ‚Vù#aVÉ…™äŲÞ0ÑÔðcÒBþV]ƒÔlI¬+^D¶Â),-v(É®L‰S“,|„ ÐÈ'GTÞ¢®²T©üSÅcÒ*ׄ2$:B ÍNXIô¬€!Xìu’|A\ Ø"°'i1E©¢<ö€¥$ÉìÐz½tLY%-(ÄæÉ\‘Ñ‚yÙ*ë_0mK’ïxb-¦]ÖLs¿ãò[ÜaoOü˜3,Àpß8Á ž™gI³Z ç&aFqÍ(%­I¡rÂ]Ö‚vKʧ4Hïß·üâ‹û²;´-]ø iç×Ù‘â7`ÅÁÆSþžBÃÌIg‹±@B©Ð:á‘'óNâÊ Ã(„îWK^ÞЉ æœp$´“›taݰ9 Ww¼à?Ú2„«1dé^6‹C'ºD0òÉœÞH6  Çá eD>²[9zVs¹£19 ê±LíŠBV瘶–ù†™¯É™yn˱ÙÁ6T4UÄÛqÁ .\çž%˜¦láCnqt™È ø±3 ïÉw)4žÝÄ©m-Rd¢„4T{f“}ä¸á £ê‚ÈA$íâ–@Å1¥â@Šo}êÊTYã’Èô *€ž 6á!7ê„H(M×η®ó´¨ÍPâ0D6) k™ŠŠÄÔÖ˜æõeúLjT9 `nØŽpo¶@¢¥õÖØ1‘$9”ù*FèTiVx:ßD@Ta­¦GP¨,Í*âï#¬–©-ò±Z¦™‰XmDÞ”¯ú·ç¿ç´pª å…ä‹Ö1ª•M ã ˆæœ (0ªÂ­K͇Bè·h)+H \Jâù'/c£ó (°4kºÈ# `ê=.0iâÌ)†bÊÇ,ã"á{Ö¤¹­)““¨`ÑMNR0i¡ wpS¹À™®*½4~Ì23Š+«ô‰Dk˜­+@EëFC‡£º ìàê/·Wï®ð¸”MÔûl ŸäpJ¨J3™m¯~ø)›ÌabjÀit;3RáfòúêoW/ñ`ª¢£ƒ *ë4ƒiËïG¶ ÏYš*öùlQ×§¾"²p”«O\!”@‘f™f ‘'DÊ,àrp.©ìLJ^ó[(.&û͘¿(@’§%Ä=‚y>š.Ó<·°u Ä)£,/±F'бP¬”ù4öËHkeŒªÀæ«‘ Ø©8É áN Î¬þìÜ©"v$“å©ÎÊOq¤R¥×EGò¨Ê*-*ók8RùKÝH}’Ù/ìF_ÂÔÔ˜§ÅP‚=0?ÓÔFE8niEli±çC~ºK ›–J²Í?ðЋ™{K É‚®KäRwÍG†>LªãBçÆèD„­ *œ¡"G‘‡Xs§ÇÕ½kH×ßs~yï²nçÈhwÄ%­Òyž²G)D® èr÷. Áõ@–'ÛÉéË“‘ÚÞ×ÚJUÎêsê%urµåwwb’axê–“n°“ׯlÁ‹Ç}`ÏQ<OÏç™R±~n‘uZbEò™¯(¨˜5LÓfà¥=ÊqÇK–Þ IMgVêø}!<Ä›"ù–ÎEN˜ƒ`õ¨,/aÞK¿Pk(ÜÍI‘Gç=9†phÉO¶¨ÓÚO8ƒ¤zÖè°~dA]^¥ºrRݵ»©Ó7›ÕØ9^nXùbýf(ïþM,Æë¬HËŒh{“r=­k‹=:þIå« ­—:n…`›~íH;ÒExFפMpºõGùµ³Nܤ=ö¾¥U=r5(޾¤1™ÂFNÂ0éNº§m÷Ä3ÔX À #åÃ^'ݘ'ùtD%OÌîÀÑK¿@±hëÏÌ(›ÃÖž‚€sB¦%8?içÏšýIGÊdEÔbÉØ4©«0ضv:¨ø:Ž­Iôìømv€d£–7Ù:s „Q«.TÀDçXíæĬV¨Z¶Á¦ÜN °ú[O>™'¯ä³=û,"t±iq1hŠik+­ôЙ7<îˆO×!ê[ê†Ôâr€Éõ‚ݘÄÎ]z²‡§?b»p8‡¶GH1]yÃÄ/›Í@êsÍ!j3wÚºA·2z ñ;â)Y>Hs«ÚµCaÛyŽÝf¿–¤¹yº(ÂFÜûÁõë °‹ÌÏzøoq[ÇÅ„cwìF;þ>´°íýŽˆs‡ï$Ÿjrd„Ãg7œ– 4$š¸Q‹ /Äw8o„!¼í£–~€Ì›„ 7u÷-Ž–ûÅH×…JµJ°áŽK%HÒ¡»¶Üˆß¡îÐUà,¨p°Ò:¼é`v}ûÚj1p34ÏñFîç\oRƒ®OBôKr‘A—CÿÙdXð6Q5Ûýp»†³gE›K4Èž;P”/ã;âŠð@NóÜ}Þ§Ý­|LßRóy³ñ.„<îy4½>w‘ߨ¸Ö4ì#Tļûûr, Füò=Õt´ Ãé¥r`ZÑ‚¾‰V¥Â°GïÔ–WÍ0œKp6;ÉT*pQo]i˵°R’ é²ñ)¨†œ¢ê(H³Aµè.?üÕ—rRkø…“ï…7h.å*§íÑHMDI¥¨ð*üe€åxßR„l°`ÍåRÉßcØ c÷<]AáЀ—¯ZlðCkÞ?x!ns1>ø Ê^JÔÜw$– ŸHa"ìYûËØ 9|ð@P&Kœ+åJ¡ˆó×fÏY%Ï2i¯_®ukN&¸&Ãñ¶yôóR‰mygª4üÏ¿ïøCGXç­_(ö÷+qqãWÂúeë¯à¼!Veì¾ÿḏ*†Jú¢ ±XÒ¥. jöñÖ¹Äs® c>{ ºªF$ïÜÅ Lž^5HDåÈÖ§’aê“û xÓÂèŸfÍ]æ¥àGЧÿDi‡†œ%' Ñ„Øòkw±Cfâ§ãßhi!¢^–W¿Šå=—?í½ý~ld÷Õ€þNлÎðÍWühñ,àK䵊)Ü öYpù~zi{AnúÿFnY¬¡£–EU¥åéýÆ<`£òC¡_lvx)üïä_”Ëœõ8Hï0 SùeG1ÿ/ÚSgiUéxãµSÝ>žIäü᡾€?+{hp4 ´áºT#â|~ò~sn¸šÀhzÁTˆÉÝüS>Áe~.ºÊ4’¢8Ì¿#ˆ1«¸èRç¿ûá_›Ô®køµéº¸$í Ž¸”¥N¼Îu¦H¤ñk¸«àóþíœö’uŠŠØiCÁÄ®D‡苹lö¿ä­!ÇõÙx“Pþ9ÎûK~À‹í¹Ÿüä2Ói…í|Ã×DîÀ#WLÿ¿˜. endstream endobj 67 0 obj << /Type /Page /Contents 68 0 R /Resources 66 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 66 0 obj << /Font << /F8 7 0 R /F57 16 0 R /F46 6 0 R /F11 9 0 R /F10 52 0 R /F14 69 0 R /F13 70 0 R /F9 71 0 R /F1 72 0 R /F7 51 0 R /F55 15 0 R /F59 17 0 R >> /ProcSet [ /PDF /Text ] >> endobj 76 0 obj << /Length 1977 /Filter /FlateDecode >> stream xÚµX[¯ÛD~﯈ïú^ÑJP@U‰ Ú>øÄÎ…æÖ89§õdzsÛ];>VôÁ‰½;;÷ùfìïgO¾ù)+&UTå:ŸÌUæQ¦²Iž$Q¢³É¬™¼ æSìÍÕšk1 3àí{sÌõ:Îbó×§Âe5 UœAÇËW)u®ªàïi˜äyaø¾™ý2 Ë2*Òb*Uk³áÓKÇðÖ\ksmøŸÆU(ó™ È2–ù5?Ǭžfšò MÅ4©G2&Ÿ€(_\EËßÒŸöUŽÆ}Ûj³Ñ±‡ÀêŒ[TšXöÙkš0MÎÚ^³h’5´”i*Z*O¼¤³åÄúsŒjY+ññ—B•¤h•_úµò\ÕoY£ HJ~ÁË«>Ú-˜¿$P7æÎçÎÀ¯®£ð—Ìf¤Ì(ãFUUWOÚÉæ ysç5iüй¾qít8^åÌ»Fùxtb׈û~S­éaí#Ø6Lä–BÑñ? °÷0ÕeÐÖeD½#ŒLÒ¯gÐ`¨t9’Bp7Íò ðÌ8‚_,%'Ÿ£F?×nàÄ­„n€M=‡4YÑÂY’èI!ª;t#f ùH%HŠk3BÜNÿ Ä<7fTQ¢¸¼Ö*¾„x¸SÆbkéÊäËßs¶$Eû”qOA¡Wðã‘ô_;õ>åÍ¢iiß‚ߟzyˆ>AÙŒ˜’Æ`Z®ªO5í"îVn>ŸEp™Š`iB°„÷ç-üÞ"ÈcNÁKô5€†2‘ŠF¤ÆSºÌýl8s¡W@иkzé1‹™m»@0в«·¢¯­¯²ß¥á™šmŽï`-sï¤æô=üìI:åíX­¡Û/U`áAŠeG7â*K…-ŒÒ“Ë‹óP[»[z±ú\9…sG‹ýÊ"?`Àþ80… ¿O´ô¾ %æ âÆ•Ñ‚ÏúeIíqöΨ„×~¨í†Ú B\ôfÎ#]¤R·º€ÙW:û1cÓÒ† >ðü”žuð8:õàÜ€’%á ,âŸRžßM0"•Tâù÷—‘)ÍKf"û”±s„=L—vHD3H#nl5ÙΊbº´øá@r–1ž¸í¥$3ÌüxÆWÐ0Ë‚¿à 'ß«2Êìw+ùŠé¡+;-toá¹´ëŽ;> jÐ,y5ˆÓdšêàç…ÛëÍ:IìÃì_ f £2·72)p'ïètü&"rê›f#ÇH(yáÈíîjg¦ù7r)”šþdî¸ ÷ ¿9Ïý/‚»ÎoÀ«ÑºE7Ê'œµ´¶#ÑN,FÓoešê ¾L,£ ²:ÖQYÀçš**Ë’ÌT)ì=ùqöäaš endstream endobj 75 0 obj << /Type /Page /Contents 76 0 R /Resources 74 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 74 0 obj << /Font << /F57 16 0 R /F59 17 0 R /F8 7 0 R /F33 8 0 R /F11 9 0 R >> /ProcSet [ /PDF /Text ] >> endobj 79 0 obj << /Length 2863 /Filter /FlateDecode >> stream xÚíZK#·¾ï¯˜›[ðª·Ù²‰¬×^Á9x}è‘Zaõ²ºµ»cÃÿ=õ$ÙRk¼É!Ø9̈"Y,«¾*ö·÷/^½­ïš´±¹½»_Ý“§Îäw¶ri^Tw÷Ë»Ÿ’ûÇ™Iºþ›Í‹Ü$­/çÉa€²@C–lö³¼N?¤‡XÛÀß~Íœ{î¡Ê ʇ—i².”qЮƒs¬á|íÇ®XÞ1á¢=bÇoeЀL<Íê"Igó²h’qy™'ü„Cv8p+­8LXBâ¯àï[\eƒäû¯¨÷çûîæ¦L›²„_“6•È×=¬iå¹çôýlž×ŽeÁb ú{PvÉç^žeD‡ýuò«Ÿá²Ûø«´#³§Ã*69ô4’;hìQvß.HÒð·æ^•).½8Ì`Õ³Êò!nÚA×uI+üuí©¥°NbÀ¶%çCõ9éG^·åŸH2$¾l,¸[ç #+üu‰ÎÛs3Sÿ÷=ËÈf5ëèfLÔíHδac']^aevœê]f*ø9ÓÚXh÷¬OÜx†TÙHÑev> òʬgNG†‹uäñê­1Ñ,¬M]æ@J$OL_Ñ*µM¡ý²û·³ºä;¤Ën‚º.O/´uÑö²Ç¢H>òáB9'î×ûÃ)ôÆÒ‚ê¤6àÈX°ÞžÖg½¼1Ü’Ë‹V·t´ƒÌpØOr°¦Ûƒs½¢÷)Ë©j"A¡¢Õ©«ïù/³y•dmÎ2n™¦ƒ´½ËªLÌÃK1„âÉ+hŸüv2‰|”&’¤p>)Ìmàç7:—ܦ&ËÇç‚L<ˆ62Ñžçý3Á{ ´" ’E³´®/=]-@Ç “½Ÿ˜pâ ɉ&Á[ &,ìøU˜D{¾–ûÓyceüIG“Éó¸JZ=Ú‰e&~ŸâL%µÎn‹ ²×£ü!ÙÓ¤|ÆáOŸŸßÅ÷fÌèï¢Ã.Òáy•æ¹!ú²)™îËÑà‰#ø¿ÿg\ýOjðÔý{«l¬»±›.‹ŠÝ´w£ÔX’áÇÎÈðcuEàf`íkµî Köè(‰³tÔYŸ_Ó‡E¡É•5ÇË̇Ó2@arc8îA°çK¨;†²"5‘'$¿ˆ…‚rœyò©Ã§ƒðF§E°°YŠ sç’•¿Nþ.í¹g#g Ñ<“ö–«´1ñy+PÕ1ôZqq¯ µAñhí:ëW%ÌÚéÒÂ)P¢G¨P^0f³¤·GA“ØÞnE·_2®|8.¢OfBŒ>d1¹²_>ndì´gràLŽ@yLj}ÉKM¹ìð|Ih-ÅVè;ìöaÌܳ±èwCø”!ØRêÂâB½T€O¢$€¯À]7ôª¸ÌùÄ@ Ð*]ç ë¾gû|©+þŠ”ºÅ8r½_“ >Ø+TVãxÏ{‚ f£x¯] ‚S˜.ØîÀÅ‹ ¦ °òQÊË–M‘ë¥Ãr¯ØÓ‡s@­!¬õ’[˜}é%Dek·"Ò?Ùof U†è±¥ý²°å-\ùßô ®7ö—GÙ%/Q}!çÐø¢äXÄ(ÜÆb¼W›`+PuÔamP BLhaÁ½+Š$­8U½­Ðqi¡)ê¥]^G´Æié¬F¬o„¤ŒI²4·•R¼Ë²Éy²ÔÔJóݵê´Ê£9TJãè:­]©4ß\ÏW„(ý9š Ú2ö,S–YZ9s±»qJ ¶ï“ðwb{MjÃÚû3yj¬@æó-€}”Ì€_r±”å X¶bÁ×ÕUjšúÂ{jÐz¹T½]ŒGS'%â6‚Éîæ+wDZ«‚j¯©¯o"¦–šáYÙ …î¶Ì@Ïu6Ï•6ÜÖs™õtØóYÀ“BåÀYΨU<^Ì)œo¬Ç=”¦ò9N,“·Zïe-\„ܦO =0‚²$KXèãÖ¢‰½Ò)d!ñ‚s)p„‘ Ú`çMÔØñÐöA2D©{‰Ž“¶F<쉽SXŽ.Ïɧ¼¦ n'iÍÓckŒE@ [gÏ* ÉôÁ5ë§qaG|<8ó¢Ýërta¸òA ϰµíùwqKa±ó¶âéò¼Lˆƒr¥>c‰xyøåLX…>R/¹¯GÔ£XÆr·FðLµËH„i7~”ÖÄf®•uG˜ÕÓ¥KÒ#²SæD1Å©D^¼+ŸiF–Æ\Æ’˜5J€Ÿнðxú £ƒRE¡EÕQ¨öÈ̯¡NwíšÌ{hJâziRÞ©O¿çÌz‘¼! ÿIˆYÒu[p¡eÐþŽdÛó(Šß®ù¢H«ÝO97gÓ2ÏŸI£‹ôî³ýEÐN×oíX#å`ng|Mò§1÷9töÄ%ƒ+{¥VaÇus…ÁIÎ`=ô XûÚ!à|=¨ËˆA…ÅP=úC`”CÊ®—œŽG~Sï;¯·šÝ¡Ô3(3¸R›<òuxÂü;Á…ÀîÅBŠTÛ‘ëj·LAWðØ.$$…öÒnä)¥*s“è½ó…öþQ3ãÛ%·xs†3ð=¥ºÌb`·Ð®ôw·»(• m^ÄþbX:±=ºW46Ñ:¿ˆ„6>U€9ó»“]ضÏ17L†K]¸û©yJá~=¥ÙñýZN)¶©?'P]\ŽØhñå@{d?ÃfÙ°VÊàþŸ!>„à‡óë÷ßZ”¤3…`íQà_{ÒË=§ؽ ”šÖIJ~B^þA¨œ9 A¨A:jˆ½ó)Ê/cÍök¨{Žho5øCøâ ÊÒº±×ßyì'ÎüXY{ Ï["ž•™KÏÅÆ=£ujšòyllñŒ ÆÎ¼m]0öÌ# VDI£Ç®>_žŠÑ2p¥ÀwDèóyòÃá[ôeΤoD¥fÎÑ×8…kh‘î£4-õ{”ZQš&(ûx…#,j£àµçéQc'lñ·„Ž«×rí¿ÖøÁ…ïp6ÂáŽ'òˆÖÅŸê\Ä$2ý4‰B~Ì[žº€#îgsS4Éßgó:ù”³&ùþzÒ¢J‹mœx_„šŸÞð—„N†š’‚æ1R·u²fOBÍŒúޱÃÅfFÅIüœ9á.³4TÑsøÀHÓ(°û—­¸c y0Íœ™0ÁÙiêú0H8„ —§®kÞ²©°ïÅ÷÷/þ#Xyd endstream endobj 78 0 obj << /Type /Page /Contents 79 0 R /Resources 77 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 77 0 obj << /Font << /F8 7 0 R /F11 9 0 R /F59 17 0 R /F57 16 0 R /F14 69 0 R /F7 51 0 R >> /ProcSet [ /PDF /Text ] >> endobj 82 0 obj << /Length 1277 /Filter /FlateDecode >> stream xÚ…VMsÛF ½ûWèVj&bÄå·o²7uǬ™Òh’²ØE IÅÕèÏÀRtb׉û`‡·À^­.>Þ$“ÔM#MVëI»q”L¢0vNVÅä«SM=§›Î‚yêv4Îéo[ÊJOãF†ûãýtf"§ÄxK¿£ìaZæØßuôßc~ _Þc?a35Yà ,d4÷æ/Ž€ÉFutûŸ¹À$ýÜé,Š}ç^ü#šä“ëzjb>u‡r{ŒÊd;lÅvÞ ÒœÃØãoÚé·Õíd>™yž›†#€XY¬yN‹¿³JõŸ¦3ß÷ò?Zï­ƒ,k„v¸§0ÍÚ¦i9@:þãMM¼ÀõƒÈ gäoÜ „7Ër }F|—Ë)¬5Î3éÏMæ‰è|%)ïئÎbWXýŽP˜ùaàÜ.iöãйÎ$(žü55‰sýA¤2ÒäÕ?`©z´dàýåï’#ˆ,˜IÛ#»G8u²Ü¬E´CÜíwEðû4ŒHXDqäQƽҨnZYà¨aPh(l+õÌ[±}½µ‰h°å´z¡$æen¿@ŽéÃÔ{:J^¼KšÅÞü{)¢÷㥉Ïù!ÖPzN 6 ñ9—&ÎÕ/!PÊBP ßã;÷•rª,¾®l®”)gÎm-§+æo·ÑH÷"£éèà3[2IĘön¢¥ì´C–ãCjR5Xd}Î>oäÎêW¬iÉü*èJz:…Ѷ„#”ïàW¡ˆ‡—&Nã“I<3$!^KBxN‚¯I 6`CÛ‰g{%²ÌÑ`p'ŸÏWåY\¢ÛªL*…U˶6©¶nÒ2Áç9ksÎXµ—•­t{›C-'¼/x2”tÜ€'¶¤TfêšF´`tÙ·\l½s¥÷°´¶è¢ëû˜ï‰‚ÂýäGæqüæÁs/¢ªÁP} â>?›6""øÓ`)Ÿ{©gÂ7!Q.$âí7tÿ’˜þ¥ô`¥_TϬÚjeiµAxñ˜¼,xîSDàǸƞvŽBTøâVÖ n¯_‹Äß*¦LLÜwÌ|nÞÁ8TŒ¿üF“+¸^ dœ7“P-±W¾·u$Žœ>17ªf7C½`~fµn¥R²poKl/ó5j/YÇ-c`K•¢f‰g ÅFæœA.5 W•B–‹Ux}Å~v…¯œö”bø§7inÂù™ªÉ;0F ã²,¸± %Ùç>‡ï†ÅåÍ ‹WŠ3 mËl¤üù‹¶ªb¥†ˆí÷[¡]ÆÂMRL¬–¯3¬¶"B µføùe{Ÿ$:ï‡*¬+ú&ƒbÎOŠÖöê![»§7ù9z˜€2—šÃÐiàøæ8pcÝnú¥\¶×þ )y–?;YÈnµ“o†Lü‹ )Åo|ž>QAdžQHÛ¦€:x'ž¿`"ÉR³îßÚâSŠÔ¹;›ç\¾ö*ÌäææM=n•žt ºÛ>?…ˆ”×¶ƒÀ6¯¬°ûZŸÅæ²RÜ´…¾#£w!ôÒSh¢wž%#ÊÇJù±‚®p°Õ³ËÔË?kNdۚø"BóQçY÷KßÞolèu–7A÷§4©Æ,ª|zî¢[·2ý´×3lxs- A©Zê57T6“8˜àaâ¦/A{6/>­.~‚C endstream endobj 81 0 obj << /Type /Page /Contents 82 0 R /Resources 80 0 R /MediaBox [0 0 612 792] /Parent 73 0 R >> endobj 80 0 obj << /Font << /F8 7 0 R /F46 6 0 R >> /ProcSet [ /PDF /Text ] >> endobj 83 0 obj [1055.6] endobj 84 0 obj [533.6 588.2] endobj 85 0 obj [585.3 892.9 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 585.3 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 1138.9 892.9 892.9 1138.9 1138.9 585.3 585.3 1138.9 1138.9 1138.9 892.9 1138.9 1138.9 708.3 708.3 1138.9 1138.9 1138.9 892.9 329.4 1138.9 769.8] endobj 86 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7] endobj 87 0 obj [647.8 600.1 519.3 476.1 519.8 588.6 544.1 422.8 668.8 677.6 694.6 572.8 519.8 668 592.7 662 526.8 632.9 686.9 713.8 756 719.7 539.7 689.9 950 592.7 439.2 751.4 1138.9 1138.9 1138.9 1138.9 339.3 339.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 585.3 339.3 339.3 892.9 585.3 892.9 585.3 610.1 859.1 863.2 819.4 934.1 838.7 724.5 889.4 935.6 506.3 632 959.9 783.7 1089.4 904.9 868.9 727.3 899.7 860.6 701.5 674.8 778.2 674.6 1074.4 936.9 671.5 778.4 462.3 462.3 462.3 1138.9 1138.9 478.2 619.7 502.4 510.5 594.7 542 557.1 557.3 668.8 404.2 472.7 607.3 361.3 1013.7 706.2 563.9 588.9 523.6 530.4 539.2 431.6] endobj 88 0 obj [446.4 446.4 569.5 877 323.4 384.9 323.4 569.5 569.5 569.5 569.5] endobj 89 0 obj [525 525 525 525 525 525 525 525 525 525 525 525] endobj 90 0 obj [319.4 575 575 575 575 575 575 575 575 575 575 575 319.4 319.4 350 894.4 543.1 543.1 894.4 869.4 818.1 830.6 881.9 755.6 723.6 904.2 900 436.1 594.4 901.4 691.7 1091.7 900 863.9 786.1 863.9 862.5 638.9 800 884.7 869.4 1188.9 869.4 869.4 702.8 319.4 602.8 319.4 575 319.4 319.4 559 638.9 511.1 638.9 527.1 351.4 575 638.9 319.4 351.4 606.9 319.4 958.3 638.9 575 638.9 606.9 473.6 453.6 447.2] endobj 91 0 obj << /Length 119 /Filter /FlateDecode >> stream xÚ31Ö3µT0P02Q02W06U05RH1ä*ä24PA#S¨Tr.—“'—~¸‚¡—¾PœKßÓW¡¤¨4•Kß)ÀYÁKßE!ÚPÁ –ËÓEŸÁ¾Ô¨o€B¬Â@ø €a—«'W $o&| endstream endobj 18 0 obj << /Type /Font /Subtype /Type3 /Name /F54 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 24 27 35 52 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 39 /LastChar 39 /Widths 92 0 R /Encoding 93 0 R /CharProcs 94 0 R >> endobj 92 0 obj [43.59 ] endobj 93 0 obj << /Type /Encoding /Differences [39/a39] >> endobj 94 0 obj << /a39 91 0 R >> endobj 95 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 96 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 97 0 obj [357.8 306.7 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 511.1 306.7 306.7 306.7 766.7 511.1 511.1 766.7 743.3 703.9 715.6 755 678.3 652.8 773.6 743.3 385.6 525 768.9 627.2 896.7 743.3 766.7 678.3 766.7 729.4 562.2 715.6 743.3 743.3 998.9 743.3 743.3 613.3 306.7 514.4 306.7 511.1 306.7 306.7 511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6 408.9] endobj 98 0 obj [565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3 465.1] endobj 99 0 obj << /Length 149 /Filter /FlateDecode >> stream xÚ31Ô35R0P0Bc3cs…C®B.c46K$çr9yré‡+pé{E¹ô=}JŠJS¹ôœ ¹ô]¢  b¹<]ä00üÿÃÀøÿûÿÿ üÿÿÿÿÿýÿÿ@¸þÿÿ0üÿÿÿ?Ä`d=0s@f‚ÌÙ² d'Èn.WO®@.Æsud endstream endobj 8 0 obj << /Type /Font /Subtype /Type3 /Name /F33 /FontMatrix [0.01204 0 0 0.01204 0 0] /FontBBox [ 5 5 36 37 ] /Resources << /ProcSet [ /PDF /ImageB ] >> /FirstChar 136 /LastChar 136 /Widths 100 0 R /Encoding 101 0 R /CharProcs 102 0 R >> endobj 100 0 obj [41.52 ] endobj 101 0 obj << /Type /Encoding /Differences [136/a136] >> endobj 102 0 obj << /a136 99 0 R >> endobj 103 0 obj [583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000] endobj 104 0 obj [937.5 937.5 312.5 343.7 562.5 562.5 562.5 562.5 562.5 849.5 500 574.1 812.5 875 562.5 1018.5 1143.5 875 312.5 342.6 581 937.5 562.5 937.5 875 312.5 437.5 437.5 562.5 875 312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5] endobj 105 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 106 0 obj [748.9 748.9 249.6 275.8 458.6 458.6 458.6 458.6 458.6 693.3 406.4 458.6 667.6 719.8 458.6 837.2 941.7 719.8 249.6 249.6 458.6 772.1 458.6 772.1 719.8 249.6 354.1 354.1 458.6 719.8 249.6 301.9 249.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 458.6 249.6 249.6 249.6 719.8 432.5 432.5 719.8 693.3 654.3 667.6 706.6 628.2 602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7 667.6 484.7] endobj 107 0 obj << /Length1 1546 /Length2 9294 /Length3 0 /Length 10320 /Filter /FlateDecode >> stream xÚ¶T›Ý.ŒwwP(îîîPÜ-@€`Ü wwmq)Z¤Xñ¢¥@q§X‹Šségç;çÿ׺we­ä}fž™½Ÿ=3û ­¦»´ Ä ¤q±ssp‰dÕe ¹¹\\¼\\<è º`˜èo;:ƒ>ꆸˆü‹! aO69 쉨q¨x8¸yÜ"Ü‚"\\..Έ¨@è ¶¨sT . wtYˆ«lg{ZçïG“53€[XXíp€´3 ¶ºÔ0{óÓŠÖ@'€Ä ‚ùüW &1{ÌU„“ÓËË‹èìÎÚI0³¼À0{€6ÈõÙ~KhAIã@gèÚƒÝÿtè@la^@(ðdp[ƒ\ÜŸB<\l@PÀÓêe5ÀKWËŸdµ? l€¿ÀÍÁýOº¿¢'»ü ´¶†8»]|À.v[°ðRAæ c]l~Nî§x 'ì´z"ü±u @AZ |Rø—>wk(ØæÎávú­‘ówš§c–w±‘…8;ƒ\`îè¿÷'†‚¬ŸÎ݇ó¯â:º@¼\üþF¶`Ûß2l<\9õ\Àn e¹¿8O&ôÿØì@0?— €äy[Ûsþ^@×Çô‡“û·ùIC€Ÿ+Ä`û$¶=ý û¹=AÔà÷oÇ#tnn€ ذÙ]Ðÿ“ýÉ ²ý?Õ ö˜p=µ7€ë÷çŸ'³§³¸8ùü‡þG‰9õdT å Yÿ’üSFâ ðcç°óðs¸¹ø‚OÿçŸø[ýVM ø¯Ýý+£²‹- ü§ˆ§Óû[ˆç_Áôר0þ{ ÈS?ƒLÿiS.~.ë§/îÿç!ø#äÿ¯÷gù¿¶ÿÿîHÁÃÉé?ÓŸ„ÿè vòù‹ñÔϰ§ÙP‡‘‚§–…vŨÅþÑõÕ·QŒÑYêŒ ‚Mxmxé0ãµÛg¦Pª þ4õ²÷üt¯Xºêpº©ŽÄêc'ÒäâcÙ:ÔÎmXS“ŽF»>t‚¿<,É£óæêûeOXµæa"µ°±·@7Š>b£œ$Ï‚ÆpèÞ5Ûðäè…y·¡U ';Ê÷9×½ °‰à`©`|kºMLC z+NS¡ºEÜ,‡åàR§LçChA¹,×âÅØYùñÂQbœ×Ù·p„ GHà]{hÕæÑ'Iñöõô»äa[Ö±~›{žoOÚÝŽ´5VP–"õ fñ ˜µSÅàñd µ¯Ž¬R« â;µ¸rgyéc°S„e³)ÃÂa'»/ˆ£“ÍŒ˜çžA.„^·K¬­ *vXÂ÷Ç£t½èoaƃ¢·o¶|Žo,P™«CxSŒh˜#û-=Fã–|ŒlÇoRÀœ=È%½¿lSž ©šö:DöÎÄ{lܻ֥OÖ¾™^ÜÎû|e0Ïy Þa •Tµ\É/êÇ]¸§x Ø·èNjÀ©N‘*ý+1§T\Ý{x{}\±äì"<ÆuÃñk§ð–¨Â7¾úûÖGÌo0A TÞõŸMß·a»l’Æ'­R»7Ú„š2`‰óí©Ã7ÇGKž‘ú ³¦ÒöŽï ™-µH븹BÏh²ž³ wREi(“ky– ó!øvÝ5~=3†yz\Y¸½Â}žÑ7Ÿ"&dÃVg>#ðfÇ6M³Çu¢Üÿ¤Ö±S[¨]°ù'µ#9²å`ÞÖ˜›§ñ¹˜Rª£Þ/ϲÎDF!¢k•MዳençA“ (¸“åy= Ä{ƒW€{Ñw®ÿ¼Ë¤Ø¤7ö><ÓVO”Œï!&®!ŸL Ù¡í<ù&¬1)Ù®#‚Æ.c ²ºŒEË@¥GNØÓiõ¦¢c™éK€çb¯m¾ ±û]hq`£i=ëÃKR“©Èä SÉxØ'±ápNѶ8Ìfº]mÊ畾ڂæÚË ÂùxSè°7fïÌt2$õÍœÏÖ\Ô³ÌXô‰i—Ô¢]È~줊¤3@ÞÍm!5ì§ŸùÜôè&N÷´\A %ðNXs«ô}Ï”_ÍHY~•1m“…%;tÑi!å¸Ã þUiþ6åBû×è"æ35Yx»Ïè".V*h‘7jíýn ä©CÐoüæýˆ•Úú*8ßjADëg‰. G“õFƒið¹ tê¥g‘$|ƒïc¶ÄÜè `2¦¢EÒåWHfw%%HÔ ¶¤#^>>þJ„E~¯õ‰ÅœºÇ€ê§TH™˜p]Áx?Ü—æÛ¾®ãXLwößNÌJ(LÄôgD{&kEŒ8à<à‘#$ßÔæ 1ä[#Ò—6'bdDz¹‹„…Bø•ñ¸ Ã}6 Šìö)`x`¹›ƒéH÷Â}š¼ÙR›CÑdOÞ$nj&[v²QÜ;Ð…². W«:Æaiñ“DåŽ*‡V:|º“}j¥Ëü§§.îu£D³×þŠçmã†ÿs¾ƒ©wÉ%tðUä¤+\“ÕÀ¦)-N[ºê㺱ص'kå(ršìvGÚk9Õ’Ba:åµO ‡·ìç…ÁÆWÝzåi ¦Ê;ä‘Pš¼Ñ›Q; ̼ž ‘Ý%9IRòŠQ@2 SÎ9^/Gúœ¡6¦%çÒ͇r?aËZ¡u"œ„˜ÍþE|9FµWcþ³#Õ…ÉžÀƒu QA=†1GQÄ-5LË0ÇÒ4ù¾œæ`ÓH%m¬Áq:…¬ƒï†¸õþË>“Ç;ZxöUŽÝž|ú(3¥…ðDûøªíyB§å©A}¼ñð·x³Fwæ_“Ø¿k,0dR`̦pœ'€«ˆ€¥ŸÃ(Äá½BË$)«ÈÇrùª?~4kàŽKx‹®b —FæZs*íMV±RÇ1È8ï·° Naçëdóß¿f9O‚¸Q‹vGÎvLrîâÅ· ˜”cªáÎýÀ…¼ï¥hÕÅ<[F­(RˆJmïÄ=ý*tÒý<1¿Àf'†‘ž‹fr• ó0išÑ»•×¢¥$ødJ³šèNà%ÿx~U]·p$Ð^e0‡ñš@K ë{£&ƒ·cEúÜÏÜ2ö›âë MÅkt<¦mflÛA3鸞 ¦rl'…o,ÑF‚w q‹_-K1Þ¹êH2¼ûй†ùÙ þ"¶»¦T¿Îý,Hâ>2åÒLþj ·eºÒ‹‡ow>¼Ç†Óê‡\  sÝ5Š_YïÞýŠ"#§«FhNÕ‚hM',e >C>–Ôuïó€ðº/ÀK=›ÎíÍdup ‡UEÆ–­ñV¼”XLé.Ö«Ë_Ÿ9§/Ì÷t €¥“ Õ"‚ɭО=gÈ笨 <¥pÖÌÁy[ß|SÖEÈÄðÞý:¾X S¸¦A+„ô „âVäoï)ªáøŒê. ž‰É/o>ô¸EÊ6_;‘‹mqØ0Ü`îrI<¾]#èNhm}¾}j>P(Áf”ËA¥p»œZ Ÿ]ÀŸªø‘¤Ð¡‹î±ã„Nu%ÎAElU¦N͸]—éÖû¾Ûè¬Ã•ég¬öB„´#eM’I|H¬c#?‹2mdÇæŠD/©û»êK‚KŒé¥›’I‹*?ã‚ÜäÛ$pºµÕ*Ñ^=«~•zî¿ gP“3ó“SS£Ù|*—…Kõb;d¿ÿ~y°•9`±Ve_ÝŽîósþ¡S‹.»*YwÙfA6Õ¼œÙ|IÔ‰Ž‰IæÂÝ´H¾UÌìenzø>H{Uz[+ ‘2¡užæ¯¯Q­Ü€’}TàÙ«mó7|÷D#T3©ÌT/ß.‰ý@,sˆÏ}œyaÕgFa‚µŒ¨ ÔoCU ÅÝbb^›å⎒©ˆKRÕšþ”Ž—ÄA|% G¯ÝX ï?~U}ëP–uníj˜%$Ö#µ•Ø›ÉÛRâжÓñŠ! ^XPcâä¹o`¢y_o4‚„³ÚG5-á_>Q‡½d¯PáÀyfäI½;f9Éñ9¸9áμ:[>íKžÊãøiö¶1,T/zj m_ Tã1f·u?À Á%ö*¯ý–ÎÑZÎ7M¢u<¹Ñdodf!Ô™£~Ô†¹Œ¾D-3±/þ1Û©˜À9Ûq·Ü­„*ýUÕr´ÆK]³ªr#Ÿ­{0툒%’räíÑL)í·añ*GÁ…W¨$\ô!j ^%§Ñ¢NÐf}o­·QE¥«»Ï”ÏLÓr'¯jn³ÄÐÔCwö á:PYÀè/èá5«xOö5ç&ÿùº~Ô´FQ××'çÏn<¦Ê¯ƒÖ3bÛ—áÆ»Jé"¥€¼Ìýè…ð‚{*‚4Ͼlp VìÌ©6ù²È‹’®]Ž“âYBÜ© F´%ˆI|zI"+ 2Ôõêd+ƯiÆ?åÙuïUØ2² ›~;>_3åì? ^j2,&ª‘H†Lþø¸¡hñ°\x7ç|’Buù®ãHÉ–if!/ýXûö‹Ülg¾{ü¯mÍs[æGžw|Ñ“U½ò*úA&ÂÝÎoD" ‰ØPÚ8ÐÑø‘h“·ùç¡Ë¡™cYûÕI‚oès?'#Ï ¤Î6Ž µËè+2­ƒ·IL¶•Ë ¤Å¤Þïmåˆ× \JÖìÛÁáÅ>ÇÒU¬8Ä/ áÒpWã¿X€›ØOR£9÷ˆf‡˜³³ð¼A@WÛ…¼‘¯{?X¨ÌܤJËkx:áõNMë׾߯ÞLØ(•ƒâÒzДµLº@:³÷¯:ÎÿJÃÖ ¨´Âõç;w"s!UîÌ[;ÎÔãþòEÁꦘÐÄa8ÕºÿÞÀSõ ÙðR·¡á =‚2$\0#¾çrov£\ÅzSd„U‡º;´™ ±ç ,[{Fw4ìnƘQ·´i–ëxþ½O§IíØçhÕâÁáêS¦ö|‹Øñàl¬mÊÙZÕK’c‡Ò»ìCµµªZÂF”Ø!ä×Q¢W-À!1aÄœ¯&Ýýi=€YH†t&*–cÓFƒûìcÃÚæ ô‹}VJ!)-CQ¼ãÔIC߇|çý^ïùŸQ`UÚ¸w˜©d›˜Æq(©ì;CÉs§Îд1ý Ÿ•%~œÄÆé=ÇåC®Ãé+ ö{Õ¶í8æ÷DHÁÔÁó7•äQiÍNï­6ÜO(êüàE”ýš,:®³_0à‹3“jVíÊ%„+~Mp—¡ã¶´¡"%óD^ë.Ø4‹¢8mi€})éU´lI»ø¾gIn¯2žñÁ5á2÷'è¾G°†ëÁ_)ÏÔÖi›T9tä+9'\ÖAMXåK_:m;Èèè™#LÄ$@\â6=ÚÒñKIÇ ïžZ¹À›Q^v̱è¿L V¯-뱎ϯY Z˜5r…G×°+ŽÎ^ù6î±*ú'£ÙZ$CÍä“–Ö€nK¹ñv‰eÎh³fÍê.%¨žLõr œM?Õ¬ªð‚öcæm6´kD—s$%*õuq6|Ì÷–j[íJœ׳-Œyb÷6¦"ýc¦ü¶¥U¬3´´¯J1Ý\l˜Vq–r}²ÎZ/!rã–º•*² Ï–Ï•"|µË?‡f¨4ïîKæSqç¢F½se¥:bSÔ]'ÄLâ¸&%ËïAC¢ƺ?XÃ+´MНì&Ëâ ‰B§IRHáÝt•nØæï¬æNê»Ö†ÑX¶î‚ÈŽ ¿ŽNýkEDZqõ‘4e^öêùb,+Yëä”°ë¾?xÎSëlÆ­S±‚oFïTþ¡¡L*;—ëÖ äPÍÂc¬Á”–MöAwJC|‘ïl›‚ÁØåب ß?|¯ÀÛñ jÀrî]iLÃd!Hq‹ddnˆ~3}!¡Ð$àí—°ÚFµ¬üéìuëƒøµºa½Û,%@ Õ ²¡2ÿ}õsáxŒo{E_VF&¤BíÛ­÷8·»EE5“)Ïè–ñv—1ÇÌ2º¢ä¯•MŠy^l“GK‹Ð>ïºkúžœ[Ê`ãÕ Ž ¯:æÏ wä~Ø'Ádeï{]LÑo?’o[pdE‰RÍääÞyp­  ”`ΖjÝ™ˆû?Œ¾Ñ,bQ•ÕsKv,„K Ýc•cd>µ¹Žr|‘8R®o¨U·ÉÀ—³rØçéDhC„ЊA˜þdI¹œz¾II›^¤¢±9ô=\¬6´[è¿è£“žáçü¶F„¥¼(§îKdYˆôÎX˜¼Zbï9g›ºµ6¶¢b‘ÿ 5ôþÛ³ô`‚Ç&ÇñùÚçSsÊR^s<5œAêwÃn_¼ö4¨Ã'w\üJýY§‘/tyÛN|VQ¾á®lˆP<ðTRë ¢9µo-ê_´LÐ|0e‘ÕÎŽR‚ßÝ쀜¿BebIrD˜ŸrÌq>Šc‹4܉›}>í^ü-'s)ßN©/×Õiõö‡îB”¯ó·bzÓ(&à,bâ ÂÚ·ÏM»œ¸8Ÿá5Wý²e@±©uñîÔÈH¤ã:z´ÀÓ 5™ÞtzIìžaq»é‘¦Ÿu/^ú¾0;6{ÌÆáéKS³Ï©WÁ/‹øpºwFþî@Gf+4¾po¿…ºRCœ¹îŽçÍ“Xx€*¤„6 ⤶äz1¶8 @™yû•a)¥ŸåPæ‹ä=ÝjËižæäéðî)í„¢_¯d#½…–è»<ì‡ÐŒ‡"PÍšÙ(cEÉo¨¯eë{%’†q“½j±ìd ÅK¬ðSNåŽ× tVsx úOȺ:Ehjô’Ë)Ÿ|V-û J¼²x騢š-¹ðŽÊB%°%Ò¼ÃlÃ3åû Ûk[²‹ŠÌ‚céûÆÆ%6]ž½¯ÝBÂ]op©Jd½ˆÅ|Åß~ï<9À®ÕF…v?ŸX4Ln5,Wúqäm™¥q+þ:«É;륟‘âH_[lXºç>x}×ÊúrÒÃZfédÍn2…è Ž9›a3hlˆwCÕÖ"ÈÑch…r'—->ÖQçm où^V(wbŒJH=Òm§¡B²P‹—‚/œ0$GÜ¿ôoîK#81„ù»ç¸ÌS;ØN•¬¶c¥ HŽVÞ>wõö]£TQJ?IJ7««|Ø«rß 'nvÓT“¼Ù¢¯„ 4W©2_²úX†$ôêÅ!7¡ËP3²Q‰Ä«‚‚¬ƒ¥wcÌBVº«F¾ÛƒåQAùrßD¸¥¼n²V^¹’¤5h¨¿Ì÷J;EÑŒÈ)Y¼¸QØ%s\¡ .p‘ÃÅ[z‹oÑ—$œõ%Ÿ^o(ìn1¼õg¥ZÝg‹ð§×Ú⻈U¨8Xªq†Úe™B«{@ñà7,¯?uÍ;õŒðùÔJ1›õJÛÉ9ÜÖ¦ÉG(pÑ}øS%Þfä„òS«‡‰ÏèSwv{‰®½Ë{c"iÍ/ãèAjyxSÖ%ô™±ìpH+:5Y+A¹êƒÐê=RòëÛN-_QÈ8˜Œ¥Íý%cLÙ³ºZÁF"íÝ|²½YïÙÆu ¯¸"„ísJC_¢!FµGèMËöðç ³ ³ÜžªsÔv¶÷Ñ`$jlÁ‘ü ä…*¦bD¯d·³ ¡·›¹‰t+ŠRbÈ‹Ûèñ×(ù‚×8íVí{Nçø®‘"Ëý/iÛ¥kË—ÚrˆœÅÍõ]\'ºÚ~dìëŸu /Â’û¤‘|ñrÖ>|,£ì滹ú\1Ǹ˜f€=z¢hǬàŠáímèc/V>D6t–ÃösæL_&ålܼˆž‰›Øc÷\uYî­F~îºcÒˆýê£|öqŠ~jþçÖ0‚5oÚðœwI³¼™ÅÎ@,ZJýJÁ ô_ýߥ~Š7½¹Z>zI+ŽwGQ‡+ «uôÐ6‡™Ó@$qNH%¤¦ÂTèÄ»'y¾¯çÈ÷”­’zVz«Uè@Ó!èâ •d¹Ì|dÖ.¿¯ÓæX´fk„ò´Åu—‘¼ºÌµv\?˜‹}éåzœ=|3D•N«§¼×"P¤A¦ä³'ξ"UuAÚ¨y?´3À~®€_z?lMŒA\¬Ì™¦sý™i¡‡`kâ  sΡ\š(> ;3&\«õ¼äõ´M Ê{ÊqÚÅÄ#‰äó¨ŸWù~Ès/›•íóõæôÆ;~¸šÈO¦Æs©%¬-—ª`‰‰4Ÿ¹­õ?N"WPʈ!¬HÈâ6d3²¡^’l ·lÈp]dê-&ÍqcKá¯Ò¨en"#4:²×[ž¥ig?HÞL[ „1™-¾ŠhEš—ê±w‘tç d’ÂrspS »çüújåò]‡ÿ)½Ï;´¤Xe9é:Øæ0´Qp£$QÅïI=ƒöÍ¿ÄVægÆh`­Ó S„':ƺ$æ+ èÌRÄ?-À$ºZÎ켃«e¬õ?ÁdE®¹k›­5à Ä9Ç·:u*3S7CwÿJÉ/¿ðV©Þ¢¸„ßµ£õ‹©š§D_”Qp²Ô: ¿þ@)ðµG ´Ê[¯Á ñ’NLÙ*¶‘¨@ˆîNËmDïˆh¢>èè'M”*‹ºš¦güEÞFÈq¯2Ea*Z6z×\:@VvAYˆ>!fª£zö«ÁYhö>ÓC¤vÙ+ð¥höL‰å²¢wõOû0ê§‹Hœ¾×ÆC‡P̺æ+ñAkÀH÷{õ:ÚQÇ/Õ™ñÜþJ¦…"–ÌJÖj^…P-çOqE›Ïø@wòó—3éI/u“<]°á à (¢ bSŸõJÆ2ÊÂC˵÷p0S9¹ó“@’½QÄ÷oïÌ^VÍ<û>Îr¿ú‚˜Î°Ä/öuÝð]=ÇRsS±ËÒgA1ø~7[G/VÖ·è$â3î°BË”GÑ]s_Å"¥Œc’✀ÖÞÎðõ7oÑCq’¨c.yg&§=ïßžß?ÓÃo`Õí;‹ªìáXàÏ—ít¼-¢`¶;oºûRz¹Z¾×פ¬,=›ÔF®9ê*Ñ;>ecsH_f$ ‰ HP:ç ÔÔbr*ÌÆi@˜¹5ÁÏÖšÇùÕ{©¤˜{¤£B?äðFÑfªÁ·4Û™Ÿ%´Í+Í®•»}¹gž,Í>1l—¼¿ö äı¢r¬ Ý ~£D¹Šû̸áö#|`…©PWѨ«ð×à-'ÏhçÏaž×ôA²0‚¡¸è0(_é¯t᳓Àè¦ÞåB5$‡WcU“çž ¶€¥¨/Y¤7ü!—8É «d@t6ÂÛm¼Ð¬ÑÕ::S9få×E_wä ê­%_kÉÙʧŸÝçAùýBè°´ 'ã ñ|gQZvYQÄÏ> †F¥Ë4^¿ y•´†u/³™w:vÅ~Ì-ÂØW½YôŒ£V{‡YC0›Âˆ6‹æƒ#N†=Q>bC I_Fzy‚o ÖhbÆßi6€âÔ–ž¬µ ·;7R&k˜T…žÇsZÞF\iT§T¢çW\ÏÏÛË}.Iȹ‰+Œ†7;× … ÷Ä= P Š–R öÙ7,„<ü Õ—¼` 3|n¹û2^Ƹýö8¹xt mæ›±>¿« ×éî ¢ûŒÇ†65r/àåw›Ç•ÁÐfœª¯˜÷­!S‹h« Ù)ùó“‡Â…Œ¨Û€i¬‰Ý^ÇkȉݳXÃþàñŒJ{^“ŸÍ•ãb%™Ž¯fê ò[ë´–¥ÚüNGfUå›l’¬ÛëGê˜5xTC[—µLÕ¨Y•YüCbÃHšü„\æê©õéZràR—N/Ö[¬÷-rÒ¡C¬ÌZD“LZM©Ëð—%+g-È–¦3zº~Ó~?Ópê×79ë¥F_¤Þ2ŒVµ| Ä(H/ŸîÒ›¯§¦o³WGÍ›{Ÿ@kf"W~€¹.:h4|óìßEÉ“(fW ‡¯M,|kpeÖC§OVµ8#ö"×2‹¹0–ÐVŽrQzªuk›‹*hTåÓG’6$bJòÑè#ZMJr\ÕæÊ6ŠÈŠ`»ãì$Æòzd·C V»FWº—‰T2AæåÂ4#ñçV¯1¿á¿æÛxà8Ã,m(³Œ˜V ¹8¡"Jã‘Î$uŠf ™0/Û;¾v¯Q_ºõºÇí ë'ò“㦤}…qI /Aê,ë“X)§YÖ²y8<'žF. 'Æ +ºªñ5i|ŽÛ3:^–NÉ‚GÑC)2“=Ù¢øòhÇ]s3r‰{#h«¤·äU¿}Æ*ÒïóLOZ1“?„ìó8&féÏ{Îæ–ÀáÛdyðmè*·X'Ÿ+LÃë<Ž<{j¤`^D¨»@ûƒ]Ž}mÚoêU|t¥sú¢@DÎY]#å oâb–)A<1=Q!^O<£QØþå½LG˜ª¢ä'Íyªkƒ#{R+(È"¢j«XàÚчÀ,<º@Sj×d!ÏÔ=J35Eæk:âò"Y’?аï1G¢îŽçˆ¼YjZ@VøŽŸÚlÿ3ž™iR=Í Wå©)«€oRpÚâʃjÃrÍf0QµãÞßqK)¾ áÔÊC~p<'ü°ìñôg_l‹g7MÙ!VõfwßÖ endstream endobj 108 0 obj << /Type /FontDescriptor /FontName /UBKXEX+CMBX10 /Flags 4 /FontBBox [-56 -250 1164 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 114 /XHeight 444 /CharSet (/C/D/G/a/d/e/one/period/s/t/three) /FontFile 107 0 R >> endobj 109 0 obj << /Length1 1896 /Length2 11786 /Length3 0 /Length 12963 /Filter /FlateDecode >> stream xÚöPØÖ ãÜи»»»;Áh qw×4‚ww ÁÝ%¸»k€ð˜¹sgæ~ÿ_õ^Q½ö^[Ö9{Ÿ†ŠLUƒIÌÜÁ$í`ïÊÄÆÌÊP×ec°²r0³²²#RQi‚]mAÿµ#Riƒœ]ÀöüÿbH8ƒ€®o6I ëQÉÁ ïf `ã°qó³ñð³²ØYYùþKtpæHÝÁæ%f€¼ƒ=È‘JÂÁÑËliåúVç¿´ft6>>Æ?Ãbv g°Ð tµÙ½U4Ú4ÌÀ W¯ÿIA+håêêÈÏÂâááÁ ´savp¶¦cx€]­ê ³;Èð‡d€2Ðô—4fD*€¦Øå?  W 3ðf°›ì]ÞBÜìÍA΀·ê 9E€Š#Èþ?dÅÿ€™íïtEÿ‘lÿg0ÐÌÌÁÎhï¶·X€mAiEfWOWFÐÞü"ÐÖÅá-èÛMß¶H‹©€o ÿÒçbæ vtuavÛþ¡‘å4oÇ,eo.á`g²wuAü£?I°3ÈìíܽXþº\{{Ÿÿ" °½¹Å2ÌÝY´ìÁNn 9É¿8o&Äl– W+++7äyšY±üQ@ÓËô§“íó›?GG€Å› ØôöÑÇè¸:»ü|þíø_„ÈÆ0›¹LA–`{IJ¿™AÿÁo÷ï öè³¾€õŸ¿?¾M˜¹ƒ½­×?ô?¯˜EKZJMNá/É;ÅÅ<>L\&v.6€‡‹à÷¿yþ>ÿªÿÓª ÿÕë?åì-|ÿñvzÿâþ×dÐþµ6t€ÿ­ ìð6Ï í?ãoÀÊÅjöö‹íÿóüòÿoöÿÈòÿ:þÿ·#i7[Û?ý´ÿ!üÿøv`[¯¿oóìæú¶Jobÿ©: ÿ,´¸ƒ­ùÿõɹß6DÌÞÒöïc»Hƒ=Aæª`W3«ÿ Ñoá-¹-ؤêàþãÁ0±±²þßÛΙټ=*.owõ§ ô¶Rÿ[RÊÞÌÁüÝcçâ^ˆ¬oÆÎÅða{[Rs石 `a¶wp} ¼‰óX88#þq£ÜÜñ?L"‹Ü߈—À¢úb°¨ÿƒÞ˜ÿ N‹æßˆÀüñXLÿA|³¿ÑŠYÌÿÙ, Á·’ÿÀ·"à¿ñÛœ³X€ÝÿásýAwpsþW‚7Šå¿à[«¿!'×òr´z{øþa¼ÙÀÿ‚o½Ûþ ¾5o÷d{kþ_±oÛÎâðOö7îÛ÷Á¿ÜoÝ9þã~;wÇ·ñwø—ú?þ«y¶·^\þß"\ÿÑúVÌÕÊô/õoݸz8ü+àíÜþß´¸ÿ ¾õçñ'üŸy2ssv~{kÿÜø·aû/þóaKÎn¿Pá“èª2ƒ×ïÄ’†¾£¯lKÑÞŠ.“¾ø·ÔÃG´&ªµ=ù>VŸÚmC\œÄíŸøv,V×GòŽ˜IStÏ÷ÅÉW;Ⱥ²Kž*ÇÉU5ëÁ㇌g]_éÏÑðù]µ½*n¤çÒi¦­hƒ ¢Yª\Ó¬9|r8W&zÌ O´ÙÛ»Ìì‰WRùÏ ˆ~'1>zì±sÞ«åšì.Ý”zø$з˜£SÔ>âÉòx >Å…qòáQ…$FóMÂh@'β®%ÁÒ‘‰ÍüCã ‰R´í©ð83µ-*ñ¡C[ìN|Ìœ:ž¢é6ã|DµÎý\見ÙLç‘~×SUöˆ´ÞÛ÷¡·´"pÉ Ü‹yý}|þÓ4õ ¯¢ŸÙMs]õ{|€ž(§P<—~2ö_õo4?œ kNBQ‡Ù¬È²àLN¿ê öWb¾ÚC•ÄLA)XÖ¬Ëwy£¾ö*=6 Y6Ü{E†ÚXpÒÜu ]4/ÕMŦ[óË÷ØÃŒ(<}¢`×È«ÖÿrÜ—%nŠ)2Z®ìZËSU ~KŽÌ£!lް¼)ºW%ᯟ*™þˆth^%nžà;%W:y&käñQ ­²lD!)yΖUÆ¡“˜ßÉ<:yÝ%ό˽P9K"˜ªI 4tBÞ êŒOKP®<€ó5y+„"C¸IϺçÎ*o¨!Uë·GÕëå6âs=eüNïä–Á¦Ž2³D­ö´Á¹òÞqNÇ÷f TØY1‘+[ct—ž¥pÕ¦ûR¤ÎX/ ¯âÍù.Ó Œ±^>7ÇØþ;Õ­Áøje7çy:E!7«’عàéAÉñ¥ÜÖÕÙ(£"?ü¦fRØîîYĦâqo-ú±Ÿï9Ê÷ýï¶e <ðB¼wÕø¡æ†àÛHQç7ǘ¿Lá˜y¶–âæ»ã§”Ò¦c ê ~â^LçRø9›^9ÃàýŒ!Ó‘Ë I×àü¡<ñÿBöY$®8JI.œö\Ð ìünŽÎkK†º!JC&ßX=½ ;_3<‡Ðˆšø ¡`9»peˆpû´/G»§Èw9¬/ŠÈÎ0‹ó0IíA*?^ý–WÇNc3*ºÑ ¡ Ci‡i–÷HƒRz,jBPžY ÌtRB$S^Šy!C†©÷Ó ‡¶ŽcåÁüXšîJ¦t¼ÀJ$›$'Y¹æ«‹Vñè]ù•d®Ä2{ƒÂðµáõÓ†ßÂÙ50uÒ¯owopbS.}Ñ–ÔcªJî©L›ÕYVHvl¨B>ËknÌõ×t»/¸˜W9®cÃù\NöDìÌ æáü²}¸2ÝÕÏ(ÔÊ0Rª-g¶ÇÑÇQ=B0ÀkÓ‡dub { gÛ†³/f¿LßmnºSî@™íâ¿È-ÒO£Ë6矛á̧Y¿H<)±ï˜Œr9W*Æ]óçãïÔ}üµQÑ9æ¿ñÛ,sAÉä]ñF$™Ÿ±°êQ•‚ô·’» aØÅ½T¬¤ÆTø]ÎÄv^o”ƒ<þ)ì# ² 0£¬¥QÙ ¤–p•?æš? ^²L6¶q(ÃÄl\ ûC{¼]:xïq<‹ ¹&©ää¦OX¯ìltMd"vŒwKUÒî ŽÂ$©z3•^ü‘»•èó—G¾Bw!+©nc]Ú~¯øpå‡TxZ>¢}Õ,ê}𯰽}¼-ÿ„ÙfžÆömŒŸ%èÀ’Œ{±cÅ/&UmœážCÛЙ¤c;,SÙи…ÂA&@ éåݹWv‰2-_Ÿ v ÷nÏýÞšÅyrn93DD޳ÚbžAÅQ×ø£«?€+Y?N]RÐ*GbÍ#µƒôò‰SU 닇•iW€ì#XÊ/4¸A­ÀîÄû>Iç’dæZ_*‡Ó+쇣·:‹Îã!*Ó»¶ b ±ì^C¢Ub)È⊦½9p_ß)£©9k^‚ÕX+ïžä¤ƒºt÷'£ØÔ;&rÎ:ʵáé–o©B• Qz¨v©È*Wã!&ùŽ/H¦$ãùù×zæq%VÉñµQq?„=•OKòÖ·ã\  ³Zd—T ¬„¯³gbnfŽsг»!–å.ä ìÇI‰å ž¸ñ‡¿Û`ÅÎÛñ‘Ã)#L$§{ô;.VÜ>bhf‚Bk•î~£hªëÓôŸ|]ÄRN,ún*ØÈ‚*bŠ5ì¿õmûSÄcЇ’|m0_á<Ê3‡FRíä u‹Žœ^ô¨~w‘`mn`$Þ&¥]m¸» º·ž‘‘âÜ>«ò‘ºSúÙOzž!7E•öq!Ó€1ÿ³gãUùw/Ár"‚\—°HšÈÞ.Eõ§~nŠ&ÉJ‹¥ä<_:g Ïî÷Û®vÂWy½TQ>WWñß|sÜMèÞg– '&æÈ]ÚKR L"ÀÎ?x—Ö`ä¨MOž^uéÍ’ÄkÛ˜QL,ø®‰Ò8jV_Æ~ò9„*èÙKÒ@©CƒOvñB-¦}.ËÙâF¦ÈèÙ€‹+"‰b{ÉG?ìÃD=Î>®ÂŽFq½¿}KZ…¡î#/u>¥ßœòwÊj’‘+‹fТ×DW@×ü"„ërÎf5º%¡1ˆ0Évsá6BÐѲ^†ãR!8” „&ªš øjç[~GáKÎÝ¥ƒªHªª;²îmŠZ˜­n6}ï÷tJG…§¸¨`¡Â޶Z´µ Už­×Нbª½v8dG_¹!«—µxEkÚØâ ·ÒlÔʞΠ¯ñM3}¡{ìzr«AìvcÀ¶­7¿Ýál8lö.1St‰HÄ}Д‡Óç‡-w'|Û‹щT~ˆR¡oE©V†ŸŸ)Z\[Hϲ‰ åCŒ~q@5ޑ嬠l‰¹ÞztàeÀ–¯}QgIsæüv|pŒ«ô«8w"cAÚ#`ðÄEkxP¯×­:öÞq2va¡±Ð¹JdP5:„6®˜HÃMÔ‡8a†ƒQµrÏ!ãW  I°O“¾œgßM1h$ 73O“Ã”Ñ ¸¸"ýÐ顤¢_%é"Ÿ:¢FÔZÿ«'ÄþÝSP‡æZgÓ {¥6ÁçJe—õ騧+´€N*ñe«è«Ÿ”Ä Œ›×ÂÙ{„ׇ )Ëä²YU¨³h9€æ2öqqôb­1ùÚÆó6’Œ©« 36U¹ÕøÍ{fÛµYHë!i*ëìbÕzD¶³6oSö¡ÊÀ5%;…¦Ð–þ FòÿüUOwÉa9»Û2Ùó_DkùûsmËõ¹÷x–:eZ§µ?Y´ÝÈc´ç®tŒÊA š´œÅ2‰œÝ 0]ãÂÉøiP¾ú"ßo: /4ڸќ{d ¿T¥xAH¥†®9öÖ³¿ÇnIüá'G¿V'.1!ì%’®b< ÛΙ‚ùÌ¿•Tc*µŸ£mÊÖ8­}ñhÊB¡à Íjn$l=!xÍ^³à”1-·I õÙïã_ADHkø½fÈ’oZ4%L1 šÁ©É~µºwaÀÏ øô«ß¸Á%ãk¿F-^Ö #ÐµØ Ys95êü»…ªþJšƒàÙ½ »`-Š:OÉoñgñ G¶¤Í •IŒA¶³›¹ðú¥À=¸EŒÈ¸ó,ÑAÚx*ÍÌæV9c6êÔÞ$ï—ûÎ{zÒÏ´„JŠSîÇ)ŸßçÏeû[•Òî·Z':$H]¶gˆTž°ú–õ³è§aÖÜþ˜H³8+ä F6¶µç^h­+TÄ¿Ié&©ìuæ¤3UX °•ÐÓ¼*†°g÷ý¯õœ:NK–IÁryaA”…Ù#šôž!¥}†O¥ûÉt7¤àlHHWN½ŒdmˆYC»ÝÈ íˆÏË*O‰A•·ít–àß-ˆWTLPòZý9úÕšº<³,Ì„af¸QAZ0ï­xÎy•ò8'ŽˆßyðñoÖmZÿ¶sv¿ÍP½‰¥@NÈA0 ô¼jÛ6¾o+Qìmþ‘äU^šR}3Á…6¥.Êâcq¬&qéWf>Ü0¸p翸í\wHVÍÙö’=Ûm¡šïÐtØ¥4Ã`DØŠzÆmÌòSÇÇjqlè|ˆ%r*ýhÔz¿NÃoú÷©|„7Ò”Kø§¹ÆO›óÖxïŠ×êü½3}œÄ Ô§­(颸½•5äW½ˆ­¾þø2@ÒßOüžnZsÒRAZ ç—*+l\0œ¼*®?­°?ßuoÁš{Ï–Ñ+9Bl']G„.×>‘´òt0P}±¶„*† vkÞŽmÅÒ]d?“»Ž'Höî3î1Ùïëå›—ë¼-N’âì¥NÓÕ×Ýo-1®¹ÒW^W{u»·—Ôç>×jú zŸäòœšZ6E‘ηíì×­•ì¥,äíŠÕeHª´`*ê? Vñ~OPÃ+ÎÍR@ÿÄOdøã›8Yiž²H <Χó}š1æ;›nAÕÁï)"»”,¦yÛTfêõAá›öíd0ô!ÛWé¹|ØØ¼ŸHW”ÑÜŒ5³ŽÒÙ‘I©ÏèaAß”¨‹Üä|oC^çYKÈ‘ø‘ ãÍñ4à/*ʾsWjyܪK-<L^\ IÇõ ö“æ M铇y²bC`l¿ ¥ÝêùøÇôÞ¬‘2‡â#ŸòŽ Ð4꽫ï3RY𣸷Ä›n(èÌ–kåx‡_AÞåE +g…çÝ1Ê&AgàwûP¶ÂÏû›fÛùáݰ ,¶â9SþØÐÞa÷L{9Hʰ™:8X:fmÎq“X‡a‘xÑ"ßµÍ! ( Ë]‰5Õ»©°•x{$Oš'³]>øDJÖSp"=„HúK ¬š(í Ð͇³iÆÆ¤“iw‘3ïïVŒ%0ð ÙæÍQQŸ‡ŸRR¨+œä]…‚úGCå~°<#UvÙvKû,–2âÜeîð!bÑ0 ‡Øçë99‰V›àFâcKDUŸ1¢r,û^iÔê³”ø<4îîùÜ×znÛ¬výnÚÛÔ©òí°øLêz2fV×Cºò‚xt‹‹Û¹Ùkž$¸XÛ)ût%]]ÁL$ß!”®‡Üæ‹5|vжa~ß‘ÉvîµÜNÙÇê‚YçK¹P©Ø»›ÙËt÷ŠÈ*SÆïú鞥ÁA›ƒèÂD›R7œ‰ñšºi~‘rŒð›ˆñHÄùë2ï¥vªSÈìˆiÿp5ü$¹fŽÒYOönž5p‹>Èmö§÷é^upj5Ñq™ó‡©ÓÂZåOÕÙT×S–Wƒ„ZºÔVc¹ãU:÷¯mÌꌳ×-ã?ÄC¢zËT>ÀÊ9Óãe//^ƯäKÂOÌ–&w„âÜ2ûˆ/ƒÇâIïŸÍuÏâ°oc[t¥ü j2OÃÂö@:˜mÆÙ¶¶j%XÜ/p`‡âРiÁºxBŽÙ´g=èß«[ÊT¸kjpFãø«!Q½>ÔRÑüy6B´ÈC¢Ëtbñâš&9ß‚$Älÿ¹åŒ* …°g î̻Ǯ¡€4Ãñk©ÎßîÔQ=Ÿ©  iÛ¹M¶æØ%y9rùY¯h·¼”Æðw!^ž Y#ŽÝ( ×7ÃB_§Ôv%œãgаþz„w?j ®è<äÃKQ[2ãùT„Áu·CLêiÕìb÷A ¡Ôw^™9}ôi,T‹ú"/Ñ@]ýdô(HøCsLî6y±â'™[y–s$J ’X²ú-¿”í@%…:$±&=þ»‚Δϟü°G+¥Õr×ø£4¢q›T­4àqÇ÷–Ý[^ÒµòÁu¼eøËçõ܆F÷_.äF{ÛËCÜÐÂèÓÕ{Ö¯¶ÉÏù™¡ýÇÓ¨+žÂ—4ß}P$Ræ«“ñbŽwsY~-„·ÑÖ#¾"vAϼ:ó•Z5O î‘låÛwLS!)ø ®Œx¦n¹y•).ÕÌðx'í)“õ ƒ¿W´þöÔÒïQ2ß?=VëLû½fQ ?ÙàÁ¿ºJ1‚Jmx Ô*AßÄýùöyËHéû´×“ŒyË­"q7Y-Õ{…5ÇÏŒéÛöŽ=¸ïýu$|]†Íud¸+k½ØàM˰áu9ad j)~œñLísÌ19㢣æ$·GK–¿^Â$‚KÚ$Å`ö‡Œ“T˜ƒYqÖ+œÈ÷0õ¼n?Æ !{ý®Ãb—ô}¸«Ÿ(Ej£Íõ[ÏÝ‘#Jiî‚Áˆû~qâÞêk_'Y½F€Ü†¢fs!^ð/v¯`&+,øÒÙ*æ:˜†¦•îΛ\‘75[­ÞqñD­‡*;™ -vèí“ÇÙN´ *$ìHÅ7Ñ¡i 7tdžØO‘.(ÉQ\©0htЦ¶AéJFú¶8Íc a½ÀÜNR¸ü~,{?uñà‡<m·YÈÅ#+ý`‘h¿4j¸»î%]õi‹Ù•w©*Ï «Z˜˜ •ñçÏãQtø-{ ís¨_ÊëO! ¦Lò²y´h²$"#d‚üï­8ØŽF6g.¨Õ}QÉ¡z^÷|§H/‹ÜH8a‡ÂëUPŠ)M{8KâÄSÝ,½êiC®Dç’RêæP%+EY„°¤ã'TÐשÇ„»Y´áÊá[} ýpzç§ù•ì[#‚*3ʰ¶ÝñTª³'L;þåVÍýcO¨Õʼn›1A8ºcíŒ> :bC;W¾¿ Ç1›°ãú)ð¸ÂúÉÜAs|à(zÑp£ A^ðwõªþA è@5…4/xû_F£`b4d¨Ç<ˆ!—©È¶©C<×<3Ãë\xG¼>žÝqó½>9î5¾õ°¿*µvyäbò›“«B$í¦ÂdaÁák>æÔgE£9PdF8’JÈä°jÀrJ~^íZÎEßöæö£—\ðGRïç‰5ÏKªå*z¸ÛÊRðÜ-±áÄžá¢7\S ‰çöQ`H>ð³pz¤\ÿP‚ O…Ùµ£ÏñÂâSN€ê’Π䚘cæ’±_}JëX¥{xí¹oé/ç˶ì£òÏ Éí<Þ^ÈâEX&þsœÊgÞµY«o ‡?t(éúê!¢Zû^pú[¨³39Ì­õèptÁ6ŸH+Sþ]ræ¡pY]¾zŸg?_Þ]S¯¼ —Ã/ÉP#T,ãV#ÌÔ1ŸŽ)v*£ÛLܲb-§Õ”ÄœÁ¡YÕlíKÈšËBlW…Á9Aü±ê´AY½RÆ»XŸƒk½ÛäXªŽäЬ4°-âJŒÂíq9ÃÈ}5jÅû¸\„x hŽÙxå BÝ^ ȧÅÉwµÉ;ÜÈÏpósø< 8¼Oy~$:Àµ1âÂÓ-Q*ÒmÞËL•¦çÔ:ž^«…íaYØ1¯ûʇ‘åR”kõØé¿5ØÍ±^6B3'˜̤*ùæg~ÇVv2ó6ÝÝ›Çe^ßÜâêl%ÝeiÏ¡«ïV»/j¯ ­U¶é"'õÊ‹Yï‹%ŠÁŒW„­}Ýiš³‘ŽÄø^´6›G¿ÛÉj¨<û`î¢Ö®“dÁ%Ç9ËØCôÖ¥%Dl®âùçC²ô¼ÇŒ—qŒîE¼RX?vŸ–ã1Ü=ËáD^™ˆç[&›1ZïµSÛó ו¹“¢ïêP?ÔLŒs"¦>š@õ+;›¥‹s}x¼£F¥#âxè‡o2VaÊóY9·ä,™†+=„w2’½ú|X©D€ÕWk$QbqĹš/93{i“V;È#±K¬?nv ŠÉ,s m<ÊF(ÙQË•þÜÇæ žsá[œÞüÁù)AŨš½Yš/Iït!F›Ý½85<Œé¨B;¯kB=@¸5!ÑGj—_NîŸ9:‰¬º…éì8¹öïòg]Pv}(²5*ã<(tàa•«Ð¹>_‰JОâôXU9ÁÕ'Ï0› ]ޏz£ë¶þ~dd¾¼W°"uu ’Øû{Ël\±¯–ÙñÛbEذ“燡aÇ ¬äm/¿…BŸP2ãÞ¾õÈ)!¶G}}ÙKUy°{‘#ñ‘1טÚÈl3?Uz|‰¡Lšç|B6` ©B§Ú܉HÜcöu­>IšÊÇÔ4?q‹aÅÆÛzGߥò)@Œ{óS·â`vFÖ솎ý/–qMpVDØ@éDŸ²9Ç5»óÈâ ßýr/¨ÍÅ(’°6“Ûé§9…Š/è„Ö'ð”¤òi«Ž»hº½:_ÍÒA(:ò™É1÷Fœj–ºFÇDZãÌ«­à+1k§˜w„x•¸0reü4ï—½Æ?7Ý#>!¿<ˆõ·fBu}ÞxI§úhíáj“~²4^…1ò±Gó#©FÇ£ÆRïœUo—m(#i|‰ k="¿¸À¢¹ë°6""gµQItœíPÌ3%RåwÄÙRt¤t¢&Ú³œ~þª}…Ò÷Óûí<þêš,úõý"X¿ ôeíqfÔæYbº¯Fgm~#¨ÜH$ü£‚k)ðíÂKl¹L_Ó$5|—ðº@ÝÃ¥¢¤?+:´1^ëëºÞêpwàà5A-÷œ™ê¢>|qžè; ·hc?t¦@å£&"a@éã¿6ºŠÅ£<š~+Zì5‹{_6]u©oéiªŽj#Ý:“[¬mÀ…¬'™-ûôûc¼IbKG3¹S}íùü ìÔ©„mÑ•EWëÞnD1Ùç¶^Rϣ»Ïmƒ¾ç·}ÆTôé½™x3‰‘Y…ãAeQói˜ëaòÐ$u2Ì^n …3ËvA%ºè'ýk´ÃìÚbô‡@[3qd4wr.—‘§÷óeTœk£4‰?žŽcú=ÒteºYPÃÍêŠÔ5¤¯õñWÅØ\ì  î´.U/¤Å– ñ„@‘_"G~¨Yi=sdv/Û™Æ#I>¨eî ¯ ç¼>ÞC‘9ÒÌ7Þ÷œçƤ™?§ ¢ÀU$Mt‰PG5–ŽŽŠ» bΦȟèOÁ{î’†PB|uÍý9ÁÉ%•n¤`|£ŽÊj‡ø«ÀOn¨ªÍß ¦ds=C!tí›÷Jíi4 ÛD§šäK¦…)ªÁŠFOìY\ÜóŒÏ{UQ-}+„L‹Òò úŒÄ±áÂßÙ¨!ð+'¢:N¡ôÅ«)/•M-pê2¸îÊË ¨JæH0Lg/ÃØ~a[Ü*ÆžÅ&4†øz/t=ͤ«CÑ#Q±¿Ë*ÍÉ?ïÌeg§·ÜcJ(µãŸØGë‹;Þ• ¯¼™œ~4ï‚Ë­„Ï‚ ÎåY+­Ç¾óko —ò_ÓðP;T2óÝn–\Ò0¬“½«Ã?xØ8Î;¢£à…ÛUÐlë¾ œy¤ßp %Vw¼„ j( ˜ÐH»ê"ÌW¼Tjq¤6ßšî´VúÜtwžOXåÆxŒ°5Ï}·ä¡?Ìù2µ°œHÞ6®VeÐMpô²Še»_þQTνŠ’ª [“¼åÓB÷ëFfÌ5Â-ùçDž*õ“à²m.Ð |)Z×ÚÁ™Þø: ʉ7ÇgæožP=N#øÃAmyŠ9/Ñmb47#x|ÔS¯”m“};qYVÕðþ›U^Η€x±Ó³9 ›ŽO~Œ¹2}"NÙhI(Yìd¡à$É#K´Ë•‹Ø9ZöéoÓqä{êLý…UQò aÂ/qÂöA¾]Ï&ŽQ´Qúª´NMU굇™ö‡+Ðó¡ãvÝ ï`¬íÎ}O4O…Zà\ЂÓrzMz¢—e¼¦êŽ~8Æa¦&)™[|y¿›ë<Ù7þ“‚•iCJ¢n-(Îk)’!Ÿ7áSO’ó(¯ýÂe©}o«2‰¾V#,V~„[.Xé¾%/JÚ^KOøµœ¶wu~&Ed“æ1xnùì”\X~çé4u¼+ä6/Ó7@Šýè¿ç²ÄŽJ Ã,¨ˆµ~ÈŠè­ò,åÖm®Ð•ÐOÈl‡aÂô/K <`'ç¿°;/‚X'¦ªúiU×Ò>kESÄ™O’->WŒ‚«ГT’fœ¯ëéð¼å»ßá8‰1Cóµ¥[¯2k2s#bŸïaûM/1±o2FöP}@—;¾ž&€÷ B­êƒ|§vÊnð…%Z)€ÉŸCêÎ0X`É›ýCV4H8Yî»®YÒûíË¡uR¹Y"éÜâÏÖ2ö¸GÇàìÇÜÑ%Ž«oãu¿ ©gÂ^ÒFï æ´ezꆨ{(&«‰e«ï1ØFªz~n'¥/Q '$ÔŠoÜu™k1Wù¢&Ó§b`™ ê®·“&¥æ(ÔdüVBEê͹™Ä±¬÷]®§‚U—0ÉëÀÊ ï§ ””z@4„-Œ·}½×È›EµlWÒk¬³Ëº3ó›Éœ]Ùá3æøüdO,Ê t€¦‚±[m^’.Ô[gÚVqs¦ÒÇ­7»¶ž¬é„¢MýÇÛ’Vè‡ùKÐŽ‘0Ãê1TÔ)KÕ ¶»W´D!C™ ¶òœ5NïøRëÜ[©ýn²M¯Š¿†&+0LïZ¹9YèKÒ)ÓXåArSB Wcë±›NÓùæš"î­eÉ•Jt¯”®bÃYÔ„=º¶Döñ8v“p©\e}Þ„¤9µN=a¥~ý®/”é|‡Ùù·9T2£ [ù¬A¢ËÈ„õ™Ü$YF ö˜¾vÞ +«L…vÝ’Î…ßþš*Ш½‚dn½ï}ûpáéûôg1A¶5@X)åŸÇeŠƒUÙJžY]ŸŽÃVGrÌ0àƒ¼üåMDæQmŸ­bÄ\6^à¿ ñˆåÝã6ˆ,vä] Çᜂã9ÐïF?I÷}öëäµ2‹jöÒ .K%„®¼ÝC§òn²KÄ£¤!ôJv¢;ÞÏŽ•¶ÿÊA7™¼Kn¤¬ˆŒÚ%Þþ ³Wq A—æ+n¬‡ ´C£jÉ܉:a¯0¶=„BKDCúå…µF[5„š…IÙä‹>`ˆ%²K?8sžrìLƒ?M–’;Eœîpñ{öôÏ# ŽÔ)±4§qêP¿DÏN,½{Ž‹T!´òÝCðU™Û“D‚§Úau]ŒT¢—2ÒW%Ä´ÞŒ=IÁtƤ£~öØ–dÖDjÕgÐ>Ôè;=ž© L+Éë_î'øM>$bWý|ϦïžÄ)DÍF«“±“}0¿‹›t Ë}£íá)üÚì^¦»O?ãPkŠZLÞá¦{kóyR“µþÔÝñ~,.n ÙR í8h#ÇÑ+cuZVã4ãtB–ÒªœÐïûJk†B°ƒUc“HEu˜Í%–rê0f Ü E°l ŠÂ¸vñ;¶„—úu\‰cºƒN> !“jkWªôD#{Õ$o…Vyø;[‡Ð+â¥UÊ¥z¶Ú¦<\Õé71og[¨ºZÞr_>¹U ïÏ˶!êŽÝ‰Ò`l\d;ìZõ'<옔½hº‰°%ˆiI$"þwtû^lV¯ÅGˆä?XmsYƒŒ•:,‰© —'oÈäP;á7A{MŒ¹¬q1L“1DÒÏ!ÉdWŠÓœ>¦æ¦ú5Í3¬Ä‘'r<ޱ ‘#5âZ¿ì|ÈLä)ú9†¨¼f‡ûhÉ0:ÖÕ¶Ò‘úí.6¬ûêÝf|<]4Ìè7´ÓA_›«ü¸¨¤&›¶ôŒN-+éwè¥ÅÅÏCc€¯¹å ïoFi€e`Ý;Dr‹ +³nOL죊€ç[L¤Åq!oê©•Ò÷]ÒïæÂ6ð‡¿¨-.³6d¸·Ãµ©ûî Ob÷~ÕcØÁyVýí+Ý>æoYP úݘ¾sH2=JþøúnYþcò·]¿ùÀÆéüÆVÔ}é™ÓÙg={RXCj¼(×á¡%àÔ„®[‚l—:”n¿ Ö‘æ»ÃWÚª‡]‘,íz³€¡Ëž÷uµ££®dœ%·è*#J6mD8]ªÏ­cóI_¿Cý|ÕU}ºdG5‰<ôØdàƒËޝ*KY õuKTKïëõÃê3û}¶¿ Qñ°â†³û1ÀÇÄ;ÐÚ3jì1"‚(람ò%-/v !Ìý iìŠTDÀT dŠŒœ:áf*ƒôõ· ½_ߤîøÎDhì¥YÑ:kæ嶪%eY2 sƌϗw½¨ë"dW»ëúÏê[æöìã°C§î*½ÉûÝÂØËgꫬÐf3ß»—<œ¿ýÓ<@H¡³Ôw;ºR¤j79 ¿$4v¹‹¿ÖF[Í©_·“œÕ²Å ©ƒrW/ªÅÖŠlù5qyû¡[> %×ÕÐ#ï«™7³ìHO4AYìÑÆ.X5c¨9û)Ëlq{ib æ×l µL‘¸´º° #¶Æ¢ ž2;:ž…" ›·%ºàf‡ôÉûâòüæòå°#¦\aîå€Ý5·üzçSÞKžè„)bV¶Ò†æy þ¶ïpKÛÂá’ÿÈMDo7g¢ýŽû³\úêm[(ìäɘ¬¬½IÜ»¯XœåJ÷¾ÝrÁ’ñ¥£Zv!+õ2ù8Í›C“Ä<Ð}¹˜À%¤}ç™ÅqÓ,ùßsòÝá##E=~ßà‚ÕSÄT\ÛùT"¸’¶\"£oæZ`“:JˆÅ ¥ÚmãóDzë9Õš0ßcdÊ vÂD[ l%Ï*ހǥÇ~”o»›8ÿ\f¶þ„î˜Ë¥ZVªëY;TÓ† [xGFØgþ)ƒ³ÂoíD¼~Ø¡³ °]ʉÁ9ÉÔÙ¿â¼ÿ'¼a.ªk®Ðêâí´”»&DàÜ\)ZÄÔ‰RÂzœbŒô}«ËÕŽ[]F£,SY’¢u§¿öð8Fðûrí€îŒW‘V¿$6 }QIBCô±Ó~-™NDüˆ ÷jÐ`ó²š÷} ÿyw&K ¤s£ÞÜô¬*vvóåÒk.ËÙ…Wý gcgÔî5²’ýK媯júκٯòíC´ªÃžåß™Ö)ߟÈ!fhUQb¦½ª3VV±æªB\´ˆä},}„Æù8TKUƒ¨—k&öX^?ƒn2ŒsÜå)Î6zGwk“Á#ço[|±ÏF?š2>á8’4ÆÝSûØê*¡¦­2{À úå…Î)"†ØÃÀëýWOÔVú¶L󬥉•r–Ób„µëm"•»…¯x’ÀëR7½œ•¼søË±ð9[cjÃ0u†I´.J/î“‚¤žwU.‚p$º‹"¢Qö9tî Å3%ÊWvŽsÎ×ågŽ÷(ÇC»ŸN®fŽ(MøZ £î¼²«‹[oåöÄÙõNt“+d‰ÿúÈ–…[D™íÏtdÌ«¢sf.8˜¬B)_жRLçê·ÑÙ›ýj°_õºã5æ9®â%nßRA–—ÒŸ½ÿœ›ŠC÷„ŒvÅÕžëx [ íø+b™0dÚWR$QÑvþ=•y©E”ˆ+¢zq°vkS§N¡{Ai™5^ßðU°`:ñÖ]%OËÌaöƒƒÞ9·D»¼A,¦OíaÄ×ì0ÐpùD ä0©kúP¿Ê™•XI§‡ÌN,’ ¦ì )ìsOâÌû_w š+ý¸ÿá’P} endstream endobj 110 0 obj << /Type /FontDescriptor /FontName /UFEQIZ+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/B/I/P/R/S/T/a/b/c/d/e/f/ffi/five/four/g/h/hyphen/i/l/m/n/o/one/p/period/r/s/t/three/two/u/v/w) /FontFile 109 0 R >> endobj 111 0 obj << /Length1 1406 /Length2 5931 /Length3 0 /Length 6889 /Filter /FlateDecode >> stream xÚuT“ë²6U:Ò‹ éÐ{/"½×!%j轋HQé‚ô®té‚‚€€ T)Ò)‚´ÝîsÎ>ÿ¿Ö½+k%ß;óÌÌûÌÌó…‹ÝÀXPÙiÓ@"Ђ !  @UWÝ¢B@ )— íûÛNÊeCyÁ‘™ÿ@¨¢`4Ö¦AcºHàž·$ IÈ€$e€@€(ý7‰’¨A|à]!À=$æEÊ¥ŠôðGÁœÑØ:?x ¼´´¤Àïp€²; ‡B]Ú掭…¸Œ‘P8 íÿ¸Ä ø}u@CÙÁ2üÃÏ Š‚{ ½„¼àn¿8 ÿJƒm³:ÂAéîC ½HÝO Ž‚A±}÷þ3\WÒùûäG88þ¢áàí!lŠ€{zôÔþ`°&ÒÛœ`h€8PZDRBóÀü Î¿ ˜ø{À~;A¿ÌXA¤ÀKw„aH1^ò†aþÓñÏ)p€CÑ{˜AúïìX3Ìñ¯3vþ(¸Àˆ]?øëó¯'0vÃ7ÿÃXXGÅ\EÍœÿå9UT~Œ ˆ@PDZ‰‰‹$%EAÿLô¯üMÿ·Õÿs½ÿH©…pD¤ÿbmßßL|þ¬ÏÝðþYA‰]h€çßûoB±_ ÿ³ ~‡üÿ–ÿW–ÿuÿÿûFÞnn¿ý<þ?Äîæÿ]ho4VºH¬Dÿ 5‡ý¥h]˜ÜÛý¿½ZhV$Ê'·5î¥÷ƒ9ÀÑPç¿öèï9`Ó»Á0¤ü×; ÿˇ•Ôû^ñÂNë· †UÕ?Kª# H‡_ò—@P(ˆ?)vøØ“8ÂêÔæ÷{½ÂB$ÀÒ 8"Q¤¿f*{y»»ÿ~í¡a~è_nÒ”€z£PXþ^lý¿Ï¿åƒùÁ ¤3SH¨l¤K]dÛIòM_ÁÕù\«æy13¨vDé¼ÕÙá ¨cåôÁnªÙeuž#¥Ol˜­æ¢˜–G†­gç¶ÆV[I§Gúßm)×÷±Ü4QZ ¼ð 4 sÅoÆí¼Ç•ïé-EiPH{âÛ«éWßWöù]ôÔªáZµ„6ÙyÙ¸`’i¢MXÉW}Î$Ç5´ 1Ížßõ‰£ã4yï¯Øî=ä' ÚN-ÆX-Š$ŸNÌU˜ˆx½f¾ÍlÅÄ‚DónìFe=ããGLiñü³ ödG+MEëc“0´«w6·8ƒ>¹ô}ØZ¨JeP%}[ L¦xg†$ÞÓ” {öÝ7oëæQ…ª®ƒ6Ñ-üè;6–&NŸ ÷A@Û5uqÉÍà…rþpQë¾â¨§’‰¢}9 2w¼õ À*+$÷`ÔdE#—>gÝ<4_bñ$ïáhQjè ºZñªâ=w {Pþ„O²ØÂÛÇ÷i/EbÉLô‰Ó©„ Óó]:~µJsåâ¾Å…‰zÖ–•Iƒ?Ùg\`TµRÑ".ÑUru&QzÕbèÎj°ãÏ{b‘[RÓ§Ÿ4 ÝÓ“//9²‚+¾¦nú>õ©ø20sIjܺìÝÒÂÁ!w¿gázƒÄ• ™nr£OD¯–›;ÏÔoýÖéá¶|Ÿgç›,ý¸oø’œù…¾Å,I)ƒG¨åS@äµÏ ¾—”áC3_pkÑí¤ýì¤)ôVÛÇ;oç­õŒZ9}ŒÈ ¹ÍEt„ÈX}ÀÉuÎÄ«cP]%ÂûʾÄv› œ‰0˜øùƒöVD»‹õýݺ‰z³(ÝT{©âaýé`ô‘•ÊêSž“uøR±c¶`[X£µ‰foWè‰<-鷤ţ±Ž5î!6eJkS‡oµ~ÜzQã;³ð6Š¼áæŒp­ŸË°”hÑÑçÝ%…ˆ…kí×¥èeû~úîól‘÷ÜÄä™K«–NP^“ç­š/¥Ÿíý>*ü8fñ¦ jO©º³!3²%ïždÝ~2†”yù²!f³HýQB%üsÉŽõåûï/Ž)b¬3…¯¯ª ˾Ù7ÚH’M£äè—àé{Q¯}SQ•}XAÖÉ™·c’\ÒüDaø‚mîµ\¸ãÚ^t-›Ü'ÜØ¤¬›üѼA“ýocèDu`Íðt‡ž/p.kQ:×5­÷¬jôÎcë XrgÈ#Ù+_ßbŠ¡l±Žþ>IbÀª³î-°~Œâþ\‚Íû¸!=N­Ý*n\éYßÁ³ž(ƒ*|—„-v›íðÊmž‘·z„̸ŽÅ…ß«•Û†Z+> 9º Ê ù ©÷™·O}Ä[t¿\]‹çþlª®õõÀxû«-ÇR뎭Λoæ60ù³d(*Õ)‘ªNËèÕ¾¢Ââ¾ ÖJ¼SÇÕ ÆSŸ·ˆJ“]'§™ûŽ-‘útEædQT¢ÚsN_ëáÞ{ëÓ` 7Oõš£R>¦ŸÐT½[&Þ¨¹dÈc|ýâÆðÒ÷k¥XfP?ý’h#†rOW{.%ÔöøÌv¦W2gÒ©Œo;? ´*UÁ¬P8L­¶_¹wœQµ¢p4ú@7¢]!™¸Dðøãƒâ±iÆ.*Ù75ëÛ•üÖb)*|8·Ò;‰=Å‚e%óë±ÿÊøQæŠxÆ?Zï|M/›ß·ÖYÝ9-?Þ»ÏI<7à5{×\ "ZYLÎ’&UEè`²òJ$ç?«%H™óŽ&YgÔ©Ãò†óµrUØRªPŒ´Hòð’EŸEùÑkKhç¹3áÿ žw`«ÿ¬ƒ@rƒî‹®6{§x ×&“£Ü:¸õ®ï¼Õ–‚SCIïðÝkÅ † 6½iöUY)2¥O OdBßWÜÙ|À£¾Ü+ÝstUe!¢³˫ʘõcNœã–÷‹î2¹ûý<˜Ô‹²$ž†þ¢Â7æ4)éY½~Q{¨ÙfUlÛ$᯸>ë¾è&PN•˜ŠÖÄD½ájI'1©ÕÙÉ®)hmµÕ¯=ë_ujbÏ.P5†¯O÷1q|åSM—¾ÆÁ™š»Ó©Sn–wCvéá·Þ<¼ÿšY&]“Îrùð"²å»f=nÀNÆÜ Çd½ÒÛDbÁQJiòiˆ¾ñ§ŒCç«ZvX³nnøÇê¡j²çÌ4é6Ô+¡)ÌþÂ8³›fžSHó+è—`ñ öõ'& -Íõz$Äê©eÉ“dfOR¶FÜWH–Þæ§Ç$Ý\aTi˜H¬”N™d¨5.'³Ò r¤ÕšÏŽºÙ·‘åÑ81Kå I1›%*Ò›èÚÈšÚv­êd†ôm“¹€‘3QN,…½¨k»ïÇùŠO:Û\Hž:¥‘‡ÉæÁùyüiw­ÔpÈqöš3Z€{;k­äÞÏrz|*uäuÚUŠ ÇÓÔ×8J' ]Y{ÝLb\x/íhß²îx¼xãKw@¡,Ñ‘9‡bß&)ØÎ¶–á…€‰HÊ€×fcäÐ÷q6oØ­i.é|jƃ,‡Hc‰ f‹$Å èDþå\AËÒ{ïïû+N‡Ö)Ù[<ÎÓº磰:f÷ò a‡•Ç,ñ,w%o(v»g›…gˆ®¹‘ê7¸ÙóhXJ kRÿ\¨Ý«ꣀSãÔKåjëRéì•0ï‹”Ùp…Œc)‘¯užõHd Ï}íLëé£Õ¿[S^*Š!8ᤒqÙ]%m$+~-Ñxý–åÊëÑûF>,¨† EÂF½“)³q­±™Pãs¿û9Ê#“W Ák'¹Ú9l„ÞèT㫇vV‹šFv¬&éN¢Ûw»Ä–r>rÁ´(ìÊfÃ\sø_–Ï3/ÍSQ».5KÎ"[ïê\ó¤‰K¡¾ç4zMlÑܽoŸ¦…6öñGÙ4‘ÖÇþ´>µ…´rŒÛøS¤©è]FLtøc9z§µ½-î v7,¨udíñÒÜ×±åÝ4²YêU¨òqˆ“'x1ÂÜjz=}Ÿ ¶³5}þ½G*Z‰OŽO±qeýèAŠÐ¶ØÀ9™Qoj´| DûÎ3´çÛÊ’Öi·®F7¹Áhà‹p%Ífé¡­ÅàôÈJkÇÊ›[yÁÃFËÿö†%‘‚g4Àü»…f›„÷ØVT©û å”0ŠìUÚ˜+âð´£ýŸÙ&Í'…ƒ;ßlc¢º]î'®q[#ŠþôúVÏdôW.'M“ÓÊBJR:Dd5P¥2*‚’ a[Š,ÁÅQµÅM¨óLnûÏH6()%Á×Ä(ýDªöÏ?¡E4G³Až×;Ä ó23lš~fð´øO²3<¸x~ÿËÌ>NIH\tÆÅ'&àÐ9~üÖä†ôÆË´ªøµÌ›’S´VGª–Ç×pW“êC›ßéw-–ɰ¿L½0¥‹R½Í÷ÉÙZ3¬› W,i]?ú°ƒ“¿ÈÿQ†Ǩ‚‘‘ µ ç …W§e.}aÊ ZœCy¥€õüµóŠªªB“Íü¢°X§·ÐðÑÚÎv™"f¤ô@Lº f7²rïªÍ™ËGúˆÍ@4gʪBWÜPlØ—~'â§Æ•GßDñç )ó„խ̦ã-æ÷ya½4,R7oUS¯ ÷Êå2¹¦¢2FÎÅWÔʜݞ â@ñmTš§…äf†ÎW™ÚwfÛ€ÕK}¼ˆ{u\‘uEÉúUE—ð¤¸x¿ôa©¸ÐÒ)½CDÏO)_"¡œ{ö¡Ù5\ÚÄ÷áâQ=Ò]ª?À,÷¾®à1Šx ]úâ{ª¦=~“–0LÚ[ycB“ ¶·*Ì`%ãüNG² Ä¡×e‹¶ÿ®^ ܯìÊÂ'2U¬¶,Ù[}hߊù¡Qä>”y$D3`çÇüš6ü‘ˆ§• Ëríó ´¹.S{v™l¿Tuÿsålr¯í;žç6HÏÒ–d3BC¼JÔ5,ç6âèàˆ‰›®G>Q“¯‡ÔÜÌŒ¼=_kÑÀPj‹)ôpù_ab1N>)ÆdïNÒM]ä? Y}:Ô1&d`Wgã9ˆ®MÍâŒøœùº a nOFÃCvXNçXúØ«,33²¦8Þ^Dh,¸ ?išY¨|jçºÄ•ðXËÛrް¹4úãÍäÀƒ&'"·³E ©GÕxFÌ­{„éК¤æ€4žFŠí€ü<í‚ꩤõ.ê2Q+ö…Ãd’„óJÙªpßBàü ´³~E9£&m6|{xV‡Öª‘]®!—CMß' |Bð‚tv "`WŸ‚×Ký8ZºÈf»©»Œv-ëÆ^~s‚¼OȈàwî 5 *'{y7æ7gú·ÖH|‡dÈiï<·¸O:LùåÎË»§Â·XËÙ/ã/û½5Þo—-wã\¯ ±fîã%ÓJz¯ò@OùŽ}2FÅëµÐáì¼yLO7ænμ¿H0UÛänIåû®Y¡©S”ަ+Üã÷Âz»‹°ÀmAªûª]Û=ê >A#ý‡‰»/%ÍhF6IÖ$ƒÌÑå«—%I’î‡Ç3¦jñóOžýÊw³â¤ àe»à—2®·÷rbZýSÚøÑ16á =b‚‹D6W5Pzrí®ŽªÏW­õd×èÝ!¿Žíx$˜é6¥"OuÊ=[‡û«¯g7Î*~û×ùC(uîì qÃÖBF¦FL™¿ºð§ÒXð\5’&úL›¦|´˜ÀÔü$¦!!/­ ùáö– Çz%¸«Øev¥¡\2xÒ"eÇvõ^ fˆÃ•ûæÕiÁŠÝôhë@Ÿ™¢±áöþ^[ -NéZZô©œSôæ½É/•fL¨“cã`矖Р^¢?v%×%žÍ‹>À\-k§8©g¶OÅÌKеé\ê~Ë\dñ)ð…©i6Š‡Þ¥²¿Óïõê!“ï¾ý“Ì‚Ôî·+²77W¼x*“8 Tf:ÎH}4Wü½­=$=LÞc8Ç”óB2çû‡±Ÿ#"(®îÀ–W½X½ªÕ‹úWE!ïÒùž\lfù¤ñ»‚‰ÅqÔÎ$Ö’+üÖD1Ÿ¹^ŽV«”6ÂÍ®^ç_¹ô(­n.I3xÁªçgýô˜ìÛè ÄaµÁÇO Ns¢QŸâ,e©ùN‰‡€G¯rtvöÑš»öâç9¾B9í8Z¯•G%^ŽjÈ̼ÞÅß»ñ¦U1û¡¹¥¿š.RðÁÞEfçÑêdâ ‰Ôœ•¼L›“§°ßØôš-j/ʰT*Oj ÂÆc¤ÏUv%neN”ȱÇ'ï4-/ȧÉ.F0CsFô•Wæ,*Ï4õ_ï?2#(i‘§ÆqõQMP(b@ˆÉ3p–:_ŸÎ§‹}7ºf+_=¾úŽ„ÙzóB•³ô¤*§dªeɹ’âKyPŸIwYL¬3Ä)sqÕWSP-Éðè¾Î©¹é‘Gì'+ùÃå—­ßîqÄÊõâ÷Î`ú«…£@d/u/¾>¢àd_×Y@›œÛ4jà ÀãzP8㨵÷Av÷ý‡:ZG'É­š‹˜}£’Òel±©Ó>Ô+V#ùNï÷3ø 1æ;ÄÂÎ7/+½÷o¬£æÞÄÑI¶0ìsk9p»ÊŠK›mœ]ôcÕpZViqL9CI¥ßqG~„&o‹¯<¤ÿFú$ÌôÖÞ<·É+ºÏº2'‚ÂýK¾°Ó]}$!œ¯]M5Ž»¡š®55ñ¼—Ñþ´µ-ÀŸ;üOô›7ÔAu>ŠÚÿMP} endstream endobj 112 0 obj << /Type /FontDescriptor /FontName /LBWBDW+CMEX10 /Flags 4 /FontBBox [-24 -2960 1454 772] /Ascent 40 /CapHeight 0 /Descent -600 /ItalicAngle 0 /StemV 47 /XHeight 431 /CharSet (/summationtext) /FontFile 111 0 R >> endobj 113 0 obj << /Length1 1814 /Length2 11177 /Length3 0 /Length 12326 /Filter /FlateDecode >> stream xÚ·Pœé-Šk‚… ÁoÜÝÝ5h 4Ö¸Npww îà.àNpܹdfÎÌœó^Õ½ÕUÝÿÚ¾¿½öWS‘+«1Š˜BÁ’P['F&f^€˜‚‚ 3€™™‰™™ŠJâd þJìàÚòþËBÌ rz–‰ƒœž  ¶Ygk €…“—…‹—™ÀÊÌÌóC¨/@ä1(0d¡¶`G4*1¨»ÄÜÂé9Ï4&´.†?Ü"6`ˆ È r²ÛÐ?èYgü7byŽi vú—ú9¥É߈ã7z¾"þÑÿ>@ é¿ þ§üçpæ¿/ïç ÿcò\ä_´ü²>û[ƒlŒMÿIÁùœÁú÷@þvy^ í¿às»Áçøÿs>t´9ZüËà9£Ó¿Z~Žæö/øÍãø_S6qvpx¦ÁëùLÿà?®a0Ø l‚6? 5á °¬ h»®!reÜãGJ댥KšÀ-V]îd‹6uÁÂ%ž€ml6¶—.ä `ý~äËì¹g¢Ô ö•a¶8ákµKž©ù9·',ÀÁ:“”êp1W›õKÏJ·u}ù~hW¹G› ÇÑÚvd8”@–Y²ßAÀÔBQ¾¸ùõÓi\´)L±böçÛs_¼Ñ>h šK±oøíÇÃ\*@ͮαTù5ŸæT§9gn/:œà<ÃÓ/Á–$%Yvæ‘U¶†Ùf:Üô¹x-[8<Øëçíq…ø-%aL /ÆW‹4‘/ýpUü÷b£ÂÛDÖΛ³Âh§«‰fŽ^´ ø¢0ñ”üH?8g3'zs–.€A¦ÛaHödü5|ýS”X3õ¶Ïo¹áT‹‚ìˆn«!#­{ tº}/é¨CŸOubÊzÚì®ØZ>--x!îuñe1#¦W'£í¨®šÅò}…¿y1R'‹qGsºl)ÈÓ¨Ã"¢Èø#íWÿ]jfËøŸ5Ú0\î׿h¯’-Õs©?záØ=2Ní›Í@w³†h½DÖìš{g»¶ªgk$†`áÊZ:8ÏĘ-S#ÏmdJšhZäª#ÛxT”U6ƒA¸qlêOáfúp »¡(Òæþ+Dœ¥»¾±[–úÉìiXÌC­úÖ!;hºø]úçø7Á!³9éÚ[]2}u‰Ë/–©ª¡Þ@ú;ô­-¼}@j2Ö²5÷–‡öR¬]ÍyVZ†4¹æÕ’¯iyƒŸ){nžzæ</Ãæ­ î[¹(º¥Y>zñ×ÈÚá•8/­øù*Ó«®½‚ û>cÎm–™Ûæ19,ê@!ðÄ·«OûfÍ ü?-åî#§)­š¬/W¨ Ïm(õ±Rx6¿úgoã[ä‡?¾YºIVWfù0áûê ®mvHä·LÉ…Æêí#mfHAGÙåB䧨?uu¡ igðl®Ý^ׯ ù¾ZتSÏþ>=VuM^çE &M7Ÿ OQCZvIˆ¿à¹ß5axd;í,hçú·PpùœÚ;¬™ÝzuáìLÆÓVq#¨©Nk'¸ããð§;Ö±Z&(´­û·`Qu½ˆÒÈh \û 8Ì.«,³®n3Q•yåØ-â›þø`Ç.ŠŠlà*^øS¬rnŠñ~1Õ™4Xî¬B±{aã#®ñím!ßÝ‘öŸ_¼€Ÿ¤(—3ÙÛ1DC;¼¼W‡N äÓ¿j¾uÏ÷™k÷ ‰2EcŒTèSã“Á¡Âúž— £‘h€_GjòËõ“‰¦ a<'¡ƒÅuˆS ©ÿr©âA«¾ ´~É~ö$pB2êÒêâ¾Xia@÷EZn¾ÚSïü»Nˆë«ú7 yl tõ_C]8¼˜J x ñ«²(8AmuØ`½T¯hxÎV/9æcCÍ*¼±ÖpäÅ`g—÷pó©•hì§¾GëcAרí}þ iˆª>7œöW^>[’Œ”iiƸä4õVâ*ªÆN|µý{–ùM_3¬Ð· /Ù˧‘ØoRÎWk—ž,EèҌ̧‚UJ«kPi ð ûV©¼Ô ³3Þæ»îƒáïQKêÒ„¡åÞÞDß D8À„%·µžW"cO™)¡h­LÀmg;ß0/_ÅDXc+¨+9.^ NßÔûŽê[͸*¯£ù¾Ôø];Sw„ÙyÄOPbƪƧ†\»·K>º£á:§?F#©ê2m l}«.M<`;õèžö¬Ñ×÷tï(‚°” d#óÔ'Tê§líy4û.+VP'¼‰—ú¦Á"\Ääã럔Ý—ôˆ§,ìðÑø”ƒƒ·ž3ãt³_Ë&AuG«qõªö“–ÖÒQÝ.'XÅUÞëT»CÝ&,+Ž“ÜòjÔÚ“áõÜëimñ…Z25wýdIoèɈ¬@ÙºxK#fÉÕ>8Isf2t˺f„p 8xlÓ´Tm¿J*°âìK+”h;õß“‡’fÈö IÓ;¨\Fד¨p\æ}T'0–÷ ŽqXá\ ¨¸æ*=ráqÜzKÄ›0Ç«aˆ'¯,Sß?„¯(V±£¦pù ròíå‚`vßøÕ§íÛ¿Øèæ,ßjõdˆïsï†QžN{žf’7J†®„6=¾Ù[ã3üˆfª’(+CÙÐ’Ø£3@óÏFÕ~÷ÕUž¬.A¬·Ù,<CâgÎÚ |C–Û¼ioWzëA ±¢K”? úÙ¬ÈL請5ߦ·VqšD¿Ûìš…ÙôEÈRDéødI¸ãËžù¶ÿ[&‡”*xÒ3 ÝFø²‡ÆÃ Œ]ÿb_öÄí(‡ DÌýµ)ÊÑRty8`hôÞ&Q¼ ÀìʵQæÔ‹3'³¾£‘/Zìy‚gjKM5{ÖjÍÞOÑšFl«¶ÖæÌ¾D«‚õó»V9Ò ›°eáÃ|‘w;s]ï[¦»pP(UÁ¨63âºðå]vÞÃ?+~† #i/ʼnº!M¢‰¯oZ‡P?@À_7_¤À—ñçTuì?xv3}H2© ¶¾fŽÌûDS„·ðQ %™þx”¸‰xþ}Ó£%€ú¨³ŽR¢4VR:†v%Ÿ ±ÌiUu’fMÉLåfEþ –Âëó÷ÃpŒ±ÕËÊ[~RøÇQY÷Ëa)¯Ÿ-¾ïÓZU¦5CP¹§c®… ð¯Ý¤ž˜b‚P0”®‚R߇[VnJâª/µM¼d ø¿Q]¬ºu4‹¯ËËÿ®gãcbÅ­¦ßùˆ<²³ŠàóEb?Øø$ŒA¬Âþ¬‡:èAûÕL^ZßL2‚C»ÞbC·ßàÓÆ*jä»Þâ`Šo't8>SQ™nöMlì5RuÉl³â¥]ÐÌ™R®ÿØ€H6ͼ¾Âá6&*îåR´oùä7¥„jëåñ wâób‰p_BÊG¿a“£æa+µP»¨v‘ŸpEŽèퟌeGã­ÂÚšG¤$,f¨w…^ûò^­U%dl ó„‘'Cõ„åã@êIúªN臘•ïÞÐ}¹Òà b”˜|³mU7ÞXîäØªlì¼ogïü|(çfÏÖS°6hèTDRž¶=á9©5Gõ‰9oD`G"F¸²%žé„Õ`Í¡¶€7‹†DZ° ¼» <²\;ÃWyõ½.'ÀËF+}ñW½ÂºÈ˜‡£‘£±WtføüCc“0®aœc.)sSòSg¦R¬D÷ç$ëã° =)#cØK)w~j#%ÍÊé€Û«JûÔº™‹“Í31DŠÏO.aM–Ø0BÉY·KÃ"$Ó;LÅŽ¼úýD|Ò¬í5BÖu5&¸8[0ÊWçä`”¼Å}F'–ÈïÀ7Àþcþ „¸¥«;àö‡U¤½ÉxhÚXö«€¨Ñ>øÍÃn)*ì»±¶Áƒ¾F…_¼Ó ·¶Ý”&j>D䥱™-Ì ¡ Âä‰ Þ[â¨Óo¹ù^|Ì9­{:sD É2د0û™Sç»,t³{0çÌa¨׿àýüÏ[:®2ètÞs¶Ë)VÂ%eÙœ|3HÄ‹y#*ÀCyü«ÿ€€2®‘3æâvÀÂó‚Ü• \Æ·;Œá3¥ÕËTlê}%õÈ,‘¼>† “¿"̲ÕPá© Õ,µ3ÒTù9¾kÖî‡y¾"¬«»þê¬^ú’#«½Íñ£P&Ó÷¹†ÛK¾êÍì~XA]YÝQEй'3XkHVk_M×ܾÈXHôû}*ÐÕw¨Çvy½l9–ñ»;Jc]&¬xû² !åq¿Wµ* í[¬‰óƒ{³†ª¯Õ|ôóIýPÑçQßΠÏ6må™lÍö^š=•É}›.í‘=ŠNtY#¿RM°`ÄSªGÞÄ¢ò9Yͬ0¦e쳓P5 Jßœ¤¾¸^>ªy /¢”ESä8hVME³tg7&¾(MÁJ˜ÌhŠ.wâ]b¼yÎF.à,q¦²Ÿ„Bs ÎÏWkYõu¨ñvu^fû†Õ—$·aõN¤Î~§¬‰À{ïÏÓÏ«¨3دÎÕ…d•'"_÷Ûúóê˨¬=ΙSAÞŽ?âÄ•ÅÙåô–Ow„$ÞþRµºJ Ø_·z»œu+¬–}ë)‹Nût.­‡@„„ëÅ?å¢!$ÎOë3œx¢ŸŸöœÐú3¶lceOj¢ûK"Lcª:完Âåº{L*7;Pijª±óm–€ç›>D8ħ—¯Æ¸­†H!¹»HüÆ7 éYͰh›YŸèIù^¦!H-~üèÁÅ¥æ`GTCbu±5*¼‰R_˜RðùÔˆûv¨ìy—|n\¹Aš‚þÂI>˜{²š9iè“çøô~‡)íªýÄ6–d‚j¢R\ž€[MAK/bØÂËï…†;½OqXôI•SJ<ÛA"^#9zèëî˜ n†á_>k ¶Fu­æp*¡[Š.lr  üØGl€‰!ÖÇvF&Ù u; ¶aêwáö#¼i? R \W"xå4ùÝ =áì‚~Ž«áDzÜðZ…ÿœ€ÅÅQɧÙ_¶vÑêG’©¶ãm¢tdµ¿©•–×¶½»ê?麺À9cºË'„{©q\¯*æc“”™óÊqFZÜ/ÍÙUš2ùÔÄM-ó°ØG‡1`5f'5‹¤º—„þ¯6¶œÕ"5*bj…&÷ˆ<±üxTã›Í¦ÂzT××ryhWíâò˜tÔ< Î-k-‡`~$öÇk…!Ã7úêŸR¼ß%»T’Ü ©¸£Y®Ndæ¯Çü‚žìÉXe"…p^§Ž€uìá=€ —þD$û&ˆk¦ \Šþ}×p–hxõjhç¾@ õˆ¯:×gnïÁÑo“§Öú:­º¹É^7Õô¶7YÈÞ5~É&Î¥±Œx2Þ¦Ãðt 'b0™ŠcÆée¶n<©ªuä×lq…)ëg™Š¿Ê!£#‹’$ô?Mb,îY¬½Áþèåð(Xú\kZ%ñ1înˆº×ï;R_QþõGùf&!æºeÆo› jFb/`ü¸ìÁê©ùxEüÆèvvN5+SUNêZJFAÄS?Ù™”Ü(ÝbÝ~`ÿмÖn"Ì.6ç–“Â5ë!¬“¾I=oBUú×”.=ÐDñ‚”ó U”Gwß›ýì]ЫEOÛ¬8ZA–Äœ*Ò’ÄSM‘¦üduäôÊE[¿‹¬ô\…e³cL:¢ìÉg’ÇP¶]¦”,ü %ˆZ?i{Î#R¹Hi…†ãƒRO&yìÔ(Irø)Ç?#{ëÞë»Ý:î5Û;‡ªÓNžHXÞäéåfN5i€3ïD=jÜRÝ5n‘lÐX.'§¨)xa]‹>ª`Q†ÄçÕ£Q1NfWù2¾70ë3N‘k9Baå˜Í(ÕшÒ1ÑðG0iWsFWÌô:ѧÛðLÝX“ŒA#²¦HDf•sç÷þÇ 5ÚúV6³M d¶>7j"wbDÙ÷Àôc}™ßw«N«MZå£ÝCJý„P§¾È~Ãî¿s²Bˆ9’×#­³Òã&ÜMÊèè¬áª™Ö sUåëC,“w$8!¼4×M84OjWuo§«Ü^ìYT--?%øs+½ióø-2ȸÈTaó"‹UcÎã!*»E>UÒT…¸ž|¸³† çЧá ÚrqäñVï5ÔÈî0d›ë–Ó ‡ôçËÕ—‡4ŒpÈ'> ™›´Ê±ä}BT–Ía”I-zg¡é•ôùHšò³I¶¼®àr0Äw@ˆò,ëÎ>ïVÆ^ ü¶Ãšf¯“ó‰¹SññÍê`ËíXèæ=Ê»‰èCBqTÉ!Ë=t¸áÑ¡ ‘ôæ—ßNͲ{×ý>rÊ­ïý0^k=Ø¥*ÁB›dá1"ź~_º£¨§4ì|±J•ÔËÄ{À¯€m 2®ïæ°“tÙÜX˜|£:*U–;—„Xê4úö]£oˆÅ^$éoÍËåW½¤ªàr#q©³¥ýLÜ2QB=Q¡ØÇ>S:jåÙpJwMt’·ÏàÔ5T($;'RÅ[ñ‹;™ØÐ²Ò³ìÒkN¢8 ˆ¢Îäúu!Ÿ/oNHî6í&U²tJJ4­…äë6N˜ð#ÆërŠ¹Ð…„:áüÝáà¶ÍºÈJZáùí<¡¹èén«ˆ)R (¨+-Y}îæûQJ-UêxÎðXÁMž8Á,{» J ßìöZwùš%`]›k]jèÇ×P.\X$ÒÊ=F%Ͳû!¤Ÿ½ó·Þ›Š)ÈQKi:C”ªnbOìè‡çéßëÕ3ز$ð`I>¨ÑÐv.Í!tövª»Þv„_Nàê\…Êm±£ôwË·•ï±YÚ/›V ë©™“5’—×±xgÁu´3ü$¼o"Ìèó&ámæ®V«Jàq^j¼”fdqI€A-ì\·+Z¦ 3{øKǧ:eò {)ªjH£û-ñ_g_ÛC{¤«Ì©üõ9¾„«+}KCW(ó!m~FÒ};nt}—߀£mÆ…â¿ Cô $Uë5QÙpr(Ïä“g=z]ªT ¦½ÉðZ%zÅðQ8ø3œ“¢eËgŸ^bp“`,}Ü#‰7œö9 ߉f.¼‡g¾iRrn®ET¶Èèá–ØŽÀSÃ,Ÿ;û¸¹Ì]ï•/rD¤/a÷KÃRS÷V÷œlÂ}q¸ƒº}¦šùnU& eò°ÖŸ4Zdñ°bËèT噮ꥡ‚’bW2ÕØ%›ºwgk2žâ”ühõÄ(3žÒ/["pÙâÕz06…ž2ÜtH´g‡5”ËtÕF¤DBƒ¼7‚(²•ìq.j¸ €š0²•éÙ[²UüMŠÎRg¹¦µÑН«²\Ë*`?‚xÁ{FŒô>”z–VFùΞøØ/NElwF>—§|õ{þ,hT½…¥ÚS¯ ­PÉÍÐêmÈÙo úÈkV(_.Ãá9uâzSÐöÕ l³J«Æ7ØÒóÿôrÏë>Ù¤U ÛZBá«`êA‰×T›ÅŠEv¤Añ‚}«M”&†é–ÕšÌ5Æß:c®ÕÕ†‘Pìä¬!†VXÑ_™ú.šJ>@oPSI¬±ž¦Å'?-êŒíQ„µ6G•Ò¶4üå§yjŒB“26 qi®ìB8­žpiswù2Ó½Íum§¡ D—ᮚ÷»šÛU(?IÚ®ßêW»Ó³–·ifòÇûòµ7ðÓ Dê"ÓÑá'ƒ[º/……¹~|Ld¢¶­ÓåÑŠªX 3M~Ne=‹>” ÒšÚñòµc`É4›È–¸×¢[lJØëê®8d|GuXûÛA¦ù¦»ÙeB¨rôÝ-örúRecPˆæÏ-W4´Àb^Éhú%ŸD%FÓ²Ûåà‹tEkw•â`êBÀ!Ô¨ßYôµÂqñmÏ Žn{ÆKXƒ½î¶§lŽâñÅ »U¾“SK4RC-ŒšC–ìULE~Ð@3Hnsi¸ÓzÏ¹æŠ¡Ç =o+ï  *êÏÔñÑGxe±K½¡õv‚oó‹WÎΊ<ï L5Gb/þŒòñµMÂ’0#B䋎7’±¥èaÁWó{Tl¦¹sk›„åz¼²úïp 4î½~<É6žÝOEç×plDvéŽè}?97ð¥Ûç2(»‡z ïŽ| 5g¢ œŒFiE ÚåÂÛ¾iåó 4ðÙ¬°%70ª½­Ñ€ÕÂg—VÛÓ–nÝ×T.¶¨¤²h‹øÀ†«Î{r7îÚεn•º †õjÌ9ŽQ2@£Ñä†à™_y1_åçßàRŸ¿‚“æÊæÁŽIæ°nlWÙà:µ÷ÛèÖG´3i%ZpW{~6d?צMãQª°6¹ª¶’’â‡Ô´¼C ·Œ¨ä¯eËöøæ«œžVDGÁ¥0uj¾ÑÂÌèLñÖÉý< @œ.Ý„Á¬œdáÕøcbôYì/ùª#ˆg#:;~Â[ itülÔi®­¼<Æx›ÞW ±­C¼”uëþ&i†#k #,uýdîíD(f ì«Á:¥ËȤÀ>¢EJ[„{¿Šp²M@…fHg©LØ­š·}c@ûùÌžÑ]¢M¾&²vFÐVÓ`p!•¾ä½ÙËú*YDÙBÃ=ˆ(7¿n? ?~Õ_ M]6zw]›õ8‡A´§œ<ÏØûàpñ³¡¯g{Xlm×Q,_÷'0¢,yŽ'Å_"Äòl5¶ýKÈFá‚p€ ²’œÑ1€¶à§Må% ëô”TN!Vœ2’•åžÌ™GŽÉ@Â2X.Ê—O!õéõ fê‹’`7kÀ] ‚šŒrzTs­¤€±Z¾*û²÷«4»ñ—Éô²ŽŸºìT£¬©¦lÐ8ø?zÎ)á¼öïMb‚ÉñqÔg]¼ê@rþɼ‚ ÙÊG]‰Žö»p©8W$©¾L1RÝMs°†%3Zßq” ªW­\_q8&òœ<@÷Šj$òm\fƒÔu=úkÜBZûæìb(Úˆ ¹•Ï«j`˜ÎØBêµ[Vì]Ü¿V:ŠÛñšËã|£u^{ñ2½—†Ò]ßu“·®h êPIÝŠEÓ¦§QãÖ²Ûõ±¦„é‹HÈ,6/Ô" žÓ¼·s’OÜŽ¶Qü °S˜“àH¬/c¨ N†6îÝ÷vê”õ/MTx:oš»×>Qœ rË\?F7•‡ñ÷ÝXûAƒx¤-.ð}O+n'…瀈¾F=ók(˜6QdÐîðjöËN*›QRSÑs,5êO;ܘŠÈI;¡.DzÀÍ‹¡ñ"%Ø×ÑŠÈ“Æ=W¹“a*‹´Öƒ$*¯¨#ØÙë•ÂZÛ°£ÃWk»¿\ÄyƘ”Ò‰ø<ç2) º pÏNk{) ã¶¾T–*!Z+¼ÆÔRo‰ÕZ0V)ËmØ|ZóGg9÷(ïî•¿&qͱeÜ%­ÇïËT1ÖÀ4Áí$—*9À'N$;çU!ÎlM‰G¿~Å~z-vн°bK¾meÙG–n†qŽìúŽß‹‹'ýd0œB;ˆÒV¦ü0ÙŠ"v7}šAç¸@îp„ðæÝÇÖòÅûŠ9Ö);™o¼Ól6q5?Ílö«eü2ŠÑñÛPæ¯ýµ|$–@Q©.ê¯UIøõÍF6ᘣÖÖJ„”®F6.z£Bw-¡êºÂètÁÑò£ŠÞz€ý K"uÊÄš(¿|.MÜKžï 1 zÍ“{Jûjfï3aSÍ~>•o)øÐEH÷}Û\ÀÆAðsšÿ’—Ÿt8eèÉF˜#ĺz¾}Rir“è.GŠamG7+‚°\14Åy29¯F¾L‚xªj2¼/-ñt’öð]£Ëpc2yð„¦Àš1NFüÒ`å!»ü”Oþ²¥6y{îÁD¨B‰4«4Ø÷é[ Æs"¤‰Ù½ÂA æMqÛ[hNe JgK8A‰qÒEs<…Þ^B?®X(–Ù‚k2½Nâ²çEŽE>5 Jn£“G8pðÀÊ13‹–þdßÜ -Ç;yQY_ÞîFn}=u¹çƒ}r;õЏL?%ÚZQRækúª™Ré7½£øF&?Tuò´Œòµ^Ïô®Rifë,G|Ϩ}MBéôÍnÅ]V°c¥Î ÎýÕOF€TàÜâ5FŠ[ò*OÎf¯~Ë›éG*©ᓚ€—9‡Bä䌾eñ%9‹ð¿‹X›>;o‹Öb½ïo­KÓÒô š{þä& ¬{’ëã«XöºÚ%éÎÑ‘1šCš_á­D8çCÖµMu$‡“Ó¸·øbèYÙóʼ~ÈD®o5ª“ê4õZ&âš«]ƒÂƒO„÷iJ]0 1k9CéIØ÷uå>&—ÜKÑÚ­ÃUél¶Në´f¸Ï)sÌ ß–[ Þ¡°ÌÕIN»×|oÁ¸U ZÃèü& =§ùvgòkE ^:/7¸‘DFa€Ì¯Ó\Êìy•=qW“›2©ˆÃ$N£ªÊðõ•Şpä“Òg¢Åø©¦v°ÊŽÙ7e¾«í„+nÞËk$™xj¹º&š 7£”+®Òéër44@ ö©Ï2µ½è/$Ä’:à¬k.,ZÌdKFÃÕõíñ–¾]‘ó×°áùä¼­™¡*Í¢ñ9úo;OOvòg{|6úÄz×·.5þ«øzÐ …™ˆð=|ÉmÖR7d¹ž„Кh^< V1qß1ïZ'^ÚÌ+/ïÎŽ¦í©æWå!B.ÉÍ&9i5$\ÑhdêÄ{cB¦î ó­mSCNÓN“Æ÷º} !Ü1Çšu˜F™?²Üï,_ÇDù%(p#ÛâÜ¥yÛ˱àÞ®Ýû‹õˆ³F&wÖÈ[ÞÇȈâkÝÛ£o•²Ú1Ñ­  >âÜ,Ö‡V@¹j1…5Íè9‚a¹±_ȶ5¬ç`:æ²ÊÝÚ­‚ ·1Áš½kõl&ËR½%ý?˜€ endstream endobj 114 0 obj << /Type /FontDescriptor /FontName /XKKIPS+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/C/D/E/O/R/T/U/X/a/b/beta/c/comma/d/e/greater/i/j/lambda/less/n/p/s/slash/t/x/z) /FontFile 113 0 R >> endobj 115 0 obj << /Length1 1422 /Length2 6370 /Length3 0 /Length 7330 /Filter /FlateDecode >> stream xÚtT”kÛ.)][Jò¤»;¥»‘”b†éîD–F¥‘”éîRácûßß9k³f­wž»ï빯ûa¤ÓÖ㔵YaP8'/ðHCCEàááçâááÃfdÔ‡ÀÀÔØŒ†`7w *ö¿¹Ap„NGøiÀ €ª‡ÀËð ‰ñ ‹ñð|<<¢;ÂÜÄy'ÄÐàTaP°;6ã#˜‹ÄÎŽ(ó÷`±fxEE…9~…²Î`7ˆ5 h€àö`gDEk ³†€á>ÿJÁ"a‡»ˆqs{yyqœÝ¹`nvR¬€nè‚ÝÁnž`à'`@ä þŒ ›з‡¸ÿÖëÁlá^ 70€P8A¬ÁPwD„Ôì Šz*ꀖ úÛYý·ðçn^.Þÿ¤ûý3ú+dm svA} P;Àâ´Õ¹àÞpµùérr‡!âAž ˆÈ áð«s («€ÿÀs·vƒ¸ÀݹÜ!N?!rÿLƒ¸e¨Í#˜³3 wÇþÙŸ<Ä l¸vîß“u„¼ ~[ÔÆö'n(ÄÕ¬"ÿÇ¡ÂþGg†‚<"Âü"<Ø{[ÛsÿL¯ïãþeäý©F ðs¹¶àˆ-ñ‡íçòp7p€ßÿ6ü[Âæål ÖpÀ lbÿ“¡Ûþ–Ãwƒx¦<îñA]DaÄ!àßYþƒÿoì¿´Ú ÈŸÞxþI¨µ…¼<¿1 .ïožhÁògeX—Є!¸ Xþ¡¾ 5âÃûÿ½¿Bþo¼ÿ™åÿEýÿnHÑÃÉé—™å—ýÿ0ƒœ!N>Tö€#ÖB†X軯²ØâáüßV8±²P;Å9y¸x~ë!îŠo°6nmÿ›HÏQà kÃÜ!?_DÏÙ[gíˆxUÜûm¹#Vþk¸?e0bÉþ݇Ôfósù…›ÈA„$øñ"ÖÖìý‹ï7G„Ì€-Ì ûç˜yynÈOÝQàvø%þ«Œµ‡›¢_ì@ôð·ü냽ÁÖØ3“0kñP‡êЖ‹JYJ/Îõ! ôÃÌ‹Ç|œC¯-0á] £O–“õ^¾˜U{«8ÓÉ«háðQSÎõ"gaê›ßZ m­À'­â–­UâäÝòxŠß9ý$ARžQºX±[’6u á',i».=FæêŽé/&â7ŠÂªa-š>U¾y¡NJ%j¸·¢ o´êÕ_üNΜ»âœ²k•¡™E¯&ñá;Ù«‚p©§¯wލ“º-晪gðúÑÃp8‘‚x/•¼F…á\>þMn±çnÖ_~VJ÷evmºJYÓ/ª5Oì×NvÍéØ\;sêÉ;1÷ Õ•’ŸfåÅ÷%Ë &U~K2›,‹€.{ó}”† ÑŒ^ ˆmh <ÕñÁiû:».pxÞ6î‘wV:ur öÖp~|á^Thh|}°–Ò-‹s?QÆóš7ü¯+ NŸEg¸†£GE€÷¶7ßaºÍ–ޝ:I®9tI&˜ÎþÚ‰³ ä+N-ì¾ø]ÚÑñ^6Í’õó’ðõµÿ¨‹™‘ø#­¥w<¾ß€9¯jŒj¬ùIÝT–££×Ò|s•«1‘‘rMþcÂhçÓ¾èQ¼ í¾—Æ²oEUÌPÓ“$’?AGe8j|zUpÇÅ„SÚQ Þ#Ïìé ~€Qð½Ñ³¤òC/ª”Ö^,À¥q±Š›¿É¡eôãZFñ@̉)ö¶²S‘¿’šZfQò§$ÇÜM²K4S&B”ƒ!W&ž]™OÆß†*•PNz”b´&“Þ÷WËð‡­: I懚)>'ßyùìåPp\MÖÓ>I¦à!^æK)϶;Å8¨(ÜŠº–Öå5$[Љ¢Y¶Þ÷kuÑŽGq—X¹ŒhÊAãá§'oÏô‡°)[ä²ãcgÄðI¿(ì;Q OUôÖ$Œ« `c›Û;韉Ÿa.øÚ>w]½ð!/».”Á{ÍKôª+Ö”µ¡…´RÎ ƒ¦rö{ö"¥,Á™Ê0õñÉ™/â½­×ü>h²[ø á–ÁJžÌ÷²+ÿRøœ¹‘Þ·hw˜TàÐÆY×zaªÑžòDÝïæ »Å¸»ëñE‹¥Üry=ð‰¾^a“|eë†ËÀjÁJá9ÙÅêú°a ¹Öp %i$©CòãPEƃï·\êþÖÛ¾F… Y0bùÄܳ8ŸÌ„ãHŽâJûá)&v5·ß—×¾èU(WyMÓª~LÊ¢` ¼³ŒëN&&ô¬ÍåÝ”nÈÕee6ÑQ­£8‡¨4Åñ°a‰juvhª; 2°±fKy9•\;…ï}-¬ûÒ°Èjÿì=Ö(ºÔE+Î ÛóhÕV:דµ;¢튠<Ë÷ç›Û)¡¸T¯ñVfÛ‡6ÉA‚oiÑ88ë:Å›~P>,L|4ÛÕqÕ~P0íadîÊ¥„lWÄhhºŒDG‰Ü[Åø’Tßܶ „ÍZL¸z[8ŠÃ Å,n™Rü’·¬” %þÄM‰ÂÒ{T6£xA©æ‰‘Æ?ò'|:äÖ::¬ž4' üj_ŒÙ¶êªTàGp6=qøÜŠiT Û=¤6Œéµ‚Z#÷´œgZɃ{:9Eµ;]ùÞ­=¤¹dêŽóöøÒ2©…"Ô¿±ŠÜŸ’äGÇË‚(6aƸaQÇH/wJSYýÇWšä…âóL½Ž…Œ•>_‰3ðkm-ºÔ‚€“ý³\í>RË w/ÜÉ^nÈܰ ½ï&ÅÔ[©OÓØšré+ÛŒðñT>võ$º&mô}Õ@5ÄdHŽ ¹{“nMñÏÆ+•Ý€—¼>ôÃw¡$m*ÕD}Ÿ­_üúBaÐú"’w÷J«A)Ô¿1AÀ ßØèdéÕ©•Â×÷]†ê1DPæ 3©7~ªkÕÈ›×Ø!6·z§²ÅC !êj|,X5z…âM~ßêç(Âv\pF#Ú[¥{Yø*›ÓnROïÝL 9o¦´Wn( ê¶y1‘<¡Ç3À!3^6©n›H×íÅÄ Ú~)1€Ô~óqÆ9Ÿévßè°¶.PNÍÀ7ÂÝ'Ï´Oß›wtàRcÍËL*G¶Œ÷Fª…ȯ²6¦‰%¦D‡÷Ë: Y5& òh›Èx»Š·&­ë™NÄ Ò´Â_´k¨ŽåÆîMMöÞÏ0´ÇVG-Œá‡u?C¶,W¶.H‰Ž¢|ÿâJ³%åëêºè:êݰK ‰ÄìÛ ñ88ˆ”Š9—ðKbŽ[oBîÈuÛäB`ŽšÜ=Á|’ÍÀ}+tùn”}š08šæu`0y› WF7Vr@Á{×ëWS»FVþ÷-Äï£_zdL“çw Àø¢GMFh h´6„3qàðÉ ]òäk·Ì¬ÒÝ%6NüÛÈ¢3ßö\ó¸Åý·ßÁ¢™Ž¡"¹çEä­>ÔW|Pk•á¼ÿôvyA„ú¶Ë±b.ŽÕf¯–£âœ½Dâž7ÊHŸ¿@¹ÿœ¢·/>ƒzÌV ‹/LrUäj‚¯$/ÈØ`>–bSŠº§Ž7jÞæ3ð Šr…o5b˨'„oZÀ.)m™€Æ/¢>ݹ š!AodYz%Á`ß!¹X»‡7;i%#ÝŠæi ÌG<5û¬:Â$Ó™?H„íÎyù‚õHyÝÓf¦³l›F1´!ULŦÔÐh’PLýp|xd·(ô¡ –Dº_­'Y¥þJ'ím§Y[ÓÖ;ÃJ<>é~jfù#Ðé2(3B¶àA•¡&ZVºV.JSléšöˆm±õ)‹Š¾¡*Ux=X¾ë´lqiêñ×pª›ËÛØ‰ËŽêç±†Š œ–Ý)&’ô4å=®šâ<÷]{æ_UŠ=$eH`WjËïB¡Št>¥‚˜\öú†«)‹×ÜâZìëë¿Ïf|5×Ð?fØ®e3óÞ‚´ù>ø3 j‡xÁþ§ðˆÈ‰Sœ½/ô~CË„žÇ0ÑòÈžt\“—Ó~œûbÓvºQÆ ·®! u›±Æ ©Å}tN•“áÚW˜¬.+`¿©|àžJ 3ËVöA_ õjEö‘Kã'žƒ«Xlƒ^‘Ø·Üè$If™¯Òš3Óˆ+ç*J¬¸4¹ð?öØO£‡•-²L¶×po¾=ã¾N<…\¨b¡>w>j›pã·íZÃLbo¬ ̈Bgs6SIŒ£ž} vE<µ:5>ÅFB0y7Í#õr½Ä×µ³¡õ¤\ÆûÑÜWUþâ/žæ•ZÃ|º•s¶ã‹Û7c´ÑH&¦3â¹*Ãs’CìvXh–g|ÌT®F­íG¦íðÜ «{6ƒîH§þ¢w}¨XN»CÀÊ.iB„ Z{’i†ž§F.¼ÄHÌ‹Z ׯ×`ë<ÕŠòìå€ÀSÎÓ3à´kULý#Í“ùäbw¹yþCГo2“–õ÷?cz;'Ò£Së2é?f.`©4kÔmŒ‚j·*°øßÑ ÷4þ”²Ï-;Â=éJóç‡ê jöø˜–ªè3 ;Ï|ue¬ÖUG'÷™”>g=ÊÁÞÖ'AOò ÷Ÿ—&8×ùF|ë;>RQO{(ÿÚ™JËÄl‘¬×c|ôδ¿þ' 5*„pÛ]M™.ÇT¦#jØŒue &’UÉò¹¿¾`±8VHß½#@ý­^*î×z–s^¦CÇÊ©¢Ìæ`0èæz‹zÄUio33½G_Ë€™S~w$¹o:Ý!NsQ÷CÂtѤWóõ² Ï—~×¥ÄíáÆRŽŠ/?W8 ^œ«ÖÖ^ õÅ E;aˆrÅ¡¥«L­úîuÑÓ²MÂÜü96ž)ùWYzÅÊm# ïÓ¦5#ŽÛ®¹ùlWP™8l­|oÇ’«E@ë» 1µûCX³Ëݦm™ƒ#µærƒ—R¤ÍÁcO‚irTî=Mw¸ª.ÕMNˆ›|ŠAÛá[¤{¢ŠµÞš"vµ "ƇwëÑÈJ%4ßÈ}ÜzmSOS5ÝW"Î?ÕÒQ©Ôsp3¶}s»”§ži݈ózç¦óɱ9‹-'’¹ïèÄñÿ̰(‘Áü ›a@'ë*xmÖ˜ëM05ÍŠ÷'ÛRõPè=ër9ÑÍ tÐ+8Ëh¼Î{EclŠñú“ªÒXÓDú¶+ÔUé™ì‡I ‡¡ß߬‘0IxçFÝ“rý”„׎ª¤˜TC³1>4õ½´úLqSZ`†zéÝx­kÔAjt ¬­.h¶•r+aÀï–ýÚ‰¤enñ¢kì„6<ÐùHÿ/»…pò\ã¼F‚Ù–ÅœÖtäÁïyB®¬ét„Ý:3ÊTû_cÍ= ³öm¶Ÿ<ý"W^ÎýôY®L¼±d^_‡Àþöó _cÞÜHÚ$‰!Æ< ’_LW¢®Ûª…_>¾!\Ÿ-‘ c‚ ’BúÖaru±ÎÈË;¨åÖCžÓH?f†§sƒòewÏ÷¯d}4óÍ63ZµY58ÆMõBÔTcŽ7»÷Êö"ŸF$¾öM9³/Á²x’Gäð¾)|…Þ ¬ÁÑ=<äCŒÆ0+Õ¶ðÿà„W>Oë€[ºÍËÞƒ'vg,Ìuócd.ËS…Ô ŠK YÈx¾Ôf¢Ž.ÃbÄCtÍÙôSòœn­t5µó¹ìâO úÏ^× Å*<¦Û¼¦â¦“ûÇ‘2Ç?ä„ ¸~}L']¾V¼óRp?LOäuiôøšÃ_”ËãëËUÛ({ç¥Î¹U&!í¦ªV˜];u.U.Ò.ñV?pW'?¼Ót¬‰u¾î‹o!Ù(AÉ­¨Î°¦ÁË·9Ž 7®‘VÙUr[XŒŸÓ²ä#c®ÿÐ&ÁbŸUH­wLÛ’›Q>)ÐR(Q‰B±M;½mÁuñ*KìÔ°Ñ:W;Z†D®Òžîîår±ÂmVWß E<¡^±Ü IŸ¿ÅÇ•n²±„¼9~öÛïªë9kª”=ýI&’¢´Ó‚äp_£Ól8ÍÍèÜm—9Má ¯'òçÅíý‘E+»ô7¦1™Ï)w/Ž î›ec:Ë lðÍÎûÈ&ùE$‰.‘VÎï†*6!ö˜*ðOÚU†<1G hŸjs ªùbö™ Ü·äå–‹MR={57¦RsANÚmÀ´kF¸f6Ez xUüÃ?êÝú»d2YÈ,p ÚvrÏ{´˜®<Î9Q%Þx Gvr…G¯òizC¥cæÝÝ}~®üµÏƒBÓòR½GÌuT‡ÉÝEõœãeM¤k¬%§'/Hy…ðù.Ú>‹õÓ_68³BŠ• ”aVÊ1=êÇâì]Þ[ƒ<ŠZbëâ§xNÕ÷‘í,Ÿ;LÕoÅúnÚÇ'$uº¶‚ å¶mªvŽskavÆröBéRˆJ¿ªƒ¢ºO­ü!BIoœ¿žšñü¨9gx¶ôÂ2è‡R°'—6ò*n>éR=§&·8piþ-ö+'»é“·¶VLK=.¶.:š†e)©×̃ÜZ´±h4§bÑ5ÕGÕa‘˲26Ö#'ùVþ¸‘åŠÉæÍïÙI†Åt¬íèó5Ì«wÑz-ˇhšMÄO&6¹×êæY·Ò!äöÅ ²À»{eFÒ†˜ òÙ"›T¯vÍf¥Sïø¸Û¹\P÷ÆéÏ›&óI+^ð/>Í9cSоʫ‹ŸìXZÆä{e1&Ú;…#žiÝ4Èð6mûÇkIýn»'¯àK… ^aØX˜Baìz:ßÌc:ço ÌN^L¬.EÃiË4ö…B¹8œn1mSÝ8S…aúXÁiAðÍc ™tݳÏ/ r·qDèiàGÞOÃÆÌ ‹(‰G¢û2ïA1A²Åœ@a¼e)ŸØ\yØ„¾‰!¼o#É›)ˆ…©ÔZ޹ +ü(%æÞÁ0„%9@/Ò¾@ÃôYÃDf/s¯’û|m=ÿÕ||=â[ê›ÊËYîX”7.|b¶fßéó¥ fÙüSx!jjÜ!EÅ㛘=‰1Ë1 ³Ð¡Ç³ªSÐÍ}è‘y>؆RA`Ýô=Ù0RγDç&ß5î…ï[’7o2¯{EI†a¨=l“ŠS~Ö³=w÷Õt†ÂaJ=Ú%hˆÔŒÇn]Z0̪Xñ‡ÝMóè$$åìq É=âãÊqZK.>:Èø3¯Û çÒFÉSñ6e—Û¯,_"¢=ƒ*©ËñO½ ° êúUyí _³‘žB0p­öË’ï*ÉÕU`벓Æ}±U)4BoÃl¥·ê–wÜC‡®{ÉÏ]V0³œ¢>ÛÏD gñn4œZ|öq’3|üÎG¶œôcè¯&á0<¶5œÔÌ)Ç[pè¨IØs×ÀœK ‡»ªÉìBŠoU¶‰bSϳ¾²à›uÄ;Ýx¶PÒŽi>èW ·sÓžpý0\‡aÉQù¶Eò;™ÎÄj× |7Þ\ýí%#ƒ#%”<… < ;k^â@JƒÑ°rÆœWmK'Søç’̆LIÔŸŒ¿ÛÒKMkРíðPx=æëCž}v)§5ƒq. pN`e¢ð/1:A<ÀÜ7¦hr¼äWKËØèø¦pé¨ÜŽÝr÷žj˅а`/ ›o>OÁÉ”}ô ûj‘bÚV^Ú´´¤AÛbz'„÷ ¿æ\Û†õܧ/vþ@׬_Nümòò[œeþUì’J¾œ-©«ª–_­Gü|}¦ãp”AkÊbbäFQø`¶i¹{{9­•_ ¥Vª÷—þŒrÂÍ'¸—BM®7½éˉ8ÐðØû]’xÆXê ˜±e¨B‘&WÏÆ»ÐC®Yqu²R—xi…¹ Ï Ý þ)XIHŒâƒèÃ1I,kòóôd\‡å9Åyo)årÉÌ‘L/¤:[ƒF êP>bµôQ!ظGß³Å$Áž•ýŸ)d‹Õ+NI+ÜîvÛsËñæ„âÒ­rïÛE’Û×ÎõR w1ùsBI¦Ùa´÷P²]Mâ· F9ËÉB: X5Âë9âLküóN–¨“t9Äû½,e6ß@á-|åDÍ’ý Tbê‘–t励Ëúx+Ѐ 0 2¤¹OUbàùØðlX @­VÔþ®Š~‚1 ³Zí ‰É¼ukA¹Ô NæMÁ~ í+$pçˆ\²?>a^¹ˆùGƒ{NÀÿ><" endstream endobj 116 0 obj << /Type /FontDescriptor /FontName /RDLHZT+CMMI5 /Flags 4 /FontBBox [37 -250 1349 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 90 /XHeight 431 /CharSet (/i/j) /FontFile 115 0 R >> endobj 117 0 obj << /Length1 1496 /Length2 7665 /Length3 0 /Length 8672 /Filter /FlateDecode >> stream xÚ´TÔ_5LIIJ— 1tÌÐ%ÝÝ!9ÀCÌCƒ´tI· ) €t‚4R‚„tJƒ/ê?žçù¾µÞwÍZ¿¹ûÔ½ûž}.A[[Æa UDÀ‘Ü`@NCCEñó€@|¸@ > é ýËŒ 4„º{Àp±ÿs‡Bw6yò.N¨z:Àü°XX ð@¢"ÜÅò/˜-@ƒ Š€C=prW_w˜½òn›¿—6vXTT˜ëw:@Æê³À¤ÔånGˆ3@aƒ"}ÿ«›„é*ÆËëííÍqñàA¸ÛK²s¼aH€.Ôêîµü" Є¸@ÿ0ãÁô`ìz;¤7Ä ¸38Ãl p» O¸-Ôp·9@OE å …ÿ VÿÀøën`ð?åþÊþUÿ ±±A¸¸Bà¾0¸=Àæ h)ªó }\ÜöW ÄÙq—ñ‚Àœ!Öw¿O(Êè wÿ¢çaãsEzðxÀœQäýUæî–à¶r(éûë|ò0w¨Íݵûòþé¬á ÷ÿ ØÁà¶v¿HØzºòÀanžPù¿BîL¸ÿÚì¡H€ HD˜_„u@}lx•×÷u…þv‚™ïú»"\vw$ 0;èÝ®¿Ä @º{BýÿÓñß ØÂlk¨= Žûoõ;3Ôî¾k¾;Ì` ºÓúõûge~'/[ÜÙ÷ßðßýåU“5ÖÔÖãüÃøŸ¬,ÂàÏ pó ‚`°0 |·üï*ÿðÿ›ûo«6ö×Ù@ÿTÛ!`Ðw—÷7¯¿dÁö×Ȱþ{ MÄ–¡¶¥oÙÜ}ÀÿÏð;åÿO÷¿ªüߤÿ¿Rôtvþífûíÿÿ¸!.0gß¿î¤ì‰¼ ÄÝpÀÿ7Ôúg”5 ¶0O—ÿõª !wã!·¿“87X€$ðÇóP„ù@mµaH‡?Bú»w{8ÃàPm„ì׋s—ýïnêlœî^»ŽýqA<îFù»¹¿0ônÈþû p„í¯iä@ÜÝ!¾¸w‚¸C‚ðÝØÚB}~ëÀËG ïRwœvwÜ_m¾¯î/Ó$à5þ‰ x!ÿ ð]¤5ù €öPÀëø/ßAäoø_¶ñtw¿cô[gwlþÆ¿ß(Ôjƒ»0‹°w¬o?¯•¡ñæ^“¸·ŸynÌÇ=VlP˜²\~¦——»¨öRq¡¬háØ¥)ëvžÿeîØµ¾ÑWà”›^ñ»=½uÒìÏSÔ™ÿ3ZÆY”B£tÙ‡beî(Út±DÝ8R¶öz@ÂÖú¾ùoÒrEaÕˆvÍÎîÚò\urZQÃ]d«õþšà %kÁŠKbØ·8³è ÍlF5‰Î ÜgÞ5D_‡‡>÷Óõƨz…®0?E+ˆÑÊ `Õ~*"¸.(óÚÎ~ào­D"«Ìn:ÅžRž´–3`´Y3첿îùB¼7Øqûåu®ÌçðF:!B×\+÷o×öêŽC>©žÐÿ¦DÁlO|Å ÿp§øH¹D[ãÜ_I0X¡}¤w ‹dãÌÑûð%0TÍ ×஽o¥D-Ù!ú´Ï²kÌ`›júòIlŸ}ÊçÚ©‹’˜ ûíÇ·ú¶R”Ѳ×NÓåC@ÉàI¹›ˆ÷ÜQ2EómÅ"˜–ñ¤­¼¢¢Ú÷òt_Óßw>´{QÇ}bkJ€Su³T­`0@sÓo9IñÄî3}Þ6Ç¥ãiõE2þóè'VmlNÉÒßÙ°³“L”›¿W4EuúoVHv{ÌÌ—ˆ6|9ÏÐaMö[ÄݯhŠf­~°†î–¯O†3£þC6Q‹;4r&Ûïçz[¬.ѯm ŸyDó°æéèE>H0Pñ½ÇX˜^[·XŒuºŸd‡Lb©7úÙ£ô.EëC‚å*ìÍA•+€¡)ù𞮞…IÜ+•O±œu¤øK6°l¬Ž. ¢ª%<ȶê¥YVXÓ’¿|‹‡õEþ›"¼švšEGílÅ¡/FD±×ѯ٭ü=®‹Ö '±œé`5“Q³ Ý®{Ñž_óZ¦ð¡ù_Ñò/WìŽ|~ý”˜‡µÏêB$r?l·ë§»Äáòƒ˜€—üGòb+†Wµ}ÞÌÝò´ Ûq–Zo;DØjÓãKŸ¬f­¢¦és%„aÊ26'âg•O*1ø´ý)‰ÞWŠÁ6Çq5]“¸]×dLžO÷ÒZ{¿¥–fÀuÇ>̦Fõ«C›}¶»>lÜb‡ÑÓtðiöÞÿÔâBþà¿fÂç¾õÞUQ¯»Æ® í¬h† ¬¦\ÐZ Bœ¯-ù–‚²—]ŽÌœÌbÚ åÃ=ÃFKîO¡K®f‹%?ã.#&섆>8µ`N+ÔÅ-/n£•µÇ|vNàð¬CtãÁŽÑÄRµ6²«–ŸgþhÛG§<ÐôN l_F pû¤j©ðP÷ª*gG³„xÏ"”a/o %í÷Xn¼kK÷£"fž ÊѰÅe„–„\EÎ eyž– ðÉFI  ”O½?v«>ù&Q]í3*`5,ƒè+t®éJÓ×Ĝˣ úÞÔ§|ÃIÞä.MyšñæÔ0oÊ ¶¢ÐˆÃ^ÌýP½¼NM)ïþJ‡žÓ±í~Kù±ü¿®: >ÁC\¡ÑÖŽŠVÜ ª—U‘¤éP´|]ƒÿÖva0-ãtÈXŽ+š—™$‘žþÀ‰²o\TQp¬H#³Öý5ôKdWpYÛ»¯3¯ë\ Ö¤ —+#‘?ïsÎ$”%DW>é:&øf!€ÿ†‹"sŒ1œ)ÎÐÔñ£dDßM¥SŽ™£N„@øÆýœ~¤•ú5¹e„a±tK Æµ Hb›nK  ]c-ƒUÆîS«Òñkk¥kåËò#ÈòÞnY»ýä$ðùŒæk@¾à!)Cº¤ÅƒfÝÜ-Ë`«AÖס¦6‡^ÈШ;{aVDÄ}*ºMž#NQ:¨ƒKèÝ¥t¬‘X¶CŸrARseΉÓȆj?.Uz?# úBͯò8Ë€ýùԎÒ'[ /ôE¶nZë÷ÈàúËF‹x’Ñóh¨Æ{Óí×Ä꽫3G[œ•.ëb/×ÃÑ@Ù胴<­Ïb>ÊX´ÁÉ#Pá É#ýáÊ%“·I¸> M Œqµâ©ˆ58Ɉåëá›ÚŠÅ„ÕvŠPºÞ×vû]è5ôª±ÄZŒîIô͈ÅŠ!.Ÿ1Ebf52Ÿ_ÁXeðw!߯N[}}}íÄ!e¡gý¨íyÑJÉwbš’7·N\ì"E觇§ð6½-å^ߣÒL›Z7y ´ °Êžž=]\Ë%7„ í¿:}¯[;-•»óÁËÍó>i€Ìë¹V¿¶AÍuùZRA2 Òò«ÓQ•ÖP4¥çÙ‹Ðw”Ø}_ìÒkæ2Ç)óÞÂ~Æ„Š•Ðt|ôBìJ]«À}9ñZâÍ ¿`N—pè@l÷MÊwˆªã£×cyórÄ.“4¶9±gš]lÂJ Ïnt¾†ŠÃP_¸ú݉¹í—$ôEØ7GÓ+QkN=$Ÿ¥ôœ–?±¡]Ý=Þ™gzc{éZçNeÏîøP3ã!k©D—xº~ÞâJ MX>÷¤q)ò&hV‘!.¢eW g¤È”þm '·ö}ú«›;êE×¾E_±3üœpVÜ)ðöÝ‘À8¨+©–½aâËžcdrd¡Ž#Úe9í™Afç¸I²ÅÔ;I\ÎîՇж)”ãÀΆ˜Jº÷b†¾Û+hÄ®¢ÔyþÔyæjBîÊ"l*êö‘X¡²wsÝd˜Ž‘¨ÁgéO¼ô°¡ÐW_/ßÔÙ-ð®(&º¿¼ÀN%Ô5ôºN¶89Æ8x…ƒzú> ¿‚;ÕuÞGi÷f½’0>+HŠk#Ö¾ËožÜ¹¾<%òü`‹ëùÍXSKiv«L³œbuVº‰K~„Åýé8’ÒyÙ¶¤}I;sÚù9UâçSÓrDJ7µž”m›tȦ|åÛ_y,Ãï«]a‰®Qѧ!V VÝ턎~gᮇ2>wRÔãô"ßM÷rf:e:̹Û rž‹¯¼Ô):/œó¼6EÌ5Á^ôy¿ºÅëKeg~€»%yŽêf±fõì(ä²¾…xý kC^ ºÕ1ÿ-F쾄 Ûú*¸rö#ý“ÞM¿]GïsÏW&¤æp­[áþ½I‚²Ðbf"Ô/ia¤ÉzZ¹ZdE[ÒÉÕ]±MohV}‰éüÁ»Š•ï±ÒûÂØc,ÕB‚UyÎ'¤é†ü—Á©ßäù™Ñœ¨¸B@¥d=9M*YŒÂJª˜‹ZÚä§®ÚŸuÜ›äzaUYCóNæn²OèYJÂH£^Ýä§š½ÖÖ72 dm$x»E\”+ðËK„Jì³µ~&ë¾GëáTTüqˆÕ(Ê~>/Eê>™±¢Ð”ÚDTmÚ…Ú·ò,ÿ´EU’å+c´lè!¥žMö\Ì ÕÃáÀ Aù5yƒôv‡™Mµ °»ÓÛƒylvíSl 7¯3qrû^Kæ­MÇEPʲ>ž¥C÷’Cêa¼ÓQFdWû§dºæÈwrMu΀“¯‹+¨c?³)c/q¿ÖË$ á$Ê2<ÏÚ@úJoÞxPÄýz4hÓ\hÌ¡@–Žþ"ÒØ)/ùã -¦·‡Ñ·Æd«M ³©XÖ··›#» +Š&³ìêDd[fí"ú?•±±£4Ôd1êÄZOJS¸q>³Ð~-$3±êIÿ²?ð+F¸¶˜Æ=ǰEeÕ@£•îâŒeA’àÝ@›Ê mn)ö¶¿Æ’½fM¨ÑþÒa1À+^"ÇäÅký î[œöâ¦Jí§¨žé L=ÙˆSß¿°¥¹˜÷¦„â ~’ £´Cò”i«³òâdÖ²KŒ$Š@É¢ŽýÞŠñ‘¡öáÇL2ìu[Ô¯Ìf)b©ÖÍ^°<[L×ój;½Ò¨eèlŬ°Ö_ÕaÝP ˆÆ3w|ˆ'SÆ–2ñu`gÞ?4O/Ïl­¼ž!ÞðåRøzxñõƒáÕÀÄjP4ÁèÙÀ1Öåª÷'N‹3&ò! ™È<_?vCCYH>ø~X/ñõÒt|•Ù—n°`±‚3–íÙ=7ÚÕÍþ›O“sÈêèçÈœ”¤ÒvF"^d8¯FHJõìƒï7‹?.p¡ïè„9m1hèŸý°fŒ’Z°±s,ºÄ5ÿØaøÁáwQ(Ûµå ÅË8z1+ÿQ©ÒŸ²]WXsX,…m¨˜kŸNNgnqD•ûÝ øŽ.¸7QÀ6¥Áí¶òÍ÷‹©äÖºuϽÔÖ.a†_.[6.ZëÄÝÒŸ=$Ä^G&]k…üRŽ3vŸÔõîØ›¸µ‡¯ín$&èõЦ7.’JØ}Ø€iµ¿ß•,1VKPßçc­aŽÎ£û|P-þ-3 `•ä„« ˆ¯XÛƒJK)ëèœC 2w)ñÒ³±Sm‘fŸj‡ñÑ5ªˆ'?ÅÍ›6‹ybžm‰3 «\ªõT9¿gO&£W0©cfΫ>ò÷¢û3¿#1h{rUž[ûÀ#À4útp¬™åµKθ~ýÙG®{q“F{Ë+³Èå‘[­8â_­U\±¯?lñ7¾pn‚Îzñz/¹<õÿq0¶m5þ&ðMN¦½bO}?Û —Íq ãX¥TÁ‹GéOOmß+ÐæÄkGëpèr0ÏÍØT]÷,.à÷ªDÎãq'õçïkO¾g¼‡d쑘+HSmõMÛxËc¡‘‰{aõ±ædF:ñ '̺¸ÚíM¬“/RýÕŽµ‡qa!(²X™ìpˆ3áa%^VùkÂA ™z엀ˇnÎÞ%OŠ“yrsG' î‘;Ó0ê ØC¸O»ªªJö^ ä~Ä!Æ»QÎNi=Ö{ñ.ŸŽï;hvïµüðÊŠšØç£k¶“ù¨­pã3"üÊÆ:þoóÜf0AFçÒÚ~È\IŒÕO£7GjûÂiÀ ä¡ñ €î»02£ %uÞ$⟘Ãñ9¢/Æ¿>¯”4#uʵ¤Ú{T¥u©ÛXuÿº¤½ TJ-ä‚7¶Æ8´ä\þüçGOFJ÷B™\«×¥4ѤB_Jt$íNt‰Ë~Pë÷ì©oì”-4’Ìøc{Р•9&¢°õ¬WT¦P4ó7÷A¡4³Ô9y[|ÓÙ©vë4&íßÏñ>™µ0Úöî¿>x©#oN™Ó÷DZä [*ãí㩃<2Ážl5FÔ)AyýÛc\XF¦'µ;ÃSgE.ljš cJ;¹oíiÆb÷sÓÎÁiïÞ¸´,ÜNîÀ ‰Rpi¶:[º G¶]ýã;nÌXT„m-T¦ ™™ÂiØ—(ä- @“¢ï ÁË%Û¾xÇZ׃Á†$hcG¡ÖS•àÝtsÑÀW‘÷ƒî?å;%Ÿ©LÒ—!©~QÒ>6|÷°î'¥žŽ¥#ùïqXT·žÆæJdïüBd€v€Ñ)Ë^N}&wzþб|<åyâÃqgR¦Ü­",QòÙ` Çãî -þ©êÅÚ¡Óm¸U[•þ¶G¹iS3ÉèbÁ-v½fÍÒ‡ ¨ã ü÷‘žeø÷ù´tÂÇ T'Ž6z™¶LåSó]ÆžƒíÀgM!}É÷Ķ#ö9vÉI›|AøÛ]{™Û»˜·òæ\w™íß¾fX(ˆíj¸ë Ó¢ °¼¢%Y .ËAßjrX=ïµ KŠ¾è –vhð„šÖöDƒƒxŒkì×¢¤žo|/wÒQžÒ$1lßѹ슃è[‹ ®i´T«ºÒ¹ ½Ùæä¤-¡€mê­vé9^J‹Îf™}¶'=þ˜¡u¹/9K‘-Iƒöô1þÕD>)s¾®i&AÚo#Çm†obþ¢..½k&CÔßðÌ©‡ÏŠût¸9–FÏHŽáóîìܾ`©l™·îuå-EËU W Pp ¸Ãø]:T}Ou¬.Éeß(Ë?M䢇=§†Ù@ŠÄï`Îaθߺ%Ù\›{שµ-OŠ8?7_oo7dÜ•18¢“_뾞&~«ŽŽgA텙ؓìs¾wúÍÞ S=e¢ª¸ k·½ÌÁ"m›"”ÑÊÌi&gçë`Eㄽ8ì|q:Dø­ÿ¦p[Œ’ÄY³¨{¬¼ò´ Ÿé{Õ®ˆ1]íq‚%ó…þ¥ì_Úà‹ûVÑÀÙR+zôˉ©´Ó[ûÌr^Åæ7©Øb±IsŸ„Þ+Ï»1P“—Nâ¼V\I¯¾ï‘ó>o«_¥ãD/µ†Ð;›µ€a"_[e#3^I]S•)bô~£Ÿí4تí ‡ýðÐæGœBƒý0f÷ÖE¿Ñ(åC¹Ôã~W÷DÇ£è5lJëáIï&Ѧz¼%ߡĿtßÊë>Ðát·Ü“ãÁ;GM·5^ú2–B ­ž“%]Ë@>)Fy¨ï1âU|†/éƒnŒ1mØ,©²uYˆ¨¯‡¾¯3Ç”âOw¶f‘žµÜþô¹¥!b¾ˆîÓLúJ•G²›ë°/aÅM(p›î!(™ºöf‡ÉªWÎvzHj²†n-ô·–<ƒh¤}šufõ{zÁ6Ùž -sdù ©r¤¿!‘;Qá–Ê‹‘ö{š›#¤aÜ.‘°‚"£B¢$ÞýV¹•zêÐÕ+t¡)Ó"‰)XQ?bLl-´¦_¤ú݉mƒWg§¼¥ïY€(§Ao_>^{Û½£^;ù5jSx1IÜa¦ü¤Šzvjÿ1Gr u4怙HoBtâP2¯ôcÎôužYÛK‚ÎÛsFAp‘gê²GËz4ý ŒdÄ4÷Çäã²[m0)‘ùH+h…›åìÙyQ¢b±nÆi\ò¡Ÿ›¢1t?1¡PHL½úJÊ[ÜîñÒIX÷ÒX#ƒ~ ßê{c]HT`¬^òc5m\¯Õ#ùì~Äi‡æ7cºxÆHOÒÁWÓ±Qntþ[¤•XÏZFì›ÓQæJ%öá®/XNGÈDÉõñõåמZy7¾oxbFH7âë=~X2ÊË?'­Âíղ䶦"z»Fù̾G…"2Á‡dªÿñ‚»åÊLw#kjDi½h°½d%ÛgôÇJ˜[).5p¤ ç\õuI[f«|B»ù¢³g$[÷õÖPTØãŒõmƒƒ~Ïhæ7G÷À`‡Â§¬©êa÷î/$~A6êôM=u¶7ÿdYÌ…7Ù®›yô^c…*Y6Y ÷,:Ö+à-á¦#À¸ˆ°aÇ=X\(ÄI½ã$JJ=Ô*ÒEUS¦÷•wý´_¾þ&çjh2]‡tŽY‘nÀƒE/ $—AÚý½w™þ潇qvðŽùО ¥àûÔ•n£Ë_µp˶U†ŸI–¯Hx'Wœ3ðZ!ßKö<Æ‚žC¥tLW§ß)‰U–"p’W5b!ob«êR¦øÉüÌhޝ.Ká½q&^ü!©h¢Ê𸙴BE> ¸ê¦T³Á`¾ñaMÓhp5¦›<«8y®ã”|ŠõYIÁp{1aBè¢l ¯A©ñöIÏC&¼!qþ«‚'y¢gbklïøwMG˫ҷqèÕï”Ïo6èÙdÊ$Ô¸ %9o`‘"ƒÀë Hç÷;Sh›mÀ.‡V*ʉhû×WS&MÎùNò¶¨¡ÄüÂpRcúÀ&N±Wh>oÊ,ðìðÈjÝÉ/^D¼äÌÏ_ õ1ÐGo¯˜@Ѝ¡Që€Ñô œ¶Œ‚r‡“]WMžžX¼t­Æ%²û!c]RHßÛÅDË\¬Û•0ÊÏ/1“o˜%–oâI\Ç_ݸ~ÝwZ÷:Ê8Ј|H[°EWö¥ˆŽ•9¢½Ê9™'tt~¸ƒ°I¿s¥ê‘ÅÈ(|Ê®‡±µ3~n9K`¹~žœÕ¥å¶|ªú8bZ¬ê¸ÂÙŠ8*{ä«/] õl÷5eBÑ}¬,Q†²ç?àŠ­Š ‘ÈÝ4ÙÛ䈮BÞ]Ž<2"ûØxºýç89Ç Í_ëhSíPZFÌEVAÅ䃙ícÌrji‘ÚÆR ÓÊ »~Âé×»a‚1÷#–‚ˆ2&Ì´œë”ÐJHP‰TbÖ´fÌ>äLLÙiØ`¥Èaa}]ÇvÑK÷zÌÏ$nBø´VslS'Ë“ñ™ˆpp“¿.Èañ¦–q™}·éd&àîiÿ¹4#Î’ÛÔb=¢‘f WÒ¢cx¤5ì®ìdë§U.0ÐÀÒ£º¢_$82°¦3èƒkލbíËä[ôð<5ÛÓÓRsjìµ ~ûSæiôÜÕ‚"2^ `Dg+È+Ù`ášÍ¹Û½r;£b ZT~?³õjW´÷݃Zû`ê¿"À¦ú^…‹#|tµëöÍM= ¿z™çEÉ¢w±¿emV£¥ÀI‚éþ¥£ÉéüňÜÄ*–´àža\sp ®‚¸HpÐ`.;gw¹A…¦¥F䮽"Ýæ§ÍÏMúû'h¸-]ö§I¿ÓâÍÁœ¦·©+“MUÌ©05õeYõ O †´îÅ€\õwu:oÞc%küڳŠendstream endobj 118 0 obj << /Type /FontDescriptor /FontName /KBXNPS+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 81 /XHeight 431 /CharSet (/R/X/a/beta/i/j/t) /FontFile 117 0 R >> endobj 119 0 obj << /Length1 2743 /Length2 23018 /Length3 0 /Length 24552 /Filter /FlateDecode >> stream xÚŒ÷tœëû  6lœ‰m5¶m;4¶£ac³±&il§±m›Í™î½»Ýÿï[ëœ5kͼ×m>ϼdD Êt‚&¶F@1[':&zF.€°¬#€‘‘…ž‘‘ŽŒLÅÂÉ øŽL èàhakÃõ‡€°ÐÐ D1tÉÉÚÚ¤œ­L,&v.&.FF3##çÿm¸"†.&Yz€”­ ÐŽLØÖÎÝÁÂÌÜ äæJc*''í_êAk ƒ…±¡ @ÖÐÉh òhlhP¶5¶:¹ÿÇ%¹““ƒ««+½¡µ#½­ƒ-ÀÕÂÉ t:¸M¿ÈZÿÎŒŽ  bnáø7]ÙÖÔÉÕЬ,Œ6Ž g ä ,)·Úü-,ó·-àŸÚ˜è™þ5÷ö/C6)ÛZÛÚ¸[ؘL-¬€y1z'7'Z€¡É/AC+G[¾¡‹¡…•¡Hà¯È b‚ŠCP‚ÿ¤çhì`açäHïhaõ+E†_f@Uµ1¶µ¶Ú89ÂýŠOÄÂh *»;Ãßµ´±uµñü˜Zؘ˜þJÂÄÙŽAÕÆÂÞ()òˆ÷›ft°122rp²€ö ›±9Ã/ó*îvÀ¿˜L¿È  ¼=ílí¦ $€Þ¦@М§£¡ àäà ôöü“ñ_ÇÄ0±0vÍ,là~[‘¦cPó,ÜÚŒ Ùc0þúüû¤ /[+÷ßâõ—AD\ILQƒæïŒÿå Ùº<éXtÌlŒ¦_CÆzðþ¯™ ð¿äÿ¢*ZüÜ%mLmœç*Þÿòpùg,(ÿY*À=ÈÙ‚f ü=ú:ŒlŒÆ /¦ÿÏ ð—Êÿ¿¹ÿeåÿmôÿo@bÎVV±)ÿâÿÿ° ­-¬Üÿ²³h-dmAËaóEÕ¯²,ÐÄÂÙúÿr% Aë!hcfõo-Å,Ü€& NÆæÏÐÿº2oeaT°u´øuØè@ û?<ÐÂ[‚GP¯þbAûô_—¢6ƶ&¿™`èà`èj=±<™@jtûk´ ô6¶N (=o€©­ܝ޲³‘þFì¡ßˆÀ ü}ÍÝoÄ `ýq0Ä~#&ƒøoÄ `øX ’¿+€Aê7Å"ýb‘ù@±ÈþF Xä~#P,òÿ¢ X~#w¥ßä]ù7yWù@ÞU#?õ'þ‹XA6 A3báhù[’Ño*‘ƒ¡±%tÕ˜:ý¦³üKÿ{1þe€r0þ±ŒÛZ†â_Ÿ¬¿(ÖÖ¿£ø5- &@Kào  êÿã‰dhmbèhþ h¦þ¤±ÿ²cï ÚÄÿQX~ ¹[Zÿá TÓߤcúdý-þp¿!Ë/èò;X¦_„ßþØ~‰Û:;üá $`öÙÿ0+¨qæîvæ@›?$@´?ü3‚ªññj§åTm«? ¨d :a~[f©Ú€6ó>(wÛßÁ€”mÿÃ%c÷› 2fº†mþ3¬LÿPÿ;, _v@Ðuû‡(û_4 ÛßgÇÎÊÙñŸ Šýï†þBÎ@Ç¿™çùÑÖ hbdõŸXXX3þψ²ÿÃù¯<ç?Ôÿ 3ýš?:ÈjÈïÙ@JŽ@k‹ÿÎ8Û/ Ë}dqÝŽÿ¦ ªˆ£Õ&˜ Õo·  †ÁÉÜøÇ"€*îäjû‡ȆóÔ|—? (2×?¤íö™wÿ‚Šèñ;8% Ãß®þsH;;€šëô×5 Ú¶ÿá¿þ*n@c¸…Y[cîÀ5­÷U‚¸®t;c¼?ÈvÔ“©è<Úœ‘`¨*Óý×n»Þ-o‰RÞ,¾x5×Á„|‹SlyòzÖQšÜi›ŸÀìÏ;¬íÅ‹G§"°ëõbï¥æg Ù Þ.E–mïüI!íÞµGÜ­¶·di$xvGq·’]þ¹dŠ.R5Bǯpš,Ç(c‹Ú‰–õÜ yúæöjÖø+¡T œ÷q$K¾§Ö:sÔÃŒÇJ™ ³c6)¶>ä êÈ$¹§Ð~¢Ôû9Ï¢üå¾y·fž|Â,DÚ¤eºwôûÌi•Já6ÝÕ.# ß™¶³¸}hÛqÕÅèúÄò•-úáhNÕ,–@üýï¦B+­·j¦“ð9§j_ÓˆšæG¾MžOuö+ƒC}twÁIß:½Õ ù{ùñÌ\™¨E͵Ö—¸Òð“ JœF!:Ô’èïh]>2g努ðÀ.÷jÔžt~ž‡mÎ"jL™“&Úù9g5šB‘»X?PÂp¾^ŽI/RìN(»É6ÑOšÏ*)Þ½3W1¦ãïþñó+ý©Z¹j˶¯ŠDÛÇát£ûõ~BA=œ·&û›·¥¥Bz¶jç审BƒQ*9Ÿ¿+ùfu8ʧˆ †CÜmÍt¿5"9Qr!²nŒFVWá}Üí(xv¥-‘!£È_í•>à ñZÐÉïÆ®òÀü|ÓèË\Ó&O †þn%Š,}w ’D¿m°Qíâ'Ž®²Ê}.‰Ç:X÷ÜÔèÑðä©§U¹N!kO€›-"Ò*†œhnË7õj—2"ì§iXÓªª«`Ý"^uÚtpíx`ä L #:ߨ± ¯˜"}U&Ÿýu6.óã¦G†À#ato…£NÍ*ú OP)¯N‡&8隨êáÛ‰Ýgö„¬)7ñE›þRÜÔ0ìm‹J²ÙøL¢¹I8f°÷ÕÚ™©(×[ù>TD¦Aþ°Æû(á°`[<3è5щJh ,¢9Dt\Õ((ë?'‘¤+é¡zj„¥ÆÀ¤œJÀÈ»=9 ¿°WÑø©MtOtÍÌJÌàÆ]…3!‰Í™¹‹±ÅG»^(‡’zU¾úÙ*òŒí%ÁÃO|jàÉ `b,ö‚Myåi9°>«k|˜O_4?_Í\+?ð…­¯¯ªJãÄ6à¿âH<„×ɆÒNø~õÁM‡•iÿŠ×iü^l‘›=¥¹ÉÛAÚ™é€å)UU Ê"³N©¼fò‡÷­;ÁìÏOÆd:kög¾:ß¾í²Ê³ºi§éßI~Ë­:«­|ÆèÛ5Ça\Ã/½”•zÍÒd,‰vñd_ÊŸÕ¹ï@8dÙ¦³`óÞAÆg¨LiJÓ/KcRuÒ¾’ö4 ÖÏlƒ#Àw_¢D “¦îf–gü¡–TÁ.©@JjõtðXÖD÷~]»ëh8:~ß‹-=2GÎLfoTér'5‚BëlûI¶Œ¹ËcÕ96%;dš:µ…º±· ±tò•³}¬h¿ç«t<¦1ÖðøÝ†Î¹Y–S99ñVÔxÏõ“ ´Ê‘k[Ün+i¶‰œŽpSZÞØ¢üíì¸@¾:³áþ{"“©Š-\Û£OÛLj B¼Õ×'DŒ±š9¶¤'wý¾d=å퓃®«̺¶%Ýk€ ý~Ióadçý!ëS1è÷\õLH°ó‰&°Åîìwþ‹«|1Ä¢@,6åKWòbº£U•™â÷Eœ‡X¼±Û”#/­’Ïæru³žÛò¨¤d&Hãk (‹ú‚òbÙãCëFõ™àr¬¼rëe#2Æ…d(êGƒÔ#˜žœ¸µ8Ó›î «62Qæ.5A*e!†]žÌS6´áLÁ¯kwVÜžÃð”$ßæ5 lÄ c‰aÃÁ’Á?x*ÍÛÆ¢tö¨¼§ÜÖW?‡0üé#íe+ ¯6û#Ieí'^L¢û¶ìþ+»oivµ^Û¡S&!¬€<ÓQC˜Øa|‹;IÊ–1²¬ìÜ„x¯ÖËŠ6¤uÙë7ñ1è·¨V%¥Ä]R|À7ójBÑêBxÞ{jT_Þ,+,Y`Óæœù­ž)¡b6aæ·µˆEöC–´)dÐï·É¼™ôu»S€¥¯=¦¡Šò¾ÿêZ1»QEáõÆ% û8Dk÷s'ÀÁÇ<î±)ÎóâK`„•'ìPñnö‹¾€é{v‰5C—IlÕQ ýª¬Qô>­R&ê’íþHÔÊhYš~#;ß`nÏ•’‘·~åŠ,˜ûjjÜó9­£¡?Q+äéå3×-h[?|(%8î2ËâÕá] [”»WÉ1zJŽù®æeEieVÑ™±h4U¹BÔ´_2“83¬³ñò`ȪÈô¾jä ”sÓ^-¿©p$ë¨gy’ëh•fƒäÓ Sþu/kO&Ó[7 å¼Ä>,?·$ÀENä{¸¬8'©¤ŒÅOò8Áy3Ü +vwð©·Ñbòõ—<ÒowŽFó䣱†\‚ã|V§¬rÉ÷êï5wæê(ÆÀÔ'vÈ;rý¤?±1R,ÞZ,L9¿5ÊÄé†ûÐßã!(‘?“ÅçN©íÖ¼¹âÄ$è\^’`'8N¯ÍúUyžÙÈ5žÆàTœw*ÄÁ2Ç©Yï%N!"ø´]ÊMÃa§“çꆪ¹)§·Â&Z1º¼‡Š—¥¿Üœ(^·q~àKI ˆ`Õˆ6yÊd°üÁ™þÊ™ætkøñŠûœÖÇc°n’ ζ¥r-Û r€1ÙÑevUIÌmütz?i `öÓ:&²ÞMÍý¦~e2ƒX—ÄãVR¡Ëe'b~pLÊí|o½ƒçgR ›&Mára¦Ç§’»¢`–r!ÓáÒUàÙ?¸¿vNÇYù‰=RBÃÏGÈ]…€s¯oä›ü,ëz:-÷¼FÝÆÓW“°Õ‹ˆVéØQÀÉS/(,j¤vk¾Þ6ØC†¡7&–o«ÐÖ³9‰àCZò¥x’…¥õNsŽóX+Ò‘3œÂ`«œñØaQýK.>ÿxÀñÇM|#¡´iBº¾úJ93«+j±•ü.é4³úZëJPß±<§Ñë×bù‚,´¡Qõnj1T݆÷’Åê·¡ƒñ:Š ]¸bÔ%Ë*˜‘ûzr€[ý¯²yÁÊd*µrߤ9¢+ai”RfµòÓž4ozÑN4íÞ­Õóxùáwóš! ×pcbઢ¹ÇW?#ÎHtÌ KDPÔÔók) Œ!K.¬;‘}Ý~MLïÐ.¢âè!ÿþœm>f“P} I1½dØÕ Ò}A¡áà3½–Â`£o“)%“ƒ K™(W³—=<‘ W^/‡ï)-V¿¶t®g™f´Úesø óÄ@¦ò«.ì’ažjxs¯çs(ªËæ™ÌÚò·Ý\Ö¸k¦dR4kÁ×SÇô®¾«K­5Hê”Þâ»!Ç›sÛ4t#Ï÷YxX"aaÂpÏ÷ưôê>Z Çg»Œïmà×ì‚Ì’MHé&un+ǦÓÈexìÚß 6![¡!x5J«û°ÿH?Ø“Ø@c:Öïá^ä@¸¡_#äQ1’|ÍÌï2Ý qxÕz3Ò­nø•èöLž5\âR–™ë ì=ÇŽØ[MweÝOAY)åV¶B³eqo 3\ˆöÃÄâýVÖO‹E× µCýMïÜX}%‚²1¼aÔŸë¸—Ä fÑßi¢³]³-\=ïÞoHèÿ(Õ†¾gJK91f“»ç,±ãcæjz¹ªwd€< k0£¼.vÃP¤ErâÓ®Øü0'îØÛäÇE¨DʩˎNúhí^AeQæ2xð±ëV½(\¦GIH_IÈЪýÈ­V?{N3ƒ«†M¶°òÍÚîéúY ZTK9r=®OÑÃǼ õÞ…"ïUƒøÓ#ru<ß’íYEfRh‰µÂ=ä'ɼòÖ·>3ô šxó6¥‰ïô¯m–ýº`¿Ë?o9 ÊAÑÉ Ú$†'ñCé¯bêé-™¡ÇÖÊÄì‹æŽ4 ¨š8)’ Y=gö4t.VÆ4ÁñÅï[µ+ÚÅã”æúü ênÈx¨Ï¿dˆä†QƦ@Ðk¾áÚvz#UýQàÝ­´6@”’Q|ˆ{ìÆ’¢=DTq¼¥ Eóyæ;îLž/ö¸6óÛÎ~NV2ÂLq=¶Qœ$8û˜¹û©|?þíð &ae𼝠LÝŸ€ƒÊÌ£ð,¥R‰²‹C V”?êIa ¾w¬í䨤9ÀÕ_€a^TÝKó«†a‰åëZÍ?*Y#U½Û{{öº¯‹ ÑU“DÉš LJ`DÓà’°Óš;E?²°¥ÉöóE+›Ž žÒx…F$uq}•¡™î$W?¾¡˜#h+˘—hÒíøRvøénÈUÐ/ž™à&à|‰§û¦xò½&fx›oìÊÚ$%Ü}z |Fw@ã ˆÄÇÜw[Žþnˉp»ô]|‹³••9ã–I‹V=Ü䉕$<{‹©U—À/T¦“Ûbiýü¹[’Æ`Êä‰<¯P­±§o€.6`Ðü2Äüñt¶ÔªKXz¯ ˉÜÝ[·¸.~Рg}u;‡Jô7—ý9ûyƒÁÔ›e-1íh¸ ‹X¼8¥|ÂTÖqfšPC]õÅûù¶¾±I³§>RÞwzŽìãh?‘Ä÷³py‰WØÁÑ(Bý‡†pîýš±KŸ˜M©&—¤?Ýe8Ò˜ä% as~?¶ž@e¬ü–“áªÎ ¸r¬Œ·©†eu¾tè)‚>ÆaÁV¶êàu²”—ÑÐ`…$Ks6šž— X¼¬ \ð+ ¾QHZe¿ø¬WíÔ=çÜüPè:öŒÏ…­L>ã³3Ü׃÷›7?’·„4Èhä NÐ ሮä -:Õæ7Ã’¼Áoæ›ãœÚӅί°Ê{XœÒ±Ø(=*Òņ¤Û®fƒq òÚ›ná¸öý_ÝßHR&¾¶~|Ç&SX"Ùa\ïíeNÐAã!²‰¦:ÞõŵE3’¦í¥L‹kìl¯g<¯7©ý²ÞÊgØÛnà|/ãõ¦ì˜cv9t¿.µOo‘÷”!Ó}òÉÀð<µ`1L»r,ìÁÔ÷Ñ̯]®8Pdž¸9»—2†ïQ½?Uí <} ÛñîõqCÞ=So€‚IÙ± qŒJf´¡FòDª´â×*ö¸ OB çkÚûpBhG'3I>ðõ òLÚ€`PWDÖe˜{ºÒ>èXÂÿÍe¸¥OOçù3u,Æì‹âÐF‘œ‘iÍRë÷ÕÀL[â·‘ …ªo[=&¯ÌÍF ·£¾8׉eO¦ðt£i<º¯4•”M@•Á}xS¬ˆä¯±ñ ÞÜPaÄ?Z¥oŒÅ¨d7"‰Ý=êìá¼ÜéMÇAâã@Âs§¼íð:Tˆèôø#J8O³¿*ö¾L$Sd(² q»‘4E²\*p•|4B9Y"kr’ŠaàÊ‘=MÇwP[õëÂvÓ­j†nå£Ú<Ëê¬×õµÊx€d³ýþN{/gBŒû­yšEq˜Nµœb§ä¥Äôèõa¿ ׉À¼ØÙû=¹WxeÎÆjž2ªóuõYh¿6VÓf1‘Fóè%úùZ*=e¨I:‚ WïZøP-¹&QŒ aðëÊ;xMøH%(`½G×ôí9Kë”\üÓÉÏŸšlÑïÅ>é§¹uF@ªfz…ShÄÖ÷‡Š «Î¾c—„×!L Ü[¾å{æTÓ‚ä›)खdôÆüQ¯Cì9{Æ¢ºèÑw1è¼-–n 9²t£°V4D%ú%ô²ÿ³j~ú£+UŹZØ0†ý4׺ÉG’üæÊeŒPsû~øüôVG[DºfX8iHqŽè-âr/ʼnˆÝ’Ç€Ö¥‡£W¿÷×™™:e´°|._Ó’ëÖÂ.A–³èÙsexæ^ì˜YíZ“Æù¸g©tåZL‚r¼KLããE,ßçd/èa§ï™&ûÒ÷~†aTtLVüÄ‘r颕&ô{Y4ž ¶úåK:\òsšb`öò*j§ÞÿÅ%~cš[²õê·'r-ËBÊlœÁ÷µà‘%™Äݳ=éµfû*­+ÛzFÅê½°ÐK”5Ô¤€Wè²û}*Œ°ˆ¸­ =t9"‡ònM|žü`ò÷IÜ‚&’¸ ¡~Þptàì:ÃV—È#/aéÒ¬8Ô£~ÒtØs+êùÊ; ᣂ%â=˜=Ëïk'¦FDϘç'$8:ßWœhn«ÒÕˆ ¢ŠÊÆÌé¦}ÞÌ"EÓ½`&ãQlÞÝü´ i®·A°u0r\åóµŸd±ßÓ¤ÀÎu2ÌâÑâЖÓ!nTÕÓÆcØ›]Y÷ŠWAëÓïs½  UQÅN1_NI»0ìNÞ6ïE]ž½„Þ¤o£G€ho.i_P§âýž4|Ųp#7™ÓòÈlP Ô®2M“z'¢vZñÁrLBˆÄè#á'ñxQ”7rISb­×Û&¶÷æ½"óîàÂ,Š6›;p#¾Û´#HæÚ˜Âäˆoz³À,²¼~þ¼Ô9­|&a¥ '©Ý ¨ Ý,™ŠÔ?M—j‰óÞÇOŒ Å“øx–{~Z?32y‹üc±:7\eKàcS•b„Âõv|hÁ~—ˆÆd­Zê‘¶¶ú椪S´Š¾=aM ç›ÉCÜ ­–Ÿ×¤ êõä®(”þ~1­K`HR7îߤsmØhÛ3²É—=¦Ò#ß²ïn"‹G·W„„ìÎ#ùŽšõ8I>#&¨Œg´$Öcè8'ÏΡ On0÷à*«rC4‚{÷»&N¾CLÑ¥Ž>ô`f-6¾»í—a%½`^«†.=rå™7Ø,Zç B*+Õï^@`ÌF—œ¸:›× h±2mN®\aû~¶®³âtúIq¨•|7úëOô͵´WŒYMˆð-Ñ]Nã·µ'Ð?Û³=÷Øwku|Þ™=À¶ˆ¶žò=òÙe›óÇz`Ëw'rœéÒý¶6ˆ‘Û.pÁ Û“«al“á´f±ÉÛ»+70:JíÆŸ'[yo풬ȫ·Â-¥Ã8h3är¶}$´y­ËB|ivNÚ®²!)Õ8ߢ©LW$qm«Ýúf›î‹*ÆoÖÑ -—ïPÇâÊ6£6Û÷ªtïšá_s LûG)Èœª—04ëbÒm`ùo9®ð2¦;¬¦/g6RüPu[iƹÔkê²£š¡ó“jן3.G¤Ùkô£1OU鑱UŸ53´¯¤6DŠ]®¥KcMå?—ßëñ.Œ»šlã28¹+*É3Å7uÌWÈVÚ«HûÁcpÿ-6]ÓË2΃‚¦ÁÄWëcËÄ—¾FAÒ½¸ë‘ïAéˆo|'å0?‹ð„ÇÓ¬[e¸¢)á“mÅò#„´W«¶OeðnÄj'×ï\`²aÅ,0K5=L~¤à…Æ[y%±Á滥†¯øBÀ†C>šz‡äß§á¢ÒOÌÉ>áÖÁ—ÿŒ˜4àvÜ)äFO©M@…²Ÿ ?ƒ©] nTlXÌ 4ÿYo>Êgk¹âdžœä§=·ÞsuÙ㓊gi!=5ÁuŽaˆþ ¬ñ%:;CófP©ìøv ý9vq¶i¡•ûí¢‡éžŠÆæ[¯ON$â²î·…væ:Ïšš:£õ'·?Ó$o)˜Ç4À¦²ÒÝ?=›\™½Ñ[Eä½ob1zãÁÅ´Hâú¬}kô™%_e!Õ|3q….•¼lÂýn{÷$aùtøëO ÷~äC{øjK<4º<üEože3÷€xs½‚ò¬·8êÎï½Ì$;è_yýf‰Ç²ëL'çZxo¾=êúÓÃ>_ûŠcaä‡Ì€ª¢å=҆׆g(òuÅG7‘kä©ñóqM«.lòY‰@dúa ØÒç}«¾¡ÀÆåˆÑsÙYØŒûä•ýYŠ~(ܶ¶ÂxÊ`,üÏ÷‡ž9ØF[ÃnaÚr¬ï¿ôÏ)ï±Ù_=áµ\}±F²ÛG¤Ëú’ü¹EŸZj×Ü¼š²˜¸u·­’Ù„ÕÙä»{Nö£ˆké‡úÚ¶Áj:¥¿ìó­ iʸhúƒó©ñÑëÕÏÈ–pR\‡GM>¢ù{E*òÕ[¸Q‰{‘­XƒXâ©K¶Œ—7ʾYWâjU¸uu1y^ÁËP5Éöxï]ƒ_`îé^jš·iGÅ“UbüÊÂcô./+ÙwË©ÉRDz6¼0’!âf€ë»GÎòQŸÇÔòêSUa;cl Ä¯÷)-.bsCG:s]ÚØµðÐáärö¡H2–A/Üo.ÛDºÝÏÆ!¨ö+z&Éæs>?óÛ:9À¨7I¦>¿‚»'Ãß  Ü›ñ ¢ždö3u¤uÀÒ›Ûº^gåYëDØg)…èÙ«%ô¡û ËÎæ‘S _‚‰O"E"A¢æ…߯ùá¶³uX.—=TB™Ìq¢¿D*UL_”;ßYbòXæ®4Ê:”é“ù¤Ö†–…æß°âË,7øA5ÏÌîoIÿ«þÛ¦«GÊ꾚ÉÁv†ä¶îF¢%Ö’˜6T ²TgêoÎBÙÅŽl6¹­RœŠJ½‘œŠri7rž„ßÇ &¶f}Ú/a‡ÈåÂѪs*C†q/Ì̸òèx;à«w™ŽW‹h)¸é…ç!P¬¡ÓŸ>2Ë]%]¤Ž}ú¨j‰zUÑ ÷ö.Ûéýtƒ‡!ÛÅæÑÅŸLÜ2ømªEO‘$q2]Õ%õ|6NÕÜ e/‡S‚i]³pÈ3z8‰ S•Às ÷‰CmÊç´ãÎÉÀ7 ÅÞ>õÈ ýBn±´æ R0¢µ½.×àý4‘1›­@\÷D2¾#Wýèé@Ô0—-´¯?â8¤ñ|DÛ¾;GÔo#—ÓLØBûÅÿ€c¢£¥»ŒV~c¥—…¨}¹8-è h•Ú7H˜,—Do…\«â;ª+±v>ž‰ò´OÒ•›Ú§Ùd?DµV™âc¿'âmõΰjÒrÆk³__Ù\j£Ù¾U[;véLž~"EÒÎ{¡ƒ±÷×BZ‡›V]*ðqeIjBcTjÿöLB•³‹?/$ø”Te ëfdYÍÓ/üzÃÑ2aÊáúËjˆ'¼|Ä£6äÕ"dÎ Ë0l¶}A!v!Öø&úO„Çé¡÷>*˜¹mH³Ã†3š&Û+…í:Z“ݵìQí2øïTNȰ(>º¦&òMÐ’›Õ­ ¶´sút‰¡ÏQ'ð仚"‰ÍØ@ ¾èçùæÈ#›”I]o…X»ÿ ©±ÛRÖr)l¶tf —.­«´ërn¶It>@’}ŠÕJÕ-êYb}\èñÌîòð•€~7£UŠÌO$@5íÃ0¡vaI.t°¡;PßÕ¦BÃ+böÔì䩼++ ª7˜FO ÍodyDÿñXÞ÷8tT{³^o£n«µlÄ6¨`[±VS¿ŽY’íÚéd8VߢÙHg¤BGibاMêFÙþ–ľoµÙЮUDA]˜zòÍÆNh^]¦ðþÄñ£û9Á7ÌÃé@¼´ÁŠqGÛˆYþf°Ý‚“d‹\(¯?¸%>’x¤TÞð½iuòÒ†ˆ@ÿt75 ,vÔ Ñ7ƒ÷EðÔ;egl¡b, ŸMfm º‘ïä3sµU‹¢–Ì%~5Ót ,¯ŠÓOk®ë63^Ö×SŠûUì›&˜díqÞø¸òáFXC5 ³ÄžN­”†ŠnšŠdBÖÏøD·Q•,xwH[š‡&˜ x±Ù.xÀ‚äÚæ˜’¹%¡¡`CÍF;¤×ÊvûïãÚoUçZêiךçhŸ”_÷ä–—ý^Št ô¢žî© çØºË¤ðy™ÒWq]@Lïi$ëµåuf↨©F­z0µ²0ˆßy†8$TÙ¿~ú²åÛ>Lä;^7΋V8ÕLmIÑ’+;Ÿ/XOœÑ½ü2RCðØ N-ïòÍë¦t-­l"ýºI ³³ŠïQ¹¬ñ­žü­8Év +½Îë’O –È;º*¹Iéwq}Íc¸½”wÒXiý{Þe~Þc9™´P¥ûf=Òd™QÙ3´¾,õ<=Œh—ë‡øu¾&ûPaë¶; ±2¹ñ*@Ül5}]ØA0òf˜öLQÕ†–èÄõ€¬õªÙê#FJ«Å¼-Ÿ¬™¹·YðÞwò :Ý`6^ƒˆîC yEIç¢ÉËï\ì2iÄCÕj•Ž£^a°P pí¦%—}³08Õbx±ýßT?žŒ3bDè š£"žjûH^™ã&`ÄËiÖ³î ÔJæâ“q|ÛÛ-´‚mµ€R"lá§ Ê'¹%ÅW¼ˆQxàpAäj!°uØv£P"¤iJ]ÔãËëíe{Óíg5m`ÿs¿«5{ƒ¾¦pR{ÒÀ²0öÍœð.›#÷zúÒËsš_y9xeÞ§èÔ`+QýpçY¿ '»Mú&\ïÇNë·l®?ÃË¡Ÿµ?-x•‚íà¦ï‘£.ë#^Œ•%!‹Iµ ÂNjÝ•Evγ%²Þ¸Âòø)WÚ´º uOJ°Ïé%X^OœÛE3Æû2šË…ö1í ¥=Z-½ îèî9q³6¶$טIÝ{ð¨VCKèúC¤–Ô'ùL<°@áê¯%B¦*¼™VXpþ7ûä<œ7C‚-ÛœCôSK ilU3.ðŸ¬pžƒ”ë9L®–@¯ýq3Â-z•ßø² öÁc‡ âÍ9^¬ k¾ÌiëIUÈPˆà¦¶ÏÝ^(3eÆ_ã)φyK%d•;O£„+B¥AR÷<Úmã/Ph¡V+€K3ø3—>^ƒÏ\‚nsKYžˆ+F­ÓI4'Îäb‚4W5rH¦@Ù`‹Ë §q#¬R8|›·–!Ϙá£w³ ZÇ©I{O­ƒÑÆI/I~U„ñ-%}á$½oò`€²Ú¢Òï·¤W5(èÒÔo_34-8>túªX‚FØ,å¤ê³·LY»ÙLj(ÕoûàlÂtŒIAâÛEÆí;6Öí!hb»,¢C%ý³HN?½ï-íá…d#Ü&êqìýç‚,Hµ.|z&ªêhp©äÚ£$íÏ‹ÔoÇìõîŠ5ÄhŒyOo´–Nµi‘óNe• Ú>¾< \Iˆ¤ƒï+Z(àyd;;#ÒêCá:+¡ƒwõÛùo¬·9Ù<­mŸŽª¹mcŰ9u“–p³°¢Ü2:‡ŒÑŒ*R«l;SßÛÝk!Vv=6Àçjºo¸üM ¨¢C³§»­ûäâqÌŸ5'WÑ­kÃm~úÃgË+û½Ð[žÍYÒ¬ÒôÆÊJ¯]Ë|„سv>7å§‘ÙúUñ7ìô÷›~eð‡ý=ðtÍgöàû²GúLÎ-SFG˜püŽjßH—ssÒ}  ÀRÁYKu^¢´ng2Ž`W¡¿%'*'Gv~~¸XWL?´6½¦–, Ť(’8ÏR KsOµ¨^G||ÀHð•Ño~ƒâùDÞ VÁXˆ_Ñ¿>:×Xþ’¯5(C¨Ã YwÅ(škÊ­5Ì#åæ½iÊukÅê/ØkÐåÍ’›"yÌ÷Õ»V²N#_ÊóLñæo£Xèjxžðl4í{ Ñ>üš OÔ,ý:‰7y>.ò2®½ñö¶ ÁœD1õÔýÊãÕI˜„éÄA³nóèfÌkq(_Ÿ!¼r ‹#Q(uëtp‰‡©7R8#t‚CÏÜœ;éU0ýÂPÀ2 ­b ‡î ֣ǩ<´'PKó-›âeÒsäCÈñ<…€øyˇs!qæ4~·‹X¶=FS¨C?h’€E2Ž$“e2¾ sßUÁCLÍ.•7ߊº6 ïñ†‚í{x…N˲©S Ù0—B†÷¹î|Î&M”°ˆ©¹ Ôðš“ó@Ý%×D^±tÖw§—þ>_1öžŠî"÷gls÷ˆy):[Ê~M"3ÞbÜ8[Xós`jšäÕ ^§yêÇTäuÛ;ŽIËÒÇOCy/¯)«?ߘ֠ôMïÆ”|NWK ¡Ç‡?<¼y>ŒU¥á,ðʼ×ÇÍÝ@qˆ‰Ô(^•n_â àMVÙQènc@Àœ4¶ÅU혰kÚ¤QåXa¶¡Ð…¯r^ÎÄ… õa€!QÅ›™šS²{ü”j‹‹uú1«5–˜â*­`íé«}£šÈ7Ùõ³Œ´I§:'ÓÚhô®š×>ÁS®q%,”6ðY&žŽL$&Z­¾wAÊ”h¡>‹!ôAUo¾G² `¥ÞuQ›ŸàÙ…¾~2ymÓëEN°›¤UB‰&t¼9zèô½°òÕÊ{͘¹6Ä›}§-ÜÒˆ${«bÍi‹ûŠÏ› WäWî×P‚HR¬Äì ݰ€bÁç×ÔøÉÍJáÁ†ßZ›ÛG;=͉Ôß2᥸K¼auLO°kGP˜7öæÖµ¸×ŽØÙ®‹‹ÇdæFXÝ(;í#¡»×¼÷#Ú[¯aîƒÁ|,˜fü¼PÀOeaãÀ‚AƒãçÀ\b¿a:IRÜóa~Z\´å‡>p—Ù*\ÑÈoÔ×ÍRÕ©0ÂE6BJ¦/ Iv­oÕXÇ®Ëhɧ'Ijðâ; „ÞÙë»õæ”ñ’Íú}3¥ñæ¼oé|ÜÔmMÄ\xzß“L±ÐÆ-ë=âþu ’ňDô×ì-ްQØŠq³Aöf^¸JQÌkÃÏ í û=W\3Œpßö@b]ãW1@oF  Bf*· Yàüye凢^â¦Ì”‡íå ÎÀŒÒóÁXC9òMùV­Æž’D=8…YãŠ6™€J±Á’ªæ·˜&´‰¸öB*³Èùq#.Énba£$!‰u™ðì;.óóBmÛã’tÕRNIb+Ö6Só`éÙþ÷ãï:sãµ=2„¹+Ñ'—o,E±ô\±‘ÇÎ_^aÁ]º‰øé¢T¶Úk¶Œ˜Ÿ ˆ)øÔ®høEuQs¾*¾ï¨ú˜àAH€€¦ÈÆÉ=/Dçø¶Å]ÕV`NDœ6U\²ˆ„ ~éã¤Nž·5l9ƒµ !¾²Ë|öÖ¸´ó+y½&$‘vm%aAæOzêÑ|ÝKgZ [÷j¦ÈÂî/Í:ž¨-ò!¶ÓGó,Ø?é¿ßë뎦îÙïV#¤tÍÍNM9¯:)\)¡Â0¦ŸïÂ^ºÜ†%pðé΋¿¢s3“Ž0à}o×n¼qH ]…)$Üý ºL± õI¹7³GÏå$pŠ‚ãкýÅ”oj„ûÂhtRû·p¹lØÅ še‰^R•'Ǿ^è<ô– aÑ¡Q/…Jê®®DÏ®'±9š…£m[¿@r¨‰.ª7G<Ó|’K¢ Ÿ;Ké±xUJû+3’'ĘhQTªy,–4ÇŽ%2âS"aXÖµ Ñv™3oß¹z¨0-hzÓÛ*Ë¢SÇm†«Šl´±{ءêhs0i%?EÛí+Š2šjE ‹¼Qs3¢CÆ¢¿2Ü–Æ©Pºw«ý¡åc’—2ÿÙ~Óq›;Q@½i)ú>sX‰^é†X=ù½ž[ÆH— ²o+,’©È–!…nïkÅ´5bã½¶§»FîÈm6£L·ZJ›Êw¡Ì™¬ÁŸúÄdÊSÜêˆoeKøsGH’$˜ær­û ô@dÛ¥wÑüä„þc<5t*ƒšÄA¾œÞaDc³C!Ëý~2\¶³ðÆ$Vƒ¿$AÂ…tRÍs)ˆ©Ú¤­aï­9S¥mAðŠ({*x‚¿öu1`ƒ›7”kM=‰ôŠÛN\óÃz>øÁl5C/&€©€J<þ‘Zb`.IIˆ¶ LéÛwÇqœK ‰IÞ}§çM¿/'AUû”Ý•ñ˜/(8Xë4D à\vEë°¡6Uöé—ÞEx%ÂݤG&,6À'{LæÇwß-]öb„Y R–_” Aoüâ»0Ð@%7Ïñ3žÕE'ˆUh vVÄ\ز¢^½z Ñ’Sw2ýeÒî*ÛÀ6aúÙ%A}Úø“v å"¬aï@SAÄ’%L*8m¥©êlKáçx§Þ³w|á!x«Áö;Õ‚¬ìÇ‹8y!Zßwð‰Céc"5rŽçøÐQ…ÊJPïªåß:FÞa7ÊèQm—8̵~=!»hX|è]°(@ R*\=ÅPÇ™ÚÈæz¿çG®BW°df È®ßú¥ð§¹==j_41ÛÊP<j8Gc±œ·íŽÒq˜€òÉ…ìÐí Øj€è§^\A wÁ¼“„`ød¨ÖL&ö2ólÉδ»$Èê$ùG³ˆiMfò¨ÐåOv­®a´,’27Ö7×ËàW*ÔšÖ†ÚXª’Áæm³¢7»³z“5ÇEbï‰.c!Rèd¾WpÜÏ95rì£O¶µƒãåäÂ<\×ÕöHËÂD°ábªŒa°`±°9x+I«Ø–ô€¥—ÙäÀ0¿*|”“‡„ª âГLW™XÝŽÍG²Œk1wR@]o ¾ïF3ªPOâ-´‘Ëîlžlô}sáÇ(\A€ ¶¦Ô{–3‚s/õØ*èwZ¤Ë!¶/Æ›ð³ˆpúB§™dA(n¢µž`ÝR¯ —²ê¶•Øc'Îvb½9IR©ý2±Þƒ ¼ô"¢šóuL ¥R]ž·Å¶[uá/½üh‰¯{°õöõwe¢Šœ9òö=ôQàÜ×¶¤eï™í~Æ_Ðcm &¶Ò¼d™8[Jëa×ÔÚ{ÐFÿ¯4}’×܇ͰËHƒ£(ˆ\¶»µ¦euäÆÔÊY“1 J=0gÂlRàÃÓ›9ˆ57A.q¼†ì«üx‰º‘(H-iP“¥’¯uw¶ò‚‘åO¯™œú)^8þõòºõ„ë°\%ÝŒTƒíG¯g\h{ñÒku6ú>výÝZ9Q…ù~x߯EU[’.ñé‰É‡Ëwüç23tTS‘žœ­¢+еèg<ïm1 õS˜/qÛëeŒR¿».î+žâ:’:¢nš‘;r(Áü,Z%Ä>{RR˜‡›¤õ4ë^8{¯4]‘Ía¬é$ô“&wàLó^3¾›\‹ž†Ë”v¤t<÷éà[ÌœÄë-#MFOeÏ\Os¥O óPöx`¡U}&ÙŒcüÊbÜÑL¦ÑÅyâ|âIdduÊê‡1îù+jŠ©ÚE&M"m&ö—Îw¸zUwË<ÌÀ®³›>J«½t©«ºvÉUfÙÜrÞq϶c›Á}ñŸ‘ß ‹ ^%¶Ùhèàý5„oäý¥k–xk9Ê ã0n’Àð èzÀW}ÕòEQ…CJ•6¼FÛ{Y"O½"ŸÛVÝi=?ZÆÖ[è(.A#Ž© |BU fUå;š^ådYmÿ²¬îüyetKšVóB²"©8ÝHõTÖ³ÕÑåÖ÷vô;ÙlÔT‡7ñ¢æÈ8<þy¶)ÇZ›®0;)™®ÝìѬá ZW¿Âùíšwz,à«<íGÏŒEî/æá•ý15ôC’ˆf#·®&ù§ÍU¼Œae ¼¥1B‹3Ïë µ&"\¼9uYÕ¸¼þ³çøÃ8çNäÚdc·‘âËKk\Túò3ƒ¤_«7ˆ™h’ÉFÕ.;σ,Ò.%Œ“Ÿùqo½UuO(Å„cÿqn[K_CÌ÷{&¼÷ÆëŽê냌¸¥ÌÃÚþéCåMÇaøûÐáý ÂïÑ%ƒP¾Ø]A;4íy<Ä0ÒГöêÇ[Í›¡[óêàðTâLkÎûžZëÇþlŠÖĤ¥ñä°þÙ[q!pŸNqg+º‘í_”ØÝ*Þ°¡½vD¥ ª‡Ì6¦5èG^ib]•ŸÐ%9âQ!Ì¢ ?—¹…G&ä~è„Ùœ³,_6ßÐhçѺ»¾zôÁפ7ž¥ªg÷Z%GDÇžbÆ2Τ1µËYcýy`·Ï_<Ê1¯JÉE§II/k‘w{­‹«jMy匜4‹(ÎJ¾ […®À&ºÏËÇŒÂxÁ ÎåJ ©¤²*ÿÚ²^—­á’å«eqæµwßfÍûy ècD‚ÛódÐ’KØèwH)nùòŒÿ¶Å*/úl¤}UMj‘æTií ~5?üÃÀ3‚ýÓÝ ‡Š^NsQ…jC*ÿq‡:uE`]Cì‘y,YìgQñÃ\|uHt¬<[D²¬S'óÂîG¥¨´IR+=O#’0t1-á¢åSóqQÖ{RAZÕéÇjßÁå!Øéö_Û4r*¬PßmÌú®vÇXwçœ ?…'˜íû‡Mƒ½`˰$d­ÌL´´\ŽË?&/cvuD|nµ‡šÞNCP\„ú†u0¥Æ xX,5w7厫¡øÃ?…,CÔ7]1|¾Nò¬šw¼¢tç­ATW C–U§¢‚5ƒ;˻̆MTDÝ%çýN/€x àj¾;&ÿe-ð9Ê›Ú3•;ÏÇÝåô­ FËy‹ìNæRý`!¾>'™Ô=‡Z3÷*/Ò8¡­!,ñtH¯˜£,–Úpw›®Z°bªî¥‚.COË!£Üìw.Ä—)O_¸ð‰a·—ƒ‹Õ<(’ëÓ`Ê›ß^Ò«¶§kƒA«99v¬Ä>ëa^ÚNË˹`%9ZšcÏQ™m¶Pfòò4¶?òhàu‘âëfp[ 5ýÔrÜXÍD«ƒ¾—Ô— æïâÈ—t۰̺¥Î» …âïpÅ.aî7ƒám6} ú‘¢¯Ì3ãam=R­ìL„£,4)›ß¤°úa…>Î*2]_ªË5 +ÒB« P¬4V,­&|mõ/ÄÒËÞ:(r ¥háñ=pév'PRl©«/_ŒñJ#:(¬™UàukŽÀä{ OÂTÿN¶x˜]šáƒ,&…qî땬ô˜5Ö]nkBSÓ¢Š%˜æ¢§¾å"<Ç„…”8â>9Ò!Íø*p´NMéEÉ5Ø1ƒAdÐ+Ód «ûP¨þÈÙ¿1¼Dóÿÿâ9ë#¾¢R¦ÿP£úº{,ûŸaŠ6o²@ ÆÂg‡ñÎ_$w0ý©¸8DÀK³Ä¢u ©yUÑ'‘¯AºúûI›2Q0ñnÈ7çÏ»3‘1="†“[´u‹ÍÅÙáïþ÷Íô+]©ð°Ô?u‹s1¤¶“éKN¶pÉ€’J(þ³]} ™ìÙ74tolÃ&ýprB‡§÷-â`Û/fšR×?°ò}_eT)Ç÷ ÀgÐ9ñ…oôç~€öóÀLåuïv+ÎKϲ¦#Àª´9àXúÉNÅ“$̯ïÓ7thÛ½7‹UÙCNŒ­à kýBHÿ¡zo [à¡ û—«Ãû¯ý£2¶Ïƒ®N&0ж£Ø&7}„¹ÇÊ€ \À@ý 6¼)>ŒçbAy—ÜÛ<¤ê´Õ}vƒ‘þåOß`î„ÄsiA$cÏfNéÙk¤lC”ð#‹Oý­0¦LtpXðœ ÔüŽôO-Ó•|HÄ»Åa Ž\Ô®}µÏ¨G•þý©âkÆárÛhéírÁ7³Õ]:KÀ"×4îqõÍ3ÂL}Ü Z€ZcT)6Û8ªöÈ›íáäÖšÿoÂÙˆÉë'8‚{›™æÝÜðînr‹\{KOÝÀjLFùyVAï?Ç }]òÿn‘xðÚ’—kÛj”U„›z….ùµÄ·P4Hº8‰‹ ž¸¹.˴ȱr}ÕŠOªÃQùQ±ãÃIM†£$µOÚ”¨Ä«Ý§™ƒuÀÞÞ"›:WÄ•%ãâØ©'¦èN®Õ;Üáž9ƽ˜åÕ ÷½zÆÌYù ¨(dµ‰O ¡Yy%ÌÓVµ/Š) ð¯&’â4urÎç…ªe¹ÎmÔˆEÕÍÓøÍc•Â|M{Å5g ú³'Lˆ¶ƒÃë 9Î?Œ/ +Sá1îr¨|,.¼Ø-¹ü …nošî)<Î }Ç/m·4 {¬žšCæ¿i,1´§ïÎIt[A|þ×L’'(D„¢IögÝ÷yêîè—è}t=Í6GëGeŒƒÒ•o¾tœpÛ`±= …R’,‰6 %ê¹Üox~ãŒ~h¬@½«V w¨!^È\1¬ÈÓ¬cŸ“›`©ƒÅ#Ó§Ä–Úþ©[>Møw‘¾U£Ò“Ä¢¹(uÄO¸Á¬´Y«/B;ÍþììrCÛÖQ6|ÂS^zA>Àìá™)¥TŒûíòsF•è*¬Z­ŒŸ-øÁ ’dýPã…!Ïv5pâûø@Ë”í°»õ•_8ƒbM+ÓÍBd^©¤ï2þ¨6ÕÐ~ŸÜ†óø1¢±ãŒõj²`¹D $d,c\jÝ–Æå¯Ç Ö!F¿=wFö{²l,…•z¤=(?òsNÆãu-šñHÖ5ðBcdâuü¸ƒËÿ_(•Ï=0ÛüSLLöpD­ÛN.fPÛl¨'þõ¬Ô(Eñ‰6…ñÙ ¦†ésC?x2;f_àû‹%jˆñBÅ8 ‘,§åõÃgLe¬ÖlwNnœËç¬(ù_Æ`KÁi'Ϻ#¬v>(ÍkDGs²Q©QHÌ»óÐÆ67ö‚ŸrÏ#¾Ÿo¿³>¤mW8d8`¬ã*r&nä 2¹U÷æ\Ëfé„vÇ£NeÌ“äw2?¸¶ñÆøçÖbå1Ú&™¥:©å:ªåô<–·€:ö©Õµ¹`Óèëû±xn÷†‹!B![—ब‘{‹‘] Ó«ÖmSAD“CÌðEr­7-;îe«g!¯zþöìª|3|_zÜúß TMt¾c€(®Ä Døeç&ë–NäæèoIÈ’Òì—φȾceÜ Ä›ŒØ" ¼PµØB8ðIªcý ¡'Ü­qm¡¢š¢˜iÀHúŠ‘øˆ¤ ²^6Âxã®®ú/ç–ƒRóu!tu("Y²•·2ÁײDŽÔbtÍÖªcœ‰Qzç%í$ü©`ý#>\T(Ž-¸Ø¹ÄRì^‘ -¶ä‰‹gÆ)^|JäÅôC®“s›§ üÞþ’/pÇ aMFì'¦`‘Kd{ üÓàÐ _ÌgE]ˆŒ¨£¸ ßÆÎ°4ˆA÷š`Ù×øH²ì1GB;J¬S`eòŽÈ6ÚK‘ Î ¥3%´­â}]†`N¨±v…:½1Hû‰Ä¿I¸ ÅÔ´h £¶ŸV®u×@¯ƒßÐÜ¢­ )'ýדC‰ù±‹¦å4dÍÿ`qÆÊ²»ºÇ–ôºíà+y¿wNÅØr!0í¥°¾¿³Ú£ ó ;Ætž‘Sèz?}OMrsšq­G“Ó‡Ç.ãF2~Z* ó_¶û‚'o Ø¥¨¡Ëé!—æäs9Ä-àI÷1PŸ¶¹ÍyÁj@ðjÀqœÐÈ#Ë:tfŸj žÌpà¶YÖ‡‚Õm“Ÿ´á…AÁ—c|•æ¤#{˜¤žþÛSâÙ¬6¦ïÚãØí‰òUÏÕçÙ›OQf4šBøOÌ´‹!ßËÞò²¡v,­èuöœ*6¡”9¹-N1Z€þý g“Ø.ØY>«¬J䔜Ým˜ê•"*$8¡JuYDõJÆór+ÅMÝarø5ªxþ‰n–&ˇªq:?]ÞA¥ ~ ‘’®%P?=Üæw‡Äf¢,(—D»í»]wE'ßI˵ÏéA›Ò”H6EÆ7 WÔ¸bù¤ÖØž½…wgÓâ3aÓ (º›vW|Í*íÒç*â}!YÜØÑÈ™4Žq†Ê¼ÍÄ-¤a«ô¿$º2Ép›ƒ0Òë3â:Úéu£3’Œ ]ÍÖ2šTÍÞé{wÊ÷j³‹ß%XµB­Œý¡âþ¬&ôx&·2Œö^¸O.§̬4Âq,¡'! ¾α“£rÇm [M)OÝ2ôDÕ¦¡;Ò´­ÚǪ(Yý—‘ž êvRÓ0™Í®Ü`æì=š2Fý*´X¢_F]Óeø©tLiu:%­+— z Îô@Ùd<é‚Å@<3ÔÓªNû óhöQi›-:Uû‘tQbösi^YyuÏ»m-õÌÚ¦ã>=0H§]Äõ,'µë .ßãÖp²SÛ+JZµê2 qz0uÇb‰|稦®þï“Gw®x"äàòøiî+5¥mNÒLUãt#Œq|«îôÈ•îJ½Â§ê®Ï@´S$YºNnk³æÒÅ—àtõ\øÊô¸$=BímçÐøÉÑFÙÚ»Uí-˜Úz‡ÃžA—t>Oá¶pö0i¢Ø(j¸ù•pt·mºç.s›}µ\æê‘¤Ü©-âŸö1g¥Ú"Ñ@¬©—YP”OɨK4œ (›2¡zÚ'ePiËt6!µüèÌySAUsm#®Q¸ ÃN’‡#Rb…§ ý+z¾éâ lö¼DˆšjFû$íÿÀ[Ö2ýÆ&e¦W’­énfò§€‡Q½ ’M.ÿ%%ÓIú@qõŽ•4ög6:PÕlhÍ0}MÆM$ûàÎ¥—¸ðx€/þ¯üå#óÌ0[¬¨èƒ \‡­5ƒ:¤‰ûAó‹'G¥ëY, íCzµÆg¸dpŽ.nZfЧ~ÏÑ|sˆh;4z§ÙÁLwî°'¸‰±GÅH꣠«»JQF ºúšjáŠ'õ–˱B#@4¿Îsm+R›JHž• lß~ØòyÇÏHá( LëëÏf'òᡞKiWzœ7”l 0z‰o%DRãQ¹‘[ç¿ò¢|cqa`N ¬þßàí fÊžyýFKêꮥ·(¤£¦Ç²¦©S3´›ïÚ7r£Ÿ.gvPœŽ&_f—(S ÑæŸÄô,Uv`î…"c}%W5ƒZŒ£ùÊJ"R$ŸÊ–,kË\ú­¦±ƒD`¥ftóæw†¼k³iŽŸ¥ÂPG3yŸŒMrN#‘ùàÚñA9ÛIKécQ^ˆ-¸yb6Èå7Àç{ ýOéu uh\4ÀÝkTÅ÷únÉ :B„èXIQ¼Ç)hi¿Â,_dÆ+¾Œ»¤·÷#|š*¤zƒÆxœ‹¾‰Ôº6Öc›Â‡ %´øwˆòéÄK[êOv—1€m¤giLÞ!ä½[ûþœ+zNþëË_sÃFð­-H[.¨*Ä6¯i6.›<œÑn¶ ØëV¹c%•ˆÈèO¬,ò–’ BT¤£Àu1¸Ñ΃¿¥RŽÙ{ƒÃ •ïÓbËò« ½úT…©©É#ŠYÑõemKÿµ¤Ìx˜›=cܽ²½uhµlT­ƒä#!/4V!œ§Õ–‡£ÃFT™“ô \ ×›º¾#o¤ 7N»Ð}tï{Ç©)Ù n`FÃFG;K¦¾£âoiàEk| Y`jx a~ßnpIg—EüøÂw·I¤<Ĩ)ØÙɸÞÄ{\ÿÝn/¢%¦ø k,Ÿw¹Ý[)nÕ‚Ñv±š–\Î[«ÐËÕÆžV¯œT]ð·)éQ0€æz¹ðt˜»$ÙÓ­Ls<­ÝÐJƒo ƒ^Ñ…êœö"{À:Ó=Õn,±>q¯øè¥ƒ¬wÄHÓŠÞsü¶Ë„†üo÷±ñB •3d~ÃF=†¿œWŠ¥ ßþ¤äÑ›I® ­\‹‰=É`Ðʇ'Y°åw»!”£fñž„s¹-`®mèY›±•?½ˆuô§¨T”•ôIªó3[´HÒ›wíiØ8J§éK+7f®'7¾™4Gxl||™†m±0F±²ï©‹âO]”m^ ¬¿ë4Œ€† L?æS×D¶g¯Õ†­É7µDÄ{‡M³1ÕŽlsR3Eá³=eKŒþl¡(E­Í`ò•QóšþWI¤ ×mfHŠ@ºoíPžˆ0+ÀÁŸÝSûº|À.YWoÉÐ88ìmª¢ø©¨Xwº#“î«õH‰aìÜ¡NÓ)µ]dƒX‚5Ùò”óÏåL™î¿]KþÁdmIZVäœý{ÌŽƒÄ«ÇìB„ä¿/óØš~åzwÌ&šUyi  >‰Ÿ¸¯Ìª/æ\BgPs× ‹+Ÿ'qaÆ&>ü;„!€¥G'>t¿["6µ’ÓÒ†ÛíNÔPí³-­Šúï;ù¼F=C,¤¬Þ#ÅN¤R_‰ˆSuæ" úrÊÔ„ÔjXýë´n˜úï:×.Ác€Hü’\1}‚§ôÙwe½Î±U¸‹sÍÆJÜyüWë¼ð†oGY…bÛÚ?±f’ggèœdžñ‡C#ï$gô‚žª@gH¶; mS¥üC#;HǵŽUgNÀ„œ«ì—Þ-óo¾]óÞ-øªºlÂRëhCvÉM‘ 50m³ Ò~–#õ;æì¬ïNzBª‘<üü4š›™ñÅS£(©›3ج&î…ˆ‡º”ICÝa Q»z üìèªhåÒ6Jì"Ì?J/ض0#ÙÏöU~µ‹?tÏNhye#ªåŽ8ÚÚyZ÷æuÞëÆ„¦ÏÚ•<6èùs£Fͧááq¬éÚ„ÐLÚ§˜•~j¤rM¢Ûül6aŒ "éY£þ î*¯¥ÄçÞrò ÖKWþeûDM“-;ëŸâüØ~N@~1½:DÆÃ3(“"OÜ[¨³ê´)š ‘ŽéŸ2á–|^3êä Nt9É8ü¼<¸¥á“ŽäÕ~åjÈÝà(+¿ãÊÛPÓ¹ýp?·ê~"…ÇÚƒê‚ñÖÀ0¸‹N½ý8q/cÁòIÕ²‚xÑ »fr‘åe(¼Å71ÈwJâdÍ–z°Ø‚†'> »ÍB÷ŒFv?¥³ýTå$Êͪ.Ò†7åÊ VkFÈdÒâŸìf¸!¯çÁ<Ù, ¶åGMlø¦wQÐ'™ð˪guBs`=²ÿ áZ‘=µ±»²iÔ†ô5gMµÁLÈpí–÷¯{Õ‰ø¸‡%{ˆoö!¾3´º¨¦W¨w»žá'²â-\…Y<¥W³ŽÉw[yv#3žú©ò? `)ã7`W^m™ò¬GµÆ»…¹‘¡¡‘é›ðÿ(µ;CFwfTû5åN¨“V@.9aáp‡ÀÎã1Ôgh¢'Ñv\'õ¦íû¹uäž^¶’1fÝÈÓý·¶cJW~×b<ØâÄ.ö—à3O”Y„1 ËÙ'ÄaÍ{í­wª˜Ú§¥M?«öÓ­|©Æ@F¹%[͹z×ÜßÄEMrGàÀë>k(ùA©e1µïû‹ 6³³E´ÏÚE7Í<×sŸ¿1‹Ž€ñz¦£è>rÏNÝ>^ Úú¯Ûüª ®+´©çë 4ªn©xy¯AÁ¨E(Û‹±ée;µz†Ó·Z*ó3¨Ù”®o-óÑòeÇ_0^€÷ðÍ0¢ª ¸u(f™â§›V«€:`¬åéa´™êÜD'Ö¼ÈfŠ˜ë\„Á+¹Bö³È6ŠMáüg0+È`jG¦’ù5Ý È£óI”¦Ï9`&lÖ#+f° íäWºõ†OWöåï9ˆ•¶• ÒÕ³•fhJSÑá’Èsõ¨°ihG¾¼ÌfFÝ•éˆ:,‰‚‰›Zxè]~Ž ±Ü5p<´÷}âÒµej5\¸ +äã:â«.hô¾Û”ŸÃ5¯ÃEqèÅ7‰+•™œ± ÐÐ"ùs„e}M|MYN)DŠ˜6ËôP9ÖµQ€µå¯DnFýäpÕh¼Ü(u ÷w7DyvAYRP˜7ÆÆó:˜ÓaKÍ> ›Ç·¼¼xN¤ý.Ó”0ô°"`!ö#F´C!/äú]Áuß.ëuã]åëF£ð=?°aÂúÓ®@’ijëz~•ÜÏ̬þyòøã}¥f¿G¬h;Y»K¸ MÿyiÜ„Z;/bä_ùÏ™‚ù•Ìx6.9EÛå뎶ÇOòg¾&‘ÄÉ0»ðG^Å”Ùù-©±¹¦ˆyÛ‚ÍŒÖN|jÿµÙ—m‹kÄu´§êÄkزÜKÔŽòñé÷¾y(hºÐüP-3!¼¹ª¢«>¼| ò–{ҕ삨=¶þAýŽMdàT ÷=dR‡P93cÖ|Ù›,é-›_ò>I”ÓfH|2Ú›1pG?“(¶YèÎÕ.Y¹v¦ ÞP‚cPÓ;2€êÇÆKiN¨·)51^‡aŠ ó¾[)Ðvfx‰!n,S×àê⛂›p©¼‹óÿðýåeÚè;U´_DßBHŸ•”Ö}ÖêTÛßIøƒÝ/:¾Œ-67†LÍÃýyþs¢ï Q oGÅSM¥ÇòÛêàŸ :2fUVÛŽ—ŒÃ„ÑoU¦¼t\ІÿÜUfb‹¹IÖÖâXŸ­«‚ŽKH±F.v›iUøšnàÏ ˜Zˆã?¹ À=…Ô(„ØCå9A÷û¡7û„4xf? \5¥JÌ|.hÄÉ%°+ù OÊ%ã=dýÿÔRŠ·ˆåëøXÃ9.¢Ã‡^qQÆÆ³’=ØSµx›ï$Soì)‰ªw= cád¶ÒãU Dê%œWA_çX5³Ù»y²²ÿ)57–îÿ 9Ç¿^fB±¯îŸÛL±Šj–n=}“Ñè³;*Êvs‘ù•”ŠˆíÅàJ”Ä_°&¡Ï·µ?™ ÄŸ­I[ ùPö$¹›è®Nåoóš'ü8¯ù¯Ýã7…½ÎóV{](L9‘ü>†Ç¨ ‰2K&0mù‘pu÷#áKÆÅ¢%1Õ”-ï½ Q¸@ÀîÅÅôlvéÉœÎÔaÕ¸:|îH/(ÚÝÏ7U ¬°R‡!6:®›ìFÅééHwpy2Û©©]3)V.Kƒº­wṲ̀¦):™IŸ#&âT—âY¾kh+÷o‡~DI%À8×"U¬ÂîÃÑey endstream endobj 120 0 obj << /Type /FontDescriptor /FontName /DGRFQX+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/R/S/T/U/W/a/asterisk/b/bracketleft/bracketright/c/colon/comma/d/e/eight/emdash/endash/equal/exclam/f/ff/ffi/fi/five/fl/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/parenleft/parenright/percent/period/plus/q/question/quotedblleft/quotedblright/quoteleft/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 119 0 R >> endobj 121 0 obj << /Length1 1718 /Length2 9715 /Length3 0 /Length 10820 /Filter /FlateDecode >> stream xÚ·PØ.Ý}ª §ÖÐf“²v¶É9C`l\ìœÂiU-.n'';''7=½æúËŒF¯‚º!Âÿr†‚€°'› öä§ê (¹;¸x\üÂ\œœnNN¡ÿ8:C…2@°5@• ä ¹¡ÑK;»xCÁ¶v°§cþóÀdÅ àxùG8@Ê [!U Ìäôt¢Ð ílÁ¼ÿK‚IÔsæàðôôd:¹±;CmÅ™_<Á0;€È õY~ P:þ¬Œ cvûÓ®ílóBA€'ƒ#Ø q{Šp‡Xƒ €§ÃÚŠ*uäOg•?^þº;×ßrEÿCþZY9;¹!Þ`ˆ-Àì¨Ë©°Ã¼`/@ˆõoG £›óS<ÐvZ>9ü‘9 '¥ >øWynVP° ÌÝ ìø»DŽß2O·, ±–vvrA`nh¿ó“CAVO×îÍñgg Ξß¿€ bmó»kw]ØÕ¤(ó—Ë“ í›-àãääâ€\ /+;Žßò:Þ. ?H®ßæ§ ü}]œ]6OE€üÁ6 §/4_7 ƒºƒü}ÿMü7BãâXƒ­`K-‚öú“dó'~j>ì0æ|š=.çïÏß¿LŸÆËÚâèýûýåP’1ÐÓeý³â¿¹W¯œ½¾l<¼6n>.€  @€àÿß*×ÿŸÚÿ°jÁåÆù "ÄÆ ôg Ow÷Ÿ2<þš ¦¿6†ðß'¨9?2ÀôÏä›pòqZ=ýáúžÿ?BþÿÆþ·Êÿmòÿ7!9wGÇ?h¦?øÿ t;zÿåð4Éî°§­Pu~Ú ÈÿºêƒþÜdU5ØÝéYEði;¤ ¶Ž_#ØMì²Öììþ¡ÿtáIÞ i8»¿56.NÎÿážöÍÊáé=q{êÕèiþûHYˆ•³õï½ãæã¡P 7çÓxqóñ|¹žÔäõÇd8Ø!ΰ§ÀSyþg(ÚïŽò 8¤›þ@¼¥Ð§ú7|âtþFBOð$à°úñ>y>=Nÿð¿+ã°þäp€þ†|<°è_ü“‚Ý¿ €ü/(àpür=iCþŸ´ÿÉå)³§ù_ô“4ô_ðIÚíŸLž¤`žÎÿ¢ŸÊtÿ|Róür?åýÏQOiù€ FÿW“¬Ü¡Ð§Çë5zêàð/%ä²B[œw¶ µoí¼®“"ódÛù$6K¿£ŸÁÌæ»ír¿ÅDIe®}¼½’JíÃ^Ù’eº”\¢º÷=hkB‰hOÖì¸óûež¨5½ÓöeŠphòÝTã *9›Žä7¿{W?½ „6¸n%ú|WwALB¼kÏy¯ÆÁŠåñðùÍoµüÊè¿*fØbucL‚Jçè ,s>Ó ÃØ(ž³àžxaÍ]^ÍâæM>R)%²¢ùÆòû­sÇÝ|öùZ¥ÃíÖCBGbDLp‰;>Íàûj/M‰hÁ·¬x :žþ™°kbÃ)…ËqÉgGMëºfÆ@?)ÌDBÌ€¯ø&f­]!É©”: ™7çdEiÍKñL ÍÒí¥zøQÝQ ‘O ]h§÷6¯Ñg…‰ Á€%BËŽ µ‘ ÒÏA—×È€Y´c;ü+–hÕdØêëØWüæV!L Yж®ÀMÛ· žAû_u¨GW¼FçN•|f"=ö·XñO.¬²‡º¿M¸Y;‘˜Á¥Ò77‹íÕµ§æÉ ³’½!lÌ¡p¿j#C"¥M}›•߯Ӭ>lnÚ°s€Îœ]»×1è‰v9ø} ¿B·âQª¶:_ð-Î;T”ø¸›øˆ wBù© a™U܈ ¥È¸¦ðÂsýê¾ó‰çǦElœ¢ì¶Ú¯°0ÔSðŒrÕTŽg âøÈÌ„šúñ„K•ÛYü~È¡‡>úöG–"+æe_®‘äÞ¹v‰wÙsš¥{¤5h>0ôï½Þ‹ê“`ŒŠxj\ÂZ· á Å뿬¸ìRriëÕÁùÇ —ô¬VÏ­9í|_¿uBþÌíy4ž—vNfË7€œ6rü í&„œLH>Ø.0&`òè|ŽY”G{šøE˜uw%¼Rÿ³y"JW–7—Ýôda[¢žŒ\ùnAçfj2Eïcr±ß²Ë4iC:½ŒŽ¼K"ô™Í¾cÏÑ~“¬zAž,:ÚóÃã^cÖ.V”K‘VÝ|Çí뻎¹Ÿ¶îuý[G„–ëöˬ0äðëÇ䤂“»yN•Ì^Œ·[ñ Üü¾.6óvXQD?'a£Ä‚ïmO)T+'M:ÔPªÎ1O)ë%ô²Êfˆ¼f² @?¬È\À%cʽ™Xûé+}š „Ù¼ðFsÌéïXI£œ4ñ„ø7¾( weÞÂOÓPp޵fݰ~Ÿ´þjúj? æ…Äp’m)Ðûý±TIôÙvK…rü¥aû>8•5$惒W-†poÑ·•ÉcC­ÝIÛá¡€Ú©}ؾ£î-u…)cTâN‚ņjíÚ/áóÕ4S¡ì…|ïP·$RÄ!;Àgt´=X¤~%Þ¿º¥”Ó(•Çœû)£2Âx¹£ ?‹B7ª_ìÉû8–˜ï’À%å¤U+O÷gëÃ0gg{^B0cÛzvyƒ ^ê4LñQÊ‹e¼ˆnƒzÅ–J X!Pûy¥ˆV¯Þ³#ÝõÀC jÇ’XÕŪin[<õ½evü0dâÛ÷ž“ɯ™%ÚŠ¥Ÿ¯å§éYâ ˜m:dç–_*‰TY½@SëIa9±+OÞSòn}Ñ/;i®Ÿ2R÷Ý£ÕDÍZáæh™ãÜëdþ»,~$cTîŵ橼J*£uÝ‹¬MÇN ?zf½áVL§“ö…Ê \\1†N Úná…R\¿[F¨9EÑxCR&’wÜëER?_­àÂXþ¦Ü±ôWéñöŠ?IQ‚ˆ"G2¸ÓIïøQ¿Jé13$xÑ%Ÿ|çeJ"ïdƒû˜ ²È±ct©šÙbõJõ˜0ðox3Ž#?σ#C!œlö>¾¹`6/…¡÷Ôå tÓ+ö×2?ÞÌ+ÛAòTÛ|‘vÍ Œ5žÃ—~Ò$Kºº[½Óáð 3X¯¿(qUA-: P­XüR½…Ÿq½˜dcDZ°‘*<-ô1\Mý;ñ­ð꘺Nîs½¶¦j¥æ£Þk¨]eÔ¦µ> +’ ìïËo0øŽ¢žM²Å'˨4ñ™Cvö«]mƒKk9(¡»¥ÁØ{¶µôÉo¼UÖß»Óz¼òç¤î­3ÆÚ' Ze eνv޵¶ˆ,,”R½Õ° gƒ¼KŒØ¯…!&™Ò6G 9Ú2ŒÎ‚ÊÒ«±öÒ,ù¬ñ¦YÔeQù^ÔëêØÝ’|uôèê 3g†»™#§s‘ÎÎ: ÀfgjUCýܹ¤q§÷&IÝ¥`lf%mÑþ9“3;n !°F¯ÊPŠiq?)’K) ƒ´/eVl "!ë·tÅ)Œ6©½ùÎÈüÑœw‘⃿²aA-k³õ¼¸™ÕªÞþ!\NY£!ØË‘{|<æ*ðè5@ª±z6‚SB|Å÷Àòm(4ÝkÒ%IÕ+{/£ñ(Þž¢ÃW¼\>4k…,öÑ +Ä ® h×k<%z]ú7š'’ûï7`Í9„Ë=Ôæ\€ÎÙkhƒoñ/kŒ¤­—É3’g¦¬_Þ'vEK/cù´]Bp|÷9µukº•OcÈ_2|ZÚ'£ª3!ªNXÙž*äRyUW9‰K«´Ý×|ÒsþÐÚJáBDÄI0ÀöÉ L 'QYÀHE†X”zaóM‘«Œ\bÎÖÏ8*(ÌÁ£˜+_·„‚ül¬™?z‘$"ã…hkAC9ô pýÖ+÷‹}ÐG±È¦×µBÝ;>ýЍK†ÞCq@òjÝj.RQi¼Èºn¦–"¶‰p”±l*»1?9ª–jÆ]ð€_Òžc oMqyÝ §œl[$"ëõ¤ž2ÈÊÂçI¼·Â&Æœ¿ÆÇ^NÎh`³^¤C¬­VõXCùNñЯ1ŒÁm æF ¿õ€--Äz2Â8Ùg;4Âj½—î<çLõ³Ô<ì0Å™sKôkG*|d=ùð‹^,xÊ/ÕÅn3´®‹jc:kïFóØ73ýö-ëÁÉ`?uö]Y!OKmøv·ÁW(> óY#MÔÍYܶ'ðïæáó—a¯/OX¢XA3¨ÚΖõI¹0_KѬA ÇO¹šá:xà ‰AʽêÒl„jRåbBD¶|bGúf^ãLH^AMK¹²¯)ëz—`‚DØþjyóŠq2c¡)IQÝ•YÅTÝÔA Ä£ð«H4¦þpïåð»{euð¸Gœ´À¢¨%}P !çðÁ 5¥ÍfÒðRG´ö[}_=²ùg5ÈËæ/*0×J›9Ò3õ­Ð‚^ìp%áËóNl$ãÀhŸGÖkߪ-½X,më? ÎÁf1“ó'¥¢•ÀÐ û×Ȇüë8 Ô©?Àþ$åmÒs9r½!U”òÉvlF롸 £àè<¼XЍÈy¿o¸B (o»1Wm Xï5; -l¾(MÜ#ãhø.rbãäà@u-wP)Ñ/ñùIôbŸg·µök_Úëï¥j-Õ¸ 4¹¾8츶wrD ÎâÉ£^9!‚&…Æ‚Q÷±-Ü©øèH­Ëntu3©‰ø/—aµâ\[‹‡F>ÃO»yAeÛþ ÎQ÷—³Gñ}t+N#&½Ø]™î38wÑ%~-ùô«±¦•ñ§ÕíF"úË›þ޼Æ|Ó*1 iÿÐ_+æÜ+Ãs÷—øx Cðèû" zy…$¤bÅ´ë˜.NM›^ŽöÄ+Ôè;*’µãì\µö¦w\²Ü”Hì|„ GøRÞ>!â's€CjÖø¢*£Ô›Òe-qiòö5U3ÿ[Ópk•Wî­Lž#ƒ €ÏòJ,XÇs´c9ýð÷¤½†œ»‹Áj1Ê’j’1ø?mj¶¿Þ„p¯Ö½ëÏC ¦‘ÂéÉ ñ¼¢§'x‹Þ ~l9 ¬)äsø,üЕ#ãéBS)bZ'…x[úÉB[Ëä½!"f=ûŒ§Ú:ѽt}¢ÃŽ7^eÀª»í>“ 椚@#ÀkC‘ˆÝVd/™SÕ†Ô÷KÒjrlˆ¶§ÄÅ_¦žóåöàÂûrhó.$°ì­„ìU»vÞ+L¸5¤ t$«I]?õ³2Ö6] º&è´ØbDTB $á¾i—Ïb<µèåúÇ܉T„O‚ThìË –¸æ[µy#0JHÕÚÏ·‰¶éá.b¾á³æX#™œs âL›Üõ»f‹}ü²uÆ›¶ôU·jmXÖŠBùûPÆ­vk–ÌDÓßjñMx°yÖ°ì¨b}‰¼9Oã Ö+,ÁGHl+½² Ë'É9 Ð*Gûn+Ð6ñÚ¡3…Áå“Ùt¢ ~`›[øÖ5/yÍä˜Mü+ê¡Î[dtÕô·¡ÕõÁÇ.s‘>1¾ŒÙo#½0¢Ž X³’Ð>‰þ…-$ ©e-öޝ9**PäuËé;vaÕoŽ#]½~Ç¥ô©¢œ%v¡®»\Å8³e ÖÅÊöRòK€“ƒG™´à×±a¼Ðð/„t\Œ$ãÍpBƇJCóÁ.-•ÜïSƒ5U*Þ¹ô¼h!´–ÜeÖ2í}´“X2v s‘ =ô*Jç|‰èùMÂÙ§ñ,»eå¹â¦»Ú+,ÆC¸Æ“8¢A=ý ‚jþ9P–AÇ9Çk·k¬#¦e!®¨£.Ξm·ÅŒÈäŸîHQß.ˆvë܆‹ä_xGš¼#’>Sz–ÏİÖã¤EO†¸•?Þ’e‚(ÕÝ;5~&à"]•_¤RŠ[¬ôªóò ç2[+à¦Èé‘_$$®ÕCc“O¾ö×ÖŽ1=Vƒ)}¼:,èu|Ûα£M§”Êf®øè|^ØF3•ÿ{ÉHóÉ–†­G¤òs}|†Ý2f$Ú›O uY1ów¬Dt£ ÜßÀæ™å¯r|<-ÈM¥÷‡M­X÷ƒ¦cÚC›XBÆnškvTÉ8Û:Ã,˜LŒ< ;Q-ŸÇ;µë¡•¿,Ö1î,¶|cÍg߇™7ϼd ,éÓ`^:Ÿ*Í…²Ìâlðꇹ,Ð壓ëúR°Å¼/›~EvÈ‹ 'u]ÊÏ,DZ¨û¹nEYg ¬n|âÙ'E˜ío§ÛøÎqî7£ªuÛÂfC¶^‹Q.ÂÿMõÖo0¯ß6à,ð§!À½Â¸ÞñŒÉ¹sñy.È–4ÓQëÐ3¹¥µw>”50æµË…õR£#~¼Œ;~a®¹)¹BüšXÎÐmç~ùŒ¡‹R .œ;&8vŽþ6NÈÊDƒFž‰fcp³âæþ²9|ãi´§?uû™O"yL^:+Mòv Y+þRkH…Þ¦BtfÒåIs,—üRÞO¯èMC¬Î}†‰™ÊQ8·¶~íäÊ^鯆|Wa´`!!/²©Ö¬¯t=BD»„·z»ö<û7ÖuZ‹_4_A¨C‰o¸¥²ÅwÈC̾î‘hùL‹ÙàDí§1_ÛÚK}Mfôа|¥a-Z»Ul6÷r–«¿ÀãóŽÓt%êÄzÑšzíø‘>x1Õq! ãe-ž[)nVµëšMgrü¨²—IÆÇìXŸfÔÔ‚eÈ(¶úv©ìˆ bŠü¨pO”wÅùfÞ°ó1)oÜÌŠ³¢„‡°òvx–,#XEʉ›zõù‡±!UkgÊ%j›Å/œ˜Ô¨”ÄÓ‰DêâR6¿òK-=U(?ª5Ž}ìÏJÏi^{uì=»ûÕÙ XO2šñéÈëÈܽ«¦x׎ â3½ ÒÊY¤…Ü8m㯪×Á?,†{öyª»³¨ _c:<ó²ŠÎÉ|Ô+šëªRès¬ã}7‰ÃsZøéìláõçì‹Âö í,dÖ"I¬®èD“iÓžKýµêBµÕb—E`ÿÎ$›øNÇø¯ŽY4Y™XŒ&”=i¨ðiÝ9ÎÎÜ¡&6¢XÑö³Ä=üÍñýf2BA_ÊD'iÝ/-"-š°²­ãšh£µ \û¡¬&?CM^ª‘oÒ¢F É/?ˆœ·wne)éX³“ʲð¥›äº;çÑÊ.>ÄP/Ô3­£b¾Ùò||Y,ÚΨï[Áß~‰E™pß:x”í¼ÏDO؈^vØõš’£2VW´¤•†v Ï‚?ÚÐBÏ»6 ­T±þ:¢œü<•‚] acét!.—íWÒ¹@÷á©¶ˆ”z´ [t8"æöñM)Áü±{ý-vA¬âò‰³éË€#S[ ©-b«ï£D<‰‹µç— Ç ¨›Œïànvóñ¨õ)ωË<³17§æÊÝ1}Zâ(q/Z:Lè5žßÂWÍ!êìM¸£ÚôG>÷’zö½´2Ð{•š8'ü²ÌŽyÛ™Ð³ÇØ)Ã9š(ª0t,{c¿±ŸÊÎ5})â¢Ñ$gÒÂ*]sÝ„4ãc(^ÍŽ }"+ Íêv°J%Ö·\…ïQ£èZÈ ÇògÒ>4cŠÇSºpjýŽîÒiÊçTç{„Öøw–šø”›ÝšÐ8 P´§éÄ[µÚf)Dü™†1I‡aã„^®4]-Ø2ÅŽ<í÷–7¢ìœ„ê‘Öšuª8D»U\F9¥Y=dŸ‰uóOÑ‚â’Dô¸¾›:=cücÔ6µ+D…©¦pÇ«F{¾†Õ 6Kzþ¼÷·c“z¹¬G`­WZiub-F¨iüÊÍjµo›³i]ÞY»°‡»R´Ö¹ÕØBŒý9kÉÛéqdåj©„˜!G¬õçÌà…`ÛpmLµ½ƒ£Ì¢ÆJÝe§Å ¬ÖË £Ñ$W»Ô·ˆSjCãC–ºÚW¥‰1D”ï^ŒnLÈÇÏèÃ7å®gœ>B‰Šwêvz¯ž§Çˆ›ÁìmM¦3\[VV™¼DYb?7[;OÐæW,“à®åä¦ T{•êé”#áQ.@–|è«ßû&¬Þ?ÔÇÛ˜ÀÑÔ2u{å¢ï¬uAï±9#Ëøma %æËt¨¼Ð$Õ,5§Téµ£Ö4=ÙÏ „o«¾T›ïZpš¾³?ø>Ÿ{à$ @T¥¾f3â»F³Qb¨ß¨¯Ï: zœŽê7"zW8SgDøÍö3õ£ìX®Îp:¸…f®–¦d/øN/-Çßs';"Bƒr¡ÈíÚ¨FühPhIgi„¦—äT® ( ®(ÆTæ »s))¸UÃa 6áÍnØ£âWéÑŒoõì©í×— 6Ï!S-;~ëÊõM1 O×ÛøN¾¥‹B%2u[ ÅÉmg”3žx#ª„?e÷»÷¿iÇÛ½^tIø@"zxÛc ÿ`3ý®M`ª¾ Ç2.pŽX)ýžB81%¼iMMû«KYÁ›å°è#Åg¯ßs¦íó09éåè 6ìú¸â?J^$*µï"'!~ˆû}*1Q²+í²1Ø^”ÅëÂøn¦¬ôÏ(4’Ø¯Ž’ä‚ÌË4ë\#3ËÅkU.ŠUÒiÚç3£*—+`ä ŒãgÐ2$r‰Ÿþ´Ÿ.àè4]Þ—VU)¹, (íŸZf†¹2õ-®V(Àó ¨nãËÛÀŸ|m9ëKFÝBÀ‘ñžüPäÁ¡w‡ÇÊÔÅwÒÒ­€%èÕ:ÞIÌws‡µíÃÄÔíµ¡­¥u–DUŸB+qN¸`Š(ÖŸbÙ£#À"ÆîÌ‘‡ùËñû…KìÈb'è?¨È>+3;êT3¦3¹êš«Šu&£$Rå:ŠM½‰ÛZEVl¿ä ÊYµ¥ìõkÛW@ünÞLÛ}Eê‚K/ˆµ3üÑ«I < !ˆ‹Iý¤YzG*:)LÖâ3ŒfˆùnÙ¢¨zÝrC M™E°S®¸}ûÀá#ÏÀ?²êЬ­½úÄ{õ¬Dy°—Ì*ÒM›ÃÃDöWâ!•f«Œ‘K*Fê/ySºÕ·¯ÚhCRйÍ#µóöú@n-+ú¨öýxŒ.$?ß°*º…Y“/R¦DÑŒ¨×ÆÇX¿üà_D+öâóîæôÙYMC”nýU—/Dä‚J"DÑÒ¾uj¥ÑZƒ¡ ^}¹àæ‚jÚÔ»;µ®W',׋AÌœ±"G5c¦pnhxf;1ÕCaàîŽ29&Š=‡pRƉº÷K‰Î`ú˜·õçe‰"Mž5}­tòü`ÒdÖDH^ÞMüËz“ÀÞø“>‚Œ¿ZZæ|mÑçœb Z™×â/Û©à*™vé_+ „ £eðÍa"xlFÂD¸f`5²“/ŒùÆaÊYÍaÝ ®0˲XCN¿ãyò“¤´]t£Ž3¹§ÿ…@½ažNTÄ*ÐÇìãÂòÞ´s¡Tx8_qÁÊxÚ&Þ¹=É{W÷¦¨–å‹ãtJëð¸¤Â«H„>þš}øiV3òpõeãíX´í¦ï68´ÎTVž—_‹Õò¤í `µÏs<Š–h=w{, cÍ÷gû¾›Ùµ£ECŒùé€rÞaÉ! mžZ kË0æÍ$7ª çɆB{Þ*ž,ºù¹ä×8› Y2'YßÊmðpº²»¿îž[æÝÝËÉϪêÆ%aô-¤<ŸZƈTÈÝ`€–Ç’ÃÎúùgR/¿ì±6jñg¼fÿBâªâîfÃÙ_ÉêÜíœ-è¡JW£RzEzéHS"ÂË_ð ¿7"Êøe¸~Ÿ‡ lÓgž¤ðöo8´‡I?Y“(¹œ´¸3ç“W’j<êH™p8=*’öÙùÏNÞr[%ª›¡; !¯®Ò«šjE‰ ÑFN|>Ÿj@3¸4ý¾ùZ¦pvNñq~Z…|Å›¬éFn&¡2Þàç’’@Þ,¤ZB˜ìmÚi­UÂhñEè j¿òîT*öEš+œ…¼ý!Sª @#&÷óZ&bïlwÝ¢t<Ù;ÕÓ´w2KAÝ(:º¢XH¥f[“C‘$Ñs ‡®‡ô§ýõéR!Èô×r®içQJÎo„ R'VL¿RF_~áB­ ,²uYÁ®ê™{îd…Mýø,¼–ÌOy´lÉy^ôÚ†>¢Ö°³MþG¡º ~ÆÔ—îâ8½"‡†C‡ê%mJý)†Þ öÑèòNž ,~‹ëú•õšQ|6^{h˜ð­¯›Ñ¦2öÂÁ É›…Õ¶«\¢å4"¥í„ï³ÂÏÙAê,¬,Ä¢¸NVÌÛüÓnGÈKäž)NEþ¾#BË‘:hƒEÛt8gÞ*{¬æ&Ø- ¯˜D¼‘pE©È¥±áí³6Ãâu ꢭÓYk{… Z ëx‚ûÓ$عÈ?c¡œ2/—^áš Î=šúÖ¿†ˆ‰¦ƒÁ$Ɔ)¥]n­Ét§`E`cÕI÷U‹~içr8R¡ù:‰€¸ƒ>qÇa§ýD|–#³éuá¬ÉyDâ 9¿’ñ=ê! 6”ä&C±¾ÓERúfé’+¡euÇ^!ˆUÁ_7Ö!øÃ©NøÍï\Ũo 5i)a·YÅ©$eX²Ÿ¾ž—¸à2Çß{Î?ï,›½ãø!GmaЧ]È¢è-›ŒFêUü¦°µq«ø¬5•ö6Šðœ{ŸÉÇÈví6ö¥5F Sù´“üEæZ—úR8Žåj)ÆÁ¾íá ªMw͵u!Ýûh›OÄý ®?µ¬\µ¶8i}z7p½ˆ—F—Òr7cíh¬1³oÊEŽ…‚ŒÍNÁöíêÐUÌ –Ê;©àj}2@Q¢þ·÷ÔíÛ0*¯‹‹ÉM½Í¯ëüš)'ÂWïÎx ›3• õt³_Ás-gb–oÊã˜ÜIÞìd…`‰ÓÔÍ~D{e£Ûc>ûé¹-K!ÍjYœ^yîŠà û,‰I.±E¨À*¥ÉÁ]åç!Eïì ‰ƒVù½å~êq€’Ê®+qWóíÄ•üGƒËOèÑ Äñ4›òó×i¥w|:R†nïå ØDiŠæR’HÉGa¢.† hŸ­næ;½X©éæZV\4aáŒáI°±Ã1R< j>°Ø>iˆVQê»ÊJá»5ÍnÌ;}ù¨Z¸ Æ‹P)½÷Diá3WB$‰8_c©ï³ß¥†ILÀïh£Âp^­ÎµŠ@ÍÉŒ3@)Êu´ä!". Ê¢Ðã“Ôwh+uòrSÃ&a€æ‰—F’Dâ7l&ûî­ âZBcÅdCI’cH¾ gÔ“2o·%9^ úȆå1@ž¼JmÍhdcZÞ6J¢ØO’Êïg$”iiO÷­Ád\.ìþ©aʨ̟mWéR‘»@×a@ž¾«r­B$IY̾a'×±ø îØï÷T(µ_Ý;ç8úÎÍföNïòbm·oÇŠ &Vµ:<>ëD™D”L„vµN FøH¼|Ø;!¾ºž@q ¼P¥A7ôÌà> ~)ùå,)®±”_·è£bø€eÝý3BÔÀíΘ¯zÖXõ”Ü·}„ç;‰‚s±©Ë"Š¥êüÎý·ƒ“ª÷ØÕNL[=ßé…õ"äÎ1ÊÌ~ñ©}TVî2ËÇÌ’Ï4Ý, (†ƒí³CpÆo£Ö3½Éž“í~¥ùZÜÒä Ž­ÔŸ?õQ ¹©•Úùr¬3BRsÒˆõ‚vrpöøÞQ_Ï¢ÄfN9%:‡¢€qMÿ¤3=ýR³Ï1 ˜QÃ-¦z¸¿G»Ðray³²ºdÌŽ¬ •?cŒÍ.ÂHµÔ÷µœÿâ^ž]W2Ä[³£_[0Ér™Œ zC> RBÑÉ›ÕpÛÝ6vY}®Ëëßx}o›ïð{ìŠ5!Ê·Ÿ:Nò¡CÒ…Õêõl„]„ºÝwV KŠG€ ßl²û¤³.=ÕFN¯Ü+ú„嫺Â+©àÜ5n3uƒÌõÅÇêÇŒ †)*üX­¾x΂OÏVÚäî/¹§M+}«¿¸³ü'Ê ¸K}ÞÛxÕ¥¥³¥;9‹NÜÈÚZœÊn\•LQ²nƒ øÍfC¥òg”Ö”+®ÓÏ¿´8`2×ÿr«à endstream endobj 122 0 obj << /Type /FontDescriptor /FontName /HADXVU+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/C/J/M/T/a/c/comma/d/e/five/h/i/l/n/o/one/r/s/two/u/w/y/zero) /FontFile 121 0 R >> endobj 123 0 obj << /Length1 1709 /Length2 10341 /Length3 0 /Length 11429 /Filter /FlateDecode >> stream xÚ·TœÙ-ŒCÁ¥± ÁÝÝÝÝ¡Fºqw·$¸»w×àî–@pMÜ™™;3÷þÿZï­^«ûÛUûT}ªê¬¯©ÉUÔE- f@)Ø…‘•‰… ®¨ÆÊ `aagbaaC¦¦Ö¹Øÿ2#SkœA0ß¿âN@S—›„©Ë OȹÚXÙ¬\|¬Ü|,,6Þÿ!N| S7@‘ ‘©Å!žN +k——4ÿyКÓXyy¹þXµ:ÌMÁESk ýKFsS;€:Ätñü¯´Ö..|ÌÌîîîL¦öÎL'+!:€;ÈÅ t:¹-¿”Lí*cB¦hXƒœÿ´«C,]ÜM€€ƒÈv~Yá ¶:^’ÔeÊ@ðŸd…? €¿ÎÀÊÄúw¸¿Vÿÿ±ØÔÜbï` ö­– ; @YJÉÅÃ…` ¶øM4µs†¼¬7u3Ù™š½þع)@JT`ú"ð/yÎæN g&gÝo‰Ì¿Ã¼œ²$ØBbo»8#ÿÞŸÈ hþrìžÌVÖ q{ÿ,A` Ëß",\˜5Á GW ¬Ä_”ò?6+  €“………›—t=Ì­™‡×ðtþádým~QàëíqX¾ˆú‚,/?ÈÞΦn@€‹“+Ð×ûߎÿFȬ¬ ¹ À h#ÿýÅ ´ü¿ß äÐgyé=VËïÏßO†/íeÛyþCÿ£¾Ì’šZbâÊïþTü·OL âðfdg0²q²x98ܼßÿŽò·þÿhÿêb úko,ÿ”[B¼Jx9»ÿÈpû«+hÿš:ÀgP‚¼´2@ûOç°p²˜¿|±þ?÷ÿKþÿÚþw”ÿ[çÿ\íìþpÓþáÿÿ¸MíAvž^:ÙÕåe*!/³þ_ª6ðÏIVZ€\íÿ×+ëbú2¢`+»¿ä,òZ¨€\Ì­ÿl¡ÿTá%¼ T8ƒ~ß5FV–ÿñ½Ì›¹íË}âüR«?\À—qúï”’`sˆÅï¹cãä˜:9™z"³¼´''À›õe@-€t6€™ qyYx‘ç °„8!ÿ®(7€Yü·éOÄ`–øq¿øÿF<f'€YóoÄûÂ4ýñ˜ÍÿF¿U1[ü ²˜ÿÀ—¨–– ¹ÙÌVÿ‚/~ëÁ—´ÿ&¿ì×î_ð%±ý?õ%1ø_ð%1ä_ Àìð/ø’Èé_ð%‘ó¿ €Ùå_ð%¯Û?í%‘Çð¿jcîêäôrgý1=/…ûþã‚=€æÈK‹sþ›ºŽ›Q"wÆÝIÁ9ê]íT:Fï%§N×;4Ä$ºêÌ u§+Ѥ‘^Œ•mIÚK‘e²G ˆám ªí÷>Æqj3»íÈ_§q§ ŠÖ¼"fÔÙóytôÑ ´…m…þ,Gëèʃ¦’}ãÞ/íQ?PömÜ.x­Vh°9wPèá“À^bϼõûž,‡÷Å»äS4F Ó[þ£\–áa$òW9rZ£Ý€>o›IùD‰Ã«¨·À°Î±½ë_¤S/©ñ¤è=íP]iÁ6{UbHâ&“9 †ÞÕgŸ›×#¹™ât\Yê8K×ZkÔ‡7N)8È´|ZzÙe±¾ ®ê_ {ºN­ò%"úxÅt1uÑÖUæ„“¤—ós1à¿©¤qw¡¯fô쥙D#ýêh'ÀzÏf$p³(¥¡¸u?¥UT¢(³I¹óeÆSOº3Ñü,†\£Fg3åÚ×ÊÖ‘7K ï7îÒ»ò‘YÔ;âFè&L-ôè'¹÷í¶öÂÁ%Gà÷+z-ëÙŽuŽðFY÷dÔÑxl žÅS]v-¼ïnéC3U²¤L¡éÕÍ:CE©¾&†‡õƒ¡6í®˜à3ú2³!$“5¦Q K"®;LZ‹;bY5^(º²Å‡‚'†ânõ˜„뀆w9V¦“ËjÇ §öÒYtþÄs›õ脽ƢPˆØÕ’ &I«Ã_³ Çû£‘*Îûü°öVµÕ™ƒÇË’êÓ/º}ó©¬bE,[•À_Ñ2¶L?áZt{3Ù 5PÔF=“T²[¾µÃ¹šUÌ8ù07{Vû¤*‡o!ÎSÑ-~¿¢½J É*aWìÎã _?%ÄyGI_FÝ¢ º-1  òâ` =`ù‰´ F›c¤NÊ®ã}èl|Åœ{öÕtú…Tµø2 W©'ÜôRÒÂò$<:*ß?kAb cŠÑÛÓ(Iµ 'qÛž,À=VMÒ…¯ó{²!_·ÑÊ›K)/cY÷Üѯ–"÷{þ•Ô)µ«¥: Ãdk¦.ˆn¹0g¢'Ýɧeð±”éyGg)`jóÜİc.«6…è,‹“Á†¾3#Å–ë AOàöÄElþÄWbQ¥uß7;¡þ¥ùøF…ÐíXÞ<)¦ÔSÆò5ÆKì¢H¦ðü™†ÛFïû#ð„ló?hÖû^æAŒ?­ºzy#~í$@ÛÊæËñï ?Œ5!&Ádûįò ÑÌžq&âã" O8K<íbˆcÛ£)’ÍmÉî` (qŒ„6%I¦ÿl„^V˨7iÿS\†U„HJYèõës ÃÚæãY–ä±ååF^µEl”-ã/=Ñq Ò¢WÇhèqE#íI%îÅÚº½kƒÁÛÇ.ä¿|æ'­‡ß!Ò3Ì<~åE×±öä,÷7g§.´N?"Zã¶aPö˜ o¾\IÕphYî’ïSUMò¥9בÕ,‚úèà‡W‚¿˜ÔSú88Ûi¿rX™"R<tK¼—­ .j‡‡[t8â6À3ÎÇT,ãm¡»z† eTø%Œ0†Æ”ÈÖW÷JÙò£\«‘p..ËT|7ð‰=‚ÀÝ"^Pô ìÒçkg²ÈI„i¤•¿®_‘ž;_ÑX>IÙ±½AàþÒ7Ϙ÷é©ö©QÙ2CŒ®VsFÜR’‡©UV›Å{mkHzþÓqhâ÷WØbµ×Q¡bØj”Öí«˜hŸ·ä×Ã/?+^\-ùhJD­¥•/DÓOÑ,ö̪xØwe×8‘¼ÙŒæ™bÖÕ xÃÝ©"ÆìŠƒâ~£Ãxí§=ª] h 6º¶–üjèù Ì$Óà/Ñÿ^ÀÚEº°-*Û%_„­ÊzVs·dÖ®u ;6RÛÓÙl<­³‡·ºÅ5!ÒYC÷ÁZ¶`ßÔ>ÔÐbµW  … »rú²ìòc´àSAëU¶k«Î¤ ä|Þ¤q-«Jëv°È„ ›YȫɌ{)û‘Æf_–!’ ´ ùÎq?³ÝÛÅ´<7ÎVH'¢]Y@£¤Y¾JÍ?u¾úŽEP’®ŒžN%eI¿ÀJЮ~Q}êüvȽ#S77ÞNU{Þ“þHÈ€bݳf(Àú™œ)ÈÝtBüÄmä­VýÒÚ2gc7ìñF¥T.L”c—jDá*1ß²ˆ =¢ø‰ºC1L)¦´Òð ˜ApÚd§6çÅÈDÍå¯d æ;‡…·:$¬Ì7M»è:Ïâé9-ŠÇá: ÛÝB¥©«y°ÌR ñá&DEwNcÝ™ ÇÂÄN²[sÞÚçP³6µ™ö)“Ü”rŠI_CüX2Ìø(Ânuš%6ÏœœÓ/o£ûÁËl».6ð¿T‹ÕŽΨ'OäjÅçûVÉú*tŒT×謁³–O_ÿžPÂA«ÐOrœxP‚ÃT\AÇÍ’c—b£5Ñ6[”½gðíQâöïû¦•êñ}µ%16t1pØiU ½Öb 'ÑÔÖ˜F6 ™¸UcñÕ_åΘË#ö¯Þ4hmp÷'»³Ïhänjn²IcÃÍù Ëî“);^ê'-U®˜„¾¾áG›I·#ÌLã.â2fSÔÆ{+ïOå3ý…†±}b©ìX*S­âƒùêDnãÿn§ù3\›@mv¡êOä¯,½j‚}²Óœeú­šö‡l;%×n1Ì·qú:)Qa?ÒdÝTÒ°ó÷™ºé: ¬T2 +›Ï<é1ð^5Ñ£DëwY}¦žNøMù °ì‡e7ù­ÒK;u’ï޹ؼóDa7EpÀÙJïY™àÝFÕø½80?ݘ2[ÈÕ7Žw…×z³”W5+ûåÿWkD¨;ÅM瀨ѻn’_:áÃŽ©éö0ŠÞÄ%¢®Nÿ±¦W m—6äD²éÌcר ÷w”"9ÝÊ ˆÔ]bí<óNÌ•à»}Ó WáÇ :ZÁK¹}zíÉ´Ûƒ1oI¨U°@ˆÐ²ÁZŒ ”¿D½Iš£çåÍâ€l±@%’Nñ~@ù#Ë9àrIBsPoR–*‘íFùA˜NR€]Û•®§pÖFÉØR—bÙãö²Qµ,°h–‘6°9EX¸A å½]¹´>EÖ„8?U·^ B™ ü!m£êlrü,gSØšFùy·åÔmËEºù¸D8¯l;œÒ\þµ>³xܵ –²0y»±ºÎÑ›ái5Û®àµg-®Ä´[D9U{w¾%Ûlôî›e=3O~r(OÙ!ítµ–÷èƒóœÑ×ÁÆW ‹][lIBãÁ§ËX\õY;—%0˜ÞãA,¡—íY†oë ’µBü*&g»u-V”d›^¹¡¦º ÑNUjâPça½#NìÙíÕø<Æ-˸ Öd•nÊQx}55\*µò¨$|04Bc.Lb Û¤¿à†zkî®Á?áe3¡Ñ²¢—Æ%;~Óü݆jë8Mµ­ÔlÑk$ƒÂ%b´±µÅ}¨vhonvh\©H›7Ɉ#]ªü)7#¿½Ù"†u_ FàUìŸÛK?.²xµØÏeth†» ü®SSâÉ{K™ =óÝ>¨®BRh\Wœ+^èËa¬¦…ò§[Yû€ ÉÒ{ŸìJ¹Èíüà"ßWðv8åvNꈎžô>9[d) ã&$‰:udÀÝv‚d¿ýMé:Vbqx¢9\UeÂ,’hƒPÿB ²Ú›üàt¿aÞÎaâêíŠÝµm—û¼!ËhßááådpRp†g¥•59l,*;ê-xF$ˆw T6A=Íåì[|)*M)ÃJyÜ)×0lVåe•juœµvŽÉ„mdý—1'! ó_æ7a¼!³›¤[OH|­+öáo_}M v=¨B%w0Îî1wÓ `x5ág ZË%+ø±ˆÍ€G2fx¦é“v!$VŒq¤.ì¾$ø žÊæyMó˜†Åó)`G¾/Ïî=‚Iž! .šŽ‡FòxûµÝCô³žè w”VIvÚ=Ó*†uîÂô“î§`WP…çA’B,•OÏ-ûýV²§àúÅpçǼ…7½IeX„¶È«MhYüè–1?Îufضª˜y›l6êD´îµôªæaå-¦4ç[=(ê¨x,J™%3â­J~(´’$ÑRƒŠq›ß/}:^sk¸ŽZši—DY˜o?F}‹:DUiaJ¿®^•Ïíǵ+HäVme°cQÔ @ŽT#ÅÃÄ ýÒø˜z‡ã|L¾Ê ²Õ¬ªp>oñE¢-\a ¿ïñ G)¾‘ÒðA %*ë'-àÌGK”¢îÕOYèsÊ‹…(µ æOÅáÈ6qL«'!w‹¯ú2>z¬oy”«¿·wGŸ´9*Ù~ëϰ€aÝØIìòln0Ô»Á*`2l`4•¢ OØ‹*ö¾ªý8÷ j‹ŠâÏ«2ò°'ú!§YdìJÎí’ð Ñ]?Ke¼àÑúAÝ‹yp‡Ê} Eò}YŽî&š4ÔàÏ›ü´¿-ë÷ÑûñuŸµªäƒõ45ù/ËT¸X_m·D¸íoŒ6s$"’u9P¸ÊÓ"‚­½—éØf²ù&'yâÃBùÛ79BYÉXDÛêWçeœ4%¸“§é€Þ²¬t­töpêœmÍÄ’•É…sùñ¤õµ‚²ªK,šîë«ÔoàÆ8â¬Â Wb+µDEÊj¿-ÊÑH0?•O‚ ÄJûôÛ>¤ ]të¿6Y‘Š-Ç]AN¶¹£‚I5HKã×?R9K¢ Î"Ü3dÓ™ñ›ÎÑ\àÇÔñtÇì£q¢ÛšéGžlP"Û¯^®XÇÁ›=¤}¶UÿÉ”±ã% õÌ,˜ÜndêäÕTh‚–LúÈ•[ÎG<?é_ïÆ5îM|G\/x­Öøˆhב³QwÏõXYaV<盓n— û(ëü¸'täZQ €V€µi/ªˆã®*¹­gÚ¹xB‹ñá+A±µK`ÊûFüÊûzþ.ÄãºúŒÜæeà›½ý1_Y§á—ÔèâB«6]/G“€pÀG‚ï«2“þaŠñ𥠿t,;ïr$Þ F®_ä¥û½Ù¨Nû%šÙõ0ipTu·,qHî”*ûÖr[FI+÷^„ÖwºWðXŒ”ÜäáŒî¨ Ëò9P™bÿÉ-òÝ&²¡Ì7 ü‘]4g¿|\ÙcÌ2(žQ=‰æ¹-L6¨N‰6Óü*@]—/¦Àô³PÆfJëýimݯ»|;©p²w½Z¶|ŒÜé-‰=¹üºYC0ÌbW&üuÍ7&ö›%n‚D¦i;WCÅ’¯ëAµüJ@ÂØ¶\1*£èX¿xJdE3lõÑãëh)|¦›Õ˜·¬æÂa”xäQ+b‚µOƒ[K†x_ÂùFá¥yÀw މ[? °ËæUåÌ0&Ó«Œ42æQúl,M×t©.‹;iX(†PùÍv‰¢euŒWs‘ÙY³£û$Ay÷x2]„ ð•½øQt…M´9úªÚ²%ŸEí˜ëMïÙæ%âLžÂô'Ýn¹ßèLT\Î8y4O×ÄÎqÃjuYyûÐsîÜüÑ}|ìË×ên怜-.DÖÊ®x¯¯ôAó¯5Ä1ßF䦔òÅ„ó†ÄýlòNÛŽí@P·×®ßK 0䤈~t’t[Ú ¼£%^¶>ËËì×è¶ÓÀÓijãMë$‘‹ôK¹ã‹8s.}—êi[`I0¼Ltä—]íH®úæÄÅ‘'H¥½Ûõìõ¥ÜªçÄAuü0€!íp6ˆ©ôødÚW6')øsõ]VÐB˜GºŠÇS¾gâî ÊaþµuÛ kAå1èJsž;=>8ÃNq~-ÜpI&–ªÕnÐQ¢f;ñõ`¶8"0c³ƒËˆk¢à½þ­Y³ƒ|!^J=0²A³ ¢Z#U‚8®Œxt2­Ï sgm½•úpZ§ŽêhØž¸9eÂ;Ü$zA¦ BTNÉ4rˆdzЖŸiN^ ý.ì uŠOnV*©ÃùC¼"ÅÑ#¬º¶#þ/ûDíšÔLÊ™¯qOz¡^*8¢eHpa9»ƒ†M Þ^¡º£°3ìųwÝÏcFíÜw”-’è×öú_Hß!/LDÌR?„ ¨[’ˈ9ËM}žï¿úLÖÇQôí)Á5ïVrÞõRv†ôëâò2ù°'ä“°£Éøò×aí¸c!½… NÔø …q—o³ÌÊ¸Ó ºXPÏé{lž%šH«½ ÇV¾?^+“ÈájpÍ÷,€è…úÈ’NûRÖÀ¢ïý52o¤$¾F½ÞH6©U"Û‚búø,Þ ³r¥öJ(ÉHþÕÅF¿z¶|–"±bù:¸ôÓ–Ò’^Å~!íhU{7ŒŽâ9í;Êà |iðDÖ˜šýÉ<±õÛ‘=E)¨ñ$ÕyÚuôŒúXj¡×§ÝCÉÎ'ÏQ-Üàùtóa¦HûnØ›3›¹ý˜¨¶îuÏŒí{ÉNzÜG"©cÎ+S4¦z##™^×§Úô®Zò£–ü×<×X¥_pIR¾JHÁ-#O>Ó9¬ç{Õ²çkyÑûí –%Ãk:¾]IG¡NjÒ¤hBB~qÖþÀ¬kÊÂ8SUJ—uoà }4¾wãiër3ôëmŽ|ÙüY “:Z Gÿu¤©–ËáUA£Zö Çê@)¢±åXJ‚7·¶$>d–|‡Ìë _S ß]VÌ,z'Œ‹î¼RLBgÃ#ÐNœë¦ „RCº¢%ÊÒÄ‚¼é –Žš1–²âÞ±mlŽpn"0¬¸b”@×Z º7ÑèØd*&t¾6%~ñX(|oÞ¨[·²oq¾s&ÁŸbo¿Ê98øÒ±8®]àû…QáaÀÏzcñpœ‰í­å+Z’`Ô¥Ý% bª¶õ“BúؼT™À%≶¬¨ï¨øw¨QåJbkDT7ËÇôx?k=qY"ö8Øl©ú+]Ú…C\jg}¨­øSÎ86~¾z>}E¼î+Aäµù¥‹ž#ØfÆg]Íò3u%+iaVAÞ–FX“‰Ô­®?N®X—;ZB|ƒø*çª.XÊ!Ý%Úù‰4Ü2„–Ov®@\=åsnÛVѧÃÓëP†ý]·`cßl‘5VœCGëÚÏ,r¬Ã‚Þ]<Ÿl8F w•éFnŸje öNнSÞc×Y ™¾ã[Êg¦ðB¼-,«,¨Áç+.º!ˆœ\áãß—áË?uu½eº;ç|2%3ŠŽ9°»©g|f'µVÕ ÍoGñ)™ ʘYöz˜zûþÃO>Éë}Öt[æEVêÛÓïô û3ü²¨EǦ*Äd·Xí4b ç*ãé`¨?šq`ðsµ~)Ç`åò$ÙÐJˆB,W ¬¾Ç¢ÑeÂÞz3SìæAÙë5Æû›r׺°<ãÍê@-ô{zò¶ÑîÞw^~Ab~<\Y \Îî ¿²"èc>nv…¼7dA(ß"‹ËK9Î.XβM<ÚÑÔÃf–eãDÁI‰!kìN”TokCsÑNõùQJTí‚ò»íê?B´,ÃüÉXnýÜLÂßÛÂ&Å¿3³“¾_ìÐÌò—ò²¥Áû_?Ôa,‡“`º|o佌í¹}ºħÉUE¦·OÏs7i‘IF]´qm‚Ix©%ùUѸãµê'ZÏ^õÙÅ%Ïb' O} ¨è¡_SmØL”;&…Û3UUô[ • pV§î}ðžZ÷A$hÒÀ±f«¤E¯F¬œ¶¼'a4ÙÕ-™¼Å#zfx*·!@YNÅÖ"<Ü  µþHj¶HÜw}®åõjjœhÆ H‰ì³à²•EüüÈ“áAè2“ðØR\IOî Åõ›=±@Âaœ]̉8ÐÍ;_?¥s2ˆ_£\”€ŸkèO<ºËj·ªƒÌÅm¤‰¾Âom¨u- ª©–À'®v¯ª8%þñ ŠrŠ•NE~sÁöF?ŒO¦™F„VÎ4^hº{êt<­àQUêÑPpô¥ØR.–õÞÀý)eø!\ûÛ£U^YªD¬c¾ª0¢ë¯û$Ö·íd6™•8ÂÈ7‡cm…CP¯߃c½¤p\wxjU]ñ×[q…ùjB²0/š3ÅW˜˜Sµ?Ñ™&5ˆìhѯº•gz¢]B a-¨µ*Æ=Åk¬³Òè,]Ìòæ)Ü h¶-8ƱÔöl-nÁÚgA¶¶Õîá|ØoŒõ[?ÇÒUiXVé=³¿½IƒÐÆ|¤CàJYf´GMO›þ~Ä»%TÖ+b]Ή½L<"1Þ±ö†¥¬ÇMfݼò‚’,ÌUÑý9 8äNÖ§Qf€)9î·‡†ûóˆ×ö¦#Û©~­"?ý(¯›nžNT±°5eêêÝck\Èwà=Áå‰çX䢘³©éö—9"úÊ:R{;|iî„_çœq’¡9U6ĬÉò¹'5L¡«–5y‡ Ü>9íý씨õ65 þóÂgÔÊõÚ±½¬®ÕOÐþE’UØH?+^ÁGŸcM¶Å\Lkå˜×Ù˜—äòÖa4ÄŽÛž©ÍGV톻œ?Q÷XU½¡¼Â›Ê,=(iÍü¨àhY¬ëz|)Ø©$Ëcnv\^šþị¨Qòf¥™’U’’"{4š[Ý<" CäpE7=ÈñËKpØK›ºÿ݈I~ÐÉ\T{dOE¤5¡r4aÎñx]zH8ïÛ!Ñ–+¡’Òû‚»ìf¼”JÞÇ6 ¢tþ{Næ0“`¯h „•±–iª†"Ùb’™g7Œ€Œ¼ ŸVØ!å[ëG/1úôÒrÁ˜g7 4Ž­š KÍžßœ¥¹@ŽºŽ̳xâqƦ|‰uˆ‡Gw¥èØÅ›üSÆØ¸™Œ¸âdÞËqt¿·å弩¶ŒZ¶%Ü<í£×FSÖž¬ÎÖž¿î«Ù! „”+¿'i¡}! éißw“Qؽž›”Ž:ʵ!6í’P7'Ï$¨9ŒÌ³9p3ôZç$A›•k_M›B¶5QZ¤ÕK‰Þº&ÿ6ÁO„¤ƒ¦‚ŒJ‚Åú”¡ôÃØ™RCj\“‚_,W„úã¼9IáÎ÷Нð]í¦q, »X[±ƒ–24º§ˆ¤HÍöa›ÿî²¶nÕkû»`BæÌÚ’T}º‡œ =±,·…#Ÿø5X•GGŠr*IܶÀRöÝú,tAô{¥¾VCÞ¥Äwy+ݼOR¤‚oo¢6<ñÌï˜0-<޾$‡æïëÝ1L-Ï„v ªe×Ëh¼‚½(zË}fÕQ RˆlII¨% Å C* ©êe46+5ωòò^…xr‚úŸSi#÷ ø˜’Žé‘F §Æç}ÚÍ‘¶u+æú7 †Ýø¹žú;¹ 6ç±fúåúòþá’ °_‹$Á]án‰ÍQp£1‹šªƒ¥7 ^¬XŽðkK•å¬j#ä}~F-]nrªõ@YŒy‹K~w¥ÊI¢Gƒ¯£ò9®8Ì¢_ûâ¹°ò]#ÈV±¿%3·©bzùÁŽ,%¼)Çw‡_Þ?¿îr«¤•‚÷ç® &§)ºòxòâ¯ñÌQ]_ä ö'ÑÕ¨zƒèh´¡§t0Ös,×Ùç¼½ ¤N_Nt#0”A¶TAŽEviï(*R2ɉ?AV%ŠhØ6X)—„’Ç®¤ú3šCÓ3»â¥7“Š®áÜôH/g–,,¸h®:)ÇÛãYÅ¿‹ÄKï–¯“Ðóù|¶´ý¥¤[{°øÎbâµ³˜S?6Á™õ£ýüÁ@ii)^ŲÖm-aõîÈÄL1@`Ÿg"Ñ[xÏãѶÆ%š£×Âæ¥6&û¥ ¢ ˜Q‹ä<#J²¼¼+tëºß¶ØûŽÍ MÛHB8¡¼ƒù-GÓvõ)lòìQ«¶ÆaÃç%i(‰µ°¬Ê5UÊ+©¦ùÒ,¡ÄC ‡ XˆCp¸þT#¸Èé§™[ƽÑGI*]Ÿ9V6ö$Õ[ˆ1¶AÏbˆÌúŸùÊ:b˪ í¦øy-òßÇtà ¬Äb`á©)Ò(ÐÖ–V#Âwüðí6º¼ŽÒ¾C5‹Úôg†“”ÍGªB!‹¢í‘ÛÂî¬Ý;`¾åÖñ»é†FpMõs “Þ±GðÛ…¬Ššf‰©Ñ%ZaÒ/Ò™‹'ŒÒÉbªj±=ä7óàv$Á.êhm©ÃKm%k7ÝÓäYÞ{ÝA—e[æÔ/Pâwa>½ZßÓN~iXVÕÔäšQRé1Þ%÷¾Ã º½ÖQï{xÖq®Üµ€¶ýßGÜSX’-ÃÄõÙ6#Öl=ñæ\{Úm­äø\sYÌœu£º;;oLŽZ©Wå†Ò”¯TÚ¼+.mù†ÄÌ-ïÏ9¡öaŠðÕ¾Ï]Ãâúì^W“©=CÈÌGLúŒ•™œÛùõ§hˆîé~ESo(¿±Æ`ß4/^°'F’‰—¿1'NZÃhÈÈ¢w mf‘чÔÀ¤ÑÀsÕþñ‡ƒÃÖ`šDŽìà½M×*4U÷oˆL ÚÈkåbÙ†\QÞÀÓIɳõa‹!gÜ ˆÈz‰_!Ö‰{Ùw·Å› ã\ÅÙú¯=2•ó7Hdp%ý õÚyW»«‹ý,eø ì­Þ5ÀVPŠ£Ó~SG_jv[°@ª·ïØ„;®?© Alµ øs¢üJø•Á•0©;Ö%ε,6²ué Ö¡89¦ìk&¡nmvP3/B)q²Ùì ®9wI‚9B¬K>õƒOßâÍB~’³»†KûáýôØhVE5ó­­~pØO¢~‡¯fæzإᠯT5I§¾×©w3K>W9y܇p—Ö6À3J)Ô•­?0,±?=Õ+‘Âø¾-ΗRJCwŒ6 ÷Ó2ÄPÝB¥ þÒµ=AêÚEØRX©vçÅèºxâTˆ¦¦Z»;¯“ð ÛÕ—0Ëp5v4kL1 ”dó@K÷oü6Àyú$‘FΨë˜ô!ˆlûE'Û^ißH0–uÁx$†yÕI Õº› dú5Í¿¶Ï‚^ô팉º-éÑ+ð¤K§HÎm%MAˆ¼¸ÈÔÐõ/N3¨˜îªefZ‚õªíºÅ¼<á¦Í‹íõÈ7^GNWáL¾ê>»ÆïûiG öÁr¹­pîµ<e0c«óÐy8å=4s¢zØÓª«¦™iE{\|î‹æ¬Ÿ+|ZÂu¿|61ÿÂÌTíRËüSÏZ$”û‹2þÇhCѬɄc·îÍ7É”… ŽûŸ6ðSÇ2H1Ä×mƒ¾òzŒµnݬۓJž­bU!”o*Ösÿ<=µÁÒÎó»EâIôCnRm@2Ø>s0BÂÂgQò³V]H÷Jž1,¥Íc|úYÕ'¥–ÛÇz²g0írUjTaŸ€d€\ºì„­ó‰<*µ ÅÁú)².‹ñ3e XAâHÙ>£8tQЩ~›ò‡¨ÀŠc¸ø(¨‹iºV —L'£ÚEÈ–¤L˜çÆÐAñЈ,hƒê&]!ßZæžÕ OTàŽæ°Ëñ|4·"·>}´¸ <|¢uÛqÉVày-™"ï"eÌ’ë®=e&‡ '™ò®E =9Š‹å¼*Ÿ˜½ä°¼“g/v^¹Qb®Ïî%HDð©t¾`®XÙØƒú ŒpìuÂ}™¸ÈÙäâ×z1¹b$;XÅt$jO’¢êÄîd’€uºo?~²+%.r&¯þ•¿>ÅëFj—IâÔ÷$$ofçWõ~œ¯îm””(í6I²sñÒvo˜÷ÉØoÝ; c¬%Èv±K¼g6°þ0z™fð«w=sgŠH:×`À»©·×z.ÅØ‹P yÐÐ1¸\ /ä£ìºxwÜê\•fœÃ #²åÈå¶þíõ.ðåïëiÝbˆ¹ãôÄãèW8\ÒŠ¸èd© CÜÍRQù Ìö©'ØéÁyÕaÂí)Xºù°çd÷£zÕ,~§àI¶7RO«•‡u¤UÁ1ÉÀ!QF¼&ñÌ™ý`x “ÌIìLß!ºQæYwâ4§˶_F›C4µB@ªiüIÖÞÆ’À-šÆvÌÊàH…Dw,= ZØJeˈí‡êY„Öü‘”?ŽܦR5¯¾DµÚÕÁ ©¨Æ"ÂÏ»§ób…Ë(¶joŒÇ¦lhÌ]´ï|BÆ •­#%Ak»–úÙ£0à+ôƱjDÈ;K°jŠEpIU.&j‡éHs1|$ln2GHúŒwåîYë«>Þbú8¯ÃV?q¾‰K 0"nëæŠ‹ÞÇ¥¢RsîežIÀ’(…æÅVêðèUz@!J:î÷'Œ@/ô çG€„u=éèâ02RõÝZéåaþ<ÀèÉ\S#ºÌäãW²ŽÐ¹j—oÈ5†gåæ·gd.è]ÇøI² på>Ö]Dû€Ò°é-ɧ,xÏDÐÇed&.§ìÑ/Bí±Ø<-ìº"ó¶Ûì'ž¥ž-„] EúMóP¦½C¶?È/qnt.ÞXX-¼ëFuo%–‘·HÒþž„°!_nqØ–¤³…&2MN½ÓD¦–ÞuÛöÉ{›mÞµ›K î<úÝ£+ZPÃâ…[ÑŽÛ¡ë¶µÕ¼úSäb³Ûw"^üs^C™º.’x™m’&iÂBÄîµØá¬Êk¥/u:¸ JZÉ£E+N(º©!{ûÒ§pÄÌH ·ˆC¸«þ÷2žÁ@¡.ØÞìB¬?l°‘ˆJBÊðÁå‡AçÇw• H‘™ý$œÿª×<à endstream endobj 124 0 obj << /Type /FontDescriptor /FontName /EUVBCO+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/C/D/M/T/U/a/c/d/e/ffi/g/h/i/l/m/n/o/p/r/s/t/v/x) /FontFile 123 0 R >> endobj 125 0 obj << /Length1 1458 /Length2 6439 /Length3 0 /Length 7418 /Filter /FlateDecode >> stream xÚvTnû6%1FFI³¤»¥KjŒ¶#H(%*H ’ÒŠ(Ò]"%!!-%©‚Ôê/þïû}ç|ßÙ9Ûó\w<÷õÜ×ýœñq™Šª8£àšhF"–ª˜ÈÁ` 10XÀÇg†ÀxÂÿ > ¸/’û_v58ƒÃÔ¡œ›ÔõóB$€i9ˆŒ ƒeÿvDûÈÕ¡þg P‚ûøÔÐ^A>W7 —@˜ "++#ò;¨‚„û `PЊqƒ#q' ž@S4 ÇýG y7 ÆK ƒ"}ÅÐ>®Š‚"ÀÆ h÷…ûøÃ¿è ¡Høobb> ™Â÷lŠvÁ@}à@à‰€ÁQ¾¸?”3܈;hª£¼åGýqÖÿã üëj€1È?éþŠþ•ú …ÁÐH/(*rº <áÀ[šúb˜@ŒŠrþåõôEãâ¡þP„'Ô çð»p(PSÅÅñû‹/Ìá…ñóExþbú•wÉ(g54 Ga|¿êSGøÀa¸[ýn« €ÂþY» PÎ.¿(8ûyÌQo?¸Žú_8ð/æ Ç¥À`°Œ,÷Âan _ÉÍ‚¼à¿_0®þP¬Ú è‚£E¸Àq?¬/ÔÄøøÁC±ÿÛðŸ;tFÀ0@'¸+ø7;†»üÙã:ïƒÞã„‚}þYÙá´åŒFyýëþ»¹ U3=cáß„ÿ1©ª¢XQq ¨¸ˆ‹ep‹ÐÿÌòÿ¿¹ÿF ˆ¿jÿ›Qå‚Êþ¡€»»¿iøÿ¥ ¿ÆEøŸ'¢q:†þ•½-X Ã}Aþ¿Åÿ;äÿ¦ù_Yþ²ÿïz4ý<=[~™ÿ+‰ð úËŽS±7hÜ\ þÛÕþgˆ àÎ?ä[u0PÜd¨ \=ÿ¹D„¯&"îl„ÀÀÜþèïàÒ{"Pp#´/â×3…€ÁÿeÃÍÌ÷”øâ:õÛÇÒ©‚¡Íœ¸”4êã àÛI±Üp:ÃëC¡1¸ Ž^(ÐíøÕOI0ä…{PžpÌ/Óòú§wÿÀ8ØÓÏ÷W€þ×á ÷ùüG0?\:ÌoáHü½ÿýPÀáp`z »y×ýõÝæã*ÖÑÕ!…1¾UË AQì´O‹ßOJ’TÁ—Ù‘ >G*©ýÔŸV4•g8ϱ[ oHb“›NCÎ’LFW›S#×Þ?ÛR©îa'c5S^ 9÷±ˆð lÀoÓåËóö»AiT@wЭXÝS>û>zrÕxí¥´ùYùÑæ ¶%ã|ùNO'˜¸‰1¢ì¤B´ûTã‡Gc´¹Ã—œºI€ÐíEX›Eñ‡'ÁsÏÍÄ}Û™y™m˜Ø iß^Ǫn¤é2~Ä–ÍŽÛúÇI|jÇ³Ó ›\&“‘B$Ýx¥ÔGȵæºôc»¼~%¢[9“ölí\¤âN¥ÔÚúø0·ïõ.v’½œ¶ö5wÛÙ3µã7™ ËÒqáº<\, ‡ü)œû>÷gü9g8µ˜Ó*¹¢AW9¯jR¼ ëq!9X…ò÷r¸¹e¹½p):EŠžÂ32çd]ûŠš{¬ÌD"ûÃv‘ö€àuFF–B¬dßj„c=ÜÊÆÜol´|òÞ& Û÷¸.’¼æ žùþ¢?ÆÁŒR:*š°ŒMÌIN›¸ÏI×Î2ÄNXLöÀ£+m->ÕqçÍ‹º”.ÐÁ#SÚ«äiBƒJ©5Äw`!$ŸSV-{\Ñ­I»Iç,ï*_;u n¸Á}[«Œ©?ïö7J jæ'Ðù׿;znDa±»Zzm*±Ý á‘Tïè¾7±ÿÙ±=«¦C£Â\Mºû¥ÆrŸÜ]ÕÐå~Çl{}”Q¿YÂqcñ¨ïÞ&õ‹«ý~êÎôÑSIÒU™mL]‚ü$)6rJ;‹—F Oä/r S¬–ó­–­©ê•·lK5úņ?)l% ðÙ—¶€Ä-Úëë[Yž~ùÐÓûY`a¹×ÈìÒÈLr…–Íä¡8˜Éq>®ÎLyÂ?úlL÷Pÿñ;¶È«ÁzSìÈWœƒE¯Š²÷ϦȄí7ë'ýUª›h ­@Îr¶í«:@‚þï–orŸ‘ôýœP œQë,oÛ’%ŠJ9Øã ȳßL¦8¡Ÿê.šòO <•-Œ„w~mk*$U™Ä­l§+=T¯ö†{w¤uÒ<}PÁFZ-šŸWJ¨±³ ä=ü”ñàZbºy©2#Òz\Œ6ûy^Þ£8˯Q׫£{ÊÖp±{£Æíë…Š¬Ê„a%~*hœÍ¸WnHŽÛ¾à½<Ô  ˆhTºÀ?¸ölc—'íM^4‘l9ËÍÀ"•+‚OŒ‚(ÉebC?Ù× ýX5w³§zÅÆç§<¯ CJwË[Øëyr[< úQùŒ[£è•õqþZÏçîºîVÕX÷Êõæø.Ó ÇD[{*–Ë'§ÂEv9¼Ð+~#5÷”-¸0@Öš—Ùé´¼öãÐHjÆ%UîÇ-6í5×.gT] å4A{—˜Õ÷ÙvÉ›䙿¡#èÓ|å²ï·)µæì û¤d_?© Ex•@HNóŽÃ¸ßµgèœ{ÈhìrFá©Ø³T‡ØY—!6‡¡ý¤béëíV¢,{>+q¯†R?WPqJ^'ÕêQò)² ‘Î*¥ÌìÓ„t3%}êÊPöWM„­ÏæäIÜëðd¥/r½y.Æú1ë8º FT8HfÝC±Þ–öªxeüOçÎ阬‹ðúÜPàIËn“G¦KJx~´'x3ÝçÛé6=œUµ©±I¨7+ï¦[êgòz3·’±5¾/6¶‘‘©m~4žöZn†}œSÄʧ[‹l鉾W‰š~J3$O³½"rúiÁO+ô…´–Ÿ°¼¸GzUéóêªM%¯›S•É)Ò” ‰ÅfiÅn‹Mš" %™=Öèñè²<%À'Ôå„m¹³kK¶xñÌÕ>€Ë ‚7 ¤™‡^ç;ƒ˜ 92–ô•â [ruHYgzÔÞQ¸­žY!–¶‹Õ¡É`ÒkÙÅ ³‹xàÒósÝâKåDrż»ŒÔÚßþý[˳3¶Ã¸VÂ*ñ“p$·Îæt+}ööWAš]íÍ$ã*V3[M)Ý—%?Ì]så=j¼Kðô™»ªU+Ö&ß ¹?k¤q£œÔIà§à½Òª<#‡‹y§ó•{yUÃA¬Wm]6Wýtaé-”z¹«ûOÜRøÚ! R¤(vÖ…¸}vÓêÚnÚÞØ“ò Û]½-Ç)¬¥.wðZS˜Û02Ærg+Æ&Éœ®‚Q6Ù>ÎEÎ +T•[tÓi[æ,J| 9%ÈóBf‚¹›ŠZÍeÖk?¶2÷ݶ°ÊÙK+Æ´¼Ô|×ê3QÔ`–Wø†Ée¦¦ed‚¡/ê8˜óðž–Á¢qظÂC£=õaO¼Ïú4f^½Öo½tWŽ(˜c`óÆOåvðÜ“7WJä&~‡*ròâ¨7l¥äd­ÌGîñ&Y ®sY§? Ûºž¬ïl3z mù=‹FlK²äa”Øn“=ß PÑ{õµõé=¥ê»2wZBöñ_P]U—7Z ¸ÛÄ-’ë¶Ò§æÍy¬Èþ•^%WÉÈár3‚Îü®Ÿš¨Ï+?¬cÇ;œ'ûĶ“wÉDoJ¬Hß!ú4aŠ'$é!ÅÍ;CË·M „«3÷S5ÊÊèÚ!Ô±Û~³ØxÅ䜦ð@€Ý\núgA+*öúâ“Ù"ß`ɯVqܯyœ–©õ±90¼hÍpÃxxKî¤üYoòŒãÇ“0 Gl'â¬ÍfPa»b;QÀ»m¬©!å=tûL ‰õÃ/4/‘ÔÊ ¡œ36k<‰n1W¼°M SÕ©ç\ƒéß}ço4“ ­ÌE_MKEdÔOoªË4ô_S?k:©üγúòp¶­qj`{•¡Êßß×°èž®<žb—I{ç@Þdüƒ48ʤ;îê0B­ú8[!$/ÅK(îÄ`Ñ•ÕjÉy°WCo²ï·ÿÄ~»ÁîŒ8Oô‹¶˜ýç€@úÕ:SkMRU…©·r¦FöÌßÉ)¡_xSTS[ÚR]¼cøºÊ™ëOPÌÈh´€÷#Èà†¿RÓÙ;¸d"âÿRæ>•†^dÏZ£½Ë¡bWÖI:Iæ~Ïd6ýq!C“›)­‰Jõ¶¢x÷=w·Õ@ ×z‰HïUlŽ®pÑQ´Ü†¸v¹:ë>8}½Rë—FV{ºláŒ6kp¨Ðà.‰ŽÒêÛ¨mE8}¡\¤z”È$XÂãw/ܘ[$Ù’/ F÷„²_r„ÆÝ—¿có¦Bùüà8ÕN¦6}µqïA»Yz^Z¬ 4oP °:ÊÜw礰Œ>}x­¾ö ;zXó1²ö®­c§ëtaeÈ’0‡çK»?&ñ£Þ㪹šTŒ¹³j­/£ñåÂ?j`ò¼þ†R8•òˆü&Í%R©l¯ëî“åÎúó±6ëÝtÆv™ {spLïWÇÖi+‘éþ#4Apñ2ŠYC»èÒ‹O§Ä”j¦û^N“cm}—¶}ÃØØ“&qÈB¡Ùu]“=5¿ëdç õQGYi;ñ 75G8´&K¾òŸ·²Ë½î)+tfؾf3Œ Ñi>´,- ô­3AHcÎ}ņ¼N^iá¼êäÍÁ<ýIeÚš[û'Õü½Ÿð­½ì)È/ì¾ö.²Ó¬˜ÿFS¿Wê3 oj /B¶[XåœH–tœRÍÉD"[|‡ºÎëÑS¡›÷}°ž½(8ÅAãE_ à²ÝiçqkÅo8L7õ¡'œc@_%#ó³™5h%òW{íPp2‹F\µ”ÀÞ½L[*«Uèé˜ 8Iðíé4YMc ‚´TÍϹÈJú—8—r&ηuh'L§oïßÞÙouœ,JLø·'܌¦m³2…qgn!ÆLÃæ·™´=\—2¾Î*€–Ô¯¢°x¼×}í…žlTƒî»ÍÚÎ]ÛãeD^·q3Œ ¹IR]ËóFEKZU©–³§D{ ·ç¢¿^Lxv çP¶=œN9­Ò‚ß¡]ܱkv‡êº[³Úñ[dcj¥pÆ‹Ž’ãlFZ›- "™zœ6ñ¼`Æ=`pÆ!Fs:=(µFþ³T€œÙöX÷3/…aèã»DA=ªR¾äl"iC¯´`Skã¬èæ.Ì5æès)•Q<¶GÁLÂUÂKP оdHáâÑž’šÊ¾S Y ?þ^z¹(|LÖX=ÿŠ'ÔHÓQý¡],IhòׂͤÏÝðŠ,¾C°Y?($PÁÝtûz…_\¿áSŸ'~»¢ƒdñ ¥5Ü>jª(_× Ö£»ûÙƒ®cÇè×eåÁ3òN¯‡* ä߃‚§c–Y…i_ãþã\?ŸiJ&'nìF+4H?G¿*}ÎøíäÙe%»Ëø Cr¢õVØJ»ŒO²JØ`V›°ÃIè¸?A*zâþ½{¦¦ZxnK’@a…K±NÃßSÁ]¥O*¡Ñtè ŸÝ‹Wl¯QÓº5™º“iãt—=JÑÑ4ž<{ï&¸qÇph¨töÚJ¹ž"9¸Sì°rUÈñV ©œT3yœÁvΓ•ФãUÃ&óÞ·¸x®h¡R«JËÑi4„­¡YCrˆâ—yõ t¨Þ@ÙìÆØ ÙgK$ß°P£Ä§+ŒzŒ ä(1“ø‘Ò™‹¬W)òF¢¨t0µùP“+Áê ±u-ç‚ Êûáœ,*ºë |’dò³#a¥žc“tâ;)–5îåMηºÚ8?ãůšÎ"áe¬Ü]- äà HürÙäxØw—~§{1J «©µkvcbåö(»¬Ã7ÌFânE –[‘‚V_ýgÛ’­ýg„ߦ…Û“o¢ô²dëõ#:è*…è+E@(Â…_v|KÏÚcªtïy?«ýí»%Æ6áÇîÛ‘oZíºQ.žÌÛ«­[*™ß\\¡RêjÔ¼eÒ Ëž®‰­ç©µB’{ |!º ¶^ƒEÖŠâ ä²3/Iur)/-8¾6é)…~:«Œ½*Àº k»™ik•³¢Ò×–uG2Ëñ½ÝË´dÿ蕨qH›à#òœó{¹Î-ÒDb2šjk E<È”®”sÖ<[«ëUÜO–(ïÈÊ´†pÎ%<ÿH3Ö…ö…ÃÞáeÛ>·$Ñp5‡mÅ.:^«Ã0Õoõü.XŠ¢\T®1Çk÷>d&ÕP¹Å¡&-‰yÕà«k›™Œ#Ù Ÿ¬w®‹R‘jz¸¯ o}B¢<¸³~”*q?‰'„=_qºÿèP£òžì÷XžÍ4¤õlÙzXˆ¡º×‹t363i¾¯§#Úd0곌{ù·`‰Üì[ P u5ÜÈV$kÂ~;ëmD×h¶)9=Y¾“„§7öÒ¾úù’„ÓŸ*‹2o¹sí6ï -ßäûFï,aÅÖ—ûrâ©£†×¿yÓ ~’Ì8LäÍ”ZXûWw¾Õ®%n`4F2kù +=ª(“îÿ¬¸ì‹Ý¿^WY¦|Õ*˜†âóékžÏ´ÚÚÂê| þ6a–·(YJôH’kÞ‹È–Á'ÉÜæ¦ànªñA>‡gǦoe `§Æ/$ô9Yd­»f?í­NoÞ2!r4Ö»ñŠÓöxw&5Üìv˜÷ ._yyƒ>ÿÞÕ ¨©·ÙØÛf£…“mõèû½ýyÈû»¶÷xñ¨«ÔZ5ö¾{ã?ãlXÎß×™}‰Wu£ÞÅŸ¢Btzh‘€o&§èVĘª™Åc"Fí²YkFáv/ÝPSÀBÿk¤W«.³Dti¿õ½sh/Å\n'Q©±µ~£ª‘ÃíHmK"UÚncW—5 ã~+[+MéÃíÛïg ˆ‚¼VIþyòzŬ`æÀw´S½«~#^öÝ­Ê™<~#ò’+NÔ"WB¡å^/:ï&a›\ò4Âc•] ó])óÅØ¹{j¢h™JŸû/¯ÒL±øú4JJ1'˜Šv™Ó÷S3:>!X¼t|¯ÉC¹°Ú–ˆÂ{$o>_j^ä ¿#Ä9‚~¥×5:[,àáXþóU2Œ°Jîx% ÜsÅsû0eiPBZ4†Î³b\‰1b:ÊîÅ~Ï›ï×ãc!ßuÄL 2mo%Á_í1"ÉÑ)Õ,Ÿw6x¾}˜5ÉK¹ó³ÌKìöç]óW<îa©€Ÿ˜{æÍס‚½má»±M«ãϯd©f"Ί.h„k¥2‡¾Ì£mÞ;ÅùÛηt*0Ú‹˜Åç¸Ü :çöø>:«•‘÷˜?)Þò®ÇEþ«8È«É ¾üi‘Âë\òü¤ÜtW˜š‰™Å¯IåLÏq®T½ào!Ê©+Q…ù=¤<.½Íó¥Ÿèâ®âx<ãÌmiµåÐMP³îò¥¦ð‹"öÒ JYn/¾ëU“Œ2- ,æãIŸìHÕ/N÷°›_žU+àáS6­3¥ÒLô ¾½1Í,¹‚¦¨ß26JHþVÿÊþäY )Qˆ®lÅŒîÁ üÇ›î]Ý–‡<´žoÎÄG÷m}ÃA<+CvúáK}ý-ý¡­‰§‚nT‚¥Žé CCIÕsÈË’zae½¿ƒòù³­wZ.ܯßUlÏ× ´5Í\’RsùbSÆDï ÍÉ?ŽîŠ{ÜzÛ¯3õšO&@½©ŠVÅlr;–›u륦„óÿ@MDà endstream endobj 126 0 obj << /Type /FontDescriptor /FontName /RBTKQM+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/parenleft/parenright/plus/two/zero) /FontFile 125 0 R >> endobj 127 0 obj << /Length1 2517 /Length2 15763 /Length3 0 /Length 17238 /Filter /FlateDecode >> stream xÚŒ·uT”m÷J7Hw %ÝÝÒ ÝH×Ò0tww#ÝÝ)%%RÒÝ!"qÆ/^ý~çüqÖ¬Å̵ûÚûÞ÷ó@M®¬Æ$jfo”²·1±1³òÄÔäÕÕÙX¬¬̬¬ìHÔÔêV à µ&ÐÉÙÊÞŽÿ/q' 1,“0Mìíï]ll6n~6~VV;++ß íøÆ®VffÀ{{; 3µ¸½ƒ‡“•…%œé¿?´¦t6>>ƹDmNV¦Ævc%МÑÔØ fojyüOZAKÈŸ…ÅÍÍÙØÖ™ÙÞÉB˜Žàf²¨N®@3ÀoÚEc[à?䘑¨ê–VÎÿV©Ù›ƒÜŒ€°ÀÆÊhç vr±3:Àùj²ò% Ý¿åÿmÀøO{lÌlÿ„û÷ï@Vvÿr665µ·u0¶ó°²³˜[ÙJRòÌ w#ÀØÎì·¡±³=ØßØÕØÊÆØlð¯âR¢*c0Çÿ0t6u²r93;[ÙüfÉò; ¸Ñ’vfâö¶¶@;3Òïú$¬œ€¦àÎ{°ü3bk;{7;¯?ØÜÊÎÌü73 ;+G ¬Ä¬À"¤?2 ÀÅÊÊÊ ,Ðt7µdùDÝÃø/%Ûo1˜‡—ƒ½ÀLèce!y9» ' ×ߊÿEHll3+SÀhae‡ô':X 4ÿ7Ÿ'+w€.+ø²Xþù¥>gföv6Ìÿ5h-9MUi†Hÿ£³wx1±³˜Ø98Ül<n>n€Ïÿú§ÿ¥ÿ/©²±ÕÊû+ ¬¹=€ïß,Àíû/×ÿÚÿlà3(Úƒ5@ûg ôX¹XMÁØþï¿\þ¿Vàw”ÿ[ðk’r±±ù—íMþ_ƶV6ÿ±lxIìÁ«b÷Mµ€ÿÞm ™•‹íÿÕÊ‚ŒÁË"jg>ðL|ÌœÜÿ[9KY¹Í”­@¦–ÿ:<ÿ8ƒ•PÙÞÙê÷`bûïPþÑ7ÐÔ|É8ƒ‡ö/¼`ÿ›UÒÎÔÞì÷&²sqŒœŒ=XÁG‹ àÅ^Y3 û¿N9€…ÙÎv€úÌí~–› À"ú[ôoÄ`‘üñ°X¤þ ‹ìÄ `‘ÿƒx,Šÿ ^v‹êöSûƒÀ>üƒøx,Æ'9›ZY™Z9™ºØþ#gcçþdecüGÎÉþ[ —•³õŸ àRLþrg7q26ÚÍA‰¹þ#þ÷9ýÇ™íßbk èìù8þ‘ÿ0Ó8·©½ x–ÿ”Èù[bkû‡äï!³˜ý9ÀÔÌìmlŒþ²Wñ‡%x¬,ÀÿIÊý[ïè^š?.àV˜ÿq—knåúWŒßj{—¿s€M,þDë-~?߀›€k·üÃÜ1KK Ý_`™Õ_\éÇ¿ x¸ÖApsþTÌ î‚ÍïƒýGnå_3_p,RqcÙâOßÀ©í\lM~_I•¾|Yìÿ Žiÿ—˜¨Ã58‡ø)i÷?ÃædûôG Þgðy±ÿ3^ξ§V÷£¡ãX†7o÷«³œ£ ÎÊ@ú´ëôB¼VˆŽÓztÂ× ›¦U!1ˆìÿœ1[¡ykº¨‹­Gh¶öq„®”Æk49ŪMT¾Ì:q“±äæ¸qÂL‰ëM“p¶Y,U }ž?…[…€ŠÖÌŒ ¿?ƒ…êq¬îM^Wå®—öµ‰¬+ˆ‰¾˜¼™†x%Ü»i½ mD™‰pz Âm7Íu#ö^œÄ iÌTk¦BÜ#ÁnzÀDD‰ØŸkC} ó|g_6¿ayVÌ͇/Joi{,[ɶq¨µ{uVó&Kíg¬þ±½P ºé[ýÙ¹afj|5f³9ç!yªôîîLN4Ö—srÒÊ1ºðÛ’P&̽MO ,®óN NBDD"§~Ó»†k¿½´¸R/h.|ÄšÖÓ'éÁ~õG}qv´\E˜Í IVfˆqF$ÏNrè X œˆ>XB¤,1\+Lï•ç J#K´h=8›|gæZ—& ìø0{ÄM§Á´}?ÝÍ’ÚªÓƒ¡H_¨túÞFëVöË V¸þ[æÓ‘“»RI{$2Ê@CqˆÜ:C<®Ôñé*q"ä.ßÇ€wÊÑ‚zQ&™t0ÜÍ%¡ß’³R4{ s<  Hå8RtÐû€·¯.²Âpm“Ñm’fë¡Ä6ô¦F3uú£±ãPõÈGB䫵i>s2C•2¤aI'„QØ);ŒÁe> !–UOŒ1Î2®ˆò(ù´ܸ‚L—¼Œ×é÷Í:y.79¨ˆZ¯l%}è»a¾”îØ…[Á0Ó&eÃk’÷Ž@oÐCŒ(Á`’È·«¤‚äžM[¸QWê¹¾%ˆö†u'¥/¸*d]Æ$/ŠSY¿Ž½6g{>?¥ÃÏÐcª˜H2Ê¢À0‡rx;d1rW=†6Žpž¼D/B{­iÜÖ8Éõ4†-óS|ÝŸ“&Î0Î…“tѾzÈÁ&Û¢ªfd}š¦öÀ t‹åiVŽÂlÏ Ê$ÿì9Ó„2ŠÂÿJX=fôºwÉî!'ó& û—ø' «á]¹eÀðtºìÖ‹Œ†何ÝYÝ•ÝÈó:¥žeX1½@:ʆæM|‘’qÅD$<8CE"ÛgåÃê‰YMT°yÑJ– <°ms±ºÈ$u^ ÅaàÉn‰/ì„=ðŸê‹_z z®’ä-´µ[as<½Æ»¿N K«Õ`>ONœEkZ¼^ #~®EÕƒÒùډ삎|Ë1\¦•£1ªsûËNý„2GÑ—ÃG·„¸íÒa}zàAÔïp±5©˜-"wð¸UCõ±RøC[9Æ>®åÔ%<Š䯢WkBX‘H?îÆåÅ$Ù©ïkj#T³)Ôûd©Ê—ÉUDéÏ:WA‹ï< mÐ̺ÛÍÇ Ò.ËÙ‡5»6}~(y,ƨñû9í™–o–ŠÈ¼^c‡1aÍÁ…ÇBrüÂäÿ¸“fü óácÂÙ^a@̈4OÓ%@jÑ|¡&•%ÝlÂo´~‹ñ®×"”C<€ã«øà—Nütœ2Í™ùºû–µ]@ø aŒb‡Ó%æÄÚukaâõÛVûÏ.VËS‡µ­Ø<Ÿ6º*ç¹#?ühÜKäàÃt…’V&oZêOà ö³s¯µ­'OÊ/\B}0ýÍ|SɤÇn¶u9»]ê–¾Iljç’°ÄôD «jøÒƒã¾sÝp û9áúWÎ6Žm~au_ÖÏ]`–)qC1–YëÄп°ó"˜Guú'lŠJŽæ:3*¨þ’‡7×dó2)^~(ò…•¾@éZ=v°“ý¢9P,´HÇñm­QUËÈNk`Ùåùx¤Ë-ÅÉÙò’ íf…d™9Ê÷ ZË”À£âØ—Í=_ÕØÔ¶wí*§—gö<ßm¯;W bœíñh7z Vßz~'‚ØGßÉóÒ£×¾äs¼SfÄq;¦týN¢—µè1c[ûJ¿@¡ŸÒ«É"Ëü,LÄeìI/'ÑÈK%Œ@Àƒ ÑDàôÚØ˜oî8¯t€‰#¯ëÁ­„íÜxº¹4ê¥çx17™è$ΤSïšÃ{ˆv™Ä@¤på*ó`+r-jã’Q±‘XIZßãäH« Ò 1|ØÊá‡ÙÍ|G £g’UüP†”˜a Såo³G¹fx5ÔßþZKO<ÿ=(IbñÛÆ¬’ô¸P†%Ìâ$Lø¦_[cœ`¿Ñ;ÎVjdkχè#‚†ýžÔŸßZ9ºÎB¾±‰EfÁ4âa*ïö‚üƒ§û}•œ¹r¦ÉIúA¦ÔQ?Gƒ›UKëC" –N,¼ëôR=J]kÖ¶èàÕÃ6p¬–O(½ÂÔ`zÈù–`᪲„ݼÄqZ—Î3aËõ'Ì\~oƒHãZýù&^ËuzJ·Ô¡ñ²»@ÓD^¶óe'h…æû.}P1SæêeË\ç{5‰@^úˆòM& a%YX Èu.™xûj‡Çñ´ÍXÓÞÍÕr¦Øžÿff‘¼­â$iã˜Ð®îÛÌ…äÞÂ$ÎQƒh× ‘Ô[ DÜe[jÒúªs"Â*B­Œqוâ‡Ùà-`°-ÚP“=G¾THèRîõk=ÑÖ­Ë”@õ+Ñ/N 9ÿ ¾J(ºªÌª¾£åCÏpq@ÝŽ5„#h:ÛÎKd±¼–@¥5bt1ržF$‰,‚I2‰á£¹ì!WÅø»/$_&‹€ÇM|Ç[b÷xŒŒynü@wãàjïTlÚ®;)(ŽÉÒ§=ØU[¶+çZ#^©zn<•ž{SÒö²Rõ“ôS‰ÌÈœKH Úëûo4lÔSÄÉÝæâìqñeöêŽz=×ysØR#R¡ÒHÃä«_úaôÈãxœ\ŒËŒ´Ÿï@±ßœyíeζcæÒX'Ž9'/ö=å+§~ö”±‚zLè?`XÎ.õ·®kýhYªmÄqƒ»?à½âu3‘7 Ucç7VÇ–™°¼‘w2ò̺XÂo@h!ëfïp³[ï7-οõ’¡,Ô!^Ào‰TMMÔÛke»!A9~Mß•Šò5,táð|¹ÏZÄÄÓ†r‹W+B®.AW+ˆÒíc—ª"H€ÖtC®Œ­…#çÐî×KÔf èîHÙ_®ãóâ•Êóy\[ã¡Fa ä2‰pý°-oä¨1ž~»¬ RÙŸÊ&v»Pôߨa:èÁ€ïñ4ûºóÒ=¿=¥äâF Ã3Ã%i§0¬$¡pÃÙe„î2ª•­¿YÖÜ%f%µþM/e‘ö ¢¢ †lßýl¾'ò±$åá§”€¼¾´_5›²>ÙãÕÂ8êG[ê+kÄ6@"Í£™±<1v-±Bò‘Ïç¥êzI†ŽZä®ÔùÒpw ŒANšü¯jnvêTm<Ž ®TŽ•Ø¤oéa7Ýuºb C¥6ŒXçà V.ëpqÈ®®Û…— |›yÕ*'’W]SÖ³r;•÷Ï‹Ã:4zóÔ[Û´ÑB3–Ü4ÔÞ* |]¯Å¾vÏ£ñì0€¡ÞÕ×@Üžðê¢(Ç"¡ I–àÞèœo–Ç¥g6#%=ûÅxºÉòP¤D««á Æ\n î‹É÷–ìÜ6µYŸ×ÜR¤ä¡h™ÌTîØ³ãæv ÌXùñ­%ÞÁTØìì¡«'Þ gkµ¤}Ãð4_õœz~JÚ~ydÂ5§Œuf²ú)|í§ì§ÿ‘ùT¯U­W$úà2,SŠ^h¯sh–öÔ5WNž/±á¸"/CŒÅV—µãUç¡ß)hxÅ˼Å?QáOþŒeŽP„ÍÅ…ˆ4$ê9RüñéðüêµÚ.obÔ9G£J"ç…I½T\¥ÚÂ%£(’ó×éGMD”_åâ÷UËWÓÐЊ?W$öù‡5†î¥V‰—kžŽm¹k’?Mh<žëÿˆn«íŒþÏy⑼Ìo;-@ëëÙ¹„Rd?Ý0yÒ͆+P¦úΪSšEÃ¥žûÔ¢/xZKV%TÆêÅԨ⭧‹Ou‡ÞKì ¡‡\Š ãOÿŒ©#³ÛÛÛîT)5:f;߯á8:`>@!û®.§™ÝÚõ&^µ}\Pe€\‰ÈH&«Ø6ý ›†¦Õžý þ.piUÓw~LZûµTÐõsR£Nôt£†ùú&E•¡¼Væ)fƧÏuPxªRðkCØló‰œPü±ÎÆ:ìù¸ï–žRêz /çÈ‹©ßòîÊoåçÐÁ6ö—ã’Y>±FN `/l!ÿêÅ‘ža¦ØÎ×øI•â5«m¶8º‡€r³1:¹§"\+/PÉëFçÇ0ú•sL€©œp±:gႆÁzûšûH-6mt²xŒx ¨ÝË&„ïŸ ŒÊO_£üå/ZæÄÞaï-i¾ “ C´?ð½W¤D–Z“'X˜æ«Ÿ•:Ø G‘¡xÎö#Ëc÷RyuÕØA•©ÙŸÆÓÅTÎ@ÐIæX¶ç˜AÔ/¢2Ÿ¬¶6ÐEš%¬ùÂíbÀZZ.ÃkQ¼3ùÿ Œ;>weRÈ$ã¼zjJ¯ß½;ú‡Èg¬ ÛŒ± «ðr^áМÂ!©´4‰êS¶¶{©Þ vy@ 7PÎÒ€åÀ·/+à¾Nt XTx»]£¼t65îŸZÞ5pþ²~ë™^8v_¹žûÉ0ë¾mumr€d‡_ðS6O#dŠx5©+÷‰—Q ý4¾Ì¥ƒ…¤PÌAiÖ÷·Â5E¯Ü÷r4‰â}yܶ–™ Ã@8zD§YR欫ý›®¤Ï3Ö™ž»Ÿº·ËìÖT§tÃü±& Â0DËgÚ¡9{‹óŸ™!Ç0Òô뛑öS½ýH³`! U^n.>?–7)Îd… ½jO‡‰KV Q=‹yÃXù>¯t|Áõ¾îì%´Ý+ÛAh9†z;û]ökBѣߘÔ ŠCò³5-*^eq¼èë÷ÑE‚;Š/6‚iñ¹hdåÂÅë8Ï£_ZÆ'äx‡¾üœ~‡Íƒ>c KmÚÁl"×ûîÒì©kD/ÐøH›á -ªíœçã0*Ùj—Ñv ЧºÞY±ïù`ù‹tx[Ú†”`Øe=q­Pðöí_9M‘5Cì¾Fk0ô\VK7õûëw>ü¤SôNšï—ÂIÌ’÷Óµ¸ASlâ$-ìr˜¹ÿ~ýwà/ÙׇƒæB¢q18ýíêЮs ¹ÄR›«‚€)‘¨NkVçAXíãcƒÝú~‡V@£È·‡uCRom ¿Ú¾ÐoŸ 2DQ¨Þ>ÁðèAŽóº½É^n¯c̵¨@y)ó ¨Lë‘þÀÙcuõÞ)XærhÜÎwo“ws2×t^ª¬ØÝè@C<È èá¯ï4õô)óódÁŠó³w,ý¤Z®s«¨|VWJ£Jß8¦úœ[á"Ù=ßøág4¸4…!öÈ—Yd6L™·›=!lt´ìšìÇ¡æ&Z6·Lw¿·WG Z• åÑ?Òܲe(â|OÚcª>kÌé7byN @õì[Oß½C{#^·\¥é¡ê:ùéB¯‹kÒߥ@ýÒùÒéÉSY¹ßZ‰0Ü5Ãtë(0 *÷b¬;9Æ=i€G6RàîÆ¯26gü½Ûl ··í¨9>Ì•ì1Ó']ý#û#8€,Ì)22毞Ä0øÝKʈÇ&,M¿øE€H+ÕO•‘Q‘£9¥$«yVÑŠ%ÀsÊrF¿úëÓ…€3ÇŒÎÏÄV‡½Æ^Í·ˆq¤º~œ32ÝSм25 ´·«špòÓ½6¼¹Aî!Ÿ/¸û¾ÆË²˜—6æÖòÓ—ç ­+aó7€’ëã¶ö\-ýu ¨V=Õʾú›¡ÄìÊÉ‘ðÚ¨J°®~–xÛJ Ìo¨i«*¾™éªŠ²MæJº–™ÝÃߦi_Mc‹N_²µÝh‚ ›GÙ°­"»£þÓWàØ‘;¥h‹äÍ,‘/=îÛ Ãèö˜ÏžqáCUâq9ÊÓ­­V߸µAÁê«å¶W{R#50@Çt|4°ó•øå%b$ÔÚô‡Æœ‘Ö`½®ÓÊ"Ö‹,ÛìÖ‹ÉÊ‘ý{Ü…mUïü[„i{€@:Öâ³SBšFήžq-#l¥$ç3äý·WN]ôä¾› ؠʌڅ´EÝÏS¯”+/﬋¤ÚjCp_¿Ù(·Úû@H?BèÅ“WËUñ¨±-|ÕÞ`f‹Mvä3Ö™”Ql4ù)ns`¯p¡²F‹úÖi¦äS7|ºážÈ?<ªg»ê¢o˜¸qEð*pbAAxÝ)â–wÞç0n”êRáÁi.øâ‹²¦‡£¿N˼V ºÄù`Ãæ þ\I\>))…7?[ÅÉ*D³Ó¤üûÖ@ø-»ôc2’·!ÒX4+oγÖʘf"Ñ9îKH QóÜö¤Å¡ }{È*ÅSJ0¡ª þMmc tyE™OÞ„JΛ`¡Œfλ³½ýÒ^ãŸÇÙ6’ rо.Ì“þî03}ÓŽ—Û²û•¬p^Š ä_~&9:î.j&š-›Áéñaqº hݪJéÂB0 »VÞ GŽJ4ÉÓ6V˜õBMlç/54´l†Æþ¢1rwATNrh¶§ÃbÔžv¢6£`s:Ô,=[ö+Ë ™:¢ºb¿4“a£Jä'—Üujýi•?Y1h¾=ª¯éWßt/ÔÀæ ô¬·?YRšõ1£bë³¼ lgÿžC8¼ù-Vîvé£ÿ©'sÿDá` )Ŷ ?öíbA©ÿYsЂ>öÒ¯mСÏaÃu×=ÜÁ5 óÅäæó±·¬¥x7œ?JC/Áu,áZ†Zhwý—Pè‚3Í„ŒWš,Þ'éÏ i( Ž¢¡­µ“”à-–jûËÉè·éB¯ŒY>$$¡ ìL˜Ä|eÑ¡V#†ÅÝ÷Õ½†ß»£º¾bk—XÛÒ¤Vú‡å/Ù–”·<Ç›öÇo~¸ƒ–B G÷ 8îg “‹ý@¿A•¿ß`·Ñ%è§7twK:>1Úè ¹”~eêCü…sè«¥ožòi'^·ã\m=yN «BtþÇÙ&…1ôR\`ùö¹*ùËc¹ŽÐÁg<Õê4¤“Çõy[xô7(ˆ–£dÛ€[ž›tIÍä‚v|þ0ö? ôÑÇðŒ`HJÁoN?$x3>9¼n>±Èï,à1> ›•2ç8ŒJk…ÓîÏÞ=q'Ñ>[«Ðï”C Ö8 à $‘ÑoÕ·YàŽåÓ³¦®/Õ¦¥2wOB$¹×0…¾_5]$X-~g"Ë¥õfsì{“üªHØ´‰Í=[‚šŽÍ%\CW–Ÿ6äÏ]o-·&Ò\ãJ×ÀgäCó/þwA±&É$rvýæ"&6Ï×dÉCh+D¨ ñ‡b§7P AÖƒ9rc-§µUÓeCœ†¯{×c‚ŽÚíøÕü¥…î}Vœ˜G3è9–M%+ä¬nFj¾xÁpý*ÎJt3ô;ˆ%Zc:ç¥U†¡L4VµØ¿èTBÂñð<|y¬b\':ÄP€·Îë'ÂðïþF$n5¸­m0uõãþÛ“/þ0E“Kâê"{6ïb¹×Ò×2/,ë’Ž|ÖµàPlH§tÁ2fØ`4<ý·Ûñ„ðhy-C?±b>©}ï‘@“áþ4Te=¸t3˜ïp3eC µº‡õâ¬)Äö)Úòá~õÚ¤/3[ö(8°ÛÖB.6-ÂX¹{ôŠ0ÿ›ý]"‘M­šÝëU™®»IšÑþB¾1 cqkb$­~ð‚U™åwÎ2úS]€ÛÊý„- ¡Â¯œÐüȨ,ÿ¥Ê¼€Ÿ»ÖÆ%»ËºÓeâ†ÌËŠ•y{3–!w©ivU ê6W¨jyìRÊÐÅlhkqÌ8(œNæø† 1¯øâ×8AÅ[oD£ü ßFáé—ŒÜÿôØæHw\˜b‰¤áž«ý‰ïNéüX¤]¹}û¬>*øÒ…M|Kº‡t7µ ^œ¢«H3,ywäÀé8Õ¶‚A9³çüÁЦñÁ¢…ƒ62;‡¡ñ-àî«÷£¤wØ{†M¶½”þ±ß©<¾¦|{´P t_ 0¬{½µAµ¨Ô«ÒRWmG–ž ¥Áôú¯¹ñ–|Ñ ’¶)é‘ÕXCñ:!e„A$¿*` òX1ZùXÓf¿©¨·ØRLLžQ°ZÝ Ñúˆ—ø(cÃE·ÿ”â<…¤_}.‰Ü1‚­v † ä¿Ê„³'£÷¡a'ø5·”¶_× 7ë é¿›ÌüSMÛ­HNÊöA5Õâö."胪°AÎþîQ>«ÔEfå[©Àåù]â’ÎK+­m!R'ThtËta;?´Á©'ä=jhÆJ사$¸CʶmJÔùÈž¸.îš@ÔJ”5@;†fq7Rüy=_ÖWÈmD‘ï¶6ˆÝ ɺÉÇæÔJ4fÉ iCýŜҙtÃWÇU¶÷T…Æû¼R:Ð#s¬aÆãSÖ´é.6Xi9ª;ÚC‰—ör¾ñJI‰ ÜYêz†u"¯{Wõwe“{,]bš½wä^ƒÄ2yëÖc^ÕéÝ»¾ÛPû ;Uµ7à VMÿçÕ]pIg†,¨Æ £ÀÁUÎ=†ÙiEpæàö!Š]oºÌëu³:Ã_Ž.wnÚæÕ±¿é,\Áò{%ûìw€³Ã„ñ]‹¶íɈ,ÙtMÃàKhn öÁLxR^ããAÐ4»)ãêò‹Q²«4:±“ÆÜí©J™–#­‰qšnŠ zDÖ{t§<€’ë«ßÔ:‘Ù[¡üœ@)]öQ»³r³&Ï`þÇeU½[ÈÌ䕨D(-&©rrd?^•5÷v~i¢@OÁg½Ä¦ŠÌ7z‘rI*€¶UÍBnЃÊD‹¸nSýgf‰Œu÷ s»¬Çî³ñÙ»b”ÛÓé.©î›¥Š ’®"Rãó> $„`ÖhÕ–¥§H–Ðñ(£±{žìÙN¿Ô¼c¥JnÉ ¾ri +ÞÜ÷à{#Ó3ÂðOéÐ %5>zÙç!s›'%6ãŸO’$ a-‘6 (c³G¯}®ÑZP ›ÜÍɧ\'íý(mi˜evµ•b€J°óÉÙ‰ ÿ@¨d¦:5?Š ç~ÙfÇLdzmª‹Ï„D2saÖî¢yÚžØÒäèHT¬f¬æj è9/–×à§{(±Œ>È^ ½oÚ¹_Eº«õ#Daåda€@‰:½/­ã"àWyVáÆÎ&h\ã2¹j]ž/|“¤éF)~óR´(X.‡_Ÿ>ÍÑ亾븮q3ðAèuü²Î&ÚØ›ý¨L’5\J<Ž­ÃªÃüµ²µæ“z¬œûSì´`—þr¹RÁ¾f{Ôu«‰¯'• Bœ¿;P' ¸,EQv> ª¤qç@Pª‡nÖº-ó2æÑÈ0S²r;èW‡7;Zù³§èD(Ôùt›ÿä†yïšÇÂg-¥t]Ȧ2h"üÉ÷ÑB1·Luo•¹>‡\jä(¶#ÂÓÀÖ5ª¿9 âní²½O¶õõ‘êã•„›Ö½Íªö«0¥–%ûR†F–_}ºßh÷÷–­ ‡aÈuS9ÛXùGmVü/Ë£_»vÉè¬Kª ÷_×á\ž•¥6ó'²œÐFÌ} ðVëÚ±¶ý’ö1@ Ñ)ÀÏ$7õ`-Ôm7¥¤tÙ€ê±DTª›4xÜBjE1’Ýòâãμn?q·xÙž¦z¸;êF€{lZ¤Oóq]_Õ&‰±Â¼Ð†øDGGÖ ôÎh”¶¶Çú^ŽF§&Š ôh«¯"¨íј¡íÅ9t™-k˜Àõ¢:Ûü)oC ±};·´ñsnþ»7M:þ¸ '¨I…DÑSNž=/ë¶q¢/i¨m+¨ª¶´†'öç[uQCóôñ1îg:·ñ†,IÁ?îWƼ9"ê]­ ~TgF¼‘¨ÌÚbrz¥ÓÈá5M!„8:ȃ7 ÄÝÏnüÌAÂ1g¯2’fà~{›ÙÝè)ðêã(¸e ˆ–]Ÿ£Y|3°0¸,ŒûÄå0‹ÈîæGø–÷l5}›#ìÝŠø¦‰&21ȕ۲p¸äËá({¹àéü' ºÎµ*9-^—dφ·guBO™ÞÂdžë;ƒ5œ«m:`öÓf TM‰sß-±Ù/K•[HТêHQ`"nýkOqy¹Ä*3ù”›/l¿!ѯXâ©’{RÆåó?'‡vçª×9:>o¬/ë_P 98ßÊsVsÕuáy=E›Æ½såï+fu«ár>zÿaò;41¤îŽJH9êÇ·Û² ·ÇgÕqjF|5­j+mØIåá>±‹É~žH-?Óž8ɆO—Âò׫‚yüéÜ·TnÏ}Àå¤×ëxÃÁaúÄþQÈ*Pý*Ö’ôט~«ðâá²7Þü˜Ë/rø°Z˜Z‰^ïÉÆÁñ&ôG™^:Þį¥Ëë¯ SªøÀémŒLô¨Nœa(MŽFœOr"ÚÊKgJ!õ]ûÞ2Ð0bkc2>So׆Ia°³;–ò.TK°øýˆÐƒôÓÞtºú£ë‡”Äz‡ôŽ·½†:iqêÁE½û!³"i6ÊžW+e_Ü‚4Ð yÛnN¤œŸXCú…5/¹¡ë§øÛ» ¦>µÆ¸eý'51.v±ý*mêènI‚H\d“—{Êú€8RŒR„dÔA?ûíä•:¼Ê•qIJŸ¢ÅÝþNÆå‡ ‡ƒÕ¬"Ú ¼nj™F©°#âð=8ã7&œ¦­¶¿h?eÚQ/Žií•xR‰Ã†íñ³ðÔ(/¤ YË,POö ÷¡wé@m¥­MèÑrPM@I“ð²ŸÒZ^¶TY>^ôÔËlyöÙ‡Sý2¹5>¨hÁ,š¹m«oEöYt )±ß½m@Ÿ<)ŠÞÒŒ+<94ô’¶»dËÈ^Ò¨aF>–‹à@¤´Å_u)rhæ· p&QøÀa² ¾¼p5¡³Ù›‰@9;:ÂUšÃ}1Ûê²Ø¿M:ºšQ?A n†ÆkH­K{PÉ”º¼ÿB©ÆB ¢ÙÒG:¦ù*..ËÝIa(@h2¾Øõ6HW":'= !Zï¥ø“a¨&q¯Ÿ¢Opç׋µ¨Ù!f”€_”uW鋾bdãÈH?¶b-¿ª+¦"{7« Šô;Igåï}–4XÆJ<LÇr¼{I^l{YÕ1â…A/øsÊŸ¶Ý=0Ad× ‘Ðrg¯ül¡z[—h㟾SçRšãpÈ~2¯bÕ'Ãóp®öÓb¸‚é9<ÛÄ‘9U£˜"PÏäP¯5Y ½îþJĭȸ®*É# ©²cgÙáðßäÐlï¥i–;¥2?ªÚ¢}FFS ½KÎŒ­ÑBìÝV»,Š|ƒM&à4Po*Ï“©9£ñ¢ Ë™ê[qe•ZèI‹1ÉvcÍ ‰V¯9jr€¯Dæáè/ßåuÍh&X-ÚõèÑ8}EG%Ëc­¨¯”ëàÈJF¿{ŒáÚtx:m›f‡b\†LÒ•áV|3Ta9ú•Ë]|\îË=˜ÁÔùN¿B3\’a˜ÎÝßèªÎ4žÒs3´rñ›@wÛ§ÎÓR%9³#¢òÒ_ կдOk Ækܦ2ö&:ÈTjoNAx>mÞOW…Î *Ûv+Sž.¸]ðK^Z-årÛå5d:BwàÁgÊÆÈjóÓôt’÷öPG‰ç}aêWÈjƒlH­ðl&CßÓ½ˆEä¥Ãvg•Ép,)0e ›áË奂wg%DšŠZXlócîß™„}RÉjƒ);™?Hl*ãOQÚF~ŸE%ô ñ•N%7¢_«)Çž’Z6DÓ×ÓyDY¬é0*£‡œƒ(ßITHÕÌîã¾Ù` þa$-eO/ SÌ—\Š¿=•ªµññÞ84 “ˆ3ò»PdjQÑ—ÌK¹õSÑ%8ö ŒÂ8á6B­­$ãÛj÷܇POï Ï«ª±æéðE(Òög”%ו·ÜŸ„ÅÄ)Éú9p8?íŒå¢êõ]N;¸*NxýÁÙ*V§úrÄY ËÁ†Wè6¯†{ý² ËoôL¬‘tšHë—¤ø†ÛjTG%È:+À=NTå£U¾©¶ mê^0ݪFÎïNhQM†JuŒ½ì/èsgénL íñ9M‹å0Ì—î3놟 lyÎGüy€®R‡Ù·Gùuó¥ö "êí6i~ƒôÖDÁ”+Í"ôÛ„¾Up†îì#S׬ûdLR8)=!1¯º$òý›kÎW…ɉËEk~‡°ÊFÜœº$ýö¼TŸí¥#ÿžˆ H¢9É¢q fñ«§F¨Qäñ’éþ³ÏôÖ#9ý!¨á+;r~C¶üO,vò0NþµÙ¼!‘ÈVrm5iݽ~rR[Æ®¬¤û,Œ%ѲOŸÎ+fxÕ5³Á6ibi*N’“‘7i"~pÍDÚó¤{õzoR03T|ó.;3Ì^s…ü<—`ðÛä—x<»zêz»<…`–L`ósp80–W, yZ:й·Jq y¼EîNêGÝ΃6µšmÀ¬6‹ä·ž·!×r‡ÐUt ¦â¯=ãÙáY‹JîþD+_;ãiœ© ZÓÄVÏ5¯¶néu¢áß0h±tÏûhDsÎWÈ ©tíŽй”ùHÖñqОíajPf;l£½/3²ø•B…hœ‹”ÊÕ.ûn/­ËiCJ §2í²DäU„P–‹ìHŒS渠•˜:x΄¢Ëa—Uó¦1Ç•òŠ“j?|ËC–ŒV€‡(Wßýúí&î8’ŒáPàiöU°ýšàW¿ 9 ÁþèNØõUDCÊ•{‚ƒ5ßÊ -_ô¸+´¤à<ƒåFkì­½xBàü’hŽÐXxxÏ©hÞ…÷šÂ’^Êö§DÊw ¤÷ôŒSrœœ Œi°PØ–t•¨"*¬,'h©ÞK 3lö;vôµéÛ¶ 44¨:z²Ë«ÞfºÐQ9BG¯Ê,"5XÉIA(gÏ•"3)+WÇÈ¥®še¢Ý5-}½ënÉœþ 5QM:…‘t—J™j‚¥·š+›êL¨µºÄY(Ö.yÜSF0k,<ŸÁOr¬ÝaÆ•0ì,µy¿Øe`’) $Âr÷î6ç |?¤=uóÓè©"$¯m˜Óõ8«‡”VÓ&Í0Y¤g¿9iÌ”Þ6ÕîåöWŠÊ åb ®./(¯ïð}™zÃÜO­ÏœT´pÑ*˜Ù¼Ò«ûýkÀ5Éwy1Õ¦ ‹äŸ-W'žÁ[ǬÔ/‘ZÑNI[§i°ýZí "\8¤¥'{ŸuúæX§ùdô+}Æ 5VãšV|.ðÎ<°QçÄè-JG‰Dµ+±|ÁH: òéZëL3Ã*¿,‘ËD›d„ÉÞ—þ¬ºñÓ¼a ™jðö[k˜)ž›u†‹ðtÄ«©Êä!!¨ïÛhٺǯq0tÒ?s+$}®ot_Ѿ±ªœ+Bx ÝŠÞgK¼)þbfÒþkÐÚLà$X0[øBTI;BM•G|¡Ÿjàg#}*zPàÜú“Kh¡\5ÛžZ¦Û')¼l¸†ìTÿ[Ñ utèifVFšÃU’þ´mÁÂæú‰`Åt “ç¤mÚšúZò},—aö牫·•̓¶ Aå×N^bn]/AsAPvLuq–îê4WE»ÞòŒ_÷ÐB*?æjŒ†trɾßý2>e÷†f«™WHâ:öÆ–‹#÷êÅAê£e·šfh(~¡w@„@b:¬A”Z å­[C…ˆ¸gJLžÛ¿îþ,õ5ŠDé€ù,#þ47†ñ×_oM&Uψë1b“~â4þ\)4#·Üüˆè_Ó,ÛÇê9ý¤d7Ú©­~Ý-¿òªïR)`s;°4¹zb— õ )xöÎ>'üÅYÀ×RÉp[Ó%hý~øCÛwï”%°~Õ× q,Ùÿ~a@•à“Õ¸ Þóãá$µ$=,È8ý6‹·730¼½ãé}¨½‹"ÎK4G‹vã* =„>ÑÇHíÌÚk‰~þ‡£X¾ÊçÃÌÃ5†œÐœcµ??/9Œ —§¤Çî`Ô±±V•‡Ÿf .Ù6l ù0ºTyQ¼åx·%¹Ž‘1å}¼Ýëùxnðþ’ƒê–$=özˆ/M$¢ÆÃ—Ôwˆò)'/µÛÊ»1S-HÌ&Ù–m3ÆÛl QnÛU† ”ÉÀwŠÝO•>§0O8>1j²JÚK5*¦­£ Æ,H¶ Q^¥èؤ1Êåæ6ë뮆s˜Wÿ•D^&â“Ü®n5£•˜œ7Fi(}ÁM›Ñ¯ ‡;yÅŸ™? ×3µ–›š0Ѫ…´iðêaÅÚàt×þpºÂc t€rˆ-ú¨G¹éˆ¢ÞVͬË0ëæÇ§¸«Ë|”Â%2I×LWîLJ(7}_¯o>mÔ‰ž(qe2“NÀN„­èØÆ µGÍ ”7.#hÑ2 Ò M™<Þ‰l -Žö±2rú:âós×ï¸K¨oŽQÅ9õ)Âït*”–d¦$Ÿ“b ÛBA´ÔÖ«½›Á]HW´ƒ#;l™$&¢ 8¡ÖCuúNš/.ɨ ÊÕU‡ÎÖÓzÊã)ú«šEZx¿´–½½ÑèÙòþAX2á)NFW…Ó;ëÖäy¬ÍdêÛá…‹µ¢ùÒV%e¬©_ô6“ovh)çWä¨Oœõ‰)öíé{L»®fƒÇªhÞ¡?/xÇBÀ¦Áf.”îQÓÅá¿ÐcÃxHFXù§ 2˜zsŒ? %Î-I©Ã˜§âbqŨŸK~ Å®„dáE?{ÈÅ hQ‘‚^sÿqêý, ƒÑý’¸Þò8ѹŠÝŸ¶ ÆONfò@0.ü˜$‡–±4Ðßð-·2õ!Jœ0æ­ÑàŽK!¯qÍ5Ã' –Möùçw>$­ŽÚ#Ô¡Ê…;þ¢¢£DYEí‹ý^z‹’O×)èfýEÓ¾}@ô‰5j…¦èS§‡„­‰NþÐòqÄëÕW&9ʤ½ØÎ æùŠCiá…cV‰eEÒoš¥ß ‰ãIjÒœôšò©Ü¼Çå¿Uη¿Pá9ïÜ\Ðdk•‡1;oÖnŸWŠçÉ„ÚEê…n&%›/9ºtÔ³ŒšSôe³%ëtù€è¼D¦¿z P ×/"õ¨ &®}?ݹÀì'*~—ÕbUD4–Ä)ùš{zÈun“;ÈÿŒf*Ë­±D9õÆõ°‘|BZOˆìh#v oýaï´Ís>Føí³‚¤´éäGÕ€ìj¼ ]¾ (§©Ü(dÔÍ×Áñ©æ‘…K×FÕ¢š ½ÈÖ $-*.[]™€ò9·º=MxŠ[ ³÷<°n©/„ïš&Ì'”òJaIb†ÖfÂ-”§˜x›[ôÆì†…g¼ïR‹9Ön¢1¬ª [ùdÓ è± ·B…6R¦ó\9Š÷×¿Žø>oÐYнx!ã|ò»ñ_ˆÿÀ6õ&èήïþ( •€+ÞèࢯK‹B0˜“{—`Ä+‰0CSì k…©ç,†ù>,ïšc Ðƒ$(åubádmð;ê P«ˆ£Æ°TÔ-óƒÅç„«ªúé x#C©û@9S«NÂßö‹ª‚•\&& ®0x»ÚéĘ`á#¢ÎŒAÕH 4ñ(­ðW¹m;iu|cþöxÒ­BûPéemGXvFÑÃR”XŽà2Ž³Ë—•Ï.¤p;aã<:ñb@Ä‹Ì/ç¼G+»eö7|e{–ôÅ®¢+7S7UØÓ“èhÁñJ\¼ß*ïnÀÂÜ¡Ö]a"É÷1#ÝŠÜ›`‘Ïë QAµMŒ°[@WyEŸÄßö‹„Ge°'yFrºÙ2Ãê}É*í¿(ÅÙ® o¼Vø-±¼ûB¹a¼Ù£¤€ oƇÀ"øËú~KÓ¿Ó.RBºÉÅ+/ÔÄ»Æò#<‡íñ¯»Ö©Ÿ„ôeeß‘XÆÞ{b·­HÂ];$¤ ³âˮɋ=b쉳ÕØù0zù ,ë±Ë%4ÅUðš05VÆ7ŸÎË•_ïm?-#çLœ±˜ó; ,4÷ûpþý;ÒI Ò³ a§±\™èg‚O(1hZ=ȼDY—ÆíáAæÎÇË F»Ü›s=¿•Üœöd)BƒDÜõ&?Ý=® FQÕc¢þ,ÓÆ™vKÆÚoQ»Æ4Ñ‘¤i¬€-uc\‚„ñ[”kn¸P\Ó æa f)Ä-ut9¢§©"ym¹¼Aá+;ˆÕx1}fšÏ©mJ¨FÖl~4˜c5V’ÏaÏ€#GßÜDŽó°‹P“}©âà·þ‘@´¹ÍŽÆ#éïÕrm #ô-ñà wCÛù™Têœô®+!·Û<»žkñ.O>AS •Ž> endobj 129 0 obj << /Length1 1421 /Length2 6033 /Length3 0 /Length 6999 /Filter /FlateDecode >> stream xÚxTSëÒ6ÒE:Ò[@z ½÷Þ;H I€@H„R¥‰TA:‚ô" ½+Uz“& R”^DÐ/z<÷Þsÿ­ï[Ykg¿3Ï̼Ï;ÏìdmN6#SAeÒ ªD E„€2U}Sk E‰89Í`h8ôo;§Ô C"dþ¡ê¡165ÔG":Þp€ˆ@DBFDRˆÒ‘^25 Ðè P§*ÒÓß æâŠÆÔùûÀæˆHKK ü({@½`` B»B=0Á 8À †AÑþÿHÁ#çŠF{Ê ûúú Pàe€Èú‡š'À̆úËaŠtFû‚¼ ŒC(Lˆ7õ`ªLµõ†žPÄ_`½¿€?‡ùWº?Ñ¿Á¿ƒA`0ÒÄð‡!\Î08`¨¡'„öC @È/ ŽBbâA> ä„üÞ: ¡l aþá‡{Á<Ñ(! þ‹£ð¯4˜cVG@T‘PEôkj0/(sîþšëŽ@ú"ÿ^9Ãç_4 ÞžÂæØ}o¨¶Ú ÆDôo›  JKHˆ‹ ÷P?°«ð¯fþžÐßÎßf ‡à@O¤'ÀC s†b¾ˆQ (íå üOÇ?WD"" Œ8A]`¢gǘ¡Î­1ý÷‚ùlù‰€¿>ÿº³Ã( ‚DÀýÿ ÿÝbamS= ]3þ?”ÿåTQAúE¥‚Ò@€ˆˆˆ@RRüÏ<ÿ:¿Ùÿ¶`v÷µÎH€ô_$0§÷7Ÿ?Êàù36¼€V0@bô ðü[þ÷€â@0æ"ò‚ß!ÿ?íÿÊò¿Êÿ¿w¤á ‡ÿöóüøü Üÿ£go4f6ô‘˜ Aü7Ôú×@ëC!0oÿöj£A˜QF¸`t.(rWx÷/; ¥óƒBŒ`h°ë_Zú»˜pj„DÁ~=w0Q@àù0£vÇ<[P˜–ývA1“õϺê0òkEÅ% //?F˜•8 P3«¨ßo‰„…H4&€á pFzýj,&¯0ý5f¿¿m“ áúmøG5°·—ù[˜­ü½þ=ýP¨L´8‡ËFº½Œì¸¨Ufôü4†¿²Ú›dÝ#ŽæšèªG¥9}_ÅRM7òÔh¡¼Éè*v\¨Û[#QEE;@‡-o$ež:9G§…Å¥"~¦Zž–ÄN_ìŵŒã(¹ü}¦p =yPÌ»ÑÅ@5ì–áƒÑ&ïz¤øœ|^Ïò1“ÃF¶Ù7aº–”XWÚ§*œ5¬³áŒÃ1x~/. ¿¾ké§þ ²´ìÜZû]S<Áƒ1ûÏ“t޳’EDíÒäáRƒŸ«HÁYBr0SO‡%Yufíöí$7k‹Ÿ½¥°àS—@×éÎ#»xógBLòópÚ»4em”¤&Ù-ÄÈ Ì(æÍô6±2 nZËÇ)×úîc$öþ£°•¡ÄedwÝÁƒÝM;ËûŒjáú­ø1ÖwWJO¢Z( ­gG ­ŒÔô®E×D[oU·–#µÒ‰Dhý¦éoípwwxÓµ#I.Î.- X™M²ÔŸ—ìg¶&Î`Ÿ~çz²žgÕøIs/DÛÄUä‹FžÓ—’ê‰`þh8‰ê… ¤x©‡‡0’ÉÇyÁïRIÂû8¶ïØð÷Î/›Ç¶fî!ÛÃíç9š¹â?ÛQHÁÓ7šQ­\‘‘¯…“ÖÖìy³ØÌ>õ…‚HN°W«'ëWiÂ:MÙ d·Ò®¼í±Lc‰”Å3%Tä)– }K™fë„Õ™‰uì§ÎË©ì„ñÇC‚c¦†”ògÒù1eÁHo\—ƒA㞈¡¥õ†™³$åáX‡®œ¨oÛGy¬ÖùUƒ"šðOUï6Æ#ºAåYv§[гÄØûÇNèxj†+¹DòO߇R|“íÑzé$S€=ŽÑ>›±çùÍ×IŒç­^ç¯e‹ëmÒ·º6|X¤Ý‰ý)hµRü=eÔ†æ×©nôû"9“2G&‘vU²4UÙ9àŠ}Ù¶êÒÇ$›mq|nn•ùÙù–w¨3AýÎÂV1„»ª÷S7£`kñQ,aÆ~ç—„àèÅnÆø…÷´Ä#kl®ú õH¥&^.>œ4N o/bWÅeëi¿mÓ}zҔǛñ•à~ÿüØÚ?­¬½8?û•ƒþhüµ£ R=Òÿš‘û4g¢nò¬mZ6TDQüPgEKÅañ°ïÜÝÀ²ÑúվƂܖJp}ú¶{ýR/ÁŠÉ>ϻ٠{ÑNr ¶òÕõžGd;Vü8Ù6<È#9óf¯ü.½ñ§6êêÈÿ\ª ÔÄw>¬”‡ü’ŠòYúž¨7‚Þ¾\ÄIÑØ®>c¦ÙE/<Ĺ҂fE-©6lûl²kJ—ì‘äçÙ5Š(Ó&½ŠÂ´Y˳Ñj1ûe%N³¬ .®ëÌhy;`Ù+ßa_þ»&{ wÅÝ‘(+}@ O7]oZ]Ô!}€ðûú¦Å€i½^›È ÖXµ,wɾ­Õ÷=›âlL/[N”W£«cê߬Ò/´(où$ùpáU[Ö é¼öOf1ªË!Â}èmÉýQx¥Hf¾ý§Òñ^³Ú n5æûºgT~nmB7ðÇ.øDjç^~!îÕΆ«éÛàˆ…7 "§ÇÜr“Ö+øPÁ=á Òø|ɰx>uñ;O!Ñ_Š évª ?p4n¸°MÆð=G¶aÍà¹Å½4ø¶?V]`"ÚÖ(cdj¦£MrVÛâ°…Ø™|¡Ä® ôݤF¡Ï sžÇ:ÑÈ¥Úºµô°ŸG;ÊÓ!b¥kÆÔÛÝ*,§8Ž¤Õ Ûh’Üä‰Ù)*·r}ñjÈÄÅ΋5\fºëïæªdnmdï)—UR®vÞ9q4Bƺòjyên#AKtQÒäÈ #gº·7«NæLûó%N×;ð˜žy•Ú&xíãø‘¸£±îûcÚÓî¿sMŽŸùŒ‘Å;BÁ­ã·FŸ4—ѾŸèÒË|®­ËJ—§«~íÌF{ƒ•o¤³B_xÓ¿/ÖDªGiqñQV ¸¥S©×€KÛãiÞ‡ç Vlo<Ò _†•Æá4¤–?Ù§-Tµ¶.r¥ÞºÓÎ&ãÉù_äx“€žì—uö • b£ Šûbà¢/É[Ü•µí¢yiõ/ÈBšL ?ÿ´N_D42„?ÆÇÏLœå*ã³O`Læ¾+a‘­Ñ]q5É­Œ÷==ïb™bد¥ü¹³XÁÿLZÚ$WðÀŸÛc%—i¾&íÉæÐ…]jÜñîëw‰A¶fÃá*^†)vÙØ+7»Jĉ˜•‚NA€â¢Ú~¨¼ÁÅÃNæÇ÷i¬yØ¥ø¶u ¹—ù‹] ×â.°ÞBObµÖ{HÕ|×Z/ìŠB‰;“¨w!ÏÕS¨‡‹{ù†-Ò¯fRqP[¬nº2 sÎÜ9ÜÝ9ìîîÖPã[ .kÈÛUq-Ó•ŽGN¾‹åž=õd…׉`5éÙ ê}ÍŒ¼“iPmµFyøs«Ë Ì=­Óøj~aÄÞ®;Å_Ù1mÝ•¤RVò¤ ŒŒYz ².îÒ¯)y¬µ“âáŽ/¾È:ÎRo~Vž$Ô’ixñlÕZ‹ÙI$µq­]k0°Å7bÐÄÞz*‹žT€è‡FÙÛt-µÅc EÏcÿ©§­·–95äÛÃ{T¢ñÏÏ 4uŒ\ë4ضàj;ìÄ„ª™Ñ,O;ßܼ)7“þ$þ*ì†ã]̯²ºÈœwÇõêÇ`“›ødáïBî•…R¬ˆº;‚NÞ:’b¹!ä; Ç>«;ݧ!ºK–9ú3ËbŠó‰Ëõí¾ÉXÒËÞ¦ðaMzHžgµ&‹©Âšr3¶ÔŠ*QÈhU[V1ŸÍ¬( t«E³‡sfn·ñ""E~Ê*îÇ­ëñ†Wq¡^¡úhùš(¾õ0'Uœ®ö$è~Fe¯º8wÓEIÚÕÉ{§ç ùCLzaØ»F©\’h4ö™·f€¬®ËÝw'SYd›½yÙ·ëe/ß›¦¥÷øtŽ]}U}P¯Ø­Ñ.œ¢Á.a.T TØÄÇ>å¯k< ¤gž©œ’ß û‚Ú^ÞðCêØ­ÎW°ú^I~ ~”ŠÞ §8*´X)¯ç¯¢(n®ª~uÅ,WoŽ Ïz1W>à¿ùi¸/•À,?^UüJÓËeÍ¿\|”#1úi­ð¸Ñ¢K#ÖW$_& _ö³íÆíú±G-é.$3Äx Å?O¨>ºibÛE;™4s»ó™ðwøf_‡H5Õé+Ù`V²£‡÷ôCrÁs<2 ÐÄ:ų‡QëT6™wYtc\¢qËû·ŠzÊ7ɨ”âs̪žf¹Ag윺AŽÆºÁ ߬{«E夯’Ÿp…^÷II ]Ï•¶o:1ø¨€ùCÔ»ù‰‡ÄKb¸àÒ§V­F “—Ó›Ú›2Š=è' Iø¦æ†ãàR7QUÛ╆’ aî–—îÅ…åØ#V5Ü/˜KhrȨ{äÌK_Î,Ié!!nK¨yR1òhâÌ;Œ _Ò§ÐÜViji~uɺ#ƒ |ÂLH¼T`¿Ýq¼Gk®p4ý­/äyíó ÔÅúGÖØQøÖég?X}‘žT§Û™l™ê}F¹’ŠHº‰>7#2éµãjöì×.©*q¹ñãý‹F §²*öƉ†å©bߺùTn‰ÕGx#Úq8šW´Hw~æ;âC8¾%bÍàxµâ,(Mm,ÜWüI¿š§¸ðzphgº¶ñ ùžëŽ]\‰ÓQ÷ÍëíöJ‡Q¹©‹±ô~\ò=4SOE­íDN] ½NÁrC¥"NýdíðåmtIã‚~DÔì=†Va§õ©Þc} êa øÏq~wíO¢–Ò ÎéÁåÍòÉ‹¾¦S1ttSå϶·Š¨’Þ ö҆ϧIsDùƒC¶Ô3zq®ùØ Ö”Ôõ¼ HqIwê•©w8C´Rîa Tù¢jP~ûÅZyrºëÑ xcû®¥·Zm$&éby,O¹š ZÚP&’âŒëÐq7מË—m¼÷5ø”[‚WÈÄr\НmDúö:LƪãØt—Åb'GØí?M}s¹ÿCXç¬>À#ëÉ.ÿcØ~³2Þãdo‡Û/Í ì©Ïø7_FÙ-Àkü>rKø¯KX½T%ïVÓe1$ íœÍã¢äkç\b‘9Òc»ò`7^DãÛ6‹=™²˜OÕÍÁ5ÃlŸé¦&Ñ>ÄD—ŸÌZ–/G€‹n;Ý9˾a›à?‘2Ë ©ZrŸ 1QÞsä騘$ço©Ð}³¡p™\³é¯ô¹ÿg«ùê%áïW ½Z²ƒàÀô#9'Ér8—_bÛžRyôÛÃ4BAHîR§÷ÆÎ=¬ž1ìÏì!çÜÞÃAÔl Kbß i{;žÛϾxBÇ;›)Øl˜ØZk8w)JnRÖÓ@æ÷fk—5ƸŸðÔWÅe2y­ãÚôº=/›}¾žítt ÛZ—¡í?{O}ïõÈN¬§W™»>q³¾R”ëÕ3±´ xïp,Úôüs'sØ>ï–°§ÉàÅòâ« v:ÑFo¢ uhŠøßb-ªÔ昛ðv]ké׌›N1Æ/NjMá”ùnýg€$z†z,òåеœBTº÷ nÄEn^­ŸÓ¨ï¬C-D§)äõi‘T¼·ÃtÀƒ‹d¦ÌE§Z†Dbæß¤²“u[¶=‹0¼þ:•|—Ï0à{›¸åÄ ü¤ÒšëjåýÂÂUâÞŽü®$åkºÌ†³‹7G;Êâˆaí¸4áūѸÌU;J¨Á€oó¥Y[=`J:éçX“¢¼Ë4ÃGö-K•Ëg^Ç>~è¾ WµåÜìÛ=e=[>Dcþ_˜¨V3ù·-‘¸ÙÐÊLM\yXo*•sµ{Ý)ê.Nþ¾~ÙôýâÝ}KfßNÁ©pjýsd\_˜RâÑÛ|ÛNYJ/§cu,ÉPOßÀmɾ ] ÖO4á¨ZÖ^Lc»ºI8~ÂÊÈ{ÝþòôÞëÝ|bÑ’æÞJI.¼qàÃÄܾwu ›¶€è«êÂç\š$2ÔÂwJzuné¶ywTjÊAÅ?>äj–þªH >¾…“ïÜùÁ<7‰8AÙ@tÝd (Oê%åÕÍ8Ã8ˆïu=¹UÁ@ïð#l³¸‰a,‚"Ò†¡cÞ­’F+Áv®.r·y_“”Ë~Š‘½¶­‘"}eWQzêy\ B-LGN<”µƒ¦˜.fÔr WwYtƬúY€`›ïO×ëãÉ9:[žÞ÷iùñ·šˆ¤”lšpÕ¼gÑkdVÀ2 ,°;8H^æ¸ÿZ×¾Àê…„.^^.ä­„ÒÁ~S¨¹ùNQ¤ÎºdÏØLÙÍzõ›ã&åýyÇâZ†ƒ[í/ þ[u‘B endstream endobj 130 0 obj << /Type /FontDescriptor /FontName /ISLVKT+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/element/minus) /FontFile 129 0 R >> endobj 131 0 obj << /Length1 1420 /Length2 6193 /Length3 0 /Length 7158 /Filter /FlateDecode >> stream xÚtT”kÛ.Ý(¤ƒÂ04HHw#-8  1ÃÐÝ„t‰”( ‚€t·twI *(ðº÷·ÿý³Ö9kÖzç}ûº^6m=+„%TGñ€xùÄrzÆ">>^>>~"66}Êú—™ˆÍŠt!àâÿ+@ £Ð6y0 §€T] HX$"ÎÇàçãû;ȃÝ`V ^€*u!b“C8y"a6¶(t›¿_NHLLäáït€Œ# ƒ€á 0Êêˆî;ôåù¯¶(”“8èîîÎ vtáE m¤8Üa([€.ÔŠtƒZ~h‚¡ñ±ôma.ìzk”;   0î‚Îp…[A‘ts€žŠ:@Ë ÿ¬þ'à!௻€xAÿ)÷Wö¯B0øïd0‚ptÃ=ap€5Ì ÐRTçEy Àp«_`:ì†9€-Ñ¿'et`4À¿à¹@0'” ¯ ÌáDà¯2è[V€[É!¡p” ѯùäaH(}ížÀ?›µ‡#ÜáÞ¬ap+ë_ ¬\€Oà0gW¨Šü_!hÑ?6( Ä'&,,( €: [à¯òúžNÐßNÐ/3¯· `õ…YCÑDÞ.`7(…t…úzÿoÇ¿OD À A,¡608Ñ?ÕÑf¨õŸ3zùH˜À”Í=€ï×ï?oOÑô²BÀ<ÿ ÿ½_ ¾šš¬¾÷ÄÿñÉÊ"<Þ< !˜âG?DDù¾ÿ.óŸ øüo«6ö×p|ÿTT[#b0 /ïonÑ‚ã/ÉpþÝAæ2ÀñõÍø„ø èèÿ[¿Sþo¼ÿUåÿEýÿHÑÕÁá·›ã·ÿÿpƒaž ©ìŠBËBü¿C ¡¤¬µ‚¹:þ·WFËCnƒ¦8H—Oðæ¢ó€ZiÃPÛ?Dú{è08TáûõÅAgññý—­:ˆ=ú«â‚^Øo-ª÷U€CV¿ÔÇ/$ #‘`O">4Éø…„Þ ´L­ ¿ù òÂ(t Ñ`@ýZ«vAc†¹Ø£·`ûËùÛŽî„:@)ï·í_M!®H$Úñ›è‰þ>ÿÖ?ê…ÍN! BìªCš¿WÊ0¸ólá-¯¶DÆwE¡Ø§_zÛªãg(;Ë>³*»3ª=S66åÃÀuø}ØÃìcVàRŠr. C¡k]ö˜'¯ïâËG(³‚ÞfD$5rü(A‡éö3SÓI†S?CEÚ¶Ë· ®Z\¶MÄöŽ€¢¯d,) ¨X™œ–}¬hcÖUT¿±YÒ4ÍU_åÉÈV4,î›ls Ý…çºi\zÿâboVwWªÀé+µŽ á²0J:*/U–œÄ:ê„,Õz vY¡¯r%Éñ÷è ÃØ—°«'AŒŒ¬qt|Ž€bF»Xf"Rë¶êHmSóØ ?{¯ËO/O¼—l’C·š‘úÉZ`,p¯¬Œ²ëYܟвÕr©T»¹<ì6°6fÍŽÒë9¾3ð¹C.wÉ(ÚùXù–¦ÖeÔÊöÇôx§&IØðÓ»:åBòc&szóêÕ$>:‹И‚O¶_¶ÌäèžSPÜ‘»çpb´5¡t²¿8%ë ϸo˜‰[ñCaÛtUÃâRYã4b C¸„B¹j9申lbY§0JÞêŸ{4`ŽXܨÝîI£T©·Â·`vm\ø2zªÈ@ÿJ=ËÄ{ÄÏy/!»õYÖ²E7«Ê{ióM€f5±U'o­-Ž?Eÿ"0ß½á”!BiL‚þ~ÝnÐt"]p¦©ØÎ냗óÊóC¼äº ¹|¯½ ÕkBLË4žšÊÏ)h“hú(ÀÅÒär\…Ñd‹þ?§^ðj§’^?Ø«tM;ͼ‘.æØÏ÷¤ Lh¨Yž¸²:„f»t- G"îêÉ] ب'º‡*eM˜»ãð¢VT›× Ïos*÷Õ|?w'3¤I·a·?š&¼_d×€UýªŒE42¼½þ‡RŠåGþKVdü*§¶¬Yª²…ûö¸|ÀÈMrª){Å•!¨ïd&?éY[Å«m¹~eûp¹,4}Ä3¡›í¶å«‹F÷Õ~"Ô‰Pþ憳ã}-qU;¼ÅÄìðNȹÉè°ÝÜ« r³¶[>ϱ)ÝX¤gIÈš¢>;_ÕÄoáa³ 3­þÉgÉ‚PDrð!”·ÜùÀñ"îÎdP×'Âr&C¢¼ ÀRM»µdÞºŸYFk!ᙌ¿×Ä€ÉÇá— !¾š U=9ûsƒ)7Ómó°“PÞÓ{X×7HëÒýô›Ì©£G`rVËçÅ>±‚©+çËHn0Þ‡—pŦ½?†|Öœ‡ ;ÙÎd˜ÐY{yJG©õ3SÍÒ]²F_Ô©"€QΧýƒd­/ŽÂpb¶†Ny¿€¬OZ¿½!ÍNT¥’%Ÿp Øð.PúŠÌWþ˜ÎîOÒk"iš’\´6ôf]¤^Á_êícúí¸ô®hé¬-½3{†—”ŸV£,$Ø9'¸ÃM ˆõ®œšMu‚© ýx™ÒÏò¾;”fè«AŠ*Óµ9rI{vŸ¼wÄoüÄuK„ÀY Øù&öñ1N9GU˜¨€8¦°„0¤3½ÝønÎÅÕñjZœš}yø¹"Jü>nX—Xc6{[.½ì˜xŸ; ‘~~c”aÜgúnÜ’ |!õ»µ Ëñ—æ>ë0ië±]¶,e ‰ïËÆŽå hî&ÐIO®Ýªz'ßè ÃÿqãÂá  å¿ïQod-Ÿ[Q ¹;äD5„ æ ȪEg£³—Ûx—L}%¾­‹\‘Î5ÅÊãq©/‰ø¸äæ6Iª^VöÅÆTø,õgÑ~‘¿Ù÷9÷ë6.êÃÍF`Ï›ýy‚ÐÝÎÖÁ–)#1a…»ÌÉ{íŽ윇qª-dCzî4‚,õŸKW‡Þ3¾]ɩƩ K5hÜSMÙt~÷¤Ì¤#dÑ4SõÓ5:˜.õTÑŠµÐš‰|ºÇ|±•)½Ô?®çæmð}z+«žé¯nÚ%S>OáOËÏlšŽí?.»uw9»o×òëY°qàc·}9p„wÎäM÷ø›Û#X|U¨=kv/E% WÿÜ‹Õ8Ú Ÿ]ng³HÝßÊÐ4.Sd!OP>"Øe—…pçetq¼O¤¯ŠÐ#Ñ6y­·¬3&AŒÓ¤L}Äîíùpl£:ãέÝ:±´ò¦DÆÌ‰×%ã;/ƒÒvŽ1&âôi(Ë\rqÄlÎŒÙIàûk?ÀT0²éÖøëdZl6"¨sÛ}.¦)þëEÁ·zé›÷J)=ô4N‹”¬òÏ‚>rGL7ßzÑ`EÏˇÌÔ0¸#;g«ã5Ú‚õHä|ë3†Ñ0Ýö;K¥Èž¨ âÒãüØ•õéÑ2èÔ“!··¹0’'ðPþŒ³êÊVIác£àAÚÕxÅÏlþsVww2kJYF/Ñãûú¾ŽrV ¹=Ä\êiY6\a]z¹´¨}í–VpjÓ^71öÐdq@{Ô^T ^¥ 1¥¡x| XwêÕ= ælŠ>Må_`˜ \œA*m™FÏ­ñ3<éŸ$¡Ç°’è +—χѢÌ,Â* oü‚[%¹Q ƒEq‡·r‹2üùÁ¹øGüsÍšBRògêß­h\34EBqÄýØzžû^Kè?--q²bî¨|J]ã!FК$Úûe³)&ˆÿ’!£xŒU´1ÂÆÂ~;V‘š,`-š;åp¹küSC7“g÷!_‰â‹‰: ãñW%÷±}×Ã< ®ØË æ#Ür÷2Icg©+³G²ù³]ƒdµñ¤\#Ú.Ûâßö?¦Ô·¼õ´ß~iô¬§;5ʈ©¹º€Ó;ÕÙÒÕê‘yƒo#L0N³ÚÊi<,%<ʲoDºñÔÇ ßñ+>®e©'ÈŒ_òvüNéÛv§à„Qú·Ï¦îlÉ"Â$šï:ÜZ`“«2Ìhí½uB¦_¨ýÂåLÂ/ؤΞè[.d0¸bÅuóen*áb2SŸ™ï›L¦<ܨ-öŒwýÖœ¸÷äkᤜ¢|áfø­&ˆ¦O„¦aŽúÔs´3OMÇ”{«Ô¶Ö³{º"î9]HæÇR®o4zz±¼œÈÃ$‘ê`ÜAØµ×‡Ê ™ñ=¡ˆ£ÅóãJ.'œf™\ݸz÷Èa°íø&n¸Ó{º ÒÖ•Tb&–ê§QçiUÀŸ°Fû=çö'©7ü…#yŸZ;z)"*¢•^¥+µ‘áJÈ4žïO«¼Õt\ò3¬^ze|©{I&¯üzë))M©ÙOøb?”œ…ý´w‡"b¿.’mÔ &ßtð%údàk¦÷’u+öôvÙ|t–P`„ôÀÔ½ªq*\¦[šà9·';~mÌ*޲0è=+£×«èK —¸}#Ï×;U¤4ÂRÖ†• ~=¡œPáŒ79¤öÍ òwO…g^eHÒ @Ò¨¾ö½i¾s u¼ÈÛ‚Ù˜\dÜ×L:á2:‡{| üØ>ó)qM oꞪЭͿÝþM²”ÅEø¢ËGw92ߘ «VS¨o-\ø`Ââ ¥È:»4iJ礧 ZØç#4p©xÐYÐö­l‡FØå u žŒàž˜,Ä Åò™Ô¯‰¬¸ØãQczƒª…¦"æI,œ0ºks½N hm÷N_wj’”å$Q®Ž‘vØOÊ¥-2—ªÞ0Ví“ì¨Û0®hi]iì™ñ×4gÓ®Ù0pK$úvJ¯Ûù³ìâVøÞª ì£]Ã›Ž¨i×eÔMÝI]”;¾÷¸PŒ“ßs{”Õ¼ûdå)¿Ö[ÂSºÂë;‰FtïÞiœ//ÿÀKPÐ*À, {]C¾£=›ô6(ZÓòÑ!¦OÇçï'(?O’µËh\ßïXŪÕÃÔ³m_M˯N,¸E79óÈÃGMòmÖÌÕ¬y£ñ ‡9 ù¥‘ÅYµÊXu¡–SF¡P[ø5i˜ì>[÷ŒØëùO÷LX–¨gæÜ®ü_æãTªE Òi;â[ŽcÒ°Íuyãq2ºÔ¦®97*¿(;ÇŽÎ{l÷ÂmìNŸ÷á·ƒeïùžp{ne-²#ÜÊå+ᢛ¨S­Aq%~š#Ï©rçRÜÅ¥c1zA<Üq&û\Ù/—QæõÝñ¯Ÿ—M~UŒrþü‹$Õ§;ØŸÙì 5ê‚W¼ã꜂g0î0†”~òßz u¯a”¾`jþ–9«åÓPýh#°%¥­'lÅSqþ`tü†ôX8Ípj[dd´ï`ëžD¯¯Q·VßnNÍ”ÞoÞóåÔŠ¥ü²- k:óØáɽ“™^ûŽ)Iü—*}”ª[<º2\ÚëY\[#ë>söÏwL¹á<…&å¿»ºbõúašÖ_þÈðg!²4ë:¼Fâ;õ€ a—¸º¤+Í¡<á–Êò9_KrUàÏ]¹»ÝY^s=­;o§DFNýcæ:øÜÖuéMŸ5Ö2Çüü6×ÅËÝ$îsù¼Î˜‚"¨fp‰óÁû¹ùèåäE÷ÀžíÎz!;¼í”Ì •£ßˆrØøÆkŒÏNÒðTWp}¿°r¦®2Ý‹ê0!˜Ž ÛhÆÎzÒÒv&Î?ÐO¬‚UJ¿†Íçž Õ1ʇ”ìs÷¶©x„_ά'%F¶Äph-³bÇÉ„© êog¾¹ÇÎËÌdìýÕf÷¡bLnî>’½•~íVqè{uľJ)C³¼FÖNÃÝŒ§sé¦è¯öqÅ †œ}›>uF©kÇÄ>¢1쥗KŠó’1}c×8‚c¡&[C"§áÁ¯E¢b[¢ò~»]Ä̉Aãe‘zçÇG©×jÄz&ª“Ãö *ïèo<œtΓÖÁ44_Ÿl_C^fP0ØÖ¬Sª%J0_“|xu³÷¤³L É´³ÉÅóýèŠ(ÒÍö<Ð[WZD[‡"¸ySð*ý.n4¡Æ…pPÁ‡YæI/ŒÐ;Á5“æ8Ò×ÞiÒ©Â(ÝÃŽ~VÄßém/oXãfœN“½.': ÷ÛJ8Kå³RŸ¬)tòȤÀOGøîT90Œškö"o=n“Hâ¼±ì–lÑÛ¤ü¨ðämÆ8"ï š>ƒ»¹0Ó°¢N$Ƴ_/~{ ™Ÿäí6ûÉÓQ¥4tBbßÆÂ…-äF¶ó%‡€êýl+¥fAv—IµšýOswì¯\àâªìa=ËåÅ,õ$ ‘E4a/'&¬Ñ^6ZúñPîEÇÛÞ~ýì\î2÷º]ÃÞZGŒŽ¬ÑÒW]ª ×7æ(Ð#س¹&в×7ƒ=–Îw` ¥,]oc§š÷1ФUÄ~Ì=!%Ì hŸJ^rê¢ø.)æ·v•7>]Ã!dõ}Ó!²>'áïta¾ K"í8+ªH…î©.l7rÓfïÓ‰Ä9Ä›äiŒ)œ«‘ü¨Ÿ¤åT™•£1\òD·­2¥?Àú ž»)9¡%´™ÜÚÃÕä>ü6hÀóÖû´õ}œâêá0ÎU_}xZy«ºÑz¶?f‹qvŽèÝ`>ǧ£;³–‰²ÄÙ¦Õ¸éIÝO9;E;Y­ ä„s­œŸ•9m$w>z/{—ÛøTZÁOß¿®”ãüúdÖ(‚âAmÔ3ýËÖµÕþÔ×ù÷:{eÓEôXõ¶WyÖŸ4w°íGÂX6Ú.óµƒ æ½åâprøôM°>lß?No‘Ý㡸‰× Ó,UÔÉ øD‡™Iôâ0å8– ÖÒ)¹[¼ (zÀUGµÂ&©QÆàˆÖfZ<”÷K¤^9(¸;Ñ‹÷–ú¢ ˜\IHRXGb™ ¢¤¦©Uø›DkôI­g¦¾„«“Kº?f›Aä.û…àv×[«täÈŒÏ÷ZïlŒ^ýM¡<Ä"Ëç–ýliíâmGhÄhHÎïÆR°Æ¡|O$˜‰”q/EÃxžõÆRÖ¼)Ç=d‰'%º=k¡¥TïïÚ•ú¾¡æ&‹×.?±SàÍ(ª)žünÔ@÷²tÀFTM‰Ê⯆%p qæUdü“Kð©G;ÓAç›mÌ5„û³Ž'„_} wìò‡Ò¶]ÝÞH9¨º2jk:go*-źòŸ^¿8¼qz´ïÕI¢Rc[X y Ò¦ZCÌD´øE÷gV[Ö–.-w Fv9[,¹bç”.èÌ«”`\¾W‹vͦ:úÔ»ÒøÔW è†BêòPøhq›„¸[-õXÇöܤx¦ë&©ˆ9Û´¸©3Œ.r'—×ÜâKÍÈëRü¶iÅ|*KIêFú[Õõအêàú¹ôŒx,e'Æ›ëtÞíA=îuøÚ­-…¶jxë•`¯çü[”påôÆãÙa«½àóm¬Ùœ>7g5‹µì¶ ž•²yûï ­€üG<ûØE%ÖsiÄðáVåÁÈéS¤à:Lóåséuef7‡pc¸MŠøI¹!±ë=Š%{¶súS†Ÿ}VÜ4»¦ ÖïKΦêë(R¸²;øáJ´=V/Ã1gDNs‘Œ£ì‰Ôzí€ùâ1 “'š7™ÐÁâcQQ ‡s ÿÍ„ÞK«O3D—–†ÒtuBU V±wƒVcpÖy€P|“½“h¶ÆuapáQN)<Õ¨¸6>ÀÔ Á ÉÐÓ¸Î%~äÉ•º¬ìqaê¤\McÔ-‘Aµ‚A÷7Iü§^<¶¯ð;ÖÝ-ýþšËö¡eÐ…HMѰþòÅÖã’±²ì•î­^bÖ ±žqª‚+6Ï>•~ÜëIKŒ©B=Õ…vã!®þÚ"} áþ6W­÷ÊW—@\÷é²NFBÜU+6aëMGíFÎ Ïѧ$Å5÷W˜!¶v*£—ã Û ,ÿééÊÝåȵéÝΪžPÐzHÈÜ*fÒzgCVq톾tÇQ\C_xšÞÃGf¤y‡“ È›*ùrî2Ý—št )ÛÈûÎúD®¶qÜ”t·4¥RûjÚ5¾ªK-˜ Š- ÝJJz„›Ušsö?3ŸêB endstream endobj 132 0 obj << /Type /FontDescriptor /FontName /TKKBTO+CMSY7 /Flags 4 /FontBBox [-15 -951 1251 782] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 49 /XHeight 431 /CharSet (/asteriskmath/element) /FontFile 131 0 R >> endobj 133 0 obj << /Length1 1656 /Length2 10181 /Length3 0 /Length 11246 /Filter /FlateDecode >> stream xÚ´Pœ[-Œ îh\ƒ;‚»»kt#kp < Ü Á%¸» îþ¸23wæÿ«Þ+ªšom=kŸµ-¥ªÆ[1Kˆ9P†¾å`eH(iÊq°ØÙ¹XÙÙ9Qhi5AP{à¿ì(´Ú@g,ø g ôÅ&i} T‚€ò®ö.¯ Ÿ ;;€“]à_gA€¤™È Ä ‡€.(´GOgµ ô¥Ï¿> Œ>–?Ób@g… dµ:¼t´0³h@,@@¨ç•`xg…: ²±¹»»³š9¸°Bœ­EYî ¨ @ètvZþ  P6sþM… irùË¡±‚º›9/{ìò’â ¶:^º4ä*Ž@ð_Áаþ€ƒ•ãßåþÎþ£üg²™…ÄÁÑ ì [¬@ö@€Š´"+ÔÊ0[þhfïyÉ7s3Ù›™¿üyt3€´˜Àì…áßü\,œAŽPVýÙþ(ó2f)°¥ÄÁ†º üq>I3Ðâeîžl_®âöþ²-­þ aéêȦ9¹å$ÿŽy1¡üÇf „xØÙÙù9ù@'ÐÆíšžŽÀ?˜_8øz;BV/4€¾ +àË?o37 êì ôõþ§ã¿ Àd˜­A`”ÿT1­þÂ/÷ï ò°¿ÈÀþÇß¿¿Œ^f Û{þ'üÏ+f“”’Ö——cþ›ò¿ââ€÷[.À[Nv'7€ïåÃ÷¿ëü{ÿbÿ§UÕ ô÷éþQQlüEâezÿ"âö·2þ^FÀwP†¼è`øü ÙyØ-^~8þŸ—àÏ”ÿ?íÿQåÿ*ÿÿ=‘´«½ýŸ~†¿þ?~3½çß/zv…¾ì†äeCÀÿªük¡•€– W‡ÿõÊAÍ^vD lý¢ó·ܬìÜÙA.Ò  ¥*jaó—–þu/=ìA` *ÄôÇ»ó’ÅÎþ?¾—Õ³°{y[\^®ìOðe³þ»¯ØbùÇ ròðÌœÍWžKZŒöàý'»¬˜ еó K¶¸£LëóîœCW}ùï ï#hWÀ£: Ax<®r«t+–¹r¯ó¬t*[ %F)·Fmû\œ—pæg{E™Úpuã~*—Eåuè»t‰¦{6kº.2bT5›´;˜q½Ã)Tã6GÍïžëb´ùøpÊm¨³’¨IS‘4GG#³¡h¯%`É­3¦¾««?rå,ú*Yøšvž%Çå=~H•)ª0d³;ÍxÔPÃf€m}ãÆzhn¿nu£¢÷µ_Yi¯•×åꊜ0|ŒçIƒ²1´&U³þ’·c†U^Kî j¢þ+ÉtJÓWܒ޳áéi˜ (PÉ“Âòö¤ú”¡&æ~Áv:žÊ"a™Iùþ%¦ž——G0FT#Ë™š³á6ä5v ó¹Ña[Ï AihõÇÕƒ¹îOÚÂFhZ„ü~?G¶Ö£ñ‹g7ÜaT<;BoÎ Ú›„lç6僞˜oæš ”Q;ÅŒžñ¬Æïp‘Ç÷¿>O[+Ûk žîË¢.Ý1äáÝQŒÛÒ°~ºÃ•X$™¸±tÞÂîÐËöêäJ–$1¦¢`¸_.öŒ/-®X¼s) u‘ &‚–s“ó!)ÝÜL=8Y|Ë)Ðþö¹áÀw¥¯[uqHŽþI]Êfé{ü=9Ys,˜Òñ(×ÔÜþÑÞI–>/^$Žóî#¹¹zALeÔ>!˜³P‘Gp‡ˆfÐL:ôÅ¥ƒ’ƒöæû½$ôì~ÏlKáMÀˆŠ ïTR•î¾â¿Ò{Z¿í•³—sÑ9ž»…p« ºö)|vo½`KLù¹µ_Í@!0™5‚)f5,螆HÄzDüG­Š—íê'££~„–C¸A-qŽùÂàAQDI†S´¸-a‘ÿ½q‡ˆØy¯lež‡ð×8$yhç,C4œlT‡0˜xçVþ_y³)hÕiÍ—ùçm‹²‚x‘â)ŽñÓ꜇¤ÔÞJP »$+4L=K~(;«3, Vç"À¿³¡Ô[‰9hV W«FIbÓž¶¬Y/‡E^îd~ßY‚ï¸Õ!¦ ’;=\"“»tÌ»,Àµaj*t±è Tvíb“ž]™`æUbkÙêpÎÈÐmGCíò™m´ßÑ™hZ(ÌÕ»>¾º°ð=6®ô|¨b*Íæ¬Z:Ü¿øùvîÆ¯Œ°ÇëòÃð¾sYZÒ,»A[Ú¸˜’Q¦Õ´î/Ñz#ò(PÓM꯰–ÈÝñ‡œ%ÚÁÆG†ñMÑ8â ÔaVŽR+T,öƒuÇjoï®4³Cà¾s"× Auª«@ꪯH‘…0ƒu8Tø­G|ŠvûUÌOU¬Œn› ýÖS»A™¹À`ãÊ­²z‘sp32-»ÉÚqëfKÌosþnÍœT ïHU|×CÔW´T(m½i8ÃÝsß±/Œ´°å½š ¿ÎoYÝ®W»ìÀ´Ó¥JƒÜeè8Í¿K¾Óż“¹n‡w›š ôÓ®ZuýŠr›or°†õ›lCXx?úõyê7·´©¸°óh¬ó’ÕÅaþ-ƒ&“IŒ}ÔãŒ{@Da‡Áõ»Á@ÁçÇFŽ©h,ó8tÉ|GBß!iÓ0ø™ö2ök‹†-pŽúRP‡ÜU¸c­'HCô§0¡FÃÒéìxŠ|FWË׿YÏK@xè8OOrZôuFižÚ•ÔUG[~vuŒ¸ç_5³»½·my^3QLzúíºØ½Ky¢1ØWzÉRW¹)n0òÈ=ªï×ñíw3¥B‹®,ÿ¶œ#²Sssrfü´4N§•++lµÀ˜`¬¦¯¨PÀ»xŠ˜³¶ÈPz÷èTäRc&ÜœD³êÂoéáúÓûÈõ7I‰˜öGœÄŒµt@Ü;ü æ¹Úë6ë§Ü´pÔÇçàý6ÊjëZôz«.Q§÷æèlºBŽþô=ôú’a.‹å5Ì#ã†Þg¯Ï·=^QcíÑH¿ò+h9(“ò`ú²ê+kBu¨]G‰fêÎ…«›Î•¸ï%ÂP3СõëÏêÆìù¡Qà PœcïévLkñš{ªéÕ‚ëcI«ʵUaŸ¨ø"µYøMÁ‡Ó†öב]~…2ØÑ.nä¶þ©ƒ v%ª¨Öd›l'D5ˆ3˜çWŸLŒ´`ȳê$yíS$ i³Ðh `Û-ê¯Ö]Ä>c½yöÇåÙh[m(€O©9ý,?œO±³¹Z?òÎËxíS þýîQ u—°×ÏMBÌf—¦“'ÛS@„ÿ@ŒŸ²É1t“‡QYÜ‘­1¨Ò&æ&*•’˜µÚ< ·Ù§mç"lïf¾2š Ç?àSÝßeØjBë}2QX#BGçE&‡„Ÿ~Y‰¡s§ÝªqÒ‹>ÖÎe^†™Œ§é—ôGk¹”¹|ØÍ [)ú`ö˜× XQÆR¦W Å {Krnñ8[¤G|ÇVØ -5Ì.ÍÙ’^Yœ.§=…OõûÜÌ?öüÛÚ"¶EÄÀþŽwšl3ótºÏ1Sß–³5cXÞk…¹— >]M=ª,öêo™,X#Ý÷N_#Ã/ÛÅ[{õ\šmÝÊšN›Ò&_qwcßa¢ŒF­áÈÎ˧³ÑC‘užuÇvF\Éx˜Ì{à‘ˆ”7à£ké¦*|¨3˜—PføîB…wÂØ|îê$¾ér¦L étPº‹¥ü&µë ×Tz÷^c<08º¼ÕN•ߊ¸¤ID´d÷¼,ä'`[†3@­n}§]¡Ý]dõM4MÊÔü@äìÜa•¢†2:O§´ð cD/%õ,?,Ðg`n8aʺ* ÑD¹|F ®÷rI$»0£!=aÑPKì¡=ÁòÝ{ÂÁÚåò”uí`!¥¬•/蜒Ãqr¨÷©U>|v®ÊèÀäðc·Kq¨ ûjÝ"_©a<ËN$^V® _àBfRQer¿¯wÖ‡·${B+;Šl‘`äo‚ƒF²;Ãó‰Y@jl‡S³égcÿ/‰Ú“•&ƒÐšiÜòÉô³ ë[á°w'ï×j8¹{ÛáÛ¦ ÀpÆÑükx\!'%Ÿ7®òßy6xXn?h6„e3x<º8”m÷,×cnè—-g}¦1u3g[R‡÷*?f“Û„E¥7×(ÓæøPÔŸ¾Áû­Ý=!µÜÕÔùŽqç›ÏÄŽ‹0ÙFcß0ãâgÚnd»~ßû‹Ñîýnä†~ä¶Àþ!ª+¸^éùÚeÏ4™5Ò} Ù ç¿®xÃÝrÖv.œ®V‡LFðúI›ÅkîPæ Šýp’xU±°Å¾&iÉÍ>$.öÙÙêÝÀ¿Ôé’#ÐÅšäÕòk-»4ù ~Lßoz·$Uê*Û›#{¬Ø;jùŒ°Ðl‡» ßΜÏý¿›2ïAô!Dˆ02¾\^4]£ qQU_¤4ïV”½ž‰R‡ûEô­iD–œ‘°ÅnÄC¼4`AûK)!ž—AZ*¯õ»³dD3àÇKûŽžwfOíU*>`×#/;¾³xúw’qÏ »À öx(§Õeî)èšqRß¡ ÇÀ»BÝêæ©|a=¸Õµ›?¿íÎn`ˆ4÷.„âÞ°H¯€Ï=¬dM…”1QcmDe¨X Í©BU›I|¸v?ó¶2ž48ÎZ?oÉ› Ôy-ët¸ä%ŽZJV" ëÊÞá¤Z.fÄÛ¶ù3ry,Ñ‚éí0møþ *wêx†Ër<Ž |?‘1ë÷z‹<[IÌ$ã³ ìµ5¯ìøßc

( B)UHEã€rtxOµ÷”·ø91}hè¨JÃb—©SB×s}‘¾æÏ›\èm9lœ‹±á.ŒŒÆ¸…K¯®@WEÉ+Qéøê=s—†Q¢žäÞò>²Ø¸Æ] knŠñjn{¶=ºêðÌc>rVæÁÞÙ6#hRÙáF¦Ïd_~H»ìÈ\×K.Lexoye´¼sŒ&<ñç!‡X“8~l¥XŠºÚ¨Ã·UF»45ßp|óüª×62è”KçW‡:á«{‘‹'©, ó+ŒCFoÓæ:M+|Øžeÿ« ö¯2‘B2G# AâøÇ­²ìµÞ.…þÆÕF+D!ôHuäy,¨CNv–ÊòúVâ_V&ØF|ƒN¬6Võ¹f1Ù°­m"cøîY‡;5g"C ¡Âò’LC2 ‘ÓtÚ÷§­Ïaàj>D¼ožÌö¹˜Ê´|Õ£;á~¿JðHzT†Mu²ƒwƒõl2Bµº 6Öz=nqãW@Mú·D3‡ žÕüÄKÆÍú´¦•ÍÈs_N“¦øŒ£ ²PC…¢ëeÚe_ßw ®¶ñP¥òÀE¿¾ Á,rçŽIŸ‚¥ ·®ws$9XÅŸ‚äc†.„«ñ’é'TKEáí.=‡ñ‡(_z›6‚W÷¾%Æi’¾3ëØ²·q‘Ë¿2|&Q‰Ê2x|;–ч¡j?N†*’ÃÓz¼¹!S£@ cô/"…›Þñì¤_ˆáD:Êlµo„õ¢ßÄ.7BnÛŽeæ*}/`¹I(tŽ…ýŠ£2+W3γ¯Ç¥xwq,qÝÇâáŵƒ.**“ÛŒŒSÛ{Ù,®,ÙýqX>!C6Û·+ª«yöÉï)¥Ú¤jÖ¦JË®´>¼Ñ­·€/ •Fßio¨^´z×oÓëñ•L<ï/ŠXzÔÇHéwaŽcx눫+òCÝŒÆÂ¸| à¤Rã Þ;ƒÌKاòfÁ'öÚ4–èìDW )G­$wé£ôÖZ—yÝ‘aïá¢ÖèÔp˜ªó³çEÉrl‚±‚ÈB)CKrõ dˆÊ¸%a4‰jpuÚBTšÿ¶,mðÄ‘òo¿4lKË×h’$Uc­®ø®c¹–‰^]¸d'Ì·‚£Æ@ ×ÓnºÔVôBˆ1©·i9èT2ˆ4‚¾ aÔarj¼óæÆ8Ò¼(R¹¾ô4TZªZ;òq~ø¹’üÍ"„@W„ãÃýUw¼-È'Í}uËþ«ªÐ§êusÙ×wS.|Ä:¦i–‹îNr»q¨£ÏÂÆÌºQÄžgã{ Êb˜4NW¢7|÷¸0HZ·îmÊ¡=KÁ¿-^\˜«Á—þ¾ !hÒÞâ0r`™i¿5ðüp£‡K Ø-’Aß¾D’º{EÙ-üÖµLaÏ"°¯/ŒŸ>•ZA1k,#Ñg‹Á“ð{œ‡PJ~âìQCî9 µ™Qt(áK” «ÒÙɽHw⦠ÍU,#Þ¶Ó¹‘.F¬PQKâ)Ç!,Ö€ôe-µÉ@è2Ñ{¯¦ÌcþÓ4ŸùŒ“$¤ü™-®¬RýŸa±i]+"C’i•“æًI]QƸÑ7¨>IÔ**„|pØu•ˆ½í‘¾PºÔdOˆЊ×­lÞh…ûjO÷ùñ‰ºô´n?å¤K^oµ‘᱆…¼­'-®ŠnwÓZº‹ ðG&hÆ`áO§o‡ÉéO/À‡ok‹ÝJÙ 6:\R_ux|eñvSgŒ>òºq•ƒlÜîýÂsö&®âBMƒìóÈîYŽÛøggÀ°­‘Ú'¸üýLs¯ÖA#.NKœKÊœ7¨—B)<õ#tuñ«èø—¶BÊx¤NcD]lªÑ¦kü(dœ|¶.FïÅh_T™u_cìïtùTI©(äãÔîvàá˃;j+‘&IÕ~2àfâ}A:~Äéz›¹È²QQNé5ys$Û:}±ï,—Ér­xÀðU:^䣮9©*6מýû„¶ö‰»«á_ɲÉÓªv}¨“Qu§¸m†â¢²8pëÑÈ­9$E³_ç%§wØóÅKÚ&Ï»ƒ' Û¢ü‘YH/4øÄÓ}Æ#5‘ä"}ˆgž÷ø}scï6yß_8šä˜ ®œþ ˜]Ÿ—­¢8¨`"O34;–Ç0…lAø ~‘§i;¤Ç1 ×z!*ÐÀ7 Ûa ~[ýd>Á…„/¥…$®u ´D>øñÖM´ÿ8p¬aƒhG—C²ah`-9±œçâBÜ|u\;RåœçàYcôù•úNS,ËB´Â ÏÖSÑÃð"´¸óô¸ç Ú﫦P÷… ·´^^ê‰Q–&íUØ#’;êgJ&i¦Ùºæ«ÄÝb¾é.VÜ”us[ª’s)¿0 1ñJ[½ª]D…@UŒQiÄÆÈMv^æñayzbÙ†´“$Û‰ú Câ0¬FŒv.H#D×U÷ccáLË1ÉA-Îø_¡¦y™§¬zÔdp½ñ?­ƒÞÒE‘mN œÙ³å\mŒöã ÙpèË}UIÞ±·º•‡ÿe}¦éÿ‹t]n¨ž$¨B`åŽôTxL®°ˆ™Ð{ü(:cï90ìÆ†ÛŠ\ $¶ÕOð쿃ô‰q_oêŠ,÷ܺCnK8;\ú%a+äx}üCAz¬!Ú Ö­Tc¾ûôMûµËÅÕ¼möª_oÆ%v~âB>„\žB÷°é“ ’dbå ÌÆØzÏj·ãïƒæî‚¸”é½´&>‘ç³7°ôO bx,À;…Þ˜Sä¤Íðu£t‘¤mj!‰ ©¾}ú2NÞ×þYÏÔîîyÈ|À§’Šœ‘ÃÏËÿ¨ »kFtÔöDH³3æ?@5˜é¢Ê=bî3´›¯­-Zþµ¤Å§ë FåÌ€¢Ä_E~ÒÅjÇ‹fë­Ê©CÇl{m¡=4àÓù…GNçwQÆþ®Ü¯‡O¨“³TsõÖçš.ðüxaOþ}°z‰!°cø:|Pé´nm¢äII¢4º}¹Fr8Ó­KY«ž–Ÿ Í\¬"o¾.è‘JSdÞâüÞž!š$S¡†Ê}¬·g¦€5„Âô”ÆqQHa þ8( "ºÇ^Õù±0öÀ˜…@ M¿’Mþj_ñ:Ì&rÝR$ójçëA>ÃùwL˜¿.ŸêeË‹;aÂg3¨=Ô©]%‚´¼GŽ´™Äü±Ldоõ¶s »Ûóø7–ê;nv€MÉp~UàºkîÚB©åe˜¨n¦d3ÔÙœ\ç&ÿo¯*réL‡·G\ö?])†ö‘.7xïyPÔ{ ÇOߣՕ9BW—k­þ¤0áÝj$ûÅŒƒÕKfÐ^Å/Ò¤a×zé*àŽpŸý æ©W°îë/7°?`fzÇò1¦’RM(ÇŒ*Øø\ç–Í ¥uÓçƒ6E¦éýüc"¡£¹Óºº Ü‚ï¾óv;iææ#ò!ðüÕ[R#Ã,àNùæ¥zŽ2}ðÜ:Z+¡ëg—ê¸BFÃܰµÖ$€ZA5ëÇ[­yšä¡«tïZ *lVŸÅºfj°þR®÷÷¦w"ß3ÇÕoаëÅØ ƒ.Êè6òh5áCqÊ´uʉ¢7°”*˜’£@_ðZ¸¿^™ÚU~¹ ¶ŽÝcW½?œ_`”8Ý^È×Ð?÷¸Ê-¦ÄžFa< ácN"ØÉͤ³£ËºH.Ùmœ™Swª¶¸õùpZãŽe–ÀT£ü‡zµ.8›¤¸Œ3¥¬Ýcž͸úÞd--œÂ×4æÄšJ)a“IRíÜP23Ðì`¸v%Ò/]›S†ÞOJTáÆçwÈ”Uo\•l5í‚ÎäÑÇh´øãü‹´jÚ‹Ò7ίùû’&gï?Ñ—vå9³Ä£©@Äãû›81XÒ¸`X|ŠéjúÌLç\6Ÿ¹RÖ¿»a¿ªú-RõÚø«6—yWÄØw*-Cûc2×ET!:‹È{¦ð¡ÐAæ³ YèÎàtM½ò `Ö4éþ2²·œ·!3 {J®.ÅoÆ “\šÒõèR*g,$:w"Ãúö÷œ+þtX¯‚äÈðõZLÝåyhŠ8æ¸3ÏUp5ŬNĈ.Jre¿±=÷~‡7%ÄB›Ó¯6Ÿˆ×Ëxa>†:rć}˜ˆö\½ÐÌ:×!“NSÒV¨qALj٬çæ˜¶]>ó“T„ŽM~ãš#™è›N9î¦ÕŒå¾t<±ñBÀ̧ßOkŸ.>jŽ~ºÍúÄÕtÿ«ib …×ä)ìh¼FT£¹…µ}o~%ÈxGÃN"Á0BþüÓ w{1¨$fSK~ëÕ˜XìÿýtóÅ endstream endobj 134 0 obj << /Type /FontDescriptor /FontName /DEFZJI+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/a/c/e/f/h/hyphen/i/k/l/m/n/o/r/s/t/u/y/z) /FontFile 133 0 R >> endobj 135 0 obj << /Length1 2554 /Length2 17634 /Length3 0 /Length 19109 /Filter /FlateDecode >> stream xÚŒöPÙÖŠbÁ=Hð ÁÝÝÝ=¸ÃÆÝÝ‚ îî܃»·àî×»»OŸNÎÿ^Õ½E0¦Œ5u­ŒXQ…NÈÄÎ(ngëLÇDÏÈ ‘SUeb02²Ð322Ó‘©Z8[ÿ+‡'S::YØÙrÿa!â4tÉD A†rv¶ik €‰›‰ƒ›‘ÀÌÈÈõ_C;Gn€¨¡«… @Ž mg t‚'±³÷p´03wóß”ÆT&..Ú¿ÝB6@G cC[€œ¡³9Ðt¢±¡5@ÅÎØèìñ?”¼æÎÎöÜ nnnô†6NôvŽfüT´7 gs€2Ð èè 4ü•2@ÞÐøOjôðdUs §ÿ(TìLÝ ÀÚÂhërq±5:@§T¤d ö@ÛÿËþÇ€ðOqLôLÿÒýãý‘…íßÎ†ÆÆv6ö†¶¶fS k @A\–ÞÙÝ™`hkò—¡¡µ“ÈßÐÕÐÂÚÐdðwè†q!%€!(Ãòs2v´°wv¢w²°þ+G†¿h@e³5±³±Ú:;ÁÿŸ¨…#ÐTw†škekçfëõ_djakbúW&.ö Ÿl-\€R¢ÿØ€Dð¿ef@g###' 3躛3üu€ª‡=ðo%Ó_bP>^övöSP@ S è¼—“¡+àìèôñúSñ¿ž‰ `baì 0šYØÂÿf‰¦ÿÁ þ;Z¸´AãÇ`üëçßÿtAfbgkíñÛüï3¨Š|RSS¤ù'å•ÂÂvî/:V3 €…ÀÎÅðù_–óÿoîK -þ‰í>)[S;×RÕî¿i¸þ3”ÿ, àO·M3@ù{øuÙA¿˜þ?¯Àß.ÿÿ&ÿ/–ÿ×áÿ¿‰»X[ÿ­§üÁÿÞÐÆÂÚã Ð4»8ƒ6Cδ¶ÿ×TøŸu–šX¸Øü_­”³!hC„lͬÿ-¤…“¸…;ÐDÑÂÙØüï‰ùo@ìÖ¶@E;'‹¿î#ãÿÑVÎØ t§8šõ· Ú¨ÿ=QÌÖØÎä¯Õcfc::zÀ3‚æ‹™ àÅÚQ ûߣ ` ·µs¹@ÙùLíáÿj);€Aè/Ñ;€Aø7â0ˆüFœÑ߈ À ö/â`0HüF,©ßÄ)ûXä#‹Â¿ˆ“À ø1”#§ÊoÄ Úß”ç(jÃßdièdlaaláhìb󯜉™ý…³…µ ð_9+ó_bPû-œ¬~“€‚6úÃÄidèø‡œ‘£¡1Ðhêü‡˜íñ–à_6¦ÿˆ­€ÎÿcÏÅò¯üÿ8€ªdü/bclg ˜cfýKbcó;ë¿&‰Áä:òwŽ AaþÏ ìé\@ë÷ÛTÓß. ØL-\ÿàøKmçòGA&f¿Az³¿HàŸ& @͇ *¹‡½9Ðö ÌâŠÔòj­ÕT‰ß³ƒR¶þkU~ëAuû£ã ›’á÷Ql .[Њý¡Áîwt g»ÿQƒ2²ÿ­‘Ùƒ^TÛÿi!+Ó?Òÿm è"f°:‚^Î?LÙÿ–YØýn+¨‚öÖ.dúê`pø‚ê÷G9™@ÅúmÌöºþQM6¹èmù—†“µ¡“ù (~ǺŸœÍ´”¦³›Ý —? ¨®@PÑÜþX·ûDïñÔówp &O ãŽúŸ+ÎØÅTQç¿_!Ðý÷_ü÷§è4†_š·3æ ²¬ j¿ÿ&„çF·;Á7C¶«žBEçµäØáòˆ “HU¸îx+”8܃ºº-Fy#¸LôâuÜRÚ¯Ôöäý¬«<½Û¿8…50Yp,T×O‡O§*¸çýâà­`ÙÞ)M–ãà‰¬˜÷þÞ­O½®¿le,d~Wi¯š]á¹ì]Ô§H€âY²\£Ì¹$ÐÎt°Ôèî(³7·3èÙ“oDÒ±4ð>'Q,…^ZÌÑsžkªÌN]8¤8Z oÐǦɽ„’¤±¼JŠ"×ù²©èÁÖÇú‰¼G;‰O¹µe¥Ð±­—¸$ÆH}Úê4zá2vU ì©kí‘>Иº¹‰Kä)4õéí …ÂÝ…9ҘȹÆÀÈ{xím.AÝhßÈŸÖNj{Žh›nÁÁoa®6 ¯‰¾„:ø`ÈAã_ïqvyX¦—’ÖxõZNžØÏ¸ï¹«t ”‹¹ ÃÃ$¦¢ ¾/Hkä ޽©yϪWÍÕf§ÓM¥=Œ³,þi…Þárï•õ4wL+…SæÉÓÍ ¿ÁÃ:a²HŠÍ½‰ÈÒ8öT ³95œ6ž#»`~¸L¦†ûþ£ ôŠI•àt›ø ø¸\èþ]·~Yî“ ÚG3šÚÇc¾¦þãp¨Y­ÌPC¿æ¦duüpùŒpI6M†åm^­Ú‚ƒ‚ÜÐþxp,<±ñBíÑ òœ()„Öƒ¹»[¤Èè$”æöV¦MaòÚ•dXÂÏÃ#WÕâ=fÊ©2¥uº_Æ¿òªÌ(6w½¡ÄTòÝ”miá÷Æ/é~ûè›2¦5Ó­µè1â>TYZ3 N.ª;s–NŽÞ°Û:ÇC…߬èóàó=v½³wÚðfÞ’^ÿe"*gyT’¨aÍv/Jïè~À-fxF®i†ÚݦykO³*cŒÿd‰Þq÷}ÅW{S/FÑ.ovÁòýìÃLÝ#†êïæÞ“Sô÷ÅÊBP?‡¿ÉÌBÚdXßê­’«¯úcG¬ z6å]¼Øˆ÷á®—dÖ'wâ-Ï¥³|—ñ• )äÄRåZ•{RPD§ç‰^û)9Bc¢Q“ÅãSwŸ…±gÙ¯©q#¶àRNÔžÏ6Øf¿òUXª½h37\€õ\šÃa¯ÔÅÊh .™:;Çzá¢}‚ä¦ÙÑ›”–µ›ÉÜ®í<õcx$i9?²>CÀü:5¼Ù(có•z‰‚ÔÞd Ð œ Œœþ…;•$0ç²”*ºÔûÓôù@éq¤)6ôCù{цç{ÔY“äžâ¾Šô ¶ñ_í~4xg–ÝdÇ‚oÉX[ï‡Ð¨pÀ§ÃFUÀ¦Âp?­N‰ÖÜ}maJp› üÑ´–'ˆ4ç̈́ӛ_+®§,D—™˜yÕS䀣—Aü AÍdæv·ÏÜ•ËÊP”“ÝÌÒ)X¾¥óÃsÅÐóhš ÿêKýFêů€1÷@MŠÖ¡²»#¶0qc?>[ËïÌ[sÐóÆ_ݘ£žïÇXü²¥tC3»=šk¹£ÇvLçsKW"âw.‹žJLÍNÈ:ísŸx-.‰))¼JcæjG‘ÝÊ_UöT€éœ¢o—£Óx[uÌ“]{>„wâ•þÂs%v3Q­±§ïõ >w'ŸieC#/½Q$~­ AWðšOùÒk¸™®>ðt?Éö–Q h#4;D“£¶0â¹6’—nŒEúÁ€EO>[Ñž÷Û·íXÝ;eåô2‘B(½j¶®×B§›4ƒÂǤû«ƒTG‚qµÜÅ”ê{´¡`¯Âýeˆ8tŽqcª¹!‰sXšþP(A´V™%ûä¨~åb$é.—mKš¥çühÓz„`õ”‰•3Ð6©2(2[3HüÙhbŠò´Ür^á*,ÌÊÈ‚ì:ÙtszXQþ+7~âXôæ)½ïî4ñ²¯ž¬ ËJÜÆ–ä«>ÙTV^o¥ÉqÕ@ïècH›ý0r#ÞJFkwê Qý‘V)Úi8Ì­î@¬ëg1ëÖoðvPP I±Ûc°a®݃¨ª#Un+ùh½Ó˜‡Ö1w}×›c€¾v;¿â»° (!´[G¦äƒ*×djÙ»-ÿ’ó©sº¢„áDáñ*<™ù ¾Wå™~nz‘‹q®¼NwÉ~º÷®[bϳ¬îúl„a»19K­Ïái}´×YÅ&[ÖlF£Fk´I’·ô…øoœý›pòò¶wZ}Wï`\Æ ›kæå ¾_w›=Κ0@BÐÃØúR÷u$¡o¶~GébØß6À3ÕúØ¡øË€P&{ áN9ÉmгF}pP½Ñ¯2ÐÁÐKô~,S>-3uºfËk¤ªÚ<‹Øú°ò7i½9}ójsÄîþ¤õÆ%V–=?+QµéËʵ̿a`‡ãð}œ!°`Ó÷Ù}+)F%Ø;Ô‹h¡B Kã’ÞeHØÝTÂHì™ÞMç)8ÕŽ¶ »PXÖÊJ¦Á †ËÖÌì™Üúлœ¼ÏÿÇd¹!CêüYÒÛ/®WùO8>ú÷6”ƒQ4A`„Ú6Ë£L¥N0E¦í©ÿ(…)YzS[E,M‘w¦áz[Faå@‹ð@¡œîy]_G* @êé '©ôÙB¶SèÑOð®0ÖzÅ÷.Øû,ؤå/D´öýaCÐ0èž´‹g$ðuŸŸd3ümF³+øü,cûË¡j‘™p½5Ôû¥(®O/âGÒÕ$èÚöº=ËïFì'Ì#»s&ºâJ.C>ApúOÎÅ{æîÿü."PçÿýÌM(ÁSuó˜«þF6¦‘tEºÊm4Ìe¿]/®g’&Hl•‚¤M HÏŒ¸Ý6Ξœz^àô^]‘˜Ìú¹v®ÃÓ*‰o;° m̶^íN¤1‘ÃD,À½¹‰Ö’´‹Šåü‘°wK€¬eú.Ñ,(\¥¨í™–  öŒ¬›S%?Šž BZÒ=Ãõ(±»Þxº|4ú Ñgá^o<°fขNîØµ„,urƒ³Ü¢Ï€Ä¤r¤bòîìî&éY°UAOåÍt߇îwáXg‰hº÷ÊèºSâRUªVîé–fFY’ñ&°ô”ÜðÝ[çåøô‘SÉdöe×üX~üz‡žî´Ë^zqãÆºŽé[ìvUM! -7¹C?+œ%ãUWE^&‰ :xø¶®‘“'Eç÷ºP}Y*?ëøˆ"6—u>iBÕ«ÇÊxÉ0LpÎÍÍ&,Ûæ÷S“^Ï5t(í÷°YÎo?|ŸOœû0óŠ¥Ýë9†f”ñW6é±u¼F \ÖÞ´1ÃðˆНÉ[XºÃbžƒ9ztæmC䥉Ž^#&¯R_rÄ+­*jêÞ]}W°ìLîÀ­Z_Þ£©‡QâÎM9•ÊÍ©%GP•ݧ$ýêÞ€)“¾ÔT»Mç¬?ébê~6ù(-Ë„%ð^®~Hà=IºQñ˜R72þù÷S㦩™lÇúŽj^3kߨóäX6Íj™I¾•ÝKÝcxÁ°}¶9I§À§›UÌ#†«­Öˆ1+5ü5—=³¿„—܈ù3T VëÄ©„]Á‡êN}©õœôI¯¯zPžt&ŽãÜ‘Lñr;ƈÑ9¤×§ç/ïƒfºîêU2K·)²qñ<½pZŽ gŒ­`}Àáskzzï‹Æ=ñ¶¥f*‰ƒT?7G(—‘ÎЖœ0䃜l‹¥@sv×¥>ËÁ¹}ºË8aM^\î£Íذ( Õ4Xo¡.êjÒD9\ËÅP7eú¶âuy®¶òþôø´'2%’¬™‚—-ªÛ:ºnã’2Áþä£pˆ#F¯jÔ …6×⛆¦Ï®–W†“]Ǫ|HC€UP;Õpt±‰ŒŒTj"­ÞMÞ‰Ùâ­®Eùt4%) 7,¸:Áa¢Ð›¿7ä b{…S½Z‰Åí„MãJHM"Ç;M˜âq¸°eÉ%/¥ó.e€-Ê|n …N'>üaG–Ðh³›¶ÚLÁ–‚søy¨¬²ŠWl6„‚‡à5˜ º’1âGóËPk`½_2SqËñ@’;aä¯úC0‰D /ÁÍ~Â23~Î\ƒé®Ž‡â¦dê]½^‘ ï$åù/¿Çn6‰.„ 'ªþWà =<"}8“dërU/B öoú®Žj +Xy!°à_M¡“ö³ËÌ?¯›1¤º·å~“Ÿ>˜ùÄ,pe&r†¡PŽñÑÞGS FF;ð¡?ŒöûÌPʹ²âŒ¢•ïº*øn|ÁW¤cƒ0ç …°8Å4뱘[U/jFU„qB²!áåÝùÄZwŒ Éö8o§¯²õ õ5HË?D å¡wù ÐÛYâçqL¾—%w'ì»V³aÔAµf°1ûDM•išŒù- ÞœÊW4b9 ^Óf§ll"üÓVxÄ*«eút3$½+Uâ{¾Õ„6>Þ uÂt_ÜÔïîëj@)³ÞG¾©Å{HÝ Ý•\©°#c=„ï] κ;‰YÀýa…5UÍPS  —¡k|°TsU S&ÂòË'!L1ÓgͲ‹Úf_/4Šôiœ˜r™©„摽ÖKîÏ·?/|ˉ-/ñT¤V-MVYPÞ5¡&Ú·–?Å^Œ1zo9ZŠáÖá¢ß_ ùŸqýK¡~§M/$ÃÁºŸÒåu‹éõVü%…œ¶„°ó’Ó,R œôF‚6ñmLÎÀÃGâ¨({D€Ê8nõÝÏ_Iio›2ÄœÂ#ÑÖàÃô«ˆv7„“X s.a®ð2MA*ãŒZ¼eíWrzYvb›óç H¼a&†@$=ßÁ&8bäï¼6-®QfŒ\üÅõ$„L‡E§:•À‰ ©FD'´aa)զȹY×+ô%àAC.œr›•«5©w¥ÅÝ £ƒýE¼’;Ø´£Ý¶ä£RX]ª«çå~zÿç `ØÒ%kŒAÉà°Ôè.¥U \¢ÚÐý“±‡zs n0¼Í7BnWDä:ïKU ãÇ4<§ô ‘ª=¾ž#ˆOÚ³O³z¼ÄtEL¤ßØd~Ê¡¡IÑñ¿bÍ%H1…š|Ì@—æ:#O•Š þX˜ì7‹^o4¦nF§)©Ä\W·5 ‘ó™TÖtI5R„l ¬tMÛ7Ð7“³bÓ²ë³[ÃØÁ/É®ûüÃ4'Ò{ýT¢‡öy±áx<Œ65§"eÛÌjXBŒó}“dJö¢a˜<ƒ=t1ÊþÄÖ!3³dÉ£ó(%ƒvŸà¶7HÑ:˜®¡#ûöœÆ Rz3-D»HD8˜Öí¥“/š%ûÒúÂÕ„æÅ³—‡ï9‰y†é¯&ñ:©û:eåî*Ï÷Ó:Šädp’ ½ŒìdÒ¥öl§?Ù5ã¹Í™\±ÄuÏEyÇ*w @»(§x,EÙWyœ·“šWÓéƒYѵ˜W5ŠˆÒÖ_×jÐ?± Á¯ÊM`+¨.¡çtC‘ #ƒ¡ãv„! ‚lž);/õÛŒ7,rÇâ™ÔDŒ6¨³A¼«™F]Ÿ—’9mX8ðÓöájÙŽô–¤³ìõD§Æ( kÌ힊0Û|¤'ÊMoÙ ¯«荒) ·¤¹-ËXŠM™áåã¸YDD¾ò( Ž¸’ðúa Õ pñuΆÆÍ|òÂx|”m o¼%÷‘ݳuÉ,òûPȺ’­¦=¶ACŸ§ltTBê|[þ•鈾Ñgx~p¦Ûxú¹Úk§œ—7ÝõänµykÌsþ:”÷aì2ih* ŒkÌú燆’ £@çÆÔ‘Ré²ù\µ€®zÒ‡žù³üùkÛŒYh7~DŠÂqMêÔZ"Ãéá YG*þpÇŒu‡þs†1Ø­ó]­Ð»*A5l‰wt>_ag°–èºÓ©±›¥Dªb¼˜†ZB„ÂSÖµúëv`ªpb)%`pmSÐÙ·‰0Úcò¸,¶ÙÖÐD_äïj‰)Üd{ÀwiÑk-IÁÔ]ë<ä&ŒL]Ó[ ðÉiçB$aªB>£N7¶KhšŸöa’Ç—*v¹7דK¡HK};.ˆ±à£DMÎ9ðäÔö¤‘=òÏCMëÖçîKY»Vy(^/=›DÎhB:£µzN»íðL›ƒr%Uj ªpž {eÆ"ÏejVH³Za:µ\P:Ò¤;§½„îqPQ­ÛGȪ)Í"mËrÄSUÖ™Ýú7„,üŠm$¡ëj3„¡Š¥¿PoÁ£|Ö¹ äæÀ¯¾±ÄµIOD,#n (‚êÄ`iqÇèÃãÕvœí”C}ÉÁ9¶ýº¦—êÔ:Ì­—,çÕË%ßß„11Æ9½&¼íÌ»¬þñ¸NÝiŽ%Χ5`ÏWwR^èeÚ¸ñÔ4á‚ìõGjßàý7·ž%|$ÿ“°ãÞÑ:ÚÅN¥ùíÈzõÞŠˆ2ðÑìw_ËE1«±ñœ£™Ï{!êÆ¾G§k–ÕßËGâ2‚C!øÌOîVêZY=ºÁ8þ¬¦³kذĚ­ŠÉ ASòëM|Gòzœv)mWׯÔ3Ç×0’)}/]dzh{†PúSå²ÅryxMÏI¼Uh'–ö ÜG)Îõ¢«8ª^âîíUd9:Lƒmx¥4> þú_5Ÿ2#ŽIg7Døû¿ulE:ãÉÍ¿ðC=Jý* W«#/¼iHàPÙìº#3¯TȬù‰(™ŒÃY¿ò-¬xHð—á¬0Ô–² 6^Q–O^•¹<&×®Trqá€Cؾӿ K²mÏD†9oòÅ“¨}‡¯É‰G#½±ïJqÔãúÓ@€ù¶ŠìÝÇ£ñpa,µxÖ”óa“¯$lìü¼ÚÏÞóuŸÅ Ú$eG)ã>Þ|M¹­%†Ô‰%ËF¤¿tó­£ÅqÉÅžÀš{k5- ”ÙÔª¯%딂ómšnC¾|t]ÑìWü•c,O•LÀZœ¾‰I†¢:!lŠ’½CJµü<{õë,·º¿€RãQa¶)Ü)*”é8ô&Ë-’þ‚®‰´”¼Î2z§ƒ3szs¡ZOv¸k¿ªÃÏ“ù±±˜¯ä %sdXLK„hÈtç„÷}5p%GBB`ˆ®õ¯¿´ç\·Ì-Õ¬Ÿ›ã~ŽnNú¡Å|’W ›#µ!Þìý"/NÆëD4Š&º±”´ ›êß72‚Fþ„g_ï¡'öZA¥‹ufÛö®.§]Ÿek\ÙÝÇûîj9ƒ¢U÷ÃSÄ®¥AáÁøí b°;âB¾R§S@Õø#æuô(¹ŽÞÚ4ëÇ…ut0^0ñm•8lϺjüÇÓ±X¶þ˜˜_3½8fijß, =Vc‚ÑQ…•~UYŒ–®R"ðò?cwwÃx:ì+SÆÚ²¸…uÉÄ ¶B}Æ‚/”Y |•1¹> °¢ûœ(<{­ÃnÉÉÝAÇ¥æ;ŽT0Uªû\k;ÈÁœžéc@×NL‰¦¾Wé·àÝÀº¯ È{ë;€ Î íØ–0çEî4mQæE]±d÷®èizòçhÒ$±ÃËöM’À¹ä´œZºmá] rT^dzéYpÛv?šò¼œZÌtÜÎøöјõ#ާËÑÉ[‚$ãéUOF…­4 »*L†Q @‹f…þê“Ê”ݪy¤63ÞQ`†¯\ÊÖ½GÓǰ&ž)feø <|‘£•ç¥ i£~¿CD]ñsû³¡œšÜ72ñ¸•Ä­ÓŒ\³; ù»§ ˆ›Wâ¤ï†ô½ï£FÓ‚‹Ô.$˜2[ª°Ç»%Äw9…ŒÅÄMœt\£(Ÿ÷š’M¤nwN”¾®_&é¸ÇÊHŠÑ}ú¨ûeî.¸ÓpÒ&éÜ\У?‚eÙ2´ää®ýËùI}&ûNØ2†ÀÒ‡wV5YkŽ@§Jdþ¹»}XË›™º$‚öþïû÷ÂÝtåËC¦æMiƒäy9 Ìß ËIhìá篌XpUff“ƒÜn~>_ØSÊ&áü]j âk÷äÃ'ŠêrœÁ8Ðï(÷€-Î_$ðazÚt0Tµc>™ì>ñë8WhT¹|kòæ¡'b• ô¥GÎŒ—Eìf»T} +qjú¤b» 9rÞ®ŽðŠ|I)Én æ’¿k³B¨2«;å¿òIí&!’.wiwŠ}¡îC($Åg÷ØÁ# \»/‡·¸ µÑ‚vKÈ9V’Æ…ß3uã,wÙ«¤œ"b¬ê†/MFGêºÝ¥¥¥úDt¼=+eX¯¯Õ@ìü¤Êv]íþåʵ±ÂÒì³ÏoU+úú˜²1 EâH¹uÍÆÁlüºd>òÑéœ$†}¨§ÔOÕ²¡–GO2ů¬ tx|'ÿE9Õ]ÿ65Óß7~ )7r¤íQ,raÙ›k«>í \¢{qSaN9ìð»Rî)܃dÝyñýWQÕ‘°xffݦ#áŠÞ,:&ú®S0cµt°ìÁ<8)á îû)æ…ß1{“&'謮s ?1íZûÛùB,q‰i‘m{óU\â¨C9¨tèi̸ýQóåíî®Cm¯ãKçzXçÎiÓDÁ|.Œ)¦pp_ªxçÅ~F‹ct-Ü“Â/ÉAm”/„Ÿc’€Å†$ú càÅV+ˆËåºùM¬èûÆréI„ã²ÂN\4W®RæÄU‘ºüÎñ¸yîwX¥láÑ?ÆI†ºG’!Ï>®ª‰Ð ~ÕzÀÒf—ÈïѸÅüò£¬Î_f£˜ºz_Ûg×öÙpw‰Ãóò~Ú5¿ÕË€ #Šb°Þ¨@E%Ñaq›¨ïÑÕ*\‡  aܵ¸öÎpÚ´ÿG7mù;:M¾ã±W9^=ŒéÏÌ&Û«0Õø{øç¨gUc à›އP^òÛƒámÙÓ+'Y°þ([M‚ÅânYÖ¬.foÞ™ÀSf=#Ák}Œ™ <Ɇ v#aA‚æŒOi³¨Ê¼ûˆ~ÕÇÂ) vv¬8âk3J,5>s*» úþwØ‹–œs¹SZŸ¼Kl 7ÚLáÔ|Á;Ó¬¥ýÅ÷µçO_!žÃ1Î/‡K=Vå”YÄ ‹²gtKhã ‘´Æ;½(ð눻„Ÿs ¢\Ž<xß“ÀÊ¡þzS4‚ož ïʽòÜb¤-˜ ywN¾±qyÑ­Ó;ÖiþíÓ5@ÁWFSa–›V/´Ä„„xñ¾‚™Œ2ïñ¾Hà ÏúØjxÇ=–äÜF¯—Ž[8²à# Oa‡¹·âŠÔsb_8üãT ßšž“!Þ&gUDzèÁtYÚè¬ó!KEKpïËØZþ”lÝ-C‘zäÚi Ð÷§ÄH”=;U%‹–tYÒø^ÿ}L"@göŒGôÛ×A¦ãf}Tïw÷£ZšoÖ j¡G³ŸõIß57×>øv²Ôv †k´kO=YTˆ~«Œ9„­P9ç;á§*E/Ž—¸„*tŠRAV9Ξû¨¼¶Ó2‹b O¦ËóÈ(N‚ÐÂá`KÀ³þúÕ Í4Ô=De$[Ïõ­%ø;–d©sÑ—žä>03†½³éuB›<âÚ[®”_;ÄA‘VnD`lÍÁýIÎma«óln{Îk•V4†²F,©Oj‹p·Ïä¤ëîWéø©g_pŸ’ 7£Ñßö:X°;Á±åϯɭ„L­+ZÄSˆø3Oœ-ªËÍGa;ø­coSåŒoœ¦¾â”%õ|°•;ØŽéwݴĆä{6»]î¿ Ô°£„ƒù‘$ŒžÝ úˆá×#'ÒÃ…`ôPàsŽ@$T±Ç¹žèAû1;òB›È—Ρ>"­jÀ…»­C$É;JMÉÙ˜ÛF lùÂàC} ˆ­4ª-+Öj³[(Ó¥‚nvøÎEËãÔb.¢$.À·ÿLªÊë9 ³ýÒ¥@°iª¯¿Íc}¹¯<þD“„9‹’À¡d£9Ï,„´k§¦Iê—í¨ºeŠß¯^õLý„µ7‡|˜@hÞCÚ—ý]¾s²a  à:oÜÞ6F¶i(–þ*o3'ZÏ"ªÃ÷Ä{ª,’›:¢,º¯Ò±\. Q¸½±®ÃQèx7H±~-~عŸÑW7G°ìZœuiÞCÁ>[â”ÒJŽ7”ú|þÞ)YXn.ÙÉÄ¥gèT8¹¶'óaCKÞx`)®ê93†¶ÛõZÆ>ŒÔ.“@$G`ŠPX(kÔJŒDpÂø(p-dwÏ·±àý´àœ§~ZòÀ=ÑT@…´æht 7°¯fý­ÂG ‹Ï%Ž V¬- Q«Âü^›ÿ}­:õÏû½XIJþ9qÃk9‘Ã3øÅô©0 ^8ù0ÿ5Ô“dp ›…žïœ9$–°ª¢¡a©Å™rn×_(ý,ôuªöbññ~ 3J·¾Ãæÿ… ^t÷«,ÓÊQ~ïÈ£ªŸ©~Cƒ©f‹ ×;¦VÁg/hÙn%í…õ.&ä²ë RTVz³·ýìÛüöÑ&v…Á÷ÏÍÚ€ˆzâAê*;%Ü–0h~v—‚*´d±‘·zÆ zžÆÏzÉǦq1b$ÁgQX¨Æp¦ȳÓÝÈ%5ºjÅÆ×VZÚr‚Õ…±Ê5IŒÖËKލh˜Á*¬Øjû–èppÅö*£9†:-í$Sx/¤n}•ËþÈÍê™p¤#œpúm‚ é®Gî*ß IûLÕ»d®ü<õÕFqÊunÓhxpÝ> ¯Ã ’“Šö¬m+±Lö² Zzì;²õ™•¢Ÿ}ìiMwÒiæxà•á]6}䂪þÐ¥„@²^ƒ½¯‘×™ËÁ)¹›]oBhµ©ˆ9vHø8|Oæ ¾N”Èß8[ÏG7dìDñRÞedÄ#`8 VêPôøNê-.}Z #švbv›V{÷†ÎET:÷)¬ƒãx3]Û^†¡(È~µºzÑXŽÙxóÕR±rÚªkl>ñ )>«(pÃQEóèWGÛÎ=$Ž˜ñ³¨hRòÓÖKXþ2ú{úÊftK,8bü[ô,S?ÇÒæBü83=§uÁúC’95™¯Ps©²)Äèg…2ì߉%&ÕÛ ~Á‡ò1ékp¹&,øcZú~ïåeŠˆ“˜ê„1 ñ°†bCÖ›‹³¥I¾Ò&o+´c§5hÿºXˆ_'*­PR&Ó_á‚ 5œ"5uå÷Oõ‚—®!K\ˆ[s$K¼ë^ß«òº8ÑpÉÈk³ÇQرùb¼ø]=úÌ’2sìÈl†~·¯*Ôj!røÕé14ãó¾ê‚Vv/Q'«É•£=ûE#1³ü#ïãèuªç9VÏG…î&èeOI`F…|•\Öʺ¨‚„(í*çiL]†"8v5jžmÎ=þâî]ÕѼ~gœ¡UІË×цít‰ƒÞü4â˜Óß Ñ?ïLá÷Aí#Ú%˜‚¥‚¾›°É éú Äj<Êj}kÒ-@ò¹ÜåäôÄŒn_Äï¬Øa’'W £ŒÆdpèÍüÁqúªíÊbLúÍDF{çQ>JÂ6–> VÚ »ÒLQñ2Šbn7}ÅÕ¿ƒ²¸.õQ½2[é÷¹Þ1ðCZ¨”<ó*.¡½ç«Á:LJèõæâf<â¨Q¥â¡‰ù†7Cf0 ä£Óç•í"ƒz›>\×åv@&þ¤ìYxI0BŠCï3»Öï{ )zzŠÂ[ŵGãæ¦1¾s B’M—ÈTUK@GÞª§Õb5™µÂ–Ý}¨Ó¹8ÒŸp>pÇ«[Pƒõú²ßE¨ŸÍmæÏ Gi¦ ox‚+‡ælôÀ¢Å¹7¾½$•9š$ÏvO1(ÌËÇ_B>&ŸŽøÎ/|û.Ñz$`ˆ±4™hiÙ€Ö Mw½­flê{³·–¨»ŸdQ7^ÔË…âS#ÿ³Õ7£öî¬,¥,ðÝì™qk _÷ž¥„¸éŸ{Vú¤$߆øX GóN®~õ|l6 Óð¡y?=w†*öt©±ºÛj‡ÌþpŒœÀ a?ܯn…O©J|›Ïø#…rR2„VÃ7Aã8}7cGcÊV/Õ€£A¹?}dXqE6œËÀPuP×Ìz*Û]šÖm$ÙûXB[‘Zj=¢KpLÓJÆÕÃyÚKëÞû4FT6÷BúèµÆyA©d nKbi¾œ~tç¨ßf€vò4Æ~—qŒ7Fúsç:“Æwæ7/ð¼!!8Êê•0ÿžKùÚŽZG÷oèÁžãuÈ'RÞoª²Œ-»Ã"éxµT>1dz2faPÝI±¤\ݹ¦ànôM1—ù¼/ò¶áËy].ì¸n…èö6&¨¢û÷ªÛ»éËÙÖ2.D%Š Lúµ½; ÄE%ž‰Ah$–'náÄC ¼ÊpŸU´e'~šg«_‰ÀP`¹gõ`Ó5ŸjBŸL>Ž`ýpú +í[ߺ7êÅ«µR¯`TÏŠž킜òñ®ø"æÄâX `ö’§" n‰èÝGüô<óºT#®ÊgU b.¨g"NîÑø±ás—Q2s7öÏÆ»ÚJ×Z¢ÏE‚ÞÅ/=ha>¦ „ôÑÑZ,ÃúiÐ’ÚÕ¶ßµì]¤ºt Æ ÈúMÐOÁã~3%Pwµ·“ûqüáù¤bTǨ©±l-„|¯æú‡ó>¯„4iwßá³â¢æñÚO“õÔù"_ÓZMºî“¾7°ð€NìáVêp¿Kyuo5áK?öŸN %ÛtuX^M±˜òpC vÅ}èXÚýy?08@‚…/\³wcð-tç'Ym!ÔögHˆ ó„‘ïÐÆî®²µxF8_¸µ·Æ’€Îé~†‹ì]mDù¶ÎYö¹FÔŬp­æüïçHÝý ÉdÍ(ehøÒi!¬(©FÄ “]ì šìÌ©nÓ·¡;`Îæw÷GD©´Ó¿aÍ!-ìhØÈÞ¸_¤É/…Xàù°&¹î„Ddï|æÃ²'pnM=ÂWf*›Ç£:<õ() ¬êG!Ê3źZG4 Pä"@½jg‡ˆi“ ÐDýìùné–CñÌî¬ËíGµZ%:6ЉGEƒ¯ÏeÎ~FƒÆ†'@ ²×Ì"G±¡d\\ânŒ'Ê ƒNƒPúì[©„£Y1ϬiÌŒú,Í*)5oê‡ _LF5 å/ý¾ˆ¯Qвɏ ¶§•Ý/Ýü?ödÆtëîöè—¼ê>+¤}uX;×…0pnúè _ÍÂ+g]²ºÿMk²ë5ÚmÊ9IªuÜØZÈáÍe©!¯j&'½±ú0—þ=+–«×rðq­ë{Çñ~.Yó]Ÿ{'ÂX ™ú<N~ñ€Iâ»&T…eß³\§Iï>Ù+‰÷5»gÊZü]/õü3ÌC¼æË…‹œ‡R ‰ýÈ1Ѭ4/YÓùRhûyÒ¨ÞÑš®ç¦ÄzÙŸ•vëì¹e&8ÇÆNb¬—B‚ý)öv…‘‘0ôx¹ËhÐ^é ‚&|„á#¡ü5\Jæ:€b¬åؽۺׯKj¦•éUÇŸUOC?¹XžÈçý(lzW"/€¢DÇ õEùg…(íAÙQ?Á¼©Êboä\½B=óšªí‡í¬™¢¢øi÷- /³ë•¦ŠøLvÁ—Ó$kÞÎnfê;­ó#_›ÆÀMîë ;ß¶ÚΔá4ýµOûA`˜ÄýŒÝ£`õQP„×SŠ;”ÛÄ3¶1pÕ¾Y£Ê™R\wíª‚`{1XÈ?VTÕC¥¸Qê–pOt¤¤¤DòeÞ«õ`MÒº Y²ÕKFŠ Ö©_u­Q¬r%ù|é@¥q;®…À?Rc ?eI1-nKÖÝÁJ±”ÔŠO‚©Ò~feVßhbj¦ý mæVF94˜™¼•¢ ÿŽÆ¶à–Oß3¾HóãÓÒJ×R%òRd iîsK9,a™¥ŒpœƒÏhvIgBŽÖ`J7OÞÒ%ј—¥jO®&hxêm´†Y¯Ðß“¯P[ïçC>®K1pE×ÚR‚]»[0ß9˜a¥ ÒÏ H0²„ˆÃ;U±z§Ö;!Ÿ:>u+ïø”õL*^9x®Í·ZÑÜsEZÎ%kˆUðk„‡gŸ¹xÜcUeI~Ÿ   Ì–m¯`Å`Ï6Ä—©âTíeÑ:k̲¸™(;6›˜­‡„k $UÁGˆB+®šÍfÖsí·G']ûzÆÚ¼?p’{¹ýǧJs“aü‹8l‘þŒeǺøÚµ iÆ.D =!Xä;Ѝ[É”•Äp(²ZÃ{ÉÁ‘"#¤^i±þ–m{k!ÿV¢æóD €ò©TYx˜Rc^ÇH}¡†•ÿ{=è¾ ¦ÔÅí*–b¾ž\~TÕñ|ÑM× XÊôí¯_ ž¯ÔêP"ÛØ Œd\R¢jùþtÅ=ÏíPM7ïÊ"y$×Á—ØŒ¿Ò2²qönÈ"Õm&úœ„)!BqÍðä²awƒ(¯6m'n¥¬¢Ä«á\WJÌŸ««ª^÷™É~iDŒCWŒf~ nt[•ˆTÁ±ñË+Ý»,%DÒ¥/Ý<±@h>ƒþö9·åÙ+ÄóoBø³iôž[,!“§MûûfÁ¬ 1M°Ì`þ¦Æö”Aë–ŒVÕ>‹›A°ÙwÕ¤†ª~ͪ⃤go,Ï öhBP <úYo0…Z Ôž¥î’6wç'µª¼š½–®ºº}*„ë‡xJÛ¦Xîý7¹%ê@cH hAë¹(TÃÉУ³„åwb6—ú…å¥l—x,€ŠPÅcÂGM!ÈÑ&sò©ggæ«tnÕ­õhŠ:#æYä÷2®‡ßì h /!Ë1‚§öslÝ)6iG¥s]o -’4íঽaÒ»¾»P™ÔÁ< ±Ë"VûqQßTèN½~7‘Xù¥žö" U+¿¼_Åþ†ãq–ÝV®qß÷P\ä%\¬˜´6äõøyžïæ\ UpÔw6J®VŠ/¡p7¥F Mì å2ŽqÑa’”ÀÓ–ÒÁWÉKÃñ13&Ý’ÅÏgŽ{—™-”"5k[?\a7=æ‰ ŽQÕCSŒÀí²ž´õº;ã:mV¢ù©è,iÌFûbè‹¿½Æ±cÖ)}÷Óú‡Qû~k†þÁ[SÃÏ-;÷ý„ÖÁ§õE[LžýgL†Ž„|_`ô—v ˜E¶˜ÃØÓT§¾vQÿÄî,'áÐ5â÷Ç7”q "íEf²{V' ±º4ÑBU!.µed• ³#[šHü¥l¿›æ-wÆaöHmfßݧΞN¶Q§«_Ljúõ¹~‡L£žÇ Åáä,3QÓs‡oŇý™Giöñ3 ¢"L;^k´d¡qï¥7 i.¡¢-’„ïmþÆ¥ÝÕÄTRF5¯ÿÛ=¥7ÂÕ9Ä‹k¡r( ÎPÏRäc4}©RˆD¨>¦¼®k¨º'zçûÊ¡»É'…©G‘^b* ¥rL]Ê.ÇÍV(@<¾ô£àóýùÊ&@]€ˆ­¯'1ÈÜg'&æìbý¨`åÐ+5;6g/ÔOìcZ“Ë"—ÕáÅ}þRò>²ÁªwX°ºÓ¨%?«Oó e?©éél¸É¯§bh€ùJ«µÇóÞÇ0ݱ“;U =#Àòõ9JÝžepúˆà±£UCu)sX@Ÿõ¢EP ‡%¶Z öu¥ñМVÅ­jñ‹‡Ô¢Ge^Ç"»¬Ä¶AEÅ=·åͤµÞ‰Ž]®ÒU•Ѱ_ë©ÿøÊŒX¤G³§£RxEd^\æcìe#¼l-Ç›ø‡-ÉUµŒJeÏ»œ<«Q1jwÌ O0S‚n?_•f)¹ ÉÂŽ0rÌ?ð9~—s°ðÎn6aýY;(&RêW…ÅœýÓ¢‡®¦Ž~\ÝøÞETXõé§ÔŠæ¸zHiŒ±½mH^U…«¯¢.yU¼wño´,Oâeá7A¤ÅØŒhÈvT%¼îÞO·tæ Äœnô[E1óx9P_MÔö]óÑ 0Ö|®˜ë“μÖ?í³‹ÛMW¯Š&íBVÓ Xþ¨ßiEž?£r ”aHCÙósC$“O¶%ÀNL½3¨Âin\9‚T׿/³¹oóŒ‚Îõwüt×'“§akæGsZjè±íß⣘7#2ò¬£/uÓ1X“R/Þi<49txw’cº@ÂaÝ£¬A ZK†KáéŒÖ×Ë‘f_M› Ä¥I6¡x ÃR÷–µæ’“àMÓ#®ûf²“´;”\ÖV ¯ÌÌÞ͉†‡þÚMöëTÛØâL°ùüæ_áGR“w/†0C +¥W×™s¯ÏÙJÖ…csû‚N+§·¹wæÝú²†j“HcOöçÅ+—Êâ¾–â9¸ß0O:©FAŠà|`Ý*¥ǵÀžy°b «Öò…L \ç¼í“¿ÁÓît Ó\±¶vÐν\“A¡~¾ûr©iD¸£ÚóFØçã5ɺÔÕÉŒX4µUjò*Ø÷ȰàßÊWr¾èÍWë,‘ÕÉÞB¤zèî7íLO%"13Ž;?U`3èÚÄ¿ÝÁœ˜3R¢uο8÷rÝe#!ϬëÙ:;SòlM‡„$%…K GGžBÝuÁ¥^‰@Pª“ðéÄ™ûI}£#³ç.t}êBbÁdbÚ¦¦.Ê…Y¿ü†ü®äž]kÕ²i𭬫i4µË5lxµä]àChPmrÄåUoo倿b¸e)®m<¸áUF„¦;(}²:Ô¨<óµ®èÃÄÞ#G âõãi7ü"Òtç_³ï(¦|ÞÏɦC[>±§L¬Ç4-"3~U£)U.ƘUš,¨Ê׃>¤­Ä”ä;0S¦5¡#—1`õÙ .I06}?ÌÁ7í£ À–h÷Sm±…z@®¨D˜»æÿÞ;IÿÃþfÄ 7¶¬Õu ]mëœNg‹‘n“¦yºê%S>0æ>…ï´a3›_ÕTá³G®'<ëa;#/¦sÎ\NœšÎcwW(ñj̆ò^Y(Ùv4RY"ƒùg,¦Ê½¥ ‘B~¾:8'f²%ٯݵ»Íô;ÂzÞ¨;Mu"TI‡vû‹‡œÅÊïˆfتOŽG¿¸×èØˆ“©tk“ˆÓ-~¼ئpÚNmf}¶¥†ç3«oØé§‘•Àÿx™Ýœ)BJ”ÈÇj·C®]? [Ýfç'o²:."]‚´”†J¢á«Åvr\"¬¶ts'XYòó|û)ßÁFûYdûGs2ÇžÉ+Xvoé(»(ÛûèE´yË dTíFF¡£©½£ŒvÏ]‰–}²_9ˆ¹Æ›H\Ø:[å°‰°Úï±ç ¨æÖ ]M©ò©¿¸J‘WŸ®'—Ãk¸zÇpZš+õEDo°¥Ñ} ©5^äÐh?ÎH1•_§‘Ýr½±·N¡+˜Ò´ßÃJr*PŸ–Û›4¸ŠäD°&ØEæ½—áÖ¨M3+ l¦ª†ã¥4+Ï AsæØÊÆ©;$ª°í'AólMhVö‚’Y»óI;‚©]X°±FÊU: ù–X’·Â¥Ib*Ëꩄ + k$°n—Xÿö(qJ7¼SN7bŸhIØâ†«%H—Å_eéÇ%ŽR\9ðœ’ÃMfÉsx%ôƒ¬ÊŒ'¢N.t=™P>G‹} ÎpK½}“ËÔ·¢ÅÈ_g2/B‘Wmp‚|Ýü×­òp¦ù?R²’K§ÕÀÛ®µö‹†›ÎÔ÷€¸yw\“#ˆõþŠ‚Neì©°z}Tlù+âI‡=—–öQ-š|™©E,ïæá˜‹`çu|¬²ƒ¹À¾Ò†@¥»&µE7r‘åöq\êQ. ·h Nj°~áV™“R %"W–|°¡jíÇY%{™!Ã;Î÷FL‘c>·n„’"ãj,g?mk½.÷Æbj3ýé^¹½ W벑ß(FYi©…ƒoØÀMæÉtÅÞÂ:ñCd ý¾"¢œè _‰“pAoóçåýÔ†µ¥*AW¿ñPYx÷V(aÚͳô ù,÷¶–”(ŒüÙ•N`[Ç*­¦ÒÒ©ÒÒõ§¬n§ÞfΕ•¨6–.8u°ÒkÐÊ\J[²¿lak$áX¾?“æ©Æt¾^Ì×b —£ég.gt˜N‚,ã²ê¾$åòŸ>ž€g¨®‘œZ›ë7Œ4$È-dµp<ÐauCäSP‡éxÒ%…h+£9 Laûðb(*ìô8Èç¶ÁSe¡ÍóvYã1ƒ/=,ª*2ñð_âz+7ƒt«ºñÿ Ú%÷ÝÃòu­­q9š-Ľ,áH<~ïF‘œH »P%OÙ$Ã7ƒ1ñµ¹–â<Ù\Ûž1ÂQºÓ|`IB ‰D«šÉDyUXÇl[—ënÈõ:¯)=FÛ&ƒ”ÌG±–d[*~:Ü£Fiã’Ì(z£¥Fœ†¬Ï@ãLÎçÏ o~G’K™¶zEî@cA#-º&`WP)µ£¬â´Ìyí< EÞ/*«Ù]­~c0ù£eb# ¤]îŒ,3—âX„é™ü•Ά¦VOr,!ÄG‚QisoØ>ßT†ï„{[g¹þS· ¥´¦qÍÇøm2VçW#'€‡-H!mkG,ˆÕˆîù[E8gj£ÅÀEÌ.ãÓo ƒîºE†ú¡;ë6¹”»Ì§ÖÙü¹óõRCðŒ7~&- ´Aáˆ6Þ ¹ôŸž«Æ·Ñ½ö¾¯ÿd@Ð¥J pK o àW1äodÿ_Ý/PÊGcX¿gÒËm Û·NTky•ÐÅ"W¯iÑ$ýtm¦Â©®`{øGÈ£œ i²æDqª»ü³ÙÒÆ¶ª`O`gê£?Ï¢, J2ü,ik܆žS$àJ©P‰Ž´ÍÞ•édÌ  ë5ò4:í£ã´* åþX9..ðDàõëÇaÇðrSgׄäT:laÜjDöù‚1õßH KL§UH`¹´R«¸TaRÓç¦}Âë|Ðñ§Á³í*”y"ÅkFl`ÀótÄ©ôuí’ëˆñaw‰ÒjÞ^‚W©efM þ*ö:â›!„88×)$Þ¯·ªq˜ÓÍTA_ŒŸ{ã`S嫚dÑÇÏ€Uyl…³9x^‚ô]Ku«·èˆ¹?¤U—Á+1´øA¨O’iù>¼u´Óhpu|ø›þг¯k(_c -UÙ&-•¦=WlËc?¯XµÄ¯a# ’ [‘Ø\(ø<âGYü¹—[TÜ‹a jbju4ÐíúêY¼wÞb¾Ùßr/þ|àІê÷&¬Œ1&p¬kßêW¿È÷˜î£þ;ô溱¸‹«ÊðW®ôÕ|›Jîr+VôÄ]µ´rû=l›ÅDvpX<Óy?1(ýGÙ[±âw\š½´¸ëo‘Ï»ÿu‘¤Òš‰úðm {º¼ÅÃv©Âã(>‘Æ3Ç€ÒSuö+ƒmŸ öèÏ¡MÌÚjL ¿qVFÙcJýŠIÉ90êÄŸþrî¹€Ÿìf&›‚'Ê.à×(dÔøO®‡tBÃ?$iˆ’Ì0ª Õ‡â¤qAú¿@6õ 빸wº´äÍä4ëDx\즷8 ¹%‘ž:æ0Šb9DdC¾«=£tÈZ^?MŸZèÃ{#–{(”!}½eב(ûJéMU¯€g=Qã@ è ¹›Ò…KÂa”Œ› fâSÄ3*~×r:ŸÝâ’z¼áë¶ú“I‚ ä 䩽v<¬&î",w±ŠR^#¡'Á;:}žháaµa“åùT)¸/)@|ÕgkçIzY±ïð(à”‰eëì c^È£ÓïúÇÐC§a¿ÞW¼5²c©>ŒÌðJÖîŸÎò¿•XWèf“nRëaËT~ÁÒ 1™Ch‘w] .ÉéÞ¬– 0é¤>5¯Æ×ÝÝT´Æn[ÓþêdÞ”‹^`Eh W˜!ª#UŒr–©É·ú:N íU.ûQèp¹5çžçðA¬%9$€]¶~­3P~gÏij™Î¯ë@ªÈÜa¬ë °¯Äóî1M¡Ý’ôŒ&Ѧû‡É‚èë C–ñXa¬,𿢦F)²Á<‘÷ÄA"Òx™/u™È~Ô­Ï3ò;®q D-)Át¶ö½‡åû!`¸38U§lîxC¸ŽÌ«½øÄ¹åvVÖ¥ø ~ë™}pG8yøùIT˜ø Çî’ 'yóA8ã¯TC{®u×Õ:à›»ÆVí•è' ¡ò"ëÔø”l×?cÅk„‘Š9#ŠÕòŸ¶;·ÊóexKv³Û%TnNŠYcmoøó__Ää­}6qéŠÓr#×Î<¥EëT ñ…"°¡djÞå³è¬îA3o°;XÁƒF_«ÖÃ1ÖBñ›K>~eÿÄ9¢îp-£(܄ĕdúßP«}`Ž.þQ„¥C6‘5ÿÃOÞX‚æƒ u@º¨zõû–£"¬Ø¯yTÜE« ¿Ë “ØŠÞkàmZ[+UÐbOV5š ”ì™,3V9ÆõÛﵤEñ¬Q@Uþòi{„ƈdsòÂ]8@€+Ù;pÊ)a¾ÝHËÌÚ/¸&ãœf\ ‹Žr¿€Y¬}ÖTuÒÛW_‰U‘Zhûis+Âýgßt[î'ÐØÿq3S“5Ùxq¶ÛS)«ìBËO‘>)u¢@FxßqB}ȪTGçO¶MH’»­Äæ-û:Áatö:SÆôbÝ{Ö÷³!ÑU ÀäbYù¶ª¦ep…X ±m:WðTÊF}í(^Wyoðb\í&:nXõqû¬ãˆŽŸ¥¹°Y±¢Ò¯rœ÷§% NåäQ•6×ËM^•·^%“¦4hHý †Ôô¡,c"ô²&VNŸŽ :_óBa{ªg¤4Ëéœ:gªj³‘·)Såy¾mEcFˆ˜|ÿ¢™x§¥ñyjÀ %'Y äy‘QÌ].‘„Ç=ûÙ…ÆfÁ׉¨Ëâl ÆØŽWÔÞ$ù4ƒg>û¯Ï«‰IÐ>E.˜V| endstream endobj 136 0 obj << /Type /FontDescriptor /FontName /TCUVVP+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/G/I/L/N/O/P/R/S/T/U/a/asciicircum/asciitilde/asterisk/b/bar/braceleft/braceright/bracketleft/bracketright/c/colon/comma/d/e/eight/equal/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/nine/o/one/p/parenleft/parenright/percent/period/plus/q/r/s/seven/six/slash/t/three/two/u/v/w/x/y/z/zero) /FontFile 135 0 R >> endobj 137 0 obj << /Length1 1557 /Length2 7560 /Length3 0 /Length 8594 /Filter /FlateDecode >> stream xÚ¶T”]6L J ŒHçÐ-Ò’Ò] 02ÌÀ0 Ý ¡ˆ€¤twI‡´RÒJK#R"¡ß¨O¼ïûÿk}ߺךû>׎s®½¯}Ö°ßÑ5àWp@ØATp¿H¨¤mh(D@ a;»!ƒüØ!H(.ýJH…Á”Á(ŒŸ6Ôð„…D€BâÒBÒ P’úÛ”*ƒÑP ¶P‡xØ•n>H¨“3 ³Íߟ@.{n ””ßïp ‚+ µÃÚ`”3ij£=4@ØC!(ŸÿJÁ%ëŒB¹I zyy €]=H§{Ü|@/(ʨñ€ Ñà/Â@°+ä3;ÐÐêñ7@8¢¼ÀHÀ ö¸&ÂîA1› Ôµ€Ý ð?ÎZø€Õ($ ôOº¿¢%‚ƒíí®n`¸ît„ À‡ªZ(o wøå†y 0ñ`4 Ûa~Ÿ TUЂ1ÿ¢ça„º¡<< °_¥ÁTYî „pu…ÀQ€_çS†"!ö˜²ûþé¬ á÷ûká…;8þ"áàé&h‡º{BÔ•ÿrÁ@€1' ($E„€w ÄÛÞYðWzC7Èoãoà ÀÏ átÄ€@!˜ÀÏŒ†QHOH€ßþ{:@íQ@;ˆø7;†8þYcš„z-@í A¿ž¾¬0òr@Àa>ÿºÿ¦™’¹®&ïÆÿØÞ@?~q ¿°ˆPLT(.% øï$ÿÐÿ›úoT ýëh ó©Ã@©? 0¥û›ú/Qpý50ÜÀÿÞAQ2Èõ¯ð-Ab {ÌÐÿ³ü‡üÿ©þW–ÿ›ðÿ÷@ªž0Øo3×oûÿÇ v…Â|þrÀÙ… mf4àÿëjù3ÈÚ¨§ëÿZÕQ`Ìp(À`ÿ”ê¡ õ†8èBQöοåòw0ÙaP8DáýuÓù…@ ÿ±a¦ÍÞs›x`ZõÛÁ Ó嘆G8üš:a1q ‰û@q ‹‰ý„0ãéñþ­k  „1쀎$àWC1A‚aà?ˆPЊ‘ö?€0@x"ÿ$€‚pÌéþD¥€‚ˆÿ\c’ºaZ‹pø7F (èAc®—Q ‚‘ý?k!  Ê ùAÄ ñoZI  /ùø¯*Ø{"‘˜»á·L1%ú{ýû"‚@¼!ö€ïö2áêÂÛÏk½ø7F妨7LR¹ùý> ;5ïWýeó#ßoèmV‹k_•MòÇ=µ )žfϵ˜¡c½âg"ä¡:ô&Ÿ>=›¢ÊûÉ¢Ï Ø‹)ô3_~ö}Æw±ÂPØ£‹žÞœŽ ÷”jd‚ÃOq;YƒvÖ¯¤èé²Ülˆ[kÞw¨è`}&XxG˜_m–7ð¨ÈnKâPçœöÃËç4u•!Öt(¢Ó|ùnO·Ö^j߈ï…ÙçÚPÐ-|­ºÎ³­Ú!Þ‡ ݲ-º™Nî_vCï@-³œ—½ dy¯L[8²¹ƒuiT€ËÊ×`õ¯¶Ø…E ÃÔÝ›~àdh¹CfšÕ‡½Òû)¸nwzçgSd:¿wò^­ìíK´_qy\ ³º'ãóDyély8,Fõa*ö/Tx*¯lpÕñ§·†*ÚŒò§¤ÆÎÏ>ÞMµÚ”¼Cšb N)n+ )ßa8c™•T1–%½—¦#“ÁÐ8Eµ°qÖa!5̘x§µÀÅæÉÔc6,ý‰);Ú™Ô §þcB’AÏgÞ%SÆÚÎ{ÓZl"º8©ó_nN~ði#˨¸É׉`—àŒÚjå¹ñE¥ \¯ÖûçÔN±}Ûײwv’±H6[Õ®uÂ¥bÁÚÌ äÙÉ—öˆÅó«àDiÅÓcÜ uK¶À¹éø¼úÝ౨°¦¼g©4†ožd tÞYßM®ë®Õëª7àÁ³ÑõY7râé¸sWQFÕá¦Ï…)ZU Õ»‡Ãd!˜vlkÒ>ZÓbrçuåÅ£“J@ØÂWÒºeà ~çGÍ©Óx] „ÒïëvGììû§ò¼¿Ž;1»Ðá‚Ò­óÍ“­FKê9žŠˆ8=ý*²¹ëå*ü ¯ë™›âÓ õ‘`|²héÀ8÷ U²•ÐhOÙEk:¬Cu¹1ž qJs¯Ç“öûx"!”ï%÷Ìjß Q<ÒTðˆ•i)<ìk{öÍvTßzÚ‡EIO”Ч-êÙÒÙIã[?zϸC='UÀ8}!V§›{Ð\Z“7ÒÝ‚lõÉ©ùÀ¼ÞVè‚Vʘù”)ǦP½v·úžþ¥ W½i$”æÑóaUÜyËw]FTŠXžeDÒ;[L‘•iáo}$>¯ç´ÇeêsnvD·EŒãŒûÙ‡ 3íöb|=G|N›—ÞÐñÛh^; ÌzМÅçW4OUÞs¥ø&%?qô½s÷º(½hîÙ¾Œe¾ÉjÜk¤2˜}áUy~Í‚}ë~[Ð5àMã(pý¬Î›J¬¿K1ÒY«›ü³èµŒi»PÒñ™éeïþ£hÈìEC—ÄWÚ3~ëG ØyŸ,n~¤08E?«‘zB-RI2×ñ†V¹´›!á{T4‘=;Imó\Ú¼Oµ†’\©/þî‹/´”“( :-íÅPé½PãÍ~‘×´ŸUR´Ý¡‡™r&nEQ¼k°÷qÝðÙH§œ’»o¤‡ôBèºï7èNòïyÞŸªU>0¶dô0(Mùá-Ó¨ª#­ât‹KÁ>´úÈüˆ³V<¥:ú®NÚÈöüšYjÊ–4œ|ÇržV+ãûµcEÆ(®÷Û=ƒ/ >>Eî-.¼.¹ ú!#—“©Yï ÎÑžP¿‡*í4°ˆ\ß,½Šk;)SO²–ºõпÓSÀúÓ„å–ç1C†—íÞ„T«Ø½º‚ó%B ’š¾CЗ“cÓçE‡ƒLngÙx£¨ãÅjfüAÏmkö¹¢#[&ÎSÈ0(—O݈%Bü‘\.ýΈWÌÑ`GaçÇh¶á~‰Çôç©×¯oŒÍïQ÷ÿ(íwQt³”ü4»5Rö ¾=û(eK2îa•cAž«)œ/æsêðr ©n:NúcnÐÊ”ƒ,é‹â¥·CªÓÛ ÇÛ*¯£–ª^+3¢V¶|ßKÁ‡}r#˜ûû€/×P¼Dë¢O£ð~ Ý"çËq›eË»ïÍqíÏ|ÐW"ú P”jÿ©†D‰bèJºlÞËñžò›Jµƒº›ãØ!Ýž¾Û=þ™ßœz)±‘ü4é”Ò 9ÛžëØÇ¹ô'+½ÊÛ–2@YÓÓ[/†)Psô º³ÕNŽ•ÏÐ_x~µéê§Û­U:E&Àk7Ö ˜Â…»3jö»XDª™gÌfYˆØÂN…Ô¢QêË©Õb‚aîÏq×N*çTöú)NX­[Çß³ùlT)^æ#ü©=×-*>Ý<›ÖSç·¿ÅæQ¼‚§{YJ‘8üN`ýù¨ÅJ63ýL},È ”ܘ?•_)ÆÙð­·* é6ŠíshAõØö~’¤þ¾vÃ,HXdóeÈ,¶Æ:ߥJëBº¼’œX‹…ëã4«Vï4ÉOæP¹˜oYÈG—I_‰¯ýp¥]úecX¦Eèš­>Ê Bãnݲ{\Z1Þ–v÷Z­A·iq>ÌZÔB;ÙòKsÈ(e1öieTü–xhAÏó·kã…ÞøÎ3©ÄãÀGR¾Ni=G‚­ŸDUiYÖÓ bò ç—É^(H\ÞÂ3xÎE7»åßÓìÕAÉámļÿ íÚPbGВДŠÍ­-¡E¨€T_¯õ×D 1M)Œ½šòáÞ<Í„—Aw--Æòµ¿C»Áã(öœZÓúp;hàd)m„Q-âΰbZ ™›u"56µìUê}åÿ¼Ûnu—8‹áƒ)rnt‡äÛr‹1x"o)oüDï $W/Þ|ŸßЗ‡(©0K2… uÿìz/ (ÑØRdAY¶¸P«.x¾êÀîË}Lí¬ñÀ;0¯o¯¤õKbÑ‹ØõÉ<½œ¬/«5œDcA3ÆÕr„ðË9@w:ø!{~ÒøMB‚·„jO[¯h!KjM­«Op§ñËÌ?ø÷Dr«HI±‰—|),¾Ð‘†›wišïæ Ï Ó óÇÃò¨mä|gPŽpð™VÓÍ幃t[wì&tT™Iÿi\ƒ,7ìo•7²›Eª¤'ÁË»£rÊH_ Í|)«ìöM¼ª+¦­^?­&£šýq:÷Ìë¸üyŽ6CLØØTT°_-2zK®Cu‹ ¥;V/ˆÖˆ«­ÈÆ É[œ*Z²ù¥Ãf ‹Ç1®¯/¨³Ø€îc-}.|zuñÎ=ŒÌ7ŽÅÂã‹3ï(ZY×&f¾u–;Û=§p£WUŽt°«#Á/\óz>Înˆ­ œLÒ¸!YÊÉj׊½£@þ,¿逛íônu<Ø›/ œïÛ<Í—¼tÕp•æ¦tSfeÖïSdV9Ø×06Rª•Yt‡€Ù½€ªœZ;£Ku:¿´Aa‰r_qK'÷HT_º„/qж‚‘¸¬}A'ë¨ì/é|ùEô/Éc6?4™¿³HI‰Oä˜\tVþñÍ¿FAoYMéPÚ’3¦Ù;¶&Üræ'‡È¹‡ï‰7Îñ tݲÜöBò¸ $de ¡*Áì ­j“¤©fö.ƒuÑã„øLD§½öÅ’±µ”üKjûÂØÛÍÊY¤žÆ5È¿<¦KÇnƒD×ó‡ÎõêN†œ¨‹òÜÍñÁR™‹L;³b7¹7rvø‘*€¯Û€¨àNOP·Àq Qb*:µ1.ñøÛ1­ýæü–Z}¼ór³ÆHêǃ7öN“ 2 ¦ËÄf²`¼÷ÈÞ=kǪm†‘‹WoU™µ”rV–|·ÉscÇ#8«¹²²–vöÓÙuâ‹LÖ¡=q,„ÅÅ¥¸ÜI\µϼ8yv_/$ž‹–=0-æ+âº2—~$øæ´¤úUE6}½ yôq`çθšã!­Ú^Y|¢è‹P¾ÂÊ'»rEb—çì¹[¬ÖU2Ä8Â烽Ÿ¼Ñ¿]t‘­%ÈÂÙ” ÈhZû Áÿ6ÖQþ¼:2’<©°A·ÝCެË~ã&)·’eKÂ?¬öl(`”½FËÍêèRûç) \©1¸Œ èšU¾™˜b~(SÒÖ¡¨Ÿq›»²Q±c¼[Ó{hðcS¶/4UܯÇ,ôÔÓ¥ÙÇΤ?ˆÆ†;ÉÄûU¡O]Êû÷Ç×ç¢áÃÖÑ]’÷—ÚÊÌΚfš¹æª¼?$q¡Þ²p˜šñ·UJð”3¼¹GTöQJ^2N`V!±>+ ÇW0¢•móØÍ‚u»gN4Ïg<¼V c9ÊK‹ÌG𝹕#TÒ?F•AHGÄóšI{ľÍ6Êg0ÌUåԽſ\íî!Xv‰ZgЏànL‚ÕŸå& xÛÔÑ2ËGéUšLˆG¤sÁJÖ>“2ÅlNø‡<œµãWMÞš°H¾]¥Ÿ©«|ë…ìtíšòåÃeüæ—VÊ=U=Zø,rx‚iy]d§ÝšûoO+k l߯ ª†9éÙ ©>݇_ºëV·µ6•Ûp¥:Í©1jr¸öµ¢š}{v®*m°Ëij¶¢Å¾¢ü±„B.äIϽJ,½Iͳ$qëIw|,ÓÒõqó–'÷Ç`X#÷§™ƒç_F/lö4Ô³á•T¾»ûd±ÌÙä7úúŠ:PöhbÃXqÿë~¢¯Q} ©‰A\óg Ïãr<µã’mB›én‘>í«ïŸV{¬#^¢6q@‘u88j²o.§g“&éÏ—zŸ¬_€Â ì?LÖ|ŽKºGŽÕxêà^öÓS|QajqwI©O]äQØÈqB” N0nv«‘)%oÁgz¬÷¶‘Ûñ¬1øº–—+Xa­ë ¦­`h^ò ’ðJ—>†&â)yâxXkÊ|ûwçþ4`-YÑU,,î'!tÂýÜ%?J•ÉŠòarR’ô›yÝ‚é*{¼òdŒ¶ Ó¼=†LÓø©ƒQ‚yÙù yª§1Mê/Ÿ ª(M?° ubê!a‰DŠ“U®>=Ñ}6BïEQWý_¿ä|›†¥€¼’D4݆Ÿ¾/ŒÅRUŽôZV`Oµûr󔥧ˆÏ0Wû†p̘ÒÃÅ+ã\H¶\Òù[‰UÏü$›¸p2?ï Ò¤l{ÒwÍ ¯…†"Ë`Ä>¤½Öþz—Y¹5ZcQöôãà Ü „Dc%Cßü!t½t‹¤]_É^ªŸ·jæ½M2™9ä!çå7¯ø¼Ìç°Tæã˜[⊞~i…"ú!ƒÇÄhû‘2Q?Goõ"ÆäAUÍ—3ÜÅZN]ÛVOEG6¾Ìƒ™¤)­³éõã&¹ /r×W WFcBŠé ¢dXHq”áC8Š>_|V›Õt8líöYˆ{û†ò3GЫ‡”&ëWX•}m ÍøÞB=“ÆM=ÃÛE<Â>Þgf+C—S%v'R¡eЖ¢3Üpæ¬ë¼®|ø±Q?>ŽÿuxU ƒÁÑÎæ“ð­6y¢£¢™ÉK<–˜;e)|Cc4Äáåè1ÚÛu¸í@ÛAVEðfBØ{–ž»Ïè¼Z vLØÍ`ѯih¥ëCïU™—È@¦U»ýtîæøM~ ’—ð9wP§:Ó4E䑬آ̋XIxèi¶õ”84=nx†¦5šÒý)Ägq KlÈ(•÷ª¦NN£¯—9ÐClÔË‹¸_½]ÔÒ§[åyVËØ„¯×ã4hZå™nÌ7žÀZÆnó¼”é¹Öhx«(z¡¥°çC):¸·Íüíé|ÀÆÌÈþå²Ùýs­˜è›~A#èåÖxÞ‚'£:¬Í)\¶Ÿ>iI$©²ÌÐä9L}GyYOQÕL˜Öîô¼NÊüL„^WŒo:a6Ñ”K Z׋·KQÈŽuËòWjCì¬ý—‚óÝó»Xû¬ÇFf9`Ô9þ;q»NПzr,]T”;§È)¼  «ý¤46R‹kªù8"lÚ“‘Þ«k(‡ÄœO]{½é ™À•7¯Òq U2 z\1Gzÿ>.IÜe'c-W\ö­]kH%äîç•w6·F"FÃ.‚CZ ‰ €»¦‚Ê;bæMñä8ò¯•Eû½¶ïô”iÆÊÆ™8ßËÄñ–\«M®±­*¿K8a±ÐkÛ’ôÕQOu’®ÖI* ˆ“Œ{×á™&$ª[0PÌ‘$íæì.Œ‘-+y.u¤®)Ó½šq{!en9ó`—ÅjŠo“NZ+©UÙ°äúÓF4;좚WÚÙpŸëšQÐx…ö}!qâ%á ¥Þ½>SX±cqËM ¤œµ!)—Ò´†XÄŠ7ØbW Å]ßÐ52”oHS°´w_;¬oú&·îv%Âè|Ñq«´3Q5|þ«Ã‡ÀŽ"œ J.ímë–{®¯û¶ú¸®Ô¨Æ-Ñ»Ò䯪LŠÍ‚l5-åd‹4尶낌̺êFj¢C£Ò„»³¨b?Ì—_ðµ%GÈýܽ¾u•ªnøæUhKê‘íÍÌÿâj>XÏ™ádöÛDédRÂ$óù·¯»\à*]7¹Û,ŠÕ%=iþL¯%§OgFktY±yWê%ÐÕæLõ‰#<8É©FökYä:·ùúWÓ[ íMu,}Ú“Hô2"ƒ—ÍMíxÍyc6™ÕG,¢†ÌúÓ³¸)Ä õÔûçÏ#˜_í¤©ZÞ jÜz©ºZĺ3y!OeÚE`+@¹3M¼‹‰_6ü¨âv_Y.¥„s ¾ñzMZǼ^«}Í¿§qÒ Ç{óZ·^Å#Ô_o"xg+z÷ík¢t;AF2p›gnâ½>^m´ Ý’Nϳèú^©ˆÍ±›£ëw´Ë÷5TÇ;-x½Ì]ÊÍuóE®¥ ¾‡?o¯4±tqÛEíkÁ@:ZK8„*°„ŠeI›Ä|>ÏÆÎ:ŸÆe ãóaÎ¥l*GGQí†u—·DÙÕÐsŒ¹ò»pd³Ø9yÊ`^Çìñ’é=ÛX°ï·'Šäß²›tœ_âz¹OçÃÚ)õIüäÁMCßy³å ô¾G,ÌQUƒÒªQ¥ÖÍ'_“ À¤ÄGa÷Äææ?lÕl<¸n{®OØà÷¾ÒÛP²G>">‡ÏìŠi—/b¨(³w'¶D,h{åI²gþUÄ£m+_½ùÏŽÇ$o.]DÓ˜&´n†©¤?8íX‰?Ôš+¾ÇWÀ å@M÷±•9¿wŸ»Ü냆о`a:ïÊ%«”Æ¿Yø9èÌlWHK7ÅýSvBÕtR£hÎ3ÕNÏ=NƒRç»|©OP8þë·í"³¬¢½K-¾´¶ýp…û¨ŒrâÚZ›v¸5ÿD<“X©Èùœ‰—UÆBÜ®<(±™ÊÔHþŸ!¦`é^§í0o­j´ÑǶ¶ý2ÇË'qOšWew»s”µš•)¯±÷¬˜û¾ùŸfû·HĶâV²_qQ&ê …¥aÚö Ë‚ïG¦Y­oÖ¹>m.|&¨««í ¹ä{â¼F5Ó—¯<$±~û"*¸¢'ÌÒÒ¯S}®CAtxÇÀQÌa”Ù‘*§®[T×°°±–#u "£Sâ!ý¾çÍξù‚›eeåÉ3,Å{¡…—álK÷f+¼É†z¬)ØÉF0ÝìYñI˜:p!¿*z\Õø®·$·ðið}Ê"à$miK{i¥– Í´oK…ÈÕVý‹Rêj‘AûVÖÑ'«1‹÷c6Ù¤F(Ùä2í WéIü*[½fP?QÒòÂk³óý3¾œ$SíÜÞéÞþÑÇÏt?z6±ÍRw¤FÉò æÁoý„x(Å|¸«º©Âô§ +³û:€o‚à€¢ÎüSé=åvO£·ŒT»W[ý°) ¶d‹Ù¾ùw?WÎtk„®C‡@ô¨cQäÓ·…*VÊwؾ°IEƒ1’ :Í ´,‘‰‘ÁL׿gP/®óáJðûˆj$¾I1µé`ßP@ˆ›a!}Õgò3}ÊÄŸ=lž¦Ÿ<’l\&kê·iž*Ï ñÊ´£$2žZàú Ì÷1oq|ïŠ jZ<¢ ö Ö™¸*/‡ëéW£–ßÓP¦ª©4ºmþhÍ/FýH6Štö€xsM랯ÿÂΕ®©•´ÂsêöË{¤w‡/sˆp™ÊzM¡/jtÌÀGúÅ> endobj 33 0 obj << /Type /Font /Subtype /Type1 /BaseFont /UBKXEX+CMBX10 /FontDescriptor 108 0 R /FirstChar 46 /LastChar 116 /Widths 90 0 R >> endobj 6 0 obj << /Type /Font /Subtype /Type1 /BaseFont /UFEQIZ+CMBX12 /FontDescriptor 110 0 R /FirstChar 14 /LastChar 119 /Widths 104 0 R >> endobj 72 0 obj << /Type /Font /Subtype /Type1 /BaseFont /LBWBDW+CMEX10 /FontDescriptor 112 0 R /FirstChar 80 /LastChar 80 /Widths 83 0 R >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XKKIPS+CMMI10 /FontDescriptor 114 0 R /FirstChar 12 /LastChar 122 /Widths 98 0 R >> endobj 71 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RDLHZT+CMMI5 /FontDescriptor 116 0 R /FirstChar 105 /LastChar 106 /Widths 84 0 R >> endobj 52 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KBXNPS+CMMI7 /FontDescriptor 118 0 R /FirstChar 12 /LastChar 116 /Widths 87 0 R >> endobj 7 0 obj << /Type /Font /Subtype /Type1 /BaseFont /DGRFQX+CMR10 /FontDescriptor 120 0 R /FirstChar 11 /LastChar 124 /Widths 103 0 R >> endobj 5 0 obj << /Type /Font /Subtype /Type1 /BaseFont /HADXVU+CMR12 /FontDescriptor 122 0 R /FirstChar 44 /LastChar 121 /Widths 105 0 R >> endobj 4 0 obj << /Type /Font /Subtype /Type1 /BaseFont /EUVBCO+CMR17 /FontDescriptor 124 0 R /FirstChar 14 /LastChar 120 /Widths 106 0 R >> endobj 51 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RBTKQM+CMR7 /FontDescriptor 126 0 R /FirstChar 40 /LastChar 50 /Widths 88 0 R >> endobj 17 0 obj << /Type /Font /Subtype /Type1 /BaseFont /WKVRGH+CMSLTT10 /FontDescriptor 128 0 R /FirstChar 34 /LastChar 126 /Widths 95 0 R >> endobj 69 0 obj << /Type /Font /Subtype /Type1 /BaseFont /ISLVKT+CMSY10 /FontDescriptor 130 0 R /FirstChar 0 /LastChar 50 /Widths 86 0 R >> endobj 70 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TKKBTO+CMSY7 /FontDescriptor 132 0 R /FirstChar 3 /LastChar 50 /Widths 85 0 R >> endobj 15 0 obj << /Type /Font /Subtype /Type1 /BaseFont /DEFZJI+CMTI10 /FontDescriptor 134 0 R /FirstChar 45 /LastChar 122 /Widths 97 0 R >> endobj 16 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TCUVVP+CMTT10 /FontDescriptor 136 0 R /FirstChar 37 /LastChar 126 /Widths 96 0 R >> endobj 37 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KYCZPK+CMTT9 /FontDescriptor 138 0 R /FirstChar 46 /LastChar 57 /Widths 89 0 R >> endobj 10 0 obj << /Type /Pages /Count 6 /Parent 139 0 R /Kids [2 0 R 13 0 R 20 0 R 28 0 R 31 0 R 35 0 R] >> endobj 41 0 obj << /Type /Pages /Count 6 /Parent 139 0 R /Kids [39 0 R 43 0 R 46 0 R 49 0 R 55 0 R 64 0 R] >> endobj 73 0 obj << /Type /Pages /Count 4 /Parent 139 0 R /Kids [67 0 R 75 0 R 78 0 R 81 0 R] >> endobj 139 0 obj << /Type /Pages /Count 16 /Kids [10 0 R 41 0 R 73 0 R] >> endobj 140 0 obj << /Type /Catalog /Pages 139 0 R >> endobj 141 0 obj << /Producer (pdfTeX-1.40.14) /Creator (TeX) /CreationDate (D:20150701163335-05'00') /ModDate (D:20150701163335-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 142 0000000000 65535 f 0000002401 00000 n 0000002296 00000 n 0000000015 00000 n 0000254132 00000 n 0000253991 00000 n 0000253143 00000 n 0000253850 00000 n 0000060349 00000 n 0000253426 00000 n 0000255259 00000 n 0000007555 00000 n 0000005233 00000 n 0000005125 00000 n 0000002523 00000 n 0000254835 00000 n 0000254977 00000 n 0000254412 00000 n 0000057886 00000 n 0000012799 00000 n 0000007447 00000 n 0000005349 00000 n 0000009584 00000 n 0000009727 00000 n 0000009825 00000 n 0000009860 00000 n 0000010102 00000 n 0000015904 00000 n 0000015796 00000 n 0000012917 00000 n 0000019759 00000 n 0000019651 00000 n 0000016029 00000 n 0000253001 00000 n 0000022133 00000 n 0000022025 00000 n 0000019875 00000 n 0000255119 00000 n 0000025719 00000 n 0000025611 00000 n 0000022259 00000 n 0000255368 00000 n 0000028834 00000 n 0000028726 00000 n 0000025821 00000 n 0000031248 00000 n 0000031140 00000 n 0000028937 00000 n 0000034720 00000 n 0000034612 00000 n 0000031340 00000 n 0000254273 00000 n 0000253709 00000 n 0000035962 00000 n 0000041841 00000 n 0000035854 00000 n 0000034857 00000 n 0000038542 00000 n 0000038685 00000 n 0000038769 00000 n 0000038867 00000 n 0000038902 00000 n 0000039144 00000 n 0000045132 00000 n 0000045024 00000 n 0000041960 00000 n 0000048620 00000 n 0000048512 00000 n 0000045247 00000 n 0000254556 00000 n 0000254696 00000 n 0000253567 00000 n 0000253285 00000 n 0000255478 00000 n 0000050980 00000 n 0000050872 00000 n 0000048815 00000 n 0000054145 00000 n 0000054037 00000 n 0000051094 00000 n 0000055736 00000 n 0000055628 00000 n 0000054271 00000 n 0000055815 00000 n 0000055840 00000 n 0000055870 00000 n 0000056190 00000 n 0000056486 00000 n 0000057131 00000 n 0000057213 00000 n 0000057279 00000 n 0000057687 00000 n 0000058131 00000 n 0000058156 00000 n 0000058216 00000 n 0000058250 00000 n 0000058640 00000 n 0000059018 00000 n 0000059486 00000 n 0000060120 00000 n 0000060596 00000 n 0000060622 00000 n 0000060685 00000 n 0000060721 00000 n 0000061353 00000 n 0000061970 00000 n 0000062400 00000 n 0000063057 00000 n 0000073497 00000 n 0000073749 00000 n 0000086833 00000 n 0000087146 00000 n 0000094155 00000 n 0000094385 00000 n 0000106832 00000 n 0000107131 00000 n 0000114581 00000 n 0000114803 00000 n 0000123595 00000 n 0000123830 00000 n 0000148503 00000 n 0000149071 00000 n 0000160011 00000 n 0000160287 00000 n 0000171837 00000 n 0000172101 00000 n 0000179639 00000 n 0000179890 00000 n 0000197249 00000 n 0000197771 00000 n 0000204890 00000 n 0000205124 00000 n 0000212402 00000 n 0000212642 00000 n 0000224009 00000 n 0000224270 00000 n 0000243500 00000 n 0000244015 00000 n 0000252729 00000 n 0000255574 00000 n 0000255649 00000 n 0000255702 00000 n trailer << /Size 142 /Root 140 0 R /Info 141 0 R /ID [ ] >> startxref 255969 %%EOF survival/inst/doc/splines.R0000644000175100001440000000612012545056257015477 0ustar hornikusers### R code from vignette source 'splines.Rnw' ################################################### ### code chunk number 1: splines.Rnw:20-24 ################################################### options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default ################################################### ### code chunk number 2: mplot ################################################### getOption("SweaveHooks")[["fig"]]() require(survival) mfit <- coxph(Surv(futime, death) ~ sex + pspline(age), data=mgus) termplot(mfit, term=2, se=TRUE, col.term=1, col.se=1) ################################################### ### code chunk number 3: mplot2 ################################################### ptemp <- termplot(mfit, se=TRUE, plot=FALSE) attributes(ptemp) ptemp$age[1:4,] ################################################### ### code chunk number 4: mplot3 ################################################### getOption("SweaveHooks")[["fig"]]() center <- with(ptemp$age, y[x==50]) ytemp <- ptemp$age$y + outer(ptemp$age$se, c(0, -1.96, 1.96), '*') matplot(ptemp$age$x, exp(ytemp - center), log='y', type='l', lty=c(1,2,2), col=1, xlab="Age at diagnosis", ylab="Relative death rate") ################################################### ### code chunk number 5: fit1 ################################################### getOption("SweaveHooks")[["fig"]]() options(show.signif.stars=FALSE) # display intelligence fit1 <- coxph(Surv(futime, death) ~ sex + pspline(age, 3), data=flchain) fit1 termplot(fit1, term=2, se=TRUE, col.term=1, col.se=1, ylab="log hazard") ################################################### ### code chunk number 6: fit2 ################################################### agem <- with(flchain, ifelse(sex=="M", age, 60)) agef <- with(flchain, ifelse(sex=="F", age, 60)) fit2 <- coxph(Surv(futime, death) ~ sex + pspline(agef, df=3) + pspline(agem, df=3), data=flchain) anova(fit2, fit1) ################################################### ### code chunk number 7: plot2 ################################################### getOption("SweaveHooks")[["fig"]]() # predictions pterm <- termplot(fit2, term=2:3, se=TRUE, plot=FALSE) # reference refdata <- data.frame(sex=c('F', 'M'), agef=c(65, 60), agem=c(60,65)) pred.ref <- predict(fit2, newdata=refdata, type="lp") # females tempf <- pterm$agef$y + outer(pterm$agef$se, c(0, -1.96, 1.96)) frow <- which(pterm$agef$x == 65) tempf <- tempf - tempf[frow,1] # shift curves # males tempm <- pterm$agem$y + outer(pterm$agem$se, c(0, -1.96, 1.96)) mrow <- which(pterm$agem$x == 65) tempm <- tempm + diff(pred.ref) - tempm[mrow,1] # plot matplot(pterm$agef$x, exp(tempf), log='y', col=1, lty=c(1,2,2), type='l', lwd=c(2,1,1), xlab="Age", ylab="Relative risk of death") matlines(pterm$agem$x, exp(tempm), log='y', col=2, lwd=c(2,1,1), lty=c(1,2,2)) legend(80, 1, c("Female", "Male"), lty=1, lwd=2, col=1:2, bty='n') survival/inst/doc/tests.R0000644000175100001440000003134312545056257015171 0ustar hornikusers### R code from vignette source 'tests.Rnw' ################################################### ### code chunk number 1: tests.Rnw:21-25 ################################################### options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default ################################################### ### code chunk number 2: data ################################################### getOption("SweaveHooks")[["fig"]]() library(survival) library(splines) age2 <- cut(flchain$age, c(49, 59, 69, 79, 89, 120), labels=c("50-59", "60-69", "70-79", "80-89", "90+")) counts <- with(flchain, table(sex, age2)) counts # flchain$flc <- flchain$kappa + flchain$lambda male <- (flchain$sex=='M') mlow <- with(flchain[male,], smooth.spline(age, flc)) flow <- with(flchain[!male,], smooth.spline(age, flc)) plot(flow, type='l', ylim=range(flow$y, mlow$y), xlab="Age", ylab="FLC") lines(mlow, col=2) cellmean <- with(flchain, tapply(flc, list(sex, age2), mean, na.rm=T)) matpoints(c(55,65,75, 85, 95), t(cellmean), pch='fm', col=1:2) round(cellmean, 2) ################################################### ### code chunk number 3: tests.Rnw:333-342 ################################################### us2000 <- rowSums(uspop2[51:101,,'2000']) fit1 <- lm(flc ~ sex, flchain, x=TRUE) fit2 <- lm(flc ~ sex + ns(age,4), flchain, x=TRUE) c(fit1$coef[2], fit2$coef[2]) wt1 <- solve(t(fit1$x)%*%fit1$x, t(fit1$x))[2,] # unadjusted wt2 <- solve(t(fit2$x)%*%fit2$x, t(fit2$x))[2,] # age-adjusted table(wt1, flchain$sex) ################################################### ### code chunk number 4: pop ################################################### getOption("SweaveHooks")[["fig"]]() us2000 <- rowSums(uspop2[51:101,,'2000']) tab0 <- table(flchain$age) tab2 <- tapply(abs(wt2), flchain$age, sum) matplot(50:100, cbind(tab0/sum(tab0), tab2/sum(tab2)), type='l', lty=1, xlab="Age", ylab="Density") us2000 <- rowSums(uspop2[51:101,,'2000']) matpoints(50:100, us2000/sum(us2000), pch='u') legend(60, .02, c("Empirical reference", "LS reference"), lty=1, col=1:2, bty='n') ################################################### ### code chunk number 5: yfit ################################################### yatesfit <- lm(flc ~ interaction(sex, age2) -1, data=flchain) theta <- matrix(coef(yatesfit), nrow=2) dimnames(theta) <- dimnames(counts) round(theta,2) ################################################### ### code chunk number 6: tests.Rnw:497-519 ################################################### qform <- function(beta, var) # quadratic form b' (V-inverse) b sum(beta * solve(var, beta)) contrast <- function(cmat, fit) { varmat <- vcov(fit) if (class(fit) == "lm") sigma2 <- summary(fit)$sigma^2 else sigma2 <- 1 # for the Cox model case beta <- coef(fit) if (!is.matrix(cmat)) cmat <- matrix(cmat, nrow=1) if (ncol(cmat) != length(beta)) stop("wrong dimension for contrast") estimate <- drop(cmat %*% beta) #vector of contrasts ss <- qform(estimate, cmat %*% varmat %*% t(cmat)) *sigma2 list(estimate=estimate, ss=ss, var=drop(cmat %*% varmat %*% t(cmat))) } yates.sex <- matrix(0, 2, 10) yates.sex[1, c(1,3,5,7,9)] <- 1/5 #females yates.sex[2, c(2,4,6,8,10)] <- 1/5 #males contrast(yates.sex, yatesfit)$estimate # the estimated "average" FLC for F/M contrast(yates.sex[2,]-yates.sex[,1], yatesfit) # male - female contrast ################################################### ### code chunk number 7: tests.Rnw:522-548 ################################################### # Create the estimates table -- lots of fits emat <- matrix(0., 6, 3) dimnames(emat) <- list(c("Unadjusted", "MVUE: continuous age", "MVUE: categorical age", "Empirical (data) reference", "US200 reference", "Uniform (Yates)"), c("est", "se", "SS")) #unadjusted emat[1,] <- c(summary(fit1)$coef[2,1:2], anova(fit1)["sex", "Sum Sq"]) # MVUE -- do the two fits fit2 <- lm(flc ~ ns(age,4) + sex, flchain) emat[2,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) fit2 <- lm(flc ~ age2 + sex, flchain) emat[3,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) #Remainder, use contrasts tfun <- function(wt) { cvec <- c(matrix(c(-wt, wt), nrow=2, byrow=TRUE)) temp <- contrast(cvec, yatesfit) c(temp$est, sqrt(temp$var), temp$ss) } emat[4,] <- tfun(colSums(counts)/sum(counts)) usgroup <- tapply(us2000, rep(1:5, c(10,10,10,10,11)), sum)/sum(us2000) emat[5,]<- tfun(usgroup) emat[6,] <- tfun(rep(1/5,5)) ################################################### ### code chunk number 8: tests.Rnw:553-557 ################################################### temp <- dimnames(emat)[[1]] for (i in 1:nrow(emat)) cat(temp[i], sprintf(" &%5.3f", emat[i,1]),sprintf(" &%6.5f", emat[i,2]), sprintf(" & %6.1f", emat[i,3]), "\\\\ \n") ################################################### ### code chunk number 9: weights ################################################### casewt <- array(1, dim=c(2,5,4)) # case weights by sex, age group, estimator csum <- colSums(counts) casewt[,,2] <- counts[2:1,] / rep(csum, each=2) casewt[,,3] <- rep(csum, each=2)/counts casewt[,,4] <- 1/counts #renorm each so that the mean weight is 1 for (i in 1:4) { for (j in 1:2) { meanwt <- sum(casewt[j,,i]*counts[j,])/ sum(counts[j,]) casewt[j,,i] <- casewt[j,,i]/ meanwt } } ################################################### ### code chunk number 10: tests.Rnw:601-611 ################################################### tname <- c("Unadjusted", "Min var", "Empirical", "Yates") for (i in 1:2) { for (j in 1:4) { cat("&",tname[j], " & ", paste(sprintf("%4.2f", casewt[i,,j]), collapse= " & "), "\\\\\n") if (j==1) cat(c("Female", "Male")[i]) } if (i==1) cat("\\hline ") } ################################################### ### code chunk number 11: tests.Rnw:654-658 ################################################### temp <- 1/colSums(1/counts) temp <- temp/sum(temp) cat("Female", sprintf(" & %5.3f", -temp), "\\\\ \n") cat("Male", sprintf(" & %5.3f", temp), "\\\\ \n") ################################################### ### code chunk number 12: treatment ################################################### fit3 <- lm(flc ~ sex * age2, flchain) coef(fit3) contrast(c(0,1, 0,0,0,0, .2,.2,.2,.2), fit3) #Yates ################################################### ### code chunk number 13: SAS ################################################### options(contrasts=c("contr.SAS", "contr.poly")) sfit1 <- lm(flc ~ sex, flchain) sfit2 <- lm(flc ~ sex + age2, flchain) sfit3 <- lm(flc ~ sex * age2, flchain) contrast(c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2), sfit3) # Yates for SAS coding ################################################### ### code chunk number 14: nstt ################################################### options(contrasts = c("contr.treatment", "contr.poly")) #R default fit3a <- lm(flc ~ sex * age2, flchain) options(contrasts = c("contr.SAS", "contr.poly")) fit3b <- lm(flc~ sex * age2, flchain) options(contrasts=c("contr.sum", "contr.poly")) fit3c <- lm(flc ~ sex * age2, flchain) # nstt <- c(0,1, rep(0,8)) #test only the sex coef = the NSTT method temp <- rbind(unlist(contrast(nstt, fit3a)), unlist(contrast(nstt, fit3b)), unlist(contrast(nstt, fit3c)))[,1:2] dimnames(temp) <- list(c("R", "SAS", "sum"), c("effect", "SS")) print(temp) # drop1(fit3a, .~.) ################################################### ### code chunk number 15: anova ################################################### options(show.signif.stars = FALSE) #exhibit intelligence sfit0 <- lm(flc ~ 1, flchain) sfit1b <- lm(flc ~ age2, flchain) anova(sfit0, sfit1b, sfit2, sfit3) ################################################### ### code chunk number 16: relrate ################################################### options(contrasts= c("contr.treatment", "contr.poly")) # R default cfit0 <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) cmean <- matrix(c(0, coef(cfit0)), nrow=2) cmean <- rbind(cmean, cmean[2,] - cmean[1,]) dimnames(cmean) <- list(c("F", "M", "M/F ratio"), dimnames(counts)[[2]]) signif(exp(cmean),3) ################################################### ### code chunk number 17: cox anova ################################################### options(contrasts=c("contr.SAS", "contr.poly")) cfit1 <- coxph(Surv(futime, death) ~ sex, flchain) cfit2 <- coxph(Surv(futime, death) ~ age2 + sex, flchain) cfit3 <- coxph(Surv(futime, death) ~ sex + strata(age2), flchain) # Unadjusted summary(cfit1) # # LRT anova(cfit2) # # Stratified anova(cfit3) summary(cfit3) # # Wald test signif(summary(cfit2)$coefficients, 3) # anova(cfit1, cfit2) ################################################### ### code chunk number 18: coxfit ################################################### wtindx <- with(flchain, tapply(death, list(sex, age2))) cfitpop <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,3])[wtindx]) cfityates <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,4])[wtindx]) # # Glue it into a table for viewing # tfun <- function(fit, indx=1) { c(fit$coef[indx], sqrt(fit$var[indx,indx])) } coxp <- rbind(tfun(cfit1), tfun(cfit2,5), tfun(cfitpop), tfun(cfityates)) dimnames(coxp) <- list(c("Unadjusted", "Additive", "Empirical Population", "Uniform Population"), c("Effect", "se(effect)")) signif(coxp,3) ################################################### ### code chunk number 19: tests.Rnw:1167-1183 ################################################### cfit4 <- coxph(Surv(futime, death) ~ sex * age2, flchain) # Uniform population contrast ysex <- c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2) #Yates for sex, SAS coding contrast(ysex[-1], cfit4) # Verify using cell means coding cfit4b <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) temp <- matrix(c(0, coef(cfit4b)),2) # the female 50-59 is reference diff(rowMeans(temp)) #direct estimate of the Yates # temp2 <- rbind(temp, temp[2,] - temp[1,]) dimnames(temp2) <- list(c('female', 'male', 'difference'), levels(age2)) round(temp2, 3) # # # NSTT contrast contrast(c(1,0,0,0,0,0,0,0,0), cfit4) ################################################### ### code chunk number 20: nstt-lrt ################################################### xmat4 <- model.matrix(cfit4) cfit4b <- coxph(Surv(futime, death) ~ xmat4[,-1], flchain) anova(cfit4b, cfit4) ################################################### ### code chunk number 21: ydata ################################################### data1 <- data.frame(y = rep(1:6, length=20), x1 = factor(letters[rep(1:3, length=20)]), x2 = factor(LETTERS[rep(1:4, length=10)]), x3 = 1:20) data1$x1[19] <- 'c' data1 <- data1[order(data1$x1, data1$x2),] row.names(data1) <- NULL with(data1, table(x1,x2)) # data2 -- single missing cell indx <- with(data1, x1=='a' & x2=='D') data2 <- data1[!indx,] #data3 -- missing the diagonal data3 <- data1[as.numeric(data1$x1) != as.numeric(data1$x2),] ################################################### ### code chunk number 22: tests.Rnw:1359-1362 ################################################### options(contrasts=c("contr.sum", "contr.poly")) fit1 <- lm(y ~ x1*x2, data1) drop1(fit1, .~.) ################################################### ### code chunk number 23: tests.Rnw:1369-1375 ################################################### options(contrasts=c("contr.SAS", "contr.poly")) fit2 <- lm(y ~ x1*x2, data1) drop1(fit2, .~.) options(contrasts=c("contr.treatment", "contr.poly")) fit3 <- lm(y ~ x1*x2, data1) drop1(fit3, .~.) ################################################### ### code chunk number 24: att ################################################### X <- model.matrix(fit2) ux <- unique(X) ux indx <- rep(1:3, c(4,4,4)) effects <- t(rowsum(ux, indx)/4) # turn sideways to fit the paper better effects yates <- effects[,-1] - effects[,1] yates ################################################### ### code chunk number 25: tests.Rnw:1415-1418 ################################################### wt <- solve(t(X) %*% X, t(X)) # twelve rows (one per coef), n columns casewt <- t(effects) %*% wt # case weights for the three "row efffects" for (i in 1:3) print(tapply(casewt[i,], data1$x2, sum)) ################################################### ### code chunk number 26: tests.Rnw:1455-1456 ################################################### fit4 <- lm(y ~ x1*x2 + x3, data=data1) survival/tests/0000755000175100001440000000000012545056257013320 5ustar hornikuserssurvival/tests/singtest.Rout.save0000644000175100001440000000366712055204303016763 0ustar hornikusers R version 2.15.2 (2012-10-26) -- "Trick or Treat" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # A simple test of an overdetermined system > # Should give a set of NA coefficients > # > test1 <- data.frame(time= c(4, 3,1,1,2,2,3), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > temp <- rep(0:3, rep(7,4)) > > stest <- data.frame(start = 10*temp, + stop = 10*temp + test1$time, + status = rep(test1$status,4), + x = c(test1$x+ 1:7, rep(test1$x,3)), + epoch = rep(1:4, rep(7,4))) > > # Will create a warning about a singular X matrix > fit1 <- coxph(Surv(start, stop, status) ~ x * factor(epoch), stest) Warning message: In coxph(Surv(start, stop, status) ~ x * factor(epoch), stest) : X matrix deemed to be singular; variable 2 3 4 > fit1$coef # elements 2:4 should be NA x factor(epoch)2 factor(epoch)3 factor(epoch)4 0.1041579 NA NA NA x:factor(epoch)2 x:factor(epoch)3 x:factor(epoch)4 1.5726996 1.5726996 1.5726996 > all.equal(is.na(fit1$coef), c(F,T,T,T,F,F,F), check.attributes=FALSE) [1] TRUE > > proc.time() user system elapsed 0.168 0.032 0.193 survival/tests/anova.R0000644000175100001440000000204212164374110014531 0ustar hornikusers# # Test out anova, with strata terms # options(na.action=na.omit) library(survival) fit1 <- coxph(Surv(time, status) ~ ph.ecog + wt.loss + strata(sex) + poly(age,3), lung) ztemp <- anova(fit1) tdata <- na.omit(lung[, c('time', 'status', 'ph.ecog', 'wt.loss', 'sex', 'age')]) fit2 <- coxph(Surv(time, status)~ ph.ecog + wt.loss + poly(age,3) + strata(sex), data=tdata) ztemp2 <- anova(fit2) all.equal(ztemp, ztemp2) fit2 <- coxph(Surv(time, status) ~ ph.ecog + wt.loss + strata(sex), tdata) fit3 <- coxph(Surv(time, status) ~ ph.ecog + strata(sex), tdata) all.equal(ztemp$loglik, c(fit1$loglik[1], fit3$loglik[2], fit2$loglik[2], fit1$loglik[2])) all.equal(ztemp$Chisq[-1], 2* diff(ztemp$loglik)) all.equal(ztemp$Df[-1], c(1,1,3)) ztemp2 <- anova(fit3, fit2, fit1) all.equal(ztemp2$loglik, ztemp$loglik[-1]) all.equal(ztemp2$Chisq[2:3], ztemp$Chisq[3:4]) # Change from ztemp2$P; it's a data frame and in R 3.0.2 abbreviated names # give a warning all.equal(ztemp2[[4]][2:3], ztemp[[4]][3:4]) survival/tests/model.matrix.Rout.save0000644000175100001440000000611012466142446017527 0ustar hornikusers R Under development (unstable) (2014-09-01 r66509) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Test out the revised model.matrix code > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0), + z= factor(c('a', 'a', 'b', 'b', 'c', 'c', 'a'))) > > fit1 <- coxph(Surv(time, status) ~ z, test1, iter=1) > fit2 <- coxph(Surv(time, status) ~z, test1, x=T, iter=1) > all.equal(model.matrix(fit1), fit2$x) [1] TRUE > > # This has no level 'b', make sure dummies recode properly > test2 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0), + z= factor(c('a', 'a', 'a', 'a', 'c', 'c', 'a'))) > > ftest <- model.frame(fit1, data=test2) > all.equal(levels(ftest$z), levels(test1$z)) [1] TRUE > > # xtest will have one more row than the others, since it does not delete > # the observation with a missing value for status > xtest <- model.matrix(fit1, data=test2) > dummy <- fit2$x > dummy[,1] <- 0 > all.equal(xtest[-2,], dummy, check.attributes=FALSE) [1] TRUE > > # The case of a strata by factor interaction > # Use iter=0 since there are too many covariates and it won't converge > test1$x2 <- factor(rep(1:2, length=7)) > fit3 <- coxph(Surv(time, status) ~ strata(x2)*z, test1, iter=0) > xx <- model.matrix(fit3) > all.equal(attr(xx, "assign"), c(2,2,3,3)) [1] TRUE > all.equal(colnames(xx), c("zb", "zc", "strata(x2)2:zb", + "strata(x2)2:zc")) [1] TRUE > all.equal(attr(xx, "contrasts"), + list("strata(x2)"= "contr.treatment", z="contr.treatment")) [1] TRUE > > fit3b <- coxph(Surv(time, status) ~ strata(x2)*z, test1, iter=0, x=TRUE) > all.equal(fit3b$x, xx) [1] TRUE > > > # A model with a tt term > fit4 <- coxph(Surv(time, status) ~ tt(x) + x, test1, iter=0, + tt = function(x, t, ...) x*t) > ff <- model.frame(fit4) > # There is 1 subject in the final risk set, 4 at risk at time 6, 6 at time 1 > # The .strata. variable numbers from last time point to first > all.equal(ff$.strata., rep(1:3, c(1, 4,6))) [1] TRUE > all.equal(ff[["tt(x)"]], ff$x* c(9,6,1)[ff$.strata.]) [1] TRUE > > xx <- model.matrix(fit4) > all.equal(xx[,1], ff[[2]], check.attributes=FALSE) [1] TRUE > > > proc.time() user system elapsed 0.196 0.028 0.220 survival/tests/expected.Rout.save0000644000175100001440000002611012257335007016722 0ustar hornikusers R version 3.0.1 (2013-05-16) -- "Good Sport" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # Tests of expected survival > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > # > # This makes several scripts easier > # Certain tests depended in the now-depreciated date library > {if (is.R()) mdy.date <- function(m, d, y) { + y <- ifelse(y<100, y+1900, y) + as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") + } + else mdy.date <- function(m,d,y) { + y <- ifelse(y<100, y+1900, y) + timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") + } + } > > # This function takes a single subject and walks down the rate table > # Input: the vector of starting points, futime, and a ratetable > # Output: the full history of walking through said table. Let n= #unique > # rates that were used > # cell = n by #dims of the table: index of the table cell > # days = time spent in cell > # hazard= accumulated hazard = days * rate > # This does not do date or factor conversions -- start has to be numeric > # > ratewalk <- function(start, futime, ratetable=survexp.us) { + if (!is.ratetable(ratetable)) stop("Bad rate table") + ratedim <- dim(ratetable) + nvar <- length(ratedim) + if (length(start) != nvar) stop("Wrong length for start") + if (futime <=0) stop("Invalid futime") + + attR <- attributes(ratetable) + discrete <- (attR$type ==1) #discrete categories + + maxn <- sum(!discrete)*prod(ratedim[!discrete]) #most cells you can hit + cell <- matrix(0, nrow=maxn, ncol=nvar) + days <- hazard <- double(maxn) + + eps <- 1e-8 #Avoid round off error + n <- 0 + while (futime >0) { + n <- n+1 + #what cell am I in? + # Note that at the edges of the rate table, we use the edge: if + # it only goes up the the year 2000, year 2000 is used for any + # dates beyond. This effectively eliminates one boundary + cell[n,discrete] <- start[discrete] + edge <- futime #time to nearest edge, or finish + for (j in which(!discrete)) { + indx <- sum(start[j] >= attR$cutpoints[[j]]-eps) + cell[n, j] <- max(1, indx) + if (indx < ratedim[j]) + edge <- min(edge, (attR$cutpoints[[j]])[indx+1] - start[j]) + } + days[n] <- edge #this many days in the cell + # using a matrix as a subscript is so handy sometimes + hazard[n] <- edge * (as.matrix(ratetable))[cell[n,,drop=F]] + futime <- futime - edge #amount of time yet to account for + start[!discrete] <- start[!discrete] + edge #walk forward in time + } + list(cell=cell[1:n,], days=days[1:n], hazard=hazard[1:n]) + } > > # Simple test of ratewalk: 20 years old, start on 7Sep 1960 (day 250) > # 116 days at the 1960, 20 year old male rate, through the end of the day > # on 12/31/1960, then 84 days at the 1961 rate. > # The decennial q for 1960 males is .00169. > zz <- ratewalk(c(20.4*365.25, 1, 250), 200) > all.equal(zz$hazard[1], -(116/365.25)*log(1-.00169)) [1] TRUE > all.equal(zz$days, c(116,84)) [1] TRUE > > > # > # Simple case 1: a single male subject, born 1/1/36 and entered on study 1/2/55 > # > # Compute the 1, 5, 10 and 12 year expected survival > > temp1 <- mdy.date(1,1,36) > temp2 <- mdy.date(1,2,55) > exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=1, race='white'), + ratetable=survexp.usr,times=c(366, 1827, 3653, 4383)) > > tyear <- as.numeric(temp2 - mdy.date(1,1,1960)) > h1 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 366, survexp.usr) > h2 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 1827, survexp.usr) > h3 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 3653, survexp.usr) > h4 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 4383, survexp.usr) > > aeq(-log(exp1$surv), c(sum(h1$hazard), sum(h2$hazard), sum(h3$hazard), + sum(h4$hazard))) [1] TRUE > > > # Just a little harder: > # Born 3/1/25 and entered the study on 6/10/55. The code creates shifted > # dates to align with US rate tables - entry is 59 days earlier (days from > # 1/1/25 to 3/1/25). > # > temp1 <- mdy.date(3,1,25) > temp2 <- mdy.date(6,10,55) > exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=2, race='black'), + ratetable=survexp.usr,times=c(366, 1827, 3653, 4383)) > > tyear <- as.numeric(temp2 - mdy.date(1,1,1960)) - 59 > h1 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 366, survexp.usr) > h2 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 1827, survexp.usr) > h3 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 3653, survexp.usr) > h4 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 4383, survexp.usr) > > aeq(-log(exp1$surv), c(sum(h1$hazard), sum(h2$hazard), sum(h3$hazard), + sum(h4$hazard))) [1] TRUE > > # > # Simple case 2: make sure that the averages are correct, for Ederer method > # > # Compute the 1, 5, 10 and 12 year expected survival > > temp1 <- mdy.date(1:6,6:11,1890:1895) > temp2 <- mdy.date(6:1,11:6,c(55:50)) > temp3 <- c(1,2,1,2,1,2) > age <- temp2 - temp1 > > exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), + times=c(366, 1827, 3653, 4383)) > exp2 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3) + I(1:6), + times=c(366, 1827, 3653, 4383)) > exp3 <- exp2$surv > for (i in 1:length(temp1)){ + exp3[,i] <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), + times=c(366, 1827, 3653, 4383), subset=i)$surv + } > > > print(aeq(exp2$surv, exp3)) [1] TRUE > print(all.equal(exp1$surv, apply(exp2$surv, 1, mean))) [1] TRUE > > # They agree, but are they right? > # > for (i in 1:length(temp1)) { + offset <- as.numeric(temp1[i] - mdy.date(1,1, 1889+i)) + tyear = (as.numeric(temp2[i] - mdy.date(1,1,1960))) - offset + haz1 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 366) + haz2 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 1827) + haz3 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 3653) + haz4 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 4383) + print(aeq(-log(exp2$surv[,i]), c(sum(haz1$hazard), sum(haz2$hazard), + sum(haz3$hazard), sum(haz4$hazard)))) + } [1] TRUE [1] TRUE [1] TRUE [1] TRUE [1] TRUE [1] TRUE > > # > # Check that adding more time points doesn't change things > # > exp4 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3) + I(1:6), + times=sort(c(366, 1827, 3653, 4383, 30*(1:100)))) > aeq(exp4$surv[match(exp2$time, exp4$time),], exp2$surv) [1] TRUE > > exp4 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), + times=sort(c(366, 1827, 3653, 4383, 30*(1:100)))) > aeq(exp1$surv, exp4$surv[match(exp1$time, exp4$time, nomatch=0)]) [1] TRUE > > > # > # Now test Hakulinen's method, assuming an analysis date of 3/1/57 > # > futime <- mdy.date(3,1,57) - temp2 > xtime <- sort(c(futime, 30, 60, 185, 365)) > > exp1 <- survexp(futime ~ ratetable(year=temp2, age=(temp2-temp1), sex=1), + times=xtime, conditional=F) > exp2 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=1) + I(1:6), + times=futime) > > wt <- rep(1,6) > con <- double(6) > for (i in 1:6) { + con[i] <- sum(exp2$surv[i,i:6])/sum(wt[i:6]) + wt <- exp2$surv[i,] + } > > exp1$surv[match(futime, xtime)] [1] 0.9557362 0.9285840 0.9025661 0.8774220 0.8532489 0.8297416 > aeq(exp1$surv[match(futime, xtime)], cumprod(con)) [1] TRUE > > > # > # Now for the conditional method > # > exp1 <- survexp(futime ~ ratetable(year=temp2, age=(temp2-temp1), sex=1), + times=xtime, conditional=T) > > cond <- exp2$surv > for (i in 6:2) cond[i,] <- (cond[i,]/cond[i-1,]) #conditional survival > for (i in 1:6) con[i] <- exp(mean(log(cond[i, i:6]))) > > all.equal(exp1$surv[match(futime, xtime)], cumprod(con)) [1] TRUE > cumprod(con) [1] 0.9556656 0.9284398 0.9023612 0.8771798 0.8529944 0.8294940 > > # > # Test out expected survival, when the parent pop is another Cox model > # > test1 <- data.frame(time= c(4, 3,1,1,2,2,3), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > fit <- coxph(Surv(time, status) ~x, test1, method='breslow') > > dummy <- data.frame(time=c(.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5), + status=c(1,0,1,0,1,0,1,1,1), x=(-4:4)/2) > > efit <- survexp(time ~ ratetable(x=x), dummy, ratetable=fit, cohort=F) > > # > # Now, compare to the true answer, which is known to us > # > ss <- exp(fit$coef) > haz <- c( 1/(3*ss+3), 2/(ss+3), 1) #truth at time 0,1,2,4+ > chaz <- cumsum(c(0,haz)) > chaz2 <- chaz[c(1,2,2,3,3,3,3,4,4)] > > risk <- exp(fit$coef*dummy$x) > efit2 <- exp(-risk*chaz2) > > all.equal(as.vector(efit), as.vector(efit2)) #ignore mismatched name attrib [1] TRUE > > # > # Now test the direct-adjusted curve (Ederer) > # > efit <- survexp( ~ ratetable(x=x), dummy, ratetable=fit, se=F) > direct <- survfit(fit, newdata=dummy, censor=FALSE)$surv > > chaz <- chaz[-1] #drop time 0 > d2 <- exp(outer(-chaz, risk)) > all.equal(as.vector(direct), as.vector(d2)) #this tests survfit [1] TRUE > > all.equal(as.vector(efit$surv), as.vector(apply(direct,1,mean))) #direct [1] TRUE > > # Check out the "times" arg of survexp > efit2 <- survexp( ~ ratetable(x=x), dummy, ratetable=fit, se=F, + times=c(.5, 2, 3.5,6)) > aeq(efit2$surv, c(1, efit$surv[c(2,2,3)])) [1] TRUE > > # > # Now test out the Hakulinen method (Bonsel's method) > # By construction, we have a large correlation between x and censoring > # > # In theory, hak1 and hak2 would be the same. In practice, like a KM and > # F-H, they differ when n is small. > # > efit <- survexp( time ~ ratetable(x=x), dummy, ratetable=fit, se=F) > > surv <- wt <- rep(1,9) > tt <- c(1,2,4) > hak1 <- hak2 <- NULL > for (i in 1:3) { + wt[dummy$time < tt[i]] <- 0 + hak1 <- c(hak1, exp(-sum(haz[i]*risk*surv*wt)/sum(surv*wt))) + hak2 <- c(hak2, sum(exp(-haz[i]*risk)*surv*wt)/sum(surv*wt)) + surv <- surv * exp(-haz[i]*risk) + } > > all.equal(as.vector(efit$surv), as.vector(cumprod(hak1))) [1] TRUE > > # > # Now do the conditional estimate > # > efit <- survexp( time ~ ratetable(x=x), dummy, ratetable=fit, se=F, + conditional=T) > wt <- rep(1,9) > cond <- NULL > for (i in 1:3) { + wt[dummy$time < tt[i]] <- 0 + cond <- c(cond, exp(-sum(haz[i]*risk*wt)/sum(wt))) + } > > all.equal(as.vector(efit$surv), as.vector(cumprod(cond))) [1] TRUE > > proc.time() user system elapsed 0.692 0.076 0.766 survival/tests/factor2.Rout.save0000644000175100001440000000345111732700061016455 0ustar hornikusers R version 2.14.0 (2011-10-31) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > options(na.action=na.exclude) > # > # More tests of factors in prediction, using a new data set > # > fit <- coxph(Surv(time, status) ~ factor(ph.ecog), lung) > > tdata <- data.frame(ph.ecog = factor(0:3)) > p1 <- predict(fit, newdata=tdata, type='lp') > p2 <- predict(fit, type='lp') > aeq(p1, p2[match(0:3, lung$ph.ecog)]) [1] TRUE > > fit2 <- coxph(Surv(time, status) ~ factor(ph.ecog) + factor(sex), lung) > tdata <- expand.grid(ph.ecog = factor(0:3), sex=factor(1:2)) > p1 <- predict(fit2, newdata=tdata, type='risk') > > xdata <- expand.grid(ph.ecog=factor(1:3), sex=factor(1:2)) > p2 <- predict(fit2, newdata=xdata, type='risk') > all.equal(p2, p1[c(2:4, 6:8)], check.attributes=FALSE) [1] TRUE > > > fit3 <- survreg(Surv(time, status) ~ factor(ph.ecog) + age, lung) > tdata <- data.frame(ph.ecog=factor(0:3), age=50) > predict(fit, type='lp', newdata=tdata) 1 2 3 4 -0.39518177 -0.02634168 0.52120527 1.81279848 > predict(fit3, type='lp', newdata=tdata) 1 2 3 4 6.399571 6.142938 5.770523 4.916993 > survival/tests/strata2.Rout.save0000644000175100001440000000334412055204344016500 0ustar hornikusers R version 2.15.2 (2012-10-26) -- "Trick or Treat" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > # New tests 4/2010 to validate strata by covariate interactions > # > library(survival) Loading required package: splines > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > tdata <- lung > tdata$sex <- lung$sex +3 > > # Both of these should produce warning messages about singular X, since there > # are ph.ecog=3 subjects in only 1 of the strata. > # Does not affect the test > fit1 <- coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), lung) Warning message: In coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), lung) : X matrix deemed to be singular; variable 5 > fit2 <- coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), tdata) Warning message: In coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), tdata) : X matrix deemed to be singular; variable 5 > > aeq(fit1$coef, fit2$coef) [1] TRUE > aeq(fit1$var, fit2$var) [1] TRUE > aeq(predict(fit1), predict(fit2)) [1] TRUE > > proc.time() user system elapsed 0.200 0.028 0.224 survival/tests/r_capacitor.R0000644000175100001440000000104111732700061015707 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) capacitor <- read.table('data.capacitor', row.names=1, col.names=c('', 'days', 'event', 'voltage')) fitig <- survreg(Surv(days, event)~voltage, dist = "gaussian", data = capacitor) summary(fitig) fitix <- survreg(Surv(days, event)~voltage, dist = "extreme", data = capacitor) summary(fitix) fitil <- survreg(Surv(days, event)~voltage, dist = "logistic", data = capacitor) summary(fitil) survival/tests/fr_cancer.Rout.save0000644000175100001440000001074212536400614017044 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Here is a test case with multiple smoothing terms > # > > fit0 <- coxph(Surv(time, status) ~ ph.ecog + age, lung) > fit1 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,3), lung) > fit2 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,4), lung) > fit3 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,8), lung) > > > > fit4 <- coxph(Surv(time, status) ~ ph.ecog + pspline(wt.loss,3), lung) > > fit5 <-coxph(Surv(time, status) ~ ph.ecog + pspline(age,3) + + pspline(wt.loss,3), lung) > > fit1 Call: coxph(formula = Surv(time, status) ~ ph.ecog + pspline(age, 3), data = lung) coef se(coef) se2 Chisq DF p ph.ecog 0.44802 0.11707 0.11678 14.64453 1.00 0.00013 pspline(age, 3), linear 0.01126 0.00928 0.00928 1.47231 1.00 0.22498 pspline(age, 3), nonlin 2.07924 2.08 0.37143 Iterations: 4 outer, 12 Newton-Raphson Theta= 0.861 Degrees of freedom for terms= 1.0 3.1 Likelihood ratio test=21.9 on 4.08 df, p=0.000227 n=227 (1 observation deleted due to missingness) > fit2 Call: coxph(formula = Surv(time, status) ~ ph.ecog + pspline(age, 4), data = lung) coef se(coef) se2 Chisq DF p ph.ecog 0.45047 0.11766 0.11723 14.65751 1.00 0.00013 pspline(age, 4), linear 0.01117 0.00927 0.00927 1.45195 1.00 0.22822 pspline(age, 4), nonlin 2.95816 3.08 0.41197 Iterations: 4 outer, 11 Newton-Raphson Theta= 0.797 Degrees of freedom for terms= 1.0 4.1 Likelihood ratio test=22.7 on 5.07 df, p=0.000412 n=227 (1 observation deleted due to missingness) > fit3 Call: coxph(formula = Surv(time, status) ~ ph.ecog + pspline(age, 8), data = lung) coef se(coef) se2 Chisq DF p ph.ecog 0.47640 0.12024 0.11925 15.69732 1.00 7.4e-05 pspline(age, 8), linear 0.01172 0.00923 0.00923 1.61161 1.00 0.20 pspline(age, 8), nonlin 6.93188 6.99 0.43 Iterations: 5 outer, 15 Newton-Raphson Theta= 0.691 Degrees of freedom for terms= 1 8 Likelihood ratio test=27.6 on 8.97 df, p=0.00108 n=227 (1 observation deleted due to missingness) > fit4 Call: coxph(formula = Surv(time, status) ~ ph.ecog + pspline(wt.loss, 3), data = lung) coef se(coef) se2 Chisq DF p ph.ecog 0.51545 0.12960 0.12737 15.81939 1.00 7e-05 pspline(wt.loss, 3), line -0.00702 0.00655 0.00655 1.14638 1.00 0.28 pspline(wt.loss, 3), nonl 2.44612 2.09 0.31 Iterations: 3 outer, 10 Newton-Raphson Theta= 0.776 Degrees of freedom for terms= 1.0 3.1 Likelihood ratio test=21.1 on 4.06 df, p=0.000326 n=213 (15 observations deleted due to missingness) > fit5 Call: coxph(formula = Surv(time, status) ~ ph.ecog + pspline(age, 3) + pspline(wt.loss, 3), data = lung) coef se(coef) se2 Chisq DF p ph.ecog 0.47422 0.13495 0.13206 12.34842 1.00 0.00044 pspline(age, 3), linear 0.01368 0.00976 0.00974 1.96406 1.00 0.16108 pspline(age, 3), nonlin 1.90116 2.07 0.40284 pspline(wt.loss, 3), line -0.00717 0.00661 0.00660 1.17529 1.00 0.27832 pspline(wt.loss, 3), nonl 2.07729 2.03 0.35929 Iterations: 4 outer, 12 Newton-Raphson Theta= 0.85 Theta= 0.779 Degrees of freedom for terms= 1.0 3.1 3.0 Likelihood ratio test=25.2 on 7.06 df, p=0.000726 n=213 (15 observations deleted due to missingness) > > rm(fit1, fit2, fit3, fit4, fit5) > > proc.time() user system elapsed 0.276 0.024 0.294 survival/tests/data.rat20000644000175100001440000001001311732700061015000 0ustar hornikusers1 1 1 60 182 1 2 1 1 60 182 0 3 1 1 60 63 1 3 1 2 63 68 1 3 1 3 68 182 0 4 1 1 60 152 1 4 1 2 152 182 0 5 1 1 60 130 1 5 1 2 130 134 1 5 1 3 134 145 1 6 1 1 60 98 1 6 1 2 98 152 1 6 1 1 60 98 1 6 1 2 98 152 1 6 1 3 152 182 1 7 1 1 60 88 1 7 1 2 88 95 1 7 1 3 95 105 1 7 1 4 105 130 1 7 1 5 130 137 1 7 1 6 137 167 1 7 1 7 167 182 0 8 1 1 60 152 1 8 1 2 152 182 0 9 1 1 60 81 1 9 1 2 81 182 0 10 1 1 60 71 1 10 1 2 71 84 1 10 1 3 84 126 1 10 1 4 126 134 1 10 1 5 134 152 1 10 1 6 152 182 0 11 1 1 60 116 1 11 1 2 116 130 1 11 1 3 130 182 0 12 1 1 60 91 1 12 1 2 91 182 0 13 1 1 60 63 1 13 1 2 63 68 1 13 1 3 68 84 1 13 1 4 84 95 1 13 1 5 95 152 1 13 1 6 152 182 0 14 1 1 60 105 1 14 1 2 103 152 1 14 1 3 152 182 0 15 1 1 60 63 1 15 1 2 63 102 1 15 1 3 102 152 1 15 1 4 152 182 0 16 1 1 60 63 1 16 1 2 63 77 1 16 1 3 77 112 1 16 1 4 112 140 1 16 1 5 140 182 0 17 1 1 60 77 1 17 1 2 77 119 1 17 1 3 119 152 1 17 1 4 152 161 1 17 1 5 161 167 1 17 1 6 167 182 0 18 1 1 60 105 1 18 1 2 105 112 1 18 1 3 112 145 1 18 1 4 145 161 1 18 1 5 161 182 1 19 1 1 60 152 1 19 1 2 152 182 1 20 1 1 60 81 1 20 1 2 81 95 1 20 1 3 95 182 0 21 1 1 60 84 1 21 1 2 84 91 1 21 1 3 91 102 1 21 1 4 102 108 1 21 1 5 108 130 1 21 1 6 130 134 1 21 1 7 134 182 0 22 1 1 60 182 0 23 1 1 60 91 1 23 1 2 91 182 0 24 0 1 60 63 1 24 0 2 63 102 1 24 0 3 102 119 1 24 0 4 119 161 1 24 0 5 161 161 1 24 0 6 161 172 1 24 0 7 172 179 1 24 0 8 179 182 0 25 0 1 60 88 1 25 0 2 88 91 1 25 0 3 91 95 1 25 0 4 95 105 1 25 0 5 105 112 1 25 0 6 112 119 1 25 0 7 119 119 1 25 0 8 119 137 1 25 0 9 137 145 1 25 0 10 145 167 1 25 0 11 167 172 1 25 0 12 172 182 0 26 0 1 60 91 1 26 0 2 91 98 1 26 0 3 98 108 1 26 0 4 108 112 1 26 0 5 112 134 1 26 0 6 134 137 1 26 0 7 137 161 1 26 0 8 161 161 1 26 0 9 161 179 1 26 0 10 179 182 0 27 0 1 60 71 1 27 0 2 71 174 1 27 0 2 174 182 0 28 0 1 60 95 1 28 0 2 95 105 1 28 0 3 105 134 1 28 0 4 134 137 1 28 0 5 137 140 1 28 0 6 140 145 1 28 0 7 145 150 1 28 0 8 150 150 1 28 0 9 150 182 0 29 0 1 60 66 1 29 0 2 66 68 1 29 0 3 68 130 1 29 0 4 130 137 1 29 0 5 137 182 0 30 0 1 60 77 1 30 0 2 77 85 1 30 0 3 85 112 1 30 0 4 112 137 1 30 0 5 137 161 1 30 0 6 161 174 1 30 0 7 174 182 0 31 0 1 60 81 1 31 0 2 81 84 1 31 0 3 84 126 1 31 0 4 125 134 1 31 0 5 134 161 1 31 0 6 161 161 1 31 0 7 161 174 1 31 0 8 174 182 0 32 0 1 60 68 1 32 0 2 68 77 1 32 0 3 77 98 1 32 0 4 98 102 1 32 0 5 102 102 1 32 0 6 102 102 1 32 0 7 102 182 0 33 0 1 60 112 1 33 0 2 112 182 0 34 0 1 60 88 1 34 0 2 88 88 1 34 0 3 88 91 1 34 0 4 91 98 1 34 0 5 98 112 1 34 0 6 112 134 1 34 0 7 134 134 1 34 0 8 134 137 1 34 0 9 137 137 1 34 0 10 137 140 1 34 0 11 140 140 1 34 0 12 140 152 1 34 0 13 152 152 1 34 0 14 152 182 0 35 0 1 60 77 1 35 0 2 77 179 1 35 0 3 179 182 0 36 0 1 60 112 1 36 0 2 112 182 0 37 0 1 60 71 1 37 0 2 71 71 1 37 0 3 71 74 1 37 0 4 74 77 1 37 0 5 77 112 1 37 0 6 112 116 1 37 0 7 116 116 1 37 0 8 116 140 1 37 0 9 140 140 1 37 0 10 140 167 1 37 0 11 167 182 0 38 0 1 60 77 1 38 0 2 77 95 1 38 0 3 95 126 1 38 0 4 126 150 1 38 0 5 150 182 0 39 0 1 60 88 1 39 0 2 88 126 1 39 0 3 126 130 1 39 0 4 130 130 1 39 0 5 130 134 1 39 0 6 134 182 0 40 0 1 60 63 1 40 0 2 63 74 1 40 0 3 74 84 1 40 0 4 84 84 1 40 0 5 84 88 1 40 0 6 88 91 1 40 0 7 91 95 1 40 0 8 95 108 1 40 0 9 108 134 1 40 0 10 134 137 1 40 0 11 137 179 1 40 0 12 179 182 0 41 0 1 60 81 1 41 0 2 81 88 1 41 0 3 88 105 1 41 0 4 105 116 1 41 0 5 116 123 1 41 0 6 123 140 1 41 0 7 140 145 1 41 0 8 145 152 1 41 0 9 152 161 1 41 0 10 161 161 1 41 0 11 161 179 1 41 0 12 179 182 0 42 0 1 60 88 1 42 0 2 88 95 1 42 0 3 95 112 1 42 0 4 112 119 1 42 0 5 119 126 1 42 0 6 126 126 1 42 0 7 126 150 1 42 0 8 150 157 1 42 0 9 157 179 1 42 0 10 179 182 0 43 0 1 60 68 1 43 0 2 68 68 1 43 0 3 68 84 1 43 0 4 84 102 1 43 0 5 102 105 1 43 0 6 105 119 1 43 0 7 119 123 1 43 0 8 123 123 1 43 0 9 123 137 1 43 0 10 137 161 1 43 0 11 161 179 1 43 0 12 179 182 0 44 0 1 60 140 1 44 0 2 140 182 0 45 0 1 60 152 1 45 0 2 152 182 1 45 0 3 182 182 1 46 0 1 60 81 1 46 0 2 81 182 0 47 0 1 60 63 1 47 0 2 63 88 1 47 0 3 88 134 1 47 0 4 134 182 0 48 0 1 60 84 1 48 0 2 84 134 1 48 0 3 134 182 1 survival/tests/quantile.R0000644000175100001440000001040512466142446015263 0ustar hornikusers# # Formal test of the quantile routine for survfit library(survival) aeq <- function(x, y, ...) all.equal(as.vector(x), as.vector(y), ...) # There are 8 cases: strata Y/N, ncol(surv) >1, conf.int = T/F # Subcase: the quantile exactly agrees with a horizontal segment of # the curve or not. # First do the 4 cases where fit$surv is a vector # test1 <- data.frame(time= c(9, 3,1,1,6,6,8, 10), status=c(1,NA,1,0,1,1,0, 0), x= c(0, 2,1,1,1,0,0, 0)) # True survival = (6/7) * (3/5) * (1/2) for overall # The q's are chosen to include a point < first jump, mid, after last jump, # and exact intersections with the "flats" of the curve. # qq <- c(13/14, 6/7, 2/3, .5, 9/35, .1) # Nothing on the right hand side, simple survival (no strata) fit1 <- survfit(Surv(time, status) ~ 1, test1, conf.type='none') aeq(quantile(fit1, 1-qq), c(1, 3.5, 6, 9, 9.5, NA)) #without conf.int fit2 <- survfit(Surv(time, status) ~ 1, test1) #with conf.int aeq(quantile(fit2, 1-qq), list(quantile = c(1, 3.5, 6, 9, 9.5, NA), lower = c(1,1,1,6,6,9), upper = rep(as.numeric(NA), 6)), check.attributes=FALSE) aeq(quantile(fit2, 1-qq, FALSE), c(1, 3.5, 6, 9, 9.5, NA)) # Now a variable on the right (strata in the result) # curve 0: (t=6, S=3/4), (t=9, S=3/8) # curve 1: (t=1, S=2/3), (t=6, S= 0) fit1 <- survfit(Surv(time, status) ~ x, test1, conf.type='none') aeq(quantile(fit1, 1-qq), matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) fit2 <- survfit(Surv(time, status) ~ x, test1) aeq(quantile(fit2, 1-qq, FALSE), matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) temp <- quantile(fit2, 1-qq) aeq(temp$quantile, matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) aeq(temp$lower, matrix(c(6,6,6,6,9,9, 1,1,1,1, NA,NA), nrow=2, byrow=T)) aeq(temp$upper, rep(as.numeric(NA), 12)) # Second major case set -- a survfit object where fit$surv is a matrix # This arises from coxph models # There is only 1 subject with ph.ecog=3 which is a nice edge case cfit <- coxph(Surv(time, status) ~ age + strata(ph.ecog), lung) sfit <- survfit(cfit, newdata=data.frame(age=c(50, 70))) qtot <- quantile(sfit, qq) for (i in 1:4) { for (j in 1:2) { temp <- quantile(sfit[i,j], qq) print(c(aeq(qtot$quantile[i,j,], temp$quantile), aeq(qtot$upper[i,j,], temp$upper), aeq(qtot$lower[i,j,], temp$lower))) } } temp <- quantile(sfit, qq, conf.int=FALSE) all.equal(qtot$quantile, temp) # # Third case -- a survfitms object, which results from cumulative # incidence curves. # tdata <- data.frame(time=c(1,2,2,3,3,3,5,6), status = c(0,1,0,1,0,1,0,1), event = c(1,1,2,2,1,2,3,2), grp = c(1,2,1,2,1,2,1,2)) fit1 <- survfit(Surv(time, status*event, type='mstate') ~1, tdata) temp <- quantile(fit1, c(.1, .2, .5)) aeq(temp$quantile, matrix(c(2, NA, NA, 3,3,6), nrow=2, byrow=TRUE)) aeq(temp$lower , matrix(c(2, 2, NA, 3,3,3), nrow=2, byrow=TRUE)) aeq(temp$upper , c(NA,6, rep(NA,4))) fit2 <- survfit(Surv(time, status*event, type='mstate') ~1, tdata, conf.int=FALSE) temp <- quantile(fit2, c(.1, .2, .5)) aeq(temp, matrix(c(2, NA, NA, 3,3,6), nrow=2, byrow=TRUE)) # Use a larger data set for the multi-group + multi-column case, the MGUS data # However, it has almost no censoring, so add a little to make the # quantiles not be exactly even percentiles mdata <- data.frame(time=mgus1$stop, status=mgus1$status, event= mgus1$event, sex=mgus1$sex, stat2= mgus1$event) mdata$stat2[seq(1, nrow(mdata), by=5)] <- "censor" fit3 <- survfit(Surv(time, stat2) ~sex, mdata) temp1 <- quantile(fit3, 0:10/20) temp2 <- quantile(fit3, 0:10/20, conf.int=FALSE) aeq(temp1$quantile, temp2) for (i in 1:2) { for (j in 1:2){ temp3 <- quantile(fit3[i,j], 0:10/20) print(c(aeq(temp1$quantile[i,j,], temp3$quantile), aeq(temp1$upper[i,j,], temp3$upper), aeq(temp1$lower[i,j,], temp3$lower))) } } # Do one set of quantiles by brute force zz <- 1:fit3$strata[1] temp3 <- double(10) tt <- fit3$time[zz] for (i in 1:10) temp3[i] <- min(tt[fit3$prev[zz,2] > i/20]) aeq(temp3, temp2[1,2,2:11]) survival/tests/fr_ovarian.Rout.save0000644000175100001440000000400112536400664017244 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Test on the ovarian data > > fit1 <- coxph(Surv(futime, fustat) ~ rx + age, ovarian) > fit2 <- coxph(Surv(futime, fustat) ~ rx + pspline(age, df=2), + data=ovarian) > fit2$iter [1] 2 8 > > fit2$df [1] 0.9426611 1.9293051 > > fit2$history $`pspline(age, df = 2)` $`pspline(age, df = 2)`$theta [1] 0.4468868 $`pspline(age, df = 2)`$done [1] TRUE $`pspline(age, df = 2)`$history thetas dfs [1,] 1.0000000 1.000000 [2,] 0.0000000 5.000000 [3,] 0.6000000 1.734267 [4,] 0.4845205 1.929305 $`pspline(age, df = 2)`$half [1] 0 > > fit4 <- coxph(Surv(futime, fustat) ~ rx + pspline(age, df=4), + data=ovarian) > fit4 Call: coxph(formula = Surv(futime, fustat) ~ rx + pspline(age, df = 4), data = ovarian) coef se(coef) se2 Chisq DF p rx -0.373 0.761 0.749 0.241 1.00 0.6238 pspline(age, df = 4), lin 0.139 0.044 0.044 9.978 1.00 0.0016 pspline(age, df = 4), non 2.592 2.93 0.4457 Iterations: 3 outer, 14 Newton-Raphson Theta= 0.242 Degrees of freedom for terms= 1.0 3.9 Likelihood ratio test=19.4 on 4.9 df, p=0.00149 n= 26 > > > > proc.time() user system elapsed 0.204 0.020 0.218 survival/tests/difftest.Rout.save0000644000175100001440000000604211732700061016724 0ustar hornikusers R version 2.14.0 Under development (unstable) (2011-04-10 r55401) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test some more features of surv.diff > # > # First, what happens when one group is a dummy > # > > > # > # The AML data, with a third group of early censorings "tacked on" > # > aml3 <- list(time= c( 9, 13, 13, 18, 23, 28, 31, 34, 45, 48, 161, + 5, 5, 8, 8, 12, 16, 23, 27, 30, 33, 43, 45, + 1, 2, 2, 3, 3, 3, 4), + status= c( 1,1,0,1,1,0,1,1,0,1,0, 1,1,1,1,1,0,1,1,1,1,1,1, + 0,0,0,0,0,0,0), + x = as.factor(c(rep("Maintained", 11), + rep("Nonmaintained", 12), rep("Dummy",7) ))) > > aml3 <- data.frame(aml3) > > # These should give the same result (chisq, df), but the second has an > # extra group > survdiff(Surv(time, status) ~x, aml) Call: survdiff(formula = Surv(time, status) ~ x, data = aml) N Observed Expected (O-E)^2/E (O-E)^2/V x=Maintained 11 7 10.69 1.27 3.4 x=Nonmaintained 12 11 7.31 1.86 3.4 Chisq= 3.4 on 1 degrees of freedom, p= 0.0653 > survdiff(Surv(time, status) ~x, aml3) Call: survdiff(formula = Surv(time, status) ~ x, data = aml3) N Observed Expected (O-E)^2/E (O-E)^2/V x=Dummy 7 0 0.00 NaN NaN x=Maintained 11 7 10.69 1.27 3.4 x=Nonmaintained 12 11 7.31 1.86 3.4 Chisq= 3.4 on 1 degrees of freedom, p= 0.0653 > > > # > # Now a test of the stratified log-rank > # There are no tied times within institution, so the coxph program > # can be used to give a complete test > # > fit <- survdiff(Surv(time, status) ~ pat.karno + strata(inst), cancer) > > cfit <- coxph(Surv(time, status) ~ factor(pat.karno) + strata(inst), + cancer, iter=0) > > tdata <- na.omit(cancer[,c('time', 'status', 'pat.karno', 'inst')]) > > temp1 <- tapply(tdata$status-1, list(tdata$pat.karno, tdata$inst), sum) > temp1 <- ifelse(is.na(temp1), 0, temp1) > temp2 <- tapply(cfit$resid, list(tdata$pat.karno, tdata$inst), sum) > temp2 <- ifelse(is.na(temp2), 0, temp2) > > temp2 <- temp1 - temp2 > > #Now temp1=observed, temp2=expected > all.equal(c(temp1), c(fit$obs)) [1] TRUE > all.equal(c(temp2), c(fit$exp)) [1] TRUE > > all.equal(fit$var[-1,-1], solve(cfit$var)) [1] TRUE > > rm(tdata, temp1, temp2) > survival/tests/turnbull.Rout.save0000644000175100001440000001724311732700061016770 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # The test data set from Turnbull, JASA 1974, 169-73. > # > # status 0=right censored > # 1=exact > # 2=left censored > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > > turnbull <- data.frame( time =c( 1,1,1, 2,2,2, 3,3,3, 4,4,4), + status=c( 1,0,2, 1,0,2, 1,0,2, 1,0,2), + n =c(12,3,2, 6,2,4, 2,0,2, 3,3,5)) > # > # Compute the K-M for the Turnbull data > # via a slow EM calculation > # > > emsurv <- function(time, status, wt, verbose=T) { + left.cen <- (status==2) + if (!any(left.cen)) stop("No left censored data!") + if (!any(status==1))stop("Must have some exact death times") + + tempy <- Surv(time[!left.cen], status[!left.cen]) + ww <- wt[!left.cen] + tempx <- factor(rep(1, sum(!left.cen))) + tfit <- survfit(tempy~tempx, weight=ww) + if (verbose) + cat("Iteration 0, survival=", format(round(tfit$surv[tfit$n.event>0],3)), + "\n") + + stimes <- tfit$time[tfit$n.event>0] + ltime <- time[left.cen] + lwt <- wt[left.cen] + tempx <- factor(rep(1, length(stimes) + sum(!left.cen))) + tempy <- Surv(c(time[!left.cen], stimes), + c(status[!left.cen], rep(1, length(stimes)))) + for (iter in 1:4) { + wt2 <- stimes*0 + ssurv <- tfit$surv[tfit$n.event>0] + sjump <- diff(c(1, ssurv)) + for (j in 1:(length(ltime))) { + k <- sum(ltime[j]>=stimes) #index of the death time + if (k==0) + stop("Left censored observation before the first death") + wt2[1:k] <- wt2[1:k] + lwt[j]*sjump[1:k] /(ssurv[k]-1) + } + tfit <- survfit(tempy~tempx, weight=c(ww, wt2)) + if (verbose) { + cat("Iteration", iter, "survival=", + format(round(tfit$surv[tfit$n.event>0],3)), "\n") + cat(" weights=", format(round(wt2,3)), "\n") + } + } + survfit(tempy ~ tempx, weights=c(ww, wt2)) + } > > temp <-emsurv(turnbull$time, turnbull$status, turnbull$n) Iteration 0, survival= 0.613 0.383 0.287 0.144 Iteration 1 survival= 0.549 0.303 0.214 0.094 weights= 7.856 3.477 0.828 0.839 Iteration 2 survival= 0.540 0.296 0.210 0.095 weights= 8.228 3.394 0.714 0.664 Iteration 3 survival= 0.538 0.295 0.210 0.095 weights= 8.315 3.356 0.690 0.638 Iteration 4 survival= 0.538 0.295 0.210 0.095 weights= 8.338 3.342 0.685 0.635 > print(summary(temp)) Call: survfit(formula = tempy ~ tempx, weights = c(ww, wt2)) time n.risk n.event survival std.err lower 95% CI upper 95% CI 1 44.00 20.34 0.5378 0.0752 0.4089 0.707 2 20.66 9.34 0.2946 0.0719 0.1827 0.475 3 9.32 2.68 0.2098 0.0673 0.1119 0.393 4 6.64 3.64 0.0948 0.0507 0.0333 0.270 > # First check, use the data from Turnbull, JASA 1974, 169-173. > > tdata <- data.frame(time =c(1,1,1,2,2,2,3,3,3,4,4,4), + status=rep(c(1,0,2),4), + n =c(12,3,2,6,2,4,2,0,2,3,3,5)) > > tfit <- survfit(Surv(time, time, status, type='interval') ~1, tdata, weight=n) > all.equal(round(tfit$surv,3), c(.538, .295, .210, .095)) [1] TRUE > > > # Second check, compare to a reversed survival curve > # This is not as simple a test as one might think, because left and right > # censored observations are not treated symmetrically by the routine: > # time <= y for left and time> y for right (this is to make the routine > # correct for the common situation of panel data). > # To get equivalence, make the left censoreds happen just a little bit > # earlier. The left-continuous/right-continuous shift is also a bother. > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > fit1 <- survfit(Surv(time, status) ~1, test1) > temp <- ifelse(test1$status==0, 4.99,5) - test1$time > fit2 <- survfit(Surv(temp, status, type='left') ~1, test1) > > all.equal(round(fit1$surv[1:2],5), round(1-fit2$surv[3:2],5)) [1] TRUE > > rm(tdata, tfit, fit1, temp, fit2) > # > # Create a data set similar to the one provided by Al Zinsmeister > # It is a hard test case for survfit.turnbull > # > time1 <- c(rep(0,100), rep(1,200), 100, 200, 210, 220, + rep(365,100), rep(366,5), 731:741) > > time2 <- c((1:100)*3, 10+1:100, rep(365:366, c(60,40)), NA, 500, NA, 450, + rep(730,90), rep(NA,10), c(528,571,691,730,731), + NA, 1095:1099, NA, 1400, 1200, 772, 1461) > > zfit <- survfit(Surv(time1, time2, type='interval2') ~1) > > # > # There are 100 intervals of the form (0,x) where x is from 3 to 300, > # and 200 more of the form (1,x) where x is from 11 to 366. These > # lead to a mass point in the interval (1,3), which is placed at 2. > # The starting estimate has far too little mass placed here, and it takes > # the EM a long time to realize that most of the weight for the first 300 > # subjects goes here. With acceleration, it takes 16 iterations, without > # it takes >40. (On Al's orginal data, without accel still wasn't there after > # 165 iters!) > # > # The next 4 obs give rise to potential jumps at 100.5, 200.5, 211.5, and > # 221. However, the final estimate has no mass at all on any of these. > # Assume mass of a,b, and c at 2, 100.5 and 365.5, and consider the > # contributions: > # 123 obs that overlap a only > # 137 obs that overlap a and b > # 40 obs that overlap a, b, c > # 1 obs that overlap b, c > # 108 obs that overlap c (200, 210,200, 365, and 366 starting points) > # For some trial values of a,b,c, compare the loglik to that of (a+b),0,c > # First one: a^123 (a+b)^137 (a+b+c)^40 (b+c) c^108 > # Second: (a+b)^123 (a+b)^137 (a+b+c)^40 c c^108 > # Likelhood improves if (1 + b/a)^123 > 1+ b/c, which is true for almost > # all a and c. In particular, at the solution a and c are approx .7 and > # .18, respectively. > # > # The program can't see this coming, of course, and so iterates towards a > # KM with epsilon sized jumps at 100.5, 200.5, and 211.5. Whether these > # intervals should be removed during iteration, as detected, is an open > # question for me. > # > # > # True solution: mass points at 2, 365.5, 408, and 756.5, of sizes a, b, c, d > # Likelihood: a^260 (a+b)^40 (b+c)^92 (b+c+d)^12 c^5 d^11 > # Solution: a=0.6958, b=0.1674, c=0.1079, d=0.0289 > > tfun <- function(x) { + if (length(x) ==3) x <- c(x, .03) + x <- x/sum(x) #make probabilities sum to 1 + loglik <- 260*log(x[1]) + 40*log(x[1]+x[2]) + 92*log(x[2] + x[3]) + + 12*log(x[2]+x[3]+x[4]) + 5*log(x[3]) + 11*log(x[4]) + -loglik #find the max, not the min + } > > nfit <- nlminb(start=c(.7,.15, .1), tfun, lower=0, upper=1) > nparm <- c(nfit$par, .03) > nparm <- nparm / sum(nparm) > zparm <- -diff(c(1, zfit$surv[match(c(2, 365.5, 408, 756.5), zfit$time)])) > aeq(round(tfun(nparm),4), round(tfun(zparm),4)) [1] TRUE > # .0001 is the tolerance in survfit.turnbull > > rm(tfun, nfit, nparm, zparm, time1, time2, zfit) > survival/tests/aareg.R0000644000175100001440000001604511732700061014512 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test aareg, for some simple data where the answers can be computed # in closed form # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) test1 <- data.frame(time= c(4, 3,1,1,2,2,3), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0), wt= c(1, 1:6)) tfit <- aareg(Surv(time, status) ~ x, test1) aeq(tfit$times, c(1,2,2)) aeq(tfit$nrisk, c(6,4,4)) aeq(tfit$coefficient, matrix(c(0,0,1/3, 1/3, 1, -1/3), ncol=2)) aeq(tfit$tweight, matrix(c(3,3,3, 3/2, 3/4, 3/4), ncol=2)) aeq(tfit$test.statistic, c(1,1)) aeq(tfit$test.var, c(1, -1/4, -1/4, 1/4 + 9/16 + 1/16)) tfit <- aareg(Surv(time, status) ~ x, test1, test='nrisk') aeq(tfit$tweight, matrix(c(3,3,3, 3/2, 3/4, 3/4), ncol=2)) #should be as before aeq(tfit$test.statistic, c(4/3, 6/3+ 4 - 4/3)) aeq(tfit$test.var, c(16/9, -16/9, -16/9, 36/9 + 16 + 16/9)) # In the 1-variable case, this is the same as the default Aalen weight tfit <- aareg(Surv(time, status) ~ x, test1, test='variance') aeq(tfit$test.statistic, c(1,1)) aeq(tfit$test.var, c(1, -1/4, -1/4, 1/4 + 9/16 + 1/16)) # # Repeat the above, with case weights # tfit <- aareg(Surv(time, status) ~x, test1, weights=wt) aeq(tfit$times, c(1,2,2)) aeq(tfit$nrisk, c(21,16,16)) aeq(tfit$coefficient, matrix(c(0,0,5/12, 2/9, 1, -5/12), ncol=2)) aeq(tfit$tweight, matrix(c(12,12,12, 36/7, 3,3), ncol=2)) aeq(tfit$test.statistic, c(5, 72/63 + 3 - 15/12)) aeq(tfit$test.var, c(25, -25/4, -25/4, (72/63)^2 + 9 + (5/4)^2)) tfit <- aareg(Surv(time, status) ~x, test1, weights=wt, test='nrisk') aeq(tfit$test.statistic, c(20/3, 42/9 + 16 - 16*5/12)) aeq(tfit$test.var, c(400/9, -400/9, -400/9, (42/9)^2 + 16^2 + (16*5/12)^2)) # # Make a test data set with no NAs, in sorted order, no ties, # 15 observations tdata <- lung[15:29, c('time', 'status', 'age', 'sex', 'ph.ecog')] tdata$status <- tdata$status -1 tdata <- tdata[order(tdata$time, tdata$status),] row.names(tdata) <- 1:15 tdata$status[8] <- 0 #for some variety afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, tdata, nmin=6) # # Now, do it "by hand" cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, tdata, iter=0, method='breslow') dt1 <- coxph.detail(cfit) sch1 <- resid(cfit, type='schoen') # First estimate of Aalen: from the Cox computations, first 9 # The first and last cols of the ninth are somewhat unstable (approx =0) mine <- rbind(solve(dt1$imat[,,1], sch1[1,]), solve(dt1$imat[,,2], sch1[2,]), solve(dt1$imat[,,3], sch1[3,]), solve(dt1$imat[,,4], sch1[4,]), solve(dt1$imat[,,5], sch1[5,]), solve(dt1$imat[,,6], sch1[6,]), solve(dt1$imat[,,7], sch1[7,]), solve(dt1$imat[,,8], sch1[8,]), solve(dt1$imat[,,9], sch1[9,])) mine <- diag(1/dt1$nrisk[1:9]) %*% mine aeq(mine, afit$coef[1:9, -1]) rm(tfit, afit, mine, dt1, cfit, sch1) # # Check out the dfbeta matrix from aareg # Note that it is kept internally in time order, not data set order # Those who want residuals should use the resid function! # # First, the simple test case where I know the anwers # afit <- aareg(Surv(time, status) ~ x, test1, dfbeta=T) temp <- c(rep(0,6), #intercepts at time 1 c(2,-1,-1,0,0,0)/9, #alpha at time 1 c(0,0,0,2, -1, -1)/9, #intercepts at time 2 c(0,0,0,-2,1,1)/9) #alpha at time 2 aeq(afit$dfbeta, temp) # #Now a multivariate data set # afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, lung, dfbeta=T) ord <- order(lung$time, -lung$status) cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, lung[ord,], method='breslow', iter=0, x=T) cdt <- coxph.detail(cfit, riskmat=T) # an arbitrary list of times acoef <- rowsum(afit$coef, afit$times) #per death time coefs indx <- match(cdt$time, afit$times) for (i in c(2,5,27,54,101, 135)) { lwho <- (cdt$riskmat[,i]==1) lmx <- cfit$x[lwho,] lmy <- 1*( cfit$y[lwho,2]==1 & cfit$y[lwho,1] == cdt$time[i]) fit <- lm(lmy~ lmx) cat("i=", i, "coef=", aeq(fit$coef, acoef[i,])) rr <- diag(resid(fit)) zz <- cbind(1,lmx) zzinv <- solve(t(zz) %*% zz) cat(" twt=", aeq(1/(diag(zzinv)), afit$tweight[indx[i],])) df <- t(zzinv %*% t(zz) %*% rr) cat(" dfbeta=", aeq(df, afit$dfbeta[lwho,,i]), "\n") } rm(afit, cfit, cdt, lwho, lmx, lmy, fit, rr, zz, df) # Repeat it with case weights ww <- rep(1:5, length=nrow(lung))/ 3.0 afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, lung, dfbeta=T, weights=ww) cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, lung[ord,], method='breslow', iter=0, x=T, weight=ww[ord]) cdt <- coxph.detail(cfit, riskmat=T) acoef <- rowsum(afit$coef, afit$times) #per death time coefs for (i in c(2,5,27,54,101, 135)) { who <- (cdt$riskmat[,i]==1) x <- cfit$x[who,] y <- 1*( cfit$y[who,2]==1 & cfit$y[who,1] == cdt$time[i]) w <- cfit$weight[who] fit <- lm(y~x, weights=w) cat("i=", i, "coef=", aeq(fit$coef, acoef[i,])) rr <- diag(resid(fit)) zz <- cbind(1,x) zzinv <- solve(t(zz)%*% (w*zz)) cat(" twt=", aeq(1/(diag(zzinv)), afit$tweight[indx[i],])) df <- t(zzinv %*% t(zz) %*% (w*rr)) cat(" dfbeta=", aeq(df, afit$dfbeta[who,,i]), "\n") } rm(afit, cfit, cdt, who, x, y, fit, rr, zz, df) rm(ord, acoef) # # Check that the test statistic computed within aareg and # the one recomputed within summary.aareg are the same. # Of course, they could both be wrong, but at least they'll agree! # If the maxtime argument is used in summary, it recomputes the test, # even if we know that it wouldn't have had to. # # Because the 1-variable and >1 variable case have different code, test # them both. # afit <- aareg(Surv(time, status) ~ age, lung, dfbeta=T) asum <- summary(afit, maxtime=max(afit$times)) aeq(afit$test.stat, asum$test.stat) aeq(afit$test.var, asum$test.var) aeq(afit$test.var2, asum$test.var2) print(afit) afit <- aareg(Surv(time, status) ~ age, lung, dfbeta=T, test='nrisk') asum <- summary(afit, maxtime=max(afit$times)) aeq(afit$test.stat, asum$test.stat) aeq(afit$test.var, asum$test.var) aeq(afit$test.var2, asum$test.var2) summary(afit) # # Mulitvariate # afit <- aareg(Surv(time, status) ~ age + sex + ph.karno + pat.karno, lung, dfbeta=T) asum <- summary(afit, maxtime=max(afit$times)) aeq(afit$test.stat, asum$test.stat) aeq(afit$test.var, asum$test.var) aeq(afit$test.var2, asum$test.var2) print(afit) afit <- aareg(Surv(time, status) ~ age + sex + ph.karno + pat.karno, lung, dfbeta=T, test='nrisk') asum <- summary(afit, maxtime=max(afit$times)) aeq(afit$test.stat, asum$test.stat) aeq(afit$test.var, asum$test.var) aeq(afit$test.var2, asum$test.var2) summary(afit) # Weights play no role in the final computation of the test statistic, given # the coefficient matrix, nrisk, and dfbeta as inputs. (Weights do # change the inputs). So there is no need to reprise the above with # case weights. survival/tests/jasa.R0000644000175100001440000000660612160143136014353 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) expect <- survexp(futime ~ ratetable(age=(accept.dt - birth.dt), sex=1, year=accept.dt, race='white'), jasa, cohort=F, ratetable=survexp.usr) survdiff(Surv(jasa$futime, jasa$fustat) ~ offset(expect)) # Now fit the 6 models found in Kalbfleisch and Prentice, p139 sfit.1 <- coxph(Surv(start, stop, event)~ (age + surgery)*transplant, jasa1, method='breslow') sfit.2 <- coxph(Surv(start, stop, event)~ year*transplant, jasa1, method='breslow') sfit.3 <- coxph(Surv(start, stop, event)~ (age + year)*transplant, jasa1, method='breslow') sfit.4 <- coxph(Surv(start, stop, event)~ (year +surgery) *transplant, jasa1, method='breslow') sfit.5 <- coxph(Surv(start, stop, event)~ (age + surgery)*transplant + year , jasa1, method='breslow') sfit.6 <- coxph(Surv(start, stop, event)~ age*transplant + surgery + year, jasa1, method='breslow') summary(sfit.1) sfit.2 summary(sfit.3) sfit.4 sfit.5 sfit.6 # Survival curve for an "average" subject, # done once as overall, once via individual method surv1 <- survfit(sfit.1, newdata=list(age=-2, surgery=0, transplant=0)) newdata <- data.frame(start=c(0,50,100), stop=c(50,100, max(jasa1$stop)), event=c(1,1,1), age=rep(-2,3), surgery=rep(0,3), transplant=rep(0,3)) surv2 <- survfit(sfit.1, newdata, individual=T) # Have to use unclass to avoid [.survfit trying to pick curves, # remove the final element "call" because it won't match all.equal(unclass(surv1)[-length(surv1)], unclass(surv2)[-length(surv2)]) # Survival curve for a subject of age 50, with prior surgery, tx at 6 months # Remember that 'age' in jasa 1 was centered at 48 data <- data.frame(start=c(0,183), stop=c(183,3*365), event=c(1,1), age=c(2,2), surgery=c(1,1), transplant=c(0,1)) summary(survfit(sfit.1, data, individual=T)) # These should all give the same answer # When there are offsets, the default curve is always for someone with # the mean offset. j.age <- jasa$age -48 fit1 <- coxph(Surv(futime, fustat) ~ j.age, data=jasa) fit2 <- coxph(Surv(futime, fustat) ~ j.age, jasa, init=fit1$coef, iter=0) fit3 <- coxph(Surv(start, stop, event) ~ age, jasa1) fit4 <- coxph(Surv(start, stop, event) ~ offset(age*fit1$coef), jasa1) s1 <- survfit(fit1, list(j.age=fit3$means), censor=FALSE) s2 <- survfit(fit2, list(j.age=fit3$means), censor=FALSE) s3 <- survfit(fit3, censor=FALSE) s4 <- survfit(fit4, censor=FALSE) all.equal(s1$surv, s2$surv) all.equal(s1$surv, s3$surv) all.equal(s1$surv, s4$surv) # Still the same answer, fit multiple strata at once # Strata 1 has independent coefs of strata 2, so putting in # the other data should not affect it ll <- nrow(jasa1) ss <- rep(0:1, c(ll,ll)) tdata <- with(jasa1, data.frame(start=rep(start,2), stop=rep(stop,2), event=rep(event,2), ss=ss, age=rep(age,2), age2 = (rep(age,2))^2 * ss)) fit <- coxph(Surv(start, stop, event) ~ age*strata(ss) + age2, tdata) # Above replaced these 2 lines, which kill Splus5 as of 8/98 # Something with data frames, I expect. #fit <- coxph(Surv(rep(start,2), rep(stop,2), rep(event,2)) ~ # rep(age,2)*strata(ss) + I(rep(age,2)^2*ss) ) all.equal(fit$coef[1], fit3$coef) s5 <- survfit(fit, data.frame(age=fit3$means, age2=0, ss=0), censor=FALSE) all.equal(s5$surv[1:(s5$strata[1])], s3$surv) survival/tests/model.matrix.R0000644000175100001440000000424312466142446016047 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test out the revised model.matrix code # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0), z= factor(c('a', 'a', 'b', 'b', 'c', 'c', 'a'))) fit1 <- coxph(Surv(time, status) ~ z, test1, iter=1) fit2 <- coxph(Surv(time, status) ~z, test1, x=T, iter=1) all.equal(model.matrix(fit1), fit2$x) # This has no level 'b', make sure dummies recode properly test2 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0), z= factor(c('a', 'a', 'a', 'a', 'c', 'c', 'a'))) ftest <- model.frame(fit1, data=test2) all.equal(levels(ftest$z), levels(test1$z)) # xtest will have one more row than the others, since it does not delete # the observation with a missing value for status xtest <- model.matrix(fit1, data=test2) dummy <- fit2$x dummy[,1] <- 0 all.equal(xtest[-2,], dummy, check.attributes=FALSE) # The case of a strata by factor interaction # Use iter=0 since there are too many covariates and it won't converge test1$x2 <- factor(rep(1:2, length=7)) fit3 <- coxph(Surv(time, status) ~ strata(x2)*z, test1, iter=0) xx <- model.matrix(fit3) all.equal(attr(xx, "assign"), c(2,2,3,3)) all.equal(colnames(xx), c("zb", "zc", "strata(x2)2:zb", "strata(x2)2:zc")) all.equal(attr(xx, "contrasts"), list("strata(x2)"= "contr.treatment", z="contr.treatment")) fit3b <- coxph(Surv(time, status) ~ strata(x2)*z, test1, iter=0, x=TRUE) all.equal(fit3b$x, xx) # A model with a tt term fit4 <- coxph(Surv(time, status) ~ tt(x) + x, test1, iter=0, tt = function(x, t, ...) x*t) ff <- model.frame(fit4) # There is 1 subject in the final risk set, 4 at risk at time 6, 6 at time 1 # The .strata. variable numbers from last time point to first all.equal(ff$.strata., rep(1:3, c(1, 4,6))) all.equal(ff[["tt(x)"]], ff$x* c(9,6,1)[ff$.strata.]) xx <- model.matrix(fit4) all.equal(xx[,1], ff[[2]], check.attributes=FALSE) survival/tests/fr_simple.R0000644000175100001440000000427712350363660015426 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test the logic of the penalized code by fitting some no-frailty models # (theta=0). It should give exactly the same answers as 'ordinary' coxph. # test1 <- data.frame(time= c(4, 3,1,1,2,2,3), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) zz <- rep(0, nrow(test1)) tfit1 <- coxph(Surv(time,status) ~x, test1, eps=1e-7) tfit2 <- coxph(Surv(time,status) ~x + frailty(zz, theta=0, sparse=T), test1) tfit3 <- coxph(Surv(zz,time,status) ~x + frailty(zz, theta=0, sparse=T), test1) temp <- c('coefficients', 'var', 'loglik', 'linear.predictors', 'means', 'n', 'concordance') all.equal(tfit1[temp], tfit2[temp]) all.equal(tfit2[temp], tfit3[temp]) zz <- rep(0, nrow(test2)) tfit1 <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-7) tfit2 <- coxph(Surv(start, stop, event) ~ x + frailty(zz, theta=0, sparse=T), test2) all.equal(tfit1[temp], tfit2[temp]) # # Repeat the above tests, but with a strata added # Because the data set is simply doubled, the loglik will double, # beta is the same, variance is halved. # test3 <- rbind(test1, test1) test3$x2 <- rep(1:2, rep(nrow(test1),2)) zz <- rep(0, nrow(test3)) tfit1 <- coxph(Surv(time,status) ~x + strata(x2), test3, eps=1e-7) tfit2 <- coxph(Surv(time,status) ~x + frailty(zz, theta=0, sparse=T) + strata(x2), test3) tfit3 <- coxph(Surv(zz,time,status) ~x + frailty(zz, theta=0, sparse=T) + strata(x2), test3) all.equal(tfit1[temp], tfit2[temp]) all.equal(tfit2[temp], tfit3[temp]) test4 <- rbind(test2, test2) test4$x2 <- rep(1:2, rep(nrow(test2),2)) zz <- rep(0, nrow(test4)) tfit1 <- coxph(Surv(start, stop, event) ~x, test4, eps=1e-7) tfit2 <- coxph(Surv(start, stop, event) ~ x + frailty(zz, theta=0, sparse=T), test4) all.equal(tfit1[temp], tfit2[temp]) rm(test3, test4, tfit1, tfit2, tfit3, temp, zz) survival/tests/difftest.R0000644000175100001440000000314411732700061015237 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test some more features of surv.diff # # First, what happens when one group is a dummy # # # The AML data, with a third group of early censorings "tacked on" # aml3 <- list(time= c( 9, 13, 13, 18, 23, 28, 31, 34, 45, 48, 161, 5, 5, 8, 8, 12, 16, 23, 27, 30, 33, 43, 45, 1, 2, 2, 3, 3, 3, 4), status= c( 1,1,0,1,1,0,1,1,0,1,0, 1,1,1,1,1,0,1,1,1,1,1,1, 0,0,0,0,0,0,0), x = as.factor(c(rep("Maintained", 11), rep("Nonmaintained", 12), rep("Dummy",7) ))) aml3 <- data.frame(aml3) # These should give the same result (chisq, df), but the second has an # extra group survdiff(Surv(time, status) ~x, aml) survdiff(Surv(time, status) ~x, aml3) # # Now a test of the stratified log-rank # There are no tied times within institution, so the coxph program # can be used to give a complete test # fit <- survdiff(Surv(time, status) ~ pat.karno + strata(inst), cancer) cfit <- coxph(Surv(time, status) ~ factor(pat.karno) + strata(inst), cancer, iter=0) tdata <- na.omit(cancer[,c('time', 'status', 'pat.karno', 'inst')]) temp1 <- tapply(tdata$status-1, list(tdata$pat.karno, tdata$inst), sum) temp1 <- ifelse(is.na(temp1), 0, temp1) temp2 <- tapply(cfit$resid, list(tdata$pat.karno, tdata$inst), sum) temp2 <- ifelse(is.na(temp2), 0, temp2) temp2 <- temp1 - temp2 #Now temp1=observed, temp2=expected all.equal(c(temp1), c(fit$obs)) all.equal(c(temp2), c(fit$exp)) all.equal(fit$var[-1,-1], solve(cfit$var)) rm(tdata, temp1, temp2) survival/tests/ovarian.R0000644000175100001440000000353111741355706015102 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test the coxph program on the Ovarian data # attach(ovarian) summary(survfit(Surv(futime, fustat)~1), censor=TRUE) # Various models coxph(Surv(futime, fustat)~ age) coxph(Surv(futime, fustat)~ resid.ds) coxph(Surv(futime, fustat)~ rx) coxph(Surv(futime, fustat)~ ecog.ps) coxph(Surv(futime, fustat)~ resid.ds + rx + ecog.ps) coxph(Surv(futime, fustat)~ age + rx + ecog.ps) coxph(Surv(futime, fustat)~ age + resid.ds + ecog.ps) coxph(Surv(futime, fustat)~ age + resid.ds + rx) # Residuals fit <- coxph(Surv(futime, fustat)~ age + resid.ds + rx + ecog.ps ) resid(fit) resid(fit, 'dev') resid(fit, 'scor') resid(fit, 'scho') fit <- coxph(Surv(futime, fustat) ~ age + ecog.ps + strata(rx)) summary(fit) summary(survfit(fit)) sfit <- survfit(fit, list(age=c(30,70), ecog.ps=c(2,3))) #two columns sfit summary(sfit) detach() # Check of offset + surv, added 7/2000 fit1 <- coxph(Surv(futime, fustat) ~ age + rx, ovarian, control=coxph.control(eps=1e-8)) fit2 <- coxph(Surv(futime, fustat) ~ age + offset(rx*fit1$coef[2]), ovarian, control=coxph.control(eps=1e-8)) all.equal(fit1$coef[1], fit2$coef[1]) fit <- coxph(Surv(futime, fustat) ~ age + offset(rx), ovarian) survfit(fit, censor=FALSE)$surv^exp(-1.5) # Check it by hand -- there are no tied times # Remember that offsets from survfit are centered, which is 1.5 for # this data set. eta <- fit$coef*(ovarian$age - fit$mean) + (ovarian$rx - 1.5) ord <- order(ovarian$futime) risk <- exp(eta[ord]) rsum <- rev(cumsum(rev(risk))) # cumulative risk at each time point dead <- (ovarian$fustat[ord]==1) baseline <- cumsum(1/rsum[dead]) all.equal(survfit(fit, censor=FALSE)$surv, exp(-baseline)) rm(fit, fit1, fit2, ord, eta, risk, rsum, dead, baseline, sfit) survival/tests/quantile.Rout.save0000644000175100001440000001306112466142446016751 0ustar hornikusers R Under development (unstable) (2014-09-01 r66509) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > # Formal test of the quantile routine for survfit > library(survival) > aeq <- function(x, y, ...) all.equal(as.vector(x), as.vector(y), ...) > > # There are 8 cases: strata Y/N, ncol(surv) >1, conf.int = T/F > # Subcase: the quantile exactly agrees with a horizontal segment of > # the curve or not. > # First do the 4 cases where fit$surv is a vector > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8, 10), + status=c(1,NA,1,0,1,1,0, 0), + x= c(0, 2,1,1,1,0,0, 0)) > > # True survival = (6/7) * (3/5) * (1/2) for overall > # The q's are chosen to include a point < first jump, mid, after last jump, > # and exact intersections with the "flats" of the curve. > # > qq <- c(13/14, 6/7, 2/3, .5, 9/35, .1) > > # Nothing on the right hand side, simple survival (no strata) > fit1 <- survfit(Surv(time, status) ~ 1, test1, conf.type='none') > aeq(quantile(fit1, 1-qq), c(1, 3.5, 6, 9, 9.5, NA)) #without conf.int [1] TRUE > > fit2 <- survfit(Surv(time, status) ~ 1, test1) #with conf.int > aeq(quantile(fit2, 1-qq), + list(quantile = c(1, 3.5, 6, 9, 9.5, NA), + lower = c(1,1,1,6,6,9), + upper = rep(as.numeric(NA), 6)), check.attributes=FALSE) [1] TRUE > aeq(quantile(fit2, 1-qq, FALSE), c(1, 3.5, 6, 9, 9.5, NA)) [1] TRUE > > > # Now a variable on the right (strata in the result) > # curve 0: (t=6, S=3/4), (t=9, S=3/8) > # curve 1: (t=1, S=2/3), (t=6, S= 0) > fit1 <- survfit(Surv(time, status) ~ x, test1, conf.type='none') > aeq(quantile(fit1, 1-qq), + matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) [1] TRUE > > fit2 <- survfit(Surv(time, status) ~ x, test1) > aeq(quantile(fit2, 1-qq, FALSE), + matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) [1] TRUE > > temp <- quantile(fit2, 1-qq) > aeq(temp$quantile, matrix(c(6,6,9,9,NA,NA, 1,1,3.5, 6,6,6), nrow=2, byrow=T)) [1] TRUE > aeq(temp$lower, matrix(c(6,6,6,6,9,9, 1,1,1,1, NA,NA), nrow=2, byrow=T)) [1] TRUE > aeq(temp$upper, rep(as.numeric(NA), 12)) [1] TRUE > > # Second major case set -- a survfit object where fit$surv is a matrix > # This arises from coxph models > # There is only 1 subject with ph.ecog=3 which is a nice edge case > cfit <- coxph(Surv(time, status) ~ age + strata(ph.ecog), lung) > sfit <- survfit(cfit, newdata=data.frame(age=c(50, 70))) > qtot <- quantile(sfit, qq) > for (i in 1:4) { + for (j in 1:2) { + temp <- quantile(sfit[i,j], qq) + print(c(aeq(qtot$quantile[i,j,], temp$quantile), + aeq(qtot$upper[i,j,], temp$upper), + aeq(qtot$lower[i,j,], temp$lower))) + } + } [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE > temp <- quantile(sfit, qq, conf.int=FALSE) > all.equal(qtot$quantile, temp) [1] TRUE > > # > # Third case -- a survfitms object, which results from cumulative > # incidence curves. > # > tdata <- data.frame(time=c(1,2,2,3,3,3,5,6), + status = c(0,1,0,1,0,1,0,1), + event = c(1,1,2,2,1,2,3,2), + grp = c(1,2,1,2,1,2,1,2)) > > fit1 <- survfit(Surv(time, status*event, type='mstate') ~1, tdata) > temp <- quantile(fit1, c(.1, .2, .5)) > aeq(temp$quantile, matrix(c(2, NA, NA, 3,3,6), nrow=2, byrow=TRUE)) [1] TRUE > aeq(temp$lower , matrix(c(2, 2, NA, 3,3,3), nrow=2, byrow=TRUE)) [1] TRUE > aeq(temp$upper , c(NA,6, rep(NA,4))) [1] TRUE > > fit2 <- survfit(Surv(time, status*event, type='mstate') ~1, tdata, + conf.int=FALSE) > temp <- quantile(fit2, c(.1, .2, .5)) > aeq(temp, matrix(c(2, NA, NA, 3,3,6), nrow=2, byrow=TRUE)) [1] TRUE > > # Use a larger data set for the multi-group + multi-column case, the MGUS data > # However, it has almost no censoring, so add a little to make the > # quantiles not be exactly even percentiles > mdata <- data.frame(time=mgus1$stop, + status=mgus1$status, + event= mgus1$event, + sex=mgus1$sex, + stat2= mgus1$event) > mdata$stat2[seq(1, nrow(mdata), by=5)] <- "censor" > > fit3 <- survfit(Surv(time, stat2) ~sex, mdata) > temp1 <- quantile(fit3, 0:10/20) > temp2 <- quantile(fit3, 0:10/20, conf.int=FALSE) > aeq(temp1$quantile, temp2) [1] TRUE > > for (i in 1:2) { + for (j in 1:2){ + temp3 <- quantile(fit3[i,j], 0:10/20) + print(c(aeq(temp1$quantile[i,j,], temp3$quantile), + aeq(temp1$upper[i,j,], temp3$upper), + aeq(temp1$lower[i,j,], temp3$lower))) + } + } [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE [1] TRUE TRUE TRUE > > # Do one set of quantiles by brute force > zz <- 1:fit3$strata[1] > temp3 <- double(10) > tt <- fit3$time[zz] > for (i in 1:10) temp3[i] <- min(tt[fit3$prev[zz,2] > i/20]) > aeq(temp3, temp2[1,2,2:11]) [1] TRUE > > > > > proc.time() user system elapsed 0.344 0.024 0.367 survival/tests/cancer.Rout.save0000644000175100001440000002310512536400525016353 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Test out all of the routines on a more complex data set > # > temp <- survfit(Surv(time, status) ~ ph.ecog, lung) > summary(temp, times=c(30*1:11, 365*1:3)) Call: survfit(formula = Surv(time, status) ~ ph.ecog, data = lung) 1 observation deleted due to missingness ph.ecog=0 time n.risk n.event survival std.err lower 95% CI upper 95% CI 30 60 3 0.952 0.0268 0.9012 1.000 60 58 2 0.921 0.0341 0.8562 0.990 90 56 2 0.889 0.0396 0.8146 0.970 120 56 0 0.889 0.0396 0.8146 0.970 150 55 1 0.873 0.0419 0.7946 0.959 180 52 2 0.841 0.0461 0.7553 0.936 210 48 2 0.808 0.0498 0.7164 0.912 240 45 0 0.808 0.0498 0.7164 0.912 270 38 2 0.770 0.0543 0.6709 0.884 300 33 2 0.727 0.0591 0.6203 0.853 330 29 2 0.681 0.0637 0.5670 0.818 365 22 6 0.535 0.0728 0.4100 0.699 730 5 11 0.193 0.0707 0.0943 0.396 ph.ecog=1 time n.risk n.event survival std.err lower 95% CI upper 95% CI 30 111 2 0.982 0.0124 0.9583 1.000 60 110 3 0.956 0.0193 0.9186 0.994 90 104 4 0.920 0.0255 0.8718 0.972 120 99 5 0.876 0.0310 0.8174 0.939 150 93 6 0.823 0.0359 0.7556 0.896 180 82 8 0.751 0.0407 0.6756 0.836 210 68 9 0.666 0.0450 0.5831 0.760 240 57 6 0.604 0.0474 0.5176 0.704 270 53 4 0.561 0.0487 0.4729 0.665 300 46 3 0.527 0.0495 0.4384 0.633 330 40 4 0.480 0.0504 0.3903 0.589 365 34 4 0.431 0.0509 0.3417 0.543 730 7 21 0.114 0.0388 0.0582 0.222 ph.ecog=2 time n.risk n.event survival std.err lower 95% CI upper 95% CI 30 46 5 0.9000 0.0424 0.82057 0.987 60 43 2 0.8600 0.0491 0.76900 0.962 90 40 3 0.8000 0.0566 0.69647 0.919 120 34 4 0.7174 0.0641 0.60216 0.855 150 31 3 0.6541 0.0680 0.53342 0.802 180 26 6 0.5275 0.0719 0.40385 0.689 210 21 4 0.4431 0.0717 0.32266 0.608 240 17 3 0.3766 0.0705 0.26100 0.543 270 17 0 0.3766 0.0705 0.26100 0.543 300 13 3 0.3102 0.0677 0.20223 0.476 330 11 2 0.2624 0.0651 0.16135 0.427 365 9 2 0.2147 0.0614 0.12258 0.376 730 1 6 0.0371 0.0345 0.00601 0.229 ph.ecog=3 time n.risk n.event survival std.err lower 95% CI upper 95% CI 30 1 0 1 0 1 1 60 1 0 1 0 1 1 90 1 0 1 0 1 1 > print(temp[2:3]) Call: survfit(formula = Surv(time, status) ~ ph.ecog, data = lung) 1 observation deleted due to missingness n events median 0.95LCL 0.95UCL ph.ecog=1 113 82 306 268 429 ph.ecog=2 50 44 199 156 288 > > temp <- survfit(Surv(time, status)~1, lung, type='fleming', + conf.int=.9, conf.type='log-log', error='tsiatis') > summary(temp, times=30 *1:5) Call: survfit(formula = Surv(time, status) ~ 1, data = lung, type = "fleming", conf.int = 0.9, conf.type = "log-log", error = "tsiatis") time n.risk n.event survival std.err lower 90% CI upper 90% CI 30 219 10 0.956 0.0135 0.928 0.974 60 213 7 0.926 0.0173 0.891 0.950 90 201 10 0.882 0.0213 0.842 0.913 120 189 10 0.838 0.0244 0.793 0.874 150 179 10 0.794 0.0268 0.745 0.834 > > temp <- survdiff(Surv(time, status) ~ inst, lung, rho=.5) > print(temp, digits=6) Call: survdiff(formula = Surv(time, status) ~ inst, data = lung, rho = 0.5) n=227, 1 observation deleted due to missingness. N Observed Expected (O-E)^2/E (O-E)^2/V inst=1 36 21.190058 17.455181 0.799149708 1.171232977 inst=2 5 3.173330 1.964395 0.744007932 0.860140808 inst=3 19 10.663476 11.958755 0.140294489 0.200472362 inst=4 4 2.245347 3.559344 0.485085848 0.677874608 inst=5 9 5.010883 4.500982 0.057765161 0.077128402 inst=6 14 8.862602 7.078516 0.449665221 0.582743947 inst=7 8 4.445647 4.416133 0.000197254 0.000253632 inst=10 4 2.901923 2.223283 0.207150016 0.249077097 inst=11 18 7.807867 9.525163 0.309611863 0.422142221 inst=12 23 14.009656 12.216768 0.263117640 0.365712493 inst=13 20 9.140983 11.863298 0.624699853 0.874238212 inst=15 6 3.170744 3.558447 0.042241456 0.057938955 inst=16 16 8.870360 9.992612 0.126038005 0.175170113 inst=21 13 9.263733 4.460746 5.171484268 6.149354145 inst=22 17 8.278566 11.971473 1.139171459 1.645863937 inst=26 6 1.627074 3.542694 1.035821659 1.286365543 inst=32 7 1.792468 2.679904 0.293869782 0.343966668 inst=33 2 0.929177 0.416202 0.632249272 0.676682390 Chisq= 15.1 on 17 degrees of freedom, p= 0.590384 > > temp <- coxph(Surv(time, status) ~ ph.ecog + ph.karno + pat.karno + wt.loss + + sex + age + meal.cal + strata(inst), lung) > summary(temp) Call: coxph(formula = Surv(time, status) ~ ph.ecog + ph.karno + pat.karno + wt.loss + sex + age + meal.cal + strata(inst), data = lung) n= 167, number of events= 120 (61 observations deleted due to missingness) coef exp(coef) se(coef) z Pr(>|z|) ph.ecog 0.7299987 2.0750779 0.2689397 2.714 0.00664 ** ph.karno 0.0130512 1.0131368 0.0137362 0.950 0.34204 pat.karno -0.0140955 0.9860034 0.0093680 -1.505 0.13242 wt.loss -0.0148821 0.9852281 0.0084811 -1.755 0.07931 . sex -0.6612534 0.5162039 0.2339979 -2.826 0.00471 ** age 0.0050920 1.0051050 0.0137288 0.371 0.71071 meal.cal -0.0002398 0.9997602 0.0003019 -0.794 0.42701 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 exp(coef) exp(-coef) lower .95 upper .95 ph.ecog 2.0751 0.4819 1.2249 3.5153 ph.karno 1.0131 0.9870 0.9862 1.0408 pat.karno 0.9860 1.0142 0.9681 1.0043 wt.loss 0.9852 1.0150 0.9690 1.0017 sex 0.5162 1.9372 0.3263 0.8166 age 1.0051 0.9949 0.9784 1.0325 meal.cal 0.9998 1.0002 0.9992 1.0004 Concordance= 0.696 (se = 0.115 ) Rsquare= 0.167 (max possible= 0.912 ) Likelihood ratio test= 30.6 on 7 df, p=7.377e-05 Wald test = 28.15 on 7 df, p=0.0002066 Score (logrank) test = 30.72 on 7 df, p=7e-05 > cox.zph(temp) rho chisq p ph.ecog 0.0276 0.1078 0.7427 ph.karno 0.1331 2.0018 0.1571 pat.karno 0.0250 0.0841 0.7718 wt.loss -0.0386 0.2122 0.6451 sex 0.0399 0.1800 0.6713 age 0.0639 0.5600 0.4543 meal.cal 0.1611 3.6945 0.0546 GLOBAL NA 9.0115 0.2518 > cox.zph(temp, transform='identity') rho chisq p ph.ecog 0.0221 0.0688 0.793 ph.karno 0.1217 1.6743 0.196 pat.karno 0.0302 0.1227 0.726 wt.loss -0.0516 0.3790 0.538 sex 0.0449 0.2280 0.633 age 0.0719 0.7085 0.400 meal.cal 0.1808 4.6537 0.031 GLOBAL NA 10.0537 0.186 > > coxph(Surv(rep(0,length(time)), time, status) ~ ph.ecog + ph.karno + pat.karno + + wt.loss + sex + age + meal.cal + strata(inst), lung) Call: coxph(formula = Surv(rep(0, length(time)), time, status) ~ ph.ecog + ph.karno + pat.karno + wt.loss + sex + age + meal.cal + strata(inst), data = lung) coef exp(coef) se(coef) z p ph.ecog 0.729999 2.075078 0.268940 2.71 0.0066 ph.karno 0.013051 1.013137 0.013736 0.95 0.3420 pat.karno -0.014095 0.986003 0.009368 -1.50 0.1324 wt.loss -0.014882 0.985228 0.008481 -1.75 0.0793 sex -0.661253 0.516204 0.233998 -2.83 0.0047 age 0.005092 1.005105 0.013729 0.37 0.7107 meal.cal -0.000240 0.999760 0.000302 -0.79 0.4270 Likelihood ratio test=30.6 on 7 df, p=7.38e-05 n= 167, number of events= 120 (61 observations deleted due to missingness) > > # > # Tests of using "." > # > fit1 <- coxph(Surv(time, status) ~ . - meal.cal - wt.loss - inst, lung) > fit2 <- update(fit1, .~. - ph.karno) > fit3 <- coxph(Surv(time, status) ~ age + sex + ph.ecog + pat.karno, lung) > all.equal(fit2, fit3) [1] TRUE > > proc.time() user system elapsed 0.292 0.016 0.302 survival/tests/plot1.pdf.save0000644000175100001440000015236212334223552016006 0ustar hornikusers%PDF-1.4 %âãÏÓ\r 1 0 obj << /CreationDate (D:20140512144403) /ModDate (D:20140512144403) /Title (R Graphics Output) /Producer (R 3.2.0) /Creator (R) >> endobj 2 0 obj << /Type /Catalog /Pages 3 0 R >> endobj 7 0 obj << /Type /Page /Parent 3 0 R /Contents 8 0 R /Resources 4 0 R >> endobj 8 0 obj << /Length 9 0 R >> stream 1 J 1 j q Q q /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 59.04 73.44 m 449.23 73.44 l S 59.04 73.44 m 59.04 66.24 l S 137.08 73.44 m 137.08 66.24 l S 215.11 73.44 m 215.11 66.24 l S 293.15 73.44 m 293.15 66.24 l S 371.19 73.44 m 371.19 66.24 l S 449.23 73.44 m 449.23 66.24 l S BT /sRGB cs 0.000 0.000 0.000 scn /F2 1 Tf 12.00 0.00 0.00 12.00 55.70 47.52 Tm (0) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 127.07 47.52 Tm (200) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 205.11 47.52 Tm (400) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 283.14 47.52 Tm (600) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 361.18 47.52 Tm (800) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 435.88 47.52 Tm (1000) Tj ET 59.04 87.20 m 59.04 431.20 l S 59.04 87.20 m 51.84 87.20 l S 59.04 156.00 m 51.84 156.00 l S 59.04 224.80 m 51.84 224.80 l S 59.04 293.60 m 51.84 293.60 l S 59.04 362.40 m 51.84 362.40 l S 59.04 431.20 m 51.84 431.20 l S BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 78.86 Tm (0.0) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 147.66 Tm (0.2) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 216.46 Tm (0.4) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 285.26 Tm (0.6) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 354.06 Tm (0.8) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 422.86 Tm (1.0) Tj ET 59.04 73.44 m 473.76 73.44 l 473.76 444.96 l 59.04 444.96 l 59.04 73.44 l S Q q Q q 59.04 73.44 414.72 371.52 re W n /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 59.04 431.20 m 60.99 431.20 l 60.99 425.74 l 63.33 425.74 l 63.33 420.28 l 64.89 420.28 l 64.89 414.82 l 71.14 414.82 l 71.14 409.36 l 79.72 409.36 l 79.72 403.90 l 84.40 403.90 l 84.40 398.44 l 90.64 398.44 l 90.64 392.98 l 116.40 392.98 l 116.40 387.52 l 123.81 387.52 l 123.81 382.06 l 127.71 382.06 l 127.71 376.49 l 132.78 376.49 l 132.78 370.93 l 137.47 370.93 l 137.47 365.26 l 155.03 365.26 l 155.03 358.94 l 163.22 358.94 l 163.22 352.14 l 170.24 352.14 l 170.24 344.78 l 170.63 344.78 l 170.63 337.42 l 177.27 337.42 l 177.27 329.60 l 183.90 329.60 l 183.90 321.52 l 190.53 321.52 l 190.53 313.16 l 191.70 313.16 l 191.70 304.79 l 194.82 304.79 l 194.82 296.42 l 195.60 296.42 l 195.60 288.05 l 196.78 288.05 l 196.78 271.31 l 203.80 271.31 l 203.80 262.94 l 212.77 262.94 l 212.77 253.69 l 226.04 253.69 l 226.04 244.44 l 227.99 244.44 l 227.99 235.19 l 231.50 235.19 l 231.50 225.94 l 236.57 225.94 l 236.57 216.03 l 276.76 216.03 l 276.76 205.30 l 283.01 205.30 l 283.01 193.49 l 309.93 193.49 l 309.93 180.20 l 314.61 180.20 l 314.61 166.92 l 334.12 166.92 l 334.12 153.63 l 367.68 153.63 l 367.68 140.34 l 403.57 140.34 l 403.57 113.77 l 453.13 113.77 l 453.13 113.77 l S BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 124.36 379.46 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 130.60 368.34 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 134.90 362.66 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 143.48 362.66 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 143.87 362.66 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 147.77 362.66 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 149.72 362.66 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 154.40 356.34 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 157.14 356.34 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 159.87 356.34 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 161.04 349.55 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 163.77 349.55 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 166.89 349.55 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 170.01 334.83 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 173.13 334.83 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 178.99 327.01 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 185.62 318.93 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 205.13 260.35 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 205.91 260.35 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 229.32 223.35 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 267.95 213.44 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 274.19 202.70 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 285.51 190.89 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 376.42 137.75 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 383.83 137.75 Tm (l) Tj 0 Tr ET BT /F1 1 Tf 1 Tr 7.48 0 0 7.48 450.17 111.18 Tm (l) Tj 0 Tr ET 59.04 431.20 m 63.33 431.20 l 63.33 428.16 l 64.11 428.16 l 64.11 425.11 l 82.06 425.11 l 82.06 422.07 l 82.45 422.07 l 82.45 415.98 l 83.23 415.98 l 83.23 412.93 l 89.86 412.93 l 89.86 409.89 l 93.38 409.89 l 93.38 403.80 l 94.94 403.80 l 94.94 400.76 l 96.11 400.76 l 96.11 397.71 l 100.79 397.71 l 100.79 394.67 l 101.96 394.67 l 101.96 391.62 l 104.30 391.62 l 104.30 388.58 l 110.15 388.58 l 110.15 385.54 l 110.54 385.54 l 110.54 382.49 l 111.71 382.49 l 111.71 379.45 l 114.45 379.45 l 114.45 376.40 l 115.23 376.40 l 115.23 373.36 l 115.62 373.36 l 115.62 370.32 l 119.91 370.32 l 119.91 367.27 l 122.64 367.27 l 122.64 361.18 l 124.20 361.18 l 124.20 358.14 l 125.37 358.14 l 125.37 355.09 l 127.32 355.09 l 127.32 351.98 l 128.88 351.98 l 128.88 345.67 l 129.66 345.67 l 129.66 339.37 l 130.05 339.37 l 130.05 336.22 l 131.61 336.22 l 131.61 333.03 l 134.74 333.03 l 134.74 329.75 l 135.91 329.75 l 135.91 326.43 l 137.86 326.43 l 137.86 323.06 l 139.81 323.06 l 139.81 319.64 l 140.98 319.64 l 140.98 316.22 l 144.10 316.22 l 144.10 312.80 l 146.05 312.80 l 146.05 309.28 l 147.22 309.28 l 147.22 305.69 l 148.39 305.69 l 148.39 302.11 l 148.78 302.11 l 148.78 298.53 l 152.29 298.53 l 152.29 294.89 l 154.64 294.89 l 154.64 291.18 l 163.61 291.18 l 163.61 287.47 l 164.00 287.47 l 164.00 283.76 l 164.39 283.76 l 164.39 280.05 l 169.46 280.05 l 169.46 276.20 l 169.85 276.20 l 169.85 272.34 l 173.36 272.34 l 173.36 268.48 l 176.49 268.48 l 176.49 264.54 l 178.05 264.54 l 178.05 260.42 l 178.44 260.42 l 178.44 256.29 l 180.00 256.29 l 180.00 252.17 l 193.65 252.17 l 193.65 248.04 l 200.68 248.04 l 200.68 239.58 l 201.07 239.58 l 201.07 235.35 l 203.80 235.35 l 203.80 230.99 l 211.21 230.99 l 211.21 226.49 l 225.26 226.49 l 225.26 221.69 l 226.43 221.69 l 226.43 216.89 l 234.62 216.89 l 234.62 212.08 l 237.35 212.08 l 237.35 207.28 l 238.53 207.28 l 238.53 202.28 l 243.60 202.28 l 243.60 197.27 l 245.16 197.27 l 245.16 192.27 l 261.55 192.27 l 261.55 187.02 l 261.94 187.02 l 261.94 181.76 l 263.50 181.76 l 263.50 176.51 l 273.64 176.51 l 273.64 170.93 l 280.27 170.93 l 280.27 165.35 l 286.52 165.35 l 286.52 159.76 l 298.22 159.76 l 298.22 154.18 l 302.52 154.18 l 302.52 148.60 l 309.15 148.60 l 309.15 143.02 l 327.10 143.02 l 327.10 137.44 l 327.88 137.44 l 327.88 131.86 l 343.09 131.86 l 343.09 126.27 l 344.27 126.27 l 344.27 120.69 l 345.83 120.69 l 345.83 115.11 l 357.53 115.11 l 357.53 108.13 l 457.81 108.13 l 457.81 108.13 l S 126.54 359.29 m 130.18 352.99 l 122.91 352.99 l 126.54 359.29 l S 126.93 359.29 m 130.57 352.99 l 123.30 352.99 l 126.93 359.29 l S 128.10 356.18 m 131.74 349.88 l 124.47 349.88 l 128.10 356.18 l S 131.22 340.42 m 134.86 334.12 l 127.59 334.12 l 131.22 340.42 l S 132.39 337.22 m 136.03 330.93 l 128.76 330.93 l 132.39 337.22 l S 133.96 337.22 m 137.59 330.93 l 130.32 330.93 l 133.96 337.22 l S 135.52 333.95 m 139.15 327.65 l 131.88 327.65 l 135.52 333.95 l S 135.91 330.62 m 139.54 324.33 l 132.27 324.33 l 135.91 330.62 l S 138.25 327.25 m 141.88 320.96 l 134.61 320.96 l 138.25 327.25 l S 145.27 317.00 m 148.91 310.70 l 141.63 310.70 l 145.27 317.00 l S 145.66 317.00 m 149.30 310.70 l 142.02 310.70 l 145.66 317.00 l S 146.83 313.48 m 150.47 307.18 l 143.20 307.18 l 146.83 313.48 l S 151.51 302.73 m 155.15 296.43 l 147.88 296.43 l 151.51 302.73 l S 153.85 299.09 m 157.49 292.79 l 150.22 292.79 l 153.85 299.09 l S 165.17 284.25 m 168.81 277.95 l 161.53 277.95 l 165.17 284.25 l S 167.90 284.25 m 171.54 277.95 l 164.27 277.95 l 167.90 284.25 l S 174.53 272.68 m 178.17 266.38 l 170.90 266.38 l 174.53 272.68 l S 176.49 268.74 m 180.12 262.44 l 172.85 262.44 l 176.49 268.74 l S 177.27 268.74 m 180.90 262.44 l 173.63 262.44 l 177.27 268.74 l S 197.95 252.24 m 201.58 245.94 l 194.31 245.94 l 197.95 252.24 l S 201.07 239.54 m 204.70 233.25 l 197.43 233.25 l 201.07 239.54 l S 205.75 235.19 m 209.39 228.89 l 202.11 228.89 l 205.75 235.19 l S 216.67 230.69 m 220.31 224.40 l 213.04 224.40 l 216.67 230.69 l S 220.19 230.69 m 223.82 224.40 l 216.55 224.40 l 220.19 230.69 l S 237.74 211.48 m 241.38 205.18 l 234.11 205.18 l 237.74 211.48 l S 258.42 196.47 m 262.06 190.17 l 254.79 190.17 l 258.42 196.47 l S 265.45 180.71 m 269.08 174.41 l 261.81 174.41 l 265.45 180.71 l S 347.78 119.31 m 351.41 113.01 l 344.14 113.01 l 347.78 119.31 l S 373.53 112.33 m 377.17 106.03 l 369.89 106.03 l 373.53 112.33 l S 435.57 112.33 m 439.20 106.03 l 431.93 106.03 l 435.57 112.33 l S 457.81 112.33 m 461.45 106.03 l 454.17 106.03 l 457.81 112.33 l S 59.04 431.20 m 63.33 431.20 l 63.33 424.32 l 63.72 424.32 l 63.72 417.44 l 64.11 417.44 l 64.11 410.56 l 69.18 410.56 l 69.18 403.68 l 70.75 403.68 l 70.75 396.80 l 79.72 396.80 l 79.72 389.92 l 80.11 389.92 l 80.11 383.04 l 82.84 383.04 l 82.84 376.16 l 84.40 376.16 l 84.40 369.28 l 90.64 369.28 l 90.64 362.40 l 95.33 362.40 l 95.33 355.34 l 96.11 355.34 l 96.11 348.29 l 100.01 348.29 l 100.01 341.23 l 100.79 341.23 l 100.79 333.97 l 106.64 333.97 l 106.64 326.71 l 110.54 326.71 l 110.54 319.46 l 115.62 319.46 l 115.62 312.20 l 118.74 312.20 l 118.74 304.94 l 119.91 304.94 l 119.91 297.68 l 122.64 297.68 l 122.64 290.42 l 123.81 290.42 l 123.81 283.17 l 128.10 283.17 l 128.10 275.91 l 129.27 275.91 l 129.27 268.65 l 130.44 268.65 l 130.44 261.39 l 136.69 261.39 l 136.69 254.13 l 137.47 254.13 l 137.47 246.88 l 140.20 246.88 l 140.20 239.62 l 141.76 239.62 l 141.76 232.00 l 145.66 232.00 l 145.66 224.38 l 152.29 224.38 l 152.29 216.76 l 170.24 216.76 l 170.24 209.13 l 171.41 209.13 l 171.41 201.51 l 172.58 201.51 l 172.58 193.89 l 180.00 193.89 l 180.00 185.69 l 187.41 185.69 l 187.41 177.48 l 195.99 177.48 l 195.99 169.27 l 199.90 169.27 l 199.90 161.06 l 210.04 161.06 l 210.04 152.86 l 232.28 152.86 l 232.28 144.65 l 263.50 144.65 l 263.50 135.08 l 267.01 135.08 l 267.01 125.50 l 314.22 125.50 l 314.22 112.73 l 334.90 112.73 l 334.90 99.97 l 376.65 99.97 l 376.65 87.20 l S 91.12 362.40 m 98.76 362.40 l S 94.94 358.58 m 94.94 366.22 l S 96.19 341.23 m 103.83 341.23 l S 100.01 337.41 m 100.01 345.05 l S 137.55 239.62 m 145.19 239.62 l S 141.37 235.80 m 141.37 243.44 l S 169.16 193.89 m 176.79 193.89 l S 172.97 190.07 m 172.97 197.71 l S 254.61 144.65 m 262.24 144.65 l S 258.42 140.83 m 258.42 148.47 l S 270.21 125.50 m 277.85 125.50 l S 274.03 121.68 m 274.03 129.32 l S 59.04 431.20 m 105.08 431.20 l 105.08 87.20 l S Q endstream endobj 9 0 obj 10658 endobj 10 0 obj << /Type /Page /Parent 3 0 R /Contents 11 0 R /Resources 4 0 R >> endobj 11 0 obj << /Length 12 0 R >> stream 1 J 1 j q Q q 59.04 73.44 414.72 371.52 re W n Q q /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 74.40 73.44 m 450.13 73.44 l S 74.40 73.44 m 74.40 66.24 l S 149.55 73.44 m 149.55 66.24 l S 224.69 73.44 m 224.69 66.24 l S 299.84 73.44 m 299.84 66.24 l S 374.99 73.44 m 374.99 66.24 l S 450.13 73.44 m 450.13 66.24 l S BT /sRGB cs 0.000 0.000 0.000 scn /F2 1 Tf 12.00 0.00 0.00 12.00 71.06 47.52 Tm (0) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 139.54 47.52 Tm (200) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 214.69 47.52 Tm (400) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 289.83 47.52 Tm (600) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 364.98 47.52 Tm (800) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 436.79 47.52 Tm (1000) Tj ET 59.04 87.20 m 59.04 431.20 l S 59.04 87.20 m 51.84 87.20 l S 59.04 156.00 m 51.84 156.00 l S 59.04 224.80 m 51.84 224.80 l S 59.04 293.60 m 51.84 293.60 l S 59.04 362.40 m 51.84 362.40 l S 59.04 431.20 m 51.84 431.20 l S BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 78.86 Tm (0.0) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 147.66 Tm (0.2) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 216.46 Tm (0.4) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 285.26 Tm (0.6) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 354.06 Tm (0.8) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 422.86 Tm (1.0) Tj ET 59.04 73.44 m 473.76 73.44 l 473.76 444.96 l 59.04 444.96 l 59.04 73.44 l S Q q Q q 59.04 73.44 414.72 371.52 re W n /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 74.40 87.20 m 76.28 87.20 l 76.28 92.66 l 78.53 92.66 l 78.53 98.12 l 80.04 98.12 l 80.04 103.58 l 86.05 103.58 l 86.05 109.04 l 94.31 109.04 l 94.31 114.50 l 98.82 114.50 l 98.82 119.96 l 104.83 119.96 l 104.83 125.42 l 129.63 125.42 l 129.63 130.88 l 136.77 130.88 l 136.77 136.34 l 140.53 136.34 l 140.53 141.91 l 145.41 141.91 l 145.41 147.47 l 149.92 147.47 l 149.92 153.14 l 166.83 153.14 l 166.83 159.46 l 174.72 159.46 l 174.72 166.26 l 181.48 166.26 l 181.48 173.62 l 181.86 173.62 l 181.86 180.98 l 188.25 180.98 l 188.25 188.80 l 194.63 188.80 l 194.63 196.88 l 201.02 196.88 l 201.02 205.24 l 202.15 205.24 l 202.15 213.61 l 205.16 213.61 l 205.16 221.98 l 205.91 221.98 l 205.91 230.35 l 207.03 230.35 l 207.03 247.09 l 213.80 247.09 l 213.80 255.46 l 222.44 255.46 l 222.44 264.71 l 235.21 264.71 l 235.21 273.96 l 237.09 273.96 l 237.09 283.21 l 240.47 283.21 l 240.47 292.46 l 245.36 292.46 l 245.36 302.37 l 284.06 302.37 l 284.06 313.10 l 290.07 313.10 l 290.07 324.91 l 316.00 324.91 l 316.00 338.20 l 320.51 338.20 l 320.51 351.48 l 339.29 351.48 l 339.29 364.77 l 371.61 364.77 l 371.61 378.06 l 406.17 378.06 l 406.17 404.63 l 453.89 404.63 l 453.89 404.63 l S 136.34 136.34 m 143.97 136.34 l S 140.15 132.52 m 140.15 140.16 l S 142.35 147.47 m 149.98 147.47 l S 146.17 143.65 m 146.17 151.29 l S 146.48 153.14 m 154.12 153.14 l S 150.30 149.33 m 150.30 156.96 l S 154.75 153.14 m 162.38 153.14 l S 158.56 149.33 m 158.56 156.96 l S 155.12 153.14 m 162.76 153.14 l S 158.94 149.33 m 158.94 156.96 l S 158.88 153.14 m 166.52 153.14 l S 162.70 149.33 m 162.70 156.96 l S 160.76 153.14 m 168.39 153.14 l S 164.58 149.33 m 164.58 156.96 l S 165.27 159.46 m 172.90 159.46 l S 169.08 155.65 m 169.08 163.28 l S 167.90 159.46 m 175.53 159.46 l S 171.72 155.65 m 171.72 163.28 l S 170.53 159.46 m 178.16 159.46 l S 174.35 155.65 m 174.35 163.28 l S 171.65 166.26 m 179.29 166.26 l S 175.47 162.44 m 175.47 170.08 l S 174.28 166.26 m 181.92 166.26 l S 178.10 162.44 m 178.10 170.08 l S 177.29 166.26 m 184.93 166.26 l S 181.11 162.44 m 181.11 170.08 l S 180.30 180.98 m 187.93 180.98 l S 184.11 177.16 m 184.11 184.79 l S 183.30 180.98 m 190.94 180.98 l S 187.12 177.16 m 187.12 184.79 l S 188.94 188.80 m 196.57 188.80 l S 192.76 184.98 m 192.76 192.61 l S 195.33 196.88 m 202.96 196.88 l S 199.14 193.06 m 199.14 200.69 l S 214.11 255.46 m 221.75 255.46 l S 217.93 251.64 m 217.93 259.28 l S 214.86 255.46 m 222.50 255.46 l S 218.68 251.64 m 218.68 259.28 l S 237.41 292.46 m 245.04 292.46 l S 241.23 288.64 m 241.23 296.27 l S 274.61 302.37 m 282.24 302.37 l S 278.42 298.55 m 278.42 306.18 l S 280.62 313.10 m 288.25 313.10 l S 284.44 309.28 m 284.44 316.92 l S 291.51 324.91 m 299.15 324.91 l S 295.33 321.09 m 295.33 328.73 l S 379.06 378.06 m 386.70 378.06 l S 382.88 374.24 m 382.88 381.87 l S 386.20 378.06 m 393.83 378.06 l S 390.02 374.24 m 390.02 381.87 l S 450.07 404.63 m 457.71 404.63 l S 453.89 400.81 m 453.89 408.45 l S 0.75 w [ 2.25 3.75] 0 d 74.40 87.20 m 76.28 87.20 l 76.28 103.11 l 78.53 103.11 l 78.53 112.69 l 80.04 112.69 l 80.04 121.18 l 86.05 121.18 l 86.05 129.10 l 94.31 129.10 l 94.31 136.65 l 98.82 136.65 l 98.82 143.92 l 104.83 143.92 l 104.83 150.99 l 129.63 150.99 l 129.63 157.87 l 136.77 157.87 l 136.77 164.62 l 140.53 164.62 l 140.53 171.39 l 145.41 171.39 l 145.41 178.05 l 149.92 178.05 l 149.92 184.76 l 166.83 184.76 l 166.83 192.32 l 174.72 192.32 l 174.72 200.43 l 181.48 200.43 l 181.48 209.21 l 181.86 209.21 l 181.86 217.80 l 188.25 217.80 l 188.25 226.88 l 194.63 226.88 l 194.63 236.14 l 201.02 236.14 l 201.02 245.60 l 202.15 245.60 l 202.15 254.86 l 205.16 254.86 l 205.16 263.94 l 205.91 263.94 l 205.91 272.84 l 207.03 272.84 l 207.03 290.17 l 213.80 290.17 l 213.80 298.60 l 222.44 298.60 l 222.44 307.98 l 235.21 307.98 l 235.21 317.15 l 237.09 317.15 l 237.09 326.12 l 240.47 326.12 l 240.47 334.89 l 245.36 334.89 l 245.36 344.23 l 284.06 344.23 l 284.06 354.25 l 290.07 354.25 l 290.07 365.14 l 316.00 365.14 l 316.00 377.16 l 320.51 377.16 l 320.51 388.38 l 339.29 388.38 l 339.29 398.77 l 371.61 398.77 l 371.61 408.27 l 406.17 408.27 l 406.17 425.95 l 453.89 425.95 l 453.89 425.95 l S 74.40 87.20 m 86.05 87.20 l 86.05 87.65 l 94.31 87.65 l 94.31 90.69 l 98.82 90.69 l 98.82 94.00 l 104.83 94.00 l 104.83 97.53 l 129.63 97.53 l 129.63 101.23 l 136.77 101.23 l 136.77 105.07 l 140.53 105.07 l 140.53 109.07 l 145.41 109.07 l 145.41 113.20 l 149.92 113.20 l 149.92 117.47 l 166.83 117.47 l 166.83 122.09 l 174.72 122.09 l 174.72 127.03 l 181.48 127.03 l 181.48 132.32 l 181.86 132.32 l 181.86 137.80 l 188.25 137.80 l 188.25 143.61 l 194.63 143.61 l 194.63 149.71 l 201.02 149.71 l 201.02 156.11 l 202.15 156.11 l 202.15 162.71 l 205.16 162.71 l 205.16 169.50 l 205.91 169.50 l 205.91 176.46 l 207.03 176.46 l 207.03 190.85 l 213.80 190.85 l 213.80 198.27 l 222.44 198.27 l 222.44 206.23 l 235.21 206.23 l 235.21 214.40 l 237.09 214.40 l 237.09 222.77 l 240.47 222.77 l 240.47 231.32 l 245.36 231.32 l 245.36 240.36 l 284.06 240.36 l 284.06 249.95 l 290.07 249.95 l 290.07 260.19 l 316.00 260.19 l 316.00 271.14 l 320.51 271.14 l 320.51 282.79 l 339.29 282.79 l 339.29 295.11 l 371.61 295.11 l 371.61 308.04 l 406.17 308.04 l 406.17 296.82 l 453.89 296.82 l 453.89 296.82 l S 1.000 0.000 0.000 SCN 0.75 w [] 0 d 74.40 87.20 m 78.53 87.20 l 78.53 90.24 l 79.28 90.24 l 79.28 93.29 l 96.57 93.29 l 96.57 96.33 l 96.94 96.33 l 96.94 102.42 l 97.70 102.42 l 97.70 105.47 l 104.08 105.47 l 104.08 108.51 l 107.46 108.51 l 107.46 114.60 l 108.97 114.60 l 108.97 117.64 l 110.09 117.64 l 110.09 120.69 l 114.60 120.69 l 114.60 123.73 l 115.73 123.73 l 115.73 126.78 l 117.99 126.78 l 117.99 129.82 l 123.62 129.82 l 123.62 132.86 l 124.00 132.86 l 124.00 135.91 l 125.12 135.91 l 125.12 138.95 l 127.75 138.95 l 127.75 142.00 l 128.51 142.00 l 128.51 145.04 l 128.88 145.04 l 128.88 148.08 l 133.01 148.08 l 133.01 151.13 l 135.64 151.13 l 135.64 157.22 l 137.15 157.22 l 137.15 160.26 l 138.27 160.26 l 138.27 163.31 l 140.15 163.31 l 140.15 166.42 l 141.66 166.42 l 141.66 172.73 l 142.41 172.73 l 142.41 179.03 l 142.78 179.03 l 142.78 182.18 l 144.29 182.18 l 144.29 185.37 l 147.29 185.37 l 147.29 188.65 l 148.42 188.65 l 148.42 191.97 l 150.30 191.97 l 150.30 195.34 l 152.18 195.34 l 152.18 198.76 l 153.30 198.76 l 153.30 202.18 l 156.31 202.18 l 156.31 205.60 l 158.19 205.60 l 158.19 209.12 l 159.32 209.12 l 159.32 212.71 l 160.44 212.71 l 160.44 216.29 l 160.82 216.29 l 160.82 219.87 l 164.20 219.87 l 164.20 223.51 l 166.45 223.51 l 166.45 227.22 l 175.10 227.22 l 175.10 230.93 l 175.47 230.93 l 175.47 234.64 l 175.85 234.64 l 175.85 238.35 l 180.73 238.35 l 180.73 242.20 l 181.11 242.20 l 181.11 246.06 l 184.49 246.06 l 184.49 249.92 l 187.50 249.92 l 187.50 253.86 l 189.00 253.86 l 189.00 257.98 l 189.37 257.98 l 189.37 262.11 l 190.88 262.11 l 190.88 266.23 l 204.03 266.23 l 204.03 270.36 l 210.79 270.36 l 210.79 278.82 l 211.17 278.82 l 211.17 283.05 l 213.80 283.05 l 213.80 287.41 l 220.94 287.41 l 220.94 291.91 l 234.46 291.91 l 234.46 296.71 l 235.59 296.71 l 235.59 301.51 l 243.48 301.51 l 243.48 306.32 l 246.11 306.32 l 246.11 311.12 l 247.24 311.12 l 247.24 316.12 l 252.12 316.12 l 252.12 321.13 l 253.63 321.13 l 253.63 326.13 l 269.41 326.13 l 269.41 331.38 l 269.78 331.38 l 269.78 336.64 l 271.28 336.64 l 271.28 341.89 l 281.05 341.89 l 281.05 347.47 l 287.44 347.47 l 287.44 353.05 l 293.45 353.05 l 293.45 358.64 l 304.72 358.64 l 304.72 364.22 l 308.86 364.22 l 308.86 369.80 l 315.25 369.80 l 315.25 375.38 l 332.53 375.38 l 332.53 380.96 l 333.28 380.96 l 333.28 386.54 l 347.93 386.54 l 347.93 392.13 l 349.06 392.13 l 349.06 397.71 l 350.56 397.71 l 350.56 403.29 l 361.84 403.29 l 361.84 410.27 l 458.40 410.27 l 458.40 410.27 l S 135.58 163.31 m 143.22 163.31 l S 139.40 159.49 m 139.40 167.12 l S 135.96 163.31 m 143.60 163.31 l S 139.78 159.49 m 139.78 167.12 l S 137.09 166.42 m 144.72 166.42 l S 140.90 162.60 m 140.90 170.24 l S 140.09 182.18 m 147.73 182.18 l S 143.91 178.36 m 143.91 186.00 l S 141.22 185.37 m 148.86 185.37 l S 145.04 181.56 m 145.04 189.19 l S 142.72 185.37 m 150.36 185.37 l S 146.54 181.56 m 146.54 189.19 l S 144.23 188.65 m 151.86 188.65 l S 148.04 184.83 m 148.04 192.47 l S 144.60 191.97 m 152.24 191.97 l S 148.42 188.16 m 148.42 195.79 l S 146.86 195.34 m 154.49 195.34 l S 150.67 191.53 m 150.67 199.16 l S 153.62 205.60 m 161.26 205.60 l S 157.44 201.78 m 157.44 209.42 l S 153.99 205.60 m 161.63 205.60 l S 157.81 201.78 m 157.81 209.42 l S 155.12 209.12 m 162.76 209.12 l S 158.94 205.31 m 158.94 212.94 l S 159.63 219.87 m 167.27 219.87 l S 163.45 216.05 m 163.45 223.69 l S 161.88 223.51 m 169.52 223.51 l S 165.70 219.69 m 165.70 227.33 l S 172.78 238.35 m 180.42 238.35 l S 176.60 234.53 m 176.60 242.17 l S 175.41 238.35 m 183.05 238.35 l S 179.23 234.53 m 179.23 242.17 l S 181.80 249.92 m 189.44 249.92 l S 185.62 246.10 m 185.62 253.74 l S 183.68 253.86 m 191.31 253.86 l S 187.50 250.04 m 187.50 257.68 l S 184.43 253.86 m 192.07 253.86 l S 188.25 250.04 m 188.25 257.68 l S 204.34 270.36 m 211.98 270.36 l S 208.16 266.54 m 208.16 274.17 l S 207.35 283.05 m 214.99 283.05 l S 211.17 279.24 m 211.17 286.87 l S 211.86 287.41 m 219.49 287.41 l S 215.68 283.59 m 215.68 291.23 l S 222.38 291.91 m 230.01 291.91 l S 226.20 288.09 m 226.20 295.72 l S 225.76 291.91 m 233.40 291.91 l S 229.58 288.09 m 229.58 295.72 l S 242.67 311.12 m 250.30 311.12 l S 246.49 307.30 m 246.49 314.94 l S 262.58 326.13 m 270.22 326.13 l S 266.40 322.31 m 266.40 329.95 l S 269.34 341.89 m 276.98 341.89 l S 273.16 338.07 m 273.16 345.71 l S 348.62 403.29 m 356.26 403.29 l S 352.44 399.47 m 352.44 407.11 l S 373.42 410.27 m 381.06 410.27 l S 377.24 406.45 m 377.24 414.09 l S 433.16 410.27 m 440.80 410.27 l S 436.98 406.45 m 436.98 414.09 l S 454.58 410.27 m 462.22 410.27 l S 458.40 406.45 m 458.40 414.09 l S 0.75 w [ 2.25 3.75] 0 d 74.40 87.20 m 78.53 87.20 l 78.53 96.13 l 79.28 96.13 l 79.28 101.55 l 96.57 101.55 l 96.57 106.38 l 96.94 106.38 l 96.94 115.21 l 97.70 115.21 l 97.70 119.38 l 104.08 119.38 l 104.08 123.44 l 107.46 123.44 l 107.46 131.31 l 108.97 131.31 l 108.97 135.15 l 110.09 135.15 l 110.09 138.93 l 114.60 138.93 l 114.60 142.66 l 115.73 142.66 l 115.73 146.35 l 117.99 146.35 l 117.99 150.01 l 123.62 150.01 l 123.62 153.63 l 124.00 153.63 l 124.00 157.21 l 125.12 157.21 l 125.12 160.77 l 127.75 160.77 l 127.75 164.30 l 128.51 164.30 l 128.51 167.80 l 128.88 167.80 l 128.88 171.29 l 133.01 171.29 l 133.01 174.75 l 135.64 174.75 l 135.64 181.60 l 137.15 181.60 l 137.15 185.00 l 138.27 185.00 l 138.27 188.38 l 140.15 188.38 l 140.15 191.84 l 141.66 191.84 l 141.66 198.79 l 142.41 198.79 l 142.41 205.67 l 142.78 205.67 l 142.78 209.09 l 144.29 209.09 l 144.29 212.54 l 147.29 212.54 l 147.29 216.09 l 148.42 216.09 l 148.42 219.67 l 150.30 219.67 l 150.30 223.30 l 152.18 223.30 l 152.18 226.97 l 153.30 226.97 l 153.30 230.62 l 156.31 230.62 l 156.31 234.25 l 158.19 234.25 l 158.19 238.00 l 159.32 238.00 l 159.32 241.79 l 160.44 241.79 l 160.44 245.57 l 160.82 245.57 l 160.82 249.33 l 164.20 249.33 l 164.20 253.15 l 166.45 253.15 l 166.45 257.03 l 175.10 257.03 l 175.10 260.89 l 175.47 260.89 l 175.47 264.72 l 175.85 264.72 l 175.85 268.54 l 180.73 268.54 l 180.73 272.51 l 181.11 272.51 l 181.11 276.47 l 184.49 276.47 l 184.49 280.40 l 187.50 280.40 l 187.50 284.41 l 189.00 284.41 l 189.00 288.62 l 189.37 288.62 l 189.37 292.80 l 190.88 292.80 l 190.88 296.95 l 204.03 296.95 l 204.03 301.07 l 210.79 301.07 l 210.79 309.50 l 211.17 309.50 l 211.17 313.67 l 213.80 313.67 l 213.80 317.95 l 220.94 317.95 l 220.94 322.36 l 234.46 322.36 l 234.46 327.10 l 235.59 327.10 l 235.59 331.79 l 243.48 331.79 l 243.48 336.43 l 246.11 336.43 l 246.11 341.03 l 247.24 341.03 l 247.24 345.81 l 252.12 345.81 l 252.12 350.53 l 253.63 350.53 l 253.63 355.20 l 269.41 355.20 l 269.41 360.09 l 269.78 360.09 l 269.78 364.92 l 271.28 364.92 l 271.28 369.67 l 281.05 369.67 l 281.05 374.71 l 287.44 374.71 l 287.44 379.65 l 293.45 379.65 l 293.45 384.50 l 304.72 384.50 l 304.72 389.24 l 308.86 389.24 l 308.86 393.88 l 315.25 393.88 l 315.25 398.41 l 332.53 398.41 l 332.53 402.81 l 333.28 402.81 l 333.28 407.07 l 347.93 407.07 l 347.93 411.19 l 349.06 411.19 l 349.06 415.13 l 350.56 415.13 l 350.56 418.87 l 361.84 418.87 l 361.84 423.45 l 458.40 423.45 l 458.40 423.45 l S 74.40 87.20 m 96.94 87.20 l 96.94 89.12 l 97.70 89.12 l 97.70 90.93 l 104.08 90.93 l 104.08 92.85 l 107.46 92.85 l 107.46 96.95 l 108.97 96.95 l 108.97 99.10 l 110.09 99.10 l 110.09 101.30 l 114.60 101.30 l 114.60 103.56 l 115.73 103.56 l 115.73 105.85 l 117.99 105.85 l 117.99 108.18 l 123.62 108.18 l 123.62 110.55 l 124.00 110.55 l 124.00 112.95 l 125.12 112.95 l 125.12 115.37 l 127.75 115.37 l 127.75 117.83 l 128.51 117.83 l 128.51 120.31 l 128.88 120.31 l 128.88 122.81 l 133.01 122.81 l 133.01 125.34 l 135.64 125.34 l 135.64 130.45 l 137.15 130.45 l 137.15 133.04 l 138.27 133.04 l 138.27 135.64 l 140.15 135.64 l 140.15 138.31 l 141.66 138.31 l 141.66 143.74 l 142.41 143.74 l 142.41 149.24 l 142.78 149.24 l 142.78 152.01 l 144.29 152.01 l 144.29 154.83 l 147.29 154.83 l 147.29 157.72 l 148.42 157.72 l 148.42 160.65 l 150.30 160.65 l 150.30 163.63 l 152.18 163.63 l 152.18 166.66 l 153.30 166.66 l 153.30 169.71 l 156.31 169.71 l 156.31 172.78 l 158.19 172.78 l 158.19 175.94 l 159.32 175.94 l 159.32 179.15 l 160.44 179.15 l 160.44 182.38 l 160.82 182.38 l 160.82 185.63 l 164.20 185.63 l 164.20 188.94 l 166.45 188.94 l 166.45 192.31 l 175.10 192.31 l 175.10 195.71 l 175.47 195.71 l 175.47 199.12 l 175.85 199.12 l 175.85 202.55 l 180.73 202.55 l 180.73 206.11 l 181.11 206.11 l 181.11 209.68 l 184.49 209.68 l 184.49 213.28 l 187.50 213.28 l 187.50 216.95 l 189.00 216.95 l 189.00 220.77 l 189.37 220.77 l 189.37 224.62 l 190.88 224.62 l 190.88 228.49 l 204.03 228.49 l 204.03 232.39 l 210.79 232.39 l 210.79 240.42 l 211.17 240.42 l 211.17 244.47 l 213.80 244.47 l 213.80 248.64 l 220.94 248.64 l 220.94 252.93 l 234.46 252.93 l 234.46 257.45 l 235.59 257.45 l 235.59 262.01 l 243.48 262.01 l 243.48 266.63 l 246.11 266.63 l 246.11 271.29 l 247.24 271.29 l 247.24 276.11 l 252.12 276.11 l 252.12 281.00 l 253.63 281.00 l 253.63 285.93 l 269.41 285.93 l 269.41 291.08 l 269.78 291.08 l 269.78 296.29 l 271.28 296.29 l 271.28 301.57 l 281.05 301.57 l 281.05 307.10 l 287.44 307.10 l 287.44 312.73 l 293.45 312.73 l 293.45 318.45 l 304.72 318.45 l 304.72 324.26 l 308.86 324.26 l 308.86 330.17 l 315.25 330.17 l 315.25 336.19 l 332.53 336.19 l 332.53 342.32 l 333.28 342.32 l 333.28 348.56 l 347.93 348.56 l 347.93 354.92 l 349.06 354.92 l 349.06 361.41 l 350.56 361.41 l 350.56 368.04 l 361.84 368.04 l 361.84 374.67 l 458.40 374.67 l 458.40 374.67 l S 0.000 0.804 0.000 SCN 0.75 w [] 0 d 74.40 87.20 m 78.53 87.20 l 78.53 94.08 l 78.91 94.08 l 78.91 100.96 l 79.28 100.96 l 79.28 107.84 l 84.17 107.84 l 84.17 114.72 l 85.67 114.72 l 85.67 121.60 l 94.31 121.60 l 94.31 128.48 l 94.69 128.48 l 94.69 135.36 l 97.32 135.36 l 97.32 142.24 l 98.82 142.24 l 98.82 149.12 l 104.83 149.12 l 104.83 156.00 l 109.34 156.00 l 109.34 163.06 l 110.09 163.06 l 110.09 170.11 l 113.85 170.11 l 113.85 177.17 l 114.60 177.17 l 114.60 184.43 l 120.24 184.43 l 120.24 191.69 l 124.00 191.69 l 124.00 198.94 l 128.88 198.94 l 128.88 206.20 l 131.89 206.20 l 131.89 213.46 l 133.01 213.46 l 133.01 220.72 l 135.64 220.72 l 135.64 227.98 l 136.77 227.98 l 136.77 235.23 l 140.90 235.23 l 140.90 242.49 l 142.03 242.49 l 142.03 249.75 l 143.16 249.75 l 143.16 257.01 l 149.17 257.01 l 149.17 264.27 l 149.92 264.27 l 149.92 271.52 l 152.55 271.52 l 152.55 278.78 l 154.06 278.78 l 154.06 286.40 l 157.81 286.40 l 157.81 294.02 l 164.20 294.02 l 164.20 301.64 l 181.48 301.64 l 181.48 309.27 l 182.61 309.27 l 182.61 316.89 l 183.74 316.89 l 183.74 324.51 l 190.88 324.51 l 190.88 332.71 l 198.02 332.71 l 198.02 340.92 l 206.28 340.92 l 206.28 349.13 l 210.04 349.13 l 210.04 357.34 l 219.81 357.34 l 219.81 365.54 l 241.23 365.54 l 241.23 373.75 l 271.28 373.75 l 271.28 383.32 l 274.67 383.32 l 274.67 392.90 l 320.13 392.90 l 320.13 405.67 l 340.04 405.67 l 340.04 418.43 l 380.25 418.43 l 380.25 431.20 l S 105.15 156.00 m 112.79 156.00 l S 108.97 152.18 m 108.97 159.82 l S 110.03 177.17 m 117.67 177.17 l S 113.85 173.35 m 113.85 180.99 l S 149.86 278.78 m 157.50 278.78 l S 153.68 274.96 m 153.68 282.60 l S 180.30 324.51 m 187.93 324.51 l S 184.11 320.69 m 184.11 328.33 l S 262.58 373.75 m 270.22 373.75 l S 266.40 369.93 m 266.40 377.57 l S 277.61 392.90 m 285.25 392.90 l S 281.43 389.08 m 281.43 396.72 l S 0.75 w [ 2.25 3.75] 0 d 74.40 87.20 m 78.53 87.20 l 78.53 107.17 l 78.91 107.17 l 78.91 119.13 l 79.28 119.13 l 79.28 129.71 l 84.17 129.71 l 84.17 139.56 l 85.67 139.56 l 85.67 148.92 l 94.31 148.92 l 94.31 157.93 l 94.69 157.93 l 94.69 166.66 l 97.32 166.66 l 97.32 175.16 l 98.82 175.16 l 98.82 183.47 l 104.83 183.47 l 104.83 191.62 l 109.34 191.62 l 109.34 199.86 l 110.09 199.86 l 110.09 207.96 l 113.85 207.96 l 113.85 215.93 l 114.60 215.93 l 114.60 224.06 l 120.24 224.06 l 120.24 232.05 l 124.00 232.05 l 124.00 239.93 l 128.88 239.93 l 128.88 247.70 l 131.89 247.70 l 131.89 255.37 l 133.01 255.37 l 133.01 262.94 l 135.64 262.94 l 135.64 270.41 l 136.77 270.41 l 136.77 277.79 l 140.90 277.79 l 140.90 285.08 l 142.03 285.08 l 142.03 292.28 l 143.16 292.28 l 143.16 299.39 l 149.17 299.39 l 149.17 306.41 l 149.92 306.41 l 149.92 313.35 l 152.55 313.35 l 152.55 320.21 l 154.06 320.21 l 154.06 327.38 l 157.81 327.38 l 157.81 334.45 l 164.20 334.45 l 164.20 341.42 l 181.48 341.42 l 181.48 348.27 l 182.61 348.27 l 182.61 355.01 l 183.74 355.01 l 183.74 361.63 l 190.88 361.63 l 190.88 368.75 l 198.02 368.75 l 198.02 375.70 l 206.28 375.70 l 206.28 382.46 l 210.04 382.46 l 210.04 389.03 l 219.81 389.03 l 219.81 395.39 l 241.23 395.39 l 241.23 401.52 l 271.28 401.52 l 271.28 408.61 l 274.67 408.61 l 274.67 415.15 l 320.13 415.15 l 320.13 423.37 l 340.04 423.37 l 340.04 429.13 l 380.25 429.13 l S 74.40 87.20 m 84.17 87.20 l 84.17 87.77 l 85.67 87.77 l 85.67 91.63 l 94.31 91.63 l 94.31 95.85 l 94.69 95.85 l 94.69 100.35 l 97.32 100.35 l 97.32 105.08 l 98.82 105.08 l 98.82 110.00 l 104.83 110.00 l 104.83 115.09 l 109.34 115.09 l 109.34 120.39 l 110.09 120.39 l 110.09 125.84 l 113.85 125.84 l 113.85 131.43 l 114.60 131.43 l 114.60 137.22 l 120.24 137.22 l 120.24 143.13 l 124.00 143.13 l 124.00 149.17 l 128.88 149.17 l 128.88 155.31 l 131.89 155.31 l 131.89 161.56 l 133.01 161.56 l 133.01 167.90 l 135.64 167.90 l 135.64 174.34 l 136.77 174.34 l 136.77 180.87 l 140.90 180.87 l 140.90 187.50 l 142.03 187.50 l 142.03 194.21 l 143.16 194.21 l 143.16 201.00 l 149.17 201.00 l 149.17 207.88 l 149.92 207.88 l 149.92 214.85 l 152.55 214.85 l 152.55 221.90 l 154.06 221.90 l 154.06 229.24 l 157.81 229.24 l 157.81 236.70 l 164.20 236.70 l 164.20 244.25 l 181.48 244.25 l 181.48 251.91 l 182.61 251.91 l 182.61 259.69 l 183.74 259.69 l 183.74 267.57 l 190.88 267.57 l 190.88 275.88 l 198.02 275.88 l 198.02 284.36 l 206.28 284.36 l 206.28 293.00 l 210.04 293.00 l 210.04 301.81 l 219.81 301.81 l 219.81 310.81 l 241.23 310.81 l 241.23 320.00 l 271.28 320.00 l 271.28 329.73 l 274.67 329.73 l 274.67 339.81 l 320.13 339.81 l 320.13 347.96 l 340.04 347.96 l 340.04 352.30 l 380.25 352.30 l S 0.000 0.000 1.000 SCN 0.75 w [] 0 d 74.40 87.20 m 118.74 87.20 l 118.74 431.20 l S 0.75 w [ 2.25 3.75] 0 d 74.40 87.20 m 118.74 87.20 l S 74.40 87.20 m 118.74 87.20 l S Q endstream endobj 12 0 obj 20952 endobj 13 0 obj << /Type /Page /Parent 3 0 R /Contents 14 0 R /Resources 4 0 R >> endobj 14 0 obj << /Length 15 0 R >> stream 1 J 1 j q Q q 59.04 73.44 414.72 371.52 re W n Q q /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 74.40 73.44 m 456.83 73.44 l S 74.40 73.44 m 74.40 66.24 l S 124.43 73.44 m 124.43 66.24 l S 174.46 73.44 m 174.46 66.24 l S 240.60 73.44 m 240.60 66.24 l S 290.63 73.44 m 290.63 66.24 l S 340.66 73.44 m 340.66 66.24 l S 406.80 73.44 m 406.80 66.24 l S 456.83 73.44 m 456.83 66.24 l S BT /sRGB cs 0.000 0.000 0.000 scn /F2 1 Tf 12.00 0.00 0.00 12.00 71.06 47.52 Tm (5) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 117.76 47.52 Tm (10) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 167.79 47.52 Tm (20) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 233.93 47.52 Tm (50) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 280.62 47.52 Tm (100) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 330.65 47.52 Tm (200) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 396.79 47.52 Tm (500) Tj ET BT /F2 1 Tf 12.00 0.00 0.00 12.00 443.49 47.52 Tm (1000) Tj ET 59.04 129.24 m 59.04 420.03 l S 59.04 129.24 m 51.84 129.24 l S 59.04 187.40 m 51.84 187.40 l S 59.04 245.56 m 51.84 245.56 l S 59.04 303.72 m 51.84 303.72 l S 59.04 361.87 m 51.84 361.87 l S 59.04 420.03 m 51.84 420.03 l S BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 122.40 Tm (-4) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 180.56 Tm (-3) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 238.72 Tm (-2) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 296.88 Tm (-1) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 358.54 Tm (0) Tj ET BT /F2 1 Tf 0.00 12.00 -12.00 0.00 41.76 416.70 Tm (1) Tj ET 59.04 73.44 m 473.76 73.44 l 473.76 444.96 l 59.04 444.96 l 59.04 73.44 l S Q q Q q 59.04 73.44 414.72 371.52 re W n /sRGB CS 0.000 0.000 0.000 SCN 0.75 w [] 0 d 1 J 1 j 10.00 M 74.40 121.38 m 131.31 121.38 l 131.31 162.17 l 153.70 162.17 l 153.70 186.23 l 206.09 186.23 l 206.09 203.44 l 244.80 203.44 l 244.80 216.91 l 259.54 216.91 l 259.54 228.01 l 275.42 228.01 l 275.42 237.48 l 318.44 237.48 l 318.44 245.76 l 327.21 245.76 l 327.21 253.13 l 331.43 253.13 l 331.43 259.91 l 336.58 259.91 l 336.58 266.09 l 341.02 266.09 l 341.02 271.89 l 355.60 271.89 l 355.60 277.85 l 361.52 277.85 l 361.52 283.78 l 366.22 283.78 l 366.22 289.74 l 366.48 289.74 l 366.48 295.30 l 370.65 295.30 l 370.65 300.82 l 374.59 300.82 l 374.59 306.20 l 378.32 306.20 l 378.32 311.46 l 378.96 311.46 l 378.96 316.47 l 380.64 316.47 l 380.64 321.24 l 381.05 321.24 l 381.05 325.83 l 381.67 325.83 l 381.67 334.55 l 385.26 334.55 l 385.26 338.72 l 389.60 338.72 l 389.60 343.23 l 395.58 343.23 l 395.58 347.64 l 396.41 347.64 l 396.41 351.97 l 397.90 351.97 l 397.90 356.26 l 399.99 356.26 l 399.99 360.82 l 414.72 360.82 l 414.72 365.76 l 416.76 365.76 l 416.76 371.23 l 424.95 371.23 l 424.95 377.49 l 426.29 377.49 l 426.29 383.97 l 431.60 383.97 l 431.60 390.80 l 439.91 390.80 l 439.91 398.20 l 447.85 398.20 l 447.85 416.56 l 457.55 416.56 l 457.55 416.56 l S 131.31 87.20 m 143.37 87.20 l 143.37 127.77 l 252.55 127.77 l 252.55 151.61 l 253.76 151.61 l 253.76 181.85 l 256.13 181.85 l 256.13 192.72 l 273.62 192.72 l 273.62 201.96 l 281.40 201.96 l 281.40 217.12 l 284.61 217.12 l 284.61 223.53 l 286.93 223.53 l 286.93 229.35 l 295.51 229.35 l 295.51 234.69 l 297.51 234.69 l 297.51 239.63 l 301.34 239.63 l 301.34 244.23 l 310.12 244.23 l 310.12 248.53 l 310.67 248.53 l 310.67 252.57 l 312.29 252.57 l 312.29 256.39 l 315.94 256.39 l 315.94 260.01 l 316.95 260.01 l 316.95 263.45 l 317.45 263.45 l 317.45 266.74 l 322.73 266.74 l 322.73 269.88 l 325.90 269.88 l 325.90 275.79 l 327.65 275.79 l 327.65 278.57 l 328.93 278.57 l 328.93 281.26 l 331.02 281.26 l 331.02 283.92 l 332.65 283.92 l 332.65 289.04 l 333.46 289.04 l 333.46 293.86 l 333.85 293.86 l 333.85 296.17 l 335.42 296.17 l 335.42 298.45 l 338.46 298.45 l 338.46 300.73 l 339.57 300.73 l 339.57 302.98 l 341.38 302.98 l 341.38 305.21 l 343.14 305.21 l 343.14 307.41 l 344.18 307.41 l 344.18 309.57 l 346.88 309.57 l 346.88 311.68 l 348.52 311.68 l 348.52 313.81 l 349.48 313.81 l 349.48 315.93 l 350.43 315.93 l 350.43 318.01 l 350.75 318.01 l 350.75 320.06 l 353.52 320.06 l 353.52 322.10 l 355.31 322.10 l 355.31 324.14 l 361.79 324.14 l 361.79 326.14 l 362.05 326.14 l 362.05 328.12 l 362.32 328.12 l 362.32 330.07 l 365.72 330.07 l 365.72 332.06 l 365.97 332.06 l 365.97 334.03 l 368.22 334.03 l 368.22 335.97 l 370.17 335.97 l 370.17 337.93 l 371.12 337.93 l 371.12 339.96 l 371.36 339.96 l 371.36 341.97 l 372.29 341.97 l 372.29 343.96 l 380.02 343.96 l 380.02 345.93 l 383.69 345.93 l 383.69 349.93 l 383.88 349.93 l 383.88 351.90 l 385.26 351.90 l 385.26 353.93 l 388.86 353.93 l 388.86 356.01 l 395.24 356.01 l 395.24 358.22 l 395.74 358.22 l 395.74 360.43 l 399.19 360.43 l 399.19 362.64 l 400.31 362.64 l 400.31 364.85 l 400.78 364.85 l 400.78 367.15 l 402.79 367.15 l 402.79 369.47 l 403.40 369.47 l 403.40 371.79 l 409.49 371.79 l 409.49 374.26 l 409.63 374.26 l 409.63 376.74 l 410.18 376.74 l 410.18 379.26 l 413.68 379.26 l 413.68 381.98 l 415.87 381.98 l 415.87 384.75 l 417.88 384.75 l 417.88 387.59 l 421.51 387.59 l 421.51 390.51 l 422.79 390.51 l 422.79 393.52 l 424.73 393.52 l 424.73 396.65 l 429.73 396.65 l 429.73 399.93 l 429.94 399.93 l 429.94 403.38 l 433.92 403.38 l 433.92 407.07 l 434.21 407.07 l 434.21 411.05 l 434.61 411.05 l 434.61 415.43 l 437.49 415.43 l 437.49 421.74 l 458.40 421.74 l 458.40 421.74 l S 131.31 134.95 m 137.59 134.95 l 137.59 175.86 l 143.37 175.86 l 143.37 200.04 l 193.40 200.04 l 193.40 217.39 l 203.73 217.39 l 203.73 231.00 l 244.80 231.00 l 244.80 242.24 l 246.15 242.24 l 246.15 251.86 l 254.95 251.86 l 254.95 260.29 l 259.54 260.29 l 259.54 267.82 l 275.42 267.82 l 275.42 274.64 l 285.39 274.64 l 285.39 281.05 l 286.93 281.05 l 286.93 286.96 l 294.15 286.96 l 294.15 292.47 l 295.51 292.47 l 295.51 297.78 l 304.98 297.78 l 304.98 302.78 l 310.67 302.78 l 310.67 307.53 l 317.45 307.53 l 317.45 312.05 l 321.33 312.05 l 321.33 316.38 l 322.73 316.38 l 322.73 320.53 l 325.90 320.53 l 325.90 324.55 l 327.21 324.55 l 327.21 328.43 l 331.84 328.43 l 331.84 332.21 l 333.06 332.21 l 333.06 335.89 l 334.25 335.89 l 334.25 339.49 l 340.30 339.49 l 340.30 343.01 l 341.02 343.01 l 341.02 346.48 l 343.49 346.48 l 343.49 349.91 l 344.87 349.91 l 344.87 353.46 l 348.19 353.46 l 348.19 356.99 l 353.52 356.99 l 353.52 360.49 l 366.22 360.49 l 366.22 364.00 l 366.98 364.00 l 366.98 367.51 l 367.73 367.51 l 367.73 371.04 l 372.29 371.04 l 372.29 374.89 l 376.59 374.89 l 376.59 378.80 l 381.26 378.80 l 381.26 382.80 l 383.29 382.80 l 383.29 386.93 l 388.31 386.93 l 388.31 391.22 l 398.22 391.22 l 398.22 395.73 l 410.18 395.73 l 410.18 401.37 l 411.41 401.37 l 411.41 407.60 l 426.18 407.60 l 426.18 417.46 l 431.80 417.46 l 431.80 431.20 l 441.98 431.20 l S 0.000 0.804 0.000 SCN 0.75 w [ 0.00 3.00] 0 d 131.31 150.31 m 143.37 150.31 l 143.37 178.34 l 252.55 178.34 l 252.55 195.63 l 253.76 195.63 l 253.76 218.46 l 256.13 218.46 l 256.13 226.92 l 273.62 226.92 l 273.62 234.20 l 281.40 234.20 l 281.40 246.37 l 284.61 246.37 l 284.61 251.58 l 286.93 251.58 l 286.93 256.36 l 295.51 256.36 l 295.51 260.78 l 297.51 260.78 l 297.51 264.89 l 301.34 264.89 l 301.34 268.74 l 310.12 268.74 l 310.12 272.36 l 310.67 272.36 l 310.67 275.78 l 312.29 275.78 l 312.29 279.03 l 315.94 279.03 l 315.94 282.12 l 316.95 282.12 l 316.95 285.07 l 317.45 285.07 l 317.45 287.90 l 322.73 287.90 l 322.73 290.62 l 325.90 290.62 l 325.90 295.75 l 327.65 295.75 l 327.65 298.18 l 328.93 298.18 l 328.93 300.54 l 331.02 300.54 l 331.02 302.89 l 332.65 302.89 l 332.65 307.43 l 333.46 307.43 l 333.46 311.73 l 333.85 311.73 l 333.85 313.79 l 335.42 313.79 l 335.42 315.84 l 338.46 315.84 l 338.46 317.90 l 339.57 317.90 l 339.57 319.94 l 341.38 319.94 l 341.38 321.98 l 343.14 321.98 l 343.14 324.00 l 344.18 324.00 l 344.18 325.97 l 346.88 325.97 l 346.88 327.91 l 348.52 327.91 l 348.52 329.88 l 349.48 329.88 l 349.48 331.85 l 350.43 331.85 l 350.43 333.78 l 350.75 333.78 l 350.75 335.68 l 353.52 335.68 l 353.52 337.58 l 355.31 337.58 l 355.31 339.50 l 361.79 339.50 l 361.79 341.38 l 362.05 341.38 l 362.05 343.23 l 362.32 343.23 l 362.32 345.06 l 365.72 345.06 l 365.72 346.95 l 365.97 346.95 l 365.97 348.82 l 368.22 348.82 l 368.22 350.66 l 370.17 350.66 l 370.17 352.53 l 371.12 352.53 l 371.12 354.49 l 371.36 354.49 l 371.36 356.42 l 372.29 356.42 l 372.29 358.33 l 380.02 358.33 l 380.02 360.23 l 383.69 360.23 l 383.69 364.10 l 383.88 364.10 l 383.88 366.02 l 385.26 366.02 l 385.26 368.00 l 388.86 368.00 l 388.86 370.04 l 395.24 370.04 l 395.24 372.25 l 395.74 372.25 l 395.74 374.45 l 399.19 374.45 l 399.19 376.65 l 400.31 376.65 l 400.31 378.85 l 400.78 378.85 l 400.78 381.17 l 402.79 381.17 l 402.79 383.50 l 403.40 383.50 l 403.40 385.84 l 409.49 385.84 l 409.49 388.35 l 409.63 388.35 l 409.63 390.88 l 410.18 390.88 l 410.18 393.45 l 413.68 393.45 l 413.68 396.27 l 415.87 396.27 l 415.87 399.15 l 417.88 399.15 l 417.88 402.09 l 421.51 402.09 l 421.51 405.13 l 422.79 405.13 l 422.79 408.29 l 424.73 408.29 l 424.73 411.57 l 429.73 411.57 l 429.73 415.03 l 429.94 415.03 l 429.94 418.71 l 433.92 418.71 l 433.92 422.66 l 434.21 422.66 l 434.21 426.98 l 434.61 426.98 l 434.61 431.81 l 437.49 431.81 l 437.49 439.41 l 458.40 439.41 l 458.40 439.41 l S 253.76 60.17 m 256.13 60.17 l 256.13 99.05 l 273.62 99.05 l 273.62 123.41 l 281.40 123.41 l 281.40 155.48 l 284.61 155.48 l 284.61 167.25 l 286.93 167.25 l 286.93 177.32 l 295.51 177.32 l 295.51 186.13 l 297.51 186.13 l 297.51 193.97 l 301.34 193.97 l 301.34 201.03 l 310.12 201.03 l 310.12 207.46 l 310.67 207.46 l 310.67 213.36 l 312.29 213.36 l 312.29 218.82 l 315.94 218.82 l 315.94 223.90 l 316.95 223.90 l 316.95 228.66 l 317.45 228.66 l 317.45 233.13 l 322.73 233.13 l 322.73 237.35 l 325.90 237.35 l 325.90 245.14 l 327.65 245.14 l 327.65 248.77 l 328.93 248.77 l 328.93 252.23 l 331.02 252.23 l 331.02 255.60 l 332.65 255.60 l 332.65 262.00 l 333.46 262.00 l 333.46 267.94 l 333.85 267.94 l 333.85 270.77 l 335.42 270.77 l 335.42 273.52 l 338.46 273.52 l 338.46 276.25 l 339.57 276.25 l 339.57 278.92 l 341.38 278.92 l 341.38 281.55 l 343.14 281.55 l 343.14 284.13 l 344.18 284.13 l 344.18 286.64 l 346.88 286.64 l 346.88 289.09 l 348.52 289.09 l 348.52 291.53 l 349.48 291.53 l 349.48 293.95 l 350.43 293.95 l 350.43 296.31 l 350.75 296.31 l 350.75 298.63 l 353.52 298.63 l 353.52 300.92 l 355.31 300.92 l 355.31 303.20 l 361.79 303.20 l 361.79 305.44 l 362.05 305.44 l 362.05 307.64 l 362.32 307.64 l 362.32 309.80 l 365.72 309.80 l 365.72 311.99 l 365.97 311.99 l 365.97 314.15 l 368.22 314.15 l 368.22 316.27 l 370.17 316.27 l 370.17 318.40 l 371.12 318.40 l 371.12 320.56 l 371.36 320.56 l 371.36 322.71 l 372.29 322.71 l 372.29 324.83 l 380.02 324.83 l 380.02 326.92 l 383.69 326.92 l 383.69 331.14 l 383.88 331.14 l 383.88 333.22 l 385.26 333.22 l 385.26 335.33 l 388.86 335.33 l 388.86 337.47 l 395.24 337.47 l 395.24 339.70 l 395.74 339.70 l 395.74 341.93 l 399.19 341.93 l 399.19 344.15 l 400.31 344.15 l 400.31 346.37 l 400.78 346.37 l 400.78 348.65 l 402.79 348.65 l 402.79 350.94 l 403.40 350.94 l 403.40 353.24 l 409.49 353.24 l 409.49 355.62 l 409.63 355.62 l 409.63 358.03 l 410.18 358.03 l 410.18 360.46 l 413.68 360.46 l 413.68 363.00 l 415.87 363.00 l 415.87 365.59 l 417.88 365.59 l 417.88 368.23 l 421.51 368.23 l 421.51 370.92 l 422.79 370.92 l 422.79 373.69 l 424.73 373.69 l 424.73 376.53 l 429.73 376.53 l 429.73 379.47 l 429.94 379.47 l 429.94 382.52 l 433.92 382.52 l 433.92 385.70 l 434.21 385.70 l 434.21 389.03 l 434.61 389.03 l 434.61 392.56 l 437.49 392.56 l 437.49 396.25 l 458.40 396.25 l 458.40 396.25 l S Q endstream endobj 15 0 obj 11534 endobj 3 0 obj << /Type /Pages /Kids [ 7 0 R 10 0 R 13 0 R ] /Count 3 /MediaBox [0 0 504 504] >> endobj 4 0 obj << /ProcSet [/PDF /Text] /Font << /F1 17 0 R /F2 18 0 R >> /ExtGState << >> /ColorSpace << /sRGB 5 0 R >> >> endobj 5 0 obj [/ICCBased 6 0 R] endobj 6 0 obj << /N 3 /Alternate /DeviceRGB /Length 9433 /Filter /ASCIIHexDecode >> stream 00 00 0c 48 4c 69 6e 6f 02 10 00 00 6d 6e 74 72 52 47 42 20 58 59 5a 20 07 ce 00 02 00 09 00 06 00 31 00 00 61 63 73 70 4d 53 46 54 00 00 00 00 49 45 43 20 73 52 47 42 00 00 00 00 00 00 00 00 00 00 00 00 00 00 f6 d6 00 01 00 00 00 00 d3 2d 48 50 20 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 11 63 70 72 74 00 00 01 50 00 00 00 33 64 65 73 63 00 00 01 84 00 00 00 6c 77 74 70 74 00 00 01 f0 00 00 00 14 62 6b 70 74 00 00 02 04 00 00 00 14 72 58 59 5a 00 00 02 18 00 00 00 14 67 58 59 5a 00 00 02 2c 00 00 00 14 62 58 59 5a 00 00 02 40 00 00 00 14 64 6d 6e 64 00 00 02 54 00 00 00 70 64 6d 64 64 00 00 02 c4 00 00 00 88 76 75 65 64 00 00 03 4c 00 00 00 86 76 69 65 77 00 00 03 d4 00 00 00 24 6c 75 6d 69 00 00 03 f8 00 00 00 14 6d 65 61 73 00 00 04 0c 00 00 00 24 74 65 63 68 00 00 04 30 00 00 00 0c 72 54 52 43 00 00 04 3c 00 00 08 0c 67 54 52 43 00 00 04 3c 00 00 08 0c 62 54 52 43 00 00 04 3c 00 00 08 0c 74 65 78 74 00 00 00 00 43 6f 70 79 72 69 67 68 74 20 28 63 29 20 31 39 39 38 20 48 65 77 6c 65 74 74 2d 50 61 63 6b 61 72 64 20 43 6f 6d 70 61 6e 79 00 00 64 65 73 63 00 00 00 00 00 00 00 12 73 52 47 42 20 49 45 43 36 31 39 36 36 2d 32 2e 31 00 00 00 00 00 00 00 00 00 00 00 12 73 52 47 42 20 49 45 43 36 31 39 36 36 2d 32 2e 31 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 58 59 5a 20 00 00 00 00 00 00 f3 51 00 01 00 00 00 01 16 cc 58 59 5a 20 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 58 59 5a 20 00 00 00 00 00 00 6f a2 00 00 38 f5 00 00 03 90 58 59 5a 20 00 00 00 00 00 00 62 99 00 00 b7 85 00 00 18 da 58 59 5a 20 00 00 00 00 00 00 24 a0 00 00 0f 84 00 00 b6 cf 64 65 73 63 00 00 00 00 00 00 00 16 49 45 43 20 68 74 74 70 3a 2f 2f 77 77 77 2e 69 65 63 2e 63 68 00 00 00 00 00 00 00 00 00 00 00 16 49 45 43 20 68 74 74 70 3a 2f 2f 77 77 77 2e 69 65 63 2e 63 68 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 64 65 73 63 00 00 00 00 00 00 00 2e 49 45 43 20 36 31 39 36 36 2d 32 2e 31 20 44 65 66 61 75 6c 74 20 52 47 42 20 63 6f 6c 6f 75 72 20 73 70 61 63 65 20 2d 20 73 52 47 42 00 00 00 00 00 00 00 00 00 00 00 2e 49 45 43 20 36 31 39 36 36 2d 32 2e 31 20 44 65 66 61 75 6c 74 20 52 47 42 20 63 6f 6c 6f 75 72 20 73 70 61 63 65 20 2d 20 73 52 47 42 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 64 65 73 63 00 00 00 00 00 00 00 2c 52 65 66 65 72 65 6e 63 65 20 56 69 65 77 69 6e 67 20 43 6f 6e 64 69 74 69 6f 6e 20 69 6e 20 49 45 43 36 31 39 36 36 2d 32 2e 31 00 00 00 00 00 00 00 00 00 00 00 2c 52 65 66 65 72 65 6e 63 65 20 56 69 65 77 69 6e 67 20 43 6f 6e 64 69 74 69 6f 6e 20 69 6e 20 49 45 43 36 31 39 36 36 2d 32 2e 31 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 76 69 65 77 00 00 00 00 00 13 a4 fe 00 14 5f 2e 00 10 cf 14 00 03 ed cc 00 04 13 0b 00 03 5c 9e 00 00 00 01 58 59 5a 20 00 00 00 00 00 4c 09 56 00 50 00 00 00 57 1f e7 6d 65 61 73 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 02 8f 00 00 00 02 73 69 67 20 00 00 00 00 43 52 54 20 63 75 72 76 00 00 00 00 00 00 04 00 00 00 00 05 00 0a 00 0f 00 14 00 19 00 1e 00 23 00 28 00 2d 00 32 00 37 00 3b 00 40 00 45 00 4a 00 4f 00 54 00 59 00 5e 00 63 00 68 00 6d 00 72 00 77 00 7c 00 81 00 86 00 8b 00 90 00 95 00 9a 00 9f 00 a4 00 a9 00 ae 00 b2 00 b7 00 bc 00 c1 00 c6 00 cb 00 d0 00 d5 00 db 00 e0 00 e5 00 eb 00 f0 00 f6 00 fb 01 01 01 07 01 0d 01 13 01 19 01 1f 01 25 01 2b 01 32 01 38 01 3e 01 45 01 4c 01 52 01 59 01 60 01 67 01 6e 01 75 01 7c 01 83 01 8b 01 92 01 9a 01 a1 01 a9 01 b1 01 b9 01 c1 01 c9 01 d1 01 d9 01 e1 01 e9 01 f2 01 fa 02 03 02 0c 02 14 02 1d 02 26 02 2f 02 38 02 41 02 4b 02 54 02 5d 02 67 02 71 02 7a 02 84 02 8e 02 98 02 a2 02 ac 02 b6 02 c1 02 cb 02 d5 02 e0 02 eb 02 f5 03 00 03 0b 03 16 03 21 03 2d 03 38 03 43 03 4f 03 5a 03 66 03 72 03 7e 03 8a 03 96 03 a2 03 ae 03 ba 03 c7 03 d3 03 e0 03 ec 03 f9 04 06 04 13 04 20 04 2d 04 3b 04 48 04 55 04 63 04 71 04 7e 04 8c 04 9a 04 a8 04 b6 04 c4 04 d3 04 e1 04 f0 04 fe 05 0d 05 1c 05 2b 05 3a 05 49 05 58 05 67 05 77 05 86 05 96 05 a6 05 b5 05 c5 05 d5 05 e5 05 f6 06 06 06 16 06 27 06 37 06 48 06 59 06 6a 06 7b 06 8c 06 9d 06 af 06 c0 06 d1 06 e3 06 f5 07 07 07 19 07 2b 07 3d 07 4f 07 61 07 74 07 86 07 99 07 ac 07 bf 07 d2 07 e5 07 f8 08 0b 08 1f 08 32 08 46 08 5a 08 6e 08 82 08 96 08 aa 08 be 08 d2 08 e7 08 fb 09 10 09 25 09 3a 09 4f 09 64 09 79 09 8f 09 a4 09 ba 09 cf 09 e5 09 fb 0a 11 0a 27 0a 3d 0a 54 0a 6a 0a 81 0a 98 0a ae 0a c5 0a dc 0a f3 0b 0b 0b 22 0b 39 0b 51 0b 69 0b 80 0b 98 0b b0 0b c8 0b e1 0b f9 0c 12 0c 2a 0c 43 0c 5c 0c 75 0c 8e 0c a7 0c c0 0c d9 0c f3 0d 0d 0d 26 0d 40 0d 5a 0d 74 0d 8e 0d a9 0d c3 0d de 0d f8 0e 13 0e 2e 0e 49 0e 64 0e 7f 0e 9b 0e b6 0e d2 0e ee 0f 09 0f 25 0f 41 0f 5e 0f 7a 0f 96 0f b3 0f cf 0f ec 10 09 10 26 10 43 10 61 10 7e 10 9b 10 b9 10 d7 10 f5 11 13 11 31 11 4f 11 6d 11 8c 11 aa 11 c9 11 e8 12 07 12 26 12 45 12 64 12 84 12 a3 12 c3 12 e3 13 03 13 23 13 43 13 63 13 83 13 a4 13 c5 13 e5 14 06 14 27 14 49 14 6a 14 8b 14 ad 14 ce 14 f0 15 12 15 34 15 56 15 78 15 9b 15 bd 15 e0 16 03 16 26 16 49 16 6c 16 8f 16 b2 16 d6 16 fa 17 1d 17 41 17 65 17 89 17 ae 17 d2 17 f7 18 1b 18 40 18 65 18 8a 18 af 18 d5 18 fa 19 20 19 45 19 6b 19 91 19 b7 19 dd 1a 04 1a 2a 1a 51 1a 77 1a 9e 1a c5 1a ec 1b 14 1b 3b 1b 63 1b 8a 1b b2 1b da 1c 02 1c 2a 1c 52 1c 7b 1c a3 1c cc 1c f5 1d 1e 1d 47 1d 70 1d 99 1d c3 1d ec 1e 16 1e 40 1e 6a 1e 94 1e be 1e e9 1f 13 1f 3e 1f 69 1f 94 1f bf 1f ea 20 15 20 41 20 6c 20 98 20 c4 20 f0 21 1c 21 48 21 75 21 a1 21 ce 21 fb 22 27 22 55 22 82 22 af 22 dd 23 0a 23 38 23 66 23 94 23 c2 23 f0 24 1f 24 4d 24 7c 24 ab 24 da 25 09 25 38 25 68 25 97 25 c7 25 f7 26 27 26 57 26 87 26 b7 26 e8 27 18 27 49 27 7a 27 ab 27 dc 28 0d 28 3f 28 71 28 a2 28 d4 29 06 29 38 29 6b 29 9d 29 d0 2a 02 2a 35 2a 68 2a 9b 2a cf 2b 02 2b 36 2b 69 2b 9d 2b d1 2c 05 2c 39 2c 6e 2c a2 2c d7 2d 0c 2d 41 2d 76 2d ab 2d e1 2e 16 2e 4c 2e 82 2e b7 2e ee 2f 24 2f 5a 2f 91 2f c7 2f fe 30 35 30 6c 30 a4 30 db 31 12 31 4a 31 82 31 ba 31 f2 32 2a 32 63 32 9b 32 d4 33 0d 33 46 33 7f 33 b8 33 f1 34 2b 34 65 34 9e 34 d8 35 13 35 4d 35 87 35 c2 35 fd 36 37 36 72 36 ae 36 e9 37 24 37 60 37 9c 37 d7 38 14 38 50 38 8c 38 c8 39 05 39 42 39 7f 39 bc 39 f9 3a 36 3a 74 3a b2 3a ef 3b 2d 3b 6b 3b aa 3b e8 3c 27 3c 65 3c a4 3c e3 3d 22 3d 61 3d a1 3d e0 3e 20 3e 60 3e a0 3e e0 3f 21 3f 61 3f a2 3f e2 40 23 40 64 40 a6 40 e7 41 29 41 6a 41 ac 41 ee 42 30 42 72 42 b5 42 f7 43 3a 43 7d 43 c0 44 03 44 47 44 8a 44 ce 45 12 45 55 45 9a 45 de 46 22 46 67 46 ab 46 f0 47 35 47 7b 47 c0 48 05 48 4b 48 91 48 d7 49 1d 49 63 49 a9 49 f0 4a 37 4a 7d 4a c4 4b 0c 4b 53 4b 9a 4b e2 4c 2a 4c 72 4c ba 4d 02 4d 4a 4d 93 4d dc 4e 25 4e 6e 4e b7 4f 00 4f 49 4f 93 4f dd 50 27 50 71 50 bb 51 06 51 50 51 9b 51 e6 52 31 52 7c 52 c7 53 13 53 5f 53 aa 53 f6 54 42 54 8f 54 db 55 28 55 75 55 c2 56 0f 56 5c 56 a9 56 f7 57 44 57 92 57 e0 58 2f 58 7d 58 cb 59 1a 59 69 59 b8 5a 07 5a 56 5a a6 5a f5 5b 45 5b 95 5b e5 5c 35 5c 86 5c d6 5d 27 5d 78 5d c9 5e 1a 5e 6c 5e bd 5f 0f 5f 61 5f b3 60 05 60 57 60 aa 60 fc 61 4f 61 a2 61 f5 62 49 62 9c 62 f0 63 43 63 97 63 eb 64 40 64 94 64 e9 65 3d 65 92 65 e7 66 3d 66 92 66 e8 67 3d 67 93 67 e9 68 3f 68 96 68 ec 69 43 69 9a 69 f1 6a 48 6a 9f 6a f7 6b 4f 6b a7 6b ff 6c 57 6c af 6d 08 6d 60 6d b9 6e 12 6e 6b 6e c4 6f 1e 6f 78 6f d1 70 2b 70 86 70 e0 71 3a 71 95 71 f0 72 4b 72 a6 73 01 73 5d 73 b8 74 14 74 70 74 cc 75 28 75 85 75 e1 76 3e 76 9b 76 f8 77 56 77 b3 78 11 78 6e 78 cc 79 2a 79 89 79 e7 7a 46 7a a5 7b 04 7b 63 7b c2 7c 21 7c 81 7c e1 7d 41 7d a1 7e 01 7e 62 7e c2 7f 23 7f 84 7f e5 80 47 80 a8 81 0a 81 6b 81 cd 82 30 82 92 82 f4 83 57 83 ba 84 1d 84 80 84 e3 85 47 85 ab 86 0e 86 72 86 d7 87 3b 87 9f 88 04 88 69 88 ce 89 33 89 99 89 fe 8a 64 8a ca 8b 30 8b 96 8b fc 8c 63 8c ca 8d 31 8d 98 8d ff 8e 66 8e ce 8f 36 8f 9e 90 06 90 6e 90 d6 91 3f 91 a8 92 11 92 7a 92 e3 93 4d 93 b6 94 20 94 8a 94 f4 95 5f 95 c9 96 34 96 9f 97 0a 97 75 97 e0 98 4c 98 b8 99 24 99 90 99 fc 9a 68 9a d5 9b 42 9b af 9c 1c 9c 89 9c f7 9d 64 9d d2 9e 40 9e ae 9f 1d 9f 8b 9f fa a0 69 a0 d8 a1 47 a1 b6 a2 26 a2 96 a3 06 a3 76 a3 e6 a4 56 a4 c7 a5 38 a5 a9 a6 1a a6 8b a6 fd a7 6e a7 e0 a8 52 a8 c4 a9 37 a9 a9 aa 1c aa 8f ab 02 ab 75 ab e9 ac 5c ac d0 ad 44 ad b8 ae 2d ae a1 af 16 af 8b b0 00 b0 75 b0 ea b1 60 b1 d6 b2 4b b2 c2 b3 38 b3 ae b4 25 b4 9c b5 13 b5 8a b6 01 b6 79 b6 f0 b7 68 b7 e0 b8 59 b8 d1 b9 4a b9 c2 ba 3b ba b5 bb 2e bb a7 bc 21 bc 9b bd 15 bd 8f be 0a be 84 be ff bf 7a bf f5 c0 70 c0 ec c1 67 c1 e3 c2 5f c2 db c3 58 c3 d4 c4 51 c4 ce c5 4b c5 c8 c6 46 c6 c3 c7 41 c7 bf c8 3d c8 bc c9 3a c9 b9 ca 38 ca b7 cb 36 cb b6 cc 35 cc b5 cd 35 cd b5 ce 36 ce b6 cf 37 cf b8 d0 39 d0 ba d1 3c d1 be d2 3f d2 c1 d3 44 d3 c6 d4 49 d4 cb d5 4e d5 d1 d6 55 d6 d8 d7 5c d7 e0 d8 64 d8 e8 d9 6c d9 f1 da 76 da fb db 80 dc 05 dc 8a dd 10 dd 96 de 1c de a2 df 29 df af e0 36 e0 bd e1 44 e1 cc e2 53 e2 db e3 63 e3 eb e4 73 e4 fc e5 84 e6 0d e6 96 e7 1f e7 a9 e8 32 e8 bc e9 46 e9 d0 ea 5b ea e5 eb 70 eb fb ec 86 ed 11 ed 9c ee 28 ee b4 ef 40 ef cc f0 58 f0 e5 f1 72 f1 ff f2 8c f3 19 f3 a7 f4 34 f4 c2 f5 50 f5 de f6 6d f6 fb f7 8a f8 19 f8 a8 f9 38 f9 c7 fa 57 fa e7 fb 77 fc 07 fc 98 fd 29 fd ba fe 4b fe dc ff 6d ff ff > endstream endobj 16 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi /grave /acute /circumflex /tilde /macron /breve /dotaccent /dieresis /.notdef /ring /cedilla /.notdef /hungarumlaut /ogonek /caron /space] >> endobj 17 0 obj << /Type /Font /Subtype /Type1 /Name /F1 /BaseFont /ZapfDingbats >> endobj 18 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 16 0 R >> endobj xref 0 19 0000000000 65535 f 0000000021 00000 n 0000000163 00000 n 0000043825 00000 n 0000043922 00000 n 0000044046 00000 n 0000044079 00000 n 0000000212 00000 n 0000000292 00000 n 0000011002 00000 n 0000011023 00000 n 0000011105 00000 n 0000032111 00000 n 0000032133 00000 n 0000032215 00000 n 0000043803 00000 n 0000053615 00000 n 0000053873 00000 n 0000053957 00000 n trailer << /Size 19 /Info 1 0 R /Root 2 0 R >> startxref 54055 %%EOF survival/tests/r_peterson.R0000644000175100001440000000270411732700061015610 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Data courtesy of Bercedis Peterson, Duke University. # v4 of survreg fails due to 2 groups that have only 1 subject; the coef # for them easily gets out of hand. In fact, this data set is my toughest # test of the minimizer. # # A shrinkage model for this coefficient is therefore interesting peterson <- data.frame( scan('data.peterson', what=list(grp=0, time=0, status=0))) fitp <- survreg(Surv(time, status) ~ factor(grp), peterson) summary(fitp) # Now a shrinkage model. Give the group coefficients # about 1/2 the scale parameter of the original model, i.e., .18. # ffit <- survreg(Surv(time, status) ~ frailty(grp, theta=.1), peterson) ffit # # Try 3 degrees of freedom, since there are 6 groups # compare them to the unconstrained ones. The frailty coefs are # on a "sum to constant" constraint rather than "first coef=0", so # some conversion is neccessary # ffit3 <- survreg(Surv(time, status) ~ frailty(grp, df=3), peterson) print(ffit3) temp <- mean(c(0, fitp$coef[-1])) - mean(ffit3$frail) temp2 <- c(fitp$coef[1] + temp, c(0,fitp$coef[-1]) - temp) xx <- rbind(c(nrow(peterson), table(peterson$grp)), temp2, c(ffit3$coef, ffit3$frail)) dimnames(xx) <- list(c("N", "factor", "frailty"), c("Intercept", paste("grp", 1:6))) signif(xx,3) rm(ffit, ffit3, temp, temp2, xx, fitp) survival/tests/rounding.Rout.save0000644000175100001440000000217211732700061016741 0ustar hornikusers R version 2.11.0 (2010-04-22) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Survival curves could fail with data that was almost exact. > # The calculations use both unique() and table(), which don't > # necessarily give the same number of values. > # Check that the routine handles this properly > # > > tdata <- data.frame(time=c(1,2, sqrt(2)^2, 2, sqrt(2)^2), + status=rep(1,5), + group=c(1,1,1,2,2)) > fit <- survfit(Surv(time, status) ~ group, data=tdata) > > all.equal(sum(fit$strata), length(fit$time)) [1] TRUE > survival/tests/r_capacitor.Rout.save0000644000175100001440000000506011732700061017401 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > capacitor <- read.table('data.capacitor', row.names=1, + col.names=c('', 'days', 'event', 'voltage')) > > fitig <- survreg(Surv(days, event)~voltage, + dist = "gaussian", data = capacitor) > summary(fitig) Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "gaussian") Value Std. Error z p (Intercept) 1764.9 163.387 10.80 3.36e-27 voltage -53.9 5.545 -9.72 2.56e-22 Log(scale) 4.8 0.105 45.56 0.00e+00 Scale= 121 Gaussian distribution Loglik(model)= -361.9 Loglik(intercept only)= -420.1 Chisq= 116.33 on 1 degrees of freedom, p= 0 Number of Newton-Raphson Iterations: 6 n= 125 > > fitix <- survreg(Surv(days, event)~voltage, + dist = "extreme", data = capacitor) > summary(fitix) Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "extreme") Value Std. Error z p (Intercept) 2055.59 180.349 11.4 4.28e-30 voltage -62.21 5.967 -10.4 1.88e-25 Log(scale) 4.53 0.108 41.9 0.00e+00 Scale= 92.9 Extreme value distribution Loglik(model)= -360 Loglik(intercept only)= -427.1 Chisq= 134.25 on 1 degrees of freedom, p= 0 Number of Newton-Raphson Iterations: 7 n= 125 > > fitil <- survreg(Surv(days, event)~voltage, + dist = "logistic", data = capacitor) > summary(fitil) Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "logistic") Value Std. Error z p (Intercept) 1811.56 148.853 12.2 4.48e-34 voltage -55.48 4.986 -11.1 9.39e-29 Log(scale) 4.19 0.117 35.8 2.03e-280 Scale= 66.3 Logistic distribution Loglik(model)= -360.4 Loglik(intercept only)= -423.7 Chisq= 126.5 on 1 degrees of freedom, p= 0 Number of Newton-Raphson Iterations: 6 n= 125 > survival/tests/summary_survfit.R0000644000175100001440000000037611732700061016712 0ustar hornikusers## check that the scale option to summary.survfit works ## Marc Schwartz reported this as a bug in 2.35-3. library(survival) summary( survfit( Surv(futime, fustat)~1, data=ovarian)) summary( survfit( Surv(futime, fustat)~1, data=ovarian), scale=365.25) survival/tests/data.interval0000644000175100001440000000161711732700061015766 0ustar hornikusers This data set is to test interval censoring. It has 2 left censored, 14 right censored, 2 exact and 8 interval censored observations, grafted onto covariates from the ovarian data set. "ltime","rtime","age","resid.ds","rx","ecog.ps" "1",NA,150,72.3315,2,1,1 "2",NA,150,74.4932,2,1,1 "3",146,166,66.4658,2,1,2 "4",421,NA,53.3644,2,2,1 "5",421,421,50.3397,2,1,1 "6",448,NA,56.4301,1,1,2 "7",454,474,56.937,2,2,2 "8",465,485,59.8548,2,2,2 "9",477,NA,64.1753,2,1,1 "10",553,573,55.1781,1,2,2 "11",628,648,56.7562,1,1,2 "12",744,NA,50.1096,1,2,1 "13",769,NA,59.6301,2,2,2 "14",770,NA,57.0521,2,2,1 "15",803,NA,39.2712,1,1,1 "16",855,NA,43.1233,1,1,2 "17",1040,NA,38.8932,2,1,2 "18",1106,NA,44.6,1,1,1 "19",1129,NA,53.9068,1,2,1 "20",1206,NA,44.2055,2,2,1 "21",1227,NA,59.589,1,2,2 "22",258,278,74.5041,2,1,2 "23",319,339,43.137,2,1,1 "24",343,363,63.2192,1,2,2 "25",375,375,64.4247,2,2,1 "26",377,NA,58.3096,1,2,1 survival/tests/testnull.Rout.save0000644000175100001440000000242711732700061016771 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # A test of NULL models > # > fit1 <- coxph(Surv(stop, event) ~ rx + strata(number), bladder, iter=0) > fit2 <- coxph(Surv(stop, event) ~ strata(number), bladder) > > all.equal(fit1$loglik[2], fit2$loglik) [1] TRUE > all.equal(fit1$resid, fit2$resid) [1] TRUE > > > fit1 <- coxph(Surv(start, stop, event) ~ rx + strata(number), bladder2, iter=0) > fit2 <- coxph(Surv(start, stop, event) ~ strata(number), bladder2) > > all.equal(fit1$loglik[2], fit2$loglik) [1] TRUE > all.equal(fit1$resid, fit2$resid) [1] TRUE > survival/tests/bladder.R0000644000175100001440000000244611732700061015030 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Fit the models found in Wei et. al. # wfit <- coxph(Surv(stop, event) ~ (rx + size + number)* strata(enum) + cluster(id), bladder, method='breslow') wfit # Check the rx coefs versus Wei, et al, JASA 1989 rx <- c(1,4,5,6) # the treatment coefs above cmat <- diag(4); cmat[1,] <- 1; #contrast matrix wfit$coef[rx] %*% cmat # the coefs in their paper (table 5) t(cmat) %*% wfit$var[rx,rx] %*% cmat # var matrix (eqn 3.2) # Anderson-Gill fit fita <- coxph(Surv(start, stop, event) ~ rx + size + number + cluster(id), bladder2, method='breslow') summary(fita) # Prentice fits. Their model 1 a and b are the same fit1p <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, subset=(enum==1), method='breslow') fit2pa <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, subset=(enum==2), method='breslow') fit2pb <- coxph(Surv(stop-start, event) ~ rx + size + number, bladder2, subset=(enum==2), method='breslow') fit3pa <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, subset=(enum==3), method='breslow') #and etc. fit1p fit2pa fit2pb fit3pa rm(rx, cmat, wfit, fita, fit1p, fit2pa, fit2pb, fit3pa) survival/tests/stratatest.Rout.save0000644000175100001440000000453711732700061017321 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Trivial test of stratified residuals > # Make a second strata = replicate of the first, and I should get the > # exact same answers > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > temp <- as.matrix(test1) > n <- nrow(temp) > ndead<- sum(test1$status[!is.na(test1$status)]) > temp <- data.frame(rbind(temp, temp)) #later releases of S have rbind.data.frame > tstrat <- rep(1:2, c(n,n)) > > fit1 <- coxph(Surv(time, status) ~x, test1) > fit2 <- coxph(Surv(time, status) ~x + strata(tstrat), temp) > > all.equal(resid(fit1) , (resid(fit2))[1:n]) [1] TRUE > all.equal(resid(fit1, type='score') , (resid(fit2, type='score'))[1:n]) [1] TRUE > all.equal(resid(fit1, type='schoe') , (resid(fit2, type='schoe'))[1:ndead]) [1] TRUE > > > #AG model > temp <- as.matrix(test2) > n <- nrow(temp) > ndead<- sum(test2$event[!is.na(test2$event)]) > temp <- data.frame(rbind(temp, temp)) > tstrat <- rep(1:2, c(n,n)) > > fit1 <- coxph(Surv(start, stop, event) ~x, test2) > fit2 <- coxph(Surv(start, stop, event) ~x + strata(tstrat), temp) > > all.equal(resid(fit1) , (resid(fit2))[1:n]) [1] TRUE > all.equal(resid(fit1, type='score') , (resid(fit2, type='score'))[1:n]) [1] TRUE > all.equal(resid(fit1, type='schoe') , (resid(fit2, type='schoe'))[1:ndead]) [1] TRUE > survival/tests/gray1.rda0000644000175100001440000000657512466142446015046 0ustar hornikusers‹íÙ{TŒkÛð‘ IåPS©vT:Û ƒ6rHh§bÛ&§è€PvçäC%Eï¦Ò”"t> Ò‘$QHo’Cìå­Ð[íoæÙ3×õÝ÷ZïZßZßzÿøÖzg]õ{ÖÓÌ<Ïsß×u5s?k¬ÍååX,–k¤”ð÷Hᦴpƒ5‚%Í#Tf‡sÐ käDæI,)‘¢g)x,ñcQ@iØ Ò¨&Ҙє¡¤±{IO½ ÛDšIš$Oy”ôœ ¥éù ”¤&‘¦PR~!Mõ#Mk#Mß@YJYIš1r i¶.¥/iN3i®éEÊ$Ê«”ͤ—¢I/O'Í3¦ÜNJ) Í/&-˜@YBzÅŒr3iávÒâù¤%z”ó(?^ â‘Æ²¯Y^ßBDz3GºEXCùGX]AZói)‹´™GØð…‘EõÉÃÒk¸Wi¸w<È£Œ8è®& ð„x¿ÄS s•…&ˆŸ'ñL…®¦0Àsâ×ILѺ7Q`šø}$f8E²…f‰ßWbÎYëÉÂ/Š#ñò3©„æ‰+±@áùa€WÄç!±púõI‹íU„–ˆÏKbéž Ua€eâó”xõo\ua€×Äç-±üÒl a€×Å×!±‚9--ð†øº$V ž*¾N‰7¼9yäM4XÅ\÷D4îatÜÃX°še4½6&½6¬aÆ…ÆƃµÌ<ª¡u)§êRNƒuÌ8NFE‡Në™qÕD_‡ð_‡œo1ãü:à?àu¼ÍŒûðŽì–Ó²[. Ì<è€ ¢YµKE™yѧ-Hš¶ eæI¼Ë19Ã1ÉD™y3ï1Óœ2óh Þ_'—¼N.eæÕl2ÿ& ÊÌ3½úÉáê'.ø@¦Ü[¦Ød¯7–h9â›Nõ8µ4Pîé›_ÊÀñí|oÞ(UfsŸæ®ª5‹æ¤¼µšëBG;ÖS"^üÚ69ÔIã« :D‚zåIJ-ïäAý{‚Í郠aû:nWðwиsìðl«^Ðôíàn©hðÇ·*ý[A³¾¹f/’ÀÝ ¿r z@ó?Œ·YqœÙúć­ξá²þ~Èsé§‘ëfæƒsN]³RÞ Î mL¾¶ð8Ï#¸¶ïn%h±ºáKâ{oð§Yù ªåf‚ó•í/ßÓ‰C?†›=P¿ .¸¿%o@C\˜ù`‘ü eð“gO®ŽYÖmžùqÒýòÆ ýºr¬ýøÏàbV‹ÂßÐúcR;ËÀ%Æ*ÓZŠÑßǺÙ5¢}¬ÊoÕ½àÒÕõïF˜Ç ùÎÓ¼oüh¥x»)þá*ÔsôôÀÁ,ôöDeýOp™Î»Ð2¿ô@ƶq.Ї‹G«†­­M+äƒÃõÑPùÄ;‰¿¡ç<È’×—›ž°Ê8¿ å­øP6ˆ6_îìP߮Љ횚)@}­)öi U1Én!{À•Jwk,CŒüÕ >¡ÙÅ–:ezè@¤¿šælÐf±S¼óZôXåçÑ*)hlÓ…‰ô—‡=ΞÛDŸ0•X¢/ í”öÛé¹M¼=r›¯Ÿd3ÀÙGÈ0ùc„•¤•,R¥Ô ´¤\D¹”GjǢ䑺²(µI·U’z;P摜# Ô& 6#åÑÆSf’†¨RÆS摆n% §<êHYJ±‚4ÒôwÊ-¤'ËIO¹QV’òHÜHOëRP–“&ÆS&Q¶P¶“&m¢ì&=#MšìHÉ#=·2šô¼å^Ò Ö”•”ݤ)Ò”ê”ÓHS9”ݤiäéS)çRºP–“fl  ¥ {2eò’ŠÒ2N€2|P Î$*îà…0Q LóÁGFpQ&¯ÙhŸ0(“ç|ð±…Çz .Êä=­r†lcê€ÚlÜ`³‘‹2—ÁF™‰€O˜:á£D/ä¢Lݰї‹…!Û™:⣻æmÚ5‹2uÅFûÍ„!Ÿ2uÆG÷lÞgÀE™ºcƒµIÊÔ!e¦‰‹2uÉ™ÿ ”©S>zZ4\”9|¡$*Än”©ãFt­(qsQ¦®£ñ¢‰ö@™:·E™1A™º—;´{ú´{ºQ¦4¢n]ùn]¹(ÓŽ£™Þ™(Ó'lÑîŽÝ&àK¦oÈ£¢ikïF™>ÒˆŠÒúQ.Êô•㨨ì[eúÒqôVµï­j”éS¶àò7fÉß0A™¾%:ˆ&ªeúX#+jd¹(Ó׎£íª©íªI`§¬áYÃé¨Åõž«×;Ñ(§}QN>hoXDoXøÚ¯ì¬_Y:ØS0Ø“#ñ_}N\Ø?®‚c¼ƒDË_Ž$ïU6åÚ¢Ìï&€ã[Ø>}A Ê v³WÕfP=tÿŽÀ±5à”pŽ[`Ư Î…`iöíjP¯t¨B3ÚÔ¿“}|kh( Ù‘’ ?‹,h›š¾||É;š þرúAOÂFЬçà¨OìUàŒ7ݳÌOú€æÏ?zš¬g¶Æî{iκuwôÑbGpvYòÎçEÅ 'SîQÈ@)8'Vq­wÂEpîK¾^%–à<·×ÉÕ­ ÅÊ’Çž Ö‚óåuîóó¢]•³W5ÜÔüsýÍ5iàÂ3õÇb‡î–~fwœxà"] Îxîz¼;ÊUúéØòyIªàâ áåÖ håË\šôÀ%º³"ãlÑpÙµ_mŠÐ·©é½OÀ¥+ƒ/yóÐLëç¿NÐJÖ¦62BuÕX¸¯­­¸fßÊ? .S›öVv–)ê³e”B}zÇ5·Ï?´ž2ûK’& h8im°z¬µÁ\®Å[ïÞ‚zEüf¦ˆÞÜ2µ3ç¸büˆÍ³–¢[¼~˜°i3š›³#¥p:xeÎì‚bpåÒcç:¾» Q3wüfµmÍ2Þ{²´Ñxp¦¥uù²âåwo4C`by}Èxz/ú³ép²ßaô·å‰RÎôòfͧ¥CèKƒ±õà­qOÙÉÔݽٿù+š’çs}Rú´æ´gâ¸J%me”Ÿj·n£¢R%ú÷OjC èU“¾¡ýãÑk³Vû‚vS¬ÂÌÓtÑ_dönõ=Ìï˜V?-rµ[e€vè{¦Úµ‚ö ºW²Ï£ó>y|%êž?Ñ6 êÝ8‹¥ŽëEõ|CŸê¯têWF'«ÎUA'>x&³k1ªé27ådÆÍèB—OçØ;x \µð’áýèöÝší±¨¿nçïíÞèa«~½¼Õ艆¥9'åÐ3%mfŠ©h¶ü£QТ[K΢•]é*æhƒ·_ÂOhë–£<ï{è³â_t9¢oüûd·¢ÿHÄë«¢ÿ4JÕ·@‡5—qÛ®–Þ{7{¨ •³HʘUt)ÐRVF'ö)ÌP{²;óGæÔ “¾¦,jh8˜ýÕ6\ØoŽN)‘j®…NÍsýðtTGÆq'Ç Zû1÷\Ð)Ýw®å Ú‡b²ù¨ÖáŸÇ_÷D'¿³µÌy†ª–ðrMÑIŸ¿^ ,C•NçDÖf£cÓ2ÂŒTPY•Þ©›æ€~{_4ˆ™ìôŸ2 íkJtÒVA»žh¬øˆvüÜ3nÐ}¤£àøµ ½·5DQF­‘¶ÖÒ®AËœ"B†Ð‹vwi çw*èFc¯%GË»¢á¾Ñ»+Р“¯&gý‰îV‰Iwÿ€n‘¾?" µu™}yE:ßÐÌè›&jìžáµä>ª®w/ª=ydEî ÐþýϦ†sÐZ×Ïö2òèYƒŒ¢þßëÝD×Ïè®ÚW²#bÐÃ9#:•A‡ýEOv;«ƒ«÷l1àœêQn‹Ê䔂kZœZ¯kJüë¶”h‘^´pú¿¾kM¼JJòçÚ3pÓœ¾=îêáì+¹=.Ù)ëêï¹së_oÊþo%Q8™¢+survival/tests/strata2.R0000644000175100001440000000126612030335371015012 0ustar hornikusers# # New tests 4/2010 to validate strata by covariate interactions # library(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) tdata <- lung tdata$sex <- lung$sex +3 # Both of these should produce warning messages about singular X, since there # are ph.ecog=3 subjects in only 1 of the strata. # Does not affect the test fit1 <- coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), lung) fit2 <- coxph(Surv(time, status) ~ age + sex:strata(ph.ecog), tdata) aeq(fit1$coef, fit2$coef) aeq(fit1$var, fit2$var) aeq(predict(fit1), predict(fit2)) survival/tests/rounding.R0000644000175100001440000000074211732700061015255 0ustar hornikuserslibrary(survival) # # Survival curves could fail with data that was almost exact. # The calculations use both unique() and table(), which don't # necessarily give the same number of values. # Check that the routine handles this properly # tdata <- data.frame(time=c(1,2, sqrt(2)^2, 2, sqrt(2)^2), status=rep(1,5), group=c(1,1,1,2,2)) fit <- survfit(Surv(time, status) ~ group, data=tdata) all.equal(sum(fit$strata), length(fit$time)) survival/tests/r_tdist.R0000644000175100001440000000217311732700061015100 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test out the t-distribution # capacitor <- read.table('data.capacitor', row.names=1, col.names=c('', 'days', 'event', 'voltage')) # First, a t-dist with 500 df should be nearly identical to the Gaussian fitig <- survreg(Surv(days, event)~voltage, dist = "gaussian", data = capacitor) fit1 <- survreg(Surv(days, event) ~ voltage, dist='t', parms=500, capacitor) fitig summary(fit1, corr=F) # A more realistic fit fit2 <- survreg(Surv(days, event) ~ voltage, dist='t', parms=5, capacitor) print(fit2) xx <- seq(1,125, by=10) resid(fit2, type='response')[xx] resid(fit2, type='deviance')[xx] resid(fit2, type='working') [xx] resid(fit2, type='dfbeta')[xx,] resid(fit2, type='dfbetas')[xx,] resid(fit2, type='ldresp')[xx] resid(fit2, type='ldshape')[xx] resid(fit2, type='ldcase')[xx] resid(fit2, type='matrix')[xx,] predict(fit2, type='response')[xx] predict(fit2, type='link')[xx] predict(fit2, type='terms')[xx,] predict(fit2, type='quantile')[xx] rm(fitig, fit1, fit2, xx) survival/tests/frank.R0000644000175100001440000000136312164373755014551 0ustar hornikuserslibrary(survival) # # Check out intercept/interaction for Frank H # age2 <- lung$age - 50 fit1 <- coxph(Surv(time, status) ~ age * strata(sex), lung) fit2 <- coxph(Surv(time, status) ~ age2*strata(sex), lung) tdata <- data.frame(age=50:60, age2=0:10, sex=c(1,2,1,2,1,2,1,2,1,2,1)) surv1 <- survfit(fit1, tdata) surv2 <- survfit(fit2, tdata) # The call won't match, but the rest should icall <- match("call", names(surv1)) all.equal(unclass(surv1)[-icall], unclass(surv2)[-icall]) # It should match what I get with a single strata fit fit3 <- coxph(Surv(time, status) ~ age, data=lung, init=fit1$coef[1], subset=(sex==1), iter=0) surv1b <- survfit(fit3, newdata=list(age=c(50,52, 54))) all.equal(c(surv1b$surv), surv1[c(1,3,5)]$surv) survival/tests/expected2.R0000644000175100001440000000107411732700061015312 0ustar hornikuserslibrary(survival) # # A Cox model with a factor, followed by survexp. # pfit2 <- coxph(Surv(time, status > 0) ~ trt + log(bili) + log(protime) + age + platelet + sex, data = pbc) esurv <- survexp(~ trt, ratetable = pfit2, data = pbc) temp <- pbc temp$sex2 <- factor(as.numeric(pbc$sex), levels=2:0, labels=c("f", "m", "unknown")) esurv2 <- survexp(~ trt, ratetable = pfit2, data = temp, rmap=list(sex=sex2)) # The call components won't match, which happen to be first all.equal(unclass(esurv)[-1], unclass(esurv2)[-1]) survival/tests/concordance.R0000644000175100001440000001032711732700061015706 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # # Simple tests of concordance. These numbers were derived in multiple # codes. # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) grank <- function(x, time, grp, wt) unlist(tapply(x, grp, rank)) grank2 <- function(x, time, grp, wt) { #for case weights if (length(wt)==0) wt <- rep(1, length(x)) z <- double(length(x)) for (i in unique(grp)) { indx <- which(grp==i) temp <- tapply(wt[indx], x[indx], sum) temp <- temp/2 + c(0, cumsum(temp)[-length(temp)]) z[indx] <- temp[match(x[indx], names(temp))] } z } tdata <- aml[aml$x=='Maintained',] tdata$y <- c(1,6,2,7,3,7,3,8,4,4,5) tdata$wt <- c(1,2,3,2,1,2,3,4,3,2,1) fit <- survConcordance(Surv(time, status) ~y, tdata) aeq(fit$stats[1:4], c(14,24,2,0)) cfit <- coxph(Surv(time, status) ~ tt(y), tdata, tt=grank, method='breslow', iter=0, x=T) cdt <- coxph.detail(cfit) aeq(4*sum(cdt$imat),fit$stats[5]^2) aeq(2*sum(cdt$score), diff(fit$stats[2:1])) # Lots of ties tempx <- Surv(c(1,2,2,2,3,4,4,4,5,2), c(1,0,1,0,1,0,1,1,0,1)) tempy <- c(5,5,4,4,3,3,7,6,5,4) fit2 <- survConcordance(tempx ~ tempy) aeq(fit2$stats[1:4], c(13,13,5,2)) cfit2 <- coxph(tempx ~ tt(tempy), tt=grank, method='breslow', iter=0) aeq(4/cfit2$var, fit2$stats[5]^2) # Bigger data fit3 <- survConcordance(Surv(time, status) ~ age, lung) aeq(fit3$stats[1:4], c(10717, 8706, 591, 28)) cfit3 <- coxph(Surv(time, status) ~ tt(age), lung, iter=0, method='breslow', tt=grank, x=T) cdt <- coxph.detail(cfit3) aeq(4*sum(cdt$imat),fit3$stats[5]^2) aeq(2*sum(cdt$score), diff(fit3$stats[2:1])) # More ties fit4 <- survConcordance(Surv(time, status) ~ ph.ecog, lung) aeq(fit4$stats[1:4], c(8392, 4258, 7137, 28)) cfit4 <- coxph(Surv(time, status) ~ tt(ph.ecog), lung, iter=0, method='breslow', tt=grank) aeq(4/cfit4$var, fit4$stats[5]^2) # Case weights fit5 <- survConcordance(Surv(time, status) ~ y, tdata, weight=wt) fit6 <- survConcordance(Surv(time, status) ~y, tdata[rep(1:11,tdata$wt),]) aeq(fit5$stats[1:4], c(70, 91, 7, 0)) # checked by hand aeq(fit5$stats[1:3], fit6$stats[1:3]) #spurious "tied on time" value, ignore aeq(fit5$std, fit6$std) cfit5 <- coxph(Surv(time, status) ~ tt(y), tdata, weight=wt, iter=0, method='breslow', tt=grank2) cfit6 <- coxph(Surv(time, status) ~ tt(y), tdata[rep(1:11,tdata$wt),], iter=0, method='breslow', tt=grank) aeq(4/cfit6$var, fit6$stats[5]^2) aeq(cfit5$var, cfit6$var) # Start, stop simplest cases fit7 <- survConcordance(Surv(rep(0,11), time, status) ~ y, tdata) aeq(fit7$stats, fit$stats) aeq(fit7$std.err, fit$std.err) fit7 <- survConcordance(Surv(rep(0,11), time, status) ~ y, tdata, weight=wt) aeq(fit5$stats, fit7$stats) # Multiple intervals for some, but same risk sets as tdata tdata2 <- data.frame(time1=c(0,3, 5, 6,7, 0, 4,17, 7, 0,16, 2, 0, 0,9, 5), time2=c(3,9, 13, 7,13, 18, 17,23, 28, 16,31, 34, 45, 9,48, 60), status=c(0,1, 1, 0,0, 1, 0,1, 0, 0,1, 1, 0, 0,1, 0), y = c(1,1, 6, 2,2, 7, 3,3, 7, 3,3, 8, 4, 4,4, 5), wt= c(1,1, 2, 3,3, 2, 1,1, 2, 3,3, 4, 3, 2,2, 1)) fit8 <- survConcordance(Surv(time1, time2, status) ~y, tdata2, weight=wt) aeq(fit5$stats, fit8$stats) aeq(fit5$std.err, fit8$std.err) cfit8 <- coxph(Surv(time1, time2, status) ~ tt(y), tdata2, weight=wt, iter=0, method='breslow', tt=grank2) aeq(4/cfit8$var, fit8$stats[5]^2) aeq(fit8$stats[5]/(2*sum(fit8$stats[1:3])), fit8$std.err) # Stratified tdata3 <- data.frame(time1=c(tdata2$time1, rep(0, nrow(lung))), time2=c(tdata2$time2, lung$time), status = c(tdata2$status, lung$status -1), x = c(tdata2$y, lung$ph.ecog), wt= c(tdata2$wt, rep(1, nrow(lung))), grp=rep(1:2, c(nrow(tdata2), nrow(lung)))) fit9 <- survConcordance(Surv(time1, time2, status) ~x + strata(grp), data=tdata3, weight=wt) aeq(fit9$stats[1,], fit5$stats) aeq(fit9$stats[2,], fit4$stats) survival/tests/book4.R0000644000175100001440000000630612350317443014455 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Tests from the appendix of Therneau and Grambsch # d. Data set 2 and Efron estimate # test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) byhand <- function(beta, newx=0) { r <- exp(beta) loglik <- 4*beta - (log(r+1) + log(r+2) + 2*log(3*r+2) + 2*log(3*r+1) + log(2*r +2)) u <- 1/(r+1) + 1/(3*r+1) + 2*(1/(3*r+2) + 1/(2*r+2)) - ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) imat <- r*(1/(r+1)^2 + 2/(r+2)^2 + 6/(3*r+2)^2 + 6/(3*r+1)^2 + 6/(3*r+2)^2 + 4/(2*r +2)^2) hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), 1/(3*r+2), 1/(2*r +2) ) # The matrix of weights, one row per obs, one col per time # deaths at 2,3,6,7,8,9 wtmat <- matrix(c(1,0,0,0,1, 0, 0,0,0,0, 0,1,0,1,1, 0, 0,0,0,0, 0,0,1,1,1, 0, 1,1,0,0, 0,0,0,1,1, 0, 1,1,0,0, 0,0,0,0,1, 1, 1,1,0,0, 0,0,0,0,0, 1, 1,1,1,1, 0,0,0,0,0,.5,.5,1,1,1), ncol=7) wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat x <- c(1,0,0,1,0,1,1,1,0,0) status <- c(1,1,1,1,1,1,1,0,0,0) xbar <- colSums(wtmat*x)/ colSums(wtmat) n <- length(x) # Table of sums for score and Schoenfeld resids hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time dM <- -hazmat #Expected part for (i in 1:5) dM[i,i] <- dM[i,i] +1 #observed dM[6:7,6:7] <- dM[6:7,6:7] +.5 # observed mart <- rowSums(dM) # Table of sums for score and Schoenfeld resids # Looks like the last table of appendix E.2.1 of the book resid <- dM * outer(x, xbar, '-') score <- rowSums(resid) scho <- colSums(resid) # We need to add the ties back up (they are symmetric) scho[6:7] <- rep(mean(scho[6:7]), 2) list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard, mart=mart, score=score, rmat=resid, scho=scho) } aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) fit0 <-coxph(Surv(start, stop, event) ~x, test2, iter=0) truth0 <- byhand(0,0) aeq(truth0$loglik, fit0$loglik[1]) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')) fit <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-8) truth <- byhand(fit$coef, 0) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')) # # Done with the formal test, now print out lots of bits # resid(fit) resid(fit, 'scor') resid(fit, 'scho') predict(fit, type='lp') predict(fit, type='risk') predict(fit, type='expected') predict(fit, type='terms') predict(fit, type='lp', se.fit=T) predict(fit, type='risk', se.fit=T) predict(fit, type='expected', se.fit=T) predict(fit, type='terms', se.fit=T) summary(survfit(fit)) summary(survfit(fit, list(x=2))) survival/tests/plot.R0000644000175100001440000000136112334226056014412 0ustar hornikusers# # This is a test of many plots. They get saved as xxx.pdf, and then compared # to xxx.pdf.save using tools:Rdiff. Pdf files have a header that contains # a date, but the Rdiff sniffs that out and ignores it. # Per the example in the test directory of R sources, use compress=FALSE # on the pdf to keep it as ascii. library(survival) library(tools) # survfit curves pdf("plot1.pdf", compress=FALSE) fit1 <- survfit(Surv(time, status) ~ ph.ecog, data=lung) plot(fit1, mark=1:4) plot(fit1, conf.int=T, fun="event", col=1:4) plot(fit1, mark.time=FALSE, fun="cloglog") lines(fit1[2], mark.time=FALSE, conf.int='only', col=3, lty=2, fun='cloglog') #points(fit1[3], col=6) dev.off() if (Rdiff("plot1.pdf", "plot1.pdf.save")) cat("plot1 differs") survival/tests/turnbull.R0000644000175100001440000001367011732700061015303 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # The test data set from Turnbull, JASA 1974, 169-73. # # status 0=right censored # 1=exact # 2=left censored # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) turnbull <- data.frame( time =c( 1,1,1, 2,2,2, 3,3,3, 4,4,4), status=c( 1,0,2, 1,0,2, 1,0,2, 1,0,2), n =c(12,3,2, 6,2,4, 2,0,2, 3,3,5)) # # Compute the K-M for the Turnbull data # via a slow EM calculation # emsurv <- function(time, status, wt, verbose=T) { left.cen <- (status==2) if (!any(left.cen)) stop("No left censored data!") if (!any(status==1))stop("Must have some exact death times") tempy <- Surv(time[!left.cen], status[!left.cen]) ww <- wt[!left.cen] tempx <- factor(rep(1, sum(!left.cen))) tfit <- survfit(tempy~tempx, weight=ww) if (verbose) cat("Iteration 0, survival=", format(round(tfit$surv[tfit$n.event>0],3)), "\n") stimes <- tfit$time[tfit$n.event>0] ltime <- time[left.cen] lwt <- wt[left.cen] tempx <- factor(rep(1, length(stimes) + sum(!left.cen))) tempy <- Surv(c(time[!left.cen], stimes), c(status[!left.cen], rep(1, length(stimes)))) for (iter in 1:4) { wt2 <- stimes*0 ssurv <- tfit$surv[tfit$n.event>0] sjump <- diff(c(1, ssurv)) for (j in 1:(length(ltime))) { k <- sum(ltime[j]>=stimes) #index of the death time if (k==0) stop("Left censored observation before the first death") wt2[1:k] <- wt2[1:k] + lwt[j]*sjump[1:k] /(ssurv[k]-1) } tfit <- survfit(tempy~tempx, weight=c(ww, wt2)) if (verbose) { cat("Iteration", iter, "survival=", format(round(tfit$surv[tfit$n.event>0],3)), "\n") cat(" weights=", format(round(wt2,3)), "\n") } } survfit(tempy ~ tempx, weights=c(ww, wt2)) } temp <-emsurv(turnbull$time, turnbull$status, turnbull$n) print(summary(temp)) # First check, use the data from Turnbull, JASA 1974, 169-173. tdata <- data.frame(time =c(1,1,1,2,2,2,3,3,3,4,4,4), status=rep(c(1,0,2),4), n =c(12,3,2,6,2,4,2,0,2,3,3,5)) tfit <- survfit(Surv(time, time, status, type='interval') ~1, tdata, weight=n) all.equal(round(tfit$surv,3), c(.538, .295, .210, .095)) # Second check, compare to a reversed survival curve # This is not as simple a test as one might think, because left and right # censored observations are not treated symmetrically by the routine: # time <= y for left and time> y for right (this is to make the routine # correct for the common situation of panel data). # To get equivalence, make the left censoreds happen just a little bit # earlier. The left-continuous/right-continuous shift is also a bother. # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) fit1 <- survfit(Surv(time, status) ~1, test1) temp <- ifelse(test1$status==0, 4.99,5) - test1$time fit2 <- survfit(Surv(temp, status, type='left') ~1, test1) all.equal(round(fit1$surv[1:2],5), round(1-fit2$surv[3:2],5)) rm(tdata, tfit, fit1, temp, fit2) # # Create a data set similar to the one provided by Al Zinsmeister # It is a hard test case for survfit.turnbull # time1 <- c(rep(0,100), rep(1,200), 100, 200, 210, 220, rep(365,100), rep(366,5), 731:741) time2 <- c((1:100)*3, 10+1:100, rep(365:366, c(60,40)), NA, 500, NA, 450, rep(730,90), rep(NA,10), c(528,571,691,730,731), NA, 1095:1099, NA, 1400, 1200, 772, 1461) zfit <- survfit(Surv(time1, time2, type='interval2') ~1) # # There are 100 intervals of the form (0,x) where x is from 3 to 300, # and 200 more of the form (1,x) where x is from 11 to 366. These # lead to a mass point in the interval (1,3), which is placed at 2. # The starting estimate has far too little mass placed here, and it takes # the EM a long time to realize that most of the weight for the first 300 # subjects goes here. With acceleration, it takes 16 iterations, without # it takes >40. (On Al's orginal data, without accel still wasn't there after # 165 iters!) # # The next 4 obs give rise to potential jumps at 100.5, 200.5, 211.5, and # 221. However, the final estimate has no mass at all on any of these. # Assume mass of a,b, and c at 2, 100.5 and 365.5, and consider the # contributions: # 123 obs that overlap a only # 137 obs that overlap a and b # 40 obs that overlap a, b, c # 1 obs that overlap b, c # 108 obs that overlap c (200, 210,200, 365, and 366 starting points) # For some trial values of a,b,c, compare the loglik to that of (a+b),0,c # First one: a^123 (a+b)^137 (a+b+c)^40 (b+c) c^108 # Second: (a+b)^123 (a+b)^137 (a+b+c)^40 c c^108 # Likelhood improves if (1 + b/a)^123 > 1+ b/c, which is true for almost # all a and c. In particular, at the solution a and c are approx .7 and # .18, respectively. # # The program can't see this coming, of course, and so iterates towards a # KM with epsilon sized jumps at 100.5, 200.5, and 211.5. Whether these # intervals should be removed during iteration, as detected, is an open # question for me. # # # True solution: mass points at 2, 365.5, 408, and 756.5, of sizes a, b, c, d # Likelihood: a^260 (a+b)^40 (b+c)^92 (b+c+d)^12 c^5 d^11 # Solution: a=0.6958, b=0.1674, c=0.1079, d=0.0289 tfun <- function(x) { if (length(x) ==3) x <- c(x, .03) x <- x/sum(x) #make probabilities sum to 1 loglik <- 260*log(x[1]) + 40*log(x[1]+x[2]) + 92*log(x[2] + x[3]) + 12*log(x[2]+x[3]+x[4]) + 5*log(x[3]) + 11*log(x[4]) -loglik #find the max, not the min } nfit <- nlminb(start=c(.7,.15, .1), tfun, lower=0, upper=1) nparm <- c(nfit$par, .03) nparm <- nparm / sum(nparm) zparm <- -diff(c(1, zfit$surv[match(c(2, 365.5, 408, 756.5), zfit$time)])) aeq(round(tfun(nparm),4), round(tfun(zparm),4)) # .0001 is the tolerance in survfit.turnbull rm(tfun, nfit, nparm, zparm, time1, time2, zfit) survival/tests/coxsurv4.Rout.save0000644000175100001440000000511612160143136016712 0ustar hornikusers R version 3.0.1 (2013-05-16) -- "Good Sport" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > > # Strata by covariate interactions, a case pointed out in early 2011 > # by Frank Harrell, which as it turns out had never been computed > # correctly by any version of the package. Which shows how often this > # case arises in practice. > # > aeq <- function(x, y, ...) all.equal(as.vector(x), as.vector(y)) > fit1 <- coxph(Surv(time, status) ~ wt.loss + age*strata(sex) + strata(ph.ecog), + data=lung) > tdata <- data.frame(wt.loss=c(10,5,0,10, 15,20,25), + age =c(50,60,50,60,70,40,21), + sex =c(1,1,2,2,1,1,1), + ph.ecog=c(0,0,1,1,2,2,2)) > surv1 <- survfit(fit1, newdata=tdata) > > fit2 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*0), data=lung, + init=fit1$coef, iter=0, subset=(sex==1 & ph.ecog==0)) > fit2$var <- fit1$var > > surv2 <- survfit(fit2, newdata=list(wt.loss=c(10,5), age=c(50,60))) > s1 <- surv1[1:2] > aeq(s1$surv, surv2$surv) #first a vector, second a matrix [1] TRUE > aeq(s1$std.err, surv2$std.err) [1] TRUE > aeq(s1[1]$time, surv2$time) [1] TRUE > aeq(s1[1]$n.event, surv2$n.event) [1] TRUE > > fit3 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*1), + data=lung, init=fit1$coef, iter=0, + subset=(sex==2 & ph.ecog==1)) > fit3$var <- fit1$var > surv3 <- survfit(fit3, newdata=list(wt.loss=c(0,10), age=c(50,60))) > aeq(surv1[3:4]$surv, surv3$surv) [1] TRUE > aeq(surv1[3:4]$std, surv3$std) [1] TRUE > > fit4 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*0), + data=lung, init=fit1$coef, iter=0, + subset=(sex==1 & ph.ecog==2)) > fit4$var <- fit1$var > surv4 <- survfit(fit4, newdata=list(wt.loss=c(15,20,25), age=c(70,40,21))) > > aeq(surv1[5:7]$surv, surv4$surv) [1] TRUE > aeq(surv1[5:7]$std.err, surv4$std.err) [1] TRUE > aeq(surv1[5]$n.risk, surv4$n.risk) [1] TRUE > > > proc.time() user system elapsed 0.324 0.052 0.356 survival/tests/r_strata.Rout.save0000644000175100001440000001146011732700061016733 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test out the strata capabilities > # > tol <- survreg.control()$rel.tolerance > aeq <- function(x,y,...) all.equal(as.vector(x), as.vector(y), ...) > > # intercept only models > fit1 <- survreg(Surv(time, status) ~ strata(sex), lung) > fit2 <- survreg(Surv(time, status) ~ strata(sex) + sex, lung) > fit3a<- survreg(Surv(time,status) ~1, lung, subset=(sex==1)) > fit3b<- survreg(Surv(time,status) ~1, lung, subset=(sex==2)) > > fit1 Call: survreg(formula = Surv(time, status) ~ strata(sex), data = lung) Coefficients: (Intercept) 6.062171 Scale: sex=1 sex=2 0.8167551 0.6533036 Loglik(model)= -1152.5 Loglik(intercept only)= -1152.5 n= 228 > fit2 Call: survreg(formula = Surv(time, status) ~ strata(sex) + sex, data = lung) Coefficients: (Intercept) sex 5.494409 0.380171 Scale: sex=1 sex=2 0.8084294 0.6355816 Loglik(model)= -1147.1 Loglik(intercept only)= -1152.5 Chisq= 10.9 on 1 degrees of freedom, p= 0.00096 n= 228 > aeq(fit2$scale, c(fit3a$scale, fit3b$scale), tolerance=tol) [1] TRUE > aeq(fit2$loglik[2], (fit3a$loglik + fit3b$loglik)[2], tolerance=tol) [1] TRUE > aeq(fit2$coef[1] + 1:2*fit2$coef[2], c(fit3a$coef, fit3b$coef), tolerance=tol) [1] TRUE > > #penalized models > fit1 <- survreg(Surv(time, status) ~ pspline(age, theta=.92)+ + strata(sex), lung) > fit2 <- survreg(Surv(time, status) ~ pspline(age, theta=.92)+ + strata(sex) + sex, lung) > fit1 Call: survreg(formula = Surv(time, status) ~ pspline(age, theta = 0.92) + strata(sex), data = lung) coef se(coef) se2 Chisq DF p (Intercept) 6.9036 0.8469 0.5688 66.45 1.00 3.3e-16 pspline(age, theta = 0.92 -0.0124 0.0067 0.0067 3.45 1.00 6.3e-02 pspline(age, theta = 0.92 2.53 2.65 4.0e-01 Scale: sex=1 sex=2 0.807 0.654 Iterations: 1 outer, 4 Newton-Raphson Theta= 0.92 Degrees of freedom for terms= 0.5 3.6 2.0 Likelihood ratio test=6.54 on 3.1 df, p=0.0937 n= 228 > fit2 Call: survreg(formula = Surv(time, status) ~ pspline(age, theta = 0.92) + strata(sex) + sex, data = lung) coef se(coef) se2 Chisq DF p (Intercept) 6.3729 0.84471 0.59118 56.92 1.00 4.5e-14 pspline(age, theta = 0.92 -0.0111 0.00666 0.00666 2.77 1.00 9.6e-02 pspline(age, theta = 0.92 2.46 2.68 4.2e-01 sex 0.3686 0.11711 0.11685 9.91 1.00 1.6e-03 Scale: sex=1 sex=2 0.800 0.636 Iterations: 1 outer, 5 Newton-Raphson Theta= 0.92 Degrees of freedom for terms= 0.5 3.7 1.0 2.0 Likelihood ratio test=16.8 on 4.2 df, p=0.00245 n= 228 > > age1 <- ifelse(lung$sex==1, lung$age, mean(lung$age)) > age2 <- ifelse(lung$sex==2, lung$age, mean(lung$age)) > fit3 <- survreg(Surv(time,status) ~ pspline(age1, theta=.92) + + pspline(age2, theta=.95) + sex + strata(sex), lung) > fit3a<- survreg(Surv(time,status) ~pspline(age, theta=.92), lung, + subset=(sex==1)) > fit3b<- survreg(Surv(time,status) ~pspline(age, theta=.95), lung, + subset=(sex==2)) > fit3b<- survreg(Surv(time,status) ~pspline(age, theta=.95), + lung[lung$sex==2,], x=T) > # > # The above line is tricky, and it took me a long time to realize > # it's necessity. The range of age1 = range(age) = 39-82. That for > # age2 = range of females = 41-77. The basis functions for pspline are > # based on age. If I used data=lung, subset=(sex==2) in fit3b (earlier > # form of the test, the pspline function is called before the subset > # occurs, and fit3b has a different basis for the second spline than > # fit3 does; leading to failure of the all.equal tests below. A theta > # of .95 on one basis is not exactly the same as a theta of .95 on the > # other. Coefficients were within 1%, but not the same. > > aeq(fit3$scale, c(fit3a$scale, fit3b$scale)) [1] TRUE > aeq(fit3$loglik[2], (fit3a$loglik + fit3b$loglik)[2]) [1] TRUE > pred <- predict(fit3) > aeq(pred[lung$sex==1] , predict(fit3a)) [1] TRUE > aeq(pred[lung$sex==2], predict(fit3b)) [1] TRUE > > > > > survival/tests/r_strata.R0000644000175100001440000000443711732700061015254 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test out the strata capabilities # tol <- survreg.control()$rel.tolerance aeq <- function(x,y,...) all.equal(as.vector(x), as.vector(y), ...) # intercept only models fit1 <- survreg(Surv(time, status) ~ strata(sex), lung) fit2 <- survreg(Surv(time, status) ~ strata(sex) + sex, lung) fit3a<- survreg(Surv(time,status) ~1, lung, subset=(sex==1)) fit3b<- survreg(Surv(time,status) ~1, lung, subset=(sex==2)) fit1 fit2 aeq(fit2$scale, c(fit3a$scale, fit3b$scale), tolerance=tol) aeq(fit2$loglik[2], (fit3a$loglik + fit3b$loglik)[2], tolerance=tol) aeq(fit2$coef[1] + 1:2*fit2$coef[2], c(fit3a$coef, fit3b$coef), tolerance=tol) #penalized models fit1 <- survreg(Surv(time, status) ~ pspline(age, theta=.92)+ strata(sex), lung) fit2 <- survreg(Surv(time, status) ~ pspline(age, theta=.92)+ strata(sex) + sex, lung) fit1 fit2 age1 <- ifelse(lung$sex==1, lung$age, mean(lung$age)) age2 <- ifelse(lung$sex==2, lung$age, mean(lung$age)) fit3 <- survreg(Surv(time,status) ~ pspline(age1, theta=.92) + pspline(age2, theta=.95) + sex + strata(sex), lung) fit3a<- survreg(Surv(time,status) ~pspline(age, theta=.92), lung, subset=(sex==1)) fit3b<- survreg(Surv(time,status) ~pspline(age, theta=.95), lung, subset=(sex==2)) fit3b<- survreg(Surv(time,status) ~pspline(age, theta=.95), lung[lung$sex==2,], x=T) # # The above line is tricky, and it took me a long time to realize # it's necessity. The range of age1 = range(age) = 39-82. That for # age2 = range of females = 41-77. The basis functions for pspline are # based on age. If I used data=lung, subset=(sex==2) in fit3b (earlier # form of the test, the pspline function is called before the subset # occurs, and fit3b has a different basis for the second spline than # fit3 does; leading to failure of the all.equal tests below. A theta # of .95 on one basis is not exactly the same as a theta of .95 on the # other. Coefficients were within 1%, but not the same. aeq(fit3$scale, c(fit3a$scale, fit3b$scale)) aeq(fit3$loglik[2], (fit3a$loglik + fit3b$loglik)[2]) pred <- predict(fit3) aeq(pred[lung$sex==1] , predict(fit3a)) aeq(pred[lung$sex==2], predict(fit3b)) survival/tests/book4.Rout.save0000644000175100001440000001754111732700061016140 0ustar hornikusers R version 2.14.0 Under development (unstable) (2011-04-10 r55401) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Tests from the appendix of Therneau and Grambsch > # d. Data set 2 and Efron estimate > # > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > byhand <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 4*beta - (log(r+1) + log(r+2) + 2*log(3*r+2) + 2*log(3*r+1) + + log(2*r +2)) + u <- 1/(r+1) + 1/(3*r+1) + 2*(1/(3*r+2) + 1/(2*r+2)) - + ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) + imat <- r*(1/(r+1)^2 + 2/(r+2)^2 + 6/(3*r+2)^2 + + 6/(3*r+1)^2 + 6/(3*r+2)^2 + 4/(2*r +2)^2) + + hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), + 1/(3*r+2), 1/(2*r +2) ) + + + # The matrix of weights, one row per obs, one col per time + # deaths at 2,3,6,7,8,9 + wtmat <- matrix(c(1,0,0,0,1, 0, 0,0,0,0, + 0,1,0,1,1, 0, 0,0,0,0, + 0,0,1,1,1, 0, 1,1,0,0, + 0,0,0,1,1, 0, 1,1,0,0, + 0,0,0,0,1, 1, 1,1,0,0, + 0,0,0,0,0, 1, 1,1,1,1, + 0,0,0,0,0,.5,.5,1,1,1), ncol=7) + wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat + + x <- c(1,0,0,1,0,1,1,1,0,0) + status <- c(1,1,1,1,1,1,1,0,0,0) + xbar <- colSums(wtmat*x)/ colSums(wtmat) + n <- length(x) + + # Table of sums for score and Schoenfeld resids + hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time + dM <- -hazmat #Expected part + for (i in 1:5) dM[i,i] <- dM[i,i] +1 #observed + dM[6:7,6:7] <- dM[6:7,6:7] +.5 # observed + mart <- rowSums(dM) + + # Table of sums for score and Schoenfeld resids + # Looks like the last table of appendix E.2.1 of the book + resid <- dM * outer(x, xbar, '-') + score <- rowSums(resid) + scho <- colSums(resid) + + # We need to add the ties back up (they are symmetric) + scho[6:7] <- rep(mean(scho[6:7]), 2) + + list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard, + mart=mart, score=score, rmat=resid, + scho=scho) + } > > > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > fit0 <-coxph(Surv(start, stop, event) ~x, test2, iter=0) > truth0 <- byhand(0,0) > aeq(truth0$loglik, fit0$loglik[1]) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')) [1] TRUE > > > fit <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-8) > truth <- byhand(fit$coef, 0) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')) [1] TRUE > > # > # Done with the formal test, now print out lots of bits > # > resid(fit) 1 2 3 4 5 6 0.50527611 0.66432995 0.79746211 0.22435805 -0.55144018 0.42933697 7 8 9 10 -0.01764508 -1.14132605 -0.45517594 -0.45517594 > resid(fit, 'scor') 1 2 3 4 5 6 7 0.2553039 -0.2183386 -0.4744295 -0.1101520 0.1137126 0.2491954 0.1057078 8 9 10 -0.4119611 0.2454808 0.2454808 > resid(fit, 'scho') 2 3 6 7 8 9 9 0.5052761 -0.3286599 -0.5949242 0.2539781 -0.7460219 0.4551759 0.4551759 > > predict(fit, type='lp') [1] -0.0105526 0.0105526 0.0105526 -0.0105526 0.0105526 -0.0105526 [7] -0.0105526 -0.0105526 0.0105526 0.0105526 > predict(fit, type='risk') [1] 0.9895029 1.0106085 1.0106085 0.9895029 1.0106085 0.9895029 0.9895029 [8] 0.9895029 1.0106085 1.0106085 > predict(fit, type='expected') 1 2 3 4 5 6 7 8 0.4947239 0.3356701 0.2025379 0.7756420 1.5514402 0.5706630 1.0176451 1.1413261 9 10 0.4551759 0.4551759 > predict(fit, type='terms') x 1 -0.0105526 2 0.0105526 3 0.0105526 4 -0.0105526 5 0.0105526 6 -0.0105526 7 -0.0105526 8 -0.0105526 9 0.0105526 10 0.0105526 attr(,"constant") [1] -0.0105526 > predict(fit, type='lp', se.fit=T) $fit 1 2 3 4 5 6 7 -0.0105526 0.0105526 0.0105526 -0.0105526 0.0105526 -0.0105526 -0.0105526 8 9 10 -0.0105526 0.0105526 0.0105526 $se.fit 1 2 3 4 5 6 7 8 0.3975884 0.3975884 0.3975884 0.3975884 0.3975884 0.3975884 0.3975884 0.3975884 9 10 0.3975884 0.3975884 > predict(fit, type='risk', se.fit=T) $fit 1 2 3 4 5 6 7 8 0.9895029 1.0106085 1.0106085 0.9895029 1.0106085 0.9895029 0.9895029 0.9895029 9 10 1.0106085 1.0106085 $se.fit 1 2 3 4 5 6 7 8 0.3954962 0.3996918 0.3996918 0.3954962 0.3996918 0.3954962 0.3954962 0.3954962 9 10 0.3996918 0.3996918 > predict(fit, type='expected', se.fit=T) $fit 1 2 3 4 5 6 7 8 0.4947239 0.3356701 0.2025379 0.7756420 1.5514402 0.5706630 1.0176451 1.1413261 9 10 0.4551759 0.4551759 $se.fit [1] 0.5331623 0.3940109 0.3241963 0.6388491 1.0026838 0.6453101 0.7848594 [8] 0.7848594 0.6401915 0.6401915 > predict(fit, type='terms', se.fit=T) $fit x 1 -0.0105526 2 0.0105526 3 0.0105526 4 -0.0105526 5 0.0105526 6 -0.0105526 7 -0.0105526 8 -0.0105526 9 0.0105526 10 0.0105526 attr(,"constant") [1] -0.0105526 $se.fit x 1 0.3975884 2 0.3975884 3 0.3975884 4 0.3975884 5 0.3975884 6 0.3975884 7 0.3975884 8 0.3975884 9 0.3975884 10 0.3975884 > > summary(survfit(fit)) Call: survfit(formula = fit) time n.risk n.event survival std.err lower 95% CI upper 95% CI 2 2 1 0.607 0.303 0.2277 1.000 3 3 1 0.435 0.262 0.1337 1.000 6 5 1 0.356 0.226 0.1029 1.000 7 4 1 0.277 0.189 0.0729 1.000 8 4 1 0.215 0.157 0.0516 0.899 9 5 2 0.137 0.109 0.0288 0.655 > summary(survfit(fit, list(x=2))) Call: survfit(formula = fit, newdata = list(x = 2)) time n.risk n.event survival std.err lower 95% CI upper 95% CI 2 2 1 0.616 0.465 0.14013 1 3 3 1 0.447 0.519 0.04568 1 6 5 1 0.368 0.504 0.02512 1 7 4 1 0.288 0.464 0.01232 1 8 4 1 0.226 0.418 0.00603 1 9 5 2 0.146 0.343 0.00147 1 > survival/tests/r_tdist.Rout.save0000644000175100001440000001743011732700061016567 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test out the t-distribution > # > > capacitor <- read.table('data.capacitor', row.names=1, + col.names=c('', 'days', 'event', 'voltage')) > # First, a t-dist with 500 df should be nearly identical to the Gaussian > > fitig <- survreg(Surv(days, event)~voltage, + dist = "gaussian", data = capacitor) > fit1 <- survreg(Surv(days, event) ~ voltage, + dist='t', parms=500, capacitor) > fitig Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "gaussian") Coefficients: (Intercept) voltage 1764.93485 -53.87917 Scale= 121.4319 Loglik(model)= -361.9 Loglik(intercept only)= -420.1 Chisq= 116.33 on 1 degrees of freedom, p= 0 n= 125 > summary(fit1, corr=F) Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "t", parms = 500) Value Std. Error z p (Intercept) 1765.8 163.137 10.82 2.64e-27 voltage -53.9 5.536 -9.74 2.06e-22 Log(scale) 4.8 0.106 45.44 0.00e+00 Scale= 121 Student-t distribution: parmameters= 500 Loglik(model)= -361.9 Loglik(intercept only)= -420.1 Chisq= 116.48 on 1 degrees of freedom, p= 0 Number of Newton-Raphson Iterations: 6 n= 125 > > # A more realistic fit > fit2 <- survreg(Surv(days, event) ~ voltage, + dist='t', parms=5, capacitor) > print(fit2) Call: survreg(formula = Surv(days, event) ~ voltage, data = capacitor, dist = "t", parms = 5) Coefficients: (Intercept) voltage 1819.28554 -55.74915 Scale= 96.84073 Loglik(model)= -360.4 Loglik(intercept only)= -424.7 Chisq= 128.55 on 1 degrees of freedom, p= 0 n= 125 > > xx <- seq(1,125, by=10) > resid(fit2, type='response')[xx] 1 11 21 31 41 51 61 -404.30257 -404.30257 -404.30257 -98.07767 -69.80767 -69.80767 -69.80767 71 81 91 101 111 121 -69.80767 -93.94023 97.43977 113.88722 -33.24278 -20.67278 > resid(fit2, type='deviance')[xx] 1 11 21 31 41 51 61 0.0933622 0.0933622 0.0933622 -2.2347398 0.7614136 0.7614136 0.7614136 71 81 91 101 111 121 0.7614136 -2.2156512 1.8511554 2.3107823 -2.0035571 -1.9821491 > resid(fit2, type='working') [xx] 1 11 21 31 41 51 61 86.38692 86.38692 86.38692 -148.70263 83.43717 83.43717 83.43717 71 81 91 101 111 121 83.43717 -137.49634 467.64123 200.98252 -34.84748 -21.05308 > resid(fit2, type='dfbeta')[xx,] (Intercept) voltage Log(scale) 1 0.2105054 -0.00703909 -1.743331e-04 11 0.2105054 -0.00703909 -1.743331e-04 21 0.2105054 -0.00703909 -1.743331e-04 31 -29.7982886 0.93975839 -1.076889e-02 41 9.6554540 -0.30502561 3.507039e-05 51 9.6554540 -0.30502561 3.507039e-05 61 9.6554540 -0.30502561 3.507039e-05 71 9.6554540 -0.30502561 3.507039e-05 81 -7.9425298 0.20791541 -6.194989e-03 91 16.3379622 -0.46035101 2.516742e-02 101 -13.8131372 0.53202477 4.665894e-03 111 0.7894992 -0.06147045 -1.308494e-02 121 -1.7672591 0.03187567 -1.433810e-02 > resid(fit2, type='dfbetas')[xx,] [,1] [,2] [,3] 1 0.001445482 -0.001447807 -0.0014173482 11 0.001445482 -0.001447807 -0.0014173482 21 0.001445482 -0.001447807 -0.0014173482 31 -0.204616568 0.193290466 -0.0875522695 41 0.066301320 -0.062737980 0.0002851263 51 0.066301320 -0.062737980 0.0002851263 61 0.066301320 -0.062737980 0.0002851263 71 0.066301320 -0.062737980 0.0002851263 81 -0.054539145 0.042764254 -0.0503659665 91 0.112188247 -0.094685467 0.2046140370 101 -0.094850975 0.109427398 0.0379342550 111 0.005421271 -0.012643305 -0.1063820631 121 -0.012135277 0.006556221 -0.1165704260 > resid(fit2, type='ldresp')[xx] 1 11 21 31 41 51 6.303033e-06 6.303033e-06 6.303033e-06 4.198946e-02 1.121526e-02 1.121526e-02 61 71 81 91 101 111 1.121526e-02 1.121526e-02 3.796054e-02 3.773652e-02 5.409081e-02 4.663892e-02 121 4.455789e-02 > resid(fit2, type='ldshape')[xx] 1 11 21 31 41 51 8.281125e-05 8.281125e-05 8.281125e-05 1.355729e-01 1.789400e-04 1.789400e-04 61 71 81 91 101 111 1.789400e-04 1.789400e-04 6.346182e-02 9.934752e-02 1.534546e-01 1.958545e-02 121 7.748320e-03 > resid(fit2, type='ldcase')[xx] 1 11 21 31 41 51 6.114509e-06 6.114509e-06 6.114509e-06 5.563427e-02 6.706055e-03 6.706055e-03 61 71 81 91 101 111 6.706055e-03 6.706055e-03 1.966021e-02 6.803951e-02 3.806159e-02 1.617087e-02 121 1.551988e-02 > resid(fit2, type='matrix')[xx,] g dg ddg ds dds 1 -0.00435825 4.361059e-05 -5.048286e-07 -0.01763187 -0.06488770 11 -0.00435825 4.361059e-05 -5.048286e-07 -0.01763187 -0.06488770 21 -0.00435825 4.361059e-05 -5.048286e-07 -0.01763187 -0.06488770 31 -6.10147902 -1.041351e-02 -7.002908e-05 0.02133278 -1.69495867 41 -0.28987533 3.893573e-03 -4.666473e-05 -0.27180126 0.04439884 51 -0.28987533 3.893573e-03 -4.666473e-05 -0.27180126 0.04439884 61 -0.28987533 3.893573e-03 -4.666473e-05 -0.27180126 0.04439884 71 -0.28987533 3.893573e-03 -4.666473e-05 -0.27180126 0.04439884 81 -6.05900320 -1.011644e-02 -7.357605e-05 -0.04965962 -1.59963182 91 -1.71338808 1.250705e-02 -2.674498e-05 1.21868456 -1.47261500 101 -6.27430559 1.141518e-02 -5.679687e-05 0.30004293 -2.03671532 111 -5.61156875 -4.155718e-03 -1.192545e-04 -0.86185235 -0.26993370 121 -5.56890563 -2.621343e-03 -1.245112e-04 -0.94580955 -0.10740203 dsg 1 0.0001604929 11 0.0001604929 21 0.0001604929 31 0.0172817991 41 -0.0006360167 51 -0.0006360167 61 -0.0006360167 71 -0.0006360167 81 0.0170281873 91 -0.0151130795 101 -0.0178836167 111 0.0081200691 121 0.0051953348 > > predict(fit2, type='response')[xx] [1] 704.30257 704.30257 704.30257 369.80767 369.80767 369.80767 369.80767 [8] 369.80767 202.56023 202.56023 35.31278 35.31278 35.31278 > predict(fit2, type='link')[xx] [1] 704.30257 704.30257 704.30257 369.80767 369.80767 369.80767 369.80767 [8] 369.80767 202.56023 202.56023 35.31278 35.31278 35.31278 > predict(fit2, type='terms')[xx,] 1 11 21 31 41 51 61 374.63428 374.63428 374.63428 40.13939 40.13939 40.13939 40.13939 71 81 91 101 111 121 40.13939 -127.10806 -127.10806 -294.35550 -294.35550 -294.35550 > predict(fit2, type='quantile')[xx] [1] 561.37687 561.37687 561.37687 226.88198 226.88198 226.88198 [7] 226.88198 226.88198 59.63453 59.63453 -107.61291 -107.61291 [13] -107.61291 > > rm(fitig, fit1, fit2, xx) > survival/tests/jasa.Rout.save0000644000175100001440000003142712536400755016051 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > expect <- survexp(futime ~ ratetable(age=(accept.dt - birth.dt), sex=1, + year=accept.dt, race='white'), jasa, cohort=F, + ratetable=survexp.usr) > > survdiff(Surv(jasa$futime, jasa$fustat) ~ offset(expect)) Call: survdiff(formula = Surv(jasa$futime, jasa$fustat) ~ offset(expect)) Observed Expected Z p 75.000 0.587 -97.119 0.000 > # Now fit the 6 models found in Kalbfleisch and Prentice, p139 > sfit.1 <- coxph(Surv(start, stop, event)~ (age + surgery)*transplant, + jasa1, method='breslow') > sfit.2 <- coxph(Surv(start, stop, event)~ year*transplant, + jasa1, method='breslow') > sfit.3 <- coxph(Surv(start, stop, event)~ (age + year)*transplant, + jasa1, method='breslow') > sfit.4 <- coxph(Surv(start, stop, event)~ (year +surgery) *transplant, + jasa1, method='breslow') > sfit.5 <- coxph(Surv(start, stop, event)~ (age + surgery)*transplant + year , + jasa1, method='breslow') > sfit.6 <- coxph(Surv(start, stop, event)~ age*transplant + surgery + year, + jasa1, method='breslow') > > summary(sfit.1) Call: coxph(formula = Surv(start, stop, event) ~ (age + surgery) * transplant, data = jasa1, method = "breslow") n= 170, number of events= 75 coef exp(coef) se(coef) z Pr(>|z|) age 0.01386 1.01395 0.01813 0.765 0.445 surgery -0.54652 0.57896 0.61091 -0.895 0.371 transplant 0.11572 1.12268 0.32729 0.354 0.724 age:transplant 0.03473 1.03534 0.02725 1.274 0.202 surgery:transplant -0.29037 0.74799 0.75819 -0.383 0.702 exp(coef) exp(-coef) lower .95 upper .95 age 1.014 0.9862 0.9786 1.051 surgery 0.579 1.7272 0.1748 1.917 transplant 1.123 0.8907 0.5911 2.132 age:transplant 1.035 0.9659 0.9815 1.092 surgery:transplant 0.748 1.3369 0.1692 3.306 Concordance= 0.595 (se = 0.037 ) Rsquare= 0.071 (max possible= 0.97 ) Likelihood ratio test= 12.45 on 5 df, p=0.02915 Wald test = 11.62 on 5 df, p=0.04031 Score (logrank) test = 12.02 on 5 df, p=0.03457 > sfit.2 Call: coxph(formula = Surv(start, stop, event) ~ year * transplant, data = jasa1, method = "breslow") coef exp(coef) se(coef) z p year -0.265 0.767 0.105 -2.52 0.012 transplant -0.287 0.750 0.514 -0.56 0.576 year:transplant 0.137 1.147 0.141 0.97 0.331 Likelihood ratio test=8.61 on 3 df, p=0.0349 n= 170, number of events= 75 > summary(sfit.3) Call: coxph(formula = Surv(start, stop, event) ~ (age + year) * transplant, data = jasa1, method = "breslow") n= 170, number of events= 75 coef exp(coef) se(coef) z Pr(>|z|) age 0.01558 1.01571 0.01734 0.899 0.36887 year -0.27413 0.76023 0.10588 -2.589 0.00962 ** transplant -0.59388 0.55218 0.54222 -1.095 0.27339 age:transplant 0.03380 1.03438 0.02795 1.209 0.22653 year:transplant 0.20228 1.22419 0.14247 1.420 0.15566 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 exp(coef) exp(-coef) lower .95 upper .95 age 1.0157 0.9845 0.9818 1.0508 year 0.7602 1.3154 0.6178 0.9356 transplant 0.5522 1.8110 0.1908 1.5981 age:transplant 1.0344 0.9668 0.9792 1.0926 year:transplant 1.2242 0.8169 0.9259 1.6185 Concordance= 0.63 (se = 0.037 ) Rsquare= 0.084 (max possible= 0.97 ) Likelihood ratio test= 14.85 on 5 df, p=0.01102 Wald test = 13.77 on 5 df, p=0.01716 Score (logrank) test = 14.06 on 5 df, p=0.01525 > sfit.4 Call: coxph(formula = Surv(start, stop, event) ~ (year + surgery) * transplant, data = jasa1, method = "breslow") coef exp(coef) se(coef) z p year -0.254 0.776 0.108 -2.36 0.018 surgery -0.237 0.789 0.628 -0.38 0.707 transplant -0.297 0.743 0.505 -0.59 0.557 year:transplant 0.165 1.180 0.142 1.17 0.243 surgery:transplant -0.550 0.577 0.776 -0.71 0.479 Likelihood ratio test=12.4 on 5 df, p=0.0302 n= 170, number of events= 75 > sfit.5 Call: coxph(formula = Surv(start, stop, event) ~ (age + surgery) * transplant + year, data = jasa1, method = "breslow") coef exp(coef) se(coef) z p age 0.0150 1.0152 0.0176 0.85 0.393 surgery -0.4202 0.6569 0.6156 -0.68 0.495 transplant 0.0741 1.0769 0.3311 0.22 0.823 year -0.1363 0.8726 0.0710 -1.92 0.055 age:transplant 0.0269 1.0273 0.0271 0.99 0.321 surgery:transplant -0.2966 0.7434 0.7580 -0.39 0.696 Likelihood ratio test=16.2 on 6 df, p=0.0127 n= 170, number of events= 75 > sfit.6 Call: coxph(formula = Surv(start, stop, event) ~ age * transplant + surgery + year, data = jasa1, method = "breslow") coef exp(coef) se(coef) z p age 0.0153 1.0154 0.0175 0.87 0.383 transplant 0.0446 1.0456 0.3217 0.14 0.890 surgery -0.6211 0.5373 0.3679 -1.69 0.091 year -0.1361 0.8728 0.0709 -1.92 0.055 age:transplant 0.0270 1.0274 0.0271 1.00 0.319 Likelihood ratio test=16.1 on 5 df, p=0.00669 n= 170, number of events= 75 > > # Survival curve for an "average" subject, > # done once as overall, once via individual method > surv1 <- survfit(sfit.1, newdata=list(age=-2, surgery=0, transplant=0)) > newdata <- data.frame(start=c(0,50,100), stop=c(50,100, max(jasa1$stop)), + event=c(1,1,1), age=rep(-2,3), surgery=rep(0,3), + transplant=rep(0,3)) > surv2 <- survfit(sfit.1, newdata, individual=T) > # Have to use unclass to avoid [.survfit trying to pick curves, > # remove the final element "call" because it won't match > all.equal(unclass(surv1)[-length(surv1)], + unclass(surv2)[-length(surv2)]) [1] TRUE > > > # Survival curve for a subject of age 50, with prior surgery, tx at 6 months > # Remember that 'age' in jasa 1 was centered at 48 > data <- data.frame(start=c(0,183), stop=c(183,3*365), event=c(1,1), + age=c(2,2), surgery=c(1,1), transplant=c(0,1)) > summary(survfit(sfit.1, data, individual=T)) Call: survfit(formula = sfit.1, newdata = data, individual = T) time n.risk n.event survival std.err lower 95% CI upper 95% CI 0.5 103 1 0.994 0.00722 0.980 1.000 1.0 102 3 0.975 0.01860 0.939 1.000 2.0 99 3 0.956 0.02914 0.900 1.000 4.0 96 2 0.943 0.03605 0.875 1.000 5.0 94 2 0.930 0.04286 0.849 1.000 7.0 92 1 0.923 0.04623 0.837 1.000 8.0 91 1 0.917 0.04959 0.824 1.000 11.0 89 1 0.910 0.05294 0.812 1.000 15.0 88 3 0.890 0.06278 0.775 1.000 16.0 85 1 0.883 0.06608 0.763 1.000 17.0 84 1 0.877 0.06928 0.751 1.000 20.0 83 2 0.864 0.07538 0.728 1.000 27.0 81 1 0.857 0.07849 0.716 1.000 29.0 80 1 0.850 0.08160 0.705 1.000 31.0 78 1 0.844 0.08473 0.693 1.000 34.0 77 1 0.837 0.08786 0.681 1.000 35.0 76 1 0.830 0.09098 0.669 1.000 36.0 75 1 0.823 0.09412 0.658 1.000 38.0 74 1 0.816 0.09727 0.646 1.000 39.0 72 2 0.802 0.10349 0.623 1.000 42.0 70 1 0.795 0.10664 0.611 1.000 44.0 69 1 0.788 0.10982 0.600 1.000 49.0 68 1 0.781 0.11300 0.588 1.000 50.0 67 1 0.774 0.11614 0.577 1.000 52.0 66 1 0.767 0.11925 0.565 1.000 57.0 65 1 0.760 0.12238 0.554 1.000 60.0 64 1 0.752 0.12552 0.542 1.000 65.0 63 1 0.745 0.12866 0.531 1.000 67.0 62 2 0.730 0.13494 0.508 1.000 68.0 60 1 0.722 0.13809 0.497 1.000 71.0 59 2 0.707 0.14420 0.474 1.000 76.0 57 1 0.699 0.14729 0.463 1.000 77.0 56 1 0.691 0.15043 0.451 1.000 79.0 55 1 0.683 0.15362 0.439 1.000 80.0 54 1 0.674 0.15680 0.428 1.000 84.0 53 1 0.666 0.16005 0.416 1.000 89.0 52 1 0.657 0.16326 0.404 1.000 95.0 51 1 0.648 0.16648 0.392 1.000 99.0 50 1 0.639 0.16972 0.380 1.000 101.0 49 1 0.630 0.17293 0.368 1.000 109.0 47 1 0.621 0.17611 0.356 1.000 148.0 45 1 0.611 0.17927 0.344 1.000 152.0 44 1 0.601 0.18236 0.332 1.000 164.0 43 1 0.592 0.18551 0.320 1.000 185.0 41 1 0.583 0.12737 0.380 0.894 187.0 40 1 0.574 0.12889 0.370 0.891 206.0 39 1 0.565 0.13036 0.359 0.888 218.0 38 1 0.556 0.13180 0.349 0.885 262.0 37 1 0.546 0.13320 0.339 0.881 284.0 35 2 0.527 0.13585 0.318 0.874 307.0 33 1 0.517 0.13707 0.308 0.869 333.0 32 1 0.507 0.13823 0.297 0.865 339.0 31 1 0.497 0.13930 0.287 0.861 342.0 29 1 0.486 0.14029 0.276 0.856 583.0 21 1 0.471 0.14187 0.261 0.850 674.0 17 1 0.452 0.14361 0.243 0.843 732.0 16 1 0.433 0.14506 0.225 0.835 851.0 14 1 0.410 0.14622 0.204 0.825 979.0 11 1 0.383 0.14698 0.180 0.813 995.0 10 1 0.356 0.14735 0.158 0.801 1031.0 9 1 0.330 0.14743 0.137 0.792 > > # These should all give the same answer > # When there are offsets, the default curve is always for someone with > # the mean offset. > j.age <- jasa$age -48 > fit1 <- coxph(Surv(futime, fustat) ~ j.age, data=jasa) > fit2 <- coxph(Surv(futime, fustat) ~ j.age, jasa, init=fit1$coef, iter=0) > fit3 <- coxph(Surv(start, stop, event) ~ age, jasa1) > fit4 <- coxph(Surv(start, stop, event) ~ offset(age*fit1$coef), jasa1) > > s1 <- survfit(fit1, list(j.age=fit3$means), censor=FALSE) > s2 <- survfit(fit2, list(j.age=fit3$means), censor=FALSE) > s3 <- survfit(fit3, censor=FALSE) > s4 <- survfit(fit4, censor=FALSE) > > all.equal(s1$surv, s2$surv) [1] TRUE > all.equal(s1$surv, s3$surv) [1] TRUE > all.equal(s1$surv, s4$surv) [1] TRUE > > # Still the same answer, fit multiple strata at once > # Strata 1 has independent coefs of strata 2, so putting in > # the other data should not affect it > ll <- nrow(jasa1) > ss <- rep(0:1, c(ll,ll)) > tdata <- with(jasa1, data.frame(start=rep(start,2), stop=rep(stop,2), + event=rep(event,2), ss=ss, age=rep(age,2), + age2 = (rep(age,2))^2 * ss)) > fit <- coxph(Surv(start, stop, event) ~ age*strata(ss) + age2, tdata) > # Above replaced these 2 lines, which kill Splus5 as of 8/98 > # Something with data frames, I expect. > #fit <- coxph(Surv(rep(start,2), rep(stop,2), rep(event,2)) ~ > # rep(age,2)*strata(ss) + I(rep(age,2)^2*ss) ) > all.equal(fit$coef[1], fit3$coef) [1] TRUE > s5 <- survfit(fit, data.frame(age=fit3$means, age2=0, ss=0), censor=FALSE) > all.equal(s5$surv[1:(s5$strata[1])], s3$surv) [1] TRUE > > > > proc.time() user system elapsed 0.384 0.020 0.398 survival/tests/book6.R0000644000175100001440000000741111732700061014450 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # Tests of the weighted Cox model # This is section 1.3 of my appendix -- no yet found in any of the # printings though, it awaits the next edition # # Efron approximation # aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), status= c(1,0,1,1,1,0,0,1,0), x= c(2,0,1,1,0,1,0,1,0), wt = c(1,2,3,4,3,2,1,2,1)) xx <- testw1$wt # Efron estimate byhand <- function(beta, newx=0) { r <- exp(beta) a <- 7*r +3; b<- 4*r+2 loglik <- 11*beta - (log(r^2 + 11*r +7) + 10*log(11*r +5)/3 + 10*log(a*2/3 +b)/3 + 10*log(a/3 +b)/3 +2*log(2*r+1)) hazard <- c(1/(r^2 + 11*r +7), 10/(3*c(11*r +5, a*2/3 +b, a/3+b)), 2/(2*r+1)) temp <- c(hazard[1], hazard[1]+hazard[2] + hazard[3]*2/3 + hazard[4]/3, cumsum(hazard)[4:5]) risk <- c(r^2, 1,r,r,1,r,1,r,1) expected <- risk* temp[c(1,1,2,2,2,3,3,4,4)] # The matrix of weights, one row per obs, one col per death # deaths at 1,2,2,2, and 4 riskmat <- matrix(c(1,1,1,1,1,1,1,1,1, 0,0,1,1,1,1,1,1,1, 0,0,2/3,2/3,2/3,1,1,1,1, 0,0,1/3,1/3,1/3,1,1,1,1, 0,0,0,0,0,0,0,1,1), ncol=5) wtmat <- diag(c(r^2, 2, 3*r, 4*r, 3, 2*r, 1, 2*r, 1)) %*% riskmat x <- c(2,0,1,1,0,1,0,1,0) xbar <- colSums(x*wtmat)/ colSums(wtmat) imat <- (4*r^2 + 11*r)*hazard[1] - xbar[1]^2 + 10* mean(xbar[2:4] - xbar[2:4]^2) + 2*(xbar[5] - xbar[5]^2) status <- c(1,0,1,1,1,0,0,1,0) wt <- c(1,2,3,4,3,2,1,2,1) # Table of sums for score resids hazmat <- riskmat %*% diag(c(1,10/3,10/3, 10/3,2)/colSums(wtmat)) dM <- -risk*hazmat #Expected part dM[1,1] <- dM[1,1] +1 # deaths at time 1 for (i in 2:4) dM[3:5, i] <- dM[3:5,i] + 1/3 dM[8,5] <- dM[8,5] +1 mart <- rowSums(dM) resid <-dM * outer(x, xbar ,'-') # Increments to the variance of the hazard var.g <- cumsum(hazard^2* c(1,3/10, 3/10, 3/10, 1/2)) var.d <- cumsum((xbar-newx)*hazard) sxbar <- c(xbar[1], mean(xbar[2:4]), xbar[5]) #xbar for Schoen list(loglik=loglik, imat=imat, hazard=hazard, xbar=xbar, mart=status-expected, expected=expected, score=rowSums(resid), schoen=c(2,1,1,0,1) - sxbar[c(1,2,2,2,3)], varhaz=((var.g + var.d^2/imat)* exp(2*beta*newx))[c(1,4,5)]) } # Verify temp <- byhand(0,0) aeq(temp$xbar, c(13/19, 11/16, 26/38, 19/28, 2/3)) aeq(temp$hazard, c(1/19, 5/24, 5/19, 5/14, 2/3)) fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, iter=0) fit <- coxph(Surv(time, status) ~x, testw1, weights=wt) truth0 <- byhand(0,pi) aeq(fit0$loglik[1], truth0$loglik) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')) sfit <- survfit(fit0, list(x=pi), censor=FALSE) aeq(sfit$std.err^2, truth0$var) aeq(-log(sfit$surv), cumsum(truth0$hazard)[c(1,4,5)]) truth <- byhand(fit$coef, .3) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')) sfit <- survfit(fit, list(x=.3), censor=FALSE) aeq(sfit$std.err^2, truth$var) aeq(-log(sfit$surv), (cumsum(truth$hazard)* exp(fit$coef*.3))[c(1,4,5)]) fit0 summary(fit) resid(fit0, type='score') resid(fit0, type='scho') resid(fit, type='score') resid(fit, type='scho') rr1 <- resid(fit, type='mart') rr2 <- resid(fit, type='mart', weighted=T) aeq(rr2/rr1, testw1$wt) rr1 <- resid(fit, type='score') rr2 <- resid(fit, type='score', weighted=T) aeq(rr2/rr1, testw1$wt) survival/tests/data.motor0000644000175100001440000000067011732700061015300 0ustar hornikusers150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 150 8064 0 170 1764 1 170 2772 1 170 3444 1 170 3542 1 170 3780 1 170 4860 1 170 5196 1 170 5448 0 170 5448 0 170 5448 0 190 408 1 190 408 1 190 1344 1 190 1344 1 190 1440 1 190 1680 0 190 1680 0 190 1680 0 190 1680 0 190 1680 0 220 408 1 220 408 1 220 504 1 220 504 1 220 504 1 220 528 0 220 528 0 220 528 0 220 528 0 220 528 0 survival/tests/fr_resid.R0000644000175100001440000000603612466153737015250 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # The residual methods treat a sparse frailty as a fixed offset with # no variance # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) kfit1 <- coxph(Surv(time, status) ~ age + sex + frailty(id, dist='gauss'), kidney) tempf <- predict(kfit1, type='terms')[,3] temp <- kfit1$frail[match(kidney$id, sort(unique(kidney$id)))] #all.equal(unclass(tempf), unclass(temp)) all.equal(as.vector(tempf), as.vector(temp)) # Now fit a model with explicit offset kfitx <- coxph(Surv(time, status) ~ age + sex + offset(tempf),kidney, eps=1e-7) # These are not always precisely the same, due to different iteration paths aeq(kfitx$coef, kfit1$coef) # This will make them identical kfitx <- coxph(Surv(time, status) ~ age + sex + offset(temp),kidney, iter=0, init=kfit1$coef) aeq(resid(kfit1), resid(kfitx)) aeq(resid(kfit1, type='score'), resid(kfitx, type='score')) aeq(resid(kfit1, type='schoe'), resid(kfitx, type='schoe')) # These are not the same, due to a different variance matrix # The frailty model's variance is about 2x the naive "assume an offset" var # Expect a value of about 0.5 aeq(resid(kfit1, type='dfbeta'), resid(kfitx, type='dfbeta')) # Force equality zed <- kfitx zed$var <- kfit1$var aeq(resid(kfit1, type='dfbeta'), resid(zed, type='dfbeta')) # The score residuals are equal, however. temp1 <- resid(kfit1, type='score') temp2 <- resid(kfitx, type='score') aeq(temp1, temp2) # # Now for some tests of predicted values # aeq(predict(kfit1, type='expected'), predict(kfitx, type='expected')) aeq(predict(kfit1, type='lp'), predict(kfitx, type='lp')) temp1 <- predict(kfit1, type='terms', se.fit=T) temp2 <- predict(kfitx, type='terms', se.fit=T) aeq(temp1$fit[,1:2], temp2$fit) # the next is not equal, all.equal returns a character string in that case is.character(aeq(temp1$se.fit[,1:2], temp2$se.fit)) mean(temp1$se.fit[,1:2]/ temp2$se.fit) aeq(as.vector(temp1$se.fit[,3])^2, as.vector(kfit1$fvar[match(kidney$id, sort(unique(kidney$id)))])) print(temp1) kfit1 kfitx rm(temp1, temp2, kfitx, zed, tempf) # # The special case of a single sparse frailty # kfit1 <- coxph(Surv(time, status) ~ frailty(id, dist='gauss'), kidney) tempf <- predict(kfit1, type='terms') temp <- kfit1$frail[match(kidney$id, sort(unique(kidney$id)))] all.equal(as.vector(tempf), as.vector(temp)) # Now fit a model with explicit offset kfitx <- coxph(Surv(time, status) ~ offset(tempf),kidney, eps=1e-7) aeq(resid(kfit1), resid(kfitx)) aeq(resid(kfit1, type='deviance'), resid(kfitx, type='deviance')) # # Some tests of predicted values # aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) aeq(predict(kfit1, type='expected'), predict(kfitx, type='expected')) aeq(predict(kfit1, type='lp'), predict(kfitx, type='lp')) temp1 <- predict(kfit1, type='terms', se.fit=T) aeq(temp1$fit, kfitx$linear) aeq(temp1$se.fit^2, kfit1$fvar[match(kidney$id, sort(unique(kidney$id)))]) temp1 kfit1 survival/tests/testci.R0000644000175100001440000001006012466142446014731 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) aeq <- function(x,y,...) all.equal(as.vector(x), as.vector(y),...) # # Test out the survfit.ci function, which does competing risk # estimates # # First trivial test tdata <- data.frame(time=c(1,2,2,3,3,3,5,6), status = c(0,1,0,1,0,1,0,1), event = c(1,1,2,2,1,2,3,2), grp = c(1,2,1,2,1,2,1,2)) fit <- survfit(Surv(time, status*event, type='mstate') ~1, tdata) byhand <- function() { #everyone starts in state 0 p1 <- c(1,0,0) p2 <- c(6/7, 1/7, 0) # 0-1 transition at time 2 u2 <- matrix(rep(c(1/49, -1/49, 0), each=8), ncol=3) #leverage matrix at time 2 u2[1,] <- 0 #subject 1 is not present u2[2,1:2] <- u2[2, 1:2] + c(-1/7, 1/7) p3 <- c((6/7)*(3/5), 1/7, 12/35) # 0-2 transition at time 3, 5 at risk h3 <- matrix(c(3/5, 0, 2/5, 0,1,0, 0,1,0), byrow=T, ncol=3) #hazard mat u3 <- u2 %*% h3 u3[4:8,1] <- u3[4:8,1] + p2[1]*2/25 u3[4:8,3] <- u3[4:8,3] -p2[1]*2/25 u3[4,] <- u3[4,] + c(-p2[1]/5, 0, p2[1]/5) u3[6,] <- u3[4,] p6 <- c(0, 1/7, 6/7) # 0-2 at time 6, 1 at risk h6 <- matrix(c(-1,0,1,0,1,0,0,1,0), byrow=T, ncol=3) u6 <- cbind(0, u3[,2], -u3[,2]) V <- rbind(0, colSums(u2^2), colSums(u3^2), colSums(u3^2), colSums(u6^2)) list(P=rbind(p1, p2, p3, p3, p6), u2=u2, u3=u3, u6=u6, V=V) } bfit <- byhand() aeq(fit$prev, bfit$P[,-1]) aeq(fit$n.risk, c(8,7,5,2,1)) aeq(fit$n.event, c(0,1,2,0,1)) aeq(fit$std^2, bfit$V[,-1]) # # For this we need the competing risks MGUS data set, first # event # tdata <- mgus1[mgus1$enum==1,] # Ensure the old-style call using "etype" works (backwards compatability) fit1 <- survfit(Surv(stop, status) ~ 1, etype=event, tdata) fit1b <-survfit(Surv(stop, event) ~1, tdata) indx <- match("call", names(fit1)) all.equal(unclass(fit1)[-indx], unclass(fit1b)[-indx]) # Now get the overall survival, and the hazard for progression fit2 <- survfit(Surv(stop, status) ~1, tdata) #overall to "first bad thing" fit3 <- survfit(Surv(stop, status*(event=='pcm')) ~1, tdata, type='fleming') fit4 <- survfit(Surv(stop, status*(event=='death')) ~1, tdata, type='fleming') aeq(fit1$n.risk, fit2$n.risk) aeq(fit1$n.event, fit2$n.event) # Classic CI formula # integral [hazard(t) S(t-0) dt], where S= "survival to first event" haz1 <- diff(c(0, -log(fit3$surv))) #Aalen hazard estimate for progression haz2 <- diff(c(0, -log(fit4$surv))) #Aalen estimate for death tsurv <- c(1, fit2$surv[-length(fit2$surv)]) #lagged survival ci1 <- cumsum(haz1 *tsurv) ci2 <- cumsum(haz2 *tsurv) aeq(cbind(ci1, ci2), fit1$prev) # # Now, make sure that it works for subgroups # fit1 <- survfit(Surv(stop, event) ~ sex, tdata) fit2 <- survfit(Surv(stop, event) ~ 1, tdata, subset=(sex=='female')) fit3 <- survfit(Surv(stop, event) ~ 1, tdata, subset=(sex=='male')) aeq(fit2$prev, fit1$prev[1:fit1$strata[1],]) aeq(fit2$std, fit1$std[1:fit1$strata[1],]) aeq(fit3$prev, fit1$prev[-(1:fit1$strata[1]),]) # A second test of cumulative incidence # compare results to Bob Gray's functions # The file gray1 is the result of # library(cmprsk) # tstat <- ifelse(tdata$status==0, 0, 1+ (tdata$event=='death')) # gray1 <- cuminc(tdata$stop, tstat) load("gray1.rda") fit2 <- survfit(Surv(stop, event) ~ 1, tdata) if (FALSE) { # lines of the two graphs should overlay plot(gray1[[1]]$time, gray1[[1]]$est, type='l', ylim=range(c(gray1[[1]]$est, gray1[[2]]$est)), xlab="Time") lines(gray1[[2]]$time, gray1[[2]]$est, lty=2) matlines(fit2$time, fit2$prev, col=2, lty=1:2, type='s') } # To formally match these is a bit of a nuisance. # The cuminc function returns a full step function, and survfit only # the bottoms of the steps. temp1 <- tapply(gray1[[1]]$est, gray1[[1]]$time, max)[-1] #toss time 0 indx1 <- match(names(temp1), fit2$time) aeq(temp1, fit2$prev[indx1,1]) survival/tests/coxsurv.Rout.save0000644000175100001440000000757612044264177016654 0ustar hornikusers R version 2.15.1 (2012-06-22) -- "Roasted Marshmallows" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test out subscripting in the case of a coxph survival curve > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), + data=cancer) > > surv1 <- survfit(fit) > temp <- surv1[2:3] > > which <- cumsum(surv1$strata) > zed <- (which[1]+1):(which[3]) > aeq(surv1$surv[zed], temp$surv) [1] TRUE > aeq(surv1$time[zed], temp$time) [1] TRUE > > # This call should not create a model frame in the code -- so same > # answer but a different path through the underlying code > fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), + x=T, data=cancer) > surv2 <- survfit(fit) > all.equal(surv1, surv2) [1] TRUE > > # > # Now a result with a matrix of survival curves > # > dummy <- data.frame(age=c(30,40,60), sex=c(1,2,2), meal.cal=c(500, 1000, 1500)) > surv2 <- survfit(fit, newdata=dummy) > > zed <- 1:which[1] > aeq(surv2$surv[zed,1], surv2[1,1]$surv) [1] TRUE > aeq(surv2$surv[zed,2], surv2[1,2]$surv) [1] TRUE > aeq(surv2$surv[zed,3], surv2[1,3]$surv) [1] TRUE > aeq(surv2$surv[zed, ], surv2[1,1:3]$surv) [1] TRUE > aeq(surv2$surv[zed], (surv2[1]$surv)[,1]) [1] TRUE > aeq(surv2$surv[zed, ], surv2[1, ]$surv) [1] TRUE > > fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), + data=cancer) > > surv1 <- survfit(fit) > temp <- surv1[2:3] > > which <- cumsum(surv1$strata) > zed <- (which[1]+1):(which[3]) > aeq(surv1$surv[zed], temp$surv) [1] TRUE > aeq(surv1$time[zed], temp$time) [1] TRUE > > # This call should not create a model frame in the code -- so same > # answer but a different path through the underlying code > fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), + x=T, data=cancer) > surv2 <- survfit(fit) > all.equal(surv1, surv2) [1] TRUE > > # > # Now a result with a matrix of survival curves > # > dummy <- data.frame(age=c(30,40,60), sex=c(1,2,2), meal.cal=c(500, 1000, 1500)) > surv2 <- survfit(fit, newdata=dummy) > > zed <- 1:which[1] > aeq(surv2$surv[zed,1], surv2[1,1]$surv) [1] TRUE > aeq(surv2$surv[zed,2], surv2[1,2]$surv) [1] TRUE > aeq(surv2$surv[zed,3], surv2[1,3]$surv) [1] TRUE > aeq(surv2$surv[zed, ], surv2[1,1:3]$surv) [1] TRUE > aeq(surv2$surv[zed], (surv2[1]$surv)[,1]) [1] TRUE > aeq(surv2$surv[zed, ], surv2[1, ]$surv) [1] TRUE > > # And the depreciated form - call with a named vector as 'newdata' > # the resulting $call component won't match so delete it before comparing > surv3 <- survfit(fit, c(age=40, sex=2, meal.cal=1000)) > all.equal(unclass(surv2[,2])[-length(surv3)], unclass(surv3)[-length(surv3)]) [1] TRUE > > > > # Test out offsets, which have recently become popular due to a Langholz paper > fit1 <- coxph(Surv(time, status) ~ age + ph.ecog, lung) > fit2 <- coxph(Surv(time, status) ~ age + offset(ph.ecog * fit1$coef[2]), lung) > > surv1 <- survfit(fit1, newdata=data.frame(age=50, ph.ecog=1)) > surv2 <- survfit(fit2, newdata=data.frame(age=50, ph.ecog=1)) > all.equal(surv1$surv, surv2$surv) [1] TRUE > > proc.time() user system elapsed 0.316 0.032 0.342 survival/tests/r_user.Rout.save0000644000175100001440000000305611732700061016415 0ustar hornikusers R version 2.9.0 (2009-04-17) Copyright (C) 2009 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) #preserve length of missings > library(survival) Loading required package: splines > > # > # Check out using a "user specified" distribution > # > mydist <- c(survreg.distributions$extreme, survreg.distributions$weibull[-1]) > mydist$name <- "Weibull2" > mydist$dist <- NULL > > fit1 <- survreg(Surv(time, status) ~ age + ph.ecog, lung) > fit2 <- survreg(Surv(time, status) ~ age + ph.ecog, lung, dist=mydist) > > all.equal(fit1$coef, fit2$coef) [1] TRUE > all.equal(fit1$var, fit2$var) [1] TRUE > > # > # And with an data set containing interval censoring > # > idat <- read.table('data.interval', skip=3, header=T, sep=',') > > fit1 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, idat) > fit2 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, + data=idat, dist=mydist) > > all.equal(fit1$coef, fit2$coef) [1] TRUE > all.equal(fit1$var, fit2$var) [1] TRUE > all.equal(fit1$log, fit2$log) [1] TRUE > > survival/tests/tiedtime.R0000644000175100001440000000135511732700061015235 0ustar hornikuserslibrary(survival) # # The survival code was failing for certain data sets when called as # survfit(Surv(time2-time1, status) ~ ...... # The issue was how tied floating point numbers are handled, and the # fact that unique(x), factor(x) and tapply(x) are not guarranteed to # all be the same. # This test fails in survival 2.36-5, fixed in 2.36-6. Data sets that # can cause it are few and far between. # load('ties.rda') x <- time2 -time1 # Here is the heart of the old problem # length(unique(x))== length(table(x)) # And the prior fix which worked ALMOST always # x <- round(x, 15) # length(unique(round(x,15)))== length(table(round(x,15))) fit1 <- survfit(Surv(x) ~1) length(fit1$time) == length(fit1$surv) survival/tests/r_sas.Rout.save0000644000175100001440000004216112113162163016224 0ustar hornikusers R Under development (unstable) (2013-02-24 r62054) -- "Unsuffered Consequences" Copyright (C) 2013 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Reproduce example 1 in the SAS lifereg documentation > # > > # this fit doesn't give the same log-lik that they claim > motor <- read.table('data.motor', col.names=c('temp', 'time', 'status')) > fit1 <- survreg(Surv(time, status) ~ I(1000/(273.2+temp)), motor, + subset=(temp>150), dist='lognormal') > summary(fit1) Call: survreg(formula = Surv(time, status) ~ I(1000/(273.2 + temp)), data = motor, subset = (temp > 150), dist = "lognormal") Value Std. Error z p (Intercept) -10.471 2.772 -3.78 1.58e-04 I(1000/(273.2 + temp)) 8.322 1.284 6.48 9.13e-11 Log(scale) -0.504 0.183 -2.75 5.96e-03 Scale= 0.604 Log Normal distribution Loglik(model)= -145.9 Loglik(intercept only)= -155 Chisq= 18.3 on 1 degrees of freedom, p= 1.9e-05 Number of Newton-Raphson Iterations: 6 n= 30 > > # This one, with the loglik on the transformed scale (the inappropriate > # scale, Ripley & Venables would argue) does agree. > # All coefs are of course identical. > fit2 <- survreg(Surv(log(time), status) ~ I(1000/(273.2+temp)), motor, + subset=(temp>150), dist='gaussian') > > > # Give the quantile estimates, which is the lower half of "output 48.1.5" > # in the SAS 9.2 manual > > pp1 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), + type='quantile', se=T) > pp2 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), + type='uquantile', se=T) > pp1 $fit [,1] [,2] [,3] [1,] 12033.185 26095.677 56592.20 [2,] 4536.877 9838.864 21336.98 $se.fit [,1] [,2] [,3] [1,] 5482.338 11359.450 26036.917 [2,] 1443.072 2901.155 7172.343 > > temp130 <- matrix(0, nrow=3, ncol=6) > temp130[,1] <- pp1$fit[1,] > temp130[,2] <- pp1$se.fit[1,] > temp130[,3] <- pp2$fit[1,] > temp130[,4] <- pp2$se.fit[1,] > temp130[,5] <- exp(pp2$fit[1,] - 1.64*pp2$se.fit[1,]) > temp130[,6] <- exp(pp2$fit[1,] + 1.64*pp2$se.fit[1,]) > dimnames(temp130) <- list(c("p=.1", "p=.2", "p=.3"), + c("Time", "se(time)", "log(time)", "se[log(time)]", + "lower 90", "upper 90")) > print(temp130) Time se(time) log(time) se[log(time)] lower 90 upper 90 p=.1 12033.18 5482.338 9.395424 0.4556015 5700.089 25402.68 p=.2 26095.68 11359.450 10.169525 0.4353001 12779.950 53285.37 p=.3 56592.20 26036.917 10.943626 0.4600796 26611.422 120349.71 > > # A set of examples, copied from the manual pages of SAS procedure > # "reliability", which is part of their QC product. > # > > color <- c("black", "red", "green", "blue", "magenta", "red4", + "orange", "DarkGreen", "cyan2", "DarkViolet") > palette(color) > pdf(file='reliability.pdf') > > # > # Insulating fluids example > # > fluid <- read.table('data.fluid', col.names=c('time', 'voltage')) > > # Adding a -1 to the fit just causes the each group to have it's own > # intercept, rather than a global intercept + constrasts. The strata > # statement allows each to have a separate scale > ffit <- survreg(Surv(time) ~ voltage + strata(voltage) -1, fluid) > > # Get predicted quantiles at each of the voltages > # By default predict() would give a line of results for each observation, > # I only want the unique set of x's, i.e., only 4 cases > uvolt <- sort(unique(fluid$voltage)) #the unique levels > plist <- c(1, 2, 5, 1:9 *10, 95, 99)/100 > pred <- predict(ffit, type='quantile', p=plist, + newdata=data.frame(voltage=factor(uvolt))) > tfun <- function(x) log(-log(1-x)) > > matplot(t(pred), tfun(plist), type='l', log='x', lty=1, + col=1:4, yaxt='n') > axis(2, tfun(plist), format(100*plist), adj=1) > > kfit <- survfit(Surv(time) ~ voltage, fluid, type='fleming') #KM fit > for (i in 1:4) { + temp <- kfit[i] + points(temp$time, tfun(1-temp$surv), col=i, pch=i) + } > > # Now a table > temp <- array(0, dim=c(4,4,4)) #4 groups by 4 parameters by 4 stats > temp[,1,1] <- ffit$coef # "EV Location" in SAS manual > temp[,2,1] <- ffit$scale # "EV scale" > temp[,3,1] <- exp(ffit$coef) # "Weibull Scale" > temp[,4,1] <- 1/ffit$scale # "Weibull Shape" > > temp[,1,2] <- sqrt(diag(ffit$var))[1:4] #standard error > temp[,2,2] <- sqrt(diag(ffit$var))[5:8] * ffit$scale > temp[,3,2] <- temp[,1,2] * temp[,3,1] > temp[,4,2] <- temp[,2,2] / (temp[,2,1])^2 > > temp[,1,3] <- temp[,1,1] - 1.96*temp[,1,2] #lower conf limits > temp[,1,4] <- temp[,1,1] + 1.96*temp[,1,2] # upper > # log(scale) is the natural parameter, in which the routine did its fitting > # and on which the std errors were computed > temp[,2, 3] <- exp(log(ffit$scale) - 1.96*sqrt(diag(ffit$var))[5:8]) > temp[,2, 4] <- exp(log(ffit$scale) + 1.96*sqrt(diag(ffit$var))[5:8]) > > temp[,3, 3:4] <- exp(temp[,1,3:4]) > temp[,4, 3:4] <- 1/temp[,2,4:3] > > dimnames(temp) <- list(uvolt, c("EV Location", "EV Scale", "Weibull scale", + "Weibull shape"), + c("Estimate", "SE", "lower 95% CI", "uppper 95% CI")) > print(aperm(temp, c(2,3,1)), digits=5) , , 26kV Estimate SE lower 95% CI uppper 95% CI EV Location 6.86249 1.10404 4.69857 9.0264 EV Scale 1.83423 0.96114 0.65677 5.1227 Weibull scale 955.74665 1055.18620 109.78973 8320.0103 Weibull shape 0.54519 0.28568 0.19521 1.5226 , , 30kV Estimate SE lower 95% CI uppper 95% CI EV Location 4.35133 0.30151 3.76037 4.9423 EV Scale 0.94446 0.22544 0.59156 1.5079 Weibull scale 77.58159 23.39176 42.96420 140.0911 Weibull shape 1.05881 0.25274 0.66318 1.6904 , , 34kV Estimate SE lower 95% CI uppper 95% CI EV Location 2.50326 0.31476 1.88632 3.1202 EV Scale 1.29732 0.22895 0.91796 1.8334 Weibull scale 12.22222 3.84707 6.59509 22.6506 Weibull shape 0.77082 0.13603 0.54542 1.0894 , , 38kV Estimate SE lower 95% CI uppper 95% CI EV Location 0.00092629 0.27318 -0.53450 0.53635 EV Scale 0.73367610 0.20380 0.42565 1.26460 Weibull scale 1.00092672 0.27343 0.58596 1.70976 Weibull shape 1.36299929 0.37861 0.79077 2.34933 > > rm(temp, uvolt, plist, pred, ffit, kfit) > > ##################################################################### > # Turbine cracks data > cracks <- read.table('data.cracks', col.names=c('time1', 'time2', 'n')) > cfit <- survreg(Surv(time1, time2, type='interval2') ~1, + dist='weibull', data=cracks, weight=n) > > summary(cfit) Call: survreg(formula = Surv(time1, time2, type = "interval2") ~ 1, data = cracks, weights = n, dist = "weibull") Value Std. Error z p (Intercept) 4.272 0.0744 57.43 0.00e+00 Log(scale) -0.396 0.0987 -4.01 6.06e-05 Scale= 0.673 Weibull distribution Loglik(model)= -309.7 Loglik(intercept only)= -309.7 Number of Newton-Raphson Iterations: 5 n= 9 > #Their output also has Wiebull scale = exp(cfit$coef), shape = 1/(cfit$scale) > > # Draw the SAS plot > # The "type=fleming" argument reflects that they estimate hazards rather than > # survival, and forces a Nelson-Aalen hazard estimate > # > plist <- c(1, 2, 5, 1:8 *10)/100 > plot(qsurvreg(plist, cfit$coef, cfit$scale), tfun(plist), log='x', + yaxt='n', type='l', + xlab="Weibull Plot for Time", ylab="Percent") > axis(2, tfun(plist), format(100*plist), adj=1) > > kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=cracks, + weight=n, type='fleming') > # Only plot point where n.event > 0 > # Why? I'm trying to match them. Personally, all should be plotted. > who <- (kfit$n.event > 0) > points(kfit$time[who], tfun(1-kfit$surv[who]), pch='+') > points(kfit$time[who], tfun(1-kfit$upper[who]), pch='-') > points(kfit$time[who], tfun(1-kfit$lower[who]), pch='-') > > text(rep(3,6), seq(.5, -1.0, length=6), + c("Scale", "Shape", "Right Censored", "Left Censored", + "Interval Censored", "Fit"), adj=0) > text(rep(9,6), seq(.5, -1.0, length=6), + c(format(round(exp(cfit$coef), 2)), + format(round(1/cfit$scale, 2)), + format(tapply(cracks$n, cfit$y[,3], sum)), "ML"), adj=1) > > # Now a portion of his percentiles table > # I don't get the same SE as SAS, I haven't checked out why. The > # estimates and se for the underlying Weibull model are the same. > temp <- predict(cfit, type='quantile', p=plist, se=T) > tempse <- sqrt(temp$se[1,]) > mat <- cbind(temp$fit[1,], tempse, + temp$fit[1,] -1.96*tempse, temp$fit[1,] + 1.96*tempse) > dimnames(mat) <- list(plist*100, c("Estimate", "SE", "Lower .95", "Upper .95")) > print(mat) Estimate SE Lower .95 Upper .95 1 3.239372 0.965006 1.347960 5.130784 2 5.183283 1.121677 2.984796 7.381770 5 9.705766 1.337420 7.084422 12.327109 10 15.757758 1.491460 12.834497 18.681020 20 26.115947 1.622573 22.935705 29.296190 30 35.812585 1.704575 32.471618 39.153553 40 45.610018 1.809448 42.063500 49.156536 50 56.014351 1.973350 52.146585 59.882116 60 67.592818 2.214072 63.253237 71.932400 70 81.233457 2.543490 76.248217 86.218697 80 98.764571 2.991889 92.900469 104.628673 > > # > # The cracks data has a particularly easy estimate, so use > # it to double check code > time <- c(cracks$time2[1], (cracks$time1 + cracks$time2)[2:8]/2, + cracks$time1[9]) > cdf <- cumsum(cracks$n)/sum(cracks$n) > all.equal(kfit$time, time) [1] TRUE > all.equal(kfit$surv, 1-cdf[c(1:8,8)]) [1] TRUE > rm(time, cdf, kfit) > > > ####################################################### > # > # Valve data > # The input data has id, time, and an indicator of whether there was an > # event at that time: -1=no, 1=yes. No one has an event at their last time. > # Convert the data to (start, stop] form > # The input data has two engines with dual failures: 328 loses 2 valves at > # time 653, and number 402 loses 2 at time 139. For each, fudge the first > # time to be .1 days earlier. > # > temp <- matrix(scan('data.valve'), byrow=T, ncol=3) Read 267 items > > n <- nrow(temp) > valve <- data.frame(id=temp[,1], + time1 = c(0, ifelse(diff(temp[,1])==0, temp[-n,2],0)), + time2 = temp[,2], + status= as.numeric(temp[,3]==1)) > > indx <- (1:nrow(valve))[valve$time1==valve$time2] > valve$time1[indx] <- valve$time1[indx] - .1 > valve$time2[indx-1] <- valve$time2[indx-1] - .1 > > kfit <- survfit(Surv(time1, time2, status) ~1, valve, type='fh2') > > plot(kfit, fun='cumhaz', ylab="Sample Mean Cumulative Failures", xlab='Time', + ylim=range(-log(kfit$lower))) > title("Valve replacement data") > > # The summary.survfit function doesn't have an option for printing out > # cumulative hazards instead of survival --- need to add that > # so I just reprise the central code of print.summary.survfit > xx <- summary(kfit) > temp <- cbind(xx$time, xx$n.risk, xx$n.event, -log(xx$surv), + xx$std.err/xx$surv, -log(xx$upper), -log(xx$lower)) > dimnames(temp) <- list(rep("", nrow(temp)), + c("time", "n.risk", "n.event", "Cum haz", "std.err", + "lower 95%", "upper 95%")) > print(temp, digits=2) time n.risk n.event Cum haz std.err lower 95% upper 95% 61 41 1 0.024 0.025 0.00000 0.073 76 41 1 0.049 0.035 0.00000 0.117 84 41 1 0.073 0.043 0.00000 0.157 87 41 1 0.098 0.049 0.00077 0.194 92 41 1 0.122 0.055 0.01373 0.230 98 41 1 0.146 0.060 0.02779 0.265 120 41 1 0.171 0.065 0.04268 0.299 139 41 1 0.195 0.070 0.05823 0.332 139 41 1 0.220 0.074 0.07432 0.365 165 41 1 0.244 0.078 0.09085 0.397 166 41 1 0.268 0.082 0.10778 0.429 202 41 1 0.293 0.086 0.12503 0.460 206 41 1 0.317 0.089 0.14257 0.492 249 41 1 0.341 0.092 0.16038 0.523 254 41 1 0.366 0.096 0.17841 0.553 258 41 1 0.390 0.099 0.19665 0.584 265 41 1 0.415 0.102 0.21508 0.614 276 41 1 0.439 0.105 0.23369 0.644 298 41 1 0.463 0.108 0.25245 0.674 323 41 1 0.488 0.110 0.27136 0.704 326 41 1 0.512 0.113 0.29041 0.734 328 41 1 0.537 0.116 0.30958 0.764 344 41 1 0.561 0.118 0.32887 0.793 348 41 1 0.585 0.121 0.34827 0.822 349 41 1 0.610 0.123 0.36777 0.852 367 41 1 0.634 0.126 0.38736 0.881 377 41 1 0.659 0.128 0.40705 0.910 404 40 1 0.684 0.131 0.42720 0.940 408 40 1 0.709 0.133 0.44745 0.970 410 40 1 0.734 0.136 0.46777 0.999 449 40 1 0.759 0.138 0.48818 1.029 479 40 1 0.784 0.140 0.50866 1.058 497 40 1 0.809 0.143 0.52922 1.088 538 40 1 0.834 0.145 0.54985 1.117 539 40 1 0.859 0.147 0.57054 1.147 561 40 1 0.884 0.149 0.59129 1.176 563 40 1 0.909 0.151 0.61211 1.205 570 40 1 0.934 0.153 0.63299 1.234 573 40 1 0.959 0.155 0.65392 1.263 581 38 1 0.985 0.158 0.67578 1.294 586 34 1 1.014 0.160 0.69970 1.329 604 22 1 1.060 0.167 0.73221 1.387 621 17 1 1.119 0.178 0.77014 1.467 635 16 1 1.181 0.189 0.81038 1.552 640 16 1 1.244 0.200 0.85188 1.635 646 13 1 1.320 0.215 0.89854 1.742 653 9 1 1.432 0.245 0.95056 1.913 653 9 1 1.543 0.272 1.00909 2.076 > > # Note that I have the same estimates but different SE's. We are using a > # different estimator. It's a statistical argument as to which is > # better (one could defend both sides): do you favor JASA or Technometrics? > rm(temp, kfit, indx, xx) > > ###################################################### > # Turbine data, lognormal fit > turbine <- read.table('data.turbine', + col.names=c("time1", "time2", "n")) > > tfit <- survreg(Surv(time1, time2, type='interval2') ~1, turbine, + dist='lognormal', weights=n, subset=(n>0)) > > summary(tfit) Call: survreg(formula = Surv(time1, time2, type = "interval2") ~ 1, data = turbine, weights = n, subset = (n > 0), dist = "lognormal") Value Std. Error z p (Intercept) 3.700 0.0708 52.23 0.00000 Log(scale) -0.329 0.1232 -2.67 0.00763 Scale= 0.72 Log Normal distribution Loglik(model)= -190.7 Loglik(intercept only)= -190.7 Number of Newton-Raphson Iterations: 6 n= 21 > > # Now, do his plot, but put bootstrap confidence bands on it! > # First, make a simple data set without weights > tdata <- turbine[rep(1:nrow(turbine), turbine$n),] > > qstat <- function(data) { + temp <- survreg(Surv(time1, time2, type='interval2') ~1, data=data, + dist='lognormal') + qsurvreg(plist, temp$coef, temp$scale, dist='lognormal') + } > > {if (exists('bootstrap')) { + bfit <- bootstrap(tdata, qstat, B=1000) + bci <- limits.bca(bfit, probs=c(.025, .975)) + } + else { + values <- matrix(0, nrow=1000, ncol=length(plist)) + n <- nrow(tdata) + for (i in 1:1000) { + subset <- sample(1:n, n, replace=T) + values[i,] <- qstat(tdata[subset,]) + } + bci <- t(apply(values,2, quantile, c(.05, .95))) + } + } > xmat <- cbind(qsurvreg(plist, tfit$coef, tfit$scale, dist='lognormal'), + bci) > > > matplot(xmat, qnorm(plist), + type='l', lty=c(1,2,2), col=c(1,1,1), + log='x', yaxt='n', ylab='Percent', + xlab='Time of Cracking (Hours x 100)') > axis(2, qnorm(plist), format(100*plist), adj=1) > title("Turbine Data") > kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=tdata) > points(kfit$time, qnorm(1-kfit$surv), pch='+') > > dev.off() #close the plot file pdf 2 > > > proc.time() user system elapsed 5.312 0.044 5.358 survival/tests/r_resid.R0000644000175100001440000000767711732700061015075 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) fit1 <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian) fit4 <- survreg(Surv(log(futime), fustat) ~age + ecog.ps, ovarian, dist='extreme') print(fit1) summary(fit4) # Hypothesis (and I'm fairly sure): censorReg shares the fault of many # iterative codes -- it returns the loglik and variance for iteration k # but the coef vector of iteration k+1. Hence the "all.equal" tests # below don't come out perfect. # if (exists('censorReg')) { #true for Splus, not R fit2 <- censorReg(censor(futime, fustat) ~ age + ecog.ps, ovarian) fit3 <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, iter=0, init=c(fit2$coef, log(fit2$scale))) aeq(resid(fit2, type='working')[,1], resid(fit3, type='working')) aeq(resid(fit2, type='response')[,1], resid(fit3, type='response')) temp <- sign(resid(fit3, type='working')) aeq(resid(fit2, type='deviance')[,1], temp*abs(resid(fit3, type='deviance'))) aeq(resid(fit2, type='deviance')[,1], resid(fit3, type='deviance')) } # # Now check fit1 and fit4, which should follow identical iteration paths # These tests should all be true # aeq(fit1$coef, fit4$coef) resid(fit1, type='working') resid(fit1, type='response') resid(fit1, type='deviance') resid(fit1, type='dfbeta') resid(fit1, type='dfbetas') resid(fit1, type='ldcase') resid(fit1, type='ldresp') resid(fit1, type='ldshape') resid(fit1, type='matrix') aeq(resid(fit1, type='working'),resid(fit4, type='working')) #aeq(resid(fit1, type='response'), resid(fit4, type='response'))#should differ aeq(resid(fit1, type='deviance'), resid(fit4, type='deviance')) aeq(resid(fit1, type='dfbeta'), resid(fit4, type='dfbeta')) aeq(resid(fit1, type='dfbetas'), resid(fit4, type='dfbetas')) aeq(resid(fit1, type='ldcase'), resid(fit4, type='ldcase')) aeq(resid(fit1, type='ldresp'), resid(fit4, type='ldresp')) aeq(resid(fit1, type='ldshape'), resid(fit4, type='ldshape')) aeq(resid(fit1, type='matrix'), resid(fit4, type='matrix')) # # Some tests of the quantile residuals # motor <- read.table('data.motor', col.names=c('temp', 'time', 'status')) # These should agree exactly with Ripley and Venables' book fit1 <- survreg(Surv(time, status) ~ temp, data=motor) summary(fit1) # # The first prediction has the SE that I think is correct # The third is the se found in an early draft of Ripley; fit1 ignoring # the variation in scale estimate, except via it's impact on the # upper left corner of the inverse information matrix. # Numbers 1 and 3 differ little for this dataset # predict(fit1, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) fit2 <- survreg(Surv(time, status) ~ temp, data=motor, scale=fit1$scale) predict(fit2, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) fit3 <- fit2 fit3$var <- fit1$var[1:2,1:2] predict(fit3, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) pp <- seq(.05, .7, length=40) xx <- predict(fit1, data.frame(temp=130), type='uquantile', se=T, p=pp) #matplot(pp, cbind(xx$fit, xx$fit+2*xx$se, xx$fit - 2*xx$se), type='l') # # Now try out the various combinations of strata, #predicted, and # number of quantiles desired # fit1 <- survreg(Surv(time, status) ~ inst + strata(inst) + age + sex, lung) qq1 <- predict(fit1, type='quantile', p=.3, se=T) qq2 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) aeq(qq1$fit, qq2$fit[,2]) aeq(qq1$se.fit, qq2$se.fit[,2]) qq3 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, newdata= lung[1:5,]) aeq(qq3$fit, qq2$fit[1:5,]) qq4 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, newdata=lung[7,]) aeq(qq4$fit, qq2$fit[7,]) qq5 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, newdata=lung) aeq(qq2$fit, qq5$fit) aeq(qq2$se.fit, qq5$se.fit) survival/tests/fr_lung.R0000644000175100001440000000131011732700061015054 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # A test with the lung data # This caused problems in one release # # First, get rid of some missings # lung2 <- na.omit(lung[c('time', 'status', 'wt.loss')]) # # Test the logliklihoods # fit <- coxph(Surv(time, status) ~ pspline(wt.loss,3), lung2, x=T) fit0<- coxph(Surv(time, status) ~ 1, lung2) fit1<- coxph(Surv(time, status) ~ fit$x, lung2, iter=0, init=fit$coef) all.equal(fit$loglik[1], fit0$loglik) all.equal(fit$loglik[2], fit1$loglik[2]) # # Check variances # imat <- solve(fit1$var) var2 <- fit$var %*% imat %*% fit$var all.equal(fit$var2, var2) survival/tests/pspline.R0000644000175100001440000000370512350363670015114 0ustar hornikuserslibrary(survival) # # Tests with the pspline function, to verify the prediction aspects # options(na.action=na.exclude) aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) spfit <- coxph(Surv(time, status) ~ pspline(age) + ph.ecog, lung) spfit2 <- coxph(Surv(time, status) ~ pspline(age) + ph.ecog, lung, x=TRUE) x2 <- model.matrix(spfit) all.equal(spfit2$x, x2) keep <- (lung$age < 60) x3 <- model.matrix(spfit, data=lung[keep,]) attr(x3, 'assign') <- NULL #subscripting loses the assign attr below all.equal(napredict(spfit$na.action,x2)[keep,], x3) p2 <- predict(spfit, newdata=lung[keep,]) aeq(p2, predict(spfit)[keep]) p3 <- survfit(spfit) p4 <- survfit(spfit, newdata=lung[1:2,]) temp <- scale(x2[1:2,], center=spfit$means, scale=FALSE)%*% coef(spfit) aeq(p3$time, p4$time) aeq(outer(-log(p3$surv), exp(temp), '*'), -log(p4$surv)) # Check out model.frame spfit3 <- coxph(Surv(time, status) ~ pspline(age) + sex, lung, model=TRUE) #avoid the missing value m2 <- model.frame(spfit3, data=lung[keep,]) all.equal(m2, spfit3$model[keep,]) # # Test of residuals, in response to a reported bug. The routines for # m-resids of penalized models were separate from other m-resid calcs; # refactored to change that. # These are three progam paths that should all lead to the same C routine fit <- coxph(Surv(tstart, tstop, status) ~ sex + treat + pspline(age), cgd) fit2 <- coxph(Surv(tstart, tstop, status) ~ fit$linear, cgd, iter=0, init=1) fit3 <- coxph(Surv(tstart, tstop, status) ~ offset(fit$linear), cgd) all.equal(fit$resid, fit2$resid) all.equal(fit$resid, fit3$resid) # # Check using coxph.detail. The matrix multiply below only is # valid for the breslow approximation. fit4 <- coxph(Surv(tstart, tstop, status) ~ sex + treat + pspline(age), cgd, ties='breslow') dt <- coxph.detail(fit4, riskmat=TRUE) rscore <- exp(fit4$linear) exp4 <- (rscore *dt$riskmat) %*% dt$hazard r4 <- cgd$status - exp4 aeq(r4, fit4$resid) survival/tests/book6.Rout.save0000644000175100001440000001431512536400472016144 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # Tests of the weighted Cox model > # This is section 1.3 of my appendix -- no yet found in any of the > # printings though, it awaits the next edition > # > # Efron approximation > # > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), + status= c(1,0,1,1,1,0,0,1,0), + x= c(2,0,1,1,0,1,0,1,0), + wt = c(1,2,3,4,3,2,1,2,1)) > xx <- testw1$wt > > # Efron estimate > byhand <- function(beta, newx=0) { + r <- exp(beta) + a <- 7*r +3; b<- 4*r+2 + loglik <- 11*beta - (log(r^2 + 11*r +7) + 10*log(11*r +5)/3 + + 10*log(a*2/3 +b)/3 + 10*log(a/3 +b)/3 +2*log(2*r+1)) + + hazard <- c(1/(r^2 + 11*r +7), + 10/(3*c(11*r +5, a*2/3 +b, a/3+b)), 2/(2*r+1)) + temp <- c(hazard[1], hazard[1]+hazard[2] + hazard[3]*2/3 + hazard[4]/3, + cumsum(hazard)[4:5]) + risk <- c(r^2, 1,r,r,1,r,1,r,1) + expected <- risk* temp[c(1,1,2,2,2,3,3,4,4)] + + # The matrix of weights, one row per obs, one col per death + # deaths at 1,2,2,2, and 4 + riskmat <- matrix(c(1,1,1,1,1,1,1,1,1, + 0,0,1,1,1,1,1,1,1, + 0,0,2/3,2/3,2/3,1,1,1,1, + 0,0,1/3,1/3,1/3,1,1,1,1, + 0,0,0,0,0,0,0,1,1), ncol=5) + wtmat <- diag(c(r^2, 2, 3*r, 4*r, 3, 2*r, 1, 2*r, 1)) %*% riskmat + + x <- c(2,0,1,1,0,1,0,1,0) + xbar <- colSums(x*wtmat)/ colSums(wtmat) + imat <- (4*r^2 + 11*r)*hazard[1] - xbar[1]^2 + + 10* mean(xbar[2:4] - xbar[2:4]^2) + 2*(xbar[5] - xbar[5]^2) + + status <- c(1,0,1,1,1,0,0,1,0) + wt <- c(1,2,3,4,3,2,1,2,1) + # Table of sums for score resids + hazmat <- riskmat %*% diag(c(1,10/3,10/3, 10/3,2)/colSums(wtmat)) + dM <- -risk*hazmat #Expected part + dM[1,1] <- dM[1,1] +1 # deaths at time 1 + for (i in 2:4) dM[3:5, i] <- dM[3:5,i] + 1/3 + dM[8,5] <- dM[8,5] +1 + mart <- rowSums(dM) + resid <-dM * outer(x, xbar ,'-') + + # Increments to the variance of the hazard + var.g <- cumsum(hazard^2* c(1,3/10, 3/10, 3/10, 1/2)) + var.d <- cumsum((xbar-newx)*hazard) + + sxbar <- c(xbar[1], mean(xbar[2:4]), xbar[5]) #xbar for Schoen + list(loglik=loglik, imat=imat, hazard=hazard, xbar=xbar, + mart=status-expected, expected=expected, + score=rowSums(resid), schoen=c(2,1,1,0,1) - sxbar[c(1,2,2,2,3)], + varhaz=((var.g + var.d^2/imat)* exp(2*beta*newx))[c(1,4,5)]) + } > > # Verify > temp <- byhand(0,0) > aeq(temp$xbar, c(13/19, 11/16, 26/38, 19/28, 2/3)) [1] TRUE > aeq(temp$hazard, c(1/19, 5/24, 5/19, 5/14, 2/3)) [1] TRUE > > fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, iter=0) > fit <- coxph(Surv(time, status) ~x, testw1, weights=wt) > > truth0 <- byhand(0,pi) > aeq(fit0$loglik[1], truth0$loglik) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')) [1] TRUE > sfit <- survfit(fit0, list(x=pi), censor=FALSE) > aeq(sfit$std.err^2, truth0$var) [1] TRUE > aeq(-log(sfit$surv), cumsum(truth0$hazard)[c(1,4,5)]) [1] TRUE > > truth <- byhand(fit$coef, .3) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')) [1] TRUE > > sfit <- survfit(fit, list(x=.3), censor=FALSE) > aeq(sfit$std.err^2, truth$var) [1] TRUE > aeq(-log(sfit$surv), (cumsum(truth$hazard)* exp(fit$coef*.3))[c(1,4,5)]) [1] TRUE > > > fit0 Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, iter = 0) coef exp(coef) se(coef) z p x 0.000 1.000 0.584 0 1 Likelihood ratio test=0 on 1 df, p=1 n= 9, number of events= 5 > summary(fit) Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt) n= 9, number of events= 5 coef exp(coef) se(coef) z Pr(>|z|) x 0.8726 2.3931 0.7126 1.225 0.221 exp(coef) exp(-coef) lower .95 upper .95 x 2.393 0.4179 0.5921 9.672 Concordance= 0.638 (se = 0.159 ) Rsquare= 0.177 (max possible= 0.999 ) Likelihood ratio test= 1.75 on 1 df, p=0.1858 Wald test = 1.5 on 1 df, p=0.2207 Score (logrank) test = 1.58 on 1 df, p=0.2094 > resid(fit0, type='score') 1 2 3 4 5 6 1.24653740 0.03601108 0.14118105 0.14118105 -0.30336782 -0.27962308 7 8 9 0.60164259 -0.16851197 1.04608703 > resid(fit0, type='scho') 1 2 2 2 4 1.3157895 0.3165727 0.3165727 -0.6834273 0.3333333 > > resid(fit, type='score') 1 2 3 4 5 6 0.88116056 0.02477248 0.06057806 0.06057806 -0.59724033 -0.16737066 7 8 9 0.38040295 -0.13750290 0.66631324 > resid(fit, type='scho') 1 2 2 2 4 1.0325955 0.1621759 0.1621759 -0.8378241 0.1728229 > > rr1 <- resid(fit, type='mart') > rr2 <- resid(fit, type='mart', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > rr1 <- resid(fit, type='score') > rr2 <- resid(fit, type='score', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > > proc.time() user system elapsed 0.232 0.020 0.246 survival/tests/anova.Rout.save0000644000175100001440000000357212164375073016240 0ustar hornikusers R version 3.0.0 (2013-04-03) -- "Masked Marvel" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > # Test out anova, with strata terms > # > options(na.action=na.omit) > library(survival) Loading required package: splines > > fit1 <- coxph(Surv(time, status) ~ ph.ecog + wt.loss + strata(sex) + + poly(age,3), lung) > ztemp <- anova(fit1) > > tdata <- na.omit(lung[, c('time', 'status', 'ph.ecog', 'wt.loss', 'sex', 'age')]) > fit2 <- coxph(Surv(time, status)~ ph.ecog + wt.loss + poly(age,3) + strata(sex), + data=tdata) > ztemp2 <- anova(fit2) > all.equal(ztemp, ztemp2) [1] TRUE > > > fit2 <- coxph(Surv(time, status) ~ ph.ecog + wt.loss + strata(sex), tdata) > fit3 <- coxph(Surv(time, status) ~ ph.ecog + strata(sex), tdata) > > all.equal(ztemp$loglik, c(fit1$loglik[1], fit3$loglik[2], fit2$loglik[2], + fit1$loglik[2])) [1] TRUE > all.equal(ztemp$Chisq[-1], 2* diff(ztemp$loglik)) [1] TRUE > all.equal(ztemp$Df[-1], c(1,1,3)) [1] TRUE > > ztemp2 <- anova(fit3, fit2, fit1) > all.equal(ztemp2$loglik, ztemp$loglik[-1]) [1] TRUE > all.equal(ztemp2$Chisq[2:3], ztemp$Chisq[3:4]) [1] TRUE > # Change from ztemp2$P; it's a data frame and in R 3.0.2 abbreviated names > # give a warning > all.equal(ztemp2[[4]][2:3], ztemp[[4]][3:4]) [1] TRUE > > > > proc.time() user system elapsed 0.284 0.020 0.301 survival/tests/testci2.Rout.save0000644000175100001440000001373212055211052016471 0ustar hornikusers R version 2.15.2 (2012-10-26) -- "Trick or Treat" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > > # > # Test the multi-state version of the CI curve > # > tdata <- data.frame(id=c(1,1,1,1, 2,2,2, 3,3, 4,4,4,4, 5, 6, 6), + time1=c(0, 10,20,30, 0, 5, 15, 0, 20, 0, 6,18,34, 0, 0,15), + time2=c(10,20,30,40, 5, 15,25, 20, 22, 6,18,34,50,10,15,20), + status=c(1,1,1,1, 1,1,1, 1,0, 1,1,1,0,0,1,0), + event= letters[c(1,2,3,4, 2,4,3, 2,2, 3,1,2,2,1, 1,1)], + wt = c(2,2,2,2, 1,1,1, 3,3, 1,1,1,1, 2, 1,1), + stringsAsFactors=TRUE) > tdata$stat2 <- factor(tdata$status * as.numeric(tdata$event), + labels=c(" ", levels(tdata$event))) > > fit <- survfit(Surv(time1, time2, stat2) ~1, id=id, weight=wt, tdata) > > # The exact figures for testci2. > # The subject data of id, weight, (transition time, transition) > > #1: 2 (10, 0->a) (20, a->b) (30, b->c) (40, c->d) no data after 40=censored > #2: 1 ( 5, 0->b) (15, b->d) (25, d->c) no data after 25 implies censored then > #3: 3 (20, 0->b) (22, censor) > #4: 1 ( 6, 0->c) (18, c->a) (34, a->b) (50, censor) > #5: 2 (10, censor) > #6: 1 (15, 0->a) (20, censor) > > # Each line below follows a subject through time as a (state, rdist weight) pair > # using the redistribute to the right algorithm. > # RDR algorithm: at each censoring (or last fu) a subject's weight is put into > # a "pool" for that state and their weight goes to zero. The pool is > # dynamically shared between all members of the state proportional to their > # original case weight, when someone leaves they take their portion of the > # pool to the new state. > > # Table of case weights and state, blank is weight of zero > # time 5 6 10 15 18 20 25 30 34 40 50 > # ----------------------------------------------------------------------- > # id, wt > # 1, 2 - - a a a b b c c d > # 2, 1 b b b d d d c > # 3, 3 - - - - - b > # 4, 1 - c c c a a a a b b b > # 5, 2 - - - > # 6, 1 - - - a a a > > # Pool weights > # 10 10+ 15 18 20 20+ 22+ 25 25+ 30 34 40 40+ > # - 0 2 3/2 3/2 0 > # a 0 0 1/2 1/2 1/4 5/4 5/4 5/4 5/4 5/4 > # b 0 0 0 0 7/4 7/4 19/4 19/4 19/4 5/4 5/4 5/4 > # c 0 0 0 0 0 1 23/4 23/4 > # d 0 0 0 0 0 23/4 31/4 > > # fit$prev for time i and state j = total weight at that time/state in the > # above table (original weight + redistrib), divided by 10. > > # time 5 6 10 15 18 20 25 30 34 40 50 > truth <- matrix(c(0, 0, 2, 3, 4, 2, 1, 1, 0, 0, 0, + 1, 1, 1, 0, 0, 5, 2, 0, 1, 1, 1, + 0, 1, 1, 1, 0, 0, 1, 2, 2, 0, 0, + 0, 0, 0, 1, 1, 1, 0, 0, 0, 2, 0) + + c(0, 0, 0, .5, .5, 1/4, 5/4, 5/4, 0, 0, 0, + 0, 0, 0, 0, 0, 7/4, 19/4, 0, 5/4, 5/4, 5/4, + 0, 0, 0, 0, 0, 0, 0, 23/4, 23/4, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 23/4, 31/4), + ncol=4) > truth <- truth[c(1:6, 6:11),]/10 #the explicit censor at 22 > > #dimnames(truth) <- list(c(5, 6, 10, 15, 18, 20, 25, 30, 34, 40, 50), > # c('a', 'b', 'c', 'd') > all.equal(truth, fit$prev) [1] TRUE > > # Test the dfbetas > dfbeta <- array(0., dim=c(6, nrow(fit$prev), ncol(fit$prev))) > eps <- 1e-6 > for (i in 1:6) { + twt <- tdata$wt + twt[tdata$id ==i] <- twt[tdata$id==i] + eps + tfit <- survfit(Surv(time1, time2, stat2) ~ cluster(id), tdata, + weight=twt) + dfbeta[i,,] <- (tfit$prev - fit$prev)/eps + } > twt <- tdata$wt[match(1:6, tdata$id)] > temp <- (twt*dfbeta) * dfbeta > tstd <- sqrt(apply(temp, 2:3, sum)) > all.equal(tstd, fit$std.err, tolerance=eps) [1] TRUE > > if (FALSE) { + # a plot of the data that helped during creation of the example + plot(c(0,50), c(1,6), type='n', xlab='time', ylab='subject') + with(tdata, segments(time1, id, time2, id)) + with(tdata, text(time2, id, as.numeric(stat2)-1, cex=1.5, col=2)) + } > > if (FALSE) { + # The following lines test out 4 error messages in the routine + # + # Gap in follow-up time, id 2 + survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 4, 6, 3), factor(c(0,0,1,1,0,2))) ~1, + id=c(1,1,1,2,2,3)) + # mismatched weights + survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,0,1,1,0,2))) ~1, + id=c(1,1,1,2,2,3), weights=c(1,1,2,1,1,4)) + # in two groups at once + survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,0,1,1,0,2))) ~ + c(1,1,2,1,1,2), id=c(1,1,1,2,2,3)) + # state change that isn't a state change (went from 1 to 1) + survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,1,1,1,0,2))) ~1, + id=c(1,1,1,2,2,3)) + } > > > proc.time() user system elapsed 0.264 0.012 0.273 survival/tests/prednew.R0000644000175100001440000000503712113157116015077 0ustar hornikusers# # Make sure that the newdata argument works for various # predictions # We purposely use a subset of the lung data that has only some # of the levels of the ph.ecog library(survival) options(na.action=na.exclude, contrasts=c('contr.treatment', 'contr.poly')) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) myfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog) + strata(sex), lung) keep <- which(lung$inst<13 & (lung$ph.ecog==1 | lung$ph.ecog==2)) p1 <- predict(myfit, type='lp') p2 <- predict(myfit, type="lp", newdata=lung[keep,]) p3 <- predict(myfit, type='lp', se.fit=TRUE) p4 <- predict(myfit, type="lp", newdata=lung[keep,], se.fit=TRUE) aeq(p1[keep], p2) aeq(p1, p3$fit) aeq(p1[keep], p4$fit) aeq(p3$se.fit[keep], p4$se.fit) p1 <- predict(myfit, type='risk') p2 <- predict(myfit, type="risk", newdata=lung[keep,]) p3 <- predict(myfit, type='risk', se.fit=TRUE) p4 <- predict(myfit, type="risk", newdata=lung[keep,], se.fit=TRUE) aeq(p1[keep], p2) aeq(p1, p3$fit) aeq(p1[keep], p4$fit) aeq(p3$se.fit[keep], p4$se.fit) # The all.equal fails for type=expected, Efron approx, and tied death # times due to use of an approximation. See comments in the source code. myfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog) + strata(sex), data=lung, method='breslow') p1 <- predict(myfit, type='expected') p2 <- predict(myfit, type="expected", newdata=lung[keep,]) p3 <- predict(myfit, type='expected', se.fit=TRUE) p4 <- predict(myfit, type="expected", newdata=lung[keep,], se.fit=TRUE) aeq(p1[keep], p2) aeq(p1, p3$fit) aeq(p1[keep], p4$fit) aeq(p3$se.fit[keep], p4$se.fit) p1 <- predict(myfit, type='terms') p2 <- predict(myfit, type="terms",newdata=lung[keep,]) p3 <- predict(myfit, type='terms', se.fit=T) p4 <- predict(myfit, type="terms",newdata=lung[keep,], se.fit=T) aeq(p1[keep,], p2) aeq(p1, p3$fit) aeq(p1[keep,], p4$fit) aeq(p3$se.fit[keep,], p4$se.fit) # # Check out the logic whereby predict does not need to # recover the model frame. The first call should not # need to do so, the second should in each case. # myfit <- coxph(Surv(time, status) ~ age + factor(sex), lung, x=T) p1 <- predict(myfit, type='risk', se=T) myfit2 <- coxph(Surv(time, status) ~ age + factor(sex), lung) p2 <- predict(myfit2, type='risk', se=T) aeq(p1$fit, p2$fit) aeq(p1$se, p2$se) p1 <- predict(myfit, type='expected', se=T) p2 <- predict(myfit2, type='expected', se=T) aeq(p1$fit, p2$fit) aeq(p1$se.fit, p2$se.fit) p1 <- predict(myfit, type='terms', se=T) p2 <- predict(myfit2, type='terms', se=T) aeq(p1$fit, p2$fit) aeq(p1$se.fit, p2$se.fit) survival/tests/survreg2.Rout.save0000644000175100001440000000642112334225345016702 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > options(na.action=na.exclude, contrasts=c('contr.treatment', 'contr.poly')) > > # Verify stratified fits in a simple way, but combining two data > # sets and doing a single fit > # > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > tdata <- data.frame(time=c(lung$time, ovarian$futime), + status=c(lung$status-1, ovarian$fustat), + group =rep(0:1, c(nrow(lung), nrow(ovarian)))) > fit1 <- survreg(Surv(time, status) ~ 1, lung) > fit2 <- survreg(Surv(futime, fustat) ~ 1, ovarian) > fit3 <- survreg(Surv(time, status) ~ group + strata(group), tdata) > > aeq(c(fit1$coef, fit2$coef-fit1$coef), fit3$coef) [1] TRUE > aeq(c(fit1$scale, fit2$scale), fit3$scale) [1] TRUE > aeq(fit1$loglik[2] + fit2$loglik[2], fit3$loglik[2]) [1] TRUE > > # > # Test out the cluster term in survreg, which means first a test > # of the dfbeta residuals > # I also am checking that missing values propogate > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > fit1 <- survreg(Surv(time, status) ~ x + cluster(1:7), test1) > > db1 <- resid(fit1, 'dfbeta') > ijack <-db1 > eps <- 1e-7 > for (i in 1:7) { + temp <- rep(1.0,7) + temp[i] <- 1-eps + tfit <- survreg(Surv(time, status) ~ x, test1, weight=temp) + ijack[i,] <- c(tfit$coef, log(tfit$scale)) + } > ijack[2,] <- NA # stick the NA back in > ijack <- (rep(c(fit1$coef, log(fit1$scale)), each=nrow(db1)) - ijack)/eps > all.equal(db1, ijack, tolerance=eps) [1] TRUE > all.equal(t(db1[-2,])%*% db1[-2,], fit1$var) [1] TRUE > > # This is a harder test since there are multiple strata and multiple > # obs/subject. Use of enum + strata(enum) in essenence fits a different > # baseline Weibull to each strata, with common coefficients for rx, size, and > # number. > fit1 <- survreg(Surv(stop-start, event) ~ rx + size + number + + factor(enum) + strata(enum), data=bladder2) > > db1 <- resid(fit1, type='dfbeta', collapse=bladder2$id) > ijack <- db1 # a matrix of the same size > for (i in 1:nrow(db1)) { + twt <- rep(1., nrow(bladder2)) + twt[bladder2$id==i] <- 1-eps + tfit <- survreg(Surv(stop-start, event) ~ rx + size + number + + factor(enum) + strata(enum), data=bladder2, + weight=twt) + ijack[i,] <- c(coef(tfit), log(tfit$scale)) + } > ijack <- (rep(c(fit1$coef, log(fit1$scale)), each=nrow(db1)) - ijack)/eps > all.equal(db1, ijack, tolerance=eps*2) [1] TRUE > > > proc.time() user system elapsed 0.732 0.028 0.759 survival/tests/survfit2.Rout.save0000644000175100001440000000213011732700061016672 0ustar hornikusers R version 2.11.0 (2010-04-22) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Check out the Dory&Korn confidence interval option > # > tdata <- data.frame(time= 1:10, + status=c(1,0,1,0,1,0,0,0,1,0)) > > fit1 <- survfit(Surv(time, status) ~1, tdata, conf.lower='modified') > fit2 <- survfit(Surv(time, status) ~1, tdata) > > stdlow <- fit2$std * sqrt(c(1, 10/9, 1, 8/7, 1, 6/5, 6/4, 6/3, 1, 2/1)) > lower <- exp(log(fit2$surv) - qnorm(.975)*stdlow) > all.equal(fit1$lower, lower, check.attributes=FALSE) [1] TRUE > survival/tests/fr_rat1.Rout.save0000644000175100001440000000734212536401430016457 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # Tests using the rats data > # > # (Female rats, from Mantel et al, Cancer Research 37, > # 3863-3868, November 77) > > rfit <- coxph(Surv(time,status) ~ rx + frailty(litter), rats, + method='breslow', subset= (sex=='f')) > names(rfit) [1] "coefficients" "var" "var2" [4] "loglik" "iter" "linear.predictors" [7] "residuals" "means" "concordance" [10] "method" "frail" "fvar" [13] "df" "df2" "penalty" [16] "pterms" "assign2" "history" [19] "coxlist1" "printfun" "n" [22] "nevent" "terms" "assign" [25] "wald.test" "y" "formula" [28] "call" > rfit Call: coxph(formula = Surv(time, status) ~ rx + frailty(litter), data = rats, subset = (sex == "f"), method = "breslow") coef se(coef) se2 Chisq DF p rx 0.906 0.323 0.319 7.882 1.0 0.005 frailty(litter) 16.888 13.8 0.253 Iterations: 6 outer, 25 Newton-Raphson Variance of random effect= 0.474 I-likelihood = -181.1 Degrees of freedom for terms= 1.0 13.9 Likelihood ratio test=36.3 on 14.8 df, p=0.00144 n= 150 > > rfit$iter [1] 6 25 > rfit$df [1] 0.975943 13.854864 > rfit$history[[1]] $theta [1] 0.4742849 $done c.loglik TRUE $history theta loglik c.loglik [1,] 0.0000000 -181.8451 -181.8451 [2,] 1.0000000 -168.3683 -181.5458 [3,] 0.5000000 -173.3117 -181.0788 [4,] 0.3090061 -175.9446 -181.1490 [5,] 0.4645720 -173.7590 -181.0775 [6,] 0.4736210 -173.6431 -181.0773 $c.loglik [1] -181.0773 > > rfit1 <- coxph(Surv(time,status) ~ rx + frailty(litter, theta=1), rats, + method='breslow', subset=(sex=="f")) > rfit1 Call: coxph(formula = Surv(time, status) ~ rx + frailty(litter, theta = 1), data = rats, subset = (sex == "f"), method = "breslow") coef se(coef) se2 Chisq DF p rx 0.918 0.327 0.321 7.851 1.0 0.0051 frailty(litter, theta = 1 27.245 22.7 0.2324 Iterations: 1 outer, 6 Newton-Raphson Variance of random effect= 1 I-likelihood = -181.5 Degrees of freedom for terms= 1.0 22.7 Likelihood ratio test=50.7 on 23.7 df, p=0.001 n= 150 > > rfit2 <- coxph(Surv(time,status) ~ frailty(litter), rats, subset=(sex=='f')) > rfit2 Call: coxph(formula = Surv(time, status) ~ frailty(litter), data = rats, subset = (sex == "f")) coef se(coef) se2 Chisq DF p frailty(litter) 18 14.6 0.24 Iterations: 6 outer, 22 Newton-Raphson Variance of random effect= 0.504 I-likelihood = -184.8 Degrees of freedom for terms= 14.6 Likelihood ratio test=30 on 14.6 df, p=0.0101 n= 150 > > proc.time() user system elapsed 0.216 0.028 0.237 survival/tests/survreg2.R0000644000175100001440000000455012271462471015221 0ustar hornikuserslibrary(survival) options(na.action=na.exclude, contrasts=c('contr.treatment', 'contr.poly')) # Verify stratified fits in a simple way, but combining two data # sets and doing a single fit # aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) tdata <- data.frame(time=c(lung$time, ovarian$futime), status=c(lung$status-1, ovarian$fustat), group =rep(0:1, c(nrow(lung), nrow(ovarian)))) fit1 <- survreg(Surv(time, status) ~ 1, lung) fit2 <- survreg(Surv(futime, fustat) ~ 1, ovarian) fit3 <- survreg(Surv(time, status) ~ group + strata(group), tdata) aeq(c(fit1$coef, fit2$coef-fit1$coef), fit3$coef) aeq(c(fit1$scale, fit2$scale), fit3$scale) aeq(fit1$loglik[2] + fit2$loglik[2], fit3$loglik[2]) # # Test out the cluster term in survreg, which means first a test # of the dfbeta residuals # I also am checking that missing values propogate test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) fit1 <- survreg(Surv(time, status) ~ x + cluster(1:7), test1) db1 <- resid(fit1, 'dfbeta') ijack <-db1 eps <- 1e-7 for (i in 1:7) { temp <- rep(1.0,7) temp[i] <- 1-eps tfit <- survreg(Surv(time, status) ~ x, test1, weight=temp) ijack[i,] <- c(tfit$coef, log(tfit$scale)) } ijack[2,] <- NA # stick the NA back in ijack <- (rep(c(fit1$coef, log(fit1$scale)), each=nrow(db1)) - ijack)/eps all.equal(db1, ijack, tolerance=eps) all.equal(t(db1[-2,])%*% db1[-2,], fit1$var) # This is a harder test since there are multiple strata and multiple # obs/subject. Use of enum + strata(enum) in essenence fits a different # baseline Weibull to each strata, with common coefficients for rx, size, and # number. fit1 <- survreg(Surv(stop-start, event) ~ rx + size + number + factor(enum) + strata(enum), data=bladder2) db1 <- resid(fit1, type='dfbeta', collapse=bladder2$id) ijack <- db1 # a matrix of the same size for (i in 1:nrow(db1)) { twt <- rep(1., nrow(bladder2)) twt[bladder2$id==i] <- 1-eps tfit <- survreg(Surv(stop-start, event) ~ rx + size + number + factor(enum) + strata(enum), data=bladder2, weight=twt) ijack[i,] <- c(coef(tfit), log(tfit$scale)) } ijack <- (rep(c(fit1$coef, log(fit1$scale)), each=nrow(db1)) - ijack)/eps all.equal(db1, ijack, tolerance=eps*2) survival/tests/book3.R0000644000175100001440000000743412350317406014456 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # # Tests from the appendix of Therneau and Grambsch # c. Data set 2 and Breslow estimate # test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) byhand <- function(beta, newx=0) { r <- exp(beta) loglik <- 4*beta - log(r+1) - log(r+2) - 3*log(3*r+2) - 2*log(3*r+1) u <- 1/(r+1) + 1/(3*r+1) + 4/(3*r+2) - ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) imat <- r/(r+1)^2 + 2*r/(r+2)^2 + 6*r/(3*r+2)^2 + 3*r/(3*r+1)^2 + 3*r/(3*r+1)^2 + 12*r/(3*r+2)^2 hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), 2/(3*r+2) ) xbar <- c(r/(r+1), r/(r+2), 3*r/(3*r+2), 3*r/(3*r+1), 3*r/(3*r+1), 3*r/(3*r+2)) # The matrix of weights, one row per obs, one col per time # deaths at 2,3,6,7,8,9 wtmat <- matrix(c(1,0,0,0,1,0,0,0,0,0, 0,1,0,1,1,0,0,0,0,0, 0,0,1,1,1,0,1,1,0,0, 0,0,0,1,1,0,1,1,0,0, 0,0,0,0,1,1,1,1,0,0, 0,0,0,0,0,1,1,1,1,1), ncol=6) wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat x <- c(1,0,0,1,0,1,1,1,0,0) status <- c(1,1,1,1,1,1,1,0,0,0) xbar <- colSums(wtmat*x)/ colSums(wtmat) n <- length(x) # Table of sums for score and Schoenfeld resids hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time dM <- -hazmat #Expected part for (i in 1:6) dM[i,i] <- dM[i,i] +1 #observed dM[7,6] <- dM[7,6] +1 # observed mart <- rowSums(dM) # Table of sums for score and Schoenfeld resids # Looks like the last table of appendix E.2.1 of the book resid <- dM * outer(x, xbar, '-') score <- rowSums(resid) scho <- colSums(resid) # We need to split the two tied times up, to match coxph scho <- c(scho[1:5], scho[6]/2, scho[6]/2) var.g <- cumsum(hazard*hazard /c(1,1,1,1,1,2)) var.d <- cumsum( (xbar-newx)*hazard) surv <- exp(-cumsum(hazard) * exp(beta*newx)) varhaz <- (var.g + var.d^2/imat)* exp(2*beta*newx) list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard, mart=mart, score=score, rmat=resid, scho=scho, surv=surv, var=varhaz) } aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) fit0 <-coxph(Surv(start, stop, event) ~x, test2, iter=0, method='breslow') truth0 <- byhand(0,0) aeq(truth0$loglik, fit0$loglik[1]) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')) sfit <- survfit(fit0, list(x=0), censor=FALSE) aeq(sfit$std.err^2, truth0$var) aeq(sfit$surv, truth0$surv) fit <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-8, method='breslow') truth <- byhand(fit$coef, 0) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')) expect <- predict(fit, type='expected', newdata=test2) #force recalc aeq(test2$event -fit$resid, expect) #tests the predict function sfit <- survfit(fit, list(x=0), censor=FALSE) aeq(sfit$std.err^2, truth$var) aeq(-log(sfit$surv), (cumsum(truth$haz))) # # Done with the formal test, now print out lots of bits # resid(fit) resid(fit, 'scor') resid(fit, 'scho') predict(fit, type='lp') predict(fit, type='risk') predict(fit, type='expected') predict(fit, type='terms') predict(fit, type='lp', se.fit=T) predict(fit, type='risk', se.fit=T) predict(fit, type='expected', se.fit=T) predict(fit, type='terms', se.fit=T) summary(survfit(fit)) summary(survfit(fit, list(x=2))) survival/tests/book1.Rout.save0000644000175100001440000001621512350326212016131 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # > # Tests from the appendix of Therneau and Grambsch > # a. Data set 1 and Breslow estimate > # The data below is not in time order, to also test sorting, and has 1 NA > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > # Breslow estimate > byhand1 <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 2*beta - (log(3*r+3) + 2*log(r+3)) + u <- (6 + 3*r - r^2) / ((r+1)*(r+3)) + imat <- r/(r+1)^2 + 6*r/(r+3)^2 + + x <- c(1,1,1,0,0,0) + status <- c(1,0,1,1,0,1) + xbar <- c(r/(r+1), r/(r+3), 0, 0) # at times 1, 6, 8 and 9 + haz <- c(1/(3*r+3), 2/(r+3), 0, 1 ) + ties <- c(1,1,2,2,3,4) + wt <- c(r,r,r,1,1,1) + mart <- c(1,0,1,1,0,1) - wt* (cumsum(haz))[ties] #martingale residual + + a <- 3*(r+1)^2; b<- (r+3)^2 + score <- c((2*r+3)/a, -r/a, -r/a + 3*(3-r)/b, r/a - r*(r+1)/b, + r/a + 2*r/b, r/a + 2*r/b) + + # Schoenfeld residual + scho <- c(1/(r+1), 1- (r/(3+r)), 0-(r/(3+r)) , 0) + + surv <- exp(-cumsum(haz)* exp(beta*newx)) + varhaz.g <- cumsum(c(1/(3*r+3)^2, 2/(r+3)^2, 0, 1 )) + + varhaz.d <- cumsum((newx-xbar) * haz) + + varhaz <- (varhaz.g + varhaz.d^2/ imat) * exp(2*beta*newx) + + names(xbar) <- names(haz) <- 1:4 + names(surv) <- names(varhaz) <- 1:4 + list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=haz, + mart=mart, score=score, + scho=scho, surv=surv, var=varhaz, + varhaz.g=varhaz.g, varhaz.d=varhaz.d) + } > > > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > fit0 <-coxph(Surv(time, status) ~x, test1, iter=0, method='breslow') > truth0 <- byhand1(0,0) > aeq(truth0$loglik, fit0$loglik[1]) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid[c(2:6,1)]) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')[c(3:7,1)]) [1] TRUE > sfit <- survfit(fit0, list(x=0)) > aeq(sfit$std.err^2, c(7/180, 2/9, 2/9, 11/9)) [1] TRUE > aeq(resid(fit0, 'score'), c(5/24, NA, 5/12, -1/12, 7/24, -1/24, 5/24)) [1] TRUE > > fit1 <- coxph(Surv(time, status) ~x, test1, iter=1, method='breslow') > aeq(fit1$coef, 8/5) [1] TRUE > > # This next gives an ignorable warning message > fit2 <- coxph(Surv(time, status) ~x, test1, method='breslow', iter=2) Warning message: In fitter(X, Y, strats, offset, init, control, weights = weights, : Ran out of iterations and did not converge > aeq(round(fit2$coef, 6), 1.472724) [1] TRUE > > fit <- coxph(Surv(time, status) ~x, test1, method='breslow', eps=1e-8) > aeq(round(fit$coef,7), 1.4752849) [1] TRUE > truth <- byhand1(fit$coef, 0) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid[c(2:6,1)]) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')[c(3:7,1)]) [1] TRUE > expect <- predict(fit, type='expected', newdata=test1) #force recalc > aeq(test1$status[-2] -fit$resid, expect[-2]) #tests the predict function [1] TRUE > > sfit <- survfit(fit, list(x=0), censor=FALSE) > aeq(sfit$std.err^2, truth$var[c(1,2,4)]) # sfit skips time 8 (no events there) [1] TRUE > aeq(-log(sfit$surv), (cumsum(truth$haz))[c(1,2,4)]) [1] TRUE > sfit <- survfit(fit, list(x=0), censor=TRUE) > aeq(sfit$std.err^2, truth$var) [1] TRUE > aeq(-log(sfit$surv), (cumsum(truth$haz))) [1] TRUE > > > # > # Done with the formal test, now print out lots of bits > # > resid(fit) 1 2 3 4 5 6 7 -0.3333333 NA 0.7287136 -0.2712864 -0.4574271 0.6666667 -0.3333333 > resid(fit, 'scor') 1 2 3 4 5 6 0.21138938 NA 0.13564322 -0.05049744 -0.12624360 -0.38168095 7 0.21138938 > resid(fit, 'scho') 1 6 6 9 0.1861407 0.4069297 -0.5930703 0.0000000 > > predict(fit, type='lp') [1] -0.7376425 NA 0.7376425 0.7376425 0.7376425 -0.7376425 -0.7376425 > predict(fit, type='risk') [1] 0.4782401 NA 2.0910001 2.0910001 2.0910001 0.4782401 0.4782401 > predict(fit, type='expected') 1 2 3 4 5 6 7 1.3333333 NA 0.2712864 0.2712864 1.4574271 0.3333333 0.3333333 > predict(fit, type='terms') x 1 -0.7376425 2 NA 3 0.7376425 4 0.7376425 5 0.7376425 6 -0.7376425 7 -0.7376425 > predict(fit, type='lp', se.fit=T) $fit 1 2 3 4 5 6 7 -0.7376425 NA 0.7376425 0.7376425 0.7376425 -0.7376425 -0.7376425 $se.fit 1 2 3 4 5 6 7 0.6278672 NA 0.6278672 0.6278672 0.6278672 0.6278672 0.6278672 > predict(fit, type='risk', se.fit=T) $fit 1 2 3 4 5 6 7 0.4782401 NA 2.0910001 2.0910001 2.0910001 0.4782401 0.4782401 $se.fit 1 2 3 4 5 6 7 0.4342009 NA 0.9079142 0.9079142 0.9079142 0.4342009 0.4342009 > predict(fit, type='expected', se.fit=T) $fit 1 2 3 4 5 6 7 1.3333333 NA 0.2712864 0.2712864 1.4574271 0.3333333 0.3333333 $se.fit [1] 1.0540926 NA 0.2785989 0.2785989 1.1069433 0.3333333 0.3333333 > predict(fit, type='terms', se.fit=T) $fit x 1 -0.7376425 2 NA 3 0.7376425 4 0.7376425 5 0.7376425 6 -0.7376425 7 -0.7376425 $se.fit x 1 0.6278672 2 NA 3 0.6278672 4 0.6278672 5 0.6278672 6 0.6278672 7 0.6278672 > > summary(survfit(fit)) Call: survfit(formula = fit) time n.risk n.event survival std.err lower 95% CI upper 95% CI 1 6 1 0.8783 0.122 0.66827 1 6 4 2 0.4981 0.218 0.21125 1 9 1 1 0.0615 0.150 0.00051 1 > summary(survfit(fit, list(x=2))) Call: survfit(formula = fit, newdata = list(x = 2)) time n.risk n.event survival std.err lower 95% CI upper 95% CI 1 6 1 3.05e-01 6.50e-01 4.72e-03 1 6 4 2 1.71e-03 1.98e-02 2.33e-13 1 9 1 1 8.52e-12 5.29e-10 1.22e-64 1 > > proc.time() user system elapsed 0.240 0.028 0.265 survival/tests/nested.R0000644000175100001440000000065012052731313014710 0ustar hornikuserslibrary(survival) # # A test of nesting. It makes sure the model.frame is built correctly # tfun <- function(fit, mydata) { survfit(fit, newdata=mydata) } myfit <- coxph(Surv(time, status) ~ age + factor(sex), lung) temp1 <- tfun(myfit, lung[1:5,]) temp2 <- survfit(myfit, lung[1:5,]) indx <- match('call', names(temp1)) #the call components won't match all.equal(unclass(temp1)[-indx], unclass(temp2)[-indx]) survival/tests/book1.R0000644000175100001440000000673312350317501014451 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # # Tests from the appendix of Therneau and Grambsch # a. Data set 1 and Breslow estimate # The data below is not in time order, to also test sorting, and has 1 NA # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) # Breslow estimate byhand1 <- function(beta, newx=0) { r <- exp(beta) loglik <- 2*beta - (log(3*r+3) + 2*log(r+3)) u <- (6 + 3*r - r^2) / ((r+1)*(r+3)) imat <- r/(r+1)^2 + 6*r/(r+3)^2 x <- c(1,1,1,0,0,0) status <- c(1,0,1,1,0,1) xbar <- c(r/(r+1), r/(r+3), 0, 0) # at times 1, 6, 8 and 9 haz <- c(1/(3*r+3), 2/(r+3), 0, 1 ) ties <- c(1,1,2,2,3,4) wt <- c(r,r,r,1,1,1) mart <- c(1,0,1,1,0,1) - wt* (cumsum(haz))[ties] #martingale residual a <- 3*(r+1)^2; b<- (r+3)^2 score <- c((2*r+3)/a, -r/a, -r/a + 3*(3-r)/b, r/a - r*(r+1)/b, r/a + 2*r/b, r/a + 2*r/b) # Schoenfeld residual scho <- c(1/(r+1), 1- (r/(3+r)), 0-(r/(3+r)) , 0) surv <- exp(-cumsum(haz)* exp(beta*newx)) varhaz.g <- cumsum(c(1/(3*r+3)^2, 2/(r+3)^2, 0, 1 )) varhaz.d <- cumsum((newx-xbar) * haz) varhaz <- (varhaz.g + varhaz.d^2/ imat) * exp(2*beta*newx) names(xbar) <- names(haz) <- 1:4 names(surv) <- names(varhaz) <- 1:4 list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=haz, mart=mart, score=score, scho=scho, surv=surv, var=varhaz, varhaz.g=varhaz.g, varhaz.d=varhaz.d) } aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) fit0 <-coxph(Surv(time, status) ~x, test1, iter=0, method='breslow') truth0 <- byhand1(0,0) aeq(truth0$loglik, fit0$loglik[1]) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid[c(2:6,1)]) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')[c(3:7,1)]) sfit <- survfit(fit0, list(x=0)) aeq(sfit$std.err^2, c(7/180, 2/9, 2/9, 11/9)) aeq(resid(fit0, 'score'), c(5/24, NA, 5/12, -1/12, 7/24, -1/24, 5/24)) fit1 <- coxph(Surv(time, status) ~x, test1, iter=1, method='breslow') aeq(fit1$coef, 8/5) # This next gives an ignorable warning message fit2 <- coxph(Surv(time, status) ~x, test1, method='breslow', iter=2) aeq(round(fit2$coef, 6), 1.472724) fit <- coxph(Surv(time, status) ~x, test1, method='breslow', eps=1e-8) aeq(round(fit$coef,7), 1.4752849) truth <- byhand1(fit$coef, 0) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid[c(2:6,1)]) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')[c(3:7,1)]) expect <- predict(fit, type='expected', newdata=test1) #force recalc aeq(test1$status[-2] -fit$resid, expect[-2]) #tests the predict function sfit <- survfit(fit, list(x=0), censor=FALSE) aeq(sfit$std.err^2, truth$var[c(1,2,4)]) # sfit skips time 8 (no events there) aeq(-log(sfit$surv), (cumsum(truth$haz))[c(1,2,4)]) sfit <- survfit(fit, list(x=0), censor=TRUE) aeq(sfit$std.err^2, truth$var) aeq(-log(sfit$surv), (cumsum(truth$haz))) # # Done with the formal test, now print out lots of bits # resid(fit) resid(fit, 'scor') resid(fit, 'scho') predict(fit, type='lp') predict(fit, type='risk') predict(fit, type='expected') predict(fit, type='terms') predict(fit, type='lp', se.fit=T) predict(fit, type='risk', se.fit=T) predict(fit, type='expected', se.fit=T) predict(fit, type='terms', se.fit=T) summary(survfit(fit)) summary(survfit(fit, list(x=2))) survival/tests/fr_kidney.R0000644000175100001440000000454011732700061015402 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # From: McGilchrist and Aisbett, Biometrics 47, 461-66, 1991 # Data on the recurrence times to infection, at the point of insertion of # the catheter, for kidney patients using portable dialysis equipment. # Catheters may be removed for reasons other than infection, in which case # the observation is censored. Each patient has exactly 2 observations. # Variables: patient, time, status, age, # sex (1=male, 2=female), # disease type (0=GN, 1=AN, 2=PKD, 3=Other) # author's estimate of the frailty aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) # I don't match their answers, and I think that I'm right kfit <- coxph(Surv(time, status)~ age + sex + disease + frailty(id), kidney) kfit1<- coxph(Surv(time, status) ~age + sex + disease + frailty(id, theta=1), kidney, iter=20) kfit0 <- coxph(Surv(time, status)~ age + sex + disease, kidney) temp <- coxph(Surv(time, status) ~age + sex + disease + frailty(id, theta=1, sparse=F), kidney) # Check out the EM based score equations # temp1 and kfit1 should have essentially the same coefficients # temp2 should equal kfit1$frail # equality won't be exact because of the different iteration paths temp1 <- coxph(Surv(time, status) ~ age + sex + disease + offset(kfit1$frail[id]), kidney) rr <- tapply(resid(temp1), kidney$id, sum) temp2 <- log(rr/1 +1) aeq(temp1$coef, kfit1$coef, tolerance=.005) aeq(temp2, kfit1$frail, tolerance=.005) kfit kfit1 kfit0 temp # # Now fit the data using REML # kfitm1 <- coxph(Surv(time,status) ~ age + sex + disease + frailty(id, dist='gauss'), kidney) kfitm2 <- coxph(Surv(time,status) ~ age + sex + disease + frailty(id, dist='gauss', sparse=F), kidney) kfitm1 summary(kfitm2) # # Fit the kidney data using AIC # # gamma, corrected aic coxph(Surv(time, status) ~ age + sex + frailty(id, method='aic', caic=T), kidney) coxph(Surv(time, status) ~ age + sex + frailty(id, dist='t'), kidney) coxph(Surv(time, status) ~ age + sex + frailty(id, dist='gauss', method='aic', caic=T), kidney) # uncorrected aic coxph(Surv(time, status) ~ age + sex + frailty(id, method='aic', caic=F), kidney) coxph(Surv(time, status) ~ age + sex + frailty(id, dist='t', caic=F), kidney) survival/tests/r_sas.R0000644000175100001440000002306712112765225014552 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Reproduce example 1 in the SAS lifereg documentation # # this fit doesn't give the same log-lik that they claim motor <- read.table('data.motor', col.names=c('temp', 'time', 'status')) fit1 <- survreg(Surv(time, status) ~ I(1000/(273.2+temp)), motor, subset=(temp>150), dist='lognormal') summary(fit1) # This one, with the loglik on the transformed scale (the inappropriate # scale, Ripley & Venables would argue) does agree. # All coefs are of course identical. fit2 <- survreg(Surv(log(time), status) ~ I(1000/(273.2+temp)), motor, subset=(temp>150), dist='gaussian') # Give the quantile estimates, which is the lower half of "output 48.1.5" # in the SAS 9.2 manual pp1 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), type='quantile', se=T) pp2 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), type='uquantile', se=T) pp1 temp130 <- matrix(0, nrow=3, ncol=6) temp130[,1] <- pp1$fit[1,] temp130[,2] <- pp1$se.fit[1,] temp130[,3] <- pp2$fit[1,] temp130[,4] <- pp2$se.fit[1,] temp130[,5] <- exp(pp2$fit[1,] - 1.64*pp2$se.fit[1,]) temp130[,6] <- exp(pp2$fit[1,] + 1.64*pp2$se.fit[1,]) dimnames(temp130) <- list(c("p=.1", "p=.2", "p=.3"), c("Time", "se(time)", "log(time)", "se[log(time)]", "lower 90", "upper 90")) print(temp130) # A set of examples, copied from the manual pages of SAS procedure # "reliability", which is part of their QC product. # color <- c("black", "red", "green", "blue", "magenta", "red4", "orange", "DarkGreen", "cyan2", "DarkViolet") palette(color) pdf(file='reliability.pdf') # # Insulating fluids example # fluid <- read.table('data.fluid', col.names=c('time', 'voltage')) # Adding a -1 to the fit just causes the each group to have it's own # intercept, rather than a global intercept + constrasts. The strata # statement allows each to have a separate scale ffit <- survreg(Surv(time) ~ voltage + strata(voltage) -1, fluid) # Get predicted quantiles at each of the voltages # By default predict() would give a line of results for each observation, # I only want the unique set of x's, i.e., only 4 cases uvolt <- sort(unique(fluid$voltage)) #the unique levels plist <- c(1, 2, 5, 1:9 *10, 95, 99)/100 pred <- predict(ffit, type='quantile', p=plist, newdata=data.frame(voltage=factor(uvolt))) tfun <- function(x) log(-log(1-x)) matplot(t(pred), tfun(plist), type='l', log='x', lty=1, col=1:4, yaxt='n') axis(2, tfun(plist), format(100*plist), adj=1) kfit <- survfit(Surv(time) ~ voltage, fluid, type='fleming') #KM fit for (i in 1:4) { temp <- kfit[i] points(temp$time, tfun(1-temp$surv), col=i, pch=i) } # Now a table temp <- array(0, dim=c(4,4,4)) #4 groups by 4 parameters by 4 stats temp[,1,1] <- ffit$coef # "EV Location" in SAS manual temp[,2,1] <- ffit$scale # "EV scale" temp[,3,1] <- exp(ffit$coef) # "Weibull Scale" temp[,4,1] <- 1/ffit$scale # "Weibull Shape" temp[,1,2] <- sqrt(diag(ffit$var))[1:4] #standard error temp[,2,2] <- sqrt(diag(ffit$var))[5:8] * ffit$scale temp[,3,2] <- temp[,1,2] * temp[,3,1] temp[,4,2] <- temp[,2,2] / (temp[,2,1])^2 temp[,1,3] <- temp[,1,1] - 1.96*temp[,1,2] #lower conf limits temp[,1,4] <- temp[,1,1] + 1.96*temp[,1,2] # upper # log(scale) is the natural parameter, in which the routine did its fitting # and on which the std errors were computed temp[,2, 3] <- exp(log(ffit$scale) - 1.96*sqrt(diag(ffit$var))[5:8]) temp[,2, 4] <- exp(log(ffit$scale) + 1.96*sqrt(diag(ffit$var))[5:8]) temp[,3, 3:4] <- exp(temp[,1,3:4]) temp[,4, 3:4] <- 1/temp[,2,4:3] dimnames(temp) <- list(uvolt, c("EV Location", "EV Scale", "Weibull scale", "Weibull shape"), c("Estimate", "SE", "lower 95% CI", "uppper 95% CI")) print(aperm(temp, c(2,3,1)), digits=5) rm(temp, uvolt, plist, pred, ffit, kfit) ##################################################################### # Turbine cracks data cracks <- read.table('data.cracks', col.names=c('time1', 'time2', 'n')) cfit <- survreg(Surv(time1, time2, type='interval2') ~1, dist='weibull', data=cracks, weight=n) summary(cfit) #Their output also has Wiebull scale = exp(cfit$coef), shape = 1/(cfit$scale) # Draw the SAS plot # The "type=fleming" argument reflects that they estimate hazards rather than # survival, and forces a Nelson-Aalen hazard estimate # plist <- c(1, 2, 5, 1:8 *10)/100 plot(qsurvreg(plist, cfit$coef, cfit$scale), tfun(plist), log='x', yaxt='n', type='l', xlab="Weibull Plot for Time", ylab="Percent") axis(2, tfun(plist), format(100*plist), adj=1) kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=cracks, weight=n, type='fleming') # Only plot point where n.event > 0 # Why? I'm trying to match them. Personally, all should be plotted. who <- (kfit$n.event > 0) points(kfit$time[who], tfun(1-kfit$surv[who]), pch='+') points(kfit$time[who], tfun(1-kfit$upper[who]), pch='-') points(kfit$time[who], tfun(1-kfit$lower[who]), pch='-') text(rep(3,6), seq(.5, -1.0, length=6), c("Scale", "Shape", "Right Censored", "Left Censored", "Interval Censored", "Fit"), adj=0) text(rep(9,6), seq(.5, -1.0, length=6), c(format(round(exp(cfit$coef), 2)), format(round(1/cfit$scale, 2)), format(tapply(cracks$n, cfit$y[,3], sum)), "ML"), adj=1) # Now a portion of his percentiles table # I don't get the same SE as SAS, I haven't checked out why. The # estimates and se for the underlying Weibull model are the same. temp <- predict(cfit, type='quantile', p=plist, se=T) tempse <- sqrt(temp$se[1,]) mat <- cbind(temp$fit[1,], tempse, temp$fit[1,] -1.96*tempse, temp$fit[1,] + 1.96*tempse) dimnames(mat) <- list(plist*100, c("Estimate", "SE", "Lower .95", "Upper .95")) print(mat) # # The cracks data has a particularly easy estimate, so use # it to double check code time <- c(cracks$time2[1], (cracks$time1 + cracks$time2)[2:8]/2, cracks$time1[9]) cdf <- cumsum(cracks$n)/sum(cracks$n) all.equal(kfit$time, time) all.equal(kfit$surv, 1-cdf[c(1:8,8)]) rm(time, cdf, kfit) ####################################################### # # Valve data # The input data has id, time, and an indicator of whether there was an # event at that time: -1=no, 1=yes. No one has an event at their last time. # Convert the data to (start, stop] form # The input data has two engines with dual failures: 328 loses 2 valves at # time 653, and number 402 loses 2 at time 139. For each, fudge the first # time to be .1 days earlier. # temp <- matrix(scan('data.valve'), byrow=T, ncol=3) n <- nrow(temp) valve <- data.frame(id=temp[,1], time1 = c(0, ifelse(diff(temp[,1])==0, temp[-n,2],0)), time2 = temp[,2], status= as.numeric(temp[,3]==1)) indx <- (1:nrow(valve))[valve$time1==valve$time2] valve$time1[indx] <- valve$time1[indx] - .1 valve$time2[indx-1] <- valve$time2[indx-1] - .1 kfit <- survfit(Surv(time1, time2, status) ~1, valve, type='fh2') plot(kfit, fun='cumhaz', ylab="Sample Mean Cumulative Failures", xlab='Time', ylim=range(-log(kfit$lower))) title("Valve replacement data") # The summary.survfit function doesn't have an option for printing out # cumulative hazards instead of survival --- need to add that # so I just reprise the central code of print.summary.survfit xx <- summary(kfit) temp <- cbind(xx$time, xx$n.risk, xx$n.event, -log(xx$surv), xx$std.err/xx$surv, -log(xx$upper), -log(xx$lower)) dimnames(temp) <- list(rep("", nrow(temp)), c("time", "n.risk", "n.event", "Cum haz", "std.err", "lower 95%", "upper 95%")) print(temp, digits=2) # Note that I have the same estimates but different SE's. We are using a # different estimator. It's a statistical argument as to which is # better (one could defend both sides): do you favor JASA or Technometrics? rm(temp, kfit, indx, xx) ###################################################### # Turbine data, lognormal fit turbine <- read.table('data.turbine', col.names=c("time1", "time2", "n")) tfit <- survreg(Surv(time1, time2, type='interval2') ~1, turbine, dist='lognormal', weights=n, subset=(n>0)) summary(tfit) # Now, do his plot, but put bootstrap confidence bands on it! # First, make a simple data set without weights tdata <- turbine[rep(1:nrow(turbine), turbine$n),] qstat <- function(data) { temp <- survreg(Surv(time1, time2, type='interval2') ~1, data=data, dist='lognormal') qsurvreg(plist, temp$coef, temp$scale, dist='lognormal') } {if (exists('bootstrap')) { bfit <- bootstrap(tdata, qstat, B=1000) bci <- limits.bca(bfit, probs=c(.025, .975)) } else { values <- matrix(0, nrow=1000, ncol=length(plist)) n <- nrow(tdata) for (i in 1:1000) { subset <- sample(1:n, n, replace=T) values[i,] <- qstat(tdata[subset,]) } bci <- t(apply(values,2, quantile, c(.05, .95))) } } xmat <- cbind(qsurvreg(plist, tfit$coef, tfit$scale, dist='lognormal'), bci) matplot(xmat, qnorm(plist), type='l', lty=c(1,2,2), col=c(1,1,1), log='x', yaxt='n', ylab='Percent', xlab='Time of Cracking (Hours x 100)') axis(2, qnorm(plist), format(100*plist), adj=1) title("Turbine Data") kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=tdata) points(kfit$time, qnorm(1-kfit$surv), pch='+') dev.off() #close the plot file survival/tests/fr_ovarian.R0000644000175100001440000000065711732700061015563 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test on the ovarian data fit1 <- coxph(Surv(futime, fustat) ~ rx + age, ovarian) fit2 <- coxph(Surv(futime, fustat) ~ rx + pspline(age, df=2), data=ovarian) fit2$iter fit2$df fit2$history fit4 <- coxph(Surv(futime, fustat) ~ rx + pspline(age, df=4), data=ovarian) fit4 survival/tests/survtest.R0000644000175100001440000000435311732700061015331 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Simple test of (start, stop] Kaplan-Meier curves, using the test2 data # set # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) fit1 <- survfit(Surv(start, stop, event) ~1, test2, type='fh2', error='tsiatis') fit2 <- survfit(Surv(start, stop, event) ~x, test2, start.time=3, type='fh2') cfit1<- survfit(coxph(Surv(start, stop, event)~1, test2)) cfit2<- survfit(coxph(Surv(start, stop, event) ~ strata(x), test2, subset=-1)) deaths <- (fit1$n.event + fit1$n.censor)>0 aeq(fit1$time[deaths], cfit1$time) aeq(fit1$n.risk[deaths], cfit1$n.risk) aeq(fit1$n.event[deaths], cfit1$n.event) aeq(fit1$surv[deaths], cfit1$surv) aeq(fit1$std.err[deaths], cfit1$std.err) deaths <- (fit2$n.event + fit2$n.censor)>0 aeq(fit2$time[deaths], cfit2$time) aeq(fit2$n.risk[deaths], cfit2$n.risk) aeq(fit2$n.event[deaths], cfit2$n.event) aeq(fit2$surv[deaths], cfit2$surv) fit3 <- survfit(Surv(start, stop, event) ~1, test2) #Kaplan-Meier aeq(fit3$n, 10) aeq(fit3$time, c(1:9,14,17)) aeq(fit3$n.risk, c(0,2,3,3,4,5,4,4,5,2,1)) aeq(fit3$n.event,c(0,1,1,0,0,1,1,1,2,0,0)) aeq(fit3$surv[fit3$n.event>0], c(.5, 1/3, 4/15, 1/5, 3/20, 9/100)) # # Verify that both surv AND n.risk are right between time points. # fit <- survfit(Surv(time, status) ~1, test1) temp <- summary(fit, time=c(.5,1, 1.5, 6, 7.5, 8, 8.9, 9, 10), extend=TRUE) aeq(temp$n.risk, c(6,6,4,4,2,2,1,1,0)) aeq(temp$surv, c(1, fit$surv[c(1,1,2,2,3,3,4,4)])) aeq(temp$n.event, c(0,1,0,2,0,0,0,1,0)) aeq(temp$std.err, c(0, (fit$surv*fit$std.err)[c(1,1,2,2,3,3,4,4)])) fit <- survfit(Surv(start, stop, event) ~1, test2) temp <- summary(fit, times=c(.5, 1.5, 2.5, 3, 6.5, 14.5, 16.5)) aeq(temp$surv, c(1, fit$surv[c(1,2,3,6, 10,10)])) aeq(temp$n.risk, c(0, 2, 3, 3, 4, 1,1)) survival/tests/mrtest.R0000644000175100001440000000164411732700061014750 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) {if (is.R()) mdy.date <- function(m, d, y) { y <- ifelse(y<100, y+1900, y) as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") } else mdy.date <- function(m,d,y) { y <- ifelse(y<100, y+1900, y) timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") } } # # A test of the match.ratetable function, specifically the # change to allow partial matching of strings # Note that 10,000 days old is 27.4 years # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) temp1 <- data.frame(year=mdy.date(2,2,1960:1964), age = 10000 + 1:5, sex = c('M', 'fema', 'f', 'ma', 'F')) temp2 <- ratetable(year=temp1$year, age=temp1$age, sex=temp1$sex) temp3 <- match.ratetable(temp2, survexp.us) aeq(temp3$R[,2], c(1,2,2,1,2)) survival/tests/r_stanford.Rout.save0000644000175100001440000000741211732700061017257 0ustar hornikusers R version 2.9.0 (2009-04-17) Copyright (C) 2009 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # The Stanford data from 1980 is used in Escobar and Meeker, Biometrics 1992. > # t5 = T5 mismatch score > # Their case numbers correspond to a data set sorted by age > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > stanford2$t5 <- ifelse(stanford2$t5 <0, NA, stanford2$t5) > stanford2 <- stanford2[order(stanford2$age, stanford2$time),] > stanford2$time <- ifelse(stanford2$time==0, .5, stanford2$time) > > cage <- stanford2$age - mean(stanford2$age) > fit1 <- survreg(Surv(time, status) ~ cage + I(cage^2), stanford2, + dist='lognormal') > fit1 Call: survreg(formula = Surv(time, status) ~ cage + I(cage^2), data = stanford2, dist = "lognormal") Coefficients: (Intercept) cage I(cage^2) 6.717591081 -0.061908619 -0.003504315 Scale= 2.362872 Loglik(model)= -863.6 Loglik(intercept only)= -868.8 Chisq= 10.5 on 2 degrees of freedom, p= 0.0053 n= 184 > ldcase <- resid(fit1, type='ldcase') > ldresp <- resid(fit1, type='ldresp') > # The ldcase and ldresp should be compared to table 1 in Escobar and > # Meeker, Biometrics 1992, p519; the colums they label as (1/2) A_{ii} > # They give data for selected cases, entered below as mdata > mdata <- cbind(c(1,2,4,5,12,16,23,61,66,72,172,182,183,184), + c(.035, .244, .141, .159, .194, .402, 0,0, .143, .403, + .178, .033, .005, .015), + c(.138, .145, .073, .076, .104, .159, 0,0, .109, .184, + .116, .063, .103, .144)) > dimnames(mdata) <- list(NULL, c("case#", "ldcase", "ldresp")) > aeq(round(ldcase[mdata[,1]],3), mdata[,2]) [1] TRUE > aeq(round(ldresp[mdata[,1]],3), mdata[,3]) [1] TRUE > > plot1 <- function() { + # make their figure 1, 2, and 6 + temp <- predict(fit1, type='quantile', p=c(.1, .5, .9)) + plot(stanford2$age, stanford2$time, log='y', xlab="Age", ylab="Days", + ylim=range(stanford2$time, temp)) + matlines(stanford2$age, temp, lty=c(1,2,2), col=1) + + n <- length(ldcase) + plot(1:n, ldcase, xlab="Case Number", ylab="(1/2) A", type='l') + title (main="Case weight pertubations") + plot(1:n, ldresp, xlab="Case Number", ylab="(1/2) A", + ylim=c(0, .2), type='l') + title(main="Response pertubations") + indx <- which(ldresp > .07) + text(indx, ldresp[indx]+ .005, indx%%10, cex=.6) + } > > postscript('meekerplot.ps') > plot1() > dev.off() null device 1 > # > # Stanford predictions in other ways > # > fit2 <- survreg(Surv(time, status) ~ poly(age,2), stanford2, + dist='lognormal') > > p1 <- predict(fit1, type='response') > p2 <- predict(fit2, type='response') > aeq(p1, p2) [1] TRUE > > p3 <- predict(fit2, type='terms', se=T) > p4 <- predict(fit2, type='lp', se=T) > p5 <- predict(fit1, type='lp', se=T) > # aeq(p3$fit + attr(p3$fit, 'constant'), p4$fit) #R is missing the attribute > aeq(p4$fit, p5$fit) [1] TRUE > aeq(p3$se.fit, p4$se.fit) #this one should be false [1] "Mean relative difference: 0.758395" > aeq(p4$se.fit, p5$se.fit) #this one true [1] TRUE > > survival/tests/clogit.R0000644000175100001440000000164211732700061014711 0ustar hornikuserslibrary(survival) # # Test of the clogit function, and indirectly of the exact option # # Data set logan has the occupation of fathers, we create a # multinomial response # nresp <- length(levels(logan$occupation)) n <- nrow(logan) indx <- rep(1:n, nresp) logan2 <- data.frame(logan[indx,], id = indx, occ2 = factor(rep(levels(logan$occupation), each=n))) logan2$y <- (logan2$occupation == logan2$occ2) #We expect two NA coefficients, so ignore the warning fit1 <- clogit(y ~ occ2 + occ2:education + occ2:race + strata(id), logan2) #since there is only one death per group, all methods are equal dummy <- rep(1, nrow(logan2)) fit2 <- coxph(Surv(dummy, y) ~ occ2 + occ2:education + occ2:race + strata(id), logan2, method='breslow') all.equal(fit1$coef, fit2$coef) all.equal(fit1$loglik, fit2$loglik) all.equal(fit1$var, fit2$var) all.equal(fit1$resid, fit2$resid) survival/tests/factor.R0000644000175100001440000000203311732700061014701 0ustar hornikusers# # Ensure that factors work in prediction # library(survival) options(na.action="na.exclude") # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) tfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog), lung) p1 <- predict(tfit, type='risk') # Testing NA handling is important too keep <- (is.na(lung$ph.ecog) | lung$ph.ecog !=1) lung2 <- lung[keep,] p2 <- predict(tfit, type='risk', newdata=lung[keep,]) aeq(p1[keep], p2) # Same, for survreg tfit <- survreg(Surv(time, status) ~ age + factor(ph.ecog), lung) p1 <- predict(tfit, type='response') p2 <- predict(tfit, type='response', newdata=lung2) aeq(p1[keep], p2) # Now repeat it tossing the missings options(na.action=na.omit) keep2 <- (lung$ph.ecog[!is.na(lung$ph.ecog)] !=1) tfit2 <- survreg(Surv(time, status) ~ age + factor(ph.ecog), lung) p3 <- predict(tfit2, type='response') p4 <- predict(tfit2, type='response', newdata=lung2, na.action=na.omit) aeq(p3[keep2] , p4) survival/tests/book7.Rout.save0000644000175100001440000000550412533657433016155 0ustar hornikusers R Under development (unstable) (2015-05-14 r68368) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) > options(na.action=na.exclude) > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # > # Tests from the appendix of Therneau and Grambsch > # Data set 1 + exact method > > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > byhand7 <- function(beta) { + r <- exp(beta) + loglik <- 2*(beta - log(3*r + 3)) + u <- 2/(r+1) + imat <- 2*r/(r+1)^2 + haz <- c(1/(3*r+3), 2/(r+3), 0, 1 ) + + ties <- c(1,1,2,2,3,4) + wt <- c(r,r,r,1,1,1) + mart <- c(1,0,1,1,0,1) - wt* (cumsum(haz))[ties] #martingale residual + + list(loglik=loglik, u=u, imat=imat, mart=mart) + } > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > fit0 <-coxph(Surv(time, status) ~x, test1, iter=0, method='exact') > truth0 <- byhand7(0) > aeq(truth0$loglik, fit0$loglik[1]) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid[c(2:6,1)]) [1] TRUE > > fit1 <- coxph(Surv(time, status) ~x, test1, iter=1, method='exact') > aeq(fit1$coef, truth0$u*fit0$var) [1] TRUE > truth1 <- byhand7(fit1$coef) > aeq(fit1$loglik[2], truth1$loglik) [1] TRUE > aeq(1/truth1$imat, fit1$var) [1] TRUE > aeq(truth1$mart, resid(fit1)[c(3:7,1)]) [1] TRUE > > # Beta is infinite for this model, so we will get a warning message > fit2 <- coxph(Surv(time, status) ~x, test1, method='exact') Warning message: In fitter(X, Y, strats, offset, init, control, weights = weights, : Loglik converged before variable 1 ; beta may be infinite. > aeq(resid(fit2)[-2], c(0, 2/3, -1/3, -4/3, 1, 0)) #values from the book [1] TRUE > > > # > # Now a multivariate case: start/stop data uses a different C routine > # > zz <- rep(0, nrow(lung)) > fit1 <- coxph(Surv(time, status) ~ age + ph.ecog + sex, lung, method="exact") > fit2 <- coxph(Surv(zz, time, status) ~ age + ph.ecog + sex, lung, + method="exact") > aeq(fit1$loglik, fit2$loglik) [1] TRUE > aeq(fit1$var, fit2$var) [1] TRUE > aeq(fit1$score, fit2$score) [1] TRUE > aeq(fit1$resid, fit2$resid) [1] TRUE > > proc.time() user system elapsed 0.712 0.028 0.737 survival/tests/r_donnell.R0000644000175100001440000000336212030334741015405 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Good initial values are key to this data set # It killed v4 of survreg; # data courtesy of Deborah Donnell, Fred Hutchinson Cancer Center # donnell <- scan("data.donnell", what=list(time1=0, time2=0, status=0)) donnell <- data.frame(donnell) dfit <- survreg(Surv(time1, time2, status, type='interval') ~1, donnell) summary(dfit) # # Fit the Donnell data using Statsci's code - should get the same coefs # if (exists('censorReg')) { dfitc <- censorReg(censor(time1, time2, status, type='interval') ~1, donnell) summary(dfitc) } # # Do a contour plot of the donnell data # npt <- 20 beta0 <- seq(.4, 3.4, length=npt) logsig <- seq(-1.4, 0.41, length=npt) donlog <- matrix(0,npt, npt) for (i in 1:npt) { for (j in 1:npt) { fit <- survreg(Surv(time1, time2, status, type='interval') ~1, donnell, init=c(beta0[i],logsig[j]), maxiter=0) donlog[i,j] <- fit$log[1] } } clev <- -c(51, 51.5, 52:60, 65, 75, 85, 100, 150) #clev <- seq(-51, -50, length=10) contour(beta0, logsig, pmax(donlog, -200), levels=clev, xlab="Intercept", ylab="Log(sigma)") points(2.39, log(.7885), pch=1, col=2) title("Donnell data") # # Compute the path of the iteration # # All the intermediate stops produce an ignorable "did not converge" # warning options(warn=-1) #turn them off niter <- 14 donpath <- matrix(0,niter+1,2) for (i in 0:niter){ fit <- survreg(Surv(time1, time2, status, type='interval') ~1, donnell, maxiter=i) donpath[i+1,] <- c(fit$coef, log(fit$scale)) } points(donpath[,1], donpath[,2]) lines(donpath[,1], donpath[,2], col=4) options(warn=0) #reset survival/tests/book2.Rout.save0000644000175100001440000001656112350326244016143 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # > # Tests from the appendix of Therneau and Grambsch > # b. Data set 1 and Efron estimate > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > byhand <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 2*beta - (log(3*r +3) + log((r+5)/2) + log(r+3)) + u <- (30 + 23*r - r^3)/ ((r+1)*(r+3)*(r+5)) + tfun <- function(x) x - x^2 + imat <- tfun(r/(r+1)) + tfun(r/(r+5)) + tfun(r/(r+3)) + + # The matrix of weights, one row per obs, one col per time + # Time of 1, 6, 6+0 (second death), and 9 + wtmat <- matrix(c(1,1,1,1,1,1, + 0,0,1,1,1,1, + 0,0,.5, .5, 1,1, + 0,0,0,0,0,1), ncol=4) + wtmat <- diag(c(r,r,r,1,1,1)) %*% wtmat + + x <- c(1,1,1,0,0,0) + status <- c(1,0,1,1,0,1) + xbar <- colSums(wtmat*x)/ colSums(wtmat) + haz <- 1/ colSums(wtmat) # one death at each of the times + + hazmat <- wtmat %*% diag(haz) #each subject's hazard over time + mart <- status - rowSums(hazmat) + + a <- r+1; b<- r+3; d<- r+5 # 'c' in the book, 'd' here + score <- c((2*r + 3)/ (3*a^2), + -r/ (3*a^2), + (675+ r*(1305 +r*(756 + r*(-4 +r*(-79 -13*r)))))/(3*(a*b*d)^2), + r*(1/(3*a^2) - a/(2*b^2) - b/(2*d^2)), + 2*r*(177 + r*(282 +r*(182 + r*(50 + 5*r)))) /(3*(a*b*d)^2), + 2*r*(177 + r*(282 +r*(182 + r*(50 + 5*r)))) /(3*(a*b*d)^2)) + + # Schoenfeld residual + d <- mean(xbar[2:3]) + scho <- c(1/(r+1), 1- d, 0- d , 0) + + surv <- exp(-cumsum(haz)* exp(beta*newx))[c(1,3,4)] + varhaz.g <- cumsum(haz^2) # since all numerators are 1 + + varhaz.d <- cumsum((newx-xbar) * haz) + + varhaz <- (varhaz.g + varhaz.d^2/ imat) * exp(2*beta*newx) + + list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=haz, + mart=mart, score=score, var.g=varhaz.g, var.d=varhaz.d, + scho=scho, surv=surv, var=varhaz[c(1,3,4)]) + } > > > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > fit0 <-coxph(Surv(time, status) ~x, test1, iter=0) > truth0 <- byhand(0,0) > aeq(truth0$loglik, fit0$loglik[1]) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid[c(2:6,1)]) [1] TRUE > aeq(resid(fit0), c(-3/4, NA, 5/6, -1/6, 5/12, 5/12, -3/4)) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')[c(3:7,1)]) [1] TRUE > sfit <- survfit(fit0, list(x=0), censor=FALSE) > aeq(sfit$std.err^2, truth0$var) [1] TRUE > aeq(sfit$surv, truth0$surv) [1] TRUE > > fit <- coxph(Surv(time, status) ~x, test1, eps=1e-8) > aeq(round(fit$coef,6), 1.676857) [1] TRUE > truth <- byhand(fit$coef, 0) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid[c(2:6,1)]) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')[c(3:7,1)]) [1] TRUE > > # Per comments in the source code, the below is expected to fail for Efron > # at the tied death times. (When predicting for new data, predict > # treats a time in the new data set that exactly matches one in the original > # as being just after the original, i.e., experiences the full hazard > # jump there, in the same way that censors do.) > expect <- predict(fit, type='expected', newdata=test1) #force recalc > use <- !(test1$time==6 | is.na(test1$status)) > aeq(test1$status[use] - resid(fit)[use], expect[use]) [1] TRUE > > sfit <- survfit(fit, list(x=0), censor=FALSE) > aeq(sfit$surv, truth$surv) [1] TRUE > aeq(sfit$std.err^2, truth$var) [1] TRUE > > # > # Done with the formal test, now print out lots of bits > # > resid(fit) 1 2 3 4 5 6 7 -0.3655434 NA 0.7191707 -0.2808293 -0.4383414 0.7310869 -0.3655434 > resid(fit, 'scor') 1 2 3 4 5 6 7 0.2208584 NA 0.1132780 -0.0442340 -0.1029199 -0.4078409 0.2208584 > resid(fit, 'scho') 1 6 6 9 0.157512 0.421244 -0.578756 0.000000 > > predict(fit, type='lp') [1] -0.8384287 NA 0.8384287 0.8384287 0.8384287 -0.8384287 -0.8384287 > predict(fit, type='risk') [1] 0.4323894 NA 2.3127302 2.3127302 2.3127302 0.4323894 0.4323894 > predict(fit, type='expected') 1 2 3 4 5 6 7 1.3655434 NA 0.2808293 0.2808293 1.4383414 0.2689131 0.3655434 > predict(fit, type='terms') x 1 -0.8384287 2 NA 3 0.8384287 4 0.8384287 5 0.8384287 6 -0.8384287 7 -0.8384287 > predict(fit, type='lp', se.fit=T) $fit 1 2 3 4 5 6 7 -0.8384287 NA 0.8384287 0.8384287 0.8384287 -0.8384287 -0.8384287 $se.fit 1 2 3 4 5 6 7 0.6388078 NA 0.6388078 0.6388078 0.6388078 0.6388078 0.6388078 > predict(fit, type='risk', se.fit=T) $fit 1 2 3 4 5 6 7 0.4323894 NA 2.3127302 2.3127302 2.3127302 0.4323894 0.4323894 $se.fit 1 2 3 4 5 6 7 0.4200565 NA 0.9714774 0.9714774 0.9714774 0.4200565 0.4200565 > predict(fit, type='expected', se.fit=T) $fit 1 2 3 4 5 6 7 1.3655434 NA 0.2808293 0.2808293 1.4383414 0.2689131 0.3655434 $se.fit [1] 1.0649293 NA 0.2864593 0.2864593 1.5922983 0.3661617 0.3661617 > predict(fit, type='terms', se.fit=T) $fit x 1 -0.8384287 2 NA 3 0.8384287 4 0.8384287 5 0.8384287 6 -0.8384287 7 -0.8384287 $se.fit x 1 0.6388078 2 NA 3 0.6388078 4 0.6388078 5 0.6388078 6 0.6388078 7 0.6388078 > > summary(survfit(fit)) Call: survfit(formula = fit) time n.risk n.event survival std.err lower 95% CI upper 95% CI 1 6 1 0.8857 0.117 0.683036 1 6 4 2 0.4294 0.237 0.145743 1 9 1 1 0.0425 0.116 0.000198 1 > summary(survfit(fit, list(x=2))) Call: survfit(formula = fit, newdata = list(x = 2)) time n.risk n.event survival std.err lower 95% CI upper 95% CI 1 6 1 2.23e-01 5.97e-01 1.16e-03 1 6 4 2 2.87e-05 5.69e-04 3.96e-22 1 9 1 1 1.08e-17 1.04e-15 1.07e-99 1 > > proc.time() user system elapsed 0.236 0.024 0.256 survival/tests/testreg.Rout.save0000644000175100001440000002107011732700061016567 0ustar hornikusers R version 2.14.0 Under development (unstable) (2011-04-10 r55401) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) #preserve length of missings > library(survival) Loading required package: splines > > # > # Run a test that can be verified using other packages (we used SAS) > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > fit1w <- survreg(Surv(time, status) ~x, test1, dist='weibull') > fit1w Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "weibull") Coefficients: (Intercept) x 2.2373335 -0.7442249 Scale= 0.4563163 Loglik(model)= -10.3 Loglik(intercept only)= -11.4 Chisq= 2.22 on 1 degrees of freedom, p= 0.14 n=6 (1 observation deleted due to missingness) > summary(fit1w) Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "weibull") Value Std. Error z p (Intercept) 2.237 0.330 6.78 1.18e-11 x -0.744 0.486 -1.53 1.26e-01 Log(scale) -0.785 0.433 -1.81 6.99e-02 Scale= 0.456 Weibull distribution Loglik(model)= -10.3 Loglik(intercept only)= -11.4 Chisq= 2.22 on 1 degrees of freedom, p= 0.14 Number of Newton-Raphson Iterations: 8 n=6 (1 observation deleted due to missingness) > > fit1e <- survreg(Surv(time, status) ~x, test1, dist='exponential') > fit1e Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "exponential") Coefficients: (Intercept) x 2.442347 -1.056053 Scale fixed at 1 Loglik(model)= -11.7 Loglik(intercept only)= -12.2 Chisq= 1.07 on 1 degrees of freedom, p= 0.3 n=6 (1 observation deleted due to missingness) > summary(fit1e) Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "exponential") Value Std. Error z p (Intercept) 2.44 0.707 3.45 0.000552 x -1.06 1.000 -1.06 0.290944 Scale fixed at 1 Exponential distribution Loglik(model)= -11.7 Loglik(intercept only)= -12.2 Chisq= 1.07 on 1 degrees of freedom, p= 0.3 Number of Newton-Raphson Iterations: 4 n=6 (1 observation deleted due to missingness) > > fit1l <- survreg(Surv(time, status) ~x, test1, dist='loglogistic') > fit1l Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "loglogistic") Coefficients: (Intercept) x 2.177208 -1.195672 Scale= 0.3847582 Loglik(model)= -10.7 Loglik(intercept only)= -12 Chisq= 2.7 on 1 degrees of freedom, p= 0.1 n=6 (1 observation deleted due to missingness) > summary(fit1l) Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "loglogistic") Value Std. Error z p (Intercept) 2.177 0.365 5.96 2.48e-09 x -1.196 0.711 -1.68 9.25e-02 Log(scale) -0.955 0.396 -2.41 1.58e-02 Scale= 0.385 Log logistic distribution Loglik(model)= -10.7 Loglik(intercept only)= -12 Chisq= 2.7 on 1 degrees of freedom, p= 0.1 Number of Newton-Raphson Iterations: 4 n=6 (1 observation deleted due to missingness) > > fit1g <- survreg(Surv(time, status) ~x, test1, dist='lognormal') > summary(fit1g) Call: survreg(formula = Surv(time, status) ~ x, data = test1, dist = "lognormal") Value Std. Error z p (Intercept) 2.210 0.404 5.48 4.35e-08 x -1.268 0.585 -2.17 3.03e-02 Log(scale) -0.446 0.342 -1.30 1.93e-01 Scale= 0.64 Log Normal distribution Loglik(model)= -10.5 Loglik(intercept only)= -12.1 Chisq= 3.26 on 1 degrees of freedom, p= 0.071 Number of Newton-Raphson Iterations: 5 n=6 (1 observation deleted due to missingness) > # > # Do a test with the ovarian data > # > fitfw <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, + dist='weibull') > fitfw Call: survreg(formula = Surv(futime, fustat) ~ age + ecog.ps, data = ovarian, dist = "weibull") Coefficients: (Intercept) age ecog.ps 12.28496723 -0.09702669 0.09977342 Scale= 0.6032744 Loglik(model)= -90 Loglik(intercept only)= -98 Chisq= 15.98 on 2 degrees of freedom, p= 0.00034 n= 26 > > fitfl <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, + dist='loglogistic') > fitfl Call: survreg(formula = Surv(futime, fustat) ~ age + ecog.ps, data = ovarian, dist = "loglogistic") Coefficients: (Intercept) age ecog.ps 11.50853384 -0.08876814 0.09033348 Scale= 0.4464064 Loglik(model)= -89.5 Loglik(intercept only)= -97.4 Chisq= 15.67 on 2 degrees of freedom, p= 4e-04 n= 26 > > #test out interval censoring, using some dummy time values > > idat <- read.table('data.interval', skip=3, header=T, sep=',') > flsurv<- Surv(idat$ltime, idat$rtime, type='interval2') > > fitfw2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='weibull') > summary(fitfw2) Call: survreg(formula = flsurv ~ age + ecog.ps, data = idat, dist = "weibull") Value Std. Error z p (Intercept) 12.3886 1.6027 7.730 1.08e-14 age -0.0986 0.0254 -3.885 1.02e-04 ecog.ps 0.0971 0.3776 0.257 7.97e-01 Log(scale) -0.4773 0.2583 -1.848 6.47e-02 Scale= 0.62 Weibull distribution Loglik(model)= -56.2 Loglik(intercept only)= -64 Chisq= 15.57 on 2 degrees of freedom, p= 0.00042 Number of Newton-Raphson Iterations: 6 n= 26 > > fitfl2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='loglogistic') > summary(fitfl2) Call: survreg(formula = flsurv ~ age + ecog.ps, data = idat, dist = "loglogistic") Value Std. Error z p (Intercept) 11.5268 1.528 7.542 4.62e-14 age -0.0888 0.024 -3.703 2.13e-04 ecog.ps 0.0818 0.364 0.225 8.22e-01 Log(scale) -0.8023 0.271 -2.965 3.03e-03 Scale= 0.448 Log logistic distribution Loglik(model)= -55.9 Loglik(intercept only)= -63.5 Chisq= 15.35 on 2 degrees of freedom, p= 0.00046 Number of Newton-Raphson Iterations: 5 n= 26 > > fitfg2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='lognormal') > summary(fitfg2) Call: survreg(formula = flsurv ~ age + ecog.ps, data = idat, dist = "lognormal") Value Std. Error z p (Intercept) 11.1548 1.4347 7.775 7.56e-15 age -0.0855 0.0238 -3.598 3.20e-04 ecog.ps 0.2066 0.3828 0.540 5.89e-01 Log(scale) -0.2297 0.2508 -0.916 3.60e-01 Scale= 0.795 Log Normal distribution Loglik(model)= -56 Loglik(intercept only)= -63.5 Chisq= 14.94 on 2 degrees of freedom, p= 0.00057 Number of Newton-Raphson Iterations: 5 n= 26 > > logt <- c(survreg.distributions$t, + survreg.distributions$weibull[c('trans', 'itrans', 'dtrans')]) > logt$name <- 'log(t)' > > fitft2 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, + idat, dist=logt, parm=100) > summary(fitft2) #should be quite close to fitfg2 Call: survreg(formula = Surv(ltime, rtime, type = "interval2") ~ age + ecog.ps, data = idat, dist = logt, parms = 100) Value Std. Error z p (Intercept) 11.1856 1.4419 7.758 8.66e-15 age -0.0858 0.0238 -3.609 3.07e-04 ecog.ps 0.1978 0.3814 0.519 6.04e-01 Log(scale) -0.2394 0.2522 -0.949 3.43e-01 Scale= 0.787 log(t) distribution: parmameters= 100 Loglik(model)= -56 Loglik(intercept only)= -63.5 Chisq= 14.97 on 2 degrees of freedom, p= 0.00056 Number of Newton-Raphson Iterations: 5 n= 26 > > # > # Check out the survreg density and probability functions > # > > # Gaussian > x <- -10:10 > p <- seq(.1, .95, length=25) > all.equal(dsurvreg(x, 1, 5, 'gaussian'), dnorm(x, 1, 5)) [1] TRUE > all.equal(psurvreg(x, 1, 5, 'gaussian'), pnorm(x, 1, 5)) [1] TRUE > all.equal(qsurvreg(p, 1, 5, 'gaussian'), qnorm(p, 1, 5)) [1] TRUE > > # Lognormal > x <- 1:10 > all.equal(dsurvreg(x, 1, 5, 'lognormal'), dlnorm(x, 1, 5)) [1] TRUE > all.equal(psurvreg(x, 1, 5, 'lognormal'), plnorm(x, 1, 5)) [1] TRUE > all.equal(qsurvreg(p, 1, 5, 'lognormal'), qlnorm(p, 1, 5)) [1] TRUE > > # Weibull > lambda <- exp(-2) > rho <- 1/3 > temp <- (lambda*x)^rho > all.equal(psurvreg(x, 2, 3), 1- exp(-temp)) [1] TRUE > all.equal(dsurvreg(x, 2, 3), lambda*rho*(lambda*x)^(rho-1)*exp(-temp)) [1] TRUE > survival/tests/doaml.R0000644000175100001440000000404511732700061014524 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) # # These results can be found in Miller # fit <- coxph(Surv(aml$time, aml$status) ~ aml$x, method='breslow') fit resid(fit, type='mart') resid(fit, type='score') resid(fit, type='scho') # Test the drop of an itercept: should have no effect fit2 <- coxph(Surv(time, status) ~ x -1, method='breslow', data=aml) aeq(fit$loglik, fit2$loglik) aeq(coef(fit), coef(fit2)) aeq(fit$var, fit2$var) fit <- survfit(Surv(aml$time, aml$status) ~ aml$x) fit summary(fit) survdiff(Surv(aml$time, aml$status)~ aml$x) # # Test out the weighted K-M # # First, equal case weights- shouldn't change the survival, but will # halve the variance temp2 <-survfit(Surv(aml$time, aml$status)~1, type='kaplan', weight=rep(2,23)) temp <-survfit(Surv(time, status)~1, aml) aeq(temp$surv, temp2$surv) aeq(temp$std.err^2, 2*temp2$std.err^2) # Risk weights-- use a null Cox model tfit <- coxph(Surv(aml$time, aml$status) ~ offset(log(1:23))) sfit <- survfit(tfit, type='aalen', censor=FALSE) # Now compute it by hand. The survfit program will produce a curve # corresponding to the mean offset. This is a change on 7/2010, # which caused S(new) = S(old)^exp(mean(log(1:23))). # Ties are a nuisance rscore <- exp(log(1:23) - mean(log(1:23)))[order(aml$time)] atime <- sort(aml$time) denom <- rev(cumsum(rev(rscore))) denom <- denom[match(unique(atime), atime)] deaths <- tapply(aml$status, aml$time, sum) chaz <- cumsum(deaths/denom) all.equal(sfit$surv, as.vector(exp(-chaz[deaths>0]))) cvar <- cumsum(deaths/denom^2) all.equal(sfit$std^2, as.vector(cvar[deaths>0])) # And the Efron result summary(survfit(tfit)) # Lots of ties, so its a good test case x1 <- coxph(Surv(time, status)~x, aml, method='efron') x1 x2 <- coxph(Surv(rep(0,23),time, status) ~x, aml, method='efron') aeq(x1$coef, x2$coef) rm(x1, x2, atime, denom, deaths, chaz,cvar, tfit, sfit, temp, temp2, fit) survival/tests/coxsurv2.R0000644000175100001440000000414711732700061015226 0ustar hornikuserslibrary(survival) # # Check that the survival curves from a Cox model with beta=0 # match ordinary survival # # Aalen surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fleming', error='tsiatis') fit1 <- coxph(Surv(time, status) ~ age + strata(sex), data=lung, iter=0, method='breslow') fit1$var <- 0*fit1$var #sneaky, causes the extra term in the Cox variance # calculation to be zero surv2 <- survfit(fit1, type='aalen', vartype='tsiatis') surv3 <- survfit(fit1) arglist <- c('n', 'time', 'n.risk','n.event', 'n.censor', 'surv', 'strata', 'std.err', 'upper', 'lower') all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) all.equal(unclass(surv1)[arglist], unclass(surv3)[arglist]) # Efron method surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fh2', error='tsiatis') surv2 <- survfit(fit1, type='efron', vartype='efron') all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) # Kaplan-Meier surv1 <- survfit(Surv(time,status) ~ sex, data=lung) surv2 <- survfit(fit1, type='kalb', vartype='green') all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) # Now add some random weights rwt <- runif(nrow(lung), .5, 3) surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fleming', error='tsiatis', weight=rwt) fit1 <- coxph(Surv(time, status) ~ age + strata(sex), data=lung, iter=0, method='breslow', weight=rwt) fit1$var <- 0*fit1$var #sneaky surv2 <- survfit(fit1, type='aalen', vartype='tsiatis') surv3 <- survfit(fit1) all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) all.equal(unclass(surv1)[arglist], unclass(surv3)[arglist]) # Efron method surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fh2', error='tsiatis', weight=rwt) surv2 <- survfit(fit1, type='efron', vartype='efron') all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) # Kaplan-Meier surv1 <- survfit(Surv(time,status) ~ sex, data=lung, weight=rwt) surv2 <- survfit(fit1, type='kalb', vartype='green') all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) survival/tests/r_donnell.Rout.save0000644000175100001440000000573712055210747017110 0ustar hornikusers R version 2.15.2 (2012-10-26) -- "Trick or Treat" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Good initial values are key to this data set > # It killed v4 of survreg; > # data courtesy of Deborah Donnell, Fred Hutchinson Cancer Center > # > > donnell <- scan("data.donnell", what=list(time1=0, time2=0, status=0)) Read 210 records > donnell <- data.frame(donnell) > > dfit <- survreg(Surv(time1, time2, status, type='interval') ~1, donnell) > summary(dfit) Call: survreg(formula = Surv(time1, time2, status, type = "interval") ~ 1, data = donnell) Value Std. Error z p (Intercept) 2.390 0.804 2.972 0.00295 Log(scale) -0.237 0.346 -0.687 0.49232 Scale= 0.789 Weibull distribution Loglik(model)= -51 Loglik(intercept only)= -51 Number of Newton-Raphson Iterations: 11 n= 210 > > # > # Fit the Donnell data using Statsci's code - should get the same coefs > # > if (exists('censorReg')) { + dfitc <- censorReg(censor(time1, time2, status, type='interval') ~1, + donnell) + summary(dfitc) + } > # > # Do a contour plot of the donnell data > # > npt <- 20 > beta0 <- seq(.4, 3.4, length=npt) > logsig <- seq(-1.4, 0.41, length=npt) > donlog <- matrix(0,npt, npt) > > for (i in 1:npt) { + for (j in 1:npt) { + fit <- survreg(Surv(time1, time2, status, type='interval') ~1, + donnell, init=c(beta0[i],logsig[j]), + maxiter=0) + donlog[i,j] <- fit$log[1] + } + } > > clev <- -c(51, 51.5, 52:60, 65, 75, 85, 100, 150) > #clev <- seq(-51, -50, length=10) > > contour(beta0, logsig, pmax(donlog, -200), levels=clev, xlab="Intercept", + ylab="Log(sigma)") > points(2.39, log(.7885), pch=1, col=2) > title("Donnell data") > # > # Compute the path of the iteration > # > # All the intermediate stops produce an ignorable "did not converge" > # warning > options(warn=-1) #turn them off > niter <- 14 > donpath <- matrix(0,niter+1,2) > for (i in 0:niter){ + fit <- survreg(Surv(time1, time2, status, type='interval') ~1, + donnell, maxiter=i) + donpath[i+1,] <- c(fit$coef, log(fit$scale)) + } > points(donpath[,1], donpath[,2]) > lines(donpath[,1], donpath[,2], col=4) > options(warn=0) #reset > > proc.time() user system elapsed 1.196 0.020 1.209 survival/tests/prednew.Rout.save0000644000175100001440000000720112113162124016552 0ustar hornikusers R Under development (unstable) (2013-02-24 r62054) -- "Unsuffered Consequences" Copyright (C) 2013 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > # Make sure that the newdata argument works for various > # predictions > # We purposely use a subset of the lung data that has only some > # of the levels of the ph.ecog > library(survival) Loading required package: splines > options(na.action=na.exclude, contrasts=c('contr.treatment', 'contr.poly')) > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > myfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog) + strata(sex), lung) > > keep <- which(lung$inst<13 & (lung$ph.ecog==1 | lung$ph.ecog==2)) > p1 <- predict(myfit, type='lp') > p2 <- predict(myfit, type="lp", newdata=lung[keep,]) > p3 <- predict(myfit, type='lp', se.fit=TRUE) > p4 <- predict(myfit, type="lp", newdata=lung[keep,], se.fit=TRUE) > aeq(p1[keep], p2) [1] TRUE > aeq(p1, p3$fit) [1] TRUE > aeq(p1[keep], p4$fit) [1] TRUE > aeq(p3$se.fit[keep], p4$se.fit) [1] TRUE > > p1 <- predict(myfit, type='risk') > p2 <- predict(myfit, type="risk", newdata=lung[keep,]) > p3 <- predict(myfit, type='risk', se.fit=TRUE) > p4 <- predict(myfit, type="risk", newdata=lung[keep,], se.fit=TRUE) > aeq(p1[keep], p2) [1] TRUE > aeq(p1, p3$fit) [1] TRUE > aeq(p1[keep], p4$fit) [1] TRUE > aeq(p3$se.fit[keep], p4$se.fit) [1] TRUE > > # The all.equal fails for type=expected, Efron approx, and tied death > # times due to use of an approximation. See comments in the source code. > myfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog) + strata(sex), + data=lung, method='breslow') > p1 <- predict(myfit, type='expected') > p2 <- predict(myfit, type="expected", newdata=lung[keep,]) > p3 <- predict(myfit, type='expected', se.fit=TRUE) > p4 <- predict(myfit, type="expected", newdata=lung[keep,], se.fit=TRUE) > aeq(p1[keep], p2) [1] TRUE > aeq(p1, p3$fit) [1] TRUE > aeq(p1[keep], p4$fit) [1] TRUE > aeq(p3$se.fit[keep], p4$se.fit) [1] TRUE > > p1 <- predict(myfit, type='terms') > p2 <- predict(myfit, type="terms",newdata=lung[keep,]) > p3 <- predict(myfit, type='terms', se.fit=T) > p4 <- predict(myfit, type="terms",newdata=lung[keep,], se.fit=T) > aeq(p1[keep,], p2) [1] TRUE > aeq(p1, p3$fit) [1] TRUE > aeq(p1[keep,], p4$fit) [1] TRUE > aeq(p3$se.fit[keep,], p4$se.fit) [1] TRUE > > # > # Check out the logic whereby predict does not need to > # recover the model frame. The first call should not > # need to do so, the second should in each case. > # > myfit <- coxph(Surv(time, status) ~ age + factor(sex), lung, x=T) > p1 <- predict(myfit, type='risk', se=T) > myfit2 <- coxph(Surv(time, status) ~ age + factor(sex), lung) > p2 <- predict(myfit2, type='risk', se=T) > aeq(p1$fit, p2$fit) [1] TRUE > aeq(p1$se, p2$se) [1] TRUE > > p1 <- predict(myfit, type='expected', se=T) > p2 <- predict(myfit2, type='expected', se=T) > aeq(p1$fit, p2$fit) [1] TRUE > aeq(p1$se.fit, p2$se.fit) [1] TRUE > > p1 <- predict(myfit, type='terms', se=T) > p2 <- predict(myfit2, type='terms', se=T) > aeq(p1$fit, p2$fit) [1] TRUE > aeq(p1$se.fit, p2$se.fit) [1] TRUE > > proc.time() user system elapsed 0.380 0.016 0.394 survival/tests/r_user.R0000644000175100001440000000152511732700061014727 0ustar hornikusersoptions(na.action=na.exclude) #preserve length of missings library(survival) # # Check out using a "user specified" distribution # mydist <- c(survreg.distributions$extreme, survreg.distributions$weibull[-1]) mydist$name <- "Weibull2" mydist$dist <- NULL fit1 <- survreg(Surv(time, status) ~ age + ph.ecog, lung) fit2 <- survreg(Surv(time, status) ~ age + ph.ecog, lung, dist=mydist) all.equal(fit1$coef, fit2$coef) all.equal(fit1$var, fit2$var) # # And with an data set containing interval censoring # idat <- read.table('data.interval', skip=3, header=T, sep=',') fit1 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, idat) fit2 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, data=idat, dist=mydist) all.equal(fit1$coef, fit2$coef) all.equal(fit1$var, fit2$var) all.equal(fit1$log, fit2$log) survival/tests/ratetable.Rout.save0000644000175100001440000001256711732700061017070 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Generate each of the messages from is.ratetable > # > {if (is.R()) mdy.date <- function(m, d, y) { + y <- ifelse(y<100, y+1900, y) + as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") + } + else mdy.date <- function(m,d,y) { + y <- ifelse(y<100, y+1900, y) + timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") + } + } > > temp <- runif(21*2*4) > > # Good > attributes(temp) <- list(dim=c(21,2,4), + dimnames=list(c(as.character(75:95)), c("male","female"), + c(as.character(2000:2003))), + dimid=c("age","sex","year"), + type=c(2,1,4), + cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:3)*366.25), + class='ratetable') > is.ratetable(temp) [1] TRUE > > # Factor problem + cutpoints length > attributes(temp) <- list(dim=c(21,2,4), + dimnames=list(c(as.character(75:95)), c("male","female"), + c(as.character(2000:2003))), + dimid=c("age","sex","year"), + type=c(1,1,2), + cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:4)*366.25), + class='ratetable') > is.ratetable(temp, verbose=T) [1] "type[ 1 ] is 1; cutpoint should be null" [2] "wrong length for cutpoints 3" > > > # missing dimid attribute + unsorted cutpoint > attributes(temp) <- list(dim=c(21,2,4), + dimnames=list(c(as.character(75:95)), c("male","female"), + c(as.character(2000:2003))), + type=c(2,1,3), + cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(4:1)*366.25), + class='ratetable') > is.ratetable(temp, verbose=T) [1] "missing attribute: dimid" "wrong length for dimid" [3] "unsorted cutpoints for dimension 3" > > # wrong length for dimid and type, illegal type > attributes(temp) <- list(dim=c(21,2,4), + dimnames=list(c(as.character(75:95)), c("male","female"), + c(as.character(2000:2003))), + dimid=c("age","sex","year", "zed"), + type=c(2,1,3,6), + cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:3)*366.25), + class='ratetable') > is.ratetable(temp, verbose=T) [1] "wrong length for dimid" [2] "type attribute must be 1, 2, 3, or 4" [3] "wrong length for type attribute" > > > # Print and summary > print(survexp.us[1:30,,c('1953', '1985')] ) Rate table with dimension(s): age sex year , , 1953 male female 0-1d 1.157372e-02 8.844000e-03 1-7d 1.446302e-03 1.027012e-03 7-28d 1.379175e-04 1.106070e-04 28-365d 2.814865e-05 2.346732e-05 1 6.169963e-06 5.423669e-06 2 3.860391e-06 3.161334e-06 3 2.909162e-06 2.424089e-06 4 2.448747e-06 1.950051e-06 5 2.210350e-06 1.692520e-06 6 1.988411e-06 1.481583e-06 7 1.813065e-06 1.298053e-06 8 1.684303e-06 1.169315e-06 9 1.593900e-06 1.087146e-06 10 1.569249e-06 1.051541e-06 11 1.626780e-06 1.043325e-06 12 1.771975e-06 1.089887e-06 13 2.062389e-06 1.199447e-06 14 2.462443e-06 1.347361e-06 15 2.944779e-06 1.550072e-06 16 3.410754e-06 1.752797e-06 17 3.819231e-06 1.928140e-06 18 4.164702e-06 2.056914e-06 19 4.504735e-06 2.169256e-06 20 4.822866e-06 2.289823e-06 21 5.086174e-06 2.410395e-06 22 5.278187e-06 2.511789e-06 23 5.335795e-06 2.613186e-06 24 5.286423e-06 2.714587e-06 25 5.198648e-06 2.815992e-06 26 5.130079e-06 2.917400e-06 , , 1985 male female 0-1d 4.429985e-03 3.701977e-03 1-7d 3.595869e-04 2.735770e-04 7-28d 6.385309e-05 5.193376e-05 28-365d 1.277308e-05 9.947467e-06 1 2.451492e-06 2.108968e-06 2 1.739100e-06 1.341882e-06 3 1.369277e-06 1.013196e-06 4 1.122754e-06 7.940941e-07 5 9.995021e-07 7.530142e-07 6 9.173378e-07 6.571643e-07 7 8.488687e-07 5.887021e-07 8 7.530153e-07 5.339338e-07 9 6.297793e-07 4.791661e-07 10 5.202416e-07 4.517830e-07 11 5.202416e-07 4.517830e-07 12 7.530134e-07 5.202412e-07 13 1.232311e-06 6.571636e-07 14 1.862374e-06 8.351727e-07 15 2.533686e-06 1.026887e-06 16 3.150341e-06 1.204921e-06 17 3.657474e-06 1.341877e-06 18 4.041315e-06 1.424054e-06 19 4.315527e-06 1.465144e-06 20 4.603481e-06 1.506233e-06 21 4.864041e-06 1.561021e-06 22 5.069759e-06 1.615810e-06 23 5.138331e-06 1.643205e-06 24 5.152035e-06 1.670600e-06 25 5.110881e-06 1.697995e-06 26 5.097158e-06 1.725391e-06 > summary(survexp.usr) Rate table with 4 dimensions: age ranges from 0 to 39812.25; with 113 categories sex has levels of: male female race has levels of: white black year ranges from 1940-01-01 to 2004-01-01; with 65 categories > survival/tests/book2.R0000644000175100001440000000755112350317115014452 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # # Tests from the appendix of Therneau and Grambsch # b. Data set 1 and Efron estimate # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) byhand <- function(beta, newx=0) { r <- exp(beta) loglik <- 2*beta - (log(3*r +3) + log((r+5)/2) + log(r+3)) u <- (30 + 23*r - r^3)/ ((r+1)*(r+3)*(r+5)) tfun <- function(x) x - x^2 imat <- tfun(r/(r+1)) + tfun(r/(r+5)) + tfun(r/(r+3)) # The matrix of weights, one row per obs, one col per time # Time of 1, 6, 6+0 (second death), and 9 wtmat <- matrix(c(1,1,1,1,1,1, 0,0,1,1,1,1, 0,0,.5, .5, 1,1, 0,0,0,0,0,1), ncol=4) wtmat <- diag(c(r,r,r,1,1,1)) %*% wtmat x <- c(1,1,1,0,0,0) status <- c(1,0,1,1,0,1) xbar <- colSums(wtmat*x)/ colSums(wtmat) haz <- 1/ colSums(wtmat) # one death at each of the times hazmat <- wtmat %*% diag(haz) #each subject's hazard over time mart <- status - rowSums(hazmat) a <- r+1; b<- r+3; d<- r+5 # 'c' in the book, 'd' here score <- c((2*r + 3)/ (3*a^2), -r/ (3*a^2), (675+ r*(1305 +r*(756 + r*(-4 +r*(-79 -13*r)))))/(3*(a*b*d)^2), r*(1/(3*a^2) - a/(2*b^2) - b/(2*d^2)), 2*r*(177 + r*(282 +r*(182 + r*(50 + 5*r)))) /(3*(a*b*d)^2), 2*r*(177 + r*(282 +r*(182 + r*(50 + 5*r)))) /(3*(a*b*d)^2)) # Schoenfeld residual d <- mean(xbar[2:3]) scho <- c(1/(r+1), 1- d, 0- d , 0) surv <- exp(-cumsum(haz)* exp(beta*newx))[c(1,3,4)] varhaz.g <- cumsum(haz^2) # since all numerators are 1 varhaz.d <- cumsum((newx-xbar) * haz) varhaz <- (varhaz.g + varhaz.d^2/ imat) * exp(2*beta*newx) list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=haz, mart=mart, score=score, var.g=varhaz.g, var.d=varhaz.d, scho=scho, surv=surv, var=varhaz[c(1,3,4)]) } aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) fit0 <-coxph(Surv(time, status) ~x, test1, iter=0) truth0 <- byhand(0,0) aeq(truth0$loglik, fit0$loglik[1]) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid[c(2:6,1)]) aeq(resid(fit0), c(-3/4, NA, 5/6, -1/6, 5/12, 5/12, -3/4)) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')[c(3:7,1)]) sfit <- survfit(fit0, list(x=0), censor=FALSE) aeq(sfit$std.err^2, truth0$var) aeq(sfit$surv, truth0$surv) fit <- coxph(Surv(time, status) ~x, test1, eps=1e-8) aeq(round(fit$coef,6), 1.676857) truth <- byhand(fit$coef, 0) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid[c(2:6,1)]) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')[c(3:7,1)]) # Per comments in the source code, the below is expected to fail for Efron # at the tied death times. (When predicting for new data, predict # treats a time in the new data set that exactly matches one in the original # as being just after the original, i.e., experiences the full hazard # jump there, in the same way that censors do.) expect <- predict(fit, type='expected', newdata=test1) #force recalc use <- !(test1$time==6 | is.na(test1$status)) aeq(test1$status[use] - resid(fit)[use], expect[use]) sfit <- survfit(fit, list(x=0), censor=FALSE) aeq(sfit$surv, truth$surv) aeq(sfit$std.err^2, truth$var) # # Done with the formal test, now print out lots of bits # resid(fit) resid(fit, 'scor') resid(fit, 'scho') predict(fit, type='lp') predict(fit, type='risk') predict(fit, type='expected') predict(fit, type='terms') predict(fit, type='lp', se.fit=T) predict(fit, type='risk', se.fit=T) predict(fit, type='expected', se.fit=T) predict(fit, type='terms', se.fit=T) summary(survfit(fit)) summary(survfit(fit, list(x=2))) survival/tests/r_scale.R0000644000175100001440000000325512141742354015050 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Verify that scale can be fixed at a value # coefs will differ slightly due to different iteration paths tol <- .001 # Intercept only models fit1 <- survreg(Surv(time,status) ~ 1, lung) fit2 <- survreg(Surv(time,status) ~ 1, lung, scale=fit1$scale) all.equal(fit1$coef, fit2$coef, tolerance= tol) all.equal(fit1$loglik, fit2$loglik, tolerance= tol) # The two robust variance matrices are not the same, since removing # an obs has a different effect on the two models. This just # checks for failure, not for correctness fit3 <- survreg(Surv(time,status) ~ 1, lung, robust=TRUE) fit4 <- survreg(Surv(time,status) ~ 1, lung, scale=fit1$scale, robust=TRUE) # multiple covariates fit1 <- survreg(Surv(time,status) ~ age + ph.karno, lung) fit2 <- survreg(Surv(time,status) ~ age + ph.karno, lung, scale=fit1$scale) all.equal(fit1$coef, fit2$coef, tolerance=tol) all.equal(fit1$loglik[2], fit2$loglik[2], tolerance=tol) fit3 <- survreg(Surv(time,status) ~ age + ph.karno, lung, robust=TRUE) fit4 <- survreg(Surv(time,status) ~ age + ph.karno, lung, scale=fit1$scale, robust=TRUE) # penalized models fit1 <- survreg(Surv(time, status) ~ pspline(age), lung) fit2 <- survreg(Surv(time, status) ~ pspline(age), lung, scale=fit1$scale) all.equal(fit1$coef, fit2$coef, tolerance=tol) all.equal(fit1$loglik[2], fit2$loglik[2], tolerance=tol) fit3 <- survreg(Surv(time,status) ~ pspline(age) + ph.karno, lung, robust=TRUE) fit4 <- survreg(Surv(time,status) ~ pspline(age) + ph.karno, lung, scale=fit1$scale, robust=TRUE) survival/tests/tt.Rout.save0000644000175100001440000000465312466142446015565 0ustar hornikusers R Under development (unstable) (2014-09-01 r66509) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) > > # A contrived example for the tt function > # > mkdata <- function(n, beta) { + age <- runif(n, 20, 60) + x <- rbinom(n, 1, .5) + + futime <- rep(40, n) # everyone has 40 years of follow-up + status <- rep(0, n) + dtime <- runif(n/2, 1, 40) # 1/2 of them die + dtime <- sort(dtime) + + # The risk is set to beta[1]*x + beta[2]* f(current_age) + # where f= 0 up to age 40, rises linear to age 70, flat after that + for (i in 1:length(dtime)) { + atrisk <- (futime >= dtime[i]) + c.age <- age + dtime + age2 <- pmin(30, pmax(0, c.age-40)) + xbeta <- beta[1]*x + beta[2]*age2 + + # Select a death according to risk + risk <- ifelse(atrisk, exp(xbeta), 0) + dead <- sample(1:n, 1, prob=risk/sum(risk)) + + futime[dead] <- dtime[i] + status[dead] <- 1 + } + data.frame(futime=futime, status=status, age=age, x=x, risk=risk) + } > tdata <- mkdata(500, c(log(1.5), 2/30)) > > fit1 <- coxph(Surv(futime, status) ~ x + pspline(age), tdata) > fit2 <- coxph(Surv(futime, status) ~ x + tt(age), tdata, + tt= function(x, t, ...) pspline(x+t)) > > dfit <- coxph(Surv(futime, status) ~ x + tt(age), tdata, + tt= function(x, t, ...) x+t, iter=0, x=T) > > # > # Check that cluster, weight, and offset were correctly expanded > # > tdata <- data.frame(tdata, grp=sample(1:100, 500, replace=TRUE), + casewt = sample(1:5, 500, replace=TRUE), + zz = rnorm(500)) > dfit2 <- coxph(Surv(futime, status) ~ x + tt(age) + offset(zz) + cluster(grp), + weight=casewt, data=tdata, + tt= function(x, t, ...) x+t) > > proc.time() user system elapsed 10.192 1.060 11.287 survival/tests/ovarian.Rout.save0000644000175100001440000003111712536400776016571 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Test the coxph program on the Ovarian data > # > > attach(ovarian) > > summary(survfit(Surv(futime, fustat)~1), censor=TRUE) Call: survfit(formula = Surv(futime, fustat) ~ 1) time n.risk n.event survival std.err lower 95% CI upper 95% CI 59 26 1 0.962 0.0377 0.890 1.000 115 25 1 0.923 0.0523 0.826 1.000 156 24 1 0.885 0.0627 0.770 1.000 268 23 1 0.846 0.0708 0.718 0.997 329 22 1 0.808 0.0773 0.670 0.974 353 21 1 0.769 0.0826 0.623 0.949 365 20 1 0.731 0.0870 0.579 0.923 377 19 0 0.731 0.0870 0.579 0.923 421 18 0 0.731 0.0870 0.579 0.923 431 17 1 0.688 0.0919 0.529 0.894 448 16 0 0.688 0.0919 0.529 0.894 464 15 1 0.642 0.0965 0.478 0.862 475 14 1 0.596 0.0999 0.429 0.828 477 13 0 0.596 0.0999 0.429 0.828 563 12 1 0.546 0.1032 0.377 0.791 638 11 1 0.497 0.1051 0.328 0.752 744 10 0 0.497 0.1051 0.328 0.752 769 9 0 0.497 0.1051 0.328 0.752 770 8 0 0.497 0.1051 0.328 0.752 803 7 0 0.497 0.1051 0.328 0.752 855 6 0 0.497 0.1051 0.328 0.752 1040 5 0 0.497 0.1051 0.328 0.752 1106 4 0 0.497 0.1051 0.328 0.752 1129 3 0 0.497 0.1051 0.328 0.752 1206 2 0 0.497 0.1051 0.328 0.752 1227 1 0 0.497 0.1051 0.328 0.752 > > # Various models > coxph(Surv(futime, fustat)~ age) Call: coxph(formula = Surv(futime, fustat) ~ age) coef exp(coef) se(coef) z p age 0.1616 1.1754 0.0497 3.25 0.0012 Likelihood ratio test=14.3 on 1 df, p=0.000156 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ resid.ds) Call: coxph(formula = Surv(futime, fustat) ~ resid.ds) coef exp(coef) se(coef) z p resid.ds 1.209 3.351 0.672 1.8 0.072 Likelihood ratio test=3.76 on 1 df, p=0.0525 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ rx) Call: coxph(formula = Surv(futime, fustat) ~ rx) coef exp(coef) se(coef) z p rx -0.596 0.551 0.587 -1.02 0.31 Likelihood ratio test=1.05 on 1 df, p=0.305 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ ecog.ps) Call: coxph(formula = Surv(futime, fustat) ~ ecog.ps) coef exp(coef) se(coef) z p ecog.ps 0.398 1.489 0.586 0.68 0.5 Likelihood ratio test=0.47 on 1 df, p=0.494 n= 26, number of events= 12 > > coxph(Surv(futime, fustat)~ resid.ds + rx + ecog.ps) Call: coxph(formula = Surv(futime, fustat) ~ resid.ds + rx + ecog.ps) coef exp(coef) se(coef) z p resid.ds 1.347 3.844 0.680 1.98 0.048 rx -0.749 0.473 0.595 -1.26 0.208 ecog.ps 0.453 1.573 0.590 0.77 0.443 Likelihood ratio test=6.03 on 3 df, p=0.11 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ age + rx + ecog.ps) Call: coxph(formula = Surv(futime, fustat) ~ age + rx + ecog.ps) coef exp(coef) se(coef) z p age 0.1470 1.1583 0.0463 3.17 0.0015 rx -0.8146 0.4428 0.6342 -1.28 0.1990 ecog.ps 0.1032 1.1087 0.6064 0.17 0.8649 Likelihood ratio test=15.9 on 3 df, p=0.00118 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ age + resid.ds + ecog.ps) Call: coxph(formula = Surv(futime, fustat) ~ age + resid.ds + ecog.ps) coef exp(coef) se(coef) z p age 0.142 1.153 0.052 2.74 0.0061 resid.ds 0.663 1.940 0.750 0.88 0.3773 ecog.ps 0.166 1.181 0.615 0.27 0.7867 Likelihood ratio test=15.1 on 3 df, p=0.00173 n= 26, number of events= 12 > coxph(Surv(futime, fustat)~ age + resid.ds + rx) Call: coxph(formula = Surv(futime, fustat) ~ age + resid.ds + rx) coef exp(coef) se(coef) z p age 0.1285 1.1372 0.0473 2.72 0.0066 resid.ds 0.6964 2.0065 0.7585 0.92 0.3586 rx -0.8489 0.4279 0.6392 -1.33 0.1842 Likelihood ratio test=16.8 on 3 df, p=0.000789 n= 26, number of events= 12 > > # Residuals > fit <- coxph(Surv(futime, fustat)~ age + resid.ds + rx + ecog.ps ) > resid(fit) 1 2 3 4 5 6 0.84103277 0.54424388 0.59670824 -0.11281376 0.75111588 -0.32609026 7 8 9 10 11 12 0.59998927 0.29570718 -2.15325805 0.76243469 0.06474272 -0.11680752 13 14 15 16 17 18 -1.22562781 -0.63474839 -0.07535824 -0.17058905 -0.22986038 -0.14654862 19 20 21 22 23 24 -0.18762920 -0.12771548 -0.53373114 -0.65480022 0.95866131 0.82111675 25 26 0.55136554 -0.09154014 > resid(fit, 'dev') 1 2 3 4 5 6 1.41281595 0.69505907 0.78916003 -0.47500266 1.13106322 -0.80757694 7 8 9 10 11 12 0.79532966 0.33122166 -2.07521471 1.16179002 0.06619519 -0.48333740 13 14 15 16 17 18 -1.56564862 -1.12671948 -0.38822221 -0.58410453 -0.67802711 -0.54138455 19 20 21 22 23 24 -0.61258338 -0.50540178 -1.03318066 -0.54976346 2.11059000 1.34157009 25 26 0.70736314 -0.42787881 > resid(fit, 'scor') age resid.ds rx ecog.ps 1 2.26503249 0.05686357 -0.10565379 -0.42661688 2 3.02525428 0.04641312 -0.08623662 -0.34821275 3 -0.06851355 0.07131430 -0.13250357 0.06167527 4 0.94597623 -0.02541510 -0.06423496 0.05971729 5 -5.41507168 0.21605962 -0.32258092 -0.39333909 6 1.48999552 0.24899474 0.14035143 -0.15380664 7 -0.68612431 0.13740891 0.28392482 0.29196506 8 0.93116906 0.08428957 0.16040160 0.18430641 9 -8.20092595 -0.51356176 0.95647608 1.11337112 10 0.95287510 -0.31078224 0.21463992 0.17363388 11 2.85526159 0.09417730 -0.14186603 -0.07586086 12 0.92721107 0.07495002 -0.05400751 0.07061578 13 -1.93962967 -0.43919871 -0.56668535 -0.48467672 14 0.63185387 -0.22745949 -0.29348437 0.38373600 15 1.41495195 0.04835392 0.04051535 0.04555769 16 2.54591188 0.10945916 0.09171493 -0.06745975 17 4.40282381 -0.08236953 0.12358137 -0.09089870 18 1.97071836 0.09403352 0.07878991 0.08859570 19 0.77692371 0.12039304 -0.08675286 0.11343089 20 1.76784279 -0.04576632 -0.05905095 0.07721016 21 -0.82272526 0.34247077 -0.24677770 -0.21106494 22 -3.48057998 -0.03965965 0.07368852 -0.26669335 23 -14.86623758 0.28137017 -0.52279208 -0.43881151 24 3.96084273 -0.56566921 0.34648950 0.44907410 25 4.30025715 0.15241262 0.22417527 -0.20390438 26 0.31490641 0.07091764 -0.05212198 0.04845623 > resid(fit, 'scho') age resid.ds rx ecog.ps 59 2.69315603 0.06761160 -0.1256239 -0.5072536 115 5.36390105 0.08039116 -0.1493686 -0.6031318 156 -0.89877512 0.10683985 -0.1985108 0.1984379 268 6.95664326 0.12857949 -0.2389036 0.2388157 329 -15.73656605 0.28889883 -0.5367805 -0.4634169 353 4.06104389 -0.70587654 0.4535120 0.5282024 365 5.50035833 0.25348264 0.4796230 -0.4413864 431 -8.06809505 0.27490176 -0.4297023 -0.5248323 464 -2.15471559 0.23158421 0.5066040 0.4814387 475 0.57065051 0.25226659 0.5518479 0.5244351 563 0.06487219 -0.47274522 0.3319974 0.2747028 638 1.64752655 -0.50593437 -0.6446947 0.2939883 > > fit <- coxph(Surv(futime, fustat) ~ age + ecog.ps + strata(rx)) > summary(fit) Call: coxph(formula = Surv(futime, fustat) ~ age + ecog.ps + strata(rx)) n= 26, number of events= 12 coef exp(coef) se(coef) z Pr(>|z|) age 0.13853 1.14858 0.04801 2.885 0.00391 ** ecog.ps -0.09670 0.90783 0.62994 -0.154 0.87800 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 exp(coef) exp(-coef) lower .95 upper .95 age 1.1486 0.8706 1.0454 1.262 ecog.ps 0.9078 1.1015 0.2641 3.120 Concordance= 0.819 (se = 0.134 ) Rsquare= 0.387 (max possible= 0.874 ) Likelihood ratio test= 12.71 on 2 df, p=0.001736 Wald test = 8.43 on 2 df, p=0.01476 Score (logrank) test = 12.24 on 2 df, p=0.002195 > summary(survfit(fit)) Call: survfit(formula = fit) rx=1 time n.risk n.event survival std.err lower 95% CI upper 95% CI 59 13 1 0.978 0.0266 0.9275 1 115 12 1 0.951 0.0478 0.8620 1 156 11 1 0.910 0.0760 0.7722 1 268 10 1 0.862 0.1055 0.6776 1 329 9 1 0.737 0.1525 0.4909 1 431 8 1 0.627 0.1704 0.3680 1 638 5 1 0.333 0.2296 0.0865 1 rx=2 time n.risk n.event survival std.err lower 95% CI upper 95% CI 353 13 1 0.943 0.0560 0.839 1.000 365 12 1 0.880 0.0812 0.735 1.000 464 9 1 0.789 0.1143 0.594 1.000 475 8 1 0.697 0.1349 0.477 1.000 563 7 1 0.597 0.1494 0.366 0.975 > sfit <- survfit(fit, list(age=c(30,70), ecog.ps=c(2,3))) #two columns > sfit Call: survfit(formula = fit, newdata = list(age = c(30, 70), ecog.ps = c(2, 3))) n events median 0.95LCL 0.95UCL 1, rx=1 13 7 NA NA NA 2, rx=1 13 7 268 115 NA 1, rx=2 13 5 NA NA NA 2, rx=2 13 5 365 353 NA > summary(sfit) Call: survfit(formula = fit, newdata = list(age = c(30, 70), ecog.ps = c(2, 3))) rx=1 time n.risk n.event survival1 survival2 59 13 1 0.999 0.87905 115 12 1 0.999 0.74575 156 11 1 0.998 0.57398 268 10 1 0.996 0.41764 329 9 1 0.992 0.16673 431 8 1 0.988 0.06489 638 5 1 0.973 0.00161 rx=2 time n.risk n.event survival1 survival2 353 13 1 0.999 0.7092 365 12 1 0.997 0.4738 464 9 1 0.994 0.2494 475 8 1 0.991 0.1207 563 7 1 0.987 0.0489 > detach() > > > # Check of offset + surv, added 7/2000 > fit1 <- coxph(Surv(futime, fustat) ~ age + rx, ovarian, + control=coxph.control(eps=1e-8)) > fit2 <- coxph(Surv(futime, fustat) ~ age + offset(rx*fit1$coef[2]), ovarian, + control=coxph.control(eps=1e-8)) > all.equal(fit1$coef[1], fit2$coef[1]) [1] TRUE > > fit <- coxph(Surv(futime, fustat) ~ age + offset(rx), ovarian) > survfit(fit, censor=FALSE)$surv^exp(-1.5) [1] 0.9977751 0.9951975 0.9917927 0.9881504 0.9825769 0.9770280 0.9704304 [8] 0.9603196 0.9499085 0.9385539 0.9217097 0.9031334 > > # Check it by hand -- there are no tied times > # Remember that offsets from survfit are centered, which is 1.5 for > # this data set. > eta <- fit$coef*(ovarian$age - fit$mean) + (ovarian$rx - 1.5) > ord <- order(ovarian$futime) > risk <- exp(eta[ord]) > rsum <- rev(cumsum(rev(risk))) # cumulative risk at each time point > dead <- (ovarian$fustat[ord]==1) > baseline <- cumsum(1/rsum[dead]) > all.equal(survfit(fit, censor=FALSE)$surv, exp(-baseline)) [1] TRUE > > rm(fit, fit1, fit2, ord, eta, risk, rsum, dead, baseline, sfit) > > proc.time() user system elapsed 0.276 0.036 0.305 survival/tests/fr_kidney.Rout.save0000644000175100001440000003205712536401367017105 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # From: McGilchrist and Aisbett, Biometrics 47, 461-66, 1991 > # Data on the recurrence times to infection, at the point of insertion of > # the catheter, for kidney patients using portable dialysis equipment. > # Catheters may be removed for reasons other than infection, in which case > # the observation is censored. Each patient has exactly 2 observations. > > # Variables: patient, time, status, age, > # sex (1=male, 2=female), > # disease type (0=GN, 1=AN, 2=PKD, 3=Other) > # author's estimate of the frailty > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > # I don't match their answers, and I think that I'm right > kfit <- coxph(Surv(time, status)~ age + sex + disease + frailty(id), kidney) > kfit1<- coxph(Surv(time, status) ~age + sex + disease + + frailty(id, theta=1), kidney, iter=20) > kfit0 <- coxph(Surv(time, status)~ age + sex + disease, kidney) > temp <- coxph(Surv(time, status) ~age + sex + disease + + frailty(id, theta=1, sparse=F), kidney) > > > # Check out the EM based score equations > # temp1 and kfit1 should have essentially the same coefficients > # temp2 should equal kfit1$frail > # equality won't be exact because of the different iteration paths > temp1 <- coxph(Surv(time, status) ~ age + sex + disease + + offset(kfit1$frail[id]), kidney) > rr <- tapply(resid(temp1), kidney$id, sum) > temp2 <- log(rr/1 +1) > aeq(temp1$coef, kfit1$coef, tolerance=.005) [1] TRUE > aeq(temp2, kfit1$frail, tolerance=.005) [1] TRUE > > > > kfit Call: coxph(formula = Surv(time, status) ~ age + sex + disease + frailty(id), data = kidney) coef se(coef) se2 Chisq DF p age 3.18e-03 1.11e-02 1.11e-02 8.14e-02 1 0.775 sex -1.48e+00 3.58e-01 3.58e-01 1.71e+01 1 3.5e-05 diseaseGN 8.80e-02 4.06e-01 4.06e-01 4.68e-02 1 0.829 diseaseAN 3.51e-01 4.00e-01 4.00e-01 7.70e-01 1 0.380 diseasePKD -1.43e+00 6.31e-01 6.31e-01 5.14e+00 1 0.023 frailty(id) 2.71e-05 0 0.933 Iterations: 6 outer, 35 Newton-Raphson Variance of random effect= 5e-07 I-likelihood = -179.1 Degrees of freedom for terms= 1 1 3 0 Likelihood ratio test=17.6 on 5 df, p=0.00342 n= 76 > kfit1 Call: coxph(formula = Surv(time, status) ~ age + sex + disease + frailty(id, theta = 1), data = kidney, iter = 20) coef se(coef) se2 Chisq DF p age 0.00389 0.01959 0.00943 0.03933 1.0 0.84280 sex -2.00764 0.59104 0.41061 11.53834 1.0 0.00068 diseaseGN 0.35335 0.71653 0.38015 0.24319 1.0 0.62191 diseaseAN 0.52341 0.72298 0.40463 0.52413 1.0 0.46909 diseasePKD -0.45938 1.08977 0.66088 0.17770 1.0 0.67336 frailty(id, theta = 1) 28.50571 18.8 0.06909 Iterations: 1 outer, 14 Newton-Raphson Variance of random effect= 1 I-likelihood = -182.5 Degrees of freedom for terms= 0.2 0.5 1.1 18.8 Likelihood ratio test=63.8 on 20.6 df, p=2.53e-06 n= 76 > kfit0 Call: coxph(formula = Surv(time, status) ~ age + sex + disease, data = kidney) coef exp(coef) se(coef) z p age 0.00318 1.00319 0.01115 0.29 0.775 sex -1.48314 0.22692 0.35823 -4.14 3.5e-05 diseaseGN 0.08796 1.09194 0.40637 0.22 0.829 diseaseAN 0.35079 1.42020 0.39972 0.88 0.380 diseasePKD -1.43111 0.23904 0.63111 -2.27 0.023 Likelihood ratio test=17.6 on 5 df, p=0.00342 n= 76, number of events= 58 > temp Call: coxph(formula = Surv(time, status) ~ age + sex + disease + frailty(id, theta = 1, sparse = F), data = kidney) coef se(coef) se2 Chisq DF p age 0.00389 0.01865 0.01120 0.04342 1.0 0.83494 sex -2.00763 0.57624 0.40799 12.13849 1.0 0.00049 diseaseGN 0.35335 0.67865 0.43154 0.27109 1.0 0.60260 diseaseAN 0.52340 0.68910 0.44038 0.57690 1.0 0.44753 diseasePKD -0.45934 1.01394 0.71297 0.20523 1.0 0.65053 frailty(id, theta = 1, sp 26.23016 18.7 0.11573 Iterations: 1 outer, 5 Newton-Raphson Variance of random effect= 1 I-likelihood = -182.5 Degrees of freedom for terms= 0.4 0.5 1.4 18.7 Likelihood ratio test=63.8 on 21 df, p=3.27e-06 n= 76 > > # > # Now fit the data using REML > # > kfitm1 <- coxph(Surv(time,status) ~ age + sex + disease + + frailty(id, dist='gauss'), kidney) > kfitm2 <- coxph(Surv(time,status) ~ age + sex + disease + + frailty(id, dist='gauss', sparse=F), kidney) > kfitm1 Call: coxph(formula = Surv(time, status) ~ age + sex + disease + frailty(id, dist = "gauss"), data = kidney) coef se(coef) se2 Chisq DF p age 0.00489 0.01497 0.01059 0.10678 1.0 0.74384 sex -1.69728 0.46101 0.36170 13.55454 1.0 0.00023 diseaseGN 0.17986 0.54485 0.39273 0.10897 1.0 0.74131 diseaseAN 0.39294 0.54482 0.39816 0.52016 1.0 0.47077 diseasePKD -1.13631 0.82519 0.61728 1.89621 1.0 0.16850 frailty(id, dist = "gauss 17.89195 12.1 0.12376 Iterations: 7 outer, 42 Newton-Raphson Variance of random effect= 0.493 Degrees of freedom for terms= 0.5 0.6 1.7 12.1 Likelihood ratio test=47.5 on 14.9 df, p=2.82e-05 n= 76 > summary(kfitm2) Call: coxph(formula = Surv(time, status) ~ age + sex + disease + frailty(id, dist = "gauss", sparse = F), data = kidney) n= 76, number of events= 58 coef se(coef) se2 Chisq DF p age 0.004924 0.0149 0.01084 0.11 1.00 0.74000 sex -1.702037 0.4631 0.36134 13.51 1.00 0.00024 diseaseGN 0.181733 0.5413 0.40169 0.11 1.00 0.74000 diseaseAN 0.394416 0.5428 0.40520 0.53 1.00 0.47000 diseasePKD -1.131602 0.8175 0.62981 1.92 1.00 0.17000 frailty(id, dist = "gauss 18.13 12.27 0.12000 exp(coef) exp(-coef) lower .95 upper .95 age 1.0049 0.9951 0.97601 1.0347 sex 0.1823 5.4851 0.07355 0.4519 diseaseGN 1.1993 0.8338 0.41515 3.4646 diseaseAN 1.4835 0.6741 0.51196 4.2988 diseasePKD 0.3225 3.1006 0.06497 1.6010 gauss:1 1.7011 0.5879 0.51805 5.5856 gauss:2 1.4241 0.7022 0.38513 5.2662 gauss:3 1.1593 0.8626 0.38282 3.5108 gauss:4 0.6226 1.6063 0.23397 1.6566 gauss:5 1.2543 0.7972 0.39806 3.9526 gauss:6 1.1350 0.8811 0.38339 3.3599 gauss:7 1.9726 0.5069 0.56938 6.8342 gauss:8 0.6196 1.6140 0.21662 1.7721 gauss:9 0.8231 1.2149 0.28884 2.3456 gauss:10 0.5030 1.9882 0.17468 1.4482 gauss:11 0.7565 1.3218 0.27081 2.1134 gauss:12 1.1048 0.9052 0.33430 3.6510 gauss:13 1.3022 0.7679 0.42746 3.9673 gauss:14 0.5912 1.6915 0.18537 1.8855 gauss:15 0.5449 1.8352 0.18580 1.5980 gauss:16 1.0443 0.9576 0.31424 3.4702 gauss:17 0.9136 1.0945 0.30004 2.7820 gauss:18 0.9184 1.0889 0.32476 2.5970 gauss:19 0.6426 1.5562 0.19509 2.1166 gauss:20 1.1698 0.8549 0.34528 3.9631 gauss:21 0.3336 2.9974 0.10202 1.0910 gauss:22 0.6871 1.4554 0.23531 2.0064 gauss:23 1.4778 0.6767 0.47560 4.5918 gauss:24 1.0170 0.9832 0.31555 3.2779 gauss:25 0.8096 1.2352 0.27491 2.3843 gauss:26 0.6145 1.6274 0.21491 1.7570 gauss:27 1.0885 0.9187 0.32819 3.6101 gauss:28 1.5419 0.6485 0.49231 4.8292 gauss:29 1.3785 0.7254 0.43766 4.3421 gauss:30 1.3748 0.7274 0.44444 4.2530 gauss:31 1.4447 0.6922 0.47031 4.4380 gauss:32 1.1993 0.8339 0.35207 4.0850 gauss:33 1.9449 0.5142 0.55229 6.8491 gauss:34 0.8617 1.1605 0.27685 2.6820 gauss:35 1.7031 0.5872 0.52657 5.5084 gauss:36 0.8275 1.2085 0.22811 3.0015 gauss:37 1.4707 0.6800 0.38936 5.5549 gauss:38 1.0479 0.9543 0.30685 3.5789 Iterations: 6 outer, 21 Newton-Raphson Variance of random effect= 0.5090956 Degrees of freedom for terms= 0.5 0.6 1.7 12.3 Concordance= 0.796 (se = 0.046 ) Likelihood ratio test= 117.9 on 15.14 df, p=0 > # > # Fit the kidney data using AIC > # > > # gamma, corrected aic > coxph(Surv(time, status) ~ age + sex + frailty(id, method='aic', caic=T), + kidney) Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, method = "aic", caic = T), data = kidney) coef se(coef) se2 Chisq DF p age 0.00364 0.01048 0.00891 0.12053 1.00 0.72846 sex -1.31953 0.39556 0.32497 11.12781 1.00 0.00085 frailty(id, method = "aic 13.55258 7.81 0.08692 Iterations: 9 outer, 63 Newton-Raphson Variance of random effect= 0.203 I-likelihood = -182.1 Degrees of freedom for terms= 0.7 0.7 7.8 Likelihood ratio test=33.3 on 9.21 df, p=0.000137 n= 76 > > coxph(Surv(time, status) ~ age + sex + frailty(id, dist='t'), kidney) Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, dist = "t"), data = kidney) coef se(coef) se2 Chisq DF p age 0.00561 0.01203 0.00872 0.21774 1.0 0.64077 sex -1.65487 0.48294 0.38527 11.74180 1.0 0.00061 frailty(id, dist = "t") 20.33462 13.9 0.11752 Iterations: 8 outer, 58 Newton-Raphson Variance of random effect= 0.825 Degrees of freedom for terms= 0.5 0.6 13.9 Likelihood ratio test=48.6 on 15.1 df, p=2.18e-05 n= 76 > coxph(Surv(time, status) ~ age + sex + frailty(id, dist='gauss', method='aic', + caic=T), kidney) Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, dist = "gauss", method = "aic", caic = T), data = kidney) coef se(coef) se2 Chisq DF p age 0.00303 0.01031 0.00895 0.08646 1.00 0.7687 sex -1.15152 0.36368 0.30556 10.02558 1.00 0.0015 frailty(id, dist = "gauss 12.35238 6.76 0.0800 Iterations: 7 outer, 41 Newton-Raphson Variance of random effect= 0.185 Degrees of freedom for terms= 0.8 0.7 6.8 Likelihood ratio test=28.4 on 8.22 df, p=0.000476 n= 76 > > > # uncorrected aic > coxph(Surv(time, status) ~ age + sex + frailty(id, method='aic', caic=F), + kidney) Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, method = "aic", caic = F), data = kidney) coef se(coef) se2 Chisq DF p age 0.00785 0.01503 0.00823 0.27284 1.0 0.60143 sex -1.88990 0.56114 0.39941 11.34311 1.0 0.00076 frailty(id, method = "aic 37.45897 19.7 0.00918 Iterations: 8 outer, 87 Newton-Raphson Variance of random effect= 0.886 I-likelihood = -182.8 Degrees of freedom for terms= 0.3 0.5 19.7 Likelihood ratio test=61.2 on 20.5 df, p=6.25e-06 n= 76 Warning message: In coxpenal.fit(X, Y, strats, offset, init = init, control, weights = weights, : Inner loop failed to coverge for iterations 4 > > coxph(Surv(time, status) ~ age + sex + frailty(id, dist='t', caic=F), kidney) Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, dist = "t", caic = F), data = kidney) coef se(coef) se2 Chisq DF p age 0.00561 0.01203 0.00872 0.21774 1.0 0.64077 sex -1.65487 0.48294 0.38527 11.74180 1.0 0.00061 frailty(id, dist = "t", c 20.33462 13.9 0.11752 Iterations: 8 outer, 58 Newton-Raphson Variance of random effect= 0.825 Degrees of freedom for terms= 0.5 0.6 13.9 Likelihood ratio test=48.6 on 15.1 df, p=2.18e-05 n= 76 > > proc.time() user system elapsed 0.396 0.028 0.419 survival/tests/singtest.R0000644000175100001440000000143512030335206015265 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # A simple test of an overdetermined system # Should give a set of NA coefficients # test1 <- data.frame(time= c(4, 3,1,1,2,2,3), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) temp <- rep(0:3, rep(7,4)) stest <- data.frame(start = 10*temp, stop = 10*temp + test1$time, status = rep(test1$status,4), x = c(test1$x+ 1:7, rep(test1$x,3)), epoch = rep(1:4, rep(7,4))) # Will create a warning about a singular X matrix fit1 <- coxph(Surv(start, stop, status) ~ x * factor(epoch), stest) fit1$coef # elements 2:4 should be NA all.equal(is.na(fit1$coef), c(F,T,T,T,F,F,F), check.attributes=FALSE) survival/tests/data.cracks0000644000175100001440000000017711732700061015410 0ustar hornikusersNA 6.12 5 6.12 19.92 16 19.92 29.64 12 29.64 35.40 18 35.40 39.72 18 39.72 45.24 2 45.24 52.32 6 52.32 63.48 17 63.48 NA 73 survival/tests/testci2.R0000644000175100001440000001177412055202643015016 0ustar hornikuserslibrary(survival) # # Test the multi-state version of the CI curve # tdata <- data.frame(id=c(1,1,1,1, 2,2,2, 3,3, 4,4,4,4, 5, 6, 6), time1=c(0, 10,20,30, 0, 5, 15, 0, 20, 0, 6,18,34, 0, 0,15), time2=c(10,20,30,40, 5, 15,25, 20, 22, 6,18,34,50,10,15,20), status=c(1,1,1,1, 1,1,1, 1,0, 1,1,1,0,0,1,0), event= letters[c(1,2,3,4, 2,4,3, 2,2, 3,1,2,2,1, 1,1)], wt = c(2,2,2,2, 1,1,1, 3,3, 1,1,1,1, 2, 1,1), stringsAsFactors=TRUE) tdata$stat2 <- factor(tdata$status * as.numeric(tdata$event), labels=c(" ", levels(tdata$event))) fit <- survfit(Surv(time1, time2, stat2) ~1, id=id, weight=wt, tdata) # The exact figures for testci2. # The subject data of id, weight, (transition time, transition) #1: 2 (10, 0->a) (20, a->b) (30, b->c) (40, c->d) no data after 40=censored #2: 1 ( 5, 0->b) (15, b->d) (25, d->c) no data after 25 implies censored then #3: 3 (20, 0->b) (22, censor) #4: 1 ( 6, 0->c) (18, c->a) (34, a->b) (50, censor) #5: 2 (10, censor) #6: 1 (15, 0->a) (20, censor) # Each line below follows a subject through time as a (state, rdist weight) pair # using the redistribute to the right algorithm. # RDR algorithm: at each censoring (or last fu) a subject's weight is put into # a "pool" for that state and their weight goes to zero. The pool is # dynamically shared between all members of the state proportional to their # original case weight, when someone leaves they take their portion of the # pool to the new state. # Table of case weights and state, blank is weight of zero # time 5 6 10 15 18 20 25 30 34 40 50 # ----------------------------------------------------------------------- # id, wt # 1, 2 - - a a a b b c c d # 2, 1 b b b d d d c # 3, 3 - - - - - b # 4, 1 - c c c a a a a b b b # 5, 2 - - - # 6, 1 - - - a a a # Pool weights # 10 10+ 15 18 20 20+ 22+ 25 25+ 30 34 40 40+ # - 0 2 3/2 3/2 0 # a 0 0 1/2 1/2 1/4 5/4 5/4 5/4 5/4 5/4 # b 0 0 0 0 7/4 7/4 19/4 19/4 19/4 5/4 5/4 5/4 # c 0 0 0 0 0 1 23/4 23/4 # d 0 0 0 0 0 23/4 31/4 # fit$prev for time i and state j = total weight at that time/state in the # above table (original weight + redistrib), divided by 10. # time 5 6 10 15 18 20 25 30 34 40 50 truth <- matrix(c(0, 0, 2, 3, 4, 2, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 5, 2, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0, 1, 2, 2, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 2, 0) + c(0, 0, 0, .5, .5, 1/4, 5/4, 5/4, 0, 0, 0, 0, 0, 0, 0, 0, 7/4, 19/4, 0, 5/4, 5/4, 5/4, 0, 0, 0, 0, 0, 0, 0, 23/4, 23/4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 23/4, 31/4), ncol=4) truth <- truth[c(1:6, 6:11),]/10 #the explicit censor at 22 #dimnames(truth) <- list(c(5, 6, 10, 15, 18, 20, 25, 30, 34, 40, 50), # c('a', 'b', 'c', 'd') all.equal(truth, fit$prev) # Test the dfbetas dfbeta <- array(0., dim=c(6, nrow(fit$prev), ncol(fit$prev))) eps <- 1e-6 for (i in 1:6) { twt <- tdata$wt twt[tdata$id ==i] <- twt[tdata$id==i] + eps tfit <- survfit(Surv(time1, time2, stat2) ~ cluster(id), tdata, weight=twt) dfbeta[i,,] <- (tfit$prev - fit$prev)/eps } twt <- tdata$wt[match(1:6, tdata$id)] temp <- (twt*dfbeta) * dfbeta tstd <- sqrt(apply(temp, 2:3, sum)) all.equal(tstd, fit$std.err, tolerance=eps) if (FALSE) { # a plot of the data that helped during creation of the example plot(c(0,50), c(1,6), type='n', xlab='time', ylab='subject') with(tdata, segments(time1, id, time2, id)) with(tdata, text(time2, id, as.numeric(stat2)-1, cex=1.5, col=2)) } if (FALSE) { # The following lines test out 4 error messages in the routine # # Gap in follow-up time, id 2 survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 4, 6, 3), factor(c(0,0,1,1,0,2))) ~1, id=c(1,1,1,2,2,3)) # mismatched weights survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,0,1,1,0,2))) ~1, id=c(1,1,1,2,2,3), weights=c(1,1,2,1,1,4)) # in two groups at once survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,0,1,1,0,2))) ~ c(1,1,2,1,1,2), id=c(1,1,1,2,2,3)) # state change that isn't a state change (went from 1 to 1) survfit(Surv(c(0,5,9,0,5,0), c(5,9,12, 5, 6, 3), factor(c(0,1,1,1,0,2))) ~1, id=c(1,1,1,2,2,3)) } survival/tests/fr_rat2.Rout.save0000644000175100001440000001306512536400714016463 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # From Gail, Sautner and Brown, Biometrics 36, 255-66, 1980 > > # 48 rats were injected with a carcinogen, and then randomized to either > # drug or placebo. The number of tumors ranges from 0 to 13; all rats were > # censored at 6 months after randomization. > > # Variables: rat, treatment (1=drug, 0=control), o > # observation # within rat, > # (start, stop] status > # The raw data has some intervals of zero length, i.e., start==stop. > # We add .1 to these times as an approximate solution > # > rat2 <- read.table('data.rat2', col.names=c('id', 'rx', 'enum', 'start', + 'stop', 'status')) > temp1 <- rat2$start > temp2 <- rat2$stop > for (i in 1:nrow(rat2)) { + if (temp1[i] == temp2[i]) { + temp2[i] <- temp2[i] + .1 + if (i < nrow(rat2) && rat2$id[i] == rat2$id[i+1]) { + temp1[i+1] <- temp1[i+1] + .1 + if (temp2[i+1] <= temp1[i+1]) temp2[i+1] <- temp1[i+1] + } + } + } > rat2$start <- temp1 > rat2$stop <- temp2 > > r2fit0 <- coxph(Surv(start, stop, status) ~ rx + cluster(id), rat2) > > r2fitg <- coxph(Surv(start, stop, status) ~ rx + frailty(id), rat2) > r2fitm <- coxph(Surv(start, stop, status) ~ rx + frailty.gaussian(id), rat2) > > r2fit0 Call: coxph(formula = Surv(start, stop, status) ~ rx + cluster(id), data = rat2) coef exp(coef) se(coef) robust se z p rx -0.827 0.438 0.151 0.204 -4.05 5.2e-05 Likelihood ratio test=32.9 on 1 df, p=9.89e-09 n= 253, number of events= 212 > r2fitg Call: coxph(formula = Surv(start, stop, status) ~ rx + frailty(id), data = rat2) coef se(coef) se2 Chisq DF p rx -0.838 0.219 0.152 14.572 1.0 0.00013 frailty(id) 57.285 26.4 0.00045 Iterations: 7 outer, 26 Newton-Raphson Variance of random effect= 0.317 I-likelihood = -779.1 Degrees of freedom for terms= 0.5 26.3 Likelihood ratio test=120 on 26.8 df, p=8.43e-14 n= 253 > r2fitm Call: coxph(formula = Surv(start, stop, status) ~ rx + frailty.gaussian(id), data = rat2) coef se(coef) se2 Chisq DF p rx -0.790 0.220 0.154 12.924 1.0 0.00032 frailty.gaussian(id) 60.939 24.9 7.3e-05 Iterations: 6 outer, 23 Newton-Raphson Variance of random effect= 0.303 Degrees of freedom for terms= 0.5 24.9 Likelihood ratio test=118 on 25.4 df, p=6.99e-14 n= 253 > > #This example is unusual: the frailties variances end up about the same, > # but the effect on rx differs. Double check it > # Because of different iteration paths, the coef won't be exactly the > # same, but darn close. > > temp <- coxph(Surv(start, stop, status) ~ rx + offset(r2fitm$frail[id]), rat2) > all.equal(temp$coef, r2fitm$coef[1], tolerance=1e-7) [1] TRUE > > temp <- coxph(Surv(start, stop, status) ~ rx + offset(r2fitg$frail[id]), rat2) > all.equal(temp$coef, r2fitg$coef[1], tolerance=1e-7) [1] TRUE > > # > # What do I get with AIC > # > r2fita1 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, method='aic'), + rat2) > r2fita2 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, method='aic', + dist='gauss'), rat2) > r2fita3 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, dist='t'), + rat2) > > r2fita1 Call: coxph(formula = Surv(start, stop, status) ~ rx + frailty(id, method = "aic"), data = rat2) coef se(coef) se2 Chisq DF p rx -0.838 0.230 0.151 13.315 1.0 0.00026 frailty(id, method = "aic 60.406 28.2 0.00039 Iterations: 10 outer, 34 Newton-Raphson Variance of random effect= 0.375 I-likelihood = -779.2 Degrees of freedom for terms= 0.4 28.2 Likelihood ratio test=124 on 28.6 df, p=7.92e-14 n= 253 > r2fita2 Call: coxph(formula = Surv(start, stop, status) ~ rx + frailty(id, method = "aic", dist = "gauss"), data = rat2) coef se(coef) se2 Chisq DF p rx -0.785 0.245 0.154 10.300 1.0 0.0013 frailty(id, method = "aic 70.383 28.5 2.1e-05 Iterations: 9 outer, 33 Newton-Raphson Variance of random effect= 0.436 Degrees of freedom for terms= 0.4 28.5 Likelihood ratio test=125 on 28.9 df, p=5.93e-14 n= 253 > r2fita3 Call: coxph(formula = Surv(start, stop, status) ~ rx + frailty(id, dist = "t"), data = rat2) coef se(coef) se2 Chisq DF p rx -0.790 0.254 0.157 9.667 1 0.00188 frailty(id, dist = "t") 64.721 30 0.00024 Iterations: 7 outer, 29 Newton-Raphson Variance of random effect= 0.78 Degrees of freedom for terms= 0.4 30.0 Likelihood ratio test=126 on 30.4 df, p=1.39e-13 n= 253 > > proc.time() user system elapsed 0.320 0.012 0.330 survival/tests/testnull.R0000644000175100001440000000113311732700061015275 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # A test of NULL models # fit1 <- coxph(Surv(stop, event) ~ rx + strata(number), bladder, iter=0) fit2 <- coxph(Surv(stop, event) ~ strata(number), bladder) all.equal(fit1$loglik[2], fit2$loglik) all.equal(fit1$resid, fit2$resid) fit1 <- coxph(Surv(start, stop, event) ~ rx + strata(number), bladder2, iter=0) fit2 <- coxph(Surv(start, stop, event) ~ strata(number), bladder2) all.equal(fit1$loglik[2], fit2$loglik) all.equal(fit1$resid, fit2$resid) survival/tests/fr_cancer.R0000644000175100001440000000126111732700061015347 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Here is a test case with multiple smoothing terms # fit0 <- coxph(Surv(time, status) ~ ph.ecog + age, lung) fit1 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,3), lung) fit2 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,4), lung) fit3 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,8), lung) fit4 <- coxph(Surv(time, status) ~ ph.ecog + pspline(wt.loss,3), lung) fit5 <-coxph(Surv(time, status) ~ ph.ecog + pspline(age,3) + pspline(wt.loss,3), lung) fit1 fit2 fit3 fit4 fit5 rm(fit1, fit2, fit3, fit4, fit5) survival/tests/coxsurv4.R0000644000175100001440000000331712160143136015226 0ustar hornikuserslibrary(survival) # Strata by covariate interactions, a case pointed out in early 2011 # by Frank Harrell, which as it turns out had never been computed # correctly by any version of the package. Which shows how often this # case arises in practice. # aeq <- function(x, y, ...) all.equal(as.vector(x), as.vector(y)) fit1 <- coxph(Surv(time, status) ~ wt.loss + age*strata(sex) + strata(ph.ecog), data=lung) tdata <- data.frame(wt.loss=c(10,5,0,10, 15,20,25), age =c(50,60,50,60,70,40,21), sex =c(1,1,2,2,1,1,1), ph.ecog=c(0,0,1,1,2,2,2)) surv1 <- survfit(fit1, newdata=tdata) fit2 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*0), data=lung, init=fit1$coef, iter=0, subset=(sex==1 & ph.ecog==0)) fit2$var <- fit1$var surv2 <- survfit(fit2, newdata=list(wt.loss=c(10,5), age=c(50,60))) s1 <- surv1[1:2] aeq(s1$surv, surv2$surv) #first a vector, second a matrix aeq(s1$std.err, surv2$std.err) aeq(s1[1]$time, surv2$time) aeq(s1[1]$n.event, surv2$n.event) fit3 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*1), data=lung, init=fit1$coef, iter=0, subset=(sex==2 & ph.ecog==1)) fit3$var <- fit1$var surv3 <- survfit(fit3, newdata=list(wt.loss=c(0,10), age=c(50,60))) aeq(surv1[3:4]$surv, surv3$surv) aeq(surv1[3:4]$std, surv3$std) fit4 <- coxph(Surv(time, status) ~ wt.loss + age + I(age*0), data=lung, init=fit1$coef, iter=0, subset=(sex==1 & ph.ecog==2)) fit4$var <- fit1$var surv4 <- survfit(fit4, newdata=list(wt.loss=c(15,20,25), age=c(70,40,21))) aeq(surv1[5:7]$surv, surv4$surv) aeq(surv1[5:7]$std.err, surv4$std.err) aeq(surv1[5]$n.risk, surv4$n.risk) survival/tests/coxsurv3.Rout.save0000644000175100001440000001124511732700061016711 0ustar hornikusers R version 2.11.0 (2010-04-22) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > # One more test on coxph survival curves, to test out the individual > # option. First fit a model with a time dependent covariate > # > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > # True hazard function, from the validation document > lambda <- function(beta, x=0, method='efron') { + r <- exp(beta) + lambda <- c(1/(r+1), 1/(r+2), 1/(3*r +2), 1/(3*r+1), + 1/(3*r+1), 1/(3*r+2) + 1/(2*r +2)) + if (method == 'breslow') lambda[9] <- 2/(3*r +2) + list(time=c(2,3,6,7,8,9), lambda=lambda) + } > > fit <- coxph(Surv(start, stop, event) ~x, test2) > # A curve for someone who never changes > surv1 <-survfit(fit, newdata=list(x=0), censor=FALSE) > > true <- lambda(fit$coef, 0) > > aeq(true$time, surv1$time) [1] TRUE > aeq(-log(surv1$surv), cumsum(true$lambda)) [1] TRUE > > # Reprise it with a time dependent subject who doesn't change > data2 <- data.frame(start=c(0, 4, 9, 11), stop=c(4, 9, 11, 17), + event=c(0,0,0,0), x=c(0,0,0,0)) > surv2 <- survfit(fit, newdata=data2, individual=TRUE, censor=FALSE) > aeq(surv2$surv, surv1$surv) [1] TRUE > > > # > # Now a more complex data set with multiple strata > # > test3 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17, + 1:11), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0, + 0, 1, 1, 0, 0, 1, 1, 0, 1, 0,1), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0, + 1, 2, 3, 2, 1, 1, 1, 0, 2, 1,0), + grp = c(rep('a', 10), rep('b', 11))) > > fit2 <- coxph(Surv(start, stop, event) ~ x + strata(grp), test3) > > # The above tests show the program works for a simple case, use it to > # get a true baseline for strata 2 > fit2b <- coxph(Surv(start, stop, event) ~x, test3, + subset=(grp=='b'), init=fit2$coef, iter=0) > temp <- survfit(fit2b, newdata=list(x=0), censor=F) > true2 <- list(time=temp$time, lambda=diff(c(0, -log(temp$surv)))) > true1 <- lambda(fit2$coef, x=0) > > # Separate strata, one value > surv3 <- survfit(fit2, list(x=0), censor=FALSE) > aeq(true1$time, (surv3[1])$time) [1] TRUE > aeq(-log(surv3[1]$surv), cumsum(true1$lambda)) [1] TRUE > > data4 <- data.frame(start=c(0, 4, 9, 11), stop=c(4, 9, 11, 17), + event=c(0,0,0,0), x=c(0,0,0,0), grp=rep('a', 4)) > surv4a <- survfit(fit2, newdata=data4, individual=T, censor=FALSE) > aeq(-log(surv4a$surv), cumsum(true1$lambda)) [1] TRUE > > data4$grp <- rep('b',4) > surv4b <- survfit(fit2, newdata=data4, individual=T, censor=FALSE) > aeq(-log(surv4b$surv), cumsum(true2$lambda)) [1] TRUE > > > # Now for something more complex > # Subject 1 skips day 4. Since there were no events that day the survival > # will be the same, but the times will be different. > # Subject 2 spends some time in strata 1, some in strata 2, with > # moving covariates > # > data5 <- data.frame(start=c(0,5,9,11, + 0, 4, 3), + stop =c(4,9,11,17, 4,8,7), + event=rep(0,7), + x=c(1,1,1,1, 0,1,2), + grp=c('a', 'a', 'a', 'a', 'a', 'a', 'b'), + subject=c(1,1,1,1, 2,2,2)) > surv5 <- survfit(fit2, newdata=data5, censor=FALSE, id=subject) > > aeq(surv5[1]$time, c(2,3,5,6,7,8)) #surv1 has 2, 3, 6, 7, 8, 9 [1] TRUE > aeq(surv5[1]$surv, surv3[1]$surv ^ exp(fit2$coef)) [1] TRUE > > tlam <- c(true1$lambda[1:2]* exp(fit2$coef * data5$x[5]), + true1$lambda[3:5]* exp(fit2$coef * data5$x[6]), + true2$lambda[3:4]* exp(fit2$coef * data5$x[7])) > aeq(-log(surv5[2]$surv), cumsum(tlam)) [1] TRUE > > > > survival/tests/data.valve0000644000175100001440000000233611732700061015256 0ustar hornikusers251 761 -1 252 759 -1 327 98 1 327 667 -1 328 326 1 328 653 1 328 653 1 328 667 -1 329 665 -1 330 84 1 330 667 -1 331 87 1 331 663 -1 389 646 1 389 653 -1 390 92 1 390 653 -1 391 651 -1 392 258 1 392 328 1 392 377 1 392 621 1 392 650 -1 393 61 1 393 539 1 393 648 -1 394 254 1 394 276 1 394 298 1 394 640 1 394 644 -1 395 76 1 395 538 1 395 642 -1 396 635 1 396 641 -1 397 349 1 397 404 1 397 561 1 397 649 -1 398 631 -1 399 596 -1 400 120 1 400 479 1 400 614 -1 401 323 1 401 449 1 401 582 -1 402 139 1 402 139 1 402 589 -1 403 593 -1 404 573 1 404 589 -1 405 165 1 405 408 1 405 604 1 405 606 -1 406 249 1 406 594 -1 407 344 1 407 497 1 407 613 -1 408 265 1 408 586 1 408 595 -1 409 166 1 409 206 1 409 348 1 409 389 -1 410 601 -1 411 410 1 411 581 1 411 601 -1 412 611 -1 413 608 -1 414 587 -1 415 367 1 415 603 -1 416 202 1 416 563 1 416 570 1 416 585 -1 417 587 -1 418 578 -1 419 578 -1 420 586 -1 421 585 -1 422 582 -1 survival/tests/survtest.Rout.save0000644000175100001440000000621211732700061017012 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Simple test of (start, stop] Kaplan-Meier curves, using the test2 data > # set > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > fit1 <- survfit(Surv(start, stop, event) ~1, test2, type='fh2', + error='tsiatis') > fit2 <- survfit(Surv(start, stop, event) ~x, test2, start.time=3, + type='fh2') > > cfit1<- survfit(coxph(Surv(start, stop, event)~1, test2)) > cfit2<- survfit(coxph(Surv(start, stop, event) ~ strata(x), test2, subset=-1)) > > deaths <- (fit1$n.event + fit1$n.censor)>0 > aeq(fit1$time[deaths], cfit1$time) [1] TRUE > aeq(fit1$n.risk[deaths], cfit1$n.risk) [1] TRUE > aeq(fit1$n.event[deaths], cfit1$n.event) [1] TRUE > aeq(fit1$surv[deaths], cfit1$surv) [1] TRUE > aeq(fit1$std.err[deaths], cfit1$std.err) [1] TRUE > > deaths <- (fit2$n.event + fit2$n.censor)>0 > aeq(fit2$time[deaths], cfit2$time) [1] TRUE > aeq(fit2$n.risk[deaths], cfit2$n.risk) [1] TRUE > aeq(fit2$n.event[deaths], cfit2$n.event) [1] TRUE > aeq(fit2$surv[deaths], cfit2$surv) [1] TRUE > > fit3 <- survfit(Surv(start, stop, event) ~1, test2) #Kaplan-Meier > aeq(fit3$n, 10) [1] TRUE > aeq(fit3$time, c(1:9,14,17)) [1] TRUE > aeq(fit3$n.risk, c(0,2,3,3,4,5,4,4,5,2,1)) [1] TRUE > aeq(fit3$n.event,c(0,1,1,0,0,1,1,1,2,0,0)) [1] TRUE > aeq(fit3$surv[fit3$n.event>0], c(.5, 1/3, 4/15, 1/5, 3/20, 9/100)) [1] TRUE > # > # Verify that both surv AND n.risk are right between time points. > # > fit <- survfit(Surv(time, status) ~1, test1) > temp <- summary(fit, time=c(.5,1, 1.5, 6, 7.5, 8, 8.9, 9, 10), extend=TRUE) > > aeq(temp$n.risk, c(6,6,4,4,2,2,1,1,0)) [1] TRUE > aeq(temp$surv, c(1, fit$surv[c(1,1,2,2,3,3,4,4)])) [1] TRUE > aeq(temp$n.event, c(0,1,0,2,0,0,0,1,0)) [1] TRUE > aeq(temp$std.err, c(0, (fit$surv*fit$std.err)[c(1,1,2,2,3,3,4,4)])) [1] TRUE > > > fit <- survfit(Surv(start, stop, event) ~1, test2) > temp <- summary(fit, times=c(.5, 1.5, 2.5, 3, 6.5, 14.5, 16.5)) > aeq(temp$surv, c(1, fit$surv[c(1,2,3,6, 10,10)])) [1] TRUE > aeq(temp$n.risk, c(0, 2, 3, 3, 4, 1,1)) [1] TRUE > survival/tests/factor2.R0000644000175100001440000000163711732700061014774 0ustar hornikuserslibrary(survival) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) options(na.action=na.exclude) # # More tests of factors in prediction, using a new data set # fit <- coxph(Surv(time, status) ~ factor(ph.ecog), lung) tdata <- data.frame(ph.ecog = factor(0:3)) p1 <- predict(fit, newdata=tdata, type='lp') p2 <- predict(fit, type='lp') aeq(p1, p2[match(0:3, lung$ph.ecog)]) fit2 <- coxph(Surv(time, status) ~ factor(ph.ecog) + factor(sex), lung) tdata <- expand.grid(ph.ecog = factor(0:3), sex=factor(1:2)) p1 <- predict(fit2, newdata=tdata, type='risk') xdata <- expand.grid(ph.ecog=factor(1:3), sex=factor(1:2)) p2 <- predict(fit2, newdata=xdata, type='risk') all.equal(p2, p1[c(2:4, 6:8)], check.attributes=FALSE) fit3 <- survreg(Surv(time, status) ~ factor(ph.ecog) + age, lung) tdata <- data.frame(ph.ecog=factor(0:3), age=50) predict(fit, type='lp', newdata=tdata) predict(fit3, type='lp', newdata=tdata) survival/tests/ratetable.R0000644000175100001440000000373011732700061015373 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Generate each of the messages from is.ratetable # {if (is.R()) mdy.date <- function(m, d, y) { y <- ifelse(y<100, y+1900, y) as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") } else mdy.date <- function(m,d,y) { y <- ifelse(y<100, y+1900, y) timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") } } temp <- runif(21*2*4) # Good attributes(temp) <- list(dim=c(21,2,4), dimnames=list(c(as.character(75:95)), c("male","female"), c(as.character(2000:2003))), dimid=c("age","sex","year"), type=c(2,1,4), cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:3)*366.25), class='ratetable') is.ratetable(temp) # Factor problem + cutpoints length attributes(temp) <- list(dim=c(21,2,4), dimnames=list(c(as.character(75:95)), c("male","female"), c(as.character(2000:2003))), dimid=c("age","sex","year"), type=c(1,1,2), cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:4)*366.25), class='ratetable') is.ratetable(temp, verbose=T) # missing dimid attribute + unsorted cutpoint attributes(temp) <- list(dim=c(21,2,4), dimnames=list(c(as.character(75:95)), c("male","female"), c(as.character(2000:2003))), type=c(2,1,3), cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(4:1)*366.25), class='ratetable') is.ratetable(temp, verbose=T) # wrong length for dimid and type, illegal type attributes(temp) <- list(dim=c(21,2,4), dimnames=list(c(as.character(75:95)), c("male","female"), c(as.character(2000:2003))), dimid=c("age","sex","year", "zed"), type=c(2,1,3,6), cutpoints=list(c(75:95), NULL, mdy.date(1,1,2000) +c(0:3)*366.25), class='ratetable') is.ratetable(temp, verbose=T) # Print and summary print(survexp.us[1:30,,c('1953', '1985')] ) summary(survexp.usr) survival/tests/data.donnell0000644000175100001440000001677411732700061015607 0ustar hornikusers0.558521561 1.000000000 0.000000000 1.059548255 1.000000000 0.000000000 1.659137577 1.000000000 0.000000000 0.561259411 1.000000000 0.000000000 1.108829569 1.000000000 0.000000000 1.530458590 1.000000000 0.000000000 0.550308008 1.000000000 0.000000000 1.065023956 1.000000000 0.000000000 1.546885695 1.000000000 0.000000000 0.668035592 1.000000000 0.000000000 1.048596851 1.000000000 0.000000000 1.549623546 1.000000000 0.000000000 0.594113621 1.000000000 0.000000000 1.010266940 1.000000000 0.000000000 1.577002053 1.000000000 0.000000000 0.569472964 1.000000000 0.000000000 1.045859001 1.000000000 0.000000000 1.555099247 1.000000000 0.000000000 0.605065024 1.000000000 0.000000000 1.054072553 1.000000000 0.000000000 1.566050650 1.000000000 0.000000000 0.544832307 1.000000000 0.000000000 1.048596851 1.000000000 0.000000000 1.544147844 1.000000000 0.000000000 0.550308008 1.000000000 0.000000000 1.065023956 1.000000000 0.000000000 1.568788501 1.000000000 0.000000000 0.528405202 1.000000000 0.000000000 1.070499658 1.000000000 0.000000000 1.546885695 1.000000000 0.000000000 0.813141684 1.000000000 0.000000000 1.226557153 1.000000000 0.000000000 1.744010951 1.000000000 0.000000000 0.520191650 1.000000000 0.000000000 1.075975359 1.000000000 0.000000000 1.557837098 1.000000000 0.000000000 0.577686516 1.000000000 0.000000000 1.067761807 1.000000000 0.000000000 1.590691307 1.000000000 0.000000000 0.580424367 1.000000000 0.000000000 1.043121150 1.000000000 0.000000000 1.560574949 1.000000000 0.000000000 0.539356605 1.000000000 0.000000000 1.097878166 1.000000000 0.000000000 1.557837098 1.000000000 0.000000000 0.613278576 1.000000000 0.000000000 1.065023956 1.000000000 0.000000000 1.629021218 1.000000000 0.000000000 0.542094456 1.000000000 0.000000000 1.021218344 1.000000000 0.000000000 1.535934292 1.000000000 0.000000000 0.542094456 1.000000000 0.000000000 1.084188912 1.000000000 0.000000000 1.541409993 1.000000000 0.000000000 0.479123888 1.000000000 0.000000000 1.166324435 1.000000000 0.000000000 1.609856263 1.000000000 0.000000000 0.574948665 1.000000000 0.000000000 1.032169747 1.000000000 0.000000000 1.511293634 1.000000000 0.000000000 0.580424367 1.000000000 0.000000000 1.015742642 1.000000000 0.000000000 1.593429158 1.000000000 0.000000000 0.580424367 1.000000000 0.000000000 0.580424367 1.059548255 3.000000000 0.555783710 1.000000000 0.000000000 1.073237509 1.000000000 0.000000000 1.552361396 1.000000000 0.000000000 0.558521561 1.000000000 0.000000000 1.048596851 1.000000000 0.000000000 1.549623546 1.000000000 0.000000000 0.569472964 1.000000000 0.000000000 1.048596851 1.000000000 0.000000000 1.546885695 1.000000000 0.000000000 0.591375770 1.000000000 0.000000000 1.147159480 1.000000000 0.000000000 1.549623546 1.000000000 0.000000000 0.673511294 1.000000000 0.000000000 1.111567420 1.000000000 0.000000000 1.609856263 1.000000000 0.000000000 0.574948665 1.000000000 0.000000000 1.051334702 1.000000000 0.000000000 1.566050650 1.000000000 0.000000000 0.558521561 1.000000000 0.000000000 1.034907598 1.000000000 0.000000000 1.555099247 1.000000000 0.000000000 0.539356605 1.000000000 0.000000000 1.040383299 1.000000000 0.000000000 1.615331964 1.000000000 0.000000000 0.572210815 1.000000000 0.000000000 1.056810404 1.000000000 0.000000000 1.552361396 1.000000000 0.000000000 0.550308008 1.000000000 0.000000000 1.125256674 1.000000000 0.000000000 1.596167009 1.000000000 0.000000000 0.514715948 1.000000000 0.000000000 1.010266940 1.000000000 0.000000000 1.585215606 1.000000000 0.000000000 0.555783710 1.000000000 0.000000000 1.054072553 1.000000000 0.000000000 1.566050650 1.000000000 0.000000000 0.632443532 1.000000000 0.000000000 1.073237509 1.000000000 0.000000000 1.585215606 1.000000000 0.000000000 0.569472964 1.000000000 0.000000000 1.067761807 1.000000000 0.000000000 1.612594114 1.000000000 0.000000000 0.566735113 1.000000000 0.000000000 1.125256674 1.000000000 0.000000000 1.607118412 1.000000000 0.000000000 0.728268309 1.000000000 0.000000000 1.095140315 1.000000000 0.000000000 1.634496920 1.000000000 0.000000000 0.550308008 1.000000000 0.000000000 1.149897331 1.000000000 0.000000000 1.508555784 1.000000000 0.000000000 0.553045859 1.000000000 0.000000000 1.029431896 1.000000000 0.000000000 1.546885695 1.000000000 0.000000000 0.547570157 1.000000000 0.000000000 1.070499658 1.000000000 0.000000000 1.549623546 1.000000000 0.000000000 0.607802875 1.000000000 0.000000000 1.007529090 1.000000000 0.000000000 1.508555784 1.000000000 0.000000000 0.528405202 1.000000000 0.000000000 1.056810404 1.000000000 0.000000000 1.549623546 1.000000000 0.000000000 0.522929500 1.000000000 0.000000000 0.544832307 1.000000000 0.000000000 1.084188912 1.000000000 0.000000000 1.519507187 1.000000000 0.000000000 0.563997262 1.000000000 0.000000000 0.999315537 1.000000000 0.000000000 1.503080082 1.000000000 0.000000000 0.536618754 1.000000000 0.000000000 1.103353867 1.000000000 0.000000000 1.601642710 1.000000000 0.000000000 0.520191650 1.000000000 0.000000000 1.056810404 1.000000000 0.000000000 1.541409993 1.000000000 0.000000000 0.522929500 1.000000000 0.000000000 1.095140315 1.000000000 0.000000000 1.697467488 1.000000000 0.000000000 0.659822040 1.000000000 0.000000000 1.122518823 1.000000000 0.000000000 1.648186174 1.000000000 0.000000000 0.536618754 1.000000000 0.000000000 1.078713210 1.000000000 0.000000000 1.557837098 1.000000000 0.000000000 0.517453799 1.000000000 0.000000000 1.125256674 1.000000000 0.000000000 1.571526352 1.000000000 0.000000000 0.520191650 1.000000000 0.000000000 1.026694045 1.000000000 0.000000000 1.609856263 1.000000000 0.000000000 0.561259411 1.000000000 0.000000000 1.059548255 1.000000000 0.000000000 1.538672142 1.000000000 0.000000000 0.536618754 1.000000000 0.000000000 1.021218344 1.000000000 0.000000000 1.560574949 1.000000000 0.000000000 0.613278576 1.000000000 0.000000000 1.092402464 1.000000000 0.000000000 1.596167009 1.000000000 0.000000000 0.525667351 1.000000000 0.000000000 1.023956194 1.000000000 0.000000000 1.738535250 1.000000000 0.000000000 0.542094456 1.000000000 0.000000000 1.021218344 1.000000000 0.000000000 1.524982888 1.000000000 0.000000000 0.577686516 1.000000000 0.000000000 1.097878166 1.000000000 0.000000000 1.557837098 1.000000000 0.000000000 0.572210815 1.000000000 0.000000000 1.013004791 1.000000000 0.000000000 1.516769336 1.000000000 0.000000000 0.544832307 1.000000000 0.000000000 0.999315537 1.000000000 0.000000000 1.555099247 1.000000000 0.000000000 0.536618754 1.000000000 0.000000000 1.002053388 1.000000000 0.000000000 1.494866530 1.000000000 0.000000000 0.555783710 1.000000000 0.000000000 1.059548255 1.000000000 0.000000000 1.593429158 1.000000000 0.000000000 0.522929500 1.000000000 0.000000000 1.062286105 1.000000000 0.000000000 1.571526352 1.000000000 0.000000000 0.580424367 1.000000000 0.000000000 1.037645448 1.000000000 0.000000000 1.670088980 1.000000000 0.000000000 0.536618754 1.000000000 0.000000000 1.073237509 1.000000000 0.000000000 1.494866530 1.000000000 0.000000000 0.539356605 1.000000000 0.000000000 1.054072553 1.000000000 0.000000000 1.533196441 1.000000000 0.000000000 0.670773443 1.000000000 0.000000000 1.004791239 1.000000000 0.000000000 0.580424367 1.059548255 3.000000000 0.985626283 1.503080082 3.000000000 0.002737851 0.539356605 3.000000000 1.062286105 1.541409993 3.000000000 0.002737851 0.501026694 3.000000000 0.002737851 0.574948665 3.000000000 0.002737851 0.583162218 3.000000000 1.037645448 1.516769336 3.000000000 0.574948665 1.114305270 3.000000000 0.002737851 0.528405202 3.000000000 survival/tests/nested.Rout.save0000644000175100001440000000227512055203536016406 0ustar hornikusers R version 2.15.2 (2012-10-26) -- "Trick or Treat" Copyright (C) 2012 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # A test of nesting. It makes sure the model.frame is built correctly > # > tfun <- function(fit, mydata) { + survfit(fit, newdata=mydata) + } > > myfit <- coxph(Surv(time, status) ~ age + factor(sex), lung) > > temp1 <- tfun(myfit, lung[1:5,]) > temp2 <- survfit(myfit, lung[1:5,]) > indx <- match('call', names(temp1)) #the call components won't match > > all.equal(unclass(temp1)[-indx], unclass(temp2)[-indx]) [1] TRUE > > > proc.time() user system elapsed 0.196 0.032 0.225 survival/tests/fr_simple.Rout.save0000644000175100001440000000615012350364343017102 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test the logic of the penalized code by fitting some no-frailty models > # (theta=0). It should give exactly the same answers as 'ordinary' coxph. > # > test1 <- data.frame(time= c(4, 3,1,1,2,2,3), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > zz <- rep(0, nrow(test1)) > tfit1 <- coxph(Surv(time,status) ~x, test1, eps=1e-7) > tfit2 <- coxph(Surv(time,status) ~x + frailty(zz, theta=0, sparse=T), test1) > tfit3 <- coxph(Surv(zz,time,status) ~x + frailty(zz, theta=0, sparse=T), test1) > > temp <- c('coefficients', 'var', 'loglik', 'linear.predictors', + 'means', 'n', 'concordance') > > all.equal(tfit1[temp], tfit2[temp]) [1] TRUE > all.equal(tfit2[temp], tfit3[temp]) [1] TRUE > > zz <- rep(0, nrow(test2)) > tfit1 <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-7) > tfit2 <- coxph(Surv(start, stop, event) ~ x + frailty(zz, theta=0, sparse=T), + test2) > all.equal(tfit1[temp], tfit2[temp]) [1] TRUE > > > # > # Repeat the above tests, but with a strata added > # Because the data set is simply doubled, the loglik will double, > # beta is the same, variance is halved. > # > test3 <- rbind(test1, test1) > test3$x2 <- rep(1:2, rep(nrow(test1),2)) > zz <- rep(0, nrow(test3)) > tfit1 <- coxph(Surv(time,status) ~x + strata(x2), test3, eps=1e-7) > tfit2 <- coxph(Surv(time,status) ~x + frailty(zz, theta=0, sparse=T) + + strata(x2), test3) > tfit3 <- coxph(Surv(zz,time,status) ~x + frailty(zz, theta=0, sparse=T) + + strata(x2), test3) > > all.equal(tfit1[temp], tfit2[temp]) [1] TRUE > all.equal(tfit2[temp], tfit3[temp]) [1] TRUE > > > test4 <- rbind(test2, test2) > test4$x2 <- rep(1:2, rep(nrow(test2),2)) > zz <- rep(0, nrow(test4)) > tfit1 <- coxph(Surv(start, stop, event) ~x, test4, eps=1e-7) > tfit2 <- coxph(Surv(start, stop, event) ~ x + frailty(zz, theta=0, sparse=T), + test4) > all.equal(tfit1[temp], tfit2[temp]) [1] TRUE > > rm(test3, test4, tfit1, tfit2, tfit3, temp, zz) > > proc.time() user system elapsed 0.220 0.024 0.242 survival/tests/r_scale.Rout.save0000644000175100001440000000502212164374514016533 0ustar hornikusers R version 3.0.0 (2013-04-03) -- "Masked Marvel" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Verify that scale can be fixed at a value > # coefs will differ slightly due to different iteration paths > tol <- .001 > > # Intercept only models > fit1 <- survreg(Surv(time,status) ~ 1, lung) > fit2 <- survreg(Surv(time,status) ~ 1, lung, scale=fit1$scale) > all.equal(fit1$coef, fit2$coef, tolerance= tol) [1] TRUE > all.equal(fit1$loglik, fit2$loglik, tolerance= tol) [1] TRUE > > # The two robust variance matrices are not the same, since removing > # an obs has a different effect on the two models. This just > # checks for failure, not for correctness > fit3 <- survreg(Surv(time,status) ~ 1, lung, robust=TRUE) > fit4 <- survreg(Surv(time,status) ~ 1, lung, scale=fit1$scale, robust=TRUE) > > > # multiple covariates > fit1 <- survreg(Surv(time,status) ~ age + ph.karno, lung) > fit2 <- survreg(Surv(time,status) ~ age + ph.karno, lung, + scale=fit1$scale) > all.equal(fit1$coef, fit2$coef, tolerance=tol) [1] TRUE > all.equal(fit1$loglik[2], fit2$loglik[2], tolerance=tol) [1] TRUE > > fit3 <- survreg(Surv(time,status) ~ age + ph.karno, lung, robust=TRUE) > fit4 <- survreg(Surv(time,status) ~ age + ph.karno, lung, + scale=fit1$scale, robust=TRUE) > > # penalized models > fit1 <- survreg(Surv(time, status) ~ pspline(age), lung) > fit2 <- survreg(Surv(time, status) ~ pspline(age), lung, scale=fit1$scale) > all.equal(fit1$coef, fit2$coef, tolerance=tol) [1] TRUE > all.equal(fit1$loglik[2], fit2$loglik[2], tolerance=tol) [1] TRUE > > fit3 <- survreg(Surv(time,status) ~ pspline(age) + ph.karno, lung, robust=TRUE) > fit4 <- survreg(Surv(time,status) ~ pspline(age) + ph.karno, lung, + scale=fit1$scale, robust=TRUE) > > > > proc.time() user system elapsed 0.304 0.044 0.344 survival/tests/fr_rat1.R0000644000175100001440000000116412466142446015001 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # Tests using the rats data # # (Female rats, from Mantel et al, Cancer Research 37, # 3863-3868, November 77) rfit <- coxph(Surv(time,status) ~ rx + frailty(litter), rats, method='breslow', subset= (sex=='f')) names(rfit) rfit rfit$iter rfit$df rfit$history[[1]] rfit1 <- coxph(Surv(time,status) ~ rx + frailty(litter, theta=1), rats, method='breslow', subset=(sex=="f")) rfit1 rfit2 <- coxph(Surv(time,status) ~ frailty(litter), rats, subset=(sex=='f')) rfit2 survival/tests/book5.Rout.save0000644000175100001440000001577012536400461016147 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # Tests of the weighted Cox model > # This is section 1.3 of my appendix -- no yet found in any of the > # printings though, it awaits the next edition > # > # Similar data set to test1, but add weights, > # a double-death/censor tied time > # a censored last subject > # The latter two are cases covered only feebly elsewhere. > # > # The data set testw2 has the same data, but done via replication > # > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), + status= c(1,0,1,1,1,0,0,1,0), + x= c(2,0,1,1,0,1,0,1,0), + wt = c(1,2,3,4,3,2,1,2,1)) > xx <- testw1$wt > testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), + status= rep(c(1,0,1,1,1,0,0,1,0), xx), + x= rep(c(2,0,1,1,0,1,0,1,0), xx), + id= rep(1:9, xx)) > indx <- match(1:9, testw2$id) > > # Breslow estimate > byhand <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 11*beta - (log(r^2 + 11*r +7) + 10*log(11*r +5) +2*log(2*r+1)) + hazard <- c(1/(r^2 + 11*r +7), 10/(11*r +5), 2/(2*r+1)) + xbar <- c((2*r^2 + 11*r)*hazard[1], 11*r/(11*r +5), r*hazard[3]) + imat <- (4*r^2 + 11*r)*hazard[1] - xbar[1]^2 + + 10*(xbar[2] - xbar[2]^2) + 2*(xbar[3] - xbar[3]^2) + + temp <- cumsum(hazard) + risk <- c(r^2, 1,r,r,1,r,1,r,1) + expected <- risk* temp[c(1,1,2,2,2,2,2,3,3)] + + # The matrix of weights, one row per obs, one col per death + # deaths at 1,2,2,2, and 4 + riskmat <- matrix(c(1,1,1,1,1,1,1,1,1, + 0,0,1,1,1,1,1,1,1, + 0,0,1,1,1,1,1,1,1, + 0,0,1,1,1,1,1,1,1, + 0,0,0,0,0,0,0,1,1), ncol=5) + wtmat <- diag(c(r^2, 2, 3*r, 4*r, 3, 2*r, 1, 2*r, 1)) %*% riskmat + + x <- c(2,0,1,1,0,1,0,1,0) + status <- c(1,0,1,1,1,0,0,1,0) + wt <- c(1,2,3,4,3,2,1,2,1) + # Table of sums for score and Schoenfeld resids + hazmat <- riskmat %*% diag(c(1,3,4,3,2)/colSums(wtmat)) + dM <- -risk*hazmat #Expected part + dM[1,1] <- dM[1,1] +1 # deaths at time 1 + for (i in 2:4) dM[i+1, i] <- dM[i+1,i] +1 + dM[8,5] <- dM[8,5] +1 + mart <- rowSums(dM) + resid <-dM * outer(x, xbar[c(1,2,2,2,3)] ,'-') + + # Increments to the variance of the hazard + var.g <- cumsum(hazard^2/ c(1,10,2)) + var.d <- cumsum((xbar-newx)*hazard) + + list(loglik=loglik, imat=imat, hazard=hazard, xbar=xbar, + mart=c(1,0,1,1,1,0,0,1,0)-expected, expected=expected, + score=rowSums(resid), schoen=c(2,1,1,0,1) - xbar[c(1,2,2,2,3)], + varhaz=(var.g + var.d^2/imat)* exp(2*beta*newx)) + } > > aeq(byhand(0)$expected, c(1/19, 1/19, rep(103/152, 5), rep(613/456,2))) #verify [1] TRUE > > fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, + method='breslow', iter=0) > fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) > fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow') > fitb <- coxph(Surv(time, status) ~x, testw2, method='breslow') > > aeq(resid(fit0, type='mart'), (resid(fit0b, type='mart'))[indx]) [1] TRUE > aeq(resid(fit0, type='scor'), (resid(fit0b, type='scor'))[indx]) [1] TRUE > aeq(unique(resid(fit0, type='scho')), unique(resid(fit0b, type='scho'))) [1] TRUE > > truth0 <- byhand(0,pi) > aeq(fit0$loglik[1], truth0$loglik) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')) [1] TRUE > sfit <- survfit(fit0, list(x=pi), censor=FALSE) > aeq(sfit$std.err^2, truth0$var) [1] TRUE > aeq(-log(sfit$surv), cumsum(truth0$haz)) [1] TRUE > > truth <- byhand(fit$coef, .3) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')) [1] TRUE > > sfit <- survfit(fit, list(x=.3), censor=FALSE) > aeq(sfit$std.err^2, truth$var) [1] TRUE > aeq(-log(sfit$surv), (cumsum(truth$haz)* exp(fit$coef*.3))) [1] TRUE > > > fit0 Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, method = "breslow", iter = 0) coef exp(coef) se(coef) z p x 0.000 1.000 0.586 0 1 Likelihood ratio test=0 on 1 df, p=1 n= 9, number of events= 5 > summary(fit) Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, method = "breslow") n= 9, number of events= 5 coef exp(coef) se(coef) z Pr(>|z|) x 0.8596 2.3621 0.7131 1.205 0.228 exp(coef) exp(-coef) lower .95 upper .95 x 2.362 0.4233 0.5839 9.556 Concordance= 0.638 (se = 0.159 ) Rsquare= 0.171 (max possible= 0.999 ) Likelihood ratio test= 1.69 on 1 df, p=0.1932 Wald test = 1.45 on 1 df, p=0.2281 Score (logrank) test = 1.52 on 1 df, p=0.217 > resid(fit0, type='score') 1 2 3 4 5 6 1.24653740 0.03601108 0.10056700 0.10056700 -0.22180142 -0.21193300 7 8 9 0.46569858 -0.10082189 0.91014302 > resid(fit0, type='scho') 1 2 2 2 4 1.3157895 0.3125000 0.3125000 -0.6875000 0.3333333 > > resid(fit, type='score') 1 2 3 4 5 6 0.88681615 0.02497653 0.03608964 0.03608964 -0.54297652 -0.12528780 7 8 9 0.29564605 -0.09476911 0.58400064 > resid(fit, type='scho') 1 2 2 2 4 1.0368337 0.1613774 0.1613774 -0.8386226 0.1746960 > aeq(resid(fit, type='mart'), (resid(fitb, type='mart'))[indx]) [1] TRUE > aeq(resid(fit, type='scor'), (resid(fitb, type='scor'))[indx]) [1] TRUE > aeq(unique(resid(fit, type='scho')), unique(resid(fitb, type='scho'))) [1] TRUE > rr1 <- resid(fit, type='mart') > rr2 <- resid(fit, type='mart', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > rr1 <- resid(fit, type='score') > rr2 <- resid(fit, type='score', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > > proc.time() user system elapsed 0.236 0.020 0.253 survival/tests/r_lung.R0000644000175100001440000000300711732700061014713 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) lfit2 <- survreg(Surv(time, status) ~ age + ph.ecog + strata(sex), lung) lfit3 <- survreg(Surv(time, status) ~ sex + (age+ph.ecog)*strata(sex), lung) lfit4 <- survreg(Surv(time, status) ~ age + ph.ecog , lung, subset=(sex==1)) lfit5 <- survreg(Surv(time, status) ~ age + ph.ecog , lung, subset=(sex==2)) if (exists('censorReg')) { lfit1 <- censorReg(censor(time, status) ~ age + ph.ecog + strata(sex),lung) aeq(lfit4$coef, lfit1[[1]]$coef) aeq(lfit4$scale, lfit1[[1]]$scale) aeq(c(lfit4$scale, lfit5$scale), sapply(lfit1, function(x) x$scale)) } aeq(c(lfit4$scale, lfit5$scale), lfit3$scale ) # # Test out ridge regression and splines # lfit0 <- survreg(Surv(time, status) ~1, lung) lfit1 <- survreg(Surv(time, status) ~ age + ridge(ph.ecog, theta=5), lung) lfit2 <- survreg(Surv(time, status) ~ sex + ridge(age, ph.ecog, theta=1), lung) lfit3 <- survreg(Surv(time, status) ~ sex + age + ph.ecog, lung) lfit0 lfit1 lfit2 lfit3 xx <- pspline(lung$age, nterm=3, theta=.3) xx <- matrix(unclass(xx), ncol=ncol(xx)) # the raw matrix lfit4 <- survreg(Surv(time, status) ~xx, lung) lfit5 <- survreg(Surv(time, status) ~age, lung) lfit6 <- survreg(Surv(time, status)~pspline(age, df=2), lung) lfit7 <- survreg(Surv(time, status) ~ offset(lfit6$lin), lung) lfit4 lfit5 lfit6 signif(lfit7$coef,6) survival/tests/detail.Rout.save0000644000175100001440000000655711732700061016371 0ustar hornikusers R version 2.11.0 (2010-04-22) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # A short test on coxph.detail, to ensure that the computed hazard is > # equal to the theoretical value > library(survival) Loading required package: splines > aeq <- function(a,b) all.equal(as.vector(a), as.vector(b)) > > # taken from book4.R > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > byhand <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 4*beta - (log(r+1) + log(r+2) + 2*log(3*r+2) + 2*log(3*r+1) + + log(2*r +2)) + u <- 1/(r+1) + 1/(3*r+1) + 2*(1/(3*r+2) + 1/(2*r+2)) - + ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) + imat <- r*(1/(r+1)^2 + 2/(r+2)^2 + 6/(3*r+2)^2 + + 6/(3*r+1)^2 + 6/(3*r+2)^2 + 4/(2*r +2)^2) + + hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), + 1/(3*r+2), 1/(2*r +2) ) + + + # The matrix of weights, one row per obs, one col per time + # deaths at 2,3,6,7,8,9 + wtmat <- matrix(c(1,0,0,0,1, 0, 0,0,0,0, + 0,1,0,1,1, 0, 0,0,0,0, + 0,0,1,1,1, 0, 1,1,0,0, + 0,0,0,1,1, 0, 1,1,0,0, + 0,0,0,0,1, 1, 1,1,0,0, + 0,0,0,0,0, 1, 1,1,1,1, + 0,0,0,0,0,.5,.5,1,1,1), ncol=7) + wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat + + x <- c(1,0,0,1,0,1,1,1,0,0) + status <- c(1,1,1,1,1,1,1,0,0,0) + xbar <- colSums(wtmat*x)/ colSums(wtmat) + n <- length(x) + + # Table of sums for score and Schoenfeld resids + hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time + dM <- -hazmat #Expected part + for (i in 1:5) dM[i,i] <- dM[i,i] +1 #observed + dM[6:7,6:7] <- dM[6:7,6:7] +.5 # observed + mart <- rowSums(dM) + + # Table of sums for score and Schoenfeld resids + # Looks like the last table of appendix E.2.1 of the book + resid <- dM * outer(x, xbar, '-') + score <- rowSums(resid) + scho <- colSums(resid) + + # We need to add the ties back up (they are symmetric) + scho[6:7] <- rep(mean(scho[6:7]), 2) + + list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard* exp(beta*newx), + mart=mart, score=score, rmat=resid, + scho=scho) + } > > # The actual coefficient of the fit is close to zero. Using a larger > # number pushes the test harder, but it should still work without > # the init and iter arguments, i.e., for any coefficient. > fit1 <- coxph(Surv(start, stop, event) ~x, test2,init=-1, iter=0) > temp <- coxph.detail(fit1) > temp2 <- byhand(fit1$coef, fit1$means) > aeq(temp$haz, c(temp2$haz[1:5], sum(temp2$haz[6:7]))) [1] TRUE > > survival/tests/expected2.Rout.save0000644000175100001440000000233011732700061016773 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # A Cox model with a factor, followed by survexp. > # > pfit2 <- coxph(Surv(time, status > 0) ~ trt + log(bili) + + log(protime) + age + platelet + sex, data = pbc) > esurv <- survexp(~ trt, ratetable = pfit2, data = pbc) > > temp <- pbc > temp$sex2 <- factor(as.numeric(pbc$sex), levels=2:0, + labels=c("f", "m", "unknown")) > esurv2 <- survexp(~ trt, ratetable = pfit2, data = temp, + rmap=list(sex=sex2)) > > # The call components won't match, which happen to be first > all.equal(unclass(esurv)[-1], unclass(esurv2)[-1]) [1] TRUE > survival/tests/bladder.Rout.save0000644000175100001440000001423312536400417016517 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # Fit the models found in Wei et. al. > # > wfit <- coxph(Surv(stop, event) ~ (rx + size + number)* strata(enum) + + cluster(id), bladder, method='breslow') > wfit Call: coxph(formula = Surv(stop, event) ~ (rx + size + number) * strata(enum) + cluster(id), data = bladder, method = "breslow") coef exp(coef) se(coef) robust se z p rx -0.5176 0.5959 0.3158 0.3075 -1.68 0.0923 size 0.0679 1.0702 0.1012 0.0853 0.80 0.4260 number 0.2360 1.2662 0.0761 0.0721 3.27 0.0011 rx:strata(enum)enum=2 -0.1018 0.9032 0.5043 0.3265 -0.31 0.7552 rx:strata(enum)enum=3 -0.1823 0.8334 0.5579 0.3916 -0.47 0.6417 rx:strata(enum)enum=4 -0.1332 0.8753 0.6581 0.4968 -0.27 0.7887 size:strata(enum)enum=2 -0.1440 0.8659 0.1680 0.1119 -1.29 0.1981 size:strata(enum)enum=3 -0.2792 0.7564 0.2086 0.1511 -1.85 0.0647 size:strata(enum)enum=4 -0.2711 0.7626 0.2515 0.1856 -1.46 0.1442 number:strata(enum)enum=2 -0.0984 0.9063 0.1193 0.1144 -0.86 0.3895 number:strata(enum)enum=3 -0.0662 0.9360 0.1298 0.1167 -0.57 0.5708 number:strata(enum)enum=4 0.0928 1.0972 0.1466 0.1175 0.79 0.4298 Likelihood ratio test=29.4 on 12 df, p=0.00344 n= 340, number of events= 112 > > # Check the rx coefs versus Wei, et al, JASA 1989 > rx <- c(1,4,5,6) # the treatment coefs above > cmat <- diag(4); cmat[1,] <- 1; #contrast matrix > wfit$coef[rx] %*% cmat # the coefs in their paper (table 5) [,1] [,2] [,3] [,4] [1,] -0.5176209 -0.6194404 -0.6998771 -0.6507935 > t(cmat) %*% wfit$var[rx,rx] %*% cmat # var matrix (eqn 3.2) [,1] [,2] [,3] [,4] [1,] 0.09455501 0.06017669 0.05677331 0.0437777 [2,] 0.06017669 0.13242834 0.13011557 0.1160420 [3,] 0.05677331 0.13011557 0.17235879 0.1590865 [4,] 0.04377770 0.11604200 0.15908650 0.2398112 > > # Anderson-Gill fit > fita <- coxph(Surv(start, stop, event) ~ rx + size + number + cluster(id), + bladder2, method='breslow') > summary(fita) Call: coxph(formula = Surv(start, stop, event) ~ rx + size + number + cluster(id), data = bladder2, method = "breslow") n= 178, number of events= 112 coef exp(coef) se(coef) robust se z Pr(>|z|) rx -0.45979 0.63142 0.19996 0.25801 -1.782 0.07474 . size -0.04256 0.95833 0.06903 0.07555 -0.563 0.57317 number 0.17164 1.18726 0.04733 0.06131 2.799 0.00512 ** --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 exp(coef) exp(-coef) lower .95 upper .95 rx 0.6314 1.5837 0.3808 1.047 size 0.9583 1.0435 0.8264 1.111 number 1.1873 0.8423 1.0528 1.339 Concordance= 0.634 (se = 0.03 ) Rsquare= 0.09 (max possible= 0.994 ) Likelihood ratio test= 16.77 on 3 df, p=0.000787 Wald test = 11.76 on 3 df, p=0.008256 Score (logrank) test = 18.57 on 3 df, p=0.0003355, Robust = 11.44 p=0.009588 (Note: the likelihood ratio and score tests assume independence of observations within a cluster, the Wald and robust score tests do not). > > # Prentice fits. Their model 1 a and b are the same > fit1p <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, + subset=(enum==1), method='breslow') > fit2pa <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, + subset=(enum==2), method='breslow') > fit2pb <- coxph(Surv(stop-start, event) ~ rx + size + number, bladder2, + subset=(enum==2), method='breslow') > fit3pa <- coxph(Surv(stop, event) ~ rx + size + number, bladder2, + subset=(enum==3), method='breslow') > #and etc. > fit1p Call: coxph(formula = Surv(stop, event) ~ rx + size + number, data = bladder2, subset = (enum == 1), method = "breslow") coef exp(coef) se(coef) z p rx -0.5176 0.5959 0.3158 -1.64 0.1012 size 0.0679 1.0702 0.1012 0.67 0.5025 number 0.2360 1.2662 0.0761 3.10 0.0019 Likelihood ratio test=9.66 on 3 df, p=0.0216 n= 85, number of events= 47 > fit2pa Call: coxph(formula = Surv(stop, event) ~ rx + size + number, data = bladder2, subset = (enum == 2), method = "breslow") coef exp(coef) se(coef) z p rx -0.42421 0.65428 0.40220 -1.05 0.29 size -0.12503 0.88247 0.11709 -1.07 0.29 number 0.00199 1.00199 0.09376 0.02 0.98 Likelihood ratio test=2.02 on 3 df, p=0.569 n= 46, number of events= 29 > fit2pb Call: coxph(formula = Surv(stop - start, event) ~ rx + size + number, data = bladder2, subset = (enum == 2), method = "breslow") coef exp(coef) se(coef) z p rx -0.25911 0.77174 0.40511 -0.64 0.52 size -0.11636 0.89015 0.11924 -0.98 0.33 number -0.00571 0.99431 0.09667 -0.06 0.95 Likelihood ratio test=1.27 on 3 df, p=0.735 n= 46, number of events= 29 > fit3pa Call: coxph(formula = Surv(stop, event) ~ rx + size + number, data = bladder2, subset = (enum == 3), method = "breslow") coef exp(coef) se(coef) z p rx -0.8985 0.4072 0.5535 -1.62 0.10 size 0.0850 1.0887 0.2086 0.41 0.68 number -0.0172 0.9830 0.1280 -0.13 0.89 Likelihood ratio test=4.16 on 3 df, p=0.245 n= 27, number of events= 22 > rm(rx, cmat, wfit, fita, fit1p, fit2pa, fit2pb, fit3pa) > > proc.time() user system elapsed 0.204 0.044 0.243 survival/tests/infcox.Rout.save0000644000175100001440000000424012164375013016405 0ustar hornikusers R version 3.0.0 (2013-04-03) -- "Masked Marvel" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # A test to exercise the "infinity" check on 2 variables > # > test3 <- data.frame(futime=1:12, fustat=c(1,0,1,0,1,0,0,0,0,0,0,0), + x1=rep(0:1,6), x2=c(rep(0,6), rep(1,6))) > > # This will produce a warning message, which is the point of the test. > # The variance is close to singular and gives different answers > # on different machines > fit3 <- coxph(Surv(futime, fustat) ~ x1 + x2, test3, iter=25) Warning message: In fitter(X, Y, strats, offset, init, control, weights = weights, : Loglik converged before variable 1,2 ; beta may be infinite. > > all(fit3$coef < -22) [1] TRUE > all.equal(round(fit3$log, 4),c(-6.8669, -1.7918)) [1] TRUE > > # > # Actual solution > # time 1, 12 at risk, 3 each of x1/x2 = 00, 01, 10, 11 > # time 2, 10 at risk, 2, 3, 2 , 3 > # time 5, 8 at risk, 1, 3, 1, 3 > # Let r1 = exp(beta1), r2= exp(beta2) > # loglik = -log(3 + 3r1 + 3r2 + 3 r1*r2) - log(2 + 2r1 + 3r2 + 3 r1*r2) - > # log(1 + r1 + 3r2 + 3 r1*r2) > true <- function(beta) { + r1 <- exp(beta[1]) + r2 <- exp(beta[2]) + loglik <- -log(3*(1+ r1+ r2+ r1*r2)) - log(2+ 2*r1 + 3*r2 + 3*r1*r2) - + log(1 + r1 + 3*r2 + 3*r1*r2) + loglik + } > > all.equal(fit3$loglik[2], true(fit3$coef), check.attributes=FALSE) [1] TRUE > > proc.time() user system elapsed 0.216 0.020 0.233 survival/tests/data.capacitor0000644000175100001440000000606511732700061016111 0ustar hornikusers 1 300.00 0 20 2 300.00 0 20 3 300.00 0 20 4 300.00 0 20 5 300.00 0 20 6 300.00 0 20 7 300.00 0 20 8 300.00 0 20 9 300.00 0 20 10 300.00 0 20 11 300.00 0 20 12 300.00 0 20 13 300.00 0 20 14 300.00 0 20 15 300.00 0 20 16 300.00 0 20 17 300.00 0 20 18 300.00 0 20 19 300.00 0 20 20 300.00 0 20 21 300.00 0 20 22 300.00 0 20 23 300.00 0 20 24 300.00 0 20 25 300.00 0 20 26 277.33 1 26 27 187.80 1 26 28 214.28 1 26 29 12.95 1 26 30 63.10 1 26 31 271.73 1 26 32 201.28 1 26 33 179.02 1 26 34 139.37 1 26 35 136.33 1 26 36 28.41 1 26 37 300.00 0 26 38 300.00 0 26 39 300.00 0 26 40 300.00 0 26 41 300.00 0 26 42 300.00 0 26 43 300.00 0 26 44 300.00 0 26 45 300.00 0 26 46 300.00 0 26 47 300.00 0 26 48 300.00 0 26 49 300.00 0 26 50 300.00 0 26 51 300.00 0 26 52 300.00 0 26 53 300.00 0 26 54 300.00 0 26 55 300.00 0 26 56 300.00 0 26 57 300.00 0 26 58 300.00 0 26 59 300.00 0 26 60 300.00 0 26 61 300.00 0 26 62 300.00 0 26 63 300.00 0 26 64 300.00 0 26 65 300.00 0 26 66 300.00 0 26 67 300.00 0 26 68 300.00 0 26 69 300.00 0 26 70 300.00 0 26 71 300.00 0 26 72 300.00 0 26 73 300.00 0 26 74 300.00 0 26 75 300.00 0 26 76 45.85 1 29 77 220.70 1 29 78 73.87 1 29 79 91.81 1 29 80 40.69 1 29 81 108.62 1 29 82 55.73 1 29 83 10.21 1 29 84 102.64 1 29 85 257.88 1 29 86 50.41 1 29 87 164.20 1 29 88 112.15 1 29 89 300.00 0 29 90 300.00 0 29 91 300.00 0 29 92 300.00 0 29 93 300.00 0 29 94 300.00 0 29 95 300.00 0 29 96 118.37 1 32 97 17.19 1 32 98 11.51 1 32 99 4.65 1 32 100 1.95 1 32 101 149.20 1 32 102 65.79 1 32 103 5.95 1 32 104 5.72 1 32 105 10.61 1 32 106 0.68 1 32 107 3.96 1 32 108 9.56 1 32 109 172.05 1 32 110 2.81 1 32 111 2.07 1 32 112 19.98 1 32 113 84.63 1 32 114 132.52 1 32 115 156.37 1 32 116 11.81 1 32 117 20.86 1 32 118 66.33 1 32 119 21.64 1 32 120 65.90 1 32 121 14.64 1 32 122 6.26 1 32 123 94.08 1 32 124 5.45 1 32 125 15.16 1 32 survival/tests/ties.rda0000644000175100001440000001365411732700061014747 0ustar hornikusers‹mZyœNeû·Ë¥¥RHÙ²e sÆ<3ó<çœ{;‰²EöHˆÒB e©PŠH!´ ½¶HZ(’l¯-’¤H‘%ï}Ÿïuæ÷Ïo>}¦™gÎrß×õÝ®[euhT²CÉ *P¨´ýWØ~[¤ýOAû¯„ýWtHï~5´ß”±ŸYN¢Òç#,:›‚ _=´|w’»FŸY4þQ’¢/W/n]äø>o˜˜]½,éÏ^7÷aI¹Óº'×>¦H½Qâ¦ÛÿAº×Ø«õ£HŒ:zo¹ß÷Qêš[Þ}ä»b¤'°XDú‰ÔàbÁYJÕ³—Ûó )wÛY½ÈÔš3»JÇÇI»Ë_û‰{Æ®;8 ùAf˦ž#Ùù­"wy?’ðëÂâ+JSÂÝæ¹r$[ÕþÖŠM¤–µë?nR9Rî1÷|Oâà€æUË'–|ºí›š’˾®”±£ ‰_Vg]Õç Rÿö-´µñQ[W÷He~ì^¨?©B+Ö¬Û˜ï/¤6œ£¼g‚³›ô%Y/zÒoŠwLí ä7zb~‹§j‘Ø2ì÷Ú_L%QüÈéó—ÆºíåÜyÛ{“שå‰Ìß6“ÚuÕ~y±Ë–+äÙOÏ}Šr·¾`ŸôW’gÔ†‹k¨;Þ]ˆ‚A‹Ü7¤^›ÿÛ5·,"i–®Ù´Å#9õ©½ÏvÎ#½ëσWí%]yÇ·þ ï“©skç‹_.%ÿ÷@OR¸¾g¥?»Ü@y/Øå¹æ’ÝÚØmFâo¼Ÿ_Ù³O8ŸDºGÚô-LfÑ÷JºíYl×¹«]æ=U(§Ãª;õ<éfêŒ 5ìîÙ[H²xa»1)Éï™Ýµ¡Ý±›(9¢Û+ž˜CªÂùœÓíú‘*øÛgKæÝB^ô^ÛH¼lwqR'q?%ï;T}Óõ¤3P:ºÝKä}_»|ɧ[“zg}Áɸ2 Ú’à}ÑEVŸ›ÖÝ®CyWw÷’w÷÷òþ/ê4Õ®©‘£3nž´‘ÄÅW”ùäes¯}nc×&dNØË·}¼Âv9sÚR]÷¸ýGæ×¹<ðé{·v¾d÷û)‡'$i7½Påy$ï¿ý{;OPÞ=KæÝ´/#ÿþ²Û¦òkʵ uþ­ãIG·ÿœÔ Uö‘“úÆ-Ô’Õ¶üûʱ«I<’×á®ÇHwßÞ»ÉkIR±Ï¢Óú—/7ÜB¢óÐjÚI©íöχ֢섫«ÇH݉zlçʯ|YjíG{í2–_OÚ=Ý_‘Wæ“É8Er¤}‹[I|í¾¾±}õ@ןÞ~šd©R';M}ŸÄòÝ}?¾¯ùµ.ØK)÷Û¿kSбÚ=fÿ\’îqKϦ¶ýÿN©ä<ûëú9$_Ç:„7æ´^9t;ù;l™ílH“®OŸÏ9Fæ¬_ÿ]NôœÃ)ñþ^]µÑVÌ$¯¡~ÅZì¿*VÝmµæ¾2ü<â&{™Ë?‘· 8 NÙåoyÒ¾w&þ³œ¼UûN »{sþ>´K• `¥íúÂÏÚë£Îu+ö>É[vÙ ¹™ôÇmƒt#³ÆÂÛ°S$æYøØÐŽD½ÞÇkeï"5ñ Û¨_’nîêâ)’÷¾j¿^#ÕfÓ–d)Û¯mlYm>lïg—éÍË$]yfÞMbÊG æÌ®Lò;ÅmFFò‚…Ï–’¾Ínó»¶ÿO·»ƒ¼øÏýÚºN‰ÑxoƶF’oßòDæY oëá¾(5ƾîÒÒ$w¸¶»‹‚—¦WüqÕ­¤ÝòþˆXTØwj8©±è»dÔ÷ ÈÜ›±£êú©¤J­},µÚ>_©æÛf!¹Ó=@Wªìà§f=‹3û-À ÊZ¿Nýeñ u ®quw‰ö5ëÞ¹lIaáðÛ+”×ǾޒwòûZ>;¡ÎÈ#(=æ^ |ÅÈ{´¾-ø¤¾Ñ´`ÏHWD&-ú[×€’÷uÝzm±ëŽ>’gûÙ ±õöö'Y û›CŒ¯‰ÆÀï0®‹ïæa߃ïÎT!Íý©v\nøu¥LÊ«þK—·åu´e:Z)³Žôa §Ëo§„Ãû SŽtûj×#Æãâ º_I ±|õ}ÃÑ$4ö#k¡ƒ¯WÉw46ÂâPGÔ_Ê}|pa ºÏgûÌa RÛ\¿œ¥ÜY=í…¾$¹Êîιi$ìjÛ $û?p¬S¨<É.îº%I|þÃëOy$DˆçÏv0°¦4iÆg]8(« OÔBK?ÿ΢äÀç{•~½O‘ažÓõ±~2®Âþe_祀#)Æ[¶åwÇývÿQDZƾ®Bòà=®¢HuÀûh^oU™ñ΢oë•ÃHgZšon¯Ç8¨–ZØ=PŒ„D?©2nƒRÖ´5åZÉ${€wRÚ‚êNò\¿íGÌõ(Ýb©+pJF¼¶Ùò~nZ8œxäíàQu¬.'Rnû¦—#ýüI·p¤—¸Bû…Tk[.ß’¼Ö>V‰I¼?Î"ÝE27Ú2©y”t•z¶¡?¢à4ê[Tµ·™ý5©§-\œìDú…=PPîîr¿[D¤äbÇ+»)ÛÁÔØ$ùùô,à¿äŸ;ôq|&×Úö}ù2©¾À“à9‡ŸÓIGuTÇֹРJn³2å¡Âämv×ë }%›Ð3 W[œ›é~5ù¯þÙ¥ÂÀ9ÔÎÉ¢ªëÉkâ3Iò°m“ õÉÿ}”EÊŠÚJ}‚Â>ŸËúK;x[ò¦­{ô§¬†}×ß¹­HÙŸ/ ó¨ª þU±NÒ[±Ï*åäßFÒQÝW%ýt ^†õ•®ÿk(qýï£Ï|.K°‹– ¬³²fvQ⻕d–A¿š+Æœ&ùŸ•C¿j7©ÉÀ39ÙÊ”Öö=¸Å0WØÃ(ûôpt׬¯­+“û~DšùC•‡ÎTOZ¹4²åD¸Ú˜$ó™b}›¼ð«$æ“©ni}t+’ ¡OÌnÆic—ût‚tD‹›IN²0PÍò/ëó øD6v¿n@aKK÷¿¾kùɶÇ}lÿX¹˜,Aúv ×%Ÿ¥\·ü-Z½ZÓ"šÕcmQoj=ú.7˜iâ2U°>íâ>—ÀÍ,ÖužÛ¾ùÇI¶N©WÝ¶Í Éë)#]ÔœD¬G_$¯¸×¿@jêK__…; ƒHÙ:pË‘q5bþÞgì:9Â8CŠ}‰~|%;?¤ÁºÇº>7úü¯”ÌvŸëIó¼t6¦Â R±âþQQ}|F¹ãÝ´¸ß¾Î› Ÿ)¶¸ŸOËÇ[™Å~¡|•€þ®]U§ÀÉ‚ µH;X?o×s:x-ë²¥ƒŠÖ?±Ñ?p=ý*ÛEI²_*ö‹2‚‡I°^Uos/B?Åë'N_ågÔËRXë/GñºG|\Éâ ÖG¾ žQKЇYœÛսƖ²ý€ºÊ+‰œ!þ{Íïç;;ôå Ò ?P±okéðmÉGÝãõ&uÔÂE^{ܲþCA„–§Ycø[çë 9•ûõ·ÊÂóë·€Oz›ýóóÿ俯úº&ˆl˜Ýwæ5z5‹yMö†NÓ"D£“ƒIŸAý&êσ«áïso¼~.Ÿø2_?è;ÌYЋõÎÓöu[ R£ XgªúÁçª?ív•jA²3túý—ÌpåzŽîO“‡õUágewèÁÅ”8hëõ.ËW;b[Eò8x^7E}™JðS: ªŸ`fÿ*¸ßE-wáä•g]Ù¼ª_tFÇâUì¿—ð{\vÝúª›Ñê1à—šÜÍÞŒu /ôAWÈ×ü¤©JV—8Vx éJÌk'¡·ÂvŽ05Éáðƒ‚ëUœƒ.ÐOr.Q þ+`ÞÖÏ£ÏýßáƒÕŽ`Èדʱw÷KV¯@§dòzÖ§òAøUéh«—½ïWÐrµë»µä³QU‡Þø¥dèv±’ýë@ûËççwþDè+É}󯺹˜töï¦ý¤˜t¬O‡c}Ò™àÝT¤çÞ#éC_é›áûõ}ÀÉ|æE°ð*™8ßqðÖÐêéoá'ÕJ\Wõ‡‘®]&&)Áü"¿rFw%ùÜWI®¯¼˜—c>Œúsä±Ë3¿i¾Ÿã÷ÞÐ%êø7õ9êÚû¿Ïa¿jb=D$_º‘ŽtÛbRQÞU“|^çuªù:N:¸m2ƒÒ9è{Ó>@އìsBÖ¹9ÜjøPŒðþÿ k凵ºŠPîácŠO–EQ)6" W‚Ä|Ç)=¿$}§ë¶ódZÁüÈe wŠSy8ÈlÙÒ‚”Õ– ‹R­Á nxyÃæP»__E‚ÅÏEª¶;ñ[×¾|G«”âÅ’½\W³ S¬Ç]Õgæ‡mHž‚hÐÇ`î‚ñ–Ëæ¼F²@Ha‘å¸íûg¨-›6‡l<גͺáEo |۸鸩T¶ë‡(ˆDëXÒÍH„ÊAô8­È²©zùæD^D*6Eº ÀDuEx ctëcžGs‡¬Gó-H”Ò+œ B~*®e±‹¯6kÁb=âò‹:PêHú‹£Ï,šHÉj£ÂXtr³éKn¿¿³bÇ… Õ(á¸<¹šRM™ìr9Üéñ*¹uÚô@h¥9$0ÍœéEz?HBLEhd¢0ág’‹bÇa¯aPÒ߃,7µÇ¦Gƒ84Å&Lì‚©3yÖ=·'Mn½n7XÓÆ Ê`%¹Ž‡ìŠÃŸÁÞ0h+JÞYÂêZ[rv.û%Rÿ Т™%‹Ù!’ÞêR•Ãd·*2Ñš öÔ4¡ BëÜû;ø•O²™l[EYŒIŸ8‘ö&é ^¾¨}4@x¨Ÿà:¨ˆë(6ïÁ+Žø‘é›KbÀE¹¬îÌ2º‡?§X´é¡X¿´ÇÃÆŠÍðdˆ§vqØÕÓÓÜÙ|›‡bê9–¨„ݾɣ¨u^"ãÚþãŽV c=ü(Ü@r(Ĥ؋û‰Ò0—q(%oƒ(–ÛħC<*Ós ~Ã~gåuÇ"Þ‡—»ÞIº7ÄŽšŽPÈ\©ËA®¨÷fÊë1LQ,¶ãþ0lÎ ‹\½2‹jÉñË* ±+˜r¡`"$™ÂN cQšE@3—Æ?B‚ñÀ¼€a@ƒH<  ?ç] QžîÃ!I×%‡o¦'BúÛÜkgUÌ…Ú5¹K&ÙÌ]w4©9<}™Å?÷kÊiàå·fQ££Çÿ‚:°hoÃÌ…-¹N›!lMF^Ê’éY˜ËœÒÒ¼îBêßÈ|e%qËK‹±O¢/‹ýþàÀ¤&#¾XNíë°8lr\ÿ¢0L¿™œÖçð>Ië–Îv¬ü0%ž„©oyæFZ½åô…XWï üÐX'³Ú¥Dü_˜Ã&)X‘©Øü{lÒD&BðÔ„èúm˜‚$›]Åx›?ÁÃIÁCõÄ»ìÎáÃùmÎÆ“:±“çàföÒ×a4ó¢â0NñzÏCü$Û`Xrèn¾ÆúÇ&FæÂôæÕ†(ówÁÌç c‘DaUЇ²ò a^dœÌ}Îü#¹®Ã>àG=Ô±åŸx8v7ÖW_@ÝäÎb2|äñPYFh™í2³ÄdjÂø溺“k¾Ãp0Ÿo{¸ £ ™).M)C>‡â憱áçéigòq" ů£4‡m!oĵ<¬y"×<‹á­ŒPÍã0ÕŸ‹õÍæB^Ý`XT«Íx/bè¦Z€w•ƒÙœ»ó¯‹gÉaOBøi„ ŠÍ¥a¢>aS;êà÷y‘,hJA~sˆ©ê¢ßƒä˜³4ã·ßÌéŒÇ­É>nÝ…gM †Ð±Ž4l–²wrí,……Øl­sxS‚ô ð‹Z~4ݰžê"ï{Î?@!÷‰ù|ãPò†Dâ=¼‡~’ã öUI§Úç’ä>›‡6áÑaQ/5׎CæÁf.‡y_vƒŽ Yß'Û>]Hf/Bp]z*7 ƒj“b“gxˆ,^/çi–ï!Éýrئ>~š›Ñïj3ˆë’à é¼Eóo)}=ô’ž “ž™H—m ÏÑêãCIpHî±)Œß?ÉæBl´^¨Z&µ ÝÆäE'\6wø ¸Ÿ%9u/ü\å ~5‡Þ²Â.ã5ÿ 3L èðpmªb¸-ZF:y y—à+ãÈú̸Yö.cëºRÝÊþa×Q8›ü°t´‰õûÃf\³Ù5#€3â<ãí/MQˆä“⡬لÐCðp3ÖkŠõ¯äPPÍÂ{i«LÚ2oÅà ӓù‚ýF‡V(ÆÁ'Ó¬[Ãö_³é 9TL{Å:9j)À}Çz_ð¡ɸ¢x(ü6b¨­'!Ì÷¢ýi’'€ÏIö‚uVÌ»JpÅx¦ßu\LzÖYrý‹q¸Ä/…ýJÆ¡Æ]¸¿ì ³l8¤Q=`š -óšc(ŸMðq<„ÎåúQ›ÑWq˜'ö3¯òá É>Ód7ôZ„SÙ\OŠÿßpˆfxª>®ë™8,a¢®DK Sò¸Nõ!„.mòb~ ü£w#ŒÌå°ÕŒdý¸ue¦p˜ó:ã•Wá£äCï³àpÃÔGˆeæcدÀ¿hZäñpG®ãC|x)U}¯9\4ç¡[Ã4†¼ð—cø£rÈ\aâ>/èáÐ)â|Ÿ)=öý†CP!¯æºˆ‡¾ŠÙ¨+Ð]úwø­ì|ÈbtgÈþÏgýªX‡š‡ÑOÁ2üÜÌǺ«¥È b¨YŸé,ø>ñ/뺫iÐãа/¼*Ò |6±?|Œýò|ä†õ£dÝ`zc¸¬ 4×»ùÅÂÞo‘(M'H³OJ„®¾ï#ɹF*ÂûAäÇøÒa~¬Ã5Þ1òáÉ<ª¯5óˆŠóö·)Þ¯€Q¥»cÝõõV³Ï6p$Õ|//@o¨ÑÐGš‡À†u‡áázZëŨ±8+ëÁ‡{<”бz•‡kŒ«ê!èg³“‡hŒSŠûG±?Mp-øÐ’a\ŠŸ_ñá@}ṎpýèŸá—÷“d~4¬tš92oľ38Î:|9ûÆðßzÂ|YŸË<̇“8ï‘ë8ØŒC ê= ¿ò‡7ìõ(„Ô¢²…äóýë+9x* XG+>T%ΠÕ0ä0ЇnÆ]þï#¤‹Ã—Çè4ëÊÔÜOŽŽÄz4Öw’ÅøFšÝÁýxXÂx©gãpšæÃ’q"—M©òÐ#¢ö+µûä\D¾ÂyÙä/‚‡$I" öÙ!«$׿æ!šPȽM®p;S.gÒ7‚'$÷“©œÖ;0lÑçaãhCÈć‹Ö ÞÛó!¬qXñÐTòP$Íáv..ÓM8'ÃÏ.õ¿ÿ\ÍR¹q+survival/tests/coxsurv2.Rout.save0000644000175100001440000000562611732700061016716 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Check that the survival curves from a Cox model with beta=0 > # match ordinary survival > # > # Aalen > surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fleming', + error='tsiatis') > fit1 <- coxph(Surv(time, status) ~ age + strata(sex), data=lung, iter=0, + method='breslow') > fit1$var <- 0*fit1$var #sneaky, causes the extra term in the Cox variance > # calculation to be zero > surv2 <- survfit(fit1, type='aalen', vartype='tsiatis') > surv3 <- survfit(fit1) > > arglist <- c('n', 'time', 'n.risk','n.event', 'n.censor', 'surv', 'strata', + 'std.err', 'upper', 'lower') > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > all.equal(unclass(surv1)[arglist], unclass(surv3)[arglist]) [1] TRUE > > > # Efron method > surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fh2', + error='tsiatis') > surv2 <- survfit(fit1, type='efron', vartype='efron') > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > > # Kaplan-Meier > surv1 <- survfit(Surv(time,status) ~ sex, data=lung) > surv2 <- survfit(fit1, type='kalb', vartype='green') > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > > > # Now add some random weights > rwt <- runif(nrow(lung), .5, 3) > surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fleming', + error='tsiatis', weight=rwt) > fit1 <- coxph(Surv(time, status) ~ age + strata(sex), data=lung, iter=0, + method='breslow', weight=rwt) > fit1$var <- 0*fit1$var #sneaky > surv2 <- survfit(fit1, type='aalen', vartype='tsiatis') > surv3 <- survfit(fit1) > > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > all.equal(unclass(surv1)[arglist], unclass(surv3)[arglist]) [1] TRUE > > > # Efron method > surv1 <- survfit(Surv(time,status) ~ sex, data=lung, type='fh2', + error='tsiatis', weight=rwt) > surv2 <- survfit(fit1, type='efron', vartype='efron') > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > > # Kaplan-Meier > surv1 <- survfit(Surv(time,status) ~ sex, data=lung, weight=rwt) > surv2 <- survfit(fit1, type='kalb', vartype='green') > all.equal(unclass(surv1)[arglist], unclass(surv2)[arglist]) [1] TRUE > > survival/tests/surv.Rout.save0000644000175100001440000000424412257335007016124 0ustar hornikusers R version 3.0.1 (2013-05-16) -- "Good Sport" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > library(survival) Loading required package: splines > > # Some simple tests of the Surv function > # The first two are motivated by a bug, pointed out by Kevin Buhr, > # where a mixture of NAs and invalid values didn't work right > # Even for the simplest things a test case is good. > # All but the third should produce warning messages > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > temp <- Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), c(1,1,1,1)) Warning message: In Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), c(1, 1, 1, 1)) : Stop time must be > start time, NA created > aeq(temp, c(1,10,NA,30, 2,NA,0,40, 1,1,1,1)) [1] TRUE > > temp <- Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), type='interval2') Warning message: In Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), type = "interval2") : Invalid interval: start > stop, NA created > aeq(temp, c(1,10,20,30, 2,1,1,40, 3,0,NA,3)) [1] TRUE > > #No error > temp <- Surv(1:5) > aeq(temp, c(1:5, 1,1,1,1,1)) [1] TRUE > > temp1 <- Surv(c(1,10,NA, 30, 30), c(1,NA,10,20, 40), type='interval2') Warning message: In Surv(c(1, 10, NA, 30, 30), c(1, NA, 10, 20, 40), type = "interval2") : Invalid interval: start > stop, NA created > temp2 <- Surv(c(1,10,10,30,30), c(9, NA, 5, 20,40), c(1, 0, 2,3,3), + type='interval') Warning message: In Surv(c(1, 10, 10, 30, 30), c(9, NA, 5, 20, 40), c(1, 0, 2, 3, : Invalid interval: start > stop, NA created > aeq(temp1, temp2) [1] TRUE > aeq(temp1, c(1,10,10,30,30, 1,1,1,1, 40, 1,0,2,NA,3)) [1] TRUE > > proc.time() user system elapsed 0.292 0.052 0.336 survival/tests/doaml.Rout.save0000644000175100001440000001742312536400546016225 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > # > # These results can be found in Miller > # > fit <- coxph(Surv(aml$time, aml$status) ~ aml$x, method='breslow') > fit Call: coxph(formula = Surv(aml$time, aml$status) ~ aml$x, method = "breslow") coef exp(coef) se(coef) z p aml$xNonmaintained 0.904 2.470 0.512 1.77 0.078 Likelihood ratio test=3.3 on 1 df, p=0.0694 n= 23, number of events= 18 > resid(fit, type='mart') 1 2 3 4 5 6 0.86225539 0.79200985 -0.20799015 0.74818869 0.65652976 -0.39796610 7 8 9 10 11 12 0.45424957 0.25475051 -1.05400917 -0.55400917 -1.55400917 0.87844483 13 14 15 16 17 18 0.87844483 0.74006941 0.74006941 0.57677292 -0.51373647 0.15162716 19 20 21 22 23 0.01702219 -0.14897252 -0.56448258 -1.15185244 -1.60340676 > resid(fit, type='score') 1 2 3 4 5 6 -0.546856248 -0.492501830 0.141063944 -0.479907930 -0.447416819 0.268453990 7 8 9 10 11 12 -0.235908976 -0.072655945 0.640826596 0.640826596 0.640826596 0.237767767 13 14 15 16 17 18 0.237767767 0.232585063 0.232585063 0.203878910 -0.165307985 0.044923326 19 20 21 22 23 0.007079721 -0.039651990 -0.181184547 -0.395076175 -0.472116894 > resid(fit, type='scho') 5 5 8 8 9 12 13 0.2706690 0.2706690 0.3081229 0.3081229 -0.6423931 0.3360212 -0.6335658 18 23 23 27 30 31 33 -0.6494307 -0.6791937 0.3208063 0.3269751 0.3360212 -0.5970995 0.3505693 34 43 45 48 -0.5525731 0.3778334 0.5484457 0.0000000 > > # Test the drop of an itercept: should have no effect > fit2 <- coxph(Surv(time, status) ~ x -1, method='breslow', + data=aml) > aeq(fit$loglik, fit2$loglik) [1] TRUE > aeq(coef(fit), coef(fit2)) [1] TRUE > aeq(fit$var, fit2$var) [1] TRUE > > fit <- survfit(Surv(aml$time, aml$status) ~ aml$x) > fit Call: survfit(formula = Surv(aml$time, aml$status) ~ aml$x) n events median 0.95LCL 0.95UCL aml$x=Maintained 11 7 31 18 NA aml$x=Nonmaintained 12 11 23 8 NA > summary(fit) Call: survfit(formula = Surv(aml$time, aml$status) ~ aml$x) aml$x=Maintained time n.risk n.event survival std.err lower 95% CI upper 95% CI 9 11 1 0.909 0.0867 0.7541 1.000 13 10 1 0.818 0.1163 0.6192 1.000 18 8 1 0.716 0.1397 0.4884 1.000 23 7 1 0.614 0.1526 0.3769 0.999 31 5 1 0.491 0.1642 0.2549 0.946 34 4 1 0.368 0.1627 0.1549 0.875 48 2 1 0.184 0.1535 0.0359 0.944 aml$x=Nonmaintained time n.risk n.event survival std.err lower 95% CI upper 95% CI 5 12 2 0.8333 0.1076 0.6470 1.000 8 10 2 0.6667 0.1361 0.4468 0.995 12 8 1 0.5833 0.1423 0.3616 0.941 23 6 1 0.4861 0.1481 0.2675 0.883 27 5 1 0.3889 0.1470 0.1854 0.816 30 4 1 0.2917 0.1387 0.1148 0.741 33 3 1 0.1944 0.1219 0.0569 0.664 43 2 1 0.0972 0.0919 0.0153 0.620 45 1 1 0.0000 NaN NA NA > survdiff(Surv(aml$time, aml$status)~ aml$x) Call: survdiff(formula = Surv(aml$time, aml$status) ~ aml$x) N Observed Expected (O-E)^2/E (O-E)^2/V aml$x=Maintained 11 7 10.69 1.27 3.4 aml$x=Nonmaintained 12 11 7.31 1.86 3.4 Chisq= 3.4 on 1 degrees of freedom, p= 0.0653 > > # > # Test out the weighted K-M > # > # First, equal case weights- shouldn't change the survival, but will > # halve the variance > temp2 <-survfit(Surv(aml$time, aml$status)~1, type='kaplan', weight=rep(2,23)) > temp <-survfit(Surv(time, status)~1, aml) > aeq(temp$surv, temp2$surv) [1] TRUE > aeq(temp$std.err^2, 2*temp2$std.err^2) [1] TRUE > > # Risk weights-- use a null Cox model > tfit <- coxph(Surv(aml$time, aml$status) ~ offset(log(1:23))) > sfit <- survfit(tfit, type='aalen', censor=FALSE) > > # Now compute it by hand. The survfit program will produce a curve > # corresponding to the mean offset. This is a change on 7/2010, > # which caused S(new) = S(old)^exp(mean(log(1:23))). > # Ties are a nuisance > rscore <- exp(log(1:23) - mean(log(1:23)))[order(aml$time)] > atime <- sort(aml$time) > denom <- rev(cumsum(rev(rscore))) > denom <- denom[match(unique(atime), atime)] > deaths <- tapply(aml$status, aml$time, sum) > chaz <- cumsum(deaths/denom) > all.equal(sfit$surv, as.vector(exp(-chaz[deaths>0]))) [1] TRUE > cvar <- cumsum(deaths/denom^2) > all.equal(sfit$std^2, as.vector(cvar[deaths>0])) [1] TRUE > > # And the Efron result > summary(survfit(tfit)) Call: survfit(formula = tfit) time n.risk n.event survival std.err lower 95% CI upper 95% CI 5 23 2 0.932 0.0461 0.8463 1.000 8 21 2 0.863 0.0637 0.7467 0.997 9 19 1 0.827 0.0704 0.6999 0.977 12 18 1 0.793 0.0755 0.6576 0.955 13 17 1 0.757 0.0801 0.6152 0.931 18 14 1 0.719 0.0846 0.5709 0.905 23 13 2 0.645 0.0907 0.4893 0.849 27 11 1 0.607 0.0929 0.4496 0.819 30 9 1 0.565 0.0955 0.4054 0.787 31 8 1 0.519 0.0982 0.3579 0.752 33 7 1 0.474 0.0994 0.3140 0.715 34 6 1 0.423 0.1009 0.2649 0.675 43 5 1 0.373 0.1006 0.2198 0.633 45 4 1 0.312 0.1009 0.1657 0.588 48 2 1 0.199 0.1102 0.0674 0.589 > > # Lots of ties, so its a good test case > x1 <- coxph(Surv(time, status)~x, aml, method='efron') > x1 Call: coxph(formula = Surv(time, status) ~ x, data = aml, method = "efron") coef exp(coef) se(coef) z p xNonmaintained 0.916 2.498 0.512 1.79 0.074 Likelihood ratio test=3.38 on 1 df, p=0.0658 n= 23, number of events= 18 > x2 <- coxph(Surv(rep(0,23),time, status) ~x, aml, method='efron') > aeq(x1$coef, x2$coef) [1] TRUE > > > rm(x1, x2, atime, denom, deaths, chaz,cvar, tfit, sfit, temp, temp2, fit) > > proc.time() user system elapsed 0.244 0.012 0.250 survival/tests/r_resid.Rout.save0000644000175100001440000003557511732700061016560 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > fit1 <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian) > fit4 <- survreg(Surv(log(futime), fustat) ~age + ecog.ps, ovarian, + dist='extreme') > > print(fit1) Call: survreg(formula = Surv(futime, fustat) ~ age + ecog.ps, data = ovarian) Coefficients: (Intercept) age ecog.ps 12.28496723 -0.09702669 0.09977342 Scale= 0.6032744 Loglik(model)= -90 Loglik(intercept only)= -98 Chisq= 15.98 on 2 degrees of freedom, p= 0.00034 n= 26 > summary(fit4) Call: survreg(formula = Surv(log(futime), fustat) ~ age + ecog.ps, data = ovarian, dist = "extreme") Value Std. Error z p (Intercept) 12.2850 1.5015 8.182 2.80e-16 age -0.0970 0.0235 -4.127 3.67e-05 ecog.ps 0.0998 0.3657 0.273 7.85e-01 Log(scale) -0.5054 0.2351 -2.149 3.16e-02 Scale= 0.603 Extreme value distribution Loglik(model)= -21.8 Loglik(intercept only)= -29.8 Chisq= 15.98 on 2 degrees of freedom, p= 0.00034 Number of Newton-Raphson Iterations: 5 n= 26 > > > # Hypothesis (and I'm fairly sure): censorReg shares the fault of many > # iterative codes -- it returns the loglik and variance for iteration k > # but the coef vector of iteration k+1. Hence the "all.equal" tests > # below don't come out perfect. > # > if (exists('censorReg')) { #true for Splus, not R + fit2 <- censorReg(censor(futime, fustat) ~ age + ecog.ps, ovarian) + fit3 <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, + iter=0, init=c(fit2$coef, log(fit2$scale))) + + aeq(resid(fit2, type='working')[,1], resid(fit3, type='working')) + aeq(resid(fit2, type='response')[,1], resid(fit3, type='response')) + + temp <- sign(resid(fit3, type='working')) + aeq(resid(fit2, type='deviance')[,1], + temp*abs(resid(fit3, type='deviance'))) + aeq(resid(fit2, type='deviance')[,1], resid(fit3, type='deviance')) + } > # > # Now check fit1 and fit4, which should follow identical iteration paths > # These tests should all be true > # > aeq(fit1$coef, fit4$coef) [1] TRUE > > resid(fit1, type='working') 1 2 3 4 5 6 -4.5081778 -0.5909810 -2.4878519 0.6032744 -5.8993431 0.6032744 7 8 9 10 11 12 -1.7462937 -0.8102883 0.6032744 -1.6593962 -0.8235265 0.6032744 13 14 15 16 17 18 0.6032744 0.6032744 0.6032744 0.6032744 0.6032744 0.6032744 19 20 21 22 23 24 0.6032744 0.6032744 0.6032744 0.2572623 -31.8006867 -0.7426277 25 26 -0.2857597 0.6032744 > resid(fit1, type='response') 1 2 3 4 5 6 -155.14523 -58.62744 -262.03173 -927.79842 -1377.84908 -658.86626 7 8 9 10 11 12 -589.74449 -318.93436 4.50671 -686.83338 -434.39281 -1105.68733 13 14 15 16 17 18 -42.43371 -173.09223 -4491.29974 -3170.49394 -5028.31053 -2050.91373 19 20 21 22 23 24 -150.65033 -2074.09345 412.32400 76.35826 -3309.40331 -219.81579 25 26 -96.19691 -457.76731 > resid(fit1, type='deviance') 1 2 3 4 5 6 7 -1.5842290 -0.6132746 -1.2876971 0.5387840 -1.7148539 0.6682580 -1.1102921 8 9 10 11 12 13 14 -0.7460191 1.4253843 -1.0849419 -0.7531720 0.6648130 1.3526380 1.1954382 15 16 17 18 19 20 21 0.2962391 0.3916044 0.3278067 0.5929057 1.2747643 0.6171130 1.9857606 22 23 24 25 26 0.6125492 -2.4504208 -0.7080652 -0.3642424 0.7317955 > resid(fit1, type='dfbeta') (Intercept) age ecog.ps Log(scale) 1 0.43370970 -1.087867e-02 0.126322520 0.048379059 2 0.14426449 -5.144770e-03 0.088768478 -0.033939677 3 0.25768057 -3.066698e-03 -0.066578834 0.021817646 4 0.05772598 -5.068044e-04 -0.013121427 -0.007762466 5 -0.58773456 6.676156e-03 0.084189274 0.008064026 6 0.01499533 -7.881949e-04 0.026570173 -0.013513160 7 -0.17869321 4.126121e-03 -0.072760519 -0.015006956 8 -0.11851540 2.520303e-03 -0.045549628 -0.035686269 9 0.08327656 3.206404e-03 -0.141835350 0.024490806 10 -0.25083921 5.321702e-03 -0.073986269 -0.020648720 11 -0.21333934 4.155746e-03 -0.049832434 -0.040215681 12 0.13889770 -1.586136e-03 -0.019701151 -0.004686340 13 0.07892133 -2.706713e-03 0.085242459 0.007847879 14 0.29690157 -1.987141e-03 -0.085553120 0.017447343 15 0.04344618 -6.319243e-04 -0.001944285 -0.003533279 16 0.04866809 -1.068317e-03 0.012398602 -0.006340983 17 0.04368104 -9.248316e-04 0.009428718 -0.004869178 18 0.15684611 -2.081485e-03 -0.013068320 -0.003265399 19 0.48839511 -4.775829e-03 -0.093258090 0.032703354 20 0.17598922 -2.349254e-03 -0.014202966 -0.002486428 21 0.37869758 -8.442011e-03 0.163476417 0.100850775 22 -0.59761427 8.803638e-03 0.052784598 -0.053085234 23 -0.79017984 1.092304e-02 0.053690092 0.080780399 24 -0.02348526 8.331002e-04 -0.039028433 -0.032765737 25 -0.13948485 3.687927e-04 0.056781884 -0.055647859 26 0.05778937 3.766350e-06 -0.029232389 -0.008927920 > resid(fit1, type='dfbetas') [,1] [,2] [,3] [,4] 1 0.288846658 -0.4627232074 0.345395116 0.20574292 2 0.096078819 -0.2188323823 0.242713641 -0.14433617 3 0.171612884 -0.1304417700 -0.182041999 0.09278449 4 0.038444974 -0.0215568869 -0.035877029 -0.03301165 5 -0.391425795 0.2839697749 0.230193032 0.03429410 6 0.009986751 -0.0335258093 0.072649027 -0.05746778 7 -0.119008027 0.1755042532 -0.198944162 -0.06382048 8 -0.078930164 0.1072008799 -0.124543264 -0.15176395 9 0.055461420 0.1363841532 -0.387810796 0.10415271 10 -0.167056601 0.2263581990 -0.202295647 -0.08781336 11 -0.142082031 0.1767643342 -0.136253451 -0.17102630 12 0.092504589 -0.0674661531 -0.053867524 -0.01992972 13 0.052560878 -0.1151298322 0.233072686 0.03337488 14 0.197733705 -0.0845228882 -0.233922105 0.07419878 15 0.028934753 -0.0268788526 -0.005316126 -0.01502607 16 0.032412497 -0.0454407662 0.033900659 -0.02696647 17 0.029091172 -0.0393376416 0.025780305 -0.02070728 18 0.104458066 -0.0885357994 -0.035731824 -0.01388685 19 0.325266641 -0.2031395176 -0.254989284 0.13907843 20 0.117207199 -0.0999253459 -0.038834208 -0.01057410 21 0.252209096 -0.3590802699 0.446982501 0.42889079 22 -0.398005596 0.3744620571 0.144325354 -0.22575700 23 -0.526252483 0.4646108448 0.146801184 0.34353696 24 -0.015640965 0.0354358527 -0.106712804 -0.13934372 25 -0.092895624 0.0156865706 0.155254862 -0.23665514 26 0.038487186 0.0001602014 -0.079928144 -0.03796800 > resid(fit1, type='ldcase') 1 2 3 4 5 6 0.374432175 0.145690278 0.112678800 0.006399163 0.261176992 0.013280058 7 8 9 10 11 12 0.109842490 0.074103234 0.248285282 0.128482147 0.094038203 0.016111951 13 14 15 16 17 18 0.132812463 0.111857574 0.001698300 0.004730718 0.003131173 0.015840667 19 20 21 22 23 24 0.179925399 0.019071941 0.797119488 0.233096445 0.666613755 0.062959708 25 26 0.080117437 0.015922378 > resid(fit1, type='ldresp') 1 2 3 4 5 6 0.076910173 0.173810883 0.078356928 0.005310644 0.060742612 0.010002154 7 8 9 10 11 12 0.067356838 0.067065693 0.355103899 0.067043195 0.068142828 0.016740944 13 14 15 16 17 18 0.193444572 0.165021262 0.001494685 0.004083386 0.002767560 0.016400993 19 20 21 22 23 24 0.269571809 0.020129806 1.409736499 1.040266083 0.058637282 0.071819025 25 26 0.112702844 0.015105534 > resid(fit1, type='ldshape') 1 2 3 4 5 6 0.870628250 0.383362440 0.412503605 0.005534970 0.513991064 0.003310847 7 8 9 10 11 12 0.291860593 0.154910362 0.256160646 0.312329770 0.183191309 0.004184904 13 14 15 16 17 18 0.110215710 0.049299495 0.007678445 0.011633336 0.011588605 0.008641251 19 20 21 22 23 24 0.112967758 0.008271358 2.246729275 0.966929220 1.022043272 0.143857170 25 26 0.079754096 0.001606647 > resid(fit1, type='matrix') g dg ddg ds dds dsg 1 -1.74950763 -1.46198129 -0.32429540 0.88466493 -2.42358635 1.8800360 2 -0.68266980 -0.82027857 -1.38799493 -0.66206188 -0.57351872 1.3921043 3 -1.32369884 -1.33411374 -0.53625126 0.31503768 -1.83606321 1.8626973 4 -0.14514412 0.24059386 -0.39881329 -0.28013223 -0.26053084 0.2237590 5 -1.96497889 -1.50383619 -0.25491587 1.15700933 -2.68145423 1.8694717 6 -0.22328436 0.37012071 -0.61351964 -0.33477229 -0.16715487 0.1848047 7 -1.11099124 -1.23201028 -0.70550005 0.01052036 -1.48515401 1.8106760 8 -0.77288913 -0.95018808 -1.17265428 -0.51190170 -0.79753045 1.5525642 9 -1.01586016 1.68391053 -2.79128447 0.01598527 -0.01623681 -1.7104080 10 -1.08316634 -1.21566480 -0.73259465 -0.03052447 -1.43539383 1.7998987 11 -0.77825093 -0.95675178 -1.16177415 -0.50314979 -0.81016011 1.5600720 12 -0.22098818 0.36631452 -0.60721042 -0.33361394 -0.17002503 0.1866908 13 -0.91481479 1.51641567 -2.51364157 -0.08144930 0.07419757 -1.3814037 14 -0.71453621 1.18442981 -1.96333502 -0.24017106 0.15944438 -0.7863174 15 -0.04387880 0.07273440 -0.12056602 -0.13717935 -0.29168773 0.1546569 16 -0.07667699 0.12710134 -0.21068577 -0.19691828 -0.30879813 0.1993144 17 -0.05372862 0.08906165 -0.14763041 -0.15709224 -0.30221555 0.1713377 18 -0.17576861 0.29135764 -0.48296037 -0.30558900 -0.22570402 0.2151929 19 -0.81251205 1.34683655 -2.23254376 -0.16869744 0.13367171 -1.0672002 20 -0.19041424 0.31563454 -0.52320225 -0.31581218 -0.20797917 0.2078622 21 -1.97162252 3.26820173 -5.41743790 1.33844939 -2.24706488 -5.4868428 22 -0.68222519 1.23245193 -4.79064290 -0.58668577 -0.95209805 -2.8390386 23 -3.49689798 -1.62675999 -0.05115487 2.90949868 -4.20494743 1.7496975 24 -0.74529506 -0.91462436 -1.23160543 -0.55723389 -0.73139169 1.5108398 25 -0.56095318 -0.53280415 -1.86451840 -0.87536233 -0.22666819 0.9689667 26 -0.26776235 0.44384834 -0.73573207 -0.35281852 -0.11207472 0.1409908 > > aeq(resid(fit1, type='working'),resid(fit4, type='working')) [1] TRUE > #aeq(resid(fit1, type='response'), resid(fit4, type='response'))#should differ > aeq(resid(fit1, type='deviance'), resid(fit4, type='deviance')) [1] TRUE > aeq(resid(fit1, type='dfbeta'), resid(fit4, type='dfbeta')) [1] TRUE > aeq(resid(fit1, type='dfbetas'), resid(fit4, type='dfbetas')) [1] TRUE > aeq(resid(fit1, type='ldcase'), resid(fit4, type='ldcase')) [1] TRUE > aeq(resid(fit1, type='ldresp'), resid(fit4, type='ldresp')) [1] TRUE > aeq(resid(fit1, type='ldshape'), resid(fit4, type='ldshape')) [1] TRUE > aeq(resid(fit1, type='matrix'), resid(fit4, type='matrix')) [1] TRUE > # > # Some tests of the quantile residuals > # > motor <- read.table('data.motor', col.names=c('temp', 'time', 'status')) > > # These should agree exactly with Ripley and Venables' book > fit1 <- survreg(Surv(time, status) ~ temp, data=motor) > summary(fit1) Call: survreg(formula = Surv(time, status) ~ temp, data = motor) Value Std. Error z p (Intercept) 16.3185 0.62296 26.2 3.03e-151 temp -0.0453 0.00319 -14.2 6.74e-46 Log(scale) -1.0956 0.21480 -5.1 3.38e-07 Scale= 0.334 Weibull distribution Loglik(model)= -147.4 Loglik(intercept only)= -169.5 Chisq= 44.32 on 1 degrees of freedom, p= 2.8e-11 Number of Newton-Raphson Iterations: 7 n= 40 > > # > # The first prediction has the SE that I think is correct > # The third is the se found in an early draft of Ripley; fit1 ignoring > # the variation in scale estimate, except via it's impact on the > # upper left corner of the inverse information matrix. > # Numbers 1 and 3 differ little for this dataset > # > predict(fit1, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) $fit [1] 10.306068 9.676248 $se.fit [1] 0.2135247 0.2202088 > > fit2 <- survreg(Surv(time, status) ~ temp, data=motor, scale=fit1$scale) > predict(fit2, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) $fit [1] 10.306068 9.676248 $se.fit 1 1 0.2057964 0.2057964 > > fit3 <- fit2 > fit3$var <- fit1$var[1:2,1:2] > predict(fit3, data.frame(temp=130), type='uquantile', p=c(.5, .1), se=T) $fit [1] 10.306068 9.676248 $se.fit 1 1 0.2219959 0.2219959 > > pp <- seq(.05, .7, length=40) > xx <- predict(fit1, data.frame(temp=130), type='uquantile', se=T, + p=pp) > #matplot(pp, cbind(xx$fit, xx$fit+2*xx$se, xx$fit - 2*xx$se), type='l') > > > # > # Now try out the various combinations of strata, #predicted, and > # number of quantiles desired > # > fit1 <- survreg(Surv(time, status) ~ inst + strata(inst) + age + sex, lung) > qq1 <- predict(fit1, type='quantile', p=.3, se=T) > qq2 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T) > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > aeq(qq1$fit, qq2$fit[,2]) [1] TRUE > aeq(qq1$se.fit, qq2$se.fit[,2]) [1] TRUE > > qq3 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, + newdata= lung[1:5,]) > aeq(qq3$fit, qq2$fit[1:5,]) [1] TRUE > > qq4 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, newdata=lung[7,]) > aeq(qq4$fit, qq2$fit[7,]) [1] TRUE > > qq5 <- predict(fit1, type='quantile', p=c(.2, .3, .4), se=T, newdata=lung) > aeq(qq2$fit, qq5$fit) [1] TRUE > aeq(qq2$se.fit, qq5$se.fit) [1] TRUE > survival/tests/pspline.Rout.save0000644000175100001440000000557312350364303016600 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Tests with the pspline function, to verify the prediction aspects > # > options(na.action=na.exclude) > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > spfit <- coxph(Surv(time, status) ~ pspline(age) + ph.ecog, lung) > > spfit2 <- coxph(Surv(time, status) ~ pspline(age) + ph.ecog, lung, x=TRUE) > x2 <- model.matrix(spfit) > all.equal(spfit2$x, x2) [1] TRUE > > keep <- (lung$age < 60) > x3 <- model.matrix(spfit, data=lung[keep,]) > attr(x3, 'assign') <- NULL #subscripting loses the assign attr below > all.equal(napredict(spfit$na.action,x2)[keep,], x3) [1] TRUE > > p2 <- predict(spfit, newdata=lung[keep,]) > aeq(p2, predict(spfit)[keep]) [1] TRUE > > > p3 <- survfit(spfit) > p4 <- survfit(spfit, newdata=lung[1:2,]) > temp <- scale(x2[1:2,], center=spfit$means, scale=FALSE)%*% coef(spfit) > aeq(p3$time, p4$time) [1] TRUE > aeq(outer(-log(p3$surv), exp(temp), '*'), -log(p4$surv)) [1] TRUE > > # Check out model.frame > spfit3 <- coxph(Surv(time, status) ~ pspline(age) + sex, lung, + model=TRUE) #avoid the missing value > m2 <- model.frame(spfit3, data=lung[keep,]) > all.equal(m2, spfit3$model[keep,]) [1] TRUE > > # > # Test of residuals, in response to a reported bug. The routines for > # m-resids of penalized models were separate from other m-resid calcs; > # refactored to change that. > # These are three progam paths that should all lead to the same C routine > fit <- coxph(Surv(tstart, tstop, status) ~ sex + treat + pspline(age), cgd) > fit2 <- coxph(Surv(tstart, tstop, status) ~ fit$linear, cgd, iter=0, init=1) > fit3 <- coxph(Surv(tstart, tstop, status) ~ offset(fit$linear), cgd) > all.equal(fit$resid, fit2$resid) [1] TRUE > all.equal(fit$resid, fit3$resid) [1] TRUE > > # > # Check using coxph.detail. The matrix multiply below only is > # valid for the breslow approximation. > fit4 <- coxph(Surv(tstart, tstop, status) ~ sex + treat + pspline(age), + cgd, ties='breslow') > dt <- coxph.detail(fit4, riskmat=TRUE) > rscore <- exp(fit4$linear) > exp4 <- (rscore *dt$riskmat) %*% dt$hazard > r4 <- cgd$status - exp4 > aeq(r4, fit4$resid) [1] TRUE > > proc.time() user system elapsed 0.304 0.028 0.329 survival/tests/clogit.Rout.save0000644000175100001440000000365411732700061016403 0ustar hornikusers R version 2.14.0 (2011-10-31) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Test of the clogit function, and indirectly of the exact option > # > # Data set logan has the occupation of fathers, we create a > # multinomial response > # > nresp <- length(levels(logan$occupation)) > n <- nrow(logan) > indx <- rep(1:n, nresp) > logan2 <- data.frame(logan[indx,], + id = indx, + occ2 = factor(rep(levels(logan$occupation), each=n))) > logan2$y <- (logan2$occupation == logan2$occ2) > > #We expect two NA coefficients, so ignore the warning > fit1 <- clogit(y ~ occ2 + occ2:education + occ2:race + strata(id), logan2) Warning message: In coxph(formula = Surv(rep(1, 4190L), y) ~ occ2 + occ2:education + : X matrix deemed to be singular; variable 9 14 > > #since there is only one death per group, all methods are equal > dummy <- rep(1, nrow(logan2)) > fit2 <- coxph(Surv(dummy, y) ~ occ2 + occ2:education + occ2:race + strata(id), + logan2, method='breslow') Warning message: In coxph(Surv(dummy, y) ~ occ2 + occ2:education + occ2:race + strata(id), : X matrix deemed to be singular; variable 9 14 > > all.equal(fit1$coef, fit2$coef) [1] TRUE > all.equal(fit1$loglik, fit2$loglik) [1] TRUE > all.equal(fit1$var, fit2$var) [1] TRUE > all.equal(fit1$resid, fit2$resid) [1] TRUE > > survival/tests/summary_survfit.Rout.save0000644000175100001440000000526111732700061020375 0ustar hornikusers R version 2.9.0 Under development (unstable) (2009-03-17 r48144) Copyright (C) 2009 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > ## check that the scale option to summary.survfit works > ## Marc Schwartz reported this as a bug in 2.35-3. > library(survival) Loading required package: splines > summary( survfit( Surv(futime, fustat)~1, data=ovarian)) Call: survfit(formula = Surv(futime, fustat) ~ 1, data = ovarian) time n.risk n.event survival std.err lower 95% CI upper 95% CI 59 26 1 0.962 0.0377 0.890 1.000 115 25 1 0.923 0.0523 0.826 1.000 156 24 1 0.885 0.0627 0.770 1.000 268 23 1 0.846 0.0708 0.718 0.997 329 22 1 0.808 0.0773 0.670 0.974 353 21 1 0.769 0.0826 0.623 0.949 365 20 1 0.731 0.0870 0.579 0.923 431 17 1 0.688 0.0919 0.529 0.894 464 15 1 0.642 0.0965 0.478 0.862 475 14 1 0.596 0.0999 0.429 0.828 563 12 1 0.546 0.1032 0.377 0.791 638 11 1 0.497 0.1051 0.328 0.752 > summary( survfit( Surv(futime, fustat)~1, data=ovarian), scale=365.25) Call: survfit(formula = Surv(futime, fustat) ~ 1, data = ovarian) time n.risk n.event survival std.err lower 95% CI upper 95% CI 0.162 26 1 0.962 0.0377 0.890 1.000 0.315 25 1 0.923 0.0523 0.826 1.000 0.427 24 1 0.885 0.0627 0.770 1.000 0.734 23 1 0.846 0.0708 0.718 0.997 0.901 22 1 0.808 0.0773 0.670 0.974 0.966 21 1 0.769 0.0826 0.623 0.949 0.999 20 1 0.731 0.0870 0.579 0.923 1.180 17 1 0.688 0.0919 0.529 0.894 1.270 15 1 0.642 0.0965 0.478 0.862 1.300 14 1 0.596 0.0999 0.429 0.828 1.541 12 1 0.546 0.1032 0.377 0.791 1.747 11 1 0.497 0.1051 0.328 0.752 > survival/tests/book3.Rout.save0000644000175100001440000002116612350331622016135 0ustar hornikusers R Under development (unstable) (2014-05-11 r65563) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # > # Tests from the appendix of Therneau and Grambsch > # c. Data set 2 and Breslow estimate > # > test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), + stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), + event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), + x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) > > byhand <- function(beta, newx=0) { + r <- exp(beta) + loglik <- 4*beta - log(r+1) - log(r+2) - 3*log(3*r+2) - 2*log(3*r+1) + u <- 1/(r+1) + 1/(3*r+1) + 4/(3*r+2) - + ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) + imat <- r/(r+1)^2 + 2*r/(r+2)^2 + 6*r/(3*r+2)^2 + + 3*r/(3*r+1)^2 + 3*r/(3*r+1)^2 + 12*r/(3*r+2)^2 + + hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), 2/(3*r+2) ) + xbar <- c(r/(r+1), r/(r+2), 3*r/(3*r+2), 3*r/(3*r+1), 3*r/(3*r+1), + 3*r/(3*r+2)) + + # The matrix of weights, one row per obs, one col per time + # deaths at 2,3,6,7,8,9 + wtmat <- matrix(c(1,0,0,0,1,0,0,0,0,0, + 0,1,0,1,1,0,0,0,0,0, + 0,0,1,1,1,0,1,1,0,0, + 0,0,0,1,1,0,1,1,0,0, + 0,0,0,0,1,1,1,1,0,0, + 0,0,0,0,0,1,1,1,1,1), ncol=6) + wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat + + x <- c(1,0,0,1,0,1,1,1,0,0) + status <- c(1,1,1,1,1,1,1,0,0,0) + xbar <- colSums(wtmat*x)/ colSums(wtmat) + n <- length(x) + + # Table of sums for score and Schoenfeld resids + hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time + dM <- -hazmat #Expected part + for (i in 1:6) dM[i,i] <- dM[i,i] +1 #observed + dM[7,6] <- dM[7,6] +1 # observed + mart <- rowSums(dM) + + # Table of sums for score and Schoenfeld resids + # Looks like the last table of appendix E.2.1 of the book + resid <- dM * outer(x, xbar, '-') + score <- rowSums(resid) + scho <- colSums(resid) + # We need to split the two tied times up, to match coxph + scho <- c(scho[1:5], scho[6]/2, scho[6]/2) + var.g <- cumsum(hazard*hazard /c(1,1,1,1,1,2)) + var.d <- cumsum( (xbar-newx)*hazard) + + surv <- exp(-cumsum(hazard) * exp(beta*newx)) + varhaz <- (var.g + var.d^2/imat)* exp(2*beta*newx) + + list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard, + mart=mart, score=score, rmat=resid, + scho=scho, surv=surv, var=varhaz) + } > > > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > fit0 <-coxph(Surv(start, stop, event) ~x, test2, iter=0, method='breslow') > truth0 <- byhand(0,0) > aeq(truth0$loglik, fit0$loglik[1]) [1] TRUE > aeq(1/truth0$imat, fit0$var) [1] TRUE > aeq(truth0$mart, fit0$resid) [1] TRUE > aeq(truth0$scho, resid(fit0, 'schoen')) [1] TRUE > aeq(truth0$score, resid(fit0, 'score')) [1] TRUE > sfit <- survfit(fit0, list(x=0), censor=FALSE) > aeq(sfit$std.err^2, truth0$var) [1] TRUE > aeq(sfit$surv, truth0$surv) [1] TRUE > > fit <- coxph(Surv(start, stop, event) ~x, test2, eps=1e-8, method='breslow') > truth <- byhand(fit$coef, 0) > aeq(truth$loglik, fit$loglik[2]) [1] TRUE > aeq(1/truth$imat, fit$var) [1] TRUE > aeq(truth$mart, fit$resid) [1] TRUE > aeq(truth$scho, resid(fit, 'schoen')) [1] TRUE > aeq(truth$score, resid(fit, 'score')) [1] TRUE > expect <- predict(fit, type='expected', newdata=test2) #force recalc > aeq(test2$event -fit$resid, expect) #tests the predict function [1] TRUE > > sfit <- survfit(fit, list(x=0), censor=FALSE) > aeq(sfit$std.err^2, truth$var) [1] TRUE > aeq(-log(sfit$surv), (cumsum(truth$haz))) [1] TRUE > > # > # Done with the formal test, now print out lots of bits > # > resid(fit) 1 2 3 4 5 6 0.52111895 0.65741078 0.78977654 0.24738772 -0.60629349 0.36902492 7 8 9 10 -0.06876579 -1.06876579 -0.42044692 -0.42044692 > resid(fit, 'scor') 1 2 3 4 5 6 0.27156496 -0.20696709 -0.45771743 -0.09586133 0.13608234 0.19288983 7 8 9 10 0.04655651 -0.37389040 0.24367131 0.24367131 > resid(fit, 'scho') 2 3 6 7 8 9 9 0.5211189 -0.3148216 -0.5795531 0.2661809 -0.7338191 0.4204469 0.4204469 > > predict(fit, type='lp') [1] -0.04226304 0.04226304 0.04226304 -0.04226304 0.04226304 -0.04226304 [7] -0.04226304 -0.04226304 0.04226304 0.04226304 > predict(fit, type='risk') [1] 0.9586176 1.0431688 1.0431688 0.9586176 1.0431688 0.9586176 0.9586176 [8] 0.9586176 1.0431688 1.0431688 > predict(fit, type='expected') 1 2 3 4 5 6 7 8 0.4788811 0.3425892 0.2102235 0.7526123 1.6062935 0.6309751 1.0687658 1.0687658 9 10 0.4204469 0.4204469 > predict(fit, type='terms') x 1 -0.04226304 2 0.04226304 3 0.04226304 4 -0.04226304 5 0.04226304 6 -0.04226304 7 -0.04226304 8 -0.04226304 9 0.04226304 10 0.04226304 attr(,"constant") [1] -0.04226304 > predict(fit, type='lp', se.fit=T) $fit 1 2 3 4 5 6 -0.04226304 0.04226304 0.04226304 -0.04226304 0.04226304 -0.04226304 7 8 9 10 -0.04226304 -0.04226304 0.04226304 0.04226304 $se.fit 1 2 3 4 5 6 7 8 0.3969086 0.3969086 0.3969086 0.3969086 0.3969086 0.3969086 0.3969086 0.3969086 9 10 0.3969086 0.3969086 > predict(fit, type='risk', se.fit=T) $fit 1 2 3 4 5 6 7 8 0.9586176 1.0431688 1.0431688 0.9586176 1.0431688 0.9586176 0.9586176 0.9586176 9 10 1.0431688 1.0431688 $se.fit 1 2 3 4 5 6 7 8 0.3886094 0.4053852 0.4053852 0.3886094 0.4053852 0.3886094 0.3886094 0.3886094 9 10 0.4053852 0.4053852 > predict(fit, type='expected', se.fit=T) $fit 1 2 3 4 5 6 7 8 0.4788811 0.3425892 0.2102235 0.7526123 1.6062935 0.6309751 1.0687658 1.0687658 9 10 0.4204469 0.4204469 $se.fit [1] 0.5182381 0.3982700 0.3292830 0.6266797 1.0255146 0.5852364 0.7341340 [8] 0.7341340 0.6268550 0.6268550 > predict(fit, type='terms', se.fit=T) $fit x 1 -0.04226304 2 0.04226304 3 0.04226304 4 -0.04226304 5 0.04226304 6 -0.04226304 7 -0.04226304 8 -0.04226304 9 0.04226304 10 0.04226304 attr(,"constant") [1] -0.04226304 $se.fit x 1 0.3969086 2 0.3969086 3 0.3969086 4 0.3969086 5 0.3969086 6 0.3969086 7 0.3969086 8 0.3969086 9 0.3969086 10 0.3969086 > > summary(survfit(fit)) Call: survfit(formula = fit) time n.risk n.event survival std.err lower 95% CI upper 95% CI 2 2 1 0.607 0.303 0.2279 1.000 3 3 1 0.437 0.262 0.1347 1.000 6 5 1 0.357 0.226 0.1034 1.000 7 4 1 0.277 0.188 0.0729 1.000 8 4 1 0.214 0.156 0.0514 0.894 9 5 2 0.143 0.112 0.0308 0.667 > summary(survfit(fit, list(x=2))) Call: survfit(formula = fit, newdata = list(x = 2)) time n.risk n.event survival std.err lower 95% CI upper 95% CI 2 2 1 0.644 0.444 0.16657 1 3 3 1 0.482 0.511 0.06055 1 6 5 1 0.404 0.504 0.03491 1 7 4 1 0.322 0.475 0.01801 1 8 4 1 0.258 0.437 0.00928 1 9 5 2 0.181 0.377 0.00302 1 > > proc.time() user system elapsed 0.240 0.024 0.262 survival/tests/frank.Rout.save0000644000175100001440000000302012164375045016220 0ustar hornikusers R version 3.0.0 (2013-04-03) -- "Masked Marvel" Copyright (C) 2013 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # Check out intercept/interaction for Frank H > # > age2 <- lung$age - 50 > fit1 <- coxph(Surv(time, status) ~ age * strata(sex), lung) > fit2 <- coxph(Surv(time, status) ~ age2*strata(sex), lung) > > tdata <- data.frame(age=50:60, age2=0:10, sex=c(1,2,1,2,1,2,1,2,1,2,1)) > > surv1 <- survfit(fit1, tdata) > surv2 <- survfit(fit2, tdata) > # The call won't match, but the rest should > icall <- match("call", names(surv1)) > all.equal(unclass(surv1)[-icall], unclass(surv2)[-icall]) [1] TRUE > > > # It should match what I get with a single strata fit > > fit3 <- coxph(Surv(time, status) ~ age, data=lung, + init=fit1$coef[1], subset=(sex==1), iter=0) > surv1b <- survfit(fit3, newdata=list(age=c(50,52, 54))) > all.equal(c(surv1b$surv), surv1[c(1,3,5)]$surv) [1] TRUE > > > > > proc.time() user system elapsed 0.280 0.028 0.306 survival/tests/r_stanford.R0000644000175100001440000000511611732700061015571 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # The Stanford data from 1980 is used in Escobar and Meeker, Biometrics 1992. # t5 = T5 mismatch score # Their case numbers correspond to a data set sorted by age # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) stanford2$t5 <- ifelse(stanford2$t5 <0, NA, stanford2$t5) stanford2 <- stanford2[order(stanford2$age, stanford2$time),] stanford2$time <- ifelse(stanford2$time==0, .5, stanford2$time) cage <- stanford2$age - mean(stanford2$age) fit1 <- survreg(Surv(time, status) ~ cage + I(cage^2), stanford2, dist='lognormal') fit1 ldcase <- resid(fit1, type='ldcase') ldresp <- resid(fit1, type='ldresp') # The ldcase and ldresp should be compared to table 1 in Escobar and # Meeker, Biometrics 1992, p519; the colums they label as (1/2) A_{ii} # They give data for selected cases, entered below as mdata mdata <- cbind(c(1,2,4,5,12,16,23,61,66,72,172,182,183,184), c(.035, .244, .141, .159, .194, .402, 0,0, .143, .403, .178, .033, .005, .015), c(.138, .145, .073, .076, .104, .159, 0,0, .109, .184, .116, .063, .103, .144)) dimnames(mdata) <- list(NULL, c("case#", "ldcase", "ldresp")) aeq(round(ldcase[mdata[,1]],3), mdata[,2]) aeq(round(ldresp[mdata[,1]],3), mdata[,3]) plot1 <- function() { # make their figure 1, 2, and 6 temp <- predict(fit1, type='quantile', p=c(.1, .5, .9)) plot(stanford2$age, stanford2$time, log='y', xlab="Age", ylab="Days", ylim=range(stanford2$time, temp)) matlines(stanford2$age, temp, lty=c(1,2,2), col=1) n <- length(ldcase) plot(1:n, ldcase, xlab="Case Number", ylab="(1/2) A", type='l') title (main="Case weight pertubations") plot(1:n, ldresp, xlab="Case Number", ylab="(1/2) A", ylim=c(0, .2), type='l') title(main="Response pertubations") indx <- which(ldresp > .07) text(indx, ldresp[indx]+ .005, indx%%10, cex=.6) } postscript('meekerplot.ps') plot1() dev.off() # # Stanford predictions in other ways # fit2 <- survreg(Surv(time, status) ~ poly(age,2), stanford2, dist='lognormal') p1 <- predict(fit1, type='response') p2 <- predict(fit2, type='response') aeq(p1, p2) p3 <- predict(fit2, type='terms', se=T) p4 <- predict(fit2, type='lp', se=T) p5 <- predict(fit1, type='lp', se=T) # aeq(p3$fit + attr(p3$fit, 'constant'), p4$fit) #R is missing the attribute aeq(p4$fit, p5$fit) aeq(p3$se.fit, p4$se.fit) #this one should be false aeq(p4$se.fit, p5$se.fit) #this one true survival/tests/infcox.R0000644000175100001440000000232112160143136014711 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # A test to exercise the "infinity" check on 2 variables # test3 <- data.frame(futime=1:12, fustat=c(1,0,1,0,1,0,0,0,0,0,0,0), x1=rep(0:1,6), x2=c(rep(0,6), rep(1,6))) # This will produce a warning message, which is the point of the test. # The variance is close to singular and gives different answers # on different machines fit3 <- coxph(Surv(futime, fustat) ~ x1 + x2, test3, iter=25) all(fit3$coef < -22) all.equal(round(fit3$log, 4),c(-6.8669, -1.7918)) # # Actual solution # time 1, 12 at risk, 3 each of x1/x2 = 00, 01, 10, 11 # time 2, 10 at risk, 2, 3, 2 , 3 # time 5, 8 at risk, 1, 3, 1, 3 # Let r1 = exp(beta1), r2= exp(beta2) # loglik = -log(3 + 3r1 + 3r2 + 3 r1*r2) - log(2 + 2r1 + 3r2 + 3 r1*r2) - # log(1 + r1 + 3r2 + 3 r1*r2) true <- function(beta) { r1 <- exp(beta[1]) r2 <- exp(beta[2]) loglik <- -log(3*(1+ r1+ r2+ r1*r2)) - log(2+ 2*r1 + 3*r2 + 3*r1*r2) - log(1 + r1 + 3*r2 + 3*r1*r2) loglik } all.equal(fit3$loglik[2], true(fit3$coef), check.attributes=FALSE) survival/tests/cancer.R0000644000175100001440000000212311745775414014700 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test out all of the routines on a more complex data set # temp <- survfit(Surv(time, status) ~ ph.ecog, lung) summary(temp, times=c(30*1:11, 365*1:3)) print(temp[2:3]) temp <- survfit(Surv(time, status)~1, lung, type='fleming', conf.int=.9, conf.type='log-log', error='tsiatis') summary(temp, times=30 *1:5) temp <- survdiff(Surv(time, status) ~ inst, lung, rho=.5) print(temp, digits=6) temp <- coxph(Surv(time, status) ~ ph.ecog + ph.karno + pat.karno + wt.loss + sex + age + meal.cal + strata(inst), lung) summary(temp) cox.zph(temp) cox.zph(temp, transform='identity') coxph(Surv(rep(0,length(time)), time, status) ~ ph.ecog + ph.karno + pat.karno + wt.loss + sex + age + meal.cal + strata(inst), lung) # # Tests of using "." # fit1 <- coxph(Surv(time, status) ~ . - meal.cal - wt.loss - inst, lung) fit2 <- update(fit1, .~. - ph.karno) fit3 <- coxph(Surv(time, status) ~ age + sex + ph.ecog + pat.karno, lung) all.equal(fit2, fit3) survival/tests/detail.R0000644000175100001440000000515111732700061014671 0ustar hornikusers# A short test on coxph.detail, to ensure that the computed hazard is # equal to the theoretical value library(survival) aeq <- function(a,b) all.equal(as.vector(a), as.vector(b)) # taken from book4.R test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) byhand <- function(beta, newx=0) { r <- exp(beta) loglik <- 4*beta - (log(r+1) + log(r+2) + 2*log(3*r+2) + 2*log(3*r+1) + log(2*r +2)) u <- 1/(r+1) + 1/(3*r+1) + 2*(1/(3*r+2) + 1/(2*r+2)) - ( r/(r+2) +3*r/(3*r+2) + 3*r/(3*r+1)) imat <- r*(1/(r+1)^2 + 2/(r+2)^2 + 6/(3*r+2)^2 + 6/(3*r+1)^2 + 6/(3*r+2)^2 + 4/(2*r +2)^2) hazard <-c( 1/(r+1), 1/(r+2), 1/(3*r+2), 1/(3*r+1), 1/(3*r+1), 1/(3*r+2), 1/(2*r +2) ) # The matrix of weights, one row per obs, one col per time # deaths at 2,3,6,7,8,9 wtmat <- matrix(c(1,0,0,0,1, 0, 0,0,0,0, 0,1,0,1,1, 0, 0,0,0,0, 0,0,1,1,1, 0, 1,1,0,0, 0,0,0,1,1, 0, 1,1,0,0, 0,0,0,0,1, 1, 1,1,0,0, 0,0,0,0,0, 1, 1,1,1,1, 0,0,0,0,0,.5,.5,1,1,1), ncol=7) wtmat <- diag(c(r,1,1,r,1,r,r,r,1,1)) %*% wtmat x <- c(1,0,0,1,0,1,1,1,0,0) status <- c(1,1,1,1,1,1,1,0,0,0) xbar <- colSums(wtmat*x)/ colSums(wtmat) n <- length(x) # Table of sums for score and Schoenfeld resids hazmat <- wtmat %*% diag(hazard) #each subject's hazard over time dM <- -hazmat #Expected part for (i in 1:5) dM[i,i] <- dM[i,i] +1 #observed dM[6:7,6:7] <- dM[6:7,6:7] +.5 # observed mart <- rowSums(dM) # Table of sums for score and Schoenfeld resids # Looks like the last table of appendix E.2.1 of the book resid <- dM * outer(x, xbar, '-') score <- rowSums(resid) scho <- colSums(resid) # We need to add the ties back up (they are symmetric) scho[6:7] <- rep(mean(scho[6:7]), 2) list(loglik=loglik, u=u, imat=imat, xbar=xbar, haz=hazard* exp(beta*newx), mart=mart, score=score, rmat=resid, scho=scho) } # The actual coefficient of the fit is close to zero. Using a larger # number pushes the test harder, but it should still work without # the init and iter arguments, i.e., for any coefficient. fit1 <- coxph(Surv(start, stop, event) ~x, test2,init=-1, iter=0) temp <- coxph.detail(fit1) temp2 <- byhand(fit1$coef, fit1$means) aeq(temp$haz, c(temp2$haz[1:5], sum(temp2$haz[6:7]))) survival/tests/fr_rat2.R0000644000175100001440000000411511732700061014765 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # From Gail, Sautner and Brown, Biometrics 36, 255-66, 1980 # 48 rats were injected with a carcinogen, and then randomized to either # drug or placebo. The number of tumors ranges from 0 to 13; all rats were # censored at 6 months after randomization. # Variables: rat, treatment (1=drug, 0=control), o # observation # within rat, # (start, stop] status # The raw data has some intervals of zero length, i.e., start==stop. # We add .1 to these times as an approximate solution # rat2 <- read.table('data.rat2', col.names=c('id', 'rx', 'enum', 'start', 'stop', 'status')) temp1 <- rat2$start temp2 <- rat2$stop for (i in 1:nrow(rat2)) { if (temp1[i] == temp2[i]) { temp2[i] <- temp2[i] + .1 if (i < nrow(rat2) && rat2$id[i] == rat2$id[i+1]) { temp1[i+1] <- temp1[i+1] + .1 if (temp2[i+1] <= temp1[i+1]) temp2[i+1] <- temp1[i+1] } } } rat2$start <- temp1 rat2$stop <- temp2 r2fit0 <- coxph(Surv(start, stop, status) ~ rx + cluster(id), rat2) r2fitg <- coxph(Surv(start, stop, status) ~ rx + frailty(id), rat2) r2fitm <- coxph(Surv(start, stop, status) ~ rx + frailty.gaussian(id), rat2) r2fit0 r2fitg r2fitm #This example is unusual: the frailties variances end up about the same, # but the effect on rx differs. Double check it # Because of different iteration paths, the coef won't be exactly the # same, but darn close. temp <- coxph(Surv(start, stop, status) ~ rx + offset(r2fitm$frail[id]), rat2) all.equal(temp$coef, r2fitm$coef[1], tolerance=1e-7) temp <- coxph(Surv(start, stop, status) ~ rx + offset(r2fitg$frail[id]), rat2) all.equal(temp$coef, r2fitg$coef[1], tolerance=1e-7) # # What do I get with AIC # r2fita1 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, method='aic'), rat2) r2fita2 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, method='aic', dist='gauss'), rat2) r2fita3 <- coxph(Surv(start, stop, status) ~ rx + frailty(id, dist='t'), rat2) r2fita1 r2fita2 r2fita3 survival/tests/data.peterson0000644000175100001440000000017711732700061016001 0ustar hornikusers1 4 1 1 7 1 1 12 1 2 3 0 2 10 0 2 22 1 2 21 1 2 11 0 2 12 0 6 18 1 6 9 1 3 12 0 3 19 1 3 16 0 3 5 0 3 14 0 3 20 1 4 2 1 5 6 1 survival/tests/frailty.Rout.save0000644000175100001440000000302011732700061016557 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > # > # The constuction of a survival curve with sparse frailties > # > # In this case the coefficient vector is kept in two parts, the > # fixed coefs and the (often very large) random effects coefficients > # The survfit function treats the second set of coefficients as fixed > # values, to avoid an unmanagable variance matrix, and behaves like > # the second fit below. > > fit1 <- coxph(Surv(time, status) ~ age + frailty(inst), lung) > sfit1 <- survfit(fit1) > > # A parallel model with the frailties treated as fixed offsets > offvar <- fit1$frail[as.numeric(factor(lung$inst))] > fit2 <- coxph(Surv(time, status) ~ age + offset(offvar),lung) > fit2$var <- fit1$var #force variances to match > > all.equal(fit1$coef, fit2$coef) [1] TRUE > sfit2 <- survfit(fit2, newdata=list(age=fit1$means, offvar=0)) > all.equal(sfit1$surv, sfit2$surv) [1] TRUE > all.equal(sfit1$var, sfit2$var) [1] TRUE > survival/tests/counting.R0000644000175100001440000000364311732700061015261 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # Create a "counting process" version of the simplest test data set # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) test1b<- list(start= c(0, 3, 0, 0, 5, 0, 6,14, 0, 0, 10,20,30, 0), stop = c(3,10, 10, 5,20, 6,14,20, 30, 10,20,30,40, 10), status=c(0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0), x= c(1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, NA), id = c(3, 3, 4, 5, 5, 6, 6, 6, 7, 1, 1, 1, 1, 2)) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) # # Check out the various residuals under an Efron approximation # fit0 <- coxph(Surv(time, status)~ x, test1, iter=0) fit <- coxph(Surv(time, status) ~x, test1) fit0b <- coxph(Surv(start, stop, status) ~ x, test1b, iter=0) fitb <- coxph(Surv(start, stop, status) ~x, test1b) fitc <- coxph(Surv(time, status) ~ offset(fit$coef*x), test1) fitd <- coxph(Surv(start, stop, status) ~ offset(fit$coef*x), test1b) aeq(fit0b$coef, fit0$coef) aeq(resid(fit0), resid(fit0b, collapse=test1b$id)) aeq(resid(fit), resid(fitb, collapse=test1b$id)) aeq(resid(fitc), resid(fitd, collapse=test1b$id)) aeq(resid(fitc), resid(fit)) aeq(resid(fit0, type='score'), resid(fit0b, type='score', collapse=test1b$id)) aeq(resid(fit, type='score'), resid(fitb, type='score', collapse=test1b$id)) aeq(resid(fit0, type='scho'), resid(fit0b, type='scho', collapse=test1b$id)) aeq(resid(fit, type='scho'), resid(fitb, type='scho', collapse=test1b$id)) # The two survivals will have different censoring times # nrisk, nevent, surv, and std should be the same temp1 <- survfit(fit, list(x=1), censor=FALSE) temp2 <- survfit(fitb, list(x=1), censor=FALSE) all.equal(unclass(temp1)[c(3,4,6,8)], unclass(temp2)[c(3,4,6,8)]) survival/tests/testci.Rout.save0000644000175100001440000001213112466142446016417 0ustar hornikusers R Under development (unstable) (2014-09-01 r66509) -- "Unsuffered Consequences" Copyright (C) 2014 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > aeq <- function(x,y,...) all.equal(as.vector(x), as.vector(y),...) > > # > # Test out the survfit.ci function, which does competing risk > # estimates > # > # First trivial test > tdata <- data.frame(time=c(1,2,2,3,3,3,5,6), + status = c(0,1,0,1,0,1,0,1), + event = c(1,1,2,2,1,2,3,2), + grp = c(1,2,1,2,1,2,1,2)) > fit <- survfit(Surv(time, status*event, type='mstate') ~1, tdata) > > byhand <- function() { + #everyone starts in state 0 + p1 <- c(1,0,0) + + p2 <- c(6/7, 1/7, 0) # 0-1 transition at time 2 + u2 <- matrix(rep(c(1/49, -1/49, 0), each=8), ncol=3) #leverage matrix at time 2 + u2[1,] <- 0 #subject 1 is not present + u2[2,1:2] <- u2[2, 1:2] + c(-1/7, 1/7) + + p3 <- c((6/7)*(3/5), 1/7, 12/35) # 0-2 transition at time 3, 5 at risk + h3 <- matrix(c(3/5, 0, 2/5, 0,1,0, 0,1,0), byrow=T, ncol=3) #hazard mat + u3 <- u2 %*% h3 + u3[4:8,1] <- u3[4:8,1] + p2[1]*2/25 + u3[4:8,3] <- u3[4:8,3] -p2[1]*2/25 + u3[4,] <- u3[4,] + c(-p2[1]/5, 0, p2[1]/5) + u3[6,] <- u3[4,] + + p6 <- c(0, 1/7, 6/7) # 0-2 at time 6, 1 at risk + h6 <- matrix(c(-1,0,1,0,1,0,0,1,0), byrow=T, ncol=3) + u6 <- cbind(0, u3[,2], -u3[,2]) + + V <- rbind(0, colSums(u2^2), + colSums(u3^2), + colSums(u3^2), + colSums(u6^2)) + list(P=rbind(p1, p2, p3, p3, p6), u2=u2, u3=u3, u6=u6, V=V) + } > bfit <- byhand() > aeq(fit$prev, bfit$P[,-1]) [1] TRUE > aeq(fit$n.risk, c(8,7,5,2,1)) [1] TRUE > aeq(fit$n.event, c(0,1,2,0,1)) [1] TRUE > aeq(fit$std^2, bfit$V[,-1]) [1] TRUE > > # > # For this we need the competing risks MGUS data set, first > # event > # > tdata <- mgus1[mgus1$enum==1,] > # Ensure the old-style call using "etype" works (backwards compatability) > fit1 <- survfit(Surv(stop, status) ~ 1, etype=event, tdata) > fit1b <-survfit(Surv(stop, event) ~1, tdata) > indx <- match("call", names(fit1)) > all.equal(unclass(fit1)[-indx], unclass(fit1b)[-indx]) [1] TRUE > > # Now get the overall survival, and the hazard for progression > fit2 <- survfit(Surv(stop, status) ~1, tdata) #overall to "first bad thing" > fit3 <- survfit(Surv(stop, status*(event=='pcm')) ~1, tdata, + type='fleming') > fit4 <- survfit(Surv(stop, status*(event=='death')) ~1, tdata, + type='fleming') > > aeq(fit1$n.risk, fit2$n.risk) [1] TRUE > aeq(fit1$n.event, fit2$n.event) [1] TRUE > > # Classic CI formula > # integral [hazard(t) S(t-0) dt], where S= "survival to first event" > haz1 <- diff(c(0, -log(fit3$surv))) #Aalen hazard estimate for progression > haz2 <- diff(c(0, -log(fit4$surv))) #Aalen estimate for death > tsurv <- c(1, fit2$surv[-length(fit2$surv)]) #lagged survival > ci1 <- cumsum(haz1 *tsurv) > ci2 <- cumsum(haz2 *tsurv) > aeq(cbind(ci1, ci2), fit1$prev) [1] TRUE > > # > # Now, make sure that it works for subgroups > # > fit1 <- survfit(Surv(stop, event) ~ sex, tdata) > fit2 <- survfit(Surv(stop, event) ~ 1, tdata, + subset=(sex=='female')) > fit3 <- survfit(Surv(stop, event) ~ 1, tdata, + subset=(sex=='male')) > > aeq(fit2$prev, fit1$prev[1:fit1$strata[1],]) [1] TRUE > aeq(fit2$std, fit1$std[1:fit1$strata[1],]) [1] TRUE > aeq(fit3$prev, fit1$prev[-(1:fit1$strata[1]),]) [1] TRUE > > # A second test of cumulative incidence > # compare results to Bob Gray's functions > # The file gray1 is the result of > # library(cmprsk) > # tstat <- ifelse(tdata$status==0, 0, 1+ (tdata$event=='death')) > # gray1 <- cuminc(tdata$stop, tstat) > load("gray1.rda") > fit2 <- survfit(Surv(stop, event) ~ 1, tdata) > > if (FALSE) { + # lines of the two graphs should overlay + plot(gray1[[1]]$time, gray1[[1]]$est, type='l', + ylim=range(c(gray1[[1]]$est, gray1[[2]]$est)), + xlab="Time") + lines(gray1[[2]]$time, gray1[[2]]$est, lty=2) + matlines(fit2$time, fit2$prev, col=2, lty=1:2, type='s') + } > # To formally match these is a bit of a nuisance. > # The cuminc function returns a full step function, and survfit only > # the bottoms of the steps. > temp1 <- tapply(gray1[[1]]$est, gray1[[1]]$time, max)[-1] #toss time 0 > indx1 <- match(names(temp1), fit2$time) > aeq(temp1, fit2$prev[indx1,1]) [1] TRUE > > > proc.time() user system elapsed 0.232 0.040 0.272 survival/tests/data.fluid0000644000175100001440000000114711732700061015243 0ustar hornikusers 5.79 26kV 1579.52 26kV 2323.70 26kV 7.74 30kV 17.05 30kV 20.46 30kV 21.02 30kV 22.66 30kV 43.40 30kV 47.30 30kV 139.07 30kV 144.12 30kV 175.88 30kV 194.90 30kV 0.19 34kV 0.78 34kV 0.96 34kV 1.31 34kV 2.78 34kV 3.16 34kV 4.15 34kV 4.67 34kV 4.85 34kV 6.50 34kV 7.35 34kV 8.01 34kV 8.27 34kV 12.06 34kV 31.75 34kV 32.52 34kV 33.91 34kV 36.71 34kV 72.89 34kV 0.09 38kV 0.39 38kV 0.47 38kV 0.73 38kV 0.74 38kV 1.13 38kV 1.40 38kV 2.38 38kV survival/tests/r_lung.Rout.save0000644000175100001440000001206011732700061016377 0ustar hornikusers R version 2.14.0 (2011-10-31) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > lfit2 <- survreg(Surv(time, status) ~ age + ph.ecog + strata(sex), lung) > lfit3 <- survreg(Surv(time, status) ~ sex + (age+ph.ecog)*strata(sex), lung) > > lfit4 <- survreg(Surv(time, status) ~ age + ph.ecog , lung, + subset=(sex==1)) > lfit5 <- survreg(Surv(time, status) ~ age + ph.ecog , lung, + subset=(sex==2)) > > if (exists('censorReg')) { + lfit1 <- censorReg(censor(time, status) ~ age + ph.ecog + strata(sex),lung) + aeq(lfit4$coef, lfit1[[1]]$coef) + aeq(lfit4$scale, lfit1[[1]]$scale) + aeq(c(lfit4$scale, lfit5$scale), sapply(lfit1, function(x) x$scale)) + } > aeq(c(lfit4$scale, lfit5$scale), lfit3$scale ) [1] TRUE > > # > # Test out ridge regression and splines > # > lfit0 <- survreg(Surv(time, status) ~1, lung) > lfit1 <- survreg(Surv(time, status) ~ age + ridge(ph.ecog, theta=5), lung) > lfit2 <- survreg(Surv(time, status) ~ sex + ridge(age, ph.ecog, theta=1), lung) > lfit3 <- survreg(Surv(time, status) ~ sex + age + ph.ecog, lung) > > lfit0 Call: survreg(formula = Surv(time, status) ~ 1, data = lung) Coefficients: (Intercept) 6.034904 Scale= 0.7593936 Loglik(model)= -1153.9 Loglik(intercept only)= -1153.9 n= 228 > lfit1 Call: survreg(formula = Surv(time, status) ~ age + ridge(ph.ecog, theta = 5), data = lung) coef se(coef) se2 Chisq DF p (Intercept) 6.83082 0.42860 0.42860 254.0 1 0.00000 age -0.00783 0.00687 0.00687 1.3 1 0.25000 ridge(ph.ecog) -0.32032 0.08484 0.08405 14.2 1 0.00016 Scale= 0.738 Iterations: 1 outer, 5 Newton-Raphson Degrees of freedom for terms= 1 1 1 1 Likelihood ratio test=18.6 on 2 df, p=8.73e-05 n=227 (1 observation deleted due to missingness) > lfit2 Call: survreg(formula = Surv(time, status) ~ sex + ridge(age, ph.ecog, theta = 1), data = lung) coef se(coef) se2 Chisq DF p (Intercept) 6.27163 0.45280 0.45210 191.84 1 0.0e+00 sex 0.40096 0.12371 0.12371 10.50 1 1.2e-03 ridge(age) -0.00746 0.00675 0.00674 1.22 1 2.7e-01 ridge(ph.ecog) -0.33848 0.08329 0.08314 16.51 1 4.8e-05 Scale= 0.731 Iterations: 1 outer, 6 Newton-Raphson Degrees of freedom for terms= 1 1 2 1 Likelihood ratio test=30 on 3 df, p=1.37e-06 n=227 (1 observation deleted due to missingness) > lfit3 Call: survreg(formula = Surv(time, status) ~ sex + age + ph.ecog, data = lung) Coefficients: (Intercept) sex age ph.ecog 6.273435252 0.401090541 -0.007475439 -0.339638098 Scale= 0.731109 Loglik(model)= -1132.4 Loglik(intercept only)= -1147.4 Chisq= 29.98 on 3 degrees of freedom, p= 1.4e-06 n=227 (1 observation deleted due to missingness) > > > xx <- pspline(lung$age, nterm=3, theta=.3) > xx <- matrix(unclass(xx), ncol=ncol(xx)) # the raw matrix > lfit4 <- survreg(Surv(time, status) ~xx, lung) > lfit5 <- survreg(Surv(time, status) ~age, lung) > > lfit6 <- survreg(Surv(time, status)~pspline(age, df=2), lung) > > lfit7 <- survreg(Surv(time, status) ~ offset(lfit6$lin), lung) > > lfit4 Call: survreg(formula = Surv(time, status) ~ xx, data = lung) Coefficients: (Intercept) xx1 xx2 xx3 xx4 xx5 13.551290 -7.615741 -7.424565 -7.533378 -7.571272 -14.527489 Scale= 0.755741 Loglik(model)= -1150.1 Loglik(intercept only)= -1153.9 Chisq= 7.52 on 5 degrees of freedom, p= 0.19 n= 228 > lfit5 Call: survreg(formula = Surv(time, status) ~ age, data = lung) Coefficients: (Intercept) age 6.88712062 -0.01360829 Scale= 0.7587515 Loglik(model)= -1151.9 Loglik(intercept only)= -1153.9 Chisq= 3.91 on 1 degrees of freedom, p= 0.048 n= 228 > lfit6 Call: survreg(formula = Surv(time, status) ~ pspline(age, df = 2), data = lung) coef se(coef) se2 Chisq DF p (Intercept) 6.5918 0.63681 0.41853 107.15 1.00 0.000 pspline(age, df = 2), lin -0.0136 0.00687 0.00687 3.94 1.00 0.047 pspline(age, df = 2), non 0.78 1.06 0.400 Scale= 0.756 Iterations: 4 outer, 12 Newton-Raphson Theta= 0.926 Degrees of freedom for terms= 0.4 2.1 1.0 Likelihood ratio test=5.2 on 1.5 df, p=0.0441 n= 228 > signif(lfit7$coef,6) (Intercept) 1.47899e-09 > survival/tests/r_peterson.Rout.save0000644000175100001440000001005211732700061017270 0ustar hornikusers R version 2.14.0 Under development (unstable) (2011-04-10 r55401) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Data courtesy of Bercedis Peterson, Duke University. > # v4 of survreg fails due to 2 groups that have only 1 subject; the coef > # for them easily gets out of hand. In fact, this data set is my toughest > # test of the minimizer. > # > # A shrinkage model for this coefficient is therefore interesting > > > peterson <- data.frame( + scan('data.peterson', what=list(grp=0, time=0, status=0))) Read 19 records > > fitp <- survreg(Surv(time, status) ~ factor(grp), peterson) > summary(fitp) Call: survreg(formula = Surv(time, status) ~ factor(grp), data = peterson) Value Std. Error z p (Intercept) 2.291 0.115 19.92 2.93e-88 factor(grp)2 0.786 0.177 4.44 8.79e-06 factor(grp)3 0.728 0.183 3.97 7.09e-05 factor(grp)4 -1.598 0.218 -7.32 2.48e-13 factor(grp)5 -0.500 0.218 -2.29 2.21e-02 factor(grp)6 0.475 0.170 2.79 5.23e-03 Log(scale) -1.684 0.257 -6.54 6.09e-11 Scale= 0.186 Weibull distribution Loglik(model)= -26.7 Loglik(intercept only)= -40.7 Chisq= 28.18 on 5 degrees of freedom, p= 3.4e-05 Number of Newton-Raphson Iterations: 9 n= 19 > > # Now a shrinkage model. Give the group coefficients > # about 1/2 the scale parameter of the original model, i.e., .18. > # > ffit <- survreg(Surv(time, status) ~ frailty(grp, theta=.1), peterson) > ffit Call: survreg(formula = Surv(time, status) ~ frailty(grp, theta = 0.1), data = peterson) coef se(coef) se2 Chisq DF p (Intercept) 2.62 0.172 0.0874 232.0 1.00 0.0000 frailty(grp, theta = 0.1) 10.4 2.15 0.0067 Scale= 0.301 Iterations: 1 outer, 6 Newton-Raphson Variance of random effect= 0.1 I-likelihood = -11.8 Degrees of freedom for terms= 0.3 2.2 0.7 Likelihood ratio test=13.8 on 1.1 df, p=0.00027 n= 19 > > # > # Try 3 degrees of freedom, since there are 6 groups > # compare them to the unconstrained ones. The frailty coefs are > # on a "sum to constant" constraint rather than "first coef=0", so > # some conversion is neccessary > # > ffit3 <- survreg(Surv(time, status) ~ frailty(grp, df=3), peterson) > print(ffit3) Call: survreg(formula = Surv(time, status) ~ frailty(grp, df = 3), data = peterson) coef se(coef) se2 Chisq DF p (Intercept) 2.54 0.187 0.0685 184.1 1.00 0.00000 frailty(grp, df = 3) 16.7 3.06 0.00088 Scale= 0.227 Iterations: 6 outer, 32 Newton-Raphson Variance of random effect= 0.17 I-likelihood = -10.1 Degrees of freedom for terms= 0.1 3.1 0.3 Likelihood ratio test=22.9 on 1.5 df, p=4.58e-06 n= 19 > > temp <- mean(c(0, fitp$coef[-1])) - mean(ffit3$frail) > temp2 <- c(fitp$coef[1] + temp, c(0,fitp$coef[-1]) - temp) > xx <- rbind(c(nrow(peterson), table(peterson$grp)), + temp2, + c(ffit3$coef, ffit3$frail)) > dimnames(xx) <- list(c("N", "factor", "frailty"), + c("Intercept", paste("grp", 1:6))) > signif(xx,3) Intercept grp 1 grp 2 grp 3 grp 4 grp 5 grp 6 N 19.00 3.000 6.000 6.000 1.00 1.000 2.000 factor 2.43 -0.137 0.649 0.591 -1.74 -0.636 0.338 frailty 2.54 -0.255 0.474 0.438 -1.21 -0.554 0.180 > > rm(ffit, ffit3, temp, temp2, xx, fitp) > survival/tests/mrtest.Rout.save0000644000175100001440000000312711732700061016433 0ustar hornikusers R version 2.7.1 (2008-06-23) Copyright (C) 2008 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > {if (is.R()) mdy.date <- function(m, d, y) { + y <- ifelse(y<100, y+1900, y) + as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") + } + else mdy.date <- function(m,d,y) { + y <- ifelse(y<100, y+1900, y) + timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") + } + } > > # > # A test of the match.ratetable function, specifically the > # change to allow partial matching of strings > # Note that 10,000 days old is 27.4 years > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > temp1 <- data.frame(year=mdy.date(2,2,1960:1964), + age = 10000 + 1:5, + sex = c('M', 'fema', 'f', 'ma', 'F')) > > temp2 <- ratetable(year=temp1$year, age=temp1$age, sex=temp1$sex) > temp3 <- match.ratetable(temp2, survexp.us) > aeq(temp3$R[,2], c(1,2,2,1,2)) [1] TRUE > survival/tests/aareg.Rout.save0000644000175100001440000003011611732700061016172 0ustar hornikusers R version 2.14.0 Under development (unstable) (2011-04-10 r55401) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # Test aareg, for some simple data where the answers can be computed > # in closed form > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > test1 <- data.frame(time= c(4, 3,1,1,2,2,3), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0), + wt= c(1, 1:6)) > > tfit <- aareg(Surv(time, status) ~ x, test1) > aeq(tfit$times, c(1,2,2)) [1] TRUE > aeq(tfit$nrisk, c(6,4,4)) [1] TRUE > aeq(tfit$coefficient, matrix(c(0,0,1/3, 1/3, 1, -1/3), ncol=2)) [1] TRUE > aeq(tfit$tweight, matrix(c(3,3,3, 3/2, 3/4, 3/4), ncol=2)) [1] TRUE > aeq(tfit$test.statistic, c(1,1)) [1] TRUE > aeq(tfit$test.var, c(1, -1/4, -1/4, 1/4 + 9/16 + 1/16)) [1] TRUE > > tfit <- aareg(Surv(time, status) ~ x, test1, test='nrisk') > aeq(tfit$tweight, matrix(c(3,3,3, 3/2, 3/4, 3/4), ncol=2)) #should be as before [1] TRUE > aeq(tfit$test.statistic, c(4/3, 6/3+ 4 - 4/3)) [1] TRUE > aeq(tfit$test.var, c(16/9, -16/9, -16/9, 36/9 + 16 + 16/9)) [1] TRUE > > # In the 1-variable case, this is the same as the default Aalen weight > tfit <- aareg(Surv(time, status) ~ x, test1, test='variance') > aeq(tfit$test.statistic, c(1,1)) [1] TRUE > aeq(tfit$test.var, c(1, -1/4, -1/4, 1/4 + 9/16 + 1/16)) [1] TRUE > > # > # Repeat the above, with case weights > # > tfit <- aareg(Surv(time, status) ~x, test1, weights=wt) > aeq(tfit$times, c(1,2,2)) [1] TRUE > aeq(tfit$nrisk, c(21,16,16)) [1] TRUE > aeq(tfit$coefficient, matrix(c(0,0,5/12, 2/9, 1, -5/12), ncol=2)) [1] TRUE > aeq(tfit$tweight, matrix(c(12,12,12, 36/7, 3,3), ncol=2)) [1] TRUE > aeq(tfit$test.statistic, c(5, 72/63 + 3 - 15/12)) [1] TRUE > aeq(tfit$test.var, c(25, -25/4, -25/4, (72/63)^2 + 9 + (5/4)^2)) [1] TRUE > > tfit <- aareg(Surv(time, status) ~x, test1, weights=wt, test='nrisk') > aeq(tfit$test.statistic, c(20/3, 42/9 + 16 - 16*5/12)) [1] TRUE > aeq(tfit$test.var, c(400/9, -400/9, -400/9, + (42/9)^2 + 16^2 + (16*5/12)^2)) [1] TRUE > > # > # Make a test data set with no NAs, in sorted order, no ties, > # 15 observations > tdata <- lung[15:29, c('time', 'status', 'age', 'sex', 'ph.ecog')] > tdata$status <- tdata$status -1 > tdata <- tdata[order(tdata$time, tdata$status),] > row.names(tdata) <- 1:15 > tdata$status[8] <- 0 #for some variety > > afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, tdata, nmin=6) > # > # Now, do it "by hand" > cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, tdata, iter=0, + method='breslow') > dt1 <- coxph.detail(cfit) > sch1 <- resid(cfit, type='schoen') > > # First estimate of Aalen: from the Cox computations, first 9 > # The first and last cols of the ninth are somewhat unstable (approx =0) > mine <- rbind(solve(dt1$imat[,,1], sch1[1,]), + solve(dt1$imat[,,2], sch1[2,]), + solve(dt1$imat[,,3], sch1[3,]), + solve(dt1$imat[,,4], sch1[4,]), + solve(dt1$imat[,,5], sch1[5,]), + solve(dt1$imat[,,6], sch1[6,]), + solve(dt1$imat[,,7], sch1[7,]), + solve(dt1$imat[,,8], sch1[8,]), + solve(dt1$imat[,,9], sch1[9,])) > mine <- diag(1/dt1$nrisk[1:9]) %*% mine > > aeq(mine, afit$coef[1:9, -1]) [1] TRUE > > rm(tfit, afit, mine, dt1, cfit, sch1) > > # > # Check out the dfbeta matrix from aareg > # Note that it is kept internally in time order, not data set order > # Those who want residuals should use the resid function! > > # > # First, the simple test case where I know the anwers > # > afit <- aareg(Surv(time, status) ~ x, test1, dfbeta=T) > temp <- c(rep(0,6), #intercepts at time 1 + c(2,-1,-1,0,0,0)/9, #alpha at time 1 + c(0,0,0,2, -1, -1)/9, #intercepts at time 2 + c(0,0,0,-2,1,1)/9) #alpha at time 2 > aeq(afit$dfbeta, temp) [1] TRUE > > # > #Now a multivariate data set > # > afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, lung, dfbeta=T) > > ord <- order(lung$time, -lung$status) > cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, lung[ord,], + method='breslow', iter=0, x=T) > cdt <- coxph.detail(cfit, riskmat=T) > > # an arbitrary list of times > acoef <- rowsum(afit$coef, afit$times) #per death time coefs > indx <- match(cdt$time, afit$times) > for (i in c(2,5,27,54,101, 135)) { + lwho <- (cdt$riskmat[,i]==1) + lmx <- cfit$x[lwho,] + lmy <- 1*( cfit$y[lwho,2]==1 & cfit$y[lwho,1] == cdt$time[i]) + fit <- lm(lmy~ lmx) + cat("i=", i, "coef=", aeq(fit$coef, acoef[i,])) + + rr <- diag(resid(fit)) + zz <- cbind(1,lmx) + zzinv <- solve(t(zz) %*% zz) + cat(" twt=", aeq(1/(diag(zzinv)), afit$tweight[indx[i],])) + + df <- t(zzinv %*% t(zz) %*% rr) + cat(" dfbeta=", aeq(df, afit$dfbeta[lwho,,i]), "\n") + } i= 2 coef= TRUE twt= TRUE dfbeta= TRUE i= 5 coef= TRUE twt= TRUE dfbeta= TRUE i= 27 coef= TRUE twt= TRUE dfbeta= TRUE i= 54 coef= TRUE twt= TRUE dfbeta= TRUE i= 101 coef= TRUE twt= TRUE dfbeta= TRUE i= 135 coef= TRUE twt= TRUE dfbeta= TRUE > > rm(afit, cfit, cdt, lwho, lmx, lmy, fit, rr, zz, df) > > > # Repeat it with case weights > ww <- rep(1:5, length=nrow(lung))/ 3.0 > afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, lung, dfbeta=T, + weights=ww) > cfit <- coxph(Surv(time, status) ~ age + sex + ph.ecog, lung[ord,], + method='breslow', iter=0, x=T, weight=ww[ord]) > cdt <- coxph.detail(cfit, riskmat=T) > > acoef <- rowsum(afit$coef, afit$times) #per death time coefs > for (i in c(2,5,27,54,101, 135)) { + who <- (cdt$riskmat[,i]==1) + x <- cfit$x[who,] + y <- 1*( cfit$y[who,2]==1 & cfit$y[who,1] == cdt$time[i]) + w <- cfit$weight[who] + fit <- lm(y~x, weights=w) + cat("i=", i, "coef=", aeq(fit$coef, acoef[i,])) + + rr <- diag(resid(fit)) + zz <- cbind(1,x) + zzinv <- solve(t(zz)%*% (w*zz)) + cat(" twt=", aeq(1/(diag(zzinv)), afit$tweight[indx[i],])) + + df <- t(zzinv %*% t(zz) %*% (w*rr)) + cat(" dfbeta=", aeq(df, afit$dfbeta[who,,i]), "\n") + } i= 2 coef= TRUE twt= TRUE dfbeta= TRUE i= 5 coef= TRUE twt= TRUE dfbeta= TRUE i= 27 coef= TRUE twt= TRUE dfbeta= TRUE i= 54 coef= TRUE twt= TRUE dfbeta= TRUE i= 101 coef= TRUE twt= TRUE dfbeta= TRUE i= 135 coef= TRUE twt= TRUE dfbeta= TRUE > > rm(afit, cfit, cdt, who, x, y, fit, rr, zz, df) > rm(ord, acoef) > > # > # Check that the test statistic computed within aareg and > # the one recomputed within summary.aareg are the same. > # Of course, they could both be wrong, but at least they'll agree! > # If the maxtime argument is used in summary, it recomputes the test, > # even if we know that it wouldn't have had to. > # > # Because the 1-variable and >1 variable case have different code, test > # them both. > # > afit <- aareg(Surv(time, status) ~ age, lung, dfbeta=T) > asum <- summary(afit, maxtime=max(afit$times)) > aeq(afit$test.stat, asum$test.stat) [1] TRUE > aeq(afit$test.var, asum$test.var) [1] TRUE > aeq(afit$test.var2, asum$test.var2) [1] TRUE > > print(afit) Call: aareg(formula = Surv(time, status) ~ age, data = lung, dfbeta = T) n= 228 139 out of 139 unique event times used slope coef se(coef) robust se z p Intercept -0.000872 -0.000905 4.26e-03 4.13e-03 -0.219 0.8270 age 0.000110 0.000142 6.96e-05 6.75e-05 2.110 0.0351 Chisq=4.44 on 1 df, p=0.035; test weights=aalen > > afit <- aareg(Surv(time, status) ~ age, lung, dfbeta=T, test='nrisk') > asum <- summary(afit, maxtime=max(afit$times)) > aeq(afit$test.stat, asum$test.stat) [1] TRUE > aeq(afit$test.var, asum$test.var) [1] TRUE > aeq(afit$test.var2, asum$test.var2) [1] TRUE > > summary(afit) $table slope coef se(coef) robust se z Intercept -0.0009538483 -0.11693804 0.534885651 0.533148054 -0.219335 age 0.0001053024 0.01795521 0.008746523 0.008734005 2.055782 p Intercept 0.82638908 age 0.03980352 $test [1] "nrisk" $test.statistic Intercept age -19.29478 2.96261 $test.var [,1] [,2] [1,] 7789.1449 -126.055872 [2,] -126.0559 2.082758 $test.var2 [,1] [,2] [1,] 7738.6204 -125.5077 [2,] -125.5077 2.0768 $chisq [,1] [1,] 4.22624 $n [1] 228 139 139 attr(,"class") [1] "summary.aareg" > > # > # Mulitvariate > # > afit <- aareg(Surv(time, status) ~ age + sex + ph.karno + pat.karno, lung, + dfbeta=T) > asum <- summary(afit, maxtime=max(afit$times)) > aeq(afit$test.stat, asum$test.stat) [1] TRUE > aeq(afit$test.var, asum$test.var) [1] TRUE > aeq(afit$test.var2, asum$test.var2) [1] TRUE > > print(afit) Call: aareg(formula = Surv(time, status) ~ age + sex + ph.karno + pat.karno, data = lung, dfbeta = T) n=224 (4 observations deleted due to missingness) 132 out of 136 unique event times used slope coef se(coef) robust se z p Intercept 2.15e-02 0.025000 8.45e-03 7.72e-03 3.25 0.00117 age 3.09e-05 0.000076 7.32e-05 6.49e-05 1.17 0.24100 sex -2.96e-03 -0.004020 1.25e-03 1.23e-03 -3.27 0.00109 ph.karno -6.77e-05 -0.000083 6.69e-05 8.30e-05 -1.00 0.31700 pat.karno -1.01e-04 -0.000112 5.59e-05 5.70e-05 -1.96 0.05010 Chisq=23.36 on 4 df, p=0.00011; test weights=aalen > > afit <- aareg(Surv(time, status) ~ age + sex + ph.karno + pat.karno, lung, + dfbeta=T, test='nrisk') > asum <- summary(afit, maxtime=max(afit$times)) > aeq(afit$test.stat, asum$test.stat) [1] TRUE > aeq(afit$test.var, asum$test.var) [1] TRUE > aeq(afit$test.var2, asum$test.var2) [1] TRUE > > summary(afit) $table slope coef se(coef) robust se z Intercept 2.119015e-02 3.05872822 1.044992929 0.955953617 3.199662 age 3.181122e-05 0.01071085 0.009280348 0.008182931 1.308926 sex -2.985556e-03 -0.49368373 0.153217001 0.151559500 -3.257359 ph.karno -8.371983e-05 -0.01131957 0.007825769 0.009654398 -1.172478 pat.karno -8.501076e-05 -0.01328844 0.007241150 0.007669582 -1.732617 p Intercept 0.00137589 age 0.19055946 sex 0.00112454 ph.karno 0.24100515 pat.karno 0.08316385 $test [1] "nrisk" $test.statistic Intercept age sex ph.karno pat.karno 480.220330 1.681604 -77.508345 -1.777173 -2.086286 $test.var b0 b0 26916.95995 -177.3767597 -791.4141458 -103.5540756 -69.1210402 -177.37676 2.1228915 0.1752574 0.4055099 0.1622945 -791.41415 0.1752574 578.6463538 -0.9726495 -0.6320578 -103.55408 0.4055099 -0.9726495 1.5095704 -0.5793466 -69.12104 0.1622945 -0.6320578 -0.5793466 1.2924520 $test.var2 [,1] [,2] [,3] [,4] [,5] [1,] 22525.42254 -109.0376340 -1294.620657 -135.7477106 -24.1718358 [2,] -109.03763 1.6505060 2.562655 0.1774270 -0.1206339 [3,] -1294.62066 2.5626546 566.194480 7.4865489 -4.7691882 [4,] -135.74771 0.1774270 7.486549 2.2974694 -0.9877341 [5,] -24.17184 -0.1206339 -4.769188 -0.9877341 1.4499155 $chisq [,1] [1,] 22.3874 $n [1] 224 132 136 attr(,"class") [1] "summary.aareg" > > # Weights play no role in the final computation of the test statistic, given > # the coefficient matrix, nrisk, and dfbeta as inputs. (Weights do > # change the inputs). So there is no need to reprise the above with > # case weights. > survival/tests/surv.R0000644000175100001440000000156012257335007014435 0ustar hornikusers# library(survival) # Some simple tests of the Surv function # The first two are motivated by a bug, pointed out by Kevin Buhr, # where a mixture of NAs and invalid values didn't work right # Even for the simplest things a test case is good. # All but the third should produce warning messages aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) temp <- Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), c(1,1,1,1)) aeq(temp, c(1,10,NA,30, 2,NA,0,40, 1,1,1,1)) temp <- Surv(c(1, 10, 20, 30), c(2, NA, 0, 40), type='interval2') aeq(temp, c(1,10,20,30, 2,1,1,40, 3,0,NA,3)) #No error temp <- Surv(1:5) aeq(temp, c(1:5, 1,1,1,1,1)) temp1 <- Surv(c(1,10,NA, 30, 30), c(1,NA,10,20, 40), type='interval2') temp2 <- Surv(c(1,10,10,30,30), c(9, NA, 5, 20,40), c(1, 0, 2,3,3), type='interval') aeq(temp1, temp2) aeq(temp1, c(1,10,10,30,30, 1,1,1,1, 40, 1,0,2,NA,3)) survival/tests/factor.Rout.save0000644000175100001440000000342011732700061016367 0ustar hornikusers R version 2.14.0 (2011-10-31) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > # > # Ensure that factors work in prediction > # > library(survival) Loading required package: splines > > options(na.action="na.exclude") # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > tfit <- coxph(Surv(time, status) ~ age + factor(ph.ecog), lung) > p1 <- predict(tfit, type='risk') > > # Testing NA handling is important too > keep <- (is.na(lung$ph.ecog) | lung$ph.ecog !=1) > lung2 <- lung[keep,] > p2 <- predict(tfit, type='risk', newdata=lung[keep,]) > aeq(p1[keep], p2) [1] TRUE > > # Same, for survreg > tfit <- survreg(Surv(time, status) ~ age + factor(ph.ecog), lung) > p1 <- predict(tfit, type='response') > p2 <- predict(tfit, type='response', newdata=lung2) > aeq(p1[keep], p2) [1] TRUE > > > # Now repeat it tossing the missings > options(na.action=na.omit) > keep2 <- (lung$ph.ecog[!is.na(lung$ph.ecog)] !=1) > > tfit2 <- survreg(Surv(time, status) ~ age + factor(ph.ecog), lung) > p3 <- predict(tfit2, type='response') > p4 <- predict(tfit2, type='response', newdata=lung2, na.action=na.omit) > aeq(p3[keep2] , p4) [1] TRUE > survival/tests/pyear.R0000644000175100001440000001630411732700061014551 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) {if (is.R()) mdy.date <- function(m, d, y) { y <- ifelse(y<100, y+1900, y) as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") } else mdy.date <- function(m,d,y) { y <- ifelse(y<100, y+1900, y) timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") } } # # Simple case: a single male subject, born 6/6/36 and entered on study 6/6/55. # temp1 <- mdy.date(6,6,36) temp2 <- mdy.date(6,6,55)# Now compare the results from person-years # temp.age <- tcut(temp2-temp1, floor(c(-1, (18:31 * 365.24))), labels=c('0-18', paste(18:30, 19:31, sep='-'))) temp.yr <- tcut(temp2, mdy.date(1,1,1954:1965), labels=1954:1964) temp.time <- 3700 #total days of fu py1 <- pyears(temp.time ~ temp.age + temp.yr, scale=1) #output in days # The subject should appear in 20 cells # 6/6/55 - 12/31/55, 209 days, age 19-20, 1955 # 1/1/56 - 6/ 4/56, 156 days, age 19-20, 1956 # 6/5/56 - 12/31/56, 210 days, age 20-21, 1956 (a leap year, and his # birthday computes one day earlier) # 1/1/57 - 6/ 5/57, 156 days, age 20-21, 1957 # 6/6/57 - 12/31/57, 209 days, age 21-22, 1957 # and etc # with 203 days "off table", ie, beyond the last cell of the table # # It is a nuisance, but tcut follows 'cut' in that we give the ENDS of # the intervals, whereas the survival tables use the starts of intervals. # Thus this breakdown does not match that in doexpect.s # xx <- matrix(0, nrow=14, ncol=11) xx[cbind(3:11, 3:11)] <- 156 xx[cbind(3:12, 2:11)] <- c(209, 210, rep(c(209, 209, 209, 210),2)) dimnames(xx) <- list(c('0-18', paste(18:30, 19:31, sep='-')), 1954:1964) all.equal(xx, py1$pyears) all.equal(203, py1$offtable) all.equal(1*(xx>0), py1$n) # # Now with expecteds # py2 <- pyears(temp.time ~ temp.age + temp.yr + ratetable(age=temp2-temp1, year=temp2, sex=1), scale=1, ratetable=survexp.us ) #output in days all.equal(xx, py2$pyears) all.equal(203, py2$offtable) all.equal(1*(xx>0), py2$n) py2b <- pyears(temp.time ~ temp.age + temp.yr, rmap = list(age=temp2-temp1, year=temp2, sex=1), scale=1, ratetable=survexp.us ) #output in days all.equal(xx, py2b$pyears) all.equal(203, py2b$offtable) all.equal(1*(xx>0), py2b$n) all.equal(py2$expected, py2b$expected) py3 <- pyears(temp.time ~ temp.age + temp.yr, rmap=list(age=temp2-temp1, year=temp2, sex=1), scale=1, ratetable=survexp.us , expect='pyears') all.equal(py2$n, py3$n) all.equal(py2$pyear, py3$pyear) all.equal(py3$n, 1*(py3$expect>0)) # Now, compute the py3 result "by hand". Since there is only one person # it can be derived from py2. # xx1 <- py2$expect[py2$n>0] # the hazard over each interval cumhaz <- cumsum(c(0, xx1[-length(xx1)])) # the cumulative hazard xx2 <- py3$expect[py3$n>0] # the expected number of person days xx3 <- py3$pyears[py3$n>0] # the potential number of person days # This is the integral of the curve "exp(-haz *t)" over the interval integral <- xx3 * exp(-cumhaz)* (1- exp(-xx1))/ xx1 # They might not be exactly equal, since the C code tracks changes in the # rate tables that occur -within- an interval. So try for 6 digits all.equal(round(integral,3), round(xx2,3)) # Cut off the bottom of the table, instead of the side temp.age <- tcut(temp2-temp1, floor(c(-1, (18:27 * 365.24))), labels=c('0-18', paste(18:26, 19:27, sep='-'))) py4 <- eval(py3$call) all.equal(py4$pyear, py3$pyear[1:10,]) all.equal(py4$expect, py3$expect[1:10,]) rm(temp.age, integral, xx1, xx2, xx3, cumhaz, py1, py2, py3, py4) rm(temp1, temp2, temp.yr, temp.time, xx) # # Simple case: a single male subject, born 6/6/36 and entered on study 6/6/55. # temp1 <- mdy.date(6,6,36) temp2 <- mdy.date(6,6,55)# Now compare the results from person-years # temp.age <- tcut(temp2-temp1, floor(c(-1, (18:31 * 365.24))), labels=c('0-18', paste(18:30, 19:31, sep='-'))) temp.yr <- tcut(temp2, mdy.date(1,1,1954:1965), labels=1954:1964) temp.time <- 3700 #total days of fu py1 <- pyears(temp.time ~ temp.age + temp.yr, scale=1) #output in days # The subject should appear in 20 cells # 6/6/55 - 12/31/55, 209 days, age 19-20, 1955 # 1/1/56 - 6/ 4/56, 156 days, age 19-20, 1956 # 6/5/56 - 12/31/56, 210 days, age 20-21, 1956 (a leap year, and his # birthday computes one day earlier) # 1/1/57 - 6/ 5/57, 156 days, age 20-21, 1957 # 6/6/57 - 12/31/57, 209 days, age 21-22, 1957 # and etc # with 203 days "off table", ie, beyond the last cell of the table # # It is a nuisance, but tcut follows 'cut' in that we give the ENDS of # the intervals, whereas the survival tables use the starts of intervals. # xx <- matrix(0, nrow=14, ncol=11) xx[cbind(3:11, 3:11)] <- 156 xx[cbind(3:12, 2:11)] <- c(209, 210, rep(c(209, 209, 209, 210),2)) dimnames(xx) <- list(c('0-18', paste(18:30, 19:31, sep='-')), 1954:1964) all.equal(xx, py1$pyears) all.equal(203, py1$offtable) all.equal(1*(xx>0), py1$n) # # Now with expecteds # py2 <- pyears(temp.time ~ temp.age + temp.yr + ratetable(age=temp2-temp1, year=temp2, sex=1), scale=1, ratetable=survexp.us ) #output in days all.equal(xx, py2$pyears) all.equal(203, py2$offtable) all.equal(1*(xx>0), py2$n) py3 <- pyears(temp.time ~ temp.age + temp.yr + ratetable(age=temp2-temp1, year=temp2, sex=1), scale=1, ratetable=survexp.us , expect='pyears') all.equal(py2$n, py3$n) all.equal(py2$pyear, py3$pyear) all.equal(py3$n, 1*(py3$expect>0)) # Now, compute the py3 result "by hand". Since there is only one person # it can be derived from py2. # xx1 <- py2$expect[py2$n>0] # the hazard over each interval cumhaz <- cumsum(c(0, xx1[-length(xx1)])) # the cumulative hazard xx2 <- py3$expect[py3$n>0] # the expected number of person days xx3 <- py3$pyears[py3$n>0] # the potential number of person days # This is the integral of the curve "exp(-haz *t)" over the interval integral <- xx3 * exp(-cumhaz)* (1- exp(-xx1))/ xx1 # They might not be exactly equal, since the C code tracks changes in the # rate tables that occur -within- an interval. So try for 6 digits all.equal(round(integral,3), round(xx2,3)) # Cut off the bottom of the table, instead of the side temp.age <- tcut(temp2-temp1, floor(c(-1, (18:27 * 365.24))), labels=c('0-18', paste(18:26, 19:27, sep='-'))) py4 <- eval(py3$call) all.equal(py4$pyear, py3$pyear[1:10,]) all.equal(py4$expect, py3$expect[1:10,]) rm(temp.age, integral, xx1, xx2, xx3, cumhaz, py1, py2, py3, py4) rm(temp1, temp2, temp.yr, temp.time, xx) # # Create a "user defined" rate table, using the smoking data # temp <- scan("data.smoke")/100000 temp <- matrix(temp, ncol=8, byrow=T) smoke.rate <- c(rep(temp[,1],6), rep(temp[,2],6), temp[,3:8]) attributes(smoke.rate) <- list( dim=c(7,2,2,6,3), dimnames=list(c("45-49","50-54","55-59","60-64","65-69","70-74","75-79"), c("1-20", "21+"), c("Male","Female"), c("<1", "1-2", "3-5", "6-10", "11-15", ">=16"), c("Never", "Current", "Former")), dimid=c("age", "amount", "sex", "duration", "status"), factor=c(0,1,1,0,1), cutpoints=list(c(45,50,55,60,65,70,75),NULL, NULL, c(0,1,3,6,11,16),NULL), class='ratetable' ) rm(temp) is.ratetable(smoke.rate) summary(smoke.rate) print(smoke.rate) summary(smoke.rate[1:3,,1,,]) #test subscripting survival/tests/fr_resid.Rout.save0000644000175100001440000003651612536400735016732 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # > # The residual methods treat a sparse frailty as a fixed offset with > # no variance > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > kfit1 <- coxph(Surv(time, status) ~ age + sex + + frailty(id, dist='gauss'), kidney) > tempf <- predict(kfit1, type='terms')[,3] > temp <- kfit1$frail[match(kidney$id, sort(unique(kidney$id)))] > #all.equal(unclass(tempf), unclass(temp)) > all.equal(as.vector(tempf), as.vector(temp)) [1] TRUE > > # Now fit a model with explicit offset > kfitx <- coxph(Surv(time, status) ~ age + sex + offset(tempf),kidney, + eps=1e-7) > > # These are not always precisely the same, due to different iteration paths > aeq(kfitx$coef, kfit1$coef) [1] TRUE > > # This will make them identical > kfitx <- coxph(Surv(time, status) ~ age + sex + offset(temp),kidney, + iter=0, init=kfit1$coef) > aeq(resid(kfit1), resid(kfitx)) [1] TRUE > aeq(resid(kfit1, type='score'), resid(kfitx, type='score')) [1] TRUE > aeq(resid(kfit1, type='schoe'), resid(kfitx, type='schoe')) [1] TRUE > > # These are not the same, due to a different variance matrix > # The frailty model's variance is about 2x the naive "assume an offset" var > # Expect a value of about 0.5 > aeq(resid(kfit1, type='dfbeta'), resid(kfitx, type='dfbeta')) [1] "Mean relative difference: 0.5216263" > > # Force equality > zed <- kfitx > zed$var <- kfit1$var > aeq(resid(kfit1, type='dfbeta'), resid(zed, type='dfbeta')) [1] TRUE > > # The score residuals are equal, however. > > temp1 <- resid(kfit1, type='score') > temp2 <- resid(kfitx, type='score') > aeq(temp1, temp2) [1] TRUE > > # > # Now for some tests of predicted values > # > aeq(predict(kfit1, type='expected'), predict(kfitx, type='expected')) [1] TRUE > aeq(predict(kfit1, type='lp'), predict(kfitx, type='lp')) [1] TRUE > > temp1 <- predict(kfit1, type='terms', se.fit=T) > temp2 <- predict(kfitx, type='terms', se.fit=T) > aeq(temp1$fit[,1:2], temp2$fit) [1] TRUE > # the next is not equal, all.equal returns a character string in that case > is.character(aeq(temp1$se.fit[,1:2], temp2$se.fit)) [1] TRUE > mean(temp1$se.fit[,1:2]/ temp2$se.fit) [1] 1.433017 > aeq(as.vector(temp1$se.fit[,3])^2, + as.vector(kfit1$fvar[match(kidney$id, sort(unique(kidney$id)))])) [1] TRUE > > print(temp1) $fit age sex frailty(id, dist = "gauss") 1 -0.073981042 1.039553 0.59814468 2 -0.073981042 1.039553 0.59814468 3 0.020278123 -0.371269 0.38512389 4 0.020278123 -0.371269 0.38512389 5 -0.055129209 1.039553 0.20210998 6 -0.055129209 1.039553 0.20210998 7 -0.059842167 -0.371269 -0.55932015 8 -0.055129209 -0.371269 -0.55932015 9 -0.158814289 1.039553 0.28558387 10 -0.158814289 1.039553 0.28558387 11 -0.130536540 -0.371269 0.06628942 12 -0.125823582 -0.371269 0.06628942 13 0.034416998 1.039553 0.80505119 14 0.034416998 1.039553 0.80505119 15 0.053268830 -0.371269 -0.43834241 16 0.057981789 -0.371269 -0.43834241 17 0.119250245 -0.371269 -0.05631649 18 0.119250245 -0.371269 -0.05631649 19 0.034416998 1.039553 -0.49980572 20 0.039129956 1.039553 -0.49980572 21 0.001426290 -0.371269 -0.13028264 22 0.001426290 -0.371269 -0.13028264 23 -0.045703292 -0.371269 0.06377401 24 -0.045703292 -0.371269 0.06377401 25 -0.040990334 -0.371269 0.38815296 26 -0.040990334 -0.371269 0.38815296 27 -0.007999626 -0.371269 -0.47650510 28 -0.007999626 -0.371269 -0.47650510 29 -0.125823582 -0.371269 -0.66986830 30 -0.125823582 -0.371269 -0.66986830 31 0.076833621 1.039553 0.19359678 32 0.076833621 1.039553 0.19359678 33 0.076833621 -0.371269 -0.16483200 34 0.076833621 -0.371269 -0.16483200 35 -0.003286668 -0.371269 -0.15794998 36 0.001426290 -0.371269 -0.15794998 37 0.043842914 -0.371269 -0.46236014 38 0.043842914 -0.371269 -0.46236014 39 0.001426290 -0.371269 0.12603308 40 0.001426290 -0.371269 0.12603308 41 0.010852206 1.039553 -1.74303142 42 0.015565165 1.039553 -1.74303142 43 -0.064555125 -0.371269 -0.45211210 44 -0.064555125 -0.371269 -0.45211210 45 0.086259538 -0.371269 0.51574106 46 0.090972496 -0.371269 0.51574106 47 -0.007999626 -0.371269 0.09475123 48 -0.003286668 -0.371269 0.09475123 49 -0.003286668 1.039553 0.05790354 50 -0.003286668 1.039553 0.05790354 51 0.062694747 -0.371269 -0.37933234 52 0.067407705 -0.371269 -0.37933234 53 -0.158814289 -0.371269 0.11248891 54 -0.158814289 -0.371269 0.11248891 55 0.039129956 -0.371269 0.54791210 56 0.039129956 -0.371269 0.54791210 57 0.043842914 1.039553 0.45873482 58 0.043842914 1.039553 0.45873482 59 0.048555872 -0.371269 0.35639797 60 0.048555872 -0.371269 0.35639797 61 0.057981789 -0.371269 0.48803342 62 0.057981789 -0.371269 0.48803342 63 0.029704039 -0.371269 0.25597325 64 0.034416998 -0.371269 0.25597325 65 0.062694747 -0.371269 0.23054948 66 0.062694747 -0.371269 0.23054948 67 0.001426290 -0.371269 -0.13680005 68 0.006139248 -0.371269 -0.13680005 69 -0.102258791 -0.371269 0.51977995 70 -0.102258791 -0.371269 0.51977995 71 -0.007999626 -0.371269 -0.23878154 72 -0.007999626 -0.371269 -0.23878154 73 0.039129956 -0.371269 0.17174306 74 0.039129956 -0.371269 0.17174306 75 0.076833621 1.039553 -0.35822829 76 0.076833621 1.039553 -0.35822829 $se.fit age sex frailty(id, dist = "gauss") 1 0.195861829 0.3280279 0.6246430 2 0.195861829 0.3280279 0.6246430 3 0.053685514 0.1171528 0.6954922 4 0.053685514 0.1171528 0.6954922 5 0.145952360 0.3280279 0.5705340 6 0.145952360 0.3280279 0.5705340 7 0.158429727 0.1171528 0.4894541 8 0.145952360 0.1171528 0.4894541 9 0.420454437 0.3280279 0.6071455 10 0.420454437 0.3280279 0.6071455 11 0.345590234 0.1171528 0.5633997 12 0.333112867 0.1171528 0.5633997 13 0.091117615 0.3280279 0.6641707 14 0.091117615 0.3280279 0.6641707 15 0.141027084 0.1171528 0.5101890 16 0.153504451 0.1171528 0.5101890 17 0.315710223 0.1171528 0.5491569 18 0.315710223 0.1171528 0.5491569 19 0.091117615 0.3280279 0.5264083 20 0.103594982 0.3280279 0.5264083 21 0.003776045 0.1171528 0.5180953 22 0.003776045 0.1171528 0.5180953 23 0.120997626 0.1171528 0.6208806 24 0.120997626 0.1171528 0.6208806 25 0.108520259 0.1171528 0.5811421 26 0.108520259 0.1171528 0.5811421 27 0.021178689 0.1171528 0.6247779 28 0.021178689 0.1171528 0.6247779 29 0.333112867 0.1171528 0.5615987 30 0.333112867 0.1171528 0.5615987 31 0.203413919 0.3280279 0.6532405 32 0.203413919 0.3280279 0.6532405 33 0.203413919 0.1171528 0.5247227 34 0.203413919 0.1171528 0.5247227 35 0.008701322 0.1171528 0.5106606 36 0.003776045 0.1171528 0.5106606 37 0.116072349 0.1171528 0.6284328 38 0.116072349 0.1171528 0.6284328 39 0.003776045 0.1171528 0.6320009 40 0.003776045 0.1171528 0.6320009 41 0.028730780 0.3280279 0.5235228 42 0.041208147 0.3280279 0.5235228 43 0.170907094 0.1171528 0.5492095 44 0.170907094 0.1171528 0.5492095 45 0.228368654 0.1171528 0.6058686 46 0.240846021 0.1171528 0.6058686 47 0.021178689 0.1171528 0.6267998 48 0.008701322 0.1171528 0.6267998 49 0.008701322 0.3280279 0.5526664 50 0.008701322 0.3280279 0.5526664 51 0.165981818 0.1171528 0.5556706 52 0.178459185 0.1171528 0.5556706 53 0.420454437 0.1171528 0.5849825 54 0.420454437 0.1171528 0.5849825 55 0.103594982 0.1171528 0.6081780 56 0.103594982 0.1171528 0.6081780 57 0.116072349 0.3280279 0.6010279 58 0.116072349 0.3280279 0.6010279 59 0.128549717 0.1171528 0.5762113 60 0.128549717 0.1171528 0.5762113 61 0.153504451 0.1171528 0.5982501 62 0.153504451 0.1171528 0.5982501 63 0.078640248 0.1171528 0.6614053 64 0.091117615 0.1171528 0.6614053 65 0.165981818 0.1171528 0.5609510 66 0.165981818 0.1171528 0.5609510 67 0.003776045 0.1171528 0.5844921 68 0.016253412 0.1171528 0.5844921 69 0.270726031 0.1171528 0.6089631 70 0.270726031 0.1171528 0.6089631 71 0.021178689 0.1171528 0.6795741 72 0.021178689 0.1171528 0.6795741 73 0.103594982 0.1171528 0.6421784 74 0.103594982 0.1171528 0.6421784 75 0.203413919 0.3280279 0.5779661 76 0.203413919 0.3280279 0.5779661 > kfit1 Call: coxph(formula = Surv(time, status) ~ age + sex + frailty(id, dist = "gauss"), data = kidney) coef se(coef) se2 Chisq DF p age 0.00471 0.01248 0.00856 0.14267 1.0 0.7056 sex -1.41082 0.44518 0.31504 10.04319 1.0 0.0015 frailty(id, dist = "gauss 26.54461 14.7 0.0294 Iterations: 6 outer, 39 Newton-Raphson Variance of random effect= 0.569 Degrees of freedom for terms= 0.5 0.5 14.7 Likelihood ratio test=47.5 on 15.7 df, p=4.65e-05 n= 76 > kfitx Call: coxph(formula = Surv(time, status) ~ age + sex + offset(temp), data = kidney, init = kfit1$coef, iter = 0) coef exp(coef) se(coef) z p age 0.00471 1.00472 0.00875 0.54 0.59 sex -1.41082 0.24394 0.30916 -4.56 5e-06 Likelihood ratio test=0 on 2 df, p=1 n= 76, number of events= 58 > > rm(temp1, temp2, kfitx, zed, tempf) > # > # The special case of a single sparse frailty > # > > kfit1 <- coxph(Surv(time, status) ~ frailty(id, dist='gauss'), kidney) > tempf <- predict(kfit1, type='terms') > temp <- kfit1$frail[match(kidney$id, sort(unique(kidney$id)))] > all.equal(as.vector(tempf), as.vector(temp)) [1] TRUE > > # Now fit a model with explicit offset > kfitx <- coxph(Surv(time, status) ~ offset(tempf),kidney, eps=1e-7) > > aeq(resid(kfit1), resid(kfitx)) [1] TRUE > aeq(resid(kfit1, type='deviance'), resid(kfitx, type='deviance')) [1] TRUE > > # > # Some tests of predicted values > # > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > aeq(predict(kfit1, type='expected'), predict(kfitx, type='expected')) [1] TRUE > aeq(predict(kfit1, type='lp'), predict(kfitx, type='lp')) [1] TRUE > > temp1 <- predict(kfit1, type='terms', se.fit=T) > aeq(temp1$fit, kfitx$linear) [1] TRUE > aeq(temp1$se.fit^2, + kfit1$fvar[match(kidney$id, sort(unique(kidney$id)))]) [1] TRUE > > temp1 $fit [1] 0.696003729 0.696003729 0.244575316 0.244575316 0.494175549 [6] 0.494175549 -0.659248798 -0.659248798 0.521423106 0.521423106 [11] -0.114492938 -0.114492938 0.800127481 0.800127481 -0.488101282 [16] -0.488101282 -0.120396647 -0.120396647 0.131121515 0.131121515 [21] -0.214987009 -0.214987009 -0.054872789 -0.054872789 0.184657295 [26] 0.184657295 -0.510007747 -0.510007747 -0.790746805 -0.790746805 [31] 0.324674289 0.324674289 -0.239374060 -0.239374060 -0.264428564 [36] -0.264428564 -0.472698773 -0.472698773 0.006304049 0.006304049 [41] -0.873434085 -0.873434085 -0.530880840 -0.530880840 0.351411783 [46] 0.351411783 -0.037212138 -0.037212138 0.442049266 0.442049266 [51] -0.419206550 -0.419206550 -0.108012854 -0.108012854 0.346332076 [56] 0.346332076 0.659300205 0.659300205 0.197278585 0.197278585 [61] 0.304868889 0.304868889 0.139712997 0.139712997 0.093574024 [66] 0.093574024 -0.209690355 -0.209690355 0.302070834 0.302070834 [71] -0.278962288 -0.278962288 0.068599919 0.068599919 0.078493616 [76] 0.078493616 $se.fit [1] 0.6150025 0.6150025 0.6160184 0.6160184 0.5715622 0.5715622 0.4393615 [8] 0.4393615 0.5761369 0.5761369 0.4834244 0.4834244 0.6421184 0.6421184 [15] 0.4574824 0.4574824 0.4813578 0.4813578 0.5119792 0.5119792 0.4764145 [22] 0.4764145 0.5532477 0.5532477 0.5195437 0.5195437 0.5534327 0.5534327 [29] 0.4775572 0.4775572 0.6364522 0.6364522 0.4708988 0.4708988 0.4670896 [36] 0.4670896 0.5600672 0.5600672 0.5641880 0.5641880 0.4650576 0.4650576 [43] 0.4904715 0.4904715 0.5448430 0.5448430 0.5570120 0.5570120 0.5608187 [50] 0.5608187 0.4996021 0.4996021 0.4831697 0.4831697 0.5452255 0.5452255 [57] 0.6057428 0.6057428 0.5209402 0.5209402 0.5376594 0.5376594 0.5911350 [64] 0.5911350 0.5065368 0.5065368 0.5290283 0.5290283 0.5368433 0.5368433 [71] 0.5996077 0.5996077 0.5762814 0.5762814 0.5782753 0.5782753 > kfit1 Call: coxph(formula = Surv(time, status) ~ frailty(id, dist = "gauss"), data = kidney) coef se(coef) se2 Chisq DF p frailty(id, dist = "gauss 23 13.8 0.057 Iterations: 7 outer, 39 Newton-Raphson Variance of random effect= 0.458 Degrees of freedom for terms= 13.8 Likelihood ratio test=33.4 on 13.8 df, p=0.00234 n= 76 > > > > proc.time() user system elapsed 0.280 0.008 0.281 survival/tests/frailty.R0000644000175100001440000000152611732700061015103 0ustar hornikuserslibrary(survival) # # The constuction of a survival curve with sparse frailties # # In this case the coefficient vector is kept in two parts, the # fixed coefs and the (often very large) random effects coefficients # The survfit function treats the second set of coefficients as fixed # values, to avoid an unmanagable variance matrix, and behaves like # the second fit below. fit1 <- coxph(Surv(time, status) ~ age + frailty(inst), lung) sfit1 <- survfit(fit1) # A parallel model with the frailties treated as fixed offsets offvar <- fit1$frail[as.numeric(factor(lung$inst))] fit2 <- coxph(Surv(time, status) ~ age + offset(offvar),lung) fit2$var <- fit1$var #force variances to match all.equal(fit1$coef, fit2$coef) sfit2 <- survfit(fit2, newdata=list(age=fit1$means, offvar=0)) all.equal(sfit1$surv, sfit2$surv) all.equal(sfit1$var, sfit2$var) survival/tests/counting.Rout.save0000644000175100001440000000531611732700061016745 0ustar hornikusers R version 2.10.0 (2009-10-26) Copyright (C) 2009 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # Create a "counting process" version of the simplest test data set > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0)) > > test1b<- list(start= c(0, 3, 0, 0, 5, 0, 6,14, 0, 0, 10,20,30, 0), + stop = c(3,10, 10, 5,20, 6,14,20, 30, 10,20,30,40, 10), + status=c(0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0), + x= c(1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, NA), + id = c(3, 3, 4, 5, 5, 6, 6, 6, 7, 1, 1, 1, 1, 2)) > > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > # > # Check out the various residuals under an Efron approximation > # > fit0 <- coxph(Surv(time, status)~ x, test1, iter=0) > fit <- coxph(Surv(time, status) ~x, test1) > fit0b <- coxph(Surv(start, stop, status) ~ x, test1b, iter=0) > fitb <- coxph(Surv(start, stop, status) ~x, test1b) > fitc <- coxph(Surv(time, status) ~ offset(fit$coef*x), test1) > fitd <- coxph(Surv(start, stop, status) ~ offset(fit$coef*x), test1b) > > aeq(fit0b$coef, fit0$coef) [1] TRUE > > aeq(resid(fit0), resid(fit0b, collapse=test1b$id)) [1] TRUE > aeq(resid(fit), resid(fitb, collapse=test1b$id)) [1] TRUE > aeq(resid(fitc), resid(fitd, collapse=test1b$id)) [1] TRUE > aeq(resid(fitc), resid(fit)) [1] TRUE > > aeq(resid(fit0, type='score'), resid(fit0b, type='score', collapse=test1b$id)) [1] TRUE > aeq(resid(fit, type='score'), resid(fitb, type='score', collapse=test1b$id)) [1] TRUE > > aeq(resid(fit0, type='scho'), resid(fit0b, type='scho', collapse=test1b$id)) [1] TRUE > aeq(resid(fit, type='scho'), resid(fitb, type='scho', collapse=test1b$id)) [1] TRUE > > # The two survivals will have different censoring times > # nrisk, nevent, surv, and std should be the same > temp1 <- survfit(fit, list(x=1), censor=FALSE) > temp2 <- survfit(fitb, list(x=1), censor=FALSE) > all.equal(unclass(temp1)[c(3,4,6,8)], unclass(temp2)[c(3,4,6,8)]) [1] TRUE > > > survival/tests/tt.R0000644000175100001440000000316612325255713014072 0ustar hornikuserslibrary(survival) # A contrived example for the tt function # mkdata <- function(n, beta) { age <- runif(n, 20, 60) x <- rbinom(n, 1, .5) futime <- rep(40, n) # everyone has 40 years of follow-up status <- rep(0, n) dtime <- runif(n/2, 1, 40) # 1/2 of them die dtime <- sort(dtime) # The risk is set to beta[1]*x + beta[2]* f(current_age) # where f= 0 up to age 40, rises linear to age 70, flat after that for (i in 1:length(dtime)) { atrisk <- (futime >= dtime[i]) c.age <- age + dtime age2 <- pmin(30, pmax(0, c.age-40)) xbeta <- beta[1]*x + beta[2]*age2 # Select a death according to risk risk <- ifelse(atrisk, exp(xbeta), 0) dead <- sample(1:n, 1, prob=risk/sum(risk)) futime[dead] <- dtime[i] status[dead] <- 1 } data.frame(futime=futime, status=status, age=age, x=x, risk=risk) } tdata <- mkdata(500, c(log(1.5), 2/30)) fit1 <- coxph(Surv(futime, status) ~ x + pspline(age), tdata) fit2 <- coxph(Surv(futime, status) ~ x + tt(age), tdata, tt= function(x, t, ...) pspline(x+t)) dfit <- coxph(Surv(futime, status) ~ x + tt(age), tdata, tt= function(x, t, ...) x+t, iter=0, x=T) # # Check that cluster, weight, and offset were correctly expanded # tdata <- data.frame(tdata, grp=sample(1:100, 500, replace=TRUE), casewt = sample(1:5, 500, replace=TRUE), zz = rnorm(500)) dfit2 <- coxph(Surv(futime, status) ~ x + tt(age) + offset(zz) + cluster(grp), weight=casewt, data=tdata, tt= function(x, t, ...) x+t) survival/tests/fr_lung.Rout.save0000644000175100001440000000262011732700061016546 0ustar hornikusers R version 2.11.1 (2010-05-31) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > > # > # A test with the lung data > # This caused problems in one release > > # > # First, get rid of some missings > # > lung2 <- na.omit(lung[c('time', 'status', 'wt.loss')]) > > # > # Test the logliklihoods > # > fit <- coxph(Surv(time, status) ~ pspline(wt.loss,3), lung2, x=T) > fit0<- coxph(Surv(time, status) ~ 1, lung2) > fit1<- coxph(Surv(time, status) ~ fit$x, lung2, iter=0, init=fit$coef) > > all.equal(fit$loglik[1], fit0$loglik) [1] TRUE > all.equal(fit$loglik[2], fit1$loglik[2]) [1] TRUE > > # > # Check variances > # > imat <- solve(fit1$var) > var2 <- fit$var %*% imat %*% fit$var > all.equal(fit$var2, var2) [1] TRUE > survival/tests/testreg.R0000644000175100001440000000430211732700061015101 0ustar hornikusersoptions(na.action=na.exclude) #preserve length of missings library(survival) # # Run a test that can be verified using other packages (we used SAS) # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) fit1w <- survreg(Surv(time, status) ~x, test1, dist='weibull') fit1w summary(fit1w) fit1e <- survreg(Surv(time, status) ~x, test1, dist='exponential') fit1e summary(fit1e) fit1l <- survreg(Surv(time, status) ~x, test1, dist='loglogistic') fit1l summary(fit1l) fit1g <- survreg(Surv(time, status) ~x, test1, dist='lognormal') summary(fit1g) # # Do a test with the ovarian data # fitfw <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, dist='weibull') fitfw fitfl <- survreg(Surv(futime, fustat) ~ age + ecog.ps, ovarian, dist='loglogistic') fitfl #test out interval censoring, using some dummy time values idat <- read.table('data.interval', skip=3, header=T, sep=',') flsurv<- Surv(idat$ltime, idat$rtime, type='interval2') fitfw2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='weibull') summary(fitfw2) fitfl2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='loglogistic') summary(fitfl2) fitfg2 <- survreg(flsurv ~ age + ecog.ps, idat, dist='lognormal') summary(fitfg2) logt <- c(survreg.distributions$t, survreg.distributions$weibull[c('trans', 'itrans', 'dtrans')]) logt$name <- 'log(t)' fitft2 <- survreg(Surv(ltime, rtime, type='interval2') ~ age + ecog.ps, idat, dist=logt, parm=100) summary(fitft2) #should be quite close to fitfg2 # # Check out the survreg density and probability functions # # Gaussian x <- -10:10 p <- seq(.1, .95, length=25) all.equal(dsurvreg(x, 1, 5, 'gaussian'), dnorm(x, 1, 5)) all.equal(psurvreg(x, 1, 5, 'gaussian'), pnorm(x, 1, 5)) all.equal(qsurvreg(p, 1, 5, 'gaussian'), qnorm(p, 1, 5)) # Lognormal x <- 1:10 all.equal(dsurvreg(x, 1, 5, 'lognormal'), dlnorm(x, 1, 5)) all.equal(psurvreg(x, 1, 5, 'lognormal'), plnorm(x, 1, 5)) all.equal(qsurvreg(p, 1, 5, 'lognormal'), qlnorm(p, 1, 5)) # Weibull lambda <- exp(-2) rho <- 1/3 temp <- (lambda*x)^rho all.equal(psurvreg(x, 2, 3), 1- exp(-temp)) all.equal(dsurvreg(x, 2, 3), lambda*rho*(lambda*x)^(rho-1)*exp(-temp)) survival/tests/book5.R0000644000175100001440000001070611732700061014450 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # Tests of the weighted Cox model # This is section 1.3 of my appendix -- no yet found in any of the # printings though, it awaits the next edition # # Similar data set to test1, but add weights, # a double-death/censor tied time # a censored last subject # The latter two are cases covered only feebly elsewhere. # # The data set testw2 has the same data, but done via replication # aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), status= c(1,0,1,1,1,0,0,1,0), x= c(2,0,1,1,0,1,0,1,0), wt = c(1,2,3,4,3,2,1,2,1)) xx <- testw1$wt testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), status= rep(c(1,0,1,1,1,0,0,1,0), xx), x= rep(c(2,0,1,1,0,1,0,1,0), xx), id= rep(1:9, xx)) indx <- match(1:9, testw2$id) # Breslow estimate byhand <- function(beta, newx=0) { r <- exp(beta) loglik <- 11*beta - (log(r^2 + 11*r +7) + 10*log(11*r +5) +2*log(2*r+1)) hazard <- c(1/(r^2 + 11*r +7), 10/(11*r +5), 2/(2*r+1)) xbar <- c((2*r^2 + 11*r)*hazard[1], 11*r/(11*r +5), r*hazard[3]) imat <- (4*r^2 + 11*r)*hazard[1] - xbar[1]^2 + 10*(xbar[2] - xbar[2]^2) + 2*(xbar[3] - xbar[3]^2) temp <- cumsum(hazard) risk <- c(r^2, 1,r,r,1,r,1,r,1) expected <- risk* temp[c(1,1,2,2,2,2,2,3,3)] # The matrix of weights, one row per obs, one col per death # deaths at 1,2,2,2, and 4 riskmat <- matrix(c(1,1,1,1,1,1,1,1,1, 0,0,1,1,1,1,1,1,1, 0,0,1,1,1,1,1,1,1, 0,0,1,1,1,1,1,1,1, 0,0,0,0,0,0,0,1,1), ncol=5) wtmat <- diag(c(r^2, 2, 3*r, 4*r, 3, 2*r, 1, 2*r, 1)) %*% riskmat x <- c(2,0,1,1,0,1,0,1,0) status <- c(1,0,1,1,1,0,0,1,0) wt <- c(1,2,3,4,3,2,1,2,1) # Table of sums for score and Schoenfeld resids hazmat <- riskmat %*% diag(c(1,3,4,3,2)/colSums(wtmat)) dM <- -risk*hazmat #Expected part dM[1,1] <- dM[1,1] +1 # deaths at time 1 for (i in 2:4) dM[i+1, i] <- dM[i+1,i] +1 dM[8,5] <- dM[8,5] +1 mart <- rowSums(dM) resid <-dM * outer(x, xbar[c(1,2,2,2,3)] ,'-') # Increments to the variance of the hazard var.g <- cumsum(hazard^2/ c(1,10,2)) var.d <- cumsum((xbar-newx)*hazard) list(loglik=loglik, imat=imat, hazard=hazard, xbar=xbar, mart=c(1,0,1,1,1,0,0,1,0)-expected, expected=expected, score=rowSums(resid), schoen=c(2,1,1,0,1) - xbar[c(1,2,2,2,3)], varhaz=(var.g + var.d^2/imat)* exp(2*beta*newx)) } aeq(byhand(0)$expected, c(1/19, 1/19, rep(103/152, 5), rep(613/456,2))) #verify fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow', iter=0) fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow') fitb <- coxph(Surv(time, status) ~x, testw2, method='breslow') aeq(resid(fit0, type='mart'), (resid(fit0b, type='mart'))[indx]) aeq(resid(fit0, type='scor'), (resid(fit0b, type='scor'))[indx]) aeq(unique(resid(fit0, type='scho')), unique(resid(fit0b, type='scho'))) truth0 <- byhand(0,pi) aeq(fit0$loglik[1], truth0$loglik) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid) aeq(truth0$scho, resid(fit0, 'schoen')) aeq(truth0$score, resid(fit0, 'score')) sfit <- survfit(fit0, list(x=pi), censor=FALSE) aeq(sfit$std.err^2, truth0$var) aeq(-log(sfit$surv), cumsum(truth0$haz)) truth <- byhand(fit$coef, .3) aeq(truth$loglik, fit$loglik[2]) aeq(1/truth$imat, fit$var) aeq(truth$mart, fit$resid) aeq(truth$scho, resid(fit, 'schoen')) aeq(truth$score, resid(fit, 'score')) sfit <- survfit(fit, list(x=.3), censor=FALSE) aeq(sfit$std.err^2, truth$var) aeq(-log(sfit$surv), (cumsum(truth$haz)* exp(fit$coef*.3))) fit0 summary(fit) resid(fit0, type='score') resid(fit0, type='scho') resid(fit, type='score') resid(fit, type='scho') aeq(resid(fit, type='mart'), (resid(fitb, type='mart'))[indx]) aeq(resid(fit, type='scor'), (resid(fitb, type='scor'))[indx]) aeq(unique(resid(fit, type='scho')), unique(resid(fitb, type='scho'))) rr1 <- resid(fit, type='mart') rr2 <- resid(fit, type='mart', weighted=T) aeq(rr2/rr1, testw1$wt) rr1 <- resid(fit, type='score') rr2 <- resid(fit, type='score', weighted=T) aeq(rr2/rr1, testw1$wt) survival/tests/data.smoke0000644000175100001440000000304511732700061015255 0ustar hornikusers 186.0 439.2 234.4 365.8 159.6 216.9 167.4 159.5 255.6 702.7 544.7 431.0 454.8 349.7 214.0 250.4 448.9 1132.4 945.2 728.8 729.4 590.2 447.3 436.6 733.7 1981.1 1177.7 1589.2 1316.5 1266.9 875.6 703.0 1119.4 3003.0 2244.9 3380.3 2374.9 1820.2 1669.1 1159.2 2070.5 4697.5 4255.3 5083.0 4485.0 3888.7 3184.3 2194.9 3675.3 7340.6 5882.4 6597.2 7707.5 4945.1 5618.0 4128.9 186.0 610.0 497.5 251.7 417.5 122.6 198.3 193.4 255.6 915.6 482.8 500.7 488.9 402.9 393.9 354.3 448.9 1391.0 1757.1 953.5 1025.8 744.0 668.5 537.8 733.7 2393.4 1578.4 1847.2 1790.1 1220.7 1100.0 993.3 1119.4 3497.9 2301.8 3776.6 2081.0 2766.4 2268.1 1230.7 2070.5 5861.3 3174.6 2974.0 3712.9 3988.8 3268.6 2468.9 3675.3 6250.0 4000.0 4424.8 7329.8 6383.0 7666.1 5048.1 125.7 225.6 0 433.9 212.0 107.2 135.9 91.0 177.3 353.8 116.8 92.1 289.5 200.9 121.3 172.1 244.8 542.8 287.4 259.5 375.9 165.8 202.2 247.2 397.7 858.0 1016.3 365.0 650.9 470.8 570.6 319.7 692.1 1496.2 1108.0 1348.5 1263.2 864.8 586.6 618.0 1160.0 2084.8 645.2 1483.1 1250.0 1126.3 1070.5 1272.1 2070.8 3319.5 0 2580.6 2590.7 3960.4 1666.7 1861.5 125.7 277.9 266.7 102.7 178.6 224.7 142.1 138.8 177.3 517.9 138.7 466.8 270.1 190.2 116.8 83.0 244.8 823.5 473.6 602.0 361.0 454.5 412.2 182.1 397.7 1302.9 1114.8 862.1 699.6 541.7 373.1 356.4 692.1 1934.9 2319.6 1250.0 1688.0 828.7 797.9 581.5 1160.0 2827.0 4635.8 2517.2 1687.3 2848.7 1621.2 1363.4 2070.8 4273.1 2409.6 5769.2 3125.0 2978.7 2803.7 2195.4 survival/tests/concordance.Rout.save0000644000175100001440000001245011732700061017372 0ustar hornikusers R version 2.12.2 (2011-02-25) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > > # > # Simple tests of concordance. These numbers were derived in multiple > # codes. > # > aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) > > grank <- function(x, time, grp, wt) + unlist(tapply(x, grp, rank)) > grank2 <- function(x, time, grp, wt) { #for case weights + if (length(wt)==0) wt <- rep(1, length(x)) + z <- double(length(x)) + for (i in unique(grp)) { + indx <- which(grp==i) + temp <- tapply(wt[indx], x[indx], sum) + temp <- temp/2 + c(0, cumsum(temp)[-length(temp)]) + z[indx] <- temp[match(x[indx], names(temp))] + } + z + } > > > tdata <- aml[aml$x=='Maintained',] > tdata$y <- c(1,6,2,7,3,7,3,8,4,4,5) > tdata$wt <- c(1,2,3,2,1,2,3,4,3,2,1) > fit <- survConcordance(Surv(time, status) ~y, tdata) > aeq(fit$stats[1:4], c(14,24,2,0)) [1] TRUE > cfit <- coxph(Surv(time, status) ~ tt(y), tdata, tt=grank, method='breslow', + iter=0, x=T) > cdt <- coxph.detail(cfit) > aeq(4*sum(cdt$imat),fit$stats[5]^2) [1] TRUE > aeq(2*sum(cdt$score), diff(fit$stats[2:1])) [1] TRUE > > > # Lots of ties > tempx <- Surv(c(1,2,2,2,3,4,4,4,5,2), c(1,0,1,0,1,0,1,1,0,1)) > tempy <- c(5,5,4,4,3,3,7,6,5,4) > fit2 <- survConcordance(tempx ~ tempy) > aeq(fit2$stats[1:4], c(13,13,5,2)) [1] TRUE > cfit2 <- coxph(tempx ~ tt(tempy), tt=grank, method='breslow', iter=0) > aeq(4/cfit2$var, fit2$stats[5]^2) [1] TRUE > > # Bigger data > fit3 <- survConcordance(Surv(time, status) ~ age, lung) > aeq(fit3$stats[1:4], c(10717, 8706, 591, 28)) [1] TRUE > cfit3 <- coxph(Surv(time, status) ~ tt(age), lung, + iter=0, method='breslow', tt=grank, x=T) > cdt <- coxph.detail(cfit3) > aeq(4*sum(cdt$imat),fit3$stats[5]^2) [1] TRUE > aeq(2*sum(cdt$score), diff(fit3$stats[2:1])) [1] TRUE > > > # More ties > fit4 <- survConcordance(Surv(time, status) ~ ph.ecog, lung) > aeq(fit4$stats[1:4], c(8392, 4258, 7137, 28)) [1] TRUE > cfit4 <- coxph(Surv(time, status) ~ tt(ph.ecog), lung, + iter=0, method='breslow', tt=grank) > aeq(4/cfit4$var, fit4$stats[5]^2) [1] TRUE > > # Case weights > fit5 <- survConcordance(Surv(time, status) ~ y, tdata, weight=wt) > fit6 <- survConcordance(Surv(time, status) ~y, tdata[rep(1:11,tdata$wt),]) > aeq(fit5$stats[1:4], c(70, 91, 7, 0)) # checked by hand [1] TRUE > aeq(fit5$stats[1:3], fit6$stats[1:3]) #spurious "tied on time" value, ignore [1] TRUE > aeq(fit5$std, fit6$std) [1] TRUE > cfit5 <- coxph(Surv(time, status) ~ tt(y), tdata, weight=wt, + iter=0, method='breslow', tt=grank2) > cfit6 <- coxph(Surv(time, status) ~ tt(y), tdata[rep(1:11,tdata$wt),], + iter=0, method='breslow', tt=grank) > aeq(4/cfit6$var, fit6$stats[5]^2) [1] TRUE > aeq(cfit5$var, cfit6$var) [1] TRUE > > # Start, stop simplest cases > fit7 <- survConcordance(Surv(rep(0,11), time, status) ~ y, tdata) > aeq(fit7$stats, fit$stats) [1] TRUE > aeq(fit7$std.err, fit$std.err) [1] TRUE > fit7 <- survConcordance(Surv(rep(0,11), time, status) ~ y, tdata, weight=wt) > aeq(fit5$stats, fit7$stats) [1] TRUE > > # Multiple intervals for some, but same risk sets as tdata > tdata2 <- data.frame(time1=c(0,3, 5, 6,7, 0, 4,17, 7, 0,16, 2, 0, + 0,9, 5), + time2=c(3,9, 13, 7,13, 18, 17,23, 28, 16,31, 34, 45, + 9,48, 60), + status=c(0,1, 1, 0,0, 1, 0,1, 0, 0,1, 1, 0, 0,1, 0), + y = c(1,1, 6, 2,2, 7, 3,3, 7, 3,3, 8, 4, 4,4, 5), + wt= c(1,1, 2, 3,3, 2, 1,1, 2, 3,3, 4, 3, 2,2, 1)) > fit8 <- survConcordance(Surv(time1, time2, status) ~y, tdata2, weight=wt) > aeq(fit5$stats, fit8$stats) [1] TRUE > aeq(fit5$std.err, fit8$std.err) [1] TRUE > cfit8 <- coxph(Surv(time1, time2, status) ~ tt(y), tdata2, weight=wt, + iter=0, method='breslow', tt=grank2) > aeq(4/cfit8$var, fit8$stats[5]^2) [1] TRUE > aeq(fit8$stats[5]/(2*sum(fit8$stats[1:3])), fit8$std.err) [1] TRUE > > # Stratified > tdata3 <- data.frame(time1=c(tdata2$time1, rep(0, nrow(lung))), + time2=c(tdata2$time2, lung$time), + status = c(tdata2$status, lung$status -1), + x = c(tdata2$y, lung$ph.ecog), + wt= c(tdata2$wt, rep(1, nrow(lung))), + grp=rep(1:2, c(nrow(tdata2), nrow(lung)))) > fit9 <- survConcordance(Surv(time1, time2, status) ~x + strata(grp), + data=tdata3, weight=wt) > aeq(fit9$stats[1,], fit5$stats) [1] TRUE > aeq(fit9$stats[2,], fit4$stats) [1] TRUE > survival/tests/pyear.Rout.save0000644000175100001440000004123411732700061016236 0ustar hornikusers R version 2.12.1 (2010-12-16) Copyright (C) 2010 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) Loading required package: splines > {if (is.R()) mdy.date <- function(m, d, y) { + y <- ifelse(y<100, y+1900, y) + as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") + } + else mdy.date <- function(m,d,y) { + y <- ifelse(y<100, y+1900, y) + timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") + } + } > > # > # Simple case: a single male subject, born 6/6/36 and entered on study 6/6/55. > # > > temp1 <- mdy.date(6,6,36) > temp2 <- mdy.date(6,6,55)# Now compare the results from person-years > # > temp.age <- tcut(temp2-temp1, floor(c(-1, (18:31 * 365.24))), + labels=c('0-18', paste(18:30, 19:31, sep='-'))) > temp.yr <- tcut(temp2, mdy.date(1,1,1954:1965), labels=1954:1964) > temp.time <- 3700 #total days of fu > py1 <- pyears(temp.time ~ temp.age + temp.yr, scale=1) #output in days > > # The subject should appear in 20 cells > # 6/6/55 - 12/31/55, 209 days, age 19-20, 1955 > # 1/1/56 - 6/ 4/56, 156 days, age 19-20, 1956 > # 6/5/56 - 12/31/56, 210 days, age 20-21, 1956 (a leap year, and his > # birthday computes one day earlier) > # 1/1/57 - 6/ 5/57, 156 days, age 20-21, 1957 > # 6/6/57 - 12/31/57, 209 days, age 21-22, 1957 > # and etc > # with 203 days "off table", ie, beyond the last cell of the table > # > # It is a nuisance, but tcut follows 'cut' in that we give the ENDS of > # the intervals, whereas the survival tables use the starts of intervals. > # Thus this breakdown does not match that in doexpect.s > # > xx <- matrix(0, nrow=14, ncol=11) > xx[cbind(3:11, 3:11)] <- 156 > xx[cbind(3:12, 2:11)] <- c(209, 210, rep(c(209, 209, 209, 210),2)) > dimnames(xx) <- list(c('0-18', paste(18:30, 19:31, sep='-')), 1954:1964) > all.equal(xx, py1$pyears) [1] TRUE > all.equal(203, py1$offtable) [1] TRUE > all.equal(1*(xx>0), py1$n) [1] TRUE > > # > # Now with expecteds > # > py2 <- pyears(temp.time ~ temp.age + temp.yr + + ratetable(age=temp2-temp1, year=temp2, sex=1), + scale=1, ratetable=survexp.us ) #output in days > all.equal(xx, py2$pyears) [1] TRUE > all.equal(203, py2$offtable) [1] TRUE > all.equal(1*(xx>0), py2$n) [1] TRUE > > py2b <- pyears(temp.time ~ temp.age + temp.yr, + rmap = list(age=temp2-temp1, year=temp2, sex=1), + scale=1, ratetable=survexp.us ) #output in days > all.equal(xx, py2b$pyears) [1] TRUE > all.equal(203, py2b$offtable) [1] TRUE > all.equal(1*(xx>0), py2b$n) [1] TRUE > all.equal(py2$expected, py2b$expected) [1] TRUE > > > > py3 <- pyears(temp.time ~ temp.age + temp.yr, + rmap=list(age=temp2-temp1, year=temp2, sex=1), + scale=1, ratetable=survexp.us , expect='pyears') > all.equal(py2$n, py3$n) [1] TRUE > all.equal(py2$pyear, py3$pyear) [1] TRUE > all.equal(py3$n, 1*(py3$expect>0)) [1] TRUE > > # Now, compute the py3 result "by hand". Since there is only one person > # it can be derived from py2. > # > xx1 <- py2$expect[py2$n>0] # the hazard over each interval > cumhaz <- cumsum(c(0, xx1[-length(xx1)])) # the cumulative hazard > xx2 <- py3$expect[py3$n>0] # the expected number of person days > xx3 <- py3$pyears[py3$n>0] # the potential number of person days > > # This is the integral of the curve "exp(-haz *t)" over the interval > integral <- xx3 * exp(-cumhaz)* (1- exp(-xx1))/ xx1 > # They might not be exactly equal, since the C code tracks changes in the > # rate tables that occur -within- an interval. So try for 6 digits > all.equal(round(integral,3), round(xx2,3)) [1] TRUE > > # Cut off the bottom of the table, instead of the side > temp.age <- tcut(temp2-temp1, floor(c(-1, (18:27 * 365.24))), + labels=c('0-18', paste(18:26, 19:27, sep='-'))) > > py4 <- eval(py3$call) > all.equal(py4$pyear, py3$pyear[1:10,]) [1] TRUE > all.equal(py4$expect, py3$expect[1:10,]) [1] TRUE > > > rm(temp.age, integral, xx1, xx2, xx3, cumhaz, py1, py2, py3, py4) > rm(temp1, temp2, temp.yr, temp.time, xx) > > > > > # > # Simple case: a single male subject, born 6/6/36 and entered on study 6/6/55. > # > > temp1 <- mdy.date(6,6,36) > temp2 <- mdy.date(6,6,55)# Now compare the results from person-years > # > temp.age <- tcut(temp2-temp1, floor(c(-1, (18:31 * 365.24))), + labels=c('0-18', paste(18:30, 19:31, sep='-'))) > temp.yr <- tcut(temp2, mdy.date(1,1,1954:1965), labels=1954:1964) > temp.time <- 3700 #total days of fu > py1 <- pyears(temp.time ~ temp.age + temp.yr, scale=1) #output in days > > # The subject should appear in 20 cells > # 6/6/55 - 12/31/55, 209 days, age 19-20, 1955 > # 1/1/56 - 6/ 4/56, 156 days, age 19-20, 1956 > # 6/5/56 - 12/31/56, 210 days, age 20-21, 1956 (a leap year, and his > # birthday computes one day earlier) > # 1/1/57 - 6/ 5/57, 156 days, age 20-21, 1957 > # 6/6/57 - 12/31/57, 209 days, age 21-22, 1957 > # and etc > # with 203 days "off table", ie, beyond the last cell of the table > # > # It is a nuisance, but tcut follows 'cut' in that we give the ENDS of > # the intervals, whereas the survival tables use the starts of intervals. > # > xx <- matrix(0, nrow=14, ncol=11) > xx[cbind(3:11, 3:11)] <- 156 > xx[cbind(3:12, 2:11)] <- c(209, 210, rep(c(209, 209, 209, 210),2)) > dimnames(xx) <- list(c('0-18', paste(18:30, 19:31, sep='-')), 1954:1964) > all.equal(xx, py1$pyears) [1] TRUE > all.equal(203, py1$offtable) [1] TRUE > all.equal(1*(xx>0), py1$n) [1] TRUE > > # > # Now with expecteds > # > py2 <- pyears(temp.time ~ temp.age + temp.yr + + ratetable(age=temp2-temp1, year=temp2, sex=1), + scale=1, ratetable=survexp.us ) #output in days > all.equal(xx, py2$pyears) [1] TRUE > all.equal(203, py2$offtable) [1] TRUE > all.equal(1*(xx>0), py2$n) [1] TRUE > > > py3 <- pyears(temp.time ~ temp.age + temp.yr + + ratetable(age=temp2-temp1, year=temp2, sex=1), + scale=1, ratetable=survexp.us , expect='pyears') > all.equal(py2$n, py3$n) [1] TRUE > all.equal(py2$pyear, py3$pyear) [1] TRUE > all.equal(py3$n, 1*(py3$expect>0)) [1] TRUE > > # Now, compute the py3 result "by hand". Since there is only one person > # it can be derived from py2. > # > xx1 <- py2$expect[py2$n>0] # the hazard over each interval > cumhaz <- cumsum(c(0, xx1[-length(xx1)])) # the cumulative hazard > xx2 <- py3$expect[py3$n>0] # the expected number of person days > xx3 <- py3$pyears[py3$n>0] # the potential number of person days > > # This is the integral of the curve "exp(-haz *t)" over the interval > integral <- xx3 * exp(-cumhaz)* (1- exp(-xx1))/ xx1 > # They might not be exactly equal, since the C code tracks changes in the > # rate tables that occur -within- an interval. So try for 6 digits > all.equal(round(integral,3), round(xx2,3)) [1] TRUE > > # Cut off the bottom of the table, instead of the side > temp.age <- tcut(temp2-temp1, floor(c(-1, (18:27 * 365.24))), + labels=c('0-18', paste(18:26, 19:27, sep='-'))) > > py4 <- eval(py3$call) > all.equal(py4$pyear, py3$pyear[1:10,]) [1] TRUE > all.equal(py4$expect, py3$expect[1:10,]) [1] TRUE > > > rm(temp.age, integral, xx1, xx2, xx3, cumhaz, py1, py2, py3, py4) > rm(temp1, temp2, temp.yr, temp.time, xx) > > > > > # > # Create a "user defined" rate table, using the smoking data > # > temp <- scan("data.smoke")/100000 Read 224 items > temp <- matrix(temp, ncol=8, byrow=T) > smoke.rate <- c(rep(temp[,1],6), rep(temp[,2],6), temp[,3:8]) > attributes(smoke.rate) <- list( + dim=c(7,2,2,6,3), + dimnames=list(c("45-49","50-54","55-59","60-64","65-69","70-74","75-79"), + c("1-20", "21+"), + c("Male","Female"), + c("<1", "1-2", "3-5", "6-10", "11-15", ">=16"), + c("Never", "Current", "Former")), + dimid=c("age", "amount", "sex", "duration", "status"), + factor=c(0,1,1,0,1), + cutpoints=list(c(45,50,55,60,65,70,75),NULL, NULL, + c(0,1,3,6,11,16),NULL), + class='ratetable' + ) > rm(temp) > > is.ratetable(smoke.rate) [1] TRUE > summary(smoke.rate) Rate table with 5 dimensions: age ranges from 45 to 75; with 7 categories amount has levels of: 1-20 21+ sex has levels of: Male Female duration ranges from 0 to 16; with 6 categories status has levels of: Never Current Former > print(smoke.rate) Rate table with dimension(s): age amount sex duration status , , Male, <1, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, <1, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, 1-2, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, 1-2, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, 3-5, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, 3-5, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, 6-10, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, 6-10, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, 11-15, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, 11-15, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, >=16, Never 1-20 21+ 45-49 0.001860 0.001860 50-54 0.002556 0.002556 55-59 0.004489 0.004489 60-64 0.007337 0.007337 65-69 0.011194 0.011194 70-74 0.020705 0.020705 75-79 0.036753 0.036753 , , Female, >=16, Never 1-20 21+ 45-49 0.001257 0.001257 50-54 0.001773 0.001773 55-59 0.002448 0.002448 60-64 0.003977 0.003977 65-69 0.006921 0.006921 70-74 0.011600 0.011600 75-79 0.020708 0.020708 , , Male, <1, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, <1, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, 1-2, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, 1-2, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, 3-5, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, 3-5, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, 6-10, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, 6-10, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, 11-15, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, 11-15, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, >=16, Current 1-20 21+ 45-49 0.004392 0.006100 50-54 0.007027 0.009156 55-59 0.011324 0.013910 60-64 0.019811 0.023934 65-69 0.030030 0.034979 70-74 0.046975 0.058613 75-79 0.073406 0.062500 , , Female, >=16, Current 1-20 21+ 45-49 0.002256 0.002779 50-54 0.003538 0.005179 55-59 0.005428 0.008235 60-64 0.008580 0.013029 65-69 0.014962 0.019349 70-74 0.020848 0.028270 75-79 0.033195 0.042731 , , Male, <1, Former 1-20 21+ 45-49 0.002344 0.004975 50-54 0.005447 0.004828 55-59 0.009452 0.017571 60-64 0.011777 0.015784 65-69 0.022449 0.023018 70-74 0.042553 0.031746 75-79 0.058824 0.040000 , , Female, <1, Former 1-20 21+ 45-49 0.000000 0.002667 50-54 0.001168 0.001387 55-59 0.002874 0.004736 60-64 0.010163 0.011148 65-69 0.011080 0.023196 70-74 0.006452 0.046358 75-79 0.000000 0.024096 , , Male, 1-2, Former 1-20 21+ 45-49 0.003658 0.002517 50-54 0.004310 0.005007 55-59 0.007288 0.009535 60-64 0.015892 0.018472 65-69 0.033803 0.037766 70-74 0.050830 0.029740 75-79 0.065972 0.044248 , , Female, 1-2, Former 1-20 21+ 45-49 0.004339 0.001027 50-54 0.000921 0.004668 55-59 0.002595 0.006020 60-64 0.003650 0.008621 65-69 0.013485 0.012500 70-74 0.014831 0.025172 75-79 0.025806 0.057692 , , Male, 3-5, Former 1-20 21+ 45-49 0.001596 0.004175 50-54 0.004548 0.004889 55-59 0.007294 0.010258 60-64 0.013165 0.017901 65-69 0.023749 0.020810 70-74 0.044850 0.037129 75-79 0.077075 0.073298 , , Female, 3-5, Former 1-20 21+ 45-49 0.002120 0.001786 50-54 0.002895 0.002701 55-59 0.003759 0.003610 60-64 0.006509 0.006996 65-69 0.012632 0.016880 70-74 0.012500 0.016873 75-79 0.025907 0.031250 , , Male, 6-10, Former 1-20 21+ 45-49 0.002169 0.001226 50-54 0.003497 0.004029 55-59 0.005902 0.007440 60-64 0.012669 0.012207 65-69 0.018202 0.027664 70-74 0.038887 0.039888 75-79 0.049451 0.063830 , , Female, 6-10, Former 1-20 21+ 45-49 0.001072 0.002247 50-54 0.002009 0.001902 55-59 0.001658 0.004545 60-64 0.004708 0.005417 65-69 0.008648 0.008287 70-74 0.011263 0.028487 75-79 0.039604 0.029787 , , Male, 11-15, Former 1-20 21+ 45-49 0.001674 0.001983 50-54 0.002140 0.003939 55-59 0.004473 0.006685 60-64 0.008756 0.011000 65-69 0.016691 0.022681 70-74 0.031843 0.032686 75-79 0.056180 0.076661 , , Female, 11-15, Former 1-20 21+ 45-49 0.001359 0.001421 50-54 0.001213 0.001168 55-59 0.002022 0.004122 60-64 0.005706 0.003731 65-69 0.005866 0.007979 70-74 0.010705 0.016212 75-79 0.016667 0.028037 , , Male, >=16, Former 1-20 21+ 45-49 0.001595 0.001934 50-54 0.002504 0.003543 55-59 0.004366 0.005378 60-64 0.007030 0.009933 65-69 0.011592 0.012307 70-74 0.021949 0.024689 75-79 0.041289 0.050481 , , Female, >=16, Former 1-20 21+ 45-49 0.000910 0.001388 50-54 0.001721 0.000830 55-59 0.002472 0.001821 60-64 0.003197 0.003564 65-69 0.006180 0.005815 70-74 0.012721 0.013634 75-79 0.018615 0.021954 > > summary(smoke.rate[1:3,,1,,]) #test subscripting Rate table with 4 dimensions: age ranges from 45 to 55; with 3 categories amount has levels of: 1-20 21+ duration ranges from 0 to 16; with 6 categories status has levels of: Never Current Former > survival/tests/stratatest.R0000644000175100001440000000314111732700061015622 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Trivial test of stratified residuals # Make a second strata = replicate of the first, and I should get the # exact same answers test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) temp <- as.matrix(test1) n <- nrow(temp) ndead<- sum(test1$status[!is.na(test1$status)]) temp <- data.frame(rbind(temp, temp)) #later releases of S have rbind.data.frame tstrat <- rep(1:2, c(n,n)) fit1 <- coxph(Surv(time, status) ~x, test1) fit2 <- coxph(Surv(time, status) ~x + strata(tstrat), temp) all.equal(resid(fit1) , (resid(fit2))[1:n]) all.equal(resid(fit1, type='score') , (resid(fit2, type='score'))[1:n]) all.equal(resid(fit1, type='schoe') , (resid(fit2, type='schoe'))[1:ndead]) #AG model temp <- as.matrix(test2) n <- nrow(temp) ndead<- sum(test2$event[!is.na(test2$event)]) temp <- data.frame(rbind(temp, temp)) tstrat <- rep(1:2, c(n,n)) fit1 <- coxph(Surv(start, stop, event) ~x, test2) fit2 <- coxph(Surv(start, stop, event) ~x + strata(tstrat), temp) all.equal(resid(fit1) , (resid(fit2))[1:n]) all.equal(resid(fit1, type='score') , (resid(fit2, type='score'))[1:n]) all.equal(resid(fit1, type='schoe') , (resid(fit2, type='schoe'))[1:ndead]) survival/tests/survfit2.R0000644000175100001440000000070212470201064015207 0ustar hornikuserslibrary(survival) # # Check out the Dory&Korn confidence interval option # tdata <- data.frame(time= 1:10, status=c(1,0,1,0,1,0,0,0,1,0)) fit1 <- survfit(Surv(time, status) ~1, tdata, conf.lower='modified') fit2 <- survfit(Surv(time, status) ~1, tdata) stdlow <- fit2$std * sqrt(c(1, 10/9, 1, 8/7, 1, 6/5, 6/4, 6/3, 1, 2/1)) lower <- exp(log(fit2$surv) - qnorm(.975)*stdlow) all.equal(fit1$lower, lower, check.attributes=FALSE) survival/tests/doweight.R0000644000175100001440000001760711741355257015267 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # Tests of the weighted Cox model # # Similar data set to test1, but add weights, # a double-death/censor tied time # a censored last subject # The latter two are cases covered only feebly elsewhere. # # The data set testw2 has the same data, but done via replication # aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), status= c(1,0,1,1,1,0,0,1,0), x= c(2,0,1,1,0,1,0,1,0), wt = c(1,2,3,4,3,2,1,2,1)) xx <- c(1,2,3,4,3,2,1,2,1) testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), status= rep(c(1,0,1,1,1,0,0,1,0), xx), x= rep(c(2,0,1,1,0,1,0,1,0), xx), id= rep(1:9, xx)) indx <- match(1:9, testw2$id) testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), status= rep(c(1,0,1,1,1,0,0,1,0), xx), x= rep(c(2,0,1,1,0,1,0,1,0), xx), id= rep(1:9, xx)) indx <- match(1:9, testw2$id) fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow', iter=0) fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow') fitb <- coxph(Surv(time, status) ~x, testw2, method='breslow') texp <- function(beta) { # expected, Breslow estimate r <- exp(beta) temp <- cumsum(c(1/(r^2 + 11*r +7), 10/(11*r +5), 2/(2*r+1))) c(r^2, 1,r,r,1,r,1,r,1)* temp[c(1,1,2,2,2,2,2,3,3)] } aeq(texp(0), c(1/19, 1/19, rep(103/152, 5), rep(613/456,2))) #verify texp() xbar <- function(beta) { # xbar, Breslow estimate r <- exp(beta) temp <- r* rep(c(2*r + 11, 11/10, 1), c(2, 5, 2)) temp * texp(beta) } fit0 summary(fit) aeq(resid(fit0), testw1$status - texp(0)) resid(fit0, type='score') resid(fit0, type='scho') aeq(resid(fit0, type='mart'), (resid(fit0b, type='mart'))[indx]) aeq(resid(fit0, type='scor'), (resid(fit0b, type='scor'))[indx]) aeq(unique(resid(fit0, type='scho')), unique(resid(fit0b, type='scho'))) aeq(resid(fit, type='mart'), testw1$status - texp(fit$coef)) resid(fit, type='score') resid(fit, type='scho') aeq(resid(fit, type='mart'), (resid(fitb, type='mart'))[indx]) aeq(resid(fit, type='scor'), (resid(fitb, type='scor'))[indx]) aeq(unique(resid(fit, type='scho')), unique(resid(fitb, type='scho'))) rr1 <- resid(fit, type='mart') rr2 <- resid(fit, type='mart', weighted=T) aeq(rr2/rr1, testw1$wt) rr1 <- resid(fit, type='score') rr2 <- resid(fit, type='score', weighted=T) aeq(rr2/rr1, testw1$wt) fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='efron') fit resid(fit, type='mart') resid(fit, type='score') resid(fit, type='scho') # Tests of the weighted Cox model, AG form of the data # Same solution as doweight1.s # testw3 <- data.frame(id = c( 1, 1, 2, 3, 3, 3, 4, 5, 5, 6, 7, 8, 8, 9), begin= c( 0, 5, 0, 0,10,15, 0, 0,14, 0, 0, 0,23, 0), time= c( 5,10,10,10,15,20,20,14,20,20,30,23,40,50), status= c( 0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0), x= c( 2, 2, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0), wt = c( 1, 1, 2, 3, 3, 3, 4, 3, 3, 2, 1, 2, 2, 1)) fit0 <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, method='breslow', iter=0) fit <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, method='breslow') fit0 summary(fit) resid(fit0, type='mart', collapse=testw3$id) resid(fit0, type='score', collapse=testw3$id) resid(fit0, type='scho') resid(fit, type='mart', collapse=testw3$id) resid(fit, type='score', collapse=testw3$id) resid(fit, type='scho') fit0 <- coxph(Surv(begin, time, status) ~x,testw3, weights=wt, iter=0) resid(fit0, 'mart', collapse=testw3$id) resid(coxph(Surv(begin, time, status) ~1, testw3, weights=wt) , collapse=testw3$id) #Null model fit <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, method='efron') fit resid(fit, type='mart', collapse=testw3$id) resid(fit, type='score', collapse=testw3$id) resid(fit, type='scho') # # Check out the impact of weights on the dfbetas # Am I computing them correctly? # wtemp <- rep(1,26) wtemp[c(5,10,15)] <- 2:4 fit <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp) rr <- resid(fit, 'dfbeta') fit1 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, subset=(-5)) fit2 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, subset=(-10)) fit3 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, subset=(-15)) # # Effect of case weights on expected survival curves post Cox model # fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow', iter=0) fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) surv1 <- survfit(fit0, newdata=list(x=0)) surv2 <- survfit(fit0b, newdata=list(x=0)) aeq(surv1$surv, surv2$surv) # # Check out the Efron approx. # fit0 <- coxph(Surv(time, status) ~x,testw1, weights=wt, iter=0) fit <- coxph(Surv(time, status) ~x,testw1, weights=wt) resid(fit0, 'mart') resid(coxph(Surv(time, status) ~1, testw1, weights=wt)) #Null model # lfun is the known log-likelihood for this data set, worked out in the # appendix of Therneau and Grambsch # ufun is the score vector and ifun the information matrix lfun <- function(beta) { r <- exp(beta) a <- 7*r +3 b <- 4*r +2 11*beta - ( log(r^2 + 11*r +7) + (10/3)*(log(a+b) + log(2*a/3 +b) + log(a/3 +b)) + 2*log(2*r +1)) } aeq(fit0$log[1], lfun(0)) aeq(fit$log[2], lfun(fit$coef)) ufun <- function(beta, efron=T) { #score statistic r <- exp(beta) xbar1 <- (2*r^2+11*r)/(r^2+11*r +7) xbar2 <- 11*r/(11*r +5) xbar3 <- 2*r/(2*r +1) xbar2b<- 26*r/(26*r+12) xbar2c<- 19*r/(19*r + 9) temp <- 11 - (xbar1 + 2*xbar3) if (efron) temp - (10/3)*(xbar2 + xbar2b + xbar2c) else temp - 10*xbar2 } print(ufun(fit$coef) < 1e-4) # Should be true ifun <- function(beta, efron=T) { # information matrix r <- exp(beta) xbar1 <- (2*r^2+11*r)/(r^2+11*r +7) xbar2 <- 11*r/(11*r +5) xbar3 <- 2*r/(2*r +1) xbar2b<- 26*r/(26*r+12) xbar2c<- 19*r/(19*r + 9) temp <- ((4*r^2 + 11*r)/(r^2+11*r +7) - xbar1^2) + 2*(xbar3 - xbar3^2) if (efron) temp + (10/3)*((xbar2- xbar2^2) + (xbar2b - xbar2b^2) + (xbar2c -xbar2c^2)) else temp + 10 * (xbar2- xbar2^2) } aeq(fit0$var, 1/ifun(0)) aeq(fit$var, 1/ifun(fit$coef)) # Make sure that the weights pass through the residuals correctly rr1 <- resid(fit, type='mart') rr2 <- resid(fit, type='mart', weighted=T) aeq(rr2/rr1, testw1$wt) rr1 <- resid(fit, type='score') rr2 <- resid(fit, type='score', weighted=T) aeq(rr2/rr1, testw1$wt) # # Look at the individual components # dt0 <- coxph.detail(fit0) dt <- coxph.detail(fit) aeq(sum(dt$score), ufun(fit$coef)) #score statistic aeq(sum(dt0$score), ufun(0)) aeq(dt0$hazard, c(1/19, (10/3)*(1/16 + 1/(6+20/3) + 1/(6+10/3)), 2/3)) rm(fit, fit0, rr1, rr2, dt, dt0) # # Effect of weights on the robust variance # test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0), wt= c(3,0,1,1,1,1,1), id= 1:7) testx <- data.frame(time= c(4,4,4,1,1,2,2,3), status=c(1,1,1,1,0,1,1,0), x= c(0,0,0,1,1,1,0,0), wt= c(1,1,1,1,1,1,1,1), id= 1:8) fit1 <- coxph(Surv(time, status) ~x + cluster(id), test1, method='breslow', weights=wt) fit2 <- coxph(Surv(time, status) ~x + cluster(id), testx, method='breslow') db1 <- resid(fit1, 'dfbeta', weighted=F) db1 <- db1[-2] #toss the missing db2 <- resid(fit2, 'dfbeta') aeq(db1, db2[3:8]) W <- c(3,1,1,1,1,1) #Weights, after removal of the missing value aeq(fit2$var, sum(db1*db1*W)) aeq(fit1$var, sum(db1*db1*W*W)) survival/tests/tiedtime.Rout.save0000644000175100001440000000270311732700061016720 0ustar hornikusers R version 2.12.2 (2011-02-25) Copyright (C) 2011 The R Foundation for Statistical Computing ISBN 3-900051-07-0 Platform: x86_64-unknown-linux-gnu (64-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > library(survival) Loading required package: splines > > # > # The survival code was failing for certain data sets when called as > # survfit(Surv(time2-time1, status) ~ ...... > # The issue was how tied floating point numbers are handled, and the > # fact that unique(x), factor(x) and tapply(x) are not guarranteed to > # all be the same. > # This test fails in survival 2.36-5, fixed in 2.36-6. Data sets that > # can cause it are few and far between. > # > > load('ties.rda') > x <- time2 -time1 > > # Here is the heart of the old problem > # length(unique(x))== length(table(x)) > # And the prior fix which worked ALMOST always > # x <- round(x, 15) > # length(unique(round(x,15)))== length(table(round(x,15))) > > fit1 <- survfit(Surv(x) ~1) > length(fit1$time) == length(fit1$surv) [1] TRUE > survival/tests/coxsurv3.R0000644000175100001440000000742411732700061015230 0ustar hornikuserslibrary(survival) aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) # One more test on coxph survival curves, to test out the individual # option. First fit a model with a time dependent covariate # test2 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) # True hazard function, from the validation document lambda <- function(beta, x=0, method='efron') { r <- exp(beta) lambda <- c(1/(r+1), 1/(r+2), 1/(3*r +2), 1/(3*r+1), 1/(3*r+1), 1/(3*r+2) + 1/(2*r +2)) if (method == 'breslow') lambda[9] <- 2/(3*r +2) list(time=c(2,3,6,7,8,9), lambda=lambda) } fit <- coxph(Surv(start, stop, event) ~x, test2) # A curve for someone who never changes surv1 <-survfit(fit, newdata=list(x=0), censor=FALSE) true <- lambda(fit$coef, 0) aeq(true$time, surv1$time) aeq(-log(surv1$surv), cumsum(true$lambda)) # Reprise it with a time dependent subject who doesn't change data2 <- data.frame(start=c(0, 4, 9, 11), stop=c(4, 9, 11, 17), event=c(0,0,0,0), x=c(0,0,0,0)) surv2 <- survfit(fit, newdata=data2, individual=TRUE, censor=FALSE) aeq(surv2$surv, surv1$surv) # # Now a more complex data set with multiple strata # test3 <- data.frame(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17, 1:11), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0,1), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0, 1, 2, 3, 2, 1, 1, 1, 0, 2, 1,0), grp = c(rep('a', 10), rep('b', 11))) fit2 <- coxph(Surv(start, stop, event) ~ x + strata(grp), test3) # The above tests show the program works for a simple case, use it to # get a true baseline for strata 2 fit2b <- coxph(Surv(start, stop, event) ~x, test3, subset=(grp=='b'), init=fit2$coef, iter=0) temp <- survfit(fit2b, newdata=list(x=0), censor=F) true2 <- list(time=temp$time, lambda=diff(c(0, -log(temp$surv)))) true1 <- lambda(fit2$coef, x=0) # Separate strata, one value surv3 <- survfit(fit2, list(x=0), censor=FALSE) aeq(true1$time, (surv3[1])$time) aeq(-log(surv3[1]$surv), cumsum(true1$lambda)) data4 <- data.frame(start=c(0, 4, 9, 11), stop=c(4, 9, 11, 17), event=c(0,0,0,0), x=c(0,0,0,0), grp=rep('a', 4)) surv4a <- survfit(fit2, newdata=data4, individual=T, censor=FALSE) aeq(-log(surv4a$surv), cumsum(true1$lambda)) data4$grp <- rep('b',4) surv4b <- survfit(fit2, newdata=data4, individual=T, censor=FALSE) aeq(-log(surv4b$surv), cumsum(true2$lambda)) # Now for something more complex # Subject 1 skips day 4. Since there were no events that day the survival # will be the same, but the times will be different. # Subject 2 spends some time in strata 1, some in strata 2, with # moving covariates # data5 <- data.frame(start=c(0,5,9,11, 0, 4, 3), stop =c(4,9,11,17, 4,8,7), event=rep(0,7), x=c(1,1,1,1, 0,1,2), grp=c('a', 'a', 'a', 'a', 'a', 'a', 'b'), subject=c(1,1,1,1, 2,2,2)) surv5 <- survfit(fit2, newdata=data5, censor=FALSE, id=subject) aeq(surv5[1]$time, c(2,3,5,6,7,8)) #surv1 has 2, 3, 6, 7, 8, 9 aeq(surv5[1]$surv, surv3[1]$surv ^ exp(fit2$coef)) tlam <- c(true1$lambda[1:2]* exp(fit2$coef * data5$x[5]), true1$lambda[3:5]* exp(fit2$coef * data5$x[6]), true2$lambda[3:4]* exp(fit2$coef * data5$x[7])) aeq(-log(surv5[2]$surv), cumsum(tlam)) survival/tests/r_sas.R.orig0000644000175100001440000002302712111775106015503 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Reproduce example 1 in the SAS lifereg documentation # # this fit doesn't give the same log-lik that they claim motor <- read.table('data.motor', col.names=c('temp', 'time', 'status')) fit1 <- survreg(Surv(time, status) ~ I(1000/(273.2+temp)), motor, subset=(temp>150), dist='lognormal') summary(fit1) # This one, with the loglik on the transformed scale (the inappropriate # scale, Ripley & Venables would argue) does agree. # All coefs are of course identical. fit2 <- survreg(Surv(log(time), status) ~ I(1000/(273.2+temp)), motor, subset=(temp>150), dist='gaussian') # Give the quantile estimates, which is the lower half of "output 48.1.5" # in the SAS 9.2 manual pp1 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), type='quantile', se=T) pp2 <- predict(fit1, newdata=list(temp=c(130,150)), p=c(.1, .5, .9), type='uquantile', se=T) pp1 temp130 <- matrix(0, nrow=3, ncol=6) temp130[,1] <- pp1$fit[1,] temp130[,2] <- pp1$se.fit[1,] temp130[,3] <- pp2$fit[1,] temp130[,4] <- pp2$se.fit[1,] temp130[,5] <- exp(pp2$fit[1,] - 1.64*pp2$se.fit[1,]) temp130[,6] <- exp(pp2$fit[1,] + 1.64*pp2$se.fit[1,]) dimnames(temp130) <- list(c("p=.1", "p=.2", "p=.3"), c("Time", "se(time)", "log(time)", "se[log(time)]", "lower 90", "upper 90")) print(temp130) # A set of examples, copied from the manual pages of SAS procedure # "reliability", which is part of their QC product. # color <- c("black", "red", "green", "blue", "magenta", "red4", "orange", "DarkGreen", "cyan2", "DarkViolet") palette(color) pdf(file='reliability.pdf') # # Insulating fluids example # fluid <- read.table('data.fluid', col.names=c('time', 'voltage')) # Adding a -1 to the fit just causes the each group to have it's own # intercept, rather than a global intercept + constrasts. The strata # statement allows each to have a separate scale ffit <- survreg(Surv(time) ~ voltage + strata(voltage) -1, fluid) # Get predicted quantiles at each of the voltages # By default predict() would give a line of results for each observation, # I only want the unique set of x's, i.e., only 4 cases uvolt <- sort(unique(fluid$voltage)) #the unique levels plist <- c(1, 2, 5, 1:9 *10, 95, 99)/100 pred <- predict(ffit, type='quantile', p=plist, newdata=data.frame(voltage=factor(uvolt))) tfun <- function(x) log(-log(1-x)) matplot(t(pred), tfun(plist), type='l', log='x', lty=1, col=1:4, yaxt='n') axis(2, tfun(plist), format(100*plist), adj=1) kfit <- survfit(Surv(time) ~ voltage, fluid, type='fleming') #KM fit for (i in 1:4) { temp <- kfit[i] points(temp$time, tfun(1-temp$surv), col=i, pch=i) } # Now a table temp <- array(0, dim=c(4,4,4)) #4 groups by 4 parameters by 4 stats temp[,1,1] <- ffit$coef # "EV Location" in SAS manual temp[,2,1] <- ffit$scale # "EV scale" temp[,3,1] <- exp(ffit$coef) # "Weibull Scale" temp[,4,1] <- 1/ffit$scale # "Weibull Shape" temp[,1,2] <- sqrt(diag(ffit$var))[1:4] #standard error temp[,2,2] <- sqrt(diag(ffit$var))[5:8] * ffit$scale temp[,3,2] <- temp[,1,2] * temp[,3,1] temp[,4,2] <- temp[,2,2] / (temp[,2,1])^2 temp[,1,3] <- temp[,1,1] - 1.96*temp[,1,2] #lower conf limits temp[,1,4] <- temp[,1,1] + 1.96*temp[,1,2] # upper # log(scale) is the natural parameter, in which the routine did its fitting # and on which the std errors were computed temp[,2, 3] <- exp(log(ffit$scale) - 1.96*sqrt(diag(ffit$var))[5:8]) temp[,2, 4] <- exp(log(ffit$scale) + 1.96*sqrt(diag(ffit$var))[5:8]) temp[,3, 3:4] <- exp(temp[,1,3:4]) temp[,4, 3:4] <- 1/temp[,2,4:3] dimnames(temp) <- list(uvolt, c("EV Location", "EV Scale", "Weibull scale", "Weibull shape"), c("Estimate", "SE", "lower 95% CI", "uppper 95% CI")) print(aperm(temp, c(2,3,1)), digits=5) rm(temp, uvolt, plist, pred, ffit, kfit) ##################################################################### # Turbine cracks data cracks <- read.table('data.cracks', col.names=c('time1', 'time2', 'n')) cfit <- survreg(Surv(time1, time2, type='interval2') ~1, dist='weibull', data=cracks, weight=n) summary(cfit) #Their output also has Wiebull scale = exp(cfit$coef), shape = 1/(cfit$scale) # Draw the SAS plot # The "type=fleming" argument reflects that they estimate hazards rather than # survival, and forces a Nelson-Aalen hazard estimate # plist <- c(1, 2, 5, 1:8 *10)/100 plot(qsurvreg(plist, cfit$coef, cfit$scale), tfun(plist), log='x', yaxt='n', type='l', xlab="Weibull Plot for Time", ylab="Percent") axis(2, tfun(plist), format(100*plist), adj=1) kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=cracks, weight=n, type='fleming') # Only plot point where n.event > 0 # Why? I'm trying to match them. Personally, all should be plotted. who <- (kfit$n.event > 0) points(kfit$time[who], tfun(1-kfit$surv[who]), pch='+') points(kfit$time[who], tfun(1-kfit$upper[who]), pch='-') points(kfit$time[who], tfun(1-kfit$lower[who]), pch='-') text(rep(3,6), seq(.5, -1.0, length=6), c("Scale", "Shape", "Right Censored", "Left Censored", "Interval Censored", "Fit"), adj=0) text(rep(9,6), seq(.5, -1.0, length=6), c(format(round(exp(cfit$coef), 2)), format(round(1/cfit$scale, 2)), format(tapply(cracks$n, cfit$y[,3], sum)), "ML"), adj=1) # Now a portion of his percentiles table # I don't get the same SE as SAS, I haven't checked out why. The # estimates and se for the underlying Weibull model are the same. temp <- predict(cfit, type='quantile', p=plist, se=T) tempse <- sqrt(temp$se[1,]) mat <- cbind(temp$fit[1,], tempse, temp$fit[1,] -1.96*tempse, temp$fit[1,] + 1.96*tempse) dimnames(mat) <- list(plist*100, c("Estimate", "SE", "Lower .95", "Upper .95")) print(mat) # # The cracks data has a particularly easy estimate, so use # it to double check code time <- c(cracks$time2[1], (cracks$time1 + cracks$time2)[2:8]/2, cracks$time1[9]) cdf <- cumsum(cracks$n)/sum(cracks$n) all.equal(kfit$time, time) all.equal(kfit$surv, 1-cdf[c(1:8,8)]) rm(time, cdf, kfit) ####################################################### # # Valve data # The input data has id, time, and an indicator of whether there was an # event at that time: -1=no, 1=yes. No one has an event at their last time. # Convert the data to (start, stop] form # The input data has two engines with dual failures: 328 loses 2 valves at # time 653, and number 402 loses 2 at time 139. For each, fudge the first # time to be .1 days earlier. # temp <- matrix(scan('data.valve'), byrow=T, ncol=3) n <- nrow(temp) valve <- data.frame(id=temp[,1], time1 = c(0, ifelse(diff(temp[,1])==0, temp[-n,2],0)), time2 = temp[,2], status= as.numeric(temp[,3]==1)) indx <- (1:nrow(valve))[valve$time1==valve$time2] valve$time1[indx] <- valve$time1[indx] - .1 valve$time2[indx-1] <- valve$time2[indx-1] - .1 kfit <- survfit(Surv(time1, time2, status) ~1, valve, type='fh2') plot(kfit, fun='cumhaz', ylab="Sample Mean Cumulative Failures", xlab='Time', ylim=range(-log(kfit$lower))) title("Valve replacement data") # The summary.survfit function doesn't have an option for printing out # cumulative hazards instead of survival --- need to add that # so I just reprise the central code of print.summary.survfit xx <- summary(kfit) temp <- cbind(xx$time, xx$n.risk, xx$n.event, -log(xx$surv), xx$std.err/xx$surv, -log(xx$upper), -log(xx$lower)) dimnames(temp) <- list(rep("", nrow(temp)), c("time", "n.risk", "n.event", "Cum haz", "std.err", "lower 95%", "upper 95%")) print(temp, digits=2) # Note that I have the same estimates but different SE's. We are using a # different estimator. It's a statistical argument as to which is # better (one could defend both sides): do you favor JASA or Technometrics? rm(temp, kfit, indx, xx) ###################################################### # Turbine data, lognormal fit turbine <- read.table('data.turbine', col.names=c("time1", "time2", "n")) tfit <- survreg(Surv(time1, time2, type='interval2') ~1, turbine, dist='lognormal', weights=n, subset=(n>0)) summary(tfit) # Now, do his plot, but put bootstrap confidence bands on it! # First, make a simple data set without weights tdata <- turbine[rep(1:nrow(turbine), turbine$n),] qstat <- function(data) { temp <- survreg(Surv(time1, time2, type='interval2') ~1, data=data, dist='lognormal') qsurvreg(plist, temp$coef, temp$scale, dist='lognormal') } {if (exists('bootstrap')) { bfit <- bootstrap(tdata, qstat, B=1000) bci <- limits.bca(bfit, probs=c(.025, .975)) } else { values <- matrix(0, nrow=1000, ncol=length(plist)) n <- nrow(tdata) for (i in 1:1000) { subset <- sample(1:n, n, replace=T) values[i,] <- qstat(tdata[subset,]) } bci <- t(apply(values,2, quantile, c(.05, .95))) } } xmat <- cbind(qsurvreg(plist, tfit$coef, tfit$scale, dist='lognormal'), bci) matplot(xmat, qnorm(plist), type='l', lty=c(1,2,2), col=c(1,1,1), log='x', yaxt='n', ylab='Percent', xlab='Time of Cracking (Hours x 100)') axis(2, qnorm(plist), format(100*plist), adj=1) title("Turbine Data") kfit <- survfit(Surv(time1, time2, type='interval2') ~1, data=tdata) points(kfit$time, qnorm(1-kfit$surv), pch='+') survival/tests/book7.R0000644000175100001440000000337612030334160014452 0ustar hornikuserslibrary(survival) options(na.action=na.exclude) options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type # # Tests from the appendix of Therneau and Grambsch # Data set 1 + exact method test1 <- data.frame(time= c(9, 3,1,1,6,6,8), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) byhand7 <- function(beta) { r <- exp(beta) loglik <- 2*(beta - log(3*r + 3)) u <- 2/(r+1) imat <- 2*r/(r+1)^2 haz <- c(1/(3*r+3), 2/(r+3), 0, 1 ) ties <- c(1,1,2,2,3,4) wt <- c(r,r,r,1,1,1) mart <- c(1,0,1,1,0,1) - wt* (cumsum(haz))[ties] #martingale residual list(loglik=loglik, u=u, imat=imat, mart=mart) } aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) fit0 <-coxph(Surv(time, status) ~x, test1, iter=0, method='exact') truth0 <- byhand7(0) aeq(truth0$loglik, fit0$loglik[1]) aeq(1/truth0$imat, fit0$var) aeq(truth0$mart, fit0$resid[c(2:6,1)]) fit1 <- coxph(Surv(time, status) ~x, test1, iter=1, method='exact') aeq(fit1$coef, truth0$u*fit0$var) truth1 <- byhand7(fit1$coef) aeq(fit1$loglik[2], truth1$loglik) aeq(1/truth1$imat, fit1$var) aeq(truth1$mart, resid(fit1)[c(3:7,1)]) # Beta is infinite for this model, so we will get a warning message fit2 <- coxph(Surv(time, status) ~x, test1, method='exact') aeq(resid(fit2)[-2], c(0, 2/3, -1/3, -4/3, 1, 0)) #values from the book # # Now a multivariate case: start/stop data uses a different C routine # zz <- rep(0, nrow(lung)) fit1 <- coxph(Surv(time, status) ~ age + ph.ecog + sex, lung, method="exact") fit2 <- coxph(Surv(zz, time, status) ~ age + ph.ecog + sex, lung, method="exact") aeq(fit1$loglik, fit2$loglik) aeq(fit1$var, fit2$var) aeq(fit1$score, fit2$score) aeq(fit1$resid, fit2$resid) survival/tests/data.turbine0000644000175100001440000000217311732700061015610 0ustar hornikusersNA 4 0 4 NA 39 NA 10 4 10 NA 49 NA 14 2 14 NA 31 NA 18 7 18 NA 66 NA 22 5 22 NA 25 NA 26 9 26 NA 30 NA 30 9 30 NA 33 NA 34 6 34 NA 7 NA 38 22 38 NA 12 NA 42 21 42 NA 19 NA 46 21 46 NA 15 survival/tests/coxsurv.R0000644000175100001440000000545212021713263015144 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # # Test out subscripting in the case of a coxph survival curve # aeq <- function(x,y, ...) all.equal(as.vector(x), as.vector(y), ...) fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), data=cancer) surv1 <- survfit(fit) temp <- surv1[2:3] which <- cumsum(surv1$strata) zed <- (which[1]+1):(which[3]) aeq(surv1$surv[zed], temp$surv) aeq(surv1$time[zed], temp$time) # This call should not create a model frame in the code -- so same # answer but a different path through the underlying code fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), x=T, data=cancer) surv2 <- survfit(fit) all.equal(surv1, surv2) # # Now a result with a matrix of survival curves # dummy <- data.frame(age=c(30,40,60), sex=c(1,2,2), meal.cal=c(500, 1000, 1500)) surv2 <- survfit(fit, newdata=dummy) zed <- 1:which[1] aeq(surv2$surv[zed,1], surv2[1,1]$surv) aeq(surv2$surv[zed,2], surv2[1,2]$surv) aeq(surv2$surv[zed,3], surv2[1,3]$surv) aeq(surv2$surv[zed, ], surv2[1,1:3]$surv) aeq(surv2$surv[zed], (surv2[1]$surv)[,1]) aeq(surv2$surv[zed, ], surv2[1, ]$surv) fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), data=cancer) surv1 <- survfit(fit) temp <- surv1[2:3] which <- cumsum(surv1$strata) zed <- (which[1]+1):(which[3]) aeq(surv1$surv[zed], temp$surv) aeq(surv1$time[zed], temp$time) # This call should not create a model frame in the code -- so same # answer but a different path through the underlying code fit <- coxph(Surv(time, status) ~ age + sex + meal.cal + strata(ph.ecog), x=T, data=cancer) surv2 <- survfit(fit) all.equal(surv1, surv2) # # Now a result with a matrix of survival curves # dummy <- data.frame(age=c(30,40,60), sex=c(1,2,2), meal.cal=c(500, 1000, 1500)) surv2 <- survfit(fit, newdata=dummy) zed <- 1:which[1] aeq(surv2$surv[zed,1], surv2[1,1]$surv) aeq(surv2$surv[zed,2], surv2[1,2]$surv) aeq(surv2$surv[zed,3], surv2[1,3]$surv) aeq(surv2$surv[zed, ], surv2[1,1:3]$surv) aeq(surv2$surv[zed], (surv2[1]$surv)[,1]) aeq(surv2$surv[zed, ], surv2[1, ]$surv) # And the depreciated form - call with a named vector as 'newdata' # the resulting $call component won't match so delete it before comparing surv3 <- survfit(fit, c(age=40, sex=2, meal.cal=1000)) all.equal(unclass(surv2[,2])[-length(surv3)], unclass(surv3)[-length(surv3)]) # Test out offsets, which have recently become popular due to a Langholz paper fit1 <- coxph(Surv(time, status) ~ age + ph.ecog, lung) fit2 <- coxph(Surv(time, status) ~ age + offset(ph.ecog * fit1$coef[2]), lung) surv1 <- survfit(fit1, newdata=data.frame(age=50, ph.ecog=1)) surv2 <- survfit(fit2, newdata=data.frame(age=50, ph.ecog=1)) all.equal(surv1$surv, surv2$surv) survival/tests/doweight.Rout.save0000644000175100001440000003571212536400567016747 0ustar hornikusers R Under development (unstable) (2015-06-04 r68474) -- "Unsuffered Consequences" Copyright (C) 2015 The R Foundation for Statistical Computing Platform: i686-pc-linux-gnu (32-bit) R is free software and comes with ABSOLUTELY NO WARRANTY. You are welcome to redistribute it under certain conditions. Type 'license()' or 'licence()' for distribution details. R is a collaborative project with many contributors. Type 'contributors()' for more information and 'citation()' on how to cite R or R packages in publications. Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R. > options(na.action=na.exclude) # preserve missings > options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type > library(survival) > > # Tests of the weighted Cox model > # > # Similar data set to test1, but add weights, > # a double-death/censor tied time > # a censored last subject > # The latter two are cases covered only feebly elsewhere. > # > # The data set testw2 has the same data, but done via replication > # > aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) > > testw1 <- data.frame(time= c(1,1,2,2,2,2,3,4,5), + status= c(1,0,1,1,1,0,0,1,0), + x= c(2,0,1,1,0,1,0,1,0), + wt = c(1,2,3,4,3,2,1,2,1)) > xx <- c(1,2,3,4,3,2,1,2,1) > testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), + status= rep(c(1,0,1,1,1,0,0,1,0), xx), + x= rep(c(2,0,1,1,0,1,0,1,0), xx), + id= rep(1:9, xx)) > indx <- match(1:9, testw2$id) > testw2 <- data.frame(time= rep(c(1,1,2,2,2,2,3,4,5), xx), + status= rep(c(1,0,1,1,1,0,0,1,0), xx), + x= rep(c(2,0,1,1,0,1,0,1,0), xx), + id= rep(1:9, xx)) > indx <- match(1:9, testw2$id) > > fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, + method='breslow', iter=0) > fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) > fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow') > fitb <- coxph(Surv(time, status) ~x, testw2, method='breslow') > > texp <- function(beta) { # expected, Breslow estimate + r <- exp(beta) + temp <- cumsum(c(1/(r^2 + 11*r +7), 10/(11*r +5), 2/(2*r+1))) + c(r^2, 1,r,r,1,r,1,r,1)* temp[c(1,1,2,2,2,2,2,3,3)] + } > aeq(texp(0), c(1/19, 1/19, rep(103/152, 5), rep(613/456,2))) #verify texp() [1] TRUE > > xbar <- function(beta) { # xbar, Breslow estimate + r <- exp(beta) + temp <- r* rep(c(2*r + 11, 11/10, 1), c(2, 5, 2)) + temp * texp(beta) + } > > fit0 Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, method = "breslow", iter = 0) coef exp(coef) se(coef) z p x 0.000 1.000 0.586 0 1 Likelihood ratio test=0 on 1 df, p=1 n= 9, number of events= 5 > summary(fit) Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, method = "breslow") n= 9, number of events= 5 coef exp(coef) se(coef) z Pr(>|z|) x 0.8596 2.3621 0.7131 1.205 0.228 exp(coef) exp(-coef) lower .95 upper .95 x 2.362 0.4233 0.5839 9.556 Concordance= 0.638 (se = 0.159 ) Rsquare= 0.171 (max possible= 0.999 ) Likelihood ratio test= 1.69 on 1 df, p=0.1932 Wald test = 1.45 on 1 df, p=0.2281 Score (logrank) test = 1.52 on 1 df, p=0.217 > aeq(resid(fit0), testw1$status - texp(0)) [1] TRUE > resid(fit0, type='score') 1 2 3 4 5 6 1.24653740 0.03601108 0.10056700 0.10056700 -0.22180142 -0.21193300 7 8 9 0.46569858 -0.10082189 0.91014302 > resid(fit0, type='scho') 1 2 2 2 4 1.3157895 0.3125000 0.3125000 -0.6875000 0.3333333 > > aeq(resid(fit0, type='mart'), (resid(fit0b, type='mart'))[indx]) [1] TRUE > aeq(resid(fit0, type='scor'), (resid(fit0b, type='scor'))[indx]) [1] TRUE > aeq(unique(resid(fit0, type='scho')), unique(resid(fit0b, type='scho'))) [1] TRUE > > > aeq(resid(fit, type='mart'), testw1$status - texp(fit$coef)) [1] TRUE > resid(fit, type='score') 1 2 3 4 5 6 0.88681615 0.02497653 0.03608964 0.03608964 -0.54297652 -0.12528780 7 8 9 0.29564605 -0.09476911 0.58400064 > resid(fit, type='scho') 1 2 2 2 4 1.0368337 0.1613774 0.1613774 -0.8386226 0.1746960 > aeq(resid(fit, type='mart'), (resid(fitb, type='mart'))[indx]) [1] TRUE > aeq(resid(fit, type='scor'), (resid(fitb, type='scor'))[indx]) [1] TRUE > aeq(unique(resid(fit, type='scho')), unique(resid(fitb, type='scho'))) [1] TRUE > rr1 <- resid(fit, type='mart') > rr2 <- resid(fit, type='mart', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > rr1 <- resid(fit, type='score') > rr2 <- resid(fit, type='score', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > fit <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='efron') > fit Call: coxph(formula = Surv(time, status) ~ x, data = testw1, weights = wt, method = "efron") coef exp(coef) se(coef) z p x 0.873 2.393 0.713 1.22 0.22 Likelihood ratio test=1.75 on 1 df, p=0.186 n= 9, number of events= 5 > resid(fit, type='mart') 1 2 3 4 5 6 0.85334536 -0.02560716 0.32265266 0.32265266 0.71696234 -1.07772629 7 8 9 -0.45034077 -0.90490339 -0.79598658 > resid(fit, type='score') 1 2 3 4 5 6 0.88116056 0.02477248 0.06057806 0.06057806 -0.59724033 -0.16737066 7 8 9 0.38040295 -0.13750290 0.66631324 > resid(fit, type='scho') 1 2 2 2 4 1.0325955 0.1621759 0.1621759 -0.8378241 0.1728229 > > # Tests of the weighted Cox model, AG form of the data > # Same solution as doweight1.s > # > testw3 <- data.frame(id = c( 1, 1, 2, 3, 3, 3, 4, 5, 5, 6, 7, 8, 8, 9), + begin= c( 0, 5, 0, 0,10,15, 0, 0,14, 0, 0, 0,23, 0), + time= c( 5,10,10,10,15,20,20,14,20,20,30,23,40,50), + status= c( 0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 1, 0), + x= c( 2, 2, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 0), + wt = c( 1, 1, 2, 3, 3, 3, 4, 3, 3, 2, 1, 2, 2, 1)) > > fit0 <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, + method='breslow', iter=0) > fit <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, method='breslow') > fit0 Call: coxph(formula = Surv(begin, time, status) ~ x, data = testw3, weights = wt, method = "breslow", iter = 0) coef exp(coef) se(coef) z p x 0.000 1.000 0.586 0 1 Likelihood ratio test=0 on 1 df, p=1 n= 14, number of events= 5 > summary(fit) Call: coxph(formula = Surv(begin, time, status) ~ x, data = testw3, weights = wt, method = "breslow") n= 14, number of events= 5 coef exp(coef) se(coef) z Pr(>|z|) x 0.8596 2.3621 0.7131 1.205 0.228 exp(coef) exp(-coef) lower .95 upper .95 x 2.362 0.4233 0.5839 9.556 Concordance= 0.638 (se = 0.159 ) Rsquare= 0.114 (max possible= 0.991 ) Likelihood ratio test= 1.69 on 1 df, p=0.1932 Wald test = 1.45 on 1 df, p=0.2281 Score (logrank) test = 1.52 on 1 df, p=0.217 > resid(fit0, type='mart', collapse=testw3$id) 1 2 3 4 5 6 0.94736842 -0.05263158 0.32236842 0.32236842 0.32236842 -0.67763158 7 8 9 -0.67763158 -0.34429825 -1.34429825 > resid(fit0, type='score', collapse=testw3$id) 1 2 3 4 5 6 1.24653740 0.03601108 0.10056700 0.10056700 -0.22180142 -0.21193300 7 8 9 0.46569858 -0.10082189 0.91014302 > resid(fit0, type='scho') 10 20 20 20 40 1.3157895 0.3125000 0.3125000 -0.6875000 0.3333333 > > resid(fit, type='mart', collapse=testw3$id) 1 2 3 4 5 6 0.85531186 -0.02593169 0.17636221 0.17636221 0.65131344 -0.82363779 7 8 9 -0.34868656 -0.64894181 -0.69807852 > resid(fit, type='score', collapse=testw3$id) 1 2 3 4 5 6 0.88681615 0.02497653 0.03608964 0.03608964 -0.54297652 -0.12528780 7 8 9 0.29564605 -0.09476911 0.58400064 > resid(fit, type='scho') 10 20 20 20 40 1.0368337 0.1613774 0.1613774 -0.8386226 0.1746960 > fit0 <- coxph(Surv(begin, time, status) ~x,testw3, weights=wt, iter=0) > resid(fit0, 'mart', collapse=testw3$id) 1 2 3 4 5 6 0.94736842 -0.05263158 0.44454887 0.44454887 0.44454887 -0.88126566 7 8 9 -0.88126566 -0.54793233 -1.54793233 > resid(coxph(Surv(begin, time, status) ~1, testw3, weights=wt) + , collapse=testw3$id) #Null model 1 2 3 4 5 6 0.94736842 -0.05263158 0.44454887 0.44454887 0.44454887 -0.88126566 7 8 9 -0.88126566 -0.54793233 -1.54793233 > > fit <- coxph(Surv(begin,time, status) ~x, testw3, weights=wt, method='efron') > fit Call: coxph(formula = Surv(begin, time, status) ~ x, data = testw3, weights = wt, method = "efron") coef exp(coef) se(coef) z p x 0.873 2.393 0.713 1.22 0.22 Likelihood ratio test=1.75 on 1 df, p=0.186 n= 14, number of events= 5 > resid(fit, type='mart', collapse=testw3$id) 1 2 3 4 5 6 0.85334536 -0.02560716 0.32265266 0.32265266 0.71696234 -1.07772629 7 8 9 -0.45034077 -0.90490339 -0.79598658 > resid(fit, type='score', collapse=testw3$id) 1 2 3 4 5 6 0.88116056 0.02477248 0.06057806 0.06057806 -0.59724033 -0.16737066 7 8 9 0.38040295 -0.13750290 0.66631324 > resid(fit, type='scho') 10 20 20 20 40 1.0325955 0.1621759 0.1621759 -0.8378241 0.1728229 > # > # Check out the impact of weights on the dfbetas > # Am I computing them correctly? > # > wtemp <- rep(1,26) > wtemp[c(5,10,15)] <- 2:4 > fit <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp) > rr <- resid(fit, 'dfbeta') > > fit1 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, + subset=(-5)) > fit2 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, + subset=(-10)) > fit3 <- coxph(Surv(futime, fustat) ~ age + ecog.ps, ovarian, weights=wtemp, + subset=(-15)) > > # > # Effect of case weights on expected survival curves post Cox model > # > fit0 <- coxph(Surv(time, status) ~x, testw1, weights=wt, method='breslow', + iter=0) > fit0b <- coxph(Surv(time, status) ~x, testw2, method='breslow', iter=0) > > surv1 <- survfit(fit0, newdata=list(x=0)) > surv2 <- survfit(fit0b, newdata=list(x=0)) > aeq(surv1$surv, surv2$surv) [1] TRUE > # > # Check out the Efron approx. > # > > fit0 <- coxph(Surv(time, status) ~x,testw1, weights=wt, iter=0) > fit <- coxph(Surv(time, status) ~x,testw1, weights=wt) > resid(fit0, 'mart') 1 2 3 4 5 6 0.94736842 -0.05263158 0.44454887 0.44454887 0.44454887 -0.88126566 7 8 9 -0.88126566 -0.54793233 -1.54793233 > resid(coxph(Surv(time, status) ~1, testw1, weights=wt)) #Null model 1 2 3 4 5 6 0.94736842 -0.05263158 0.44454887 0.44454887 0.44454887 -0.88126566 7 8 9 -0.88126566 -0.54793233 -1.54793233 > > # lfun is the known log-likelihood for this data set, worked out in the > # appendix of Therneau and Grambsch > # ufun is the score vector and ifun the information matrix > lfun <- function(beta) { + r <- exp(beta) + a <- 7*r +3 + b <- 4*r +2 + 11*beta - ( log(r^2 + 11*r +7) + + (10/3)*(log(a+b) + log(2*a/3 +b) + log(a/3 +b)) + 2*log(2*r +1)) + } > aeq(fit0$log[1], lfun(0)) [1] TRUE > aeq(fit$log[2], lfun(fit$coef)) [1] TRUE > > ufun <- function(beta, efron=T) { #score statistic + r <- exp(beta) + xbar1 <- (2*r^2+11*r)/(r^2+11*r +7) + xbar2 <- 11*r/(11*r +5) + xbar3 <- 2*r/(2*r +1) + xbar2b<- 26*r/(26*r+12) + xbar2c<- 19*r/(19*r + 9) + temp <- 11 - (xbar1 + 2*xbar3) + if (efron) temp - (10/3)*(xbar2 + xbar2b + xbar2c) + else temp - 10*xbar2 + } > print(ufun(fit$coef) < 1e-4) # Should be true x TRUE > > ifun <- function(beta, efron=T) { # information matrix + r <- exp(beta) + xbar1 <- (2*r^2+11*r)/(r^2+11*r +7) + xbar2 <- 11*r/(11*r +5) + xbar3 <- 2*r/(2*r +1) + xbar2b<- 26*r/(26*r+12) + xbar2c<- 19*r/(19*r + 9) + temp <- ((4*r^2 + 11*r)/(r^2+11*r +7) - xbar1^2) + + 2*(xbar3 - xbar3^2) + if (efron) temp + (10/3)*((xbar2- xbar2^2) + (xbar2b - xbar2b^2) + + (xbar2c -xbar2c^2)) + else temp + 10 * (xbar2- xbar2^2) + } > > aeq(fit0$var, 1/ifun(0)) [1] TRUE > aeq(fit$var, 1/ifun(fit$coef)) [1] TRUE > > > > # Make sure that the weights pass through the residuals correctly > rr1 <- resid(fit, type='mart') > rr2 <- resid(fit, type='mart', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > rr1 <- resid(fit, type='score') > rr2 <- resid(fit, type='score', weighted=T) > aeq(rr2/rr1, testw1$wt) [1] TRUE > > # > # Look at the individual components > # > dt0 <- coxph.detail(fit0) > dt <- coxph.detail(fit) > aeq(sum(dt$score), ufun(fit$coef)) #score statistic [1] TRUE > aeq(sum(dt0$score), ufun(0)) [1] TRUE > aeq(dt0$hazard, c(1/19, (10/3)*(1/16 + 1/(6+20/3) + 1/(6+10/3)), 2/3)) [1] TRUE > > > > rm(fit, fit0, rr1, rr2, dt, dt0) > # > # Effect of weights on the robust variance > # > test1 <- data.frame(time= c(9, 3,1,1,6,6,8), + status=c(1,NA,1,0,1,1,0), + x= c(0, 2,1,1,1,0,0), + wt= c(3,0,1,1,1,1,1), + id= 1:7) > testx <- data.frame(time= c(4,4,4,1,1,2,2,3), + status=c(1,1,1,1,0,1,1,0), + x= c(0,0,0,1,1,1,0,0), + wt= c(1,1,1,1,1,1,1,1), + id= 1:8) > > fit1 <- coxph(Surv(time, status) ~x + cluster(id), test1, method='breslow', + weights=wt) > fit2 <- coxph(Surv(time, status) ~x + cluster(id), testx, method='breslow') > > db1 <- resid(fit1, 'dfbeta', weighted=F) > db1 <- db1[-2] #toss the missing > db2 <- resid(fit2, 'dfbeta') > aeq(db1, db2[3:8]) [1] TRUE > > W <- c(3,1,1,1,1,1) #Weights, after removal of the missing value > aeq(fit2$var, sum(db1*db1*W)) [1] TRUE > aeq(fit1$var, sum(db1*db1*W*W)) [1] TRUE > > > proc.time() user system elapsed 0.280 0.036 0.313 survival/tests/expected.R0000644000175100001440000002300412257335007015234 0ustar hornikusersoptions(na.action=na.exclude) # preserve missings options(contrasts=c('contr.treatment', 'contr.poly')) #ensure constrast type library(survival) # Tests of expected survival aeq <- function(x,y) all.equal(as.vector(x), as.vector(y)) # # This makes several scripts easier # Certain tests depended in the now-depreciated date library {if (is.R()) mdy.date <- function(m, d, y) { y <- ifelse(y<100, y+1900, y) as.Date(paste(m,d,y, sep='/'), "%m/%d/%Y") } else mdy.date <- function(m,d,y) { y <- ifelse(y<100, y+1900, y) timeDate(paste(y, m, d, sep='/'), in.format="%Y/%m/%d") } } # This function takes a single subject and walks down the rate table # Input: the vector of starting points, futime, and a ratetable # Output: the full history of walking through said table. Let n= #unique # rates that were used # cell = n by #dims of the table: index of the table cell # days = time spent in cell # hazard= accumulated hazard = days * rate # This does not do date or factor conversions -- start has to be numeric # ratewalk <- function(start, futime, ratetable=survexp.us) { if (!is.ratetable(ratetable)) stop("Bad rate table") ratedim <- dim(ratetable) nvar <- length(ratedim) if (length(start) != nvar) stop("Wrong length for start") if (futime <=0) stop("Invalid futime") attR <- attributes(ratetable) discrete <- (attR$type ==1) #discrete categories maxn <- sum(!discrete)*prod(ratedim[!discrete]) #most cells you can hit cell <- matrix(0, nrow=maxn, ncol=nvar) days <- hazard <- double(maxn) eps <- 1e-8 #Avoid round off error n <- 0 while (futime >0) { n <- n+1 #what cell am I in? # Note that at the edges of the rate table, we use the edge: if # it only goes up the the year 2000, year 2000 is used for any # dates beyond. This effectively eliminates one boundary cell[n,discrete] <- start[discrete] edge <- futime #time to nearest edge, or finish for (j in which(!discrete)) { indx <- sum(start[j] >= attR$cutpoints[[j]]-eps) cell[n, j] <- max(1, indx) if (indx < ratedim[j]) edge <- min(edge, (attR$cutpoints[[j]])[indx+1] - start[j]) } days[n] <- edge #this many days in the cell # using a matrix as a subscript is so handy sometimes hazard[n] <- edge * (as.matrix(ratetable))[cell[n,,drop=F]] futime <- futime - edge #amount of time yet to account for start[!discrete] <- start[!discrete] + edge #walk forward in time } list(cell=cell[1:n,], days=days[1:n], hazard=hazard[1:n]) } # Simple test of ratewalk: 20 years old, start on 7Sep 1960 (day 250) # 116 days at the 1960, 20 year old male rate, through the end of the day # on 12/31/1960, then 84 days at the 1961 rate. # The decennial q for 1960 males is .00169. zz <- ratewalk(c(20.4*365.25, 1, 250), 200) all.equal(zz$hazard[1], -(116/365.25)*log(1-.00169)) all.equal(zz$days, c(116,84)) # # Simple case 1: a single male subject, born 1/1/36 and entered on study 1/2/55 # # Compute the 1, 5, 10 and 12 year expected survival temp1 <- mdy.date(1,1,36) temp2 <- mdy.date(1,2,55) exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=1, race='white'), ratetable=survexp.usr,times=c(366, 1827, 3653, 4383)) tyear <- as.numeric(temp2 - mdy.date(1,1,1960)) h1 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 366, survexp.usr) h2 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 1827, survexp.usr) h3 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 3653, survexp.usr) h4 <- ratewalk(c(temp2-temp1, 1, 1, tyear), 4383, survexp.usr) aeq(-log(exp1$surv), c(sum(h1$hazard), sum(h2$hazard), sum(h3$hazard), sum(h4$hazard))) # Just a little harder: # Born 3/1/25 and entered the study on 6/10/55. The code creates shifted # dates to align with US rate tables - entry is 59 days earlier (days from # 1/1/25 to 3/1/25). # temp1 <- mdy.date(3,1,25) temp2 <- mdy.date(6,10,55) exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=2, race='black'), ratetable=survexp.usr,times=c(366, 1827, 3653, 4383)) tyear <- as.numeric(temp2 - mdy.date(1,1,1960)) - 59 h1 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 366, survexp.usr) h2 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 1827, survexp.usr) h3 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 3653, survexp.usr) h4 <- ratewalk(c(temp2-temp1, 2, 2, tyear), 4383, survexp.usr) aeq(-log(exp1$surv), c(sum(h1$hazard), sum(h2$hazard), sum(h3$hazard), sum(h4$hazard))) # # Simple case 2: make sure that the averages are correct, for Ederer method # # Compute the 1, 5, 10 and 12 year expected survival temp1 <- mdy.date(1:6,6:11,1890:1895) temp2 <- mdy.date(6:1,11:6,c(55:50)) temp3 <- c(1,2,1,2,1,2) age <- temp2 - temp1 exp1 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), times=c(366, 1827, 3653, 4383)) exp2 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3) + I(1:6), times=c(366, 1827, 3653, 4383)) exp3 <- exp2$surv for (i in 1:length(temp1)){ exp3[,i] <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), times=c(366, 1827, 3653, 4383), subset=i)$surv } print(aeq(exp2$surv, exp3)) print(all.equal(exp1$surv, apply(exp2$surv, 1, mean))) # They agree, but are they right? # for (i in 1:length(temp1)) { offset <- as.numeric(temp1[i] - mdy.date(1,1, 1889+i)) tyear = (as.numeric(temp2[i] - mdy.date(1,1,1960))) - offset haz1 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 366) haz2 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 1827) haz3 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 3653) haz4 <- ratewalk(c((temp2-temp1)[i], temp3[i], tyear), 4383) print(aeq(-log(exp2$surv[,i]), c(sum(haz1$hazard), sum(haz2$hazard), sum(haz3$hazard), sum(haz4$hazard)))) } # # Check that adding more time points doesn't change things # exp4 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3) + I(1:6), times=sort(c(366, 1827, 3653, 4383, 30*(1:100)))) aeq(exp4$surv[match(exp2$time, exp4$time),], exp2$surv) exp4 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=temp3), times=sort(c(366, 1827, 3653, 4383, 30*(1:100)))) aeq(exp1$surv, exp4$surv[match(exp1$time, exp4$time, nomatch=0)]) # # Now test Hakulinen's method, assuming an analysis date of 3/1/57 # futime <- mdy.date(3,1,57) - temp2 xtime <- sort(c(futime, 30, 60, 185, 365)) exp1 <- survexp(futime ~ ratetable(year=temp2, age=(temp2-temp1), sex=1), times=xtime, conditional=F) exp2 <- survexp(~ratetable(year=temp2, age=(temp2-temp1), sex=1) + I(1:6), times=futime) wt <- rep(1,6) con <- double(6) for (i in 1:6) { con[i] <- sum(exp2$surv[i,i:6])/sum(wt[i:6]) wt <- exp2$surv[i,] } exp1$surv[match(futime, xtime)] aeq(exp1$surv[match(futime, xtime)], cumprod(con)) # # Now for the conditional method # exp1 <- survexp(futime ~ ratetable(year=temp2, age=(temp2-temp1), sex=1), times=xtime, conditional=T) cond <- exp2$surv for (i in 6:2) cond[i,] <- (cond[i,]/cond[i-1,]) #conditional survival for (i in 1:6) con[i] <- exp(mean(log(cond[i, i:6]))) all.equal(exp1$surv[match(futime, xtime)], cumprod(con)) cumprod(con) # # Test out expected survival, when the parent pop is another Cox model # test1 <- data.frame(time= c(4, 3,1,1,2,2,3), status=c(1,NA,1,0,1,1,0), x= c(0, 2,1,1,1,0,0)) fit <- coxph(Surv(time, status) ~x, test1, method='breslow') dummy <- data.frame(time=c(.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5), status=c(1,0,1,0,1,0,1,1,1), x=(-4:4)/2) efit <- survexp(time ~ ratetable(x=x), dummy, ratetable=fit, cohort=F) # # Now, compare to the true answer, which is known to us # ss <- exp(fit$coef) haz <- c( 1/(3*ss+3), 2/(ss+3), 1) #truth at time 0,1,2,4+ chaz <- cumsum(c(0,haz)) chaz2 <- chaz[c(1,2,2,3,3,3,3,4,4)] risk <- exp(fit$coef*dummy$x) efit2 <- exp(-risk*chaz2) all.equal(as.vector(efit), as.vector(efit2)) #ignore mismatched name attrib # # Now test the direct-adjusted curve (Ederer) # efit <- survexp( ~ ratetable(x=x), dummy, ratetable=fit, se=F) direct <- survfit(fit, newdata=dummy, censor=FALSE)$surv chaz <- chaz[-1] #drop time 0 d2 <- exp(outer(-chaz, risk)) all.equal(as.vector(direct), as.vector(d2)) #this tests survfit all.equal(as.vector(efit$surv), as.vector(apply(direct,1,mean))) #direct # Check out the "times" arg of survexp efit2 <- survexp( ~ ratetable(x=x), dummy, ratetable=fit, se=F, times=c(.5, 2, 3.5,6)) aeq(efit2$surv, c(1, efit$surv[c(2,2,3)])) # # Now test out the Hakulinen method (Bonsel's method) # By construction, we have a large correlation between x and censoring # # In theory, hak1 and hak2 would be the same. In practice, like a KM and # F-H, they differ when n is small. # efit <- survexp( time ~ ratetable(x=x), dummy, ratetable=fit, se=F) surv <- wt <- rep(1,9) tt <- c(1,2,4) hak1 <- hak2 <- NULL for (i in 1:3) { wt[dummy$time < tt[i]] <- 0 hak1 <- c(hak1, exp(-sum(haz[i]*risk*surv*wt)/sum(surv*wt))) hak2 <- c(hak2, sum(exp(-haz[i]*risk)*surv*wt)/sum(surv*wt)) surv <- surv * exp(-haz[i]*risk) } all.equal(as.vector(efit$surv), as.vector(cumprod(hak1))) # # Now do the conditional estimate # efit <- survexp( time ~ ratetable(x=x), dummy, ratetable=fit, se=F, conditional=T) wt <- rep(1,9) cond <- NULL for (i in 1:3) { wt[dummy$time < tt[i]] <- 0 cond <- c(cond, exp(-sum(haz[i]*risk*wt)/sum(wt))) } all.equal(as.vector(efit$surv), as.vector(cumprod(cond))) survival/src/0000755000175100001440000000000012545056257012745 5ustar hornikuserssurvival/src/pystep.c0000644000175100001440000000571012545056257014440 0ustar hornikusers/* $Id: pystep.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** Returns the amount of time that will be spent in the current "cell", ** along with the index of the cell (treating a multi-way array as linear). ** This is a basic calculation in all of the person-years work. ** ** Input ** nc: number of categories ** data[nc] start points, for the data values ** fac[nc] 1: category is a factor, 0: it is continuous ** >=2: special handling for "years" dim of US rate tables ** dims[nc] the extent of each category ** cuts[nc,dims+1] ragged array, containing the start for each interval ** step the amount of time remaining for the subject. ** edge if =0, then the cuts contain +1 obs, and we are strict ** about out-of-range cells. If it is a 1, then the ** table is assummed to extend infinitly at the edges. ** ** Output ** *index linear index into the array ** if *index == -1, then the returned amount of time is "off table"; ** if one of the dimensions has fac >1 -- ** *index2 second index for linear interpolation ** *wt a number between 0 and 1, amount of wt for the first index ** this will be 1 if none of the dims have fac >1 ** ** Return value amount of time in indexed cell. */ #include "survS.h" #include "survproto.h" double pystep(int nc, int *index, int *index2, double *wt, double *data, Sint *fac, Sint *dims, double **cuts, double step, int edge) { int i,j; double maxtime; double shortfall; double temp; int kk, dtemp; kk=1; *index =0; *index2=0; *wt =1; shortfall =0; maxtime = step; for (i=0; i1) dtemp = 1 + (fac[i]-1)*dims[i]; else dtemp = dims[i]; for (j=0; j shortfall) { if (temp > step) shortfall = step; else shortfall = temp; } if (temp < maxtime) maxtime = temp; } else if (j==dtemp){ /*bigger than last cutpoint */ if (edge==0) { temp = cuts[i][j] - data[i]; /* time to upper limit */ if (temp <=0) shortfall = step; else if (temp < maxtime) maxtime = temp; } if (fac[i] >1) j = dims[i] -1; /*back to normal indices */ else j--; } else { temp = cuts[i][j] - data[i]; /* time to next cutpoint */ if (temp < maxtime) maxtime = temp; j--; if (fac[i] >1) { /*interpolate the year index */ *wt = 1.0 - (j%fac[i])/ (double)fac[i]; j /= fac[i]; *index2 = kk; } } *index += j*kk; } kk *= dims[i]; } *index2 += *index; if (shortfall ==0) return(maxtime); else { *index = -1; return(shortfall); } } survival/src/survfit4.c0000644000175100001440000000205312545056257014677 0ustar hornikusers/* $Id: survfit4.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** C routine to do a small computation that is hard in Splus ** ** n = number of observations ** d = number of deaths ** x1, x2 = ingredients in the sums ** ** If d=0, then new x1 = new x2 =1 (fill in value) ** d=1, new x1 = 1/x1, ** new x2 = (1/x1)^2 ** d=2, new x1 = (1/2) [ 1/x1 + 1/(x1 - x2/2)] ** new x2 = (1/2) [ same terms, squared] ** d=3 new x1 = (1/3) [ 1/x1 + 1/(x1 - x2/3) + 1/(x1 - 2*x2/3)] ** etc. */ #include "survS.h" void survfit4(Sint *n, Sint *dd, double *x1, double *x2) { double temp, temp1, temp2; int i,j; double d; for (i=0; i< *n; i++) { d = dd[i]; if (d==0) { x1[i] =1; x2[i] =1; } else if (d==1){ temp = 1/x1[i]; x1[i] = temp; x2[i] = temp*temp; } else { temp1 = 1/x1[i]; temp2 = temp1 * temp1; for (j=1; j= dtime; istart++) nrisk--; for(j= i+1; j=dtime; istart++) { atrisk[sort1[istart]]=0; nrisk--; } for (j=1; j0) { matrix[i][i] = 1/matrix[i][i]; /*this line inverts D */ for (j= (i+1); j #include "survS.h" #include "survproto.h" /* my habit is to name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP pyears3b(SEXP death2, SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP grpx2, SEXP x2, SEXP y2, SEXP times2, SEXP ngrp2) { int i,j,k; int n, death, edim, ngrp, ntime; double **x; double *data2; double **ecut, *etemp; double hazard, /*cum hazard over an interval */ cumhaz; /*total hazard to date for the subject */ double timeleft, thiscell, etime, time, et2; int index, indx, indx2; double wt; double *wvec; /* vector of weights needed for unconditional surv */ int group; int *efac, *edims, *grpx; double *expect, *y, *times; SEXP esurv2, nsurv2, rlist, rlistnames; double *esurv; int *nsurv; /* ** copies of input arguments */ death = asInteger(death2); ngrp = asInteger(ngrp2); efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); grpx = INTEGER(grpx2); n = LENGTH(y2); x = dmatrix(REAL(x2), n, edim); y = REAL(y2); times = REAL(times2); ntime = LENGTH(times2); /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); wvec = (double *)ALLOC(ntime*ngrp, sizeof(double)); for (j=0; j1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(esurv2 = allocVector(REALSXP, ntime*ngrp)); esurv = REAL(esurv2); PROTECT(nsurv2 = allocVector(INTSXP, ntime*ngrp)); nsurv = INTEGER(nsurv2); for (i=0; i<(ntime*ngrp); i++) { esurv[i] =0.; nsurv[i] =0; } /* compute */ for (i=0; i0; j++) { thiscell = times[j] - time; if (thiscell > timeleft) thiscell = timeleft; index =j + ntime*group; /* expected calc ** The wt parameter only comes into play for older style US rate ** tables, where pystep does interpolation. ** Each call to pystep moves up to the next 'boundary' in the ** expected table, data2 contains our current position therein */ etime = thiscell; hazard =0; while (etime >0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k0) { if (death==0) esurv[i] /= wvec[i]; else esurv[i] = exp(-esurv[i]/wvec[i]); } else if (death!=0) esurv[i] = exp(-esurv[i]); } /* ** package the output */ PROTECT(rlist = allocVector(VECSXP, 2)); SET_VECTOR_ELT(rlist,0, esurv2); SET_VECTOR_ELT(rlist,1, nsurv2); PROTECT(rlistnames= allocVector(STRSXP, 2)); SET_STRING_ELT(rlistnames, 0, mkChar("surv")); SET_STRING_ELT(rlistnames, 1, mkChar("n")); setAttrib(rlist, R_NamesSymbol, rlistnames); unprotect(4); return(rlist); } survival/src/survS.h0000644000175100001440000000422612545056257014244 0ustar hornikusers/* SCCS $Id: survS.h 11252 2009-03-19 13:46:26Z tlumley $ ** ** The S.h file defines a few things that I need, and hundreds that I don't. ** In particular, on some architectures, it defines a variable "time" ** which of course conflicts with lots of my C-code, 'time' being a natural ** variable name for survival models. ** Thanks to Brian Ripley for suggesting a machine independent way of ** fixing this. ** ** The S_alloc function changed it's argument list from Splus version ** 4 to 5, and there is a different one for R. ** The ALLOC macro allows me to have common C code for all versions, ** with only this file "survS.h" changed. */ #include "R.h" #include "Rinternals.h" #include #ifdef USING_R /* typedef int Sint; */ #define S_EVALUATOR /* Turn this into a "blank line" in R */ #else /* ** Splus definitions, to use R type calls */ typedef long Sint; /* ** At this point in time (Splus 8.0.1) I need to add a little ** to the Insightful definitions. (They are in the process ** of improving Rinternals, so this may well go away.) The ** two functions below are defined as "not supported". I need ** only certain cases of defineVar and eval, so can safely map them. ** I am using the 8.1 R*.h files courtesy of Bill Dunlap */ #ifdef defineVar #undef defineVar #endif #define defineVar(a,b,c) ASSIGN_IN_FRAME(a,b, INTEGER_VALUE(c)) #ifdef eval #undef eval #endif #define eval(a, b) EVAL_IN_FRAME(a, INTEGER_VALUE(b)) /* ** These two refer to undefined functions, so use the 8.0.1 defs */ #ifdef asInteger #undef asInteger #endif #define asInteger(a) INTEGER_VALUE(a) #ifdef asReal #undef asReal #endif #define asReal(a) NUMERIC_VALUE(a) #endif /* ** Memory defined with ALLOC is removed automatically by S. ** That with "Calloc" I have to remove myself. Use the ** latter for objects that need to to persist between calls. */ #ifdef USING_R #define ALLOC(a,b) R_alloc(a,b) #else #define ALLOC(a,b) S_alloc(a,b) #endif /* ** Prototype for callback function ** */ #ifdef USING_R void cox_callback(int which, double *coef, double *first, double *second, double *penalty, int *flag, int p, SEXP fexpr, SEXP rho); #endif survival/src/chsolve2.c0000644000175100001440000000163012545056257014636 0ustar hornikusers/* $Id: chsolve2.c 11376 2009-12-14 22:53:57Z therneau $ ** ** Solve the equation Ab = y, where the cholesky decomposition of A and y ** are the inputs. ** ** Input **matrix, which contains the chol decomp of an n by n ** matrix in its lower triangle. ** y[n] contains the right hand side ** ** y is overwriten with b ** ** Terry Therneau */ #include "survS.h" #include "survproto.h" void chsolve2(double **matrix, int n, double *y) { register int i,j; register double temp; /* ** solve Fb =y */ for (i=0; i=0; i--) { if (matrix[i][i]==0) y[i] =0; else { temp = y[i]/matrix[i][i]; for (j= i+1; j #include "survS.h" #include "survproto.h" void agmart2(Sint *n, Sint *method, double *start, double *stop, Sint *event, Sint *nstrat, Sint *strata, Sint *sort1, Sint *sort2, double *score, double *wt, double *resid, double *haz) { int i, j, k, ksave; int p, istrat, indx2; double deaths, denom, e_denom; double hazard, e_hazard; double temp, time; double wtsum, *dtimes; int nused, ndeath; int person; int strata_start; nused = *n; j=0; for (i=0; i=strata_start; k--) { /*non-deaths */ p = sort1[k]; if (stop[p] > time) break; resid[p] -= score[p]*hazard; } for (; person= stop[p]; k++); for (j=k; j #include "survS.h" #define SMALL -200 /* exp(-200) is a really small loglik */ double survregc2(int n, int nvar, int nstrat, int whichcase, double *beta, int dist, Sint *strat, double *offset, double *time1, double *time2, double *status, double *wt, double **covar, double **imat, double **JJ, double *u, SEXP expr, SEXP rho, double *z, int nf, Sint *frail, double *fdiag, double *jdiag ) { int person, i,j,k; int nvar2; int strata; double eta, sigma; int icount; /* running count of # of interval censored */ int fgrp =0; /* the =0 to quiet a compiler warning */ double loglik, temp; double temp1, temp2; double sz, zz, zu; double sig2; /* add "=0" to keep the compiler from worrying about uninitialized vars */ /* double g, dg, ddg, dsig, ddsig, dsg; */ double g=0, dg=0, ddg=0, dsig=0, ddsig=0, dsg=0; SEXP rmat; double *funs[5]; double w; nvar2 = nvar + nstrat; loglik=0; if (whichcase==0) { for (i=0; i1) { strata= strat[person] -1; /*S likes to start counting at 1 */ sigma = exp(beta[strata+nvar+nf]); } eta =0; for (i=0; i0){ fgrp = frail[person] -1; eta += beta[fgrp]; } z[person] = (time1[person] - eta)/sigma; if (status[person]==3) { z[icount] = (time2[person] - eta)/sigma; icount++; } } /* ** The result of the eval will be a matrix of 5 rows and n colums, which ** we re-index for convenience. Note that the parent routine has given ** us the address of z WITHIN the evaluation frame rho, we just keep ** replacing the values it contains; expr then acts like a function of ** z. ** Actually, if there were any interval censored obs they take up 2 cols; ** icount from above contains the actual number of columns used. */ PROTECT(rmat = eval(expr, rho)); funs[0] = REAL(rmat); for (i=0; i<4; i++) funs[i+1] = funs[i] + icount; /* ** calculate the first and second derivative wrt eta, ** then the derivatives of the loglik (u, imat, JJ) */ icount =n; for (person=0; person1) { strata= strat[person] -1; /*S likes to start counting at 1 */ sigma = exp(beta[strata+nvar]); sig2 = 1/(sigma*sigma); } zz = z[person]; sz = zz * sigma; j = status[person]; /*convert to integer */ switch(j) { case 1: /* exact */ if (funs[2][person] <=0) { /* off the probability scale -- avoid log(0), and set the ** derivatives to gaussian limits (almost any deriv will ** do, since the function value triggers step-halving). */ g = SMALL; dg = -zz/sigma; ddg = -1/sigma; dsig =0; ddsig=0; dsg=0; } else { g = log(funs[2][person]) - log(sigma); temp1 = funs[3][person]/sigma; temp2 = funs[4][person]*sig2; dg = -temp1; dsig= -(sz*temp1 +1); ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(1- sz*temp1); ddsig = sz*sz*temp2 + sz*temp1*(1- sz*temp1); } break; case 0: /* right censored */ if (funs[1][person] <=0) { g = SMALL; dg = zz/sigma; ddg =0; dsig =0; ddsig=0; dsg=0; } else { g = log(funs[1][person]); temp1 = -funs[2][person]/(funs[1][person]*sigma); temp2 = -funs[3][person]*funs[2][person]*sig2/ funs[1][person]; dg = -temp1; dsig= -sz * temp1; ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(1+dsig); ddsig = sz*sz*temp2 - dsig*(1+dsig); } break; case 2: /* left censored */ if (funs[2][person] <=0) { /* off the probability scale -- avoid log(0) */ g = SMALL; dg = -zz/sigma; dsig =0; ddsig=0; dsg=0; ddg =0; } else { g = log(funs[0][person]); temp1 = funs[2][person]/(funs[0][person]*sigma); temp2 = funs[3][person]*funs[2][person]*sig2/ funs[0][person]; dg= -temp1; dsig= -sz * temp1; ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(1+dsig); ddsig = sz*sz*temp2 - dsig*(1+dsig); } break; case 3: /* interval censored */ zu = z[icount]; /*stop roundoff in tails*/ if (zz>0) temp = funs[1][person] - funs[1][icount]; else temp = funs[0][icount] - funs[0][person]; if (temp <=0) { /* off the probability scale -- avoid log(0) */ g = SMALL; dg = 1; ddg =0; dsig =0; ddsig=0; dsg=0; } else { funs[3][icount] *= funs[2][icount]; /*f', not f'/f */ funs[3][person] *= funs[2][person]; g = log(temp); dg = -(funs[2][icount] -funs[2][person])/(temp*sigma); ddg = (funs[3][icount] -funs[3][person])*sig2/temp - dg*dg; dsig = (zz*funs[2][person] - zu*funs[2][icount])/temp; ddsig= (zu*zu*funs[3][icount] - zz*zz*funs[3][person]) /temp - dsig*(1+dsig); dsg = (zu*funs[3][icount] - zz*funs[3][person])/ (temp*sigma) - dg *(1+dsig); } icount++; break; } loglik += g * wt[person]; /* ** Now the derivs wrt loglik */ if (whichcase==1) continue; /*only needed the loglik */ w = wt[person]; if (nf>0) { fgrp = frail[person] -1; u[fgrp] += dg * w; fdiag[fgrp] -= ddg * w; jdiag[fgrp] += dg*dg *w; } for (i=0; i0) { imat[i][fgrp] -= covar[i][person] * ddg * w; JJ [i][fgrp] += temp * dg; } } if (nstrat!=0) { /* need derivative wrt log sigma */ k = strata+nvar; u[k+nf] += w* dsig; for (i=0; i0) { imat[k][fgrp] -= dsg * w; JJ [k][fgrp] += dsig *dg *w; } } } UNPROTECT(1); /* release the memory pointed to by funs[] */ return(loglik); } survival/src/coxexact.c0000644000175100001440000003516612545056257014742 0ustar hornikusers/* Automatically generated from all.nw using noweb */ #include #include "survS.h" #include "survproto.h" #include double coxd0(int d, int n, double *score, double *dmat, int dmax) { double *dn; if (d==0) return(1.0); dn = dmat + (n-1)*dmax + d -1; /* pointer to dmat[d,n] */ if (*dn ==0) { /* still to be computed */ *dn = score[n-1]* coxd0(d-1, n-1, score, dmat, dmax); if (d1) d1[indx] += score[n-1]* coxd1(d-1, n-1, score, dmat, d1, covar, dmax); } return(d1[indx]); } double coxd2(int d, int n, double *score, double *dmat, double *d1j, double *d1k, double *d2, double *covarj, double *covark, int dmax) { int indx; indx = (n-1)*dmax + d -1; /*index to the current array member d1[d,n]*/ if (d2[indx] ==0) { /*still to be computed */ d2[indx] = coxd0(d-1, n-1, score, dmat, dmax)*score[n-1] * covarj[n-1]* covark[n-1]; if (d1) d2[indx] += score[n-1] * ( coxd2(d-1, n-1, score, dmat, d1j, d1k, d2, covarj, covark, dmax) + covarj[n-1] * coxd1(d-1, n-1, score, dmat, d1k, covark, dmax) + covark[n-1] * coxd1(d-1, n-1, score, dmat, d1j, covarj, dmax)); } return(d2[indx]); } SEXP coxexact(SEXP maxiter2, SEXP y2, SEXP covar2, SEXP offset2, SEXP strata2, SEXP ibeta, SEXP eps2, SEXP toler2) { int i,j,k; int iter; double **covar, **imat; /*ragged arrays */ double *time, *status; /* input data */ double *offset; int *strata; int sstart; /* starting obs of current strata */ double *score; double *oldbeta; double zbeta; double newlk=0; double temp; int halving; /*are we doing step halving at the moment? */ int nrisk; /* number of subjects in the current risk set */ int dsize, /* memory needed for one coxc0, coxc1, or coxd2 array */ dmemtot, /* amount needed for all arrays */ ndeath; /* number of deaths at the current time point */ double maxdeath; /* max tied deaths within a strata */ double dtime; /* time value under current examiniation */ double *dmem0, **dmem1, *dmem2; /* pointers to memory */ double *dtemp; /* used for zeroing the memory */ double *d1; /* current first derivatives from coxd1 */ double d0; /* global sum from coxc0 */ /* copies of scalar input arguments */ int nused, nvar, maxiter; double eps, toler; /* returned objects */ SEXP imat2, beta2, u2, loglik2; double *beta, *u, *loglik; SEXP rlist, rlistnames; int nprotect; /* number of protect calls I have issued */ nused = LENGTH(offset2); nvar = ncols(covar2); maxiter = asInteger(maxiter2); eps = asReal(eps2); /* convergence criteria */ toler = asReal(toler2); /* tolerance for cholesky */ /* ** Set up the ragged array pointer to the X matrix, ** and pointers to time and status */ covar= dmatrix(REAL(covar2), nused, nvar); time = REAL(y2); status = time +nused; strata = INTEGER(PROTECT(duplicate(strata2))); offset = REAL(offset2); /* temporary vectors */ score = (double *) R_alloc(nused+nvar, sizeof(double)); oldbeta = score + nused; /* ** create output variables */ PROTECT(beta2 = duplicate(ibeta)); beta = REAL(beta2); PROTECT(u2 = allocVector(REALSXP, nvar)); u = REAL(u2); PROTECT(imat2 = allocVector(REALSXP, nvar*nvar)); imat = dmatrix(REAL(imat2), nvar, nvar); PROTECT(loglik2 = allocVector(REALSXP, 5)); /* loglik, sctest, flag,maxiter*/ loglik = REAL(loglik2); nprotect = 5; strata[0] =1; /* in case the parent forgot */ temp = 0; /* temp variable for dsize */ maxdeath =0; j=0; /* start of the strata */ for (i=0; i0) { /* If maxdeath <2 leave the strata alone at it's current value of 1 */ if (maxdeath >1) strata[j] = maxdeath; j = i; if (maxdeath*nrisk > temp) temp = maxdeath*nrisk; } maxdeath =0; /* max tied deaths at any time in this strata */ nrisk=0; ndeath =0; } dtime = time[i]; ndeath =0; /*number tied here */ while (time[i] ==dtime) { nrisk++; ndeath += status[i]; i++; if (i>=nused || strata[i] >0) break; /*tied deaths don't cross strata */ } if (ndeath > maxdeath) maxdeath=ndeath; } if (maxdeath*nrisk > temp) temp = maxdeath*nrisk; if (maxdeath >1) strata[j] = maxdeath; /* Now allocate memory for the scratch arrays Each per-variable slice is of size dsize */ dsize = temp; temp = temp * ((nvar*(nvar+1))/2 + nvar + 1); dmemtot = dsize * ((nvar*(nvar+1))/2 + nvar + 1); if (temp != dmemtot) { /* the subscripts will overflow */ error("(number at risk) * (number tied deaths) is too large"); } dmem0 = (double *) R_alloc(dmemtot, sizeof(double)); /*pointer to memory */ dmem1 = (double **) R_alloc(nvar, sizeof(double*)); dmem1[0] = dmem0 + dsize; /*points to the first derivative memory */ for (i=1; i0) { /* first obs of a new strata */ maxdeath= strata[i]; dtemp = dmem0; for (j=0; j=nused || strata[i] >0) break; } /* We have added up over the death time, now process it */ if (ndeath >0) { /* Add to the loglik */ d0 = coxd0(ndeath, nrisk, score+sstart, dmem0, maxdeath); R_CheckUserInterrupt(); newlk -= log(d0); dmem2 = dmem0 + (nvar+1)*dsize; /*start for the second deriv memory */ for (j=0; j 3) R_CheckUserInterrupt(); u[j] -= d1[j]; for (k=0; k<= j; k++) { /* second derivative*/ temp = coxd2(ndeath, nrisk, score+sstart, dmem0, dmem1[j], dmem1[k], dmem2, covar[j] + sstart, covar[k] + sstart, maxdeath); if (ndeath > 5) R_CheckUserInterrupt(); imat[k][j] += temp/d0 - d1[j]*d1[k]; dmem2 += dsize; } } } } loglik[0] = newlk; /* save the loglik for iteration zero */ loglik[1] = newlk; /* and it is our current best guess */ /* ** update the betas and compute the score test */ for (i=0; i0) { /* first obs of a new strata */ maxdeath= strata[i]; dtemp = dmem0; for (j=0; j=nused || strata[i] >0) break; } /* We have added up over the death time, now process it */ if (ndeath >0) { /* Add to the loglik */ d0 = coxd0(ndeath, nrisk, score+sstart, dmem0, maxdeath); R_CheckUserInterrupt(); newlk -= log(d0); dmem2 = dmem0 + (nvar+1)*dsize; /*start for the second deriv memory */ for (j=0; j 3) R_CheckUserInterrupt(); u[j] -= d1[j]; for (k=0; k<= j; k++) { /* second derivative*/ temp = coxd2(ndeath, nrisk, score+sstart, dmem0, dmem1[j], dmem1[k], dmem2, covar[j] + sstart, covar[k] + sstart, maxdeath); if (ndeath > 5) R_CheckUserInterrupt(); imat[k][j] += temp/d0 - d1[j]*d1[k]; dmem2 += dsize; } } } } /* am I done? ** update the betas and test for convergence */ loglik[3] = cholesky2(imat, nvar, toler); if (fabs(1-(loglik[1]/newlk))<= eps && halving==0) { /* all done */ loglik[1] = newlk; loglik[4] = iter; chinv2(imat, nvar); for (i=1; i #include "survS.h" #include "survproto.h" void survdiff2(Sint *nn, Sint *nngroup, Sint *nstrat, double *rho, double *time, Sint *status, Sint *group, Sint *strata, double *obs, double *exp, double *var, double *risk, double *kaplan) { register int i,j,k; int kk; int n, ngroup, ntot; int istart, koff; double km, nrisk, wt, tmp; double deaths; ntot = *nn; ngroup = *nngroup; istart=0; koff=0; for (i=0; i< ngroup*ngroup; i++) var[i]=0; for (i=0; i< *nstrat*ngroup; i++) { obs[i]=0; exp[i]=0; } while (istart < ntot) { /* loop over the strata */ for (i=0; i=istart; i--) { if (*rho ==0) wt=1; else wt= pow(kaplan[i], *rho); deaths = 0; for (j=i; j>=istart && time[j]==time[i]; j--) { k = group[j]-1; deaths += status[j]; risk[k] += 1; obs[k + koff] += status[j] *wt; } i= j +1; nrisk = n-i; if (deaths>0) { /* a death time */ for (k=0; k=2 special handling for US "calendar year" ** edims[edim] the number of rows, columns, etc ** ecut[ ] the starting points for each non-factor dimension, ** strung together. ** expect the actual table of expected rates ** edata[edim, n] the subject data-- where each indexes into the ** expected table, at time 0. ** ** output table's description ** odim number of dimensions ** ofac[odim] 1=is a factor, 0=continuous (time based) ** odims[odim] the number of rows, columns, etc ** ocut[] for each non-factor dimension, the odim[i]+1 cutpoints ** that define the intervals; concatonated. ** odata[odim, n] the subject data-- where each indexes into the ** expected table, at time 0. ** ** Output: ** pyears output table of person years ** pn number of observations that contribute to each cell ** pcount number of events ** pexpect expected number of events ** offtable total person years that did not fall into the output table ** ** Scratch -- allocated on the fly ** scratch[edim + odim] */ #include #include "survS.h" #include "survproto.h" /* names that begin with "s" will be re-declared in the main body */ void pyears1(Sint *sn, Sint *sny, Sint *sdoevent, double *sy, double *weight, Sint *sedim, Sint *efac, Sint *edims, double *secut, double *expect, double *sedata, Sint *sodim, Sint *ofac, Sint *odims, double *socut, Sint *smethod, double *sodata, double *pyears, double *pn, double *pcount, double *pexpect, double *offtable) { int i,j; int n, ny, doevent, method, edim, odim; double *start, *stop, *event, **ecut, **ocut, **edata, **odata; double *data, *data2; double timeleft, thiscell, etime, et2; int index, indx, indx2; double lwt; /*this variable is returned by pystep, and controls the "on the fly" linear interpolation done for the calandar year dimension of rate tables */ int dostart; double hazard, cumhaz; double temp, lambda; double eps; /* protection against accumulated round off */ n = *sn; ny= *sny; doevent = *sdoevent; method = *smethod; edim = *sedim; odim = *sodim; start = sy; if (ny==3 || (ny==2 && doevent==0)) { stop = sy +n; dostart =1; } else { stop = sy; dostart =0; } event = stop +n; edata = dmatrix(sedata, n, edim); odata = dmatrix(sodata, n, odim); i=edim + odim; data = (double *) ALLOC(i, sizeof(double)); data2 = data + odim; /* ** ecut and ocut will be ragged arrays */ ecut = (double **)ALLOC(edim, sizeof(double *)); for (i=0; i1) secut += 1 + (efac[i]-1)*edims[i]; } ocut = (double **)ALLOC(odim, sizeof(double *)); for (i=0; i0]) * 1e-8 ** The events are counted in the last cell to which person years are ** added in the while() loop below. We don't want to "spill over" into ** a next (incorrect) cell due to accumulated round off, in the case ** that a subjects fu time exactly matches one of the cell boundaries. */ eps =0; /* guard against the rare case that all(time==0) */ for (i=0; i0) { eps = timeleft; break; } } for (; i0 && timeleft < eps) eps = timeleft; } eps *= 1e-8; *offtable =0; for (i=0; i eps) { thiscell = pystep(odim, &index, &indx2, &lwt, data, ofac, odims, ocut, timeleft, 0); if (index >=0) { pyears[index] += thiscell * weight[i]; pn[index] += 1; /* expected calc */ etime = thiscell; hazard=0; temp =0; while (etime >0) { /* ** The hazard or survival curve (temp) calculated within ** this loop don't depend on the case weight --- the ** whole loop is only for one person, and hazard is a ** function of time alone. Once computed, however, the ** total hazard added into the expected table ** is weighted. */ et2 = pystep(edim, &indx, &indx2, &lwt, data2, efac, edims, ecut, etime, 1); if (lwt <1) lambda = (lwt*expect[indx] + (1-lwt)*expect[indx2]); else lambda = expect[indx]; if (method==0) temp += exp(-hazard)*(1-exp(-lambda*et2))/ lambda; hazard += lambda * et2; for (j=0; j=0 && doevent) pcount[index] += event[i] * weight[i]; } } survival/src/coxmart2.c0000644000175100001440000000370512545056257014655 0ustar hornikusers/* ** Compute the martingale residual for a Cox model. ** This routine does the same work as coxmart, except ** it expects data in inverse time order ** only does the Breslow method ** exists for the sake of coxexact.fit ** ** Input ** n number of subjects ** time vector of times ** status vector of status values ** score the vector of subject scores, i.e., exp(beta*z) ** strata is =1 for the first obs of a strata ** wt case weights ** Output ** the residual for each subject */ #include "survS.h" #include "survproto.h" void coxmart2(Sint *sn, double *time, Sint *status, Sint * strata, double *score, double *wt, double *resid) { int i,j; int n; double deaths, denom; double expected, hazard; n = *sn; /* ** Accumulate the weighted score in reverse time order (data order) ** Temporarily save the resulting hazard in the residual vector */ denom =0; for (i=0; i=0; i--) { expected += resid[i]; resid[i] = status[i] - score[i]*expected; if (strata[i] ==1) expected=0; /* last obs of a strata */ } } survival/src/coxfit5.c0000644000175100001440000004304012545056257014473 0ustar hornikusers/* A reentrant version of the Coxfit program, for random effects modeling ** with reasonable efficiency (I hope). The important arrays are saved ** from call to call so as to speed up the process. The x-matrix itself ** is the most important of these. ** ** coxfit5_a: Entry and intial iteration step for beta=initial, theta=0 ** (no frailty) ** Most of the same arguments as coxfit2. ** Allocate and save arrays in static locations. ** coxfit5_b: Iterate to convergence given an initial value. ** coxfit5_c: Compute residuals and release the saved memory. ** ** McGilchrist's method for frailty with a fixed theta, but for ** space savings I assume that many elements of imat are zero ** ** the input parameters are ** ** maxiter :number of iterations ** nused :number of people ** nvar :number of covariates ** y[2,n] :row 1: time of event or censoring for person i ** :row 2: status for the ith person 1=dead , 0=censored ** covar(nv,n) :covariates for person i. ** Note that S sends this in column major order. ** strata(nstrat):sizes of the strata, cumulative ** sort : sort order for the obs, last to first within strata ** offset(n) :offset for the linear predictor ** weights(n) :case weights ** eps :tolerance for convergence. Iteration continues until ** the percent change in loglikelihood is <= eps. ** tolerch :tolerance for the Cholesky routines ** method : Method 0=Breslow, 1=Efron ** ptype : 1 or 3 -- there is a sparse term ** : 2 or 3 -- there is a non-sparse term in the model ** nfrail : number of frailty groups (sparse terms), 0 if there are ** none ** frail : a vector containing the frailty groups ** fbeta : initial frailty estimates ** pdiag : if 0, then for the non-sparse terms only the diagonal ** of the variance matrix is penalized, otherwise the ** full matrix is used. ** ** returned parameters ** means(nv) : vector of column means of X ** beta(nv) : the vector of answers (at start contains initial est) ** u(nv) : score vector ** imat(nv,nv) : the variance matrix at beta=final ** if flag<0, imat is undefined upon return ** loglik :loglik at beta=final ** flag :success flag 1000 did not converge ** 1 to nvar: rank of the solution ** maxiter :actual number of iterations used ** fbeta(nfrail): fitted frailty values ** fdiag(nfrail + nvar): diagonal of cholesky of the full inverse ** jmat : inverse of the cholesky ** imat : cholesky of the information matrix ** expect : contains the "expected" for each subject ** ** work arrays ** mark(n) ** wtave(n) ** score(n) ** a(nvar+ nfrail), a2(nvar+nfrail) ** cmat(nvar,nvar+nfrail) ragged array ** cmat2(nvar,nvar+nfrail) ** fdiag the diagonal of the sparse information ** oldbeta(nvar + nfrail) always contains the "last iteration" ** ** the work arrays are passed as a single ** vector of storage, and then broken out. ** ** calls functions: cholesky3, chsolve3, chinv2 ** ** the data must be sorted by ascending time within strata */ #include #include #include "survS.h" #include "survproto.h" static double **covar, **cmat, **cmat2; static double *mark, *wtave; static double *a, *oldbeta, *a2; static double *offset, *weights; static int *status, *frail, *sort; static double *score, *ttime; /* Hp-UX really doesn't like "time" as a var */ static double *tmean; static int ptype, pdiag; static double *ipen, *upen, logpen; static Sint *zflag; static double **cmatrix(double *, int, int); void coxfit5_a(Sint *nusedx, Sint *nvarx, double *yy, double *covar2, double *offset2, double *weights2, Sint *strata, Sint *sorted, double *means, double *beta, double *u, double *loglik, Sint *methodx, Sint *ptype2, Sint *pdiag2, Sint *nfrail, Sint *frail2, void *fexpr1, void *fexpr2, void *rho) { int i,j,k, p, istrat; int ii; int nused, nvar; int nf, nvar2; double denom, zbeta, risk; double temp, temp2; double ndead; double d2, efron_wt; double method; nused = *nusedx; nvar = *nvarx; nf= *nfrail; method= *methodx; nvar2 = nvar + nf; ptype = *ptype2; pdiag = *pdiag2; /* ** Allocate storage for the arrays and vectors ** Since they will be used later, sizes are based on what will be ** needed with the frailty terms. */ if (nvar >0) { covar= cmatrix(covar2, nused, nvar); cmat = cmatrix(0, nvar2, nvar+1); cmat2= cmatrix(0, nvar2, nvar+1); } a = Calloc(4*nvar2 + 6*nused, double); oldbeta = a + nvar2; a2 = oldbeta + nvar2; mark = a2 + nvar2; wtave= mark + nused; weights = wtave+ nused; offset = weights + nused; score = offset + nused; tmean = score + nused; ttime = tmean + nvar2; status = Calloc(2*nused, int); sort = status + nused; for (i=0; i nvar) i=nf; else i=nvar; if (nf > nvar*nvar) j=nf; else j=nvar*nvar; if (pdiag==0) upen = Calloc(2*i, double); else upen = Calloc(i+j, double); ipen = upen + i; if (ptype>1) zflag = Calloc(nvar, Sint); else zflag = Calloc(2, Sint); if (nf>0) { frail = Calloc(nused, int); for (i=0; i0) { /* once per unique death time */ /* ** Trick: when 'method==0' then temp=0, giving Breslow's method */ ndead = mark[p]; for (k=0; k0) { imat = dmatrix(imat2, nvar2, nvar); jmat = dmatrix(jmat2, nvar2, nvar); } else { imat = 0; /*never used, but passed as dummy to chol */ jmat = 0; } for (i=0; i0) { fgrp = frail[p] -1; zbeta = offset[p] + fbeta[fgrp]; } else zbeta = offset[p]; for (i=0; i0) a[fgrp] += risk; for (i=0; i0) cmat[i][fgrp] += risk*covar[i][p]; for (j=0; j<=i; j++) cmat[i][j+nf] += risk*covar[i][p]*covar[j][p]; } if (status[p]==1) { efron_wt += risk; newlk += weights[p] *zbeta; if (nf>0) { u[fgrp] += weights[p]; a2[fgrp] += risk; } for (i=0; i0) cmat2[i][fgrp] += risk*covar[i][p]; for (j=0; j<=i; j++) cmat2[i][j+nf] += risk*covar[i][p]*covar[j][p]; } } if (mark[p] >0) { /* once per unique death time */ ndead = mark[p]; for (k=0; k0 && newlk < *loglik) { /*it is not converging ! */ halving =1; for (i=0; i0) { /* ** Compute the size of the hazard jump at this point, with the ** total jump saved (temporarily) in "expect", and the Efron ** amount in "weights". It applies to deaths at this point. */ ndead = mark[p]; temp2 = 0; efron_wt =0; for (j=0; j=0; ) { p = sort[ip]; if (status[p] >0) { ndead = mark[p]; temp = expect[p]; hazard2 =weights[p]; for (j=0; j 0) { cmatrix_free(cmat2); cmatrix_free(cmat); cmatrix_free(covar); } } survival/src/agfit4.c0000644000175100001440000005435712545056257014305 0ustar hornikusers/* Automatically generated from all.nw using noweb */ #include #include "survS.h" #include "survproto.h" SEXP agfit4(SEXP surv2, SEXP covar2, SEXP strata2, SEXP weights2, SEXP offset2, SEXP ibeta2, SEXP sort12, SEXP sort22, SEXP method2, SEXP maxiter2, SEXP eps2, SEXP tolerance2, SEXP doscale2) { int i,j,k,person; int indx2, istrat, p; int ksave, nrisk, ndeath; int nused, nvar; double **covar, **cmat, **imat; /*ragged array versions*/ double *a, *oldbeta, *maxbeta; double *scale; double *a2, **cmat2; double *eta; double denom, zbeta, risk; double time; double temp, temp2; double newlk =0; int halving; /*are we doing step halving at the moment? */ double tol_chol, eps; double meanwt; int itemp, deaths; double efron_wt, d2, meaneta; /* inputs */ double *start, *stop, *event; double *weights, *offset; int *sort1, *sort2, maxiter; int *strata; double method; /* saving this as double forces some double arithmetic */ int doscale; /* returned objects */ SEXP imat2, means2, beta2, u2, loglik2; double *beta, *u, *loglik, *means; SEXP sctest2, flag2, iter2; double *sctest; int *flag, *iter; SEXP rlist; static const char *outnames[]={"coef", "u", "imat", "loglik", "means", "sctest", "flag", "iter", ""}; int nprotect; /* number of protect calls I have issued */ /* get sizes and constants */ nused = nrows(covar2); nvar = ncols(covar2); method= asInteger(method2); eps = asReal(eps2); tol_chol = asReal(tolerance2); maxiter = asInteger(maxiter2); doscale = asInteger(doscale2); /* input arguments */ start = REAL(surv2); stop = start + nused; event = stop + nused; weights = REAL(weights2); offset = REAL(offset2); sort1 = INTEGER(sort12); sort2 = INTEGER(sort22); strata = INTEGER(strata2); /* ** scratch space ** nvar: a, a2, newbeta, maxbeta, scale ** nvar*nvar: cmat, cmat2 ** n: eta */ eta = (double *) R_alloc(nused + 5*nvar + 2*nvar*nvar, sizeof(double)); a = eta + nused; a2 = a +nvar; maxbeta = a2 + nvar; scale = maxbeta + nvar; oldbeta = scale + nvar; /* ** Set up the ragged arrays ** covar2 might not need to be duplicated, even though ** we are going to modify it, due to the way this routine was ** was called. In this case NAMED(covar2) will =0 */ PROTECT(imat2 = allocVector(REALSXP, nvar*nvar)); nprotect =1; if (NAMED(covar2)>0) { PROTECT(covar2 = duplicate(covar2)); nprotect++; } covar= dmatrix(REAL(covar2), nused, nvar); imat = dmatrix(REAL(imat2), nvar, nvar); cmat = dmatrix(oldbeta+ nvar, nvar, nvar); cmat2= dmatrix(oldbeta+ nvar + nvar*nvar, nvar, nvar); /* ** create the output structures */ PROTECT(rlist = mkNamed(VECSXP, outnames)); nprotect++; beta2 = SET_VECTOR_ELT(rlist, 0, duplicate(ibeta2)); beta = REAL(beta2); u2 = SET_VECTOR_ELT(rlist, 1, allocVector(REALSXP, nvar)); u = REAL(u2); SET_VECTOR_ELT(rlist, 2, imat2); loglik2 = SET_VECTOR_ELT(rlist, 3, allocVector(REALSXP, 2)); loglik = REAL(loglik2); means2 = SET_VECTOR_ELT(rlist, 4, allocVector(REALSXP, nvar)); means = REAL(means2); sctest2 = SET_VECTOR_ELT(rlist, 5, allocVector(REALSXP, 1)); sctest = REAL(sctest2); flag2 = SET_VECTOR_ELT(rlist, 6, allocVector(INTSXP, 1)); flag = INTEGER(flag2); iter2 = SET_VECTOR_ELT(rlist, 7, allocVector(INTSXP, 1)); iter = INTEGER(iter2); /* ** Subtract the mean from each covar, as this makes the variance ** computation much more stable */ temp2 =0; for (i=0; i0) temp = temp2/temp; else temp = 1.0; /* rare case of a constant covariate */ scale[i] = temp; for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } else { /* scaling is only turned off during debugging still, cover the case */ for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } } if (doscale ==1) { for (i=0; i population ** of the earth, any beta*x over 20 is a silly relative risk for a Cox ** model, however. ** We want to cut off huge values, but not take action very often since ** doing so can mess up the iteration in general. ** One of the case-cohort papers suggests using anoffset of -100 to ** indicate "no risk", meaning that x*beta values of 50-100 can occur ** in "ok" data sets. Compromise. */ for (i=0; i=time, ** and are thus potential members of the risk set. If 'indx2' =9, ** that means that 9 subjects have start >=time and thus are NOT part ** of the risk set. (stop > start for each subject guarrantees that ** the 9 are a subset of the 27). ** Basic algorithm: move 'person' forward, adding the new subject into ** the risk set. If this is a new, unique death time, take selected ** old obs out of the sums, add in obs tied at this time, then ** add terms to the loglik, etc. */ istrat=0; indx2 =0; denom =0; meaneta =0; nrisk =0; newlk =0; for (person=0; person (nrisk *110)) { meaneta = meaneta/nrisk; for (i=0; i=time, ** and are thus potential members of the risk set. If 'indx2' =9, ** that means that 9 subjects have start >=time and thus are NOT part ** of the risk set. (stop > start for each subject guarrantees that ** the 9 are a subset of the 27). ** Basic algorithm: move 'person' forward, adding the new subject into ** the risk set. If this is a new, unique death time, take selected ** old obs out of the sums, add in obs tied at this time, then ** add terms to the loglik, etc. */ istrat=0; indx2 =0; denom =0; meaneta =0; nrisk =0; newlk =0; for (person=0; person (nrisk *110)) { meaneta = meaneta/nrisk; for (i=0; i maxbeta[i]) beta[i] = maxbeta[i]; else if (beta[i] < -maxbeta[i]) beta[i] = -maxbeta[i]; } } } R_CheckUserInterrupt(); /* be polite -- did the user hit cntrl-C? */ } /*return for another iteration */ loglik[1] = newlk; chinv2(imat, nvar); for (i=0; i1: estimate multiple scales (strata) ** strat - if nstrat>0, contains the strata number for each subject ** eps - tolerance for convergence. Iteration continues until the ** relative change in the deviance is <= eps. ** tol_chol- tolerance for Cholesky decomposition ** dist - 1=extreme value, 2=logistic, 3=gaussian, 4=callback ** debug - >0 causes tracing information. Can be removed ** expr - for callback, the expression to be evaluated ** rho - for callback, the environment (R) or frame (Splus) in which ** to do the evaluation. ** Output ** beta - the final coef vector ** iter - the number of iterations consumed ** imat - the information matrix ** loglik - the final log-liklihood ** flag - success flag 0 =ok ** -1= did not converge ** u - the score vector ** ** Work arrays ** newbeta(nvar)- always contains the "next iteration" ** JJ = the approx variance matrix J'J, guarranteed non-singular */ #include "survS.h" #include "survproto.h" SEXP survreg6(SEXP maxiter2, SEXP nvarx, SEXP y, SEXP ny2, SEXP covar2, SEXP wtx, SEXP offset2, SEXP beta2, SEXP nstratx, SEXP stratax, SEXP epsx, SEXP tolx, SEXP dist, SEXP dexpr, SEXP rho) { int i,j; int n, maxiter, ny; double *newbeta; int halving, iter; double newlk; double *loglik, eps, tol_chol; double *beta; Sint *flag; SEXP out_beta; int nvar, nvar2, nstrat; double **covar; Sint *strat ; double *time2, *time1, *status; double *offset; double **imat, **JJ; double *u, *wt, *usave; double (*dolik)(); /* will be pointed to survregc1 or survregc2 */ SEXP z; double *zptr = NULL; SEXP out_iter, out_loglik, out_imat, out_flag; SEXP out_u; SEXP rlist, rlistnames; Sint *iter2; int nprotect; /* ** The only input arg that is overwritten is beta */ out_beta = PROTECT(duplicate(beta2)); beta = REAL(out_beta); maxiter = asInteger(maxiter2); n = LENGTH(wtx); ny = asInteger(ny2); nvar = asInteger(nvarx); offset = REAL(offset2); nstrat = asInteger(nstratx); strat = INTEGER(stratax); wt = REAL(wtx); eps = asReal(epsx); tol_chol= asReal(tolx); covar = dmatrix(REAL(covar2), n, nvar); /* ** nvar = # of "real" x variables, for iteration ** nvar2= # of parameters to maximize = nvar + nstrat ** nstrat= # of strata, where 0== fixed sigma */ nvar2 = nvar + nstrat; /* number of coefficients */ /* ** Create the output variables */ PROTECT(out_imat = allocVector(REALSXP, nvar2*nvar2)); imat = dmatrix(REAL(out_imat), nvar2, nvar2); PROTECT(out_iter = allocVector(INTSXP, 1)); iter2 = INTEGER(out_iter); PROTECT(out_loglik = allocVector(REALSXP, 1)); loglik = REAL(out_loglik); PROTECT(out_flag = allocVector(INTSXP, 1)); flag = INTEGER(out_flag); PROTECT(out_u = allocVector(REALSXP, nvar2)); usave = REAL(out_u); nprotect = 6; /* Create scratch variables ** u = working version of score vector, overwritten with u H-inv during ** Newton steps ** usave = a copy of u, after each Newton step. Returned to the S ** parent routine, and also used to "backtrack" when we need to fail ** over to a Fisher step after NR + halving didn't work */ newbeta = (double *) Calloc(LENGTH(beta2) + nvar2 + nvar2*nvar2, double); u = newbeta + length(beta2); JJ = dmatrix(u +nvar2, nvar2, nvar2); /* ** fixed scale parameters were tacked onto the end of beta at input ** copy them to to the end of newbeta as well (survregc1/c2 expects em) */ for (i=nvar; i0 when in the midst of "step halving" */ newlk = (*dolik)(n, nvar, nstrat, 0, newbeta,asInteger(dist), strat, offset, time1, time2, status, wt, covar, imat, JJ, u, dexpr, rho, zptr, 0, NULL, NULL, NULL); for (i=0; i10, in which ** case step halving isn't quite enough. Make sure the new ** try differs from the last good one by no more than 1/3 ** approx log(3) = 1.1 ** Step halving isn't enough of a "back away" when a ** log(sigma) goes from 0.5 to -3, or has become singular. */ if (halving==1) { /* only the first time */ for (i=0; i 1.1) newbeta[nvar+i] = beta[nvar+i] - 1.1; } } newlk = (*dolik)(n, nvar, nstrat, 1, newbeta,asInteger(dist), strat, offset, time1, time2, status, wt, covar, imat, JJ, u, dexpr, rho, zptr, 0, NULL, NULL, NULL); } } else { /* take a standard NR step */ halving=0; *loglik = newlk; *flag = cholesky3(imat, nvar2, 0, NULL, tol_chol); if (*flag < 0) { i = cholesky3(JJ, nvar2, 0, NULL, tol_chol); chsolve2(JJ, nvar2, u); } else chsolve2(imat,nvar2,u); for (i=0; i #include "survS.h" #include "survproto.h" void agscore(Sint *nx, Sint *nvarx, double *y, double *covar2, Sint *strata, double *score, double *weights, Sint *method, double *resid2, double *a) { int i,k; int n, nvar; int person; double denom, time; double *a2, *mean; double e_denom; double risk; double hazard, meanwt; double deaths, downwt; int dd; double *start, *stop, *event; double **covar, **resid; double temp1, temp2, d2; double *mh1, *mh2, *mh3; n = *nx; nvar = *nvarx; start =y; stop = y+n; event = y+(n+n); /* ** Set up the ragged arrays */ covar= dmatrix(covar2, n, nvar); resid = dmatrix(resid2, n, nvar); a2 = a+nvar; mean= a2 + nvar; mh1 = mean + nvar; mh2 = mh1 + nvar; mh3 = mh2 + nvar; for (person=0; person0) { /* ** This happens when a penalty is infinite. (Which itself ** is often how a penalty routine signals that it was given ** an illegal value for a parameter). In this case the ** updated value of beta will have already been set to 0 ** via cptr1 above, which is of course the correct solution, ** but the U and H use for the parent routine's Newton-Raphson ** step are infinite as well. We force the u and ** hmat matrices to dummy values that will cause no update ** (none needed) and more importantly no infinte/infinite ** arithmetic errors: u=0 and H = identity. (Only the ** relevant columns of each, of course). */ for (i=0; i 1) { /* ** Get the penalty for the dense part of the matrix ** Note that penalties never apply to the variance terms, ** which means that indices go to nvar, not nvar2 */ for (i=0; i=0; ) { ndeath =0; if (status[i]==1) { /* process all tied deaths at this point */ for (j=i; j>=0 && status[j]==1 && time[j]==time[i]; j--) { ndeath += wt[j]; index = indx[j]; for (k=i; k>j; k--) count[3] += wt[j]*wt[k]; /* tied on time */ count[2] += wt[j] * nwt[index]; /* tied on x */ child = (2*index) +1; /* left child */ if (child < ntree) count[0] += wt[j] * twt[child]; /*left children */ child++; if (child < ntree) count[1] += wt[j] * twt[child]; /*right children */ while (index >0) { /* walk up the tree */ parent = (index-1)/2; if (index & 1) /* I am the left child */ count[1] += wt[j] * (twt[parent] - twt[index]); else count[0] += wt[j] * (twt[parent] - twt[index]); index = parent; } } } else j = i-1; /* Add the weights for these obs into the tree and update variance*/ for (; i>j; i--) { wsum1=0; oldmean = twt[0]/2; index = indx[i]; nwt[index] += wt[i]; twt[index] += wt[i]; wsum2 = nwt[index]; child = 2*index +1; /* left child */ if (child < ntree) wsum1 += twt[child]; while (index >0) { parent = (index-1)/2; twt[parent] += wt[i]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); /* sum of weights above */ lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean - oldmean); vss += wsum3*(newmean+ oldmean+ wt[i]- 2*umean) *(oldmean-newmean); vss += wt[i]* (myrank -newmean)*(myrank -newmean); } count[4] += ndeath * vss/twt[0]; } UNPROTECT(1); return(count2); } SEXP concordance2(SEXP y, SEXP wt2, SEXP indx2, SEXP ntree2, SEXP sortstop, SEXP sortstart) { int i, j, k, index; int child, parent; int n, ntree; int istart, iptr, jptr; double *time1, *time2, *status, dtime; double *twt, *nwt, *count; int *sort1, *sort2; double vss, myrank; double wsum1, wsum2, wsum3; /*sum of wts below, tied, above*/ double lmean, umean, oldmean, newmean; double ndeath; SEXP count2; double *wt; int *indx; n = nrows(y); ntree = asInteger(ntree2); wt = REAL(wt2); indx = INTEGER(indx2); sort2 = INTEGER(sortstop); sort1 = INTEGER(sortstart); time1 = REAL(y); time2 = time1 + n; status= time2 + n; PROTECT(count2 = allocVector(REALSXP, 5)); count = REAL(count2); twt = (double *) R_alloc(2*ntree, sizeof(double)); nwt = twt + ntree; for (i=0; i< 2*ntree; i++) twt[i] =0.0; for (i=0; i<5; i++) count[i]=0.0; vss =0; istart = 0; /* where we are with start times */ for (i=0; i= dtime; istart++) { wsum1 =0; oldmean = twt[0]/2; jptr = sort1[istart]; index = indx[jptr]; nwt[index] -= wt[jptr]; twt[index] -= wt[jptr]; wsum2 = nwt[index]; child = 2*index +1; /* left child */ if (child < ntree) wsum1 += twt[child]; while (index >0) { parent = (index-1)/2; twt[parent] -= wt[jptr]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean-oldmean); oldmean -= wt[jptr]; /* the z in equations above */ vss += wsum3*(newmean+ oldmean -2*umean) * (newmean-oldmean); vss -= wt[jptr]* (myrank -newmean)*(myrank -newmean); } /* Process deaths */ for (j=i; j 0) { /* walk up the tree */ parent = (index-1)/2; if (index &1) /* I am the left child */ count[1] += wt[jptr] * (twt[parent] - twt[index]); else count[0] += wt[jptr] * (twt[parent] - twt[index]); index = parent; } } } else j = i+1; /* Add the weights for these obs into the tree and compute variance */ for (; i0) { parent = (index-1)/2; twt[parent] += wt[iptr]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean-oldmean); vss += wsum3*(newmean+ oldmean +wt[iptr] - 2*umean) * (oldmean-newmean); vss += wt[iptr]* (myrank -newmean)*(myrank -newmean); } count[4] += ndeath * vss/twt[0]; } UNPROTECT(1); return(count2); } survival/src/coxmart.c0000644000175100001440000000443412545056257014573 0ustar hornikusers/* $Id: coxmart.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** Compute the martingale residual for a Cox model ** ** Input ** n number of subjects ** method will be ==1 for the Efron method ** time vector of times ** status vector of status values ** score the vector of subject scores, i.e., exp(beta*z) ** strata is =1 for the last obs of a strata ** mark carried forward from the coxfit routine ** ** Output ** expected the expected number of events for the subject ** ** The martingale residual is more of a nuisance for the Efron method ** */ #include #include "survS.h" #include "survproto.h" void coxmart(Sint *sn, Sint *method, double *time, Sint *status, Sint * strata, double *score, double *wt, double *expect) { register int i,j; int lastone; int n; double deaths, denom=0, e_denom=0; double hazard; double temp, wtsum; double downwt; n = *sn; strata[n-1] =1; /* Failsafe */ /* Pass 1-- store the risk denominator in 'expect' */ for (i= n -1; i>=0; i--) { if (strata[i]==1) denom =0; denom += score[i]*wt[i]; if (i==0 || strata[i-1]==1 || time[i-1]!=time[i]) expect[i] = denom; else expect[i] =0; } /* Pass 2-- now do the work */ deaths=0; wtsum =0; e_denom=0; hazard =0; lastone = 0; for (i= 0; i #include "survS.h" #include "survproto.h" SEXP coxfit6(SEXP maxiter2, SEXP time2, SEXP status2, SEXP covar2, SEXP offset2, SEXP weights2, SEXP strata2, SEXP method2, SEXP eps2, SEXP toler2, SEXP ibeta, SEXP doscale2) { int i,j,k, person; double **covar, **cmat, **imat; /*ragged arrays */ double wtave; double *a, *newbeta; double *a2, **cmat2; double *scale; double denom=0, zbeta, risk; double temp, temp2; int ndead; /* number of death obs at a time point */ double tdeath=0; /* ndead= total at a given time point, tdeath= all */ double newlk=0; double dtime, d2; double deadwt; /*sum of case weights for the deaths*/ double efronwt; /* sum of weighted risk scores for the deaths*/ int halving; /*are we doing step halving at the moment? */ int nrisk; /* number of subjects in the current risk set */ double *maxbeta; /* copies of scalar input arguments */ int nused, nvar, maxiter; int method; double eps, toler; int doscale; /* vector inputs */ double *time, *weights, *offset; int *status, *strata; /* returned objects */ SEXP imat2, means2, beta2, u2, loglik2; double *beta, *u, *loglik, *means; SEXP sctest2, flag2, iter2; double *sctest; int *flag, *iter; SEXP rlist, rlistnames; int nprotect; /* number of protect calls I have issued */ /* get local copies of some input args */ nused = LENGTH(offset2); nvar = ncols(covar2); method = asInteger(method2); maxiter = asInteger(maxiter2); eps = asReal(eps2); /* convergence criteria */ toler = asReal(toler2); /* tolerance for cholesky */ doscale = asInteger(doscale2); time = REAL(time2); weights = REAL(weights2); offset= REAL(offset2); status = INTEGER(status2); strata = INTEGER(strata2); /* ** Set up the ragged arrays and scratch space ** Normally covar2 does not need to be duplicated, even though ** we are going to modify it, due to the way this routine was ** was called. In this case NAMED(covar2) will =0 */ nprotect =0; if (NAMED(covar2)>0) { PROTECT(covar2 = duplicate(covar2)); nprotect++; } covar= dmatrix(REAL(covar2), nused, nvar); PROTECT(imat2 = allocVector(REALSXP, nvar*nvar)); nprotect++; imat = dmatrix(REAL(imat2), nvar, nvar); a = (double *) R_alloc(2*nvar*nvar + 5*nvar, sizeof(double)); newbeta = a + nvar; a2 = newbeta + nvar; maxbeta = a2 + nvar; scale = maxbeta + nvar; cmat = dmatrix(scale + nvar, nvar, nvar); cmat2= dmatrix(scale + nvar +nvar*nvar, nvar, nvar); /* ** create output variables */ PROTECT(beta2 = duplicate(ibeta)); beta = REAL(beta2); PROTECT(means2 = allocVector(REALSXP, nvar)); means = REAL(means2); PROTECT(u2 = allocVector(REALSXP, nvar)); u = REAL(u2); PROTECT(loglik2 = allocVector(REALSXP, 2)); loglik = REAL(loglik2); PROTECT(sctest2 = allocVector(REALSXP, 1)); sctest = REAL(sctest2); PROTECT(flag2 = allocVector(INTSXP, 1)); flag = INTEGER(flag2); PROTECT(iter2 = allocVector(INTSXP, 1)); iter = INTEGER(iter2); nprotect += 7; /* ** Subtract the mean from each covar, as this makes the regression ** much more stable. */ tdeath=0; temp2=0; for (i=0; i 0) temp = temp2/temp; /* scaling */ else temp=1.0; /* rare case of a constant covariate */ scale[i] = temp; for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } else { /* scaling is only turned off during debugging still, cover the case */ for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } } if (doscale==1) { for (i=0; i population ** of the earth, any beta*x over 20 is a silly relative risk for a Cox ** model, however. ** We want to cut off huge values, but not take action very often since ** doing so can mess up the iteration in general. ** One of the case-cohort papers suggests using an offset of -100 to ** indicate "no risk", meaning that x*beta values of 50-100 can occur ** in "ok" data sets. Compromise. */ for (i=0; i=0; ) { if (strata[person] == 1) { nrisk =0 ; denom = 0; for (i=0; i=0 &&time[person]==dtime) { /* walk through the this set of tied times */ nrisk++; zbeta = offset[person]; /* form the term beta*z (vector mult) */ for (i=0; i0) { /* we need to add to the main terms */ if (method==0) { /* Breslow */ loglik[1] -= deadwt* log(denom); for (i=0; i=0; ) { if (strata[person] == 1) { /* rezero temps for each strata */ denom = 0; nrisk =0; for (i=0; i=0 && time[person]==dtime) { nrisk++; zbeta = offset[person]; for (i=0; i0) { /* add up terms*/ if (method==0) { /* Breslow */ newlk -= deadwt* log(denom); for (i=0; i maxbeta[i]) newbeta[i] = maxbeta[i]; else if (newbeta[i] < -maxbeta[i]) newbeta[i] = -maxbeta[i]; } } } /* return for another iteration */ /* ** We end up here only if we ran out of iterations */ loglik[1] = newlk; chinv2(imat, nvar); for (i=0; i newtime and id=nid. */ for (k=0; k eps) eps = matrix[i][i]; for (j=(i+1); j=0; i--) { if (matrix[i][i+m]==0) y[i+m] =0; else { temp = y[i+m]/matrix[i][i+m]; for (j= i+1; j=0; i--) { if (diag[i] == 0) y[i] =0; else { temp = y[i] / diag[i]; for (j=0; j0) df++; /* count up the df */ for (i=0; i< *ntest; i++) { for (j=0; j0]) * 1e-8 ** The events are counted in the last cell to which person years are ** added in the while() loop below. We don't want to "spill over" into ** a next (incorrect) cell due to accumulated round off, in the case ** that a subjects fu time exactly matches one of the cell boundaries. */ eps =0; /* guard against the rare case that all(time==0) */ for (i=0; i0) { eps = timeleft; /* starting guess for min = first non-zero value*/ break; } } for (; i0) && (timeleft < eps)) eps = timeleft; } eps *= 1e-8; *offtable =0; for (i=0; i eps) { thiscell = pystep(odim, &index, &d1, &d2, data, ofac, odims, ocut, timeleft, 0); if (index >=0) { pyears[index] += thiscell * wt[i]; pn[index] += 1; } else *offtable += thiscell * wt[i]; for (j=0; j=0 && doevent) pcount[index] += event[i] * wt[i]; } } survival/src/coxdetail.c0000644000175100001440000001473312545056257015075 0ustar hornikusers/* $Id: coxdetail.c 11357 2009-09-04 15:22:46Z therneau $ ** ** Return all of the internal peices of a Cox model ** ** the input parameters are ** ** nused :number of people ** nvar :number of covariates ** ndead :total number of deaths ** y(3,n) :start, stop, and event for each subject ** covar(nv,n) :covariates for person i. ** Note that S sends this in column major order. ** strata(n) :marks the strata. Will be 1 if this person is the ** last one in a strata. If there are no strata, the ** vector can be identically zero, since the nth person's ** value is always assumed to be = to 1. ** score(n) :the risk score for the subject ** weights(n) :case weights ** means :first element contains the method ** rmat : if first element =1, then calculate a risk matrix ** ** returned parameters ** ndead :the number of unique death times ** strata :the indices of the unique time points ** y[1, ] :the number of deaths at each time point ** y[2, ] :the number at risk at each time point ** y[3, ] :the increment in the cum -hazard at t ** score :the weighted number of events at each time point ** weights[] :the increment in the variance of the cum-haz at t ** means(nv,nd) :the matrix of weighted means, one col per unique event ** time ** u(nv,nd) :the score vector components, one per unique event time ** var(nd,nv,nv):components of the information matrix ** rmat(nd, n) :has a "1" if subject i is at risk at time j ** nrisk2 :the weighted number at risk at each time point ** ** work arrays ** a(nvar) ** a2(nvar) ** cmat(nvar,nvar) ragged array ** cmat2(nvar,nvar) ** wmeans(nvar) ** ** the 5 arrays a, a2, cmat, cmat2 and wmeans are passed as a single ** vector of storage, and then broken out. ** ** the data must be sorted by ascending time within strata, deaths before ** living within tied times. */ #include #include "survS.h" #include "survproto.h" void coxdetail(Sint *nusedx, Sint *nvarx, Sint *ndeadx, double *y, double *covar2, Sint *strata, double *score, double *weights, double *means2, double *u2, double *var, Sint *rmat, double *nrisk2, double *work) { int i,j,k,person; int nused, nvar; int nrisk, ndead; double **covar, **cmat; /*ragged arrays */ double **means; double **u; double *a; double *a2, **cmat2; double *wmeans; double denom; double time; double temp, temp2, temp3; double method; double hazard; double varhaz; int itemp, deaths; int ideath; double efron_wt, d2; double risk; double meanwt; double wdeath; double *start, *stop, *event; int rflag; nused = *nusedx; nvar = *nvarx; method= *means2; ndead = *ndeadx; rflag = 1- rmat[0]; /* ** Set up the ragged arrays */ covar= dmatrix(covar2, nused, nvar); means= dmatrix(means2, ndead, nvar); u = dmatrix(u2, ndead, nvar); cmat = dmatrix(work, nvar, nvar); cmat2= dmatrix(work + nvar*nvar, nvar, nvar); a = work + 2*nvar*nvar; a2= a+nvar; wmeans = a2+nvar; start =y; stop =y + nused; event =y + nused +nused; /* ** Subtract the mean from each covar, as this makes the variance calc ** much more stable */ for (i=0; i0 ** isurv[n] -- individual survival curves ** ** cx and cy must be sorted by (event before censor) within stop time */ #include #include "survS.h" #include "survproto.h" static double *y, *nscore, **newx, **surv, **vsurv, *isurv, **used, **tvar; static int *strata; static double ttime, /* Some HP compilers choke on "time" as a variable */ **imat, *mean; static int death, ncurve, se, nvar, n; static void addup(); void agsurv3(Sint *sn, Sint *snvar, Sint *sncurve, Sint *snpt, Sint *sse, double *score, double *sy, Sint *grpx, double *r, double *coef, double *var, double *xmean, Sint *scn, double *cy, double *cx, double *ssurv, double *varh, double *sused, Sint *smethod) { int i,j,k,l; double *start, *stop, *event; int cn; int npt, nvar2, method; int kk=0, psave; int itime; int person; int deaths, nrisk; int need; double *a=0, *a2=0; double weight=0, e_denom, denom; double inc, sumt, km =0; double temp, downwt, d2; double haz, varhaz; double **oldx =0; n = *sn; nvar = *snvar; cn = *scn; npt = *snpt; se = *sse; ncurve = *sncurve; method = *smethod; death = method/10; method = method - death*10; y = sy; start = cy; stop = cy+ cn; event = cy+ cn+ cn; strata = grpx; /* ** scratch space */ need = 2*n + se*nvar*(2+ n*(n+1)/2) + nvar; nscore = (double *) ALLOC(need, sizeof(double)); for (i=0; i0) vsurv[i][itime]=0; } return; } /* ** Note that the subjects are sorted in strata order */ pstart=0; for (ic=0; ic= ttime) { temp = -haz*nscore[i]; /*increment to the individual hazard*/ if (death==0) { wt += isurv[i]; totsurv += exp(temp) * isurv[i]; } else { wt += 1; totsurv += temp; } isurv[i] *= exp(temp); } /* ** The variance is computed as though it were the Ederer est, always */ if (se==1) { /* Do the variance term (nasty) */ for (j=pstart; j<=i; j++) { temp =0; for (k=0; k #include "survS.h" #include "survproto.h" void coxscho(Sint *nusedx, Sint *nvarx, double *y, double *covar2, double *score, Sint *strata, Sint *method2, double *work) { int i,k,person; int nused, nvar; double **covar; double *a; double *a2; double *mean; double denom, weight; double time; double temp; double method; double deaths; double efron_wt; double *start, *stop, *event; nused = *nusedx; nvar = *nvarx; method= *method2; /* ** Set up the ragged arrays */ covar= dmatrix(covar2, nused, nvar); a = work; a2= a+nvar; mean = a2+nvar; start =y; stop =y + nused; event =y + nused +nused; /* ** Now walk through the data */ for (person=0; person #include "survS.h" #include "survproto.h" void agexact(Sint *maxiter, Sint *nusedx, Sint *nvarx, double *start, double *stop, Sint *event, double *covar2,double *offset, Sint *strata, double *means, double *beta, double *u, double *imat2, double loglik[2], Sint *flag, double *work, Sint *work2, double *eps, double *tol_chol, double *sctest) { int i,j,k, l, person; int iter; int n, nvar; double **covar, **cmat, **imat; /*ragged array versions*/ double *a, *newbeta; double *score, *newvar; double denom, zbeta, weight; double time; double temp; double newlk =0; int halving; /*are we doing step halving at the moment? */ int nrisk, deaths; int *index, *atrisk; n = *nusedx; nvar = *nvarx; /* ** Set up the ragged arrays */ covar= dmatrix(covar2, n, nvar); imat = dmatrix(imat2, nvar, nvar); cmat = dmatrix(work, nvar, nvar); a = work + nvar*nvar; newbeta = a + nvar; score = newbeta + nvar; newvar = score + n; index = (int *) work2; atrisk= index+n; /* ** Subtract the mean from each covar, as this makes the regression ** much more stable */ for (i=0; i=0) { for (i=0; i=0) { for (i=0; i #define SMALL -200 /* what to use for log(f(x)) when f(x) gives a zero, i.e., the calling made a really bad guess for beta */ static void exvalue_d(double z, double ans[4], int j); static void logistic_d(double z, double ans[4], int j); static void gauss_d(double z, double ans[4], int j); static void (*sreg_gg)(); #define SPI 2.506628274631001 /* sqrt(2*pi) */ #define ROOT_2 1.414213562373095 double survregc1(int n, int nvar, int nstrat, int whichcase, double *beta, int dist, Sint *strat, double *offset, double *time1, double *time2, double *status, double *wt, double **covar, double **imat, double **JJ, double *u, SEXP expr, SEXP rho, double *dummy, int nf, Sint *frail, double *fdiag, double *jdiag ) { int person, i,j,k; int nvar2; /* nvar + nstrat */ int nvar3; /* nvar2 + nf */ int strata; double eta, sigma; double z, zu, loglik, temp, temp2; double sz; double sig2; double funs[4], ufun[4]; int fgrp =0; /* the =0 to quiet a compiler warning */ double w; /* add "=0" to keep the compiler from worrying about uninitialized vars */ double g=0, dg=0, ddg=0, dsig=0, ddsig=0, dsg=0; switch(dist) { case 1: sreg_gg = exvalue_d; break; case 2: sreg_gg = logistic_d; break; case 3: sreg_gg = gauss_d; break; } nvar2 = nvar + nstrat; nvar3 = nvar2 + nf; loglik =0; if (whichcase==0) { for (i=0; i1) { /* ** multiple scales: pick the right sigma for this obs ** The more common case of a single scale is set 6 lines above */ strata= strat[person] -1; /*S likes to start counting at 1 */ sigma = exp(beta[strata+nvar+nf]); sig2 = 1/(sigma*sigma); } eta =0; for (i=0; i0){ fgrp = frail[person] -1; eta += beta[fgrp]; } sz = (time1[person] - eta); /* sigma * z */ z = sz /sigma; j = status[person]; /*convert to integer */ switch(j) { case 1: /* exact */ (*sreg_gg)(z, funs,1); if (funs[1] <=0) { /* off the probability scale -- avoid log(0), and set the ** derivatives to gaussian limits (almost any deriv will ** do, since the function value triggers step-halving). */ g = SMALL; dg = -z/sigma; ddg = -1/sigma; dsig =0; ddsig=0; dsg=0; } else { g = log(funs[1]) - log(sigma); temp = funs[2]/sigma; temp2= funs[3]*sig2; dg = -temp; dsig= -temp*sz; ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(dsig +1); ddsig = sz*sz* temp2 - dsig*(1+dsig); dsig -= 1; } break; case 0: /* right censored */ (*sreg_gg)(z, funs,2); if (funs[1] <=0) { g = SMALL; dg = z/sigma; ddg =0; dsig =0; ddsig=0; dsg=0; } else { g = log(funs[1]); temp = -funs[2]/(funs[1]*sigma); temp2= -funs[3]*sig2/funs[1]; dg = -temp; dsig= -temp*sz; ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(dsig +1); ddsig = sz*sz* temp2 - dsig*(1+dsig); } break; case 2: /* left censored */ (*sreg_gg)(z, funs,2); if (funs[0] <=0) { /* off the probability scale -- avoid log(0) */ g = SMALL; dg = -z/sigma; dsig =0; ddsig=0; dsg=0; ddg =0; } else { g = log(funs[0]); temp = funs[2]/(funs[0]*sigma); temp2= funs[3]*sig2/funs[0]; dg = -temp; dsig= -temp*sz; ddg= temp2 - dg*dg; dsg = sz * temp2 - dg*(dsig +1); ddsig = sz*sz* temp2 - dsig*(1+dsig); } break; case 3: /* interval censored */ zu = (time2[person] - eta)/sigma; /*upper endpoint */ (*sreg_gg)(z, funs, 2); (*sreg_gg)(zu,ufun ,2); if (z>0) temp = funs[1] - ufun[1]; /*stop roundoff in tails*/ else temp = ufun[0] - funs[0]; if (temp <=0) { /* off the probability scale -- avoid log(0) */ g = SMALL; dg = 1; ddg =0; dsig =0; ddsig=0; dsg=0; } else { g = log(temp); dg = -(ufun[2] - funs[2])/(temp*sigma); ddg = (ufun[3] - funs[3])*sig2/temp - dg*dg; dsig = (z*funs[2] - zu*ufun[2])/temp; ddsig= ((zu*zu*ufun[3] - z*z*funs[3])/temp) - dsig*(1+dsig); dsg = ((zu*ufun[3] - z*funs[3])/ (temp*sigma)) - dg * (dsig +1); } break; } loglik += g * wt[person]; /*if (person<8) fprintf(stderr, "i=%d, g=%g, dg=%g, ddg=%g, dsg=%g\n", person, g, dg, ddg, dsg);*/ /* ** Now the derivs wrt loglik ** Remember that the "x" for a sparse term is 1 */ if (whichcase==1) continue; /*only needed the loglik */ w = wt[person]; if (nf>0) { u[fgrp] += dg * w; fdiag[fgrp] -= ddg * w; jdiag[fgrp] += dg*dg *w; } for (i=0; i0) { imat[i][fgrp] -= covar[i][person] * ddg * w; JJ [i][fgrp] += temp * dg; } } if (nstrat!=0) { /* need derivative wrt log sigma */ k = strata+nvar; u[k+nf] += w* dsig; for (i=0; i0) { imat[k][fgrp] -= dsg * w; JJ [k][fgrp] += dsig *dg *w; } } } return(loglik); } /* ** Case ans[0] ans[1] ans[2] ans[3] ** 1 f f'/f f''/ f ** 2 F 1-F f f' ** ** We do both F and 1-F to avoid the error in (1-F) for F near 1 */ static void logistic_d(double z, double ans[4], int j) { double w, temp; int sign, ii; /* ** The symmetry of the logistic allows me to be careful, and never take ** exp(large number). This routine should be very accurate. */ if (z>0) { w = exp(-z); sign = -1; ii=0; } else { w = exp(z); sign = 1; ii=1; } temp = 1+w; switch(j) { case 1: ans[1] = w/(temp*temp); ans[2] = sign*(1-w)/temp; ans[3] = (w*w -4*w +1)/(temp*temp); break; case 2: ans[1-ii] = w/temp; ans[ii] = 1/temp; ans[2] = w/(temp*temp); ans[3] = sign*ans[2]*(1-w)/temp; break; } } static void gauss_d(double z, double ans[4], int j) { double f; f = exp(-z*z/2) /SPI; switch(j) { case 1: ans[1] =f; ans[2] = -z; ans[3] = z*z -1; break; case 2: if (z>0) { ans[0] = (1 + erf(z/ROOT_2))/2; ans[1] = erfc(z/ROOT_2) /2; } else { ans[1] = (1 + erf(-z/ROOT_2))/2; ans[0] = erfc(-z/ROOT_2) /2; } ans[2] = f; ans[3] = -z*f; break; } } /* ** In the Gaussian and logistic cases, I could avoid numeric disaster by only ** evaluating exp(x) for x<0. By symmetry, I got what I need for ** x >0. The extreme value dist, howerver, is asymmetric, and I don't yet ** see the appropriate numeric tricks. ** Perhaps a Taylor series will could be used for large z. */ static void exvalue_d(double z, double ans[4], int j) { double temp; double w; if (z < SMALL) w= exp(SMALL); else if (-z < SMALL) w = exp(-SMALL); /* stop infinite answers */ else w = exp(z); temp = exp(-w); switch(j) { case 1: ans[1] = w*temp; ans[2] = 1-w; ans[3] = w*(w-3) +1; break; case 2: ans[0] = 1-temp; ans[1] = temp; ans[2] = w*temp; ans[3] = w*temp*(1-w); break; } } survival/src/doloop.c0000644000175100001440000000362412545056257014412 0ustar hornikusers/* $Id: doloop.c 11357 2009-09-04 15:22:46Z therneau $ ** ** Program to mimic a set of nested do loops ** ** Usual calling sequence would be ** init_doloop(min,max); ** while (doloop(nloops, index) >=min) { ** some calculations ** } ** ** The result of this is as though the code had been written for "nloops" ** nested for loops: ** ** for (index[0]=min; index[0]= (minval+i)) return (minval+i-1); else return (minval-1); } nloops--; index[nloops]++; /*increment the lastmost index */ if (index[nloops] <= (maxval-depth)) return(index[nloops]); else if (nloops ==0) return(minval - depth); else { depth++; index[nloops] = doloop(nloops, index) +1; depth--; return(index[nloops]); } } survival/src/survfitci.c0000644000175100001440000002304012545056257015126 0ustar hornikusers/* Automatically generated from all.nw using noweb */ #include "survS.h" /* allocate a ragged array of a given number of rows and columns */ static double **dmatrix2(int nrow, int ncol) { int i; double **mat; double *d; mat = (double **) R_alloc(nrow, sizeof(double *)); d = (double *) R_alloc(nrow*ncol, sizeof(double)); for (i=0; i0) PROTECT(vmat2 = allocMatrix(REALSXP, nstate, ntime)); else PROTECT(vmat2 = allocMatrix(REALSXP, 1, 1)); /* dummy object */ vmat = REAL(vmat2); PROTECT(nevent2 = allocVector(INTSXP, ntime)); nevent = INTEGER(nevent2); PROTECT(ncensor2= allocVector(INTSXP, ntime)); ncensor = INTEGER(ncensor2); PROTECT(nrisk2 = allocMatrix(INTSXP, nstate, ntime)); nrisk = INTEGER(nrisk2); PROTECT(cumhaz2= allocVector(REALSXP, nstate*nstate*ntime)); cumhaz = REAL(cumhaz2); nprotect = 8; /* allocate space for scratch vectors */ ws = (double *) R_alloc(2*nstate, sizeof(double)); temp2 = ws + nstate; ns = (int *) R_alloc(nstate, sizeof(int)); atrisk = (int *) R_alloc(nperson, sizeof(int)); wtp = (double *) R_alloc(nperson, sizeof(double)); hmat = (double**) dmatrix2(nstate, nstate); if (sefit >0) umat = (double**) dmatrix2(nperson, nstate); chaz = (double**) dmatrix2(nstate, nstate); /* R_alloc does not zero allocated memory */ for (i=0; i0) { newstate = status[k] -1; /* 0 based subscripts */ oldstate = cstate[id[k]]; nevent[itime]++; wevent += wt[k]; hmat[oldstate][newstate] += wt[k]; } else ncensor[itime]++; } else break; } if (nevent[itime]> 0) { /* finish computing H */ for (j=0; j0) { temp =0; for (k=0; k0) { /* Update U, part 1 U = U %*% H -- matrix multiplication */ for (j=0; j0) { kk = id[k]; /* row number in U */ oldstate= cstate[kk]; newstate= status[k] -1; umat[kk][oldstate] -= p[oldstate]/ws[oldstate]; umat[kk][newstate] += p[oldstate]/ws[oldstate]; } } else break; } } /* Finally, update chaz and p. */ for (j=0; j0) { for (k=0; k0) cstate[id[j]] = status[j]-1; /*new state */ atrisk[id[j]] =0; } else break; } itime++; } /* return a list */ PROTECT(rlist=mkNamed(VECSXP, rnames)); SET_VECTOR_ELT(rlist, 0, nrisk2); SET_VECTOR_ELT(rlist, 1, nevent2); SET_VECTOR_ELT(rlist, 2, ncensor2); SET_VECTOR_ELT(rlist, 3, pmat2); SET_VECTOR_ELT(rlist, 4, cumhaz2); SET_VECTOR_ELT(rlist, 5, vmat2); UNPROTECT(nprotect +1); return(rlist); } survival/src/init.c0000644000175100001440000000510612545056257014056 0ustar hornikusers/* ** This file causes the entry points of my .C routines to be preloaded ** Added at the request of R-core. ** It adds one more layer of protection by declaring the number of arguments, ** and perhaps a tiny bit of speed */ #include "survS.h" #include "R_ext/Rdynload.h" #include "survproto.h" static const R_CMethodDef Centries[] = { {"Cagfit5a", (DL_FUNC) &agfit5a, 20}, {"Cagfit5b", (DL_FUNC) &agfit5b, 19}, {"Cagfit5c", (DL_FUNC) &agfit5c, 1}, {"Cagsurv3", (DL_FUNC) &agsurv3, 19}, {"Cagsurv4", (DL_FUNC) &agsurv4, 6}, {"Cagsurv5", (DL_FUNC) &agsurv5, 10}, {"Cagexact", (DL_FUNC) &agexact, 20}, {"Cagmart", (DL_FUNC) &agmart, 9}, {"Cagmart2", (DL_FUNC) &agmart2, 13}, {"Cagscore", (DL_FUNC) &agscore, 10}, {"Ccoxdetail", (DL_FUNC) &coxdetail, 14}, {"Ccoxfit5a", (DL_FUNC) &coxfit5_a, 20}, {"Ccoxfit5b", (DL_FUNC) &coxfit5_b, 19}, {"Ccoxfit5c", (DL_FUNC) &coxfit5_c, 5}, {"Ccoxmart", (DL_FUNC) &coxmart, 8}, {"Ccoxmart2", (DL_FUNC) &coxmart2, 7}, {"Ccoxph_wtest",(DL_FUNC) &coxph_wtest,6}, {"Ccoxscho", (DL_FUNC) &coxscho, 8}, {"Ccoxscore", (DL_FUNC) &coxscore, 10}, {"Cpyears1", (DL_FUNC) &pyears1, 22}, {"Cpyears2", (DL_FUNC) &pyears2, 14}, {"Csurvdiff2", (DL_FUNC) &survdiff2, 13}, {"Csurvfit4", (DL_FUNC) &survfit4, 4}, {NULL, NULL, 0} }; static const R_CallMethodDef Callentries[] = { {"Cagfit4", (DL_FUNC) &agfit4, 13}, {"Cagmart3", (DL_FUNC) &agmart3, 6}, {"Cconcordance1", (DL_FUNC) &concordance1, 4}, {"Cconcordance2", (DL_FUNC) &concordance2, 6}, {"Ccoxcount1", (DL_FUNC) &coxcount1, 2}, {"Ccoxcount2", (DL_FUNC) &coxcount2, 4}, {"Ccoxexact", (DL_FUNC) &coxexact, 8}, {"Ccoxfit6", (DL_FUNC) &coxfit6, 12}, {"Cpyears3b", (DL_FUNC) &pyears3b, 10}, {"Csurvfitci", (DL_FUNC) &survfitci, 10}, {"Csurvreg6", (DL_FUNC) &survreg6, 15}, {"Csurvreg7", (DL_FUNC) &survreg7, 21}, {"Ctmerge", (DL_FUNC) &tmerge, 7}, {NULL, NULL, 0} }; void R_init_survival(DllInfo *dll){ R_registerRoutines(dll, Centries, Callentries, NULL, NULL); /* My take on the documentation is that adding the following line will make symbols available ONLY through the above tables. Anyone who then tried to link to my C code would be SOL. It also wouldn't work with .C("whatever", ....) which I use in my test directory. */ /* R_useDynamicSymbols(dll, FALSE); */ } survival/src/survConcordance.c0000644000175100001440000001166212545056257016255 0ustar hornikusers/* ** $Id: survConcordance.c 11166 2008-11-24 22:10:34Z therneau $ ** ** For each observation, we want to know, for the subset of observations ** with longer survival (and only those) ** number with smaller, bigger, and tied x values ** ** The input data is sorted, largest survival to smallest survival ** ** n number of time/status/x values ** time ** status needed to keep track of tied survival times ** x vector of scores ** n2 number of unique x values ** x2 sorted vector of unique x values, smallest to largest ** ** temp scratch vector of length 2* n2 ** ** returned ** result number concordant, discordant, tied survival, tied x but ** not tied survival, and incomparable times ** (bigger survival + smaller risk score = concordant) */ #include "survS.h" #include void survConcordance(Sint *np, double *time, Sint *status, double *x, Sint *n2p, double *x2, Sint *temp,Sint *result) { int i, j, k=0; int start, end; int n, n2; Sint *count1, *count2, *count; int tdeath; int nright, nsame; n = *np; n2= *n2p; count1 = &(temp[0]); count2 = &(temp[n2]); for (i=0; i<5; i++) result[i] =0; /* redundant I think */ for (i=0; i x2[k]. (Draw a picture). ** The root of the tree is element k= floor((n2-1)/2), with value x2[k]. ** In general, for any subtree that "owns" elements i to j, the root ** of that subtree is element k= floor((i+j)/2), whose left subtree ** owns elements i to k-1 of the tree, and right subtree owns elements ** k+1 to j. ** ** As we update, count[i] will be the number of data values in this ** node and all nodes below. ** ** We walk through the data one survival time at at time, comparing each ** to all the survival times above it. ** If the time is censored, all those above are "incomparable". ** Otherwise, we need to find the position of x[i], among x[1: (i-1)] ** We do this by updating the counts in the binary tree. The count ** vector contains the number of x[0 to i] that are in or below any ** given node k of the binary tree. ** ** Tied death times are a nuisance; we have to refrain from updating ** the counts until the end of each set of them. Thus a vector ** count1 (up to date) and count2 (lagged). ** nright = sum(# values to the right, each time I take a left branch) */ tdeath =0; /* current count of tied deaths */ for (i=0; i 0) { /* ** Walk the tree a first time, to count this observation's ** position */ nright = 0; start = 0; end= n2-1; /*start to end of sublist being looked at */ if (tdeath==0) count=count1; /* use the appropriate count */ else count=count2; while(start <= end) { k = (start+end)/2; if (x[i] == x2[k]) break; if (x[i] < x2[k]) { /* take the left branch (smaller numbers) */ end = k-1; nright = nright + (count[k] - count[(start+end)/2]); } else start = k+1; /*right branch */ } /* ** At this point x[i] = x2[k]; we've found the number in the ** x2 list */ nsame = count[k]; /*provisional */ if (k start) /* there is a left hand branch below here */ nsame = nsame - count[(start+k-1)/2]; result[3] += nsame; result[1] += nright; /* # values bigger than x[i] */ result[0] += i - (tdeath + nsame + nright); /* # smaller */ /* Is the next survival time tied with this one? */ if (i<(n-1) && status[i+1]>0 &&(time[i] == time[i+1])) { tdeath += 1; /* Yes it is */ if (tdeath==1) { for (j=0; j #include "survS.h" #include "survproto.h" SEXP agmart3(SEXP surv2, SEXP score2, SEXP weight2, SEXP strata2, SEXP sortx, SEXP method2) { int k, ksave; int p, istrat, indx2; double deaths, denom, e_denom; double hazard, e_hazard, cumhaz; double temp, time; double wtsum; int n, person; /* pointers to the input data */ double *start, *stop, *event; double *weight, *score; int *sort1, *sort2, *strata; int method; /* integer version of input */ /* output */ SEXP resid2; double *resid; n = nrows(surv2); method = asInteger(method2); start = REAL(surv2); stop = start +n; event = stop +n; weight= REAL(weight2); score = REAL(score2); sort1 = INTEGER(sortx); sort2 = sort1 + n; strata= INTEGER(strata2); PROTECT(resid2 = allocVector(REALSXP, n)); resid = REAL(resid2); /* ** 'person' walks through the the data from 1 to n, ** sort1[0] points to the largest stop time, sort1[1] the next, ... ** 'time' is a scratch variable holding the time of current interest ** 'indx2' walks through the start times. It will be smaller than ** 'person': if person=27 that means that 27 subjects have stop >=time, ** and are thus potential members of the risk set. If 'indx2' =9, ** that means that 9 subjects have start >=time and thus are NOT part ** of the risk set. (stop > start for each subject guarrantees that ** the 9 are a subset of the 27). ** Basic algorithm: move 'person' forward, adding the new subject into ** the risk set. If this is a new, unique death time, take selected ** old obs out of the sums, add in obs tied at this time, then update ** the cumulative hazard. Everything resets at the end of a stratum. ** The sort order is from large time to small, so we encounter a subject's ** ending time first, then their start time. ** The martingale residual for a subject is ** status - (cumhaz at end of their interval - cumhaz at start)*score */ istrat=0; indx2 =0; denom =0; cumhaz =0; for (person=0; person #include #include "survS.h" #include "survproto.h" static double **covar, **cmat, **cmat2; static double *a, *oldbeta, *a2; static double *offset, *weights; static int *event, *frail; static double *score, *start, *stop; static int *sort1, *sort2; static double *tmean; static int ptype, pdiag; static double *ipen, *upen, logpen; static Sint *zflag; static double **cmatrix(double *, int, int); void agfit5a(Sint *nusedx, Sint *nvarx, double *yy, double *covar2, double *offset2, double *weights2, Sint *strata, Sint *sort, double *means, double *beta, double *u, double *loglik, Sint *methodx, Sint *ptype2, Sint *pdiag2, Sint *nfrail, Sint *frail2, void *fexpr1, void *fexpr2, void *rho) { int i,j,k, person; int nused, nvar; int nf, nvar2; int deaths, itemp; int istrat, indx2, p, ksave; double denom, zbeta, risk; double temp; double d2, efron_wt; double method; double meanwt, time; nused = *nusedx; nvar = *nvarx; nf= *nfrail; method= *methodx; nvar2 = nvar + nf; ptype = *ptype2; pdiag = *pdiag2; /* ** Allocate storage for the arrays and vectors ** Since they will be used later, sizes are based on what will be ** needed with the frailty terms. */ if (nvar >0) { covar= cmatrix(covar2, nused, nvar); cmat = cmatrix(0, nvar2, nvar+1); cmat2= cmatrix(0, nvar2, nvar+1); } a = Calloc(4*nvar2 + 5*nused , double); oldbeta = a + nvar2; a2 = oldbeta + nvar2; weights = a2+ nvar2; offset = weights + nused; score = offset + nused; tmean = score + nused; start = tmean + nvar2; stop = start + nused; event = Calloc(3*nused, int); sort1 = event + nused; sort2 = sort1 + nused; for (i=0; i nvar) i=nf; else i=nvar; if (nf > nvar*nvar) j=nf; else j=nvar*nvar; if (pdiag==0) upen = Calloc(2*i, double); else upen = Calloc(i+j, double); ipen = upen + i; if (ptype>1) zflag = Calloc(nvar, Sint); else zflag = Calloc(2, Sint); if (nf>0) { frail = Calloc(nused, int); for (i=0; i=time, ** and are thus potential members of the risk set. If 'indx2' =9, ** that means that 9 subjects have start >=time and thus are NOT part ** of the risk set. (stop > start for each subject guarrantees that ** the 9 are a subset of the 27). ** Basic algorithm: move 'person' forward, adding the new subject into ** the risk set. If this is a new, unique death time, take selected ** old obs out of the sums, add in obs tied at this time, then ** add terms to the loglik, etc. */ istrat=0; indx2 =0; denom =0; for (person=0; person0) { imat = dmatrix(imat2, nvar2, nvar); jmat = dmatrix(jmat2, nvar2, nvar); } else { imat = 0; /*never used, but passed as dummy to chol */ jmat = 0; } for (i=0; i0) { fgrp = frail[person] -1; zbeta = offset[person] + fbeta[fgrp]; } else zbeta = offset[person]; for (i=0; i 20 && *maxiter >1) { /* ** If the above happens, then ** 1. There is a real chance for catastrophic cancellation ** in the computation of "denom", which leads to ** numeric failure via log(neg number) -> inf loglik ** 2. A risk score for one person of exp(20) > 400 million ** is either an infinite beta, in which case any ** reasonable coefficient will do, or a big overreach ** in the Newton-Raphson step. ** In either case, a good solution is step halving. However, ** if the user asked for exactly 1 iteration, we should ** just return what they asked. ** ** Why 20? Most machines have about 16 digits of precision, ** and this preserves approx 7 digits in the subtraction ** when a high risk score person leaves the risk set. ** (Because of centering, the average risk score is about 0). ** Second, if eps is small and beta is infinite, we rarely ** get a value above 16. So a 20 is usually a NR overshoot. ** A data set with zbeta=54 on iter 1 led to this fix, the ** true final solution had max values of 4.47. */ halving=1; for (i=0; i0) fgrp = frail[p] -1; else fgrp = -1; if (event[p]==0){ risk = exp(score[p]) * weights[p]; denom += risk; if (fgrp >=0) a[fgrp] += risk; for (i=0; i=0) cmat[i][fgrp] += risk * covar[i][p]; for (j=0; j<=i; j++) cmat[i][j+nf] += risk*covar[i][p]*covar[j][p]; } person++; } else { time = stop[p]; /* ** subtract out the subjects whose start time is to the right */ for (; indx20) fgrp = frail[p] - 1; else fgrp = -1; if (fgrp >=0) a[fgrp] -= risk; for (i=0; i=0) cmat[i][fgrp] -= risk* covar[i][p]; for (j=0; j<=i; j++) cmat[i][j+nf] -= risk*covar[i][p]*covar[j][p]; } } /* ** compute the averages over this death time (a2 & c2) */ efron_wt =0; meanwt =0; for (i=0; i0) { fgrp = frail[p] -1; if (fgrp>=0) a[fgrp] += risk; } else fgrp = -1; for (i=0; i=0) cmat[i][fgrp] += risk*covar[i][p]; for (j=0; j<=i; j++) cmat[i][j+nf] += risk*covar[i][p]*covar[j][p]; } if (event[p]==1) { deaths += event[p]; efron_wt += risk* weights[p]; meanwt += weights[p]; if (fgrp >= 0) { u[fgrp] += weights[p]; a2[fgrp] += risk; } for (i=0; i=0) cmat2[i][fgrp] += risk*covar[i][p]; for (j=0; j<=i; j++) cmat2[i][j+nf] += risk*covar[i][p]*covar[j][p]; } } } ksave =k; /* add results into u and imat */ itemp = -1; meanwt /= deaths; for (; person0 && newlk < *loglik) { /*it is not converging ! */ halving =1; for (i=0; i 0) { cmatrix_free(cmat2); cmatrix_free(cmat); cmatrix_free(covar); } } survival/src/cholesky3.c0000644000175100001440000000506012545056257015016 0ustar hornikusers/* $Id: cholesky3.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** subroutine to do Cholesky decompostion on a matrix: C = FDF' ** where F is lower triangular with 1's on the diagonal, and D is diagonal ** This is a specialized form for the frailty problem. The matric C in this ** case has C[1:m, 1:m] diagonal and C[(m+1):n, 1:n)] is dense. ** ** arguments are: ** n the size of the matrix to be factored ** m the size of the diagonal upper portion ** diag the diagonal upper portion ** **matrix a ragged array containing the dense portion ** toler tolerance for detecting singularity ** ** The diagonal portion of the matrix is unchanged by the factorization. ** For the dense portion, D occupies the diagonal (of the full matrix). ** The factorization is returned in the lower triangle. ** The upper triangle of the matrix is entirely unused by the process (but ** because of the compressed storage, this isn't much space). ** ** Return value: the rank of the matrix (non-negative definite), or -rank ** if not non-negative definite ** ** If a column is deemed to be redundant, then that diagonal is set to zero. ** ** Terry Therneau */ #include "survS.h" #include "survproto.h" int cholesky3(double **matrix, int n, int m, double *diag, double toler) { double temp; int i,j,k; double eps, pivot; int rank; int n2; int nonneg; n2 = n-m; /* number of full covariates */ nonneg=1; eps =0; for (i=0; i eps) eps = matrix[i][i+m]; eps *= toler; rank =0; /* pivot out the diagonal elements */ for (i=0; i #include "survS.h" #include "survproto.h" void coxscore(Sint *nx, Sint *nvarx, double *y, double *covar2, Sint *strata, double *score, double *weights, Sint *method, double *resid2, double *scratch) { int i,j, k; double temp; int n, nvar; double deaths; int dd; double *time, *status; double *a, *a2; double denom=0, e_denom; double risk; double **covar; double **resid; double hazard, meanwt; double downwt, temp2; double mean; n = *nx; nvar = *nvarx; time = y; status = y+n; a = scratch; a2 = a+nvar; /* ** Set up the ragged array */ covar= dmatrix(covar2, n, nvar); resid= dmatrix(resid2, n, nvar); e_denom=0; deaths=0; meanwt=0; for (i=0; i=0; i--) { if (strata[i]==1) { denom =0; for (j=0; j0 && (i==0 || strata[i-1]==1 || time[i]!=time[i-1])){ /* last obs of a set of tied death times */ if (deaths <2 || *method==0) { hazard = meanwt/denom; for (j=0; j LARGE) return(LARGE); return (x); } survival/src/dmatrix.c0000644000175100001440000000075312545056257014566 0ustar hornikusers/* $Id: dmatrix.c 11525 2012-12-07 17:20:39Z therneau $ ** ** set up ragged arrays, with #of columns and #of rows, ** where nrow (second arg) is what R thinks are columns ** but C thinks are rows. */ #include "survS.h" #include "survproto.h" double **dmatrix(double *array, int ncol, int nrow) { int i; double **pointer; pointer = (double **) ALLOC(nrow, sizeof(double *)); for (i=0; i0){ fdiag[i] = 1/fdiag[i]; /* this line inverts D */ for (j=0; j0) { matrix[i][ii] = 1/matrix[i][ii]; /*this line inverts D */ for (j= (i+1); j #include "survS.h" #include "survproto.h" void agmart(Sint *n, Sint *method, double *start, double *stop, Sint *event, double *score, double *wt, Sint *strata, double *resid) { int i,k; double deaths, denom, e_denom; double hazard, e_hazard; double temp, time; double wtsum; int nused; int person; nused = *n; strata[nused-1] =1; /* Failsafe */ for (i=0; i1: estimate multiple scales (strata) ** strat - if nstrat>0, contains the strata number for each subject ** eps - tolerance for convergence. Iteration continues until the ** relative change in the deviance is <= eps. ** tol_chol- tolerance for Cholesky decomposition ** dist - 1=extreme value, 2=logistic, 3=gaussian, 4=callback ** dexpr - for callback, the expression to be evaluated that evaluates ** the distribution function of the random effect ** rho - for callback, the environment (R) or frame (Splus) in which ** to do evaluations ** ptype - 1= sparse penalties, 2=dense penalties, 1+2 = both, 0=none ** pdiag - 0 = the penalty matrix is diagonal ** nfrail - number of levels of the sparse term (0 = no sparse term) ** fgrp - which frailty group each subject is in ** pexpr1 - for callback, the expression to eval for sparse penalties ** pexpr2 - the expression for dense penalties ** ** Output ** beta - the final coef vector ** iter - the number of iterations consumed ** hmat - the cholesky of the penalized information matrix ** hinv - the cholesky of the inverse of hmat ** hdiag - diagonal portion of hinv ** loglik - the final log-liklihood ** u - the final score vector. Usually =0 at convergence, but ** useful in other cases for a score test. ** flag - success flag 0 =ok ** -1= did not converge ** ** Work arrays ** newbeta(nvar)- always contains the "next iteration" ** u(nvar) - first deriv of the loglik ** JJ = the approx variance matrix J'J, guarranteed non-singular ** ** Notes on hmat: H will be p=(nfrail+ nvar + nstrat) square, but the ** upper left nfrail*nfrail corner is a diagonal matrix. It is stored ** as "hdiag", which "hmat" contains the remaining dense portion. If ** H = LDL' (see cholesky3), then H-inverse = (L-inv)' (Dinv) (L-inv) ** where D is diagonal and L is lower-triangular with ones on the diagonal, ** and L[1:nfrail, 1:nfrail] is the identity. ** The return parts are hmat = L[(nfrail+1):p, 1:p], hinv L-inverse[ same], ** and g=hdiag = D-inverse. See coxpenal.df for more. */ #include "survS.h" #include "survproto.h" SEXP survreg7(SEXP maxiter2, SEXP nvarx, SEXP y, SEXP ny2, SEXP covar2, SEXP wtx, SEXP offset2, SEXP beta2, SEXP nstratx, SEXP stratax, SEXP epsx, SEXP tolx, SEXP dist, SEXP dexpr, SEXP rho, SEXP ptype2, SEXP pdiag2, SEXP nfrail2, SEXP fgrp2, SEXP pexpr1, SEXP pexpr2) { /* local variables */ int i,j; int nvar, nvar2, nvar3, nstrat; int iter; double newlk =0; double (*dolik)(); /* will point to (*dolik) or survregc2 */ double x1, x2, x3, x4; double y1, y2, y3; int golden, goright; double newpen; /* pointers for the data regions of the input arguments */ double **covar; Sint *strat ; double *time2, *time1, *status; double *offset; Sint *fgrp; double *wt; /* copies of the scalar input arguments */ double eps, tol_chol; int n, maxiter, ny; int nfrail, ptype, pdiag; /* Variables allocated in this routine */ double *jdiag, *newbeta, *u; /* variables for the callback code */ SEXP coef1, coef2; double *cptr1=NULL, *cptr2 =NULL; /* stop a gcc warning */ double **JJ; SEXP z; double *zptr = NULL; /* structures and pointers for the returned list object */ SEXP out_iter, out_loglik, out_hmat, out_hinv, out_flag, out_beta; SEXP out_penalty; SEXP out_hdiag, out_u; double *loglik, *usave; double **hmat, **hinv, *beta, *hdiag; double *penalty; SEXP rlist, rlistnames; Sint *iter2, *flag; int nprotect; /* number of PROTECT calls that I have issued */ /* ** The only input arg that is rewritten is beta, so no need to duplicate */ maxiter = asInteger(maxiter2); n = LENGTH(wtx); ny = asInteger(ny2); nvar = asInteger(nvarx); offset = REAL(offset2); nstrat = asInteger(nstratx); strat = INTEGER(stratax); wt = REAL(wtx); eps = asReal(epsx); tol_chol= asReal(tolx); covar = dmatrix(REAL(covar2), n, nvar); nfrail = asInteger(nfrail2); ptype = asInteger(ptype2); pdiag = asInteger(pdiag2); fgrp = INTEGER(fgrp2); /* ** nvar = # of "real" x variables, found in the coefficient matrix ** nvar2= size of the dense portion of hmat = nvar + nstrat ** nvar3= #coefficients = nfrail + nvar2 ** nstrat= # of strata, where 0== fixed sigma */ nvar2 = nvar + nstrat; /* number of coefficients */ nvar3 = nvar2 + nfrail; /* ** Create the output variables */ PROTECT(out_beta = duplicate(beta2)); beta = REAL(out_beta); PROTECT(out_hmat = allocVector(REALSXP, nvar3*nvar2)); hmat = dmatrix(REAL(out_hmat), nvar3, nvar2); PROTECT(out_hinv = allocVector(REALSXP, nvar3*nvar2)); hinv = dmatrix(REAL(out_hinv), nvar3, nvar2); PROTECT(out_hdiag = allocVector(REALSXP, nvar3)); hdiag = REAL(out_hdiag); PROTECT(out_iter = allocVector(INTSXP, 1)); iter2 = INTEGER(out_iter); PROTECT(out_loglik = allocVector(REALSXP, 1)); loglik = REAL(out_loglik); PROTECT(out_flag = allocVector(INTSXP, 1)); flag = INTEGER(out_flag); PROTECT(out_u = allocVector(REALSXP, nvar3)); usave = REAL(out_u); /* the working vector 'u' gets destroyed in chsolve*/ PROTECT(out_penalty= allocVector(REALSXP, 1)); penalty = REAL(out_penalty); nprotect =9; /* Create the scratch vectors ** u = working version of score vector, overwritten with u H-inv during ** Newton steps ** usave = a copy of u, after each Newton step. Returned to the S ** parent routine, and also used to "backtrack" when we need to fail ** over to a Fisher step instead of an NR step */ newbeta = Calloc(LENGTH(beta2) + nvar3 + nfrail + nvar2*nvar3, double); jdiag = newbeta + length(beta2); u = jdiag + nfrail; JJ = dmatrix(u + nvar3, nvar3, nvar2); /* ** fixed scale parameters were tacked onto the end of beta at input ** copy them to to the end of newbeta as well ((*dolik) expects them) */ for (i=nvar; i1) { /* ** There is a penalty on the non-sparse terms ** Create the vector coef2 in the contained data frame ** (pexpr2 is implicitly a function of 'coef2') ** Since scale parameters are never panalized, only the first ** nvar of the coefficients are passed in. */ PROTECT(coef2 = allocVector(REALSXP, nvar)); defineVar(install("coef2"), coef2, rho); cptr2 = REAL(coef2); nprotect++; } /* ** Get the loglik, score, and hessian for the initial parameters */ *loglik = (*dolik)(n, nvar, nstrat, 0, beta, asInteger(dist), strat, offset, time1, time2, status, wt, covar, hmat, JJ, u, dexpr, rho, zptr, nfrail, fgrp, hdiag, jdiag); survpenal(0, nfrail, nvar, hmat, JJ, hdiag, jdiag, u, beta, penalty, ptype, pdiag, pexpr1, cptr1, pexpr2, cptr2, rho); *loglik += *penalty; for (i=0; i y3) { x4 = x3; x3 = x1; x1 = x3 - (x4-x3)/.618; y3 = y1; for (i=0; i y2) { /* toss away the interval from x1 to x2 */ x1=x2; x2=x3; x3 = .618*x4 + .382*x1; y2 =y3; for (i=0; i *loglik || y3 > *loglik) { /* Success - keep the better guess & compute derivatives */ if (y2 > y3) { for (i=0; i 1) *flag= 1000; /* no "non convergence" for 0 or 1 iter */ *iter2 = iter; /* ** Put together the return list */ alldone: *flag = cholesky3(hmat, nvar3, nfrail, hdiag, tol_chol); for (i=0; i>= survfitKM <- function(x, y, casewt=rep(1,length(x)), type=c('kaplan-meier', 'fleming-harrington', 'fh2'), error=c('greenwood', "aalen", "robust", "tsiatis"), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified'), start.time, new.time) { type <- match.arg(type) method <- match(type, c("kaplan-meier", "fleming-harrington", "fh2")) if (missing(error) & any(casewt != floor(casewt)) error <- 'robust' else error <- match.arg(error) error.int <- match(error, c("greenwood", "aalen", "robust", "tsiatis")) if (error.int==4) error.int <- 2 # these are synonyms conf.type <- match.arg(conf.type) conf.lower<- match.arg(conf.lower) if (is.logical(conf.int)) { # A common error is for users to use "conf.int = FALSE" # it's illegal, but allow it if (!conf.int) conf.type <- "none" conf.int <- .95 } if (!is.Surv(y)) stop("y must be a Surv object") if (!is.factor(x)) stop("x must be a factor") if (attr(y, 'type') != 'right' && attr(y, 'type') != 'counting') stop("Can only handle right censored or counting data") ny <- ncol(y) # Will be 2 for right censored, 3 for counting xlev <- levels(x) # Will supply names for the curves x <- as.numeric(x) # keep only the levels # Allow "new.time" as a synonym for start.time if (missing(start.time) && !missing(new.time)) start.time <- new.time if (!missing(start.time)) { n.all <- c(table(x)) # remember the original data size # remove any obs whose end time is <= start.time keep <- (y[,ny-1] >= start.time) if (all(keep==FALSE)) stop(paste("start.time =", start.time, "is greater than all time points.")) x <- x[keep] y <- y[keep,,drop=FALSE] #make sure y remains a matrix casewt <- casewt[keep] } n.used <- as.vector(table(x)) # This is for the printout nstrat <- length(n.used) # number of curves <> <> } @ The computation creates a list with one survival curve per element. Since the number of curves is usually small the outer "for" loop is no particular worry. One of the odder things we do is to use as.numeric(factor(y)) as the response variable instead of y itself. The problem is that if two times are within machine precision then [[unique]], [[factor]], [[table]] and [[==]] can all do different things. We have been burned by this in the past. <>= survlist <- vector("list", nstrat) xint <- as.numeric(x) for (i in 1:nstrat) { who <- which(xint==i) if (ny==2) { ftime <- factor(y[who,1]) indx <- order(ftime, -status[who]) #censors after deaths survlist[[i]] <- c(list(time=type.convert(levels(ftime), as.is=TRUE, dec=getOption("OutDec")), .Call("survkm1", as.integer(ftime[indx]), as.integer(y[who[indx],2]), as.double(wt[who[indx]]), method, error, length(levels(ftime))) } else { ftime <- factor(y[who, 1:2]) # Be careful that no one person has overlapping time intervals if (error==4) { <> } survlist[[i]] <- c(list(time=as.numeric(levels(ftime))), .Call("survkm2", as.integer(ftime), as.integer(y[who,3]), method, error, length(levels(ftime)), id[who])) } } @ indx <- order(id, Y[,2]) #ordered event times within subject indx1 <- c(NA, indx) #a pair of lagged indices indx2 <- c(indx, NA) same <- (id[indx1] == id[indx2] & !is.na(indx1) & !is.na(indx2)) #indx1, indx2= same id? if (any(same & X[indx1] != X[indx2])) { who <- 1 + min(which(same & X[indx1] != X[indx2])) stop("subject is in two different groups, id ", (id[indx1])[who]) } if (any(same & Y[indx1,2] != Y[indx2,1])) { who <- 1 + min(which(same & Y[indx1,2] != Y[indx2,1])) stop("gap in follow-up, id ", (id[indx1])[who]) } if (any(Y[,1] == Y[,2])) stop("cannot have start time == stop time") if (any(same & Y[indx1,3] == Y[indx2,3] & Y[indx1,3] !=0)) { who <- 1 + min(which(same & Y[indx1,1] != Y[indx2,2])) stop("subject changes to the same state, id ", (id[indx1])[who]) } if (any(same & weights[indx1] != weights[indx2])) { who <- 1 + min(which(same & weights[indx1] != weights[indx2])) stop("subject changes case weights, id ", (id[indx1])[who]) } Now for the real work using C routines. My standard for a variable named ``zed'' is to use zed2 for the S object and zed for the data part of the object; the latter is what the C code works with. <>= /* -*- c -*- */ #include #include "survS.h #include "survproto.h" SEXP survkm1(SEXP itime2, SEXP status2, SEXP wt2, SEXP method2, SEXP error2, SEXP ntime2 { int i, j, k; int n; /* number of observations */ /* Data passed in */ int ntime; /* number of unique times = length of output vectors */ int method, error; int *itime, *status; double *wt; /* ** output vectors */ SEXP nrisk2, nevent2, ncensor2, surv, cumhaz; double *nrisk, *nevent, *ncensor, *surv, *cumhaz; SEXP ievent2, icensor2; /* integer counts, in case of weights */ int *ievent, *icensor; const char *rnames[]={"nrisk", "nevent", "ncensor", "surv", "cumhaz", "std", "ievent", "icensor", ""}; /* ** Get copies of the input data */ ntime = asInteger(ntime2); method= asInteger(method2); error = asInteger(error2); itime = INTEGER(itime2); status= INTEGER(status2); wt = REAL(wt2); if (method==4) id = INTEGER(id2); n = LENGTH(itime); /* ** create output objects */ PROTECT(nrisk2 = allocVector(REALSXP, ntime)); nrisk = REAL(nrisk2); PROTECT(nevent2 = allocVector(REALSXP, ntime)); nevent = REAL(nevent2); PROTECT(ncensor2 = allocVector(REALSXP, ntime)); ncensor = REAL(ncensor2); PROTECT(surv2 = allocVector(REALSXP, ntime)); surv = REAL(surv2); PROTECT(cumhaz2 = allocVector(REALSXP, ntime)); cumhaz = REAL(cumhaz2); PROTECT(ievent2 = allocVector(INTSXP, ntime)); ievent = INTEGER(ivent2); PROTECT(icensor2 = allocVector(INTSXP, ntime)); icensor = INTEGER(icensor2); if (error=0) PROTECT(var2 = allocVector(REALSXP, 1));/* no std wanted */ else PROTECT(var2= allocVector(REALSXP, ntime)); var = REAL(var2); /* ** first pass, from largest time to smallest, count up ** number events, number at risk, number censored */ i = n -1; temp =0; /* accumulates number at risk */ for (j=ntime-1; j>=0; j--) { nevent[j] = 0; ievent[j]=0; ncensor[j]= 0; icensor[j]=0; ctime = itime[i]; /* current time of interest */ while(itime[i]== ctime && i>=0) { temp += wt[i]; if (status[i]==1){ nevent[j] += wt[i]; ievent[j]++; } else{ ncensor[j] += wt[i]; icensor[j]++; } i--; } nrisk[j] = temp; } /* ** Second pass, from smallest time to largest, accumulate ** the cumulative hazard and survival */ <> if (error >0) { <> } /* ** create the output structure */ PROTECT(rlist = mkNamed(VEXSXP, rnames)); SET_VECTOR_ELT(rlist, 0, nrisk2); SET_VECTOR_ELT(rlist, 1, nevent2); SET_VECTOR_ELT(rlist, 2, ncensor2); SET_VECTOR_ELT(rlist, 3, surv2); SET_VECTOR_ELT(rlist, 4, cumhaz2); SET_VECTOR_ELT(rlist, 5, var2); SET_VECTOR_ELT(rlist, 1, ievent2); SET_VECTOR_ELT(rlist, 2, icensor2); UNPROTECT(9); /*once there is NO chance of a memory allocation, we let go*/ return(rlist); } @ Let $Y_i(t)=1$ if observation $i$ is at risk at time $t$ and 0 otherwise, $s_i(t)$ be 1 at the point of an event for subject $i$ (status of 1), $w_i$ be the case weight for observation $i$, and $u_j$ $j=1,2, \ldots$ be the unique death times. Define \begin{center} \begin{tabular}{r@{=}rl} $r_j$& $\sum Y_i(u_j) w_i$ & weighted number at risk \\ $d_j$& $\sum s_i(u_j) w__i$ & weighted number of events \\ $c_j)$& $\sum (Y_i(u_j)-s_i(u_j)) w_i$ & weighted number of censored values \end{tabular} \end{center} For (start, stop] data there will also be $e_j$ which counts the number of subjects who enter at time $t$. Also, let $f_j$ be the total number of failures (deaths) at time $t$, not weighted. The cumulative hazard estimates are the Nelson-Aalen-Breslow (same estimate, three different papers in three places) or the Fleming-Harrington. \begin{align*} \Lambda_A(t) &\ \sum{u_j \le t} d_j/r_j \\ \Lambda_{FH}(t) &= \sum{u_j \le t} \frac{d_j} {(1/f_j) \sum_{k=0}^{f_j-1} (r_j - kd_j/f_j)} \end{align*} To understand the Fleming-Harrington estimate, suppose that at some time point we had three deaths out of 10 at risk. The Aalen estimate gives a hazard estimate of 3/10. The FH estimate assumes that the deaths didn't actually all happen at once, even though rounding in the data collection process makes it appear that way, so the better estimate is 1/10 + 1/9 + 1/8. The third person to die, whoever that was, would have had only 8 at risk when thier event happened. The estimate of survival is either the Kaplan-Meier or the exponential of the hazard. \begin{equation*} KM(t) = \prod_{u_j \le t} \frac{r_j - d_j}{r_j} \end{equation*} <>= tempc =0; /*accumulates the chaz */ temps =1; /*accumulates the survival */ tempv =0; /*accumulates the variance */ if (method==1) { /*KM survival and Aalen hazard, 99% of the calls */ for (j=0; j0) { temps *= (nrisk[j] - nevent[j])/nevent[j]; tempc += nevent[j]/nrisk[j]; } surv[j] = temps; chaz[j] = tempc; } } else if (method==2) { /*Aalen hazard and exponent for survival */ for (j=0; j0) tempc += nevent[j]/nrisk[j]; chaz[j] = tempc; surv[j] = exp(-tempc); } } else { /* FH hazard, rarest call */ for (j=0; j0) { temp <- nevent[j]/ievent[j]; for (k=0; i>= temp =0; if (error==1) { /* Greenwood */ for (j=0; j0) temp += nevent[j]/(nrisk[j] * (nrisk[j]-nevent[j])); var[j] = temp; } else if (error==2) { /* Aalen */ for (j=0; j0) temp += nevent[j]/(nrisk[j] * nrisk[j]); var[j] = temp; } else { inf = (double *) Ralloc(n, sizeof(double)); /* scratch space */ for (i=0; i0) { /*nothing changes at non-event times */ ctime = time[i]; /*current event time */ if (method <3) { /* variance of the Aalen hazard */ for (;i>= /* -*- c -*- */ #include #include "survS.h #include "survproto.h" SEXP survkm2(SEXP itime2, SEXP status2, SEXP wt2, SEXP method2, SEXP error2, SEXP ntime2, SEXP id2, SEXP nid2, SEXP sort1) { int i, j, k; int n; /* number of observations */ int nid; /* number of unique id values */ /* Data passed in */ int ntime; /* number of unique times = length of output vectors */ int method, error; int *start, *stop, *status, *id; double *wt; /* ** output vectors */ SEXP nrisk2, nevent2, ncensor2, surv, cumhaz; double *nrisk, *nevent, *ncensor, *surv, *cumhaz; SEXP ievent2, icensor2; /* integer counts, in case of weights */ int *ievent, *icensor; const char *rnames[]={"nrisk", "nevent", "ncensor", "surv", "cumhaz", "std", "ievent", "icensor", ""}; /* ** Get copies of the input data */ ntime = asInteger(ntime2); n = LENGTH(itime); nid= asInteger(nid); method= asInteger(method2); error = asInteger(error2); start = INTEGER(itime2); stop = start + n; status= INTEGER(status2); wt = REAL(wt2); if (method==4) id = INTEGER(id2); n = LENGTH(itime); id = INTEGER(id2); nid= asInteger(nid); /* ** create output objects */ PROTECT(nrisk2 = allocVector(REALSXP, ntime)); nrisk = REAL(nrisk2); PROTECT(nevent2 = allocVector(REALSXP, ntime)); nevent = REAL(nevent2); PROTECT(ncensor2 = allocVector(REALSXP, ntime)); ncensor = REAL(ncensor2); PROTECT(surv2 = allocVector(REALSXP, ntime)); surv = REAL(surv2); PROTECT(cumhaz2 = allocVector(REALSXP, ntime)); cumhaz = REAL(cumhaz2); PROTECT(ievent2 = allocVector(INTSXP, ntime)); ievent = INTEGER(ivent2); PROTECT(icensor2 = allocVector(INTSXP, ntime)); icensor = INTEGER(icensor2); if (error=0) PROTECT(var2 = allocVector(REALSXP, 1));/* no std wanted */ else PROTECT(var2= allocVector(REALSXP, ntime)); var = REAL(var2); /* ** first pass, from largest time to smallest, count up ** number events, number at risk, number censored */ i = n -1; temp =0; /* accumulates number at risk */ for (j=ntime-1; j>=0; j--) { nevent[j] = 0; ievent[j]=0; ncensor[j]= 0; icensor[j]=0; ctime = itime[i]; /* current time of interest */ while(itime[i]== ctime && i>=0) { temp += wt[i]; if (status[i]==1){ nevent[j] += wt[i]; ievent[j]++; } else{ ncensor[j] += wt[i]; icensor[j]++; } i--; } nrisk[j] = temp; } /* ** Second pass, from smallest time to largest, accumulate ** the cumulative hazard and survival */ <> if (error >0) { <> } /* ** create the output structure */ PROTECT(rlist = mkNamed(VEXSXP, rnames)); SET_VECTOR_ELT(rlist, 0, nrisk2); SET_VECTOR_ELT(rlist, 1, nevent2); SET_VECTOR_ELT(rlist, 2, ncensor2); SET_VECTOR_ELT(rlist, 3, surv2); SET_VECTOR_ELT(rlist, 4, cumhaz2); SET_VECTOR_ELT(rlist, 5, var2); SET_VECTOR_ELT(rlist, 1, ievent2); SET_VECTOR_ELT(rlist, 2, icensor2); UNPROTECT(9); /*once there is NO chance of a memory allocation, we let go*/ return(rlist); } @ survival/noweb/Makefile0000644000175100001440000000403512545056257014732 0ustar hornikusersPARTS = main.Rnw \ coxph.Rnw \ exact.nw \ agreg.Rnw \ coxsurv.Rnw \ coxsurv2.Rnw \ predict.coxph.Rnw \ concordance.Rnw \ survexp.Rnw \ pyears.Rnw \ residuals.survreg.Rnw \ survfit.Rnw \ survfitCI.Rnw \ msurv.nw \ survfitms.Rnw \ plot.Rnw \ tmerge.Rnw\ tail # coxdetail.nw SFUN = agreg.fit.R \ agsurv.R \ coxph.R \ model.matrix.coxph.R \ plot.survfit.R \ predict.coxph.R \ pyears.R \ residuals.survreg.S\ survConcordance.R \ survConcordance.fit.R \ survexp.R \ survfit.R \ survfitCI.R \ survfit.coxph.R \ survfitcoxph.fit.R \ survfitms.R\ tmerge.R CFUN = agsurv4.c agsurv5.c concordance1.c coxcount1.c \ agfit4.c \ coxexact.c \ survfitci.c # coxdetail2.c RDIR = ../R RFUN = $(SFUN:%=$(RDIR)/%) CFUN2= $(CFUN:%=../src/%) DOCDIR= ../inst/doc all: noweb.sty doc fun doc: all.pdf all.pdf: all.tex noweb.sty texi2dvi --pdf all.tex all.nw: $(PARTS) cat $(PARTS) > all.nw all.tex: all.nw echo "library(noweb); noweave('all.nw')" | $(R_HOME)/bin$(R_ARCH_BIN)/R --vanilla --slave # noweave -delay all.nw > all.tex $(SFUN): all.nw $(CFUN): all.nw $(CFUN2): all.nw $(RFUN): all.nw .PHONY: fun clean doc fun: $(RFUN) $(CFUN2) noweb.sty test: $(RFUN) echo $(RFUN) %.R: echo "# Automatically generated from all.nw using noweb" > $@ echo "require(noweb); notangle('all.nw', target='$(*F)', out='zz')" | $(R_HOME)/bin$(R_ARCH_BIN)/R --vanilla --slave cat zz >> $@ rm zz %.S: echo "# Automatically generated from all.nw using noweb" > $@ echo "require(noweb); notangle('all.nw', target='$(*F)', out='zz')" | $(R_HOME)/bin$(R_ARCH_BIN)/R --vanilla --slave cat zz >> $@ rm zz %.c: echo "/* Automatically generated from all.nw using noweb */" > $@ echo "require(noweb); notangle('all.nw', target='$(*F)', out='zz')" | $(R_HOME)/bin$(R_ARCH_BIN)/R --vanilla --slave cat zz >> $@ rm zz clean: -rm all.nw all.log all.aux all.toc all.tex all.bbl all.blg all.out -rm noweb.sty noweb.sty: echo 'library(noweb); data(noweb); cat(noweb.sty, sep="\n", file="noweb.sty")' | $(R_HOME)/bin$(R_ARCH_BIN)/R --vanilla --slave survival/noweb/figures/0000755000175100001440000000000011773425616014735 5ustar hornikuserssurvival/noweb/figures/balance.r0000644000175100001440000000232711773425616016511 0ustar hornikusers# The balance figure for the concondance document btree <- function(n) { tfun <- function(n, id, power) { if (n==1) id else if (n==2) c(2*id, id) else if (n==3) c(2*id, id, 2*id+1) else { nleft <- if (n== power*2) power else min(power-1, n-power/2) c(tfun(nleft, 2*id, power/2), id, tfun(n-(nleft+1), 2*id +1, power/2)) } } tfun(n, 1, 2^(floor(logb(n-1,2)))) } set.seed(12345) temp <- sort(unique(floor(runif(40,0,30)))[1:13])/10 indx <- btree(13) xpos <- 1:15 xpos[4:7] <- tapply(xpos[8:15], rep(1:4, each=2), mean) xpos[2:3] <- tapply(xpos[4:7], rep(1:2, each=2),mean) xpos[1] <- mean(xpos[2:3]) ypos <- rep(4:1, c(1,2,4,8)) pdf('balance.pdf', height=5, width=7) par(mar=c(1,0,0,1) + .1) plot(xpos, ypos, type='n', xaxt='n', yaxt='n', bty='n', xlab="", ylab="") temp2 <- c(13,7,5,3,3,3,1,1,1,1,1,1,1) #text(xpos[indx], ypos[indx], paste(temp, " (", temp2[indx], ")", sep='')) text(xpos[indx], ypos[indx], temp) delta=.1 for (i in 1:6) { segments(xpos[i]-delta, ypos[i]-delta, xpos[2*i]+delta, ypos[2*i]+delta) segments(xpos[i]+delta, ypos[i]-delta, xpos[2*i+1]-delta, ypos[2*i+1] +delta) } dev.off() survival/noweb/figures/balance.pdf0000644000175100001440000000467011773342756017027 0ustar hornikusers%PDF-1.4 %âãÏÓ\r 1 0 obj << /CreationDate (D:20110218110759) /ModDate (D:20110218110759) /Title (R Graphics Output) /Producer (R 2.12.0) /Creator (R) >> endobj 2 0 obj << /Type /Catalog /Pages 3 0 R >> endobj 5 0 obj << /Type /Page /Parent 3 0 R /Contents 6 0 R /Resources 4 0 R >> endobj 6 0 obj << /Length 7 0 R >> stream 1 J 1 j q Q q Q q 1.44 15.84 486.72 342.72 re W n BT 0.000 0.000 0.000 rg /F2 1 Tf 12.00 0.00 -0.00 12.00 16.13 24.43 Tm (0) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 48.32 130.09 Tm (1) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 80.51 24.41 Tm (7) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 112.70 235.98 Tm (8) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 141.56 24.43 Tm (16) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 173.75 130.21 Tm (19) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 205.94 24.32 Tm (22) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 238.13 341.76 Tm (25) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 270.32 24.43 Tm (36) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 302.51 130.21 Tm (38) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 334.70 24.43 Tm (43) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 366.89 235.87 Tm (44) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 431.27 130.21 Tm (49) Tj ET 0.000 0.000 0.000 RG 0.75 w [] 0 d 1 J 1 j 10.00 M 238.36 335.29 m 122.48 250.67 l S 251.24 335.29 m 367.12 250.67 l S 109.60 229.51 m 58.10 144.89 l S 122.48 229.51 m 173.98 144.89 l S 367.12 229.51 m 315.62 144.89 l S 380.00 229.51 m 431.50 144.89 l S 45.22 123.73 m 25.90 39.11 l S 58.10 123.73 m 77.41 39.11 l S 173.98 123.73 m 154.67 39.11 l S 186.86 123.73 m 206.17 39.11 l S 302.74 123.73 m 283.43 39.11 l S 315.62 123.73 m 334.93 39.11 l S Q endstream endobj 7 0 obj 1327 endobj 3 0 obj << /Type /Pages /Kids [ 5 0 R ] /Count 1 /MediaBox [0 0 504 360] >> endobj 4 0 obj << /ProcSet [/PDF /Text] /Font <> /ExtGState << >> >> endobj 8 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi /grave /acute /circumflex /tilde /macron /breve /dotaccent /dieresis /.notdef /ring /cedilla /.notdef /hungarumlaut /ogonek /caron /space] >> endobj 9 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 8 0 R >> endobj xref 0 10 0000000000 65535 f 0000000021 00000 n 0000000164 00000 n 0000001693 00000 n 0000001776 00000 n 0000000213 00000 n 0000000293 00000 n 0000001673 00000 n 0000001857 00000 n 0000002114 00000 n trailer << /Size 10 /Info 1 0 R /Root 2 0 R >> startxref 2210 %%EOF survival/noweb/agreg.Rnw0000644000175100001440000006411412533651334015045 0ustar hornikusers\subsection{Anderson-Gill fits} When the survival data set has (start, stop] data a couple of computational issues are added. A primary one is how to do this compuation efficiently. At each event time we need to compute 3 quantities, each of them added up over the current risk set. \begin{itemize} \item The weighted sum of the risk scores $\sum w_i r_i$ where $r_i = \exp(\eta_i)$ and $\eta_i = x_{i1}\beta_1 + x_{i2}\beta_2 +\ldots$ is the current linear predictor. \item The weighted mean of the covariates $x$, with weight $w_i r_i$. \item The weighted variance-covariance matrix of $x$. \end{itemize} The current risk set at some event time $t$ is the set of all (start, stop] intervals that overlap $t$, and are part of the same strata. The round/square brackets in the prior sentence are important: for an event time $t=20$ the interval $(5,20]$ is considered to overlap $t$ and the interval $(20,55]$ does not overlap $t$. Our routine for the simple right censored Cox model computes these efficiently by keeping a cumulative sum. Starting with the longest survival move backwards through time, adding subjects to the sums as we go. For this routine subjects we also move backwards through time, and subjects both enter and leave the sums. The code below creates two sort indices, one orders the data by reverse stop time and the other by reverse start time. The fit routine is called by the coxph function with arguments \begin{description} \item[x] matrix of covariates \item[y] three column matrix containing the start time, stop time, and event for each observation \item[strata] for stratified fits, the strata of each subject \item[offset] the offset, usually a vector of zeros \item[init] initial estimate for the coefficients \item[control] results of the coxph.control function \item[weights] case weights, often a vector of ones. \item[method] how ties are handled: 1=Breslow, 2=Efron \item[rownames] used to label the residuals \end{description} <>= agreg.fit <- function(x, y, strata, offset, init, control, weights, method, rownames) { n <- nrow(y) nvar <- ncol(x) start <- y[,1] stopp <- y[,2] event <- y[,3] if (all(event==0)) stop("Can't fit a Cox model with 0 failures") # Sort the data (or rather, get a list of sorted indices) # For both stop and start times, the indices go from last to first if (length(strata)==0) { sort.end <- order(-stopp, event) -1L #indices start at 0 for C code sort.start<- order(-start) -1L newstrat <- n } else { sort.end <- order(strata, -stopp, event) -1L sort.start<- order(strata, -start) -1L newstrat <- cumsum(table(strata)) } if (missing(offset) || is.null(offset)) offset <- rep(0.0, n) if (missing(weights)|| is.null(weights))weights<- rep(1.0, n) else if (any(weights<=0)) stop("Invalid weights, must be >0") else weights <- as.vector(weights) if (is.null(nvar) || nvar==0) { # A special case: Null model. Just return obvious stuff # To keep the C code to a small set, we call the usual routines, but # with a dummy X matrix and 0 iterations nvar <- 1 x <- matrix(as.double(1:n), ncol=1) #keep the .C call happy maxiter <- 0 nullmodel <- TRUE if (length(init) !=0) stop("Wrong length for inital values") init <- 0.0 #dummy value to keep a .C call happy (doesn't like 0 length) } else { nullmodel <- FALSE maxiter <- control$iter.max if (is.null(init)) init <- rep(0., nvar) if (length(init) != nvar) stop("Wrong length for inital values") } # the returned value of agfit$coef starts as a copy of init, so make sure # is is a vector and not a matrix; as.double does so. # Solidify the storage mode of other arguments storage.mode(y) <- storage.mode(x) <- "double" storage.mode(offset) <- storage.mode(weights) <- "double" storage.mode(newstrat) <- "integer" agfit <- .Call(Cagfit4, y, x, newstrat, weights, offset, as.double(init), sort.end, sort.start, as.integer(method=="efron"), as.integer(maxiter), as.double(control$eps), as.double(control$toler.chol), as.integer(1)) # internally rescale <> <> } @ Upon return we need to clean up two simple things. The first is that if any of the covariates were redudant then this will be marked by zeros on the diagonal of the variance matrix. Replace these coefficients and their variances with NA. The second is to post a warning message about possible infinite coefficients. The algorithm for determining this is unreliable, unfortunately. Sometimes coefficients are marked as infinite when the solution is not tending to infinity (usually associated with a very skewed covariate), and sometimes one that is tending to infinity is not marked. Que sera sera. <>= var <- matrix(agfit$imat,nvar,nvar) coef <- agfit$coef if (agfit$flag < nvar) which.sing <- diag(var)==0 else which.sing <- rep(FALSE,nvar) infs <- abs(agfit$u %*% var) if (maxiter >1) { if (agfit$iter > maxiter) warning("Ran out of iterations and did not converge") else { infs <- ((infs > control$eps) & infs > control$toler.inf*abs(coef)) if (any(infs)) warning(paste("Loglik converged before variable ", paste((1:nvar)[infs],collapse=","), "; beta may be infinite. ")) } } @ The last of the code is very standard. Compute residuals and package up the results. <>= lp <- as.vector(x %*% coef + offset - sum(coef *agfit$means)) score <- as.double(exp(lp)) resid <- .Call(Cagmart3, y, score, weights, newstrat, cbind(sort.end, sort.start), as.integer(method=='efron')) names(resid) <- rownames if (nullmodel) { list(loglik=agfit$loglik[2], linear.predictors = offset, residuals = resid, method= c("coxph.null", 'coxph') ) } else { names(coef) <- dimnames(x)[[2]] coef[which.sing] <- NA concordance <- survConcordance.fit(y, lp, strata, weights) list(coefficients = coef, var = var, loglik = agfit$loglik, score = agfit$sctest, iter = agfit$iter, linear.predictors = as.vector(lp), residuals = resid, means = agfit$means, concordance = concordance, method= 'coxph') } @ The details of the C code contain the more challenging part of the computations. It starts with the usual dull stuff. My standard coding style for a variable zed to to use [[zed2]] as the variable name for the R object, and [[zed]] for the pointer to the contents of the object, i.e., what the C code will manipulate. For the matrix objects I make use of ragged arrays, this simply allows for reference to the i,j element as \code{cmat[i][j]} and makes for more readable code. <>= #include #include "survS.h" #include "survproto.h" SEXP agfit4(SEXP surv2, SEXP covar2, SEXP strata2, SEXP weights2, SEXP offset2, SEXP ibeta2, SEXP sort12, SEXP sort22, SEXP method2, SEXP maxiter2, SEXP eps2, SEXP tolerance2, SEXP doscale2) { int i,j,k,person; int indx2, istrat, p; int ksave, nrisk, ndeath; int nused, nvar; double **covar, **cmat, **imat; /*ragged array versions*/ double *a, *oldbeta, *maxbeta; double *scale; double *a2, **cmat2; double *eta; double denom, zbeta, risk; double time; double temp, temp2; double newlk =0; int halving; /*are we doing step halving at the moment? */ double tol_chol, eps; double meanwt; int itemp, deaths; double efron_wt, d2, meaneta; /* inputs */ double *start, *stop, *event; double *weights, *offset; int *sort1, *sort2, maxiter; int *strata; double method; /* saving this as double forces some double arithmetic */ int doscale; /* returned objects */ SEXP imat2, means2, beta2, u2, loglik2; double *beta, *u, *loglik, *means; SEXP sctest2, flag2, iter2; double *sctest; int *flag, *iter; SEXP rlist; static const char *outnames[]={"coef", "u", "imat", "loglik", "means", "sctest", "flag", "iter", ""}; int nprotect; /* number of protect calls I have issued */ /* get sizes and constants */ nused = nrows(covar2); nvar = ncols(covar2); method= asInteger(method2); eps = asReal(eps2); tol_chol = asReal(tolerance2); maxiter = asInteger(maxiter2); doscale = asInteger(doscale2); /* input arguments */ start = REAL(surv2); stop = start + nused; event = stop + nused; weights = REAL(weights2); offset = REAL(offset2); sort1 = INTEGER(sort12); sort2 = INTEGER(sort22); strata = INTEGER(strata2); /* ** scratch space ** nvar: a, a2, newbeta, maxbeta, scale ** nvar*nvar: cmat, cmat2 ** n: eta */ eta = (double *) R_alloc(nused + 5*nvar + 2*nvar*nvar, sizeof(double)); a = eta + nused; a2 = a +nvar; maxbeta = a2 + nvar; scale = maxbeta + nvar; oldbeta = scale + nvar; /* ** Set up the ragged arrays ** covar2 might not need to be duplicated, even though ** we are going to modify it, due to the way this routine was ** was called. In this case NAMED(covar2) will =0 */ PROTECT(imat2 = allocVector(REALSXP, nvar*nvar)); nprotect =1; if (NAMED(covar2)>0) { PROTECT(covar2 = duplicate(covar2)); nprotect++; } covar= dmatrix(REAL(covar2), nused, nvar); imat = dmatrix(REAL(imat2), nvar, nvar); cmat = dmatrix(oldbeta+ nvar, nvar, nvar); cmat2= dmatrix(oldbeta+ nvar + nvar*nvar, nvar, nvar); /* ** create the output structures */ PROTECT(rlist = mkNamed(VECSXP, outnames)); nprotect++; beta2 = SET_VECTOR_ELT(rlist, 0, duplicate(ibeta2)); beta = REAL(beta2); u2 = SET_VECTOR_ELT(rlist, 1, allocVector(REALSXP, nvar)); u = REAL(u2); SET_VECTOR_ELT(rlist, 2, imat2); loglik2 = SET_VECTOR_ELT(rlist, 3, allocVector(REALSXP, 2)); loglik = REAL(loglik2); means2 = SET_VECTOR_ELT(rlist, 4, allocVector(REALSXP, nvar)); means = REAL(means2); sctest2 = SET_VECTOR_ELT(rlist, 5, allocVector(REALSXP, 1)); sctest = REAL(sctest2); flag2 = SET_VECTOR_ELT(rlist, 6, allocVector(INTSXP, 1)); flag = INTEGER(flag2); iter2 = SET_VECTOR_ELT(rlist, 7, allocVector(INTSXP, 1)); iter = INTEGER(iter2); /* ** Subtract the mean from each covar, as this makes the variance ** computation much more stable */ temp2 =0; for (i=0; i0) temp = temp2/temp; else temp = 1.0; /* rare case of a constant covariate */ scale[i] = temp; for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } else { /* scaling is only turned off during debugging still, cover the case */ for (person=0; person maxbeta[i]) maxbeta[i] = fabs(covar[i][person]); } } } if (doscale ==1) { for (i=0; i population ** of the earth, any beta*x over 20 is a silly relative risk for a Cox ** model, however. ** We want to cut off huge values, but not take action very often since ** doing so can mess up the iteration in general. ** One of the case-cohort papers suggests using anoffset of -100 to ** indicate "no risk", meaning that x*beta values of 50-100 can occur ** in "ok" data sets. Compromise. */ for (i=0; i> <> } @ As we walk through the risk sets observations are both added and removed from a set of running totals. In order to avoid catastrophic cancellation we need to make sure that the terms being added are of modest size and that the weights are not extreme. We have 6 running totals: \begin{itemize} \item sum of the weights, denom = $\sum w_i r_i$ \item totals for each covariate a[j] = $\sum w_ir_i x_{ij}$ \item totals for each covariate pair cmat[j,k]= $\sum w_ir_i x_{ij} x_{ik}$ \item the same three quantities, but only for times that are exactly tied with the current death time, named efron\_wt, a2, cmat2. These are used for the Efron approximation. \end{itemize} The three primary quantities for the Cox model are the log-likelihood $L$, the score vector $U$ and the Hessian matrix $H$. \begin{align*} L &= \sum_i w_i \delta_i \left[r_i - \log(d(t)) \right] \\ d(t) &= \sum_j w_j r_j Y_j(t) \\ U_k &= \sum_i w_i \delta_i \left[ (X_{ik} - \mu_k(t_i)) \right] \\ \mu_k(t) &= \frac{\sum_j w_j r_j Y_j(t) X_{jk}} {d(t)} \\ H_{kl} &= \sum_i w_i \delta_i V_{kl}(t_i) \\ V_{kl}(t) &= \frac{\sum_j w_j r_j Y_j(t) [X_{jk} - \mu_k(t)] [X_{jl}- \mu_l(t)]} {d(t)} \\ &= \frac{\sum_j w_j r_j Y_j(t) X_{jk}X_{jl}} {d(t)} - d(t) \mu_k(t) \mu_l(t) \end{align*} In the above $\delta_i =1$ for an event and 0 otherwise, $w_i$ is the per subject weight, and $Y_i(t)$ is 1 if observation $i$ is at risk at time $t$. The vector $\mu(t)$ is the weighted mean of the covariates at time $t$ using a weight of $w r Y(t)$ for each subject, and $V(t)$ is the weighted variance matrix of $X$ at time $t$. Tied deaths and the Efron approximation add a small complication to the formula. Say there are three tied deaths at some particular time $t$. When calculating the denominator $d(t)$, mean $\mu(t)$ and variance $V(t)$ at that time the inclusion value $Y_i(t)$ is 0 or 1 for all other subjects, as usual, but for the three tied deaths Y(t) is taken to be 1 for the first death, 2/3 for the second, and 1/3 for the third. The idea is that if the tied death times were randomly broken by adding a small random amount then each of these three would be in the first risk set, have 2/3 chance of being in the second, and 1/3 chance of being in the risk set for the third death. The variance formula is stable if $\mu$ is small relative to the total variance. This is guarranteed by subtracting the mean from each covariate before any other computations are performed. Weighted sums can still be unstable if the weights get out of hand. Because of the exponential $r_i = exp(\eta_i)$ the original centering of the $X$ matrix may not be enough. A particular example was a data set on hospital adverse events with ``number of nurse shift changes to date'' as a time dependent covariate. At any particular time point the covariate varied only by $\pm 3$ between subjects (weekends often use 12 hour nurse shifts instead of 8 hour). The regression coefficient was around 1 and the data duration was 11 weeks (about 200 shifts) so that $eta$ values could be over 100 even after centering. We keep a time dependent average of $\eta$ and renorm the weights as necessary. Since it would be possible for a malicious user to have a stratified model with mean x=1 in one strata and 1000 in another, which would also defeat the use of the overall centering, this check is done per strata. The last numerical problem is when one or more coefficients gets too large. This can lead to numerical difficulty based on a small number of observations or even on a single large outlier. This occassionally happens when a coefficient is tending to infinity, but is more often due to a very bad step in the intermediate Newton-Raphson path. We use a cutpoint of $\beta *{\rm std}(x) < 23$, where the standard deviation is the average std of $x$ within a risk set. The rationale is that exp(23) is greater than the current world population, so such a coefficient corresponds to a between subject relative risk that is larger than any imaginable. <>= for (i=0; i=time, ** and are thus potential members of the risk set. If 'indx2' =9, ** that means that 9 subjects have start >=time and thus are NOT part ** of the risk set. (stop > start for each subject guarrantees that ** the 9 are a subset of the 27). ** Basic algorithm: move 'person' forward, adding the new subject into ** the risk set. If this is a new, unique death time, take selected ** old obs out of the sums, add in obs tied at this time, then ** add terms to the loglik, etc. */ istrat=0; indx2 =0; denom =0; meaneta =0; nrisk =0; newlk =0; for (person=0; person (nrisk *110)) { meaneta = meaneta/nrisk; for (i=0; i>= /* First iteration, which has different ending criteria */ <> loglik[0] = newlk; /* save the loglik for iteration zero */ loglik[1] = newlk; /* Calculate the score test */ for (i=0; i> } else { /* Update beta for the next iteration ** Never complain about convergence on this first step or impose step ** halving. That way someone can force one iter at a time. */ for (i=0; i>= /* main loop */ halving =0 ; /* =1 when in the midst of "step halving" */ for (*iter=1; *iter<= maxiter; (*iter)++) { <> *flag = cholesky2(imat, nvar, tol_chol); if (fabs(1-(loglik[1]/newlk))<= eps && halving==0){ /* all done */ <> } if (*iter < maxiter) { /*update beta */ if (newlk < loglik[1]) { /*it is not converging ! */ halving =1; for (i=0; i maxbeta[i]) beta[i] = maxbeta[i]; else if (beta[i] < -maxbeta[i]) beta[i] = -maxbeta[i]; } } } R_CheckUserInterrupt(); /* be polite -- did the user hit cntrl-C? */ } /*return for another iteration */ @ Save away the final bits, compute the inverse of imat and symmetrize it, release memory and return. <>= loglik[1] = newlk; chinv2(imat, nvar); for (i=0; i>= #include "survS.h" <> SEXP survfitci(SEXP ftime2, SEXP sort12, SEXP sort22, SEXP ntime2, SEXP status2, SEXP cstate2, SEXP wt2, SEXP id2, SEXP p2, SEXP sefit2) { <> <> <> } @ Arguments to the routine are the following. For an R object ``zed'' I use the convention of [[zed2]] to refer to the object and [[zed]] to the contents of the object. \begin{description} \item[ftime] A two column matrix containing the entry and exit times for each subject. \item[sort1] Order vector for the entry times. The first element of sort1 points to the first entry time, etc. \item[sort2] Order vector for the event times. \item[ntime] Number of unique event time values. This fixes the size of the output arrays. \item[status] Status for each observation. 0= censored \item[cstate] The initial state for each subject, which will be updated during computation to always be the current state. \item[wt] Case weight for each observation. \item[id] The subject id for each observation. \item[p] The initial distribution of states. This will be updated during computation to be the current distribution. \item[sefit] If 1 then do the se compuatation, otherwise forget it. \end{description} The local dmatrix2 function makes it easier to declare ragged arrays, which allows for the nice \verb!x[i][j]! notation for arrays. <>= /* allocate a ragged array of a given number of rows and columns */ static double **dmatrix2(int nrow, int ncol) { int i; double **mat; double *d; mat = (double **) R_alloc(nrow, sizeof(double *)); d = (double *) R_alloc(nrow*ncol, sizeof(double)); for (i=0; i>= int i, j, k, kk; /* generic loop indices */ int ck, itime, eptr; /*specific indices */ int ctime; /*current time of interest, in the main loop */ int nprotect; /* number of protect calls issued */ int oldstate, newstate; /*when changing state */ double temp, *temp2; /* scratch */ double *p; /* current prevalence vector */ double **hmat; /* hazard matrix at this time point */ double **umat; /* per subject leverage at this time point */ int *atrisk; /* 1 if the subject is currently at risk */ int *ns; /* number curently in each state */ double *ws; /* weighted count of number state */ double *wtp; /* case weights indexed by subject */ double wevent; /* weighted number of events at current time */ int nstate; /* number of states */ int n, nperson; /*number of obs, subjects*/ double **chaz; /* cumulative hazard matrix */ /* pointers to the R variables */ int *sort1, *sort2; /*sort index for entry time, event time */ int *entry,* etime; /*entry time, event time */ int ntime; /* number of unique event time values */ int *status; /*0=censored, 1,2,... new states */ int *cstate; /* current state for each subject */ double *wt; /* weight for each observation */ int *id; /* for each obs, which subject is it */ int sefit; /* returned objects */ SEXP rlist; /* the returned list and variable names of same */ const char *rnames[]= {"nrisk","nevent","ncensor", "prev", "cumhaz", "var", ""}; SEXP pmat2, vmat2, cumhaz2; /*list components */ SEXP nevent2, ncensor2, nrisk2; double *pmat, *vmat, *cumhaz; int *ncensor, *nrisk, *nevent; @ Now set up pointers for all of the R objects sent to us. The two that will be updated need to be replaced by duplicates. <>= ntime= asInteger(ntime2); nperson = LENGTH(cstate2); n = LENGTH(sort12); PROTECT(cstate2 = duplicate(cstate2)); cstate = INTEGER(cstate2); entry= INTEGER(ftime2); etime= entry + n; sort1= INTEGER(sort12); sort2= INTEGER(sort22); status= INTEGER(status2); wt = REAL(wt2); id = INTEGER(id2); PROTECT(p2 = duplicate(p2)); /*copy of initial prevalence */ p = REAL(p2); nstate = LENGTH(p2); /* number of states */ sefit = asInteger(sefit2); /* allocate space for the output objects */ PROTECT(pmat2 = allocMatrix(REALSXP, nstate, ntime)); pmat = REAL(pmat2); if (sefit >0) PROTECT(vmat2 = allocMatrix(REALSXP, nstate, ntime)); else PROTECT(vmat2 = allocMatrix(REALSXP, 1, 1)); /* dummy object */ vmat = REAL(vmat2); PROTECT(nevent2 = allocVector(INTSXP, ntime)); nevent = INTEGER(nevent2); PROTECT(ncensor2= allocVector(INTSXP, ntime)); ncensor = INTEGER(ncensor2); PROTECT(nrisk2 = allocMatrix(INTSXP, nstate, ntime)); nrisk = INTEGER(nrisk2); PROTECT(cumhaz2= allocVector(REALSXP, nstate*nstate*ntime)); cumhaz = REAL(cumhaz2); nprotect = 8; /* allocate space for scratch vectors */ ws = (double *) R_alloc(2*nstate, sizeof(double)); temp2 = ws + nstate; ns = (int *) R_alloc(nstate, sizeof(int)); atrisk = (int *) R_alloc(nperson, sizeof(int)); wtp = (double *) R_alloc(nperson, sizeof(double)); hmat = (double**) dmatrix2(nstate, nstate); if (sefit >0) umat = (double**) dmatrix2(nperson, nstate); chaz = (double**) dmatrix2(nstate, nstate); /* R_alloc does not zero allocated memory */ for (i=0; i>= itime =0; /*current time index, for output arrays */ eptr = 0; /*index to sort1, the entry times */ for (i=0; i> <> /* Take the current events and censors out of the risk set */ for (; i0) cstate[id[j]] = status[j]-1; /*new state */ atrisk[id[j]] =0; } else break; } itime++; } @ The key variables for the computation are the matrix $H$ and the current prevalence vector $P$. $H$ is created anew at each unique time point. Row $j$ of $H$ concerns everyone in state $j$ just before the time point, and contains the transitions at that time point. So the $jk$ element is the (weighted) fraction who change from state $j$ to state $k$, and the $jj$ element the fraction who stay put. Each row of $H$ by definition sums to 1. If no one is in the state then the $jj$ element is set to 1. A second version which we call H2 has 1 subtracted from each diagonal and so that the row sums are 0, we go back and forth depending on which is needed at the moment. If there are no events at this time point $P$ and $U$ do not update. <>= for (j=0; j0) { newstate = status[k] -1; /* 0 based subscripts */ oldstate = cstate[id[k]]; nevent[itime]++; wevent += wt[k]; hmat[oldstate][newstate] += wt[k]; } else ncensor[itime]++; } else break; } if (nevent[itime]> 0) { /* finish computing H */ for (j=0; j0) { temp =0; for (k=0; k0) { <> } <> } @ The most complicated part of the code is the update of the per subject influence matrix $U$, which has nperson rows and nstate columns. It has 3 steps. Refer to equation \eqref{ci2} for the mathematical details. \begin{enumerate} \item The entire matrix is multiplied by $H$. \item Consider the scaled matrix J whose $k$th row is the matrix H2 scaled by the value p[k]/ws[k]. (Probability of being in the state divided by the weighted number in the state). If subject $i$ is currently at risk and currently in state $k$, then row $k$ of $J$ is subtracted from U[i,]. \item For each subject $i$ who had an event at this time and went from state $j$ to state $k$, U[i,j] will decrease by p[j]/ws[j] and U[i,k] will increase by the same amount. \end{enumerate} If standard errors are not needed we can skip this calculation, which speeds up the code considerably. <>= /* Update U, part 1 U = U %*% H -- matrix multiplication */ for (j=0; j0) { kk = id[k]; /* row number in U */ oldstate= cstate[kk]; newstate= status[k] -1; umat[kk][oldstate] -= p[oldstate]/ws[oldstate]; umat[kk][newstate] += p[oldstate]/ws[oldstate]; } } else break; } @ Now update the cumulative hazard by adding H2 to it, and update $p$ to $pH$. If sefit is 1 then H has already been transformed to H2 form. <>= /* Finally, update chaz and p. */ for (j=0; j>= /* store into the matrices that will be passed back */ for (j=0; j0) { for (k=0; k>= /* return a list */ PROTECT(rlist=mkNamed(VECSXP, rnames)); SET_VECTOR_ELT(rlist, 0, nrisk2); SET_VECTOR_ELT(rlist, 1, nevent2); SET_VECTOR_ELT(rlist, 2, ncensor2); SET_VECTOR_ELT(rlist, 3, pmat2); SET_VECTOR_ELT(rlist, 4, cumhaz2); SET_VECTOR_ELT(rlist, 5, vmat2); UNPROTECT(nprotect +1); return(rlist); @ survival/noweb/all.pdf0000644000175100001440000253267012544502454014544 0ustar hornikusers%PDF-1.4 %ÐÔÅØ 1 0 obj << /S /GoTo /D (section.1) >> endobj 4 0 obj (Introduction) endobj 5 0 obj << /S /GoTo /D (section.2) >> endobj 8 0 obj (Cox Models) endobj 9 0 obj << /S /GoTo /D (subsection.2.1) >> endobj 12 0 obj (Coxph) endobj 13 0 obj << /S /GoTo /D (section.3) >> endobj 16 0 obj (Exact partial likelihood) endobj 17 0 obj << /S /GoTo /D (subsection.3.1) >> endobj 20 0 obj (Anderson-Gill fits) endobj 21 0 obj << /S /GoTo /D (section.4) >> endobj 24 0 obj (Cox models) endobj 25 0 obj << /S /GoTo /D (subsection.4.1) >> endobj 28 0 obj (Predicted survival) endobj 29 0 obj << /S /GoTo /D (subsection.4.2) >> endobj 32 0 obj (The predict method) endobj 33 0 obj << /S /GoTo /D (subsection.4.3) >> endobj 36 0 obj (Concordance) endobj 37 0 obj << /S /GoTo /D (section.5) >> endobj 40 0 obj (Expected Survival) endobj 41 0 obj << /S /GoTo /D (section.6) >> endobj 44 0 obj (Person years) endobj 45 0 obj << /S /GoTo /D (section.7) >> endobj 48 0 obj (Accelerated Failure Time models) endobj 49 0 obj << /S /GoTo /D (subsection.7.1) >> endobj 52 0 obj (Residuals) endobj 53 0 obj << /S /GoTo /D (section.8) >> endobj 56 0 obj (Survival curves) endobj 57 0 obj << /S /GoTo /D (subsection.8.1) >> endobj 60 0 obj (Kaplan-Meier) endobj 61 0 obj << /S /GoTo /D (subsection.8.2) >> endobj 64 0 obj (Competing risks) endobj 65 0 obj << /S /GoTo /D (subsubsection.8.2.1) >> endobj 68 0 obj (C-code) endobj 69 0 obj << /S /GoTo /D (subsubsection.8.2.2) >> endobj 72 0 obj (Printing and plotting) endobj 73 0 obj << /S /GoTo /D (section.9) >> endobj 76 0 obj (Plotting survival curves) endobj 77 0 obj << /S /GoTo /D (section.10) >> endobj 80 0 obj (tmerge) endobj 81 0 obj << /S /GoTo /D [82 0 R /Fit] >> endobj 104 0 obj << /Length 887 /Filter /FlateDecode >> stream xÚíXMoÛ0 ½÷Wø(³*êËÒ± ša[nÛ†£¥F'pœ¢ý÷£$ÛˆÑí¡Öv—‘ERy|¤òqqr:•@N9(™,~'€Jc Š Ë“Å2ùA¾ïÛ›ê&Uœuš ÆÉeÊ5)Jÿy*AŠ•‹/æinȾ)»jÓìÒ_‹  tj•â>€Ô`ÈŒj,Dÿ‹ÔpâÚö}ðœ,®\Û¸bïÍE Ü.¨·_ì c ö!šp*Fã6I…Ô!Z9§Z™$“œZf£ƒÙ¦IÎÅGNÁK­æÚ2 ‰‡†*YîÃÏ Ç” m@»°Ÿ‡ÅlX¾Ìð.øšm¼¯[ôdùÚ{tõîAw§sspÀLXE™°ø¸ýFŠ'ƹwÏ ¹Ý^¥™†œÐ4SŒ½ðCZûw½Ä Àô$2Ì'+å”DI<¿-Ê.R¸-Ú®B7±Fp¡®®=î®®"»=ÇiÆs.9sŸÕ!Ô!§bäô¬Yºv·i²OUíƒI~2iÈá sþ…öåó„'Ï<•=y¤×÷JPRŽ¢Ø— ÇjÐX”¥)]r¤ë²u˪ìÜ2RµCEª¼ª$Ö¼ŽÊxVúSL/²Säxv€ˆÙ6Bè¿ C®ó•„²†ˆZ°ïHÆjZOPÓjb”ü¦Ü´Ë¢)=+ó׫ÅÏŸq¹~TúV¯eß=p}£Bôc¡z¹Åy‚X6pr¨/:ú¸ úï;:¸óã†+Ú]˜^pÁÚcÖy´>+KW»¶ãÏS‹ÂQÕûÖÅ…Eµv÷Å }`úiý%ë›ÛUË}á}˜ü­õ÷çÉ>?$ŒÈ>ÞhL¤rÈÙ’”¾øìð¹ÀŒd°‘?‰/3òõ¥ØÖE“}u•kñ„F¿¿Æÿ0Ië1fì1³Ízš‰ëªf;L[í®‘kù»#vQ—8Þ™@NÛ ò¤$³ÌÏ»¥ïÉ9jOlÙÿub„1¤GR0$¡Gð²­šÁÃÄÑp,šeLÇm½ébrr o,@⨊œe?RÛwÔßõÂD…‘z”V1‘V® ROúìp©gh éÖ®]¹°CXC£úA©#S?X¼’iû-–‚QÃ?¸ñä|qòÚÉŒâ endstream endobj 82 0 obj << /Type /Page /Contents 104 0 R /Resources 103 0 R /MediaBox [0 0 612 792] /Parent 113 0 R /Annots [ 83 0 R 84 0 R 85 0 R 86 0 R 87 0 R 88 0 R 89 0 R 90 0 R 91 0 R 92 0 R 93 0 R 94 0 R 95 0 R 96 0 R 97 0 R 98 0 R 99 0 R 100 0 R 101 0 R 102 0 R ] >> endobj 83 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 497.365 176.874 506.276] /A << /S /GoTo /D (section.1) >> >> endobj 84 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 475.448 173.499 484.359] /A << /S /GoTo /D (section.2) >> >> endobj 85 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 461.555 164.85 472.403] /A << /S /GoTo /D (subsection.2.1) >> >> endobj 86 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 439.637 231.613 450.486] /A << /S /GoTo /D (section.3) >> >> endobj 87 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 429.619 214.041 438.53] /A << /S /GoTo /D (subsection.3.1) >> >> endobj 88 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 407.702 172.171 416.613] /A << /S /GoTo /D (section.4) >> >> endobj 89 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 395.746 215.466 404.657] /A << /S /GoTo /D (subsection.4.1) >> >> endobj 90 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 381.854 223.824 392.702] /A << /S /GoTo /D (subsection.4.2) >> >> endobj 91 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 371.836 192.552 380.747] /A << /S /GoTo /D (subsection.4.3) >> >> endobj 92 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 347.981 205.039 358.829] /A << /S /GoTo /D (section.5) >> >> endobj 93 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 326.063 177.41 336.828] /A << /S /GoTo /D (section.6) >> >> endobj 94 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 306.083 279.442 314.994] /A << /S /GoTo /D (section.7) >> >> endobj 95 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 294.127 177.829 303.038] /A << /S /GoTo /D (subsection.7.1) >> >> endobj 96 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 272.21 190.041 281.121] /A << /S /GoTo /D (section.8) >> >> endobj 97 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 258.317 196.15 269.165] /A << /S /GoTo /D (subsection.8.1) >> >> endobj 98 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [111.716 246.362 207.607 257.21] /A << /S /GoTo /D (subsection.8.2) >> >> endobj 99 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [134.63 236.344 198.668 245.255] /A << /S /GoTo /D (subsubsection.8.2.1) >> >> endobj 100 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [134.63 222.452 261.377 233.3] /A << /S /GoTo /D (subsubsection.8.2.2) >> >> endobj 101 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 200.534 232.944 211.382] /A << /S /GoTo /D (section.9) >> >> endobj 102 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 178.616 148.661 188.966] /A << /S /GoTo /D (section.10) >> >> endobj 105 0 obj << /D [82 0 R /XYZ 96.768 705.06 null] >> endobj 106 0 obj << /D [82 0 R /XYZ 97.768 667.198 null] >> endobj 109 0 obj << /D [82 0 R /XYZ 97.768 538.212 null] >> endobj 103 0 obj << /Font << /F15 107 0 R /F16 108 0 R /F29 110 0 R /F30 111 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 116 0 obj << /Length 2544 /Filter /FlateDecode >> stream xÚ}YK“ܶ¾ûWlÉN•†¾É$ª”ãHUvœ‹¼9E9`IìIL@P³ªÊO¿À!gi_@hôãCwƒó·ÇïÞ}Œ«‡( “4Ÿª",òò!ÏŠ0N²‡ÇæáßAt8FQœ?‡$ œ5‡#´Í\»ÎŒ‡ÿ<þüîcùP…UçÈ"©Â2‹Žq–qÄ,~Ñ6Åi0OØfA}ˆË UãYóØXžwV5òU=”s›YæŒ,k5wj-)¢`ž³ÐÁ˜ï3¯»Xs¶j˜ÂÃ1McPerZ5<k@‰‡#˜¡JSh£°ÊD÷nPçnìÆóáp~«öòÀÌ–IƒêF&95}á^7Ér#c8΂XLUl³$Y-ªÒð'>¶>d Öp™¶¯-œ&až'~íUdºט·Ð)‹ ×"í,Õf¬õxˆ‹À¡¬ršU“"·oT‡ÙV£’IަV¿\zåíc8ñµ.vS/^;jœv`’‡YœÞ),¯ˆ ý‚ûjÄ_™%ÇãTí­ãå ¶1èà(>Ÿ²ÓßE O|¹ý)?»ö­øÄHË;Í£ªLAò”9!d[:5.ƒXŒ $änù¨Û©7xÚ•Åm[ëQL`åÈ bd­ýb†ºæ;uÚªýH[Òô†7 Æ¡ÅsÀ4Çs =“cÒÄH¾¥Ü€ÛPoCê3t x¿VS§í†)lÕo±“ „Œk·¬k«é~aŸÍ‡’,çvì3h·Zæ¬ådWIb§ì¶¡„e<::±ÕW°¢”3{̧!ì][½ö…Ũu³=q&¬rÛ8ñ2íSÜôF¥¤ ¹n ÏåÞßû÷s”¤W‘¹DäÖ’ Œà§^òš4“^¸é»QO}×ÀäÔYã•…¨Îø}J°ÀÊ.ØTÕ£™Ï­?ùîÊÚÈi ûG^Š8ÙÑŽ!VÂm„¼—ÈX+?”„ý¤ 9²\HÕ„bäÑÆ!J$s”çpÖá5¥©åмQé7c;NL&Áº‹¢ëý…>k9¥.ŸÞòøÚši×sD½]`e‘@IÁ&AŠ˜Å²z¢³§Aõ!F<,±x ‹ÕÇÑ 0œâŸ˜¼D à²vL)¹†³ãqÄ Â€×wrFÎCö/nä†„Ç€Æ ®ñ{r×Èan‘H¯0ÂäeË>†Þ­ÕU©Ü%³Aÿ) ±Ò“%ݬúI–a8¤K()ÛV÷ÙMêCKa‰ïʯ¯ûž\„I£‚BBågŠ%}ÒL«y ¸öLÚ×çhØJ,[ „´ÿq:÷è ¤\; ­xHÇÕ¨†ê¼Ÿ˜D‘WM3?6Ìê  9h'ÀzÎXØ6ˆÃS"±!K 1CÆj¸/©%[RK&PÎ$žgˆôiämÈ%-æ¨=¿CeŒ–Hƒ}.žus¿u‚JÅŒ»÷O÷“„ ºý0p·#‚FEË*pÚ{ÒH°¬Vù¹º× (\gÀ*z¹Ì–sIµßREn Ò,‡w_Ñ\áö¼ªèb(·‹Ê/’(kF‰×œX*_BÀáÝpé}ø¦WRWÅ/º—'Pì,郟7_;|ߘñ(e~U†1œ»1„3> ³Êg(þÔ۲¡ä¥ç(šBB¤ˆ8\ù$˜Xô€É4ämƒ³¾)ñCDZ†GÑ.ènbã#©ÉpнiººC@çºU8K¨ƒÞ“šaÐ¥ 2m’…wg?¶+¬í¦½€À‚£Ggp\^w«l Ö6Ü^¬„8=‘-}p¿¹\üÔ®„’«&3^÷Rb5lå„Åô¦f‰¶z-‚€R š‘àXx8Þ%¤ðYªý@¿”¡xzÏóHŸD&®0%ðÀDMñ:Ž«|uÒ°•Î]Xˆ"ªiô‹³ªŠ™ž;vÇÈÃOÛ“1ÇÑe‡×esÃBéËÈN‚µÈð»(nž9òM®£H ]«ÿ;w–®fâË~ ßgs˜Ó/pf¸¤Kø„„òL¯?.Fž$áZn†9aƳT¶‘kc-°•›ÀÑ=äø·ýÂú$1Deh,×~4øq Ý" þ)ŸØ úßX<‹8 Ë(a´eri_^4c&Õ-pÏ6ßÈÎþ”Ùe— üÁ-gTŽÅeÄßBJñ,àÁ¢mÿ?Á.¹_C‘i ˆe×Àaµ§¯rpÊGfì/wQ %ÄÓp»E¥„ô8Ý¥FzM±ÿ›¦[ü Ïœ# ~œ¨a÷j‹ÿ8ÇbÖVŒÏß+–¾•+Z„ÕAÀ¢jéÎ]Ãæ7K6ü?[E¯±Ä?Mv‹jº™ü2ãUh! D£tý™fÑh±h$HHòõç™°,öà²áaYF~U·;d³||¿|Iª»|ý½«eó¿¹]E—Ó‹”Ñ@~ÙK,ÞïnGÏByú–gàì»WÝ[7ÉhšŸ&íd0ªPÕüXÄÓ(,ÓøîóñØùÕ€PgM/#×éé½Làùo48f|#³ož õ溌õ œô†£èNPÄ ÒÛKÜðþñÓ¿>ÈfkžæÉ½ÿøÃ/¿~ØÝ=‘Ÿg nFß<· œWhЮ5Í{ÔD(aÈ™=a|Z>2>ï àCA ÃØVÉÿÈ|ë´A¹º •=£ÕpÚŸxïðéæ½ÞŠ7÷®÷­Nü^œ"_z2þ|H¤…!Fbˆ† éâ-yeXïuNÄ™ÿšHÐc¸ñ÷Jöj^i“-b‘õéj“Pã7îþJŸÉŸ›c¬†çºEsñ¬{ÆçœÛù4UCv9&i˜%…OT°î»ßýØÝÒî endstream endobj 115 0 obj << /Type /Page /Contents 116 0 R /Resources 114 0 R /MediaBox [0 0 612 792] /Parent 113 0 R >> endobj 117 0 obj << /D [115 0 R /XYZ 96.768 705.06 null] >> endobj 2 0 obj << /D [115 0 R /XYZ 97.768 667.198 null] >> endobj 6 0 obj << /D [115 0 R /XYZ 97.768 357.443 null] >> endobj 10 0 obj << /D [115 0 R /XYZ 97.768 331.582 null] >> endobj 114 0 obj << /Font << /F29 110 0 R /F8 112 0 R /F33 118 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 125 0 obj << /Length 1363 /Filter /FlateDecode >> stream xÚWKsÛ6¾çWÈJ%3"Ræk¦:¸™æÐish5Óñ¸>P(!&–%ùßwñ$AÓª/âCÀî·‹o¿]þ²y÷åÛ*šd(‹Ãx²)&A¸BY8‰£…ðÇf?yøôþó" £O∩¾Ã§¼lsõæJÝ:}S7¬Æ¾/š¼ÂŸ7¿M–“E ,2Æ Ý_ô’ŸúZåbwüg-wògZ°¦jË|:×ÿN÷¹èΘŽ‚»gÞn9î‘æ(ß ÂèLså?Y¡0Œ|Àqéík^–z©2@™³^Êwj÷bl;)ôriAÆó<êk³S=pÁjÑI‰L?äÍ¡­0µ©äúÚà[ÒàýÔùdOàªö³'#xP™ æ‹Ü hxa’ÑòYß=a\Í1Îæõ9§ï=qá"~ôqäɬJ ³Šíq‰f:}8»cNx£h©:8³‚ê¼KŠÚ9¯ñŽä¥ïZ󆋦ϔ]Ùr÷,ÄÕ¤~ðNÇZ&…4]Î =È[IFi¦;eðQ)*ÆŸê\¦+§±ï—¢OVšëVÌ“t¾¶4;ÇÌš4ÿ‘?á.v}w"œlKìøhòÇ.õÑã(7ÏDY+l>A| ð ÊxÎGÜà1Úôkæ†YZUyË\ˆF^û‡àª[‡ÏÕÉ}ÂeìùÜöt,Œb¦ xUÞ ½Ä“:àQ„Vñ [<¦'ÿØ)>#x)aÕyű6Òi•åÒ;«>ª!µr`Î*~ 'S–ëòõ¥cÒe5ª×¬pôS×™sÕ¡|™²E¸é|Ó‡‘ÄÙ45X¾Ù, ‚ΚVù×µõ‰AvVWtqŒkêwÖÛÅ被Q³’AchN¹Ì3µò7²ìüh\% fP'm™ û»§•f3³ñsc6§FzBNp¹¿ZÕÚ1*fk}k«Ž21´)Ž)+2‰LImI\ñƒ`ðuõ¬Ú¨Á«jTç“U˜QóðÓ–úÞ¶ãò¹fÈØÞXÍn¹¹aEãH–*Ûæ8©jÛ’l@ñÙÙ‹OÝ÷Á5_8žvÒdø—sbÇÀ)5j¬f.Ðî\îíq—„«èo!h‰éA¨aÅÙ»"JTX˜¥WE)ËÐ*I†š¤¨¿@ëæYí Œjjê³ÇïMïK|°ÍŒõlŒ*ô‹É¬v£™sì…=·üup‡sÔïoI*Ôˆ¤Öf‹Û¦4*ô÷Z8`!O¡‡,o ú‰ªGáÁj>H¤σ…ðØaŒ‚h ’{ÏN¿¦5|¿{…‹8AYvû&•}-A½dôœÍ˺ïñbHñ¾¤ß&r5¿©Ôì„ÖBÛ’ÌkPT<}ËÁÞ RC„Êô½5ñWÛœ¦~NÖÿ4æ 'H[_Ž8˜ 'Km¶ýwWÆ»çfU3s8HrÉ›"’ oÖ³F~…ÌôëôÈ¿v¬¥kæS8§0ˆºPrú•]zÉ7Ü΄¹®Y#Æd&KQ­¬ÌÐrÝ¢, í‚©­ Y³ô9CQšúŠ0qÁwâÿy œ§±ãêÐq\¯g3¯âA>’lPr° ÑšiáÞ|cD H^ƒ+\m]?8Úž³Å ™xД©ì‡H“S.U•ÄìƒMÔqoÅ ¥É­ZUÞbËÄ‘“=3˜.QFv=ë é-ZF™çØŒ®ÿŽ ·½±4¹Æ—:§ûÁ·–Ìeÿ+]Ó®pýYöõr¨¦}Ò% ƒÌP ÄQWrÝ»_7ïþ4?bX endstream endobj 124 0 obj << /Type /Page /Contents 125 0 R /Resources 123 0 R /MediaBox [0 0 612 792] /Parent 113 0 R /Annots [ 122 0 R ] >> endobj 122 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [206.983 139.676 212.85 151.631] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}bothsides2) >> >> endobj 126 0 obj << /D [124 0 R /XYZ 96.768 705.06 null] >> endobj 127 0 obj << /D [124 0 R /XYZ 123.92 143.163 null] >> endobj 123 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 137 0 obj << /Length 1299 /Filter /FlateDecode >> stream xÚ½WKoã6¾ï¯0²b1×zZ*ši‘Š ‡®H‘æ@K”¥F ’²“þúΈ¤#9rìì=‘"‡óžùF¿.>}¹õ‚QLâÐ G‹l三ÝQ̉ ‹tô0þ<™:³`6.Ôd¸Á¸f,•zû—^hêÊ™ÞH%¨¢Ä¼¼Ù0ñ¢ò¢^é[VJC÷7\'e#—ú„g™dÊ|lY±Ê•*gò¸ø}4M‡ÄÁžf2çMiTXÎìTH3Ç4{JA-3.ªaÞ‡­YÆöŸ§†³RÍX0QI£÷™\³¤ ¥qü°/#7¾ ;¯=Í}û6áÏë|Ú«å=aÑŒ8Q`é -+òÉ,ˆO‰KHœÀíÇÅ´¬Äü8Ý‹æÁ»M &‘w¤ D ôžG_ë~„±L3ýœ•ÔôiÅõZÒ]óLxµn”)=ª—.C(°‚Ö 2+0ùh÷1~صŸŸœÙ¡Jå"=^'@ìÇC«Ð¹ÂÛÇc‘¡õ .ðú§_-Rñõ+ÝÙo&ÒÚ]´6 ÉUŸi¡©¨-ŠÚoxÐUðúì©a/ßkxÚ¸c ÏÄœ–%ßšîÛ”ªX—Ly$|Ýú«/¹ãHËqkìÜòú˜¾¥µ²\Ⱦñ;P¾7ZP%Šç _éÆfá–¾èÝsÉ6 o_°ŠovÓGÏwè-«j:¨ª©¹“×€XýÞ<ÒUƒgÖ@Ú-‘eÉ쌓Indzeig­ÕÇ. ’ܳÜF…¬˜º×»*x˜v“þÒî³¥nî uÉ!t~ËδÒ÷Vs§sRÎJŒÓMª_¹Cðöúîë͹øcN Þujtx .¡b5à SwiÃö7§Â¶š„§»S&é*8+‰®7Ò‚éñæDP®è›Þ° }âÇþ"‡ñÂè»ÁC³ÄÄ ßÈñˆã}Ÿœ5«iù?ȱ(;๘ÌÝ}IsqÔ¯Ò%eE]ÈüGBd27êf8Ü»3¤ƒ3×Ð]C§÷Ã:~"Ų,_ð¤) {!kýÍ„àÂK½fú …MaÎ0ÁqO™Ä$ Tm& î·úŽ5–“©9VË"e†/vlË®µÚñIìû}«ÛQ4%…z÷æ~‹d^h 7Ð|}š¸ó13W ƒZ"D¯•‚6â™îú¸²Rßê2Å3ª2Z”ÍÄHÓߨ,^¯!h‘ÀÁN9ý—ŠTÂxà;ãë4µÌÇ3ì….8¬õ6nd“ ‘¹þªx+Úö8w>Ó?þxµµôŽ7 ¤ùN4^äl ©9CalóÊb¡|›D~LBdz„8ÊËvòl¥dzm›Å$òÇ$Ú{oŽÿa‰1©½@‹)÷¼n]‰C d¯M¼]cázX/ž–ïãÕ§›Å§ÿÀ endstream endobj 136 0 obj << /Type /Page /Contents 137 0 R /Resources 135 0 R /MediaBox [0 0 612 792] /Parent 113 0 R /Annots [ 129 0 R 130 0 R 131 0 R 132 0 R 133 0 R 134 0 R ] >> endobj 129 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [233.135 486.376 239.002 498.331] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform2) >> >> endobj 130 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 223.362 197.159 235.318] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X2) >> >> endobj 131 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 211.407 191.928 223.362] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 132 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 199.452 191.928 211.407] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}penal2) >> >> endobj 133 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 187.497 202.389 199.452] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 134 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 175.542 197.159 187.497] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 138 0 obj << /D [136 0 R /XYZ 96.768 705.06 null] >> endobj 139 0 obj << /D [136 0 R /XYZ 150.072 489.863 null] >> endobj 140 0 obj << /D [136 0 R /XYZ 123.92 226.849 null] >> endobj 141 0 obj << /D [136 0 R /XYZ 123.92 214.894 null] >> endobj 142 0 obj << /D [136 0 R /XYZ 123.92 202.939 null] >> endobj 143 0 obj << /D [136 0 R /XYZ 123.92 190.984 null] >> endobj 144 0 obj << /D [136 0 R /XYZ 123.92 179.029 null] >> endobj 135 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 156 0 obj << /Length 1665 /Filter /FlateDecode >> stream xÚµXKsÛ6¾çW(NgLMCXà[iÝ™¶Óz顚ÉÁñ¢ ‰cŠdÐK{w± -R”“´îÁ¸XìóÃb¡_Vo®>f³%[&A2[mgË”¥I6Kâ”a<[mf7Þ¶V‡¹Æ™Woi4”…643{áfy]Í ˆ¢ÒÝš¤‰’º™ûAêÕ•v¤ûy{Bb]:ʶ.Ë%<à?¹!ê?žFò%7Enj¥ÙÜyæýÙ®u®ŠÆÕ®ã-ôûùíê÷ÙbæsΖ±óI–ò ÉRà #ãz…¦‘”=eéH5Ÿy¡çÄP-Ë--iǶ§ž¤¹(­'IOsîiÚ¦$†È)5 m¼úÈ££T Åð8‹÷Ä&G,!ËÒ-C^?6{]›½.6ROHÌŒgQÇ_8ñ9×6p~°tÃW€Ãq{Ÿñ¢”ÕÎìq&ŒQ8®¤:@Ø-Çå½K¯¾„%Ž´ð ¥œÚ€ïIÚ»0–/XÀƒŽãÉ4üí´ vÛÒ÷BÓ„Lv—Á"ÁñÐÍ'# ±ýèÓhÐ7VT•T_éò ÿÜv^ö8LHÃã«hð'LdJTO8¼-4«ÎÂä6sÖ 'Ü:ÝåªÏŠG>”ÿ TÇ …\¸0ß÷Ú&¤i¤P.ðPìˆ(uÞîe—“­q;ª MT±Û;AšdlG;]Ž/Î É8Ë¢`höŽ¥¥<ž¹Ÿ»‘¶”Ð ‹Ô¢:Z¥jEóeGõ9œn8õ{"ˆŠÊI,n‡Q¢Ò膭dÉÂ[íÝÊ¡¶Ê6ÒIÜ*ÑmêÌ‘ DÌVJU;M4`±ÚŠ;kÝ!GlECn„ ÑbáiihB%göØà¤gi„Fº/0Þ´úh"EŽ:÷ôÕVÅ_­ã—÷Ö ž1¸²"œð¼>àʺ¨„)ê 0ñ” aM¨è&˜@º1B*0aèΑ«Ñ•#uUÝ^i [ ”$–FQÄs©µÜ@:¢(séññHiweh#"N]±—+iã3ÐY@’‰F!À«kª'˜ ¾ Úw×0¼ö\ ›Ýp u3F¦’x&ž ô©Îhɲï„1âc§JíQ {>¼vCˆŽõ?N½RPCPDp¬ò²%pºÍP Ïàf®ên\`Ç|ZD!*†TNª3Œ­–˜oI˜À5 çÌž/È|B%þ’³¤Ç{‡›”$¸ÙŽdã>uqhÊ'š+Ù”"Ǫ&‰Ý1q‹£þ–²qÓ¹ŠÙ"ì¹ éIº¦CÅPh¬[“» hS>94ƒ5[ -0ý꫺T8“Ÿ! k9œºíáŒoâíQ“Åi6º*hÖß p‚I½ÖR‘Éx µ=Üû4ÏBϱh#”qÜ­›<Xáq&•{lá+G»öÒV‡;ý—9Ò … iÕéÂMÖ6enìS¬´V+•á‰ê@MØqÇ×õs¨¨ì{1¸«àÛK-YòM-Y_Ýÿ×– +©Õ4ì&Úʈj@ÓÌ Qê“^˜ËsÍC5!ð¹Ýë5~gKí± dK~Ú€`ãÑ–¥ÝlŽ[‹qT {®mS½ŸŸ¤l¹ÝbÆ MݶUŽø±Žs~ßµú®;÷õ ±ÛÐgÛшEÑ·´£¹mé,B¬áøÇå/ªÕeׇֻÂq@ïºN4di8êDë5î{Á½3vs¥t/…r1® 5T…Ö ÔŒ‘‚‹Êg1®_á‹¿F 6°ÜWç4ÛWN7hKG:Û ŸaLÐdÚbÙ§y˜[ŠÞ €ó­mrdÂhéë­áþ8YGØ'tŒpÛ»eˆ‡ÃÜ{Er×5Õ4×MK¨a]{,že|ù45½îiœzyè>Q¦ËÍ06_~phS7öµ±êúÿ.|BíZxÔ»¯C«Ý,¯+zkŸF®VGѼ.:>Ý?+¦‘0ýî2‡+o¯i¤bzîœB‹°Ì–/¹;Ëߨÿúú\bù"bÉ"}Qi²$£Ó7ë½mfmZìO#ï§\Ÿm} F‡—þáh¯{š››âövôZ677ˆoëñ±²f¿^ù‘¥vì!ùIÕøËÓø'†ÞÑ1R/¾ªůQŽ&¬ýwHü×s4lƒt@Ðû> endobj 151 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.868 102.639 633.823] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}bothsides1) >> >> endobj 152 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 438.555 102.639 450.511] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform1) >> >> endobj 153 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 438.555 186.698 450.511] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform3) >> >> endobj 157 0 obj << /D [155 0 R /XYZ 96.768 705.06 null] >> endobj 128 0 obj << /D [155 0 R /XYZ 101.643 625.355 null] >> endobj 145 0 obj << /D [155 0 R /XYZ 101.643 442.042 null] >> endobj 154 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 163 0 obj << /Length 978 /Filter /FlateDecode >> stream xÚÕWMsÛ6½ûW0iTGBð{¦¾ÔÓ:>ê’Qu IPbB²Ó_,” ’íº™I.$ˆÝ·ûvÉ?–7>†±—“Ë®ý2Ë&ûr’R@=ÃŒ·%¼¾Ã0ôïàù½Økfn[Hô\íû>*`O“mï{=5j‹KNð–ïx7c™¯ÉÃ" ¦xiDò(²‘t}#¿œcf9É4hÓùÆõ㌥>/umás)ÐYÑtM·Á9…‘Eþ°´ßÁõa¶ÐWnNÊÄ Ž'3!|¯€‡­yz„³&<¢šwfÐZ÷Zq¥$¢‡×‡i8 ¾ &´XøGá`Û™ac˜ŠÐ¯öòJ‘,‰ŸIñIJð8‹¿h÷nƬ•‹I[¡9›"ˆ#6"zN²ÞƒyÓÉœ0±×dkxŸµ‰S…ÙbøÖ&ëõGó§,ƒu`nÁ§T1¿m+`ÇÓž W`柑|êúca{‹TÿÉ “:œN‹,ÝëDƒÀaM£q«xu êâ´†ñ¤§C?œãi®J‚ИnÉ ½ Z½pïÃTÝ£±Æ”ƒ&£3L­4»¯­ ¡ìy¡Ñ¡;Ä$„ ÉÌßmK7.oþN×À endstream endobj 162 0 obj << /Type /Page /Contents 163 0 R /Resources 161 0 R /MediaBox [0 0 612 792] /Parent 113 0 R >> endobj 164 0 obj << /D [162 0 R /XYZ 96.768 705.06 null] >> endobj 161 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 168 0 obj << /Length 1198 /Filter /FlateDecode >> stream xÚ•WIoÛF¾çW¨NP’9æp§ešÂ.ZE hEZ™t©yù÷å ‰RšƒÀY¾·/óôyñîêû£%¡Ž›Q¡(ŒGa!× F‹l´´òñjñûÕvpŠ£`ähÀš½îr›W)­7¬Újxmì û_ÃàùŒìØE‘ãlŒQ€2ÛÍØÜÀº±õw»Yò‚fäõr¥ˆœ>ük}¿¯ž¿9S‘ü¬Ùžòú/¶äRà„VEý¯@àþqÍS¾¯åù,þ¶#}qŸªâ1çŸÆ6¬÷ú’½ÝŠZÄZ³Šd < X°žç ‰•™^¼¤ÀFÚ%ô¬8לí]S‹€ñº¯0ø_—„>ò¼ã£iŽ{B-á…H“DlYFJWRÀvsÊ¡ÆZ‰ú©¨Ý—¥\·ÄX>~Ôß”¾즠'hóSÂ\%.>ô S¸è©¾Ýטp»æ~Ñ𲇘‰Zhõ/ €H|MU½œÕFäöRæ¦B~xN«e±Z­ú9_ʰÎ&ƒ¸}]^âU“2.•|{È Ý¾]G«’;kãÄGn$¾J´»+*ó0ïÒº¾îL&cg<-s•Á€‘‘”ç­9IªÍâšð““ŸTH„jä?"YÆ©?4”Íœ˜5î: Aùƒµ¢*V“r3ì* EqCÍj:=Õ_´¾èl{“E‡§G; bH™ãv²£m•‡dPˆétö½R§î‘”c“’yÑy‹¢ùBƱ·Oè‘Õš<º ¢MW}IK0=]W¬® %É:©zÙÿc &às¼é§y1Åá§Ú¥•‚O«ù¼£»¹WÊ[|x£B)ot´ú##ö"á[Qá ŠãX«IØ»ÛÅ»ÿ+¼ËÞ endstream endobj 167 0 obj << /Type /Page /Contents 168 0 R /Resources 166 0 R /MediaBox [0 0 612 792] /Parent 171 0 R /Annots [ 159 0 R 160 0 R 165 0 R ] >> endobj 159 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 653.748 102.639 665.704] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform2) >> >> endobj 160 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 653.748 186.698 665.704] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform4) >> >> endobj 165 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [149.449 474.421 155.316 486.376] /A << /S /GoTo /D (coxcount12) >> >> endobj 169 0 obj << /D [167 0 R /XYZ 96.768 705.06 null] >> endobj 158 0 obj << /D [167 0 R /XYZ 101.643 657.235 null] >> endobj 170 0 obj << /D [167 0 R /XYZ 97.768 477.908 null] >> endobj 166 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 179 0 obj << /Length 1201 /Filter /FlateDecode >> stream xÚ­W[oÛ6~ϯ0: íˆI]—ú¡’¡CÑ­dyPly¦ãPEÅÍ~ýxx‘%›ö0`O:&?~çÊÃã_W³;šŒ T¤$-Ö#\`Tàx”&"jg±ÝÛéôÏ(Áã0!IÀkž¦Ó›ñÃâ·QHR„< ±:–X83À¹ùlCl°ÑÅE½7ˆé¼Ç¬±³;eÂÁ¨D¨ dxþ/klaß–Ó—šÅ!¦i˜¢<‹P–õeÝr–»]½ Ÿ«çZ¼y¬ÁŠ5I3wŒYÞ~•âű2‹PNœM|a˜LŒó_˦1¹6ß5ÛíŒ$7Õ³‘÷r̼–칚G7¾`¬kaU^# Cÿ;ɦÜã~£4êÜ_ûÜÏ)Jóô¨*Ö¥,ïÙƒ­”9>Ô—Ùkd)¤ÝeÞúò•²mŸc:5»ˆŽ/[QDH2Ô²‚ ê”&¯Mƒ°nç=WlX·˜“úwj"„qa:ír1›ðš‡«ª”[þ4÷ÎZ;qÇàʇYÍæ%K»µNV«Þ‘‹(§»®rpç,7Âû÷.µ:M[HÓpGÖm¼±"ìÓÕÛ¶š}—þ\@A9Û/Î7¬Þõývûñ3ø)ÀBÐt¯/ØÃà¬6ß{ã>}YÜþzûM3ðsÇ­!SëE8( sWÜõâÿ^‹ósÅÈøªúÑ…Îa°¯7GÞ®c”ÇôíÊ;ÖÈPT²ÜÛ•RDé°)+M#L}Çì.ï7”a°6C…BkÔb£Ê’"Ð!S™¨_ÌïeÙØ¸Uð-Íç‘I#¨wÃ"öc’µxBã0Ži³²Ðz ß 6=çz¤Ö^•±®í¹º®ÖƒÖ ¥®ç¾›Çe ¶µ®ÕÌ£9ÞCˆL ‚øj-Á.Cc݆¹’ÑhGïšî— ‡‘æ÷ÓÍTKJÝ>4TK¿mñSŽy»N©Z¥j\ƒØ¿¸}y°¬šæz˜ì8pòZ•–¿!C£—lrÜ`}CSœ£<"—&ÝÕܸ”¨Û|üÄþÄør×®ì(õ®iÅëw´yç+¼ï·|50gˆ–ßÜ„|XbM-$ö/Ÿ¢õ[HÎ Šªm¢D%öÒ¤H£%ÅÑÁí¼ªåk÷—¦Þ{ÇÍ,r{íL†÷Æ÷O¾&¸$óÈ‚'¦»z¬êöqgƒ?û\¬ô¡‰™™.+˜Iâ_ÕxF¡c*ö"…s_‰7§wg"ð4«¤äynÎ瀻º]\ýI3³€ endstream endobj 178 0 obj << /Type /Page /Contents 179 0 R /Resources 177 0 R /MediaBox [0 0 612 792] /Parent 171 0 R /Annots [ 174 0 R 175 0 R 176 0 R ] >> endobj 174 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [238.365 593.973 244.232 605.928] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory2) >> >> endobj 175 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [233.135 319.004 239.002 330.959] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return2) >> >> endobj 176 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 223.362 102.639 235.318] /A << /S /GoTo /D (coxcount11) >> >> endobj 180 0 obj << /D [178 0 R /XYZ 96.768 705.06 null] >> endobj 181 0 obj << /D [178 0 R /XYZ 123.92 597.46 null] >> endobj 182 0 obj << /D [178 0 R /XYZ 123.92 322.491 null] >> endobj 173 0 obj << /D [178 0 R /XYZ 101.643 226.849 null] >> endobj 177 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 190 0 obj << /Length 797 /Filter /FlateDecode >> stream xÚ•VßoÚ0~ï_‘§ª…&Ä& d4“6 ¦MÓTu> endobj 186 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 187.497 128.791 199.452] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory1) >> >> endobj 187 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [238.365 187.497 244.232 199.452] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory3) >> >> endobj 191 0 obj << /D [189 0 R /XYZ 96.768 705.06 null] >> endobj 184 0 obj << /D [189 0 R /XYZ 127.794 190.984 null] >> endobj 188 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 198 0 obj << /Length 700 /Filter /FlateDecode >> stream xÚµV]o›0}ï¯à©j‚p¸&mR:mšöPååµd3M "TêþýlìPL mÕí‰ßsÏ=÷§õÙâ:"Gœbê­·àqìQ’ ,Öw^zQ,ÃËY! /óY@0¹x¨DÙäµ~h*sûôí|q9Û¬¿z¡ NL$qh²ºQá‡ÛÊÄû!3‰6§zWåe÷”ãââdÆ*£é(ÚÂ**  ✠°<4&ßR_UÝàTlÜÈ·ÏØM5™´4îKAÆ9Š’dYÂPLŽÈ¨6,kq¸ IEGRÑ‘Tø~‹¦}È*-ŽHuÌÊA†_.˜8”ªÀ¶a Ó÷_AWÖ<RÅ÷fœ+9b›$ËtÚÊt§TgõQ½À::(<Ьæ•ÐOçç=A§Jj‹aòòÍkyâOK‹B†Ÿ¬ ³QÊíÒL÷ìü#ºh›ïê/LJ ž¥¦r¹ä6¯kÝlß·÷>÷µÔ ÚJ*Ê»ü©Ëy ]nµÓ,R[yp‚<-fÔΕÿy½øð„£48¦ éôjá1‚x0-…{éîIé3¯§ªÐCܺ㬘)70~Sí¯UÇUßÒRÝm,W³ î_¾¯WŸW7m„rÌÝì‡[c7ŸN²\¥‡>OÃaŠ€àIéùGäÔÝË»D¼µ†AÃú¢Æþm°Òþ$!–ÄÇbn«§Ûê±l‚ü†uÞ<Ö¥óLQu+Eh‚È Q"byl 9“ívÕm°Ï÷Uýǹàj£4Ä*riàBŒB6øÓ“?­­7ëÇçòƒÇ^ ëc<„Êîlµ>û ë½Ó endstream endobj 197 0 obj << /Type /Page /Contents 198 0 R /Resources 196 0 R /MediaBox [0 0 612 792] /Parent 171 0 R /Annots [ 193 0 R 194 0 R 195 0 R ] >> endobj 193 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 163.587 128.791 175.542] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) >> >> endobj 194 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [233.135 163.587 239.002 175.542] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return3) >> >> endobj 195 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 129.714 102.639 141.669] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory2) >> >> endobj 199 0 obj << /D [197 0 R /XYZ 96.768 705.06 null] >> endobj 185 0 obj << /D [197 0 R /XYZ 127.794 167.073 null] >> endobj 192 0 obj << /D [197 0 R /XYZ 101.643 133.2 null] >> endobj 196 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 204 0 obj << /Length 1893 /Filter /FlateDecode >> stream xÚ¥XYÛ6~ϯ0ö¥vkEŠºPä!]lÚAìi€¦h[¶•ÕáêØ£¿¾3œ¡d­/ yXˆ‡Ãof¾¡÷·Ù‹Ë·ž?Š8Áh¶ W:q?t$¬Ì–£¿Ç/_N¦ÂõÝñ›,+ºI&S_úã<ÉËêiòÏìÏ‘;š áľÝpyHúéæãìújö,UMš³•×ôÑhús²hÊ n®ß¼¿ýòé-¨b¿ž³]œ2üîÃìGí¦Å2y|}Ò^U>6웫Ý´õ{Ç-±c¹3¸*ì½þýú¦wöÄn>}#-t;/ß µ“SÒBŒ¦R8®dlCz^°£ç9QèéFaQ>.ʶh¦YZ7Ó*iZˆÑ¾máŽVxWÊfýcf_››AfFN\ðò%ÝŒ3ãfÃ)‡nÐè|¢öº‡óéóõUu2ø·×³o >ûxóíú}o÷ºü­Š£Q;m@X6©À†´6N§Ïi#ž5ò,•ÎÁ\è<©Oa};»9‹õއ þîÃïÍÏÏï®6ÚØ¿(ª´¾»8Ä3†Å¾a ÉÅÏ:,÷íš0ý´aoß0…î´å:iÞ4M•Ο'ÀÍ·hýö)Ÿ—™M†îÐ=é9± sm±­Ê‚æ‚£ ŠîŽÜ!ªh‡(b'Þ¦þšDÞÀ‹ÂqQNd4~ÀIÔqŽ›’ W  ¬ÒuZèÌŠ«D/yeõLÜ_W:‡ +W€¸l×RÙN¦p\Rmô¶&݇´áµ"a7–ºÑ¼WKæ6åÄJ aŸ·i†.„±9šxTìÜ]yŽ/¥eÉ/ûàéDnlr 1|t L‘_Áýš´D4B¾âÑsxUä;(DWeEGhš_™€ðb^¼–IF:Uòo›VÈ ¸ú½EFÆï§€›ÆÇUº@‹wüÆÞaLÃÐ Ÿ¨3DŽçºN²´àÙFÿ§«%ÉeE.üw&—a¾Í4®>Õ4m¬ùªÌ²b<‰.蛸¡IªE²ÅBò|¡æ›þOǘeÃû·[ZÐì¢QÐðpÜž•ñ‘¸õ}¶KRó®„õžÀEgã˜KÇ/0…LI¶6@Xfm^Ô4«ȦU粩÷ŠÂÖVûw?ñƒ±®R=Ï>loW¨b¥Jp¸h2È5f¯é&¤Z$É’ý¶wDY#Eäø±7„Ž®þÃïtYâªlÊ¥ÀÇaÝšÌÜT“иA)C‚f£Þ×ðCÈÛÁBº0Ž»·R³I‹»z2ÏuD¨z­¤²ÎÔtŒÉ>rüçìC.PbünEkÝ6„päqÝTÀXŒŽÂƒ†DÍà`?À¾´xˆ°bÇN>á._‚ZàN»hLaAóBÇ ëî!8êvVRˆÜa ZçÙm£@wF|d0þh·Õ˜Wi¾Í 8@:Ò ‡ÙbcbŒeš,ÃïøiAÆX²x'”Ò7ê´L.IO6]¶¡²6™HXÐõ.œAÁäxx ÔdØexEÕð®Tð˽çâØ±jàÉÙE.žoƒeYü‚zž©*@L)ÿ0yànˆ0Ð2˜ˆ6~„Ùu‰š%°á"æOÌû»> ¥ñJYØŠôtAS‘Ë ‚ôh±eâbfú7Ž ®hR¤F,%%J!ÜEË4ÿ*<µ“"ævsAÛ:ŽC݇MÊeÎ÷ Ÿq‡6Ùgð¹GMŠ1\õ*Þx´O2,ïSêmià³!Ä­µ)¥dKmœBi¡óÐJŸ=5Ee$€+“œ6è(±Ø°&ƒ_*sªëµN ’#¸0ã5Ís]¥Á`PoX±žê ÙBÃÕK”IÇ€ñ¤-%"[îJÄ„5˜³ÌúœÙ嶇 ¼s dðªIÃ\Qp  ‡qP®]Q›ç;¶‡—œZB™îðçhúd ù/ìP‹¤®uõDò´À¢æ LS”†³©¤q|ˆ.@·kN81ÅßÝniÆØ-i¥Ì·mÃê%‚{¨îû>á;Øç}èÊ?Áö±ãöÿàpGÊq£hÈoñMgî …oŽÚÄ 'H м‰ß`.é£çËl¡Á'RÌl0hëÖ%ì…÷[A”„"Ÿñ™@ÌÄÇØcm²¡†%M¯ª2g¡ Œ‘[fæ (^‚¶÷áuó~!z`%t°Î'H_˜ >·¤!Æ3ªë¨ØYÒbOo?[^‘ƒmOªšt¸´p¨í »-Œç¶ÉÁxÁhñÃÌdP¨ø•JoŽ®ÐP ¿=’ä3“ ~ä!´V$ íÈ£ªÍi¤Wp ¶#/Æ R&ƒ<[Ø8øê UQ3 Ç·érÉú©WÝÿß”‡Ñe®®ët]x’)€¾×2? M]¡eó£@„½3ð«§Í”~fÀ2׳Á~öÉz(dBµB=TàÚ‹ëÙ‹ÿΨɜ endstream endobj 203 0 obj << /Type /Page /Contents 204 0 R /Resources 202 0 R /MediaBox [0 0 612 792] /Parent 171 0 R /Annots [ 201 0 R ] >> endobj 201 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 549.05 102.639 561.005] /A << /S /GoTo /D (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return2) >> >> endobj 205 0 obj << /D [203 0 R /XYZ 96.768 705.06 null] >> endobj 200 0 obj << /D [203 0 R /XYZ 101.643 552.537 null] >> endobj 206 0 obj << /D [203 0 R /XYZ 97.768 263.132 null] >> endobj 207 0 obj << /D [203 0 R /XYZ 97.768 233.551 null] >> endobj 208 0 obj << /D [203 0 R /XYZ 97.768 200.096 null] >> endobj 202 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 213 0 obj << /Length 1723 /Filter /FlateDecode >> stream xÚ­XmoÛ6þÞ_‘¦-b£¶¦w[Ø2`/ °a†Í ¤ù@Ë´ÍU¢ ŠŽ“þúÝñHY’å´Ãö!щ:ïž{îŽò‹WßÜͯ2/KÃôj±¹ÊfÞ,_¥ÉÌ £äj±¾ºé—ãiE#Éùº&QWt]§álÄ鿣ÄO|M7‡ý„d>ª…Ì­ÂqÎGm™Iü¯iEè®m!×üɉFÏ>лqàÌTJl…dÝm*U¢æ¡`Þx:‹£ÑÝxŽJô˜?±r_ðñÃâ×+ÿj^–Ø(EÉÀ²9†Lß!@M·;&×$ÕbmUª šûæà:áD™†lbŒ³-('a2zO—Z+¦ÙG?ñkþ— óT+M&Û™ ÂØ ÓÄY<î¸rÎ*ë¨ñšÝnX®1p”03RD—‚?š\µgИ†iäÅYÜeapÁd]‹­´²ÖJ¬Ú>Ú¨ª$I;íòÆ× |0z§ÔÐÂQI ‰¢È!*þ  #ÂIH¡n^ᲂeÅk÷æåJ®…ÜZ?ªž?–=\å|¯'CÙ‡[#‰þú!ÀÊ™.¹#(.åUq(% dŠb!>Á@‘„&mAÐÎ[šAIÍ\Þ–C™õü8p %tŸÈêŽaFý`$+ZèÒu˜‘î^U{?WvÝåžrg’ÌoîÏ»X¡!‘ID0‰&‘B¶¥ŒO‰uXš5=#´Mu µ&ñqœ$#¦[ö¡N‰Ók¦Ò•0.tJ½ b/‹1Ó1ãY2ZC´€GYÓ-l˜ž6Ä•¼’TÀ[ ­vëEÓZ¬=’V"¸ã,!!D%éQ2/:O–t¡üEæi6úEÓ"õºÚ‚ÞŽÇu)´`…øÌ-r¦FÅQa¶}™Û¢³¤„•£iÜQ«ïâꚳ¦ÎvòâPkn·4°x–Ëq‹©X4pZ§w¶ñ¥-•țϚ&•WOûÝ´dŸøt9`.¡ãDNY ´Ñ޵[‚/y~öÊht] çê¢×BÚB¹>uØï¦t ›%¿­çeÀîè ôKè•´ˆ©7;Pe]Û®]Ñue[=Ûóõ·ÖÐ-]§†€”=R3¹-¸Wïy ©ûaÞÐ,¹±·‰µmÛVŒ°Nãû—[½3ãw|ûÈT}Âè{ë±³ÔËZ§^7£mpú% :7nkM3ËÀ`¡díyèdNq©]µø^6ëâ}Ñàê±î"v÷Ão~BwƒÃÐA \:ܦ¢†˜|8}}Ôå¡ÐgÚ¶câw€°¢¨ŽPü³ŸyÉ<ëºÿ†MçCÆ»³£5 fd’J«§ª£}ÁÅÏY¾#é±i•m0t³…b%Ÿœì € ^‰i³ÅÅ]0Þ ;”Õšž-²ßEþ©ñ¹¾hÎB9Àqç „e½6ÃÑØNrl±=SZäb¬«»^2ùÜ*ZK6oððØ*.xË\ZÒM¥Ö\Ý Åî[–Í©iX÷bò` Òw‚©î¦ÂC¨Œ0ü¥±øã¯ÜžÚ0œƒË±¿îmßuµ÷bkt·ÌÑPólÒ®O³ÒØìzc)MÚUÔy7ÏíÙ’¨ÇtÞíöPYïë‹Y´YºqÉh'eÚàËСÂÈË‚—šuÊ °a˜½Ø§¦YæE³^KE öН»€¹ðÛ©¸F5ì”×g£Æ}5 vØ£®Ý“ñ’ðåè?öR6Ø\ëêΕØkó-ÑP‹O«Õß<·½#?(“ªx>|«å@št[ W ÚÐÙ£M´pê û©9§¿ïÃö ÕõO0]®;Þÿ7¦¶U–]·¨¡Ó!ø,žrÓ”˜Ä.^ëúD#yLm/ÊÆq´•}dRwÏ\òP®¸êÍ–Vo9*Ý{xúÙ ñºi øý5<]Pág·g¥›µå}ûÍ9ÏF|Ü 7¿è»·}T´$ûÌU5>P¬_ƒ´ë_ûU84Vð>³Øƒ³\w«—zŸ™Ñg¶NíxÈž³±œtÌç|îôf±™ƒÁû—¬<ü\>_ÃêRÓ­ÿæÔ1.1Çrc£Ï˜9èpàÓ êÛl:]W•<çuÍÔóÓ–Í9²=̆ËKút>lÑÖ[÷Û.¼ò]g4Ó—•ýQ¥[Ð's†‰Å‘=×—pºäS/‡Ë{ Ùkãìä´Ñ­ùdxøe&›y)~ªGø!‘Ù DÅW¯þ8™n endstream endobj 212 0 obj << /Type /Page /Contents 213 0 R /Resources 211 0 R /MediaBox [0 0 612 792] /Parent 171 0 R /Annots [ 209 0 R 210 0 R ] >> endobj 209 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 562.092 102.639 574.047] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X1) >> >> endobj 210 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 562.092 171.007 574.047] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X3) >> >> endobj 214 0 obj << /D [212 0 R /XYZ 96.768 705.06 null] >> endobj 146 0 obj << /D [212 0 R /XYZ 101.643 565.579 null] >> endobj 211 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F11 183 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 221 0 obj << /Length 1266 /Filter /FlateDecode >> stream xÚµWKoã6¾çW8éîÆ^XŒÞ¶€:‡ Ðb·è!R¤90e‘HA¢â¤è/É¡lJ‘² ½X|Ì›3óº>¹¸ ¢Y‚’Øg×ùÌs}”$É,ŽVÈ—7×ÙìvŽ…¨ÿv#÷f¹p"?šŸá¦¡[v&Ï<8ùÑï$ý··§OYÍ«»ÅÝõo3wæxJ"#ðš½’ŠÙ³þ(êKWÉ<Ê}¡9åLԸ͔òÁ­ƒ§õ“¢!ß­aLòwŠ»¸Z[‘ •Hù1Š<Emv 'ð×s±#°hˆh+´pÂ0˜ÿš›3^Î eiÑf¤fYwL…Z%óF`AJÂJìÎJü ¶Ž¶&p.vX~óÍøÂQ”F>ãæ¼ Øh\ÇÊñB”„a?b”Ií YáGI>;÷®­×²›G¯¶±x%'øÁSTª7÷Ý(,7nAïÝòØÈé}+È$T —ƒî¯Nš¾ØÑ¼ww‡Z”ª†‡AòZi¿@eÌ`aÝ5õb#¥„q¼Ò`×a–Ö£ ü¶Ä"Õ _{»œf:W:Qï%öœww÷Æ'— /Gf#Ýù/ÝÓË3=kÛce]lº½dÀºkMæa%Qaãæ/Üh}úþZG>mUÊ9 {TøkÙy^ãCÁ¤·ùjeet endstream endobj 220 0 obj << /Type /Page /Contents 221 0 R /Resources 219 0 R /MediaBox [0 0 612 792] /Parent 223 0 R /Annots [ 216 0 R 217 0 R 218 0 R ] >> endobj 216 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 572.909 102.639 584.864] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 217 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 324.697 102.639 336.652] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}penal1) >> >> endobj 218 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 127.721 102.639 139.676] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 222 0 obj << /D [220 0 R /XYZ 96.768 705.06 null] >> endobj 147 0 obj << /D [220 0 R /XYZ 101.643 576.396 null] >> endobj 148 0 obj << /D [220 0 R /XYZ 101.643 328.184 null] >> endobj 149 0 obj << /D [220 0 R /XYZ 101.643 131.208 null] >> endobj 219 0 obj << /Font << /F35 119 0 R /F8 112 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 228 0 obj << /Length 1109 /Filter /FlateDecode >> stream xÚ­WKÛ6¾ï¯pÕ ÞlÆÔËRZ] $@{,`Ûh‰’ˆ•)C¤½»@~|ù”E[V|ȉ3ß|óàè¯Íݧ¯A4KAûñlSΠ€ÔŸÅÑøbcSÌžïIùß*Z=,#?ºßa^·E–yÛ³¦}õôòÃm=‡pÙµÔ·áÃ÷Í?Ÿ¾Âp Ê÷C®¢ÙJk)õÍÒ_ 8I8[BҨǣHTü}³LÏ;RÕ|®Ô-¡Ø¼/ ç¸Ó».õXa./zyû¶¯8ÑãH=¸aøa™¤âÒ” Tu¸r*¬œBWx5bî¨~©jh±õÀ¿¡ÜX{©‘ .$¬âÄ(§Éõ¹Þíä*x7ðûSY¨ºõËf¼ÝKm{Ä8>1à}£/´}¥ÃèöÃ/‰hh ÌhH'ܵH[)å>O «¾Cœ™¶,ææƒPb§yKy×6æëK‡°ÌŒ `ß\š€Ñƒ¹Üµ¯€¢f*¢Ê¡çÔ¦!€aNevpú<¦Å£§Ð ²bZz„ºn-G€Ž!R‡ñQxνÁ;‰îéyA]jª<{Í2)G””ÝYúa´Žî)â!­Ó ˜ìúÛH;¶VI޽àa2˜E²$”áŽã)s­Ä†µ¶\¼X³=¯‘ aûö-ftÎ{ Ø­)]„è6¹9òh‹±ýtÅi¾ëáÜÄäÂ}éÞ2ÛÙ¼gOãÊyÃt­àœÊ±nŽú{6Âüj¥°:|»ð¼‰¡º‚«íÀ*I0Ý$‰H»³êÉñnïbS!xêI7ú†'ÕmÏ‹r‹9šŸmS—|ÛÜ;QªéÅE¶ù÷Û—S]wϺ!&:lCZÞçƒÈ =‘¹ ¼¶ÀÍÏ<ý²euùLœ?“$‰|›å äÀP»û²¹ûð_Îò endstream endobj 227 0 obj << /Type /Page /Contents 228 0 R /Resources 226 0 R /MediaBox [0 0 612 792] /Parent 223 0 R /Annots [ 224 0 R 225 0 R ] >> endobj 224 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 488.369 102.639 500.324] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 225 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 488.369 171.007 500.324] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 229 0 obj << /D [227 0 R /XYZ 96.768 705.06 null] >> endobj 150 0 obj << /D [227 0 R /XYZ 101.643 491.856 null] >> endobj 226 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 233 0 obj << /Length 968 /Filter /FlateDecode >> stream xÚµWKoÛ8¾çWx·év#VK²ŠÍeæ°Ç…‹6p}`%Ê&J=@Ò‰ýïKФõ0-Çh÷’0Ôpæ›o¾2ÿ,o><á$IäG“e>ñÝDn4‰ÂøâË2›¬Þå˜û3‚K)¨)ÊpÊ+ÊþtB?|÷·£~»ï…•Åh½üwâÌ=°˜Çó@j§ˆ0¤Ž¾Æ}úÍ Ý¯jýæýµÀ%æbßk‚¸}÷µÛ÷AÃÙÒj_o¥Gú^}å‡=ܱ´¢èNo¥!°fè!%;ƽoy âpÑ÷‚ðfËQö°üïó'ƒéã7ïpëx±/Èõú'7ÊrPøÆ.êðuêÔ€ú"{U)‹3'•<%çzn²ü;âÐcÍxX…_UÎÿPÔWUÊ&Ø‹…jsž=CÚ‡)uêJøMľŒ›jt³Õßõ]Àº&é2P@NñÞøt¥SžÉœíŠ‘¢ÌhÃÉ Ç$à…#¦ÐöÜ ´ÍZ»ß[¤ä´"3^DAº­ˆ4žI6^m|Ùx ¾ýIfÈRž‡&8Wß%h‚Ê 7R˜¥ÊsœbTrÖ²ÿö­ž! ”;BŒñ‹4Þuø‰I2Ús~,°Åƒ!q[V\¼Šš&æJŠj½_•ä VE•‰Ž×,­ó ¾¡fúlV•È&¨ÊîýI<Ë•…é[up¹EíV«g$;Wã—]¦–ßµ!©Ê ÒŸù–f…Œ‡&á#?-«!5×… ¬¸¤€NUÍ%Ñ–þtú S_)¦Ö¶0í½uêÇF£úâ4®âø~8€-ð­¼ãš]Ë=TŽÍWOÁ-4Oxv·žRÚç½v}ßf¦/»0‰7¸Š®lz/ŽÁ"ð¯îú˜r\•ƒ9È9•¹F;=NÏiÖ>ŽçúúÔœ‰Ýn_Ò´ï™ù!žVa0A"/è³nÏãq K62°<1výd11I@GÖNÄ{=ôëAË6YjÒ(,ô®£ÉbbfÄà$ßâò.7ÆnÒ"_­¦€íè3˜®×BÉaÜãiä„ „Cu¦[¿fßú´`¼’w³;]b“GC«˜D…úó/{ÎÍE˜çaZ5AMÏ=/škðŠ†ˆÚöî„7Iùo¼j»*ÛŸÁ¯_G£÷Ÿ%C ߇þõõͪJ7¦rw.^×ñ(¬E¢Eøšp§}uœ J\…Õ}·;~té0?ª¯RËo)òA%wxèc{ºR[‹nEñϤ|FkÁ4v^( o>-o~­œÑ endstream endobj 232 0 obj << /Type /Page /Contents 233 0 R /Resources 231 0 R /MediaBox [0 0 612 792] /Parent 223 0 R >> endobj 234 0 obj << /D [232 0 R /XYZ 96.768 705.06 null] >> endobj 231 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 240 0 obj << /Length 1305 /Filter /FlateDecode >> stream xÚ¥WKoã6¾ï¯pœ`#×z[@S ]t-¶íÅE dsP$Êb+QHÙÉ¿ïCÆ’,'z9œ÷ã#õóöç/›EJÒØÛr‘&$‰7‹8JˆD‹m±¸s¾–+7ˆC'ã+?qžqÓ¢¬¨ZDÎaåoÊv•úJ§%A;Š;ÞJ\x7ðM|Gdê|¯%'ê:*úZ ²r£ q~* £"k¨@ÙŽDB§Y¹žC3nlçmó¸rAuË)×N«‡Šåj_­î·¿.Ö ×óH™`3嬟úN«eó<‚µ<«ëgEœ^в¯‘G¹ h¢ïöß×^(ÉÊsråF« <=Và¬?ÀÉyÛ)rþ¡¹J’¢†·žçª±4<¨KòÆø©¹ýýÏoßn´Z/ö‰Žu["Y·»Eá#65„‹˜“b{› ëÓ³œ©¶ø)‰¢'YœO­.xi³iRŒnÙ<Šq9—OK Í£zµØà;*ûÎöŽ%ö¼¦BŒÊÀéa©T£sˆŒ/0j‹¼ƒÖáoáôà‚UªNo×጑ÞÝ]?]ßß¿`¹²à¯IêO’ˆÎÊ¡êË¢åׯihŸ¼²)ûi‰³gUë«ÔÀï±5m“A B§xbB–=Z½’ï½RgRqôµ)'wHû‚éG['VL± :Óž~L¼èõîŒàÉî½Ëe5£ÖuÓMKúÒ2.ÖäÕõF|:ヱ½9¾"ìXß!ÇÂMÌèLÌ&Yc×%¼Ïûf@mg02›GÇ™|»s]º›Iûäá–×½Ð×ðÐ_›ïaf–â‘æ,«1çWVîÝO9#0,ÖÑÖ)ÙÉ«= —zñØž¤êÇièÏeÆw5%Öç“`Œ7Ës³Stí£> endobj 235 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 607.92 102.639 619.875] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 236 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 432.578 202.389 444.533] /A << /S /GoTo /D (model.matrix.coxph2) >> >> endobj 237 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 121.743 128.791 133.699] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X2) >> >> endobj 241 0 obj << /D [239 0 R /XYZ 96.768 705.06 null] >> endobj 230 0 obj << /D [239 0 R /XYZ 101.643 611.407 null] >> endobj 242 0 obj << /D [239 0 R /XYZ 97.768 436.065 null] >> endobj 215 0 obj << /D [239 0 R /XYZ 127.794 125.23 null] >> endobj 238 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 247 0 obj << /Length 1476 /Filter /FlateDecode >> stream xÚXKÛ6¾çWl6AVlFOÛ º9H€ôÐCëCm´LYl$Ò({AûÛËáz­œÍÅâcf8ïJþu÷êýç(¹IIº×7»ü&#’†7ëdCB½±;Ü`5FåõuzÚ8>.VI˜x*oŽ~YáS/dŠKñ·Ÿø¹¬«¶¤Kܪ¨^púÝ$) H´ Ç9ʹÛÌäã©xnìÝ÷VÁœ?¶TÉF«g’k”üK6çÎ T0òg[Ÿá©xŬ#¢ªmúÐþÃ=:²Çk>~ªkY¤È¥MB.[qxNI=vPâ0 ¼#¬8ÂDæøT…Æ·[<Õr_² 'n± õÁÊ Rßk£µ*–0K½}«œ`3” ¾ú©áV ¦—t^tDÎ0;/’µÆtK¸V·%kæ²Ò6 Ô4š÷€×'÷$Ä¢’Ëi[ôk)è Æš†‰Œá„ëÐOÔØjpF•²LDwô’‰ik nCvy-Ee…QÂdV XIM MØGx†idR6ƒ4KFí¥ÈJÙ 1i}lû`€—5ÛN'l´n,·%ÊíqùQÈšlSørw^¸½BOfQ¹Î(Ã3ðÑȲ…Ì™îÛ¾¼Á Û/Ö‘ˆl7‰𰢪æÄ°yÆhš UøLÙ¼GƦ[¥ñ8ÅÚÛÓ~²EBM'Î…‘O‚¨;<ŸñN_l) ƒIÃ?H¸J†Ç—¼6䛢_hLFu’‚ µœK¤ö)H6ºCóê÷-hÔLµµ½µ¸"så6HCÉÄQ™¤›ªô]þþÞ×Öõº÷æ mY€õÖ¸ J½":0+4+6‡È5 ’ð‡€Lô›X0[¤?ؾåå4û²æG.¨Ë¬yý0rvûN5ãèŽÕÕµ –‘^‹1ïëXjscvæj$c ;¬ÿ#¶82\áJO¦ž-IH"?¾BÀ€-Boîý.âÕœþhЃf˜W˜FÉ“ i7Ó>sÜ&¿âMÓ1޲ùúvXxcì­ÄªÓ$ÿ`øÁ¤2X‚oÈD¼…‘ŽTß;MÐdï4×þ.vùB…b‡9€:_ÇŽPM*ûw7@àægè6—‰®Û÷Ùšœ¾îOûXêûv†°£™xÍï·Vn¬laZЛ™ô%}6W¹ TéûQ¿5°+ d~——¥kžO³yÆoÌxíˆ-ÜžÃHyfµ¢ß¦iÓ=uhðJo…ä÷ÇpñbC!s–~ªÝ~ì©d¾ Ãðttìt'X‹ÕÐÑ1$FIîhv™I-í¤ëü¼bôúŠáìYºd0áÞ#UW¸åÕu]vv ¼pW´½;_ô,$ÌKºYÎÇœ“•û¤â½W]ÕšQ§°Ÿ/jp× "Þj®ùÝõw­xU>.„l¾±tQmlú›_›ï¿åF6WsFͽ‰¥zÙüK!ü„ãï³ Ñ¯³>à+%Û­ý#Ø€à«O»WÿU ™ú endstream endobj 246 0 obj << /Type /Page /Contents 247 0 R /Resources 245 0 R /MediaBox [0 0 612 792] /Parent 223 0 R /Annots [ 244 0 R ] >> endobj 244 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 462.466 102.639 474.421] /A << /S /GoTo /D (model.matrix.coxph1) >> >> endobj 248 0 obj << /D [246 0 R /XYZ 96.768 705.06 null] >> endobj 243 0 obj << /D [246 0 R /XYZ 101.643 465.953 null] >> endobj 245 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 253 0 obj << /Length 2168 /Filter /FlateDecode >> stream xÚµY[oÛÈ~÷¯PÓ+Ñdînj Ø&~(¶»q.¼y %*bLI.I%F}¿¹"©‘dï%@2#òð\¿sæÌÉ·Woo„š¤$Õ\OnW&%1’M´T„›Éírr7-V³¹âjú+Uô/EM¶û²´û¬i*»®vÕf_f¯›¼ÚÔo<í«eÖdÿ(³ºÎëW bíßÙ§ÛN8%)ç“9c$UÊK!‹u¾xø×MøèÅÂóͪ“2çš01\”=Ï–Û¾ˆú1_Yéù¿nš¾ oo˜ìù³”èDN¨³ò$×Î9…F³µ§ü«—õ~çצØäaWeÛÚZîl¡Cg5'úÛܯù׬ì{j‘•eP8öù/ï7»e^’*¯wÛ:·|z^}ZC¯¦~ÿ ç¹Ï²Ã~DÊ|û¹YÛsÆáŒJ¢irÖáFmôXÿ|ó8T¿m²íç2'­Êc[¾óÚ~²#ßèó¦@âë¯Yå̹¾öŸ;BÏ—<äùH¥ÍêîîðݧO1Çåe_`ã_ø€Þ½é1 –Ôë]Õ”Ù}^^ßþüï/йÍM^‹.\^ƒñš»œgCÆŸ[i½ ]‡HêÞÇ‚˜Cn-vOëù =F² %̨–¾ˆBCªÒaø¬Ó_‘`Ê+8ýØ—uÌ2¦ˆNc–„޼ºY½ˆÙ¾½á)*7Rsǵ[€PpbZãÄ ßÁŒOÙÅC$jú˜U àî”ÅÃL0€ªX£açÿ]zþfhgÄð¯ó&ØÁúéiOºT‚N4Ia,…$2U0‹ª/K“$å“Í5´’é4²8£Á‚Á[¹ÿ‰È…bJÆäÙÁ’¾à_)“1ý˜f­4Ÿ×"ÑÓûÙœ›iî4ë°©ŠúÁïêÅ® a¿ÙÝ×yõu¦4Êi±ÛÆ|ªárŒk3ÐZŒ'Àm§73#¦m»[”óÝ=:åwÅv†Å¦½}ˆsv¿é Êš€H! £,qŒyJì8v3f€·±3Лi |ße{€ˆ é SäK/l™gͺöûlcEÁ‹¬Í¡8tqÛ‹H 5Ýûzx"._ZM+¦ñ«,Ü-EÒs7ž/vΉ_g lº-ZVµ…ÝW} 4n_”¥÷P±]æO­/sß Ì<¤•z=«cˆTžLÒK›Z)²M43-á;¨­“é÷vÑ~IÚ剚`.¥Z]¶puVÅ“˜aw.>Hçö½‚墠e2½]çNõùÕJr›ÜW@»EtÐÚÈ“¶b·sy°ÄQ)™9|bÚTÅýÞ%ªûºÕ¡Yuˆµ•Æ ˜eBŒ²Ì>“Ó"Ctäœ#ÚpŠÖÏ1—(i±<.|ödÁ˜J’«/g³­9cÒÓU”’Drw óÀ+éQáüÁÛV'v¬´UG#I«R¹û Éé‰ò$&R\:Wä¥s%RÞ£]‹F0‡úNiŠºSèEÚ¿•6´ƒóä˜ E‰ ÷Ýúíï®D9\\}¸½úï•-_tÂ&èP°ª„P­&‹ÍÕÝ':Yâ%Z/"R3ùæH7!‰M»ròñê§«ìíp ºã%Ò|8‡˜ƒ‹“d/1«í)вßÞ8¯c¾ƒÑ¬« Oa¶sEþK$¦)š£aè/†Õ†3 ãú«ß‘x2AÕªHÂ¥mu}Ja{Ôü3wMÄêÚZ·sw–ÝæqûKãoªgzWÞõ0¹¹Ôð¼ñ"¾­‹…íŠ×g÷8óKm°Ò7ÑÓGâ0…Çà #äI±p,Où¸[tÝ$Iñ[jyº ž[넇ˆŸ±"Å“A+a%PN;šå»m¤8Â\h% 5ÜÍ,سF/í„ÏSÙÚŸ(â4𿔤iU\žŽlÏžÕ‘¿@\¬’ér îûØU4¦C—3ʤ D:RÂN3Pá““ã‡9W¸iw!“:Ð}sCcmÕ`À?Æ&HÚÄÀÁN?—áú¹Ûî%•£q „Õ{(r(oü—ßX37pÀ&\K뺸/[âý=èëáuñ¿˜¹ %NPÞ°.ñVÕ.–Ÿ×puþúO£ÿÝñtÆšN³í2Öo ß> EêãL“8–}.;ÙTôˆBX"ü m:lœOöºÇüÆS–ã ƒ}Ò›«,ó§¼o7‹Ú8£z£Å¾ªònüÀ(PQæ [ÑÝ(KO?îüób½Ž1ÂTr&HÍBÒ.ÙÞTŸ Œ@Žžü%Pž8rÏñ0>£~¿Û—KÿjÙ¹N;ƒ³O|€’)ÿà1+ª:<â<¸Â36¼C”~g7TŠÀÖ^ï—ÆgºDsÉ\{TwÂ3bÄX´G·ÇÚƒÕ:"8?Sjü`4WÅcé„£—qÎdÆe·Soü[â¡I<4è„TÄ7ñ©ôÿkÁSbL趘i­ÿF—,ÿ endstream endobj 252 0 obj << /Type /Page /Contents 253 0 R /Resources 251 0 R /MediaBox [0 0 612 792] /Parent 223 0 R /Annots [ 249 0 R ] >> endobj 249 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [154.306 462.466 160.173 474.421] /A << /S /GoTo /D (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform3) >> >> endobj 254 0 obj << /D [252 0 R /XYZ 96.768 705.06 null] >> endobj 172 0 obj << /D [252 0 R /XYZ 159.176 465.953 null] >> endobj 14 0 obj << /D [252 0 R /XYZ 97.768 388.763 null] >> endobj 251 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F29 110 0 R /F8 112 0 R /F11 183 0 R /F10 255 0 R /F1 256 0 R /F7 257 0 R /F6 258 0 R /F9 259 0 R >> /ProcSet [ /PDF /Text ] >> endobj 265 0 obj << /Length 2526 /Filter /FlateDecode >> stream xÚíZKsܸ¾ëWLåÄ‘=0ñ"ÀrT•ÚJ6•­\RÑMÖ"©íR!9¶ôïÓ9䣑¬Ù]»’žúñ¡»ÁÆO—gï~¶‹”¥‰H—· Î3\,m˜zqY,®¢Ë»r¹R±nÚmïª&ºmÚÜ÷æÍæ~Ûg}ÕÔ4ÖÜÒ@¶^S¥ÛÞteßÑhžÕÔÛgŸ–ÂF~•ŒŠÏ®«}¤¹ë¦þÕO®6%[®’ÄDÏ*¿nÙÑô+~í7¼kp/ËëË_+®Xª”œ¥ÚóÔU›û5l-av[æÛ¶säcXÛ,…‰¶ë%²;Ó¨¿Ëzîî—+¤±,¨ÝßUÕ¶÷TæMÝUEÙf7ëÇ¥•ЭdýÈzÞýÌùDâÂH&Rµˆ‰®¿Ò”©N,ÓBãbV‘ÌšqâýrÅAPõ×kqÍôn3XŒÉ7ÄÓÀ‘¯eÝlª:ë›–:¾Tý]`wãLX3qd%6#™eÖߡ褌²º¬«c–š‘õ;)ƒ‡q@Ùr¥“è#¨ àQÞw(z¥Áuˆnu»X ÎTÊ_­ý\ȹ„±ÑE`Mn×v˜Ûúñ"©H‹3ÐÀvÅTHs12%äb2é$<ú j2Á0™¦»Eâ8¤±4ÙçMà¬ÚQ¥s9ÎöC•q|³q ÷Â"z:‡œÅRœàBÁœzÃìî$<Ÿzn„4nwêaÚÆ,‘ˆ Íbå5þw2\m”‰=¥u™ƒæœýíòì?g¨ÓxÁ\¦LK¹Ð©`ZñE¾9»ºŽ þ²ˆ@vñÅMÝ,¤_µõâßgÿ:û }Õü|€}c£-XPn¦Œ~ˆ¹ °NÊX:/ÞH} p‘2³,“ÂÒ¤àáä@‡ÅÍo.âoBŽœ#çk=€üSÚ3ô ³‰yB£æòäš„Ärª‰ÐÉqûíV2a Oa/!Ÿ¥Pó?`,¿0‚‰ga&ÆÎ\ÂŒŒc¦Ô‹O¯9†©™±âE˜yl¦Y²™ÁuøÜµþßKŸÄKk!Ç 8äæ?&u±›1v€›hfz´J‚Lî!#‚ܨ–šÌ:¹ÿßw‚–¥@ Nñ¤$ÇO$®à¨¶4˜é$µ¯2¾{Rá°ÜdÒKOç§€é„ûAêgú"×úÝêõ4qÝaï| .AòIï'&Ù /bŠovùz¶×!—oË×ß%оãÍ_j GY¦Ï3ÌD=æ,§N 0K¸ŸHþ"0‰Eô‘XDL9xÝÁHçã{ˆ…Å1T‹ßÕƒ`™< ÕÇ‚a%àƒ×šè%KÓ•° ŽÿÁÀ{Yþ" ~ ÑÂ%ád@´Ï¿”½ì–ýƒÄÆGÍk¬YlÄÀÈø+oe#Ä"aö„7ùÉ­ìˆ%'²„r ÀW‚99 ˜w7¼ï#Hø#ÍéŠÇ Ç¡`HC0›:KÄ÷ÒO*Ê(KáR;Ÿ1Cãû'ùlV.Ѥ£<»ï·-¦i\¯í‡Õú»¶,ifÛlûª.;ð¹5¥0é¤íŽ<m×S#ÏÖùvõði@MàŒ°Ïý¥ì62Ñ?z¢©êH$ÒZˆL†ÔJBsaß52&R~('£c0 v%v†FnÇÍßáj4"ù¢1w2~´¤¦ö¦¬m3TÐ#5)yFÄbNk9¨K`°¾Fœ†°òy f'[oQ1ØnnC—E›Â±â¿áû£éÇ æ@•t •·Y; ¦Øc+„y Ö#¶óÜÙˆ2LŽÕ¦¬}~R*—^ÅnJÓaòÌIÛFc²QiÀ+eÁðƒŠ$]ašçƒè‹Ê¥kCiÍ„xQ$$OˬÖaë1gìXdÃ\Á‰ÞÊ' « ¥^›¶Ïˆ%/ˆ†øt"ɸ£Õáh³€éZ %YÌçù?– °;€Žò!oVÎæÖ„Û¸ØÙºÊ/¦-vAH¸sóôi³½Á 5H4öPÄ(ëªî©§xKåØQûŽégç]Þ´eh Ød½;ó‹Ä°4ÝKŽïvÙd“p3F“ñKÝ]¥)“Ææê¼¨ß;bPDªÜ« º¥qd¸¸¸ˆ‡_‚`¨K°èµK1×=~?Û£¦Ù#ž7»uë•óC(ˆ‡Ù`AÅŠ#jabôîœzîKÙR!7¬}U¼­¯=Gïfü|ÍÎùH›gëkÁ:…è§ GT©ñˆôu½³­SênFø ÖiBä*Á¤ž‹î|Ov@W ¬ëó=$+îUµÜÂf("eÁU'{=óg3ÿ¹Þ)|¤çÍÅþþß´û¾PÜýÞ™Éà×€È÷¶Ã!xpñ§óÅçÑ‘árn„\´!•Šêò¡§š7dØÐӸà öÃü¢l+p›:Êúê³]ƒý.H uÊáKîWvæʌ־ÏÚ¾ÂÕ[êΚ.)É«’ /zCÁ -ïL¯ƒ§Ão†ÐEò$jj|C„µ[ðVî‘ ÔË,Ç=ûpyj«¬÷x†8è—S÷ ƒ³Ã‡C8츃a~¹t.äµv¾“è "ôðÞ'D€{Îã#BœÐ•èPg†(^¸8m$^.`•à×õ‘K‡çá ôši}+«|͹|c}DT€ íÞc 8{€>µóã'ºÕî}ÉøgnkàþÃwa$ÿtt™A ž¥¬õL“°çþ¾mî]BÇìø+ÜÞ@€s‹$>©šn†m÷ò Ê.ówfA]»`ôg´¤@ÜuAê?]W9…C X¡KƒL&Ñ?4ö¨²U9,ÇGUØ(²¼_?R#VîIÆà%à˦1°Tô¶NªÎÐ1Â3Ùè&±+·l¨ê¢§¼ôo¿`p,¿ÜU9ÊöŽšyÖù%¾ Ð…úf<-°Aƒ!šQ•°C2²³3Ó“’|À¹zRS°{ø¬‹è‚–¥×Wô’zð• N騹­ë2/»Î}Õ‚‰©èAxÅEÔUd`æÈbšxÑáçY‘4ñÇÔpûI~‹Œš›¬3ò¨¥!ä5Àä.UÝøÐõÑíÛq9¹é?¯Ö(òUƒ.ÆXXEÀU9†õ+¬ðU×—»è˜.¾..‡ø¿Èúì- â-¡AœU,ýëËP°ÿ a2?}˜èçà9 D¸yóÙÝg4<Usí;}2¬æÆ0+Žª‹‡÷¡(úg¡ÝK£`X |˜Ç—î¸;ÁoÛ¶(@£üHÕM¹¹Bæ‚_¬¾Þ…È3ò§ ¿Bb¯Ê©b\%OÊ1üµòÜÙ¸ö$4 ð‹Öÿ¨çéðSö¿ n±K endstream endobj 264 0 obj << /Type /Page /Contents 265 0 R /Resources 263 0 R /MediaBox [0 0 612 792] /Parent 272 0 R /Annots [ 250 0 R 260 0 R 261 0 R 262 0 R ] >> endobj 250 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[0 1 0] /Rect [479.309 656.239 486.283 664.652] /A << /S /GoTo /D (cite.Gail81) >> >> endobj 260 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [159.91 457.627 165.777 469.582] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur2) >> >> endobj 261 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 187.497 102.639 199.452] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur1) >> >> endobj 262 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [159.91 187.497 165.777 199.452] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur3) >> >> endobj 266 0 obj << /D [264 0 R /XYZ 96.768 705.06 null] >> endobj 267 0 obj << /D [264 0 R /XYZ 500.749 626.778 null] >> endobj 268 0 obj << /D [264 0 R /XYZ 500.749 604.994 null] >> endobj 269 0 obj << /D [264 0 R /XYZ 500.749 548.595 null] >> endobj 270 0 obj << /D [264 0 R /XYZ 97.768 461.114 null] >> endobj 271 0 obj << /D [264 0 R /XYZ 101.643 190.984 null] >> endobj 263 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F10 255 0 R /F14 120 0 R /F7 257 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 278 0 obj << /Length 1190 /Filter /FlateDecode >> stream xÚ­W[oÛ6~ϯ0ºÙµh‘º¯M hîaƒlѱb[2(*uþý/ºP¡¤Ý“éÃÃsùÎU¿.¯_üp’¢4"Ñd¹™à @I€'Q#7Ë|rëäø¶(óÓÝÔ Iè\«Ÿz]1z[ºøn¦ëê1c&á”{ÿx¡—»x®Hew’Ïõ9?d¼?žà þ0½[þ6ñ&.Æ( µÅF1I™KÁ§_¾¿îõcÉûjíí½ôåÍ&}Â/™4MHK|%‘)²7û °½høâ qv± È‹MÕ÷ŠÓȈÖáH±0ÊVJó´‡gU¥¿JñQŠu²UÍjOû¡«(¹öIûÖÊÖÙÁ³™Òðb™AÇs(Fi˜68wVÄF•Ñx8{³›[SªóUÇÏ‚mHïðR¾ØX@uÓùql¯-×åó|A7ª\€Ù'ÌàéË÷Ú@õãBø!ŸBÏYÌ@ÕL¼Ò¿Û6š c´µ c,{RÇ=¬(ëÊ&Ÿ—w³Å‹ÅFŒžtíA* ¢KHaLZŽÅ¬æÅ~oº°êòñpl8Õ~k#](âÀUÒȼoˆ³¾uhíR¹¬12n "략椨¨»Æ»è¥öJεWòÆöÚÕB_Edعtb?Di’ZÜÚY›²k}p¾7“˽¹¸{}ꃷôêW 0ÛÈã¶-õ1¼áÂ÷M”ÝX±û¿¬x°ÓZqyôŒÒÛ6z¬3j0âÈ÷¸dØ„PŒ}àJP˜ÆŠë÷jJçÛÔõÃÄÙTLbÕ+Å᥺[M];Ј€=š&¾ƒ97c/ª²ÖZ¾*Φ¦øàüÛÞ Q)8eä9®û(‰F*Ý&±0hŠªÎ7ê§æ¬YC$5r²2·¶n…þXÿsõ>ÌÂ(醀R)„²Çi*X‘É-(QÁVˆl$I"·”¯Õav÷Nè·œ´†¢„y¡¨»o€û£Ò¡¥>Œu÷S=ßç?Bt´n… švh¥(=­éQ_xÏÔ«ŒQEd4Ëݪܷ9ž³Üê˪áZ8–—€™ºzZèšË•Â!¡´Ò¬ÖJÔßuuÌOZæÆ:Ä;ñÈRQ‚¾< S$óV=%qrh*ô”­¹ETà£0îæzaK_Ô-Q×*A„¼QšþT”ë}“ëà#t§-Ú~²ùfr¾«öxƒ¶ï Öè,ë‘U¼³Ûmøó_zâ‹¿`S©µ)Ïq¬uç`윇ä©N.£°¢Ù¶Í"â_BÒ|„Å4Pzóùï?ºÎ/ƒ$ZjOm»Å\/ýÍQkô B·¡T˜â560 W›MMy;P{º* bݹ{®b™)„FÆcz¬-ªxµÎlž`:TuxqI‡Æ€ˆ—œù(æó݇³ßÒD¡m“7¾6fÃÏÂÙ¬€îWv–Ýß·+­ÜÊëázk6RG(ƒ•¤(I¥ŠÈýèêóòê?šš € endstream endobj 277 0 obj << /Type /Page /Contents 278 0 R /Resources 276 0 R /MediaBox [0 0 612 792] /Parent 272 0 R /Annots [ 275 0 R ] >> endobj 275 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [102.002 211.407 107.869 223.362] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur2) >> >> endobj 279 0 obj << /D [277 0 R /XYZ 96.768 705.06 null] >> endobj 274 0 obj << /D [277 0 R /XYZ 106.873 214.894 null] >> endobj 276 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 287 0 obj << /Length 1124 /Filter /FlateDecode >> stream xÚ½W]ê6}ß_Á#»‚'$$B«J•öÞnUUW÷òP©­*“p7±‘í°¿¾ã؆$„À¾ô‰‰3öœ93>~^Ý;„Ñ(õÒ8ˆG«Í¡—£8Zx¼Xå£?Ç9¯Ö¹ŸFA4~P´$kK…U%—÷S-¢ñìÁ,S¶¯”1s¬°õÝÿ½úuä¦yiÔw2ßl$QË>?ÊàÀù¾Ž)àÐ+^ ÕV/R¶5O|-­±1¿Y%aÊùŠÏ`—¤ *îK±È×äzçˆüÈÜæÆÈkñòè_wT¤Üp¸.@ÎÒR8{ÀÂÂ~µ¿9?’'ÙËn3ز§vvKÉK ô§!O˜ ò¥[4V•k"Ú•’Õú_’)éú­´UH}¦ÝET/ޏ£NSÒÝâaÜ€Q’’‹w ‰œäÆÞp‡¹èü-ó''YÛ9ÂZXº„Àï-L·u ±WË’C„K^9ȽˆpQ4ÂÈ¡°œ`µêCùƒÔ׎òBµ[Äk¡0Öž¹µá§½y6nJ‰ß$ÛÇuwºË·‰è•ªk§ÚôïŒG7šŸßž~ÿºúå/ó¯%õ Vr+9v/Ëx!õ֌ûË;­ð·âbù £K„Þo.Ÿ¡Ëïz/¼²û:s™³[¢ko„ŽAÿ Õó¦ƒÓªß¹P-å­ßàc ãG¶ƒòå½-«Í~Fé‹ç!hkê%Ðÿ5ˆiÇ»§ÕÝùõÞþ endstream endobj 286 0 obj << /Type /Page /Contents 287 0 R /Resources 285 0 R /MediaBox [0 0 612 792] /Parent 272 0 R /Annots [ 280 0 R 281 0 R 282 0 R 283 0 R 284 0 R ] >> endobj 280 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 271.183 191.928 283.138] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 281 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 259.228 197.159 271.183] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}strata2) >> >> endobj 282 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 247.273 191.928 259.228] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter02) >> >> endobj 283 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 235.318 186.698 247.273] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter2) >> >> endobj 284 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 175.542 102.639 187.497] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 288 0 obj << /D [286 0 R /XYZ 96.768 705.06 null] >> endobj 289 0 obj << /D [286 0 R /XYZ 123.92 274.67 null] >> endobj 290 0 obj << /D [286 0 R /XYZ 123.92 262.715 null] >> endobj 291 0 obj << /D [286 0 R /XYZ 123.92 250.76 null] >> endobj 292 0 obj << /D [286 0 R /XYZ 123.92 238.804 null] >> endobj 293 0 obj << /D [286 0 R /XYZ 101.643 179.029 null] >> endobj 285 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 300 0 obj << /Length 1948 /Filter /FlateDecode >> stream xÚ•X[oãÆ~ß_áGÚkqy¿¤XiêI°+$Ú¢S#{jŠ#pHËî¯ï¹DÊÔ¶yØåÌ™s¾9sîò_Öï>|ŸæWuXIqµÞ^ÅQÖu}Udy˜TWëÍÕ߃7×ÿ\ÿp]­â8¬óœ©77׫8棈ܨWy’㞿ãæE¯ô†×ªïÕ+/÷Ötƒî…Ùž ýΟzór{éæ$Ž¢@u›¢;ƒ4;Á<2ºA £[ý0£Lmì³ê?²ì†Uú¼ùóÝ·?â—ŽXÅ·ÌÓNoüÎðèOK×ttùš\”å§R'˜÷tõ¹^ j&÷é§õÝßî>ã…¿|þy}÷Ý—›qßšF 7,EÚø‹èv»u>ÎÞÂGó%iXÇñ1°øz··½ê%>žu3XïÌ3¿ø'5¶ŸÛô·ã}+ä¼UÔù—j[Û é=zFœäÌ´Ýž„¿þÔvs¯Ï,9Ñäý$æïý_y„bM¯Áò¼¶ã°Ŧ «Q ™[ŠÏeëL\Šú&3…gN6xþÕ7¿y°w.!_›¨0Îï'_üJ.ö©ôå÷_Îræ"î¸¨Ëø)b ÿ .7— èܾoË],U‚Jæ$¿Z%&Ê·ö¡5O@ýü¤6í}²1ÿfÐnͶU·;õb ”.Ë.º@ô»ø”nßÛT çÄúáûjÒê°ŒS-Ã:MYv½!K£`Cu,Kê`¯œÃÖ‚T¬÷øÅÚˆgÊ1áúDhÝ1ÝÙ~ð2÷×pòÊk®t`„,O¹U µ×ÏÈ]E.çZ‹«ƒ•âŠt¶C÷&%¬â,¬³ì¬t V^ÿ&c¤u bqÖ»M^¯Š´äãÑóuŽ­“ ›jÞ­“,,ò l=-ôoLšæaš%žë`Ú–ÑÅ>¸©‚˜ijàíàU€®Ó Ín™¤Uƒ¦ydj§ž¯—'|Ñ{ÕSA;‰QгÐPX@³$ƒn»Û·zÐí+ïÝ#KR:¡Û¹oà’¤ð.Ã8bvKØN0þI¼oèå2‘Krª³ðò^6^±î:ì°Ö ”éZ½g—ß‘’$&6Ä iUOAATH²ûfl„BíïÐÊø… Žƒ¡‡«×èCnÈR¹ÂPD#´GÇëypr'DúÖö¼Píj©˜µ¬tx”$GÞÝDØ«FãkÓ<øí‘r ¨ì”¢èƒ?‰Òª }}žqV±?7ÉaµáèP…A=h;ºyÚûÆIÜÚtÈ%#FS  îwzcb0‰¢0‹ç®„{ ¡Ó®Ò”üðö-YFUù•ÇT!6Ž®ZIêÌ<Œ %ÄØ[Ãäa^§0ÌQÈ-t/~•ÿÂl-õ ·Îtæ%˜Aˆ’õé4ëa3R‚râÃb£{Ã6à›”^¤Qm3¶”‡!ÖÌ<øzBŒÓ·lß©'¯$0Átà˜NªÛþ‰é‡ OçÞޒp9ïé6ú…yìØ{x)íÐrЬüü<¶¯ÌwœdKFœ…2/9‚¸/f‡A<î˜@`°ÿ%{> ƒA˜s¿AŠï7¸ælŒ&ÉçG,KƒßÐPr£;KœÈë2 ¹ñ$ô ãNÊØêÞ-ÿ:åz¤¸vñ¬ûtÅó"-v ®–à‚YŠ%ª#¢ö%c×Q%¬‚o;©¦‡GŠxà3s æoxñ™f»äŸ²œX¯Q(!êàË:(½”Žx cQH‡KsèO–šf$ÏFuyCÔcC*’§æg 5舦×Üúè{²–•¤ÆišÛ'>££³;Í#Z„ãuí¨”X@ZoqšU¸œàJ]ž«¹À&Ž/rBö{æx‹|åy@ç"Áž¢³§‡áþpd&a ™a<öA`ô®B®Î"Ê7O„¨÷¢˜»Çº„¾vë¡cÊv친âF°È…Œ¦É6…_–[e°•d5^c˜Ä4§Ã‘âô cª\gÁ—ñ8ÊŸXú±ã—¯±ðáTå`Gý§çËÀ&"•Òÿp!Ë_?ÜóNŸ2ì(Gòh楿úŒ÷0Žšýà8º7†òæs;0Åû¯:ù,Ê –Mf«H~b& ý(¿®Š³1í8Êé—Æ¾¬¦¿°fpEf§QÍ,üV›¡}\öÒ(,@UR‡U%zMd|w·~÷_òq¨= endstream endobj 299 0 obj << /Type /Page /Contents 300 0 R /Resources 298 0 R /MediaBox [0 0 612 792] /Parent 272 0 R /Annots [ 297 0 R ] >> endobj 297 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 115.766 102.639 127.721] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}strata1) >> >> endobj 301 0 obj << /D [299 0 R /XYZ 96.768 705.06 null] >> endobj 294 0 obj << /D [299 0 R /XYZ 101.643 119.253 null] >> endobj 298 0 obj << /Font << /F35 119 0 R /F8 112 0 R /F11 183 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 304 0 obj << /Length 1100 /Filter /FlateDecode >> stream xÚµWMÛ6½ï¯ps‰?bZ¤¾,l´‡ šr ô°5 Ú¢×ÌÊ’AÉv6È/)’¦(Sv°M†hŠä¼yóf8ú}q7û臃$Š‹Íz$I2ˆÂ þf‘ ‡UÍp½åh¢p˜ÂûÑz¡7œå -äs+"GõV ö˜‘¢–ãMÉžJ5ÏFËÅŸo0…$¡²T“Ý^‘›òaË”YqÄŒâUNÎgËAVÑï¤kù ÒÄË®·Ê 7àÀñUÌOa‡gÃU™Â^n:NJ‚œŽEòÀ3¼¿9o´9½!î}q¨HvÏg¡Ø6û+ˆ¼˜&‘K¬€M“À ¶áÓ1¦‚G—i {̨#®X@êš’ e•¦dUÙÜ(6 rê¥h:÷AàGýÐéƒ×ƒ8ˆûWyA€!²×À?)#¶Þ#ùÌ >’+ÆyY¨i¬Ü§õ[åÿúÀŒà8?›xMþmçml2v-,_—V¦èõn=[Ké½K¢.ãã‚ÑêYA0 h°\$¬½óÞ»´'gð< h/¼L\»ñ÷ %™ªb}e‡ /zÙŽØå«ÞÒêfB+4{=Õ£¸,0¯åÙ1Ê èÔxÁ“û§ph¾ŠÃnEX—¬-aäšÇ§-ÕÅVHDV§§ ²¾2†B¢äjºêRwIòdrÓ¹Iz®Ñõ¡ê£ƒvOŠ\%'m*²œùñÃÊ8í­*KÍxÅ~6Ôºµ—•Å[%¿5+«^‘ýW•8ÊH›¦;Ez&-L y~KÑíŠöKŠÊÏT¯W—ÎV» «Ëçò¤«^®q­2aGv%{é´ æÖXssk…3†_ªÆ A”tÊÛ¬î ›ÚÍM•Óµ®U{Öt;²ñi.V×ùî¤nuK©!ßɰŒ ‚ž×^ªäkÈoDÇáõ`Òð/~3uÁNToÀ_ZÍ"gêfœçZ·Žée£÷?›o‹Ë8ý›"Áõöp ˜û·»+Ôí®ŒŠ+.$º¯UèO4Ï•Ž„mr­LÝV% €hn{@+™€ÿ¦á¥uè{Qd–I”£îÛB–7³Aë².KÕEaöDÞœ)í°â¼á]EÅ!Ï’€˜•‡s®Œ ¦/ÿ4ÉÚ¬1zg’¦Ü˜Z!ª+Ÿ÷%-ê³Óåe®»»|a^Gç†'´ØƒmlÀõ$4_fFž%í&Wlï*Û¹³ÚZM}F=âšÉMBçg4Ÿ9ÜAýg21Hüô?ŸœÂ¥Ë  ð’¸Âù…º¤¨”!Õ!×Dê|Þþ°1ô9T>ƒÔ?š.+@Q“1â‰åƒ÷.E†Y&þ‰ ¯Zú%ÙƒÑ4Fhø×hîóKèY®ã„ŸF(ž7Šx¶Nm¾¿Ä@´gÜÛ HD¨Õ±õ–•‡§­Ü‰EU‘³Dï!ê¾šÇ ñy¶ú>˜óï®Æä‹wwwÿC'Ë endstream endobj 303 0 obj << /Type /Page /Contents 304 0 R /Resources 302 0 R /MediaBox [0 0 612 792] /Parent 272 0 R >> endobj 305 0 obj << /D [303 0 R /XYZ 96.768 705.06 null] >> endobj 302 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 309 0 obj << /Length 1716 /Filter /FlateDecode >> stream xÚÍYYoÛF~÷¯`ߤ(\sï]8 Ú4 ¢/m]´€c´´ŽhtHÊv‚üøÎ”H‰T”Ä)ò"-w‡s}sìH/ÎON_©H#-ˆˆÎ¯#-‘*\"Byt>.÷Y5Æ”áA5K«°Ê– c&Ôàõu}h ã—éf±5xoÈ`jÒjV>µÔR{’Iîhª"»ZWY¾*·|ÃÕ/ùÛE6ÏÙ Nàÿ9âãÈoÌà©(«°Â18Ù¥9Óáåù¯QÅ#̓uSSdwC.ißDŒÕ€Ò3•ZM3·ÃœÑ~ïÖ«yÕ= üO_aÜp(N$H$Ç^äož®åwÄ™ÕÜù8ði±A‹K”Pé©Þ$,qadKÄ%ñgÿI$–TY±B’"1åȲ@Fd)¬kMHލØxL;ì&@Æhd=­h¿R JNšµuj2£ˆZã·Dë=' „nùè‡@ÄÚ®Ä ™-_&I(ZÐÚRˆ¿aŒm˜ñ¤ gÞÝ¿Üg¦'¼>&¸IƵcÈÜÆ>×S Új€M!¬‚¶?cˆWE'¿œŸ¼;ÁÀ7‰ÀtþÔ2⊢DÓh²<¹¸L¢)BÜ#ªUtïH—æHHËEôçÉï'/l!hTó’J‚nJnx¿‰%`ͤò`†äºéòƒI!JôQÑ.¾>ÚåwíÁAÙ;õ`µ5†â#ÃXì†qJŒ0‡¢Ž$máØ¥˜½ª¦Ig:*ðmí“3d!«fqM[1¿²D ˆK}LÈRèMø@ÄÒD"ž{ŒˆÝÍmÖ Ù—‘äó›!Œé.²A&è .Áû>cˆêq‹O/üË@ˆ1±íÌ2æ½…G3xÏfŠ„ÔÎܳ‡#O‘À¼DxߨUWñ’=XEœ'Ð*ÉÁH (áüPñªy1€–ÈÏ Ö € èÑ ú4SÙb:愈Jp^ª]wrXbbs˜Ä‚…lAqD¯úæEk$ØQÐ3‚°Ô '„! ó{€^uA¿_šuüõ…=Rma‡-ÛÅv§Q ”¨FGùD‡bß[ƒ¢Ðˆ5—G•¥O5(4(öè¡ w‘pkN"Ôÿî@ö r Õúq(`Š"äë8ﹽÔö Ò|e~i‹ ㎯¨Q¢O(ÒÔ ¬ =ý—;;Ú ó0ݸq›a>0àP1ÉL=áÚÃIº˜¬©Òë•§¾‚y&gÿ0É—·ë´rtéfðÅ{Æe³* Œè ® e^±3ã¾æž÷u^øÅº¾0×û§šGQ¬o+oD´f¬=ÕæÝÚ”Uå­¾öÆ“úǼ3 ÿp•–Y‰:ê–ý@Tá>ó$pÕîÈ> endobj 306 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [159.91 486.376 165.777 498.331] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) >> >> endobj 310 0 obj << /D [308 0 R /XYZ 96.768 705.06 null] >> endobj 312 0 obj << /D [308 0 R /XYZ 500.749 630.473 null] >> endobj 313 0 obj << /D [308 0 R /XYZ 500.749 595.547 null] >> endobj 314 0 obj << /D [308 0 R /XYZ 500.749 563.205 null] >> endobj 315 0 obj << /D [308 0 R /XYZ 97.768 489.863 null] >> endobj 307 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F1 256 0 R /F10 255 0 R /F13 311 0 R /F14 120 0 R /F7 257 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 322 0 obj << /Length 1180 /Filter /FlateDecode >> stream xÚ­WKsÛ8 ¾çWøVÙ‰hQ¢^›fûÈL÷¸ãžÒÌŽ"16c[òˆ”›ô×$!ë9é´{±H@àãGøÕÅò6g)I#?š­g”1’0:‹Â˜ø Y³;籪çnè‡Î/ôžn¼k;{úX³º\^‚”ÎïWÿ,o)ë›ô=â§ÉÌCcVeè4¨ßj,ÖäÉ-Ïòåx™âvºXjc3—z) “”¤!F]л§{«wÓî~.¨>CYðLm®ìjY ¹Å±Ì«š_J©²ZáR±ç{¯7ÖV¯ŒÛ8 ¾šAchcŸ=&Cfe‰!—æS¶Äc—u±ÿn?AgíßÿþÜð|ûYòúS©x]7¥wi ëÀšnN©q17&[CÝh¢úÛSõ·qçS»p4$^¿‰€0ö ’çUY`€¼ÇL‰#³êŠï¯ŠîÿEb’¦£BYí}oƒÕì)Ù—è]MÆêo§õ_a Á3Ó;à > endobj 317 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 414.645 102.639 426.6] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter01) >> >> endobj 318 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [102.002 295.093 107.869 307.049] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup1) >> >> endobj 319 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 295.093 171.007 307.049] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup3) >> >> endobj 323 0 obj << /D [321 0 R /XYZ 96.768 705.06 null] >> endobj 295 0 obj << /D [321 0 R /XYZ 101.643 418.132 null] >> endobj 316 0 obj << /D [321 0 R /XYZ 106.873 298.58 null] >> endobj 320 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 333 0 obj << /Length 1218 /Filter /FlateDecode >> stream xÚÅWKoã6¾çW{ØÚIL‹¤DIMŒÛn€ía{ñ-›ƒbѶY ôÈãßwÈ=£¸‹E ÉÑpžßpÆ_6g«éÍB*¡f›ÝŒ;‚…a8S®ÇD0ÛijÛy²[,=áÍ8žsŒ^“Jëµ'¾¸Ûü¹ºánO@ ™ ¼™ƒWwÈ1P±T‚ùÊ-9g¡ç‘ŠZÖÎÕbÉAÕ|užÕÇû†ž“†3ª’<+ñ|¾š°ÂJ?ªÇ"Yà»úu›¿.wI–”‡ qÊenØ2'†Ü˜TÁPÑ~Ê_á°PpX% 9GÆÕ¹2²óüœ\ÿ®Ÿuq‰ffÛm~|J£$ÃStŸ×Uó!¦½Î¶š¢E<Õ»¤(‰¹¬ô#E›CDÔ—èíò¤M Êü¨óŒÄ"JC•· ßȦˆlØåEg–î²H^þˆî'GÍ&íXMQAvÎÄbÇŠ¿Îž£¢9\\|V©ap­aȤï“V…Œyßë*ºMíjÊÐüx¸À¥nœF4-è[ ß…€ÀÇ©v¿ç Ì_KéƒÙ-†Ì.…´CnáJð”øÅfÖl¶ÈÓ]q¤gFö³ù×bÏ|ˆ²7h°ô9Éö¤ä 1n£Bïê”nÝ£%õ‚Ï+$åE¬ Ã’»,tGÆÍ"skŽ„µ¦Í¶. mM3b@ݾÖe‰ß""UТq_æimÞä@+ 1¿Z½\0n³q¤–ƒžã̇j¬+÷0¶ .Í÷iòx‰Ôw¡îè„M ¼ÜÎàfò¼ðØ—`<ËC=Å‚ƒJÊGfƒ:qoóf6]~|›ŸøÒ¸æÌ ]ÕE†<Æ[n­©"$Y#Í ÂM¶FÇÐjóˆ ô£¹õ<ºb±–%e§ÁƦL›Q|}9,®Q0À¼’â§«­)dƒ"×ÿ‘g¿P:ë'ëGU”„_ ‚HU/ ²P£ ËIàÛñDE ×vº@È·ÿ¾2ÍN6L£k3 ÷h€„Š5µP”ƃaÅÛ¢ëdžÓ ³!] hóNa$ªÈDn%&üX¢yKîA…ý´øŠ—¤$ãl‚6=#§4ºŠ²ÁKõQö†=Ú6Ž÷"=—9R:ôè!ôû5BÂSÌóøhVÀ§ªFp…N¢ÀbÓ¶‘Ωi45¹dÜpIÜ4Üftù„"öM«åS†ißx5ÕÛÌÔůºîy½¦Y¬Gû¸Ñ4:q²ÑqåC§E#Ó/éc7˜ýÿ˜´Ö]/ý ³Z³úf=tfÍ\ÎWކRxú@ÏíÃPÕ;4÷¯ª^îF¿1ŒFq E5ô åžBúRIÆe@g;oâ”uÄõ.1 d¿}8õy>Ì#OXÓl›Á…æ>Û.ìGÝÀ½ }*ý牎"‡Ïá­ŽJÛCžêòñMX”A^šá²yÙL ©.L>[<ôüïÿ†ÙE÷¥Yù²ëF·üne±n˜¿kR¬ŸÈÑÏŸ•ûñ/ ¡ó]ÿ¤9€ž·ÕØæ&M»¬ †¥€—Ê‘£­åƒHY/&q ¿èäþ»hÔ2S¦·H#UÒ£ ßÙ×ÍÙßÈcN endstream endobj 332 0 obj << /Type /Page /Contents 333 0 R /Resources 331 0 R /MediaBox [0 0 612 792] /Parent 324 0 R /Annots [ 326 0 R 327 0 R 328 0 R 329 0 R 330 0 R ] >> endobj 326 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 617.883 197.159 629.838] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 327 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 334.944 102.639 346.899] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter1) >> >> endobj 328 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 227.347 128.791 239.303] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) >> >> endobj 329 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [138.615 119.751 144.482 131.706] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 330 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [206.983 119.751 212.85 131.706] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 334 0 obj << /D [332 0 R /XYZ 96.768 705.06 null] >> endobj 335 0 obj << /D [332 0 R /XYZ 123.92 621.37 null] >> endobj 336 0 obj << /D [332 0 R /XYZ 97.768 462.023 null] >> endobj 337 0 obj << /D [332 0 R /XYZ 97.768 442.098 null] >> endobj 338 0 obj << /D [332 0 R /XYZ 97.768 409.664 null] >> endobj 296 0 obj << /D [332 0 R /XYZ 101.643 338.431 null] >> endobj 325 0 obj << /D [332 0 R /XYZ 127.794 230.834 null] >> endobj 339 0 obj << /D [332 0 R /XYZ 143.485 123.238 null] >> endobj 331 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 346 0 obj << /Length 946 /Filter /FlateDecode >> stream xÚV[oÚ0~ï¯`o@‰‰í\ˆZ^V1iÓ¤VmTMj«*¦‰S%†Ñ?;v.ÎRû_rüËwα¿‡ÓÐ ð7×è8`æÀçúaw®OÃÍè%ü5ý!– …l (F ¨äèzd¹È>Û®M9Éçó4:ʉ؀ê×2'ÑÛÕÈ‚Bd8oô]ýà[¢&'zk%Ê6j±•š«Éx*ÍØ'M`äoò¦–×jH²MBßžàKiPé”é=¶ï³[÷ø\ب’˜Ž)WдP#ËôFœ±É7µß «-E áLN@àz Îðx¯*KMÆ:Ë[QžÛWZÿ5;Dyµ¸¼¬<T<µ1–„GOôEëià²deüºÔ¤©½†Å)ÒŠ¦ca²yV‰fв8')a¼íy'âVŸy½ùæ@‘˜Ø$IAz@‘ ‹Î±]à­:ø2¢í°kjêÔ1‚V¦X/Mñ¶È’A%KiÄ'’žÉ^Æðª'o?Éj7oEQ0;ç±û¾id—ï¹Áw¯_gäëœ)‹{_!tlíÄu£Êádôß—!Âó[•©ÌÉ ßçLÍë G¢J·D/dƒŠ8͘Y B­µØ3Ï­A{Rc¬UÝG"Ûs³$jE_H›FÖÞýlÂiylÊCÛ¶›v«;,ݰ(©Ú•Ñ­‹É™jí¨Üj¼–3ß©H Ç8;ZkÊh±íóàµ0U³vÒº3à mZE³,åByS`Û6§©dMÍGš WúWÉ´œ<ÛБÚEaƒ‘å`wø“„àA~HΕ¯ )'S rCÌszœˆDBÉ»<ö¡e«Ã™—¥¨øHS"ŽÆåQX^_ªï; p3Žï2_0ÆjCTbbìh«Ä¶È§Z¤àù>)M@??íHâÇý ?¸ÃQsÚ\yèùÀ†¨7939eYœhž”šÖ©SUtB£}¶k°Ó?áWnEcW÷àÝ5­f@•“4Mô¸§]Ç)¹½{©ûgßC©ªÄ¨(HºLHç ÔîYÙrGb^TòzÔÏZðî[¨íÐÝým¸¸ ¥;¹”ÕFI’Å5“ñ´7îtéàÿQÖ€‹ðUˆ‡·÷¯‹ß ®>kOšëdê<Ôã}Ñ4ihø*ž´Ÿ‹èÔ]³¸Ò¥D3ij§ W$)E…ÙgÛm>ðd“À²¾°Ò†|)x±/þY´ôå endstream endobj 345 0 obj << /Type /Page /Contents 346 0 R /Resources 344 0 R /MediaBox [0 0 612 792] /Parent 324 0 R /Annots [ 341 0 R 342 0 R 343 0 R ] >> endobj 341 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [102.002 330.959 107.869 342.914] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 342 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 330.959 176.237 342.914] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish4) >> >> endobj 343 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 275.168 102.639 287.123] /A << /S /GoTo /D (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 347 0 obj << /D [345 0 R /XYZ 96.768 705.06 null] >> endobj 340 0 obj << /D [345 0 R /XYZ 106.873 334.446 null] >> endobj 348 0 obj << /D [345 0 R /XYZ 101.643 278.655 null] >> endobj 344 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 351 0 obj << /Length 2211 /Filter /FlateDecode >> stream xÚ­ÛnÛÆò=_!ô‰B-†Üå5AÒÀ)ÎAѱÐhƒ€’ÖyÑY’q|¾þÌì W¤´Nœ¸†–»³sß¹ù§õ³çoe¼Èý<Éb}½áçy¾HâÔp²Þ-þòÞ½ÿm}ùfýwº*»¾)jÕ-W±ˆ½WôSTU»ý]mûV#ÔÕúýÕŸï.è,‚ÿ^.?¬ÿ½«0ôó˜q_]®?ø¿~ýùãå/'$AÀ¿õí›}aðÿ°mÕõOŞ㞌Tœ#-ë¢2^yŽ·joªòöÌ aîTÿºïu¹±HÏû¿"ö«ûzÓV¼w$:Ã'¤Ÿ‡„nhºíÁΈn\ÿ(L«~Ð¥máž¿ù‚ºÝ*o Tê§‚®J?\.`ëu³Sºk›ÕÏeU-W2½¿ƒ0ê;”MÜÉg¾ÌrÂñÇ^5p!Œ¼~¯p!½nПÊOË}–vvE_(‹û¢£3ä¼ë Z“aæu}{øpz¶íp¨˜J{=îÕ‡¡/ú²mFre× ªã›Zñb·S;¹Šdæ½vÙñ Á‹ô=@KÀÞà5)mìۥȼ;úè[:ܵü½'0AÜfèD“m :‚Û}u¿Ì¤‡LD0[ VlñkÏ_Ÿðƒï0…²f†î–"õxÝ(µ›r$Y|,]þòß¡ Äe_×—2:¡oTkÄR5åÑæp` äƒÕœ™´FîÓ‘{]v·´ûûäSa4‰‘ŸG0šûa$‰QÐ\zî~¸’ “Y#E‘¨ࣼÙÉUÜ톚(Hsâ7ˆ!Ü»… 5­ØÕÃ)Wqìç2]¤~4>™w 5~˜&‹ÕìŽÁ‚Eâç©Ì,õC)AD?Ê#‚*Ȥ/Ãd1ÒTÇ.TSM%~È)¦»½‚ÇpNRd¾H£Q§O dÑ”Þ+Ô0˜åó߸‹pî'°Œö¡“v¥_£ZK§´1 ’©‹fç"löÒÉwˆíp‘ÀA&÷ÙA+ö“ÜééHøY˜XB0Ô‰Âè Ð’ss†çrP8“ëG½ÌÏrù5±¢oK¸ÅJòï‘Kür½€‘¤ô“ÀÏJÈ—·²phÇÌÁa1²aÑõªlTÁÁó Õ®ÄâÎWõ0<Æ2ú¶ðˆÈ)gLÃ#îÖªhˆ³IÀÓBR×eÑÛè8sêD‚ÿD×¹ ,ÌKÙï‰ê.‘‡É]"„œØŠDÂ#‘cÏÈQ¼™…é½"v%‡ƒÆˆóå9M%ÂÚRùc±µéJh·Ì»F.Zúýº…Ø&&ñ—€%låLL4Æ ~}ìÑXZ¸ÅMãä±âÆñv^sØœÈ=6L( §Æ¦íÏ› ·ìiäg|ìâx *ŠÓ¹³ý6g‹Ð÷ûÒ´yYì]ƒ:È ÙÉ¢ÿ­õe"255pºUMײ#ïá3}Ô$æNU J]Y7â\Æ:ë Ü¿î ‚ž£:”Í ‰f;¦ihÀÆRPn€F, ´£¤Z:†®Â@ ì Cb4û”˜£ˆËÜªÚæFu=íŽm==<®‘ÐÄmiwƒ$õnMz/ô®£í~ ¿)Àk†X U¶–£P'a¡6ËûÚöÜ^7*F>*:Ûgc|Ór¯ýûnc_syìÖ.`P!¹xJnÚkCÜgâõLdÜÙpÌ4ð$2`‡›=©’rLÏOyÀ­ vQ^‹¶ÁÏ1Øjúä›™W©âø†h‹ƒun”d"¿#?F!ƒw§ŽTàb5N9 ˆV¦ò#dˆÝè¶¥ ðýžàÊ*W%1MNp»Õ8Sr‰kËLð@Áb}§\莫LÇó1Hä™*hGk‹´4#6ÈïÃMÊ1R´˜½¨ä¤$3Ó0º$¾ÀãÁ¡À¿Ù¥‹±A]cÇ:þOéÖéJLjÆXßÁËÐpÛ:`JS¥ »gýì(VËÓgfÙ,¡:Ùa"3vÐ+1Në¶r¨ –ò8&SÝPõÝÜR'ý3¼vttDI<¶{Èí˜ÒŒµ;dÌô„È c.<Â6GÛ¢s6“ÝÅÈq¯øYí™íyœ @<ï|—LpÆe­ú=fIáíJ?ÅÈád Š+Ç’æõ8p‡g !õ&ðØ _ýçôÄ2ˆW—×Ú­A&5ãM·¨º;úÏÚy»ù‰´>7tóÁzäU%bU˜Ø*wv…”D2_È€~–eÜdxöìrýìÿɹ endstream endobj 350 0 obj << /Type /Page /Contents 351 0 R /Resources 349 0 R /MediaBox [0 0 612 792] /Parent 324 0 R >> endobj 352 0 obj << /D [350 0 R /XYZ 96.768 705.06 null] >> endobj 18 0 obj << /D [350 0 R /XYZ 97.768 546.438 null] >> endobj 349 0 obj << /Font << /F35 119 0 R /F29 110 0 R /F8 112 0 R /F14 120 0 R /F1 256 0 R /F11 183 0 R /F10 255 0 R /F7 257 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 355 0 obj << /Length 1076 /Filter /FlateDecode >> stream xÚ­W[oÛ6~ϯð’K›•äH–Šf@4@‹b­‡ èú@[´MT ‘Šm ?~¤x(‰4m§X SÒѹ|ç;ý1¿zýÞ2”%Q2š¯FÙ Í’t”Ä3MãÑ<}½Ý¼ú6ÿøúqš ä¦(Å£@ àuMÖhE…´Þ¥( "#IAU|JÕ½MâM£p4 C”űkeGñíÛ‰þ_5ÕRPVýÄÁ~¬ïàŸ‹ lµâDÀ­¨9.Y%jVŒ[Ëa£8JmË;B×ÁA¾$bÃr¸¨Ù®Â%áÒzè‰~w( f¶¾•…À©ì +iGx0v:ùäŸpí¼²dE‹ÉÑ+`‚ \;X¾ŽÃo~Y¶ÝÉF^YòDªc½S¯,]iå(.ZÛ×ïïåvë:d’mÕÓë\ýÚ;*`ý÷Àö!–“BwTlô)€—0-ššðë!,Ñea¨]ºÑr_˜Glˆ>ä’K½» ð– Ûè°&¶Kå°‘zInH˜ÓeÏéI( Ý>S f‚Qp€¡*7·ºt *9v¼cà'4jV‚§Øx*Ì3Zsq)i©Öb£NºÖT,> endobj 356 0 obj << /D [354 0 R /XYZ 96.768 705.06 null] >> endobj 353 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 362 0 obj << /Length 1482 /Filter /FlateDecode >> stream xÚ•ËŽÛ6ðž¯0œ¤‹1%Y¦ ’Ch²=%9heÊ&"‹†H­×ýúÎp(ÙÒÊy\$’ó~pføçí“—ïÃõ,cYijÛrƃeÁ,ŽÖ,Hg·›ÙgïéÂ_kÏì-aÚ¦ÚÝçU몤¾-¥yV(áöÚäÑÖýéW¨ÃiH,ki–ŽNÑŸstÛˆÅ×Û¿f«™Ï9ËÖë^E¾Z¯<é¸Ë¡”{QÕ¸£Ú)^+3@Ú禑¯:5ÙFµw•“»QBw:±+ ô“ªäF–§‘Ç4HÏ·n³W›‘¿ vÚ5Ûv/jðÖ„dž!‡/`î >œèþð’z”‡Ç(s2l>OKPe©…ù±˜£ÛÑ?)ìŠ9µ8jÓäâæ²6b+šI66Û†øìm^UÈò­FKK—¤,¢ñÉ¥Úƒûw:¸­³k9%—|3 ê³uÀ„F“–“Wa¢Þô[{]®ñuž@Æ{avjss3e£êùcÓTùƒ4vɬ{¡jÓ¨ê™8è_#0ª +vªúIµ8â#ÿ´+ j Áà¤ÑOÀü„Q‡kÈãB 2TÌ6=\¾lhÙye;´£4;Zýý†Ñ‚b -àn;.ÒáE—o6Ò(ÍqC ò¦†Ì ƒ½°™£m+´ˆäcÕš)Û:®ZÚ>a¯1¥  Ú=¿b3¨ždNu¤Ê«­jÀ´=•8 àùF@uÙKR!”‘$ð¤­QIГ{•M7(ÔQ’Á9p1mµ) ½É©à“‚-­ÙA‘¶I¯-²ë£$!øJЂ²-¡lC€C¸4)B ­ªÖH›‡€ ¡u,ôk.B±u×ϰ&·ºµ¥w¢oåZ“ë .À´Èx—>™½¢a¶‚C#ší4Ùtìî".78¹¡_¨køÑÊ l–g"9ØI%(ã]]•tÐÞX»Qh2k¯½Äȼ£AOMDò|ÿm[{ÿàІàpÑä盨îÈ vÁšÅúšR8è£~Ðyn\×â,IùЦû¼ŽM4cäi‚—p°¬Í~lsžpÍyÎïø\Ìÿ©#öYNYå[ÇÁMa8Âv²Ø1m#y)k+2qÈ77«)õD¥Å÷ù4â€lÞ¿ùðéÝ#Kñ9Äâu©G†Þé³-=ñ¼{]uÙ¥ Ü´‡&ôòºWïrì¥Ã߬žQÂVIx󈞦`¿p™‚w¢t­Ù“ü.£VŸ:„Ö]ÁÕ0rmì@>ÿ ¶•ü6t» Ç€&&úl…6×½wçîŲ ¥TÏÙNí¿w÷å3jøuY¨ªÊZÜÌ—›š£—×üU§„éà§î¨›þqb6‚9¥.0‚}Î8„;\³4vüC[ ž¼»}ò? Ýx endstream endobj 361 0 obj << /Type /Page /Contents 362 0 R /Resources 360 0 R /MediaBox [0 0 612 792] /Parent 324 0 R /Annots [ 357 0 R 358 0 R 359 0 R ] >> endobj 357 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 450.511 191.928 462.466] /A << /S /GoTo /D (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup2) >> >> endobj 358 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 438.555 197.159 450.511] /A << /S /GoTo /D (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 359 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 313.026 102.639 324.981] /A << /S /GoTo /D (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup1) >> >> endobj 363 0 obj << /D [361 0 R /XYZ 96.768 705.06 null] >> endobj 364 0 obj << /D [361 0 R /XYZ 123.92 453.998 null] >> endobj 365 0 obj << /D [361 0 R /XYZ 123.92 442.042 null] >> endobj 366 0 obj << /D [361 0 R /XYZ 101.643 316.513 null] >> endobj 360 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 371 0 obj << /Length 1332 /Filter /FlateDecode >> stream xÚ¥WKoã6¾çWi‘·×$õ°¦@l€-ÐÚܲ90%1Ñ $'ýõ>äHŽœî¢ÃÔpf8o†ÃßnÎ>]ãh±EÛ„$‹›b E[²Hâ/nòÅmP.ïn~_„d ;xbŒ¶ñxëÓõf¤a‹RL (ÄsÝT|RJƒš)mVQÐŽ¢‡­¬[†däÜí åÈOK rùâ¾”fmÎdŽ–aÅÁU×ô;íH®D¾cµF§©g™Ññ¸Œ“€•^ý®?:¤wµVÈù3 Iˆ#´¢©K•c¤Éˆ‘¢M/ÖŽ•’—a!Z¡ª¥I„¢-˜…ןÒvéR¤h‰OAâ¶jp¯ãuðs¸ cL¡'žéN~ê³£}øøÁ-²Žnõ“ûëŠBqíÖ^Ú5Fô•õ#+ ¡l8kì`ó³­§hPY'¹“w»ûšü¹7uÿž›Ç©tÅêÚH^±²aRÓ•L7(©‰ôËÊû`Lñë=e¥Õjî´–ï•–LÏnf÷¢Í͹ª“ñ6”›OÀ¢ÔƉ©¨Ï ø-ZÍKn³Ðp]uùåå/d×^LÜçÜhYÃM¨×6–wÙí-ËœÕÂg͈·»ºnºœ×ÃqG0ôg{”30 ¡Ä4šQ ¥m&»²—îã–ܹ€á˜˵œIÔKž ƒPå,½Cqu®¸G–:~pycêyÖ=÷29÷i¼°¤‹×èžRAÙ¥t¶óMbud4¯ŸQgƒß94߀0Uù¹hÏfûòp7‹i¿ÝW"«my7Õó篃Œ¹0dZ(§œµÙQu«|ºzÝDcÁP„u?Œ)16-ÇSõ?`ËZˆLðV+ßá._{ØiŒ=1 7 ^Ø0‹)‡×  Æ@^½ÓáÆÖ8•i®ô,…æ`ŽÓ£c }Þ°w«dÒÞ]+ý%%Ý}&–>ßaqÈÏ~â¤QqŽ‹ð¿J/ùþ¡#Å0Rh&ìXvî€=ì^¹¿ñøa?[34€`{ÄßXXž%Iƒ n&Þ–À†Á. “ ‡%\ 'ÎË ™[€K3-º†Ž0Ž7Á/b/×Ù_GQøöBW0ÿÄ)ÀŠ‚D;Z ΕÞ}]c ÚS¼¸ýa„r<ÖíÚG[BÆï—š;ŽÎÜ÷Ó2\HÁW¥;oT7ØÄ禚„¢<òݤÉÛl’áô0H)¯¿â.&t³låÉŽwƒafÄ3ýЭ¬fá§½(øË¡÷&òÁ”ÓÊQÌÔøÖho·#£ßÚ Ð[ãÍ¡«¿9²·î(Æ.ýüÙ± ÀãŽOÍNËÞòdl¹@ùå¾bÚÃîwš¦ÓP9Ô`j aâépàwö‚V kE¿«™æ€£dC‚ë冘ªHÐLKµxv{#•£|ñ¼ìÑõŽh`âø ÷/YYZL+“’™8¼@ #Jà@á•)ÑôõËìèU×9`¯&é73~Á%·Ýj&ö4«·à5o|ücBÍ¡"…ÞDyÖ0}+înîÞ‚#ŽQB ÷Oê"‘Ü?Y†ú¶IÎòÒ£7Òì3…P‚3HÔÏ·¾S ÉG3ú(Eãï}¡Pðô¨]ÿ Ú¬ÞåÃèaªPõËÜÅ0å<7ÓÅߨ:ÿFÖ^vºóìǵ™¢Ä´QŠÑfãP{Ýœ}¾9û±|èY endstream endobj 370 0 obj << /Type /Page /Contents 371 0 R /Resources 369 0 R /MediaBox [0 0 612 792] /Parent 373 0 R /Annots [ 368 0 R ] >> endobj 368 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 601.943 102.639 613.898] /A << /S /GoTo /D (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 372 0 obj << /D [370 0 R /XYZ 96.768 705.06 null] >> endobj 367 0 obj << /D [370 0 R /XYZ 101.643 605.43 null] >> endobj 369 0 obj << /Font << /F14 120 0 R /F8 112 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 376 0 obj << /Length 867 /Filter /FlateDecode >> stream xÚVßo›@ ~ï_òÔ¢„ÀHP4MšÔJÛÓ¤åaRWMW0¸èîH·ýõ»ß! };ƒí³?¶ïÓîfù°J¼,ÈR”z»Ò‹BdYæ¥q ·+¼ÇÛo÷ß¿Þ-”Üâ}YóøG˜„ço¬§'4¿[¬¢$¨æä„©üŽ¢pøqŠ9?žv_¼ë4öQdIòú¶¨÷gÒI^8'eÉ€x¯ŸÁ9ßòËå‘2N…tù8mW¤˜òÚâß5•r”¬­_8ÙÈuœ4@q—Ã+ÏékÏa9n Ü#©¹|ˆâAÉÖ› á„Ú¬ÔE]D( P¶ñhd‘V¬;n0›ÿšæG Œt۱Ϛ]ñ[&¢ÎªF8^·;0|£ÚÑšì¹Ì«ë¶]Ï °ú‚RN[¥br)HÿÜ€Vò}E½¹“Z¦ï×BØšª,}Š÷{( µ)Åôñ$ ¨IÇüåXdwaë™4…¤ž¤¸t *;­†ÑE>hÚâ¿ÛÓK%Ai翃øe®^¡¬yÝÂ;´ =·òˆ¦-:xiÚâCø6;T;U¸9ÕÝ~kšvéc vzØîÿíì£>3SzC5^“–´ÐñÉåtФù™W¤1iŠvŸ†¾ܽð‰~`§Z…MÖJJºŸ/–ñ…å¼nH‰Á Xú¶»=gïIZ3—cê‹qbõá$À» €R3ÓÝ5é×ñðÕäž—ž¾Sæj÷Lb¨çýyLòà3cxU 0³,;Û—„æÀìNiá Lk^‰kêüÜ.p³FGŸ ïig§yþùõb–¦œ(ž0+˜eªÎ½=4dßÔ‡wL á<ìíA[»1)oÛNlëœsñ• Þ»$×í{F¬t0JÈ35¤_Õ›L:E›š“U´…+k.–ˆ¹ ¯0µLïy‡[`OF¶y”¤²3¯ló8È6È*ÌråÌD>ëÝIÖÔ t'*ÔgæÑ 8»Ì`¦sêš³_fcñÇi —&¾ýTDñz\ɽõ8R¡_÷d×·Ï@í›Ð¼E´ª7é‚ÏvòŸÀ¾`XoûE·Èòa3Ìdk´òqlâ•i…¤âÍýîæ&­ì‹ endstream endobj 375 0 obj << /Type /Page /Contents 376 0 R /Resources 374 0 R /MediaBox [0 0 612 792] /Parent 373 0 R >> endobj 377 0 obj << /D [375 0 R /XYZ 96.768 705.06 null] >> endobj 374 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 380 0 obj << /Length 811 /Filter /FlateDecode >> stream xÚV]o›0}ï¯à±MóÎÈǪNê¶tÊ´uUЦIÛT¹à$‚#Û4í~ýl|àÔ<ÅÁ÷œ{®ïñ…ÑÙàvzÓþt„F^´ô4ìO‘7 Ç}$7¢Äû}>è\ôBž§DèÏþ®—¸Hô"¦¸ð¼3¸ø}õ|¯ýiLEÉ Ä_럂Ñ-ÿã‡~LŸ1Cr\9‘r÷¢ÈÀ2¦ù)È5+šóy!HJ˜êf$ÙÈb‚p¼‹ù‚à\Aå–7Ò8AóÇxEs«Ð=Vn†‹˜´hÆ/™ 쀠.]43$”Ç8'Í `3¨ævÓ³bSBÛ1KË59Òbi&¬´‹ÙÍ7•‘—ì¹ùи »Á5¦Kh¹²»UÏR—•UóÞ’,]™ªDÃ^óAÓå’wÁz«Ê)]óü.š}™-ªãR»í`Ô Fm‡Í°À–è:¸Úms†COüÂc ŽWðgƒcÒ]IW÷ú=Ø« ¿Ù>aKË×þiwcîÔèã5Ýý¹ÎeLèLâ´>98R}ÍÊ's ;êDÁ"8Ïi¬"j3œvô´«=B]9Ýý ¦Ë} ÅÝh[×NçÑ›‘=@¦d8}=²AÈNök ­çUÝÐÝ1<Í“7¹k3Ð?ÉÔ• sÅK*b¼ §©éf ¿òVcé—”§F È¢fA ™ ú÷ %å&Ïb,HP3ψ¡eºj-`KŒDX¤4+R;Óš&Ùòƾ¹+IIì¨]å[üjžd0<-EVìöy»$ (Ù¤œ$}x> endobj 381 0 obj << /D [379 0 R /XYZ 96.768 705.06 null] >> endobj 378 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 384 0 obj << /Length 816 /Filter /FlateDecode >> stream xÚ½V]OÛ0}çWäq´äËmÒD¬ Ó&¨DhCÈMܑĕí0öïgÇÎ'I×jÓžì:÷Þsî¹·×þž˜—3Gó ß®n4Ì h®³0ÿÆÚÇ(ƒìTw€óa)—˜äí‡åX8׈Á©ü¿BrvªÛ΢ÞWÇÜØ>?} ¿i–¦Û¶á;­èàÀr×:˜ô ÆÐDJ¶3'C&é’{ö¬6¸`»BåO)"VD;a\Æ ~»º ƒ‹PäDÒ„v•Ì^®a†bññ>¸¸û~{VƒæüyŒ*—ïf(bÓéàg!è ÝᇠoVOÁUÃHaZj‹]šD\a”QöÒ¼æÜ¸ÂYŸ®„síÛ-…ªEѶeý­V˜¦8ºçYcQdK@5²U¥åZtô¨x`¬_ösjMD b¦x›&/ÇTbvH®û‹"A‡Ë¢&C0§Çðÿ“Ú”°ÃŒ%£±"ш!ÊŽ!ìBØÞËV‚Ö˜%늮"ôÞW ŠM · É´\÷v‡ ½å+0Õd® ¹wð%X‰@%£Qß„!rßÅ_󘊶ؖ|ñnwÄßkF`ÄzC^4˜ÜmÎäÁèYÝ ¸¹] ­\µËà ¢½xÜ>y„Æn›RÂgüf,Á ;+*Ä T]9pǼnÊvÕ_Á-¥Frˆ}Ì Šâê×t*Vi51§ª?Q²}fô!yïáÆ*ªh¶òà‡2ì£d¨wbºý˜æ²Õ8ãþ¯¼0ÿo ô“®ú²!¡¹%›†IŒiSUÖåÒéOÏãîîkO›7° *‹ªï`J±ê¸¦n9‚¶üJëá¡ÚJ_x°Ü±j«3/½6ülnxÏ×ç‹']fsaw„'¿Ü['í endstream endobj 383 0 obj << /Type /Page /Contents 384 0 R /Resources 382 0 R /MediaBox [0 0 612 792] /Parent 373 0 R >> endobj 385 0 obj << /D [383 0 R /XYZ 96.768 705.06 null] >> endobj 382 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 390 0 obj << /Length 1070 /Filter /FlateDecode >> stream xÚÝWMs£8½çW¸rò8cć¡&ÞÃNm{ÙøjÙd*Xr!‘¿H&;5sØÚK¢Õýôú©»ýûþfóD‹¤1ŒûãÂ#°ƒEm”_öùâqydÕ§u£åß^ä]pÅÝy_Ú¥öõžÖçÎÒÝ´ö?=íÿ\@¤Ð_¬}¤‘ö)ðùÒšßíÚç+&§BðÇvûS»¸jGtà*xÆ^PõHžŒ‘ ñ¥‰±ž BŽî.ào^ƒ¬ù¿[Ýu p£þ¶~=×#.9¾²ÓŠ_†Zn4ì UÚ:CfÓ˜þÀ(ˆ ó&ÏGж«Íž¡KF¸]Äñ/IÞæÁ{ú€^Ò$AZç­‰#¡µŸÀ‹¡ z˜9}À«è'28'ÙöqFo,4q%8€Ô™9L~¨µ%ëÐI¸¾O=Q‰OÃkçèCÃj#² îcàGpžú„!t`ĨdCè©}!\«‘–ï:uEqn4ªiÏëÊnÉñ¡>Ô«:šŒ§ƒëÆ)ËÏVÉX O£K …½žòòìÿIyñø“ÿ±(åU„’<ÌGΚ¤Áïü+Ü'‰ÜÏRoAšF3 6÷äžJZÌ‹Nxóâp¹ÙurWURo°%÷ JAeç…÷å98÷1c.µñOÞæP ×·?JÊd—hzÔ‹Ÿ&`ÙK2}4˜‚$‰MHR'jöJRBÏ[~ÃbP¤ð;²î–æh±x7Ó€~ÚÞ¨¶¬ôöûöQ²“¢ÃzÍY}(qv _ØÂ²ÿT„?ú˜éÝøí¢¢ôw(¶eî€3vÆZA˜HŸ•QÙQ Íô½êYlû”蒜‰p«6ªXMu­Þzž®Æó™qm_ *•3ˆaú¡™{(ŒjØ…]ê ÂèÔ8aš9Ær‡Q% ]ý}'­ëМGs.ÛÇ»¹¦ ÂË(-özËWö6Kã™åØ´¤‚½ªl]ç=ÚFËï#Ý ¦»W-ݱ¨OÚú•5æ:ÎÁRf\ gmˆ²†Øv®ÞG©á@¥hîH9³-™hHï–úÓ”Ö—Az¤+Ô'úyÂT®—óbü‹†W§¥¯3V°Jõ‚TÓÒðä¼€¹0ˆ¸Å¨äcá:•±þ5á€édš¾SóSO^C* ˆq›íA*t+Š{H¶¡éUèt$"\+uO8‹RckKÚù'€$.Éÿ6Žªä¼˜ˆ§` ‡‘ÒD~:97$NûÝ‚8 í<ר‘2¼ùcó» endstream endobj 389 0 obj << /Type /Page /Contents 390 0 R /Resources 388 0 R /MediaBox [0 0 612 792] /Parent 373 0 R /Annots [ 386 0 R 387 0 R ] >> endobj 386 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 151.631 191.928 163.587] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter2) >> >> endobj 387 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 139.676 202.389 151.631] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 391 0 obj << /D [389 0 R /XYZ 96.768 705.06 null] >> endobj 392 0 obj << /D [389 0 R /XYZ 123.92 155.118 null] >> endobj 393 0 obj << /D [389 0 R /XYZ 123.92 143.163 null] >> endobj 388 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 398 0 obj << /Length 3090 /Filter /FlateDecode >> stream xÚÝ[Ýsã¶÷_¡GzÆbðM°3yH:ɤ>´7iÆñOâÙHéØì“úÜËc [òv[óœ MpN·Éº»7àO÷¡­J Ch1k«’Y[%.·UúÕ·jL+‰7_@ë)´íBš˜éäüm÷a ÒV›m¨Y9÷Mý¶Í*ÖŠ®È}HuáM‰èݶÃÉuU~ƹh4îýbãæ"©Œæa¤‡iù§lÕá;nªO3ÐúX¸äïÐÌÕ¶iòʧ ìXçL ¤/·â š`b¬Éx!»WnúÕw·W¿]qx‹-øB0¨U òkÅckìbµ¹º»g‹5 áX¦vñÑOÝ,DœÚZåâÇ«\}»_õ¦¡êQž’1)™iGœd‚RÁ n1?éa;¥‰žx¯ùŒýû{ëúˆÝg7ý×ÃÞ!õ—øL"ôYzÊúaY‚…º”X8ŸÞ ¹K`toŒ;4V'¶¾Ù[¬]Õ 5}Q•¯ºº 9š…¢TõTþy¸ÈR9>„œZV3 vZPE?äm[d´°Mñ)»À,ÿú!ä,åýx PµK/ÜÕLk•ôdB‘–ÅâK+Cý+_küZ‚§Çã@Ô“huñÔ qÊ„C²HBÔÆj€r,d4ç.°"x¼tu^ôGY ™uÊ7c/ÄWp› ¦jÒ+Ø·&¯‰é˜ÛÁ?Ž‘±M†9Ý! ë4Ñàîï>@ÌüÀ•Hv25'¹23¸R/peÇ\A”HlÐ’HmÎ’Õ)K–jÏ’?„-9adÉúœš:DL–fÕÔz);&õËoU!Rö¥‹mPÀb ¶jÒ øÎ1ްÎMlÙ´R{:\1ñit4çaMÌ‹júQÍÌŒjöü¨vdMlÔLBš€Ü§gЂŒm“õ&ÁíW&Lp¯“/í5H1Yp¶•NVC3}AŸ”s* Bøs…ÔR (¦í\aõ aõ[ÅLK—¬ÀØÞ9kÌ=yÓþ+ÃG'[m©>ÃONÇ̳Nc§c¦š3ÕŒ˜©^=fNhÌsÙ×KmѰö3 Ò™Lй³:u˜IT¬íô03­‘L U’\ÈTÄV¥oœûC©ê_æôš‘ɤœÞW—å{áf)TT®ë¾¸x†ùãΧ©ýÓr1O¹‰!ó|H¿Ñ“/Šè!3JMìNSãP~AÅ«‹KFó9ÁÜ^"˜Ûó¿‚þAƒùÝ‘L¼Íå±h¾9ÁÀ ÍÌ2Lü¯Ë°“ºº¿ ²­Xú%©¯ (ËÂ[âÕtU†ddúUìêþxb—JÃ{úTbçÎM‰<‘Úá`KP¼”ÎZß>³C†°¯ÂU)öxæåþÙ‡9ùöEë9§óy—gÿ?ýsŠV s|⹆ñ1ãç,O8‹+!Øžp›‚°âTì.¨Lê?M‘¾í‡¹@”S@:Ll¥œ2c†Òø½©éKëÉ·aú-sÆQÉ༡¹ÿ@;|³òˆ•ö7gi”f˜DG¨)ÙH/_þ± ’‹RÓC0ª"ŽÜÐÅ °E|æÈÞ>õCkl0üWƒ,ÍÇ¢ÍoBg* >m‡:⬋ύ¥tÂwÑî)òq1Ä»½j}ÈWÄ6"sz쇻ŸÒÒË`ZJHÓóŠÛ™çžËx…—¨WAÿH(i{ŸsGd{ ¬ o¥ìÄèÝÅŒ_›ëÈßsE@°]ö½¡ 3 Js…õØ"K¤\“‡¹;ÔîîÁF†©.0\ü6Äñ†*éÑFÐ1Eù‹cèÝäÞ· åáIã@Dñ•.³±5bï¢=té Ø¶EVâ°nþß”_ìC\Ýá%mbÌΧ!ÂA‘åÜï—ÃEƒ:ò4Ìa´‚Æû’pò}ã}ßù¹eÇüœÉØðAŸ¶1—OÄžKo.I¡8PîÎd¶³ˆjEséBûøqÃý1bZÚ)l'‘ƒ#fÅ zÑ}ð0;ÌFè¦VÄ&•s·ð@Wªdjª·Z"°‘Öµæ×é5ä¯ðƒ¯Hxu¯é ÿµ›¬,qhUožËb5LÖ•9¡ïì tëÈnËÌÅ¡£=–ñ3q©.?Ž4E…‘è7½p®í]FÛzC¯C™X½s^áÂÞ3Ú!Ÿ„õ‚oeÔ—cWQ­JHÈ4ÈTnC’p很7-”¾@lj5¬ð賆ƺÉô`8xö¡ÂOpnŽÍqéñÂ*(Äe/ÄXkíÔø¶í6+ V†È»XLìÈÒKöBŽ´q*èHx.¨¿Ëž¨"ÇÇš–òI˜(òɪr·*œTÓv£ÕˆSñ•DßióU]­oBX>B]‰ˆï½>ÔyC³ÞêHX§¢d¢7J×ó~BD *„+$½!vz1u64»¾êÇ-ÂÓ»¦FÐ}EÏÎ]Ò…‹ƒcÑ(åe„´ÄüáH•’›z»;ù^drW‰(ŽžGŒ $Àäܤ¼_oKBWîvR9 |?:v;¨ú__¸ñ6÷•Ž’ûÐ{˜…{ë@šN ™àïbÓþž"f^Nfàö«ÚÃÒ÷vhÇŠ€œd _ìÔ@ œ¦;îùèéñ«OýBÝ È%ØW3v,3BÉM‰AC9żÙöYßuPqžÎ…ów%Í-‚å–„uò쑎è°Y4›Üååݯ&z7ëÁwtYy(…?&9çê©öÿ¶YÓvÔý+H@Ü}˜˜þ¬ÝB¹ªiû®k²Ö*¡Mw#xxq]ôshy»Ì^ùRë|îºiëß{$¨£†Œ}ÆQ º®ß•jÛnøõPB\7ôŒÞѸË×qÈ‚ñ‡(“c—²îG!-¶VømººÛŽÎFÛªíh¿áÉ#øO.h¿ŽÎo¤Ê9rßðïy‡óÁϦѷù*Û¶ùx|D7ÿ„ÂÕ!©‹¬ ˜Úž´óPlÁã¼I'8—¯CvÍ¡¼2®ãððáÀ|gÕ}à<Èuª’—¸Þ!·ôô@ÕÇÈÄå÷âêØŸVôñ*oè¼ËPõôÆ‘¯?‰|áè$OÞÅ´ûìøC)háãTÙÛ‡GW—k}ƒ}“Ú?ñÀ{8¬`|6±´ Ë°Bšþ³újœ<Ÿ endstream endobj 397 0 obj << /Type /Page /Contents 398 0 R /Resources 396 0 R /MediaBox [0 0 612 792] /Parent 373 0 R >> endobj 399 0 obj << /D [397 0 R /XYZ 96.768 705.06 null] >> endobj 396 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F1 256 0 R /F11 183 0 R /F10 255 0 R >> /ProcSet [ /PDF /Text ] >> endobj 403 0 obj << /Length 2251 /Filter /FlateDecode >> stream xÚ•ËnÛÊuŸ¯º¹t1>Å&.Ð ôíâh t‘d1"G"mŠT9¤íܯïy MI´œnÄ™3gÎû5úõîݧß6«ÜÏÓ0]ÝíVyægéf•&™FÉê®\}õžn§íÍ:cOã'òJ=ÈÊšOº–Ugõ Þéò¯›ÞF3¼oñw¢OõP1ú7Åt4ðw{³¦Ë|Øí„"HPÕ»—ÅM˜y•n÷Æ2`è&yÍŸD9Ò7ßïþ± Vk¥ü<]‡úTÃ<óJsdæm)Â2¼¸Y+¯CÀãM’zº¯´$£Èûë OÓ¥¼9ê~¨‹±Ñ=ï›ÇLºúF¡Z™»=T"FáX%ŠÁŽ·)™L×6ÂjËlA·O¿©xæÚ(ýMƒÆ¤é· Hkîpœ;”ÈQ$; H˜bÁ  À(h¿u”…~F§f´#\ ïM>Žðl³Û¥œv»Â2öFô+.TÈgU7ö ™9Èñ–êÖF—¼Á@A„ o‘°Wà¥$H¼»Jî÷fßc (ÏÖ]Ë0°7iû-PiQ;Ï/DÊ”‘ ÁÝØ"s¥<ÅÍû]‰ÎZ½˜% Ì(Ám5ó `´žfGtãÀ°0—í€òíg=H ¨y D‘dÊyÙ ú2Tê§AâP8ښшºE76¢­‡ÄBœûi–žJØyž2 ð æê¥ÐNCô¼,Äø¦¯Û=%Xàýçfy†Ï9‚(Kñ"\b©yþBnë—4FüIª^ïåÄÍ¥¹Â<òs5Ë›P]Úkã‡ÁdQr=ÒëMÛõ‘Œ3:@Ïfž©÷•«€$Ž ¡¯T"L™^k c­î öh‰<ôþ]·…qàÕqÎ!#ž­ó á -2¯³¶Þ6Úu=#jþtSu7Z>†4”s¬¡ø­œÍ2¢úrÓÎ1±)ùò³h²|#R'fF·KYõ|‹1‚R˜!¸¤5 &špÌk¤àôŠn»Î+ 7å}Dxä=U5÷FŸ› o5”6pPšÑ¯9yᘫ@¨¶„èOVŽ„œSÓ0ð$~?.6›ªFcG®o™/<0ˆb¯ìZa›ãÔ…MâiûyŸÒçB27ÚŠ>Ò[A¦B7 :öÄÄAlhYø§ŠÊ1êë@á¢o{,ÖJËg{ãVcuN”~Oí2@¾Zp -Ü®á|2bI±s-6©WÖ$ÅØ hÊ ÜjK K*¶ðÅÚdž=°Ÿ²8nЕ|·v %nÄX»­ ²á¼h¸¥Ð^Ú=¥pÆ@y{ÙA1ojÓûKá!æIc-XÚb£É0¼ÒGNlÓ ¦ø-¥A ?ËýŒŽuè¼%(›N[ÌgœJj±è&¦DÊBo;žPˆ¼ÍD½Ä) ,í–åhN)ŸÎ]©ë0dxÔ Êü–ƒ+ YþÈ;JðhJJàÈdúƒ)y6BèïæièÚõ¿ô±²\9ê¡¢¨‹¦‚¨œÕp®ùSŒƒTÊÚMcˆ¶Ø¢8óãlTls\‘% ˵>jýt*T§×µÂƒÞ×­†.å/dH‚¤jÃ’VŒ¥óT…‡Ï4Ìêý®âµ.Ëñ¸@^I³¬®…Zòµ[Ž‹4óžÚŒ¬$Lhx¯oƒÏ¼«¿´ºw›f™:—C…¯/Çg· È:Ïýè|Ü¿Öß™6r\p¥~!]û~’û~.÷ýroÞ–û,‰ÀÉHöõ~Ž¿‹:ÈŸDYƒ(yxVµö’ȯÉŠÑ…ó¿Z°ÍÑôÐ`&ñöK M¥<½j«0қ読Ô^kéYhý±ÅW«Î"R0zzÏ€½> endobj 400 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 464.458 171.007 476.413] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) >> >> endobj 404 0 obj << /D [402 0 R /XYZ 96.768 705.06 null] >> endobj 405 0 obj << /D [402 0 R /XYZ 97.768 467.945 null] >> endobj 401 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F11 183 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 410 0 obj << /Length 727 /Filter /FlateDecode >> stream xÚÍVMsÚ0½çWøÔI ¶\ÉÖ.i=s£ ‚-ÆIúï+ëÃ`c¦Óvz²¤}Þ}Ú]=éëüaò!‚!öæ/ „xÅ*Ë<õ£Ñ“à x¤iúä#ˆ%+²Ò…ýî˜ÄöÀ_?› “kð´œ÷ÏC@s9éZå¥,¨œÏÆ<ý€Æk"e¹ÈºØ 2Fs&é —¼àåë0‚½†QûGV”"¯zÉL§ù©dé³²‡•ƒÉKøå² *¯Æ›A4 å‡$!͸GËÊ|JQÈpaâ-»sº9Ódo,—‹ãr6 zX%À d…!ˆ1j&]'t<î$Ð(Èx暆*ø‹ÚXìÇQ“׿hÞzyd>ïŒow²ìóvÑ0.vàvIy]M>Íßhá&ãq_Aa IsGˆbÜŒJ|yMp´*¨2]m wÝ×d÷ÓwÃ3UOq#0l^gTVö7ÂÛñ¾¦ÒÚ¸ßå|Ûµ} „­~Ö½ÛÓ;[ä\ÈÏ·f’‰ÂêPÎXÚÒ¨•5¥"·£:QÊ…‹•úÝý´*Â%ëp1l ¤„¸OòŒ5OºǾ&Wiê`›¼ò´Rš»–vï'ÙRseß³µ´)y߉Ò$-¸¦Ãû.…¢:‡w\]ýëW¥ÿS}C¨S¡¿ËKSÿÙ# ˆQ4¬¦q ’vš¯Õ.t¨Ûbª³£*bV¦ç<%jU0úÚùþmÁôÿ™`â+ÁôÿÁŒûÓ¿K0ÑßLÿ fË„ïÐRUψ;d`-²ãI²Ö©¥o¬ [æ„QÍ:EË]ŸïÅUË)9ül mʨú±¥ Uþ©}æ}²Ü`øöú§Ü¾MÏÔËœ>8A²§WUsÒÆÌ‹rW‰ÃpY1&ãI㽆@T ' IÃ.J*à÷ùÃ/ƒß endstream endobj 409 0 obj << /Type /Page /Contents 410 0 R /Resources 408 0 R /MediaBox [0 0 612 792] /Parent 406 0 R >> endobj 411 0 obj << /D [409 0 R /XYZ 96.768 705.06 null] >> endobj 408 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 414 0 obj << /Length 829 /Filter /FlateDecode >> stream xÚ½WMsÚ0½çWxzJ Ëà¯Iè¡3ÍLzé…™(ÓQŒl„Ab,AÒK{%´2ر¡™¶9IHë·O«·«åÓôjø0 œÄMB?t¦™ƒÆc7#' "×W;Ó…3»î oæÓ/Žç r“VIVröãYÞ ?¸žxwmF‚Ù“Œ—fÿ»xT~Ñ{¶Ç¥ýÑï«]¤?> ñ)aßsý$V ̘ÔÎ4Hw…u¯ØŸÑy“QØÂhU1Z2Zý=£¨Î(Ý`©IÍV–W¸·ü@} ¼%>jÙjú<úuÃÁr)þäbŠÉ–”‚38~q/d‰%žÑÃ8·ËÝQIb7 Fg£‚¢ÈG¨~[/%šóVÂ4;ò’ogÛ¹Y¸7ƒ¤r wøõT\´â”T5¯äe«1‰:ïv~Dè™á™Ð|)…ÚjE[Æ7ƲxÚA«-Ó;ý~g*)u$C©²÷Gn‚лæUCÅJóׇí¥\9ÕâîRgÊMèë„sÚ4¯ÓÇdÏy÷0_UTcB5Ôz*A²'L*/“ êr«ë ÏÆ8ŒÜ$ Ú²¶!pvw¹T×"sö»Óêm¿º öwV]xRÍÿ·àüÎzý/ç·V÷1RÏqë–÷Ö’¡*Þ“Zu+êåØ {m^zPòAær P ³Ä9©Š’™ìñzkK ‚͹”œ™9߸xf­ØâÖL3új‘î9*ßÀÁf'¤åÂép­lÍŸ‰5X«ºÎv,•ÔÒÒ’ø`Àê#“d])¢ƒâi‚Б¶Ri3h½†'ˆ(K K8ú“ ã¤B©xªÃý´ñ“„¹m½Œ%ö•‘z¨« ¤XAÊ—êe5 [¬ŸzxowyN„öt#¤quäAÀtÕ ,?&}ØJ+¢xG¨´ò˜¦»²~‹xñÁ̪¹µýp¨FÁÑ—9)+÷•¨h—×ðR‡‘á'¡Gx—mÁÇ£ ;¶=„> endobj 415 0 obj << /D [413 0 R /XYZ 96.768 705.06 null] >> endobj 412 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 418 0 obj << /Length 699 /Filter /FlateDecode >> stream xÚÅVÁ²š0Ýû,}8$’yÏE;Ó·èÚe:ô5Šà@Ô÷ùMHTAít¦]È͹''÷žðu1òßÃÈ!€`„ÅÊ83:8š$f™³sº?¼xŠÆsõ Ÿ‡Ax{š”§b __’Åw'p<‰ôÂŒå^-qõR‰e]••ŠÈl¼ª7öVœÒêò2™È\r¹ÿ§mÞ(ˆÄT©ck! œÍ̬é’%OLѱز™íu3Ûöf¶¿l2üا\p\n“‡›‘¡HÅl©ÃÉœB&ʺVSÐ2Õ!Ýá «GWVð8+¾ B@ Šð]Ûž]W¥ú’ejPÑú˜óZ×VÁK5:ªGZè0&´R£ë±¦y®;áD‹ Â%Šo˜þÂÙžªÑ¡øVR¾í«”æ¬á|}2Mù¦¶–ëu¨7™®Ã­ê²xÛÕ鉟†ë#§ðn}Šy€ 1Ó›ë²âp©r%VºlucÛ(½<$ó9 Ç¢ð]Vx‰:I¤x“‰•@O×&ØÝS¾)3_ÒÊÊ㯜6ŒR† ¨–9z·Þséª*‹ŸgnåSÐs¾SÁñ™²õ†×BWt“xhÌv ¹y¹nØ"ë[-óì÷õ5U‘éo¦Ü¦ñ_™ßÏUœãuÙDƒµ¤ù(Å>¤áLh¥’Ûù3+Ÿ³¾‘;6+”þ!|¶ÇN§¾¡7¢·fxj"ÈÉ›¹÷¥i¹¹Èk°`0É]%ó6ñ.Ò^W^ ¤kŶók_6—¾s78žpˆx>um´ µí Ê;tý\\…W©hÖ<“¡z°¼k`aHÜ10…Ú5 ÜûÛÑ·ÙÀ70[T¬ÞÝ]ÏŠìÓl¥Àÿü¡ÂC?T×=nÆSÇâ@‰xÄjÉ´)üÑ·Åè7§IŸç endstream endobj 417 0 obj << /Type /Page /Contents 418 0 R /Resources 416 0 R /MediaBox [0 0 612 792] /Parent 406 0 R >> endobj 419 0 obj << /D [417 0 R /XYZ 96.768 705.06 null] >> endobj 416 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 428 0 obj << /Length 1626 /Filter /FlateDecode >> stream xÚ­XKsã6 ¾ï¯ð­Š7fD½Õm.»ÓN§Ó“;=¤i‡‘i‹‰Qr’þú‚$[²œîv{2H‚ø€>úãúÝÍg?\¤,¼h±Þ.x²8òQ3V֛ų­›«Uè…Înè>ÞºpôøCuM?xÿVùÕýúç›Ï<85é¹ÌK“…KÆPeäu•¦ÌãÅŠs–†ä5+E{§îïïÇs»îX}gT+p–z|v)à, üß,M½(öú£Ü,ñì²Ú\­8`ãÔ[œYÖ•]!ZUW8SÔõ¥å NNà@˜B Ñü﹄;ç:VÕÅöŠ;¹Dùc#uQ_y±ó ‚N§[<Ø%i,'Öòh£ªä0iÌ‚x¸dÕÊr{ËÏO„ÌõÒAO£Õª&ߺmTÖ¯x->daNn³’™ÔZ4¯ß_­‚ p`îûž ±‘(>›ÈÍèÒçÌO†{¾¥ìã'1‹ƒá¸Oç–¸ƒ›VÉ FFÏVE?ã‘»P$p–/ô¹ZËÁSSw»|p‹ÂiÏ]‹‚™É¡œJ#æóþ¸6Äf¦ƒ”_„½/p4mfµ(IRUÖÈ<µ8ÕÖô‹ÚÜÑ]©¯¡8ÜØÑ´†7ˆrVwÅ5¹/DÖ›&{ϪÍqYàDi6wE«ö…Êl 2{–IÇø‰Î*˜1äAúàÚH¦²Í/-уy›Éf½QSq¢ ùíÜU&œ…Ç,œ¹Ê”¹<U ØÒ¥( Q˜žsÑA”0ž|­ùB4;:\]:GðŒE ‚ÉÑØ!¾7((c"(Ž(ƒÝ÷*ì(~Ð÷ªÄyQåñŽÎ{¼íˆ É© £ÓReI<E춪 VÐ~škaÊ<ø¶õÇžéÆX0*ûÅ*â,N&=¨ïüŸUcĈƧ=Ç5ŽŸs•å(æB£°QÛ-Ô|Õß ‹‘³ÆPbüxëƒw Œ`Ôj·Ÿ1,Hƒ·ÐXAùQ2v[Ô»B=ݹ÷é-þTò¹x*ÀÃ8ÐÑâ Q²UƒŸA³åV Ðáðo ‰ßSÄa8y4 φ@úži3¾¬O¢ÈÌ7y“Îꦟ“ýuN<Ï1!50!uÊ„1!;¸Yvšl'¾#r@Ù zŸ%1ÓǸuîõ«KQS¼sôG‹!ÕÁćËti¹-Än´!ËëBê§WÏÊ•¢2§¾îc/þ2jæèæ.-Ëu]äÑ„Ý,Px®;\CÍ~ƒ£‡Wʉ¾DTu–ÖÞðù{[êÌ€DDò[nõÈä•ßßR¬é¥8‡û$;Õöè»/¶u!ö·îF$°;|“PC N¡¸G7°<±ý%ŒúKÌVUJçs&e±7î0@õòt†…OŽ0 FZθð"Æ ¾E™ùîýÛ~3tƒÙŠK¥VÉ—i“ŸM²%Ýú¯òÐÕ%$±¢Î&ê®íLúîd•‘“¾ûµ¹¢þ°=~\t+©G =³Ü×}g±«3%7”‹â_FÃu.Èî³x%u)íG߯&ªŠcxò¦÷1UÊYžõí­ôÿ{TÖÅÆÜô´#ÒÜlwxCŸª`è LB½ü¸|ãµ:yšæc.´¶D4öOÈ¢pP"+ÝÈÇ}óìÇÕn܈Fºnw²Õ()äç5Žì +Žd‚ìl ÅòLµfíõ*ñCDቨ¥ìðÜÌúJ#;ÃÑí~ŒK’?Ìg#QjÎoê ðRC¥9ÒtaÛé^ 6$ãL½šc@žÇ°ï;ÕÎñopsÒ6ÀN€Zñ"biì'¶{1qhxzµÐ=iÉÉèmÆ¢4‚ëQûzöÑèÅþðÊ¿Âë*þ.Û ±¨ºàƒ :!\fÞܪª‰ª–¢É ª9.ز5 ¤‘:棅/®“š‡ãþ”6ë³ÿ*¨œN˜¨æ1ùPÝÐVCL“…Ç"/ÅÍÎ`¹â,ð $ŽZG@OAB@ s†`íÀ#¶„ÇG ½ò`žÞÈÝÅ®fž—8Tö¿È€š ¼zXÑ {Cï~\¿ûïÙB endstream endobj 427 0 obj << /Type /Page /Contents 428 0 R /Resources 426 0 R /MediaBox [0 0 612 792] /Parent 406 0 R /Annots [ 420 0 R 421 0 R 422 0 R 423 0 R 424 0 R 425 0 R ] >> endobj 420 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 502.316 102.639 514.271] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter1) >> >> endobj 421 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [159.91 502.316 165.777 514.271] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter3) >> >> endobj 422 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [102.002 478.406 107.869 490.361] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup1) >> >> endobj 423 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 478.406 176.237 490.361] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup3) >> >> endobj 424 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 299.078 128.791 311.034] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 425 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 299.078 202.389 311.034] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 429 0 obj << /D [427 0 R /XYZ 96.768 705.06 null] >> endobj 394 0 obj << /D [427 0 R /XYZ 101.643 505.803 null] >> endobj 407 0 obj << /D [427 0 R /XYZ 106.873 481.893 null] >> endobj 395 0 obj << /D [427 0 R /XYZ 127.794 302.565 null] >> endobj 426 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R /F7 257 0 R >> /ProcSet [ /PDF /Text ] >> endobj 439 0 obj << /Length 2364 /Filter /FlateDecode >> stream xÚÍÛnãºñ}¿Â݇S9‰‘"ui6mqÝíC‘>mcÓ¶6²äêï¢?ß©[”œ=@ö!Öh8Î}†ÊÞÝ~LWËb¯ö+,Õ*V ‘Z=ìVŸÎÖ†ÁÇuU½ÞD" ÚãšfýåáÏ·9± S*]…´÷‰(Æg(¦ü*0±ÌòÖԺͫòßUдºnié’{"w¢…KsqÈ:צió“n—daÈRÅý±ÿ¹$ªd³,‰R$ŠW‹á×Q©Ðñ Gdã\ÎÔ³‘L†Ñp’â/-€‡ˆÕftXXJ<óf¦`¯ª!nàš]–±4‹ý‘¦Õ/’ñ¨÷J^"H ¦B…s–)傇EQ°­Ö‘;÷ªËr 8UMKP lM¹† ×tÛ­iš}Wж‘&Ñ+š¡X6X¯*vËÚD1˜®'c¤ŒH2e€P8eÄ ‚Æú#æA¾§'¹ €^Ê´—2õRÅQ7€ HF‡¢æ5¨}°ÐÎù,—Œ4e¡ìÕy}!“ƒ¿¡´—w±.x´§¬ÍÁ”X%Œ“÷‰¾Ü p­9o(x¦X"2Ö.ÅŽºxÎËfZ¤‚O%ì c'K¨@–Š!·º1“eðBq¨j(E'¢>i:}ÁaØW8}>¯–-øÑõB$ &†ÂÔVtÚQ=`E;¥>X#ú±k½ ®8ÜTE‡Å”Þ(Ñ ="GÅÓ\'ãÚd­ sÕNŽÕ!âIÈAð&/5A¨w±?d[Qîצíê’p6ê0¯Çe%òeeT_*€jW.Ê}UŸlXrC‡o8”ù«…l©êÂÉCàåqaÍú”-{ jÝ̶/kq3Ó^%¨KˆTƒJþk:[a.ycn|‡²ÖÎ _šoNãC%œV{QÇZ·Õlk[]V„#ˆ'*'AþýRP²„#TT]M$¨¡Š™õ¥Ð$àBSLBo7ÅÊRâIc§Cžå¡!Ø:¤ª·Ôy rC¬‘¿[=¸ŠY%¿ý±˜ï“Ó¥’¦©bT ¢à¯æÒVåæoú|ll‰¬A¬¿TÈçÓrý ‰D Oô<Ãho»B×vϾ®›Ò„ïÂe¶€´xò}¹8dÞ´mó JÌ¢$þ8ëõ#|þq$Ieðiï Of‡I¬&Þ\©†Iœ3g}¡¡Zc¶9ªf/N2¤¼ˆ¢ zt2ë6DW¥mŠqZ¶dð*èÑ_Z¦±Wb5—É…¹!üV—¸éˆ^@ô¼êjÀpˆ½Åáñ{z~­r8p8‘<¸@[4Ú±„Éx!{É€ˆg^5Ê‚›xlmÀÛ׃“a† ÊwZ¹äØ;·C›E¶%Ã[ãg(Àjw@©Ûß6„Fö;ƒnŽ„Àhiܹ—ã@HæàLù•÷càªE=wU5ÊmÿZºµÂÁmøÓÕßÍgñ$ÌÒd¸¥öy+7hÃn*cbtk\¸l ³>¬ï)cÎ’”O…¿½Zo”ÀÏ8!TTp_²ÐÕ풺܇ô¼[oâžžÝ=§'„‰cëÙ[Ûó621ÈðùžæhÝ)ï%»ŒÎ¼BSÝó;·ß>Ü{ݾáëÝŒ]u}=¤üÄÂæÕx°àÞƾ¬¢™A^ñ¥œùRïvÝyá¨JP&ßr&Ø–EqŠ_L „ˆîj_hïzlUaš§ïÂÏR7„/ŸuíÀ¶*þ‰d¨ôÝbÎî#íõcƒO¾Á_h?Ð}>ó/·¥¹OÈÿ¼Ío(@üôÓă÷÷ák&Žá*“‰‰gŠÃØ)x?CúàÒX ØÙ2Èü{„€›neþ‡Ü5O½}^æÍqÉ_KÄÔ_«M&™š U‡…‰ÇÖ¶!I¯&aûŠõðÿ"iô«¬×íxk™ÛÏãÜÚdã"}=¬Û'ö!ñšˆ\%Uö–ˆÓ¯ë·W¹«8CÚØ­æ¾gSúÂó›©ðn‰|«P-Çû¤„ä÷¡«ùLÿ2 ülœ >õ3šòsþe’‹ÈÎý¯ _º,8º>!âV¸ƒn¯@äý´$_?·µ9™²}+à%g©œ•§WÂpLbŠÆ,&(ãJ¼å¾ ÜТE«£)ì݇ÌÄZ6´îfä®]80b~FÄA ¤{÷óûÿj)Ë endstream endobj 438 0 obj << /Type /Page /Contents 439 0 R /Resources 437 0 R /MediaBox [0 0 612 792] /Parent 406 0 R /Annots [ 433 0 R 434 0 R 435 0 R 436 0 R ] >> endobj 433 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 346.899 102.639 358.854] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter2) >> >> endobj 434 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 299.078 128.791 311.034] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) >> >> endobj 435 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [143.845 251.258 149.712 263.213] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 436 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 251.258 223.31 263.213] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish4) >> >> endobj 440 0 obj << /D [438 0 R /XYZ 96.768 705.06 null] >> endobj 441 0 obj << /D [438 0 R /XYZ 97.768 667.198 null] >> endobj 442 0 obj << /D [438 0 R /XYZ 97.768 641.295 null] >> endobj 443 0 obj << /D [438 0 R /XYZ 97.768 619.433 null] >> endobj 430 0 obj << /D [438 0 R /XYZ 101.643 350.386 null] >> endobj 431 0 obj << /D [438 0 R /XYZ 127.794 302.565 null] >> endobj 432 0 obj << /D [438 0 R /XYZ 148.716 254.745 null] >> endobj 437 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F7 257 0 R /F10 255 0 R /F35 119 0 R /F33 118 0 R /F14 120 0 R /F30 111 0 R /F13 311 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 448 0 obj << /Length 1589 /Filter /FlateDecode >> stream xÚÅWKÛ6¾çWø¶òf%K¤žH¶E»H€´h¤Î)- ®L[ÜèaP”½Û_ß!‡’%[›&½ô"’ÃápgF?®_¬ÞÒh‘yYLâÅz»âÈKbºˆ£Ä#°³Þ,>9yÑ6å“?üÈS7õÉ›–Á«åŸëŸþÂ%ÔËdß6réF$r ÿ­ÿ Wâµ>Ö/^¾ÔÇõéÕÛ k@|d)ÈDaÈ2QÒÍ2&É /‹¬’M¹¹çŠ}â ·8XÚ çøÈøqñré`ƒÓ]HˆQ‚Øž,µ'¿CJÅ-ÁØy)ûöœoVE^¶Üzìò*$¼ÆÁýÚݳ+Ïüï‚ÿ3LÕØÝp‹<¿5wê,vgö}øë®àùç-—ïjÅ¥ìöJ[ˆØBﯮ{cpÜ7¥Pvîº8nÄ'ª°;ˆÄY!NòZÉÒ½ûW׫µ>~ÞŽP=JW×’«NÖ(yx¬n@»%S¢©§×§ãÛ}/è»$ðÂ8EÑ¿³%Iƒþ€e$‹C8ê™=!ÕN2 ¬Y‰‹{¡Ú›¥K}êäMµï”e¸EmÅ'—­%6[» g¬Þअ 8OUÅ•÷2”½Bò’³^HÅ«F>Ž˜¡—…áè:o&,™Å6…C!ÉüdéÉÏl_²Úý… óª5N4Ÿt©HþÓ«ðæbG v~bb—„Ö èŸ7LÍÅ-öR:üm`™¨ÀZûÍq3›¤Ú?˜ 1<º‰šÖƒEÓ•Öôû±=XsSs77½oì½¶HÚ+ú÷nPÅž¯AJ2/MíO]HõÞ‹7ëÿ–õ† endstream endobj 447 0 obj << /Type /Page /Contents 448 0 R /Resources 446 0 R /MediaBox [0 0 612 792] /Parent 450 0 R /Annots [ 445 0 R ] >> endobj 445 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 469.252 102.639 481.207] /A << /S /GoTo /D (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 449 0 obj << /D [447 0 R /XYZ 96.768 705.06 null] >> endobj 444 0 obj << /D [447 0 R /XYZ 101.643 472.739 null] >> endobj 22 0 obj << /D [447 0 R /XYZ 97.768 301.002 null] >> endobj 26 0 obj << /D [447 0 R /XYZ 97.768 274.587 null] >> endobj 446 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R /F29 110 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 458 0 obj << /Length 1923 /Filter /FlateDecode >> stream xÚ¥XYã¸~Ÿ_Ñé]À2°æè>ôÃì¢Èñl¼O³ [¢[LtA¤»§ÿ}ªX%Y²•M&ûb‘Å"«X¿òÇwï?Fþ]!Š4L»"Yšß¥I&Â(¹;VwŸ<£Ä¯~Ûýçã_ÞÌÜa.²$¹ó‰ïøó>̽_ßÜAà}øÛ?Ø¢(òLÝŸ› DZg¬ì*9î jûÑãÓþG(š”};œ­ª ^éy`чh!+ª»J¿èê,›[U“Xdy>©*I G«Ësê¸y¿?ÞªYÝw¤Ý©ç5«[u¨Ô@v•êð¶vRufÞË>I=9ji•ÙÒ™uXél÷Qè½Á©ðQÎD0)Mö“n jð6\l¶T·f+šóø¢A³Ä³²@Ù»-[vE+ _öi ·ú/ŠÆ¡ÈŠü÷©JZ‚ýºrS7–AÚ¥,IuºÑ'JEš¤“>ú’ ß[E&FÕ²ëH1¤IçÙ7œÎí†è¶––Fµ¼ÞÓõÄ®ˆH±ÁûNcßòa5³÷g öûCùÞ±ÖÌI߀Lƒ„'YâIÿzÅ_9V.¬ÀÈâÂ}‹$¹˜\Zù¤mY´é«¶5Ù»o*5N†F%Gî1¼xºòEÙ;_Uj;ûbQøÙUöU·^b‘„Åä…Q ,U»Hˆ¬®"‰ê«_ÎÃáƒJ3‹ wLr|>ÏÇ‘iHâ*F>4Íêvp]·G°1äx©@íÁ=/^y­²5[ËpÁÕIMê÷WìµjZä³2lß ^˜Í ƒkƒŒô¯Ùé2¾EžÍ•sü¤­(û¯C½q^Zˆ,Œçdàã’«ãf†2eŠ0×Q¶–tHÂÄûÓÁ±¾ˆSff»ŸÎ]‰%àW?ñÁ-ÚhO§^+VžÁg>þå‘)eß„îìƤ‹ë>R£Ô$‘¯u´oƒâM/P·Ü yý넸úP¢‚÷Mÿ|ÏÛp|XÎ!nu7Ϻ¾S÷°'Ø>×U¤åmtµÉØI!}`4Hc˜]‡oø1ˆî¹Êž6yÒLDÔð0‚ ˆó'µì-÷m¥-kQ/?É»V÷}ú|þ¼Þ,èd«œÙ8"œ=àù’÷]§K d¥JGc—Ò8p¹€ƒ³Qã–Üþ韪´k¡Ah)̾Ûñ|G«Oª”pœ·ÖÒ°À~qk:Òû?dÚî·“/ÞL¾ƒQö4ÉWÀ Þ‰ÁûãÒ'¤L(œ{:îÓ$ºž×ß aJeÌÔÏÍíbÛW.|qLN¸Ž‚q:oÆ+ÜÏj&=—-ÍÒü¶àºå¯ÜQÐ¥W Ô æCÙÝ!‚Ç> endobj 451 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [227.904 356.862 233.771 368.817] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 452 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [233.135 344.907 239.002 356.862] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result2) >> >> endobj 453 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [233.135 332.951 239.002 344.907] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 454 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 169.564 102.639 181.519] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 455 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 169.564 207.619 181.519] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 459 0 obj << /D [457 0 R /XYZ 96.768 705.06 null] >> endobj 460 0 obj << /D [457 0 R /XYZ 123.92 360.349 null] >> endobj 461 0 obj << /D [457 0 R /XYZ 123.92 348.394 null] >> endobj 462 0 obj << /D [457 0 R /XYZ 123.92 336.438 null] >> endobj 463 0 obj << /D [457 0 R /XYZ 101.643 173.051 null] >> endobj 456 0 obj << /Font << /F30 111 0 R /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 473 0 obj << /Length 1648 /Filter /FlateDecode >> stream xÚ­XÝoÛ6ï_á¸A#+’’,ËË€èPìe)6 Í#S–VY2$9N€bûîHê›NŠ­~èxw¼ûÝýËí›÷x°ØMÈÂÅm² Œ“ [„Áš0øp»]Ü9oWnÀçs-õ¤IÍDUy¨2ј²0“¤*÷â}¹•ùêþö·…·p)%›À°oäþ@5ÍÏ®ã/^à-哈›åµÞZ>T²ÎËS·– £X!µ1­Õcó|c¾{ÑÄ)ò.þ–qs¹—MZn G¥GËïýêÌâ2ÌBÇ2všpd¿‰2£ÍٱЀ±"J,|Üìox…¯Øé«È’\fuœº‡JMËÎLB䲘íZY \p¬/ð:ä¢p÷2“ÕYÓƒ°}VìÜTTŒMÇrrûå®’²8•å¶;ÚÔ€–¬îR>î¼èÚ”zÅ—jBDµÃÒMýióï÷áEÙ—^³k~­ÿ3øS<îÚ÷÷z¯ãd (ãá,ÑbÕ­²ºÛâôQT( Åž'¥ñÂõ‹€Úø„úáX3Ãy|éÎ?àb?2 é‚ÙÂ`6;öÇ@<«iì˜3¤˜± ìØLŠÅã†Íͯ¨Õ¾Ùkå³ÿéLît.ù¢–$Éš^l\ ™_Y@‚_Ú¬¥üÒÑ‘Á:šÎÑ@Ý YSp:þ4Òü?®\ß BÊ­ž5¥+ÙT™|\±È‘zGè!.+¶vžq…)1ÇRCUVÙ.+D®W[ѲrÃ;®"¿c•Ÿ‡@îÏõL…´Å4(Ï=F<À¬õ_óëQN¼¨C=¦ÊžŒ¼bkápÂ`fèŸç !F|Ö~ÔùÌwV.(]6©N«<ŠH@'ÑFáëÀ9¥YŒWLq‚Fê¾ú›æ#õ") ¡šf…•Aq‚&ԳΡƒl¬=çc2!î­«XäyË5Î[idÈ TZ©º©@\ŽGpü‹Gy-µD_Ú yR˜Èv)Ž x¢mø:Bz—‡89sËGÂÃÎSO6¤òuçêÒÆ‚2„ìçE*2ÍwQI­’Éþ¨¯gš)Üî 7 oÈìÜ‹ZqùtHçr؆D¼¤| Ð.1 °/zAZ‘R[7µN*±—Z­S–›M gÌ]‚pÒ¸N½–(…À…,^=ÆÜN|bìñö{ªPØ6×l¨s¬mF ¤¨w†jð®îç¶=ö•RlµÒV…üx<³¾´À.¿ƒ†›Ë@Òöb¹h…jÚ™v:¿špóñh­z æü 8%lÝAí)—•çJr˜rÞ£eЩ¢ßc¤h*6¸FÔ^ô¥À€lÓö³2j-£;lb}&$ø®é¸%L¶2Ç\Õ+bkÖiÂb ØdÆ~&&À[hÝû ú¨‚D éåx°µZ%ÔëÁcspõ[‚íQêqÀܤ±MEMtbË‹¬&Å1ÏÑ€–jðbèìkÓY\Õg"¯¯Ð—šÓ¹~fXïì Ûç¾öû¦ÇÑŠ¦û% ²u²s˜å/O˜›zÆc¢,›çÝ»3z .þßï5l™$àÿ«–å™®›yÿÅþµ¥¶^ûdÒᢨÌÚ+tFæ}Zô ÚTúïŸ?}‚"‚·Z“cnžðX&•Áͺwþž—µ^C"k«¦¥{†ž®ÿÜzÕXw5…;ºßªõ"®$ˆØª*ÏÛâûYcFE9ªÊ ½>fºŸðu¡€ï»J> endobj 467 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 313.026 102.639 324.981] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 468 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 313.026 207.619 324.981] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 474 0 obj << /D [472 0 R /XYZ 96.768 705.06 null] >> endobj 466 0 obj << /D [472 0 R /XYZ 101.643 316.513 null] >> endobj 471 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 478 0 obj << /Length 919 /Filter /FlateDecode >> stream xÚµW[oÚ0~ﯠ´­nœ‰´¾lj¥MÕ&úÄxp‰Þ‚ƒ‡€Ô?ÛIœÊP÷1ññ¹|ç;ç˜/“‹»GèöBú¶ß›Ä½p Æ~Ðó½1°¯7‰zÓ›åílòýîÑñ[rÆ^Ïy‘mbÂÀ<Ý®—£³b-Žhª¡´|uˆH­^G««î+ÞZðm—/ =é‰oGžíÝü²<‹ä€IR-W1ÿ†ÕGlïÄãóH<Ó—ßxΦÓÁn0›íµ[ºZœäØà¹íèAåWlp|äq¸t]Û«4 Èp¾NiŽgMnììSýößÂqÙÂcg7X|ú$žˆîª-”ƒbÙvå.樂& p­]¶-ÚP÷;g麒îM‹$jhÊÄ"Ãs9ËŠ¹|Á–XË҆ǘf}e¡“ƒ‘Éà” %h; „GéRãºå¸êÌÙš²'ð4oÎ…H„º?œÓ=µLª5'ÔɽïÒyªËÓ,-딘,µƒ¥zN…­k:…3ñâõUWºm󀞖ìGD,³ÍÒ÷’fdA(JÐÄŠw:õÿ’ÃnÉ—L‡+Ã{/!i¢o`ó?¾ö»Nú”˜,–,?^ûÂ,Ù‰aZÝ8;ñXçUÛ­±n1–íy( *‘Á!tÛa5º/Y…Ì@oT±yZPFèbpbB´n?KD£Ç’ |k¬2L €¸ j ôûÃÆ?¹ìôò ê9îéj»AWC•lÚ…Òª:ÀêÅÉsu˜W½£‹=›¢†>WâñMr×õ»$¹X%ˆál¨FFTi¿ø¡L½à$-¥5ôç{!ô€åu¥Õþ§dð()ÑNÚ‘¬ÇJ…A—6òDäMx„FdC¢B¡§Ð˜ü|~8XûâdKmOEÊFsËZd¬Ek(jß`<ëæ**V+5éPRàãC¢«óÇóÓ“6 Ý¢´r¹…6ÅeE¤ö ïVC;´ö;è3>Ttu§W!pöÜ¥™91éš>(äšì)ÆQIIØRñµ4TŹ†Î¸=ÒªJÏmJ“#r/—´i0á¤QˆÆ Μ¡Œ*ÕÆ.qþMã˜×ºNAë¿Ý˜ešŒŠÉ)vΜõÑS¢û˜Y9ÁÙ*7]ò®Ù~çç/—üêÍïN²P%Ä s£y¶¦}ЮfþwÎwù˜°C°åú•àÅÃäâ/Äÿ’} endstream endobj 477 0 obj << /Type /Page /Contents 478 0 R /Resources 476 0 R /MediaBox [0 0 612 792] /Parent 450 0 R /Annots [ 469 0 R 470 0 R ] >> endobj 469 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 653.748 102.639 665.704] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 470 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 653.748 207.619 665.704] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) >> >> endobj 479 0 obj << /D [477 0 R /XYZ 96.768 705.06 null] >> endobj 475 0 obj << /D [477 0 R /XYZ 101.643 657.235 null] >> endobj 476 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 487 0 obj << /Length 2643 /Filter /FlateDecode >> stream xÚÕYKܸ¾ûW ¼¬N¦e‘z´´Ép€õÂ9$@0Ax}ÐHìníèÑ¥y‹ýí©Õ-ÆI޹4Kd±X,V}Udÿñö͇Oa|•ùY¢“«Ûý• ´ŸeÙUï| #·åÕÏÔÖl¾ÞþéÃ']ðêÄW±º ˜iÏ3iÛ„\m•ò³XdÙ!oµÙlc{¿ßr;¶ÜëÛ“)ª¼¶?qpkúÆ^3Ã{;ôù¿‡~Å=ßÉLkJ¦š±ª““¸vè«»òÆv]Ýž[Ð\Œ?³€v%þf›¨È»u\{Юëí¥Øq±Çª®™:æ¨òƒ3$²rÓOÆCY]¿ÐAL¶²·É\wµËD`žy -ö,G‰}¡×tt 0:ó–©®•Ar\<„0 ¡rL'†óœ9"X˜„¥6‡£]¡atâ}Þ³Ìjàñcn¹CÄ(nªvÖ[tõØ´kûuQROç|â#mK˽ä†!¨Ú5 Šºé4±íKÓ3‡v=Üz®Ö_óày™Jç"×™ Äà@»Š6›­ò:>ÕKÇ…)hÂÍ9ù®¿å¸OøÒ²Ï£€_rÁúé.vèhÇþa_ ~Ñ=Ž[k†ñ´ ©ÊWAâ&U+ŠR#Çpð¢€YÍ1ïfáý3ž¯,L8Êíï’hq¼¦9ͱch™þÞº¼øW÷zTLÑø% ' ü‚Z\}=]úU;åBvûŒ¢ü‘(ñ7«k’3¾=§£?¨ÿ9Ìa>ów*?’C£øK4¹/4àœ1Ç+!v†ä –ÉÒ˜†x{SŽm™3ìrÑË“TI O•yäKå4hgŽ©ïÆ¡"À±Œ@-€z0Ï*·~5X& „òg~ T‚ãEefzvÜþù£ìÞ:+pÖT¡©…ë5yOëö]Q¤¼¿oҴДóvÞÙZ€ÚêÐ2–jD °$‡ŽXÅ”UœÔûD‡™7 Ø08i™DŠjFhþeúŽŽn„sÆ£²èJÆUBNG¨úǶÌ»êÁŒ˜?0¶Xähc¤ëR^š–s€°5lò’Σ•*³Äï-VÑÒÛ]’º§c, umý,+´Òƒzx¹qny@ôÕáèlŸªJ¦lUØk ¹âó4)_­Fe{±âCº ¸g·k¯›Æ!:r84ÏhÝtî8©>E±•¡rÒÀå’ôkçåB—_àÄôìFq ø^ðWûXúkûåR$ ¼:'MRÅvç3WÓ™ræ\€@Ó8kþF]xŠ0ذ2wßçU=8›¥\#ᱤ FOä ¹;KޝÍ]+kÛµ"<ÀÁóaìµ‰ÏøîÑ2ƒ±ÖÈ@ʆ­ :t5 NbÁt©c*ÆJáñ4k鸻†©Á ˜[3Ui" ÊC&|Në7*P6R1 ŽTaï¹ËB„¡d Gq¼²íûA€;Txì32¾öÏÌPš½im%…c˜°J$ED8©w9•Ú÷ä®y_JwÑ5'ÈŠHE>®m¢íÚíä=P2Í­ …R) `˜|ÉÀêkîÈa4 (¬Ò ÂpŠ(ËlÝ(²êêÞ%èfD—Ž!“È)êiK¸ùY´È.AD¸pÍI‹@èÐâ"»)—! 6&o-÷±× erw‘‰¨¼¥˜hD7Ÿ«u™XÃj¼ÔeÖ¼L½îÂhºW79ÜJžà¤vQÈ¾Ž²¥†ÇXëZÖ’î¼n%Úº€‹3i’øa–ÍírÎá]Ur²"”º\RtÒðýx4-Sà¤ÅXO¹Q{æé„¡¼²mµ‹ü(ˆÏÛÞîR{­ì?öU6Áé¿Cé‰lŒT uÇÚ雟ÄÕZIó{æÃ{_”4V+%’®/Ã[Ä"|rIf~¨¥¥¶:Û‘;0…'­!©ª ¸‡<š™\“m6Çç³ùèœ ¸)SáB=¼VâMBìµãÝ€­)é^·Fð6øq¸Lä¿8’Ü WA™®÷Q”y¡yY”À]ˆ½ØÖïeG6o„ºtbu‚Ϫò·DÐ¥”®–1²­y\ñ¡0Žý@;râ`*ZK¬ýÁÏj ŸkÅo¬À©L 3•‡9¤ÄìE>Z#<0oFeÎ3¦OÄã(˜\pìçÑ 7£"öa=3¬ÁC{Lþ»8Éü09_[+Im(åo¡ë\ 3aÌÂÒmÞ_ZãÐB6\ÄdpzÄ^­–ÕŽžŠ'º[ñÕúóËô4`Bã©~š_R41Ð<«ÚðÅ ÅVÖus¸«é½ §ô²8毣©ú¼gœf¿¦e§4w»µk(?2€šOyƒ¯)/ïPqä‡ñÑ¿Êmò Œ¿¾¿·æéåÁíR?Ч#FˆS)cÂ[‚àwbhz4ƒ¯œ›©Â¯ù›®e¬j8‹ ˜ºÕ0M¿ô¤ÔÕÄr~z Ós9ŒjP´#5Zy«qÕS8¯–ÖtªýX¥ H=½¸“ÌÖe™ÑB_ëÉh˜ ÔùŒè. 7ð*óŸ¹ kòóm$àê[¾Zþȹ“ÇùôºƒX*Ç‚ŸJ“ó»ÐÍôð™LÀF/Då±t$@äÛ èÙ­EQ%û˜Yá.—‹á¬Å f+¤½ŽÅ÷#©¶ðÞZÅéŒ^ý»äº çÚ ß„Tу\1¼åa9sw;ìD$Õä¸ÊÅ 8öN¼8að–ø4Q B+¾tUì õZÆ)`œ0À¾ò®F¯Þ»ÿ‹wµ‹¿KžÎC¿üÂm Eá4ts¼ö?JœÀ†¿õö¦_«)ïÉGò'FÓ‘³ É—?¤~MÜ{Å(xC˜äÛ9ý¯¹_6%¾^V=²ÞÕƒ©†™x—›žD`·\–nÙ¤Ñ~1Tt¥P‡ÎX7ØSð®`3(£À¤dýÛé¿-N9±Ä¯ÈP@4ägû¥¼§ÅßITl£ðÀ—w¿¶ïoÚ׿ÙÏE~°úâŸ÷M>¼¾Zp­®/gè¼Küþ©3?Måÿ h‡Œo~¸}óoqa€µ endstream endobj 486 0 obj << /Type /Page /Contents 487 0 R /Resources 485 0 R /MediaBox [0 0 612 792] /Parent 450 0 R /Annots [ 481 0 R 482 0 R 483 0 R 484 0 R ] >> endobj 481 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 492.354 102.639 504.309] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 482 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 492.354 207.619 504.309] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup6) >> >> endobj 483 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 187.497 102.639 199.452] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) >> >> endobj 484 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 187.497 207.619 199.452] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup7) >> >> endobj 488 0 obj << /D [486 0 R /XYZ 96.768 705.06 null] >> endobj 480 0 obj << /D [486 0 R /XYZ 101.643 495.841 null] >> endobj 489 0 obj << /D [486 0 R /XYZ 101.643 190.984 null] >> endobj 485 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R /F11 183 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 493 0 obj << /Length 2412 /Filter /FlateDecode >> stream xÚ­YY“ã¶~ß_1‰ãŠ”ŒhÞ"+™‡MÕNyS~J&åTçCB²s1ðÓieƒ£Õµ©†ÞþBüjÿ4) ò2šß¹­¸Â_´e¾‡ƒ^„f( 'N¹é[äžçe+.Yoâ¢X©®^›A²÷¼] -‚°Ìü:ܵ5o,¯ê[£íjgû–iÃAH½5{Ó©†Éµx4.sÒ‚Õ±Â*{Š•U¨ñëy‹H÷ŒÂÝɾ˜—u†Ð‡¹Š¹¿+à"|“”N’°æxÀ¬Q]¥ƒõ&MÓÕë"á Èv=:­kB* ‰Q‰#WäÁÞuäO| X¢é¥I[ïͯ7Ûòy”^»•#'Ž#oÌT”Ýc>$eNÏÜ|…‘:P `»AUŠ<´8ó§DÛkMp4ï„*¤Q¾’˜3=,"¼šáÀ£Á´zSc’ÃèjÝáU ¼Xõ8Ý?hÙN&áàä´õ›†G@Çvåo)Nó|¦«šSׄ3Õ- =) ISo‰©âÑóÌbÚ  b;Z… ÕàN¿’2–InÀÙ;%•ˆI=-Å+‡AWl! Þ7å=àc=*²²‚W´«Í‹©Oª¹{øÇ¿>-Dh¤çêx!€ç2‹WŸwçshШA²[š!µ[®~EP½˜K1yu|Ûr,€†î²Dl¸å÷4WCÏÔªg/*ÓñBß5DøÇ™nßh^$ËOªáè[A¦\ƒüLNg†ǧf0G/ÀÐ^xc ÆÞq¶gæ1ö rA€µ„ª´duù¢Jç8þ¢¢4(ÓsF†ÐDÄ?7rX«ìÇC§ÊÐøäó¥“ò%(/šŒ¨.J=ÊÉžWÀÅ)‹4ï}Š,&ñ$ÔñC7n05 lpCÒ©Dh µ~Þy†…ˆcQxþÞ4Ôânì3¸èN_HA°ô‰³&Á¸îTURÌT"Cš¢ hü"’9¡XàÕ/(¬\÷Àà¶ŽfÀ+ÂÝ:‘Íu*ñbjh•Tlž¿'Ë\Ð,É>_ó|-/y? ÉúSSók•W—3 s…TH) T‰Ï‚¤‚îoV½&Yá‹ÅXi4´Àõ¼V‰(h7 ²²Ûª^¾íŠËóIb).®[ Ò= M¤B&Wö§“Ç®óÁ<¾ÌxÄÌK ’‘sꑟ>Ô!g³8™Dûø®ŽSñ8Ü}5øAEmÊž35¹1òŸF®“©±((ÌÙ§aæ} ”ñ@·fð¹!Ì/-æá|új_lóá!Žxñ`$½:þ•>4ÄL€·E³2+ÌÊôÜêãdþv Q=³KÉaâN{N«*~Á¹xÌý Ü€‡‡¡ÇZ Ý—Phzž«gÓ˜AŠ. ²@í„¡“Ýôz0Z„ð‡ í-KØk<ñ/bêÀd>«M©@Ùš'•ò'ˆH8Vô° €ð=¹ˆ%&œ}•¡'80“‘fjÖ†H ¼7V‡ê¶ŸÏN×il~¨«Ž1qvÀ5ù:æó’zë?Ž•ÒRm1Yð ü’Å"Ÿ&Ñê£4 ß4Â,Øæãç9[‡W¸ÂgRY2âsTzlSQ%ïù PæN5qiM?/½ z„çzt~Ã^+”eA§ÓôlT;¿C»Ù 2™=1ÓË€Àî™úzÙµïå6ñË`×™dA‘‹)ik>=|ø/ƒü$z endstream endobj 492 0 obj << /Type /Page /Contents 493 0 R /Resources 491 0 R /MediaBox [0 0 612 792] /Parent 450 0 R >> endobj 494 0 obj << /D [492 0 R /XYZ 96.768 705.06 null] >> endobj 491 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 501 0 obj << /Length 1696 /Filter /FlateDecode >> stream xÚ­]oÛ6ð½¿"Í DFkÕÔ·€mÀ> ð—zOih‰²¸J”'RIóï{Ç£<É‘“­Øƒaòîxß<Þéçí«÷×ÙEîçI\l« Æ?eÁE§~ÆÛòâÖÛÖbµ£Ä+¹á¿4\k¡z\•„9ô¢¼_ʼnÇûeL/wƒúrOVGR-Ô*H=ã05_™w 'H4¢EšÌ3ŽT„1£.š·‚P]_Šž€\ÿCu·ýíbÍ"?"øg~;s>mXÔk³ZIì•iÙ)ÜF^WØŠÁÅ'F/L×ëK‚a+æmyrȈ¾ÕþjÌû8ìtÑ˃‘j¿DFry¿Zç‚”ügžÐD«:c-ÚÌm±ÎìïÑSè8?‰ºžþ]ˆb :˜Ö|°Ž'=hƒÖjÇŽüMÇiA1GˆU°COâVÏ Eˆó›;6æÃ’Œ 7`}"ŠÚ¤ÞCÍ !NpètÄ€‚­[9&Ÿ­=â€c—½›dæ‚]ëŒPŽ9Á¥“ÁoZu;pOâý% ãXá?¸*· ý$8)sp]‰ƒØû~MÿÕ  å³Ã`F ´Ù± N”øQ”ÒªuÖ,ˆý8Èæ²gEh*«ÀLôå„ôrTc¡Ê•÷6 Ÿç5ÒeÔ5¥‚wcˆ|,¹=ázåÊØ2Åßí-»»[Ê~cæ s+ï–Ôª_æ´-7Er¿A‡;g<>ÀH‡é ãÊêBÕ‘?nÎådû&ɞͅ<òY”ž—úZj_ Mƒë1V£%Vã`ãçA0gðR˜gnQ·ac~£o­ŠÇzIÐ9M'z^Ù䜲Oò{¦ï{kµ¼[pý¢¶û¥XBvò¶š³,ÙË,Oj/æÔ‚A®öðõA’76q·XxÆœ-šAC%ºz’—Ésy‰’Þ,'ȉÚÛ§Un¬}§º¬-×Å&¹<ƒÐÏÙS•Z©5´4¸Tâ6Õhñ–¤/ß’dnC['ÅMûøò Ôn‡ýÆ›VpEZÃÓÆÞwE/8¶ –Ü¥Óж´tº.‚c}©OSòˆur‹NT•,¤PFŸ á’ú(ݯzà6µ NÕ ªôµéQÝ—ëŸ~ÿøáÌ^Ñh±À.H|³gcÃ4ô徭ó.›SW›î€‡.ÿpѲ, ®°Ý·ë ð EI«‡ÚÉiÐÛ® -kî.‚o\*Ø|¿<“ðlr‰N‚Hë…—ä5”-y/Ë7.9ñE]bZB{ïÃÄrè”Ç»yæ*È<×aEË–Kÿ`ñŽ&~šg³& û˜Ÿ¤ËE6›åfK0ëþ®WYhç¬ gÞŽÚú϶qå}© Œ³(´“4ˆyÄ3ï—{7D±#m éo:aª»Vàla7Ø£ãÿ^Ž3Ÿ=¨è&a©xÿH;¢À~öRAÆfß’âÌš(·”|¸ªzP€·¶e^ºê6JÁ ÄÁ õ}7‚ÅpIƒ0µÝ5‚{ñ÷ áÆ€c¯ÎïÆñ15¥üÚq¹!<)”À;`É;¹¡UV#CC6€æ“#BŽŸœ‚™uœÁÊç/åûÖÚ²Ð3CoWÌ*‘Ä»#(' ¹¶Ú‘-…Qâó8¨OèzÆ¢Õ»£¯BìºÖiR;>n"ÚK¥à™²Cof£ô+}A(†Ñ~nìw<ÝU‹cx}„&yjG"[Y3U¶z8à7¨Nn;~SȦß"Qo®Z:SwúFkš¿Ewœ·`ÒÝ C¼¤ÏFFnÒû£(ýgêÆ7f/–ô?Ïf,÷ah>_@áí¸··tÒaŒÍ£‚Øõ²¸|®ãÈBè8‚“×köM/òã VûY–¹A=GºW¶¯¾”A@ endstream endobj 500 0 obj << /Type /Page /Contents 501 0 R /Resources 499 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 495 0 R 496 0 R 497 0 R 498 0 R ] >> endobj 495 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 578.032 102.639 589.988] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup6) >> >> endobj 496 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [243.595 207.422 249.462 219.377] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata22) >> >> endobj 497 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 127.721 102.639 139.676] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata21) >> >> endobj 498 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 127.721 223.31 139.676] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata23) >> >> endobj 502 0 obj << /D [500 0 R /XYZ 96.768 705.06 null] >> endobj 490 0 obj << /D [500 0 R /XYZ 101.643 581.519 null] >> endobj 503 0 obj << /D [500 0 R /XYZ 123.92 210.909 null] >> endobj 504 0 obj << /D [500 0 R /XYZ 101.643 131.208 null] >> endobj 499 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 510 0 obj << /Length 2490 /Filter /FlateDecode >> stream xÚÙnäÆñ}¿B ‹44ïÈ` €ÙÀYl?dýÐ6gâ°<$Íß§®&9#jmçawŠÕÕÕÕuWë¾û§W¥_fQvõP_…Qì—ÑU–æ~ ÕÕ=SßnÒ(õ¾i`ÚÊ<›jT |…Œï{ĵëV¿TjPŒ=ŒýÀУæ_Y˜IêNô51úýá_WÁÕ& ý2];¶÷Û±Á3ƒöôði¸ÝýC6ß} “å…ÒÐÏ àM´ÔËÝÚýø}¼ÝÄyâ½ìÕ€PÊ׌mµ®î,o‹‚ìá^!HÀ”t’áºbØ]ç¦dm„Ë¿¨óÙ)Èy¯yEµÂ4”BX5ã Ùm°|š‰VîãäN┽#F¨äv÷Œ8ªŒ26ª#ç…X=s86p1Ù½ä´%O€ÍS°Ó>Ü35 H—'…÷3ˆ×½GÌÆÂ]ÂF=Â9~6p³Šqµík7"[¸_è‘9¶ÙaìáWêœiÈ~ˆcGÒ–îXˆ¡…ô¦Ý1ÖYê3 ÊWówÌÆÀM¦‚»•A€Aú›Í™ø’J“ö 4Ž"œn4EYî‰ü¡œ/B§žE•:ˆPÛ‘åƒ,ìÜ`ô6‘ö3…Î ã! ÍÌ>O3|ÏöRUu¡#°Æf͔܀ŠÇ’ì“R¼6)cï#› ­=ÉÁm+¤çÌÌ||4 ²8˜ÝÞ%XxäìÅŸùé'Ìàwü9&ÙO“ÐÜéi…Ž=„l½v%à¢pœ9…¢¦5Ûû9§sÄ Q €»½-]a’øyèêo?vϵ¡²?làÔÕ³ûà ~·µmí›v¸÷˹j-‹Q~dŽó$e§û±Á Väèn+õ>Îb?³ £Z—‚¨~’Ç¡’é7Ž€q)†>QýK§ŽŽPÌ#I ” —Ä4CwZ8Ù§¥Œ[9Š{´§V¨C\1"T —ÖX“$è$›¬ÒÞ+*n¶{ò×wõE ºÉßz B“þÆ!˜æ,UZL.™J’H%I¤Åì‘i~fk¡n…šÓ.Æ{–x£ð—0¨vûEôõÂÏìQU 4#{ Ð¬ ðNsCöÙKtôÜ(Æ®lͶLÈtŒ:(F‚ È·ƒíN?¬ 7¬FÖwö§õý =>ÐŽ ÖÚ Cß8ð¢=‡¯©òr7¸6NÄϦ³íbŠ^èÄH€u}å°tQýqÍPƒŠH"Nq|ŒZJª™ueÑW”QH}CJÏYP<­œ^d ¹ãð+÷ ]߃»ÒÝ<’ÀJo;ÙN-,4c=„ß‹Û+è û ¨hŸ­bÓ MYZàa¤wZ–vÊ m˜¦8þ’é–´óT…Ÿ³Ùˆ&Æî®ý?rœ¿¨·EÌSS(fR7— @º‰ÇÔ'FÈô®{ .Pˆ,)iPà.”J(¹}K!ì> †rüâÇ£m¿`Ø·Z=ÖôL#…Xfo€ß3$Y0ð¸ÉnùÃ)iLÛ›J3,Q~v…;ÆÜð«ÁÖö¦Õ7ŒëyD@ •¤éÐ5%Èêì4ÔVº‚¸FoœÄn| ª\/*,wzP‹ÜÃÈ;I‡^¬ òsŸù«tTäRÁdFAõ@ÔüËÕ.âΕ1–§9˜zÂ(®ú—dÙ•BbkÊ ŒPÑP‹À¤ü8X·>ìô\9åƒ%P«jT.°‚J¨*)ÉëžNeÜUþ§–;‰Û†?ì¡0½’ \{ ¤×øÔu}VÅ¥êGP³çx !¤Ô`œ·DSõ¹f…]ßÍlV®$%C,O¼_Ñ-iî¦ï8—×+DáK£8Œ3u¹·kì# S¿ü™Ko7ò9{/lBï­T$Pê}byVj™†ÝnpÞB‚Œ«ÅÓBáJqÓC¿úì˜Ù ºÃNgüå6OÖ(øµåŒ¡Tô"žÊRÂkÐ>[%öåe‘BÛ” ½™.C2R6Z y÷ ‰Á»ôƒó–}™=’©³„À;çµÂ‚t½Å*Œ1Ó–—©õÊHչ躰ƒ™yá|êm Î3É^™l9HúEž^Œ×þÖ¾÷a­5¡aæçe>Í?+c22NÁ½´í M¦ï¿ÓL©þgfoò,ùêS|úE¿Ï¯zßÙqß~Ë¿c$>¸¡ê!¢ öË"þúŠØ²è28 Pœî^ÞþóË«ó(Œ3º}>'£ø€ÄËÈ[äÝ÷8ûœdeùVÿæ54¼kéOsÊ¢‡ñc\ò}ß½­(×Ùñ(Ÿý:2˜Ütãú8”~’^¼‹¸¿"Gy{Â] §éîå‚Óþ̓…¼”„Â_e¶–›œôÏJ¾0ˆøoPÐ{7úXèßûK8W}ÝðQàGeqQÆ4>á-m(“¤~¦¿‚`Àžý:Ü£l MÍCíœ/y·òä'~§G%ü|z ݇>ü£l@½ endstream endobj 509 0 obj << /Type /Page /Contents 510 0 R /Resources 508 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 507 0 R ] >> endobj 507 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 223.362 102.639 235.318] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata22) >> >> endobj 511 0 obj << /D [509 0 R /XYZ 96.768 705.06 null] >> endobj 506 0 obj << /D [509 0 R /XYZ 101.643 226.849 null] >> endobj 508 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F33 118 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 516 0 obj << /Length 1039 /Filter /FlateDecode >> stream xÚ½WKoã6¾çW¸N°¶W”¬PÚ"9A­÷äúÀÈT¬%"e;Àþøò%Y´èdA{Ä3g¾ypüûêæËƒL„^8Yå/rM ?Xm'ë9-®Xºúëëý?n›ÕŸºKº_@Z²ÈNàs.æþRP@ ÊÄšár/Uåé÷ïÂsAâAU¯B¾­:M‚ö{òÚa|VA” <ñ—¦jsš·UÆŠºšv'ò×¼ ¯Ûj (kCJïWG}~{üû^¨|y€Ë#΂xé›0/JРŽßäù pÌÅðÖ3eî™&”õ7¨ÄBk…›’zÚç-WM+|ßÏÒ3È ô’Ø4¬BIÒ~¥NÒúùÎØXcBÑ´aq l!Ä  <î®"Ô„ðä1±”/¦»´}fb{䫃kF»3ºÇY »ïlZâ|™‰UKˆX è.¼„AÈcù¦—Ð@'úâPI–¯ÛtÔ¼t]"–í„1V7§‚@Ð3•©¬¢$¼.£Ë`—˜Ú½Ò`U-/K]±·Q±¶! éâÅþ¤}ÊUdî o-.>}}|´­È¯Ï(õÞ½ìRë¥Ø²OO~hZÖÓ™!›ó†ÚÐyÍ‹÷>/V?V˜ð"ÍúB¼Ý… 𽎽C´ß‡ÜÒùí±nɶ;<è Ê¿ônÍv¸9VúÜë‚ÿA àOvP–!BLþà;ëL©«÷¾µÈÝY±u]¥Π§@ßQpÑÀú¦"Яµ’A;1A†ÆÞmG!Ó©c+^¥’×MÙ’ -ýx\Uz¯ã^Õ\¯áfcjòœªtæÏ•0»öÐŒÊ|iä#­ŸJjñÿ­ú|(³ßJâx —€Ь‹@â{Jê©^xñüÈsÀ÷—üy€Ë ‘œ“×EìÏå¶?Ç'npÆ” ·Ã’ÐÐAG]Ο¼ñíÐç}ûâÉÖ'…Ÿ7uy—[ßZÊœ¬Áˆ ”°yý½c¾Âßi#äâ(è  msÈ ²ú´ß9 ¦-a¶9Á ÏQ´VañÝ€M»Vƒ˜‡âê+|™Ÿ>é˜_Ž’“ˆ¿;oöÈÇ=ØKÜîÛ®µp‚KPy®Må“PB3Å4n&n[1T½ º1Í2ÀI«¯–’:¾;O]^ikqíÝ5t£1é®n˜–Ìß'?óœ­å¦5¦jiÿ:¹g¶dÜÔ¿aÄЉàza+JÊjñCÉ>áãùÅÕ®bÖ·ˆ!7C£»VÍÉ¢–EÕXÇ·)^ ^ÌÓóûk³È6¸«½µ#£!Gø¥œ£DâóDZBS7ÝܯnþÁjÔJ endstream endobj 515 0 obj << /Type /Page /Contents 516 0 R /Resources 514 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 512 0 R 513 0 R ] >> endobj 512 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 199.452 102.639 211.407] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result1) >> >> endobj 513 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [206.983 199.452 212.85 211.407] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result3) >> >> endobj 517 0 obj << /D [515 0 R /XYZ 96.768 705.06 null] >> endobj 464 0 obj << /D [515 0 R /XYZ 101.643 202.939 null] >> endobj 514 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 522 0 obj << /Length 1199 /Filter /FlateDecode >> stream xÚÅWKÛ6¾ï¯p$k·2#R[E] v-¶=4Îi›ƒV¢×DeÑ)?þ}Iq(K²ä½ô`‹¤f8~óqôyu÷éû£E! G«õ»EQ4 ƒ9"^0Z¥£—ÉÛôÛê·OjztG3ŒQ€ÍÎL„,b3ùyfžë8‘¼øË Ü‚îôÃuÌ‹¼à=ß®‰z`ûÓµ â¡ lm4´4ËS¶gigV¸ÆÂCá"P;Tªëÿg!AóÐoGÑ´±eB°üMszHUL-爋"‚A=4êBò*¸ñjÉM3‰‹·rKsifÛRÀè„wõëÆæftv¹úóëøvaÖò¡'„wU é@ŽÂ¹:iÿZŽp€|ŸX‰÷oEüj6—6>–šç>.XüšQã—Jm€ƒŽ_i[žÒ Ñ£BK"€‰1KÇM …ÇÊË,«†i}8B8 –.ÿøúü Þ 8ˆ:³û8³ŽÙÓ¤«ãH_Môȹ2º)€RÀÃ¥Ð*¨ƒ#éKä6–;jõ-¶Â¦özqfÖJZðÝòñ×ç/ߦ3¬ä'ïsnÓ!i‘Ðü^â3š¿É«í—K·'õO5náEYìÙÞÎ^Kºµn!$Ͼ¹HâŒ/ ¸„ê–Gótb&äjáëµ ²7¯æUãXZŽ…×r»ž3óK#K½&[‹Ö çç݆RsYïÆnOÐ'rsAÞ+bÚñ\Ðû[Ê2–²búqîåiJÕëwKàµÜϤA‡I¿´@£õl˜=›r*,¨,ËÆ2ÙtHkͤ¤i#ê!J ¯d¥}_†8Nx™KuŒo ô©ÍòmÂâyv²URÖš€z°dožPŠ:7ß¹0º88‘ÿDZtQS…Ü(’¾<Ä'[ñ±,(0yoÏ€î)xX •ïKªÃ+˜ø»mŸ«„$¦°ÍÚ‡>ØPéúœë[¦o™Éu(X)$ü¸Û 5¨Ž"8Âó í‚M†vÖ©qט™žÇ©¬c?D~Ði ¨6âœmkÐ85ïµg ²No¥Ž=¡–Y2paÍp€UÔ{m-šrŠæZLõžgÄžä=××5SP=4XÔKš¯õ¼®<½—ç”,&:LcSnú]ë´¨0ªu5© k!–fñ ¥ÀX>Åz¨þRãæ©z¤”% QU€^=’¥‹¦3ß &ºõª°Ž}ùþNЄçzGÏÿ<_Q€nÄôR·[ž›Uuid È; /%Sæ ô Ln O"ZÌëæ ˆ*,ά—Ûb7DžÚ´XOû¡·­û¹%dÁ] Eà÷Ü—&Hˆd®6Íú(ø¯-s7 Õ]ºáüª¹ÈGªœ† nbL˜µ›…ª†FÝfŒçm&—$|Þtü»¾€ÄLnš{Bór²tTu-’6(c®Ý‹ÙfíÏ‚.3“6k\tzuç¢æ÷Ó²î ž-õ_ÅÝ&²ñp—Ð%‘¾+ãÖïÁˆ‘«gŽ]䇸woiЂÿ«AkS,1r=‰Ðb±0ND Þ=¬îþ°f ó endstream endobj 521 0 obj << /Type /Page /Contents 522 0 R /Resources 520 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 519 0 R ] >> endobj 519 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 235.318 102.639 247.273] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result2) >> >> endobj 523 0 obj << /D [521 0 R /XYZ 96.768 705.06 null] >> endobj 518 0 obj << /D [521 0 R /XYZ 101.643 238.804 null] >> endobj 520 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 528 0 obj << /Length 1587 /Filter /FlateDecode >> stream xÚ•WK¤6¾ï¯hi¢w¶1ÏŽ2‡DÊJå¶}›ÌÓXCdÜɯO•ËÐÀ0›É »\oUŸß¿ûú-ˆV;o³xµ/W~°óbß_ÅQâ18Ù«{‡×_»‹œ¦,;®m~ué»]?ìÿ\mW®ï{»ÈÊ\gLǦàµwÌ´׿·ÑvÏÕ±c{X2 ù÷×·”B5íÝ·ßþúñÇÃÚõßy/:Rs•óV¿Ål—g5G{×ÞVÎQþîJß͈ì¡#¨ùë7?%Æe[oÇü©µ1N2AÞ€;ðvFÃ%ùE‰îiê¿¶èVN¾íãçÖdž—Ƴû2É}ïßKZe:Ûë~{a4u¡ãhds³­Ÿ[n·çLMwP C(î’6Q/:]eGÖ‰öé}Ëæ$ {òÊÅùqâ‰é2ZË…+ûOó¢U.%úƇ&_3K*ÇVA?‹¼ˆ¥SC•õ(I^š„½0\Eè–âzjŒ¥[ÏOc‚Œ¥¡·vÖV<*Ø•ú^ µüÓ2OGw^⣠0˵¯ I„[óC™Õ¸ œG±öMë¦$†Ëš¥N£ž¼µGÌùnɺWæªq“©IçD9 É :ëKÂJKbôœ:®,cÕœj+$ëÎkk¢/Bo†ÓÐéÚ×.Kbã>~ñæÅyE‰”(èïìþ檢qâ€næ(HÉ3R“[¨ÆÀ£•ÿl ÙðHåå…Ì´h$üG,…íIÓ10V,®­ÙËu£&íÂV‚¹ ߸ÞoÞƒÖ0W³e+âh$¦Ó‡N!´%Q†”µ-4j%2m ä2^ìÎw¾kb7÷u´)³N›knÝX'šc{êU€ò#‘³º¶U ª0zÆyÜ,5B]ePA°3êƒ Å:°„Ê.„,ÄY§¬¾ÛOÝD~‡±ïühˆKZ«fÓ “”'Ú5PäIJ…îhu±.ëŒêÁêß“DêÈìhéh‹w2qsTW ^D]#üdX d0è¥CG‘ûh¤£#2ö,Á­üpÚ†ÎE9¨z¦¿ŠÜÙz¶Ó&ðZ3‹þg3 'ÍlÖ¯' ïl·c^²µ4©¦ãPªæ²ÐÁ{€agË™þÂå0‹aøTYa'²¬Ÿie'5è_ú× Ú®þl2·5cÿÙƒÅ[¢Ô\t5ötpëŒþ**h.ÝÔ²qææ©<¹€-¸ejÄÕõÓÖ¦De‘ ä÷† è¹…<ƒb'YC ™¨@ã8ñÀ̯3Z{° ½1"ñøÛ°×Í:\ts!˜ºQ#z(Z¾F¥&ò¥Â‡FŽ8¶$”Õ=PÜÐÔÆmºQ•kqv¨FæÖá&¢dN º$5R N ö´cÓicÖ¦‰âá#ø—ŸÈUXv„¿ðK>Šº7do#CS\ÑRñšgt "E?Ð¥ÅÅ›>)|'a4lË brÙÎKÓ”bŠ<{÷ÇþÝ¿Þ÷ν endstream endobj 527 0 obj << /Type /Page /Contents 528 0 R /Resources 526 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 524 0 R 525 0 R ] >> endobj 524 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [248.826 534.197 254.693 546.152] /A << /S /GoTo /D (newstrata\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup2) >> >> endobj 525 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 410.66 102.639 422.615] /A << /S /GoTo /D (newstrata\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup1) >> >> endobj 529 0 obj << /D [527 0 R /XYZ 96.768 705.06 null] >> endobj 530 0 obj << /D [527 0 R /XYZ 165.763 537.684 null] >> endobj 531 0 obj << /D [527 0 R /XYZ 101.643 414.147 null] >> endobj 526 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 535 0 obj << /Length 1072 /Filter /FlateDecode >> stream xÚÍXM“£6½Ï¯pvSµxb4 >ªB©Ê’Êi}Û ²­Z,Â^&•ŸcÀ`ÏÎl¶rÂÓ´úõÇë'j~]ßÜ݇‹E¾ã/Ö›E À> ãÒÅ:[|°R&dQq±]Úñ­š/‰µ_ÚÄb-m?ÂÖ{–ª¿ o²¸¹¾•d™öçB=©Uï˜6—^W<©±$"ù¤…øˆ‰—A@¦B«È-ÓŽÔ ­šUŒZI.ÍùÂä¸3~“‡¼nÓ$Ö¬ëßxa‚"ÚÕ‡ XG\×µ¸TOÏ,©òGc˘äi’k“‚Rnõ.©“ÐOy¨Ž*ùú?‘»»'^¯Å þr úN»¸~ÏÅEa@Á±uPá6¼FiÑ”;{×»‰°ûÈ…(æ7aé\ظm…MpˆBèÄ #|³´©C­˜âôøá'™@õE^s3j{Â;ø¼9 ñ³­Ÿ`Hk®¦Nq³Ò¶ÏŒ•3¸ÄÅ( Ñ%`‚‘CÈS?z%q‰öI]ñ¦S-Jû¾ù PW«¬*ÊøþA7ÉCQHL¾ŽÇrÉô‰~䜉m½{ Ç'CWÌêaŠ’§èÍDñ.Anèt•m§šN`0n@F]¸a×UbzE~ˆ™¨µýýÀ]÷Çù í‡<ïÅ’u•ÔÉSµã£Ì\¢M7ánÍö£ 6IZ·ìlÕ_,’=““¹VMØU…Aˆ<î.‡–ä2žùÔ{pÜpdà>L;UáêäSÎÚP–&Ã…žF09vƒ°Œ¦®³ÂçIYæ§’º…ƒ-¼¾{_“ˆã¢–ðŒ4’!д9Qqàº"*…ÞHTþ>*mî—ù—(ª½‚#=¾.— ⢅¿sLÁxEžÃöölýX²8~Wæ ïæ(îPD+ÇòqpÎq2¬£#ÜÚð“~¨ŠoG¬Î«*m»=?=QÂs®àÙßoŠés.Õ|ð¡,Y—{.ºE!zP_ñ¥}Ÿ4Ý{g…O[º 3£µ<ÐLЄêh÷ù13]{økWrŽPy±¥¡_‘ÌÈCı©i†æu·ôÅ ÆÇÀþUÏ /)^ÖÛäXðL–ÅÞ\Ko U¤mõm‘”É–É9’‘gfböñvH±‰yv—eÓ ¿É¨iN7ìµ5JøW&–ã;äm¿<ïÿpÉôrÍ–Ó[+ÅãÿÓRÙ—Ë¥ðu]Öi³|j|Ù³S†ÚþÏ™,Æqï3ð­,YÊ»«+MäôÍ/2$bš¾ûÛC.ÛNOïãÑwg'æ.ã‹JrjÅÜ•y¾ž­¹z ÒyMÝökê~޽ªng®î—ˆˆˆýÍ¥cZ,ŒªôÿõA|Sz…ðáÖúQO9Þü¶¾ù?‹y endstream endobj 534 0 obj << /Type /Page /Contents 535 0 R /Resources 533 0 R /MediaBox [0 0 612 792] /Parent 505 0 R /Annots [ 532 0 R ] >> endobj 532 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.868 102.639 633.823] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 536 0 obj << /D [534 0 R /XYZ 96.768 705.06 null] >> endobj 465 0 obj << /D [534 0 R /XYZ 101.643 625.355 null] >> endobj 533 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 539 0 obj << /Length 2282 /Filter /FlateDecode >> stream xÚ…ËŽã¸ñ>_ј°h{õ°^‹Ì!;Øzƒä0èœvçÀ–h›YrDªÝÞ¯O½(Knõä`«X,ë]$yúðÓ—$½+7egwO»»(Œ7eYÞei¾‰a橾û=èµ÷×J5Íý:Óàokþ~FÌ·§ßî»uœlʈéÿ³OÍÀGw9é WÝñÔµºu<4V¾-_s4uݼ–íŽÊUƺ®¿øÍ£hS¦"­ië×¹˜´î0 Wvè_VŒmÕQ[IJzEø[bÉs¦ÕuéïÑϸé7aŒÚ~RÎõHqä ±+Ü@²tD¼'¿½!­v¼ ÉÝ|Åï(2 Ô({£Ï‚Äd‚q›ª{õ–ð¸Õ{úëÆ‚+¶q]SþÍöø=aæ§/Å$úÖ)]´€Š…ö_Ý}\gu…ÁzÕƒ8I‰FTõÁmà:FÕŒBsT\ˆÂØ”3]»¹_gÀè½ë˜Òc,2Ócˆñ¨èÈ "—±¸ 0¨†£nQ$§k¦8h`¸ ò`M»’$ ZÐ(MžÙzˆv´”À3Œ#wèµ€Ÿñ³ únp¦Õ–‘ƒ¥Í"Ea2µÐÔo‰1/÷i¨†IÔÎéž'•ð&±^ypdýjÍùÅ›(ÏÅ1‘1…ˆP{äŽJD°3xÄ‘`0PüAÍ1•iHIñÁóàxÁÛ€cgen˜(h€R#‚˜VVȻߔÂ*I&q—Ù& Sð I^ݧQн¨Þ€¬°j'D)wx“Y¸I·ãRÐy›Á£HfDSr ˜%›"žûžuCOš£fó×ú'·5P8¦¨(ÚÁSiÀâÙ óäÉ<3â™b`GÀ@ñÇjäÇl#$ò^°¨¨¥šì8ÎdžG/¬Å•ëa´}¥ÈÃU·r;££c²ÆºaðkÓÐ:Õì„)7±”÷Ó7Ž‘Z9Ф\Ì]@8ŸNŸ]‰G>)ƒÇOú<ŠòiÁ(‘Å®†' ¯aTÊ1Nµ5[žª(?8‰¼ŠÈN;‘Jùö†-Ø’î!qÿAW¡Œ7$æ3Ç¢XÂbâ°7á ù˜Ì•MÅbð•ÕIør 0,Ƶ¯%àf®ÄˆŒƒÚË‚¸5¨™åIÊX¹º‹ *? '¹dó(ö»á0‡ QÊìþÿãÅ©œÅ¨(B‰¹œ%ì­[ ‘«=ÖqCĸ¡oÑ“q)âtCSóä88 æHä5̽áIDrÒÊ qp1/¦Ð'82PHM`žòº#Ï5I^w¶ºêÚzI«±޶ƒP¤µõØyÆaˆ™¾æÑÂâcŸÒËöM•|DYÓ$ømÀÞPF¨j*½QÞ4ptà@};F\ýŽ£^¯Ï½qN·ÌÄ´žGbÏj.ÜÓ÷õ¿¦½ÙðÔËB[ËŒIüªº¦¶ðnh+ [Õê"ɿݔÛíü|ÈfË ;ê“2=w~ÚŽZNFa–R4œ-#¨øe‘:‹ø|ä| åIÆ?óño ˆ'ž,ºý¶Ò˜;`·GY¯)1+êY1ÙT»×"ŠZ>4Ys4òͳóá ‡¡^£d@¢Ng±¸P5ÿÇ?á99ô@4é3/¤›ê­ö‡ÔÌÖ"Ö竨 `é—*˜cC8©žK+Œ¤XAî(ÖZˆ(x©øN“‹4¼ZúCŽf»Ö,ÙœIÊ`<ñÿ™Ž}0özK¢Sâ"öÁ0KËH æBâÜȰ ƒïŸ(:¡=ûêp¥Ú•¾Ð|šFR€l…XÅ|áºÖ›Wbe¸µù<Ãýœá½nu]ÚòWO’)4làšø°YtóD4D«àhtK\¤ìÄ‚¢í”%¸»áÍs¨ŽgÏJ@¾ü”åâiÆ&ŠtóãW¼ þ.'@ÙŒí·mðТJ7”“нpsĬf±[=ú1á1%D[Lt&LÀ‘Fù[G}Ä;7á0µ³ÊÈáõ 8ùb ;8s€æGÆÎ„gwÞeÇß«%˜ßq° L#[Ù#®¤ÇÚ¹¥sä‚SÝA¹›Þñòï#ã$ܺ“Ü•ìx#Ê=ž‡ü!ܺy7B¯dVÖ蛚dýí "¬¤†ì1o Pyä ÙÎ.s•\µü áýºóÅô$X\¦R«)ú}[}†ëPcZÍÏWˆ™ÚÇ|àZyh!¥ÏÀ#ÇMÆ®0Ŧn‚Û©Êy¾×ü÷´RÞ/>roô¼Ð•œgI)×Üë̯¨ÉÁT×׺Ƿ‚° :<ºõDpYõŽîÿ™ñ’NîËåxÁæÁ¨#qѤ£›+`üŒbüXþ—>c BÃ’ñ·>m°ê»¨»•$™LÃùúhsbo}§¢²Ç„ͺÄ]x~fiæÑ1 P«1êƓƶƒ“Ç3Dã kç1jI9î?»i§ M±ïÎ>Ô—ÇÛ±OÚ×›“àEîÌ“tA¶“z8¿É>É&$p'ÏÇ;¼¼NUÝëé°`e‘n¶Ñ¸ÀÇô†ãÖ|âp.òM™ÜÞco7›½”ù€˜¾¾Ê÷ì<"öO†Æ~÷—xöGcÛ¢Æ}äbfø*ûF"|Œò¸…ÏFÐ~e`j.ñl£x±¥rkúôôõß¿úwÄ«f`Ÿr4ÚnÁªë8L6eqcÛ·O®/ü„€âÌŸA_ž¦vnô‹nì|ÉR¯ä‡Î²Ìßáƒ5¹ ­ùï DX°K»,P»wô&=¼#ÑÄ[Üò§Ë_´7Á 'Ç'íó_¼=Ó·O·^ë·ú”¦£æ¯öÑÏ7;ß8> ËM‘ä7žŸ½²e›¢È`C , &KS¤ûðëÓ‡ÿµ,‚G endstream endobj 538 0 obj << /Type /Page /Contents 539 0 R /Resources 537 0 R /MediaBox [0 0 612 792] /Parent 544 0 R >> endobj 540 0 obj << /D [538 0 R /XYZ 96.768 705.06 null] >> endobj 541 0 obj << /D [538 0 R /XYZ 97.768 535.193 null] >> endobj 542 0 obj << /D [538 0 R /XYZ 97.768 503.866 null] >> endobj 543 0 obj << /D [538 0 R /XYZ 97.768 461.968 null] >> endobj 537 0 obj << /Font << /F35 119 0 R /F8 112 0 R /F33 118 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 548 0 obj << /Length 889 /Filter /FlateDecode >> stream xÚ½VÉnÛ0½ç+ 7 bHŠÔTZ >A/uO®‚,ÛBµ;E>¾¤Hj‹ì8qšƒ Qμ¾Y¾Í.n§&¹Àµ°5š­Fˆà4²¨ 0ß™-GóI”.÷×ÅtòÙïÝ& 6¿!…Ë}æ{žü]ÊåG‹Å<gÆéXñdÌ¢$¬),k-Ç0e­u¦E–%)&$]dcA„ú@P&ÿo½,Ø„y>`Ò=ÛÓ®Ž0.Thþð!£ƒÀk ýüq @ÄHŸÞÍÃÏž«Ã@å,åu¬{©,L¶ºô¨|€éµ¨M'¼aá¤ÎûÛmüØpZ9º*Ó€EY*6ö‡d‘º*NÄâ騫%§]ê'aáM¿Þÿ¼k¤ Ç–Ó ®¸«ó1 -¡!;¯C%IÃq!néœXUzÞ™¥‘UüÕýè d•ž÷ šL%ïTRÅ{ÂjO¶âU°š¢ßM?Ò¬ê,Ýh€âÈÆK ¢Ñ©g‚Zªé'º…·›l»ª‹'‰Š"J×â3Z6ZŸžä;*@ZÆq³Ýˆ\])Ãy¶«\ÂbïË¡†)sïX˜öóºòšή¿+ŸåѾӯ˜"Â0TLÜ“¤F,$I¡†¶té4D}m<.ÞPˆº7.h9ÝÜ\FIïÖ˜n ´Ë?~Ýßë .Ûúhcohš«Ë&ôÁ¡•F?<¸'6xˆrª8+N¶íá((­ž(bh›oÔ³É]tcúòäaTC:}þ@§ÎÇsÿímA瑚: ´_ì ¬ÓÂê“SvÃçæêµ„ØÅÝìâ,úÇå endstream endobj 547 0 obj << /Type /Page /Contents 548 0 R /Resources 546 0 R /MediaBox [0 0 612 792] /Parent 544 0 R /Annots [ 545 0 R ] >> endobj 545 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [238.365 582.017 244.232 593.973] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 549 0 obj << /D [547 0 R /XYZ 96.768 705.06 null] >> endobj 550 0 obj << /D [547 0 R /XYZ 123.92 585.504 null] >> endobj 546 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 556 0 obj << /Length 2725 /Filter /FlateDecode >> stream xÚ­ÉŽã6ö^_áC6PfÄE¤¤“F7&3“à …9LwPÉ,[Yr´ÔÒ_?ïñQ›MwU¹”)òñíÉúùîæÇ2^¥,ÕB¯îV<ÑL'j¥cìÜíVŸÖ=ž~Èûã!ûºÙÆ"^ÿ´¥ß¾*‹¶ûÅQ™Nå ŽÛöewK}•wE]áÂ3üá4ýì±áÌíæËÝ?Vþí_¿}pûx„.9Ò‚E7nK_‚;¡ÚnÇlÓ¼]*Äf¦b½ÄvUPO$%ôa©~üÈÕÌÛ4eÒ "j¢²'À…Ŷׄ –P{Ò¦C Ko¡dËÖ( Íx,€ i'„-;CæÔ2©{Òãhš¶k².s|GKÝЖïš…Næ\ºˆøMX’*‚û¥ÚlçëÌÿÖÍ®¨²æ¿Äºí›ÇâqëuVÒz3‘¬-Ô÷ L¼þŸÍ;Z*º- Ä(N¥¥5\GI ØÒä©©7[€Üõ¹ÝÑÔ=î| èYJÁLš¦@î 4þù륬ʰÈÈГì–&SÅ®—J@™Æ£Ø±[&Ô³Ã\Û.§A^ƒ["ãue+d¸ki!k¬Írœ?øiúi‹j_zRdÞÕ }w‡¬#ry (ͺˊª¥äÝ‘oxZÃÖ€/yGÀP•1‚–eHŸUG¥ƒª"p¸yáv©5Ûl¹Öôcf?*ë;ÇÒÙÆ™ZœR¼n°h ÈÃ>";ƒüð]?x° _“üvo›Ö z¦eÄûš9qã4P¾ ]·¸50 `iìøë³Þ;Ʊ0‰ãmi–ØiiÆÃ•uµ/*À´÷Xk¿a´;~9þú#êLhÒ76kAE¤ðxÉþ# ³@‘Ñø©(K2¤B: ÉŒJ–WÅçˆKÛL†€ÉQhã„ñÚ”$´û½omC€¥ÀÙ@¦ë¶¦eï©0òTŽY×Ï4§<Òȹº£Yû-Ev_Z‘Ë¡´4è«mʈú~®©û®¨ÐžÒ¤ƒ"Xú"B0ȳRÂI’Áä(–„¾°:¶€¹ºkü†Üç%õÔØ]Ö »нÏÐ 87ÔMú#ËPìÃÔÎæÙÎb%`!™(žta êï2^+ç2?r1Ÿ¥Ç¼~>°…¸Ì’‰a’›¾?m@š¢ƒ‹vÀämšèþPi4…¦¾%Ÿ|UÀiê:˜èB‘ ‹ôY!B0nÌ™—ÌKÊ1ÅN'½¿†ú†&ÊÌñ‚Kƒ}¤ãѶ-Í{ß›ƒø "ôÏÞ’wÀJ 0ÍSýŽ/!†Ab\ðã21$³€H†)Ã^‹jW<»>+©m F³xªsƒjE*§¨g01@6n-Mí ”U-ˆŠ¶‰"׋ Yqm–RŠ Üí4àY³·-B8Z˜šÂ9Q¥PïÊSAƒ²ÐÆQá2¹C뢆T!GÏû‚–‚aG|ýa*—^OZ¿qo+ÛøD„ôÓW؉ƭí|­ð»—ý‹Ûåò F±ªÜº6{:6Ûgê—Á-P± ‚¼G‚ º—TD”á—\ -Fèß_ǦͳF §nÏNÈ\%ÀZ‘Eùà Á«*è»qåÀû‚Úï.*‚ïE/ÏPÕJDàe- †vÇS»â]÷ÇŸP¥—„A á¯%|¾²¾¶+‡®!;8oFÅSêÎË7nÈ:hŠöP=Ù’|°‡t>÷âFã§ ÿ)sHWçqŽÌ#Å,} Í1ÕK9T7rVø9f¾Jz•.:ß4wQš|T1nCöóRÜ–ý±¢ñÛzOm…Œ’¡òÆØ*ú*ïhyȸâ{ážÚ&¨³Ü¬ñ(>G‘¤,fVš¥F&îà$!åKÛ,¤6`^É4÷¼Š€j¶xÿ¢Ýù®›1À¥ôJ@ë! èï—ôÀΉXh™_jº˜Ïµüäú€KŠÐª±Ó»B¿…Ú…‡£F µ¯rG]P;&BLîctãE›–¤,;Êl©9àBŠA;š„"·ƒ÷ðË«ŸÏÇ6S½NÍ.¢˜i ¯vô÷C®B˜Òé2™ÄQ€¦d‰Å õÛàɰN×Yx }LLŽÝëÎÈã1ã‰Î8¿ž¸ðª2å è/cüÿÉ í.QqÉç3\×ø‰_áǼÂÏ3ª'àIÉB•sd¨Éø/ghaÙå9†3 †¡úo Ý <,ÙP £Ø«à9d;ÂófçnCÑ­¦ûÀ¡™Vâ,¿áC‘¼Â $åw0rl!5(R¤€j$†€»Œ4Îb95íxOoóŠ‹é*¦n%À>¤¥$ƒ\]ÒM™†cýp©@õbNwu«)1_¿ÇžÔõ‹žòÑÝ®w¶¼¥>“(AB^“Wl¢¿Ç7Þ–±‘Ç eîðÂÕµ×]†Ýuî:m’Ъ¦6ÇÚjÚNM5”MÉ㳦z<=hw`&“u¡NÀÀÞ)³~ ö±£éü!^"èõÑfÕüT3[ª›b_Ðñ¾vþ²XãÉëx´Qò»§Þ¿9ÃSø+éfvŸ G1¼¾Y „Çoµ•¿cRr¿:»^ùжþ*·îê&apÚ±í%±“ˆ¤J-RÂô{°î@VÆÔ=B] “$]Ï5ÃH+£Ÿmßøë¿ Ľùy:8‚ިÂz,Yã¤ÛI¾ØÙÏžJL-;å¦I~Ĥ&k¼±'7vMX—ý1¼ ,š³é°YägÇzúôlI5üýuçtxWók®b Qx Ú]Ð ‚š&¦ës÷¦sdWÞþðáÎ xðý«¾VèýÅ)s÷¿[º@=orLµSÎ+Bïns¼ï†»?Å¢ä,êíóÉ)`þÎ9N~üËñ³ðƒÇ¬éýÇò­ûüVB« ão>ÄŠ(…M}ã5´Ò½œì»wü›Oê®÷_ÈÇäSSï¼? ÀôŠõ9™Äº@ŠBD[Ä4ýÿ@  ÍŒÀf ".ñm`l†öåÿµYK> endstream endobj 555 0 obj << /Type /Page /Contents 556 0 R /Resources 554 0 R /MediaBox [0 0 612 792] /Parent 544 0 R /Annots [ 552 0 R 553 0 R ] >> endobj 552 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 163.587 102.639 175.542] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 553 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 163.587 218.08 175.542] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) >> >> endobj 557 0 obj << /D [555 0 R /XYZ 96.768 705.06 null] >> endobj 551 0 obj << /D [555 0 R /XYZ 101.643 167.073 null] >> endobj 554 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F7 257 0 R /F11 183 0 R /F10 255 0 R /F13 311 0 R /F1 256 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 561 0 obj << /Length 2337 /Filter /FlateDecode >> stream xÚ½ksܶñ»ÅMÛD”¢c€IOÓNÛ©gœÉ$3Ú|pì)Eât´yä™KNú㻋]ðqG©î+,€‹Å¾_8ÿñæÙ—/T´IýTK½¹Ùm„T~*7:Œ|™lnŠÍ+¯Ü]n#y?QPvþ!ëÛò?$ütøùç´Ömszö;a÷¯o¾þò…çÌâØO”ÜÄhG(Ky"? GŒ_šÖÑ~ŸÕ´c.À˜6%Ãks_d}†D7[K_E 6ÂO£ˆ–ýoº¡ý@ø¿ÝÒÚ ½iQ<ºiµe÷Ž¿.Þ\8@²É¢<ÔÙÁtxß‘Ÿ,áXTe×#Æ·ýæ›ëQ|¼IÖ{ŒÇÜ!ñÍ#öÕÒuô”y·¡ð“ЙE⇬Ý/å|ø×“£¿ªL}×ï®x/û eq8yS}µŒˆÇtÚ5í,Ê–ŽÏ×iœFÄJ ã'ÕMC_„úÄeý#þGòá:±>òUyýÚÁÕÅäX¾Š7n³vMC´Î«ëòõ’òÝp8·"a|1ÆÇ÷€‡XøÏ ýÙÕg´kà%¼vE€¢w¶º²âÈ[.eâ°~U¾~#W,º]»s·fW‹(–adÿ/|Ó¶K»÷­~Ôô_ŸÊSžg4¹ìS}z’ÙWÿ­Db‰hªÎ¬“ÚÑ“•o X‰O)XyÃz,"OÿŸëÅJ àñ„µó“,þw’ç—Oúž¥XMÊu÷ÿ¢I ŸN‚SøÕ"þ‡1Œ>™Z² FY±“Œ÷º@˜Jhø†6UY›·Ê{;t=³ÛË­L¼æþ|À?æùå6ÖÒ{IçûŒObÉd´Pk[›<ÒÀ’ÔeCÑŸË GMùr_ö{¢ÙÔÌ£má=}­„¦¥¯Âd½/N¦ºXPPtÐáP»øŠÀ|áÒÙU•B0_FÚËÚ2»­Œ¹Õ*ð^\&¡×ð¹Ér”hO_­5×=ûJ,Æ@ sW8*Ï(ÁFûi¬D‰|‘D ‹¦!×™s žÎp:sÌÚ¬7ÕGáÞújÜÇ`„ô±>ÇÍû!+àZ™Ó'L ‡5©c˜ýI½ ¤|%â¹Ø+K9RT3ŠBù.+_ vI°Â‚_û2Hµ±”~”*ÂA÷ÉÔ»š þd$ uO„-|vg3M1¼Ù*YJ‹\;‚§1LŠ!Ç+Aâå8=« õ(o A w¶ó‚²ª"KG_Óiê3-T® ¾xZÖtJþ³€Ú0Að¿!ïMï ¼¬½ïÚ¢¬³ö#¡ca¬f1Ÿ«ƒãI U_«2‡iµR MUqö¡~Aoñ™¨pTšd$ß$<"¬2*™8™ ŽvÆš@åCš£oE¡Ÿ†áRÞo‰+öÃ@ak>¯<*öµ OY¬¡ÆÂSb5Ô©w(»®¬ïèãÞ•:Ø[gã¦+GÌ<ÜgGÚÆhjãk¥êЇsŒÌÃ+Ôyìj?ˆ¦Á{¨sò²éZ§Êcé!›3ð}É©B‘,†«NüÓ8γUŠP=ÿÄÙÏå!×Rñ/,¨QdkH–MxÅWŒÅ)Ô—·P°™yF bV|Œ3XIÕ·¢Óç4öAÄ …›™!àk¨Ë÷£”­ôö„ÄjoMnÇ4 ?m´ð<Ï`~BR@ä—68a×q—°ù ß·jŽ ¾ëØìh¥v¤D aE uG—¹̹ÛF‘m9½af¦ïÀ€JB‘jÉ1C¹&êoÁ{Ãáh#¿±N?%­©×2~Œ(L¼¿(LÈè„ì›Ø§ÎTØØÖª4ˆ—!ƒù F¼%ý¾C ¾ÖQ¡X;*¶xÕ´ëÌ=Л;_- 4Þi&€ðX›ýZâ¨\Ÿpâ‡cøæÍ ÚŽÂtG+U2õPõÁ–ÌÛinÀ÷ÑŽ×ÛδƘÂ|ß–½9n>ÂCO H{ÿ “Û¡w¸Ž;fäpåÝ~&懲+yÛ±+Û¬¢ºµz.ŒñÃe¢¬ëâJ3+Þ"ØuØØ>ië6 QÐñGÚQ„µ ·ºlçƒý±uÑ0ôè¬ø1gÉ0ôãt,Ø6Îü%C_'Ñ“C©(eAä©/8¸ÍWÆÈñ¶«•©&'†=´æ¸p°<³#b‡ÞŸ s[!(!Χ»´j› (–õb´%UÒÆPó/rxö’3 Åf`¯3×ï3׌éòˆhV¥^©ƒ$v©k04q@¢ºTÑáݬµÃç®m0U“dÖ‘´±”Œ*ÍDk2¹SÉ,#^©ÆÀA<¬<ë± Ùr›7öÙ Ö!”´Å´Ú¦ « hå§‹˜´¥‘)Ûîδe–„¶g,û›+ÑÚ1׈Ö̉)na˜ýÎ^ÖcÓ?­ô³–Fq8ãîËÑÕ $ÍÆ1ÎÀŒ·‡ù†›(¾ÈgÃŽÑð§J•^h |êiþFKÙ[Öp ŒÆÈÖ"dª>c2ÝìyYˆ¸â7Qkù®Üã£-ECâÅ[ЇØS_à”¯<n…d˜ˆ·0\â AÁDi;oê&Vx>º+0k÷|ã•íØ‚Q¾±RÐ"èd> endobj 562 0 obj << /D [560 0 R /XYZ 96.768 705.06 null] >> endobj 559 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F10 255 0 R /F13 311 0 R >> /ProcSet [ /PDF /Text ] >> endobj 566 0 obj << /Length 1518 /Filter /FlateDecode >> stream xÚ¥WYoÛF~ϯ঑‹áòVå¡@¤Eƒõ›¢+re1!¹—´å$?¾3;ËS”Ó´6—;÷7õëí³Wo£ÙÊZN0»ÝÏV¡Ñ,ðCËqýÙm2ÛÌÓbáDóJ”Ï^‰KÏæUš :©˜gBá9œ§Šî8=viE‡,ýŒšŒÄ^f™Ä÷‡´¸3Jd.d!HËÃAÒí2q£SUâxlvj;Ò„ Èz%î„s%*¢È½¡@UÃÄ ±c¯w‹¥ïÏ?‰¸¢›6Úf)” ƒ7pÃD¾Q A! PKf‡–¿AURò=ôÈÐ'1AçBbÂ+nA8/uà²E’±JÌ]Y>°éÍ9Žà‚×á( c¿0jÖ°˜H¿×“!¶j¤?ØŽ÷o-¦h N¥ çÝžîLá¿Py[‡”挘= !f@#븞幫!²öWf¯çSGàaB#Ü:ôèr /ÌþêØ˜M/êÄÑeAÇ?b©…€”h¨’ž;]‚†ÓãSñä¡ .ÃΕÀn¢7U—÷éÀ»n¨UûPŒºT¡eʼnÇUöH½Iã¯5ðü`û6€‹ÍN“EËdÊœ’T$ÀÂnŒü¡QkÜÆ™ ˆËÚ`˜AdÖ¥È\¯›#l敪 `ɬ‰²ÂààÍ1 <˜ôX\+ ý¦¨¬}ZY±<ËXæÇºz@uÖN–‰ÙƒzÛѳ6ý £ÍŽüaU¥P«¾ãkHóT)ÂxL5€Dúöž©²Š:Ë:²fAÕŽ Q²Ñ ªÎ*’|½¤gU–=êìA¨YªpÀ AœŽP•æåä˜Ã=/sÞ°(a6­Éå”M]çx9Å|Ö ±Ÿ€kéÃb-•B`©Šaûºˆq‰èz´>?6U•0àð/޽2UŸ'ƒ¹ë:0í˜ó¤¿®ÁìeÃî)*^Þ‰æE)ÐÛ“CÕÞ\u¾%qHSIºßÃ@(Œj;øëœgjjùb&‡ïa­Èm¾@„^˜˜wmGz KÜ·¶[ç-m|‹BÉrd”—v~åtW:ð.¨†Kû?áÏáebÄþ_hðuƒs§o•+ Q½Z™©•3y“WtVðƒö§¯Åž²¶o ÑÚdžì—‘«ãÉb¯¬Èž,½(‚eŒ"D4ÎqÅÛÍF‡¿I·Ûí”»ú»h ùpHãCÛ`?u){Ó´Ø&½a[zùÙX±¾^÷xí¥Ìô§_&Š»JÛE—Pb½¶ŸÄ)üaœ.fòŒðrà?½,‡ñOTÊq¼—¦£¯Êt¹© j5¬‹Z5àéuËTþ_v~ŽN w5e)KÓ»ÚD!ÍF§ÌsiLòfÌLÊ*žfÃÔ’ô ¾´a{ÝMåáG…óÝEÁΓºÁú¦Ç)VSŒÿLÎnÆÕºÁZÛ¶ËTS¯ú“7.Ecqrô¶†7l{^NCÊËó:û‘I3U‡9? »´ ¦Ñ?oXžNàDò†¬5\›e ¼ßkZœ2ÕãQ¬×=ñò¢!MÔf>ž›éùköÑE= ]«šÂ’v—‘l‘F?“o"µªÖ05¤‹ÂI5Z|>%Ë.]qâ$ß~Â@Š·Š‹§'k77I)ë·“Võ¾ˆR›tßKôGç’BgB!|Óü]ç ãÁ¿&Úç×Ïû_Ž˵iƪ Ì\m_-Ao´L“ËSÇTa> îTc u á‰Zaðý_:àÌб”Éвu³‡¾ {#§µñÇòÏK®Ð¯®i›ðUÒK ¼[¥ƒŸÕ¾oE6åâO—”ú+d|öÛí³&ø` endstream endobj 565 0 obj << /Type /Page /Contents 566 0 R /Resources 564 0 R /MediaBox [0 0 612 792] /Parent 544 0 R /Annots [ 563 0 R ] >> endobj 563 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 574.047 102.639 586.002] /A << /S /GoTo /D (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 567 0 obj << /D [565 0 R /XYZ 96.768 705.06 null] >> endobj 558 0 obj << /D [565 0 R /XYZ 101.643 577.534 null] >> endobj 564 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F11 183 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 570 0 obj << /Length 1782 /Filter /FlateDecode >> stream xÚÕYKÛ6¾çW½DF׌(>$õ¡²‡¦èZlö µéµYÚè±üú9¤,É´7‹M{E8Ão¾ù†²ß]¾zsÎÄ,%©Œäìr3£œ“„Ó™ä‚DÉìr=» òÍ|!"| EØ(²É[Ñùõå/³($iÍ”’T4/òF„å2kȽZµU­oÛì¦Pf‰¶ÎÚL/qÕ´jwwE¯¯õÝ™Y‘I"édÅ6ß©åªÛ5ÝN/Е΅~ _Ý¿Žß,I7Ÿ1ü%^öoãÃÃw¥{WÝ«²õ¼fæOø\©²©jÏ‹øàè›MWßbÕ^Kc›}µVxã_±]U×ãE¿Ô& ÷Y½¥Q4ÿqº m9›ž ÁøØ½*…¯ÿ=RŒÖ–ß…âI —ž7甪|áô Gr°0NáÊHJå + }É×6Âe¾†Ì ‰1ñLÔ$,†ÐÍZŸÏ$_r¼·Z5]Ñ¢—ŸcNU¥Z&È#èù34yŒìàÉ hî®T:ÝîY/nG¢Ÿ~±ƒ:8\.’„Šè$ ÂŒŽóÔ9´]™é桯ÇðyüöeøZCöÚ¢P¨ò¶Ý.ñbJ —=¶ô¦ª÷„Èq˜—x¥o.3% I,Ò“`8òŒÜçåúq¼¯‡m¾Ú"Ë%x½Ê¯OÃr×ã5NrêJû<;[×ÕÝòüç_{õˆ{HÆéz:j~ Õ†™Æ,`·"tí'Ç4õ•ö8ˆg™:iµe¶SÞ4"³o=Ù€}ßVþR°‘$£)‰B1ÌE¶êèÊRä ^Û­Ò¶j¾ˆ’`­pÞðÏn“ÒȽâd·FP‚`‚°HôôëÊU›W% Î)5ŒšƒË-:ÊVmÊ÷„wk G5 SúªþLæ É¢àÒÍ«:TÖ*»­jƒ`Q h…òPüóû¹Ÿ˜sdµÂúÁ‡¬¸ùR¡òƒ\\ÔªÔ£6_YÑØcÌÄ;ÈoQi£3Y¹ÆGï7uUBÐ> endobj 571 0 obj << /D [569 0 R /XYZ 96.768 705.06 null] >> endobj 568 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F10 255 0 R /F1 256 0 R >> /ProcSet [ /PDF /Text ] >> endobj 575 0 obj << /Length 3186 /Filter /FlateDecode >> stream xÚÝ]“Û¶ñý~…úÒòšL?¦qg’i2Ó<4MãNÓqü@SÔ‰1E*$äóµýñÝ/P¤DŸïZי鋰‹Åb¿°Ð—/®ž}­r•'Q²z±]å©J“l•ØTEÆ®^lV/ƒêçcáꮽ~õâ›g_ë ¶Žb•çÙj«ÂþAÐfx‰ ­^­s'1£Ý Z¸JTžš ÑR¥M´Zk &K× ÄŒ2:]MúR1 ?™ÔßHY•fñ©ì!J?†6\\1KãUÈ8n‰ŽA†# }½ŽÒ,x¾ {–MW‘>Rò›?-lR¥ó¹IûM¦³M~€Æ®,‘VaªWhsü(m‰'´®¾zqõó•.•^™8Q$žÄ ‰Y•û«—¯ÂÕ¿Y…ÊÀ¹Ýê~‡*r›Õ÷Wß]}yne#)“¨Øä¼¤—BÚ-bíF¨ÞW¸qu½ŽÓ4øã–ÝŽYÓA/„‰Ah»¶¹gœ®•¡ÓJi0.àÇ|x¡)»B¨ü]6ÝÀüÆÁ¶ë÷ ]sÄ9Š6ÊNE¥ˆÃ< ÚΤã0 X ¹Þ´²pW··ŒÃ\à)2)Æqp<ÖÔ‚$ªâˆí¨jy$äFŠM+“ö¥ÌÞtÜ+Kºê00Ø Ë¯ë¡*=À̸­Ül£r¤(¢ÜÎXaonW·<¢«uÇ–Dyð… «úÎ'ÂxtrÐòz1ì\D<6X7Ý-CÌj|ú²¯††”ðŽ;ªT¦p2\í‚§ ®èŠa?~"Ê. ïêÉîà3&Aß·È®G8^=(R ¼ðWÁ˜ûµ;†î°«a² '€]5ÕÃOØb]‚¸R²·A?†:n‹†1ÊcïÅFó‡Ë°R2”ÞÑw ~B1dBDÖ31s)ViÈÃ"ðœùíÙŸ–p,{ö}·¶Oðìååb©²Ö€J²«»]Õ/zÔLeáÈ×ãböâ‚&ÍWÓ¸ŽÚhŒh¯‰½Ÿ‡®¡$]A à=!#ÁØ$ø L•Ð:IAv®Í¤BQ–_m{T1O¦_e!ÝèWH«ðãm]±oÙð7µÜPêÈmjÔNçí]ÓÂbÇvƒ&ƒ óœD‘[ìhÖN6> 8 rmªÂí† ànè«¢¡€1nœóOnî¼Ûø'ö°H°kC ]ô4FÐõ›1Þ\S0JB|ßñh½õèS¤êŒCÃÍôüdw»n)’lÈñgGÅ £Ä^z(*^êmšª,zÀÜR‡æ”ÖzYDà¶›†¡]!=õc´«¾’¯Ãp¬6ÝÚÛ$xp ß]¿xÐJëìñ)Ê%‰Lü!W’@F>޶ÄC¨¢l’Îêô’H¬ÂØL%ŒYU\DÞäìØ;ëÈÚ ZâÊB’™>+Ð=›‡ÁþÀŒèƹZE¤ydN&‚1×s{:€ÁJòè6·c½Û`SµÝ¾†píG(úN1Ðôƒ€x¶G™Ù<ˆž™ùŒª(Q¥vK‰¡€t ÷èÚàT0´§tÉdýf)’##(›N·€‚Ó±$«Ñîû‚xÅè²Z"|§~1ŸJMRe裄hso.õÉ=wf9¡è(¤ÁïnÑ–Õ¸Äýt 9µp)6‰dµFV‡–Ý„¡t©¯0ˆb®l#¹³e®ÈTCéôˆ½›Äç-—©D„9B–¡ÏH#æáÏKú)¦«õí÷åø1‘GV âGT ⛡Ü\Š XåÔê ¤’O/ó'Ý÷oXWOžxª§¹‚;Ó/+ýHe*²ñÿÆÃÅ"„(’νR­*us–ÓBR'i®\ÅBÏæ«St#Ùg²Kyß××™ È‘ÆY0ÔûCS ×!¸G‚ßÄ öl W0$×ûÌ'Ø„¾_(c&XiN>¾ö|º ròÉ+ÈO+Õ‘ð·þ‹¦9;¡c[ÿ|˜tAÆ!ÖŽ"†LÒä&¹a2²ÈU¡ŒîìB¸“¤’eHïæÊVvû%'ÓÕ‡ZnJ.ØAu´Ô>EÆ!Ð-¬Dá©ÀÀ3·´Y𮽭éÝvܘŽÁ£bÍlGÊû‘œ“:üýf8Ñ^¸< »®w°æ±…\`V/»+<à„ht2ÒÜv}ívÞŒÇû(êmPL~pÝá•X1˜žZ(! ]‘}(ÉEhLrÿ@™UHeh†êPô\ÓÁ/ª=â¨ãvš…£ˆòD@ÏÉôúîuS Rh%¹ ½ºç¯ïÝB•} 1ã"jÑ7ÅŸ:¦¸ Ù›ç)ÍåŒMú l©»‘TàÞ0½µä*Ó㵯ŠkÙàó5·ã™|î¸ã×ÜàÙüþ¹;ÕT¦²¢‡¶“‘,^—ât\÷Û%;ËÒä)õŸv©†tþÞ„òàä$D% ZñÆßVá‹Ëj$B¹¯Ö˜sš_äÌ­W鿬¤ÒWû¢n¹ ŸeÑ”hÙpƒ`v·fÔKZ½fee­N­Õ¥Âfœû[EÈ5i;ß`ˆ.ˆ2Vy;î ÷C9¼}öc¸¥K“‘«=+W°aô×Hä^¦¢Þœ,…;YøÈɉÐØ“þ jŽJ8¡-Ï ë!+ “e=ËÂéíö1z¶tÎTf“§)pÉfž*7Ä%_ kÊ*BzêÆ­Îɉ ò=P¬sòä0’ê»#IÀžÁªï;y;Ña¤R{VÊ£òRîµ€Ïó8­½@' §…áH'Ó"6•VB¸~˜ÇJ:.Ö…ÀÕJ\¶ 9Ü…R9êeœ•\&øBYvâEŠØ¦‚8yS¿­7ÕF-é¨ Øc= GÉ'}×¾œÏ'}U6Ç i$ ŒJÍõQ×¥M›ßÍŠVÈHÉBJ§øJj'ñéo˜qÊûÁ˜˜|¢%Wïàt™ü}z€68¾içò¶]hM5áø—Œ3"•ý ˆTnÇó˜XÑÔ#™È¨$>ixd¼¾VI6&g“  DÇ, –â`ÅI’±ÊDé<µVtFœCC>ÚçtÚ¿–Î×g-7 òÑßÕ Öª­Jr=©xÅ&ç‹@ßíýZ¤:;Õ1uv^½ÔWItÆj’/Pð¥ãNRêÆd©EÀÁ³îNTýõ…+Å8·tG©4ëñ“œ¤Xr©ðY@º'ŽÄ¶Cf@^çRT¤rx×ÒÝ_ûç²²Á_ª}EáBjûS7WÈúèJÍM¾/äw >ŒÑFÅÈ|vî]ç¼kóèráš(Zojò Ú1]§‰Š’è,ošÇÇHû ¹1æGPB{ÊvÃàÛ£;àÃvÓ:ÐÊe4òÉÏ`@Ì »ðÚñ;„ùÍÐÅý5üEúˆà@{JÇßÍ—9ò{1’êxFuͲn(ëa&3n¨›¤6IÂi3–üRŽh)jøtŽÎàRQ4cŒœ˜(-Büg^8cvšaÂ,Ôï†_„klj˜ŸT£”˜@ILަ â™þ3;ˆ„ãÃ?|Cv¿ó ІÍÿLP…äróJ,‡sþ’6ð.e8ÏÚv"Îä}wÜâë K×(­G»«ŽeFç¹8TØš9{GçfYýöØÒ…môþ†ûÞI{çÀ- ˆÜý¡’Ï·×Á×äA5ž½nF*6£Þ.°¾Ž­¬>«ö´@wÎg[v òX j_¸¾~‡ø£ýÚ—)báŽÃœÊýËOèç½Z¼B±A<4{×ï™ÏW°…»ðóüùŒa¦¹ÖÓh76ÀͧrÌDˆx{h×mõ¶jÝœ Èì- Rè»;°„îÜo‰U9Ê“.+LmÀ„ ͯˆ¾ÔUµÆ[×”{ µìf4øñ4ŠðÕIY+í‰öÿhû7´qik endstream endobj 574 0 obj << /Type /Page /Contents 575 0 R /Resources 573 0 R /MediaBox [0 0 612 792] /Parent 578 0 R /Annots [ 572 0 R ] >> endobj 572 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [133.758 237.989 139.625 249.944] /A << /S /GoTo /D (agsurv2) >> >> endobj 576 0 obj << /D [574 0 R /XYZ 96.768 705.06 null] >> endobj 577 0 obj << /D [574 0 R /XYZ 97.768 241.475 null] >> endobj 573 0 obj << /Font << /F8 112 0 R /F1 256 0 R /F11 183 0 R /F10 255 0 R /F14 120 0 R /F9 259 0 R /F35 119 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 583 0 obj << /Length 1141 /Filter /FlateDecode >> stream xÚ¥WK“Ú8¾Ï¯ j¶*ò ;µ\v*sØ\rà6áà±VÅØD–¨Êß–Z~ȘÇT ÓþÔÝêÇ×âßÕÃÓ‹ãB+ôm´ÚŒˆíX¡=ò½…eËU2zó¸Ú•Õn2óloüÏ ×M•Ç‚ù¹7?ÂA1§)ÁòI ¤þ ô—Fó†ú)‹JO¢ÐïÑzõßh>šb…žö-ç¬üizÆ;Æ‹wýô.qS$‚íhãË€Rv¯Rq·Æ å®<.2¹žÚP,—v½óé…¸|Ô)5D›ù + =Óž »½=™0ÓÁ ÆçcßyÃqUzˆ·hCí÷™ô}®Œh%õÚÆiè4½¨l‡ŽcÏ¡‰ ¤YIÔÙ¾E<ûjp<(ãž²„f"2³»cªŽ¶Ù¨ƒu²ùdå“Jˆ©#–;Ë‚ ¹V9ûUQ•å×)Y×ÊtvÑÑxõ·r ËA¥ì÷#šæ‚Ÿt;€Õr°¾òäØK7¤‰ÊŽrV›&_2šoEÚÈ;nu`;*Ò"Y~Š‹¼Q.>M•ÕÀ±üÀ7+„W]Öéß,Õ9ÿ:)øÌÀס;£’› kİQ^hŽ8QÑŠršÜIQ7ƒÒ½é\z•A]ßÉlXÛuÚ0kã†ÑcQ‰Kx©µWº¹øCÑ<²ü¼Ø®{’ôjí²'<>\ÛÈØ^ᨮ?ÊÃN2”ko[5‹ ’í&âÓi‹ýòeý‡¤%§'Ñ%—ÐH¤½Jo¢]%ªòŒ›ö§l› ?ZUqQå]º€‹ö<€¦ã  «Bè:îøÛw¹:Ð{`)qú«bEa"–)PJž5 ÂÅr­Id)L ý;±~Ç `"y'ÐÆ~̉Ñƌæ;‹ì4 Ü1ÚõÂq”'ˆµ—_á#q× ]·Oµó€s´ _I)%Ž<€\J ¤• èEÍ äyùÝë¸¦ÞÆÅn_Õ6”gRº¸`Q¦ @ÅZê¤üT׉ß8V°ðêymËŠ9ŽE©al Þ =Ë&d‹yØ£=ì"£5Zþœûé¦8íi}W!¦{èZĽzW!žåºÍÀ~üeo›Œ²2Ngß9‹©åû‹¡×m'E3)‹Ó¶£´wª± žP®æº<ékPŒ¸Öt£Zkp¦þì)ëYêzÆ´¹8õ Ä £¨´pçm# ó pRTo™º+HÆÄpþ]ÜFw]i9ø–f5Ï¡~‰Xׂ¦§v_ÿ"Ü+‘™°2êPM.— ÿÆõ­Šò .߆2%(STŒìÔ¹Zé¶ó|(ÕÞ‰äD²/Q;›Ò@ÈyMƒNzCWé½4[Úíe\pÚü +õÙêñÅiwø ”8Ÿ»FxX<ï…ï^+Ì–’rzh¯Ë·kò}/ôx¡þ‡ Ô€šS9t¬¹CÙ­ p³¯þP<|]=ü”‰ endstream endobj 582 0 obj << /Type /Page /Contents 583 0 R /Resources 581 0 R /MediaBox [0 0 612 792] /Parent 578 0 R /Annots [ 580 0 R ] >> endobj 580 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 366.824 102.639 378.78] /A << /S /GoTo /D (agsurv1) >> >> endobj 584 0 obj << /D [582 0 R /XYZ 96.768 705.06 null] >> endobj 579 0 obj << /D [582 0 R /XYZ 101.643 370.311 null] >> endobj 581 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 587 0 obj << /Length 1054 /Filter /FlateDecode >> stream xÚ­WKs£8¾çW¸fsÀ#ƒxW†=lÕÎa¯ãÛ>ªÈF ˆ ;³¿~Z´ƒ#gsØ‹%Ô¯¯ê–ÛÝl¿Ñ*#YLãÕn¿¢^Bb/^ÅQB(PvÕêO§HÕ ûË‹¼—alaõ7ë¿w¬¼•ëû$‹®qÒ«¬@ô×nD#'Çå,Ø0qµÚ vdB*ïé¡ÿ‡ž—‡¢_èi Ùó%í‘ Ùj’8½Ñ¯Ôo¿úá,®nLIGÚdŒ&ȸH€K=’Q‰­.þE+_\\‡—ådû£üñÌЧ4%™/…êVô|x²zþA†;4-!à·*yÆÓ•k³ áx5P¯€Ž˜¦P+\¨ànâ( ¼`iŽïY3°sf~͵;˜œË½k+צxœèÇÃCßϹÚþ#8IJÂ(Z ›Ë…¨”¡;ˆõ[üqÖ…&ÐggÔ³3 À)ìÙ06réMÃ9e#7QïNêûÇìÊHÞ²\ý˜Ì AŽ@^𜟋ÓAJ&†®Ï…Zµ(8ZôU~Íßrl”Â̓¯:,à¦b…¬s\¬ñVÁÊ‹ççF9ê[ˆ:7JÊ®ÉM»ØP}ú Ãëe¡#ÿüîç9 MÌÆ­¢-SòÔÞrQÚ"¡ø¶k]ðé~•ÎÛЩ—š,oW³µPÏ)úÃØ2±¦)”’:òÙ!IÖ|ÀÝ~¥ä0"Ì00”˜ÄÇVý>¬]øe=rt{\GÁ¿F ÊI[µRçTórM§FcVØÔç*ý‰8¥ÀI†­MAa¥ìzPzT—î§¢·!VÄBâÊŠRµˆÂ¬ /ô3§b¢kõ¾•æQhÕ¦grìÅÚwàÛw&ª¯%ˆ¿ŸZ²vãÈÃLLbS0•'i«……sA’:e‡ù+¸À.*~¼‹Žë(vŠfdZf¯€I¢ƒ 'ѦGP= ú]ÄBà CãÊ1´ ^Dp^p­ð5Y vâMƒ§:Êx̾k\‰*6û‹aÊÒ”5xr¥ôg%]ëQÏX’&0âËç n\hѤ$©áã–™¯…†!Ç =’^ÌŒ_àú6cÅð:Rö¾‘ú“­ YXŸûNv3öY?v¼BNí… Ï7®†%0 LÃÃO|`éA¡'+õ=oI8Il1D1»¸As݃º§!½7·à­Å§öêËbjo‰·í‰ÅÍèã›Çy+4÷¶à¾²Š{À¡r¶z^б}`z®wºQ›î4=M±O˜á7h¶¶*û kæá.™pä{K:Åâ ¾·éÔ\ž•:Ý`3g¸bB‹_„Ù}þ|% ¾Â_ˆÄÔóþíÐH(ñÕƒˆf$MSd‹Åwóûîæ'É£Tš endstream endobj 586 0 obj << /Type /Page /Contents 587 0 R /Resources 585 0 R /MediaBox [0 0 612 792] /Parent 578 0 R >> endobj 588 0 obj << /D [586 0 R /XYZ 96.768 705.06 null] >> endobj 585 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 591 0 obj << /Length 1040 /Filter /FlateDecode >> stream xÚ¥VK“¢H¾÷¯ðˆíPP@¦ÛËÆNGì^6bÃ[Oo„­¥V«à ´½ûë'«²@KAæqzf~ùå— ¿Mï‚' R’Æ,L—* :ˆ£„0Ø™.ÏžZýˆEÞ×0 ³…œ•ëgõ‚K“I«'›Ýi> _¦ÂO)I#kIn ‰nÚ¤Æ&Üž¨8GÇ"±¬kK<âFi}"¸G“Y^â`†¯RÉŽŒkÞ2èØE~¾öùQƒ§>NåóÛËýAx ™åúÆ'ܤÝÐ=´DvÁÓª%2Ÿ…$e¬…Ðks,&4b½<±Kž^U1—óRå΋|[f A¢[Ä•, ‡µJ@es÷k?·Ì'…l'áζ<ª‡£Q—N”ŒyòSñCøV¡ö]æJISBÙ…PŠjg/jÌ}m&o6šÍ£ž¿šb@©à^w¤qBxBoEêSC¾Ò.˜#H×ZÁ«àÍK€Ú¶"7ÙµZ¶4 .5»Rû4M/rÖN0ë,`aa×1¸Ï¤7²¢‘Ýî?Î-¿çÖ¥da°Ÿ¬â°¯­\¿Dê…‹7—šF_Ý>hK7¶‚§ñ™‰”$”é”D¡ù;T§¡n¾á8“r¡û¹Þ7Õ¥åZâî—å¡ö0ÛïÆØ‡ÚÇdè žxd{ÀR•öÔaUíd¦Ï•…íx‚¤B¸±Ï¶óòð‚ü³Ð³k‚ÀMo2nª~¾}xJŸs/_ê·ðò×BÞ‡QìÄ®êÚÐüeêßJâ’íOún©v²8u  oýk˜ µ±,~¦ñ^àÚ¬´0Ö8P/–†2μ÷a,€ákXðeK`t–°°¸77éú"mzjVJ­œ¢ µí ù -ÔÀ¹¨¡æ¨=HµZÁ€NÂjJt½¤Kö°…~Ì}°>|ÐBÝT¹€mΰn´sDÁëŒöá¨yÒn®pJD“,(¹ƒú¨ó%PþPÏ•luá|ŽÔU[ý»Á‹®‚±ö°@¶Ð;Û# —7°ÔBOÛ×ážþš¥lû_?{à:á¯ÌÚ¨Ê}UWɸ°ždæM·q?ÿu_:³ü5¤I[[Nc^ǵ˜„Ÿ¡ó qÒ⨠ø`£¨2ìȳlU [w¸eW™*Õl«þ—®ºÃŽÿq8©ø.À´³×6„ÌЖÏ!$ŸŠæ‰´C"ú”þ·Mxc}—½X¯EjU÷Z÷4Ç$KH> endobj 592 0 obj << /D [590 0 R /XYZ 96.768 705.06 null] >> endobj 589 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 595 0 obj << /Length 967 /Filter /FlateDecode >> stream xÚÍWKsÚ0¾çWxÚ O¡‡%Û“péLsè5Üh2ãD&˜‡II™þú®,ÙXF@˜ifz²X­>}»û­$¾MnF÷ {ŠÞdæJ‘¸'x€(ãÞDzÓÎv·&ÝÇÉÑ}ØpeŸxXûüÄ“…q…Å_*&~Ó·;ä”wÈhOô¨_ÔdeêÏžjàx“ó¸ν!á(òkÊôÊCƘáŘßÙÆëDÛŠ$_oÆúk牊p«?”®9èüNW+=zîiØIôbniö’'ë$ëÂ\aæ6–OØ™Çâ\vIGi-°ka”Ÿo ©´¥Í›YËï½ËE'ÎÓ8{IeŒØ®GœI-ÐRb¥<‹9ÐKô¸ÍÓ¬r3Ò6!i™§ëdÛš_'q¦G@O4•é‡ì°ªÅÞU+Lx5Å<5½A1 tàŸ¹tA¡µ#ŸÁB'¡Fš›6ÍÎFapØêu»Ëß¹…ˆù´òKý¦€ê£l¬Õåc¶Úá+$xµ“‰–÷µßšqUê}“Jífx©yH³B{IQæñ=Ε€oÓ.å@äf÷¼2Û÷öDkD@¶àè±¶·KL´×ƒš¸Jå¶ü„UÝ ®åzµ${â$ ÐÁâP€™£B+?8f‘¬ßn]¥¨ó˜†Ø²ú.+š°ò=Ó£cÕ±ç«3åÊ‘Ú)OVÛÄÕ¹ª:ôl9ü‘±7½ª|V~u~w²¹¾ rxc·y?ýÀ©•Q¿¤ÀÒ[ŒÊÇáeñõ‚µÎáÐ_¿-À‘üwlç ŠN£ðl®¢± u/—V¦Žzëx 0„YhžE”×öÍ÷ÉÍ_åžH endstream endobj 594 0 obj << /Type /Page /Contents 595 0 R /Resources 593 0 R /MediaBox [0 0 612 792] /Parent 578 0 R >> endobj 596 0 obj << /D [594 0 R /XYZ 96.768 705.06 null] >> endobj 593 0 obj << /Font << /F30 111 0 R /F8 112 0 R /F35 119 0 R /F14 120 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 599 0 obj << /Length 1967 /Filter /FlateDecode >> stream xÚ¥XÝoä4¿¿bš=ºnì|ƒî'x Ò!éÆÛäšMVqÒnÿ{f<ãl’ Üxhcíñ|ÏÏûýý›»÷A´ÉD«xsØ(?±oâ( Vî‹ÍÞù!ïþxzúk»‹Tä}ý޾ú‘g3/«;ú Q!õíÇ»vI"¿¥O¯§·øWîŠo·Ýÿ|÷^†AüÍNJ‘E,Â#nÙì”/2¥þ}Iþ¯Sñdéî½Ê6t@¡<; ÃJD’ñÎP¨-îK½ÝIä:]Tûž&GÝ—- ¥WÏt¡a*‚4#^ÈÁîYøCøQ{íf.öíùT^sL|‘f‘Û|š}_µÍv§2 GA•zŰ׆HÏÛ(öò®j&ô[•x¯'˜$mR^{pçíÕº˜ž­ÇëеG:á‡v ·irä»u-¶»0ð²›]ö¹kòîq8êÏ÷À>B 9ãø f"Ì:×>Àfåé`þ+Û‘ˆ³ÀÙ†Ü^§ÍP÷t ªŠ´œ>û¼®i¡oW<“ï`Ä ÿÅ#6h¤Û$VtØÉØ2±š$RÒÆF¿yŸ_ó c‘#¿_Oè_óu›ÞíÅþ04°ŒìÈ螇¶£ÁKYíÑÌ%MÙÇ30¯ ³Ð¦‚%ôDÆO¢ö¥[wßüÁ°Û–ü gxY]y›]a¥²ævÎi_G‘FLìò`0qýO_†= ùLH¡–P¨­X[…"óÖî·5ö<+>L„¼Š›YªÌÃgbÅë²ׯj~=h‘\ߊ£1\ë±~·0P]5:ïf×¶Ýš:w.žš4"“®H¸£›w2‰Š?[Ä®2OëB^V̾uy ÊOI%#'}>]ßÒáXÚ°í¬ð DšŒ{ w•ï­°J…Lå„“ü¿“*Y|±ÉPQ,‹šÊêh»I4¯l± 6ñÿ­±Ç)þàÌ3®õóŽ>[Ü^wGHÁ0 QÌ0Œ°Ôàì˜÷]u&ÒKÕ—DmMh<$/49"ãÌ  ¤÷G)ySÌŽGPYëáØÑú°&½ÑOHAéè@Å›{î1þ¼»O{ÑJÀ F¡ ¥F ª4×¥!I4F܇Úýè Ž¦¦uªu§º†¢}ó®`Çu]Û™©3'‘0©¥ë…D™·‘\ä“B4•<ò¡oÊ‹äùLÊÐ+XZò'ŽŽ•1UóHË“öO²Ó\ ¥H£Ä1¯+é ^" ¬ÆÅ¢â"oÛ°}­+ìCºgrQÊmŒ^7ÿ@„É¢ùG¾‹h;$ ÀˆúQ„MRœÍ/ álÅ'Ú5J;Z^[ƒG˜WùØÑ;×-!ò["up´oâÂܹŠÂEªª¿ášŠ(Žç-´äº¸ïã“$Œ¯D~²ÒèûÌÝv—„<¬ó“Yi!Ö]å¬ûÇvËà„fÏXô¥=¹øžW+RÁ¥«Â¤Ó¹åöŠÄ‹9òR‚j`W¬]ŽÁãƒñu;éoiEøÏ¬VRkf„N T›½^ͨàÒW.Ýo÷‹¶³gµu7z§Ÿ£ÔU)ùÚ™”« ýH‡:Óù«BEé豺žARôvA,5æ¤.æv.s4>9e†ñê8™}â ®O¼þ[./ãQl#OF _93‡W®ä¶‚G8,e@ΟŒÛ| !j:ÓÓ¼Ï-8Qó÷Ùè@èÚ¡¯>g3ˆ¤9ä­Ošn°Öz£áYeòî•– tã‚®õì-3–X{±IÑñ}Ü)|Ïä—P/ˆ”sòKHÌp  Ùyy·xÇô«/Ê(¡úÔ³%›ÖKÛü5ÕøÒЮ7¸Ò$2ÊÍ×5yÊšô]ø @{g£|³†Pzfs~D¡‡M‚àýÊ`ãÝߣ*ç¥npš`^UÅ׆N¹hÄ1©dy•Lj»êÑ4œ˜¡{®¸ Å>Âì)æ6WTÌÛLR…c½Çʧ’¯ÁŠÙŠhf†ýý`u{Æ ˜CNïט{J9¿Ä¢lo‰­‡P÷‚f6ŠáË’ŽtO€Jó.Û`| „²z,k¾y j†œïÚçF»ý9K”ó^cK6ŠèଭíÅJx.”}}÷þ»_~ûq¥¥Ã0_5‹º>úʽäÅZ˜…@ˆø‹ŸcT³#xk7ÎЬ÷“Öì5­2v‚‘®Â6‹l]@‘°p ™AØB@—Z>7é‹]nˆ=i ÔÚ¦ˆ±ò<lÊ1÷_eJDÁøkáú[ª<Ô,Zü¾F QA©|ëPmÇšŠ}H¯ÃF…¤Ô íà ³}ü‰{Ÿt}Ñ)‡Oósez×IW9IåO-òkEy|Dÿ­|ʹòÄûÒ¥@Ú¼¹éÙ$J ™,°Ø`ÖŠ®L„’cü´˜Ü_W T$dº}€·÷î÷FèáXu¡×Ö†›$;ƒA«…_ ®±:×úY׿S™u]±W3I…¡ý)óKSiì øC×:I+ÔöùuëR= “xN ¤}´YŽqŒko~¼ó>3ο endstream endobj 598 0 obj << /Type /Page /Contents 599 0 R /Resources 597 0 R /MediaBox [0 0 612 792] /Parent 578 0 R >> endobj 600 0 obj << /D [598 0 R /XYZ 96.768 705.06 null] >> endobj 30 0 obj << /D [598 0 R /XYZ 97.768 581.12 null] >> endobj 597 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R /F30 111 0 R /F11 183 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 610 0 obj << /Length 1420 /Filter /FlateDecode >> stream xÚÛnÛ6ô=_á¦"1+êJ õ€mh€{0°î‘i‹, " °ß!%KŠœ%}xѹßùóæâã É’ Ylv‹,%iÂIœ’ Œ›íâo¯Xþ³ùõãM˜ àBÂÒxá#@ÓŠ­Ì5ÉëÇÆˆ&Iƒ¨ƒ–Ž\òil¾,"4J_§U<Ѫã8ÃŒù„²xbAcVû9…!/²€ŽãH”JÌð BŸ…D õ‚~ŸJJšx|]ÝÏ„$βïãd“kŽQD¢,œÆ#aÂfL71ÖÃrñ¯Š7iËfýdžîƒÿoàÀ§÷ç’…žX®BÊ<¥y«q[ïpÍëc«Üï© ÜÝr%sÜòv<ˆj0Ï¡æf_XRïnI=YíÉrE¾·)дB·RÜHG\ns¨—+¸ÞŠÏ o¡ÎSÔ@%L½­H=¡Py yE㘄ڱ âÈ+¸áб âØãø‹]·çZt*\Ú¾Wõi¨Dw:7*€4 ÂLbP%(M;·lLφ$£}Õ·P)£^î„– W'^g¬œ‘¯Œê£;X{TŤ)ñ'iQ·`ç(õòò¨ÀÊæ`ƒåÃØ÷xÓ´uÓJà‡»¥y+¸;¬ÂA8¸÷÷6’àg^£/Œ,±.Çè°j˜Nõ†üë¹ 3âCö¸ÿÆ),ònL wBiÃæ©±Fk¨„-¹¶ÖÛž$µ }˜eX86I+âוý43r'VîÛÒÝTBlnuë-2ÅúƄP˜QÛ,Z­qó°<Á¶âà¬cDÅ;éท%ž”„Îí¸nsT¢½RtvžÁÃAî Œ“™Y CpPeØk.+¼¨+AfJÁ‡Á/ÉÞVV_êèá[Ǿ„‚×èùÎûNV…h¥VϽ++Í•iÅ];^ÌUj¥ëÆÎE¿·òÀÛ'ÄvÕNãépÌÝ|y+Üÿnðê'OdÞ·þS[L1@^¿pãÖᜠ”$‡{#C<áÌ|Ñá!È× Ï:9¹Ù±šµuìNrß7Iƒ3Pe䄃„‰´B‰ üÐèÓÈÏ ÓѼ0>6Éy§KE*ðN+óWqtCþK³Œ„i:é-޼zêYáí»µ{h”uÝŽÄÀû­Í"JXÎÌôô#.¥¨öº˜ð‡Äü³t>Ñ^{îcúKuÏK¹ÅÛBÞç&ÕÕœàÿ?)ƃIÑò™±¡q7Ƚåúgѹ§ÎôÃ\L’ì\BóIÜíå½pY!Ýjkêã8ôÚmvõ±ÚžÁ>Ô0Æœ5æHÊäÆäÙ{dð—ÙJÜ›Ž[±ÈÓ÷ø¡>×cŸÀëYõZ‚øBwNÓd.ÞÞ»Âà†Gú¹ù¾L¥²cÄH¼¾iõ=–iðuõX:âuU>½$œá:°F;‰ÃS!]»ü½êžPWsd‡ñ{ÊR®µMùÍ p¯‘¼T¶ã|px¡,a /ŠxR–Fc|#‹ (™Œ!\’À‹Ï›‹ÿJŽƒ endstream endobj 609 0 obj << /Type /Page /Contents 610 0 R /Resources 608 0 R /MediaBox [0 0 612 792] /Parent 619 0 R /Annots [ 601 0 R 602 0 R 603 0 R 604 0 R 605 0 R 606 0 R 607 0 R ] >> endobj 601 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 582.017 191.928 593.973] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init2) >> >> endobj 602 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 570.062 207.619 582.017] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata2) >> >> endobj 603 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [227.904 546.152 233.771 558.107] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected2) >> >> endobj 604 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 510.286 223.31 522.242] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}simple2) >> >> endobj 605 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 498.331 218.08 510.286] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}terms2) >> >> endobj 606 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 474.421 202.389 486.376] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 607 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 378.78 102.639 390.735] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init1) >> >> endobj 611 0 obj << /D [609 0 R /XYZ 96.768 705.06 null] >> endobj 612 0 obj << /D [609 0 R /XYZ 123.92 585.504 null] >> endobj 613 0 obj << /D [609 0 R /XYZ 123.92 573.549 null] >> endobj 614 0 obj << /D [609 0 R /XYZ 144.841 549.639 null] >> endobj 615 0 obj << /D [609 0 R /XYZ 144.841 513.773 null] >> endobj 616 0 obj << /D [609 0 R /XYZ 144.841 501.818 null] >> endobj 617 0 obj << /D [609 0 R /XYZ 123.92 477.908 null] >> endobj 618 0 obj << /D [609 0 R /XYZ 101.643 382.266 null] >> endobj 608 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 629 0 obj << /Length 2235 /Filter /FlateDecode >> stream xÚXÝoã¸ß¿"›. óL}ëZ?\ÛŠ}h€;`w22m«+S)%NÿúÎp†²-+i¯O¢Èáp>~óAþåñß’ì¦Uç7›'¢Šoò¬1,<®o>G½ÞîYœE^Ðw0½2ÛV wÐu£Z÷e™-µÝ»"¸«ÛÁõÚÞñ¿yÿõñ¯7Ë›…”¢Ê˜·ßt¿°äÞ=ýK×ýw=.}^àé4þŠ ~ø$Ó3y'ì¶Dr¡Ò"^‚Jò’P·NÓi,Á¹zç¡Ñ.RÒæfCt¨uÿz`FW¬»>‚Uz½¾óJŸ‰/OYëV÷ZXíqz4á[¦šˆ<áöx%,íÚ)'\oU¯.é?6N˜¡mñXÕ÷öʃÁµ^‹ïˆÅ[¢á!Ýfãtÿ;¡Mþˆ÷x¿èf»ë'^Ræùµ×ìµZùóØð«àfqwuLNÇ%TÝ7ƒÓëY<Œ$sbšË-­6Û~w&¸¹Y*rÆOç˜Ú7Î5f‹C«7ÚjSë“JßO_DÞjuëQz{Zw\ tëÔþÐêÛ÷‘õÆæ½êëPvF$oåY¸U¢ ð¯D–²mwÀ/‘edô±§Q¯Ü7uüQˆl7ôáŸÆ1}Ç‹º†`±CÝÓ.£kÈ‘vNÙWšZ{¬#1Ø^›û˜»pˆb^îc °A°Ž§Âùg܃(À=ŠÂj!SQ¥é¥åð $‰Êâþ$%m`µ¹NEY)Ò¬ï÷ÖÝñ°»¶_\‰´HÑ—¥L{q¿HÓ4úõ¾L"Mü_š¶¥ö‚ò™~°éÚ¶C^s/!)zºÃ4Úu6u?ÎdÔEƒžÕ¥AŒbÎÕU>Jºé,p/€-ò~¥£ôêŸ q—‘O‹ø[Dc@Ð.òI%ÏkÇSm¶FµôãûÜ<ßgyf^Ñ4h¦8´5̳åyˆ~-Яexƒmü·îp}A'=泃ÖÍË“WLšc ߨ¡E×çiä:šd×ÀLgÚWF¨þšþ<˜áûD6¢mÝ×Èz§™ åÃr+Lì ƒµ|òà=8¸AñYµr:Ð+–H1­Ó–Ed¯Ô+8ç¡e_WŸ~úÛ?~¾v}ò ¾÷¢&U@à™»p32ú&—g.ÁwÞb@ <2¨¢P²pLÜ >xŽN3Ç_¬zz‚dɤ=‘Â?¨Æ,ùeï;2è$Š=(;ûþ”£ÕíÙ‘IôÛ}IXÕ>ÙÀL;pÀ…}iú]UÅG±ß`ŽÔÃÖï^ƒª3aJÁ‡QZ$¿3J“°aŽËÔ{G£ˆ0§a¤™ØA»·VvÍ+ÖvKxRåçƒ(Ño¼Κd_±îxiÇ1 B†T•çX+¥(ò*¨ðÛµŠ‹@ráU¨Z¶9>pTúЇ¢Ìdz'q1\ *˜UQ†L /Xý–ѯ;È#>~ÑCd ŠSÿùF@€]dÚŒ“ÎècM0Ù&ãJ_E«c©ñ™Q;­„”É;jËXäË<º òì!'ö)–`gîz¤‡¹R•‹¢,onbŽ×2¤™yè m©õâ¡V%cÈ'¾£ÀŒ¼{¯\Ê$ Œ [ý ­Ê|2J’S]Ö¹Ùø‰ËäÿŠ ±ÊýBúqÃAZ1â0ž7´|ôXÇò<kt!‰„ÌŽØc郀*È»1t"Ëâ²BÂÌê(¡w©1gyð^v?¦ µ_å©ýRT•º8y~ì• [ë5ê˦`0EƒÁéÓk·ÆE.–ñÏ-4{ÊŠƒÕë¦î!³Ì8 ËòЗ—øô^¿<èƒÐQeiÉiŽŠb3aõÔÎq™‰ä$è'޳[)‹S(`táô†G˜ïÑvE½ìš1rðŽúæÛN&6ªvß¹žÆì¹¼¼Öl¬Äïià-Mh`“LL¬¦Üwæ¼þ¤ ÚRɹ™ky½Í²¤¸Ì‚4Å MëÁóBs=fOÄþKKþ–òèL®V†ÖÑJY’û ÿO)²kÖ4Eц#ç+­'ß…H(e¶,ä`ÖXĨ,Ím´îÅÛ=ýOÐX”ᄆäIi™GOÊßq-‡_ G>p‘ê‹LÒã­¿ær{F;mÎXj[ºÃt†ÚUõï'YïãË£6[?l˜ªa™bÝ`.Áê| ’Içšëøü’.¹oÎí}¶R†]·×uYŸxMýŽW‚cd€AO(@LkC+#ñ>`µ¥¶6Òxk]p†[£å›·§äÏÜàÇ wYÑß7¤Tjdpѳè|ÿÜtƒ£IjÁpÔð áɯù{ÔsøŽ]aC7 ú‘œ÷­E;“K¸ô(Bš€Xûq|\YníaþV Å .Mã&Žèt™‹I+Ûïºa‹¢JÆ…ÌNwÝkirè1rù¾0þ àôùD•™ùx5gäeEYeá>;<ò}¤ñ™×7ƒaO{É}ë€#«ûÁšÙç½x)Êtò¾÷¬¬[½!ò¿7ú>tÕ˜^oµšZ¾¡PYЏ=ö'¼f2:“*ñе>=ÃÛQ‚sô«x6Üf`ì/ðEPcf ažT1—û*91wøÞ±ædÅrLü­ú±‡²ÜÝ5|1ýñ¢ÒóÅêýeqÇÛíiuloèõ34C 柽ˆà3–)dQz®cåçe‘óü»Ç®I>a\3, 'gcGÙÌ…ð Vdœ¢iOŸŸµØo.ß²Æ í4Î¿ê®æ_u¯Å–qq=â~3Uȳб“;öÜcãçÏw¯w_¿þ79ÏÞW_ÏQ'²AöÈËì=Ñd.Ÿ¦þ@'> endobj 625 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 175.542 102.639 187.497] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata1) >> >> endobj 626 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [175.601 175.542 181.468 187.497] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata3) >> >> endobj 630 0 obj << /D [628 0 R /XYZ 96.768 705.06 null] >> endobj 620 0 obj << /D [628 0 R /XYZ 101.643 179.029 null] >> endobj 627 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 634 0 obj << /Length 938 /Filter /FlateDecode >> stream xÚµW[oÚ0~ﯠt*­TÜ8$!‘ÆÃ¦µÒ¦=mì‰ñà‚Þ’Ù¦Ði?~vì —­{ŠãËwÎùÎÍ~?¾¸øDtÆqz=Ø ü!påÊxޙܼÜö}׿yÛ¿íCÇwnR:Ç À[ÁÐL|—3i|§·ôæ+šqÜ“Óðv:þÔq:}Aä¬%zÆ KDõùö`°¯bÊô¤Xb=Hè‚ÌôéOŒ7f‰d˜ëáNèÆ¨ž)™ëᜮŸ´¡ì§Òéþz›.rµû®"Z–Œí6ºA¨7c "‡c¡ÿÿ6¶½¬ðhÔ˜¥¼W_RG.SÂ9ÉjœáÍ ”ó™o¸¾®btñv…gÏ»å ¥L°Y —ˆ.}'PP­1c†³™BÖ{긭’ 7¸8!ð#Iq.=¶ñg¶ÔÔ\KÒ¶–±T•jÂA¶N5¤O?¤r“IoÛ›N÷MPl0Y,ß_ q̱¨sè»§s¹§É½« ª….׉€ ùr¡<Ï«ër¥Å~ÔŸ ãy#‰(# ’¡Dÿålr‰¡׈ ö‰½4i[ÙLa]r°Òè¶"P…ÎÝèR‚_ŽŒ;€<–üsAWêd÷ÃŽ~b¼™QѰŸ++õˆü2#dvÙ¹ÙMdÙ“éÜÝiÖ·©¶µ‘“"ÁȶdÇ+•Ý#cý®Êhk‰Ÿp‚08> †×.ãòXèF’y>]7´vÀT8áØ÷*½àA½}ÙšþÇéª. ”- ¸,M%\Y1VU¶hPZµÞ)l'8[ˆe^•¤7ψ)<8å!'“òÐtj•SqcǺLÉ]͘—”‰ɾ7Rµ±%"dW^­-nÁ t·8›(Îpž–Îç»f;¦KjœZý­§ŠYªÔ…Ó+y¶â&«,£(ŽMÓª¨jG¨kªWZ ·¨â¨Ñ3¿ÿ’òŸB”csì©[صd {4áÁ:Wkºg³¿ª*þ+DdÕS{a’»a'w'ô‡çºí¿xôÜšZ¿GlhÖÍ+EõºS6Ê¢óÎPá LdûdÇòá´–æŸÝÒþ¥9˜_ØæìüŒÛÈQ=9éëBßì§öPýÆXœ|÷ùëîÄpŠH&_&~éZÈ—Ná‰gÌJ§Þ-Kôç+V‹3óéþ1¬>&‡ª«‘0 5b©ã‹? /Á¥ endstream endobj 633 0 obj << /Type /Page /Contents 634 0 R /Resources 632 0 R /MediaBox [0 0 612 792] /Parent 619 0 R >> endobj 635 0 obj << /D [633 0 R /XYZ 96.768 705.06 null] >> endobj 632 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 639 0 obj << /Length 1526 /Filter /FlateDecode >> stream xÚXÍnã6¾ïSx³ÁFb­)Y’UÔ‡¶Ø-{h½h7F¦lµ’(tœ¼}g8”lÉJÒö`“šÍï7CÚ?®ß}º £Iê§qOÖù„¡Ÿ“8JüëídãÉ<×ÂLgQyßÏhOï׿Læ“c~9ÁƒþS_îî‡/¿}FÙOwlqfgÌÁë¿¿#Áå™\ê',ibK’ú*§ÁÒ;Nga{;Åh·å†Ó.W²Â]䙽 R-ŽC)ÉYàý>].<'xœ‰Çk´`ˆb$é‚àˆPØ[Qú¹â•£6ŠÈ™Ðº¨w·@™WÔÄ•j+T§S7c ?],ú ȤR"3å3¨JO<5¼ÞÒ>癑Jãó:ª>dèê"Y„ # 1”ÙÊúÆT-„“?¢´(v{ F/çÌÛ»Œ"ïÑ~+Ë™Ÿ i³r]Š‚)µâÛ·’?ÌDAr²ò°ÅüÅ1JW-™V N’ÊSù\ásomÃ8hâm|ááP”ÛN ñ3^–D:4…„_ÏÎCŒBÓ3„/úÞe‘òÔìÇb,ŒeŽ9dè *csOÖXØ1Ò$ ±Èà@¤¢@åjw¨eÚV8ð'iöÜé5BU: jÕEŒ½}xDGQrAYR$ý8@rUð‡RèNmÝ )¦\® U슚—ôÔ6Ô¡ æ•ߟiâÕ’øPam¨àM¸)rZuªA2ï=èYÀoUC?÷Z«T M¶ èÃþR †—#ÙØè+ÞÌ»©nwP¬Þlo44£]Ô¹<àeÐŽÍz0ÆN7·v'\ÍÆÜy+·ôãi$pjª-Èf»â±òN5žÕodˆ({O3µàáÓXãÑ¢›ÏÁfv¦µÅˆmIÔ'~š.úÌdmÔÊÍûÀµ±mngìv«d³²¿× ¿³åš1~%ÓÿÊÏ¡gcZÿ›gÃÑëšÿÍÑ; bŸEcйTêD_(‘„ìÿÕn“þ_,ŠüåÜ ñ^’æÄþáòîóúÝ?³U endstream endobj 638 0 obj << /Type /Page /Contents 639 0 R /Resources 637 0 R /MediaBox [0 0 612 792] /Parent 619 0 R /Annots [ 636 0 R ] >> endobj 636 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 514.271 102.639 526.227] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata2) >> >> endobj 640 0 obj << /D [638 0 R /XYZ 96.768 705.06 null] >> endobj 631 0 obj << /D [638 0 R /XYZ 101.643 517.758 null] >> endobj 637 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 646 0 obj << /Length 1936 /Filter /FlateDecode >> stream xÚ­ËŽÛ6ðž¯Ø$V ¬ëeI@÷Ð Š"hY ‡4®D¯Ù•DW¤ÖvÑï ghKk9-Š “óæ|÷ñÓû/7_îºZEXF««E…eÆ2:¹3ÀgoYœß.è¿—[³\~¼¥}×ëÚuŒâñ‡"ß}ˆÒ‘‘Ë©èG"™œc/Ã2Žà?[Ž&èõÚÈ6´º–MH˜‘nw¯iEÔšXL™°š—GÁG›é0óú–ÉÙ¹d{ØÊ»;Ú_Ëý¼.ëë ¾ˆà qY€<'g=狲 “¹â5Ç“žÓ¹ÀÅSmÆj—'o~–;b¬…´Úò£ÖkÙËŽ½n†þY=‹†v¨ŽWÑù•9àÍ?/æìújæ§¡¨¬Ò]8YOýî}Þö^9’_ÿ0&”áÃvñÔŠŽD# e˜G Èþ¨ þ_7¼–FE°»‰‹@⺠jM°N[^H<$®Œ]-zÞɾ׽¡µs;rWºÝVàq £×L½ßÞ,P æ?6âÏ£0Å‚žoâ$X¡j·¤>¥a™¦/RIu(Nb¶ æVôVu¢aÔ€ªÌÄ¢„ÖPQÚÝDxì(¨ + µƒ!Ø‚@dNö‡`^š÷^çò]tÎèUÞ}fŸé²2× ´â¹­t3´àÐ$IС”BÑ(ºé*Œ¡Isß8œ‡? ã4öøp®$²(L³ô,f½NÅÕˆ0 ‹<ó’¶•Þo7 ß×fäË0*RO¯fr{"ðއNæËär¿n•Ø?r£ÂþqÞv^T˶Y«‡Ï·xÖµ0dýB`Æò„ôécȰ9ÉcÃ^Ï[FýÅmL†ke/Í€<WËh2@„gULõ^ŠNz!:ñ\x²0²IxÀûeålWçâã¥OŽ’«‰¸XAîÝ* e'{,@WÂ0¸O®Ø mÃŒØ1`½¤Ïâÿ÷z€úf”Õ¬DS ° 5Xn”N› ºsÃä Îé¦íZ÷Ä'E…eºùJÇÁk€úÏVמœlJò˜êþ¹Å`Ö!êL FVÔ&««ºZ=û~û4´'"ðft¤ ³@o1ΪØ[ç°ƒ¶¤³Í¥ÑF6ì.5aÉÀhµØî2\€Ž Ý/ÞÿìÐwÌÇ6$£—Ç|µÐúΪVºXÝÊhú÷c ×4…œ¤í¶?z%q90sªGi©_ üKÝ`j¤•7˽kƒWsQþÅ..av ýàú·Òƒqö-ƒÌ)K•èåzhxã\§UÅì4J—¦øïr|4­ÊàÔ^ÜÖe¢ãØ0%¶-E3áG·VuÆJÁ8½¾œ·Çbršçs\äÁMµFc=ìà€q™;¸HÅ\”<½!û¡»1¬ã"ô²u¯UGE–ó­ápkÆï7ÊÌÿ "®ðšúÇ Ýsy¿î]ɤå(PIó½&É–H­Ô.–@§<=á:¼}âžìÄ•»ÝãÂ]C‘Hîáêæò;õ1A:ÑJÂcæ²TCÿ¤Px€ %üLªrúcn;“ÌWI6r)îØÔGˆ½T6)wØNËâ¡T¥¬¯Xׂ‹ÀkQí ý¿EIŠ·’7lÀ©hÏLEïÍœÎù¡€¾§\»UKͪ(¹§Ð–“‡=\(ÑÀ"r¡#Q´Àøî\_s[ÅÈ$ o±©¨ªµn˜æhŽW¼¥žÿ„•KÎP'•îÐîw˜¹]S‹Þ¸-®,C»îŸ¸Ïv<?’i·N\ø _áê2)ÒSãÉS÷æyÂÎ.œÄÌ£¬Ýd†´^ #EÓCaµòÃʆsgý‘¬9^eiðôh]%–E‘{Aø¬Šâ •ø8C΂ ú«ŒÈx(ˆÙEgv¾õà~§š†¹zå(,ñÎ…È2ê€T¿°à›‘ÜŒœ{–ã =ú(r³}´-P–•ùN–EiðéøB¸ áz#ÐzßG©ÎòcãÀög‰~ƒýàr€ËúADÏœ˜W éÕéÐ]MèØqúlÊF­x¦wžÎê&#–‡Xwm¨iPrÈ`å¯XøîÅçO|bšÓ˜> ÃåX¯ ‘Ha”46œýºGiþßžÿòR;óæH_¾9ò°L^¼9ºïMß C§`Rá _7µ#¸ôaòõiÊ f;Fz\4p‹}È32è’ ýTÊ~òRáïn,ËÍð·¸Ïnîüg¸ýéƒVÛà­ÀøŠ¿­@ÿmO\´âwç?|2úÊcÌѼ¥÷——ÒêÁ¼éyñÕæ¿pø“×zxhœï»£¯.9OxÂSêŶbòü‚wpƒÐ2, þR’;E¯Þß¿úEé›C endstream endobj 645 0 obj << /Type /Page /Contents 646 0 R /Resources 644 0 R /MediaBox [0 0 612 792] /Parent 619 0 R /Annots [ 641 0 R 642 0 R 643 0 R ] >> endobj 641 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 432.578 102.639 444.533] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected1) >> >> endobj 642 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 384.757 218.08 396.712] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected22) >> >> endobj 643 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 187.497 102.639 199.452] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected21) >> >> endobj 647 0 obj << /D [645 0 R /XYZ 96.768 705.06 null] >> endobj 621 0 obj << /D [645 0 R /XYZ 101.643 436.065 null] >> endobj 648 0 obj << /D [645 0 R /XYZ 123.92 388.244 null] >> endobj 649 0 obj << /D [645 0 R /XYZ 101.643 190.984 null] >> endobj 644 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 652 0 obj << /Length 897 /Filter /FlateDecode >> stream xÚµWKsÚ0¾çW0“dò-øÕ©/IfÚcKO)cËØ)ØTô×W²$@ÆÆäÑclKßî~»ßjýetvÿ8°:òl°;£¸ƒa€<èØ–ƒ€¿E§ë%ÑMÏëúsO^ bÞGùr2#¿úV?£ùZ\ç1ð ¿›ñè[§ßéaŒ©ÿ@oUËmu9q¼¬ïí“çGôŒJo;g ¡jLПYÔ÷Ÿ\êèYò1¢28޾ÿ|¨=‰8-Ð0cr;¯™1ß<‰p;²ÅA94M5m¯ÜþáÓHËÒÐyÄ盌¢R–½B}+éÅe!¸û§e!·Ï«<亮„wJí=ŒÎþ_¯d endstream endobj 651 0 obj << /Type /Page /Contents 652 0 R /Resources 650 0 R /MediaBox [0 0 612 792] /Parent 619 0 R >> endobj 653 0 obj << /D [651 0 R /XYZ 96.768 705.06 null] >> endobj 650 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 656 0 obj << /Length 1532 /Filter /FlateDecode >> stream xÚ½XKÛ6¾çWA‚ȉ¥ˆ©GQ_ZtôT {(ì+Ñk%²äRôzÝ_ß’zÙÚÍ éa!’3œ÷|Cïo×ÏÞ^E|‘YLãÅõfA RF1O ”ëbñÁ+7KŸSî} yØÊ`SjX‘åÍõŸo¯ÝŽiÄ|ÚkË1‘ï3ò£…OHq'ÿ^¨­Õð«o¿9ª Wnsص‡žPý¹Å¿hþ}øDÞÔj'Rn…šJU·e]Œ%yÈ×KZ­ Õì×W73ÞÉ¿›s†AFé”q¾:o*üžPãzM dq?Hkæ4Åß“(ö͉*ô4 ¨'‡dØÝkû©åñáú®nŒzsên˜¤Ì$«•öÊÍTAûB¡BÞ¸T6Ç÷PHÿΰ—¯_ÚEsûIæ¦V^º«šÎ`´d.eàƒ*ÛÏM—œM´©G"ßZ,.…²jåŒ>„Ó'Æ¡a§J?Ñi4Å~¯š‡¾‘t¹“®È/xÔ«>…'—Bz³2^±$“3ù;©·M±~•7u«E­_¹ë›u×`§Jnô°SåÝV¯­*LË“‘MxggòzÔ|ÿÜ~³œ€KîÄŸña&DÎ…ç ÏÝuu¨äšþŒ¬õyÄž07¿ÀkšEzŽW.ó{%‹Ù&躌N:)wÏ{ï¬nQ ïÿ‚بôç`9ý10?ù°,î$Ó'øµ²éLþîÏDâý2Ò¯ƒÆ~êž\ªì5Pñ#`ü=³€ú÷¦\\ŸêðË_’¾ØÓqc†)ðI-ßÕ2¼$QD=½•€òn©¤[6{] á&òŽ[¡íqÙÚ¯’ú jYŒO#O¸ÎŒÆ¸, !]óª%'0W„.ï奵xØO¡2G;¬šã¶Ì—4õ¶V[ÞìöBIg‘ Ì¿m¥ò]¤`\¤<9+¤%=ád‡ÌÓûn1!úBà‘R‰;i©4 :6æB {v€g:Xú,bÞ_xS(]:¹&H°*—Äs˽à^.*s‰{ïÞ:é*Í4Èe'ÈäÂXh”z'k¼ªçJ8o,©RhDãØ; [Ü$ÈãöÀŒ"±;×0ŒŒdÕTŽT–FÖZ/ý,@W(õÞm,· rï: ÕH½TŽªäÇ0mI‡¶¬ïœ2Kx²ûÎNÇxÍfÎåhj6F‹­] Ð\K[!%Rµt¨tk7 xZ¿›0>ØÍØÜ±ZšCUØí­¡º‹ò²lë ĵbç2z,õ¶7`ooæ.ÇP–3 î¢€1‹óÒ¾]Ù¶°• ‰†’qGZ¶=YêSĘw‘ØÚ "E™ê;ÚMUÖR(»v}Ø(idüä€CiÖµìßK?I=Ìåes“4àl« ý¸ÇèI‡GãPÂÖvµ¨ï:gÊ:‡Dš±ígŸ¦ÿ¶k#hD1ŽJuöÔdŽ£—bˆÙ¿6æ}pÜ»ÒÅsë„+‹oKW_ˆ`J‚‹ƒ5”rJi÷ AW–D)²² cˆc˘{BÍ™¦l1âYÏ(%!ä"²v©+ (gsº&rÀf3bÂ5ý D B†—vã*XxHæ‚HÞ›½®gŠ ÊqÈÌè=9â!c,æûσˆq36)ÉÜO˜ù@&!D“ÈžŽf4‰æœ0,æÄGð±“aÒ¿_|RèHxDЩt,+Dª pG÷C&ëà(³puÓ®ŽñLŠv-gf%¬íd‚EѶ‚Ì|ƒU÷uƒj¬d€Ðì BG÷D§pÒDxo«víÇ’XÃD€Fe×`E+uµ=·mSp ^#Ô,1+œanÒY!hÆ¡µ‚ú¡WÁBàBâ´ f/ /'[(GÝÅ.šVÛã\*-ʺ:Íù×=¿ºHÄZéÀ¹ãx Y¨?—5ˆ¿ëp.pÿW A èG1t¬“›ô{öÇõ³ÿi­è endstream endobj 655 0 obj << /Type /Page /Contents 656 0 R /Resources 654 0 R /MediaBox [0 0 612 792] /Parent 659 0 R >> endobj 657 0 obj << /D [655 0 R /XYZ 96.768 705.06 null] >> endobj 658 0 obj << /D [655 0 R /XYZ 273.931 216.887 null] >> endobj 654 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F33 118 0 R /F11 183 0 R /F10 255 0 R /F1 256 0 R >> /ProcSet [ /PDF /Text ] >> endobj 663 0 obj << /Length 2146 /Filter /FlateDecode >> stream xÚÕXKo䯾ëWN`QÎL/Ù| ë`Y >HV» ¤EöˆŒù˜åC þñ©êªæ#JVœØ¬®®çWÕEþxsñîýÎIEËØ¹Ù;¾/EâK'Ž!ƒÈ¹Éî/…n®¶A´s‡BÓâ“ç‡-®äÎU=½ìÛŽŠýЩ¡DnåÏ-òé¥n¯¶ð–ëʾw¬ SvU²èFë\çâjù;÷ýÕ.p­®²éÕd̯ú~¬µµW WŸo~r¶~(Ò0„§/Òˆý2–gè{n¡rZ€á¡—Z‘b„˜Õä*¾,]CÊ-n?Òùäúè&¾<”CA,™ñþþ*Š\Õ•ƒ4š÷î½ïÏ’ ÃH$±ïxdæ‘XæiŠDœ†v¢eêª&'mŠíÓYÛ°e¹&êÀ6]ÉÄÕ»ŸulÉ)V2òEAZ«Ð”i“,-€¢kUé~Å—4~¾âJ*¤7¹J¹–n§•Ê2ÒÆõ¹ð8Ék²w ¤Ýÿ@-ˆÜäyî ÙÌ)ω†ôc5°UeUY‹¶|› %ª¹³/"DøKvÂ’eL“·8ÀlÂ0´ì+edSÙL‰JÙx#NZzvmea’NwîaQã“´±/`ÝÕ=£ÑL3æýBu``6|G—1øa€‹”!Î1Ö—æ’fÓ7Ê|Ö=%h’¥<ºÄó™fËÈŽ$‡¬=Š-Ë.5NEr`J‰›® s‰2”ËÒÇÚ‰ c8Ë—½hƪÂe{‹SγVï÷eVêfèq&¶s±#=ðÆ?¿µõþûmÈ]™¡obß®ñÇȚo2~4}µíW|ÛFBKY /þNßvô¢šGZüü=ÑÀž–”L\yzb±â³ÚÃòbÈp,Ø~ázú&ZŸc4‡÷û-ËÚ£×6Øê•Po–æ½Àµ¿³Þ~·@›ˆÔ„® h¸lô~àÏ“¦Ï÷#áÅÉ« KCá‡ñY/ØïMSŸ‡gNcÎ &LfkBÔY¤çÖÃå#윉´o¿¥g§÷ºÓM¦¯¯¿¡ýo^pIz—]ðªKþ.·å* a dª¹doÆÞ¢²e3è;ݯªÜ˜b á}‹^‹¾\7=&>ÅÔ~íNSü(1ß È´W¤f}Ï×çy)ŒM§Ï0Þ˜®j ®ï°ç–ÛrÂ+`%Võ]yÏ:üÜDlˆµÂþY[!˜ö~ Ž®}€™£vüîgÕ¡ß,ãQ|gZω÷eΗ°?.5ˆ$Û>ÖjÈŠy7 jºÖ½±ÑpY=›Ï¯Tjô†J}Ö8€@…€0q<áè÷¼€TϺ1–cß ÅuÎ7^©üÿ¤Óü·ÿZ×ÿßÉÿþO”J¨&K¿ªâ¯.„7ØüFÀã¹7`>úúÛux<r.«Ã%Ñ~ûç ¢ã@zùÒ]'"HÎkdñkË2 p ˆ- íÖ|«‚ endstream endobj 662 0 obj << /Type /Page /Contents 663 0 R /Resources 661 0 R /MediaBox [0 0 612 792] /Parent 659 0 R /Annots [ 660 0 R ] >> endobj 660 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 454.496 102.639 466.451] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}simple1) >> >> endobj 664 0 obj << /D [662 0 R /XYZ 96.768 705.06 null] >> endobj 622 0 obj << /D [662 0 R /XYZ 101.643 457.983 null] >> endobj 661 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F14 120 0 R /F13 311 0 R /F33 118 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 668 0 obj << /Length 1969 /Filter /FlateDecode >> stream xÚX_Û8律ÃÝ¢N/q-ÿ÷]ç¡ ì=`»@;Ø-0;G™èƱ³’Ò™ùöGŠ´c'Î{/±DRùI1úñöÍû›$»ªÂ*ó«ÛÍ•ˆ“°Š¯ò¬c`Ü®¯î½Y¬²8 ~²è`UøA¤½Qk}XÑwmº= ¶êù…(?¼ûu§6Ç¥ÿ¤O·ÙXå÷·ÿ¹Š®VB„UÆÛªÆªK»tÿUµûG£[%Mˆ|]»ÎØ9=cóÁúvG#úz½öƒÜÈtÏ_;‹Ã ΰߥ9j{‡r8õ$6疸׽º¾~k´}zÛ‹¿¿éè,ª4i:üâ ILNkU$agìoN‚çh©$È[öƒÄówSÌõ+³àïÌu²]KÃÝÊצ3½£•¥±ÜïM÷2ƒÁ‰™ÞöUAŠÖû›r ð㨄/Ðb–»Ý‚ùi&·ˆËàu¿XÁG];ev–FY½>Ȇ§ÒøQ iþ  vaUϨª3OábUTEð©%ò×}s°´Öm5««;¿ãš—BB­Ùž-ïó# "ÐÍ:$»¼×" «4z½9´µÓ]»„¥²Õ5Ú²]¬’²@»Š@Zš(kU‹L§eÓ¼"±¤M‘;2 §ÓíhÄÉ6;Z¡^œ‘µC«‘Ý h›Q©U{i¤SDìÍ '±ÇûW¦X ð›?K38ä81|Þ°ÒŸÅJIúÐBå?OSµä*øXÖÛÅ:Úõ‰èšÁêΣÇx`n€§ð¿£ J6ÔL¾µ 1Éò@ï(;yâéE°Ö¿G"QFµµ²V™Hƒ´¤é ¼c KÖáTÑïÓ“7@íuûÈ{µ$ÏbnËdK_×»f¢JÓ<,À! _‰Áv[«EüÏç\eð8”ךC÷LRLµCãÀ0<…/Iü‰ÂÇ"˜ÚÇC# $Äôë¢L€%Ä}B£äƒ´ý~¸éÌå ,”¤Ùû <ð­q(ŠbZ"EÁ'D6zd«`‡F p2ª!æÖG< dK_M®)S«½["-G ̵Ï|cÕ0¨»æ°c™’­¥)¯€ÖmN|c“¤3ú ŽÓà·E™ê-Å\ï^‘LB6FÉõ+¡°îZE¸`i ¹Âð"„„“3fHnpô \(rHÐ$GwD H/‰ý¨qÉ 4Ú2É;”š°õ1‚POã—JG’$Øt>ªÛ‰ÚóoÖ •މ[ͺüîçwé `Îß)" Ó\ô—2ãw¼¨Ä4–pP5«b¤+‹Â,*z]ßÎ7«Â(€åŒ†<Œâ¤ðÒ9p1ô"Xáε€Ùq5Èü:ã4Ü£y~Üʰ?K¸¹¿/2(œFK(TD¡ÀäÊ s@ñp@xó çÃðˆ‹>Câ>>bŒlëˆ8ôv4ÕvÆþ¸ŒÃÚ†ÈUQ^@c•Y˜ûž`tÐÏÚâBwò‘;EŸg3»VY˜‰éž¿BpÆI’"󩊤ô6ÂQUÅUæ‚7έƒ¬ › a•‚kXsl§p¦ŒÁ¦×}›‚7Èþqˆ†oGSŸwEy”]kùص²Ø\Rˆžxz]÷™”tkTE†r”Â}y;ˆÔ¾8Ÿó–&u·Û …Xr¥·=¹ùŠ}¯¶Cóýs ºŸè¤^î|ƒ‰>¸"ª“çÙ‹°Li·Öc{Ž’†I1$$ZI¨µj¸›a/^äûAœBíÞË~KÔ<…ÂÄ÷PD}‚•;ÅìÔzà/\\Ú’\ß±DØtr+Ò°Z„HDUÆâ$´±Úa."ÔYË}éM+jmÊ“›ÍÏ('ýÞ0–ôñvг×D »†3æãÖ¦uGVÊMUÕû#ÝâE-û¬‚(Η Ôý—×·ÝìßDa ¡·¿ïjÓÜ7ÕŠeß ë!ç‹éd=‚á u‹'moÊÍпÐÔ¨‡´ç¼7»Vô7øÄÓÈ@ÐàìL`&Y —@rŒÌ¹¸Œ1.‡ú6Ô(h¼ø‚\KßÍŒùt’‚öIï¹òsOŽ#WSÉ¢G¿œ—gCxáŸÚ¨oز¯ùØW°1ë½Ø×äÌjøs¢.‡ÊS °è9äÆÚ® ‡¼#_zG˜ùÿí·¿ô\ÄPâãòOÿ‹(…û¤8I8Ź‹cá9ûCà-ù°jTûè¶>©AÉ¥¿ëVÃMµP:¯IŲ5Ýóõç/¿üÖ¿\œ¾HŒ­õ®…t´Ó?ðc˹€à“È Ùk]’È@ïM>Û-ÿë¯ þãü eÓ? ÍçÊ`‹—Ž“êÏŽ³„»·ÌÏÚÀ‰eèåݾ¿ŸôTع»¿AÝJ´—£Â6{\gG{ïçðz9Õ†Àß-[Û®oî/EÈÝRßO× Oß±F&¡Eá‰%ý{ÌIÚYu¾ý…絋V}—¬Y2DüêÒã;9äÕœ-sG=÷ˆtþvF8Rf~Ò%yX@;¾Š«°,KÒXd(øæ§Û7ÿ’:‰h endstream endobj 667 0 obj << /Type /Page /Contents 668 0 R /Resources 666 0 R /MediaBox [0 0 612 792] /Parent 659 0 R /Annots [ 665 0 R ] >> endobj 665 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 307.049 102.639 319.004] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}terms1) >> >> endobj 669 0 obj << /D [667 0 R /XYZ 96.768 705.06 null] >> endobj 670 0 obj << /D [667 0 R /XYZ 97.768 487.926 null] >> endobj 671 0 obj << /D [667 0 R /XYZ 97.768 456.045 null] >> endobj 672 0 obj << /D [667 0 R /XYZ 97.768 426.102 null] >> endobj 623 0 obj << /D [667 0 R /XYZ 101.643 310.535 null] >> endobj 666 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F13 311 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 676 0 obj << /Length 1773 /Filter /FlateDecode >> stream xÚÍ]Û6콿"톞3\\[þÆv°b÷°— Å}è:@g+‰ZÇN%ù®ý÷#E:q>®½®°'IÅoÒ¤¹yòâ:ÉfUXå"ŸÝ,g±HÂJÌò¬\Ü4³7^ΙȂ?£,²*\j»˜`VÑúÓb<¿¹tÊlì%Óo¯®ßÎßÞü6‹f ¤UéœAŠ[£F¾¨ûÎ:Ù¹‹=ƒáaƒØýí;U»ïë^-—ºÖªsö†m”ìF¶ ÍæêÆ“Ö/®ãt¢$Ça•±~+B)'x/¢Ö2,EBx7ó2 úù"ÉK°EœvÚ®ñTÖ ÷sQŠöR ݺÉcÕÇ­ìºèÊn.Êà6ÚZÝ­,á莉­™¾QvhùÝ­´#³¾£kƫвvºï.½qVizh/I*¢ ß"¦l[!« îÛVnÑÉxëiâ†x[:Ük·Fé_È·à‡,¿ÌqàÂù¢ˆ’àõZ:FX÷CÛЛÑhoæ1Z·H¶Íž9àc„4Ò4cLMQÆôÆË÷kÕÑnªEh¾·[Ï@Õ£šŸQÈ2x=/SGÒBØñ3ÇÂÇ£€R×è;Ý ²¥óV«Z1uiöHŠ9u¨žB‹x?394þT•œTqkÄÐý©îæY4} úLˆ ~… è‚D ¾Á bÁÞdðFˤÂ'à†"âÆ‡<K:m®Q¼Cž–ÓóØl-1"èº_Ò þ¶ÄW³xí‹Ò·º!è²7ãR÷­î”4眇•@×®7ÄI\FÛ÷¸K0/â ¡½¯.¼õÙÂÆ%1#-àÍÀ>aÝR [Ô躩Ú0ÈŽ0„–4RS¶ MšÁµ/ gµ‚Dç8sªr¢J(‚D•RnDbýÍÇÂHÐÞjؤl`Ž"VÝ£Zé°àÙöLˆ˜$¾Fãz¯[æqKŠ3^O‰ÆÈÈ~¯UüÞ_ß‚ ƒ7é5ïç>Y1©“* nwNƒ·9¬F-{sƒéh/yq4Ÿ9Þ`WC¡»Ä2·¾hú[u´¹Õ’ñÇ:ƒûÁTØFùŠQ>¯³'{š¬ëcuƒ ž$È>Ið^q8 Ý{Ýñ…W8a·øK«Lèy+ÉLX$¥l‘T‡1uá£å‚PÐHˆrqtå} W²õâJO„ѽæX“…^l¤3ú#Ed%F6‘ñ!H¯ø Ü·mý@D5 /—']½‹”ÃO—·•rv4ÒZn·ôáƒ, ø`ã‡XÐÛ5¡$ù% Ë"Dª$uÿq»^,µÿDŸÒË«°鈭™\vDn‡pEj"©8TcÚ¹O[^qCCîy  ‰£*,“bä²<#Æ¢,¡aÊOëäas„]ÖØKÌŽ#økº¸‘žU_ÕAŠ."è$マo*ÃSmÃnh[Üïz”’Âkðɲ‰}Îbq’…UYºéÔdäOÄ)î!kLЃ’eê½m»#¾¦¿ÿ’«T‹)  _·ª[¹õ—]ý'ÙÿžûÏùŸû[Ü Ú^ØçÏ9“&Q2Åz@0‘FaVeŸ Wa^¤;toäS¹ÆÌîÄİ޲åÙË]Ÿ‰à;…½ í±4ãê (nîMß­¦>}ö3OC"ˆ§¡Éìt(j —ÁwÌ›cèÕåôù¡\’–]Ñ÷°ö^Â7âK‰ý`|€í…¶ê/qFèǤÚ^aóÿu…Ý'Õ·ßcñZÀ«IPÚÕ^Èã!…*ÕX-IQÍH<á%ÉB‘  E˜g,I&¢1¼zÙwо68¡œáôÕO –Óøí›´„¨Ù?ò€ÝCGߥE°’ðýÇ,C µDývha"rªER"J¦P>±_¦ÝÑ¡$„ùl6Aú„Ê‹³!½Dr@ñÝ©«Ó0®ò„‹¶°¤w÷ÜûŸ=xNm±¡Ø"4€ÂFɬé„VeâþÇÇg,V¡LÕh1{Öbyñ vŽ`FUùƒ•ç ††òÅ"©Ò0:®~Zä_=LŸ˜Ø…†¬TôFYœÓ”°^L’ÍAÜÅ7Êÿó¿¡œ0ùG^áO’#Ï21»›4MÛ<"-ûÁ|eúaË#"á›r˜Hò,I"Ç»'¿Þ<ù'šeS endstream endobj 675 0 obj << /Type /Page /Contents 676 0 R /Resources 674 0 R /MediaBox [0 0 612 792] /Parent 659 0 R /Annots [ 673 0 R ] >> endobj 673 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 472.641 102.639 484.596] /A << /S /GoTo /D (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 677 0 obj << /D [675 0 R /XYZ 96.768 705.06 null] >> endobj 624 0 obj << /D [675 0 R /XYZ 101.643 476.128 null] >> endobj 34 0 obj << /D [675 0 R /XYZ 97.768 173.286 null] >> endobj 674 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R /F29 110 0 R /F11 183 0 R /F10 255 0 R >> /ProcSet [ /PDF /Text ] >> endobj 683 0 obj << /Length 3287 /Filter /FlateDecode >> stream xÚ­ZK“ÛÆ¾ûWlåblE„1 'r•"[UNªâª˜9Å9` , 6¨•+>ý$¥æªã0NcNL‰î\˜%&Å)6´Y ûÃÃò¬je!F©½›MúN$IWWTq©ÅŠ¿^Ê/ c•ÍÌ›b,4‰öô÷«ô»äÓÉÿ3“߯“¯R·FÿBW7Zè¤01 ­Q7+ë÷U¿PVŸ¨ꫵ1¨`É"®IQŸïu\‘?Èû™Â˜I ñåB Waê6pÔ™¾Ýæ¶(/mö´¤¼Úsß])tîÚã±mxx‡¬ÚjWŠL´p¡ŽR¡Êñ–ÿ½$IEà0ás ÿ4¾?WŸíx•í¹Ba¿Hq|Ê»ü¡.I±@•¬ •5Ë`½e¸ -ÇvmØ{ã@Q‚µM5Tèóp¤}<›"š& ȬºÝW=|µãÉ]¹¿ß(µB¢‚¾‡õ^¡„EMúæ€D´ôË.ãvǶÖÇ™_NØ/sŸÀ65%ˆš°3¡&yäÉÁª®W‘­ìSLñg`¢óýûoÙØß .#_•ý"Ú¯]™>4y¢.„y Sr³psCüq¶.œ‡³S0TQðî>5è‡éÅ âMlå5ZNQJÆ€äúT”ýlYn æÔNp•ū#»¼iÚU±°Vktyó À³Ã>£P tíÊnÈ+Á+H¯Í$ñŸTM?äÍn5¦:&©úROp6Qçñ·Þ•MßvL;œäÀ£x ÜR?¯ÀLô.:ùRïâBë5ê€Ð³ü@Öá¡ýÅ‘??fF”8œ, 2˜™Ð7%›\ #†Iÿ|‰‚üê]ÅnXóÙ:tnäuÕÛt Ç\°îIy<ƒ¨\D‡h¡dJð¦nÊ[ˆÜ¶z¢2~-WÉÓQ¦€«£«ÇŒ¢MPXÿ{;ÈiÀFrBÕãZ*à £p_±Œ;Ç%$2“ÚPEÉ™{•Ó8G4h½$@Aå¤7ëÊe œ·Qÿ’J-[NØC >…R„¬iõ`½ûŒX4ze‰;?=ôeÇΑ~G~y›ÉòÐ@¼€Ï¢‚Pe *Œ)%Ìï‡.ò³/òºoý·XCˆ!i¬Šù´™|0VJ°3¯”`¿mêß®  XÙGfŒ#D)¦¢DÔ§ÂŒ,e–xnQ¯6&ˆdfœ/qðbÌPõ°K’JF´EÇÈqÃýë/ą̃©oWaˆQã„?®§8Ziâ®R»¦Kèš‘O{÷~¶X¦ñè–^¯ip6…Ï+L™0MìK,™Y}ï÷«¼DQ|ûÛU´¹' m ËÖK¼(›öX5ù@Ès1ë„Æ £]ªiÿ鈿l¤‚¨›sÆ «… VFÇkí9²ÇyZO]»ïò#wvdÓ=w,?ØX» —îÑš4´Š+<…Mƒ_u\xAäâNgUM"I¦ ˜ý¼$ÛböÆg´`I\ b|Y åë4æ,‚Ç A Ú5÷žf¿Dt×Ò+«FÑ?œ^•‚,ê·/?ê˜ÇŠò1?Õ2qaóZ4_<ä})´òb’ÜàÀŒ\˜KñrSîõŠÀT–fgøä}6§–D }Ó'öbvªfuåI lâÙr’ô{ñq­/{ñ€_œz_ÃÛå_ŒÕIÂ…/ûµXFóáâ‘_B*€õâ—ÔÕE ®Õ1ï°Úåéü‰â&ˆ5þØò¡>¨äy ï™; £ §‰ 5p²nWI´©)@ís€¸Â2”&ê4@ÝËNùóp>º@£É=\¼ÔµÍ?­âšÄÝàGŸ%ŠÌ0¤£)LrJ¶Ñkî×àÛÌ&‘ÝÙ·D¹££|ͳ}òIê®kû~R¸çi@eîŒÖe”¥ŸÁî"ÝM ™i‘(z!ÍUÿ˜¥Åéò‘ô}uT»½( [N:ŸñGLë=eQRÏ­Š2/P"•;Ã_¢(øG™wäî©’ú¢H &rbÐ[¡ìq0#+æY”MxJ~[ÈÛJÞ=øtˆàQމ3}ŒI8ìž*XΟp™œ«,_L›Ìèšè%Û×ab–n}=”d ‡|¢Cáæ<ûêyˆÐY–ˆ]BCîpv9pãYp0ŽÉ½CÆ×Yø’îÄ !õÅl9¬¢ø:SÖwÕþDqg…‹W‰¤f=æ<ú×Î ô[`{ Ë>œ³“œ:¯šªÙË¢†Ÿ¬S´>iˆ-ÒÇj"Ι9I‡ R’OÙæ º|®^†àU]È%}ãþf¿äF¾Ëë;¼FdÒåÝ!g{÷ÿ˜nû°¿í!#w¥à¥Ä¯ò|ŽŽnÑ Oç]ñe6îݽrR†ÊÝÊWü%àÄÝä¯PA"{ÆŽäüÿYí‘~´!Æãù›hqþ&%‹0½V–uG•¼ìù¥xcÜKÐR ñ·I4…µ:‡.o‰N:v¹ùåfÛ‹R‰vš=؈8±5#»ìBòt5ß’$€c\ºW|äØ eÁƨV OJdy/ýå‚,$,ïJOìá­K(n›‰®v! }:t’ÉTŒRœ+”óã+…Bý‰Ü¼Tª„º uÁ4JnŽŠúf\iò´¦â~š1„]È–ÀºGe\á‰7I)òâSò¹ì"6áÿ5†Š§c)k,5ã’ôüë;g¹2ƒ`ƒCµ;âË’¾fŽÕhùEûó4ôÜ/?>AæÕS®œù\Ù#CÜŠ&ó þc…IñÝzü÷†žþ½¡“ŽQ‹F©?còW@øAî¿”¬U lÑc¹æûiWöÞSþÙù¯9X* é=ã ¢—ªÄ¬,9Ä&ó›h0ŸSSøqùûæþÝÌu=E†Y½ÏZ¡kç/Û'n¢fHfªD^®„&9¹\ÛðÒÞ»c…ÝÄ¢#Ž“Æ´&QŽÏ61¶bUèSp¢Z)Òo s%7ʬÖòÀè§&*Ëj/GKL‹æ½¯cÐ*ä~™*²‹…S *Ì=¼?0궨öGnøO˜É¬ý/‹³.ÓlSéà®$÷ ó’pÀÀO˜ h‘¹|-·ó#w|¦|M-¢“rß0<õÂFä…æ(½1¯3\ÖÉÖó hÈ1‰¸0®’nκò¯Ø.{?—-ÇeÁˆçå\ˆÐD‡Öün!¹›M¹¥Ê{ Lð> |bI[ˆ£Éó=] c•YRÅ/„ÁV¾6s<õé[YxíÂMÀœ•Œ›×«uéÌÍ \ãaƒ„ì„©5ü.¡ù¯~Ø~õ?›Ã endstream endobj 682 0 obj << /Type /Page /Contents 683 0 R /Resources 681 0 R /MediaBox [0 0 612 792] /Parent 659 0 R /Annots [ 679 0 R 680 0 R ] >> endobj 679 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [144.449 247.826 151.423 258.674] /A << /S /GoTo /D (figure.1) >> >> endobj 680 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [326.275 223.916 333.249 234.764] /A << /S /GoTo /D (figure.1) >> >> endobj 684 0 obj << /D [682 0 R /XYZ 96.768 705.06 null] >> endobj 685 0 obj << /D [682 0 R /XYZ 97.768 339.925 null] >> endobj 686 0 obj << /D [682 0 R /XYZ 97.768 296.643 null] >> endobj 681 0 obj << /Font << /F14 120 0 R /F8 112 0 R /F11 183 0 R /F10 255 0 R /F30 111 0 R /F7 257 0 R >> /ProcSet [ /PDF /Text ] >> endobj 690 0 obj << /Length 179 /Filter /FlateDecode >> stream xÚM½Â0 „÷<…ÇdHJ~š8l ¨R6ÄPÚ!ÑVxJ¤Êƒ}¾Ïg «±$x'œEP¹%T-yÊ ™€Å8Y2¥Evh%ìzrëoñÉ™Û@²Á o•…ЀRRXã@i3ÂB gZÜoŸ!2®µ¦r͸19ÝÌòZ>Ê®Šõ¬ÞCœ0Cû&ázî]ϸBZÇ—`—p‡Â{ \š\hó‡?ìù#O9 endstream endobj 689 0 obj << /Type /Page /Contents 690 0 R /Resources 688 0 R /MediaBox [0 0 612 792] /Parent 659 0 R >> endobj 678 0 obj << /Type /XObject /Subtype /Form /FormType 1 /PTEX.FileName (./figures/balance.pdf) /PTEX.PageNumber 1 /PTEX.InfoDict 692 0 R /BBox [0 0 504 360] /Resources << /ProcSet [ /PDF /Text ] /Font << /F2 693 0 R>> /ExtGState << >>>> /Length 1327 >> stream 1 J 1 j q Q q Q q 1.44 15.84 486.72 342.72 re W n BT 0.000 0.000 0.000 rg /F2 1 Tf 12.00 0.00 -0.00 12.00 16.13 24.43 Tm (0) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 48.32 130.09 Tm (1) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 80.51 24.41 Tm (7) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 112.70 235.98 Tm (8) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 141.56 24.43 Tm (16) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 173.75 130.21 Tm (19) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 205.94 24.32 Tm (22) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 238.13 341.76 Tm (25) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 270.32 24.43 Tm (36) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 302.51 130.21 Tm (38) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 334.70 24.43 Tm (43) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 366.89 235.87 Tm (44) Tj ET BT /F2 1 Tf 12.00 0.00 -0.00 12.00 431.27 130.21 Tm (49) Tj ET 0.000 0.000 0.000 RG 0.75 w [] 0 d 1 J 1 j 10.00 M 238.36 335.29 m 122.48 250.67 l S 251.24 335.29 m 367.12 250.67 l S 109.60 229.51 m 58.10 144.89 l S 122.48 229.51 m 173.98 144.89 l S 367.12 229.51 m 315.62 144.89 l S 380.00 229.51 m 431.50 144.89 l S 45.22 123.73 m 25.90 39.11 l S 58.10 123.73 m 77.41 39.11 l S 173.98 123.73 m 154.67 39.11 l S 186.86 123.73 m 206.17 39.11 l S 302.74 123.73 m 283.43 39.11 l S 315.62 123.73 m 334.93 39.11 l S Q endstream endobj 692 0 obj << /CreationDate (D:20110218110759) /ModDate (D:20110218110759) /Title (R Graphics Output) /Producer (R 2.12.0) /Creator (R) >> endobj 693 0 obj << /Type /Font /Subtype /Type1 /Name /F2 /BaseFont /Helvetica /Encoding 694 0 R >> endobj 694 0 obj << /Type /Encoding /BaseEncoding /WinAnsiEncoding /Differences [ 45/minus 96/quoteleft 144/dotlessi/grave/acute/circumflex/tilde/macron/breve/dotaccent/dieresis/.notdef/ring/cedilla/.notdef/hungarumlaut/ogonek/caron/space] >> endobj 691 0 obj << /D [689 0 R /XYZ 96.768 705.06 null] >> endobj 687 0 obj << /D [689 0 R /XYZ 265.261 246.719 null] >> endobj 688 0 obj << /Font << /F8 112 0 R >> /XObject << /Im1 678 0 R >> /ProcSet [ /PDF /Text ] >> endobj 699 0 obj << /Length 2458 /Filter /FlateDecode >> stream xÚµYK㸾ϯ0CÔØ6G$õÜdl‚ 2Á"‡lßf&€lӶвԑäééÝ?ŸzQ–d¹'¯½˜d‘ª‹Ué?>¼yû>[å*OL²zدòT¥I¶JâT¯v«Á¾iïÖQýÑÝ­mžuQ~¹3Yà˜~rý±¹[a§îÖIdƒ‡£LíÊO¡¶®uõV(Û¢f&§¢ïp~>ºúîóÃ_ß¾×z¤‰R•Ùd²"²d¢/öü|Ù1·ªh¢iW‚û£ŸïÛµgÚ®¨Ä Ÿ»žG0( AäjmÀiž­ÖZ«<¿ìš¥X˲6 \±E&G&;ï$ ªC¯ïDQ| ãðCÍkاðéV<(,74rUƒ>3íYØáò®/ÚžÉEϤŸî4ØÂÊþÝßvL®§Œ‹z7âVT¼êü¤@-MæêHåQ45–vÔê$ØVEו[0üyaBßpÛõM+ô-(CNp¨‘ŽiŸp¦ð-7UY?’…;?îÈc6Céž]á;¸·Åý£™êд`ûIØò®€¼b·+ëƒ,óiK—ŒìΈmÏ«!ßu<޷͉{7}ëPPC—•câ©(y¿¡XñFc`UAA‰ …ÖíÖ› #' /l»{ì§Á?\Ó\¿¥Ýb†çžL §Æøh19ìAñ8Ü\Î&)+sOOm#QLÎ7ÁO|†pv̳v¸QĽ”Çà&Ü Z‡!H½mÑM¾-G’Ö1Å,޲êÔйÛ9áDá:ýX˱›°ÙE…áÆÞáqЖÝ#÷º-G%téÐŒW5ç¾s=úW‡êþìÙÁ¤,Ü/²í¸_JÛ´‡¢.;Še$³.²%x´‘ã(kŠ …íÎñì|o›s‘ÔwrÞ[?Ô#d_è8Ê÷¤´?»¶á\Pȧͦsí—»v¡/›ZòÉÀ°u§Fò(D$¤þ[¹ãz,‹.ù úe½s_±ëS`MOŒÉÓãâ°LT„)ÍC>긂—¢UO¸—y˜_,“æ‹@ ûÚóÜ2ÇϜڊg[2aý’^ØÜÌ Ì:±Ä¿LÃŽ‘NÒIC_óÊá“§‚Ž"lÝšÏ_YídÉuÁB²ɶ9‘ǘE_nÏP«¦¸¢{¹­All¥0„ѹÇ1ù:GÓu9ŽB°4‰‡r{]ŽSFù¨_ó€Úš«PÏÜV¾»æ•©– /ƒç2¥t)GÉ¥w¨?”ô.9òé€{:þàÒ–2Ú×L6a¬²0zÅäDŹõóRœ è–xiÅÿ>+oÁ"«HéËÒV±²fÀMßAÈåa )ŸçÁ±ÿO쾭ϸ<'´ÜrÉNn0J3eÓt–z.µ¡T\VÞ5 ˆû!ù ;{wÅ‹6jƒ¿ÒW]ÆFf1%„‹Ëœ#­²Lƒq:clfŒË&'i@7Œò„ç™Jòd&ý†YŸÂ0\‚½q>D’öÀm&)SY<Ö…é+°³¹ -6cßT>;AQ*ÅD>9ñ(t82rP}@ÉR[ן[ªƇƒ‰GÁe$ÏC[ð\×´=ÕM !äÞ3'L˜¬äçGu¿<ðûÂn„? k±Vl F²¨€a9èJAmüV›¡"tKõ˜ŒÈ‡K’öÅ H«C—XÂd‰e]€zu<§ïñæ†Ú8ˆd¼X?ð¢pöýÖÉV$E`Á•¡¹ ÑTÒ0Ö¤½Úá¾äç’Àô¶øU3€Únb*hÞA­[ð‰mw÷\K/ZL¹—®,Ú.5'@oŽtQ[ÔÖ­$–-K‡’Ú‘ª@^´8Å[ÕDi ‹jÏ hÇRï/ωµ£ƒb#£†tP7äª8xqÒÁûЮ‚1Bªë’* g•Ý´Ž´¹Æ¢tÏPËC3¹$ªáÒ}2[Ká—% –‚ä:Ú¿“G‚„ëñ†û­>·rËØ¾=Wý˜)ø\®°£‚¬ÒÆŒŒXy'c«,2Õ†¢ôšI²¶gl¢IÂ[˜V6ϧ)‰%¬c8Ú¿_s»?×[ĸ˜´ëQ>+ ¹dv„ý‚ë,S¹N¦ûÊa5׺'”þxÏc¹Ö@ ~ÃÇÉѲùâÚg@9=¿’ÌD¢êµk¦Bô÷‹«wî ÏÛÄúªiZT¨jrÁ½A† IhcE§Êšb¿7ÿ )·¾Ç|Ó6Ϫrõa®‹þ®^_ÐÆìreãëtyçþ‰µ÷äæåñ}fÿN™·4£½ú88ô# üüy&òbýw܄ߓó<ç¿B kýã-A7öì"yM¢ý§ÆB\ɽžp5Êió}#r%*_ Ü$Uyÿ‡ÛL Þš+ÿ—Ý"ž¿Æöýÿ³#3÷®MùMOÙ¼¼dÍ7ù%~ãÒ²¸eõ4½þÝí]Ûd3ÆÅ¡ 'èÒ+±4::œ!Aß/¨Šè;»Ô-J™´uöc0™+“ ¹ñ¥tÕ®¹ô´~'Øö|žÉKµJãáê˜I.Œ¤Í¥5’#BÇ2ö¹SËØÈ8‘±•±oÓg›¨,ŸîÐÉò‚§„Ú- Œ\åN£«€ñ ¾«:“Ëš‰OŒYº+“2^ÁÛxA &ïå=Õ (¤5—÷‘*ZD˜™ˆ…¨G)˜ãöÎ_PÀuœŸØ5aklº#bõ–ÎÞRt‘R—ÿ `¿U¾fIð¡çÿ 'ÜÉDìì›ö„=k4…¯ ÜåGFìr¶‚œ-b¤[njMé~„8zýxs.)v­õ—aè4Ðk‰˜úG`ž>!èŠý½ h§âkyÂH?ós<Ò¤ä£=·­/u¼ï0,E `Ít•¨<µášÐ( ¹Ù¨È&þ=dá:¥«Ñ’÷Ï Hª—î¾ùèò¹XN¥õ7/¨ùå}B‹>ôScOü ½måóµY£â,žÁ¶³ÿÃñ'_G¯úIˆƒ `4ÿoMïaráÁkï/þûäßHÑò> ïU4z5ÇEp¸:Š=ôE÷ŒÌÊŸA¢¦ NdŸ-ŸÊŸ-:uÃÓÿv6 Œ¬…=û&xŽçà9V‰Ž_;ç‘ÒÑ ;÷ÀüiÊtžšgB&‹*$pŠ¢W5БVFÏj@¹¿€­úÝ;ÆX"ö:ðu¢Uh >†Àu%ciŽ ßüùáÍ¿ý^dí endstream endobj 698 0 obj << /Type /Page /Contents 699 0 R /Resources 697 0 R /MediaBox [0 0 612 792] /Parent 702 0 R /Annots [ 695 0 R 696 0 R ] >> endobj 695 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [128.528 466.451 134.395 478.406] /A << /S /GoTo /D (btree2) >> >> endobj 696 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [202.432 231.886 209.405 242.734] /A << /S /GoTo /D (figure.1) >> >> endobj 700 0 obj << /D [698 0 R /XYZ 96.768 705.06 null] >> endobj 701 0 obj << /D [698 0 R /XYZ 97.768 469.938 null] >> endobj 697 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F14 120 0 R /F33 118 0 R /F36 121 0 R /F35 119 0 R /F7 257 0 R >> /ProcSet [ /PDF /Text ] >> endobj 706 0 obj << /Length 2811 /Filter /FlateDecode >> stream xÚåZmã¶þ¾¿ÂߪMÖŒø.Y¹KM‘-²h $)¢µµkåü²‘äìÝ¿ï ‡”%™^ûî’¢@¿¬iŠ"‡óòÌ3ã}uwõÅ©gŽ9#ÌìîaÆ¥c†ó™Ñ– xr·œ}ŸU붺žk¡³ú>Èu¾½½ðÉifSâ[úòY½¼ /,ý’ïþ:ËgsΙÓç6•G›· ?IÃÏù·/žó_¼ájpIa×ûE´b¤†¹†ÛóñfÛuõÐÑ‘_Γ’Ó—§ÝsÕ|6PŸ¸žsX4¸õ¦ÞâkþᜇkmçþûâÔ¼Fº‡½×KÔ+dªúÁNíážò¤Vs–ç½Õ_+Õ²BÊ¡RS¾Å «ãšÛã= z Ÿ ºùqb/ë³Íi@‰‚÷•1‰[9¦¤˜(g²‰ËM¿‰7'P$ϼ÷ü¸¡oàsˆæaÔ,aXnÑ”Á°¥?´i³,ûð†ýŸÊºiƒw8ÉŒžxÇÈ‘0TäKt¡\ðOœ îâ'=¤à¤‡ü]CÓ%Íõþ‹s?ä\… öm½}Œ»†÷_„®#&ûsŸžz¿Þ”èÖ!¶ÀÀ]|+éú“âÃí\!Kðfv¨Í÷ø‡ƒ€ã>‡$¬o Ïׂuÿ‘‚º\˜Sç'œºG“îxÉ´+F1¶ˆî?’ÇŲIÊ Zþ·„GæLs;›V]"Xä3™XnÊiмÿð«ê[ÂC°/`P1{öK7p®µøÚzöÝÕ߯^!Ÿ ·20‡¢.0† kÅ1äØÁI8\GT9Vè:Xt»MêÝ™o0ÉrŠ…ãý„±Ï)^¥?<ÐC½€ñ8r{L†m ]|ó!ÿ ' 1ãP#V±«{Ð1âV* pºO« 9Èî}°p°Y!ñ”pZS·o1Ë<ûKŸ]a~QniðŒï†Ýö$ï²ìÂD·òùF-Bk«è&a @wú&’ /Ûºüϳ7€µéJk¦•¹¦ÚTGH/ÃM¥‚J!0³ºÃ÷Çî³ néBðþ+ L,G¤‡ìd‚žnpÉË»‘:éF’I€‹ÁšKÐâ]â4bM‰Ó&hš—B H$,HP¢`|ñ´ˆlíZ¶ÌØIŸ<‚“s5‹1ÍÀb°(äˆÑf60ßÿ‹•'ÇùQåè“ ðbPWŸS‡­ÎÆ”úô˜‚<‚e@Œ0Si° ­ÊG0ý9hLB©úœ¬àÅ™½1¨òGÝpß ¸@÷XƜսq.¥}û’òo€ 3qÐÄIÿ=ã€ï @;…Äè`Xu26»¥Ã&QÓÄžc‹ìÕfÔ(…g¾Bv±ñƒM½Sù–úø.„`ûû@w$Ãgµyª›zQ®éñë¯ßÜÄKÃ`¿%¢~Nð̃S¢¨G<Ùí»@šëª%˜y¬õÛ´kB-#tÁÓ74àý ’T?Lv“òh›ƒÝå¤jeÖ‚ñ!1<Écú5<¶¥^ß—Fš…†oqb¬…¯Z¨S¼Ü¡°ñª¤Öìaæç}‹ÊrEV.—¤\‡Hlyj&t¸o“iÍöðLλX›1UJ4*Ͼ«©Å2|¼ñ.HnRÄÜ n“äIBàz&Â$ƒ1Ïå(@SÉ@A=9X“bÚà¶´LÌ·LNE¶üÒÇÏvI;Ôõ‡†šGÎñ//MˆUF#ƒÉ ,ÚTeH‘ŸºÇð}ŸLäàÊœ<£}(Aöœ»g³ : 0Ù·U—üU‚©a{ùœW…CJpV›D[%à*ùüA}=—r‘ˆÂªr½¦™ qjiv±oš@×ïiª›õæãVƒùÔ¤ï qëŠðsUaâá…ÍØî¨ÛÐónUn“ÌÓ0Áí‡5ÿÞ&-4¶âÍ@ªº«6$If… }rx\ÞSÚò°_„_íp~‰mð$`+¤E}°¥÷)ÌËOYºˆ½eKÈì‹ò{û‚œ±Úx'z{7€i"Ô`H‘O5™[÷¾C¿v—·›‹½;€&êØ*Oê‹9«<ÕV a©¢‡ íµqƒ-{~¢ è ³Ò~°<É~¡Ô.)϶zN †#°ÄA¨=_Á¸ŽŽÖ,W“ŸÛý=VWÙÏÕ" /ŠõRþ&=(Ó”RÙ?±L«ˆŽ`ò¶¯8ŠF$ÃXh@fˆìòbÍý”j ŸÍ‹[ ÷›@'HøÇ•@æ"–­/bÙÉ\Ño™‰YL¶Â‚üb îSYÈÙ³6„=•=kC™‘ÿn6<®¶ ³~]ÿ}+­Oð3ep*`¤ÿšË†3YèK”mÇÊ–Ie A'õ9Pø§W®ÎÚ¨\ƒ˜µ)Ë4¶VÎåE£ïw©¼ð|¢Å!ý‚ŽSñ'æÀzð¥U|xŠÿÀæÐÉø$þ#&üGÔowØrCHJψêp‡PÂâìsL'é¿cÊícŸ`w®¶ŠT`|}ÿQF Òk7f”*•詨&*ŠYÊâ?6Ñ¿‡Èl]•Ë––ÒÿN!ÙÿÉ$'?¬©ÑÏcá€m|+¬Éš% ×õ6ìC„Ï×ÁpA$@ßüaNó?Ñ%«È¡&êy~ ›Ýþ’>öH¢ûØÍB€BIi p“-úŽù,ìP endstream endobj 705 0 obj << /Type /Page /Contents 706 0 R /Resources 704 0 R /MediaBox [0 0 612 792] /Parent 702 0 R >> endobj 707 0 obj << /D [705 0 R /XYZ 96.768 705.06 null] >> endobj 704 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F1 256 0 R /F10 255 0 R /F7 257 0 R /F9 259 0 R /F13 311 0 R >> /ProcSet [ /PDF /Text ] >> endobj 713 0 obj << /Length 2341 /Filter /FlateDecode >> stream xÚí[[oã¸~Ÿ_á·ÊØ1—wQÀN¶èíëh*[Š­Æ–¼ºLvö×÷‡²)[Nì‰3H¶y$‘‡äá¹ó£ó×›w?~4“„$šëÉÍí$‰I¬ÍD«˜p¡&7Ùä—¨YUu»JËl:BD·U=ý|óÏ?2 ÔœˆØL(é)èD“$ÆRH"9™1÷pTÅØ<„5 ˆ~šÎxlFgdŠP6˜ñ©Â¹'œÃ„&Ê×Õ÷Ÿ¨b~ê``„9fiéÅðï(‘1pÀaziü¦~ºÙ°ªÌ„1’t÷#ÓÅ„ 1&¢pC‚‡‚-Б%1±|L!jT!2˜ÃÕèØjöMô«}¢\,º¥Ãí-·'‰ê`{^Cq0îNƼ<ùñTÖ†ãÉ, úp¬jF‰b¤þ¾Šæ×T´ú6Ï;T4{6MÿglFˆëhZ â5ýx9ç¡ÈB%Ù¥ã×¢svBïªeò­Š¨¨d|¦êùwvr5¢ú§†ÄVûœ ùà-X?“ —ÇÛSDò‹\Xž¬å[°~ýÁZž¬å[°~aÁúiN®FT?ÖDæƒ5XŸ^ñ™êäo±úcõ j#.pdƒ¸î÷Ôµ¾V¸þ¢.ËËîD~"ZK"’—ªñ£`ýË µû:ãZáZÿýÜœÎ5»Ò éóº“ty- ùli€EÁÑ@Ö<&‰âyš¼®§wæQgxš:«,2ç«H´úT4Xô‡§„Ä (ç{|Ô/øGÒE¶K9z°>f¡Äp$õëx¢:ÃmNg2N„sÓ{¦õ(FŒޣěG½yÔÓ=ê±ãÀ#gI5¸¦8äÎCч1ëO Ûo~¦ ¥Î–“½Ù °;ø³/DoȧFDU= É£¢Í7 ¾V·ø\§õ2÷ÝuZÞ᛽ô3lA‰IÄ™U±:y3cwÌ¿·ËÊ(]¯‡Üµ«|Ïšç¼(už6¾c>åqôuŒWp~%ù7©gx‹$¬Þ¯ò—¾·×Hž4Ë8.óû½ÐQA<1„ƒ;ÏlhPç³·kÇÜs6Â\\Átcn£’AÓËÄ)%=-@a˜¥—Tß'tÍÃõA×RÇ (+¾ÎsáÄ&%‹šnƒ-Öì³ùµKkd–E÷E»*ÊÁ-ëªÛâkáéºrag+^æ™Îb­¢›ž¾I7¹OTA™k¼B¼‚³¼.¾L•ŽÒ¶¨¬ªfƒ7 )^!º Ä/½ú¡YàWOæ‡uSm݈,m=)ìëøÒQ“8.  $ç!#=Ðõç“ù‡}tóJÐ{M—|Þ¼6ò¸¶ßî,^>6ÂÎQù Ðøµ±‘nôø 9ÍØˆ8!6rÕ‚«‰õ¼ÜpNjêäÔÆµÌëGl ì\ªKvÎóà|y¥ü~½fr­¦¾è€ùû¨±s£ŽÍꤞ:3²‹·ÔÅN“×Âü>Æ=1X`E3*53 +’(š/EVÞ,ÿÿØòåi˧ÊúùQ•=¤„v¯’D£Ý³°xÆ!öŠÍÒ“âÙõŠªåØ¯+zMži[ßœùÁUáÈÊáÌj ~„5 ¶Ãš„Gs ©-ò ß,œó;á_Ôø ­p`¾Ã¦°#»5 ãÑ+ÛÛ#¡Ðë|ÍŽJÏw÷óùß»÷¢°]uÑÜagZûy¶ˆT^Í'ÍD²@5V×kll*|æ¿v=†_ûKDL·Ö¦×EîaÝÿvM‹„Îw%ë®»Ì> v×y‹âÛŽ“ÞjÚó±OGp›yÍ †Tël°9È:ö™ötð²Nk¶)¯7øfQHû öÈt€È»Ï½DlhŠ%8ÃL Ëp©MÚÖ[|®Šå*÷C¼ áÍc’N"v.k?Zq:êê»ëð>/w¦m{Îwf‡üŹ~rF9úg{„Å "u½+P¢ç­^œuŒ°* åøeð _¸ YÇæjzò÷Oá°*鯌‰Æ¿¬eZ½Ç4Z¤>­••µ€{üÈr—œ< Ú¼ümVW][ìÛÓß² Ãjs0`Þ­ïðÍYxØåÂUUß¹¬(£EÝ´A§êq÷ƒÁÀÅÚæäòݬr[HŠRÒþFÈoæ?)z‚›²¢Yt.)PÖ£Ñ"¾o«JßÖçJßûÑRC8wÉ úw"³}£‹ìnDeZÅæ¦«¿}¨qiHéûþ "C‡¹/°Ú}<õnC£÷°ª\ø€u„ 1ðÕ´µÏ@RÊèå¶w0d¡.ÒùzwåQçhv‚vgùƒ/®Oˆ€}!w·L»òÃ%Éncmn>à kɃ cjEÛŒ¬?óÓ#^J_G¹ |çù›Êß`Y%vu^Z”΢–¶Ù×¶Kûæ$nû ñ·]zàÖU@%Øou뺫ÚR¶­77¾Yäe¶_&ös`¶³4Yž¶+?϶ι£wÅ•´’¡9À|°P4dL6ÜF…ÙeöÛ±x8„ÉŽ”„õ–+üöã®Ëú@¼*0‘ô¥bÑ µçÓÜŸŸ€0§A³¨j¿J7E–j×ó5°±ûvD½p¦Üþ^[þ\A Iuwr¾¨ ÓâamêÄma˜¡N"šzáw¾{÷÷›wÿUaî endstream endobj 712 0 obj << /Type /Page /Contents 713 0 R /Resources 711 0 R /MediaBox [0 0 612 792] /Parent 702 0 R /Annots [ 708 0 R 709 0 R 710 0 R ] >> endobj 708 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [208.141 275.916 215.114 287.871] /A << /S /GoTo /D (equation.4.8) >> >> endobj 709 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [224.001 263.96 230.975 275.916] /A << /S /GoTo /D (equation.4.9) >> >> endobj 710 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [286.28 252.005 298.235 263.96] /A << /S /GoTo /D (equation.4.10) >> >> endobj 714 0 obj << /D [712 0 R /XYZ 96.768 705.06 null] >> endobj 715 0 obj << /D [712 0 R /XYZ 500.749 564.411 null] >> endobj 716 0 obj << /D [712 0 R /XYZ 500.749 536.458 null] >> endobj 717 0 obj << /D [712 0 R /XYZ 495.768 429.853 null] >> endobj 718 0 obj << /D [712 0 R /XYZ 495.768 401.901 null] >> endobj 711 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F10 255 0 R /F1 256 0 R /F14 120 0 R /F7 257 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 723 0 obj << /Length 1425 /Filter /FlateDecode >> stream xÚ•]oÛ6ð½¿Âè&'±"Qßëò° ÉÐaèŠÖÛ tÅ Èt¬F&]вãþúÝñ(Y²å{Òéxß_<þ<q}x“ÌÍbOæËI–¸IœNâ(qYMæ‹ÉG§nÖÓOóß®ïÒ!‹ÜÐ&ž%)T®‹)KÕtx¾Soò‚_!œ8;Äóòa…_]ÁR*<‡ç}>!§3ø[p:”KBëÕÔw¸…ç?Lg!DµF‘“W ¯é?WüH‰^ñžüä[Ûòª¢ƒŠ/5Ñ‚9‰ƒúÊj¡¸¸Bß'3ϼx2ó}7‹l\r±Ž °B‚ÐQ­“„'ÇH Epë ÝÌK@hæ²Ø'¡…lÄ4` â$äsã8kCN.Åu£_ †¿3œš¢-à8¸n@`|g‘ ôU_É¢¬ © 2íºl%KA˜§³'º\s{hc20ä)†ä”¹(8¹ç‡ýxXgÌsY”›+¸¸7M¢6…duÁý¡qè†YW§¥•“vCÙŽ×óÙ0Ýß•¢¨¬ÌˆEÎ˺QÛ÷îê¥að,iL¤ïo?¼%²¾uÿx‘w8Ù_Ñ÷€Ùi8¨zÈR,žØ ©À&` pÌåꊥ~ëÒrÄg(ôÀ Žê¹Úê´ú>ÛïãUg z5pø˜µÀr·äèC¡¿M/®z’.ds_Ù _Ø 3p­sÝÔ¯Æâ?äÙé–E@ìý¬Âm][†õ^å¢ ÄF¢ßƒY^|}sӀ؃”àš€{^ÉåÁ6²`~/·üâú9£ª5ÏÛ¸5=XV‹ÞŸà;üë\d›ù§Â ¥¹^Í~” ÑÖR×ûÛÐ4ë{®†¾6ëNn“©W¥Åld—áƒ?Æÿ´C ìùÔï¾Qݹ}­2æ® 7Ö%w5¶âÛgT¦)ÆS^¿š?p…Œ‡æ-¾°¾»ýéwä‚î>«-0½~3¿ýõöò™0à칆ý0ªí¼sÔ6¦ƒ”K£Qηïþ˜ßþ2Gñ”´aˆªJñBK$4äý‡·¶#4çlÄŒ´Q7H1Òdl+ÔœDÝÕy)ê¶ y;®àþ_纔¶VåÓqYö]ÔGÉCUhF.™ø¯ñ)Ø…) ëk]~åryàw½-µ#…—ÏÍE³Û´6–7ž;åôµ6µØË˃å ãcùɪõ\ïÿ‰ÆDšl€ÐSq6Å0Hoèäè²Öó¹Û>»íg;©GDfž‡ñ·®üYºhþóM¯¼Ïç‹–­ãÒÙe`‡É˜? ÂÃéN7‹`Ïö‡+Ï ˜e™óhç=ýÀmªK1bd£(äzÓh[èˆÀ1Œ_-éÛlh«Îµ•¬[[£›Ö²´Â{K;lˆµ 7¬uOÓб»ù;)¾×DŸ›µR7€„¾ØÉÿ_×½cQ-ëGàð™ƒ(Ê÷»-š­žâë÷â–.óºµN6qÓ”u5@+­$´9ƒ%] ÚO}7W?vJMTCür!›‡1,§fs6pïÝTò¾æŠÞ!”£È*¤èKkÅð¬3&Ç`m”ÜÀ~¬)J³Ôw½4:Qæn ’ôðV‚äDIFuƒ- Vé¤éhÎZ"zÀ·˜b®öH)…=Þ­dmAÜtKûÆ"Li…U¹zજ RÚ© {¡¢QøBS 6F* =«9Gß–·#–ÆN- ·–Âí á] ™‡ƒ½å䉉§üK“Wj+æ+WKFÖÝ4 L2 cÛž~6yiѦð;ÌîXAQè¦QÚUÞWqâ|>­LZ>îYNjÇ ý{M€—…ðDdÔsJ,)Ã÷óÿ; Æ endstream endobj 722 0 obj << /Type /Page /Contents 723 0 R /Resources 721 0 R /MediaBox [0 0 612 792] /Parent 702 0 R /Annots [ 719 0 R 720 0 R ] >> endobj 719 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 602.275 171.007 614.23] /A << /S /GoTo /D (concordance12) >> >> endobj 720 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 219.709 223.31 231.665] /A << /S /GoTo /D (concordance1\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work2) >> >> endobj 724 0 obj << /D [722 0 R /XYZ 96.768 705.06 null] >> endobj 725 0 obj << /D [722 0 R /XYZ 97.768 605.762 null] >> endobj 726 0 obj << /D [722 0 R /XYZ 123.92 223.196 null] >> endobj 721 0 obj << /Font << /F30 111 0 R /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 732 0 obj << /Length 1746 /Filter /FlateDecode >> stream xÚ­WKs›H¾çWè´Av30¼ÊëTe«6UÙÃ^â›ã‘=².[Ù¿ýÆrR•ƒDOOÓÓ¯{š¿®ß]|ιʓ0Y\oyªÒ$[$qªÂ(^\¯7^±^/o¯ÿ¹ø Œ£¤ÎUœ§‹€eîÆÞžÁÿö–e‡J­‚4ïD]½ô£8ñÜ}ÉDñøØÔÍR{¶pÂ[ÕOÕ2L=§–~¬Cï+캢q¶º“—ÜDË®hîʶãÚ½°¿qp¸·«e˜y÷̲-?Ûºqåº{»u ª? õ ? T’& _k•džª(vðBdàܦ~£îî»5EžkÊRq ´ÆË¨ÜP ¬PT~ tú2\HE]´¾lX­u|ë‚ë²pàóæ’©%³«’|DËj~B´—>¬Ê¶S¶ÛM¼ ½²ë,k0täÜÛ~oÏÞ‚Õ:ËÆîiL^xŸfP8ˆ‡±wÆWÀ”jéd&²¨ì¢P®ýºòä¶+v˜M݈'ì1£°M+éÒÃt%±ÊÒ¬;Ç^.}¤Þö¥AcÓ Ö›iÐÐ’–cQOU‚ ˆ šß%J×è‚I Ö›u¶®x…š$öÖHho ”ˆÕ°ìÁ¢[H}§ ‹B×if\XÀ,–óÐ`œç{[6ÏË8†|Í@/&êÚ§=Û:Ð=ª4š)>1ó‚& $ I.Y¢ì,!¸œT Ú‚’ ›m„·ªÙ.ª5ï²æÇ] .Þêª#Í<«J%dÂ#ÒÛIÞcO6Б¤ 7g(öýÐVëèh‰€ŒÌQl ß®{,ü¢™kÁ ‹G•¯vo¡ßíþƒðÍñ4&5u[2£ˆÌva.f×ò¤n)oqL‘K1¬ú)„ä3†c ¼HíÅ'Cv¸¯Úݺ)eëu,H×0©WqZƒ«¾Éb†>nìÊcªWÌKF6R=²qÑû† Á²É¦ñ`€ þººëÄ ÑØóy‰UG— nÕò»†PHØ‘U]Áë¢b°€2î#„ªôˆª¦É }‚ݵmj0µ&ÇPT,MäûVOüÝ!EʤžlI¯;Ÿ¸Üg@]ÝÅB´®|d O ¹,(ÖÈ# ÀÍuá æ‘ pï}U;fíË÷L|Ǿ¨VhÌ=&>Œèvýþœ¾Y±Fªü>–¬Tõ¶cšËgRSÍyJA†~ÿ‚ 4°×c‘¡€,l‰ø,øñX ¸åhX;JR¡Œ·§~ŒBÏ„"Ü¥ fF•D ¬¼Èþƒ¡ÔÈ5Xó,ÄíS{€±07ÔófÜ|¶«Òç04m17‡ø‘VY¦yù—¼;ÂH#ÞAk¦¥¬¡-µ¥c‚.Ad Á8%AÈaÈÚ–+·[yݲ—5¯ͨŸZ¢o MrŸxñ>óøV ø€÷(±)$–—¶›AB£à×ùXÜÌýž©$ëçn%Ããº&#RAŒ{E/¹:ú;¤¹¬“QjÜJiæ›3YCÞtÔ_4nÖäX%puˆÈ, tœCƒŽa$cuÊ’÷r^2 ŒAq§ªï|«Rû‡ºy˜QÃìlú7ì\Ô‡*¯ä¶‹”¦Éhtå6< bqÛ«Ê×—¼¶¯‚Ë~KÏMÛQIÐGi3c…ŸEØiºd FÕxÆL‰ÙÍѨÖî©…‹þêJ¿bIn”6É)Ct BÝìBÆ_˜Òy@Ç¡ +‘ê&tÚv²K¥Ho×¶æÙ…|ÓDÊDɉo¯¬u ñeê?ø)ŽnÑÑñ^£Ä'ÂÞv*|ÿ•€„y“R~25a ߟA6iáƒÜœ_ >æÓâ?$‘‚œÉ¦ÒÉL ú@<|Üv”øC ìÈî&º±¿º t¹KMª`¤# Y„rïþ¾~÷?9ÁLy endstream endobj 731 0 obj << /Type /Page /Contents 732 0 R /Resources 730 0 R /MediaBox [0 0 612 792] /Parent 702 0 R /Annots [ 729 0 R ] >> endobj 729 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 432.578 102.639 444.533] /A << /S /GoTo /D (concordance1\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work1) >> >> endobj 733 0 obj << /D [731 0 R /XYZ 96.768 705.06 null] >> endobj 734 0 obj << /D [731 0 R /XYZ 97.768 617.44 null] >> endobj 735 0 obj << /D [731 0 R /XYZ 97.768 601.223 null] >> endobj 736 0 obj << /D [731 0 R /XYZ 97.768 562.756 null] >> endobj 728 0 obj << /D [731 0 R /XYZ 101.643 436.065 null] >> endobj 730 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 740 0 obj << /Length 1156 /Filter /FlateDecode >> stream xÚ¥WKoã8 ¾÷Wd/'˯Øèv€Y`Ø=-°9,ÐÉÁ‰•Æ3‰Ýµå¦ý÷CJt,%JÚbO¦(>>ŠEÿ±¸™} ãQæeIŒ› B/ FI<÷ØX£g6»q;_ŠBbËqàåãV´j±©›ãvKõŠ6ËJÔ'Ê¢áDå\ó9oʼZóÉl¼\ü5òG.c^¦£³ï~ìß‘Ï?zÊuÏPqöEZ|iêe,kÊŠ’0NÀMæ^–Ŧ»CÛíÙ=8²@©wÅžç•ò|O±ă¿œV…²*ø‹!œ—‡riJ'JºSRc©doIÙK«}ña /0 ^­ ëm¹+ …`¢…uË  2s¬ßEiªöÔ–›!³šðï ËF&WÅ„‰1ÃÄð¥Þ‹š1жøàACýÙÿŸE™<å ¯„qJG.C_ #PÊïΠ~h¿½|’^†Óê³ñ'Ý»=}Õ§ÁË|)I}Yn…‰]ŸÁw‡Ä¨šBD* ×fW Ý+w–|œÜG[B¤Ýàoh¹§JŒ@l¨³@Á6NØÔé6––˜¯êg~­ÖwgCº¾P]úè'D‹æ¬ø¡oºOœšé`ßt>Öêö¯M^ý|äS”Ûç¶=¹¨>Á$2bta·ïIò|‡úŸ )6Bq CužDc^s«ÝÚ#¤îj“¼KV>MkQéç­[›\Û¼vÅâ·®ØÉÕ]×]%¢“¾÷9[#ÊL•Îûoxª{‡ýÀOá›zqFr œ-Â$c7H‚–rZ@â¡y#¦°˜'N+ê'•Ü‚É#WTÙªï])È@[îË]Þx€=J/$PVê+z·ùcÃÙÔ(+Þöcæ€sEHé†1æÎa €krW7BµGyY™§í±\ó£ð{ç~ùgQÓ \0ÕB¥wØ!ÖǬz{U4žƒÚåž“Ù]]=re ÔêËófW—däÜ&õz -_×=s…Q½*Z;Ùž­¸1yçUEͳþí}=/HÌúýçqœ888®ðÃÈ©jŒô€‹ÐÙæ­âŠ­>‘·®wݾj=KºAæù! „a8I•‡-aL4¹ÐKçq`]7Ž®¶Y"‰¼( {áÒ±aí^•Œ> Ì2øçë¿Ók=¸Ä7Êv^!SAÄÀ:ˆ`JOÁÀÄÙ±!*g-`#„,òX”˜$±`±ŒÎL¨ Hþ…ÑŠ%ؽ:ZÁˆ&ñéˆ@“JI.Ð÷çô¹0~ëC‰kÃÆEùjªËuѲï3rñ$š%Öw®¨»U?–NðŠ°©¶èS3³¢kiUà–‡iíЙT)õõ &˜¼#\ozƒ¶N êµySA¾çSmÐèð8lYýˆµâ»ú@:¢ä}>åäebtïr> •N£éAžG%\Ý76O¼9Ž™Ð)Ò”ZD¡àÍ×ÅÍ/à‡ñM endstream endobj 739 0 obj << /Type /Page /Contents 740 0 R /Resources 738 0 R /MediaBox [0 0 612 792] /Parent 702 0 R /Annots [ 737 0 R ] >> endobj 737 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 259.228 102.639 271.183] /A << /S /GoTo /D (concordance11) >> >> endobj 741 0 obj << /D [739 0 R /XYZ 96.768 705.06 null] >> endobj 727 0 obj << /D [739 0 R /XYZ 101.643 262.715 null] >> endobj 738 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 746 0 obj << /Length 1388 /Filter /FlateDecode >> stream xÚµËnÛFðž¯z)eWô.ßDêCZ8…‹" µ `-®,&ö®@®¬¦_ß™!MR”zÑ®æýÚ™áOËWgoÃx–ûy$³åz&ƒÐσY§~ˆe9»öJ³»½WóEÄž.Ua7¯çŸ–¿ÎÄl!¥ŸÇLöáâã{"Z™¶Á$Q_ÖÉÞNÒTÚÎÂ;©tùOGƒ¦II4šDœ³UµÙ77"_áGN Õ¶VjÀT4—Úª;U#£CG¹÷vÀzuñæ7äÚÛã,hú€éòÝòâ—‹+äCÜqÎÆÔ68ÆŠÈÆší³Üòyö~`mõ ä¤³Çc‹ePéÈÉÿv"Ep⪎Mkª•Y?q>k©)ì h‡‚'9צ~²±:¯é_õ#lS ==}²t\WŸX­ðÅËÄÇS"]XAèQqMÓ*tø³·2êu«õ†H¤GúY!'X¯L]z¥‚ÅÞÔ_&dæÂO¢¤å¨Xd¿G.òȃá³ùã]¯*ç­VvW뉊±´a9d¾#¬o¬ŸÊ¨0ÕrE†¡·­ ´ÑÌ[©¦©ô#o5ȦÐwª!ªJÓ~ø¾6GŽº¸îo„L§ŒÊ“° ™“DÞ=Yc¶¨8 ½;Cðumè&è°†Ëþ\±‰¸•ª'lËü0ïòv#„8´L _&Iú*ÝXU”ôǬ'”Ç‘Ÿ‡ÿ‡jË1>Ty{ͳÐãxi¥Ê>Uèí5Q™µ‚]÷Çgz`@LÒ®®§œVye¾Ñhª ÎüDÈVnáÊ*ó,Uaî5»ÛfUW[Kp×ÜõVè’`{¦~ÄBt­¼E‰pm…+]³êêÍm£jÈì¶2š•5eé^kŒíõˆ£õÓ‰ZOn˜Îê Ê ¹¹Gî7J+xU3¤uîJãqô-ÔU|Å%‘XäØ`B®f¶ççsnCn1«ž‚—J\¶¥èÀ6mЮ=óÞC]Bzfr]¡^†EƒqIƒV ý©U?sLÉ©4Å™)†p!m`€a@šÎQ¿oìvbVÂÑŠsïÒN.ö<9a™@Èh])RH¸F«Â®`C Ýö;ºž¸–rñmÈ*Z4_x³ùö’ر€.ã¾Ð&_ZsL¤p‚702žÝ™âïLñs;Ó@ä9õŠ<„‘8ž/œ‰Þn‹;$ 9áSób½ç³è•Ýð>ý$SÊ{äÉÙ‹¶VýºƒË©¹ŸA ;·&§h’úy>v{oßE÷ÅS{ìé¥f@­Uݰ.ÁOžÑ R¤êI¹§êÛùô¶]­Ÿ¢@_f×hê§ósy$Rä~¦ÏFÀ)jj\švÁß*³m?—ÌŽgØ ×ÜÅu·, l·- ËøGœ£P©ý@Œö°8ö3Û{(ýŒß@#Ý«‹å«ÿ˜qO endstream endobj 745 0 obj << /Type /Page /Contents 746 0 R /Resources 744 0 R /MediaBox [0 0 612 792] /Parent 750 0 R /Annots [ 742 0 R 743 0 R ] >> endobj 742 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 390.735 223.31 402.69] /A << /S /GoTo /D (concordance2\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work2) >> >> endobj 743 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 189.489 102.639 201.445] /A << /S /GoTo /D (concordance2\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work1) >> >> endobj 747 0 obj << /D [745 0 R /XYZ 96.768 705.06 null] >> endobj 748 0 obj << /D [745 0 R /XYZ 123.92 394.222 null] >> endobj 749 0 obj << /D [745 0 R /XYZ 101.643 192.976 null] >> endobj 744 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 753 0 obj << /Length 864 /Filter /FlateDecode >> stream xÚµVËr›0Ýç+èÆ“À #ñ0Œã,:ÓÌ´«.¼£,h,Ç`\ë´__„‘@;MVbîãœ{¯t>/o¦¶« ð§-×tà;PóÜ@tg¹ÒÂÛIöøÎt‘{»`Kõ…ÉÑü.Z~Ó,Í„.·Xç;øÃr­9ûLJ„}ß³%cËdÒz…a™†ìt±·ÎDòg4¬r˜>BG@b# ÑÌXFìˆÖ´-Øžœú©<î!k)Áå»ÕÇ™Lȉ„V4EJƒ”ò$– Ê&3I²~‘l蟗0d<£ ÔFœ1“›ÑßÃFä-FAHJ­­4xÚ$»•d€tŸiA!í“Û©Îþìðš÷‰`ªOU¾“uÛgÂᦿHqÝ V[ZC¨[m§.Éná6‚õƒ5Ðw¾è¶7AàÊ@q3"±tŽhÂ*Ö@sU˜ñ5Eyût4©µ„5ùÊ–xÏW¶Éóf°NÈBƒ£eûÚÚ°¶ª2b0L•ßúЂ¹˜+JÒ)곪&µ_ØÏ×VŒ¹”e…@Ö¶jÓî°¿ëݵ‡ Õ§ûñzìÊ£<[>±ãá€ù­ÔúWµ¹îÂÛÿ)âl{IÊ]¼D¿ËRî“Ú\¯¸æÉp'Ò]l6·KÍoÛÆz[)nmr»Á É~{%sJ6üŽøÛ\ÕlÅ¿Ž1IòŒc‰æ¿ñÏJÔö«¨‘~üo¼çÐ]¬‚/±ª&w\9ÔÇ6ÇæÑ}méœ#›^§løþ^äO¸Ix…c²)Çh•ÄHºH¸~HùSÑ!ôQ&DzV!(L£h±€}•‚Ä}Q”¤ÃzÄóDθq\€,4¢ê¾kb«…@MH§FŸ‚‹¥†§às{æs{Ÿ6_œö4äÇŒ„v¤ÎH§ Î6Pµ=°—^Ô!¦g[„&˜?O9/÷ËXÇôK•«–S.îÌýpåÂH°.!aHшРcþN)Á7¦$Žõ‡ˆ,èÇ9«ÿ¦>§x·mž¿ÎÝ]¡ãóÅ fú6pìÙ;é4‘\â„Ë-©:zëœãX{!z½¸H5;Õèz!¦ˆ‡w%¾ºa¯Œ<}ôÅ2CÚHn@Ó €ïû, ß«Þ|YÞüØNÐÚ endstream endobj 752 0 obj << /Type /Page /Contents 753 0 R /Resources 751 0 R /MediaBox [0 0 612 792] /Parent 750 0 R >> endobj 754 0 obj << /D [752 0 R /XYZ 96.768 705.06 null] >> endobj 751 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 757 0 obj << /Length 1040 /Filter /FlateDecode >> stream xÚ¥VKoã6¾çW¸iQøQÑz?šMíbƒ¶(ÐC}s}àJt¬D"'ÿ¾¤8²D™J¼íI9óÍ7ç×ÍÍúÁ f JB7œmö3'Qû³0ˆ+N6Ùl;ÏiF^Vàó{õ9âšP~·ØmþX?8þÀžYŽƒ’T¥ÈÌrm”¸Îô‘k813b‚Åè«Æ/_9-9IÇõPâ8J~½Tç³L-øôO$î T_à0§¼)óšÀ SN«òØpØ}ÆuŽi ËuGmèÊÙæ?v`ß©_»Õj%öCÀãX¸ 4…bŠ\¡$ ts'Ö”Ä̾3ªŠ¬$˜jqå'¾µwkרy­I³ªæî6ßéÒáTQ‰—­Ùá©°Ý*í”ø Ôäö¤ÿ/J26®F­·mTHy‘i îràŸ,FËi=×_Aöê¤W5×E¾ïËb üˆÉÒk+C¹Õgu5HY«gNÃI‘Þ€õ/öÿ¬8_wDõ -Jg‹–#mMÔ•ô@)K‡qûîlèÇÖP°.!¿Ãõ-á«>µì Sy2õ#Cü¥é ¬>7ª¬$£»é^Ù Ý¿ßuÃ÷ú§{É×3Üp¥ô`èX_jnO{ýâ¢u´nÌÒ—ö.v<‰¨ß-JNjÑRo¾dBçÊ^._kLŸ®¡<áò3c£b–êKk  Z¶ºîÒÆ·/ãeŸ)ж@o2CFÞ{VÝñi&ù€º ßÀ, °†Q.ß:¼vJ¹fÞH«†ò­?êCârb~Ð<Þ¬U]]ã¡uqîÚ±øÆ(H@î/*:µÇó3®V§:§OlçL}y¾pæbÏ™gr#}Œ=©#–V5a? ’/ç—Wµ- ,ŠB©0©N¸ú‘<=.,7žW9]ˆ0èÎÚa¨5KXѪ•ψB©ö#Iùv!Õô%¾¯‡å7Ò²ò<å˜çƒ¦Ø(±àЮêªá9U0†(‹)6ŒÜ’D<GA÷’±¦~þTQ¡¬Ö /Ÿ1ƒ|n¨ ð^ù»(²=½Æ¶Ô›×ißД畬^[Œ…eSàŸÔI†¹XJTÇ‘ï'£Ç\ ± Ìš/Œpø¡áÔô¨·xÁ›º SQ¬;ò©­˜!ûóô€R±/ù«žÐ6èï~&Ú+›VÇWh5{ý{°%8‚¡MdU*šåX:¥ÍúСßbE^ŽbRGYÅÙýÃÇ?ÿþÜEeÜÊíÖÙít(Ì5.qnË*#Ú×âÿvãH¢“ïÛ®•3–ÓG¹”ÉÕ§Nê’]VÁ-ãµ½íS9lN¢Èõæ&(Ž¡øãH Þ|ÞÜü m“A endstream endobj 756 0 obj << /Type /Page /Contents 757 0 R /Resources 755 0 R /MediaBox [0 0 612 792] /Parent 750 0 R >> endobj 758 0 obj << /D [756 0 R /XYZ 96.768 705.06 null] >> endobj 755 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 761 0 obj << /Length 960 /Filter /FlateDecode >> stream xÚ­WMoÛ8½çWxµEìÚ´$[²´XØÉ¡Z q…WF¢ciôGþ})’Ö—%%z1%jæ çÍ<’þ¼ºšÝϽQBßõG«ÍÈ :‹‘ï-+¿¬ÒÑúaŽn¦Ë¹g_ Ärþ¿íÙÊò=†“›©çz× h™×T>GÅ´tnâÕ—ÑÔ_‚0\Œ¦ŽÄ÷ n®­ÿê . sƒÂ_8ØA†ˆ(f ¨În­ŠU5Á ¬6V+×ܹLUül:ç4E “L*u‹!¾£„#ëb)¾FÊ6Ú²ðú##[Ä2¡ûy†xܳƒUÏfv/ ¯•`î0%°Üh“F•dDØ~Ðd¡:ã€ìs¯™þ3ûã"ˆ±fíæÅ³ÃýëÌgdOÞöëíß©%‚K8׎ç²èÉ[X ìrO•ú¸××2•F)ƒô°möÕ\I~ƒ/mÞU Ê«,@Ôo`¾¦)Žæpƅ橤":“XΔ‚‚GŠQ3C"2Q¹ú.Xú‹Î–‹Ìh\©ðè*]¯ÂÎy%È(¾ÅZ¥Ý›. ÕYUúU7=uíÊćžÐÕôå¶Dqz‹!×2QáÇ­]jÜÕŽEqÞwÐ*Kà/$ ‹¹¶ ‚Âðênuõ íb‰ endstream endobj 760 0 obj << /Type /Page /Contents 761 0 R /Resources 759 0 R /MediaBox [0 0 612 792] /Parent 750 0 R >> endobj 762 0 obj << /D [760 0 R /XYZ 96.768 705.06 null] >> endobj 759 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 766 0 obj << /Length 1668 /Filter /FlateDecode >> stream xÚÍËnã6ð¾_á"1Wõ\4‡6í[ôÖÜ’‹¶…µ)¯D%Î~}g8Ô3’³›S/æpDÎ{†3þãöÃÇÏ"X$, ½pq»Yp×cI’, b|¹Íwα̕fU]>Ýj]”YªÖr¹ ¼ÀùmEë¦VkêÞ ÜÓác°åˇۿ?~æ~‹ÇFþÂ%ú:2dÅ“˜E€³$°’ädðK^1Uï÷¯÷CAN¿®Só…Ïðæ>gÀÿ.½Øy!¸'ÄÒ‹œÓqGÛ²¨u®dÅll|hƒß÷zWÔ[¸ <ßyF¢’à´´À“A–/´[~Sïi£ ‹,¥Ñ á\áy-·²¬,%•},J‚³¢'Ù R€˜nëƒT¨…®ú<„ðòê -•Î3QÈžËrR ÊU‹! =ËâX‚ rò â*v:†4ªtY?}Z®|éž ÿ¾‚9öÚÞPU]!ÏÈ‘*Í/\ì[àbáØ ’+W[+HÑ Ô×Á¦™„»d'(‘µ®„²¾n‚!™nAßýšn ?8TkˆÜÐbÂn¤®Q}eo¶Ÿ–AžÈ©Ï@ F;®˜\6¯x/k„ﲈ‹&¿o&R/aA6î]wŠLÂü¤}5ÿ|œ1ƒ¢ÑÑhS| ‹»Îæú5 ,D-Ïš }ˆ¶|Ƹ“¶Eã¡7 ,c2…‡Ck. ,©,-3òr&Ÿr[𣱬3–ã~È?>o9?IÞ¶~Ær&:ç˦Ñ‹ B]„À†ðÂebe/´ª>çzG)3ûF9ÑB¡´¯¥%Ÿg’ÊCµìÊ–¥Êx - „Ô„¢Pñ½$™«¯)Ј‚özvÂèFçÌÅ=æû­ÍŸs£]â;&Á £µÝR¹„ÍwYbâùFµ&Y#X¸ k4¡©6dDø zÉ·ŠÍ”â<†pŒÎ wAêøHC"|S÷–Ê:U¯tø¦!N µ¶ºæ¶*Ù—*sFjÕyÔ¥”S<"wÏÔU°GÅC½§m>³b {FkèTÒ*YæÕ×Fß®ûÇŽßgÇdx¡bág¸nØJµ.ÇÜ/õËQRËMC‘.K4ýåœ8q u÷ qˆ‰h<ƒâ³Ø· d'Ôݰ‚­,Â{˜¸ëÓTU”f0¨Uþ­6}?šv0°Ž((oH¡uëÓ«;a3XdrÄØDÞRÞºƒæl½køß†NDõÁ*H ÿÇpÁ‰i(œéo‘ÊMo€ãmš qyõ@¯&„¡ˆG¾‡î•Úk¤±eDé†9÷Õi˜¬°MÇãFÍ<;6Û™¿{ð¿•@ø“s× |FVïfð÷ô…II¨ó³(¦ÍàlLÓŸ#GaܱJÇ™Èî"yà¼+1Ù–¢.Ï’äÿ—äÞ—oæÅ¨>Tíü@]މgœàÁÝ~øÔW` endstream endobj 765 0 obj << /Type /Page /Contents 766 0 R /Resources 764 0 R /MediaBox [0 0 612 792] /Parent 750 0 R /Annots [ 763 0 R ] >> endobj 763 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 330.959 128.791 342.914] /A << /S /GoTo /D (btree1) >> >> endobj 767 0 obj << /D [765 0 R /XYZ 96.768 705.06 null] >> endobj 703 0 obj << /D [765 0 R /XYZ 127.794 334.446 null] >> endobj 764 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 770 0 obj << /Length 1225 /Filter /FlateDecode >> stream xÚWKoã6¾çW.°[›+Rïź‡.ì½5·ÔE¢m"²èŠ´ûã;äP‘dËÙl ©á¼é?îo>ÞÑ$#YÌâÉýzBãˆ$q0‰Ãˆ°tr_N<ò%¯ªüÈÿRȺM™×góÙ"b‘§´Øñùluÿç$ $HÙdA)É¢esEJyx¬¸‘?i Enÿ‚QÔšoxc8E]òçwòV¼Þè­Ù¬È¥X<&¦d£‰Òr¿ ½Ó’ÉídÌŸûxGÃ^!1$ð᪠2²}fWŠ·Y•{cqú­>æ•(õÐ|ã—~Ù;îµlpÓ«ÏôšoÌ'£ïò$£YÄM¯vB)QolI¹Ølµµe¿ǵ+JDZ\ú­S“1/å?/pm¸Í%þü\ñs¿ãqA›|ÃÉN–¶_:G[SlÏi[}ó›!+Ýä:+䎣òYÍHœ„`ÒÚZT`A³”$Q0Œ©‡ú,I¥´D¡KÓ³[{õ™ˆ~„VÅ]ËHüð¡m*sÀVm,=b´ú¼ðÇæ¦ëä–qè1ûUýÛèÎl´ºpÑiªóW¯Œ—õ+옼6¾žÎÛº uNÒ‚—¤êiH16ÅñÏBBÃdèÁTéÒ:°0 ½:X”FÄÓÿ7õ# ±7û : ÃŠbë S ¶Î …³îÉöÁÅ+~‡²:w˜~> ¿ûPj Þ/Jî^á¬ój—¿àf›ݱk8ù¨Æ|iø]®aæß÷‰+eÝÈÓ²wX“ˆëx^Èj]‹ûAíåƒ[QãJ?©½Ò4ŒI½=Õ,";ë“'Î÷ÃO[QôЉË%®èǃX]›l+1_]GŠcs>/¹_Þ­ZܰÔÕ=ôæ}ÂÞŸôý›_DZ6sÑ!Äœ®~s; EÎÃv}Õ7¸Ϊ\ŠÝà¤Jw…ž»V~È‹4 ÎÒñà9+ßæ\/¤5 ~wp®~î…À2èp„a€F²ÐyAʠýÛgèç€z¼Ðž/AyÛçË,f̆ç…IÓõÁ›]Ey˜Q'™xyí”æýÄo8ž‚k;˜‡ Ë,ØõU©ÎùöñmÉÞ\æ;IÖå{ÇõVÚL—bß{´¼²=q2~ĉ÷xÐxª·ÖØáJ½Zº£¼:Ò‹;Ô[ˆšãþ$WŽƒ’¢àŸ ¸(–N¾æO‡ ÚµvJkç€L)´5ÎÔÃ„Ü ãì2Á†hn0{¾oyãX[ïðÞÀ‘Ï\.€ 9L‡Xή*VMãôîd»Sr‰Ò¢ÈmÿôÕcr Å4¦YKnúÂò0…?I“ÈF™f F—Ù'åÍíýͦÙG endstream endobj 769 0 obj << /Type /Page /Contents 770 0 R /Resources 768 0 R /MediaBox [0 0 612 792] /Parent 750 0 R >> endobj 771 0 obj << /D [769 0 R /XYZ 96.768 705.06 null] >> endobj 38 0 obj << /D [769 0 R /XYZ 97.768 197.39 null] >> endobj 768 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 778 0 obj << /Length 2098 /Filter /FlateDecode >> stream xÚ¥XKä¸ ¾Ï¯hô¥]À”ÖòÛƒíÃfЃÅb‘éÚS&@Ô¶ª­ÄeWüèžþ÷!EÊe»<;›Ù“-‰&Å×GÒ9¼ûáSv“‹< ’›Ãñ&OEšd7IœŠ ŒoåÍ?¼B7}Û™æYìö©Ì½C¥wû(ˆ=Ó”æÅ”£ªi]ŒÝË.È<ÝÓZuLXêîÅÐQI;Ú •îèýص'z;ïö@ÓžÇZ ¦mh³S³ÔS½“¬¤Ç"ú±@¶ ä]à½ûçá—ÿf/¥ÈcVä·ÇÝ>”©§šf¤Þ¼ðê¡Ì¼Ú5¢Twéfv·âã¿Ú›¹Ï>êŨ žÛŽ~Öª*Ú|à›~0sm»ÏZuÏš?J–VˆbâÑU+È—(“ÒÎB*~öc-ÆÞC{Ø”N#Ï·úú<”®àvšÏ®Ôiظ‹|\jØh]B$màp„ð»Ê·R u†úõŠp¬7ÁñCÏ»àå^¼h”PT ¯”“I ü$Ÿp«Ÿ›y«(0=݆Cª;©­"#…Ì‚ "r=OÝøXz Þ`‰ÉÄÔÖLe’N¯ Ý®õTܵC5›µ­Æ¯bK¾ß®46UÉÃÔ©ò“U%_¨’;X²9EäTàé’:wý›¡RÔÓ&aBIµ r¼â¯Zzíé<Ì€ã._Æâ\xŸ°¡qÂ'Ê Î"ìSá‡KH ˜ —›¬Üät)=íW =wís§N´;ö¶œÀ6Ÿ§ ˜Æk $mÌGæñF*Iõèx˜£“eÜK?™¾·mö°Í”/€ñ5åå EТl;n VÉØËhÝ  :6?] R²Üì|,û£4XW]ÜÄ ´uŽ)R®YxV¨ºà@`É‘ÞǪ5…5_ÒP̧<€æ2 ç ko‹wÅžLöž¢”“~¼´ôP«¾Ù¿írfQ´‚w;—,Ò¼h+èÉ9“ÇnºÞ·m>µúÈ€±Î•—=×—k8ÏEM¹ü·ºt¨=õØ.11`”«6š;í†ÍÈ>´‚P"N«ø›é¾©_gD¿ÓF¹Hó ®y>J®ÇBÜdü„7‡Ÿ2¡ ÃsìyÃB>Ž|ü‘>‚šMa!%t­ŠùºÉßg™ÕEîLbËm‘+“Å@‘®Ìò`Y.&Ùù÷F­p>jEó!ã¦ji´aá qž8 ?Z¢eÅäÅñÓ…í8<3 Ö÷D¦Å³à “Xñ MñJV ?οS&€6D¥‘å¶R+ƒüþnQ¦1}õGEAg–®ßç tÊS‘DÂA.²,#ºÜ†Ä»‡Ã»ÿª¯R@ endstream endobj 777 0 obj << /Type /Page /Contents 778 0 R /Resources 776 0 R /MediaBox [0 0 612 792] /Parent 784 0 R /Annots [ 772 0 R 773 0 R 774 0 R 775 0 R ] >> endobj 772 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 163.587 202.389 175.542] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 773 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [206.983 151.631 212.85 163.587] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 774 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 139.676 207.619 151.631] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format2) >> >> endobj 775 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 127.721 207.619 139.676] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 779 0 obj << /D [777 0 R /XYZ 96.768 705.06 null] >> endobj 780 0 obj << /D [777 0 R /XYZ 123.92 167.073 null] >> endobj 781 0 obj << /D [777 0 R /XYZ 123.92 155.118 null] >> endobj 782 0 obj << /D [777 0 R /XYZ 123.92 143.163 null] >> endobj 783 0 obj << /D [777 0 R /XYZ 123.92 131.208 null] >> endobj 776 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F30 111 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 793 0 obj << /Length 2546 /Filter /FlateDecode >> stream xÚYÉn䯾ÏS“ƒ¨dšæ¾ÀÑÁ<€“œb9ŒH‰¬V3Ã¥]Ej¹äÙóoÅ¥›rºÖ_¾¢þòðî»ÅMé—Y”Ý<oÂ0òó0ºÉÒÜâô桾ùä=œôÝ!ï× LŒyrÔ/EŠà@™§©Ó|Åò’Ñ­~¾K3O %¾1¸›Ì ޵ TœFR[`w²ÚÜ w²ô˜KÕ…¡Xª–÷€ /bÁ¥…é4M½CôZ^¢XÇ3ȇ|¾W;>j£ûŠÊ\æÍv‚‚Ÿ€Œ§ÆòN«U-C.qÖÍ8N½³2cS± /sqo¬ª¨jœÏÅ0s5ó¸rš™É=Œ_6½; ¥S¶Å20Bí÷RI'Šs 6 Ü×O<O !&¹Ô¬•OóÀûxWÄà_Þlzì…•ÜïüÒ7T.püó,l ÏÈX* ¸Xk5ž^BŠ,p]Ÿ/´“‚ïOö:fÓÔÏ¢¹5À©»2®\§Š¹­MNAˆ}MZž= ´=/?²)¯¥ (i²´¡®Sæ 3npI¬m£ ‹Ä/â‹NX$¿ÖG±c¢ž¤ ß³¬ó¼ik…Ö~mä2h5{5\¥ˆÐ/šAêÂ\>j· ®â)3L#<ÞQU¥Ï8!øòÆ“©o~›Ü©ÇG£ŸEo'á¸Â $ˆ*egÏXÝÛfl8¾ÿŸñÞˆ&—›„ƒc`Þ캄 {6Y<_z6Ÿ=»gÏ_0—Êì‚LǦå)™5*¡¼ BpÉèJó!7n,XDË ìLT![Å ‰Â5î©yiv Nz l-dO;u>^ÁÉ#gúÒp´F0Û/R~µòoª[ÉåÓœ®æ&Òq°`8§Î¶›ž‘$«²ÉÀes§”~’…[sKÁÈåÎý‡Ë^yð'þ™yâI¨÷ð÷Ç8Ký(ý0߸‡?†Kuü$Ýêÿtàμùõ¸ «CQš¹@F„º7üùp!Àý^ÝÛŠ,÷Ëò¢êý„–N±7SîÀHr'-İ)ä"°Ew Y±t#j¨ßoÒsTìôèÀÏ‹¹„U‚ "Q<—‡#,,ŠÃDœõŒZj#H#B‘¢b¾¡˜ Ó¼Îèw0µ6[úØ&ÎÊZž'h O§÷ræ$|»¤Ø¼«o¢Ã|·ÃSSµ³ õþÉ¡QäB몈ó‰±I-H ¢\Eª¢’¯çO)ðeK܇›Ñ‘í4UQ±–bøIBÆ•èQnÝŠyFzÜ#±žÉ÷úe/|(n©¨ pYˆÄEOà©zó@>(<éa(>Ê3à ¹_²8cY »YÄRZÆîáTd¨×|£qÄ%vO¤CÂI²%l,Þ”‡ úi/a*” yª0d,¿±4nÈÃ#dfsÍiÄ«°Þ@I½©"ËY¾#ä¿ 3.,0Z’‡NW­‚‡Ñ÷0‰ª…IH À2|( ÍÍkRÙ-O\@‘.ႹþÁsÉ2mzzw- ÊÊùWª‘ýš ÓjKñÔ) 87ÐÍé[Ð!§Ñ¼ÀÔ|Øð›ÂÈ2zájc8©³½–ÈŽzÌù&?vðÊcÛŒŒî¡MùÌ‹±„»K8úé–kÏTK%–Xrö¨Ûñò …(·˜†8ti(;Lˆh`BöG 'gîÁ(ØåB#´É3—…-Ö)†&d>€Y4[=þ/Ûæx§øãƒØ•,Ä‘Ê~±¾àˆŽZˆ¢ˆÊ bp0g­¹ñ¢ªkËçDüK4p( çaæ]7‘(öÃ8›ßEî›$ˆ ŽŒ’”Z.òЯª;·"k8µ5O]ƒÙ‘!Îý4ŒÖ2Ü·d1ò¹†ZWHjGp°`CGÙwýtNÊ­èÝ3>Í—Â" 5•ªºƒuÝHço 0¢¯Ø8àÐÊÃ4WPjì×6ˆâЗï¬_ñh“—‹„tUpH-r æ …v[ŠÉ îŒà_¤ ]ðaY®ˆÓ¯;1Kü È×Í+ññ½•ÍaD-Ý“³\Ь #°.A…Í> W0Åá§ö—Ì(ŸµÈÍ}«;C3}DLJÕ‡Ÿ›N+ÇÖ½´/ĜΖëV+h¤ˆýlþ˜î¦=Jg…]gdÌKçGÃRë§e3?…ëN Xýce'›J¿(?¡³4ßKh,ø)Å –œþBK3>ÄEt‰<a û:µFÚâÜi'x¿cÐ?ñ™füž‡Îôëù”©yIVÎg3œ©·KÖ~R^X@¿ŽF1ظ|W"TYL¼À”QwŒ#À< Êz= úȵçÌN7¸¯nIé^EÇ@{¹¿ÂmFÿ65F×+°… FADÿ+ùEQ°‚e„{ï~|x÷_öø5¢ endstream endobj 792 0 obj << /Type /Page /Contents 793 0 R /Resources 791 0 R /MediaBox [0 0 612 792] /Parent 784 0 R /Annots [ 789 0 R 790 0 R ] >> endobj 789 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 622.437 102.639 634.392] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 790 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 622.437 176.237 634.392] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 794 0 obj << /D [792 0 R /XYZ 96.768 705.06 null] >> endobj 785 0 obj << /D [792 0 R /XYZ 101.643 625.924 null] >> endobj 791 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 803 0 obj << /Length 1360 /Filter /FlateDecode >> stream xÚ¥XIÛ6¾Ï¯ð8 ,cJÖÖÖÚ Z=n/î86m+Ñ‘rf€üø>nÚL{šä2”žÈ·~ïãóüº¹{û>ž$( ½p²9L’Ea< ƒy~0Ùì'[§&œrò˜Ñ¹ë'¾sžCêTH˜ñ-ÔÓŽè‡Ç¹ëÅŽ>RS^§ô<’½ê2O+qX‰òRžÙÓÌl!9Es7ôVÎ6=lþxûœê¼õ— Š“p²TŽVÏ”ÔLíëåÈ7{ê²ái¡-6¬87qýU€0¨t1FI 3ÀÀØçN“J=‘úØä¶9üG!‚õ”2õìУü;\ ˃ZUÀð°3«Wsø˜ªLZÌ¥éb’·Úå¥Z…¡ !›2kCóR´uUÐ\¼€IÒ…ZÓ¼4|¥îïÀ\gýrÛï‰ží£ ¿6ÒŒ™hÿ'’Çã/x⛃ÅáG– ÓÚGÆSÞÈÆm Û 6p6íÀUëÝ7_¾è´Žàp¿¾‡[¶3=ý½RM÷=+Z±]øÕ»Â¦ûŠfIâŸø0ìŸ_FÜ¢;ù±Ñø.J~5“üþ½¢Õ¾;ÑÝ'PÂ[¨ÖÔ¨ F¿FX–Ñ#ÑDÞÙÓ¤ùE«GKá !žíHK£?ŠÑ¶¥¿´mÓš&²Xæü{}>´ Xºÿ‚{äœ& e‘Ƶ˜îÓ<í±õòz—J H3L­VU»ò©:Moµ,\ŸQàßnÙ(‚– 3¯tåÿê±Ô4+Sœ)kv'“X`su˜ú©ÎÊdžñQº”•‡kµÖÙMÕ¦®¤ž°Fmø ³?…(§¨¦¬* 6,¨bè‰ü~Òm'€!‘´£µôc‹Þõ}y{ç„ïäð#2 Ä­‹ttR:šÔeõþÁ]6ìÞ¶4-%\I¹e""ųn™‚¨+?oï;͆£ÉèŸa Sʦ儫=üÃtѹr™™• ×nNL–àîÅŸÎޤś^’.ŒÞ@”·[ñ ­û0ªŽZ“—ˆì÷C4h§¶_ïáƒõ—+d-T„ñn–ëÞ¤3m×÷^­Ö³µ¼fµdWf©]ϳkô×Ϻ(X3ºì;$[îÃá¼Ùú85Û‰ Ïp¨ÐØR¨.þç£~Kû€˜nâX¹šøbãÝo›»ÿ4.·¶ endstream endobj 802 0 obj << /Type /Page /Contents 803 0 R /Resources 801 0 R /MediaBox [0 0 612 792] /Parent 784 0 R /Annots [ 796 0 R 797 0 R 798 0 R 799 0 R 800 0 R ] >> endobj 796 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 612.57 102.639 624.525] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 797 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 612.57 176.237 624.525] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 798 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 564.749 207.619 576.704] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap2) >> >> endobj 799 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 522.242 102.639 534.197] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap1) >> >> endobj 800 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 522.242 202.389 534.197] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap3) >> >> endobj 804 0 obj << /D [802 0 R /XYZ 96.768 705.06 null] >> endobj 795 0 obj << /D [802 0 R /XYZ 101.643 616.056 null] >> endobj 805 0 obj << /D [802 0 R /XYZ 102.999 568.236 null] >> endobj 806 0 obj << /D [802 0 R /XYZ 101.643 525.729 null] >> endobj 801 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 815 0 obj << /Length 1950 /Filter /FlateDecode >> stream xÚËrä6î>_áx·buí´FïGU¼‡Ífª&—ÒUIÕdt‹Ý­DvDÉÿ}TKjy<³› ˆgÿg÷æÝû8½)ý2‹²›Ýá&Œb¿Œn²4÷#8ØU7=Ù¹ù´ûùÝû0™F™¦ÑM@D¢˜1Û¦À$¼Ù†¡_¦Ì¬—í9ÚlÓ(õ~ØÒZɳèŒü#Hƒn/š€¹ëÙ§öpþéøa/?÷÷gaz»»ÝÛÿo‰È^ÉðíÛ¦6ýò4t§x5þÀÇá§O+Ê/D;Zy·QÆ WŽÞ½/¦vó((,}R2Ýî$7Û$μƒîÚMTxC#ñ°ÙÂV6±O„« ­<Í6ô*ÚiÕlBï7¹W+Bö'ÀI‘-V¯ øI[]ÉÆ?t¢•×:d¡_ä…#n–«Tf褻WôrªY^< Ã6+Àòhn4RYƒèÛ$ ˜A.l'BUtÂ$׉^öâ¡á/7i扮FŒñ7Û,ϼß6Eìñy¥Õ]O_>!g¡6Qî1ƪD-¬€­êÃóâ6ÝÕÇZ‰æJ¼·kljµ‡ã  Ëà >£ÝÃËõ„Çëq¿ï$(Ÿ“‹Åp¶aºçøýú½ÂÈ[Ñwõgâg ˆ—X•ÓIs¶êje55¢Oï·säJ«­~@mk=böàTÐVÕc+ò‡5;@CšH—i²OJOTUÝ×ZÑβC,îÓü.î´j î‰ÀšWË('Oƒ”‘Õ`%bˆá8›Çì:2?JbGÛ顯ËþM¢Òýh&ÓƒµE·Áx%ß/sðý’UÏ8Ý‚µSO`$‚U/oK‹5,¬ø† RŽ.­fgVýœ¦ÿ¤íp¦õ* $÷á@g‚–YjAÉjãX:©¥ÒÑåž-gT Ó+ëG4·d%¶¯éáËØ²xnž|ÇÀãÄqì$%Š8áT¸Nh SFG¤5¬ÜR´5èBÁó4Á‰u"\é5±Ò ‚{½Ô~^ä$扟+›ÄP ƲàQ~>oì‡ó¶kÅy…i™N?©W\`ÆóžÍnQ&s³ýƒJ÷œ#—Ê®;Ñ=Ó˜Üw?AP4™+¦ýðÄŒ. Ú×EM\º¸’Oà³xø€4¯ö7õ>@ÂFªcBˆX’áé¿i —…aÃ(õÓ¨øb[‰Ÿù"ãÚXtWÜOÀÄëw²kQ'×;íÁßÄÙÈûÛ[¢†yá'pÁ,¿ÌH³åçwÿº›p5òì0ä3Ž,zûO~Û…áÏxÛ!"Ì\¥r%£Ÿ©ò‚Mƒ¯ï覑‚ &É\û“Ùf­#Ȧ4€r›3,tØQÅ\`Öñ±Š$†¯["´´ÓÐÏŠ?m7¡VëYìg—ô¾Z}Í•ûFª-nðÁ¯îïïÆýÝÜÔB=O$Üw÷á4ÿ½jÿÙǯXýZd–@ô}7¦_gmÝU²»›”»¹` ŒK÷öWŠ8W»•Ò£ë>:Q½Ä¶m55 pùDôÙ¯8s_·òKu#„qº¸ô±«µ¸(ÀDÙË÷±­ìU?ãeóhý  jîg”ë·uoý5H¨»Õí·¹ñ¨¢{‘èeGY&çÌ݋ªì‚Cr0,ýƒœ7h0FC»‡ÏÀ-:Íí ¯ðÍ¿Ê,kxœò7ÎÆ…›ÓùhAúÁÐÆX›±öeé¥öÁužqL 7°–ö™ÍèH• }Úž;B,Ö5ix;üiCÓÁƒ…˜j¯»Jð[ž¹¿Lô^·ç¡·§] 'UÂ#¬<åѦҊ!h­b…ôD;šÖN°NIq± mku!]ib ³ˆ£1ˆÍõDXúYX:¢Åˆšð“$ÕziǦ u°¶‰CÈZÉ"$­b*•‘U W«ƒªÿ~¦…üJ“±¿˜„Þî„%IU1„ó)®Â,’qö= uÕÓšBmc~­©ËÂñZãË.Ö'~R¯›9ÈÂu3#ãÑýqãæ`Å~áïÊþòáEA,Êßü´{ó7‘Ö- endstream endobj 814 0 obj << /Type /Page /Contents 815 0 R /Resources 813 0 R /MediaBox [0 0 612 792] /Parent 784 0 R /Annots [ 809 0 R 810 0 R 811 0 R 812 0 R ] >> endobj 809 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 494.915 102.639 506.871] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap2) >> >> endobj 810 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 494.915 202.389 506.871] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap4) >> >> endobj 811 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 361.701 102.639 373.656] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 812 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 361.701 176.237 373.656] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) >> >> endobj 816 0 obj << /D [814 0 R /XYZ 96.768 705.06 null] >> endobj 808 0 obj << /D [814 0 R /XYZ 101.643 498.402 null] >> endobj 807 0 obj << /D [814 0 R /XYZ 101.643 365.188 null] >> endobj 813 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 822 0 obj << /Length 1689 /Filter /FlateDecode >> stream xڵɮÛ6ðž¯p\ –˜Ñ¾y— Ðrh A’ŸEÛDdÊÑòì—¯ï ‡ÔbË/MÚ\lj8œ}áð×õ³WoÓYƲØgëí,KX§³8J˜D³u>ûàœ–~êˆå*ðç$‹W¾S·÷µhš—ˆrR´Ó”-Uñh û²6–Qìð¢5}óÆ¢Z•³å* Rçí2"9ý—+àTÛ‚7²4 +Þê ¿/ ®r³”ßä¢ÕòÓú™;[yË"£áA€„š2ž ì\Ȭ<§¦O^íÚƒP¨mC©wB§_ZY è¿zë…£"·„eA@Üö„Ä”€¥Iˆ¡n«q>®À¾íq‚^œ±Ä-¶4ä¢ rÂVy•øÌý±âï—«Èœ×+ú?”¹(˜87ß4ÝÈ=¼¤E%êc©j±¨7eBU² b²fª- $óþÖ)¹%\DB ïBáÈgX}¶ ƒƒYú·©d]KµÃ%ú´FN7¸ynÈb7y’]°8o³å1*u<â÷ð£ã;ó]–ùVèØ8¿)xb.$DaE$9ýiÉ „"±¡¯rŒg½eAÿJˆ\äóZ¯F•vSºkDoŒ( ËM‰%Í0HðéÌøav×äü˜y‘÷¤ã"¨kcƒ_¸íÀ›Jž»¨½a­,d^?É) ˜ïGO†È_è}~è/Œ'›¦Ò[6›Ç#eßÝÝ¢’»}³è]dÞû/½OSéÖ{©‹²ß‹Bìx17X¦o…Ëznh®µŠ¯ÝÎÒp=ÖüØÈ#Õ¶,ŠòDëö؇W'è4ß_ ¼{®{«&¨c)USåhÄá8¶l«ä—VL—´ẍl…‘;A§_HÕÑ2~>ðó%¨j×ìï"öÅãBËï“ø{ÊäíÜ®ËJ÷ ä}‹Ã¬§iôÒn:^öÎø€?¯‰:ù>Y!ÿ‡Põ–yÞµ]ž £o ùØ2»è¦¼Ùìêñ¡›öù¯õ¬tm¯½±ñ^B>Urc“<[Þí#ÊH]y×mLo©¿-C«lJ•K¼“ñ⺚MnNÙl>Àœ¯¥üoZÊû×Û² }œOe}_Y'Oîùç¶J¨ùO*¥cOì).œðürcRT öyË Vÿ,iûË¢•­O‘çwÖÎ×W%ïÆUéÉk½™¨ºHmY`š°Õ²u=´ÎÝ‚¤Y áp`B|757|Ÿ[ÓøB±;ë9!YÅ‘`41¤4&Áÿfé'Î^lp¨øL ²m Êœîþˆ³Q”álÒl›œ¯g#šÙi™Ù–ÜÈ$ ›rkDëAµjóHÔ Y†cGjAµ¼‰ótñNøŸ—JÐê±iZ\GIæ²4ìf*–š×†NÝO» oÕ•„æàG‰³ƒU;âÆéO+çå|Ô–+(,_1´Èç{) æ%! ý̲úóZ*%f=cæzo`e#L&ÁÅSÔœæH ‡(4få8y‰ÊÀš¶R(¯^c¸D³„²ªkäÆƒ8N¿éiüLœk(úÅ„¦~²4ŠŸÐ4eIÔ ^føª÷š¨±=˜ Z$(xT!NfþG5µpô PIpwmîl!LTn65^ 4n†§ˆvŽß©Îb„‹3X_ðC' .j~0+²³ EÂÁ\u"@´Õlñ-žhª;©°­âWguüè­~J\Ïùß2ÄTCƒ¾ŒK a\ô–4Ó#«{Bè"ŒÎ•ôÏɲXdN JPQà Ì,Iµ!X¶*ï¶+ƒ WÒ€*öà›¢…³“/)u.µ/$xu ˆ“os˜"Îk °>Õ%­p8ÅU¢ÕAH[¤éòáe>óºÐ=l¯c×ñ㧤‹…ÁÒל;³zŽ‘¼Dkœdm¤9I‰€+í=3tÄ,‰“‹‡¥Þ“ÛJG¥~3")DÁ£ŸÈ‚Ð> endobj 819 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 623.861 102.639 635.816] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 823 0 obj << /D [821 0 R /XYZ 96.768 705.06 null] >> endobj 818 0 obj << /D [821 0 R /XYZ 101.643 627.347 null] >> endobj 820 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 830 0 obj << /Length 1402 /Filter /FlateDecode >> stream xÚÍWKoã6¾çWx³"1WÔ[EsØ.6ÀE©÷¤902DC¤çßw†¤lI‘¼[ ‡žD‡3Ãy~úmyñé.›å$O‚d¶\Ïò”¤I6Kâ”a<[®fÞß~ìë Ÿ/‚,ò¶øñävW2=§žµ¥7L;ÍžK·Ê~Ÿ™â+»´ü¡Çìv3i»ǯXµ-9è£d¾ˆàän¬²±§îʉwvSIø¦ÞŠ—£fÜÀ:‡›ÎHåÎÖ ’zÕnÂ;æOËßgþlA)Éc÷f´< C¯Ø5 ¯‘Q—o–$ëvÅÐý<Ž=&Jûj$£t³0N ÃÈûºâ wÄŠë4N\ÔüéŽF€O¢<²vl,K˜tXB’¥10µköü°]²Úîàíï%f>¡YÔò '0žxkܱÈ’ú¡sJbäž5à—8ˆ½_öË´n0?–¼©Ô¥]kØp/Õ5sÌÁ-o³bšÙe!kÍDíßát ô­£WLQ¿ ¸0î6ñFtu”´v#…¬Va¶ù|ÏJü6+K÷Š -wk¥P©>«;VhÙ¨Ûåý÷¯S/k{  EŽ– m âE!û¡]8Ö#ñZÐÐ'iœSÖ&÷™hçl¬%y˜0%HIpΔ’0 úyòÐ7¢á[ÔS±ƒé$¢âªãÛzB÷Àâ—1åP)y@ÁÖ¼Úö 093&ÀÍ w¦"z?xj¸ºÿm†Bqë%O½áЪ—8mò°Ý\ŸI£ÀIžÍBâ'ƒèåÑÝç?þM$S÷÷´¦^ishÚÉØ«¡Ô÷{oþSïJS£½E®×ŠëãÃÏ´—(ÙJK“|—_X]K튟³Ò®^…Þ¸‘0C޽‡é^Or½Î}­ —gT›gAÓ`ú'Úå¥ÚòB°R‰WæZ&%Óþ*yý¢7¸²êþ•OÜìpOí8hÃêUɽi±8ÀqÿEº;ã$ŽÅ½k?«ßLb¶)øÕã æÏ“‘j¸>ÜÎN޾ä×ÔÐéy"Hƒ¾G7|¯íì6¡—–á=ÍŒ>—,S`äW€$.'jq1¦ÿÇÝ!v‡£Éßj˜Xb5(¸Žú¦#) û°âOƒ¡^†ä¯àBÉʤ ’¡,á¼C$úÞ·µå)䯀'm·á È_‹¶Rʵ¥BÅ€š·öæ .ÁH×\Y]Ò1;ˆ™{x1°PÛsLFgŸX9t‚„ í.4"yõÝ E ‚ÃÈw0,¢ұݛ-k‡´Wr°1q¶ë‰óøÜ<¤> hÐr|<ÉtÕg\;;Ñ‚œdY2˜c¸ª_º¦Ų́Ÿß6ÏŠ7{fSÎôQ¹nPí”ë×ÏŽò"ö¼¾ü9`ŠÅp¶?¾ÇQ.'ÈZè“$úKísï¾-‚¶€ï¡öN¢®Ñ.išßH4-ë9pÃìvÚzöföù.@méM{ð3vƒ á 3—[=§Á4üj-*ž!5† —7#âžÞY“ü(~°º®¦6|×§½žŠÙ»972îÇæÜÇÙšŸ,o⇡ý`ޝ‹ÎiŒ¯-JiÚÍñúT€©p8œŒ8â^ùJN8fZaÅ©š‡À°jAá‘ob¨Ó4!i€cÛPf…ç 2^|]^üª«¿ endstream endobj 829 0 obj << /Type /Page /Contents 830 0 R /Resources 828 0 R /MediaBox [0 0 612 792] /Parent 784 0 R /Annots [ 824 0 R 825 0 R 826 0 R 827 0 R ] >> endobj 824 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.299 102.639 633.254] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 825 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 621.299 186.698 633.254] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) >> >> endobj 826 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 271.183 102.639 283.138] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 827 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 271.183 186.698 283.138] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute4) >> >> endobj 831 0 obj << /D [829 0 R /XYZ 96.768 705.06 null] >> endobj 786 0 obj << /D [829 0 R /XYZ 101.643 624.786 null] >> endobj 832 0 obj << /D [829 0 R /XYZ 101.643 274.67 null] >> endobj 828 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 838 0 obj << /Length 1426 /Filter /FlateDecode >> stream xÚµWKÛ6¾çW¸Ù"+±¢‡%ÙA}hƒM=>¤Øî‘i›XItIz½ù÷ጴ–VÞEÐô`pDç͙Ͽ¬_½û˜f“e¸Ì“|²ÞNâ$ —É$ÏŠ0ƒõfr¨ít–%Yðw”E•lvnT#BQ:¥øˆñGLFº£iˆÁH«6=Þ·ÄôðÐ]º]ÿ>‰&³8—ë“••Cat¹ß}ŒçgîîˆeqÆçI´€öæûCO“EpšÎæÉ2ØjC„ÛOã@]kÃT©ëZ7ÕWú:Z¹Aj”ÂÊ÷ÓYž¶†ªfG<‚{4÷ê~šå¨Xì bNgE•±Ž™%«Ãp£x·—­I¨Y<—óyßeÑ D°.â`gôñàíÀ/Òm”øRI *çé½SV¢K]rí<¯?(6ǪBºì.ùã7oX&n¯Vר®;™IU;h[ÖiŸõׄïxûØBê'HÝ·¢U(knWøEíJT¯/„yÖ3 ?:ûÑöŒqŸqP­Ô#Ðôšb~ûM£z\Ÿ&11^m¸ò}g@â¤ÍÝKÉQÌ’ýÇ/• wˆp«Ê˜ßZUâ×gÆæO^ÿⵑ'§jIsi¹ Ó"ï몥ÛëÍjõšîFád„ôðe4ÔaR;kg1Ø›,‹`ôÞöeßË7Hø äßÐ!þ>IµÛ»§í‚KŸƒ×7 üÑoЗI;S=ÝH?B €ëAðèÌ`jï˜í8õÅóíy¯Ct–/ÍèÁ bÄ­Ÿýdó—^"ýˆ¾PˆhˆøBW[•d@æ´ð°½ˆ káŒz ^‡bXÚIQþåØ%úÞc0äð²‚p2Ë“àÓ¶ ˆ‘]d‹¡¹þ +2;(«ôôS‘öS1òÄíñ‹•Ž‘‚& `JQóÃ]CY¶tæm~D):ˆ@í©8:ñ9h‘#´€É‘X@޲‘Å|€,Š"œÉå&W+‹…à‡FøüoÕÀ€8š‡yôül]¤a¾ÌÖ&4ÊÞ¼Íf<=ðžòúíÿ6þ/safø½;)‡¸B¸rß…»ßüŸý?{ÁÚäEk3ëŠT®÷Ý_dì# lfÜ&hù Ø|½‘[í±¼¿æ®ùb¥,‹ÒÛ1O®‹òk¦QÜ¡›Á¤~”hä?Gi;ˆÒ—Žÿ~´v”øg`\?jœ4ÝB@>'¬ýÿÉÎîÔ½ÿW|ŽŽ:O —>"q–…‹âÛLɆeŒ¯~]¿ú "y endstream endobj 837 0 obj << /Type /Page /Contents 838 0 R /Resources 836 0 R /MediaBox [0 0 612 792] /Parent 840 0 R /Annots [ 834 0 R 835 0 R ] >> endobj 834 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 542.167 102.639 554.122] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) >> >> endobj 835 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 239.303 102.639 251.258] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format1) >> >> endobj 839 0 obj << /D [837 0 R /XYZ 96.768 705.06 null] >> endobj 833 0 obj << /D [837 0 R /XYZ 101.643 545.654 null] >> endobj 787 0 obj << /D [837 0 R /XYZ 101.643 242.79 null] >> endobj 836 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 844 0 obj << /Length 1116 /Filter /FlateDecode >> stream xÚ­WKoã6¾çW¸h°¶Q›1õÖ¢>´‹Ø"è¥.°‹t\‰Š‰H”JI‰ýï;#R±©ÈNŠìŤșo^œ‡ß\\ݸþ$&qà“M6¡žG"N?$ÜlÒÉÝìçù’®üÕ,+Õ|é;þLñ[^7<ÕŸU)dSë}³e9U<á)ïÍæS¹Ó›L4ÓzþmóÇd5YRJb(M¨Ú@¥å“|bʈÁ1D›7){QÂh’‹sR·EÁÔžÔ­z±dLªÔ¤¿. 3—÷ÍöÑð¢ºlDÁ჎q>p^Ù̬ªT¹³˜ú†~”f·+Ûf7µ9ØçY½ÅæY‚ù¦Å@2&Êœb¯ê„å|Ì''sàµç¿ŠIä†gƒEjƒŒOE!~,bP¦DÝàíŸßÞ.úÊõO ¿«4´ÊÏ8Ÿ¯ñgÑOõ¸6e’àXC¡ŸÊº¿ÕŸúG. ¢eèçµñ÷ §-ÇøÞàsYEÏâ]nJN{'sMèÇñÿ¿kÆøŒk"«¦Äs±¤Â™cènvU7Š «êµë͸Á¶’vmV y_+|wÆÒt@\· =à­èl×­Î ‡$Øë%)‹ª£,%—sXqRè†âÁ×UktÃTs£ðpÙÖ\u>[RÄžg‡›A©hHÜ€B`©Vö€PCÞÞÌ#×Ü­ÐÄZï*¦‘´937 ”’ëÏzI˜œ6=¨ú´ÌôÊ #gu){x¾×ôü±SOíõÅ~•mž’CJ*«¹ÎöcÉÊUÖµv¯fô‡0“A(2Uz×y7Š5f×°ï¹ÙbTÚæˆŽ+Lmµ‰ÆÓÃdª¯X nM-tWO%ßnÆ4Gf×ugIÎÐ×õtàáHÇ6:ðäD}?z»[“°Á‰K¢Ðï3sïªe&¤¨·#€!<ßëÉÅH@¼g‚µ~raH¼Ð9Ý >óC(ÁžËîÈ.…å ÷ª®ýŸVig«³³UùòZ{ÛÛì{›ýëì÷vSÕ¿?|ÐëOÐie›çÝàgF˜¢xnmÏ*â©­åú5keI¾Â× Û6Þô:劫éÉ0ê¿G¸0ª§sŒåo€?¢žŽi¬eÄüí9ƒ±-U3ÊÞe jÌ/'‰®?MMþô-¦æÓôÜñ¢íÐ8$VhÆêjÊ8B‹ëÍ۵¿U endstream endobj 843 0 obj << /Type /Page /Contents 844 0 R /Resources 842 0 R /MediaBox [0 0 612 792] /Parent 840 0 R /Annots [ 841 0 R ] >> endobj 841 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 267.198 102.639 279.153] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 845 0 obj << /D [843 0 R /XYZ 96.768 705.06 null] >> endobj 788 0 obj << /D [843 0 R /XYZ 101.643 270.685 null] >> endobj 842 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 853 0 obj << /Length 2016 /Filter /FlateDecode >> stream xÚXI㺾ϯh¼ËØ@›OÔ.$ Þ‚L9Ì›[¦ÛÊÈ’!J½üûÔFÚ’•™I†¹‹µEýåñïŸâêN§*IóøîñpWªÈË»<+Tœdwû»/›|»Ó:Î6ÿØ&zc×wÛ]Rd›wš›Ám¿>þí×Oå]¥ª<ΑMt·‹µ*cÍv»‹³|s†¿ÒóÀ…÷-ΑM‡~›NˆM·çÁèÛ7aPVöÜ4¼4/[$oy¥î‰fï¹ vÁeÄ;_·q±éE*3Œ"@Xßò?›|ÛfÙÆ<[TõÕZU™Xl<šl”'›“ùFòlr4ˆé\°o£:䌳ÁŒB8š§v«ñ˜Þ¸{4w$gâÍ뱩Qø#OIR<óÏ®!«àøó¬œúa4mCZ¿_Ø[qxùâ7ØTiT‚>¤*6WÓŠ‹wžv¦9úì–­®TRÄ·\‡[¶y¤J`&¤ä¼$AC™Ó¹%•’\Ü¡6£XÙ3AÓñÖxd»áxá)µÝ¥iÊÁˆGÌð’˜òÈ™®Ù ˜È5ŽÂfÖðÿ v5Ý3/¼m††‚¸f¶ÎY¶wcÚö~§K•ä çÃåBWcH§1¸£_ ‡,U:*®Ãááyc§$W¹QsÞ“ÄD΃uS;òÕ£Ûü&Ù²`[âüµieôÄ„'ÿ$朘»ó§!ªh´o `\¶e+T‘*t%VÈYн…œ>qdÆe¶9 ý‰G\TÊ|æ?XÿÁHÜ" ããJ¼(&JŠ+Z¨±B˜Ž'MW·Ó^HýbA^ðÇJ¶C$‰öÆòæXIv!ù{ß P‰¡:¢EÿÀ9÷²O£ÏÅFRÇqõi%1FÖŶ²)„ÝûÌ~ò…+aßêé§â¬ zœÌùV…8Uq_ ‹ä>8A­eu\äRÝ®ÒzoFƒÕ¸L? ½-ú@pt²ê¦'gG™tF™+ŒÎcåAboŽÉMÁQ3i<ýLšu=­tyÑS,¢€f'Š,gªhÚPú…’ ׉<Ô‰{ñ–•M@;ógÛÄ®VäcMòF^Z©¢ †b°Ó¾‘èÏdaZÉaä%ÇaÏÿŒ½0 c듟كlW<$åæmƒ Ò]m„½…2`‡2B/×_]Ü íßÇë‘kbYÅóLò¶7\ƒ}iúɵï¼ü{¤Óqô ñW²î›·}@µ(NU-°ÉAq^1t\ª¬Ê½¡?GÁQÜÌ5QBjÏœé@u‘ÃàB=+‰$Ï 1…¤—žÀ ƒ õÞPr»€á˜BZÃõüdqgšQ{ˆiY¯¡ v¶¡Ö}>öS‹E·ðe^#ZdGBಠBïeè—í(,‘Õh ‘Axø°åjÉš¢Éâ² ÷Ï$Yç ½ˆAMÁáÅÕ €Cã”.Âêà»™Ðvï®:îD:ò&®³êO?Œ&I]Œ?09ÄŸT½7úG÷ÝövZ¥…þé¢'ä7XsZ‰à¬Tîè5“½`8é(äœíqB3éE3>JG#ý!Þ,Áío8˜ÖÉ‹sbÜÃUAõñ¬ÓkeYx|5UºdŽb”ü:ŠUyiÏï—×׌[’(­/¦Xƒþk>AI¦òd@rÃ.ƒçßwü˜:Bµß£,Â~Jð=ï0<"¯颯½"ZRÈ#m@Jœ¤”9"ÛýÚÃ*”XÏËÔ—XÒ æ¤ÍCò}´/¶? õGVç#ìè9ûÌÇŸþü÷Ͽɑ·Ùì}6»Ä"/#ÛgÀ›QA’{c èY°¿ˆðø÷žÎ¾ÃMç•›òT¥U:s;dÑJ£þ³Õý ¢]¹ªH‰Ûüª¢„'õÿ]uhºÆ×”‚ ^*U$PýJ=¯´­°ÏF¾J+xønÐa8¾´Ã¡+7Þ <Ìz0o™è<ðz o$F|£@šó ½Üù,ÃÕý¾åòï„«ùFpЬ ÿQCßËó“…<Ì$C~ˆR\ôÁ é¢èCJ`-Ž+èRÍàÁÁ!,w{„8ö¸%’bÖ;Þ£$6ŒKl'ñµLÏ’ró¯-Ô{ØtÌçÚ&‹+Ü™a‹”lèƒJ™Ï.^ý&'?èî#ÀˆÐõ†gS#ï´¤¬ ûÏæÙNµ?§6®žà»ç)}I€Æ‹¨‚2^K «ýÕîʇŽ4…Z,¾t¬4#•Ê/Mº¯¨ÌÙõüÿê?;/d×<¬Ø”¾8©ˆ]bl|'¹p‡Ö®.S•èÅã'!èû)ù_(/T¤èÁU}ŽD'3ÖGქž | ^&ÅÚa\ÇÓÿíÜé{‡àJ¨jßnŽ7¸²àò…ÀIß_1ņ,ÁÔZ3a" æ £a€ô ãªTé4Ÿ«Õh9¼íÄtŒÂ=Iò@ÆÀßWùȱÂàôå‹þúu®–q ù"Û_I¹àâì \*Oj;*üøŒl« ?üöøá?ãk× endstream endobj 852 0 obj << /Type /Page /Contents 853 0 R /Resources 851 0 R /MediaBox [0 0 612 792] /Parent 840 0 R /Annots [ 846 0 R 847 0 R 848 0 R 849 0 R 850 0 R ] >> endobj 846 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 295.19 197.159 307.145] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 847 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 283.235 207.619 295.19] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 848 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 271.28 202.389 283.235] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 849 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 193.571 102.639 205.527] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 850 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 193.571 171.007 205.527] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 854 0 obj << /D [852 0 R /XYZ 96.768 705.06 null] >> endobj 42 0 obj << /D [852 0 R /XYZ 97.768 667.198 null] >> endobj 855 0 obj << /D [852 0 R /XYZ 123.92 298.677 null] >> endobj 856 0 obj << /D [852 0 R /XYZ 123.92 286.722 null] >> endobj 857 0 obj << /D [852 0 R /XYZ 123.92 274.767 null] >> endobj 858 0 obj << /D [852 0 R /XYZ 101.643 197.058 null] >> endobj 851 0 obj << /Font << /F29 110 0 R /F8 112 0 R /F35 119 0 R /F30 111 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 865 0 obj << /Length 1051 /Filter /FlateDecode >> stream xÚ¥W]oÛ6}ϯp„aN‹õe ˜ ì£y(†nh½£ÍkÓ¶‰ÔHÊŽ‡þø‘"%Q¶¤¸íC@…!/ï=çð\æ·ÕÝëÇ š$ ‰ýx²ÚM çƒ$I&q1Ym'ŸV˜åü•ùÑÃ/®ÓÝg/òò”ó”ìÕç $G¨~ô¡v©?í(ËË Íôô”!ú’áiµöiõnÎ7&.„ ‰"}(Î8~å΃Ȼ!ÒL¿¬s¨bº}AÓÞª‚!r®!˜«"ëÈ”m1›¶µ¼MmU¾,B¿š Z¨8ÎßgŒ˜kƒ¡BÐðˆ¬mDJ‰•c¥nâûH ÔñU±]ú2wx7)Êt´Ÿ„ª¸Þ0&{qP_j‹U¸§Ç¯_õxoQÎrT´+{WÔÇÛè½~„¡%¸À¥Ð L®Jj§—t4é^‚¸›ýqЖ×fõ៷ý–ª‡–e߉Xþ‘KZ)ÉÎ&†ìò½ö  ʲZ†I=y[Bè¤_Ó6ŠuµªIø=m“2ªk‘«T³KñÖ`É3b4Ø’b›G°˜‡5‹¼dGü\¸‹²p•Pzb&‘Ü×[R]q’€yÔÅ|ß'‰>r´ehRÅñøëŸuØÌÆ5åàCMšÆÊßòn(|DYÅ‚¹|b˜°c(ÇÝ=ZÔ¹†s¹í…¨üÌŽÜÓbeÝ“Ó-Î~Êjì€S†yA · xÄ$¤ÌªŒÖ×Òj×94ËèÉ-ÍœHsƒ|^rc¨(¤7vµ/x@ý#IÝˬ>Ju³º–Ø"ñ"õ–Øó8¾b¦MÔàz£¼Ç{$ÒcS‘BDÛ° –xÁ qŽKŸ›2_Â…lhË”±5¿Ç ÖÍ51YQã¹,׸ÍÊœðCøþ;zΌƛ€|—f|.zâºÖºZr­óž­`Ëe}QÒýAL€0^¼Í9  ç/¾AÖŸfðiX`=ÒR›J£¸ATä?Ìh%^æÍ±Ë¥iñ?ëANúO5ð&ѵ¼ñºÝóâtBŒ˜vU ^5_½{vý 49ô ÇìˆÔcɈõ”Šƒa›Ô–)ݱžÚêϾƒWš9ú8éé Œ/ôé4 m”²ao¤3Y)H;®é‘Ó\¤ª§›}ÿ–˜«¥ªõ8#ÞäFsá÷ºß×"ÞÐ’‰úô¶‡Í_Í+¦R¿»Á„S†·ðÖQMçctƒ9ïÕ¦Ìè µ²((ÃÏÞ’G±±;9éª0zºÙA­vG®·DÞ–ž»ù‡üÇÃLaqa­F‡¶Œ~ q~k7Ë¿µ‹›m]›é=šáª<> endobj 862 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 498.331 128.791 510.286] /A << /S /GoTo /D (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap3) >> >> endobj 866 0 obj << /D [864 0 R /XYZ 96.768 705.06 null] >> endobj 817 0 obj << /D [864 0 R /XYZ 127.794 501.818 null] >> endobj 863 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 872 0 obj << /Length 1580 /Filter /FlateDecode >> stream xÚÕXYÛ6~ϯp¶"6£ûºiZyØ:h4\›^‘IC¤÷è¯ï ‡’-Yö6@_úbÎñÍÁ¡Y¼zwSN*Våq>Y¬'Q³"Š'yV°8É&‹Õäk°Øˆé"6\á¯Å`ÈR*oøÒŠÆïÙ}£P[7Ç`É’` º1 5 ó£(Nìžh±…cক¨G쌓”•Y~ÁÐ’Yg飬kâz瘊žX'©t’œ¢àOé•xÜh㉗N¹‡i–¼‘àlC0ÅiÂò°êÃÄàVxŠx¯p®£}ñhÔÏ´p©‚÷~F8û@$niÖ©±fލÅK2;1º‘÷RñšVê¸8 ~*Â(øcZ&‹i³ŸLÜšF>ÕÓðI 6ò‰jO» QvæVFÓ1nh­´õ ý‡ ÇL}pÖÑ}‰‚FPÉÜtÀjåù­õ^­ºÏ'ø+ÌÂF€ªÖÇÖåÞºzœ\•J€'ø{;Gð¡?£Â.]Ä(þ¢ô(‘¬bÖÆGq~D’°²èªÈîYðÆÌ!÷÷»v9$vÕ´ɉ·k μ`!Tøžß~‚Œ³ YqËi µÓr© ­’8y8.]™s¥ªGÕ«{Ñk"BÌ7ÜŠ$¢6beU±¤èÒz=bå¼(Y ì{bŽlùyN#î°u …²Å¯ß1hgÞÔÁÏ «º7ïÍ ©ëÅí—­Ž—Œ’}›ÂžgŒŒ `òò¢‘QBýªú!) ݈Çf¢žc:6¯»>éàšrÚ"J³Ï™}Ûgç$¼¹=ùGçîÇ ŒCH¨O(jã­ë!¬ u¥5=t½Îo—JËÛ XÇaª2¹ v™°0_ûæý§ßGÑ^ˆfk𴪾±îã qôIík¤ÜÚGǵµT¯×P:S;'9 n¬Þ!ƒ«\¹šêB`ÁÇÙ£´Ju—öÜöÊ-¸IþU'w>.Ì1f͕ى¥äµqüÞ¸cü%ˆj¡îíg$æ< ù*­Þ¸#H6\­jÑU¸ö×/M\ÐOÇCdE§jrõŒƒ‚Êãbvûu†±ðÍ1pT¯¯á&/4ƒXi]Ž Vó;?â÷œEÙ9¿QÔ0;¿ê£ô§©‡"Ù§Òµ·­›-tãÇvHÔ‡Õáê‡ÜËZ”zíJÏ™¤ýá’ H‡Ä¯ *¿\ ròꇗ½b±8,ûwûg×ä¦ib/æLœï¦Ð©POýº¡MjÃ`¢´šnO·½]ÖõvЕÐx.¨ÅÏøÀqYs3ÖþÇÐþWEWíìc'&Äô«iKãÞȺROnÅ|Õ¶Wjå;ÿ&ˆ2– ë¶ÓÆ©›º7Ø0ºÆžluýsk6N@)ÿÜ“‡wDAÙ ïöÖ“bfpÒÓ‹†ZàÚuaEåÒÏ6žôÐã*½,Yûó‘ç—ÍòÙX5âØM¦ðx‰BÉ•0Ih8Q¶éù³;BÕs™g 0Ó ›ç´ÕIÈ¡Bø~r“šk¹½Ì6Gäy¼o¿jÄ…¦Ã·ŠlFû_Û@kížpÐæº=IO¸ÿQ»«ºÚéJ:©¶ÿþ:rðÈõux®åƒrWÅÅ6Ä÷¾c½ºÒíåeÅ=¼Þ„y=æ³?-ˆp50š©söhphÿŒ†ð:Ýà§±®ôv¢ÿº éäðú‰."èþÜèxbΈÛF‘91ñó—OŸN‹)þá'Ø T¬,K:…Žç«‹WÿÙÍ  endstream endobj 871 0 obj << /Type /Page /Contents 872 0 R /Resources 870 0 R /MediaBox [0 0 612 792] /Parent 840 0 R /Annots [ 867 0 R 868 0 R 869 0 R ] >> endobj 867 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 574.047 102.639 586.002] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 868 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 574.047 171.007 586.002] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 869 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 223.362 102.639 235.318] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 873 0 obj << /D [871 0 R /XYZ 96.768 705.06 null] >> endobj 861 0 obj << /D [871 0 R /XYZ 101.643 577.534 null] >> endobj 874 0 obj << /D [871 0 R /XYZ 101.643 226.849 null] >> endobj 870 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 878 0 obj << /Length 1970 /Filter /FlateDecode >> stream xÚ•XÝoã¸ß¿"XXˆµ¢¾µí=´×.p´}Ø(ÍcѶz²dt²þï;_T$GÉå^¬áp8ÎÇCÿíöÃç¯i~UGu‘W·Û+•¤Q\y%0qÛ\Ý­†¦=¸ëužä«¿¬ù;lõfÎi†ÓCg¾ÇyŒâðU×÷·¿^ÅWk¥¢:Uÿ/:hoÛ¸(¾aN/ß·” 'ßôú`æºÍÆu}ìZç?¾CÑv°,„«Z&Ûž¿êËtåç¯*›x©Î"• ŽÕ°ÄÌë2’$ŸïçÍáxျ»áQ[w×ÞßßÏL,xI»XØïm½CUÉ ?ùÍÉBK_±L²üMkU’GyR]ëŸÛK'Xð$ò“¹¨öÞ^Z †‡¶÷îÓ«±Å\»»Ü¶3ýÎ„ÓÄzY ì3W°¡ÄöÍó9’Wm€ä~aB¼‰w¿ß?~§L÷|ö‹0]X±[ŠSCu&sAPiÔ%E¤òäͨçPã¿G?¤´pþÕÚ^N—äµ|I/cü'™ÇxÉåïHôEöMŒYˆ,Ûþ¾ø]­3UYº8EÁT‹Q¯¦`• ''ük¸NªÕÓõ:M3@^ü¦+¿7Ll†Ãñäµo‡ÞE×ë,ËW·Ïs×kXÚÈP?ÐÔ=âa{íXÀ™£¶Ú›†‡ Œ›=^çÅJÛVú;žM^ÌJ™¹³Ãéè¾¼â7<›xà{¬Ê%'ÔE’™N“©êrsd!~ã×_«UÃ%Á'E]¢X£V4U‚K¼±‡¶É'ÀÚ@²c¸€ÒY‰UQYeÁ,çµõ/Mhë±Gûö²ï…q²eë÷’$uUe1ÏkÜ‘w¸·ÊØÛU3p£šÇ¶ß ŸâDP`ä.Žs£a#’D;0]òºZ}½® Ö–e@€œ„ô¾‰Z2õ%¤$U©b¼ÝÉ>šÇèä^º%|JGßù}ëXõƤ…-»Ž÷ä5ìŽêé³B*6äIL¶a+ÙëOÍ™IÃǶg<ÚdÿNäùv ZPðìø’ÁYíß§­oXÞ™R-b†f))W{Qs‚åùê¼h¡֜볂oQ$U$ ¶†ºHÀ­Cϳ>ˆÙv·—LÀážÌGʵˆ ý³5‡€âJ2«!œ×VcFëî08Ùö`‘v£»î¼xG€hö*Ê Ð ™—#˜ŽÙôœžÕ«W¢„Jã†5UBMµº”³P3V ˆSÏÈ883ó¦ÓÎ-à„ªÒ(+ëÑ:lO^§Š(ÏF™ÅÄHª2ŠÇÔøƒx™äÉ3nÀÍ ,ÉJð@@_»ÁïY›®sB²¢tE9ðWÊÓ²¥ |CM¹à!Ä̤ө,ª³ìÂOuI™ÔéBÓö'U—øðÔFŸòœAå6k×mï˜ÃÑ(![Šl&RTHðýB[ ½L>íÑq²A]žóq—Õc,“2Áµ%š'–míp-híR<´Öï%åÀ19ƒ®mm›‘ÁDZäL„E(KóÕ·ÙøâJã”­N² v€ÜåMʰ E8flÅ%ÿ¦Û b‹Š%TÀ@G3òH}£•ª.b_N§%çG Ì.°HÂÍ9}ÐÜ–¿üº y] +<’€nÏ2æ"ç;ôâC×°bluÉöÎ|œ.¾Á¦/Ï\Á^Oßàj ¿Þ¸›<¼Y Ó±ÿÀM®t>’†…ŸiÓ÷Ô¡ÑP{ÍÔ¤kÁáø¾ú™?Ø#iÎùNô·!ë¹:y”•¨™>ÈÖœÚ~zFÒ¿”üÖêN¸*)00/{Ö´ŠjU†. Üàþôì«y?W¨¨,Uå s¬ZÅøÅ.h¦=á}F}ņ;+vh&rYU¬B/à‚’q Ï?wò Xý¯0 ówB'o]Gàxª•Q•Wò1ý ôá·þv ½ endstream endobj 877 0 obj << /Type /Page /Contents 878 0 R /Resources 876 0 R /MediaBox [0 0 612 792] /Parent 840 0 R >> endobj 879 0 obj << /D [877 0 R /XYZ 96.768 705.06 null] >> endobj 876 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 882 0 obj << /Length 1428 /Filter /FlateDecode >> stream xÚ¥XK“£6¾ï¯py6eObËÌ«çÇV%UÉaÖ{HMæ ƒl´D!1ç×GOl0ff*å Zýøúën៶VŸ¢I âÀ &Ûý$AD“Àëù“m:yœo3|¿ô‚h^âã½Íq­9Úå˜éû”Ê» <#F•©]2ú~Ö—Dˆºá<5«V¾BŒa³鋳‚úfOk»·”.q¢. mÌæ¿ßAܳ|Rî£Z¼…÷OÛß'Îd !ˆ}é³)-Þzž–Ô˜K5ž§t2ü²ÐO5J°Ô¤¢Õ'¸¾@QªO®Q›i/¸ñ@úBP T'á[&´¨aíZa¸N®­81úüž¾V`£‚[†!X‡n7Dš4"C¾ëÏX&2"¹°p"Kè‡þüîkÃŒˆ@U]İY1oޏ6+t¯¯9.<Ó÷θ”‘±5­¯UMSåCJ v•”ÀDk HÁ 1P#Žõì†Vq ¼0°Pì°‚>X¯]+qGÊ$oRã~©pÂ€5»q ÎY7 ±R“ÈœŒÁ¹ò¯„üZû(–+JJñn`Ï%„²ÉsEq¹‘Ÿ*eëÐõïFc€,¡³Ó ôŽæ©¶Êx~2ø¨È†<¬÷(ˆJ¬rQ¬ym`k‚òî6…a«ìÇ[aõ̆ÂrQ„°+ˆs6Tan ïނ䋖_ ~ÛÉ‹^Ülà-2Œ¡ÐjÙlÖ¯ ¼…K>¡ò$/gOÆÈäÄ òÂwTW'ºL´âœ”ý¤º¸¼ùòÙôZ]°L‚â¦çG=nj¹?éqavdë.y}šé‡qÛ¦L+C]‹u+ ‰­›ëé,›b‡ën¿KÑi°R…gP 9g¢­˜íp%~qàWÕ8êèo˜Õ.F*ãÅcë{N%¶›éYɘ¨0F؉q\'}-O²sGíùW úkÎz3•£j n˜”[ÕŽÔv œó`±UY꿴〻óGaÝòu‡˜õÊfÕ"Üä~aß~3úh93ƒ¬h’Ì9;ÒŽI®ß7e -YGSc£MGJ{¹E5áY9IÆÒ´Ã9=óð¥•Á›RØÎ F½Ô|®ò†]ò6Zb«î£c*ƒ™Žy‘äâÌÕÍ#E•cÃÜ=;Ø·(g´µ×­Á—œ\ðÛ,k[#Îˆî‡‹Š³E¯Dc0w{ŒxSã¶Z‘a¶kÖ¯¶#É󞯂WFJG«Ñé{½E¥#*y ¸Âeªö…r•¢A£NhÞ;4¨Ú•õ ŽqSQ4ÓÅeÊZX\LbqÆ"i;†<¼È}u(ȳRF„};Z%j$vu0N+åÏC·©¤ÔæI}"èŒ>÷]–Åžž˜Ñ|ÓUüBg~‘$=9À£wL0“ÓºÁ½±UÐ×&­Ï‚qÄæØÖÀƒ™`p "Ø›`»s)´&í½Œd&»™lj3}2—¯¿;‡üð¸ ytŸž:Ӯç§[Ž$G×®HIµVé”eÑ7MoÉé~Ï0ïjÙ]÷ñs<âcNß(óÖ‚W´¬e¸ÚÌV3M[iz>ˆ£¸÷Q“)7S…Þ#ÑËÁý·OH=†Ý©ƒ¢.ÙA²Í Zò ˆdL³1ÎE‘àœ7Ê9ÁJ{œS]~a'Mƒr14£âøÎúªY¹Ðq†˜õ¾Ü˜šZã5gºÁ·Ü‘|TÒÃÝT©{{¾69A¥Ô p±;|·ß½z Nû@cy=íÒQP¤§Ñ„ûˆÏé|sÂÅÙ©.1jflx\v‡ï3"yû9¶„Aà‹osBT§ÉØÿi'#±@ “£û¿ô=à 8–žü#ÃÓº¡ãIÉ¿n?ü+ÄA endstream endobj 881 0 obj << /Type /Page /Contents 882 0 R /Resources 880 0 R /MediaBox [0 0 612 792] /Parent 884 0 R /Annots [ 875 0 R ] >> endobj 875 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.868 102.639 633.823] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 883 0 obj << /D [881 0 R /XYZ 96.768 705.06 null] >> endobj 859 0 obj << /D [881 0 R /XYZ 101.643 625.355 null] >> endobj 880 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 887 0 obj << /Length 825 /Filter /FlateDecode >> stream xÚ½VKs›0¾çWxœÌÄîÉC¦é¡nsè1õ¡×Âf†H ’8¿¾zÁ„顇„µ´o¿Õ®ôuuqs?Ç£„>òG«t=ùxÜY%£õärê 躖¦œˆ©ƒž|vÌ÷1‰1¢]yJŽ@-þq± gpöx$Q)ÀéfõcäŽAˆk×rñæz È#©©5¶F£…Ò±*-W$ç¬PÑÇˈ^[¼iF#EÔ~‹¢dE™ÕøO™ÄyÄy2ƒ@Ší€æ0¨±Óq·ƒ° 7hc~XÏb–ó5ÚlÚ´¾Ù°ë¶–I4!„u‘”Â7 Vy_˜ï8Þ‹‚eTðqÛÍF¢„$Äò“Væ,O*>¹e¦¬9ÑcNxwM(Ù?9zÞeñÎZí¬ùÒ2Í’Š}&½iI%eÁòHdŒVHÏÌS‰¼#t–šmUýˆÕ§L£Ø,~ªUEÎJzz‹ªpiס[ÌB¸P§âT9¶ÎVEÞsÀ gQÞwú©jvásB·B›Ë¢ñõZØlzH®Úö*]c4uæþO.5‡(·ì vβ¥”HÙŠá›Mg±Øž&šjÕJ/JR€>éο¥UßÏl©~f:D`¼Mé¶vU;QWÇÖñ®ëö•Óéô½µ­è‚¡Ž­7cF)¶bÂb™öë?÷åAÁþÝW«öQÑ ,•Á²P9rh8 æÀüv"Tµ¶¤Ô¼N;"œéÉRhsžÇ{f ÛË P«~¬öL²íNðPN Á¯’ì©P'Gª Æizÿ8‡=Í3.ªöN‘š›zbS˜'ûxSu8%Ê+”Iz%/rŠ»»kr T\óÿ«—Ó5w (<{í§· ö3ÜB÷yC=KÕÛÿºËÍÜT^Œxz‚ô‹Âò0†¼ñõÁiÔ÷Ø·mºæ_kŽ žq~ã)Õó0ƒïݑҙ¿˜W—ÔÞ5&eùÀ°/!ºóSõyÖ¼ÑÿaØ¢Ža‹ÿ÷°mœ¤™á2h>H\€=‰ … c]O)^|_]üiˆ  endstream endobj 886 0 obj << /Type /Page /Contents 887 0 R /Resources 885 0 R /MediaBox [0 0 612 792] /Parent 884 0 R >> endobj 888 0 obj << /D [886 0 R /XYZ 96.768 705.06 null] >> endobj 885 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 892 0 obj << /Length 971 /Filter /FlateDecode >> stream xÚÅVËnã6Ýç+ÜÌ`â#RKƒº‹E×^´p½`,ÊVǦ ‘š4óõ%yIYŒiÇ t#Qäå¹stÉŸ—wO_’lR¢2'ùdYOH2LP1ÏTê#¢ºáØPóÍIê¬ —½ ¦¬Ñœ ’¿¸QTRTwô`ˆ¸5Å‚¬n£œ yžú>>€ WN=¦ðÒadÜÚ5oSr£Sò|tTÍw΂rO0i·Ô¡K‚UžÃBa‰ê¾2v›£S‰$;?;0ùS 24¸6Ñ—µatí‹K¦…÷ ¶áÁ4C±pUaþÂ^VÜñã×~Ó!öŽ Þ3¾•»0Ðb/°Ž“ •Ey•vŒ3çEwæÍ¦Úت´|ÿêFvIåÑhRfoª³¡û½_ÊûªEzö>÷­3éX œ øÒe°ÎDÃ7®1ý}¤¼BÛ®©,•LöžÜT×mRZUûôx¿V> endobj 889 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 486.376 102.639 498.331] /A << /S /GoTo /D (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 893 0 obj << /D [891 0 R /XYZ 96.768 705.06 null] >> endobj 860 0 obj << /D [891 0 R /XYZ 101.643 489.863 null] >> endobj 890 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 896 0 obj << /Length 1186 /Filter /FlateDecode >> stream xÚíWKoã6¾çW¸‹VbFÔ[h}Ø Ð mÑö@K´C¬D¤äÄÿ¾Ce[¶œ4Žôˆœç7OÉ߯îãtV’2‹²Ùj3£EF²"™eiN"à¬êÙÚl:¶nø²ãíîãp{0køŸaJç­~ž-²ˆäY2[PJÊÔkŠÍ|‘FiRá3D³Ž;åAíá‘&gîË’Äy> Q{ƒ#€‹¼ IšÝ¨¾ûÈ_v¼êx¿[àÓ!Þ!ÏÁ oD©åòÓîÀ™6ŸFǽmÿœƒ yËË7ÂÙ7=kÏômk}ýY*Ó‡±»“ÂDò§:ÝNe/ IÑÛèjU©^v¡ï¹ì¦òêd_A’¾ o Ÿ0e„¦ôÕ®H¡[¯zb ¹ÆÆB‘¤,ña ÷®l %Era›Á‹2­ÙÁÁø Ë~µh— þ™Ó]²–›%À©íÉfö~Ô#¾\r¾ˆ²¨ OòÞ³ÿ…ƒ‹)ÅþêDO¥ã?Ÿè‹Ä Ð{²ÿÿx߸576ÃE}œ÷–ç"¢ÉFüÇqYœjm¸Þ³N(iÆp¥VÏm{«5$#¬²Šavvz>–G¹W–²ë7\n»'{:êjOì`ßð}"¿Ñd­ªy3î'G›k§Ì|ÍU|ée çe å··":ŒÕ/æâ÷¯8ª©h˜1Ö+¸:ùœ ûar×öþµ•3š8É¢cnÀO‘2ñ€r¨7sŸ«Š7Ü®SØ-qžó2 ˜hzÍ‘°­?µ žQ•4è¦Á¼@<Îþê‰O„ !`òu4½Þóíµ©ðé±Ú½ôÍ—Tƒ&±gì˜f º9 ´¨PdsD—ÑÃÉ¢ ‡žÌá³1ŒÂ[²júš”d“!#“öÜk*Í Áù¸g>`@ã"qÈž\þ’à×y\¬{ûân4j»J·¬ñâLº2 þàãÁ…Úcüi–?uÈaQ¨uÌŽ¥¤=Ï£æªÁóÜŠK£4=€)]*™žê,Í·šeø4b°*º'{]¾é†6…ÜâµÃô‚>c°²áZ¾ò\‰w^)©ZQùæŠãóÆÉC&ÉЫyžj-ºëæU‘_ü'Øä4.,Œ¨ÌHs»ˆrR… Býpü¨{v£ÝiAâòÔí>ÂAÃ…¶QélbŠ”P8 ƒžœ8B.ÂËó†µ÷¬$GŸjƒÃ÷s„NnÑÙëaçôùäð|!AÌ$LK81:–ÞìÜÜÃKŒ_ÃK Ò³„cüáÆ×Se×– ’žvü~žfÐØý”zʇ‹_;£€<èÁ­CØ5ëØT°eLr8еæ{Ád5& ýÇ,œ†e·ÓʦõE´°0x4ânCF`f@ó©¾4¼­YQs†'ÓÁgƒ~ø•p°{Ã.ˆ0s’{yì/8`ìV}q¦qGE1¡e>~ éZ³Ên†/ÏVžéÚÏh¥Údl- c»é™Ñ<% |UÛI’&hކ™eÞý¸ºûkÞ1Õ endstream endobj 895 0 obj << /Type /Page /Contents 896 0 R /Resources 894 0 R /MediaBox [0 0 612 792] /Parent 884 0 R >> endobj 897 0 obj << /D [895 0 R /XYZ 96.768 705.06 null] >> endobj 46 0 obj << /D [895 0 R /XYZ 97.768 283.567 null] >> endobj 50 0 obj << /D [895 0 R /XYZ 97.768 204.025 null] >> endobj 894 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R /F33 118 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 900 0 obj << /Length 1774 /Filter /FlateDecode >> stream xÚ­XKoÛF¾çW¨NS@ÄðM ¨-Ð)zJÕSÓÚ\™[S¤Ê%¥¸èï¼HŠ䋵ÙyÏ7Cÿ¼{õî}è­¶î6 ’Õn¿Ú¦nšd«$NÝ ŒW»bõ§Sì×›0pt§Öí~}÷>»x&næ¥+IÕz¥‘sXo|GuíÚwÌg<гéJ¾ljÍGm³2çÌ›ãz¤Žny×’ðoD2éXEµªxsDmYS$âÇU¯Ùã³m‚ÈÝ‚Ï7þ|ÝÉÞF)òÝV¥Qz„^ƒÈÒA«­)zUYÖÂæªÒ_=¢}/W&Ø‚¢Z!ÑmÛ´bìž)/ýB¼XXžäF×xÕ-Y(ú²…óÌkÚg Þ­‘›Ni83cÁód×\Ññ®2µVbűՅÉ;0jIG9‹BUäÊê…*ˆ%þ ^ežÑr]™’ÂÅ‹…±ÇJåú0ø‡ô;·øÐê¼ÆŠ1O%‘Ù%EîÌUî9.è»a~I=ªe9 Tý¤—eáWδ¥:rV/Dƒ·Ð0onÓòc¾ENÍìíÐÿ$ÁJ“cZ•ט„•gF‡œrÙáÜoíiÐÞ+u\ÂÓ(t³t,§æˆ]–:©‡­ÃJ¨M{òüÚ"ÇÎßpÇ'¦<0èÈ3 ~C9îžØà iã Î……˜ìa€ÙdXª/I*•0< <È3s¬†É#«b C"ÍçxÄ>9?ˆ=”‚?…ê_XÝA|‚$å /)Ñp1&TŠ-‚Žº†K…w§5ž\ôMÜN#îØªD¬"rÅŒ:Àtç£ÚL|rlâJˆ#¦Æ2*ùÎ7§ Ôÿà·ã„¾J“­øÙ à[Éî`ÌlÞ†#Èà• _\DÎè‘|÷Âka8®È÷Sƒ¦ìòlIT৘¡·ôR¡hzèûs^ 8gSU\t.Âæ_Ý6.3AŸ„  €VMß™Z60ѵ˜(Q3;ŸÊœÇ» &„œ‹¢Äa ÐÔDŸy²Õ>õC»µ|DE„|9Ü•)L'xC/h¼‡kC{—ƒ»R?žÔìøk„5R9Cë¹p :â³WÁq7ÃQ5Ã/sã‘r ýJÛzCÅ üŸUÕ¸@AP|¼Ë›Ã±ï(‡)ÎR‹x¥ prh¬0É‘Es84õ¬T’Ášª²½¶Õ—©#3èªg*âz@ãöj¸Exkm'•åG%ßM|ñ_)Å—\¶mÀÚøf´vmßžZý´Àsë¹I4¾0Â2¾b ’¢¡ëSŽ^8à5 N;o>o– äõòáÆ÷>N±D”}¸ô¦Móˆp¶ˆS·–Ò‹7¯¯sŒö'Å\ÞòE÷rÔ9ßcêaÖÝËÕ}¡OFÁÇ# Dß]/Içrïa4©dzC[;îÏœ°ãž¿9„ãUÝó°A‹#í¤¢÷ ¬ÿvÑ<ÉñÃîã¿Èá3?¼ÿé·ß‡Ã3BÀÖìÐu]ä¾%A’¸i4¢ó~!M6aâ¹Ùözð±òÒ°AãÅ ìÚ’z½J×O]9Åî ö¹ãŸý£nvÎLXtt³ðÄìQЦ†îc:äìqÑ=¼C£PøÎ¢«b°]sÄ·w¦>!zJºMú Aã¯;Qk/•ÀúÝÌ7wøØú³Jû°®¥¸ø¤ZÊX ±²õ½°×¶£ÿŒã ñ±zp ½ ÿ,:ˆ кºmÎqù¤7§àâ—7z¿7ø/r¯?=¹Ìà!^”Ÿƒ_õA;ÙÍ áß–þWƒ›Z@ÁIhM}åÂCSèJ°:Äa Ðä²VšÓœ맑„h sO–±š¾—"å«_v¯þ½6> endobj 901 0 obj << /D [899 0 R /XYZ 96.768 705.06 null] >> endobj 898 0 obj << /Font << /F30 111 0 R /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 909 0 obj << /Length 1721 /Filter /FlateDecode >> stream xÚµ]oÛ6ð½¿"Ë TbVß–€uà 4@†½Õ@Ò<(m³•)O¤âØßï¤H¶Ú¥ö$òxßw¼;êýúÕÛë(¹ÈEž†éÅzs„‘ÈË4Y‰ÖÕÅ­÷óbø‰ï)+xucË$L¼ª‘†VºaPÙ´­4‡FW´· }‹þXn6ªTRÛ+>è1MWóòKgvzÓ¹¸[ÿqá_,ƒ@äɉZv§ô–Õ¨ +[Z*Í ìdϪ’bŽQYƒ¸žì—%};m ½­¥0Yª¢6ŸAÖZ¶{suÉ— ^[„ͱUâ„„µÔ[»ÃÓ")ÿêhcú`ÊË ž*è@·KæwçH—=-F1ˆÖض°Å”¸°¶}6ˆ`ozKß8»˜lÎ_Ã)»æþ‹,íëQpç=¢AÓRì¿+s9à^Ž9áFi¹Aið]n=î7™¥Ú5«?“¥ç1VFè®®qÉ(ì9ДÕTÆõï~ü€¬ß^ñè"úD’³#Œ(aD"[Å€GÎ0í²RpkÎ9Å‘HVQ¨(Q╈ãÕÔ/ÂIñ¿ p©jNJ&2ÿTJŠtõ_¤l”Vf7#&‰…‰Y‰Ü_Mÿ%âlD‹ç¡ŸAô$¾kÕbA ó²Ö¶J>,ÂÌ“q ¿VÝw‹À³ªÑ»‘ŸzÇ*{GHÊЊbE«=žvµU‡ºç¨öÒ@­Ž£È[÷Ì+ážxÍ~ßhÞFNYô!iE[”XíA,ò8žºuÕ[âçÞ(W^%>–Á™«ýxl𽤕¬å^j` Ðl8R“&¾ŽÁ×Óµ­ÜŠÞ?èsîùÀÏEe=¸/9äNæ<í¤Âª¥%º´·òé‡ZÉŠ k 1N¦.¨ñÂ-£4A3Ü×IÀŦi÷¸J±ƒ9{ ¥1bŒÕB+Z&±ï}t¾AèÔ<*ÚþL~öƒx`P˜qP«µy–Ù«Sh¶7™ 6J¯(È9+puäŵ–N2¬0¤ø5ÒÎ%J‡(§Öy¤¢X$É€…À «\„á€S½˜“ÓnÓÔuƒúYëR+ýò¯VRUoñ€5x³ÀXô5« œ& „[¦°uDQŒN±t½[ÇN)ŒpJ¿â[ ,~žYä‰ù.Žˆ@îwËZòÒ6NŠÚ‰Ë'ò—ðTøù·çèye_*pÏUšòIÁyéI©N:áì5¿½±¼ã)gŽ›¬ Ïy§|G ~ …WÕkJß±'. ÂD$ÏI:WÊ–Y$Ò,=ñ eóDËjÛtºDûQ“G”þøäqÝΉ‘ÅጫÎÙ…©kXß1&HDš/°¥÷ÝܤUÍŒoø¬•éZ9ŽëOÓÀNÒ“¦Ç&#s"žZ&T¡Kyn[0K4ç *ût^ïr¨AÑ´&Pƒ‡ú®qö©Ý…Ðla¾Ò‘ëðih_Á _I‚w…e¨{&8X?ž8¶²ÂYê%‹JÝèM-ßÍß"Ðrê#…+kæ™…¬®"‡†l§ m]?ÍŒsSpQ]y4ôÄ®†yB3KâŠÖÒ [~yªŠ½=ڜɵ›ˆBï†1”%5e55íËDaºuÙMUa‰‰÷ «¹Šº&¦]AÈÏz&\Û^­¸MÀB>ÚVîyó°HRpo')m'ˆ'BPséOi¨}sWyA åýmŸIµLäÙÐWÐG/!“Ûj–Á¸‡ —E+ñy‚@ʆ¾­„ñ ’ÕÍ¥ðlï,pǹ…"òO =…:Šxʉú)|iÑÝì£ç/—©øö_ÆIè‚ñiG]×!Tª">7´çá\Ó®M°“ ùéÝ5A”žÿ ¥Þ{Ya½˜±»UntƒÍ^íᥬ·sy0Œ õžæ†¾ôoú4»~¨—ñ€Q7z{Å´ Žá¸*ù•ìýˆt*ð sÙðŒB *xymYV RÅÐÆçÞkRÿ¯P&8·*Š(€aˆjÕãL æ‰ðÃaÈœËÀäy–<*¼MÈðÞÝgæN¥…ÖClе³´˜¬+Ž;Wå+&ÒÌÆ…©i¿*òŸ#VUWÔFpZ Yt2Ý8?f}ì2x–á{‹sÏé €.ëŽUÎÉëÿ¢ç-0’nš'–šÐÝÀ½?²±?²“Ç ãM¬–æŠe’¡#Uû?DxJUâw_Kþå‹(ƒ ‡ð¼Ê2²6ð3<|õaýê?ìQÊ endstream endobj 908 0 obj << /Type /Page /Contents 909 0 R /Resources 907 0 R /MediaBox [0 0 612 792] /Parent 884 0 R /Annots [ 902 0 R 903 0 R 904 0 R 905 0 R 906 0 R ] >> endobj 902 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 498.331 176.237 510.286] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dist2) >> >> endobj 903 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 486.376 176.237 498.331] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}data2) >> >> endobj 904 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [175.601 474.421 181.468 486.376] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid2) >> >> endobj 905 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 462.466 186.698 474.421] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 906 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 362.226 102.639 374.181] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dist1) >> >> endobj 910 0 obj << /D [908 0 R /XYZ 96.768 705.06 null] >> endobj 911 0 obj << /D [908 0 R /XYZ 123.92 501.818 null] >> endobj 912 0 obj << /D [908 0 R /XYZ 123.92 489.863 null] >> endobj 913 0 obj << /D [908 0 R /XYZ 123.92 477.908 null] >> endobj 914 0 obj << /D [908 0 R /XYZ 123.92 465.953 null] >> endobj 915 0 obj << /D [908 0 R /XYZ 101.643 365.713 null] >> endobj 907 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 922 0 obj << /Length 1284 /Filter /FlateDecode >> stream xÚ¥WK¯â6Þϯ ÒU ñÄNâ$Õ°©4#µËŠYQ&1N(6\æÇ÷8v {o»€Äöñ9ßùÎÃÎïËOŸ¿Å“%”ÐÉr;I"ÑxBÃ?œ,³ÉÊ©8ÏÄÌõqäÈZ=©³™¹$v¸žlxZÏ`xR¼áš¹Ÿ8˽8ÍBê°&g›‚ÏÖË??ÕƒM Æb‚'ž6'ò]É´Ø52’ ߣPn .5",r˜Štæb‡¬Ñã|kØ@=ů2N Ù0Éæ0 ž³9J½Z«-¯¹h‘O\G“^0JBÃPkÜ÷[ã~`ˆHeÝèé×\îGüŽ|äÅQçQuïrŒÂtë¼à%¯”f)´™´ÖC–WyµÓ¦d˹‚r84õ8—\K‹”fmÛã,‘ÈÙë‘8BXÃÐù‡·äI¤=¦Ñ$´nCxžSð“ÒÀ¶33RaººQS yvd…PC•v ý®rËìëT mœÃN¦¥2Þä:‰d®ÉÕó%“M~†t£Aèü±í5FÕ«ÂÆ„°‘ø4@œ2§ÅQHÞÜGÂõ£yIb³ $°ÛE³œW7q(ë–‰ŒÝBnÂ÷šfn¨¦68Ç*Ó¯y5V- E^à÷iÃòG'6:ˆõ‚s­sHç;¼#š@è sÔ^pŸ¤i]êÈו¡é-8°Ø 0òCÅ&LÃfW"ô:>(Žúè4¢q3¨Ð¨G‘E4ÀBgvLv!¥f+$QHBço/ôrªcQ¨×zE!_z²a«Ÿ>ôó‹«ŸFÚðF­¬á…"8z¸Ó ªÚO|”à¡£³½M!e 'DºW/òràs½˜ªñjðPW‚OÍì4ã§œUé0ñZ7? L;ßZÿFð?§è2Ðòó§~þr%¥{콌Z»öì×Á1Kÿj5=O×ë;ŒÄƒDÂ6Ðrk³TÖP…hÛ°’ { n§¢åÚ} Ïå´\®aÞdmì#÷é½ÉZ—8ÛÒeÌ-~f[w¼rÛ…´ú{by¦,{u•x“…$~Š{¢^d—˜R¼=V¶ÁÑ‚9ƒcâØÜÔÊe5¯Òz y­§‹7=e•Úàõzíj.ê]Z7p©‘=q,•<,n²Žï²êµÌñz¬t®J÷ƒ*ÐÅbšWpÞœX1}Ä2‘GŸ³œÀÐ7ݾüõbá[¡Ç±t“_é&¯²²œÑ…#­¿ó>þz(&K¯ÕøZá÷Yx¨ÿ†¤›PïÆHj«û! ‘È|+‡º!î äi¾Ç â ¹âiL¯†Eè&мE}‹(<‚è^ÿ,7Ãöñ05ï¸\°='îõÒÃSõA,û*3‡‹9]˜iÑ*µ¾øv¸Ü5›öÅs_÷š÷eá8Oy¿>~µçá}Ú6¢· ¸ÎlÇ•dÕ®àHxšÃ^Ù_ò¦Ý!£Át׈‡×†v±•·s+WY}‘ê}ýVë.xµ“ú†£6Á©=£ é¡Ó€ÐÎo|)·«U¿mmÛ¢·iôPKO>´«ù Ï öu# ¶áÅbù×÷¯èÐZlÍLE¸„otr à‘’jLKÉÎ77­1óíGöØå¦ýN\éÝëÿÓl?ÐCðM¢Z¾0> endobj 919 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 597.958 102.639 609.913] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}data1) >> >> endobj 923 0 obj << /D [921 0 R /XYZ 96.768 705.06 null] >> endobj 916 0 obj << /D [921 0 R /XYZ 101.643 601.445 null] >> endobj 920 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 930 0 obj << /Length 2651 /Filter /FlateDecode >> stream xÚåZÝoÛÈ÷_¡¢DF£=î—Dk 8 y(ŠE ôÐ$´DÙìQ¤Ž¤çúÏ÷7»K‰¤––Ý8‡}±ÉÝáÌì|Ϭ~¸½úþÔ‹”¥±ˆ·Û’¥bkÃ6n7‹÷ËÛ¼Ùµâz¥…^þaåþÛµë·^D‹ç,Õ¸j»&ë²14A¶ÅýnWßý+_wßµë¬Ìé“ïßq5`n‚àÞŒø_‰üsüÇ98w€ÅÖ¡ÿé¨Êó {Ĉ3±ûÙ6„_9fä1tŠ÷ïß<¾ùøñzÅAuùÛM]½éÜþ}“ݹ§Ç2ÿ”—­{Y×»}]åUÖðE˪CYÒ£=H˜…ˆ×üIÞvõ&/Ù.ëš‚¾œnßúÍíÛž• Zl»–eÍý×ÊqqFxGÑt” á°/¢ÿ 3:õ6ö_¯T¤À`ÛÑ“•Ç®®Üj“·Åæ•n§hÝjw-’å—ýõ ÿò# {¯«6Çi”ËÏÅš Ü×Mþó¡œû +{ŠjYÕbW7Ýgú®n~ò˜¶uã ºkŸ8ƒb©Rã3רm€^Æ!0Ä=“éÑ_vµ[Y7yÖù]Bi7rÜb½uk›¼)>]ëx™uøOÇmÝ ‰l-Ç´°oj{þ5ˆÕ=s ýÇu"{6­õ‡˜X¸Ô)3QÜ;Á™7çZ”88ü ‡ªw+'™*›t,‘^%yE¼CÚÿ¶M½sOîè8Ç&ÿTdÕ:÷‡jÝdöd?€N–›±¦¸†ö–VÓ¡¶ ¨É»CSµÓOmˆ¢§½`]T¤¨.$såýA¿œK"a*5ý~ѵy¹=qìdbH¤ÆË$vK¥Ëü1[wÄ»–Ë2ߺG±´§ÅvSÜ?ôBÃ+ÔÑB×÷VßµyÓÛ…“°d•ßv$’e¶‡Uì›ÂÙ6vÅÆ«¤¨È·=Q¿Úã.'´ñ ºmÙLȸùƒp<0KŒîå×´ÍÊ:{—JX‰£¤Cº¢ºqÂלÉiTFÓîË>¿¹yCƒbÆ›™èƹfQlžÌ 0}®â1©/YcpoåDý‹¸6¾eÁ}†Ð|Œíy7Í §M0bsÕZ”Dñ;ç·ös èiŸÀ<•²¨ò¬a{èµXwu3sü—¤Þ! ²[(“ûL5&4…’cì#særLÑM׬\lœ192ЈNfÌ㳨X£2Z±Hž‘1,ÌW¤BÉKoîÕæS.€ý ó6‹Öm"¸oüJåV²²ßõømžrkë¬%ÿ])™öôãeÛeM‡â Ï O}&“Äð}ôÇ ÂGѨQ:ɧèBÔ2Oÿ-óBÆ>øß qŸ^]Üw°™[ê«….:¸®ËÃŽAÖÞ¯Q9ŒÕ$Œó´?Ç»ó@{Çmò­’QhzÙ"€&y©Yþµîòž@Öy¸­ÃÓY‰Ø¥¬ñJ%‘—'3‰ñr(»b_Ú9'Z=èmIS÷ÖN:’å Ç’š'™¥ë“fÝL¾êòf瞊ʯ¸1.z×ÚÎÕ6yiÚÆ­:}ÕºGùïØgM>B÷O=šxÒ†ŸÁÆîü¡{…Ã5[oPG+|³§\Eå+,·'­º„ ?æD7?0Z)˜:5ûkÍ—Ô¯e‹*·;Ÿ»È°t¶MW~—DÌP=§b–Sézϲ[q¼›G£•N¸Q,¾L´’¦Âð—œ&Åôä+G²(X›5*¼81j+F;Ù[„fݾ”•'+7G+7+§jÕŽ#ÙϬÝËç¦èº¼r Yðç%\L7:Æ&ú¡W'£öN+3šýÏèf‹¥œÆ*öýc¨BÐ_êgYž˜ÐzÁˆ%š …Èdß_ €àñ5;á_€°¤¼Û¤M˜á5ÒdØ@OYC2_™F‡Ð¶þûoñ¶%Äòwçú§â%ýë'ôŸ ó€£XQʯ@Zë"|™o| èÕ)Bšˆ“ Î PÀIÓs¬¬Œl’Ä6ÊxŸýg€9§ ˜¨Â¿¡¤P{š éY >Z¢ð˜.2DHïÈL'¥÷äüÔÆ­×yšY"`$êO ˜*´’YýÒˆgz±=¡š$*Du|Æ´äC²3£*afúF¦¹tΓKéÌÏgœ˜z£.‘„›À¤p7u}ÛþŽ~UA>'4÷7SZŒ.ϵ½<§íÓÅ=ÁL.î[·jmâ±ñvçZR{#¦…k<Ѳ&ܯXüAíÏ hï°ß÷•<_úͼšü²Àý*€L]£Zs݃½…vÁñ*Úm¡ž€GHx&`MãfD'› ÔãÇ_ºüx½²ÎÂU`Ì¡OîH1ª x Ðq%~rÄyÔOiþPi© endstream endobj 929 0 obj << /Type /Page /Contents 930 0 R /Resources 928 0 R /MediaBox [0 0 612 792] /Parent 924 0 R /Annots [ 925 0 R 926 0 R 927 0 R ] >> endobj 925 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 462.814 102.639 474.769] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid1) >> >> endobj 926 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [175.601 391.083 181.468 403.038] /A << /S /GoTo /D (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}deriv2) >> >> endobj 927 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [180.831 379.127 186.698 391.083] /A << /S /GoTo /D (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid22) >> >> endobj 931 0 obj << /D [929 0 R /XYZ 96.768 705.06 null] >> endobj 917 0 obj << /D [929 0 R /XYZ 101.643 466.301 null] >> endobj 932 0 obj << /D [929 0 R /XYZ 123.92 394.57 null] >> endobj 933 0 obj << /D [929 0 R /XYZ 123.92 382.614 null] >> endobj 928 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R /F11 183 0 R /F13 311 0 R /F10 255 0 R /F7 257 0 R /F33 118 0 R /F1 256 0 R /F9 259 0 R >> /ProcSet [ /PDF /Text ] >> endobj 939 0 obj << /Length 2816 /Filter /FlateDecode >> stream xÚí\Ýoã¸Ï_! *ßE\røk€Ã] ÷P hÞ²ûàØJì±SÛ¹Üæ¯ïP¤mѦdÙqœôºO–%z4Ÿ¿Iÿr}ñé³É,± Tv}—1D3ȔԸ̮‡ÙM~=*{…šËùø·žTy‰Ÿ`òráôçaÄb4s÷Ÿ§ø•Ùü¶W¸Q³W°œô MyþÙ2üb9ê1¤×ûzýë§ÏŒÕx DHžQÏÅRçV+W¿PIQ‰ѵ1”²]2L£Ì¬Fˆ9Ö ¿J2E(ˆ×ó´&‘àÇ]ñØYý¼K E³ì~Út)¦ðƒåô2Á”ÙÃÜ[éªb‹¥Ø*¸ÑÄr›Œ¡±‚“¿„‘4SÄjn¼-©dA1å=¥d°Z!ÍÍ «„Œ‰Ï›å`œ€ÙøúlÞ+8çyß}ˆ|>¾¹Zú»ƒrº˜ÍË¡ÿ6»]”.´æ«ÐœMçÄøLQ «µN;É=Iù¹y Ü ¾ŒPM²OÊ»´ØÓ’¤¬ Ln5Œ iÀý¹W nîýx]Õ1ˆ[0"¬¬Yçâo×ÿ¹p‘C3–1à„ÍÒÂfƒ‡‹›¯4âÃ_3J¸5Ùs5ô!c¨<å´;ÉþuñÏ‹_ÂÆúaô>†PRãï …tèQ…ï&†‹C4 ¡ Ò¢YRÅ*‹vT*Ûä\QªËù…r˜¯ 5ÄRå°Ì¶"a\Uyn¡ §:B`Æ#÷ES:Ç Mh³œ Þ+wð‡5áÏK* ¨ŽSÇ®Jµª0¢ƒJ1¤jUª6„k+µ{bM™±«d±ÝÐJoÛM¦‚U Ðd—ÄmIác2ˆÍ1)-#†²“Ä$ↈú 1ÙAÒŽ1ɹ7&¡5& bÒ¼Ö;f%Ù˜%äb+®YÂ4¨(cv-qÕç*‘j95-Thºø úA{D¬hÕ=¥{ w„‡™}†­ :•aJîŠÕ…j6lZ¨-(ÞÅ$Ì2L LË gÕ*Eþ¼‹°X¼‰Èë AM á#¯K@µ Z²½Õ3èm™TœH ­@m‰ä¼¨)òc2) ªÓ@åTŠQS1$ÐÜqÁPVÖkÙ +I±›%z&Á.Âa6€6ÙV¤ê²mÀ¹.Uó»2±È‡9^ æ)å{k*~xME“ƒ…®E•|mQ…¯±œ œZaÖÆC¤øw+ªt•¡Š* ›:)º½Ä° aüÊŒ5!GmÌUk)¥°Œ3»:©¿×§´ãÍxU0î Rœ×ưó ì­SúqA N^N‚~X®J,}?úu®+úÕe;ú‰®ègÎ7£ìP9ì¯sí)ë\تs?ø¯ªˆC–ØïŸ.èÖàÓ‚(,v"=WñØì§«ˆáˆ6òÁuþ©Jb‡ð²©&nÊ’f_–䇈xpÿâ|YR´eI3vÔº1è#óýv†£øNŒ Š¿DëµÕbˆÈ¦5 ZÂbÄE”Ãc‡”tm­Éì>ÕèUœÒ.G´¾€seA£NN-Dâ> ‹‹­Žå¥W‘VÌÏ¢N{;´ª·îUJŽr¢‚º˜R¡T‹)¥à„£²"S¾[Ym«‰P—^¥Xe:¢Y)[ŠÀàÖ®ËN!ÊkÔ[DÈVuH+Ÿ=ªÓytÅ]þCʤØwú§sÀT´ÒYÜ’k»zÃ>çɪ:ô|Ée‡`‰=·íSœ ZÓ÷6çû´9 Ó£×K*4:UëTŸw˜ê+{À\¿¹ÓiqÉßeÊZ=“ÒÚ_Q˜‹Œny/_͆ @o{y;Ї¥£½õe{3_ÄùN/!¶3h,ip^~X1§݆-ú9gérVŽÿ˜Ü³ÆaZêɱö-€!DwŠt `m‹—B`i„g§KWÛ’A{=ɪNÞ§=á 4±«% |µAW ˆq¨lÄë Énø"÷vy…ÄT ݺ¼oýÐ ýÎá•8ô£â)·,èó„4ò›Và¯Ê|W ûÔÿEKùU+N§Ÿ2g8}v»µNQRuäiÔAmLìIZÂ@vA(FQJÓ6m2Œ&…£¾O›ÞgÚš˜WW"Í:8¤÷Õº|þ*ƒ¶Õñ¶ãóÕR²ƒ”f¿”©~g¼ÚºŠš¹ÑXPس.‡·L‘±zCôá–W{2Ûê$‹~mÛ ¥@Êðõum ¥à2U6¹Õxù–mѸW[,Ù¬¡4u;ïä³qq¿ƒ©-ѾsAbFØV•b ®Rs8o4ƒÈ¡íÒ ?o‰…;0±gÕϾž}hÐN¢3þ’C3­ÌÙܳ‚ …RJ}4ÃHÿ+pÖ}›v·u™÷«µO¾½Üžb‡½‚óÙÁ³JBû# ç$¶}åT ­­ÛH7)¬­0ˆ~Ÿøœoâó5µìŒ@ÎLÇ~X©/ö5xÚ¹ó†9ü°FgÍáhC޶Q8n{]è þÝOU&_ºsåîb0«Î‰KO–˺‡#D@K¡gˆ±kÇ/¶(ÏËIù[Oʼ?]ŸÅ»Õëfãú=ì/Êßûƒå%^k]= —«Çõ£¸ŽõÍ™Tü6ì/ûaüúðm$ ·Ýh½Qâ%µWÔwz­%yzô\—s¯Y®X«âCÌåtØ*[?h><-ç^7“”,ÕùÚB ëÏÿ»[þ€ñ|Ü¿¤LœV½b}XNg‰ðF koÌ N¦ü"°WEp²ÅÓm=ÿ.¨°² ½çÑxà^7r·Tå/î³z¥»XôJ?ºz—»ÕŸLüEòß܃õÐÕxR5\@ 2ÄU#ò5íià7uy“¨Ê$îÁÆ$îÛìÇ-ÜYïÕ°¢Ò½ œ—?;š¥¿^ο…‹™ûtç«Á™´ò–¡¿U™âéø±ƒþtPN&«Cå8äÖ ”žã齿]‰^ÂóÙ#ºÉrõêþxâÍîÂQx4/–óñí“{s7ºÑPB2ô‘¥ãÈ©ÃæÏãåÈ_5d#¬6Yú„{d~B(¤hmíqUdÚ¸?.pUšHA:†˜Ý,oþ´K1ƒLʨ£àã6Dñä)|Ý%^£ÝÿJè8¿§Ô ¬ØZÿ®¬êÎÿ{H從kwí|Ø}›zoÃh«"ÉäáÇu«‹ñ2|.6´0b—`A3ÿÃ*œžÜírŽÀʼnpo…•ÿ/þŸ <$kÏHøÓ‚àêx õÉ¡„ûî3jëœD]ž¸6c›Q…øÈëólzý©V©r[›]cÌc„‰±õ\ú¿­'¯ endstream endobj 938 0 obj << /Type /Page /Contents 939 0 R /Resources 937 0 R /MediaBox [0 0 612 792] /Parent 924 0 R /Annots [ 936 0 R ] >> endobj 936 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 235.318 102.639 247.273] /A << /S /GoTo /D (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}deriv1) >> >> endobj 940 0 obj << /D [938 0 R /XYZ 96.768 705.06 null] >> endobj 934 0 obj << /D [938 0 R /XYZ 101.643 238.804 null] >> endobj 937 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F14 120 0 R /F10 255 0 R /F7 257 0 R /F1 256 0 R /F13 311 0 R /F35 119 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 944 0 obj << /Length 1232 /Filter /FlateDecode >> stream xÚ­WKoã6¾çWØ¢x-Z|é¬{(ÐzèÉ·]P,ÚQW–\IN6Áþø5Ôƒ²$m†èápæ›™Cò÷õÕò–ËYD"Ÿù³õnF'›ù2 &ÖÉ웓âšÝ¸’Iç«‹_-ºq© ¤ó)9Ï(~Œ³“ªn6ë?gÞÌ¥”DÒØxð´Òò–ŠïÑ’=ªXð\æ< _€IQ¯NT^lóß=éé_UÇõ©Z­ôŠSó>‚o ¶é'¾@@°ÄùT¦û‡e[•WE©’&$É÷¸é£ó±#zæˆvˆ`Äý`àHýŒ·õT¶ÆVÙøt ~¦vo2ÊÏŒ¦;•UJë¼üæ-ì<¹ ôhѧ‚F#Û…Ñ¡\ëžcí}~JóZ•À˜ÆœK#N$cc XUÍŠ½×Ãì²nå›e•î±"?zç&KÅ>KLåiŠOti¤C}Cd„ÒÕÈPq òÜÖàunêYÏmƒ±ÓefÔƒ’ý%ªŽ±"LBÃ׋N/0IXLú”äï5Ä'Ã7Lâ÷Üx—$Nö %$㚌KrÇþcQ šZŽguAbï%”äŽaQ¸$Q½·i`¸òŠÒ"ý¸ŠÌÍHŒvù0'ƒò´ –§2ÉjË3ÞÑ_y»¡÷ަv–äeq™ûXð¦¼/lÞá·P¿Ì϶™Í*k§kàCZ]Æì©k`Ï_´õ;öfÀCÌç9×F ‡‚Z;=]x÷ü f? < £–^íß³‰# 8aL^n±gõ¹À/ÅÚu dËx[%U¦6æí}š'ÚϾãUz¿–Ãû¶>AH„”ãT Öj:ƒc¨rû¡»˜ÚvÀ¦UǨ@Õ¤ƒ~ym]ÛªtÀ­um‚–·áðòä‡$úˆ HÄ9Zû«¸a¡ópyà;OúÒ÷”ÐÙÆysaA¶=e£™¨NÀíåçw3.U•&ð¯jWFé^›|À¹½ …S”#‹ÉîηPŸiJ`ª4“Oi–¡¾QBq‘+Ã_A"!ìdm‹ìt€8˜/#|G•øo[ ‘ïõ·©Ê5Ìz¡U"§Òs¾€£®¡E1€äuÃÉklµ^„#PTôKQnh©eÀï2ý‰â\)ÍFµÀï=â4×_>¨ ÷ñEŒh‚;åçE†àÊZ –ÅÇÊ\`u‰ —£K[Š¡ƒsÎ!>ýáD=~JëåmÐZƒ® žBgÝ”ÄU|P­<=ž²¸N‹%i…::hÀÔÈv `€$€A– VG)rá±0 ñƒÉ–?Ì Ùf¢¬K·1È&ŒIA<°dTÓ‰‡‡¶%Z…fSúDÊQ»ÔŒé.ÏGµZ]'ê1ó­ºîw¥åœRI<¿+ênêÙ B…o»z~ˆkoô,3®´÷çÅ ›qqÿ·ÚÖŸqy¨º&:j/ey~ur›Ž2o(¬ŸK3ÒÇ ÞzžŠòGš›¦Ô÷‚‰ëòØ>à˵²l|á?e­…¬ikM¤ŸÍÝj{ƒnì¦÷¯JÝÛÛ>VRíuQzgÖL^ÞêÚ"tÄk”ÿúeÞ5ÍŽF5©r%ºñ^âg‘”Žøg” èúŽ‘0 Q‘Òf+]ý±¾ú!"Ù endstream endobj 943 0 obj << /Type /Page /Contents 944 0 R /Resources 942 0 R /MediaBox [0 0 612 792] /Parent 924 0 R /Annots [ 941 0 R ] >> endobj 941 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 223.362 102.639 235.318] /A << /S /GoTo /D (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid21) >> >> endobj 945 0 obj << /D [943 0 R /XYZ 96.768 705.06 null] >> endobj 935 0 obj << /D [943 0 R /XYZ 101.643 226.849 null] >> endobj 942 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F33 118 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 948 0 obj << /Length 630 /Filter /FlateDecode >> stream xÚÍWM›0½çW ­VùÐâ`c¨J•º‡^Ë-Ë $‹”¯Ú,›þûb²˜É¦Iµ‡ÈØŒgÞxæ=œïÁ`úhÃEÔD6ðA‰Pñ"ˆÙˆÏ·,›‘ÑWSqÂÒ|ö€B5¨a?6¡E¬ÑšÖö±„§ñk´âã0øiX† !ðHé?](£§b/ãérOPXN!®£8Ûå¶…²Ðà›ÛÝþ7”U·7ÜWóƒï8–²”Sþ›eâ©ZÉsá¦úµ•¯Á*žG<)H2vˆ½}ûõºæ’[l"½Âa •€Íª¶"+˜¿vD„¶âö2 Zpm§Q¿FÉœËf±ß5I. 4{M¦4‰ùobDO‰Q§ Ù€º'dHaÑ%ƒÞ@†è%2DÉ<È6¬O ˜¦ôfrôUFÝ ÕŽ~P¢²d½k48;_¦Ž¨¦5/%~ëZ¢pž*ð—h—tËžëõË$À¢®ŽŽwßTÛ®ȵIOn@zr éIûÝ£÷˯1Üîöq>ˆ*—+ß?tr71]Îm~·éücoˆH`_Ç”ÅUwB F¿Þ¬2e—:¨?qY³«âÖ‘8¸ø{ƒd•vÚÂrð#üq|>u endstream endobj 947 0 obj << /Type /Page /Contents 948 0 R /Resources 946 0 R /MediaBox [0 0 612 792] /Parent 924 0 R >> endobj 949 0 obj << /D [947 0 R /XYZ 96.768 705.06 null] >> endobj 946 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 957 0 obj << /Length 1580 /Filter /FlateDecode >> stream xÚWK“Û6 ¾çW8ÛÌDî¬Q$õ˜ii'9´“S¶“C·®EÛêÈ’F”v7ÿ¾AÉ–VëMr±Aˆ‡ð¿ß¼z÷)]e,‹£xu³[q±„G«X%,ju“¯þ >•.Ëoë"è të( 𧦓ºéŠ8éØv¦±þ`Gÿ:Ï‹jOôv½á¶†VNŒ)öüï,IÐUîyë²ÔïÖë( îÝ•–vl·Þ¨8øÏl;Ú)r¶þ÷æÏwŸ¸<3oÃ%ˤ„ÿ”¥<&óÄ(â3FÁÒD­Bbhm»ÙUa "•d¡káe©™,90¼G†¼•)ÅáœeÊ»ù—­ó‡ŠTð€¾? o8å*vÄrªØL4§Ý¶¥ÿß6Óõ¯‹r#Á2î•|lœÃ‘­î;"\°‘8–ü‹{]öÆ]T3ÎÖØ¾ì^zù벪/K¤ë; Û›J3½Å BkœEOÍㄉ„ÎÜ-xB3©¼jâ¹[* Ï,ò%í׃ã†'\2¬8ê®-qAWN±¨f:ÛúáĶ$Ù”ÿ2o~·4Õ¾;LoÏ\3“µ_rM²,â>îCàIÃôù 6æà¥Öä/FÓ§Òþc5—¢)¦¢ôrCÉâ0y^çÉ%çéüõûê´°]Ý ÛÕ×¶2˜.½«}r¼½½z.Bó<Ê[’ ¡µý‘âãè;üð³!ƒW8Î([A=2Ž«‚Š œ*88Îq¦PeÁÛÁ—¾½/î×q¸ò ¶…µˆcIV:}WÄYyu7øiG1”t¸srÙáR»¾r_ ­°(kK k:ºÒ7´APÓiy4Xþ!Á yî5ÔtÞ´@È@ô¶îËÜËk$ÜuR<ÇôI]M»wNEwï®Û#JíK½Xa1ŒxUÈT›¾3´ èâ/Ý”ºÚ|6…¡ÏÑí^!ßxÕ<6ZÖˆQÊa”eëL8¹Ïrç ðàCuÉLåt‡Â5üwÀî\4yÖQW(¢‡/˜Õè=m‘%ÁĦ¿²èºÒíã‘CK.‚»Þ³îÚú8¹Ô´õ¾ÕÇ#Á2n¸›¦µž€HÜûèáqṈj{&+ ö¨59å‚ãÏ7Å">¸¶À¿¯òòºâh®— -Ýg.Ãp¤ ÓØ×ΠdL(<‡®¦ö]L@™s›ÎHœ’Ém×$éνÔÐb †{&R·\È/î[Dr¯tšm§»Þb¾\ùGÙ¹ÆSf.ZV0î}ÇÚL»NÉ·fE™Ì,)!³´?<ù[0Ú»ÓùÐ&í%VÞ$ñ²\½ØE…?Ø?aÝØÝ‚$‘BùŠ~´y’!Kã©g“ê<〮unu'Œ±ç°)IX*^è0 {ËfÍÛßÖ|6×S\ùç\]ƒÊŸQ9ÖøóŽe¿Ð£=çh9sôÚa»m‹fÉåYÈbO\¾ÚdyjÐw+óÆ-¨JCÆS5S•B¿¬²ŸÔ…_Õ‚¢X2™ÉY#ÚF¥,Vjš·¾øfCËêh7«À¿¦?‹8ÕCµuø ÕÛá'8üôÅ —PzqЧádô4…x1•¥çæ=}tØže"áXeª¡‚øÎã@æÁa|ú½+jÛÛgY;9¦¾¿}%t¹èDü™œå¶³-“ç^†^«;MôÓÇaåýè<ïÇ[cjº¥ƒ¸Œ*X6é0œÂbz¶’ªšöÍú=k 7¶…îŒß$L¥µv:ă×OO‰_¿EûËqr26 Ýaótj¶5mÆãåñ8»s§xw(©‚¯Syq^Æ™Á°[OÆX8?¸Â'•ÀcÀl¨ÛÙ|ýÎ0øÒì~=NàÕHŒ°xŽ.Øà—fœ·G× EئFËóôÖxñƒ¼—4vü”mtÐ-ÌKEð Ü@…T‚ÿܼ\ÕKÀ£fÀ“ Æc9õ{^Ù‹èóøÚHÎ"ž]F0>‚ZqiFéÇ7”²‡/wIe–Á”|çØ :©×9}aú™ž=-o\ &8Ž6KÓ”´p.‘óÕÇ›WÿÄ|ʽ endstream endobj 956 0 obj << /Type /Page /Contents 957 0 R /Resources 955 0 R /MediaBox [0 0 612 792] /Parent 924 0 R /Annots [ 950 0 R 951 0 R 952 0 R 953 0 R 954 0 R ] >> endobj 950 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 634.933 102.639 646.888] /A << /S /GoTo /D (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 951 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 250.754 202.389 262.709] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript2) >> >> endobj 952 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 238.799 191.928 250.754] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula2) >> >> endobj 953 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 226.844 176.237 238.799] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}Surv2) >> >> endobj 954 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 151.631 102.639 163.587] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript1) >> >> endobj 958 0 obj << /D [956 0 R /XYZ 96.768 705.06 null] >> endobj 918 0 obj << /D [956 0 R /XYZ 101.643 638.419 null] >> endobj 54 0 obj << /D [956 0 R /XYZ 97.768 420.407 null] >> endobj 959 0 obj << /D [956 0 R /XYZ 102.999 254.241 null] >> endobj 960 0 obj << /D [956 0 R /XYZ 102.999 242.286 null] >> endobj 961 0 obj << /D [956 0 R /XYZ 102.999 230.331 null] >> endobj 962 0 obj << /D [956 0 R /XYZ 101.643 155.118 null] >> endobj 955 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R /F29 110 0 R >> /ProcSet [ /PDF /Text ] >> endobj 967 0 obj << /Length 1009 /Filter /FlateDecode >> stream xÚÅWMoÛ8½çW8NѤ€Í˜²%KÁæ²@ ì¢è¡pO^‰¶è•I/IùcýRüEÉvÒæ°—X$‡ó93“ßg7_Æa/ID½Ù²'O`/ § P+³¬7À…ÀŸ†a>ÀO‹ÙŸ_ऱeŒ@ÀÞB„vËÊz¾G¾‰öÚuD†2ÖFË3~†¡¢Ö£Üümh~ LW2ÿkŽ„äH"õ­Ù·i¥ÙQÙ6Hrr0%ßUÛôVm”V ”̈¦¬°MË&HÔ¾AEôg®ð§n7ƒ‹þTœ–Döý‹Y–4•„QÍÛž`¿2ζϳï?>»ƒ´ƒL@2š\N0R4âq+>ê^Óü2B3ÇF"¾Âò8 @4‚W À‚(öpg|Ïr"|p'iç¿›Ÿ=ãüh>Ñ +¥ Ž8J%®€WH’¶CfEX°IQa¢|)'[pF}ŠTÇøÃ" bŽ– DzäÔõ†­!³²Ý2A* fD¨Ä+ÌÝ #)> +ñfë>žŒö^Š6Xh3åäô4œ#³ùè¼ ôâYDëvŸï†Aèªúú/—8~ß~|ýj/ýô*‰µûµTáa>Öú*Öº¼!-ѲЩ†œP>~´ÜO«kŸ‡ÐÉÏá´tGYGžFO Låíkœn[¹Òe@ŸÛíYr•ësá’U‚õmê²JYãÔ~g [ž”Ù©9ñÖ†l0êi‹~V«ë’t<›ãRRî‚$¾š…’Œ§Ó7ךvùUÄèM2Z_iéÙW¬},wb‹SâÒ\Šê Ç$ÎêÔ¹Ãç”dÎë»5sóÁz «Põgñ6ã–̹O¾žî ™é+€Ñk€åv‹;pz² ¦'ßs¶‚í»Pz² ¥'ß•–›ýÛÆ2³m03ÛE;›Ýƒ_ëMAFœÚ|{9=äÈfF‹£_7ÕÞ»”ÇwdW Wpÿ³ó«Rôÿô¶×JÁ)BcÜ*éÿSÙ¡‚ÑØ™?>H•/ö÷‘nqVeÞgc:@MÜuO®‘èd\›_ô¤í*X㔯¨Ñ£»*äzKçú|ý‰üú³EBb£4M¡?Ð[¦1˜(<·6%óÞžSý—P ­ÀÏ÷f|_-Îâ¾®jú²pvïq®3|£Ñ·í7¶Ô]ßÁù8¶jåKÝ&º†5Ãüɪ…m°Kåv¹tO¿‘ùø)XøÑàX{¸ÖØÎrë£ B>–´úq×BÕõ–îíîIQø^ ”ÛŸUZÐÚ‹›uq€qœ¨+I@ǰ²¼ù<»ùšÊß× endstream endobj 966 0 obj << /Type /Page /Contents 967 0 R /Resources 965 0 R /MediaBox [0 0 612 792] /Parent 969 0 R >> endobj 968 0 obj << /D [966 0 R /XYZ 96.768 705.06 null] >> endobj 965 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 972 0 obj << /Length 753 /Filter /FlateDecode >> stream xÚ½V]OÛ0}çWd1*5¦N§™è¤M‰ íeÝSׇº4:Q>h™øñóWš8qÓ* Ήï=Çö¹·þ>9¹¼±Ã²1Y9ÀE¶XôËdnL/r¼Jz¦c9W¦x¦8ù3pðK„ÉC¾dãÍY–§~îÓ1ì‹Yu¨7›ü0† !ð™÷ ãFÞ‚Da–³t‘Ÿ$Ñ …d¾‘  ò0&œgåðz\:ó/å_ÉkÙÀƒ‚6\ˆ0N³ÕÏhª”Wc98?Ïy'ÕçreªúŸ¿ïîtëÄQ†{¦;rërn#KxÎÁRñæŒôL¹N-€L™â™Žos–‡+ÊhÁA‚S¶ç;‚HÃì©¥LÀÝø“\ÝŒ-Ü`’Åi=H š(Ô>¾OaHE€” §Ûƒ— 9Ø”ÆÁŠâòkE`ZàY–*öALTª¥±žº2*låçi¸‘µQ¤ÏuK6!e´¼MÈ-4\¦çÛE*ã© ¼•Äù‹A@éäpégbp1£ G8Èñ\â/­9Ú'òcU¢Û<1‰^t§y*ÉœÆkY·k‰ÌcòYÚdí—†ÉcñŒâ 7–‘á .•dÅ}¤a²Šu"v™#¬,QV´<¥rÆc5ãõµj 5ø+T½ÅZ‚j¬›ow¿®÷y¶É¹Ã# ‘ÓédÑ^=lÖ³ZÓï ãvï÷™ê1û7;dÿXƒšœ¦J_B0Ú-bfKƒ€õ2L]ªÝbŠ$iW:KVùúÎEGñºÍÃÁЇ¿vóìí\A±Zú›D­˜Ä»–JÛÁàÁŒÿRµÓY@vzѡׅÜøØ¯ÊèX†T–1ä5Ú|/ð8±ß¤ xÒÿq¬Švñ6ü6ÐWß"º<ö{Ú:®§ßé`Õ—]v=è-¶ü~Œåößðb³7¶ÎÍmxT¿þy5#õšc—^ù<›yr=9ùÕs¡ endstream endobj 971 0 obj << /Type /Page /Contents 972 0 R /Resources 970 0 R /MediaBox [0 0 612 792] /Parent 969 0 R >> endobj 973 0 obj << /D [971 0 R /XYZ 96.768 705.06 null] >> endobj 970 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 977 0 obj << /Length 1465 /Filter /FlateDecode >> stream xÚ•WKÛ6¾çW8NQÛÀŠ+êe íæÐ ’¢(Ðî%Øì+Ñ6‰rIj½ô·—¡lÉrê\LŠä¼¾ùfHÿrÿêö}TÌ0FEšF³ûͬX£u–ϲt¢8ÝW³‡eŽð*À8J—¿‘}Mxð;eT¬ï?Þ¾Ïg*²(3²á,À9Šó‰Ýïè*ˆ×ñ²i¥r³²mš–»y'a»Ý¸Qùó²ÏŸCœX¡h¹éx©˜cÒ¦vnFܰiE³ŠòeWÑ#½F§ð®±íÊ„ºÑK¹^╳h$tt6 âˆm‘¥‹E·ÒK>L‡iÈ*ý‹ÏI˜aâ\§Î 0H$ň%z‡q­‘Th$E¢‘Ô  °#Q0.[^¿8•d¿¯•`¨uãPú Êp-M–O²é ÀßA>«–/€i{Áz«š G…h…›6TJ²¥nŸm€Xê”`Z®,Yþkö½¸]Šp´º}“Q%¯Q²ØwŸìäHŒòuêa6»a*0èdÏ5æ!ÂyÒ§e"á…wœBGr`ÀyŒÂ,òan¦ˆ‡9JG>¼³-ã4Ò†¨r‡J½nbõæú–•åðããP˜HÄICäð\XÖKË7 ùBÝIÎJÓÌÔPQéNH·°lKËLt‹©¶ùÆm—‚Úr²öa©Ý¿¸™i4ee°Ú0¹3Ýž¶µ78ª¤O1Œ„Cjϲ¡ÚžÞC&Áš©ä¡'M[Ñm w†yê{Nõu"WF¤´ˆ1À–…¡fÿäì¿=7‹ž† fR–§CÓ GÃ^˜U~:·¬ŸaÅ@Þ9õë­u÷î_0eê͵°Ÿ©xñÕ½-U£Lqú– ‰Y„|³é5}©O>÷ÖšŠS÷©¤[_ЖÐ&´ƒºûË>&ÒP±Æ—§Á¨“Žïæû_|1•@K{áè´êJº`@S»7{sÒű'yJúáFAÿ阠Õü’MEÍÍqJ[Ä–:>GÆ#ÿxIÁ·k~~Âä‹~4CqúLj´'BdáÔFN%£Ãs„ŠF¥•Y:oÊ6¨¹TBWÀÜ3.yëØ%çZϳ>:¥ì³ÀÚöô×§¨X\“Üšò­²%ªeŽùý”óØr߯ïðÀ¹£<_Sž¸†À ¶ÇÂSÄE 4Ñ(³ê¬Ãzòøë혴ÀÛ¶ € k íÁxÒ\º$>º•%‡®Rãµôx *÷-—ô*H_3‰|ñ}êá–ÍŸ  wPxÞ¦¢™FÚÃÓß´Tó3öÁ‘HzPWF4‡n;¿ªèêÜuëŒ ƒš2,èÞ•-¿T+cļ…i>o6æ.ðv¶DpýÌ´˜þa©cËÛq×9AÍp.Âax ôr-x1ÿöswî;:`ÇáÛš"¹§%#µC3hSŸ.vcé‡g"ìeð6¼ôh‹C´Ö/²o=ÚpX <ίé0£îÀ[7L•Hé/À¾EìÈ3`¬þâ¢?é/ókKÔÎýŸúNžZó'üôE7vÄ7ÐÊOü»Ù·;ÁÈS}™ lÔ盇‡¿ÇÉëoâö±K´è¯3žçîÆksòÕ¯÷¯þ@þ›Ä endstream endobj 976 0 obj << /Type /Page /Contents 977 0 R /Resources 975 0 R /MediaBox [0 0 612 792] /Parent 969 0 R /Annots [ 974 0 R ] >> endobj 974 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 570.062 102.639 582.017] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula1) >> >> endobj 978 0 obj << /D [976 0 R /XYZ 96.768 705.06 null] >> endobj 58 0 obj << /D [976 0 R /XYZ 97.768 667.198 null] >> endobj 963 0 obj << /D [976 0 R /XYZ 101.643 573.549 null] >> endobj 975 0 obj << /Font << /F29 110 0 R /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 981 0 obj << /Length 1108 /Filter /FlateDecode >> stream xÚ­WMâ6¾Ï¯ ìY)dã0¡ •Jݪ+µÕž6‡E,“p7±SÛiÕß^& 3=ÙIü~?ïã7ïÓ‡w&É`Χñtn ž„óx0}JÂx6HóÁê±(ÞŽ“8yüilV(û%QŠXÉón$äCXÀ *øH~o×éƒh0 œ'‰Q„·æ°.Ù‰½Þú¼þ²X˜5j_}öoa&¨6ÏP¥uØùâñ9ùU@‰YÅ™ û½èòOúŽ:rÁ €ÊZ¹*Šõi|*]\Æ÷æá§šÔ~Ù:h´QíùðÙ<—5f·±æ¡Y”¼aóÊÄð°±k~³¯GÈrnÍÔUE™Õ¼¥ì, „Ç9ªÊ0(7/‘x®œlW—ˆt'ª=æRTÒ!ú&3• )‹Œ¡><¼*g¤Öxˆ´T“7)øîx:Åi4g“¥:­fkŽxPÏf2EÓ~{ÈKçj멇Æa™ÑšLvCóâûwJÉV‰­þ©Ñ¿N ÃÞ¼4 É !ÔV±vøÜP±?«gWé̱Å-¹•và`|€¶³ý ½¸®ö.Ö*OË4è~U0Î}|⾋ßðZ‰ÔÜGêrŒMc®ïD8äaK> <=a‹ÊÊ—pS F.p< œ[,¢ž¶ö jº‚Ñ㧺äj«Œ8Y‹,p­_¦qžÀÕ~‘ßÃÌýLXrI]idÙ,•ìaîÑUMðßµ=U ²à¢Û‹š^²Q†§Œw•‚ £ÔæçS«ç^!V”åH×f¬R³ ⵩®xã û˜qÑ‘¤N í:RuVðöâèr6Óзmô`Üvÿ­û¹áÕ-4ñV¹£v À‚(h0Ldã0œ]pl -Zùö̵¾0±µ לíÃ쎔¹Qž1aW–~´ÍͪòåµSÐ2Úbh¯#.Š÷¢»¡¹^ëÿRAö¥±\L0 åöÞ X²`YAMMæÍ;"¾r=t¶ ¸«eNg%K+¿7¤Y*Šƒ’˜ó¬Ù˜ ýö³ v²“µÈϘ¥ ¥£‘¬—N§Ó-1± eçl÷9RËñ ÚÉ3Ýc«V£0³éü–>ü ™<>I endstream endobj 980 0 obj << /Type /Page /Contents 981 0 R /Resources 979 0 R /MediaBox [0 0 612 792] /Parent 969 0 R >> endobj 982 0 obj << /D [980 0 R /XYZ 96.768 705.06 null] >> endobj 979 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 987 0 obj << /Length 2442 /Filter /FlateDecode >> stream xÚ­ËŽã¸ñ>_ÑihhkE½dÉ"L. dÈag°%ÚfZ–Qê™Éa¿=õ¢,¹5Ù=äbSUÅbU±žüËÓ»>$Ù]Vyœß=îTš†eªîò¬cÀ<5w?nì/Ÿ£,ºŸºÁÔý±³ÿ1ÍnŸÅYà¦áվꖿÆos”j÷ËÓß~ø Òç}…U¬îöJ…U&œL¸!Z“ØóF¬ »©mq9šóånÔ£qx Jtu«óWðŸöüÿ€ìø€'ßžeZg~—šÌ!¬Îîþ‘Á2Ûˆ“°Rê­2Xh£ÇqÀÿ³0zèt¨ëÑö݃×nf™*¸¡d-8YcÞ³–õ7xϽgJÂ.xÖºm×ì~DȆí|ãÚ£Í /—Žø8*ál S%ÓýÔÕpIÓe·ËµJ"h†Žö,øüRõ; û‚?èÄëø©t”M=ÿ?Ç"ý¨¡§ÿbÇS?yê“PüóWõÀ‡‹(qýá†ø°S@pÞÁS«ÃÝ>MÒàédÛY¥a•¦k›ÈÚÁ~ؤùïlÁ·_HúA0œÚÑ^ZA F»¾søUr* ÃRÁâÙÆ ô2ôÏ­9ó‡•sÁ ¶;¢¼)Øôp³ßM`³,þ̤‚šxåÁéFîFzå¹O!©gӡ𸵀[ ) %§À¯jp‚ÖÃqZnP,*`º^š¿do4ŠF”*ø;ã;cÙCW óº5t;8ŒùD>þè8ÎI$!éz¼É/^P=n¹¾EyÒ’„”+-$7î2Á–!½XÑÔ#Jš)1î•ݺž‰½ô$qàŸÄ)I„¢8²â£‹µå,B."”,BF"0à™|Ú€ &B7ßÒ_óA¯T K Æº‹k”ò$º]@Õý0ð©=ˆ4:¹¡) FÌÎÄ€9\UÐÎúÓi/ŽIŽ ¤Ä 4'gSË 3 ýàù;§Bj7t=e—$ jL¥~ôxŒŽÄ„a8—ƒU©ÎošºP_ûè·KÆü5˜Ov0N*ðyœeq°ú„ãЭD\ݱ‚1÷y¨` ~–¥”Œ2Œá²–¨J\o`FÈi» ‘u/îmñÁ+,䬘9aEà¬gNöÂaÌ‚—P%l'´V€ì‚Ã7þ:šÎ ºë¦*¿–§™ÛœR!&„þKÖÃØÂÏ^þÍW46í¥¾‡×«z°‘y±R‚÷î³$ >ìÊ”Ø%yì“F®¨€ ŸSQƒïLpÌâ;÷9€­–„Ñ2â³JÒz‚Èî0QŽ „’÷ Ôñ{:c„ bÏU£Á$˜C–²\í=¥Aø‡beϬ$îý‚N;Ü´¯‡Á7>IQpM.°zqµÂõµ¤ÁQ:+àŽÈìßxÍ(PGOV—½õÐ;áƒFû‚“qÔr½- ©ï Þ\] Aè ×®áÐÈ[]©Øï»o«ÄYJÎPPþmãkóÀˆ¹šõI¶×ÚÉ Í„t]/Ó¹~à& ‡j÷±ó[õȔΌ>×âWÎlj™·Jhñ N%s}’è.«"))ù…iT€ÞS¦zy­Ð£BT…Q^®/ak¤)]]ÙD[¢¨(,`%DŸ65WQÅšg¿Cóìÿ®9Ô|ê.Sj?Z\T×#}MJ½Ã¨!ë5²gèÏL«àì™zR\úÙ/ÉÏRï0ðé‚;ÚÈE¬Dýý†z0žª,ù –pu3ž.çÀ‚æ°‚dŸÜVËEn€”Wô1¯¶AMã[jŠÝ·â”eXƱ?­{+MV³× g2r…X;Úû²¸„”®Ùxˆq0_cW õ'°aß x¼Ü3GìLÁ‹º‡|z'¸Ìo¯†lPÁ„Å[#fpLQ‘tœÜ_GCé‰"ž †ä üµÚ£ /à8XèkÍ~ì÷€Øƒž'߯‰n ôxÂy4Qiðådç~‘Ð-7†µ¤ÿ%{Î^¸â9oæ¼Qøxb‚e„!Œ­E£——Îà ¿C›á7™öQHà$MawáKÅc‰„‹4ŒòõE4SÉüŠÌKñ(ý¢2[ƒ"•—8Í‚ÖBs Ç‰"M*ÔŽq(6ýŸŒr-(®Õ~Vç ‡8(Áàð\­‘e­#¶ÝP‘×DôÂ|í09öAØ&¥1s°ÿþèKý?+åT¦»Ü…TŸ©:4ÄŠ+|~šæ¾Â”~pйQÃŽ¯¦(®?~d89X“ú¤"õu<뺾³µÞ|ò¨g×°n£[‡ÔŸæj=#Ìý²oÈ¥‰Æt>O¢àÒ”£ MîñúºöþáLnõp}“úUHÍW¡Ãü=LóæÚ>¯ß[–rI°ütÁþ‰F™`ªJyüæ¶AºéþøƒP þ¾™ò©„A;9éh+lmÏË.2JרZšš^H ÌØ;|¡áÕwû^ÿ5hhŠP+‰#y¼¡†6õí7 ÂÔ–b¾ÍÓàÏŒ•÷›Ñî}Ó–õ0þm –ü š¼•܃xÏ`¹&"„ÕXÝ¿èÀæVrQK:iü½Îðú`’ÊSTº Tü|ö'¬sÐpÂOñ5ñÐYï(ß#CŸæÒØßS$ôÖ¸¦ºq™Õ²\/ÁoÓ^Z"¯!͘&ÜÒc°Š!¹9–¹Â:ЏJ]ýÁša2V 8b|e RÅ·wNp%'8è=¡‡6̆ =¿¤>øRV‘æ%kl›ã(øtu,š4P¼ÁËé ¶Í¦»bµÉ¨h È=˸´Hóß}Ç®ÒÞÏ”2AE«;]-pà\ hÇ0ãSñƒOÖy Å&°×õ¥ S Xã ó€ÈTãÁ'ÝÑ£ rãFVr 'šn^m?ˆ<‹÷bÚn¾ ÞÌdX<ç‡r>_bÁƒ´dZÒú7J$Ò]C8\ئsµË5è&OMòiÝ|L×è¡yäO`À5iø…™]|zÁß5þÉ­eè>^dÓâÀ†¦X-*”Yˆ*¤Y˜òŽ¡T…¸w}z÷_|`ü endstream endobj 986 0 obj << /Type /Page /Contents 987 0 R /Resources 985 0 R /MediaBox [0 0 612 792] /Parent 969 0 R /Annots [ 983 0 R ] >> endobj 983 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 432.578 102.639 444.533] /A << /S /GoTo /D (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}Surv1) >> >> endobj 988 0 obj << /D [986 0 R /XYZ 96.768 705.06 null] >> endobj 964 0 obj << /D [986 0 R /XYZ 101.643 436.065 null] >> endobj 62 0 obj << /D [986 0 R /XYZ 97.768 396 null] >> endobj 985 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R /F29 110 0 R /F11 183 0 R /F10 255 0 R >> /ProcSet [ /PDF /Text ] >> endobj 991 0 obj << /Length 1196 /Filter /FlateDecode >> stream xÚ•W[£6~Ÿ_‘¡ÒfVJX áV5ÛÕŒÔj«Jm*u5˜àØÔ6­´?¾6Ø€Ãî>DÄøÜÏw.ü|¼y÷àïW©›FA´:«4vã(YEaì»puÌWwåÛ§ã¯ïvÑ„nç&q¸òzÖÒsø‡_zBCà>q/ДH‰ çD$Ájúî.ðW[ßwÓ0¼Ô² ƒðî§mÿ,ZœqDð?^èý½éß}RÏWˆN%gê„rý‡qÀá¦Såû¡ëE±©Šnà!“×/ ©ÞÖAºVü뢂5§m (ONðxUkÁè÷Ò½  +<8ÿøëÞzŸ\¸óC/L\˜dÑ„lbcENR¿/þßÉÃÖx!ìG£}˜`8o`'¹"¯*Ñ-kA507“áP“æ4ù³d?ðS7Š÷:»…%ýÛ/¬¶ÁÎMýž®†¼$¹™çð¬t=I»¤óZ¥àôÜô-?¼Ý¾çÝAJ ]ÔÝ¢<»TôœC¼Uyr.œ+Ñ c¯€b)ÂùWŸ{V^B…I\ 9d¨Uÿê_½¼`T(ЉbÄú'k›†PU¨ }¯k$%Ùhš-P˜‚4ÐÌjÄÍ<w='`aÄí e xEM, Ìß…nš¤‹(ëë<¹Ê¯Tú^E‹Ô5Á³Á¢Û2HÕ;N†W ÚÜþ¨òðþãŸ÷Πº*E|­D¢ª‚' *9ÉÚbÑ·D«SøÜ*-ÏJù ¹ ‚ÓàÞNã©ý¶"À‘­ÂYjS&½¸·äÆZœ'K†<£\›8—èñtow$‰ó-xeÝÖ5•#aÝŸß¼Qiò.#-æ¢HÇ뫪‰.íš@'ع|A¦uÌšÎ8i¤Ù `víâC[·•ÈüyèÊ!Îà€Û¦í‘¡Á^s£GXR³ÍYØ Ë[’²u58Îfôj£ZoºI™8e¡Ø'¾C•¶žÑYC8Ð&0ØÖÆä&BIœ8Â&ˆ0%¯ˆæºµÜ Zf2}zÜàŒTß“-k¸ëµ&cÅnPJ¾ žaźŰÙ}·–¸Ô}û%û&ô׈15wú½g,ü/_t‡sq[U&ÉHÖ¿3-¢°Ã¨÷QY„çLnkÔš-¾ï&ï’Y™Ø°¤Û,ÛF,ÇÉâÄjé…ñÕŒ– “3K§N&Lm‘ˆ—ÓÚæPž+hóŸÃº¹HæŒïöéÐq4 ÄjƬ6ÁHeN\£®'V«Ù67ãavPXÃúY^¾8¦ŠkH¼¥(­sCÀð…Q RQCœ]Ǻ–MtôP›5êÆˆªYe&T[Œþk»^ $ÔkHQvMödïÆ359ÊìD-•èfêÐeýƒÚRi¦Å#rÀlul¦a=LX(ïÓ_/ï[4f³À–ÇksL!–¢_Üdõàt¦YIÚ*7÷$Ô3J4°1‡§a½Ó·²ëIŸ†aìÿ(½?¼Io½øÊñT³1™î·ñ^LOù‰$‡mÒ+òORÞÜoþ )l endstream endobj 990 0 obj << /Type /Page /Contents 991 0 R /Resources 989 0 R /MediaBox [0 0 612 792] /Parent 969 0 R /Annots [ 984 0 R ] >> endobj 984 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [149.449 653.748 155.316 665.704] /A << /S /GoTo /D (survfitCI2) >> >> endobj 992 0 obj << /D [990 0 R /XYZ 96.768 705.06 null] >> endobj 993 0 obj << /D [990 0 R /XYZ 97.768 657.235 null] >> endobj 989 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1001 0 obj << /Length 1422 /Filter /FlateDecode >> stream xÚ¥WKÛ6¾çW8^ Õ6kE¤ž.êCl€=­ $ØìAkÑ6»2劔7ù÷åpH½,oÓô`x4Îãã<ÈŸ×/Þ¼ ãÙÒ_&4™­·3BCIgIœúT/¬‹Ùwu½ˆiì©=“ IÁ>+Ë|ª(¹`I¹¯š²@úÁîà‡c%%(Ùâbã֤ʕ¥7ùÁRÛº:´†‘(r•»E»_ÿ2 f Büel}å[øÄAÉÄNíB š"¸ºZ‘î£oþÇþ×ìØí»Fø›­½\‰îCªêØmš?Õ•ØY_U÷›÷€"hß,ʡ벪¨nÿ»/‚M/”!?tÞæJÕðÿÑ®ÍQ©19þ¡ÄU©•!YYò²÷Þ¼#Q/™Z z»G‘0鉄~–FZÃiêÓ–«·ï›êpl›Ð¹ ü$JÜŽYO`A‚ØOi¦‰Ô_†××YèéT “Ä;ä×4ó|¥WÈe¹ä¬FZU¸†rìh¹u¾Æ#.B솽çb'‘æÂñ¬v $W¼y‰k¨´¬t²¹Bá⚀°,‘¸ö&:3sµ/ò—Q4LÈ'Ð]Õú,£(€ÏâŒÈRo”L!i”E$ó´â™ã¶¸àŠƒ£ æ2rQtÌF"}2‘lTU[àÿÆm t4­-¯ØŽ9yb´Z+¾öGÇhþR÷gׄu§_†6±žö¬66Bï #2:Òµƒ<®^ HTK…«F6‘d‘·vFŸ„D¹.Hyâ\äísX:Ù\‚µÓu µÐ0kR$üÜ:&t™²Â˜½ß0#påu ‰8Ñn .7XuÃ_b=FµUL ³E>  СgÜö°aÅ6ÍR'7‰¾ë/µL=¡¿¨«ãÑ1ˆW 4vp ¬ýÄ«¥ËÁŸ ΜA†è‰0s ‡øÃ2œ*ðE¥qÀ’)8/`:݈ „Ï7¢ø¬MèŠ2? è¨ fè@Õ ‹5 üèòìÈÅøÃλZm¿=7O‚ÈO‚Ô©ßNØ_d¡ŸdÉМÑ GÅú÷?o§Žbr¬ôx¯í™‰Ӵϛܛtv›‘]¾e¥dбƒÊý#ÿõ%ÀFþí¦£¾ô¡ ˜í…>‚îÝO¿þq;1‘7:sÆÐLöšq“snýFöÒÕ @šém4vÓ×i-ÙI»Înä<ÅĦ*͘áÕŠ^@ˤNü\v‘Ø¢6ÿ¯ÜqâM­Î…äÊ4(ø>2{‘ÍÃ_¬…4ói0JO.ŠÏÃø  í»sw·óŒzÕOƒ—X<ÿᲦ ?SnËÈ'ÑóÕ–†>¥ñhçµÐ·sæ¹£z¬>å:Z6:eïÒ\}'Û jÅÝ ±…CÍw&ÊüRôÏ„{?:˜+‚Ësªt¾ã¢KßWö¸lË+šcÉ7:¬¢ìÇlàHKe®çó·¹0cîó“{&8û9ÙÁl’‹»›6èQüÀͤsž¦-,½Š¿rF‡/H2¡jK‚f÷ˆÉ•Ë]‹ „ÝvÒ{Û&dOóè¥ràðÆÄÆA¿·œw7äXL]ï¹”&íÃÇuXLvœ‹jégaöü€Ê4|_ß>¬V¼;ý4ÃÑ0…CHá¢Áú¢_A«¥¨LwƒÖ€â£ùÒ Î߉L¤-‘,L] ({b|·WrÀ“özý¬ª·´1BLú“åîŽßê×H¿5@Bc?6拾1þ¢sA9IæëÕ…œÑoâS²œÕºÖŸÄôß§x01¿ÏMYmÃì쿉¾à ªÓ8ËPŒPãõ‹Ûõ‹†Hzû endstream endobj 1000 0 obj << /Type /Page /Contents 1001 0 R /Resources 999 0 R /MediaBox [0 0 612 792] /Parent 1004 0 R /Annots [ 995 0 R 996 0 R ] >> endobj 995 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 593.973 223.31 605.928] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 996 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 510.286 102.639 522.242] /A << /S /GoTo /D (survfitCI1) >> >> endobj 1002 0 obj << /D [1000 0 R /XYZ 96.768 705.06 null] >> endobj 1003 0 obj << /D [1000 0 R /XYZ 123.92 597.46 null] >> endobj 994 0 obj << /D [1000 0 R /XYZ 101.643 513.773 null] >> endobj 999 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1013 0 obj << /Length 3229 /Filter /FlateDecode >> stream xÚÕÛŽã¶õ}¾ÂèK5HÌåM¹èM‚Ù z™¢ 6û µ5cíÈ’#É;Ù|}y¨›MÙžd&ižL‰Ô!ÏýF~sõâ+&†ÅÕâævÁ¤$Z²…ŠÂE¼¸Y/ÞD›ë·7ß¼øJ¨ÑBAt/(.höõ‡Û¼ýâõ2_¯6Ùê?˜@6”(Ù‘Û‹¥‘$æ‹%cÄÄ߬iÓºmÚjØŽQFUû1Cv¼sst±ä‚vúòø·y™7›ÀtL$;:£„3æO F'xñ•“‹$LÀ*M¸¸êuy½äFFí&³m¯¹ŽöE›[J´Î®ÒÆìtÿ¶ÄoVi±ÚýêÖj_×Yi¿hqfWg®c¥EV®üê⪭j.“(®ïðÝãÊúùhLCPƬna›dƒáÉöM^Þ!&%‹Aƒ$ZïWí©/¸!ÔLÙ_äÛpBD·U½ œ‰IC4µÜ€_ÿæø%:z:VBX¬Coê"¤) Xsx!•ÄÅßûEt¡ˆI„¶«–‚0P%§Äp¯&Í_pa2ZpuÁ« EØ APà°HÆ .zÀž“DõÈ¿>†’1ú“7ê¬?;†À8Q}Oðs0 Ç@$Qv¾õ†ÔÞÉ“)­&¬¡$ˆõ4<Ž/ä1bOstóøë ¤p˜§¥?’Ÿ po`}—\ƒf{+üŸMVg¡-8¡">Á <êÏ7`Kb¥ö‡E`ƒÀf>໑ýÄ5å:°ŸNˆ½Ü}ÌØˆÃ ÛÍ5³æÞŽ<ƒ}¹INë5k›¶uþ¹^Jc¢Ï®ÁJ¶8‘¥+ûí?ÉðÄ_pÛä[¯ó/ö³uöe²Ì¼““”$zjfY¶F¶‚•Âe´ÎÓ»ªL œÈŠlë·jp¾º ©¸iNãÓô‘ƒR¼³?-¥Ó­%gDP96‘eë‚mÈ@¤ñ¸¼¿^Æ:ºI‚¢l1Zy‰ÔÎx=ïŽÀÈvÿ7€B ì@L93£xäØÚp0À¶¤cÛCl)¦Ƀ8 –Œi±þ.‹Y‹3…5K]k.ø…Ô}œ=#@wÑ, ‘}BOE:XûaÙ=Åãéùý £Å!9a%à5&°TÛÈ`™pF2N¼‘|˜1’6°½Î}T# ZëŽ&ÏrZÂ0^ŒÖ®ªý` ,Es÷vï Þ;аH5N€ñ€ß8ªÞ5YýÁš¸loÚ:mò6¯JË×¢¥¶C0Cn­g[ƒ_é5ô}ÈpÖ˘µó'áAÖÅy¿ü>è ‡äé¶ò(6{8: ô|Jj A†“äPäÀ³Oƒ&l©‘rjÊçOOŸ’»§³"IÐÂo÷x‹¿gÈ”Ày”:A&ÐO­'îÖ/=ð9–Bó“"y5mšó®zyEr)ABœ{%úXêQFý€ÉÄwÔ@u„êO~·éBx‡òÚ3BÍ1ÂæÆÉñͬÐ\ ­½rtú‚zÖ@ ԫ΀=ë!ûc”M…9µ+µ‰(CбèŸ>` pròAIƒe?Ž]ÑÆœ]ò^¨€à:kVuîmGºa|‹"Œœåt³Uã_UÈëÀNÍÙH ÜÒ§4ÀÆæ#ó³”TDw•KäÝ®mø¼õg +¸(.Í b@~•ôm«“À`NuP› ÀTã=?‚*…â#ytã‚ê!¶•”ƶÆÌȇ0¦ÉS±-dÇCÞ›Ö~·¼—Û¨©ðE»q&F£ˆžl>’¸|dö |·³ghö[’%¯=ÃG+¾>¨?ÚŸªô§ÝX¥³+á ÌÇL ™àfÞYT”Ei›¹°ÃÊ29æÈì,̺šËvR|±ªŠ"Ý5™Ÿo=´P0ø6Ýi¹ü[–»HÞ8I/iòí®pñ…‰V×K¹šš}z°O^i@:Ý  ¡Iöºw)‰e’˜§/9l<â©7z†»d}BÚI"Љ–KÁЬ¼k78æ8õw/ºT† ç^áSZäº\Ð>:ƒ Ó£8þ¼òy_–®‰;ž ôËÏ,©)Hê»}ceX8/Øž´ÎSWŠ´o| Ô1 Þ`¹Oc¹¯+êÁ‚¬uǃÀ*Ì–iôXºÆaUâ”)A‚œ¾¤N,ìAöEæ2e}1c¾vrã?°ÍHˆ…zÆ”ÈZRp%â@ÜgóTP7-N‡šhÞÛ8Ìì}ª}‹ôÞ¹:hæ™= ‡Ü›¼„œ<Þwb]ÂÀ‰½gž¯ï³&X$`2m’C®ãQùw°"¤éA|qÔ‹çÉðæ«¨â4 ÇK„ÏŒÿêb”](Y¥DÇÓæ>äÖl‹a´æÉü8,¹úòæêÇ+Ëmº` ©™Y`³)ÈÈj{õæ-]¬aò›Dÿðýƒ[º]pð¨ŠÅ¿®þqõ¹kýLüµ}ñ…y³½‡-Bñ¢M )<0.…ˆÌœ*t2Ø”[6Ènç@­SËÞ/­ ?a~ÒÖ9-JƒR,͆CG?Wvò$yT(?ƒ°½—6n‰'¤MqXÉNI¸r &kÌ®.‘¶øœ´qÁˆPH'¸ÈÁø¨ò›H8ðøÒÊû‹Ö4^†dNiq!ƒ~+¹‹ÏÉûýä,5çÁ†¯¦ý1ÔÎ’L¬™•ì¹!b\þŽô>RsëL$²\a›ç„š °¸®U1§æÂj¤E]oøZ.gµœYÓÁÅ‘–‹V,…-’síÚZ”™Çpä’‚?–‚Ö[X›É—‚€!>ÉéÊ4ó\a`tcá<¾¡â‰¸¢KÆ< 5+Ц©ì9'x¸ƒAÙ2Ùå„=×a—wä³t䤊@µ>64G=ê¾Þu¦ì*O—]Õ3¢õ+ýËÚ%ÏÛxhо*Ñ»Q‡»"wfžÍÈV¨Êaul" ‘ý&êxzØš^¤…€tLä #|å¾3¢:qŠ—+¢Œ^†øn@ûŸ¿D ùÓ¨aòhÍÀ™°ä1ªø>x¡Ž0Ʀ§G“ÅüéXðV8-öÓÚi-&ÈãAi‘b&¼<·¹n2näÂnÑWrÑRœéËjw[ëWç™¶5C'ìw•4ÏTÁìÅ&ÊOà 6»çÅ!­ªb¿uŸ™«±³ž¹ùa“º®:¾ ]¯áœð¡1WµÙèßOè–Ü$37=b!]CƒOêʇ*ã.µÀ¯Tlþ6„/ .LÍÿ­Y;ãÌ]ŠùXò²xN”ô=¸ë]ÓŠa:|Õ@úò9,UVmù´m|ÍAj.‚¹ZW]ft¶}&@AÔ™æYogmyZ$Ñ—?Únj¯6 ðQõÃ=×Ù®HWÁ:.£¶B§~ûë»ö\¾É@»"<¼*RWüwÊv8rt/ZUÛÝ~„¯ïzÁ’•oüº×y뤢›¡}ƒÕ /5îžõˆ3eö“k±è¡Î[×G2´“x[í[|µJëìv_ñ½¿ƒŒ‹3|7n%À£m%¸FgYºÖW8Ÿ–ø\£YnÖMŽ—Wp&ûñuÝê$ºé¸ð±½W7mw›}rù.«mUaoWJ5ΘÝ{ìu¤Uæ°¯Öu\,ˆžÖñ©ï7(m¯&îÚá®ÈWÝ1㘹«ëvAwzüt䨦{¯³:ǦCÛu–ÆH©ÓöQ=ì"XÜ‹`'xHÞø6‰¥¼ëŒÐ³çn.PN3'n.(°6zص(î»nKûâOOQ?lò¾{ ŽKp¬²*—?gµ?\U:„ò}cßbÖ_A…‡UU{LíwNQ'BØ(ì—kÏLþ%04ÿÝÞ-ÓÌŠ+EŽ¿þsß,´‹îóžÆ)«}Smýh½˜ ²›­ó~kš~ÁtK/8qoñ\dïSlrýÀ„,²[ßs·B\/ÓßMý)«W¹ëÃ,’YˆáR+XØûOØ¡žklâÝ©ú›³Ì]U±^&鼌½ÔµÅz… üĉ< A’šÑð66}šV®ªb¸`ŒZŽÖ$òRå-D~Ûm v¶?¡“¡U[|ôª ¯°ç+»‹@=Ž#Šbf(áBMé‚÷"Rf¢Î<“‘Øzr‘žz:À¯Oº÷ËÝý•^ÜídkÌómê—¾G²Þß—ù­_7j÷×`tß|¶®õ¨=:Pw‘ð‚SJ„½†É Ñ]oƒqÞÕôþzR¦, endstream endobj 1012 0 obj << /Type /Page /Contents 1013 0 R /Resources 1011 0 R /MediaBox [0 0 612 792] /Parent 1004 0 R /Annots [ 997 0 R 998 0 R 1006 0 R 1007 0 R 1008 0 R 1009 0 R 1010 0 R ] >> endobj 997 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [238.365 653.748 244.232 665.704] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck2) >> >> endobj 998 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [248.826 641.793 254.693 653.748] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}startstop2) >> >> endobj 1006 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 605.928 218.08 617.883] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 1007 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [190.795 199.452 202.75 211.407] /A << /S /GoTo /D (equation.8.12) >> >> endobj 1008 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [137.303 175.542 149.258 187.497] /A << /S /GoTo /D (equation.8.13) >> >> endobj 1009 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [435.09 175.542 447.045 187.497] /A << /S /GoTo /D (equation.8.14) >> >> endobj 1010 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [100.646 139.676 112.602 151.631] /A << /S /GoTo /D (equation.8.14) >> >> endobj 1014 0 obj << /D [1012 0 R /XYZ 96.768 705.06 null] >> endobj 1015 0 obj << /D [1012 0 R /XYZ 144.841 657.235 null] >> endobj 1016 0 obj << /D [1012 0 R /XYZ 144.841 645.28 null] >> endobj 1017 0 obj << /D [1012 0 R /XYZ 123.92 609.415 null] >> endobj 1018 0 obj << /D [1012 0 R /XYZ 495.768 308.965 null] >> endobj 1019 0 obj << /D [1012 0 R /XYZ 495.768 282.44 null] >> endobj 1020 0 obj << /D [1012 0 R /XYZ 495.768 255.681 null] >> endobj 1011 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R /F1 256 0 R /F10 255 0 R /F7 257 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1026 0 obj << /Length 2254 /Filter /FlateDecode >> stream xÚ•ËŽÛFòî¯ÆÀšF4ßìÊ@’çÙ1r˜øÐC¶FÜP¤–lŽ"`?~ëÕ)Qì©««««ë]E采7ï?f‹ÜÍ“ Y¦º8—¤SwÁÁ œúÊ“å¡Æõ ØjëÇBX÷èiÖb9rÿ«»šÓ÷3%a”8T-©l¾Ã Ȉ‘9JÑ@U •Az2F”²DHñG¡ð×T"ÉgÀè¦:}Áܨý¾¦GÞbª¦d¢çеê-£jy“HŒ·™ÂÄýcÍ‹œ¦ÅŠž'ο‡ž ”ßG¥žQûN÷b@>BCLh9Ðõ_FÈ®  GÄ€aoVÒOazÕ4p½ðTú*®¹Î÷0rÃðÔWU÷<ìFÒ¥Îî\ØñeÑ1¥DÊlÕä2à%nÓÐÄ@$rä‰'"/=(é‡g‡m%iGØZ«R.µŒÙ¨ÂЬ8q]áŸDÎ÷ ÑVwLɯ-U³+á–zRÐàà‰cé WrSík=çáŽÓãÁÏîóTmüm ü­ ÃtFCàŸZ§o;¤<#9üøÊgí™WϘí´1!MƒÓ^â&ç›lÆâe"^L,ÿŽBQ—2 ç6jw¡Lv”‡°Š‘'u0a°êÁ™’[ŠSí°`áÌö°µœÐÑrò"˲*´}is#®Ó48E,Ìuý³q&'*Îc|g+òa2ØGhCÊÓV5¤—èNÝ8¼¨rô4¤CÌì0A°á\ ûnç'™AË?gæäÜõòp"2²–ºs Ý'‘‚'\B’3éD°û ¦»IvÊûR+³ý%JœË´5ŸÐT Щ>P¥FŒjxÕ6¢}ç\hàXXà¨À&"ða8µø9f©cXqÆtºV3É\/=3åM»Gix*òQ…- ÿ=æf`IEñãüâÖóV4b¼‚OË8öAÀ¢˜9LÆÁêfi|®ÔÝ˦2?~ZÉl³jër†»v‰Ï_ŸÕŒú¾kéÖЮ½ìbZ+[+´ú±ó¯v(ùÝ‹=ݘîxÏxf˜ ÙTTds0#}")›?£N‘QxÎØu „ånìûÓJõv®l¿å׺vÀ2‹`»Ùˆô]‡.F°À˜d`ÀAw`PúNÄ•KæÐ ª÷ ªN®*ø†N;U¬oÈäƒ%q(zšõˆ!}` t74ÕÖÆWÐNwò’U‚b’h Nsr|šfX€#eµÙ@+¸mœ¾­_¬ÒT“z’ÛÒ<‰=|3šÓNIlpKžá¾á¤œSYÍæn6×7kÝ<›-Þ¬õ‹®{„6–ÇŸd¢ß§Õs§©Eá†>„I«mÛ[Ëq.Uñ‰?æ_øuO´7g¯Éw÷ï@Á>v8Þ’TS.}µ»1^ؘP½ÛÀäÕUÅkºJ´áLÅf­f§RŠQ«Šµ¤9îµ cÑ‹îÌ7$EA–ª_?üúõ'A”ºX?kóyo‹Á˜åŸº¸›uÂí0˜*¹¹ … ts_"A&‰ùPà‚rËû_vêBuþñoz®¨Ôtía-ਚ5E[¯ùñ[z½|™2ççȉo_ìOu3÷¾¿°EשãD¤²Ú­É0͸ŠN77#_ªrúLß²·Ï¥… ð-_ÿÒrg7 Ìj"(‡)[ã”s`köZ /´N,çÜw3/ÚVúŠ#«šË¹´LšG÷ú)¶ »s&½ã€åk†™ñú£ÿmŠ­6ð([B]!ðeô¸'ë—[3aÏj=Â{…µú¤•‚~œÆÎ[þÒŒz®XÈv; ­I*ýúF¿¼Í%O!úÁ¢ó”¶âf[ Ÿ¬·µºF—³?ä]³.©ÝÎr>à/MFKHCéx²ñc3ÜÎþ—AÔ€û›Ë¨äÍÕøòÁ»YŬñȩּúß5ãv1Ï书ɫãLš¹Ñåïfò=3‘ü$4o?¬E¿Fm+‰ÆânœØn¶Z|åtãT-8X†[W¯]=˜G¾ýíÖuSéòº5œT\¯«³A§„nˆ³_ˆSmÈÌü DÊ7?=¼ùV+½4 endstream endobj 1025 0 obj << /Type /Page /Contents 1026 0 R /Resources 1024 0 R /MediaBox [0 0 612 792] /Parent 1004 0 R >> endobj 1027 0 obj << /D [1025 0 R /XYZ 96.768 705.06 null] >> endobj 1024 0 obj << /Font << /F8 112 0 R /F11 183 0 R /F13 311 0 R /F7 257 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1030 0 obj << /Length 1334 /Filter /FlateDecode >> stream xÚWYoã6~ß_a`±…’µQ‡ ^ ]taô¡0ZçÉuF¢-beÉ){Ó__IT7/&Åc曃ߌÙ~xø§“e°œE³Éö0AI,4™¥ó ;Û|²óèáÎO£Ôû+LÊ\HÅwtíÕâjŠut·ßþöð %CY( ÂÙb*)uÄVŠÌ‰J".K5ÉHÅê†äê« …I§ 1AÁ2Õ 7' °4ªŸ}5näÐ>:´‡~AoÀOæA8o¡÷£Y€ÒÈþ¸›Né…\óÑÞæ²yÖ|4Ý;P™Û3uû¨<Ï‚9º§"ªÈ>GJFŠ´m·Ìa¶,Á¼VcC°åµn¾»¬ˆ‚÷¨Ž=S®&µQÕR†«Œ¨/ÊÔȵüCݘ-ŽÏçòÅÞ>Ò‹Ù­9.Ywi”gŒcN˜ ¦UþÃŽH³¬™“ï0o(û¾Ÿª–æ2•|”ÎSï#/ jÀçšVœ4#¼¨™>ˆµñ ÔB£ƒ]1 fÁ…d¼nŽrÌ®|ì€Í¡Lšº“f™ÝR¼è’­”À€)kO0šw1vJNpn¹Tù„S’szÒöüÔ»·Õп„ÚEôÞE’T˜í^tá‰. ·£8DZqH1Œ6zƒ%©Ž\Š9ÃËQ,‘º»¶ïX1ª/)ã°ü*ùíHH•(A‚%—v.ô’”{mucAð>TLÓ¨%7uZD@Ò¡,¨0LÖ ™Üõ¸VQ…©¦-§À:†¿?Žò>«OϴœÖ&­¯”úÕwüqnÑÇ.îšS|ìPÙØ‘›úúg{bý)G#ûêú¦ªèµ.מßåù…4LØiS›öPEHnJ\çœ'ý*½ñ8íLÑÎÁ´É.t1†UéV+Ôƒƒ’„F)/¼à0[Ö?FÀÒ·j›üú|KRE®2%mÃü½ó#†&‡®²äk20“WŒ²æXOMú>‡­ùL\ÞÑxê2wà±^¦!s ]m2løîÐÔ'—ž'[ÁÓ§ûOk#ø@¦ QõkÏ9Àæ«NØ'ÅáãÆa´ì÷Ñßܯ#ñò!v®*àìM89GýÔÎ8m/¤õ&p=£ÿ‘²¬—=›ú]nÛe‚!rIÁÒ*í­Lyeþ{oÂ߯Ú$ðûU;n|¾­z”~CÝûq¼ôÍ(NÎn8“™ æ" ‡ÜýÎÎ4«KÃ¥ºÌuW”Ø©ixæÞ?ÁÇßÑ]¶…Á2rµ®z ýW;ÂmŠm߈¬ìNI¬3Âeì,ŒX˦†’òû§‘R.›¾»úý*¼r´dº³•ºªZ3çqØè•|U÷®ð«Eeí©Àÿ®ßr­þãòßbpnÌœ+‘>µ…× ~ÂÏâe/¼úNü\àG®Ï¼C{‡¼*㲪ÁÂ6¡ØÃ­–AEƒT'Åö܃ÿjµhêöXh@Û0Öƒñ¢^ôIì³L0¦P@˜0>^foõe½* U~±²•W}¦Rc‰›£ù™Ë$UxÌgE‘,¸ó“hæýA„Ï¿`þŸÀØ2Zõ´ÊIS¾tß_ýL!ÉÉÆ…€*œ%/ÎEKÃ1 “’a"Xñ æuÕ/ªINžÛ㱓n›³²T8#p=\½8ö¾Âxµ*šTùMpÃô W#W@ã+2&rGº„At |ÿd”ò( ƒx!<ÇÁB¤›„€¢6?üºýðäÀH¯ endstream endobj 1029 0 obj << /Type /Page /Contents 1030 0 R /Resources 1028 0 R /MediaBox [0 0 612 792] /Parent 1004 0 R >> endobj 1031 0 obj << /D [1029 0 R /XYZ 96.768 705.06 null] >> endobj 1028 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1036 0 obj << /Length 1393 /Filter /FlateDecode >> stream xÚÅXKÛ6¾çW8Y ±Q[©‡¥¢F‘ ÐréI°õ–h‹-:egüø’R¯•f»E/&5r^ß ‡þùöÙ«7É$õÒÇ“Ûí!ì-žÄÑÒÃA4¹Í'wÓÛ‚Îx‰§•õ¦[.`ò§ùÕl¦’ˆ™ú+rM+Ékàȉ$0cŒösäžÓVÂy†“)½Ù"ÄáôÝ, ¦Ý%Rê_ Éa;-«Ú "{Bªz3[Dñô/šÉ—Uç(Êv¶¾ý}²@¡—†¡‘—FÖp­o$ÊF~¦¹2-}%BÙ©¥Aª>è=0DŸv2G…KÆAif†š×öTíIÅ+7“¨|kÇMEÅi¦4'’ñÒÊ&¥e;aæ4ÈyùÒ)«-ðµüàTÒÆú}3+I¤Q90. ‚pÊdE÷[…0„økêžTøÀ3Á´ä–*)X&÷÷@Va2#‚‹9Ì+{4/DT”žVêÕv€¨TÄJE”€Š°q‡%ð’e¤Á†Zœ¶LþòÛ‚åYA³#g¦¾‡ÍfŒ.¹`¤‡â!0T"ðXU±r§§,W¿–¾|‘U^Yï÷í²aÑ'cßK1†‚5ë ©}nöç0±«´”îTýT3A-IJ=Qò}|ë|Ñ^8É8PFÖ¤2ÿ Ì?.`ä"§ôžåÃݯÁD¤èÓÃâ¤ÓS£ždj5<3Y(äƒZõF§áµxÔ—ŸiÙo_[Ùš¡+šùH˜µ™Û˜ìÉnçTR›X¦4¹ Ôt+òíëÆeÃŒ!Úß ~º3f¬´Z¹2¶äï`x®AœDÔgt7È1‹7fOÇ3ØÊjcùO£fwKÊ{S½›=VôûžÏW]*Äßùe`(M¼e¸ÚŽdØ-—^à¾Vç‚÷Ýiíü†+µ¢ç‚eſԸñG> endobj 1032 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 607.92 102.639 619.875] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck1) >> >> endobj 1033 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 275.168 102.639 287.123] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}startstop1) >> >> endobj 1037 0 obj << /D [1035 0 R /XYZ 96.768 705.06 null] >> endobj 1021 0 obj << /D [1035 0 R /XYZ 101.643 611.407 null] >> endobj 1022 0 obj << /D [1035 0 R /XYZ 101.643 278.655 null] >> endobj 1034 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1043 0 obj << /Length 1380 /Filter /FlateDecode >> stream xÚÍXÉŽã6½÷WN6`sDjâÉ LÄ@¸}%ºÍ´,½t¾>ÅE Õ’å’C£¹«_-,ùûõÓ‡O®?‰Q`²ÞO0qQL&"ël²™¥g~¡ÕföÛùÒ'þ웥þŸ•j‹<;¾óû†Ùm· ½g椞W"çJ-ÂÒvýãc9A4YbŒbߨbRŽÖbê䕲—ƒ¨¬5%U™ Ëì=ŠöL&,Í|ø„½Îý–ØõQŶÕhéÃ1ì€81;­Ã ::\…þÄÑ0æ>~^îYÁªÃ€ÍÈGn0£Ñïiôj•…(v]çWú†ë3/ôH¨pZsaø(D©G‰aDc4GÞNæL¹ûƒ¦BYsŒÀðÁ“]-_;{.RÁÊB:;ÍÁO†nšÓ#-ÆøÆ‘ N&õÝö—_†ìaûª‚O5€Æäfƒ·ÛÍÆ˜Ünk£ þ:Šöú¨<È*tLg79‘Zå¡1´NŒ"7¼‹6ŠPŒÛ^‘Ú\i™wÍ alA y79Ê“Ó)ë3Üeþ¦ÔiGªyúO§D=¬»7^^WëE‹tU¤cÙb):ÉÒg…83=Ò¼¢JI€°?"êDì(»ÝlŠ“ ] zKþ¼5Üu©{PižTU×»«ÕT$»œN[Í­so7ƒa€÷œ/â0—}ûŠJù6ä'‹þÀ*jCžraÏâë¾_jŠMéYÍ—.öÛŠ9øÍÄŠYìõ2œi£ û¡oÊŽâX=9†«©”œ¶ÊzH ÄYõ*uÀÁQZêžzHXiñq-Jj@MP«IiQ•|5®A Œ#9qz1„<@s‚0»¯Ç1¬Vã!"Gî?‹†ðbä“ðÿýcÃû:ûðír¾“£HËaHɧÖO »! endstream endobj 1042 0 obj << /Type /Page /Contents 1043 0 R /Resources 1041 0 R /MediaBox [0 0 612 792] /Parent 1004 0 R /Annots [ 1038 0 R 1039 0 R 1040 0 R ] >> endobj 1038 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 609.913 102.639 621.868] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 1039 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 609.913 191.928 621.868] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) >> >> endobj 1040 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 127.721 102.639 139.676] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 1044 0 obj << /D [1042 0 R /XYZ 96.768 705.06 null] >> endobj 1023 0 obj << /D [1042 0 R /XYZ 101.643 613.4 null] >> endobj 1045 0 obj << /D [1042 0 R /XYZ 101.643 131.208 null] >> endobj 1041 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1049 0 obj << /Length 946 /Filter /FlateDecode >> stream xÚÍW[OÛ0~çWT€DÚÄÎÕhyئ!mš¦IëË´í!4µh.KÒR¦ýøùJbšŠxØ“çøó9ß9߉ónzpqåú#I€ƒÑ4!CBÈ(ðCˆù›i:ú1>:>öÇŸ“¦U³kÖ^ªY’¦j 5¶sª&³²ÈXJ‹™~¾NŠ´¹<ý5ý4rF!Hüù¡?ÎË”.¨Æ+5^³¬Wkám>1”Ë›¹šge­&_MKƒ©ñÛð™Ï]dæ§ã; …Ü>CÂòâ y=ê ÃÀãÛå¶LYXäÁÈs5~  ›6…´Öq¼j¼åÇ›7NýY% {Çsá#¿ˆ,@VHŸ/°¦Ð™ Ãc`A· Œ€t¦ªéÊlÀ.$h“yV{_Ñ8>© +N¶0…°}í¤ 9 œÐö®¥ye{'ˆ8Ó•2@—æXD`ï› ‡Å²¦dÁÁ‘³(ï¨ÎE\åÉÚ˜Éø8ßp(b›ÈÃC`äF¶«Ëª¢5‡`…qÞƒ@À$ÛŸšÎI§,‘Ó¸ŸÜ-Tƒ(â¹ là›!Â¥èYÉ]”7[S‹|è»SK<ˆ¼ÐÎÐQ²*™nSæºgò“Äá’ìCµ–Ó¦Inh3T½ëµe–ÑEC­Ä12ô›±¯9aÒ×F ‘9´ _è&Ž ùå­žÐu¥ªJ†±^Khùæ|[ý^(«m^àWó¼Ð‹=õ %$%„&Z=ÒR)ÅÎ…ŒRIÃ*à`C¢ÿ#]€]Úp}H"²»íiýX~ÝÍK›ì^®ÕÂßîcÉ ¼ËïQSÑ3ߊYÒ ëFÐíîYT›µ<%B‡Þ ÑD}â8¶òñ÷¯Ô„ªl%KÃÚ4zPvg;žÒ*zš É~o¾ö÷ ¼b`xÿÀð¶’Ù·˜6Òo jm¨iøCM¼zãèZÅKzоñ%-¯i‹ ¹‹$Tì=êŽETúm¶%m[ ¾ë…ÍáÎ~.z~"q[ OBe8×{×u9é§Gã”;—ª•´”+3qƒä£´óÆÙ²˜µLÜæ… kÔ8§5"yxùßï‚òȹf0èº0 }ÓHõ/Âûü¶›WËv™80ðv°¤ȇÛ{¬J\ˆצ*-E„ôQù›0å´hë{óémYN']&—~`2¯úá®í™Ð“Ô¬îúíp †¸##ÛL#@$„HÜ.1¿¨FšV„CazðazðÊ}¡Ž endstream endobj 1048 0 obj << /Type /Page /Contents 1049 0 R /Resources 1047 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R /Annots [ 1046 0 R ] >> endobj 1046 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 127.721 102.639 139.676] /A << /S /GoTo /D (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 1050 0 obj << /D [1048 0 R /XYZ 96.768 705.06 null] >> endobj 1005 0 obj << /D [1048 0 R /XYZ 101.643 131.208 null] >> endobj 1047 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1054 0 obj << /Length 1237 /Filter /FlateDecode >> stream xÚÅWËnã6Ýç+ŒL 3¢~£E4‹¢(Ô;O´DÛœ‘(I9I¿~._2å(if5‹D—äåá¹OÒ¬¯nïÓ|²B«y2Ÿ¬wœ¤h•Læù%°°®&›é§›Çõ_“x2íò~r–'ùT´¯¬ØîvV B´ÂŠ%á^è$µ¢‚=ÛšFvÄÜ&õÔ:5TZ‘·å‰©sH )A¦opÂqO‚–’µüÞ²­OžÏ¹ó¶/öÛIÆ÷Ž· Dõ#b?;R*°p}§Ù[¥Ï³Pù ð*õ?Ê•x‰¨Öƒ!Ö ¹ª_ÕôDk©1v~÷)4D ö¬w‰WtOÅÇÅ‚—m]$o›ÀP5„@Þ5T0cœa½‘­PzÔqö½£ýA›(yô&Á"|²j.uNàŸ0•ØÒàRõÊ5|¯ú$³(ß2¡cÕpãkª¬z/,¯¬?Q^EŽÇúEKO*ºŒ=3Ä¢0’Å™«›—]söËm8£´òQZlðÖÑÂCpLì¾AÜFÎ#µE»%[V3åÊ¢#ÛvêÒ*G"cíÞ@–&>_øÆÑP²ù”}ì o%êÞ5š­¤âDBA~Hð8ÙSÔ´={nÈêº2-èz`ü×N*ïß´|Ï’mC[îRy=SJ®GQ¶?(9æ˜KJÖQ#´àõXê]‚hŒN~dØýîœ}åMr-0v,¯|ne)ØqÜ´»hèŽÔµæw';q‚å’ùbÐù­VÙ©UЕ6QØ#6þ;ãp¹3ùðÎsßpÍÍVɨ÷µ·£ÿKû÷Ó}adº/޾6:[ó ´‡¨oBAO—ÓÍb6p·3îÈS< "PÆç+ë#m‹>—uWÑâŸßõäíÙ©Ažð÷7ÿµÑLbx„à‹XA+ñ‘*\¸#x°0‹òÛ1쨪*äwÛã{…Þ?Ö±Éá<Ç‘`ò›E)|¹Ôÿvô0\+Ž‚{Ë) 3=<ŽJÊáö)Œž•½5çÈÁÌxîuÍüW!È‹çeç"ß³›Â\ÍöÚŒø0 #éþÙôÉ(^¤ÃôÑŒùɸhÈâ/u¸/žââ—zùögÁ3|6Ö!÷V1CÅÞéP)IŽ–+›%JôŒÁ´»Y ïê4™VÔî^›µ™K”®ì6ÍpmÞÂiOý×¼ø´Ðí—˜¬ßÂvýH„ézô%ÆYíd}‹ê¯:¸ Ùv¢t²f•,5+3¬˜0﨧ɸWd*äƒá ™¾r rk¿ ùv`ôâ¼ó£w2Ñ{ôif¹¦)¸« ˆÁ„nRúkAA³·&Í­©b××OvŽ>+H2x£ ¾’p†VÙEPï™Êš¦ïuœqø7™îº sóØ1ê@œ>\´–l©r° Oᥰ׸™‚´µ¦.Ó)º™eóåt}p§ â€EpJC¸wL›%…û†`.wÐ68‰ŽécöÓ^l­£ôoA›ËáoOøÍ‰³Êì{Þ°ÇÍ×Ç×Y›‚˜¦^·Š8çÀ浨OPpÜ‘õ}œ¢4niв$µ›q²Ô‹W®¯~æw endstream endobj 1053 0 obj << /Type /Page /Contents 1054 0 R /Resources 1052 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R >> endobj 1055 0 obj << /D [1053 0 R /XYZ 96.768 705.06 null] >> endobj 66 0 obj << /D [1053 0 R /XYZ 97.768 186.508 null] >> endobj 1052 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F30 111 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1063 0 obj << /Length 1550 /Filter /FlateDecode >> stream xÚ½XKoÜ6¾çW,Ò‹ÖðÊ"E½ä Ú"ñ¡@l²k¦»’«‡×ίï gôÚån ô`‹¤Fœof¾™!÷ÇëWW„Z%nÊpu]®’ÈÂx‘+ý`u¯¾8÷ëÛ럯>øáLÎwã(Xy$ÐöÍc©»L“àbC»±'I–€ÏlõV›@¸¾«n0Œt•íú¼Xo8¯Qég÷þµE§·ÚHßMÄ÷á«ø›|Ÿv~²l™xn¨‚ï™±I”ˆd‰ùóûß#¼£’/ð¦å²ÓûB^®7–çâuÓ‰3ë×q2-W¼ :OxÊ ½ˆ„'@º´ëÛÓ-2|Qœ®: ³26ðx@‚„ÅÃBz®L¢ÁÅ%AzÈØ_jù'-²]Úç"."J1”ÿZYVïúîRÖ]ßT]qà*é¼{”P[ú8ž“8€´G1X’LwͶßÕZÆN×®7¾ôœ®†§Hœî¾ AS÷® z›6káðx”(ëÝ®Æ=ºÚºëRžóaûNÝ@ZÑó}Xßgçk‘u´p#|õ­È_“ÔGzôí±ž¬6Hñ, ‡€ëš·¯K[ ‚PDÉà2Ð"-®a¡… #|ßiв@;|5."KÂÁXH«Ü‚Aåz‘?qŠÂ≱˜iRƒíÝ,X¨¿d`@.ñæ† ßw½8`°¥¦P"ò=¨³£çÞ1 µŒœb`„Y½ë÷¹Ä.§ºVYÃV4Ï“×Ì xÒ[ØÈÒÐߦj¿§YÛO4²ÚÊø¦šÚk5Uø£ómrq/`®e]Íór)0@S"0ò­ñB€ã=ò³h1üÀ¹¤o<¡š¶ãOwÅ|´`Âéq‡ áÃzƒ¾ÖSØQy}b¹­Ѿ¸äîëJ,ûGÑeVÿIå&¦×ùO¾Ô$ö:£Â8Æsdmñˆˆ'Wà;Œÿ¹°¢®¬jíKç ‹}W…cÉü¥ß£’;ãÕäE}¥ÿ§õI ü„ŠX9i–îz ±Âkæ+ã‰È;ËÌV+Î%mßAoáThš>Ÿ1ž¬YÆÄt}‹åÊ'Ë?“Ô<Š2*2ª¾k‹ œ‹ª1L9Þ[Nô¢‚è¥ë©Zô:tXaùrlz”p Lt:ÝÑ„>…¡ÇXam–ý°LÙO•î—à4s¯Gk@þ w¼G›>ì)£rPÓë¼o¨BÁ˜Ú|Ê=%àv/Ò)z)Bxn¹ÉC3Ž–McR5çwÖ7 g%“í{qÒ:K»€ÃüÔ.~J[Vu0™¤·÷³\šPLij!îcÔmmƒ´-Pi[ÃcO²ŒíXªç}RçDÀu¾ŒŒ6X¤nA» *¸q¨h‰ ¼41&¹n¡wÝõyX1i O+Êr9d9ö@¢Œ¦˜Ãäˆ^¸3Ð ¬4 ƒ©©…È2X‰†Ì¨-»uÖASsð](ß êÍ-xy=/°ZYø%7ˆF~},)(b¬\ÜñóãsÊÀC“H)xÉ¡æ [–Vn‹ÔÁ¦e¼ÄHK"™è›ÌÄ æ|¥“4+û*cwÂlŸþiªwKSs¬€g‘¶Ú”ý„=ŽÛð=ÂLšt»5‘ƒñT‹Á%|®0Tƒð=0biND¬‡R-¡È‘Eàu¥–·ÒYa9&Â-IMGÕ§/úöË×[K…¡?&«zdѼ²›ƒzÃåʵ\%6!h;Îé—ÿðŸîÓ‹kÿ,ø®.ƒWt¹G×™éfF!Rf "õLCLÛ™ÔV?"WÍEºßßaìçRM‘3òïEZ^½¸²]…±$ŒD^÷w;Fvq1P ÐUÇŠA ß½§5Psæ¶,ýÀõ§Æm»Ôl$ÐJ´*ò°¹~³@ͬ[" ol¶-¤òQdö£ |I¯ßÒm]îm 3ãO˜Ø¡ÈÌ xªË“ïð3ü³ã‚ºD畞Õy®>£x¡ñ¸Á*¸/KsäŠáîf@™ ä«÷ׯþ$ ï¿ endstream endobj 1062 0 obj << /Type /Page /Contents 1063 0 R /Resources 1061 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R /Annots [ 1056 0 R 1057 0 R 1058 0 R 1059 0 R 1060 0 R ] >> endobj 1056 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 617.883 202.389 629.838] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix2) >> >> endobj 1057 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 570.062 223.31 582.017] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare2) >> >> endobj 1058 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 558.107 223.31 570.062] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) >> >> endobj 1059 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 546.152 218.08 558.107] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return2) >> >> endobj 1060 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 215.392 102.639 227.347] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix1) >> >> endobj 1064 0 obj << /D [1062 0 R /XYZ 96.768 705.06 null] >> endobj 1065 0 obj << /D [1062 0 R /XYZ 102.999 621.37 null] >> endobj 1066 0 obj << /D [1062 0 R /XYZ 123.92 573.549 null] >> endobj 1067 0 obj << /D [1062 0 R /XYZ 123.92 561.594 null] >> endobj 1068 0 obj << /D [1062 0 R /XYZ 123.92 549.639 null] >> endobj 1069 0 obj << /D [1062 0 R /XYZ 101.643 218.879 null] >> endobj 1061 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R /F30 111 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1077 0 obj << /Length 1147 /Filter /FlateDecode >> stream xÚ­WË’Ú8Ý÷WP¬h7[~€‹a“J²˜ÕÔ ‹©êé…1+1’G²!¯d]ù…ytUV–¬«sßGÒ§ÍÓü«Œ"; Q8ÚìG.òìÂ`a#¹°Ù^'{ÆŸg &ÿ9CÖÎJÏÈ”³³™¼¼ÈU÷ùmóçü«ëwEË‘`Z¤£tE¶·XŒf®kGh=ÆÅ+yÓèkýÙ­Ôf‰ÔÜéÅ¢ ËVvôv†ì¦FÈí r\”œ*×¥EÊÇ[àa|ÙÆ–ëÈYÊïÂŽ<¤å>ã$‹9~žyž7‰³L ü ÛëE +§ç ˜ÄœÄÛ {@ýÌõíÈ÷ÜÓà)x¶={¹L*DÉO{R$d¶;.‘#Çýzš‚ôÀºJÐ,òl7ôºq$´€R™êï7ø~7ßï²–Ü`Læ–þsÀs’èIÆXtG,ôÄšÕD­,1è¤ G cœêvn‰'do´ÜÂûØ RâxnàH ¤ä›5µ¤G*›¹À‹ÂØCA0¹cl~5žÞqŽæœ8)úq£åq‹yW=ˆ‚å²ÖÀG"D‰w©cÙNqa¢Hñ¹šÖ‘<§æ€o©1R¾9U|,áq‘¤· nãY¹ð¨ N'M9ǧ8Ã4ñ“ ‘a»áìwÀ­TrBSúoÿŒùÎ$µàä‡Ǧ4R"úE’³:Êød•Z- [~å&ã¢Ü~«3áæñÿšeå”PÝê…­ªs¡ ö½ÂîØcôB²÷®a ÷® U­[Tô3Û.{‰ß‚7…cS9ƒ5y5àçJrš$Ÿ19¤…iœ„•u“ì/ù ²"ï‡6‰në5YáƆíû@¸ï×ðY½(¦®{C|RmÝì]aÀ{,fX¤à!•àc¼O IɶŒ*ÿÜ }—mÅ´<ñHJ’ûÙ[RË,.È ßæ‚A¬;™Ug…éPÖë§¿ªÌ…ࡳÊŒî´5i1«š¶*Jë[ŸÌ(or:måÿƒ™¶*¨)¸‰áø#~Ÿ Àr|»žJJþ/ñ Eòl(,8KgÙâ g`*#‹wàŠ;ESÛ¶ëÃóÃ5m%ƒÒéºÕ4 lXïcÜ ©¨Ï±š®*„ùI6£¹Dä~z-w†Q!†ç” úbŽòØ &°¼÷®n´ ¾ôþcšnAÿóåß¿`kFÄEØêæí+Q Pºë¶5”I|4ÅQ“`|¥ £T2”¹Ëð éõm=ôBsÛ ·^hòy±tÀ˜ªcz<ÓªiÔ@—ú²4V×*9©¬³‘uc5–,)‰±––.×ãñÐD¿ë·Žy:¯3„|;”*¯¤)—tŒ@ë©5Ö¶µÈ±ÉQÂŽ9£Í±w¯tx 0Ĩžª¢Õýû«´ÍÐö©5Öv®®_’”õ(5›*¥õNþ‹§ª¶ÈçŠìår©A]ÏQ’O_6O¿Ò53p endstream endobj 1076 0 obj << /Type /Page /Contents 1077 0 R /Resources 1075 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R /Annots [ 1073 0 R 1074 0 R ] >> endobj 1073 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 558.107 102.639 570.062] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare1) >> >> endobj 1074 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 558.107 197.159 570.062] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare3) >> >> endobj 1078 0 obj << /D [1076 0 R /XYZ 96.768 705.06 null] >> endobj 1070 0 obj << /D [1076 0 R /XYZ 101.643 561.594 null] >> endobj 1075 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1083 0 obj << /Length 919 /Filter /FlateDecode >> stream xÚ­W]oÓ0}߯ÈcÖÑ4vš´ n 1•!BiâÒ@¾”8-ۯǟ]’ÙYéxilçž{Žï½¾Nß'“‹¹á[¾=#X@k á¹3 :®ÄÆWó¦8…ssw:v<`ÖÓm6%_ 83‹$§FU5_^· Ó”¯k¾€7ˆ–b}u:v]óŠp-9¸3a_ðÕ¦¶ˆ!ôÌ@:À\ù&x Ð.‘´+¦ \mbó•‘Ñ÷à½1SËŸNÉX¾+¶Î¸‡9™3'dR¡2 #2f,_“·w|7ešD„€È%~'`Ú °Mü“þ7ÜÄñZ&Ž5Ÿ¹ÄÔMµ]'8JÆ1ŠÒ°B Ÿ¾myÓ=".]Ës¾]ß±€×ÛnŽ“ Ó(»fX_åýDÕ7ÛµÙ HFà%CÛ=\Iò^ä(ð׋›ËàG5&áP =&‘tg `]TèYo—ƒÅÛ EѰODOƒÖ·!bˆ½tru,.ËÁm8Êqu÷¶j…œyàÃ3þÈ•—ÇLºpy(¨FA½>ºé¦VÀغ·Ãd,¯¯)l‡õ$î@ZdI|P”P¶rÏ3<EE)ÂM{}&y‚“0哲BÛ0E9=êt>š¨ÂZ*÷YQfʳ"qR¢(ƒ&[¡ª+•y©•ÊdêéžÎ™f¯»2¡cù€ƒ%5ißE´×[—¡ŒkðtÀú9SÖ಄ÅJôs½¼vâ²wsÇh?„¸JþPÖO_n_ˆp°½Ë =@ês-3”…êbd´úrÞÇŠO_ÙÁÐ&hwµý¿»«èPZ‹èO ä³uBè\–@ÜdÙ];­CYÝꂽ vK}޶¤*ô&Ô+]ÒZq{²¸G]s|I‹PNú%|ž*Ù¸3­,Áu®*©f¼«íÀó#…RïZ™Œú ;ºÉ6á½&vª£0’'Ì]+ËN°|ºTÞ×´ÏŸ×ë¨ q´á“-Ûã`'ÜÕv*;.šU*ïÖpø®~0vjG4ÖÉ=*ÖÐÁxa”õîJ)á¬ÍõÕýBb÷³<]j©¡5Xw´x{…wÿýg¶.È‚Žm( òÛ¿ß)p£Yœ±C ’O޼¾Ø¸9BAg¿= ½ÿ’¾k×1È¿"ò÷Ã᪀îΓEpòj̨ endstream endobj 1082 0 obj << /Type /Page /Contents 1083 0 R /Resources 1081 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R /Annots [ 1080 0 R ] >> endobj 1080 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.868 102.639 633.823] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare2) >> >> endobj 1084 0 obj << /D [1082 0 R /XYZ 96.768 705.06 null] >> endobj 1079 0 obj << /D [1082 0 R /XYZ 101.643 625.355 null] >> endobj 1081 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1089 0 obj << /Length 1690 /Filter /FlateDecode >> stream xÚµXKsÛ6¾çWèHÙLãäÐÎ4‡^:ÓñMñt( ²`I¤Â‡ç×wÁ‡ ;é´ åî·O,ðÛÇ»Ï\Ì2’Å,ž=lg4d$˲Y,Â`ça3[ë]þ}¾LŸðïK(ÂMÙ®òæÆW7Ǽ©Ô7¦w‹ºÉù nàDÞÏþ˜…³ã$£Èýî©þú;?ʵåUÊÚ~\68ø.«G†8npz”DzzÅñÍ]'R’ «À¶¬zàêSx3õ‘uÓÛ[£ 0¸ûL£¡UxH‘[d‡$#Ã-h˜‘”'cÁçz©­éÂû´)Šë>ðÏüóüó;à³7Á§)¸#ËÝ;Ûòùmx:88ઠ‰;à±”Ú*€GcK»óh¡ŽÊž«'‘ÁXTÔ(¤AŽÔ_[S/®*4ÆÞ(ÞÔÀ7¢)ø Iû%Ž µùÒã4‰‡—EÓ•íø·nÔQÚ{ÔÈ » J cW‘ –,8ïÀÐhq_ÄqÏœxvÁª%¬#R5*o<¡U³1`B‰‹”P6¶â4.bø3Jé&P¬Ûª²îÁç_Y7;"Š»ŒL‚U^«5Gù˜pF@‰ÿùfƒƒBžqP·+à… ’Æ§Ž¸éøÛ°ÓÔ²1yŽ(§VŸßž8©sC ¨ÓŒ]·Ç]dßåÒ%«ÄÞ˜Fž¬"&mÒÀ*Qnmì0`29qÎ òØDC{Ô¿+ãqYé?fHLüª—1uxÅ%]FôWì,¥ip˜sã=vÒ1“TN“db"Ú‡é=Á¦|ꙑK|MøÓCÎmUäÔà¶Â%Ulä7e˜òòÄQŽ’4õé`¢òWÏ;…ãæq¸gÁãd­£°ÔFض5ˆ ¾þ :‡(²Ç¥õÔ—&¾s5‹Ýùù 2ʶ\êÀñ¼–&U.qêÄкâl5íNXï+ –Øî+Ù8!¢˜ð´¯e=Ù¸7cè"aTLhüS*z¤†Pä]v—Ùp±R9\3ìòà°‹Ìa75‡µ«î%¨Ý7}E·Ø:ã7–Bm®4•°pâÖ.⿵ ¶pžó†Ü±‚ØW/jÓš#Ÿw‰»‚KWlµîn`&«LWxMÍ$$Pôÿ÷c±¹Ô2"Iè8¢±:b¤7·«ÌéšiHBÝj3J8¸ o1b<NÒAßÙV/p¯X«…=ô<œ³Ä‘ûBùZ‹!ËOXâ3Nh+ê;×ÏßÝ` µ÷ÜžÂ>{‡vw®œ˜f¿ªò×ú­›¯<5ð›vw÷^ªÂZcdÚ›u—8fmga ú4­þw7í{·îó½€¾uG…¾=ËÄäò¹½K˜\‡»ÿ¢:0½UL¯,×{ GuOWüñ’Z;,·‹°i›ÚļoL8~ÝØ Hb›¿zvÞ•µßÒVD}­GáåÄWkhâ›÷uŠŽç!=¹þ˜éͲ·_B² ÊÉä!äÒOt©ù]5¼õ›,÷—F¸q|5ðCŒ&Ý×~ŒëçRm@ö @F¡a;©a@8O@3®ô‰æõEÿø³ß?ÞÞzp6›øù­un–{¿ÉÎÍÉbiqýƒî¡`ø ÕzŠD¸=æÕÞéu´ŠÙðZIÓ“á£þc¯Þë{Ñh„Р³Ÿ~’‡.AV•Ì÷ž *8áo Í{šâg”›÷„¿?|ø²nK® endstream endobj 1088 0 obj << /Type /Page /Contents 1089 0 R /Resources 1087 0 R /MediaBox [0 0 612 792] /Parent 1051 0 R /Annots [ 1085 0 R ] >> endobj 1085 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 319.004 102.639 330.959] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) >> >> endobj 1090 0 obj << /D [1088 0 R /XYZ 96.768 705.06 null] >> endobj 1071 0 obj << /D [1088 0 R /XYZ 101.643 322.491 null] >> endobj 1087 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1095 0 obj << /Length 1531 /Filter /FlateDecode >> stream xڵ˒â6ð¾_ÁÑìÄëåGíNª’ÔnmrJ%äDæ`°Á`Ol3dÿ>-µlx<äur«Õêw·Ôþ~öîþ3“Œd1‹'³Õ„2N26‰eB—“Y1™ëéãì§I4 ÍEÜÆàî?óøì,'i"€Î4‡úe¥Û¥—ÕþùЪpŸ·µ^ªOúRyBšugµRA‰Ì8”dRþcÁ‡ç"oÕXVÊÔ ¬åIH9%”3g¼³þþý4”L³|§j7XêZ•-.Ô € ÂyY8 U6Uí°ÕÁ‘V« Fµnv5ÊѼ¿ï"qî‘UUãö> ¨?–twx:`}š‚AqgüjÈø8!Y&}qÛ§UU·l®? é¤W'•T«÷j¾}D̃;¾4ØW4£,",KGUË2“ؗZ=MkB’œ ‹šëTxtN¹ï…éìа‹ËfÞñ~ CßâIŽg$È*t [ü¦—ÌÉCÓ»éÛÈúæÚ϶þPH{ËJu¼Ñ*SÍÎç}ˆGñÈóO¯¢r!@=5NE­òÝë¬éí¬»µIŸ»»¿Ë2=£Ë å˜æ’B“IjfŠ'"ØMY¨¯¸x™Ê8Èk/ž {Y”­9žp,WƒÁf“·º*q'¯ÝNObàŸNcz¦ ¡´Ï÷/׺RFRPÖØVrÆš»®c”nqç¹V¨ö“*—ª³ÃµlA÷kBÅž&Åþóµ*Щ{MÉ£„ˆTŒ›"£SÏ5`$H« \ä6‡-ÔâWåK£üW‡RÿqP›<@èyI¥Ñ d pÌ/Õ”%ÁqÈé”É{E¶×š¦$ꯤj5À<%»ÙÖe¡¨Kg±ÂpÔ_‚44°v¸†hP" –ð`ˆUz«îÛCÓâÊM×d©ã…õ›‚”6 ðÄe;ßÊ÷í7™º4„]0ÌâsmóXÔyÙhSnÇFÔRô'%ñ"(eðkubyí o!ÙÉ¡L z¯#Áf½GÔ“ÚŸÊ˜ëÆž’‚déE1}úh£¦×{§AÃnͲ(XÕù»€Y7˜¿y¹VYµGèÕ §’ÆF‚œÓn¿­ÆÙqI’žÙ€‡$‘Ýî7†S†ñeiöŠëç£Ä÷üv°ƒž7êÚLñ˜c¨¿bXX ‰/„è»&Wb|:µ žµ«}»,àš™x£YAé¨ ”fðx¡¢Ô®áƒ„æ°op¯­CZ?®p]V¸_• ÚÐŽ–è‚ØáJÇv0 LBjŸ÷‚×ã°t¸Œ ³Bu½BÆÐ¹íZË´L¦4ø—‚P übœ®êÆ]„„R÷Q1K[@/ó§'„¾0ünòÆIpœ‹Ö$†rܱl tv%ÀªÐùº*sÇ-ï”iœ¾®Û$}._>UL®¤6W"ÙE2î 2=FFû$p¸u…ß…ÑÄtIÑ(#)OFïzÁ;_ðÏÂ<2ßÝ0‘ ½ñ%IœÝ0Q„ü=<×ÿPºÙÐ$”7——‡ýBÕþÐ>2ç+·ÊÛŽv„X[z†²ñK+`n‡ß/W”(õÒ£e:Jr‘~ú”~og^$Hÿ^H9‰Séß0;Àµ?¶·þdØý?’7‡ö?´_ËÊ¡bÜ 'Œ‰Ë°¯ÿg\È´“?]úäÜÈ7]îÁ%Ú,kýÜ¥èp’½ñ÷d÷8/7û‰Ü{3ÂSè¹,³S‹=F97”ï>ÍÞýb$5 endstream endobj 1094 0 obj << /Type /Page /Contents 1095 0 R /Resources 1093 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R /Annots [ 1086 0 R 1091 0 R 1092 0 R ] >> endobj 1086 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [264.517 629.838 270.384 641.793] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}matrices2) >> >> endobj 1091 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [254.056 617.883 259.923 629.838] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update2) >> >> endobj 1092 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 319.004 102.639 330.959] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}matrices1) >> >> endobj 1096 0 obj << /D [1094 0 R /XYZ 96.768 705.06 null] >> endobj 1097 0 obj << /D [1094 0 R /XYZ 123.92 633.325 null] >> endobj 1098 0 obj << /D [1094 0 R /XYZ 123.92 621.37 null] >> endobj 1099 0 obj << /D [1094 0 R /XYZ 101.643 322.491 null] >> endobj 1093 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1108 0 obj << /Length 1523 /Filter /FlateDecode >> stream xÚµWÍ’Û6 ¾ç)|ÉŒœ]Ó"©jÒÍ¡f’œ2Ý“³Y¦c­mÉ©uóö R¶lÙÙ¤ÍE¤@ øûý«é{2’%,Ý/G4‰IšðQ§„ÁÊýb4 öòYVz<‰YÜÜá¸×³õãÛñãý§Q8šPJ²Øq¯¶¹žÕ›…Ò¹–³Jîqviÿô=Ž,˜°dŒõe~EÆž©'jåFI]²Ru3+u¹•77—µÐ—kINµÌ™¯EÓ˜$Ù DObp1X p[®r‰²¿„qXY§»C¼Cz t@%•a’‚±VÌrHY%}«¦oPJµÂyQow­.«¯øû‡7Ó¡P/ëæ`ðÓ]øÿž~«lÈýïÍÍ%»yHÒ8»j7 3"¸è‡¡ç(5{ràzwÉA)'ŒÅWõÄŒð÷ϧåv‡’ÍÙ¾ç€uç€ußëï8@¼Àé%Ë|BÙ´{z¼š•vÙ§>÷œyjZ‡þÊo†ØÅqb4íÓÔ{…sÕnq¢këJ~׸ÿRBÎÍ$áÕ2r”Ä/Pt _%—¥þqø:ÀF'îuº“£Íœˆ´Û«Úæôåó\N†`RBÃÄo*Q!e1ÉÂ!·ží‚Ñe3>ÿ ÃÐG»3’R“ω#çžûÄ7â"ØÖJ›YjËݦ, Y¸´Ë·T/‘¢q—áO˜NH©¬½ÔR’¥Á„!%qì¬Ù Yµs|<É©,«/![ W˜34×Mùó í•…ˆdQÔ÷ÂùR"DW`oÇžfÁ~Uc0te~Ã`•+œTÞ¨1( f‡¬Œ[tÙ¸ç};7Ñæ< ¶Æ#íF—ƒ³67´oQ¢IDÒ”zÿBi˜uëÄ!ò]$Ì›rg,sÆþQWª\2€Ž^i q®àäÆ"3÷–3ð÷'$íWµ’ƒ6&$:ÜAëscÒÕÐg…9` ã4F"á = >°Sëbï¯þ¦ç1|òMë~wpU™ aýè½gPE?ê*Ÿ›zžÏË ”QËidSS iÛÏzY!ݪ5e˜.ÊgðìÂo(|;aÞ­°¯+3šm4pì•…ÅÖlšÛ*`!Þ)Œ¼ Ž ¸ ¦øÇ¥3ÄVh—¬X¨ç¡:½FÎC•@;Á»‚©žEÛ4Ú›oHÊ5ŽM©ÖŽb“î¹ÅÛž;Ó‡ $ »„nQø£B…Kçr£*}tE¬ ‚uÙS/‡L£„GÝòi(ÅA…÷ÄB7y¡¥óɲ©·¸ò0+o="™€,¨ËSîòôýXðÀ®24/ÌMÁMØQ¤£,‰)ëÝz§5=äÉÖ¨c•/P|^9¥Ï&ä.”H²a½²'Šy^øm $!È/HqÇ7¦_@cPÈ¢OC¡b‡6D;›/‰ã©}¬¼O‰°vÍ,%éÉ“a_nº{¡€·›’Z_‹ÌÊ>lBOrÂ_;âAPY]„ÉD•oÝ]’oëCáÑà ýäÀÓÕS4êKÞØmyÓÔB†Ü^Q@¬j”JJW»bA¤¹‘åŠZ—;_Ž,€êrÑnìõk\I…ë®À °­‘®ÜÑÀmlw½Ê÷GV³½©&C ¢lò9È2Ð÷3£$‚îæW7µ'mz¯G½s×NÈI¢á‡óÛ=;¿ÅH3ƒö…Âým>"¯ß¼î½°'ýÝiç®Ï(\sô3/ñmç^ð‡®Z\}ŠÓ0ÚU[ïx¥¹7Ù²w>g§m×YšýŠwóOX‹Ýêƒûe†ÃÃË4bPÝ„®kä‘a|õçý«uÆû endstream endobj 1107 0 obj << /Type /Page /Contents 1108 0 R /Resources 1106 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R /Annots [ 1102 0 R 1103 0 R 1104 0 R 1105 0 R ] >> endobj 1102 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [248.826 378.78 254.693 390.735] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}U2) >> >> endobj 1103 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [227.904 354.869 233.771 366.824] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P2) >> >> endobj 1104 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [465.506 311.413 477.461 323.368] /A << /S /GoTo /D (equation.8.14) >> >> endobj 1105 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 151.631 102.639 163.587] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}U1) >> >> endobj 1109 0 obj << /D [1107 0 R /XYZ 96.768 705.06 null] >> endobj 1110 0 obj << /D [1107 0 R /XYZ 144.841 382.266 null] >> endobj 1111 0 obj << /D [1107 0 R /XYZ 123.92 358.356 null] >> endobj 1112 0 obj << /D [1107 0 R /XYZ 97.768 299.245 null] >> endobj 1113 0 obj << /D [1107 0 R /XYZ 97.768 277.572 null] >> endobj 1114 0 obj << /D [1107 0 R /XYZ 97.768 233.372 null] >> endobj 1115 0 obj << /D [1107 0 R /XYZ 101.643 155.118 null] >> endobj 1106 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R /F11 183 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1120 0 obj << /Length 1033 /Filter /FlateDecode >> stream xÚ¥WÍr£8¾ç)|Ä“EF¨oÕnÕ¦2{˜Úƒç”Í9`l @$•}úm! ÆÄ3sBjµû¿¿nÿ¹¹[=Po¢¶ØìØuQàâó|Dàe/ž,ÁO%yÊž×Îçåóæï…³°1F¡§Ÿ÷Eµ´=âYÿ:ž“e’«½fÙ—¼‘àÝýþXp+ÃÅ ‡=Šý~­¾Í)O‡ç§Ì<|RŸDÒ?hÍZ=`·ç‰Íò™;Ôð¢.Ïzsvfä‹veh¡6PnÜ™´{ˆ…X«f×­³=ÈÃÂ&w˵Òþ/c°GŸSß2ª„:M©›­¨¢¦î«â¤Nü•WïE®é÷*­³6=”!†É02&2"Ikuúª>9ç±:=£U+Æž’3ˆð¡‹ðaáC?‰ŽðAGØÖ!Æä@AÉ«ºÈG.³ƒ©‡Â ¹JÞT^0dÐaÁ0ƒéþ¬82š`òz¯è ]„]6«Æ§ˆoè_– êl׆ ÝÐÕ³@’QÎÆ%n …–² WoµüH!7ö缨ª4‡w[süPwP´´1˜ný‘ëˆãT¤…¾u1ãÑ.éú%ƒÚž/ºô\t—›ã"æÌ×A@ Æu0(ƒº¨¹VýÂä"=ñ3X™*’ÔkBI0˜50 õýëZeµ5u§öwg®'üÛ{‚MöD·e§’lj¢*Þ4^5§-×éJuοÏ%·8Æm· ;.˦ƒó·>÷Ès{´Ó.FרÓÊóƒìµî2/̘2š©³ÂÆ:…ºQÛ§¦ØÈ<~¬ukn+eÈ!=>¹„àÂâúV,I`Aª¡d­¦„¯o)0‘´ßÚ5§%¼4ÇH¤¯òšžDÿEU¬˜·’þ®è"òE—cNÒD¡î©ˆ¡ŽkEy|©Wy‡{V“ ×ïLyé—B¿CPs)ೈÇKZÉq @žK‰õu¯ ¬¡Ù±+”ržK"Ön%<×óPšŽ“ÿ¾R ªåœ>BãwEÝ‚ˆ%h¯°|ä5`ãIn-¡P?’!”wù†& @– ܈֙èªb=Šß3îÕMõºOÅ.µwÅ©l·ÿ™„Z 1é›NÔª”Úm­’ðKFˤ•‡4ŽÇw=dšÞàÙA-éù’ë©4“g4ìgv£©ÍÅA¾7ÔØ Q@ç šC ÍtpnÚ†@»+rX/ënžÊœ«íð†ÙyÛÎ-CÛ_H ²%Ý¢Òcow{p¿¾´=‰ò>^—Í^Ûy27-ÔâotŒÿ5±_ÚËðœª3¼OþWz6ÛÁ&<Š(@‹ [oÀ´`X%çÝ_›»ÿ9LJ< endstream endobj 1119 0 obj << /Type /Page /Contents 1120 0 R /Resources 1118 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R /Annots [ 1117 0 R ] >> endobj 1117 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 229.34 102.639 241.295] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P1) >> >> endobj 1121 0 obj << /D [1119 0 R /XYZ 96.768 705.06 null] >> endobj 1116 0 obj << /D [1119 0 R /XYZ 101.643 232.827 null] >> endobj 1118 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F11 183 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1129 0 obj << /Length 1572 /Filter /FlateDecode >> stream xÚ¥]oÛ6ð½¿ÂÀ^d§f$QÔǺìaE tØš¢õ†mQ(m+ÖD*Yöëw§£dKQ²5}°u<ïûŽÇ_6/Îß8Þ"b‘ïú‹ÍváØ!sÝhá‹€¹\,6é⓵[~Ùüzþ–Gµ€ýÅÚqX$ Ù¶j–ká ë³-ì› û­n~*•޵ì—gg°ïТþtó…  úhYÔ. _‘È‘nHg±vA¦’̽QÍ?¡ã, ÄÂ&Õ6·ÛL'Ù:©ŠºÕrÝÖ)h3ÃßqÁfö'³«‘µ×\ ¸ÀXøc_œ¯È¥«F˜•º2Fî ªˆu“%RõèXt—å9A׆²Ž•’©ÁÅÉ Õy§‚ý=q˜ñ·YpôÃv.úޱcÁ« :;´îƒ9ÕrU6™:L¨Kõ€ÜŸ1ê0uu8M®UÒûøŸ‰ˆPŸ_S ¹ÏleÛ£JBRÑògû?JI<éFá2n{O„odi-U•SQQ×f‘ë>4„(ÏŒåwºÃMb¶*rÃê’Ùë9v«Û‡±E³…Ú1pÆÜ™jÌœ1ï~Ëfžÿüún¤n›rFP0ÏßZÖaÀ"Îç«Úˆêà˜>y¦ôSEùþÃÕæòõcÚ íEqx2EÄŸ—¯?þõþ¥á]Zahñ7›ˆ/7_áÌæêÃ×Ëߎ, Û|»úrŸÉÄé™È[Yêçrq{.‰,UÕ̰ñÿn¾Ø^ž«Šg¾ÔžËE˜ï퓚üñî$ÚeÝTZ&&9Î*%Õ q ƒ\µçnAÁÂȸîlæ.›°­÷MV.¹ké¬Ü-×<ô¬¸LàVWºCw\æ¨JÈxDì6{9S$PÔ^Mê®P™ 8 å”äpu|έ4ûl;š.=kÛTmÄ3âBE¶7÷P÷™ï¸=ÙKâwÝjQÄåÒ ­{BW[úv÷/n+(5BRïávÚÔ([ʸÉï‰.®kÁ•l®Û9tñ0BÚžø¶ÆÕ;pÄÒ Yh»ÿձĤc ‡qwÒZ ´ûÍKÍœ1\0Gºeu}y:§Ï$S³Í{`ÛZµE7÷3<#›ùÞ¸+/Ö‘Ç„}‡´k•4Y­ç¦ÛaŽíOœzšJëJËLýc-@DÖ‘÷š{ö0•"SDÓÓ‡ ¤ª—@vO+ÊÁÈ {uáè$@[íÒ±š[¨Ïª®!:0®ah0½=§OïæD½?Jb•g»=f¼Î¼¾ædWÚÜ´P[ÞdìP™º¦J\×Ây¹«ˆª4‡[®=O>AÒA7T]÷çÚ<¥ý}ŒçnñÏœPº‰uL„Ô—Iã.Bq^TÊð‰ó;ô[Çâ^#†Âm®³:—½ü¼-JÅæzì[´Mt8Žå\ø7@Ueç²2[§!|Öm×ÜŽÍâ¼?ýŽZI’ÁdšÒ„ ŽQ÷Âøcê;Ýp9Òì(§¬NéçÌŠ›][œDxÒ,<® þDæC»ñ‡vc¢á{C7Ø»d ‡¯¬·˜ Û¤3y5†Z£–wøWbð²šðÅ"hºlìŽ4‘†¨WƒSÔþ8¼ZÄ$©;—f>PviëÙÌjZçsW\=)>%\òÉ~æÊy’çú¶LtV•IJ "»¥îE¶J6Aø«C€ÌÐMw³•AeV¹“"Å7ã‰QTӎχ窴æ§Ä8E:E2E;Y_æŽûé½,‰²/ZÄî¤Ù†Nˆ‹sO+êoQa$‰asôÓEœ•Šv°'̽ÊL"ñÐ$$Ú¤òaË$âÀJpt\G×wÚÀJâ®Åšów}“ie²ž€Ù0jõéPÍŒ85 ÇÁ‡Èƒ²òJIföéíÍþoZ–1LF3º#Ú\¤ÀPó8ƒÁ.I0CKfæSÀÉeœš©§2ÃLihKjÎØ±òÌÜ6]³Öúþ.SfP:zÍV”f™ôž”ù½y¡ÂP•ãöyÜ9ÜÇÍ—›ÿ.‚ÉŠ endstream endobj 1128 0 obj << /Type /Page /Contents 1129 0 R /Resources 1127 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R /Annots [ 1122 0 R 1123 0 R 1124 0 R 1125 0 R ] >> endobj 1122 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 619.875 102.639 631.831] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update1) >> >> endobj 1123 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 454.496 102.639 466.451] /A << /S /GoTo /D (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) >> >> endobj 1124 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 246.823 202.389 258.778] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary2) >> >> endobj 1125 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [206.983 234.868 212.85 246.823] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript2) >> >> endobj 1130 0 obj << /D [1128 0 R /XYZ 96.768 705.06 null] >> endobj 1101 0 obj << /D [1128 0 R /XYZ 101.643 623.362 null] >> endobj 1072 0 obj << /D [1128 0 R /XYZ 101.643 457.983 null] >> endobj 70 0 obj << /D [1128 0 R /XYZ 97.768 323.106 null] >> endobj 1131 0 obj << /D [1128 0 R /XYZ 102.999 250.31 null] >> endobj 1132 0 obj << /D [1128 0 R /XYZ 102.999 238.355 null] >> endobj 1127 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F36 121 0 R /F30 111 0 R /F8 112 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1137 0 obj << /Length 1154 /Filter /FlateDecode >> stream xÚµWßsâ6~Ï_ÁÑ›KnËÆ2¥3íL2×Næî¸'ʃbˆÛ•ä8é__ý4–‘ÉuÚ¾`k­ÝýV»ßjùeyuû'ƒ9˜'a2Xnó)˜&³AOAŃåf°ºÙ\/»}ˆ’Ö¾̦ñ ÐXEŸ·„Ù˜UO,¥¤äZÅ1 `X%b¬Æ«»a!7 Æ0ˆ@NÄ óØ@®@ãtøq‡ñÍcýÜVyÊI‘ÿÄÁËHËæmC‹r±üòí^|†"h%Md[Èq'`2é ʈ§û~($ßX4Ñæ=à4I/€1"IŽÿ´íåž0×¹^e˜ùý¨ J_õ+z**®_Ó=¢(嘨9Þ!Nž±YÔX+v$E™^49g*aÁ.(¿ùÕx@Y^ ŠyEs³@Æ6‹­~–#‚^‘œã¦v±!)vÝ&Ú-ÇÇÒM¼Ëp¾ã{™ 7]È9:b¦¶ #j“cH+ûåþ•LôÚ“Øq€y]'½Âh9441Ç ñ c$ßÉW DvBG,¼ÏßÍ™ãŒÙcsC`å-+¾ìõº~ý\¿BÛwnm_„¼Ê2ÅÞ÷ŒSÄQãÐÃq¨á|v‘-ó9ˆ¦Ó‹w)yÑ>KŠŸÿ ‰—ŸŸ “™¡Õ“¥Ÿá€†Ó¡€…D§ædޗg¨Ï›–"f™Œ²¬¨íw|4o\eÓg9)’Ÿ¹•tjôz£ñPã,û>œêèÃý|×úL:¥æñ–­`Ÿ»M7fåÍÄ.ë qEqøX”¿ZÌȶq‹ªÑ÷1›ñBöŽ`XixÓ9¸¡AsX_7–½M£—^ˆgœò‚¶ù"´'àª[éê°¶Jªy:QN‹Ú¥ËbÑIä©Û¶²ü“éÅ'‰¼-€‡Ÿ¿ÞûñJ7]´R¶FêN—?k¿nZ÷诮¶–®Â@{6èµÓÿ]§Km¦Ô-ÖF|~ÆJü6ð~‡UYâ3wJØu¦„\%o¹]âÜ•v])ṫNû Þ¾ÿt͇}5_#š›ËhøUÏ€¦S>Dñë÷½mHEž½öö29B’g;Ǥb…‡=mÿ»qû¸ê3ãðâ%‹~Wí´É?0îŒ> ÿ©zYVä;P¾_¼¼çäˆO:æv¬fîU Œy*æßÅÕ ëBÏ&vµ3ÙÙP ñkáØÿþ^"Ʊ*F A´u©2‰Øê¨4[‰§å›8Ó"ËPÉðâZ¯¯åÇQÿ]p¼ä-sí [íFÞŒ"õȹÓÈP¥±ùÔLÝvþß`zçðIÌ ´Ã9E¤Ut®Ý|P¬ôõ2+xÏ#?: »ç¼ÕAJs¢&ÊìµMçoî‰*xûße±8«åÛ‡Y{˜BÍæƒñ‚Ù$Ò8`4•;¯î—WŒ9¤ endstream endobj 1136 0 obj << /Type /Page /Contents 1137 0 R /Resources 1135 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R /Annots [ 1126 0 R ] >> endobj 1126 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 653.748 102.639 665.704] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript1) >> >> endobj 1138 0 obj << /D [1136 0 R /XYZ 96.768 705.06 null] >> endobj 1134 0 obj << /D [1136 0 R /XYZ 101.643 657.235 null] >> endobj 1135 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1141 0 obj << /Length 777 /Filter /FlateDecode >> stream xÚ½WÝo›0ï_ÁÚ¨k$p±‰ùÖI›ÔJ›ª½,{Êú@·!%€€´tê?° i} öùî~öÝïÎÎ×éÉå…5x6²µéƒm ÛÒlìDW¦ mv<Œ ŒðÅo›!‰ó%ôò…OWbp~ο‹4Nšåb”婟ûBÙàß¿noÇwÓïš© 4fdl8®#B X[VâY-Š»Ê²€¹‹b Ëv°d©Ú2b1ʃ5ÅDДA˜pöDH¢Ù5Hƒìiko\ÜaˆCòL¢\ G-¶œ“(‹SÙˆK«Ë8‘2ÚòñÈU”¤ÈBª¢œô k?Oƒ‚MŠQ’’g–Ý2ÃÛˆºCžK‘KG]ˆž,ÇVÏ8à7< ô;>ÊHHæ9YðY…¯|$è•Æ/º ]}aÇaè'ŒIl–/I-Þ¬£¬+¡Ý,ga•ˆŽTž‹àD›0dÃU£ùöÆ¿'iñ3¬ÃW—‹Ê†›/·?¯»8ÐJ‹‚ܓ׶‹‡Ò1˜LP¥qÖD8ÛÜgó4Hò:ÒYÏ‹°ÕhÉdÚmÊ2²´™Îd%auýŠýô0~¾Y/ý?m\:ÓõÒIÄ~OrÜ>Hc½dHšÊÔ&¸«½¡ÙÞŒwÈP\Ù»6³I’ªä UL÷ _?N¿lã”§œvâÛb`G§ßv‡l1ä+m0°ák‹¥÷q¾ÜÁ×RæG µ)Hš~N²†Ù¢>ÔØÒÛDeBD{䜗ºÞpCñ4?z­jÚ/Ô• Ú^‰¢ S–—w­yºá¨úd•S±›¿›cõæ„««§ó3ðL»pœÊ›ë®Y~èí2aÒÙª§Ð‡ûÄj¯"ßÑ.VÔÍî†aÚ0Ô60ØVGìîÆ ‰Ú¿t E8ºGËè&äÞ–¡n‚#˜>uÛ»A:C¾^»ÿA´w ×;Pé@â 5‡{_CÅÞWš+ßA΄҆b²LØâzÐr™æÉõôä/ßΊx endstream endobj 1140 0 obj << /Type /Page /Contents 1141 0 R /Resources 1139 0 R /MediaBox [0 0 612 792] /Parent 1100 0 R >> endobj 1142 0 obj << /D [1140 0 R /XYZ 96.768 705.06 null] >> endobj 1139 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1148 0 obj << /Length 1184 /Filter /FlateDecode >> stream xÚ•WKÛ6¾ï¯p½‹DVŒ©—¥¢>4EöPè¡rHz%Êf+‘®Hínþ}I)K²ÖN/Ãùæõ é»»Oé"CY$‹]µÀ8@,’xƒ‚0^ìÊÅWow$+?LbOtM“·ßWiä!ѵÏßÖ8’°•³òŠL#`³êX!)gv*ŽykTG^+uÞ,XÞðŽ­‚Ô“ Á+X/øÊW«%A+?Sï#—ÇÕ_»ß>ŽPEê‹Q[Û J=ÊŠº+I ³#õL­Ã oKÒÂPr+”køgýc… Þ4ܼk «¼âîèÑJ6\H+×íemåö5_hÿXàÊ(Õ ¼—#Uûï謽YÚÍõØÁR†Þ •ú@y‚ÔDE™`ƒBIÛ’æ’€¤ ý“±ˆ[14ȇ' ªBA §y‘0ˆ„(ÝÄJÐè¼WTeÝwÕp©3[£$êOP«2~KåRœ…'“[ dQUU±÷‹_WqßÖñšïÿVy„ aÇa‚·¤Ü>ýúùÏO G Š¢lUä5Ùb{Œ¼JÂF‡&¹Q!ÏÙö@ä'gÃ{k%2{ïÕMà(BHïÍ.HŽ—j&p>Ž1JÒ±-‘hÌ™M+ÐöþD™â •BO*ê¢çÜ08CFiNB&ùIŸ]Φ©È™µ†Õßa´'ðí„&«É¡f—ŒŽ‚ýbÙ£û>U­ú=Èš¦§–2{«ÅdÆTÀ·$§–š+¥u×a: !i]ð¦BU2u&‰ Ç›Ø4sœ ­°ÞXˆ}ç0L×"LLŸº‘"*ëêZc}JŒÄÀW—ù%ãŒ,ÚA´$iNcQm£Õ9È>ÐÀü>PúLLŒ•ù¾&c½é¡ÉeK_UŒ”rï¾Ï²éÔâœû•.‡¼“',ý7™‡;¡Ï]@«d¶†AÀ3åßP!T+ÕcÓ1†õ>!&^g( 7W™™¦ '¹F7ÌF1®¸„´¥ÇOgù¥‘€yÓ 9¦Õ´|31çÃ}Ây+GÈ3ŽúÁeë:̸; 씌ª túјg¼!#sø<+9W,æÝ1+:ãï½­5(NÕ[į¯ Ä X£_}›úi†‚4›ío¿µ¤o"ý3cpg)?»Bv·/ösº”©¯†?Ûí8‡ºã©ûîEú™S°²ÝZlò¯ÞÉkÎÛ¡‚+ Ó ZŸŸê³&êULæ—äz<ÜøÌ>‹ŠcζôGFã…*¸~¨vö¿3-x÷iw÷sqçG endstream endobj 1147 0 obj << /Type /Page /Contents 1148 0 R /Resources 1146 0 R /MediaBox [0 0 612 792] /Parent 1151 0 R /Annots [ 1143 0 R 1144 0 R 1145 0 R ] >> endobj 1143 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 607.92 102.639 619.875] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary1) >> >> endobj 1144 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 607.92 197.159 619.875] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary3) >> >> endobj 1145 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 165.579 218.08 177.534] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows2) >> >> endobj 1149 0 obj << /D [1147 0 R /XYZ 96.768 705.06 null] >> endobj 1133 0 obj << /D [1147 0 R /XYZ 101.643 611.407 null] >> endobj 1150 0 obj << /D [1147 0 R /XYZ 123.92 169.066 null] >> endobj 1146 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1156 0 obj << /Length 930 /Filter /FlateDecode >> stream xÚ½WÉnÛ0½ç+Ô$‡°‘Z,5ŠHEoõÍõ–)›©,"íëKq±ÓŠÜ9$©á<ÎãÌúÛôâöÞ$ŠœiæÀ q($¾L—Îì#'›Ý5§rã†(üøÙUÏÃô-KqNnæÓïŽç¸‚$l­Ä‹¼»TNYÐLüòBïe Øçyõ.1^bŽÅVr9ò@‚TÙ¶Q3œòmY¹c•ÍHÍæxAr6Q @7„÷·÷0hãPã;.òA¡B[)Ë „{3¼ eŒ+Õ '€¡ïq˜¿Ò_f„!d,®–$Ãûœ+´”l[Nî¿þøy§g0Sô»…ÅǶ™Û.HÊOŸðéܘÑbù çÿ1E²mY“ø y-t\ÕÜeJÊ~_êóCòH Þ+Bš'Õ¸‚‹G¡ÆŒæ%Û—{Æ—€”õút¿Yã×Ã0ß>‘úã~·£žÃE!žê9\%ÄIÒ&â•,IŸÍæós*KxéÙ\’Ü¿·q ‚0< &°6˜—ô¹‹ÖÙs+a§2g4Z–ÛÊ]k\uq½åjn+i) °ír¥J£¥-ÑàjO Ó%y.” ~*äR•ÎP(ƒÛþÏÐ<FBì, Ð,Iy… GÓÓ{x« Ê Ù¸åZåWŸØ¡×ÜaíöÔ&­}¦o«òñ&õñ_R/½àþ°‹ÍX:—7žÎ¥]óãN¢!È7ZxR¨ÛÛ°œP¶R¢v:08óÓâ½Úl“¤þ¢Lû ò¬Ë¦­›n©¸TQk­3Ídܪ ü8©®Ì Žcµx•åÅÝôâ¦cø› endstream endobj 1155 0 obj << /Type /Page /Contents 1156 0 R /Resources 1154 0 R /MediaBox [0 0 612 792] /Parent 1151 0 R >> endobj 1157 0 obj << /D [1155 0 R /XYZ 96.768 705.06 null] >> endobj 1154 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1162 0 obj << /Length 1378 /Filter /FlateDecode >> stream xÚ­XßsÛ6 ~ï_‘­»FÞbVÔouuï¶»u»Ýó–ä•éX‹Ly$e7ûë”"9JÒÞúˆ@à#ðTòëå«·ãô¬deeg—›3ž¥,Ïâ³,ÍYšËõÙUPoË4Jƒë0 ¿« S]ÓàzSÛšö(5¼püYÜ\þùö#OÆãåiq’« ™Lb.yX²"ÎaÁY™ú Vîöä‚¿_Ò³ÂÈü‚^&®jµþÌo0„{ì¬Ûï_tælNœœh…¬Œø4Àí̹N0<—ŪÛmÅ¿CÃLŠÐáôÃî—PGRöMP«…Ï?eeQ>_|ž²0;)¾©ÚNÙé±;ÕÔÆbìFì÷Í=®”<Úz'OF#Õ­ÝŽÍpO½oDe[¬|¨åÞ±ârvÞ5!B·C¬ƒlÌêÁ £Í¦ŸÐ’¢2%ïùZö ûnE1+9\¥! ¡ÞI¡˜Tk̆ü. ÙÒ0xó†ì]EŬíÁ qÆiœl™fsªB/ÿì‘—J4ÍI)Ò(©ßstT‚AëVÍôÑ {êAõL¯“ý?ºÆÓoͤ֓Ìg¹êŒ§°Çšé¥Ó_Û×  3ж~gëÙ­¯n¨§¾‚·qNT0¦ÇÿÙιév;¡ï:…”ž?Õy/NÓé\*Fv%Ëy Ö`Ås²ú]‹O‹eÇnQÍ»Å2I—}ÃaµÄ%ˆ ‰T;S<ÎC:Œ)7Q#ˆJ–äqo$ôm·“jõ!|XúhR˜ûÙ6OSç,@ùCo=²ld³"Oû˜]ÈôrC^·G3ã¹HY‡ õÌYÑcÒ¬¨äEÎÊ8~šÁ»Ú˜ZÝ:Ô~z=5ñÄeaöìÄ/b–é4ÜkŠöwgüÔß×Õ]OÝSîÂø×í^×ÂzËÆ<ÏÛÎî;ûÒ(©¤2­–ën»‹tJs#ÿA[Ñ´êv…“£•sÞaâË'=·uµÆƒ›DÙáS—Â×uˆ£”q¹E&F b¯ZµÆuT¡¡´×Âüëp öˆ­Ôê;¸æ¢,  ð•·Þ dû=™Õ í(iñסñV×Ò÷t­÷Mƒ¾ç®qÌw–¾-!¢¡w׸pßs {'$Ô~[#ê5©%òE¶¥' @ƒ½ç6Ñï†*:~ªÿŒw=·Ó¤:%x8ƒ§´´SûÄŒò°w;Ì•¹c*)×þGxH¦ÎôÔúë0Œ/”ùN¨i¬Ðž}íæD7 `/ññ+ú‡®³Q»Â<÷øšúÎ=jšñ¾]GÐ×%u¬c<žÙ7ÞXG a‡ï5ÒQ:}¦—ÉÞ¯bú³+‡ïžþìJàó¦ÿ¯OÜ |õÛå«ÿ…)V endstream endobj 1161 0 obj << /Type /Page /Contents 1162 0 R /Resources 1160 0 R /MediaBox [0 0 612 792] /Parent 1151 0 R /Annots [ 1158 0 R 1159 0 R ] >> endobj 1158 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 336.936 102.639 348.892] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows1) >> >> endobj 1159 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 336.936 191.928 348.892] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows3) >> >> endobj 1163 0 obj << /D [1161 0 R /XYZ 96.768 705.06 null] >> endobj 1153 0 obj << /D [1161 0 R /XYZ 101.643 340.423 null] >> endobj 1160 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1169 0 obj << /Length 1467 /Filter /FlateDecode >> stream xÚ­XIÛ6¾çW¸I±›%K¶ÛNÈ)z(¹9>p$ÊfG&]’Oþ}¸K”e'@‹ ñãò¶ï-|ÊoÞ?¬'°)ÒbòXO6+°*Ö“"_4Ë'Õd;ý0K×S9[dE6•äˆ-•ëŸtzž¥«)¶ô‰”zõlW/³<Ÿ¢¦Å®+Âq)›¯–}I`ædœÈ I4Wä*™ BK|©áˆ¤Qq³Ež¥WLÛØŸ³>t[Ÿf Åȸ»M™<º· RÏv¾PwP$“„`“;ð«Ä´²÷úˆe9È`¡n›[D;»É§3áôÃ_Ÿ?j‡6K¥ØXg¡à) p—>KÄÝ3‘‡:r¨Úou’| #Ö ,òk|‡ŠÐötÀ꣠•ÐÓ«…æð¸Héì%â–-=oŽ*å{Yñ«Ý´å³÷²@ðâ#€qŽÅ‰ÑÊ4ã~,U¹iÃÀù@Ëöuwi—:u¹¸ÇrÀP‘ºÆ‡ $Ë S.Rh è.*>©kÎt>„µ? FÕ„É6uäÀ˜]¥*W{ï·E\¿Zý«S¦_Û.]´÷ú4J)ÞÞ¶çÑÀžôF¨wc]Q ÿÍY‰…«1h7>«Q‡ßüìùJÝ«zIã%«lj_'=> endobj 1165 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 607.92 102.639 619.875] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows2) >> >> endobj 1166 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 607.92 191.928 619.875] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows4) >> >> endobj 1170 0 obj << /D [1168 0 R /XYZ 96.768 705.06 null] >> endobj 1164 0 obj << /D [1168 0 R /XYZ 101.643 611.407 null] >> endobj 1167 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1174 0 obj << /Length 1280 /Filter /FlateDecode >> stream xÚ­W[s£6~ß_á&;Mvb„cgêv¶ÍÌv2û²îìCêÙ#$ _úë+дÛé‹–tÎwî¿®ßÝ?MýÑÒYνùhŽÐlæ,fh4÷Oí¬ƒÑËíõ‡‰ïù·O<Ó ædT¼êõ‰À3©WX?DL÷‘Œ/ú- aH2bÎPó£^{co<O›õï#w4AÈYúµfäúîíÕg¦o‰<®ô›¤ z™rj„Kϰ1r†­Ha"™…A/æeã!(Û\öŠùÉH猴d‹<;Ò#Žýú§öþe<k°¬ „Õ ¨‹š$$ XJ’xôk¹s!*ž›g¤:>¡Ö{È…4ÑRÖ“–À€’1lW2ÞÌ„'Uª” •³øm­LµaÌ›0T,2ü&%!Æh“[cU åãM‚ž´ìø“Õnüiù@Ø^FE¤E±YX0¯Yžl«\ -À¢K¶$IŠš²qšfü\ÈÞ? -õOø¶Rïîc¹ogžËUZ–ΫÅÔ™/æM¥ ‘V7;΄Tå}—Õ «,ÉÊ+´¼¿”b&]rT¡ÑË ÝlšÐÄ¢Àè–æ­*±3ºƒŒøKšàÒX4.:7¥?»s¡¸ö-º´Ò«ÎÌ«Bf)iseÚ͈ôi6Wê|¨Œ˜š^PëÇ[~$¿Ào1’ê8Gö/’q1nefH3#ŠœÓŒA9±§GÓ*¡<Žd'Mo’‘É} ‡DÄ3Iêm €Á,h¡Ò‚p"Ø èR­ZFVÌ h9ijBE‚å.2Uª‹Jt&sÈQuŽŽdÛ…9«RíΘRùÞÜQ±nd”7u–A ‡uÓþ ‡åq\¬õýa‚—=ÊÜ¿B³ÆT;3_ -……úDcî*ïÌœ¹ûÐ6G‹þN{àRÛ [ò¡)­Èó²?lLбx3¬†j§¾–¸KQ!ÇaMme10^`Ö¤ö¯“Ãæ®Ã°Ã¦ÇÏ-Ýû.GÏ"8ÓæA“µ-qÞÜA¾76_±¤Îæ##SIæÖÅ>ãq@,r l„ÿ·U›¦æOÝEàd^ÕŽjÊÓÁ’7þràý3kCÛaa&‚áq–ŠWÆOЋÌ"ÁìÒìÿÐûRÂÓ˜ôŒ×®Æ/hì¤n1€þÇ^H{>„„Ãý0Ï”°Ê)…Ènヘ R©MW›+_¸4l‹g‰Ú¸ éU„DшÕÓÇ篟Ìlæí@Øl€Ö 2ÍTÁUì„]Nøât ¼®"sÿc}x®³ô0Û°k¾sþJÙD èVÙ¤]šÍéšéjô­å&Þ¦zô,wtEÖ<Á\­P­¾ƒf$Õ”¥ÍC5ç³GUÛβöÉ­­›Dÿ'·DßÅ-aŒ ‘K åg`qm^¬KGEÈwÜù¢©ÁV5gì@4uù/êLO›dÀ×lY5ÆŒœÅTÈ2úäTÅ®›úÔ]Ü.ÅX×ËíþËå¼mºÐ¾\nÜfWì¼ÍàÊö:Á—æWØ—?žŸ»Eë9Ð%µ¦]†´ôнZØtìa¦œ_ô¸¥³X,´&4›'ß}Z¿û9Gƒ’ endstream endobj 1173 0 obj << /Type /Page /Contents 1174 0 R /Resources 1172 0 R /MediaBox [0 0 612 792] /Parent 1151 0 R >> endobj 1175 0 obj << /D [1173 0 R /XYZ 96.768 705.06 null] >> endobj 1172 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1180 0 obj << /Length 1086 /Filter /FlateDecode >> stream xÚ•VMoã6½çW¨N€Ø@Ä­KEshÍ¡Z`×E݉ŽÙ•(C¤’l}ù)‰ít/ eß̼73ä/û«q È¶Y°?pƒbdI ¶y°¯‚¿Ö„Vop¦ÛtýS¨ÿ÷´&Œ‰ÒHmŠÜü½ÿõÃL&Xá6X0!Ešj´gm˜Oì °ƒ±°VÐøü„O›p›¯1â›0Îáš±^”­Ú¨Ìç¡íô‚õÝË—&¼aú‡öIÄš®ÿÁ%g@,ó|½· -­¿éU‡Q­WÇcÜaZ+Âf¾Of¸{ÙXÄIKå¯Ñº=è”ã]ò$ "8ù²L<Œ‹ì — D+,@¼ÛZ¸²oŽèß%`,à¢ØZÊ8F•ˆ2ŽýQ •w‚rc/)\b ðé …ÍNüäHx4þ²i| ߥSw"ô Yß4¨ûæ." JqHÈG´€L¬Á½4Â"0‹]Z0xu+úÐÓRJ*kº}’Es§78i03ëSÖv¸ºøùñóÇ;å &H’™†¬D5¾‡æ~ã˜:‡"×¼k0¢÷Ϙÿ~²1Üš8Ú»•æÍŒœé¾m`:TŒOý¦d¹‹ðé2£Ùð…MÚBÆû¡GÜÎä‡À0©ßt«l¨AEwÌâíIž^‘ªDÔD¤ºW®ž°™G WFG9äbvXgÁVC¡ A<¸Lˆ0@ûº–K%Á¾²P?¹<­hKñʧ‘ÃSC#ôY®UiMi™)#1CâÝE ó\DŸ]ÒEæÑiÊ¥KM” }´_) ýÝôŒ»<¤‘ÅYEL[ZDÃpÇów^Nº‘#×µö€ªÊºjLœ7'oM2è++c”ïµí¾úDÓs£f¢“ Í–é—ÚB°ß Þ‘·q®Üè›jÒ ¾ì+ÒP¤›Ÿsñño<>ÞMký†qÄGÊý$K¢fzvƒPÏ4õ÷nÒgK=ÕØ…–În4› üõõЮ„–uÏÔª?Í‘£Áh{~êù˜òÙ1 Ĩ•µæq-v€Ú2gU)¹µ´·þÄèÐOg Ù™Ó£Ú"d‹L'§zƆ;Ì-ÛEÆøE@Šô=‰\zþÄ¡.ûZ¨ixo_-K)mÃHǬVò aù3Î+ùйàë•ðãŒó1 ŽølOr郣#i–û³ý`9vçéÍwÏšï5¦Ïü¸°zgöNFº<Æx‡8º4cñ¬I‹ËW¨šÑ³;”*èÙØð±Å–ý×au BÛÌÔ›Ýp^¦Ô¤˜ÝF+OBÑûÖ;%ñK¸8|ï…²º’øØq•ùNÒ~ÔØwc¿uŽ¿aø.Aþ÷å5»®Î=ˆ“éƒX¼}ƒèkÑÌã,OAçïáék=O@!_5±D޵Lixõqõç¨û endstream endobj 1179 0 obj << /Type /Page /Contents 1180 0 R /Resources 1178 0 R /MediaBox [0 0 612 792] /Parent 1151 0 R /Annots [ 1176 0 R 1177 0 R ] >> endobj 1176 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 574.047 102.639 586.002] /A << /S /GoTo /D (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary2) >> >> endobj 1177 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 119.751 128.791 131.706] /A << /S /GoTo /D (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows3) >> >> endobj 1181 0 obj << /D [1179 0 R /XYZ 96.768 705.06 null] >> endobj 1152 0 obj << /D [1179 0 R /XYZ 101.643 577.534 null] >> endobj 1171 0 obj << /D [1179 0 R /XYZ 127.794 123.238 null] >> endobj 1178 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1184 0 obj << /Length 1002 /Filter /FlateDecode >> stream xÚ­W[s£6~ϯ If7;ƒC§LgÛÙpCk¹µVwŸf¾ë?ü^bı^#ª¿ù‘G®×Œ—Ç„Küi½üÃr¬„ ò}̓¤úÒߎïd˜îø^®Ý~ƒbãX“RJø=',©šöბeõ[}R½fø_IAYNwq›Aõ'5{|‚ó–½^äw}•ž©¾ÒqÉÌ[@à K:q|(´Ø_fÆ›gœg@¡ÃÃÍõi²Gt7èÉA©ØeS?²eÝ—Aû%Údý§êhÂý?èQ;Q1Cµ}d¹ŽHw@]}·+5E ÏKÉŽÉ;vÊœ±þLÕ›êmä©f(:`6˜Ù‚pîuuØ „§gª`o¼Ò6ú@#t§l&Fèˆ`è9`áG3:½ð1ÁñدTz¯RÀg ¤yÙ¸ãÙxÈ”q"Ïn)( ûçÖDJlE)oíLY^¶/PŽå^Š \°z5r[”ø¥¾Ïøà²yŸå¯¸Ù‹Bì.Å% E\Ü‹qqçð£ž'¼=÷òjõ¼^_S*‚Ëå¢x(²Á¼‘à\˜u@¼$ßúÒz:wL÷tªØö¶Ì‹øéóŸ}´«©‹)Vë¡jTˆªÑÜ« Uepr<ìÑ©‡'¹õùʶµq¶ý'x ®ƒ7ã£1”“ì!ô,º:^|Ð@߇|çìÜÀ‰²½ÞƒS`–Æ¥h¬aížH¼Šªj¸2x°QÔKw0±þ˜×fbóµ‡â›ä4D\R©a6•ôïÖÍ^,ìÆêóYG¡‰Ä©°î¼ï+R×9õéÙi𳑡#D‘ú2ôá"¤_G%íQ¨eMC<¯Á͈”³!„©ðuÿݰÛåFTȯƚeuzd•9d^ìÑ‹!²'$%N£&¯ôªÈE¼õrƒE#lpw•+‘))™yeRË̮۪û‰BŽa½!‰˜oYe 2OÄÆÄuÌn#*ËÊÈѵêú»Í©Y½¾ï˜ g¡Ó!î´ *Ùgÿ—‘f1€Ñ’y7;Tˆ¥lÇnÒ¥ÒáýèÔC}’Jüª'û8††÷×Ïv+^³ŇFªÉ~r:nêrké)ê ÕmFߕʕozÔëËWMMçý Žçêò¢Õb;°ÁžIU5ÐM¦aøÎ4TÜß`Y­Åˆ]Oa['!gˆIÞrCó‹Î}yóæËòæ5”Ü endstream endobj 1183 0 obj << /Type /Page /Contents 1184 0 R /Resources 1182 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R >> endobj 1185 0 obj << /D [1183 0 R /XYZ 96.768 705.06 null] >> endobj 1182 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1189 0 obj << /Length 1160 /Filter /FlateDecode >> stream xÚ¥VËnã6Ýç+Ü ˜‘˜õVQhhÑEѸ«™,h‰¶…‘HA”»_ß{IJ¶<²“ ‰¼<<÷Mò·ÕÍý£ÎR’F^4[mf4‰H”³(Œ‰+«|öÅiyUÿØÕ5oæ‹Ð Ÿæß¬ ‘uC×½3‚MÑéXÜÓ»»¼‘õòñ×?ŸžçÏ«?îip¢pá¹$õèlA)IC«p‹@»äM,íŽ!¼T|B“zÖ ÍÏ"ŸÇ\Úõºá/cϳs¯b~Fßµý½YÑ UÖU;öïdt‰ÎZVlÌÜüC¡ˆèÊÇH ÚœðFgAo×@­ß®¼aÀÉ~kÅG-(å+?êÿ>ÔwI&W³AÝ”$~<‘Íþ†ƒW<¸TØÙEý¿ 9zw!_OqÃZv5Â!I“ôz„iHÜè,Â*“hÇéDY¨u—¬®ËŽm‹Š+¦’‹m»;5h2ÚÆî1û†e­Ä𺠯ñGg©EH;èz&_a¨mÒ™’­i´ÁÀÞ ñZô{Ïqãù$¥ô´¯YYž9…ŒÒvŽ`ÂQH1ѽÃÚ÷ªŽKWJS›ýnÇN­²ë6g‚p‘còѾ_Ü9´jè:Ÿ>¼öˆMb‡Ò(I‚©³v´åìº-éÆš¢ÉJ¦0É.òسܪ®ªXs ªk^ r•º½T¬êòûG/Ñ€øAä — }¤íù"EÁ†¿JÙ¶…ØÎ~ å †/óÈsXi$Hæ¾çpe˜“³$מMÒïóEÄÎ:$7C†¿Ä©¹.çÔ R§2K¯;Þp3¬Á‚¯. °Ñ‚8rÊBpu& f˜r¾ðbGbî%N«Á Äà JÉŠëƒÁ(Ê‹¼È`{ì´“AßAP™/"?qV»(ü4vÌ?qòŽ›A+Í?“uÁsÊ@få=dÇZ3 0´µí£ÑØhÇWÅÖ°$ø¦ÿRXžM'‘éýŒ¤¬;K/¤0ÊBlx*DƧÝäØRgÇ0”/øáF¢¤ùW(ë2üîŒD;aª-Õ⪒â'!PYn :SnŒ›Hüªõ®Oè!~&HF"”,­˜µ\ $•¡ímeýŠœz÷4\Á鎜¾ïÈ&ÇËÖ÷gÚFÜ)Ó 0Ô¡Ö>ÙùI^aº¶éÈ|ÃZ ‚ÐyÄÔ4ª5Ì9Çrv‡q²6ëp}hÏK#od×êŸèwÓm)/;{jF'Ÿ$qØ_³C[ÁÉ2A$Hý\LœÁ#¶¥9墘¸ôìùpªh¸:ÑßîÞ^•Ä )ö²>¤áèûF°=—«¿ÿy°7§’ÐKÆEŠÀ¥?°”KzW¶‡åp¿æ8ž¨îŒï(¹5OÿIä^ÁÉða˜N`7¤w¿<>Ê`z¸€ÝW¬wÿP7MÔ%DË‚öðJXÞÞö[F³=Û«åç§ÏvJ¹ôÞJxÑúgï­dôÂðRB1ÎA×V`€4ˆzó°ºùR<‹ endstream endobj 1188 0 obj << /Type /Page /Contents 1189 0 R /Resources 1187 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R >> endobj 1190 0 obj << /D [1188 0 R /XYZ 96.768 705.06 null] >> endobj 74 0 obj << /D [1188 0 R /XYZ 97.768 316.942 null] >> endobj 1187 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1215 0 obj << /Length 958 /Filter /FlateDecode >> stream xÚÝWMÛ6½ï¯p·AVVŒ(ꃪCdEÑCëœÒX‹¶ÕH¤ ÒYï¿/¿äŠZzÛM›¦íÁ¦l‡œ7oGo6W¯îP¾ª@U¤Åj³[Á*]y R5±iVKFz*ÖqžæÑ7±Í_?%yÒµBê ¨?ëŸ7ß®’U !¨rç¥ÝÙuÚ”°=LŽëúF> ôÆ[ž&ê Ð÷!$ôÂëÍZoû‘ ‡vK:û“ŒûcO™´¿¬Wûܺã3î&I×ñ{Ú\O¾ºƒÙ ˆxÚ_íök‡Š™¸ÌT¬Æ`踌åH˜Øñ±{p\% ÈŠiEë\ÎsWÈSyñ½¾¢e{ý¸kG!LfúåK;~Õ ÀŽ]§mN/†‘~LÎà»ÐÏ(ŒiàÔ‹dþ9$Ì— ò8„Æ9È`qR\}Êf[Þ÷œÅŠâóofà;öÉ•Ë"[ìS þk¦ÿ@PG¶•-g!üÊÌx{е%8­|Új¿@Ç>Ó6?¼{Ò]¯¾å¸ù»ØØŒä>à+SÇNÐļZdl:ñô¡k&Ü]+ý8&lMAÞÚÿt´µ ðÖì Q*Œ}üz2~¨‘³ßò®†î¹“¿?ß7ú9€ç–ž.ÌhÇ@¶=­u.œ£“PBJ/¬°”¯“Û¹€œÏpêÉ)¼ìÈ” Ta«ó³h™¬ï^÷ã´§»5‰i:%f¢\šŠì¥¿ã©ã{üŒæÖÐ3—ô>ñÿRŸÿ‡¥þ©¾Ïİþ[Äiàªôž¡N¾˜b„‹lÙ¢Ÿ<üaAc œ{­Û,uDŒ>%=Ÿ¥ÑrÜòàùþµS°¯]°q s´qý%s¢>u¦¶"Îm©ê³SD;#_¬0—…Êt.QL]KzšY:óu=Ë:3¯oÇcû˜/Ëi'¨C¾'òY®.uóšÁs΀j.bUa¥µzÓñuœâh»V_Ö1J³ˆïô¨sl¦jÿ–Ü1'«~PveÄek5J1w“E$€;¬T.ÏìuU;ñ|~dõâ…’s ù/ Š<ú•np‘´þÅt®‘öH†¥fJÖ±±¿4c¤ºo\i*9…ÉBÎz*.jBѽ†«vâeÕ›dŸ¦lþìº~’³^Ô6',Ðe·ì@ÇÖòkµë3ì×óWÓeŸ¢š%²Oõ)¶·[ªç‹Ç1t£½'ªk'EJ²Ò `ìP…Y©-¯Þn®~°>üò endstream endobj 1214 0 obj << /Type /Page /Contents 1215 0 R /Resources 1213 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R /Annots [ 1191 0 R 1192 0 R 1193 0 R 1194 0 R 1195 0 R 1196 0 R 1197 0 R 1198 0 R 1199 0 R 1200 0 R 1201 0 R 1202 0 R 1203 0 R 1204 0 R 1205 0 R 1206 0 R 1207 0 R 1208 0 R 1209 0 R 1210 0 R 1211 0 R 1212 0 R ] >> endobj 1191 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 605.928 223.31 617.883] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms2) >> >> endobj 1192 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 582.017 218.08 593.973] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup12) >> >> endobj 1193 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 570.062 218.08 582.017] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args2) >> >> endobj 1194 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 558.107 191.928 570.062] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx2) >> >> endobj 1195 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 546.152 218.08 558.107] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup22) >> >> endobj 1196 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 534.197 207.619 546.152] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions2) >> >> endobj 1197 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [175.601 498.331 181.468 510.286] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw2) >> >> endobj 1198 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 378.78 128.791 390.735] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms1) >> >> endobj 1199 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 378.78 223.31 390.735] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms3) >> >> endobj 1200 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 354.869 128.791 366.824] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args1) >> >> endobj 1201 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 354.869 218.08 366.824] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args3) >> >> endobj 1202 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 342.914 128.791 354.869] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx1) >> >> endobj 1203 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 342.914 191.928 354.869] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx3) >> >> endobj 1204 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 330.959 128.791 342.914] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions1) >> >> endobj 1205 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [201.753 330.959 207.619 342.914] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions3) >> >> endobj 1206 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 319.004 128.791 330.959] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw1) >> >> endobj 1207 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [175.601 319.004 181.468 330.959] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw3) >> >> endobj 1208 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 259.228 128.791 271.183] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms2) >> >> endobj 1209 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [217.444 259.228 223.31 271.183] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms4) >> >> endobj 1210 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [122.924 223.362 128.791 235.318] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args2) >> >> endobj 1211 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [212.213 223.362 218.08 235.318] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args4) >> >> endobj 1212 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 139.676 102.639 151.631] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms3) >> >> endobj 1216 0 obj << /D [1214 0 R /XYZ 96.768 705.06 null] >> endobj 1217 0 obj << /D [1214 0 R /XYZ 123.92 609.415 null] >> endobj 1218 0 obj << /D [1214 0 R /XYZ 123.92 585.504 null] >> endobj 1219 0 obj << /D [1214 0 R /XYZ 123.92 573.549 null] >> endobj 1220 0 obj << /D [1214 0 R /XYZ 123.92 561.594 null] >> endobj 1221 0 obj << /D [1214 0 R /XYZ 123.92 549.639 null] >> endobj 1222 0 obj << /D [1214 0 R /XYZ 123.92 537.684 null] >> endobj 1223 0 obj << /D [1214 0 R /XYZ 123.92 501.818 null] >> endobj 1224 0 obj << /D [1214 0 R /XYZ 127.794 382.266 null] >> endobj 1225 0 obj << /D [1214 0 R /XYZ 127.794 358.356 null] >> endobj 1226 0 obj << /D [1214 0 R /XYZ 127.794 346.401 null] >> endobj 1227 0 obj << /D [1214 0 R /XYZ 127.794 334.446 null] >> endobj 1228 0 obj << /D [1214 0 R /XYZ 127.794 322.491 null] >> endobj 1229 0 obj << /D [1214 0 R /XYZ 127.794 262.715 null] >> endobj 1230 0 obj << /D [1214 0 R /XYZ 127.794 226.849 null] >> endobj 1231 0 obj << /D [1214 0 R /XYZ 101.643 143.163 null] >> endobj 1213 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1243 0 obj << /Length 1246 /Filter /FlateDecode >> stream xÚÍWßÛ6 ~ï_‘µ‡ž34nä‰=ìt@¸á°—eOë\GI„³%C²/ÉÐ?~¢H'Qι­Ø èC ™¢>~¤(RùyñêýmœŽò0ŸE³Ñb5bQæÑh–ÎÃÈ.,–£?±OÒ( >MÓ©0a]´Zìàcwe:ýhg ~¨´µ,jnüu\ûq‚c%L ë¿þ~ÿEVµ-Z·ÿ\ü2šŽ&Œ…y:Àá;KBvU…&*µåúÀÁn}ËÏ£iå™t@+TñœžäyÏç¾EBö™³IO˜î®º¦¹¼ŸÒœXšyÄ|¬õÙÞÜìi`jaŒk˜®:韋ø”^óG.Û×_Ee4ImZ0§‘ä ®lˆäì"³yÚ‡¼©T;)U]+9)ôÚ ˜ÌÒ0a‡ bÀm@Lz…$“ÍÃ<Ž}žRÎ÷´ÎYïð(š¦5÷wﮜl0'µüœt‡ü?ä¤H­3驪›”>Û¸N/‘£¤Â+Ãà¢YÈRöl Ýü*•\…B¶¾·îû8èñ@Ä ìŒ'̺¼© ýЯ¯M*[‰hZª%ÿ‘€ŠÊPã ‚.¶ŠÐ à N·®)ó¶4.•¼&?›Ž&À—v$j7¤_ri”¶€ä6mÍP’{õ\Z»¢ÅrMµønß5×Çêñå ŽÕW¢ Kµk6×Ϥ}Ìì-Žž=ð(Â,Ï/ÿ“±ÁK¿ÒÄRäEÒp:…®!A«‹¶ø§æU Üò’å‚Z:ä³N4xKZ^7¾žæ °ctÈ—ëvƒ|Ÿ†¼FÑÕûÓG†êè·TUÒË‘:uùxD_¾ùä•széÈÍþuäd Ü'|ïIT*—¡f°cÖJhÓî}¬c?ÁUò^jµ½ñânž˜9Þ‚ìć<œ³˜^d{•/žÍƒ]]ìpfß]Íå8Ê‚%ÂàØ¾ÄY«plt' àsQ–Rèq‘x§³ ¨P‘MîC¤_)ºÂµ"UW‘œÝñÄ"(,Mˆ–„y’øÁ]l€}厂ÏÏ¢Åo@wÄ´Ì5.Ø"Yr\Ûa˜o øx<‘€ 0.!v ”¶?ÊJ4c4®»¸U2ìbû–PöN3_n ̓Ãq€p ’BKë=£GåYFÙ®eŠ5ôÅ8š+­j;cd D\r]íX€÷%®hÕµÂâJØÍ‚КŒóà!uÃ{ /…;Y» ½×™¶-Y¢Ä‘¦ÕB¥v±ÜxTBž5\ò¨ySwz0œ&‘ý\;úÓ ’T¢‘£ªæÕ%í¦ }%{olûÿKa¸_`}r!¬€"êø !ûWL“²æu!$åè™[w8¶0.bqÒÇÐÊîpœÃA,IQá¨ùÊúãòv! +ÿ4e‰­Àd{ÃÂD²pžÍ¿Å Þ_Õ“ç ©cáû–j¼Ü»êO½Î…‡@4ÍÃÈVÁçšË2ûXˆ|NÔf¨âÁÔV¼‡³÷#V;È÷š„1˜ÎvG5œCÊ: uè}UQèîšš˜zìßÔƒ•¨µœ­è›„à]BÞƒ<élžØÖÁ32̲ é²$ÍW¯þÞÎZà endstream endobj 1242 0 obj << /Type /Page /Contents 1243 0 R /Resources 1241 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R /Annots [ 1238 0 R 1239 0 R 1240 0 R ] >> endobj 1238 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 561.523 102.639 573.478] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args3) >> >> endobj 1239 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [186.062 561.523 191.928 573.478] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args5) >> >> endobj 1240 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 151.631 102.639 163.587] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args4) >> >> endobj 1244 0 obj << /D [1242 0 R /XYZ 96.768 705.06 null] >> endobj 1237 0 obj << /D [1242 0 R /XYZ 101.643 565.01 null] >> endobj 1245 0 obj << /D [1242 0 R /XYZ 101.643 155.118 null] >> endobj 1241 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1248 0 obj << /Length 944 /Filter /FlateDecode >> stream xÚW[oÚ0~ï¯`bt%nâ\¦1i“Ú‡©ÚËØã! ¦Xä‚â¤À~ý|KHB`mœŸïܾsl¾Ïhàß±œÁ|=€ÖøÖÀA.°Ø‡ùj°ï h"s‘˜ä÷²ÐxOò\a¶Â™\³•\ì2’*ÙQ>Þƒ¨À §ü´&Í÷Ëù90 >R·ïrÓC>¹èØýüýò¢ ‡R‘SåKzÒë2‘ãx—4ñòà5ÂåÙv)¦ë5ÅyS3,bZÄ\5äÿ̉ä68Žk]¦‚k¹$Ê+ø9¡yä¥îã3´ë…2mà˜.C”HrK£–†7Žç(“ŽŠ;'1.½§üe!ÂÍȲËE²>yDÔ!fòyˆƒC=ж³Èò{-ºD‘a!Ÿ”‰y^0ÿÅûƒ|È:1©.ùŒª°7b·â5,“u‘Õ´÷ÖuË%QÜg9"«7{ˆµk³Ð¯Áª\L8Mþw ‹°ýÊk¨MV y6Ó•ÚqïÛ}¾Bl»Šf¥4¯ÆG,WŒh ñ‚HïÊBAð•&½¼Ü0Ðßx÷4¿ûЦ›v endstream endobj 1247 0 obj << /Type /Page /Contents 1248 0 R /Resources 1246 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R >> endobj 1249 0 obj << /D [1247 0 R /XYZ 96.768 705.06 null] >> endobj 1246 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1254 0 obj << /Length 1308 /Filter /FlateDecode >> stream xÚ•WKoã6¾çW¸NÈËz ¨hhÑSë=m÷ H´ÍB¦ ‰Ê:ûë;ä²dKvöbó1œù8óÍpôûúîéÅg)I#/š­73Ïó‰—Æ³(Œ‰;ëböÅy(«í!—«Å2ôBgÓŠ\òJüë†î~(.S×ýåø¸XRXvî›ËyVâVž5Ìž­åŽÕ8yeeõmñuýçlé¹$õ¼Ù’R’†Ö,/˜\¾ß0 F•Ž1¬JzþYÔ,¯¶‚gÅPβzÛîÁÚ\«UêÜ¡&»þôBƒž¿–iJü8ÊnQpàX«.BVZð þ+˜¼!ý;ÂX™=ÝT Ž~]v·C{Rßyàñ–•Ü\þÎ=\¹9° ¥Æ…M[¿ Jæö¶¦ÜÖ¿ÝOp=Ñ–¥>Ú¬înwéXJCâF èÓz6#.]¦¡Á™ïQñÞÎèª Idœ<©÷& º·£ü¦CÁ ¯ù>n÷¦H7¦Í˜Mzr)‰©R°â©õxáǾSd2ÃQ#ë6—mm66U­c·!ÆümFŽÊfµr(+i”T‹¥—8¬Á©°ë\äe[…MÀÀ­Î•´ŽŠ‹üI~ˆÀY5æ|]µ’ °¡  Hg‘“Y ʼ(RÆx­†Ptªý¡•æ¸ÚË:™uŽ;LjÂD1ÕÌ‹S'ª,Y¶ðbçMm\â–¬pÅÍìàJèP ³ÖAÁ- œ‹-.–èÍ\C&cÙ„ÑóÏ}¦ÜïÖQ(Ñ…F[H‡6Zªœ.û¼è Sâ2D) ¬ÀÊD’8¡Óïùž7 ¯î¹:^{ÎÝ”$~|õ9O¸pô±þáøó‰É}³³É‡ö8|hû žŽMž•¬×‰Ò‘vêòR^Dhè]½S]4¾Vý+î§ön 'I7 çýûÉý:ÕmaÛçShÀ¯èrmkhœ÷C-ÆÕîã¼åíõ‘÷hT—}5(Ø&kKin^™Æ ÛÝ— Ó«–úñÇ^Ùü«‡KËF&¯ÄF}-äÌžÙsÙŒuy÷æÅFAâƒÝ×^‰òÝŽÌVµÝêo{ù ð.ä°}·«Ã¨(y¡ ¬Vô#}4¤Іe§M)?’ËÎ}ý÷ççIâLœÞ$øP—dûÃï2™ï|hôñæúãäµ’»¹Y™+ÿw1˜2ÜN'U 25P†龃xÿ;è Ÿñ-ÀùGb¡ ­V'”×ýòÛ_ÿ<ú>ÐÆe§5b[»ëfäÇߠɰ? qÁ†ŽÃ§$1”%x÷¼¾û~ñ3v endstream endobj 1253 0 obj << /Type /Page /Contents 1254 0 R /Resources 1252 0 R /MediaBox [0 0 612 792] /Parent 1186 0 R /Annots [ 1250 0 R ] >> endobj 1250 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 332.951 102.639 344.907] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx2) >> >> endobj 1255 0 obj << /D [1253 0 R /XYZ 96.768 705.06 null] >> endobj 1256 0 obj << /D [1253 0 R /XYZ 97.768 420.125 null] >> endobj 1257 0 obj << /D [1253 0 R /XYZ 97.768 402.247 null] >> endobj 1258 0 obj << /D [1253 0 R /XYZ 97.768 386.307 null] >> endobj 1259 0 obj << /D [1253 0 R /XYZ 97.768 370.367 null] >> endobj 1234 0 obj << /D [1253 0 R /XYZ 101.643 336.438 null] >> endobj 1252 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1264 0 obj << /Length 1133 /Filter /FlateDecode >> stream xÚÅXIÛ6¾Ï¯0R ñLmFµñ¡ 2hŠößÒh‰¶ˆ‘%C¢¼üûpÓNÉ3ŠžD“oùÞJÿ¾}øø EBÏöÛýZ6Ãpá¹>°‘»ØÆ‹ïËÃã¿Û?Öb !]½™ˆÍÏÈëÜF ðN( NiÎÖ%aÕi}ÄÅK©è{Ò8Э/P)f:ÀµkÛNèÌËrNçèNi¿QÊ>Ò–{êè—ǵk»Ë¿¥ ¹ÄQ‹,gjqJqDbµÎ3õòlOc’Ešx‡³¸4A+LP$ŸÖê[Ó?–k‰ƒ•ÚIIv` È+¶É¢ª8~Ì¢<52ãûœä«×3£{ÅBܧ%Ȫ#)hT«=Ê«òº$l¶û*”yÁF·j6!ìA½Mô}± ks¸ð)4ÄZ/©¼v Õ_eÑÒ»š’%¤vZã½ûêGÀ×–×´J)Íô)»†ºi”¸MÌn}¬c‡fŒH±jêýª$íMƒÃ¸¦6´¬å>/ºJTÆâÕ*/|þ:!À… )ácδ‡O›áqp«I ¯¥µ^‚&C&u…®Ï]#œ¡µ\–Õq(“k¡ù¿dùE{ã’`ÖÐÝуw<z$Xß<Ðsí2¬õ¯ÒtNÉ3‰X>K`Æ!ŸRåпÍý YR[•×å&¥/Ä(I»†\£´Šiv€UMQ¥˜¦[7D–ÏuÊê´€m7exo¨Óœ@š®L£ØÛl´‚ïß×éÕ­]šª©A¼õ„6ìóo2sXE5ƒU'ƒºëN`JëÚ$†¤uTÎÔO¦ã·¨¯?­GÉu´îô†6]ŸP°¾)ÞД)áØZ!ý¢=ÑÌàðê0 N]BæP]ñèi¾O¶¶ÏwßW°½kuz‰ï[͉fý‰Æê±Ç'ĵrfØ ‚1O†º`òÁ˜íèÂùúÂGß>/éE9*½Æó±!³ÁĢwx œ¼L@¿f¯sС Q<ÿCŽRÈuäØ"¾²3ˆEAJ¦V¢‡õŽä¸®–²—÷Å@¯9pÍy3Z‰_áò˜—,½©“?ÚÁò,Û‘¼œ«o¬¿Ê’FkÈj¡øZwÐ5t@è8}hÅÌ9Ž Û ë蹦¹!Û²kÑ| |,½& o¯às~t=>lP¼KÅé@¤F± ™@‚‘"ñmŽû‘f1)4¥_|%¾â<’œFX¡)5mx¼Õj¸šø:VY”†ˆj™jXÂ.4MÕÞŽ£þ’(Ô‘þ¤ë܉câÝ„¸JÒŸ¨ ¾¡ƒ‚¯öU1ÊVã8?ðÞòtlßð¿y:šßOOáT&l~è¶Ôѳ<òg«VðF1€SV¯ëH’ëˆðù·¿¾}¹÷¼iµÞ<Ü>L÷GÅÛs|éÝ»ÿ© tƒz!°,ñG‡>TÞyø²}øÊGy endstream endobj 1263 0 obj << /Type /Page /Contents 1264 0 R /Resources 1262 0 R /MediaBox [0 0 612 792] /Parent 1268 0 R /Annots [ 1251 0 R 1260 0 R 1261 0 R ] >> endobj 1251 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 641.793 197.159 653.748] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks2) >> >> endobj 1260 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 621.299 102.639 633.254] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks1) >> >> endobj 1261 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 199.452 102.639 211.407] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup11) >> >> endobj 1265 0 obj << /D [1263 0 R /XYZ 96.768 705.06 null] >> endobj 1266 0 obj << /D [1263 0 R /XYZ 102.999 645.28 null] >> endobj 1267 0 obj << /D [1263 0 R /XYZ 101.643 624.786 null] >> endobj 1232 0 obj << /D [1263 0 R /XYZ 101.643 202.939 null] >> endobj 1262 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1272 0 obj << /Length 900 /Filter /FlateDecode >> stream xÚµVKo›@¾çW8Ž'Qذ¼ RÝCÕäÐcãžÜöª° vI°”ß}šcœ¨éŘažß|3»ß–gw®?‰@8Ád™N ã‚È™þ8ìÃòy²ºÚeÅæÚòÿê‹%Ÿ¿mßfÂÅb¶›IÉÛ›|Ji³›1xý´ü1±'„ ò•³fÄYóQgL!n®-È<ìÝqÿLõîzF]–c³º`×\)v`8J”¶Ižçˆ„7üZcž”N¬-Š€;˜/á#cÍCà±<:ù˜Áëm\ÅkšT§BVž…£ERóãY†6ƒo¦ñßKÊ5•=âC>,þºêÏ¡û5³å!ŽÕa{ °ç£u„.Bÿ«Þ“/¤-K‘Ét¹×qH'IF’£ÙQ:áæÇi+5.[•%)“5Š3ù¢ÒDÔWºË”b³ú\–J”Æë6ÁЍ"TyÐö¸ òϽ0½¼LÐ!j↴ø_^daÎvw! Là^‡…ÐvÐãx/hÇNÇDx›Tˆ‘Õ­”MI]½¤ˆædª'Ïœ@6P€^Ð#"+©K‡2ó;å_¦2²ØWµæÕmzNÒ¸Îè {}Ƈł¡ŒµU< E„s_O™Ô?$¡ué 8}àÁ½`÷èi……¬<œƒÈu»•_„J(ehïëÏc¶Äˆ.6¦ºÚ2ÁÏšw=(X—«M'˜žšåÿ¾œw9ÂÝÎRÇæ_ÌÆfëÁ©Šš!¡Û[$ÏTõÛøE‰ýÄóJMþ÷¢?äUŒ7*àºÈ˚Ƙœ‚•³–­íÇc [!5&»Jø®×#Kw#Y#+)-*%cãˆ^´Íš½%ƒéŽà Íto^÷ô{ÔŒCøZZ4ÉËÞjÏㆃB(Ê“v‰Üȶ÷4UÚ3æ=‰ à0`¶nwë­áûÖÜí¦;ãH<7«õàÚî(\ÜÐ9Äz׿*z%+ªä_± QŠ0¢­ˆ§ ö·xÿj?ÃÂÌ^ÄZ9O‹l{¨…]¼@òãÝ>¡ÅÂþü½#3ƒ½Ì´ô„CJÈtRGlO®T9®³LÔW—¥¸ˆ½Š£÷Ä Ô<0Uç¥ã•||,W’¯ÜT<ÆMß}»{Í¢|¸lµrΆêÔeÇ0W<»_žýÅ\3 endstream endobj 1271 0 obj << /Type /Page /Contents 1272 0 R /Resources 1270 0 R /MediaBox [0 0 612 792] /Parent 1268 0 R /Annots [ 1269 0 R ] >> endobj 1269 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 404.682 102.639 416.638] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup21) >> >> endobj 1273 0 obj << /D [1271 0 R /XYZ 96.768 705.06 null] >> endobj 1233 0 obj << /D [1271 0 R /XYZ 101.643 408.169 null] >> endobj 1270 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1277 0 obj << /Length 1398 /Filter /FlateDecode >> stream xÚ­WKÛ6¾ï¯p²@#6#Qo F‘¦ °EÑCᢇt\‰¶Õ•)Wµ”__‡´d[òæƒMj8šç7ÃÑ/ë»÷Ÿo“8 Ál½™9Ô%1~H¨ëÏÖé싵?®{ÿY>ö|ölé8$ö5KÍ÷‡n¾ô©oý´Ä5ùÇömE_ a“•UÝIªƒòÎô.©-õ:çR¿A1Ï+ŽâGL(™Ør0£ªšòå̌ӃÈj¾Zÿù×'MŒ”{E0–‚JˆŠã Êlƒœ xŸUU&¶°Ý4Þ˜ð/öˆãR”±qlº„Rÿ:®íD\Û¡CíÐÚ¡„¡µoº¼Ø*Æ©ˆ]%­Ûgâ;¤LÅ9îϬ QiûµdGmÛN§õ‰UY¢·E;æáýñ5ør‚ÀyÄng}¡$Ò€8þ…{gâ¾D?vUÂr¾³³î|õN¼Ó/Êü¬äKÛœ=­àϤž»þ¹emµ‚¿!ä¥1ý¶`7š‚/¿Yá:…ÙÈ%AäßÄl@IxÓˆ;Ü•@lªŸWšðVRÞölK…B[»h÷¡qìÞÌÓÔɰ§[Æ«ö¼W–=f5é<ܨü[e ù|Ùq-"~"Ûª€†±Õ€É4²­b, ^H<šDõ^õ y zFì TXÅ\þ½ÌihiÝY¥ÏÑ‚§LóW‚³çn¹™/=XÈpLÀ”z7w,í@ž nŠix¶ŠL€ÖZŸ–ES÷¼u¡µ+¹±8ç§cü˜Û¿À#K¥—ºJ†Z!ª°&±Ðñt âû±‰Ð1«wÒ–ë@º>‰£AØÕUpR±ÇÃe›O,×jYÉö¼æ¥áe5îR “#;s¹Gß‘é°ï–0`¡PR¨µ)ÊÔ«08vH|yÝœ·åv À/…6FDRJ~À¸*K$!«eN}ßµþÞA;dO³T¼Ó'YZ”‘^Iß4©”[¾ýŒ›BkmUÉü ¨õ;K ëšGÂÛðråh&]ßÂFÙVQë D …ÖãØ•" *Á׳˜Ùì@U#Ò’§R”B¹jXI‰‚œ•êwuš\ßÚ1]" ˆJÀµÕݸX¹möÜ”’P tA‰bËæJ=à’gØžŒOËâ0zÃ˨ÙñHì]`_5×u-Á[-PÎ)I¢Âƒ´Àó¾$u£ ‡¥rõ(EÆZÛI¥œƒUîtAÃJ!QxºH ê–½×bC)U±fÏFÈ;1¬ó¡l‚!½ TþXJ`ëžÏp©j€¼º½µ%ºŸ—|_¼¨:ƧTÂ…‰$tà`0P" YÄÙÞIÒè{D6²Ñ9EÖ¥o[’íż½Å\bµ¤ïÏ„‹ª(Mè/¥¥Eï‚™ÞŒ+€­v15j;Ô'>nÞëŽ-/ ;TîT‰º)E_5*”÷,ÏõU.oÀ£*|Lê×ÕéùãÃX¿87#ϯ̘‚×$•²âÌ rÄzWŽH h¤ŠîÈÌŒ B­Ùþg›®·õãî_XÞX1€bSˤ½žJÀÅFûÜ~ã¼øZèzÎïô9'Îä\lë]®WÒ&V+7åÛY¥¾Ú•§»U÷úTx!‘ð6øE†t¡{ æÄÑ'75^‡NÍîô&¨Ì¤x)y‘ç,ÑŽ¼ðò©0^íd+ùZ¦ë¥âÿ5\$0yöÌ'z– …™~ ŒÞê€xÓGÑÔ%îôQ> endobj 1274 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 319.004 102.639 330.959] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions2) >> >> endobj 1278 0 obj << /D [1276 0 R /XYZ 96.768 705.06 null] >> endobj 1235 0 obj << /D [1276 0 R /XYZ 101.643 322.491 null] >> endobj 1275 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F8 112 0 R /F33 118 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1282 0 obj << /Length 1116 /Filter /FlateDecode >> stream xÚµWKÛ6¾ï¯p³ b£+F¤ÞE ²‡"·øæú •éHY™tE:¶þ}ù”E[r6Àö`‹"‡óøf8üô×òîÃcÍ2Å(ž-·3† á,Ž€ÄÊr3[Íïô#¾¬p·ð"Í‹ëkèñÁÌ‘”1‹;\´zÈ+³¡¡ô™é!Ý^®ÕÄŒx·Ç`±^þ=óg„ ‹Œ+ïöZäOO?Ûÿ#œëÄ|Û`òWll禥{æn…úñ»qû°c‡Ô%¬<­Ë¾Kõkùg°¸×›ö´&ܨ=VØÂÒ‡TÚf¨0ž¡dР[¶¸àF¨0ÈqlÂÜHÉkJœb½³6J äïjòn`N9,v~x„á ËY`˜uJÍVK¸u0DVâþœ¢‚ñAäzX3ÇyiY¹ë¥)ˆãÄÍéÅ,¥ÿ§\›BzF…`gqQV9’Ù .3Ü]i6šº•2zL¬R5o¤™;ç½—?³ ‡\Àê!d¹^}A÷ÂqÜ0<¢ÅFèV–¼HÒ׆×LœVd}…t<…ôÈi¸¯søJ65ã*¨\Æo¬uy§=š¶G,ÝXºvBch;LqÜí³‹Œ=ÀÊ=ë™yJiÀë¶åˆ £mÿbÅQ  4ËnÖJDIê†3l¿Õ ÷µÊ[ïÔ<Á0atÓ,DˆPvQ'œÓJ‡¼=Ï+Ü …_œÎÿéèA ‘xÛ E/Äíž6²‘«2 Ô96—Ñ÷H¿$úb¿oéi¬ŠNvb‡yE7ù›’Æ Âߘùmî«Û±{EàôŧÜéýéΕ›¿¤zoÙN‡×H ls È‚@K-å…Á¼¤ ¥óz çœêé¢ä‡¢i:ý&NæBµŒ¼“å+fzuK[³¹ZÀ¹Q´o(e ?³<´h" —%·1öQÒDW€á“ÒH=‰5Æ ‚!ÈÂС2iŠ‚H“þ`Ê <Ù–F4†)Hý¾Ä둌;ªr]ÖÁeÆJxE«†écK¼x[”ÏäoÀ”D3zÂíø%z S ï+•6öjâåØMw–Õö]èjfCOš„ÄðÖð’‹>}¬¨㱪˪÷/Ïë)¦[´»yxÙ¨»¦iTãxK€wXÝXS°âø&|%¼[k¹îl¦ T3ï-Õ6|4òr|¿oŠ[ò^óÊa±zkMÆ8ÁD?W×ç¶n·½Éî®Ìÿ ^JÏã§/_?;¼`øõAèù 1n2®í™r ð»[€ð"΂J05üÓ¼M9 Ÿ4Ho_hg´ºUýðÝrDe_B&¦¦@”jÔÖç“T0•ÛTE|ëFA0mB~æù{öþl"uA HÅ B"ê4ÕûaJÉ»ÏË»ÿ{:ä endstream endobj 1281 0 obj << /Type /Page /Contents 1282 0 R /Resources 1280 0 R /MediaBox [0 0 612 792] /Parent 1268 0 R /Annots [ 1279 0 R ] >> endobj 1279 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 283.138 102.639 295.093] /A << /S /GoTo /D (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw2) >> >> endobj 1283 0 obj << /D [1281 0 R /XYZ 96.768 705.06 null] >> endobj 1236 0 obj << /D [1281 0 R /XYZ 101.643 286.625 null] >> endobj 1280 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1286 0 obj << /Length 909 /Filter /FlateDecode >> stream xÚ½VKs£0 ¾÷Wp+™m\lÞ;Ëeg¶‡=ç–ö@ÀI¼K0ƒI“ÌôǯŒLé{ö@eÉú>Iù9»º½s}+&qÀk¶´h ò¬À ƒYnÍíB”\Ý;¾“KÕðJKûýÍdê3ß>`IÍ¢h <óŒ=M&‹žM±Ëx´F{Nf¿­)sH̨5¥”ľ‰Ê ÅKÿ$ªšCÅýóIN?¶X¢©*)·^‹L¯6iý—4bÃ[×Öèé ßý-}ê­¼NwÚrœÏñŒ=/•¬Í¢ÊÖ‰6˜gô”ݦ¥§U:¦'tû13¾OàépÝÞQï¤æÓ("1 ŒK€.+4ì5Ç I{^æ:."ù1íì‘Ú¥ÜF݇yÁËU³ÖÎØTcÞíûuëo£±26°fCkæwz^õL–K"ÊæBÎB—0æÃ­ãr$Ypq÷r_é®ÅU’àûZ]_g*ôR¸ $qì÷ýtqÛn^ŠZÁm7º†R…Üñz¾[KÐi4—®·ÎŸK‰±AÊOîÛè]uzµNŸ¥µ­ªÑ Îi=‘£ô Šø½˜cóã ³œçDZ€PŸ¾Ü³0ô/äúÝóM3ò2|ª‰ÞÝ:%v¡mNI>‹ÿ±™@å“ ¦#ðÒMÃF½>Ø¢|J, MÑ)„jÚ"$úkÒ} ýèn)žÈb‹zÄõvlTí2{0u&G§>Ö+s¢>FIÄÌ´Ÿ­aüº~ÔÙ·òr;¡v™5B–¨ØMXd§ 9WbUòWi-·e+‡vŠ*ÅäÒ(ªÉ”…6ÏĽC½ •U-þF K£Hh‹œ›s„ ØHÜYlEa¢n+}¬©õ?Øv]€ª#G6$’—šƒæzvž6)Z VdÉql®àÚf?í½½6@÷<Ï@‡½ªðÏä€i½‚?LUßzl"AEcÏ  b†ä)<¯–Üntiô;"hÖü"Uõ!EmՑ؉¢@iù1ø wìjŸ·Õ‰øåØUB4;xl6„­ä¶Î ˆ¥¬Q(ùw'~¹)ô‹º1åªtw¦ÅkÀÞžO‘T"?‹AEÙ%’÷ >gÛk³ýe˜Ò<†§i AOŒQàþ1__SÞhÑÐ'^ÂìñG™™=¾¯7¯~Í®þ8ên2 endstream endobj 1285 0 obj << /Type /Page /Contents 1286 0 R /Resources 1284 0 R /MediaBox [0 0 612 792] /Parent 1268 0 R >> endobj 1287 0 obj << /D [1285 0 R /XYZ 96.768 705.06 null] >> endobj 78 0 obj << /D [1285 0 R /XYZ 97.768 269.398 null] >> endobj 1284 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1295 0 obj << /Length 1779 /Filter /FlateDecode >> stream xÚ½XÍ’£6¾ïSxC쪱 ¬I*{ÈynI ÈFY È3;©}ø´Ô {Æ[S{pYj5Ý­O­þѯ÷ï>|Šg K¤³ûÍ,‰X$ã™ #&üpvŸÏþ\Üj¹ò}‘5*5*·“`‘§&Er« Š´Å5S4о©ÔÒ—¡\¤NJÕâÚ_^襗K•¶üמ -E¼(p9m–|¡–ßÿñáØu4XÄ ‹=æ¡­:Gžáž¸ÇÉ;Ž» !’yqÜ1˜Ö¤9ã,:*J«|BO˜‰¤z.HHÆÃÞff+?âLúÁlÅ9KÂ!ò"IŸ—"Z¨gœì­yn”6eÏÀ‰‘-W\Fø'»YÚl;UY8Mkñõ$áY|{IvPW$²Þ aS+ày“x¡ZÀ9Ýçò[@[vØ9MýBN )‰Ñ;…{æK‚“-çŠT9YŒþîvîC.¢—ޱç Ð窰á~9ê„8SÜàÄ9Aª­õP©uбniÜî=¿ ᢳ´Di0(É‹íR}(éÎ<€ŽhÒC¸nŠ $À8¼c¡WøovªÙ*k-ß!½.sé|•ºÒÍs€LÕ®á@ìwvÎìØØß¹ƒ®†ÖH´æä>Ž6v lÐëãa»ÑôÁXµôÑÁL^r3öw¯ÛÒ¹åAȸ úûW“ÀNbÃÏy0ܹ>‹yðIøwr 2I£‹£°WäNeBžï3Îû«®'¶6’³¦XB( k!GF¾°9T™ÑuåNvÆ»‡±è½Œ1a|;Nê}çC{+­=úÄx+€­ðúض™ØËÊ1ßèßÜm¾KMV0{K¬é><úîdv1p¸]œ=¥•ii5þ7*«·•þøÛ½ÊtJª?«ç§ºÉé MB¬cLkæ`X7]ÌtIQ®6º"-t?]t nÆÖ’F{} ê’qhmR¥­\±g—6ŸIËÃ?*3í³»zC´ÀTõh¡†ÌÛYãÛ4©‹Þ0ôâÈñÒ¶Õ[çbs ó»s×Ϻà2<,¯y $€¸¿_Æv—ÚüŸ‘š•`‘%qJG_8 ‰ïË”=qÌDÒ»ðö‚AbT*€W{1ó 3 ª›5`Û·ò%“\ŒÙðAJºA)ø÷€°³ï~IÌ¢cÞ¾dM‡TNâ4ÏߎŸz§½¾PB =ñè¡uWÁóY(nw>á³(¾<9å|ß‚_$˜ô¾€½…W0^Â÷oÆ0´e|5†9x ¦l çÖÞ<«mŸnû°‰)†É.a9&Ê”I(ý´]Âú÷ •Ï/¥À Úuþ’j¿™Þ÷º*T£M{ZfÌí“Èü‚=v‡ÖŒ³^zD‰|²5qô ¿^’ã’lyù°Ÿ'$Á¨0›­dÌ|h~GP¼VTÀФNhJX$N5E ’oÔµ‡n‹×jJòd¬i{^8çp¯°ê‡mä/öM½…£±½Mõ£A:”-8ØÔkù¥{Ê–úYJ<‚jpXz¹CE¾J¹—¹2DPZ_ÉáëPBZèYÕJª‹&ÛK[ùù}¨’µÀ}]•­²ÐÇ’v,M}0®(¶ tR¥kíž!£n–‹[>× ?PNœÊLÕ£ xÂØaîþí_àCUk»GG„´*©s/‘DÚ·öóÁQ!ÅH|‚°GLæã8ê GF‡8¬äµÛEv*ÏüøPø[ ÉÀéΣkól‹ºO·Ô¶Ö =WÔ“M«7~7xe·z=4ú·ö¬2bÞiñ†ŠXVW¦©ËËÍ«Î×sÏGýézŽÿóa£êˆõÞÒ\C˜jp}í]H‡<‰®÷­‰`"’—“`©ª­)\´*©îýš_¿A·¬ht6ÁÞq ÿxl€…§ÈÇþ~Ì¥ÇOº°q5¯VuµªÔâØ#-€Yª9&ýÕHßdêoYVÀµÊŒÛÄ士7^eY/²Ò@dye-rbúÈ¥jrßÌ0û*v›mÎ ¦Õû·³l垸ǶuÅ7\®þˆ®Öù Ðzøä> endobj 1288 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [191.292 328.966 197.159 340.922] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 1289 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 317.011 202.389 328.966] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addvar2) >> >> endobj 1290 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 305.056 202.389 317.011] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) >> >> endobj 1291 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 203.437 102.639 215.392] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) >> >> endobj 1292 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 203.437 171.007 215.392] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 1296 0 obj << /D [1294 0 R /XYZ 96.768 705.06 null] >> endobj 1297 0 obj << /D [1294 0 R /XYZ 123.92 332.453 null] >> endobj 1298 0 obj << /D [1294 0 R /XYZ 123.92 320.498 null] >> endobj 1299 0 obj << /D [1294 0 R /XYZ 123.92 308.543 null] >> endobj 1300 0 obj << /D [1294 0 R /XYZ 101.643 206.924 null] >> endobj 1293 0 obj << /Font << /F8 112 0 R /F35 119 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1308 0 obj << /Length 1291 /Filter /FlateDecode >> stream xÚµXKoã6¾çWx]`a±¢·¬¢AMž}K}`$ÚV– ‘Jšß%“²ì¸»íÁ9$çñÍpfèß6wÏ^8K4öãÙf7ó\ßIÓtG‰ãÑl“Ï^ûåvóçÃ3LOÝÙʜԣ²f¾\E~´øeE_&eû—¹9“Ì»'Ú\í›ÙC–ÈÄóœ4ÒrÊmÃc_JáÔ]UáXR‡ÔòׯZBý«ÆÆ“Yƒ®®J!­³ZœŠA/\ÿ”f¾ë¤¾g«'dsÄ“su”Øå šq#i ” a)õbó^ÓÀ *_;ià±²ÄÆ¨J!Êzãæ(˦¦Úcg;?]ÌŠÑn«4u‚$¹î‡˱Dµi@E¯j:¡áxíÍ¥òúþ¸d¨›:ë ¹jèz ÑÛò~"qO«÷€ª¥èÉQGÖÊwö¡X´M·/¬­‚k[­ƪJüú™…:~•¡Ì ß!H ÄÇñka2¿þ…øÍX=Dk†¶sƒ·Ž\¯‡2_ ÙrvG©oG)ð–vÈqPºx»çNÖÔ²m*mI6X¡ ŸôÌ©¼´šº¹¼ÚᦡՄF“â.äËï¨x½—Åyħꧼ9‰¤g'8ÛìP€ÉF%~ϺBeÞË®ËiÒõUiµ½»¦«s;È0Ÿú÷6¥Ö¨Ôyl¹àµü4k R±ý²F™Û`ñ7¦ z_¼¼ÌË|¾ÝÞ–Ý”í ½ ‘óíª\Úòã²U’Т)`³’UDÙµT“ÂØ¥ÒŽÄîH§9k«’·¸¥ ÑU.½EÁ¤ž€âʯ^è¤ahC›CBöºԽR¤ªkÀ¯2»®ÎŒu™g .HÝE¶/:uXú‰A+Ñ„ZQ5Mù g€“¹NT…*0ÈŠ5" $6úKe\¹ö Kâ2øÖkËìlÜ7Ê AhˆtU¥â{Vý (…সýÞ“;é&~\Á†ÅEíR6Á€ÃESs"( áK¢›®Õ‡*0ø¾+ ÐLÍD"áC»bhºßÀÐôÉ´B'©ØØ8ë$êKᵂ p9g‡N˜ýær"åYÜ)ŠâÄq=²qÝ·ìu¸zŽãô¥Va o ÷@ã—7£B’^5óþ¥¦2ø0ƒš6<á¨ÌSÈüýL7K| ÚÙÒU ¹Ò?Q‹34ÀÊ€>…m/Kyyñ¶ÛÑ›¥¨–ÈŸ"T²4¤›>‡1±úèûzSíit'²öÅN¤g#¦Ë9 ÓÖòú­lkþ~Cw#K³ÛâuV5bÌßìïà\V11b,ØñX©× ÕÌú´…ä¿O–âT±èøç\2 Xžÿ0„ ¼¥«Ñ¯+ÍìQ—ÁùüÂëUå(嘋æÛ‡K Üœò7PÇ\$‚ÝèšeÖK#Ç‹PúÝõš4ò¢wÞý¾¹ûÐUÂ7 endstream endobj 1307 0 obj << /Type /Page /Contents 1308 0 R /Resources 1306 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R /Annots [ 1304 0 R 1305 0 R ] >> endobj 1304 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 259.228 102.639 271.183] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) >> >> endobj 1305 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 259.228 171.007 271.183] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 1309 0 obj << /D [1307 0 R /XYZ 96.768 705.06 null] >> endobj 1303 0 obj << /D [1307 0 R /XYZ 101.643 262.715 null] >> endobj 1306 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1317 0 obj << /Length 1806 /Filter /FlateDecode >> stream xÚÍXKoã6¾ï¯H½E!¶VO[Z4 ´@hÑc€²9Ðms#K®$Çëb|çE=9è!Ñp8r¾ypè_Þ}øÆw©›®‚ÕÝÃöÎ÷7MÓ»U¼v˜yÈï³/ã v¾x±§Ê ~Lã– ‰l¯³g |û7zøã.ðÜ4ð鈴hiÚêˆ+Žªi53UïN]¶8hpílÁÁD©úñõ.O Ò¿NÜôŽôÏʪ•åü©uVíJóÎyÜ^Žz6<æ‡ÏÉÀô¥ïEîÊ[‘º«•>ìõ|y¡ÓfÕ©œ‰Óòø ÚÚ|C:pž‘r¾£ ³ÚZeó` S4¬¶¼ê¼W²þ,kÞÏ}83H¿ ˜yUj€# ¼¡lÔ†;­sÉÝ£ÿ*¹ÞÒTYÑ-'Û f´©ÀXʃ^[¢oÙǶAVŽ0ëû©ÃNüÎq N* Ÿîo`þʪ©Ý¦a/@bl(¼6á“DKC/Ý=ÿ¨áG¢Ú@ç÷¢ði&¬B7õýQ´þ.ÆËÅÆ?ÃØÀ´÷¸àmòfÑI—L-Ôºû½¥²njNV·gÄUÙ^L:mèŠ ÷¿¤¯ŠU£ÿ>Á ʨŽEiµ}mþ¦YÝs¹F1LI‹±ioEæ5lÇœUck˜Ux:‚ÚÖÆÛXeU]Ó5dÅIÜ:óf2 ½{g•“©£m ð˜|âB“59„ð[•ã¿÷<ŽXñõ¯}WeÓ}w•\UÍák¦ñE¤n’HÇçÇ Ê½ûíáÝ¿{h_ endstream endobj 1316 0 obj << /Type /Page /Contents 1317 0 R /Resources 1315 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R /Annots [ 1312 0 R 1313 0 R 1314 0 R ] >> endobj 1312 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 576.704 102.639 588.659] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) >> >> endobj 1313 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [165.14 576.704 171.007 588.659] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) >> >> endobj 1314 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 330.959 102.639 342.914] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) >> >> endobj 1318 0 obj << /D [1316 0 R /XYZ 96.768 705.06 null] >> endobj 1311 0 obj << /D [1316 0 R /XYZ 101.643 580.191 null] >> endobj 1319 0 obj << /D [1316 0 R /XYZ 97.768 471.192 null] >> endobj 1320 0 obj << /D [1316 0 R /XYZ 97.768 425.567 null] >> endobj 1321 0 obj << /D [1316 0 R /XYZ 97.768 393.834 null] >> endobj 1322 0 obj << /D [1316 0 R /XYZ 101.643 334.446 null] >> endobj 1315 0 obj << /Font << /F35 119 0 R /F8 112 0 R /F14 120 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1325 0 obj << /Length 966 /Filter /FlateDecode >> stream xÚ­W[o›0~ï¯HÓJm¥B±H–Nš´>ì9oŒ'8‰Wb2ì4›´?Û\ ¥ÚZŽ}®ßù|ø²¾zz™ù“Ð LÖ» |wÌ&¿p¡¬“ItOhòëÁñ¡ÿÉQÏ,OpþÝó=Š/ âè–$Jb^pƳ“Øâõ·‰7qpCß(Ü©ÍR¢¿å£²q½RO†ò4£{÷Bøa%÷H•æOª~zóšÿЛ¹ár& †vjK#D,g®À¦K7Úh–sµâ¬2[:ÒfbôËHºúÛŽmÃ$ižRïŠC±%(znÖ(={Kh]Û–®“ó)%[Äq"“« ¹t€¿ðïo3á›Už]Ô•!ª…Ñý†Ò3þlMQ+9BU 4cБbºç¹€B9W®Jéjå• ùi–çVk:Bv8eX»RB×ÓÏ­x6íédkj÷µ¬Ã~‘íT›è<‚²C%Ãdâ»ó9ìOà‘0&r£¬(±'bàÍÝÀ[ šÔ,ý&D4umE/«ÕLçy#Š3¾ÁP¾ß¦ˆ1éh.- .#Œ ª KOñ¦U­­X‚…bA㸠ƒ"+¥‘óQhÓx-ñB ž'(U? ³¶cTâíL“ép«ÖŒ›ÚX$ÂŒ‹–2yïƒË ÐN™  /é°^µ ÿÒd.Eh5k{j¡ÙnÚ‚Ì5÷]˜šÛq#Ï Ú¤ænÄ;B1kݘì„h;ñ£[D“|E56Ïñ©NÒˆÕeöf«k7LÙiÔáLVªÒx<3­t£ß<â,ÍU»+jL5¼³ “ÜÉIF™%•PL9tƒ’DÙ¢Ï^Ñ$ ñL“oZõ­.þ:̼@h57i·Ð ÁûsPq$ŠD>¹˜(ãxd ûfšn ûˆ©Ô`tBv綘ÞáYØŒëˆøVÃó±>ñup:¢Õ õ‹îŽ$wJáÁcÑæÌ-Šrô½<†ÿm6êG‚- ×+î0·T] Ç'î£Z[„®: ÑMÖ­ªSA/¦Äç‘+&¬ö'RÒÇ}æËÄ4Wk"NþãMWÿ‚ð…YE¶-Óš\¶éÉNB Á×õþ`—uÎ[ÌÄ$¹†îr¹Tû€ÊW_×WÇÖït endstream endobj 1324 0 obj << /Type /Page /Contents 1325 0 R /Resources 1323 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R >> endobj 1326 0 obj << /D [1324 0 R /XYZ 96.768 705.06 null] >> endobj 1323 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1330 0 obj << /Length 1473 /Filter /FlateDecode >> stream xÚ•WKÛ6¾çW¸NÈÀZõ²d´@H=ùæì–h›]™tEÚîö×wø½V릃45œùæÁoÈ_7>?‘tV„Eç³Í~F¢8,Šb–g«0N²Ù¦šmƒÃâyóûç'øÛ F³%!a‘yV+æ¤êâ<$a+´ŸÐC¢0&1J|R/–YœÔ {Þ(¿RÒº6 Àp¦i<À÷Nê{”ET¼š«PP39Q]íJõà¤*ª)Ùnµ<ëŸyõü IÿgíÄQXÄdhFÁ£iÎ+§éJë S¬¼¿ÌEg(vÓÛ‘—Gï\Ãܤu·/OæaÍeJÂõØqŸœuO®W$©u˜+'õ‡\Äëà¶X&q87ÑGæ& £µ›ÝŒœl^ÂÅ2M£ài±Ng´\Ä«œHHЪâšK;is¸œ˜0ûuO—·ð="©M¥ùcÒÓÔÂàÕgÍOþÏ™BîÝTî;¼®Ò°HÓa 3î¥+C7ÓÒÖLÊKÓt²Ši(Ž$#5I´ÒT³ƒlø?~·õmQÃ?îôxhѸõz^,a`&†ë<ø¶¡ð>ÂìºÈòÀTP«²Eû%e=Æã–ûtþb°F©Ñ)pFOQÒ&Æ­såFõÂÏd¤2À’4Øñ#ŽÔ ¥´¾]mþg¢dg×M.Š5 t¥pãuí¾Ê½F|;gÛý9±æÀÅaàd#Or …d&Bæ¼pº«™[©ù Öˆ}'IZQ³ðêl­˜Pp ¸šƒÿ¨p´9¯n èAHøOóÇNæÎŽ9¯˜næ~Ó‰+e [XÕG©<"ø(˜  9Ò e!SžË‹.¥­Œ$ ¤ÅZBÁ†L`vAç…ÛyôÄš÷D’p½ÊWµ 8[Be\i3¡/ʈS”æðS ͪ‡$…7+÷#D5´N¿GivöHvÌVØ¥·æV4ÈNYCÄ›|{Ä#F[ ÓÖ+¬Ëªj¯Éï% ¨c(<9_{q½˜q!þØlP>#D_½òi`FäÛ¾‡Â¿àmƒ§U Šž¿®3ø Ôž4H»5ð yx“Ú½FŒ‹ï„*[eþÆfO¥y+ýø6DóŠíY3oY˲Ø0&pµÀ#~ë‘ .µdòzÆ0WåC¯Mœ\tž~“ÍM^M¾ç%‡¸û 'ye’è^dèä Þy^àÖè{q­d¿‡ö“ÑëpÓÆ¤ËFI/Š Y™ ¸q»ëýtj){Å{][h‰Œ|¾c5Ƨm,&CIW¦Ø&܇xÈÝ]Îo¨ ×.ÔÝ8bã;»7ÀËîOVêîP´•¦$†úü¦3µ”óÞµYeÃl¿ŽB_¢‘7³àïbƒ†‡¾í­§þkäÆOŸÆ—m³Òyoüd•†q7ñ¸×ëµ3GòÈH~ømóá_îò¬÷ endstream endobj 1329 0 obj << /Type /Page /Contents 1330 0 R /Resources 1328 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R /Annots [ 1327 0 R ] >> endobj 1327 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 522.242 102.639 534.197] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addvar1) >> >> endobj 1331 0 obj << /D [1329 0 R /XYZ 96.768 705.06 null] >> endobj 1301 0 obj << /D [1329 0 R /XYZ 101.643 525.729 null] >> endobj 1328 0 obj << /Font << /F14 120 0 R /F35 119 0 R /F8 112 0 R /F36 121 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1337 0 obj << /Length 1376 /Filter /FlateDecode >> stream xÚ½XIÛ6¾Ï¯pÜõ`ƉÔ:¨H€HQ ‡øæú@I”ÍFU‘šè/)’ÚFö4Ѓ!‰|Ë÷ßFÜ^Ý}Fþ"qƒÅ6_¸ž"Ï]~ ÜÙf‹Ý ׇ´Àœï(Ý_¯}è¯ÞÓò½~Kÿt|g)²ty«–isRŸrÝU¿ëýö·»Ï®7TG ôÑÂÑâsM2±¶4k×±o€ä¬ÖJ”RRâÑŸ´ÔOAI)Îiu}àÑE­±\/+äTiñ¿®õ³$Ox·k!ì÷J’;â¢y”rP°Mq¡¾”¼Î9-MFrRïÔú~¬H°JüÜîΩ 'Sï6ÎEŸ¬¡bÜ@Ë=´Êô–;Vy˜qÚÔð2{v§þµti}‘™8I0'ÝôDÌ{Æ®…¥"\l–UMY½´ç:§þ FVÝ´>©Å¹ cµØÕZÒ”Û£9ƒ6ôŒË¥h®_s\6FÓo¬4,ÌĈoRI>°šJÞ¡•A§ÏõCåMý‘Èl0‚„‘Óš‹N©q1ç1+ Y”¡Q2á¿C2’þh§™Ë—‰íÜÆ~¢âØ{L¿ñ —æ(°DEïM>pç< H@X“p5.9+ ö´nªKÖxÆš¯¢¦©(:+†¦ÍY¦Â-c„ÿb5©±H[­I™‹e‹À7¾\Ðy;qg«–0>T¼Tdœ$4WĨ·/*y6#[>•¬ ò§n–¸¡æF\t–»ÏÞkŸaD xND›Ë›Í°|ìZ!¦þsÆ”“U-#Üõ¾yz¶`»šjRø™º0. _›$£Ýá‰Qàv¿ŽfŽÙDpš2Ãõ˹h‘ö¯Üûqdðq}Q>·ɾÿJn3k¢lf¦üHEE•ùKAIfàg²¼„~/–Tb™Öƒô!ùðR;rä5³{oâCÿ>K³LpúíMx3IË›äbÚªÍw¤ŽÎ•¹fœn¦=Û¸o#þf’ò—“ÖäÍ(×ú¤×™;“.È‘3]<¯ZÇÉÑòÖ½§³ NŠV_ŽSÁêÎæã£ª9UðM;Šº÷žY î£Wf[øàädœs`…f¶ìfƒ‘Á¬ÙÎ[2aã %œãšÚnÁY-dÒÌœjoq4±—7§vBÄUU¼t~¶MÐŽ.yS¦‚²R<÷S¤aΚª j”Ⱥý©&Cà$fÆ $D¡g'gq"õ¬q–ÑÎÈ bÂŽšê³—Qé»ñì4 x%ku!ˆÒTd=„¾«n!͉”×0Z ½ÂæføÐ~tX³´ócoþP£//<´ô)n¸šÝZ…ú¡Xבs„Ô,[÷BÊÔÇãµH|Tz]/Ë$ŵ»jH‡µ}ª4ƒA¨‡À³F@¥Nоº>œý3ÉA„ÂÉß Ãô‹|€\Ù¡$Œ"Mæ­¬«OÛ«ë¢ˆÜ endstream endobj 1336 0 obj << /Type /Page /Contents 1337 0 R /Resources 1335 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R /Annots [ 1332 0 R 1333 0 R 1334 0 R ] >> endobj 1332 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [196.522 235.318 202.389 247.273] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin22) >> >> endobj 1333 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 175.542 102.639 187.497] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin21) >> >> endobj 1334 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 175.542 176.237 187.497] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin23) >> >> endobj 1338 0 obj << /D [1336 0 R /XYZ 96.768 705.06 null] >> endobj 1339 0 obj << /D [1336 0 R /XYZ 123.92 238.804 null] >> endobj 1340 0 obj << /D [1336 0 R /XYZ 101.643 179.029 null] >> endobj 1335 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F36 121 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1344 0 obj << /Length 2064 /Filter /FlateDecode >> stream xÚµXKã6¾÷¯pz@ ´5|ˆ-v7˜,v€ä H9Læ –h[[r$zºûß§ŠEɲ­vÌìIT±TõU±^Ô·7oÞIµÈÓ\ ½xX-¸i.Z™TÀÆCµxŸÔ«»¥*ù•)¶uÍÚopUtëžœöþC(¾_røbÉyš«(¨÷í?½õmKü»¢y¡Õ§¢«‹Ç­ëéµnèY4·÷qÕ­ËmÑ÷ïëúC$Ý–Åv{;ª\Î霂¯êÕ Ÿ}±ßo_#¢°Ñ2~´ê«³¿fJvküTÖÿÓ”¯ê>m;×Õ%¾¾ÔMy„ýõ×—zƒ¢:÷ ÀÞΧόkÚhAT8˜Uvn7šüŬܽ¬QÖ¿–ôBéë¶A4ÏQÔºÛrÞ¼ãÙ4„¥Js›/É[ËI”ƒB•2mO5{·ÛŸ*öc  ƒßƒÎÁ’çéKØ6±ó3sÛÞ¡<°¨ÙãÇ,>òŒE³~Z¿îÂõŒÉìôpÕŒ&¡S®øU×)( Ÿubѳëa78ò5÷¡ÛO%w.%OÙyÞ2çë;&°Ð°ûbrE¹¡UÙ3³ˆÏ¿çó7ïì„/O GX Xâz;ã6ÎR“›Á±¾*/%ñøt§0Ë¡lßÖñix6Q6$P=(ò-)€"@„ÕÁ:GœóT›³„ }z·Ì2•ü\ Ô—ˆpSxZ=Ý!0ZoŠjØŽ`Ví¡µÈôT7kÚîK×ìfE^É’¶‰rúÃ#Ä…N~Ì/}€È5@Ìu8lÉÕp]wôŠ`rUÕûd—©>Ëßû¢n.Ã8!`½ùßÃÍï7˜{lÁFæv‘ (T6[”»›÷Ø¢‚MH’°õXwÈN·‹Ÿo~ºù{úI8pJ£@”Ež…3ÍKž §‚Œ²ú¨–,SWð}\䟅OBnh ¢4äRTÍ£{ Ô*ÍÌY€èÐ\¥‚ùE2H¬X? àËM8A:~ÅÒæ©àöKÀ“¹H-‹ºmô7©$„†¥™:ëx€Œ Jv¢YP–sA$±‚t?„$5*zØz˜x²&JUx¬IF'½óDzªCÍÒ¦Àlþ„çˆRSú¹Ž Ò¶'r»¢O±5ˆ{[®´DÑ÷æŒbî-5ü¨h*’‚;öž«aK¿!¦²EÅcô4À8@Ù›ii–È1²Áb$Åù=mpi„1”A™üØl_ˆ­ ½9 HF'm@AѬ}‹&l¶¡#”PºzÚB“p€ÃÍ¢!ÚÔ«hQËâЇrŽœD¾1cg7·|Üz.J¿Ò©Ûâ¶À4†¤âE˜©9Š8~eÎQÌDãÑmЕ:W¶]üo!®àI'‹Iý9ÎsÝØ¡N>¤ ·|Ü!Œ&džóª®Ie½ÔõO˜µ0¦ÏÃd†ú5M|Ò<—‹T0Š«žVáœ- úÌD” Jþw ñQ ´¡¦ÃkYP}ƒ% İèB>Er,# j1¬Ö®Áù:8TÒH9•ý«Is Xãx²Â±xƶÞÛb Þ¡TO'Ö`ëtðm·ÕXSN¼ÝSÑ8É›¨ç— ú8“ {<ÎÁà@Ùà 9|pÈõ$ †ãCZÛÐWúbWÕ0ÍÇý"¯œ‡ÝÎUpRFæ,u”HOXUÑÛãáJ5 èiû¾~ W>Lü|š¿ =æ<7ɶþ¨ÁáÀ’™»nø. Èè¦T¹“ë‘ y \åy— Wi“_î¬L"–ÎE€]üê±Ýç#½ÑÂ">ìp+fzÄÒ®æ{ÕiI%,4•áÊ4¶èäñêS#ÃÍ0$Œ3Ú¬ýÕ˜é†fÕ 7À±Slêû§"'q:zÿü}ÔàE$\²ðÖ¯+uˆ·—ý±5ä, ?— q …a^=C}ÑyøU=‹Ëq–+“Z Ïܦ°¾6Ί\¥P"`ýÊ<Ë \ij|ÁazW6¯,k<@ª¥Éo±ä õà9Þ/*îÌ!Âõ+8gѬÓ»Nöpoí.S›†e@ÇoçdIø&Ójœ1 ’±.öX|„< Îf»ç|^`™™Ô+yQŠã¥lI^Q7ãýsXC-Ȭ“R –øWÈæ:T¦³ôø­¥¶e®I^8;ó'nPK¼Dk¨R«®=N=ÿN—cæÂ¨ÌBi9#Ί–(ÚbÄAW ÿÇSá:ûËÂ9Ü"™‚ô6Í`~¥Öc"ýÅœ < endstream endobj 1343 0 obj << /Type /Page /Contents 1344 0 R /Resources 1342 0 R /MediaBox [0 0 612 792] /Parent 1310 0 R >> endobj 1345 0 obj << /D [1343 0 R /XYZ 96.768 705.06 null] >> endobj 1342 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1351 0 obj << /Length 1570 /Filter /FlateDecode >> stream xÚ¥XKÛ6¾çW›•‘µ¢—e«èhƒHQô´@޲EÛÜÚ”#QëÝþúÎp†z-½ Ѓ!j8ojüûý›÷Ÿ—“ÌÏÒ(Üï&ÙÂ_¤ËI:_øQ<ŸÜ“•÷e7Åqè部./§Eèåj-½gz‘Ú¼œ§3x0grK{µÎ+MËMOD ™±u|á ` =:K“Àûû ±O§à8 À’pþÂ0=¢"AFƪ)¦€æq:áJ曣¨‰d\q üWT¼Ê5s‚-‡ÏÌ#æ$J1À¢ ¥–'c©ÊÐæ^‰u@´1Ø5ÎCî8ŨÃ8‡[ù 7„&IBï7Ú®›Ít6O½±ÕDÙ–Í‘í `9æŽâBÄSY «:W 5cxykÃá%TʰF€;\¶Ãò"õA*&ó“yEEþ‰Z—,a¡Áú`oÁ†%vÀ©¾E‘oñ퀱ˆ3ï‹"nöj¶@ÏÍ2J¸äX² â—šÞveň鱗¶Ž1<ŽíiޤÂ`k°-÷MÙ ú8áÔþ <±,h«¬ QÝ2õP6ûƒ]3Cç¼ðY޽S[ @U%S7]ãèl™‚j¦Ø6†¨i=|ÿ9Lz= ý]øY“¿b‰ÓKì/s`4 ú$ª½˜AH9ô¥™¿ˆË-YÝ|¤®e¸3AŸ-"?J¢aè¥*žp ö×=±9¾ó@ê糸»K`:èHð(Ô^“$j}!–²É±#£Uñ•ûÁк³sø ÝË_&ñÈ9,q=©óóùøŒ¶ÖÌÊ@]ßÒ&¾„CÒ®Q[-K…Ož!×e¥‘Ô(ù­í¦ý9Ãå›{¢¦nÁÔÛQ ÁX›Þ[Õœ6Øâp·äøU%6ãSIOÔφ£ØÏB²û–6é^c< ¸0—š«¬Äy¾TÆçðOÆ©‚É¢¸ "§óœthú—¡Â.rý¨³ÓóÅÜâ6å8ö¢A@–‡a؈¾Ç¢¸æà A­‡¨Âw¬¸Kãx==@¯GöÁYÕ_á´uI„]èU—ÆaÕÁM}fú7¡]q6·±ñ0¬X¨;ëÅÜ_­ÖëQ^f([‹o¦Bñt±Ìl™m0¯|”0Èë®ÄîuƒŒfʵsi±†õ‹ºƒâ½Y¸¾f€ºîÕ†ÍGbÊôOÚ\¯]þCÒåym“;Àé$L¨ü$†IÓâQ0ðÁém]æŒÑäà{—²\ˆm0ß¹œÃ ñÓ`ñêí¼Œýt™Ž2ØYhã$G»èK\Iyà@6‹?‹Â¡’½_øQYı=<ÎnüÀ8cµŽË¸»JÔ‡ö«CnG=ŒÓ²Ék! w§¦Â’ÅÚ+s©Žó $éÎ-ز`Ë›ém¿²-—¨õÝ͹’eusµåÅèZ,Ïï^D~!ßÁ½w øÍº²_®ãËÝöV3½ îΩjYˆú†¿¤*¦«¶ùòɳßÒÞ²¢Ø;ÇW9Fb>j,”ùwË5”ëàÆ> endobj 1346 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 587.995 102.639 599.95] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin22) >> >> endobj 1347 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [170.371 587.995 176.237 599.95] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin24) >> >> endobj 1352 0 obj << /D [1350 0 R /XYZ 96.768 705.06 null] >> endobj 1341 0 obj << /D [1350 0 R /XYZ 101.643 591.482 null] >> endobj 1349 0 obj << /Font << /F8 112 0 R /F14 120 0 R /F36 121 0 R /F35 119 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1358 0 obj << /Length 1135 /Filter /FlateDecode >> stream xÚµWKoã6¾çW¸°Ñˆ«·¥¶AÑ.6‡"Ø“{rmQ2±2åJTýñ%9”,Ñ´ôq0H‘Ãù†3ß Ç¿.ï>>yá$EiìÇ“e>Ih'“8Z ?ˆ&Ël²šíæëåoŸ‚x  dM\à{RÄÁYF™Â#¥qŠ~ØIS­.2ÔõR`â,|ä‡þÄñ<”FÚ”û¹ùÑL Á„r=2uÆKÓvÿp#×ŶÄM³¢t «(û³­˜nÛ=϶ÓX“Ÿä…0>›žü)ßE©ïaŽ”máÐOŒ-+iÃ¥Vއò(gRJ릙žŒ¦ÝœÁPêy€ÀÈë ®Çb-ïVâf ØZ\ÍX³û¬" Èñ‰Ð@ñ¦Ô_äMXøó¿÷øèªÓ 6ø¡‹¢´'On¡ƒã{)Š#.FÄÚ²”SpNØ9çÌa59HqWÛÊêêU—~ÁðÈ7Bc]ò`Ónøñ@5æ_0ôËÁÉœŽ¶e “JE[§Ü¥C…kÒÅLÇôâ+>ÚŒ‚+®(ËÞü•´p½Û( ¶ã\›´¥¶DÞMÅD:Ûz–]‰­3Ê  K„ kHÙUbäEÞU¦D¢NÝŠ•éw0ú]¢h.ï0+ºŒÁ9ï¢Ó'H‚¾ÔˆÝ\!£«j’YJ„È0(—“¨« BXÌ¥šE‚B¡Ãȳ“Áà¤ÿ½÷<ü—–Ná°´¼/©ÎCã¹)J‚ÅÕØ$‰À‰/'qIXÁwºÎP $üð¾$Fÿ ÉniþòËךàòëmõ†C °ÂZÖ,¯Ç}÷&h&b÷m+fG˜H|ÍJ÷aš’xYº«ú•AàOú–UË)#6÷4¼ªqAоÊH÷`‚ѹɔÊÄÂá\lšU­xk¦# ØtâÚu(Âéž¼*ºüp¢O8>…¾¥{¼ö˜o7¸!ý U/¨ ? yÙš(NŒlíUZÔAiÖËêjÝŠå§Z>ÈnX¸T‰=7D±ßf¥{ÖkXû £÷nUº’a÷ˆž¬ÆbÅ׈O”Ñf7w‚ ˜µ9†³ÍÜOfGXSMcsEV)€9¯éÜ›mZ.‹²ZbÙH(˜©ÆÄzó¥a86ã½=læþŸ=¬¼Ü ¤t­—ΞžÓ[°¯ ×¢Yßhñ†ãš?«ƒ:¼¶eÜ%ÈmÕê^ÍÀT·šÃSù†NaÜ‚]Â<õ‡&4$LQ‚‘tNcâñù/¿??ÛN+š\= Ε{(¯û Ø«‰*ÂWy¯]¾ìhz¨ÅŸÁw<î °Ú£ ŒyUÃîVJ‰¼øaî„!üÍ‘r Ý‹†^ B¯‘ *‹T ¡ÿ:M$8·eIˆÂ4¸•%‘‘%ñ¹ž‘% Ü8\y˶œVLFíMÓ !t©O)»×{HÑcŠÚmÁW­‰æñ›%kÞÕKï6^‰þV'^ @΋C)y÷yy÷7V®k endstream endobj 1357 0 obj << /Type /Page /Contents 1358 0 R /Resources 1356 0 R /MediaBox [0 0 612 792] /Parent 1353 0 R /Annots [ 1348 0 R 1355 0 R ] >> endobj 1348 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 653.748 102.639 665.704] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin23) >> >> endobj 1355 0 obj << /Type /Annot /Subtype /Link /Border[0 0 1]/H/I/C[1 0 0] /Rect [96.772 267.198 102.639 279.153] /A << /S /GoTo /D (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) >> >> endobj 1359 0 obj << /D [1357 0 R /XYZ 96.768 705.06 null] >> endobj 1354 0 obj << /D [1357 0 R /XYZ 101.643 657.235 null] >> endobj 1302 0 obj << /D [1357 0 R /XYZ 101.643 270.685 null] >> endobj 1356 0 obj << /Font << /F14 120 0 R /F36 121 0 R /F35 119 0 R /F8 112 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1362 0 obj << /Length 451 /Filter /FlateDecode >> stream xÚuRMoÛ0 ½çW9É@ÌêÛÒ°] ´Šv‡ÀÛ%íAµåF¨c¶ÜØŸŸl9[“a‹à{||&y]¬®n™H4hIeRÔ Á´Ö‰9ЀU²Cëøƒí_ì:ÍècßzlKïºö |ÜÄ,QÕw?>Ûo7&éSqwuKø»V”pàœ'86©#åÌMF4AYh±Ø)3 sÏYùÌÒº2Þ@Ý›C0ƒúYíW{ôÖï»jØÿãí¢ê%R¨N‚gÆ%ÍQ DÐ$cad\FæÖÖ¶·mi‡X¢ÎE)EI¤îÈSðŒ1z€4㘢/ËûÙ¸&Ìî. ûñÙµ fÚjI.à÷åÝN¤Á[׆„ä Ý»×”*d·ïÒ,Dñ»”—¦)ÇÆL‹b¦îúŒ/§Ê½ýÃl6Ï•àZ],¦k'¶ï»&Íc(èäÇ” Ê…‰„¶gpìßÜ[*$2M„?žx ýt~#ïl •5“;Àr3ìýY…ËåRÎêÚuë{÷jþ]W@؉¸‰ºR}È1û•ã|{­ÈÜ'ü+Å“°q.PÅb!‘b‚W7Åê7ò]Óå endstream endobj 1361 0 obj << /Type /Page /Contents 1362 0 R /Resources 1360 0 R /MediaBox [0 0 612 792] /Parent 1353 0 R >> endobj 1363 0 obj << /D [1361 0 R /XYZ 96.768 705.06 null] >> endobj 1364 0 obj << /D [1361 0 R /XYZ 97.768 606.357 null] >> endobj 273 0 obj << /D [1361 0 R /XYZ 97.768 582.543 null] >> endobj 1360 0 obj << /Font << /F35 119 0 R /F14 120 0 R /F29 110 0 R /F8 112 0 R /F33 118 0 R >> /ProcSet [ /PDF /Text ] >> endobj 1365 0 obj [892.9 339.3 892.9 585.3 892.9 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 585.3 585.3 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 892.9 1138.9 1138.9 892.9 892.9 1138.9 1138.9 585.3 585.3 1138.9 1138.9 1138.9 892.9 1138.9 1138.9 708.3 708.3 1138.9 1138.9 1138.9 892.9 329.4 1138.9 769.8] endobj 1366 0 obj [649.4 739.7 677 684 700.6 827.6 533.6 588.2 758.1 480.3 1228 880.8 702.8 739.7 658.9 671.3 670.1 563.7 846.1] endobj 1367 0 obj [680.6 680.6] endobj 1368 0 obj [446.4 446.4 569.5 877 323.4 384.9 323.4 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 569.5 323.4 323.4 323.4 877] endobj 1369 0 obj [527.8 527.8 583.3 583.3 583.3 583.3 750 750 750 750 1044.4 1044.4 791.7 791.7 583.3 583.3 638.9 638.9 638.9 638.9 805.6 805.6 805.6 805.6 1277.8 1277.8 811.1 811.1 875 875 666.7 666.7 666.7 666.7 666.7 666.7 888.9 888.9 888.9 888.9 888.9 888.9 888.9 666.7 875 875 875 875 611.1 611.1 833.3 1111.1 472.2 555.6 1111.1 1511.1 1111.1 1511.1 1111.1 1511.1 1055.6 944.5 472.2 833.3 833.3 833.3 833.3 833.3 1444.5 1277.8 555.6] endobj 1370 0 obj [339.3 339.3 892.9 585.3 892.9 585.3 610.1 859.1 863.2 819.4 934.1 838.7 724.5 889.4 935.6 506.3 632 959.9 783.7 1089.4 904.9 868.9 727.3 899.7 860.6 701.5 674.8 778.2 674.6 1074.4 936.9 671.5 778.4 462.3 462.3 462.3 1138.9 1138.9 478.2 619.7 502.4 510.5 594.7 542 557.1 557.3 668.8 404.2 472.7 607.3 361.3 1013.7 706.2 563.9 588.9 523.6 530.4 539.2 431.6 675.4 571.4 826.4 647.8 579.4 545.8] endobj 1371 0 obj [565.6 517.7 444.4 405.9 437.5 496.5 469.4 353.9 576.2 583.3 602.6 494 437.5 570 517 571.4 437.2 540.3 595.8 625.7 651.4 622.5 466.3 591.4 828.1 517 362.8 654.2 1000 1000 1000 1000 277.8 277.8 500 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 777.8 500 777.8 500 530.9 750 758.5 714.7 827.9 738.2 643.1 786.3 831.3 439.6 554.5 849.3 680.6 970.1 803.5 762.8 642 790.6 759.3 613.2 584.4 682.8 583.3 944.4 828.5 580.6 682.6 388.9 388.9 388.9 1000 1000 416.7 528.6 429.2 432.8 520.5 465.6 489.6 477 576.2 344.5 411.8 520.6 298.4 878 600.2 484.7 503.1 446.4 451.2 468.8 361.1 572.5 484.7 715.9 571.5 490.3 465.1] endobj 1372 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 1373 0 obj [777.8 277.8 777.8 500 777.8 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 500 500 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 777.8 1000 1000 777.8 777.8 1000 1000 500 500 1000 1000 1000 777.8 1000 1000 611.1 611.1 1000 1000 1000 777.8 275 1000 666.7 666.7 888.9 888.9 0 0 555.6 555.6 666.7 500 722.2 722.2 777.8 777.8 611.1 798.5 656.8 526.5 771.4 527.8 718.7 594.9 844.5 544.5 677.8 762 689.7 1200.9 820.5 796.1 695.6 816.7 847.5 605.6 544.6 625.8 612.8 987.8 713.3 668.3 724.7 666.7 666.7 666.7 666.7 666.7 611.1 611.1 444.4 444.4 444.4 444.4 500 500 388.9 388.9 277.8 500 500 611.1 500] endobj 1374 0 obj [525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525 525] endobj 1375 0 obj [703.9 715.6 755 678.3 652.8 773.6 743.3 385.6 525 768.9 627.2 896.7 743.3 766.7 678.3 766.7 729.4 562.2 715.6 743.3 743.3 998.9 743.3 743.3 613.3 306.7 514.4 306.7 511.1 306.7 306.7 511.1 460 460 511.1 460 306.7 460 511.1 306.7 306.7 460 255.6 817.8 562.2 511.1 511.1 460 421.7 408.9 332.2 536.7 460 664.4 463.9 485.6] endobj 1376 0 obj [694.4 666.7 750 722.2 777.8 722.2 777.8 722.2 583.3 555.6 555.6 833.3 833.3 277.8 305.6 500 500 500 500 500 750 444.4 500 722.2 777.8 500 902.8 1013.9 777.8 277.8 277.8 500 833.3 500 833.3 777.8 277.8 388.9 388.9 500 777.8 277.8 333.3 277.8 500 500 500 500 500 500 500 500 500 500 500 277.8 277.8 277.8 777.8 472.2 472.2 777.8 750 708.3 722.2 763.9 680.6 652.8 784.7 750 361.1 513.9 777.8 625 916.7 750 777.8 680.6 777.8 736.1 555.6 722.2 750 750 1027.8 750 750 611.1 277.8 500 277.8 500 277.8 277.8 500 555.6 444.4 555.6 444.4 305.6 500 555.6 277.8 305.6 527.8 277.8 833.3 555.6 500 555.6 527.8 391.7 394.4 388.9 555.6 527.8 722.2 527.8 527.8 444.4 500 1000 500 500] endobj 1377 0 obj [670.8 638.9 638.9 958.3 958.3 319.4 351.4 575 575 575 575 575 869.4 511.1 597.2 830.6 894.4 575 1041.7 1169.4 894.4 319.4 350 602.8 958.3 575 958.3 894.4 319.4 447.2 447.2 575 894.4 319.4 383.3 319.4 575 575 575 575 575 575 575 575 575 575 575 319.4 319.4 350 894.4 543.1 543.1 894.4 869.4 818.1 830.6 881.9 755.6 723.6 904.2 900 436.1 594.4 901.4 691.7 1091.7 900 863.9 786.1 863.9 862.5 638.9 800 884.7 869.4 1188.9 869.4 869.4 702.8 319.4 602.8 319.4 575 319.4 319.4 559 638.9 511.1 638.9 527.1 351.4 575 638.9 319.4 351.4 606.9 319.4 958.3 638.9 575 638.9 606.9 473.6 453.6 447.2 638.9 606.9 830.6 606.9 606.9 511.1 575] endobj 1378 0 obj [625 625 937.5 937.5 312.5 343.7 562.5 562.5 562.5 562.5 562.5 849.5 500 574.1 812.5 875 562.5 1018.5 1143.5 875 312.5 342.6 581 937.5 562.5 937.5 875 312.5 437.5 437.5 562.5 875 312.5 375 312.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 562.5 312.5 312.5 342.6 875 531.2 531.2 875 849.5 799.8 812.5 862.3 738.4 707.2 884.3 879.6 419 581 880.8 675.9 1067.1 879.6 844.9 768.5 844.9 839.1 625 782.4 864.6 849.5 1162 849.5 849.5 687.5 312.5 581 312.5 562.5 312.5 312.5 546.9 625 500 625 513.3 343.7 562.5 625 312.5 343.7 593.7 312.5 937.5 625 562.5 625 593.7 459.5 443.8 437.5 625 593.7 812.5 593.7 593.7] endobj 1379 0 obj [272 326.4 272 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 489.6 272 272 272 761.6 462.4 462.4 761.6 734 693.4 707.2 747.8 666.2 639 768.3 734 353.2 503 761.2 611.8 897.2 734 761.6 666.2 761.6 720.6 544 707.2 734 734 1006 734 734 598.4 272 489.6 272 489.6 272 272 489.6 544 435.2 544 435.2 299.2 489.6 544 272 299.2 516.8 272 816 544 489.6 544 516.8 380.8 386.2 380.8 544 516.8 707.2 516.8 516.8] endobj 1380 0 obj [602.1 726.3 693.3 327.6 471.5 719.4 576 850 693.3 719.8 628.2 719.8 680.5 510.9 667.6 693.3 693.3 954.5 693.3 693.3 563.1 249.6 458.6 249.6 458.6 249.6 249.6 458.6 510.9 406.4 510.9 406.4 275.8 458.6 510.9 249.6 275.8 484.7 249.6 772.1 510.9 458.6 510.9 484.7 354.1 359.4 354.1 510.9 484.7] endobj 1381 0 obj << /Length1 2136 /Length2 16464 /Length3 0 /Length 17744 /Filter /FlateDecode >> stream xÚŒ÷p¤[Û ÇÎÄV‡Û¶í 'èØ¶mÛ˜8™˜Û˜ØæDódï{Þïÿ«Î©®ê~®u󺱞jrbE:!c[C ¸­=#7@DNX“‰ÀÈÈBÏÈÈ GN®jîdüÏ9¹:ÐÁÑÜÖ†û  ÓÇ™¨Ó‡¢œ­ @ÚÙ ÀÄ`bçfâàfd032rýGÑÖ jàbn £HÛÚáÈElíÜÌMÍœ>âüç@iD`âââ ýÛ d t072°È8™­?"XTlÌNîÿã‚’×ÌÉÉŽ›ÁÕÕ•ÞÀÚ‘ÞÖÁ”ŸŠàjîdP:\€Æ€¿(ä ¬ÿ¦FGP53wü—@ÅÖÄÉÕÀø8°27Ú8~˜8ÛÑ*R²; Í¿”eÿ¥@ øwqLôLÿu÷oë¿™Ûüml`ddkmg`ãnnc 01·ÄeéÜœh6Æ)X9Ú~ظ˜[~(üº@\H `ðÁðßüÌíœéÍ­þâÈð—›2‹Ù‹ØZ[mœáþÊOÔÜhôQww†7×ÒÆÖÕÆó?ÈÄÜÆØä/ÆÎv j6æöÎ@)Ñë|Áýsf t°122r°s€ö ›‘Ã_TÝí€ ™þ:þààíigk0ù ô67~üÀy:¸NÎ@oÏ?ÿ‹à˜˜ÆæFNC ©¹ Ü?Þ?Ž&ÿÂýw0wh3~Œ€ñ¯ÏŸt?&ÌØÖÆÊýõ¿[Ì *"®.,CóoÊÿ Ûº<騨tÌlŒ&&vVÇǃ÷ÿúùoþÃþïSEóg÷‡G)[׿H|Tï?D\þ=”ÿ^*ÀÿF·ý˜g €òŸñ×adc4úøbúÿ¼›üÿ›ý¿¼ü¿ŽÿÿÍHÜÙÊêo9å¿þäÖæVîÿÖø˜gg§Ý³ýØ›ÿ«ªü×B ÛZÿ_™”“ÁdžÙ˜Zý·ŒæŽâæn@cEs'#³ ѺðáÜÊܨhëhþ×… cbdü?²3²ü¸T?zõ·ø±RÿRÌÆÈÖø¯Ýcþ軃ƒ;ÜGë?À“écInÏ6€ÞÆÖéÃðAÎ`bë÷WGÙÙ Bý qDþA\±ÿ"Fƒø?ˆÀ õú°“û/âüÐTü}hªüƒX?Æ÷¿ˆëÃÎàÄ `0ü}D7ú/bý°û¸h¬ÿÑþ«Z Æ@&ð¿ð£ ÿjÔ˜?ù¨Ÿ£ÙFÌ“à‡ “?à_Bó\²ü]þˆñ—ÜÖÙáw*¦À¤ÿ ÆúQj3w;³‹ö3ó?àGÖÀòXþ?êcõü(õ¹”ãÏl¦6ô‡üƒœí?É|ÛþøƒŒÝ?âDì>¶Óö3}°ùƒ+ÓGêŽÿÄû ]þàÆö¡îøqµýcðáóŸv|\ NfÀ?ÊùAÀÉÕöƒÎÀú.À®ôêÃú`ÌîÝÿ¡óaêtø—ïÿÙ#g‡wËß7ÜÇpüÿý"Ý€FpË ¶FK­¥ã_ò“<ß0{›Ê‰Ž†õÒ éçíÝjîÔ;‘t< œ÷Y4K±§ÖsÌã¼Çz¥*³c7Ž6ø-êø …§ðQŠ4Ö¢gé·¨M¾ÅŽðØá• þ„yuEl‡ 2:¾L?Ù.4²î\W£L+u£)yÅg¢zÏ`òR “ÄjÒ‘Rýl|sßÇWR•MPcš=„Ýgçj]Ô^ ±ØÜLч“Ö„w|h;Ù>ì/P 3²É& ¤+ò ˜³*ùÞbÊ#ŽÑ EžÚïùôßîé,À-B½¾9Yëlp÷ûqb)N>*6%^üz-îŠ3ÑnüŽ~‹8¬ßɶ;íÄüÆ’ØÄqD+ÅŒ%üéWE6ç°tÚ&ã<ºµ#¾˜Æ´Ò“±æ1þÌ€À§¹×ìÎ_å«Jç¨g¬¿€Ê‘’Ñðºsôév¾úÓøYþ{œç9ܶÇÜÍj¸ÎE¢‰ZÚÖéz Ãsk¾ ¬©0V7¬Ã ó˜ûN²¼>DÁ•®²\¦¨Š0=öfÛJ~“y˜Ý.Õ3TÁºŽµöÊ$¯Ó5CÁ¤/qÀón‚æäÇWH-_EÉËxñLÞ{1邺ç ¼­zyŸÓ2?õ¹ïö0²Ò4ú30s ;pNo_\d– ˆ~uœaÒe/•ìcxÕ¤„‚qO¹ : õ7>Sœˆ‹6sIª·? ºXFŒ8µÐ‹á{yTti˜gòäÈ×"(È1p>á\ÂÌ —I ­pt.ð“Täfc9¥Œ¸òPÏì‡Y±ÁwÁ‹‹# kì=1«¿ €1S7žáAýœøŠÅÉ­׿m¥V Ý$Up# íZ†»3nK¡uøµg¦,·sã;‚£±$Pü[êgÌjǃ#ÄÙiUc®ŸÂ[E‘ÈÁ+Ú1ãñd)¦uLí/&HÐ:GÉЌǰ¿ ”§ôP(3^d±,PHLj­]7˜ÛIÛmƶɼ"ßgTQÉB'%êN{Vh–|}b —HvEkªÜ[õR|`>1]kpýmhÙ-þ=Þ_¾²ÄÄ‚'ÆX€4Ί„†ý¬‚ ×ì Öóñ…úqãüW*Î <­ŠÒÞ'æoíÏ•ŸKmªpt¾Öé*Er}ÚYoíe·Â±a4ä ±Öí¸¼¢=eX)-ÖùBÆŽ«Mã¯&A:Ø%—¯pF‰8UG/ú²-[³SŽ­sK½-øàÁò¹äÅt2û· KæÔ5¨ˆ.rHÊä› K¿¯uØ¢›o uÃ`É–”¨# 1±–iÈv5Ÿœ¾Þé ÷=…bïÌ×êHúûœ|AÓ¿+KhxÀyF3"ïFò¡D'^“vYtü„ä6Äe¿àz0½å¹›Ç‚Z=é8™ˆ"”û@¡D×gpø9œ""]ޢޣ¥%ÅIkóBÇÀ׋-Y®¤…¤>€º³©‹àlå„÷{Ôd²hLT=m»ìo 𤸳ÑÎ]c w§¹#Ù$nŒ½÷f¤à›ÎÚ‚+ÞÛGË$ p#ÆRñbÔ°´‹®݃ú'·VŒ2Ø‹K±ž/V5ß¼Céé€Òð‡õC·"qÉÐ¾ßøˆÊdv1E-lj¤HÜÑ5CñÖD›¼(:òÃún-ù̳;z—ÏAÐÑܾG¶:÷ à} ™©MЦ_¨·:f{t'¿ì !ú¶#JÎ! òY• ^L5z¨ˆH6ƒ×ÊF—²Cd§Œ; t逭éNkæ [™þýa™ û…jÂaù’óóEÿæ‡dn^ûWÐþèßÏýMT(pm;M³1õ¹# 5`ÅàZ>™"‡©p‘jôˆb‘"íÛ,¹Þ ïî”þ’1´øJâG‘m·ñ ·j›^{ÒFRUó6zäóE†+ó}ê@™¯+#9ùýÈ˯xïÞǦ<“]ñuH•‰‚E?¢ø$kü*{NŸ/ÊV­m¸Æ™ÒŸèBï÷óž1Ê´‘À@+ ö!:YpÀð È­»ïx| : Ýкl¯#dfYk›ÖT ¹…œÅ A–qqA Õ5H./…«äRÂÅ æÑþκxoí°á¤"Í>Ù4ÔZÎ~罘P£4Ý¿]zÜ ,úªÀœÙúZ„½êaÈɸ ?vK‡­ªþݳ£öA_ò".R/T»ˆf%<´³ƒ˜CðÂ% Øë~ Ãlåçïâæ–°(%NÓe|ftŸÙüU†Cz+ £`„Ù\ÃãÐyFuºd«^Ãa‰Nâ#Šé›ìÀYjƒÀ*Eê]ðÒVÎØ®ågó—ŠF *Ïvh¾EpC6«6s'ÞµTS®æ•‚P56PM¼E_·=ñ¿Š¡¬FÐ0vT“zà»Ð}šKG–óšƒÌˆ7–~ÐùR®ü©BÂÝ©™Õ˜yS>v¿Å“ÓêQ' k9\Í©—þzq)tÁÇ7õàp PßÓó½£Þ·v»¥ÞWâ_óé Ž~'~|ËënÊ1äô›nÑD\H—+ûþ£!ª u©dÕÎ.f+-’jÈᢠ!Ïgoéw|ïbl4/¯1mÞ{ñ¾I6­b·Ýd©×eµe‡2gGã !¿9|TXS- ×TŸ•-?[€/ØþÝ õ(ûk¶°xhç­,ùOk5ü¬Ž#´a _5Ùw"[·mòmïou²Ní힨×à45Ö·Èèº%(‹Öß OAõñP¦í¡D¦©›¶kˆÄäj¸èÝfáÅeík”?³Ì·G­‹¹õ]µ%´€í†­ô ØÚÓa “ú:1Nc¼ÃÁ <­Ñ$ÊøUÖë½Õ¹™TÕweè™b1:ÃL=s3 ¶IÅÓh î» áËTÂã—:‚+¦þÇQSÒÖ1{û夠NV½…™Ñ¤®ÑÜ&|ƒ.„Ý¢–îiϪ‘’LÔ -âý¯t+V˱Ôíö ™Ø9{øËmKùN²" ¦³pÜ6†Ò°aO²mýö±¸IC‡lzý\àåòÊêõÒH‡U@Œ­ë8›:ò³í©ïh_“‡|SW!°X["wyíI4œ„uxò…J u_ ! ënñL°G\Ýݽ$Ñó=OBÔ|0y­ºÆ Þä® Ó@:ýPÎW[—¶ôg:úŸ'çùÅ'#ûS#\”BG,rßPpÁžª³Ù1ÈsŒüÀI‹†|ãàS#Æ Á`³áñ¡Ÿ –U˜2®§€¡€ƒ^qxgDG=sìã“`÷×8Ú¤Æ@X‘ËâŸåÁ„¯´x¯$4ËÕ2–шJlXáY£Råc/b!W†ªTçH5Hy$Ñ!õü®Çë.ÏõÛ^dXô¬'‰Ô‚sŽ«pæaSv †;:°Ñ\Ÿ„*ú¶´x]hÊG¡jdEöÚ“'iJ V9ÁbÑSðÝ#éÝDf—s)ºÔJ“½—u¤öqȲGŸF±¤Mñ²»·¹VE°qËF Д™×¸(=ô)glÑšÊ_VŸ~”zBs}­æÜÂ@ŠÜé_G¥ ÷ÒŸ£©ÌKp18ÙRÙOåáP#·äß•EþzîdYä +Ö»Ôè§&©Œ88A"ž~òKù{¨×šûÔ…³Ï¾ŠY…e— «:ô\Q(Æ1ªŒFÛd6çUi’o/ Z è3Êü—½¥xº_òË”ÞiÞðó‰ô¿ã5Ì+0 Šfƒðø@]Kð+p}Ƴ´ÀY+ççòÈün<ŸFàJ|²Œò$¦*hcÆÉ…­›è8¦0sT2˜/^C¨¯Ä’†lí yºÂæÛ§PbÚØO´Kd‘Αm[zðšU®×`ÊòÅÓêøù:¯–8/›"Èârr÷#)H‰fl’œœµæ”&ÆŸÅÔ†ˆñ±Ætˆ6âÑ\ÅÞ?TÖ,Ÿ±·Uh,FÇ )qÂ"þªçE w³¢ðÇ+[¸©B>÷6ÛáÛo(Ø$aÖiÕ¥Ýó«tR–~²?¤ˆøÂñ²½²ôµ¾ÖNE€¼ÅÜk,òÑÉ DæÓ÷CŽï1k–ËqdÚ¦7`¹i¦Ü•™uy1¸Û˜Áð\ÔG›ªæÚ”µ$Ÿåèu]‚„ÁN>°—)™`™§ªÃ)ñ4•4 Ûîèáý±×ÛU.…ĶxªÒ¯‘Þ°,uׂXk7F7²è1àîÍÃ}Áê'Õû+I`õòV&mj7°{]ÍÚv6ýf™¹Ä,Õ›Ê/{÷Ð#q‹‡¥öþâv5Hü‘6϶÷r ¹,gïHBÉ}Tu…ãšWlõbç­Û>\ÃG¤¡àÑú Îo¾Cà.½1ùÂ#ÿû7žM´®Øæf²½+½I°<~Ú/YôâÏkIE ¹ä I}Xmè´ ç˜j“*•åŒH'éxQi*U¦:»-½Ïé§ë34fœô¡èÄ#% ñï¬Þ4ã#7ùiÆ"ã ù"š–Ÿ¹¼°FLUôˆI«ûNjsuYó€…9Ä}r#ŽžMœñ4¸éÍž®É¦§WÆ[ph|;!ͬŸM€šƒNZ­jV— ôÝ2ßÃÒÞ“*Ñâl9ÚÍäûÎ~'Pƒvl ‚E†¨:*QvŠšïÏ(ï(Ëv» 2y/KôÛ9‚.”}ÁTßå NöÀ¶X¬¦e°ÀGŠªA†˜EójÒõ*<ˆ¬Òý>ªç}OšÓ$0‰ïÎhP‡¦D(W(­ç3‹Òš×zÝî:°¨Ìî¦ÖMS†)íÙÚ”!颿t…£òD€ÀkŽ®m1œdÔÿªá"S ‹3¼Z妩9H &eÌ Ó}·†2¿ _*m´Ã=B£BØØE H=²3(Ù„ 9vÔ¥H­YÝÑͲüý«W¥AöÂýlCÿÍâa,My±‰÷bp=Þ:ãÀÐ\у•i ݳæN°»!SH‹×k‹Î¦"‰H*“‹„¥M°X‰}pc4€‡yÍÛiÜJ„ ÑW·¹3(´üù˜ŸSÐÝ]IåÔyå"iJÓã-Çúx™Åmñ&Ü\†8ºÊÑ9r:Á8²~™Ï[¤VSà·UG`ò¸º¯†±ûú*›4R\ýÌ‘åÚ)ãé̼J[»™ÊºLë^4U ¤¡ßâS9nxr£U‹á¶ã%Þ÷7OÐSn)φ¯»·ùŸu iñÕ_6LKÀù¹Êî§'Èk&ÄÜ…‹®[6ŸˆVxZ“Ç‚BO2°¦1½òñö!'©¬ uœÈTý±ªÅ°F ݨë¥iÊ*­S2£ܧ–¬…¿Ñ×T`R¼—yXxSVOFyOR*jz`“‘±¶v7È ï&$l‚B‘°W†öè' ºøw Ôê ±„rUN+ ´*—¦ú0úºyŒÁÕ't£šóõG4^ 0ù&ú ºbñ«›ö«Y1¦q%Ö°óºr6…ï©0.”ßEënVd +P|#w.–•«xV%2 ¤ûÔU‘¶ßÜõŽV«Þ›M ­}¶2¾Öi1GmO‡yEBN{î I¥*)?!ØÛSn ­f¹§_7ßÙŠN|U-—‰…Xû-ê¡l°v34G øR÷Z¸˜„¼x>ê6˜%"Øþ•(g7É«Í8%Pú âÓR=¹‰¥ .SÖ.°|0˜*‰~GÇœÁJ2€ î")·ÿ¤Wk¸pù½ssy–z÷ÅçíÌáþlùÊ«š›»YR>NX¡GÍãÓÈš¤á¥D¨ä§š_dìd¬N²È5Ò† pRçòRØ" Yöô2é(´ƒUü~" îpVÉ0Ÿs¬Má)lG™¶Ë¸ƒÌßfƒËPö ê­{ÖòS~ ¹é`‚L<aáç3Œso-ÊüŸXxŠh,Ý ü&á{0_Ëo¬]ŸÝÁ—*붺t½eaìm·¥ Òo–/¶pX÷ÖÕE„…ýËdŸÝ&˜õË*q(u˜Gwµö:µèçÖà$î×wðüO€¸ÙQ ÎMMœÇˆH,[v)/œ(¢ö´‰ƒëHÚ,r7ÛŽPýN hSå=-ª<‡ç ÏMH"<Iº¢¢µÎŒë`Ð…óEJ/Ú|~o£ÅŠùÔ2"jö –y I»¤S臖­ßá©ä`¯~=Kà+~ÒB$"QþÂqY¢ñÖ¾6 2w²0øý4ü"}—‘Fp²~^u]^uHà+ªhõÜß ]ºÂÍݧ•gØÙ®±À ±M€íâ;Î;™6•㫼1o6iLÍôÖ±Z…ÜOJÎiDÂtC±.¥SIÔ}(d'ú^öCܳ…ü>5ôñ%„¡\êý|·]ß©\óŸ¯Í :F¶Sá*Ü㤢þ÷|‚2wBû̹¨ÍßörŽª>™ÊHZ+,§U”ç~â‚QŒËëÄ…ny&²íJžKY šV@7_”*íçÌQVS)uÇd>Ù¥±‰&NL$¯¾Ù¸ŽÔŠÍ¤ØFÑ䵟#òŽ}}uQn8ŠLïŠú:ÏQ|Z<@—n"ÿÞ‘Vf®ˆøèŠÛÞ™óÀ–®!ed|}1|.ʘkqy#³ûJá Ù^ Wêb†‘ÄUõ ¯ãSÝ0D9‚‰]Н/µ#òø tCÎ¯Ž‚¬NÜyj#ûO ÞxKÁ>}M¨º4Í2Šfœ´.Æ¢$ÕØé¹oi(áÝÖI`’˜ªûQÑh>Ü¡ŸUÆW¹ó• ŠÖv²Ôl”ÙÔ¼óÔò­˜ËIdá+" Å]¬uÁÇ¾4Ö®ècF¿$Ïà&_0]”9‡@2¢{ÁØÝ <–À°&u§§=cÕò…:΀0º¸`‡ñ¯‡Q{X@u ² ' ”¨%S{fù-ѹoðÑéÁWÉFM2#*4¢:|n™ËÆôUúõæ¬ mè«&fD1¡ÜĹöL>Z¤óS‡5ù—i•(†ÿ"Ã3‚ºtgðÖåh/™œÖé:¯XŒÄè!_1÷䊃,>NÝÐ#iNKë’'/kóÛן)ž^fB=¥‚ŽnùËjåK|Œre’(A@˜¸ªÎë:£q>Ãð.|.ˆMí/KÉ3î­ezKo¹k ö¯ žëGFÑ[bý(îÆ¦þýŠ+“*ZðãhRVBëYZ^³«¼î®®_Xb¼“?¹zð„PIuÄN7XfhSܧÉóZÿXêtç ±ˆëCï>ÂQÝòÕk‘Ó^„çzŽîìq|÷¶¨§q¬ä㌒)ÍŸËÚq”l‡¥8o¤ì×Ö`ŒBff®]œW0¤=‹ã ^+1 ƒ@-#ú.ðp\ Ìj~2†éÅx¯ƒ¡D6yüÖ]ÓÆþ”é~S(TÙDù{ð½M‰Ehóª°ÿÛqðƒ“£¤½ÇŠ×£!û`T¦®xÅÏö)bž—HMŠô{mG{U(K»ÉkÈ }ÌE^Y"'ô!hQ­e¼ ;Çü¢†Ö ýžß”52ŽUÌ€ò;çRäH;2-áÅQê·=‹%‚Óü/D~Û”øiKFÌ/÷[i =8µ§^Sgš¤v †X—…&ø°ÌtWlž¢–hÃaŸ®‡ ù›|DögüºÎ×ß"|hža?qÎ åNÓ“}Å(J„Z1:@"tDTω*ùÅ—ÞáBî2rúT‚ïÚ£'˜å‚dRÓuŽž—››tZroµ€;À˜òT(¬'åéâ‚nní3²`’ŠÄµ}‰AÒÅ0fI9ð¹®Àª`?ÙsôÓ.E+B½W$rµœ¼:¤ÁÃIJœÿ½ ÀŠ$D VY˜—Óì'9¶œ0qÌûV0¡©9u†@ýl¸!Nà'rá·fšøµ–ŠÝîä³€R¾¦7!,ƒXèo‚ŒKÆÂ¦®Žûõ*Ö)"Ý,R¹LéDu\ÏYFn¤Å¯ª¡‘¤ëÑ*f¶C¡±&¦ß~ ÷‰¶Øž’\çÒ“…âÕÙè§]ÓÓŽIŽk ÷ѧNœ†Oé!€Óûo°ù§Í0ˆ«cÞ'eê^Ǭ ´Q,\PCC ZÙ ¯¿³”©ê~½ùr[\{` öyn¶:׬Né—g~M7¨5êú€LÍ5ß-ŠÓ[£=Ǫ̈„?>l_vøÖÎʬ$Šu¿jݦƒ¼ ÍÜTgEo}Ãcõ.¿¬mürk‡ LÅçâ—ö¯m¦ÐË9WIF¦a.üʻyxŽñ2Q‡'‡0 ˆ{Üymž§÷Vʇ1 âše™)bAø9]®­p…ÇèÕ„Ý ±ˆÜÛ¢3²>Uåd¨Í4u€ÇÁrü®>٨ƈš¿q¥ ¼Ÿ0`z fÓIbÔ”9õð4hÛëÅ$½E[ñ®IH¹•muésÓýľËñžmu¾ÿ96¢^@®§Ð ;Gñ¨¦Já.7åi:½Øq.HÊZgxl¦âueOêÍ {®¼ 6%Š÷1ö2Þ¡‡i»Ã>š‰Ì:Gý .§;‡b¿åF|†“Ë–=øg¸üÂIÈNøbxr¼-Q6Oõåqš8…ôÕâ¹NHŠÎ Ÿ'FXFZL*ip¶²Œ]‚ŠWjusz¿±¤‚‡JAG®´Õž!H|ÿvl£tR iob­a[ÏØ][¶’“þÔê3ß–žV0£ÛÖ¹íûOŽÂÌQ46B2}"®0nPïü–E39ù«údLkþXø¨\w®Í±¬-jeSñ³\jsÍ+L(³<âógNÀ4ƒø'Zš-=GDîẑ¯Êèc_ãZGKHµtq=WOè-c#)l …µ“ŽWOɾÜd}Ÿ&«[(Ó~‘$Èê) ›À³«œÊ¸Q£¯Wº÷·äßX=Ž"ÉHÀu¿+JaŽù¬4 êrºŒë«kÜfŽÐiþÉ’‰òÉ &ET”Çt?S´‹eOd÷F‡ÌÕ­Àü-2ü²+¸Î…Óñä+HãRÂ$èh²éðÏž\®„wó¼-ÁýI“Ž:Æf` =<ƒ·=4‘I…{+•ö‹ €ír “Ó{ÿ£@º %¤éÑ|ë ãÆp’ü‹ÙªÖ–š.Ú!iïíí¯uAÆøÐ¼ïNXODœ=Õ\âb|XáõMÐâÇy¸_ k‰¿  Ø êµ6Žož¿€†þÜͪÎyô)fq±8ó7T³Âm6u|D‡U1rÒ&ô}·]¹" 2:íåwŽãêòéç‘Sè…ëvãÇn¬;Ü0Â_ËâYqm/ŒÑ‚Ò1ª$Cbà ±Fß=HŸŸIŽòr¡ÖUBfº)Ä£D}W–3å c˜ÀÔκTÌ:6–—€€“cümyfÕ0ûF…o¡B˜]¿&ó•&rŸŠŒe|­ÇÀP_ ¿šË§ Ê ÛÙ(—“êàr€•° GÐy;$Óˆ¡}ÕTÏ'ØìŠAg!MXÉU°ÐêAïE€ ‘Vœ³ILŠE‡Ûš?¬óf¢wÆ0{?æWiÍ[Þ Á¹}ÈÅÊC¨…MIïIõdºŠ0¥cZY]Rëv†¿“oÚÌÙb£NÓ`®?]ÎöG&÷žna†…–ÖŒÚZb#,ü$ûLÊÉûE´5üË-¾XïB$XbyÇXÛ¾ä`MJ¡ây/¤>oo^9;ÔV¹¿t¦Ž³‘ÄžÐrâ¯á!ð¨ŽÞ8=ÿÓO«û%£e¡“†1Ë@2BÁU¾t(ŒŠ ú'£ñk¤û…ª„:7 -¯‰.UuBŸ´We$¬ÍPè®!pzМ²ýøÓìVŸ˜ßâ¸/©ÉŠ!v”“» ÍJ;§‘êÁU¤V™Ðï‹CNY™¥Ý®Y}º ¬ŸîkXä®:¡^1Eì1‹åy_lâ @5Ñ~³ÚiU–„0_4Þöl;=y)~ëàØ™ÉŒÕ~·n¾KÛÚî…!Ö“qÖ=©¨£NYoÏ“ gW«ŒŸ¦£õ=Û-UN ßìxÔW^LªÊ|s'šµ–üîíh³NNbP¿çŒœxö¡ Èä”ì$Ÿ¬½õ þ.Oíõ€:8ŽY…ÂîJüòCO³ðj™œuåŒË“-J÷ö’C<ñ9ÒZˆ‰œȲP±×ÚPIe0Õ¨×EÛnGpyv”¬ûÎ[Éù”«ebßA8fÈdÐ>¶ËÜDþJ¦¾³”Ù7Fn“Í Ç~ó½ ª1-Rî”(0ŸÇÞç^u?¡7ä€Ú™µ¸62匴zÑ€²œpöŽqZ,ó»¼ûP\;ø<ž)2ùõ–;Œ‹ßƒâô0f0Íø[‡«ÈàŒºy³‹pgxÂÀNÆêÃMFÄ;KòJY¦õ«Vs+heùö+ÆoŸÆRLŠ÷tÞÓs–Š­¬"ÔÑJ³&¦¾»ŸöÅšÙ`_>ד‡ j¶’„éDH'ÏJ¨r"â'Øm"œBtP¨%7jÌ™´v¤ÞõÆ'­0X™‚è ¾¨#‡{wbrš®3àâ Ä#Œ ˆ¤¯±˜¥‡:û×*è§N½øú!d" 9¢b—Ž u ŸÇUv¿Náf @'Îñœ7ýPj*ékê%Ï\)¥]™{ÍÁjÒÃkéÃKò{àõÏ"³ë¢¢‘MŒTV‡X¡FŸŒ%I~~‘ÂjÔlØÕß!Tä.Vl9Y/ÛÂ?¥6÷õµ¬uZM‡WÌÒ rMÓ}åäê]Qx‚L€ @ˆÛXe#š þ\ö¼Yñ™²(ºuY@¶Èîo¿ƒÁØÐ¢ê¹¸XlW÷4Ô•ÁõUÃgÓºD¶;•`ÙbÌ­[©oiž¹·¡*ˆþr¡æ4±·§œFåRC}1õKÁ47mrœ„/#o“’´ÈãÒÈ ~F ²Ìº†+Sz¸«…«—¶ 5:ýæ„€Õ÷~U;ï%ÃIo¦ô“5–|&5#¿äû—>Õqm¡#ÀЬBeÖp‡v¼ûlÏõ¼ØÔR=ï·¥$Y?A¥¸£…~A…¯§×iþj÷tyMl6ÍJ„!S@+pÁ ÏTRª6^  Ñ QdÃYF˜•¬’Ÿ3äïW¢Á¼+£: 8õ¶ˆ N¹qöñ¿\+š_×¢¾‘(;T#PED\X»z=Ið2"$½¾ÏÀàvä¤Rh·Õkn‹Àžf$ÒØùf8LóòŒJ«'ñntV7+â1ƒýl e§Žáé_-[{龿~Ú{Áúqv¹@É„sOQáŠè^†G s^ ¾±a‰vÏ´è`»Œ”kÝl ‹* Û~¨¸(ÙÄW.oM{*§è!X³ì«×‚¼y~|)SmeHJ …±´Ö é‹jΊƒšóð†¿ÔuYŽ…ZÓ2 ƒ‘ç¢zö^kË%ã8 ÚS81êr…ÍwÎnlGÜ.BPÏ­ ¥>KlW.«Ônr>ê)á|´‘«²FŇÈX ÀüQNÅ,ŠÏèB¡ Çi»tòYŠÒ¡ûû• xܼuûk¯àê‹ÆTj¹jK5#cR¢ŸelÐÌ47šóå³Y2P~N üÒ"Ä&Ì3‘¢"1K’Áûýcì{à áÙôºl!¨y@œéîLpíɇséK&N¸Z”¾è=JÌ[ƒ$Ž·êUGŽFðlèP]ôP([RÏÙòlÍÒæ{ íAÙ~ß»‘ÂñøøÔÏ9pÃQ«¼l9Pàk?êOKø²\CË cÒO£©¹VPl ý"„¾ˆ*mü̾]Á?AºüÅ<$Ùò¢ ¾’ ¥å½ñ<ʉÿð¬û;~+$wDOQF¹gcä¯m7 `™ÕÒíkòèµôúšIkôIpÐg-\ÛSùIÏ;ƒB )î¾W?÷¼‹£ÿõöª~5§“dðÇwüöû3+JË9ÃB’@Õ¶+iܦ JA(÷ÙXS¨^¾ë ²7ôzÓ©a[\Ý?@Ÿ‡o ÄßB$ Ãü³‚ §M*IÎíì–ùC¦‡`±†R}uJ¤Ê°¢“ªA»Øv†~5ÊýY¥³±ª ã­CºÑ´;ë7E>/­ï«/—ŠÙ!Â#*l¥„/툵£²âÈ¡»ÓÄl>„Œþ8¸®ð†nÛCÂüÁgv—n‘³´ÂÔvÞéÈð©f&£ªhÙ—ùT¶Á”°eÉ#Žén²¨ºX“eÝ3þSl]KºC˜ ¯T[9J ZÝÜÏc9!Dd g6óÎnÛÎ)iÏßœzrîÌóëâo\Ù~EŒŽ*&ÜéK5imñ-Ìð/A"yó”òw¿Kú·Ø^X~NY:ÝpÏÐ ùìËíªÿ²ûɰƹ›Î86ÑÖÎÛF>ÇÚMÕny’Qࣇª=Vh×Åc¹–$çÙÕQdl½yI?£Rªq\”yúî¶•-3þ1Ò[#´ålµrq¹œŸ ‹o1NmÕy«¥Då;¨mñ-ÄwšG úݬfWÿ!¸”HÃô#jl÷´>5è6Q§<‡`=ym¶–^ð¸ùãÛÂÇAÜÙ¢:þJÌ™k\9ê }:Îl;D[°œÛT‰^FF+^¶âù¿Fr·d/Û¢Ótv\]ÅÒÈ-”’b3c÷úŠ„DÏãódø@Žê|Õ¾¸Ç—û,=ùzëxÈÙâm#¥#N=ü]Dó¬ÞÆuùùë—û]º®ºã­Ùlýz×2XÊmÖŒ]É6e†‡ÅžÊè£ ²=3+4Í)˜>ìÔ Øôw?_ËæA¼‚:m?1»ºrš†_ÎW¢†ÅD1aï]b¬êYý, Xí#MØ­­k$¸Knð»\7kð]\’ ÔiF.¤S0Âñ»46‚î´áC´%²âéœ;†$´gn½}r¶òb® ›¥ÖCÚ¬'à¿Æ7^’‡“Äh;¿LŠ"ž>áÙ;·®™Y7€½‹èBÎà,²æyvÒ(0z¿›³dzqûË%h*Åéã—ÑàEŽßÃCÍW~™‰Ezè<Sd§ÀÄz¶¦ÐE*ª»^¾ìÄhDç}èÀäì·§|¸³÷`Ô ñìYWÛ£Ug²kâÀp´ÇÑíGÎWÔ‰'¿¢Ù™hìÿ Øé¥¦e¬œòOsùΚçÇšƒìk+Ö9ÍÖš¼¤(TTO/ö¹¿[p„ýµuEÁÍzîdÖÊä ömN|;-¥„å+äy í¹/õ0ÑûýÈM_è‰ú Â3nÔ‘gyÍü˜¤<ô5Ü9É BŽÍšB#*‡ÈIªçÝÇú[}5ÑÐ{kq9ÌOfK¤˜`®@Ï É#g–N^¹­}҉Ƹ­i<4ÅìÛÛJØöĨ¡@NbÉ2D6¬“yæ¶Ïk– Ðjɽ_ì¤ é-ß‹ãM¾èÀ³ÑgSæo‘àfeì 4óÙÈ(ûz%\]Rò^o¿Œx³·_< º;X‚¬‘d¬Øu€ Êt5Tܻ׌Õùz}ûÒoú¼¿ýæ{‡NÅSk« I+õ­Ìu¿¼Åj$Cñ£ƒM¦±o†‰ §¼Ã߃«ƒí´§2dÞÌÎãNüZ~ÿÊô®šƒ0~á¨sQÈéSBȂ鱹¿Ccd Üä9Ê<ó¤¦ûÒ_ L˜€]—Ÿ·²·®1ö·=Aýí£³¶1Ôfø¹e€¿zæX$:D>%ÉÅ Lúà0¿¢j ÄB…ZƒwÞļX”­íuQo¶Rx–vT§‡~ã!!’ÔúLÕµNèÎÕç››<0tž„_!T¿QÿB)5Eã,ðª,páïç·c¢¶ÙÈ EXмÍÕ¬ó–›5ÞQ£W\lå¼ #¢™<¼šÙœ¤¸í¿°L#šM£~ÅHHHcxt{üµŠ½Ÿ=xk%OÀ´®uVdâ*UJGû0«ó­óAE#Oð»Ç+,w¬n^ö)´,+'ƒÉÀv«ÖϘvdª‚pD Ê¹JsËÏæ–FeTIq~ (ÑvºƒçÊÇlâ¡a¥háІ ï'sgÞz¬n¼a-3üU'•+ÄUO¼Ñ.—±‹Â#~-¬ô½UDÆ=ê×§çÓ^묈¹Ž;{Ùù¼gÓŠÕbïs}U(ª‚gìÔ ]ÍþúÃŽUw¬ ¸jdѫǻÁdÞ0གྷ‹7“J`¶ÌO—¹•ÔºlÞ¹Ch+f ÜRDy-½ác¢‚½ç¤‚7äD}?n}îÉTÍF7Þô{Øëö~Ü[°e_±Ï ÕW£ØîÖI®0ÒC•Ž$Z=º<îÑ©7ç5Å%í¬ìã) ÏÎŒÕͺ¹Ðt¼ç’ªo²Ó¶3|ø<Ý\\ü‹"³h iŽ¿ ÷ÖfñæšâÖHoL{|2Lv{)¶ÌKt§Ÿ…& õíR§½VÇŒÛËd~}_ò–2,KöF5b¦_ö•ŸpÝá5ý‰Î¼þ„.‰ùIú+ú×ez×rMd3¯T.ë±¢ÒBÿiE7ÖWø±±1™öûÃá»Ø…¼ªû;…,m_S‚ÕÓ4×Lž1®8/a”›ó“©W4Hõ{¼×Xªrl8Çy‰÷À¼ËºŒ#kÊËöQJ8L&§Ø¬gb¬“øÁ¥ŸýnÜJcâÀñ¼±¶7¶ZŽmâ¾#Nóg}<žR#"ðˆXfrF3ÛçöÀ­’H]ÚØŠk[1z›L(o?މ*åG½Wc?«±¿%8PBÉøß®¦ŒòÔ5f:TA°dz åð—Çîi‘Y ¸¼Ÿœ|E¿‚pÝ:D<^v>éå°X.ëá%¼q HSÒuŒŒ‡ãž"Rìþ}Kæ×Ô¡Ž>ïs²äwÙ.ˆ•·Ž'Õ¢®®ÞЖœ'¤«þ4LšçQ!¬¶¦ÃóN®”mš'm}—ü• tíÍ̾g•Û’¹!Ë9¨[YK•)Ì\n§ëÆ~ã%]¸ !Tm“)²Z®˜Âô»vÆ…ŸËêáÞu…A(9Aw;pèu\$ųI@{cÉ+ Õ/;!â½s'9ÆÖö¶e@ÿÙ4Ò¸,‰‡Í§?ÓÆ…©…ä]Ôo1ÓêÈXO¬ ‡Ÿø’R½¢í†oáXU§¾Äbqc‹YÉó·6-lë®=($¤%øÈ2Ô„"‚ø¥ ]…ž¢8®Ÿ£¦¥,0)7Ž‹§AgqJ{™ÿ u&rl„S&ïñDßzwÓ3¹‡B…½‘2m¿×ªÚÞAc Þ, Kºw~=`%oé¼0»w«2’Hœî‰K^[i9»àëeƹê÷ÂH½ô˜Ÿ£›¥¯Ë´gçá3Ͱ…?ò7^==]mš´£'R>d(J$춨âB%MTMUK§!5È&+ŸRó´sý%ÏCÕ!–ý˜æ÷´ÛjÍ…>µù–·4Óo‰V­¼Q(²ö¶?·¼Æ7ÂõQG}`vI¿è`'ó! ¼óã÷/ pp'¬Ëµ6Q¹Ñ0uø Ž„Pô†8ÑÞWÐy±ãG½<­úú¥ê‰uYõø•ÜÉAÞ[=°q}¸ÉˆU„}¶Á—ÔçGyLº7Ýpñëþý1ZÌAŠ’ΛÏ0‡™/7^¿›3Ϧ‰Ú„ç¿É¡ÅW(3¬0½dV‚‘(敉cµÙVŠ%Pk|ÂqÏ 3mº1ð@> ü${£®'³h¿pƒh? ïç–Ųô à¸ë¥ªX#ÙCÙ’‰!Å}?¿Ù˜Äý /=ŒM¼;ášoÍÈ_µ@ræ®p½7îÑá e ¶ÊªS‹Âuý’Ô8âÀ\$XG ‹Ó+[P ¯Å€Š,ç7I…tV|T¼KÎû<iÕ줉vhÒŒ§7YZ†F`îõŸy€ºøç%2HR«×/o3ÊÏlÅô!PæÅ{À6‹@ˆCjvœÈ“! ðè®´½@òÑ·1šP>;Áœ yóÙ'(p±˜fÎåüUXšpô¯Y¶b<¹DF×sW"G‘8}bš˜t°‡e°.Eî§$œÏ-Øíô¶$=|5N¼ËËzqU¯×C¡ßyÚ@Øð…%QƒénR‚ߪT1'j­'ôÞ)ˆ-w·K«jA·ŽÆlp/ÇðO’|V7~ͤ¼åÑÚnì­ èY¼Ø"|äÏlÞÐî&ì&îùºZ»‚HY4ySsþñðæ?YÝ&~'D4QqÄõ±AʹG²ÛÈ Ù›èXÇý¸Ž‘÷>¤ÔÁ±¡o'5|B…8m3rçÙã›lYcµ­{SÀ¢,¬aÏÆ¤fU±1R­¥Ê_­ŸÛõtTø¼7‡ƒ©€ypOT‚’wêÜ«ÝyÐ2{¨çª] 6ÒØ:¯VͳjC4 Ê„­_td±!y‰¹õ¹X“C‹ðic[¶>n0£†Í$Òu òPB>á!=n¸YösܶޤBH›Î"cŽý…Òçe¡0E8œ¿¬îœzç™5º³Rjí߯UqÒ͇Mùºˆ}žW-3ÕŒbœ!.­cÞ É›ã^4\sûæÆQå@5X;qÕÒ5Üx…k¸9Ú#¬çÒšw¥¦ª´!ÁŒ`6ä3%!ɸ2³ÛU˜éN¬zÿ?Ì3üÑýd0ôq‚‰ƒÛ€t›j¤^(—reæ½ Àì_CÜ’³›kÁ¨Û!ð³/púEóWø#½Œ¬Ê¾Råûýüh‡]¯p¬´]{,ÜŠ"9P *nná9›ÂÛ¤Êc,FÄ|/¨ç’²+õwnõºñZ/’ê@€leÀ½ Æ2E2í £ŒA÷úl*LówMUk¡&ÝѯþÃ8xÕW`nÃ0±=¬ÀÚá–à £³jg62/› …_ýT×^@ié ‚!‘ÁçC—òJ&Ófp¸šZ…†! ^×/èL>¿²·ÝnÚ6 GxÓ¶ÁE†QµyJ‹õ}Ÿ^7ˆé]O‰ëšõµ·,ñ8Í=ôp1UJ*Ò‹ çˆtÔ&ú y¦[:ÃQй3W¡fdºa•Pâ‰Å?4mZ±Õ®íâðÝt³wÔ\:!iÍ78D*9Ó'[—‡0‰-úÅ®h?Ù~p Ù¸?ø³Õœ“¨ÍPüÚÚx•â­ì÷!ÙK˜lÊ×ñm»:‘ ü]5;î…Âu}ÄìB¿CçéëÏŸ‡ýŠ'Äã.³S½Ž %q÷ãï±y¸|? (~Rþ1Ën,]lv+yÁ¤ ê·‹ÙpJu€ Ú`Á‰/šªýk©Á[¹].!GÝØáúèzÖy½èÖŒ3â-¥l7kú:MR¯jºCd?¦¤Ü;üi¥"?Iÿ‰ûƒ{ýi/'&'¥•2}Kô *ný…5 G\)ƒÇÝpWDcÒôÖNЬµÜ›ƒ@ ú¡Do¯_fcÉ÷ª„«/Ò–€"ÕoÂ@åב™¤Ry׎u*jÙ·€ÞÞÚ‚‡xT¿åãlRæ¦;E¨kVe‚]ùM—@~‚G Ü…O!d¥¥S¿ 7ŽR1aù 0+õE è)z|ãîçláí9°$²®qÎ —l+RÒ°ÕûA<• Ûk»µÌþW5ø«‘0ñÛâäÆZ•ºOÿ¬oŸÅáM2݉BìuœBèzÒ$ 8MÞ‡Í%oÔCE'ì:!”ú0N!€Ù6O£ÕÞP½Û í3dâ6ì*KäÞT ŒG„4NuÜ·ÉÝ šì¥“]ýõÁV¦> þþ‰%CŸè¦?ãÜŠjrÊÝçNf ŸGK½Ìz~¬¼øb™—‘Á6¾Ð@ALu®\3VåwØ£‹âôðÍÄ1møD†ާ@¹wvvÄT7éÂSµ¹rÞŒ¾q¬,Iœ|L«dÜ€î (3áèÓv)7Š\ endstream endobj 1382 0 obj << /Type /FontDescriptor /FontName /TCFVBK+CMBX10 /Flags 4 /FontBBox [-56 -250 1164 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 114 /XHeight 444 /CharSet (/A/C/E/F/I/M/P/S/T/a/b/c/comma/d/e/eight/endash/f/ff/fi/five/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/period/r/s/seven/six/t/three/two/u/v/w/x/y/zero) /FontFile 1381 0 R >> endobj 1383 0 obj << /Length1 2081 /Length2 14139 /Length3 0 /Length 15395 /Filter /FlateDecode >> stream xÚ÷Pœ[Ó ãnÁÁ!¸»[p îÃà îîÁ-8w î‚»C‚»;„`—s^Éy¿ÿ¯º·¦Š™Õ²v¯ÞÝÏSP“«¨3‰™Ù›‚¤íÁ.Ll̬ü Eqm6v++3++;5µ†•‹-è?v$jM“³•=˜ÿN —7›¤‰Ë[ ¢= çj `ã°qó³ñð³²ØYYùþhïÄ4q³2(2äìÁ g$j {O'+ K—·sþó@¤°ñññ0þ³9YMÀEKÝÛ‰@[€º=Ð äâù?t‚–..ü,,îîîÌ&vÎÌöNÂôŒw+K€Èää2ü% dbú·4f$j€†¥•ó¿êöæ.î&N À›ÁÖ ;¿¥¸‚Í@N€·Óê² eø_Á ÿ `ü»96f¶ÿÒý;û/"+ðßÉ&@ ½ƒ ØÓ l0·²”¥˜]<\&`³¿MlíßòMÜL¬lMLßþ.Ý -¦ 0ySøo}Î@'+gfg+Û¿4²üEóÖf)°™„½ìâŒôW}’VN à[ß=Yþ}¹6`{w°÷¹ØÌü/f®,ÁVŽ® YÉǼ™þØ,@..VVVn>ÈòZ²üu€†§èo'Û_æ7 ¾Þöó7 _+sÐÛ’·³‰àâä òõþ§ãÀÌ è0YX‘þ°¿™AæÿÂo÷ïdåÐc}?6ë_Ÿÿþ2x›03{°­çŸð¿¯˜EQQJ^Kƒáß’ÿë·÷x3qq˜Ø¹Øll|.V€ïÿòü·ÿQÿ·UÅÄêßÕ±þa”›Ûøþ%â­{ÿâöïÉ û÷ÚÐþ÷%û·yèþŒ¿>++ðíÛÿç%ø;åÿßìÿÅòÿ:þÿ·"iW[Û¿ýtÿ øÿñ›ØYÙzþ;âmž]]ÞvCÑþmCÀÿ7T ô¯…··5û¿>Y“· [Øþ·VÎÒV 3+ å¿†è?·ðFnk©Ø;[ýõÀ0±±²þßÛÎmÞ*Îowõ· ô¶Rÿ{¤hoö×î±sqLœœL<‘Xߌ‹ àÍö¶¤f ¿gÀ ¶wyK¼‰ó˜Û;!ýu£Ü\±¿LÿB<‰?ˆÀ"õ_Äà `‘þƒØ,2€Eözã”ÿƒÞ8ÿ‹xßXTþ v‹ÚôÆ¢þqX4þ‹øÞXLþ ·Ê€ÿEõÅìð­6Ðá[[XþueÞŽ5ÿÿBVâ9þ‚nÿ øËoïêôü·‹À·J-ÿ 9ßÔ[z:X¾=OÿD¼Ù¬þßÄØüòXlÿßÄÙýloâþPq½¥‚ß&ãþ7µöNK¶ÿ÷[õÜoÍpxÛ:û´‹í­üˆc{«ÕùÏy!Û?Äp½…;¿=²þ$¼qþiîÛŠ³¸X:þÑ¿7.îöÿHxÓàúø&ßí·ñþvö7>Ï?õ¿ÅzœþEö?‹turz{Iüý¨zÛ’ÿà¿ßH ˆ´¼`±® éøõUŒÈiwBh–zW+žÉ{Ù©Óõ7|2}MVкÓXòpÆÏm)º[ѲgïãÖøð¶$ÕöGŸ'£µéÝv¤¥)ÜÉÂc±úï$ˆÄL¢{>ÏŽ>š6ЭÝrÔ¹Ž®¼h*ùX¿Üûe<ê¿—ÿ [ØUÝ«á–G~*ŸaŠù­X2Ggš=OçÂD‚ðóÂ}îön3gò•L.É÷$†£È[wƒ=öaÞkµRƒÝ¹‡€Š@ŸúslšÆ[ü EoÑ»´8N.,ªØœÄp¡Y¸ÝÄ‘³–¼{Y°|trsÿÐ(P¢}{:,¨ºE->|h«žÓ…™[ÏS2óÌfT€ä¬ÚµÿŒëU5—å,2`ãrªÂžþ-Ñ–4ä–N®ë ü e‘{‰ à;Ÿß mCâ«¿h»iž‹^¯·‰êiϯKßÉI0½ Sa)âaJ“6Kò8ãÓ/ZCÕ˜¯6ÂDe1ÓPòÖcµërÝ^h¯ß”¾ƒEV ÷DQ 65v]C–ÌÊõ@Ó±Öür½`˜QùÇHJvõü¯zŸ¿g‹›bŠŒU*BÙ‡ÔñÔ”š˜¦DäÓ¶„[ܔܫðO4L—Í|B>4«7Kô™–-Ÿ: ý`èþI½ºbD))yΖ]Á¡•TÐÁ<6uÝ-nj˽X=G"˜¦A 8|BÑ(â„OGP©4ˆó%e+˜2S¸Y׺÷Î2¸‘!íã‹{Íëå6ÒSUüη©-ýM-%f‰":Íýs¥½=âÜξŽwh°sb"W¶FνeŠa*Í÷åÈ]±žæêž¥› ݦïÆ¿ó¹’8ÄÜ©l ÅUruZ Wrµ,‹š’œXÎkëU™‹2,ñÅon!ƒíé™CÚa*ðúø~ü)Gå¾ßÝö‡"w¼`¯]U~¨ùaøv2´…ÍqæÏÓ8@¶rÜ7üÔrBºŒ LA]ÁHî¥ .ùsõ³ ^Oïd:óX!é{üt*“"ùk°‡ù¸â¨ey<²šóϰ »¹Z¯­™j¨Y|ý°ºº™$v>@<ûðÚøÀá Y}»0%ˆ0?púç£ÝS”»\V}‚gs”L'˜¥˜äŽ@åþW?„•ÕñÓØÌªtHhȺڕ=²ÀÔ^óÚ`Ô'–" Vj°dÚà3A)/dðÈoµ:}ôÒ¶ ¬|˜þå™îŒ'*‡ ¬$ò) ’Ÿ×|õÑj!Ý+ß²¬Ÿñ Ç^ P|Mx½ôDB8»F¦®÷ëÛ=œÅØTËŸ5%u™jRz«Óç´VäSkPÎò[šòü4\ï‹.”ëÙp*ɉýL2ix8 ƒ÷a£Ît‡gjc-×”Æîq|çø:J0ÈkóÙòÄöζg_ Œ¿n’±ÛÔ|§Ô3ˆ:×Í‘W¢—N5žcÆ?eÄYH·~–xTdß1ãrªVˆ»æ/Àß©/ú½QÕ5î·ñÌÅ\T4F,݈À"7ç3V9ªñE.,»«†]ÚKÃJnJƒßEàLê áõA}D`Ã?…} °yG€"(ÌøÁ¯Ú°bò£pß;Ì5?z>¼™&l£†É¹p¸Tö_ñvV{ÙlNµÉe'7ß…u+ÎÆÖD&cÇy·Tt!íâá(í@’*7Ó¥Ÿ¸Ûˆ>?ð» YJõiÓ xƇ)ýJƒ§ã#,ÙÇ3ãZÌ_Óæ,‡mçÂ"»@Ñ–È„û*E*w[ÑÕŒ¡¿4g ‡ß–BÚPù´lû¾enŒˆ_‡¸‘I˜Øa˜ç.¹E¡í±apI÷Ë „_ý ííÃme$f»Y:[á8?Kào€9#ö RçOߘ<4Õ † x>Mo@W²jŒí±LucÓ*¹¤§#Ãæ^Å%êŒ\C²¿Ø1,âžÛ½5‹ÓÔüJV°.‰ =çWóem£O.~®½85]*AË\‰5÷´N²ËGN1¬Ïî–¦Ýþ8¬¤|C‚õЋìN¼î“Õp~Q‘Ì>AëIårzc…ö;x©±h=¢1!¶–šÉÂî5&ÙY&•ƒÌ¯h;Zöõ3›K˜³$X]‰?æß“œtÒ”ïþ`›Ftg¢`ଧZ+ži-L¢¬‰Ê¿×‚ꊨq1f’ëü<ˆlJ2Q^p­kwQfž_×/ì¡tZ–¿æ°ç ]œÝ*£°¬\d)|3ks3{œ[šMر"'p!Ç o‡4AF,GðÈ?ÒgC€»`ÈG§„0–œéÕë¼Xr»{‹¡…ÖªÝ|3ÇÐU $ŠÖfÞGú8‹-¦ž˜Ë0ÜT±‘VÅ”$©ƒ ¿oûQÆø½{f@ò¥Qßê‚çAnÜ*kÄq4ÍN _0ïÌýäù^ýÍ÷.ÂJF“Ûݤo“Ê®.ÌMÃKW•ÈPa~ŸUéHÍ1ãŽüÇ{žaWåŽ !Sÿq¿³'£U9Äg‰ YA®Ë"Xd /ç’†S_WãÅ¥€ò. ž/ò]³ÈÐg÷ûíW;a«¼žÊ€(ï««øBŸ\7czÒ>fÉ0bb޼å½dyªdì‚Äô#‡]FÊ̪ó·lI¼öYµ×ˆ¤¢> Ô¦€1`-|ûIB0uà“§¤¾bÃ;u¾Kª1óÙNæ72%†Oú\\áÉ”ÛËn8z¡:“ ô8û¸ò;êE†Ä ~àÖô*EBíGF^šÔ3ª—ÔÕddC-ÌÀ%ÏÉnÿî~ü„ëJÎUyoúe¡qˆP-É3áv{BÐÑŠn¦Ãr2¡U]T%QáÕ4ΧòŽÒ–‚»[>MLE{tÝË­8G7 8sïûxJO§°$o®Ì޾Z¸µ Sškø_ÃÔzítòω¾rEQ«hõŒÖ°±Ån£Ý¨ûp:‹²Æ7Äô™þ¡ûѵ©Ç•Û¶Áìv‡³ñ°Å« ¨àž„ûKCN¶Òm‚ð)t/ZD+BéW”"Îû6Ô¯J°Vçg æ×æÒs,%"ÈcŸí1&ÆŒLÆÉs¢n‰¹ÜzéwâeÂV®}VcIwâ,<>8ÆUü]š‚;‘¹(íî?tâ¢5¸òoÐíQ'u‚Œ]\l*vªR‰7 FŸPH¢e‚&úŽ4 ÄyWõóžCÆ·H’`Ÿ6c%/ÜC9d( 7“»@›Ë”Ù°ðSúW—»¢²F|¤³\Ú¨*Q[ÃïÞ`oxÄÇÀNµ®æöjM‚„j%çõ™ØÇ+tÿ.jñàËè«TÄ‹Œ›×Â9{á„×ö‡ò©+²'kÐæÐs-ìâ¥ÇyhåjéšÎÛI3§¯ª€l*²«ñ›÷̶¹ªsÖÃÒÔÖ9¥* Hlgí^¦ìÃÕµZkŠ vòÍ!­ƒæŒ?Y8ü úkïRxòYxz«„Fí™öÕ¨"÷6À)ÄrâMÄ¢à=ÈWn1ÖÃ>Ël5i¾¯9 (Þñtù4iTë¡ñþü£7–Ý„÷¯·qНÖ=ûð¥9(KD'ÔñÒPxš!¯+Ÿd¶†ÜšƒàDŒÕù7Žùå@jcMÃè¿ü·ø2+ ¢sGWuîõÀC©MU7ƒkP±Çس]› Á¼GT¿»))@SB˜ó6$È$—ækæ!y)qÑ9£ÈÕ˜eA (*îDR|%·“TJí!+‚ùÏ¥¯Ÿ(¬µaÛµ¡Ÿ—ޤjN²çž(ÈOÞ›ã3d’C~‹H˜BaáÀìÙöàI¹ä¦$ê~5!¿Žú´M9„õøŒ<œ<ìZ„ešßÂ’´.Ó]ÿ“ðAå^¿Lߨts{žêÎI¨‹~\$ž¥˜ðž/V0ÓKü#ÆÑñ2O5s.WÙ6¾xÓ¶š, ½¾ˆÏ„_“k›†Äã›Å^Éq¥cšM²PJØI4@—­#,¦¦cõƒÞyuÆŽ´ t›º›èLÕa“ûþŒ'å³YÆ+TßèyµóȺ_¦>4è!â׬]ìù ÿî• {[¡È4ÖÏŸÞ¡¹»Û29 _Dëøòl+w6Ùã=Xî’i›ÑŒ4o¿‘{ב÷³ë¸hLRиõ,–IäìÖ@€éúNÆWŠÔÄå~Óe±ÉÆ•öÜ=ã¹&Õ3Bú(-dÍá[;)vkR¿¯ìûµ1Þ[òœó*æsN#zðñoÕoZ¿Ø9¹ÝfªÜÄR¢ 'æ"˜x\µoÝ·*|kéÏAö¬+O þz3É…>­&Êâm~¬*qé[a6Òƒ0´xç·´íTH^ÍÙþœ3Ûm®R`ß|Э8Ë`H˘†vÆmÄòCËÛri|ø|˜%b:ãhÌz¿^ÝwæåT.Ü yÚ9,r¾i'rsÁ±t­ÞÏ+SÁÛQLBmÆ’Š>ŠßKI]NaÕ“ØÚðKÿçÁD’FÈøbRú)¿)y]h1œß*¬°qApr*¸~tÂ~|[4ߊÖÜz· _)‚a»è;õ¹ö‰„,¡•f‚L”0–êʨcØ`·ÜàØ~Z¸‰ìpr×ó~¸KÀ=&¹nTºyîµÎßRæ$)íÄ^î2]}Ý-ltÉ“¾ò¼Ú«ß½½¤9÷¾vWÕcДKÌwlAbhÝE>ß¶3®[+‚¥ÌeQìKÕdHj>ÂT5è ÕðöeüR Å+-x͖Ljä'2è/'/ÿާ$ ‡y¾O»(Æ|gÓã-¨2Ô—*²KÅbš¿M Tk ÛwüÞ¾ÊÈãÃÆæ$û©„îj¤‘}”ÁÖ„BFsöT¨dHSâ*ësûÂðºÀZÖHÌ o†§QUÑÇ]ýÑýV™BjñÑŸdêâêXH:î›àY.ú´E¨+6Ä»ígäô[]oï ˜o7kdÌ!ø(§¼£ƒ´Mºˆ ½ãF&Ôf¨~íþñ¦ÛòZsÇs½Â® ïò£F”²ÃòïeS 3+Ä}([ᦧýªMàvAX,‹­xî´6´Wè=Ó^.r—l––°Ý)n ë04/ZÄ¿O“Ñ,RŸJ ¸Ò…XC­‡[‘·WòÄ¿eJ!ÇYÇ;B²’ùW°¤·I™¾µ@3¸Ãl$‡v|\^ %…Ö!iÅd–ôî§‘Ä;ƒ:[ ÈÕݶ=ÒÞKåŒ8wY{E|HX´ŒÂÁ¶ÇÄºŽŽ¢_q#ð±%¢¾ž1¢q¬ø\}$8¨Óc,óþÕ2¶»ç}cTOù6ܶÙz=t·i?¥*·Cã³häÉ™Y]é+‹â1Ì/nçç®y’ábm§ÁŠî»Ú‚YÈ>èݿòZ.ÖÜñÙ)ÛGø}F§:¸×ò9?<|-š³qºô“ Š-²»™»Ìp«Š¨1eìÓËð( ÜÂÐ mNÛp"VÀkî¡ýMÆ1Êo\$Æ#ç§Í`´—Ö¥F)³#¦Ùïb)¹f†ÚÕ@ޏÀ°õ>Ðuö‡×éÞ× ´¯DÇN:Ó§ÅuJ‘_sš¨¯§ , ®†?jÓXŽâŽ_RîÚ¿þ½1§5Á^k¸‚ÿ+ÍK¦æô `¥”;3QñüìiÞi¤þJ±,üÈla|G(Î-³ôZ Ô’@kùÓ|J§ÂÔ7Y%IíQhÁ$…³;Y‹MìÖUÆŒÀç>¶éô %&Řô,4¸.©:è õ [ä&Ú ²o? QiÑqúÁ iü±æ{³äʪˆƒlH³c£d/óï§FH\~ük¶üòìº`Õ®£ßt¯Aá]´Ú*‡I‚aÍÆj °= Ö»KÚ­>gò‡ïŒq'´Kƒ€GYwŽtøB~0¼š'D÷¼š5^ÔéJPı¬TIšã³Ã¹Ìù¨!P ŒuH¢–Êš•gò¸ÉïuàFžQ_F»gdÜðø„úò5oµ,¸ÖyÔ|‡£áòd9`hFn.V[.Á–²iqQÀ=lÒa›Í÷˜NÂĮ܄ÊeZ‡~-/I¬¤ËFV|”ítÁ"¾à°gªuZý—Rð¤ËkkÉ—„Ky‰E“~µ”[±³”ùñpæò¡-£fçŠkÇ’­.Gª‘o¢?𼇫÷K»U“u"Ýðçó+Q¯º–Õ…Zhë[çTE¾Â䈻Ü{oM€\A=‘/ *!Ï®£ÜW‚›*€بb RÙø^×ýšº[90€ªô`§[JE=îqKÄ+pŒS\í÷≜íw- Û8ž¨E¼®ï‡‘áŠn,¼w¬rö+¦Òc#Iܰn"Bî O¸ãyäP—ßÔU|ZµQÁøÐMú4ÐÊESÄÒ-–7 /¶£Ú?¥sÎæûO"öv¥¯<@fì¾—î6ÁŸ±dMßù3ÃbÜ8Am¸øŸ .FÐô:Üç:s£ª~Þ݌ݢšÿØ?ù ‹aBèY±š.6ÚzÔ?r~ŒFcï„“X> wäd©†v2²²È-ßp£h÷ˆÎ\à˜ù•"ë2g™ÜÈÌâsð9[⎥Ů Ý/Û‰2ëŠ?]¸z!¡3®ógèìFhˆq{6¿Ë_ßñ‘”‹Z ÄüšƒL÷ñc;{sÅd•ùnÍ.©l›9ä̾˜À—fTû g¯©ðM‹¶ë°;üoæ¯bõ•Mãj&²_µÓ—дø:Hõ4ÂJ\"d·üg±Ð‚³É Z5§ðN˜®ç\d#uEÕaª²)I¥pÂÉS‹ôëÙ íyï.O`– (‘ Ô»Ñ h‹’Ë²Ž›>—ÜÆÊ•1|˜nåp–Šnj\àÀæ§]yéaòËûïÔš8àkXž2+òÝĪ"ˬÞÈâÐl{™à;=¡mVûAµ–„™žE@ZÄ’N¦5ç-ÏpÞÄÖ‘ “¾X$Ó¢ Ÿïõ e¹õ‰&¦x Ó= ²ÑÞ=ˆK ë 2žª~óç#¯&•EðGñ4,Ãû5AÞ ³´Ò„™#ÓóöÃäö3å7Úˆ†¬ý Vhb¬Ž.«’gfdH³rxò¶;Æ÷f®ßà5hŒÔÓú¨p‰%V³žZå ͎'ûÆËûSÆ å§ h|ð‰QI:?¦Õ+*†ÁšQy‡\x¼Èåi–\ÂÜ *w ÿ’ã%ØÂ¿?¨‰0ûå<¡7ŒxëÝŒ-©ÿº~ƒå/[T¬h'!0&5×™½Ê"¼ÞWñìx¯úÐfãøq¸t±›Å¾ï ¤ÂÏôþŠd"«Ë›÷P‡_¢52¨{2,Íã›MÞïAÓNH“cÕ6àmäHgIKŸm’PÎ9ªf ÇpWúh®â)m˜K<5mA––ÂÏ.±~44×®³Ò)SmÃfqÜŒž@—4ÇÑÃÞ*¿w€ì3I]ÿíóÉ7ƒ›J³ÏCl—ßFpB )»cC5_ÝÁV'm¦oC ¤M”Ý{†{¼îr‚m³öô§†'1r¤¦Y¹3o 13iŽË¡¤/¹N°jü]ú‘ÜLél>–hÑžPÂ|‰?°…¼Š¿k¡þ…üðXõw|#.>ÏÏ×%¯`QŠz"ÜiÁo,Mî]×[pñ ð€Œ§>zw} ©¢Ïâó=J[¯ìÖe* %{þ­H'$»<Æîö”’¥™ZñÁàk]¯JbCåš®F—Ók·]»Ã>«ÝlÃ80s`‚ú%ËËÙa¶»©¿>¦ƒŠ¼†i}ăiBO–iœ—MÞÈ(iáFý’Î˵ãð–‰ñ¿á§Ôú›µûU7±ì\»OlF£gˆ°W kß1¶Û}b N«B\“Vo¯è±I ÛßôØb)4ÒyÈ®610´mÞq4Ãe1d,Ýû %_Çj>Ï Ü)ŒÖÂ2oFªk#»·¯Eg×Ú©ó™HÌ¿^Bvž/Ñ:‚(ô>L'Èp⾂ÉÏßE‘œúˆÏ5d3Δ#îÑKW9÷8B¬Â]ym…Ød T û–Q<ƒ¿Ó9ûø¡=™S˜‰~*عÿ¼ï`c)AódXõ›úö¨–ЀMY]7½iÆ罿tÆg¦‰]C¹« öæßºþ”ƒ"w,³ \ÖPÀoI·Á+Ÿén|ا¢™˜‚{ø¬Ë캄ÈêÀ–íïNAêwú5¬?QýûàwT}ì‹ñf—ØQ²îâ}Åç/;3ºÔúu¬z‹ýŸ/Ã^£¶Çe˜†½«aÊúNÎnêÝ!3ÖQƳ¥Àc.èS›1‡F¾uZ"Ѿ¼ îFNL:x³ì–bl Tí}0Ð7BµéAÎG†¿4ÊÖ¼u]š,”èÔèÐæóœ/LÃÛ)•:øÌ)“!ƒ?´ª…å0ø šŠãýðÝv«¹½º¬ò™í.Q¦û‘”lý9oƒ>$ßüÃûá1ƒ4Cmï&¢ó/>$r„Öù^±öÐ^×™"›é›n#5­ÏñîÚèZéCÔ—‡ÓKHu¸ä­Ù¹¦(!» m¶rõ=•)"PåߘYºØz&9Úí IÛ0‡ Åj>bäÜp·ïA¸Hðkk(iµT濺=Qº^[rqBÖPV cêecoæ”å ¼7¾Ð½+ó&8j|ánãŸÍiù½¹ùÕ&R"Lðt88ãü)-‘Îc¨4çìø„M»ª¼Fê¦7QÕÞÄÕnï¸(—爾†þ[aÝg …¨ôùß ¬2Åž™ñ|ìÔÅ¡¡}ƒ\ÇD#ûƒ3œ6âã’—èŽÇFâÖ7rüEØ=«Ïx Ü’9pî=íºñú'M_Ô©¨Úkµ­©£Ý9–-§å4¦{Kr*yÔm•|f'‚‰Hª%CÑϼÞjìí¿Dv¯S.Pcï‰T rjÓl0žœÈØè¤ÆtVžÇ7Á)'ª­"×6¶÷üøÂoM©ª"uéÐR¹ E¬åÁ怖µsÊH$œßiè¼Üw£ë…Ñ1=Ô©´9:±‚ß“jMïÁÌ¡E–uÏãc—"ƒ¾}Î÷Q5™±×‘Aà„¬Ýuí¼# £ÆþMWç)вLÊv©i nÉ—$Æ©/AçãË*¯ypÛiÖsqÙ æcžcŸpd ´Ì·k Òîeå¨î.-yHT.9ä•l‹ÎÃÀØ[Úì;¢ê×¶Lê—AY±_òÑéy@DÛùiÎÙûDËì÷÷#{SAÒÛù‚ø¥²ê°ÐìٔиÃÐ8g~Kví}¼P“ÀŒªçëÅC/CYIûªsœíö¥D¿Ì~ìG æþAò¹Åš5¦Lv¼ýüá7Ủ`«zÅoõ†¢L(³ØT凜íp«ž0’"öÊŸ×Oµõ +È^¯‘YW…—º,%F}ÍVuó­;¡vÚÒfÔRÚ› &¾«ÜÆ[dŒÎžÑ[…æ}ùäbe‰Ð%©r½^çud6zìó¿Š5Íø6¼KàhMëÔ’>•Íß’‹à‡ƒŽó—T4çÙD~lÝ5¤œò—ÎÒÏÒ†s —«O ãeç*ºÒ¤Ô7j£îËKá•b&õÉàµa|É ôö¹yòƒCˆ¥p¯èoÑYjVį éûXÚ²œ^n[~NþhóSY@6eö8~]>]χ~FL T— žjHÓ@ðLë=1…ØA÷æ|ù!ÓÅí–Õ°üºyîTÉëX5é aÔ6Û¥_¬Ubd†á3W´Ff±fdÀãÄ…\)\o~®Ä„nb[á<7Hjð/3¥U÷;=œs:;ÃQc­N~ž[ôXßz}Oƒÿ>õ]ø‡Õì÷yޝÞ^2­›×ð†?°a±¥…åî~åX7Zçܼ²3•XvÍB_i¹•œ}w •B”Z‘ÉS.%q̯1øÍ•†±T(D\¾§-Æî[ZÝÒqï*uPß¼œ †M²í9¦::' Mµiê{wEMýiÙ l¦A|ÇŸ·¬‰lVaßSN™ßŠ£V[¯õ\¡QL§3öȳèj*K¥®ùjUÒÒˆ²Ìpágk€8ŒT­ÄÞâ ÑÜ =©äö!V³õcq«˜D°;Ù,Á³bàÇö‹›Jà5TÉjcÃv/Ë¿ö¦¸‹ÉÚßãë¾€× ˆÄû¼íP“5ö=SÔž•g±»—Fvi[©Ñu9îr#+ãRŒ•êœ8röÍös€XôÅn_»8S‹˜Îk1«FF)FϪ×Þ]†Ý8xïkCØéOöhÕDÞüyñòyh¿®+ ïwÝŸÉF¶Ž}’áxÁ”¹™ADë…ƒ}ÎïÂ×3 +¿ÓcÏ©öÑ ‰hÃöâ,ÇO8íSa8hàR3}ü¾øžLß»A°?ºi«ÓrüJ¦&¡íýÆ~J›Ë6\]]ô,×hmW°gÃR@ŽxÙæ²°É¯ãµÜŒ ó‰H ÷KCyoi†\p«ÒC wZ{þ^DÃåP­ëRg^¬éó¦šñ5j<¿Šˆ_×¼Ÿ>Ö¢¤È5¿¯„e,Q,ÆÂ6ˆ¾ZB²ÎZËgª<€$Rì4,¿:dyÁ0gŠìHUžæ†§s éyÆôkG“žß†7¦?íJΟ._5Ñš,nÊãõ½7'áE«Îv FzÞMlý>{âsÞ®ôB9|¨†zQ˜>°§˜ a 9C´õ×>Ï#Ç]¤v‘ÀЍ±zþ© ÉŠ ë[ãƒvï yý Ðçx\7”®Ï쀥0SЫ¶W*¥A|N‰©GxTa\eŒ7‘!+BörcL±Ðòy¹c*­ÃÇ£‚x’¬Y-±„úrÝò6 ‹œ¶Î`Úí~êdÆÑwwÝ.*¡“ó¾ÜËÛ$Ù%ϲûôÑIX#T¤MÑñø(Nä§ wžzäl†£7”ɱ–‘‘o¯¦\ê}¼ŠÊ[öÇòpv!ÅxÐA -ˆ×”úÓý¥úÑ*i÷ÝÇbïö\èwD1…Ôœõ'½´;žRáîoB$òŠpÊÛttb:6=É( ¾sYíÌÁþšøÄ±ÂòO§ó7r° ¹—.ÈzÉ~ççå$f%Õ×ügYý,S(ß-4à©ÀÎ26F77I“;Ð>I"è‰8Ôðsü†>¸«Þ»JN»×¼g¦íÊ"½“‹¦Ì%z;¦‹Sf6IŒÒ^v¹`xÊX°oÞn…ûë1;½êªB&{Z!ªm‰”íóÓ˜º™Òš5ø®[û8£ì8†åbò0Ï.”éÆëï;;®Y£…ýÁÕ|ÌT›°“ÅŸ.NË<ñ4¢Ð5_}èÐ7»î–UXvÎ$©¥a•H‹CÉüœÛ F¹’^€ ë‹´c|²eeè÷£ ñZ_fæ=wMÜkFb2‰Ë2»¦‡äñnemD…iD÷D[dí½«U”\N®UÚ<]¿Bê« JÒw)›µ;Ã9ôêÏD“û¢”E³4åǦ¬«8G¥Rðî ðŒÐ6¦wpä~P†$%‰Ðev# -µCT÷çÏv\¬88M´Z¹Bdò5ã®Ñ¯V¢§^Fwh¤m|$B’/Õé7ô«d%D”­`•ìzа?,ðÒû/óFC3·64ežk6互LòXW˜¢ÿ>vMj£#×q™a¨Œ’O^]-7”öœ'`ß»ú^\©ấ)… 2MïJ}¯:ž ð³PœòêŽëÉ›ü2ò@Ù³A°œ¸I[yì’ìvmv§˜‰}y>± ?nÄp~Œ÷õæ›®BU>íom‰;±tSúËF }FTÅZÆãzÓÛI b)nÎj läoƒ5VMÿ¸UzH,üd»$F¶V raìù-š!Þ/ç| ?B‚òà°ûˆ„÷iƒ«–Lç¿$h„êUPˆwjЧ膾_¸No¾*~’ßd‚Ù$‰ ³œàÇSaWó»x¯§æCáuî¢òNR{1}~?ª ›r~"\ 5aË5>'mCÌdgZ²hŠ'ï…ᯮȓ$ðU©K¯Íê{€Œ—K‰>ÿa›Ú‹…aq ¥Vž_½á¬¹ ç!æÕQt#ç7lÍ·ø(Y4´Ñúô<Æ Õ“œ9!yúFAdt×íïKªˆ »µA±~øõS¸Ù˜B@¤óâ…°ð]½Å óµ(ÕËĦ‰+ŵœöññæ0QŸã%ÌŒ½¼ç_ó?8 í÷¼P6£F/FÑˈ™l»"ÛÄ–:P.GŸ/ß ¾bφ·ä#š;¾žeúÜä¿ÇBæÅ⪣ÓéP‘Y«…õõ…ì+Y}Q¶³¦ûQxÉEÙXQd‰ïnµß\R3Àà…¨ÿZúäåè/± WEoLïº8¤ÁY~Nw•c FΪâÐCmË£Z7þgš<óö­´+yeK…'*»:uç:ïã|Hb UˆAKßýð‚>Òrv}rùdcCªÜô}ÅÍh¹v¹uüŠË™W­ôt‡ç”ã¨IºtnÎÄ­ÂPøHçç3÷Γ ­³fÎ+‰,xa$Ž,À×m#9oü$ ÈÇ)Ÿ³¾ÞÆ2M;à—~XjfßôʰS×PˆN,œ vVÜ&ÈvÜýns5!Ó‡nk“Ü32.ÀÁ r.u_Æžö½±É&‹•,øãö N 9+Ý2Û‹ÞŸŽ tžh«.9îL «åG@矲«nZ®= n…·¥Ü{Ý~£MzUõºŸn#{¨ 8ôh¡‘ÂïjЉ3¾PXd® Ößkµõêåê ¨¸/¹‘j†Ü}éÄnŠXd}°¾SµŒ,F_èÕ‡õYÚÚÀ¥a0P~¹îM:ÔÑ¶·v؆ 9Ê7E44Ðu“±a•«K}«ó €¢@” Ï-¼õ›Yƒë¶#P0ׯ€ùÕè/oƒYžÌþˆNðÊb ÿ ÞÀF»:2”¯ßAžð ¦m)tâÙšov–5¾0TîÅ.ÐS|4æÛs›¤šFbßl%vQr7^[¯D¸Z¶/Ts'Ht¦?ïn½ ª™am*hGæD•!„F=Tz>˜>Þï UÆS8B'àM«/W #°1Ƈ¸À±º¶â­µªˆõÄ'B$`d{¼÷@Å.…BOîDÊlfÉ»´bÌ­$Ç‹;>9Ì Ë"$QkŽKϳÛc‘‚ç'*(K’jCãW·Ò༠q|04‚N޼ ”†½ò‹¯³A´él6ëJÆ3qRÖ´||´/˼ġ\z. ‚û5B%}Ý*yRå·ÚfoWq–þ/>edë›Á)g²)áœB!ø3ëQœ(À$uƒV%]œòÐá’2‚©E\g´µT N×!iµ–_@"î ¸Ä0MñºXËÒòq[…:‹¤Ó,Û*[T;kܹ·y? p§7uÔ«0‚~S \§úDRrî~úüm¿t䨇MEsyŸÉ±ðªÓ%ŽøÔ=‡ˆÚºs9´JNæYej:?Ý™¶µîŠrÜÑ[!) èu¢. Ù`[R‘4øZ‘µL®_I”®txÿ â“£âÂ…š^FÈ b»XTÅUšeß#‚ɯAÓžQ(âä-T·RyºF¹Ôv‰Œ“«i¢‰¾+$UJÒAp14gj®³J®V[ìyg3r– ßQShÛW~v¼ÓøâéÏqd%ètpòÝ£Þƒ ½v=nvgøžý©42çò"ÊÊØ™ô6hŠPÒªbg-d¶ƒ;îÚiBSOJ¥–¼®eõ1Õ§'rh¶®Aáö J ÝíÓß2¦%]Œ§õ>1GùmŠÙËù'—ÎÌlÃU­4XRO[wç|»ÈÆ›¾ ЪŽnèh-T*:»½³þ56u4ÞÔiHùúÐν‘*QÏq@iXqÃ,u©Ö"’ÑS˜-,÷a߸³³oN‚yªqÆsIVz<×Ebáë‡phÁÜÅG¦XMת«ÞpIÒÊ/e×±Œ!Ùx¡’JòfýËö§Ìp³<Ó-K¨,ʱM_Ö«"±î娇‘²¸ª¯ãt@/‚ƒt¿9¬l¿ÀÙ@ü2¤Êe1Xò¬šŸESŽ4x:»>2S ™â&Øè„ê$ ž2ð²€ÐøµÙÉÌšçAB ÁÇ@ž­¬¥”.\7¾ç‘N– . àDšN=)âÆ™¯à ûIßå >ÛÓ:Ö±@~RÊÏ¿“¾åTË)]"¬¥»‚¯ÄçBbÔ„^a³B®ò²cÉijØ3³–¼"CÆž”Y5€_æ f{áà!Щa:9MÝh:²Ýð*}{½‰õ>oÌFÕ@u7ñ\xH´¦ç3øê·(òqØ7‰ ÚÎe‡ÌÀú8ØèfŸê¸ »­E€Ñ©Ü1ç„Fê±#3tÜzí£õRBH9}·†Ù³ÜKþÚ?D·î Û:¼ÂhGûÑ€"»ùÓŽÜ:Ÿ¤šòEµëê§zxhn:û(ú½˜ ƽÑF¨~…ë$%RÁm9¢¹/Pîó”šïiÍÅèKÌóÀŽw!Ïœ”7Yq±NûÌTIÝViåpHÂÍn—ðü æ Ú®w£~JÅpl"Gaê öŸïñfÖë6ÚUÑ—½ÜÖìË8ø›:ÒN–cCÊ!Æ„i9\èŸÄs20…òx,´aÔ³ï/ýd[þ½D‰öÿ5ã± endstream endobj 1384 0 obj << /Type /FontDescriptor /FontName /MMEKWT+CMBX12 /Flags 4 /FontBBox [-53 -251 1139 750] /Ascent 694 /CapHeight 686 /Descent -194 /ItalicAngle 0 /StemV 109 /XHeight 444 /CharSet (/A/C/E/F/G/I/K/M/P/R/S/T/a/c/d/e/eight/f/fi/five/four/g/h/hyphen/i/k/l/m/n/nine/o/one/p/period/r/s/seven/six/t/three/two/u/v/x/y/zero) /FontFile 1383 0 R >> endobj 1385 0 obj << /Length1 1650 /Length2 7348 /Length3 0 /Length 8410 /Filter /FlateDecode >> stream xÚvT”].Ý *0€HÇÐÝÝÝÃÌCÌÀ0t Ò RŠ”€ ÝJƒ”‚ ]’Ò‚”H]ô«ÿûÿ»Ö½ë]kÞ÷ìçÙûœgÇYÃʤgÈ#AØAUp—_ ¨­läðó òòó °²ÁP.пì¬&P¤ —ø†" BÝÚ”@¨[¢6Ððt@  ¨??@€Ÿ_ü/")PyÁ m^€õ `UD¸ù"aލÛ}þú°ƒ9@qqQîßîyW(ÁÚ ”#ÔõvG0È`ˆÃ (ß…`—rD¡Ü$øø¼½½yA®¼¤ƒ 7À†r@= H/(ðK2@ä ýS/+ÀÈæñ`ˆ°GyƒPÀ­Á†Â=n]<á(p»;ÀP]  ë…ÿAÖúƒÀ ø39 /ðïpzÿ ƒÿvÁW7ÜwØÃ\ ]-^”Š‚C~A.ˆ[æ²»%ü>: "¯Ý*üSŸ sCyðzÀ\~iäûæ6ÍÊpˆ"ÂÕ Gyü:Ÿ ßæÝ—ïÏâ:ÃÞpÿ¿Vö08Äþ— ˆ§Ÿ1æî UWú“sk"øÇæE„ùÅDEDPwÔìÈ÷k#_7èoøË|«!Ðß á°¿• „ÙCo_þ /(…ô„úÿ'ðï€ÀÀ(€Ô'ø'ú­jÿÇú¶þH˜À’ÿ¶ý€þ_Ïß_Ö·AÀ]|ÿ¡ÿ.1ŸºŠ’¼¢ן’ÿ>!€¸?($,þ;Ðß)øKþo«öçñþ#¤:ÜÿCÅmúþRâõgk°ÿ97€ï ƒ¸mh(€ýŸþ·âæßþÿ¿§à·Ëÿ­ùEùöÿŸHÅÓÅå7Îþáp+ÌÅ÷OÆmC{¢n‡Cq;"ðÿ¦šBÿ˜hm(æéúߨ: t;$òp—¿ óPù@!z0Øñ>ú«·á]`p¨ÂöëÎðùùÿ »;°óí½âq[­ßôvªþ½¥2Œ€ü?a‰ùÜÿv% ðÞÎ)êó»½|¼pêÖp+/`@üª©€€Ï C] ö(;˜ƒÃ/ø„ÿÄŠú_ ø7ö»ÂÿÅoo» ꀹ@`n. ß¿1AŸÛíeÿRá?½ÿD~ü$&~ !O0ê¿ã‰ÿ†PPÔ?v1Ÿ‡§«ëïÛüœøÿüÛí_i{"o‚úÝú·9ÿkýûŠƒB} `‚Ù)X2Ì©6¬õGµü=ožõaéqÖuÓLÿYd›çOÜ4Žª¡KÈSù´ïHçW•ÙO保üwšêq#›Ÿê·\\Ú$Œ®·ÌŒP÷}*Ü‘¯ë¥Ç¿Ïc$·på`âŒÙ„Þ¡Ášçî)F¢÷’â‡wªO]oéçSëúU"š„—¥c<ñÆqV!ŬùvÙ“´Ì8(zW¨ eoQøsÑ8ÁÞlj‰‡ž:ùÖ køP2ÂÂák¯‹wìä_¢0D5ÐÔITtù°-81^9‡<õ+{Æ)Zdæéåý¼‡8®x6â‡ÃéØrãË}J.¥ Sù¢Þ奉:†=úµþI½lÉVkd6™\á2:îMB.Y¾*nZåòã½õ ûŸBa;b3çþ^©*ÚÇ/ÿøò†9+¨ü+vʶ÷s¯ò/ý‘×TÀ†ëžu44"W 3ç»øÎX ”“[½Â@*¥œø¨EÚ>ËAS„›Ëê#ö½o’TcÞ¡+R¦Wºfó¤ø%ÔnÍŸÂp>S{ç÷O¯È=Åg¼fS§ÝKýÙAþÒ÷MHM/Ç¢­§z™|Ú(®b—÷ž¾žÈÂzIË¡¥WU)@õÖ®Øf—Ö:nÂ=ñóŒâþ“6'Ë *µÚ‰:“pí';±¢!Ýà™ Ô‰…Âúsö›°•"û<­! –zøª…]HcÙ|‹_ö3mß``t’—!±RÁ™:T÷;S ]íû ãå$6Qý½Y^¾§!1Á,RÜ“Ïû+2Oüdîn´Ý?Ó}áÅ3÷Ý«`™ãÒȨ@‰“¾E]+0·Ñý~õ—’Ïv>Ó2g§ôž$mI•·$†w¤]¬'vŸ"$Þ¼©Ü.T~[û\¼g™(ÿ(ýÓÕ)qT¿eßueƒ!É÷‡[ñ’©$Ì}"콯ë4ïÉ*2 ÉH:8r´O©‚š¾á‹Ãõ_3.tJ…ÊÑlDd³S0JÍ¡GÅgÝ àŠàœÔ댤xøºÂîî ¾\bYU'q¬mÜì^W9…ª_ðDÖaÑçÌ…1Â×¾~ ð/3FÙûzÅ ñW^¼Û±öÒ”=\ˆµ ÿ= Ã¢¾_Ɇn%>ïýá¢;\¯Ó-ñè»Ùv» V±‹Å¾&0¨ƒM‡îg_ôò{Uœ|ë@²¥lrðùÀM`Fð¸¨ÎgŽ^Ô Gá£2euöGó)Ú–w–cmowŒìK,Ûw:î ú-lùçÍ"I§*;ÌÀ_{ û±‹zË*0ÎíׯhνáFûF ³ì›Ãt) MË ÃIÛ5¾ÖÁ<6gžBeîžë4…'M§ý /Åxæ^œ1äœzÉ•ØÊÿ>:Xj嬨ºñ£WNå”+®ÅÙ~QÞ¢¬*Æ é6SGó; ®ý+˜‚?Ï·n²íè Eè?Ô$ÚÅç9Ú<¡{=ºŠ·U}MKÓùúîÐÊ[§\°jE°ÁŸä@[s!é±Íé…Íl*hÁ¨CÓ} fˆß¢DÁ­0Ÿ2µÞ†# qš^¹&s5Ç‹Ö,[ÁÊœNL¡é2"•|_½¹[Áe)”¤pĉv?­Ï](HR4¯îöŸf¸©,†áYËÃiÁ´’ÑÅCK­õ½ó²ÓƒGlþq—z&ëƒò¸³€üÖôFñ“Š”PIi¹qÑEøÏ*‚ºÜ“I†yne²¼†Ÿ¡\-¬UG6$ ÄÃÍ¢ì„á>Tš"ÏR/Ýßuë_P¸,HÐTâ¿ú`ë¾7áû –{d£[ÐŽ%ºEùE[“©C¸šÝc›Ö^j@™¯EÍ{ÑbGÆ¡¾¸gH §ˆZȨ+δ.)F(7·ôLâñ§ò‡Û<‰ìJOøùß”Ø;+ÒãRZ†äV¦ÏûÐÅ-°IûDt™«}M "ð )Ž!§º*÷ŽåÍY±æQZçMÿžùŒzh“®Øû g5[ÝZi&³¯¬”íebXòð·S%ÕHCœýcÝÐï•N†uàŠHÝ|ýQ„{À¬´Õá0™¤\¦FL`ñq‰U†¬ÈÌ›g·…bÜ5{ÁUùid»ºât,¼=º½Kž¦ÌŒL»k~—$?”¼8…ä׳y .á*µU'¯yQ2¿àÁdÚpXMÀ¤w!¥•duÞn"f {îµËŒú–õ³öëwXò)"S:þÁáïY›Ó¬ãi•.jm%7dÔwZë6 úÖ™^ä+Â6gº3ö#ý41.´žÁ<ßtzヅ—ÓV \lê.Iì]¶§ËÑöer+Yϼf¾”Ç\[\U9ê(ÔZ?)¢<àõ} ;1ì¾ÀöŠ{ÿ€BòÞÖ´ù¥uY‚m xèY[cV ¿p„VfbÈíU%2ßû…·äú `••rVftìˇQ°bÇ×ô gæ‚dAFÊUD­§ñ4w”HÔšÔ=¡wBx¶q>J}=è å)Ôgí0‰ÈŒˆÊçÖÞ_sN²åKsÒØœ¬ó¥/.4…¢°é÷•廥 k†%µu'@}c\&f,W_Ùq/л‚f9›bhS†DÄ3,bª­ [Z£žTÙ2 |’¥¤ŽEyªl¾ôžF pôÐ3F3Lñ­ñ¬bêzÕcö’uÉù¿Ž X7³ŒÜ O¾??B÷ ŠðGûÆ„?Xà÷AÊí‘À‹²ëW°o¾{ ß¸”3ðîe5ëX¹AT€‘L“S‡õ h‡ž¤Ìßfì¡ÝlŽ=Rêþ#ôÁ=½AÒÓþžŸµs18ù9âBÆ„·MÇÈÓÇԭͦütØW’«/rêdÕâuhÞø/÷ïm`:ÞËíSϘgqÕÛô­{J3‹÷àI±$U¢ËŠ™¡ƒºóö>QâäJ]Õ¥§@û??}<¨¾>Jüzyš*.n݇~޶º‹SåèL¢Z-&eÊÅœX%]Í0xŸ vÿ}òݘN:‰4UJóÕã«°æïªuè~{hÑþ¦h†@˵žF|3æ²0£¹ª†ŸŽ·5LÐ&íœÐ齪*/-,#ÂWtäiVdk“è|ùжæ¶Mܧ¦~à/AÂL›ÏŒ¨››êtðñ”SJ& MÜž%í »ó¬á¯ æ¥EÆß[£QެŸˆ«Oš¤®Ñç_M` äæ{ÉœZc:?âb×JT˜Kno@'–= Fg'Ke¤m%ilÓµ®•Ü»Kèd˜ w ÙƒÄÙÿ4ÆYð££Õ ÿ¹C*Qˆd.Œ‹Ý×~™bßÑB íÒ”ÞÌÀ¼·ÑBäY áNöЛ+±çpØ—‹ ÅPÕÕô;I#x¼¶ÑÙ„oøòÑAêÉX 7£Á¤}Ònãµò;«¬Ê›|z5³×kãŒAI‰õo–ú‡ãëâ3÷Ñ íçÞiÖ_aÞSt_DïõÚ®jX ™:¹†ÁNÙrÑЂj…€ŒËíôoB½[å==ßËøõ*Y‹6ßì|8€¬ìú?=¯À£[5EQ}¢íF‘› a§È?ã²â¦!v³°{ÓÑý²©Âìb£|wwæ²óíðcU –{çÓã5]ïÉ·+žK[›‹’s¾H˜B>Ñ¿0™êχ!{OJOЗ!Úâ ãð»œ-zµ€úöû /–cì–Ãʶš}°î@I‰N倎ºNØ?©ñÂÏ­Î2‡|ÁÞ$™~ÔÙ$}NRhCÃø¸Ú´lRïy=~šq^¥ÛÖGåÐûw6×^Qñ–;÷«?­\ 9HMmµÐ¬}VØn>·Ø±êÀP b( v ÍF6FÜ¡S.È £‚<¢èõ.}ƒ1 kxã x)‰{"@DË,Û»M`mkSCýÚ@‹+¦Ç€nµ5|ìù‚#äþqVñ<2š£,H˜¡HðY¼l`D×jŽÖóO|å€ç›$Œ€vGãÚ»—#ÐZ:ײ=n>ÈZ&}Ì=³šè]ÙwND/LCOÓ6\të]ìØUÌņTÉ~.Õì UóöC IðÐj­K¤jjS(íäü?~(”gDç5Œ"þ`/‡è<Õ*èËÉà[øÚ‘ÚÝK¡«V •_)ŒÄê?f!•pÚ_'¨%,êi¸sß|­sä‘=²~K»žFçǔɘú謰á•Ï£l=ù­áÉ›¥ ò9šÙŒØž¨Ûd[‹'˪¶ Fi‚»j]B+ÙÓl£PMbÛÒùçl®7e‹t+‹¤dN+M¢óˆå5=wòè$2 ‡¡±à½­åÿs港:x‡¸ ßæfx·°1—GU–§¾^5ɼêÙ†­\Iýâ¤TNÃÆõZ®¨†§öž©­ƒE=A®úù5ö ݪ‡ÚÃv#Ûæ+=2•¾hi¬·O†éZŒï¤båC÷vf.¿7¡‰EÈqJqÊ6¬m0$&ñîúæZë9&ÐèL”)>º?îþ¶4§pد­ÖN¨7èÿÂKB¾ÝEplcöafx­¥}éÉ&¿äæ ¥à¼kkŽ+ã0ýn¦Ú*ùuÖnAšBÜqH]F%~±Nyƒšzrøó…QÓ—ö¾ÙD†¿sz˳1Êf /„ûRéZHè®]OÈ'¤–'µ úIª A+¥¤XÅI|6ÄY<Ë#J¹lˆZ¦ûºkúl±5Zçr™D¯gx™Œ}F¾Ý8IK)¹oÔh¾ZJy˜ã)¾u5äž5¬$%±h¨Š%[Ѳkãl=¼ ¯]c³ÎùX—MGKbUVtõ`д­ƒö3 |#r‰?[ÌILÆw»$'´Ÿ> 9‡Oû‘’0 ò!/ã^aL¾ÚâèuÔ}“¢« é"mÚH ¾ž"Ú Ôó™]D¿ñUÓò3ïcÉ—í”7ôŸ@‘ÏVS\0tîÄä!sŲGªÍ¿!}Åcc{e·Á޳9éL?ÃÁ”æX_ãÂuãHÛ>ÿ’ŸÌºßiÆÎÍH·jü™ÎÞì;ÉDxõêÑ—ÙLv­âFB °à¬“W¤ß±cÌØýÉ-ñ­7©•1÷D§(,NÍOqÐ×ãë7}ÔíZ.•àfz“reL®ø€sÎÑR5ä+ºPü¦nÄq; W¡ïÓtæ2 –fZ¬æ9T/“>(ECÊ*¸-q.Ë++#U^mç†D9 šCGj:Ú$ éâý‘iVþûa7­Ž¬^â'l(jÜcÊ¢ú"ýÞ´‡O~ªÜ¸õN}žà3]ßÉh<ÝQiú¤’’ù´“G¿PÙ´uDI¹ÖoTØ#‡…×G HjÀ¶S^#qñ`!ë ’s«Â458'ëûñb¥±]Ç +kå/Ü€nY-gDma‚neáÄ5¬_%>:Æ'mˆ],Ú¸rNyÒýSÌŸ7ûîÝcæÕ¬šx`ÂáÝâ]ŠgÖô_×0h„òÝPN½1ÝÉŠ©™ï#SCb‡z*îN¨2B7£×ù˜ X!,ß)ñ—ä˜@TÚŒvß•kø+º²0Æpe«¬+væ…ãm;‘gêO ]2NxÉûm}è¾sÍP„>p·ÀrZ­yµ…2Õ¦m«Ö,•ì«ê{%ÿ‚Èc÷¡û¥½¤9ÁÄ [À «tIA{?9:¡e£ì–ŽSfáì&051!jËS_ÖÓÛ¡}|¼z…)3± #š¢µsÅGVè »^>Ð>Ê«g[kÊ?4Ê&pÁ¿156ìuáíÂk uy6¼G¾@ßËTiž‘á—5ÅŸ©™ µîV"I¥u—‰SAg}FZ”M´[@êú¶´…nV?û7®Ò+Û5|‚faâ ç/F"E!Žxï1:¦Ê¬M>ë''Âv_¼$Qb6µìæ£D©˜RÙˆwýòr5`UMÅov‘• Z0ÅÒË'hÇ.™NƒHTŠ%+C½_ò/®;ê6éåÓ«SçCw‡æµÈÀ!-Ê ÉÕj")äŒá#,’gX¯ æ§@ܶuIÑ=d™â…V»ïJ)6²îN`ä5ÅJ{ó|gÛRR!u°“v¡{¡{ß{ÀO2QJsï•Ù#‚‘ ’/ÍÕÎùúï3”1]Ç\÷yª|Ú-]}‡~t§6Ø’q¨—ƒP=þ“B"†üC»NÞ-ìùEÓÈîwþjÙ‹¾A¤­“ûÅŸºæy§ÎÝ’ÚoÐO?ñ=íy—)@³á**vív«(Çzõ7ô‚“ãö߈|0!Þ>ŧެ 4E•­_Ç‹ºŸÎ+Å,> y5 ò(;ÞoÈŠY¯Ú½‘p~pÙâ›ÔÊ…Š´ ¥é€à¬ø’9Šâ“jVZŠ^_Õ7œ#öŸðú´ïÆ ¬ièÀs§\شؾjq¸Ç²¡­c¶}]<“åÌßÖo™ÅF£¡mpò/õUæÛa‘ ZÈ%CG\hRc@åï—óŽh›žEÖÇæF¡´Ag.ƒ„hfàž}Ù.039ÉŠžÈ³Ö`1[Æ›OBP}4Öœ÷oÏó×lgFZú{Mdy õw#)ÐJ6R«UÀÏ¥"¶5&¿T˜Ð"$ŽŽY;žÍ­,¡–<ô϶ÅWwD ÑyÖKÛˆÔ1ò·Í D.Š µj]kËX¦rÊp†(©6?V#5{Øçñ6™Öû=çáŒü”wƒk’÷¶×<Ø+âYôfÛŽ]kŒFbK7~7­ç(Ú)É"¯ÄåáqzÔç'OˆoBW×=<ª” ûÖAÓ8Ÿ]ëmgy¥r9[ã Ÿ¢)]ˆlÄË–ûl.ûf}3R¥|TÒ3¹éÌ»qê/©j*NÕ+x ãcùü”ðÛÈkøq0™ÞôžÅwÄ«(K^l±C$â÷ôm¶ÖÞ!JußNXï2Û››7»-‘Å¡§,<îzDEb¶só@èJ“BÁäLt_w=M Èï9œ©‹ÐÖ­ÅÁȦ:/zº Ml·å·í11RØV˜n.W߈Éã/~©°/ò` c¢XŠ)&a¯qu¹_:Urù ­a8{XW~mÁl©âBU·óp(3Ы¸Yš ÍÙKf$V¦l€.$MÍPâxt¾˜&ôÝÇFºjlý#þ]ÿ©gÓú7”¦Vk‚Ëjs„±ÒÕiÝ)¬Æãü›¶*•'ånµkµ>:ôy‰ûé†.ÍG\˜÷Kœêo(£wÃÞ¸ZcßÖU¦á Ê>¼3.™-_ÜsL¨ü—H«?©iÔwetˆ*ˆ²Ù6lWø”L›çº— ØCjFÄK4hØm|wb²¶2Äð5Ͷ¨;ö¶UµÆ R·6f¸úm¼––Lí£ÅxÜŽ))£ú–×™'ÕîS§+Æ7ÑÛ~N£ç¥tñ¼V!™”i,·Êðê8YŸÐl¬wóE‡×…hÈpŠ:a^¬õ¿Ò‚–Šâ˜ãöÎå¸àR2¬Q¹ªwÖví¿àXšd>{VxR¤÷Àâýa]Jfô/‘ÌëE˜V~Ͳš8 þN»FNÁ¤Œ—?§(fCoÁ)¯ŽôשÇt÷¼ Ð49—1äÎD€öLÍšS„§Ð*ܬ.Îòǰí#e½"V?ƒ‹¨æNÊåÀ¥3wžÖVl¿½2öIÚ»DB^©*Gù7Ã[bŒ;09sljǩ¥Ä.'pÃô±™fÃ9âòÞ/.¶A©_¿Ÿ‡íÜ´´CžZ(¿Õi$[NIËG—eç裂¸Æ¹šè¶t+éiV#¶”®$Ĺkza5Û³©"gc'vöª$Ø96›]³Þ¤Ýø|¡©í†nî5O“t¹vÈ'xÚv&0ãB­§–±HV6•~ê®Tô_.\*¦_^S¡®¯Ä©J¶ßÍØä­²`©c¡¬@òÅ’,ÔÒgŸÎŸ•–Øî…žeê+m Š¿ŒH›Ù›œàu^?£>6öN­]êI~pÒ¶úð½vï˜?·§?í)Vš³b‚Ó7 ‡€îÄ÷*çú]åg÷½6$®§ZUŽ3G>À-4ì$ syX(4j|­þDæ³LGvÎG" Ãöí%L¡ö}Ìç~lx½ï).{ Tßómü´V3fy'%“öxZ«öðµu+ÜõŽžþzjÈó]N¾3ùC»èïRq‹[)›l[]§"Ñp¢{b£k d9ÔÉ{mc²³¸íÞC–HR…“ègó[bRœM¦ëÖHDœÏ•™¹§Ä‚8º¦HÄ8XÙuñ‡/Uãq4QªÀu~\‘ö$´ì˜,È7×|á¶ÊÊ\r¿žž/o,ïám;•l‘m¸oš?¯X2CãÅÎã@“7skÔ^ÝÆݦ-"R)~Pµ¿,\6±QÞNæD4Šþö^ƒBƒŸ[Øè8KÃ|ŒG…踉߃á_£€Üâ5DÙ¨ÜñGŽ–N¯­º6ϳfAÍÓJ^ÃU„™Í÷VJD“¿Ï«Ý䨖ª<ê ’oϘOŽ\ Pà;±Ó{^ âÞ‹ˆ5ª!«âyû‰¹¦EÁˆŽ7‡ýG[=íÌÞZÓ¦©ÛUÞEKœÀÏt.r—Qº—O<ÃZ(pÛ}*Û "¶ž-o‘øÎÐ4wIïQƒ…ÆOx‡›*ú¯väSÔ¢–‹pÊz7gÐÓ,ßH¹ºß¢c%pË ÙƒzŒ+á\ù6´ÌC!z.r‚܆Ññ‘*›%Ü[Óý?hqê²Æ§?!8­|²Qþò]uï‹ÓùᮘæYê…?s´+ÕNbÿõúðè]Ÿz—H¯K=e´Êž·ZÚQËdÖ¼…go·ÏSŽFi<Ü£5Ï}häó¸žêOõY4v~KH¤>©sÉmÁ#t Ú޲0L*f°>æ¸Í|Ó³¯Œt¶øvi¸Ï\¢=:²ö/äZ0Œ‰6.4êKºb´ƒF1ò,EEýeš…'r…Չо‹ÅõLà¦Ú8Ã*!EÉ…:ÁTSr$:Ÿ˜Œ’]ïààsý/Ópw endstream endobj 1386 0 obj << /Type /FontDescriptor /FontName /IFDACN+CMEX10 /Flags 4 /FontBBox [-24 -2960 1454 772] /Ascent 40 /CapHeight 0 /Descent -600 /ItalicAngle 0 /StemV 47 /XHeight 431 /CharSet (/braceleftbigg/bracketleftbigg/bracketrightbigg/integraldisplay/parenleftBigg/parenrightBigg/productdisplay/producttext/summationdisplay/summationtext) /FontFile 1385 0 R >> endobj 1387 0 obj << /Length1 1936 /Length2 7190 /Length3 0 /Length 8357 /Filter /FlateDecode >> stream xÚµTl6L·tƒ¸R’tKwIJ ,°Ä.± ‹4ˆ+HHI(% ! -)Š !¢t*)ßêú¼ÿÎ÷=‡Ýkæš™kî{æ†ïª±©¨Š3Ò¦‰D D!@°@Í@ÇÌ €Áâ@0XŒ‚Ï Žò„ýë à³€ùúÁ‘¹?(j¾0( gS‡¢pL$ ‹ö@Ä)9ˆ´ ƒeÿ!"}åêP¸3ÀÐE"`~|jHï@_¸« W蟟'ADVVZäW8@Å æ w‚"P”Ì WÑ ê 0E:Áa¨Àÿ¤PpC¡¼å@ €€ ÔˈôuUÀQn˜Ì׿ øÙ4Àêû§7 ÀÌ î÷—Çé‚ €úÂ8ƒ'Ü †ðÃÅ Î0_®<ÀTG`ä CüEÖÿ‹ øût äßtGÿLGü †:9!½¼¡ˆ@8Âà÷„Œ4õ( JE8ÿ$B=ý¸x¨?î uÄ~i‡4Un ¸ÿnÐÏÉîòúÁ=6 ú™wÎg5¤— ò£ø©Oî sÂ| èŸûõ@ AÿB8ÂÙåg#Îho9é¨ÿM™(~Û\a(€$ –––À|0Œ“èg ³@oØ/'ä§×EH7Òà‚kwá¾(‚ü þ0Ê úÓñ_DœáN(€#ÌŽ øg†¹ü…q#à ÇlÀ¸ „À??ÿþ²Ã ™3áø›þë–AjZF*zÂÿôü¯WU‰áªŠŠ‰‹¤¤dR²R€ÿ¦ù÷þiþ—Õ ÿ[Üùt.H€ì_=àïŸ>üÿ ¿Gðß †HÜDÿÀ, vÂýü?¯Á¯ÿ¿éÿ™åÿ¾ÿ+Iíéù‹ ð7ãÿC€zÁ=ÿ¦àF­‡·$ˆÿ¥ZÂþZj˜3íõ¿^·&*Wܨ‹B$€`‰¿ìp?M8æl G9¹ýšœnWÂŽ€#ýà?Ÿ\ü?>Üò9yàž?ÜýrÁp»õß²'¤óÏ%“”@}}¡`Üœ‰IJâ·­Î0̯€€$ ÀµpAúRü¼Y)iÜTý4ýBÒâοH ÿF8Ÿéo$ ™ÿF2ÐÍ‘,.'ô7Âù#YÈé_ô³cóÁþ€bË¿P'ÀîÿÛ/ùÓDûþ€£¸þ% ·¡N²[ ·îÅûÍÀÙà@œr? Nºç§Ýë7Ä- èT¸í!ÃqqÿþpãÄzÿvKán 4Á©ý£NšßúÝ:®6*ù‡§ýÄ)÷ÿâÔü†b¸ḩ<ð·4\èm˜ï_¹ÿ3kNh__܃üëUÀ â?ø×ëƒa`NÓ“H'ù;î5wZŽ«T8D¿ŒˆwI™^ ¯HPÄ{â/ : Y=J9$òdRsÔÝ”*m ñï>àЈ—ðUm’Ò3DÓhǰo_(ÊQí’ Œí»Ø“âXÓ¯©¾ÖÞ¹Zvhŧ¤6LozÉvùØzeÕLj„ƒ«¬ ‚ÜhuRP–ù{ä¶‘¶gè 9’wýmê`÷Hl:¥î«ú7'×t­)ë7*Zé&&»U/I†½»«EªÙuä°*•–k~%öÝ­UO¹ ¿]¤(öɵ•o"EýGÓú„É¡<<äëœlE®÷7 XòáË+×.1|‡ËìZ·`Ï„Ç&’9Ci¾ÙØŠ—® X#ÛOÉ|í+,0 ðœx1gcà÷Rç.ã˜à)ª oª4å™2>Ü+d’òC¨ˆù°ùú“bNYú–9Ù°Þ¬fjúø«­2O¤-©Úï‰Þy‘ü »-FÖŒ­´"Ô¼™÷4+X…Qqõîî8Êçóßq|[BtÏÔ×vðué=Ꜻ~¡o׸yë_“ÊÂ-î•í–‹ÍÉÆL¼*øQLà»6¸ÿ­›³Ìåvcª0ý}vù¢åðmë7‚ùƒg–Ç .ªò$5 ¯[%ˆÄA༖5Ytpªe"7Éí«M…¤¨.5ë¸ó–èÜŸ…Üi½*{ æ7µ¨3Î ñ'›ìô!aÞ÷Ó ©72&jÉ6j•ÊmmHÖ)\²2™V5cH?-çêüJÞd!µ>o—Z-Í£ý Qâ—pŸ &lпO½ðE3¼+;u"¡äx^‘õ°ÿ‹Ðãmã 8&üîAž-F*¾¦]ÍÐÌä~ õã“™ë5õ…§àü„_ƒI¶$?¬_ 4ž‚5EZÂÏ>Ѳn6›MgÒ…­ìófFA[ó˜n’VަÍE;â~h,3êÊ­;arê×FÒV0nDÏ /™ñµ2RòxØ5iN¶DFßd5ì_äªV¤©¸ß=ñŸ`ø¶éòÒ7ìueH(èÃc gA‘~ʯQõô¯öüñàn¬ããÃÇʬ €ÅŽ!‡5DOB×ñ˜G¹NÇĦîôrX{"Žö¦Lú7Ö’>;ŽÀ¿63Xî#ЗŽÅŽ®{¸===v²$ìÝYåk¦2\¡* ¶+îoùpêºWH»Dܯ—¿úBIˆãÑ^ýŠ{H1& U"Í;ÈÚ€„ÇmÖ¥˜P0r¯ù'©:Ž÷ò$f‚ºrì#D _‘»€’´y!rƒ±{®ìâØéŒõa®žXÝøšŒm Úy¡Ÿ­¡Qþ.‡ˆÊˆ¨¶­E†î¡æ¾I°²tß³4Ÿô4tŒëæ@ìm¯ áÆlëÆ¶[€ú•B$ÕyS¬T—|$ŸVNÿÖÑö\Í!Õ΋OBµBÎ)¤5O|X¬‰ÀŽÐÔÌbØFÒW%¦hÑÄtüä¨U=»mÍ(É©o²¡N– –6 ›“c–¥e©CVëŠ<Ÿãt’o/èSX±+Ö}î›ßÄO]ÿ¼–ô#_T˜Õ–/“YT^¦cÆ.qÅaH±Áè¹äp,Å·ëû³ô,'i¦9ÙW„óºf'’2ÜWìî%Õ%H͸à­fzU²ÛÙR{†ºýì¢ÓTEþÚô¡¾ PER@þ)Îâ‹ÄF%i$—²Mñרð³ÁE¦ã&'w06­&âûÉ ñˆK4‹- ÓR½PFÇð]5–6:D²6…v.8ÍcyêhôÀ§Õ˜v8&)ºOóEµ\Ëx`| ¸^ÊõæÒt¾ýrué‹H/Wï‰v z‹ºR®‡Ç4¨^-1JàÙiëÆ½¸‘©íœ ”Å@|ÃxÆl2«G@Zsoæåy%õîÚ°Ì :½à£ê}øX•>¹p°Áµ'rŽî|X‰#!¤!{\w~‚Lc ô2z?’¨Ðe~XPT†}aj(›7)˜ŒÅ_Üò£òÖOðl‹À3§y ÈÏ_=›¥ÚEHkÚt§ß%¼¬ƒtN2gÜù¢»dX1ý”¦†®Á§Ùv¾ÓôvÈ*ŒO‘3ÊκÞß È:;aØšÝ<íýpO:Œ"MN5AL4˜²¢¬ðu:Z‹–€K:º­­)WIáˆ&îŠ$UmºÌû÷.GHâXKÄÉìÐ5@@Ów‹1ˆµé÷;¤­KÆz#“!5}èâùo{º­w=!Ü­¡ØY|R;{å¤:úŽ+#C@ºïÆïÕíN6&ØÔWhxމ“ƒ6e"´/„!o2»Õíî˜@¼‡Ï•V•2·¿¯¾->4W~༙,L»mÔûêKÞÙ×Wå%®ÞöPæÄ%ù¼'\Øï6Níß²¸ï’AuD†‡™LÌ» Ûmˆ‰îAxh¿Õ–c`îÃ/smëIÇŽ}>|syÕ QjØ ®ZǬD\~PîdÌl=á=ïO®Ù~¸DىǚW‰”ôÓ!°"k¢ õY8Yذ¾›®]3Bù^ƒ‰PfŽ ZŒ¸žë¡–›@û‰(÷§;SãM,¥‚ã­=‚Š÷:÷²_IS²ÆÌ™mã;?­­˜9Þ  Ow—Øt^=F¥gj\žKZ…p5)¸@Òf¯ûÆ2ÏGæâÞM¶ðAüUEúU{ï5Ñ’BEï!‰Š^'R—¼X?«§ŸOwÑžR"~Íç‡(±Ió|眇ïI§ùbÃ>·å˃fŠªŽØo5Ø:ÅÉ è*ø.|^t-žˆ¾=i =±9]tU0#ŽFlÎm@‡úYúõÓÉ£(e¼YÑKI]E6¹ÚÉ¥¼Oo"ž~+²ž2§}™ò«~õ?U‡-{íõ¯TDT~XfÌoÒßV“‹)"*³ì 0&zÀÞXp};~>IÌá¾û8,¯zyõ‹Ê§S{ÎëÝ îÁ$“¥1ž4vwÄ+lj±_Â/ÀiŸŸš•ü`eÛ9ai_Ò“®|Hp¬¦©¹Þ«×E¬i„äkºÆl”}¸.PâýÄ×&ÈÍLôQ)) éÛÔÉ…ŽäÉp»RÂzcþâ©—ïÈÂZ VÔjD£·Ç1[vP?æN‡K©l""ôc‰Y¼bgW¾x±°öÈ’“r2ðlœ½ºõH½WyŸB2ÿ°§%ßhy´6¾\D9É|ë-ÿðyšîÁ¢¨áòæ¨Ùs•t>ím0ƒ‹ãƒ]Ñlÿë´[ ƘÀp·:²úýá"®ËY#½¶7k«Ýäľ.7'½Ô”ïÜìÖW/ŒÚ>BžÆûhÒãFßÊ¢ˆÝ~?úô„@'ôÒš(ñgî8Ù rß+‹ß_÷/¥.¡³2^7ѽTÞ:åÀަŽèôÖ‘êÚªšp`òÃz>çM03îrО/I[…ßWôWÛµ›¬tÉR pŠì–IéÙ}æ_s>YGÅ¥ÛÑãb‡´¸ßƒqœ /’oø Ñe ÜË' RðYÅC êÂO_1¨ÿ±Ù=·ìFIi¬8æôže©,h‚,ù …Ì3a.¯nð¥?6{\â0Æòa¢º³;Q-éIs~gUZXC6ñŽºÐiäà¾ÁÓ+f£ZtË…¨±æ{¶±Å(½ cýqKK±öèHG³q+?5F¶·%-’Gw¢¯.§ÛSe9>¬Sïmx.P(ŸI ¿0-è’üµÀmÞ žÐ0ÝBÿ¬ÿI»h/¹Œ°ÅÀ;Ëà+Bpû÷}nA‚ìS†ÎžoòÙòVy¸C,ü#¦¤c¤€ˆÜ¯E‰G|0« 0{[åú妄ÙÅo-?hÙÍs£¾®Eõð)tISg‡› ´•Ó%š&ÚÝ¢i”§Œ=r;ˆJY…\£,"•Xl6y{xHµ=<2ܳtÁ»”bVïÅ>•‚(m÷™æÙþáÁß²ѨŸ?âsàÍ’5v0bž«þä¤ï¼“tü9^ÀŒhX(u×èm>¾¬l¶lØ>žþœ7¡[m{–e¹¿¯5Ù kÓÆaÿÊâVW-¯BÏ–˜LÞÂÎòlð¾øÇm…mn¡š—[ ˜„Øë=³¥ !µµÅ«›¢I¨¨ŽyeÛYmJÍDÝ»Y7™½MðH]Éu$Yˆ&y}ù6¸oÅ¢Ó°7Î’Ò:ÔQ}Køy¸½iµ{ëÝÔ¬Óɹšµ¿Ä×ıÕKxÒ` )ºÔ¶ä9 .c¢ÙÕ*«ì‰OÇ–s€½Ô¿àrÇ(·Å¬ªµ2Ë_Àó%íã‰ÕÍRÞë˜ÜË‚]õ·Êf=Ô¡B"¾Š:»në¬ãƒšã¥DT±ó¶6BfÓñŒwV©H»¼ ­S•ÒEÎܼò’)N¾´†]})”¹.yE­Úùì€z¶PÁôm¸ÀËo޾m/ëoØtK±¨‡1âj³Ý'(-Ͳ-_šW¦»~è}p6Û!V\~èPîžÓ¡ ½V‰â½ùäŸùaÔsh»_$T"û€Ïš@Œ‘»DÔâп'ÈÒÖ¬EEÿ~i?_â dž§I`÷!^׿ª.$Öýe» ˜vÁ •ßT>/N@š“ÉôŽØÝ}æñÁ–6¿Z>F(Åp›ck»ØàËóòˆ)~¼àÝ~:.÷}B ã¯uÇ›O±'=íû½…WýÔðC`\¢³#Ó̫̄c6®h™{ˆ‚æÜ¯›ï2…7£·’É"~põ.XÞTÖ;úÐ=¾x‹éÎÛÓ¸ù•¶žb„/Sˆ ÔYâ+z®?‰r+­ z‘ÜvšpŽéó:Õ ×uRHR²â7á Ô_ν÷¹¨‚Eþ"õþĺ›ì<)?ß‘ ’‹HTR¹CµÚ¹_ï}Ž?\={ÄÚåF‡Eë£ ¾~˜´Â¶pÞMzïøÆA«¬c½º+HfWÿ ŽäÝà)Îøƒ÷`^ó±ndÒÈh|iq/ui)Gû yÕ'컽>1¡ñÕ´ûÊõCiÐTŠ!íóm*m+qâÍÉ—[¥‹*gXi&„2…ˆê¤2—sï\°¹SÉ'~ îû¸ žgrì§]UÜ$Ûs=R·ÛŒÂ{a§¹ sà’«Þ|ô®ÄÚO/S*£x’îT_%pŒcó ]:ˆ|‰ó•Õ6Ðç&èu.]Þ^Ý: ‰¨$mf‹*ïðPÓ'Us•ýqÿË=Šô«á«ÕŽàª  xL'ªs§…È9Τ©m-” "+|*´¦:_õ´nh‰ÞüŧJÏÐôÛËñZã•G2ê┬W½…ÒY„Ø ›© Iò^s¸?$nýÐs®ßaôÂAŸ WW?E™(Pu NRìX`8RG¿;®…Õù?;³ drQˆ‰–{ðÀÿ±¸uñCû9™ã/·AnÔb¨ò‡qn<ÌŠçCÑmÚ{>Z%RÆžl_çïôîéühX"3Þ*Öš¾t`†)wK+r“2›Df|ã!vzò‘A%ÍÑçÛ­Ái ‘‡©¢»„‹ìÛ†NùŠ/üÞ‚#‹!«3–‡ùˆ—BO%G:>Ä[ÓÄË8ÊR<ŸÍ“qû‘þr8on5CºµÛñˆÿŃÛK3,½ŒÇi´dóióçÞÞ¨†;Ù[Ùuyr‰µSƒ‚B ²%æa«HS›l±8Á¸UqßvjwÈg1}\"Ÿ¹ HT°ŠÒIÖŽÕT/nb¯Ö]>HðT"øHäí‰æ<à· ài|¬QhUV´£dAwµœ…ì+5'Ø'dâa×,ã²GÐäø»jÓ¼þç[#îÐf¹é]ä…÷²¨Ái­·4 i“ùd·õuUYßNH6ñ™85Ü­d€&ú3té³¹63œjÛµƒú0 7 }g¼¢I Þïoá‹_-•×2¯~eQsv<\q±Ë´Õh2Ò^T²à’P?F¶Nædûš&l2L‡(cϪ*ïû¨Ä vÔ<"dÖŠÔ^j5 7x1nå@²œ¶}Ò½ ô¾*Š‹Ö#% áGúˆî îˆñgùl‘¦|½~~FÏõKïs¾šÛ6”{»!¯|€Åµê?æyÖƒj{x¤†<Ú°19J…O¹ßCa¹Í_… шäy}ñ~ä¹,ŒMÆòN[¬nqh;‹¥ÆÁÊýIwò­VXÅÑZþÒoÏkék!Ϻ@Îe ›–â|Ôò53à$wpß•}ŽèÈ ããö˜ÞÄÀK—yʆ_âY"¶/ïÀ#'«­§ÔÏÕ“~|ÿh“ìƒ>gð¬#ÁÔg*¶2Õ˜ˆq¦»<‘hâË(¸ŽôàÀnÁï_çÔ<7å7AeuLʦp°Mæò»ü=Ö5JVJ±æö+ëµ®L~ôŸYyÀnØÍÀ½Òpßæ£ø(ÄIÓêdxטº¶ÕÂý$%àR@éòs{Þ–4¼žˆ—2¢Èî[ =â”{‹e-ë…³âoaÄTƒµÑRH"­ƒ)oìâ†ô.Äù5k4¦*ÎRÑ öÐåÛ$Î"ãF Õ"4açâÚ¶ºŽ(q¯[·©± ÿ%^%Hô)7f–þÛÐ$õkÑwhÿI&Bîý·®ŠvXÎTú®Ë6?©N˜¶»ÁAÓÔÖ(—?lt qp9ó}Ú­ bа*F߃Û'E~Ó Ïò&3è%ÖŽ!ãT®ÜˆÝºçUªÏn&í?ˆN„¬©¶N/›Ö(êg,0/õô¾ö¹Vk‚WÌ=.íŠ÷& /~»>>,ñú\£r·¡»TÊ'Q¸˜;sùwLmâg† ³ïÜä/óko' =\h·ïí¦Ü¤Zñ«°Â•ÍÐ+{&¼g!~ÄÙ¼ñæ-zîÞŸ«Ì¥˜¡kwXÐ+»½`›r›«6PŠž úåõ+M—¤»ÞïÂàA¨óƽnŠÞšFÉ‹åšÊïJ”_µ6’BÓ8ù…}–î½]ã’IXÈ“W?JÂéu´wºDÊV6v<ú„/³±kߢîÎÓ“¡=ˆU¸çØ~„¬šw9µ6½@l e=vçaÇzyÖàÓâézƾV—¶ðfG Ûzšhc½ALÄ-‰§¤¶ºâò!Âá²ŽÉø*Kùž¸ËÜÅd'óΤ’®,¨¸“2”ävÔ®m¾—+ùØ}"Æs¥Œq)ˆÌî36u ܉’ãzñœD€ßgU ónÍ—æ*Ë~CÜçZ"4y1K3báVå8–RSP®Ž]´µ9 ðhå@ó¨kç:,Ï®w°›kc1ׇ®÷ÚßïŒ?‹6_qÜVœQHà %æCÛíÜ?_iâVšÊ°>´ƒˆÊމ¹ …l¢`ü®j®Ôh/<…•£yéúž¡äû™ØöÍx½ò+@¥Âãøž1sRãzÅ“Õ9ží«\¢ zØVruñµ Cé8™±v¨WN{ò÷‹6ÆàÂ2áÛ\›uÏìG°Ë–,é*¾ï®ú·Ìî@´èØ·%÷ÅCï¡ãÙWAþnjÉŠ}»B•Ñõ Ÿˆ¯*œ¤nó†(u[\6?:UˆZ˜y™Y¾Wš°oýÁyž Kí»¸-ar.ÜÉì~LªtmÆÌog6&žG‡ËƒC7ÓjItn«Ó±ë@Å_á]äÀ°ºt˜Äh‡˜Ä[d‚M)Wˆ!¬Ý¹‰Kýi$ç„kœ\Ò‹[Q¿I»WÓÈziÞôÌž? íÓnù$ß3ÉTg«AólËCÌÃJF¤iã)d»e>ß¾—DѪz™í ‚nÍcN>Ö£m§¥G‘óN'¾X)«uD%V"=.±ðÜ”;ÀœYW;¶YÇh–Í<–mÓ³y–¾H¨‘&)­vý̧M¶ÎS"Ê-%NyøŽÑGÐ ~‰:´LP$šæ‰7e÷] KLZú& gçÛÛZÚjxÉ6¬ªš}ºéÃåRµ̧K×S‰,ûPzñ\°ÖSw™%SVbÒFýÕJùóDsWîÕ7Wí$c¸Î~$QFœ°îVåz›U©ÖwcŒvX‡ÛʾV= –k󡞬™œ¤ÀŽ= ECMµ7„zÅed·P“1¶¶rúƒ¤â±(Ōԑ§4Ùaýv­S'áçšÔÅ;iû9±÷2ÉXë“7çÇJ¢ŠuÂä…£Û!Å‹|¬ÿVA[wp‘±7%g/Ûß=5V¹| 8º%“M3pÿT‘Ôvy´Zì ¹ãÄðÒq ßËœc‘”tLœq@Ó£yÉ=Å»J´Ú˜ãõC|/•g‚±odG6]ÄØ£‘N2ºîë¤j__Zïœ3/1¤ÄI*ãÍÙ—WíËÕ'õE5³=‡XÓo Dðå·²ö‹tà«­UÐpM^Wzû*Ãlh²9iç=8žqœy…šÈC\¿Q·×suµ‘íõ ZëÙQ»dõº‰eÝýIJü¯n“Ž”@nb¯‘ùöd ðÔôÚ>M,õóÒ`£› €èZ÷sõ“Þ53ïv«ž½ñaÅDÃØ»™øW]­*¿^¾4¿w}«PèrOÙ—Eǽ[m…ÓôYeçú“îÜr+×/ÏNgëö-œp¹Ó.c‚V„ñEOºæÅ¹‰`¯÷¼[íé>`n»1šSÓÚ4Ä\÷,žœöÙñb:ÒÓ¼È)òRyuvXª%îc-LÊ’˜ëT%W˾-ÔÙVG;û¶ÿÖFtý-yÒpÝñà‚ðª7W‡,­æ6ü8Ë [vc]%ÅX‰RÚµî®?Ú+ßPž¼Æ£tŠ;)­Ã}ve3ú|Þ^OTÅ2¢æ]?uØ7fQp«rµ. ¨Èq}vK6s,öZWA endstream endobj 1388 0 obj << /Type /FontDescriptor /FontName /CGOAVK+CMITT10 /Flags 4 /FontBBox [11 -233 669 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle -14 /StemV 69 /XHeight 431 /CharSet (/C/I/P/S/U/X/a/b/c/d/e/f/five/four/g/h/hyphen/i/k/l/m/n/o/one/p/period/r/s/t/two/u/v/w/x/y/zero) /FontFile 1387 0 R >> endobj 1389 0 obj << /Length1 2187 /Length2 16524 /Length3 0 /Length 17839 /Filter /FlateDecode >> stream xÚŒ÷tk× Ç6Ûp…mÛ¶½bÛ¶ÛVãFÓXm›q²÷‹î÷ûÿ1ÎkŒ¬çš¾æ=çýd‘)(Ó šØÅìlié¸Â²²’Œ f:&822 gkàäpdj@G' ;[®X; ?d"†Î†²v¶)k#3€‘‹‘‹ÀÄÀÀùC;G.€ˆ¡«… @– eg t‚#¶³÷p´03wþÈóŸG…1%€‘““æow€  ÐÑÂØÐ kèl´ùÈhlh P¶3¶:{üO sgg{.zz777:C':;G3>J€›…³9@ èttšþ¢ 3´þ›@ÅÜÂé_ e;Sg7CG àC`ma ´uúpq±5:>²”%eòö@ÛËüË€ðïæéÿîßÞ²°ýÛÙÐØØÎÆÞÐÖÃÂÖ `ja È‹ÉÐ9»;Ó mMþ24´v²ûð7t5´°64ú0ø»tC€˜ "Àðƒá¿ù9;ZØ;;Ñ9YXÿÅ‘þ¯0mµ5¶³±Ú:;ÁýUŸˆ…#Ðø£ïôÿ>\+[;7[¯ÿ S [Ó¿h˜¸ØÓ«ÚZ8¸%Eþmó!‚û#3:X8Ø™9X@ÐÝØœþ¯*öÀ¿•Œ‰?8øxÙÛÙL?h},L_p^N†®@€³£ ÐÇ럊ÿEpŒŒ cg€ÐÌÂîOô1Ðô_øãü-ÜÚ ãÇ`øëóß'Ý 3±³µöøcþ÷Ó«Ê(*«Pÿ›ò•BBvî/Zf&-+€‘…Àþñàó¿qþÛÿ°ÿ[ª`hñïêþQÒÖÔðWŠ¿X|´ï?L\ÿ=ÿÞJÀÿ¦³ûh €âÏüë0°2üaüÿ¼»üÿþ¿¢ü¿Îÿÿ­HÌÅÚúo=Å¿ þô†6Öÿ¶øhçåµûXÛÿkªü×FËM,\lþ¯VÒÙðcImÍ>–‘…Žå_r '1 w ‰‚…³±ù¿†é?§ñ‘ÃÚ¨`çdñ×ÅóáÅÀðt»glõq¹8}œÙ¿T†N‹èü÷ñþ…«ö¿uˆÚÛ™üµ“L¬lCGGC¸‘ø@¬/Æå5ºÿ=óz:[;çÀg€©#Ü_ÍÆ  üKô/Ä þƒ8ô"ÿEì z±?ˆ @/ñ1è%ÿ ½ÔÄ —ùƒ>bÊýAœzùÿ"Ž ÐGLå?è#¦ÊôQµêô‘Aíúà þ}äÓøƒ>òiþq~Xþ1~02:ÿp~ÿ±þ…>.¨8|½ÉøQ¡ ÐÚùŸŒzàøQ4ÐþãÚ³³eü#ü¨ø?>e˜þ~tÁìω|(Íþz} 퓸ÿ€ݰüüH`õøÑë?ñ>Xÿ5nô4mþ ™>òÙ¸üÑ~\ô¶ÿ€jû?Á>`oèèlahmbaú‡ëGJû Àî½ú«Yÿ ðWkþÁôQ²“…Ù?ZÍöÑH'kC'ó¸|Ðtþüðùg¡4Üþq´u»ÿ~„óøü áù7üŸÕ2vqtüؽ¿ïĽûþûÝºá–æíŒ¹ƒ-ëƒÛjqÝh÷&x /Ò4˜h'Šô`œûE§õ·”³³–¥+Ä–~2ŠéYvÉ 9<ä¬/Üxí66z°ÜÑŠšÅÍ¿ßÎ&zÝãÏ#·€ä«§p•:öƒ(àG¡tÃò›˜õ+“!{©ï[Lt#Ç(c— i—ëì®-Ë’ÁÂãT;ÝVrn5RÙc½ýô%wÛækÐN´NDª\±4Oç#\‚[ ÊæàÈÐ*êOüž¨)×À½KÒPgTQî!<èµÕ¤—ÜRד 4/#qt“þÊ Ê”‡z¹kóÝë]"Фnvsƒ$ÓhÆSÏG|TûU ¡4ǼÂýH“•œ0BðæA¾Ä{'µ¾t ‰ÞõoÛøVRa!Ò˳}S›/.Ò©’Ú“Á4² ¼oU»§>ßÉ:b\MCÁÚ¾Ô¦ îŠRceQû˜ Zg¯ÞQ„̪3is¤™·«ð•¬r·ùk 23#ºZ¤U«"Ñ?„.Ö¼óòdy+¸éì"_v¨‰²A‹öÜlXõÎ(ʺ`¤n4N¯‹öt³Ùºó®qßÒ•´±+öùæÙnÃz¶¹€mÇ Ì)?îŠ ŽùD×Aümeçgøo‚*°82¹¼Ê§Û¬ñ»vf8ײ€è§Ï×° €œ}£'d…&·Út—Û`A?<ßÍû_[ª¼¿‹ ÃØ«sÓ|+ÆW'uVÛ.'êVˆ…ËÞ¤lÒ®¼²¹-ú E,b †âPÐQÂ×èvÁÍÛÖÜ(Ê™:ÜÙ3„ö2Þ:Îòƒ¬“[f3\œ›,^$BíN}Âw£w;}ÌK!<µ¢§ÀQè¦kub³Ï`»MÙ™½›†×GvZbNòñ”ÁIEd&L×­nÑr?¾ÍHðýñù} ³’߯™ŒÔ~ÖXÇh©UdV†ÕÅhÔÙš%UaèmÐiSjäGÙã¦ta¬MÎ`™tY¯Âîñ`‡žfþY¹9¤€ÒÏÍþvúØt^`ø<¯×kÍâ©vtsh«t³Ù…¯¨i óôN‰OV· k棌LÏ UŽÝpbžüúÕ*F/Ú(!ãÒs´Æ~4„E_jçmõy`â‰ñ‚zjuùTý{=Óˆ½]–ÈsVex§;߈é‚DÓ6¡®Šè]e†ÒP¤/=õ3üî.Ö1 # eÍšc×Sc5Á¾þ673[‚Hí~5À¤‰¨ÉŸÛ„¥ P¥y‰•‹fçɃÄXú+Õê<7µç$&´Ft š'¢»HqfíŸÏÐaÀˆm1Fò©uB…<„økî>ž7üØ´µ‰çÒRú%v†ÔªÅúnLÿÖ†T%s‡ögPÞ¶•¡Ì¨ÎêcšÊ¤£×¯t Û¼ˆØ¹±e/#•ޱvS¨Ðí³¼o^2Ó'äáÝÝp#Ùœ;#%›O÷çß7øøЗwUò‡ìΕÜ2f¶¸àÂ2­pô×°m‡øx=Ÿ°<žÏZ|‚óœ÷—5 ‚ÚˆÙý¯œƒ7MŸ|ºÑö§’(kC9 ”•f4’=°Ñx²œ™,“åõÚ·‚ÚPȺ¾ªf·…lzñÍÆÐçUÕ” ÛØmì1LUùäÛ¯`›x™c±ðE¢EB빉”\ oM~‚Z˜hNém t}燱G°`æÆ«êoÄ>? ÞþFŸ"þ&m|´¯wË÷2ü¢b7Å[D…‡Þ2<úäQä¿DbEš®:gXâ_ïŸÍªÈ¤ÅÌõ)»wˆGS|Ê‹ñáÝXMòs-8…¾£,ñC„sˆ‰¿ÐZ…ÜI(“®±…áÖËÍ;ïþ¸cúW•s=ª é¥:oÛEÍ7ôï8²'…Ì4TßFº²úÐUs–`׿³¶7¢"52|âÀÙÊ™|¤ƒÎõÕj±&~DC¯ŒÑ—²uûŽF0\[ %„¼YŸó¹%¬kólKhÆÔ@¸»¢u ûøïŠÑ’fgaQ|÷´QT2rænD:²,jù™m¿+÷ùŽåÛ ËcúíFÃ껥 U¦Ùt˜bE]=,EÉ÷h@­"¢*a°°½Ñá¸÷CŸ0'ðˆ\L›" ®à%ä1âSŒ#HäYZ{›þmu4ê´©<Œúú/°²Ü2—G†µûø+ú„jòÖßÃÓeßçaêh7d4Õ´*ŒNY4r´Ç\Æ,±øDg­êýë‰4ú»eâ:›òãUSkï2w²aÖO*xC¶Óo™?ªï/N!‚«Y‘v!̄޺„€ÝôÝ}\ÏÖ€5¹jòdœ$ñ9UFï¡tôœ`܃S’ Ç“bžI¤%þª‹ôÒ`àÖ ¦lî›yø°põn>uô TdO+FÉV‰5/šuv,«ÏSÝ ë•;ÒÀu<¾SÚbó·å¨J5}&x„¢&µ2 ÉÓÆò\3M«óGK]0•¤ZÓ6ýÌ &‹†ÅýtTàÙ‘²÷Õ±ùxsš´X¨^Œ>2Eë7`~}Ü_7gz‚Ñ…A×É .á6Ù;£-‘¢‹%w£”âèR¨þ•ÕªýGkÎ@ŠUñÖíói‚j›¨5uý"Á/û ÝZm3Ýh04¤J@X›Ymðo1Ý1ö¾£—Õ—ðP«‰‚¯6‚žK¿² L:帉wÓ}?AM%”uÚo´ß¥b¶|/AòO”û•[ûsÛ }ΆÒ-¸q?;h`°ÛiÉE† ¢ÕðÕÿýMµ„Œê‰ Wgñ¦Ih+q9žgäxO!šÔÜò›÷À‘[ŸsµEÙºEu‘TÅlNóË, Cµ¸ôÒéP³-Û2ħN§×YBøúbä‘ NdkðcÖËøs!¼Š'¿·óxòÕ»—Î+Õ¸6Ìú!¶0œ¢kÙO&ú|¯oI«åÍž@~…bXŠÜH*Z‘ &cB¹½^Ä Y弡´g‘iMÊ@ænEô*œÎåÏyt ?AûÃÇÊWfJàପqTœ÷§^¥ùâ1¥UMf«,‡ÑLüƒ€“öƒ»ø;]|( òÉ'ùûÐ ­¨sËš1 •Õö_ˆ4¤†A8J+µON¦I…E‹:6þÆVŒvuÔû~0acûþ=¢ÇaFQ4ÂÕ7}䡯è'’…™³iŽ:+M½ÃïÛÏJ@;ׄ­6Gl{þÓÉÙêš,Ó9á L¨ºòùVuÈ;ûÐÙ+¨‚ ‰!Á[gµRŽ´ÿÚ!x f^êèÒn)5žøfÕÞ:&.j>K~È(D߬Æ5HÎÞàŒ"J³ÓI4TIÕUr†?Eöìªù‚CÕs¯ÊJ+:…³’h÷CÅh{­‹u·¦¹ë¥ƒ)©“¯òTÚݹ¯xsXß¹ÿ[æQÜ/ï)õ²p†Â1Þ}Ñxš¶$º &½Mdžb®\ | ¾*àá²ÂØZÃ,wÍýâàB~°zÖíÕwÙ-Á O''#Ÿ¸œè¥×>§ ýD§vz°Ôœ±ð®`dùÑÞÊTëó¨usq#Ð;CŽ¢ŒfRŠõë!=Ufô‡ŒÆÙß;7ÂÄ•ï®]Q-¿-ßPAbøÓrŸVGñgš÷YéÊœ¸:uq¹%˜ªC:êù­»ë1Ðvß@Šïo‰€0…+Ç´ÎŒ±‹ô8ôƒñF<¡É‰«¯VÏô{^PGSIv™yèÁ_ÇÀwN‹9ÄÉP#ì'Ú‡Oše¯¸f’Ÿl_{I•ýOpY‰*rÚ’#·—BˆRš|wE`gH8¸üò¯ßoœ ðsõŽkyÍ„/ âÖøO<¡¹côuÛ}¦‰Äšzçø¾›C6árvq³ØÖ´Ça\.äG!^NÒßÐWƒ'ŸH3©šÙâ? ™;"ÿ& ‚%VqŽ"ùO«÷Ó•™øÞ‹¿1ˆ¦mMÀ‚­ 0î6Ǹ[ÔFªU8 ĵÔT&}C™·?0+’uóÐݘ×ÉZubr°9ãÏ¡[,–nzºãöúý[¥•%%´;·÷k©\ÁÅ,Ê&T§ÕÉæO“M÷IsQÞÁ ÒðbÝ]G€JB·bÛ¹dàóY&ÓÚçjÄâ0ˆô·ã~mØÚT¸¹c—WVU¥«Mð¸ÿH¾DàVŽt…{·k(Ìæ©uôSy®OÛx¹vÄöÉ9SåŽ]e£€ˆ¤×á$Àës;[Í®Óff³à“•–6AŠ•=^¤FxX;«Ç”Ï¥¨u6­#£X}¶ŸY‘ f"‡L£µ„—¾ð­0Ú¹e&âu½Q'«',”·uX‹Á»æ7er‘eíð‚Iw­+]ˆ›-}˜ÄЀÊ_:mϘr¯“v„‰j à«üNÈ+½öÎ/ú¸/›™IÞ@}Qè¨CÊKª½Mï>oÏÜ r²­©š_åù Ì ÏP~/ÿ-Q•c`°ß.ŽÊÞ ïQ´øÎÚªÔHÀV‘úfŒ[)>Ú6'!‚Éc2´+x©É‡àÇ¡ôÑ¢´šJI%ÎVtN²½(W§ÛHvΖ¨1|bn_U¯äR©AƒËá$ J%÷‹õ„£8ðŸî°¬StËæe$0m"Ÿs—†ÀŒ¿ûö´bI#ì„¶\s9@` ™{H¯5¾‰qóh¾çþ yñíó¾®lÞ€EE¿U¡wìz-!¬¹üitÜ>XØø¼k]EfíyËaèët(¢ ,Ç"­Ó´v½Ï/Jr¥d;}ŸL|¼eE²–{›  œ÷ä-Où~+¡ …áC“4íªÊ/ÂCé?šr¡[”ùZtžüã2¡j{ýHü× jO"$Hs† 顤ìÝ–GH ½,¾wmsI.¯Î$d¼3"ú‡ÕEÁ!ÑcrVn+(ÜNî§pjn^ÄLñ??OveG{òÐzÒ«ß»ã;0ßKÒ¢ eýÃ{S°1HFêÞÏi·F5ÛéðÎ2aSu ©#áÞ“3Çò”¿öPÄ’aSäsÁ yÝë‹Ûú!£–Kô÷ûyßQ^©Sk¦å9÷B}Æ>åëÀoyè!ÓðºëG÷TªBþøÚ½‘Ï&f)´¼ÃÁÂ|°'ù#$O;öÖC$i© ÚF½DŸ$Á(Êð=ÅÛUÒ`ù[Ÿ…¡¡·á€S-T¶Š=1ó ù¾‡@ÇÖöL =TXwŠýÝWç)xwøä›ßÔcXTìM“úŠ<·Ÿ]äý[ƒT@V¬æXSM4œžR$bë‚L¬ÔíÙ÷<”¦y ¶Üáyoi³\Ã!^Ä'1|j5[Ï=Ógok&i)-À®lÔæ7(ªd2RvÔÍ¢ÊAß'xéë»ÈÅT|Du¯Ú™ó7J•ÉðÈe[ Â^ÉáëG’6½9ÐeÜËAˆÙDXßßÑb.õÜ\Ò“‡W@aón´G…ÅëíKc<È€ÖÏ`Ý|QÛ8îÄ£ÖæR5È8tŸOHEøUìsÐu*mYßJÃ#ªŸ§kÐPñjeƒI8Ѝ_âóyFà"Ô@iуŸL+?CãíÜβ²0H »P%£«”ǨÞÞÞ¹¾y}ºÖYE]Þ oú’…NÞÔ=Áýõ*öA£ås^Õ¨‡´8†þ9CQÀ¼Ý­g‰zÐ娍× ”z;õ5DªtC{×˼ „ÛA¹“C˜JU§y ›ãòñµL"9—ºfäÏöôxÀŽYEÑw¾èˆ00ièLDgæ9se9.ŽüB¬/fG%0ªüë”»³JFdSÏËvÑÍþi‡|‘lˆ;?})Ó`åà*Ë5ÜXÇé‰G·¼­ÒË(€ŸIé?­KUvŠÃ·l& þîÉÁcýmGbn?¥rDöóz”V)‚¯˜jâ vºìtˆ‰5ܼýð¢Í[’ŽÕ¥Ç@Q¡ð~>Òઠ«8•ÿ¸’˜lâfŸÿÛܘÙNì±}·éò˜-øH'ë#²IUQkü™ôÏ4m8ØF¡î&Ò>/ŸÃGØŽˆÔO6á&âžD1›$[/Øhž%UzHÃbÞ&<2—ÂÎA%Rì-Z÷ßÕŒÍst?Y“¯lä‹õ­qµö¢È‘’ E‰p•D¡‚„Q‚Ac~ÚÉ”uO|$Ö—I}âÇë5A£zâ ›g¸`Òew§8mc±‡¯Vàh§Â£ƒpeÅÏq5âùÛ +£ÊpAõÐ$}–¼@Ý,À¤£]¶«˜ùzø&ü¹ú32DJþ€¿õÊÅÀ”Žïg%jLR×>»‘âôe³Ð ˜7€Çã¼CCôoÛÜDJ>Æ”üZ‚òBhýGos×(ÖzbLg’/ס{­t â¢ç0ÊR“ê£Éc8rˆ‚ZÛSŽ!Û;DÃvg-X§EĹA¯bÓ}´×”y¤zšH[uÞ_J催HòfÇ㶆žTUL“À7¶¢UhÝÙ@‚,ûcòähPI)[ÞúiOÌ©÷ #¸±Fæù//^j#JäfvHÍ’†iÜ+¯»S}‹ÃÛ¶’ú[Ðé=ðãޤ½ß§;ókž–LðoÅOæ³»Ã[ªÞqÔ ( Ïtñ &ûÁjpöƼ¬u’ÀtüRü˜²¾™ï¦îùZhu¸¯ :ͺ}E0_ŸýYfé±”Ò%HíΜÑ]ù8t» +niw:'$bªÞý§8 qU{Uòiè 0™¸äº©{ ž± ¥òë¬^ò}9D °_ňÀ§" ¬ûžƒ86mŸk¯»8cA9¹ÒfÖZÙ”%.\¿¾Ã&®6>`”JUÑIçç—¶„-o)è,± £3ód<|’æÄÙð@k]9•ñ£5_AœºXÄfÕ•sœ_L±t•’óëMǤŽ0ù…wtDß„T ZJÞ@ëcšÂ3~§’¨À.ˆìWéÅ5qŒœF±&:AVÚÙ›27%Åý0ôIÈN¹ŸXµ]ýeÏuÀ’¯þngkÅü”Ä^zfÔBŠøþžÔÁQ#kSÕ³q3}I¹K)m—ï}î·Ô:ž g†sß+42€(ËÚ€pe~¬èF¨§¡‹?üÙ®Âe÷„býq›Oª¤8~Øcøå÷ËʪÜÖ[Љä¦ŽjÖ¶þ+õ’|f ÏU²þIêÁd¾a-£mðEG'R~1‰[RÔƒª‰³è•oèÕõp]ç„™ˆ=â&èIfRºñlïR{ÚX…>í£qü!Î˜Šˆzh–ŽJ8¬Â{ô›Eøk¤“˜Ž·95oaTD6›è‹!•u‹ ½¿0ÏÝà,âÞ[ÛtÄÈgp2$¹Š$Hv´]ú¸)çB×!`Ù·z•‹ÁÐõqz6ɸ~„*=L%¨õ[zS-‚ÚfÐ)i“§Y¤£V¯7‚B_%Mšqm`±µ¦dÇÿåcArµÙzwŸ+9ÑÉwÙ‘fÂÔl¢æÛ¯B8ÓÒË}¨©³Ÿú$äX9‚Òùò«—Ö%yªÙ¹^KW'´AHþé!ùkEùÛIÙs¯ñÞD¢Æ.¿s>ùÓÀî‡sÖ§Læ¯`-d T\BŸÃ–šAó°Üí9¦ìAµ5í ¼N-z¹e£ ÈVXøW©Å_©À˜ Ñ(xî ë´ùM£Ë²’j Ý£þœÖl° dÎZ´“xŠPÕ¤ÒXlÚñTìR~2…(íܱu­ëhGŒ4ßWNßËKáRNUkòÞ/´Ã5WÙ»­r«“µiLô@lõAßúà”aMç—碣l°b[bE/;4»ÒÂÚô—R…ä™11L݃„—í*/ˆ©+®à^9¦@Lv¹4"]*#ÌT ¾{fÛ—ï¼'_+¢.kÄ·c~dà)Å´ƒ÷ǰå>!íá–Å㜂H‰žO;HFÔй¹¾ÓÓ‰JRãþLG`y».÷å¶ ]w”®ˆ³ˆÂ2Ø£Ž b¸5¹zÕ68/Žj¢@jWm±õéÊ¿mqR%-ñ:ǯ ƒª™bÙù«Hì­Ha †ÒùþÁ4ÍøEÈ7ø[¦hvqÉç%÷ÍÒ7‹[‚Rn8Ò«8¥qú"Žp^¼a¾X‘ýù&ï¼…/VÔ ÇH”KNrŸgïBEœ¾~ü6‘h/6såe/AOàUµC~ú±%ÅŸ§LÒü10|ÓFBËà .5ûOèVÍÇ\¼¦o* ksw@ÂÕ“]vpNÞá/Ç„6äQKíXµRü4$ÕªuTB£Î|ÝHÅŠ|tNƒð]uŸ#ª`Ƥ€-…q™gR’®]‚¨Ît>ƒ8™%S´ ‹Þq‹š-ŠÜ‰Ú¥écCU`1aµ ÙçY[Ý/RRܰ…íV\ƒg'rb±Ô‚.Ú´÷¦,,ÇG;ù}Ø8|P;¡û}†ÉbAnËJ¶ûé:0’œu3D˜ÝoÎ:¼½Â#Ë•lí«Îv Ç\—Îu1Ò6Õ>÷ÆËb•où†êÖãÚƒÑé(>!<Š&”ÆŽlP ,‹cq ¨ÚS¿ŠÄE¡UÅW›CPâíê4s짺&؇R³ºýz©h^Rp=“­«š°˜«ž !­{â¼ká¿gj  Geö3(ëÿ¬¬š~Ò²,S‰ %×sµÃá<Ü”¼°OÝØÊ½#7 P=~ Æ›ä…H'¦Ÿ¦~`]þöë[®+Š`¾bŽ*<(d¸i«.|69%åŒÐt'iÛTƒ'Î/hÐÑqÉ3so¦$„*ʘÜBCÍ(I×u\H0Û’€RÒÅïÇõ —Üg-¦uBM uk9ſ–ÝÊ}]•ÓíÂØ3ôã /Úëó×/GÌå2<; ÞŽšcÞñe`®"ëUb¯:/#8´¶îº€`äêdF”+Š8X‚¯2“P$AnNû±Ú¤YŠ• wÜr›y<5¼–t2=:¨Xjšš„D«bÐZÈɪ{ ç‹õo-–Mz'4žõ.ÉÏU”$˜sâ%”K‡2 Òö£° êîE(;mb]¼†“W„„hkföXiOÒd UK³ÈVY\ûÖU=Ž$*;€)ÞÚJs‡?ÞBÙAðΙÁ˃$›}œª üR"âË#P+ûTÂÓÓZ“©fœW€¶L*ÌÛPêBy’ëµiMå÷WžÈùõÙô²Òõýsk÷üT$¿rÍÓ1» p#OQ ¤>[ÿy¨Á w€*}™³ÿÂç»eгÖOšNU jBް>}ÙÚ”1MÞ¸,NùO€ÝjLZV¥–=4R·Ž$.S€¯ï‘J¹›Éf%î¿5o{nxó-ÔºÓéŽXÐŽ K4gx·Šêé=_¨µ~~MVDˆáv£Eÿ}üÈY34LˆÜöM×MM<±è޽ndXí0<&²;]̘¡ l[¯­‡µÁãv»è;/׌ÕCL ZºéÑBp:í!­YÕd[çÈ‘§±lÎ*bËv„æüùÚyŠœÔŒ"göKb¿o¹,´‘Ų°î•Yçg2»Vº þÓEÎm&Õ¬µÅ!÷_H½s‰üóK¨R9§r×°&Ž`ï''ñz]ò}S)¿Pu¤_·!T¼ÛFÃï‹ß7#“:t…z'‘+²8u0ÎãÒ“ß Â˜!;ÂÙºŽVÔ;¾Å¦-èÿ®òåÇWúz?«®Þb±õUQÈ/|wÓjò¦i5–díÀ·2-´÷¤›9Yîæßñn;•YÍ¢½“”ÙjãÊUò5[®ù”É3z+0\èüõÏcVÌçã\`§‘¹|lsw¤.CŸÉ Ç“O:F8C<öŽ¿üD˜±T´áëxÔù³n}ÁÚtÐiÐßê-t@ÛyAæ|:)©<[¤l»²!Fb¹þüÆ Ò?5.f‹¾‡ÅÝ¥º%›À¾Õ*à-˜zOÈ]nü‘[­nåv0Ì[©"d>0[2Vt8 ¯4³|àÓ¢·¸Ä±Xä»×&\„1ù³iš‘Ø^|YîïåË̱¬¦€˜€B3ùõ’õ­²&Qå[öu>ä•’­¢*ì_'­Z“Þ3si’¦@n·W¶#^éL8·Õ«ï!ßqÂuBP­6ÞÉ·Ðv‹)ÐW¥¢ká‡BÖ×FšC“4ÕEpH &Õk¦#,gkˆÕPo`v ÷úÅ$Þ¢ŸºQ/ìÏO9ò ŸÂ,DüEüGzüwÇ g¤‚á ´UQÀï½Áˆ¥›Û)º­d—,G:½Ìc‚N£á]?§qDDÝeúÃ`ß’/¥ûë ¼¶-ÝaП’Xo‚‡‡×Á&Øñ­ÈN¦g“V“l¡g&‰)¹öÄ¥<Œ^¼™MÛ £°U‰{5=x¸Œ<9½œ´’¢þþ.ü©¨Ié Ub±%fy÷,šñ5¡S•¶‚zΪ*Õ&=ò| öÙŒ³u:y Н§Wù'âE+Zm¡tÁT- œŸÙãk*T/}‡Û§û-E³ëÝíg ¯ÔŠ|'ÙÍ b& C-¢ì lßæ(\AéÓoï08WâŸ+Ø­ŸQÇénÉóS8vé¯Ï㟨àô± {DªH-/gDC×á:^˜Ü©ç( !ÀÙdÐnõ†MPeÌ|­@Õ;81£ ŸÆ{­î'I2 ºH=·$Y‡¡ >¡²k~Fìp*9ÚöyVA É»Û ˜Æé`¯ÈyͰðøsÙ]‚/½ígüº:E‡eðRH>£™€‹©9e”g¢Àx>Fë  Js•R:¼±}ñú2õJšÇ^UȪ鹙2qŽÚs(ï2YV³Ê| Öýù³ïôï6 SCÙüä<·<Á78øÔ+è²B+…䩯ÒxPWfš<q…/ ‘CñãnXvÉNˆ04ßãqò¡›Ü‘¯óœLíÅÚ*¦àϤy0FÏm6šA:æ2ýiu2¿Qs¡ƒU—³Ãê×CìÈ×ó-ï…o¥´ØÀª[‘V7¼tÚc úKæSú©“àÊÙÜ,ćø¼ºžÎ+ÈÐÙëÛEÉ·nר·Ò°|KƒWž·Ä/ޏ‰S=ßyPyŽ FŒà÷¸Iâ÷‚ƒ¯AÛãTå² # {ˆô#V®*²7:ÌwièánÛþz?À©}!ïrÒÖ þeV‘çݘVò°ÒØSñ`U-m@K;‹µç·á÷/áf¡K`ü+– —­ˆÌÃÅØ»±ÀIâ‚iañTvÇç¦ãÖµí…å¡wšVZUóºªé±%k”—âÝF„Aˆ©zmFI£J~æyM¡Ý¨³ß½„'ýùé$B/L ™? ¿wÝF2sæ9Þ^%}-)¹¶*¡|Ò|ëÀdÇ „(7-w»ã¯×# EÂÅ ÉHÇêáoYWÎ*þA×*™°Ü 3éŽâš?ˆ0Gr>€$Á0sl1–mÔJÒuñ5i`Àa¯ÁN2£;¢®7àÝŠA cφpý¹ûþNÚz}%kâ;" }O¾¾NN6$¨’–€†+ùM÷Hư|H±C©YCõ²Å¨Šv ³Œq®j|>ø8«ä—:ÑïÅ¥)JµI¨5Ëa<º½%0Û€í©G±²…&W¢ 2Ô4Æd12ɽMÅ#êU7ªE(¬ïã5`mÊøs¨»Uï6þʵ·ñ~Ÿ*Lf«í;2 &«•R˜Ê—¢|ô±£ÅLDù=ëH‰ÓPÔbæ šÑÀn ðpÒÄ èH÷Ú8ö(`ŠGqü¢7ËŽITÅk@ÏcÐ,@`kØ}á>Šb0_±m™ƒE‚¬-]]ÖŒ*Ù&®OÙ'1˜¢,±Y #ÕOžÛ]p„º‡ÇS›£¾ sðcâ<ÿºñO»_ÉÈ%”yz–¾Å?6à«ÖÄõZŸ/ |štÐy@.ŸD9 wJ õ?\˳ãAßðƒ`Ö̆LÂmJÄš, ÕIÍ!Âi G_ybMRƦŽ©/,¿âs†qòUZ±mó Iú )/W*TYšçLcã|7Uý{ÚA¿´Â!¡‰ çƒçzüØýKvã¾Òß8Y¶éG=I–‰f. l“Ú öíñË=³$´¸Ùg¢†Ý~Xm~B€èQƒ9Žìü1|ÌU2+㺲|r  ǨS>qŒk=¢%I‰Z1-pö â| $±ƒ«Q”×Q»“œ)ù‡‘öÓ&1bë.LÔ…ÂjM„‰2zÏ(^Êl|¿À<Ɉ;Í=î<Õ›³Tký7TTõÉz•ÛsFôD”IX‰šëX,°ªbožšfð_ŒÙuUæTiõI\ð๼d=?PN‚Ÿ¶P¯‹Lä¹XüÓ“j]žnL³ M$6‡VAÇšôIÈÛ»Ž+Kàî“ZGÎv/‘ô¹Œ!ÏÊ=×{›ôÚhŠt¨K†y+cÌcëZb­ H‡©Z‡B6@ÃIC0èG3…»{*Z"ƒØ}®U‚~Ò²¢¹,©Ý¯æ“-r!ïS¿’P‘§I3UrX#Kóq•#.cÙ ¼ãX«²)˜ªŠ)2>Í5ãbòˆæÝï~}¦ÀÁ¯LL½`Qdãòðð²ë¨D`RŒ.ó9ˆw ’Rp²J€Ô¨ö4+g¢¥ÏÑ¿)„Ýav† ûÜÕžèB˜LWHî6|ê ·­}$(ÙЄéæÁ…—…lUX*Ì1¶è~së?Þ¯k%%†LŠ^ZÖ¨¼D‹<<ðBfM\GƒŸíîDz}†Ç_pôËãPæ•» º …gÄüĺ §޹tâwÀùÐý€¬K¦‘@M׉½pt1¨ï2¾Ô€ÏŽ<*hŸÎ˜üU9á*#>ðŠÅ-=2C{¬Wâ5ûmMP£*ê»[Þ‚G‚àï2.²HlàÏ|EoyËVûlûsˆuYV#$Ž.÷O2HB#Ÿ£KÆ8@1VêHmX…îÔx²XU\˜0`Úê‹;ÜõÝ­(W„â2Áf ´[û›Wjq£Rº†esOytt|³Ò3Ç·=ˆ÷=à ªašqL3žŒ@CXv NÊŠ˜%/ü`|­t`"½J “bôI>È÷![W()aÊ)£›EU’]¤Y Zô-‹ås‰£¥¼Ùß>?rf#íL`†íI`@è‚€°Ê¯²™h¬œ!ãÆÕÐS.¶.A:qDÓí2Þá'øà.9ã8> ET_X·ì1«æt‚}¼^È»Ù^¦¹ë°â7¥;hO*DsßY–š Ø$h3(ëÃ5ᬩŠÔ”+¥]‚дy¦âºMa§%cඬÎu©Ç~u©”m@ŠÎÓ§ôUøž… Ée™ýÙ¤b×ÖnW­3êË¡aj_EÐþVÏæŠw™õØâ™ƒïÜT ÜÑï/= áÑ:¬0Üm4z\¯ïî念^Û§¶íxB %R½( ÕsiM…×Ö|¤ú±º‡šîó¹F0‹®âlåÞˆW0 ÓÃBñEÞçÂ@â;õàVao2¦ú¶ :÷±CuÊ_Iyˆ*eß75#u¬šQb¼ç©æƒ:t‹Cçe·¬N n˜M) Nda/±éðn¡FÞ†e:)Ù¦Ù›ÿÆ_ëæZ,ÝÔ‹{ ÂôöÊ¡aŃóWœÙ½íL?(!ÿrЦò2 ³“],d,/ÒëÂWÐ.¾Ñà»Ù9ihä*ýÛ§:ê3D#޵?³³SûÐTÐ~u6»æÝ~,}*‚úRÈôèDb¯060qõcUàøÇó =ºGw’ý/EH™5-pY _n*IûthB¥é´‘· Ó»žd.òPBè#eD…¸ñ¯Ïèg šóhÈ¿$Î"¶ÈY…â~QÇyRH–]Ö+´yKMž›|ñœÃƒ:Ó×sïv±`…]ö}Lu39u’ ×E\M¦ÃÙ}¥˜ûÅÝt+žoG‡JÄÛÿRS_. ÛÉ¥Ç$] ª‹óÚ‘¼ºÖ_ŸÃ†‘Kk“ýÊŽ…½¡ÊSÔdû²2xÖA²%+3Ub/;v{ɇT¨AžúR›4þÍŸJ }82sò=š "ÒT`ÃÆá¥×:1¾>y„ä¦G*&g}‡Ô™Î‰¦3QÆÄÊú3}y=6>®É7òp…xÔwÆA“À¢[Íhå‹óg"½»È%áž~Únc*¢+;…'ý´­D•Âú+ošÞÜÅ`ÀAº_€UÔñã­{/qhxvù"%«µðÙ/OÀé!ÔÑÝo5¢SÄþ‘PdˆÐq½j¯þÇ–+Z˜-Èš/Ærˆ¯‚ /_°p¸P—òঀ#ª B›ÞîË¿ÔAæð'È­A±Çx£Pàï«$j-q=¯Þ}þwj«Õ/n[ÇÒÃ4 n\}Ò¼ç_øe1TO­¡©[h‘Õß¼,õšå©SS®¥ºËe£iï'©ECTš¬±UëÒ"ù˜'}>+*ëdÈ‚FåEGûÈFHD0¹ÿVgw4¶vßȞǥ& ›ñÊöª\øÄ2í™~#âјÓD6tò&òJ‚÷ü®yœR3óvâ†õ PK(®Ð/Œô%ñ¼ýwøN^'¸Ôíu•¤ÃIK² WܹáèWubrÓBHðE&¨{´klxlÇé¢GBVÞ¦s ÌšÙv•<£I̽à„dŒ\®J•å¼Ï"ÔÝ𕯑è7j”Rï "ñæ¿ ŠÂQ@½¼ŸX’‚ô„CŽ û¥QLü4lãÊ A.º,fåa€t²ðàô…1‡ù-Xp^Ãg¦¥JA”£/Å]bðÚ…Ñ~ZA.LR<M p7_ÏoȺ¤ ÚȺE÷4%èsûíÛƒ¨2löÔ/¶–ÆÈQ@#2°3Ýiçà á§(hd¶‡mz‹âÛ`CÖƒ‡Óé9ø²dªdðúî¥ë¼`ƒ$8/”wóIñkF¿„NL´ÇÏôÅ„}ÞÑF£úléÒ«L/‡RüÍè—ý·SVN ’ðo?šï©ïݾk{¬Dkst;« ²ÇÙLvÅj?òÜ%X³w­ó®öm?©Í^6Š þvs‰l‡•¶ùµ¬³òúè fŽ+6Ýþ¢aé©T4æn2:á$q½'î•—‡Mx²³%|Ķ#¶ml­}ÃGŸMÓ¹uìð•›¡=Ceª’§læ6i!õ)8²½§ºG¬âF/KTn?Ø1õžÀ)ç³Àí>¤ÃŠ5´ßr¯h\̆*·±¥üeªà=Š—sM•ph™–5W×§ª`®UõÀ~>8©çV1P)Cf…}8ˆsëåç:Ž:äTr(±8É–OKÕ–ÞÈoÀôì¯PÒéÏ•åÃö¢+åË]ô é¡ýoD#ÌÉ¢œ§µ×•Öè8ÆÍthìù‰}zA¡ÌÓ«×8‘O:ŸÍç~íq°?Ýs@%öóRW²÷d{ÉÒ*5›„€/FC²×$²_kµ¡oø 2–z_dÙFÜF\­#Ï×03æåÙxšDÚ»Åæ§‹Ïµ˜ÐŠˆÌÎ8±õŒ NéF.оÍÜß"~½úbé"TÊæÛËjƒ¢uÚè“×ÉÄB€dðŒw4+ £˜$E¨ JzJ¸6:‘ÏÑ Ù>'¥F3p>—ôC’@ÐÇÛ š4,2›[õê‰za·%Ùw¼© ¨ ta¤î„Z’¬Q)çåqBÞ…»`QˆX®âjú¾Ù-‚K`éã{ ob¼ý´¡ˆ¡]ÚûXÊ5ÓXØUµë׿‹e½I…ÞOàì–²’À—ÍïEN e™Ã󊆲ðržh¼q“æ9%É+ÝAjc&Žõ¥ÿ¹É–‰YäôjNÿ CÃî·ÑBÔÙ¥^ìN/ØŒ•2ý©ý|¦&C…ðQ¹>r1¥Ü1ØÚ“Ë{+9˜/t°7=äÇo€G? ³Ÿ.f…uŒ‡O…Hp9ಅ=ü“¤'Euç|P*Év©@ìÔgm#,hÏôGãß÷“±5Oä€IÂ)ƒ€~P67?Ë †€Bi6“B¯ µè®áÛL‹Þû\ÎHìÒ‹q·—¯Ê†vª5dgƒ.g±‰Ë>˜Zª®² ,‡½ºI,{EÎZbÈ”ä Ð‰Ò}¡æå '”f\d%EhªÄòo³ÝðÑEéxc¨©ÐvÉ•`çn?Á6‡B6‰äi–¦dxé°;“1C=-£{öî—r´Ë“ð-’ÉæÅÙm»^—7{ƒ ìEL^…‘'ušQ[íi ª)I3êÖÏ O Bv¿›¥Þî^Så&°Ó‰ÕƒÁ)Å$"Ðêóë“£qj§bʰÏ㺠SÏÄwÎÅÞÈÝ1!”áæs¯e‡Uy c1fýÔâJ … ¥£Åœ† Ý:ß9juÕr¼¾C£ …éó8 Mû›àÌÖZÀAX„Q¿yL/ýpÐtr€ñíê{0„Ó™{ȳ¬¸P úõUn–"ýZK c³p‰çNw¨ !†À¤»õ#'+‰q­hÑå\9ªxcn‡eMyÛœüX*¼ß‚· € _Cá*E½ŒlKzÇ·âö£þÖf9~¬jÐ^ƒ¼tŽ‘Ûqó&]ñÎ^Óº]ÒïÎ~-‡=Mè.9Œ5?aENΙ-9òi™gkßøœqǸ sj|F_YµÎ̓Œ_rBÝ6› kògúe•6¸­Æ7ß! †üã\ŽF¥qî]0·Uï¥^†åkĶ8&ckUÍ%¹>ÛzP,–Ûy·Ž=_¦´–@î—Q&ÓLÀwøU¸&YàËᎎÉÅ úùeÓ5²w>Ñ \ä^©ÚP^Ú-ê,Œž ™Ý¥¥÷ú;ÔjГš©ÔˆÜ¤Õìà—AQjU «^FΛU8¤Ž@·ªŸÎ@àù®ÇŒ]ÒÔ¬ÞÍ,ǹÄ8^<© ßcÕ¦B’îç´lå©Ç)èœX·‰PE¸¼:hAöP]4~Ø”˜ö/˜C¥Sç'Ê{d:ißýžX¢¾Q³èE._Ûgüà¬Á§å/­NþUfdå|í¤ë9K©TìY÷#qüã^¥‰ï‘[˜ßZ-bË&ãAæÓ—…–ÏàV»ÑI‡'Ü>ÄÐÄgi*¤E('¹3{³XŠ\)—cÑ” æ—zMQt±‘Ê)uC^M±|ÉÆ£{ëÖú]DûV‚ÖÿO ßûýW²núZ†#e jG˜BI~áÒaéUÅ/- ñµ@ÒÚ’b½Px&®‰+ا;ÃePß_qlQ&]=›¯QÍ ÑE¨“ì;j¹Þ$ånQØïòß§ƒѪøÀ¡He=xMðfæÙ'¾¤Ä½úEA=¯&{œXõL¢jzµŸæ½.ÛVsïdŒo‚ØVœ‘W×N ôc|@&´_ÖdƒÂÉ —R˜Ú¡¤Rå´‡e.M«ýÔH=Ľ¤ÎNÂSŒ÷~àTǦØÑGÛ¦À:¡B¿áèfm­æb`|>NoÂh‰Û^-$|›y¤ˆoïk'2'Iø/–§vÈËaxþ}8£ÉÃH¬k]Œù+$/ü4q’‡òpDç_åæäc%2˜[|z\=œi&ó‡ã¬Î~6é¬7ßè\µ@(&Å)I*ÜDBÒ…¨ŸûjÀ"6eQ£ŽªÏædç =µ;B{ ô枦0:ÀêæØž–dž¡a›K%©; ŸÆ¢¦ÞÊîydL÷]zy!å|ì´Á,TàÚàÚXé+Ú»¥Óí÷(íiÝ×ìö~×ÂB :8Çc²Ö*Ï!…èCÓ9­ïÕf!ÅU•0¡ŽÞÛmþŽ…„^{É2IO¨´H ž¡—)=5»_Ž×™lô Khúzu‰WÝÒ: íf‚W˜G}ÚÐÆ7Södš»hs%œY~v¦ { ‚ ƒùñÜ•Ÿ,b4²S£m‚]eЦ… ÐîHŸ^Þ~¯F«Ôà¹ùLÀkÛ”ø9ÀK cüà} v‹é‡t·)š•à*€ÆŒ-І/uØJÈоJ{3ä\ÛÑ`:u曀™h¿%»8wIÿº¡àݘÜ)¢7ï÷X¨Ý1Îߣ?%Vô ÷¯újoËFò(åS²Î’UfŠ+ˆ-¿ušoÖíª6»G=¥Üf8œdÍ\2¥PJbIPŠ1²MôüÈ}åç}ˆõϱl°©å=Ùé#­¿lj Æí—ÆáA“ €ÿôR"1à T¤–RŒ<ñOˆøØLèàÈŠðÒf¤Î…Á‘‘a ¨iC‚XÔÇE©(¢|Ò¯heÓ¼Ê\µÛãžÑßËîþˆ|¢É£}°‘Ès endstream endobj 1390 0 obj << /Type /FontDescriptor /FontName /TALQST+CMMI10 /Flags 4 /FontBBox [-32 -250 1048 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 72 /XHeight 431 /CharSet (/A/C/D/F/H/I/J/L/N/O/P/S/T/U/V/W/X/Y/a/beta/c/comma/d/delta/e/epsilon1/eta/f/g/greater/i/j/k/l/less/m/mu/n/p/partialdiff/period/r/s/sigma/slash/t/u/w/x/y/z) /FontFile 1389 0 R >> endobj 1391 0 obj << /Length1 1481 /Length2 7224 /Length3 0 /Length 8213 /Filter /FlateDecode >> stream xÚ´P\Û-Š» Ú@pwwwwm Æ¥qw ž‡4@pHp‡àÜÝ‚ë#É9÷Üsÿ¯z¯ºj÷žºæXsŒMM¡¦É,néd’qr„0³³° $••å¹llœ,ll(ÔÔZ`ˆ=è/7 µÈÕ ìä(ð_ ’® äÙ'„<ç);9Üíìœvv^66ÿ߉N®) Ø ÌPpr¹¡PK:9{»‚­m ÏÇüý  ³ °óóó2ý.ˆ;€\Á@G€2brx>ÑhÐt²ƒ ÞÿjA'd8 °²zzz²ÜXœ\­E虞`ˆ @ärõY~¨@± P´lÀnüšNVO +ðì°[€Ýž+Ü-A®€çÚòJUgãŸd¥? L€¿îÀÎÂþŸvUÿjvü] ´°prp:zƒ­V`{@UF‰âa-%íÝœžë@°=Ðü9á÷ä@€Œ¸:ø ð/xn®`gˆ‹ØþDÖ_mžoYÚÑRÒÉÁäqCù5ŸØdñ|íÞ¬6kçèäéèû—av´´úÂÒÝ™UÛìâ’—ú+åÙ…òÏp³ññrò±@.—… ë¯öZÞΠßAö_îgþ¾ÎNΫg °èùÅ× è@\ÝAþ¾ÿø·…Âΰ[@æ k°#Ê?ÝŸÝ «?öóò]Á^C¶gî±Ø~ýþófüL/K'G{ïÒï—U]KAA]†ñâÿÄ$$œ¼¾œ¼fn¶gºrñxŸ_üÿÝå?øÿÆþÛ«ÿ5Û? å­œìl0<_Þß8<þ¢Ý_’¡üû§g.ƒtÿP߈›ÍâùÁþÿ,€ß%ÿ¼ÿÕåÿFýÿHÆÝÞþw˜îwüÿ:€í½ÿJx¦²;äYÊNÏâpüßT]Ð)+ƒ,Áîÿ•‡Ÿå!îhýLqfv.6®?~°› Ø d©†XØü!Òß»x>ÃìRsrÿúârTæ_¼ÆŽJs ƒšìn—#¹Î—N®ÛŸè ¯ëÚv 'åsi._ìÅšp¿-ì¹þUÚÙùE<Å•˜¾…÷îÎoÜÙHWPÒOu¥œÍç'`Á³Z±yqZã-]°îI(ÇBÕztD¸B…ó+Êáb j}Kmཾø'~y£#Ø€Ô¤‡BSxXªË‘Æ×wÈùg±,4¨¥U Âôn¾NÜ—ˆùç…@ùÍÈRr/X˜ÆåϬDM¶-ã_7ÒŠ‡¢Ï Qvåìkܱýd»‘Š’¾%ÚeoÜÀÒ`Á¸Ð°í‹}Ó?—©’…9ë•V>Mü2X-ƺn?"œb@ “A¸÷þÍû‘ ØšÌ×Â4A#Tì´7"íO2±ŽüsÒZrçp·=Q³x½Ïê¢=÷ân 1þ䣯ão¦ã‡/µFPˆ[$²âbæ00ð¿KÚÎTöÕÄO*q¡ ÛØk] ^"-aúXe¸¬_{–ÝŠð¡°cçtÇÒ7´àW)QØ#’UÍŸg-‹cj?ˆU…*MNÏ}7ìk½ãô†ßÁÄ3 ’õ EȪ~1"=ÔŸ¾•6°l}œ ¨ÖF%µ¨õDR$¿`‹Äéá˜xD|zêåó‹!Þq.úFY/½M¸¶sO‡¬m¾d.Ap½¾9ªM¨:ÚÂAŒo›¤"C}tþÈ¢äGd±ë£[ø„Eÿ@ì`r5ÑGÇÐõÁF—IînîÀ‘R»î ‘®Pâ+ kU:ŧ“¶”›Åvñ&áâ4-Œõ¥\eï •ÄŶáa-"™GHTÑ‘õ¾‘ùU»:U” ¹©Oõ'è³D<íKîìÃÎC ë¾7,Ó[ǽù‚</rÝŠ:Ì¥ÐJár¶ñ„¯V˜köåj{7Ù$MRú}m¾cxÔ›Èý‰މù´~©K°éøUa‚ä|wçmÇQþ¬»®± ‹B ´uµŽá*Etßgê÷øZÆVƒP@úâ ¬õÇÂqTvG¤â–™Ÿ@)³*!¼Üo¬Ä0t}'es2×ÄŠˆ)œc’¨a³ÁQ¡õøü+n˜ ­>ß–r=eQâ†}äC«œ…5Á³:€jGœÝȯÕuKùðF5iø@ýCWQí^7ÔE½aµ¤`äÉ^ÑiC¸g B(SŽ|^9½;¢µw„f/zÊl&”KœÀÁaE†jÙù´:?‹Yv"õê (MT MƪôÔw¹fX^Us}»%ÂjŽAauý¡8êÙSMg&ÌÂî•wžÞ'~%WüX¦LïmÂü§r©.?Qõéò^_>nÕxœÞЮ4-ºMqín.™yb~<½1Kl! ñ þLèGŒ÷Ðù>?’—:vÔ™ß.ÂÓØP\KïV…°Pp‘¦Ï®í‰žt³êã§&L¤T!7²|élÙ^OìÎz¸IcÁ»®àÞž•^b$:1.y,mHž0|\ ¦N¸S$C}:…Xp¤¡lX’>ÔÅ:›4% õš“BÞe@=ˆôzRܼcpåJº¡œ·+Ù™ì¦D{áM î• ½ãÞåMNZ„ž9†JGSd»ÉuIKà»Àø¤šæš²içEŠs*–¯ZŸCû.Ðä²è¦ $G/†ÇY‘·ÌêÞX´óíÌÅ÷‰º°[”P?ª>©ržûv¨ÕÊ™ ûüÈÜe-!Úo>@¶JÇkÔR"b¾ÚÀÔ¬Îz‚$±×“X{‘xmµL«m © %qÒ\t•O—ÅÊÂïEæ)jóWås^T÷–NØ’ko޲nŠÝß̦+ª!ò@k£w–è[4ǶFuæŒpںLjì©í‹W_þ™(ŒÖ~'ëÜ7¸·¨ÆÏÿ­®Ë&x@û­»vÝ='Ù…ƒßô¸ìíK*(¸#‰ÿãµ9W,½ÙgÏtE»”×nµãîÐ:‰s¸’ÍŸ…˜KÌÒòÒ8‡«C$®ß…KrƼŸhî:Ò¼LÙÄq$jÓM,=…úœÞXMB±vËZ[¹}ÓÌÌ÷Âlx"oþ«º@îÆtvžo÷¡…”á jÇMZΜÂ)/OÏDP¤î|8µ(ô±ÎÕŠt„?ì@L@ðØg-å=e Bl]•á{;/dˆQF¶;ñ¾Œ/oìçI{=ϸnF¾" ´q”àWçÚöYèÁnJ,Wôá¥2eJÐÔïGL¡‚•ºRËcœTÔ®†“MRËV»è ל»Ù=±(o›(ÍX Rqªî,;×g‡—ÙjmÞ"«ÒOïÜÞo‰Ý3Œ|é$ÆÕ›+Í’Y²hÉYŽq°Ué¹øü…º§-ý5%(Ãê Q%…“=}Lµ znéù–Ñ))8öEžt&Æço*U-ÀÈÒ*.¸–¶¸Ž`ß¿Um ñkŒçÒÔ×=[ɹ0ç‘þñ¥[G) Ûß‘vËSä#Ñ–¯ÂF5ôv×J°å£æ…xñH|°“"rfá–`“ïÏú¢Ð=gÔñðŽVÑ>:Žªf™”û䀋c„ûYÿ‡íäŽÏGX[Ò°íž4x¦”èÚ¨ú«ÕíS©q}Hhã»ï…†0;Ž£¿Î™AçÑ<ºa×ÖH(jû„»yçhya¢åÀA5Ö¼O'±Ófx F«æ!¼ÍܚŚy4(n?bÞ˜ÄÀé¯f æå"Øš¸©i8wÂMÖ: yס¬0‘s03݇“¦cƒ¢[Í éÑ›€6«³ÈOŽŠ$þò® ¦Yš˜¸ ºn+ªŽt?ôÆD(!+Új ÐxˆOB›u…,´€…¢9%qâ2„bp͵@Jèo<Í í¶²ú4Î8‹EVDØnÀ’6؃œäŸ?ÄÈR3³¯kî‡c"ˆ ãž6K˜×=äÄ5n0F^ˆI¦ºÅ›Ž L_h&ݹ¦g–î¯00cnÜî=ô2’xc=|G·7˜c–Ë:1e“Ì7¼>Õ!¿š*^slUt-^ûÐ6©X­—-+q--Õœìu €—wfþMu¸Aºg ÓÃê5ý‰Ü¦‡å\WÙ.™LH!Õ[yËRÝi,-øãÉѱý¢W"xÈB©ÖÞ¾µufZG0]ä]FíM;å:U袃¤´R'À[üU`z¸xþËÏ:*pMÈ©ªÙ0ù41¥jcVÅtòZ: $aõ ©î‹²å•½a$÷71S7Õ1:2ñSÌ×Gf=É”d½.*‚l8.½‹9U¯ð©âe›&òºaH".HÀ7}>aŠr‚5¨Ü&‡ZZ_²¨sfs't:T-羘àw¢Aã€úÉ`Ýó¿……GL] &k~§ôYmãɈ¦¡üæž5m·!%¡¦—ýîpÛj¶QÌ­®!v—ºÆjð€Îm|A‰ê·— .Óÿ°©bÁ[>™–n'ëh „t½2ëĹñÛÑm Šv_Ì'Vx¼D£ô­€œ”æôTܪ…ÊsŒ¯½6³ð¡eËt³…uG¬ÛŸ.Yï.À× È°'íS®œV]<Hñ"Œµi‘ð .£Fò 1 ¤ó/ï"qgÖg&gð0§ŸfÙDÞo–ø¸4 5´žUÈ8±5öQº~ÁÖ¼V«“G±vÅ­ýøq‚< ÊÀpN0[~tAxh„ÑÎì’ƒ–ÄE·µãİ’=e~OgôÕ«óÐZöªøÐaë¿¶.בh˜¦©È¬œ )Ñð$åÊöíÓÞ¹zkN|Y.K<ÄõšùâpѼ. ô•ì£Á^¸TR±›Ä"ç1Ðô§ï´Y=N?’—C%ÃX•1H¹nkX»S‰ÌYNŽ9Ú #ò#Ï85ë ó (Í[<©Ó®9à YùL™æ¸—ï£!~b¾©0¶<}H#ÛŸ)ùeW >Ñ;ÌoQóJý'.ò9TQoGçÆ¥|ËÔ|‘¸§|WÊ‘ïàéW,åJ´œ0!ÍÍ„üløQ Ã’¾†¸æT’t‹-ôwt&§Ò©ûO˜°?«WŠ;ÃTß|¸ª U§ g–—c…³Õvuy„=a)Œ°±œ›= ,Š¡BúPñt"|h2Û)Fv]÷ d¸lÐÇÍ\°*@Æö}Ðe%aw´±”©òû/ ûÇ ²ÔZÙpµ¾&êÀ –«<6sñvR¬~*(z=G°m‹•·ÀÀ6#•“¹¥Y,×>ÿ…:eV%ü´ýÎŽ•Ã*` –†ÉÊÜçøq"©š¸¹O]{8‚<¿ÚcXÙž>'Grø#ƈڣ0óÐr×ôõw‰Š Ö×oÊÐÄâô…;Ñ:¹whç8ãáÐÆR¦ñt|iõA%J­ªe Ó{¬Íù!'n|ðÀ¦“D]Œôêl…¶²ÅˆÇ,ÔÃÜèlv`žøÞñÕá­¸·JžÑvZ«Œ2Ó¤¡f°¢ÂqôIãvÏÁRÙAÄëð„ŸäK›dÓ\lÛL†5J{2{xÏèl 7AÕ¼HûÒ'Œ£3:4©\ ä#VÑvOw¶˜½‰P—í¯Ùt¯¥ß¦›™ Šƒ'óD¶Ô5¨–Ã_Á«ÌÏÁ_xÞ£ö¨¦ª+*^-dð¢ì÷¼Ê|F§yÌm*n:Ãyq‹O]÷ŠÂåòCB,ª>l£xï=÷:F¨¦Ð#òFWïŽÉŸÅýýû¡u«H‡•®…uyyµ¦— æHÝ{uΟE™œ¿bœ¬·±V'½zR¶«á‹q@óÁ0n"f•Q¢ÚPfçØžDrÔÿ<&*¿/뺴· jÆA@%ZßÖ.Dg—YHªyJÞœV±ÁÕR(TC´K>»kÂróYàB§Ñ"[-J O¢Ê†âéýj±ôcf÷À°I–f"®Ä=Þ€&§;Ñ×µ‚‰øÜNÆÇs…Mð‚IòÖ4 ^pQÊE~R˜îE„ì~|áq°Û˜¬ðŒÝºy÷plOÆë©þãµáTtzñžÞòØð¡Q’ƒÄÐÇü<®·x¢ox"ÿ ~Õâ~ˆLt@ß‘·4ç´uU°©1ŒÇT`»]`Íw£~&8ÐÀЧk6 ^õü}ä„|Í5)<~6;Ö†ATè þ÷]f5ôƒ_dùfy’8xpܵwË•\NUè}8SÀÝz©Ipv‹N)ÿmvK¾s®ü ‡“%o£˜gVJ¤ï„¶Žä8©§¨žy²¬iƒ¾$ÿâì>;Çu{¿À å5Vƒ=¸XSÎÉüZ.ú¨W‰‰èT°‘±Ûkg˜­­¨%¦.n†íBéÚÚÌ´Øv¦~'ÆgÛ&N?>±Ë¥T(±ë`ùyï4»ö”J©+†¹Ï‘"»ô‡0²ç ÚÜÌ“øÑá{À…ÛCÍÕ›•wf~‰A,jÐë˜yø+õdÊÌ*¬‚€㟹ä?˜ M?Yu™ÓL­ô:[q¥9««è¸kc–%¿½£fbMS%ŸŠË'»ˆª©0 ™Ó­Ê¡ "KÚKµr¾ìB»Ž¨I2nþˆ7* o`MÁ§lÜX½ ×gV1JEÖl xá04µÍºQ·H¿“ &´)^”#”éŠê 5Heñm“äìÍ‹¾}´çíf…í|MÚ«µh˜Ä!*s͹\¬‹õúÃ%ƒìÚmn]ÜwçÊ*"7GŽÉß ª`ºe`TÓ0Õ§”݇a­[kÓÈJaƒ§¿b( 2RO(£¦úOãè®Åû|£³wSëKØQò2åCž&ûG$«dhWæ·¼NZÈA)íS>!±TËþ÷ùÙ»¨|”dî-dƒÉ¥—U˜â(h:ê?ÄC­J¹1ç "+gƒÂ*%ƒÃ®Zpa–Þ¢ŠjXÑ$¤öZÔ’8TIa|¢Ë:—¤ñuÞ2‡ŸvÛ¡+Û@Wx3 ¬CeiÚÔÛNÉ:µ >¥éˈ\lÇaÕµ„2dhÈÀIAAÊ /‰ƒ¢ø¾èN4…·CO„¨¿c…$•Ð\u ° ùÄÃ…c)̞Ό!ÛZ´íTøUDÀ­“jYx7N¦c‰Œ¦_Ù@ì é Šõjî >/g2¹~Š"òS~‡î2{"Òœ 25_shßãŒT6ùVû6ö˜¨Rï>úÀYhÂl^hÛëñ¹I  ‘Wplž~5ÜSI@¹jy0 &_eò/L—7„š6oÓnNzF ‡œê"*¾j‰•{Ó»µ€³Ÿ3›&}œ\tìæÖjFg´@)Í¥—©|@etU99 N¾Ô ÀCÀ=­š$7cáÃõûÙýÚ‡™W¶J^ ¶Ë9?þ ûqåXEZ†…qᥢ]7¨ÀnS3ÉPD0~FD3?˜•œËc¾"T’wÚŸÖˆùÜ™LÆó)ÔJt§ŽjuÏ-8aä®ðôÊy )Óß>²ßf.r4“}«á¤߃Ï^BçeR¹·xþ1â&ö‹&ÞPt† Ô·é3v-V qƒÐ —€7D¶Om ImÖÁÅò»Ø1o¯2ÐauÆÙžÜ{´“OR¼áo®W†½aZ¡h¶£ÂãR dŸÄºª]óø\Ùzc§š#昳ޖ— $Îþü°N‘+¨ÝPrc®@ Ãr»t—YÏÄÛ¾ÑzAhq&S1ò}´á®f0ÇÝeA©¨_ N|VR}wÆ‚ñÐaÆ1á×<Ôpx¸Ö˜+8²ÄU¬¨¤ÑÞìmËÖ´mX-—‡˜G¥»hÉoïW9W>ŠJ’³#1Û+âœS 7y¦?ä‡Dt8ÜcO%) ý¨ž¢ì ”0ï!¥ÃYïârIyèj$Vj¿ó…JÀ’RjXnšMÀIñ&Þ’ ïuî]Æ?¼|á‚9!g¥@˜t™—^~±7ÎõÒ< [³ƒ1õˆÚ„a!rlXÝïKº 8M @¹;ôb”)û7/yä×*¼j ‰G Vk3úÙrMÃFdÜé<9ÓÚIZ“tí`à9Žœä¡Ò¶¡ž|"Üà_¼í‚*å6û\ù-ðÅúG'#Â…M†˜¹/¿y)=tôÞØ*§Õ̜áš´8Òëi¥1©ê9ÏYr)i³MÒÕ:<ñ§*›³0·¬T*±\漤•Ó82u©‰åZ  äI.’¤Ç#Â7"Ï·ŠaÇ&ÿuÔ—CeR¨È+ˆÃ§ß0'-u"I‹™“^ó`¹Bó#hm^ºW ûÑùÛCtºqù2!á’XW½ÿ+í«©åK}±;ÜJ̺ù¯˜èÜôFðúÔ&ó1‡»ÅÑѼxRNjõâì+zt€Öœ.½èŒ©6„„}Ð_¢÷9,äEη íßó5"%†ÈZ] í|?Äo*œ/@(Gªá.ZÛm3=ê¬t¼™sãЧLW%voñNþ4 §š:àH5ó"¨ã¾,ö#ºÒ†æÉÃ'›Ì³Ïµ™á?¦çvvŒ6iØÉL‰ÔoX2Ô™½¤hÔ(Õ~6¤ÍˆHA—•·®WèvÒÕÆL3HìŠû¬£‘U濌Y_R.G£œÀÀ"tÁÅ»Vå–ËŽÇ‘®áŸ‰ Þwµ ,Ô‹¼VIJ®¼qþ¹G°˜ÃƲðF–„~ÑÔ*à¢w›é¢ô‘à–ÀH»)Š•¨’K2 "‘×쉛,̲”®RD7I}FY gIRziÈím'º¼ /Z&Á$gè8á|åK®m˜?K®•½`œÏº€¥¯*!w‘r–6Ì“å 6›¼,¼×Ë:Ïý¬O46&¡\¯‘€"í_×™ÚaºU lcÕ.Hu¿žB‘ÑWdóìINâ;Gúµ}46ì͈ »&-R¤æèy8ÔØ.å¼´}RÍ€A#”0‰’_~u Ãz4©ô†7kFW{ Ÿ?vFJ|ÇŒ_ÉÚS/qœKŠçͤ#꥽ä×â¢,‘o×Kª|=ú ÖgÓÍèßJ³‚Ö"Ò…´žñ c)áµN¿Õýû¦A‡W‘ZE¬©¯ßŠ]Œ‹-áKBúØnc²â˜K+ç»ÕZú¾–ÞÇ…öu*“lÞ7Ö<2°\Vœ—‰¡ž³YÖküUIú&Ó]}¦ h·Ð D|UîF¡ª( ¿™³qÝÜ®ô1!©æÏ2ùF»tí^b|¼5òo7Š/Ù!C1 'ׯ6ô3:%ŸÂ½IiL5Hõæ¡%tæ8Zó@ê;>Mïß¹Bšò'¹´lÉ#ÎË_\í.áyïÉI»ŠåLÑf°qf¾Jtð:\¶—)Ø3Ï ¬)>áµa'Jåv\Ê–jNÆÃ¾ö®SBT{s4,š­ >pôÌánăñO"ýµÿõAóû endstream endobj 1392 0 obj << /Type /FontDescriptor /FontName /QTJJQF+CMMI5 /Flags 4 /FontBBox [37 -250 1349 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 90 /XHeight 431 /CharSet (/c/d/i/k/l/u) /FontFile 1391 0 R >> endobj 1393 0 obj << /Length1 1768 /Length2 10206 /Length3 0 /Length 11324 /Filter /FlateDecode >> stream xÚ´Pœé-Š‚Üww îîšà 4Ö¸ÜÝÝ%¸·Ü!¸wBp·KfæÌÌ9ïUÝ[T5ÿÚ¾¶|”¤ÊjŒ"¦`c $ØÖ‰‘•‰… ¦  à `aagbaaC ¤T9Yÿ#PjA`[¾ˆ9œ^eâFN¯v `[€¬³5€•ÀÊÅÇÊÍÇÂ`caáý!Ø nä2(0dÁ¶@GJ1°»ÈÜÂé5Í>4&´V^^n†?Ü"6@‰‘-@ÁÈÉhóšÑÄÈ 6Üÿ+€…““3³««+“‘#ØÁ\–à r²¨.@SÀoÂE#àŸÌ˜(ê Ç?åj`3'W# àU` 2Ú:¾z8Ûš¯Éj2ò%; íŸÆò0þê €•‰õïpyÿ²ýÃÙÈÄlcgdë²5˜¬%Iy&'7'€‘­éoC#kGð«¿‘‹ÈÚÈøÕàÊ’"*£W‚Ñs4qÙ9929‚¬SdþæµË¶¦b` ­“#ÂïúÄA@“×¶»3ÿ9Y+[°«­ç_À dkjö›„©³³†-ÈÞ(#þ—É«á™9Ð ÀÉÂÃÍÎÃÚ€n&̿ë»ÛÿP²þ¿2ðö´ÛÌ^I½AfÀמŽF.@€“ƒ3ÐÛóߊÿF¬¬S‰Àh²Eø'ú«hö'~¾È  Çòº{¬–ß}|]/S°­µû?æÌ—YQTJD[‰þOÆëDEÁnOFV#' €••›ÀýúáýßQþæÿîH•@ÕÆòO@[30€•åO¯Íû—¿Ö‚毓¡üw Eðë.4ÿ¬þN“×ÖÿçøÃåÿoïGù¿­þÿ$élmý‡šæýÿGmd²vÿËàu•^ÏBüz¶ÿkªü󔀦 g›ÿÕÊ8½ž‡ˆ­ùëŠ3²r0±pü)9J‚Ü€¦Ê '‹?é?³xÍa ²*ƒA¿_œW/–ÿѽ^‰Õë«âø:±?UFޝ'èôÇpcàë‘ýw¶&`Óß×ÈÆÉ0rp0rGx]ˆWÄ ðd}=[S Ûû`f²;½º^9{ÌÀ¿ÇÌÃ`Vû-úñr˜þA¼f“¿çoôz›ÿè×Ïlú/È `þ ¾†6ÿr±½¢ßoèk¿ÿ1á0[ü r˜Aÿ‚\fËÁ×â¬þyÌÖÿ„-Åúwãþ.þÕÙúÏWæŸ×g¶ý|-üÁ×€v¯k þ%Ö×ÿU/ëkŽÿ‚¯9þýµ@ç Ûk2·Á×xüÿk€&ίþãî^§ûüÇÛ ºMçÀ&ü–_ÛokD\wÆ`ORoµÙÇ õáú$¦ 6âÕ²2—äÊ${Y%õ-;Eío³Wç/<·ëHêÝ9®I$÷ÍIŒcç^®!g;ôA(E }{+djÞ§F‰êEý¥g!Á• ³D’[6¨]ñë·š’LylB^ÍŸ›ªN-ƃê;œ—¸Ô9›61[‘ÂRÓÉä¾Þ!Ä»V£­ ® ÷uEøÉºøïœR;¡KðŠsÀUk®ä£<æ|v9Jçi,…!*M«7E›P»“ѧuãR=l³µë\ÀßíkyTö˜)ò#°žˆ Õ.ÓPe´çW]í…ߊì%ÉË0.´ß›ö˜*öSál⟅çÒí,„ÕÖ½å(MÂ÷j÷¶¬|—éoçI<4Ø¢I+懫vF[ˆ ‹~¢s­œPk“š&ê•}Šè1OøQ3uWžb~¤û¬nŠ#TïM‡&úh5]2H)è;)6b4Ê*’¯ °ÐZÈóÆ ³…™—W6Kµ‰Ñú̬ –ñÒô£?ÊÛŠ§µJ >‚§^ƒIœOf?H²Žèî-o0+ïâsÃ>¶ÒXÅ9 ïÓÀ§ÇêH7î—6„~õ<(üæ8³PR˜Ã[·z3‘¢BHÙ]¨‰8QÚF]ùnÚ>[ëíŒü•heP \(èLË뤃Ÿ¡¿P^ WÙ€=ó5œ¸:xì.›…âÕWºß¥ÍM¢¬ZÈG=Ý‹ñK`­;,þ}r'ФñÊFG|†â€§Ì„ @S{ø—ê$È*™•úZL/ä5 «h„Š*¸¢9ÐÇ´¢ìCZ@Ú§x³£² yÕ$ÿ-Ò•¢Y‡bþy;M¡Á/ü.ô#­±–™ã£îoõ w,Õ \#>Ôóî„ áí#³A”ç!û’oéñȦ°`t&êÃx0OÈrÀqçËZ³wá^eìçâ0|›š5=®ËöYÊ9Ž&MÝY!Ÿs(%JÊÍ •¼~y°9ü R™Ûží1¹7OzŸÿ‚'¼õá]´¹>àeœP‚ãBŠw—-} rTdzéº(“ Á­éÊiA8B¯a¨¸†ÿPz+@lï¹°Ýç{þ9Úï˜99”ðLl f’ÜÃÓO¯ûðáÙR¿FÞó[úŒ¤%¾ÊŽ@Ÿ|=L!ˆGôL¯!~5æ¥9s£¸Ÿ'È`‹L„4>-—Ñú¡²—Š/¨´–©yS*íQ¥3® éy¤À–°þ:¸IeL.é/ pQ6RºÚ/¼*.P¦öÏ{Übz¥ƒ™ÙZÉòXÃ3MÃež˜)e®A<¢ÔµdP\7¶·fê%<@˾…®×úFÚ‰*ÏôEbh/–=è`r!C¶DÎigLÒ¾ÝB®õ؃rôt9ßW•¯I‡Ÿ©Ï¨+mñ1öî)T¡Çt‚ªÆÕ °Ý [gà–Ö¨”ß0}+ÖLØ Íóxfc ·c¨­ã œzÒXN0É©êD^u>b‘;ÿJjéÞJ,N±UŸK í•6ºuÎîå<Ãæ¡i‘t°)v#®4Û`¿<7tØGí$Õ¾U›}F‚ºÉŸ¡Aù\]ÂiÌÄÏÖêMqFˆƒÛŽ-†e¥?m1 «YïfÀ¸â³–b÷¡o©è%*ú>èû©×à¬U3 ê´Dm”ÖÆÒÔçVÖq·Ÿ—¶w¶igÐÚ–aèB5&¢Ë¹©W­'и§Š®Ñ¸íÜNŒn‰Jr^LøÝ²Ò)”ci\L–7AÒ ˜²§Óžõ OøœBjîˆZôнƒË©-/bÚp½éÑŠNH_Íø}3T:AQÛ³-Oþ;’éá)¤½zW±&X<ž©]Oa”¦jÑkÔk´à¥L Öº®“‹ª[ÞQÕši¡ÌŸ³.öΈ˜^"Mó-­Š»â5˜œXú˜%×c2-þPR¹é{KÀ!\øžúõÃäêùÔ Ü¬~ÐK¸?_ÑeDzt8ggând)üXÙ÷& (+äœyUÔÁS¾ã¶„}#YË÷MAãY bè6“¦£ÒÊð36&Åš7O*[gþü È2C¯šSfûòš€:íÁXr9dõµ£`îxB×uÉ'ˆí#ÝŸ ämÆi÷vµxæì–ÄŠ)ÄÔ~˜üÉêYK›½ ÙŒ“ÚÅNO>s’¤‘A_!–„ÞŽä뱑ô{Ó3*#’<\ü”_æ•Ü%î‡Þ4Óü}Sø•£èœ íÓ[C[÷ýŒΜfdwd±–.ÌfÛ&^%öÍ× Þ8 ý©!AúoE²Ä’¦ ¸”_ëÂˉF÷øÞî›K(DlCÔ:¿¨ÄÛiúaÛQqëñÚ/£«€e~==6h&ÃÄ(°}4ê¹tÑ@T>nø>yÕš-2oJÆ8”ÝÁ'¢ªjº<Æé_^ÀœV½…¼õê-eL´[p“:~:.GÊ‚ó¢†Û‹0ïôXÀ¶þR’r{zÈ»C0^GÞ\œÞ"Ò(&Y™–¬c“¤øS8’ £xÁq®5öDÀŒô#á¼,zîÔ´šÔS3nëÁ¬ vô”»XL{•–ã"örOÍ&UØ?utÉ1_Ùã¯À‘ Ôý(*Æ/@ ÍÜ +Jœ/o{W°“]¬É¯©¹Î2#êxn"¥×¾Šø/pg|¬͇¢³uà—ÜFF·wB™ÝñÌ>ž" ÞBÚëïÆŸûÝé†@ßýD]—å mxÁþ Óup>C³»ÍZ>·Lò©ûÀcFÕRÄõֹʈ‡ÜIq¸Æ>ïvå³!äjRfœšRæw%¬üCá¸ÊΈ†6‚mwt"OÖcɾòQ¸äžÚp}I9?_oY¦ÛïÂDèkK¬_·ÄÙ) ¬ðüXN‹±º2dŒæx`òÊñÂïª ã‚픨84ˆuƒ*Ò¬>Ú‹~"¡* À ­zÊNüФ¬®õÁ›º•¥ÿ 0°$–ã‘8¡iùõ –ÐÑ*¿p|WÏKx» „é8™²)ÑØ€V©× Ù³Ÿ}Ý,+HµN&êöWÍ$}>üxØÛ‡SÜoÇF\#¹Ýbæ@·UÖªÿtžVùž’‘ÙÝËé¶9õÙ¤ãÇ'aCÉÀâÛšE-äY”ÕyJHgûJQcÈXC­5àrÝCeEÍÒq’Úò)æ¤ÆE'lÅK5¼À~]‡YÅɶ =Àãð8æsð‘k•N*º DÛ*)nyÀô!££ÖV}œá̇©êþ烑ãÒMÉúÙÉ4³ZÑæ9³Å Þ•Ï4´uÕid*Që”Æ“¸}öôñúÊM\"ß·IÊz½×a•ù¤h``-–¤éd½È U—f rœP†úZ¥ . 3‹á<ÖxHÄ龨2ÜnS2–?W¡«Î­ÎùLä+{ô©ŒZí¬ÁØ¿™Q Õtâ|ÕßÃ>¼AþòÁZ.&' ªÒ߉›üãöQ~Ù;7Å$MqeK§•‘štõEž£ÓÐôqxä2Q\‰¹– “*XÕh´ 6ÅwýÖ’«$ÔÍÛî^»Ã׋ìDµ:’à\ÃîvŽáÈaI”í—C;ËâÈ_¹ C §+†R®'êB}ûàù¥õHœÑNiH z²5(è‡wÚw9—A* ¤ÂíÁAì7ÌH-³üî…IO½“@|IùTwÊÄ˽Hù“;ó*Å¥¬¶"œ[X\,˜â9Ükòïåw—s|¡(Rúµ´°›jèïxÛ‡uÉKëE,hkð«>ÌáDàí~( Š_JVsi½~¦$KQ9 ߤìøÔÂ=ásðFQtÌFaI˱è¸[ÐÒ ïñŸéñ'—¤¶ÔL>Πï¹3H¬ŸÝ­Ïj>ô}”õ C»é»€»ßVb\¡QY¿q"9ÅP,|zK€›‡=Ãê½{öE`ý^ob›Âh g©”>‚–&ÖÍžEîjÛï ÷ier¾Ã©2,×)#!¶¸ Ù)YÁ/¡rî݈kÛÒÕpˆˆ›Þ†P‹äSþ•1 ×Ä@¨Ð¢‰Á¸eþ=ÂÇåÍY‹÷Œùþ4úe‘$|†žcBÅ/¢pópTy­ovV.¯gžß.òJ÷Úk°ß1@°¦˜û¶›žŠ7"â‰í|S½åt‘Û•¸i®’Þ7ïݵÔ×òÛ'¡Æ£Âï:Å>*Qrky$\¤[~ªí>Ñ0×±oÜ9Þ‹‰Vë"L®_;ÕÃ0›Ý)µ iËEËŸ°‘õ òVS„eñý8­äßòy“0ŒµBàD–¬é‚$ĵ´Î@÷÷›¿(s®ÿ*·Dp‚÷“ìý#$=;ÎS[«þ:ÓÇ÷ÃL¼ÝD1øO“a›oðÃgn+–Ù»¡=)†04Z?=”¤vf™8ÈÇ‚OÆ©ÞãÛľPÿr³Ì»w9ù~¬»¤<ýX¶~þ¦%òÃîºä¦ù—³fOá;ëàœ ³ëš]_°çÕéø‘áD›w[F*{‰d×—vš“‹²ñr¡œ‚÷F§Â+Áõ¦ýÄÞQÊa*tªtó3&]K‹ÈèÝ2! HŒ±½¾ÇÊ“£d°Nd]ó9I²-îI{ýLú ©w†ËÕ—3Â1«ô ãÂ:|û¶3+w'ùªŸÆŽÚµ¨y,A…ÒXgƒôÑı¸Hi%M¨”Xò«^÷ÄöÖ®E9Á/üØçÎŒ ”cßs`±­ ÈTÀ´~ õ×ßH+*Š~@d.¿EGz’Ni¹Pá+Ê&ÂDÚg™ûÕ$¾Êº¹)Ç×Gÿþ‘ær!ô0Pû ¹¼¾–}kñˆ“̺¸¦×h¾(Dí¡Ð!t gÕ÷ÅQá‘­kÄâÈŒ„ÐmgLÝ žpmä/YŘ)‰»ø¬þ k²Ý祡J÷ªõQMEí,ÅèßùÎ QpÏùP¾æ ÞXBj²oå5êØœ‘;“·„Ë®4)A~)¼ò!‰X½(sà7–*ŽaŸâ¶Z™(ÙßF’…ûó|¦¢•Y:ŒޝÈ,B°¡y_¢ÇÁaßäc¯Kì<¼Ã¢lÀ·øùHh‰Ì9¡å8¨±£_8²P=>»°›žFfÌE£ýâmªIº&ØÀ§¤£ä,Ç,``ÚÙó‰(ÑŒ›]™0ñv}ÿÕŽHhtÇQmG9¸çc]’û²ü–Ι WaÔâŽ+•aû±˜ “«oµHEÐÒìRýó¾z×/ù½ã¾ÏK(õ3žðŽ($}PŸ-c hºvKËpÙ(NXúü™ðâ ÁužŠþ0´×xÀSgsŒÆ}ì˜QêMœ{cÒ@ZäÀRáA:^À}á"C *#ç½\µg ‡Ã·¤=Ü@ªA H_×1VX s‘•_•CË ó|¢I${ÖzÒÈÂâìJ—#ƒœâW¾@¥¤:ã;[K2Xž Tûh㪘¹‰mµ'ió!^dRN[û& µÙ4/>Oþ´ÍCK@ 8üÚÜit.`ÚÙ«?ñÓžµµÙO¯/:5•;É þ»9šR'Ñw£èÈéBéqÀWjüÜßxªœõxú#¯wÁCì™b0Û5öLy¬ºFeAQûøðqŒ­‡|2œŠ`0’n@è75…ƒÍN£_·w<}„}dViæbò31Ó d% ¹1Ä֘䙇ùp¼Ø ¢¾ –Ú~ß|Ù§*—j¯l%ðZ} ÕKô1Æ–r¾ÂQ¬^›MZn2‰ï¼sþŒŒÈ¦¤8n„—sii¢–jJ^RÖûxLÖuzäßà×ËwtBwŒÙà΂|Ôéý+õèøÍs©øG†WÏö­õ} ¾cm5n ª*BŒßÏÐ?¹,¶o»ÍbÃî:ð…múêœz5=(a¬>LÚÕæ;¡Â§¹{û%V*ÒSŠší?Uî;#Ô‡y wš+e½íˆì|ÐÚŽèé ‹p@jÛj–÷¼si~˜c^p/§È†ÞŸÎᤠ@ë"?Œ‰Ç¦ž,Ô6ÌD {,ÀdØÏ°}_¸«LîœIáõÔ¼±êb3d¼nŒ àDÒ#ýq<7¿Æ†ÌY,åªúXþŒÓEõPÍPMÉyHù“l_Ø_þ—ìxm¬Í²VšgÏ]mF5…††Çé¼Å¼v;²çtDÜGeÆc«–Ö,!ôìÌlµ_Ç~vd¾A_Ù¥ßÀ!ª)"·¨¨8çÔ<…#ªÚC8F}gn¡Ü* •’ü°ê jzIl0çÉn\s\¼€GÚtÜfæãÒ@ 阣 •Ê;‰°1ô‘¹xˆÍÍ ]]Bñ»³Tp $|m¥ì¼P^§[#ŸÞ̸¯”0mÔH… ;‚„Å=!Ù8ýWK"ùq¡ëWªØ†g¥Ú5dx }ŒSE¼tù25‘Ô.¢yñiøSÚY]Nõ”¨.¿ˆ  Ÿm-âJgà96‹H °'2Njµ¢ps€%ƒdÏÕÅ J½7¥$D¾½êEtUÒd8±Sx 8»Â´Åø`åAÀœ.ëõˆÍ¿aaÉn䋘h¾mÁ¥òm5^ñðö¨ýYÇUõLuq+vÖÌa¿¸oÙ·”Æ9{a¸Ú-k¯æÂn¦úÐa/7öÍ|®ÏZÀÛo‘£?$_ ±ô°.»ÿ–»hÿˆÿVÀÃѦèÅ’°< ýˆ5‰<õõ@_Ù$#í1ªnYTþ¥E©tÔ*įlþ²²¢¬¼ºW± _y:6¹_æ<¸À¿¾9:©‰Ž1Æ¢ºµ:Žï(ǕΩOûð2„ýÞwIvõ\f’k…¥s²€¬:Ÿdˆ¡‚›ŽTœ`ò!íº×™¨['ã0õ%\~úÅ ó}¬âæõ‘%_ß°7î“x.'¡_F¯ˆ–(•W‘Ô§vù›üŸÃŸ¢xeòôIЃuDÖE…`9­ßÄbÕ­eŠùb”…oÚå;â £i DE–íŽnå™ê­Ô´úëSŒH°‡yo!¡¸¸å¬Ò)¨ïîT€A Ö`¤L[BG¢¾fNQ×Sø——å1d¥O·\|s­œ°?QùŸÖÀ'ëL2Èp°ÁFA5d-y×Óþ$LÇ@Aïšãéå»ÜãT!põÎûð?]erkÆÏÀ“˜ÀráÜFr­œ7BûÔ¤yr9 ^Fø—u¾°ã çš@ ®·k&‡då˜"/É’6a›²ƒn¼QÈ'ðÛܽò³5"rÜ9‰‡¸˜kX= ˜c§Ö8KØA<œ>¶ãÛ½¤»Ö³Ézx-'bmKÙcÚ!ï0o¸ù_³We©>î5»vá:—7.‘ãøÿ²˜ áJªðõ‹«Â;ÂÉ‘‡2W«ãÚÑš¹±«¯»B“;“÷ÚÞ† fßNø2pæ(G}Ëîo™ ;iY– ·ÇqÒ¢»¹“~2Á$ï‘ÙžêÖgõn͸”$™AÚ ß'°a½ÏFg‚´$T»0Š„‹ÇéšÙö“BíØd4R ±Õp†tódm÷íÍAn¬¿ä„ô×Þz¿¸þl žË,6Ál€èœ ÁŒtU8äV*ZæÖ®¶2dDÜO¦¦M)Hîb}ǵtDYM ã¦ÝÚ1Î9&;VÓ»ƒfôz®[_T;ÉóÜ(RbM%,¬hç)9%Wmmp+7­Ñ(¨çéçßÔÜ'Ô+$ßС%WÕ¼ûRD2 Ð÷˜sÈ® ~“žiî\pc÷Žëþ,ôf]Õu;ou5霚½é¢^IzOòZ4e¨„˜‚ö0Ã[ó¡Ù½ý6¾BUéÌÕÅÀØÿ“ãWÎ׺!®dr…Ug‹líƒ:‰ð)´Ñ/yƒäí ½LLî%¹jL69ÅŸù4g×9,¤d¦JC|y…îuæ2Â3s–ù tñ?ФÂ+úG‚®5õ¿î×^ Æ™õĆI¨"ÑÊ©„)Þ|¬’ìé9ÞhÏbWÝ"º^J- P¸c©Bª@K¼qÌå½5©;<«¦’ =ÚaÊÃ;é@Áª”…$8$¤h­äf3ÆØÕP¨zn;A:„ËûùaäÙ]!ᥰ‡ëìë“Í& ¬TjK³©gaC‰s^_ÄÕNU1Oñ†²“TÏÒœ´ãCeÔÇzx3¦ÓTáCE ã2ì!+æ¸ ó¥½.ß’8-@[íÆÉƒ²©ŠøämòñÀ¯1ÇrÃênY’¼•o›íªðÉ„|_’|$”œp©¨ŸPï´mÿ!„æ¨NÉTŒÕ !ë·]:oû²‰–Œ_ã eP ’O_hrÅE7–*,¿7.!øÈlê™àé&Ç Ýw7}šèUÕÀà11R>'4˜•ÊÇW¡« I^-甼T„%Öÿ~s&‚ìÃr}pO‚ÌWÎÂ®Žœ—·S w‹+ì©ëZGºÚê¶ž‰Û}¾­g‰ƒR«ð±Ý„ãÜ®†Äà¸q!ؼˆCݨš[âlã¯g ò“ ^íÑõýü©‘™7;‚hÑõmª on´£úô3uݧIË&ÐÖ´ïPžW-9:Q¨øèÆ‘Aψ/&À¾äÝŒ²·û3…zsý7õÂ0ñ¹¥ÌA(}%\Å’¶F³‘ªr«jöP•nå AvÔ|HµÒÚ’ˆGõÊ=MB³ÔãЭäÎ[DáJBšE3¿2<-‹Íûgî¸bòZlW¹Ð3([¾‚x>Ôàæ› 7ƒZ™¾—Ôc‚Cbß³`¶i;#ë)nõ#æÏî0ˆ²æöÞПވžlñ?å°˜ÒUg= Í)ç­å0 ú—x!8XCP¡¦üѨ%Œ ‡³¿æ^™dLÛÐ9§ND|ú—ïíø2®Ü™¾¹¤Ž`ËUèWŒJ]|ÜcL?ãòHŽc“â‡<•C?ˆ,«©¹–†5U’Næ,µmÕµNHFì†"êtlS2ë"E-ØZ¼Ñf™æ¨1útÉ’|û<`ùbrÁtÅÆ1…‚šŠÖæÒ<§¯`ˆÄ£UéAÓ ™.»Âà“Bc-zúî3½3Ũv¯¼ÉýË%›^Îmº{/qº)soh‘Õœ¬–%šTï­e×ÍT# ¬e€1f4xG·Y2šåqkkêÔè\I‚¹[$nÍÈ(R3+Ò’èb¤¥P˜Ù«Ùʼ¿éûU­s6I~½²™ØÞñ…µ¥ÔÊr"?ß0gñ¡gAðõuÔ‘ xS¤2«õìº^¤½j!‘D}]@ W¬Zƒ¯^ Z¼š­yîý7ÔåøÇ¼T---ÁûÓm½¸Yé‡Y²ùSí52‚zìÈ~Ñ=ðÂd\{ìɆEÚØ—V‰IoÀœ¶Q~àä0gÒÉé»1úÔÄfå¹³)yí_ÁcúG¼xÒß0®o²ãEï@ݾ"G­'ɃÚèS™YŒõ©hkyÂLíÕ"d4yé¼¶Ö£+N͘þïmö¾WÅÀ“Uàcc–±Q“G¤7•êÉ9úHTEÈéàã<©j†iSá¡Hx¯ÄhÐ7n|sþØ»òáž™»&™"ÄÕ÷}G)=Y$J[½ÇQHÝ€>"„¿‰Aè²é–Øšf«&… zÒòºà5Í5ö-¤]ªæ§BŽDæn—Ä™­Kÿ5´a²|¸V ‹J (z§/L“¾ Òø<xEœMäÐoçÿ±óÜìIÓ_Ý4:NÓ›:ÁâFŸuØÞ­÷yÝnáƒXÜæÅDÎøB˜´õKQÉôÀó§µí6yXm†ÙH%ú½ý¨|lí;öœµû¥º3½"4¾jY+´¯·§_Ú G ú}¦ºß»8vòëëÚ¡×)˜ÏG¦Š¢{Ý .wçjT³>Y<Œ†-G‡P*º6ž¬r¾¸ôƆ41äí‡òj!.èd¨‡­p!làÙ¶êeTlÿº–Ag÷Ä]iñaÿˆ‚¹‡­op·˜ø¼ÝÜ|ù$äÊo¦»vŠqŒk¶rµ ñàô"‘øA½­P¬•œ UÀ²¿oÕÅ,µú²,Ów]†Š'Û'y #€@Á¼%gŸüi7@* V9 ¸/°&Ç®÷ŽÉA7¸; =è³d·òVxfÃïú´õË–Â,ñúŒsA›\…²ïÌUz9$BSs-¦YKò^Ç1Ò½õïÔû¦-³ŽÙ½æ+¢@õ¦íCù/úL˜žjn¡÷à#r4xùºô‰¦“锕_'¹ ÛXË?;”a烘Í䯭„%® ½ sdÂ2^W4éÙ¥@XÍ$ŸP|‘ÊÚ?‡ÀÒѧ®«[|cŽZÕ~¯¬£ÒÒŒ¶´etÒ¸ØïÆÂäãšÁAayÔ¹&(ÉðüX˜uoÌËLñ,Ï%ÝFfÁ/=èŒt0y¡|“¨i–²,àƒ›¼}ùýfžŸ*ai”J°¢³ØXÏ/O¤‡Xù¬Y ˜ƒaOº½ŠÎû¬†[-¡ ±][¥«òŠ5ÄZâL…OâÅ-±)8³åįë>–þž¡†=ŸéìtzyÄq è «#!Ù?ð´"\ 6¥Âï)xPÌ‘U™ê…ÐφU¿Õ’¹ÉDi¨.¶èŒ@Ñ! ê <6dfRŠåO»Y*¼øÎ°ÛsuŸ%ün‡süa¹õí UÛ^FòR*þ…BSs+Âg«wêŒr³e5³ýnÚŨª =Ç*ð"jvÒÛ· µ¥iäšÉñ}YÎ e4†é”O$úmõï²ÚO: ‰ûú=[ EÛf|‹Ø1°ÇÒ*Èø ˆ€_Èfá ÈáNhPz†¯év”„êF~3Â8Ó ë¦mhz/­ßDs´`Lð“i›Añü>©|KÒêê´-)ÛM|á ‚k{Á/€ßDÇ€˜–ËpúUú°¥™· ãöL¥¯¼HsÊægú͵d°µË/Ôh¸‡°è[vAD$3$œCºT”5áèÓ šÕDz(+Ï•¥Q;TFl~&zdjèHDHSôÀ炇ïÛ o $j[ϳûxÓØqÞªxîûDBÑ~øšUç·Œìuå¨Ôb&Ytjc2¶ÎÆ*ñRŠûOØ'‡Õ½ë›Í°âFó‰îWß ñ¶Ip{ûDˆèѳËtDªc™WÞèFã¶"ÉU¦§xʲ9Äö%–2§qh”p`3˜b7“Dö Ê Jðy8ªFùŒG’Ë«¬bÍ׺€je}šòóˆK»,"¹’¡r¥Œm³@³oÆþác‡;h œ±öLŒc.ìÛV—ͨR“ïò7ÌC†uQß’¾˜VÐ×XÌ–Òí]:×5`ÕÔ$¢¶¾?¼š‘G´UëâŠ3ºJNT£S¯hÄ`=M<ºàŒ¦v¾ÞÓÉ0K"y¬Å·n~#²H"bRÈTW0oݲ‚¥)á ŸÐBù®_‹öɦ¿9g f'An+Ó¸`µ6mòýg¯ _Ί¡,lv¡\‚Á×nN‚ù‹«ƒ‘«REIKÂuÞºÞS3‡1è­ õ³Å]¶€-b”XLôh‰ÆgŽãͳSn ×u0š¨t@Ž÷§‘ÕÆóÌãÁÀÖKýµÂ¥á9›ð¼÷Ñ€¥‘~Ïí§Ä(Ûãmá±»ôV»ÊB9¿;ºS°–Є endstream endobj 1394 0 obj << /Type /FontDescriptor /FontName /NBGAXO+CMMI7 /Flags 4 /FontBBox [-1 -250 1171 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 81 /XHeight 431 /CharSet (/S/a/c/comma/d/e/g/greater/h/i/j/k/l/less/lscript/n/o/period/r/s/t/u/x/z) /FontFile 1393 0 R >> endobj 1395 0 obj << /Length1 2887 /Length2 24532 /Length3 0 /Length 26115 /Filter /FlateDecode >> stream xÚŒùPXÓ Cpw0¸» îî\www Á!@p înÁ]ƒNp¿“Ý}—ì÷ÿU÷U0OëÓ§»Ï™*(HU„L쌀âv¶Î ,ŒÌ<9ef33#33+<…ª…³5ð1<…:ÐÑÉÂΖçG ¡3H&jè ²“³³H»XXØ,œ<,\<ÌÌVffîÿÚ9òD ]-LrŒi;[ <…ˆ½‡£…™¹3(Íÿ>¨i,ÜÜ\ô¹„l€ŽƆ¶9Cgs  (£±¡5@ÅÎØèìñŸÔ|æÎÎöãWªÁØÂÑØÅÆÔ4 ÿs€rÛYƒ¦î_*¿ A·ì+¹ßãÈdò1¾Fð?‰YXA!€6&†NæÈ@4ACû§ŒówЪÿÛÜßFîÆÖ†6¯FÌÿ™€nî?H€NÌô‚B™þÙC‹?²ƒð+ä`û ]_k`ù-x¥ÁñÛÜÎÅñl ³? (þkì 0÷°7þI$û#?3è,ÿ€ æ[ýAM°þ‚góGi xÌrµÝèAµÛ½’9ÛýG *ÆþU fzþmÿ3Cì,ÿHÿ;Aì Öö ™´{m>;è ì­]œþˆ’8¼¶ë7r:ýu‘ý+d}þ§‹Ü)ìœ&FÖÿ!ÄÆþªø?SÍùæ¿öÜÿHÿkÌò{þh# ¨+¯Ô9@NN@‹ÿÎ?Ço ëÍäq²xÝvP œ¬ÿ3Ý, V¯iA¯“³¹#ð.°þ6°°6ùco@pv³û#(¬Ë4®@Y·?â¼Ýÿ€ Œ@йz¾òEò:þê?†±‹#¨éÎ=ë åüþë«è4†_š·3æ ¶¬ n»­"pcø1þ~†â‡Æ'¯%Çv—{d˜dšªÌÀuÇk¡äáÔÕm1ê+Áeâ'¯Ãæ:˜°–D¥ÖïGýxå©­ð‹“Ø‡B_û áÞ2¨ îz?9x«XA4ƒwJSä:¸¼CVüŒqëÖ'áþµ¿te4tþ‡Òn§ Âcé4CŒZ´N@Ñ,EžQÖ.)´3!,-ú™;ÊìÕõ zÎÄ ±t<¼ÏQ [¡×‡ Öwsžk媬N]xäxp !®ÐG§(½„÷S¤q¼Š WÝ›ù ‰sèSWP÷Y3ª,”£l{k\G—:Xvr“ÁT;‰5%˜úޤ XU­úQÎ5lV@ÂýSáµ¶kõnSËÔ·ËÜÓ__³HZæ‡þMÝ^ukÃ# 7¡©-·Ã#ÛÝ>EýoÍÜXhÅ̵#WWx2S¡JÇÞti¥0Qé—]“ß=B†ðyž ßã³l03xÕ >úö„#×ûü£OEý™|ÇÆ,Ú‚µºnÐK§ü‹QgêÙYž9›˜Af(u^†XwDÞ÷X*%ÞZ`ýlP)ÓÙF6£h‰±lÂÇâù y!»”Dï…ÊáH@Ë;î|Æõ µÖ8Ž|%’£(†±ýú€!á >îk“ý­ë²2a=¬ŸvêwgnáÂÃ7U¹‚ó":”ýsºœÒ$„¢Þ|¹Ùžë…3";Vv%±iŒEÑP}¿=ÖõåѾTJ”‚ªpI®_æ'_˜÷’î¢ha/^µ¿ vÄU£?km»Â-5b&êÚGŠÌÝ2ýöáFõóg|aÕÛÏdž`½ ÓcÛ$ߦ¶É½¬+tŠØûš¼ÑiÖñ”$ Ûþéëï©£#ŸM#›¾«¹ Õ-¿­É˜® B “Á´³1còI J¨2¿ËÒa¾Ì'f[nyf ÞÇ6½¯tÒ©ýŽyÅRö^§K œ|]T \ínr÷‘39gÚ]"OÉv°Œ =oÇ¢Šb>)›da ^– §F;;ír»Ð†$Ú4D3Öx- l›oŽ s¨66å\C‘M,$˜§ mãyRY¥Šj©¯VDzLÚ¹Œ²×‹ :޳ž,@}²w²ëpn^rŽ ñ"Š©dY|ÁÌTœ#)Öí\%œÜ»ê%ÀN‰o|o<aÊÀ·/‹ˆ…¹Ä6í…¯õ§Íi]ãÝbæ÷aó³ï¹Áë?ýaëë««3†ñ .¸R±ôGs¡´“;.Þ¹+ò±³Ì^¼wž¸•[FAâåLknòq”qaùÉö®&e‘Ž]'é\Q;5ãsíK4ÿìkLD¡³îpê¯ÓҲˮÀîJ¢¡#Õò¹úôkÕ#ÖÀ®q>ó:aÙ/9é;¹-æòÑXW/ÎuܤyÛ^ÀLJátR6¼÷7IYªÓZ2Œ«2¸Ã4Ýô8#2^†á"„%²Ûá‰=V¨‘#e¨B{Y˜gÔS+9¥ÉÉ­~Þ—71$âlhwáÎdž£nñañdF(Y)Ì’Œª<Ãn¤GñÒè]ì|åÊYûp=¿»$¤å†ÍrÒ¦·Ò~AêïA*›záî/^èË—IÂ6¦Áý6q³©sf–ã\A)P²ýq¢ïòÁZõЭ=q·<×D^G¤)£`|Yáz~B°PƒÕxÿ‰¬Étå6Ý¡ï‹:Sôð{ÁšËûcæ„n­<;Ï;òã»É›AоŠÎ©a,·Íºv¥½ë€Jý')óo(.ûyÂ6'âÐ8<õ,Ȱ‹)&°%Cœ7]Ëß„âHÅÁ¸²¶*¿Ü(Kgª«²%µI=šËß×Í{í( ?±@ȺÏÙ`AYÔé¢"•;:°élÔ˜ ­À-¨°Y5¢`^úE{o~Ó——¸žÈdzÕ›eÝN¡(ÆšÇC¤.D£",Ë´ËWŠ}ÂQ*’-”¿~“eÍ[éõ­ š¬eQÊV:6 ìXð;/åE»´î>Uê}³7FBÏ~2Þvrúó3©hëÏoãÛ‘ Ž6 ³ö‡½*ÝÌÇé šâp²áK®+ºNyâ©aéWçŒa|(c!¢-ÝŒA¯Š•£4²÷åõZ+… ¨PbÃÞWWç]Ìk ŸAFB¯T`TÈÞ° o{÷®lž(è¨74Ä,ç½Îû ARñ{ñ[Õ<£‡Oñê­©­Í*»³–¦+bÖHšöKçRæ¾él>ݲ+±àTúʸ·Ô ›ŠF  NzVÇŸL>fØ"{ÆwÃÂTäïðÌ¿•ÍöÄÓýˆvVêˆYø¹4ÈU^´#JN‚‹›\JÖâ™2QhÑŒ`ÓÖšÓ ,¥ÄqñB;Åö¾­)´ ?òÒ¸vÒ)6¶¶Eîî£1NÜaàTÛDHGŽËw<òI…GCN3WƒIĂЦAˆ™z„k!”¬ìnè%Öž»ä³¬§µnk!G ÜF/ù?H¾ä`ŒŒiôÒŠ£ë®1}1ð(Vª NÒQJî!§-]µPÅ>Ùד\›èçË„ªP¨~•o‘ክ‚¥SN›?Òv,ÌxÒºêÇ8Örú´Õ,¾E,ŒºYÔ S¼¿„Wû|tñ}6#Ô5‡"MU[ÏþAIpEjiÙ"ç%%³K»˜†«²ã1×|Ü6 ±ï§ÕìŠaOƒjï9•¦£ßìz“­¾m2„´l6,uŠ |í^î·ÉtøŠzyB/ñúõ•3=« £ï=¶¾ØÇ²Uùº°+†jyàÍý^áè®[§²ë«-»ŸÙ/Y>‘ûa8Ú½œ8%cö \ü²ý°A›Öú¾äfÄéêÌ.ÓHÐ+Þc–DD„8Ê Ç–QÃïCÃÑ)Ñ.3Ž-†‘}ˆ™“â'r†)ëªñÙ JY>ûNHô&k Dïfߌºwû÷ŒÃ}) t¦ãƒžÅ®AÄ›úµÂž•#a9¡—¬®³m°Ðo.Ú®Âz5Á óIÞáÍØÀ§¬„áš¹-‘ÀÀÃÞrý‡ÓôPÑõ=Éy“Vqgm'<ÿ³<Ž8%Ë•„ùn¿1R<)qmýþ¤DlÝP;<ÐôVÑÝ_2$ËûF㱎wE‚hU ƒ”Mø’céâq÷vSR‡t¦Lú–%e<íØ˜Cþ–»ÔžŸ•KºUøé¢Þ‰ â$²ÁŒú²ÄK‰Ù™_»rëÝX¢„SS±29·.'&eø½G%E=^¤ëðOËžkâ(Ù>ea=BeaCëÎC÷¯ú¹ ZY<µrEUë»'+˜§‘0[+Pê üŠ˜, B6ú—òH|¾$ŠRj¼õ*Ý™WÒl&‡–\/ÚCu–*¨hƒóšc\Òz{.¸h‹U昂ªyd»Ð‹Ô¡ð¸í,$Å ‹d›•*¥ÿ[O?dÅ 3¡Þ7{ –7Æ€¨zò¸XF8ìû'°¯¡{•¸*¾ ž?iߺSÉ>>¿ì³ß MWhCÖ]}á/¦^<*D½æ+žgHéKÁL÷²¯AbÔÌ#¼ã-X+Jo>– &Y©R5Ÿe£òfóÅ9Ø.î¸8[ËŠ°$jö8ÆðSáâú¦ v¢&=YDTèºÞæ+²ôú‡ß¨°Ž!°•I§È-'[SŸÏԓõ‡•~0ÙØËsÈpk<#½izW¿FJ¦Tl|hž©bw‰QóéìïÛë½,†ÆTL‘|KÑJ>:»#™—‚ÕúQ<“ƒ'C±o\(VÕtDôñ^xiTJ—À_šå@v1Ó‚fŽ<ª­*k^ªÅð#ÐWÅñÙà §¨%Q27ÉKÄ óV²d G ;*DÃ?am}Šþ63ž!«7¨qXTÒ2ì3ê¶–#V ûj ü.cÿòä|UUÞ„Ujå²u/eJßÞrzõ/`éÔŽ¸g†Aã`£Àçm)c0AIʾ¨¶„H «-Ø;Œ€,ñÀ·:Ûê5¥lý†B$mÛ<ç3t˜ÙÆ$ùîgP)FrÏó›L¦>lë)‡ß¾ n“J”¤ UìQ™Ê9MÁÌkj¨=ù<^×76iõÕÇè!øÏ.PXŽ ’HvœF)H¾À}$Ö¿kˆâÝ«ÿå¿%Ý䚇\ »JO¾SвB%b.À‘Ƭ‚ûDØzü­ºû\0±7 .ݰ¼ÎŸ3MÈÏ82ô§µxõ¯XhG°Æ"²ƒ…I[-/Ë6oë7®„•-TRÖ¹O~5$?êó®f{F޼’>C‹U}:}?no¸¯‡°h~¨`aÕx’±m{XñÎÅc•ÜñWÊÚE² ¼©±C/G·‡/ÉÓšî\Ö=Ámײ}x‡s.Í®ä¯Ü…¾êJ} BÀOBqõŽg‡Ihlþr—ý> žðÍó>Ï ì É:±jÇ®‹]`DH“žÛz¸ß„c<(üåU·šØÞãVjn´Ù`J'Üb3̰ßZl/H ®Æ˜N=Y5)3f\=¹RùÚ0¢S©l÷ˆ¶Ž.Y{<ãr%bFÔºšùZÑoó³^¯¡JÂ…C_"æ¶~‰åËŠ3Sô,ÂóFÌèÖü{D K…m¸þƒ(¿=4Ý®i©·99á}q?¸®à©ÜÁ¢Þt³„Ì5/GtÛàa+䎔£Ã©P1›©Pບ…K¹ÏÚNï¯ö©Ó£it-GÔŸ„¢W×<±1Ë&D0Ê…k€D!g4ÄC†Ò8zLš­ÃÒ‚áó­ níÙ"—X•=\n™¼`´>U™CMò7³áÄ… í­7îQƒAùRÔ)/m–¨²E¥R]Æõ>ÞæD]äñž¢[j=qn­Z1tíOåxÆO÷ú& ú3Y:Õ[û-»Á \nßà{C–cͯ†ï×¥è-sœ0e{L=žå¡BŠÔ®D5˜îËÎïqÃï†:2$"@Ìóܵø%kŒƒîc9R½7,øùçßÏe÷T£ &í‡m˜ÓÇO̶´È^ÈUÖJŸ¬u;• ´¸~kÐÓyŒ MÀšRÙ,–72 ¡[iëøœmG ³T¤׿9uav`6f¸ó1Î¥N•{‹–8 "²ÝÑÓ3röåù-iø4Ì väÚÂù_>¿øn7ŽÕE¥ç¼S[Z4Q|¦ŠD#‡pT‡„†öLqŸÞ|ÊOHRôá…Š$i“¹ž!WÍmD¿¹×ÙúћM„ ć@àMƒëò>Pˆêô"IºÌr¾(õ?M~¢ÊRb2âu'kúHÌvA¬2ÊSji„v¼BÑä,ÃÄ?’§ðr’Iè¨þ= Ï]·ººŸfë4§»^×ß:ë‚îU{/oRœ•^ž²˜ â$R§F^©[ê—äìØå$q—€*Ï±à¢ø)Ξü ŒŠf÷s _9M«ù†ÆOôáä›Ä°&EÀ/«n´b”¡€õž=³×glmÓ:ð ‹ÎÇÏÏZ›±8â¾úîÝ1AjÙÞQTš õƒaÁ‚(jó¨œR:ÄiÁ{«×ü<Ãêúo†ùç¾pÓJ1û`ÏÔëz-bÁž²©-{œ»ìˆgšBŒ.„\)®ЈŅ¿C |T+̼w£©<3Èü†å0˳abIVØ\µŠnã0ˆP˜Ùæd‡ÄÐ,˜/!Á»MZá­4öf·ôþàÃ/O'ï@Ÿü>Ù¹:ŒÈBÓÒ9›¶¢!¶ÓØù3Ö¾÷xñóÚ_M¥3uV¿bU¼ý…m|´Œëÿx‡â ý͹Òej ChÏñ9«²kªòoTÚµ‡^†ØîW]ÈD²ú÷¸¸LøOJÁuÚ«ßùÔº5yÞB.¯Ó]SlÔÀË·® «pp‡Þu)Í&íïËüj¶¯ú³mmGϨD£z…º––í]~»Oƒ¸¢‡)OâXÑ«åYÈWJ‰ã˜Ê+d"%B`ÀàÏ` çôÔùfý eô)2S†Ÿv,@†oaM£På‡bÔX¥P©UPvßê=Î×ÉéQ±SÖÅIIE®nœ*¢c­a†Zñat1¹øÝô!ርrt0ÝsV >¥æÝ-ß]s½M¢íŸ£GÕ~ùƒd˃^&_ìݦ‚ Á,î-ˆí`¹ÇÔ¼l=¿ùpªè^¼WÔß>ƈ¨W´)©Ú+Ê+kEÞ(ØÁ¨ØÑrCNd}h}¾$_Ò¨§tC£l ŒŽo[C–n¼òè”ùlËAß™•K¹ê9 mÉAº¹¾‹)ÛY¶?sŠá?lÖã&CŒ@JVÈêM©ÇÒqù4¿€†9e¸ÉÚG ¢Æû¦ÜgÐ-8eª)=^—6öÀ“•½ÄøæzP–üœuE¼ºìÐoÑ`«xƒ‹¹¼L¿w ‘9SjòâtñCP«µió§ª5ŽŽÓ ýcç_õ÷H#m”«±ùϘ[µlê/XóZo¢¶]Åv¹r¹¯?€¾Ùžîy$ ®#Öñûd÷5Û£Ûû*ö(çWmÏîëílΈ”ø³ÇäûíOoF¯{ðÁ…B®/¾áq™|ËhŸúbwsáÆ@­Ýø|¼]pgŸj@ùNu-ÒZö #xŽRÞn€Œ¾ MrUø —½³¶›\XZM%þ7ßXš Ó5)»÷ù¦Ûâʉ« L¢a«•ÆôñÄò­[{Õº7ÍJ/Ÿ LǨ(œkV°´êâ3„ìa®¹…Œ.ßfͰ›>ÚJ @Õmd¦]×Ó _œR¿”ŒÈú5*ÃY«'‹}¢Æˆ’‚§öÀ¬•¥¨}!½)Zâz)S–`ªQq«÷~iÂÍd‡€ÉÀÙÍXIY%©©k±C1Wy¯2c†Ïà¶%!SËÛ*Ñ“Š®ÁÄÿƒeëdÜ@£ùÞøËÑŽ L$Hÿ)yl¡Q¾¨$º ë,7 eBŠíäã°‡-Å)ÑuD$YœH÷íÏ”–çhÊx\îALŸ‰Ó/ú0?É÷ö¡p¡fyòú Ö,í ìPŠh…žë\Žc„4#¾±ÇÕ«R]\®"?³qý˜^p Ãï;øõsÚÇ(¶Xz©^µ•ƒnbVŒd…kýINO1|˜T«ºZ®¡#–ç›–Úxá–=M÷T5·à¼}É$ä<$¯‹ìÍu!*µ´ÎuÆê¯Ÿ3¤®©XÇ5Á™¦s2=|M.Ì^H­£ pšØŒ =yX–Éܵ¯"Ø EÑ–ÒÍ·RÖøáÓ)Ë'0ovv“Wï@—¿þ$J¿%?ÆcP¾CèÈúê·8½E¶­ÏG€·ÁWDZkhzËc>ßCY䆫.!Wø¬zqO5ðSð/Eö¾#t~¼ô½Ûçï\s*Šy—T]¼ºGÞðÒð…hB¹¡t¯ê.z‰2=q6¡e݃‡OF9/ŒÂøÍ¶ìqßz`d3¸qu^ü¥Ê¾éóÈøƒà+Oë¼3¿{„íF@§“˜[Ôú¹;AÁS:§BÒâ—¹ï‡Å6‡ÂŸS|‰Ù'EÒlnkÓã}Z:;*ÆûtR¨³Ep¤d†|+þþ]r#fô‰žƒ}ɼ͓p4®Ï&¦§ð­nñ‹ÊT‘C?t] J„„Þáç–T—øÉvÒí4Ãxƒ·éÃ…¿ÉE,‚·B‹òÆ#À±I;Ãóš.á­H›ãÙ’]ž›«$dõœ339dIÑ´\·AÓ»†"Ÿªij5틞ÑÇ!œ¹•n+Ñ<Ÿê¼Á^ƒ7/rq%¤mxݾºõE¬ŸUD:î‚d¾ÇŽP«ª.„=Ãw£ªýê$_KïSéH$ce6¦óÊ#röEÂ~’•ÑæºÅBBûŽ3`Èì,ôBâö=)ŸÓ›íd8C¦[„EðŒž!Ø ø v€Ç·õ.rí99‘É|Q§Lk–³YŒfÅ«ûS Qµa™¦¹ô]ï@Pd7@r'\_Wï•ÓAðY™ùƒ*Îɉj]À™é¬ ä›Ô³ë‘[ ¨”…«‚ Ž(°n륕æu;z}/Ö˜5^¼å¯4W8®R~™ïP×Á‘„h•Úõë]Õj(Þ^ïûâ%¼PíxÑß Ôv½Ü{ìÑ–ß•÷ $é®â¹ß>Þ°jÓbà6}½fàôw0CJ”š–´%)á"­€¯áÄ2ÄH°"Á í®[UÎLÂùÒhÓçø´ÏP“è ;®eš}úðA+–s™¬ÛoÓ÷—±F;LR —„¬†Ñ¬âÕø™1 ª£…º¢^aÆ×l(øõRÄ›ÇÍð½5pÁ­ÓÈþAG†×!÷—KÞ.vH=ܤ)³æ¹éÄçcoûrÕå³<…ú<>ðÄj®äOCNlZçEÙãf»yV—XÎͺdkuD¾y­ØÝK!¯œ ý@Êæ4Êê—Þ’Cj¦3Ö²¶’S.âG²XÆRØbÄÔ>÷È%RÝ âŽt}Ö·<½ÅÊmaÜû>LÄfAÁ®žn³ã7—qù9Ì«wÑ‘L¨õý*ËãÛq¼s’vøð(Ó8²ùÄÌ-yÈl°ÀÕ‰à»j9ØŸ÷rr•.¹? ¹‹ÕÏ ,‚í¦±…K¢hé_íªèÞbögÈË}ú¤¥¯i§=±6¨Š-dù`¤yžÜ.%66é4 JéìcoÎ=$8|vP®%TAš(Jduôôé(2g§Î”Ž\iÄð¡˜Ìáq}ÚEgòîù‚b ’ß¿X;ïéÍ ±« %Qô¢sKõ´}‰zôŠ#‚ó|äwéÚ_umaÞ°ù§âé§ãG¥¸-h+½Õ#[2\Œ%Õ5–R"²‡¿¼eñ¤÷ç¦áÝ´ F]³,œÆùʨʈ ,׋ËVã ‡zJnJ9 ±ùÔ»´Ì (¶I³ïäÈNžgÝi'—s*‚òû‰2Ú€Ì&€>¯Û c¨Ï¿ðn]±P‹°*#o2¹ÄŸóÏÍÅj…ÇhBO  v¿Žßg¹5঎5!Ç—’áDì@•ç?W}],Xc¡þîÏôh…Q󳻩çÛáØ¦P´ ý1½Ý«cn­iu3æ‹ÑФ¹¥­üç3 ®¾{$AœL°¯ñÌp©¶Î\¡v“õÒ­ÉÚ·Ùb=0|âØ#Öþ\íab<Øà Œ‚UÇ©Á6n Þ'Ð-P!ì7#æ$Ïõøzmû¥XŠÄ}6ŠY[¸‘[ɲÝmT~;Jï…ô5lï´XìñÀ­Ñ¹”[ZU3;Ïäè}Ùbsz “š³ó?ôÑ® VvÞNèÅâ6¤uvUN&n홦!ï"™„ƒDF·ãb*œ ïî}‚z2 0ä(‚ïwç#Š÷”Ù°ç:c4ÇõØ›áwèdxí³ãß ?Êš{+óQ„Òdm‘XãjñZÙ<'°ò6M#×M&×ßßà¹ÅfQ¸[Ž)ÕCóYZz›kpO!†¯¸-Ñ_yJ8‹'èŽ2ʼ5&[íƒ+`:£ºeUˆY§Mm"‡8Æ8ó}(ç8ôª&å&Ò5~däÔfƒ/">‹úß‹ÀK7–V…á–éS^MôñÇO†¥ò<Ñ­ äd“.#Ôö^_¶Ë/#…–&Êp:ð'7þPÇ¢ÄR£åš““H”»QÎÂ<x7EqßwÁÑ4íñ° –C]ëxàîzF Ï*=_’v—i“EÌ!„'êè…›ª›½ª<ݰŠàÛÅB`Ö)K!­š[êš¼£^ÌPÄÈR¿ç86I\á}çÈ̼(ê•`¢\užƒ.$;8žqjñ8QfJpÌÓýÈlk ™=n;úL<Ë’Z3ˆå×”ó Ûn0ÍðgkË@ÃeŒ Bk†ƒ˜ø£m¬«%7í÷+=<¶PC«1ßc…\ôZ¡Ã!BG[×°™éé/?<µ8Û{¯M÷++de65)„SIL(N µJß_TíïZY ¥‰-ÊÖ^”Á*¤ ئ½Öñ2Yâùœ yìá͛ɵòã0×ÚÄ@°YÞ섚BébQ„} Ñ³Žø#*<ºWîéäq]Çœ_$*U&ߘrvlE äª:Ì}3oOÊÃ;H´A0â¾Yjf©>Ü"'Q EïZ]SÞ½2ßF¤™u”Ü o3‚9cóIÕï¨ðçØ÷äU¤gàƒš‡ÕZçV¡ìÎþ`Õ2Nñ‚LrÊ~¸\5*ý‹£]P¯Tsà|é´|]AA“Z„PwŸþázzJ—TÖêûQ&ÊÄ—¸(±cCªH[ùÑh^m,˾D•ˆ·{°ûr¾FZÊŒli!ŸMK„Üi’+œêÊ=”ÆI2ÃÎðÑ+‚Þ»-ßóBë>6ù?ÎzàÔ'ˆù Äê¶Ž¾oº»>C÷}cp^¯Ó‘Ïä TÕÍj¡ò–%ÄjŒÌV¶ü¹%ÿmyñáf{õ•ÊW¸}™#á‹Ã¯¢ÈŒ6_ž ü”*Ö}†;Y* Ÿ2üòd xˆIÔ5¯š˜ýÙ‡ItJƉoÉ®‡Ò…ŽNCƒ!#q.TŸ¬7ÅD­£‘2p,(/ 4-uVåܼ Sí‡ÙÃ8BÝ\â¬YñPI$sƒïG%î¬?Þକøù8½SðK#©n æ˜îv€ç;'ÓŽÄ>!Tcl]>ÔŒÜîåy}Ôz£®·£/â'ã´¾¢Œª¦µØˆEËê?ò{2 ?yý6xJæy)hÐ÷ëK¯RߥÝSaœJ¢Ewo+à”žÒÔì‹·1‰K…aÔ1i5ÇØ{^¸&õÕ,°Š¥x*‰ÚÁ¥Á”?­¦‰Í)Ÿ×²_e„D®¡[õT6°Åœu±%ëõÌ 2I8bÚÓÜäÙ¤'—mvˆýä]÷"“;Fi¢cå°°ÌEAiþ¡íœØí°íÔS„Ë<Ð;Î;¿go€µÙ‰éÚa¨ÚW¤~E¼•Ü8I7œëþà$;ÀBöѰžÏS(Ñã}„÷‹ 9^u«~îe)CÀL c›b¬ ¯„›€ ¸4%âL¨pMj\Ç3QMÿG18”|ŸØ+¶hxo¶¶–LlòþÊ©¶a˜ÐѺü6O°æ\]·lí¨d÷OþVœƒOéLº¹BŠiß«_*ûÝ4”Æøö?‹~æ‰yŠxß=X•cy&p¤æ×¯ðŽ]Ô´[Ôµ¾·Xx\Wz+(6ëò^W"mY˜²çW;Ùçdg‡föÝFùo w ¢OLsbÃÚ1saÏê™…jÇㆧîQê˜ü:Ci[x})ñmÔ›•Ó+S¹ ‹Ö`DÇDM¾÷EµIœc¥K%U²-ý4§T+UŸRæ»áüs™G¿=gK" áú9ÄÆ[|Š8Gn^ǵ›E¡D‘Œá_›ËÚ;=Ü¿ÙPwd®é))oÄ*ŒÜs”8z[¨p:9GŽæ”{˜IhxI5¤~RD ñ¤jYÔGäH/¸U»žáî®_zµõ3ªÜ|ujë>r6 ‹¥”­@weÇŠK9+èq‹Î4úå—0:ÿ@*ùÔµ /ý ¯åPÈŠé'à{t›VÏ‚s2ܪ¯QÎäq–TX7Ígr rç :ËðF< #Ö¾4žæ)Wüa* Ò‡ë9`IzG3!®|Ž91÷r3èn`˜³/„[•@oÎt¹ÌDZF¤ÙF™Q­sT¦‰\L·HEÍÕÞFkP‘õN´Ö˜‡q ‰+,e‹Øå_ƒž5ScË×=ÎՃΦÁQÁkP{6½7»’‡ÐÁ<ké™S!ˆwî.ôÔ®kè`΃‹@P­'ug¯„šfö£Øe™W+-Âe”—_Ò*“:Xgl]¯Éç=s]ÁOŒÊ~Ú .Áb9Rd‘ ÌñÌ vÖëų^HüiŠÄè5Ö[7äyãoçd‡uN™X<îir$¢´Ç”Š€Ç¯ÛqA?‹0’ê:‹Ç'û–ÍgªèàK*:ixmnÔMp§#·Ø`V³tkt‘œÉl>p´Õ„ü¸‡× Pqtñ¦S_fkŸŠÒ+_xø%ž«:w\Õ¢¡°[ÔÜSk!:œÌÅ 0„›¶G}*Ñ*vÚíIƲ¸Ø±ÎªŸm#Y™;2gƒ6@—{5CÊŽj¾ß¸º™#0 5ÝØ:q¹}m, iJþ‘±ú=P“ýÓÞŠ¸K“ò²ë3ê—®å”àç!«ÇìÌ¡o ™,4êyu€\ë;9Yˆ”#ÍÂøM’^¨NÂNä"B¸lùsHîïˆq>¢ÐÏN—š b…×f¶ Î|Ñ$ëP°øZ4òO¹ÆÕxX\QTŸP|¡"º}Îq0|a› jVU|å!&#º?³Uéb|ƒCµÊ&”¾”Ö4+O9„2½A’a%;Õœ} ìdêsAÚkW'D>ÚýÈ †DÓ¤‰¸{>&¨Ñg2À”„ß¿¾©J6‡˜ˆì-GX¸Ë„µ²AÑV¹$7>2o¶Ÿ’FÃÔ£Éõ*¦‰O2’b±æÀ2±â+Éa<Ù›qò`ûúŽ ÆæáÉ»©æ÷nç¡1¼yŒH×÷Õoá?Fô ´†!¼ jPfö+DÀ¬96@¦Ë…´Ò‰ùÕÕNŒZd¶š1ÑöbŶCG+“©à–à E#wŸ GAù =ÌÜ QÊós‘IS¬!ø^¯Šµ>ÉRqüјRðÄ[žŸB‚ü”A#²Â]³EkAaKµ”CüRÄr–Ž3õœ}í庩20åfîyßdJ9"¸¾}))†Ÿáé{®ë™{ ÷uO<"gŒ\pÒpáüB•(w‰)!” 9yYè"… ª–ÿÐO΃®X"´µø<ø¡ W¦¯½$k”·Ð÷\~/©áû$:ªvs7¼V›·[SV*¶Å‡F„‚ÚA檉6ì’Z¤ êm›@HêÀþ¯¸‰If2.ˆõzÊ1„â&òÀ޵„‰°¤eCä}…îÏíu[ºø®™xMúT³!—ù„¤fêªtu[f2K0¸Gj´ŸíPüˆ!Ƨ›ôTÐ^ûÆÑhÇC^É¢8.ú¯ûx±)0äõcª¦ E·1%Üï„G£H I£˜ÄiUMxçàð¹Mo¶Fa¼9Tˆ¨Kzó­íGã"#Ân╲E‡‹èº`÷á‡A–éùQ¬lü]~5%˪î|lèÄòÞf`æcsµŸF+Ï8lmâfÅöa|‡Ôc޹̾¡J<&®ãå®îP=Ìy'_øöÝÚõ#+2§çA3$‹®Åà”9Õ-‹­V7}úæ51Gê˜*h–kû†îÔ?×Pø|M£7b,/˜ÞøàК¬>æjf“’–ÄáùÉ×@õª@•–k b¤Ð‘&‚}à³¹Ûè#2ÎUbŠö£Åž¹­¢U‘7·¢BÊçwsª¥\Þ¦šcØALÐ×ôÝÙ£wÉC:çgôN›HÊ›—CR3;êøo˜`Þö¶Þ™ò©e$f̨. ˜e´¤Â(t]$\YòO°£Ã• 3¡Ûú±3FëRƒÄ^„Dç9¤;´_»?Šñm(1 OÚcï¶úbÖv‰ƒÏ+ü"³^˶q`ʉNÒÇÆ×=èÓ –êYi[)*ö³M@û¤u§êhn‹/Ñ%'ž¢)êzz¨çËý…&à;å/„”B#é£h*ZÎI?gšÐ“Ü ÁUÙ¯‹Ð§qGgùÖ—Y(ŠsÝüØA×ÎLTU[æÏßõžE2}épØ|Ùæ˜Q. ÓV™Ã¨ ôžÊëñ\bûW3¶f÷ÉŸ(t#d¼}‘æ~1sôß&0Êu¾2ÊVíVÈÙbž"<“"«]Î;—\6ø•@†H©%·‰»t¥Ÿùr¸ßO·¾½KØÍ×òѺRî1W96&àêÉL[Çtî¹A÷ƒÒxö“àµkRŒÍ·XÎ7F•|{˜ŒƒéÎGÞ­­ßÌÒ!; ³X(ÊêÙu‘³š“j• c°½¨´Í¦Ã…zÿf#çöÊ£>N"jõ‡I.m[ëD»¿|g-Þ{ÀŠSê¢$8I÷:)Þ%*wþØ»õ÷Yå„ÒRžÞ1·²i PÖ!EÅTÒšß-93¼…•½ÔãƒS8œZqjú°>šÍ3}é~ˆ‚ãfÊH> Ñì¶AØ",ÃB›üBÙUdÀc° .¼úM¥f3Lâ. Lf/¥oNR«¯ %Pl…®ÝýWó6¦j'9ÑÿþÕƒŸm‹4×N±2ƒƒVûlÚs· gY‰.KׯÀÇVJ/d5k½ýB##¥Zî•®a%*Ö¥mĉØpõÈ—)ãŽld®ËZs˜É¾dö.[“/ Bˆs£%ÌOŠ9… M÷¾!¯Í ]~b ™¿ #™êXó««ÄAæpoï(L¨Ùpëì7ó³¦\3/í–Ï®»ÜóäÚí@mÚÃa_;¸…‡q™UÛ»ìº V úí™<ŒºL²Æ‰p)ßmÚv —¾Òd¹ÖÄHnl\ùÙÛW4+ß륥«Ã°©yéò0ªÊs„ÍyÈ̼šŸLÊœªua U‰†Ÿå:¾È¢W<ÆajüzgÔÞQ< cô‘S»ƒŽx¼çÚ„«¹ þ$x±Ÿk?v¶z{âŸ^&úFü|0…†;@*#Û³)¯|G_à–Æ„…È1·…3xèh䯞Z²*yTg}Þ-µLׂ6ý+³Sƒ!†‘|ýRµÛ5Î4á§øˆ:Jçƒà h£Á€úì­„&õh:KE4lÄã}XÕîÌpµ åʭКÙtf©_Žª­ÏeaÌ¿ÊæšÚš:Dí)á§õï‰×åK3µlZÝàV-ÇA¾(p9‚Å+£ý6­JN}rÀyÌîÕÄèyßü©!$®>ãás¡F2 …H1$f-dNo]7¨D¢ðuá˜ÔFtËE×u¿}Iû`R›Ü«2꜂fFW‹gϱWR yèé׫xûvÔíi“—믻¨íýI×9i:ß¾¢à²âŸ/Ë.\n÷èv`æÜC{6ŽìþäòP½8õwh>ïˆ-MŠ‚æ5ªÖâÌuÁÇ Ç£¼¬…i"JÁáXxS†EØì×âÞ1X å™Ã)†¹Ø`‘|ìW7—Ã>2šiªÙ‡'m©F—o÷ó¬ab¼ßÌÓßíu5bÌ,x‹ƒ·@Eƒ©Ðs•ûž­‰q¢|qøÇ„ãAÉ »Çqû€(#Õ²¾Â‰¶¤—ÂTÓdò0Ýi¿D¢Ñõk»%ÿ†7·›€ûÜÔÎßLUÙ¦ôMÎ`Yxc;ÝÏuˆKƒéYŸ©I€j^f›‡q™ ñ7[CþH,<0áyÛ È}ÿãBÑv7‡pRL.Yª7‚åIJxÀŸr ꛪÎû‘oÇÛ+ifâ~j˜å"³¨¼)(Ë÷ µÛRšæÇ÷M#‘XÛ]šÖŽ•éó!Sª3AÓ…X_zçbZœð¼òPðœÜŒÎƒûöµ‹£–%q¿ ‘¿S(è«>Ú ù"6¢)Œ>D?kqþ Ðg¨…,ÇgKüX„aŽ"j— «ùœ$ ÜÙªe#‘±ˆIG»÷A€ª³ÃÊ+z*%"sæóÝ›–¼Ÿ›íËCåË·øQÕuƒT¼RÂq“S&l‡ŽÏ ±„aLŸµÍšÎixN²tš–>¦ð 0dŸ¸±Æ‰eDàÇtù3h,T(¡½Øiõ£ÜUfÿ½ÞD·“YäP}îéý^º”Ù‰r®wù,C!Ú -Q&Y2'óˆc´|Jh\2lzbs„vK*R †O‚Ä¿óãÇÔÏ4ïUØ#í/õÔùƒŸ• #‰“á{o™]F "ô´\ØñÍa4ïrº añ ®²KßÞ¨‘©ð±rXå,(ÜH$FÛ¼Oeu¹"P›OdDEª¹ºî@ð©ºMoP}»X}l1#A ‚:¸É-Ôfž¾Œkn¡­æ·nlU#&щòá€{ £“ÎèW3 PÍ£ð‚E(ÚÄaE¯–B˜Z%M²£_÷)[jèråYÙç^„9öÿ $÷ÛÝ.Jó4?=ݼ'ü€¬_ÜHž ´ÀÚËf¢©QîóΧÎwceÑxAq"™yEgñâËæk¼>å´DPÊÓäánvärÏLýTà×Ð÷šqͺŒžO_'Om8m–D*Œ—7MÓuGoxì¿/Ôï´éä‹>›9q›­1…³ôû"C­¤ã˜-¿bpP9÷͆e)Sù ÖÕ"Ã’ó« *1\÷(žð”[mÕï‚䟅ÛèË‹ñÅø­ÜFoçrîzåÙ…ð“Åín5vÄ›N—”-ñr¯—a›N`&öru—É)Ç‚¬f-i& oûÀ’…—Š{Áäß/¹™VçíÔ{ (Ú*ñam{ _f8ð½FŸ\ï´ä÷ÔKäX9ºáî1RÓB?õïFo¥2XDxµás@B2°Ö¹íÔ[IÚêŒní8›5E:ÚãW?g¥b6Ôœç®ôÀNµž°®x«ýe}î|Çî ôÉú, ‡ˆ€¥áïbù©DC!µ¨-›« %—]rEšÜ;¶},2ã†fåsÌDÛšLÓSxÝh´¸²Ãˆˆp°nðÁÚõg‘.,J‚+Òœ¤k£ØúùAiH*j/gUÔJáÌRJk^77ø3….œÓ8Ào˜Ô]!»IÎût·ÕñÌ¿JÊ›`£À3_Õ5³Dq?¥(@móñ(â¦+,ù‡¿{J{€cF-í{V)ÌnÐèâÂþOÑÁ4"NWÙ_ö‚mHÉÉA-¿<”mrÑqÍ6ÎÖ˜t„}R£ðyüÁã¶&Õ"1 âU4ð9è æÌâȳ: æ$¨ÞGb:ªaûo¡3]õ©Êcbéy½T…\1u&>N²aöqŽ®RÉÚi?Õ&ƒRÌ­Éêk ÄÍÚxDNV|ïfË|h¶,kIX\‹)èzU»´¥‡oz¡ßš«ú2ÇúW~ûgõÈßFýZø}]þ¹iåÏe ZZÝ/r¸Wb jÄǺ:uzülÊJŸî׫ûLy|úZíÝÎüvà¥k«OÒÐQZóºþ ›ßúD|ZC%nƒÙý87u’ýùzL´5Ú\¥6ÎgsÆE¾~Z{ªê,ߤgÂZüÁômæs†x•%ð•UŠì*’chƒ™‘B1ʾffD²^÷ìBDU)6KXý#mœ4ؾtJpƒŸ…˜¨ŠÑã·¦ØËß`kc:ŒØ\ 6ñoYv×xˆ§G€æ¾:—øÊÓeóî† ÓcÌË6‚Á>YE TlM9à67šRq©ÁÎ]¢ú‚j€ËT`D«rœ_U'Tn¹›YÐF@ƒŽŠ×ÂŒ¸Gìu°8†Ú6âØÙÚdÒ”RHeô¥žýxh;ÎèµÀ'ï™ÓÊ…ääq)kÁX#Ú—Ài›"BÜ™L #i¾t*+åà…àxaœ Ó!Êݘ´®"®'ì"E•TRÛ’’`rët˜#æ1&ëÒ†yóâǪMI&CIëÄë®W€«)¤Ä´&´5«þ_l)³ÖÔ%ØûÃ'#n‰l{’oY‰Ûf!%âiè Z˽’73‡º‰Áß3‰5QÄ]×Mܼˆ-”œÜ?¹-û ë20Ž}VuDÜ.–¯pzŽô¨Â{wÊ+vu伡_Ÿ2ÊúD•RÅûuPŒ‚ÔøJºù›¨åS“wÕ‘3O©2”ãl:NxÕ¤ˆø9,õYÄ5…îNúÁJÉánã©?å.™­£µÉUº=õðËv™KkbeªdZYÄèìCΊNYŸË²ñÓ«Ô)°ýýg½‘à}IÈ…«c¨/Î ×+ï ‹1ÔÕ¤?r·\S\Xd;0i¼]»Rˆ\Ú4Ô(Í®}:`}"Ä¡?EË±Ô Ã.¾búÜ36ñÙ‹Úö’ð‰u%»#RÉwFäTöÕJÞÝ2ҬłÖûDhqèŠo€·ê¶# ‰L‡zÆi3tù.‚ ‹¾«RU*×µþ'¼Aúòùœ%#{VdF©““ê#¯v˜‡@ ûº§¥­·³ßdÖ¨èU|®”/ESaŒÁÙÚHT¯íz(þ’÷k¤äGWDP:V?®3È4KÑø)¢,·àNÛës\yÜ‹{¹8FM›)³ç³çùÒñ–8\ùEÖ¯eLòKUü|ÂÒ×ë]`1PþÜ¿(eÅ„7V)†U­R]íŠhøâú0ñbŠ˜žù#0§ä»ð[ò,VŠÐžé «QòO•³¡ø"RP]Wß”¦UߪüÇMÕ&‹ÿ ž5 Ô'1fw“¶ÃŒÂÒý¸ÄP]º‚ÛÃGö™ÂiM2äɈ%òï0ÑxÛÑFR*ÚÃýšëµ,ç¬ß­zmƶ0¬—Ll áà/úigGÌ+›É¿ö1¿yf‹,Ûj§’¹0íK ÊÕ?5YP]&ÖÀ4Šk=Þ,ëÑõ=J¯¨Ö·* 4¶“=76.Y,%L}‘-’„ó OÖ¹„„H|¨Uš¶`a˜Ñ~ᤑ™«(ñ0FvQé~œ£àp:6„ièà gpC£ƒ‘¨aè|ešÄüc†³ë ï #ÛI'ÆQ*9¡FA Büª%0ÛMrWª›×º!¼-HÝ×-îáqö-›„ŸÜ(Š«uÉ" í¹Ð´Kìº%^hc4_=@ÿnØÜ7vžY¥¬Î“¡5:Ò×T˃§Ý>Ç<=§P…¸œÁ[$nŽÙleËt3nD/0=B™ÆëüöAûô·àœ€…(X/õiUÕªoù|2Lé/º‹ºCòºvÇ“÷•е„—°ßoЉ†‹æœAj1¾ÆqÊ·Õ[äвªÔš5ªØÓ@YiÔÇ:⨡J+¥,hØý„ÿ`¤hdûÈÁÈ1l/â 𤛋XPgæÝ‹j¾EÛ»”Ç$ž×³|}Ï#²«¯šå=w¬þ‚µ,zªíËœÉÚB¹nþb:*Zh¬Í9’QáëîW¢¥Ã´Á÷¦q¾å?Xˆ¢;®"Rò¹BGH;ÐoÐtØã²×uói±‚ôøüاµ`Ÿ~ ^¸©aC¿cVƒªðˆ~>ìxršuâÆS$b½ç"qÂŒ= aáI³Å™ß½!ŸíäDß1H«Ð.3ì° ,Ž€Ün¬VÐtt7زîÀ.Iùu¡BKCùW¶§ç?÷ ˜ÝÊV£&b3” æsê6ÿãRŠ—ØG¡ä\ n‡K6`øþC €Ýß›äçð¦ãd®nx¼N—‡‰ôƒ³úIq~.%!¥ýÁŒ øÈ;ÆÈ„‰è›À™›„/µÄõñ+Q#(šXu=³cP„‰y©÷u­Ù÷¿g¬&i9Pœ|Í_Ŧº3·HöÉHØÕê(46iH²¦î“Z~ˆ­Oñ¬‰søõŠ#hÁ§”Ö…nÚƒl¼DĉPö48|ðG×9® °è`£ë(‡°A윦¡Îjn”<bý;„m8Í"ÌUÃv‡Â`ž¤ôCioŽ–ÔQù2±]FÛú®šý_Ûž$¬Ýp3|fÚuÅ>O±è×:zoA7²Ølœ¡kê„Öhû×re‡µ:òp†ÆyhÅ8wO%ˆèEt¶ö¡Ó“ãà9õUzo°Ùôì5ˆOõ£ø#"›%ILhù:Ë9þCÛM‹ý‚Fòh-=ž†ðìi¿$áüçúdEŽ Ð|d…²9–¾Øñ½’×)b9!<Öó®¸M0"ÕžÚ„E$mZò¶­µaÎò#ÇÌÏ ]狃øMî4Üé^àŸé™6•ë^T÷¾ãý7sÔ.£^¦w¸ Âh…Ïü aŒ§´œ`Çu(Ò¹Æm%èþòâ’‘ŠàläƒüÇj\M¦èˆA›˜ÂRò:”iF Å,w°ö š\O!”J˜eyÅZ-¨WèÃGCø«­[/¶™HæðLr}ËÈY1Ä0*š'œ:BFMiwßœ"%¹”i£“¯:æ¨PÚpŽ’ÀäÓà79K7ÎŽÊÿÞ&¾”‹Ö?¢³*ÑÍ:d*ÖN»ÉñŸƒU·kXæ?çiÑç"€(<âµl˜ßU×—_UÊõŠ—Ä8á%ˆ˜ïGF¸ƒë7G÷jÈÙé X¼Ô®Qð\¹œe°ƒnÑ/OÚ|Üãô x·/ãn'–ˆ§ê‡^Ä LøºK¿ ÀÀ8q/¤r-!B*qLÄÔ+ÅÈGMK 8e:Dn;m¤’.aGÚ_—еþ¢¾v“Oå‚'V<Å3Õ»pÑ0Ÿs y¯H²1Îõ˜h¦R°æÄÛâÆvÞ|±­Ç"Ü“$­>6|?míÆö ¼J×Þ¥^:%k‰ bÀëø›­–u©•ôæ›i›mmEé5©öqJù*\=ç¤eéç drvºdò™—æ<•4ÎŽEE3 9ÐŒÊ,f›¥Hžäѡ޲Ë"µ«JöÙÎÃÃêm±‚oÄIx2”à°¨í_5êŒÛ¸ëá–NI\ ›ò– \;À;zG÷–ÝSÒU¥Ìñò‘ð… ‘ó‹Mw1G!LfîçHFñcuFzÕ³w¸ÂÉ'=_êXIv Ʊ¢=¥uP×V#uº—ÃuuÓŽŽÓcõ‚WàØ­qß;¯áiŒ+–+l%3¯€ßjK»Ò)°ð‘x×>­ôð#ÒÙH“°uÓmúÜfÒÄ»|˜ía¹dÿ!,ˆ1§8Ð<é¯í ¢ýŸ)Öý<ˆœˆ.²ê¡ãb‹çUœ²½_€ÚqèJ=j0Ú]Xôò ˆÃY4锎ÞÚž˜œ#çš7ÛNÑ鮿ô¼c£Zæ˜zÆôhç‰/Aú¨M1•“–"3yàéaKBÜŸm†?òÁLÌ+¹ø¥¬@ûQ¯z·m›bR„øØbEW »ÏN˜AŠÑ© 'JÏú÷s±óõ…²¶{9Bac«H¡áQ6ÚrÜm`üwŠ7x¸Ø›c &ÓCâîÆ¡dÊXÂ{ÈÃsG‹@­¡"ÌáqÄÇ1(Þ«5«Ÿ¯()3ùóhŸŒ¢gQ)ïÞå@ê‰ÿS1K¸£ð )@‰èH­)[ßÅó羃يü@1·r†h®ìÿJ ×Ú¯Æ!ªü(qî¢È3vÆî>°K /­ëÜ\& a’Ã>‚¬J*Í·uUØl4¼XžòbÖ,¢{ ëJ§ìJ‚8¢¯PþÀýߦ†^‘Ÿ©­½êfË€iÃ"ƒ6|±÷¡) *"­Î¼˜Y;ÂLö£WΜÉbŒ}“±\šn°¶œp.xÆ›}ÌèNCxG™Ã€z3¤ªˆFñ­§Ï¯AJ¬§{²>íè¶œ˜kÕ4O•¨×ÛŸ´å)óÇB…UUX^#)~üÄny²ÿ¡ÙSK‘|™Õh¡ Œ( <Áiѯ’Ò°¹lvK[&$ÚTšØõåÚt° ¿¹Î¾üÆLÞOµÐÏjjjtçõàp´q2%ìœÄSÚ݃ü*äÖ2ù *`®7 {ôÓíÏg¤ŽT(móoˆN.¼ÒÒõwÚo0TÍ•û‡ÊÕO+3¶P„ZQáÕ¿.‰k>ˆ“‰,;-Â}T«U#ú§z†Qï‡ IT1îê-¼Z(oÿ&AÝq~äOƒËÑg_½a?©îÎCZþ;ìÜoMù”+–Ð-NôÑ+Ž’€œ´l|ul㽺®K!2¤®ý¸M,žžòàmnÒ¸0G%§²ž€R”“‰ç‰{‡cUŒ]ßjÆÍ7Èý†öcQÑPNÓ’Uy`Å×Ù_PcÈD3ÌY«@ÂE &øn±A^»¤ÛXí063xøë´Y–ÃçlB¨Á'ŸEKÖ%kw€šË8àú$ÿ»+æøÊΉ*º ::·FG:Ê\ç=¥¬Z&Š\]RLð¥m|c²X2~ ¤8³%b,±1á¾”Ä­É zÂÓ®—4ã¿ÇxfFr‚:2h»¦ÊƒöÔôö´:D4ŒÆ|–ÕDV-Ë;UæoïaTåÒª¸K¡ÈLÕ\ƒºDê~ü°+Q*0¼ ¦¢UˆŽÿ^ w¾2G'ܨÅKoôç•bO«R3mÿÇä…д¥6ν î‚p°¥=´ˆÐ¢[NY6¶ó—r+8 C¢{Tl’œA\Æ‚Ýaè \“x¨W%þ iü™60ÐÇØ¶Åiø–·[øÕþt8€!_],SI¦ÝN³·ºÙ^ó†ÖÎûi ùD¯•ø8Kñ£Z‰^A†N1 ó¡µ˜T'j˜ýÑV£­Lœ_T—~¼?ƒF5r¶Äºvezœ‚ó4ü'¡Ø/š4‡(ž¯+ú#!`ïóÍxœ® „<ël¤˜¿çŠÙ𒞯EüØ<)ñç‰èHÜšÎ0÷ÔÍ߉)êbø¢|õœÙd-(¶q_½>Òéè Á}„Òäë;6çùâ‚°~vèt:æu¶¼ógæ¨v`úáPßÑxÓÃ5{‚«PÏ^OÆ0õűý’ûÊÛU³þÛõáÁ³û峌ùõNšÖv—f»¼˜Alìm}(Ø[(Ô ¾3I„;«g’ãò&å^N*nü8cU«ÞµUî^§ óoQKòVË­IDÆh(Iÿ/ß½yaÂÃcÇ+£PÃ¬Ä ÏÀ‹üâJ²ÑÜh8xp»¨S¥ãÙH©‡»4¥v–í2vüZn×$îAlæÿ^²þšÉbƒð ˜‚¤Ùfb¸ØDã·’-KnŠÏUŽÝ]ƒ»¼qXeåÑC…á¯ú‰žÂ7À„ö/à_e#—’ÅÃ]Ñžv¡`ªEMåt©ñå̱¹¡ÙÝùÙ œNt+mµcø»1 2>¤…¨-Ú5Dˆhiø‚½Í10%@™yKɼd"ܾ@×µí¿–öwŸ^⿦m/J×TH{ɾڥ‰hP~@%É[¿uw›ú€"š\¼v¢Ü â]\ɾ5/[ÌDŠضúTŠ֤ݟ½ü:ÛXyYÐ_aó¶P5ƒøBɯ\RtZ¿·¬æMÓdü@Å\ê ?«%Ø1ñ .3ЀÈc]QaDþÉŒô³n3vk,`Rú/~o[x-éIPÖHª™ÆâÉ ’ü1¸  Æ­‡5¢§ˆalí|<‚šì‰ÂÐy²•«‰QtYã=Wñºr_ÿ^Î\ŒÝ M¶ˆí|nSbtãýjPç0ŽëZqÅ‘e¬ŸZ2ôúGš ­çÖ\DÄ^n8ÛÄ?€.¥kö¨ ûlgrzL”½JÚ—‡ÜýQ(«‚.’Rç»trØ}Y*ï×tö•r|“Õn} À\ œoÛûáöól!ísß"ÏKâ§-ñëŸëfÚžmE:Ø} ˆÎ’‰W õl ¼ä38âzîéqECj¶œWoOúC²´øÇ¥[=´æßocÐ.’±¬nî±6!Ýy^®ã •Kr—Š.(ZÎcü°ˆ1ìLxJ±–q ðð˜™ÌÇL^‡¦[ž*$?J""€²ð†èމ^_"€ÅÀy)¢–Üê˜4‚m»­_ž—·ï‡ø;·þýöZf²0 —.sIg)P¸q®™!˜Åà§oüêéÛ—ÄãÌ*ñÖÅí:/6Å‚y-,ŽBŸ‘%òÉ(à?ÛÔ]ñ’õîìf‘ˆâ,ÍÄ䜖€”O3Þžï Yé3Ju±«ü¨¿æ_Z£ Ô+JÙñG÷ä>“rz›šÎÖ?V–g!äö$Ïv5óýpNn¹z5OjPC¾ÜN(yϸÒýC]ɬ¹mߦ¥*…6kÃ/äZåTœ«É0]it2‘ÛËîXË}{f¸XÀË; ¢´ôÚ3Ï»…ìOîÖ¾.T‹ßDtšôi §Í·ëªÊÈ3¼ìнXË:×Î…+îê‚è¢ ãFjÙÇúxjèe_6éµ°ûÏê/OÉÝ0ÄyÞ©Xy žic.äôã ?È:Nu¶knå ^Kð´MÌã=¼S­ ñòºÄ\Ñ„ÕqfÞî}ñJGýÆãènù?!&„²†[¶67&Kœ]b„Ð!QýÇœñkBNàŒÝ×ý\÷}aRÓ»’Í´âÏ8ä4ز#2qœðø$ªÀ¸¶ #$|޵oý6A$©$ÂéѲéXü<̘⌸šßç‰æÑ×®X¨q™2¿[ㆹ‹!ÕÇ)û“¡'áº7õ×fT@–Çm”1ŽÖ¿öÅòWÒv#Ù ZåÙ¥j¯möþJÒ@9ÕÔ ®`ùÉðÜUtZ¢Ô ¿˜ÿ8³y¸³SßÿSZ~\oíêl$1­K×è`Æl†uòåiÓƒ‰‚é«2Ê΃ýöqšèȿ܅DÊüRT@cÈŒ[½ß̸oŒr:LÇ 5s¾"¶¨öqJ£"š¡,Þ©²/¤>;㼜‹E{ÅZÆ)´94Ph¶@¼Š¼U4àu•fà@åsö!œ‰V/šÌ*(.‰¶u¿(ä"lðC’ƒ«“)މ›O„w|Zê&ðœ9w4±[Ó‘Í~ñ²âßĶL=ÐÔÀøÁ+Ã6~µÁÝ‚„¼¹—A]ìïzQ,ï»?6Øß÷ÚÐÔãD}ÕPbz@£IéPÇŸ‚â˜~CªJê ~¯HsÈYlóÒk²%ÞBl3ïöqœ’Ì8ï>È!ÕŽF3%’Êš«½ÚøÒŽk¬Ó†+iÂUoJÁ† ëˆ8òš…‡REMe¬u¼eh'Û¼4¥¼ˆØ·Û¯½âaýUªgNÆÃjN¤$id½1(ug0 `lSîž„ii‰ãÈ}‡‘Ë(b…KÈLxa_æ~j?)…ÇÄbÞ-l±¬Ûzo#RÕƒ§Rõ£{y“‘uïýÛÒÄÒ+?Ò‹È{ d®È¥ê'ªøµ"„¢wæß–€[ž9M7´Œm¯¹/ÁR¬ÅuC™ãÐV_TêÏT Œ@>‘uè—¶ç!Nùës¡ˆNsÇn%‹]èbFHùCÞóúm~¯ ¥æ48/Ç/ ¦¤.{¥I«}—ÿW–x~Þiᢲ7%¡.´b’é;O!/—ÀßÌsV½lÇÔ¹½d&Ym¾|}îä®0E£ÚÚk–é¢ùNäôùO¿q˜^‹ù37 Nèl‘{þ…©É)7óOº[“W>¡ƒÄW°x°ù£ñAODì½ÔÂMT­A uÆÞ¢‹Òâ„eùM€mºÕ£¦÷/\$õÚùj{žg0—ß1J6érˆ·ŸIöÏ9ŠãÕ`H™£mÂCò„óKøî ]‚[[päYš1Ìk·{b*lÄßÈóŽÜŽ-¢‚@Ñþ»OÐ-œƒ?AÚ@ÑvS.9¡óI'fjHÛóèÂ=nðæ°Ë‹BP˜ñº©)×ýÚ¥#Òdû/DLíU1©\‘cÎØ›2ƒ#m ËLP!ußLR´Rx¶óÛC«D ”à+¨ p¸ $ŸœBd11€5|gמ˩-ð)Ü+¬¦žžçظq7¯C€“ųï&"¥rJYNóïäPzN!d#PøB™jE¿!5 ·%Bß•:ˆȇ"èæpã‚ör<§ñz‰Ä®ƒs‚ý ÔØþ[`-¡èh÷˜VŒaq¦:\@ÚúÁSK–³ÙßÐ:ÎTIg+H{¶‹ÊÆßŒs®ÍXd;ú˜ùEŽ î þfÀ)@¥ës)È;sþöè¸C2¬ zw„–M·^kp²ÄϾm¶Û¨—05h¯ˆê>B h°Ö±»rCî‰ÖVÌ:§sQ‡™n {saêä@Ñ…4tÒ5ç‚¢‡•‡aõ°¼pÀ ¯tU•,ÿ=÷¨ˆ¢­Ãf®vøæ¦èä¼ðî$2/ žl¶À+]I÷Þg@6ÜQ–foþ°M3¨­Árf‹'T8iòUÀD„}ŒÈ-ü5¾ãÝÙ)MAn‹Ø]ã'æ>Wóp–*mÒ©÷`M]ðbß®%o¿Ý•yÉ”ÔûrÀjZ ®ƒZGð‘H97à0íúǵ¹Ž²ëd½¶m¼Ã>ÛmÍ€Za ÷û¨¾>éM4Ø-¶’ŠK£p¥)¦ û—äiÀýPþ4¸™ à†ꎷ‰0_áÒR§R†“ÛK™^ôœº½7~3êî#5¿L“¦€-¡ù¾?Jʨ82¾²¹XèßAm¤ŸÞÇ1,#“úOþ_ ò¶Ñ\úö4œãkéV«äIäÐ\Û'Güíéχ.e/Úw&D?7»jÓ‹|SE —/A4ÄÕr^çˆLeès/dᬤ]e¨cfK‘ö_Ûš¢}¯_y)GKWÈÏwL¿'×K¶;jÙ§ Ü7"q¾™öOèŒ5‹¬Ú3>ÐŒ×1ê^‚{1`!Óˆj]HhX{˜é“ðô0Íùõäß]U†‹š³w%õÎeì4—ÒØKö"Xoß>B*†‰ÓBÎKÍeì ådÀ:¥À&ª†×˜jàn[Í‹déàü% ]a¿>_Úm䊒d¶_Ùç¨À •dÿˆ6~w¬áB ¥vr'µšØ:o2¨"‡™®¬È{8a%çA;}‘š$áÓÙh–zGe{­›3%ËFÇ­hK{á3p‹µN‘¶ÌÏ=æw §å¯ðÄ"¨ñú‚â#1ˆVà$ §JeÈ“>‘ëÿÁ±ÖËŒïZ·cwP3G˜Qð0Vè5½ËÍA*}j“p~ÊÊ%^˜Îv_Úï0Ø­»Ê»;y­ Ó4Lv»Š¾ÿZ*äy _‹D"¯¡¿v42V2)sKaÿ RÖ¥¶§ú¤dWá„Ã'z-š.Áà­>é—Ùq5©9—Ó9 xb[ÞvÙñ›àVëÒž?áLQ§¶ Q%Í9nÙþc2‹·Í¤|ú|Ü‚®œ† ç†mNœZjȵCzù=»;7z endstream endobj 1396 0 obj << /Type /FontDescriptor /FontName /LYNLCD+CMR10 /Flags 4 /FontBBox [-40 -250 1009 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/Lambda/M/N/O/P/Q/R/S/T/U/V/W/X/Y/a/asterisk/b/bracketleft/bracketright/c/circumflex/colon/comma/d/e/eight/emdash/endash/equal/exclam/exclamdown/f/ff/ffi/fi/five/fl/four/g/h/hyphen/i/j/k/l/m/n/nine/o/one/p/parenleft/parenright/period/plus/q/question/questiondown/quotedblleft/quotedblright/quoteleft/quoteright/r/s/semicolon/seven/six/slash/t/three/tilde/two/u/v/w/x/y/z/zero) /FontFile 1395 0 R >> endobj 1397 0 obj << /Length1 1604 /Length2 8341 /Length3 0 /Length 9393 /Filter /FlateDecode >> stream xÚ´Tî6L#Ý 2¤{tH‡„t‡ ŒcŒîîNA”îét€t‰€„  "! Ê‹þúÿ}ç¼ïÙ9Ûçºïë~X™tôyåm]­Á\¡^> $@QSO@ ñ‚¸¬¬„3øO7.«îq…Jþ+A!î|J Ä]ž¦+ îá ˆJ ˆIA Pâ¯DW¸$@ ä ±hòÔ]¡`w\VEW˜b;毿N€€„„Ïïr€¼ ±Aš „ØåîD3@ßÕFøü‚CÊ€Iòó{yyñ\Üù\áö2œ</ vÃ=Á¶€_„Z ðÌøpY÷?üú®v/ ¸s8ClÀP÷» ¨-¸; ¯¦І¡$kü‘Àø³7>¿áþ¬þþ.ÙØ¸ºÀ@PÔ`q´ið!¼<ÔöW"ÈÙÝõ®ä ‚8ƒ¬ï~ßx$¯ Ýü“ž» C¸ó¹CœQäÿs×ee¨­¢«‹ ŠpÇýu?%ls×vþ?&ëuõ‚úýiØA ¶v¿HØzÀø ¡7°šÒŸ)w.Ü|ö`@ŠIÀn°·ÿ/xøwPà—ûŽA€Ì°»#€Øï~pýÜAž`îðûwà¿®€ÀbƒXƒí!PÜÐïÜ`»?ì»áÃ!ÞsàöÀ_Ÿ¿ÿYÜÉËÖêìóOúïùò›h™ê©qÿÁøï˜‚‚«7ÀWHÀ+("ˆ‰ÿEù›ÿ_Ü{u@?ïüP jç øƒÂ]ïþ¢áù§*8þÜNÀOÐr½“2ÀñòŸE€6w_ÿÏúÿ]òÿ'û_(ÿ7åÿï…y8;ÿsüŽÿ ˆ³ÏŸ wJö@Üm…¦ëÝn@ÿ7ÕüÇ&k‚m!.ÿUC€î¶Cjïüw!î Þ`[ÂÆá ý5…;xg¬ãêùõÖx€Àÿ‰Ýí›ÓÝ{â~7«ß!ðÝ:ý÷He¨«í¯½€àp.ðN^‚""?»µ{ÿV6€Ÿꊸ+ÜÑ Ø¹ÂqMTLÀ¯þËõÛ¿³ þ¶$Äü ¿-á»ØÝš»üã¸S)?øoSDÀoñÿ+~Wâð)ðCÿÁ“ðß½•ÿ ßeÃÿA»ÃF8ÀÁÿ¿+Gx¹þ«àî~ÿ˜‚w>ÿÀ‹ø}Áð?ÒÿÓ28üî)ù-ê»~þeÿ~·À`o° îÊ¢«ÍÃpÇÆðîËzy:/ÞÝ)é9Ö]ãlN^¿xÇ5!vgÝóÐ ø…|ÆøñÛmeŽs¹UÆ~Ÿ:š±£:Ót»¾ûßpúètžSJH†š ¶· M}e‘Jã¾WÜy/+]Ķ”»@¡{È©¥–’gIk|>ñs>•¡f}h··ÙñçÒTJß¹Üï|yúqišgŒœ‘kRãmñŸ=èÌ93K ¨1k?Ýu_/í{ôÍÞ£~pûˆÒzÓqyy›–Zôå;A‹N§7ûõv› ¨ŸÌnÑ(†êÛô 1vã‡ý ƒfÕô“.-ìêS“û ²F9å³TÞ³¹&àÏ6t0È‹‰Çøýψ²ÄÞÈêf³Qæ £ ™ Ísfý,å¦qÑb'“ÝZVÃ*‰ï²+=G›yÀh,9™hϹâþ8m»n¡pI>K ;šjÿäÓz,ÿ¢qÆr§­Ò +é\‚2G„ï’ÁÿJÓ»_²¿dïíô±i€Þ‡iûÑ]‰Àº7ˆgÃk _¦J ö(¸ìwY.;ÆKÿä…‹Žbå3•þ‘^9Ì.|¨Sn ï>øh5v¸DûBfðݶz~|“|gþTvU”ùVW5yƒaÌ ôLŸãj‘s Xú—v½Ãoí_ÉØFÝådɡ왼›¹âd3µ[yo.d ËÓ[{F=tjÕ`}úª‡zýF(G†›Á‡²LÎ/4WªgíÉ´÷×øÈ# †ÄÒ¨¯[½¦Ó‚8e;ÐËïmfY“‹Y¾wÊͯ8WXmC€«Õ—ÎuJ°’ Bß÷¢tsÅ?7u~ð~´aé­ÍdíFñ>áx¹ó|ÐVšè¬CŽ(¦9ŽàÊFË›‚äûŒf›ž†g9ï»uüY9FÛ ]¾t.Uå1’J³uƒp?TSž©'ºg‡?e(A6¦>ÃôI Z¡õ÷Ó -NjΟÈRÈJsTûF‹B<–-˜Eû£ Dsó’ÿ6[NLOî…â£0G*}7Ÿ4êgÀ$£Uþ¬;¬zv›Û;ÃsÒ$ ²Q6I¢²(D¢Ä ™jÙšÔR4WÌÅÖ!|[†¿÷N¸Ù£ß/L²t Êu’;ÑŠLÜeTt¶´ÄB¬¶òºIÈuÎl€oßáÈ5tý|]Øæë¥ O³!—ØÐ¡àØÉ—rï®{˜Q)6ªÊŸVÒ&œ¥·ªJÝ%òªf.kq×°…b‰ç½ÿ$×Û°øpt²êXã.,†:LKÿ9[ &Y2ýî0KÿÂ7V4ƒži§ú­{pˆzS9Mhl›Ç”±«,-‘®8Ù—Š=CðχÝx¶eXío'\q&i;à§nž{ô‰¾ÏV\Œc¥658g×;‹¥uœE)p]`Ь„hüLâFù’&©ý³…iõM.ìÃ\Á\â. Ÿp{bs•:.a¸~rŒ¥Ê´P¯J#Ÿd¤TÓM§9•z÷+ë»§¿Ô‘(ö¼ÊÊŽ"ÁÍ$2^fk< ¦FUÃZÏqßäÌ3cbO,@ YÕ¹J¾îñjm0*£hɽÌL‹´~bîœëÍ申§¢§ì™4ŸÞ½µrâ2Ào娭°È$Ñ[æ5ïçP®9”Wgr—ÄsÉ|[‚ÄÇ4ûY9¦ìÞ¬5ïÒló¦ò„oíJ?j zîësÝÞ(¶ºÌQïÚÝOÈ_h6aot¤“by½›1ïJÆ0Q·Wl®l7ß…½£³Y!ü¾‚.iJVHR° ±r²âÃF÷è“—Ô…ã1áQ}×#>Í9B?ꊭ‡9ò@èµuj­¬ç™©Ö·zÖÈvÙ°([!Ëaî÷d)<ÒÌ0¢oÓõèPÚ°3Çz>ƒ <”ó(ê}¶*’(ñÝù߆á<×h‰>äüKªš«`=âR2ï(3&Ëa¢W^Ê %»†ÙÏãm—Ûç ÂÂtüÖµZ·‡ì´Ï.Я íöœ|=h©Òv˜V—}–ÓÔ-cù’T³`.Ù‡‹›]ÍÑ>W+i«Å|œZ’®1G·Òó.dЏ±eêÑÃCMÏ ŠÎiá:)Ä éÝ‘Ø+¤Ò¯V²%sZ¤Mµf©¿6:¬´Dd 8ˆ€SÞ©jEÇTkÔÊx´Û°EÙ7Ëh¶¦Iœ‹›#sqÃo#µ´?R¯2{ƒPp&´ òïíòvq´1©·´™½ò1ÙÀé)g²¨ó]z+·Ä×\qE/rƒ2Í›”¦¤Ñ,ò¾¢º{Pãfú²Žÿ>üÃËPâ}ûÉ:Ö´8o@µíÇÞÌ>ïÂyù6Ÿ†³':§‹ŠoLm×9¹$ó/]l­¢‹‹å5¯u¬Ãã#y¡¥)QuŒT æ–²˜g{¶ñ9pyV Ñ~¦µˆ-Ù —¶2ŽH2Aƒ¡Ã5ͺ³g¯DÝSNÔ«y$ L17gä17’£]3g¡{Uç»Q\ª6¬hba£hÕ¹Pø˜›8Ô\+.Rm*Ï1Ì~- žƒÇ‰O[G–>'=•U†%&nÜHŠcŸÔ]}dç|ýTx…áUÀcÓ¢:îÛEK›wFq¢#¤@e³ÏØr2ÎjÈ:8 ÍØT3W %yAWùɵ7žå½ í‘cìWþަ¤s+ÓYf!W;_;´l‡®ôðL+e(.‹Š˜7k½dûaƒ[Í·“iƒ?¶-y” k}LOÝsC¶ðF¿²[üÔmž´Y¹¯ÜË­)=±ŠkD¾çP?Œ Þ¡amïã“xz¶…Õ:Æú'´#5ÉowÞ h(ÔWM“2«ï4¢¶|é;ý9xÒΣ¢RLÈQ¦L"ÔIIR‹™i(=•dœÙí© ”ÓËÎ;"ÙÇÅ%ù9âÕòUŠ V±±P&ZDcG1h¢² ¤Â¸‹+à€ËçÞùËŽ!¯¥£›ƒêº$zw}Õ^1a¼i…“€U´zM´`ò1™ÂX†îÖíSPï'ðjH~ˆÿÆOQ½Z‹„‘–3ïIn¿ð¾Mÿ:¶c•‚eÔ—q¦¬ŒV"ÛjCLM¸xIޝ¸–5úLL¬‘×jt…£®FÓsû#ÃGJÿ¦6÷¡ø)üZQ¦!{f¨ít”yšÉNx”#Üv%1ËuÞ•ñÛ˧‡é®‚„ÛRë]•8hÑ ô£ýDh÷£¼5“Õz-q{ÎjÌYl• >š-_áÏ:µm†¦Aüµù>(KxYë£uº+`û&o-æŒ53µäÚA+]¼!-'Þ\›´Æ¶g3v|]K1¦ ’°5)Fž´ v G@SB3ôk+òRjÉWlIKPÙ‹H[zO"90½Cš‡nWó•ƒî×÷¯"Ä©ˆ´ ÎÕ•&ÂÓS³cz«rÊ{™B’©ÇÑÞa?°@m}DÞ;Ü—²¢l@&8梑Ây1QG? 3Ý·{Ÿ:ºÙ«ÿœÍ˜Âψn¥kíé-A%áæÐËþ¬gÆ6¸!»iË">)âMtüø‹XFÛu/ÏV^v ~ Í#æ²|@ËÈ,‹¯C·—ØÌ‹us˜Ië2èø%ýyê½GžôF#ƘÇ_–ùˆ±Ÿ°›_Œ$ž™…Æ%0ÄD/úï‘J°a?ï&|g#Új Öí„¶ñ4DbtIwøÍcÑzè©ÍÓBƒµõ<À/©ndvU†¦ñÊ|Q®ëu^×ñ8I§Ô[­õkv#8ìºttqÆŒÍ*`zû…Fžþ‰6‚s@lå1$ÎÔ(ÂBÛdƒQ~ehøìF{âa`ÀZ9Q;É¥­Lxìí½ Bdæ£}ç+lLTgÃWϲ±Ü$MtY}¥‡XÊ QP=¤Ú(ÔEõT².&šß"£<]Þí4Q±ž_ v4Í’[TKë(„ß¼}*øvtÞéÇ99Û§4¼ƒ‡à±¥TÌ2µÌËø ®]?gÊNìwFšãÜ|0“þ{ÁÄ´Go&,&C)‚…$ÏU“£¾q:eä W4Ùåã^®éÉ(ÒwnhZ\[DÚj(x´sxe ³TÔ¹ˆŽç_é'ýÉ÷½G\{Ë ZñÁå´ä²ã?‘g³«ÚY¿ |W_:Y„‰Í`F²Ò“yÇÎL —‰Å‰¿n; nˆ+q \üÙ“§ä{PõÐñ¤^ãºgÊJ_ïI«)a߬—Ö&Õņ§]²ªÀwªÌâ„ÓZbMO¢ ~5ªûxö÷Ó€šv´~詧 1à¤%„é{©BaJ À#ìëOg>ç; »Ò°àòçòB²Ê‰|zÄ „¨˜»!Y˜6Ó†þÚ_˹; Xšá3Ò+Q1ɵtÐäÍD¬%<›[üó_,áEdi%&–¿Éè>×ZR1ƒ`‡Uo|{žbŸ “ö‹œ{J4ö 8¯:)ÃñºÙøg®Ì×?×Ù¼m¤¦¹]gǵQ.:@´ùP»j)•õ\Odғ׫–kW“h9)úê4S Ô¨ø>$Jv§8T]Á.¢&ïƒÞ 4lˆßŽ*s³°Gj!ýp`7®ø+ÞÅÄ¥=“är—‡¹Ô71{”ϱ ð4³ž‡ëÔ4„ZÃæ£‹¬|ãý~²ç>öÆ9¦àÎIÅ}M%õY ±”*¡•³ÒÜpVSe(\wÏ”>vÕ#ß#1¨.‚JÔ§*+¸–ê{+4ÌŸµ %¸sÙØŸË-º8Ùy–+НOŒ’…GHÀÌä’,IÂ#è/à !b°¶*ÁÖŒP]ª&±ÆRXA¥­ÜN=‹þ[ÙUs—˜,ó¡wIÃkOÖÕ·éknÛÛ{jï=´T‰ØQ›¾$R WRÄ3™ˆÎƒsLºNùƒÜ/}&órŽ8Ö$bŽz€};î+ÙÑCXß<0cöΨ>Ô»–¨øD?)¥RüªŽRȶÑç¢ÃJ‡±]ˆìÅÎy‚!ßÛÿùU ¦X]X¢ñ’´L]¡x~Arž«xUâr+ú0,±ÇSç='™JÝÍö®5+Q£k’6"$(©cw›ÚÙ®[Þ û}©ÌøbAÄV c@«\ôÓé¶Æí[ÌŠScr¶團ÌWSKÁõ9ñ‹ß¹‹¨XÆ“÷ OŸU(äùzb[Ñ[(ŒZØpï‡ÌÄw†7cr…M\µÔîjÒ;º#¿-=™û‰å:¸všxêÒi„[ÁSf`Þ}ŠU²² ?f†'®¾Zi›’æ.ä;@œkŽdKØ8¶ÄRˆGoèÇÀßZ>ªCw (ˆ¤u[-|VŒ ²ÒFú»mÇØfc½Ûš: EŸìdÙúÈê?«©wÝÆkO·Y‡_ñPtOóÚ¸`Ð>ðkð7S€G¥yƒóW×î•[¬ù{ÚgÎÊD‡^imíý‹áÜÁñA°3ÛÕ&gò$%Fòâ|¼ŽûÖ‚šhà‚Ý.ðÒÒåDoÍÄ[¼C~?KªOšñD¢«Ð'Kyn‹($wýÙ@ãô„õ±NÚKx]ëÊ™…ÿÐŒð™ï¦±f¬þéúä4ôS(ê:mجSîUîÏ¡„{àùžý„¡…KFù’ÅGyËNÍüÖOη÷=ÕÖ¿í£µá¡é’¾#=ÍCUj˜ÿxúÝéÒ=³ØA¦5Ç*"¥”P¡SÏèȪSÙ>6.C„Û¢b육2#‹ö,/j§åËÕ‹ Œõ” ¶þXs¶žûò¨‘,Äñ¡ ó¬×‰6Otð¨p0hœœ­GuOæ„éÇcUõ+65ºI↢KHxƒm°,ôæü|У:e}´;õùö¸7 ˜­n¬,BpFœRž; ÊçÊì²Ñ‡Y6®ïÓèùÎHÛ‘Ädr^Ú;ʯ§±{GY+èØJÕm—YÎ&òÌ1 y.ìº,ã©3¥4pIÕ6ðë7¢  Ik"%LÌ×ô„¶®å¹µÂl?=xï‰òHGù<Õü˜¨hjÎLK+T‰Ža{à£5E|‰?#é—ÇdDfãøD8o]SÌÉpcG†q wù,È•S¼ÃÌK|£°ðÙÜ”±½;ý§š×ê†$>#&=åd2…©ì%¯Źž‘&\Ç–…Ä1á[•W-˜¶ ³&ሃ•ÏŠïâ+¢/ ׊úd,(:ÿ¢9 *°ë ‚úάõò6ƒ™¡W.;äï´ëÑ~®DR{ xi{pi^:¡xÛÄæ=»5*™ï—¨Vp¯/&:)žúúu)h!÷¬8Œ/D?‹{ŒJލ'6åÉÐŒE_žµñFS±Ö»2Ø hð30Õ.©Û9iÚ5‡«¬”€ß ÃÞW„KžÔŸ2ìÎêcH—젤쓿Gæ·Ð¡SŠûÝOqQ4\n{ئ‹øI·g^k¶QIJâ8’ÓìoŠ®+Ì8¶§(Eaž^þL庳{­|Ÿ…;7µ<‡\«ùQo÷"nùyã«x¦¥ŽMüøm¿áŸâßo=ðyr<Û|F8œ< ºX”òÜ~6‹ëèVAit¿6zYRXFZÌ\·ˆIÙÜYNˆ#–Î+ô*Ì èi_#xôI;ÖIyAd›:èÜÈU0VáÁ'û’F^¹G2¹é sïиYi´ž>«Ëx3¹ª ˆÄslýÆnãmÄãD3éM̺¼° :èߣÁÀÔÙ÷ÝiBFçùÊ•>ñS¢oX¤.©^ãà óS“ÿPx .×1î5¦7×ÿBQZ°`3s(l´Éëz™wãežÀ—fMéƒ^3òïÉfÁSGÍV©ŠáckNšŠ¢{†ÙÐÛD}_˜0« äŠyÝ‹I;—YɱÊo®6ö9Ÿ)š=ÏÝÉKÜ2ö£1²±Xàò£ñy}ÖøÓï‹ûQ¡ ÷æ›9Ù=Xµêg”¼©Xñ4j]Ö±cÎê4 ²Ü¼º¯ñdvì}kÕn\y6xIäc •D˰Ïj‚pãÑ3"yCµƒÁ¯Ã,FËaÙeÕeð`QƒB Àz|Ÿh¿…OA³TE~ÇÑé­ íh«¢=I=áÌpOô¾Œˆù ª6ø{W÷)9ò“P¼É³ w&]"³ÎQÈÖpNU2ÌZûÏÙÁg€›ˆ¯Kå‚Õžu½ÔQôí=žI¾ßDöÖ¤’¯Î¦,ƒa¶ü£güÂS¾Óg—Z|) 伡dv—¢ô]DIX(^ïgl,¼=•¢ä*¤ú¿LGmã¶lq%‚x+0²+±RBçvùÙX±hFÇŸqŸÞB%«Íq ²md‘%Í» ªJósŸ/¿OëØ°¾9öø”µV¬1žâ¾:‡©–‰ýF)ÃÊúöç}­ëÍ*¨•;6yÇût2÷Ù]µ×Ÿ¢UÕ´ 2îUo¹eŒLORH6Ì^ˆf¢æéÃ>7):®\¢cîõŒ„¡Ð½à$˜1…ÀF\§Óõw†@>Iv:G¾›‡°æ ÆI¦Çé7sŒ:ôÑG·êGGÈ¢KËvÍìPõ÷n¾ˆš³œê1RʵÆv}ûΠ!L‰,å¹mÈ]ºD;ž‡øÇ­wÍé¼B!̺å,~Ø·MšÉÜÁªM )±nZ0°gJº¦ð»$fzè~Q¤s“ŸØc³,QõðF:ë/™,9$k–Ïæ®- V|–GÕ°/4˺œxQš²«gÚBˆuIÕ>£× aq¤?òäõ‘ÈlCÊ”Ç •ˆÜ2ÀÒ'Cþ;°þ áT áÅ9N«:M$ÔûÃ~{_°BlIGÂÕ#$vMüõ]BÅÃp º¿|jàzíiz¨vâ°ç]Z¼s;QÅMnãx&0¹6h^;K¢HúÕ8I•…v$ã×Timú\·¦­ J3UýWé ¡µÕ­WsxŸ6W]>1ÇFý굓ÒJýùØÞbuòDè‹ÏæØ@cSAó˜õ’Ñ´Œ¡¦%¬ â_cÏ{ì ¼Sgm´‚j™·ŽŸ%÷>–½}cöþé ýúéžPTé!‘d´/Ë{?ºÖ¶X˜×Ï'<¹ÃZ±Ân ùWß…ÑÝ?è}(W"¯28IÅ‚8èXðœäÏZhXóS‰gÊ"?Pqµ"ô r… J÷{d÷c5L€8cÌãÜGQX´±Ò mI\÷ ¤¢Ü#Pi€Òj~y—µÛwsšbÞzÕ~÷] ˜IÖä®b»·{&î­êZ1¾\¨•Sš>--‘¼Jþ Ëä 89z˜áó£c+ÜŽ‚G0.]ËH&Ëpt¸ 0˜âLÛNI¤ñbS9jëÙ<ƒ uw›ÌÂ/™âRIž *)4~tîVr‚ØF¼Þ,ý”¸C JŸˆ Vذ,Ä^Ðfµ }ïuŒjJì5íHÜ}âî ¬ [sw$ÞÙ3¤ZghWñT"žM$Ö\òô n=q3óÖÚÝ»g~›‡ö ¢ endstream endobj 1398 0 obj << /Type /FontDescriptor /FontName /XNYRHV+CMR12 /Flags 4 /FontBBox [-34 -251 988 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 65 /XHeight 431 /CharSet (/J/T/a/comma/e/five/h/n/one/r/three/two/u/y/zero) /FontFile 1397 0 R >> endobj 1399 0 obj << /Length1 1620 /Length2 8804 /Length3 0 /Length 9844 /Filter /FlateDecode >> stream xÚ¶TÞ-Œt·„ÔÒÝCww0Ä ÒJJwHwwˆ¤tH‡"¤„<ô×ÿï[ë½5k1³OÞ}î>wAK©¦É"aµ¡7VvA€”²€‹••–V ìæúÓŒJ«rqC!‚ÿ r™»=Ù¤Íݞ┡€‚»€ƒ ÀÁ+ÈÁ'ÈÎàdgø+ê"6÷[”Y PÈ•V êäí¶±u{jó×O½%€C@€ùw:@Âä¶4‡”ÍÝlAŽO-ÍšPK0ÈÍû?%è…mÝÜœÙØ<==YÍ]Y¡.6¢ ÌO°›-@ä rñY~¨˜;‚þ`ÆŠJ в»þaׄZ»yš»€O°%âú”á±¹žš4啪N ÈÁJ0þœ €ƒ•ãïrfÿ*†üN6·´„::™C¼Á€5ØP*±ºy¹1Ì!V¿Í\¡Oùææ`s‹§€ß'7%ÔæOÿ¤çjévrseu;ü¢Èö«ÌÓ”e VRPGGÄÍõ×ù¤Á. ˧±{³ýq³ö¨'ÄçO` †XYÿ"aåîĦ ;»ƒä¥ÿ y2¡þc³¹xØÙÙùx gÈËÒ–íWy-o'Ðo'Ç/ó?'¨Àú‰Èl zúBõq5÷Ü\ÜA~>ÿvü¡rp¬À–n ‚úOõ'3Èúütù.`/€!û“ö8ì¿>ÿ2~’—âàýOøïûeSTJU˜þ`ü·ORêðaáâ°pò°¸y|Ü¿ÿVù›ÿ_Ü[ÕÌÁžýŸ‚òk(@à O³û‹†ÇŸª ÿscÿí }’2@ÿòØyØ-Ÿþpü?ëÿwÊÿŸìUù¿)ÿtwpøí¦ÿíÿÿ¸ÍÁÞ<)ÙÝíi+”¡O»ùßP]Л¬ ²»;þ¯WÞÍüi;$ 6ì {¬ÔÀn–¶Hè¯[x*Ԡ®à_o €…ƒý|Oûfiÿôž¸>ÝÕoèiþÛRb µúµwœ<¼ssoTö'yqòð|8žÔ äõ[Ù6VÔí)ðDÏ` uAýu£|ì6à/ÓoÄÿ„ÔþA\6Í¿‘€Íü$`³üq<é• ô/ø”ió/È`ÿ >U²ÿä°9ü9žÎù|ª ýä°¹ü >Uvýä°¹ý >5rÿ|jäñþg„–î..OOËo‘?Í÷/üû¼@–¨KóPK¡P»úÐΛZ O–Ý ‘Ú]Ý4Ÿ%—.÷[L¤· 5YÁë.Wo?öc¯lËÐ_Š/S<ø´5"E´'©wÜùÞ›&hLív .~"š,:h$C!eÑÿâûàì«d×ö¬G6ÏÙS­ïÆs@Ö«a°üóXøü®ú—^E´ûòi–×Ú±FA%³´ùÙsDTˆn,dÈŒ¸§^X³—W3¸¹“ L¨~‡¯¹Š} 68ßü˜{µZ©ÅéÚKLCl@Dw‰;6Eç#¹—¢@¸àSú.»•u”³àA!Ûø0 uQ’Þd7ð½Ý„b’ôáÕF4(¼kôËu "òé+`§T#rÌÝ0}Y¯>‘vGuRhòGVKXlƒ«ß›9÷#…©’ °E Æ{÷Z[ô›ç.©ø¨ô‚:9åñ~Ìî^š‹E}ݧ62ƒQ.ïù›$õqô1÷×ÕfÄÞÊ.äa#|VKçëÆZÍìû’*jcŠü½³ƒ÷µ¦c°ðMì—ážGƇ©êyrÖû°{è§ÕHízceàûfæûõýíºÝ±!gÌZ sÍÖXG´¬Ixoqè­nIå5`Ä…Êçï 2‡’öjÆ&]62åÚ˜O,kS% Ÿ:Êf3Îl6`½¨ì7•€AÂôª‘“±ÊØ~Ÿn<þš9¥æúUÎѦ®¦3kèxÙKF3.=ù¢×¯€Æ&^ܺM²ˆ™¹eþŽÐƬߟŠ­ ‚¡í4é› ‘ßò«ÎÓ®fÃ/€í¼ùbó•¼Kíê:Óêô=ZäU™1EÜêœÃy‚ÑÚCqR‚O´ìðõHô,%Ñø¸îqýÅ[‘b,±ÓÇ^¬¾éjAáFÍ;[ôŸ.Ëa”\fp¨ôE˜_ÊXYŸDÄD¸eÏI¯!ãaO²øx›¼UoÇOÞv$Å¢ôŒ—Å” Ø@ìÚËFõ}!Øk²òüøÊTÞ3odÑ:PüîËF`@Õ=­pjœn °Ë8\¾vòbä/ÎTÌ„«'åd²®*£€òÈ4l]¾‡$^ìeõà¦(ƒuI ÄØÏj*.Jr¹áøg$—+yOÎN(9ÖÍTB%TÓïùNX@Y‘Iѳ\þTsÚÉJSÅZÓ%. ds„Bi¡,ãm“ב„¢öo´|E.Šò¡¦qVÝ_½:¡DZì"ÆÜÊÌ è ;Œ7#%Ãá|'¤v/Ý•y&îë&†ðbšô“›1¾}Ÿ¶xNŸ5—“‘²ô12æ¤ …á£}0Vy‹Á„㔇8 PU í\ë¥q]Ëñ4{Êèòr“€Æ<ú–éB_Lœ¬D×Õ1&VBañÇŽ·¥ž%ºúZýkC!ÛÇn”ß}g'l‡™™§°ôl½y*,¹h‹ql3ŽHÖøì˜U½f"Z.WÒ´Ü#[—»ß««¿õ{y®'¯] çäOXŠ˜8ÿ¶¯ìahºÚqå´2I¢¼ä‘|)_š;Ôˆ°êt&hD`™­\&ÜÂr÷ýY0š4ÈL[*_Kv\Ó´B¿Ö ã™_–«Ü3ò?‚Âÿ@ºÞ§zÿ!p—±`(D7‹]qT:\+½­a9TöÓèùŠÖ\ÈIêŽý "ßÂûY–üw?{á~6©ZgJ2ÔÁiOIYËð³6`Èë²û¬m}}w–¼‡Ò„'Yw=&‰§AìDmÛ±Šˆƒ¹ß³¥hºqÙ£|qµä«-½–^1Ã8~Ä8¸tÚ7­æåØSëB¾ÿ|3†’MoD‹ø9_—F¨$›;>ºçgе¿îˆNL =ÄäÚV¼ÿ½±¯”E¡%Tj¸WÈÑM>·-!ß­XŒ§ÎqV{¿lÑ¡s 7: ÖØ ûK y„À° N*§¶Ò‰ŒMØÀðTº…§vç%¨ÐdíåEaŠ9Ëâ¦zd]lÂ7]^Í»XzR¹­ÊÇzJÛT½¢ei«ùpl@ðØp3’²'©þT°sQrÅK(áZO˜E»ÆTǤY^jÕ„Ÿ1Q<Nbh´WQ¯æË%NïÜüÓº˜È2ç®[§BàwõÍù££Æ3Ú‰…:©Ù÷«ßåý”zã¢4 »«nm3Ö÷’J¹é•ÈŽ“÷«bRñrYK*©ùØsRít&#Û§ëAòwÌ^#}*|ï?ë¤y}ÃZmMŽ›vYUÁª³ƒÚÃKD6·7¥SLÄ>N ÿÐ`÷3\åËœÉ' ¨Ù4jkô• ¼=÷äšÒÊÛÔÞä”ŃŸ –ÿJ¡ê|©òv©jÅ, íFs*ÃáEV:_1¯)§²ˆ.!bï§…—,ãKåÇ¢0YuP7~”S0¥ÓNK+|»p]N÷ õÔEöþ‹H ‘÷òŸxÊ Û´9wJ¯=bÙ~$ê¥F‡K—÷POJÇe)0iúÊÚËЙ$bd£–UT™ÜræÍpØ„MˆÒ̈)fØmÓCû)és UÔwÀ²?z´ÃÄçªWJxi‚w®%–]'2ˆ»©"ƒÞ¨¾0ªÄLÕc;ŒR ‚ SêQw¿4xÛÍ2õ4Üœ§7ü2|É ýI>'$­þu³‚²q_.mû¬f^qMMƸÚ~IPŒCú§à€#²MW~‡fÀn„_¢3ìt+€(^‰»ó(0>SFLäñY;BM7øh…0õÇO4oÖÝ>XìYë:…s/[78JPá‚‚ –h7Ésuœ"q}ØP­æhÄ3¨Þ2ªÆ±S-—&µ÷÷©d«“ØoTì‡ëå":´_é{‹eo”Ž.u+—ßã{°l/›ÔÈƒŠ§YHà‚ZRÅÄuÐ_;TÈReK Ñôä—"–»!Ò7©Oã¢&Nó4‡¯iUœ÷ZOþh½È°“ŽDž)¸6d“J¸¶ÃU£¡Ì„â5ÕôÂ;û0ÿ\Íq(Dó®#þäÙÎOIÓÑ[`Í9³û|ÙÀÂ[ˆ’Æ[þƒî –Fëk¬¡Yì˜ëÓ«¹ùî-η¢c!§Ë¸¼ Ù™;—¥°8>cÁìa—ÙÆt Ä)É:¡þ•Ó½úV+Ä*rÉ-(i¢ô“UÚø´ù¸L¤É}»ý†ƒ=£ü1Òñ,»mÙæ\%´³èÉá2àÊƒŠØþ‡/-ÅÈLáš ç<0~Xzj ¿²׋lÛ0ÊP‡X¥QÏSçàœò¢bà2ö$TlnvavÖ:½I¾õY°mÅ1‚e1-ì“û~5¥“©XNŸ¥‡^ 3ʸ¿]x-¢øƒð»Èy]“<'¯Óçx•;wë€åëëX±VÿãvÒÖêB|­üJàúÊ<‹¦%…C™—Z&^"špÐ쓾#щ¨é¨Ûð%5И‡*æÁEÆci7’ú–žtÙö,?k@«×A [A¯¹] ½‹L!Ê?õ–[2òÌy¸Œ)Í۾Кxx™1øÈ?§Æ™Rýù‰›3°ZG¯ûڥª÷ø~Mâ0€9ýp:˜µìølþ¾|F":Jä`ƒ);x.¯þ¬šß[ ¡oüö ‹ÆiͶ}Ƚ0‰ú|¥=Ë—‘R€í <»a¼$OÓæ0ä,]»Œ6”#…„ÊÜìä5á/|-MôCš¢ÅI±ˆ0µšÕ¨ÝU¯…–"©"|2䃽µµy¶•vZ¯‰ádÜ‘¼9Žk&0Ü,qA¡$NRAÍ:zˆl¾ß^eIYò¬>ÿ”½|6Õ‹? ;Üéü>Q™êèNSÏ™è»c²nmZ–8õÔbÂOƒ°WjøåÈðá¹»CÆÍ~ÑV°¦³˜+&Ü>Õ£OýÁ1‹nu« âµ£áùkÔ¹ñÈiÚûpa ãkJ9ñ#W…Éžq©«ŠnxîâÏ?“ÜóÈ̺_ÊO‘/Î/?§P ÿ‰z~4‘XÞA0Ú_æÁH¼Psû<ͦJð©Qæ1ã §w©67ò*€ß ÓyCð÷¿•o)ák ,‚¾X°Š Qe\¾mAÅ{µ•rtû@éÅh´³:Š-Ö¸G©Þ4Ø•+ƒpÑ·&Š(š9Šé¸ÂH ¤Êåë²w[* dýÊ¢º1êŽØ%3º·Ô!Á²ñLÜQ; Ç“YR[º_”F0coÕgé5±2NàhEÑN{?¤¸žÂV—„1dß¹?» þñåÆÛ>àäæÃwº\ÅòÙ³ MÌXî7šÈ“­—ë"&uj\ãÎ5A@’Ñå8\‘›öd¾]FƒÖ‚‡lë‹iÆL—•Só>Ic;(Êo±˜G ‰u“gz©¨aTG‘¯èI²µq¡ÏûCd£§L6|;öM-‘Îb­0ÄÆ• ìR¨sÓZqïF²2&ƒò=ç¾\å¸Þbsòž¸×\ÑkË&ý"ø•¯V'[W2¢I®Ž«Üýý…Îù1ÝB¿ÄûAßg»ØsH‡cÄÈœtÖ(ôd!K»KZ¤4í÷š'EŒñùirAK¤ãœÙÑ;>щLÑ*’k$47ËÇŒ„uÞì‘_¸9íiœ«Ü:ĘÝê¦}im•„RϹ7POQH×ý¤I^m.t3r‡Ø­Âú®kX㤭d¿›VR´)¦ÍJîQ? P""À#-5Mz•{UtJÁrDqýIaJ/( 7S(¥™Ú“×þ ø8ÜÂàtÉÞQ™ÿÑYȤ}4Î×XHçó#ëWOeVÞ–ò†•æû"Ÿ"Žt^B\ÉbÁžwA„^ ðù`žÁ;s±ÒïKÄd¸ ü#k‰ŠSèܪ¾ÚÅÇv~»×ÑaUeÄê«÷©ßçtqH«RöqÍœ ¶Gg WÈ<Øf‹Û;§ý(íoƒõ˜ë‚P†«ƒ€Â®¤ˆ•º)¼}2§¡Î†QL%®° ©HUÙo„»Z•ºaPõëèSY«#E¦°Ýgª—õFÂc÷ð›,|]ô ÛÇ<(Z0çÎÆzúü‚vCȘT6ËÌÄù]/P–÷±~®m6y{$zzÎâßN¨Ñnºðh(ŸÅ9%6.ÂÔ6>è{·W;°[~#0õt£0Oî\á66šp¦N¶¸®fòÅ-IhNv<÷IœŽ0é«U½%oç|h7lF”ßDÞ±Œ¦²Åº¾áÛÚ¶VdI´y•ökâ|VJn½†,‡FÁqôoK™`Ã3Üu!ØZt·² ú˜ð‘°ç"¾ëh)Ú‡bR^†þkæÍj¥‡ÇXƒóá ¼â`°æ³hý²p#Aä(C Ï’/ÌX˜_˜HI«Ý ~ëªnÖŠûâ¾5²©V¼ÄŒsò3?³ !ÜSéëºSÒ+ë6 «M¸éÚ>û¶fª6@ic¬SµæDFxÁ‹RéHý¥—êbê¾>‰­¯q fh˜¥™šñQ8ÌØñ—3£fºôL úóV_‘ÙÏRÊ4ø~^ò¶It "Dš‘fô¤ò-&Ž!<£«Æ‡—’Ž-£À¨¤µ¥qÕ­¢yì¡x#7hì‰Bû‹'"#¹‰ÆV“ñÔ)uÍGGÏm›_Ãí"é¶2ãaá{´È­¹¶!hb…³&úÀÁ»˜·L%W SWÓiŠWí^a8²àØù#‰g?—s¾•LÉy÷©|EÉòæ¥úâ£èo˜¡FÉb¦Ö¯RS¢ i^D´”ö=]u.íX¤¦‘ž›Ýìøù †ÔETéqA7ÜLXnbõkqS[\…o( žäˆhÅN¹¨¹øß«Ù _ln°L îŽPQN\5_¯ôƒÏ 6«¨„ä»BŒ«ýÜ{¨ð/¹—У‚oêû¸±ŒÌ$Óé<È BØG;º÷L™·ŒšôÝÞÀ·áò=Ô› "¨o(§Ï´{ŠÔ—Ü£[Ú¦.3¶kF—eaks“¢€­Ü âÔ>óaKÚˆÕø:Y8ImF‘’[ñæñ?Ÿ¹fØ®1èJ†.Dr}²´¢“<ÿhÒáðFÇPæxNAà!ôÒ¥åU¡Ï>äËqȃ¬=ò™ŒhI¿Ì²rµf¥uË`ï=Ð:DXIMí…Š`ƒ½^këæ¦†ÈOª£ŠT<œ°oÞUw—?˜Ê«Þ‹l*ú.^^”nÇgí£l‘'W«£¾˜“°ªð"²\†Û)‹›2Q£fÙ°k—ä]˺Ýug_à=$€ëàŒš¬BÉ›û.ò¨ÿ™!Ä`ˆð¬Kš‘vºë ØEìT|åë€<µ&¿üÚW¬”ê”CÅY…¼U'x0v8ÁâÆÑ¼N 3„mFD¼CÙ€a®Bƒ½±DŠ´Ø!#ÉÚiõ7“T݃>†‘žRûÇT÷äçW ÆïºQ|Ôß鄉ÞãÔ9Šðš„j’MœÅ­Es,]£²>§}ŸÇG÷áGî}‹: z‹Â-Hé‘$e®s¯ºÝ} Ìd^'+Óª˜‚ðY´{& K°rär§bôÉÀºO,õåÈ´h‰D¹‡i,6y!öÚÏSns~ö¼kh‘R)—€—K{NñÝš°o_7Œñ‚ãí4‚ȵíE£kÉIK5d›ë&[Yž|ô91eã±á.GK”ªð#uÀ•C‹&íЮ.`á¤&»ž÷¬¨¸Î UËÖwQxOã>ÜÄ#[Ô—~ ËÏÃÔ+msºSJj¨ø%k/þJy Ù¯` J(;áÔ ­ÌmLL¼ÖxwGO"ñŽ£í!šRŸhiµP²©¢›[+&Ÿå» `xK™©ó'%©Ã¯»çhf¾gXˆ5_üiGݪ&ŽHެ±?Z£÷@»’Ú‡ô¥½­¹ŠÈüÓµCt…é/]AõÜL=Kަ¾'–ëH=±ôêI>113„i£YŸ»n,^ áÓúFœû`/FSà#Y€n$8IÈóúý¦Ñ2Mý¯´¤Å-¬X[ñ ßžÕ·ôys½TÀ™ÏM± ¥‹Ô:<õÓšºÂºè]”ub¼ƒG'há‹ëHî&ã\RÐ Ž%1'_0mn¡FŸ?ÁPs4Ó[>HeŸ2‘þ´߯fc«B¨z,_øfª38â§s ?úC‘i¢ŠØàà~©”å–«œ¸'ŠJzÙ3±ø›%"ß}˜’¸‚ù e$¤÷£|PdíÝL ™¢MxÓÎÒ—uixMÀPÜÝ·;Å×ñÙݽãϬ‹ä¸ëu>®‹£#ÔâfÞÔüt:í,P¤àÏêµ]ŸCêlèÍ¿m¸I0?&–'Ø ÝgŒ£ÕÈŠP‹sæÜ ó™^ÏÊŠñl¾Ã§¬¹~þÿ"ÍA†4¦ëÒ&¶jÙI—²å´¥á£[À¤ºŠ¿„n ì Ù dN¥T÷‘^Éhþ™ŽP(ÿkìÅïðçÏ`´’²¸EŽ7êÉÚ4ÿ‘L \ endstream endobj 1400 0 obj << /Type /FontDescriptor /FontName /KOFBFN+CMR17 /Flags 4 /FontBBox [-33 -250 945 749] /Ascent 694 /CapHeight 683 /Descent -195 /ItalicAngle 0 /StemV 53 /XHeight 430 /CharSet (/F/P/S/a/c/e/g/i/k/l/n/o/r/s/t/u/v) /FontFile 1399 0 R >> endobj 1401 0 obj << /Length1 1395 /Length2 6101 /Length3 0 /Length 7050 /Filter /FlateDecode >> stream xÚVTì»'¤CJRp4H  éF¤ 6`°Ñ ©„tƒˆHƒ4Òt Ò!%J+]wúÕýþ÷žsïÙ9Ûûôó{Ÿßóž±3ëèóÉÛ"¬ÁÊ8ŠO( x¤¥'…ù@!Bvv(Ê ü‡–Ý슄"à’ÿÍþÈ B¡uŠ ÚM ¨»9…‚¢’‚b’@ @”øËá* P¹CmZüuŒ$d„pör…ÚCPè*\6ÜA 1ÞßáyØj‚´@(†®hrè#l `”׿RpICP(gI~ ÉpµÈÍ ð€¢ =0ìê¶ü‚ x ‚ã'd@ È?Ôú;”È @+œ 6`8à·»еújšmg0ügÍ?x^ @_ðïtFÿJ…ÿÙØ `Î ¸n°ƒ:ÚÊšü(O/·ýårB"Ðñ wÔ dvøÝ8 ,¯ ¡ñý‰iã uF!ù‘P§_~¥A_²ÜöÃQHÂ_ý)B]Á6è[÷ø=VG8ÂîóÇÙ ·µûÁÖÍYÀuq«)þéVþ£³£"@ PL\v€=m ¿’x9ƒ©Ñýûù8#œvh`?¨ýC胹ƒ(W7°ŸÏ7ü["ØBmPk°=NøOv´l÷‡Œž¼+Ô` DOüõùûdŽæ–-îäõûïá 詪j«ñüü·IAá ðáø„D€h®Äпgùÿ_Øku@Ð?{þ“Q n‡Hü}wÁpÿ“\® 7àß#Ð<¸þ¡½Phƒþü“ÿwÈÿÆù_YþÚÿg?ÊnNN¿­\¿Ìÿà ‚A¼þ´£Yì†Bo„½ðÿt}þc‰µÀ¶P7ØZÕP ôfÈÃíþ¾D(Rê ¶Õ¢l 诠Ó;Aá`úë™ Çüz×lÑO =©ß&0z•þ]R nƒ°ýµsB"¢«+È‹=x´$ðD/§-Øó7¯üp @ÃóØ!\ Íó@ýXýRþ–ÑuPˆßò¿ Ù¸¹º¢—ï7Ð]ü%ÿÞt0ØlC8;…°‘ q¨ i>«gðà[‘ùľþ$•›ÏgÖµÅí‚/‰»<#hÉõD>©¿“laM‰ëXnîÞ•ÏvÃ{¼ÐÆݦKߟ–qzcëM„3©{Fs·å«» îòÈ}õ½rñ5 tÄnÀlSgÏrq'Õɦ<óèRñ¬î.žz1µ®ûµ\Tƒègñ8_”a¤Y`Áû[ëד´,¸(>Füû?¶ÓãSrÈm¸•òvpöª¨/öbÜ4£¨¨0}*¬ªØù%Gt…ùá´QÏ7z1qÒV¢Ä¶§Ãö:Û¶i’~bÞüÌy% «OâR¹ENŸ«1¢êw7x31Ò‡ó »?ónònç¼’§êKHú;.%vü'bß)ž Ú›xz¤å¿½ˆ÷“—l^NC:bÄsTÜXö >Xû^plm.Ÿ- ×msp 4’æ›…ÐOáÖ?dgßq[(ÿ¤Ç°)rLoHöÊ”¼Hëêê}AS|°Æ)þ›'tâ1®8±Ï¬‹òôøùŠ “¦û¥°*QˆÏPª(ù¾•gy¶=åë?Ž÷Ýs±ª¢nwš“y× ¯Krô4vPÑ8êþòÙØEÕÌZ>¹l@90׳"f1®6Ý#ÃÏ,iŒvªHZ.à¶£g‡-×J´o{» †7Øçë§ÁlÞ7;‘×?Ô¦ØöÍkïi¤ÆŽFc™2ŠÛ׿’ƒ“·¥„â-&ñæ•,u2®«dŽ”¾ýȱv+™W;ªyäθx›ÆØUS#sбIÔWLð@ï`¿¥xR:æU„ÂÃq´à«²pY‹{°)ºÊÂJ˜½‡:(ÑÚê1Á2AH[»˜¥JàÙ¤®–íe­È0äêøß§B¾~iƘcõ £¸éêø¾C#cDáþ‚Åî„}ñm¯7öåØ)x£bÌV«ƒ…Ç"nkó,ãÍðpb¶¨<È¿÷Š/ÿ:¦¸­¹S<ŠÆ3«»Þ~Òó¹….§ðf﫟TçôÒgŠÏjË—Ä6–Œ;^“ÌXø±Õ™K|3ãÛƒt ÕºÍÛØ¥œŠÜtkÛįwãËÀYO/4õØh4±ÔA”:ÞfM?%/x¦HèVu’àôñ?¡9¾nÃ"\þ–ïÈ|ë&î@h °ú>ç‹(½`áÚåaÑWŠŽÎèá–½ÇøêVŽùƒ]»¿¢ÄüãzF3Ëlç#äyd…)Leua0…|vŽ7êÁó¯É ½]zÀ’%ÚO¸Qµ¹x}‘kÁÉÅ3ìßÎéí‡I•å[(sÓI¶6â5ýšÂ‚aÄ\æjÛL£w´Èlsß?¾p,à~Ù*·ÖVµo§_%òê&Iø‘ØÞºÔæ<7Îéd”àpI‹©Zkr•ç3Í+3ê;sŸq X?M_ÐÐݽ&åµu%scÖfxÃÙ–;T™®ŠÎ‹l9ðE^VØÓ@r’ÖUÚN“ɘûé8o»œDÈï=ÆrÁk“P$ƒ+;ëoB‰õ{f»†gwŒnÅ,Kl¢ÿ %óŸÛçvX©¦â¿Ž­jÁÛ$¼½fZƒ(uºÝ߈ð¼9s'^Œé ½ò4Ð7Ïdß¡eÕ|mÆúÂ2ÊñÑ„©gˆñmQ‡¤ 2w ¦A>HÔèÙðæ|Ùœ†#ñËü¤ÏÅ.î€ËÍt/LnjNƒäæd:gyû„ëËÕ 3óFÊDhgC\…4ðÛrWóëN,£Ž› µøádúXÁ>vÌÌøIwÞ–µ;ŠGõ‚cÙÌaoMhÀìÖâ¶sC]h‘Ú? çÜ Áw y¶¥vOöÉ^0ƒgãQÆq‰Æ#×WO²…üq^ت4¸®ƵÆ-‰Ôœ¿ÜsXx•9®¨^¥Ýë x?X.•«®ÖÍ!RñPAèM§ÖbˆT7꜓r™ÄŸã¦I_ȾAO$Âú¼¶ãº¨°ÐÖV\DöÑźÒU­2¥"£«ë½µ^Ü©~~L#šÇqT”é"y(–Ä[ïNQà ñˆ IÆþâM°pçÎiº“¤ ݽí’jóZ‡ò5e*ƒÅm’8}CV ­ ñ럷U…>]<öˆ;/yFŽ -ý‘9Qý¤A$’¬Ðß”¹É;'Vg®Žœ;Žœ{­Ã²™Ú1î:òÈ7½´”Ì&Ãâéã9<þ§göÚsÅ1 ä0&UómÉ£U±\%w»V5›«òétbZûfæËêÓš̱…šǧŠ×V4`OL‰F¬AX|ÄãÒq;·ìg9:Ód!çú\§ÐЭz¥‚8øi'-rÆ4*±bÔnü„Ê^Óîà¨({ñõùµóšÉ§ÐÉ;µ/HMÃ-0Í `û–ç˜`†DZ”uJ |ÄÛ”•Eæ•dH`h\vó%CñÙ’ð+^R)¹ÝÏ·GY=æs¶IΩ̊I·–zô$Pqêî Âz¿“8{…dp'¶3–[[侫Ûì-'¤ýYÒ Nºœ#§ Ð¶ÉùáX¦Ï*ƒ–%ÐïðÐÛ”Ù#ךÝÒ=1¢³+¤„cáÂ<ˇ@¡šk¨\û§4Nð‡‹ƒë0Á õû;]Üš‡3)LeùõÉRVäœÔaÂ]w&'o©Ý• yà;؆ϕEŸ9{²9ª7hÀû‚¸ÝYL?0‘‘‰1ª3CN€‰{Sß’¦7{»wWyÜ n°6&ÛLû‚²²¤ïªúß?lG~c!o+÷¹!ðN/§É¨×á)»õ !¥4Zq‡.Ù,¼mN….¡·[¬quãs©üƒC.Õ).Æõº*ÇŸ‘IDöóÎw®QÂt"Äw/g!ͪÝ=&oÉë¬ÓJ˜f‰/δ™03ÝíÝgܕþ_[Ö‹§ˆd{¥ˆ ‹)?tÞÙ»¡uÞÓŠÏz}d˜i#0¾+ÚÿÃiOã:ÆeççÌGµžc†ŠE:ØK¾TéÓí c¶ùKÜð UëÉ ÔwDÁT`Ú|VÖ³*—ÈpÖC¼š³•,Øü!\²œáçc6N½îßojæÇiŠ•r暦²Q¼úÝù X;&Se¤êW\ØÒ½Û;êÕøáOÆ·ÂsSKÔžPîíz~Q±ÄɼFPØ=·]2° 9´Ÿ¼ƒgGÿm3Ÿ(YvÀ/NîcxÌåÉàÙ’(GFºgÖ1éR/$’Pý¸Y=“XA;Ì‹a[[Ë_è:Áž©›¿…0Šèê|Y¹KU¿QÝ©æ ’—eUŽ”zúÄdõõ²,߀]ºnÜ`_„þ,¶ñí¶Ç4!Ž|Áb³¦Å6Ýõ\þôø|A°«N"ϲ%`oΉËÔcP™Hw‘.ƒóöÖ1>ÂéE}ý¦Ùžª~ ×°u7ªŠV’8tþV÷sO‚9ßðpBJŒÙÀ¢8ͤic©JFìva7g‰iÖ‹_ß‹»¤ôäL>¥v9—Õh…ËÒ™cðº¯äJug9nû›˜¿ˆÆåå;já–…gë×å5œŠñµ;6ÁªÉpˆ5nèy…ýŸµÛèÈž:Å,(¾%î4æ¤ Ò/=Û z“í³T1ì°&¦›Ø0é¯!³nÿjY©B†j_wÂì±uÅý@[æÝHÎÎTã£ì½K¥z‡^5£yv ñ¢Ñ8µNÙÈŠÙ)Söø ¾4ÞhÏ2†ÕÆnÍm¿óŠºýÌ¢)è°H–»m§¿Y„Ò)ðÒZXè¯{ž×8óÚa/(¿ê ÅÚ]er7ÓBÆù3)½ÎÚâ2­B­ˆ¯¢YŠøäÔú`‰Ã?TAý¤í>»xThå`Äììš~KdßTçG¤~Ä´îÝ£FˆÈɉ8¡<235#ß©pq9kɽ7§"¯)Œ½Ö©9¡?PÈ÷Œ¾Òtëâ9p±3“® >zAëH''ÎòÌe&ÂÁñÖèÇK1Gp¾ÂAn»ü9]©÷UgTÏú£|cepJ ƒ«ªý¸ýÌeqÞû+¶1–ÕÀòüúa…J6÷- ¯j{\– ãùÀS¼À€©èú¨!wC­‘ulb|»{Ûª‘ØòoNjR:ë?rHïÆ„Ô¶¿ Ácø@·‚3=¸?(ýªáKÁAÛ…õž.aÜý×}Ÿ[ÈI¶â,—¾äá©I–Yñá¦ÌR5‡ ±^ §ëä\ø5þpG¥GæZ£sðC’t™ d‘È—cñ,ÄdÏ…KøÜ9³R‚›“žH,8 fñ(\ÀqƒŒ?àÌŽ|)ŒOY¾VûQœžg¦û;ROÍ}Ô‰çîõÇ¢.=DÆj+˜Wí*jje­ô0»@ç¼Kmy–ýÛ¢iÌß^³´>h*û°-iÎon°L'|ôÎ;Ä~áúJ,_¾ÊèÃØ§Sbg¦C ÷û\†.IèPÍÈÀQ«‹Ìý*…cd!ÇýÙ ³¤•ðÒݤªo>½Où/oá¦ÉX¾ØÀ›¨®ª\{Ó„á1pm0Þ¿Š((¬,f¦9-:£m 'W{TçÛm{NåZé| ŠonKfWçÿ0ÍWƒó–¨‡CË©)»ò'­†%[Ü(“Ó.7Y=MlqÈì½{¾6å*³…l$zæ4œö*™×-D~I"Mß²‡Šænw¼kù²Øß´ÉC§D½¸“*²Ïž_?+ 3„©Ô^Ž¥Æ?Jþ\X/ÚS[ÇØ¤@4eù°±©¤û3ð*­7ït)3ßôag/ꛊéã£æÃUçÁu!vœ—%/v.RæÙÆW1úbm3’= ”¾~8¥Rß–ñOè}áþŠ4Ô^þZzº-Õý›µ{@½õS0<È»¾\ûWjÒ~zÕ>ZøR2qÞ4€ç·Õ\æ_­4÷†Ê™Œê¿F…ƒñ=—„Îú[·GZò§oçW×ÈÍqãæßlá^:eY´s"®sfž,¥JêIGôBéËʼ Ä޵¹˜c“¸á…ð’gWQ¾¾§Ùy£.9SµœˆòVå =ƒ­Ì‘. À~NµG¤gZ{ekªA«nõyëX CAD@ÙPËŠ3oéqàldöÕÉŸµñëTÝ*ý·`/ƒA1«à´QözŠÞž„?&Eˆ´nº,6^ÎëZ±Ž4;-–éBO5£\ˆ\~7êyÙ»ð ØÈ0?îhɬ¥í…4¦Ç¶'÷þáÞ}ð-Ç•t¾¦“,ªâ0j.Q›ê¢#I.oÇ4ö÷»Íq8^nˆ0¾-ÿ$nœ:©÷¤‚íÍÚ¤Ã=¸œ³âƒ­¯Ú`òÄíªÔÖ…ötŸ°«O¸ñDAµ çv›&G]—S?< i¬%b,iDW?¶‰q4ˆ‰>eò™©ýŒ¼ÁR÷Áiê]Ÿû} 6IØ|Ào²aë;v{²XÓ¾CoršãžncU?úñhsËÍ 7ΟL?£\-®oò±SI/yÖÔ% ¢ XÕ}êa~T.‹_ñ1ô°ñÞAî²Ý;ónË$§!“Bîsæ…ñz%Ò@‰”úö¯:ŸUì÷sC[p­ð²>êú| ¤¤§‚g…GOf*ÏùX´ 5Ûá¾XÝÍú* âîØgœn0”ûêxgšgõî¨R@ßÍ®åéž/%ײcþ¡ÜÕšAÕj’[»òXÑ]~jzê˜t‚Ë#Áàø1”åQô"U,,§éV‡ÛDü©:kЮÂÉò6Ž˜$õåG7¥%Ý*¥‘À‚ ³;M¼âóiRÊ‚‚»bëÑ:ƒ˜kéK ðØ±¡Ó®“ÈÞ"|úJµ‰Íç㟕¾ up‡-a˜Ìf2 EB¦ßáfÓ´:Ý`N±³ »•Màï³PÛrµ²ñ\æD–âš »CÌm#Gsà!ä7'|m{ªY>i¦Ü걉Q¯Šg¹Ì”‘èÙ{…;Á7šê‹Ô÷íyŒ©£VNSöŠâ(ñó `qAG¤Õf5âû°Îî?>\2‰þJŽ+Ê:„ÄJš?`hfjÄÖÍnuÜðŒ5EA1þ-÷&„yó¡£(Í£ hM%PŠ Œè›&Vºò)èÑúŒœ/R÷Š;ÕUF,:*|Àë·jJ)O~î{X^¾ëZ…yºÇ=²0µM"eð‘¶c–asb¤tÓñÛL5hQI§ñ-ë—ðï×»7ÊësSÊÂÌêÎê%ø5Éw3È^wãT*½?º±Vú ûËn¨8+ëmßl%ʧò˜1pg˜§Zªk"%“ˆÂyåªÉ:±‰“sût[Y¯ô ¤„ϸ”ÙÃðêÌòŠ?"åŒnÞ Ïø/&ZEZ+Zà‡Q ¾qã7˜Žþ/-óÅ— endstream endobj 1402 0 obj << /Type /FontDescriptor /FontName /RHUHOI+CMR5 /Flags 4 /FontBBox [-10 -250 1304 750] /Ascent 694 /CapHeight 680 /Descent -194 /ItalicAngle 0 /StemV 89 /XHeight 431 /CharSet (/one/two) /FontFile 1401 0 R >> endobj 1403 0 obj << /Length1 1492 /Length2 6671 /Length3 0 /Length 7671 /Filter /FlateDecode >> stream xÚvTnû6%1:”–I(½¤»»¥ÆØ`À6b£F£" ‚¤4"Š´4ˆ”„”¤”¤ H)ÿ©¿ø¿ï÷ó}gçlÏsÝñÜ×s_÷s&pÅÄ\LÕã× ±bq°ªRå [TX±~($†ŸòmÆÐQ¤ =ÏŠš÷=TMëë`˜]Ö!V‘ž¤ Ä­z¹>¸ët&ùÌG]ÚÕÝe3ïb&VLWŸËèSŸ•½K´L°,Èsy2ÎÆKŽã¦fÚ ¤;8eÊ:çÑK„n%Jâí$ðkcçg·cã&=`z7r¯¶ž®Çú_R8W0f+Š“œm#rЛX¢’UB¦\¡ÜKzeÕmñûV+u}?ZT;¶r:ýéêÑòðJoéÕµ3ðANo8Ã~‚»Œ©ö%oëÙSYõ£W..ÉÄEèñ]áPT<¸ö€gÏ÷î´?Ò4ö>{zå•-)­á«°nu™S8yW0›÷ ²´= !ͬDÓ]pFåš*ªçX^s›“LtoÈ!Ê1¼ÆÊÊñ^˜“êkH¬§{é(‰‡qÀzó¬Ï ÕžçUÑÔU'åˆGï~öÅÂ.󦢕‹Ç­cS…²SÓÇïò0·q r“Q%zv¦¯Æ§9o'Ú=«{Ð ÚO2g¢¥NPN«!‡…P|z°bÝí†iIÙIù1Îñ¶ò¥K;òº;Ü逸ʔyr«¯QJH+/å˜Ù¿þíáã«•’K“ÉmÞHÏ”zgÝñ½OÎm™5íšå–BÊÏ5Ö{ÔjFˆ»í3mõÑ&} “äEï¡Nü» Ïhûp®,1“’÷L)3ÔØíÌA8)šõì’ÀŽ…¢ÆÅù£c…Ÿ9 IÎ÷Ky6‹ži²¬£º™ Jœgë¡£á€Ék3+Î…JÃ.òë žû,Ü7Ñå«3dJZ· ô¤ëìÞR™{=6“Ål7¯'ã±ûliZÑ]uºj~ð¹”<«¸™*(àXÒ ’°j«gªoáxòù}wÏ'Áù½mÔn#;¢‘â—Ù= 0›ó\\…jD|øÉ”ùžÁý·\Q´Áú“†Ü¨<…/ ³öÎ&©D7ê'üU«_0šý?µyÊü¹¶¶mêtUIú¾Y¿ÊyJÑ{2®8­ÞQÖº)=Bý`÷J@Þ¨ãF*Í1ËdWá¤ÿAÒý)Q±‚(xÇ—×£M”Ê ³¸å­ å{Õ>pŸöôÆ'‰å\”Õby¹%”šÛ;@þƒÙ‡‰—’ó…%1¯+FÙŽ‰€ñ"Fâ'?j"º•†bc¯)æFÖõèêruïî¨óúy££ªr%¢ TxÉEͳiÊM©ƒ1;£·¢çúa‘Ê?‰÷/=]ßáK•C&WÆqcðŠLõ‚Ðc“ :jÙØÐYÇŽSk'k²Gâ$>ïAEU`J¦ƒ\ÁÊQß‹”g0Ø*)è{åS^ÍÂëôÚ Gy«ÝŸºêºZÔb=šéÖ^Çwš?tN¶w¤ç8|*RèÍí¶¹f¢îñ`Z›B.³ågw9-«ý08œöp—Ž>çÃ&—ΪÛÄ}§ˆbcÐîy$v¥ÿ]–CªäÆOêG–¡Ã˜Ó<•ªo7é´?:ôJ˽|\Šô.†îSœæ…ñ¾mN3»vS1:9dÀÓðùiN±3ˆA.§Á½”"™«m6b»¾Ëq/ÓŽ”>•ÓóH]¥ÔîVö-´‘É,¡{Ôk è(bK™í|¨&â;¦– [›ÉΕ¼ÝîÈÌX¸òªBŒÇ±.wæ«^¦˜ñÅûHZ;ª•^{òQð3›m!Q¯;²u8”ƒšQ¬*³êbÖ!·Î5^ürJ’ëzæm*l±t•]«ýÐÂÞ{ÓÊ&{7M¼Ûü\ëmK¨ïxmPƒmXnÁ+6ÄtMóð¸ÀÅÞè£`žƒÛÚ† ¦acŠ÷Lúw5†¼ˆ>í0Zx÷ؾñÖ[>¤au‚Í™>‘ß™'òHÝX.™—?–gçÆ1¬÷ÛKËËÙXŽ=âIÌ2çÝøvI¬}j²víêKR–›"·÷£GGsÕIŸ¬ïñfÇQn¾ôàéYêéµ4Ox”¶V–æ0ÜLo|TÄ1  ¿ïMiuúí4ø<äL×Áεb- \SƒiYuµÒãW© Ù½{¹9†5Zøs GKfPÀÞ¥ Œ4ú9ôCW¾æÓ+×7%H}üÉ2ĵ*¸mTÓó´BïJ[«9’0„&X+J§ù¬’‡ûTE$pòfoÿôÕí ë›åá/Vò[¥˜Û>( ¿®²U)9nì-‘>)]g¨Ãë6ß›$Qb~ta-ê%fXýÕÇ­hU Å6µcå‚×í¼f±ŒU€v7§>ѰÚÓï>âš!@4æÝP†…Ì ¤²M¼Ë;©èû¡Øïðì¶Ÿ¤l$cÓçuC°Í«©®v¡û1Öß2Å7¥ŠïE‹ï49 sW_êט ÙµS®Þ¾%{ëž5dø0­†‚É2¤ßÃá8n:X¯…ù(-÷#ùçûJÜ_X48r”MœÎw±Ã˜¼oúñiÉñ×`eñuÜDsT³\ówrï¢ØXÌÉ•XÚÅž$Lò…¤Ü£¹>¸tÓ Bº27ÍyP£¢‚©D¹sí½ë¶šø¨%Ò Éü1'ã“™ =w}Ñq¾\¡_°Ô›8Þ—|.Kô¤ølQŒV„Ñ3"¢EÊk™orMãÇR° ‡\Çœ,¦Ña;âÛåÑÀ[­ýœi!eNÝ—™÷Øæ“ëz >Ó:GÐË×…³Ï¸l‰$yÅ݈Â6$ÍÕ&+® d|ð›»þšJdpùc mzòaýÔ††lCß%³¦ãÊo|+ÏfÚ0š§†&àQ°w)ºìÝ]MûÈ®©ÊS‘InÙô·NÔM¦ß)ƒ£Íºâh‡ØµûYŠ!¹¹h~R —‹V™-Ö<û»5,f{¸½ÇŽ[ ¯d%øbžµÞÙ«ʲ¬Ô™k^j’®*H3ΞÞµ|+¯ŒyæCSÍ`mOÿóíÅ/+<9>ñ$E¬¬&óDo± ݉—k:zÍDýŸËÞ¥×t!Šê^m´“fQä8PêÌ<Πx´×=‘ÅrTÀw±ÉÝœÉLµzKI¢kŽ<šÒx\Ù‹ž×gN~Dã³q̜穔S(ÏLO6_) Q¡³Qkì Ø¨œ"Š¡ÉI}¥bJMaI±ÜmSߘ:6_Ù¾rg,UÐY¯ÿÙv9ñ;=fý%A Š¢/^è ƒêïÎ'ˆã¡KeQlÏ&ÂÊpMñz/<4†Ø r{½úŽÖ²çÄ‚5ï[=uÖ¯ûv‡³n©ƒ.õ;QÞ.ßY¾ÂLÓ|+âÖéæ ¬{ïòåB-.ªötÉÊcÑàT®ÉSa2°SߌtùL·À0˜”Ì&Ṫû /Xÿ¸@±.¬Ï€"ïå$?¿wW!ÜîU„®bÿ(ÍA¶6c½~ûGÚj+"ƒLÜá&,ì2Ÿó32Œ¡s½î§ûÚêñì°îÝD³A#ñ+½òRÚ†’…­>§ºËm²~,BÞÏîžë³j˜}p¿epJ—pg†q©´T %úùPe¦.ùqO§—Ø’@ sõníI›EFnz¬,4w@°2ÂÞû3ÎEq s4rïR} ,1+ú°êkbëS[ÇÍÜ༘3)i /Ê6•ñ¸Oc؉»RC›R„ _±5åÓüüÓ?ºâGýuåz•a… Æs”rénç­ÇKõ?F[mw2XÛdÇ-Áwz¾:·LÙˆNõbH‚‹SÑÌ9Ä”üœ=%§S7ßË÷v™èmí=·ïÂÇw³IˆBæ ,®ê™íªã®Rýi$òUÁ[SCúOC/ëÞÚXÃê³;ÖÝ1%‡öª_FæáÈo/>-e¢¸å\|gû@¥ÚZýcÿwÏ8 [†ÉÚü›zS]c©Æ5Jú"‘´\"Z3ßÄ©ôg·óÏ?XOÓÓMoÌYÛ] IÃÁ¸LòOÇÍî±%¯Y<…d×<~TÚõ]>ÌLߎO¸¡5|Y{¢øËµ-Üò/»K \/n]²¹3„Ñ}}à0PRèZcƒPÞ9kôCñ»x§G¼÷®S°t9óõ¥”êhmîW_ë™%¶eMzÞ;,ZÜŸWÐuémT‡EѵëM}ÞiOý~iÅ ¢H¹.ErÕdr”c×&Ô]ì@rE?ãÐWù=»Ë!Ìs~#‰k©1‹B“—½Yj¡ãWì·Ûx†Ý[ˆö2Ì}YH?^ÄÐRQáìf [ÈüÕ_:åÿYÄ`†/‰YKâo§/–Ö*v·OŠ'øuw˜-Š¥sÇAÚ†«æ>"ú…¤ü‹]Kx’çZ۵Ȧ2¶önnïµ8O$ÇýÛŽnDãÓ·8ÙÂxm"ǾOÁæ6Øt<Ý~™Qõ/jÐb\ðDüWý…î¯WƒîºÏؼ´ËÏŠºjçnIqƒ¢º–¶Œšr-Owˆéîx…åjéÙ¾HœSéÖPÝ”j3q»NQûŽE8ýW C—VµóרƴJoà´73ÝåD¹:¬(´6SD1y?}¼"Ú#`pvYœñtj@z•úv150½åÕ¿†³,aYâ;Å@ÝýjÒ~Ô\’¢éƒ/´a“«cœ˜×ØKì1?¤UGˆ¸’‚ÙDªD¡šizS!½¤ ‡»Êêª{.5©m®Aàïd– #FåL Öò.xAM´œ5î9ÄR„f_\²ûžòR¨"éºwTÑÍ;tõ‚å¼M7¯–ãòåû„IŸø>Æíˆ PÅ+>,©áõUWCû¹sÞÚËp=B¿,-Ó žVpy9X¥þB *­¡ÎÀ_*Ó°äå–Üò#­VXj”'Ìük™ßp³¼è'jÉ•Ÿ¢:µ„Ÿ œ@_ð’…BgÏ*ci9w$cí7ÙÛd/«ö¶f†Ke:¿sx¾’ê³=-iJ¢Îþq;ǵY†L\V«ÑS}´@™Ô“vÍ\eƒgiw¾ˆ;áˆö‰ªLoˆàY$ò²àœo›Ï6êZQ±o }CWwÚRêdæ·9¸SûZ §)ÌAç˜^~igôƗʬ: SjÖœ ™£P ¦]ÝxÄê4œ•ðáñZÇšý(¥fÁ§ÇêÐæ, íÉ{Ÿóƒt±Ç!H"!¬bÙånÒfåm¹o±|íZé(Û™Òµ°# ïg\2v_N‡u¨` gÝX²¯Á’9YÓöªAàF®B93î›™oò%;ÇA ° ©©‰²í"ý¹³çŽÕáÔ‹’.»j*üe†®µ[ü×7ýRïš¼µ†ÙžïÉK¤]ýêÃH60+õð ™?Pbeë_ÝñF§–¼ÕÅ®íK-¢œT^*Ó÷IiÉ¿wµ®²T…Ö&˜‘æÓéK¾OL::" þvaÖÆ‰ÊV”’Ý’¤FÔZ·G#³ä†‰)r7yix›j|Qð¬ØŒÍ‡ýøÉ±Ÿ’<²IknYOzª3^oš‘9›ê_Ác´·3»ä¾F(Tž_gÉ»M=ù ÛoÿÚdþxK#¦ÁñYÄ`òî–ým~"†*õ-—Ýo^†¦ÄÁ‰â<ýKy{º3ωª®×#üiÊŦH¦³ #GÕ,¬î“±ê”ÎØ²Š´yë…šæû^¢¼[ôØ%cJúloÿ€æûÑ|Ìé «67µ5hT3qº¥cM¦ÆÔeê†X:ïµp…q2–ÜÛºøn†4 DFä½ HñÏER×+e³¾ešìYÁ ÷{;vµ¨<âà +H-»0ˆ^@ ‡–y?ë¸m”‚oBìçjFĪ ŒrÜèòĹ]y»k¢™ØJ*|©—V'9ü|¥¤ÙÌÅ:-YúX“,œ;¿Ó⣛_iMÆ%)XΕXúÁÃ…y†1/ô;GfŠ=«"N^Ô¡ÂH«ä–£Áݼæñ÷,HʈÝaö*SfÜ›ŠvH$¡'î~õíj|,ä[À®¸¹á#{ãø‹]V5æA5ǧíu¾¯ïgÌr„Ÿ”z‹ßü´cù‚Ï#Œ^d+(Oê01>/Í2Ç׿Óå—9‚±µø‰=Ù:<Q¸ÕÔÌñ£Þ|0o4Ú"ëh;T'fƒxÂ-\: µXž¸^Œ¯Žáà〗޲¦lŽ “£|fµ3d@¥yjV½‹í&IeÉ 5tÛ‚xü\*dÚ². 6ïæ«=šÛá7BP­¦€4ŠåžuJÙÞ*Â{Õm|ã´¤aÑí?Ÿ×ô\;©-â?OÎSÚ÷ܥ꛽6¥fÞÞÖOMJÒ æxô¢k^gû‹FÐ0¥ïM»¯‘ĉéQá/..^¡ pe©ÿ$´ÖX—Ï*VP s;…#afTMM 8èSäŒä™ºÞä¥êZ9l„iÉÍ :»Víà¡Íp’f£ªÅ!¢c¦pŽšiï|Ó‘.Sž]lÊ‹Æe^±ZOx<ÒBŸÀ©çÚy[ÊöU‘-0ùĨÔQ,ûy}¸µjs[úð{Ôe[ò:œù:.…DñjÚS}¶n$M`ð…ZëÊ«QIƒ/žœ³2 °ÚmÀn(|xV:K>/L´†éÖŒôÉZ5ù4‹º~%bFJ̾léD'£¢è£}ö”$"³Ý D­"-ùÚSJºõˆí6“™l–ÇÕq˨÷Wï½Lj,ÝÅ» åpÀEœݛͻ¥õÞŠ)õ’ÝÛK ±ã×ö8âÚàe¢“φþ8ƒYÌes·Ûhî]•õ÷]%ýW¿ùS]ɹ‰úÞ¦úî–Éì|LÖR»ó¨wÜCµÁú i×°EÛÖŽœ¯±ÅJùD1‡'÷7Èñs)G¯>A’»>¡ßo®®óèZŠKZykÈnBç9´ìFG.]  pï†ZÈ]ZWëTšT÷(J£Õ6§œ +=™7mym/ú ÉøÑU_1m¹¬3©Nlð2Ödäæ8§/ài®…¥_Ÿ\]m¨R“Á¶s¾¡h•E .V̘½m?\©±Óƒ,Ú| ÙYz†Èê|xá ¢§vº5»ôÈUúþ@ÞÛ›]ÚÁ’¦pÏŽ ù–Ó­‘!Ó«%Õ ìXÅqIÚ›Ýd¿Åµ ­© 1Œó*Eò¾m¡‹!¥z=oc67ˆ.¹Ðœ‹WpRÁò¥' ͵S`u$ÎX§ÒŸ7úËтӟB Qù“¼ìyIûv÷™Èíw8³¿SÆÔÀ—èÉIX—Èn|H Ujrs«xLcüöU¥î©ÑˆAìÁÂñ6în?÷X‘pÝÒùºy–>2$×û8o1ÒíR7‡Ûç{7›”ø6`Åùg s€ÈÃ$ÍÏTj …ÄRcĹdôséî5mŒ+­¾d É\_NMÅ,ê+8á(|C™d_«°ëìYûÆå endstream endobj 1404 0 obj << /Type /FontDescriptor /FontName /FSMIQE+CMR7 /Flags 4 /FontBBox [-27 -250 1122 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle 0 /StemV 79 /XHeight 431 /CharSet (/equal/one/parenleft/parenright/plus/two/zero) /FontFile 1403 0 R >> endobj 1405 0 obj << /Length1 1710 /Length2 7936 /Length3 0 /Length 9054 /Filter /FlateDecode >> stream xÚ¸TÔ[6Žt§ˆ€à€t Ý Ò- 3È ÝÒ]‚t H#¡¤€t‡ ´´(ð^ï}ßûþÿk}ßšµfæ<ûÙ{Ÿ}ö³ÏoÖ0ÒiêpJÛÀ¬@ 0(‚“‡ (UÓ1ä€@>. —‘QŒ€€þÆqõA.p0 *ò_ Y%‰ÉY"D5ðÌàáðˆðŠ^ Pøo"ÌE g鶨qžÁ  8.£,ÌÙÓlg@æùû+€ÅšÀ#,,ÈñÛ ír[[Bj–{2£µ% ³ƒžÿ Á"f@8‹ps»»»sY:Á¹`.v¬w0  ‚ƒ\Ü@6€_%Ô-@JãÂeèÚƒát`¶wK @ÀÖ (éâ µ¹Ù:ʪ gô/²ê_ÀŸÃðpñüî÷¯@`èogKkk˜“³%Ô µØ‚! €†‚*ÂÁ°„Úü"ZBà0¤¿¥›%bi…$üÞº%@AZ `‰¬ðO}pk°3ÎC~ÕÈý+ ò˜å¡6²0''Çýµ?9° ÈyîžÜšë…¹C½ÿ^Ù‚¡6¶¿Ê°quæÖƒ‚_¸‚”åþpî0;Ààç€^@ÖöÜ¿èz:ƒ~ÃÈ|½aÎ[d _°-ùë ·t.® _ïÿ6ü{…Ëð[#V ;0÷?Ñ‘0Èö¯5²ÿ.`€1)?ð×ëŸo¦H…ÙÀ Ïÿз˜û¹¢ž¬2ûŸ’ÿ1ÊÈÀ<Þœ¼ÂNa €‡‡G (ÈðýwœNàï꣚–à?»û¯ˆÊP[@ø¯"§÷w!n”ÁòglXÿΠCê`ùüM€ü@käÏÿóüvùÿÓþ¯(ÿWùÿïŽ\!ßv–¿ÿ»¥âù‡Ô³+9j0ä„@ÿ—júk Õ@6`W§ÿµ*#,‘3" µC꜓ç)ðé_8®öÙh‚Ööiéïf s@ÀP& þuï ½€Àÿ±!GÏÚy·À‘-ûm!'ëßyå¡Ö0›_#ÈË/°tq±ôÄE*¹âxó gÕäñ[ân.( t kôØÂ\p5–¹OnK¨•‹¥µ#yGÚ"~þØøÿÛöWCÿ6ò!MpäaáŽÈÚÿÇ )Ln+¤b ÿ/È‹„‘@ÿ“‚ïüïè<ÈÌVÈÖþ pƒ^¸‚Ý,! (Rdã¼<n»_W2ÈiFêèoÃSa7мÀHQýÃFî‚<ÐS‘ ê ÿàGž d÷ûðïBÜÎWøèÿj‡µ«‹ òÆú=7È^ý½þ}=‚@ kÜù˜µhˆÃÛ¶«ijwÎÍa̕ՎÈxÃÞ~Óìko{U¬Lʼn26U‡Ò4çÊÂÆg|¨Ù¯Fd.§*å"Pä{×eŽ9ó¾Ÿ~°ÑÚÀ¨ìFÓ"×IŽ´hˆ,ŒÑ™“ f~† vñm¿Ù*¼é€o ïœÅÄt©Ôˆ霎=Ô—†t «ˆACâXRãtW]e½ÑÔ—àØK™¶hÊïžëƱƒŸ?ÌîëMã;ÉW鮨 #£¼ïõŒ.g(±Ÿ’×;uÙ@< ¿O®u0þñœ®å“GæYºß¹ 6'FÚsÔ¤1È0VÛ \ ¦UóAr»ÑÐ0ë:ïÔ&a%¥—†ª¨ ÑX~Á|f[c¯,b2xxMSÄ!Bý_ d4!´ÎC¶ ?ÅRdTŒdšÊhêt«®CO8¸Î zÁÏ'AÏvš?¶ v…ÃiÈ'㟷gîõì&4)ó_L‘è³É jUfaæ¬q9q“þÚ¹9èšï:E¤„çÆèãôùˆòtIfuU·QÒÕq EB¥&‰ ÉËhÞ,T |z-Ô,ž"ô[;¾'§µG\»Ž§ÕõÃ,ÞÊ嘛ÛL%·ÃÈÞ¢Dùâi<¸q/þ玌áDà±àÌr®Ì­ž¿9Y©ŸI®mÌØóí³†£ÜçÒuFÈäÎ ÿuó‚O4íF}±Òßíw·3§l{(5\_ “Y *ŠôÓ‡A*ãø[e?÷‘Gê÷Yª¶5c»t}²m<ûŽÏc|)út#ÐÙl§AýÈùGù 9)ó\¨0ßz•Z“°7:~„¾¿Á¹)åXCï ” ÅÞ¶w"©²ð ϘáesIݵ¢óñE÷£Ïd­µ¸Ÿõ{Y_¢)ýpÓä[úÂý\Cࣧ4¬³öÐooÛÔàµ\Z f„áÓ•’Ó°f’|øéOùÏ5åUox×Íy[ðªZÊ`J)Þ¸<”÷v™;Û\)}ÞÁ®.® ÔÓhgÊ'Ÿd´ÄM¢žý`J^O‡“¬ØFy@ÉË‘Åú’/ØãGuÔ§8¸­šÑ$–.A8>kÂä_.Sl9¢Ne° b¿@3^ ÃÜ¿¼n™Û™4½ 2›ehbŠÞ<=1%‚¤Äm4Á[˜BB>pǯ­™±fÒénöXœ¢®VÎÕ­>lסÏÛIùéj†¢‰+Í’! #N²˜ï]ÌÐ]Ǯʈ«m£¿,#3寿ô÷õŽïÊLeù(Lxï¦ÌÚrÄ:°¸^?ùeÚ»œ@½,åÈžþ½üëq¥evU½ðAÐÆf¥þÂÆHp§eY¦éÙŽä4>êÁ‰"’”nO,p§áÖe^‚©ýõSp¼À Mó€NËùçC<õe‹‹ÀeÛÑ¢:£Ôξ ·9ZaG|O’Ø J‰žÎ¢Tr³ë€$Êžì1‘cíÓÊÏYV^в’ôŒA> ÏEît;µ©Ã1CEÚš; [M-"wíS¬í±òOævŠl˜+»7;©9[ñ?ò¢pS÷¿Åú†ÏwRGÏ-Qà; …Vý¤êÉW%}6c=tÚ0 »?Z•¥ªT4Õª²tvژÚ~Žõ"‡}vÑÚÐ3¥wø]QnÖ=\2sµOÑ7œRuX@R}Ìjæ³ìÑÚ±‹2„Né@!‘äçÚçäèd,Á†Á¡=—Žê †ïæÄvd|ëR¿:Ö-vc­h°,LšñÎ1Q¥Þì;…¼C‰!ú ñq †M~Ü/{J©µÙJ^’çùŠÌ')nÁí1é»_|a.mO²|ƒuìÔÛy´Dµ¾½Yîª š{ˆ¹P´ŸJ Ì°yMÙú¯nê‡ÛôоøÅû‡X¹9¦ M¨f$[ýk¨,®Ø¶¸ÅÝq!ü!í•« vi˜>f¬û(ˆH3wX_Ñ­/ î$%ô:Ýêëõé,’°¹sðÔÍŒ ìnÐwêßJ[¥|QòÉNv2á4áhî}kæ=a/Œw›º²ËþÜ=¿Ö„Þô0Œ ÛAšptŸËè´òË<˜CµÞÊ™‡C^+æð»OÍÌÛoøÝÊY9Õ#-#4¾ÀËFMN(ì¬í¤Yl̰}ĹÏÃ}E+Í&Ïÿ$Í&ü[!,6ÕÛT¶\‚áMÇËŽn,‚­ÖŠ2‰á5Å©þ#ð`¸h¸ v"׫â®e÷9Æ9›ÇŒŒ0ˆk&*âÕH×^-¾ª~¤ßÙÇ»ŸWN÷!–#¦Œ÷÷¹è‹-âß•™¿Ñj?ò§Úp~ÑK;ˆ…ODå÷äz‡öhbø ë·­Gùºy(Å6<*¹ùŽŸ¢>ûWäcSC^åŠohV¿ˆlŠÛ³,3ÐúÊè2ùðÙÛ–¹ç…Tù]|eO`OGúä“È^ã¤'…4ÁrOä§‚«Ø*JÏk—ÓÝŽ\4&mF ‚P¼B’ 4[á`¶jêýç"çÛým™"lÃ˪4¶¨Ã1ä݉L–‰¢ì4Þäï€á8¬c Ò‰vÝ5*l»ÏUºPõ™½æN?IÓ,°â@/Fñê¦uÛ’Iº3âUGVŸí|5“—Þ‹ÀÅ^-5óvÑ}»¿¬~í“…’úó;¿~Èží¸^Ôî Í@ßËŸ>âwÑ™Èß3]ü>šÝ˜;ðLáo%´±¾›Il¸4ìvÎÒ5Ä}*xùId‚³Î§ñR«Òæûs}åV(¥¦hæ§"´³W[£P¹è¹ƒðí©(; 5^ƒ6fœœOŸ…Ãê¥jj×RïÌw¥'üùJÖ´È¿vÊÐêŸù1 Ëç·>ˆwÇ}i* ¨ØyåíŽQmNÄÏwY¤ph7ÙY÷ô•LÆÎFÖ¾tiéjû‹·£ÇCDWÞ/?mÀŠ¢ /Œ;äy£iûp §ò”cF§7Wàl½ ãÑk—Âã„ÏÜ,¾à·5Ôþˆ¡8ëô÷¸Td¸Ë¥)ÚåòÍkÁû”ÜTJ±4Ú¡0œ‘l]S›™$þ°êƒ-ŽÇlCíå\˜ÜÛî˜=‘ÚB­,Róó/3½¬›Û¥ºÕ™”Òr>H<§ûè”ú6(5y@hpG?¡/«´sõJhÊ‘b:CÌó*[Ý•À2k·¨´½—«¸ï¡^_Ôá}KÜì(­lÎJ¡vEäߨ“¿ug˜:m  ŠÁÄ̈›f*e3‹¥éO`þi-…B´öð'SÓ€ØÊHoaÚeí8ÕA éÝî|9{²5aU@£XžŸ'³Ó>\,Cïk °3yj‰VÛÂùs±z–býÑ*Fº!joéð{ÓzRè)ßs‰gy^’órâêW¡÷Ú­ 41Þ/(¬ Y$èç…Øv^v ;–zò]ÁÖ¢®P–§‘Jë]qrîk-W¦…øíÔñä{6¯½åÉ‹º_ºÎS®fp1ë¯nÛS!²/¼LŸîU“£[Û6ç]W¿“E7H•::ýÁ÷ê"Íù1D«úu’ÖVF$ºÕj)9·ó齆Ó×t“^t(Œ ¡›ßNK“$vB÷È´¦°îòTb½™¼DŦ;s0NÖÁ‚ŸFÔïe|cf{póây€î=è BŒz:Úñ„[²Œ×‰’5ghÔö¬¥ú eu-ÿÊó Ëzµã‚w95q ]›m67 ¾âÑrf»-SÁ^9íòòH~åÈ1¾`§¡ò(F«€ášá­ø}ª/A³Ù’»6R|֬Ħ|åT©„»Ó¦QÙÔ|•uSe|-Ê•µ)XMÇpqn§¬á¹ïp“«@BEÜa §A¤(2_®Éš¤¶:³¢W±^`Þf›?Î$Ég­fi¹Î*jÀSSkÅm¨ !xÑçÆÔX=x)»²É§ŽÒ¨j¯s×ÕtõNxok­¹ÒÌyÉÎùªª,OïeZ­Â¹Þ•&ý;•‘°0öж à ™$°‚%ålCM©Z/jçˆ%ü”ÅP™3>bñæüPÔbš|{Kz [I¤þÍëUC%+ž¤†µwJýÞÍîÁýÚf†ã™”„¸¯J§R•äæOô©$O°<ÇÓZîk.3*ˆ¿ êêï“ Ç¼˜¿ÈST™Ó´¯óS¯7nFW4ßÅ>Ö}àµÏíÍÁž$,ÝÜhYþŸŠ.vkz˜=ÄüŸâÂÓ¢™|jâ:£°ýR>+–oÌ å¤íäUcª¹m®°8P J‚×WßFó¡ß²LGL{µP?< Õ˜K Šp"äŠ.ˆ­:ÆÑ6î K‡©‹9<Ù¾eÚîiã©"›#|/êû˜è8ÔDÍÿ•õ ‹È(®Vò"4lÌ(ã)­J„]8zYïNaWÙ6™Tt¶ne2`šÙò‚ž±OÅÇBKÅ÷KþÇõW¹°ì¤HBßS¦€›!a®›™’wÛVTn2Öìþòì’øüÅLá³ç-šcP»3e…mÉ.Dò\<¦ŽžÆˆu¡¯¬“qÑ— A‰@GƒkÇ¢ü²vÔ¡çÕÌÁoiŠd‘w‰é•¼U_R…Ù8,Âg ùˆÃ%ð3žPS}KG0?O±!f—¬=VßÀÄΰ‰òî5=‰vjyÅN‰×÷8çÝl,y‘ •åKCÔ0LÃÔ‹ÛÇî0g²³¯tò=š¯$aG{4‰„×Nªè³>Ø%ÉD½Šé×<Œ=•1ÓŠÓ(KâûÞÉ&ƒÇWì }‡ÆÐ´¢ÜG¸{—k® ºµ˜Â}œÎð~Å"È—S˜\‹»§hS­#9÷¡`w¢¦á-ñ¾ýšiT±Õq'ÎÍ×w»滼bãWé}ìèÄûˆG]å5Æ£Ùµ”ÏònË4¤ Õ”ƒ–¿ ŠæÜ0ƒÃ¦M¨Z¸­ÖÇ»OÔôÈüw#ìŽÊ›¼Â¶©¾eMⱫžÆ3>DxcÅ]kžCyXq·}IýÖÙaü nnÿ€1ù¤j”}.j¬!)yëŒ_Б|å6é £?–R¢ G¥ÇˆÏ¼îqP¤”#¦ B²¾>‡1|`_‚×b$0ö0ʼncyÜ^‡ÓÀˆ4ŽmäÙC榚ËAñÒ%wõÍWu–r‘H†+ þÍ÷nó±Hy`Ê2_äØ½Ùæøw»·ÂÀZ[ù>QgzñÛA[ëY®,GEøÁ´ìÐ^àzŽÆ÷o{¤ƒØ%ΉéÞÍ+¥Ê²©*7²hÜî5¯œîcU, êL1¼ÑúA;.«z–šÐüÀ#ÝçKG1M¨Bór}Ȇ|’Äp˜O£¨:Q/¾ÛÆ«Ç ÕºÆÖ3˜¦GA Ô(zÅ4=*‡˜• ÌϨc„›Ç´ˆ«VÂW#Oóûoumè“Ê ¬Ø·?h„™ÎAª=¾0 ¸Ÿ¬ <+KÜ)§B«Eù!Ü6‹ÅNÊÝÔ¶X?c¨Ëx×ßo/šGáûWZ3"i>·Jœþ5,·‰êG¼÷Üðq½®7u›—¯‡€óŸï[=¹Èºgœåå9š8Í€â©\tœæO³Ñ–Þ·`é •òŽÇð4è<ŽØ¸N¨Þö”Úê½kÑ[½Æþý—B·’h¿µw걘§Œ•`„ŽÉ#®õCª,|ê(›Ó&}±Ýuc×¥ku‹>Ýÿ’ÙõpЇœNb‘ï°>†¢»­ÀlúMòCÖé Nï&¸–™k^bíÒ®z";{Ø#´z±ÏÜeìÆÖÚntnÞådÑÏ–C²¬Ž/¼D[jÓ•=§Mäg1> íF:»”:ªá7IaJu˜Þ¿æKI‡tF"t.·ziXw¸µû¯–çß—Ó?ämpÅM—%ò_±÷Ì×ÀKŒN˜©°ï×¶”œ§Ù¶ŠÐzKuv\£!Íuè½ÄSRya¹Q¢/¬ù1r‘©x™¨ßÏ‹ Ùþ¹~ù@~‹c¤Ï;A"®F#c½ÏøÌº_ÿK¡È•Þ<ïx³ÄOÄìÇ$z¢N-ƒÖ×Á7çã OÙ4¼~´rY7Ÿ:X'W2ý\Yš›%‹Ûßß$M6|˜Qqõñx—Cš:¨¼¢•Â-ùóSTÆ*©%œ«^ûVvú*!‚béF)T‘èá ¢±ÆNÓ¡dúž]‰ý5¸éÒñ¥âY ­R@Fàªz•I§¤ÙSæÃ쫲w£ß3‹¨LX¸;[ê’¥ÆÑ--šßá_(¥S×\EµÄ|Ž K8½¥ñNí}°à{«Ôéd„€%‰_Þ¬èXIÈÕÊéo꘷K,ei–о‰J²iµQ˜î´¶Æe ¯HÇó9&³§}¦å†-ÃLeò ­nø>ã'®·|;ÕÏX±¤ƒì(Þ²ëvD–¿?,¹Bi Œq®`3™Ít-µ!8TàìßÚ9¶Zý³Ž>ÇQ!Däã((¡TDç| ×ÐÎyvÁWF“SÅV§ëúhm› €n‚Ÿë€Óõ‹‹‰°€ߺÓÁšÎX©ÆÄ2‹$X9Å&ˆ¼Ö°æ@¡_6g=£È©¯Œ…õØ`` _Ïÿnb?f3îÎ-©Æ‚„ŠãíS®Š€Ö—÷%šZ <¼]Ú×$ƒe¡Š¶%£0˜S³B—7–ÙmÜù±n©ˆ£d±l^S$:¡;o»ÚHQDTâÛ×2Â%Mk ºólÑLìû¦À Ï(÷ᒀĀъ.ÉçùÌL\ºüሊV@¡Ò[2[ª|ÑÛkÊBÇp”' S9 Æ_W´<¼_Ã>ÎjHŸÞ× ‘Xd›<Æ2ÇVNäÄëÄ5ž -e¤…ç¢~Ìœ®]VÎJѹû l›jsÏÅ¿ÆI%çÓ:™–zuæX’^Éz8}Seu°{c.Þe‚š‰HÚFÛú^¶íõTº‹µ“Ôæ')š¸aqéª31½[Ú^ð-á¹ÈpI湞'ܸ“îL•)]} \<´ñh6!{ìÏ6ÛDÝSÉY0:Èé²À²ÞôsJUÏ].ÛJ”ßÁd;Ó *òŽ8«ùÁ~àÑ8apÌûðPµÔŠÒ$3åv,—SXµe>™µ7'D§à½Pç'³¢ü•_lN)¬ü± a/£ðýTô©ÈO•P£áœaZ™™2?ÐîGüT0]zf”“° “bᕃ^ÓÊÿøzþ]kãí'ZÌÎm^q^r%K_‹Ö»>]±óÉýe‚¡ÓrÜê&3³¼ï»(tO"BÀ›{±/Cï-íe$så¡ —Ž1ûó‡÷+ñ×Ã}ƒ¬_‘´O 4·åÙ6µÐ°ú?“ðy² Ø¿ü°£ëˆbk7­¬*W+‰ÕAõæyrÎøþ‹˜½ŒÑ3 y”ÔÆK·÷v{£äƒö³zÝ‹q{2¥}Mj\bš¶®ÃrÔSÔ1öÕÆÕ“Ulæìù1Å7¥v¼kÌŽòDl‡ùU†¯£jo")G]â%ICfyؘ£äw¦3¼ôû¾±wæ$•õýŽxSá3oq_l P­ˆE|ù¾-AQëu»¼s¹ç‹¿U÷%.ãªæjŽ¡?H;·{³²ÞQÀ€Ò¬áoYðJŸ™èÝ0ÖÎùY‹oGƒúR ßQï•s™É,¿N¤ êʤ SÀerñ+P:“LS¦«Š±7¨XQh‹ŽH Ã8¿:‘!L-¦ðMc[=>ãÿÂûêôu ¯ú·8¦ü¶dv™RÒ,‡ãÎ>% ™L3×inÙ¤Œ,ù¸£ËW²?'rɳíeB²8„ôC»ÊÌ›ÛÇ™ì®ëÑÎFÃÀÐŽÚ,±¼Ï0‹W»*Xx³¾¹Ã5Œñ&0sÑé£âkæBUI/hðFH(9q¬<†ý}áXYl…UÁ­A¼èG¾2ÛÞôp‹TSž~¿…š¶-¹ÐŸÂG>´uør.Ü-µ«gÉ)†yì.>¦~ÎÕUÄæ?ÏÜŒ¶pÓÁ&  oüµ{KFÓ®ŠÁBO‹ý×NÅ žC2ÔîBf%Ãsb߃v’qB|>*òóˆNÙþ?Ù~¢B endstream endobj 1406 0 obj << /Type /FontDescriptor /FontName /XWGUCI+CMSY10 /Flags 4 /FontBBox [-29 -960 1116 775] /Ascent 750 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 40 /XHeight 431 /CharSet (/angbracketleft/angbracketright/asteriskmath/backslash/braceleft/braceright/bullet/equivalence/greaterequal/infinity/lessequal/minus/negationslash/plusminus) /FontFile 1405 0 R >> endobj 1407 0 obj << /Length1 1456 /Length2 6442 /Length3 0 /Length 7434 /Filter /FlateDecode >> stream xÚtTTmÛ.]ÒƒÂ0´HH‡tIã0 0Ä Cw#%Ý!R¢4"Ý ‚€tw*(ðú¾ï÷¿ß9k³öZ{ïçºë¹žûº6m=>Yk„T Gñø$òzÆb!~A"}ÊúLÄaEºÂp‰ÿå „‚QhLŒBûi à57GH•‰Iüíˆ@JÀî0k€?@ ‡ºqÈ#œ½0[;ºÌß¿.7ôàØýßáY'(Ã`”Ô ]vè! 0(Êë_)¸$íP(g ÐÃÃìäÊ@ÚJsßxÀPv]¨+éµü" Ð;Aÿ0ã'âèÛÁ\ÿàz”  G wEG¸Á­¡Hº8@OU å …ÿqVÿãpð×Ù@ü Òýý+ þ; œœÁp/Ü`s„´”ÔùQž¨û0Üú—#ØÑŽ»ƒaŽ`+´ÃïƒJ²:0šà_ô\!H˜3Ê•ßæø‹"ðWô)+­åNNP8Ê•è×þ`H(}ì^À?u€#<à>-l`pk›_$¬Ýœp˜‹TUá/4DôÌŠˆ<@]POˆðWz}/gèo#èŒfàçãŒpØ I@ý`6Pô‡ÈÇì nP?Ÿÿmø÷ŠXà (€Ô'úOv4 µù³F7 ó˜  µüzþù3GËËwôúûïþ•´MTuxÿ0þÇ&'‡ðøðD|D@ ú%&.ðûwšàoò¿Qm0ì¯Í ü'£*Üxð‡úðþæáþ—,¸þnÀ¿+h"ÐZ†¸þ#}3úúÿ€ß!ÿ7ÝÿÊòÿ’þoHÉÍÑñ·™ë·ýÿ0ƒ`Ž^9 ¥ì†B…=ðÿv}ý3ÊPk˜›Ó[UQ`ôxÈÂmÑç ó ÿÁa®J0O¨µ6 ±û#¤¿[®áƒCµ®°_7:J@à¿l詃8 oWtÃ~› è¡úw]E8aýkúED`$ìE$€™ ˆÀ„Sk¨ço}€üp @sôØ D¿ÚŠÎ „:BMØ/ÃoLÚþºÞ H¨‹º/оN0¸›ë?ZA@g$ }x¿m â†D¢“þÖzׯßP¨'B4;€< µ¯ mþ^%ËèÁ·1‚·´Òú,Þ¸'RÅ9óÂÇN?SyÂEî©u9ÝPšö§Òðñi_FžÃfÙA‹©*y( Åž5¹c¾ü‹/¶Pkƒí˜XÚ³5ò£æÛOMqLï%=™þ&Ö)´U±YxÕêºeò`à(þJÆ–ÊˆŠ•ÍmÝÊ6fc_A ›‘ÅQ MÓÝôU Æ®°¢aqßäšcè/¼ÖLã2ú³{{Ò„N_>îª-§¤§òVcËJ죧NÈVkpÅà”ù*_š‡¾È)œsÛ©f ÄÄÄ×M/à(a²’…Eb&"µn«Õ55 röK²ýôöÂ{Á!5BÚŒÔOÑcûÅ`å\¯’Á„ÖÍÖ“gÊuK3)î£{`cöœ(½¾cº¡ƒN.ù¼E£h—cRM­Ë¨å­ŽŒxç&)ب9«N…ˆÂ¸ÉœÞ¼z ±¯Žè4¦ð£Ý—M3yúçtòwOŒ6g2•Oö¶0ƒ¦å<äÉøL{O²p+ïa(n™®hX^ªhœFu‰–òB(W¬æœs—L¬ê?·äà Y > ¯×팾 ŽQ®Ò[ølvmüüåé#ÃKõl?ü¡ #ä½t¤ÿæœU«n VµÏâÆë@EÌš[zTÝüuhq(€±øïN#•Ç%îÖïäM'3„?5•Ø{·x»,³Ü§QÅK)ª ž+ ñÞ­ô\¹&Ä´Jçû¡©òœ‚6‰fðˆ\"C.ÏSM¶€ñsú «n:åþõ½Ý*·ôÓ¬™®½/ Ê„wµK“@vçý°מÏȱHV=ù ç;êÉÞ‘*9“!–Þ8¼¨eµæ5FâóÛÜ*µßÏ]À)Œé2íØgYãâ‹íßaÕ¼,g–AÞÙpË)-€ògˆx„%'6q•[WÞ,]5ˆÂ}s\1dä.5Ý”3ŒâI‡4t³Ÿô­®àÕµ^¿´»¿T–1æ•ÐËqÛêåE£ÇÊ÷V@Á"YÔ‰HÁƺ‹Ó]- 5{¼…Äœˆnȹɇрú¹3va•fm÷¾cCS.úm°Xߢˆ 4U}*v¾ºIÐÒÓv nñJý£ï¢%¡˜d¿%”¿Ô}Ïé"Žn*¸ç#aó¢üZ Ðê±v[é¼Í ‹¬Öç„§²žŒ8‚Þ“CŽÄŸE_|õôÓü\Ý—»77œ q7ݲ? ã?½ƒuq Áq‡¼c_¼›q“5}ô0LÎ.tÙr^âK!œæ¹|¾ÔB|3„ñš¨$”¼”«86=ür ;<u¶û”iBoãí}*õxñÕ,ý%{¤ñE½år:8LÖ–|޳9BpzÈÿdsÒöí}(IN¢•ù¤Ç=áwoƒd®ÈüNœÄ¡ý&R†i©)Å«#¯×Ä$¡>¾¦ïAЮK@ÿj Ö¸Îêbñ[³×‘áxIéµ*"ÂÝsÂÛ¼”€XŸªéÙ4g˜: ÈŸŸ9ã,ÿ»cY¦î­Çâªã -a®<’¾ƒ÷N¸ñyHÅ\»_Ç>¼uŒ“KΕÀG..$)*) íÎè4fͽ¸j‘¨¡Å©ÝS€Ÿ+¡$îâ†÷®þ=ùä•8ÚŠ®·s8 ¤ïnfj ‰–+³‘'¨œ‰ìpÊAx¿ó3¹‚ޏÞ'2TGê> Ö6y¥·¤3.y §I…úˆ;ÂÇëþøzM&ÝgÒúéM‰LY“¯J'¶_§ocLÆê/ÐP–»æá<°C¸0å$‚ï®þSÁÈfúØã¯Sh±9ˆ .íwy˜§Ù¯„ß`èelÜ)£ôÔÓ8-V²f,8 îàœi&M~gÍÀ+€ÌÒ0¤“›³ÓñþЊõPì|óÃh”~û­•ò³¾@¨0âÒóüØÝ|'æ^9tÚ T„ØýMŒØ&˜iÖE]Õ&%zl2L»¯tÀ0gͺU[¦Ä6öhéˆßÏïU”‹"ÍíÆ”2/«Âpð“e6Ø¥·kë㯽rÀJnmÚë&¦þ@šl.hßã—c•ÃWé#Ì©F(>?Cöm¡u¯Â9Ûâ“FØ-“Ô ÓcHåMÓè¹£µ/þON§ˆ0…¬%»È*" `´(3Ëðª¢ÿ6`i^TÂpqÜ!i^qf€ ˜+ÿÈPp®Y“@DZáLý»5í[¦¦XŽ„?Gßs_ák)]àÇ%£EnvÌmÕi«|·´&‰þ9Š$ ?FdÉ(a¯q°qîçW¦†¥ÙˆçM;^îX‡üÔÐÍâÛ¹/PªtÇb¦ÎƸgüUÙc|Ïí° Ï’'öÆj¥Ö·Â£\JÖØEúÊì¡\ÁlÏ0Y]< Ϙ¶ë–Ä·]ÂŽÔ†Ö7^ñ[/Œž–àô¦E17×rû¤¹X¹Y_ ²nðmE©'hVz¸G¥E?°í‘¬›û:}Åǵ*ó™ JÝŽß.{Óé’ðáÍ£³iºM9D¸”b3«³!«Ègùê'™m]HOÈôKá‘u_x\ˆ…›Ô9ý 3D ‡WìX ^¬ e\Lê3‹=“©Ôûëu%^ñnßšw ¾MÉ+)mD6A1}#5½ sÕ§÷Y¢íYøj»¦=Ú¤·´žÞ¹×yÇùB*ïV,åÚz£—7Û‹É|Lbé.¦m„ÃW}¨ìˆßá‘E-^À5¹¼hºUJMã ë‘ãp;ø&n„ó{úJ’¶å´[Ìl5æQçéÕÀŸ°F‡]—Nƒ´Áõ¢±üƒwmÝ.ïú)"+£•_g(·ŠáJÊ6žïͨ¾ÑtZôR³üÒ,„©{‡Y6¿âzÓœ„¦Ìì'|¡IJÎÆyÚ¿M¹WÿŒãƒL¡é*ðKôÉÐ×,ŸI$ûfìéíòùèl‘ H™¡ém5ã4¸l!¶ Ás^/Nüº˜QÐ{v<&ï—Ñ—@o ‡F¾¯tÕ$4¢Ò6Oªë‚z"¹a¢-L7¹$Í ë ¬§¢Ÿ^fJÑ A2¨ÎÝsKu¼g·…s0yÈx¯™u"duw±}çSãš_×›«Ò­Ï¿ÙúM¼˜ÍC˜Üc‰£»ô¬À˜«NSd`5BtÒr¿µØ&»,iJﬧú¼Ç;:%|Gk·{Âôª[“¸<7‰reœ¤ËaJ>}¥L톩zçñ¶º-Ó²æˆÖ•Æ®™`ms=áª-˯d¢_·Ìš½o1ÁÃÕl-oV¦ö:‹d»žÞÕö茢„¡OMì¢éâQJþ¶2–Ôâ_/}ÎÎÝì>yrL'¬ú¶á³È-lFœá­ÈÂ:ìßMr舛ö\FÝÔˆ×GÙîéûLˆÄ8û?w@)SÍ{LERž j½"<¥/Ú¶¡K4¢ûVã|ié^‚¢V!fY yÜ«Ú[‹1Úó¸Io‚£U0­Þaþx|^ù~’ò`ŠX¤SVãún×Ò¬Êz˜zv+é5‰e÷è&>ø“IÁÃÚ¹úÀµïc¯5>â°6¿0²¼"«Q¯)ÒrÎ,iÏ¿" —ÛãèýôàÕó]¶EêOsîW/rÉqª‚G Ók;á[M`} á˜/îñÁãf&t­K[uiTI.?ÇŽÎbŸì>N7àsømÉg~›/ÂWE‹ì·jéêÉHñMÔ©Ö°„² Í‘WŒt…KîÂâña<Ü f.‡V\¹/—Q ݉¯‹Ã¦@ÿj&ùÁâšÓmìBúeŸ¸zçOtL¡e¶DA=b˜d.˜›¿eÍjù¾«y¸ÔšÚÞ¾ì¥4¿¿ :~Mr¬ œa<µ+6²Þs´óHbÐר_mè´ fÎÉ´èûrêÅRyÑžŽµ ûôÈÑàΛɧ~‡®Pi>)Füª”j›|º²<ÚØÜÚž×pÎwM»ã˜C“ Þ^]±{ÿ° Ko‹¯xøäg²,û:¢Vò;õ"1a„º”Í¡á¦êÒ¹@kJuÐÏyÖÞìuï¹¾¶í7Óbc§1s]îkº ¦OëXb~~›ëáŽåm’ð½|^ g JEÔ0ºÆyaƒJü݈|õróŸ÷ÀžîÌz#•º|ì•-ž¨D¿ â*´õ‹×˜˜¢1ä«©äù~aíB5ZmºÕe1F9I¶ÞŒœõ¢¥íN ™¿§ŸX «’4~ZÌ1>©gR-Ýãí¼‹ŠmþÉÿív1 7·eVÝéW-µ?½ÕÉa»PÕ· J7žÎ:ç)÷ëaš¯N¶®!/2©N‡mŸÔ®Q>N”d¹&;iyy³kÐ]®…dÞÞà‹âû~ôE” éåxô[WFL[‡"¤yCø*ã­.n4¡Æ…hpaË,Ë”7F]Hí”—LÇê[Mz5¥GøÑÏʸ à[½Íó¥u܂ӲWDûäþ› ç÷ò¨|—R4EîPù÷é©ßžª…SóÌ^ä¯Åm‰ý7V½RÀb{´?ž‚í8×Ð3º¡a‹¢§ÐcwWbvÔ‰”âDÎËó…o¡¡óSü½f?ù1ª•GNˆÚÙx°EÜɶ¿äP½Ÿm£Ô,,Ôî1©yìðÓÂû+¸°"wØÀvy1K½ÉFdÍd:È?Õè,ÿPÖq ¬ð¦çï켌~z.™wÝ©á`£ó€ž¬ÑÊOSº0×/æ(È3ij¹6Ȫ 7ð {¢‹ê±ëm­ð­6wqí=ƒ±­·_h‡z¥™PQ‘ïàºNS&º‘ë®´Gù]]ÌrQ2£âßÓïˤFÂ,·?ƒék+}£BËo}´«§ ›÷™k¬8W#ýRÃ÷©vÿi’gµ¡ÂÏ2N¬Œ§ÎG·˜x¿¾vW¯2ØkñiŠ”]àbV+)·®å´¤´í_—¤Dñ<žÿÑ]aÖç°Î_upN²ÂìXhT«YóE^?75'I?ü¡ÃQÍ6~ÙoT*Ï>o91Ef>fÒ j´®ŸíѸåSÄÆúCˆb†Ÿ6Äo‰h†On7t¸k1ãf-ùG/lî*ˆ% ËÿcH£¤†ØÛÀÙð~Çä΃©1y!×u W‰a¿\¿–n¡an/8Þsý×âÂM”ØeëþB=ú¢d¯JJn;~Íž$DÔ7²åñVõZºÄp’w5h‘ÙfôtŒïhGÔ^31ç±È–ew+‰ïóÔäS-sHi”3:a…·›–Œä›c‰5¨‡w&ûñÞ0A“Ë) „$Å5$–™0JQz†ZU°I|¡VŸÄæÓô—õ[ßä“îŽÛeyÈ}!¸ÝóÆ:9öÉ÷{OF¿þ†H¾bA°õ u/GF»dË ù!4wžŸoh¬lÃܱ¦·mà´\<ÿF8ƒšêZrT kO¹d¼û‘7AÛIóèPÙñ¹E ’ûh5LnrZWW÷~Ç ¹Ó±m; ])è,³šdà.¡Ì-+ÒiŸÏaÑeˆÞ8úÒ¯`-o4‘ET§Ô³«â=RJUx¸L0R°†pnYÍ¡­¦C^6”ôœ­ÏdóÙ ƒC÷gYK†S@*¡ÇŒÛ DOÈeïæ[OmU›­[HB$ºšôΦÊt±¸v®Opvb)ØãP~'’,D*¸—€ÈâQ<¯ciþÔã>²Ä“Rݾհ2ª÷¬ö¥„~¯©yÉâµ+Nìù ‹kK¦¾½£Q6d+þX™ÊòP –-hqæ]ü$ä'°¹g'ó~÷ë-ÌŽZ½Y!§“Ñ—ßÈmûüÆ‘ô-7÷×ÒŽêO<Š˜´5]r6”cÝϯ“oœ.Ä{w«ÖÚJí«¶«ÕÞb&Zø¢û3»={Ó—–·ž‘#§‚#–\©{Ztæ]F„G0¡Ð/E»j[}êSe|ê'|C!}y(z´°E|«÷qÚ±ŽÝ¹#!¿‘_M:÷«4¦Ý̓÷K›W–‹³”~ôèñÐÞ٫ٺ+Ê%ݘ¶–¼µö½3®ËáÅZõJIÛr©ÙËcç’ý^ÄBÈüAÊwZ±,@9{ܨg©ËYYVGÄ2r¾Ùµ\ÒcëÃ>‡ðP&Û¢®,‰#x¿?\íìÐ tgB÷ÖÉ)øt÷©nü0†²#–È—¾îÒš^°)rDv.äÚCJõŸ¶8ÅâÏŸ§X¤m·Ö]©¼dQkùízõžL›O ’zÆ¡ÑcQyc«•O° É®*°Õñcœ+ÌQ›ÒT¹EöÉ,{'¢tΞ%ûÆíìÆ7­6ÖN²¢úz´Qq©23-žË¤¤¹m¢&¸àû^¦ãïÍdŸÝÄ·÷öª91pÓO„ÖV¬íå«GzO&ðKs3®X"æÞ˜ìѸ‹°W¹å–bÒ`ÈUs³†‹†ä—]¹¬0¥+2kG|PöÓüìy¸ê{¸Kú&Œ\&ÃÕß}¯?˜lpºÍ_Ör½Þa_¤RÀ:ºÑ&ÍÜüùh²˜di^éÏñ‡v‘©ûÁ&!< …nN<°;<ù3”…9T¦PîÓÝtw­™úë—#ísƇª+¿ë(_æ—«ö’kˆ?¿š=â 8æø$ãÑn4IÌ–¼žM{o—rjaAKõmˆ.`Þ”ÃÙ¢þØŠ2|HøJ¸)Ö“Gïd™nï`Ï¿¥àâA¨¦,uŸ§rw[pTM8×¥ú]V>ã{9QºC×>Fqͧ{úXÙ¼´©ÇÍ_‹?;WÛÅMˆÓš!ffB)©ßÑàhGĪнÆ;Ø<Ú'O÷ ¹Ã èOú}²Lmrúp¦3T'rÒr˜:q“)Œ7Fÿ> endobj 1409 0 obj << /Length1 1754 /Length2 11981 /Length3 0 /Length 13085 /Filter /FlateDecode >> stream xÚôPœ[Ö £Aƒwww î„àÐ@#ÝHãÜ,„ ÁÝÝ]ƒ»[ h‚»^Ι33™ïÿ«î­®z{?KŸµ×Z›–R]‹UÒb”ƒ€¡¬œlBimEN7*-­6jü·•Vèì‚€…þ°všA_d2fÐCðÖÕÀÉ àäâäâàpqpþÛâ,1sYTØo!`  *­4ÄÑÓdm}Éóï#€Á‚À)(ÈÏò·;@Òè ²0TÌ 6@‡—Œfö-ˆõüŸ "6P¨£;»»»;›™ƒ ÄÙZŒ‘à‚Ú4.@g7 %௒ªfÀJcC¥hÛ€\þ¥Ð‚XAÝÍœ€=Èvyqq[/ÙZŠÊ5G ø_ÆÊÿ2`üs9N6Îÿ„ûÇû¯@ ðßÎfG3°'l °ÙjrÊlP( À lù—¡™½ äÅßÌÍ dofþbð7u3€œ¤Àì¥Âês±p9B]Ø\@öÕÈþW˜—k–[JC€`¨ ê_üd@Î@‹—{÷dÿ§¹v`ˆ;ØûßÈ ¶´ú« KWGv0Èɨ(óÍ‹õ¿2k ÀËÁÁ!À%:€6ì%Ðötþ­äüKüRƒ¯·#Ä`õRÐd|ùCõv1s Î®@_ï?ÿ‹P99– (Àh £þ7ú‹hõ/üÒgà=ÇËøq8þúýçdô2a–°½çÍÿn1»’–ކ’6ó?%ÿG)%ñx³róX¹x9œœ\<þ—ƒïÿÆùÏ ü»ú¿¥êf ØýQlþ«ˆ—Ûûw!nÿLÃ?kÃøß ª—yþ;þ†¼/ÎÿÏKð·Ëÿ¿Ùÿ+Êÿëøÿ_Fr®ööëþeðÿ£7sÙ{þcñ2ϮЗÝP¼løÿšêÿµÐ*@K«ÃÿÕ*BÍ^vDlý2笜b%Ó À4ív·,áý'†À‹Öà æ.#ZâÔÂéŠÛg gÛ1ŠšÖ ð}ÉðÒŸ¨N6P$¿y˜Óÿ0fê)lŽ\LÒÎóâæ¥/î.)‘Ç•CAë]@‘›ü\Íö [ÉG¹jh2ÞXÛ%)}:%e8·^]ëÜ{žsÒe8*È9Ä"ÕF¶å¯“ƒd;oo²8­ÒÉmíÁT@£¥ÄŒ<½'?¤ÔÍ%õá?‘eá'ܳ·Ò³7 âijÂT ôXÞeBæ8Ïû{õ†£LäÃ{çP»#2ªÌÐÝ”©Íàµî>³f[ïÖ‡ùBÏZ6%¼‰è’Š&ÞM_¦›F(¢ù,g»`O‚¬¦›¶H‡ÃŒžƒkDà ½ª~žS?Û§Ÿè%?‡éTXÅE¢•øŽg9Å[°„H”&y‡× óuû¯‹G¾Õê¨.î­q»¤I¬¹ÕPÑœV«E£f¥ø¦»¨'~³º‡ÌsVRÊÍWìš [8îÖJÙ¡b¦aàÅðaKÔ¼>Š5]_‚ï1¾@ª"èBW?ÀOèD‹Ì$:_A›ðø Ù+>ç-·œ$í¡Ä'»ÌØ)Šó)Kö¸£ ëóÞ\ÃWù"†÷´ë àq=† <^WÅ7ëtSk–9Š(¹Vz•­Ð£ä[#Äön.ŠK8ó³ý¢ D] ¸ºI-¿µË¢òº×{t‰¦û6ú.ò’T5[´»˜qý£ÉT“6G-"Ïu±ºüü8å6Ô™IÔ$)HÚãã‘YPteiXrëô™fMÍGîìe_ _Óî³Ïq¹Rä‹* Ùm&ÁÎFs5‚Ô°é [߸‰>šÛoÛ½h¯:®,‹&t7Êë‹rÞŠ0DÇóŒ¥Â Û˜ Y“hXÍÝ5C‚ªn|¤j¢Îƒ•a:¥(nMÃùáéi˜(TƒÉ•Åòö¤ú”®!é~Á~:™Ê$f™KnþšSÏÇÇ++®€åLÍÕp ‚‚Ë|nô»½ï»Û›ÒÐêèõÃ…Þ®']Q#t ¿‘±íÍ<ÀòÙ Obï®0Ùù›Ž&aÛ…­·AOÌ7 MtʨÝâFÏx6c\äɃoÏ3ÆÖªöºB§Á h+w ¹xw“¶4lŸnãp¥— †Ä¦n,·±;ßeyus'ŽÊSQ0ܯ{ƉW,ß¹”À„ºÈBËyÈyI‘”Honfœ,ò² tó¾4ú® ôª/+Ò?iÊÚ¬4Àß““¶À‚)r|A-ÑöN ô¹ñb‘p\w—Ÿ;«—$ÕÆí‚¹J¹Ô w¯Ðß· u餀d££À4ßË@Ïî÷Ͷ•ÈÆü“…|±ÚÙU¬~·¼Ò_´·:Ëa½“óYô¼ Irž†jô'aV²s= ´ð÷¿þl):`7Nž¶Ô•BG~o6„ûÖ«¦Q!‘ã£kú6~@Õ„VÃ2$šý)½5#/Þ÷â}!ÃÊ@ÌoïåÛ(“š<‚U³Î=6°é{µfuÖ^Ååë¢ñ/4áFhÙ¥ß4wëe/Ô/jOõlj{aFÛ¼Œä°ð²:¼–köv”Ç|¨¤e+Ý}¥vÒúÚòöË9ʹéÏÝBx4„\”¾¸·]0ƒ¥güÜ:®æ ˜Œš@¡d³–מ†HDïŽÚ”.?«»NÆÇ+ÞøXwâµÆu;æ‹‚¿‹¿’a8EÛëôߟtˆ€]ôÊRå}GÁ!ÎEǘ:g¦áb§ú ƒ‰wnå_ð/‹‚–Q“ö»ùªÀ¢mQfR<…Ã1~jò°¬«4ÅëQ—ÏJ 3Ï2ÊÎê KE5¹ †ðï,Å(ß-Å ¶¨„kT¿Kb×µ¬Ù,‡E^íf~’Ú]ï¸Õ#¢ R<=\!U¼tL—¼,Àµaj*ô°è!Vöìa“œ]™`æVbëØêqÍÉÓýŒÚå3ÛèŠÐ™h[(.Ô»>"\Xx Š×N{>T1—fqU­ü>¸a]¸ñã/#è3ºü0zà\V@§‘4Ïñ¾=uRRÅ(ÃjVG¼ÞÈ< Ôt“²sÖ»;þ½Bû½ñ‘a²S[<Ž((e” ïµ§2[­{ýÈ,4dM¡¬äiÿšZ:†ÕÔZ”˜©Üj•U;¡ZV>Ôÿ>Œ˜ªè*‹—#±F›æMér,ŽûX‹Ap1!ÿι'f.kpüí\y»³ï¯M ÁFèЖ „’&óÓ‚4LÀŽ”íæÈþwE7¦½6vzQ {ÉÍÊWaú.u#”MðõtÕÌ ª½a¯ôÃBÀ†PNÒft½ë´K gþ/,pkX¦ ‘ˆ}/1>Zͽ”Œ‘—ŸMštÓˑ͛¦h. p+ûÈDi xý öÁ±ž™ nÄoãx¼ÃyˆQø­pÓ×Éð·\uSžFlœ·È[0ôÀìhzŽÝ]Õ6¨éÇÖëýý{rÌΉÜsoª‹˜ð P}%WMxeŠLÄ9¬ÃÀá¼>©ØŸ±#êXé½6@ûí§Ž÷eæ‚ß×nU5‹œƒ[i9”ˆ7ŽضZc™ ôjg§H{G²¨ã»þDC ¥BmïOåÄí]hƾ0TÒÁV\öj*ü¶¸mu»Yí²û¦ƒ¾(Eä.OÇõÝ"=57'÷WbÆOMårZ+±²ÂÖŒ Æê2ED Bïá)cÎCØ#CéÝcvÑK™ôp³ͪ óÒÂ# õ§÷‘›dI‰˜öG\DŒµt@Ü;ü æ…Úëvë§œÔp´ÇçàƒvÊjëÚ×ôV=âN5î-Ðù4¥lƒÙ{èõ%â$\&‹5Ì#ãw_X¼¾ÜöyEMtÄ "íäWÐrR&å dÖWÖ„êQ9ºŽ?.Ì49 ×·œ*q%¤ÃÐŒÓ^Cë7Ÿ59òC£F• 8ÇÞ³˜ÖR5÷T³ë×Ç2V7”ë¢>QñEó¨ð[B§ (ï#{ü å±c\ÜÈmýS¾'Ø•¨£Y“n±ŸÖˆ¾šÃ<¿údb¤CžYß•äu@‘€®}ÌB£[4)ˆm;z´l°^wññ‹ìÙ—÷GûzC4xvXÃi¤ü÷bÊ”¼ÅØÕæ‘wn:ŠOfó-Ü£2,ÚA¿ž›´¤ÍM?&o–§ ˜À¡¤€ä“sø&£²x0""Kë»Ú º¤›¸xTrbæzË,ÜVSœ®‹¨½›ùÚx,œÀOõ`a› ­÷ÉTa©"~Úe%†Þn›ÖHG2æX7‡yf:žfPÆ7½õRþòa/(j¥ìƒÙ`þ͸ÁŠJ7–5½j(nØ_Qt‹ÇÙ&9â?°ÂVj­aviÉ’ñÊär9íC,|ª?àaîÚ÷ooø)$& öw¼ÓfŸ[l¤ÓŽÊ[ÍZÖŽe‘Ð v/A{º.š!úvTYì5Ø:]°ñfüÀ 8{ ¿jo!ê5pri¶}«`:kJCðùЧûuWZÀ̰ޛNJwÉä_$vƒCôÚ*"Z“Á1åmvêVD%Mbâ%{çe!#€ŸÚ0\Jìu›»J†¯ÅÖÉbh’g‡"ç~W)k©¾æí–]cŒè§¤ž€ú -Œ&ÌXW%½2Q-ŸÓ€kĽ\"Î*L/@LKX6Ô‘|èˆE´‘0!ø^»Z~‚º©,¬’¹6Æë}‚Sq8þ\êj{>¿PetèrèÚëQ.ȺڴÈWi˜Ì´‹WP,À¸šTT™Ü¼;aÏÅ[Q8¡UG¶ŠH0ò7ÁA§ Ù„ãýÄ,(;1…å݋4Ò8¸#]{«Öô>´f·|úæõÙŒõ­h˜È‰ÄF —2O¿q|ûlÎ8F`;¤ó¤äË«|ÏCËŸÚ aY .ÎUÛÆ}ËÍØúU‹Wó>a³˜úó­)£@µ®ùÀÏí¢âr[”© ü¨šOy0Æ~wOHG­÷@ eþÃcÜÅ–3)€ã2L–ÑDf\ü\ûBϯ{IÚý_<Ðh ì.q}ÙÀÍJ7ȯ”{¦é̱þ˜CMÈ^¹ÀmLŃ"è–«¶{ét½:d:‚ÏOÎ,^{—2û•d—ÓÛ!„Š¥mð5IlkNÖo¢bŸÝíþø—z=Šoô±¦ùÇuüÚÊ.Mº5éMÏàVdK]ú³Ž•ûÇ-Ÿ—Úmñ`¤ùèÌùܿٔybé&|#ïËíEÓ3¾UõUVûnMÕë™0ÅxtPñ]¸5ØŠ3V äáTˆ—V,è`%9Äó2QG Å 7S^<Ý~²Tyàèywþô•½ºCÅìúWÇËÅŽ`OÿnRž¹7£.°ß…úÜêÚ­‘¼½ùb-ýK¡¸7,rkàsO#+“GcaU OxDöUj*FòÙghã«c¶¢á€¢NÊñT{„GF"™·ÿº§ôÅ*º´¹OÔD;aVCKŠpÕV?®]WBbîvú³Ò¡çYÛ—í·f‚u^«z.9Éã–2•ˆÈú w8i†–Ëéñ¶íþŒÜ+´àGz{L;¾¿’ZÁ]…ƒ&žáª"¯#_ÍWlÂZâÝ2ïv3ñd·àãÆù†Ž¡Ú; Œ‡ˆŠ!ô«J §Ï’CÝ +p2‡Mû<ªRÅ®Èg[žPÌ /#ÇÒ¦Á‚Dª¼ØÏlâÝ#ñ“3ì+ö+ïíWéÒvñˆq"4ý>ê¹KQæ\Ä_߃ú˜yÝß)¸á®T· ëP |¡¸ÚˆV%"ìZAÄi:á9K _eZ}Õ{)êmþK‡ï#„Y™vtUô±òA×Rã6„íê[âäRØYÈ¢ñ£R®B\³]ÞÁ, ±ãf´Aü3CÜÝóáÐ<ÈcÂJ‡@ùÓG=lë\ ÌPV¾ràòþbS3³â›šþ$è ¥ž‘{ÿQƒƒ¤êHk€òæÅVü£Áì»IåeµÓüÜkÏ·¥ Œ¤ê°ð\04"ú­6ôÒ(jä•DiGÏ5o$Áï_M† uÆ}äçÚ%Yã‚wÀNä‡Éˆá%¨6R|Ç5ƶMûK¤ÑÓbæâÕŠ¥ñj!<ôµ‹jo O®s¯jžvcœ“g¦eà#„BÈíšž¸ÄõSŽXî4òßí¢(6Ýï×HÉî#¾m§sn50Gyó½RM«_xÆÏINÐbú&¿ÄÂ"™¨y.¥ÝRy ל£ÿópvÔ×5›>ƒù“jÍ ¼·1¦ùb4¹‹D[ÞÂÒô@{E }6KÚB_ö\ ÝlY‡äs0B ›¦üÖó¨d(°º8f /wŸ+EÌUI8¹ ©hPþZ ÞãSí=å-~vìúk4•QÉË”áŠë…H_ óç-î×íÙì\ËÃ]q W®@WEŸ×¢Ò8ñ5û. ã„}ŸûËH?Æ5îQXóPLVóسwõékÂ3Oø(Z™{gÙŒ¡Ëf…ñ½³¤ú‘WHg²´JÍÂêQ1©æÌ4 %['V¬1™ÖŠ?,BJÏS”F“nÉ8;gTž••YÂê^™:&ÖÒƒCÕÌ#PŠ ¿"i‹²/Þ_ܸ Nö]öuRD]ÕCKÄ¥ókÔ­ä ®¾£¬°™ýþÌ M¹°+cÞu‰L:’úJ¶ÈƒxÑ–öwãä»ß}ÝÇJzñœg©1kŸw³ØßÙA%Ú3ÙýyÏnÝEz40h¬Zðþ>O)¾\è0®:&G, {{ʈ¨i‘ºÎ…K…Ž¥Rˆêë¸eZJ³Iºž1Æîgd¼{—ªˆ>OÂlgøc‹a(Lȵ[ô.©ÊàKŸ¬ÍÁ(ö]&O¶‚ë“«ÃáÆ+ p$ù• TlH>;È^Ë;êgÌd^meš·o»’°CMè:ÐZ«Ë ·“3ÓÅÒžÛ Sž¬ŠÝIÌÄ|LŸH¿Ž ɹìÊ_×Ë,ͤ{o{Md¾iqü8K"tâÏK0 ±0&vŒn£X‰ºúQ‡o«Š~ijþÑì¡ß62è”[o§S“á^ìâI6Ã<Ä cÈÑÛ´¥NÛ ¶oÕÿêƒ=BRHÆx4H ÿH¼æ`@ªcJ$+éÁ®åîî5žJJù²ˆ @Ü%eì”+Zب1I‘ÓhÒdÐoó f+\mù¨¸áØ'šŸY<1¥nC[º$ë7q| A)§@EVW9 e|’ŠÖ”±/b¹æËÝé÷…É¡™ÇuûŠãž83qà¿`t:ê5aÇ<ˆŽ·”B¼"¦uƨ¯½o_/„:ÿ&Þê6Ï~¥²0%œÂêï*ã9Ê—LÄèöFòëïLYÆ}RøÚ¯š°س2W}p”ß@Þ˜g¹«_ç%öÆ´„:qs:"5´üîXcÑΡò̆Þuh°°ÎÏÁoóÆKf,%N·ÐR¤0l²P«ðBD²:Hµ«?òÆuUŒ‹Ê%1+~̤µH=VB žw¦{Së>ò¤ö¶ÿîûe(öÔãjF1N–¥.ûL«Åœµ5SÄoªqé¬ý,Ê=Z²~{@\…•øú¹–¸ù¡øU ÀÂÏ:4µEêŒCz@âàiÛut»@[W¢~”eÉÆÐÛML°µ³ð«NŠo™a·¶ºò¬€-;q×ñtí±§wâºÉǸ `ŒØÂY|dµ«GˆfS/·}iŠXì4…N#ZYp‘z> (ª|«ï{÷j 7=æ×ÃøÂOvà䇸Ìl“¤÷Ò eë>ëXõ¿«ò -ž´óE’°JXzšãk²°J÷Ø-÷óbo(·Õñ‡êM5¢ÕÌdÔGöq…½¬0‰ê.‹x—Aidb@ªø„8Xz1ˆ‹(ÆÝ¢€ #Wþ÷É€O}JE1\CxÈO‚ŸôÖÌÞcq´Q¤nhÞ”=¢ûÉþi—-(ÎOŠÊgGXä£Ä„¹ì_dÙ 2½²av‰ëó2n²—’XȆ>ô.ïå0"Ø£AþìðÁ—Ç艢ˣ0s ¨(nŸ¤~6iɦ½åÓ†Þè'ø±=çPÐî‹d{ú;ǧJÓÝ&FN±âÆ&„/ð‹ %)ÅúÖ[Ú?ÐbL[ïã|jÞÝFzžšæ|x~7˜Å–}ì6Àõ îþ¹Íy×E47ÙéÎáèøÃVFb…^9*@}±èØ9{û^?c¤¢pn¹àLÕsÐÄ¢(¹ÿ|~ß‚{J¬Õ¼(,€­X+1ë›Õo .{î#ö0[,™ñZLáEL÷…ò“ —v.ÉBŒûDHEe’«?[‹·Ž›ˆÈVI®{wJF7wr%\(Ô‘VòÌÒo.ÑÇËS;æþ WЧ}ÜSÚúÒFvô^©Æk?|å5mŸu~"ã(¹ÿÈÉ7¬Ü%ý€Ë̃5Ë$ Í9ç“ó,ŒaM´—…Bþô‰€À³rÿ÷ö±š‘ä_Ø4±¥|P'UÉËÇ®/"Ú»„¯PôÕÑw.3Í{š‚¤k¬Lï9¹÷¶ÆN¦Dæ¾QwSÎÃö;ómDûn® ]ˆµßµŸ įdiT.lVIÒÑóërÊ'£äËÛw`Ì1Ár#7êªÛ³½7D"[ØðI÷a4 ÞO ©ûäZ©@íÎ}ïØƒ¥–ü߻ٯ¿f ÔõžÅ(gÜ&Û®åprôÀI9;-!ë—dÇÑvŸ$©@Æ _EäVðø3Í –w–ä>zdœIJ!•7a9•"Õ`û¡CVËo¡+ŽëÏ7ch%'æ*°vJ´ˆçL:Ñn¹µ™Q-6‹ k‘çk"÷íEfo}ºð™OÕW³ýzpÞÌÎ]š˜W J_ëkï ŠLà ՜»%´o»3“´f‚<-ŒrZ`O8ÖÒ²ÍOoŒ&2Ús³óˆôÎád Ê|&ŸÀÀÎÛû¡ (-à ‹M/ä¹½M®vF„>ƒ‰_v€±OÔÕØüŽo7JФ¯Æ¸†BL7;>Å›Bþ ÖDp оүö~ëD‹om­ñt©ÇY^˜aHdYÛøõðЋqÜ3ª=)B›UWuy¦{C×Ö\KøÙߪ¥s[w ¬"À%YB£XH.ü’}^Îc‘_–£³£x­È&…OÞ—à=Ù:;ŽÕn ²µùx3PÊýfÆÐ÷u®qƒyêå–Õøž–â]©wᮕ¿ÂuÔ}¿!dßcþ=FÓ§É#œµÍâÜT†„§a[,–[‘WÔ^Çw¼Àþ<ófJ¤ã—µZ´<¯ãé‹á°|¸]È×ïL—LÑi).ª*š=@DÄ‘×Iåô‘ …I|´Ï²Øýê¹$ÛŸ¨}à+ %xÃy¢2ä'z™†¹ 5ú +¾ oÃ3ħ¹â "ƒÑ¨ËÔ™–Py³(ÎckÙ0¨KI•? WÍõ$1 0“tÔ›¦f8hòÕø ¯**¾·”U°µy"=ŽïîwÀ­(EMwý|Ú-¢<Æ"jMe®ï]ÔÛʤ̲fB £R¹\ š¢÷±Å¬RâògwMZŠè¨ø­N~Þâ¹±Rá©þ{€kQw®·àØ´‘JÇmÖçíhf¶ÒØ™ÖÔhæ|0¾×í…Bðsn[ñ‡èWé¼×S-¯Ãê𱈟l)‰eu$BžiÏÈb„3úóR‘ºRnUE/n|ó¯×†‘YVKÆ·ÖRÂbÙê‚N Ì´1 •uw¸Ö]‡*ú9ópo%gܽZ=õü‚YÐh19o Ù·ìíèMÊz*uQ×€è½Cé´íÝÙæO¤‚?Væ{Zë™ÕòÞÅ8î_-xm\ñÇ"*ܽúh.Gù©ƒEî4jÖRŽfk±\iýdËkFûKý‘¸%¼EReøç *çëçrÇ/¢Þ ¢°vÂPyöÖÆ¯:£åÇA®:Y†»• Q¿ŒZÙ`~áf Eä(“¼·K÷Ô'ïilnqáŽdPÛ,æ7xIÌÇ ]˜ìáÖØ,jSÀ/béõV§Oc»Ú³ÄS›Óïþ-Sf¢Âã1§Î âÇeïý½Õ÷ŽÉË6½e“J;úïJ`Œ½€Õ#Ø€œVÉ•)èûèG†Œ›d%d[gíûÔ>‘fÖNâdq°z²™ñ¶ZÃtWZÑ©EäN8Q8ÿ}%U%ÌTGËqnÂ…Ñ@Ø•Ìâ0­,´¨GD-fÒ*ô¦ëáW‘Á[À&B±M‘ŸícŠ2Àpχªt›mï+—#äyÔÕ¯yŠÑT)¸m ÀO9k­‚Ë%Û gÊèÕ° iÔ›ƒáýàÝïs¥E’æ¨!Φ ƒæûqí†qz E“‡VÐÆ-Ñ¡ÑBßlékkjÒƒŒ¼Ðq, Âü0GƒØ´‡Ÿ0dŠ7¼ÈÍ#Ú<ûtÁÝÈNÔð¹Ê´‰2‡%Ы–ŒKÌ=¸Uýð\&íãJÐÒ¾^q.‡ a‡à`i8(sˆ[œðÖä+×Ðgšxdפp¤ËtžnÎ%øY gn±’fÉS~%¸R¡ôá:‹Æ –è —Y‰ÓDø{„~]ôC|D§ñ¸QÃÔÌM9é¹Í¥j¶?(|Ï®±aññWf—cï׳IÄFRâh¾²cq¿ºæ,¯9gËü`ßÕUý…¬a@vꪈé¶Ë6àY0A Ló¶Õ#ŒëØÄŸ[˶ý²ZK}¡*³„$>‚'àÿëlwÅMu…{»/øÀ*¢8ıþm]Í乃ëþ–vzÄÇ\gãñƒVæԈr–òåE—›2Ü"£€drÁ<8 ÈÜTîñu†0Ný¶à/D¹Aª¤ÍcÓ-¥ã¾ªš{xP’›à¶½jì·Wõ|vINÅÌæÎç¾ 1;zh®UíV‡´yidQÎîzöQ¾Ócêµ7å¸DÚ’R1ñu—E›¿HOb»ì˜u>~kÉú|ûõÚܸ'JäcöÞ-•X5RƒŽ’U©wA#ˆPtô:ËZ6‹Ø¶rGÁvXCת4ìô‡¢Šr` ¶W•×€£É;'+®ï ’*?©­Å´ß&"O m¤ûÅ’šJÀØ{mÍw©0ÿæ¬SÖ‡~Ýáûh§$~V^!qùbÆŠ´qqâ(-©Á»ÀŸ©3i`H›¹B06ßûÖ$Eõ7¬^yFtFígTL4FY¿,ì[ûÉ^‹C\pQ¾Z׺—רª=UËU_ïqð^¼»é+%©#:>Œb•Ðû׸âG]©Ø.To$F e:M£ãºb;¤×qsÞêpúoáÖÖ •  G Ôza"þâOJªÓ¾'´¥kdžæ“ŧTâÐ%WÛ$ùœŠ:"˜?RP:÷û[ô„^¥á3ßX™=ûõi“Õ@5£P~z¤ÝM¸Ö—W©ˆÄ’êo [„v“70L"•ï#ËŒÄÓý øÍV±§öç+´tÄ®ÏûoL N?ª,6<úˆ¡-á«Ö©vÙÕ½OÓÅS„w7éSvœÛ]¡xMÚ8ÅÆ çå_Ÿ`^‘b(Í¿Æøª®5²^ãdôe?*R÷öËÂŽÉ6ÿ“/,¥vZ±†ØæÕ5G ‡©öTö¶üò”íï¤ïÙÒ7íYnTè›VPDµan•¬¶¼ îÃ]°¯¾}‚Â)G•MÄÏ߯{µ¬ìL_ŠÛ÷äÈ[Õ³W{Ù]®fhêÞ¸˜Hmo¸ôúáÆ“šÒ’ɧ{Ä|¿×™£Õýt–Ç‚ye¦tªg-0j Âëú–·ö´úÒ æfxðÑ„ƒ®x‚ ù39¬RÛ½sðdàð[÷û-]‘ I?†1i2òÄ tÕ1¨ùϺ“‘ÔC¨¨÷=Ò¶—›·Ú¼æky¾Ÿ èúTº7¤D¦¤R|q—Ø<½t1¤ ¦HdlaW·¤‡džÑ ŠE©”.I/kÛ<'å­Ð˜L^²Et[©FœX÷ÍÀ&Ƥò“TFÌZ‹¢¢u‚•»‡rÚÖ;8“# ìmç6°»v ´ÞK¸²ïe8Üfæ™Þ¡ÑPh»Ò¬‰Ì`}Å­ PÍ«Vßñâ„u*O`E­î/&_xÅ$©è9;n¤õÁ1¼t@"ªíàX,/†T,l#³=áìD zÏL=SÈJ¦%dÏeÒ Eåb¯× vO»bK›žûÁqß|"=Ñm“'¾>ëSºc+Snh çÍšêº~óÜzJ'”±pz_¦\Ú^½¹êß‚qçv·„N­Û´S%#Z½þáYa¿ÎmÊ Úw¼«è„¬@/\¯p;ë{Ëe±]ÅOÏ¢ú1r°–¯J®bë :š<Å*Ë“’9¤èy’dë_žŠhhjIq1¸å·l jºÈ¸ïX|S×6yÓkó](ÒÖÉúhÕÕ*”vó 5D–$J*îÏgµ$(Ï­<œoк4™}DÐ#~@Ñ0'R]]+ \¢êÚ2†zìñ™”ýæ³›uÿrâä}þlPÆž?•S’«g¤ù@|i/n*E"BôÓÌGrëö˜ßæ e×'ŠÖÉà´{âå[`õÀt4y ûÙ“T6âM4]ˆ Å‚|ôøtšà öª›°€ÜnZÄ.Çýk~ãqW›éfåâ)†Š7‰•Îz– ;³oû#[°«½Åt`“ƒ Ifåâƒö -Ó7R8Z,g¹ê%x TvÕØqpøÍ‘Qû‘'B½ßšfõt¨*a^à»ãwÃóñ4T|Òä_õ,¨»×IÑ Q2yHֈч†E„ZGÊ;ž•íƒU3PîÞRƒ²úˆàå}>‰ÚmRÙ6¹žéYžBÇGNrI§Z¨˜jŒ»O‹+¶ãÉ@Ù‡pŸb§rê;]ì×ÙÕgØ»÷ä«ùí~Oci|íâðëT¤ÖL%[(ˆ#¨œ!ÿ®þ@φEž çÒÓƒ†é f~TŠ@C½Ž?6AàŽ‹‰•©¦¼9òøãÖœ‹ÓN³àx‘듇ùÔG^7Ag’°ZŸÏ Y„l¸^_Scc²¥èxaj’ø,:Ý+*ˆ7ñÛ}xøòbµüüYæe:ôv´'­mÔ`‡õ3«¼MíÐj»Ç5rÊuöïê# (WçpZ‡\.¿Ø"Šê×β Iž©ùñ\§à_Ý3¤K…dð’¬sCnn’ßÄ'M¯ÇG›1ñXºÕˆŽöÂjÚÕ&¾^EŒÐ_-ÜêÙú}‰ÃaÛoh­¹ÎnbUÏ@d­i¯ ú=÷½ …/[#QwýÕˆÂTU{¹²dшóV%ïÃØðÆÞa,^V)JsuìÄõd,¾©{³vÏi'É2ÿ݆˜õxV~~Õ^Å|µŠfì¥ItzT*Í]¸RŠ]òø‰ðŽlrüZéñÚ*È»™Âƒ¾/BýÀÄÄÞØZƒâ‡Ã>êAͬ¾²‹÷‰šöìå’Ö{6Ö U± "‚˽"]±ãO½BàÀ¹BX.K£÷¬ÄÉ %i…̘¬~’¢ƒ67¯¼ó!ùÞFC-BðÝ’èŽs PnNëéP|'k]?÷½rÊí`8QýeƒS-ˆ8ýPvÒÕ õWHľ¾<ÎÖyÏý0Æuú*I0^`Q"ŸnÞñ•Qy<›6ÌM̬´kZR[J¦Èé°O_†Ÿ: ’±j¡šÇbêämlIòcãSœ½á ½ÇY¥'M©!‘–Xêorî˜ù·Ã>þMõ®îFÃÆuÏŠ±6º€« CéQ^BµEfÈ%—äTnb°«æQIšÔ©‘ƒ÷°§Jx/ „e.ÀгOraé´À.^²}\¢†ù@Ø!‹…³õØžfû o=àõwÄ;]µ¦ ØÌx°T¤óP—Â× —a <*Ÿ°ÜŽ®®2M‚Uâµ~óˆrŸ‰(Ýà $ˆn~.'‘¥ïÏ»Ôùàs쌬Sdao‚‡"úÈÇ÷ib®ej›skàÇ1ôvÂbÁÙ†€|»/¶u÷ííO.²†ÓzËÏŠ08n>í,+ß=H†DôÐ-c®¼B á ·¹×‰¬¿þèåa´+Ò8ºïãAô\þYN’Lj@àÙ\âFJzZ aÏ š”̤ŸéCkÏä¢xŸ³P›Kƒ†½ÝÑ]30éF^°ó…SV@|T ñYLÆ`IXl|sx˜žøþÇr„q¼…½‚=哸áîTn6¸%XECôGM ¾"kÒDÎ6Ù˜ÕeµÕº$”’[ËõúJŸu»hÍÉm¡q5\Å)‡©Qp³~SöU˜¹’æ•8Ep™Òæ—¥Øyß?޲1£´[kU9U ö¦›ò¿§ý~È7îÌÉQÙR·)•Ò€Pgæ¡«bˆéÙ…=œÌÓ>¾ä$ïw¶KÇF/ogß± 2}Ò|eïûÿQ°—Ö endstream endobj 1410 0 obj << /Type /FontDescriptor /FontName /KSUQKT+CMTI10 /Flags 4 /FontBBox [-35 -250 1124 750] /Ascent 694 /CapHeight 683 /Descent -194 /ItalicAngle -14 /StemV 68 /XHeight 431 /CharSet (/B/T/a/b/c/d/e/f/g/h/i/k/l/m/n/o/p/r/s/t/u/v/w/x/y) /FontFile 1409 0 R >> endobj 1411 0 obj << /Length1 2895 /Length2 20602 /Length3 0 /Length 22235 /Filter /FlateDecode >> stream xÚŒ÷TÕkó SRÒ’R›é.éné† l:6 JƒtwJƒtKw—tw7J#¼ÛsÎsôü¿o­÷]¬%\3sÏÌ5qß?©È”UEL팒v¶`FV&>€˜‚š+ €……‰…… ™ŠJ ¶þOŽL¥ttÙÙòýa!æ4CdâF`ˆ¡‚-@ÖÙÀÊ`åâcåæca°±°ðþÏÐΑ nä2(0dílNÈTbvöîŽ s 0$ÎÿþКÐXyy¹þ:±:‚LŒl F`   $¢‰‘5@ÕλÿÇ­€lÏÇÌìêêÊddãÄdçh.HÇp-*@' £ Ðð‹2@ÑÈø5&d*€šÈéo…ªØÕÈ€¬A&@['Èg[S # *#P²Úþm,ÿ·àŸâX™Xÿu÷Ïé_Ž@¶621±³±7²uÙšÌ@Ö@€’¤<Ø Ì0²5ýehdíd9oäb²62†ü•º@RäÀÂð~N&Ž {°“ÈúGæ_n e–°5³³±Ú‚å'rš@êîÎüOs­lí\m=ÿ‡Ì@¶¦f¿h˜:Û3¿·98eÄÿ±ˆËÌ`' ;躙X0ÿ  ænüKÉúK áàíiog0ƒÐzƒÌ€_ÈžNF.@ØÑèíù§â¿™•` 2Œæ [äßÞ!b ÙßÒG@‡2~¬–_?ÿþ¥™0S;[k÷ßæµ˜YNTI\QòÍ?”ÿUŠŠÚ¹<9ŒlììNvn/Àû¿^þåÿ?îI•@ÿäö‡?[3;ïß µû —æ‚öŸ¥¡ü7‚¢dšÚßïËÂÉbù‡õÿó üuäÿßäÿòòÿ:üÿ7#Igkë¿ô´üÿèl@ÖîÿX@¦Ù Ù ;È~Øþ_S àßë¬49Ûü_­ ز!"¶æÖÿä$ rš*ƒÀ&MÌÿÚñn ²*Û9~Ý7FV–ÿ£ƒ¬œ‰äNq‚4ë/²Qÿ(akbgúkõØ8¹FŽŽFîÈ,ùbãäx²BvÔèö×h˜™líÀ#;o€™#ò¯–rq˜E~‰þF\fÑ߈À,öñ˜Å#^³Ä¿ˆ›À,ù±˜¥~#6³ôoÄ`–ù8̲¿$¹ß’‹üoÉEá7‚䢸ArQúñ@rQþ ÑU~#HtÕß]í7‚DÿA¢«ÿFè¿$ºæo‰®õ/â…Xý‹Ø!–F6ö}ùuþk‰käd™€Mœmþ•³²qý£ƒ¬MÿÊ9Ø~‰!# r²úíâÚø£ŸÆFŽ TL€Ö@3ðbÎÄ/æ¿ÞXÿ[Áÿ±çeÿWþ@X›ü‹8!ɘØYC†øßœ9~Ill~×â×t3ÿ.;„«©µõŸ9C®=æß´!óÌ üOP®_zgÈ-ñ¯H~é¶6ú£Š,r™ýö±0¹üáö—ÚÎùϰóßA zó_O;ðO‹ßä E´p··Úþa‘þ€ä-ÿ€±°úBêõ›¤0Ö¿–ü·RÝ?Aîxæß¡8!¾l!—Ãï"@BÛ:Ûÿº–ÍÿH òú0ÛýNâÓîS¬¬¢ö¿ÕöOÛÿôŸƒõé»yY˜!S ùøÃ”ë/Èîw9 …µ·vþƒä3ŠÙá7ñ_Èèô×õ÷¯oŽ_B;0ÐÔøw‰ ÏØßÂÿdÇÎûô¿Ù±²B\üÑœ“‚Äüë#ÌÉÄÎñÏÆ@¦ÃåIØõ †8uûB¢ºÿ!Mõø3Ä“Ðñï þón˜8;Bº þëi‡<*ÿÃ}¿n@ä…Y;þËꀖÛJBWƱ·ST;ÉtŒž Ž­Î÷h té~kŽ×" ƒ]Ë[´W‹¤?=šj‚¿Æ½k~ðz4ˆQ™ÜiFžŸÀíÏ;©é%F"bTÞõúéà¥îkÛÝ.K•íà̃¦üùÕ­k”[MoÉÒHÐìÎ»Ý .¹—%ß#Þ‡ëúNSågÌà“Éé±ÎÝЧ¯®§°²ÆŸIecÞ {G°ç{j¯³EÞÍx¬|Qcsê  $ÐÆ'†½Â™¤öÝO”Å›ó,*_{;—ÉBǵ6ÒKê5ÜNv§#/ƒ…Çbí¸À+5BéÝ\Ãì«Ù‚”¾£jhO_m„6‡ÿÆÌÕURê³RcAþî@0ÒMˆãS—(EwÏݸ+­£+¹£“êq!³M$äMœåÑñŸÁÞØ ðD?vy:¼ò,ÓŠ)«<»-Çí§ÜvÝT[ûJ%NÑòRÐ1^å¥ÖsS Æ\U½âÐEªàm6†BÒí¤Ó$X”|¿Ääp±ûÄq’3¢Ì#÷àáêJTçn?^ ÃiŒÕ@jis¢ó-%”!Ž;+ov°D®Šï–BXvÉ´\x²YòzT!xï¦Ó $çÁ“ÂüMõýÑÛ†}Á£P¸iíŒ`£ ID¡ŠWءҜZÌ‹[ÚÕ}yûy9Á½qи„£ù:ÃcÔÙ2/»8ögn®Q÷áÑñ-ì­ÌBuÊÒ©pE‡.+$»„ÌURäŠkô>F ¨N)û5v<£G}É{U²©3BÔ· WIñ!yD{ªS{Þ}Èm "¼:¸ªßWƒZ\'fê*«nçë ?9/Qs ²÷w[ÌZ{÷¤ÑÕ¬%“ÁϱˆìÅaiÒºÛÝýÃÛ>רÁ)…†)z7›ÆÜ]­ò$ìÁãeT&ÇW_¢íÍŸÿ´‘ìy½rVJœQ›ÔN¸8“2È-)÷A7(Ÿ[W[dYáAI-”‰?r%:\zè (H³*“ß»:ö6{ײWKóJbι”´%—³¿Ù~)ZT¦¥`#'T€ûXœÍmÿ®5‚ƒÅZ:ez†ãÜYçÕU«µ;15s'ƒ­E<ñmp(q17¼6]ÈâGJh£qúÆý1 ¥²i_¾a(%@9%0íŸ/¹_öE1]d±×ûɳ¾â£p3<ø»ÒWâuyà[ŒiÓ¤®Âì·_ÒÒ9G¿·||CXŸwjÉÝIu~NÂÝ|5€IG=2¬Æ 5áòúýò„äAÕMtkx¬ݹՒú*B‰Œ —]¸úédá0ô¬æ®7{l9ÌÙYìíÂ¥›º•ÜKF‡“¬DtŸ˜ÖSοû†I¸ùiÑ|(¹9ä ‘4ùøÖÖ²m³u~Ö{?Ú•%âñvdŸÝÇ8KF/8=´Ó½±š/rdÕl6§xù0,nçð¢à¡ÈÌü˜ªÝ>çAtAªDBKãY5S=ŒæZú¤º« L£á¾ž$ܬaëØõ&¹‘,ó)²›ŠøsòJο3éT» þåÑÂ3MBtY´’ç¤dò§n£4߇Ûq~Εô®fÏÒþÕGI\&‹äî7Z„Y—Ö~F%E®# ŠåïíG ËÊ~l¦*ðVÁo`Ëš3v%ÛLBiqêñRûU&Þkì7ȧá@¦÷Ô—ãÚgø¼ŸW ‡N³Óe¸n¡ß…¢¦%Sj+ý h½]ÿ³u4ö2ò¦;ÛkåzvË7Ùçzn@SDb²†Fû®T‹µi÷ºôSöûöÉ/EÌÇJ÷/Ô© ©‰<hñåzù˜ÄŽA£r¼ŸÛݤ{_¹lJþ؆»d^OËŒèËGM÷Þ¿Œ†C cËY8ÄÌ=…гÀf;œú|­ò©Ü PÄéÔËÞG:µÎÞãë®×a÷Š8çtdýÞdß7ÿô0σ°– ‘Lä .ßÛÚé¹|bX7`êOè5€*%ûìÁܾ/MÓ¦Mði*e^a,'"o=2ªOc$$:Q€{ͳBÞ~³ûFŽêÖÆ¾ÕÐ2»ÀqŠš]Þ£˜­7‡`ƒö³ã@¥g³#^ç{íg^‰’¬Ùi©ò çÞEÁuC;‡¶Qf¿¼hn›‰Qª^éf¼¤s_„:4/˜œDÝ̤Œ%™\5øSì:ðòEå­¬äê<xÍÍÍ ­¼J©[}¾á£) ÑçNS^"nw¡X®<æ~lrðq´»¡Ц;ضQ}"eŒ5ß™=Ö[tô†Çg|&Î#goµM>H¨Æç&ï‘D.Œ¿üê>Gã™jD3>à Žr>˜Ïx×~«VRß4UtŒdVV…šK‡Qþ„ƒ,=bÒ] ÊF¡î'-ìË2ÄŠ“!…W¥Ê"}f1>jäfDáäpBÀ›¤'ÍBêX•¶‰þ£—zElò>‡A=P\p(Bt<Ô#8Ì¢Ffþ׉qdx_YBêÎÁ ½a°š,Э¬k”\”8 Þ¾k¬–; Mn®½Ÿ”mU@èÊÚÀ^±üƒËXŽšó»}»&¾•Ë·ÞÄuµÃE´[ªCã,Lq푬«t/MpÃü)¾ÉÀò7û(ëáaµÖ]4áª@·?%<î'|•eˆ'Jˈó½Þ2ÍxœãñZ^’Ï\]9Óåÿ*‚lÛ·o²VáFª9ÍJ&Ä·Ž³Ù”¸ƒ ¦ é^Ç¢jš¼I0Umêx¤fG¡½CktI÷âÇ(°–Œ÷ˆ=ÇJ\.WÎyÑÂÆþHa½H ·Z“¾CÇ%,jÇŽ="T4™ãCx¬(‚¨Å&ALD_TVÕCUÓ? ðw´(O éòºÊ®&{ðçø^„âž¶'`ê½¼UÁÒ›”)W³r;O³47ΔŽ3Ed¢åCîÜ<+M'b Ÿ€I¢Ê³wh/ù!ˆûñ‹ þ‡â§~쨉žcÚ&—]yCP]ÓUÎÀê°tœÚ²ØÏñ9@‚ÿ„æ¯Ï%¶{cƒhˆ¼#ªç[³Žú/”4ÉÛ^oú+ª;ú04¬û’{°©ÓgÜ41hE“úæÔÂ?N©£ç»3€C¶0âiÓR",èÐ-=c'šö¾q# ‘Ú’Aü7È?a 8×ÞÊ’¨]Þ—ÅI‡à@óim4àÚ6¾š÷|¬bôEo¹½GÌ?óV~<Ž÷à|.”u«Ià™˜R!ZÚ`ÂÓõ2t^yÖÁ !$­+üAuÐÄÞõÈÝ¥;ãhë¤(KZwø6~™ò3[²ÌêKõËš—mJ–íI­¯Ë×wßÔ"h_ÏL8+̨'…Ñ•Ü&'~ï\G(‘½ÐR¿Nã©=î`íü oJ!+ÏŠ+ôJ¡v@èyošqáÈ»N4¢³¶&3“†‰©,ÇÚÖŠ—æÖ"Î’b8µ*äÆßúªî\è™" ‡äù pÎH;­û=\-ãô2_n~ ±R'ZqÞ5ßþÿ3'löÛ†Ã:a"~Gø®¢Ý@If­‰› mœÙ3ZÎÃÑÔqß ÕŒ0§u„ô‹[vmröâ6`nªã¦V5£x‹&ë5¡‡'AÓ‘q]þ”‰•²1â‡>_¨yd–œª®îÛ‚QÂ-™©2²T5ÿÆ0•Ê)†|¿¢cæì}Q¨¼ãM_j‰dxžÎš”þG$×÷7éÇIó‹= éë ‚`-õ&z༞֛‡ 1yô ¾vx,Xu÷}¿+6!”¤•D˜%®÷uxÍÆ9#hŒëÁ[é€@‚I͸Ns¦éC*¦{¨úçÚS‚løZB•ò(bܼꉺÃó 44Ôb» Ù¤HNêvœ&/ mÚ»€Ã iy¤AáÝà1nS¥îÜÝ¡° £˜K‚Šå2¼L>'<*zrx’d÷ƒ¹MK^EÝÉ}œ3+茺q¡wÛò$ÆÓ8 æJ¶4<8ƒ%eådx¯%¦ƒhø‰ŸYqÁ[AéCi¾i÷¬÷ˆŠ¦¾\sß‘çŒû¨}N$Gïj?f™†7c¡ÉzSÃOsU4ñ¢V¿À]`œöìûÇûNN©Ž—@ ƒh£1ý—„¤H¦Io¬KÕÈ=I@\•.ŽêJK¸Ÿƒ¡£Íà÷²üJ,ü×Ì™SÜšs*'Oú¦Þ³ ]š‹b¿ó-Ŧ°GöÖ’AÓñ»ë ah›ºI>SQ!EP¶ Üh¨AïÄåE£®†€Ç”Bb•ûR­G¢®Õ<éuYÔ^Ž’P ˆ.îd+f£T ½`‰O²'x>y’?¯×¬­B]\½7R„ß4òÅrlᘋ›&¸3m+IêŒßs©àÄ®9üšÎ Äé7SyÓ`òVóìTº¤ÃCü”:=ac擺Ä/QV!Ø£—.íU¦×UYÜ ¸hGh/<¿&®-ÜÙ«ìhîC/±ÞC»ª&¼KÙ Ü]¨r¡áÞ…îžóN;ÛÉØ¡}EµÔÌ1’áà᫼qÕrÔŒ’ÇBüB²JÇaÌpҦͳ š§ŸÎ5 Þ€ãYsØèDfÑ<׊n϶Lüç*³cJ‹<”éß '©Î©ì˜Ò“îY+žàÍG¿² .Ɔqmu6è-„Å'bYÃgÏ7h·é~ ˼¿öñÝÅg³›Ÿ‘/hdá>7o ¾r°,¹H¡CŠr›«ÓxëZ÷ZâÃqä·ñ”ØBf Ü$I‹©kÞ˜M±šÃÛvû!{ìé4›oÔ^ÜW²Üny%Ûbgá‡â ØÎÈ#gÉñ -D (d^E0&ql ¯_³3Èd×it[ÖÇÙDÛ«¾û¥Ø€Î$ (txîÿòù ÔCõz[ÿkv/‘É -Žk „yU"é¡dÙÂÌºÏØTé…Æ:í˜JBݽü8Y*®ÁkMýeœ¹Bä™;^>ü3å2\§0ñš04º¦¡ûrw?œ¼Ù <߉й¾–=#h®Gâ€ØÃU°ØaÔÒ8cŒ½H>ÝÇ °=tp‹xúf¥¼Î—לVâÚ{`Æ 1ä#Ÿ¾h…©m¯)"ù sÁc.My%Åð<¢`èî-uXŽ7$Dg»übõ{bê3÷†ÜK2Ñ¡ÖHkèA¦e»+’qܺçd¹†U¡Qmâ–KýL; ˜Ù³xTÍS# ªþ‡þõŸ7Hdhmë6RM.îæ,¼‚…µä$¬'ºeÀ±/2õ(ªN˜ƒ¢2j á3Ó.—X ÀýºR$•f+kJ¯2ÐÍ1‹ƒýy+ò;7¨IG»-éûw!5).{i½þçÀ… Ž(âþ`±ñMòW¡Ruš #·s”.úQ\¤þèæa ;bb?>*¿bùFî˜Jè”6¦*V¾ ó¶ë›ûʃÎôô¾c+e%§|üª&¦ £àîL¼Ì)k°)E:–,ï)uŠLl E`¼×`ÄåÙY1¹±Ó:ÈA¡çµ¶íqwQûnW8pb„6¿Ê¼oÞî¡§ð–òXå®á9«îiùËý‹Š,ØMY>Ëö?36dÅØ:>v1±h~%Œ!r¸5"B¼oÛ§ƒb§ð<±ïC믩½åwm3 >âçs,e©ëŒ¬¿aúlh|XD ¾.f=dª÷œíŸê4™|¬ðÜ.Hs9¾Yn\Ãñ˜ý,pÿrê"q`"wŠwÄ~¿Î—<ÝØ\Ÿ2”Lé~0Q2›£~ ÐÓHÄïšý6-˜»²Å›‰yõ‘T9O4¶AƒžX[l°/-T)óPÕ×鈥æÀgÆ( ï«ßØlÇWøÕ€*α„FoîO~ˆÓ}¸ Œiôx2båQž¬MA"¡ÉkÚ½5ÛåŽJ1´Rˆ/_Û&cqm‘b·D}æmq®cŠÿT¼©&%¡q•ïBE…ÞaÀª¶$O8€Òp©qW36sIkÊ»#¢f˜ ’F(òǘ¬o‘Òñ3 <éÁ¡Ž+Vîpk¬¥–ÁÂOMy>Ê‹½¥ENÊÞ÷àóÓñ¤RÞ ÎÂMêÕæìÉX»”»+ÿXx4 ŸÒ!õc~õ˜tÝæŸ´€2ÉçM¬8ÐDQå9Æò̈A›ÉÐú"Ëa…ãÔtNCéø& œú3x—›.€n͆:L^cäÝ4ê–n¢Š+Ë×6iɰ(%du4píØìMnØ=„:Ž€ÓíR¸ÓOÙG¶Ñ+ú)N_ù²ô“<»y%âz°'‰`Fx&WD·À‹G5NKHì±Þ]¾»ôÆE~N~8‹xlî°zžœawè ¡šbñ®: Ý_bÖí|'IÐÝ”i§ÞR©Dù’Þèë­?F}>)c¼íXÇÏ(YM“âæWðXBR÷ŠòŠô:¾þ´f(ß©¤÷e‘ó_Nt¾yLÖ´@#‘ÄŸ#®!ZÐ"‡Ž:z“Â2›Öf}S®zÔãØÅc~„;ÉŸí 6y^‘¦8u¨.ã+‹§zìSzàh(}öTb²Â†fö£ÕòðK™Ü·²Ð÷lU,ÜÄq"üà.›R"$l¥Ù¼ ãÚ’FUа7Ô…_Æë®¾½õÜE€3HC™'€ äàguÒâG„ø¦Š¸ÒA¶¸ÓJhŠ×P©é }b¹=Ë ^3_ï”VÊ4C–_d´ù~Œ>Èê‡gZ^öмcwÐ[Ø/¿ß ~•ÛG½½àòã”(aµ€?“¹ï±x¢¶¿°µµÉùÜàý*î)ø E–ú_0a!«ž°·Àd9kSž4žð‘1¥¾ôV5‘ï>ü¤½a_³¢y5P¥ê;ƒ–ÙWŒ€’%å(ò ”[võ§µäDÜÙ˜@Éâã.jiS)V"qzJ'™˜–K·LÑuSo¶Oµ%ʦØ4¨Ûw[ó¶î —2ø)j/Y.ÃÐ#4ÏwEìîæËÇ^$îbK­Ü©jÜV•ñŸ_0 4GF+:sŽ{½¥ÉyüRL“¢ò܈ýô®n¨8ÐaKaõ OMQ±…Fý¶3·ZŠSUë‰I©Ø­U³…Ï„‹Ÿ*üŠÛ]³ xt&ü¨°‘xÀsî+KþhV¹çd:Ÿ›)]êºs™.AÌM’WféúÜ2ÿÑ삘mбR\{ÒYÈì„/-:Óï²æÒ¦ÓÑô •ýÄ4<÷“–v’oH[Î2¢-?©Ù¶Èøë.7Ú´ÌÂÐÓ&ËúÊÎáÁ3§iú0~è ècÑ'mCz{ÍK£.…|9îåmë73QxýJÄ‹^Q×õŠj¤•£iJªsææŠÍ“$ƒk·†y/È›6«‰{ø&o„x„©~Œ=¾Úf&úì2¾åÿA3;iÓŠ6¿Ì(´ITûÚµÝr„£ ¸Œ tqŽå¥›¯˜ÖE¢k…¾(2Žvßz­Ÿ? üøÓi@Ùӽ𭔾|SXôU_±¬GãÝiø‡/Óa|(›£õR–žˆBøÔÐËЋÄOv¼\‘ß‘ö[ ´Žfòz»±S¹Öì÷ÛIH½ž®îv1ŒfUXjsG¯/ u.“39ÉÖo£IjR$uùÓÖ¢#Ј4wßøÌ(è<ß}^fº1Â95âÞ(‚ÈÏ`<© Âh¾•µnc¼B÷­÷Y«û÷ü´Vj(ózÌém,õÈs'/¢Ÿ‡G`– Ÿ#ÛḚw¦¼J™ÜuG‹‚Ô¯åÅ‹ÞàE*èpŸ ¬^*x;͹S5éÏÛU_¾kXxïmyÌ~Ôg£fD fþɆâÍ66‡ÝŠ?"}üEyB²wIexÌÎŒÇï[×rªÊÕ³KÚ¬#ÔÚå»%/`²/÷1~}2˜ËÀpIB€ þ ¡uÒ"f·jKüžV*r(xƒïcDB«÷ 2³°º`×*`úûšƒJŠB‘j?Ž@’€DŠö­F¬ËúQÒìa´y9èœüÛ(L 5<Ê~>þ ôH^>ÚÆ»TcCb1Äöõ.`¦†¿†ï¤ðF¦&©Ç•06¯}|¯‰JõQfÅPºO¯+ÁT@kCФÌý'·”ï4ü½š0[×?üHѾd”JйˆÔCÙr!âÛY-Fs>:*tIãvö¿3ÖºÍþ@.µpäT£#ˆ">ùdÖdq»ÿ8E¤Œ|í›#qŸFà‡¶và™DÑÌÖn`›ÆÓÌ9ɵ§…–÷Ýʰ"繺²ƒ‹?d(KúH&‰ï³Ô‡N¬Î˜æõÚDž€Gvt¤Z¤¿×™–xšààXí¥ÙWsùþûgÙë¦À®…Aôì ëJh’ôÜà,ùó˜êæ{ryšFoÇ‘{ÿu,d*/£ž ˜°Ù×ß®1㦒í_Ëí$tê<´¿Ý²ÜîV`‰½åhÚ2¢ï­žhô4 k ç +½œòóͧªÜ3ì¼|È?ÇôBò›§5¢hxûÅa»‡¢^×ë“‘]‘Ëè¾ðä×ãâí="lpɶoXB˜:V,J4¿VMùu9"‰æ .b™hx£:çñÛ˜KÈ…-ÅØ't¼,¥µÝF-ÙX™Ö±œŸËoÖK['Cqá.SÚ`òð)ñÙ8õY6ÄV“ëö’BmJ‚5|ñDp-.’­Ñ³ôáÎoH$þˆNÁ,×½jý FÞÑÇ õr?n;Ÿœ[õñ-*]«Í$’ô# ?¯f¤wé€5)‡`)ZS‚Ý6s´×mX/‡[^Ì™Ãsh”ü¼×ïl# Ùc~?£ùIŠ”$:ÿt¢DìJÂ8¯ŽšFÔ¹%1pD÷Ô<>{?5P:jÞ7ü¥LÿÙ›sƒº ‹–±AӸ©šyf* ½dH«=n[õ“V×Ä?Q{Ûég#ÕÞÛ"Ñ@mÊãxNfíŠöhX„–Y÷x§ÝŒ‚g´ÝfÔ{–ù/°Xï\F­<ˆ‰ÐõpØ¥eqåßËn€¹*¥0˜äøÆUÄË&=V²uü¶q0‰ªVnj?Xñ”‘Ó®Ð~â #Àb‡žRj-ɘ­ó(i³éŒÉ}Ðöœ,û‰¢1èàpÉ|âs9…ÃíܶȺaÿã{í=eV9â’ÿ¡‡þ|Ù·ÜSû‚ÖyK-tTDðóAì˜32š‹É„Àô@hƒÉögWlÀF+÷µxÜuŸ„8ï òAm€ÄjÁ'L ÖHZöû2…O¼Ÿ=ãzçê*-?ŽFÌjJQØò*¿öšáÏ®²G´d,Ï;C`!¿(á|àC6»[!°‹‹–éYßÑ-Op‹¼+IЉYØG>èt8U߆Øom¶ã²(23Ý凌ç*z'ߤj‘ ìp¾³GV˜B[ýp}‰xI€‘±¬oÜ%Ï5ס³iD\AK±üLª¢ÇaEvX‘fËpñЛÊ9ï\¢g´ÓE¥ú=‘t©óÀ‹€MïýÅü7s-™Øì )Ÿ3Œ™Vœ6b¹WÚ±¨þøJPa¶™5KŦl}æà`; ‚£ÆÂRëdâØPÒ6•ú£–Å+t1ÕÆ£tË2ÖþŸç;>üè’—€* ÏÎñóå»R^?6(~Vçéú(É…ø½ÏVã­Ø³JÌyïɪh…\t(Ä J„ï[Á7óÚU û5íZR ØÁ%ÍE‹s\$„+•ö®_0x9v¬ –üO–>zzˆ¼]£Â܉]B<ÛÞ¨ þŽnï–¨|÷~w†%­ºõCÁòåáp_'åd$êÔ'9Äx°n$Ïqƒ¹¡FxqExrZ®æÏ{öVã±Ò9zÙÏ,‰XCÙÐ ˆlßF"}Q‡õ›zï¾x¼N«MsÝLrEøæ£º»Ì¨ãš4§Y0ì§9•O©Ú*%9§ù-˜Ïføâ-˜ì¾ßOŒFŽN©Ž´b°çˆi¡Èé%<¿hb‡VÒtÑÜ1.øŠµÛTz>°_«Îr<ªåõC7ãd‘3ž¨ïQt©—föí³âQ-²3Œ¼»QÓîŽR~«¬~s»jáÁVžˆ{F[õŽ—°Ê3ŽÔÎ)ä³'úê2ú<ޤ˜däÛW&apô·+…"bÕ­ÞBD±æE‹ô*&ƒÒ“Ñ>” 5ûÜhôÍ˪µ´,³’†PÉ8…¶<ÛO-µ¢oq'¸–&ÑVÌ8ÔU¿Ðr!ý\êü•§éˆ¤µÔÿ­‰VÂlK\ÃÀ#œ˜¦’Úï ÉüI:xñÛæ‚º*I•¤ùþ¸b\®Ø7Ͷ3ühiþlfýÓò LPVØDÙÜëd÷÷ Ó K  ²Lôe—YÕÑÛ5¹· "–œN©äD y[®CZ"î ¦ô@›uhĵ]’ÿÃÖkÝÑÌüäj†ï \ñi‘<Öa›°.6] ïù³sÃc;æ ,Xu-ß±¦­·bÌV#üˆÊÿ\Ñu{Š?Õ¼žÜþàdji@×óþC2¢ãB£Y•‚LÜÎ:°…˾g‰nÉZ|0²ä¹J¨ú‰äy˜®â9q},9ÏüQɽáä}KÊÜ)WWyæ· ¦‚“*9Z1ŒÊK"¿ÜY­„ëï³6ÈØsícø[é*Ž S™Q´èqzoÆP:;fµ1Üö%´© mÄÍÝÄ——Âr!W‹¿~Ò‘Yl”“KÁZΜ}ׄ—`Í•m¹2tejÆ{Çää9.5f`ZyO qÇO8à Áë¨á°@Ž‹l7–Qî ×;¨³×û±SñŒ!Á´·ÎV>‰¢î©«NO"tƒÄ,YøÐàl4än°v+2`ïÀuËKSÉdÌMªÛoEX0°©Å1äLqYäM’PAЄæË³æO²µl”0ÉoÈÖ’brTMÆs˜ƒ-†}Øê›ÕÃuôJ:3´4I‰¢Æ®p”VÃ8Ïçx>PÛ„Ãå¯&W¢ÎœP”h\nÁ¼¦Q°<÷= ¾©£93}›‡ŒïÜõ©óq‰Å×[€p3‡(‰‹z®¥‚&7 ¯.êÓ·æ Þ“##?Û§·ÀÍN3[åòärê*òC÷תÞT¸4c0__â³gÏà¡À¦; Å~;‰GV™S«MÛ˜ûˆAwz;WñÅœ¸7âG¦}òkÔýöî¡Ù'­[­çNÛ *Ä’E¤zcJ|Âá‹ Y¾c€žñ9”Ýu“mq£Ä½BÓèËŠZ›=ʾ1Á×ÒÌG_xZwA¡¦$·ï¶Ÿƒ’à*m|ðˆŠÎôTZ¦8]Q—W#ýY¶ÈáÔãûE¿ÝfF]³,»¸AIµÜœÑ 1ËÓrm Üê¢k­8 âø$€’:NIAý–Õ•äZñ³Þ1;g‡K ñE½Ûãú†‹ Óǃ'¯(‚’v£]Úo8`+´8r’ÑA8ÌD]“D bÜIgúÀõg2㙑E ê6ØÛ²n’O <Þa4I8˜ ÎX^N!(åT^ÒA–Ø·x<¥ë1°†}…»'Âîv° =Ò£÷7¶RSæÍ°zš‡‡co?mVcgç‡Ë¿v å•cgZØ·À3«{Áo·}ßoW3o¬*#yÚ€aMb™- |µJ?å¹¥ßDš×´ãå=Þ°še ©E¡8ðpËKñ»ÿ›´3¾B ^'µg˜šBHÄœ‰êvjk.ͲÿuÄçôÝ83—꣬ļU¿|À€äbþ#¾e[Lm¹éˆ>¬Ó†u}« t­ù^ÞO: J²si*Šdtyùºvb»xÇ2ñ wµÍ]b˜Å?Ҁ؇Þàý*÷¦JÝb ªùä°6û+:ЩÏöãpI¿ÔÈ"é6[¦Ù0îI,b~'Ûy9V‚¨rñú9ö= AÅf±ó|ÒKêô–ððƒAŽ£ŠK|ݺ4°Æ󱃜 Ù=ñ戶 ‡Vè …ÿÇÂ~ýœ¢vKv¹ÞcÚýý°uÄC…Vb$–úC›¥‰ê»Ýþ “ Áï™:äãpV"îHcæBhñ¬r©Ø¢$i8ýª:ä¨D¬-µ2=² æóR÷ÛЦ=ƒÕ*–ÖtØF @/^/¬¨AÞékÝâÔé™–Æ%çÝ;ù!ùF»´âÞðöȱ}ìʼn¡¸Ëø™w·«¤¬]ÅGªµ~%†Cå%”G-•©×©ï%_ÒE°½k¹d—‡…ùjò­Q*™ËÉœ¯„ñ’ºÇxÉtß@Úã6˜V‰­×ÿü¦¤_râ:ŽYˆ»±‡»B «;ˆUSâ{òÇö‹´mð{ø‘#fQ€·1r•¬øxN#,X'³ä:>ðfšå¦9¶ßø!ÆY•àÚc÷%98Cq—ë„cÐB­0Û¤uäŒ0»öê¾ú8mVÜÊ'BÃ)3úGº{ÄnÙLùõ77g%”ù%ª$­ÝJ¥ GÛÆ˜Ú»©Ït¼í£'ºù‰>Æ¥=¼ÖîHO}ý=³Xp"l”¨ìiÈ-ÞŸ›R‡^*×óÓÞåº0ÜB«ÈsêG‡BIÙ%ý^‘šÉ¥\¦(ÊEçÅÂCž·ôK1¾µácÜ-gé\‚w 3¦ŒàPrÕ|–ƒ™çí@¶áì‘%ç$$VÕ>+_ìpC¢“ÐÎ gÐ¡Š‡cß <‹°bLÉ]’.íîÈ£—FàeaþÑLFØÅ^ÍôªÀë»rK­‹‹ D)ƒ»ÂWÂå·“;ʪ¥i…’åW®»NêýÇêÁ=vð€"y¯í¤ƒè¿7§Êo× ÂÃ+ÒSecÅŒ¬AÌX&Á첟¥Ù2f£/^45£¥»ÒÙåx «Úwšþ”òêÒï`Ž*…ÁÎ|E_Ê»£Ü—§Q0k‘ÁlÛa®­ò15W4Aiô°üUŒˆ3Ýê|?ïN¦·q’AÀ<Ô™âdѪÍû6Ê®«\?:ŽmûŒ$Ö´ÀE³öÓ 2öðÉñòC½mŸ oô8’"~’|D/Úb­Íúî>Ó‹O¼)ú¢Ž¹±uéÛM@貤uñ[´©Ú¬ÕŽÓ–¹úwW–`iïLºrRÁ*æLjñyç0•Qhmí‡Ç7a¤–‡}‚yµ;ÑŠìÚ>tH8®è‰ßò_¥GEÿhãÈóP-yOJùLÉ*ÏÆWËÎuïm1ú#ÿ¬Jf'@¢÷ºàiÆï¡/ïÉEzáôSE-(†ÿ‚›æSеW"Z4çz]v-ÆD¿AÀ*è‡gÅË@>Äzkç ±5…ÀDè©óö`?&|U(ÂÑ1ÌÖ$ÑDœèlïRá>vÕNBá8“ ßÛw3qÇðTô)$Ì5·x&ETÝžñŸ£¯Ñ$J½,Ñ¢;b¦žÒ)÷yoÕ%»ÖKÍ»Ý9›˜]ët7c”ƒ\ €WlK8 ÈÔ(± ”ß=üíí Zx Y¡c¯ŽÌ¡E´`h'çÍ?qÚ*mRìš9 KÜOË>­/±|ßtËL›¯Ü¿%)Ô‘'íö1Þ¤Ï •ªrnÍ)ÌÆønY¾ònÅ=UäüÎM¤,„£>ø™º'úi›Ðb6óÊûi)á=ÉWývyzz™Î“þZ¹¹EUr˜&´úm"!#©Âž›YYá6âêûŘкýªðÊõyÙÏ_¶X¾ûpC1åÇ•“¼h7k'nfçI¤1Sd-ùÎâ–¹plÊzÁ©˜¼ éíbnûf‡•µKˆHö~ãZ´Ñ²äîzÙOÀgû“Œ,]jl¦ÓkXÉÔe1ÕÒìhôÇÙÏt°ö7‰« ± ºœbÎþáÍ!ÅÏ…ÝY968J.ßëZçi\óDÒšâ§eï‹3ù  bÀîu-Å›î_™Íö¾É$ðôÄX§N‰ó«¯=´PÝ0ÇYWhø¸ž*.D[ªrh IñÔ¥_íqñçVåœ[m÷¶&ĹcÑ[¿·îó0Ü´ƒþKM>÷Á®+ÓHrþ‹û-O“ÌëL k¨Æ†Z’hsáù)£–ïdWŽcãù¾÷{·0áüLiXzјÅn9Õ Â<‚aFŽn©ÝK•¯y¹Ý{o/ßÁ¿ P´dkaû¹Óa¬’aë·ÖrßÁp›+î;¦…ZÅ8êZWë¹wS#,ó}œqö«Ô•jÉË×JÙÃìfH"•ž³Ò’^9Âä®s•ƒ1¯9¹LZfqCÓ³°:×›ØO›jµ3¬‰Tggƒ’òöú8…ø•£Œ˜-¦tg‡²Ÿ+FW“3>Ã(Õ\Êî¾—Y†kÓ$o\Ì8fóL|ާÆéæÙö¦¶vñ Á*øÔk¸™¾F¢ú6ê„›w¡õµ5åf7˜¾ŽKÎ ø ¶Ý0rëÚÞç¹Ç•–wzÜ÷e´Y Ö —àøKçDrO.£uÖ…„N^ ݼw|³AwK´†=ÆŒ¹;0¯ô²pgý£ä1kežßpÊÀ¨žàðQöÛÏ£°'±E7m¾5ÃŽˆjMµSÏÏË|¢ Àtº¤:ÀÏ=%mÞ³.oމŽçà'ÇxÌøÚÔ¹èÐÛ è¿RÒÕÜÝ ð¯0ãHºƒHZ6¢NfS¼ êDœöÅÎE4Sœ1¾^ ¬Á0Švê¿Ä¸2 ö”X(’"ËÕ{X¶ í.n“;¹Õb¶ô×Ö•[$YžßþøP4ÄŒ°1D,cŠRàˆšÃâXéŠp+Ðkiô[ð]k³|•À^Zc¦Ý‰ù»ñˆamC“ÿ3õ vcAfbGŠ g±ÄSëƒ÷ÅFúÌLÄr[zU#xPv¸nå` N­˜Úw$¼.­§ž·]ŸÙ`{ròmJ9ÙÓlr»—ž¨á8Ø~¾ˆ}*Jø6L45wùÉPv°ó'{Ÿ?²ÀéÇó‡x«Ô˜®•ËeÖq¾¡ú }ïð• Ç» 4® ›` Þ9 »”÷QìJ/¥srd³:˜ó—S×ooRÚ+›CéÞ¾`pjo˜ÅÔÎX{ÚtV)Jëå/ųeRïVlï§>°JÔà׬Nj=ÉCчu¡sD †/4ddõžlñ‘tBï,^îU‹l¼Xkù>nþ¹™Õ™¶*œåü D~pý9ÉS¬ÇŸâ›üÌMz×e5æ‰k)nsà÷ˆ7ØUµ[rò³ZXu±&;þ¹Ù Êtð˜éÑ*£Œ)øìL11¸óØrT¡«£Cœ—ÝŽÈ0”ä[dñèè=ç-8üɹšš(7tÏK™^³xÜŽÃ35§Œy;fGd;Éa³tá€$µ«5¼^dÙŽøµœzÖ†&Ž¥â9ˆ¹.[Ž#øàž\_?«vûì°‰ØÛ¯g1õ¿ Ùîy©Ÿš$oÞoÌ$&dÂ$(5÷õ*ú»©P+"“ù¡Àá§·#ê0ñÙ‚[¥”íP’WÖ›©-‘•oñúX(½?ü`pz`TX/ÿ˜â°‚u%EMë«x«“_ M( öS»™ô šø<œ¶ k£胱‰‹ßò•b¿að„Rjœzü±³:åJ\:BLüFa™Ê8–úH—áMª u\£œ{ØYù÷—ä‹"n'5´ë¡kŸœ´ofúðm±¹oüžÊR„ƒt¨¸œEºpR.=&¨VÅ,7]ôØêùÓ9£oûË’êLc}ä „[: ¾ ÒÅZ£‰Ê5J–£°%'?™cdšM¬ö0-J.‰} ]Q¦'¡·N19 º~îªÄ8×lZÅ¿½c/$¶x·}‰ªÑÆõÒ`:#^.¿;™cîFvâT.##™í¹Ã"ðù¼%Ži^"4…i”ˆ¾õíG|O¬=oƒæRC%M_¿ñún™2{ŠÚY]ÚÛDCDEãéK²H=kµÇ<ÿYª~!Y†ÙTœT³\g7Ó~×h1}SÏÐ1zl'6¤ØðY4t¾âÈ‚q!ß&¿Hr=$‚Kç-ýY^êœsÂØEuË~ÛC—&:•*ÐGÄ–ö¤Çœ»ÔÅ{ ’Ä-„yöÉT©Rê<~aÿÞ@܉ìÞTCµ=ÅÃÖŸñÅñ›R$Ê7ý Q†W†ï@Bn”iÈ—WÓI W»êÍŒr{q·ÎŽÙILîbMj]аÉ+Ðé3g®æ¼xnW#|Òœöƒ,”¤‹ÉW{1zxúí¥aV:tD0(÷­ôE‡ƒú .• ñg¤eÅ€%51íÈ™.?­WÖ+Ù/ÉR°*wHôgë;¤¦B?Rã⢤.rT/qa[Í1å½¼B*ÁÄÍÁçîüáY½~ºçKÞcÐüÎé¡Ñ‚œn–ñ4µU"-.< '1Âú1ELÅS€^;A—µ½‡þiʣѨG÷5ômÏw'!NÓô}‰~Óò…ûÇŠ8NÄ­ç™P™„O‘σãÈdµùûN®Ï¯’ç㯩LT›´çaqÎ0Q̼56¦RÒÖ°–ìÚpæáêB‘ä;ÐN©ÈÓ´‰ÓËÚ›ÑHè˜=ž*¸šÝL8‹ù$5ÿ4Ôk{53 %ë1)o›í±êÓCá|¥5^Ê `s#r/Ù}ÀšCs°í2˜%Žv0M¸Ï»,^$+~”H|;©ÁÚ´wGµAÉôõ¸SØùÔßü3ìÀÆ‚ä7NHòë±¢aOú¤pªÆÅÝsº"gõàRX;¢x½g±MÉšCOɺ—»Óo\DWó”à5¡íÏå*EkÁÏ´²þ—™¼h–*’¢q¢e©Ãõ Ù¯ƒ»ñ©Wü;¸sÔFÄŒËzq?÷÷j¶›bÀ¶l?vèuć$µ­o>€BáÈ8H÷Þš·õ㽇ò:ld)|V/ïW=¢°ëq‡èè×NÜ)}tÿÒÀwËÑ­õ:vSáŒO»Ùý¨¼:ä ÏÈ Ò yáÖoÕR¢Hî ÔC¶oVLÿû ”²Ï©'aE*³ãólâsa•¡†%¥ïúýi™:ë{Á?÷w?¶º°Þm))âþ7_$áÉ8c‡ë¶y¡„q:®ÁðWîCAÀjy@±ÔXàÔI<•¨6ˈ'gû<5-J* ­WtëBBÎf :Vèâf–p›  Ü“«ÂX†-ñ”ã²KFUŒÒáh²õÕûÉ÷Õ£¸ÛÜéÌ&ûR¢$9omÞÆx(, C·Î3cí^´ž<±¼@ÓÊŸ3`ù¶|¶^®† ƒr…ƒBf âgPf} ?ðÁ¢œ»0üÝHñ5uÖÑ_Uì•9U‰8µz–̆œ=ËîõÍ÷Ç—2ÑÓûfG&!:hñ¸ŽÔÙ‹ F 9¥ÁЈ^ØÀ<Ô¾´x‘¨|É^ª±xd2" èÞ FtJ€!?»Ì$üˉìíOÎ[y]4åÞfý¦ý«P¡šÒí× B:†Å|ºèùQ¼†Üç«Ôi—¼ ÀÃáЍ/âU€kΡɎØN^Q87ö´¨}¡ØPO`ŽŠB<×Ï0{ã4GÉôÅч¯&Ý{R D­ŒëÄÜ'»vRIÌw¤-6Û}ü\›>=lÆII–,?.SÙ¬å½ÍjZYwÊ$™x®ÅGd–¬JÕì{÷’éQR鸓5Çñ5΃RégÿÎàãÃH¡Ó&zô ?úŠ!x Ðj4ly†áwÛ²ïéÍ'?Nô4Õ3üšj€¾b‚Åï·¶Sᔫ²:ÍcÔþ4æÌ„î¢Ow»3‘ª­÷žÇ4ÐõYâJËz6Z ¹Ö°þ›Í úq?³L çqƒ/CÙ2kMÂWë1¯¨ó¿°i!$^4i1·²fË«{³ÇIž_Û[d[™mÍÍWñÐ_¢+1 ôO¨|=±‘(ñ¥ü LEk†‘;õ?¢®t!h0á–XUÎØ¯æ»+ry¶³ïd7®æ[Uàæ¿õÚG¶©ßBš7LòåîJâljH‡£‡«Ïoâ1È íg¤T|ý /èõC–»£€ž’ÁüMux·hÀîM’r:„±[wîåa¦¿‡Có˜ï”Xz¬&ÐA{¨v7÷ŽVm~Zω„Aê oGÌi5Bg+‘Ô+¶[;rßè_g‡­™$G(Ùr•^š†%ñÉ–;¿ °ŸÁ(¹M³Rdவ¦|è2„«=1C–nö_¹€N>£dHÑ”nò'–ƒ²¤Oª§Ô[1.÷©lFh8ZƒÛ5- I´¯(&±¥(Î#±ñ¥LJ¼„‰—´Fl¹Á¹†Ô\ï}ñ7¯6þé%ú0}«ZðiØ%¸wr5ãçþÈNŸK,k»çížñÏÀu<ºLŽ`w0áÁh;k¸Úâ7êVÿS]i׎äwXåV´‡:Ù&»= á’ït¹FLpÊSS˜è—½Þh«.· Lª”[æEdµÖ*—ØÕNÖogþƒ¼Ý3Gåô§-^ÌOd-£›­?hÜå{­\zÙ•˜e…2ÚÜðÃ<†hø «”h¾Zâ¼} ™s¡CÐC@±zq0c´²Ï;+j¦¨ß-”åw&•Š掤€M4ïûññ‚JK_;ŠÎ—k^QøNŒ½È*Q®2¤¡,’ÒÝð$lÂ)^«xÌT5%.K3¢>*¶ßÒquø‹üÖž®ýE=y„R»Ï<¯ B<-vâL«¹‘)äßlý‰DŠŽ®í}â¾ÌZÞX}…ŸÜМlŸ‹D ¦BUaøtÊØTX…ÙmYVÓ®  ó¾š-É#/aÐu8ãÖ¯y)›Üù†…­ €0Nt8…;]˜r½›.Eçì[‡Xc~ 2©îóñоVô5ÖZlÚ@7®í9Ó5Y~•2ó`P…ðx?ôxNKÃÎÀ5,‘[Kû%éüÇVÑßÀ -Âü蕳F=nôÞ;¥ÄPÒá:ÏŸ=¯ŠaªŠ6ȬR5Y¸{÷>ô?[¸hý8ל%=}8•U.87,û¶·¢mBN^~ø çëí ~þš©“Òr®*\ð¸ï‹9–ÝJèóeM+¥µÄŽnðgnV­å—šåòÉnÎO‚¿‘zXkzA§âÔ”w¬×míhçGîv¡cê—Öxí+²ì.4>C°Åu¥^}h›Õ¥ÔëZ9‡÷å´g™’_‘Þ#ökƒ¿ã=\[‡ÜªæË,xеFù‡epçÒ!ˆ Ï=¢©éx.¿‰hÂÝÇ.¬Þ ¶nl­A¶*ûëÿŒ“-èf‘ØrÄõ*:””‹YƒahpÓ”Ò˜“KqZy½%úݪ³_5™­ikXD‘+ßSãÙéãÔ ¬."`·lóÌ2;¥K-PmPd®xEýËD³%jø%Îzý¨Ñˆ*ˆpÛÁA!®FÕ#ƒäÇ£BiíSWkQǯBAnÉf¯„-×IáëØÑÆž xÖDŸïáÖ¦ŸíD['Ó4f’î¤6~*#7=s}\¦Ý´’ýc9ö”ÃLt´IÈ‚—±o‘D;I}nÞ°éù'¼ÿ±j¡0áÔX‚_Ï„.{ˬϲY4Î0xb±yIŰ#ËUþÖ„¾Ý@ ”ˆT>ûDZ<*´À›Œ:N.Ö¸ª¸']Ìq-Ÿˆêº¤H¬ƒ]`à‹šëv––À¿<ž•$+·ÄðBY×I¤nAP÷³<ŸÀêB7[¸VT{}_þæl':H¿èN·\y}CB!ì‹ÔÏjà¦Ójí΀ªnb'4[`ôr8ÀQ(ßž¥CüýFŒXçŽq³7‚ÐcT€ÈH‚Y«L,bÈŸ’D(}=)x";_Øœ|´6õµzhÄÛlµÉN·+‹sÀã"‚·¤¯fè:7÷¿ÙHÿ":XFÿi¡Ó Gš¬òî=kƒ«FþKâºô÷Qí-$žA˜üT©YÃJß-Æp’Õ¶t¿AQà?BÍ ›ò/Å(†Eþß·qÓÏæ{Åi§?WÞzrdÔ¯ß÷£¹Þ&ö›ßÑuR‚\F/@nGü’„°Ÿ†8_…̶RuÖ5žnÄÐU´þú|/¦ÔŸÉnÕ"¡·ñŽåèF¦Æà€€¥¥Þ‘/LMê½ ýjÚýà|Æ=Z Ìà|ÏL‚³×½agaÎ!³ŠÉûÈ4ªä—æ€7þñ6ÄÕÓA³œé¯Š­¢,úExã—’ôkÛE6øo3@:͵"#|º§’yÇÀ_È¥Aª¬LøD7üÔPÕÀ¯£'%àÛ§T“/ Ž[ú±‹Ž¥§3|*áQÈ\­ Qé¢`ì­|°Bm²!GáHòÎYm--ës‘/aÚΟ ÑÀzcx?*L¤r¶`Êøþ?nÔˆ½óôÓ° …I9d0¡ýþB/ˆgH¨èÄ]DÑMæò£º{kÊ4ôKé¦ÜEˆííš*A@ 2D1ÈXjÒø‚Óª²`̱ÿÓêø‚ºÆöèƒóh´Ê¹… ¿‡Ð»NUš¯‘?áÏí@>ª•ÃÈÔóêiõ®&±H~5F¬–ÇÉ8³7s"Ø}‹Ñ“KW`±<ˆ{{õm`†-}5 fÌmI´Ú‚òc‘2áèÂ|!lå?Ü!sO¶Ïvþª-Xú[ªs¨[™ˆÎêÏ­Cè\ëü=2#”^¶­…!¶Ò#_ñøq…T­@ÄâäFä|m üZ5[h Nv}¿W”øóS3ò§Ï>F³?àßsGŸR3 L^›±wت}'„¤`¤Æ£’(Hö RHs}F\µVRCDcŽYÓNÌÔádÎ|)xÙŸÖuí*ÇžÒ g·Ôb­ ÚDèFAâÈݺ°%B{…Câ»]å¢äô]¾Žiæ‰-Yòz}þõ ìyÁw×…Ç1$3›sJ‚ûq·ITæÕ¡ù¬²s:â¤â0H:º¶"sm‘(ÀdÛEsVØÆ$/¿ß5ÃJàõœ{÷(FÉs²ž‹Óura+ÍMö×Y| F˽N†ö²¡z‘L|ãB­/¯&|Ï0á ý˜f9äß 'ß1ûvÚðúwΗÛ£â—Äd‚ ìø*ÂÁ;ðÆÒžÉn‹²fÐì–ñ¸˜‰N¨ ¬Úù^°Dr- ",á49¾- å¢Ï‚GŠrí©Ah+¿³àÈÆ S(Ø`®¨ J6ó3ãïj‘¡ŽFZS?5ùízUrˆÞ2½¾ôóó2ˆã¶ ¯ƒa>Ùˆ€Ìçʬ„Ôø«òDYxÞeêI¨Ý¥Š’{Û2c–ðB¦–©é’ 刢Çàcú,›Ú €¤åè¶YCùƒfZÓ¯V¿€åZ:)¦ïË¿¡¦€«“¢­tA¨Üc}'ØÍÉ3Áiì1»f'&MU=À‡ì 9nzéœÈR8³ :Ÿ(¦æÜ£&ùi¤¢§c–á~_òÏîs4éuð,õ}Q”Uͨ´íê××/'{²€i¾ë $‘%‘ÑpÚÆ?Ú€à „mì¢\ZðY¦šªK–³˜•KZ‡ÌÇéXÝRÁ§ß•¥U¾š9íŒÔîd¸ ÆZ°p¡bÁƒ (R]Š2 jïÈ¿GlDòQ"•€J(6YG*«ÛNhâÓØdŸízÉ}!S3°¿¸k`N^3ý™¨ÿÊ_h³†Þœ¸">¶5üÞñ“ÓÅÏxEÜ*¸þlG¢œ*è”ðEÖé,qäÌ-yø®t‹PìßÇ$íµ„v³qU “Òu¡ŽU)zåYj\*¼˜˜^4¾Ð}õ{äòesæ\®ä[æß‡³µR–·GÈè^2£×`& K ‘r6JxN]Ya "ßVÂa ˆ½ùÒÂgrlpiÔ°†ÉT]<ŒÓ_–V€Â *èošÑ©[èÙ#mµñóè¢f ¸Xž‰¼Z°r-^'—¡äôRlÀ„j Ž‹ÈÜç¬M.ff[¿lR è ·éô\U5¸ Ãe_ˆlŠêAŠDŽª’¦ãáh¨åµäü­´úóæ¼OŸÀÎòå´úYd|iÍ—‘*aQ¢«ƒF9”w¢¿³p*(€Iþöµj4"ðíÝ,“$·„‡ÅR’³–8Á^r%¬eÅyÀ¾EÊ2„H{ëÚã–¯‘©USêëý—}>ÃâØù·53ð…(ð…é.£A]%µ­{¯L±Hz÷;6M“x+ý°±ùs4§_­S2;wù  (M·Y’}ÔÍhÎ?‹y)Œ7˜%¨µþÝÀ‡y²m9ösr´WºøZúÔ&j}IñÕNyÓãÒ~ÐHU'ž5?ñ‘qï_)´ôw <üJò7û~†ëqh9ê2äKg"i0 û³uB¡7ì'…a}0ꢶé‰NˆD¬³ÝfŸª™'¬ØaB%iÃFÌ4n†Á¤`¨šuÙû@ä®XI|›*•6‡U!³Xô5ó¸b)0±uÐÕ¡žÍ|Óa¥/#Ãû…H쇽×| µÑìqë—±ï‰| ói¦ ÿ’@ë¦Ûkôf° º„Qµ©<Âa™þ›Déò€uy ì>2}2Ê”¤Ýîğó:­&°å®ìì ¸W–•Ro]zªv‹L—µ’¾‰S7QT°‚ý¿MÛq*¶WкLåÆ6°•+ƒlýS€N³ÇTSú¥ á‡PÎoŸ“oüî0wÏ:óãUwŠÏ¾³LUãýùyù7a4¥Cf-0‘ l«ºa}v«1 ÚÎ}{Úƒ óÔšr× 9aÓi'EoÈ‹†¥’=T@š‡šý< ý›z­ öFܽAjoÞr^Š„nZ­QýsKÎùÆ[­(ßëXÉù ûä9 €(7YtììWµÚÜßR> “@L¸l=ñ½ endstream endobj 1412 0 obj << /Type /FontDescriptor /FontName /KBODNF+CMTT10 /Flags 4 /FontBBox [-4 -233 537 696] /Ascent 611 /CapHeight 611 /Descent -222 /ItalicAngle 0 /StemV 69 /XHeight 431 /CharSet (/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/R/S/T/U/V/W/X/Y/a/ampersand/asciicircum/asciitilde/asterisk/b/bar/braceleft/braceright/bracketleft/bracketright/c/colon/comma/d/dollar/e/eight/equal/exclam/f/five/four/g/greater/h/hyphen/i/j/k/l/less/m/n/nine/numbersign/o/one/p/parenleft/parenright/percent/period/plus/q/question/quotedbl/quoteleft/quoteright/r/s/semicolon/seven/six/slash/t/three/two/u/underscore/v/w/x/y/z/zero) /FontFile 1411 0 R >> endobj 111 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TCFVBK+CMBX10 /FontDescriptor 1382 0 R /FirstChar 11 /LastChar 123 /Widths 1377 0 R >> endobj 110 0 obj << /Type /Font /Subtype /Type1 /BaseFont /MMEKWT+CMBX12 /FontDescriptor 1384 0 R /FirstChar 12 /LastChar 121 /Widths 1378 0 R >> endobj 256 0 obj << /Type /Font /Subtype /Type1 /BaseFont /IFDACN+CMEX10 /FontDescriptor 1386 0 R /FirstChar 20 /LastChar 90 /Widths 1369 0 R >> endobj 121 0 obj << /Type /Font /Subtype /Type1 /BaseFont /CGOAVK+CMITT10 /FontDescriptor 1388 0 R /FirstChar 45 /LastChar 121 /Widths 1372 0 R >> endobj 183 0 obj << /Type /Font /Subtype /Type1 /BaseFont /TALQST+CMMI10 /FontDescriptor 1390 0 R /FirstChar 12 /LastChar 122 /Widths 1371 0 R >> endobj 259 0 obj << /Type /Font /Subtype /Type1 /BaseFont /QTJJQF+CMMI5 /FontDescriptor 1392 0 R /FirstChar 99 /LastChar 117 /Widths 1366 0 R >> endobj 255 0 obj << /Type /Font /Subtype /Type1 /BaseFont /NBGAXO+CMMI7 /FontDescriptor 1394 0 R /FirstChar 58 /LastChar 122 /Widths 1370 0 R >> endobj 112 0 obj << /Type /Font /Subtype /Type1 /BaseFont /LYNLCD+CMR10 /FontDescriptor 1396 0 R /FirstChar 3 /LastChar 126 /Widths 1376 0 R >> endobj 108 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XNYRHV+CMR12 /FontDescriptor 1398 0 R /FirstChar 44 /LastChar 121 /Widths 1379 0 R >> endobj 107 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KOFBFN+CMR17 /FontDescriptor 1400 0 R /FirstChar 70 /LastChar 118 /Widths 1380 0 R >> endobj 258 0 obj << /Type /Font /Subtype /Type1 /BaseFont /RHUHOI+CMR5 /FontDescriptor 1402 0 R /FirstChar 49 /LastChar 50 /Widths 1367 0 R >> endobj 257 0 obj << /Type /Font /Subtype /Type1 /BaseFont /FSMIQE+CMR7 /FontDescriptor 1404 0 R /FirstChar 40 /LastChar 61 /Widths 1368 0 R >> endobj 120 0 obj << /Type /Font /Subtype /Type1 /BaseFont /XWGUCI+CMSY10 /FontDescriptor 1406 0 R /FirstChar 0 /LastChar 110 /Widths 1373 0 R >> endobj 311 0 obj << /Type /Font /Subtype /Type1 /BaseFont /GXPZIQ+CMSY7 /FontDescriptor 1408 0 R /FirstChar 0 /LastChar 50 /Widths 1365 0 R >> endobj 118 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KSUQKT+CMTI10 /FontDescriptor 1410 0 R /FirstChar 66 /LastChar 121 /Widths 1375 0 R >> endobj 119 0 obj << /Type /Font /Subtype /Type1 /BaseFont /KBODNF+CMTT10 /FontDescriptor 1412 0 R /FirstChar 33 /LastChar 126 /Widths 1374 0 R >> endobj 113 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [82 0 R 115 0 R 124 0 R 136 0 R 155 0 R 162 0 R] >> endobj 171 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [167 0 R 178 0 R 189 0 R 197 0 R 203 0 R 212 0 R] >> endobj 223 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [220 0 R 227 0 R 232 0 R 239 0 R 246 0 R 252 0 R] >> endobj 272 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [264 0 R 277 0 R 286 0 R 299 0 R 303 0 R 308 0 R] >> endobj 324 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [321 0 R 332 0 R 345 0 R 350 0 R 354 0 R 361 0 R] >> endobj 373 0 obj << /Type /Pages /Count 6 /Parent 1413 0 R /Kids [370 0 R 375 0 R 379 0 R 383 0 R 389 0 R 397 0 R] >> endobj 406 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [402 0 R 409 0 R 413 0 R 417 0 R 427 0 R 438 0 R] >> endobj 450 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [447 0 R 457 0 R 472 0 R 477 0 R 486 0 R 492 0 R] >> endobj 505 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [500 0 R 509 0 R 515 0 R 521 0 R 527 0 R 534 0 R] >> endobj 544 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [538 0 R 547 0 R 555 0 R 560 0 R 565 0 R 569 0 R] >> endobj 578 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [574 0 R 582 0 R 586 0 R 590 0 R 594 0 R 598 0 R] >> endobj 619 0 obj << /Type /Pages /Count 6 /Parent 1414 0 R /Kids [609 0 R 628 0 R 633 0 R 638 0 R 645 0 R 651 0 R] >> endobj 659 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [655 0 R 662 0 R 667 0 R 675 0 R 682 0 R 689 0 R] >> endobj 702 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [698 0 R 705 0 R 712 0 R 722 0 R 731 0 R 739 0 R] >> endobj 750 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [745 0 R 752 0 R 756 0 R 760 0 R 765 0 R 769 0 R] >> endobj 784 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [777 0 R 792 0 R 802 0 R 814 0 R 821 0 R 829 0 R] >> endobj 840 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [837 0 R 843 0 R 852 0 R 864 0 R 871 0 R 877 0 R] >> endobj 884 0 obj << /Type /Pages /Count 6 /Parent 1415 0 R /Kids [881 0 R 886 0 R 891 0 R 895 0 R 899 0 R 908 0 R] >> endobj 924 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [921 0 R 929 0 R 938 0 R 943 0 R 947 0 R 956 0 R] >> endobj 969 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [966 0 R 971 0 R 976 0 R 980 0 R 986 0 R 990 0 R] >> endobj 1004 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [1000 0 R 1012 0 R 1025 0 R 1029 0 R 1035 0 R 1042 0 R] >> endobj 1051 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [1048 0 R 1053 0 R 1062 0 R 1076 0 R 1082 0 R 1088 0 R] >> endobj 1100 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [1094 0 R 1107 0 R 1119 0 R 1128 0 R 1136 0 R 1140 0 R] >> endobj 1151 0 obj << /Type /Pages /Count 6 /Parent 1416 0 R /Kids [1147 0 R 1155 0 R 1161 0 R 1168 0 R 1173 0 R 1179 0 R] >> endobj 1186 0 obj << /Type /Pages /Count 6 /Parent 1417 0 R /Kids [1183 0 R 1188 0 R 1214 0 R 1242 0 R 1247 0 R 1253 0 R] >> endobj 1268 0 obj << /Type /Pages /Count 6 /Parent 1417 0 R /Kids [1263 0 R 1271 0 R 1276 0 R 1281 0 R 1285 0 R 1294 0 R] >> endobj 1310 0 obj << /Type /Pages /Count 6 /Parent 1417 0 R /Kids [1307 0 R 1316 0 R 1324 0 R 1329 0 R 1336 0 R 1343 0 R] >> endobj 1353 0 obj << /Type /Pages /Count 3 /Parent 1417 0 R /Kids [1350 0 R 1357 0 R 1361 0 R] >> endobj 1413 0 obj << /Type /Pages /Count 36 /Parent 1418 0 R /Kids [113 0 R 171 0 R 223 0 R 272 0 R 324 0 R 373 0 R] >> endobj 1414 0 obj << /Type /Pages /Count 36 /Parent 1418 0 R /Kids [406 0 R 450 0 R 505 0 R 544 0 R 578 0 R 619 0 R] >> endobj 1415 0 obj << /Type /Pages /Count 36 /Parent 1418 0 R /Kids [659 0 R 702 0 R 750 0 R 784 0 R 840 0 R 884 0 R] >> endobj 1416 0 obj << /Type /Pages /Count 36 /Parent 1418 0 R /Kids [924 0 R 969 0 R 1004 0 R 1051 0 R 1100 0 R 1151 0 R] >> endobj 1417 0 obj << /Type /Pages /Count 21 /Parent 1418 0 R /Kids [1186 0 R 1268 0 R 1310 0 R 1353 0 R] >> endobj 1418 0 obj << /Type /Pages /Count 165 /Kids [1413 0 R 1414 0 R 1415 0 R 1416 0 R 1417 0 R] >> endobj 1419 0 obj << /Type /Outlines /First 3 0 R /Last 79 0 R /Count 10 >> endobj 79 0 obj << /Title 80 0 R /A 77 0 R /Parent 1419 0 R /Prev 75 0 R >> endobj 75 0 obj << /Title 76 0 R /A 73 0 R /Parent 1419 0 R /Prev 55 0 R /Next 79 0 R >> endobj 71 0 obj << /Title 72 0 R /A 69 0 R /Parent 63 0 R /Prev 67 0 R >> endobj 67 0 obj << /Title 68 0 R /A 65 0 R /Parent 63 0 R /Next 71 0 R >> endobj 63 0 obj << /Title 64 0 R /A 61 0 R /Parent 55 0 R /Prev 59 0 R /First 67 0 R /Last 71 0 R /Count -2 >> endobj 59 0 obj << /Title 60 0 R /A 57 0 R /Parent 55 0 R /Next 63 0 R >> endobj 55 0 obj << /Title 56 0 R /A 53 0 R /Parent 1419 0 R /Prev 47 0 R /Next 75 0 R /First 59 0 R /Last 63 0 R /Count -2 >> endobj 51 0 obj << /Title 52 0 R /A 49 0 R /Parent 47 0 R >> endobj 47 0 obj << /Title 48 0 R /A 45 0 R /Parent 1419 0 R /Prev 43 0 R /Next 55 0 R /First 51 0 R /Last 51 0 R /Count -1 >> endobj 43 0 obj << /Title 44 0 R /A 41 0 R /Parent 1419 0 R /Prev 39 0 R /Next 47 0 R >> endobj 39 0 obj << /Title 40 0 R /A 37 0 R /Parent 1419 0 R /Prev 23 0 R /Next 43 0 R >> endobj 35 0 obj << /Title 36 0 R /A 33 0 R /Parent 23 0 R /Prev 31 0 R >> endobj 31 0 obj << /Title 32 0 R /A 29 0 R /Parent 23 0 R /Prev 27 0 R /Next 35 0 R >> endobj 27 0 obj << /Title 28 0 R /A 25 0 R /Parent 23 0 R /Next 31 0 R >> endobj 23 0 obj << /Title 24 0 R /A 21 0 R /Parent 1419 0 R /Prev 15 0 R /Next 39 0 R /First 27 0 R /Last 35 0 R /Count -3 >> endobj 19 0 obj << /Title 20 0 R /A 17 0 R /Parent 15 0 R >> endobj 15 0 obj << /Title 16 0 R /A 13 0 R /Parent 1419 0 R /Prev 7 0 R /Next 23 0 R /First 19 0 R /Last 19 0 R /Count -1 >> endobj 11 0 obj << /Title 12 0 R /A 9 0 R /Parent 7 0 R >> endobj 7 0 obj << /Title 8 0 R /A 5 0 R /Parent 1419 0 R /Prev 3 0 R /Next 15 0 R /First 11 0 R /Last 11 0 R /Count -1 >> endobj 3 0 obj << /Title 4 0 R /A 1 0 R /Parent 1419 0 R /Next 7 0 R >> endobj 1420 0 obj << /Names [(Doc-Start) 106 0 R (Item.1) 206 0 R (Item.10) 541 0 R (Item.11) 542 0 R (Item.12) 543 0 R (Item.13) 670 0 R] /Limits [(Doc-Start) (Item.13)] >> endobj 1421 0 obj << /Names [(Item.14) 671 0 R (Item.15) 672 0 R (Item.16) 685 0 R (Item.17) 686 0 R (Item.18) 734 0 R (Item.19) 735 0 R] /Limits [(Item.14) (Item.19)] >> endobj 1422 0 obj << /Names [(Item.2) 207 0 R (Item.20) 736 0 R (Item.21) 1112 0 R (Item.22) 1113 0 R (Item.23) 1114 0 R (Item.24) 1256 0 R] /Limits [(Item.2) (Item.24)] >> endobj 1423 0 obj << /Names [(Item.25) 1257 0 R (Item.26) 1258 0 R (Item.27) 1259 0 R (Item.28) 1319 0 R (Item.29) 1320 0 R (Item.3) 208 0 R] /Limits [(Item.25) (Item.3)] >> endobj 1424 0 obj << /Names [(Item.30) 1321 0 R (Item.4) 336 0 R (Item.5) 337 0 R (Item.6) 338 0 R (Item.7) 441 0 R (Item.8) 442 0 R] /Limits [(Item.30) (Item.8)] >> endobj 1425 0 obj << /Names [(Item.9) 443 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup1) 405 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) 407 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup3) 431 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 393 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 395 0 R] /Limits [(Item.9) (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2)] >> endobj 1426 0 obj << /Names [(agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) 432 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish4) 444 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter1) 392 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter2) 394 0 R (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter3) 430 0 R (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 365 0 R] /Limits [(agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1)] >> endobj 1427 0 obj << /Names [(agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 367 0 R (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup1) 364 0 R (agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup2) 366 0 R (agsurv1) 577 0 R (agsurv2) 579 0 R (btree1) 701 0 R] /Limits [(agreg\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) (btree1)] >> endobj 1428 0 obj << /Names [(btree2) 703 0 R (cite.Gail81) 273 0 R (concordance11) 725 0 R (concordance12) 727 0 R (concordance1\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work1) 726 0 R (concordance1\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work2) 728 0 R] /Limits [(btree2) (concordance1\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work2)] >> endobj 1429 0 obj << /Names [(concordance2\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work1) 748 0 R (concordance2\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work2) 749 0 R (coxcount11) 170 0 R (coxcount12) 173 0 R (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory1) 181 0 R (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory2) 184 0 R] /Limits [(concordance2\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}work1) (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory2)] >> endobj 1430 0 obj << /Names [(coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory3) 192 0 R (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) 182 0 R (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return2) 185 0 R (coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}list\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return3) 200 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}bothsides1) 127 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}bothsides2) 128 0 R] /Limits [(coxcount\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}alloc\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}memory3) (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}bothsides2)] >> endobj 1431 0 obj << /Names [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 143 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 149 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 144 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 150 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) 230 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X1) 140 0 R] /Limits [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X1)] >> endobj 1432 0 obj << /Names [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X2) 146 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X3) 215 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}penal1) 142 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}penal2) 148 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 141 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 147 0 R] /Limits [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X2) (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2)] >> endobj 1433 0 obj << /Names [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform1) 139 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform2) 145 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform3) 158 0 R (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform4) 172 0 R (equation.3.1) 267 0 R (equation.3.2) 268 0 R] /Limits [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform1) (equation.3.2)] >> endobj 1434 0 obj << /Names [(equation.3.3) 269 0 R (equation.3.4) 312 0 R (equation.3.5) 313 0 R (equation.3.6) 314 0 R (equation.4.10) 717 0 R (equation.4.11) 718 0 R] /Limits [(equation.3.3) (equation.4.11)] >> endobj 1435 0 obj << /Names [(equation.4.7) 658 0 R (equation.4.8) 715 0 R (equation.4.9) 716 0 R (equation.8.12) 1018 0 R (equation.8.13) 1019 0 R (equation.8.14) 1020 0 R] /Limits [(equation.4.7) (equation.8.14)] >> endobj 1436 0 obj << /Names [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup1) 315 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup2) 316 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup3) 325 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 335 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 339 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) 340 0 R] /Limits [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addup1) (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3)] >> endobj 1437 0 obj << /Names [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish4) 348 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter01) 291 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter02) 295 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter1) 292 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}iter2) 296 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur1) 270 0 R] /Limits [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish4) (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur1)] >> endobj 1438 0 obj << /Names [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur2) 271 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur3) 274 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 289 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 293 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}strata1) 290 0 R (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}strata2) 294 0 R] /Limits [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur2) (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}strata2)] >> endobj 1439 0 obj << /Names [(figure.1) 687 0 R (model.matrix.coxph1) 242 0 R (model.matrix.coxph2) 243 0 R (newstrata\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup1) 530 0 R (newstrata\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}fixup2) 531 0 R (page.1) 105 0 R] /Limits [(figure.1) (page.1)] >> endobj 1440 0 obj << /Names [(page.10) 199 0 R (page.100) 866 0 R (page.101) 873 0 R (page.102) 879 0 R (page.103) 883 0 R (page.104) 888 0 R] /Limits [(page.10) (page.104)] >> endobj 1441 0 obj << /Names [(page.105) 893 0 R (page.106) 897 0 R (page.107) 901 0 R (page.108) 910 0 R (page.109) 923 0 R (page.11) 205 0 R] /Limits [(page.105) (page.11)] >> endobj 1442 0 obj << /Names [(page.110) 931 0 R (page.111) 940 0 R (page.112) 945 0 R (page.113) 949 0 R (page.114) 958 0 R (page.115) 968 0 R] /Limits [(page.110) (page.115)] >> endobj 1443 0 obj << /Names [(page.116) 973 0 R (page.117) 978 0 R (page.118) 982 0 R (page.119) 988 0 R (page.12) 214 0 R (page.120) 992 0 R] /Limits [(page.116) (page.120)] >> endobj 1444 0 obj << /Names [(page.121) 1002 0 R (page.122) 1014 0 R (page.123) 1027 0 R (page.124) 1031 0 R (page.125) 1037 0 R (page.126) 1044 0 R] /Limits [(page.121) (page.126)] >> endobj 1445 0 obj << /Names [(page.127) 1050 0 R (page.128) 1055 0 R (page.129) 1064 0 R (page.13) 222 0 R (page.130) 1078 0 R (page.131) 1084 0 R] /Limits [(page.127) (page.131)] >> endobj 1446 0 obj << /Names [(page.132) 1090 0 R (page.133) 1096 0 R (page.134) 1109 0 R (page.135) 1121 0 R (page.136) 1130 0 R (page.137) 1138 0 R] /Limits [(page.132) (page.137)] >> endobj 1447 0 obj << /Names [(page.138) 1142 0 R (page.139) 1149 0 R (page.14) 229 0 R (page.140) 1157 0 R (page.141) 1163 0 R (page.142) 1170 0 R] /Limits [(page.138) (page.142)] >> endobj 1448 0 obj << /Names [(page.143) 1175 0 R (page.144) 1181 0 R (page.145) 1185 0 R (page.146) 1190 0 R (page.147) 1216 0 R (page.148) 1244 0 R] /Limits [(page.143) (page.148)] >> endobj 1449 0 obj << /Names [(page.149) 1249 0 R (page.15) 234 0 R (page.150) 1255 0 R (page.151) 1265 0 R (page.152) 1273 0 R (page.153) 1278 0 R] /Limits [(page.149) (page.153)] >> endobj 1450 0 obj << /Names [(page.154) 1283 0 R (page.155) 1287 0 R (page.156) 1296 0 R (page.157) 1309 0 R (page.158) 1318 0 R (page.159) 1326 0 R] /Limits [(page.154) (page.159)] >> endobj 1451 0 obj << /Names [(page.16) 241 0 R (page.160) 1331 0 R (page.161) 1338 0 R (page.162) 1345 0 R (page.163) 1352 0 R (page.164) 1359 0 R] /Limits [(page.16) (page.164)] >> endobj 1452 0 obj << /Names [(page.165) 1363 0 R (page.17) 248 0 R (page.18) 254 0 R (page.19) 266 0 R (page.2) 117 0 R (page.20) 279 0 R] /Limits [(page.165) (page.20)] >> endobj 1453 0 obj << /Names [(page.21) 288 0 R (page.22) 301 0 R (page.23) 305 0 R (page.24) 310 0 R (page.25) 323 0 R (page.26) 334 0 R] /Limits [(page.21) (page.26)] >> endobj 1454 0 obj << /Names [(page.27) 347 0 R (page.28) 352 0 R (page.29) 356 0 R (page.3) 126 0 R (page.30) 363 0 R (page.31) 372 0 R] /Limits [(page.27) (page.31)] >> endobj 1455 0 obj << /Names [(page.32) 377 0 R (page.33) 381 0 R (page.34) 385 0 R (page.35) 391 0 R (page.36) 399 0 R (page.37) 404 0 R] /Limits [(page.32) (page.37)] >> endobj 1456 0 obj << /Names [(page.38) 411 0 R (page.39) 415 0 R (page.4) 138 0 R (page.40) 419 0 R (page.41) 429 0 R (page.42) 440 0 R] /Limits [(page.38) (page.42)] >> endobj 1457 0 obj << /Names [(page.43) 449 0 R (page.44) 459 0 R (page.45) 474 0 R (page.46) 479 0 R (page.47) 488 0 R (page.48) 494 0 R] /Limits [(page.43) (page.48)] >> endobj 1458 0 obj << /Names [(page.49) 502 0 R (page.5) 157 0 R (page.50) 511 0 R (page.51) 517 0 R (page.52) 523 0 R (page.53) 529 0 R] /Limits [(page.49) (page.53)] >> endobj 1459 0 obj << /Names [(page.54) 536 0 R (page.55) 540 0 R (page.56) 549 0 R (page.57) 557 0 R (page.58) 562 0 R (page.59) 567 0 R] /Limits [(page.54) (page.59)] >> endobj 1460 0 obj << /Names [(page.6) 164 0 R (page.60) 571 0 R (page.61) 576 0 R (page.62) 584 0 R (page.63) 588 0 R (page.64) 592 0 R] /Limits [(page.6) (page.64)] >> endobj 1461 0 obj << /Names [(page.65) 596 0 R (page.66) 600 0 R (page.67) 611 0 R (page.68) 630 0 R (page.69) 635 0 R (page.7) 169 0 R] /Limits [(page.65) (page.7)] >> endobj 1462 0 obj << /Names [(page.70) 640 0 R (page.71) 647 0 R (page.72) 653 0 R (page.73) 657 0 R (page.74) 664 0 R (page.75) 669 0 R] /Limits [(page.70) (page.75)] >> endobj 1463 0 obj << /Names [(page.76) 677 0 R (page.77) 684 0 R (page.78) 691 0 R (page.79) 700 0 R (page.8) 180 0 R (page.80) 707 0 R] /Limits [(page.76) (page.80)] >> endobj 1464 0 obj << /Names [(page.81) 714 0 R (page.82) 724 0 R (page.83) 733 0 R (page.84) 741 0 R (page.85) 747 0 R (page.86) 754 0 R] /Limits [(page.81) (page.86)] >> endobj 1465 0 obj << /Names [(page.87) 758 0 R (page.88) 762 0 R (page.89) 767 0 R (page.9) 191 0 R (page.90) 771 0 R (page.91) 779 0 R] /Limits [(page.87) (page.91)] >> endobj 1466 0 obj << /Names [(page.92) 794 0 R (page.93) 804 0 R (page.94) 816 0 R (page.95) 823 0 R (page.96) 831 0 R (page.97) 839 0 R] /Limits [(page.92) (page.97)] >> endobj 1467 0 obj << /Names [(page.98) 845 0 R (page.99) 854 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected1) 614 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected2) 621 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected21) 648 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected22) 649 0 R] /Limits [(page.98) (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected22)] >> endobj 1468 0 obj << /Names [(pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 617 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 624 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata1) 613 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata2) 620 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}getdata3) 631 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init1) 612 0 R] /Limits [(pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init1)] >> endobj 1469 0 obj << /Names [(pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init2) 618 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}simple1) 615 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}simple2) 622 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}terms1) 616 0 R (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}terms2) 623 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args1) 1219 0 R] /Limits [(pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}init2) (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args1)] >> endobj 1470 0 obj << /Names [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args2) 1225 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args3) 1230 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args4) 1237 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args5) 1245 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw1) 1223 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw2) 1228 0 R] /Limits [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}common\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}args2) (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw2)] >> endobj 1471 0 obj << /Names [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw3) 1236 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx1) 1220 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx2) 1226 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}firstx3) 1234 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions1) 1222 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions2) 1227 0 R] /Limits [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}draw3) (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions2)] >> endobj 1472 0 obj << /Names [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions3) 1235 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup11) 1218 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup12) 1232 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup21) 1221 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup22) 1233 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks1) 1266 0 R] /Limits [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}functions3) (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks1)] >> endobj 1473 0 obj << /Names [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks2) 1267 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms1) 1217 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms2) 1224 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms3) 1229 0 R (plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}transform\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}ms4) 1231 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 856 0 R] /Limits [(plot\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}marks2) (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1)] >> endobj 1474 0 obj << /Names [(pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 859 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 857 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 860 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 855 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 858 0 R (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) 861 0 R] /Limits [(pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3)] >> endobj 1475 0 obj << /Names [(pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) 874 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}data1) 912 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}data2) 916 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dist1) 911 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dist2) 915 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 914 0 R] /Limits [(pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1)] >> endobj 1476 0 obj << /Names [(rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 918 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid1) 913 0 R (rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid2) 917 0 R (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}deriv1) 932 0 R (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}deriv2) 934 0 R (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid21) 933 0 R] /Limits [(rsr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) (rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid21)] >> endobj 1477 0 obj << /Names [(rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid22) 935 0 R (section*.1) 109 0 R (section*.2) 1364 0 R (section.1) 2 0 R (section.10) 78 0 R (section.2) 6 0 R] /Limits [(rtr\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}resid22) (section.2)] >> endobj 1478 0 obj << /Names [(section.3) 14 0 R (section.4) 22 0 R (section.5) 38 0 R (section.6) 42 0 R (section.7) 46 0 R (section.8) 54 0 R] /Limits [(section.3) (section.8)] >> endobj 1479 0 obj << /Names [(section.9) 74 0 R (subsection.2.1) 10 0 R (subsection.3.1) 18 0 R (subsection.4.1) 26 0 R (subsection.4.2) 30 0 R (subsection.4.3) 34 0 R] /Limits [(section.9) (subsection.4.3)] >> endobj 1480 0 obj << /Names [(subsection.7.1) 50 0 R (subsection.8.1) 58 0 R (subsection.8.2) 62 0 R (subsubsection.8.2.1) 66 0 R (subsubsection.8.2.2) 70 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 781 0 R] /Limits [(subsection.7.1) (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1)] >> endobj 1481 0 obj << /Names [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 786 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) 832 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute4) 833 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 783 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 788 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format1) 782 0 R] /Limits [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format1)] >> endobj 1482 0 obj << /Names [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format2) 787 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 780 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 785 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) 795 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) 807 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) 818 0 R] /Limits [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}format2) (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5)] >> endobj 1483 0 obj << /Names [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap1) 805 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap2) 806 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap3) 808 0 R (survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap4) 817 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 550 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 551 0 R] /Limits [(survexp\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}rmap1) (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2)] >> endobj 1484 0 obj << /Names [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) 558 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 462 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 465 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata21) 503 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata22) 504 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata23) 506 0 R] /Limits [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute3) (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}newdata23)] >> endobj 1485 0 obj << /Names [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result1) 461 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result2) 464 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result3) 518 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 460 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 463 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) 466 0 R] /Limits [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}result1) (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3)] >> endobj 1486 0 obj << /Names [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) 475 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) 480 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup6) 489 0 R (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup7) 490 0 R (survfitCI1) 993 0 R (survfitCI2) 994 0 R] /Limits [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) (survfitCI2)] >> endobj 1487 0 obj << /Names [(survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 1003 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 1005 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 1017 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 1023 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) 1045 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck1) 1015 0 R] /Limits [(survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck1)] >> endobj 1488 0 obj << /Names [(survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck2) 1021 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}startstop1) 1016 0 R (survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}startstop2) 1022 0 R (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}Surv1) 961 0 R (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}Surv2) 964 0 R (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula1) 960 0 R] /Limits [(survfitCI\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}idcheck2) (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula1)] >> endobj 1489 0 obj << /Names [(survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula2) 963 0 R (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript1) 959 0 R (survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript2) 962 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1) 1067 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) 1071 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P1) 1111 0 R] /Limits [(survfit\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}formula2) (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P1)] >> endobj 1490 0 obj << /Names [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P2) 1116 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}U1) 1110 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}U2) 1115 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}matrices1) 1097 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}matrices2) 1099 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update1) 1098 0 R] /Limits [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}P2) (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update1)] >> endobj 1491 0 obj << /Names [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update2) 1101 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare1) 1066 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare2) 1070 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}declare3) 1079 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix1) 1065 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix2) 1069 0 R] /Limits [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}update2) (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix2)] >> endobj 1492 0 obj << /Names [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) 1068 0 R (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return2) 1072 0 R (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript1) 1132 0 R (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}subscript2) 1134 0 R (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary1) 1131 0 R (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary2) 1133 0 R] /Limits [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) (survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary2)] >> endobj 1493 0 obj << /Names [(survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary3) 1152 0 R (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows1) 1150 0 R (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows2) 1153 0 R (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows3) 1164 0 R (survsum\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}findrows4) 1171 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin21) 1339 0 R] /Limits [(survfitms\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}summary3) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin21)] >> endobj 1494 0 obj << /Names [(tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin22) 1340 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin23) 1341 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin24) 1354 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addvar1) 1298 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addvar2) 1301 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) 1299 0 R] /Limits [(tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}addin22) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1)] >> endobj 1495 0 obj << /Names [(tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) 1302 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup1) 1297 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup2) 1300 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3) 1303 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) 1311 0 R (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5) 1322 0 R] /Limits [(tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5)] >> endobj 1496 0 obj << /Kids [1420 0 R 1421 0 R 1422 0 R 1423 0 R 1424 0 R 1425 0 R] /Limits [(Doc-Start) (agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish2)] >> endobj 1497 0 obj << /Kids [1426 0 R 1427 0 R 1428 0 R 1429 0 R 1430 0 R 1431 0 R] /Limits [(agfit4\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish3) (coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X1)] >> endobj 1498 0 obj << /Kids [1432 0 R 1433 0 R 1434 0 R 1435 0 R 1436 0 R 1437 0 R] /Limits [(coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}make\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}X2) (excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur1)] >> endobj 1499 0 obj << /Kids [1438 0 R 1439 0 R 1440 0 R 1441 0 R 1442 0 R 1443 0 R] /Limits [(excox\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}recur2) (page.120)] >> endobj 1500 0 obj << /Kids [1444 0 R 1445 0 R 1446 0 R 1447 0 R 1448 0 R 1449 0 R] /Limits [(page.121) (page.153)] >> endobj 1501 0 obj << /Kids [1450 0 R 1451 0 R 1452 0 R 1453 0 R 1454 0 R 1455 0 R] /Limits [(page.154) (page.37)] >> endobj 1502 0 obj << /Kids [1456 0 R 1457 0 R 1458 0 R 1459 0 R 1460 0 R 1461 0 R] /Limits [(page.38) (page.7)] >> endobj 1503 0 obj << /Kids [1462 0 R 1463 0 R 1464 0 R 1465 0 R 1466 0 R 1467 0 R] /Limits [(page.70) (pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}expected22)] >> endobj 1504 0 obj << /Kids [1468 0 R 1469 0 R 1470 0 R 1471 0 R 1472 0 R 1473 0 R] /Limits [(pcoxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}finish1) (pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute1)] >> endobj 1505 0 obj << /Kids [1474 0 R 1475 0 R 1476 0 R 1477 0 R 1478 0 R 1479 0 R] /Limits [(pyears\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}compute2) (subsection.4.3)] >> endobj 1506 0 obj << /Kids [1480 0 R 1481 0 R 1482 0 R 1483 0 R 1484 0 R 1485 0 R] /Limits [(subsection.7.1) (survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup3)] >> endobj 1507 0 obj << /Kids [1486 0 R 1487 0 R 1488 0 R 1489 0 R 1490 0 R 1491 0 R] /Limits [(survfit.coxph\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup4) (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix2)] >> endobj 1508 0 obj << /Kids [1492 0 R 1493 0 R 1494 0 R 1495 0 R] /Limits [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5)] >> endobj 1509 0 obj << /Kids [1496 0 R 1497 0 R 1498 0 R 1499 0 R 1500 0 R 1501 0 R] /Limits [(Doc-Start) (page.37)] >> endobj 1510 0 obj << /Kids [1502 0 R 1503 0 R 1504 0 R 1505 0 R 1506 0 R 1507 0 R] /Limits [(page.38) (survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}dmatrix2)] >> endobj 1511 0 obj << /Kids [1508 0 R] /Limits [(survfitci\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}return1) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5)] >> endobj 1512 0 obj << /Kids [1509 0 R 1510 0 R 1511 0 R] /Limits [(Doc-Start) (tmerge\\unhbox\040\\voidb@x\040\\kern\040\\z@\040\\char\040`\\discretionary\040{-}{}{}setup5)] >> endobj 1513 0 obj << /Dests 1512 0 R >> endobj 1514 0 obj << /Type /Catalog /Pages 1418 0 R /Outlines 1419 0 R /Names 1513 0 R /PageMode/UseOutlines /OpenAction 81 0 R >> endobj 1515 0 obj << /Author()/Title()/Subject()/Creator(LaTeX with hyperref package)/Producer(pdfTeX-1.40.14)/Keywords() /CreationDate (D:20150630070147-05'00') /ModDate (D:20150630070147-05'00') /Trapped /False /PTEX.Fullbanner (This is pdfTeX, Version 3.1415926-2.5-1.40.14 (TeX Live 2013/Debian) kpathsea version 6.1.1) >> endobj xref 0 1516 0000000000 65535 f 0000000015 00000 n 0000009110 00000 n 0000625222 00000 n 0000000060 00000 n 0000000090 00000 n 0000009169 00000 n 0000625100 00000 n 0000000135 00000 n 0000000163 00000 n 0000009228 00000 n 0000625041 00000 n 0000000213 00000 n 0000000237 00000 n 0000047314 00000 n 0000624916 00000 n 0000000283 00000 n 0000000326 00000 n 0000073586 00000 n 0000624855 00000 n 0000000377 00000 n 0000000414 00000 n 0000105327 00000 n 0000624729 00000 n 0000000460 00000 n 0000000489 00000 n 0000105387 00000 n 0000624655 00000 n 0000000540 00000 n 0000000577 00000 n 0000162654 00000 n 0000624568 00000 n 0000000628 00000 n 0000000665 00000 n 0000188025 00000 n 0000624494 00000 n 0000000716 00000 n 0000000746 00000 n 0000221827 00000 n 0000624405 00000 n 0000000792 00000 n 0000000828 00000 n 0000249135 00000 n 0000624316 00000 n 0000000874 00000 n 0000000905 00000 n 0000262904 00000 n 0000624190 00000 n 0000000951 00000 n 0000001001 00000 n 0000262964 00000 n 0000624129 00000 n 0000001052 00000 n 0000001080 00000 n 0000284473 00000 n 0000624003 00000 n 0000001126 00000 n 0000001160 00000 n 0000289341 00000 n 0000623929 00000 n 0000001211 00000 n 0000001242 00000 n 0000294055 00000 n 0000623818 00000 n 0000001293 00000 n 0000001327 00000 n 0000316370 00000 n 0000623744 00000 n 0000001383 00000 n 0000001408 00000 n 0000337604 00000 n 0000623670 00000 n 0000001464 00000 n 0000001504 00000 n 0000355887 00000 n 0000623581 00000 n 0000001550 00000 n 0000001593 00000 n 0000380554 00000 n 0000623505 00000 n 0000001640 00000 n 0000001665 00000 n 0000002681 00000 n 0000002947 00000 n 0000003097 00000 n 0000003247 00000 n 0000003402 00000 n 0000003552 00000 n 0000003707 00000 n 0000003857 00000 n 0000004013 00000 n 0000004169 00000 n 0000004325 00000 n 0000004475 00000 n 0000004624 00000 n 0000004774 00000 n 0000004930 00000 n 0000005079 00000 n 0000005234 00000 n 0000005389 00000 n 0000005549 00000 n 0000005708 00000 n 0000005859 00000 n 0000006190 00000 n 0000001713 00000 n 0000006011 00000 n 0000006070 00000 n 0000618393 00000 n 0000618248 00000 n 0000006130 00000 n 0000617230 00000 n 0000617084 00000 n 0000618104 00000 n 0000619404 00000 n 0000009288 00000 n 0000008938 00000 n 0000006313 00000 n 0000009050 00000 n 0000619112 00000 n 0000619258 00000 n 0000618824 00000 n 0000617521 00000 n 0000011000 00000 n 0000011358 00000 n 0000010868 00000 n 0000009424 00000 n 0000011237 00000 n 0000011297 00000 n 0000017714 00000 n 0000013020 00000 n 0000013258 00000 n 0000013572 00000 n 0000013806 00000 n 0000014040 00000 n 0000014276 00000 n 0000014938 00000 n 0000012848 00000 n 0000011468 00000 n 0000014511 00000 n 0000014571 00000 n 0000014633 00000 n 0000014694 00000 n 0000014755 00000 n 0000014816 00000 n 0000014877 00000 n 0000017776 00000 n 0000033809 00000 n 0000036250 00000 n 0000036312 00000 n 0000036374 00000 n 0000038404 00000 n 0000016942 00000 n 0000017179 00000 n 0000017416 00000 n 0000017838 00000 n 0000016794 00000 n 0000015048 00000 n 0000017654 00000 n 0000021404 00000 n 0000020716 00000 n 0000020953 00000 n 0000019179 00000 n 0000019007 00000 n 0000017948 00000 n 0000019119 00000 n 0000021191 00000 n 0000021527 00000 n 0000020568 00000 n 0000019289 00000 n 0000021344 00000 n 0000021466 00000 n 0000619521 00000 n 0000047252 00000 n 0000024045 00000 n 0000023067 00000 n 0000023390 00000 n 0000023712 00000 n 0000024107 00000 n 0000022919 00000 n 0000021637 00000 n 0000023864 00000 n 0000023924 00000 n 0000023984 00000 n 0000617668 00000 n 0000025953 00000 n 0000028080 00000 n 0000025247 00000 n 0000025570 00000 n 0000026015 00000 n 0000025107 00000 n 0000024230 00000 n 0000025893 00000 n 0000028142 00000 n 0000027054 00000 n 0000027376 00000 n 0000027698 00000 n 0000028202 00000 n 0000026906 00000 n 0000026125 00000 n 0000028020 00000 n 0000030798 00000 n 0000030418 00000 n 0000031043 00000 n 0000030286 00000 n 0000028312 00000 n 0000030738 00000 n 0000030860 00000 n 0000030921 00000 n 0000030982 00000 n 0000033123 00000 n 0000033436 00000 n 0000033871 00000 n 0000032983 00000 n 0000031179 00000 n 0000033749 00000 n 0000042320 00000 n 0000035489 00000 n 0000035722 00000 n 0000035955 00000 n 0000036436 00000 n 0000035341 00000 n 0000033994 00000 n 0000036190 00000 n 0000619639 00000 n 0000037876 00000 n 0000038110 00000 n 0000038466 00000 n 0000037736 00000 n 0000036546 00000 n 0000038344 00000 n 0000042197 00000 n 0000039797 00000 n 0000039625 00000 n 0000038576 00000 n 0000039737 00000 n 0000041428 00000 n 0000041661 00000 n 0000041823 00000 n 0000042381 00000 n 0000041280 00000 n 0000039894 00000 n 0000042137 00000 n 0000042259 00000 n 0000044401 00000 n 0000044180 00000 n 0000044463 00000 n 0000044048 00000 n 0000042491 00000 n 0000044341 00000 n 0000046954 00000 n 0000050334 00000 n 0000047374 00000 n 0000046822 00000 n 0000044573 00000 n 0000047192 00000 n 0000617959 00000 n 0000617376 00000 n 0000618681 00000 n 0000618538 00000 n 0000617814 00000 n 0000050488 00000 n 0000050721 00000 n 0000050954 00000 n 0000051556 00000 n 0000050178 00000 n 0000047571 00000 n 0000051187 00000 n 0000051247 00000 n 0000051309 00000 n 0000051371 00000 n 0000051433 00000 n 0000051494 00000 n 0000619757 00000 n 0000403780 00000 n 0000053401 00000 n 0000053107 00000 n 0000053463 00000 n 0000052975 00000 n 0000051704 00000 n 0000053341 00000 n 0000054942 00000 n 0000055176 00000 n 0000055411 00000 n 0000055645 00000 n 0000055878 00000 n 0000056475 00000 n 0000054778 00000 n 0000053573 00000 n 0000056111 00000 n 0000056171 00000 n 0000056231 00000 n 0000056292 00000 n 0000056352 00000 n 0000056413 00000 n 0000059040 00000 n 0000065484 00000 n 0000068653 00000 n 0000058746 00000 n 0000059102 00000 n 0000058614 00000 n 0000056585 00000 n 0000058980 00000 n 0000060578 00000 n 0000060406 00000 n 0000059225 00000 n 0000060518 00000 n 0000062604 00000 n 0000063144 00000 n 0000062472 00000 n 0000060675 00000 n 0000062837 00000 n 0000618969 00000 n 0000062897 00000 n 0000062959 00000 n 0000063021 00000 n 0000063083 00000 n 0000065546 00000 n 0000064726 00000 n 0000064957 00000 n 0000065191 00000 n 0000065607 00000 n 0000064578 00000 n 0000063317 00000 n 0000065424 00000 n 0000619875 00000 n 0000068715 00000 n 0000067180 00000 n 0000067415 00000 n 0000067647 00000 n 0000067881 00000 n 0000068116 00000 n 0000068839 00000 n 0000067016 00000 n 0000065717 00000 n 0000068350 00000 n 0000068410 00000 n 0000068470 00000 n 0000068531 00000 n 0000068592 00000 n 0000068777 00000 n 0000070888 00000 n 0000070124 00000 n 0000070359 00000 n 0000070594 00000 n 0000071012 00000 n 0000069976 00000 n 0000068949 00000 n 0000070828 00000 n 0000070950 00000 n 0000073646 00000 n 0000073414 00000 n 0000071122 00000 n 0000073526 00000 n 0000075148 00000 n 0000074976 00000 n 0000073819 00000 n 0000075088 00000 n 0000076969 00000 n 0000077203 00000 n 0000077438 00000 n 0000077915 00000 n 0000076821 00000 n 0000075258 00000 n 0000077671 00000 n 0000077731 00000 n 0000077792 00000 n 0000077853 00000 n 0000079864 00000 n 0000079570 00000 n 0000079925 00000 n 0000079438 00000 n 0000078025 00000 n 0000079804 00000 n 0000619993 00000 n 0000081155 00000 n 0000080983 00000 n 0000080035 00000 n 0000081095 00000 n 0000082316 00000 n 0000082144 00000 n 0000081252 00000 n 0000082256 00000 n 0000083482 00000 n 0000083310 00000 n 0000082413 00000 n 0000083422 00000 n 0000084870 00000 n 0000085104 00000 n 0000085522 00000 n 0000084730 00000 n 0000083579 00000 n 0000085340 00000 n 0000085400 00000 n 0000085461 00000 n 0000098691 00000 n 0000098815 00000 n 0000088975 00000 n 0000088803 00000 n 0000085632 00000 n 0000088915 00000 n 0000091561 00000 n 0000091916 00000 n 0000091429 00000 n 0000089097 00000 n 0000091795 00000 n 0000091855 00000 n 0000620111 00000 n 0000098753 00000 n 0000093019 00000 n 0000092847 00000 n 0000092039 00000 n 0000092959 00000 n 0000094198 00000 n 0000094026 00000 n 0000093116 00000 n 0000094138 00000 n 0000095247 00000 n 0000095075 00000 n 0000094295 00000 n 0000095187 00000 n 0000097223 00000 n 0000097456 00000 n 0000097689 00000 n 0000097924 00000 n 0000098159 00000 n 0000098395 00000 n 0000098877 00000 n 0000097051 00000 n 0000095344 00000 n 0000098631 00000 n 0000102795 00000 n 0000102857 00000 n 0000102919 00000 n 0000101613 00000 n 0000101846 00000 n 0000102081 00000 n 0000102317 00000 n 0000102981 00000 n 0000101457 00000 n 0000099012 00000 n 0000102552 00000 n 0000102612 00000 n 0000102673 00000 n 0000102734 00000 n 0000105265 00000 n 0000104970 00000 n 0000105447 00000 n 0000104838 00000 n 0000103168 00000 n 0000105205 00000 n 0000620229 00000 n 0000107751 00000 n 0000107993 00000 n 0000108236 00000 n 0000108479 00000 n 0000108720 00000 n 0000109267 00000 n 0000107587 00000 n 0000105583 00000 n 0000108962 00000 n 0000109022 00000 n 0000109083 00000 n 0000109144 00000 n 0000109205 00000 n 0000129041 00000 n 0000135262 00000 n 0000111802 00000 n 0000111259 00000 n 0000111500 00000 n 0000113127 00000 n 0000113368 00000 n 0000111864 00000 n 0000111119 00000 n 0000109390 00000 n 0000111742 00000 n 0000113670 00000 n 0000113732 00000 n 0000112987 00000 n 0000111987 00000 n 0000113610 00000 n 0000117748 00000 n 0000116722 00000 n 0000116963 00000 n 0000117205 00000 n 0000117446 00000 n 0000117872 00000 n 0000116566 00000 n 0000113842 00000 n 0000117688 00000 n 0000117810 00000 n 0000123750 00000 n 0000120673 00000 n 0000120501 00000 n 0000118008 00000 n 0000120613 00000 n 0000122716 00000 n 0000122957 00000 n 0000123202 00000 n 0000123446 00000 n 0000123935 00000 n 0000122560 00000 n 0000120783 00000 n 0000123690 00000 n 0000123812 00000 n 0000123873 00000 n 0000620347 00000 n 0000127052 00000 n 0000126748 00000 n 0000127114 00000 n 0000126616 00000 n 0000124045 00000 n 0000126992 00000 n 0000128497 00000 n 0000128739 00000 n 0000129103 00000 n 0000128357 00000 n 0000127237 00000 n 0000128981 00000 n 0000130927 00000 n 0000130625 00000 n 0000130989 00000 n 0000130493 00000 n 0000129213 00000 n 0000130867 00000 n 0000132907 00000 n 0000133145 00000 n 0000133565 00000 n 0000132767 00000 n 0000131099 00000 n 0000133381 00000 n 0000133441 00000 n 0000133503 00000 n 0000134960 00000 n 0000135324 00000 n 0000134828 00000 n 0000133675 00000 n 0000135202 00000 n 0000138152 00000 n 0000137797 00000 n 0000135434 00000 n 0000137909 00000 n 0000137969 00000 n 0000138030 00000 n 0000138091 00000 n 0000620465 00000 n 0000139377 00000 n 0000139742 00000 n 0000139245 00000 n 0000138275 00000 n 0000139621 00000 n 0000139681 00000 n 0000143344 00000 n 0000142798 00000 n 0000143041 00000 n 0000143406 00000 n 0000142658 00000 n 0000139852 00000 n 0000143284 00000 n 0000148339 00000 n 0000146169 00000 n 0000145997 00000 n 0000143579 00000 n 0000146109 00000 n 0000148036 00000 n 0000148401 00000 n 0000147904 00000 n 0000146305 00000 n 0000148279 00000 n 0000150559 00000 n 0000150387 00000 n 0000148524 00000 n 0000150499 00000 n 0000154093 00000 n 0000154364 00000 n 0000153961 00000 n 0000150694 00000 n 0000154243 00000 n 0000154303 00000 n 0000620583 00000 n 0000156086 00000 n 0000155878 00000 n 0000156148 00000 n 0000155746 00000 n 0000154524 00000 n 0000156026 00000 n 0000157565 00000 n 0000157393 00000 n 0000156258 00000 n 0000157505 00000 n 0000158968 00000 n 0000158796 00000 n 0000157675 00000 n 0000158908 00000 n 0000160298 00000 n 0000160126 00000 n 0000159078 00000 n 0000160238 00000 n 0000162713 00000 n 0000162482 00000 n 0000160434 00000 n 0000162594 00000 n 0000164543 00000 n 0000164777 00000 n 0000165014 00000 n 0000165252 00000 n 0000165487 00000 n 0000165721 00000 n 0000165957 00000 n 0000166680 00000 n 0000164363 00000 n 0000162862 00000 n 0000166189 00000 n 0000166249 00000 n 0000166310 00000 n 0000166371 00000 n 0000166433 00000 n 0000166495 00000 n 0000166557 00000 n 0000166618 00000 n 0000620701 00000 n 0000169792 00000 n 0000176410 00000 n 0000182614 00000 n 0000185484 00000 n 0000187963 00000 n 0000169259 00000 n 0000169495 00000 n 0000169854 00000 n 0000169119 00000 n 0000166803 00000 n 0000169732 00000 n 0000173300 00000 n 0000171168 00000 n 0000170996 00000 n 0000169977 00000 n 0000171108 00000 n 0000173004 00000 n 0000173362 00000 n 0000172872 00000 n 0000171265 00000 n 0000173240 00000 n 0000175637 00000 n 0000175874 00000 n 0000176112 00000 n 0000176595 00000 n 0000175489 00000 n 0000173472 00000 n 0000176350 00000 n 0000176472 00000 n 0000176533 00000 n 0000177868 00000 n 0000177696 00000 n 0000176718 00000 n 0000177808 00000 n 0000179812 00000 n 0000179578 00000 n 0000177965 00000 n 0000179690 00000 n 0000179750 00000 n 0000620819 00000 n 0000182319 00000 n 0000182676 00000 n 0000182187 00000 n 0000179960 00000 n 0000182554 00000 n 0000185007 00000 n 0000185546 00000 n 0000184875 00000 n 0000182825 00000 n 0000185241 00000 n 0000185301 00000 n 0000185362 00000 n 0000185423 00000 n 0000187668 00000 n 0000188085 00000 n 0000187536 00000 n 0000185682 00000 n 0000187903 00000 n 0000192733 00000 n 0000191742 00000 n 0000191893 00000 n 0000192226 00000 n 0000191602 00000 n 0000188234 00000 n 0000192044 00000 n 0000192104 00000 n 0000192165 00000 n 0000194884 00000 n 0000194946 00000 n 0000192621 00000 n 0000192361 00000 n 0000194824 00000 n 0000194336 00000 n 0000194481 00000 n 0000194581 00000 n 0000197724 00000 n 0000197873 00000 n 0000198145 00000 n 0000197584 00000 n 0000195045 00000 n 0000198024 00000 n 0000198084 00000 n 0000620937 00000 n 0000220164 00000 n 0000201357 00000 n 0000201185 00000 n 0000198293 00000 n 0000201297 00000 n 0000204099 00000 n 0000204254 00000 n 0000204408 00000 n 0000204870 00000 n 0000203951 00000 n 0000201529 00000 n 0000204562 00000 n 0000204622 00000 n 0000204684 00000 n 0000204746 00000 n 0000204808 00000 n 0000206663 00000 n 0000206817 00000 n 0000207238 00000 n 0000206523 00000 n 0000205017 00000 n 0000207056 00000 n 0000207116 00000 n 0000207177 00000 n 0000211583 00000 n 0000209814 00000 n 0000209333 00000 n 0000209876 00000 n 0000209201 00000 n 0000207374 00000 n 0000209572 00000 n 0000209632 00000 n 0000209692 00000 n 0000209753 00000 n 0000211368 00000 n 0000211645 00000 n 0000211236 00000 n 0000209999 00000 n 0000211523 00000 n 0000213364 00000 n 0000213602 00000 n 0000214024 00000 n 0000213224 00000 n 0000211755 00000 n 0000213841 00000 n 0000213901 00000 n 0000213962 00000 n 0000621055 00000 n 0000215264 00000 n 0000215092 00000 n 0000214147 00000 n 0000215204 00000 n 0000216654 00000 n 0000216482 00000 n 0000215361 00000 n 0000216594 00000 n 0000217977 00000 n 0000217805 00000 n 0000216764 00000 n 0000217917 00000 n 0000219955 00000 n 0000220226 00000 n 0000219823 00000 n 0000218074 00000 n 0000220104 00000 n 0000221886 00000 n 0000221655 00000 n 0000220349 00000 n 0000221767 00000 n 0000224344 00000 n 0000224580 00000 n 0000224817 00000 n 0000225054 00000 n 0000225595 00000 n 0000224188 00000 n 0000222009 00000 n 0000225291 00000 n 0000225351 00000 n 0000225412 00000 n 0000225473 00000 n 0000225534 00000 n 0000621173 00000 n 0000229016 00000 n 0000241199 00000 n 0000243674 00000 n 0000245470 00000 n 0000228485 00000 n 0000228720 00000 n 0000229078 00000 n 0000228345 00000 n 0000225718 00000 n 0000228956 00000 n 0000232281 00000 n 0000230793 00000 n 0000231027 00000 n 0000231262 00000 n 0000231582 00000 n 0000231901 00000 n 0000232467 00000 n 0000230629 00000 n 0000229188 00000 n 0000232221 00000 n 0000232343 00000 n 0000232405 00000 n 0000235996 00000 n 0000235934 00000 n 0000234764 00000 n 0000235083 00000 n 0000235403 00000 n 0000235638 00000 n 0000236058 00000 n 0000234608 00000 n 0000232577 00000 n 0000235874 00000 n 0000251220 00000 n 0000238378 00000 n 0000238083 00000 n 0000238440 00000 n 0000237951 00000 n 0000236181 00000 n 0000238318 00000 n 0000240189 00000 n 0000240426 00000 n 0000240664 00000 n 0000240901 00000 n 0000241322 00000 n 0000240033 00000 n 0000238550 00000 n 0000241139 00000 n 0000241261 00000 n 0000243612 00000 n 0000243079 00000 n 0000243316 00000 n 0000243735 00000 n 0000242939 00000 n 0000241432 00000 n 0000243552 00000 n 0000621291 00000 n 0000245174 00000 n 0000245532 00000 n 0000245042 00000 n 0000243845 00000 n 0000245410 00000 n 0000247902 00000 n 0000248136 00000 n 0000248372 00000 n 0000248607 00000 n 0000248841 00000 n 0000249440 00000 n 0000247738 00000 n 0000245642 00000 n 0000249075 00000 n 0000249195 00000 n 0000249256 00000 n 0000249317 00000 n 0000249378 00000 n 0000258467 00000 n 0000261293 00000 n 0000253963 00000 n 0000250840 00000 n 0000251282 00000 n 0000250708 00000 n 0000249576 00000 n 0000251160 00000 n 0000253201 00000 n 0000253435 00000 n 0000253669 00000 n 0000254087 00000 n 0000253053 00000 n 0000251392 00000 n 0000253903 00000 n 0000254025 00000 n 0000258171 00000 n 0000256420 00000 n 0000256248 00000 n 0000254197 00000 n 0000256360 00000 n 0000258529 00000 n 0000258039 00000 n 0000256530 00000 n 0000258407 00000 n 0000621409 00000 n 0000259717 00000 n 0000259545 00000 n 0000258639 00000 n 0000259657 00000 n 0000260998 00000 n 0000261355 00000 n 0000260866 00000 n 0000259814 00000 n 0000261233 00000 n 0000263024 00000 n 0000262732 00000 n 0000261465 00000 n 0000262844 00000 n 0000265187 00000 n 0000265015 00000 n 0000263160 00000 n 0000265127 00000 n 0000267276 00000 n 0000267507 00000 n 0000267738 00000 n 0000267970 00000 n 0000268203 00000 n 0000268799 00000 n 0000267112 00000 n 0000265310 00000 n 0000268433 00000 n 0000268493 00000 n 0000268554 00000 n 0000268615 00000 n 0000268676 00000 n 0000268737 00000 n 0000270696 00000 n 0000274504 00000 n 0000284411 00000 n 0000270406 00000 n 0000270758 00000 n 0000270274 00000 n 0000268909 00000 n 0000270636 00000 n 0000621527 00000 n 0000273748 00000 n 0000273979 00000 n 0000274211 00000 n 0000274687 00000 n 0000273600 00000 n 0000270868 00000 n 0000274444 00000 n 0000274566 00000 n 0000274626 00000 n 0000278205 00000 n 0000280177 00000 n 0000277914 00000 n 0000278267 00000 n 0000277782 00000 n 0000274885 00000 n 0000278145 00000 n 0000279885 00000 n 0000280239 00000 n 0000279753 00000 n 0000278440 00000 n 0000280117 00000 n 0000281245 00000 n 0000281073 00000 n 0000280362 00000 n 0000281185 00000 n 0000283167 00000 n 0000283399 00000 n 0000283639 00000 n 0000283877 00000 n 0000284112 00000 n 0000284781 00000 n 0000283003 00000 n 0000281342 00000 n 0000284351 00000 n 0000284533 00000 n 0000284595 00000 n 0000284657 00000 n 0000284719 00000 n 0000289401 00000 n 0000293993 00000 n 0000286166 00000 n 0000285994 00000 n 0000284904 00000 n 0000286106 00000 n 0000621645 00000 n 0000287269 00000 n 0000287097 00000 n 0000286263 00000 n 0000287209 00000 n 0000289044 00000 n 0000289463 00000 n 0000288912 00000 n 0000287366 00000 n 0000289281 00000 n 0000290947 00000 n 0000290775 00000 n 0000289586 00000 n 0000290887 00000 n 0000293699 00000 n 0000295669 00000 n 0000294111 00000 n 0000293567 00000 n 0000291044 00000 n 0000293933 00000 n 0000295943 00000 n 0000295537 00000 n 0000294260 00000 n 0000295822 00000 n 0000295882 00000 n 0000298215 00000 n 0000297700 00000 n 0000297939 00000 n 0000301888 00000 n 0000302128 00000 n 0000298278 00000 n 0000297557 00000 n 0000296053 00000 n 0000298091 00000 n 0000298153 00000 n 0000621763 00000 n 0000314698 00000 n 0000302370 00000 n 0000302609 00000 n 0000302765 00000 n 0000302922 00000 n 0000303078 00000 n 0000303678 00000 n 0000301699 00000 n 0000298388 00000 n 0000303235 00000 n 0000303297 00000 n 0000303361 00000 n 0000303424 00000 n 0000303487 00000 n 0000303551 00000 n 0000303614 00000 n 0000310345 00000 n 0000310409 00000 n 0000312981 00000 n 0000306340 00000 n 0000306162 00000 n 0000303826 00000 n 0000306278 00000 n 0000308083 00000 n 0000307905 00000 n 0000306489 00000 n 0000308021 00000 n 0000309802 00000 n 0000310041 00000 n 0000310473 00000 n 0000309656 00000 n 0000308181 00000 n 0000310283 00000 n 0000312201 00000 n 0000312440 00000 n 0000312680 00000 n 0000313107 00000 n 0000312046 00000 n 0000310584 00000 n 0000312919 00000 n 0000313043 00000 n 0000314396 00000 n 0000314762 00000 n 0000314259 00000 n 0000313231 00000 n 0000314636 00000 n 0000621888 00000 n 0000316431 00000 n 0000316192 00000 n 0000314873 00000 n 0000316308 00000 n 0000318347 00000 n 0000318588 00000 n 0000318828 00000 n 0000319068 00000 n 0000319307 00000 n 0000319925 00000 n 0000318174 00000 n 0000316542 00000 n 0000319547 00000 n 0000319609 00000 n 0000319672 00000 n 0000319735 00000 n 0000319798 00000 n 0000319861 00000 n 0000321967 00000 n 0000325968 00000 n 0000337540 00000 n 0000321424 00000 n 0000321664 00000 n 0000322031 00000 n 0000321278 00000 n 0000320049 00000 n 0000321905 00000 n 0000323582 00000 n 0000323280 00000 n 0000323646 00000 n 0000323143 00000 n 0000322142 00000 n 0000323520 00000 n 0000325666 00000 n 0000327911 00000 n 0000326032 00000 n 0000325529 00000 n 0000323757 00000 n 0000325906 00000 n 0000328240 00000 n 0000328567 00000 n 0000329146 00000 n 0000327756 00000 n 0000326143 00000 n 0000328895 00000 n 0000328957 00000 n 0000329020 00000 n 0000329082 00000 n 0000622013 00000 n 0000337476 00000 n 0000331039 00000 n 0000331360 00000 n 0000331682 00000 n 0000331839 00000 n 0000332602 00000 n 0000330875 00000 n 0000329270 00000 n 0000332160 00000 n 0000332222 00000 n 0000332286 00000 n 0000332349 00000 n 0000332412 00000 n 0000332475 00000 n 0000332538 00000 n 0000334360 00000 n 0000333978 00000 n 0000334424 00000 n 0000333841 00000 n 0000332726 00000 n 0000334298 00000 n 0000336366 00000 n 0000336692 00000 n 0000336931 00000 n 0000337172 00000 n 0000339302 00000 n 0000337792 00000 n 0000336202 00000 n 0000334548 00000 n 0000337414 00000 n 0000337665 00000 n 0000337728 00000 n 0000343117 00000 n 0000339606 00000 n 0000339670 00000 n 0000339165 00000 n 0000337929 00000 n 0000339544 00000 n 0000340818 00000 n 0000340640 00000 n 0000339781 00000 n 0000340756 00000 n 0000342337 00000 n 0000342576 00000 n 0000342816 00000 n 0000343244 00000 n 0000342182 00000 n 0000340916 00000 n 0000343055 00000 n 0000343181 00000 n 0000622138 00000 n 0000352868 00000 n 0000346790 00000 n 0000344545 00000 n 0000344367 00000 n 0000343355 00000 n 0000344483 00000 n 0000346249 00000 n 0000346488 00000 n 0000346854 00000 n 0000346103 00000 n 0000344643 00000 n 0000346728 00000 n 0000349199 00000 n 0000348660 00000 n 0000348898 00000 n 0000349263 00000 n 0000348514 00000 n 0000346965 00000 n 0000349137 00000 n 0000352932 00000 n 0000350914 00000 n 0000350736 00000 n 0000349374 00000 n 0000350852 00000 n 0000352326 00000 n 0000352566 00000 n 0000352996 00000 n 0000352180 00000 n 0000351012 00000 n 0000352806 00000 n 0000354369 00000 n 0000354191 00000 n 0000353107 00000 n 0000354307 00000 n 0000622263 00000 n 0000355948 00000 n 0000355709 00000 n 0000354467 00000 n 0000355825 00000 n 0000357438 00000 n 0000357757 00000 n 0000358075 00000 n 0000358393 00000 n 0000358628 00000 n 0000358946 00000 n 0000359184 00000 n 0000359417 00000 n 0000359736 00000 n 0000360054 00000 n 0000360373 00000 n 0000360691 00000 n 0000360926 00000 n 0000361161 00000 n 0000361399 00000 n 0000361637 00000 n 0000361870 00000 n 0000362103 00000 n 0000362423 00000 n 0000362742 00000 n 0000363061 00000 n 0000363379 00000 n 0000364713 00000 n 0000357112 00000 n 0000356072 00000 n 0000363698 00000 n 0000363760 00000 n 0000363823 00000 n 0000363886 00000 n 0000363949 00000 n 0000364012 00000 n 0000364075 00000 n 0000364138 00000 n 0000364201 00000 n 0000364265 00000 n 0000364329 00000 n 0000364393 00000 n 0000364457 00000 n 0000364521 00000 n 0000364585 00000 n 0000364649 00000 n 0000373628 00000 n 0000375302 00000 n 0000370939 00000 n 0000377393 00000 n 0000379210 00000 n 0000367324 00000 n 0000366307 00000 n 0000366625 00000 n 0000366944 00000 n 0000367451 00000 n 0000366152 00000 n 0000364824 00000 n 0000367262 00000 n 0000367387 00000 n 0000368766 00000 n 0000368588 00000 n 0000367562 00000 n 0000368704 00000 n 0000370391 00000 n 0000372484 00000 n 0000371003 00000 n 0000370254 00000 n 0000368864 00000 n 0000370625 00000 n 0000370687 00000 n 0000370750 00000 n 0000370813 00000 n 0000370876 00000 n 0000372803 00000 n 0000373121 00000 n 0000373692 00000 n 0000372329 00000 n 0000371114 00000 n 0000373439 00000 n 0000373501 00000 n 0000373564 00000 n 0000622388 00000 n 0000374922 00000 n 0000375366 00000 n 0000374785 00000 n 0000373803 00000 n 0000375240 00000 n 0000377094 00000 n 0000377457 00000 n 0000376957 00000 n 0000375477 00000 n 0000377331 00000 n 0000378916 00000 n 0000379274 00000 n 0000378779 00000 n 0000377581 00000 n 0000379148 00000 n 0000380615 00000 n 0000380376 00000 n 0000379385 00000 n 0000380492 00000 n 0000382760 00000 n 0000382996 00000 n 0000383233 00000 n 0000383470 00000 n 0000383705 00000 n 0000384255 00000 n 0000382587 00000 n 0000380726 00000 n 0000383940 00000 n 0000384002 00000 n 0000384065 00000 n 0000384128 00000 n 0000384191 00000 n 0000393144 00000 n 0000402831 00000 n 0000386417 00000 n 0000385885 00000 n 0000386120 00000 n 0000386481 00000 n 0000385739 00000 n 0000384366 00000 n 0000386355 00000 n 0000622513 00000 n 0000389402 00000 n 0000388635 00000 n 0000388870 00000 n 0000389105 00000 n 0000389719 00000 n 0000388480 00000 n 0000386592 00000 n 0000389340 00000 n 0000389466 00000 n 0000389529 00000 n 0000389592 00000 n 0000389655 00000 n 0000391056 00000 n 0000390878 00000 n 0000389830 00000 n 0000390994 00000 n 0000392846 00000 n 0000393208 00000 n 0000392709 00000 n 0000391154 00000 n 0000393082 00000 n 0000394932 00000 n 0000395169 00000 n 0000395405 00000 n 0000395831 00000 n 0000394777 00000 n 0000393319 00000 n 0000395642 00000 n 0000395704 00000 n 0000395767 00000 n 0000400695 00000 n 0000398266 00000 n 0000398088 00000 n 0000395942 00000 n 0000398204 00000 n 0000400162 00000 n 0000400397 00000 n 0000402233 00000 n 0000400759 00000 n 0000400016 00000 n 0000398364 00000 n 0000400633 00000 n 0000622638 00000 n 0000402767 00000 n 0000402469 00000 n 0000402895 00000 n 0000402087 00000 n 0000400870 00000 n 0000402705 00000 n 0000403842 00000 n 0000403539 00000 n 0000403006 00000 n 0000403655 00000 n 0000403717 00000 n 0000403966 00000 n 0000404306 00000 n 0000404435 00000 n 0000404467 00000 n 0000404615 00000 n 0000405055 00000 n 0000405466 00000 n 0000406102 00000 n 0000406430 00000 n 0000407067 00000 n 0000407463 00000 n 0000407801 00000 n 0000408488 00000 n 0000409132 00000 n 0000409770 00000 n 0000410201 00000 n 0000410511 00000 n 0000428377 00000 n 0000428747 00000 n 0000444264 00000 n 0000444618 00000 n 0000453149 00000 n 0000453517 00000 n 0000461995 00000 n 0000462311 00000 n 0000480272 00000 n 0000480649 00000 n 0000488983 00000 n 0000489215 00000 n 0000500660 00000 n 0000500952 00000 n 0000527189 00000 n 0000527807 00000 n 0000537321 00000 n 0000537587 00000 n 0000547552 00000 n 0000547804 00000 n 0000554975 00000 n 0000555201 00000 n 0000562993 00000 n 0000563256 00000 n 0000572431 00000 n 0000572809 00000 n 0000580364 00000 n 0000580618 00000 n 0000593825 00000 n 0000594097 00000 n 0000616454 00000 n 0000622736 00000 n 0000622856 00000 n 0000622976 00000 n 0000623096 00000 n 0000623220 00000 n 0000623328 00000 n 0000623429 00000 n 0000625294 00000 n 0000625468 00000 n 0000625639 00000 n 0000625812 00000 n 0000625986 00000 n 0000626152 00000 n 0000626834 00000 n 0000627685 00000 n 0000628192 00000 n 0000628647 00000 n 0000629613 00000 n 0000630910 00000 n 0000631922 00000 n 0000633005 00000 n 0000633631 00000 n 0000633845 00000 n 0000634063 00000 n 0000634910 00000 n 0000635753 00000 n 0000636599 00000 n 0000636970 00000 n 0000637147 00000 n 0000637324 00000 n 0000637503 00000 n 0000637681 00000 n 0000637866 00000 n 0000638049 00000 n 0000638234 00000 n 0000638417 00000 n 0000638602 00000 n 0000638785 00000 n 0000638970 00000 n 0000639152 00000 n 0000639325 00000 n 0000639496 00000 n 0000639666 00000 n 0000639837 00000 n 0000640007 00000 n 0000640178 00000 n 0000640348 00000 n 0000640519 00000 n 0000640688 00000 n 0000640857 00000 n 0000641028 00000 n 0000641198 00000 n 0000641369 00000 n 0000641539 00000 n 0000641710 00000 n 0000642324 00000 n 0000643182 00000 n 0000644200 00000 n 0000645463 00000 n 0000646317 00000 n 0000647676 00000 n 0000649038 00000 n 0000649895 00000 n 0000650726 00000 n 0000651557 00000 n 0000651905 00000 n 0000652086 00000 n 0000652297 00000 n 0000652691 00000 n 0000653562 00000 n 0000654423 00000 n 0000655726 00000 n 0000656651 00000 n 0000657562 00000 n 0000658202 00000 n 0000659096 00000 n 0000659980 00000 n 0000661032 00000 n 0000662601 00000 n 0000663670 00000 n 0000664568 00000 n 0000665453 00000 n 0000666318 00000 n 0000667177 00000 n 0000667381 00000 n 0000667747 00000 n 0000668111 00000 n 0000668312 00000 n 0000668430 00000 n 0000668547 00000 n 0000668662 00000 n 0000668867 00000 n 0000669156 00000 n 0000669366 00000 n 0000669581 00000 n 0000669879 00000 n 0000670151 00000 n 0000670269 00000 n 0000670475 00000 n 0000670720 00000 n 0000670896 00000 n 0000670936 00000 n 0000671067 00000 n trailer << /Size 1516 /Root 1514 0 R /Info 1515 0 R /ID [ ] >> startxref 671394 %%EOF survival/noweb/survfitCI.Rnw0000644000175100001440000004750612502643304015676 0ustar hornikusers\subsection{Competing risks} \newcommand{\Twid}{\mbox{\(\tt\sim\)}} The competing risks routine is very general, allowing subjects to enter or exit states multiple times. For this reason I prefer the label ``current prevalence'' estimate, since it estimates what fraction of the subjects are in any given state across time. The easiest way to understand the estimate is to consider first the case of no censoring. In that setting the estimate of $F_k(t) = 1-S_k(t)$ for all states is obtained from a simple table of the current state at time $t$ of the subjects, divided by $n$, the original sample size. When there is censoring the conceptually simple way to extend this is via the redistribute-to-the-right algorithm, which allocates the case weight for a censored subject evenly to all the others in the same state at the time of censoring. The literature refers to these as ``cumulative incidence'' curves, which is confusing since prevalence is not the integral of incidence, but the routine name survfitCI endures. The cannonical call is \begin{verbatim} fit <- survfit(Surv(time, status, type='mstate') ~ sex, data=mine) \end{verbatim} Optionally, there can be an id statement or cluster term to indicate a data set with multiple transitions per subject. A multi-state survival fit has a status variable with multiple levels, the first of which by default is censoring, and others indicating the type of transition that occured. The result will be a matrix of survival curves, one for each event type. Subjects are assumed to start in a "null" state, which is not tabulated for survival. To change this behavior, give all subjects some other initial state. The first part of the code is standard, parsing out options and checking the data. <>= survfitCI <- function(X, Y, weights, id, istate, type=c('kaplan-meier', 'fleming-harrington', 'fh2'), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified')){ method <- match.arg(type) # error <- match.arg(error) # if (error != "inf") # warning("Only the infinetesimal jackknife error is supported for CI curves") conf.type <- match.arg(conf.type) conf.lower<- match.arg(conf.lower) if (is.logical(conf.int)) { # A common error is for users to use "conf.int = FALSE" # it's illegal per documentation, but be kind if (!conf.int) conf.type <- "none" conf.int <- .95 } type <- attr(Y, "type") if (type !='mright' && type!='mcounting' && type != "right" && type != "counting") stop(paste("Cumulative incidence computation doesn't support \"", type, "\" survival data", sep='')) n <- nrow(Y) status <- Y[,ncol(Y)] ncurve <- length(levels(X)) state.names <- attr(Y, "states") if (missing(istate) || is.null(istate)) istate <- rep(0L, n) else if (is.factor(istate) || is.character(istate)) { # Match levels with the survival variable temp <- as.factor(istate) # append any starting states not found in Y, but remember that # if istate was a factor then not all its levels might appear appear <- (levels(istate))[unique(as.numeric(istate))] state.names <- unique(c(attr(Y, "states"), appear)) istate <- as.numeric(factor(as.character(istate), levels=state.names)) } else if (!is.numeric(istate) || any(istate != floor(istate))) stop("istate should be a vector of integers or a factor") if (length(id) ==0) id <- 1:n # these next two lines should be impossible, since istate came from the data frame if (length(istate) ==1) istate <- rep(istate,n) if (length(istate) !=n) stop ("wrong length for istate") states <- sort(unique(c(istate, 1:length(attr(Y, "states"))))) #list of all <> @ To make it easier to keep track of things in the computational kernel that does all the real work, we reset the states, initial state, and status vectors to all be integers 1, 2, \ldots, nstate, where ``1'' is the first state. The status vector will have values of 0 for censored. Per earlier discussion 1 will often be the unnamed initial state, which will later be dropped from the output. The statename vector is not modified. <>= if (any(states==0)) { state0 <- TRUE states <- states + 1 istate <- istate + 1 status <- ifelse(status==0, 0, status+1) } else state0 <- FALSE curves <- vector("list", ncurve) names(curves) <- levels(X) if (ncol(Y)==2) { # 1 transition per subject indx <- which(status == istate & status!=0) if (length(indx)) { warning("an observation transitions to it's starting state, transition ignored") status[indx] <- 0 } if (length(id) && any(duplicated(id))) stop("Cannot have duplicate id values with (time, status) data") # dummy entry time that is < any event time entry <- rep(min(-1, 2*min(Y[,1])-1), n) for (i in levels(X)) { indx <- which(X==i) # temp <- docurve1(entry[indx], Y[indx,1], status[indx], # istate[indx], weights[indx], states, # id[indx]) curves[[i]] <- docurve2(entry[indx], Y[indx,1], status[indx], istate[indx], weights[indx], states, id[indx], se.fit) } } else { <> <> } <> } @ In the multi-state case we can calculate the current prevalence vector $p(t)$ using the product-limit form \begin{align*} p(t) &= p(0)\prod_{s<=t} [I + dA(s)] \\ &= p(0) \prod_{s<=t} H(s) \end{align*} Where $p$ is a row vector and $H$ is the multi-state hazard matrix. At each event time we define the off diagonal elements of $H$ by $$H_{jk}(t) = \sum_i w_i dN_{ijk}(t)/ \sum w_iY_{ij}{t}$$ where $N_{ijk}$ counts the number of observed trasitions between state $j$ and state $k$ for subject $i$, $Y_{ij}(t)$ is 1 if subject $i$ is in state $j$ at time $t$, $w_i$ is the weight for subject $i$, and 0/0 is treated as 0. Row $j$ of $H(t)$ describes the fate of those subjects in state $j$, going from time $t$ to time $t+0$. The diagonal elements of $H$ are set so that each row of $H$ sums to 1 (everyone has to go somewhere). This formula collapses to the Kaplan-Meier in the simple case where $P(t)$ is a vector of length 2 with state 1 = alive and state 2 = dead. A robust variance for the product-limit estimate is based on the chain rule. Consider the $n$ by $k$ matrix of per subject influence values \begin{align} U_{ik}(t) &= \frac{\partial p_k(t)}{\partial w_i} \nonumber \\ &= \frac{\partial[ p(t-) H_{.k}(t)]}{\partial w_i} \label{ci0} \\ &= U_{i.}(t-) H_{.k}(t) + p(t-) \frac{\partial H_{.k}(t)}{\partial w_i} \label{ci1}\\ \frac{\partial H_{jk}(t)}{\partial w_i} &= \left\{ \begin{array}{ll} \left( dN_{ijk}(t) - Y_{ij}(t)H_{jk} \right)/n_j(t) & j \ne k \\ \left( -dN_{ij.}(t) - Y_{ij}(t)(H_{jj}-1) \right)/n_j(t) & j=k \\ \end{array} \right. \label{ci2} \end{align} where $H_{.k}$ is the $k$th column of $H$ amd $n_j(t)= \sum_i Y_{ij}(t)w_i$ is the weighted number of subjects in state $j$. Equation \eqref{ci0} replaces $p(t)$ with the last step of the compuation that created it. The next writes this out carefully using the chain rule, leading to an recursive equation. The first term of \eqref{ci1} is the formula for ordinary matrix multiplication. In equation \eqref{ci2} the derivative of $H$ with respect to subject $i$ will be a matrix which is non-zero only for the row corresponding to the current state of the subject. (I've skipped some intermediate steps in the derivation of \eqref{ci2}, ``left as an exercise for the reader''). The weighted sum of each column of $U$ must zero (if computed correctly) and the weighted sum of squares for each column will be the infinitesimal jackknife estimate of variance for the elements of $p$. The entire variance-covariance matrix for the states is $U'WU$ where $W$ is a diagonal matrix of weights, but we currently don't report that back. Note that this is for sampling weights. If one has real case weights, where an integer weight of 2 means 2 observations that were collapsed in to one row of data to save space, then the formula is $U'W^2U$. Case weights were somewhat common in my youth, due to small computer memory, but I haven't seen such data in 20 years. Below is the function for a single curve. For the status variable a value if 0 is ``no event''. One nuisance in the function is that we need to ensure the tapply command gives totals for all states, not just the ones present in the data --- a call using the \code{subset} argument might not have all the states --- which lead to factor commands. Another more confusing one is for multiple rows per subject data, where the cstate and U objects have only one row per subject; any given subject is only in one state at a time. This leads to indices of [[atrisk]] for the set of rows in the risk set but [[aindx]] for the subjects in the risk set, [[death]] for the rows that have an event as some given time and [[dindx]] for the corresponding subjects. <>= docurve1 <- function(entry, etime, status, istate, wt, states, id) { # # round off error can cause trouble: if two times are within machine # precsion then "unique(time)" and the "table" command may differ # solve this by using creating a factor ftime <- factor(etime) ntime <- length(levels(ftime)) # If someone has chosen to set the OutSep option to ',' (France) the simple # as.numeric(levels(ftime)) will fail timeset <- type.convert(levels(ftime), as.is=TRUE, dec=getOption("OutDec")) ftime <- as.numeric(ftime) nstate <- length(states) Pmat <- matrix(0., nrow= ntime, ncol=nstate) vP <- Pmat #variance A <- array(0., dim=c(nstate, nstate, ntime)) uid <- sort(unique(id)) U <- matrix(0., length(uid), nstate) #one row per subject P <- as.vector(tapply(wt, factor(istate, levels=states), sum) / sum(wt)) P <- Pmat[1,] <- ifelse(is.na(P), 0, P) cstate <- istate[match(uid, id)] #current state for each observation nrisk <- integer(ntime) #to be returned wrisk <- double(ntime) #weighted number at risk nevent <- table(ftime, status>0) for (i in 1:ntime) { atrisk <- (ftime >=i & timeset[i] > entry) nrisk[i] <- sum(atrisk) wrisk[i] <- sum(wt[atrisk]) tiedtime <- (ftime==i) if (nevent[i,2] ==0) { # all censored here Pmat[i,] <- P if (i>1) { A[,,i] <- A[,,i-1] vP[i,] <- vP[i-1,] } } else { # do real work # A bit of nuisance is to force tapply to give totals for all states aindx <- match(id[atrisk], uid) #the id pointer for those at risk ns <- as.vector(tapply(wt[atrisk], factor(cstate[aindx], levels=states),sum)) dead <- which(tiedtime & status >0) #the events at this time dindx <- match(id[dead], uid) nevent[i] <- length(dead) H <- tapply(wt[dead], list(factor(cstate[dindx], levels=states), factor(status[dead], levels=states)),sum)/ns H <- ifelse(is.na(H), 0, H) # H has NA for combinations with no representatives diag(H) <- 1- rowSums(H) H2 <- H diag(H2) <- diag(H2) -1 #version of H needed for U and A, rows sum to 0 if (i==1) A[,,1] <- H2 else A[,,i] <- A[,,i-1] + H2 newstate <- status[dead] # where the transitions go, will never be 0 oldstate <- cstate[dindx] # where they came from U <- U%*%H #first part of update U[aindx,] <- U[aindx,] - (P*H2/ns)[cstate[aindx], ] temp <- P[oldstate]/ns[oldstate] #the extra update for the events U[cbind(dindx, oldstate)] <- U[cbind(dindx, oldstate)] - temp U[cbind(dindx, newstate)] <- U[cbind(dindx, newstate)] + temp cstate[dindx] <- newstate P <- Pmat[i,] <- c(P %*% H) vP[i,] <- colSums((wt[match(uid, id)] *U)^2) } } list(time =as.vector(timeset), pmat=Pmat, std=sqrt(vP), n.event= as.vector(nevent[,2]), n.risk= as.vector(nrisk), w.risk=wrisk, cumhaz=A) } @ The above function was used to work through all of my test cases, but is too slow in large data sets. Rewrite it using underlying C-code, but retain the former one for debugging purposes. The C code appears at the end of this chapter. The setup for (start, stop] data is a bit more work. We want to ensure that a subject's weight is fixed, that they have a continuous period of observation, and that they don't transfer from a state to itself. The last is not strictly an error, so only warn. <>= if (missing(id) || is.null(id)) stop("the id argument is required for start:stop data") indx <- order(id, Y[,2]) #ordered event times within subject indx1 <- c(NA, indx) #a pair of lagged indices indx2 <- c(indx, NA) same <- (id[indx1] == id[indx2] & !is.na(indx1) & !is.na(indx2)) #indx1, indx2= same id? if (any(same & X[indx1] != X[indx2])) { who <- 1 + min(which(same & X[indx1] != X[indx2])) stop("subject is in two different groups, id ", (id[indx1])[who]) } if (any(same & Y[indx1,2] != Y[indx2,1])) { who <- 1 + min(which(same & Y[indx1,2] != Y[indx2,1])) stop("gap in follow-up, id ", (id[indx1])[who]) } if (any(Y[,1] == Y[,2])) stop("cannot have start time == stop time") if (any(same & Y[indx1,3] == Y[indx2,3] & Y[indx1,3] !=0)) { who <- 1 + min(which(same & Y[indx1,1] != Y[indx2,2])) warning("subject changes to the same state, id ", (id[indx1])[who]) } if (any(same & weights[indx1] != weights[indx2])) { who <- 1 + min(which(same & weights[indx1] != weights[indx2])) stop("subject changes case weights, id ", (id[indx1])[who]) } @ <>= # We only want to pay attention to the istate variable for the very first # observation of any given subject, but the program logic does better with # a full one. So construct one that will do this indx <- order(Y[,2]) uid <- unique(id) temp <- (istate[indx])[match(uid, id[indx])] #first istate for each subject istate <- temp[match(id, uid)] #replicate it to full length # Now to work for (i in levels(X)) { indx <- which(X==i) # temp <- docurve1(Y[indx,1], Y[indx,2], status[indx], # istate[indx], weights[indx], states, id[indx]) curves[[i]] <- docurve2(Y[indx,1], Y[indx,2], status[indx], istate[indx], weights[indx], states, id[indx], se.fit) } @ <>= # Turn the result into a survfit type object grabit <- function(clist, element) { temp <-(clist[[1]][[element]]) if (is.matrix(temp)) { nc <- ncol(temp) matrix(unlist(lapply(clist, function(x) t(x[[element]]))), byrow=T, ncol=nc) } else { xx <- as.vector(unlist(lapply(clist, function(x) x[element]))) if (class(temp)=="table") matrix(xx, byrow=T, ncol=length(temp)) else xx } } kfit <- list(n = as.vector(table(X)), time = grabit(curves, "time"), n.risk= grabit(curves, "n.risk"), n.event= grabit(curves, "n.event"), n.censor=grabit(curves, "n.censor"), prev = grabit(curves, "pmat"), prev0 = grabit(curves, "prev0")) nstate <- length(states) kfit$cumhaz <- array(unlist(lapply(curves, function(x) x$cumhaz)), dim=c(nstate, nstate, length(kfit$time))) if (length(curves) >1) kfit$strata <- unlist(lapply(curves, function(x) length(x$time))) if (se.fit) kfit$std.err <- grabit(curves, "std") # if state 0 was present, remove it if (state0) { kfit$prev <- kfit$prev[,-1] if (se.fit) kfit$std.err <- kfit$std.err[,-1] kfit$prev0 <- kfit$prev0[,-1] } @ Add the confidence bands. The idea is modeled on survfitKM but with the important differences that we are dealing with $P$ instead of $S$, and the ``modified lower limit'' logic does not apply. We make the assumption that $\log(1-P)$ will have better CI behavior than $P$, with standard error of ${rm se}(P)/(1-P)$. <>= # # Last bit: add in the confidence bands: # modeled on survfit.km, though for P instead of S # # if (se.fit) { std.err <- kfit$std.err zval <- qnorm(1- (1-conf.int)/2, 0,1) surv <- 1-kfit$prev if (conf.type=='plain') { temp <- zval* std.err kfit <- c(kfit, list(lower =pmax(kfit$prev-temp, 0), upper=pmin(kfit$prev+temp, 1), conf.type='plain', conf.int=conf.int)) } if (conf.type=='log') { #avoid some "log(0)" messages xx <- ifelse(kfit$prev==1, 1, 1- kfit$prev) temp1 <- ifelse(surv==0, NA, exp(log(xx) + zval* std.err/xx)) temp2 <- ifelse(surv==0, NA, exp(log(xx) - zval* std.err/xx)) kfit <- c(kfit, list(lower=pmax(1-temp1,0), upper= 1- temp2, conf.type='log', conf.int=conf.int)) } if (conf.type=='log-log') { who <- (surv==0 | surv==1) #special cases temp3 <- ifelse(surv==0, NA, 1) xx <- ifelse(who, .1,kfit$surv) #avoid some "log(0)" messages temp1 <- exp(-exp(log(-log(xx)) + zval*std.err/(xx*log(xx)))) temp1 <- ifelse(who, temp3, temp1) temp2 <- exp(-exp(log(-log(xx)) - zval*std.err/(xx*log(xx)))) temp2 <- ifelse(who, temp3, temp2) kfit <- c(kfit, list(lower=1-temp1, upper=1-temp2, conf.type='log-log', conf.int=conf.int)) } } kfit$states <- state.names kfit$type <- attr(Y, "type") kfit @ The updated docurve function is here <>= docurve2 <- function(entry, etime, status, istate, wt, states, id, se.fit) { # # round off error can cause trouble, if two times are within machine # precsion # solve this by using creating a factor ftime <- factor(c(entry,etime)) ltime <- levels(ftime) ftime <- matrix(as.integer(ftime), ncol=2) timeset <- as.numeric(ltime[sort(unique(ftime[,2]))]) #unique event times nstate <- length(states) uid <- sort(unique(id)) P <- as.vector(tapply(wt, factor(istate, levels=states), sum) / sum(wt)) P <- ifelse(is.na(P), 0, P) # initial probability distribution cstate <- istate[match(uid, id)] #current state for each observation storage.mode(wt) <- "double" # just in case someone had integer weights storage.mode(cstate) <- "integer" storage.mode(status) <- "integer" # C code has 0 based subscripts fit <- .Call(Csurvfitci, ftime, order(ftime[,1]) - 1L, order(ftime[,2]) - 1L, length(timeset), status, cstate - 1L, wt, match(id, uid) -1L, P, as.integer(se.fit)) prev0 <- table(factor(cstate, levels=states), exclude=NA)/length(cstate) if (se.fit) list(time=timeset, pmat=t(fit$p), std=sqrt(t(fit$var)), n.risk = colSums(fit$nrisk),n.event = fit$nevent, n.censor=fit$ncensor, prev0 = prev0, cumhaz=array(fit$cumhaz, dim=c(nstate,nstate, length(timeset)))) else list(time=timeset, pmat=t(fit$p), n.risk = colSums(fit$nrisk),n.event = fit$nevent, n.censor=fit$ncensor, prev0=prev0, cumhaz=array(fit$cumhaz, dim=c(nstate,nstate, length(timeset)))) } @ survival/noweb/plot.Rnw0000644000175100001440000003353412536335130014734 0ustar hornikusers\section{Plotting survival curves} I found a problem where plot.survfit, lines.survfit, and points.survfit sometimes did different things. This is due to copied code that later changed in one function but not another. Since they have so much code in common, this section of the noweb code consolodates them so as to restore order by using common code blocks. First define the top level routines. <>= plot.survfit<- function(x, conf.int, mark.time=TRUE, mark=3, col=1,lty=1, lwd=1, cex=1, log=FALSE, xscale=1, yscale=1, firstx=0, firsty=1, xmax, ymin=0, fun, xlab="", ylab="", xaxs='S', ...) { dotnames <- names(list(...)) if (any(dotnames=='type')) stop("The graphical argument 'type' is not allowed") <> if (missing(firsty) && !is.null(x$prev0)) firsty <- 1-x$prev0 <> <> <> <> <> plot.surv <- TRUE type <- 's' <> } lines.survfit <- function(x, type='s', mark=3, col=1, lty=1, lwd=1, cex=1, mark.time=TRUE, xscale=1, firstx=0, firsty=1, xmax, fun, conf.int=FALSE, ...) { xlog <- par("xlog") <> if (missing(firsty) && !is.null(x$prev0)) firsty <- 1-x$prev0 <> <> <> <> } points.survfit <- function(x, xscale=1, xmax, fun, ...) { <> firstx <- NA # flag used in the common args conf.int <- FALSE <> if (ncol(ssurv)==1) points(stime, ssurv, ...) else matpoints(stime, ssurv, ...) } @ Block of code to transform components of a [[survfitms]] object so that the standard plotting methods work. <>= if (inherits(x, "survfitms")) { x$surv <- 1- x$prev if (is.matrix(x$surv)) dimnames(x$surv) <- list(NULL, x$states) if (!is.null(x$lower)) { x$lower <- 1- x$lower x$upper <- 1- x$upper } if (missing(fun)) fun <- "event" } @ <>= ssurv <- as.matrix(x$surv) stime <- x$time if( !is.null(x$upper)) { supper <- as.matrix(x$upper) slower <- as.matrix(x$lower) } else { conf.int <- FALSE supper <- NULL #marker for later code } # Two cases where we don't put marks at the censoring times if (inherits(x, 'survexp') || inherits(x, 'survfit.coxph')) { if (missing(mark.time)) mark.time <- FALSE } # set up strata if (is.null(x$strata)) { nstrat <- 1 stemp <- rep(1, length(x$time)) # same length as stime } else { nstrat <- length(x$strata) stemp <- rep(1:nstrat, x$strata) # same length as stime } ncurve <- nstrat * ncol(ssurv) firsty <- matrix(firsty, nrow=nstrat, ncol=ncol(ssurv)) @ The xmax argument is used to prune back the survival curve to a small set of time points. This is a bit of bother since we have to do our own clipping of the data to prevent warning messages from the underlying plot routines. A further special case is when we are drawing lines and a curve got pruned so severely that only a horizontal segment from the curve start remains. In this case I need to reference the firsty arg. <>= if (!missing(xmax) && any(x$time>xmax)) { # prune back the survival curves # I need to replace x's over the limit with xmax, and y's over the # limit with either the prior y value or firsty keepx <- keepy <- NULL # lines to keep tempn <- table(stemp) offset <- cumsum(c(0, tempn)) for (i in 1:nstrat) { ttime <-stime[stemp==i] if (all(ttime <= xmax)) { keepx <- c(keepx, 1:tempn[i] + offset[i]) keepy <- c(keepy, 1:tempn[i] + offset[i]) } else { bad <- min((1:tempn[i])[ttime>xmax]) if (bad==1) { #lost them all if (!is.na(firstx)) { # and we are plotting lines keepy <- c(keepy, 1+offset[i]) ssurv[1+offset[i],] <- firsty[i,] } } else keepy<- c(keepy, c(1:(bad-1), bad-1) + offset[i]) keepx <- c(keepx, (1:bad)+offset[i]) stime[bad+offset[i]] <- xmax x$n.event[bad+offset[i]] <- 1 #don't plot a tick mark } } # ok, now actually prune it stime <- stime[keepx] stemp <- stemp[keepx] x$n.event <- x$n.event[keepx] if (!is.null(x$n.censor)) x$n.censor <- x$n.censor[keepx] ssurv <- ssurv[keepy,,drop=FALSE] if (!is.null(supper)) { supper <- supper[keepy,,drop=FALSE] slower <- slower[keepy,,drop=FALSE] } } stime <- stime/xscale #scaling is deferred until xmax processing is done if (!missing(fun)) { if (is.character(fun)) { tfun <- switch(fun, 'log' = function(x) x, 'event'=function(x) 1-x, 'cumhaz'=function(x) -log(x), 'cloglog'=function(x) log(-log(x)), 'pct' = function(x) x*100, 'logpct'= function(x) 100*x, #special case further below 'identity'= function(x) x, stop("Unrecognized function argument") ) } else if (is.function(fun)) tfun <- fun else stop("Invalid 'fun' argument") ssurv <- tfun(ssurv ) if (!is.null(supper)) { supper <- tfun(supper) slower <- tfun(slower) } firsty <- tfun(firsty) } @ The data structure for a survival plot does not include the first plot point. Those routines start their computation at the first endpoint, and leave it to here to decide on a starting location. The points routine doesn't have to deal with this nuisance. \begin{itemize} \item The initial time value [[firstx]] is the first of \begin{enumerate} \item a value given to [[firstx]] by the user \item [[start.time]], if present in the surv object \item if a logarithmic axis is specified, the smallest time >0 in the object \item the smaller of the minimum time or 0 \end{enumerate} \end{itemize} <>= if (missing(firstx)) { if (!is.null(x$start.time)) firstx <- x$start.time/xscale else { if (xlog) firstx <- min(stime[stime>0]) else firstx <- min(0, stime) } } # The default for plot and lines is to add confidence limits # if there is only one curve if (missing(conf.int)) conf.int <- (ncurve==1) if (is.logical(conf.int)) plot.surv <- TRUE else { temp <- match.arg(conf.int, c("both", "only", "none")) if (is.na(temp)) stop("invalid value for conf.int") if (temp=="none") conf.int <- FALSE else conf.int <- TRUE if (temp=="only") plot.surv <- FALSE else plot.surv <- TRUE } <> @ <>= # Marks are not placed on confidence bands mark <- rep(mark, length.out=ncurve) mcol <- rep(col, length.out=ncurve) if (is.numeric(mark.time)) mark.time <- sort(mark.time) # The actual number of curves is ncurve*3 if there are confidence bands # If the number of line types is 1 and lty is an integer, then use lty # for the curve and lty+1 for the CI # If the length(lty) <= length(ncurve), use the same color for curve and CI # otherwise assume the user knows what they are about and has given a full # vector of line types. # Colors and line widths work like line types, excluding the +1 rule. if (conf.int) { if (length(lty)==1 && is.numeric(lty)) lty <- rep(c(lty, lty+1, lty+1), ncurve) else if (length(lty) <= ncurve) lty <- rep(rep(lty, each=3), length.out=(ncurve*3)) else lty <- rep(lty, length.out= ncurve*3) if (length(col) <= ncurve) col <- rep(rep(col, each=3), length.out=3*ncurve) else col <- rep(col, length.out=3*ncurve) if (length(lwd) <= ncurve) lwd <- rep(rep(lwd, each=3), length.out=3*ncurve) else lwd <- rep(lwd, length.out=3*ncurve) } else { col <- rep(col, length.out=ncurve) lty <- rep(lty, length.out=ncurve) lwd <- rep(lwd, length.out=ncurve) } @ Here is the rest of the setup for the plot routine, mostly having to do with setting up axes. The [[xlog]] and [[ylog]] variables are internal reminders of the choice, and [[logax]] is what will be passed to the plot function <>= if (is.logical(log)) { ylog <- log xlog <- FALSE if (ylog) logax <- 'y' else logax <- "" } else { ylog <- (log=='y' || log=='xy') xlog <- (log=='x' || log=='xy') logax <- log } if (!missing(fun)) { if (is.character(fun)) { if (fun=='log'|| fun=='logpct') ylog <- TRUE if (fun=='cloglog') { xlog <- TRUE if (ylog) logax <- 'xy' else logax <- 'x' } } } # The special x axis style only applies when firstx is not given if (missing(xaxs) && (firstx!=0 || !missing(fun) || (missing(fun) && inherits(x, "survfitms")))) xaxs <- par("xaxs") #use the default @ <>= #axis setting parmaters that depend on the fun argument if (!missing(fun)) { ymin <- tfun(ymin) #lines routine doesn't have it } # Do axis range computations if (xaxs=='S') { #special x- axis style for survival curves xaxs <- 'i' #what S thinks tempx <- max(stime) * 1.04 } else tempx <- max(stime) tempx <- c(firstx, tempx, firstx) if (ylog) { tempy <- range(ssurv[is.finite(ssurv)& ssurv>0]) if (tempy[2]==1) tempy[2] <- .99 if (any(ssurv==0)) { tempy[1] <- tempy[1]*.8 ssurv[ssurv==0] <- tempy[1] if (!is.null(supper)) { supper[supper==0] <- tempy[1] slower[slower==0] <- tempy[1] } } tempy <- c(tempy, firsty) } else tempy <- range(ssurv, firsty, finite=TRUE, na.rm=TRUE) if (missing(fun)) { tempx <- c(tempx, firstx) if (!ylog) tempy <- c(tempy, ymin) } # # Draw the basic box # plot(range(tempx, finite=TRUE, na.rm=TRUE), range(tempy, finite=TRUE, na.rm=TRUE)*yscale, type='n', log=logax, xlab=xlab, ylab=ylab, xaxs=xaxs,...) if(yscale != 1) { if (ylog) par(usr =par("usr") -c(0, 0, log10(yscale), log10(yscale))) else par(usr =par("usr")/c(1, 1, yscale, yscale)) } @ The use of [[par(usr)]] just above is a bit sneaky. I want the lines and points routines to be able to add to the plot, \emph{without} passing them a global parameter that determines the y-scale or forcing the user to repeat it. Why didn't I use the same trick for xscale? Lack of foresight. And now there are hundreds of lines of code that have an xscale argument to lines() so I don't dare drop it. The next functions do the actual drawing. <>= # Create a step function, removing redundancies that sometimes occur in # curves with lots of censoring. dostep <- function(x,y) { keep <- is.finite(x) & is.finite(y) if (!any(keep)) return() #all points were infinite or NA if (!all(keep)) { # these won't plot anyway, so simplify (CI values are often NA) x <- x[keep] y <- y[keep] } n <- length(x) if (n==1) list(x=x, y=y) else if (n==2) list(x=x[c(1,2,2)], y=y[c(1,1,2)]) else { # replace verbose horizonal sequences like # (1, .2), (1.4, .2), (1.8, .2), (2.3, .2), (2.9, .2), (3, .1) # with (1, .2), (.3, .2),(3, .1). # They are slow, and can smear the looks of the line type. temp <- rle(y)$lengths drops <- 1 + cumsum(temp[-length(temp)]) # points where the curve drops #create a step function if (n %in% drops) { #the last point is a drop xrep <- c(x[1], rep(x[drops], each=2)) yrep <- rep(y[c(1,drops)], c(rep(2, length(drops)), 1)) } else { xrep <- c(x[1], rep(x[drops], each=2), x[n]) yrep <- c(rep(y[c(1,drops)], each=2)) } list(x=xrep, y=yrep) } } drawmark <- function(x, y, mark.time, censor, cex, ...) { if (!is.numeric(mark.time)) { xx <- x[censor] yy <- y[censor] } else { #interpolate xx <- mark.time yy <- approx(x, y, xx, method="constant", f=0)$y } points(xx, yy, cex=cex, ...) } @ The code to actually draw curves for the plot. The code to draw the lines and confidence bands. <>= c1 <- 1 # keeps track of the curve number c2 <- 1 # keeps track of the lty, col, etc xend <- yend <- double(ncurve) for (i in unique(stemp)) { #for each strata who <- which(stemp==i) censor <- if (is.null(x$n.censor)) (x$n.event[who] ==0) else (x$n.censor[who] >0) #places with a censoring xx <- c(firstx, stime[who]) censor <- c(FALSE, censor) #no mark at firstx for (j in 1:ncol(ssurv)) { yy <- c(firsty[i,j], ssurv[who,j]) if (plot.surv) { if (type=='s') lines(dostep(xx, yy), lty=lty[c2], col=col[c2], lwd=lwd[c2]) else lines(xx, yy, type=type, lty=lty[c2], col=col[c2], lwd=lwd[c2]) if (is.numeric(mark.time) || mark.time) drawmark(xx, yy, mark.time, censor, pch=mark[c1], col=mcol[c1], cex=cex) } xend[c1] <- max(xx) yend[c1] <- yy[length(yy)] c1 <- c1 +1 c2 <- c2 +1 if (conf.int) { if (type == 's') { lines(dostep(xx, c(firsty[i,j], slower[who,j])), lty=lty[c2], col=col[c2],lwd=lwd[c2]) c2 <- c2 +1 lines(dostep(xx, c(firsty[i,j], supper[who,j])), lty=lty[c2], col=col[c2], lwd= lwd[c2]) c2 <- c2 + 1 } else { lines(xx, c(firsty[i,j], slower[who,j]), lty=lty[c2], col=col[c2],lwd=lwd[c2], type=type) c2 <- c2 +1 lines(xx, c(firsty[i,j], supper[who,j]), lty=lty[c2], col=col[c2], lwd= lwd[c2], type= type) c2 <- c2 + 1 } } } } invisible(list(x=xend, y=yend)) @ survival/noweb/main.Rnw0000644000175100001440000000464712145504442014705 0ustar hornikusers\documentclass{article} \usepackage{noweb} \usepackage{amsmath} \usepackage{fancyvrb} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \newcommand{\myfig}[1]{\resizebox{\textwidth}{!} {\includegraphics{figures/#1.pdf}}} \newcommand{\code}[1]{\texttt{#1}} \noweboptions{breakcode} \title{Survival Package Functions} \author{Terry Therneau} \begin{document} \maketitle \tableofcontents \section{Introduction} \begin{quotation} Let us change or traditional attitude to the construction of programs. Instead of imagining that our main task is to instruct a \emph{computer} what to do, let us concentrate rather on explaining to \emph{humans} what we want the computer to do. (Donald E. Knuth, 1984). \end{quotation} This is the definition of a coding style called \emph{literate programming}. I first made use of it in the \emph{coxme} library and have become a full convert. For the survival library only selected objects are documented in this way; as I make updates and changes I am slowly converting the source code. The first motivation for this is to make the code easier for me, both to create and to maintain. As to maintinance, I have found that whenver I need to update code I spend a lot of time in the ``what was I doing in these x lines?'' stage. The code never has enough documentation, even for the author. (The survival library is already better than the majority of packages in R, whose comment level is abysmal. In the pre-noweb source code about 1 line in 6 has a comment, for the noweb document the documentation/code ratio is 2:1.) I also find it helps in creating new code to have the real documentation of intent --- formulas with integrals and such --- closely integrated. The second motivation is to leave code that is well enough explained that someone else can take it over. The source code is structured using \emph{noweb}, one of the simpler literate programming environments. The source code files look remakably like Sweave, and the .Rnw mode of emacs works perfectly for them. This is not too surprising since Sweave was also based on noweb. Sweave is not sufficient to process the files, however, since it has a different intention. The noweb.R file contains functions that can tangle the code (extract a given R function), but the creation of the pdf still requires the noweb exectuable itself. I am working towards correcting that. survival/noweb/residuals.survreg.Rnw0000644000175100001440000003543512145505666017460 0ustar hornikusers\section{Accelerated Failure Time models} The [[surveg]] function fits parametric failure time models. This includes accerated failure time models, the Weibull, log-normal, and log-logistic models. It also fits as well as censored linear regression; with left censoring this is referred to in economics \emph{Tobit} regression. \subsection{Residuals} The residuals for a [[survreg]] model are one of several types \begin{description} \item[response] residual [[y]] value on the scale of the original data \item[deviance] an approximate deviance residual. A very bad idea statistically, retained for the sake of backwards compatability. \item[dfbeta] a matrix with one row per observation and one column per parameter showing the approximate influence of each observation on the final parameter value \item[dfbetas] the dfbeta residuals scaled by the standard error of each coefficient \item[working] residuals on the scale of the linear predictor \item[ldcase] likelihood displacement wrt case weights \item[ldresp] likelihood displacement wrt response changes \item[ldshape] likelihood displacement wrt changes in shape \item[matrix] matrix of derivatives of the log-likelihood wrt paramters \end{description} The other parameters are \begin{description} \item[rsigma] whether the scale parameters should be included in the result for dfbeta results. I can think of no reason why one would not want them --- unless of course the scale was fixed by the user, in which case there is no parameter. \item[collapse] optional vector of subject identifiers. This is for the case where a subject has multiple observations in a data set, and one wants to have residuals per subject rather than residuals per observation. \item[weighted] whether the residuals should be multiplied by the case weights. The sum of weighted residuals will be zero. \end{description} The routine starts with standard stuff, checking arguments for validity and etc. The two cases of response or working residuals require a lot less computation. and are the most common calls, so they are taken care of first. <>= # $Id$ # # Residuals for survreg objects residuals.survreg <- function(object, type=c('response', 'deviance', 'dfbeta', 'dfbetas', 'working', 'ldcase', 'ldresp', 'ldshape', 'matrix'), rsigma =TRUE, collapse=FALSE, weighted=FALSE, ...) { type <-match.arg(type) n <- length(object$linear.predictors) Terms <- object$terms if(!inherits(Terms, "terms")) stop("invalid terms component of object") # If the variance wasn't estimated then it has no error if (nrow(object$var) == length(object$coefficients)) rsigma <- FALSE # If there was a cluster directive in the model statment then remove # it. It does not correspond to a coefficient, and would just confuse # things later in the code. cluster <- untangle.specials(Terms,"cluster")$terms if (length(cluster) >0 ) Terms <- Terms[-cluster] strata <- attr(Terms, 'specials')$strata coef <- object$coefficients intercept <- attr(Terms, "intercept") response <- attr(Terms, "response") weights <- object$weights if (is.null(weights)) weighted <- FALSE <> <> <> <> } @ First retrieve the distribution, which is used multiple times. The common case is a character string pointing to some element of [[survreg.distributions]], but the other is a user supplied list of the form contained there. Some distributions are defined as the transform of another in which case we need to set [[itrans]] and [[dtrans]] and follow the link, otherwise the transformation and its inverse are the identity. <>= if (is.character(object$dist)) dd <- survreg.distributions[[object$dist]] else dd <- object$dist if (is.null(dd$itrans)) { itrans <- dtrans <-function(x)x } else { itrans <- dd$itrans dtrans <- dd$dtrans } if (!is.null(dd$dist)) dd <- survreg.distributions[[dd$dist]] deviance <- dd$deviance dens <- dd$density @ The next task is to decide what data we need. The response is always needed, but is normally saved as a part of the model. If it is a transformed distribution such as the Weibull (a transform of the extreme value) the saved object [[y]] is the transformed data, so we need to replicate that part of the survreg() code. (Why did I even allow for y=F in survreg? Because I was mimicing the lm function --- oh the long, long consequences of a design decision.) The covariate matrix [[x]] will be needed for all but response, deviance, and working residuals. If the model included a strata() term then there will be multiple scales, and the strata variable needs to be recovered. The variable [[sigma]] is set to a scalar if there are no strata, but otherwise to a vector with [[n]] elements containing the appropriate scale for each subject. The leverage type residuals all need the second derivative matrix. If there was a [[cluster]] statement in the model this will be found in [[naive.var]], otherwise in the [[var]] component. <>= if (is.null(object$naive.var)) vv <- object$var else vv <- object$naive.var need.x <- is.na(match(type, c('response', 'deviance', 'working'))) if (is.null(object$y) || !is.null(strata) || (need.x & is.null(object[['x']]))) mf <- model.frame(object) y <- object$y if (is.null(y)) { y <- model.extract(mf, 'response') if (!is.null(dd$trans)) { tranfun <- dd$trans exactsurv <- y[,ncol(y)] ==1 if (any(exactsurv)) logcorrect <-sum(log(dd$dtrans(y[exactsurv,1]))) if (type=='interval') { if (any(y[,3]==3)) y <- cbind(tranfun(y[,1:2]), y[,3]) else y <- cbind(tranfun(y[,1]), y[,3]) } else if (type=='left') y <- cbind(tranfun(y[,1]), 2-y[,2]) else y <- cbind(tranfun(y[,1]), y[,2]) } else { if (type=='left') y[,2] <- 2- y[,2] else if (type=='interval' && all(y[,3]<3)) y <- y[,c(1,3)] } } if (!is.null(strata)) { temp <- untangle.specials(Terms, 'strata', 1) Terms2 <- Terms[-temp$terms] if (length(temp$vars)==1) strata.keep <- mf[[temp$vars]] else strata.keep <- strata(mf[,temp$vars], shortlabel=TRUE) strata <- as.numeric(strata.keep) nstrata <- max(strata) sigma <- object$scale[strata] } else { Terms2 <- Terms nstrata <- 1 sigma <- object$scale } if (need.x) { x <- object[['x']] #don't grab xlevels component if (is.null(x)) x <- model.matrix(Terms2, mf, contrasts.arg=object$contrasts) } @ The most common residual is type response, which requires almost no more work, for the others we need to create the matrix of derivatives before proceeding. We use the [[center]] component from the deviance function for the distribution, which returns the data point [[y]] itself for an exact, left, or right censored observation, and an appropriate midpoint for interval censored ones. <>= if (type=='response') { yhat0 <- deviance(y, sigma, object$parms) rr <- itrans(yhat0$center) - itrans(object$linear.predictor) } else { <> <> } @ The matrix of derviatives is used in all of the other cases. The starting point is the [[density]] function of the distribtion which return a matrix with columns of $F(x)$, $1-F(x)$, $f(x)$, $f'(x)/f(x)$ and $f''(x)/f(x)$. %' The matrix type residual contains columns for each of $$ L_i \quad \frac{\partial L_i}{\partial \eta_i} \quad \frac{\partial^2 L_i}{\partial \eta_i^2} \quad \frac{\partial L_i}{\partial \log(\sigma)} \quad \frac{\partial L_i}{\partial \log(\sigma)^2} \quad \frac{\partial^2 L_i}{\partial \eta \partial\log(\sigma)} $$ where $L_i$ is the contribution to the log-likelihood from each individual. Note that if there are multiple scales, i.e. a strata() term in the model, then terms 3--6 are the derivatives for that subject with respect to their \emph{particular} scale factor; derivatives with respect to all the other scales are zero for that subject. The log-likelihood can be written as \begin{align*} L &= \sum_{exact}\left[ \log(f(z_i)) -\log(\sigma_i) \right] + \sum_{censored} \log \left( \int_{z_i^l}^{z_i^u} f(u)du \right) \\ &\equiv \sum_{exact}\left[g_1(z_i) -\log(\sigma_i) \right] + \sum_{censored} \log(g_2(z_i^l, z_i^u)) \\ z_i &= (y_i - \eta_i)/ \sigma_i \end{align*} For the interval censored observations we have a $z$ defined at both the lower and upper endpoints. The linear predictor is $\eta = X\beta$. The derivatives are shown below. Note that $f(-\infty) = f(\infty) = F(-\infty)=0$, $F(\infty)=1$, $z^u = \infty$ for a right censored observation and $z^l = -\infty$ for a left censored one. \begin{align*} \frac{\partial g_1}{\partial \eta} &= - \frac{1}{\sigma} \left[\frac{f'(z)}{f(z)} \right] \\ %' \frac{\partial g_2}{\partial \eta} &= - \frac{1}{\sigma} \left[ \frac{f(z^u) - f(z^l)}{F(z^u) - F(z^l)} \right] \\ \frac{\partial^2 g_1}{\partial \eta^2} &= \frac{1}{\sigma^2} \left[ \frac{f''(z)}{f(z)} \right] - (\partial g_1 / \partial \eta)^2 \\ \frac{\partial^2 g_2}{\partial \eta^2} &= \frac{1}{\sigma^2} \left[ \frac{f'(z^u) - f'(z^l)}{F(z^u) - F(z^l)} \right] - (\partial g_2 / \partial \eta)^2 \\ \frac{\partial g_1}{\partial \log\sigma} && - \left[ \frac{zf'(z)}{f(z)} \right] \\ \frac{\partial g_2}{\partial \log\sigma} &= - \left[ \frac{z^uf(z^u) - z^lf(z^l)}{F(z^u) - F(z^l)} \right] \\ \frac{\partial^2 g_1}{\partial (\log\sigma)^2} &=& \left[ \frac{z^2 f''(z) + zf'(z)}{f(z)} \right] - (\partial g_1 / \partial \log\sigma)^2 \\ \frac{\partial^2 g_2}{\partial (\log\sigma)^2} &= \left[ \frac{(z^u)^2 f'(z^u) - (z^l)^2f'(z_l) } {F(z^u) - F(z^l)} \right] - \partial g_1 /\partial \log\sigma(1+\partial g_1 / \partial \log\sigma) \\ \frac{\partial^2 g_1}{\partial \eta \partial \log\sigma} &= \frac{zf''(z)}{\sigma f(z)} -\partial g_1/\partial \eta (1 + \partial g_1/\partial \log\sigma) \\ \frac{\partial^2 g_2}{\partial \eta \partial \log\sigma} &= \frac{z^uf'(z^u) - z^lf'(z^l)}{\sigma [F(z^u) - F(z^l)]} -\partial g_2/\partial \eta (1 + \partial g_2/\partial \log\sigma) \\ \end{align*} In the code [[z]] is the relevant point for exact, left, or right censored data, and [[z2]] the upper endpoint for an interval censored one. The variable [[tdenom]] contains the denominator for each subject (which is the same for all derivatives for that subject). For an interval censored observation we try to avoid numeric cancellation by using the appropriate tail of the distribution. For instance with $(z^l, z^u) = (12,15)$ the value of $F(x)$ will be very near 1 and it is better to subtract two upper tail values $(1-F)$ than two lower tail ones $F$. <>= status <- y[,ncol(y)] eta <- object$linear.predictors z <- (y[,1] - eta)/sigma dmat <- dens(z, object$parms) dtemp<- dmat[,3] * dmat[,4] #f' if (any(status==3)) { z2 <- (y[,2] - eta)/sigma dmat2 <- dens(z2, object$parms) } else { dmat2 <- dmat #dummy values z2 <- 0 } tdenom <- ((status==0) * dmat[,2]) + #right censored ((status==1) * 1 ) + #exact ((status==2) * dmat[,1]) + #left ((status==3) * ifelse(z>0, dmat[,2]-dmat2[,2], dmat2[,1] - dmat[,1])) #interval g <- log(ifelse(status==1, dmat[,3]/sigma, tdenom)) #loglik tdenom <- 1/tdenom dg <- -(tdenom/sigma) *(((status==0) * (0-dmat[,3])) + #dg/ eta ((status==1) * dmat[,4]) + ((status==2) * dmat[,3]) + ((status==3) * (dmat2[,3]- dmat[,3]))) ddg <- (tdenom/sigma^2) *(((status==0) * (0- dtemp)) + #ddg/eta^2 ((status==1) * dmat[,5]) + ((status==2) * dtemp) + ((status==3) * (dmat2[,3]*dmat2[,4] - dtemp))) ds <- ifelse(status<3, dg * sigma * z, tdenom*(z2*dmat2[,3] - z*dmat[,3])) dds <- ifelse(status<3, ddg* (sigma*z)^2, tdenom*(z2*z2*dmat2[,3]*dmat2[,4] - z * z*dmat[,3] * dmat[,4])) dsg <- ifelse(status<3, ddg* sigma*z, tdenom *(z2*dmat2[,3]*dmat2[,4] - z*dtemp)) deriv <- cbind(g, dg, ddg=ddg- dg^2, ds = ifelse(status==1, ds-1, ds), dds=dds - ds*(1+ds), dsg=dsg - dg*(1+ds)) @ Now, we can calcultate the actual residuals case by case. For the dfbetas there will be one column per coefficient, so if there are strata column 4 of the deriv matrix needs to be \emph{un}collapsed into a matrix with nstrata columns. The same manipulation is needed for the ld residuals. <>= if (type=='deviance') { yhat0 <- deviance(y, sigma, object$parms) rr <- (-1)*deriv[,2]/deriv[,3] #working residuals rr <- sign(rr)* sqrt(2*(yhat0$loglik - deriv[,1])) } else if (type=='working') rr <- (-1)*deriv[,2]/deriv[,3] else if (type=='dfbeta' || type== 'dfbetas' || type=='ldcase') { score <- deriv[,2] * x # score residuals if (rsigma) { if (nstrata > 1) { d4 <- matrix(0., nrow=n, ncol=nstrata) d4[cbind(1:n, strata)] <- deriv[,4] score <- cbind(score, d4) } else score <- cbind(score, deriv[,4]) } rr <- score %*% vv if (type=='dfbetas') rr <- rr %*% diag(1/sqrt(diag(vv))) if (type=='ldcase') rr<- rowSums(rr*score) } else if (type=='ldresp') { rscore <- deriv[,3] * (x * sigma) if (rsigma) { if (nstrata >1) { d6 <- matrix(0., nrow=n, ncol=nstrata) d6[cbind(1:n, strata)] <- deriv[,6]*sigma rscore <- cbind(rscore, d6) } else rscore <- cbind(rscore, deriv[,6] * sigma) } temp <- rscore %*% vv rr <- rowSums(rscore * temp) } else if (type=='ldshape') { sscore <- deriv[,6] *x if (rsigma) { if (nstrata >1) { d5 <- matrix(0., nrow=n, ncol=nstrata) d5[cbind(1:n, strata)] <- deriv[,5] sscore <- cbind(sscore, d5) } else sscore <- cbind(sscore, deriv[,5]) } temp <- sscore %*% vv rr <- rowSums(sscore * temp) } else { #type = matrix rr <- deriv } @ Finally the two optional steps of adding case weights and collapsing over subject id. <>= #case weights if (weighted) rr <- rr * weights #Expand out the missing values in the result if (!is.null(object$na.action)) { rr <- naresid(object$na.action, rr) if (is.matrix(rr)) n <- nrow(rr) else n <- length(rr) } # Collapse if desired if (!missing(collapse)) { if (length(collapse) !=n) stop("Wrong length for 'collapse'") rr <- drop(rowsum(rr, collapse)) } rr @ survival/noweb/concordance.Rnw0000644000175100001440000006463312544502656016250 0ustar hornikusers\subsection{Concordance} The concordance statistic is gaining popularity as a measure of goodness-of-fit in survival models. Consider all pairs of subjects with $(r_i, r_j)$ as the two risk scores for each pair and $(s_i, s_j)$ the corresponding survival times. The c-statistic is defined by dividing these sets into four groups. \begin{itemize} \item Concordant pairs: for a Cox model this will be pairs where a shorter survival is paired with a larger risk score, e.g. $r_i>r_j$ and $s_i < s_j$ \item Discordant pairs: the lower risk score has a shorter survival \item Tied pairs: there are three common choices \begin{itemize} \item Kendall's tau: any pair where $r_i=r_j$ or $s_i = s_j$ is considered tied. \item AUC: pairs with $r_i=r_j$ are tied; those with $s_i=s_j$ are considered incomparable. This is the definition of the AUC in logisitic regression, and has become the most common choice for Cox models as well. \item Somer's D: All ties are treated as incomparable. \end{itemize} \item Incomparable pairs: For survival this always includes pairs where the survival times cannot be ranked with certainty. For instance $s_i$ is censored at time 10 and $s_j$ is an event (or censor) at time 20. Subject $i$ may or may not survive longer than subject $j$. Note that if $s_i$ is censored at time 10 and $s_j$ is an event at time 10 then $s_i > s_j$. Add onto this those ties that are treated as incomparable.\\ Observations that are in different strata are also incomparable, since the Cox model only compares within strata. \end{itemize} Then the concordance statistic is defined as $(C + T/2)/(C + D + T)$. The denominator is the number of comparable pairs. The program creates 4 variables, which are the number of concordant pairs, discordant, tied on time, and tied on $x$ but not on time. The default concordance is based on the AUC definition, but all 4 values are reported back so that a user can recreate the others if desired. The primary compuational questions is how to do this efficiently, i.e., better that the naive $O(n^2)$ algorithm that loops across all $n(n-1)/2$ possible pairs. There are two key ideas. \begin{enumerate} \item Rearrange the counting so that we do it by death times. For each death we count the number of other subjects in the risk set whose score is higher, lower, or tied and add it into the totals. This also neatly solves the question of time-dependent covariates. \item To count the number higher and lower we need to rank the subjects in the risk set by their scores $r_i$. This can be done in $O(\log n)$ time if the data is kept in a binary tree. \end{enumerate} \begin{figure} \myfig{balance} \caption{A balanced tree of 13 nodes.} \label{treefig} \end{figure} Figure \ref{treefig} shows a balanced binary tree containing 13 risk scores. For each node the left child and all its descendants have a smaller value than the parent, the right child and all its descendents have a larger value. Each node in figure \ref{treefig} is also annotated with the total weight of observations in that node and the weight for all its children. (These weights are not shown on the graph. For this example assume that all weights are 1.). Assume that the tree shown represents all of the subjects still alive at the time a particular subject ``Smith'' expires, and that Smith has the risk score 38. The concordant pairs are all of those with a risk score greater than 38, which can be found by traversing the tree from the top down, adding the (parent - child) value each time we branch left (5-3 at the 44 node), with a last addition of the right hand sum of weights when we find the node with Smith's value (1). %' There are 3 concordant and 12-3=9 discordant pairs. This takes a little less than $\log_2(n)$ steps on average, as compared to an average of $n/2$ for the naive method. The difference can matter when $n$ is large since this traversal must be done for each event. (In the code below we start at Smith's node and walk up.) %' The classic way to store trees is as a linked list. There are several algorithms for adding and subtracting nodes from a tree while maintaining the balance (red-black trees, AA trees, etc) but we take a different approach. Since we need to deal with case weights in the model and we know all the risk score at the outset, the full set of risk scores is organised into a tree at the beginning and node counts are changed to zero as observations are removed. If we index the nodes of the tree as 1 for the top, 2--3 for the next horizontal row, 4--7 for the next, \ldots then the parent-child traversal becomes particularly easy. The parent of node $i$ is $i/2$ (integer arithmetic) and the children of node $i$ are $2i$ and $2i +1$. In C code the indices start at 0 and the children are $2i+1$ and $2i+2$ and the parent is $(i-1)/2$. The following bit of code returns the indices of a sorted list when placed into such a tree. The basic idea is that the rows of the tree start at indices 1, 2, 4, \ldots. For the above tree, the last row will contains the 1st, 3rd, \ldots, 11th smallest ranks. The next row above contains half of the ranks \emph{not yet assigned}, and etc to the top of the tree. There is some care to make sure the result is an integer. <>= btree <- function(n) { ranks <- rep(0L, n) #will be overwritten yet.to.do <- 1:n depth <- floor(logb(n,2)) start <- as.integer(2^depth) lastrow.length <- 1+n-start indx <- seq(1L, by=2L, length= lastrow.length) ranks[yet.to.do[indx]] <- start + 0:(length(indx)-1L) yet.to.do <- yet.to.do[-indx] while (start >1) { start <- as.integer(start/2) indx <- seq(1L, by=2L, length=start) ranks[yet.to.do[indx]] <- start + 0:(start-1L) yet.to.do <- yet.to.do[-indx] } ranks } @ Referring again to figure \ref{treefig}, [[btree(13)]] yields the vector [[8 4 9 2 10 5 11 1 12 6 13 3 7]] meaning that the smallest element will be in position 8 of the tree, the next smallest in position 4, etc. Here is a shorter recursive version. It knows the form of trees with 1, 2, or 3 nodes; and builds the others from them. The maximum depth of recursion is $\log_2(n) -1$. It is more clever but a bit slower. (Not that it matters as both take less than 5 seconds for a million elements.) <>= btree <- function(n) { tfun <- function(n, id, power) { if (n==1) id else if (n==2) c(2L *id, id) else if (n==3) c(2L*id, id, 2L*id +1L) else { nleft <- if (n== power*2) power else min(power-1, n-power/2) c(tfun(nleft, 2L *id, power/2), id, tfun(n-(nleft+1), 2L*id +1L, power/2)) } } tfun(n, 1L, 2^(floor(logb(n-1,2)))) } @ A second question is how to compute the variance of the result. The insight used here is to consider a Cox model with time dependent covariates, where the covariate $x$ at each death time has been transformed into ${\rm rank}(x)$. It is easy to show that the Cox score statistic contribution at each death is $(D-C)/2$ where $C$ and $D$ are the number of concordant and discordant pairs contributed at that death time (for a Cox fit using the Breslow approximation). The contribution to the variance of the score statistic is $V(t) =\sum (r_i - \overline{r})^2 /n$, the $r_i$ being the ranks at that time point and $n$ the number at risk. How can we update this sum using an update formula? First remember the identity \begin{equation*} \sum w_i(x_i - \overline{x})^2 = \sum w_i(x_i-c)^2 - \sum w_i(c - \overline{x})^2 \end{equation*} true for any set of values $x$ and centering constant $c$. For weighted data define the rank of an observation with risk score $r_k$ as \begin{equation*} {\rm rank} = \sum_{r_ik} w_i(r_i - \mu_n)^2 - \sum_{i>k} w_i(r_i - \mu_g)^2 &= (\sum_{i>k} w_i) [(\mu_u -\mu_n)^2 - ((\mu_u-w_k) - \mu_g)^2] \nonumber \\ &= (\sum_{i>k} w_i) (\mu_n + z - 2\mu_u)(\mu_n -z) \label{upper1} \\ &= (\sum_{i>k} w_i) (\mu_n+z - 2\mu_u) (-w_k/2) \label{upper}\\ z&\equiv \mu_g+ w_k \nonumber \end{align} For items of tied rank, their rank increases by the same amount as the overall mean, and so their contribution to the total SS is unchanged. The final part of the update step is to add in the SS contributed by the new observation. An observation is removed from the tree whenver the current time becomes less than the (start, stop] interval of the datum. The ranks for observations of lower risk are unchanged by the removal so equation \eqref{lower1} applies just as before, but with the new mean smaller than the old so the last term in equation \eqref{lower} changes sign. For the observations of higher risk both the mean and the ranks change by $w_k$ and equation \eqref{upper1} holds but with $z=\mu_0- w_k$. We can now define the C-routine that does the bulk of the work. First we give the outline shell of the code and then discuss the parts one by one. This routine is for ordinary survival data, and will be called once per stratum. Input variables are \begin{description} \item[n] the number of observations \item[y] matrix containing the time and status, data is sorted by ascending time, with deaths preceding censorings. \item[indx] the tree node at which this observation's risk score resides %' \item[wt] case weight for the observation \item[sum] scratch space, weights for each node of the tree: 3 values are for the node, all left children, and all right children \item[count] the returned counts of concordant, discordant, tied on x, tied on time, and the variance \end{description} <>= #include "survS.h" SEXP concordance1(SEXP y, SEXP wt2, SEXP indx2, SEXP ntree2) { int i, j, k, index; int child, parent; int n, ntree; double *time, *status; double *twt, *nwt, *count; double vss, myrank, wsum1, wsum2, wsum3; /*sum of wts below, tied, above*/ double lmean, umean, oldmean, newmean; double ndeath; /* weighted number of deaths at this point */ SEXP count2; double *wt; int *indx; n = nrows(y); ntree = asInteger(ntree2); wt = REAL(wt2); indx = INTEGER(indx2); time = REAL(y); status = time + n; PROTECT(count2 = allocVector(REALSXP, 5)); count = REAL(count2); /* count5 contains the information matrix */ twt = (double *) R_alloc(2*ntree, sizeof(double)); nwt = twt + ntree; for (i=0; i< 2*ntree; i++) twt[i] =0.0; for (i=0; i<5; i++) count[i]=0.0; vss=0; <> UNPROTECT(1); return(count2); } @ The key part of our computation is to update the vectors of weights. We don't actually pass the risk score values $r$ into the routine, %' it is enough for each observation to point to the appropriate tree node. The tree contains the weights for everyone whose survival is larger than the time currently under review, so starts with all weights equal to zero. For any pair of observations $i,j$ we need to add [[wt[i]*wt[j]]] to the appropriate count. Starting at the largest time (which is sorted last), walk through the tree. \begin{itemize} \item If it is a death time, we need to process all the deaths tied at this time. \begin{enumerate} \item Add [[wt[i] * wt[j]]] to the tied-on-time total, for all pairs $i,j$ of tied times. \item The addition to tied-on-r will be the weight of this observation times the sum of weights for all others with the same risk score and a a greater time, i.e., the weight found at [[indx[i]]] in the tree. \item Similarly for those with smaller or larger risk scores. First add in the children of this node. The left child will be smaller risk scores (and longer times) adding to the concordant pairs, the right child discordant. Then walk up the tree to the root. At each step up we add in data for the 'not me' branch. If we were the right branch (even number node) of a parent then when moving up we add in the left branch counts, and vice-versa. \end{enumerate} \item Now add this set of subject weights into the tree. The weight for a node is [[nwt]] and for the node and all its children is [[twt]]. \end{itemize} <>= for (i=n-1; i>=0; ) { ndeath =0; if (status[i]==1) { /* process all tied deaths at this point */ for (j=i; j>=0 && status[j]==1 && time[j]==time[i]; j--) { ndeath += wt[j]; index = indx[j]; for (k=i; k>j; k--) count[3] += wt[j]*wt[k]; /* tied on time */ count[2] += wt[j] * nwt[index]; /* tied on x */ child = (2*index) +1; /* left child */ if (child < ntree) count[0] += wt[j] * twt[child]; /*left children */ child++; if (child < ntree) count[1] += wt[j] * twt[child]; /*right children */ while (index >0) { /* walk up the tree */ parent = (index-1)/2; if (index & 1) /* I am the left child */ count[1] += wt[j] * (twt[parent] - twt[index]); else count[0] += wt[j] * (twt[parent] - twt[index]); index = parent; } } } else j = i-1; /* Add the weights for these obs into the tree and update variance*/ for (; i>j; i--) { wsum1=0; oldmean = twt[0]/2; index = indx[i]; nwt[index] += wt[i]; twt[index] += wt[i]; wsum2 = nwt[index]; child = 2*index +1; /* left child */ if (child < ntree) wsum1 += twt[child]; while (index >0) { parent = (index-1)/2; twt[parent] += wt[i]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); /* sum of weights above */ lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean - oldmean); vss += wsum3*(newmean+ oldmean+ wt[i]- 2*umean) *(oldmean-newmean); vss += wt[i]* (myrank -newmean)*(myrank -newmean); } count[4] += ndeath * vss/twt[0]; } @ The code for [start, stop) data is quite similar. As in the agreg routines there are two sort indices, the first indexes the data by stop time, longest to earliest, and the second by start time. The [[y]] variable now has three columns. <>= SEXP concordance2(SEXP y, SEXP wt2, SEXP indx2, SEXP ntree2, SEXP sortstop, SEXP sortstart) { int i, j, k, index; int child, parent; int n, ntree; int istart, iptr, jptr; double *time1, *time2, *status, dtime; double *twt, *nwt, *count; int *sort1, *sort2; double vss, myrank; double wsum1, wsum2, wsum3; /*sum of wts below, tied, above*/ double lmean, umean, oldmean, newmean; double ndeath; SEXP count2; double *wt; int *indx; n = nrows(y); ntree = asInteger(ntree2); wt = REAL(wt2); indx = INTEGER(indx2); sort2 = INTEGER(sortstop); sort1 = INTEGER(sortstart); time1 = REAL(y); time2 = time1 + n; status= time2 + n; PROTECT(count2 = allocVector(REALSXP, 5)); count = REAL(count2); twt = (double *) R_alloc(2*ntree, sizeof(double)); nwt = twt + ntree; for (i=0; i< 2*ntree; i++) twt[i] =0.0; for (i=0; i<5; i++) count[i]=0.0; vss =0; <> UNPROTECT(1); return(count2); } @ The processing changes in 2 ways \begin{itemize} \item The loops go from $0$ to $n-1$ instead of $n-1$ to 0. We need to use [[sort1[i]]] instead of [[i]] as the subscript for the time2 and wt vectors. (The sort vectors go backwards in time.) This happens enough that we use a temporary variables [[iptr]] and [[jptr]] to avoid the double subscript. \item As we move from the longest time to the shortest observations are added into the tree of weights whenever we encounter their stop time. This is just as before. Weights now also need to be removed from the tree whenever we encounter an observation's start time. %' It is convenient ``catch up'' on this second task whenever we encounter a death. \end{itemize} <>= istart = 0; /* where we are with start times */ for (i=0; i= dtime; istart++) { wsum1 =0; oldmean = twt[0]/2; jptr = sort1[istart]; index = indx[jptr]; nwt[index] -= wt[jptr]; twt[index] -= wt[jptr]; wsum2 = nwt[index]; child = 2*index +1; /* left child */ if (child < ntree) wsum1 += twt[child]; while (index >0) { parent = (index-1)/2; twt[parent] -= wt[jptr]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean-oldmean); oldmean -= wt[jptr]; /* the z in equations above */ vss += wsum3*(newmean+ oldmean -2*umean) * (newmean-oldmean); vss -= wt[jptr]* (myrank -newmean)*(myrank -newmean); } /* Process deaths */ for (j=i; j 0) { /* walk up the tree */ parent = (index-1)/2; if (index &1) /* I am the left child */ count[1] += wt[jptr] * (twt[parent] - twt[index]); else count[0] += wt[jptr] * (twt[parent] - twt[index]); index = parent; } } } else j = i+1; /* Add the weights for these obs into the tree and compute variance */ for (; i0) { parent = (index-1)/2; twt[parent] += wt[iptr]; if (!(index&1)) /* I am a right child */ wsum1 += (twt[parent] - twt[index]); index=parent; } wsum3 = twt[0] - (wsum1 + wsum2); lmean = wsum1/2; umean = wsum1 + wsum2 + wsum3/2; /* new upper mean */ newmean = twt[0]/2; myrank = wsum1 + wsum2/2; vss += wsum1*(newmean+ oldmean - 2*lmean) * (newmean-oldmean); vss += wsum3*(newmean+ oldmean +wt[iptr] - 2*umean) * (oldmean-newmean); vss += wt[iptr]* (myrank -newmean)*(myrank -newmean); } count[4] += ndeath * vss/twt[0]; } @ One last wrinkle is tied risk scores: they are all set to point to the same node of the tree. Here is the main routine. <>= survConcordance <- function(formula, data, weights, subset, na.action) { Call <- match.call() # save a copy of of the call, as documentation m <- match.call(expand.dots=FALSE) m[[1]] <- as.name("model.frame") m$formula <- if(missing(data)) terms(formula, "strata") else terms(formula, "strata", data=data) m <- eval(m, sys.parent()) Terms <- attr(m, 'terms') Y <- model.extract(m, "response") if (!inherits(Y, "Surv")) { if (is.numeric(Y) && is.vector(Y)) Y <- Surv(Y) else stop("left hand side of the formula must be a numeric vector or a surival") } n <- nrow(Y) wt <- model.extract(m, 'weights') offset<- attr(Terms, "offset") if (length(offset)>0) stop("Offset terms not allowed") stemp <- untangle.specials(Terms, 'strata') if (length(stemp$vars)) { if (length(stemp$vars)==1) strat <- m[[stemp$vars]] else strat <- strata(m[,stemp$vars], shortlabel=TRUE) Terms <- Terms[-stemp$terms] } else strat <- NULL x <- model.matrix(Terms, m)[,-1, drop=FALSE] #remove the intercept if (ncol(x) > 1) stop("Only one predictor variable allowed") count <- survConcordance.fit(Y, x, strat, wt) if (is.null(strat)) { concordance <- (count[1] + count[3]/2)/sum(count[1:3]) std.err <- count[5]/(2* sum(count[1:3])) } else { temp <- colSums(count) concordance <- (temp[1] + temp[3]/2)/ sum(temp[1:3]) std.err <- temp[5]/(2*sum(temp[1:3])) } fit <- list(concordance= concordance, stats=count, n=n, std.err=std.err, call=Call) na.action <- attr(m, "na.action") if (length(na.action)) fit$na.action <- na.action oldClass(fit) <- 'survConcordance' fit } print.survConcordance <- function(x, ...) { if(!is.null(cl <- x$call)) { cat("Call:\n") dput(cl) cat("\n") } omit <- x$na.action if(length(omit)) cat(" n=", x$n, " (", naprint(omit), ")\n", sep = "") else cat(" n=", x$n, "\n") cat("Concordance= ", format(x$concordance), " se= ", format(x$std.err), '\n', sep='') print(x$stats) invisible(x) } @ This part of the compuation is a separate function, since it is also called by the coxph routines. Although we are very careful to create integers and/or doubles for the arguments to .Call I still wrap them in the appropriate as.xxx construction: ``belt and suspenders''. Also, referring to the the mathematics many paragraphs ago, the C routine returns the variance of $(C-D)/2$ and we return the standard deviation of $(C-D)$. If this routine is called with all the x values identical, then $C$ and $D$ will both be zero, but the calculated variance of $C-D$ can be a nonzero tiny number due to round off error. Since this can cause a warning message from the sqrt function we check and correct this. <>= survConcordance.fit <- function(y, x, strata, weight) { <> docount <- function(stime, risk, wts) { if (attr(stime, 'type') == 'right') { ord <- order(stime[,1], -stime[,2]) ux <- sort(unique(risk)) n2 <- length(ux) index <- btree(n2)[match(risk[ord], ux)] - 1L .Call(Cconcordance1, stime[ord,], as.double(wts[ord]), as.integer(index), as.integer(length(ux))) } else if (attr(stime, 'type') == "counting") { sort.stop <- order(-stime[,2], stime[,3]) sort.start <- order(-stime[,1]) ux <- sort(unique(risk)) n2 <- length(ux) index <- btree(n2)[match(risk, ux)] - 1L .Call(Cconcordance2, stime, as.double(wts), as.integer(index), as.integer(length(ux)), as.integer(sort.stop-1L), as.integer(sort.start-1L)) } else stop("Invalid survival type for concordance") } if (missing(weight) || length(weight)==0) weight <- rep(1.0, length(x)) storage.mode(y) <- "double" if (missing(strata) || length(strata)==0) { count <- docount(y, x, weight) if (count[1]==0 && count[2]==0) count[5]<-0 else count[5] <- 2*sqrt(count[5]) names(count) <- c("concordant", "discordant", "tied.risk", "tied.time", "std(c-d)") } else { strata <- as.factor(strata) ustrat <- levels(strata)[table(strata) >0] #some strata may have 0 obs count <- matrix(0., nrow=length(ustrat), ncol=5) for (i in 1:length(ustrat)) { keep <- which(strata == ustrat[i]) count[i,] <- docount(y[keep,,drop=F], x[keep], weight[keep]) } count[,5] <- 2*sqrt(ifelse(count[,1]+count[,2]==0, 0, count[,5])) dimnames(count) <- list(ustrat, c("concordant", "discordant", "tied.risk", "tied.time", "std(c-d)")) } count } @ survival/noweb/refer.bib0000644000175100001440000016526511773346736015076 0ustar hornikusers@string{annals= {Annals of Stat.}} @string{applstat= {Applied Stat.}} @string{bioj = {Biometrical J.}} @string{biok = {Biometrika}} @string{commstata = {Comm. Stat. Theory Methods}} @string{biom = {Biometrics}} @string{jap = {J. Applied Probability}} @string{jasa = {J. Amer. Stat. Assoc.}} @string{jrssa= {J. Royal Stat. Soc. A}} @string{jrssb= {J. Royal Stat. Soc. B}} @string{jrssc= {J. Royal Stat. Soc. C}} @string{jscs = {J Stat. Comput. Simul.}} @string{lifetime = {Lifetime Data Analysis}} @string{NEJM = {New England J. Medicine}} @string{scand = {Scandinavian J. Stat.}} @string{statmed= {Stat. in Medicine}} @string{statsci = {Stat. Science}} @book{Andersen93, author={Andersen, P. K. and Borgan, {\O}. and Gill, R. D. and Keiding, N.}, title= {Statistical Models Based on Counting Processes}, publisher={Springer-Verlag}, address={New York}, year= {1993} } @article{Andersen00, author= {Andersen, P. K. and Esbjerg, S. and S{\o}rensen, T.I.A.} , title= {Multi-state models for bleeding episodes and morality in lever cirrhosis}, year= {2000}, journal=statmed, volume={19}, pages={587--599} } @article{Anderson82, author= {J. R. Anderson and L. Bernstein and M. C. Pike}, year= {1982}, title= {Approximate confidence intervals for probabilities of survival and quantiles in life-table analysis}, journal=biom}, volume={38}, pages= {407--416}, } @book{Anderson58, author= {V. E Anderson and H. O. Goodman and S. Reed}, title= {Variables Related to Human Breast Cancer}, year = 1958, publisher={University of Minnesota Press}, address={Minneapolis} } @article{Barlow88, author= { Barlow, W. E. and Prentice, R. L.}, year = {1988}, title= {Residuals for relative risk regression}, journal= {Biometrika}, volume={75}, pages={65--74} } @article{Barlow94, author= {Barlow, W. E.}, year ={1994}, title= {Robust variance estimation for the case-cohort design}, journal= {Biometrics}, volume={50}, pages= {1064--1072} } @article{Bartolucci77, author = {Bartolucci, A. A. and Fraser, M. D.}, year = 1977, title = {Comparative step-up and composite tests for selecting prognostic indicators associated with survival}, journal = {Biometrical J.}, volume = 19, pages = {437-448} } @book{Becker84, author={Becker, R. A. and Chambers, J. M.}, title= {S: {A}n Interactive Environment for Data Analysis and Graphics}, publisher={Wadsworth}, address={Belmont, CA}, year= {1984} } @techreport{Bergstralh88, author= {Bergstralh, E. J. and Offord, K. P.}, year = {1988}, title = {Conditional probabilities used in calculating cohort expected survival}, number = {37}, institution= {Department of Health Sciences Research, Mayo Clinic} } @article{Berry83, author= {Berry, G.}, year = {1983}, title= {The analysis of mortality by the subject years method}, journal= biok, volume={39}, pages={173--184} } @book{Bickel77, author={Bickel, P. J. and Doksum,K. J. }, title= { Mathematical Statistics: Basic Ideas and Selected Topics}, publisher={Holden-Day}, address={San Francisco}, year= {1977} } @article{Binder92, author= {Binder, D. A.}, year= {1992}, title= {Fitting {C}ox's proportional hazards models from survey data}, journal={Biometrika}, volume={79}, pages={139--147} } @ARTICLE{Blackstone86, author = {Blackstone, E. H. and Naftel, D. C. and Turner, M. E.}, year = 1986, title = {The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information}, journal = JASA, volume = 81, pages = {615-624}, annote = {parametric survival models; non PH} } @article{Bonsel90, author= {Bonsel, G. J. and Klompmaker, I. J. and {van't Veer}, F. and Habbema, J. D. F. and Slooff, M. J. H.}, year= {1990}, title= {Use of prognostic models for assessment of value of liver transplantation in primary biliary cirrhosis}, journal={Lancet}, volume={335}, pages={493--497} } @article{Borgan95, author= {Borgan, \O. and Goldstein, L. and Langholz, B.}, year= {1995}, title= {Methods for the analysis of sampled cohort data in the {C}ox proportional hazards model}, journal=annals, volume={23}, pages={1749--1778} } @article{Breslow93, author= {Breslow, N. E. and Clayton, D. G.}, year= {1993}, title= {Approximate inference in generalized linear mixed models}, journal=jasa, volume={88}, pages={9--25} } @article{Cai95, author= {Cai, J. and Prentice, R. G.}, year= {1995}, title= {Estimating equations for hazard ratio parameters based on correlated failure time data}, journal={Biometrika}, volume={82}, pages={151--164} } @article{Cain84, author={Cain, K. C. and Lange, N. T.}, year= {1984}, title= {Approximate case influence for the proportional hazards regression model with censored data}, journal={Biometrics}, volume={40}, pages={493--499} } @book{Chambers83, author= {Chambers, J. M. and Cleveland, W. S. and Kleiner, B. and Tukey, P. A.}, year= {1983}, title={Graphical Methods for Data Analysis}, publisher ={Wasdworth}, address ={Belmont, CA} } @book{Chambers91, author= {Chambers, J. M. and Hastie, T. J.}, year= {1993}, title={Statistical Models in {S}}, publisher ={Chapman and Hall}, address ={New York} } @article{Chappell92, author= {Chappell, R.}, year= {1992}, title= {A note on linear rank tests and {G}ill and {S}chumacher's tests of proportionality}, journal={Biometrika}, volume={79}, pages= {199--201} } @article{Chen91, author= {Chen, C. H. and Wang, P. C.}, year= {1991}, title= {Diagnostic plots in {C}ox's regression model}, journal={Biometrics}, volume={47}, pages= {841--850} } @article{Clegg99, author= {Clegg, L. X. and Cai, J. and Sen, P. K.}, year= {1999}, title= {A marginal mixed baseline hazards model for multivariate failure time data}, journal=jasa, volume={55}, pages= {805--812} } @book{Cook82, author= {Cook, R. D. and Weisberg, S.}, year= {1982}, title= {Residuals and Influence in Regression}, publisher={Chapman and Hall}, address= {London} } @article{Cox72, author= {Cox, D. R.}, year= {1972}, title= {Regression models and life-tables (with discussion)}, journal=jrssb, volume={34}, pages= {187--220} } @book{Cox84, author= {Cox, D. R. and Oakes, D. O.}, year= {1984}, title= {Analysis of Survival Data}, publisher= {Chapman and Hall}, address= {London} } @book{Cochran76, author= {Cochran, W.G.}, year= {1976}, title={Sampling Techniques, third edition}, publisher ={Wiley}, address ={New York} } @article{Crowley77, author= {Crowley, J. and Hu, M.}, year= {1977}, title= {Covariance analysis of heart transplant survival data}, journal=jasa, volume={72}, pages= {27--36} } @article{Dempster77, author= {Dempster, A. P. and Laird, N. M. and Rubin, D. B.}, year= {1977}, title= {Maximum likelihood from incomplete data via the {EM} algorithm (with discussion)}, journal=jrssb, volume={39}, pages= {1--38} } @techreport{Deng95, author= {Deng, Y. and Quigley, J.M. and Van Order, R.}, year= {1995}, title= {Mortgage Terminations}, institution={Institute of Business and Economic Research, University of California at Berkeley}, type= {Working Paper}, number={95-230}, } @article{Dickson89, author= {Dickson, E. R. and Grambsch, P. M. and Fleming, T. R and Fisher, L. D. and Langworthy, A.}, year= {1989}, title= {Prognosis in primary biliary cirrhosis: Model for decision making}, journal ={Hepatology}, volume={10}, pages ={1--7} } @article{Ederer61, author= {Ederer, F. and Axtell, L. M. and Cutler, S. J.}, year= {1961}, title= {The relative survival rate: A statistical methodology}, journal ={National Cancer Inst. Monographs}, volume={6}, pages ={101--121} } @techreport{Ederer77, author= {Ederer, F. and Heise, H.}, year= {1977}, title= {Instructions to {IBM} 650 programmers in processing survival computations}, institution={End Results Evaluation Section, National Cancer Institute}, type={Methodological Note}, number={No. 10}, pages ={101--121} } @article{Edmonson79, author= {Edmonson, J. H. and Fleming, T. R. and Decker, D. G. and Malkasian, G. D. and Jorgensen, E. O. and Jefferies, J. A. and Webb, M. J. and Kvols, L. K.}, year= {1979}, title= {Different chemotherapeutic sensitivities and host factors affecting prognosis in advanced ovarian carcinoma versus minimal residual disease }, journal ={Cancer Treatment Reports}, volume={63}, pages ={241--247} } @article{Efron77, author= {Efron, B.}, year= {1977}, title ={The efficiency of {C}ox's likelihood function for censored data}, journal=jasa, volume={72}, pages= {557--565} } @book{Efron82, author= {Efron, B.}, year= {1982}, title ={The Jackknife, the Bootstrap and Other Resampling Plans}, publisher={SIAM}, address ={Philadelphia} } @article{Ezekiel24, author= {Ezekiel, M.}, year= {1924}, title ={A method for handling curvilinear correlation for any number of variables}, journal=jasa, volume={19}, pages= {431--453} } @article{Fleming81, author= {Fleming, T. R. and Harrington, D. P.}, year= {1981}, title ={A class of hypothesis tests for one and two sample censored survival data}, journal=commstata, volume={10}, pages= {763--794} } @article{Fleming84, author= {Fleming, T. R. and Harrington, D. P.}, year= {1984}, title ={Nonparametric estimation of the survival distribution in censored data}, journal=commstata, volume={13}, pages= {2469--2486} } @book{Fleming91, author= {Fleming, T. R. and Harrington, D. P.}, year= {1991}, title= {Counting Processes and Survival Analysis}, publisher= {Wiley}, address= {New York} } @article{Eilers96, author= {Eilers, P. H. C. and Marx, B. D.}, year= {1996}, title ={Flexible smoothing with {B}-splines and penalties}, journal={Stat. Science}, volume={11}, pages= {89--121} } @article{Gail81, author= {Gail, M. H. and Lubin, J. H. and Rubinstein, L. V.}, year = {1981}, title = {Likelihood Calculations for Matched Case-Control Studies and Survival Studies with Tied Death Times}, journal=biok, volume=68, pages={703--707} } @article{Gail86, author= {Gail, M. H. and Byar, D. P.}, year= {1986}, title= {Variance calculations for direct adjusted survival curves, with applications to testing for no treatment effect}, journal=bioj, volume={28}, pages= {587--599} } @article{Gill87, author= {Gill, R. and Schumacher, M.}, year= {1987}, title= {A simple test of the proportional hazards assumption}, journal={Biometrika}, volume={74}, pages= {289--300} } @article{Grambsch94, author= {Grambsch, P. M. and Therneau, T. M.}, year= {1994}, title ={Proportional hazards tests and diagnostics based on weighted residuals}, journal={Biometrika}, volume={81}, pages= {515--526} } @article{Grambsch95, author= {Grambsch, P. M. and Therneau, T. M. and Fleming, T. R.}, year= {1995}, title= {Diagnostic plots to reveal functional form for covariates in multiplicative intensity models}, journal={Biometrics}, volume={51}, pages= {1469-1482} } @article{Gray92, author= {Gray, R. J.}, year = {1992}, title= {Flexible methods for analyzing survival data using splines, with applications to breast cancer prognosis}, journal=jasa, volume={87}, pages = {942--951} } @article{Gray94, author= {Gray, R. J.}, year = {1994}, title= {Spline-based tests in survival analysis}, journal=biom, volume={50}, pages = {640--652} } @article{Green84, author={Green, P.J.}, year = {1984}, title = {Iteratively reweighted least squares for maximum likelihood estimation, and some robust and resistant alternatives (with discussion).}, journal=jrssb, volume=46, pages={149--192} } @techreport{Hall95, author= {Hall, C. B. and Zeger, S. L. and Bandeen-Roche, K. J.}, year= {1995}, title= {Adjusted variable plots for {C}ox's proportional hazards regression model}, institution={The John's Hopkins University, School of Hygiene and Public Health, Department of Biostatistics} } @incollection{Harrell86, author= {Harrell, F.}, year= {1986}, title= {The PHGLM procedure}, booktitle= {SAS Supplemental Library User's Guide, Version 5}, address= {Cary, NC}, publisher= {SAS Institute Inc} } @article{Hakama77, author= {Hakama, M. and Hakulinen, T.}, year= {1977}, title= {Estimating the expectation of life in cancer survival studies with incomplete follow-up information}, journal={J Chronic Diseases}, volume={30}, pages= {585--597} } @article{Hakulinen82, author= {Hakulinen, T.}, year= {1982}, title= {Cancer survival corrected for heterogeneity in patient withdrawal}, journal=biom, volume={38}, pages= {933--942} } @article{Hakulinen85, author= {Hakulinen, T. and Abeywickrama, K. H.}, year= {1985}, title= {A computer program package for relative survival analysis}, journal={Computer Programs in Biomedicine}, volume={19}, pages= {197--207} } @article{Hakulinen77, author= {Hakulinen, T.}, year= {1977}, title= {On long term relative survival rates}, journal={J Chronic Diseases}, volume={30}, pages= {431--443} } @article{Harrington82, author= {Harrington, D. P. and Fleming, T. R.}, year= {1982}, title= {A class of rank test procedures for censored survival data}, journal=biok, volume={69}, pages= {553-566} } @book{Hastie90, author= {Hastie, T. J. and Tibshirani, R. J.}, year= {1990}, title ={Generalized Additive Models}, publisher ={Chapman and Hall}, address= {London} } @article{Hastie96, author= {Hastie, T. J.}, year= {1996}, title= {Pseudosplines}, journal=jrssb, volume={58}, pages= {379--396} } @article{Heit99, author= {Heit, J. A. and M. D. Siverstein and D. N. Mohr and T. M. Petterson and W. M. O'Fallon and L. J. Melton III}, year= {1999}, title= {Predictors of survival after deep vein thrombosis and pulmonary embolism}, journal={Arch Internal Med}, volume={159}, pages= {445--453} } @techreport{Hodges99, author= {J. S. Hodges and D. J. Sargent}, year= {1998}, title= {Counting degrees of freedom in hierarchical and other richly-parameterized models}, institution={University of Minnesota, Division of Biostatistics}, type={Research Report}, number={98-004} } @book{Hougaard00, author={Hougaard, P.}, title= {Analysis of Multivariate Survival Data}, publisher={Springer-Verlag}, address={New York}, year= {2000} } @inproceedings{Huber67, author= {Huber, P. J.}, year= {1967}, title= {The behavior of maximum likelihood estimates under non-standard conditions}, booktitle= {Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability}, volume={1}, pages= {221--233} } @article{Huffer91, author= {Huffer, F.W. and McKeague, I.W}, title= {Weigthed least squares estimation for {A}alen's additive risk model}, year= {1991}, journal=jasa, volume={86}, pages={114--129} } @article{Hurvich98, author= {Hurvich, C. M. and Simonoff, J. S. and Tsai, C.-L.}, year= {1998}, title= {Smoothing parameter selection in nonparametric regression using an improved {A}kaike information criterion}, journal=jrssb, volume={60}, pages= {271--293} } @article{Islam94, author= {Islam, M. A.}, year= {1994}, title= {Multistate survival models for transitions and reverse transitions: an application to contraceptive use data}, journal=jrssa, volume={157}, pages= {441--455} } @article{Jones90, author={Jones, M. P. and Crowley, J.}, title={Asymptotic proporties of a general class of nonparametric tests for survival analysis}, year={1990}, journal=annals, volume={18},pages={1203--1220} } @book{Kalbfleisch80, author= {Kalbfleisch, J. D. and Prentice, R. L.}, year= {1980}, title= {The Statistical Analysis of Failure Time Data}, publisher= {Wiley}, address ={New York} } @article{Klein91, author= {Kay, R.}, year= {1983}, title ={The analysis of transition times in a multistate stochastic process using proportional hazard regression models}, journal=commstata, volume={11}, pages= {1743--1756} } @article{Kay83, author= {Korn, Edward L. and Graubard, Barry I. and Midthune, Douglas}, year= {1997}, title ={Time-to-event analysis if longitudinal follow-up of a survey: Choice of the time scale}, journal={Am J of Epidemiology}, volume={145}, pages= {72--80} } @article{Korn97, author= {Klein, J. P.}, year= {1991}, title ={Small sample moments of some estimators of the variance of the {K}aplan--{Meier} and {N}elson--{A}alen estimators}, journal=scand, volume={18}, pages= {333--340} } @article{Lagakos84, author={Lagakos, S. W. and Schoenfeld, D. A.}, title={Properties of proportional-hazards score tests under misspecified regression models}, year={1984}, journal=biom, volume={40}, pages={1037--1048} } @article{Laird81, author= {Laird, N. and Olivier, D.}, year= {1981}, title= {Covariance analysis of censored survival data using log-linear analysis techniques}, journal=jasa, volume={76}, pages= {231--240} } @article{Langholz91, author= {Langholz, B. and Thomas, D.C.}, year= {1991}, title= {Efficiency of cohort sampling designs: some surprising resluts}, journal={Biometrics}, volume={47}, pages= {1563--1572} } @article{Langholz96, author= {Langholz, B. and Goldstein, L.}, year= {1996}, title= {Risk set sampling in epidemiologic cohort studies}, journal={Statistical Science}, volume={11}, pages= {35--53} } @article{Laurie89, author ={Laurie, J. A. and Moertel, C. G. and Fleming, T. R. and Wieand, H. S. and Leigh, J. E. and Rubin, J. and McCormack, G. W. and Gerstner, J. B. and Krook, J. E. and Malliard, J.}, year= {1989}, title= {Surgical adjuvant therapy of large-bowel carcinoma: {A}n evaluation of levamisole and the combination of levamisole and fluorouracil: The {N}orth {C}entral {C}ancer {T}reatment {G}roup and the {M}ayo {C}linic}, journal={J. Clinical Oncology}, volume={7}, pages= {1447--1456} } @incollection{Lee92, author= {Lee, E. W. and Wei, L. J. and Amato, D.}, year= {1992}, title= {{C}ox-type regression analysis for large number of small groups of correlated failure time observations}, editor= {Klein, J. P. and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, pages= {237--247}, publisher= {Kluwer}, address= {Netherlands} } @article{Lindor94, author= {Lindor, K. D. and Dickson, E. R. and Baldus, W. P. and Jorgensen, R. A. and Ludwig, J. and Murtaugh, P. A. and Harrison, J. M. and Wiesner, R. H. and Anderson, M. L. and Lange, S. M. and LeSage, G. and Rossi, S. S. and Hofman, A. F.}, year= {1994}, title= {Ursodeoxycholic acid in the treatment of primary biliary cirrhosis}, journal={Gastroenterology}, volume={106}, pages= {1284--1290} } @article{Liang86, author= {Zeger, S. L. and Liang, K. Y.}, year= {1986}, title= {Longitudinal data analysis for discrete and continuous outcomes}, journal={Biometrics}, volume={42}, pages= {121--130} } @article{Liang86b, author= {Liang, K. Y. and Zeger, S. L.}, year= {1986}, title= {Longitudinal data analysis using generalized linear models}, journal=biok, volume={73}, pages= {13--22} } @article{Liang88, author= {Zeger, S. L. and Liang, K. Y. and Albert, P. S.}, year= {1988}, title= {Models for longitudinal data: A generalized estimating equation approach}, journal=biom, volume={44}, pages= {1049--1060} } @article{Lin91, author= {Lin, D. Y.}, year= {1991}, title= {Goodness-of-fit analysis for the {C}ox regression model based on a class of parameter estimators}, journal=jasa, volume={86}, pages= {725--728} } @article{Lin89, author= {Lin, D. Y. and Wei, L. J.}, year= {1989}, title= {The robust inference for the {C}ox proportional hazards model}, journal=jasa, volume={84}, pages= {1074--1078} } @article{Lin91b, author= {Lin, D. Y. and Wei, L. J.}, year= {1991}, title= {Goodness-of-fit tests for the general {C}ox regression model}, journal= {Statistica Sinica}, volume={1}, pages= {1--17} } @article{Lin93, author= {Lin, D. Y. and Wei, L. J. and Ying, Z.}, year= {1993}, title= {Checking the {C}ox model with cumulative sums of martingale-based residuals}, journal={Biometrika}, volume={80}, pages= {557--572} } @article{Lin93b, author= {Lin, D. Y. and Ying, Z.}, year= {1993}, title= {Cox regression with incomplete covariate measurements}, journal= jasa, volume={88}, pages= {1341--1349} } @article{Lin94, author= {Lin, D. Y.}, year= {1994}, title= {Cox regression analysis of multivariate failure time data: the marginal approach}, journal= statmed, volume={13}, pages= {2233--2247} } @article{Link84, author= {C. L. Link}, year= {1984}, title= {Confidence intervals for the survival function using {C}ox's proportional-hazard model with covariates}, journal=biom, volume={40}, pages= {601--610} } @article{Link86, author= {C. L. Link}, year= {1986}, title= {Response to {J}. {O'Quigley}, correspondence section}, journal=biom, volume={42}, pages= {219--220} } @article{Lipsitz90, author= {Lipsitz, S. R. and Laird, N. M. and Harrington, D. P.}, year= {1990}, title= {Using the jackknife to estimate the variance of regression estimators from repeated measures studies}, journal=commstata, volume={19}, pages= {821--845} } @article{Lipsitz94, author= {Lipsitz, S. R. and Dear, K. B .G. and Zhao, L.}, year= {1994}, title= {Jackknife estimators of variance for parameter estimates from estimating equations with applications to clustered survival data}, journal= biom, volume={50}, pages= {842--846} } @article{Lunn95, author= {Lunn, M. and McNeil, D.}, year= {1995}, title= {Applying {C}ox regression to competing risks}, journal={Biometrics}, volume={51}, pages= {524--532} } @article{Mallows86, author= {Mallows, C. L.}, year= {1986}, title= {Augmented partial residuals}, journal={Technometrics}, volume={28}, pages= {313--319} } @article{Mantel66, author= {Mantel, N.}, year= {1966}, title= {Evaluation of survival data and two new rank order statistics arising in its consideration}, journal={Cancer Chemotherapy Reports}, volume={50}, pages= {163--166} } @article{Mantel77, author= {Mantel, N. and Bohidar, N. R. and Ciminera, J. L.}, year= {1977}, title= {Mantel--{H}aenszel analyses of litter-matched time-to-response data with modifications for recovery of interlitter information}, journal={Cancer Research}, volume={37}, pages= {3863--3868} } @book{glim, author= {McCullagh, P. and Nelder, J.A.}, year= {1983}, title= {Generalized Linear Models}, publisher= {Chapman and Hall} } @article{McGilchrist91, author= {McGilchrist, C. A. and Aisbett, C. W.}, year= {1991}, title= {Regression with frailty in survival analysis}, journal={Biometrics}, volume={47}, pages= {461--466} } @article{McGilchrist93, author= {McGilchrist, C. A.}, year= {1993}, title= {{REML} estimation for survival models with frailty}, journal={Biometrics}, volume={49}, pages= {221--225} } @article{McGilchrist95, author= {McGilchrist, C. A. and Yau, K. K. W.}, year= {1995}, title= {The derivation of {BLUP}, {ML} and {REML} estimation methods for generalised linear mixed models}, journal= commstata, volume={24}, pages= {2963--2980} } @book{Miller81, author= {Miller, Jr., R. G.}, year= {1981}, title ={Survival Analysis}, publisher= {Wiley}, address= {New York} } @article{Moertel90, author= {Moertel, C.G. and Fleming, T.R. and McDonald, J.S. and Haller, D.G. and Laurie, J.A. and Goodman, P.J. and Ungerleider, J.S. and Emerson, W.A. and Tormey, D.C. and Glick, J.H. and Veeder, M.H. and Mailliard, J.A.}, year= {1990}, title= {Levamisole and fluorouracil for adjucant therapy of resected colon carcinoma.}, journal= NEJM, volume={332}, pages= {352--358} } @article{Moreau85, author= {Moreau, T. and O'Quigley, J. and Mesbah, M.}, year= {1985}, title= { A global goodness-of-fit statistic for the proportional hazards model}, journal= {Applied Stat.}, volume={34}, pages= {212--218} } @article{Moss83, author= {Moss, A. J. and {the Multicenter Postinfarction Research Group}}, year= {1983}, title= {Risk stratification and survival after myocardial infarction}, journal= {New England J. Medicine}, volume={309}, pages= {331--336} } @article{Moss88, author= {Moss, A. J. and {the Multicenter Diltiazem Postinfarction Trial Research Group}}, year= {1988}, title= {The effect of diltiazem on mortality and reinfarction after myocardial infarction}, journal= {New England J. Medicine}, volume={319}, pages= {385--392} } @book{Mosteller77, author= {Mosteller, F. and Tukey, J. W.}, year= {1977}, title= {Data Analysis and Regression}, publisher= {Addison-Wesley}, address={Reading, MA} } @article{Nagelkerke84, author= {Nagelkerke, N. J. D. and Oosting, J. and Hart, A. A. M.}, year= {1984}, title ={A simple test for goodness of fit of {C}ox's proportional hazards model}, journal={Biometrics}, volume={40}, pages= {483--486} } @article{Neuberger86, author= {Neuberger, J. and Altman, D. G. and Christensen, E. and Tygstrup, N. and Williams, R.}, year= {1986}, title ={Use of a prognostic index in evaluation of liver transplantation for primary biliary cirrhosis}, journal={Transplantation}, volume={41}, pages= {713--716} } @article{Nielsen92, author= {Nielsen, G. G. and Gill, R. D. and Andersen, P. K. and S{\o}rensen, T. I.}, year= {1992}, title ={A counting process approach to maximum likelihood estimation of frailty models}, journal=scand, volume={19}, pages= {25--43} } article{Oakes93, author={Oakes, D. and A.J. Moss and J.T. Fleiss and J.T. Bigger, Jr. and T.M. Therneau and S.W. Eberly and M.P. McDermott and A. Manatunga and E. Carleen and J. Benhorin, and {the Multicenter Diltiazem Post-Infarction Research Group}}, year = {1993}, title = {Use of compliance measures in and analysis of the effect of {D}iltiazem on mortality and reinfarction after myocardial infarction}, journal=jasa, volume={88}, pages = {44-49} } @incollection{Oakes92, author= {Oakes, D.}, year= {1992}, title ={Frailty models for multiple event times}, editor= {Klein, J. P. and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, publisher= {Kluwer}, address= {Netherlands} } @article{Omori93, author={Omori,Y. and Johnson,R. A.}, title={The influence of random effects on the unconditional hazard rate and survival functions}, year={1993}, journal=biok,volume={80}, pages={910--914} } @phdthesis{Parner96, author={Parner, E.}, title={Inference in semiparametric frailty models}, year ={1997}, school={University of Aarhus, Denmark} } @article{Pettitt90, author={Pettitt, A. N. and Bin Daud, I.}, title={Investigating time dependence in {C}ox's proportional hazards model}, year= {1990}, journal=applstat, volume={39},pages={313--329} } @article{Prentice86, author= {Prentice, R. L.}, year= {1986}, title ={A case-cohort design for epidemilogic cohort studies and disease prevention trials}, journal={Biometrika}, volume={73}, pages ={1--11} } @article{Prentice92, author= {Prentice, R. L. and Cai, J.}, year= {1992}, title ={Covariance and survivor function estimation using censored multivariate failure time data}, journal={Biometrika}, volume={79}, pages ={495--512} } @incollection{Prentice91, author= {Prentice, R. L. and Cai, J.}, year= {1991}, title ={Marginal and conditional models for the analysis of multivariate failure time data}, pages ={393--406}, editor= {Klein, J. P and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, publisher= {Kluwer Academic Publishers}, address= {Netherlands} } @article{Prentice81, author= {Prentice, R. L. and Williams, B. J. and Peterson, A. V.}, year= {1981}, title= {On the regression analysis of multivariate failure time data}, journal={Biometrika}, volume={68}, pages= {373--379} } @book{Press88, author= {Press, W.H. and Teukolsky, S.A. and Vetterling, W.T. and Flannery, B.P.}, year= {1988}, title= {Numerical Recipes in {C}}, publisher= {Cambridge University Press}, address ={Cambridge} } @article{Quantin96, author= {Quantin, C. and Moreau, T. and Asselaiin B. and Maccario, J. and Lelloucj, J. } , title= {A regression survival model for testing the proportional hazards hypothesis }, year= {1996}, journal=biom, volume={52}, pages={874--885} } @article{Quigley89, author= {O'Quigley, J. and Pessione, F.}, year= {1989}, title= {Score tests for homogeneity of regression effect in the proportional hazards model}, journal={Biometrics}, volume={45}, pages ={135--144} } @article{Reid85, author= {Reid, N. and Cr{\'{e}}peau, H.}, year= {1985}, title= {Influence functions for proportional hazards regression}, journal={Biometrika}, volume={72}, pages= {1--9} } @article{Ricci97, author= {Ricci, P. and Therneau, T. M. and Malinchoc, M. and Benson, J. T. and Petz, J. L. and Klintmalm, G. B. and Crippin, J. S. and Wiesner, R. H. and Steers, J. L. and Rakela, J. and Starzl, T. E. and Dickson, E. R.}, year= {1997}, title= {A prognostic model for the outcome of liver transplantation in patients with cholestatic liver disease}, journal={Hepatology}, volume={25}, pages= {672--677} } @manual{SAS96, author={{SAS Institute Inc.}}, year={1996}, title={{SAS/STAT} Software: Changes and Enhancements through Release 6.11}, organization={{SAS} Institute, Inc.}, address={Cary, N.C.}, note={Chapter 8: The PHREG procedure} } @article{Sastry97, author= {Sastry, N.}, title={A nested frailty model for survival data, with an application to the study of child survival in northesat Brazil}, year={1997}, journal=jasa, volume={92}, pages = {426--435} } @article{Schoenfeld80, author= {Schoenfeld, D.}, year= {1980}, title= {Chi-squared goodness-of-fit tests for the proportional hazards regression model}, journal={Biometrika}, volume={67}, pages= {145--153} } @article{Schoenfeld81, author={Schoenfeld, D.}, title={The asymptotic properties of nonparametric tests for comparing survival distributions}, year={1981}, journal=biok, volume={68}, pages={316--319} } @article{Schoenfeld83, author={Schoenfeld, D. A.}, title={Sample-size formula for the proportional-hazards regression model}, year={1983}, journal=biom, volume={39}, pages={499--503} } @article{Schum87, author={Schumacher, M. and Olschewski, M. and Schmoor,C.}, title={The impact of heterogeneity on the comparison of survival times}, year={1987}, journal=statmed, volume={6},pages={773-784} } @article{Segal93, author= {Segal, M.R. and Neuhaus, J. M.}, year= {1993}, title= {Robust inference for multivariate survival data}, journal=statmed, volume={12}, pages= {1019--1031} } @article{Segal94, author= {Segal, M.R. and Baccheti, P. and Jewell, N. P.}, year= {1994}, title= {Variances for maximum penalized likelihood estimates obtained via the {EM} algorithm }, journal=jrssb, volume={56}, pages= {345--352} } @article{Simon11, title= {Regularization Paths for {C}ox’s Proportional Hazards Model via Coordinate Descent}, author= {Noah Simon and Jerome Friedman and Trevor Hastie Rob Tibshirani}, journal={J Statistical Software}, volume=39, pages={1--123} } @article{Smith96, author= {Smith, P.J. and Hietjan, D.F.}, year= {1996}, title= {Testing and adjusting for overdispersion in generalized linear models}, journal=jrssc, volume={?} } @article{Solomon84, author= {Solomon, P. J. } , title= {Effect of misspecification of regression models in the analysis of survival data}, year= {1984}, journal=biok, volume={71}, pages={291--298} } @article{Solomon86, author= {Solomon, P. J. } , title= {Amendments and corrections}, year= {1986}, journal=biok, volume={73}, pages={245--245} } @book{Spector94, author={Spector, P.}, title= {An Introduction to {S} and {S-Plus}}, publisher={Wadsworth}, address={Pacific Grove, CA}, year= {1994} } @article{Stablein81, author= {Stablein, D. M. and Carter, Jr., W. H. and Novak, J. W.}, title= {Analysis of survival data with nonproportional hazard functions}, year= {1981}, journal={Controlled Clinical Trials}, volume={2}, pages={149--159} } @inproceedings{Stone85, author= {Stone, C. J. and Koo, C. Y.} , title= { Additive splines in statistics}, year= {1985}, booktitle={Computational Statistics Section}, organization={American Statistical Association}, address={Alexandria, Virginia}, pages={646--651} } @article{Stone86, author= {Stone, C. J.} , title= {Comment to paper by {H}astie and {T}ibshirani}, year= {1986}, journal=statsci, volume={1}, pages={312--314} } @article{Storer85, author= {Storer, B. E. and Crowley, J.}, year= {1985}, title ={A diagnostic for {C}ox regression and general conditional likelihoods}, journal =jasa, volume={80}, pages= {139-147} } @article{Struthers86, author= {Struthers, C. A. and Kalbfleisch, J. D. } , title= {Misspecified proportional hazard models}, year= {1986}, journal=biok, volume={73}, pages={363--369} } @article{Therneau90, author= {Therneau, T. M. and Grambsch, P. M. and Fleming, T. R.}, year= {1990}, title= {Martingale based residuals for survival models}, journal={Biometrika}, volume={77}, pages= {147--160} } @article{Therneau97, author= {Therneau, T. M. and Hamilton, S. A.}, year= {1997}, title= {{rhDNase} as an example of recurrent event analysis}, journal=statmed, volume={16}, pages= {2029--2047} } @article{Therneau99, author= {Therneau, T. M. and Li, H.}, year= {1999}, title= {Computing the {C}ox model for case-cohort designs}, journal=lifetime, volume={5}, pages= {99--112} } @book{Therneau00, author={Therneau, T. M. and Grambsch, P. M.}, title= {Modeling Survival Data: Extending the {C}ox Model}, publisher={Springer-Verlag}, address={New York}, year= {2000} } @article{Therneau03, author= {Therneau, T. M. and Grambsch, P. M. and Pankratz, V. S.}, year= {2003}, title= {Penalized survival models and frailty}, journal={J Computational Graphical Statistics}, volume={12}, pages= {156--175} } @article{Thomsen91, author= {Thomsen, B. L. and Keiding, N. and Altman, D. G.}, year= {1991}, title= {A note on the calculation of expected survival, illustrated by the survival of liver transplant patients}, journal=statmed, volume={10}, pages= {733--738} } @article{Thomsen92, author= {Thomsen, B. L. and Keiding, N. and Altman, D. G.}, year= {1992}, title= {Reply to a letter to the editor}, journal=statmed, volume={11}, pages= {1528--1530} } @article{Uitti93, author= {Uitti, R.J. and Ahlskog, J.E. and Maraganore, D.M. and Muenter, M.D. and Atkinson, E.J. and Cha, R.H. and O'Brien, P.C.}, year= {1993}, title= {Levodopa therapy and survival in idiopathic Parkinson's disease: Olmsted County Project}, journal ={Neurology}, volume={43}, pages= {1918--1926} } @book{Venables97, author= {Venables, W. N. and Ripley, B. D.}, year= {1997}, title= {Modern Applied Statistics with {S-PLUS}, second edition}, publisher= {Springer-Verlag}, address ={New York} } @article{Verhuel93, author= {Verheul, H. A. and Dekker, E. and Bossuyt, P. and Moulijn, A. C. and Dunning, A. J.}, year= {1993}, title= {Background mortality in clinical survival studies}, journal ={Lancet}, volume={341}, pages= {872--875} } @article{Wahba83, author= {Wahba, G.}, year= {1983}, title= {Bayesian ''confidence intervals'' for the cross-validated smoothing spline}, journal =JRSSB, volume={45}, pages= {133--150} } @article{Wei89, author= {Wei, L. J. and Lin, D. Y. and Weissfeld, L.}, year= {1989}, title= {Regression analysis of multivariate incomplete failure time data by modeling marginal distributions}, journal=jasa, volume={84}, pages= {1065--1073} } @article{Wei93, author= {Lin, D. Y. and Wei, L. J. and Ying, Z.}, year= {1993}, title= {Checking the {C}ox model with cumulative sums of martingale-based residuals}, journal=biok, volume={80}, pages= {557--572} } @article{White80, author= {White, H.}, year= {1980}, title ={A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity}, journal= {Econometrica}, volume={48}, pages= {817--838} } @article{White82, author= {White, H.}, year= {1982}, title= {Maximum likelihood estimation of misspecified models}, journal= {Econometrica}, volume={50}, pages= {1--26} } @article{Whitehead80, author= {Whitehead, J.}, year= {1980}, title= {Fitting {C}ox's regression model to survival data using {GLIM}}, journal= applstat, volume={29}, pages= {268--275} } @unpublished{Winemiller98, author= {Winemiller, M.H. and Stolp-Smith, K.A, and Silverstein, M.D. and Therneau, T.M.}, year= {1998}, title= {Sequential pneumatic compression or heparin is effective in preventing venous thromboembolism in spinal cord injury patients}, note={Submitted} } @article{Winkler88, author= {Winkler, H. Z. and Rainwater, L. M. and Myers, R. P. and Farrow, G. M. and Therneau, T. M. and Zincke, H. and Lieber, M. M.}, year= {1988}, title= {Stage {D}1 Prostatic Adenocarcinoma: {S}ignificance of nuclear {DNA} ploidy patterns studied by flow cytometry}, journal= {Mayo Clinic Proceedings}, volume={63}, pages= {103--112} } @article{Yau97, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1997}, title= {Use of generalised linear mixed models for the analysis of clustered survival data}, journal= {Biometrical Journal}, volume={39}, pages= {3--11} } @article{Zhen94, author= {Zhen, B. and Murphy, J.R.}, year= {1994}, title= {Sample size determination for an exponential survival model with an unrestricted covariate}, journal=statmed, volume={13}, pages= {391--397} } @article{Aitkin80, author= {Aitkin,M. and Clayton, D.} , title= {The fitting of exponential, Weibull and extreme value distributions to complex censored survival data using GLIM}, year= {1980}, journal=applstat, volume={29}, pages={156--163} } @article{Bernstein78, author= {Berstein,D. and Lagakos, S. W.} , title= {Sample size and power determination for stratified clinical trials}, year= {1978}, journal=jscs, volume={8}, pages={65--73} } @article{Bie87, author= {Bie, O. and Borgan, O. and Liestoel, K.}, year = {1987}, title= {Confidence intervals and confidence bands for the cumulative hazard rate function and their small sample properties}, journal= scand, volume={14}, pages={221--233} } @book{Billingsley68, author={Billingsley, P.}, title= {Convergence of Probability Measures}, publisher={Wiley}, address={New York}, year= {1968} } @article{Block85, author= {Block, H. W. and Borges, W. S. and Savits, T. H.}, year= {1985}, title= {Age-dependent minimal repair}, journal=jap, volume={22}, pages={370--385} } @article{Breslow72, author= {Breslow, N. E.}, year= {1972}, title= {Discussion of {P}rofessor {C}ox's paper}, journal=jrssb, volume={34}, pages={216--217} } @book{Breslow80, author={Breslow, N. E. and Day, N. E.}, title= {The Analysis of Case-Control Studies}, volume={1}, series = {Statistical Methods in Cancer Research}, publisher={IARC}, address={Lyon}, year= {1980} } @book{glim2, author= {McCullagh, P. and Nelder, J.A.}, year= {1989}, title= {Generalized Linear Models, 2nd ed.}, publisher= {Chapman and Hall} } @incollection{Hettmansperger98, author= {Hettmansperger, T.}, year= {1998}, title= {Median}, editor={Armitage, P. and Colton, T.}, booktitle={Encyclopedia of Biostatics}, volume={4}, publisher= {Wiley}, address={New York}, pages={2525--2526} } @book{Huber81, author={Huber, P. J.}, title= {Robust Statistics}, publisher={Wiley}, address={New York}, year= {1981} } @article{Johansen83, author= {Johansen, S.}, year= {1983}, title= {An extension of {C}ox's regression model}, journal={Int. Stat. Review}, volume={51}, pages={165--174} } @book{Little87, author={Little, R.J.A. and Rubin, D.B.}, title= {Statistical Analysis with Missing Data}, publisher={John Wiley \& Sons}, address={New York}, year= {1987} } @incollection{Little98, author= {Little, R. J.}, year= {1998}, title= {Missing Data}, editor={Armitage, P. and Colton, T.}, booktitle={Encyclopedia of Biostatics}, volume={4}, publisher= {Wiley}, pages={2622--2635} } @article{Nelson69, author= {Nelson, W.}, year= {1969}, title= {Hazard plotting for incomplete failure data}, journal= {J. Quality Technology}, volume={1}, pages= {27--52} } @article{Peto72, author= {Peto, R.}, year= {1972}, title= {Discussion of {P}rofessor {C}ox's paper}, journal=jrssb, volume={34}, pages={205--207} } @article{Prentice78, author= {Prentice, R. L. and Gloeckler, L. A. } , title= {Regression analysis of grouped survival data with application to breast cancer data}, year= {1978}, journal=biom, volume={34}, pages={57--67} } @article{Self88, author= {Self, S. G. and Prentice, R. L.}, year= {1988}, title= {Asymptotic distribution theory and efficiency results for case-cohort studies}, journal=annals, volume={16}, pages= {64--81} } @techreport{Therneau94, author= {Therneau, T. M. and Sicks, J. and Bergstralh, E. and Offord, J.}, year = {1994}, title = {Expected survival based on hazard rates}, number = {52}, institution= {Department of Health Sciences Research, Mayo Clinic} } @article{Turnbull76, author= {Turnbull, B.W.}, year= {1976}, title= {The empirical distribution function with arbitrarily grouped, censored and truncated data}, journal={jrssb}, volume={38}, pages= {290--295} } @book{Breiman84, author= {Breiman, L. and Friedman, J. H. and Olshen, R. A. and Stone, C. J.}, title= {Classification and Regression Trees}, year = 1984, publisher={Wadsworth}, address={Belmont, CA} } @article{Breslow84, author= {Breslow, N. E. and L. Edler and J. Berger}, year= {1984}, title= {A two-sample censored-data rank test for acceleration}, journal=biom, volume={40}, pages= {1049--1062} } @book{Broca66, author= {Broca, P. P.}, title= {Traites de Tumerus, volumes 1 and 2}, year = 1866, publisher={Asselin}, address={Paris} } @article{Breslow74, author= {Breslow, N. E.}, year= {1974}, title= {Covariance analysis of censored survival data}, journal=biom, volume={30}, pages= {89--99} } @article{Bryson81, author= {Bryson, M. C. and Johnson, M. E}, year= {1981}, title= {The incidence of monotone likelihood in the {C}ox model}, journal={Technometrics}, volume={23}, pages= {381--383} } @article{Chang82, author= {Chang, I. M. and Gelman, R. and Pagano, M.}, year= {1982}, title= {Corrected group prognostic curves and summary statistics}, journal={J. Chronic Diseases}, volume={35}, pages= {669--674} } @inproceedings{Clarkson89, author= {Clarkson, D. B.}, year= {1989}, title= {Computing extended maximum likelihood estimates in monotone likelihood {C}ox proportional-hazards models}, booktitle={Computer Science and Statistics: Proceedings of the 21st Symposium on the Interface}, publisher={American Statistical Association}, address={Alexandria, Virginia}, pages={464--469} } @article{DeLong94, author= {D. M. DeLong and G. H. Guirguis and Y. C. So}, year= {1994}, title= {Efficient computation of subset selection probabilities with application to {C}ox regression}, journal=biok, volume={81}, pages= {607--611} } @book{Delwiche98, author={Delwiche, L. D. and S. J. Slaughter}, title= {The Little {SAS} Book}, publisher={SAS Institute}, address={Cary, NC}, year= {1998} } @article{Ducrocq96, author={V. Ducrocq and G. Casella}, title= {A {B}ayesian analysis of mixed survival models}, year= {1996}, journal={Genet. Sel. Evol.}, volume={28}, pages= {505--529} } @article{Guo92, author= {G. Guo and G. Rodr\'{\i}guez}, year= {1992}, title= {Estimating a multivariate proportional hazards model for clustered data using the {EM} algorithm, with an application to child survival in {G}uatemala}, journal=jasa, volume={87}, pages= {969--976} } @article{Henderson99, author= {R. Henderson and P. Oman}, year= {1999}, title= {Effect of frailty on marginal regression estimates in survival analysis}, journal=jrssb, volume={61}, pages= {367--379} } @article{Hougaard86, author= {Hougaard, P.}, year= {1986}, title= {Survival models for heterogeneous populations derived from stable distributions}, journal=biok, volume={73}, pages= {387--396} } @article{Huster89, author= {W. J. Huster and R. Brookmeyer and S. G. Self}, year= {1989}, title= {Modelling paired survival data with covariates}, journal=biom, volume={45}, pages= {145--156} } @techreport{Jaeckel72, author= {Jaeckel, L.}, year= {1972}, title= {The infinitesimal jackknife}, institution={Bell Laboratories}, type={Memorandum}, number={MM 72-1215-11} } @article{Kavanagh94, author= {Kavanagh, B. F. and Wallrichs, S. and Dewitz, M. and Berry, D. and Currier, B. and Ilstrup, D. and Coventry, M. B.}, year= {1994}, title ={Charnley low-friction arthroplasty of the hip. {T}wenty-year results with cement}, journal={J. Arthroplasty}, volume={9}, pages= {229--234} } @article{Klein92, author= {Klein, J. P.}, year= {1992}, title ={Semiparametric estimation of random effects using the {C}ox model based on the {EM} algorithm}, journal=biom, volume={48}, pages= {795--806} } @article{Kyle93, author= {R. A. Kyle}, year= {1993}, title= {``{B}enign'' monoclonal gammopathy --- after 20 to 35 years of follow-up}, journal={Mayo Clinic Proceedings}, volume={68}, pages= {26--36} } @article{Kyle97, author= {R. A. Kyle}, year= {1997}, title= {Moncolonal gammopathy of undetermined significance and solitary plasmacytoma. {I}mplications for progression to overt multiple myeloma}, journal={Hematology/Oncology Clinics N. Amer.}, volume={11}, pages= {71--87} } @article{Leblanc92, author = {LeBlanc, M. and Crowley, J.}, title = {Relative risk trees for censored survival data}, journal = {Biometrics}, year = {1992}, volume={48}, pages = {411-425} } @article{Lee83, author= {K. L. Lee and Harrell, Jr., F. E. and H. D. Tolley and R. A. Rosati}, year= {1983}, title= {A comparison of test statistics for assessing the effects of concomitant variables in survival analysis}, journal=biom, volume={39}, pages= {341--350} } @article{Loprinzi94, author= {Loprinzi, C. L. and Laurie, J. A. and Wieand, H. S. and Krook, J. E. and Novotny, P. J. and Kugler, J. W. and Bartel, J. and Law, M. and Bateman, M. and Klatt, N. E. and Dose, A. M. and Etzell, P. S. and Nelimark, R. A. and Mailliard, J. A. and Moertel, C. G.}, year= {1994}, title= {Prospective evaluation of prognostic variables from patient-completed questionnaires}, journal={J. Clinical Oncol.}, volume={12}, pages= {601--607} } @article{Mahe99, author= {C{\'{e}}dric Mah{\'{e}} and Sylvie Chevret}, year= {1999}, title= {Estimating regression parameters and degree of dependence for multivariate failure time data}, journal=biom, volume={55}, pages= {1078--1084} } @article{Makuch82, author= {Makuch, R. W.}, year= {1982}, title= {Adjusted survival curve estimation using covariates}, journal={J. Chronic Disease}, volume={35}, pages= {437--443} } @article{Markus89, author= {Markus, B. H. and Dickson, E. R. and Grambsch, P. M. and Fleming, T. R. and Mazzaferro, V. and Klintmalm, G. B .G. and Wiesner, R. H. and VanThiel, D. H. and Starzl, T. E.}, year= {1989}, title= {Efficiency of liver transplantation in patients with primary biliary cirrhosis}, journal=NEJM, volume={320}, pages= {1709--1713} } @article{Miller83, author= {Miller, Jr., R.G.}, year= {1983}, title= {What price {Kaplan--Meier}?}, journal=biom, volume={39}, pages= {1077--1081} } @article{Murphy81, author= {Murphy, V. K. and Haywood, L. J.}, year= {1981}, title= {Survival analysis by sex, age group and hemotype in sickle cell disease}, journal={J. Chronic Diseases}, volume={34}, pages= {313--319} } @techreport{Pugh92, author= {M. Pugh and J. Robbins and S. Lipsitz and D. Harrington}, year= {1992}, title= {Inference in the {C}ox proportional hazards model with missing covariates}, institution={Department of Biostatistics, Harvard School of Public Health}, address={Boston}, number={758Z} } @article{Ripatti00, author= {Ripatti, S. and Palmgren, J.}, year= {2000}, title= {Estimation of multivariate frailty models using penalized partial likelihood}, journal=biom, volume={56}, pages={1016--1022} } @book{Searle71, author = {Searle, S.R.}, year= {1971}, title = {Linear Models}, publisher={Wiley}, address={New York}} @booklet{Seer81, key = {Surveillance}, title= {Surveillance, Epidemiology, and End Results: Incidence and Mortality Data, 1973--77}, year = {1981}, howpublished= {National Cancer Institute Monograph 57, U.S. Department of Health and Human Services, Public Health Service}, address= {National Cancer Institute, Bethesda, MD}, note = {NIH Publication No. 81-2330}, } @article{Sellers95, author= {T. A Sellers and V. E. Anderson and J. D. Potter and S. A. Bartow and P. L. Chen and L. Everson and R. A. King and C. C. Kuni and L. H. Kushi and P. G. McGovern and S. S. Rich and J. F. Whitbeck and G. L. Wiesner}, year= {1995}, title= {Epidemiologic and genetic follow-up study of 544 Minnesota breast cancer families: {D}esign and methods}, journal={Genetic Epidemiology}, volume={12}, pages= {417--429} } @article{Silverstein99, author= {M. D. Silverstein and Loftus, Jr., E. V. and W. J. Sandborn and W. J. Tremaine and B. G. Feagan and P. J. Nietert and W. S. Harmsen and A. R. Zinsmeister}, year= {1999}, title= {Clinical course and costs of care for {C}rohn's disease: {M}arkov model analysis of a population-based cohort}, journal={Gastroenterology}, volume={117}, pages= {49--57} } @article{Tsiatis81, author= {A. A. Tsiatis}, year= {1981}, title= {A large sample study of {C}ox's regression model}, journal= annals, volume={9}, pages= {93--108} } @article{Verweij94, author= {P. J .M. Verweij and Van Houwlingen, H. C.}, year= {1994}, title= {Penalized likelihood in {C}ox regression}, journal= statmed, volume={13}, pages= {2427--2436} } @ARTICLE{Volinsky98bayesianinformation, author = {Chris Volinsky and Adrian E. Raftery}, title = {Bayesian Information Criterion for Censored Survival Models}, journal = {Biometrics}, year = {1998}, volume = {56}, pages = {256--262} } @article{Wei90, author= {L. J. Wei and Z. Ying and D. Y. Lin}, year= {1990}, title= {Linear regression analysis of censored survival data based on rank tests}, journal= biok, volume={77}, pages= {845--851} } @article{Yau97, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1997}, title= {Use of generalised linear mixed models for the analysis of clustered survival data}, journal= {Biometrical Journal}, volume={39}, pages= {3--11} } @article{Yau98, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1998}, title= {{ML} and {REML} estimation in survival analysis with time dependent correlated frailty}, journal= statmed, volume={17}, pages= {1201--1213} } @article{Zahl96, author= {Zahl, Per-Henrik}, title= {A linear non-parametric model for the excess intensity}, year= {1996}, journal=scand, volume={23}, pages={353--364} } @booklet{smoke90, key={Department of Health}, title={The Health Benefits of Smoking Cessation}, year={1990}, howpublished ={Department of Health and Human Services. Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health}, note={DHHS Publication No (CDC)90-8416} } @booklet{lifeus40, title={United States Life Tables and Actuarial Tables, 1939--41}, author={Thomas N. E. Greville}, howpublished={Federal Security Agency, United States Public Health Service, National Office of Vital Statistics}, note={U.S. Government Printing Office, 1947} } @booklet{lifeus50, key={United States Lifetables 1950}, title={United States Life Tables for 1949--51}, howpublished={U.S. Department of Health, Education and Welfare, Public Health Service, National Office of Vital Statistics} } @booklet{lifeus60, key={United States Lifetables 1960}, title ={United States Lifetables 1959--61}, year = {1964}, howpublished= {Public Health Service Publication No. 1252}, note={Volume 1, Number 1} } @booklet{lifeus60b, key={Vital Statistics}, title ={Vital Statistics of the United States, 1960}, year = {1963}, howpublished= {U.S. Department of Health, Education, and Welfare, Public Health Service, National Center for Vital Statistics}, note={Volume 2A, Tabl3 3B} } @booklet{lifeus70, key={United States Lifetables 1970}, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-115}, note={Volume 1, Number 1} } @booklet{lifeus80, key={United States Lifetables 1990}, title ={U.S. Decennial Lifetables 1979--81}, year = {1985}, howpublished= {DHEW Publication No. PHS 85-1150-1}, note={Volume 1, Number 1} } @booklet{lifest60, key = {United States Lifetables 1959--61}, title = {Life tables for the geographic divisions of the {U}nited {S}tates: 1959--61}, note = {Vol. 1, number 3}, howpublished={National Center for Health Statistics, Public Health Service, Washington, U.S. Government Printing Office}, month=May, year={1965} } @booklet{lifemn60, title ={Minnesota State Lifetables 1959--61}, year = {1965}, howpublished= {Public Health Service Publication No. 1252}, note={Volume 2, Number 24} } @booklet{lifemn70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 24} } @booklet{lifefl70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 10} } @booklet{lifeaz70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 3} } @booklet{lifemn80, title ={U.S. Decennial Lifetables 1979--81}, year = {1985}, howpublished= {DHEW Publication No. PHS 86-1151-2451}, note={Volume 2, Number 24} } @book{Lifetable, author={National {C}enter for {H}ealth {S}tatistics}, year={1965}, title={Life tables for the geographic divisions of the {U}nited {S}tates: 1959-1961}, volume={1}, number={3}, publisher={US Government Printing Office, Washington} } survival/noweb/survfitms.Rnw0000644000175100001440000004535412466142446016033 0ustar hornikusers\subsubsection{Printing and plotting} The [[survfitms]] class differs from a [[survfit]], but many of the same methods nearly apply. <>= # Methods for survfitms objects <> <> @ The subscript method is a near copy of that for survfit objects, but with a slightly different set of components. The object could have strata and will almost always have multiple columns. If there is only one subscript it is preferentially associated with the strata, if there is no strata argument [[i]] will associate with the columns. If there are two subscripts the first goes with the strata. The little [[nmatch]] function allow the user to use either names or integer indices. The drop argument is important when strata get subscripted such that only one row remains and there are multiple columns: in that case we do \emph{not} want to lose the matrix nature of the result as it will lead to an invalid object. Otherwise we can drop columns freely <>= "[.survfitms" <- function(x, ..., drop=TRUE) { nmatch <- function(indx, target) { # This function lets R worry about character, negative, or logical subscripts # It always returns a set of positive integer indices temp <- 1:length(target) names(temp) <- target temp[indx] } if (missing(..1)) i<- NULL else i <- sort(..1) if (missing(..2)) j<- NULL else j <- ..2 if (is.null(x$strata)) { if (is.matrix(x$prev)) { # No strata, but a matrix of prevalence values # In this case, allow them to use a single i subscript as well if (is.null(j) && !is.null(i)) j <- i indx <- nmatch(j, x$states) if (any(is.na(indx))) stop("unmatched subscript", j[is.na(indx)]) else j <- as.vector(indx) x$states <- x$states[j] if (nrow(x$prev)==1 && length(j) > 1) drop<- FALSE x$prev <- x$prev[,j,drop=drop] x$cumhaz <- x$cumhaz[j,j,, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[,j,drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[,j,drop=drop] } else warning("Survfit object has only a single survival curve") } else { if (is.null(i)) keep <- seq(along.with=x$time) # rows to keep else { indx <- nmatch(i, names(x$strata)) #strata to keep if (any(is.na(indx))) stop(paste("strata", paste(i[is.na(indx)], collapse=' '), 'not matched')) # Now, i may not be in order: a user has curve[3:2] to reorder a plot # Hence the list/unlist construct which will reorder the data in the curves temp <- rep(1:length(x$strata), x$strata) keep <- unlist(lapply(i, function(x) which(temp==x))) if (length(i) <=1 && drop) x$strata <- NULL else x$strata <- x$strata[indx] x$n <- x$n[indx] x$time <- x$time[keep] x$n.risk <- x$n.risk[keep] x$n.event <- x$n.event[keep] x$n.censor<- x$n.censor[keep] } if (is.matrix(x$prev)) { # If [i,] selected only 1 row, don't collapse the columns if (length(keep) <2 && (is.null(j) || length(j) >1)) drop <- FALSE if (is.null(j)) { #only subscript rows (strata) x$prev <- x$prev[keep,,drop=drop] x$cumhaz <- x$cumhaz[,,keep, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,,drop=drop] if (!is.null(x$upper)) x$upper <-x$upper[keep,,drop=drop] if (!is.null(x$lower)) x$lower <-x$lower[keep,,drop=drop] } else { #subscript both rows (strata) and columns (states) indx <- nmatch(j, x$states) if (any(is.na(indx))) stop("unmatched subscript", j[indx]) else j <- as.vector(indx) x$states <- x$states[j] x$prev <- x$prev[keep,j, drop=drop] x$cumhaz <- x$cumhaz[j,j,keep, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[keep,j, drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[keep,j, drop=drop] } } else { x$prev <- x$prev[keep] x$cumhaz <- x$cumhaz[keep] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep] if (!is.null(x$upper)) x$upper <- x$upper[keep] if (!is.null(x$lower)) x$lower <- x$lower[keep] } } x } @ The summary.survfit and summary.survfitms functions share a large amount of code. Both are included here in order to have common source for the most subtle block of it, which has to do with selecting intermediate time points. <>= summary.survfit <- function(object, times, censored=FALSE, scale=1, extend=FALSE, rmean=getOption('survfit.rmean'), ...) { fit <- object if (!inherits(fit, 'survfit')) stop("summary.survfit can only be used for survfit objects") # The print.rmean option is depreciated, it is still listened # to in print.survfit, but ignored here if (is.null(rmean)) rmean <- "none" temp <- survmean(fit, scale=scale, rmean) table <- temp$matrix #for inclusion in the output list rmean.endtime <- temp$end.time if (!missing(times)) { if (!is.numeric(times)) stop ("times must be numeric") times <- sort(times) } # The fit$surv object is sometimes a vector and sometimes a # matrix. We calculate row indices first, and then deal # with the cases at the end. nsurv <- if (is.matrix(fit$surv)) nrow(fit$surv) else length(fit$surv) if (is.null(fit$strata)) { nstrat <- 1 stemp <- rep(1L, nsurv) strata.names <- "" } else { nstrat <- length(fit$strata) stemp <- rep(1:nstrat, fit$strata) strata.names <- names(fit$strata) } <> # Create an output structure if (length(indx1)==length(fit$time) && all(indx1 == seq(along=fit$time))) { temp <- object #no change temp$time <- temp$time/scale temp$table <- table if (!is.null(temp$strata)) temp$strata <- factor(stemp, labels=strata.names) } else if (missing(times)) { #default censor=FALSE case temp <- object temp$time <- temp$time[indx1]/scale temp$table <- table for (j in c("n.risk", "n.event", "n.censor", "n.enter", "surv", "std.err", "cumhaz", "lower", "upper")) { zed <- temp[[j]] if (!is.null(zed)) { if (is.matrix(zed)) temp[[j]] <- zed[indx1,,drop=FALSE] else temp[[j]] <- zed[indx1] } } if (!is.null(temp$strata)) temp$strata <- factor(stemp[indx1], levels=1:nstrat, labels=strata.names) } else { #times argument was given temp <- list(n=object$n, time=times/scale, n.risk=n.risk, n.event=n.event, conf.int=fit$conf.int, type=fit$type, table=table) if (!is.null(n.censor)) temp$n.censor <- n.censor if (!is.null(n.enter)) temp$n.enter <- n.enter if (!is.null(fit$start.time)) temp$start.time <- fit$start.time # why the rbind? The user may have specified a time point before # the first event, and indx1=1 indicates that case if (is.matrix(fit$surv)) { temp$surv <- rbind(1, fit$surv)[indx1,,drop=FALSE] if (!is.null(fit$std.err)) temp$std.err <- rbind(0, fit$std.err)[indx1,,drop=FALSE] if (!is.null(fit$lower)) { temp$lower <- rbind(1, fit$lower)[indx1,,drop=FALSE] temp$upper <- rbind(1, fit$upper)[indx1,,drop=FALSE] } if (!is.null(fit$cumhaz)) temp$cumhaz <- rbind(0, fit$cumhaz)[indx1,,drop=FALSE] } else { temp$surv <- c(1, fit$surv)[indx1] if (!is.null(fit$std.err)) temp$std.err <- c(0,fit$std.err)[indx1] if (!is.null(fit$lower)) { temp$lower <- c(1, fit$lower)[indx1] temp$upper <- c(1, fit$upper)[indx1] } if (!is.null(fit$cumhaz)) temp$cumhaz <- c(0, fit$cumhaz)[indx1] } if (!is.null(fit$strata)) { scount <- unlist(lapply(newtimes, length)) temp$strata <- factor(rep(1:nstrat, scount), levels=1:nstrat, labels=strata.names) } if (length(rmean.endtime)>0 && !is.na(rmean.endtime)) temp$rmean.endtime <- rmean.endtime temp$call <- fit$call if (!is.null(fit$na.action)) temp$na.action <- fit$na.action } if (!is.null(temp$std.err)) temp$std.err <- temp$std.err*temp$surv #std error of the survival curve class(temp) <- 'summary.survfit' temp } @ Grab rows: if there is no [[times]] argument it is easy <>= if (missing(times)) { # just pick off the appropriate rows of the output if (censored) indx1 <- seq(along=fit$time) else indx1 <- which(fit$n.event>0) } @ This second case is actual work, since may invovle ``in between'' points in the curves. Let's say that we have a line in the data for times 1,2, 5, and 6, and 8 and a user chose [[times=c(3,5, 9)]]. At time 3 we have \begin{itemize} \item nrisk[3] = value at the next time point >= 3 \item nevent[1] + nevent[2] = value since last printout line. However, if there are multiple strata the curves for all strata are laid end to end in a single vector; our first row for a curve needs to use all events since the start of the curve. \item ncensor works like nevent \item survival[2] = survival at the last time point <=3 \end{itemize} At time 5 we pick values directly off the data, since we match. At time 9 we report nothing if [[extend]] is FALSE, or the value at the end of the curve. In this case we need to calculate the number at risk ourselves, however. This logic works out best if we do it curve by curve. <>= else { # The one line function below might be opaque (even to me) -- # For n.event, we want to know the number since the last chosen # printout time point. Start with the curve of cumulative # events at c(0, stime) (the input time points), which is # the cumsum below; pluck off the values corresponding to our # time points, the [x] below; then get the difference since the # last chosen time point (or from 0, for the first chosen point). cfun <- function(x, data) diff(c(0, cumsum(c(0,data))[x])) # Process the curves one at a time, # adding the results for that curve onto a list, so the # number of events will be n.enter[[1]], n.enter[[2]], etc. # For the survival, stderr, and confidence limits it suffices # to create a single list 'indx1' containing a subscripting vector indx1 <- n.risk <- n.event <- newtimes <- vector('list', nstrat) n.enter <- vector('list', nstrat) n.censor<- vector('list', nstrat) n <- length(stemp) for (i in 1:nstrat) { who <- (1:n)[stemp==i] # the rows of the object for this strata stime <- fit$time[who] # First, toss any printing times that are outside our range if (is.null(fit$start.time)) mintime <- min(stime, 0) else mintime <- fit$start.time ptimes <- times[times >= mintime] if (!extend) { maxtime <- max(stime) ptimes <- ptimes[ptimes <= maxtime] } newtimes[[i]] <- ptimes # If we tack a -1 onto the front of the vector of survival # times, then indx1 is the subscript for that vector # corresponding to the list of "ptimes". If the input # data had stime=c(10,20) and ptimes was c(5,10,15,20), # the result would be 1,2,2,3. # For n.risk we want a slightly different index: 2,2,3,3. # "In between" times point to the next higher index for n.risk, # but the next lower one for survival. (Survival drops at time t, # the n.risk immediately afterwords at time t+0: you were at # risk just before you die, but not a moment after). The # extra point needs to be added at the end. # ntime <- length(stime) #number of points temp1 <- approx(c(mintime-1, stime), 0:ntime, xout=ptimes, method='constant', f=0, rule=2)$y indx1[[i]] <- ifelse(temp1==0, 1, 1+ who[pmax(1,temp1)]) # Why not just "who[temp1]" instead of who[pmax(1,temp1)] in the # line just above? When temp1 has zeros, the first expression # gives a vector that is shorter than temp1, and the ifelse # doesn't work right due to mismatched lengths. n.event[[i]] <- cfun(temp1+1, fit$n.event[who]) if (!is.null(fit$n.censor)) { n.censor[[i]] <- cfun(temp1+1, fit$n.censor[who]) j <- who[ntime] #last time point in the data last.n <- fit$n.risk[j] - (fit$n.event[j]+ fit$n.censor[j]) } else { # this is for the older survfit objects, which don't contain # n.censor. In this case, we don't know how many of the # people at the last time are censored then & how many go # on further. Assume we lose them all. Note normally # extend=FALSE, so this number isn't printed anyway. last.n <- 0 } # Compute the number at risk. If stime = 1,10, 20 and ptime=3,10, # 12, then temp1 = 2,2,3: the nrisk looking ahead # approx() doesn't work if stime is of length 1 if (ntime ==1) temp1 <- rep(1, length(ptimes)) else temp1 <- approx(stime, 1:ntime, xout=ptimes, method='constant', f=1, rule=2)$y n.risk[[i]] <- ifelse(ptimes>max(stime), last.n, fit$n.risk[who[temp1]]) } times <- unlist(newtimes) n.risk <- unlist(n.risk) n.event <- unlist(n.event) n.enter <- unlist(n.enter) #may be NULL n.censor<- unlist(n.censor) #may be NULL indx1 <- unlist(indx1) } @ Repeat the code for survfitms objects. The only real difference is the preservation of [[prev]] and [[cumhaz]] instead of [[surv]]. <>= summary.survfitms <- function(object, times, censored=FALSE, scale=1, extend=FALSE, rmean=getOption('survfit.rmean'), ...) { fit <- object if (!inherits(fit, 'survfitms')) stop("summary.survfitms can only be used for survfitms objects") if (is.null(rmean)) rmean <- "none" if (!missing(times)) { if (!is.numeric(times)) stop ("times must be numeric") times <- sort(times) } # add some temps to make survmean work object$surv <- 1-object$prev if (is.matrix(object$surv)) dimnames(object$surv) <- list(NULL, object$states) temp <- survmean(object, scale=scale, rmean) table <- temp$matrix #for inclusion in the output list rmean.endtime <- temp$end.time # The fit$prev object is usually a matrix but can be a vector # We calculate row indices first, and then deal # with the cases at the end. nprev <- if (is.matrix(fit$prev)) nrow(fit$prev) else length(fit$prev) if (is.null(fit$strata)) { nstrat <- 1 stemp <- rep(1L, nprev) strata.names <- "" } else { nstrat <- length(fit$strata) stemp <- rep(1:nstrat, fit$strata) strata.names <- names(fit$strata) } <> # Create an output structure if (length(indx1)== length(fit$time) && all(indx1 == seq(along=fit$time))) { temp <- object #no change temp$time <- temp$time/scale temp$table <- table if (!is.null(temp$strata)) temp$strata <- factor(stemp, levels=1:nstrat, labels=strata.names) } else if (missing(times)) { temp <- object temp$time <- temp$time[indx1]/scale temp$table <- table for (j in c("n.risk", "n.event", "n.censor", "n.enter", "prev", "std.err", "lower", "upper")) { zed <- temp[[j]] if (!is.null(zed)) { if (is.matrix(zed)) temp[[j]] <- zed[indx1,,drop=FALSE] else temp[[j]] <- zed[indx1] } } temp$cumhaz <- fit$cumhaz[,,indx1,drop=FALSE] if (!is.null(temp$strata)) temp$strata <- factor(stemp[indx1], levels=1:nstrat, labels=strata.names) } else { temp <- list(n=object$n, time=times/scale, n.risk=n.risk, n.event=n.event, conf.int=fit$conf.int, type=fit$type, table=table) if (!is.null(n.censor)) temp$n.censor <- n.censor if (!is.null(n.enter)) temp$n.enter <- n.enter if (!is.null(fit$start.time)) temp$start.time <- fit$start.time # why the rbind? The user may have specified a time point before # the first event, and indx1=1 indicates that case # the cumhaz array can't be done with a 1-liner if (is.matrix(fit$prev)) { temp$prev <- rbind(0, fit$prev)[indx1,,drop=FALSE] zz <- ifelse(indx1==1, NA, indx1-1) temp$cumhaz <- fit$cumhaz[,,zz, drop=FALSE] temp$cumhaz <- ifelse(is.na(temp$cumhaz), 0, temp$cumhaz) if (!is.null(fit$std.err)) temp$std.err <- rbind(0, fit$std.err)[indx1,,drop=FALSE] if (!is.null(fit$lower)) { temp$lower <- rbind(0, fit$lower)[indx1,,drop=FALSE] temp$upper <- rbind(0, fit$upper)[indx1,,drop=FALSE] } } else { temp$prev <- c(0, fit$prev[indx1]) temp$cumhaz <- c(0, fit$cumhaz[indx1]) if (!is.null(fit$std.err)) temp$std.err <- c(0, fit$std.err)[indx1] if (!is.null(fit$lower)) { temp$lower <- c(0, fit$lower)[indx1] temp$upper <- c(0, fit$upper)[indx1] } } if (!is.null(fit$strata)) { scount <- unlist(lapply(newtimes, length)) temp$strata <- factor(rep(1:nstrat, scount), levels=1:nstrat, labels=strata.names) } temp$call <- fit$call if (!is.null(fit$na.action)) temp$na.action <- fit$na.action } if (length(rmean.endtime)>0 && !is.na(rmean.endtime)) temp$rmean.endtime <- rmean.endtime class(temp) <- "summary.survfitms" temp } @ survival/noweb/coxsurv2.Rnw0000644000175100001440000005764512470201064015555 0ustar hornikusers% % Second part of coxsurv.Rnw, broken in two to make it easier for me % to work with emacs. Now, we're ready to do the main compuation. %' Before this revision (the one documented here using noweb) there were three C routines used in calculating survival after a Cox model \begin{enumerate} \item agsurv1 creates a single curve, but for the most general case of a \emph{covariate path}. It is used for time dependent covariates. \item agsurv2 creates a set of curves. These curves are for a fixed covariate set, although (start, stop] data is supported. If there were 3 strata in the fit and 4 covariate sets are given, the result will be 12 curves. \item agsurv3 is used to create population survival curves. The result is average survival curve (for 3 different definitions of 'average'). If there were 3 strata and 100 subjects, the first curve returned would be the average for those 100 individual curves in strata 1, the second for strata 2, and the third for strata 3. \end{enumerate} In June 2010 the first two were re-written in (mostly) R, in the process of adding functionality and repairing some flaws in the computation of a weighted variance. In effect, the changes are similar to the rewrite of the survfitKM function a few years ago. Computations are separate for each strata, and each strata will have a different number of time points in the result. Thus we can't preallocate a matrix. Instead we generate an empty list, %' one per strata, and then populate it with the survival curves. At the end we unlist the individual components one by one. This is memory efficient, the number of curves is usually small enough that the "for" loop is no great cost, and it's easier to see what's going on than C code. First, compute the baseline survival curves for each strata. If the strata was a factor we want to leave it in the same order, otherwise sort it. This fitting routine was set out as a separate function for the sake of the rms package. They want to utilize the computation, but have a diffferent recreation process for the x and y data. <>= survfitcoxph.fit <- function(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat, id, y2, strata2, unlist=TRUE) { if (is.factor(strata)) ustrata <- levels(strata) else ustrata <- sort(unique(strata)) nstrata <- length(ustrata) survlist <- vector('list', nstrata) names(survlist) <- ustrata for (i in 1:nstrata) { indx <- which(strata== ustrata[i]) survlist[[i]] <- agsurv(y[indx,,drop=F], x[indx,,drop=F], wt[indx], risk[indx], survtype, vartype) } <> if (unlist) { if (length(result)==1) { # the no strata case if (se.fit) result[[1]][c("n", "time", "n.risk", "n.event", "n.censor", "surv", "cumhaz", "std.err")] else result[[1]][c("n", "time", "n.risk", "n.event", "n.censor", "surv", "cumhaz")] } else { temp <-list(n = unlist(lapply(result, function(x) x$n), use.names=FALSE), time= unlist(lapply(result, function(x) x$time), use.names=FALSE), n.risk= unlist(lapply(result, function(x) x$n.risk), use.names=FALSE), n.event= unlist(lapply(result, function(x) x$n.event), use.names=FALSE), n.censor=unlist(lapply(result, function(x) x$n.censor), use.names=FALSE), strata = sapply(result, function(x) length(x$time))) names(temp$strata) <- names(result) if ((missing(id) || is.null(id)) && nrow(x2)>1) { temp$surv <- t(matrix(unlist(lapply(result, function(x) t(x$surv)), use.names=FALSE), nrow= nrow(x2))) dimnames(temp$surv) <- list(NULL, row.names(x2)) temp$cumhaz <- t(matrix(unlist(lapply(result, function(x) t(x$cumhaz)), use.names=FALSE), nrow= nrow(x2))) if (se.fit) temp$std.err <- t(matrix(unlist(lapply(result, function(x) t(x$std.err)), use.names=FALSE), nrow= nrow(x2))) } else { temp$surv <- unlist(lapply(result, function(x) x$surv), use.names=FALSE) temp$cumhaz <- unlist(lapply(result, function(x) x$cumhaz), use.names=FALSE) if (se.fit) temp$std.err <- unlist(lapply(result, function(x) x$std.err), use.names=FALSE) } temp } } else { names(result) <- ustrata result } } @ In an ordinary survival curve object with multiple strata, as produced by [[survfitKM]], the time, survival and etc components are each a single vector that contains the results for strata 1, followed by strata 2, \ldots. The strata compontent is a vector of integers, one per strata, that gives the number of elements belonging to each stratum. The reason is that each strata will have a different number of observations, so that a matrix form was not viable, and the underlying C routines were not capable of handling lists (the code predates the .Call function by a decade). The underlying computation of [[survfitcoxph.fit]] naturally creates the list form, we unlist it to [[survfit]] form as our last action unless the caller requests otherwise. For [[individual=FALSE]] we have a second dimension, namely each of the target covariate sets (if there are multiples). Each of these generates a unique set of survival and variance(survival) values, but all of the same size since each uses all the strata. The final output structure in this case has single vectors for the time, number of events, number censored, and number at risk values since they are common to all the curves, and a marix of survival and variance estimates, one column for each of the distinct target values. If $\Lambda_0$ is the baseline cumulative hazard from the above calculation, then $r_i \Lambda_0$ is the cumulative hazard for the $i$th new risk score $r_i$. The variance has two parts, the first of which is $r_i^2 H_1$ where $H_1$ is returned from the [[agsurv]] routine, and the second is \begin{align*} H_2(t) =& d'(t) V d(t) \\ %' d(t) = \int_0^t [z- \overline x(s)] d\Lambda(s) \end{align*} $V$ is the variance matrix for $\beta$ from the fitted Cox model, and $d(t)$ is the distance between the target covariate $z$ and the mean of the original data, summed up over the interval from 0 to $t$. Essentially the variance in $\hat \beta$ has a larger influence when prediction is far from the mean. The function below takes the basic curve from the list and multiplies it out to matrix form. <>= expand <- function(fit, x2, varmat, se.fit) { if (survtype==1) surv <- cumprod(fit$surv) else surv <- exp(-fit$cumhaz) if (is.matrix(x2) && nrow(x2) >1) { #more than 1 row in newdata fit$surv <- outer(surv, newrisk, '^') dimnames(fit$surv) <- list(NULL, row.names(x2)) if (se.fit) { varh <- matrix(0., nrow=length(fit$varhaz), ncol=nrow(x2)) for (i in 1:nrow(x2)) { dt <- outer(fit$cumhaz, x2[i,], '*') - fit$xbar varh[,i] <- (cumsum(fit$varhaz) + rowSums((dt %*% varmat)* dt))* newrisk[i]^2 } fit$std.err <- sqrt(varh) } fit$cumhaz <- outer(fit$cumhaz, newrisk, '*') } else { fit$surv <- surv^newrisk if (se.fit) { dt <- outer(fit$cumhaz, c(x2)) - fit$xbar varh <- (cumsum(fit$varhaz) + rowSums((dt %*% varmat)* dt)) * newrisk^2 fit$std.err <- sqrt(varh) } fit$cumhaz <- fit$cumhaz * newrisk } fit } @ In the lines just above: I have a matrix [[dt]] with one row per death time and one column per variable. For each row $d_i$ separately we want the quadratic form $d_i V d_i'$. The first matrix product can %' be done for all rows at once: found in the inner parenthesis. Ordinary (not matrix) multiplication followed by rowsums does the rest in one fell swoop. Now, if [[id]] is missing we can simply apply the [[expand]] function to each strata. For the case with [[id]] not missing, we create a single survival curve for each unique id (subject). A subject will spend blocks of time with different covariate sets, sometimes even jumping between strata. Retrieve each one and save it into a list, and then sew them together end to end. The [[n]] component is the number of observations in the strata --- but this subject might visit several. We report the first one they were in for printout. The [[time]] component will be cumulative on this subject's scale. %' Counting this is a bit trickier than I first thought. Say that the subject's first interval goes from 1 to 10, with observed time points in that interval at 2, 5, and 7, and a second interval from 12 to 20 with observed time points in the data of 15 and 18. On the subject's time scale things happen at days 1, 4, 6, 12 and 15. The deltas saved below are 2-1, 5-2, 7-5, 3+ 14-12, 17-14. Note the 3+ part, kept in the [[timeforward]] variable. Why all this ``adding up'' nuisance? If the subject spent time in two strata, the second one might be on an internal time scale of `time since entering the strata'. The two intervals in newdata could be 0--10 followed by 0--20. Time for the subject can't go backwards though: the change %` between internal/external time scales is a bit like following someone who was stepping back and forth over the international date line. In the code the [[indx]] variable points to the set of times that the subject was present, for this row of the new data. Note the $>$ on one end and $\le$ on the other. If someone's interval 1 was 0--10 and interval 2 was 10--20, and there happened to be a jump in the baseline survival curve at exactly time 10 (someone else died), that jump is counted only in the first interval. <>= if (missing(id) || is.null(id)) result <- lapply(survlist, expand, x2, varmat, se.fit) else { onecurve <- function(slist, x2, y2, strata2, newrisk, se.fit) { ntarget <- nrow(x2) #number of different time intervals surv <- vector('list', ntarget) n.event <- n.risk <- n.censor <- varh1 <- varh2 <- time <- surv hazard <- vector('list', ntarget) stemp <- as.integer(strata2) timeforward <- 0 for (i in 1:ntarget) { slist <- survlist[[stemp[i]]] indx <- which(slist$time > y2[i,1] & slist$time <= y2[i,2]) if (length(indx)==0) { timeforward <- timeforward + y2[i,2] - y2[i,1] # No deaths or censors in user interval. Possible # user error, but not uncommon at the tail of the curve. } else { time[[i]] <- diff(c(y2[i,1], slist$time[indx])) #time increments time[[i]][1] <- time[[i]][1] + timeforward timeforward <- y2[i,2] - max(slist$time[indx]) hazard[[i]] <- slist$hazard[indx]*newrisk[i] if (survtype==1) surv[[i]] <- slist$surv[indx]^newrisk[i] n.event[[i]] <- slist$n.event[indx] n.risk[[i]] <- slist$n.risk[indx] n.censor[[i]]<- slist$n.censor[indx] dt <- outer(slist$cumhaz[indx], x2[i,]) - slist$xbar[indx,,drop=F] varh1[[i]] <- slist$varhaz[indx] *newrisk[i]^2 varh2[[i]] <- rowSums((dt %*% varmat)* dt) * newrisk[i]^2 } } cumhaz <- cumsum(unlist(hazard)) if (survtype==1) surv <- cumprod(unlist(surv)) #increments (K-M) else surv <- exp(-cumhaz) if (se.fit) list(n=as.vector(table(strata)[stemp[1]]), time=cumsum(unlist(time)), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor= unlist(n.censor), surv = surv, cumhaz= cumhaz, std.err = sqrt(cumsum(unlist(varh1)) + unlist(varh2))) else list(n=as.vector(table(strata)[stemp[1]]), time=cumsum(unlist(time)), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor= unlist(n.censor), surv = surv, cumhaz= cumhaz) } if (all(id ==id[1])) { result <- list(onecurve(survlist, x2, y2, strata2, newrisk, se.fit)) } else { uid <- unique(id) result <- vector('list', length=length(uid)) for (i in 1:length(uid)) { indx <- which(id==uid[i]) result[[i]] <- onecurve(survlist, x2[indx,,drop=FALSE], y2[indx,,drop=FALSE], strata2[indx], newrisk[indx], se.fit) } names(result) <- uid } } @ Next is the code for the [[agsurv]] function, which actually does the work. The estimates of survival are the Kalbfleisch-Prentice (KP), Breslow, and Efron. Each has an increment at each unique death time. First a bit of notation: $Y_i(t)$ is 1 if bservation $i$ is ``at risk'' at time $t$ and 0 otherwise. For a simple surivival ([[ncol(y)==2]]) a subject is at risk until the time of censoring or death (first column of [[y]]). For (start, stop] data ([[ncol(y)==3]]) a subject becomes a part of the risk set at start+0 and stays through stop. $dN_i(t)$ will be 1 if subject $i$ had an event at time $t$. The risk score for each subject is $r_i = \exp(X_i \beta)$. The Breslow increment at time $t$ is $\sum w_i dN_i(t) / \sum w_i r_i Y_i(t)$, the number of events at time $t$ over the number at risk at time $t$. The final survival is [[exp(-cumsum(increment))]]. The Kalbfleish-Prentice increment is a multiplicative term $z$ which is the solution to the equation $$ \sum w_i r_i Y_i(t) = \sum dN_i(t) w_i \frac{r_i}{1- z(t)^{r_i}} $$ The left hand side is the weighted number at risk at time $t$, the right hand side is a sum over the tied events at that time. If there is only one event the equation has a closed form solution. If not, and knowing the solution must lie between 0 and 1, we do 35 steps of bisection to get a solution within 1e-8. An alternative is to use the -log of the Breslow estimate as a starting estimate, which is faster but requires a more sophisticated iteration logic. The final curve is $\prod_t z(t)^{r_c}$ where $r_c$ is the risk score for the target subject. The Efron estimate can be viewed as a modified Breslow estimate under the assumption that tied deaths are not really tied -- we just don't know the %' order. So if there are 3 subjects who die at some time $t$ we will have three psuedo-terms for $t$, $t+\epsilon$, and $t+ 2\epsilon$. All 3 subjects are present for the denominator of the first term, 2/3 of each for the second, and 1/3 for the third terms denominator. All contribute 1/3 of the weight to each numerator (1/3 chance they were the one to die there). The formulas will require $\sum w_i dN_i(t)$, $\sum w_ir_i dN_i(t)$, and $\sum w_i X_i dN_i(t)$, i.e., the sums only over the deaths. For simple survival data the risk sum $\sum w_i r_i Y_i(t)$ for all the unique death times $t$ is fast to compute as a cumulative sum, starting at the longest followup time an summing towards the shortest. There are two algorithms for (start, stop] data. \begin{itemize} \item Do a separate sum at each death time. The problem is for very large data sets. For each death time the selection [[who <- (start=t)]] is $O(n)$ and can take more time then all the remaining calculations together. \item Use the difference of two cumulative sums, one ordered by start time and one ordered by stop time. This is $O(2n)$ for the intial sums. The problem here is potential round off error if the sums get large, which can happen if the time scale were very, very finely divided. This issue is mostly precluded by subtracting means first. \end{itemize} We compute the extended number still at risk --- all whose stop time is $\ge$ each unique death time --- in the vector [[xin]]. From this we have to subtract all those who haven't actually entered yet %' found in [[xout]]. Remember that (3,20] enters at time 3+. The total at risk at any time is the difference between them. Output is only for the stop times; a call to approx is used to reconcile the two time sets. The [[irisk]] vector is for the printout, it is a sum of weighted counts rather than weighted risk scores. <>= agsurv <- function(y, x, wt, risk, survtype, vartype) { nvar <- ncol(as.matrix(x)) status <- y[,ncol(y)] dtime <- y[,ncol(y) -1] death <- (status==1) time <- sort(unique(dtime)) nevent <- as.vector(rowsum(wt*death, dtime)) ncens <- as.vector(rowsum(wt*(!death), dtime)) wrisk <- wt*risk rcumsum <- function(x) rev(cumsum(rev(x))) # sum from last to first nrisk <- rcumsum(rowsum(wrisk, dtime)) irisk <- rcumsum(rowsum(wt, dtime)) if (ncol(y) ==2) { temp2 <- rowsum(wrisk*x, dtime) xsum <- apply(temp2, 2, rcumsum) } else { delta <- min(diff(time))/2 etime <- c(sort(unique(y[,1])), max(y[,1])+delta) #unique entry times indx <- approx(etime, 1:length(etime), time, method='constant', rule=2, f=1)$y esum <- rcumsum(rowsum(wrisk, y[,1])) #not yet entered nrisk <- nrisk - c(esum,0)[indx] irisk <- irisk - c(rcumsum(rowsum(wt, y[,1])),0)[indx] xout <- apply(rowsum(wrisk*x, y[,1]), 2, rcumsum) #not yet entered xin <- apply(rowsum(wrisk*x, dtime), 2, rcumsum) # dtime or alive xsum <- xin - (rbind(xout,0))[indx,,drop=F] } ndeath <- rowsum(status, dtime) #unweighted death count @ The KP estimate requires a short C routine to do the iteration efficiently, and the Efron estimate needs a second C routine to efficiently compute the partial sums. <>= ntime <- length(time) if (survtype ==1) { #Kalbfleisch-Prentice indx <- (which(status==1))[order(dtime[status==1])] #deaths km <- .C(Cagsurv4, as.integer(ndeath), as.double(risk[indx]), as.double(wt[indx]), as.integer(ntime), as.double(nrisk), inc = double(ntime)) } if (survtype==3 || vartype==3) { # Efron approx xsum2 <- rowsum((wrisk*death) *x, dtime) erisk <- rowsum(wrisk*death, dtime) #risk score sums at each death tsum <- .C(Cagsurv5, as.integer(length(nevent)), as.integer(nvar), as.integer(ndeath), as.double(nrisk), as.double(erisk), as.double(xsum), as.double(xsum2), sum1 = double(length(nevent)), sum2 = double(length(nevent)), xbar = matrix(0., length(nevent), nvar)) } haz <- switch(survtype, nevent/nrisk, nevent/nrisk, nevent* tsum$sum1) varhaz <- switch(vartype, nevent/(nrisk * ifelse(nevent>=nrisk, nrisk, nrisk-nevent)), nevent/nrisk^2, nevent* tsum$sum2) xbar <- switch(vartype, (xsum/nrisk)*haz, (xsum/nrisk)*haz, nevent * tsum$xbar) result <- list(n= nrow(y), time=time, n.event=nevent, n.risk=irisk, n.censor=ncens, hazard=haz, cumhaz=cumsum(haz), varhaz=varhaz, ndeath=ndeath, xbar=apply(matrix(xbar, ncol=nvar),2, cumsum)) if (survtype==1) result$surv <- km$inc result } @ The arguments to this function are the number of unique times n, which is the length of the vectors ndeath (number at each time), denom, and the returned vector km. The risk and wt vectors contain individual values for the subjects with an event. Their length will be equal to sum(ndeath). <>= #include "survS.h" #include "survproto.h" void agsurv4(Sint *ndeath, double *risk, double *wt, Sint *sn, double *denom, double *km) { int i,j,k, l; int n; /* number of unique death times */ double sumt, guess, inc; n = *sn; j =0; for (i=0; i>= #include "survS.h" void agsurv5(Sint *n2, Sint *nvar2, Sint *dd, double *x1, double *x2, double *xsum, double *xsum2, double *sum1, double *sum2, double *xbar) { double temp; int i,j, k, kk; double d; int n, nvar; n = n2[0]; nvar = nvar2[0]; for (i=0; i< n; i++) { d = dd[i]; if (d==1){ temp = 1/x1[i]; sum1[i] = temp; sum2[i] = temp*temp; for (k=0; k< nvar; k++) xbar[i+ n*k] = xsum[i + n*k] * temp*temp; } else { temp = 1/x1[i]; for (j=0; j>= pyears <- function(formula, data, weights, subset, na.action, rmap, ratetable, scale=365.25, expect=c('event', 'pyears'), model=FALSE, x=FALSE, y=FALSE, data.frame=FALSE) { <> <> <> } @ Start out with the standard model processing, which involves making a copy of the input call, but keeping only the arguments we want. We then process the special argument [[rmap]]. This is discussed in the section on the [[survexp]] function so we need not repeat the explantation here. <>= expect <- match.arg(expect) call <- match.call() m <- match.call(expand.dots=FALSE) m <- m[c(1, match(c('formula', 'data', 'weights', 'subset', 'na.action'), names(m), nomatch=0))] m[[1]] <- as.name("model.frame") Terms <- if(missing(data)) terms(formula, 'ratetable') else terms(formula, 'ratetable',data=data) if (any(attr(Terms, 'order') >1)) stop("Pyears cannot have interaction terms") rate <- attr(Terms, "specials")$ratetable if (length(rate) >0 || !missing(rmap) || !missing(ratetable)) { has.ratetable <- TRUE if(length(rate) > 1) stop("Can have only 1 ratetable() call in a formula") if (missing(ratetable)) stop("No rate table specified") <> } else has.ratetable <- FALSE if (is.R()) m <- eval(m, parent.frame()) else m <- eval(m, sys.parent()) Y <- model.extract(m, 'response') if (is.null(Y)) stop ("Follow-up time must appear in the formula") if (!is.Surv(Y)){ if (any(Y <0)) stop ("Negative follow up time") Y <- as.matrix(Y) if (ncol(Y) >2) stop("Y has too many columns") } else { stype <- attr(Y, 'type') if (stype == 'right') { if (any(Y[,1] <0)) stop("Negative survival time") nzero <- sum(Y[,1]==0 & Y[,2] ==1) if (nzero >0) warning(paste(nzero, "observations with an event and 0 follow-up time,", "any rate calculations are statistically questionable")) } else if (stype != 'counting') stop("Only right-censored and counting process survival types are supported") } n <- nrow(Y) if (is.null(n) || n==0) stop("Data set has 0 observations") weights <- model.extract(m, 'weights') if (is.null(weights)) weights <- rep(1.0, n) @ The next step is to check out the ratetable. For a population rate table a set of consistency checks is done by the [[match.ratetable]] function, giving a set of sanitized indices [[R]]. This function wants characters turned to factors. For a Cox model [[R]] will be a model matix whose covariates are coded in exactly the same way that variables were coded in the original Cox model. We call the model.matrix.coxph function so as not to have to repeat the steps found there (remove cluster statements, etc). <>= # rdata contains the variables matching the ratetable if (has.ratetable) { rdata <- data.frame(eval(rcall, m), stringsAsFactors=TRUE) if (is.ratetable(ratetable)) { israte <- TRUE rtemp <- match.ratetable(rdata, ratetable) R <- rtemp$R } else if (inherits(ratetable, 'coxph')) { israte <- FALSE Terms <- ratetable$terms if (!is.null(attr(Terms, 'offset'))) stop("Cannot deal with models that contain an offset") strats <- attr(Terms, "specials")$strata if (length(strats)) stop("pyears cannot handle stratified Cox models") if (any(names(m[,rate]) != attr(ratetable$terms, 'term.labels'))) stop("Unable to match new data to old formula") R <- model.matrix.coxph(ratetable, data=rdata) } else stop("Invalid ratetable") } @ Now we process the non-ratetable variables. Those of class [[tcut]] set up time-dependent classes. For these the cutpoints attribute sets the intervals, if there were 4 cutpoints of 1, 5,6, and 10 the 3 intervals will be 1-5, 5-6 and 6-10, and odims will be 3. All other variables are treated as factors. <>= ovars <- attr(Terms, 'term.labels') if (length(ovars)==0) { # no categories! X <- rep(1,n) ofac <- odim <- odims <- ocut <- 1 } else { odim <- length(ovars) ocut <- NULL odims <- ofac <- double(odim) X <- matrix(0, n, odim) outdname <- vector("list", odim) for (i in 1:odim) { temp <- m[[ovars[i]]] if (inherits(temp, 'tcut')) { X[,i] <- temp temp2 <- attr(temp, 'cutpoints') odims[i] <- length(temp2) -1 ocut <- c(ocut, temp2) ofac[i] <- 0 outdname[[i]] <- attr(temp, 'labels') } else { temp2 <- as.factor(temp) X[,i] <- temp2 temp3 <- levels(temp2) odims[i] <- length(temp3) ofac[i] <- 1 outdname[[i]] <- temp3 } } } @ Now do the computations. The code above has separated out the variables into 3 groups: \begin{itemize} \item The variables in the rate table. These determine where we \emph{start} in the rate table with respect to retrieving the relevant death rates. For the US table [[survexp.us]] this will be the date of study entry, age (in days) at study entry, and sex of each subject. \item The variables on the right hand side of the model. These are interpreted almost identically to a call to [[table]], with special treatment for those of class \emph{tcut}. \item The response variable, which tells the number of days of follow-up and optionally the status at the end of follow-up. \end{itemize} Start with the rate table variables. There is an oddity about US rate tables: the entry for age (year=1970, age=55) contains the daily rate for anyone who turns 55 in that year, from their birthday forward for 365 days. So if your birthday is on Oct 2, the 1970 table applies from 2Oct 1970 to 1Oct 1971. The underlying C code wants to make the 1970 rate table apply from 1Jan 1970 to 31Dec 1970. The easiest way to finess this is to fudge everyone's enter-the-study date. If you were born in March but entered in April, make it look like you entered in Febuary; that way you get the first 11 months at the entry year's rates, etc. The birth date is entry date - age in days (based on 1/1/1960). The other aspect of the rate tables is that ``older style'' tables, those that have the factor attribute, contained only decennial data which the C code would interpolate on the fly. The value of [[atts$factor]] was 10 indicating that there are 10 years in the interpolation interval. The newer tables do not do this and the C code is passed a 0/1 for continuous (age and year) versus discrete (sex, race). <>= ocut <-c(ocut,0) #just in case it were of length 0 osize <- prod(odims) if (has.ratetable) { #include expected atts <- attributes(ratetable) cuts <- atts$cutpoints if (is.null(atts$type)) { #old stlye table rfac <- atts$factor us.special <- (rfac >1) } else { rfac <- 1*(atts$type ==1) us.special <- (atts$type==4) } if (any(us.special)) { #special handling for US pop tables # Now, the 'entry' date on a US rate table is the number of days # since 1/1/1960, and the user data has been aligned to the # same system by match.ratetable and marked as "year". # The birth date is entry date - age in days (based on 1/1/1960). # I don't much care which date functions I use to do the arithmetic # below. Unfortunately R and Splus don't share one. My "date" # class is simple, but is also one of the earlier date class # attempts, has less features than others, and will one day fade # away; so I don't want to depend on it alone. # cols <- match(c("age", "year"), atts$dimid) if (any(is.na(cols))) stop("Ratetable does not have expected shape") if (exists("as.Date")) { # true for modern version of R bdate <- as.Date('1960/1/1') + (R[,cols[2]] - R[,cols[1]]) byear <- format(bdate, "%Y") offset <- bdate - as.Date(paste(byear, "01/01", sep='/'), origin="1960/01/01") } #else if (exists('month.day.year')) { # Splus, usually # bdate <- R[,cols[2]] - R[,cols[1]] # byear <- month.day.year(bdate)$year # offset <- bdate - julian(1,1,byear) # } #else if (exists('date.mdy')) { # Therneau's date class is available # bdate <- as.date(R[,cols[2]] - R[,cols[1]]) # byear <- date.mdy(bdate)$year # offset <- bdate - mdy.date(1,1,byear) # } else stop("Can't find an appropriate date class\n") R[,cols[2]] <- R[,cols[2]] - offset # Doctor up "cutpoints" - only needed for old style rate tables # for which the C code does interpolation on the fly if (any(rfac >1)) { temp <- which(us.special) nyear <- length(cuts[[temp]]) nint <- rfac[temp] #intervals to interpolate over cuts[[temp]] <- round(approx(nint*(1:nyear), cuts[[temp]], nint:(nint*nyear))$y - .0001) } } docount <- is.Surv(Y) temp <- .C(Cpyears1, as.integer(n), as.integer(ncol(Y)), as.integer(is.Surv(Y)), as.double(Y), as.double(weights), as.integer(length(atts$dim)), as.integer(rfac), as.integer(atts$dim), as.double(unlist(cuts)), as.double(ratetable), as.double(R), as.integer(odim), as.integer(ofac), as.integer(odims), as.double(ocut), as.integer(expect=='event'), as.double(X), pyears=double(osize), pn =double(osize), pcount=double(if(docount) osize else 1), pexpect=double(osize), offtable=double(1))[18:22] } else { #no expected docount <- as.integer(ncol(Y) >1) temp <- .C(Cpyears2, as.integer(n), as.integer(ncol(Y)), as.integer(docount), as.double(Y), as.double(weights), as.integer(odim), as.integer(ofac), as.integer(odims), as.double(ocut), as.double(X), pyears=double(osize), pn =double(osize), pcount=double(if (docount) osize else 1), offtable=double(1)) [11:14] } @ Create the output object. <>= if (data.frame) { # Create a data frame as the output, rather than a set of # rate tables keep <- (temp$pyears >0) # what rows to keep in the output names(outdname) <- ovars if (length(outdname) ==1) { # if there is only one variable, the call to "do.call" loses # the variable name, since expand.grid returns a factor df <- data.frame((outdname[[1]])[keep], pyears= temp$pyears[keep]/scale, n = temp$pn[keep]) names(df) <- c(names(outdname), 'pyears', 'n') } else { df <- cbind(do.call("expand.grid", outdname)[keep,], pyears= temp$pyears[keep]/scale, n = temp$pn[keep]) } row.names(df) <- 1:nrow(df) if (has.ratetable) df$expected <- temp$pexpect[keep] if (expect=='pyears') df$expected <- df$expected/scale if (docount) df$event <- temp$pcount[keep] out <- list(call=call, data= df, offtable=temp$offtable/scale) if (has.ratetable && !is.null(rtemp$summ)) out$summary <- rtemp$summ } else if (prod(odims) ==1) { #don't make it an array out <- list(call=call, pyears=temp$pyears/scale, n=temp$pn, offtable=temp$offtable/scale) if (has.ratetable) { out$expected <- temp$pexpect if (expect=='pyears') out$expected <- out$expected/scale if (!is.null(rtemp$summ)) out$summary <- rtemp$summ } if (docount) out$event <- temp$pcount } else { out <- list(call = call, pyears= array(temp$pyears/scale, dim=odims, dimnames=outdname), n = array(temp$pn, dim=odims, dimnames=outdname), offtable = temp$offtable/scale) if (has.ratetable) { out$expected <- array(temp$pexpect, dim=odims, dimnames=outdname) if (expect=='pyears') out$expected <- out$expected/scale if (!is.null(rtemp$summ)) out$summary <- rtemp$summ } if (docount) out$event <- array(temp$pcount, dim=odims, dimnames=outdname) } out$observations <- nrow(m) na.action <- attr(m, "na.action") if (length(na.action)) out$na.action <- na.action if (model) out$model <- m else { if (x) out$x <- X if (y) out$y <- Y } oldClass(out) <- 'pyears' out @ survival/noweb/coxph.Rnw0000644000175100001440000006570612545011561015104 0ustar hornikusers\section{Cox Models} \subsection{Coxph} The [[coxph]] routine is the underlying basis for all the models. The source was converted to noweb when adding time-transform terms. The call starts out with the basic building of a model frame and proceeds from there. <>= #tt <- function(x) x coxph <- function(formula, data, weights, subset, na.action, init, control, ties= c("efron", "breslow", "exact"), singular.ok =TRUE, robust=FALSE, model=FALSE, x=FALSE, y=TRUE, tt, method=ties, ...) { ties <- match.arg(ties) Call <- match.call() # create a call to model.frame() that contains the formula (required) # and any other of the relevant optional arguments # then evaluate it in the proper frame indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(Call), nomatch=0) if (indx[1] ==0) stop("A formula argument is required") temp <- Call[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called special <- c("strata", "cluster", "tt") temp$formula <- if(missing(data)) terms(formula, special) else terms(formula, special, data=data) # Make "tt" visible for coxph formulas, without making it visible elsewhere if (!is.null(attr(temp$formula, "specials")$tt)) { coxenv <- new.env(parent= environment(formula)) assign("tt", function(x) x, env=coxenv) environment(temp$formula) <- coxenv } mf <- eval(temp, parent.frame()) if (nrow(mf) ==0) stop("No (non-missing) observations") Terms <- terms(mf) ## We want to pass any ... args to coxph.control, but not pass things ## like "dats=mydata" where someone just made a typo. The use of ... ## is simply to allow things like "eps=1e6" with easier typing extraArgs <- list(...) if (length(extraArgs)) { controlargs <- names(formals(coxph.control)) #legal arg names indx <- pmatch(names(extraArgs), controlargs, nomatch=0L) if (any(indx==0L)) stop(gettextf("Argument %s not matched", names(extraArgs)[indx==0L]), domain = NA) } if (missing(control)) control <- coxph.control(...) Y <- model.extract(mf, "response") if (!inherits(Y, "Surv")) stop("Response must be a survival object") type <- attr(Y, "type") if (type!='right' && type!='counting') stop(paste("Cox model doesn't support \"", type, "\" survival data", sep='')) data.n <- nrow(Y) #remember this before any time transforms <> # The time transform will expand the data frame mf. To do this # it needs Y and the strata. Everything else (cluster, offset, weights) # should be extracted after the transform # strats <- attr(Terms, "specials")$strata if (length(strats)) { stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } timetrans <- attr(Terms, "specials")$tt if (missing(tt)) tt <- NULL if (length(timetrans)) { <> } cluster<- attr(Terms, "specials")$cluster if (length(cluster)) { robust <- TRUE #flag to later compute a robust variance tempc <- untangle.specials(Terms, 'cluster', 1:10) ord <- attr(Terms, 'order')[tempc$terms] if (any(ord>1)) stop ("Cluster can not be used in an interaction") cluster <- strata(mf[,tempc$vars], shortlabel=TRUE) #allow multiples dropterms <- tempc$terms #we won't want this in the X matrix # Save away xlevels after removing cluster (we don't want to save upteen # levels of that variable, which we will never need). xlevels <- .getXlevels(Terms[-tempc$terms], mf) } else { dropterms <- NULL if (missing(robust)) robust <- FALSE xlevels <- .getXlevels(Terms, mf) } contrast.arg <- NULL #due to shared code with model.matrix.coxph <> <> <> <> <> } @ An increasingly common error is for users to put the time variable on both sides of the formula, in the mistaken idea that this will deal with a failure of proportional hazards. Add a test for such models, and bail out. The \code{variables} attribute of the Terms object is the expression form of a list that contains the response variable followed by the predictors. Subscripting this, element 1 is the call to ``list'' itself so we always retain it. <>= if (length(attr(Terms, 'variables')) > 2) { # a ~1 formula has length 2 ytemp <- terms.inner(attr(Terms, 'variables')[1:2]) xtemp <- terms.inner(attr(Terms, 'variables')[-2]) if (any(!is.na(match(xtemp, ytemp)))) warning("a variable appears on both the left and right sides of the formula") } @ At this point we deal with any time transforms. The model frame is expanded to a ``fake'' data set that has a separate stratum for each unique event-time/strata combination, and any tt() terms in the formula are processed. The first step is to create the index vector [[tindex]] and new strata [[.strata.]]. This last is included in a model.frame call (for others to use), internally the code simply replaces the [[strats]] variable. A (modestly) fast C-routine first counts up and indexes the observations. We start out with error checks; since the computation can be slow we want to complain early. <>= timetrans <- untangle.specials(Terms, 'tt') ntrans <- length(timetrans$terms) if (is.null(tt)) { tt <- function(x, time, riskset, weights){ #default to O'Brien's logit rank obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } } if (is.function(tt)) tt <- list(tt) #single function becomes a list if (is.list(tt)) { if (any(!sapply(tt, is.function))) stop("The tt argument must contain function or list of functions") if (length(tt) != ntrans) { if (length(tt) ==1) { temp <- vector("list", ntrans) for (i in 1:ntrans) temp[[i]] <- tt[[1]] tt <- temp } else stop("Wrong length for tt argument") } } else stop("The tt argument must contain a function or list of functions") if (ncol(Y)==2) { if (length(strats)==0) { sorted <- order(-Y[,1], Y[,2]) newstrat <- rep.int(0L, nrow(Y)) newstrat[1] <- 1L } else { sorted <- order(strats, -Y[,1], Y[,2]) #newstrat marks the first obs of each strata newstrat <- as.integer(c(1, 1*(diff(strats[sorted])!=0))) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount1, Y[sorted,], as.integer(newstrat)) tindex <- sorted[counts$index] } else { if (length(strats)==0) { sort.end <- order(-Y[,2], Y[,3]) sort.start<- order(-Y[,1]) newstrat <- c(1L, rep(0, nrow(Y) -1)) } else { sort.end <- order(strats, -Y[,2], Y[,3]) sort.start<- order(strats, -Y[,1]) newstrat <- c(1L, as.integer(diff(strats[sort.end])!=0)) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount2, Y, as.integer(sort.start -1L), as.integer(sort.end -1L), as.integer(newstrat)) tindex <- counts$index } @ The C routine has returned a list with 4 elements \begin{description} \item[nrisk] a vector containing the number at risk at each event time \item[time] the vector of event times \item[status] a vector of status values \item[index] a vector containing the set of subjects at risk for event time 1, followed by those at risk at event time 2, those at risk at event time 3, etc. \end{description} The new data frame is then a simple creation. <>= mf <- mf[tindex,] Y <- Surv(rep(counts$time, counts$nrisk), counts$status) type <- 'right' # new Y is right censored, even if the old was (start, stop] strats <- rep(1:length(counts$nrisk), counts$nrisk) weights <- model.weights(mf) if (!is.null(weights) && any(!is.finite(weights))) stop("weights must be finite") for (i in 1:ntrans) mf[[timetrans$var[i]]] <- (tt[[i]])(mf[[timetrans$var[i]]], Y[,1], strats, weights) @ This is the C code for time-transformation. For the first case it expects y to contain time and status sorted from longest time to shortest, and strata=1 for the first observation of each strata. <>= #include "survS.h" /* ** Count up risk sets and identify who is in each */ SEXP coxcount1(SEXP y2, SEXP strat2) { int ntime, nrow; int i, j, n; int stratastart=0; /* start row for this strata */ int nrisk=0; /* number at risk (=0 to stop -Wall complaint)*/ double *time, *status; int *strata; double dtime; SEXP rlist, rlistnames, rtime, rn, rindex, rstatus; int *rrindex, *rrstatus; n = nrows(y2); time = REAL(y2); status = time +n; strata = INTEGER(strat2); /* ** First pass: count the total number of death times (risk sets) ** and the total number of rows in the new data set. */ ntime=0; nrow=0; for (i=0; i> /* ** Pass 2, fill them in */ ntime=0; for (i=0; i> } @ The start-stop case is a bit more work. The set of subjects still at risk is an arbitrary set so we have to keep an index vector [[atrisk]]. At each new death time we write out the set of those at risk, with the deaths last. I toyed with the idea of a binary tree then realized it was not useful: at each death we need to list out all the subjects at risk into the index vector which is an $O(n)$ process, tree or not. <>= #include "survS.h" /* count up risk sets and identify who is in each, (start,stop] version */ SEXP coxcount2(SEXP y2, SEXP isort1, SEXP isort2, SEXP strat2) { int ntime, nrow; int i, j, istart, n; int nrisk=0, *atrisk; double *time1, *time2, *status; int *strata; double dtime; int iptr, jptr; SEXP rlist, rlistnames, rtime, rn, rindex, rstatus; int *rrindex, *rrstatus; int *sort1, *sort2; n = nrows(y2); time1 = REAL(y2); time2 = time1+n; status = time2 +n; strata = INTEGER(strat2); sort1 = INTEGER(isort1); sort2 = INTEGER(isort2); /* ** First pass: count the total number of death times (risk sets) ** and the total number of rows in the new data set */ ntime=0; nrow=0; istart =0; /* walks along the sort1 vector (start times) */ for (i=0; i= dtime; istart++) nrisk--; for(j= i+1; j> atrisk = (int *)R_alloc(n, sizeof(int)); /* marks who is at risk */ /* ** Pass 2, fill them in */ ntime=0; nrisk=0; j=0; /* pointer to time1 */; istart=0; for (i=0; i=dtime; istart++) { atrisk[sort1[istart]]=0; nrisk--; } for (j=1; j> } @ <>= /* ** Allocate memory */ PROTECT(rtime = allocVector(REALSXP, ntime)); PROTECT(rn = allocVector(INTSXP, ntime)); PROTECT(rindex=allocVector(INTSXP, nrow)); PROTECT(rstatus=allocVector(INTSXP,nrow)); rrindex = INTEGER(rindex); rrstatus= INTEGER(rstatus); @ <>= /* return the list */ PROTECT(rlist = allocVector(VECSXP, 4)); SET_VECTOR_ELT(rlist, 0, rn); SET_VECTOR_ELT(rlist, 1, rtime); SET_VECTOR_ELT(rlist, 2, rindex); SET_VECTOR_ELT(rlist, 3, rstatus); PROTECT(rlistnames = allocVector(STRSXP, 4)); SET_STRING_ELT(rlistnames, 0, mkChar("nrisk")); SET_STRING_ELT(rlistnames, 1, mkChar("time")); SET_STRING_ELT(rlistnames, 2, mkChar("index")); SET_STRING_ELT(rlistnames, 3, mkChar("status")); setAttrib(rlist, R_NamesSymbol, rlistnames); unprotect(6); return(rlist); @ We now return to the original thread of the program, though perhaps with new data, and build the $X$ matrix. Creation of the $X$ matrix for a Cox model requires just a bit of trickery. The baseline hazard for a Cox model plays the role of an intercept, but does not appear in the $X$ matrix. However, to create the columns of $X$ for factor variables correctly, we need to call the model.matrix routine in such a way that it \emph{thinks} there is an intercept. If there are strata the proper $X$ matrix is constructed as though there were one intercept per strata. One simple way to handle this is to call model.matrix on the original formula and then remove the terms we don't need. However, \begin{enumerate} \item The cluster() term, if any, could lead to thousands of extraneous ``intercept'' columns which are never needed. \item Likewise, nested case-control models can have thousands of strata, again leading many intercepts we never need. \item If there are strata by factor interactions in the model, the dummy intercepts-per-strata columns are necessary information for the model.matrix routine to correctly compute other columns of $X$. \end{enumerate} For reasons 1 and 2 above the usual plan is to remove cluster and strata terms from the ``Terms'' object \emph{before} calling model.matrix, unless there are strata by covariate interactions in which case we remove them after. For the first strategy the \code{assign} attribute of the resulting model matrix then needs to be fixed up, since we want it to index into the original formula. For example imagine the right hand side of \code{age + strata(sex) + trt} where trt is a factor with 3 levels. The assign attribute from the modified formula will be (0,1,2,2) corresponding to the intercept, age, and treatment columns. The final $X$ matrix has no intercept, and a proper assign attribute of (1,3,3) since trt is the third variable in the original formula. The dropterms variable contains terms to drop before creation of the X matrix. It was initialized far above in the code when we dealt with cluster terms. <>= attr(Terms, "intercept") <- 1 adrop <- 0 #levels of "assign" to be dropped; 0= intercept stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars) > 0) { #if there is a strata statement hasinteractions <- FALSE for (i in stemp$vars) { #multiple strata terms are allowed # The factors att has one row for each variable in the frame, one # col for each term in the model. Pick rows for each strata # var, and find if it participates in any interactions. if (any(attr(Terms, 'order')[attr(Terms, "factors")[i,] >0] >1)) hasinteractions <- TRUE } if (!hasinteractions) dropterms <- c(dropterms, stemp$terms) else adrop <- c(0, match(stemp$var, colnames(attr(Terms, 'factors')))) } if (length(dropterms)) { temppred <- attr(terms, "predvars") Terms2 <- Terms[ -dropterms] if (!is.null(temppred)) { # subscripting a Terms object currently drops predvars, in error attr(Terms2, "predvars") <- temppred[-(1+dropterms)] # "Call" object } X <- model.matrix(Terms2, mf, constrasts=contrast.arg) # we want to number the terms wrt the original model matrix # Do not forget the intercept, which will be a zero renumber <- match(colnames(attr(Terms2, "factors")), colnames(attr(Terms, "factors"))) attr(X, "assign") <- c(0, renumber)[1+attr(X, "assign")] } else X <- model.matrix(Terms, mf, contrasts=contrast.arg) # drop the intercept after the fact, and also drop strata if necessary Xatt <- attributes(X) xdrop <- Xatt$assign %in% adrop #columns to drop (always the intercept) X <- X[, !xdrop, drop=FALSE] attr(X, "assign") <- Xatt$assign[!xdrop] #if (any(adrop>0)) attr(X, "contrasts") <- Xatt$contrasts[-adrop] #else attr(X, "contrasts") <- Xatt$contrasts attr(X, "contrasts") <- Xatt$contrasts @ Finish the setup. If someone includes and init statement, make sure that it does not lead to instant code failure due to overflow/underflow. <>= offset <- model.offset(mf) if (is.null(offset) | all(offset==0)) offset <- rep(0., nrow(mf)) else if (any(!is.finite(offset))) stop("offsets must be finite") weights <- model.weights(mf) if (!is.null(weights) && any(!is.finite(weights))) stop("weights must be finite") assign <- attrassign(X, Terms) contr.save <- attr(X, "contrasts") if (missing(init)) init <- NULL else { if (length(init) != ncol(X)) stop("wrong length for init argument") temp <- X %*% init - sum(colMeans(X) * init) if (any(temp < .Machine$double.min.exp | temp > .Machine$double.max.exp)) stop("initial values lead to overflow or underflow of the exp function") } @ Check for penalized terms in the model, and set up infrastructure for the fitting routines to deal with them. <>= pterms <- sapply(mf, inherits, 'coxph.penalty') if (any(pterms)) { pattr <- lapply(mf[pterms], attributes) pname <- names(pterms)[pterms] # # Check the order of any penalty terms ord <- attr(Terms, "order")[match(pname, attr(Terms, 'term.labels'))] if (any(ord>1)) stop ('Penalty terms cannot be in an interaction') pcols <- assign[match(pname, names(assign))] fit <- coxpenal.fit(X, Y, strats, offset, init=init, control, weights=weights, method=method, row.names(mf), pcols, pattr, assign) } @ <>= else { if( method=="breslow" || method =="efron") { if (type== 'right') fitter <- get("coxph.fit") else fitter <- get("agreg.fit") } else if (method=='exact') { if (type== "right") fitter <- get("coxexact.fit") else fitter <- get("agexact.fit") } else stop(paste ("Unknown method", method)) fit <- fitter(X, Y, strats, offset, init, control, weights=weights, method=method, row.names(mf)) } @ <>= if (is.character(fit)) { fit <- list(fail=fit) class(fit) <- 'coxph' } else { if (!is.null(fit$coefficients) && any(is.na(fit$coefficients))) { vars <- (1:length(fit$coefficients))[is.na(fit$coefficients)] msg <-paste("X matrix deemed to be singular; variable", paste(vars, collapse=" ")) if (singular.ok) warning(msg) else stop(msg) } fit$n <- data.n fit$nevent <- sum(Y[,ncol(Y)]) fit$terms <- Terms fit$assign <- assign class(fit) <- fit$method if (robust) { fit$naive.var <- fit$var fit$method <- method # a little sneaky here: by calling resid before adding the # na.action method, I avoid having missings re-inserted # I also make sure that it doesn't have to reconstruct X and Y fit2 <- c(fit, list(x=X, y=Y, weights=weights)) if (length(strats)) fit2$strata <- strats if (length(cluster)) { temp <- residuals.coxph(fit2, type='dfbeta', collapse=cluster, weighted=TRUE) # get score for null model if (is.null(init)) fit2$linear.predictors <- 0*fit$linear.predictors else fit2$linear.predictors <- c(X %*% init) temp0 <- residuals.coxph(fit2, type='score', collapse=cluster, weighted=TRUE) } else { temp <- residuals.coxph(fit2, type='dfbeta', weighted=TRUE) fit2$linear.predictors <- 0*fit$linear.predictors temp0 <- residuals.coxph(fit2, type='score', weighted=TRUE) } fit$var <- t(temp) %*% temp u <- apply(as.matrix(temp0), 2, sum) fit$rscore <- coxph.wtest(t(temp0)%*%temp0, u, control$toler.chol)$test } #Wald test if (length(fit$coefficients) && is.null(fit$wald.test)) { #not for intercept only models, or if test is already done nabeta <- !is.na(fit$coefficients) # The init vector might be longer than the betas, for a sparse term if (is.null(init)) temp <- fit$coefficients[nabeta] else temp <- (fit$coefficients - init[1:length(fit$coefficients)])[nabeta] fit$wald.test <- coxph.wtest(fit$var[nabeta,nabeta], temp, control$toler.chol)$test } na.action <- attr(mf, "na.action") if (length(na.action)) fit$na.action <- na.action if (model) { if (length(timetrans)) { # Fix up the model frame -- still in the thinking stage mf[[".surv."]] <- Y mf[[".strata."]] <- strats stop("Time transform + model frame: code incomplete") } fit$model <- mf } if (x) { fit$x <- X if (length(strats)) { if (length(timetrans)) fit$strata <- strats else fit$strata <- strata.keep } } if (y) fit$y <- Y } @ If any of the weights were not 1, save the results. Add names to the means component, which are occassionally useful to survfit.coxph. Other objects below are used when we need to recreate a model frame. <>= if (!is.null(weights) && any(weights!=1)) fit$weights <- weights names(fit$means) <- names(fit$coefficients) fit$formula <- formula(Terms) if (length(xlevels) >0) fit$xlevels <- xlevels fit$contrasts <- contr.save if (any(offset !=0)) fit$offset <- offset fit$call <- Call fit$method <- method fit @ The model.matrix and model.frame routines are called after a Cox model to reconstruct those portions. Much of their code is shared with the coxph routine. <>= # In internal use "data" will often be an already derived model frame. # We detect this via it having a terms attribute. model.matrix.coxph <- function(object, data=NULL, contrast.arg=object$contrasts, ...) { # # If the object has an "x" component, return it, unless a new # data set is given if (is.null(data) && !is.null(object[['x']])) return(object[['x']]) #don't match "xlevels" Terms <- delete.response(object$terms) if (is.null(data)) mf <- model.frame(object) else { if (is.null(attr(data, "terms"))) mf <- model.frame(Terms, data, xlev=object$xlevels) else mf <- data #assume "data" is already a model frame } cluster <- attr(Terms, "specials")$cluster if (length(cluster)) { temp <- untangle.specials(Terms, "cluster") dropterms <- temp$terms } else dropterms <- NULL <> X } @ In parallel is the model.frame routine, which reconstructs the model frame. This routine currently doesn't do all that we want. To wit, the following code fails: \begin{verbatim} > tfun <- function(formula, ndata) { fit <- coxph(formula, data=ndata) model.frame(fit) } > tfun(Surv(time, status) ~ age, lung) Error: ndata not found \end{verbatim} The genesis of this problem is hard to unearth, but has to do with non standard evaluation rules used by model.frame.default. In essence it pays attention to the environment of the formula, but the enclos argument of eval appears to be ignored. I've not yet found a solution. <>= model.frame.coxph <- function(formula, ...) { dots <- list(...) nargs <- dots[match(c("data", "na.action", "subset", "weights"), names(dots), 0)] # If nothing has changed and the coxph object had a model component, # simply return it. if (length(nargs) ==0 && !is.null(formula$model)) return(formula$model) else { # Rebuild the original call to model.frame Terms <- terms(formula) fcall <- formula$call indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(fcall), nomatch=0) if (indx[1] ==0) stop("The coxph call is missing a formula!") temp <- fcall[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called temp$xlev <- formula$xlevels temp$formula <- Terms #keep the predvars attribute # Now, any arguments that were on this call overtake the ones that # were in the original call. if (length(nargs) >0) temp[names(nargs)] <- nargs # The documentation for model.frame implies that the environment arg # to eval will be ignored, but if we omit it there is a problem. if (is.null(environment(formula$terms))) mf <- eval(temp, parent.frame()) else mf <- eval(temp, environment(formula$terms), parent.frame()) if (!is.null(attr(formula$terms, "dataClasses"))) .checkMFClasses(attr(formula$terms, "dataClasses"), mf) if (!is.null(attr(Terms, "specials")$tt)) { # Do time transform tt <- eval(formula$call$tt) Y <- model.response(mf) strats <- attr(Terms, "specials")$strata if (length(strats)) { stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } <> mf[[".strata."]] <- strats } mf } } @ survival/noweb/tmerge.Rnw0000644000175100001440000004554012512763753015253 0ustar hornikusers\section{tmerge} The tmerge function was designed around a set of specific problems. The idea is to build up a time dependent data set one endpoint at at time. The primary arguments are \begin{itemize} \item data1: the base data set that will be added onto \item data2: the source for new variables, optional \item id: id variable in the new data set \item additional arguments that add variables \item options \end{itemize} The created data set has three new variables (at least), which are \code{id}, \code{tstart} and \code{tstop}. The key part are the \ldots arguments, which are one of four types. Tdc and cumtdc add a time dependent variable, event and cumevent add a new endpoint (event). A typical call would be \begin{verbatim} newdata <- tmerge(newdata, old, id=clinic, diabetes=tdc(diab.time)) \end{verbatim} which would add a new time dependent covariate \code{diabetes} to the data set. <>= tmerge <- function(data1, data2, id, ..., tstart, tstop, options) { Call <- match.call() # The function wants to recognize special keywords in the # arguments, so define a set of functions which will be used to # mark objects new <- new.env(parent=parent.frame()) assign("tdc", function(...) {x <- list(...); class(x) <- "tdc"; x}, envir=new) assign("cumtdc", function(...) {x <- list(...); class(x) <-"cumtdc"; x}, envir=new) assign("event", function(...) {x <- list(...); class(x) <-"event"; x}, envir=new) assign("cumevent", function(...) {x <- list(...); class(x) <-"cumevent"; x}, envir=new) if (missing(data1)) stop("the data1 argument is required") if (missing(id)) stop("the id argument is required") if (!inherits(data1, "data.frame")) stop("data1 must be a data frame") <> <> <> } @ The program can't use formulas because the \ldots arguments need to be named. This results in a bit of evaluation magic to correctly assess arguments. The tname option is used if someone wants to name the three key variables something else. The routine below could have been set out as a separate top-level routine, the argument is where we want to document it: within the tmerge page or not. <>= tmerge.control <- function(id="id", tstart="tstart", tstop="tstop", defer =0) { if (length(defer) !=1 || !is.numeric(defer) || defer <0) stop("defer option must be a non-negative number") if (!is.character(id)) stop("id option must be a character string") if (!is.character(tstart)) stop("tstart option must be a character string") if (!is.character(tstop)) stop("tstop option must be a character string") list(id=id, tstart=tstart, tstop=tstop, defer=defer) } tname <- attr(data1, "tname") if (!is.null(tname) && any(is.null(match(unlist(tname), names(data1))))) stop("data1 does not match its own tname attribute") if (!missing(options)) { if (!is.list(options)) stop("options must be a list") if (!is.null(tname)) { # Changing a name partway through a set of calls? if (any(!is.na(match(names(options), names(tname))))) stop("cannot change names in mid-stream") topt <- do.call(tmerge.control, c(tname, options)) } else topt <- do.call(tmerge.control, options) } else if (length(tname)) topt <- do.call(tmerge.control, tname) else topt <- tmerge.control() # id, tstart, tstop are found in data2, if it is present if (!missing(data2)) { id <- eval(Call[["id"]], data2) if (!missing(tstart)) tstart <- eval(Call[["tstart"]], data2) if (!missing(tstop)) tstop <- eval(Call[["tstop"]], data2) } if (!missing(tstart) && length(tstart) != length(id)) stop("tstart and id must be the same length") if (!missing(tstop) && length(tstop) != length(id)) stop("tstop and id must be the same length") @ Get the \ldots arguments. They are evaluated in a special frame, set up earlier, so that the definitions of the functions tdc, cumtdc, event, and cumevent are local to tmerge. Check that they are all legal: each argument is named, and is one of the four allowed types. <>= # grab the... arguments notdot <- c("data1", "data2", "id", "tstart", "tstop", "options") dotarg <- Call[is.na(match(names(Call), notdot))] dotarg[[1]] <- as.name("list") # The as-yet dotarg arguments if (missing(data2)) args <- eval(dotarg, envir=new) else args <- eval(dotarg, data2, enclos=new) argclass <- sapply(args, function(x) (class(x))[1]) argname <- names(args) if (any(argname== "")) stop("all additional argments must have a name") check <- match(argclass, c("tdc", "cumtdc", "event", "cumevent")) if (any(is.na(check))) stop(paste("argument(s)", argname[is.na(check)], "not a recognized type")) @ The tcount matrix keeps track of what we have done, and is added to the final object at the end. This is useful to the user for debugging what may have gone right or wrong in their usage. <>= # The tcount matrix is useful for debugging tcount <- matrix(0L, length(argname), 8) dimnames(tcount) <- list(argname, c("early","late", "gap", "within", "boundary", "leading", "trailing", "tied")) tevent <- attr(data1, "tevent") # event type variables @ The very first call to the routine is special, since this is when the range of legal times is set. There are 3 cases: \begin{enumerate} \item The id and tstop variables are both found in data1, and there is no tstop argument. This means that the user has already taken care of the work. We just need to note the variable names in topt list, check that the data looks good to us, and go on. \item Adding a range: tstop comes from data2, optional tstart, and the id can be simply matched, by which we mean no duplicates in data1. \item The most common case. The row counts of data1 and data2 exactly match, there is no tstop, and the first optional argument is an event or cumevent. We then use its time as the range. \end{enumerate} <>= newdata <- data1 # make a copy if (!missing(tstop) || length(tname)==0) { # This is a first call indx <- match(c(topt$id, topt$tstart, topt$tstop), names(data1), nomatch=0) if (all(indx[1:2] >0) && missing(tstop)) { # case 1 above, just some data checks if (indx[3]==0) newdata[[topt$tstart]] <- 0 if (!is.numeric(newdata[[topt$tstart]]) || !is.numeric(newdata[[topt$tstop]])) stop("start and end variables must be numeric") if (any(newdata[[topt$tstart]] >= newdata[[topt$tstop]])) stop("stop time must be > start time for all observations") # If there are duplicated ids, then we need to ensure that each subject # is a sequential set of observations, sorted by time (just sort it) # If tstart was not supplied, we need to correct our "0" created above baseid <- data1[[topt$id]] if (any(duplicated(baseid))) { indx <- order(newdata$id, newdata$tstop) if (any(indx != seq(along.with=indx))) { # sort the data newdata <- newdata[indx,] baseid <- baseid[indx] } newid <- !duplicated(baseid) #is this row a new id value? n <- nrow(newdata) if (length(tstart) ==0) newdata$tstart <- ifelse(newid, 0, c(0, newdata$tstop[-n])) else { ok <- (newid[-1] | newdata$tstart[-1] >= newdata$tstop[-n]) if (any(!ok)) stop("overlapping time intervals for a subject") } } } else { if (missing(tstop)) { # case 3 above if (length(argclass)==0 || argclass[1] != "event") stop("neither a tstop argument nor an initial event argument was found") tstop <- args[[1]][[1]] } # case 2 and case 3 if (any(is.na(tstop))) stop("missing time value, when that variable defines the span") if (missing(tstart)) tstart <- rep(0, length(id)) if (any(tstart >= tstop)) stop("stop time must be > start time for all observations") if (indx[1] >0) { # the id variable is in data1 baseid <- data1[[topt$id]] if (any(duplicated(baseid))) stop("duplicate identifiers in data1") indx2 <- match(id, baseid) if (any(is.na(indx2))) stop("'id' has values not in data1") } else { if (nrow(data1) != nrow(data2)) stop("nrow(data1) != nrow(data2) and data1 is missing the id") indx2 <- seq.int(along.with=id) newdata[topt$id] <- id } newdata[indx2, topt$tstart] <- tstart newdata[indx2, topt$tstop] <- tstop } } else { #not a first call if (any(is.na(match(id, data1[[topt$id]])))) stop("id values found in data2 which are not in data1") } @ Now for the real work. For each additional argument we first match the id/time pairs of the new data to the current data set, and categorize each into a type. If the time value in data2 is NA, then that addition is skipped. This is a convenience for the user, who will often be merging in a variable like ``day of first diabetes diagnosis'' which is missing for those who never had that outcome occur. <>= saveid <- id for (ii in seq(along.with=args)) { argi <- args[[ii]] baseid <- newdata[[topt$id]] dstart <- newdata[[topt$tstart]] dstop <- newdata[[topt$tstop]] # if an event time is missing then skip that obs etime <- argi[[1]] keep <- !is.na(etime) etime <- etime[keep] id <- saveid[keep] if (length(etime) != length(id)) stop("argument", argname[ii], "is not the same length as id") # For an event or cumevent, one of the later steps becomes much # easier if we sort the new data by id and time indx <- order(id, etime) id <- id[indx] etime <- etime[indx] if (length(argi) > 1) yinc <- (argi[[2]])[indx] # indx1 points to the closest start time in the baseline data (data1) # that is <= etime. indx2 to the closest end time that is >=etime. # If etime falls into a (tstart, tstop) interval, indx1 and indx2 # will match # If the "defer" argument is set and this event is of type tdc, then # any event times are artificially moved left by "defer" amount wrt # doing the indx2 match. This will cause an insertion that is too close # to an event to be labeled as itype=3 (or itype=2 if this was the last # interval for the subject) and so map later in time. defer <- rep(0., nrow(newdata)) if (topt$defer >0 && length(tevent) && argclass[ii] %in% c("tdc", "cumtdc")) { for (ename in tevent) { temp <- newdata[[ename]] if (is.logical(temp)) defer[temp] <- topt$defer else defer[temp!=0] <- topt$defer } } indx1 <- neardate(id, baseid, etime, dstart, best="prior") indx2 <- neardate(id, baseid, etime, dstop+defer, best="after") # The event times fall into one of 5 categories # 1. Before the first interval # 2. After the last interval # 3. Outside any interval but with time span, i.e, it falls into # a gap in follow-up # 4. Strictly inside an interval (does't touch either end) # 5. Inside an interval, but touching. itype <- ifelse(is.na(indx1), 1, ifelse(is.na(indx2), 2, ifelse(indx2 > indx1, 3, ifelse(etime== dstart[indx1] | etime== dstop[indx2], 5, 4)))) # Subdivide the events that touch on a boundary # 1: intervals of (a,b] (b,d], new count at b "tied edge" # 2: intervals of (a,b] (c,d] with c>b, new count at c, "front edge" # 3: intervals of (a,b] (c,d] with c>b, new count at b, "back edge" # subtype <- ifelse(itype!=5, 0, ifelse(indx1 == indx2+1, 1, ifelse(etime==dstart[indx1], 2, 3))) tcount[ii,1:7] <- table(factor(itype+subtype, levels=c(1:4, 6:8))) # count ties. id and etime are not necessarily sorted tcount[ii,8] <- sum(tapply(etime, id, function(x) sum(duplicated(x)))) <> } @ An argument of \code{tdc(etime)} causes a time-dependent covariate value of 1, one of \code{tdc(etime, x)} causes the created time dependent variable to have a value of x. <>= # Look to see if this term has one or two arguments. If one arg # then the increment is 1, else it is the second arg. The myfun() # function will later compute totals by unique subject/time pair # if (length(argi) >1) { if (length(argi) > 2) stop("too many variables in an", argclass[ii], "call") if (diff(sapply(argi, length)) !=0) stop("different lengths in an", argclass[ii], "call") if (!is.numeric(yinc) && argclass[ii] != "event") stop("non numeric increment in an", argclass[ii], "call") myfun <- function(x, grp) { temp <- tapply(yinc[grp], x[grp], sum) ifelse(is.na(temp), 0, temp) } } else { myfun <- function(x, grp) table(x[grp]) yinc <- rep(1.0, length(etime)) # each counts as 1 } @ A \code{tdc} or \code{cumtdc} operator defines a new time-dependent variable which applies to all future times. Say that we had the following scenario for one subject \begin{center} \begin{tabular}{rr|rr} \multicolumn{2}{c}{current} & \multicolumn{2}{c}{addition} \\ tstart & tstop & time & x \\ 2 & 5 & 1 & 20.2 \\ 6 & 7 & 7 & 11 \\ 7 & 15 & 8 & 17.3 \\ 15 & 30 \\ \end{tabular} \end{center} The resulting data set will have intervals of (2,5), (6,7), (7,8) and (8,15) with covariate values of 20.2, 20.2, 11, and 17.3. Only a covariate change that occurs within an interval causes a new data row. Covariate changes that happen after the last interval are ignored, i.e. at change at time $\ge 30$ in the above example. If instead this had been events at times 1, 7, and 8, the first event would be ignored since it happens outside of any interval, so would an event at exactly time 2. The event at time 7 would be recorded in the (6,7) interval and the one at time 8 in the (7,8) interval: events happen at the ends of intervals. In both cases new rows are only generated for new time values that fall strictly within one of the old intervals. When a subject has two increments on the same day they get summed. This is odd but possible for events, likely an error for time-dependent covariates. We report back the number of ties so that the user can deal with it. Where are we now with the variables? \begin{center} \begin{tabular}{cccc} itype& class & indx1 & indx2 \\ \hline 1 & before & NA & next interval \\ 2 & after & prior interval & NA \\ 3 & in a gap & prior interval & next interval \\ 4 & within interval & containing interval & containing interval \\ 5-1 & on a join & next interval & prior interval \\ 5-2 & front edge & containing & containing \\ 5-3 & back edge & containing & containing \\ \end{tabular} \end{center} If there are any itype 4, start by expanding the data set to add new cut points, which will turn all the 4's into 5-1 types. When expanding, all the event type variables turn into zero at the newly added times and other variables stay the same. A subject could have more than one new cutpoint added within an interval so we have to count each. In newdata all the rows for a given subject are contiguous and in time order, though the data set may not be in subject order. <>= indx4 <- which(itype==4) n4 <- length(indx4) if (n4 > 0) { icount <- tapply(etime[indx4], indx1[indx4], function(x) sort(unique(x))) n.add <- sapply(icount, length) #number of rows to add # expand the data irep <- rep.int(1L, nrow(newdata)) erow <- unique(indx1[indx4]) # which rows in newdata to be expanded irep[erow] <- 1+ n.add # number of rows in new data jrep <- rep(1:nrow(newdata), irep) #stutter the duplicated rows newdata <- newdata[jrep,] #expand it out dstart <- dstart[jrep] dstop <- dstop[jrep] #fix up times nfix <- length(erow) temp <- vector("list", nfix) iend <- (cumsum(irep))[irep >1] #end row of each duplication set for (j in 1:nfix) temp[[j]] <- -(seq(n.add[j] -1, 0)) + iend[j] newrows <- unlist(temp) dstart[newrows] <- dstop[newrows-1] <- unlist(icount) newdata[[topt$tstart]] <- dstart newdata[[topt$tstop]] <- dstop for (ename in tevent) newdata[newrows-1, ename] <- 0 if (topt$defer > 0) { defer <- defer[jrep] defer[newrows] <- 0 } else defer <- rep(0, nrow(newdata)) # refresh indices baseid <- newdata[[topt$id]] indx1 <- neardate(id, baseid, etime, dstart, best="prior") indx2 <- neardate(id, baseid, etime, dstop+ defer , best="after") subtype[itype==4] <- 1 #all the "insides" are now on a tied edge itype[itype==4] <- 5 } @ Now we can add the new variable. Events and cumevents are easy because each affects only one interval. Counts are more work and for this we use a C routine. <>= # add it in if (argclass[ii] %in% c("cumtdc", "cumevent")) yinc <- unlist(tapply(yinc, id, cumsum)) newvar <- newdata[[argname[ii]]] #does the variable exist? if (argclass[ii] %in% c("event", "cumevent")) { if (is.null(newvar)) newvar <- rep(0, nrow(newdata)) keep <- (subtype==1 | subtype==3) # all other events are thrown away newvar[indx2[keep]] <- yinc[keep] tevent <- unique(c(tevent, argname[ii])) } else { keep <- itype != 2 # changes after the last interval are ignored indx <- ifelse(subtype==1, indx1, ifelse(subtype==3, indx2+1L, indx2)) if (is.null(newvar)) { if (length(argi)==1) newvar <- rep(0.0, nrow(newdata)) else newvar <- rep(NA_real_, nrow(newdata)) } # id can be any data type; feed integers to the C routine storage.mode(yinc) <- storage.mode(dstop) <- "double" storage.mode(newvar) <- storage.mode(etime) <- "double" newvar <- .Call("tmerge", match(baseid, baseid), dstop, newvar, match(id, baseid)[keep], etime[keep], yinc[keep], indx[keep]) } newdata[[argname[ii]]] <- newvar @ Finish up by adding the attributes and the class <>= attr(newdata, "tname") <- topt[c("id", "tstart", "tstop")] attr(newdata, "tcount") <- tcount if (length(tevent)) attr(newdata, "tevent") <- tevent row.names(newdata) <- NULL class(newdata) <- c("data.frame") newdata @ The print routine is for checking: it simply prints out the attributes. <>= print.tmerge <- function(x, ...) { print(attr(x, "tcount")) } "[.tmerge" <- function(x, ..., drop=TRUE){ class(x) <- "data.frame" NextMethod(,x) } @ survival/noweb/survexp.Rnw0000644000175100001440000004077612514444605015504 0ustar hornikusers\section{Expected Survival} The expected survival routine creates the overall survival curve for a \emph{group} of people. It is possible to take the set of expected survival curves for each individual and average them, which is the \texttt{Ederer} method below, but this is not always the wisest choice: the Hakulinen and conditional methods average in anothers ways, both of which are more sophisticated ways to deal with censoring. The individual curves are dervived either from population rate tables such as the US annual life tables from the National Center for Health Statistics or the larger multi-national collection at mortality.org, or by using a previously fitted Cox model as the table. The arguments for [[survexp]] are \begin{description} \item[formula] The model formula. The right hand side consists of grouping variables, identically to [[survfit]] and an optional [[ratetable]] directive. The ``response'' varies by method: \begin{itemize} \item for the Hakulinen method it is a vector of censoring times. This is the actual censoring time for censored subjecs, and is what the censoring time `would have been' for each subject who died. %'` \item for the conditional method it is the usual Surv(time, status) code \item for the Ederer method no response is needed \end{itemize} \item[data, weights, subset, na.action] as usual \item[rmap] an optional mapping for rate table variables, see more below. \item[times] An optional vector of time points at which to compute the response. For the Hakulinen and conditional methods the program uses the vector of unique y values if this is missing. For the Ederer the component is not optional. \item[method] The method used for the calculation. Choices are individual survival, or the Ederer, Hakulinen, or conditional methods for cohort survival. \item[cohort, conditional] Older arguments that were used to select the method. \item[ratetable] the population rate table to use as a reference. This can either be a ratetable object or a previously fitted Cox model \item[scale] Scale the resulting output times, e.g., 365.25 to turn days into years. \item[se.fit] This has been deprecated. \item[model, x, y] usual \end{description} The output of survexp contains a subset of the elements in a [[survfit]] object, so many of the survfit methods can be applied. The result has a class of [[c('survexp', 'survfit')]]. <>= survexp <- function(formula, data, weights, subset, na.action, rmap, times, method=c("ederer", "hakulinen", "conditional", "individual.h", "individual.s"), cohort=TRUE, conditional=FALSE, ratetable=survival::survexp.us, scale=1, se.fit, model=FALSE, x=FALSE, y=FALSE) { <> <> <> <> } @ The first few lines are standard. Keep a copy of the call, then manufacture a call to [[model.frame]] that contains only the arguments relevant to that function. <>= call <- match.call() m <- match.call(expand.dots=FALSE) # keep the first element (the call), and the following selected arguments m <- m[c(1, match(c('formula', 'data', 'weights', 'subset', 'na.action'), names(m), nomatch=0))] m[[1]] <- as.name("model.frame") Terms <- if(missing(data)) terms(formula, 'ratetable') else terms(formula, 'ratetable',data=data) @ The function works with two data sets, the user's data on an actual set of %' subjects and the reference ratetable. This leads to a particular nuisance, that the variable names in the data set may not match those in the ratetable. For instance the United States overall death rate table [[survexp.us]] expects 3 variables, as shown by [[summary(survexp.us)]] \begin{itemize} \item age = age in days for each subject at the start of follow-up \item sex = sex of the subject, ``male'' or ``female'' (the routine accepts any unique abbreviation and is case insensitive) \item year = date of the start of follow-up \end{itemize} Up until the most recent revision, the formula contained any necessary mapping between the variables in the data set and the ratetable. For instance \begin{verbatim} survexp( ~ sex + ratetable(age=age*365.25, sex=sex, year=entry.dt), data=mydata, ratetable=survexp.us) \end{verbatim} In this case the user's data set has a variable `age' containing age in years, along with sex and an entry date. This had to be changed for two reasons. The primary one is that the data in a [[ratetable]] call had to be converted into a matrix in order to ``pass through'' the model.frame logic. With the recent updates to coxph so that it remembers factor codings correctly in new data sets, it is advantageous to keep factors as factors. The second is that a coxph model with a large number of covariates induces a very long ratetable clause; at about 40 variable it caused one of the R internal routines to fail due to a long expression. A third reason, perhaps the most pressing in reality, is that I've always %' felt that the prior code was confusing since it used the same term 'ratetable' for two different tasks. The new process adds the [[rmap]] argument, an example would be [[rmap=list(age =age*365.25, year=entry.dt)]]. Any variables in the ratetable that are not found in [[rmap]] are assumed to not need a mapping, this would be [[sex]] in the above example. For backwards compatability we allow the old style argument, converting it into the new style. The [[rmap]] argument needs to be examined without evaluating it; we then add the appropriate extra variables into a temporary formula so that the model frame has all that is required. The ratetable variables then can be retrieved from the model frame. The [[pyears]] routine uses the same rmap argument; this segment of the code is given its own name so that it can be included there as well. <>= rate <- attr(Terms, "specials")$ratetable if(length(rate) > 1) stop("Can have only 1 ratetable() call in a formula") <> m <- eval(m, parent.frame()) @ <>= if(length(rate) == 1) { if (!missing(rmap)) stop("The ratetable() call in a formula is depreciated") stemp <- untangle.specials(Terms, 'ratetable') rcall <- as.call(parse(text=stemp$var)[[1]]) # as a call object rcall[[1]] <- as.name('list') # make it a call to list(.. Terms <- Terms[-stemp$terms] # remove from the formula } else if (!missing(rmap)) { rcall <- substitute(rmap) if (!is.call(rcall) || rcall[[1]] != as.name('list')) stop ("Invalid rcall argument") } else rcall <- NULL # A ratetable, but not rcall argument # Check that there are no illegal names in rcall, then expand it # to include all the names in the ratetable if(is.ratetable(ratetable)) varlist <- attr(ratetable, "dimid") else if(inherits(ratetable, "coxph")) { ## Remove "log" and such things, to get just the list of # variable names varlist <- all.vars(delete.response(ratetable$terms)) } else stop("Invalid rate table") temp <- match(names(rcall)[-1], varlist) # 2,3,... are the argument names if (any(is.na(temp))) stop("Variable not found in the ratetable:", (names(rcall))[is.na(temp)]) if (any(!(varlist %in% names(rcall)))) { to.add <- varlist[!(varlist %in% names(rcall))] temp1 <- paste(text=paste(to.add, to.add, sep='='), collapse=',') if (is.null(rcall)) rcall <- parse(text=paste("list(", temp1, ")"))[[1]] else { temp2 <- deparse(rcall) rcall <- parse(text=paste("c(", temp2, ",list(", temp1, "))"))[[1]] } } @ The formula below is used only in the call to [[model.frame]] to ensure that the frame has both the formula and the ratetable variables. We don't want to modify the original formula, since we use it to create the $X$ matrix and the response variable. The non-obvious bit of code is the addition of an environment to the formula. The [[model.matrix]] routine has a non-standard evaluation - it uses the frame of the formula, rather than the parent.frame() argument below, along with the [[data]] to look up variables. If a formula is long enough deparse() will give two lines, hence the extra paste call to re-collapse it into one. <>= # Create a temporary formula, used only in the call to model.frame newvar <- all.vars(rcall) if (length(newvar) > 0) { tform <- paste(paste(deparse(Terms), collapse=""), paste(newvar, collapse='+'), sep='+') m$formula <- as.formula(tform, environment(Terms)) } @ If the user data has 0 rows, e.g. from a mistaken [[subset]] statement that eliminated all subjects, we need to stop early. Otherwise the .C code fails in a nasty way. <>= n <- nrow(m) if (n==0) stop("Data set has 0 rows") if (!missing(se.fit) && se.fit) warning("se.fit value ignored") weights <- model.extract(m, 'weights') if (length(weights) ==0) weights <- rep(1.0, n) if (class(ratetable)=='ratetable' && any(weights !=1)) warning("weights ignored") if (any(attr(Terms, 'order') >1)) stop("Survexp cannot have interaction terms") if (!missing(times)) { if (any(times<0)) stop("Invalid time point requested") if (length(times) >1 ) if (any(diff(times)<0)) stop("Times must be in increasing order") } @ If a response variable was given, we only need the times and not the status. To be correct, computations need to be done for each of the times given in the [[times]] argument as well as for each of the unique y values. This ends up as the vector [[newtime]]. If a [[times]] argument was given we will subset down to only those values at the end. For a population rate table and the Ederer method the times argument is required. <>= Y <- model.extract(m, 'response') no.Y <- is.null(Y) if (no.Y) { if (missing(times)) { if (is.ratetable(ratetable)) stop("either a times argument or a response is needed") } else newtime <- times } else { if (is.matrix(Y)) { if (is.Surv(Y) && attr(Y, 'type')=='right') Y <- Y[,1] else stop("Illegal response value") } if (any(Y<0)) stop ("Negative follow up time") # if (missing(npoints)) temp <- unique(Y) # else temp <- seq(min(Y), max(Y), length=npoints) temp <- unique(Y) if (missing(times)) newtime <- sort(temp) else newtime <- sort(unique(c(times, temp[temp>= ovars <- attr(Terms, 'term.labels') # rdata contains the variables matching the ratetable rdata <- data.frame(eval(rcall, m), stringsAsFactors=TRUE) if (is.ratetable(ratetable)) { israte <- TRUE if (no.Y) { Y <- rep(max(times), n) } rtemp <- match.ratetable(rdata, ratetable) R <- rtemp$R } else if (inherits(ratetable, 'coxph')) { israte <- FALSE Terms <- ratetable$terms # if (!is.null(attr(Terms, 'offset'))) # stop("Cannot deal with models that contain an offset") # strats <- attr(Terms, "specials")$strata # if (length(strats)) # stop("survexp cannot handle stratified Cox models") # if (any(names(m[,rate]) != attr(ratetable$terms, 'term.labels'))) stop("Unable to match new data to old formula") } else stop("Invalid ratetable") @ Now for some calculation. If cohort is false, then any covariates on the right hand side (other than the rate table) are irrelevant, the function returns a vector of expected values rather than survival curves. <>= if (substring(method, 1, 10) == "individual") { #individual survival if (no.Y) stop("for individual survival an observation time must be given") if (israte) temp <- survexp.fit (1:n, R, Y, max(Y), TRUE, ratetable) else { rmatch <- match(names(data), names(rdata)) if (any(is.na(rmatch))) rdata <- cbind(rdata, data[,is.na(rmatch)]) temp <- survexp.cfit(1:n, rdata, Y, 'individual', ratetable) } if (method == "individual.s") xx <- temp$surv else xx <- -log(temp$surv) names(xx) <- row.names(m) na.action <- attr(m, "na.action") if (length(na.action)) return(naresid(na.action, xx)) else return(xx) } @ Now for the more commonly used case: returning a survival curve. First see if there are any grouping variables. The results of the [[tcut]] function are often used in person-years analysis, which is somewhat related to expected survival. However tcut results aren't relevant here and we put in a check for the %' confused user. The strata command creates a single factor incorporating all the variables. <>= if (length(ovars)==0) X <- rep(1,n) #no categories else { odim <- length(ovars) for (i in 1:odim) { temp <- m[[ovars[i]]] ctemp <- class(temp) if (!is.null(ctemp) && ctemp=='tcut') stop("Can't use tcut variables in expected survival") } X <- strata(m[ovars]) } #do the work if (israte) temp <- survexp.fit(as.numeric(X), R, Y, newtime, method=="conditional", ratetable) else { temp <- survexp.cfit(as.numeric(X), rdata, Y, method, ratetable, weights) newtime <- temp$time } @ Now we need to package up the curves properly All the results can be returned as a single matrix of survivals with a common vector of times. If there was a times argument we need to subset to selected rows of the computation. <>= if (missing(times)) { n.risk <- temp$n surv <- temp$surv } else { if (israte) keep <- match(times, newtime) else { # The result is from a Cox model, and it's list of # times won't match the list requested in the user's call # Interpolate the step function, giving survival of 1 # for requested points that precede the Cox fit's # first downward step. The code is like summary.survfit. n <- length(temp$time) keep <- approx(temp$time, 1:n, xout=times, yleft=0, method='constant', f=0, rule=2)$y } if (is.matrix(temp$surv)) { surv <- (rbind(1,temp$surv))[keep+1,,drop=FALSE] n.risk <- temp$n[pmax(1,keep),,drop=FALSE] } else { surv <- (c(1,temp$surv))[keep+1] n.risk <- temp$n[pmax(1,keep)] } newtime <- times } newtime <- newtime/scale if (is.matrix(surv)) { dimnames(surv) <- list(NULL, levels(X)) out <- list(call=call, surv= drop(surv), n.risk=drop(n.risk), time=newtime) } else { out <- list(call=call, surv=c(surv), n.risk=c(n.risk), time=newtime) } @ Last do the standard things: add the model, x, or y components to the output if the user asked for them. (For this particular routine I can't think of %' a reason they every would.) Copy across summary information from the rate table computation if present, and add the method and class to the output. <>= if (model) out$model <- m else { if (x) out$x <- X if (y) out$y <- Y } if (israte && !is.null(rtemp$summ)) out$summ <- rtemp$summ if (no.Y) out$method <- 'Ederer' else if (conditional) out$method <- 'conditional' else out$method <- 'cohort' class(out) <- c('survexp', 'survfit') out @ survival/noweb/predict.coxph.Rnw0000644000175100001440000005407012350313207016521 0ustar hornikusers\subsection{The predict method} The [[predict.coxph]] function produces various types of predicted values from a Cox model. The arguments are \begin{description} \item [obejct] The result of a call to [[coxph]]. \item [newdata] Optionally, a new data set for which prediction is desired. If this is absent predictions are for the observations used fit the model. \item[type] The type of prediction \begin{itemize} \item lp = the linear predictor for each observation \item risk = the risk score $exp(lp)$ for each observation \item expected = the expected number of events \item terms = a matrix with one row per subject and one column for each term in the model. \end{itemize} \item[se.fit] Whether or not to return standard errors of the predictions. \item[na.action] What to do with missing values \emph{if} there is new data. \item[terms] The terms that are desired. This option is almost never used, so rarely in fact that it's hard to justify keeping it. \item[collapse] An optional vector of subject identifiers, over which to sum or `collapse' the results \item[reference] the reference context for centering the results \item[\ldots] All predict methods need to have a \ldots argument; we make no use of it however. \end{description} %\subsection{Setup} The first task of the routine is to reconsruct necessary data elements that were not saved as a part of the [[coxph]] fit. We will need the following components: \begin{itemize} \item for type=`expected' residuals we need the orignal survival y. This %'` is saved in coxph objects by default so will only need to be fetched in the highly unusual case that a user specfied [[y=FALSE]] in the orignal call. \item for any call with either newdata, standard errors, or type='terms' the original $X$ matrix, weights, strata, and offset. When checking for the existence of a saved $X$ matrix we can't %' use [[object$x]] since that will also match the [[xlevels]] component. \item the new data matrix, if any \end{itemize} <>= predict.coxph <- function(object, newdata, type=c("lp", "risk", "expected", "terms"), se.fit=FALSE, na.action=na.pass, terms=names(object$assign), collapse, reference=c("strata", "sample"), ...) { <> <> if (type=="expected") { <> } else { <> <> } <> } @ We start of course with basic argument checking. Then retrieve the model parameters: does it have a strata statement, offset, etc. The [[Terms2]] object is a model statement without the strata or cluster terms, appropriate for recreating the matrix of covariates $X$. For type=expected the response variable needs to be kept, if not we remove it as well since the user's newdata might not contain one. %' <>= if (!inherits(object, 'coxph')) stop("Primary argument much be a coxph object") Call <- match.call() type <-match.arg(type) n <- object$n Terms <- object$terms if (!missing(terms)) { if (is.numeric(terms)) { if (any(terms != floor(terms) | terms > length(object$assign) | terms <1)) stop("Invalid terms argument") } else if (any(is.na(match(terms, names(object$assign))))) stop("a name given in the terms argument not found in the model") } # I will never need the cluster argument, if present delete it. # Terms2 are terms I need for the newdata (if present), y is only # needed there if type == 'expected' if (length(attr(Terms, 'specials')$cluster)) { temp <- untangle.specials(Terms, 'cluster', 1) Terms <- object$terms[-temp$terms] } else Terms <- object$terms if (type != 'expected') Terms2 <- delete.response(Terms) else Terms2 <- Terms has.strata <- !is.null(attr(Terms, 'specials')$strata) has.offset <- !is.null(attr(Terms, 'offset')) has.weights <- any(names(object$call) == 'weights') na.action.used <- object$na.action n <- length(object$residuals) if (missing(reference) && type=="terms") reference <- "sample" else reference <- match.arg(reference) @ The next task of the routine is to reconsruct necessary data elements that were not saved as a part of the [[coxph]] fit. We will need the following components: \begin{itemize} \item for type=`expected' residuals we need the orignal survival y. This %'` is saved in coxph objects by default so will only need to be fetched in the highly unusual case that a user specfied [[y=FALSE]] in the orignal call. We also need the strata in this case. Grabbing it is the same amount of work as grabbing X, so gets lumped with that case in the code. \item for any call with either standard errors, reference strata, or type=`terms' the original $X$ matrix, weights, strata, and offset. When checking for the existence of a saved $X$ matrix we can't %' use [[object$x]] since that will also match the [[xlevels]] component. \item the new data matrix, if present, along with offset and strata. \end{itemize} For the case that none of the above are needed, we can use the [[linear.predictors]] component of the fit. The variable [[use.x]] signals this case, which takes up almost none of the code but is common in usage. The check below that nrow(mf)==n is to avoid data sets that change under our feet. A fit was based on data set ``x'', and when we reconstruct the data frame it is a different size! This means someone changed the data between the model fit and the extraction of residuals. One other non-obvious case is that coxph treats the model [[age:strata(grp)]] as though it were [[age:strata(grp) + strata(grp)]]. The untangle.specials function will return [[vars= strata(grp), terms=integer(0)]]; the first shows a strata to extract and the second that there is nothing to remove from the terms structure. <>= have.mf <- FALSE if (type == 'expected') { y <- object[['y']] if (is.null(y)) { # very rare case mf <- model.frame(object) y <- model.extract(mf, 'response') have.mf <- TRUE #for the logic a few lines below, avoid double work } } if (se.fit || type=='terms' || (!missing(newdata) && type=="expected") || (has.strata && (reference=="strata") || type=="expected")) { use.x <- TRUE if (is.null(object[['x']]) || has.weights || has.offset || (has.strata && is.null(object$strata))) { # I need the original model frame if (!have.mf) mf <- model.frame(object) if (nrow(mf) != n) stop("Data is not the same size as it was in the original fit") x <- model.matrix(object, data=mf) if (has.strata) { if (!is.null(object$strata)) oldstrat <- object$strata else { stemp <- untangle.specials(Terms, 'strata') if (length(stemp$vars)==1) oldstrat <- mf[[stemp$vars]] else oldstrat <- strata(mf[,stemp$vars], shortlabel=TRUE) } } else oldstrat <- rep(0L, n) weights <- model.weights(mf) if (is.null(weights)) weights <- rep(1.0, n) offset <- model.offset(mf) if (is.null(offset)) offset <- 0 } else { x <- object[['x']] if (has.strata) oldstrat <- object$strata else oldstrat <- rep(0L, n) weights <- rep(1.,n) offset <- 0 } } else { # I won't need strata in this case either if (has.strata) { stemp <- untangle.specials(Terms, 'strata', 1) Terms2 <- Terms2[-stemp$terms] has.strata <- FALSE #remaining routine never needs to look } oldstrat <- rep(0L, n) offset <- 0 use.x <- FALSE } @ Now grab data from the new data set. We want to use model.frame processing, in order to correctly expand factors and such. We don't need weights, however, and don't want to make the user include them in their new dataset. Thus we build the call up the way it is done in coxph itself, but only keeping the newdata argument. Note that terms2 may have fewer variables than the original model: no cluster and if type!= expected no response. If the original model had a strata, but newdata does not, we need to remove the strata from xlev to stop a spurious warning message. <>= if (!missing(newdata)) { use.x <- TRUE #we do use an X matrix later tcall <- Call[c(1, match(c("newdata", "collapse"), names(Call), nomatch=0))] names(tcall)[2] <- 'data' #rename newdata to data tcall$formula <- Terms2 #version with no response tcall$na.action <- na.action #always present, since there is a default tcall[[1]] <- as.name('model.frame') # change the function called if (!is.null(attr(Terms, "specials")$strata) && !has.strata) { temp.lev <- object$xlevels temp.lev[[stemp$vars]] <- NULL tcall$xlev <- temp.lev } else tcall$xlev <- object$xlevels mf2 <- eval(tcall, parent.frame()) collapse <- model.extract(mf2, "collapse") n2 <- nrow(mf2) if (has.strata) { if (length(stemp$vars)==1) newstrat <- mf2[[stemp$vars]] else newstrat <- strata(mf2[,stemp$vars], shortlabel=TRUE) if (any(is.na(match(newstrat, oldstrat)))) stop("New data has a strata not found in the original model") else newstrat <- factor(newstrat, levels=levels(oldstrat)) #give it all if (length(stemp$terms)) newx <- model.matrix(Terms2[-stemp$terms], mf2, contr=object$contrasts)[,-1,drop=FALSE] else newx <- model.matrix(Terms2, mf2, contr=object$contrasts)[,-1,drop=FALSE] } else { newx <- model.matrix(Terms2, mf2, contr=object$contrasts)[,-1,drop=FALSE] newstrat <- rep(0L, nrow(mf2)) } newoffset <- model.offset(mf2) if (is.null(newoffset)) newoffset <- 0 if (type== 'expected') { newy <- model.response(mf2) if (attr(newy, 'type') != attr(y, 'type')) stop("New data has a different survival type than the model") } na.action.used <- attr(mf2, 'na.action') } else n2 <- n @ %\subsection{Expected hazard} When we do not need standard errors the computation of expected hazard is very simple since the martingale residual is defined as status - expected. The 0/1 status is saved as the last column of $y$. <>= if (missing(newdata)) pred <- y[,ncol(y)] - object$residuals if (!missing(newdata) || se.fit) { <> } @ The more general case makes use of the [agsurv] routine to calculate a survival curve for each strata. The routine is defined in the section on individual Cox survival curves. The code here closely matches that. The routine only returns values at the death times, so we need approx to get a complete index. One non-obvious, but careful choice is to use the residuals for the predicted value instead of the compuation below, whenever operating on the original data set. This is a consequence of the Efron approx. When someone in a new data set has exactly the same time as one of the death times in the original data set, the code below implicitly makes them the ``last'' death in the set of tied times. The Efron approx puts a tie somewhere in the middle of the pack. This is way too hard to work out in the code below, but thankfully the original Cox model already did it. However, it does mean that a different answer will arise if you set newdata = the original coxph data set. Standard errors have the same issue, but 1. they are hardly used and 2. the original coxph doesn't do that calculation. So we do what's easiest. <>= ustrata <- unique(oldstrat) risk <- exp(object$linear.predictors) x <- x - rep(object$means, each=nrow(x)) #subtract from each column if (missing(newdata)) #se.fit must be true se <- double(n) else { pred <- se <- double(nrow(mf2)) newx <- newx - rep(object$means, each=nrow(newx)) newrisk <- c(exp(newx %*% object$coef)) } survtype<- ifelse(object$method=='efron', 3,2) for (i in ustrata) { indx <- which(oldstrat == i) afit <- agsurv(y[indx,,drop=F], x[indx,,drop=F], weights[indx], risk[indx], survtype, survtype) afit.n <- length(afit$time) if (missing(newdata)) { # In this case we need se.fit, nothing else j1 <- approx(afit$time, 1:afit.n, y[indx,1], method='constant', f=0, yleft=0, yright=afit.n)$y chaz <- c(0, afit$cumhaz)[j1 +1] varh <- c(0, cumsum(afit$varhaz))[j1 +1] xbar <- rbind(0, afit$xbar)[j1+1,,drop=F] if (ncol(y)==2) { dt <- (chaz * x[indx,]) - xbar se[indx] <- sqrt(varh + rowSums((dt %*% object$var) *dt)) * risk[indx] } else { j2 <- approx(afit$time, 1:afit.n, y[indx,2], method='constant', f=0, yleft=0, yright=afit.n)$y chaz2 <- c(0, afit$cumhaz)[j2 +1] varh2 <- c(0, cumsum(afit$varhaz))[j2 +1] xbar2 <- rbind(0, afit$xbar)[j2+1,,drop=F] dt <- (chaz * x[indx,]) - xbar v1 <- varh + rowSums((dt %*% object$var) *dt) dt2 <- (chaz2 * x[indx,]) - xbar2 v2 <- varh2 + rowSums((dt2 %*% object$var) *dt2) se[indx] <- sqrt(v2-v1)* risk[indx] } } else { #there is new data use.x <- TRUE indx2 <- which(newstrat == i) j1 <- approx(afit$time, 1:afit.n, newy[indx2,1], method='constant', f=0, yleft=0, yright=afit.n)$y chaz <-c(0, afit$cumhaz)[j1+1] pred[indx2] <- chaz * newrisk[indx2] if (se.fit) { varh <- c(0, cumsum(afit$varhaz))[j1+1] xbar <- rbind(0, afit$xbar)[j1+1,,drop=F] } if (ncol(y)==2) { if (se.fit) { dt <- (chaz * newx[indx2,]) - xbar se[indx2] <- sqrt(varh + rowSums((dt %*% object$var) *dt)) * newrisk[indx2] } } else { j2 <- approx(afit$time, 1:afit.n, newy[indx2,2], method='constant', f=0, yleft=0, yright=afit.n)$y chaz2 <- approx(-afit$time, afit$cumhaz, -newy[indx2,2], method="constant", rule=2, f=0)$y chaz2 <-c(0, afit$cumhaz)[j2+1] pred[indx2] <- (chaz2 - chaz) * newrisk[indx2] if (se.fit) { varh2 <- c(0, cumsum(afit$varhaz))[j1+1] xbar2 <- rbind(0, afit$xbar)[j1+1,,drop=F] dt <- (chaz * newx[indx2,]) - xbar dt2 <- (chaz2 * newx[indx2,]) - xbar2 v2 <- varh2 + rowSums((dt2 %*% object$var) *dt2) v1 <- varh + rowSums((dt %*% object$var) *dt) se[indx2] <- sqrt(v2-v1)* risk[indx2] } } } } @ %\subsection{Linear predictor, risk, and terms} For these three options what is returned is a \emph{relative} prediction which compares each observation to the average for the data set. Partly this is practical. Say for instance that a treatment covariate was coded as 0=control and 1=treatment. If the model were refit using a new coding of 3=control 4=treatment, the results of the Cox model would be exactly the same with respect to coefficients, variance, tests, etc. The raw linear predictor $X\beta$ however would change, increasing by a value of $3\beta$. The relative predictor \begin{equation} \eta_i = X_i\beta - (1/n)\sum_j X_j\beta \label{eq:eta} \end{equation} will stay the same. The second reason for doing this is that the Cox model is a relative risks model rather than an absolute risks model, and thus relative predictions are almost certainly what the user was thinking of. When the fit was for a stratified Cox model more care is needed. For instance assume that we had a fit that was stratified by sex with covaritate $x$, and a second data set were created where for the females $x$ is replaced by $x+3$. The Cox model results will be unchanged for the two models, but the `normalized' linear predictors $(x - \overline x)'\beta$ %` will not be the same. This reflects a more fundamental issue that the for a stratified Cox model relative risks are well defined only \emph{within} a stratum, i.e. for subject pairs that share a common baseline hazard. The example above is artificial, but the problem arises naturally whenever the model includes a strata by covariate interaction. So for a stratified Cox model the predictions should be forced to sum to zero within each stratum, or equivalently be made relative to the weighted mean of the stratum. Unfortunately, this important issue was not realized until late in 2009 when a puzzling query was sent to the author involving the results from such an interaction. Note that this issue did not arise with type='expected', which has a natural scaling. An offset variable, if specified, is treated like any other covariate with respect to centering. The logic for this choice is not as compelling, but it seemed the best that I could do. Note that offsets play no role whatever in predicted terms, only in the lp and risk. Start with the simple ones <>= if (is.null(object$coefficients)) coef<-numeric(0) else { # Replace any NA coefs with 0, to stop NA in the linear predictor coef <- ifelse(is.na(object$coefficients), 0, object$coefficients) } if (missing(newdata)) { offset <- offset - mean(offset) if (has.strata && reference=="strata") { # We can't use as.integer(oldstrat) as an index, if oldstrat is # a factor variable with unrepresented levels as.integer could # give 1,2,5 for instance. xmeans <- rowsum(x*weights, oldstrat)/c(rowsum(weights, oldstrat)) newx <- x - xmeans[match(oldstrat,row.names(xmeans)),] } else if (use.x) newx <- x - rep(object$means, each=nrow(x)) } else { offset <- newoffset - mean(offset) if (has.strata && reference=="strata") { xmeans <- rowsum(x*weights, oldstrat)/c(rowsum(weights, oldstrat)) newx <- newx - xmeans[match(newstrat, row.names(xmeans)),] } else newx <- newx - rep(object$means, each=nrow(newx)) } if (type=='lp' || type=='risk') { if (use.x) pred <- drop(newx %*% coef) + offset else pred <- object$linear.predictors if (se.fit) se <- sqrt(rowSums((newx %*% object$var) *newx)) if (type=='risk') { pred <- exp(pred) if (se.fit) se <- se * sqrt(pred) # standard Taylor series approx } } @ The type=terms residuals are a bit more work. In Splus this code used the Build.terms function, which was essentially the code from predict.lm extracted out as a separate function. As of March 2010 (today) a check of the Splus function and the R code for predict.lm revealed no important differences. A lot of the bookkeeping in both is to work around any possible NA coefficients resulting from a singularity. The basic formula is to \begin{enumerate} \item If the model has an intercept, then sweep the column means out of the X matrix. We've already done this. \item For each term separately, get the list of coefficients that belong to that term; call this list [[tt]]. \item Restrict $X$, $\beta$ and $V$ (the variance matrix) to that subset, then the linear predictor is $X\beta$ with variance matrix $X V X'$. The standard errors are the square root of the diagonal of this latter matrix. This can be computed, as colSums((X %*% V) * X)). \end{enumerate} Note that the [[assign]] component of a coxph object is the same as that found in Splus models (a list), most R models retain a numeric vector which contains the same information but it is not as easily used. The first first part of predict.lm in R rebuilds the list form as its [[asgn]] variable. I can skip this part since it is already done. <>= else if (type=='terms') { asgn <- object$assign nterms<-length(asgn) pred<-matrix(ncol=nterms,nrow=NROW(newx)) dimnames(pred) <- list(rownames(newx), names(asgn)) if (se.fit) se <- pred for (i in 1:nterms) { tt <- asgn[[i]] tt <- tt[!is.na(object$coefficients[tt])] xtt <- newx[,tt, drop=F] pred[,i] <- xtt %*% object$coefficient[tt] if (se.fit) se[,i] <- sqrt(rowSums((xtt %*% object$var[tt,tt]) *xtt)) } pred <- pred[,terms, drop=F] if (se.fit) se <- se[,terms, drop=F] attr(pred, 'constant') <- sum(object$coefficients*object$means, na.rm=T) } @ To finish up we need to first expand out any missings in the result based on the na.action, and optionally collapse the results within a subject. What should we do about the standard errors when collapse is specified? We assume that the individual pieces are independent and thus var(sum) = sum(variances). The statistical justification of this is quite solid for the linear predictor, risk and terms type of prediction due to independent increments in a martingale. For expecteds the individual terms are positively correlated so the se will be too small. One solution would be to refuse to return an se in this case, but the the bias should usually be small, and besides it would be unkind to the user. Prediction of type='terms' is expected to always return a matrix, or the R termplot() function gets unhappy. <>= if (type != 'terms') { pred <- drop(pred) if (se.fit) se <- drop(se) } if (!is.null(na.action.used)) { pred <- napredict(na.action.used, pred) if (is.matrix(pred)) n <- nrow(pred) else n <- length(pred) if(se.fit) se <- napredict(na.action.used, se) } if (!missing(collapse) && !is.null(collapse)) { if (length(collapse) != n2) stop("Collapse vector is the wrong length") pred <- rowsum(pred, collapse) # in R, rowsum is a matrix, always if (se.fit) se <- sqrt(rowsum(se^2, collapse)) if (type != 'terms') { pred <- drop(pred) if (se.fit) se <- drop(se) } } if (se.fit) list(fit=pred, se.fit=se) else pred @ survival/noweb/coxsurv.Rnw0000644000175100001440000006042112505342110015452 0ustar hornikusers@ \section{Cox models} \subsection{Predicted survival} The [[survfit]] method for a Cox model produces individual survival curves. As might be expected these have much in common with ordinary survival curves, and share many of the same methods. The primary differences are first that a predicted curve always refers to a particular set of covariate values. It is often the case that a user wants multiple values at once, in which case the result will be a matrix of survival curves with a row for each time and a column for each covariate set. The second is that the computations are somewhat more difficult. The input arguments are \begin{description} \item[formula] a fitted object of class 'coxph'. The argument name of 'formula' is historic, from when the survfit function was not a generic and only did Kaplan-Meier type curves. \item[newdata] contains the data values for which curves should be produced, one per row \item[se.fit] TRUE/FALSE, should standard errors be computed. \item[individual] a particular option for time-dependent covariates \item[type] computation type for the survival curve \item[vartype] computation type for the variance \item[censor] if FALSE, remove any times that have no events from the output. This is for backwards compatability with older versions of the code. \item[id] replacement and extension for the individual argument \end{description} All the other arguments are common to all the methods, refer to the help pages. <>= survfit.coxph <- function(formula, newdata, se.fit=TRUE, conf.int=.95, individual=FALSE, type, vartype, conf.type=c("log", "log-log", "plain", "none"), censor=TRUE, id, na.action=na.pass, ...) { Call <- match.call() Call[[1]] <- as.name("survfit") #nicer output for the user object <- formula #'formula' because it has to match survfit <> <> <> } @ The third line [[as.name('survfit')]] causes the printout to say `survfit' instead of `survfit.coxph'. %' The setup for the routine is fairly pedestrian. If the newdata argument is missing we use [[object$means]] as the default value. This choice has lots of statistical shortcomings, particularly in a stratified model, but is common in other packages and a historic option here. If the type or vartype are missing we use the appropriate one for the method in the Cox model. That is, the [[coxph]] computation used for [[method=``exact'']] is the same approximation used in the Kalbfleish-Prentice estimate, that for the Breslow method matches the Aalen survival estimate, and the Efron approximation the Efron survival estimate. The other two rows of labels in [[temp1]] are historical; we include them for backwards compatability but they don't appear in the documentation. %' <>= if (!is.null(attr(object$terms, "specials")$tt)) stop("The survfit function can not yet process coxph models with a tt term") if (missing(type)) { # Use the appropriate one from the model temp1 <- c("exact", "breslow", "efron") survtype <- match(object$method, temp1) } else { temp1 <- c("kalbfleisch-prentice", "aalen", "efron", "kaplan-meier", "breslow", "fleming-harrington", "greenwood", "tsiatis", "exact") survtype <- match(match.arg(type, temp1), temp1) survtype <- c(1,2,3,1,2,3,2,2,1)[survtype] } if (missing(vartype)) { vartype <- survtype } else { temp2 <- c("greenwood", "aalen", "efron", "tsiatis") vartype <- match(match.arg(vartype, temp2), temp2) if (vartype==4) vartype<- 2 } if (!se.fit) conf.type <- "none" else conf.type <- match.arg(conf.type) @ I need to retrieve a copy of the original data. We always need the $X$ matrix and $y$, both of which may be found in the data object. If the original call included either strata, offset, or weights, or if either $x$ or $y$ are missing from the [[coxph]] object, then the model frame will need to be reconstructed. We have to use [[object['x']]] instead of \texttt{object\$x} since the latter will pick off the [[xlevels]] component if the [[x]] component is missing (which is the default). <>= has.strata <- !is.null(attr(object$terms, 'specials')$strata) if (is.null(object$y) || is.null(object[['x']]) || !is.null(object$call$weights) || (has.strata && is.null(object$strata)) || !is.null(attr(object$terms, 'offset'))) { mf <- model.frame(object) } else mf <- NULL #useful for if statements later @ If a model frame was created, then it is trivial to grab [[y]] from the new frame and compare it to [[object$y]] from the original one. This is to avoid nonsense results that arise when someone changes the data set under our feet. For instance \begin{verbatim} fit <- coxph(Surv(time,status) ~ age, data=lung) lung <- lung[1:100,] survfit(fit) \end{verbatim} <>= if (is.null(mf)) y <- object[['y']] else { y <- model.response(mf) y2 <- object[['y']] if (!is.null(y2) && any(as.matrix(y2) != as.matrix(y))) stop("Could not reconstruct the y vector") } if (is.null(object[['x']])) x <- model.matrix.coxph(object, data=mf) else x <- object[['x']] n <- nrow(y) if (n != object$n[1] || nrow(x) !=n) stop("Failed to reconstruct the original data set") if (is.null(mf)) wt <- rep(1., n) else { wt <- model.weights(mf) if (is.null(wt)) wt <- rep(1.0, n) } type <- attr(y, 'type') if (type != 'right' && type != 'counting') stop("Cannot handle \"", type, "\" type survival data") missid <- missing(id) # I need this later, and setting id below makes # "missing(id)" always false if (!missid) individual <- TRUE else if (missid && individual) id <- rep(0,n) #dummy value else id <- NULL if (individual && missing(newdata)) { stop("the id and/or individual options only make sense with new data") } if (individual && type!= 'counting') stop("The individual option is only valid for start-stop data") if (is.null(mf)) offset <- 0 else { offset <- model.offset(mf) if (is.null(offset)) offset <- 0 } Terms <- object$terms if (!has.strata) strata <- rep(0L,n) else { stangle <- untangle.specials(Terms, 'strata') # used multiple times strata <- object$strata #try this first if (is.null(strata)){ if (length(stangle$vars) ==1) strata <- mf[[stangle$vars]] else strata <- strata(mf[, stangle$vars], shortlabel=TRUE) } } @ In two places below we need to know if there are strata by covariate interactions, which requires looking at attributes of the terms object. The factors attribute will have a row for the strata variable, or maybe more than one (multiple strata terms are legal). If it has a 1 in a column that corresponds to something of order 2 or greater, that is a strata by covariate interaction. <>= if (has.strata) { temp <- attr(Terms, "specials")$strata factors <- attr(Terms, "factors")[temp,] strata.interaction <- any(t(factors)*attr(Terms, "order") >1) } @ If a variable is deemed redundant the [[coxph]] routine will have set its coefficient to NA as a marker. We want to ignore that coefficient: treating it as a zero has the desired effect. Another special case is a null model, having either ~1 or only an offset on the right hand side. In that case we create a dummy covariate to allow the rest of the code to work without special if/else. The last special case is a model with a sparse frailty term. We treat the frailty coefficients as 0 variance (in essence as an offset). The frailty is removed from the model variables but kept in the risk score. This isn't statistically very defensible, but it is backwards compatatble. %' A non-sparse frailty does not need special code and works out like any other variable. We also remove the means from each column of the $X$ matrix. The reason for this is to avoid huge values when calculating $\exp(X\beta)$; this would happen if someone had a variable with a mean of 1000 and a variance of 1. Any constant can be subtracted, mathematically the results are identical as long as the same values are subtracted from the old and new $X$ data. The mean is used because it is handy, we just need to get $X\beta$ in the neighborhood of zero. One particular special case (that gave me fits for a while) is when there are non-heirarchical models, for example [[~ age + age:sex]]. The fit of such a model will \emph{not} be the same using the variable [[age2 <- age-50]]; I originally thought it was a flaw induced by my subtraction. This is simply a bad model and it is not clear that there is any ``correct'' behavior in creating predicted survival curves. <>= if (is.null(x) || ncol(x)==0) { # a model with ~1 on the right hand side # Give it a dummy x so the rest of the code goes through # (This case is really rare) x <- matrix(0., nrow=n) coef <- 0.0 varmat <- matrix(0.0,1,1) risk <- rep(exp(offset- mean(offset)), length=n) } else { varmat <- object$var coef <- ifelse(is.na(object$coefficients), 0, object$coefficients) xcenter <- object$means if (is.null(object$frail)) { x <- scale(x, center=xcenter, scale=FALSE) risk <- c(exp(x%*% coef + offset - mean(offset))) } else { keep <- !is.na(match(dimnames(x)[[2]], names(coef))) x <- x[,keep, drop=F] # varmat <- varmat[keep,keep] #coxph already has trimmed it risk <- exp(object$linear.predictor) x <- scale(x, center=xcenter, scale=FALSE) } } @ The [[risk]] vector and [[x]] matrix come from the original data, and are the raw data for the survival curve and its variance. We also need the risk score $\exp(X\beta)$ for the target subject(s). \begin{itemize} \item For predictions with time-dependent covariates the user will have either included an [[id]] statement (newer style) or specified the [[individual=TRUE]] option. If the latter, then [[newdata]] is presumed to contain only a single indivual represented by multiple rows. If the former then the [[id]] variable marks separate individuals. In either case we need to retrieve the covariates, strata, and repsonse from the new data set. \item For ordinary predictions only the covariates are needed. \item If newdata is not present we assume that this is the ordinary case, and use the value of [[object$means]] as the default covariate set. This is not ideal statistically since many users view this as an ``average'' survival curve, which it is not. \end{itemize} When grabbing [newdata] we want to use model.frame processing, both to handle missing values correctly and, perhaps more importantly, to correctly map any factor variables between the original fit and the new data. (The new data will often have only one of the original levels represented.) Also, we want to correctly handle data-dependent nonlinear terms such as ns and pspline. However, the simple call found in predict.lm, say, [[model.frame(Terms, data=newdata, ..]] isn't used here for a few reasons. The first is a decision on our part that the user should not have to include unused terms in the model. The second is that if there are strata, the user may or may not have included strata variables in their data set and we need to act accordingly. The third is that we might have an [[id]] statement in this call, which is another variable to be fetched. Last, there is no ability to use sparse frailties and newdata together; it is a hard case and so rare as to not be worth it. First, remove unnecessary terms from the orginal model formula. Any [[cluster]] terms can be deleted, If [[individual]] is false then the repsonse variable can go. The dataClasses and predvars attributes, if present, have elements in the same order as the first dimension of the ``factors'' attribute of the terms. Subscripting the terms argument does not preserve dataClasses or predvars, however. Use the pre and post subscripting factors attribute to determine what elements of them to keep. The predvars component is a call objects with one element for each term in the formula, so y ~ age + ns(height) would lead to a predvars of length 4, element 1 is the call itself, 2 would be y, etc. The dataClasses object is a simple list. <>= subterms <- function(tt, i) { dataClasses <- attr(tt, "dataClasses") predvars <- attr(tt, "predvars") oldnames <- dimnames(attr(tt, 'factors'))[[1]] tt <- tt[i] index <- match(dimnames(attr(tt, 'factors'))[[1]], oldnames) if (length(index) >0) { if (!is.null(predvars)) attr(tt, "predvars") <- predvars[c(1, index+1)] if (!is.null(dataClasses)) attr(tt, "dataClasses") <- dataClasses[index] } tt } temp <- untangle.specials(Terms, 'cluster') if (length(temp$vars)) Terms <- subterms(Terms, -temp$terms) if (missing(newdata)) { mf2 <- as.list(object$means) #create a dummy newdata names(mf2) <- names(object$coefficients) mf2 <- as.data.frame(mf2) found.strata <- FALSE } else { if (!is.null(object$frail)) stop("Newdata cannot be used when a model has frailty terms") Terms2 <- Terms if (!individual) Terms2 <- delete.response(Terms) <> } @ For backwards compatability, I allow someone to give an ordinary vector instead of a data frame (when only one curve is required). In this case I also need to verify that the elements have a name. Then turn it into a data frame, like it should have been from the beginning. (Documentation of this ability has been suppressed, however. I'm hoping people forget it ever existed.) <>= if (is.vector(newdata, "numeric")) { if (individual) stop("newdata must be a data frame") if (is.null(names(newdata))) { stop("Newdata argument must be a data frame") } newdata <- data.frame(as.list(newdata)) } @ Finally get my new model frame mf2. There are two cases. If the call does not has an ``id'' argument then we use the semantics of top-level functions like coxph: get a copy of the call, keep what we need, change the called function's name to ``model.fram'' and evalutate it. then we If all is particularly simple we can use a simple call. Otherwise get an abbreviated form of the original call that has only the calling function, na.action, and id. The calling function is always element 1, the others are found by name. Now manipulate it: add the formula, data and xlev components (the last might be NULL), and then change the name of the call. If the original call was [[survfit(fit1, newdata=mydat, conf.int=.9)]] the result is [[model.frame(data= copy of newdat, formula=Terms2, xlev=myxlev)]]. If there is no id argument we use a simple call, except that we allow the user to leave out any strata() variables if they so desire, \emph{if} there are no strata by covariate interactions. How does one check if the strata variables are or are not available in the call? My first attempt at this was to wrap the call in a try() construct and see if it failed. This doesn't work. \begin{itemize} \item What if there is no strata variable in newdata, but they do have, by bad luck, a variable of the same name in their main directory? \item It would seem like changing the environment to NULL would be wise, so that we don't find variables anywhere but in the data argument, a sort of sandboxing. Not wise: you then won't find functions like ``log''. \item We don't dare modify the environment of the formula at all. It is needed for the sneaky caller who uses his own function inside the formula, 'mycosine' say, and that function can only be found if we retain the environment. \end{itemize} One way out of this is to evaluate each of the strata terms (there can be more than one) one at a time, in an environment that knows nothing except "list" and a fake definition of "strata", and newdata. Variables that are part of the global environment won't be found. I even watch out for the case of either "strata" or "list" is the name of the stratification variable, which causes my fake strata function to return a function when said variable is not in newdata. <>= if (missid) { if (has.strata && !strata.interaction) { found.strata <- TRUE tempenv <- new.env(, parent=emptyenv()) assign("strata", function(..., na.group, shortlabel, sep) list(...), envir=tempenv) assign("list", list, envir=tempenv) for (svar in stangle$vars) { temp <- try(eval(parse(text=svar), newdata, tempenv), silent=TRUE) if (!is.list(temp) || any(unlist(lapply(temp, class))== "function")) found.strata <- FALSE } if (found.strata) mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=object$xlevels) else { Terms2 <- subterms(Terms2, -attr(Terms2, 'specials')$strata) if (!is.null(object$xlevels)) { myxlev <- object$xlevels[match(attr(Terms2, "term.labels"), names(object$xlevels), nomatch=0)] if (length(myxlev)==0) myxlev <- NULL } else myxlev <- NULL mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=myxlev) } } else { mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=object$xlevels) found.strata <- has.strata #would have failed otherwise } } else { tcall <- Call[c(1, match(c('id', "na.action"), names(Call), nomatch=0))] tcall$data <- newdata tcall$formula <- Terms2 tcall$xlev <- object$xlevels tcall[[1]] <- as.name('model.frame') mf2 <- eval(tcall) found.strata <- has.strata # would have failed otherwise } @ Now, finally, extract the [[x2]] matrix from the just-created frame. <>= if (has.strata && found.strata) { #pull them off temp <- untangle.specials(Terms2, 'strata') strata2 <- strata(mf2[temp$vars], shortlabel=TRUE) strata2 <- factor(strata2, levels=levels(strata)) if (any(is.na(strata2))) stop("New data set has strata levels not found in the original") Terms2 <- Terms2[-temp$terms] } else strata2 <- factor(rep(0, nrow(mf2))) if (individual) { if (missing(newdata)) stop("The newdata argument must be present when individual=TRUE") if (!missid) { #grab the id variable id <- model.extract(mf2, "id") if (is.null(id)) stop("id=NULL is an invalid argument") } else id <- rep(1, nrow(mf2)) x2 <- model.matrix(Terms2, mf2)[,-1, drop=FALSE] #no intercept if (length(x2)==0) stop("Individual survival but no variables") x2 <- scale(x2, center=xcenter, scale=FALSE) offset2 <- model.offset(mf2) if (length(offset2) >0) offset2 <- offset2 - mean(offset) else offset2 <- 0 y2 <- model.extract(mf2, 'response') if (attr(y2,'type') != type) stop("Survival type of newdata does not match the fitted model") if (attr(y2, "type") != "counting") stop("Individual=TRUE is only valid for counting process data") y2 <- y2[,1:2, drop=F] #throw away status, it's never used newrisk <- exp(c(x2 %*% coef) + offset2) result <- survfitcoxph.fit(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat, id, y2, strata2) } @ If there is no newdata argument, the centering means that we need to predict for x2=0. The second the most common call to the routine. <>= else { if (missing(newdata)) { if (has.strata && strata.interaction) stop ("Models with strata by covariate interaction terms require newdata") x2 <- matrix(0.0, nrow=1, ncol=ncol(x)) offset2 <- 0 } else { offset2 <- model.offset(mf2) if (length(offset2) >0) offset2 <- offset2 - mean(offset) else offset2 <- 0 x2 <- model.matrix(Terms2, mf2)[,-1, drop=FALSE] #no intercept x2 <- scale(x2, center=xcenter, scale=FALSE) } newrisk <- exp(c(x2 %*% coef) + offset2) result <- survfitcoxph.fit(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat) if (has.strata && found.strata) { if (is.matrix(result$surv)) { <> } } } @ The final bit of work. If the newdata arg contained strata then the user should not get a matrix of survival curves containing every newdata obs * strata combination, but rather a vector of curves, each one with the appropriate strata. It was faster to compute them all, however, than to use the individual=T logic. So now pick off the bits we want. The names of the curves will be the rownames of the newdata arg, if they exist. <>= nr <- nrow(result$surv) #a vector if newdata had only 1 row indx1 <- split(1:nr, rep(1:length(result$strata), result$strata)) rows <- indx1[as.numeric(strata2)] #the rows for each curve indx2 <- unlist(rows) #index for time, n.risk, n.event, n.censor indx3 <- as.integer(strata2) #index for n and strata for(i in 2:length(rows)) rows[[i]] <- rows[[i]]+ (i-1)*nr #linear subscript indx4 <- unlist(rows) #index for surv and std.err temp <- result$strata[indx3] names(temp) <- row.names(mf2) new <- list(n = result$n[indx3], time= result$time[indx2], n.risk= result$n.risk[indx2], n.event=result$n.event[indx2], n.censor=result$n.censor[indx2], strata = temp, surv= result$surv[indx4], cumhaz = result$cumhaz[indx4]) if (se.fit) new$std.err <- result$std.err[indx4] result <- new @ Finally, the last (somewhat boring) part of the code. First, if given the argument [[censor=FALSE]] we need to remove all the time points from the output at which there was only censoring activity. This action is mostly for backwards compatability with older releases that never returned censoring times. Second, add in the variance and the confidence intervals to the result. The code is nearly identical to that in survfitKM. <>= if (!censor) { kfun <- function(x, keep){ if (is.matrix(x)) x[keep,,drop=F] else if (length(x)==length(keep)) x[keep] else x} keep <- (result$n.event > 0) if (!is.null(result$strata)) { temp <- factor(rep(names(result$strata), result$strata), levels=names(result$strata)) result$strata <- c(table(temp[keep])) } result <- lapply(result, kfun, keep) } if (se.fit) { zval <- qnorm(1- (1-conf.int)/2, 0,1) if (conf.type=='plain') { temp1 <- result$surv + zval* result$std.err * result$surv temp2 <- result$surv - zval* result$std.err * result$surv result <- c(result, list(upper=pmin(temp1,1), lower=pmax(temp2,0), conf.type='plain', conf.int=conf.int)) } if (conf.type=='log') { xx <- ifelse(result$surv==0,1,result$surv) #avoid some "log(0)" messages temp1 <- ifelse(result$surv==0, 0*result$std.err, exp(log(xx) + zval* result$std.err)) temp2 <- ifelse(result$surv==0, 0*result$std.err, exp(log(xx) - zval* result$std.err)) result <- c(result, list(upper=pmin(temp1,1), lower=temp2, conf.type='log', conf.int=conf.int)) } if (conf.type=='log-log') { who <- (result$surv==0 | result$surv==1) #special cases xx <- ifelse(who, .1,result$surv) #avoid some "log(0)" messages temp1 <- exp(-exp(log(-log(xx)) + zval*result$std.err/log(xx))) temp1 <- ifelse(who, result$surv + 0*result$std.err, temp1) temp2 <- exp(-exp(log(-log(xx)) - zval*result$std.err/log(xx))) temp2 <- ifelse(who, result$surv + 0*result$std.err, temp2) result <- c(result, list(upper=temp1, lower=temp2, conf.type='log-log', conf.int=conf.int)) } } result$call <- Call # The "type" component is in the middle -- match history indx <- match('surv', names(result)) result <- c(result[1:indx], type=attr(y, 'type'), result[-(1:indx)]) if (is.R()) class(result) <- c('survfit.cox', 'survfit') else oldClass(result) <- 'survfit.cox' result @ survival/noweb/survfit.Rnw0000644000175100001440000002310312334154566015457 0ustar hornikusers\section{Survival curves} The survfit function was set up as a method so that we could apply the function to both formulas (to compute the Kaplan-Meier) and to coxph objects. The downside to this is that the manual pages get a little odd, but from a programming perspective it was a good idea. At one time, long long ago, we allowed the function to be called with ``Surv(time, status)'' as the formula, i.e., without a tilde. That was a bad idea, now abandoned. <>= survfit <- function(formula, ...) { UseMethod("survfit", formula) } <> <> <> @ The result of a survival curve can have a [[surv]] component that is a vector or a matrix, and an optional strata component. A dual subscript to a survfit object always associates the first subscript with the strata and the second with the matrix. When a survfit object has only one or the other of the two, we allow a single subscript to be used and map it appropriately. <>= dim.survfit <- function(x) { if (is.null(x$strata)) { if (is.matrix(x$surv)) ncol(x$surv) else 1 } else { nr <- length(x$strata) if (is.matrix(x$xurv)) c(nr, ncol(x$surv)) else nr } } "[.survfit" <- function(x, ..., drop=TRUE) { nmatch <- function(indx, target) { # This function lets R worry about character, negative, or logical subscripts # It always returns a set of positive integer indices temp <- 1:length(target) names(temp) <- target temp[indx] } if (missing(..1)) i<- NULL else i <- ..1 if (missing(..2)) j<- NULL else j <- ..2 if (is.null(i) && is.null(j)) return (x) #no subscripts present! if (!is.matrix(x$surv) && !is.null(j)) stop("survfit object does not have 2 dimensions") if (is.null(x$strata)) { if (is.matrix(x$surv)) { if (is.null(j) && !is.null(i)) j <- i #special case noted above x$surv <- x$surv[,j,drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[,j,drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[,j,drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[,j,drop=drop] } else warning("survfit object has only a single survival curve") } else { if (is.null(i)) keep <- seq(along.with=x$time) else { indx <- nmatch(i, names(x$strata)) #strata to keep if (any(is.na(indx))) stop(paste("strata", paste(i[is.na(indx)], collapse=' '), 'not matched')) # Now, indx may not be in order: some can use curve[3:2] to reorder # The list/unlist construct will reorder the data temp <- rep(1:length(x$strata), x$strata) keep <- unlist(lapply(indx, function(x) which(temp==x))) if (length(indx) <=1 && drop) x$strata <- NULL else x$strata <- x$strata[i] x$n <- x$n[indx] x$time <- x$time[keep] x$n.risk <- x$n.risk[keep] x$n.event <- x$n.event[keep] x$n.censor<- x$n.censor[keep] if (!is.null(x$enter)) x$enter <- x$enter[keep] } if (is.matrix(x$surv)) { # If the curve has been selected by strata and keep has only # one row, we don't want to lose the second subscript too if (!is.null(i) && (is.null(j) ||length(j) >1)) drop <- FALSE if (is.null(j)) { x$surv <- x$surv[keep,,drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,,drop=drop] if (!is.null(x$upper)) x$upper <-x$upper[keep,,drop=drop] if (!is.null(x$lower)) x$lower <-x$lower[keep,,drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <-x$cumhaz[keep,,drop=drop] } else { x$surv <- x$surv[keep,j, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,j, drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[keep,j, drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[keep,j, drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[keep,j, drop=drop] } } else { x$surv <- x$surv[keep] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep] if (!is.null(x$upper)) x$upper <- x$upper[keep] if (!is.null(x$lower)) x$lower <- x$lower[keep] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[keep] } } x } @ \subsection{Kaplan-Meier} The most common use of the survfit function is with a formula as the first argument, and the most common outcome of such a call is a Kaplan-Meier curve. The id argument is from an older version of the competing risks code; most people will use [[cluster(id)]] in the formula instead. The istate argument only applies to competing risks, but don't print an error message if it is accidentally there. <>= survfit.formula <- function(formula, data, weights, subset, na.action, etype, id, istate, ...) { Call <- match.call() Call[[1]] <- as.name('survfit') #make nicer printout for the user # create a copy of the call that has only the arguments we want, # and use it to call model.frame() indx <- match(c('formula', 'data', 'weights', 'subset','na.action', 'istate', 'id', "etype"), names(Call), nomatch=0) #It's very hard to get the next error message other than malice # eg survfit(wt=Surv(time, status) ~1) if (indx[1]==0) stop("a formula argument is required") temp <- Call[c(1, indx)] temp[[1]] <- as.name("model.frame") m <- eval.parent(temp) Terms <- terms(formula, c("strata", "cluster")) ord <- attr(Terms, 'order') if (length(ord) & any(ord !=1)) stop("Interaction terms are not valid for this function") n <- nrow(m) Y <- model.extract(m, 'response') if (!is.Surv(Y)) stop("Response must be a survival object") casewt <- model.extract(m, "weights") if (is.null(casewt)) casewt <- rep(1,n) if (!is.null(attr(Terms, 'offset'))) warning("Offset term ignored") id <- model.extract(m, 'id') istate <- model.extract(m,"istate") temp <- untangle.specials(Terms, "cluster") if (length(temp$vars)>0) { if (length(temp$vars) > 1) stop("can not have two cluster terms") if (!is.null(id)) stop("can not have both a cluster term and an id variable") id <- m[[temp$vars]] Terms <- Terms[-temp$terms] } ll <- attr(Terms, 'term.labels') if (length(ll) == 0) X <- factor(rep(1,n)) # ~1 on the right else X <- strata(m[ll]) if (!is.Surv(Y)) stop("y must be a Surv object") # Backwards support for the now-depreciated etype argument etype <- model.extract(m, "etype") if (!is.null(etype)) { if (attr(Y, "type") == "mcounting" || attr(Y, "type") == "mright") stop("cannot use both the etype argument and mstate survival type") if (length(istate)) stop("cannot use both the etype and istate arguments") status <- Y[,ncol(Y)] etype <- as.factor(etype) temp <- table(etype, status==0) if (all(rowSums(temp==0) ==1)) { # The user had a unique level of etype for the censors newlev <- levels(etype)[order(-temp[,2])] #censors first } else newlev <- c(" ", levels(etype)[temp[,1] >0]) status <- factor(ifelse(status==0,0, as.numeric(etype)), labels=newlev) if (attr(Y, 'type') == "right") Y <- Surv(Y[,1], status, type="mstate") else if (attr(Y, "type") == "counting") Y <- Surv(Y[,1], Y[,2], status, type="mstate") else stop("etype argument incompatable with survival type") } # At one point there were lines here to round the survival # times to a certain number of digits. This approach worked # almost all the time, but only almost. The better logic is # now in the individual compuation routines if (attr(Y, 'type') == 'left' || attr(Y, 'type') == 'interval') temp <- survfitTurnbull(X, Y, casewt, ...) else if (attr(Y, 'type') == "right" || attr(Y, 'type')== "counting") temp <- survfitKM(X, Y, casewt, ...) else if (attr(Y, 'type') == "mright" || attr(Y, "type")== "mcounting") temp <- survfitCI(X, Y, weights=casewt, id=id, istate=istate, ...) else { # This should never happen stop("unrecognized survival type") } if (is.null(temp$states)) class(temp) <- 'survfit' else class(temp) <- c("survfitms", "survfit") if (!is.null(attr(m, 'na.action'))) temp$na.action <- attr(m, 'na.action') temp$call <- Call temp } @ Once upon a time I allowed survfit to be called without the `\textasciitilde 1' portion of the formula. This was a mistake for multiple reasons, but the biggest problem is timing. If the subject has a data statement but the first argument is not a formula, R needs to evaluate Surv(t,s) to know that it is a survival object, but it also needs to know that this is a survival object before evaluation in order to dispatch the correct method. The method below helps give a useful error message in some cases. <>= survfit.Surv <- function(formula, ...) stop("the survfit function requires a formula as its first argument") @ survival/noweb/tail0000644000175100001440000000007611773344220014137 0ustar hornikusers\bibliographystyle{plain} \bibliography{refer} \end{document} survival/noweb/ratetable.Rnw0000644000175100001440000004331711732700061015715 0ustar hornikusers\documentclass{article} \usepackage{noweb} \usepackage[pdftex]{graphicx} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \newcommand{\myfig}[1]{\resizebox{\textwidth}{!} {\includegraphics{#1.pdf}}} \SweaveOpts{keep.source=TRUE} \title{Rate tables in the Survival Package} \author{Terry Therneau and Megan O'Byrne} \begin{document} \maketitle <>= options(width=60, continue=" ") makefig <- function(file, top=1, right=1, left=4) { pdf(file, width=9.5, height=7, pointsize=18) par(mar=c(4, left, top, right) +.1) } @ \section{Introduction} This is a short introduction to how the United States rate tables found in the \emph{survival} package are generated. Data for new calendar years is regularly added; this is stored in the data directory and processed by a function there into the current rate table. Much more detail on how the rate tables are used can be found in a series of technical reports from the Department of Health Science Research, Mayo Clinic. \section{Data} The US death rates for 1940 to 2000 are based on the US Decennial Rate tables, published by the National Center for Health Statistics, within the Centers for Disease Control. The detailed address of the web pages changes, at the time of typing this \texttt{www.cdc.gov/nchs} will reach the top level, from which the documents can be found by using \emph{rate tables} in the search box. The structure of the tables has changed slightly over the years $$ \begin{tabular}{c|l|l} Year & Age & Race \\ \hline 1940 & 0--110 & total, white, black, other \\ 1950--1960 & 0--109 + first year & total, white, non-white \\ 1970--1990 & 0-1d, 1-7d, 7-28d, 28-365d, 1-109 & total, white, nowhite, black \\ 2000 & 0-1d, 1-7d, 7-28d, 28-365d, 1-109 & total, white, black \\ \end{tabular} $$ The first year data for 1950-1960 is a separate table, with much more detail than the other years. Publication dates of the decennial tables has lagged, however, with the 2000 table not appearing until late in 2008. The 2000 decennial data for individual states has not yet appeared as of this writing (Feb 2011). Data for 2001 and forward is taken from the annual life tables, which appear more promptly. The annual tables contain total, white, and black, do not break up the first year of life, and extend only to age 99. First year of life data for the later years is obtained from Report GMWK264A: Deaths under 1 Year, by Month, Age, Race, and Sex: United States, 1999-2007, which can be found in the National Vital Statistics System (within NCHS). The log(hazard) for the later ages is extrapolated from ages 70-99 using a smoothing spline. The source code for the \emph{survival} package contains a \texttt{noweb} directory with the code for this report, amongst other things, and a sub-directory \texttt{noweb/rates} containing the raw data for the rate tables. The decennial files have been subset and normalized into a common form, and found as \texttt{usdecennial.dat} and \texttt{minndecennial.dat}. The US data contains the $q$ values for all combinations of age (0-1 day, 2-7 d, 7-28 d, 28-365 d), sex (male, female) and race (total, white, black). The Minnesota table does not subdivide the first year of life, and contains data only for total and white. For individual years the \texttt{noweb/rates} directory contains individual files for each year. As new rate tables are published a new file is added with the following procedure. \begin{enumerate} \item The NCHS source is a pdf file, with tables for each combinations of race and sex. These will be saved as individual files with names like \texttt{us2006tf.csv}, \texttt{us2006wf.csv}, etc for the total female and white female tables, respectively. \item The code below, with names modified appropriately for the particular year, is used to create a single file containing the $q$ values for each age. \end{enumerate} <>= basename <- "us2006" suffix <- c("tm", "tf", "wm", "wf", "bm", "bf") #total male, ..., black female newdata <- list(age=paste(0:99, 1:100, sep='-')) for (i in 1:6) { tdata <- read.csv(paste(basename, suffix[j], '.csv', sep=''), col.names=c('age', 'q', 'n', 'deaths', 'L', "n.older", "expect", 'dummy'), sep=',', header=FALSE) newdata[[suffix[j]]] <- 10000* tdata$q } write.table(newdata, file=paste(basename, "dat", sep='.')) @ The infant mortality data is also updated as new information becomes available, usually by hand editing new lines into the file. Updates of the life tables and the infant mortality tables on the NCHS site are not coordinated. \section{Rate tables} A rate table is an array containing daily hazard rates $\lambda$. The US tables contain annual death rates $q = 1 - exp(\lambda* 365.25)$. (Technically one could keep track of leap year, using either 365 or 366 as appropriate for each calendar year. We do not.) They can be cross-classified in any way that is desired, various attributes of the table describe this cross classification. The attributes are \begin{description} \item[dim] The dimensions of the array \item[dimnames] The labels for each dimension. These two attributes act identically to those for a normal array. \item[dimid] A name for each dimension. For the survexp.us table these are ``age'', ``sex'', and ``year''. For survexp.usr they are ``age'', ``sex'', ``race'', and ``year''. \item[type] A vector giving the type of each dimension. One of \begin{itemize} \item 1: a categorical trait such as sex, race, smoking status, etc. As a particular subject is followed forward in time, this trait does not change. \item 2: a continuous trait that changes over time such as age. As a subject is followed forward, different rows of the ratetable apply when calculating the expected hazard. If I start at age 2 with 5 years of follow-up, the rates for ages 2--6 will be used. \item 3: a continuous trait, as in 2, but represented as a date. This allows the routine to be more intelligent in handling the multiple date formats available. \item 4: the calendar year dimension of a US rate table. \end{itemize} \item[cutpoints] A list with one element for each dimension \begin{itemize} \item If \texttt{type==1} the element should be NULL \item Otherwise it contains the vector of starting points for each row of the dimension, e.g., for the age dimension this defines the age range over which each row of the table applies. Because we only define the start of each row, the last row of the table implicitly extends to infinity. \end{itemize} \item[summary] a function to summarize data. \end{description} \section{Creating the US table} \subsection{Data} First read in the US Decennial data for 1940 to 2000, and fill the $q$ values into a temporary array. The first 3 ages are the q's for days 0-1, 1-7, and 7-28, the fourth is the q for the entire first year. Change the array to one of daily hazard rates. For the 4th row, make it so the sum of the first year's hazard is correct, i.e., 1*row1 + 6*row2 + 21*row3 + 337.25* row4 = -log(1-q) <>= decdata <- read.table('rates/usdecennial.dat', header=TRUE) temp <- array(decdata$q[decdata$race=='total'], dim=c(113,2,7)) usd <- -log(1- temp) usd[4,,] <- usd[4,,] - (usd[1,,] + usd[2,,] + usd[3,,]) usd[2,,] <- usd[2,,] /6 #days 1-7 usd[3,,] <- usd[3,,] /21 #days 7-28 usd[4,,] <- usd[4,,] /337.25 usd[5:113,,] <- usd[5:113,,]/365.25 @ Note a change from some earlier releases of the code. There are 36524 days per century, so I used 365.24. However the year 2000 is the exception to the exception: ``Every 4 years is a leap year, unless divisible by 100; unless divisible by 1000". So over the lifetime that these tables will be used 365.25 is the right number. (If they are still in use in the year 2100, some one else will be maintaining the code.) Plus, using .24 confused everyone. Now pull in the single year tables. The \texttt{temp3} array is made full size, even though the single year data is missing the subdivision of year 1 of life, and the older ages of 100-109. Thus the 4:103 subscript below. We keep the data for all races in \texttt{usy} but kept only the total column for \texttt{usd}, the reason why will appear when creating the table by race. <>= i <- 2004 while(file.exists(paste('rates/us', i, '.dat', sep=''))) i <- i+1 singleyear <- 1996:(i-1) #the data we have nsingle <- length(singleyear) temp <- array(0, dim=c(113, 2, 3, nsingle)) #age, sex, race, year for (i in 1:nsingle) { tdata <- read.table(paste("rates/us", singleyear[i], ".dat", sep=''), header=TRUE) temp[4:103,,,i] <- -log(1- c(as.matrix(tdata[,-1]))/100000) } @ Pull in the first year of life breakdown. We use that to compute the proportion of the first year hazard that falls into each interval. Then rescale to daily hazards. <>= infant <- read.csv('rates/usinfant.dat') iyears <- unique(infant$year) # extract the deaths for total (all races) # then scale each year/sex group out as proportions deaths <- array(as.matrix(infant[,3:8]), dim=c(4, length(iyears), 2,3), dimnames=list(c('0', '1-6','7-27', '28-365'), iyears, c("Male", "Female"), c("total", "white", "black"))) for (i in 1:length(iyears)) { for (j in 1:2) { for (k in 1:3) { deaths[,i,j,k] <- deaths[,i,j,k]/sum(deaths[,i,j,k]) } } } # Partition out the total 1 year hazard, and then rescale to daily hazards usy <- array(0., dim=c(113, 2,3, nsingle)) indx <- match(singleyear, iyears) indx[singleyear < min(iyears)] <- 1 for (i in 1:nsingle) { for (j in 1:2) { for (k in 1:3) { usy[1:4,j,k,i] <- temp[4,j,k,i] * deaths[,indx[i],j,k]/c(1,6,21, 337.25) } } } usy[5:113,,,] <- temp[5:113,,,] /365.25 @ \begin{figure} \myfig{extrapolate} \caption{Plot of selected decennial and annual year data, along with a spline fit from ages 70--99 and the extrapolation of that fit to age 109.} \label{fig:extrapolate} \end{figure} Now we want to extend the yearly data to ages 100--109. There are two reasons for this, the simple is so that we can match the decennial data. The more compelling one is to note how the expected survival routines make use of the rate tables: namely that for any continuous variable the largest dimension of the rate table is used for values exceeding the dimension. If a rate table extends to the year 2007, then for follow-up in 2008 and later the 2007 data is used; if age extends to 99 then the age 99 values are used for all earlier ages. Thus it is not a question of \emph{whether} we will extrapolate for age but \emph{how} we do so, with a constant hazard after age 99 or one with a more rational basis. We have noticed that the log(hazard) is remarkably linear after age 95, and so do the extrapolation on that scale. Figure \ref{fig:extrapolate} shows the results for the last 4 decennial years, along with 2006. % Because I like to manage my own plots as floating figures, I need to % use plot=F. Otherwise Sweave tries to do it. % <>= for (i in 1:2) { for (j in 1:3) { for (k in 1:nsingle) { loghaz <- log(usy[4+ 80:99, i, j, k]) tfit <- smooth.spline(80:99, loghaz, df=8) usy[4+ 100:109, i, j, k] <- exp(predict(tfit, 100:109)$y) } } } makefig("extrapolate.pdf") matplot(80:109, 100000* usd[4+ 80:109, 2, 4:7], log='y', xlab="Age", ylab="Daily hazard * 100,000", pch='7890') for (i in 4:7) { loghaz <- log(usd[4+ 80:99,2, i] * 100000) tfit <- smooth.spline(80:99, loghaz, df=8) lines(80:109, exp(predict(tfit, 80:109)$y), col=i-3) } temp <- usd[4+99,2,6]*100000 segments(99, temp, 109, temp, col=3, lty=2) ty <- match(2006, singleyear) points(80:99, 100000*usy[4+80:99,2,1,ty], col=1, pch=2) lines(80:109, 100000*usy[4+80:109,2,1,ty]) dev.off() @ \subsection{US total table} The US total table will have an entry for each single calendar year, and dimensions of age, sex, and year. The data for 1940 to 2000 is based on decennial values, interpolated across years, and for 2001 and onward on single year data. <>= years <- seq(1940, max(singleyear)) survexp.us <- array(0., dim=c(113, 2, length(years))) single2 <- singleyear[singleyear > 2000] xtemp <- c(1940, 1950, 1960, 1970, 1980, 1990, 2000, single2) for (i in 1:nrow(usd)) { for (j in 1:2) { #so what if loops are slow, we only do this once ytemp <- c(usd[i,j,], usy[i,j,1, match(single2, singleyear)]) survexp.us[i,j,] <- approx(xtemp, ytemp, xout=years)$y } } @ Rate tables store dates as the number of days since Jan 1 1960. This is historical and not every going to change. Users should make use of the \texttt{ratetableDate} function. I can't use it here, unfortunately, because this code is run in the process of making the survival library, so I cannot count on the function's existence. <>= if (exists('as.Date')) { # R datecut <- as.Date(paste(years, '/01/01', sep=''))- as.Date('1960/01/01') datecut <- as.integer(datecut) }else if (exists('month.day.year')) { #Splus datecut <- julian(1,1, years, origin=c(1,1,1960)) }else stop("Cannot find appropriate routine for dates") @ Adding the attributes is the fussy part, since these are what define a rate table. Users making their own table ``mytable''should always follow the exercise with \texttt{is.ratetable(mytable)}, but again I cannot count on the function being present. The summary function is called with a data frame containing values for each of the dimensions, and prints a message giving the observed range of each. Its primary function is provide a warning to users to invalid input; the most common is when someone uses age in years instead of in days and all the subjects are treated as being 1-100 days old. <>= attributes(survexp.us) <- list( dim= c(113,2, length(years)), dimnames = list(c('0-1d','1-7d', '7-28d', '28-365d', as.character(1:109)), c("male", "female"), years), dimid =c("age", "sex", "year"), type = c(2,1,4), cutpoints= list(c(0,1,7,28,1:109 * 365.25), NULL, datecut), summary = function(R) { x <- c(format(round(min(R[,1]) /365.25, 1)), format(round(max(R[,1]) /365.25, 1)), sum(R[,2]==1), sum(R[,2]==2)) if (is.R()) x2<- as.Date(c(min(R[,3]), max(R[,3])), origin='1960/01/01') else x2 <- timeDate(julian=c(min(R[,3]), max(R[,3]))) paste(" age ranges from", x[1], "to", x[2], "years\n", " male:", x[3], " female:", x[4], "\n", " date of entry from", x2[1], "to", x2[2], "\n") }) oldClass(survexp.us) <- 'ratetable' @ \subsection{US race table} The US tables have not been consistent in their race breakdown. \begin{itemize} \item Decennial tables \begin{itemize} \item 1940: total, white, nonwhite, black \item 1950--60: total, white, nonwhite, \item 1970--90: total, white, nonwhite, black \item 2000: total, white, black \end{itemize} \item 1997--2006 Annual tables: total, white, black \end{itemize} We will create a table for white/black only, and use the nonwhite data for 1950 and 1960. Then create our usd array as was done for the total rates. <>= temp <-array(0., dim=c(113,2,2,7)) #age, sex, race, year temp[,,1,] <- decdata$q[decdata$race=='white'] temp[,,2,c(1,4,5,6,7)] <- decdata$q[decdata$race=='black'] temp[,,2, 2:3] <- decdata$q[decdata$race=='nonwhite' & (decdata$year==1950 |decdata$year==1970)] usd <- -log(1- temp) usd[4,,,] <- usd[4,,,] - (usd[1,,,] + usd[2,,,] + usd[3,,,]) usd[2,,,] <- usd[2,,,]/6 #days 2-7 usd[3,,,] <- usd[3,,,]/21 #days 8-28 usd[4,,,] <- usd[4,,,]/337.25 #days 29-365.25 usd[5:113,,,] <- usd[5:113,,,] / 365.25 @ Building the remainder of the table is almost identical to the prior code. <>= survexp.usr <- array(0., dim=c(113, 2, 2, length(years))) for (i in 1:113) { for (j in 1:2) { #so what if loops are slow, we only do this once for (k in 1:2) { #race ytemp <- c(usd[i,j,k,], usy[i,j,k+1, match(single2, singleyear)]) survexp.usr[i,j,k,] <- approx(xtemp, ytemp, xout=years)$y } } } attributes(survexp.usr) <- list( dim= c(113,2,2, length(years)), dimnames = list(c('0-1d','1-7d', '7-28d', '28-365d', as.character(1:109)), c("male", "female"), c("white", "black"), years), dimid =c("age", "sex", "race", "year"), type = c(2,1,1,4), cutpoints= list(c(0,1,7,28,1:109 * 365.25), NULL, NULL, datecut), summary = function(R) { x <- c(format(round(min(R[,1]) /365.25, 1)), format(round(max(R[,1]) /365.25, 1)), sum(R[,2]==1), sum(R[,2]==2), sum(R[,3]==1), sum(R[,3]==2)) if (is.R()) x2<- as.Date(c(min(R[,4]), max(R[,4])), origin='1960/01/01') else x2 <- timeDate(julian=c(min(R[,4]), max(R[,4]))) paste(" age ranges from", x[1], "to", x[2], "years\n", " male:", x[3], " female:", x[4], "\n", " date of entry from", x2[1], "to", x2[2], "\n", " white:",x[7], " black:", x[8], "\n") }) oldClass(survexp.usr) <- "ratetable" @ And finally, save the tables away for use in the survival package. <>= save(survexp.us, survexp.usr, file="survexp.rda") @ \end{document} survival/noweb/casecohort.Rnw0000644000175100001440000004312612073122107016100 0ustar hornikusers\section{Case-cohort models} Prentice proposed a case-cohort design for large survey studies such as the Women's Health Study, where the population size makes it infeasable to collect data on all of the cases. In this case we might collect data on a random subcohort $SC$ of the full cohort of subjects $C$ along with information on all subjects $E$ who experience an event. Let $n$ be the size of the full cohort, $m$ the size of the subcohort $SC$ and $d$ be the number of events. Let $x$ be the covariate vector. The kernel of the Cox model's score equation has a term for each event $i$ of $x_i(t_i) - \bar{x}(t_i)$ where $\bar{x}$ is the mean of all subjects at risk at that time. In a case-cohort data set the naive estimate of $\bar{x}$ based on the available data will not be correct since the data set is enriched for deaths. The methods below all use a Cox model as the base, but ``fix up'' the result to correct for this bias. <>= cch <- function(formula, data, weights, subset, na.action, subcoh, id, stratum, cohort.size, method=c("Prentice", "SelfPrentice", "LinYing","I.Borgan","II.Borgan"), robust=FALSE){ Call <- match.call() method <- match.arg(method) <> <> <> <> } @ The call processing is a little unusual due to backwards compatability with an older version of the program which used separate model statements for each of the [[subcoh]], [[id]], and [[stratum]] arguments. This was a bad design because it does not properly handle missing values. The obvious way to check this is [[is.formula(subcoh)]] but that would start by retrieving the subcoh argument, which will fail if the variable is part of the data frame. Instead we have to parse the call itself and look for a tilde. If there is one, then we add this to the current formula minus the tilde, otherwise add the argument as is. <>= addterm <- function(oldform, new) { j <- length(oldform) oldform[[j]] <- call("+", oldform[[j]], new[[2]]) } newterm <- list() if (missing(subcoh)) stop ("the subcoh argument is required") else { if (is.call(Call$subcoh) && Call$subcoh[[1]] == as.name("~")) newterm$subcoh <- Call$subcoh[[2]] else newterm$subcoh <- Call$subcoh } if (!missing(id)){ if (is.call(Call$id) && Call$id[[1]] == as.name("~")) newterm$id <- Call$id[[2]] else newterm$id <- Call$id } if (!missing(stratum)) { if (is.call(Call$stratum) && call$stratum[[i]] == as.name("~")) newterm$stratum <- Call$stratum[[2]] else newterm$stratum <- Call$stratum } newform <- formula for (i in newterm) newform <- addterm(newform, newterm) # Formulas can also have cluster and strata terms in them # The next few lines are standard and almost identical to coxph indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(Call), nomatch=0) if (indx[1] ==0) stop("A formula argument is required") temp <- Call[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called special <- c("strata", "cluster") temp$formula <- if(missing(data)) terms(newform, special) else terms(newform, special, data=data) mf <- eval(temp, parent.frame()) Terms <- terms(mf) @ Now do a sanity check: squawk if they had both an id argument and a cluster term, or both a stratum argument and a strata term. <>= extras <- seq(length=length(newterm), to=length(mf)) names(extras) <- names(newterm) tdrop <- extras if (!is.null(attr(Terms, "specials")$id)) { if (!is.null(newterm$id)) stop("cannot have both an id argument and a cluster term in the formula") temp <- untangle.specials(Terms, "cluster") id <- m[[temp$vars]] tdrop <- c(tdrop, temp$terms) #to be dropped later } else id <- mf[[extras["id"]]] #will usually be NULL if (!is.null(attr(Terms, "specials")$strata)) { if (!is.null(newterms$stratum)) stop("cannot have both a stratum argument and a strata term in the formula") temp <- untangle.specials(Terms, "strata") if (length(temp$vars)==1) strata <- m[[stemp$vars]] else strata <- strata(m[, temp$vars], shortlabel=TRUE) tdrop <- c(tdrop, temp$terms) } else strata <- mf[[extras["stratum"]]] subcoh <- mf[[extras["subcoh"]]] Terms <- Terms[-tdrop] @ Now do a few checks on the retrieved variables <>= n <- nrow(mf) #number of observations if (is.logical(subcoh)) subcoh <- as.numeric(subcoh) if (!all(subcoh %in% 0:1)) stop("subcoh values must be 0/1 or FALSE/TRUE") if (n > sum(cohort.size)) stop("number of records is greater than the cohort size") if (!is.null(strata)) { if (method == "Prentice") method <- "I.Borgan" if (method == "Self-Prentice") method <- "II.Borgan" if (!(method %in% c("I.Borgan", "II.Borgan"))) stop("invalid method for stratified data") strata <- as.factor(strata) if (length(cohort.size)!=length(levels(strata))) stop("cohort.size and stratum do not match") subcohort.sizes <- table(strata) } else { if (length(cohort.size)!=1) stop("cohort size must be a scalar for unstratified analysis") subcohort.sizes <- n } if (robust && method != "LinYing") warning ("robust option ignored for this method") if (!is.null(id) && any(duplicated(id))) stop ("multiple records per id not allowed") @ Now create the data set that will be used for the calls to the underlying coxph code. It will be a (start, stop] data set, even if the input data is not. <>= Y <- model.extract(m, "response") if(!inherits(Y, "Surv")) stop("Response must be a survival object") ytype <- attr(Y, "type") if (!ytype %in% c("right", "counting")) stop("Cox model doesn't support \"", type, "\" survival data") if (ytype == "right") cdata <- data.frame(tstart= rep(0., n), tstop = Y[,1], cens = Y[,2], X = model.matrix(Terms, mf)[, -1, drop=FALSE) else cdata <- data.frame(tstart= Y[,1], tstop = Y[,2], cens = Y[,3], X = model.matrix(Terms, mf[, -1, drop=FALSE))) if (any (!subcoh & cens==0)) stop ("every observation should either be in the subcohort or be an event") @ Y[,3]) if (any(!subcoh & !cens)) stop(sum(!subcoh & !cens),"censored observations not in subcohort") cc<-cens+1-subcoh texit<-switch(itype+1, stop(), Y[,1], Y[,2]) tenter<-switch(itype+1, stop(), rep(0,length(texit)), Y[,1]) X <- model.matrix(Terms, m) X <- X[,2:ncol(X)] fitter <- get(method) if (stratified) out<-fitter(tenter=tenter, texit=texit, cc=cc, id=id, X=X, stratum=as.numeric(stratum), stratum.sizes=cohort.size) else out<-fitter(tenter=tenter, texit=texit, cc=cc, id=id, X=X, ntot=nn, robust=robust) out$method <- method names(out$coefficients) <- dimnames(X)[[2]] if(!is.null(out$var)) dimnames(out$var) <- list(dimnames(X)[[2]], dimnames(X)[[2]]) if(!is.null(out$naive.var)) dimnames(out$naive.var) <- list(dimnames(X)[[2]], dimnames(X)[[2]]) out$call <- call out$cohort.size <- cohort.size out$stratified<-stratified if (stratified){ out$stratum<-stratum out$subcohort.size <-subcohort.sizes } else { out$subcohort.size <- tt[2] } class(out) <- "cch" out } ### Subprograms Prentice <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.00000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators ## Calculate Prentice estimate ent2 <- tenter ent2[cc==2] <- texit[cc==2]-eps fit1 <- coxph(Surv(ent2,texit,cens)~X,eps=eps,x=TRUE) ## Calculate Prentice estimate and variance nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort X <- as.matrix(X) aent <- c(tenter[cc>0],tenter[cc<2]) aexit <- c(texit[cc>0],texit[cc<2]) aX <- rbind(as.matrix(X[cc>0,]),as.matrix(X[cc<2,])) aid <- c(id[cc>0],id[cc<2]) dum <- rep(-100,nd) dum <- c(dum,rep(0,nc)) gp <- rep(1,nd) gp <- c(gp,rep(0,nc)) fit <- coxph(Surv(aent,aexit,gp)~aX+offset(dum)+cluster(aid),eps=eps,x=TRUE, iter.max=35,init=fit1$coefficients) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db <- db[gp==0,] fit$phase2var<-(1-(nc/ntot))*t(db)%*%(db) fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$coefficients <- fit$coef <- fit1$coefficients fit } SelfPrentice <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.00000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators ## Calculate Self-Prentice estimate and variance nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort X <- as.matrix(X) aent <- c(tenter[cc>0],tenter[cc<2]) aexit <- c(texit[cc>0],texit[cc<2]) aX <- rbind(as.matrix(X[cc>0,]),as.matrix(X[cc<2,])) aid <- c(id[cc>0],id[cc<2]) dum <- rep(-100,nd) dum <- c(dum,rep(0,nc)) gp <- rep(1,nd) gp <- c(gp,rep(0,nc)) fit <- coxph(Surv(aent,aexit,gp)~aX+offset(dum)+cluster(aid),eps=eps,x=TRUE) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db <- db[gp==0,,drop=FALSE] fit$phase2var<-(1-(nc/ntot))*t(db)%*%(db) fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit } LinYing <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.000000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort ## Calculate Lin-Ying estimate and variance offs <- rep((ntot-nd)/(nc-ncd),length(texit)) offs[cc>0] <- 1 loffs <- log(offs) fit <- coxph(Surv(tenter, texit, cens)~X+offset(loffs)+cluster(id), eps=eps,x=TRUE) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db0 <- db[cens==0,,drop=FALSE] dbm <- apply(db0,2,mean) db0 <- sweep(db0,2,dbm) fit$phase2var<-(1-(nc-ncd)/(ntot-nd))*crossprod(db0) fit$naive.var <- fit$naive.var+fit$phase2var if (robust) fit$var<- crossprod(db,db/offs)+fit$phase2var else fit$var<-fit$naive.var fit } I.Borgan <- function(tenter, texit, cc, id, X, stratum, stratum.sizes){ eps <- 0.00000001 nobs <- length(texit) idx <- 1:length(nobs) jj <- max(stratum) nn <- stratum.sizes ## Cohort stratum sizes n <- table(stratum) ## Sample stratum sizes d <- table(stratum[cc>0]) ## Failures in each stratum tt <- table(cc,stratum) cens <- as.numeric(cc>0) ## Failure indicators subcoh <- as.numeric(cc<2) ## Subcohort indicators nd <- sum(cens) ## Number of failures nc <- sum(subcoh) ## Number in subcohort ncd <- sum(as.numeric(cc==1)) #Number of failures in subcohort m0 <- tt[1,] ## Subcohort stratum sizes (noncases only) if (ncd>0) m <- m0+tt[2,] else m <- m0 #Subcohort stratum sizes X <- as.matrix(X) kk <- ncol(X) ## Number of variables wt <- as.vector(nn/m) ## Weights for Estimator I stratum <- c(stratum[cc>0],stratum[cc<2]) w <- wt[stratum] ent <- c(tenter[cc > 0], tenter[cc < 2]) exit <- c(texit[cc > 0], texit[cc < 2]) X <- rbind(as.matrix(X[cc > 0, ]), as.matrix(X[cc < 2, ])) id <- c(id[cc > 0], id[cc < 2]) dum <- rep(-100, nd) dum <- c(dum, rep(0, nc)) gp <- rep(1, nd) gp <- c(gp, rep(0, nc)) w[gp==1] <- 1 fit <- coxph(Surv(ent,exit,gp)~X+offset(dum)+cluster(id), weights=w, eps=eps,x=T, iter.max=25) score <- resid(fit, type = "score", weighted=F) sc <- resid(fit, type="score", collapse=id, weighted=T) score <- as.matrix(score) score <- score[gp == 0,,drop=F] st <- stratum[gp==0] sto <- st %o% rep(1,kk) Index <- col(score) tscore <- tapply(score,list(sto,Index),mean) pscore <- tapply(score,list(sto,Index)) score <- score-tscore[pscore] delta <- matrix(0,kk,kk) opt <- NULL for (j in 1:jj) { temp <- t(score[st==j,])%*%score[st==j,]/(m[j]-1) delta <- matrix(delta+(wt[j]-1)*nn[j]*temp,kk,kk) if(is.null(opt)) opt <- nn[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var)) else opt <- rbind(opt,nn[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var))) } z <- apply(opt,2,sum) fit$opt <- sweep(opt,2,z,FUN="/") fit$phase2var<-fit$naive.var%*%delta%*%fit$naive.var fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$delta <- delta fit$sc <- sc fit } II.Borgan <- function(tenter, texit, cc, id, X, stratum, stratum.sizes){ eps <- 0.00000001 jj <- max(stratum) nn <- stratum.sizes ## Cohort stratum sizes n <- table(stratum) ## Sample stratum sizes d <- table(stratum[cc>0]) ## Failures in each stratum tt <- table(cc,stratum) cens <- as.numeric(cc>0) ## Failure indicators subcoh <- as.numeric(cc<2) ## Subcohort indicators nd <- sum(cens) ## Number of failures nc <- sum(subcoh) ## Number in subcohort ncd <- sum(as.numeric(cc==1)) #Number of failures in subcohort m0 <- tt[1,] ## Subcohort stratum sizes (controls only) if (ncd>0) m <- m0+tt[2,] else m <- m0 #Subcohort stratum sizes X <- as.matrix(X) kk <- ncol(X) ## Number of variables nn0 <- nn-as.vector(d) #Noncases in cohort wt <- as.vector(nn0/m0) w <- wt[stratum] w[cens==1] <- 1 fit <- coxph(Surv(tenter,texit,cens)~X+cluster(id), weights=w,eps=eps,x=T, iter.max=25) ## Borgan Estimate II score <- resid(fit, type = "score", weighted=F) sc <- resid(fit,type="score", collapse=id, weighted=T) score <- as.matrix(score) score <- score[cens == 0,,drop=F] ## Scores for controls st <- stratum[cens==0] ## Stratum indicators for controls sto <- st %o% rep(1,kk) Index <- col(score) tscore <- tapply(score,list(sto,Index),mean) ## Within stratum control score means pscore <- tapply(score,list(sto,Index)) score <- score-tscore[pscore] ## Subtract off within stratum score means delta <- matrix(0,kk,kk) opt <- NULL for (j in 1:jj) { temp <- t(score[st==j,])%*%score[st==j,]/(m0[j]-1) ## Borgan equation (19) delta <- delta+(wt[j]-1)*nn0[j]*temp ## Borgan equation (17) if(is.null(opt)) opt <- nn0[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var)) else opt <- rbind(opt,nn0[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var))) } z <- apply(opt,2,sum) fit$opt <- sweep(opt,2,z,FUN="/") fit$phase2var<-fit$naive.var %*% delta %*% fit$naive.var fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$delta <- delta fit$sc <- sc fit } ## Methods vcov.cch<-function(object,...) object$var "print.cch"<- function(x,...) { ## produces summary from an x of the class "cch" call<-x$call coef <- coef(x) method <- x$method se <- sqrt(diag(vcov(x))) Z<- abs(coef/se) p<- pnorm(Z) cohort.size<-x$cohort.size subcohort.size<-x$subcohort.size coefficients <- matrix(0, nrow = length(coef), ncol = 4) dimnames(coefficients) <- list(names(coef), c("Value", "SE", "Z", "p")) coefficients[, 1] <- coef coefficients[, 2] <- se coefficients[, 3] <- Z coefficients[, 4] <- 2*(1-p) if (x$stratified){ cat("Exposure-stratified case-cohort analysis,", x$method, "method.\n") m<-rbind(subcohort=x$subcohort.size, cohort=x$cohort.size) prmatrix(m,quote=FALSE) } else{ cat("Case-cohort analysis,") cat("x$method,", x$method,"\n with subcohort of", x$subcohort.size,"from cohort of", x$cohort.size,"\n\n") } cat("Call: "); print(x$call) cat("\nCoefficients:\n") print(coefficients) invisible(x) } "summary.cch"<-function(object,...) { ## produces summary from an object of the class "cch" call<-object$call coef <- coef(object) method <- object$method se <- sqrt(diag(vcov(object))) Z<- abs(coef/se) p<- pnorm(Z) cohort.size<-object$cohort.size subcohort.size<-object$subcohort.size coefficients <- matrix(0, nrow = length(coef), ncol = 4) dimnames(coefficients) <- list(names(coef), c("Value", "SE", "Z", "p")) coefficients[, 1] <- coef coefficients[, 2] <- se coefficients[, 3] <- Z coefficients[, 4] <- 2*(1-p) structure(list(call=call, method=method, cohort.size=cohort.size, subcohort.size=subcohort.size, coefficients = coefficients, stratified=object$stratified), class = "summary.cch") } print.summary.cch <- function(x,digits=3,...){ if (x$stratified){ cat("Exposure-stratified case-cohort analysis,", x$method, "method.\n") m<-rbind(subcohort=x$subcohort.size, cohort=x$cohort.size) prmatrix(m,quote=FALSE) } else{ cat("Case-cohort analysis,") cat("x$method,", x$method,"\n with subcohort of", x$subcohort.size,"from cohort of", x$cohort.size,"\n\n") } cat("Call: "); print(x$call) cat("\nCoefficients:\n") output<-cbind(Coef=x$coefficients[,1],HR=exp(x$coefficients[,1]), "(95%"=exp(x$coefficients[,1]-1.96*x$coefficients[,2]), "CI)"=exp(x$coefficients[,1]+1.96*x$coefficients[,2]), "p"=x$coefficients[,4] ) print(round(output,3)) invisible(x) } survival/noweb/exact.nw0000644000175100001440000004101212514444605014732 0ustar hornikusers\section{Exact partial likelihood} Let $r_i = \exp(X_i\beta)$ be the risk score for observation $i$. For one of the time points assume that there that there are $d$ tied deaths among $n$ subjects at risk. For convenience we will index them as $i= 1,\ldots,d$ in the $n$ at risk. Then for the exact parial likelihood, the contribution at this time point is \begin{align*} L &= \sum_{i=1}^d \log(r_i) - \log(D) \\ \frac{\partial L}{\partial \beta_j} &= x_{ij} - (1/D) \frac{\partial D}{\partial \beta_j} \\ \frac{\partial^2 L}{\partial \beta_j \partial \beta_k} &= (1/D^2)\left[D\frac{\partial^2D}{\partial \beta_j \partial \beta_k} - \frac{\partial D}{\partial \beta_j}\frac{\partial D}{\partial \beta_k} \right] \end{align*} The hard part of this computation is $D$, which is a sum \begin{equation*} D = \sum_{S(d,n)} r_{s_1}r_{s_2} \ldots r_{s_d} \end{equation*} where $S(d,n)$ is the set of all possible subsets of size $d$ from $n$ objects, and $s_1, s_2, \ldots$ indexes the current selection. So if $n=6$ and $d=2$ we would have the 15 pairs 12, 13, .... 56; for $n=5$ and $d=3$ there would be 10 triples 123, 124, 125, \ldots, 345. The brute force computation of all subsets can take a very long time. Gail et al \cite{Gail81} show simple recursion formulas that speed this up considerably. Let $D(d,n)$ be the denominator with $d$ deaths and $n$ subjects. Then \begin{align} D(d,n) &= r_nD(d-1, n-1) + D(d, n-1) \label{d0}\\ \frac{\partial D(d,n)}{\partial \beta_j} &= \frac{\partial D(d, n-1)}{\partial \beta_j} + r_n \frac{\partial D(d-1, n-1)}{\partial \beta_j} + x_{nj}r_n D(d-1, n-1) \label{d1}\\ \frac{\partial^2D(d,n}{\partial \beta_j \partial \beta_k} &= \frac{\partial^2D(d,n-1)}{\partial \beta_j \partial \beta_k} + r_n\frac{\partial^2D(d-1,n-1)}{\partial \beta_j \partial \beta_k} + x_{nj}r_n\frac{\partial D(d-1, n-1)}{\partial \beta_k} + \nonumber \\ & x_{nk}r_n\frac{\partial D(d-1, n-1)}{\partial \beta_j} + x_{nj}x_{nk}r_n D(d-1, n-1) \label{d2} \end{align} The above recursion is captured in the three routines below. The first calculates $D$. It is called with $d$, $n$, an array that will contain all the values of $D(d,n)$ computed so far, and the the first dimension of the array. The intial condition $D(0,n)=1$ is important to all three routines. <>= double coxd0(int d, int n, double *score, double *dmat, int dmax) { double *dn; if (d==0) return(1.0); dn = dmat + (n-1)*dmax + d -1; /* pointer to dmat[d,n] */ if (*dn ==0) { /* still to be computed */ *dn = score[n-1]* coxd0(d-1, n-1, score, dmat, dmax); if (d>= double coxd1(int d, int n, double *score, double *dmat, double *d1, double *covar, int dmax) { int indx; indx = (n-1)*dmax + d -1; /*index to the current array member d1[d.n]*/ if (d1[indx] ==0) { /* still to be computed */ d1[indx] = score[n-1]* covar[n-1]* coxd0(d-1, n-1, score, dmat, dmax); if (d1) d1[indx] += score[n-1]* coxd1(d-1, n-1, score, dmat, d1, covar, dmax); } return(d1[indx]); } double coxd2(int d, int n, double *score, double *dmat, double *d1j, double *d1k, double *d2, double *covarj, double *covark, int dmax) { int indx; indx = (n-1)*dmax + d -1; /*index to the current array member d1[d,n]*/ if (d2[indx] ==0) { /*still to be computed */ d2[indx] = coxd0(d-1, n-1, score, dmat, dmax)*score[n-1] * covarj[n-1]* covark[n-1]; if (d1) d2[indx] += score[n-1] * ( coxd2(d-1, n-1, score, dmat, d1j, d1k, d2, covarj, covark, dmax) + covarj[n-1] * coxd1(d-1, n-1, score, dmat, d1k, covark, dmax) + covark[n-1] * coxd1(d-1, n-1, score, dmat, d1j, covarj, dmax)); } return(d2[indx]); } @ Now for the main body. Start with the dull part of the code: declarations. I use [[maxiter2]] for the S structure and [[maxiter]] for the variable within it, and etc for the other input arguments. All the input arguments except strata are read-only. The output beta vector starts as a copy of ibeta. <>= #include #include "survS.h" #include "survproto.h" #include <> SEXP coxexact(SEXP maxiter2, SEXP y2, SEXP covar2, SEXP offset2, SEXP strata2, SEXP ibeta, SEXP eps2, SEXP toler2) { int i,j,k; int iter; double **covar, **imat; /*ragged arrays */ double *time, *status; /* input data */ double *offset; int *strata; int sstart; /* starting obs of current strata */ double *score; double *oldbeta; double zbeta; double newlk=0; double temp; int halving; /*are we doing step halving at the moment? */ int nrisk; /* number of subjects in the current risk set */ int dsize, /* memory needed for one coxc0, coxc1, or coxd2 array */ dmemtot, /* amount needed for all arrays */ ndeath; /* number of deaths at the current time point */ double maxdeath; /* max tied deaths within a strata */ double dtime; /* time value under current examiniation */ double *dmem0, **dmem1, *dmem2; /* pointers to memory */ double *dtemp; /* used for zeroing the memory */ double *d1; /* current first derivatives from coxd1 */ double d0; /* global sum from coxc0 */ /* copies of scalar input arguments */ int nused, nvar, maxiter; double eps, toler; /* returned objects */ SEXP imat2, beta2, u2, loglik2; double *beta, *u, *loglik; SEXP rlist, rlistnames; int nprotect; /* number of protect calls I have issued */ <> <> <> <> } @ Setup is ordinary. Grab S objects and assign others. I use \verb!R_alloc! for temporary ones since it is released automatically on return. <>= nused = LENGTH(offset2); nvar = ncols(covar2); maxiter = asInteger(maxiter2); eps = asReal(eps2); /* convergence criteria */ toler = asReal(toler2); /* tolerance for cholesky */ /* ** Set up the ragged array pointer to the X matrix, ** and pointers to time and status */ covar= dmatrix(REAL(covar2), nused, nvar); time = REAL(y2); status = time +nused; strata = INTEGER(PROTECT(duplicate(strata2))); offset = REAL(offset2); /* temporary vectors */ score = (double *) R_alloc(nused+nvar, sizeof(double)); oldbeta = score + nused; /* ** create output variables */ PROTECT(beta2 = duplicate(ibeta)); beta = REAL(beta2); PROTECT(u2 = allocVector(REALSXP, nvar)); u = REAL(u2); PROTECT(imat2 = allocVector(REALSXP, nvar*nvar)); imat = dmatrix(REAL(imat2), nvar, nvar); PROTECT(loglik2 = allocVector(REALSXP, 5)); /* loglik, sctest, flag,maxiter*/ loglik = REAL(loglik2); nprotect = 5; @ The data passed to us has been sorted by strata, and reverse time within strata (longest subject first). The variable [[strata]] will be 1 at the start of each new strata. Separate strata are completely separate computations: time 10 in one strata and time 10 in another are not comingled. Compute the largest product (size of strata)* (max tied deaths in strata) for allocating scratch space. When computing $D$ it is advantageous to create all the intermediate values of $D(d,n)$ in an array since they will be used in the derivative calculation. Likewise, the first derivatives are used in calculating the second. Even more importantly, say we have a large data set. It will be sorted with the longest times first. If there is a death with 30 at risk and another with 40 at risk, the intermediate sums we computed for the n=30 case are part of the computation for n=40. To make this work we need to index our matrices, within any strata, by the maximum number of tied deaths in the strata. We save this in the strata variable: first obs of a new strata has the number of events. And what if a strata had 0 events? We mark it with a 1. Note that the maxdeath variable is floating point. I had someone call this routine with a data set that gives an integer overflow in that situation. We now keep track of this further below and fail with a message. Such a run would take longer than forever to complete even if integer subscripts did not overflow. <>= strata[0] =1; /* in case the parent forgot */ temp = 0; /* temp variable for dsize */ maxdeath =0; j=0; /* start of the strata */ for (i=0; i0) { /* If maxdeath <2 leave the strata alone at it's current value of 1 */ if (maxdeath >1) strata[j] = maxdeath; j = i; if (maxdeath*nrisk > temp) temp = maxdeath*nrisk; } maxdeath =0; /* max tied deaths at any time in this strata */ nrisk=0; ndeath =0; } dtime = time[i]; ndeath =0; /*number tied here */ while (time[i] ==dtime) { nrisk++; ndeath += status[i]; i++; if (i>=nused || strata[i] >0) break; /*tied deaths don't cross strata */ } if (ndeath > maxdeath) maxdeath=ndeath; } if (maxdeath*nrisk > temp) temp = maxdeath*nrisk; if (maxdeath >1) strata[j] = maxdeath; /* Now allocate memory for the scratch arrays Each per-variable slice is of size dsize */ dsize = temp; temp = temp * ((nvar*(nvar+1))/2 + nvar + 1); dmemtot = dsize * ((nvar*(nvar+1))/2 + nvar + 1); if (temp != dmemtot) { /* the subscripts will overflow */ error("(number at risk) * (number tied deaths) is too large"); } dmem0 = (double *) R_alloc(dmemtot, sizeof(double)); /*pointer to memory */ dmem1 = (double **) R_alloc(nvar, sizeof(double*)); dmem1[0] = dmem0 + dsize; /*points to the first derivative memory */ for (i=1; i>= for (i=0; i0) { /* first obs of a new strata */ maxdeath= strata[i]; dtemp = dmem0; for (j=0; j=nused || strata[i] >0) break; } /* We have added up over the death time, now process it */ if (ndeath >0) { /* Add to the loglik */ d0 = coxd0(ndeath, nrisk, score+sstart, dmem0, maxdeath); R_CheckUserInterrupt(); newlk -= log(d0); dmem2 = dmem0 + (nvar+1)*dsize; /*start for the second deriv memory */ for (j=0; j 3) R_CheckUserInterrupt(); u[j] -= d1[j]; for (k=0; k<= j; k++) { /* second derivative*/ temp = coxd2(ndeath, nrisk, score+sstart, dmem0, dmem1[j], dmem1[k], dmem2, covar[j] + sstart, covar[k] + sstart, maxdeath); if (ndeath > 5) R_CheckUserInterrupt(); imat[k][j] += temp/d0 - d1[j]*d1[k]; dmem2 += dsize; } } } } @ Do the first iteration of the solution. The first iteration is different in 3 ways: it is used to set the initial log-likelihood, to compute the score test, and we pay no attention to convergence criteria or diagnositics. (I expect it not to converge in one iteration). <>= /* ** do the initial iteration step */ newlk =0; for (i=0; i> loglik[0] = newlk; /* save the loglik for iteration zero */ loglik[1] = newlk; /* and it is our current best guess */ /* ** update the betas and compute the score test */ for (i=0; i> } /* ** Never, never complain about convergence on the first step. That way, ** if someone has to they can force one iter at a time. */ for (i=0; i>= halving =0 ; /* =1 when in the midst of "step halving" */ for (iter=1; iter<=maxiter; iter++) { newlk =0; for (i=0; i> /* am I done? ** update the betas and test for convergence */ loglik[3] = cholesky2(imat, nvar, toler); if (fabs(1-(loglik[1]/newlk))<= eps && halving==0) { /* all done */ loglik[1] = newlk; <> } if (iter==maxiter) break; /*skip the step halving and etc */ if (newlk < loglik[1]) { /*it is not converging ! */ halving =1; for (i=0; i> @ The common code for finishing. Invert the information matrix, copy it to be symmetric, and put together the output structure. <>= loglik[4] = iter; chinv2(imat, nvar); for (i=1; i…ùy•r°ƒögô‹ÞË"wEîJ®«XéÏô+¿zÕ<Ð>¨Ú1>­æÑ‡v‘ï³ù»¯ã=¸X½^¶[ n‡þ”ò?©ê˜Ö¡WÙIBNîxŽSô½¼þ‰êr˜Bêj^°×¡Æ•óQîiG¿Âö­ùçšïп¬ñY öëÖ 9Õßù+ëe?•ÿÐQÙïó_êˆGŸ´Ÿ·È#?úÄÿ´’+ïoÅÊqz?€½N%‡þòóŸPã+å±/ô/ÏS￈ŸîG^¤Cä¬ú•}ì_ZêÏÄ_îúyB¼u|ðü`¼Î'<اôþ}È/Ôu¼{¥Ÿ+xž°/ëýzt<õs»ÚžÎO<÷ºóÇçìâ=Ï >¿ô:èϯ°ý z‡~ѯ×=™¨ÔõÓùª?Wz-öaÇgOÞ³P©ò>ö¿º0O×ùj¹°q›Ǩú·ÚŸŸí~EçV_/›?›u¹î›½ŸD½´?aö6{?yÙö«õÚÙªç!ª|Ôýv»åï‹þ<}Ÿg»VÞ³VÏBì·&í…xT<*ŠGÅ£âQñ¨xTVÜl!èxq‹Û1c}Ôì¶°'€°#È>м/„ý9àˆíh³ƒªÂáŽT!è¨t‚ŽUÀ±Ž[8‘“9åÈéˆT; ~Ö³çÜÙñ_Ûé Ê}|þ£ìÚ†w¼4ÈÛdÃl„ÉB&hŒ‹-e~aþâý'¬ïEx÷q•‡œ‹McÔè 3Êú»ÎïwQž‹ ]}žÇú¬w=£êu‰בûü3,ó3«)?û=Ÿ?¥œ£;úÍ!óó ¼ÈÎùO)Në)‘•˜Ýß“ޜ؜›s¤zMLŸHº!;/zÿŒÌ%‰+Ö 1J‹؀ΤÄàÑ» ú;d|Næ.ˆ>ÌóÏÑ|à'|—5À:\£yB~®×å­­uA€ŽG4ÞÄä.Í?#õ{j-ÛE'~rÇë5A¾ÂŸ¤ÈäÅöŠ7ž5^'ä‰ñwV|€_I?YLIó"G˜Szñœ›%ò sÊÐSRFî—äZ‰ô 6÷½µ|ÏQ¿-ïYGRîAy_òÖò¾$±f_8ïKbçÉŒ«:ç}É Î{è™òÖò^û;"e[Þ#fì¯-ï'×¼‡Þ°¼GLŸH9,ïŒÎ{Ä((ï ^³ÕoËû’èÃÜ£æ=üzªt¹æ=ÆsÞÃ÷X?×¼G\\óžóÄøkË{ÄÓ5ïKÞÚsŽ˜q¾qÞãYF™ßù=¯òoWݶž¿]tE±$k«»êq™ƒ­Ï5N®ö6âŸË˜(1Úè¼\cenQtE™kT7K~½z]æªêæK€šeoõÿÃðýD~ßB¦ (ÞÊỉzó­Å9®´©ï0VôÖB¯õ»Š:#A_M,›¿B– ÿú@TEsurvival/data/ovarian.rda0000644000175100001440000000115312545056257015216 0ustar hornikusers‹ r‰0âŠàb```b`f’@ÌÀÂ$X84{~YbQfb³0ËÄ|@,åàÛÀ1 t2”_¥?@è] £7@é ºq„n†ªow€ÐPZBwJ@è®zTÝD(½BOº¡'ë@è¨ûJ&@è2¨yePûË!â0ÙCœÁ€NÃ.>±âԸ܉Å>H|‰š{[þÁ!hþ„i»gœp˜{~ÕÙy-^ë–¬8rÊÁS»ÚKÙò ƒ9o‹/£µƒOùû¹›tù|ßf‰Éù—9p[ÈÖî÷pðžvú·èÖŸ„Óšqó¤<ùØãó/)8ø^X.ûlÖŸ¶ÕfUüœm[W–ëàÚÂ;öLŠƒsQHª”Ãn>g€à¬ƒ×!Ö=Œ‹ܤ|/}\ùÇÁ÷tJ”̇ Î:K«*\'v¼ÂÁß¾ûK¤õ<^,—sðU_ð“3{Ü_PO¢aáBH&¤ŽXqBîAŠ7¼é}„ÌGW‡ËÞÿ$g©_=9‡õ³ê¥½]£q>;<ÿ¿•ó3vó³Ûßÿ–ÅðH=ßjçõÜÙá}®›W»‹íßæí@Ÿ÷g1,çÏW{öœ÷›ç‹Nï7å½èòc¨ß¯;}Õo£¿Æ£ÓÞ·z9¯Çéá¾÷ÛËqÊ'9¨q˜mîGçg;½ázÕÏõ3Û÷c$ÖîhܹêOíøx _#ypú(?³ò‹íï'ýýV¿¾Øþú¨Ÿ¡‰Oi\)_zåôBOz~6,G;#=YlÿÜÈqýÌðþH¿ö|ºN=¡ßÉþº}r}sûèök¤×^Óåý˜ éÉc8z?¾ÕŽI¯zrZãët.ìyoÿ¬•§=ê9;¼>¢ßŸwº½~ßú‘ìúÈNøõT¿Óïý³ë.?#»¸>—p€ËQôc¡?“Û;kÏ[ÿFòÛ¹?ªÏúUvøŒ]7ú#nðöfÃçGúzfûò#yá¸>ùúëvrtŸú9z»g†Ï¹žŒäò¬µ»^w?îú™pQÒ7· NÂ3UÞΓýõ~ž_ ïG¹±r $¿8ò×g‡÷G|;3¼žpÛ¨½E(ÆîÛó£~Z“}Hå£Þx¿tt¹‹¸,Йüê·ù}¯ßé?kôÙs#þqîôROê×Âè;3,ïõŒôd6O<¦ñ­vfvßûçí{}”óþ,†Ïõä¡+ß”KçÞ¾÷Óë?cåV>ÕkÇÈßÅö÷]_ÓõØÏÔîÂŽVnD¿÷óÌð|$gìhíŒê ôöìl|îL(ç×½^§{fG§waG¿Îs©^.Ñ{&”KôY;]»¶°rëѳ°ãÙa½QþÎ:¹ïôxùpžø0¢ƒréyëgÒË©vh²_8;ñþÂŽg:×íís¯=ïßÔú­?‘_Þ™ö¼÷ÓŸK×vtz݉_QŽÏ|jz#¬ŸÑÿz½©?gì¾÷ó̧>Oþ¸‡“ßëâŸ3áþÂîûu¯ÿÌöôG¿áõº»ú³–ëâ”Dï;žc?¼tßûãýMíYÿÞc×{õDÏù;:}©>§ÃÚïÚS+×ã]ú¼_éykô<弞p¿‹#½>§3•sú§ž;]V×NœÙþØ•G££ç7ºvvÊÑÞÐnº? ×{ýñú½ÔO£ÛëéÑÝmïl¨Çé?cG¯ßé›…rÞŸ3¡ïŸ]Oò<êG:ZùÑ8…~tý½÷3õ»ÓΧ÷B?»ã•è[X=g†Ç.^LýìÑ?žwí~j7Õ¿^vØËϬœ?çtA¿Žï¦½™áüLxÎÎß}vx¬öÎZù3vÿ̰ܻ½œÝôÙóNÇèèÑå圮@glßÛ³úGãäý™ ]þͶ¿ŸÚ÷~ê÷~y{ÞÏ@_âk¢¯G—Ëej'Ž£÷kÎízߣ眮Ô_£g4¾Î(ÿÖ~¢#é}W¯ŽÑuoßè­öR½NçÌž_„c~¯?ôÃËÇqHõØs£çý¹Nùø| ·W¿ó-Ù¹dŸ#]©=¯ÏËú»öÕé±r±}£·çO¢? Ïè²ö»öÙêïúÝN½#yöö¼_N¯»xºB»‰¾‘Üûu;Žú—è°òÑŸùóvLö¿'©ŸÑO†óˆƒB½£zœ®N===v’çÛ_ŸìïÏ Ëuõ±Ó^O΢~x½SûÃõ]i\¢\½E·; ÇÅöç½ñLòÏ­EŸ÷ÓûátõÊ/¶?8Ëë·ëÉž$þD?åôœ^ïúY?ζ?újíwû;³ò^OèOŸhߨ?]·û]î÷üzìåfv\„ë=ÕOoߟ³çGÏ>ý†—wzzôÙyÏŒp›ß_ Ë%?1Ò®û¹—7:GýïÐûcå»þðL¸ï×Cÿz~«‡Gºþ3ô;ÒµCz¿9”]·cÄ+Þ¿³Ãç’\'›ü^”§D—ÕíÒlxŒãêMôŒÆÙ¯Ÿ>ŸøÜó ]ü`ýŒ~1,×Ã/}Ñ^ûõŰ|’nÿ½_~ßêíùýPÏž}zíí§zvôzR?­Þžõì]¿hߌÞ8®þœ]ïù½._œ~ïgª×ééÐÛÅ+~±ýõ‘Άå{þ|²~y?¼\¢×èŠü å«ÞÐ~×Îy?>§ßúÑ•'§Çéõò½~¦vÏ Ë'}ÉÓÂŽ3+îúè‹þÑêrÚëÙ·.?­ÞoNå®ÎýèÏÏ~êç"ßSûcϯ&ûÏÏØÑÚázÔw§6,çåÓøNµk=ûÓåÕ“ìÔ»ÂñÝ”ÓóïR;ïZØñ¬•ãxfx>ªo1<ÏŸ±ûºGíÍB=F÷¨}îÏìºÓqvûöŽQùpÿÝv}Ä·N?Òø8ñy/gíÚ9³³ò£~„~ÖѯÏ:õy¿fÃû>±¾Þ}/®§qïµ7ê—÷cñ©¯Êy}©Ÿgí˜è·ó‘¾9^ŸÓoíBŸGvÃÛµöâø;}gB;³áõºïýñ~%:­þÑø†ç{v´gFõ…ò=~$º{rí µSíGrâõp\ ˧ú¢Ûõ¢Ëë³ë={ùaíôÎGG¯ßûáçgÂýÙÄsúwvû~uÇÇû3³c¨o4þÞÞÛ=³}ùxê‰t.ìèí{=‹OÝßä¯Rý£úÝ‘_ÖN]·û^>ÉI7ÙÑÛ‹ãsÖÊuè­Þ„»Fts<ûòÚíÙ«È£³ËoÇÏîWç~õ›ú¼];ïá³h×¼¼Ów6~¿Ÿè°þEz¼ŽóÙðzª·k÷¬Üh<­ÉÎE½ðúŒÎ~Œô‡~ìxœ ?•³û]9å"ß¼žPߨž³Û×ÛõW^Ÿ·Û«÷l(Çy¢ßÊS.â"?ús¡ºÚ‹øýl¸ß¡«§o©¿;õ[I/Gý²v£ßñ~žÝþèåcÿR¿ü9£3êƒ_õWyïç,” ýêâºð|¼?³û©³öü™ð¼ß·ö½ T9?·çFt/¶?:©Sýsò‹Ñ/y?ý~ê§ÓŸúaÏyùd{v­›7 Ï%¿2ꇗ÷úgÃó¤·=7¢ÓÚ‹öËë;»ýõ.ÞðþzÓxtõÖêëᣑ=5º»åφòg¶?ïᣈC8÷v:דþ¾;gë³ó*ïtŸ{õ;==;îç]{îýðþùñÌðõ!•súB»Õ_»_×Ó}ï·µ×åƒ{öi²ßïörú". ôwËy;Þô|Ä+Öï_ן[}Õ§W÷¿‰ë:þj»Îù¯öráøMö¼«>ú;±^/ïç©þH7t†òß”î{¹N»Õ»î÷ßeåÒÑûˇzߪÞ«'ÝOí¥vb¿R»VÎË'9c ×å~Ô+×çT>¶Ê9Fôxûö\Ò#Þé‰vÀéíô»×N¯|âë»ìzì‡?oÇQ=ÖN’Ÿ¨wDO #ÙQï׈ÎN¦ò7ÚS¯/• ôôô3ñ»kÿS¿zýðv¬½®µóÔN´Ÿ^‡þ(©¯ÏèKtú¸z»#ú;÷9O~'ÙÞŸrúFÏÙýžŸŒöשÞžÝô[{=ÜÓÃWSùÙÓ·ä'ÉO%Üð®p½gO»öÊÎßeå&ûY(ïÇÐ~âCÂÉo$»×«7ʱÓêöñÌö×Gö¤sLå«}£g¤Þ¯3Û?—â('ö{ñL²»=¹Mz>—uqX¯¯/”ïÉQÄFwÏMÅ3éyÊ÷ìC—nê]OåÞå×Þè·Ï|êó.}¡_Såç]vì>èHõôô¹‡X½Sí¦Óï'zÎ Ë'~Ú·ç¢_·r#zìù·y;¡ÝžüWûN‡ÑÙ“ß_y?Ãýd8çØ³Ÿ ÷vãÀtžè¦¾ÙöÇn|ž÷~öäÔËG<ú•èçùÄ·wÙ±‡[b}vñ„•óö{ñÀÔñOüëÙí(zªÔ»ïåΤwÝÑáõ‡ç¦ú›®½Ø¿˜×°ö§æ³&ã¨ÐŸ®|[}±¿NŸ×cô¼\½éâΞïæC?“žõðÕ»¬|â[Ïôâºâ¢„_¼ß=ÿ2Õν˟³ú{¸4»×Ûó^_´‡¡ùÑÚíÅË1^µòÉN¦x©Çòüˆ./ïíú£¾Úýdg?ïå7»v3Ôýz g²ýí×}o'=ߣ3õ»×ßÉ?÷亇ëzj?”‹þ¸C×ó./èŒü Ïê õLƯ^~âõèçì¹wùýPÞËÅx8ôcj¹nœú‘ìß»fÛ—KÏèÔùÛÏl|GçüzþÜŸ U.”û™íË¿#”¯úý¾]w:ª};¦~%ºÞé玢7Ôëõx}Þ~õËÛëÐ9âCx®Ë«ÿV.ÉI¤#´ãç½£×ïô$yòkýMãå°CçˆÏVÒ›‘\zR}IÒ¸úó#ú;í%9Hrôn¤×Þ~hw$W~ßúŸäp$~Ýûãô†çFö¢COï^ý‘_ޤŸ#9ðúÂs©~çOêgo#]ÖŽ×ïívå/´›ü]Ò£c?ÃýžMöÙÇg$ωî=³í¯÷äpª}Ñmå{r˜Æ½ç¿"ŸfÛ—Kv)ÉM´ûv?kÏžÇzBù¨©ÞÐßä/ãøÎ>õõägRSûé~ovýl8O|Ù»P®×N´·V®ç^Kx¯‡ÿ“]K÷§âìžœwåçÌöçSõ¡g_ãø:ÖnÒ¯Øîl{ú»twú™ìRÏ®$?Òóǽ~÷ìV’û®ü9Ý¡Ý.néô¿‡/R¼Õ‹Ã¢Ý´ç{rùš®{ÿíØÃ“íI¸ßggÛ?Ÿä¢'gŸÙ¾|/ß15Οjç{zŸä²Ÿõôð]‡TOÏ®î4èù¯ÉöÚÚ鯝¡=¿—pzÔ÷pžâ—Wz¿¼>/èŠy«wÇù­Ðÿžÿï[/ï“ꙚÇéâÐPoïzН¦âê©~³çO»øÀë±ö&ó7­ýú‘$—Sý^Ï=ûÝ‹÷z~²‹ÏlOo7OdõNõ·=»ó á¼g'£¼zûÞŸý¡¾.îÆù¡üd\ß©§gíTŸ{þ£›ÏIåfÃëÉN¦8ajœÞÍ/9=¡½É¸-ÐÆ'Ú]{¾×¯Éú°S:ìù8ÞïÐn´+é9+ŸðäNóIïz×{òس_ÝÙi¼•ô¤›? õôòSýÎNãÊ©qrÒ£^~`j¼Ú‹'zv0ÆcÞ^¸Þ³#Sãï.¿í¼ç¯ºq¡Ñ95¯ãüN}ɯ$û25NKø3òÙêëáñT_ÂIž’œôâ‚©qÖd?I=3+ï×í~×Ï„ã;NG g§òÔõ'Ö^]úÕß®}óëÞŸ@ŒO¼þ™]?³ýõ©ü‰åR}vÞ‹û§>×££'̆çSㄈÏì¼ç7“ý×$¿;5ä§g§¢ζ?Ƽ…õcj<íB§_IO#.ýêå5¦Æ‹—¥~'º¼ÿv½›' ýš/tûáýµû=>¥qîù¥ž_îÅy]?êéæï]=ÿ³Ó<ÆÛuüÆ3vÔõ_5³ó3ã—çœzßnÏ£]¯ú¼‰Ï½ÝÊ{»½v¼_#zí¹_Êû}¯ÇùœèKíûs£qýLô|cz~¶ýy¯ßÞþèyëG:&¹JtŒÆÓžKןF×]=:¼Þ¤?IœI"ß]IFòÕ¹ù{æS_OvaÔÏTOx.ò¹ó|”ãÔ?{.õ#é½×H?Ï|jºFõvÚIö0ÖgôE½ ÷Gããý°çzúõ,µÛ£Ï¯‡z“½›*Ÿ£ú=£þZûÑ>x¿½ÉoôüJ´»©Þ@gÏ$>Åv¡4NÉÎöìZêWÒ¯‘Ûó=ûžvx¶ýóÑN¤ûÖn²©|ä—_õôÆ1éM|ÎÛw:Ý#ù²ò½öãøÚõ.î´v"¿¼~£'êQ¯_¡Ý¾éùév®‡G"Î ô&;•üNl/ôgÔÎÄv“>';3â›×oç·X?z¸¹çïzø²gg¦ÊkijáùTO׿žÞïêkègÏ~'»˜ä,Ê¿·fûç£ýòòFw¯|OŽ’üöî÷übÄñNo(—ðÇhÜÎl½ÿ‡~Lõiœ“Õïý°ûцc´Ï¡Ý^¹©|éé;:}á~òÓ“ý Ýþ¡ÓÏnÿÂõ$WÉî'ùë=—øÐËÿ%¼ÝÃ]àåÂy¤{¶ýyO¾{ù»¤‡=ÔÃ}=ÜÜ•{nj¹è§C{=þôðD’Ó^œ²Ó¸º×íTîb~Äé ýœê‡R==ü›òœ)I¸'ÊE ·ïŽüièçd½ô¦qŠ8"Õcç=ýNí'ûq–µÛÍßý½|DÇ&ýëáþhw½½ÐŸ^^"ùÓ.NòûvŒv.ÐÙ³/^.Êw #ò?Ô—èžÊŸnb¶ýs=ý¥×ÚKz”øÑk?â²D_ 'ùƒ(çSï‡ö"?3¶gýJrŸú=Â7ö£^ùsvŒzÔ¡'ÙóÞyâ{²SÝqògõ$ÿýR§_]»žÝ7zºãwfûç{v/ùˆ—þ©8a*z¸2ÚëT ¯Kןè Ï¥zzþ§÷|Œƒ8†þNŽCŒÎ©¸?Ù‘n^!Õkôwí¨õ/ê…’OõK=»›ìÆä¸ÙÊ%¼Ç3ôÇù‘ìsj¯g7§âˆh·íùžœuý¤•ïáÉ©þk¤·©~ï‡'º{õõðKO÷ìmWOB}1ï”î;Ýþ.ŽJý²vzz’è|¹úÕÃ+Sõ¿çç§ö·'_£ñ´z£ å£ýKíX}‰O —øõdo»q¦§ñ‹ñuh¿Ëo?µ“úÓ»žÊ͆ýLþ»‡ßºøÊêIãÕ̓Øõã';ïÅÉ“ý„÷Ãé÷úŒÞ©|ìÙï$§_³Ùžc³ÙßÛþ³ký÷iÿí^ÿ­ÿ֟տ϶ Xÿ­ÿÖë¿õßúïsýï³í«×ë¿õß+ëï³m³Öë¿õßúoý·þ[ÿ­ÿÖë¿õßúoý·þ{¹Ÿí˜zý·þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿÖ?{þ>Û1ðúoý·þ[ÿ­ÿÖë¿õß+áï³ÙÖë¿õßúoý·þ[ÿ­ÿ^IŸmì¶þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿÖë¿ñßg;w¼þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿÖë¿õßϦ¿Ïö¼Àúoý·þ[ÿ­ÿÖë¿õßúoý·þ[ÿ­ÿ&ÿí]Åq{g‡Úqÿ÷ü–ß÷[¾ç÷´ÿÍgËouuׯà?ï´òû~Ó÷|çï¡xUò]ßù›¾ïwüîö¿Ûßñåç¶óÌ–¿Åß\èÈù[íè׿.”›züá8µ|¢ëÓ¥çk;íùýTÏË=÷#í}](ÿç^.zô¤sžwº¾Âî{½‹=i|¼}¯çÓ•žÜOí÷?[_.ÉgO/¿ÖŽ=ýKtöƳ÷œ{õzÿßêIú¹ÓöÒ±Çÿ^?“=éµ;õè|ðë~þuvôú’¼¦þ{ûŒ“×O¹E¨?“þ¥q ý$={žŽ;¥;GO¼|žd/½<÷wêÇ“¥rŽ/zr˜øáýpº¿ÖÊ;=î?{üLõM•¿©íøñåâ¯^{=ûèýx[¨g§8mª^õpCj?ÉûT9zŸãT9Jý†¾ÅÄr½~÷è*‡®‡©žþtñ²ãÝd¦ÊçT{²SyIí½Ü£“ã´—kw?]úÝ~§r=~N=:¾K¸7Ù³Ït<”êqz’êáþ¤GIo~ù}ÇÉÞŸ©×§Æ/ÞþNå±GW*7Õßôô(ÅÙ~|ÇÄò>ž½x¾wìÉqŽ…û±ÇŸÞ¸$ùMö<ÑýrqDʸœx\ÐÃÁSã¶ÅD>}º¸»w=Ýïéc²·Séä˜pNÊ;LÍkí4ŽNöó3å<ŽÝi¿všoÛ©?ðë;SvŠóz87ÙéøÄÛëá‘_½ÜNõÆñizn§~€úñ?½¼Fz>•›jßýþü"´çö¡×_?Ÿz½W_¯)ÎHÏ%<¸S½ŸÚ^ª§ù¼^½/·|Oozþij^ƒóE§Ý„C}^_î©ñf/_éý^„ò=º¦æÅzõ¦ãT|çô÷òS妧ŸÞ?§£wìá´ž=õãNãn§~"µ›ìT/ÖÃ5½¸8éC:NKüúÔ|S/_05nï¿¡óüT»<ç&zw*¯^_ϾM•Ó©zУïåÚ¯©öØïï4®sû˜â«©~®§·½þOÍÏ:¾Lô$>ôä ‡—{ó¸^®wÞÃgŸn<èú½S=š‡Iþ®÷üNíÕ;ÂsSñqϾõðÍÔyY×·žþ¼\=rúœþ—ë·§Æ—©~}jÞ,Ù^Üšü¢ógêñåÚaoo§ùª„O==;4õüåæý^.nLå¦Î/&üæqJj'é{¢gj?R;;½ï妶7µß\ïõ¯—íù“Þú/ϹëûT~&þ¤x{j=SóBé85Ÿ5u=ÎNûŸüîÂÎÓuÎëßi<”ŽNÇÛ¬ž—kGý8Uî?Sý›ÊÇös§ýžêwS¼½S=˜ï%»<Õþ{ž'•OñîT{ìÏõúéÏyü²ÓõSùæv;ÙÓd×{y³tt}Mt&>M•ÿ©|Kø¾×¿©qµ{88ÝOþhÑáÇTýHø¹·žgjÓýæ»àCZ>O•«Ô÷S×ë÷"Ô3•¿)ÎMÇ©y•ÄÚ³T¯ç“vêǦʡû£Ý)ŽÞi¹¾òë½öz×Ï©¼·?µŸSõgªþùu×+÷ËN_òÛi]ægj…ÓÙëç¢s?Å÷SÇ£§'ÿÖ‰×{qqû:;r½—ŸŸŠ#zø'ÓÔù¾â ïwâSÏN.&>Ïõæ={åR^¥çO“u>öì¶·;UNz×{ò=•¾^Äù’êéÕŸèKþàÓ¥{ª|L=&ùÛi~,Å1.'nÏ8_tÚí{óN)¿–ÊûõÅËOµ#NOo«7>)ÎëÑëýÜ©ŸYXù^¼˜pnòSõ!ù³Ex>i×ñM:¦|E¯ÿNwª·'O;µWSùÑ“ƒ4ÔÃ~þƒ•›š¿Ix ÉÇ"”Kþ°×ÿ4?÷_=z{ñß7X¹ÒqjÞ>áúTÇ©ù›Þ±Ç·nÚiþ Ù¡©ò·Óvwº¸÷\/ŽHyˆ$÷SõÂéKv¿×ßžL×§Æ é˜èšŠ?vª— gôäm§¸rÑi'É{ª¿§/ì<Ùõ®—Ÿj§ú·ôüNù›Æ)å5ÿzò<ÕîìÔ¾N]?ÛÃMSqš_w;áø!Ñå~5µï~?Ñ•ôa*=Î^ÿ§Æ‰©—û¾ÃTü;µ¿=\çõ/ìþÔ¸,Õÿ»Ný ·:SÛóv¼;ÕÇwX=SŸó|@ºïýÞ©öä*ùïžôÖ¿'ºz¸'áNïO§®›CÎÞfÏ¥ú’¿Or3u¼_w¹êåGRÿ||SÞ»›ôfª›Ê÷GO=NŸÛéû;ƒý¹t>µþ俦Æ'>¾>žçz}~œZ.Ñ™ä-áØ”t=[tø’èêÅ…SñêNÇÛùØó”_XùwØõÔ¾Ï>MõSõùåÆ)/ÜÃí©^?öèIöȯ÷ò¤®Ç=¾Ó¸±WÎŽ£üùd÷§®ËñþúÑçE¦ÒÍqaÇ®›ìù™ÄŸÞq*^IvÎëIö6áµ~ôrSí^ÏÎ%=ŸšÿKíú1ÍŸ/:Ï9{x³GoNç‡Ó׳céèí.ÂõÔŸ„'¦æ_“ßt?”ìG©/÷˜èîù=ן©v.ų~\tÚß鼬??Õ¾øñåæz×§ž'yô¸§_'ýHýööRù>p½êåéÓ¹Ûý¾I¸¡‡k^™çûxìÿ¦uuSçE>]ÜÓëçâeÖãã05þíÕÓëß"”ëÉcoYÂþ<çSó½qN|Kþgj<°S9Iôz}‰Þ—kwÓñväzŠ{õMÅ Sí½÷7á®TOî„/wj‡ÒÑùºÓø$µß³Û^óm§yèž_íÑ5/B»)þKt&ûäõ;=Sǧǿ©ò—ⱞ_JÏõèéÅmSíùNýÐÔ8ÚëÛi^Ûù¾èÐÑÃÉÞ|¦òB©ÿnÒs~½/ì4˜êÿ¦¾Ÿ’èèÙ³©8?_n>¶ï¼Ü¼Ž×¿°ó$ïo›XÎÛIþÌåej½=|éíÿô~ËÔûýžü'»ìyïô|:N³?Së‡Óx.:ÏïŸû1Å ×ôpoÂ…^ïNóVSõÛå.µÓ“é~1ÑÑó—^®ççzzЋ»vMíoçDï¢ÓÏ—›ŸŠë{|vÚ«w®ïÔo¾Üë½xh§í÷pRÒßÞz’þÜÔ|@²ÿ=û2Õo»?q»‘žK÷“\M§‰Ý?LÅ} ;&|Üë_OOvžä4Õ›®¿ÃŽ©¿‰=»ŸúíxÞåt§ö7§âƒ4^=¹ïù÷}é˜ì@¯Ÿé¹Þy²ƒ oöpÚÔñqú§ÎÛøõE(·ÓxÎéñúzôyýNWš÷ðöÒý^Þ1ÙïÄ—TÎÛqüëõôÖw¤ö¼]Ç™)/ðV;¦xj§úïåR;SŸÛ)^Lz—ì{òŸ)nýtç•“ü${ýró·Sã>——·†ëÎ'§ojþÆëï­œÚß©8hª>§ò=:üzz~j^ë3¿÷ÚIrêMòéãŽ\¿ÍÎ{r“ú¹°óŽçÂŽôÛí³×—âú¿V>•ócâ‹Ûµä÷zõì4O”蟊ó9N]ÇØãkÿîÏÒ}Η%<ºÓø±×nÂë)ÎÊŸ^¹Ow÷ìÅÔ<àT<·S=ïÑÑãK—Ok¦Æ³éù^½é~òï‹N»Sýq¯ïoÏN¥yš^ëõ$}âˆýIô}ºv'ÉmÂÿÉŸ&|°°rÞ~ÂA‰¾©÷§Ú—·4®iœzüñrð£—÷÷zÞfÏMÍÿ§~öü`²ß½¸Æûáõz»©|‡úõE§?©ž^?]’ïá{ç{âÿÂÊs=áŸd·©'ùe×Û©¸Èé™*‡©¾ž|N§DçÔq›ê'½¤ïiýBª§×¯tt;Ý›ÏHå½¼aÏ/õžŸŠçÓñmö|‡½Üõ¶SñùTz’ÿLí¦ë;Í·õìJO©žž_ôz{rŸßaçNG/.ô|eî®êážœøx%ýLã¹SûÐ׉îÞ:™^»©þ~uÿõ鯱½v¼|Â# G/ìúÔøÁùÂy iþ,õÛåÝýêËõ;.GI®ß^á~Òäo{zât¥~stþ'¿˜ða:&=Oí%{–òQ~þ6;¦ò~}ª¾8îèÙÍ^¾¬7ŽŸ.Þä¡ý©óÁnß“ŸKþ 7&¾öú½S{™ìTo=Joœ“]K¸Âíˆ÷cjž ßÂÎS=ÉO%ú¿Ý.$;’ì÷¿Áž£É>x?¡ûáù^|’âÃv>öä2ÅÙ^Î'î4nšz=ñ¥w=ñÕùçòåýòç{83½^\ì|M~ËŸëʼnÞ^üÜóÛþ\êo¦r½c@—LJ=úvºÎÍå!éÏ/ìzÂý©Ÿ½¸1Ñ“êõëé¾ã/çsz®‡‡ÿÝ>-&ö?ÙÏ…•_Ø‘rÉ%\ÈóINz¸ª]I8µ'×;Õ«©xßÛá|aõ¤v’¿NýHçÉ~/¬œãø^¿üzŠG½ý^97Ç SóRÞŽ÷¯×¯tÜiüÞ³Oéèã•âN/÷6;Oz0/õø45>XØs;õ‹~½§×Þ^¯þd¯SüŸÖ³¤þz;=œ3µßiS<œøîô:)NMë{vÕqßNõ£'ßSñH²G½<ÑN×!¤qHõ¸Ÿïña*q»±˜ØNòÓSý¢·ãÇt?µÛó¿é˜ä%ÙÝä—¦Æói<¾ÎŽ=º9B‡£ßŸªg½ûN_Ï.%ÜÔÃo å“:þ)/´°çzxïVÎéÜ)^LÇ^Ü’ðOÂ=}NíOõWSóc©ý$_=\߻ދwi÷B¹^œéýNø=Ùõ4NɾLß“?ï•›zžìŽóÇûåú³°s¯ç­VÎË÷Ú›=דƒ…]÷qKýèñ3á÷‹^óq*îñ~õìíTýMvßéõø+å1zzÒ[ŸôÖù“ø˜ô¯çïœ>¿ÎqÚórŽgzrít¹_íåG“œ%û÷rí³ë‹Ë{ÏϤvzrßó7nG_I¯“¼¿Õî'º’H8)Ù›©ýM÷“^%ÿÒ‹¼ßè¥×׳cÎg/¿è”ëÉ}âOòïþüTyïáZïG¢+•›Š¦êËÔ¸¾‡WRý¼g·vjÞþv=ʼn©ÿ=º{zšêw9íçÔxÍíêNóK £/µ×«Ïér=LõOÅÛÉO8Îx›Ý^_©y?öÚëÙñ$_.ÏSåÖûÓÃUÞ¾ãý¤;åK¯ÿ~L~0Ù=ÇÿÞ©~±GÿÔ¸kê8¤öz|èá´…Ýwº¦ÊyÒç„+=nìùÿä·Üî¹Iö8]§çKÂ^.õ›ë½y¶©ãîãœô1á˜d/RÔ‹3{qÌNíR²oÉþy==â>Ç ÞŸ^?¡ž?íá€^<›Æ9Ñ =<Ô³û”ƒo©ŸnGRèýz‡“Þ&üÙç^\±°ë‰Ïo³óÔÇY=¾öôÞùº°ó„{zvÙÛù»ïòþÖP~aÏy?’òxÑéOüt:zúããÖÃþ|j×éõrN—ëMÒ‹äÏ“ßJõ»?N8;ᾤo=>N—xÿ|œS;Éra×9Oþ-ñi*îíõsj=ÉΧéÉ“—OñÑÂî{ûÜïÅ‘‰Ž©øŸvzøÝëïé]zÞíÇÔ<×NqX²ë 7LÍ›$¶SýZ„v¦Úm×úÝõë?xÜçíôì{OÏ“üsþuç½ÿ½üÀNq¾?—ÚßéÑû;µ?=%›äµ'É^§çÝ_'ùéñ-é‘ëçÔùïžýKãìõ¾ÍÊ':{|]XùÞº©ëCR;NßNó| ‡»>§qq>õò)©ŸÎ—w§#Ùyçg‡¤xÆëOvÔíPÏÏôì¯_ïñkê1ÙãDýZtøã÷{öd§ñÉNã­tÎ1Å1=9ñç]¾¼Þ„ë\ßÜ®zù¤=ýNv²Ç·©ö²‡wÒx{èõ¦cÏ/¤|•Ûõ^üçãÙÃC)>[X9·c©œ÷üŒëWÂwI¾zqvÏIϧÊaÏŸxÿ§âÞtLýééÑbb>šj§Ú¯ä?¦Î£ï4îïá9¯ßϽÝD÷âe¶Ó»Ÿø•üǯ³cjoj|Þó7i~¾ÇÞë÷~÷Æs*nJí'{àÏOÅÅ©½^¼Ì1áÎß®§r‰=»èô$œšìËcÏ/÷òÛî½\ò‡Þžëo*×Ã==;Ü“¿žøâÏ%9ìáèNráüv¿¶÷{tº¼öpú[íùÄÇô\ª·§½x!Ù£Þxûü ?—ìzoöô0=7Õn%9j.š:ï›ìÑÂêOxÖëqºR~¼7íãñ6»ŸèéÑÙ³Ÿiü]{í;ßë »Îq§qÞTÿëv,ñ¥'ÿÞ_ïG¢wa×{yÇ‘=ýœz}j½=>ð|/ÎKò³°çS»‰þÆ]þ|ªê:×Ódÿ{t÷æã’½IþÓ‰ÿŽG’¿Mõ¤vÒói\{þ;µçõ¹½t}w>§ø,ÙÑ…]ß)½©×#/ï~¼—Oðqðþ:žèÉŸ{þØéä<Ù·©þÔå½'ŸŽ¿½Þ>JõºÝL8£7N^>áø ‹™Î_P?z÷Æbîï½{Ví,f«öû¹.~žZµ·8¡öàÃAñeß²õw_?ýŒø¾ku¾¸gŗŬÚ=ýÉÕs§ŸVÿž}ÏÞ%ï—œ~|ÅÏÅ.õïyÑsHçWÿä—ï>ú«þõbŸÆíiññ øöœÊ=%ºgwÙòͧ?):v«ÿů¹Ú;¨#ýÚи~Rý‡Dÿ•Wý‹½ªw¿ê}­î¿fu^ãù Ý?¢zž[ÑyúŽê9ºâÃbcU>žþ„øÌ8ÏÉrß©gÏÏ¿û[ÐõýŒÿŠŒ_ññ€ê{NÇ—ÄÇÈ÷wÝý¾³R¾Ó/ª™®k|áGéÝ®§¯ë9Æç¨ø:SûoÖ9òyHüyAGñqqJý=Èó¢ãÉã|Tzó,üP?>¡ó½+z'$OÒ›Ó—$G·¤7Œï^ÉÓ³êï¥ÕýÓÄOúyQô!7$3ñEÏ/vk²{‹ÝªGvîôñí*ö\+¾=a~97= ò/Èž=§qÆþß½Ð_UnqDýc<ÀQȵŽèø?>¿õ×ÿùWÿ¶½Õ?ìãUéÙ†Ÿ®ßø ¿õwÿÓ/*\Q|—<ì½Â-‹×‰>ü©pïé{¿ÿþO>õ/Eñ¿¾Fã =7e·žU¹§ðŸâô+N(ÜŽ(<‰¿?ÁØIÙ‰Ó7u¼¬qD?Ð áupYÑ-{Sþ¹ã¹#²²SÄ!…ë/áƒÂÂÛÄ3ÈíŽÐó“‹}²›5>7Ä'ä¿öŽåÕoÅ5>×V‚RvèˆäK~h‹?ò3ÅOèû$õ˾ïÓ¸àV{ø-ô:ÀM—Õ¿þõ¿ðô®¸F~ùô“Ô Žï˜?Ç¿*)½*þ+ÿ‹BŸOˆ¯OwˆŽ'ÄçO"Çð]ô_ÿÿþ¹÷<÷Så-öaW5ÞÏ?€?$/ðãÙa»‹™äŽçoÈîÎE?váüÛøÞÿáϾ«èÁî”ßP¿…ÛG8t;£ókô>ËïŽn,\€Ý?®ñ#y`if6ÿ}µ/{Sxÿ¨è)¿*>cèvãæÙ»/<ˆ¾bÿ+^E¯¤'Š?K¿h‡¸ ;„þÌE×ãàTpµÚ™k¼S¯êþ–¡ßÃüHšϒ§À_á'tœ‚žƒ7ˆ·OüÖ}êï㪼¤ø¶p8ü¥_´ïvõ˜Æ{b|ªx–¸¼®¸wqrˆ³Ê®î"^"@^ÁÔ/þß'œ>G>Šø ù×óø¯Ç Oü“\/Nª?Ø ü2ùœWK/Àר“gìÐ øüŒžßøCÕäø~ÆGü8H\C¼A>ºÑp6y ìíãbäo+0þDÞ$—ÇE'x y»ºLÃýÍ[úËxÿ¡%àý埰÷ž? .Ô¸/6À‡ä;ÄoprÿÆ/(/Pù9úœÜºô=úëÿû?ZrEü Ý-~‰èGþÁ»çÉ[—ŠŸD·ìUÕ3S±GØ‹sò³Èíã²ûw4¯’üœ#ÿ!:/ÿÆw~à‰¼SyXâ)òOø]â4ìïqìûÐ.–þ[ì–¿ÚM=êçÇwµ³{‡”~aßÈkÝ&¯Až…ü‹ä¿FþVyÂÅQñüx\þx1çyÑé'zõŸüòÝ¥¿øügå“É_è:xæú o,ŽË>—îÀò:bÀǧÄÏã’Oü¾â°ÓçådwÉ¿È[0þ¯ÑuäœñþLúL¾”< úârwí=Ï-9T~=ÜÓ.zTù´/%¾RÿY\QùIp¸Š|ìRáÑsci-~ß•^ÞÐ<ÁQów•w&¿(zna/,¿Ê¸"ð[ñz_ñɦO“ØÑÊï ¯ø#ò¡Ä…²S/’G>·¥]âð0ò.}Ùn½ÈË3æ÷®“ß%¨öÈ÷ƒg°·GtýÃØ}òÌ*‡Ã`÷Á¯3p)öMüø"ì£ä¨òä7$'àŒËÄaªg¿Æ{Büü(x¼8ˆ<ñ(~E|/\#~¿ˆß×9vsÿ0ÏXyròV%ŸðWý,\@>.A܆>‡!n’<]ØSòä/*ŸÿªÞ ð¿\ÕñÕ_Zøˆ8ÿNpa¹¼ãëÿRåíð+•A‡þ¾pöm“wüþŠxþQzÉwà׉SÁ!äñ°ÿÄÄÐ ®Py;ú¿øO7¸¹´ÊÃÞÊ=yEüaá?ò¸Ì·íÚ믧ñ…ü‘Í x‰üã!»ŠüÐþùU¼ZvyÀÞÝø£Ë<_@¼Yóªòµ†ù^äïâj>m±ùGüñ—ê?#3Ζg…ÿèëùÖÈÛäU¾ìüòwk^~9ÝñËÿNùͪ—Ó~÷¿¸5L|Bþ¾2îùHïÇî12Ä)[q·Æ¸¼…ýdðŸ•/”ÿ:¦sü™üôé[ô[ô¡oÅæAÀ»øCò¥k²_Èómâ³OÄ¥ðãqˆê¹ >×|—úÏ|~>Òoä¿ô–<ñúNž‡8Wí^_gÑ{Eõ*.­üÆuøcúÎýñÔé[¿»°íô5˵åò“Û9}Y|$_Š^2OG¾þ㪇ñ,>H>èòžÁŽb?°ÇÈ“÷ýèyôê%óO —¿dž ?PxÐür„Üs^òeò_ùipÐp´ðß#Œ7ñqŸúünåmÀªýfþ üÏ: ê©õ6:g½ó)É0ÿJþ ¹ùëáé?tã/TüYã/»‚?…?è)öúÁYÌw㟰·àMòjØ ì?|Å~3îÈ+xç¦â3è`~?Jž½¸L\ŒœZþLù£ŠË¯ï Çg‹?äDv{BžùªuŽ'Åoo_ß!Ø}äñ¾<“øúÅÌ €'…ÿ_E^ph+¿ Ÿ‰C˜—"C¾:Ègúú4ôÿŽé!rpK|Þ§öÁàDì=þÜ~a}ó¼ð“öXÇC|…>`+N½×ï†e¿ºü9~» ]5äQÕö›ñ_#ßð}­uà?éÃEärߔݡ¯øßZ‡|ÿ î–]º|×¼ñéëÊë¯1~ÌË“' ¾b¼áç×çË ÿñéGwâ0øÃ|Ñ0®.¿=Á?€çËž2ßFÞü%x ûcódõ|é›ââ×Xž!/¾Ào“ÇÐ8{^ê†ââsòêàsüÿå»ÓV¿Ëߘÿyù MÞKr‰^£ÿøpžøsEëaÐCðylÆ}¬qaœÈç2ŸH¼x–yùs®€~×±T^yŽ'MNÉ'°n üÇèN)}ÁÎc¿_møçë’ðv¥ÖE qËÖºSɃðX]Çn0ßC<†=¸õ‡î©ÂO/Ÿn-³Ioþèé-®½µ ûÿú?¯üõ_øÈ›Z„òËN_XÍsÔºsð+ãM?Jþ{ËÛÕ¼¼Æ‰õ†5Îè¡pEÙG_ßú˜Î±ßµ¾Av™ø:À¥µn»&9¿E†v~q¹Ç¿ ç¿Jï+ïHžGýcÝéaÑI>B¾U×+žîc¾¿„þç ´>£§µnx8ÏVrþ¥]p†÷?ôÑsˉ·×Ô8£Œ[ùOñûQÉÉcÈ…ú‰¼Qï•UÜ\v¸øûª÷;j|‰ÿ±ÇøðÛõ¥Y{ç}#¿‚^÷bÞþúj~¿èc^¾æÓ…f’³šOGΘwÂJ ?ŽÜz^ˆ¸ÿ >Á^íÅNj>}øIÖE¨~æÙ+Ñ9yWÆ=eü‰—7_'ýW?ˆÈ‹Õ<ˆä¥æÛ,ÎÞO^^ò~ÄßcÇ.󾵎Œüù äø<‚¼_Ò{TÈváÕ ŽANY·Š>Ý\­û¨ø²æ#ÁÃà"éµÞ›©¸|ˆ£ÁÿûÈ磟’#ø,»]ëʼnG¡“þ?jü~ÆÆ¹ÇaonÚ:ðÞå»ðÃåÏoèyáfôú[òØQâ ³;'`·<ߪv±cÈû'‡~¥Þ{ÒøÏ¼é.âDÑA>~¿ù«…ƒà÷ùžª÷Vsò ÅÑI>uħÈG ç;¶äWô#÷ußA‡úÏ®+ý¤ý=ÃùÂZ¿U×^Ÿa?ÈSÛ¸@78 »Ê8ÒOÅ1•'aþOó…s…WkÝ+ù.ì8ùlä {qó+ï&àË_•¾ëùZ§ñ½øï.P¿ÏH\„=ÃH¾?¬<5ë*îQ½Ì£¢G—•çÇ/ÞwwyÊ—W\"ÆüGõ¼VëHVëGë=2ò|µ^V¸Ü¿Ž&/…^©^ü x‚q¬<²ô‚|*ó¡¬/À. ï5£ñ¾®3ó8ÈÏ%åseÏK^ÀC…/T?~Šùôàªî³>|VëÈ««?äcnçIΫ]=Gœð¤ù)W|¨|MÙCæåÈ7í'Ã7å?ð¸ýÙnÜ‹\ó“5.ä‹ê½AäVãÀûÍä}àvî¼ä»E¿É·C §¨äh·òÂÂ+MNkAòIþ±òn:¿%ûWñÚaG½Ëú&ì¯üôqäQô“Ï=&úÐç»Öåè7–\bï?ÖºìÏ÷ÔúBѾ’×ÔyÙ3Ö3@ççª!ÿ<Ïx€k÷Å?òÄÄñ´˜Ïاóó«ùâß­¸ž<¼ž2Îø/pv¤ÞÕóÄ!à8ô }®yCµ[yWò.øfñzôù^ð¶ê%_†_|Àçc4.׆ý,ÿÃsØwâoÆ•~"/¬7Â^Ø{“Èõâ ð‰øVv}CvãˆÇUÄÿÈ ùJâGµwqõ¾H½ÿ*{\ó†ð;H>è…a}‹CàDæ÷ýƒ­õ*æ‰>~Sÿ.½þ®ƒ©õøéÚßAÏÁü-qvƒø€u Š‹*&ÞÀž~|§v˜O`= óœøypéA•»ô/ùßþñ³¬øžeÿ è†5¿Dþn(Ç¿2ŸÄ<¨½wZ¸ ¿s‚<8™|8¤ö‡—þ߯öuJÈÕƒ=×sû,Tñ1vÀâö3!ßÄ8’‡'‰}©¸]ãYq“îc¯™#¯ùù4ÙaÞ3¬ùp‰ÚG®ô½ò àbâ5òmè1x•|;xx¸ž£æ­‡« ì5×9ZoÎÃ^U¼­ê;…?`þw˜Ol½÷μ³úQùñ÷Û‘+úý¯ý Æ·¼O{e¹Lë“Oo½oK^<ŽðëÜ‘ó¼áî[½¯ ?à7óXäÈ£]Ó~è ãÇ|]½¿O>^tâßÐWÏÿ€§‰ãàÏqÙ%ì3ò~VòD=Ä›àmìåýµ¯î“d y%·õ^'¸Ñâ[ôøaÍ‚ˆÏÀ—VóúÅ¿Š¯ÀäYî.Ëû­5®_ ¼<ö{rÔø$~„oϘ]Ak½±øŸ‚îûñƒÇšOÇÏ’·3ûêóàÈïǑϧT¾Vò¤<,ë¸k=Êò ä½Ì1~µ¯‹äýÕz~æÌo~ÒìvƒõÜ·÷µ«õ>›ô|[÷Ÿˆƒ‡Š‹t2…%?O?+/¬óZ/ öÁS¼/K9ì òÆ8ÔºKõù¾Ú÷ˆ¸nˆ·ü®áá'\ž™Ï"þ°üþ ¼I^›søA°—×yž8z‰§‰÷,nÕ{b%O5?l^zKþoÔ{£CZø¯Ö¥›Õ~5ä¥5¯x»ÞWŠnp óÓZG\ùyìôÖxëù«Cy-}¨õN«xîôáÙÚ‡‚yò#j‡ç+?#¾}ˆüƒpq ë[n-áý7}|qXú®yÍÒÛ;f¿Î¡Ä36_‹]*»€ÿ!?,z‹Ä«’“Êû*]‰nŸ?d¼Áø‡G˜' /à÷ A¯­Çßšç“ü#Ä¡Ÿ0ù®sÄãÚ ì¸Îó”è_å%÷×Vãvúšîß´8zžWyâ^æ…ÙÿŒüóö ›çe\éôñÞÇ>ñCû Uþ¼Úý”—üœ×>|Ð ÞcýóÜ´K¿Ð#òˆ¼Ÿ_ûGÊmÅïìW†ý'nÇ®]?ÐKò"Ø ø‰Ç¯|@û R/v¿ Þ&yTúDœÈú™Ú¯FúX8UåkŸ+ñu>¬S¹=Ô‹z¯†}w‰W‰—5UëàðØâq=‡ß«qAn‰¿ôÜ5åÑKžEßµ»Ë¾ºìxÿÁ<á³ê7ö‡øy*ygžüžáB(÷>âåõxÏÜ\ù›/Fo/`ï5~%?²çsâ6µWûhˆ~öÂ!O“'ÇNbT_íó NÑõkÂÕWõ¾õ~³Ç†w3µ[û"?«‰ÌÑû Èþšx òoØM›¯ªù_½¯Èº{ík4Þç\|þßdÏj¿.~´äUrüÏÈ “'˜½#ÍûÔä—ÀýøÃ+Ë4κYvù¼Öo–^«>øRïÓ‘gZ­SÞZ×C<Ê<^°SµóCàMõ_ü ã„Â×Ì—2N†õÕ{ÕŒ;å°ƒç„çj_>ð*ùo­[Áòü Óì ÷+ÿ@^OúŽª¼»Ú¥?–çqý(¹«÷…‰ˆKÀ)äcˆ·l\®©ßìóIý¬oe¾È8P/øì¨Êã_ÀÝØ Þów׺'›ÿ€ŒtÓÿŸÒºSìë^Oü,¹ÁŸ`¿ñÓŒËýz¦ì»ÞŸüÉÇÃä±Ïè©Í×pyº¬y\ô?Åx^_¾ø·ÿcÙsô¹öy/˜|QŽ<ãDܾ¤KÂ[Øè$ÿôaòX¢{YëÛEÏý+zk_-Æ‹vj}¯è¨üåó0¼mûén½§!;RöY×ác­Ï$_iy;øE>=­úÈáGõ<ñŸáÊÒ« Z†F/ê}âò±àò{4Ÿ‚?Çßh<Àᢪç^ò€«ïUTÿï˜\VüªúÊßWáÇÕ.yzò›äÀÔ{KxÿQÑñÀR ~üß¾²uó£õyµ?ªžç¾ò_5/Ux ÿgù”‡É_é}è„.ìu­#ÿjŸKõÿ„<]Xº½;¸Çìë«7ùÍÓ?¬vXï޳vj_–ÚïVçÈ?ôÀ×òï6ÏzcÕÏÓ·–oG4×_yŽ¡-=ÄÎûþ"éý¸‰Îë¦7µ/3ñ–ùiò4·ÌUü) >e½t|!zwê«u¾6OXûŠK_M^Ÿ|òF~ ?‚½Òõ‹Ú¯RóÖ£ý‡‘÷ûÑ á¬êŸôý¹¢ùµg‡z°Ø«ñtûIó’¼WßkÝùuò²ä7UîºÙIí_Xq óå7ïº_\öñ­×¯}m¬ü ù2ò1Œ÷h=‰å¡}ž¯ì"íIîÙ;Qï‰OeÇ‘+ó1ÓKâUæáWáá æÕ*ÏÂ< ~ŠvÈŸ3o ~Á—‡È÷¢'Ê+‘ÇÝ\ÍÛÔ¼õAò‰ÒcíO]òGN|À~žSžLû'ÿ¯,W]müŽÓ÷YÞ ûYß=P¿°³•ï%Ÿ£vÁŵþB|b]v­¾WÀ|ygñ‰yðZ!:>N\§zÀ;Œ7ûüqΑ•OÓ>q;übZâfìEíŸdùìãùCö½ªùdü„ø$\¶8)>‚ûÐ/ì¸òpóè«èû(yò\àkñó~â9ö ÿ‡àù›ÄÅ*=EÿȻǸr˜çã{[xWýe]v »}Uù'òL—´Î|[vÜ ù]·×èxý´ï@láoê}ÁÚ/IõR8‰~N!>Tÿ‰W˜ç{3ýdÞ Üiò!|U~Èå¦ð™òfìcYû¿’/±8êQüˆäôcß¾›–ù’’³óœƒKɈ®GÉ çN_W½ÔWóöºÎ<\í-~Àß÷iÝ~ˆõ!¾_vüÆzzâYê%߀ÿ¬ý°ß¬{E/•G^Ð'ä±¾S¡ûµ_ø“¸FyÀŸ#:jßBæwô\­WZí+XßE#Ê>,¬—‘'/¿iùaðqh­7¡Ÿòwð½ò 6{Aû€Ô>”àõ£ò´ê?ïOƒc¯Ýmæ++oÁ¼:û.àç˜ÏÄn}ˆyAÝ?ÏøëºÞ+[Ücó9Œ z\ù Ùƒún"qšêgÝ­¯/ºeò‹~=‹þ©~ÞÃ< n¯«Šg-Ž@ßYÇÉ~ÇäGÀqØûDoÍÞ5›_UëðGä+±çø³zO]r}¿ÕûÁ•ÇÚõš·$Î<Íá!Nª}1*¾c?X§Džÿ:ô[ë]äç7<¯þÀ’"že[ãø q ôê¸Í{ÑÏÝŒŸèøØµþ¿òª¿ÞÇ£?*ÎCNk¿Æ[÷k_É=xûRq¡Ž'(ž«}yTyƒŽÉÓX>ñêÝ0ü7×<À~@Ø¡ò?*O~šï%C÷yÙÕZ¯€ý4ûŒ\ù>Æ7Vã¶ø9'ìÁ5ò_Äi:²/Œ¯+a=ï—ð~3Ï_ÎGÓþæ>ůŒû¶ì'b°[æsˆóòô³ì(yÙOòÄ䉰/ï×ø0„¾çü;TàNÆÿPó#àÝÿ)Õ{c¹Lå‡ÿZù­Â¢;Âz„ÂáàâAå­ÁiÏ¡Oà •ÇÕ>R¦‡ÄǵŸ+vP÷kßê…¯ð_ãöåêzø’Éò î¯^×<í׺ Ëc?ÈWÞ}!Þ€>⃊×uŽÿÏ_’\ÖwÝ ¯=¦¡ò7\_=·µþWôÜÒüwíÁ<’ö9.=¥á:“­}åtÄÿ”Ý“‚Ëm?¬ZÿTø–qÖ¸½_û«ò]®z/Žù=}#ïôØ;ïô5žä¡ñ‡{ O· Ø“Z·"~g±{÷¯ ýJʼnŒß“ÎG‹¯£ÃîÁ×'°{–©ïqP½ÌŸÕz/á.âòµ§èD/Ÿ0{‚½gÀíÌ× ßO›ÝÃÞÝú‡Š«„_ —Îçœ~Tq|áÈ8°î—÷¨jŸ ôVõ#/…ëD/vÎ×½ßÐþôÈÃ?S¿±cµ€ô¿Áø:î$nÂϱ¯=ö¨ä„¼¹ánøŒÿgü_0{ îFÞÑß%lxrÄþÝÇkòÇØWÿ]½O¢zÈC+nÚ›Œ«è!Ït»€è(ùª8?ɯã‡évìóGj‡õ(…ûÔN­wêËÖ~5–¯ª¼ðÏ9á™À?숞Ü0}¨ui¢Kï§—~Õzfp«úWßù^åõk=~åEe/+нDnjŸDâê!ÞÊÛ¢oäÁ7Äw¯Àô\çïëÜÔ~9Îoªæ ÕïÚçKåÁ}Œ3ó¦Ï3/#¹ò¸©Þ6¾·³î•ù껬8ù­yJÓ®ÃÏËÈ ûÛ«<ö½~fh§ˆ?ê{‹øSè‚k.ó•|?mK?‰×åÿÊŸ¨¿…ïÄßÚÿYëDð+Ø?Öy_Ô¼ òrAóåÇÄp¸„ñ-y#E|k8Ü×ëÔ|3ùæ!m>¸äDóú>oðÂ0ï]táßÐkòwè_ù5ðçæË®*.îØúžÊ?¢<.ü©÷]ÁõÚW†ý½÷ûÄ'Ö\×:vü0õ÷]½ /¿»Æ¼ü…/ÈïO³¿MÙOå]ùß1ì”áâô‘|&8Œï±^âºp'z]ëÂîn'²Qýf?çŠl~}){ žÔ>ØÄ—5¿\øOrSïÝ©¿Äî§ù®3öynâ 䋼tÅyàGò­Ÿ‹fgˆ°·ø{­(ý'Π\}·EíàO. å·â ð&ë™á_ÅÒÇÒókw§~}ÙiôžýÁsØÛZ· Æ£§zß|Kž½àºÎù~$ø€ýÅ%.¥}üí*þ*=$NÞgþî>ã#ë<ðsìk€ŸÀ~ñ€¯3øy-éó}õ});ïSÔ>ÒÌûH~‘/¾‹NTë±çÌ[þý¸„\#W¢y©v¤—Ðó®Ùÿ …ðs•Ãn±žEz„`¯‹_jÿ²ÖÏb‰ÿñ·4„žëû·õÞóû—ÚóŸØÚÇLý({€ŸÕó¬'¯ïk½ß+¹%ûèøE|Ǽöé•\ð}âÿ{WqU剛ȟn–|ÂwøZû°®ÖuÔ~¤7´Ž¶öÓýØ/pó…ÿ°wäA”_ñç™O ¾¼®ñ®ïÞ©¬Ï¬ïæǨùÁ7‹ŸØaöãdß¶zÏù!´yèŸÔuµGÞšõäÁßgó›õžë†ôž›í[qòÃ*w¿ðÁ5}_Íß'†^Þc'ÏD~8º™Ç"ß_Ù¯ù¿µÊ—=Âp2t`ü»Hø!Æ<н¹¢õä{oiÿ±š'!ï§öÎËo_}âîÆÇWCž—ùžÕ¾p[ë14>Zç\ï' Ïà/âûÞæÖ¼„ú‰ß«ø[õ°îæ#WŒ#ý0o†žâç/®>41š+\Jü…½¡Ÿâùª½óÊÏ¢?Ê#Öþ’ØÝ²¿æÑkÆ9"ôí|s•~(ÿ&ûTßÍ$þeÿsâ¾CZñyYòÄ“âï)ÏD\Å}ðãêß«¤~äL8·öú¤ÙAÖw0/@\ŒPx™ñƯ\Ô>:Œ÷ æÛ”?`>”ù7µSv­Ö±ª?¬»`”<3x\…}®ý 4ßļ z¯÷··¾§jö¯ö·—^}@vþœülí?&ûô¯ÉëÚ¼ÕýànìˆèÅ/`—‰_×ëäÁ%äyÀAõ]uõ«Þôüp½ qÆž_±øìaåùÎËßs˜wb\Êÿã´nÄ¿w}©ïyþô5½/®#ÈwïÁŸÌÔº}ü ú vž‡oØcä¬òâËuòÁÌ—ÈÞV/9@^nÊþ\^š7Ý[ü¢^ߟµæ¡ÁIâíÕú:â¦Õþä5þç×]3þ‚ŸÂ§gÚC?+/)zÀÿ¾þƒ¸†üí“¢z±÷ôSû”WÜŸ­ý:m½Á%éûI±ºûKö‡*;¥~¢”'^¤è•üaÍ;o=&þ~PåÀ»©=ô ú‰·kÿN‹ïùGó :²ßg}׎¸IyNåíj=+ãŽý†?µÿ¾Žà úA{~§òD·ïbŸøî,ùÚCNÙÇþŠì4~ yôùŠš÷÷èˆÝä;"̃£÷ÈeíëB^tGÈÃ}Hv¢ÖQ3Ϥú®ËnÑîMáŸzIt£Oø¾[Â|¸“<=òž‘Wû¨Ùmä“÷›jLôü¾R=ðý‚¾g½ª|‘äóãä ”'Ežë=oìŽäáÙ¡ÿÛÚ7ŒyË"÷÷Ùüýû¾SÂ8U^RíÞÔz–Úÿ»¶ò#µn¬Ö!šÄîß2ù§0…ýù òŸÄqÚoð~å½ë=tåUj^uoª‡þVHÏ}Dùy¯§öcAÎ-ÿRßWà|ÿïÝ2^çåçÙ¿™÷Kê;ÊârÌû)ð™õQõZü°è½¾z¿¯î3¼o®Ð÷¾¶òž¬ÏÿÐ+ùóʻ݅žû´OóÒàâ¶Cœ]ï£Ò?ìü¬üŸž«ý'™w!Ÿ¯yP®×þpj÷ƒŠkŸyåOdj]_­“/²nCv|(y¤Ç$geOÁÙòƒà]ú^EÁWì'Ï÷ùˆ?*o§qB_À“È øýAò.«ï®–cOÿ½é)|&OÀ÷=àý½)<Ï<ÕÙ#ü剻ôÓ²ÏØ]ä­äAx®öåâyp¹Åñ¼ïD¿”'¯ý.Ó< ¸¨¾·)>‚‹ Ÿk®ð}ñóQ½7ë뢡8è¦Öu‡­ýUˆ÷ ?’Ÿ»¬q¯}¿VýXœÂÏ7lž´¾Ç£ù]ðñj­'d^–yÓ;äÈ{NèÇzÏ_ß‘ºWý.\ªöyϘ|%ùXò Ø=Ö«°¾œ÷~Î?Ë;Ôº_×fë®°ß>O…}d<‘ûòº^ûqaµ¯ öÏÞÚÚÃò¤ì3cœê=•c½)ë&÷k|4üZý“üa_ê}lì™øñàï§/h‹yìÀ9ñ•ù4ö¹ù ù³úþ¥ê'­ùÑ.OàW|ŸòB¶ßßÖ¾ ¢ \ƒ_ñü=ü»¡ýJë=PÆÙÖcÕw8u$Ÿðaå¿Á×ì»B<ú%Ø äLíã_®óÞ€ú½5K&>ÔþäMô]‡=Ì“ oÚoŠy äûUëéÕÍwWþ€ëøkè¯}˜Å_òwµ…åð.úÇ<rH~¯æGÚ{h[ø†| Žð»ßÃUþKrPû’(ßPóð>ŸB>^ãt¼or ø¯zNzsAãQv^ü(Õß‹Šÿ®2>gÉŽð>íá«ßÂñøÝšoZN§ýÃÇӺѲƒÈ3~çý»ˆë‡ñTí›\ysÝÿ°âúú.…åÅ+Ÿ¨ù_ÀwÈé/ãQëžð:¿bòÄ|q ñ‡ä°æ¡ð3>ÿWûÁˆŽÃî‡É+ÏÆ8±ŸƒïÏxÎè¬÷µ´ù»Õþƒšo¬ï:¨öÑï’Ï©÷ɃiÞŽýY¯ãûª£Ÿ5Ïza8Î5¬ŸðýÑjþ ½ÐuúuEò‡¿yÌø]ß±Ôyí¿I¼#zè'8†÷ê;ÄC}^=¬©õHâÇùÅÊ+“ÿ°ï¾<¬yúI¾ƒq¾¨ùƒûõ~4x’x‡ùfü-~˜ýäàw­ãQ½•/§? _ªv.k~òœò£ä[$Çõþû/áoÉÏÔºk˃b瑯«ò—à>ŽÈãÂú[¾S´Çì8þÑòÆ…¯lÝz/÷Âýý_Úcž;θ¥øþÖúù¡~Õ|æEÛ‡¯Ö©Ýû˜¿gŸ,ÖOh__'ZzB¾vµaúǸÕ~G´/:yOºÞ£%^6?Uïó’תþÚwñöxŠü0y ò®6>µŸœøVy`±3à©zo‚8Uv‚uu¼Áúýš¯&Þ•z«øƒ¿Á¾×÷塟÷êÜoK¯Õ¿úî#ññúÂüqƒèÇb·|ÿfôŠy<Þkw}oãïÑ?ì]}/Oñ?þØ×¡€ßÑ·Êsª_´Ë<ßé½!ûˆ~s¿öuÕx?oô“?cÝr­g²ùÍïÕzô Ž¿}TãVû¤1¿Ç¼ù6æ“Dr^òŠ¿â÷Z—vMúü„éyí»îN%.¸Îü€òn¼—Ëú%ÖÖ<„æÀàüáÄWääšÆ~×:Pò™²{ìóQû1ˆ~ü*zÈ|#û6bw±{5O º®Ê¿Õþ¢º.y(ûòáUÞª¾kWß'ï§qF?ÐÃzÎæÛþwå‹X?À~oÌÇ×øb/‘wüo퇨úø.óžåß,Róš¿@oÉs“×'¦÷ý«^òPÄgõ-=I~¨¾£&|Êw_Ö—€{ÈÕþ€ÃyÖÊó×{—Vºˆ»ê» Ìó)Lþ³ô›ü‚Æ ;ÿ€ø‰þQþƒZ_ ŽÃî(žªþ==¬ÇJ¯á3yTð3òG| ~dÿôä!•WÆ~]Ò{&Ä[Ø×ÚïSô0ŸO¡ü†ÚÅa—Ñ›'¬>OC¹+ûZó5ÿ€ŸÃÞcG°»–!ßvCó øô|z ù®ÅIâ%ÉÖ{׆OàøÅþŒeßÉc(Þ§_ ¿]‡~ñƒxñió+¾¾ƒõúØÉZ—!ú>°¾‹·õ}b³3ìÿE>?‰Ÿ¡~­Xœ1oFÞNë%™w=Ç>[â7þ¾ÞkÊKÉ“ÏsÕü•æ3êû¤ò÷ð‡q²yÛ-Ü«çj?p­¨ïô¨~òyÄ»µï¸Ÿü”úCž„ïù¯}½Vî“~²N™¼xžuµåoÕü6rGûŒg­'×9뎈'˾’_ZùõŠÉ—×z+µÿöñÛ|ÎÄûÌÛ¨<ë`°7<ÿ¢áÞ{º—ù›7¨ï^3~̇¨<òÍzÛ}jçºÞA^°+—[¾¶ò`äà«pþO w0G¹Vœ¼=jóêø/ì6r|CûÛÚ÷^ª\}GAí—‡Ô¾óÄ]† ·kœ°çØ â-_ßÁ|1ßç¨ï0o¢òèßû¤wà.æ_¡úýûÞàUââ/òoWµ. 9÷Ôºùñú^x›u¦'ðéŠáæë8b÷èï»b÷\µ~wëû˜Ã¸§öïÂ~×~9Ä·Ìëä» ´ƒ},~’åy/ñ¨æy Ö¾hÄËâ'ùÞÚ߃üŸêá9åÊðž$õKïù¾Ná­úÞœâFÊû~Úà*ßÇJïýnÀ~W?…7}£Cy(ûF^ªèQ}äÇ?ìD}' ¹~&ŸÁ¼ùRæIo›}Áþ"ì7gAt’gy€qBŸE§G¹ì®ìãêóÔ˼ôÕ>pæ§5[ïýÐ?ŸÓwd¶ö=çRvŠüäÿÿ}žú®«žëÃ}\j@Æ¥ö9Oš/»£zëý#Ñ}¿½w¬¨õä3,ß…Àžw3†ŸÒzóZƾ4–W¬÷zñ¢Ë¾g]úN¡{ÁOª@üÿŒé)yßÇœ~>¨ü`­ûÞá{>µÏúFÍ|¢Æ‹õ[G=¨ñ­}2†ùØò?ÄÍœÂì ý¯õZŠÇŸ3»]qW7‰_Ì÷Öþ¬[“ýÆâ—áû‰‡Õ¿ÚOþà×`ÇÄÚÍâ«z¿–<0q‰èÁ¾oßéýÉšçE?YwÇú æí‘wŽä_˜*»ÞÃÎÁ¯!øÉ‹šG¨üŒÕƒ=ÄþâÉ×ÒóàGüI­ËœÎ5Ž5¯lù¨gÍ€sàß5Åy¼ï>«}eð×6?F¾Bq|í¿ Nô|.õ0oʾWÄÏ_èOí Ëú¿Â=ØoÉòÍ8]4ÿðˆúG½¶yÍ3ÔzLÝ¿¥þÇû>íÄ5µßr¯ùZ¿ò=ÚŠ~3.OšÜÕ¾¼w§vì½úþÜs®zϦp6ñØ›¿ï=î©w–ýÆ?»|×~«¢¿‚Ü]a]„îóëØÈçh>rq„x‡<y7ËGÔúòXо×{µ:q–Ù}_çÃ9yóÚ‡ÿ„¾‰Ž‡”·}KÉ#_V<Œ_Â>ŸWÂúVòEµ¿-øÇì6ñSí·!ÜŠÞ•G.Á•ºŽ]e~¢øÉsĉÌߨ?ð…}àßY"ß/øpUò†~’§©ýÀqà ñ·öƒÕ‘ý8Kpî~Æ•<ÐJjÿ‚Ú‡Ûð‡üK}çžÓN­[Õ~«ŒÓcZ_Áø+¯±xƒÚÁî‚á/þÜI>ä!éís¥yÌúžÎ7[ÜÀûÀØòßÈïeí‡þÖ÷°âyüy4ä¤âm•Ï×üŒÎYçZïí¨ÿèO}ï }%þ#^!¡q)~Jî—ƒS%õ=L?Šž’WV½þìëÐÃZgfø »O}ÿgì:~µ¾ÇJYùÜ‹zÏúj?Q•c?ÆÑwmñWÌ w*‡=æ;?ð¡ì¾øL^ŸzX¿É÷l°“Ä£åDó°ø×ÒOæ_5ÿ€ÿzÔü x›÷§Ü^_0}$¯"~×>`àÚkÄ_ÄS¼ÏhycìcÅÇØIµÃ:M}/}±_÷y÷œò¨?Œþ¨äãj]‹Ö÷Õ÷t¤Wí{àBÖ€³.ÛwJëû˜Ø5p‡â´ÊªõT~QבÛzøÔòÕØß‹6Ž>ÿK½ä j?Kµ_ß5~ã7vò‚Þs¯÷=TžüÆeóËÈqÉÇ´Ž\èßaÔ¾:£ý¦j¿>õ»òšØs=_ëqÈóáÄwò?䟰WµÏ%y[—=Ø#Ç¥µoŒüþ^ò›øp®ô±p¿Í7º½ƒ®z?Gôáñ Í^Û>ú£}¨ïU¼¸±âS½o’/G~~BzzEúRëMLN4޵¥äš|°è­ï ƒ·Ínb×+¿€ÿb> ¾å{džßë;%6?‰¼‚Cñä‰Oj_vôLïGC_í Ž_•¿ ŸÌïùúþš÷/C9®õ¤ê=Åš_µù´Â˲wø·ÂÏÈ ~UüÄΣOå' ×±oŠëßA'Ž)ùb^Ëpy}çNïÙ½Šùì€Î‘[ÖµT“Ú¿Ÿü“ð¨¾ãUù*ø\ûzcÅÇ‹ÊG#¿”óýSj¯Ù!ö!_F^¿ÖGhjß}åÓØOû|O‚ušõÝ?ÉÇ>âW•;¯ùÊÿÀû5ä'¥àÆyf^‡ü>ëìÐì)yŽZ/©ëõ½Aøb8 ¹ª|ðqW½Gnx zk$ÖéX¾‘rkÝ5óÉ·¦÷Äñà(ìùþÊçO(o˺´ÊGHN°Ïµÿ žÇÿ“¿CŸÀMå×u¬ýÿÌnx¼Yq½Îµ¿Îâ-z?ˆCn*/­ûØäûÀüýóuàçß׺É-ó¡ØÃjß8òäîÉ[—\`§DtW™<¢Í7Cï]aÏÀ[õ~¥Íc2ÿu^ó‰È•åŸOßüÊ»t¥ï¼×ŸÐ{c•—Ô¸0ŽŒùÛz_ÓæY‡…ª<¡ð‰ïÏ¿Séï÷"gÈq?ú~\*y¢䙼68™ùËÒÛY¯¾ª?µþUõ–ŸÕ}ðÓapymõ¯öcDÈß/ªÈ3ö××ßûz(òçG±+â[íK¬òøUæ> 8qd˜'9&~ºŸbýÆKf?(üð¢Ù“Z7dñrù=‹;öqy{¤ç˜7s?*ÿ¤óúÎ…ðTå°O6Ï_8‚¼9y/Ê_ÐoµIy‚+äM-žñx£ú#zô>kÙ…Ê›éùúî+ý?"Oj¯¾®öÈ÷”>H?iŸþáß°÷èxê‚òŠôÿ–í=Øëú>.ñŠø\~\ë*\ó[øÝGVy¥Zç_yæ D?ö´ü$yÉÏ%í³ÝzÚã ›ç‚oˆ~Æ«ö[f^Vr~Íä=f~=þyOäû!Ž€¯Â]ø™‹ú~öý2é?ññE½WN> ~ÿ¸®co˜/!7²^šy¯Š˜ŸS¾Ï÷[?‘o ®ÿ`×À/›Ýaœ|ä‡þx¾Öàñ#â'òˆ|~@þzxÎñqY}·Vû’±^ªð…ö9„NòEø—›¶.ùÒ¾€•Ÿ`ž…#¸ùd^ìŠÖÛñ])Ö¿bïx/÷&ßÈï‘'¨<vÞâØZ¿£u\zß½ìçýZÇÅzNìqìÕ^ªx°æÄ'ö±Ôº·úp}/—|£Êß«z¾w‡Ÿ«}mEßc_ì~‰|þ‹ýìÂUÉ õóPíC N'èœü ïÅÕwhTïOò½&ÕƒŸ!_øAéyÙ} ùô÷1ùïû™_‹Ì;³n¾áwê;z¾¾k'=‡'þd@t=€ÝÅ¿³þÍìAùw•«}%„Óà£è_¼–ü´ŽÏØZσ¿0\>@î±ïàßæùÞëkð³Ø‰ùPrvAó+õþ4ëÿ°{–G¹nþ ûŽªï;ÙürÍÙßÂaGü½|Žõ^,þýF^©|þÃè•Îÿ™èdtásì‡ÙyÖíÖ¾‰ê7ßÿ¨ï†¨Kš#ŽºOôÜ0Q¸N|xØæyˆëÀ½ø§[¼Ç7|/¤ÞûF>ÊCÙòïØ/Ö/³œ[ã¦v‰/Ȳ^ÄçïgÞg¯<œÊùºjâAä ¿‹\sÎ>^äjÿRè•ò%ùmêAžê½bòCø-ù?ÖÓñr}·Bó´Ì^ÿg_Òþ+Tö”üô®Žõ=5æ_¯Êß©Ýsª÷!â;]ß"wµ¯¼Æ¿ZûÆaW$g/Iîð/È#ñ)íùzNä‹y1òQÄÍ×…«*>1z˜ÇbÜà7qþ÷ÑÀ5ØåuËÿ`Ÿ>`öªÞ;Òuäá­çd=ëA³ÏZÏ7Z§Èú‘½Ì7»ðçš?þsz ‹ý€{ñwŒS­cœ°¾¼?ß_Ç>Â?øó¤õ·p¤Ö7ŸQüPû5YÜ_xŽu Œ_íßL¼¥ý+°¬{`?ÚÃj·Öƒ(Ÿ÷Iü/qˆÊi¿’Ñw=àùì)ãÝ"ŽcßGÁ¾Õ¾˜àäU=E~< ù¾Å1âQÖ‹‘WÃÎG$ΊÎGìûl…ÓÁ«¢“ï_'ŽÓóŒýÅn×DîhŸõ$<Aï#”?'Þ?ðØ ð8z´NÆòµÿ»òÚØúK¾<‡\UÿÉcJ>J~5þìã|Cò Þ±ï1þ•¼§ïÛVúaóvÈã“ü_í·nv~`Ï©‡ï?Ö÷E8÷šâYúWûü‹Å9‹CWê—°%ûþ“/Ã_£¯È“ö©ýñðo¼OÆ{º|w ûóÆ‹yPË “'Ä"÷ÌOûºöó²óØ9¾'UØ÷Tô’×e¿?âÀÚ/ &>âØ×œHÞ{JýúÞWåJÏÔßûÃ|ÀÃòà-Æûìï/w=¤ùFÆ ?SôÖþ¨¶^\pMûH]"ÿ(y‡ïìW€]¨u‹’WðÄ'†öºòá¬ÿÁ.ð~Í×ÙCúAþò^åßÈó€™ßfý󜇉cħGµ>²ä4¿Zy/p±è¬¸SþÊ¿Cξ¼÷¿@X¯ vÏ)È:ïëzžëäÓ³Z׊\j]|.{ÉóÊë_έüŒêe_!Þƒb_rptÍca·oü…â¿z¯‰rÄ“*=!^¹¬O|†õ|·âyÕŸ¾ëQßC±ï†Õþ âkÅ×Ìë@‡ç5Žõ½òÊ¢\…Ÿ­u.øiµ[ó“’çÊKS¿êÅ‚Ãë{Ú6¯ÈýšÏ^ÀOöÏgৈ¯'ìü±>EåXŸVëØl|ž2:7ûË÷ljßVÕ‹=å9Ö½ƒ'ß/½»¬8÷„'¿ Wb‰›±'¼÷Uï§I.ˆ{Ý~×{X¢—ï×÷j4þÈE}ÿ¼Ž|Éÿ?4õžÌpß(Äîâ¯àW­ëP°[È'ù}pöîCš?^Æ_þÃO×û˜z¾¾÷kx’u¹øWðî*r¼ö«%N½Oøˆ¸ä)³3¾ÿz½÷¨¸â€õ«ÖIhÜ.)¾¯}Æ…;À5ï²z¾‘x|6{ã{§˜¿]íKʾ[û“«=òàÒòñ|7…òä)'†ó¶‹¹ú½5?¾Ø£z^¯öÙxòò„äÕlÞ€¸­Þã`žVÏi\ê{¥Ò‡ÅIîKžj¿µ/ܳ8´znqþ¨ëˆÛwÉÎ(Þ[¼NëKà×(¿«xxqäWæ3É-îYµ¿øBÙ¡ÃÔO^}Eoíó8|ÚÏÅœ|¼Ö5±ÿ’ä¢Þ70ܺØG¾šüü¿¾@üc÷â”øA\¼ÿWÿô]ÌÅõ[ùÀÅ\íÜ£zˆ›éïÉ x”ëÇ%¯Q}G%gÐPå^KžD÷™ÇW¾eqB|Q~‘¸Ÿ}”ùíâ¸äKë)kÊ^©Ãè‹Ù­')¾ê½»j—üvÜB½ÇÕ­c^¼E|Ò:ƒÅ!Ñnd¿Ë“Ü×8jÛÅ«Uß1õüJœD=¥ªïU¶?B|.~Ôwú°¿ÌSÖwŒå7Xª÷=oú½~Õûêû‘Ó‡}ØM£òaÕ_íÉ>-ÈgáêÚQúÎþÙì'¿¾ íÀÖ÷´À ȑƕ÷ú˜ÿyµú¡u*‹#ØÑÁ<ß#¹GvàåuT>±Ö÷ág…¿Ç¥µo”êå=zž§]ìÐ=²÷Øiì©ðÌâˆô;ùÌÐÔ÷%‡õ^úQß·}{†y©Å»}%o¨~/ ?šßZ}AŸçŒ»Æ‰üœðÜâ”ÐþPA?kÞNyµ-?!ºÈ³‚ /êõñ½®}ÌkНûuýõÌGÿÐ.|VûºÎÔ|Bù;øpT÷ï`—tÜ…=}:¥çðwÈ|Dï÷`7¥Gìw‹þ“w.}QûWé'íh¼x¿‹÷ÌïQûà‰cØOâÑq~[ú÷Iòhô_zÊw$ëç‡ò^ã²»¬û…ŸÔ¾ö¿­ý¿ý=pyWË‹ÃÌÿJ¾‘Ù×ÅeçÈ—Ô¾âÛìŠÚϧ†ózŒ_½W¸[|Aþð‹àš#ä¿5¯Ng_cí¿WûŽ2Ð .½=”ë­yôBãû:p†ÚyÁÆ›}Þ_´z°Wôç¤ô–ý`ȳø:xòS|_‚õ{5ŸÁ8‘ïR}sô}&î¼—}ß"7Ä¡5®öŸÄ›]þ 'Ô÷‘wô‰ö±×à¾Êcã?Ôÿú>»ÅÝ|—¡ô¾I.‰°Oo”d^ùÿ˜öK ^üy­‡·u à€Ê?€ët^û`Šî òàVÉIᕇÿµß®è©ü#ù&=Wïq“ÖóõGú£~ãωw¾Ü*>°þù©õx¢¹Ä“_€Oµ_q½ä¢ü¡ž¯ï Â/õÜÜgøäú¯%%ÿðƒý2¯ÃÄ]ðü¢q¨ïOZþ’ø±üý0^:}Ÿ¾+Êþ_Äoà/öýœÓžúQNæ?ÿêõc·l¼˜O@îßzOArûú©ûµ‰èTÞ•}Ý>ní+$»@\Vóvâ3þØÚY=ਛðx‡x„ù@ñ¹Þ?w#â vÁþäåoÅGÖY“—ƒè…Öñ/o|¿·öiǪÞc¬×€Ïø+ñ…|/rü,yÖZ×ÎøŠžƒâùâ)òEØ/Úe6x}+ÿÆ8/iœÈû8e\>Êú~ü­Ž{†t×¼ vºjß ä»ù'†ë°ƒÃN âó𵿱Öç°Ï!x¾k]óâ¸ü¬â”­ïãïÔïO ý v¥ö“µïSUÞ¥ö'džßð4ø×¿·Ç÷@Áí¶?Jí‹Îü88?O¢¾¯Š¿NÈÿcw;âæÍ•*¿È~(Äà òÌ¢õ= =O>‰ø”qÅO2¿E¼\ñ„è|ih§KÁïä•‘ä:Ž‹>Ö#€üòòÄS¢gNT/þmûÇV¾ñyìùË7×~ ¢/ü´øjÕ»x\BühòB¼4ÇjÉ“’­ù*ݯ÷—©_×k?$ò–Ò;pqy‘ŠÛÉ‘¿Ò_ñ£ö¡[­SÀ®Ž!ž`¿·=Øaò´ÈxEòµËìâ)‹ Ù/¿¤ñ]üâm=G¼F>R|\¼ZåùÞD͉‡†¸ië»+è¥ÚGïxϺðƒGtý”ÊÕúfü‹äæyð¯ésíg½šG©u¼†ð«‹×«]ä‡rð™÷È Ýøäü„žâÁ{àž¯<•î™ä’ï W?‘ïA‰ïà+r z±;µîÁð,ñ qR­Îã=ˆšÿæÝFë‹yŽ|@íkñþåÆöԃߴ±ø¢ÌÓVþ…ïÂcÏ|ßÌ›øoò~Äqä³$؃²÷Ìß0¾"Jž|?Œ*>Ü˼¸YõÔ¸Ô>RäCÀ³²àâ?ðëlèo{$'›èòŒÝcü° ªŸ8ªü9ñ“îûwÎȳ²?'ûw‘÷?’ÿ'oòñ¡ìqèÞýѺÉZ—‹~ƒwêû·à&æu~D|:¥~½dö¹æ­4þàôŠSDöûŠ|!/ø1ô—ü9x œPûîçÕ·ìºúÃw]jkâ$âa³×ø ×ïOšß9¥ç7ðkà¹!î¨x9Å#¯äÅÞ‚}±|â ¦çØÇô~0öùyqø†y$ðyuµƒœ£OÄ…¤oCSy$Æx†üú©ø£Öƒÿ\Ë'V9â<õï¸Gõ‘/Ãþ×>)Ò·ÃÈãp~¾æI™×¨ïÔ—$I¾{ þ’‘à~}GÜE^rˆw g͉§âuµÿjÉEí»¢vÑcä§ö»V?xૉÛ%ßÄKÈ¡}‡›}Å¥µO|d¾†¸Ií€w)Wv=’\I˜‡,9`¾c?ó2–OƒN} òÌø9ìlù#éËQì»Ê3O.DNv‘Wa>A|>&1ú±Ñ÷!iõœä+ö‚ß¡w˜ðïÕüßG8ŽÕx\OIþ7˜]Ä œ#‡µ[|å;)ä óˆå/ȳ°þ•üq$yñzYõŸ$¨çj}#r¡sì3ytÚ#¯Pûï’¿¢¿øQôZö<÷„ÆuŽ ï^¤]ß^ôStc·ñ/|ßîÚê=¶š?™K¿jÿ’÷óÛ¬wÆïΈÏÁyä1$?Ï çÄÏ¢×ó¸øõ³Îá øÜFPû’“ày{ÿ²æ‹Á··´¿C½_Ìü#ø9îe= ïdó[û½ ñ×Öû+Øp6óxØú‹Ýfž {,9/|ƒ^’€Ÿâ78®òäʼnðª;\xZë>iãSóÄq<&yyøƒ}«÷¹±/âöË× ðÿÙ!.,Ü.!¯Äú/ò'O íRõÿ¯uï•'¿Iþ>Y>‡uyØoÆ£¾£¤#zE~˜xp˜÷.  ODÿ}Ý0öîq‰ìÞ=’/ò¸ø!p$yþšo3ûVyHñOó‰µNõæÌ3܃ÝfÞ }•®qWû¯W¿n óÅGüû¿cj?Gä[õ¼„nµ[ûà5®ØÛÚ·ŠyAp”úɺâ«Â Ãñ,yõ}jÐ{Ö+m7ª]¾{½¾?yeüg}ÿWíVÞ܉>‰ʃÖ<ó¾¡}®õî—ø{$µÏÖÿY>–8äß™±¼pí¯…þÚ9öûTñÍ=Ã<¾}çfkžJãE}”¯}ð4[ÞcƒsòúàDñ;Rr¡òØ?ìâ‹Ãq®8ëi³ãð©ÞÇ1½Ænë}­}èt‚߆þr1GÔÞâ+›7¿£ŒS}_š|“Æ™üëŸÇ°‹zý=&¾=!~øwžSý•O&N„ÿä³Ô~­_U=¼WÏþÂð]íÕ>%ºÏ¼ vÔ÷!ü~Gõ¼–rzþAéņä=GÏ^eù€g†t•¨õÆÂ#µŸ2ñ’á ð?óžÌSÖ>5Ÿ{°›6$VyÃ-µ­ìÉ~Ù=ô“ùBÖo G¬W+;byRü rÏ9ïÙ°.ƒùóš7²¼ôÕzx=w€ùdüºÆ™ù$®cŸ÷sÌÓ©ô¡ö+´¼'yXÞ;W—Ôwx5®¶ÏjÅû”»3´[óa¢¸ù¨Z× ýÀß¾¶ã­á|¶¿/YñÝÓf/kÞ~ƒ«$_àƒ™äãM_»ª}FŽñØÇÛC;2ŠçmÿÖ­8ºð—а¼3rÇú“ú>ƒ®“Wa^¢ò‹ºŽ^0ÎÐW|!ÎÄŸk—ÚÇÐò…ÌÇî¥zžü,öºø„½WyÖâ+¯AžFãëq'ö“üö>ðÞÑ£foj5éýð:ÖûŠàæû-Ïwe8TtW~]í05?Ä_‹ÃÄYÃ|ÀhßVÿ"ïO}Ô⢊¯¡ÿ,¿JüÊ<óÊç3߯vÉg‚'ÀïäÉãó¾ x«¾ Bžx€õÄ b0OYq|íÛ-ù@ߨ©ò£ÄýÄ!Èr®U¿ÁãÈ/ñÕ'‡ã_ø¹¨õ–_@¾êýFì ø…¼¸øËwTçÚϼ˜úcµ>£âÿîB­kÒOõ磒›šw‡²oÈï Áwægø¾vñM|g?2üà£CÿUö³âä\‰ý–Þ¿€7jÞNvœO¾š}f‘;ð.yðõnò<²CsoÏòj\çÀgè¨|ÃpÝÂÖþFÌ'¨ý£ÄaøEñó?ð>v‘¸œ£zGìzDüD|…ß*=c¾Lö–uäYÙ§ƒ8’<ö›zyƒ<ï›’ÿd>|Î{%3Óä©ð²ÍËÔ÷ðð‡øu觉/ íUÍ{ƒ/È+o¿µ >œWÞún†ç͇þ£ü ú|ÿjæÚËö¯<0|¨u§j¿òÃ:'T¾RõTJ~„üy\òðä[É H>‘sâß§„þûû¡w\$|—ë&ñù p,ø<íé¹ÚwCý©yDéùò²/µXô‚°ȃ޻ÚÊGÂ?ü|×sâãâ°úÍøR~Ðö½(Ãüë•oÚ«ïžàGÕ^Í»šü?i~=$oÃ>Ø­ãÛzÿ‹øTí`÷Á9ìRþXv³æ;8JŸñì÷€_E.jŸ›?(¹”ü£+N´¼_½×¬qÓºÞ›*ùG¾*ß­çˆð£¯Æ!5ZySñá? Ïâï1œÛüÀÖ¾”΃#^$¯8Ä¿5ÿ®šŸe˜uèqí/`ùdâpã>ÁïÜâûB'ô¹eœ˜/fb}Vú_ûêzíŽ0ûRû@˜¾×ºXÕÃ>BÄ;àúSø@õßæûpØëÛC»qúxßú¼ÖþÅCÿWv™ø‚xмëóJ.ñ#Øqìó)z¾p)vxÚèB®ÈÏNÃЮüšò*•ï××wq´O8iƒù8òÄ•è·ÉE½.ùä½ò7´¿3/,}ò÷Uð7ð£ü ù4â6ÆIòN^p•÷ÝÉç‘Ï Ï©v.3Ï!¹-ü¢öXGIýèö´òéÄoÄ3†ÀW¬ÓÅð\å ÉóhœßZç¤ñ¹¶Ú?`qH¼§vZ¾çöÐîm釯³æ/m^yðø–}ÓÉÓ_îÛÚ§;@\­úˆ§Ñ[ú ÞdžØ×UTžFv©âRÑϾXŽßN(/‚?Û¼¸¢Kóȵ¾„¸–8óžíóÌ[xBòOž’xþ”ú_ßïRùýF7ö÷þy~ŸÑQï¿Jn´¾¤æSÁš+»‹½¬õä‰g5Þà2ìùkÞ ­~ØóÐA9ì—Ç9Ì×ðþè³Cý)ýÀ3¿@ž›yèúN8ZqóâàJüSí]Uý5#ù`}PÍwhÜŽÊÿ½IõÖ÷§5^o‘œƒêûÎâ;x«òƆÃÑoòZøaò‹è³ï¿W¬çÅß³Ot‡(;¯þ\²|ä-íÛó¸á}ôŒ¼ úXëj°£äÁÏøÉÕIÉ+qvU8´p:ñv<@ü{ëþÞñ$ï+‘‚õþ+¸x@ü!?Hñàvô9g ó)/2O ÿ_ì0y ô]þð$x“<øBÏñ>YåqÔ.å°—•oñ¼×Omn使ßÈóÒÿzû@~ÀòkøÃZ·K<$½#ÿ¢o_õç„ú…ÜÏ<ó?â‡ö (y-?Å8’%ßÊO±Ä=øu ˆŽÄzþ0~EGÆó顾Ö8½ \‡ƒË›ï!O%ûpNçìï@K•ê-ë+OTïÇ‚ñ[ÂÐ³î»Æ{µ_¶ÚÃ^³_qDÅßà7‹wê½üùéè/<òÄа¾o´/ ö幡8ýñ.*½);¿Á©à@ÉÛn£³ü¬ú]ûXˆ¯Ìðž.8v*þ‘|Ô÷ðw¢›q×û[ùõ{ó…±ÛØYüñ ûeð ߇#¿A\À<;óêØÅ]¢ïÃrþÝ(pîÈŸû¾á5îâ›ÏӢ׌|¯ïÒÑü†è"N©u7â/ù§Š7À¿àVòrÖ_Öoø÷+>$NU;Åíõ¦ò䥡þÔ~ðèÆƒË[Þ²òíàä‘#/©ñ"ONðù0èÏÄ ØŸÍ_&ýbý8—ü-ñó¢¢»ÖOh\žÊiÙ©Zï¬úˆ'lž¾öÿšaO˜ò¼“žWU>_×±ß'‰ßUŽç+ÿ.º ÿ’äG‰‹ÁËõÞ#q5þWtn—~à×ðW6_rq”x—uØyí×Ï󵿾úSû— ‡è™ä¡¾Æ|y ä|©ÅSàïz/ƒy =Þž .=ÄKàWuŽ©ïÉN!¿øÚWXãYßEGŸè=Ÿ'‡v»ðqe}\õò¾ôNÚóò£à-â½6¯ ^ ö\}zï`^rôpü?ÇÂAÌÀGòsæÁ}¬?!ßž•#ÎÈíÙ¸½7h—8 ¼§òèóg•×8°O)~\pM~’yÞ|=v{È×Ú§Üþ¬ý°›à(ã ûÿêû£àõû¦ègžÜTû)ãÇDÏ-íCþ½cü­÷ÀËä'%Gè}­ŸèÈwý“Çͯà/Ë>ª·ûðþ×þÄEØ/ð±úsSß-&NCÎÉ#á¿ê=ð3zgñ{í»BJ|`À>Qù^âcòCëßeæ=Í-܆>á÷,ùFð×õx©ýʊ϶n»ô‚|=퀧˜÷¿¹ú®7û¹Ôú}üç íK\ßÃažWòŒ]/`Ïj^ ¿¨ññü2v²ð‰ø®f߯ñ-?£z¡¯ö-¾¼Š|®Æ‡<ïy°xœ~à‰G.ÜM¯}cùpráep ã­ñ<@ ÞÜW`_ÑŸÚßVvô¤úI~?Wß«}<Ïûmà]Æ|¾Ä.Ø÷´\Ÿ*ßðÔÐUúÞõïô]7û=«õBâ;ö¶ìrm~‰ïGpŽýÐø•œ0NøûÞAÿôÜâ>â5‹ã¸Î>¼¬¿ó÷è° ÷^þ¥¿ì×ÜüMÕopU}wÛp…¯ë€/èu­g—þ²Î‰ý KŸÖÿÎ/m½oiù#Ö"/Ä=…#%7È+yãÚ7UõÚw2k-ùpÒÙMðçÕ?ñ·ÞkŸÉ“T¼2Œ“Ê?²Oñ'ß[|Æì%rVï'26Ì#nå÷‰kà§øÃwìá v =¨õÄ¢üXqówÌjÜÎÅÇòª§ö×Ðuí+_yèò[†3±à^ôû–Þã@^êý<ô‚¸^íÖw Xý‘ÝÆ®‚KÙ×Xó[ûøýÎâü*þùòy–‡ þöxæ©á8ð^@ÅMÏ ýû;”?¯uÞj¯ôÄüR­ÿbþÂòwŒ'ø¾³¾¾=¼úÄÖ| ñŠÊ_¯ÑƼ®ìÉ#–ߨ÷}³a|š|G}çÎâŠ=ê8¬Öe¨‹«ï¡ì‡Æ³öw²u8‡mžˆýÔðû¼×@¼öaÅô‡÷NbÕ¿Í.Á‡ú^8rн\Å_§/3žzþQpšåûž5û@œ=¯ï·‹6?=š)?~!çä3Í_V¼dyòqäS}=¸¿¯QûÆ!_ê÷»é’ßQþÆÞ3)ÿI¼ß}¿ËóÄCô¼%ú Så[±S¥G'ðÖ fwðßÔ‹±8cq’ü’êî>¸D{ïy¾ìå'Ìnð/ÿ^=òVóžOÀ“ïÑuÖ/2/èߟ"_Äü&ò^ú(ú/“_–Õ~™ð‹8¿Â<´èÆnÓÏš§Æ>‹Þ›è¹ùoêÁ¿=5Ôï­ü(ôÓì¼ú޾ÞV=Øð;ë¥Ø§–8žxüæó6…ëÔ¿ÚGüó\Ø=_ù¸¡üm}Çò7õ½(üx•yô_ý!?ŒÞã·àG­Ò9úD¿Ëkÿ@ä ú¥þ“Ï­}åõü¯ý›t^ñí1ÎðƒñW™Ÿä~­7 ïoñ"ù¿šgÓ}ò¼ØaìWå#™$¿‰óÒ‹7Oß[}Ïr ç‘ aŠ=Üš¿”|×>F*‡Ô<Ì0ϼµ“z˜"jóaØ=æ]ë½;Ö£ˆÏ™ý­õÝV8æIÉÕUÆ <ªqººÊßÖ¹WLý»Ê<1ë,ÎC~ˆ?ý;¯ä«k_RóSª¯ÞÏñï­ùøãG7†ó[û£íLÅ‹¬ãc~ÁåüÏúÓz/V×±‹ç,ORûá°¼Ûyü«ò<¼_n¸xqÔìºï_ƒ_¹ºz¿¹ÖÓûº Þ³‘½¬uDµ^ÏaÿË?0¾àRkíG+yÃo¾D>@ý¶ïålå‰-~|ñ óàdõ9§^Âî“§f¾½Syöû"Ÿƒý½©}çÁÿð{ɼßgUÜWùÍ£C¼[úN:)ýfýþÔÞ§ØÒòE6U/óI–‡f=°çó‹.Ë£³. ½©÷°ØKéñ9pÍ0®©8 ùG^w›ž×ze=^ù@žCO?ؼٖ«èíÕ>ìàqÃù¾¯IùEÃçõÝù'ðV퇩8æ-ð ܆öùp¸Ôü ücßA¾sB> y`¾­ì8 œ^R¿Ñ÷é÷?ZN<ñ·üò$¼Å<ñ*yìqíkÉx‘—}Œkå#e÷‰âªzï@þMö¸ÞS`½lí`¸”¼0üb}.8¼Vï=ãÿÔÎQ˳|àwõ^~í‹€Ü íQ­g`ü‘ËšoÎ+½Q¿îÓ¾gŸ0»ýäwxŒü‚½^ö¾ä ?:Œ÷+.Á>Tž=&®²ü¸ÚדU~|]*Dÿ·Â9â?ïÍUžUþ :ê{Ó’ÓÑ~<俈ÕòîúŽÞÖºKâlÕëï?`/ßú/VÇSäÉË0>Œ÷p}DùµZL~Åò·…óðÿä•t$ÎçýÆ ¾Ö{ÿŸÞX}—±ô«òªo†¾ íOɯ¾/±õÞ¦øÅ~dõ] É]á2ì±úŸæ»åôï9“[òšäÅ+¯£r•÷µ±ž¯ý’…³_M¾¼FþKü¸6|Ÿ¾öKoÃâAöG§}üãYïKj|à£ãÆQ>|Ž¿’}}?yËÏ1¾ÌcW@\­þᯀ»4Þ§äÏN‰në(_ö‹¼8¸Zö‚ýy^0{Ÿ±÷´Ë:7òŸÈïG4þž?vcÅ=â¯åŸø~ãV”ü ùJp´ãõ‡ïÒ™>~«øs˜Ÿå¸Nž ¿Á÷šjžMtñýÛúîïpÞ ôýcv ñZ­ÿÕ>³5/PëdÀÌ‹ò<ø_ñÇQµOÜvEypì®ø½˜cw‡õÔûMÌ@Ï-ÙqæK¾#æ½™'®bß\{¥ö=çûu7—èüµ±æ«0ŸÎ%~„nâ(Éöùðýf,7Ê÷b¿Ñkìþ»böië=ò#–7ÿ„å+¤#v÷&þBôïzïÃò¾/´Ö¯2Ÿ‡~“÷İnKå°ûô÷AæÿE'ü«ïžn%_VxCý`ÿ öñgüðû؃[_~wB´ìFí÷˸‚ß5n•§׉.Ö]<9ô÷[ß5Ãî‹NÆñ"ö†¼ö½GÞÔOå5+¿[ûÞ ÿè7þ‡ü‰Å+ç±³ÄÑ–¯y µ‹”}ÔõkÃ<Æâò2ªçºöE)Ü< û»õž²ä9zôîôÛï©ñð|^­ŸP?>ª÷¡ ãÏÉn'~¬õž:^f‚öȉ¯èQ}ßÇâfÖa\Ç_‚‡–Û½þomùøK¼€Ü ׿l­£æõNÿëUÕ_p3ùÊ'Ý^bŸÀ7¢\yfñ :˜ß&®|Öä¹¾ ßÍ~=7ä×VüÆuòØ7ù‡·0~äß /2.äsë=@úž€_°Óèq¦Ú½‡B¼IÿïÌso­³8éä¤ÏìWˆùèJ®·¾Û¨z9úû™µŸœøn¸÷û— ö+ ¢‡Užç%Ÿ ž%‡ºŽ¬ù7âÙÕú¥ì&úA8¯Ösƒ-Ï_û’·PyúYëåÉc§D7ûOm0?Á|Ó*__yP}÷oëûÒÌ軣5oýƾ‰¾ú®öS~n7zoyoŽøyÞ›`ÿØò×j‡}(j‚p$ëe‰ãjßpÑ?jŸOøLüF>\߆ñrÙ?ò0Ìÿ]{ûÝE¶ÖY>yµú9¸ºÈûоûôý·š—#ŽQûà2ô܇ݬïGâ_„k¥Êã×ë})Ë¿”ÿ$®$b}q6ö}ªÚ'ßíÁpþ¦î_á:qœìû×0~•‡ä(œµûnØóU[ûIŠÿ|ç 9!?þ¨uߺÏû…e?éÏPN}?ÿ­uá–7®øÍò|•¯TyÆ“~2MœÊ÷a+O¥#ã >Äÿ¬Š¿c+…Ò8ÕþèØqòýjÿ†ê­õÄðA÷ÏÑé!ûõá_ÁÛØ pÌ­Õw JŸ¼¥OÌ¿‚G±»7Vß ÝZw&ºw“'FNÈ;b˜'$Ϫx=¯8Bã^ë#,nzÚìZ½*¹cÜõü_å¨û²|*öÛóÊO¹ßÓ>çŠÏ¬G®ùò–äå¸/º®ƒGÕ.vï ôü ÅãõóCž_¨q±¸¸Mû•?ìcôZÃ=·dç‹?:â§À‹×4¿öñ3è)ü¾ñ5K€ü7j?ÞÃ/‚«È#'Gƒk+þS?Ék”ß&>@ŸÔÎ1›7o×ûðQúǺxpPáä ¿*zêû+Ãv·ì²øO¼^ëÀÀ9*_q q¡ÊÝXf-¾÷/Ÿ>oùâ^ÛwÌóe·ÇZ_§þ!ϵ®w(gµn‚¼ûù¾ycâQËKÏ ßµ®\t²?GÍï3_‰ÿP¹ú>+ñùgµwÿÁ¯~Ë•ï`´~ûÆZOe¸ÖÖŸöa&ßË÷3Ø_à8þJžnX¼÷Sèî/e·j~8ùô Ü¥ö¯ß ¿ÿéëKsú·ÿãb<‚|‰ïñä‹E'|¨¼œíRójø©òߨïÚïÛò>Ð~1žûÈo1/B|.:,Þ„!®úÅ­}fàóU”8é²é1òQñó ûL/ ·áFüê1ò€6½h¿ÀG䱇ØÇÒ­ã©õ7âcÕG~[þ÷Âö]/š¿>"{G|ø”óUüd]ñ"|ª8CtÖþŸ¡g•¯K>ÜœIÓÐÂSÄÍØ÷–áéÛtë½zòÏêÿy3ÏÉþ+ìÃ~Ióœ5o†VÑó–ÏEO°{—lLí› îCîÑ‹¡ýÚZ:œ÷ªy<Å‘WU\"¹fXìÊWïÕlå“5.¼Š¿©÷5ɇ0/#>#§ìËÂ÷7~ñÜõ»¯‰¼kq’¸RrùðÝèáWþ¸Úóý¨Éçû÷ß¿Œf~à[ë£ÈcõkkŸñ—rõ–ôàUè1þMãWyrònä±Ð³WøézŸKý>$:ë=xòýf·™?ß×>£ÌWˆ^ð;ö«òØ3ñ©ö!'¿GÜ(:? <`áDò¶Ô˸ˆ·”W|PëÔêý!ÉÑ«}´·öK$IœIœˆ^Yƒ½×U\OÞˆñÄŸÚ| ¸”ø{ö¤v²æaÈSkTÓ¿­ý<‰{”wªy‹Ój¿Qå£jŸ<Ùôàcø=É—ïÓEÜ}r¨q}HõÂOüç3ä)iÜ­þÕ>Æâö_kÿ]Ëo?²Ú¯”wDNj_ò„Ä?ºîqÜu³e-O†]ö÷ýjž‰<œÖUÕ¾øÒ›ú.“é ¸ <€Ý‚_µŸxÒqŸÚ½&ë}ÙëSøãažmkùÑ]ëºÅ¯Âð;Žü1߃_Òõ®¾7\|"OòaÆYt2Ï ž°ïÆß¿š¯Q}—ì{u7WþºüñVå›ßÖ¾¢äiÀyâÇ…ÕümÍ[ž=´üÅ}ºþ¬ùcì üª÷žo“ójh>¾õ=â)ñ«¾Àøê¼Þ›¥}äYöœïÕwXͯbë{§ÈóœÄ×äoEv±òÃ|ÎéVùÌѾ,È‹üÙxŸ$ò€àðþÕòz”/ùa>Vö=¯<öpäôMá“úNì>~{åï'c'ëûyÈ•pîAòàKä€ü~AòÀ~,Ä?µßã…Ü#G–÷«õ3à ÑÃ÷^õåÿò“ï®À/^"Of~›ñ®÷é ‡á÷á/y!ß'­Æ‘¼&~d8O°%äÀ'f¸½·ïlÕù#+\XÏÕ:Ëw•¿j¬Ù üü3C»½•wÅÞ ¯°þîüj€²Œg}W?§zÍßV>¼Ö/j¯ß}å[È‹Tœ‹—3/„¿¦_‡©¾GÀ§àcË“¡7ð ‰<‚‹ÞnÜ}‘lk}$óÄ¡Œyp¡øB¾zàóÙþÞeíC*¼Çì&ñï=i¥ÑzðšWS}õÝ*ÕG^´ò¯ºÏ{å|oí†éï±ùPpÍ%ñ‘|úÞQ/öâºäácÚïËpÁhÝr޽­|ŠÚGá÷ÃØ#?¶Z_Qß­xÁøËøTüjñBù1ðrôÆáú–üË»‰»Ó×´Îã“f˾Z<Œ}¬x†¼íêú…å´Á×ÿ¥Ò»ŸÒ¾Á'°§‡é}ž»®x’õoȽïC|;ž¹fGìüy‰y!3õ oõ}.ü#y"ÆÝæ%*ÿ(?E¾¸öÕT¹ŸT?]ßK>¤àˆZ_¢þÛ{Ô[ßãÆoIÏ‘£û4¾µ?™¾wsín:ÿ«·Þ"n'n#_#9¨|6ó*f'™W,œ-¾“—f_LÖcÌÉ‚#‡òZù§ÂêßEôJrþE^Õ>dµ£Ïu‹ÿ´ŸÌÖ~Èø'嵨/Æöç,½FÞk‡øÃ~änþú»€¨Þ§?ôÛóQœûwȉçöã_È÷Z?o›~”|ƒÚýðê»Uå?¿Tòˆžùw¶ë}bÏg“_Çj<Ñ÷úž¹žÃoÕ<…ùð‹¿ÇÅ{ÜWÿäÝ‹£ä™—ðuÄõøEp€ÅµþUôbß|} ã̾äY±õ^óGØaòIj§ò¬Ä¥â{½W_Éw^çÕû£àaèGkÿ,p–øÏxÖw…ˆ›¨Çò‡…3Áiʃž·ü× }ÏÓ¿OŒœ’?ÁÞÿOßö=Oá}eÙÛÊ ó&[ï›Jߨvï_jÕ›îÝšŸ4ÿøŒG}/ÜF¾¼¬ô‡ü}ÿ¶æ·«’G¾§]ërEÿ‡YÏ]R;àVìì¹ÕüÃîß7MþüûèµNý–â1öeà}À'‡z¼µ^R×kÿNð¤ž#޽aÜÊN‹¾GŒ?¾v³ä;H\L¾Oσÿj½8Èìù¡úŽ®è®y4å…niŸOû.Wõ¿Öý‘O’=ÆøZœ„¿å}[ö½Å‚°Ce/ɾï!åè'|g~@û=—?'ŸSß÷4>î5?æñ!òÌ:ÝYÅA…?‘ònÌßÒÏÊ›Àü+ùVæ°äËÁ‰’ËʧŠö ©÷PTþ#Ê/<¼Üæï¹Ÿ*=Ãîn$þÖ_ï±ïû5úûö7­õ¦pãÎ>óµ¨ÙŸWªïî·ô0þøAŽä]}ŸâCâäZWbã\þ :Õ~½G)}ÅÏØ>#[ß9EO<ß¾ª¾'Éã’çÓø‘¿¨ýT¿ö_®÷ø®Hõ_r]óèä TO­û`<Ì~^µï/vÓòJ7/Ñ: ò·èÁµ•Ê~­âíg@¼hr.? حµè5ñê³ïlÖ:ÞŸaþ•/÷ý_±¯7´Ïq¶¯K®ùò äÕ”Kó6äOÉãÖw,ðk*O¼UëâT_}§Gú~©÷lÞ{H?jÿ ÑSë&®ãç}Ýp­[6|{Qëœ}]«}‡f ÏÙúüL½—`qÐG•7«÷~‡ùò_؇ç‡ò[ù]윯‹¨| 8€ùü›#óð©ÞÓ%Ÿby\ä»Ú“Ü\Ä™~ó}+æñÅ/ÑϾ]åçÁ{à;âð­å›^"V/«|íëny‚Zoļ q–áL_¯Xëù‰+dˆ7ë½:ì#öœ@hù‡¢zÉC g¾ÑÆÓüáÀ­ù éö}ÇÿXvq\õ]Zù¥ò/µ^œ|¢ø}Ÿüï!ÏL>j?}Å |—–qyqˆ‡ê½@ìè­áïUøj8Ÿ¸eo gVý¶yŽsZ×ôœŸïìï{Ö>­ºO#ù]ßOþÿÍâ_}—”|&qã%þ:BŸ’yåÚczÈs— ×Œ'þz˜×«ü¬­c¬xæEùmü.ߣ*Eœ¢<¨}{ë=k“;â]ô¢¾GF׿;ž2»G}•G“¼Õþ/ÌããWÄÿ‡ùêÚñ–ÙKpÚóð>œÇÞúø[v>ŒÞË'¯e¸†8’ýzjŸî¡_¯ü›ÿ7<_úÎ{Ñûl·ÖEÈ>…üNõ÷À17ÕôÿÕÏk+¹­¼/ùð$þ°æ‡ñƒÂáûɳÈ/dÞÁòXÈ òqCï·\]Å=5®é;µÎùQ=àÛúnrcß›"¾©û†Kéï#Q?ñ"ûwßÚÕ­õ ²ß¬{d<êûwo¡GäK™>ú‹­ï>Z}Üg¼k=ñŽÆ ÿÉøÕþÂäm·¡_«u³Ø¡ðåß}éïú©ú^E­ƒ5|‰Ý¬õjà.õ¼Çzvp‡ömíS«¸më»ijÇÖ l}¯Òâ4Ö-|œ¼ªÅõýPåakŸÖ‘ àäZòXyÝGOÉk>¤ï"—¢Ÿ<…ü4v¢äSÏc‡àyèZ_§qüÀ*º•oÇï0ÏŠ¿Q;žŸþ˜öQ¨<y^ò¿Ø3ìòn¸«Þo%jù–Z¿‡\N"߀Ÿ£öèã:"÷ð‡þ²Ä“Ø%úkô—ÝP~‹ï³Þ¶q¿’G©}&4ïaï!Vü¸Wüyj¨5ÏÁ¾—•§@ÿXç£÷תðë\ÈWQ?vü#ZW…}ø¨ðïW_Ôü,ÏCÀµäoëá<­/ÿÇ<Žöwz~0?¢úίÖ/ÖþJeçÈϨ½úî(ùbÆYüÆ.ÖzuæÝ‰Û‡x}ôÝ<èAÿjžF=–÷}ÖôáŽñÇ¿ü!ö»Rýõ>œ—OzžyUûµ_…ê)œE~ ¿‚^Yþ~ÐOϧúú@Ÿw«ü=ùlì=y/ðy2æÉë[Š÷*߆|‹>äàù!Ö'ˆnæë{ «ï+,ˆ.æ§Ïk¾>Öþ4‡÷i~’ù¢š%®‡?Þ¸¬ýíÁo…ÛÑ_]'®-û©~|Tý¨uêàòäÝÀźþ qÚ*·8$ùÖûï§oj]6z»_~¤æ#È_k&õ×÷ÓmžºÖh\ßð_¯¿Ä|àp]IásìüÃÇ/Ƚ/³•÷1<®õ}®hºÀÝø'ô¹ö‡^®÷±‡Èó쪇|²}7{ôýèâ{ŸÌ ú~Däïë}µcûíÔ÷]k)‹—ð§5Š >Ång¶q*?¬xs?v¿©v<¯ Ä-µ.¾©žZOoö¡p>öÈâ$òP¬W®|5vüK…üùË_`qŽä¦æ¥+¿bóAÌ¿±ïóo¶Îa´.yÁ¯à/ž79©÷ÝÔ¯GÙ÷ž8AzÉü#|½¦õ0µO¼ú¯¸nk]q?öPþá ~¿%zïUþ ÖÍnŸg¯øîë§ÏŸ£/Ð ßÁïµ.Sv 9yÎÆã–Îo ÷BWáõ ½-~`g¡Ûâô®ü±Îé×#úî*ë[ÉWð|ÅóÌ÷©âàW›xžxC×ýûC¶.§Öá×¾×ÒƒçLß¿šgÁÞހϵ_ ýQùúNë4ÀZ—Vë®™$®!ï ÷eÁ|_“¼ïÃþÖzKÕž?dq=ö¢ö§A¯E'ó*‡È3ã7ˆOä—ù.xí)¾>´Úw+hó¾nßæ)O?F¼¤ñgœ+?ƒÿ'?¿Õ_æù©Ïסß¿ªÞÚÿ—¼¸?†ŸÅ~YÜöˆáK_»Åû&•GßÙ|ö ñùþúþ6㌾ŠNÛwyk}4xLå¡ëC²ƒµ^OöùœQ^UöÃ÷ÙD?û÷–ê»jÊoóÝAübíŒ=&Á|v…¸Žù<ì°é±ï/GûÍè£åÕ¾È)¸ùÇm çmÙ§äOr½õ>ð"yÓgŒ?ô³ÖÉ`o-Á÷õj|ò„Ãy’wìñõ»Û¡üêê'yžò»†'Yo\y>ÃÕn' 77&ž¢âC}oƒ8Bü`ŸgìôóÖ/ÿÎví ýÌ0j—~‘ϲu˵>ÝöÝZ.þ]ÓõKà'õ××5Öþ_àõ“ø¶¾'®°î|M¾N÷‰±]Í#ÔþAÚW®æï°ìëYï-ï€ÇÉû‚³´ý©¼ ö»ˆ½]W™Gb¾\øšý|?þò†â/öçÑ·,'Hžåû/ô–ïÆúwhX7Wyop¦ž¯u”àѾÖ÷ûʧ?kr¨uÌ5O‡}¦bóüÈ1vñœ•óùÄúNò ¾òäù!ÍÿT^ŠøVçÔ:„úÞÕûøÐ¾nÅf?ðÏŒ#û„ÖûÖàjò‡Ø+ž³¼yæÇx?½AÎ*Ï,ºß·z/³æïÈó×wÄÕïàó„äGUÿû—Óû÷¿Xû˜×þ?zþÞÕzÐZ÷Sóø!Õ˺Vüë±yù’ñ»ög¼É#`Ç$|/’xõäÐo,mgx¿¤¾¯î|"ßÅ:hÉuŃ|7êžaÜ·õ=ÉÉ.ògêWÍ¢—àNáâíòÛÌ‹À¡¯}X·WùCüv–¸¹!Ÿˆ|Iþ|¾ßýq}oZã€~^â÷-/Wùâ ÓãʇËoÛ¸»Œë5[ïÇZ>ç²êe_ü9ëÍØ¹®ï$k:êá;Èþþ3y®z@íSí<®´µ~÷®oW[œß߀]¥>=Wû0~¢ï¦Þ'ªýÅ,ßN>ù³|ãÖ¾O+y¬}a˜ÿ¨õþ¢ÿ «ïP3odó£>Ï‹?àûuì§Yß‘ÿ˜Wçý¶Z—ߘ' /L¾Z~?eû½mùEæñ‘_ä”<5øŠöEŸâìZŒœv&äEk½äp¾¨ìL½ïO~†<yiò^C|¶ߪò¯¬w$_[qŽúÇ›üù)å5ê}oÖ’çÝ0yÀU>Uø ºÈïû{ÃôÿAíçG^ì¢üÂ!òUàòs¢“xŸýÁ{è8ù„ÚÏDÏ€¼¼ÆõE&×5iïCm}§‘ylÝøîò§/ÞŠóÕÞ9òÔ’£zŸÇð&ùäíƒæÇÀgþ¾ÿ£äÈiÁ¢Ÿ}tGù$øLÜ/» ޠߨQä¼öõÀO©•ï&¿­þ2ܯ÷ɉóÉ·‘_Ýô÷Y³èøõƒÚ{÷àÖüà­uløaË‹Öü€ž»©ù;ß/ÿQß)“¢×Œó7EOí „Ü›}Ã1?É< |ªýÑì'ù?Ñq]ëÒ®iÕC䉠WçȃïKçûÂÕwœÅwð(óí俎`?<¾½ä©ŸÏAÏuNüôÔ°þ­< öIôÕ¾WâC½W&ºkÿdô¿%~’ß"î‚?Ú—qq\K<$>Ô:Mp$tÃ?õ¿äû ~W7꯷ǹ—Õ3ù¸®ï”/›vì]÷c¢ßù´Sº}\’¼¤£÷ëŒË=V®Ç_׋öÏù“ê3;:=*Wòèå\þ’ÝH|*‡vÙ—¤Þî¦Ýg|æ®TOÒïY‡þ4ÞÎÇM+ŸèpùrùIô&=HãñZ;&ºœo›vœÛy¯”Ozä~>öêOþÀ¯§vçáúÌî;Ò¸'½x¹v®çÿ\ˆ~wªMöÌínÏ;>˜Û}ÞÛ{“=çüçºÓ“úåí#'ÎÇ$×IïœÎž½äèãèôyûÉîmÚ±çg’þ'½òþ:‰®‡³Ð¾×gýɳ÷ƒsç·ß³'I¾Ý.Ó¹gVþ»îýöv6í¾•ùsʧþÏÃ}×#ßžèéå,”ó~:½›V.ù·¹SOz>Ñ{“~R¯ž¯qØÜ¾|Ýw½ðz®©xËý§ëÁ|x½Ê¹]÷vôqØ´ç½Ó£îs¹s>»ÿíø»‘½±ñW² ÏÅç9wy›ÛÑëç¹™=—ôp"ž³ëcÏ~:ý<—ú—ü$Çä§“šÛýM»ïúžä,ésòx¹ƒvôö~u¿îöhÓÊ%;·iÇÎM8Áùšâ ç®ÿM~:Å?I®¡+åG6‡×GvÅûëô»~9ÿ’_p»’ìsê×,´ãýv9äzÊ rníÚs‘>—“„¼>ÞŸëå½½îHqßÌŽÉŸØ8Œôv§í'{âø0àΘg…^ׯ^ܾSÑ‹Ýn§ú^¶çb<ê~~b^b„ƒ“]òþmÚÑå7ÅÁί¹]Oñ#å’q¾§xÛÛwùNç^OG/"žñ¼›û§„¿œŽ¹§øËã&¯/éaʯ§ø/ém²ó#|èå硽ž?„#6C¹ào·Ö lOÄà —zûs;ïÙ›¹]ïáß$‰®d§’Þ¥ñçùÙÄr.ÏN_šG›˜Ƀ㿞ÆefåMNGòÝóCÞïä_{ãM{ó@'ÇägÒøöìXòA¯¢M~¡ç÷ÜN¸ÝNõLÅ-3“_â¾÷ÃLJr´ïþ4Å%®=H¹©ø¦—Ÿ÷¼€ËÁ¦Ó<ÓÇ1Ùƒ4>÷=Sý¦áîø|ÀáG>­›vžæ1“Ú´z“|ùuÇ.Ç)¿æåçV~nG/¿>³ëÉÏôp÷Ìν¿^¿÷#Ù ïW²oÔç|Oø/ó¦•ç|nÇÍá±ÆŸû´Ÿp—_÷øsnõºsþ$\¹i×~Û´ûs»?³ëÉ/¸ÿNúçí¥<çÉÏnZ½É®ÒÞÌν?{ûþ<çÁ^8]#ëüóú99Žv'á–„S2ÅsÉN9ÝÜw{eãÛµãs;wþ:~r9OñB¿zù¹»œú8½Nx æ¿ÓÑé£]÷§3+çö'ÙÉ„?6íºË£ËÛÌîûx9ÿ©fG·+s;¦|V¯óp?àj§wDO/_&»Gùä—œnÿTÿ¦]÷zRÞ,á씇 zó vóâ®o›áÜùïú—èOú1³ëiÅãÖà¢ÿr¹Krer4z>Ù·9zø5áš„ogv?é[ŸÍì|Ó®ë|„ǼžÊGŠ»h?Å1 ×ÎìÜéæ¾çÌŽs+çô»½˜Ùý^þÒŸw;Ÿæ‹7­׋dǽ]ï—ë§×ãö)Ù¥^^ŒònG]î¼}Ï7RΟŸÙõd×ÝO-ŸòË)/–ð@’«ŸÓß™Õïý˜Ùóη„ÇÝÞù}—Íás1~õçæÖŽÕ?Âñ>½|TÈOtqM'¯íüñßäuDÿf8÷繞ò‰ÏÎ'ŽÞï„Û\½]ç·ã§ßíõÜî;ª×åÏå=ÙMÚ™ Ëqîríîlxñ€•É‘!îµÃs›ö<ç½ñäèøƒ£/å½þdÿ>r¾Žôâµá¾Û}÷CÐ5³ëãîüågvÝíwOoœO®ÿ^?å\޹žèNþ)ÅeŽBÞ8åb¾‰þ¸Ü»¼z; 7z9¯ÏåÙýÖÜ®»\yÿœOF×ȯ»ÝãèùD—Êìãó®©}¿Ÿð_©ɯάž”wMñ­?çzEý¢{ò:'“‹nÞÍÆ!Îg$>¹ü»}KrBù”rþÏ>·oì«Ës/¯âtöæÍ'õð@Нgv=Å#).K~rfÏ¥|ûÜŽI\îçv=ù9§?µ?³çzyq§Ûå/àÝèŸÎt>ø¸8]´ïô%:f¡=·Æ÷Rþz\Þ6í<åŸ]}r{àýr9à9çkŠgÃrÝ÷ üy—ßD·ÓE»)>4ºêܯ{?Ü®ùuèt;Iûn§½]—ó„o8:=>.ï´Ãós+ßË¿%{œèJø¯ãïâ|²×ŸèOùÇô|*§û1þIy5ÇŽ©?é™Ûáà/»óâ½¼ë¦]^B<ØÍg¤ü™·—òNnR|çøÄûçöiÓΓÝIòÔ‰ÿFvÞqB'íæ»þõ3Ù§Ía¹4ÿçex>å|ü¬½n^6ÍǤ~'<àí÷ÚMñzòOóa¹8‘òy”›Ùý^|·iÇ4?“ü¼Ë÷NÔ—üròVOÌ%=tûáö4ÉÝÌê ã×G&<åò–p~Ê+nÚõd¨/ù—“0nïRŸ÷×Ûs|áG/¿¿i÷ýÜí[òw)®Ý´rΗŸ” øÝÇy„§R|‘ìw˜?ŠóÇ)NLòßË+¸èåÝ7ûu§?ÅÁ ?&¹r?ëãæãe|S²+ANã¼åfá˜ìmòw wÏíÜý.÷C|ÜÍw$ü}!IqÕ¨þÔ{.âÈM»ïúåã¿iÇ©ñD/oÆ'á·s)ïðÆÌÎߥüôÌ®{?œÎ„Ÿ\Ϩ×ù1·c²S3+çòœüc½NŸÛŽúœÇ¥)ÎHý¦Þd¯Üôp§®G=´8¬Wn4~)o’üCÊ'¼âô÷ⱔ߸§s?á‹Dï<ÔÓ“s—4®IŽ]¯’þ¦qsþ͆LjGÝþpìÍC¹¤½ÞKñ•ËSÂQ={äô§xÇíåf(—≹ÑòA£þÍŒn/ïòµiÇ?¤ü<õ¤øœó„}<¼_)Žâ<á:׋dOÓ8©øž…ÓétmZ9o7Õ×›×÷~¸œ&¾:]nSüé|šÙs3+×ÃO ?lÚyòÞŽ[Òëž?ôþvò¨1?ærðpw‰]y_o7=—ìS²WnGÝÏέÜfx.å“}çèvÕÏÎw{ºi÷½Üwþ%êôYù‘½žÙõM;Où+Ž®G ?§|Æ,”ãº/×}|¨7ä GýIóÔNŸëeÊ nZù„ƒ\ŽCÜÛ[§qgGŒÆwf÷9Oñœ·Ñ1¢;écò×®<×›ošY¹„G“¾{;s{Îñ[²›É~¹<¹žÐÞÌŽ.?Η?÷_n‚<ŒèHù£/®KIþ:ᨹ•§”ã¾ó‡ëîŸçvžôsÓîS¾7O±iå9O87ù™'?çÇN~b$ïßÇõ)>÷þøøÎì˜ì¥?ŸòTs;¦z’~¹ýä>ô9?“ßuys}qþº}èäé#nyßæVåfvÝqŠóon×½èN¸?ñ3á¸M»îõ‡øzäw¬ý˜g¡½„c“½Ú´û)Mþr6,ŸðTÄ-iœ¨×ê‹óvî/ævœÙó~ŸcŠ {ùŒ„¨7­óØ ýJzíöžúCÜó_Ô—ìM/ß›ü®Ï¥ñq;âå­Þ¢ÛùéürúüyÇÝs£;ÈoÌKz}i~Ðõ£Ûƒ™{?æv}Ö¹îõõâèM+—î;ßÍ_ÄyhïWºîx`Óêu;ØóÛÞ×[ʹw½éÐ7jßñÖ<”ç˜ôÅä#åͺùs—³4ì²ëÛ÷W:?Sœ=i>Ûéóö6íhò:âŸ×ŸðËA¢ÛåtÓê ùÀ]¡½¸ŽzfGç“_ïÙgÊ›¿ŒzoÏE\òçö¼÷cêü‰Ë™ëE²;ô'Å;Võ8^K¸(ÅÍ!oó`Ðéy¤÷¯ŽúC½®ß~nçƒýõ;{½I6ÃÑõ“£ãð¹=—ì™Ë×Ôu ž§U{îï¢ú¸ÎíÜâjÇé†Î¹=7·£—÷~ø1Ù}çOš—sÿòcE—×ïúãvn"®ìâ×s—'/ïr—èwùÙ´ëœû¸»ÿây×+oÇõúgV~nç)ï“æE>Iã–ò/nSþ¢‡6C½)ïàúâr< ד¿Hø6Õëõ¤¼£ÉóHŽgžM;w=NóUœ;ý®gÉ¥|þÌʧ¼æ›­¼Óáx7Å73;ãó ÎGŽa<ãüjÀ»£ò!ŸçRÞ<ø³hÿÜ^$~rî~*åIæV.Íg¥üÚÌèI~žrNŸ_çœç¼=®'ÿ“ò:¸¾x}>žnoS|ÕËï&>†y©´N*>7 t8}Éî{ü²ʧüËÌŽÎçdO¼þä7{y?÷w®Çiþ¸·.Ëû“ülšGîå57·ooòûÁîGÿ—p¤ÃÜÊ|˜ìþÈNl†#÷{q™ÛÛÙö÷ãúo7Å#œÏìèúÄu÷Ûn¯{8–v½¾„+Üÿ¸±çcÜãt&ÜÜËÌþç$¼ÍÑøçòçQ/S_—ôÃõÁùÆ-®ßðs×——§Þø¹Ùñ¹=çöÎýÏoÚõ€ã⺨7‰ñiâKªúS|:ôy»³aùˆK{yÎnOãê~cÓÎ]>Â:”‘þQô¿ŸKíΆ÷GrìÏd{å|Hy§$/ÉŸ¥8gÞy>Ñp÷³7ÄxÆý•ãØ„ÓŽõzÓºëϨþ¾éæ‡ÜîÓžó9áµ€Ÿ»ùãþôñp{¼i硽8þn?ÿ7‡÷#=^¿ËeäÏõp õmZù0¿×Õ%9ðó4®~îö%Å?^¿Ë•Óçv{Ž)Oãt¥üHŠ×{y /—âÄ Ï#ûœò®‡³ÐõÏípæH>ÜÿÌí9¿Ÿp@¢#·Û‡ž˜…z\ß9&»`|ñžû¥Ùð¹˜ow{òÊ=¼pÙ¨Þ™•OòïþÊõÍø×Îíz’§+Ùa¯f÷S^=É«÷Çï»ÞÏÂýÙðØõ7.·s;qXÄk´?·û=ktŒüš×ŸðÓ¦•›Ùý·Eÿãõ9ýÉÞÌí¼?ÏÃ}÷'.ÿ!žñÏõ7ù9ãs\OêþÕíà›‡åâ:o?åÁ.qû3Õ^̬\Â_›á¾Óëráã–øêç ïÐn²×Þ¾×ëãåþ:Ðç’]M~-ñÇ屇?“üz}^ÎõÌË»ž¸;=ŽgR¾*áG÷#Ý^¸ð8€çzv…ãæî‘üܾ\äcÊÓ;<®œ…£óeÓŽIîSºìÉ·Å)ÏÓ•ç£Ëµó#ǦÝîÌÂ1èͨ¾°'οx;‰_›vîz`8aÄçPÿˆŽž½Lùµ„7ø8:?z~$á§¹•›‡ë.ú’ôÓÇwÓêq>ù±“ÏÕÓ[5·ò!^M~s$ßiýVŠsÝ.ûÑév$ÆI¼òˆþ$Ïi~†væv}êIvÉû3דýJy¯™•›…ë^¯Óéý x1âhè óQo“?¤œÒƒ¤o©ÞTʇ¸e„ë]6í¹´kÓÎ=ÎóûN¿ÓCùä·Î”ä¸iåçÃû£ënϹžü€Ëмþ®í­sv}fç=âˆ:çò<];ëô'ùpû–ú›ÊÝ3,íí¦ùˆÍáyŠ[º~™c/Ïïõqîí…¼X’¯8.ޝÍoD¼5Û¾½Q;þ\Âg›Fד_tú<^·ñ•Ÿ[;óa9×Ë8_ðb×/wâ”'Iøµ‹‹“üøóÉõò.óíéêÒ×ǽ|ýÜÊmÚ9÷Óúä^\Àqn÷Cþ²—߉|Hv*å 8nZy®SoˆCc~ÍéòöS>:R¾~6¼Ÿðl• zõÄÛu<4·ëÁ®Ä¸Ôó‡É_rLùR§/áôÞ<çÜŽ&ÿqþ&áUçgÊ{uì]/ñÒÜ®O­×3;OóU3;&ÎùNÇ»ÇÅø€výy·=}qýs=ž…ò=9ãèñ}ò³iÝŽãÝ€ãFõÌÃuo'Åͽ8ÃñXšïí­sñ8?øƒ‘ÞÍC9Ž);^OqÒ(÷úSœ½i÷>LöÅñÙ¦ÏíÜéJþÆýù¦“_˜Ûy×›Ÿ¥8*á™Ùö÷Gí§¼Cê—]µëòÒ›Iòâ:§ƒ~ø¸'üòñ=¯¿—ïâºë%å oj¾'åÛÒüµÑ3ªÓÎSÜŸôó´Î!ÅË®‡ÞžÓO{i=d°si¾9áòîúĹ]Où<Ús¾™}ëæÁ¦Ž‡Û»4?çz‘ôÑå½g§¡#å’Ý óÙUßÜ®»|ð\ÂI›Ö¯”‡ðçR~ÆévyLq¿›ë…ë3õÍì9ÇAÜŸÛÑŸ÷ñàú›í8³£ó•zÍ^Ç8)­×wÜãã¹iõ¸}q>y½IÏ­½ˆëæv>³zi·çÿܸÞÐŽ×ëãíú3·ö:ë·b\C¹dg¼ý”ï‡îY8έœãUoÏÉÿ»\oËÇü]’ßÎ|IÏ.øêø…ëI.­_ѯ¤ø8øÕè÷“½uyìõÇ颞N¼ó©]·”Ky—[êKz›Ö‰:Ÿçöýó~øxôæù<^J¸~ÓŽÞ7ŸëÎlZyï‡ËgÈwFyíÍC8nq¾%ÿ?³v½IœÏ)¿‘ìѦÍž9ßbÞsn÷S<ê|›‡ú<^pÿC9ÏM+—øcõöâ•( ¯ÏíØËWÌí¾ËíÍìH=.o!’üešŠx¬gS|“úåÏ'¼ô!Ê·ÖGÆùÎ{ë86‡å¢žzRíúÙ[Ÿâå¼·<7·ëIoܾͬ\ÊQÎíâ¦]÷çRþÈígÊ%;?³ûÐáxùÍvîvËå¡—p:’?7þG Ý®G½|…ÙÙî¼s’»M»ìpÔ«W¸ÝO¸ØõÊõrÓ®'Üä|æ¾Ó›ÆŸò ÿoZ¹Ùðzôi«£oÑO¥çƒ¿r~Dûêt:^óãÜÊ'{âç¿Ûç¿É}¬ÏõzzïÎÂužOãÇýOIyé4Û]W“æfvŸs—CÚ7=ór´»iÏ»½Oñr¢ÇèíÎÇ»ºNùçëÔuCIÞh?Ùu÷wɸ~%ûä}4î^_Ÿ…ûi|æ¡=®o†~mZ¹Gy{^.ä¯b>˜co'\ô×ííPoðÃ1¾žÙu?ëÑ#~šÙu÷s.½ú]ïÒÑí,×§æË7C»^ŸëíÌÊ÷ÖÑ9êÅ#nߢ¾õì|Šz¸ »œËÅü«÷Ãëq}á¹™]w>m†z{ù‹$?o¶ò\w}´¼yÊcÅ|•ËEÊß9^ Ýu^nŽÞo®Óõ;]‡ŽqDÒó¹{ñiÂÃ)çç~¤ÞÄÏTOÂ)ÞÙžðPZ¯itFœ`z㤴®‡rÞ/—}Þ_ÇI”KyÛ„«Sþz­_Cžg¤=\•â——´NfÞi?ÉMŠ}&ÎÇEü“üÙ,´ãrB{!ß6—w‡uúq=¤ó+å/Sœò¯#¼éþßÇ£W§¸jn÷S~ÅéqyMv:éû׃v>·ò³P.å¼_Ž?’ž[ùnžÎéLír}ÓŽi>¶§ïs+Ÿòìé˜ø’ÖQ{¿gF_Ç¿ÇyŸM;Oq“ã Ïo¦|UÊçÎí~ªÏèëM½]ÚÛ´ëÁŽNž?Hñí$ýÔ1æ¿þðcÈïGœòеiåŽHóÉ.†u:1~OüÞ´c/ÿœâ®Ÿ'œôÙí‚ÓpnœGwÿåò”ìÓ;³rï8)™Yù¹•w¼@y—çٰ܈¯ Çøx¥uq!Ï7âßÜêóAÐ;çx?S¾2ÅK<øñaÂiΟÇÇ|Jj7ÙažKò˜æœï^ŸÛCÏ_:ÿ6í˜ð-íøsç$¹äù¿ð¼ãÀdŸÒûÊÐ5·ó4?3³sGî'üëö*àÓ˜ÿ±r×¹>Ííºõ'Ú‹ÔþÌΓ½ò£óÓí õnÚ1ѹi×yÞûíô'¾„yטðö¡7Ùm§#µãüt¾ýѦõ=Éž9ŽHãâõ”¬rÉÿ$ÿÞ÷J묈7’~¦õâ´“òñn÷7ízò¯)}´“pŠŸ~íÅa)Ïõ Ä}qÅæö×]Žãó´ã|œÙÑÆ9Ù‰¨¿óíË'ÿÕ¥Ûó+!nÉËåÜÊYý#{ì[o=Ã(‘ðt¸\öøò§1ŸÞÃ?Ng'/ßOñøËÆ}Ä—Ç{ýNwÏ®q¤Þ4žA_Óz˜ïóvÝ]9÷úR<äôܗλvÃíCŠÇæÃû)žŽùï^ÜD;÷Æz8&ýMù]÷Ka~¡Ê÷æ5’^lZyèu;’pç<—æïzö|sXntîíx?Â8Gy÷ú©×ýDÊ+¤þ¸ v!ê]ÊC[}1ßëòµií E»ëÏ›‰y¬T¯·ë¯Ç›v?Í;¤ñ›…vfÛßù±`Gýö8×qÀ¦{½Ð›æ/Ün„|^äÓÌÊsN»žóò.nÒü”ó1Å÷s;Oþ/½/ÉÏoÚuúæ—Gt$¿pėwâŽ#½ |ˆqk/‘ò÷®WÔ›p»ë_ß´£ã2®§õ,)þÙý™•ãèãbó,ɧ¸·‡»“}ñú¢|¹=ÙܾÑ|‘ÇÕ›¡?ÜOëó̾ô!É]›„'F×/ 7Ï?õõçtãÉ Ýx8å ƒ¿‰xˆ£ã¶Ôÿ'ˆqeÀg1oìÝHçv½÷Ÿ÷Ãä&æS~Æõzg¡¾tôz6­]§ÇëOñY//À1ùË„KÌ.ŽÆ{6<á+·Ï^Ï{âñ»÷Ãë¡~ïõOÔ—h—.èp=Jó5.WN¯ãîú»y©¹ÝOþ:Ù5êq:¼þ\g{û!o9êÇfxÎî§øÔ¾µ7¢wfå7ízZï5³ú¸ŸpÏ%ÿÜ›÷ï囿¡\êŠ?¬#?”pcÂÇaØÈîÏìzX×ó$.ópœÙ1Ùi‹OÚïølöÆ÷ž~áçßý¾þ¾åï'N?û]w%Oü•åï¯Öù?¥ç8rýiÕÃsª·ø¡ëäùåÙ¿Sí¾¨vïèœöt•Âè°òuýù!ÝE‡Õ³ØÓéÇSÆ'çåis¿îtz}jg±w8®§oÛ8p|Úž·qs¾OX?é÷Eëå8ª¿‹Ù^Æ­ÚƒŽÛöü;§œè¢ßUïm“äÃés9a\8Þ6~¾nG·í9Wkwd÷¨~¤qpy÷úC{#»çö?ô:WïÑyÿâ·ËŸO*ox"ú—§Ç‡í•\ª=ʛ޺_éõÑo——ÇëC¾×ùCA®}Ü9×s>n#º\~®=Ð<9Ÿ9‡¿­ÿWŒ.—ç«·›ôïb‡.Ú½b|£^®{Îßd?¼?ÔëöÁí²Ë1÷½ß.o´KùD‡áܮܸ}syõþݱþ¹=s>ûxxûm|h—ëÏÚx§z¼ý¤7Ôãô¹½á:t$¾ù8¹\ú}§ÏùçýóúœÎÔï¯Ëo¸ßµ·®gfßGýp{‘äøœÑçãäGʹ~ä'Ñé|q~&~øóÉÿ›\D{œäÁûåvÀõ%õ;õÏŸsúh;|ÑÚ£ßnÏy>ùžßô~ôä Ñý¸ÑçÏ»wüDé˹@Gâ»û§$‡~=ù—Þ8[ùŠÛ½ŸI8—<ìOâ_ÂwÎÏ$ç)ÿðˆî»q¹ ä2õÏéôz;z5ŠK7ú¼½¤Ÿ´wgu=Úy¯oj;.—œ»¿õþsîty=Æ÷_\ïM®Fù4—o§ÿ¶µŸì†·k~}ñz{^ãoùÅ~k—qêÙÏC{ý9÷?ÉŽ:{Üúçãâòfõêéµïòmò±8Ò±3g§|øèºëgÈ›Ôú Êy^ËùïǤnß’]MqOȯƼšûUoÇës¹c¼\è¿×ïùD—›åoêº?çüs|nímå­ÑçQRœžúííûø{?¼o'Í·¤qÖøfr4ÊwaŸÈ³Ò~Ê×rîýÖxŸµó$ŸNWÂw·ƒœÜ1þ`ÿ^4ú<ßÛËOûuèóçßøOø™p`Ê“:xN×Gv Ñ™äËÇÑùFÿ\¼}ÊÁg×#{œ¯.Þ`·¯[=&_£ù®¯ù§zÝ.¸_IùÞ|œ·kö²ò©©]øçý@¿ÐäÉûev¤æ{]ŽÒ¸Ü¶zþq»VóI<ÏÑýÅ󯝩ò¬r5Oåv%á?§çºÑïôž.ßSç¸ß/ùçÝθ^y{\w}p9¤¼ã»Ÿ§87õ3ä!³a?FvÇäz$'InÎÏ£¤yZ—ÛÄÿ^Þ8Ô;Ê[%âòçãèúìˆû—KoÏù™î§ñ‡Î$ǽñòs§½>«§g×»ù–”‡u9 v4Îû¹ßñþùŽv‚rè[П=V>Ž‹÷ÇíËÇ_O¾ôôÇå{•âÎ$¯n?:åz~`—Pß#ýrÚÕ'/yð~9Ž÷z“Òžû¹DŸËÛ×Ô¾«óÙãé0n‹™ñ;ýy¨ßN¯?ϱ“ŽþÚùëöÚåË툷§ûž?Jþ¨ø—äŽö WŒâÿäï©ÿ‰w‚܇ùâïhß}<\Ž]~ÒüËí`7žïÌÏÇø¸˜¿‰ö7É¥ßwùé»÷»ÇW—s·/ÎwçGOoÒæ¹„zãæãíõûxX|å&ô3ÆÎO—ŽlÏ®HãéüNë/NÔoï¿çͽ^oâ£û«;ÛëCŒ{Œî®¿Jv6ùyo']Oü£Ü9ÓçÔ¯¤~¾S½pºmœ¢p>Ò®ó/Ù ÷wÆŸn~ÝÛKG쟷çtÚy”«À§ˆ3ÎÙów¬þžÝt}t¾z<æö+Í@O/¯ÉsÏZûÆ·‘ü¹^8ÿSÞ¡ÇÿçC;Þ®Hõ$;›ôÂô7®ÃH|y*ÄKSéHýOãçòdüJñF7oä8ÈøÓµG~Üi?“O|Hãéç6®#»áó–=939Ž8(•Ký戧õF6¿^óÙ»lü{vÄñ¯Ë{÷d§{óñÞ^jÇímâ‡óÁë÷ùfï_Ê›õæ¯ÂýÅ,ð§Ç_ÿ0³v¹þ%Æ—°Îcyÿœ§7Ånç¹N^Ùô,Æ9SåÀ׉Ý6ú\Ó:ªŽýr{q…ëg°g#;{{È—Å)]Oë5±z’ü˜}w}ˆ8îJàŸÛ#‹g'·¯Å ããtÛÚ±ñ(9OùÏ$oN7çW†tð’›ÉAÄI½÷%¨×ìTwþù /Ïé‰ë=>þþ~f²Ûª¿Æåy;vìV/¾NýŒršÚuý¶~Åúzë_“¿uùëÙu·S¾nz%¯‹/Z×¾Œz²×è òå|ÙëÖÿ„LO'çãŽÅ«¯>.¿6~ wDyJã}ÝøÅøÀÛÃç«ÜÝ÷uˆǸ\]4þ¹=ìñÙûÓ“ã¤Ïœ;~J8Áí?Ï»ü»ßIô†uÀ£çzôØsºþC÷£ÏéHër)oòí/ëØµ^v13þ%{ïvƒ~›Ÿˆ~ÏÎkþ2å Œî—pLvÚåú]^Ò:-ÆåŽÕðàH>:ö>â:—óäo¬)Îëå‰j<:ëìêý,“»ÄÇØn²×éèrè|è¼gí»âS—^œáåýýÖ©£7··¿ÔßñcInºrçôß¶þM·˜·éÅ©ÉN~"§1?rÑÆÿúÿ ߯ø(ñ9ù9·ã´ã|îcÒ§ßq›á½8n.ÿ½¸ÌÇÙâ‰ä/Rû)*áxçKò#W>µ}Œzm~¶‹'xÎëç~x¿jT¯ã’©ö3Õgô-fßNwàÄ®>nÄ ngÒþOÉ^9]îÿ;ø}ToÂáFGן$¼ûBè§ãðïPOƒ¨7®Ÿ)®šÊç'GìG’Ÿ ];ò¸Õãô»ß ëÚc¿ŒÏ›F} œê_ÌtnñóÈ_¸Ü¹¼›¿Ùñ^ÞÕÇ;Øÿ¨wÖ¯‘ß÷¸§—oòqöøÆõËýö•ŽÝHrëãWãüˆCòN·?çòêãg{¹ŒzLÿ;ùå„Wãú ×ÓÛå#ð9âA×]¯} ƒ½íâ‹+Æ“×Ïsôñr9 òÖÕ;ç«ãï;ÃûUA’¿Hy¬Ô¯4^¾¤Û¥À‡Ñ8Ü^Oó\5ÞIÎ’ž%ú¡Ëùèà:|væ|¼bräõÙyôŒç9Óƒú{ÚsýâèòqÅè4}ÉËÏœÑÓχ¦öNêáþÅíÇßùžðAW^:qMl'ÙW—?ïo/oäø»c?bÜøåÕèNzå6ñ©ÃOûºë;¬\Wî\BÿF|P{•âz'?Ͻi\L¯ãº¡oFûv$:l³íÇuTñk4.½õ´Þ~°w1ÎKtyÙE£4^Ðåí¦#õ§ömœ.‹úâõ¥ñJy"çgh¯ëÿì¹È·4Éþ¥ü‰ë··cüKýÉ«Ÿ»ö~¤þ¤q»²=ÿGv;é¥ÓgíFûíýðz½=×»D—_7yïò­—'ñqöx"á§'à‰:¿9z¼A½I¾“øuêÄøïvÌûëõ¸öäÔŽ±žþLýMrâõ‡çj=Sâo§¾(‡=½ù§žÝÉ•—wyOt$}åâ™ÉvÏüWêgÂùÑ¥|²ñ+â ”ÏHy×dGŒÉÅ~$ÿ’Æ»'w®o”{6ø}o¿ã#.py°qŒ|2ú»øÃå7É‹óÁùÚMüù‰Äw§/k8vãÎ;ÄñêɉÓÛÃ÷”s¹¡=ÏO¸Ü%½p¹qûyÛÆÓùâG—›ž%>û1áBêñxB÷³@çf'Gó¶>Îü}OÏŽŠùz»>7§ßå9ù¹4ß4u½ªç±½Ÿ)ŸìÏ¥|½/^ô+åÓC=½yýÑú'žcˆË“Óãü'. íO§ôñɇ×ÎSþh²]àèýêcÊw›ýK|O|Y|ñžß9¦ùß×ÛxÆxññP¿ÓŸÆÑ¯»<ýŽãéãÂ9|v»Û{ïãŠñ#Ø?’ü1nÇÂx‡y¤Q=½éÊqXwó¥=}KóT=z(wNõÞüîèWÒó.ÿ~Kô;Ÿ¼>èöù)ŸW·uY]\:q¾¥kÇ\Ï{úåôš=ˆô%yt½žZ_²;Fߨ³·£qëù›Þü@§È¿$/.çi®àèãéãnv¨÷^ÙT¹|Ùö>èkÔ¿`_Fö0ùÝÄ“øn~ݶv¯Ø¹W’§Gç‹™®;]!0Z7ÚË‹†òÌŸDÜä|õúzz11^ˆóéÉÞS¿ûÿž§~Ž·­=ýNùû ‡)ßäø%öïúöü‡K¢«³Þ9Æqÿ“ìqÇ$>;Žõ~&;éúó¼=ç~â¶Gª·#ŸÉ>ÇñŸ×%½ò˜üq\ѳ£‰îdgœOŒGÂgI_}=Oâ¿]=Ÿô+ÅIÞΣ+Ô7ŠÃܯ'¾=¢ò.‡É]Ýü"ånûçͯoßnÌs$¾ù1ù±¦z^Î/JN¿(ïãi×c¾Þùî8Àéó£çÙ(Ñìd°û‘½x&Œg—oþ|·('>>ÎGïOâ«õ'¶ãã`ã5ÒçSx.æÍ/noÏzñt”K··Œ‹÷ÏÏ^NýüñßÇÿ¡?¼?rqüzô‡ñéÖæ¹ãxQ¿éGÌg„q™<ïâíLäcšŠü4ùõ—ûçtt:è×9Ó#¿N{ÉF=òv“}´þtq¿óÙùé~Úåá©íù“Úïʧ_w:½ç†ôõìa*7ÂK<ïqƒËU§¾þ#އóÇì^×Î÷ê t&¿šÊÅñLt¥üèEã«óÍèÕïø“òÞÎD{åÍëKדŸòò©þT>s’3ίs“ë>˜Hïd<èæ˜â¸O…~ô,èCŒûyÎñy’Ç‘=üšäéJ¨'é•ó?ñ'ÍG¸œ8_ÎYÿý¾ûëž|&¾„zâxYûU.ùƒ‰ó‘îž¾8¿½†âw%|œyîYãÓâµxpDòQS½†‹ºqyo|‡~Z¼~{½œªwSå"Ž·Ç-®Ï½ò&ÏÝ8$ÏùÑér½ïÅÎ7—ÿT>Ñ‘ô‡qíùÓ Ÿ=9ßOñúzþÆíU²£FW‚Þ¸‰|s:ݯ9ÿ\O¼|Ѓô€Qý·uß¿‹œäÂñ¢•ñƒûoõäžë‰.¯Çõ-Õß‘Ãn¼ãýöçB\âánüèrü[Çuã —“$ÿaÊh|²qðñ˜ÏMzõ¥çÿ;y»8/–øÑ“çÔŸ€ÃGöÍåÍëKó4ö\7®÷qéø—®Š~Ûçm’¥ñë©ô÷ùíǫ矻v#ÈSô»NW¯Ä·DçDùèæ£Œ¿ÝçÜïN”£n¿oÎëåM?{ö»7þQžýá˜â¥^åÞÛw~¥ñIú‘ø‘Ö—:>qº’¾8]”{v{9ñ-­»öü€C’ë¤?ίԟ·{¹N¿¢œ¦ùIÇ©‰ïNßD}IóvÝ8)ɳëO¯]“·®óúI’·ø·›oLt&½Kõ¸\%ùLãÁ±‡c?N̯uó‰¿=ý´ñv·'Ÿ.'½ùÒ+Ûó=Úÿ`÷zt×õ;¿½ñMÇž¼'¿ïò—ø×‘“Éñí$ŸäÎèLõöâŒ^¼×Í¥%áÊ{ž³§7‰¾+ÛÓõÒûíò¬ö_4ôó]{₞ütíjOŸ òßÓ‡P>Êu¢Ë¯;ÿ½Ý4?Ƨg7ºôÙ¸ÆýRzýw¹ úص—i=í&¿øÙ³›¦W#º~Mzéýñ|d¢³Óß^¼×•N¼?ºžì’÷·'¾žÔû•êM|rzlœ¦âÄɸ”þM;z¶S}‰ö(­ûôvŸ ãîýp9÷þ=oôûyÒƒ$)_–pDâ³×ëõ¤q»ÚññãØ}¿âÙíùÙµë÷±âóž¸Îœç|ÞÃëóz]¯Ò¾ÞI>Ã{^1®vÜÔãG»ý½÷o­ž‘=ëíÛëÇ´þ>ñ-ô{Ô¿gM¾CfD§ÉÙâµ&)¯1U¾œŽ´î1ñ-½ÞÛµûü_ÑÏ:=ÏÛýëÃñžúÞ䈞Ý·ƒIzâþ3õÓèŽòÊ9ò”ð=çõE9á¹Cýö<ûÍŽèíõú{ãÜÃ#ÔããðLʆõ3©ÿ]:m¼G|6¹¯s;õ×õÏîGyèí àúÞ“§ Q¿ŸÁõâ‘îøuä¯kR}Éî¸áW·b<ä|6ùLqwÄ7©þÔÿ`g’ž÷èéâ†tî|ð~ítœ½\àK´ó‰¯É>¤v{øÃñK£H§óÓéLúšöIzîí¤¸(OàW×ÿõìX§þî~þáÞ>)ßòC)ι­r) ñNä—ÓÁùu—ÀÇ$ç)ÞKzã,³«½xlÄ¿ '·{óOöôhä¯|\¹%ÈwÇÎFú=¿Ô4^nW’s}09‰~!øÅˆÓ{òâôÒÎóvLqMª}M~Æùc×'ûm«7Æ].WGûýä(;ŒgÏ®'?ìú“äÁé´|d̯¿ÔÓ—¨ï}ÑFô%»žúŸòK==ƒ>çS’cž÷öýùÄ÷”géé§ë›×ä;Å—1Î÷ñp¼Þ±kÑO]±ç’¼zérß³ß!_—üT/ï§ñs4nN/Ç€[zþ9æUo}÷¹'ï)Ñá«·¿ØcÏy|™p¬ó#´Ûãgôƒ»þÎñQÊ#õèMyŸ„'¼Þ©ö!É‘÷³÷<çŽ;Ò¸st=µv£ÝMüÐõÅ,ô;‡ÉE·ß!?ÒÍ£ôèHòàõŸ3þ_ õ¹¹mõ÷øÞÁ1ž˜ˆûºþ¡·ž'ŧ8m$GÔ{=Ðä}4Þ>nW&Ú•ž¡~ŸGëWFó®—‰o.¯6Ž‹{_Â8OË{qv¯üh<}÷ì\/Ž´ül/>èâL·Ã®O wûñJh¯ƒ·£=I÷{q‚Û3ӳ´®ÁåÅä;­/Šùé$§»’ò#©þ^#Ú7│ÿiÜ_z8»Ÿ»ü…sÇEñ»yÜç:çè7×ý~ÀÓ‘Ïçl\¼ÿNG²«^ÎÛ±qì^‡ž;ÖO®'þÚõî¸ymN~Lqg’+75ÙÓþ­ë/l¯ïÉžvã—W“óQy•k'Çg³7¾w¶þ­ëßú·þ­ëßú·þ­ëßú·þ­??âÕõoýû|þ­å|ý[ÿÖ¿õoýû™ü­ýÎú·þ­ëßú·þ½r~k¿½þ­ëßú÷Óó[Ûן™ßšÏëß+ñ·–Ûõï3ñ[ËÑú·þ­ëßú·þýÌÿ^)þ÷•Bççûïse>Wèø\ù­ù±þ­ëßú·þ­ëßú÷Ùù­qØú·þ­ëßú·þ­Ÿk¿5>YÿÖ¿õoý[ÿÖ¿õoýûìüÖ8lý[ÿÖ¿õï•ñ[Ûëõï§ã·–«i¿5Ÿ>»¿ÿw:>ëñüìþÖü_ÿÖ¿Ÿ}¿µÞ¯?¿Wª\½Ré^ÿÖ¿õïóÿ÷éÚ§µ}[ÿ~6þÖrÿ¹õ{¥ŽÇ+•îõoûßz›¿µü­ëßú·þýôüÖöõ•õ[×ú÷Jþ­å÷3û{¥óósþÏ5zÖ¿õoý[ÿÖ¿õoý[ÿÖ¿WæïsW~®Ðñ¹ú[ógg¿5¿~fŸëëÍ~¦åÁÛ[Ëãú·Ýo-ëßN~kyùüø­Çñ•õ{¹ãµçÏÏß+}\_éô¯¯ìßZþ^¿4Nëñûéù}¾ñõó­?Ÿë¿5¿×¿Ÿ‰ßçz¾÷gºõoý[ÿÖ¿Ÿ®ß+ÅÞ®ÃßšïŸÛ¿ÏÕñù\¥kýûüþ½Òåî•NÿÔßÏ–~~¦ë¸óg×o=¯ìßgzÿðŸ©ßç*]ëßú·þ­ëßÏÎßÚ/­ëßø÷JÑ‹W Ÿ­üYóig¿5¿Ö¿Ïæï³-Ÿíö׿õo»ßg[.?ßÞË]ÿÖ¿WÂï•ö]ÌÏvûëßú÷Jø­õägço=îŸÙßšŸëßú7þ­õâ•õ[×gö·æç+ã÷ù²ó3]ÿZ~׿õïÓÿ­÷kZÿ¶û­Çwý[ÿÖ¿WÊo½žiýûT¿Ÿ-rñ³¥Ÿ/÷·æÏççïóu\?_ûµþíì·–ƒõïåü^rsl6{ã{Ûñ[g³}»Úñ—·¿ß3›í=ÑÎß>›ùâÙl×÷ÍfGo´óÿd6;ü`»÷êÙl÷ÿØÊýüvmÿl¶ÿ×Ìfþàl¶ç»Ûµÿ{ûÿïhïoÏ}m;ÿ•­ŽŸÓž{}ûÿkÛßÞöÿ¿±jïÐ{W$ù/ZÙÖæì5­¾'ZÝÿ¢ÕùÖví Úÿuûû·íÞ¹ö×ÊüÇ­ íÚÏme>Ü®=ÚÊý²v~¶]ÿ]íï¶ó§Wí,»}þQ»þƒíÒ×´ ­Þ}íøÎvý¥vÿGÛ³íúîÿ_ûû}íüµãÿÐî½®=Óþ´ÑºûÛÚõÃíø·ÛsíÙƒËêÿ×ví·¿oj'ßÓþþAûû?ÚßZ¹×îWûÿ¯j}må6šö~åêÞì{Û߆֞ýŸÛÿÿ³ö÷¦ÆƒÆ§Ý7Ûÿ7[»KÿëvÿHãÑ¿j翽 SkôXŸ£‡»ÛsGöÌf§~s£ó›[ç{þJ+ÿ'Û½F﮿·âùñû¹ÿy»Öú~¸å®ÿg»ö­Î/k×ÿ|këJ{î§Z[ml÷´ög¨]ÿPû{v~O+ÿ-íÙŸlm´úw5y8°˜ÍNþóVî¿iÏœ]×Þ?Ýè8ÓúóŸ¶zßÑ®ýîV¦ß®?׎ÿ{£ã]íÿ·ÕÓdáäÿ£ÑñÛ3ϵ¿÷4¾|¤•¿Ôúôõ­î6¶Ç Zß5ìjòr Ñ{ìb£ç“íÙ6nûÛ˜kegíÚñvmÖÆädãݾ_ÔʶþíÿSí¹ÚŽ¿}U×îËí¯ñ~+à©Vþß·¶¾±=»¯ÑÕÊìmÏîý›­Ì_m¿Ç›œíû’öÿ¿Õhh}<õg[¹;í™&GÛxìYòà·Íf'¾aŃÙw¶ëׇ¿°µûbã]koߣ+ÙÛׯþx“cÇ6:«ó‹Zÿe;~áŠ_G?Úê?ÞÚiûÿ—vï¿jþ¥Vϫڱ=3kr|àÿlåÞ×Î_O4Z6z}{ûkúr°×¾FÓѦwûšìÏþe;¶>|Ÿô¢ÉÓÆ·¶Û8íj2~à74zšLïk22ÿòöÿ_׎MO´:v·1?ÑÆz׿i϶vö-í@ãÉ‘vÿT“Û=»¿¥µõk[Ûmìÿ·í™ÿºõ­ÉÜÁ_ѱ·õ÷@Ó‡C¿Ýkrqâ´g ÇoµºZŸv5Þžj¶awÓ…=ËñøËíø¥­Ì/mÏ7y>Ñäôàßim¶¶vµ¿}ÿn%K¾jü7½9ØÆf÷ò^“ëC½]kmÎ~r¥kûߨŽßÑêi2°1oe~l%ƒ{þL+ÛžÛõk4þå¿þ¼vü½ízÓÃ#·‡_öþ­¿¤7Ù?úÀJg–r¾§Ñ·Ñž=Öhµ¶O4Û·»õiß#­ÎÿØÚhòt¬ÉòÉ6¦§Z[ÚXΗ¶ð¬lÙñ&g‡Ÿf­Ž}§ZýMžN¶1:Õøômí~ãõ‘¦;Ç›Ìir¶ï7µöž[ÙœãÿŸV¦éĬÙÂý­¾ÆóM‡4Ù9Üx~â+ºvÝßþßú»¿Éä‰fö·±Þ߯rßÒf5¾lÿ?òÚóMµvNþÎÖ~»wê«ÛõÛízÓ¡]Í^oúµ¯ÙË}KXv·µ1kcz¢é¡6>G›LìnÏÏ¿ªÕÑìå‰6F‡š~íj¼Üÿ]+Y?Öø{ü|»Ö|Çþ6®»ÿ»ÖN“¯£¿µÕÕú>oý=µäÍ—¬ìÍ®¥?ùª•Ži6ïÀ½í¯éç‰F×üdãÿÒ®4Ûw°t¢Ùì#­oûš½;²´?KûÞx´÷߬äu)“ÇþZ{®õcOãǾÿfe/O´±ÙÛê?Ü|Õ‘fŽ´qÞ÷ß·gŽ5u¼•Ûßdæx³ã»[ÿ÷6zf?¾j÷ð’ŸÍïìjcµ·ù‹o\ÉÞ®f]«£ùǃ­ÝýKÙo<Ù߯êP³ ûßÒxÔúº{)›ÍGÍ–6»ÉÙñFב_Ñžo¶ï`³‰'šNjöñp³¯ûÿpãIÓ¹“MG6š,ìotkò¼ÉÚüÝí¹æƒwÿ…VÕO´óVÿ¡ÿ­õãO´ºMÇš?Òx¼«ùˆcÖ]Kûu½Õóº•]Úº£_¶’Ý]wV²:o6~ïRÚñp«c£ñöT“¯Ãÿïv½ÙÛÃËóæÿN6Þœ¼³Ò¡C®ãm\fM†Ž¶ºN4v¼Ù´½mŒ7®µú› ;Üììî‡ÛñwÈGýH+×ìÓ¬×Fã÷Ѧ³foOýñv©ÑªÉü®Vçñ&+{›Ýßê8Òèžµ¶v7Ÿyà Wü;ÒÆõøÒÿ=¾ÒõC¦½Ç[Ÿ÷6º÷ýÅöìÚ_“Õ‡ÚsM/6ÿ4ùÝÕú~¨Ù²ƒ/4Z~×J÷·¶Ž´qÜÛž;Ðxyò;4öM¯w7œq¤•9Öê›·gf­ÌÁ›òm w/Ë49>ÚÆc£Éûžfc÷4{rø­žVÿ±fëw·v7Z½í|£ÙÏ#O6—47º.1U³wû_Ž·vw5Ÿxêt+÷ñÖÞŸ¿ Íf{–~ÿ6Úšß=Ôø±÷]+lv¢µqè7.ÁS+Óx¾ÑäíDÃc{›.môû#­¾f£4Üt¨Ù²£K™\â¬ÒhjüÞ¿ôEí|Öè;ù¯VüX↣ÿS{¦éÙÑfÃ6–ú¼´µí™ƒMŽ´¾ík´ú+\2k~sWÃ{O6,¶§ñè`³Ó§~ÍʦYâ¼Vö`ãÓ¡&‡ûNØýýí¯={ô‹VØjWóë{šœîjøñಭÖöÁ7¯pÉüÿÕΛnìmÏ[b£výäW´zš9ذÁÆó­Þg[;Íçíiôm>í@³…Ç?wµ1ÜÓ0æîöw`©GK|Òäs÷¯_éç±XÙ¸“Mæw7?ÙtïTó§‡ý'›=™7hºv¬Ñ½¯ÙÚCM†ö4y=4_ñpöÝ+ÚŽ7}ÛóÍ+±üÃÊÿh8qWó'M\=³çûV6Ok_ÓÝ#MÇO¶1?ÐìÛþ¥|ÿã•ìXúïæ{7½:Òìùž&÷ûš¬Ì¸•o|Ýßx´»ù†ͧnvðп[áÚ­î“M®ö¼u%«Gš iºshIKãù¾f·5|½¿Ù—½­®mLN~reûv¿´jg£ß¡†ß6Zù]K™o×=ÖÊ5[yôKW¶êhÃ:'›Ηt7hc³ÑÆïØÇZù¥î5›rb‰ËÆ?Öìü‰&;§~›ô¿ùì%vû…í~+s Ùª=—óæŸæÍFm4»¼g©ÍÇo|Õª½=Í'iòp°éÛ¾&s§O5Û³Ñ|óɦ3'~ÝJw–v|£éÁ|‰ïš=>ØÆàP;h~ÿDó£§–xûç­pÏ®ÖçýMæ÷4ö.ñY“­ãK»·”ÿ¦k›]˜µkM.Nþ+?¼¿Å#§–¨Õq¤áÍYóŸ_±ÂÏ{}'¾q…­N¶1;ÖdêÐò~Ã#m\fK{Úì×ÁgÚý%¦j ~x‰Ç›<i|;Üpĉ¥¶ºw7[¸û‘¶œ5]ßÕÆjÞž9°´Ílr{äÝ+=9ù{V>ùà2¶hvô@ÃÏóf뎶:ÿ–•_<ÚüÒF»°Ùƒc˱i>ò`ëÓFÏ}w‡ÚØîmvdÞú|°õ†N-mDóS-¾ÙûÆ•ß<õ5+Ÿx¼ùÁý·WqÈ‘%Þù¡öÌ2†Ú¿Â'Ï7ZÿN4¾nú{ä;V˜ûèþUœpli³š\îm~g_³mû~h…;ö7Ÿlíï_ú‡ÆÓS§VöðÔ÷®b™Z=‡V±Í¬ùó½Ë2MÞw-1âIÅmx…[5›r Ñ{處_ÛÓ0ñÉfN5[w ùï“¿h…‰Žÿ›•?YÆ«‡›oÙÕäèd³ýû®ÜÝxxâ{W<›/u¹Ý;Òút¢Ù†e,wªÙ½M¶÷þØ ƒÌÛq£Éü°Š'–¼9ÒÚ?¾”ýf#N5™·¾]âÅvýH£{ö_®bݶvš-<Ôúv²áºùŸ]Ñ9o¾åÄR=q¯â©/]ù ¥¿=ܰמ¯XÕ±·ù¼6&mü÷¶ñ=ÐìÃñ&ç§+|»·éÖ®÷¯ø:oü8Òüù®v<þgV¾òpkëÈXáÄ¥­ZúÊ=Íœjc¶÷j{æ™U¿6Ú˜n÷6Úlò¾¿ñg£ÃÉe\ÑÊœh¼ß×h=õúÕ8ì{z5Þ{š¾m,mÁR&›]8Úì߉†6–ºÓ°Ô‘e Ó|ùÆ ´¸sc9ÖK|ttÅ¿CYaß#Ífk㼄!K~ìÿ‰•Ìš:ñkVöå`Óµ#ÍíYúÀf5zN¾j%K'þúJö¿´¢åDÓ™£Ínüƒ•½Üó·VvâÄXùÀÆ›}ze÷6ÛóªU?–öédÃ6§¾sË[>׎G/Çâø kžhz¼ÑtnÖîlrvbÙ%öi6zÞÊn|<Ñ|åÉfÛ6š_Ühº½ñçV¾pãµ+¬r¨Û‰öÿCmìNµñ>²”Ó¦?û¿~…ÙN6~îir¹çwjüZ\³ûC² ß²âõÉXÆc-¾8¹Ôí†öߊ†e ¹Ñ|ÒÁ&~ãÊÆÞÅó-~Ùý¿®ìà‘f—O¶q:Ñê;ù£«6–qýž[Æ_ÍÌ—Øo±Êl´19ÑìÁÁßµŠ54]æ^~Þ*–^ÊØÉÆ“cK›ð7VvvÙÖ¡FçÁûîn~rþÐJþ.ùÝlõþGV~øä×JΛϘ7y8Õäåä»’íCÿq…Í—AÌÁ¦ƒ'Ÿ[å)–>æÀó«8ri7ö~b%ÓGZ5þìnã²±››+,xâ¿[ᣣÖM¿O4û¸Ñäïø§ý»•\.cä»ö aŒýÿv%óK½>Þh9Ùh:ò+Ì{¼éϱ¦ã{šlnmíi²zàç¯âêo[ñîDóÏ_³Â³Æ×}MÞÅJ§wµñš5:ÞÚ;ÔdwO“Çc_•_ŽÛ‘åÿÿÅ*çrê÷¯b¼†£O>¼²Û˱ÚXÆ ?¾*¿Ñ|݉fsOüÐÊö,sR‡›}8ÒìÄ‘fƒ4~_Æ-KÜÑ|Ξ¥L5ý>Õúu¤÷Ɇ­Nµ~žl8ñTÓËüV‹|Ç*²ÑÚ=¶ôM÷O5p´ÙîY³WG–z×ìÇ©ÖÇÃ7V÷÷7<·¯a×ù³«øp£ùµY³É'[}'›ÞÎV˜7^mñÓ©f6}'šÜø!ñ·Ñzìw®ìÁÓîYÆÅß½Òƒÿeeö4½ÞÝÆãx‹IŽ6{q´éÚ¼a½ý;\æ™þà*>;ÖÆüP«sO«s“ñyóù‡—yŠe,óKVxäxÓ÷ã m4l}àw¯lÁÒžh<™5|sò¥U<±Äþ‡ÚXüç+L¹ÿÇVqß©FëÁ&“ÿÿ­½eâ—®äsÙ·ƒßÐΛ¿:Üdn¾´EÍ6hÿßÛ°þþ†ë·þiváhãÁ¡æ§-ñsãÍ|)Ë­OG[\¼§ñþ`“™yÓ³C_¿ÂF'›¼œlò´¿õ¡¥>¿e•{˜7L±-cf ÷¼ÕÇ#ÍÆío2y¸Õ±ÿõ«z7þØ*þZæjö6Lrhi[¹_¿ò‡K<~ê¯üü©ß·Â¿GšÎjº¼ÑðÂñ3«xmw›ƒ¿wEßRV-ûמ;µe7–þoi‹7Ú¸œjã¿q}eK–~ýdÃûûŽ®pÉR¿–±×6ž'¯îùOVãr¸á‚·¯0Ï¡ÆÏGWùÒ“­¾“ß®8|¶ÊñÝÕ•¦‡›,œZÆw«nüW+]>³Ñ|ÝÉ_·ŠÁ5\°ñÅÿW{çf[UÞý5gÎÌémε+A%F˜lA£b,õ Dà^¼4±!ŠÁ‚±`A£’Øc%Q¬ˆF,hbì¢1Šј/|ïïý¯=wŸ=§—™3x÷óœgföÙ{­w­õ–ÿ[ÖûÛ|‚¶ÅtÏÊ[¥‹ÀLü"óøØÐŽÍ¯Kö|U6>h˜ÝXùAŒß/ùZÞSñ¦ö %kà9â,-³åù×Ê/Ãn»5ºó?V\wùÞ#v z+?6-K¸ceOÍey½©µ÷Í1|×Ú)û½‚Í·ý‹t¾ÿÊÕ’ó²ÙÁöGþ—>%¹wx?ùŽmÃTEîÛ:­˜ÝhØø–®–¬7m®Úç*¾Ò>\8·f¾síҙıšï àúe›§Ã_-ð™éÌ•w s¯\.­a¶¢ñ ÍX“¸"qíš=WÝÇlòºb2V2Ú« »`‡š&—E“ƒ–­gûoëY¾P¶¼a²Qµ6›ç¨d»úÅW°³uÃm£¯az¡™O¬ÜÝÆÉÏ)–Æ\4ÛÓz–Ö”ø}Þø öÅZª¶ÞoŠî|ù%ù ­÷Ë?Zx½|•ö QwÍ-ÖþZ§¶Ù¼1Yé$éiø¹`m7J’ë¦ñJóíj ÑÄ¿1?­h<Ã'¹¥lNÃ|ã¦é²%r‡‰ÿá×ek+oý7ìÞÊOe›Ä2MÆÚ6—áÃZk°nÛtRî;¢káÕ²÷îã¡Óž+^\4»[ÝK¸±eº¾`£mº§|¦ð¿—^±õm›­Ý _¿i:§ió_'¦ÌÛ<-÷°±7‰»˜OÕ6þ+½+×E<ññu0{ÓÂß°ui/ì7š[¯“+ÜVþ0þ&cX:E+\£˜nÃôKýÅòÕª—FÙfŽÏŽël<Þ´ù^@/Ú{uÖótáuæ¨e:må ÒïM{¯jú{1Ýž;H4#?m£³½Ex‚u#w²bø¤i˜§ò#é2rÙ”}_>M|³dr]-)V†oÞ2½Œ¯”¿tcñ(Ñ®ÄO©bk>­˜(:¤i¾PëHa.ð¾lËìY㣒…¼ÙÁšÍMë"{Öæ®¶$¿xŰlÕÖµy Å›è…[k~ÛfÿV¾(^†o‰ï2fx`ùuZðSóÅPÊÏÓ®¼ÊÚo©zAz½v¦° ¹C‹á׎X¹ b?Öœ|Ü~Â|àú%bÆ—K§5Lw¬˜î¯ü±òV¤ýÈáà¿v¹0Yÿ¡ü× þ‹é‡–ÍmólÍ%ü„ýj]#~ÂÍÿ[ľωºÖÚ'—Ð8Z¼ËºŒw[Äx.Xy†h©›<â› cÌk }më·hkÒzæ]–Eök¦‹*— /‚‰Ú6•Ë­’KýÙ­Ö+sª“ÁÖÞ[ô7£ÿ^k˦àGÔðU§5]ù¸æžØ v›ù$žæ|jýWŒ¶N…_‹&0xÉèl™.]Þ_±¡°>rd8·½ñUÅîÈG¹þz°ôPƒX5yœ+¤³ñÂ’bÈðÇÊÃ$‡àYôëÒÙ’_rõ÷(¶µB¬è}Ò»Ä É©`É—7ìÙü/”ó­ì®öøÉpùZ+|üèªÉYËÖ¼ýÉv=~­" {híáq°%þ6 H\©fëVy™|™@|üza¦öÒæ¤nØ´þ6éÓŠñlÉÖºúMñûŽŠ+‘ÏCßàC/^!ƒœ˜ì>ªõ(b×>«Ü»ç#/R|³môMïT>)üV?3ÆólŽ›¯¿!7¥;iŽ ‰IÃôÄJE|„î^¾Hrà¸ÈÚlÞ([vÛßÀË5»/‹²lsW7Ú›T0ÙmýŽì)ï!e³Û¥«”@îß©\KÕÖ£~7åõjè°ƒ)ÿ­øµq¨ü°Æ_(.Ƹ øÀfC–̦å~çÑäŒW2ÛÐB‡!=ÝÄ×2j狟Û_–­Â¿¬àW4ýT4½d8¦ô@á|GÇ:†ÍË&Û Ãµõ»+OUú¹ìºëÓñ%³Så ”çw;ø0aô@Ì|ÍkÎ> ×Zxë¢Îüoå_Zo‹>­Cãå÷‰×-Ý ^ï/b·%,Eœ¢iv`ioÅmÚE|ù ¨wî#¹‡6ޏë8ÍKÝd´þkñÞ’‰¼Šc;üÝWé¡_sgi+öwõá òî­Ÿ+®V5¹ÆG!o³pzä%›«Ò[´N+P»5p±ù™9â6¬‡£m<Õ³ÿFz¥ô.Å\[f—á#[Û:ù-{¦`|PÏËVbËËÄà±÷÷ŽÆß©Øœ­¯7Εüb3é9(`“Lîšð5úа^þ&Ê .–ûœò®•#D_ ?fú£òrñÖ²ÑÐ$Þn:¸nïåÀ£Fûa}¶É·}•×X›GE»dö¬uóH·aJâ4ÄÎëE;ø»òù«f+F{õ7¢­òű­Å×Jw/á{ý{Ä4§(V Î`fÊøÉ7“ž#¿á®ÇQ’™<§!µÈòÊAQ'44Ç-ìƒÉj=æ–ˆ—X{KPl·q[a[0ñ⽤o‰?Ó^ÝänÁæfÊV­5¹nǃðÈ_Êž![Øã•‚æ¸t¾ü‡œ­Iøº°_¸­ê„À´5kgÑø6_иŠmÙoäÅuåƒ%wîÞ'Æ»ó²‡èæ>À9’Gž¯¢wï&^Æ¿uY:Xxl¿AÌ€œõM³UÅmŠ…/?Yµ(í«4g>ŸÔù¼^¹i|ªÒ£…µ 9&7€<6NRžºñWââ«„ ÛfŸ©9€ÏÀÃíÛ‰·+¿¿G úµöBÙnbeK·À×%§Fì­õ)Ù×Å·ªÝÊ%²ÇžK!6È;¦—ŠºÂæì˜;IëU}JÔeε!¯]7£6¢…î g´6yeQø‚D_Îät…˜Øç„1ÃÙÊ¥¸f§ñoÂ%ÌKÓÚmÞS:ÛuŠÍGójaÌ‚ÑÙBGÖ]4?ºhú¹iºgÁôS¸»äµ¸OÄÄ¿î*Y€¯‘pÄÂAŠ#ümäûC=I“8=µmåÍð_É©@7:ýVù¶t^õ1ZcrqÎsõ =Y˜„òŠé±ÊTÅÚ“Ÿ¬{©æ±m6øµ×˜ —¯¼À§mãµ"¹îÿ'lW´vVhãiò)±ßŽÙð៬|¹Ç;¾+žmÙ—¯U ±bµ Çwc<äÇ<ïvô µ;`Š…G«í6>–­_í ’±–ùâFÊ?¥¾X5¸;omU?/[Ü‘ÿ$¶Ò6¾¯î),ÚzƒÑWrÅäÌ]´ö©GsürGáÛ僅ß‹þgEzÛ}ÓÉ+‘­^£š¥âõŠ-B3º•øpî©ZÇþ`…²x)ÀLJËÏ?æX¯ÿŽshv~Ñt×›ýgxv‹Öªfc-~F¹…&|dz£uçhÿñ¥áe|:pá‹ÿwœoz1¼1x ÕùroÉx’7 †ÅÇÊ:}5£‹ø{ãùâ9ükðu;Ä;™CÆ»ôµh÷¾¨xJñ¥²cµj«is[½‹ü¨¶a»¢ù åÃ4'Mò]Dýh6Ï1´ÑT7LXºCÔ9F+1íò¯…ÜÿþY´ÉÖx“t}Ã|Œv´ùà%l_Ëæ¸zé r9ðrÓd{¾3:K?‘d-òø…§+®Cómš—JGà¯/ MþÈ[Ôñ]?®÷Às• µŽ•O¨V¬Hm8 ü1åð©=¶cü_x«l9ëÒú^ç«e«òoV$&ÒB÷ؼ ç—o1–ÍmÕži½Yyƒ¢asj––_&]Ú2lH>©n|W<]˜ÐóP׫ž¥JáaÆ%r™ÇC´À›W-QóVÂðýöÅd¼Ñ‚Éñ‚ùRùç+>íqJ›óšñDÛìò‚éÖò;• ÀV×m^ŠÆ[­Fû¿§ò‘ø6í8ôú¢ø}ëÓæ§hk³ïP“ƒ¸DsE^ NÞñ¢—õ'·†?LÝLËpXÓôÒãáúa‘wÍ?#c_ ïÒlj—Ê®4¾&™ÂgkØ8‹U |U§~†8ôícmØüPH-ñär™ËïJ/½O1„%|¸‹å#,^!½¶`ÔkÍ®W>¯ùt™~—t6|AŒßÚ×¶ÿ2Ù=púòKµNÈlÁè*~[X†œ¥Ë¼Í‹ûFÄÝ.rlkT0þ¦Ð팹fkðkñ ¿SBܺòØ7ZçilÔG¨ x£èeØÖìý¿”Ží­wÛçÂÝ…‹¶áCZôJÑž§¾?²JLÇd¡iëÜø/É*¹Í÷°¨«ñÛï¥vJÆs•¤÷Kð¤µá8ü@ýÜò8É9;|.b\­_¯">Œ×–¯F-\ñPMz¸ekZÿ¨ô˜¬I¾ï|ÙUâÌkƒù7»·ðYÙâzä`Ïep»É|õK’æGëíK×8ËØ0ŠñWù«¢8Lã,á%d‹cŒ‰Í3þ«“¿ú‰td;Ê=x“˜ßâJ—ŸöçÉ)}Ýþ´Ö¤òû’æOÍ„ÇP‰¡añó÷—Π†²üVщo‡^­šéybò½·ÑúRÿ Vj>M9FžWGžy´ÖÎùÜž]²õic›ïóÎäóȧÚ;%ìÈJ´»û)¾B| nº©e²Ñ„¯–½¡µö©ØçTOÌ|ð\1Ö¾`ÀYÄ4Á‚-tþ&yB³Å­–j/–Ÿ*¾ Ý4? tÿÈc¶6Mûž\cóOdÇšÈ~ùѲƒž¯¿ŸìsU5=ÛúcÙSâj-󟪼ƒ.;Pº–ñmYT¿Š¾&FøSÅ—VN“ÿ„üz<•<ÅUÂ"Ô·¡'#öþPËCç(ž"û‹^÷<Èý%ïžs³y(êfa5ŸÙ4:jFOó~—.?ÈË»ûÀgj¾-úÏØfô0xÚìyo}¨I†gñÏ©±ûÖ°—¦×ÊÖ1BhGJ_Š1ƒ¿¯ðŸ¤ˆõ#¡Öt‘Øã6éHrH`æŠñDõ„ó›Uì¦fóÔ &q¾ô*¼]´yôŒÙøÂe/ËÆ{Ä„ð¥‹ÿ+=Șñ=ÁNðJã'â7| dÑãžä)*š3lkÏ}Ö›Kþ’˜üZ¢¦ª&ûçú1¯µfoM£©Oxê_ÑG&£|7|*ü`û=ÿDí‹´ž#ÿ¡n:=O­…µ½ð]ñZñ¼hë‘}ÆŸ$àÅ;Ë~ƒ#VªÒ¯äêÀV`;üù†=[=Oz]R"wûû h¶¾ø$Õ†â3bêßUL€ýˆðlýëÂâø%Ô¥W­=°PËüoö«±ƒükÚŽ>cÍtGÅü[|ôšÙ‡¥WjîÐÔ´ £ñƒGÞW|ã5ˆÆ[y›ËšáÕæíÄ3^‡q½ø“ñ6¿¦½?è$bW¾Ïæ.Úÿ#NM|Lçùïr¬Q2¢q™|Qø–±VmLy£¡ùƨ+>¨z9dÀët®T­RÃ|àrp®tö~ òðaéô˜‘œ!òÀÞô<†_P»Dëìñ„;+wŠîõÚ”ÏÈ>?n¼Dãa-?Á1øâ/R°õgòáàѪéžÅŸéžûV7W¼ è13ã©ÜK¢ ´~–ÁÚø× Ã£[ÁÄÛ<þNÌÈä ü܈qC0.uØEô‚Étñ–²¥¼t²AüÓÇfs¿ß_¼L|Äs/èG|¥gk>Ý'{ÆÜƒ×ŠÆ¯ˆfð0ü_EîΗ\‘ó†ˆÇ3çð¿Ëñíc ÿQqM­ïÒƒÔ7~sK nãfÒ‘eÆdcá Ù“¦­yÕ|êÒ¢ì)¶©bú¡Œž.ûÖºi´Æµß‘.—°ŸÔq„­Oé ñjý—ŠcC{Ö<Ÿõ­ÇÂ/µGÂãsÌý[•k©UÞ$ÝÉX<æm~£ûÿªºöâ¥1ÞöÉ$ùEƬa«‰½R—[y«òu؉šÉ¹çlˆñϺ^ëÕx‡ô§×k©}&Ä2ŸÇT¾ã$ÆoÍ}%ãØþÂTßN<ÔõÓßKß³'8¼$âß ›C‘ù·–ÁEw6|%òªÙ“ÂÝ„¡È͸±U¾—cÛ??nßþFmzírþtÙn|¡Æw$Ô7ÌNÖ¾l¿“¯¾DÍ•†x𼿛A5Wȶºo@^Ä֬Ʒ~ªo×ØÐ«Õ_JÖˆ–Ì®UÿNïú¾!“¥*1ä£Ôn,b럃‡±•ÄÏ.•þ'&7ÿÀë@Rü?#ÜG¹¡Ò÷ÅeÃ䵛ǽlÄ1ÎЧñVÙ!Þ]4<œ3þa¿køká™Å›«Þ˜õ^|¦rÚ`qöÊ“hÚœ/“WøI”1ð™Ù\jïÙïç1 b1—k ÏÕZO]ü¨õ]xä÷B\Sësùåyûh\Ãæ! oLnÀü^Ï´§ñ ð@éYä<à‹ZElÓKÄ ¹‹¥KÁœÄµ™-^ù±ì û°»Ð¼h2W³ûßèV²+ f‹+`IxÚtuKÄ=ùº˜ßÀç=A¼XûŽÖ]Ä~Ú•s¢_w•ü÷yÉUoÕÞi0ÝNlÂçñQòЕü?U¶¾¯¿Rx ©l¨ïðÐa²9ÔH±¦øEÕyñ‘Q¶Ë²áŽc¾)߈0ö¯ö:Äk£ß í·Sܶ@Nweœ[˶ֈáM9båÄ^,<€ý!Vξ×Éà;_&z8› ùa¥=Cü„|•áÚÊ-eûš‹šgtº¨ývÉ üRÀw=I±Eöé ³ë_’­ðz4ì²éAÏE›¬ÇéòO=òÏÂŽËì½â?EÿœØó4ï@Mü·»O– 0Ÿà|ÿŠ=Sú‘j ¨ð:’?‰ºßx­|¤lˆãô÷ÊÆøþ,³ÇÅF¹{›ô¥ÇÅ+Ý Ïóþâ~º·øyÙ¯#=3Útxlÿñ¹hì4ºì‚_BìÌé~¼{¤p_Û|ÊÂM#O'9Ã+ž@^¾£]ê:À¨ bÛØ°Ãec<–´3®EYsÕº“ìJãeñÞ ÅÿäÛ&_ìqðxã¿P¶Å^_oï!ZsÇ毵•ßÌå—F=ar—³÷ƛԖbÜþác?[ëV*a› '×ÐUWÇù1›à5Œèè‹å繯ð‚X·zºêc9/Sã‡O¶hsQ¶±7lÌ쨙]ò<ü'4wìáÅ‘ñ|«ñSåXéš%dõSñ2í}a>‘;×ù&ÛìgïžÛÙã´7„¡ÜÛåãº\©µX ¾/>Xc¥Žšø y?ö:Q]ÿ°hi|D5~M£ìY´ï8wÂ÷C쯚ã b&Oµmz ó„ù*ãï.ÆœÙNõîF/á â¿!§õÏk?eÕÖ¯dï6Ÿ!ûäãÝ¡1PY8TõmÎ'$ÛÌþê}ÙŸK| :¹[°6½~z?/—P 螪‘& F¦NŽ5¾&ú„9¦ž‚wÉR—Eì¬t¨tn5按ÛÕ£/í¹…ß(vÔzG”0æUq O’M]"nc:’ýÑèRrJ^Ó½&¯BˆØ;zàÊ· ;šø1¯Ö¾{×7ß•<£3±Ä=ÃÂlµ+¤¯›’l°Îó5GˆÇÁ“ÍÓDWîµá5c6nö?°Ö³tLÓì:ç'ÁgØòÔCyÜì}‘N— w¿Ûìê¢Ù Å¯Iø~x¿&]Ìš±Ÿªö–hwL€=B{>± øã쥫›\/’Äó¾¿ëÜ(7ôKlÙôôÒI7iØ·ïØO® ,m~99ðAÕðÑÒ¿H·¯‚É]¿Õä3±¯Éã㯕/K~ÊãE๛j¾ÜξEú€ø?ñ€e¾{³æ»ö^ùîÄüy§rø¥¯Ëßs—ìƒÁ›ÆÓÄM‹î'ÿØÏ¢°5i}.ò1ø¬&>ðü ñF›§ÖçÕ&5b~FΡÒ1>·Wþ‹³\ˆuTéçF¾˜Í/±XÏ©¡~Ñ{ø å/KŸS+ɾ”ús"¯ÝO¶ÌcÈøWfo©5¯›ÎZØ¡¾=O6?]úÝ^™l濬ŸØxbIMrý;4·ÅŸ©-p õ$žc8@|Úx‚äÑsíO“^*T2ï¸í™â¿&rówq/ñ=´¿ƒ$ñ{ô}ÑhkÁÏÆã-ržg;ƒ™©_a/¸˜üØÝñ‹éú1Ê’l·ûÝä? Ч¹þ6ìNL±’³!?ë˜åtùjàkλñý¥wÑq^ž¿û¤r²eÓ Ë/_, Ÿ ?g}Q÷ÛjH¯¹¿Þ”œ;æzœæœüOÕ…¯|*¸o¿R95 åUGÃa!SÄûœ½.E¿ž«8›Ç>‡ O“nqgXhÁxmåí}bR{ÄXÞ¹Z'?ƒÌPQ|°r/a-jã q…òe§À¥ÏHçSÏÿT·K¨‰jÙz•WжîµÃdó¼¶ê;Q—ßJ±÷«î}õ#¤o’º<Ήßû¾»H^¡Ë±Ú“õ¬c’tjí—Ѷƒ—%Gƒ¹Zó†íà\ ßÀ>[/EùÄ Úxð£ñqƒøØñTûC tÕx­ýý¾è÷Xœ­öMíµ@¿_jîÔ8ÙÓXy¯l‡ÇˆY›Ì“C6‰áPãíÏÿÍ.ëµ€fûwï:¯ž&¹Á_©]븾Vk*¨qp{×Flš¬>yãÎÚ‡>¥´ŽÝg/D;‡Ž{†dÐ÷¯› P?Ⱦ]¯õú°dÒsØ¿Ÿƒ5©5hß_ö“ñ×÷‘¼ƒï‰UÞ¥vÙÏëøÓìRóÊ'5ß ù¡Ýã@·V®ƒÓY[üXl‡ïùB_ÝUkâ8ðŒhÿÿO{üÈŸ0×ø¾sïàgHÃ<®hþuÔ÷B‡ó2iüŠÿƒL’SA?–_.¿šý4µç«>¡`¼QÛOú£„þš÷éÆÒu>HÝx¹°,ÛÓþžæÆ÷,Ù\×lÞÊÄ ÀG·‰u!õc„ïÿ]:cÑè¦N ½=àEðJýñ Åñ=¦eï/ñºÉyí©ÊxíF[z±H¼•8Îã"ãäÝ=¤ƒ8k¯qgÕÚ’Û¯š-Y¼.î»À熿_±ØM$g‹¯ÕÞáæ™²Ãðbíåûpû‡Š§]oš¼UM·¶Þ¤õi|Rr€Î/™Ý ¶¥D\è'ÂGž~ø“œUŽú‹XÚÞ›×#y•tÃ"1°ˆñD €M¿FyEΖk ÃÍ·lý¥Ëeã*È~9ês„#Ꟍøá­]¸J²ä6`¿Øÿ1²y 0ã^ê¿ö\}ß,KÆÀW^_qÔ&;ËÄÅÈ|Okëµô¿/]ìõ¿ï“½‚‰­4Þ ~!^³`x¨‰}l»Ÿáð(ÕzÑ×]’“mÉ&Ð5n‡m^È×;6¼Qǃã S/]"9a\g»ºEÜ‹oz‚}åu{ó|“aÛÆbf§¼~Û½ÆDfýú1™—”Ò6Mç.0Ÿ?Rÿ^ÿbc\4#¦L=›ããe| lÞ§…GJf{—Ê.’û!7èg\)û¿Ã9ž³¸A6²lr]0Ý]c}ÛäûÇË&QÇB‚c×_JŸ²ç«¼U}WÁÊÆ×åJÖÓÿmÅ7âÿ'̃­iÕ¥k_m¾7õÈxÕuŠÅvö2ÔÞmðQÒEĺÑkOx›jW<®orUûX´)öŒ×Ý›~¬¢Ã‰Å=Ou—Ä£9¿Ãëgo«X±t2óUý#É–ûÀ‡ ›Õ^';áõ{[4ÿ rß÷”½ôý,oVM;ãÅ÷©0wÿç-èŒ'öè•ÍÖmœ?ú¨øÙ÷HÞSºŸxgá¤Ø“[Ü;Æ]?£¸šÇŠáÿ#ÄÇ‹_~Ãï)}[:Ìká¯ò†/Æ#•+…m¨—E9·§þ ÙÎ~ùÕã\‰8žœë™ øìG»\}/%Îþb÷;Àèô—Êv»Mþx:ß&üióYiKßQŸçûÈŽ‰ºëSÊ?/Ôµï°tžæ9¨)áýÚ»œ)CŽ ½^Cç-ÌErL®‹Ï–ù/ùçHïc¨ôz6ü{¯}þðÅ÷ž"¶C-> Ûä_Ľjö~ÙæeᙺVpÞ6œÜ<\þ"‹úóæQQ‡U¼B¼b©.û¿ÀŽÜ\º„s‰±^.~p;†îo?F¹ÿB®²tõ:Äæ5÷‡â}ô|Ñ>[ ï éDp ×1šÅžNl"8þ$n‹Íõ½üö¾ôòuZ'pùþ¦ÙöFUö—sT=žqŽð¹GÏÛý™j£È¡Á5ä4mJ?“~÷‚YÙ;‰^'VÊþH|DßKüsͽ纟íÀ±’7Λ#­î×l§d«f˜ŽXˆÇŒ>"AŸqög· ?Áƒè›Ö¥Š À~œ½¾HžèÞÁ÷&yÍÍ=#ŸÜC÷9¹ú÷ÒØXæ?i¿AñYÎ'Á÷ Îòº6Ã+^¯júºI,9ưmè*jÒ9¯ˆ³½á!ßû÷~Éã5)ÏQãìû^ÁÍUáIß“õõÍ´¢µÏyR^³w‰ô?z›ó‘9O£®6,âçAü§lçŸ9¦ú©t@¸Zë”'7òw’]ÆNluy3Wÿoéa³§‹¯ðŸJðµñ~Ó|Εô}Ðiö¡j|W&v€ïù„3μîõ«Â[ÄÒàã…ý4>â•è#ÇŒËÂ@îcŸ?CúÃ÷fÕ¥°Óì«Î®­ÞåüOjè°ž{y¸ìrìç~á§‚ LW”L?Öð‡‰uØ|· —ÇùL.‰Ï.à7—nË~¡X‹ÚH×;'´`Ÿ’é£%dÆú®ðìÑ0ÝÁžæÀñèíe/¼¾ùqZs?3ò{²óydÿƒÒKð{}ˆÝà—»ßþÜàû•9/y«¿[<=Ôˆºî¼iÔA¯T Ôc™{Êöµ ÛéöØú¬þ»æ™ýœä\ßï%[@&¶Ú÷%ÞÈN¶…<‚çº)Ùk0Äë*}þ•Úgd3ý¬>æèyÒõØ¿Äw Vªns vFÇ¿#¼‰¬‘_#g“ƒ¦uÄ·aý[Ø£?WÞÃÏÓü@ÌbG¼Á0ÕÂá#ú»ˆŸQ8Lô‰á_ë©%bg&Ç ‡('š»Ðž7¹ Ø1Ã[ä´Ø3ËYøáÔãV÷Ñ\»ÀáºjhX/þ?ñÎ'ç $þ—…Ç‹ /ÕðEÔÚ{Žßä3nÑ7Ô7xþýâaöñWðAmœX?TsÕ¼N|FmŒ×¡>R¶‰X<û®ýŒô›ÍE¸wÔ¡¦_8?‚º b á©‘o^¥<6=Ø:ô.qIp6g,ûÞ&ð€é¿*q÷ï ?øÚcÃoz»øua/éò/¾wÈìc“<Ä×4wC;(ÎïŸFܶ$ýïu>GÊî7È= œL½–ŸÍx Ùê£L„CŽ˜³ñg5³¦Õ—©îÚý9›çªñŠ×"ÿ±lgÍÚ¤6×÷+|Q²à8à=‘ÿêÖu§;½H¾g©9v19ö½&& œÙZ_Ž6åÄh 3R÷àqmÓO—oI ‹©åB>±Ïž·ÿx„ç¸ß| |†qO³a~ühXÍsY/–›j>S>ik?µ‡ÞE_q6rþÕ¸ ôZïOhüMãSê²9„ºbx|9 þg aŽk²`KÎâwlClÇæ§ö¨ˆÃ&|Åùçž8Sõ ¾¯ôª(w†±¼.×ôõ½ÔTà‘¯ˆ?Àøz¾¿Ãæ§p¾ô'úß÷T™®hâ×0¿*;ïyÛk]ñ±|ö«eóK•°z&ŠŸ©WSL§Q®ôú.£³z7Éçü{­ûuªëôø¦é°âÍ¥Ÿ} Î?[6ÞÃOÀNÕÁÄ×êÿRO%Ïâ{°°Õ¦o¯îrØ[V5:¨…Ìÿmð8k-æq½æü«Qæ*ÛFL ¹£Æ“ø*y1p­Çˆàç…¸&†{VŒ—«×E>¡Î`»x}Ä©OP]YÓøzé_xÈa²œ½=ª™žð³o‘â×Jî='güþD±ö¯Ó÷x0ï›)Ûx˯¯´¾¥g‘ûÞ ³Éu[§æ’eƒ£¯/V-8øÁ÷­˜M¨™Í,~"b=üâØWIÿaïªKGÁÇá7’o?ïWâ ?Kü!²ùœ}ãgÜØø8›š\çò€éáUô2:ƒ˜ÿ'‚ù¯›n-1Ÿ[d‡=U’.åÿKà ‘ÏÇ7m˜>#OÊûèNç/ø…Üøƒ"Æ2ÈÞ\ö¥»O¶¯Ú¢^˜ Jí¾QO0v³%œÍXü¡x®nc«üJX˜ó¿8 Þ&U™>[S›¾\6œ&%n@lËkˆ/_­øµÇ(YEG€¯ˆ ?$?¤BŽâLéV÷I +,c[ÁGÏ_å¯U¾¸¼tvŸ½,俼^Åübê5ë`už1¬I.|Éô~sýÍò±øß!œ5Òˆzÿ‰zgüIþG{n–î-þCÏñ¿.^)<΂¿ñ_ñ ]ÿή›ßT5¯˜žð3î$ºð8#ž9òZ$ÃûmÆo>×4<[˜†ØõìWf^ø-ØtΠô8ûyaõì¿…OÊ6òçg¿s”¬->*þÃ;T7E¬’ÚMrÿî ì+\Ýð¸ú~j#oôU³é¥[åJG8_Ý^ºÎÏö»úÁ>û^Ú˜S“Ö°ŸWEù6ÛXÛ[v×sYO¬ÿ^ŒC³ÖqŒ^w¡|Ï/>]2æq’–~A‡úþ‚S”‡Y°vüœÞªmõ|ùÃd£9ëÖsN¿/S3„œ{;fï+ØëkñöЧ9n>d&gaQ×I¼¬‡œ¸0¾æKóÿšr¶æä®°¹.¯çéüg¯•Ï ïb“ˆ{R#í ¶ãq™òŽ––xí³6Ÿßóì?’¾£?þ·Cý é“Ê ®BB“¸€é/òó~öYòI©•ós«®scpì¥DÞ™kþ+qûß‘szÝä#~ÙWù×ÙæãQ»Î>PâøúÄŠ<vް°cù(ßNŽÅ÷½W6‹}½íè_ßÕÿUý>;¶¡f¸@µdî“Û@‹6¿Å·Ø\_òïûÊ* †^#íç°V—ß‹IÌÁeÚè$ŽHüÛýIè[š^©?R>„Ó^„Füƒk´~–ƒÍ99AâjäøÜ§3Û’WŽŸñ{»h§n.yF'ƒ!±S–ô71cΛ£þ'XŽüò—ãܾQµ^>ï†E¿åñ¬Øf%b xôaŠç;¯/È.6lýÜšûÙ}Zµ­¾Oë¾Z ÿ¿Æ[£zS}Žô4zÂë‹°Û‡Ëñ½ élÏ5C<ä5×Öˆ[¹}x^Ô7û˶lA׿$'ÇÿrŸû_UûžËù¾ô ù0â Ô.R„®öº‘Óô¥à!jJ¨-£Æ‚3Óµ3¾çÊÖ®ôç²™ÄjϾPø‰úw?KæTÙJì¬ï™Ú!Dô7.ˆ¹ä£_(^ ™%ßà9¹{jî°©ð.ñ/2þ?:g½î5–ßÒzRkˆ=[97ê´É÷,¾^ö¯y¥|—&y,òW ÿ`S°3Žg ³¸O÷û¾'uK”Éý5gð>ÿ‹½·ø‘ì=*FŸÑ÷¿Iy;ÎéuŸÝtí–œì%ÿ‹…ÿ=ÂÿˆÄ·e¿gý’ß®øÿˆø¾lBýõë(uùœOéµøùæ¿ð¿/ü¬[›%â ·Ó£ë°õÔÜ0Ó™]®™Ìóÿ/‰RÓígÃi’³âü&tL õÅ0ˆóÓù0Wüˆx™yfÏqÏý^Œu}B|@œ‰3Þø{ä|ý»€ûøù~õU»™|^?cŒø#kÏÈU^ú’üœ¯©ñÇÿ¿íõ Ň~Ö™É4µpij—(>¦¦¿rn|Ö06g û^[üvóÕ—‰õí+=ì5Z9ű¼®ü»1–²‡øÝèûb±9ØäÇ(.Åù¶ìݧÎÑóT×Gü^:[ø< ÿÃÆé0Þ`o²Ç8?ïý$ÚsrQÄ<ÿ(ÚP0Ë}Ëôó¯ÖÁÞ£+ØÓ„½ÖÜt{ñâ¸âÇQÖÌ4Ž Ê…Ù÷ë“ÿ¹™ûÉ4ñÇ™7•ïUFO?]øŠ:¯<[¸½‚ºî«.vÄ<¸®|SìÓÆÃ™]¾gÚt4þ¬×ù¾YöÒë]î!šù_ì-X¹T5] “«¢½¿Œ/þcù_üßalû9Êk¼¬Ýò%šß§},¯‹ÀŽC:7®õGEqz0WÜ~˜p’ŸñÍ¿×îbŒwüÜtÛG„A¼¶ž8Ê­u–£ãœ ¤Ñ÷ļöÄÚ÷ºá#¥»©›÷ºŽ3"¾<_öÔÏ@3Ù&–ç¶ù®²ÍÎûkmSýÒgÔ퓳\±þêÿ*=\3šŠï ~¾º=Ãÿ=õ3vª‰ÿëÂÙ0~q »Ç™Ž?#z®ðç)4m Ø{N|ËÿÇñxÅq$1½=4Ç †û /«¾;gnú9ög±žÌmsÆóê¶þáâoS:ßøåÑṳ̈Ϫ^Þ÷‘Ù}þG³×¾FþñBêOË·.›¾ç(û¾ÚÃØ—½Ök&}Ò~³ý÷¢gö3k~™ÖüϺßõoöš”žiÑ7ëv³×´ùbÔv³ÏzØï'mgTz{õ;èûi÷?k~œ}0¯ze³ÏK¯kÜq ûü çÆ¥cÔvg}?{M{¼Í—ƒèîÕÿ¬éš”îiÏë¨ÏÛÞ ~†}oZôf¯yÿ¨tÛïFñõ¬é›u»½ÚŸU¿£¶Ÿ}.{Í+ŸLª/Æ•§AßßXæaÜyÊ^½î÷znZó<¨ÿiËñ°íºÆåËæ×ì5mþ–ïf5ã¾7mú§=Žng\>éõܤïOªïz]óÆ_ÓêoRý0,½žÔÞ¸ýÛ߸óÛ‹®iÉÇ´ûWþ†¥gÚó<è~¯vz½7­ùŸöø§¥¯³× ý¶^zsPÓ²Ùûã^£êaßW~'m§×sþ7êüN[ÎGå—aû•ŸÇÕCƒÚu<Ó–ëQû—¯FmwRyéu+oÓâ¿aŸ¶½qõæ¨ÏMëùY¯ó¸zpP{Ãö×ë½iËï¨ü9ê{Óâïqçg\»4.?;ãÎ{¯ï‡¥7ûܸïÍš¦Å'³jgÐ5m=>è½qߟôš7ºÇÕÛ£ò˰ß÷jÖýõz~Ö|2ªöý^Ï «×†íoÐ5).µýõÆ/ë-?ÓêgÒvgý~¶iÉÿ¸ò6n¿£ê»aÛÔß´×oÒyYo¹ìõܨãU_K÷¨ãš¶ý×¾ Ûî ~Æå§õþ{Ð5¬>™tž§e/¦Õo¯ï§=¿Ã^%¯£^³žŸYñÉ´®qíâ´ìJ¯¿ÇmwÔk\û³Þr3ì÷ö?n{“êŸaéšö5é|jG&Å‹“~?¬Œ«ÏÆíwÖt ûÜ´ú¶ýIŸÔï°íNzM:Ï“®Ã¤r:©½š”F]çqå~P{ã^“â‚Iñİó>®^Ÿ–þDרË7£~?èš–^Ÿ¶~WßM Lº®Ãö3ìû“òG¯ç']·a¯qõÇ ö¦Í_³¢kÜçõ3èïIõö¤r8èþ¤z~Üuöš–ž–½ÔÞ¨ãt==7+ýÑëûigÐ{ÓÆ#Óâ×Iõâ¨×´ù{Øþ†íZúpØûö3-ùèÕþ¤×¤8jÐó£®Ë¬ìÒ¸Ï ûü¤ü5-=3®ÕúÛï¬íø°ôŒÚϤöxT~•™•^'öz~Ü~¦mǽfe&mwØkX>œ•¾÷šnž•ÞÔþ¤8oØþ¦mo†½FÕë5¾AýNzMJw¯ç{µ?m¾oM«ŸIûŸ6þ¶ÿQŸ›6ßeÛUûÃö;m{8és³Â£³Â‡£ö3è½qåo£äd½øcØkX}3j;½¾Ÿ5ßLodÿžö¼ŒúÜ ÷&å·Iíò°ÏÍŠ×Û.{«ççmã^ÓÖÇÓngVíõj\½2).÷¹qÛtœ£¾7é󳯳½î¯·þœôš7¹›5¿LÚÞ¼Û«IñÜfããqýyÓ»“úsój‡‡íoÖþù¨×FÍC¯û³²¿³ò'æÍ®dÛ™6¿mT`Ü~6‹þÞ,ý΋œMjÇ'õï&µcã^ÓÖŸÓº&¿Yã•͆F]ÏiñÕ¼èË¾Ö NïMû½QõæoÿL[>‡ÏI×y½økVøc½pß¼ùÇë§¶ÝY=?îµYôЬpñ°Ï{ ЧÝX¯Í‚‹6Jgƒ}?-~œ´ýõÆëã¶;k»8m>œ·ùÜhz6ªŸA×zÇ­GígÖôíÆAëÓÿ¬íÀîk¼«—2oúrØk½åy^Ç¿ÑùƒÆµëÝþ´®y¡sÞý§õÒS³Š'¯ÞŸö5i<+Ûδžôެ㳳¾æÍï\¯<ËF_ë=ïë=oë-ó†÷7ºÿi_ó>žy§o½¯y?lt¿óÊ'‡Û¨k£×Övx³¬Ãz_íï»>ÓòK6úš÷¼Ê´¯ÍBg¯k³ÙƒõŠwnöu½±]óÎW³¾ÆÅÙë'š÷ytõš·Í2®YÇ3׫õêVu6Ãö»Ñó5èšÕ|ΪÿÍvÍ{ÞpÞâö³¾6 o6~c»&•ŸiåQ× ¯MKO›§Ü}vš7}²QþÈ´®ÆÕórm´:«ç7 'mTž|Ü÷çÍo™W|;­kÞò«³²ïó¾Óº6KüyÞêw½7ìóů½ž7ÖkÞæu³ø?óJW¯k½çuÞñÜFõ7lÿó‚ËzñÍFÏÛ ë· WÍÊ¿Ø,ãtÍîžvžl³á©Yé·y‰ _6_>/u;ób—v_»¯Ó5®ÜÍ+Žßè~¦Ùh¿tRœqc©˜ÖµYÖoV×n;¾>×fÃå½®ö—6ªÝy»ÖlV»±Ñï¯w»›õÚ=Ï{mV|¿ûÚ˜k½äu½õ÷¼òçFÓ5/ñ…aqϼÄç­ÿa¯ÍBgöÚèøÌ¸×¼é©QåiTùö½Šûüf᯺6:îxc¿æe^ç5>°›ÿv_³¼¦evóç|]Ó®w›Wý¸Ù¯Í¶7ÖëÆÎß›_~[qÏo«¿vc×¼gÞèõº±Ëżãìy¯›˜×~oì×oû¼þ¶÷Õÿšwþ˜wúv_óqíæ“Ù\ó;¯‹µö¸Ì~ÉÙ§&ŸJüÔíÓHÝoôx>}/Ÿ<÷,u”Ü/'ÏÆûÙï—3í&4dûÉö—~®ûO~O?ßHýìÖn1þ¬gÚL~–# Ý•.cÈwi—{¥ø|#>Û­ÿZêûôý…T| ñï|¤5=oy¾Kæ7Þ¯§~vОî/Þï¶ÖÙÏêýÔ;ÞGBgfNŠaí8k™¿ ©ñ'÷÷RcM¯yÒGºýnsºÔ¥¯¥d®’yÉЛŒ«ž¥+C ý-¤î'Ï5RtûÏ­õÔÏ¥L¿8–욤y:y?—ù;ÍS«ý'í§ûIý^IÑÜO䦜C2_é9OúJd%½¦åÐÉké~“gÓ뜕ùä^>Ò’ô[)’j+Ñ7µÔ˜Ót!.ñ»d\ÙvÒ'<±úNü®ÖÎÿê\gÚíŸÔ\Ô2Ϧçd•¶Ô|•Sw¬Eê™åÍ]ç4=žÔ»ÕÌßµÔûéµHÞÏ¥ž-fÞ©­E"i™Id1_9ôÖí•Øsèð,ÏtÓG«s›ù®Òå÷ô÷ÒÝå°KoæSt-w¡7mÃ*¡ OÄŸi]í6&™Ë°Kþ+™g‹™6:Æžº—n/«ÇVõQê™´ÎNë‹Õõ‹w´:çsuü©ö“ç’¹ÈÎWºí|üÙ¡[’{a—§uݪLñTBo53îôø“µÍÚïJØ¥?Òò“¬ÿrèÔ¹ù }ùÔ;Ë©q¬‘™°Vç%»žã™ø\7“È}Z¦Ê©þš=ÆžØÖ¬rŸš‡¬îMÛÉ4ÍÉœ­ÊRjüK©ß×ðyj¼Œ';×É÷Yûæó“š›|¦Ï,?V3}¥ù©C/&cOµÝa·ã¥é馯ʩyJ¾_JÍmb3²z)é•ÿ3÷³ý•2÷™JÖ7W³üœÌG‚WuMªïEÚŒ'?WíNBSò{ê^Z²ó“Å…i“ö+ViM·“î7=]hÈ%ãH'i7ù$ß1o ¦K?çz6tòHšÞ´ÞHÞMë¼ÄWJ¯sšžJæ^z~VõeêùBè´}ÌóRÈÈMº¯ØV½Ëý,¿KËjú“¶ÑévÛ”}>Û—ÿš—,ßg±O‡ÍÎŽ-ûIµ“Öƒi¼‘|VõU꽄ߓ9LžcmR§e,Ñkx2þLË\G»™çÒ¼”µ§sÔ¥´g}Ò¶"ë3¦ŸÍú«z7Eû>Óvvm’ùMÆOõYL=—¶‹iž-¦ÛOýžè·„/¼•´—ð÷ªMJµ™¶ÇÉýJºÍÐ);cJžO~O·vÉdòn=ý^öºØîÔwYŒÑAOŸ6Ós–Ä’q¦ù*ëuÄRýfe0»îég;0e7zS÷û²:y<­×V1A¦ô3‰<åÒÏÄï;æ?dpvvL]h]åÝtßY:»´YH3’wÒ|ØMþÓ4§õz·1&Ø©:å™÷±5svñôšñfþNèÉÎãRª´_´†GC†ŸÒ4d¾Kó~2ÞU™ú.±[ÕôóégºÝ òÞHÝ[Î<׌ãË¥žI|¹þOýÝÁ«]h^m+uoŸ¤–S?WûIõ—æ§Ü€O gca«¶/ÝVš¶Ì½ä½4ZƒSCX£W³¸>ë/fq^ktÓ½iÌÒ¡ÃÒŸù#û}¯ç“ïB1ì'¤x¸GÛ¹^ßgžM¯ñR¯v†[è\—¬6ˆîT?Ånß÷¡•¾ÁëéþBç8K©÷VcƒÆ—½ßk,}è+eÛ kñ[WyÎþžz ßöGÒggûòc¶­Ï$ò×GÉŽ!ì²É1Âä^7zB ‡eî¯bº>4 ”‰Ð…×Ã.;ZJÝëˆcgúè°-}úèЗÉÏYÏÐigÓú+ûw5óݰk•å^¼‘Ð•Ž©­ÁÌ]~62?“q­‰ï÷è·”´RyªÌ3ÙµHÛšA6«/:ôrü;­/’quðõ€>Òñ›UŸ é?dtJv<=ÚÌÒÓuÝRϧó¹aiïÓ^W:CÊo²¸§ƒöôýai ]tfBS¶Ýíx<1COB§¬®y.¬õcVã©]žÉþ‡><—ô‘Ð:ç§§ÍÍ<—¥}5–Ÿ7—ú{9õN3¬é7¤ò‚IŸ]ž©¦ÚèéGeî­æÚBFežK¯_òürXëç­~º¼›K=ÛOû}ºÐœŽ³ÕR?W}ŽY^meèÊö“´›]·nò—yr¯CwÄO7½Ù….2:×£—/“K=“½×¡‡B—õÊ´¹šNßËÐ’í#»KãÈÕ¸Rò|裇Cw^Ncå¤\ØÅÙXõ*m!å·eÚLâƒz(óLbûšÐÐ [Cèä…~˜$¡µ#nߪŸ›í§Kº©Ûs!#Oñ~šÖ$gW }tošÎ.íwÕÁ]úkd~\aW—ú}5†Ð­½Ð9GÞnü{5_Ûƒæ\æï5¾AècèÑV:ç—Гõw:úës/½f}õJ–¶ÐÃþ÷CŠÞZö^>Ùiö¢)Òî±Ý°V¿ Š _²4†µ6¼CÂZ>ì©#R÷~JrÎ}±o!ö:ç¹—ï‘KÑŸŒe9¤ä"¤jj†¡§Ùû]þÎb»\¯ïºýÞ…öJ¯çR÷³:6ñ;òH=ÞíÕf®ßwé~»Ð¿&ž–þ. ‰âÏAÏåzüÞ!S©ûùÐç0œ ®‰1d¾_¹÷è'›ËÒØ¡»´ŸØŸtÌ¡y&ÛÿP:#ôŽË¥é_c¯²ý„N¾-¤Û ýãYùÉÊPÇÜ 1ž5:;ÄúÉ.ãJêM»êùô3ƒúϦýƽÆÏ úw±GŸizÒ5Ýü§\ü™Ö éø}š¦nºp v ¸¬+VÈöŸ¹×AK¿÷³m…!çº×»Ù÷C}<àžç {oP»a8ÒﻎšÔÔýÄîÒGÝødœ‡.±õ©³DèŸÉeß ©øÝ ¶3ý$6®[î&—j¿-ktlè“[Hµ¹Úvê»51ÝÐE¿dÇÐíý~Ÿ°6Þ0ÔÚ Ón—{®g²´ÎâvÍÙ0ù’¡bb£ÒzøYI[¡;6îë—eî¯ÁÄ!e‡F¤§W¼®>è™ïeñã<:l[úé«§Ò4 ø¾×ø‡µË}ãÊa‚¼k·÷C§-å÷Õš«QiÐoÖ7ïÀ\Yšº¼ß+ÐK×­±sÃŒcܱÅ}GöO§LC6ö™ëÕ~Z;òFé{YzÂ2׫ÿ.´tÅ$Ù6²ÏÅ{Cç²zµÓ­ÝIۜŻ¡ÿšeëð²2œ­ëå÷Ń¡‡:Ì82ïNd¿Coû“K~†ùûn4wûnØ>¼³ºa-ÙïÃ6i–Ÿ°–Öĺ¼“Îç÷U÷ù®cÏeº­ÐÅ?XÏOè^DZ†–ÝCÙªëêñýP¾KŸ÷×`º~ôô{¦-kìHjlý|ž56©ß8zÜÏ óþ¤ï ÑfÖìåsç2÷ÕÅ ôÕgñ }ðþmöâƒ:/(6´­ÉþzÔÍõ¢/tY¯Ð».b”RÖ> #‹¹n¿÷z¦W»a„õ }jꇡ¹Ë;}cÈC¶Ñ3:äûY,5’†Ð#Ð22ýÃÐÒå¹5þѰïÎÛ'ìÊÕtÕÓWè-c}×yŠýÌYö¢5ô°3£ÐF”³0z]ØH:¯WÃÐ6M¹nt…L«aÖé÷÷„´'¹~¸»gn¥Çó5Ê¡“'Fö¿ûô“÷ÙÐYÏ›ýnæ±âQÚl1m¡ûZäÆyoœO˜2ìÓÏT×mÚíM›Ž0Ù¾š±ÞíE˘måfIë,ÆvÙÉ¡ã-aòO£Ža„çsÙ¿»ÜcŒ«{׃în}Í¢ÿYŽ)¶ß+LBGŒ…ºÖ5Ír¬ëõ #úaú¶#׋®Ð'÷4N›S¤Ù}¤Qé¥ýq¾å™qÞ=b¯a8ŒÔ3÷=ƒjÒúÖKL›ž!èÕ7;ÿ&ˆ…{ËúÑ7,ÍaJ5Óþ„ëúÄÜÃ.¾Ì¥îåº=ÛçÝ~>h.óìÜ`©>4'¾u®Ûx‡xP}b¶¾s( 1ýýtu®×wS¦¡+ÆkÇÞ•ž^cèv¿×³ƒúôݸï±ë¾§1ú´aâ{–ö÷GK!þ6@¯vÏ0´ŽÚæïôÛÇÖ/‡:°¯Ð»FmÃë †ý„X_<…v¦^KfdËÆ¥µßóãÒ&Œû†Î:Aûÿ{ÒŸG˜r. Ýö¤ÏŒÑo. ØÓŸzvæ>~Èœ{1àÙ~X#}fAÏõ }°Â„ã„ :Η›´ßì{!bȰkOÍP8ô©1ê÷Ì(ßïþ¬ß'ôßK•Ë>? ­qrHÃú-khéw/Œ‰=»µ™þ®ß÷C´=ÍǸz`lúGéšcÖ§MÓ¤5Œp¦Ù$4Ïò:uÀºÄ‘½†ˆ[Œ3¾üŒHóÔj6‚þÔ;¹¦{ÖŸ0†Ì z'ôƧnSÆ}¿Û˜Ò÷BLÚíõãzµ™m;̰F¼Ëý‘úêFcHåü²Ï†îgŽU_Øë~˜³|sƒÏô˜§O˜O‡Éð@·=¢sa·7Ã' !añ〱®êŸÐ‡Gª_»3S¿³6…LBç8ïNÒ߀vûþoˆY~²cšÕ{ô½.cžÖ˜¦ÕNŸösÙ_¼7SÖ­Ïi¶?/Ÿ¦u£ûŸ·O¸‘ÖØÏÓ'l¢9™&­¦2s¦Ön:›·ò^+»tÿwì8&þQyÀñ;>íÄ­§îØyf¼U~ÀŽíÇmÛ~ü©[OŒwJܶSN=þômɃ¶³ãèÇoOn4zÊ©ÛvnßzâžØzÚ)ÛN‰·k¤§Ù“[WÛ¯râ‰{¸íX{;¡¢|Èöc·}êñ;NKÞÜrÈö'¶óÌ=·n?fÏCwÊŽíÇo?.~·ümÛw×xÈi§Øvœr¶·¥h~è¶'Ÿ¸õ”“’ ݶóô]Tþ|Û)'¿3=ðü#N8~»&m1™´…¤Ïc·mÚoÿ§'V¿YÚ¾õ$°¯Ä›‹[K¦fñ”mOIh:eëI'Ÿ¸í®gîLÞ=aëÉ'oMº8qëIO8&ù«pì‰Gßõ¸''3tôÎm[O=~û®)ÏŸtÜê`–=íÔãOJ¾Y:Æ}bÒÎÑOÜz²­M†êÒÎgÜ5¡|5 |à ß; Ç”ÙzêÖ»»Ó^Ñ$„ÿûÿÛý½ñá$ survival/data/pbc.rda0000644000175100001440000014010311732700061014304 0ustar hornikusers‹ì}˜Eó÷ÎæÝ‹$£¢`BQèãrâ’Q$šÓ© ‚‚Q@ (*** ³‚ sEŒ¼`Vüª¦«öv{wvfv÷îðûÓÏ󻺙鮮®Ô=qûä èt8N‡Ó§9œ.ø×í„?šÃ‘ê?l8ümÿ6dÀ‘»äaVÓë</Àð°)²M¤ËvŽL@ ÐÐÀ|÷4ìh 8p à @3@sÀÁ€€–€V€C­mmíí‡è 8p$   +à(ÀÑ€cÇŽt8Ðp"à$@@O€äzry€|@ P(”Je€r@o@ Pèè èè8p `àTÀi€€A€Óg†††À6Ž€‘€Q€3gFΜ 800000 0pà|À€ .\¸pàrÀ€jÀ•€«W¦¦®\ ¸0 0p=àÀ ÀLÀ€Y€Ù€9€¹€›ó7nÌ,,Ü X¸ p;àÀ€ÅôûK÷îÜX ¸ðàAÀ2ÀC€‡–<XX XxðàIÀS€§ÏVž¬<X XXxðàEÀK€—¯^¼x°°ðàMÀ[€·ï6Þ¼xðàCÀG€Ÿ>|øð`3` àKÀW€­€¯Ûß¾lì|øðàGÀO€Ÿ;ÿììüøðàwÀ€?þüØøW†¾4ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø×0o@ükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿZKÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯Aükñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÚÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÿÄ¿ñ¯AükÕˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ â_ƒø× þ5ˆ}¢ùy6>.ðeÈâ @,úÀp<:Ã\Ÿù˜´ø·{œ’2Aß.tjK6Èî{e€>œØørÄL6øó ØÑ ¾Öà6é™ Ǥƒ-²ÀFiP/ó §éwüÐOб|£بä‚ °Wä?ä ÄBð'ä øoŒ'~á‚öÀø=à+>ðÉtÈ À–~È]  N6è âÇþᜒ:ËßB^É‚q»A·`Ÿlè+Æ— ¾ãƒXqc~,”¾êþ™`_Ø99"â? úÉ‚ü–±æ½Hæ´èÕ ù+ˆ6ß íþíÄ|}dAœ;Á3AO™ÀÇ :Ï€|‘ŽzŸK%§CŒ§ƒ}3 –‚ãÈ‘~ˆ±@g©käÀ ˆ¥ 3tóCžÂõˆo§Ì»éó\OÓ€wlá‚8M‡|í}¤Ã¸<`wè. ü3r¤ƒ­Ò!æÒ!¤Aî÷@ÎñCL§AÎKƒÜ»¸P‡0'¤>ƒ ç4˜<à“NÈ}i`—4È nà—†ëðã Ä” b5 òVàA¹ä ‚ìÁä|á§üç†|ÿñ‚‚àËn˜C‚ · èÂò!~ƒ÷AÈ%lù(: ÂX‚0æäÜä^?ø¨ æø“ôê„y ù00_êçŠÄ˜ òcb*c €î¯ ƒtÀ>Á¶~ˆ ?̃~È‹~ÈŸ~ð?Ú|Þ¿Tú–Æî‡ñúÁ®^ÐÚû!¸ 7ûÁOüGÜ0gù!÷;ûËœæƒñù`îñÁø]0ÿø‘9Ës§âÞ1ëÝyÁž>Èi¾ÉDÁVN˜s}³]?}h‹»¤_za®ñ@ì;aîóB ya^òœæyÛ ¶óÂ|çð‚¯{aî÷‚^˜ÿ½_½ ¯úõÂ<0¯xÁHîeœï1î9<¯RìÌ›×70Ÿx`\nà弿Øð€žÜ/n˜c\-$Oæ°¥låž.ЗÚ:AØ×“!טÝ0&7Ìán°¯{-ò¬âÌû]à_.ˆU´Ë‚¹?ÆŸ¶÷Cü¤ÁZÁ;KÊ.ãÝk… Xd€Ü Q÷àWÙÓý0gy@Ï^\o@|¹ <»lg€þ‚àç~Øvß ˆ« Ì0O¤ƒÝœ0O¤£¿Cü;!‚à—i`§4°™âÙ ë• äç èÄ2`NöV’ƒ¼ð æ˜ü ŸÿU9GaŽÆ5†×x ?7ø¥ó´ñCžôC,ú f<<ƒ|0.÷,és^X‹ùΑk"ø‰øyA^«r‰Ö^˜ï=XzÈõšç² êæäK7ä 7èÀuŸÌnhï†ùÜ srŽäv÷6vC7¶Ü炾°Ætƒ¿¹ÁöNð+×[5ó‘~nä J§SQûÂá0ù_mo[mcT×aP×HF#™cµ1’ÇaP?Þ¸âñ¶Ò_,>ª=ÊþxýÇjkÄÏlüê¶Ú¯ÊGåo$ƒÊϨM¬ïX,?ŽÕ6XýÙ˨ž‘>Ìä1“+\—*O-l¬ïXxp>±d¶ÂϪTZŒývlK†x±fÔ.^Nql‡S-6áûU¨Ô(cù8+>K&+Çi~½.êcíS÷«m5“¶f¼ÂyÆâ¥òÕ&Þ>µ­QŸjµžQ«uôdt܈g,9­ök¤×x|âÙЊÞãùY¼qÝÌ­êÓŠ¿Ù‰+r&3;þaÅ/âŧ•þŒâÆJ,Yµ™Õ¸Ñÿ¯vÈ‹Òû°wá_pÓîåCwùçýsƒ(;¾Åé#Ö>'ª¦,Yuí†CDé¯ßúâ¦ËEnÆe·m»LTΚùWÞ§¯ˆÒÆÜÝ{Ù‹¢äÚüYÇä]-òs´ÆvÜ"ªî~¯Ç7K¾%Ÿ ùzkåBQ>{Q‹éâQðáñ5n7C”5¶©êâKDeÇÓNêûãÅ"¯Û‹÷ÞµûWQ²ÿÅï>Ùÿ*Qò[vécGME/¯ýîŠÕOˆò?ƒŸµºà!Q™¾«æ¿Ne­GxÞ¨¢ô¯×ÏÝ1q¾(è5züVLüáÚ´ãD‰«ý˜·:eŠÒn=Wç‹‚‹P°N¢âç¾øéðÕ"ïѾÝÓ_»Qä½´âåN=]”ü{ÑØÏz÷ÅK:ÿrÿøE‰oÕÊû®î!Š*Ž:e¿oeGÏÞoÆ™3Eo”¢tªèU).Xr¡!¯ÞYä…ýÉâíãºü|Âf!>ü~Û-kEã&¸ŸiÝMôzmæâ™þ-ò¾ž¾àÃQýDÁ¶¾ÿl…¯{ý•ÅG‹¢Ùãò@¢÷Üi7¬l¹S”¬k:°ù¢M¢$íXÔ˜(Î?öò«ž(•·ƒ–ƒDw»íì#*†t…•cO^}äé¢bòaíw +yüø®û›EáñÿÞ[ÜãhÐÇ—Ã÷Ÿ(Dï+¶Ï©˜!Š<£iç^'в£:pM·1¢hLñ-;þ8T]Í DnÏœK:ÿ*Šy¿,mGÈ_ßç´ç·5Å¿/qXt½(Úþ\§v­…¿ÿóä§ûŠÂ^Y ›ÑÆÕwÝí?Q±øé_}~„¨Üôaúm+rD™¶vö+~'Jwä?›}ÏhQˆ^ØÿSQ2´ásV7½î^:üš5{DÞcC÷9|»(!?«¿zÍŸuŽ(ØøxæùAÑ÷ptì=¢WÞÉÇÜ~<èõ—Ч¶w[ » Ct£êN7øg7q¢>ìBQÔý‡F“Z¸Eñjו­gÿ ~´ðÉGºíe¤ß"ô‚£>ùWùx÷ãï9žÆ«Šª–·Œ8Ó÷•¨èµþŠcÛlŇ?sÿkC_ Ôè ¯È<õóU÷ydïÒ§Ý !!ŠoÛÜ,Ðl–(ýÞüÛG=,òu3_!zOó¶ÚÓó\QÖÙ³ ­(B³ š"J[¾Ðóå^9sGgˆJŠ“Ò'Þø9Ó=SlO{öìA‡Š’ _»}«V‰Â[Ž»þÞ¿—‰’ Kw4ݲ\dT‡WDÞSJö¼äoÜ| ˆ, š7ï5LŽÃ0‚E^ÁkGïØP!*ú\{GÁoÙ"­èºRä_²åÓ…__qØTõ‰OTœ{j@ßñ爹_åŠÂ¡º£ŠÒÀÝ­Þ‘! 9Ÿ|º{QÙÈÅ¢0£MþðS·Š’6¯ÿré¼Q¼àÁé»nì Ä÷u¼¶é*Ñgíš%ߊw½·íœüż DÎåý?ùæBŸ(: uÚð•¢êÒî¨Q´qÐ)‹.ÊU¸{ÅB‘Ñå+Ë7\šù¤(b¿!-Ør²ý‡?ˆ²GæÜŸûæÝ"p[a¡(?ëÖù¿7/J°öØå"•†!JP¬•‡.ò”ÿgÑûâûp$¢póyÏ}>äZƒ  »(ê÷Æó]ï^*Jú£cœ Š~:í–àòÑ£`üå׊Rú'.JéŠyod½Ø2ûQ ^9ÿõ^¢ ¥ð¾È§¸É/zâ®ÆµÅuÙ…ß”ºx¨ÈkUòöŽ;&‹^äçE{¡ÇŠ*Ôò)¿Šœ¯O_sAõxQQ¡3Å}pÀ‰ÒÜó¾j5ä½÷ßvÍÈBQ2¨dÈszZvŠ‚®o] ©]œ4óÀ¿~/)eS-|à²7E彜m “‹<0ÎÉ7ö¢ù«¸ÿv‘w½¸¢7H5é ¢Í %_vЧ9¢ÿv~Uôg;õñõ¢O—ÑKŸ=éJÑûžõËL}L”Tl{aÁåËEÑ–Gž›²äqQr^:f8QHy¡øÀƒ¹ñ+Q…aüÎë"Óþ7÷ˆòyÍs›ó»èì¸b÷~O‹œ2L¿ˆ| ÏÏ–ŠŠ·–ß±qÚQ¢¬ra׿MrD!zcU+Ñûdø†È…dxþF—ÈvïùÌk.ÊNuõ+ê’-Ê/zù¦ìƒ?½pon3QqÌþ÷-? »È)ÿã”q·måÅûãŽÛEþ)¯ö[üÍ;"çH}"ÓÀ—³D.åßò-Wmm½æmQ¹ðŒß/h÷”(Äi¬ýBQFrç=òýÏ­cDel”¨ü¢ûÚn zº¾@|©'6!(åaù¦‰ÒU§ý¼nbCQøÏ Žc§ˆ¼‘…•/g\.Ê.]ßöŠáQÕ|Ñ;¿=ÿ”è‹Óp«{Dî/m¾ëÖž"G3êJÑs“î}ó-QNz)F¯i3Q”ãlßc¨è³XwQÒùÖÿ=üʼn"„˜"gÜo7öî×¢ Á1¢ÆYÖ^dQõt“F<ÚæÉ'¬h´QL×'\qBaëï–Ío*ŠOœtϼ·¹]ÿÈ»Nâ´{Ò'¡qäv¸h÷¸ QœñÁcÛvž'r®/Ïßþ½WÝ e?…6¼÷âE¯ˆÊ gÜ4ú½BüþÖg—tÊGEE]²vþ5QöðÔ){ƈܱ7¿yþ¢œâ¡è}b•[¨Üõí¢òŠ@ñ9YóEÞ¤Ò/2·~ Šq9´f‰([Ð9‹ÞÒP"gÙNȘ¹¢'ç9HÒÇyöƒxú4õˆÈÅôÛi(‡E„Ž(X¨/D/šÿ{÷Cß,rh•Ý7~2c(ëª'`ÑcÎû-Žœ6Dô&?*o7øÁ­T‰ œ¦Ž{JôúwxJ3!hé-s¢²çnX @ž¡üÝg­¾åó|Ìjÿ' i»€ÚÉi>STÒvÑ ªŸCõKhÕç~¸o3ÿªWF혖R}–—÷Ò~O‰ÒO1Ñ":Þ‹å!ZE틉²¼¥J»J>NüY?å´?—å#ý•ÓqÖO¡²G|X/¬WæÇúQõÌõó•ñ±ü½ùÔqq?‚ê±Þ¸?ÞÏò©|Y¾RÅ~,ûë¡JoÙå)£ýl—bEO!?TüOåÃöáþy?ë½TÑG¾b·bÅ~9Ê6×cyË#ýÕáp…ß72ºvï¼ÝÊ1;<A¼ë?VþOVN«í¬èËÊõ»üã]‡³Ã+Qß0ºžb&s]øN*‘ Yëj¼Vc·>tï:c}Ú.U2Ôµ~­Æ•ëèµ<ÖˆGÀ½cFž?r >)œ-ê{ñ‘BùÏ(¥¾gø˜¡“¸zˆÉ¨¡Ã'ÃGŽñ1ÒðÛRfE­“h›DÚ%RÌx$ÒGªå²#c2}Çj›Š±ü×K¢:H•îŒøÔ¶m’w}ÅMÂ¥zþpD?&áû?–ïåGµÊK­k[í'£þcmÇjkL±Æf6ÆXrÉOæxzÅÏʱx|¬Ú;O3ûõm$«™Œfs©™¾Œlh$«U1âi&£Q;ñdäofv6:¯ßx±Ï_ÔbÕtc§†6ªÞ‡ÿñç£ðb%gYõ©x±mÄ/|Ûj ˜åúx¼ÌòŽ>Vò¨]Ä:f%_õoÄÇn1›“¬ð·ãñljuŸÕ¹ÇŠN¬ú•U~\Ï̇­øx¬ãVcÙlΌקY±šKÌx˜ÕÇ÷Hë=×ùHl¯çÏŽˆÒss$µZT>‰7ë×”Ïfkõ¬–Dù¤ª3~©¶SÒü7Ǧ©îϨž™ýüÝjIµ] ûÙl±ž:ž:’/Ù¢ŽÏÈ_ìêÁ®= ù¥ˆO]•ÿмõ-gmç¿ÚæSßÅêü\WzKx~Þ\»üëºØÍ›‰ÎÏv娫ò}–ý9Éã‰ö›êR‹ýÉ×™§~v.À²°ç¯£ôÒóï X6Š´£ôÒs;ç¼·Iî¹_t¤zÔ^¤Ëö=·?nÏ”ö µoY_4¦v_Q?$‡èBÇj}’÷â×…äáã$hHíh|Qrí¢v<¾fÔ>Hòðxv¿ƒ¨ËCz Õgù›Ðq²gÏ=Š^vFÊ!šÊzÂMív+rs?NE.æ÷=Õ£þCò[-ù²¼<^¶;ËÁz;šäe>\ŸÇÁõZÿ.H¹:Ðþ‰éG¢Õ?ŒíBüxœQ=’S4Vü”õJÎÜóªÏõx<ý|œhz¤BÛ»#ûUã&t\±›8’¶y\W^jÏþù{¤CãRüV4'~Š^Búc}ûHξ,ß.Ž;’˯؇åeûðx\Ôžõp|u¤¾8°ýx?ñ éŸã¾ñÍ&ÊñËãa=è7äw,?ïgÿ(úa{±ÿqžb?þ*’¿h§ðc:LÑëîH}‰nÕ‘zlKrE†2p¿ì—E>E®¨¼ è;¤÷¬H¾!=é“åÝ?2®Bó×ã<É—ó‹8Œøþ)gˆŸ’OCí(Ÿ‡æÃûq?ˆïï‘ò„ÆÏqÃù­ms~hIýe)úb½7%ù˜Ïn;p\òüå§z<Þ‘z Õ§z¢Id\„ò×ãxâ<Äú å{ª—ÿêy“ÐH¶o+õYž(;wÙáç¡qñ¶,W+’“â;´náñ¶Rò˥ؑý”õBÇy=ågêø}‘vVý#Dÿb?ˆGˆ/Çë6…ªë-žÕü¥Ì«j> ù3WûAÈh=È~Ã~¨ús(ßIüøø~¤ÏCh?ÇW;¶‹÷JžÅ%Û•ã§sdýÐø™*~Êßê:‘åæù¼«"‡GÉŠžÅ¡‘ñÏãÍ£JÞ QöƒÖJü°<œ§Ø¯¸þaÊ8X•þؾ¬¶?ûwDÆQ(ïmŽ—h™§D¶§4_ˆ†‘ñZ²Ÿòú„ågÿd½©q·K˧㣿³¬øY`ôüäòn‡þy_ü¼°þé\X;Ñâ¯8ôÏãgеªe—øcüÔ(JŒŸiþΡž?E¬B?+‹í`¦pãç®ñ³ÎŸ;ä'maôú'€ñÛœï8äçoñSÞËò¼ø Û±†C~.÷PúÿK‡ü䯋¶5‡þ‰bç(Ún ¸ð?Ù~F?÷ëÄÏt(ÇëZâŸÔÅú:ôO-âç§7§CÿŒ0~:Û õ¨«é¤£•ý³Ãø)fm-?ý§C~ZÆä…vúçqŒó¡-~>??þ†Cÿ,°ãÕ°qøÉgül+ê?“«ÞyµC~ ù-Ò7~‚?íŸþÝ<ñÍiRçú'ž±îk0®Çúg~—RŸø¹ÞNÄã/ÒÚ>½¦ïO¤# ~ê¶Ÿ~À!?™ YJÿä1Ú« êág›©Ñ¹ õ¡Ñ6ŒÇ‰/éág²ñsߟ:äçñÑëúçšñ˜ö+Ùh;Éø‚ÃáA¿¹ìŠŸ¾†Žá'º¿'þ±ˆâ'ŒñóÛ¿SÝWÉføÿ ?9ÿ lŸ%ã'$ñ³Òºà§©ñ3îèWA’?‘ŽŸåÆÏsã'½ñóçøimüÄú{5}êþ~õù­´~.Ùý·ÜçBÇ> =â§Ÿ'“®¡øùBÚEÿ´øsÒ?^i_§SÚß¼ÿö=-ã?™­÷‡öÿÐ??Ÿí¾l ñŒŸ½Ö튟ØÇÏãçÃwÑÿëò3×È ?Û¼ñÃÏK£ý±ý‡ü´5~VcwùÊžCú*¢q¢üŸH»âg³ñ³Äºÿá§fñsÞø‰èÖ4†c)öŽ¥8ÅÏ‚÷8 ®—ŸVv^C>‡ŸÇÏoŸL:ÇÏVWHêýá'ä’ãv/i˜ÀÛìÃÞ_§£ë@"ûÆA~Y£Hû xÒø»?îªï¦üd‘Ž[Á“DíÏÖ« ¿úñèø%ƒի¾Ù´®Wë(2‰¿“·‘tøUd>Üøâ-E" ¬çÒ/ÞG.éõD&ñËл vóÝÔPÿÙÔ.Cs‚RûÉ›IòóxÒ¨}ðI}‡Ð6Ÿ†ÖéiúîÂOr°øú ·òqâŸAíÒI®,’ËKõ3h¼Ü_&é=¸EWHèz"çu/Ÿe‘¼™|^Aü³Ù^$ŸZßËö"ýs=µ¿,² Û9ía}€¢!əΔû£öYd7ë‰ì–Füüdÿl¶?’_žéƒûc¹2I¯ì’ƒ×µÙÔO¨=é/øÈ_2ˆ¯ŸŽ{È><^¶_&ɤñøýd±¼ÄíœÍÇY¶?·g»’¾²Ù˜é­·ç~H~–›×߬çl²3û=ËÃqÄvg?ñ‘œiäG™j?¤'–—íÏz éG‰?¶'ë•íÈõCz'}d‘÷ìßiJ¿¬G5ß³ÿd“}J~I£~\ÔŽçŸe;³r ù)õü—ó Ó¯øÏì·>…/Ï_œÿØ}´íeTäç8òòüJzb=? ÉO~‘Éy•õÇó5QžYŽtžyäy™âŸóI¨Úï$ùx¾á<òÚö±Ÿ’½üÔ>SY7pœ„æKö/’—¯¿p¼g)ãá8`ýpì/\ŸÇÅó)Ç™ó,ËGãc¿c¿TýWã8¥vœGƒ4~ŽG¶ç™ÐúGÑ#Û3@þÊ‹ä~ž_x\4NçêõžÉò³ÿsþc=r~`}’߇ô¥¬K8ϰü^ÒGϤ—lÖσ$/óç¼Çó`ÇEüyÅzQÎćý–×W¼Ndy™?o³?sœp¾gI£qó¼ÉëV'õËúOWìÅväùŠùxIÏ¡ù‚ã—ösÞäñó:íÁúæñsñð:ýƒìëf9”õ!ûƒ‹çW%n9îÕu:矿¡¼Í~Çù—ÇÍëeýÍúáù!´.%}{8.Ù/yýÄë|%?±Þy=Ärrü†æe¶—Ô—£æ=#ü‰5üù;¼N%ògÚðÚÜÑüÙ:¼×òðgèðzž÷ãϪáÏóáO…àGµÚS=üù­·äµ-ýçã°>þÄþŒþôޙǟeÃïAàOס,ø~ø³ixÍ nõ?e7Ä!Ž®Cþäþlàb‡þÓ{øSaúOõJc¸]ʨ_ÿß0ŸÿÀë‰ø“|]¨¯fÔf0õQBýô£ýø3siÜø äµ ýçÛð'ôðgÞJIø“.xmʯ@7'¥„¶¤KÔÛ¹4”¯#àOÎáݼ‚?ˆ?…‡?ׂôs<ÉÑŽx`FÍ$]–Pü©@üÙ?üù‘*Ú2âO^ ò­¦þš“nQ_¥ÕÒðËx· Zq#µ=QŽKÿiEüù¿ƒ¤t}ÞA²t'¹ð' ç“}pÌ-òºå!$;b—´“þ“u¸ý é ë~D}á5ºÄíZI¶ÀŸ8:ŒÆÒ’ô‚׆Ð×zRßådW´S{ªß“ø&¯ êcÄãøsvÉæhÏí’7^ Óõ[Fò@ãÄzøs‰øþt%^ÛÂkØø“Šéd wéþ ú?@cœAcÊ!ýç’®Šˆ×%´o ɾŒä{…äů¨!¯oé?YHõG“ý¾‘ãÀkº-ðçÑç$œJ|[“܇x}ðRª×šê¶!£žÐ_®$™§í‹Hx¯F`~@½¢OãÝôóëȹdƒ«H‡x¯MŸIu‘7~Œ ÞÊޱéíþ‚CÆ-ö±1Â!}sÙêPÒ[:Ö"Ì®iÜØ÷Ó¨«}Øûáàët ?Ô7Äê/å¼XÝRn?±]nO‘7ªÄ ÙJL[+écòÆ›XüŠÜ¾a¹¤³…¤W](éœÉ’^CíÖÍ×ï\‰©gÉí[vK:S.Äm#ˆþ-éô’ΣþŸX.åzü¹=e©¤÷w“tãöOåöã¯ÒøÆJúü(¹Näíù›%]Dò½¼S®C®/·Ÿi$Ç9·ˆä_%¿8·Lïgɹ®­¤«Ž•õg5&¾×IúÒ,I¾%é ½%Ÿ©òFª˜Er,ÿ^ÒŸÊãOu“üV“ôùuEˆ×I®—š ç §l÷ìKrü×!;}nÙ@ò‘\7’\×ÜIú~WÒ[©ÿ)Ù¤×'i›ä^,o‹¹c$½˜ô4“ôç‰DÉ.‹¨Þb‡¤ n’ô¦Aćì>Ÿý‡äç–ôò%dŸÛH®Y-‰’¼·Qý»IŽä_wÌ‘7XÅ•ä—sI¾¥äß·õ#=‘œ×ÊïZŠ…äSÈNs·ÒøˆÏ’{ùÁAÛó%Ær’žî¢ñ]Kþ0“êÍ¢vóiœóIßwÝ éõÔžãðZÏì!´Mr\Gt*ù뜒Þ<…¶)žQÿìwÊÄÕìÔßUdŸym$NãY@í¤í»Hž¹Ôÿtü:Ýù·^&þÓéøMd×ëÈß"»,t>ˆ^CòrÞ¹”øÍ"þ hûnÿm4ž›¹=¿h³¤3IþidÏ«I®›™ñ½ýNÒÉ7ŽÏ[Fã"¿¾ŽêÏjFÇ9ž¨Ý5ÄÿVûãÏDPÞ¼’ú™JúḸ‹òölÒ÷ ²ó½oS¨ŸEÔžíýÅÇÔµ’.&¾3IŽ»I/Ó¨ÝÒßBâ7òȼÕ4j7‹ê_Åñà ½ÐöÊ•4þo%½™ô4‡¶g“ü7‘¿Í"=Ì&½ÝFÇç¿L_©Ç¨Ý !é=4Î¥¯w¥“–j7‹¶¯¿SÒ¤ß)ÔߤICH_dçÉ¿–\Jú£ñ›ÚÏÏ“ôÊ7siœ? äƒÍbÙz:Nùj>éýž»$Ú’ôArÏ$ýÍ&}]FǧÑþ 8P^˜Í”Æ3ì°„ô2ó%åÕÔÿÕ¤¯«IOÓi|G³I3ÈoÙŸÒ¸îüˆ»É_dºªÉ3w>Nã"["Œ×>Ü7¿YäWSisÉη’WÓñ™ÔÏ5ÔÏ<¿~8/,]Dãb}’½fÓx&R»4ò¼{Ç#å­9”ï9¬ÔßZªO㹎ä›C~¼ðªGóÚ×èô ݾ»öÑ}tMž:ø¼hh¹þ4T^? /PŠÞ•òzá byqŠÀ†Ñûx¹}"Õæ–ëða¯}Ÿ—ëõ~Ód½“å¢j­<>øyݳϱrÿé“dý ×I>ÜÏ‹äve¬wõ;†ú=gµ¼þxÚjYoÄ3’Ï©Ëë°gÞ-·û •×qÏ ñŒ¦ë¢ýè:æ9U’ïHÊß}äpQµI¶5WRÖ?­…¤Ãþ–üÊH®ÊåõÕÓß§qŸ¡ûËëÈgNý9]^o0Znó8¼'Û³œÌ—ÇÑw¼ž{ÊgR§Mû¹Þà…’/÷ËöÊ_!·Ï%½{äZ,úTîŸLvãögÿ(ÇQÚ’ìBüØ>E$çÉEò>DÖg½ªtùËÇ~Òé›ù²?±½úµ“Ç^-ë³Î}CîïOãKóÕðÝò8ÛeÉÝ—äd; ¤qŽ¢uÏê·Â!·¹ýƤGê÷Ô_#íÍzaù˜ïÈ+%¿³n•û™O¨Þ6ɇýõÅú`9‡P\R¼pœŒ"}•-‹ŒKŽÖËÄÇd}®ÇLJ¦ËíÁ/EŽ“·ûÐý§ÓY¯¤¿aϧPÜIr Nãü$’žEúÉûžÆIr±<|üÔ¹‘íú“>Q=Õ/‡w–rŒ¦õ<Û—åe½õ!»q;_åýäÇR¿´})­‡û‘¼l‡¡doÎs¬Ÿ“?Œ ΋ìCiÝÌqÉíx¼c—Èö,ç §Êã<Þ$·«"~<~WÈ_hìÅÔÏŠ#¶Ó9+‰/Ù“åáž·¹þÀj¹¿„hÊgü³5¨Ý`÷ªw2mŸLÇÏ$z µ;ƒhÕ§'‹E?¢gåù Ñrjן¶‡8"ùk)ç™´]Fü†R»óZFòg9û‹Èúg0%~§Óñþ´Ÿí=úNIŸR0qVËÈýCi»m öÃx<›é8ñçŸJÛD'Ò~º{&†ˆÈq°^G‰”{¨lWßëÈ}týÿ‰:jž_À žÇð">.ñþç{€gé޳ĕ!Þó¿ªÚ¡—!yÿ7ñâÞ¼Зx=Øè¸×!ß5yÞ!ïµâs#‰N‡þ>ŒÎßUÂì8®Úq ïÇv¥ÿïqÈû¹Ñv1Éx:m_æï à8Î ž—’¼x¡'­©ô?^Ðyð”C>£ñ$ñÀ'µnò}'Ê÷ãñ^;.. äÿø††“"^èÃì‡ò&=¢.†9ä½ÚnÔ/4å‡k ée¼˜t…<ðþ->«@ïNé÷‡q?^Û]xOû4êÿ_DïÀàû0§;äó!øL Þß§ì¯ß/žF2 ¦}xaj( ßîÏsà}k¼<ð1/PÍ¡q OœIz;›øL¤ç<ÆÓ6êàrê«”t‰ÏI¼AcÅ!VÒøó¨ÝL² Þ)8—ä=šŽ!?zßı”èT¦i3–g6õöïA|ÐKäq|Fï÷6ÓiToý¿žtˆ÷äÑ÷ðù‹ˆ7ËŠÏÓ Oñ{F¨;ôýÍÈ—ö½Cgø 4.Œ“[©üu¾ZBº9Œd8…€÷øñ¹”³I7UÔ×õò ÝFøü>Óð¾C>€Ûh¯BêûBâ‰ý¤~ÆPãH·Õ$+>g„þv•<®¿†mVlç‘LÈw2µ™HüPÎu¤SÔÿ×€ûiìÿ•Ô7ÆûÿègèÛ Ò—§’þS¿èÏWP?g’¾Pfô+¼Ð+ò¹ìŸQ˜@Õï2ïÃÞGÍ|´Æ!ŸuCÿx@wàôwZÑïþ’ïáá{ ºoã eÌ}øÌÊK=ö4|–èmz_ý ŸµÃ¸(ïÞi8à~|'óÎ ÏË÷ 1éïêáûqèc˜v@þ?Šž‡úR¾oçxÝ¡?˦}L>¶Ë!é<Φ}ø»M0Ÿâï>9>”ãÑßÓÛ ØäÐßÕç4|Ïó&Ì¥øþ¯ÞÇWð€q»…þoH§6xgåû°w¯¤gž0G¬’mjÏÉwõw†«¨ŸÙô<¾Ó:[¾¨¿Kˆ±‹ë|ç°L¹X;]¾wˆ¹T¿ß Æg»fÝäПcÒßÓÙ4Ô×ãòH|öOëJzC= 0†ñ&Ð}ÎÓ^Äq“nÐ."ùaŒú{À¸yƒxàÍù™4÷VJ{ã{’øþ!¾Çˆïêý`>½AÚD;\¾Ë¨ásgcȶø.äLíx)£î;ø Ûdkìë]ÉËñ#É~¦¿‹s$¾'ûƒä…ïBâܫˑ/m¤Ï§ “ó )¿†ï‘®% ý–H] Žõ÷yÈ9\÷Y|Ž ýýU؆õŒþÎe™ô|÷T×>k‡yûg‡||‘\«à»Ýz>þ‚üoZ蹎i÷9ä{å[i¼x³ÿÙVÎu2ü çÌÃkÈ÷šQ\¡¯ô'ßj/×ZÆ×RºÎñÚ;®?`m¥ás­“(ÎŽú‡C¾k kÍ/e×炯ȧ0Àjß{ÕçµßÉÞøîöþ’¾Ö¹…Æsί9ä{¼ÇH¹õqà;¾M¥ìúï ßKcE_BÝ;I7äOøn,>ãk2|OW÷ß¶Òæú:ð=òÇödK\Yã{®h hË–$ûòM´;òÃyëKò§ÄçØ/ÉŽ¨sœ{wÈ:Ž_‘RýIÒ¿ôuÒD¹Ïù¶Ÿ¾¦ÀXü„ìŠÿcNÛM|°æ‹;iû9i}ý~3Å ú®Ÿqý„ë;ÈYø ín‡žø]‚å²¾.Ãfégº?"o|—¼ZúŠ®ô›†äSm2>·Èvzœ’uôµÛ.)£¾ÆÙ_׋Cæ:Ì…+ÉÐþ¸f”÷ø;bm’×—äe/‡hC×#ÛÒu,ú’hMõy›i{¢Üž¾sâ«ÖoËü¹ÒO'®OûÛŸöÔk®O|x»ƒÒ^í?4.e¼m•úêñvüÚ+úh«ÈÉúëÈrSÿQúPú5Ò›ZŸõ¦¶ãq²:Qÿí¾m¾mLø·7h§ú÷ÓAÛåaýÑñŽêq–‡å ý!»(~Óžý@áÛžôÁýµávDÛð~…?×kÃzS¨êOü}>u¿*?Ë’ûWü9döÅ>lWæò{â²ËKÛüý ŽgÖ3óc?åq3ÿPP+vWý(j<ŠýQôÓZÑ×Õc~Äÿ0¶‹2žP?¬_EÞÖì§¼M|;±}XæGÛ‡(üáã<^nÇT©¢Šÿ…ÆËTՃŸÇÑQw;e<Ì¿³j_Ö¿ßí9YOªœªŸ©ý±üíý„ê+vŠ¯ÚŸ¿¡þ9C~Çñl0©íx?o‡æögŽ7Å!¿£úQü•zQ~¢øykE.Uoj|q<¨|Bý©úUÇ¡úóSìÊ·*_Ö3ëMµ7ëUoµEϪ~ÔvU9>!}øITœ*ãë è#j]Ä|YŸ´j§Ú‰©:?p{–[õoå8ïg=wPÆ’K3æË|”m5vUü&¤Õ/é8¯ûÚ+ããí¨üGüUês?X?U;ºÂŽ…#ü¸3F[—R/\VÍ€o,]ÇÓQx}•§jµuŒªŸ©öq*mÔ}Zœú±úԱǪú¾j§ðþüÓ(&cÅG,QõËT>jì¨ÇÔÿÃûåñb*ÞXÔ‹BµC/!h›÷»Â÷)ò„û¹3ƒ¶F9-V<©q®ê)–^â_SøÇówXÿFüT™ÂsTøxÌü/^\„ýï&?ÌÆëžs‡Ž‰™ñ†|cÚé=‚þsO=o¹âÿÞI“‡N>om¹&OœL•\CÏä:®I#/¤}C' =y$×÷ž5rüÐÉãøà¤ñ£GŒœÈ=#GŒ;”6ÜÃFÍÿ?kܘÃ1ÃÎ;ú\f8|Üøñ#'Ò–è˜s:?k\H¼¡“&3ÉGŸÉÕÆ:y䘑|Ì7~⸰!z`ˆ8–&Ž» 3« ¯ë>ýï¿{Î]ÿ’ú¸ApÄÐÉC;š8Tç»Gaç?lø¤‘ ^÷n$yú8j.ÂÆX¿3šËx à‹¿ q€ß“H³‰tB†MdÆA– d'€ahH6NìgM°?¡i8À&4Þgn†æqpZØDË$ÑÊ"I­´±¶„vÑÞ’@G’Ä¡Ià08Ü:›àˆ88Òº(èjGYÄÑpLŽã ÐÍÇÛÀ I¢{œ˜NJ=,¢g„r Ð+äZD^‚ÈWP m ÈÅI DAi àýörBo6P© Ê"úXDß$Ð/ ýmŸõåg/(8UÁi00 ø¼ÃéIàŒ0às9CÄPÃÄð0BÁH›eų̈Œ¶ˆ³càc‚±qnŒ3Áx˜m`’ML¶óbàü8¸€p¡ ðÄ‹-â>Oy™\WØ@u \iWàjS 0Õ®±‰k-â: ˜fÓ p½n°‰a˜i7†a–‚Ù æÄÁܸÉóÂp³ÜbóðÀ"ð¹ç[ðHÁm6p»Ü¡àN,¶»w‡a‰C>ߨâ^Üg€¥q àïµó»Ö"Ö)XÏ;ä7GãáE xÉ/‡áxÕ"^#¼nL°1Þ0Á›qðVÞ&¼c›Lð®Þ#¼¯àÆÁG >¶ˆO ði |f‚Ïà³ê›M°Å¾´€¯lUðu¶)øFÁ· ðùø„ïß~ ü†ŸÂð³ v*øŸ vv+ø…ð«~Sð» þß~‡ƒñ ö(ø76Bï¨ÐÂà4+ÜùÛŽ±à ƒ/ü  ‚q&ßSÑ‘™aÈ2A¶„¡a 4RИ°_ 4‰ýã iàï'…ã Íâ ¹Cþ!¢…‚–„V ‰ƒÖ1Ð& m íâ =¡C:†¡Sð=¦Ã‡‡¡s ¡ß‡ê] GÅÀÑ1pL Ç)è ¾×¬ÿ¶# qs¸vGÂ7Õž-±xÙ/ÆÁ]&b ¼ï‘ù˜=èödßbîuƈwZEööáÒj€×µMž²HdÛƒsq 8ßfýd WÛƒgIlàuzgÛ4zÇð=cFæ¿öи(9¤³†&†´‘È~Ê:ôu }™5øÿˆÆÙ‰ÃS‰ÆÎ$1:qdŒ°Ž†'›ÃÿV|¤?dŒFßšÃùB$‚íÌ‘•a >æÀßÄexÆ@eløŽ‹ôw¬£±?9ø—£Ñ à¤ÄÞÐ\kày36îˆìÁö|ÛŽH î9‘Èn4ëhø‹þJ¾Ö‘pß ¼Ï­à]‰M­Ãuq$ß[CƯÖÐp\8½[‡ï1Ñ5®tw$¼#¼Òøûî!À¶Á7‰Ãùc ðþHƒ¥‰Á9& nHÓìÃÙ!Ù í!ói{Àû®é™ÑØ"nŽFÖÇÖÜ?>ÒGZƒïoc:ÇGÆyÖáz"¹Ö‘õ—Mðº|ÖåÖàz]"«'èv‰9\Ý¢‘ÕÏ:ÜÑÈʳÏÕ±áê‰Ìæ±üÞ™;lâ;‹xÂJíÁÓ 6ÇÄFæ,›¸¶Iæp5 ã‘È<)ñrZƱÑðÌ çŸ5ÈøÖ6Y‡»º¬ïÑg¬®«$2–[GàAs¨¯*dL0ÁtëÀû‡~_ ðúªûèhdôˆïM±‘ÑÍÜ«£‘Ñ |åAkH[hŒô=°Ó">³Ï»±‘þàQ<†;-b`~4üëâÃõF4ÒzEÃÛ!ÎÏ",ôcb@Y/à=ÀôÖàj‰ôÎÑHƒ<ø+>Ò>¶€·MðH&ÛÀ\‹¸ŒPb'´ŠFÚ45†ëº¤Ê-¢‹ šÇ††„÷ÏH¸»E"ÐÒ®&‘HÛß"‚±áÞàñáì ü>bðX\m§˜ÃõM$‚e‘Àï62‚"áž‰à‰ :‚ŽgHø÷HHøn%ô©óñ^ˆW~$ü•ñáºKB½Ž¸UÂcײHn6ÁMaØŸëøÀµa¸*>œ["i€^±áÿ®Üøð /¬‡Å†ï„xö†× ÷‰‘ðÿd |¦Š¯ýøÎŽƒkà_c‚9±áPÏ}Ñðm„8áøhø6DÃlÄȽ |gÌ;-þ#´ÈaþæÞìH¸•pµˆ„ïgc¸¦FÃûG |/VÃuŽ„ï¥8÷«ïÑ0àó„ôüRÄþ›¢á=)>ìãô880ކ÷Š(Іg«‚G£z¯ÙI°Rœ6`ôÀ‰QýTõ›JØ-õÕo¬¾“)õ¥ï½EÿõeÃDÆ’*þµ![]è*ÕzO¥,ÉŽyoñ±ÿQ›º«ËR:Ùí°7ø@²¥.åßÛ㩮䬫1Õ–Ÿ×e©ë¾íØÆðv‡ùùF*eJŸDư7ú¼YIT&+úID†úò»%YÛÕ·ÝS]ìúImô•¨Mþ 1›h~3âaWÆDâ2•%UzOÖ^ÿõ8UK}瞺šgSQ’ñ«õjÛV×hÉæh£R×¶«Ï8Mf\©öƒTåʺ*fýZ=f&³êëÉÜ oŸlþ2â—ŠXM4¾ë"?Ä+ñôo'¯ÙÕŸU›¥B7µaçTÈ¥Ž/•öOdýh¶V°Ó¯_µ$š7÷Æ¢~C;ը˲·È¯Ô¥îêÛv±xíí>ô_-õac»úÜÛý6¥>ÆóYçvåLåXëÛßSa‡T/ÙöñøÖ·Ýê£Ô¦ž’‘!•cHŸTÙ®®mŸÊ8©ëøJt|vJªÆ—L_µQêÛævÆYºOEItL‰ò®+ýÛ-{ƒ]Ô¾‘¡6dO•=êÃîñÚ§RWuÁ+ßDôeG'Éö_›¶®íM%ïúÔSªJ"vI•ÿØéËn©K;ØÑG"rÛé/™±Y­ou*s@mÙñ¿«\R!W²>™J¿K¤Ôf‰Ä@*|Æn»xû­òUÇn©­ñqÉÄ{Í¢|è.ÿ¼nѲã[œ>bísIÓª)KV]»áCZzÆë·¾¸érÛ47㲿۶]f™VΚùWÞ§¯X¦¥/Œ¹»÷²mÓ’kóg“wuÂ4?GûalÇ-¶iÕÝïõøfÉ—¢ä“!_o­\Xk´|ö¢ÓÅ!!Zðáñ5n7£ÖhY³a›ª.¾Ä6­ìxÚI}¼8å4¯Û‹÷Þµû×Z§%û_üî“ý¯Š¦¿e—>vÔQôòÚï®XýDÓò?ƒŸµºà!CZ™¾«æ¿N¯5ZÖzäçêEKÿzýÜç‡hA¯yßÖ>6øÃµiÇÕ-qµóV§LË´´[ÏÆÕ¹Å¢à"L ¢hÅÏ'|ñÓá«SFóíÛ=ýµ£éK+^>áÔÓk–ü{ÑØÏz÷­7Z¼¤ó/÷0e´Ä·jå}W÷¨3ZTqÔ)û}s°!-;zö~3Μ™rÚ³DéÔíU).XraÊ($Ç«w¹¦…ýLspSñöq]~>a³1ýðûm·¬9¸ÞhѸ îgZw«sÚëµ™‹gøwÊiÞ×Ó|8ª_Ò´`[ßÀ¶%L _÷ú+‹¶M‹f˃%IÒ´÷Üi7¬l¹Ó–¬k:°ù¢MÉÓ´cqE—2ZœìåW=Q”4­ü¸XbPÍ]Põn—ç›ÒÊñ|±¾Ï@ËT¼ÙlûÈɇY¦%¯¶éšï>J9-(ÚzÙys]õF󺞿Ñyç'¶iù‘_5Yxý»µNóõÓ`­ÎhïÓÎ>¢bH×”ÑÂʱ'¯>òô(Z1ù°ö»†•[¦yüø®û›-ÓÂãÿ½·¸ÇѵN‹—|9üpÿ‰)£¢÷ÛçT̨3ZôàM;÷:Ñ–Õé„kº©uZ4¦ø–šr*ôÓ–inÏœK:ÿZç´ø—÷ËÒvÒüõ}N{~[SZüûçE׋¢íÏuúa×ê-üýŸ'?mÜטöÊjðØŒî)§•÷]wûÂV,~ú×AŸ‘2Z¹éÃôÛV䤌–ikg¿òçwuNKwä?›}Ïè(ZˆWûšrZ2´ásV7½î^:üš5{B4ï±!ƒû¾Ý–ðu"*Võš?뜤iÁÆÇ3ÏÈFѾ‡ã…b7ïäcn?~SÓ’_*žÚÞmA-ì‚&:×Z^3¢UtºéÄ?»Òõi¦°ÎhQ÷Mjá®uZ¼ÚueëÙ¿Ò‚‹>ùH·]µFËh=Y[´¯*õ¹È¿êÈÇ»oÊhާñG?K­jyˈ3}_™ÒŠ^ë¯8¶Í†”Óâß¹ÿµ¡/%L ÙŒ®ð¦ŒæNžúùªÎûÙ¦ytÞ›(-}zѽ͆mJ9-¾ms³@³Y¦´dô{óoõpÊi¾~º{E­ÑÞÓ¼­öô<7å´ÎJö,hkNñôhÐÛ´´å =_þáåÍ™;èѾÝ3¦•t¥ô‰7~ÎtÏL-ØžöìÙƒ Ñ’ _»}«V‰Â[Ž»þÞ¿—%MK6,ÝÑtËrSZQšz%å4ï)%{^r˜ÒŠ7n>RV­Ñ‚æMÄ{ Ó-Ó ÇaxgÕæ¼vôŽ –iEŸkï(ø-;išg/®+k^²åÓ…__Ÿ2ZúWSmÔ'¾¤iŹÇáŒZk´øŽ?GÌý*×2-ª_h´LKw·z{DF­ÑBznC¥¥Ÿî^T6rqÒ´0£MþðS·FÑ’6¯ÿré¼CZ¼àÁé»nì`HÅ÷u¼¶é*Û´ÏÚ5K¾4¥=è~–UÚ{ÛÎÉ_Ì» Šæ\Þÿ“o.ô%MK§h6|eÊiÕ¥Ýq¦Š¢E²è¢œ”Ñ*ìnÅB‘w{Ó‚Q´|eù†K3Ÿ4¤Et^jÚ‹®W%K ¶œì_ÿᆴì‘9÷ç¾ywÂ4p[Èä…)§ågÝ:ÿ÷æãE jcìò(š¿JÃ+´¢Ãg廦4ï¡ ‡<åÿ9aÚûâû0sX¦…›Ï{îó¤Œæ¬Á´ÝMiQ¿7žïz÷RQÒOOH-úé´[‚Ë lÓã/ïè¸6iZúÊâ£GœpxÊhI#}âKÍ{#ëÅ–ÙïX¦¥Omï6ÿõ^¢ £tÀû–i>]ÿ¢EOÜÕø¡)£|ÿ¬ìÂoJ]<Ô”æµ*y{Ç“¦½èzŸ-êÜ ¯™Ò*\-œò«!Íùúô5T7¤º¦Å}0Á~dJKsÏûþ©Õ§Ò’÷ßvÍÈBQ2¨Óº¦N-Ó_tÒ‚®o]ý˜»MéI3üë÷’RQ6õÐÂ.{3aZ@ÏÕ$Ks¶58–Ø ÓÍI˜áßίšÒÞpò|êãë¦}ºŒ^úìIWÚ¦½ïY¿¬ÁÔÇ¢hIŶ\¾fÞ~a-£¸H”æ=òýÏ­c iet´Q‰Ó/º¯íÖðø-кÀ˜~©ßø QA÷Ù¬Ò<¼Šã›fHKWöóº‰ ¦…ÿÜàX1vJÍYXùrÆå!Zvéú¶W wˆªæ‹Þùíù§¢h_|}¥Õ= ÓÜ^Ú|×­= izý¨+-Óžžìpï›oY¦å”‡TZŒ«ä6¦åø¶T¡¦´Ïb})J:ßú¿‡¿8Ñ2Í¡_0‹¢9ã~»ù³w¿6¥ ¾À+†´âÚˆ–µ×/Ú¦UO7iôÇ£íDéc3NXÑhcˆL×_H°MO(lýݲùM£hñ‰“î™÷Ö£ ÓÜ®ÿ nÛç:Ë´_8é“hjg*ÍípÑîq<†´8ãƒÇ¶í'k¾mš7©ô‹Ì­˜Òb|=tÍ’(Z¶  j0D{ˉÐ6ÍY¶ó›OÉ ÑžtD¥¥{z‚Eö‹¢%ƒ>‡ŒõHÍÅÛ§Ö‰òý¾9NmC´`¡þB@íEÏù›ÑÞýðÄa³!Í¡÷Kó¯»âÆOf¬3¥e]õŒ!ÚcÎû-Žœ6Ä”ö¦u’-o7øÁ­T™Ò |ü鸧Li¯7ÁŠ¥™!ô|… u8\êï4Ûý6¼f³íÿÄ{§¼¾e«­qYk"ß'¨ïqïÃ>üWQßñ•ŠþkSîd¿Rúª¯¹·Êµ‰Û°¾å©ïñÿW|6U2&›ïöVýìCjm_ß²&*"<ë{\ÿeýíã©o¹êK{ïºÖ]Û>Ûç§û°ÿ·ñ_^wý_Á>[Ô¿Î÷6ì 2ìCíÚ7ÙóGµ­Ý<¿7ù{- pþ ÿgËíìŽy,ÿ3JÖpq xG >yÜDøo£æ·ž±¬%ú¸dãÂÊPÇ¦ÃØäp¤õr8ÒRåPg tû m¯†íɰýôºÍáð_K:ö2»øî8öˆÜ­] õ·À¾°ï]ècí¿ö·üÇ`ž·A†÷‰×Óp|*ð:Ž=òí/å÷?@ÇŸ„㣡-Èë†x@&ÿ6GMÁ±]ÇwÀñpüU)§Æ< Æø(Œñ‡#ó\ª¿’èSÐî*è·=´ƒþ½ Ÿo.´íwa\™À;{.Õ êgGôáN0¯èذï,dõAÿþÁ ƒçA†Rà÷6Õ}ê^uÿ‚ºK¡_èß7êÁ¡ ðHƒ±e}_ãpd‘Žt»,ózØwìÛ<އþží ¿9ÀäO;Æ m3‹` ?; PÛUж7´´ ¶Ó&¯Wì?ðe¯+×?Àk<ðúdz*Ì^gÕè^/`síàñ#ð<À?| —6ŒÉ <†ð‡Ì g}˜Mýû ÍÐækhSmî§1€ï¤¿mŽ «þõÜà·ö[,+àØôÿ£Ò.äÙô; ø¿È•‘ºø!¬Í¡ôÿ¤ãÏw.´[m`·ÿ2h÷=ÈöIÿdñ†ùÖÿêƒm=ÐÞw0Ô¿êï†ú0îôO ~ï2‡£ÈÛð>‡£qË0`ûÍÐ62žíÁg@_0®´ÕTï¨7ê½Mq>àƒ8ñßõšA=°Qú¯ÐÄsô×`t˜\ËN¿ làéý=|@ÐSl—ží'SÝYäsàΙÐ/ŒËõ¼ÀËþÛ§AÝôiT|Oƒq:?ƒº0&ĉâßÿr˜®Àß\`G7èÀ‹ùé8¾xu—ñ™þ– 9"ë;ÃÉ +²ÚÇãt_AÛ#¡ÈNèà ¶ô€ }‡¯‹`,Ð&-ŸôÑx=¼Ú: G5ÚS#«³!ð„ØñÀ¦r¢t8dñ¦Clg€ßeaf=7€8j”Am×J¿×Ç qç…øñ͇¶‰ƒã oÈ?'‡ÅÃXÊÿ^»¿ÔÿMç|‚qzr¾uŽ•þì;ê]NÇ1.ÏY€ã[žÐ—Ž=ÇÀ7oÀ1°§ç!ò?ô×÷ÂülîdùÁ?´i”w /y ßÑ2gµ0["¿> ä>üü&í8ß«TüPƒ|â{ºÁv°‹wðj u˨äF×´šÿ5Ð¥ø¸GÒ¾¥R>——ü¦ íǼ öv<ƒ ÿÎaú‚8t‚_¹¤ù¨ô ¶ ¼¹$ý-ªû$ésEÔ…¾|­¡.øsê¤Á¾tÈÙè÷ Ù “†CÀÖy] Ôõþm¤cWµ»àX)ébÖ¼‚W_à•9'dÉ‚9¡A_jó:´u~Osø‡Æí_íÀÇÓÀ2šÊœŸÝÚm “ç>ÈßnÈã^hãƒy>~}¥³ÝpN¹2,QOÿ“ºö\f³‹É'{ʧÈÙÚ=0ˆ#÷½ÐèÝrøÁσ¯Ó W¥ƒOe‚œÙý©Íó”—0_Cùo?š3€_ðͰz¿o˜‡ÝЇbϺñƒlAÈÏiçʹEŸ†qçÈfÚùÀ{´90Ì/Ñß 7èÛùÍzöC, /Á¦é0Î ðÝÌaaºƒ}NÈî~Ð7ŒÃ_õ!.Ò ïdC¼4Ü+ WçWPdñ<¦KŒy°¹ûnZ›ÀšÁë‰ äš4˜«ÒÁ2[‚üGÙ0Ï4<‡ÚbŽIæëЮ“9ã=TV‡ÉãvÍ—›N\éN< /¬O|ÿ0OÁWÓ!î2o†>ÿ¢¶ëeÌ8y<ŸRþÇ\}zþ„ö0¯ù!†À3°…êÝKóЫdÇ;¡^ÍÁ‹Óx>:ÌáÎíP|ËkÌG~Ì‹Ò. ÓÃå´Þ8;Ì“jÆŠëFç6š7!F|Ýe MÄ{ú¢0Ý<ÖæçÈ'ž¿A/×ÒöÃ4~È“î>p æŒÕ¼‚¬ÿ§dL‡óÖJȧ x`Nõ‚?ûO‘k¶ øBø`F×r¥Ý཰¶W„ñz–òæG2oz 6¼ßÑzƘÖÆ:Í€™97û>j‡9ÖÎv´ý"ÙÆæ†µt냸óÃ\„ù# ü9ý')OÖr®ÔË:ʯè¿EÐîYZÏ€v3Ò f€L™W²yü Éì¤mÈ]ÚxZGÓšòhâ¹ü/dË¢u¬–€þ\'@ýÛ¡>ÔóÂzÅ8Œô„µ~:ä© Ža,ÏS^ÞFëЗâÈ>l'׉zy•ì‰óððÀzÊw*Ô½›@=Š bÕ ùÅ 6KŸæoPß zr_N¹¶·\ÿ_:ø[ƽ5~¤ÇÉ4fð#äFÿr:þ„Ì¿úãêù ù7RüWP=ŒÝKin=ô:ö­ ³ÍE‹8æUd›™a:A;À9‡ûynäƒ9Ôë¢`%ôreô¢º8gÂÜíÄó%Îë›\à§nˆ/ØÌ<°n ‚Ìià‡=¨î 2ϹšÈ9× rø@GþO(׃~3Ürþ ÕKçy“eÌùÀGüׂÕPrEæpª‹ñó†K@]ž³WКí Z¯ÁýÕtlчhŽ›¸¡oÏh97ú>óŒÈ.r™m~ ~ 6帾5¬ Ê3ø®¢9ã]Z@œzRwéüÍEó»>p~z’Ú4€6Ci#Ó!ö3;QûµPòŒõ¿)LV˜œ½WÓ\scÐ6çñzûòAœ¯!¦<¬^ï|Æw­!ð“Ö:Úõ°mú­U@>?¬c0??¦¶«¤mœà‹îv2Ï{£u3è3 l”cÍä\ù£òzò’Ö2~ÈA°w:¯–Ñ:ôgƒ^Ÿú @®Lû*Lî´¦{‡ö-“vÇ÷?Üà#ž(WþKÇA×N˜‡\p.à)’×BöÂkc0»AF/øoƒÌ9A¨›žIõ ~5˜o\ÇÊ\*óeÜãµ:}N‚üæƒüæ‡|»§ÝCõž%ß…9Ý ¾êv>>o@˜+ϡܨ«&tM ×­©ÎÓòZ–ëIºNkøFæ—àiÐÏFª‡kÄSéúÛÑtÞÕl y0¯K­¡5®`Þòð5H,+åÚB—gðûÀaí‡×Ì÷zûk÷Ëk¡rX^,^|°Öñϧkf,Ï:Š…ÃdFþà?Þ—Âø-–ó¨û²7ŒË9>À×aî¢5ÕP|Õ³êÀ|6OàúÖ Nð ×/Pr¾ï ¹Î阯»A¼j°Öv‚O¹aìôˆu?äªÀÛòzAÚ˜0¾ƒè<ªˆÖŠÐ§ÿ º¾Fç®èÏzYOsßf:wÃkF`?äþ`.ÕYNþ†ýw§yæ,ÿ!tüVZ® [ÀÜ×Tœ?Ð5jœ×`må¿™Ž¯ ¼¶Öq0?ù)®àû®5>’¯áÀ|æ†|ëüãÅ5Ω÷´£h\ œÓÝ0NäZ_yŽ<˜øðù ò˜ððzí¢Ñ91¯ÖÈ~]ÇÓö*Ò×gtþ³<ÌõŸ»à|™Czi)sµòžææøPZCª÷>åi°“¯SþÔºÈë>.Œs¾¦ý0ÕÅsötþþà5ÎJ’¯Ÿ£¦÷@NðÂzÍ: pþÀë*x-¦]ëxÖç0ž ]{ÑçgàáâÜHñìb_{\ú‰^6Ñ9Ðï´½‘òLsÚÞEsÂÅaãXQc½?¾Âãx6lý¶w>Jÿc¼¥uN/:÷®’÷-ôö¢æ|ŒûÑ}×ýéšß€kô©·»’Î}ž¡mÖÎõ#.˜sÝ…5×Ï5˜h´qÐæª‡¹ªZ®i\·ÑºÿWy.²­ 4¾†Äºâ9s^kÜBç`ŸÒµýÏÂtë4']kÖýw²¼žäi"×ÜzyQž¹²iíÃvZ*s^`ÎÒÊé\üÚ¾æk/¯ …ô±7ö?&/ä8çtÊsä5k½€>µçèÿ5Ò\EòW/oÑ:4üZã…aÛëäZÂ5]®C¼Gþâ°~—„é|Ú‰sÍotbãÈÇóâÓ´¶ÂyæaÏK4ÿ²îÑ|€çÈ#@ûaÌÚô?øv»<Çôô¡>2ŸyxÍ×ì>–:×ׄÃd…üçtËs%½,¦5Ù‡t_âßËk£ûI¦ÏîGpF›@çúyÖ·´cÅ)ÏU«ä\ Çõ¡´ž kϱ¦¯ÁÏ\nÚæö—ô{Kœ_ѧ§Òü“K>Ò’ŽÍ$;`lð9érí¦—HfE¾® yŽŠvõÍ :»ˆ.—sº^ðK;c3âaÅÆVc0V{+1‡Yôh$SÿjûýÅ“ßJ¬õm(c<9-ÈÏÇ­œO˜ùa_¼­»±«½‘m#ö)ý™ÙÌ–?Tˆägö|ƒ]Dð6ág”s“ñýXã°RÇNüŪ«¿DtſښìñúTå²R±{¬>ŒbÒR')W¬úá|ì+¹%UyÃp<ÕÑuÕÿ­öcÖŸ¹¬ÆˆV;LK¢1o%ìøšN«mðŠQ7Þ=¸([TÇ®g4¾DŠ™¼fqoÌ É`W¾jÅài´ÏޝXÑ#Æv,™ã1™ÉWW%}é9œ·ú¿™é׌W¼beÜVÚÅ:·mÞf}Æú?ÑbfÏdøµ í¯6¡:²^<>vóU"ö2âÅ7¬ÿTÄp¼xŠ%‹•‰×VíÃ,έŽÑJl™Å’ýZ‘ÓÌÌÆl¦s«y$žü±øÉiXª£yñN„LTGóIE1Ê V’¾Ñã·®K•,qõ_¿¯Dr¤]Ù¬Ž1Öµ[²ÇàmUßFñ«½¿x:%_\Y]ðU‹]¶#¿Ýq8âÔ·*“™œf|íôo_<þ*5ÒÝ¢òL¤½{#›Ã²=bðÉ£:²ŸX60å¡Ôµâ#j_¶}´Ú¤²#ù{Ív|?îX-ÈjVâéÆLwvå·ßÕñåŒåo‰–xñaæo1Ku$S›W×Ku‰5Ö¨myìð¶ó±dØÛ‹Õ¸1ʦz­6îÓJcì¶³ÏvŠ¥ñTÇ>Kþx|­ÊjIoŠLñÚÚáofŸð¾cé!ª­R7–N­æ¨dKmòNi©–Äj¬§ª?³Ïç˜ÚÉvŠÙ}•º*¡~«k¶W»y-îñjÇ^óŽQ.1Ú§î·âgLr¯Q^²S ùg¬Õ6™$Ó—Ißµ/fö´âÏñø¤ê\#»šÍo1ëWÇï3‚G¬ºÕ6d k‹Æmk¡žÙbÅ]Ü}qxXñ“¸¥ÚšÌá}ªýGµ·Úw þ±äHÆCÿW' ÐÞVªë¿_£ø¶ój}u_T½êHÞv棾ãÅ–•¶vó\"¼Už©ì/–¢b¹:Aæíb§¶ôo.M6OÙ•ÇáH¬+s’]~Q6¨6®g¨§mìʬþãéÕ0?T''g¼¸4…;SQŒî)¦’w<žñbÒ¡üo&ŸáÜÀ´:Nc™’)vb"fŠSßjÿ  }ÛgãÅ„!9¬äk+1fUGvüÁH®Xu,å‹uÍúL¦˜õçpÄîÏʱx}m[õ9³bU×j}+|U™“-FzO™,Õ6ŠÑOܹÏ&»º³3ñÚ&Ú·^ª£yªü3ñ^sÏŸåN¦f¥çæÈú†t³A;3jÂײ<›cÕÞûŠ]9Æ›*9Ìô^WzUíhd_•¦¬ßë7UrüWü:Ùbæµf¥¾õoµä¿VíQÛŨËóK’4Õ%årÚ̵=®Ð¶E´*ÕqYGR]Œä3Œ#9S-O²û몘ù±åu£R¯¶Šÿú^O$Z ×ñfë»çf4A½Ùåcêw©^o›Pµ¾m~ÿ1¿««y7Ôß^¦—„ç{‹çkµgÿ•Rgë9³ùÈâ|•òu¡MûšÊ¯ôóÿK±¼­åu›Õ’²ë› hªóLªh¢ëDe;U%Ùu‘íuz-Íïfç/vùXåoõxTýzŽ»½=ÿÕÕùÝuk¢ýÔvIÖŸRv>´—\ÏÙ[ŠY4k—¬]Ôþì^M•}Swêë|±®Öã¡í$¯ÿ%6ÐZÒwªìiÕ?RíG–çƒ~Må4 fý[õ[»çIµ½Ž³:î}¥nK¢vø¯Ú-Uç}vûIu}»eoãk¸þ¨§ëÆV×C‰žî-~`µßÿJþ6×6[«·¯Ø+ÿÿ0+ ç¯Íñ©ÕvõU’•£®®ë$Û~oõ˺’ÃöyE-Ï¿É^©­²·ø…ÕRÛÏ£¦ª$z]*eýZ¼/bø\T=ÿjõ¼£®žÏµk·”÷[ÛùÇ"­«þMÛýþ|—„Ï›ëë¼qsríR}¾™h1»¯Uslj»_›íS]¯®‹ÝëL{Û8j=ï3Mô>‰Åúµ=¯%+w²Ï9™Í£‰úáÞºÎ4*ÿ9ÊeÞ·;Õ×ùoÔñ$óLÒó^ŠžkJ¶Ôöº»®Jm¯ëcðÑßg‡Ë qLÙ({î\€eaÏíDÙÏöŒÒ‹poÐë‰t¢ÙÊvCY¯ç¯D™ßïÄçoY/Dÿ ã¥Û"ƒh6É·O“´F:Þ„úoLô(êo»2žï©ßm$ßWÄ·y¼»h¿ªÞϔǹ‰¶YÏÌ)÷»;rüb?'»‡õz(mw z$éápªßQ‘ƒå"} ê‘Ñ…ƒø¦)úVìÒkªÇÇÛ(ãcý±ž¸Ö×NEï¼Ívçí=ŠüÛ•ã¼­Ú‹ë3®÷‘ÒNµµ é#¨Œ»)é«â/l_UÞVõ¡Ê­ÊÃrê¿!õ(Éi$‡JÙyüFõTj'´-“~:Óv;’K(úP)Ûç¯H*\DYïjÜóþTý•íÅí»P½£ª#ü;”˜OÕϤíPþ"=7â|Cû9Ÿ´¡ú­‰iâjü«vP· Ú…òØ´½?ÉÏ~yˆ’Ž!ù»Ðvg:Îy]õµ_U%ïqá<Ò_gÚîEõÒ~_d\ ?Q¶3û9óåù€íæ%Êã¥ü§Ö åΫjdª^wTõÁTÖ‹š_Ô¼ÄqÂþÇò¨ý«òªý©~¥öËüŒÆ¹IÙf9Õù‹ç‡t¢ä÷Qó˜ê'fùë1öçfDy>ã¸d?*ödÊr²¿ð|߀ü¼‰b¿ ôÈòý¬´oKÇU{±=Ù_QýÆÔ¾%у•üМó7‹çí&|œøt¥ýÑ6ç5îçHe¿š÷xÛŸ!}6 í¦Jþ8„öFû;°]h?éCt"½Í¿Fó±ºm4ß1ßÝÊ6Ѩ¼Íãfûó:”òL¨>o³i½"NRä3¢ì/ªß(ãͬ_ž·Øl®×šè‘Õ‘üúQ³Þ”uR(òüÊël£|ÈÛjÞâõ†?¶=¢ò33[Ñ;ÛkÅOÙ«5õÛž¶Ùï>@Ö;‚ê)„æî—ý‚çÃŽTŸíÎ~}$mw'¾ì?l'^G³Þ8ïp=^tUôAõCóçÓ–T¿Kd…ƽŸ"7¯§ò¤§/&çS5ã!Êöäó‹# âT_ÔõŸJy1š—XŠCù\­Ïòóº‹ó%ÏÇ®ˆ”“©šOŒæQ£¨¶SÇ­Rõö‘ߌGù¯ÓsñÂõ±YÁžè‰Ëv“Ÿ€àèVêÉˈŸ8^¨~ñ=@¯ç}~¾E¼©ÁÓøð1êçè:x:O=òê{Aoˆ‹±»GÂ/§'/–Ÿ)r½)ô+?ÎGûvëJu8¯ÇqâqÔ»›U)ëyÄdpXÏáñïàóNâÛÌšuD½¥¿+ø ìv[ô¯_/Š~à7¨ã¡{è˜<`éûB]ŸÖ7}Ãü+xÎàféxõ{|&ßîí§þ€,ø±½õÜÞºo¯Ãòñ^ZÇžš‡|lkÖ¥ãÝ‚~Ôáèºp=âþDõºp\ì üÀªn5`Ç5ñ÷0_´ŸX—ˆzK\qc"ýWî ãDÿŽ^á¿ÉLjÿgôë뀿ŽòˆÇu|ˆç/ò—û9®©gøïà—X×ýaÞõu{É}[Gÿ‰ëç†ñèCUð¼ÁOÍGßÕï>#GÆ/êôcëêèèß á~zÎ×ùƒxßæ>OO|ý%ñ(~x´_Î…ïÈÿ=·íªçv>=?¡ëÄQèE‰o5xJ½™8—z ÷¯1?uè#?zŠzæCЫ˜¯c·±‡/зw÷Dõ’ºü+æ;¡~ZâqêàL´ãº¼ŠõjùÃò~€z8ï xïɺ©ß”¸]|ÞRÇÔϨCkýüÃÆ÷w¼7¡±ç6ùxG‡¾Fÿá |÷y•÷6[„ñ£ÿF?8Žù ü!>Xø<¢˜Sc·Üñ0Äoů”õ"â-òÞ]éÃ|ÑOÄ>ú‰wÖø¢g CÄY{‹Þ.rÑó䇻öÛm¬×Ôñ£ÿÆo“ϡ—3ÿÞAâzëâuú˜‚;èÍ#jôº.›È¯¬ ô1o¬'G?ãž‹qx]Üì«6OvHÝ|8¾‹øµ_°‡ˆ«1_‹ù|™ÈŸ×ñ1Þ7‘ÝÄ8*Ú?¸HœÎw ØË“ÔǺ=z¹zÞƒï†=þÓ߸剼*ñÁÌ Q?B øzìª訓c]|ùýMœ?æïô!/D®%¾ñBñ+ñ}süN ò¯®nãø¿‡ˆÏG«‹¿¢$nDOb<ÔEŸ´žX¯¨«?DùßøqµÔ5/v€ÿ'¯#ÏÜVý'Ï_ uÂǺ:ÕD8¾O)r/À ò"Õ+J¾Âw‰Ô}w ü‡/¡îPüÇ¥×yê“ʃ¾{;0ØAÝû¼ØGý¬ûÞ&ÖóšýôäÏ{]ܯÇqÞ]uïUãû?êõ¼·ÞIççûA¿ /è8>·„ïN¾³ ß×Lø,¾vùsm}Gþż*¾ßïïŠ_C_ü.âÏýç×G¢~¯ ôj"ºÃwq…ÿðo¢ïKâyú‰â–‰òê³tLœÂº‹Ü?Qƒ/uôoìú6²ŽT¾#!î¢îí¾ÄøøÞ~¹ü§ÎßkÆ8ä®ÀGâ ¾–œc|5Äãu5ëÀ¾ÂzÆ*O\…y’æç;ÅÁµïÙîÂ~Ôó>Šï׈Wɯè÷?,Ó3‘¾ÞÖWWÇâú?±N‹c÷è%õ,¾'$îÜ6èƒð¦à@ø. ¾Fü*þ§äMþNÞó5­£øUƧðï3ƒý® ãÔõŒB?÷ÕÕ'àG]ý>æ èþ8Ö¿ŽÑ+Ö»¥ÖÉû9ì&~ÿy3ï—U'(úƒÿ`½¬+ö¢‹ï‰gŠüáúG\çï"ßð'\¿AÇ׆ã8ÇêËz‰ø®MuÂBq;ï ˆÏàù'ßaσº8<¯øNuÛ`Ÿ7;â'óc䕼OOø’:æVÁî6öû¬ºø"æ‡uù|®ñ7â¿{ÁO’ó=Ç,ø©cÖË8虾©Åõºúb\ë—…õÑ£o±.'ÊóXúH|ßg>²Ÿÿž)¾GW¤?µÿ>èÞþu”À Þ‡òó'?E?Ñ×ÙÁþêüpÔ;ô=ØFóá7Ñâuê»å××uõÜ“‚þFùÐÇïWâ{äùwQßêôPú²þûÝ~¹—üþáw±Þ«<×óð¹Å< Úqø^¹è úUÞ¨ÇÞ„õnêw}‹ïS 7¼G@®ïx®î|\w¬ÛÔ}—Áõð½ÈX%~¯oøTü ö+¾Q.ßé„~aÐ3øð‹@G¨=O‰Oñ;Ä•äÄAØ/ß3|/Ý?©RÖIÝ„zó`Ô¹êò}øòûº8«ŒË{øÝ>Ó/ßýž˜óuõ~òü¯ô¼Ø9öÒôUÁ°šuûÆîÁ‡øtÅïûb½?Úcü÷~áßýnâ„ýæû â–]ËÏëý‘]šâÿž¹ª‹Ó½§Ù·÷üýGkw[û¤Žmí ~¾ÿ™Æ]­ý½UÏó_±öMk'Y»?<3Yý=´õ¦5£±ažµ±uï3=Ïîa혪jê-vs÷ªêÌî¹þ>ûß5Ö·Ÿ­néùû¿¬Ýeí:¾KÏo³9×X³u4^#ÚÆ¬b~Ú»Þž¿¿¦þŽ ¯¡zPý•ÖnÖß×õ\ÿaÏßß¶ö#k·úyk¶¹§Z³u7,£hŒæsÕ§­ýFÏ|·fÞÿ´vµµk«,›ïYûŸ|¾ñüpïC5cüºÔ¿L<š¶þžÆkÔ_³¡1úÇKò|µµ=íà%Ö/ç^ב†µÇºò¹öië¯7_hç]Ÿ¨úäÝ8w::ö¿{Wçæoý“ìk­Ýlw'¾¿úI•åx‡µïWY&ß­’ÎV×÷Ü÷ðœëÁÿhÎUê—X;>Ðs²µ¥£Oµv‚5ˈ[{JÏ}ÓìÛ}׊¾ßå5¥µ™4×s/Õ}Ͻ×Ú‡{Æ”Îÿm<Û2}gY{«µ[ûù÷îÓó÷­YTÐx¼µ'õbžcC¿éíÃx@ó[ûA•uò_3uïwÂóŽECUÒãæŠ Œïüš^³–Y8wyÏßó±Aëï鿯ý=ooxìêFõµvþ¾gC÷ö<ó§lÏ­Ý— kf»ŒÕܳ{Nî¡û™ÖžÖsüè*ãö#2¯’¼¼º³Êöù»]Æø¯p¬/ª×¯¹úy%ß&ØÉ}óµ5´îÆÀbûï6ÝÛœ`î¿äç«ëCkî{Å8c\`íåÖÞ¨Þb†ê’õ×›AvÍ÷m`=æsœeÿ»î›¥ã½­Íszlævë¸ý«*Ç6îgîë¹ÇŽk{æh‡9çôüýãúûR,âsr£ó†­ÕzŽ¿níkWUÂo=û½ü|c¨gŽ «yþÆÚ­ùïî˜û4݇ÞÍPÿvÝÿo¹o¾Yç›ûÖªÜÚéü‹­Ô3ÑÖX¬kçôcmÓ¢ïfÛέ³¾«sn‹¿õy­i'®Õ8W‡õ¼£÷Xçžî~\>üX§CmkWèž‹5Þïâó=ã42¾4ºÂ™_»MÛN×uã<çqöS5Ÿû]s_}:Ü÷úž¿÷íáÕmU–Ëß Ñ®{…)ÆúŽYí|=ùá×÷ïW_¶~—Aúª‡ìü·óZ’¯><±þª0‡ÙYÃe·¨~½Ù„s®Ã7fžU·T9Žöœç¥5c|ÕÚ:kôœ{“5Ë©ª×Y»ÔÚÖ>(ºn#ùi[K³Î'¶ÔOÒø7þÃÚG­½·ç™ç«ŸÇ Ï^gí‹UÎ;Þá§uïøŸôÂ}÷QÖ^`íè*ç@~üí Œý³*ËØóÇŸ›óß­Ïm`ìK³×ëYë›ØæAxÎw¬ú°žy[µ\ÁÎ=ÊÚ<µ…UO _%lÏ6/Òù›ªó4©kSzÆ›«¾ðVÇK­ýÊ}žµ¯¸[ûµËzÆ…Öó°é0Æjk+Õ?W¾ùpkO3œûî >µ qÐzL×±ÇL-k‡VÂÿ*ë‹ë¦Åo•Å™Õëóùæ i­µWÊž<^¶<1Å¥ÿÓ‹Ù=ôxàÏÙ–Ó¼šûþV¥º2M±Õë¼ϫRNÝx‚µÖ¶w;Ñ5i>ÇÔµgª\•c7‹­ÜÏ4¶ÛðýÕUÊ}f.Ì4¥ØêÄž1W…9,¯lhüjöèÃaêÍÏ5c¾ðÄ Í½~þ„¿ÈxÙ˜žµ9Z÷õŸËqubõ ðÂóˆK]Ç«³à¿»¼Ê~fa³ñÙãžæ!=4ÞžÙ2ó6aüñЫܶö“žûV„çŽß•m¸±¸Ê:ãy©çRîG¿ÞoËl¡íÿ¢è÷FÛ§íø°ªÇÿÔÐé>Ícf¯àº¾…µi>ºy‹ôéã¹v”dú–*û¨_‡ñ¯ä_çšžëÏ8ÖyLãy†çÉ=¾!ñù©ú»Ñóœã°ÛÛ—åOîëÒñÕb=³‹ëÇðÒ|{Êϼ÷xïæü¼ç-¹ö£ë{}ÌÚû­ý[¥k³ö¥*c¯ñ"ù¶7~:oF$#·Ë¶Î;¶>Z;¿·‘>X®Þ`|¯+ý8Œçx¹Ž±ÃµaÙ¯Ç1¯í9ß[Ãy‰pré÷ήãUªq ä'gÜV)&ëáçAý÷åøÛzÇüvï8÷flÉñ§Î™¬[óûÇ©¾¤Þë4®^sþ»¿5ŸÝèÉ3V…õÛ¹öÎçžsgIþO’ æªß1äv.£žø=Õ5Ýyíõç=çÝï@_­gG{Æ1Þ'?î¼sûq_²{¯W9f¾ÂÚ«uü2k¯ê»çÞùZë™UÖ3¯åx½‹8º„ÏéõÈ»«µ«RM¹†_÷Øì-:÷NkïñµXóz¯×š?)ž?ÜÃÏÙ½~ñÉš^¢þt­ÍßÂü@çÜ=ÿs{ôüÖí+Ô»·SÞ{]^Sãµ=ë o=QývÒ×…®Ky u ž¹Qóy}êž*Û%5Ì_©w;½½Çþä¿ä7o¶v‘éžk®4¯“-ÏF¾ÞîY¤gÃ/î¹÷÷ÒÇ£µVǯ“Ͱ¿¤ç¶Êôºïô¼3ûgëÏîáõi¢çrûÛýþ‹uþ8õkÔŸ¤~m%ßi½ÍÙ4ýiz}Éc+Ë—Üo準õ{I¾¨ßÓÚ“«ä«ªŸi-wdù§u< æTŠÕ:§uyÍ«ú–µ¯ö¬cg]»5ó’ü¯õc¯ûT%VwôgݧäØÊü…¾†MþÁyûEÝ»µz¯ •ø¡Z/óßKV+Mÿý^âW÷_°~‰Ž “›î+&ó̓t^u°_y cúÚ¼­ZŸ3¿¼‡÷—U©&ZÞtõÞ§íë;|½¾&ÜÕ¿Øk¼4Ïë~<Ç¥UÎíýË’­üY÷znzïú{’rœtóº…Ûÿ/tïctïQC=vý¼µØõçè¾F¥ºv•ß ÀS·¯wëo¯Åy%¬ÖÛ%u`-ÚzÎë þ¶Ùk4ŸVÝ러]¥šy¥ëgãŸíú˜Ž “›æ‡›îçóÞÙeÂW>îª#~óʱé¯;yþŸ¾þºòmÐd½cŸ–NŸ<6ªëC:îühtßÕïÿÑúoðu~’ÎÚ »êȱÉzn4/`¬­óCé¶ÑõßæéùöUéúýþúŒÏ|á­Ï›œnßwl”^÷ Ó§ÇŒU:ž&:¦éù©·¦?Æ&Ûÿ쯱©¢£|C¤ù§ûé#î›ôu'ücÓC?YãÓóM ãOðoŠø;ìT¯¾bl(‘ÎXKãq~Jºíñ囤®3ÿÏcCO ‰®–ÖÓÑñtž×}ôÓ5뙬ûï$ø¯ómÑ_þ-BbÓ‹Æ&k=|k3U×§rÌ}âÇ$ø§õL‚ÿâÇ$ñyD×'ëù¢§©ùÚ’SGßR´õ|Sç[zž~ªø§#'ôYóLï:?¬ñÐKäÂ73SDWò,òŸYß=¿§é~ìqŠiþ=Ç7Ç-=‡}@ù7Ššw†ôœ›…½NÁu?tñ\‘¾IÁ¾ÐŸ²÷‡èCnÃâ3ü™Šþ¡gè—øÈ·U#:Æî;ôzþ7+üGO¥MìBüi…oÆv7UÏméµÎÉš™$þŒÀÇ<ÿ~÷k^þmè4ÝÇ·~ða²ì¨Ø…úÍ;IóÂ÷Q‹¢OݰNìR|Úï!]e|É{XrG§ê˜o¨é‘ý¨øÎO ób¿5=÷3þ Éq Í?¼Ò8ÓćiàžÆ…oÅßè:x?=ÆŽ;Ïë¹²' xÄ:YrzÊùHÇúøƒžë‡ñâr¦‡¾Ò³¬?,}(xË8q}z?½é:~ƒõ‹SÐoÆ?ñ›ØÉ4=Ïü#]é{ÑSßn€¯-Ù18ÓÑýð•q°Cô¿->àâùåßz³nä'}Â. _rB¯Ñ+ô¨Ø ò/5¸PäÉ<è©èÏ;àžžÿFÅžGŸˆ«À¥é—¦ƒCÐ¥~”õè9p?Ÿ‡‰·$_pþ7Ñüƒž‡üz¾\ʼn^pq}A/ƒ|ð«Å¿ŠØ#ßÒC×®—8”8)ÓYô¤ÈSò› ž`×|SŠ¿ÑºÀiäÙE_ˆgÀø ]Ä-ÄsÁ#ßâ/Cœ€=¡'å7eˆkÕOÇNÑâœà?§ã§d·ÇÄì ù`ØøVöÚww¤/èëšðzºøŠaïôà:ú_Î#Oè€?ÄGè1ø'úñàãhÀ'øÞÑ}Ý ü1qQ[ë!kd~”o®ñ%Ïá= ö=½C¿5_ñkè¯æ†þs¹i=eoäg·#wøXâ ô:u} è"®Ð¸%ÿ’žáG ^£çÐî“ë~Æ™ÞbGàó‚+Ò;æ‰yë/‘'þ»à)q¤®GüDÿñCmG\îÆ½{¡§ä«º|+yr ôN×üÓ‚ .1r@Îø/ì;Ôx%%Óyðü$Ÿ蜌ëÅo¼d>ü6þ Æ+à=t‘ÿÍýècÈKþ¾yƒïØöa'q½È‘ûgäñÖï9Œ¾#gÆçðCš‡úõ ü!õô }&~CŸ5±çáÀÅ0?öRâ$p\@'ñ)ñ,ñ)ë þÖ}Ô›è›Á~J¾òã_'á߉_‚Ç=ôás©·„úÇØtÿ ™|¸ŒŸ+çÑ'ø£ñˆÛà¼ø«¸~¿¿…|‡¸8„øjDëæßº< þªà%ù$øJ@÷7ˆ¯¨?R'·È7à/ñuE­¯b^êcÄY俇øyFô·è5v«q˜—ø<+~ ~S÷@Žâ÷£çmêOŸ¸œ(ù¸Cüð¼ä7áù²Çnèc\Î|¥.&¾Ä8 ¿DÜ^âÊPCï £ÄcàëDñÛÈÿÊúˆ3àkˆcñÓØíhðƒàwÉÈ'µNþ-e±#âü ùT¨£ƒŸÅ¯PO>p‹øœ‹ô²^ðšz+ó‰~ð³Ä¡/u‘à#>!—R—ÃÞ±KÙK ûÆß·àgÉ«°p\'îò*u<êä×øSð“uhœiOKÞÂüäGä­â/úH?P/Oøì.øóâ¡Üàyâpú3¸Aê_èë§®Îá[à~ ?…ÞB?ügò‡ÜÑäBœÔ…¨S7Ägüê-û=Ï—=¦„8½(uœ GØ=vãúXç*õ'âô–<»!‘œŠŸGžøeÞC§âð»á½ç‰ßYoùÍêåàv¨Ëb¥NJ~D\(~à‡Á/ì]|FÿÀ—X¿Ö8Ø3ù}©»„ºpÁYÞwé>ê¢mè#ß$ÿ¤þÆ8á=ú€‰yg¬«u^L~¤à ñBŒu½ð‡ø“÷Ô%ÄOâzü_© R@¯Ñ—àï°ó’O£·Øµô´Ô]¡G×±¯¨ŸÌOž a_ä»àç'},;…<€ûο%/y5ë’û+ñ×Ã{×ò}ö$:ˆÓJŒü?€Þ£·Äw!~+ß[…x{ÄŸ”÷8z{Ákô*äÃå; ê&ÄíÄGÄ'Ä‘:_Þ—kÜòý‹ø@—ú8¿«$®Šß‡ÀøJÞïɉzo‹8ÿ†½á?±/üõòQÞgïQ÷ îòÀRç x_¾ƒâ»0õ%Þ8Pò+âsÆßC]1Ö×‹ß þ²Ô¿;þŽx/ÔÂou¬¯+~•º8që_ {%Ð}å{[Þ{W„87úô àdxï^ðž<ûæ}|ß±ëPÿ(tÿÿ#?Ñ×7‰kèyŸN=ºyO¬ç›¼'çEöîO ü+ïAÐ˧Mý”u“§âu½|÷€|ˆëˆçćò^ ?¤õÐ3qõ«â÷Âûžò~¼ ú…ÿ+ö C¨Ôˆ‹É#CÜUê6zŽ|§ì勼°ü)þ•ºù!Ï!Gâp˜õÕÄ]àiÉ‹t?ù_ñ¡NSäJÜ ó¼çàûì‚ÃŒƒÝeþì÷0ùvÎûɧè#v„¾ƒ³!_,t‚Søiô<‚oèGˆCcùþ.Ô)Jþ@œ:"ó€ÇÈ ;'nçñ[Zƒº‰Æß‡Cù"~ »ÆOSwåýpy¯C_Ïó]-ôÓ%¯%Î õâò]©øEÜBüKÞŠÜá;q/8){á»rôÜ*õfèÅo1ü¸_êð‰<ÿô´ì Ÿ±øâOpœŸñÔ%Ë÷?á=lñã¬ý Ž†oòþ.à@yû¤nð¥Ä1Ä™Ì#þ`÷ÄŸøŽá3qBù>2ä3“‚¾ :°gð®ø5ðýÆÎ‰?ñ7!N‹ßíGþ•ïB©|÷ŸÁQÍןˆCx?V꟬‡÷âGù^óbñ»hâðò]$ß àONàG"ŽGÿOþQø ΡgÈ‹óäġثô¡üû+ÞÀwèB¿Á[É“ñË{êà[ø®ªÈ?H\n‹=¹?<ô±¼/¤ŽÅ|!~,q±Ž«€'%~ Î N"þÅ.°â êè |C¡û ïQc\ ü,ñ0uqÍ•÷+áýVù4øžÇßï`Ôí nƒ!^*ßé>âÔXÇ-8ƒÝwb7:¦¾…¿*ü žê³å}¸ƒ}%î/q£Æá;Ž¢g:n¨NúÖò6ð ¼Ocˆã°‡‚ÏÒ#ì­Ô#ÐcpŠx'Ĺ¥þÊsw £ï!.*ù5zÆû —ï•cÞê°å;vî#ÏõŽòïáˆW©à?ÉKC¿Ô›³üª²wv÷sÖ«ªÇü¢ª†÷¶¿?QUcªª}lUµn³f÷¶&UUó_õï¦}¯#ß;ë«UµÕvßÛ¬½ÒÚ;¬í`Íï=ÛÚ{¬}Jû3,´æû]hcù¿§~®µmíïiÖëßy·öÊû8´v²þ'Ö>kí©y€-|Ü™¾Ï¡Ñ¶½õFc÷UF¿ÕþŒÑmÏMÙ§ªF®«ª-w¬ªi»Úñ£í¼Óz«Ýóë_aý“Gë!›k몚ü.“½3ÉÖ9ò>£J¦gê9FÏçm.[sû×F˵֛®´÷Ìs¶Ïzáã·çn™î=æLdž29͵g»GX3:Ú´v£µ/Û3·çÖ:Ãú«µ·×/ò³ÃƇ½ÁÚ Ä“gäÖ1ë­í»­5­Ù[—XoHÜZb÷˜= Ͱ{L÷‡í¾Ñ'رédÇ×r½ôÑìpÈt½s…=ç×l­í+}B뇬›õ=¡¬™n·)›4Þíb4޾×xf<u{ÛÙ6½I{¹ùÞ¦KÛø>æF_ÓxÜô}2L/šTÕ£·4ú®Íz3d²sKvaôt޶ó7Xû¶ËWºo¼éšu s:¦'íÑÜwOÍï`v9jºœí´åûø®O7¬ií#ìð}—lÌÎÌŒßiÌ~cKÓ©IËL¿ {§˜½ŒØ8#ÆãŽéaçqYÇ·±gº‡ä¹acwMƽ]ã}ÛNÏ3ÜèÞd÷›v S;wæ{÷œ[U[Û³[Fmm2Ÿ¡ý£vÜKr6žh÷M:gãv|L;n6ß»U3ÓÚþ¾]– ™<ÛŽ¯³k†K-³ÓÖËlßÏúŸö#ú¡5ÓñÇ~+Ûርq’áÏÉbH{¦}yÍvZ³|/Åìך&Ó¦ÍÓôýÅ cšÏ³ößÖ^,ûZå=+®Ö·2Vuµß̰ÝÛq<›cc»1> ÛsCfCGeœh›íu„!ÝåvlqEëíY_›¾þc¤†KÍY;#ÿb–Ù×ÜžÚî®Éú?d6ÚþCn-Ó…Ö÷}|Íák5,k»Žú¾°¿ÉúÚ8Cûùø>lF_ÛcÇÚ5£)ýþÅ¶Ùø^fio ßsÓäÚð}“nÍ1WÒÓÙŽcètkˆC†«SM¿¦x çvo:büòíl½y¬ï4w\ÏÍv›Ží/ÊqƒË¨y§â0íØõ|Áì¡kzÓ1^·=ž·¸ªe4´ž$Ì1^5mÍ7Vy¿(û;ígíûú…¦'¾Ç`ÚëÊ÷-ó½ö~™çk[l1lúÝUë<ÙèõøøÙO¶ ‹ºæëºïé7CFn“Ü,¦2þ™¯1»éÞ#2y¶ìZÓc5ýŽ‹cƒó¦µ4ó9ù­Ù9¾uLõ½¯ZnC¦Ãmçã‰â¥ñ*í¹º¯ö®2ÚZÛk¿.ßkÍÇåÄo7Ì6/ʲh~1ëJk…õ‡äøÝõ´µ<ÇìîCF ûF=§1{2}ìZ|Ýy«èõ}Ì §›[gY:ö6}oQßÍ÷Á÷ßµ¸J¾Ç0§eç[Æ“¶~g ==ç m“CÊöκ’äí±«ÇŒ#’ñ®ù~÷Am÷í®Ë“ï›L[Èt·,_jš}w-&i]ã÷äç~™±¸iëKùâÃÖ;.ÖM2Ýìö­ßLj™ÞvŒm‹›Ú†Q-Ç÷{rã¹b×±ýoÙ·¸îuŽÌñ_ò'§k}ÇgZ ž±0ÑøP–²¯›»µ5·_-ÿð•Ì›„̱kû†;M®Ÿou_“‡Ì.;Gã¹aãóËÖ|ëÐÙÿ¦ØÃ÷¸4þµÝ×_R®m:Òzš5ŸßmjØüÑd“ËdgªÉfä'Ò‹96iyñYÇ:Ë9²ûÏ”ó½M¾àe¹µœ7O”µ±›ÚS­s°r ‹A»Æ®á\ÛsÌ»òz;¯&ïFgf\HûåÛ:Ó~¾‡­ë½ÅÚ-Ã…¦û‹aºîÇOκãkmýDqÏqЇlž)F爯ÓñÕô´ë¼6n9.¹oøYŽÅš÷åxlÄômÄh1üöo²ö~K´^žu©õ«g Y,0l¹GÇböŽÇ±®ÇJG–c¹„?WfºZ;flnÞ¤ÜÖó³õ¡ÙÙoz  k˜Ðq¼Òomu,nšjø2lX2Íâ«ßÍ0üØÚcgÇ·µ—Èçš^7?—cäÔ/Ó\.kÓÿT'Y'iyöT£wŠorâò7>N1Û÷-;‡µo§ç_Ž][w÷«ÊÅ,?è|Nvô›¼¶¶×/Üoyüíñ¢ß=!ÓÒºIúwa®§ ±·¡ïoë¾Ãƪl}þûéo³ëêò¼G^ÃdÓ6ÿÜ5^v ó;&—³«á§å|Îç²u /Î5™aÓí®­±ÓÞLÍ4{=§cúí?¡6j6?dqø–1ùZ<:Åx7b<Ý+ëiÓÆhÚµæâ̧äK¼îâ¿å¿]áûX7÷¾÷»Ç #†½#fw#·d›5ìëX¼9Åæšb>m’áǰóËäÝôXN{¨vL‡FŒ“Œ¶Q³á®bÜ!Ï!Wþc1QëË;†·ó†ù­û„iv­í¾qažûßßÒhhzìe¶Òv?dëi}(ëEÓòóÎ[ÜW¤\ßó˜_f›bÌ#ìo‹['U©îæ5¥–ùÖöÂ{ÿDz¾»Ž¥šÚG³ïè¼(cF×øÓzIŽ<ÿë\”ñÉ}_Ó±ã{Ùf“oŸ¥ÜÈ}çUŠvÌ÷4?£Xæ"Å€æƒRîãûEþ6û¨®ñ¹¹CŽç:–[y,z\Žš†çÍ×eŸÑôøÃÁd3lö1ls-RøÝ\ðØ8å†GmóϱZÛýÀBŨ^ÿ80Çâî£=Îì¾(Ç ][c×óc›{Ô®M»(ÇSÓmþQÿŽl»íXþÖ\»éLÎøä>ÐkVŽ­[û¤=3MÛ„ËŽ‰¾ÿªÑ”öÊö_âôØåf¾kÈãUÇCÃ¥–É£õ‰ì'+»FsÛä?õ]Š£Lšî[]gÈzïþÖk—ž·~güyUŽI‡æg¾´¾Ÿq²å²º;Ç#–‡¥ÞUÒ¿ƒ³z®×Ô>¢m¯Õ®Ë˜×²ø¬õ#;·[μÞäùP×èòZF×b³ŽëÍ[².uî“þœ—s%•“ÿ»"×ÁÚfŸmËÛ#9çðØ·ù'ŸW汽ד ïÚ¿È1Á­môòŒ·Í÷‹ï3sÌæ~#a÷{W¾Ty˜ïa}vÎż6’tÔ}¸­«ý õM£Gûu{œßÞ&Ç»»¬gšÌ|ïÐæ¨ÆÙ1ã[ú«ù™—iÿRüFá ïÕêyƒÑÚÑ>«mùþ;Ç{©víq¸b'ÏéWj£qKÆÂTGtLxD–yúFß›ÖÖëy^ã§Âêg)®4¶Ï’þ¢ØÃeí±›­cò§ ×´wjªQyÌiºÚ¶\µ3Ãù`ü¸>Û¥çÎî;¼>›tçTÙâvŠáì~ßS¾i|Nñ²Ù\ª í ÜÚ%ãPÓ°xÔtdÄâüaãmW{aO¶ã©¦£f£Çæ<`Øó_³Ÿa³£[ï°éæÐ§„­ r\æyoªù™l†\ïŒf¯ÿ·ódŸß²ØËó–Ö-òc^ÿø–ã­µ5™?­ä/lþ»—Iÿý÷=O Úþ[ú-ÀRÛðöqÅ é‚c‰×^‘ñÅ}ÙˆaרÙÁ$Ï»~mÍß7˜½u ×G½†ã5H“ù³ëG™þ·ÜvmM­'d™´²f>Ëëwÿwø]DóáÍC³>4;'gzRÍaµtöMUª·¯Íu‹Ë›<×õ:ñ-™§#ŸO7þó}^¿öš¦Ç5ÃS¼3¬øá‹9Çndº½f6Õ|ì´ϱ֨Çß´-‡i?Z±§cË{Ÿ’óqïÒ2žŒÚõa£±éõÏI͹cÂǵ…C.ß§ÙëbÛçxØkämeÞ”íÚýLú½*~ÇÇqÊòNçÑt]°x®ó•ù»¦$¯§äø'Å<çz8Åô—fÿœÞu™~uíZÇë‡^ÏX”u¦}zÎ_’Ž›n7-mjýÌèï¸R<åy±ïü‚œ™ßštu~6åÙ§çøÔýcÂrß#úÛª¥ø¾ð^g\•u£y¡âŽéYfC;YÛ'ó$åGîë¶Îât¥:épz§ñâ*åN-¯…í“}aÓ1Í÷òö½Ãß•u«{Y–•·¡ï˜œÍŽš-ÌXbò~wޝ§šþN1_5ÉhM{ƒ_“ýÔë•ÉrôµÙŸ›=vnÊ>xØøØ}µüÏG2]îCž;Z~êû¹ßG^ßÙ¼=×m_Ÿglóü£óŒ÷´-þèxþoôzÐ}ZÂlÓïögó}΋T£÷˜Þß šޏ_¶<`øûY¹&à5ÖQµ-ÞéXÙ:F6}æUwmŽƒÛ.ïy9^šb¶1Ýhža¸7ÝðuÚÓs<—j —ûÌ.'¾O1žwMvSÌ&†,Ñï:¶,ÎéXŒÞ>"ëU˰¹ýüwäz™Ç“3w=_p¼ry».zÂb檜ãz¬åüý˜tÐcOKo KgZÌ7ÙâŽÉ¦ÿ[˜-Mò|Öcæ;³Ÿé•õÇó™–ã¹ÇŸþ.fr^GÛyîï ý=ÛY—Ü~ºŸ²füšžãvÙ=Ît{1›ÙBôÛ³Ý;r]Æõ_{g/KœÛZïºO^”¯Îý’ü§J¿èÂü›~Éžùüs׿û­r¼î?â¼üü°ÏäþØmòs‡–ç[>+ß¿ìËùxÙ]¹_ù¦|ÿ²úç;YÏð@>~®Æ_þrõZv¨;NÇÏyz¦ç÷œåq½¥ç~­çY÷_¶ŽŸ£ãCÕï£olž{Sþ-¾AtåÆVîªõjÏ{²ø(~þZñåö|¼ZtŸ½4³æK¹?ãs’ËÊüüQßÔüâÇ1â÷¢Gæó'k½‹®u|¼ä~¼Ö{"½Öù<ñãhÑÅsÈñÍ‹\Ð'äƒüÐùãªþ¾ž ëGêøÑq”Ž…åúþá>úSÄWÖÏú þлNÏiGÏÕz4ý‘’Ó‘ú†~­ùN”^¡o̧_‘;Lô):OÖñ‰ºÿy:†®•:¿øòÜŸsiÿøÏPÿ,ÝÿL­ýeý‡JÝúE‡±gj¾ÃXü}QÿŸ':–Ê.c}!/ŽÑ ä°Dt,Ñ}KÆúùpÂÚ@üÿŸ»ežqV-Ê×±§3„#§–ÊîÎÆNd§àÌi¢oeÞ´dì´YùþÕznµèZ%>`'k$—ƒ¾E9sT¾~”è}:|‡_¬[=xxÚ»rºô úÎ}ô'h§¾_¸)ºÀÏ#5/vÉu켆/']’‘Ûs£<´ž•âç¹ßѱî?Kò:ã¨~þž5’ÏŸþ{ñSü¢?Kë=Sóœ&~Ÿ.úþ èܲŸîcµ~ä‚]û×uέûŽü§—ôëcúeò²žStù Îâׄ÷§¬è?ΣžûN¿ÐãÅ’#vÝ#p±àŽèZ„œÔü&Ë¡ûéý~ãù«§‰îý¥ޛσ#ø})Pô¾¢^˱²3øý8ñŸèá)’ËQêñ'ô¬ë8é|‡ÐæY!?¶zç|ýÄYâ«îƒÿøeìyAz€þ‚×úõ©‚kG!oÝ_ðï¡Lô|S\uØŠ~~`ŸÌ‹ÿ‡®âß9’#牳ÎÔs§Š¾3éÅ×Õ:>C÷?Gr>I~½ÁïG1yÃý³ñKAOð×ëW~Š>±Nô==ãö~{*8°Jz}qÿzW^ÜߣßП°‡ðù›ýë+ç%?ýZèØbñyõÌ|ß©?”ž·±'ä»â༞Պ¿NÕuâVì~©ðø}'>;U¸½â'šW㬾±nΣ'Ä¡Kˆ÷Võ÷EOÔƒ£ø-ì~EÞ„¾Áwè=Ir?Nýñº¾dVîÉïÎZ”ûSuLÜ œùÁ<ÏréÁ¡Ò7ò9ø }Äa\Ç_R—xVè‰×¯íïÑ×· /W¿A8):Ñòý¢çØ¢uº®ñ–è˜øôÂ[ò:È[/xgÍ9ù?Î:Ö<ð½+ù­øÀýÄ¡úÕ»xóé:¦>sP8FñóøGzô¹¬_8Lþ¿–J¾ÑÒR¿¡ç:y)ô€ç¥#ùáÿ‰óõ§Äõè!ñþŸzñz„>ÓS×a~ú¥ßÈãp8ž8{§ÐÓ¥ÏÌ‹]q~/Í¿z쑼|$Ù_ãï­qÆËÇúåȱ1£¯7˸û‡óàõüi¬ó ðƒøÜ.y³ø‰Áÿc'è7y!ò×ÉOÖ}àõ:âò—ÃÂ}ÔsÐê«ôÈüÄñçu¥7WöëñõÃXxÎXÿùg©GŸ(9¬Þ' ôâq õ7ü-8s´Ž‰ë+ñ5ö@_üÅ•ýþ£Ô×õˇ88‰u'1?׉—°üþÿ1¦ûöUÿTõOSÿ Ý÷4õà|¡¼è‰‹‰«ˆÉ[ÀíÓÄÏs.Ï=öIÿ'<穲Káu…sE×9²Ó3tL^ëÒGᓎ‰›‘S‰»ý¸ƒÿ!Ï!ž9umÿúðÇø9ÆÇo#äž·€gäãÜ{îÇN=ïcÐè@ÿ˜·¼÷]¢gäáà |Ï몋GJžô%¾TütAvlëã…8®ø¸\ü^¢qN_ÛߣÈyqž>Æ÷%Ž@žŠwˆ ‰xÿ°$ôè5~cÞ—àÿÉSx˜8ù¤pL¼ó§RWzz¿^}ÒsÄ{Ø/çÑwüvEÞN~DßÇ¿RW*vªñ©g,½èq)yOy¿!» ç‰ –KÞôÄ;¬»¡ÇΈŸˆ¿¨£a‡„ó¬ºx ]œ'.Xªu‚ëøí3EgÌŸˆ‡x®¼—ÑüôÄ ÜOœ_8A|Y<+ߊøq²ž§Þý<Í{²ŽÏÕ¸çeG=vøC]ˆ|ƒuóþ¹þêÿ‘GW¨ÿ¿YzwºŽyßëAÔ»ð£ä¯¥®Œ}ÝÔëØ?øv¬Ö ½ÕºˆKÊû'ñóÔu¹§NMϼ‘òôžúä1?Nž‚½c_ËtŒŸ×á?ñü“dð¿Ì1q;þ›8&Ö©˜\"^Šy2ñÝÒEýë…_àãóÞ¨Ô­·êD=Œºñ8uÊÕšœ¢'^×KIãAx¾³ì »#î…>âYÖ úÔç!Wäý%ß uèézyŸ¡qx?²tVÏ8gK¿Ñ?Þ¢×øßóUO(y¦zòDìÿ b—ØúÉw#+D'x~.è>zè[­x|`™îÇ.Ÿ犟PÏúÁbÇŸ¼†¸ë$è^×O?zK<Ã:Ë÷3ò§ñýzÃw<à.ñQ©€êÁKòpì;EŸž¢žóÄÙä/hâê"Ða7à%zÎÂüþ?Fû\õ<ÏûX¾÷ßÐ[æ¯Áû³u:ãâÛioåå:Öyâp¥ÄïZ7öë%Ä?Ô+àãÞêñCð¯¼Wå;£Yâ«è–q^t>O=8‰¼‰Ç°7æç}*ñ6ïC–Jã·/zs¾Ÿ÷1ÏeGàâÙšÿœu¹?St'–÷’²ä{üöL€¿(yµî'‰£È¯éÁ×R7ïj½¢…èÇoààõ*p„x‚}ŒyõNòâ'ðS¢—úutÖSêï:NÇïÞˆ¯ñøùR÷>ÀOìŽùxOJ¼FœBþ~¢î#¿'oÃÞñ3å=¤ž/y?té¹àçñ|ŸRâsùa¾¯"!?DN±ÞŒ}–zu.È•ü?…>À¯ò}‹ø^ž :~càuð? .Óƒ_Ôo÷øð¯Ô;ôx‰ýS_-ß©AŸÖ~•ï2tq~½'N§™Ÿ¸ª¼‡Ñ1ïÑSêÄ àè+^œÇç»â/üLÉË%7ü8ß…–÷—’c©S®í_oyÏvqè%GÖ‡Üxž÷œØ/õBê$cŒG>ƒ‰ïùð»Ô!OU¿@ý”¸ })ßµŠžò^N=8ļø»Øcø{울~…îî®úùÂ÷WEß„CÑï”÷[z>æ™è úÍ}±îŒýSw¡‡âzÞÏ–ï­DÏR=G\ŽüÔ:¨O ÇÄmØ!=vG<Êó¼¿Ó}ä×û†çÁƒR'S^"Ÿ’‹à[±ó v_ñså{]Ç®ÐKüï©Ñ¾Ûäyü ú_òFõÔ­Ê÷TAOÀKò7ò?êùç¿,¯ó¼?f>¬Ô¸à}Á)=ßøÕR–½»¬¿|/¯c¾ÈÇ‘{éuI‚¡ÞJ]„ïùÞ“x”8±à2ô]ÜßãwÊw{zžïÂòˆùtùŽ˜ïëužý!Î[¦qÏÊ=ß«®}«4ëZ¬câBÆa=äeå»Ý_ä¥cò´Çiþø¾ü#nDÞÔ-ˆçJPãœ{qO¡¼Sß“o@7þ‡ø{ÄnÐgp©¼‡Ôóà:t—¼ìòÜSß)ßwê>òÃ5ªk> /x¿’/œ"úˆËsêËû&Ýn¡çè?øx`8.ù´è]¡óçÈþˆ·ˆ“©«/_ÔOõiÎÓS—@ÿÎî.ž¥ñÖöß_B¾Ä‹ËÄgìˆ÷?ÄñÄGè1t'€[àïOc~EÜDÞQþ½zôëµzOŒþó½Ü"­;&-8½¶ŸNÞw`'Љ<À â ô ûâ}z‚>âO°còâÑ1£/å»3è σÔOáù?þy–úÆÚþõáÁIø„]üYÛ¿žêù÷TGÇúÇÁ¯àWÁUâÞ¢êÑ“¢O:'ùw§éyìžï'ð àòÄ^À¾» _D‹4>öHÜõD]§î\ø¥ž|ïR©Kœ®ñÊ¿ËÐ:¨¥}G¯ÁQòP¾£âû*âÑë˜: ù[ÉuŒÝâ—Ñä–…~‘®¯ÐøÄÍð•¸yïcà'rÁÎc] Ô'Ô£§§2žî§nÊøü»œø'v|JÝ@Çà zÿýùó=Òóçk^ä½Äçøô9aè;ëEË¿»»8Ð¥óøwÖ‰<Ð+ôžüe‰è'Îà½ÙR‡å»±*ôœ×}àt`ŸÐ‹¼©s‚èï2ÝOž„ÿg}‹uÿ²Eý=ræßñý$óá/±+zÆ/õLõE^ºNž¡—å}øÅýÏ—øOë/y >€‹ø±už8‰ú"òdýø™òMÇÔ›øþ>Ÿ¡óÔ•ù÷ ‡†çÈC±+Öµr‘ÖÝkû×½¢ßØ|‚Ѿ‰ëˆ+þ—ï¦ãò^Wã•8c]îÑãò]‘®/¿\ë¨øO{gW_±öyk?­ò¾HïµöBkÿlͽ¨kÒ9úûmUÚ/©ò¯iÎÔs0þÅÚO´gÑ5Öü_í¼ÊšKÁ3dG¿¶<¯®± ïC’šö±¬®«òMÞ_Qå½g.ÌcøÞÉi”s´g…G¿¬Ò~uÕoóü [OõµíoÒ¿‘×ÒØ¹Êû;ú¾v¾ Á­}ÉÚ;µFø\îMlüÊÑo‰îýS•÷¾©Êû û¾¶¾ŸÔ-UÚC¼úƒµï[ó@ë]Ö.µö"k‹¬½²gÍóª´×Xcræµï£“xÿYkWYóc®è´ú¦]ߺJ{²¤=w¶µæÿ[»^kwÚîµvs•öD¬ÖYûqæïe™îýQ•÷úªhý¬îû–µTy¬OkMŸÏÞ§1|´E»Yaã™âõÿä½e’xó[\G^`MÖ‘dw©Æÿ¡äòt—u•÷5/Û8&¯)­íÉÒ…'Ù}÷Hì™Æ’Û°ÖéúùV­Í?¸üˆµ³$+ßcæÚGËe~šóXô8м^2÷=ŸyœhÚNÏL—ŒýÞ÷[»HçÖˆ7Wç=sûh?Ge]vY¤½B}½/³vA^gõ3éÇ­UÒËÄ¿Oª­ïýš¼GköþƒUÖAßwéoUÒµFK÷º½ÅÚ§tŸï)2Cºàò½&ÏÙØ*ëPû{ú{è0ùúÙ ß·ã3Ö|Oì›2ï“Î9wÚß|_õkóÅG·k³óÊ«»çê¾ÝµÇß“c­ÅuîÊü\Ó÷Õ»9ïÔô½õœÆ;µWÊŽy?´žsóšªWƒÜ~wgž&ù}´Ê¸ã¶æ6øssù~œÇkÝìØ÷–\žçö>aÕõ¼ï¡³“õw[ï›» ½ÝÚw$û7K¿LWª7XûP•uÜt¢ru\;Jºå¶áã·²î4ükŽ/Iž/µv™µ—ëíåÚp;Ú;óª±Èú»¤/?Ë48ÿ}ÿ¤4—c©óõh]÷hØ÷v[&ÛÙCÏ»M|E²øaÞ7¦±P{Èt²Žû~¾iï°ökÉõê½ó ½‡gLþÚi÷¡K³Ü’?p ºKº ýÐɶÖäøæq‹Ç n·þA­aCåúîò‡6nŠ…|­/Ët¥øæ âók¤Ž}ïÖŽéž5ýFüuÿç_…»šî&›s¾z%ãé‚ãÔ¡¢ÇõcLÇ~~±úc²ÜÒß®WhÝ·fÿšâÿ³$¬|»äéÕŠsósþ.•Ç“_”/ó8àmÒ%çÝ+µ£/ùy)<ö˜/ì|Š|ÃMÒÿ§e:¿“^l¡æ÷Y\Õ8\´8ö9fÝ":<¸7Óš°ÿËâËåvÍÇ:RòwÿT gJ7 Õ83Ô»qŸæë×…ŸN“û÷wÇK‡Ü˜oh(LϹoþ7éÇ+ô·ë€cè¥Z“ËøMzÆã7߇ÍãVÿ÷‘ŽóSeËî<ýS–wŠ·Wo–l}ŽnÓªã=$^âïTógã.-¯žû—P÷ø^ú¾Çþ?K/œmÕ㚆p9a×ûôÌ«õ˜vQîŽ|Hkó_{:Mc¿Cç&K¦n+—ˆ°Ô›=¡ AÂj‹‹“„®+ß®2>8ï?/ù~)é¿eã¿§”|üs²¤ç·/wËúäìºä˜î6ó]­ÍeôañåÅâó¹™>-“îŸm2¹Jzèr¿Mcü ÛTÑ;Ï!\ÝG¸ÿÿOÙÁçÅ÷‹ŸÓ=./· Ç}3“®´ö3íxè1˜ãÔYâ‘·cÖ[÷õ kܶ¾’u"Ù‘Ç鎱/ߣ]—…G wý7'ž°Þž:†Ü¯u¸Ow|p,¼®*~´ñáø¸Jzá:ï8ê9ãDׇ¤ŽƒŸ’ÜÝ_–ù‘|Õwó8Iw}¿fêI»ŠÏ^§:^úêºqŒôswñhšìÝíÐs`ãgÃ0$å ^“ðµÌ’=šßL¾ÈsÏ)<~öºÍÁj{‰îkEÏ—Å×k%?ã[Ê;Ý~_,[qÝ}f•ÅãÀw‹?w¨dxX¦/ᆯßc—Ÿd9¦¸Ãý*ÞÉ_9Ÿÿ%ó#徯—Þ¹s|ôߊhêþVësÇ}r÷ÝÈϧxÁmË}¿ÙfòCžoë7õR¬áõ%¯g{Làyžç®ÃŽqŽ-ž‹z¾~W~>Éý}™¿äm©ê¾ÁcB½QHrÛE|˜“iHvã8æñ¶ã¨Ç†_«Ö×¾ÜW{þæøéù‘ûgyÔýÿ©õºZ[åϲ<‡Ó—lÀíÖíÍíÃc‘›DãÅÇzϙݧ»o{ƒhÜ]4º¯ñ½™?¥±½6âùõç…/nƒŽËÎéÒw——ãòó%K?öøZoZ’=ß'}p\D¯®©Ê{„TǺJô^Ÿy˜â!¯]ÿîÖØ^»kg>6fU9÷÷xÂcÏ©n”>¸¾C|ñqÌòK±Åtñßó>5ÝÏ>Zs»þ8Vz|àXå6¼¨Ê~àÉÅc(Ï Ü‡ºþ¹?>_´ø<ïIÿȶš0úñ²eãåþë2Ù³ÛŽÇÛ.{Õ‚>=Q8|0ÁñÉ#bIÆ‹d§ÞnÈ8šjÑÝlŸ)÷r_O â>Êsš©Yw’>9zŒîù–Ç_–<Þ›ùœðmìÝÇôüÏëzž'ï#½q›?SkuŸåöòNÝ»‹ðÇ}ã¾Â9¯—zÀñá9âç)â§óeŸLwò¡_mn[WK6{º?tût{º3ó:ù÷Éž_yìÿ+éô4éÞúw¡Æ¼'ë]е šÏÈ¿»à~ ùÝäµ&=ëf^¥uÝ!Z~‘ïK1äc2MiOÍz•êEN¿×Üý.ëKªÏß(¾ù;Ïqv”>:ïÜOzåm’cÐ-ÒëK²ž•£Ç…ÛVëã´yy]Iç?‘å—0Õùç¶³šcη«õu†5²ÃÐô{$·J÷Þ,»–dôOºÏ}‘ûý•’ç$·¹3sŸÖæµ—óX·M¦Íù•|Â/uÞc·o÷óO•îVâáº*ÕvR¬áuÉË$“­Eûãuïýëçó÷<•Þý%~Ýe°òOÿô;D ïÿÜÞ*¼pòøØãóë)Þž)Ý}a¶™TsüH–gòa‹ªõyõîZëÙäÏ¥#^‹}{–]’õÝÒa×mÇáyz~jUr¤73ÅÃûEŸÛÓ3¿RmÞ1öÙÞÒX¿ÈºÓЛòû9Æûó³dËÛ ƒ¦Kǽ¾õé5¤¯mÙ–ç;Ž]¶¯±%y!@'ã2z´}xŽq9Fà™_×çÓ×èóÀ:užùgëüìp‘[àá'ëÔýðanàö[ÖÁº‚•ñÿÐwô |@Ïv€.Æ ÏÓóÍ#ë-|G.Ág#'øøˆ~D{-t‹z¶@ývŒä0;Ðqûçºæzá_Áך>âzáK°ãÙ5ô²Þ¨‡ðùE¾ÌøùSì4èË‚@'v2o¢u„ã2´÷ˆ¬›õb‡Aß ^a_ÌíO|‰tÌ ô| üˆ÷E9¾Ö¥çY?z¼ à×üp¼CXϼpŒýG{Œúù97Ðϸ1~°·Ð} ú¼}ìëü=üeþ>Äñ#¿ Ÿƒ>•øGÇ»nðû ö†Ç;ÆîtùE})| ü„Ž…n”õFûúYæ~(êyÀÙ?qgÔ—HöäÀú‡¸q Ÿ@ú‹^@_ïСû÷ôǸ…uD{Û!:'Ð]W*zƘ/¬s >ø2°î~=ˆï›âÆàgˆ£K~Áú‘+üe¼h^ü7òŽu»˜ÔÕgãºËøèY7Æ1îðw~¯N?‘/ëŠ8ßÎt ¬~{/ñ0öH¯ëO¼]zõ ýC¯Vôß7;Ì_ôzÂqmÆaÏXè#Þå}uÑ/è‚_oñýâ@/Ü?'Œ;‘c}µ~<èõ@ü­câp.êçž@GÔûh?A~Ñolôxà}[ 3Æ‹1^ˆïsòð¨Á>Þó`A?JÝ$êOÀ!ôfÇ`WE¿ƒÄ:÷@¾ü^Œ¯â ô)àð@ÜϸÈ]ÇÐë„gê A_"Ÿc~_êЂ<çz¢¾;|‹zùñù .‡õPO‹ïùäÆÖÇ+¸QƒÅ^Á%äøõ*Öûâ{†8^çc\É{ÿ¹á¾¢aÞØ|‡×úX·ˆó‚ß)õú 1މß#ÁÏœá~øAÏG;ä8Ìñ¾ïp«äq¬+ò3è;úU÷ž~a7ò<Ù9èO´ÏÔ½‹yT]}mäWÖô+Æ[ñ}ãüð|ÄÁo×Áýq½ȵøøÇsA~1ß*ò€Ï:Þ.ð3Æ×uß Ç‚§Wÿâwï/jôi€øè)þD×ç‡ñô)G|g=ägßWÿ‚Ü:£¿¨û^¨ä š7â0yc¬ àz”Kà9®‘K‰okøí½®û~¥ÔÃàgèâŒ|Bà÷Ü@o|ï6ðYàÛܰþøÞ¿àr w®¹>PÏ þ¤®Þë%×yô™z]©ÛÕØaÌ?bœF_Ëwšgç`?ØgÝ÷Hñ»Ö¹Ðõ-úÉÀ¯è×â÷U8¯Gýã|ÆzüÂpßÀw›oXgÌû ¿¿ë‹qÜ@>Ç|Á ŽÄñ88ýwüþ:Þ7¯G~G=ˆüÁ£\jôv@.‘oÐÖóšíã}ÁÞYÇn¬/öçbþÉ}ñ;ËR7«Ñóˆ3u8V÷~$ú©ºïÑ¿ºïžËwœA©;ļ¦Ô/Ñ‹(Ç@oŒËbü¿ˆÇ‘kÀ—ºzz¿_‹uœ¨G±zë¡<ëõ¢0_”[ùþ«Æ.p ~Ezu~~ä_è£}Æx7~÷×±ôÖÉ 9£Ïè7ã‰õça½1ßÞ•ãšuÄ8»/~0ȵî=ÑÀ÷¹a½Œ[÷=s´‡hGø]3×wŒtÕô‘¿Ñ Ô[£Ý„uÕéÏ@ÜÍ1ó×è}¬—ÐSƒ‡Å`7Aþ±Î鈸óÈ"ϰžø}Düž±¶ŽäQ÷~/Æ=EÑ×€³Ñîãû¾(‡ûŒ~ øÙøÞ}àß;ÐC_œúkì!~—9pÀ‹X'­ý”ýˆrVOÞ_òË€EþÐîßÍLdÿçÁOpNçùî4ÎO_ë·‚cþóµ¼q> Ø{Î{ŠòŒz^ô‹ù¸?ÒøX÷Ý[ÔLjû1ÿŠùlô?ëŽzäÌsȱVïYg?ÒSû½ü 8¿?«•3øô¦ÎoÔjäß› ÔEƒ¼èãûîX§Wk¿ªñåý¹úX'øw=a¼HÝw€ÑŸCßNqÖp:Öƒâ|ußS–øˆyâ}èü x=Pg8°Sã@ÞŸ ü‹8=÷¡Oÿñ½ú}ÐçÃþÑݯx§Z¿wv[­µ­½ ­îÎ7Õ6eþ:«wÜæ8÷ÿ½køß+ò¨¹‰ÏÇÏôº6¶õ®¥Z­½÷÷žëí7‡Ç¢%êãx:iÙ¹þ=ò›HGz×e=‘=mˆ×ÿW¶òÙ6ÄÏVÕ¯'ñ ê÷xóDý‹:6_7æZŽlÌ:ƳÉ::#þý¨[¾oʽ|ß½O6‡å׆ZÄ«‘á¦È4ò¤ŸÿÑó§§Ér<½ìõ›b÷aåDrÛÔ16•—:¿©zû÷èùæÐþ÷ÚÒæÒV‡yãamäk/ØÔõ§›*û éZ¯ªÃ…h1îÚ~Ot†ç6êü†p!ÞWómH>uq@]ìº1¾1òðæ6×¾ÆãÛ?ÒvÇ“ãæèÛ?s'²çMñ-›KëæbÅx÷Ö]ÛØØpC÷oŠ}7ÖÿO6GGÿÜß/‰6UG˦Æÿ´ˆg;öÆðk¼¸4âàxó·¶M¹ws1tcó–8ÞÆŒ¹©ºùÈÃ6UŽuò†‰äýÂÖ8ÿDþ`slx"û¨cÎmêú7W+ÿMÉ£6EŸÇ³ÍñhÜ]ØTÛ¯MDÆhç6E§7ÅŸn¬­×éâÆÈace1^ ÛO·6'>Ú OîÒÿñtzsðuc×Q§Ñ¦¢~m*îo,½utl¬MnÞý½ú°©v±¼÷ ucmêœÙ䯬mcíh<[ßX{ÞT9nªOÙXlÞ{ïùñÞaLDÏæèçÆÞ¿1ãwïÆÐ;ÆmŠ¿ÜÐù É­.vï9ß¶¾íûÖûžÜ[Zç¿/Ð\¹ÌþïuWœ¿få™Ë9:oÍâ5矧£Öšs×èÖâS¹§uÞò øsÙâ õçÐâó–®\³œG»§-_½xÍÙ\B…Gà²W´vö|ß°‡lm¢Vj,9ß™™3çwÎìx½ëÍâ̲9¹|ÒqœQg´4Ò}wͱÑîŸÇ©Ltõ£Ýz»ÞL·§ôi·Yî¾?é¾_œí쿎õXÿIçXµX»ç•»^8û§œ¡ß£ÿÒ»4À{ìïìÉQ®Ó uÁÚ侚N4ê»õF«kUÄ£×[n=~’níÓDòÀ¾Ø•þŸ§%ßj~>öËÑDz_§œ|äã}„|iã¸ã¸Å¢ËâWUí[*_%ic|£âaÿ,‰ÿyåWûn1Éó/Š^Sù¦2oCüïH¿'z¥“Ïñ×:y¿ÒÆ:^–ö‰»ò2S9ž<ÞO¥ÿZæï|àkj¼†ï7oØùÇz°âuØù‡ó_«ëÿ–׬/»>¹•5’…N¦hë~œÏgDy^.˜¿sÊ¿ˆ£ãê~í¯ã¢}]ùaqü‚y‹{ç‡qO¯¸Ð¥¾ò×õêõÔù÷[~w•ù!^‘Ÿ®óô¾ÑqP—^Í)ª~?§ü|Õ_´/u~˜þäàºôqG<}Ü¡¸ÎÑû§HuݸÎQýEë¥ùº>½.EûŸ‹¢õ‘yÎ'¸~:%Šëë/DϨqÔ…ë:\Ï}&ZSmOså_Q~à©üËʲ  ÿ)ÕÖu£×µçáäkѪêÇ:žÉÏK"Q\_bý°^:Šjëú¦Õ¸®gRé)Õv”VUÞ§T?î0y—ó~ï¬ÆÕqp=Vcõºj?ÝÆ}nEõCÁÓû±ZgAÖëŠâ~ªªüõqÔû9ì3ŽxEŸ ½?NôC§T[ï¬ÇiU'Æš/ç“…¬òd^Æ/‰âw€y—û$ú/ˆî·xÞJòó¡¸Ã} îGñ{Ëåß=f‘ÇÞÁy]?Üå‡8à¡^Ä’'Šó2øàÜVýø õ-¨qÔ‹x˜õÆ÷ ëSù_@<ÅǺ^íäçƒûèUæ³¾N>ÔyQtVù£Nðõ÷5×'ÉtN÷«zçT?âê}Aê_88_,ì\'~A=¨ûÇë‡ùˆÇë¤N¾¿çñx«¼nÊø%Ô#ãX/ý9âþÅçŠëˆq•'ö7¸ðÃ>@]ȃûAëðÎçí½|¿ìGÎ'hà|¤ÛEýý^ƒÆë7oÐñ~ýƒæ=hïgØu{ßñ"¿aëï—?ΰëxT¼a÷C¿×ûîËAÇË/òöóRÌÉ?ÚN·êxÔ»„éu–ÚͶ˜åÞÃ¥öóGuiµ7·`ŸhµÓ6ŸE?L›|ôT^ÛL7Þr-¥0Ç57{šzi”þrò¢N®¥ítj½™néçVͧSo'ûB¦Œ¸0 †#€ˆ`Ä0jbŒºÓ´\Z†–G˧Ð iE´bZd2 † C†!ÃaÈ0d2 >>>>>>>>>>!!!!!!!!!!111111111152jdÔȨ‘Q#£FFŒ52j`”ÜéikºÖ4Öô¬é[3°fhÍÈš±5-͵4×Ò\Ks-͵4×Ò\Ks-͵4×ÒŒ¥K3–f,ÍXš±4ciÆÒŒ¥Kó,ͳ4ÏÒíïÓþíÐþ!íÒ¤=؆z}¦C—Ò¿”þ¥ô/¥ýËè_Fÿ2ú—Ó¿œþåô/§ÿQú¥ÿQú¥ÿ1ú£ÿ1ú£ÿqú§ÿqú§ÿ úŸ ÿ úŸ ÿIúŸ¤ÿIúŸ¤ÿGôÿˆþÑÿ#úLÿéÿ1ý?¦ÿ)úŸ¢ÿ)úŸ¢ÿiúŸ¦ÿiúŸ¦ÿúŸ¡ÿúŸ¡ÿYúŸ¥ÿYúŸ¥ÿ9úŸ£ÿ9úŸ£ý+ Œ[±Êø¯Bw®CW2%óW2%óW2%óW2%óW1óW1óW1óW1óW35óW35óW35óW35ó×0 ó×0 ó×0 ó×0 ó×2-ó×2-ó×2-ó×2-ó×1ó×1ó×1ó×1ó×3=ó×3=ó×3=ó×3=ó70ó70ó70ó70ó72#ó72#ó72#ó72#óŸgþóÌžùÏ3ÿyæ?Ïüç™ÿ<ó_`þ Ìù/0ÿæ¿Àü˜ÿó71ó71ó71ó71ó733ó733ó733ó733ó_dþ‹Ì‘ù/2ÿEæ¿Èü™ÿ"ó·0 ó·0 ó·0 ó·0 ó·2ë"(|¶.Âoë ¾[·Bá¿õ 9[_ƒ"oëëPå^èÐmÈ߆ümÈ߆ümÈ߆ümÈ߆ümÈ߆ümÈ߆ümÈ߆ümÈ߆üíÈߎüíÈߎüíÈߎüíÈߎüíÈߎüíÈߎüíÈߎüíÈߎüÈßüÈßüÈßüÈßüÈßüÈßüÈßüÈßüÈ߉üÈ߉üÈ߉üÈ߉üÈ߉üÈ߉üÈ߉üÈ߉ü]Èß…ü]Èß…ü]Èß…ü]Èß…ü]Èß…ü]Èß…ü]Èß…ü]Èß…üÝÈßüÝÈßüÝÈßüÝÈßüÝÈßüÝÈßüÝÈßüÝÈßü=È߃ü=È߃ü=È߃ü=È߃ü=È߃ü=È߃ü=È߃ü=È߃ü½Èß‹ü½Èß‹ü½Èß‹ü½Èß‹ü½Èß‹ü½Èß‹ü½Èß‹ü½Èß‹ü—ÿò_BþKÈ ù/!ÿ%ä¿„ü—ÿò_BþKÈ ù/!ÿ%ä¿„ü}È߇ü}È߇ü}È߇ü}È߇ü}È߇ü}È߇ü}È߇ü}È߇üýÈßüýÈßüýÈßüýÈßüýÈßüýÈßüýÈßüýÈßüÈ?€üÈ?€üÈ?€üÈ?€üÈ?€üÈ?€üÈ?€üÈ?€üƒÈ?ˆüƒÈ?ˆüƒÈ?ˆüƒÈ?ˆüƒÈ?ˆüƒÈ?ˆüƒÈ?ˆüƒÈ?ˆü—‘ÿ2ò_FþËÈù/#ÿe俌ü—‘ÿ2ò_FþËÈù/#ÿe俌üWÿ ò_Aþ+Èù¯ ÿä¿‚üWÿ ò_Aþ+Èù¯ ÿä¿‚ü!ä}ŠCŸ¢ÏÐb(z }Š~CߢçГPôÚEPô…‚cèï àú7Pp ý=|Cÿ ΡÿïÐÿ„Šûzü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Àü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡Áü‡ÁüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀüGÀ? þaðƒüÃàÿ0ø‡Á? þaðƒüÃàÿ0ø‡Á? þaðƒüÃàÿ0ø‡Á? þaðƒüÃàÿ0ø‡Á?þð€ü#àÿøGÀ?þð€ü#àÿøGÀ?þð€ü#àÿøGÀ?þð€ü#àÿøGÀüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁüGÁ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ üÇÀ? þQð‚ü£àÿ(øGÁ? þQð‚ü£àÿ(øGÁ? þQð‚ü£àÿ(øGÁ? þQð‚ü£àÿ(øGÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁüÇÁü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Áü'Á?þ1ð ücàÿøÇÀ?þ1ð ücàÿøÇÀ?þ1ð ücàÿøÇÀ?þ1ð ücàÿøÇÀ ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§À ü§Àü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Áü§Á?þqðƒüãàÿ8øÇÁ?þqðƒüãàÿ8øÇÁ?þqðƒüãàÿ8øÇÁ?þqðƒüãàÿ8øÇÁügn…b‡3¿ Åg> Å.g> Å>gn‡b§3÷@—@ÿŠÝÎ| Ú†~ŠÏ<„> Å®gž†®€®bç3› [¡;¡ØýÌètŠÎŒAÏ@§¡øåÌ¿„¾ ý×PütæßB_ƒþ{(~;ó 3ÐÿÅgþú:ô¿Bñë™ÿ=ýPü|æ-èèÏ¡úýÝhAñÿþŸÀÿ‹ øÿOàÿ‰(þŸÀÿøb ÿOàÿ ü?цâÿ ü?ÿ'¡øÿOàÿ‰Pü?ÿ'ðÿÄV(þŸÀÿøbŠÿ'ðÿþŸ8ÅÿøÿO¼ ÅÿøÿO¼ÅÿøÿOÌ@ñÿþŸÀÿ¯CñÿþŸÀÿç øÿOàÿ‰ Pü?ÿ'ðÿd Šÿ'ñÿ$þŸ\Åÿ“øÿO@ñÿ$þŸÄÿ“K øÿOâÿÉ6ÿOâÿIü?9Åÿ“øÿO®€âÿIü?‰ÿ'·Bñÿ$þŸÄÿ“CPü?‰ÿ'ñÿä(þŸÄÿ“øòU(þŸÄÿ“øò5(þŸÄÿ“ørŠÿ'ñÿ$þŸ|Šÿ'ñÿ$þŸ<Åÿ“øÿO^€âÿIü?‰ÿ§ZPü?…ÿ§ðÿÔ"(þŸÂÿSøjŠÿ§ðÿþŸZÅÿSø ÿOµ¡ø ÿOáÿ©A(þŸÂÿSøjÿOáÿ)ü?µŠÿ§ðÿþŸ‚âÿ)ü?…ÿ§Î@ñÿþŸÂÿS¯BñÿþŸÂÿS¯AñÿþŸÂÿS3Pü?…ÿ§ðÿÔëPü?…ÿ§ðÿÔ9(þŸÂÿSøêÿOáÿ)üßu‘âÿiü?ÿ§Añÿ4þŸÆÿÓPü?ÿ§ñÿô(þŸÆÿÓøº ÅÿÓøÿOBñÿ4þŸÆÿÓ+ øÿOãÿé­Pü?ÿ§ñÿôÿOãÿiü?}Šÿ§ñÿ4þŸ~Šÿ§ñÿ4þŸ~ Šÿ§ñÿ4þŸžâÿiü?ÿ§_‡âÿiü?ÿ§ÏAñÿ4þŸÆÿÓ øÿOãÿ³-(þ?‹ÿÏâÿ³‹ øÿ,þ?‹ÿÏ@ñÿYüÿŸ]ÅÿgñÿYü¶ ÅÿgñÿYüvŠÿÏâÿ³øÿì (þ?‹ÿÏâÿ³[¡øÿ,þ?‹ÿÏAñÿYüÿŸ=ÅÿgñÿYüöÕVçwó¯ß£-ï#úž£ïúey?Q ~ßgôýÆåÑÎq¾7é{‘òSžó|Ÿ2ßÏô}Lßô=Mß“TÿÂ'æ«ï‰Š»¼Ç)õ~OèñDŒ÷=Д¯¾Ë’*G}˜¯Ý–¶›×µO¾—*?íì{¯Åî­¦òóË{²3Íñò)8à?óŠ_C_íôdðó½Wçkÿ´·ø½^pÓö=áäSâúhŒ/ïårÝ÷Š ç‰'ú‹¹^ì×}Ϲ̫ð+yÈuóPÿ·¾÷¬}ÅéuíRâ7®û§~´oñ8N=KÊ8ßïÖo¾·Ýõ>uàY6zÌŽï{âòÕN'ùžµó´¿þ³.ø^zyÏ[y\÷}võSþ`Ì+ï‘s]û”ºÖnòUùæ{ëÊ‘ÿ·äÛnÊ¿ïá—ùŒ÷}þe1O*>ýn|*G½J:|¥E~à²_»;yÌ3ÌËšÞ%®¤Ž¾ÚÇyE>óÒ>Ö‰’W´Í;ÿîAÿ¥žÚqãüûŠ\K=âºy$?ÿ~#ë|Y7†šzj7óÏñòUý=ØjÎÓŽ^/þn5õÍ:¡ß2Þ»ÖÅvSÿ²þ„¾%†šxĹ\¾Î£m[ÿ–Åõå•ñe}i5õPÞb\‘Ë8ëKÙWˆÃ6òÊß«ˆ‡qþLÆö´>–xk5ñ,y%¹¾|¦©§óK=œ ½ú)7ð)Ç¿÷Ñn¶ý»£Zœ%ŽÔ#ÿÎGœ¶­‹™e_Åõ´sÊñzÉËЯԗÐWýR¯ò÷J´s=*ëäL“¿ür(å”u4ñÐμ*ûˆvWê§þÖ¹Gí&Ÿ²¾C»ö©ò <ò·Ž”zóÊ~N¹Ç¼T®~SßR¿Åói°Ò6ŽŒ_ý”õ®Ôæû÷qe}‰qeÿBûa)×Ë}˜zÌ4ù&îâÆ—û$çÙ5ŸÕÿ›­¦¼ÜOå><ïoÊzO»ø±ó¡ú£¬´ÿ&æ¥>…/óŠ?¸n^—õp¦IÅ[ö©­&ý¢½J^Ó¿zåßý9^=r_]ôjÊ)üÅ;ÓÄåxåæ>¤ìÛBN®?¹nH3ßÿ² ½õsþݦòü»ÒR‡ãrŸ©=å—u"ïóË~’~íà|¯—ºÜnÒ¥ÐRo>åz‘y¡Ìò§ß:Tö›ò‰~í ŸeCÑßjÎËû-íRÚAå›÷¥ŽÏ}HîOÔ§¬ûCÍqög~g½ÉvîÃÕó»µqÊåzñsÈ­w¤þöÿÞ´£×óï—K܆þâ2Þï}€üÔ[¾yžáúPp´›z+ßëeŸò•§]Ë:Ä8×çÜz_”vﺯ æK¹Ïe~žoÙ/_×?ñ”:I»¬ÇÊ ~%¿†šòòœ¦œgp=Ï9ÿï\l5åe¾ä<×õ¼¿T¯åê'N¯3.×ïòwîð×nÚÅz•ûº´›qUò¹Úÿû7^ó>,÷•òµþë×Rgfšóó¾R=µyTÖAæ/K|ê?Ôœ_‹Ãå3ÍùòMÿ8^;•zNÙçÆõÜç¥ß²ÞhgýSÖÚîÏò|¤Ä×˹¶ýÌ/ûñÚžiê“û(ãÁuu©òÅÕnâ,ë3×+ãG?UïWè7_ŠÜ™f¿xõK9°­>Á׸,û%Æç¹ûliÙ 5åvËÎ\ZŸbÆ}?æ[¯Ëù|Œ¿ÏQê xJüÍ4qfü ¶šóºÎ‘o|—¼þ®sf¨÷óyÿUÎ[‡šó'ŸÜçy]‰'úgëJÚ-ç=@9ïŠqeß5Ó¤¹Oë:¯urþ6®«¯ßwI{,jâÍ}Eîwc¾ã¬[yŽœßS1ÏÔÓxËýµñ_Îû‡šú.EnÖë¬ÆÝWƒ_ÆyÙo«úkýê¹Ärúó úH\מâÓ…OëÒ|ÔÃüz4Úy?¯ÝõCÙÿ·šøJ™i^_¦þ­æüR¯_ó9Q®KÛÍùæG©“3M¹Yr]òzYÇiko÷cù¼4ïÃÊsõ œ%¾àožWêc]Ìú[î[ ú·ìWäÛjê“Ï#ÕK?¸N•û'ñ„Ë£Ïß²~ÒÖþY?ýŽÙ2ùÑ.ç¥A—ë]îcËóš_ò(ð–çíàc»ÝÔ+ã=Ï Ë¾`¦ÙÎs\?Ë}‰úC]ò\ª<ßÜeÝUÿÐS{µº^ÎËࣽËy\Ì_9¯H½ƒ:/ßË(÷ÐÜ÷wÙZêíò9¯C­›å<¤Ýä[ö{‰—qÚ!¿cWâg¨9¾ö(ß›Èçc™_êõ Úe)ùÐ6.Jü·›òÄû mó±Ä3ýæ¯òoÙשm數jê#¿‡c|×¹[»ÉÇëÚ#÷óµu"ïÏóŸc¼öÿ,×ÿ@{È*¾?D¯ûh/?ÈÿíþTOõ/öƒÞ 5>ôwñŸó"^ï‹q·›?÷Á·Ä©ò´·qMûŽÈ'Çë7ãÉëÚÃ<3®ÍWq+_;Þxä¿Xý"îõ§vÎ|+ò‚ŸŽ“Ïâˆ7íüy®/ œÚ3óþÎÈ ým½p¼×õ£¸Œý§ï >‘çÅÞÚù·$~ë_•8¾+ìxoÄñ­œ»á;v0¾˜Ÿxµùæ|í±$ìQ®‡¾Bûþ°—þº]û„н"Nw[à¸/êÌ=aý!ãÖ8%¯Šå#.ó̼ÔN%¿£~6ÖåZ7Œ§»Œ{í|îxÊ84OŠÝö1δS‰÷Gôßv±ßü*ë‘quD<Æí=¡çÌÿGšuZ{?X?¢n˜'e }r_qWØ%ãú輢¯õœqƃþ4.ÔSª}•›uÉùÖEñ—õðß5qçúáuõ2.ÌçâOóI¾ÚÙù¶™g¾Zgõ3ú–x(qíz¡à÷ùð÷—ÃÎeÿ`>Gw±vµm‹+ü¿ÃŽæííÁ¿¬ã‘§ÖãS¿ßÛ´Çû´Èß»W©'ôçzšûTý¤Þ‹COé]aý!î²>…ý ÖÓR¬SÖGã¶û¶;_‰“ˆû’¿â6¬gáõÓ^ÆÏ½a§¼?}Šô»x¢~+ÿ¶wGøÁ:Pö_QWÊ~Qùâþ²¿1~b½(ûOõVO×AíÁør=¨ñ*Nã(÷‡úQ;dœÚ.þ ?[ÅáýØmêÇ8ýUâ#òÿŽÀWê…yqá¾Á¸,ù,ÿÀ×u¿bÜ[W"¿¿ú_ÞgÙ_äVü]î³Ä!ÎŒå?æ)×嫿ò.š‡æ㬃ÆK‰hîOŸ²_Ñô‹_kOãe â_~ÊYòGÍO¿F½.ëŒqi¼FüéOóÀ~ã¢ì#µ_àöºë9øÊ¾ì¾àSöuÎ{Š#ü±8ÚÆ½÷‹Yoå£=_*ëºã¸®¼Ï«gÔ›<ÏÈýµö+ç5òÑNŽ7n"?¬æÑÁÝKÜYWÏÈÇR§ÕGÜ‘¯®^§ñ >e {k7ãõžÐC»˜ÏYoÄ—çˆå~P{©¯ñú•<µÿÍN{ ô)ç`ÇÅŸôÿqÔ¯û÷Ñ.õÃ8W¿È«’¯‘—å!ì\Î#ÕG¾ú9ü}gð¹ÿšù&ß’—Ú͸ˆø(ñöP¾ë„öÏsØâ߬ò·ÎD•z­†šzÝúºnëñi߯èÿ¨úËëw¿û¯õõþˆÓ\Ïʾ‡ëî;ƒrnñhüˆGùôßí’OÖýÈ;êK©ËÖ{ñÝñ•q•uÖu®ì'ôsÄíýð¿M»:N{)Ǽÿ8Ï|ïº¿ä¹Ÿí’Æ!× Nó?ôq\ÙïÉϸUnè]öýê©™[ØËü+ë¡ñb¼iwýõßûÅÏF})÷M≼’:ϸ»?ø”õÈqòÑŒw¹1NÂ¥¾§‘W¶}žRö‹‘åþ‡þrÿyPÎÌ3ë1ö6Nè/yRžOˆ'ÖgóÛü)÷/Ê ÿªwÖ±;ÏÝ‘¯yîQÎ%µ«ñ¤\õÓï•:SÎ!+~3ÿèz³8ìPÖÅÀWî/´üJÜ[Oˆ³²ͺþº;ø•8ÕÖ…Ð×8°>8î+4ýUÎ)µ£þ4ï#nÔï ¡—òËsYøå¹uYç‘Sê†õ”v9o ?k¯ÛœßjÚ3ÏõÄ‘÷yâ)÷³ÐÛÂ楸îŽëž£gw†=²ÞÕöK´‹ú¤Å­ÞÖKäßÿfC¿×åœÙ:¦ßõ·u>üTž£F=·Î”÷1´{èYòX~áOýRžSˆß8Ó߯SŒ+ÏÇ#O´©SÁg‰õ2òÆu»Ü@Ëy¿u1ä—ç{ê§üÌýqn~\Qÿo>ž×•û‘ðç]ÁO=>ë«çÆî;­[]÷±ÖõÀïs 㯜³G~+O~e_(Eþ}1^yŽ+òc*ë¬þN}Ã~ÊùläEáƒ<ëK>ÿ´þæó¤¼ßs=½'ø–õI?2Îûв¯¡ý¥Wî§ì—?ô+Ô’GÖ¨7wF¸#ìZòÁ¼ŠõâÎè/ç´óý¢²Ž2Î}d±›ñuÈõ¨<·ò:rÄ£}r¤ýb<ø ~õ)ûjó“ù¬ÿÃe½¡¿œ7F5OÄc^•zoÜjOõw<|Ês²°ÿ—¯v°.|Ƨ~4¾³n>óTüe=qw‡œ{BÏ<ß(ëZıë\>O]q`Üåsº¬[e?b|ļò<@Ê8ë@ž/{ÎïU‰»œ÷WêXYï"^óý–|®žï…•ýµq§ÝÌýbégç^ýUž¯ÅõÅÁ'ç•x{æù½ëZyßB?gªGÚ/ä–ç†ê¡¿´íÏ«Oø/÷¥ÿÖƒ§ýS¾ú‰×x@¾øË¼ÈWí[ÖGó&ô׿æ#yÔâ=Ú¯ÿ¼£Ï×Ö¹þõοA×*íó9_¿Àõ·b¼íÿã~^¹nÛù¶•{!äÿœö›ÜEå¦þ‰ãBПWæÉÏqï6ǵ+ø«÷ù&mߌži¿´Ó;}ÚK¾-úm"æÕp¤Ýß`\ÊÍqi۾5æ½ÃŸïVp¦=ßF¾z¤?ÅYÃ[³Gâòú/c¼ýiùê?õ’f~$ÎäO©ßØ«}SÈ/z¿~Q¾z•üaž¸íO|oŸÌÓ´Ï{aGýõv\¯ù=iÆAÊWn-¯3ßÔƒy%>3GÆ|ôOW~d>dý̼þES×å®þ¬g?µ8ø2žºò+í—vWoýþ^Óž]qcûÝÐïùú'â³Ø9ó'õÊ<3®#ŽÛŸäzÆ}­~d;å÷ÊóÌïs=âô|ðÏx‰úPì&Ÿ7Ã/Ù®é™uÆþÌë´Ó»Á'ûSä“miæWÆOúÁþÈûª?Ó‰;æ•zœò*u¾+ïÞ«à–õç‡}ºð井öþ)4ùÖê”ýÊ˺™q›yU©ß]mÇé‡w*úœ <æÉï…?Ô7ÆuÝOü2ìUËõc¼ëIÉ3óîs1>ïãþ b§^÷÷o†²ójõíݰK®û里°Sõþ¯W¨ÅC-¾Ó9ï½Ð³¶®ÔâT~¹.êWë¹u¯×¹‹ú(W9Ú'åfþžoêÛþXàJ»fýìe×Z;íý™?éŸ^q¨þ'æaM?ùhÿÔÚu™ñ–õ1®·ÿ<â˜ùíVÅž¡O—qT;G“Oíþ-ÇÉ×þ ÑN»¤<¯¿툷®üK~ŒëºŸÌzñó°Km¿©ýn ÿhÿßFŸZ}©­cöwóDœæýCíÜ.ìÑþæ¥?R~åܱˎYçÄ[»ŸÐ_ʵþdœ·Â~_G™G¹¨Ý¯ÕÖô[Ðö?Áï¹výýÃÐ'õËüéUŸÒÏ៮¸L½Ó.·âº|2^ÞzЯÂîEßZœ”ºV‰óŒËÿ5þôâÿûÇ˹hÚ3ó¬VoµWäwÖ¯ê}[e|W]¨CèÏß}3n³>!¯¦W»qõfèWÛ¯¿úfÜ&M{ÔÎKÅ[©ãå<$ãèÀ_»ŸÊüÏø‚O×}[ê™õªv¿–r¢¿¤=v«åòË}—×s_"òÔçú=Ï5áßµ^¦_sªí´G¯ç‰•uK=|žÛů×ó å¼ø¥ï†þ—íE癕ú×µNÖìù^ÅÑ+ÞmÎ÷}„‚ãÖÈ÷ôkÆÍ¹Ð£vôvEŸÊøŒÿjýìwÿ¸•ßÅ7qÔêiâI¼µs)ùX—Þ üiçÚõŒý÷OcªícÞjâj·.OÕóþþ7BNmŸ”úÆŸrkõJšç"™•çåï¾>ñžu9Ïé¿Ôô›vêª×•û¥®û9ù.®èŸþ>ߜ׵…_«þ¯=ßóþæ“1.ðVëPí¼¢×óãð–Ð3ùç}|ÚË|ÊýSí}‘7šó»öQi'ëÈ/›z–}A-ï3OÐÏ÷ªçêµúP9çëŠû¬ßiÏŒ¯ZýÎx«Åe­^äyÖùàßu?v«=Çï¥wê•ójó3j|´ë{Íx+qœv;âÒvœÃVï—ÎÇõ·Ã®µ|7Þ>ã*÷å>*íTË‹ôKÔ“®ëêóG/ãÀúœqoÿoǼ7š8ªëOêã¸éÿŒãÚ¹›z﬇òOû'.ç¿[±_ÖMç;>㣒7ÕýA-î3nkq–~«ó–sÒ¦½ºö±ŒïÊãZÞ(§R?»â³×ýwàöœ®ö|»<«ù¹v> mµnî¼G{®Ó︲ïMÜç?ø\_|®×õií|X~ú#Ÿëد½lÛñÓõþ…üÄ—zeœý<äh/íÿFg\É3ýz•8˸ˆç¢]çD?Ežô¼üšã ÎXç˼Ÿ…^ùÏëïÞ´‡øjûÕ¬'i?å½ýqÕj5Ç—8Ê|½PÑ;ÎUºžçåû}9/ô.qòvè‘üò9·ú‰Wªßn æ}W+øJÕ»öž›x._õúIÄ»4ý—ñòNØ1ësÊýYo×>Ôûçåü|ä\ðuÜ›!?íð“À¸’oö«÷›úŸþ·£vþYäuÆ‹úçú¡þè[žË¼~ȸHf~§ÿ”“q[« é×Üw:.ã#ùA»ž‡d½·~e½È8Í÷¹Óÿ¿­ðŸüK<¾ø2Njï…‰#ííõŒÓäû®ç†?kêYîã’O/þtÕAý™ëZ>wÉõµö¼2©ù˜ïcëßZ¼§}ÔëBÐZ¼Ÿ=òù„üÄ/í`|d¾½ÕÄ]òÖuÈ}TÖ£x¾Kù‹+÷¡êzçßÙtáªÄU×}IÆmîŸÅë:™ë¡úÕÖÿ\wÕ7ý™y—uB\¹¯øEê]‰ñ)_Ü¿ ûœkΫîÎÛÏxííü¼/È:ê8iäOWþ3¿¥¼¬ÏéÄ븬ögxèªo_:nºÖû³î&~iê§Þêi¿v¼ãjï[ÙŸzgݸPÑ#ý$Ÿ\§—ûDǧ.4çuÕ=ù¾¸Õ;÷»Q[üJþ&ž¬so„ÞÚ%ëLÐrŸäõȯ®ºWÛ—Ôäõx¦ºß û”ºm¿vËüVÏÚ>8÷ ¹®ö—zfÞÕÆeÞÿ²׉?÷éõvÝz«i¯®çG©_äC»z溠k÷BOõË:›ï e¾§\'Ü·§ŽWîO¢vSŽó3¯sß$¿·C^¯ó%äûëZ·ó|RùÚ»×þö\“©à,rc~¹þNø/ñdœdüfÜæsןF]¾ó]¿­›±.´ÕS{8ï§ý¤òÉõOùù¬¶?Ïy‰?õz#øÖÆÕžGf^ë—ŒížþËqšrºî[àÓõ|'õÔžÆEÜO”xÊ}S휮ԧ¨ëOyÞã:¡}ߥΩoe]/úz]~ú)ä¾¥>.õ~7ÆG<œ¡wõü$ׯȫ‚Ãzõ3ên×}€~”Ä_±÷OBÏŒ»ðo;í[{>!_õ®ñMz!øj_åVΕ»ÎB¿®s©q‘ñq¾¢§x2ž3.ÓOï„~‚_ÊÉûӷßïôð‡ùoÚK>q®\Æ'>ôÈç ]¸OÖGã5÷{Y/ß {¦>꡽ÞhÖ]ûÌ¿s1.õ)r#ž2ÎküjÏËk8kñ‘üµ³ò£>tɧi§”'¾b7qÊ'õϼO}ùŸzf^©GÆ›ãЧÜGîòsà¬~#⢫Þ纃¼®÷ r¿”ö¦ÿß ÿ¿Ó\ÊßÉfœÕ¾#’û¤¬¿¶+õ ë|#ç%ÿðO¿ÚyEÈÓ.E®ã·ÜZþf>ÕÎ’ßOÂN¹®D¼”¸”¿qV;O|7ôLšöP?ëý­Mù]çaÁ¯ÄKÖõTN-Ní×? û¤Þï…¼ìÏýGÖ¥Üwæ~&ã¤×¹zÆ›ö4ÏrÝpœ~ªÜÇWÏmä—~¯ÅgÚ)ÏÒ>¶ßŠvÎ9]ÏYÒŽ¹îôª[ÉÿÖ¯þêéså\7{ùí͘—ë—ñþðW=I½n©ØAª¾yåøÌ[çõª?Æa¾_T{)©þ<ze€.‚®ê(ÖVG†:te«yý)hþ{¿þ{T1ïÚƒ3º„ö¡+: ´Ÿ =‡>ͼeÐï 5ÇmäºÿîÈ3CM=ÿ º†ë?êô´ÿšë?\ϵ›8žjÒåô?Hû1ÚÚg×ý¾ÐêsPô|œþ•+:tãüNØôÿ íÇ ;;Û«˜ç÷·—CCW2kèñ4z®¢ýúhU¼èùÝà?ÈxñkùóÝ•v›ëKiƒÌÛ$¨~—¿¸ñs9[ 5~¿8U_ä<ÿjS?ç=ÍxãчáièfãŠü˜·¾Æ·þßÕ/ÒÖ~ê«>Ÿ®9סÚáYèŸAͳ‡‘ç¿$Þ‡¡æ©ùþƒÐ¿äz/æ¯|•§=¾ ß6í¨vÚÒ¹Pì»´Ý”c\m†Ïzè³í&ë‘ñmëGíü¡K·õãj®?ÏõU´×C7¢ïzäèõ°~<.ŽÐãkâ„êý •¿ù¾¹ß¡ßøxázÒo|lƒ~ú]ø¬Eóh)z[×¶Âç9ú¿xVÓ¿¹æ±8žW?ÚO0O»¬…ÿ:ølèPí(>ãÃõÝýƒu\=v!ϺótÅ«MýÖÁw5|WB­/îkÜw/Eÿ?….’«©'ëI·ƒ¶ûð×ÚË8؈\÷ æõ×ÕŸ¶þY ¿5º 9k¹nýÕ«i»X·¾×jâ´PýÎ5å¹.f=^ƒþƽõo9úÆ­ùµ7ôs_µšyÛ^ïPãS{?}>ÖÉvàxœqÖ?ãËýÖó\'o»þžm#ú™7®_êËxóÍ}ëšz©û~íà¿«ã¾É¶õÂ:ZÖ#ú­kú»øŸëÖië‡qe\®wÞ¹&óßuÈû ïô»÷«À«ÿôÿêê¾Òz²z¦©_‹ëÐßü°._î—ïºéýúÙ¶.?>÷Cúß:æz,_ýcÝ(û1ƹ/—uÇzåý“÷+ÞÇ©ÏZø´šý[†š|×rÝøs½³ÎX/ä·ýµuÝ_µ>Y77s];¹oYó*”þõ´­‹®gî õ¯z¬„ûŒ]íæx÷®s;ik'×uç º ê:æýu×Ý7Ú¿ª¿oœ[7õßÁÑÎÖ-ÇÉw|×tôþ`¿ú»oñþÜ}¤ëï_5Ýwçâ5ÿµ¿uDÿZ?ŒcêPÑoxÜG׿¯ó寿ò¾Âý›÷êgº^µh[OÊ}7|̧­‹šxݽHÛzåýfÔÏ‚ÏÆsíéý¬ñæú¨ïãÜGª‡uÑý³8X·ŠÕ«ÔuËø÷Ü@½ÕÏ<æ\¥èï~Ôø\1Óœg÷ñƒŽC®Ï\GÍ‹V϶ êú`ÞXïïý°y”ùÐ¢í¹­ø}ža=ñWü`~èà~~ÚÃûÝç᯽—Òö¹Œëœ÷wfšz’7E¾õÛzu{ô?ÔufÏ`‡¾H[{ëw÷Aò÷ÜÏûüÕ3ͶöÑ®O®>çðÜÆ:éygY/ Þ‡¸¿lqÝúnþx>ëýaÆ»ñç<§BŽë‰ú™æ·që9‚û9íê9”ur9üÊýzr^ü#÷5ÆSžË·åy†×‘gÝò\C½\ Ï£‡ñíþ×:e|º/ä|¤è;-÷wPë˜~¶ž™×î{]ró,òô‡ñ¡?Y7Š[àk^yŸh^¹ï±N¡O×yŽûÖçr]»Û¿ ºˆþ•ïAú·¯èPí°„ñ›ÑÏý®ëßÊЧìàë¹ë¡þ·.ä}œqìùµûýl\¹o+÷íŒs?œëòÖr?à~Êý„ûAó <·¡½ýõu?=@¿ûd×!ómÏ™·Rô7ÎÜ϶šò=÷XÍxë“û.ŸsxNçýݧcžçæ…ûDícÞ»z^áýšûÜ%Á×<^Äuós ôpß`\jwí©ßÍóÆýÎ6®{>èþÞûgã¡Õjö{>ãù„þuœõË82~ÌkÎ+K|ZW½pÓîúY¿zî¢}?Áüv½Ôå=‚5¶œëÐ\'V£‡ç"®‡æ}¹÷1ú׸p½ÑŽËè/ûI¨uA?úüq€ùæ­çq>Ó>Ö¹Õðiq}%üµšü=_Ò?>g-þŸv¶ý­˜çþ×}ÛrÆOî¬Ögõ„za½°ßç¥î3ôŸëì‹èm½(çå3MþÖg¸®¬{®/%ÿák~’/%^]_}ίôƒr|_À:ëy¹që}Šñ™ÏÙ¬#ÖKŸï×®ú¥¼_B¿ë¹çÊóœ[»˜Wîï\‡\w¬3ꉼì¾rž6Сæ•çˆÞ9ÎóÀAç­hò5ŸÝ?{ªýÜgZ‡´›÷Ö…–z 5ç[g]7q}ãÜß¹ß_ÉøÕŒ7޼otýõy…÷ÛÚC=–@Ý·ê7ÎÑ>X·é×ÏÖIÏô‹|­_æ«ûwŸ§º_TŸŒkó(ïK©§eœú—÷ˆ†šzz^•ë‡mãÂý¯û1ë¿ûFç•}8ã=oô>A;êwós‘óé·>³¾}p.uT¾ö³î{Nf\xß`œz>âûjå¼½<¿Yòó½,ë¦çÚÖmë±û[¯kWóõ¥ðw½×N®§®Û« Ö+Ïçlxü]k\7ÊýV¯ºp½ïK/—o ÷õŽ÷Fÿ]«sÃ…ú}XÎ[®õï;þùþ]é~ïFóǵη«m¯ëåþåF‹“ü-tž\ë}Ý|ñù°øy¾Æ_oûÙëÏBñ½Öùw£äÅBçñznøÑ¯ó›ïýÄGöýw­íóQÝüpþ.a×¾GÛjݸߣõ}íý{´þ=–ç¶y>Œû°}V¼ þ=ZÚù=Zãùºýí«àl7ù\­ïÑúÿŽQ>êQ¾K)ŽÐc¾¾Gë߇ôû=Zÿ.äJ¿GëwÝæû{´þ]f~Öú|½~Ö¿ËÊïæôúmùþ#|¯Õ÷hõÇò=Zã¯ßïѺèõ=ڥȽÒïѾˆÜümþ=]ÏïÑB¯Õ÷hÓùúícÜŽ×:t¡¾G» ¾óý=ZëÝ úƒã§¾GÛ¡ õ=Úò½úçú=ÚåȱNû½«+ý­ó?ôߣ¥}­¾Gëß?_íïѺ^ñ÷há»»Ýÿaý-ûÎö+¡ÿƒÐ…þ­~º^¿Gk^èïÑZ¯ô{´Öki¿ß£uØó{´ð»Q¿Gk½3¯—ïÑ®îô;Þ0ߣí(ô¡ù­û³Û÷hÍËkõ=Z÷ý~vò®ö÷hÅe¾äwï\O®ô{´ÖÛåB¯ô{´Þ_î÷h]ÏËw(ÛM=.÷{´æÏGߣmŽ×¾ý~ÖúQû-ó‹®ö÷hùÖoï+ûþ-óþ±~6¿úÑ÷hé‡ï\¿Gk°®ß£-ç7Ðëý{´«gšø¤¿ö=Zõ¯ïÑZ·¯Ö÷h·q½×÷h—Âç£ïÑvhí{´þ;“sý­÷IµïÑz?tµ¿Gkþ]é÷h—ÒÖž}¶CK\Ñž¯ïѺߺÒïѺïºÒïÑZÿçû{´åœë—û=ZÏkߣõ¾q¡¾Gkž_÷ߣE¾þžïïÑz¸Ðߣõ¹ú•~ÖºýÑ÷h;´ßïѺžõúmyŸ‚þ¾GÛ¡î›Ê>½®ö÷h]çÛÎCÿ´ÝϬdÜ|V{ô=Ú¯ïÑê§ømí{ýRWÊg¡äÌuÞåʹR\×JîÕ’w½á™ëøÕï Å¿ßüÍ—ž½~s•S›ß/¿…¶çÕÒ#åô«Ï|ë—¿…Šë~s÷ù²×|ÉŸï:9×ùó]÷jóûå7ßãæ;¿ç*·ßñù[¨8¯º²P8:^kã{]¿Ò¸™o{ôúÍW|ΕßBñÉß•òí5¯Ö?×ëó=ÿJqöj_­:׫Ýëú|É›k~-”¿jüæªïBçÍBÇëåÖÅÚ¸Ÿ~ùÏU~¿ãû—×çÊÿjûc¡û*oj¿~ëT¯ñ OW*¾ì9_zö+¯¿~ù÷úÍuþ|Ç×|×^zôú-t>^ë:>×:Ñ/ÿùÆ}¹z_­úR“[k_)ß«½.\íz?_zôË'ûçÊÿJãüJëø\íSã·PuåJõ›/ûηŸú•—ýµv/y ]æ*çrù\-?^®ý{ÉŸ/½®·:ÚïøÚ¼+µÿ|ÅK¯qWzýjש«gWK^¿ó/WN¯yÙ_ûÍ·¿ç›Ï•ÆS/½®”ß•úçJõ¹Ò:Ù¯¼~çõËo¡ìÜK¯~û/W~¿üjí^ã®V®ý:o®í÷·Ðyt¥qz¹õv®í¹^ï¥ÏB×¥~¯ÏWü÷â?W~s­K½øÎUîB×^|ºÛ·Þ|ñÿnjÝÚúäEú±Wn^»ãâ,ºø¿Oqñækø¯[V®_ëîX»×ÿܰ~•óWo_»r'oÞ±uã NøØÖ7ÐØ±óƒqëvýfך‹6„^ŸÜ¾eϽêö;ïë4tñÿ~õ«ÿ÷w·þ ð©5+w®¼wÝö‹S.¶Þ{ÿÿ„òÙÙXsurvival/data/leukemia.rda0000644000175100001440000000056311732700061015341 0ustar hornikusers‹í“MKÃ@†7›Tm RPèµx¥ÙôK[AÅ‹<õº´[(&)4©zìOöT7ÍÌØ$ToB³ï³ÃìÌfwçån(ì¡Í㌌›-®ƒ±jY«)}O'1j;ÖVœ±õ7¸È¨í‚^ƒöAoA–‰>Â|\Oô”¥µžÑsÐf¦Îè èòß/SõpÿýÏd‚Šß_ýÛâ¶iÑøŸuf%Þ?[_ÏÎÆ‹Yêú<õ¦¼PS"b¯ý$§A¤MÁSyžþ†3•¤4òdˆ9(óDŽ¢Ù\Ó*H_a¸ N+šú —†‘Œ!ÌŒÌúò|öÞØÌQƒM‚@pZm„B¡‡p À&‘C$ˆ\¢Q›¨CÔ%êQ A5ÕTC¸Éß›_™óµÇ2’É\ÆÇ–;ã#O-^•?•ûÞþÝ¿-îÿö6Ç—R¬›s,×ÀFìÞ³%\²oÑøÙêžðâf‘survival/data/bladder.rda0000644000175100001440000000664511732700061015151 0ustar hornikusers‹í]ùzÇo´º0²¸Å!Y—-¬lŒY‚Øñ…1NÈEZb%)«…8ù+”¿ü8yŠ<¤ Uik«ï™Y %Óß÷SïLwWWWW׌¦zjîüp~ôá¨R*SÙPŸÊúõÏLÿéSj|DçÃÖ««ÍÖœRýGôñƒºxf»Ž‚º¯iÔ4CÐVc1Ê4Þ`8È0&ð¦À8Ã!È#ˆ£ˆcÇ5NhL N œrà´ÀI)·Óg-8çÀy ˆ·3¸„¸¬qqñŽïjÌFâšïE@넚×X@,j,1,k\׸xñÆM5n!>Ò¸­Q·àŽ?Ѹ5>Ñøi~¦ñ)â3†ŸGàsÄ_2|¯5îi|£q_ã[ßiüBã—ˆ‡üJã×ˆß ~Ëð;ßküAã GˆÄ*CSã± à{5?kü…a]ã‰O56›Õhil ´5ž!žksÄíS…ÿ_dJG…¢èº¿\o>o®ƒÇpvhs½±Ò|´‡£›o­­nü°ö´‰g†Û߯m´›› A®¶²ÞØ"j;}‡# åËg¬\òì’§òðSdœ*±®MgcäÁÇ’ÚV&—œ|²°­o9”B÷²mÊuƒ ëS’´cÆäã]ÖóÙ'Hƒâ˜ú‚'ñãì<<©“õ9Î~ƒ®Ãã!¬ËË©ñ‰Æ{ÒÂÓVžêg´€ßaäïãú˜°Ðšd¿G=ª |‚§àÖ=Çê“<Ùñ ê¬ ÈÁ{qy™Âó‡ó˜b4€WxÚ¾Àêel SÈ<퇧úðd}ëÁüŒ!Ïà¸ÌÚ“\b0'g±=OÜŽgÊ´ë–z”jâêÀ|ƒLO±s#8àó¤¨?¤:úågp¼mºß#}‚9:õ.²z£˜CÝ£ø{û?ŒçŽ#°.a+ŒxJ®±¶cØÚ€^€ÉË“øZ¤ë­ J¤;>â c£¹¥cÞdžÕ'oÈkÏÁ¼r9˜Èiéð5Cc8ÍŽ¡Ï£ŒÎŒ(“ii_cç¨ ·OtEr¢¹‡¹€y¬aNºsy Û¿É6Ño²M'T·mšPµHÞ¿>Ì3Ìá©:ò—¶é’²Û&òüAW±Ï«ªÛ6½£ì¶é]”é1èËö»¤LÛçnª8ÛtKÙmS]…mÔQªc‹øÿ×|í ãyúßåöG:[Cþ¦‘¶´Mç”Ý6‘÷5d›¸÷5d›® ¼G1'Û4«Â¶iÛÃ<-«ŽmZVaÛVðª‚7•ß'‘×ý ŽsLÙmyÑ)ÏPd›È“Î=çÜ6ÂãÓØvRulÓ£´M•ß6A9·M´.«Žm"ï·Í6‘\É5îØ~ƒã Ðwñ#û‘©l>c墛گ”Lyù(KeÑ+«™BëÖÕ>¤g¡ú®y Ù‹²Ç¿_è†ô ¬~c×S¯õR¦T}IåK^]ò¨;躮2•m7cé—5_±×;J!;’Ê«ì/–~½qü¿î;ûfS~ÌEÔ™wŸYp·Zv×¹áþ1gü¸án~~,gL:6vÎ,Ežz5gœ7ΘS°ñÃ31Sé»Y´ÓÅ’» sšæŒæž)0Ϙ],•cdhNAL_ž"óÌŽ²™saŽÝêÉá¢QÙóæ©uKFe?ËFss86bº0‹’è$™¯¤ÊžÙ1•g–=+׬cÏ"òÏ‹´ócÉÝé‚QÇs 0Eç)2;]6Θ–$iåšt<Ê6ïîtÉ(òÈ9Ÿ¥1zæ6WeÒ%;ßÒótá¹™gL6bVŠ©Æž;˜ë©ç¦Â³<=ž$UAÏmOA£—ÄXWQoï.Í‹¹Gº1b޹ª›ëÞ39I槬‹UÒL¤\p“Zå»àç{ÌÄÌÇFéŒÅÐ1±ÇÆì‚ú%‰.Ÿ>ÇÜþĘ”² mYÆØSÙs;_ÖÚ)K%òõ•´ êXYÆ3¦‹|ÿ¬—¥užkÜ.ØÕ‚Â,ëÚÔ»UYPb1?òÝ廿–5MÊc:ŸI~Ö«ÒÊéØéSr”Çž¯+å?vü±rˆå»Ž;½êiôRç/Ô^>SåZ¶|ÊâCÊ!•¯:îö¬OcŽ;0ëòÑ+«ÿ¢z’ªÿeÙ—^ëi¯äY]¢—×^¦ö+ûÛ­~‹æ’ïX¾ŠÚTþBv¯¬ëkÑõ"ç+ïõ¨(ß±ó”Wî.zòÕó§'MzõüžÌÖVñ×H»Õl´Ÿ4Ÿ¶é­ô§Ïž°¬âøTà3ådõ¹À_*0«¯î |#p_à[ß À¿êʺ%ªò*¯ò*¯ò×+¯«*¯ò*¯òý™«øÿÿC!Ml¡ÆŠ %<–­~ˆ W˜¼H•G*¿²¾DH~¡ç7Eû/{¾‹–ï¶>ÈþBarB뫨>å?U1aúŠ„× …ñ+ªo¾V1ö…ìjÑqözµ³!½ ÉUÖ•‡ô ´nŠ^'RÇ›Z¿×zm ˜ÂoHRõ)U~!»’*Ty–m—˶Eõ9D/b6u<¡û$IÏu¿ {&ýIƒâXú‹¤ÿ§&Žy¨ ÷ÃÁCÿÈð?©!€lþ >qœ‰c äò…@á„@üb®°B¶pBç¤ÿhRùýI#¬ÿ e† ’þ#àû‹Å1ñtma‰(Aú‡ÀÆýAÀ÷ÿÐüÍb¹ôßdª;t‘ QD÷\äÛ ­#i—BþTé/¥Ð;Z{‚G®é#ÍT·ôê„å±ùC¥¿3W(ôÏõ—‡ö¡ð>2Ô óã ëãBŸê„û±ù'?þGú+¡\ú,9BþI[¸ Žq,õq íêž&z®ºWôåóݼô]í]Ï‘‹–‡òØTT>½ž¿Pê•üË’oÑT´¿¢þŽ¢ü¥êWÞùÉ›zŪüz•M¯Ëú•¿.úûº¥ØõjK?5íýïÕøbÛïUêõýEDâÿÿÓó¨ê¸:®Ž«ãê8ï1¼&0@/ oEd­|¯C/=ÔšÏ;oP°jÐá_*Ï{ ôµçyµý"Ã0RýÑ2PþBñ@>s½\òbŒ@¾@À_à_{ÎûÿÊ3mìŸ8«ÌÍüç•ýëΡ ûüá—o3~ž øô ‹•Ù·a^nŽ—â}_h¾£º7¶ßUæ——S7°Ëê®é|ú=åÞ`Î7“ƒný¸×«¼Êÿ—òºªò*ÏŸ+ó>ƒœo.¶tfûÊ82Y?´ËÇKˆ6¯ÃùÒÉîÛ !ûðõ%eà’‰ë\ÌØ¤¬¨Ü·¯È&¢Ú4WÖ÷l]ºfÛ ÁeCúìÓMÛÜÄÌEl]>OÄ'ñ8ÀÊ‹lò²ÍOÌJyÇÎ?£o³•ÔWÛü»dgãÏ·f|¼Æêyì|ÇÈ4TOê­­·sP‡ôÙ–öë7‰ÇÔo *Õ½Q†R?æ1ß„ûÝ@J™à•on¡rÚÜâKÄ'×QJÔÆŽ{ñ:H|s¤<ߣãçdâ›7(±ïÐíÜoÐÆ¶ýþ º¼ßžãºL›¾€vÊ÷æR¾3Wd£Vì·Ívû›f±H—]ß/+ðÝ2§}ÞI!G‘,9¤\å©ÿ?æmçj¢J»Íwˆ¼üå•W¨ŸT>\ü§ö[–ÜËâ;TÞ«yˆM½îw¯äŸ—Y/”JЋ¤çÍõ@¿²œŽ«˜¢åòÝ«¢©ýIú®ã¢ã.»ß½ž—²ô&¯ 9 È”¦Rwí.Š®£0ùµŠ’j¯ÿJFi êëŸÝÕQæy=ƒÊ]±-ÍOíÀcL™q7?Úãñ.û…Žw¨˜WšÄN• ·²†Ÿ¢Ê4—ËB<‚¶ÄâFªY5Ëö|”@õ×7Õ‰ÓQcSªäÂ#SÔrZ‹Ò¶Fí¦—\=?ߢ®‘zJn@êµÕø³ù¢Ëh…uE._¶Tü†Æ›™ûžGSiÙ‹mhv¶ž›ÜÅ ÍÕò®õ+7¡{ÎU‹,´Ø±°ke=‡–£¼ ‚í hÕ‘reÒ<­{X—Ë~8£Í’…å)'J´ã9æ³”yí}˜^uwK¡CÍ(÷ßÎè_»Ÿdî¢s®|Nêç›èÊm}Qú§1ºéî;3‡0`qãõ€52´1Ɉüùe˜ˆ]ä2/#•êÑOnâ0/dR2+ ý4…L‡¯@Û~BD•¹>43‡U¿³Mº§I]Ç’¸ *ý_‹%”ž÷zÕ ¥]¢6!úPÈyùøÏcP¢¢ýô<ï»ÿí—&˜Sìï_úßùU=3Yç¡®ñæ-KQhúiôùƒŒ ´T \ÚC†bïı!dÊʧ[$ùŠ,9Q#€l[„-342îj@òØG³#ŸFšöƒ[(@:Ï[œ¨Œ‚»¼ÙdP„ŸÝYèìY þÙSú.%©7oãÑöD›I”™½û}5×å‹lÖÆñ˜nô÷žÇ|¨ :Ô<Y„'޶ºGÝáAeÇ›…BjTýBªµ’×(pë… wª ¼ kb¤F %#ñ6ÔL.ñøjsJ?݉iB·ݨ'ÿ\BüîkÔoN­°nCÃËÊÒ†4¶hÌü®AzMŠ)‡d ï¢Ùñ—VU2}h~(’¯ŒŸÍG¹âü›ãÐ"/9Y»ö#ZÆQ¥¤ˆR£•ÿÍkwºMÐÚ54Éër"Ú8Šýgx4 m/’FszvïîLnñØ£ý^Y­¡P1:†3°ü´lAg=M ¼…®ZLÑ›gðñCnnq‰CèÂñXˆéI/Ê•]#‹5ÿ‚´ÜRKfö`v¹»ŒÂ†ÒðiA'{ï…šïÐÒ·JY¢ Mÿ¨Üoè¬Býõ©–ôj¨«òÑPDMIÄZU.áîiN®P%öøáÒ#¨ú ò ?ý Ôœ%\=Ä u!Äaž?u¡qQr¢àÊ"´„=Ö½cÞ m™|z%óБصa󺂫ó˜®@Ïý“sç¹ /Ô¨ˆÃ+$¾ˆ¾þÆ ƒ/$²îTzÀPLíG†‘d¦Ê/.º ê½¾{‡aøf®]ñŸ@ˆZ†þÅša”h÷P7Œæ5­¶‚±ÏÂzæ0PiSw» &ÂR,¼0„ÉœÏ\1ž0M¿È~vfʼ2ÇsŠé¥3ÖámÔ“¶Xäé¶<ÉRK_}|v¿Ë«:¯´?UÁªíÖ%‡µÁç=“ò¦°KF¯}¶Î–'Î(WÁŽÏä®*á{øûðoéI~ø7d;zÈ ¾p9!ñ£ušçQH|•3%ƒ I)+ï›B²R-»_·â%ë6áaJ¤Š¹>å×¥†Ôi3S/)ëvXüwªàÒ½àñþÇP„GÚÈßZ°!_¯á*52>ضò•\EfJsÙo9È’JHÚ¼âŽl§ëî;ûÞ@£ú 'BwäòèŠx<™ÏF+·!K²˜ëTÈ¢ªµc{“ÝV"NF¡ªúM;ÆIÉ n ž¡A±6A†ű()¦Tàÿ_ýIÿ¹ó´8Ê>=’ -‰r'ï(Ù2È¡|&'»è+<þdyLÿe*ž£Ó~?ÍŽJÞï$/×H¢JÄ¿°3µ‰x²½b<몪ñ·GÞüê¾ÎoÏ jTè,‰eýF-‰ï½Ùã¨8«cþŽ u9 fÒtèQïÝó‰ÌŒh Oû„9аv9ÓB“/?¿bÀŒ¦ìW˜…ØýÑÌë™Dá¹h®ùx2¯"-„¨œœ“¡%õwíCŤh¹'=±æv­¶=o0Ü‹E’¦õò×hËÿ¸ Âqílè}<ŠüÑþmŽ—:’äÅ)¹S¡Óu‚cþþ;è¼6¾µâW†®nÌiþNÕèvîY¨bˆ †› ±Ð2ßZêŒÿÅAü±N|ÉPü…E¸¬ßJJ“?W]4…²œæ .¥m¨,z¿Ll µ~zv½¹ãÐ(°Aü‹š<½s»õ5´é 5ièAÛ™ˆ‹Õ—ö =Çò©ö‰ÃÐ1ý¾Û¹i:÷"ÒU•© kà¼ìjÉ=è~לR## =ÆY¯83 —óVQ'ôn­üžÓØ~æzm¯y`ú±ýŸ Z‡EðåÔÂÐ7'K¿¯0öÚfA‡F,èŽÿwF^MÖs‚1žWÓE4`\@Þø†L(•ẼIß™4Ö¡|˜ÊäÌJ93Ür7²¨aö“{ç¹z˜¿ÔäWàOãÓ[¿Z`éÄswCX>ÅjöÏ­ VÞÓû¿•f†ÕÔ#®Ý*°~>\¡x"6ÖíNиÀÖ›Õ»­Æ?`粈ø×øøë³û 1¦þÅ.0t$†!œ¶ÜîþŽHþ‘‡ÄóÔgL ÁT@.k>êÖF44ŸùCóOÉâ¿Ü8¡‚&¥Æ»âL"h¦N!Ë“-ƒæ|kÎЊ¡IÒsºG§Ñb¦-œ?èZ¶þV¶3íD«’Ä3iñÄh×ñÃý·ÚŸúœèn…¶ý·ñm¼Ú3yÏý@?æ[q èø·*‘]à0:ç‘]$¥zˆ®_õR\ðQ†þ•Åô–J4B5µÏ]vªKªfÃJPé~çÄãY&¨”;kÊz¾*¿î„ÑMA5%Þgf|µ÷¿Ù‡ù@üd‘m¥´±4Þx˜mBÅöWÿA¯^ñ%g.èÜ|lg|ºÛNç@oÁì€âoè·zþï¤ dzD Üs;ðLÓÿüW¸H­Z¾wK†\ê7ˆéÃPcw¹àuÆa–7A©0\’É,V #ÖÔ5Ûž0J|h‹ì³ŒVd›K¤ÙÀØ·LÍÙåz/óÝVºÈ|µ”¦T W“Êa:+Üÿöµ¿0kW7=ÉÈ óG/W&ßè…ÕÓ"çÙ²`Iuöõ{MX¾}2j—ŸVFÕbÞÓÜ‚µ³—ãç<Å`ƒÓ³ª}6WCâêzõ`{‰ñæ¼Û2üe¢Ÿ}´‹ðïôcúÇ‘àCÿ›5HXþ“Ö-¥‰) CM¤fÂ,ë’…tU°üƒŽ9j䥈T–|W<2®!µ¯’ü7Áe¤}ÏOû¤ énÀ÷ëgðHv^”ôÉÛÈ@T 64H„Œ>‡´—J‘i:FSÔFYî]³P8Ⱥ©'_¦G¾{Äʯ“KîªðØwä¹ázÓà y× :?j ¿›µÓãŽÿP`*€\€ð> ½h¹EJrEnæœæ#çE±ø›Ü+dQbŒ9ó‚ß}<êÅ:¼†eb¸m…dUQŽ"íëê­Ç(ŸÐ‘œ“:€Ç"º žú£"§ó¼ˆç2*ѦڢŠÄPñ¯”÷xòJl¢ÿæâÈ·ŠëïP]8E£ÆS7îñ.QÔÜžßôCm_å 5Ѿ¨3k%ùœÙ õü.Ÿ©¤ÿ…äÊN]‚Óhøþ »9›–?)DSÒ—l¬Lhæá¢Љæ:^Vÿ™™£…´‹zþª'Zò0Êj{'¡+J—ÍAk.þPµŸ<” »^Ô6„¶V ¢"Žõh½äÝÞè‹ö}¶2ß}gÑŸP”ýD§f1¶'¶ÐE6^å¤2%º¾íÉ#Ò@·ÇíB<ðÁËkÉÿÖÑÚ-—RÇ)%Nj܎Òz!ù›,5{¥e˜þ»U<ï =fµ†õ˜§ä™Ö^hPIôg„VjÂÊPñâìɦ1(ÝŸ‹ú%Ò1¾Ö¼„¢±ôá…?ÆPt~Ÿ?^[ŠêrßíAñô†Å£¡£PZ”ýPxš÷àú‹[Ž»ù÷c÷Ðwƒúø>ðÐî…¦OuË!ôÌÐB³»Ðsã´:§k7Je@Û³6ÎWí\ô×™Œ€Ž¼¬EñéÐù°8 c€º ô?©Õ‡®EMÿë¬ß ûSkÊÛÍOÐUx§ËÍzßäm5¾ì…¾×"Q½¦Ðßœ–p}=…ýÁ^~ ¾}­gÑü€‹¹7%x`äÔ`Cyy.ŒêSŒh„1݇¼ÿNdÁ¸S÷÷ˆ q˜ˆ.‘ á‡ÉÆ JLy¿aZ)äkKà,ÌTˆ~43ùsa,d¯¥Ò`~áeëœ,ÞxÿGü5,Fб¾€Þ[½‹“^°z¹ën¿š!¬ ì'˜5†ó Ç·ÿåÃæþmáÇqî°]¡"ÒùEv³ëŽ{¨Â^»²Ø‚<3Ž};Ù‰„f­¡J»×èóh’º÷+$‘l%ã¿Wˆ¤×µJÒ‚‘üiþÆ©k¿‘òÞ=²)á¡$ÝBï:¤ùsùe`ü ¤S°?û׉ Äžìp¨Gfê¤áÌÈX;ò’æ026Èp"ëR$ËÕ:ä8ó9 u5‘["œâ}œ ò&M<¹âȃü6Œ+[¿PPyòð(Á36ί+9‡¢±ìåS:¹(qÈìù>J˜©*¢Œ¥¯y¡»/ÊNÔnŒ®¼qMÊcYøÐÌ(æšbšŸp£Ý‘E ¾WF hÉNì»/ê†VGÎ7R]}ƒÖ´Í–¹hÃ(RçL–†¶B‰e‹qTh§¿µJÛ´öá&‹ íGÐaúÌÛâkèt9˜?xÅ]Äjª:Ñê-‘øG|Xç_õCÜvˆÊÛïJ£LµÈ£Ø‡H5SÑŸ@³·þ|vÌ€þ»êËwË¡ÝþaRàkh¶=oýõáh°TÖu¨5ñ:Æ¢zÀ9“ÜÕ¯²Péíû¼ÐáTx^;|õj=T¨h£†ŠÒ×祯òA=…×Ô<èë7±Ž€úÉç+lÏ¡ùæëÀšh}åj–"yÚ#ÙÓfOÅA§ÅÛѼû¡Ð-xDòÎK3èÙ ßµô€¾›¡¤É¶Ð_õât¥†3 ¼j¿ßĨÄ$4ŸoÞƒAÍ zê Ÿ`ðZg› iùÊÐ÷Â0ó¥Ìº Nô›J«†#]ÙÁX¥$9+Ká£^ÊRª†±5Áâžl˜`h9ìÊW“žªëÛ“`ªóó‰h²n˜ ú#špCæLÈ ²À»åÝDŽ$XŒkíj< ˇŽg>w‡• b¢]ýþ°ºa`¹ôŒÖS^n{ÆÃf°ZÔ— jؾªAP|vÄ,Yóœ`o®1ö$yÈQl—Ç áý!‘ï?X‘h³éÜi ’DJö6ÎýC2Ís²{w‘‚Ëòm€?R1Uí™ßÍDjéðÛÕjÌHëÃfqóßK¤svþ·þ‡ÄŸ  v¤KÚ,¨ÞBF=®cÄÙ1ÈTÕdƽ‡,S޳±®íͤ#û†Š›Ê r1—KØ@KAù:äm] ]Y»„üpÁ_ÈŠŠÆSï{¡Ð¹N’o«{(¢·¸S!$‡böâ–5Í(‘°šìcˆG™xE}÷PÆýcÂyC”mk~<3òn>–"Å~©Á¹ë ŠpüŒveÞ:§Ðqs—÷§:^÷¹•%ì‰NKoæ•þn¢ËÝ›á¦è:ö‚}/µÝ²Jš‚B"1l¢!·Ìß ­æZ¾—F”ÐàÈ¢?-†dkT%0åÓP|¹PºVÌœ¸æÙ Å¤jïÃ;¨+ݺ@ýïTõj¤Éœ©…ò˜¤UI¥z(Ù‹øÂ‘ÅŽåOœÎk@¡ë©a™Œb(˜Éãõ×ÕƒBU6bc=(>¬·icJ%ÕÏ߯ëA¹Ÿò~æƒ|þ¤™éé‡:²waÅÍ ¾zi”äÖ24³}ÙûÞ-æum³Ð`ña©ç´ÝÓüáqÚªh¬@0äÞJƒŽåÒ¯?%ÏCgL»qWGti…7zKß…nîH­²›MÐ#žšÕÐË/æþ[Þú k¿¶ †þW%LÕâa`®T³Ëâ9 –DÞ¸eÃOø²‹¬`Ä_R¦lÞFÏM®Ý•p€±Û_’#|ù`<ãyæœÛ ˜˜/*º3FSV­c*U0=¸|¬å6̾ry¿çâó´å"ìþ°v^1öÈXòœ-ák> Ë sêøVfÈWìÒaÍä*aŸ—¬WôZiò3æ UQ l³™ îUôÁ.ÁË‘,ù$Øc S#«N}kV¾dg$˜ mV?Dªôßþé!qöÒ±€D$5ª|é)Šä‡5¹êÆŸ!ÅVÛÅ*[B¡±ÈÝÒùyå¬WzþÅ#o¾*ò*A)YŠoõSú(C>ò&ùÇ<ÊÆ™UvýCyµ“/OßKÀã´M ‚¨H˜Ò´¯Ò‚Jì"  ¿QÅæa²ç-<™K|,V¼ÕxÍfööPýÚ§Ëç3iPcÂà9ïeÔ:Ícº.G‚Ú›çO‡fÐhòÖå#…Ѥ²Sž#ÍÍlÇfšóûÐܤ&#{R -ô““ô¨(ÐÒ@Þ¼¯K­ÌxÜ'AkÇí#çÑæBЈùQ´ìêU¯,B»ŸWR‚)³Ð~÷øêâÑ èèüû¡å+tš”I*‰D——ÑÁ‘o2Ð þ”» bôìÎëéÌètF5îûVJÇýû d㎔¦±DwÖÙ`:gî¿L¹Vèý“dŸa©mì2ZŠ ŒÐ$nr‹dªêöZ•MŸxø$»Ðiã«P¹ÐÚ“õ‡*\†‚Y¡Üûâ¤+i”kñ\Õö1‚òû¡î¨dŸrñ‚êÈjrMÁt¨›”¼í;M+æ¹ãT¥ÐJ÷O·}„ÚVçŸxž¥†ŽÏÙó ¢ ËÏš#JfzäEØ.^„Þ”óþBfþÐ7ØŒg¡ÿ­½¿"Ã0ôo×Ù8WÁ€èÞQ¦:^H›9“û)é xä¼ap†×#¾ï6 ­\SùįÃÏBÎå\%=à¢U¨‚;ŒŠÝãë(‡1qÞã´A30®qŠBçî­ÿìgK¡¼“;-ŸÇ}‡éW´>£—aö|‚je½ÌÜjjàŠ†…¯ Çã-»`é?‰ó^´_ayçHÉ¡Og`5D€n½bÖeæØÄפa“ä_È6}Øúw1Ô˜'vð×»«A°w…öÈìýmØŸú{Ô‚è ê’«“•¹"Q®hcd¸’ø{®æ_E2â‰íLH>¸ÔÈ$ú)»/„«ÙLá¡•®º!G¤•¤KžÞx%lšã=$¼Qª\H×üÒÊÈU<¯·%­õQ^`'k1ŕ丗‚ ·¶•$]Þà ÷×/„.Få‹ïŽ»º|GÕ/’´ÛÛRˆle"ê#¨Öñ¯èØû£¨!ìÈKˆšñ®Û5Ä1¨MÑ[”箈:á^YWsöPI€Š•úê}²fwO m¯TÅ%.£ñ‘†KícthÒ7ËCÀn‹föKVãšãhnÄ cqº -´È¨â˜ÑR³“27§ ­LœËNCkÏ×O_qððÁñ¯w´ÍÓý;çøív2¼rçÐÁäyÅ#ÂTtü« GÐYbY³x]Ò·ŠZ+Ÿ¡ýgq>átë´]’…a²ü Ö|Ð*ðaŠ%]#Þø5‡¤¤Ç;,ôŸÃDû»¢M~è|Úéx<šÏH4Q[;ðPŸXû¡2áÚŸžÏvPö #-<æ”°2\z-E‡—vèùÚáwçÐ_yø}ÚÛBß~gÐ>êÿ/ Ï)žP¹ÙÅ-Áß+µ.B™19ÉN8Tº§M_>k5gšx] ^Š©5l® ÊÙúz Y{n¹'•Z¬Z¥uÒ] Õ‚A`u¢Úd§Ö+Mg¡mÁk‡QÚ_¨d¹Ó"ÇcAÇlHzÛ@ç7Н¯” +û™ëW¢Ð"¡˜x÷:ôTΗIùYAÝŸËÁýûIiÛ ÂÀ0=»jJÀA_¬Qô«í‡á×þVƒ&ã0£VMÂM£ïî=d±úKÇ­ÞRÂÄ¡#ÚóÔö0yI€%à=Lm嘲YÑÃLfЦÓcF˜“Š¿uC¿þ<9~wûç:,Ö¥®xÓ]ƒe6ÛÔ‡Ö¤°âv£ƒî¬.xû¤Üõ£FãW`ãwü%*¾<Ø:Š_pàì}ôž”þÚ´Úð¸(À¿(ùÁý ÃHd¨á¶‰„µåqH|’Ö£ú‡3’´iëùñ´!Ydö1³óHá›Ó`*€T+£…*Oúû¹üýZx˜59±™é:ŒéåË_"½&S·«¤12Œ±¨j3]D¦rïñÛÇveÚTjQä ²›6‡³þDνÐr9¾~äÙÑþη‘/EUDÈJ®:»°†BAŽ2ïX}Q$õfæ¡Ù§L7°­AÉ4§ôcQúÚç!zÆ”ÕèŒoä>rÓמÌÜeÇc™–c:¨¨±Ö|ÝO|ù›z«u•ÇéÎÅæàIèe >ÚŒXz©¸¢BÕOŒ}o†êF$R¸QK”mlÕ»µ ?ÑŠ/Ç£®O‡µÛ ê E¼¬åïBƒÕma›ær4ê¼báL8Œ&5즌WØÐì”y¥È ùéc2­êhá+ôn¬ª-¯ŸRÈ cF«GÍëªäÑ:mï$O?ÚT×mZ¾ÊFÛ¥#KO·Ð^pÞ~Fó4:\àÜR¯EǶÃA#2èì0!¥S®‡LCÏýF7?¥ë=¯1š²î²šX:vz>½t¥^|‘Š¥Eò.òçL`raæî¾ÚôØÿ-i¬Vv²fhøPj–ž}jÓÎJ©íG@Õ$õx|hT´Ûì\ï„rëÑë‰rPvÖ»@çé (3Ú9ä8eÃL!šO¡B ˜ö¶L;T=Ì–S\­ƒÚ1I#ábuhœñ¨ì ¹-a‰H/mÎ…Aƒ—B¡ƒ¢òoÞúSè,}Aõ¬ê 'ž¯PŒºüzšlGSï?Þ r¶C…öЗ2èñ\ð ô-Å&ÑSA?-“H<ôëÓý¥¶Ý…þ†<Zó5H`«çñ‡ÁøÔÇa«0TpÃóg ?~Ÿ Rí#§/Ò×-Q7ŽlìÜ0viPjDÆÜøp¬`’{£øÿc˜ú-F©ÿfÊ^íö…¹ØCÉ{<×áOm†UÏw ð»pÜû Ë÷wìñº ¬Òì6ìläÀÚ/E?§ÐjØxø;È/·¶î'’îÝê‚tAÉãñðwËøÉsØ?7éӬЄÒö-)Ht-DVœªI¨~-²†/!iy,•\p’¿yÄÄÑ7ˆ”¯˜¤%¹ñPŽËF Ò,‘O[uÒñ%<ûSéƒGì›ô*Ó þÃé§A¯Ç‘‘Bá•êQd 3>K­yY8ƒ“!k=w`À­ndOJ'ï tBηÖÿ~ L"w?Ýé7 FÈûH}0̼ùvÞ$›™Ë À¥â×û¢(txˆþP# ¦q·AÑápuÂ6Z” ûáDn‡R~ׯû7£ I±H±âY”=ËÏP]œ‚rS•£…Þá±`ól7ÕPAFgèKßUÙòŒð¤¼Ú~•Yš‚ÒÛ‡œÇ'·¡X.Q^`  |ìZ΄_6ƒÙ)?"Å)ch¿8‰··Fà÷Ïþ»µÎÿAQvÉ#‰ I(n¡µZ„ 2¿Ò­Žƒ\˲ܜä…Úóùe”viPï!®¸dþl «.Aó1óÜ„ÐÂ3{Cã:´,{êgDBëë|µnéhSËÎý š mmOM‹úã:×r’$thñÕ¿ÊM†NeàQH†.q)š[M¡[ƒ¶3,Kz„]  †ÞZÝžV6Áƒüz|ÜkŒ.SRŸ¹XzrmyÙc†³_¾z@—#¡áó40:¯›W0ãbõ¶ÇÓ`"ÊÏ0£"¦èoòíu¾‡éŠ?#¿ŠjaÖXðDS ̿ګww9 ×="ï°ÃÒ³rš~j+Xîòi3¿vV¥OׯOvÃZ¼“v˜B#lpð3VùxÁfÉUÃuØ~®¬ò¬Ó vŸsû*®VÂ^‰â‡“$ yðöo}Ú]霒HF¢ É' Ÿ‘Ä:½Ö°ï’‘Yú&üÖFòá—â]O‘r´:¶PÈ©™}ê´´j‘öBá¯â ¤ , ݾ‹Gz^ú?¾åƒ >`‘„LºƒvŠ>Èâ)›'ÞqÙJGöÎ^BN¯»o·¦“‘Ç6dëœ,/ò™˜‰Þý‹ìX®Þ:ȧÄ黚ó("'-ôZõŠ…wn- j¡$=cåÉ xtäypAÊ,;y•›Çc2Û>¨pbˆímî žÐlUy×-‡Ê—Ÿ*ž¢X@ÕŠÌå0¹aD-üü8#Õ¶Ô,ŸÇ¢†—ñO3‰:Ôœc¼ºœxµïÏæ»©º ‚I b¨·Sz˜Ô† ÚÓÆÙJ’ШÒK>M›Mê\ÎѨ¢Ù…¢wwKÂÐüÆël‚B}´x”šrû~:Z¦?åxvl­B"xú¨õ×QÝè.´à¦1™TF;k#²IIq´õ¢±¹ÉŠ#ÄŸŽ £“ž‡ç3jtnz‘õÁ ]ƒé#éêÐ-õ¾,Ç7VŒݨ$ý莺™Ç®LE‰×Î?„ò‘ŒCÜÓ:U&„§|MLy «ÒAlyÃZؾùíC=Ÿ‚€Ùx¨±$â•;ÿ*¿ï'WåÍCy^‡¬ÏÇþí]²+4„Òó®qGy Ôº=éŠG-”.4WH|¬‡rÉgOÞ)«AeÌgÉÒy¨–*aˆ‡†a¿öæ>hþ«ª g¨ ­U1v;Ð~ª'"_êtéß~ãÂ]i§¬Sý­ {KÃ\,ÿ ôÒœ\ùl ½§Ì¶»Bï”ýTÁ è]•ò½G }œA´_6úbÝ ö.ß~SòÓ„§ÎÀ€rç˜GY. š.Öß}CfΆ˜uÁð[Û-Ÿ¼‰gÞÈ·€Ñ¤°&µ äµó“ MŽ{ß2ÏÁä€ÙTQfLgTfÍr6Ãì—çפø¼a~¢Â”ÁU ]äßnÄÂò‘Ñe?}+XyÑE kÚ¡ Ü`ƒùú%;CØ¢® ÿ<(;²¾Ñ?V³áï­‡G´C¼àßbs²Çz ÔØúe ýA¢}Okª qqÐIÛ3Hz_(0FUÉ=$œø3"¥ó¯×Ü„6x((ГæIÒ|ß—WÓ>‚tÄ¿FÎ|§Ç#*wå’냑>LÏÝ6¢¦-‰½ö÷étôU¢©QdÞ±ŠìEÖßå~…^Dvw— /‘óÒz1e!r§,’Ù’,#ï¹™¨f‚äëÎËŸhÊE}iáÚm*{ž[ú…Ó~}»}E“GŒìº2P¼62ØzL¥øe?'_ËCéŒøÞØ+«({Ü ò´$Ê}â)}?ŽÇf>> èÁã#qa²v&¨˜µ­;JŒJß=Ó‰™KPe¦âέWóöJ}T»\t˘·ÕëÊ{Z|úPS®õ¯÷èÔúpƒî£ê°µ}…º9íégIPßv&ªgì1RùÝ-î}‚Fm„̯ÝÑ$KYpêâ 43¹d‘“ææÅÔ47lÐÂîs—̱´ôùoT’Œ ­îÛ±Û¡u{òzè‡Ü£séÆÐŽ«´å«£9ÚûøœN¤ÒG‡’^ŸøSè$¾0èmrß=ÙÕ[EWAG-K~t³ºpêt= ºsÝSï%Ä{Uõ›tk/Ðò!iAÊ ;Ȥ¹„DgòÎÚ«Áh†Q᬴uŽ>®.†•É1çO[Pe7fÐaüÊy'‚$BI :ƒúÉ(|i>{пsãz23Cˤеko}Óo z”"ß7ŸƒÞ¬ïJ…Ð/ÛÞæB­“”ÊÂ"0ômz‘ì OËø…gÀ(5ˆðMØn¶E©Ë&Œ§”Ý~é “’šk£ö 0ÕwüûÒ]˜ñý¶*æs)•Ì5$Cð§É¨õób!,ÑËø‹—½‡e“š ÃQ°’4 *" «ÿA¤DÖƒ>çÝûJ›LQœ~ÔOakœrúAj-ìôŒ[ø»©âûKÐö5É…g‘`òÝŒ_©]˜ØCân:#ÁÿÄ‘ôn_©(’[Ý w•ý†”&¯5 Ûäöt_x»Ò”ÇN½ ˆt¬’û±G<ö]Qs(Òïj·,q5!cõ¤¬R23jy… ÛñÕk³âÈÑtûvÑ׳È]ŒÏIŠ‘wÃSÜöëoäÏ¿T±»ˆ‚ïw_º]<¿‡_É<ÊF1 âÔðP2”xÈ\©1}„²§zð0iT!ÿÖ(ÊX¸³|å×hªGàñöpe3»S›*|QiŸ™0趪=h&PXD¨¤}iÿ´Õr©Ê2£;PC¸é¡ÿ·k¨ùåð¡Å6%Ô6÷Ì XC]b¿Ù`jmÔkÚ U‹°Gƒvoû³®£QŽãj¸)š4¼{@Fê‰f×¥nõjõ¡y,¥bJº9Z|Kçм–½e—ÇКúgsÚhžä¹h¶7/ È¡]~× —!´ß*.Ë9è…–s“'CÐ©à‡™cáºh™œ ÖD×ùŽ˜6Â^tëkž Š‘züìubüè3KbáÚbÉ4'…)½äÎÆ35s™°ItÖN¹Ÿ±ŒUÐÌ–íCÐìµ ×7Œ  jï%ƒ@_;TÜ]ñØZ†²$W•¥”`(Õ’ÊU: %—lÏ1}y%þ>L %{ÑŠtdPvüc…æ¨xÎKÕûŽ^³Pßuo´êr 4_î“r'°‚VYƒÖ›W ­pž9ÿô{èpçN_ËÎ1B¿²«tÐÍ'Yôi±zýëÍ•­¡çÛ4:èpCÏ ŸÌùëO §[;Dvø.ô,Žð™Ç‡@¯-å“ù'4ÐG¸3©ê}CdŠwúG}ä‚ }aP2"á!Ý( å‡ÖÑÙúÁp=½ûQ‡0ÒȺò÷Í #Ë/M£‚qo5½BЉWDYæaªÞñÏn ÌÔ 9 XÁRòzˆQw>ÂCÂ&˜ñ i®nÜ’ø6އërìë0pÉsÞYREzo éãþ-ÈPí¹*WVLZu¡ÛÑæÈÜUßl(„¬·u ÃÉ]ŸõÇÓÌäT{¦d°ø ¹/7¤Òœ†¼úO”ؾª"_Nkë 5àKü’½1‡‚ß§yNgI¢p@§¹uâ õV¶ýÚoˆâ÷¼/š)£d‹âp}p2J›~>ýþ*=ÊøþçŠr²ô+…Q~ùîk Wðxì¨ä÷aTt|EÚÃ~•¬µÇ­VªQåA¬’úõm<ùg@i~ÅÕ0i Ç•Õ_ RD?ÔEMBê›?-N¢–_¯j¯‡ÜЙAÝŸ¯?%"B}I&Û—O•Ð`lu°sk²{õƒ©Ð$ÑC¤]ÍŒ oQ.Ñ¡¹U|Ð Zœ"üýð>?Z^ßj<ýù Z%E½iù0ƒÖMÝü9±h{ˆÕ\WíÌT8Ê¡}’ÎéÃljè°{C°c¸NWZ&x§óÀ¨cõ2ººS‡}góA·S îvtWg÷Þ Ã{ ü‡:ü³Ð’õäeŽ  ü½\,‡„»ÇªFa„•¯ð‰´^óI¼Mou—¾„H…CŰlÜy (ã(ì¸F Åç«ã„áw“û% Èg½ÃµY ¹&ºŠéQð³û-˼© üÜ]h– ¼yæ"ŒÓ\Œð«Û¼ïlbÍAOlT&Y¾¥f”×(8 ¼¢(My÷=T.’¼§•sê˜ 6©¨ÕJ˜êÈ—„ºÉ·ñ1JÐp•hˆa3ë£CÞÔÅCÓ¿’˜¸†Qh~œRòTð/´°°|7©©…–T÷óÁ´]Ðz&ÏÊoõ´Y¸í¿¾ír2yêçò¡ã„Åž’ô t^ŠZŒîõ8ðÞ†‹õCYÐÃI\s‡»z£ÞZJ@¿ò‹‡?40°“pŸmn†‘ ?·#¥ £^¤=Rá:0VÊð=ÎT&t¨ª*ûj`ré&÷u>E˜ŽÆÏW»òaöÛ#w™OymáÑhquƒE.¥ì•õIXª¨É6Ð{+tVÏé¼#aÕkP¡Æ0 Öº•\Ô× aãLÛKAaØ’îy$;ì9ÏÏpOÁ_)”¾#ÿ| J-2Ï ë_»=$ìðùSòωö,:‘dïÔ¯óO7¬ÚmbJ± )òýV ¶•‘ªßc®áT'Ò°G×<å(ÅÃ~Jó!•oðYrÆÒ?pæ¬x»Œ&õSTª:Èì¸ÑôNö²~$Y“kGŽ7Ãe‘ûÈ^Xy·;ò¾¾&±4Œü–~!—/  r]¾\@aÝXºÒáï(ú@Ɔw¤ %9®=E©ïçs¯îE²Ì¯Š(;"v- ðÊ?›mÐú„Ç/±œÒR`DÅó7OÍŽ¦¢RôóºÇëN¨Òæòåºò?Ý7„QíjDZËwYP}ÊüÔ E!Ô¼L™´@±ÚŒgÏǽGæ²ã”Éî¨÷AnóÆï—hOáš”lF)ÆP†&M“ïþ5Q¡ÙýWŽrÕ2hžæ÷Ë(Þ-Ú>}þIh„VT+&±ihm0‘yž MŒ3ÝêhÛUª¤}CíyŠå#Ÿ[*˜¹MkúÄÔÝίUT7 »L_~íx%ôp|ëîË| =Ù Þçæ¬ 7ôxûiQèóL‘dR„þïbÿ5ÏàœíS_&¢©8±ó†ÿ¦ÏÁ¨´ƒ>UÿAOŒ:M][Ï1g¯ÃŸG´nW›*aI,›¯w aù[)U„¬:pÞ]–‹…u‘ ŽÿŒ`“CèmBlã çš‹ °EM÷æV+ìm~`õ›ž@ÃÔ‹¿×‘0ï‘=E¬,Æ> [sBR2#SÃ_žH6qzVzýRŒQ|ßýT‡ÿªT5#ÊŸî<üxm©ŒµéþM¹0ì"½Íƒk2¤Žþ{ð™©b–dŠ‘ù‡gû%dµëÿ¤ýÙ¹š©¯˜°"'Ý±ÎØ_äVÍÐ-ëTB^‘“§œ‘/nn–®¾ùw¨^¯Õ£àýùú·¹(¬~xrû×CzŠlxÅAËCÿ•JÞ†õ;Ñxt¥l)ç†,ÊÔ½ºúšè7ÊÉiµ¸'û |[l€÷!<®%ݦ`þI´nŸµ°Ç›"Ü)ͨ"ä²B¼‹'#&f›_ù¢Í“§/ º÷¦¢î·lÔhÒÞ'+©C- ÷T ¨]U¨°uuí ~ýKG}uGQ4(ðWùÃ"‹FÏŸØÙG“ðHñÆò>4ÓgŒ*óGs[ß©ê˜n´8oýv¶ãÀGÚç'Y+£UþËÙ·óÇz5>Ût怇 MLÙÒ§Ñ.ädŠŸ#Úw zÞUDÇ㋚¼èô&±z˜¡]Øä~¦6/¡kbAõÞÐt‹—NH¿üÝ=3q‡šc¨ËUÞkµh‘lÇdDŒ3†áÌk°Ÿô/¼¥†Èÿ³yþ†¶…Ï]޽Pó!PôÝW6(÷¾=Hâ %,Hnýz …5v»«A½ÏýLN8ä^ޱ½ŠÌSKÐtæäX92Õš›Bιqãâßð3±Þ%ðo+äͲQî°@§Þç%=(^k·xœ¡à­aK/ÂNC…ôǸ˜¤ywŠ•JR…jÉÁ§#òÏñæ£JïÔÑ¿ýÍŸÙõZu®©÷: ÁÕ‡•œ” /`4£@Óes6{š×ЬÆòL]'šÿÞgu;ðΖ~#©‹­ç 5Gî1^´} ‹½í… jÐÉJ§\«±]÷­ìÕuÆ¡{rø¢žqô>«}œ%ýšf’<-ƒŽ¹©t¶Ã0lب³ý®FîÙº>ÿ£§¦ÂÙ`ܵåå'G˜$-ÿÝcþ¦>5xד‚™ß6Ÿž›WÃ\ërÖZ1,0Óßñ¼Kª;‚•™é°|§n/1uVšx¤ÏvŒÂšBѾêYX/øÎ\¢’ ›7>jœ3ˆm'¢/w³¯Âî)À¦NØ‹ º{[e ˆ‰i£Ó‘Ðù'sõ÷™vÒÞE’ks£ûH¦ª¥yñÿý.¬ÿ®`Œ ©Ðõï‘UG¤¾t_êVÒæ°õŒÈ\@ºDQ}^ÿ|¤gð÷~%, ­šScð™ºz†î +'óÅø_y}{CÅXu ¹.*7,g!¯N´Tä¡oÝQà£XðBæîÊ (Ôªf-—…¢dÎ$yj(À+¦ ïQŠÓÎF5·¥·û>±¡l„çÅm?”W¬ª/R©Çㇾ‡3ù¿GE"‹-Ú©QTbçx–r.UŽG¯! ž¬ÉðûãqÕ“>È8 zj›RÁïÔ¯Çëñ|žûœën§‚Î[žºfaè:;8/¥ ¿DœûVå¡›â8{O? ô©#J^…Þd‰û/xœ¡ÏƒÉ<@~$TFê‰`Ðz£m„±†¾>nî<ñFÔ²‡ÆƨýyXa‚–‹EÁ”B¾§ñ¼Ì|ØV*"ä΋G®s]: ªåôܰ8r®—¿\–Ó²,w~Àj‚CÉ6¬W­¶8D [ŒÁF÷Â_ï óíÓ׈¦é["Uß/Lbþ‚{¨»ÙRÊÛ‘´º¿ðÓí5$ÿ¸GF!) )?|…©ÛHUá¬k>´Š4»£Ì$O‘Þòp@Á"?2 „ÏU,D&ÑùºýbÜØRgpYöèÅŒµK!ë«Âó¢†E¸_öÊÙç/å‘}1àŒé´rô·]ë.ÔDnŠ'·¡yIB¾Ð$ ß52âD3äo¹]¸˜Š‚Ö ¿½<†Â4zwb~½E‘?3—íG£Ø:ßñc78ñ ”¦·Ÿ ëö®¹oŠÒá³Ê— Ü­tÚìð'ãa×tÿÉSïQnåî­¼Æ6Tøüg}ë‰*½¦´YOîDå ‰äê;QUh6:K31§ìÈp®7ª+0¸’tð¢FÌýV2¶¨Épñ Ê,j…„ ×ø£ŽØ©#[f¨ÛviŽcñ!.¿Þ#^C¿n)Ú¡ñµ Ér÷;hªùûõÍÓ]hf{ìx"Çk4¿ñ)@W¨ -â »¶Ûв=YΣ˜­Ø,‹îr|xúQ·,$¢M–7C»í ÚÒ’œYR”B»ËEùiž÷ñd'pz]Eˇûž„Îâ©nwjÝËè˜ËpİÒµ³ðáMôýÔ ïp}Í]ýÑÉŽ|Wµb.¹åÀ_¢¿!>ƒ^ÐÓá-x°š¼&ïœÒo&äGÕz-éd(Ÿm‰yÓLÈ›ú"»üP”[éúÙ7òÞö¤r†œÚõCÍþ®­•ßS(æYÍ,û’,L!«çWâjJÈžº’DÖLè‹C¢Ÿ¬\ cÓÚj޾ÛC·ž“ýÉJcO 6 ò ÍÅËÇ8QÕÀÛ1W¯q¶Mb¬U ÕϧҸ‡«;ß§®»£f¤«Ûç=¨}Ñ©Nðg$êš¼ë_àD}ë³=BhÐÐ_>¯ÝlÛ.wÏŸ¶BÓ77ö 7ö Y»nyäåóh±ÿkÖì9´¼hþ•bþ¯Ñíœ|!ˆÖâoǦ Í3WS<1%çxhít {¨Ï ãÉÏoiF Wê>gjo<•p±FúÍ6:Æ™?O÷C§›YT!ô$ðá=;GÒ!÷ŸùUmÏÄÇ¡¶(&¹«~|ðêFh¾ÜEªÜ?›Š i‹ Õx&».ï!´½êid»¯íÑ3Ô²AÐqýÄ‘ëÆÐQ~ƒêîM»×€çF+Cÿ^ÿ@âš?—ùÏ<ÎúPÙò×ÐõºþùÓ¤FøUzAö‹1ôðížø±Çz{÷xãe}™1….÷ÏÂÀƒ×ŽÅÚÑ0Øš:<þ †ÚÇ^ƒQ Ž8éâ*ç­¯ù#Æ“fÙ¾´ž 0]<%`”´³©Þ?¢²„ùZÅûÌSë°x9[ÁÒ¬–\5è0†U±!îvÖ°nT}¨wF6#žÒoy#lïÞzî×K »Ñs鋲 H¬¬Î­!Š$ |ñÍ/‘4àš(…òS$?÷ö¦ªÝ$Rž;^ûËJ©ü3CÏïp ME¯ÃŸu;¤ß•;£Þ Êu»÷ÛÝF¦½­‘Mò™¸ï\Ty8½2&ÿ¾‹¬²Ù!J¸Ÿ†\ 5¬Ù¿CT`r|½ÎÈÏ\-çKôEž*Ÿ?^ŸÏŒ›ýhí/äOžÔ¾:„‚b+™—4¡PçC&õµ I׺•5%Šbé;ºª´(ѧ['ðRZ ï~èŽÒN3ο¼P&ÿü@Tü æÓó¶·QE¹ÄzÞ£)¨p<âê³ò¯¨¤<¢ï߇ʎëg·«P¥lÍ0âO¢¹Ïú¶ª“Š…o>G‹ùÄÞNⱾŒÀþHÔ:«ÉÙŽÚ›ŠÓ´k¨›YЛ €ú×WÙ-~¢¡õ8_ );tKçàBSµÄRoTÐÌž½âa¢-áö/}ð’D‹ ަ÷—Ñr>ÿžšn,ZeªL6!øð…JlšÑ8ÚŒåy+Úv í±3.ÚÂh—dïÇõí7Ú³fÚüâA‡ Vw7›?èH,\›Ã…Žóísl4è”ÒÛíøð-ú¬ðéó/o¡Yþó;r BÈw¬/>p6]|Do/pAW®úíCË„<š¿Cÿ{êþþ3gæ?\$Jýt‘j…ÜËÑ»Ô!›ß‘\ìG.d–]íÞž†Lƒ&Fÿãæÿ¬YЬ ¡EÆC›l¢›‘yMÁ.»F èN½jÔC(eañ±Šž„²´%ϰ›sÿÌ»ù÷U±!\kP­×ï¥&p¾Ëë±g­Yý3ª–F f~ü3ªn˜¢6âÒWB<¢æáöÏ\«¦;—<ëR–ÿ™wõÞÿ™‡c½r³"‚ãÞ;•ôm7TÌ «-÷¶Dt/t3œ¥·Î‡ß¬›q‘:&ÿÌ«úÏ\ª¡Sj×Ï<€i‹ˆÌ0:5Ñ{Ý%ÆŽ«Í€ÉŽð£©z0CQy ^½f­r¨ï‡Á¼SAMžh(,ÄÅó~™Š‚Å¥^Ã[µå°ìÒÃH¡V÷é¾1þc k}QÝý°Ñ¤(ë­ [£v»ÌG`ªÏ*ïE"·Š›…:*H<˜hð¨ ÷¸oË$vÞC2AªûúbHþw…¤8g÷’ÙrF3EjNÏ+A'¤=ú++–”éCÏì!‘BF‰„¿Ú‡‚é—Ìdرdnнrüè3d#9âêðœ Ù¯y­v×ù#§XöI‹cÈ#yÎéæÛÿΓÚVqûÍ厂ã‰Wnu¤¡iŽÓä ŠÙ‹=“­A‰YÉ—-TÝx(sñð7”þÕ}ŠøÊzûÔçžc@9韹 Õ¨ ±ÃkµhJò|b­6{Pùì°Þ„ŽªüÐ7ò‘¡D¼|žö\a1ª‹vŒ‘×H Fþ†ÝD²j^cLUÌ@m¥tæŒsY¨+Øk)ieúJl‘žã„|*Å&Æ«“Ûhúqd^Å\ͦ3µÁö Z•É{¢\Š–Ñ¾!#qÊhE¾y1:{­oЩ;þË¡Ée/±l´Ë®~q3&íy˜¤—Ð!„h—ñ¶õÿrøy·t~ì$_ÔrA›Æ“Êô( bN·¤k;•>ï]™¡ÇRpçW´ ü¤5ªmºµ¯/¾÷¥+‚ªçAëóûL ÂçgزF6”ù •©} ߆CÙÞy^€âwD d>7¡H¯5¸CÓRh¤¾›&rP<€oI¥éJ…ÄP~пˆ~€Ð3=øzˆvïC­³»t«£ü0¸r$Üš ü yç|dd¤évÊA«D„èŠ7óÿÏ‹š¯`Xƒv¿“'h³Ç Ãü±r-ot$0”æ?™‚Ž`ç-®Ð¡—åRj/\þ5œÐaøq'Ù=:ŠÂÏ7_†Î Ní\ešÐ¥q…QH~0 e·9ÝYIv@ïeuÙJè“uœù² G u1²À` ÏAû7a˜^óžß)i‰iºÍH cå§V4ŸÀÄl«­I;L[•NÌúÂ,³úyíp'˜÷®ûþIX<8ÁùÒ˜–ÒL^œ „•^?ó™µ°ö·Šñî#?ØÔ÷pvîÚ„íü¼ÍÙMIØ=UõÞÞY‰y¸ü$mÕãzþ’ž¥ürùû$Ç5Ɖ¯ËH)/NwgÞ©ôïˆýy»ˆ4¾&O—e^ ]çäü+C+dø`C!W“‚Œó4$¸Ïbÿ„ÔÑÇ®~Òk½Ú¬§ÆIôhþ Ûδvêl3²­¾s¼Â‹‰½yÃ:ŽÈUs,ì>¡Gñ”~èNzýùtB¶D7Ÿ ,Íç›s](¸?tè@> •rJyh5 Hˆ¾­•ŒŠ=O<ï·Š™ÄÖ,Çïá!rßæGμ(­m1õŦebÍ$Üß*àa*Tßçxå"ÓÓ>ÜñBó¿¥ÞûˆQIí Æ‹·®¨|fÉ-¬+UrósÙÓ5»ó»S8Pmû‰v|vj¸¼Øûc{‡àà Ê^Íë¨u鯉èJêзÕv¸ì î÷ŠþP?â÷õ™þR4ô;øÑú« ?9X¶ïåy45KØØw@³‹¤IuÊ+hþü0ÝŠhZdgb¿–«¾B$ˆV ìÄ[ DðáÎ#ªéÇh3ºò'5Ïm5®›Í‘µ£]âÀãs-ОEôv¾à"Á‡ ¯É¦Õ úðÜPþö_>üGzñú’Øf/DV£Ù¥±‰þÞËÈýÙŸ¢ü*l4´ü"ר„.zÇ—trß þïíhŽ|¨òžÄ@{9(cÛH–¾`Å©•ô¼¹P@«Z¹}÷ ä z[Û‹@ÖÛ ó#ã‡!SV(H:2JoYD2@Fưž»Kd¾:+æøÂ²Ï­wJuí@Þ¾ã×ÒßzCá'ÚÅϹ÷ࡉeÙòB™”ŸQªŸ3”›´/ž<ú*œÏÒ…ÂRå½ßùU¿V¡êµõŒú Tßûýä¦e$¡7ºÜ8³– 5zs™ uP+uGõIA0Ô±P³þ.üõVÓ26“ðƒèÝÌÅéh ïåÛá„f…?ÓƒZ÷á§õÕ8þYh¹±óÚ å´6?z÷W½Ú鸤 œ³Jî#p£O>ÕÝ(b¢¼ö‰À«ºLGœôUÙQv3‡tíkRvÞ0$¦:ÿ*†KêÉöŷèǽ˜oõ{a܈g¥±ï Lj­š¦è:ÃôEïå¯DàOºñEƒœY˜§¡"-tôÄê;ãÍ`1Æ6_#!–é^²Ô÷ÖÀJ¶žé¬y^ýô˜T6¬Õ9;;YaëäÎÚs2üý cèÛý ‰Øïp½W{ĬµŸãž}ùÛc®YHZÕü‘úë=$£îÓÔ”1Ó¦9s H•:~Äð‡ ÒÔ¾_Øø‚ôoçäy·arx]èÏ[d Rd éKBæ;dj‰éJÈú¾8epé<²SqI,Œú!GúøJf‘'r'Æfe\¼‰¼sßéç·ÿ"ÿçßQ=-(øR6PÇz…os 6l è† ö£0k”xü$ÙDÒ Y:¦)1'¢tÚ§JÍ£(k[IÉPN”6eï&%*0Ä'}缃Jû 8¥+PùÄšSþ×TéX>æÑ|ã‹\l ªk—‡N¡Æèãi…¶bÔ|7,´éwµÆÉ{òéìCï>¢þK¾—Ž— ÐðG Í7J4°$l²hú}6×Ã]ÍÉæÍëƒ2ÐÂäë§¿§ÑòSsÞ]Y´¢?–®…Ö·K*níóú—CÃîpË´ËXÔúÓЃö\^oúdÐáÓ{-ÿ—ÃÏý—ý”0à«´”û^´ -ü¢ü1ºßÚôaGÐp½‘Úº£ž¶ÄŠCsÌæOÖÓ¡–ÈUÆÃ³ ªh<\¿~*ÈÕqþ”Ë­öN¢ ðMÄtqJ|šÐž»…·vŒ'Í ððy½ˆQ#(2jb½ì%éÉÂÌ9ìPVTäÎS.G¸ß~±ÎX6Bž¶Ü¯†åŽ¢È?aÐ,*s•Øhõÿç'&t¼MVJõ7¼ªyqußå„ÏÐ~©!q¢7 :”78m~¸C‡O²×JÌ-èp0Ÿ.¹½ ÔÝ+²•>ÐÞ¼Ìiõz ڗ£æ" ×­O¶É&Cç^©ç½ç ³óŒex|t-Ê­<:uºu¶.ÙCÏÔų5Uð;G’üZ¯?ôÏðµæò ªEE§é*Êdu÷Ø€‘Û<»Îf0vf™/z|&ž.Ⱦ’; Ssò~r{áχ›‡(ù/ü)EÍÛ}e°0ÍAÒ¨ –>ÍZ]¯€• @¦ó¡Ê°ö¥b™›{6þÏkTȆíkëçNì?»Ñá5áÚH¼gcÖGz IÂêqÿ'’jÇÊvXL 9›QJèž]¤$Þ±à}ò©è4¿]9V‡4jåy›J‘.©üÊd<+g&ÿ»¸¿d ™á>ÕÚš³=È\Ì~íKò+d5‰6ÞlE¶…æóOtÚ=#õÙË)È«åøîÏä*áŠoTôDž¢£‡ÏÅ"ìýÍ*¡‚ü/¥™îÍñ¡ …CÍ‚ %eƆŒ£ˆ'ᘺ™‰bçŸ|ŸéG‰€¼ÃÃü1(ÙÆ%­Ù‚Òâ…»Qæ¡’—f¬Ê.ɓמ‹B¹'‘êô_4PÁÔ™îü TÒº¼Ð’?‡Ê®³7² Qåˬp‡V*¢âáÛùü.O­?òèFSgWO0|XwJê“ jÝ8i{ÔÜu„®$Ŷ‘¢îØì@òEÔÿ¦µTz1 Ó®J²è}FãìûfW(³Ñôtœaµ¬šÝœ{ÁÐŒæ‘çÒ5{Т„ñE0Znë×W}@+õ##óÐ:¨ô¢Ëc ‡‡…Y–ÑVRífÕ#´ûpæEæˆ Ú3“j°{T£Ã‹×å±wÃ^Ý/øíôù°ãL’E2úÒ‡ò•ª£¯Úñ'ßs‘«oTÒmk6ä¿ ½gïüs=êÕ5æ…O<†Ê‰Œb”’Pšß1,Áå ÅTT»O‹h ÿºA^H}äXj:I Y<¯bõ‰R üèÁwódÈÜäín‚ ö5Ï"Èø“ëtÐѲ*uN±îêB®õŸ3LPЮÓÁÑv“Ð;ý+ ï\†ÒT¡ïA]PVóÕi…Êë/É¢%ôÄnÎ÷ëaË„ïÓ|úXõ¨ª¹²u4½ª“X¾÷Þá€ï…52VCPsö̶Ö57¨Õ¡s¤Š†º7×ø@ïzšâø‘®›qËô4j©m©JBs_Ùƒ'Ëðsã¬SKÔWhÕciR˜ym> Ã’«Jàz‚äÌá¹¼yÈ>t¤~E€¢kÄkèùPɼ–}”e¯Ã[ŒÃçØ`ðû䥲66Ò` 6ƒ‘UæSÔÒÎ0Vc“¢ÀíÙVt†îÃTc=!f>ƒ?Ôcõû/éÀÜ¥ Ÿ×•Ù0¿˜qƒØ†µ S{߆¥\Æ™°b­OthÖhôož‚õ~Åô»²K°Ùöjn¼ÃþR¿‘É÷æ€Ý'ç¸6ćðþÑcH’{£FhešgÃM"™[‘\ùÑZTR*ÜñkùáTê6wyGk)Ëc&H—Ð×sŽ’—È>g “–ŠÖ«[%È,z‰~]xYuÔ\N”áþ·N{‰|Cù@§Pró´SÌìõBÞ¯Þ50& ÿÉi#›(9T.‰cþ¥ŒÂºª¿~E¢hQ{ØJŽ—7˜£ärã5=‚Ÿ¥n…pº¢¬2]úëO(G+åw’ÄåW´dµPqÁë€ËÞ‡¨lH}ëéþ›¨2{ÄÑ_1Ã߈ÈÇÕSÕ©©ÛðãÛ~»¦¨ù;FYû'7j·Ë{ˆ¨[÷|0b~õ§$w¹~E#ÁqÖeÍ4q_µµ#CÓñõ{ïŸEs^ÉS6•Khá,}E™ -ó¸³Ì‰¡۳ɧ^oÐúniÂðè³9Ô×ÎM­D»4®¿§·yОÃUä‹:¼Œž4û_?ãäh_Æ€ÌUåÈÖýhqô\7ï=äßWâûÌQþúí7R ›ƒ_‹‰§3š¡Y†ÁòôJa_;–8ôAå“ôDí^Pþ$mrÆ;’À ³ÇÅÃPr‘¶QzÀ Š4ÈO×ßúÅ¥¤â䇡À¢o'â$ê¯3þ ë…₨ùtY(3\ä—z•It7SU¾BMæï[‚î*Ððæ źf4“ö‰~ö‡Ÿ7šÃœ/ðCËÈ ½‘´)?T8@ðMûq•åÊ=¡CàÜ߉[»Ðá¤))“F Õz¶Õ ½‘ØïƒD3¡G:Ý=˜íÑK‡žº?…¢ýVLŽÐññ/{¶B4tzI)qr‹C×ÓV ®†ð«{ãæ¼o5ô\yóÂaÊ~+^OÙ¢f~§×ކ,0°(lgFÚ CøRm}Ëa„/Aš&7ÆÈ–t΄=U[Èg˜ èÞæ¹ RkSïÀü>÷?.I°ðÉh’Zû ,>h5ùxV4žþKû köå%yª°‘v”V1Ô¶)¶W(<‚ñ+×8!ÑD‡UÌß~$¹ý–¶˜šI…ç©„#Ùì{-´¡BŠÆ–ˆJÁ\ÜÛBk¹<–€Ô»òcDï–ÎêW¨™1+2\ôrr{G‡Œß4•DòÞâ>Q–ÃN¡‘9Ù©‹¥‚YÑ$:+§ÙF²Ÿ¹¼íDö$WÏr äK¥t™F®ÌµTÎÝËÈ“ÛüŒ»ÿ5òÉÝÿ¶ÇîòûYÜô›=+­xCì) ½4>ß0D‚"Ö–‘ßÖQLsõ3í‘ï(qÒ!m®%ãÏô=DizþU[ë”9ÇSÆþ¦e{Åî{Ô Ü{?oõÑ¢‚QÝIę̈¤ŸÏÚz•ÏOú| ¦G•w“!+W>#Š=;ñšàÃ?Éçrè ¥ÜÙbþÌ~H'ŽZ>….F_Pç(]ÀY½¨GξôÒÄõÇR¯†¬Å aonFËül|@Œ¦×¢ã^Þ}‡fþÝ׬ÅÑü]¬Eè󴨟øÅ¸'“ì»hëƒVšFQ´A†hìkV*܈6#-1×|ÑßH‘<’C»wc!;«áhÏ4°÷VQ:<ßžQ¤Ÿ"px†¡E“ô¿|Ø*jxô+úrþxþbÀM[_—Ëg!çNv[›üXöRæc‚Žo±% YáPç7ù:=Â*§ Õf C©]§Ps¹˜Í=°v‚¼žŸ%Ø!ûûþëd¶Ë+1e{@2TÕ 2K!ýýqÝY®çîC¯þh2ô("×e¯¼6¶#ô r¢úƒXÈ>A±Oþ“…kP|NîCyl”üðw ÚÊÔ¥•Ϧ:@¹ì‹Z¢¨· –0[‚J%Ý.­H¨b"¿¸KÏ UÓÕû=€êª€ßfC3ð=Πè >ÔÜbL¿p"jk^©eô ünt¤;êsš›.BcBjòṓÐìÊà´ø5~Þn?qFZÊ>lzDCëÒ¸&…>´{æt°Û®A'ѹÉâ@Bž;ÎóJñ to5ZY˜ŒÀoC‘%EIèï ó´ÿ ƒ–kªó;0Ô“ÏI}ïŒx'óZ…}ƒ±c/c<`B0»þÉh˜’5áK—‡×ûâ£|oa¶Ìt#õž7Ì¿9Ê/AW óÖ¯Ú,­`éÔ_ÍÒXÞÐ/2©Õwü:òF°~ö6ÿ‹fØ4Ÿ‹”‡í«¬óïag°»qÆõµ­e{³¬#‰|ˆË#>$Ýk·.x[ɺº0?ëBŠ&Ú¼Ù2ÜÛZ[¦Ìs©G^úzzéÄ9¿GíBW—?£+qKýÚK‚G÷µµôë½y‰,“Õ ¥ûq¿âùp^Žûx ÜHÔïE7r½¦¶úxyí? ë"?­F1«M ŒègÙ>@¡)â°ŸZPT†½P(ÜÅK5{ô®†¢d˜Š‚*;J;èIé-¿DYöã‚Oêfñð¢™}*Õs”ÿ5‘ÿÑã;*6½}õeH•çåInT™Ò•2•nCl ipæBuß׳Ï%“ð˜JÊÞV±!Ôb³6ájCº”¬£Á~¨Çåšd*hpLÞ×ôÝÑjwà,C“¢HâÉÅÇhFI¢ÕõI Í•/\Ý<Œ7SžäDËj7âÔ>´âé}hµˆÖÊ}sþåP'£}ûŠÚ}ÑËê’pF{öãÇ SÑ!XႽˆÑÿrø9¶6Ý( òŠ´ÊŸ1¢µúUú–ä{¼\d3#ÛÃÃñ±D'á—ȉÛ"EÈm_O1ÎÁ÷F;£c+P¹W*j¯Âc(§ç ™ˆÜ ¥š’Ô|n¡8µîÓ°Y(|ùøÖ!ÏBW ÉX¹Õ23‚ƒ tùh¢m¡¨â.}Ãg(ý-ÒŽ~kP©IúݯGjìBc WÓ ASÕÂúÍhª`ùæÇ*?mO,Ž"pîëYþÚ¸lªª÷C»ê8qIB^ìê¿) j¿‡ÝÊ# }óPÃîÝgÐ,ÊÌÂíŠý×¾ùCû±Èjï¹{ÐÿËäç"+th×-hC'óG|T,t ìæÙ»¹À/·µ‡f_º×d­0"ôæ™ùù} S¦ƒT0¸¾X—/ClÁÚZ¡Oa¸VÐÑŠªF?rÞÖOì€ñêƒÃ¬‹k0ÅûàuÁ76˜ù>šûTù1Ì}I´ºÅ g4éjlÁÒ ¶J°¼ôòÌ.X£-Ë?egxÞ.JÀÖÄc’–é°“¢ã»y‰~d”Ô}G’S×{¿AR²ZÕnZHVµÙ7õ™)¢{Ï– ãÞ°ö©›o!uñ§}ä¿…ŽÕÌUºqŒ½x †Œ‰çË£^Ìá>¦›Íһȑ¬r·ÇY¥öL\xúÙ:£¾ªmô {,·Ã’xr•¤˜lÅí)˾Si…ŠòЬ <ðm‹?Y" K#û¯/æ@žÆÙ¸ò³Ê}d&ª÷D7drV)\8œé…½dÚ”œÎv0­ÐÑÒ&xÄÃ…~´n®$½™·G.ô@vßû7Õyj/Yøf“î2Ýò¯-£‚oVn)KF¾„>úÛÁíl”½öò:³låÁQÞ´‹PPf¬s~*í¢zË/9@•”×¹~cP½ÇbñêåU¨î<|«`T¾§0ï¤%ùCÍò×ék1'¡Î)A±Ñl êÇçD–_Áâýû.Å?ƒ¦îÿl^hÞVZø-Ê BÉTÐêlµÞo mý>5øA‡³q~dgté\¢(î¼ ÝºÌ/“X 7è…­ áüègb/úMèmSm¯\xÅùUZ$ÀMFˆq0~댖Ա’ÆÚÂx(Kà&kL†µ4&=´„é2¶'é0Ë2w¥BIæµ*~XV]‡…×í»1’2°D™)H,ËÑ—›<"aõÈ&+õUIX›ìþ>õ6* ú×NlÁVוmá"IØ9b¸¿ä„yŸ“ðšOEâ%‡Ì6¸'…ïmZÿ]$»!¼+s¯ )ÎÆþ¡ ½®Gɤk®"õ5òd*ßP¤M·Íü1«‹ ‚‰UâíȘ®ôu÷ùIfRE–gúw. »#[#}½¢Â< ¿lf¹˜ÔSOi]E^F±z_åYä+Ž:ØÆº‰Ï;¾öT¡ÐÓ*³Ú#%(òƒÙÓ3‚Åíâ—ªÃDQRþÓÆ¥Ÿæ(-¿¿Ž¦¡eÖ£#¼§‡ñpç5E"¦&”/y¦Á5ƒŠ_ê©V IQYúð'ÉÉ*TYZÂ7ÄÕwÏ"µ'Pýí£7ô&wñع£¯ù,=QËŒñtÿ…'¨s|JbêPÏÝIžú£Á¼]cí…F­ïC¾/Ç¢)é›.ž* 4¥¡¥©ÖFsË7¯I]G‹çb÷lŽè¡e[¡°²£•(ÏYoE=´~Xé”ÿ8ï_µ¸Wš·¢]²ç%#Û$´gÓRY~8ƒAºJõ9úÿËáç{7´«.a@±¯éŒZ5'åÔ“ž‘øªÐMvØ>–ºJ¶Y]n¥FÓuÎИꕽ®¤ßåwM¡$ìÿ!ƈ( Vɺ ß²ž<×{E‹~%P0HôòJ äk©_À5È ­;ž=Kx­Oý¶ \ý’>:C©›} ‰ë ¨èßbµ»Õ5{ Ê<ï Býô|lÎÒIhºã.©Ùª ?UfoßÿË-±j+oÙ@ënŒ÷ëFhæ ¾m:í˺e­¼ÐÁ{ãõ§vah¯á,ŽÛé‚vsãw7/CÛ´Š:‰¹%a‡Kî/Fé™þl€öÑHÔèȬÞac‚΢QöÚë÷àÑÊóŒÎ<辿ÓóTqz•÷¬í='}†×ý=w`À®þ‰¦Ûm¬Üé“N´„aÄà8AÕMKL ׄñ3ú…'Ì5`2_XÑŠ4fNpÞ1¬‡9¹nƒñ¿aA(3ûêÂbK’ƒæç«°œb:0«Å¥Ñ»S«°ÁJ–Ce2[ 9iW{aÇë2ÕÁ$ÊîŒHEM½²“Ò¸gÔ¡‚fc ÉbÌn†ñ"Å9o>êt¸÷Dìý탦H}‡S–ÕiÛk•L:A&´0ñîd|dDëÈô—ñ^ÈL#2û-¾]&BVîüØÚÝ{gžÊ¹A-ÂL ÓTHÝô§Þ»¾ÁoN·(äèo:ÿ‚r?y†œÈf†¬Ï¥e™6¥íà{ÛˆÒ•½LºL¦ í éþÀwTf"'¶/ÑÒv“(ƒ:!£CvµáªdóŸO¾y¡ òN¬<8/qŽkƒÏ6Û@‰EÐI!ç‹P*ûÄ÷”yG@Ùô/×àÚ(ïSx%=ù^¤ šXÀàúÇÃ{ìªaØ¿=á£çÿÿ/FýÅ>¯{<µÏ§ý„‰ +ï¸Ö<˜æø:uï™9ü¹ð|²s‘æú"ç·×­`á˜å%WrZXL?/ZüžÀ­ª¡€ûËs°Òüsd0¼Önp¸»iÂÆQãÎ]Ž2Ø‚ÞAû“9ð7¸øt°á~$â늿Z‰ÒÅFÞ½À=†ãzªš-HÆríž±G3RIó »²#åúg=‘+^HMôsÜ%Ai•"ßývBúª¦ïz­ ÈxÖ é…þ_Ü'ùöîLÀ$²j,›{VlÎ9DÆf·‘½%Ãî¸ çӛđÃÈó|ª¤ NùÜÞ¼l?‰Ч‡ïÚS¡äcÚ2Ê9O/U‚b«»4Çr ñ`[Ñ=—§(M=áÜô~ e:‡¨ÜhºðpQƒ9Ê¿“ª ¨bBŧ+1¤•ùuJwÍ/¡Êß7Rþä¨Fs‚U–ùª—\Ñ{yÅ…4ˆºß§@­'õ7$rÕP'ƼV=…O+öËçv5£ÁŠÞ55“n4–Øc.ÓpMÅÿ|Õ®E3]–6µa4¿<òˆÈ9-â?º×çÍ åð^ýþq´’>—xí-ZûW« Ú5þË¡ÆS–w«¶h÷1^ üñ´g‘ûþ¡P^ØR(kéü/‡Ÿ{UE1 \÷$Ëδ:Õo_q[y½œ¤I÷ÂÖ»Ûún+ŠÐ™%—fu~´Å½t†ê7¦.âLPA®™Æ¾žel»ãÍ3P2Pô 3—Š4Ɉ¡@:èòÛä=&-<ù r³Õ¸ÂM ùõ)Q¢´›lø%Ü!ä[a¯’ O¨ð^p‘ÝïEÈ$£±d¯¡>äÝ«~4É×D“Ÿ…Ÿò—­Øyy-eÙ¿±Üê@ïñÝDÿä {Ôº×Úy’ƒÃMu#Û sÝ,1òÛ'‡Ïh¢@YZ½O‹. °—Y7˜¡ðßÓÅwÂjP´sŽþ7¯/Šw‡|<`}%©ìJ{¾G ÔÍLú”amŒHFÙ®øÇ(g.ë›çᎠZµùß¿A%Û[©Ã¼Œ¨|µ;®Càª<êé1_ù€H~Ov‡à÷Rfû먡+Q{žàÃñ¼»*Ũõ¶Ë߯ê êx ”>cG½3ѵ¶[Ðàäø]ôy4r Ô¤ÎØF“Óµ¯ÆFGÐ418b隥E Hðm¡y*?ü@‡£9<ô§[ñø>7׿KgÐÊ(ÐÝ£‰ŒàÃÒ¦Ý,‡bâ{žI¡íQë͸J ´‹s7 iŒöô×NE½e&øPøB±i:µ·%¡'º.dAéÙ…¯{,¡Lg¿YÅ£y(ÎP6 •ƒòÙŠ=á!¨ÈéÜ[ò<*ïÍpfPC•Éš´ÕÐÿ{ñµªöUøn8­¶Âk 5 ß\Ô¢ ŽƒX¢#rê¦FûȦáÇÚ&eB-4í°PôÜ&‡ŸVw¸"|¡Å`èN/ÅUhÍ0;r+Ú8aâù:Ö^w±°ÙB×@-Í-èaµã tY…ßü½<`€5è¨ý½»0À ÷õ5`˜æ¢³RU3Œd}ÐÊ}Æ®¶ïù¥?'»E#¹`ÊKWß@=f¾tšIi„9™dS— ˜Ÿiùu¨omy$o³ž„¥~Ã.úŒ#°röSfnáMXÝe"eJpõŒ¡fòß°ùA§:ä/Âöġϧ¡vƒ_úm(‹ÄÚ—We«döóBm²0’f~û{ñt’G–×8ö)Ÿr=ç>p©^Ä––u!Í·Žg‡ ž ×0+Þk6‡Œ\¦Ÿ1!S7…‘êß0dþiAqu‰Y·NÊ»\º‰ì¶#ËÅuW“4Õ²Åǹ7|Æ úÓÞOÈšúÒ`VƒhmÉæØ}vG`ñ˜ K\ã,ÓE5õð_…U®'u®¯°îÝìð›ú$l±É?½øUþ®Öšÿ@¢;:kŒ2÷„vI×è‹îyÿøGï>S$3ØI™IR@ dóç®H9(•²ç*RmD±#­e^O“YÒϹJ”|MBF»n§%Ædj¡­­Öº‚̧žÏ\A–ÞÆgTæÈ–¦ÏÖJ:Žì÷µ<qBŽk¦ÏÜE®§§úmx,‘'QhŽú¢òÑYkçqg ¿é Ûéç(f²)M€B‡3úü\Báþ³±\¿iQ43B†ßí"ŠÕ8!xvþy⟤‹Rê¾ç’¥Qz«hÆ-le[¿ôA9Mö}üñʨpìL(Ãò,**r9…ʞퟆv9QÅ»S÷*ß;„EÂÙàIðacã•õýQ¨¡öL¹TèÁ‡stÍñ‘¨•¢ø2‚_u^ú’nò žo5TG?S·!k4zþ£¶ÖÈM^p hÚK£i’dI4+ñ§¼’ŽæÍ×{[­£Å¬É ]çgW  á"E+Ó7ÔãG"Ñ:äñ´xŒÚ º¾¹=Y‚¶G¦'Å. ]Ìâ' |†öt&›Óè@(}¯O(8^ ´zþ_>¬Vîé©D_³L•Zw4Qß™,/ÕAö»ëWÝ8<`U©h[ÿ`*´ñrÐËÒ„ï!~Ø÷"Ê÷ª‰žÚ ÅÕæD|ÊßâöÕ¼2rn¼¼hZ™¯ÎÓ;Y=ƒôÈ—¢°CšbòÈv¤ö»4k›ÑAªr¾LýóÀÿ÷â¼ñ @ú³fò¨’|Èü´TÕÓL9çut\ ¿`F¬_dŠŒZŠ¿R…Úп"qð-†ºè|L*”&X«¬ªÏB™—xša_;”+ÙHÑø Bùf€­"Ó~¨(Ê“iÛÇBðâTö4ý7B_ä{vªoQNg6Ã÷ò·¥s VÞ]_rz/ÔÓX𰿀äM‚ÎÄÐD49I- ÍïÆ>Zk¨ÂÏ’§A?¢ïC+ç¢àì/ h &¯$œS&i¥j¹úÐup"Î0-ºO”y;× CoYbˆ]T#ôÛüâÉ´3†A–0½-¥zE,ä'!#Ò×”öÃè\™î(%Œ—Ú·¿¸î“ÍÉlk¤0³¯®›Ø%fý cÞµJÃ|¸™ÖZn,’´Ë*êÀ’ѳ¯­{ayÇÇ÷ù7¬PFæîhÀº¸W ]œlRg¤ÔÅÔÀ6Æ­ÊÉ<…ß×´¿A¢©ƒ¥gʶ$øj”½7’šÓÜ ´¬AróÇ|’HÉõâ'sÅ=¤âª(Ô?W[ÁÍ ?Ú;K«% ñ¯Ò&²;rhz#Èþ­¨šŠþÄe…æïb4)¢EÝ;óŒréÝ÷lgÔÿ—ÃÏ\(Æa@}ÛÕóîËxüžaJ–c(rWEë°*6Ãæ†%Õ)FèP[®=âV ¼kµ)²PuAë>ù€:”ï9Q-^Ð¥¼ýGím„â#ÜÓ—^éAAq ë«‹+×ÖRàÊüÿy5üJx d¯J„ù¿€•ÊͺtsÈçV‰L…’£5v˜·¡<³¤ÿx"To^‘œcöºaªm‚Иäí@õšG¢ÍÅNÜ„95ÞTÐúQsø:Í´õ_¾M]EðÚ—H—œ®ŸÐþM³È¦ ÚeÆ‚||‹¡-²ÜlPú$´‘ù¹¤R•z¥”°êœ.´‰þ–ÿá1 måÄZR=‹Ð,þ)÷ÙeèÚÇuЙ?÷¹Ãš~±¼21ú, ÝÅÝã±KCÐûynŸˆÜoè[ö?Ã< ?ô?,ô¤ÂÐYI>oëU‘t“H †1å÷š¼ü0ñA¤]¦iJë^^(?5 ú9•`ÞÑ𜊜,,l>oäÈKe±A60+?Š æRÃú¡ìLíól°ùóúzûGø[§eP>DN‚§$w\‘x±†u¡} ÷øñX‹ÛA2ãZÛêHÞ¸6»*ò )߯«0w"U®çwž²y¤e–þ’Úré³ý­=¡FFÝ— qd*z:wêJ92Ö7r«º#ËŸvõþ#ÈöNJ9Õ~ Ù¯ïìÉi˜AŽ ù&"­ZÈåsd:qÜyâ uª_ÜG>Ò¢LÎ ¿&©WÂT; ÄΉöiT¡§ÁVÖ9.»tï› 9Š¡ùÙ ÷#vÌ^°Áƒî[Ên •™KÒJ¼WÔ”=·ö‰öõW”“ß}÷¬… ÔjwµÔPÉy—ö¢?*ßlΜëîG•+-Ù/âFuvçf>ü.%ÐÓZ†ÊN>!®º’_G­ìð7Ÿîm¡ÎGîA³s‚¨¿`¾¯* RòfO ö™ø¿G“üË™ñmwдæ±êã4kðø“g æ=É÷¨ö Åêä;:Ã<ν“uüÇ´²øšdÿþ Z‡ü|¤¼6ïéu ÑVé~gæ±·h÷Æ3ÕˬíiÅgí ÑáÉL‘ÌtC¶ù©ÿòaùîýŠZôu˜IŠÜg†ÆIéûûpL-{«å<¬¼¥ûÜ~ï8´ÞT–]–‚êNñÉó .ÿlþ¡"b*š«¬oÛ!_2+NÇ‘ ²ûâ.óÄÆC&™ª‚Òñ«Î&îšü:”à½'§¸“áKsÜ@ÇøB9óeRdN½¶‚´ î>£{ ãO^ës_¯}ïÄs|mGQéƒBýåK…Ÿ„ œ ü¦Œ²’‹¬Äq~™ðmë ©ÙV7”V½þ?âÞ3(Š.ìÚ%‰$I$$ƒs΂ ’UTDTATPDÀ,A‰"HÎQ@²(9#*9ªœñ­z¿ªïœßç×®®šžš©êk¯µºw¯mœsªÂæSÛGC5(dd'„t1B{%•›;ϾhL†Ú3|2MNP÷B(!ð™!Ô­jÿ’Ðý›{=¯©HA³PY‘‰·>´2ËÝY<m¤EÖ’í¡ƒ3°3n~>IÉ>­¸‘•Ï¿‡®åÉMò½ çUÃk_Eè ’Š¦ø Ÿ ©’ø_~‰|aá‚¡ˆ»™ÅOaä9ÕyaEcM˜j‹„ñ̬oORH`ÒìÞå¢P>˜æ ¿V; ³ƒ('²œà›§—9 ·=|ïk]^W} ØžW÷3£å—¶¤`™ßXÍdù3¬Ü·3s'«5Úq'ûý°þ”ª’Å•6¸îƒl'2¨’ hÁQKòÖ÷»‘(f.ú™™4’HV;Û¤ ¶Û/´kqWÓxYП÷¸;‡ˆ™Ú7)3dRÏs"u›¹ûÃŒ7H«dõú†±,Òßt°‹âTAFéîo#3?‰Å!>¨¢YD^§Þ}zÙ|ô…o'÷à¾õ}ÛZe>ÈÕ‘#7ÈA…<—׿µÐG ŸìnáÉ‘`ìV{}î4 )ï8¿{¨Š"M,V%Üçðà{×]ù)ÁóÖ[Ò¾ ¥Î½sS=Yˆ2>[ɺ2ý(g¶Ï­_# „ÎÎïÛ¿ lDPœ¹uUö=üõçZ'ªâÉ]ã³”¨Î&kÎÚÛŽš,!½vÊQ[a³€þ±+êž1òlîFý’§7û*³Ñˆè„ãë#äh"ÅsÒ’MÈ6¯Ð£Ùk™ƒŠh^ö(ã5Ùy´æÕRèD«=ËEéCh­§éf_ômb[†Ý;þq¨Ø~Ž[eRôùXKЉŽôD±Ë$Á—Æ\[Rùáð·xòs¼Þ‘rWä;áû¯šguµ"WÞîÙ&9°YýUçÅD(ô\wHþ½y Ztø:?…Cí'UêUKÂu?a&Ãñ *ô·d^Òã¦<…;_áqðâgÞ@(âf›å:vðÖ}æ‡Â;–'Ê÷yýáÝ‘ïk…ð~Æ–ín=”9z¦sõUBµI·Ù»€ú§"æ¬Ö 9ª¬UÖÔÚ\»'tª¡£ATó=ßtÒÏUÄ>‚®›äÜ)Ï¡»a¨aêÛô$Pèh_;=)CIÜnнÓ}…Œ ºígF-³ «h±Aùñ-ÂyÝå‡N*AWsŠ„¡)á|—¶é‹ )Ðsh-F…b zy‚SƒâÆ¡Ïðû{…°$èO®97Þù/Fˆß¾òyûÔ¸ÇÁ VCJ…ÂM=a}ôÑîìúNØŠ.ðÀo·à÷ås …«¤H¤MÔ ßb„Ä}É“÷ÿ\FR·‚þL±7H¶•9¬ajˆäZeV•!…ge/Õû·HunfEp›iº^‡Þ¶°Aº‹OG$Ýý‘AT0°¿<_Q1m¤0!“"“t¤Õ,2Þ<±^À¬ LN7  »÷»ùéwç‘ãïÚLPrß»¾š­‹ÜMÓÂrƒg[ƒ*þ 3ò)}=·ùcnÛܺH†‚d“lc…i(ôìôQŸ(â=}æÍneµ-R¹Z~ÅÎ0p}õÊÄC]<ÒÕ¯rQ².:œ5þ&J›}÷¯y2„²ÂS2ë$ŸP$%¾ßÇÃÞj¥Ù¾¨x±¥ìWá3T>ÑÆ—”ŠÐ—VÙæFÐÊ؞ŠÛ3¨.+o@?˜NÐÃß%ç}LQ«bõUÉ— Ô)-©dïE½*N÷­ 4èâõ¹‰²h4eÛ£¶&U;húùZœ•jš 8tþnAó©Þ€ÒËh±sõøžjr´hp`»¬ÖÖ5šÝ´hsÏûáþGh;<æõü Ú+ì­W!5E‡¤ü*wrt¢!Wrõ6AM&›CÆ RLŒíü?8¬¨åÏlÃp_!&÷hLçžìïç¬e«ŽÜ> «»]œ\ÚAg‡.…ì>¨ß÷)~•J*Çw·Ælr[ýÍn>(º±«´}Þi½q-œ¿yœè,í ç~MÞ‡|zxëU¾¢>¤ Y5'/:î†Ì>*†“mõùSûe_ÈÑðcñ¢6üŸ÷1PNÞQ Ò ¦ZÀû oß}ïÀ]Îï•tP*zYi=Ay´{|Šþ罩÷g¾˜™#Ù4œ *[ÂèïÂ+Âü0½-²oœM™ŠBõ2.aAPóB0Œ^Æ‚0hª)¸7þ½ßDÕð2 ¼ "¿a“š0ï Î`háﲬ™:Mȉ7KÈÄ# ]tÏö} oBN|—Só>-ĶÚ=Є.›–ÆÑcw ¼Å%®„Þºs5M®{ ä¶2×½rø"ŸÍgÇsŒO?g]‡é«.ËÌ…0úXìõ¾ÕQš|B18•+Ê"ìS×K„0ëÃL$ƒW®k!Ì•ínýÌ ßE‡ôõßRÁÏÎddìÜ„EK5¥l_NXª3^]`Ê€1)Åž°š°ZMòn ÖÙSŠöóõÃFÿ@…¡Øú~³Ÿåü>/»Të€DR¯­ÇF<‘¸ë÷7ÿA$½ý"¸úpî:vvñþW5Ü­~ÚŽ„ù:R*üùÚÉSÔVq¦­ëqÏû¾V¤G—ë7Þ›#ðR‰´sî-îû-$€ÌÆÝe§uùRå:9î;efêà…\Ê:Nw5­‘GÂÄáFÛ4òÞ…ücÜÈ_ø;MÓ û·Øw=lG÷3þ‡Õ)ñ M[cmá)<Ôf”ZH-€šHÅe Œç¹$e”“+›×õ{… d©îÇÖêðÈÜÍUrTÐXú!郪¦VM'Q]ž{çáÆsÔT§=þú€jŸÎotœcEÝiËF#‚h@{Ñ4fÏ&Ù¶â‘òB“— >üަµRÔ«hÖ¨¼È`ãŠæ}ýï厡ÅrªÜÃ÷çЊSmP=ò'Z›Å<5°6ñËÛŸ ÿãð°Ò½e¢ _jJœ¨½CàpiÉûÃ'ª+aŠÿ/~ý"ÍÉŠ×{\n”ŽN¡erñæ 2äL9{H÷D,lò½JÞ34Ý«OU_³†f?£ÞÕõ.¨=¬ø`üí ¨"òÿªtiç¿^Œ’;aÖÄ~§ XKá#wïÆ¿~Œ»È„;\ÿëÁø¯ccµuåöð=ÿÛoó¿=ÿÛƒÓÆqÈœ"bô_oÅ÷ûræbÐå{T¤EÒºsL˜«Ë çò‘^©¥ÿÆÿýÝÝ\t§êþlÿ÷ù.©`ëý?ý;ÞI¾Ù`¦÷ßç{Š]ˆ&Aïð·ªYŒÐÏ¿’´™õ¯·&qH8Þ¹:vô¦`DD¸§aa´P*|ÏŸ«0~ê~„õ¸?Lž^yq¨¦ Ž8»¨ÃœÄ!K¦ð~˜ï½3øÐ¯~»{ùÚ%âÏЩ×ተ,ü’Öü|¬Š—PòÍÖÂú½+Žì1°¥µÜ•AY¿Õ Dù„÷#‘è Ä‰rH\zº@Q˜I5ô‹¬“äÂç6ÿzjÄnúܶD*YG^1Y+¤¹Ëcæ×­€tjUŠ^ð  ˆ î5"ãŠwš#“°cÂ7dþ¨Rðææ.d½±yå8ݲ»øõ²¥ ‡]TJ:w;rùþ Ó<¯†Ü&7•»Ï"ÏbÑ|hN=ò‰¨¿=ü—Ä#ó5Pàûey²º¨Jœ7ˆ"Ú±÷–,QTüä>³èS}:}]ß=ú '_ dF@ë|\J+ŽËeР,{kíßõ³òJ !_&'ððÉŸåHT¼\ßHïÊNM±í/#4m†e ôðýÂõgú´¨~ˆÚâÈ‘rÔÞÅÐÜqµš-óÚG N§mʘ¦ê蜽‹[7ç[øÑ˜1Å„ÓM)(ç¹F¡é\Ø8Oô=4›Ñ.Ò:ëæKÌ5Õþ›hIÁ%¼G­D㈿šFk»ÞQý s‚¶§(ª8Ü¿(Íþ¯¿¦—¡Ôzw1þÇíÀi¯šÿõLý×[ÃWíåá  cO÷[Ÿâêý¿{kFÒ›.³ÒÁÃKuŒŸ†ùŸi)çá§àC¶ç¹°ðk]Õgˆ–N˜dÄÉÃr ÅOzcZXÝ?oýš¡ÖîHçN¤À,;ø‡ºýÇñ¯Zæ¡RÞAØÉhÕÔVAâs5w·1IÅ7‡EËl›\0·¸ɻǒ7Ê£‘¢®yŸ¸ª+R}mSÞ¹1‚{díîíÖFº¹ÚçíÈý4êj·6îupçºrë"2[­RVõû#ëÅlû£'‘}öú“¢ÍÈùjèšžßä®0JˆÙwy¯s[]wG~k‡â{õQðìJ—U’3 ïtD¾sOFÑ åè^ýt½µÞ×ÐöSä¶ÚžnøÄ¶Re]^ZîN'¡ëÛ®§K¡G¡}á±ôRð¤uÝøo챕û¹+æ,t?¨{Ù_k]ŸmKäý¡+¶hdÈxºª–âb ÛŠÜgIê"ôð‹¹A¿½´ÅüÚBŸœÃ~±Sèò8f}}çcc&À—9·éNµ§0ÄeqíܱúâÿŠßFÇg´¹Ia<Ť-ˆX &S$Jæ'Àt¯U¡(ù˜Ó«ÙËËó?³ÃÏ7|æ5½)E*X _ªÝõ–e.UZåÏÁª¬ûâKˆ…õ8ÚÃm”.°%+ìšCÐCžñÙÏ)ýHÄ$óàˆÄ&?‹Ì¸"¤âyœÕ ÁHVÃ~%Ãrv랦iX¥s¹£óHÖ"Y‚ÆaCž6ÕÌ<¶ŽØªys?…_ù›3°s©,³(ˉÍ6Òº Éï7ë9 º‰4©ÜîEƒzH§Y}¬ùÖOdØ#|ô û2ÞT~¡çêˆL/š½Mr¹Š›1Á…à/C÷:ãÃ…ìŽ*ƒ®SÈá(_œ7‰\ç5Óªh[‘;Ý{ïù^aäÙþÉ<<ïˆ|²ßm›g‘ÿzÚ‰JÚMøÅ¡\6]„BI¢GĪQÄãdh®ŠÚW)ç^A± —D©×rx¨Gõ I½%J–?}âÞ‹Ò*Oe®²¢, ërÁ[[”'+·×á=Œ‡9Ž* ¦BEv Ñçq¨´N¿²Ö€`wNòò“k¨šì/¨¶”óŒûà0jÄ´fX£–“ãÓ6ÉAÔQeÊãñF=‰1x›z ¤ lbï Ñ‘áÏRèÑD"öêžR4 J7]’¾‰fžóNœm’h~4soŸÎ Z8{tuQ±£åqÑ y¯´:k•Ýr­£?é„ å MÇý‡hÛ£4ï¹+íwkÆZ¿uEõÈkI¼Ð1rk4~5º‹¶ŸNÊâQ±]¡õÐE;dùéq ›´­\?††D'Í’¸¥ùY¹ç“ñDX–ëÒ?¾>II9e€Ú§¢÷ïÉ›j!ÙÞð¡¬xȘŒÞÓ‡:ÛY„ ÊWJqG9Oãæ¡¼9Èö½fr9ì'deTéÚŒnÜ d€ôY¢÷¾1Ijèøb²i½‰ßúGBîsÓ®[ÉqsæÏ»BÎðþlÒólzø ËxþÈæ!Š­Z]ƒÒ¥+º3žw¡\À/šï<;TX;Q8ÅÿÍ­ß8¿@åJÆu…?dPõÐÛ¦™ì TÃA'æF¨ûžÈ=òj“sÜ€:¦âZ¢R¨÷“)‡h nO¥ø ͤöÇ«ÎCËeØñ¦h#}2ÖJñO]Ôi;Í¡³áBŸ@ ?a>HûÇ¢½’µëÖЯæ}Ì{£ÂG­»®õýÓÃKtwÆÏìQîöDO9{{ÁàXÏ” Gijo|ï…Éå5F{)˜|å̬.Ìíê5?ý§æN݈û?< K.]†…~b?rwXR™tô«€åäX±«Wlae§»eZà¬oi¤¸õ†øqYÁ—°%Vš-*w ~ÅñÞTxÛ;ÆÙ{ ñ!×Ô{ƒZHÒæµîd7I¶íÆÜVo.Ì¡ )4£+=Ï#•ïKiΤÉÜ Ðå.E:Û}"¿x‘áðØIµ=ȸö”ô¬/2Mæ= ½„,[Ÿv©ò?CvÛôÖÙ‰`ä¤Þ5bFlƒÜl¿||Ë©WšÚp•æ2òý:ч‚ -½K£(‘qûÁ@?ŠÚŽ%÷“¯¢xf¶ÊÜ9 ”ÌùÌ­4´ÒgÔ û¢¬ÁsfÇ”—¡s2R8‰‡©qýV AåýÛ§ç{CâXG-QíÖõ´¤ÉjÔ¸0!¬`…Z7ÕÈ¿PÙ NqËü{Ô[ó*/ ™GC‘Ûñò'~¡±R”Ê©+šÒ{ RþñB3’“6£÷ÑlKQþÃa4ßèÎI×½‹;Íj޾ъN#Œ¼+­…ùógÛÐFçÇ“ÇÄÌh{:‘\è~(Ú=—ÿ½÷ð"Ú0+g5FGA¡õxyt:x¼£æ:·8HýádA–“Q%…FYmQ!¸Ë Í ½®ëAö?tÑީ̰6q0†Vmñ.„|¸Y{¬Ly&:'tÊLafŸrt•Å!˜»<GÑ¡¾óε Ó×ÉÛ5ßßg`ñnëÚ6M,ˆ 8ŠIÁªZQuÖŒ¬_?°¤ÐE[ô·Dý!èá³—îm}ØùXÙUTA„ħs²Ø¥Ã”$,‰Wæ=’½Ð8— $·é^md-G HQ%¤Òg”0áGšû8»v"¶úÚ¶Ì2P3ßÞ÷5¯± ¼ëG&æfwUkd.¡«ŠjåCÖ`Ñ™àyÈn/8J­…N·ÄšN!×¥“?vìï wnÏ}bä%¦£B>EpëÕ<€ü1výá¥](HêUÉd'B/ªsÛ½Þ¡HàÑAí‘Ý(zÚs·´æ9»_q’¼-Î<1´ÖBÉÆbŸí¶~”Fâ|êRN”YS‰vö)D¹YsÝK壨°høŒ§<L=|§$þ•j ^EÙ:"ðÊìž¾¦„ªn7–»mP­‡Ô€Q`5|d°·ù –ÐHR›8%j¯-•T\)Fݾì§ÌÂK¨ß³Ý½Øˆ†uÎ:/¡q…‘ƒ•Išª©ìþBN†fPâÆE4g9W2Zà‚”ÑrËÑ’ä|Áùw´"N>vöZ“wr ÆÏ¢ ãé@…ãh+HNqøåÚÈÖÅæShïl{óó¯yt¸–Òœ1s  8Ÿ±G§9V¯)ŽxT¸ÌœûQ†=²rOÑJC­Gx¦™Ég<Ÿ.úÀÒC·còèx3ÔÀšµš¤I~”PÁ¥I+, ,zùΚèBáËÇ¿ÅBþ»œi%1ÈU ˆî«ü«skeø2;x7 d õé¨ÚBzÔtgž‚ŒÒbŸžä·ðvmÜ]î¬4ä–ôÕ .AÁômÿx‡ãP´×’øR…|h mÕ¡6 ä¾#'-Ç¡Ìó °õòe(3T.!èóz°âîßPégÃãäd U42/H߸BU&ƒÓA[¨Ö[`¼KÇÕ3íÛ±? æÛ‡¸ QBÞ5Év à†zW¯mRßoÐÈôs‰Vpš¾µÓˆ…–n—dºv)ø¸þ%S¼×úŸºÉÔ»aw`‡'-8M;‰©r^¸fD’TÁe‹$3%ö_óGrößw´2¬‘‚tþ.ï¥0¤b5°oñhDš êžQŽHG'¿G^¸ èÂ]Ó¤’‘±¥|BsÆ™ÞÜqÿfA‚,\rBú‘^q,Üž9JÆt·Hãþº®ž¯±õÈ»;L7›—ù:©]/–AÿFÖT~Öaº¿†¢”r/l'9P\#.@­S%O Déž{Ò*ÙYË(Ë6¤TSqå¾Hs=þ憇oQÇ-äN¢R"eDÉj9ÂÀÀe>éBT©K²7ðDõ?‘•'ˆo£æ†è£åNÔa6(žwF=Õ•ø3±ªhàö@öÚu&4òré+»*†&ڥѿO¾FÓØ«¯²_¡ÙeÓ¿ÆlÐü´ä¹ófæhá}ÂUšº-½6;&‰FÐêäÇòŠ× h}ÉúÂE"´‰½ý¶÷rÚ¦ßKeIÒE»æ·Éçtï¡ýRo^U—<:ò)”ˆ¨v£“ë*í^tNŒgõ‰eÂHçŸtJÚÑ<@õðù¶:dçú¾Óâ¨kûWL‡Ü€Î§ÆdÃø¡ñòì1x´Bٕ̬¦Žoœ t`ý#‚#|8ϳFÔ½ Šnlìa†Â«¦më6Ï  ä-ՆЕ¿=Â=ÓgÛ!/ö¡“Ƽ»:sŒá ”ü|úH*¶ªƒHCižuCÃ{aÃâ´Ö5ò¸ôð]ô“s–ÖÐy嫸üýtèújµÈØÐ=› ;)ÐÛfvhžú Ø(ˆŠôÿ{×ã¼Ú^´ýí9Ôc`:=ð]‰â4Ñb,Ð.¯¦Ï€Å'œA,±›°lCÔ{ÔÓVueJ*ÿ„õð÷móï?ÃæŸj‰+/zXúö9ì¤ù½pøq‰mëÒÝ¢_"Éü¬mGn/’%8äH•@rs¥fƒÌ¤ ‹%.úˆT&Æï¿”|åþ ¤SH§S}ð -2P’–Q¥²!ã•Ez¢ d¢ûñ{Jò72¿#:ªÐ‡¬ç”\Âêu þrÏD¦D.r8u¿%ÞF®°G**/ >³ø‡x^ßä%DUáEäÃXÿñcÞÈw¯ž–÷[¤*Òðn.@¡œÈ¾ò—¯Q$ÒdÕö5ŠÞ ·îóF±¼ýÚ_ÉiQ‚öëý)y”ììX³><€Òª®ÒwªùPfΟâÊ}œ’ÏsŒF…Eª‹£x$„Ì ð*ÅÙÚsÍê Ḛ̂©çªêxø!?‰jï­–wí'B ­‘ë7/\AÍ÷ø¶ò¨ý¦£ÎùÞ4ꆆ캘ã‰ú'&²Ì-vÐÐ¥7C‹»M¨¥_vGSŠŽòšÏ¡h:O¼"¼fUÝæyúм çʃÁ}h‘Z»pç¹ZÞî—eæ*G«°Î…&>´bS0OqC›³”º‡ÑöÜžãiS,z›Ý• ‹êùÕ_ÄÑáîeeâ+èøú€ÅÛšFtª¾\ºå;afæä,w‡Q?ªAüäÖqdòfô|¶ƒ°øk%ÒÿõUè82ç\jÌ 5“_‹KV% <‰C;%åÏ.y$%ÅA¡r©J¸™8ä»ÙlŠL퇜'·Ï3Þ‡·9¦ONx7@æ†+Šû ƒmOÁªHkûž`oùWƒ+%=àí°c}iÐ0äv9‰˜YþÝãû8?(:}ÆÓ²] Jô¿óZr–raXÈÍE5(ìÝUwÙÊGòÙEHœ büÎ,b“¿¿È1.€é©; ^%a0»E~ì¹÷k˜W^Œò!…%2W:ÏÂB>õô5°Ät‘úÃK2XdöZË{ +#ejB ñ°æ^ÕÇ2N.Þ›-³°µ÷ZT†« ür—ÉWʇ?KwƘö^E¢©ÌCääè¢`²’±ì|"¡zŽ»¾ä¦Üø‰»kŒ<¸ân!eÏ9:÷œJ¤áO¸tcå7ÒÖ“ihX ýŸÅŸmwd|4(éèˆLçN‰o ÝD– ¥žþ=ÈÖв±¯ýrø‹+ÙÞ߃û2U ]Ãç뺦—ÜCä{±ððôÃ^x;_kkàŽÂ” —̘ð@ñ7ŸkÔPìËé—%hPRüäª5Jý¾7ö(8ejvwð>F¹«–ƒ?Rmñ0G/é¿1*ÑyÒîúóá°À„=ªÖ‘ÞcÏ݃êìnÅÕ£æ &)šÁQÔ>½÷¹ãü.Ôõ8”võSžôv¿CÃD…ÓwHÖÑØ÷ùÇ©±ËhÊ™¸ò^”ÓçŸG³Ù[ŠZöôhÞCve3#-jsxjÑ£eÑù¦„â£h•Snb»S‰Öo"m"Oì ÍÛ%5öíT´}§óK¥Þí*cà“ÚwlÒ‡–ó¡ÃTwÑ~*ft"Ñb?š{#©VǽôÐlúI£Ø©wȦÒzVŒý$¬:ï6rš” ä2ÎËwˆX ‘Ô¤ëÏÆ4Ô0’$ñ¤@EAø@ÓÒÁ/ö^¸:$o ½É€"»Ù)Ó}ÉP¨{™¤‡ù(°f)S@žKTÛ/¾Ÿ'X¼!QëGàíØI•6(iµ†C1B÷}*Ìz½AD{v2 >z§í½ÿ>qߨ¸ó :מÏåílA·-9÷æ¤&ôL¦FÇ,ûCŸU—#Ÿ ô«…î´Øü7ö…“Î0¼™‡Þë)¡üŸ+ 'ÉFî”AE>õ—iÐÃêÃþÜÒz2Þ¬<ÎÓ‚Þ _"S0 úÜâ/4C4[Ü¥*}øÜ.ÔXÙ _×HiîÏàlð̧t0|žÉÛ€QšôNþüzKª±ž{BV‡×‰/¤Á”ÆÙÏÎ 40ãÌïj›Ú s,t‰ôðÝRˆy ³8"ŽO ŽÂâ[Ÿmé½f°|¬ïƱç`Õ„ænþ°~Y+óIrlίÖôð‘öÌûrØ9³/pg?ãÀË7-HÒgÆzEbÉ®Ÿc}98…ä†OSÚñ)R(T$Å9#•5µSd¡-Òäî'N÷ž@:=MnÇp.d _ü¶åš…Œ—Z´kÚ‘‰joâsÞgÈüv‘û̳[ÈzÊ ÿÆò²[­Î˜ts!‡Ëá1 ÅH䊬¾¸Ä‰ÜՒжÈK­j± UŠ|š#A W2‘ÿAë ¶›(ÈH^_¥,B%ú¿ßŒ'¡H‚êžP]M)zÁåí‚bm!ûse·Qâà‘˜á@”ü:s%çéW”V«ÌcmF™¡r§P./X`zK Ü;bˆG„,±•ÑäK¡JÎÛÂbרzPc*åÜuT{ø:iƒ•5¸Ý7ùP³x&ªMðj»¿mû΄ºÂúžc‡‡PŸTɱ½P VÜË™G Ñh¨‰êÅ›Ãh’ÖøÍÎnMÓ'ºv Y5õZŒ ší±Þ®„*Œb¨–Œ–œl7êÐòWWï¶#Z 'ÜØåŠÖµIøy6m2;êIn£íý_ñ¶uËh>•<ðñÚŒ|u}‰J*OgeÑÑIU¸Ãv¿º=Æ…zíËö3D¸÷£²ç‹=L°èrM V_Ú/Å”¹Qj@ÍõwŸ©F‚¡œ-KO‚% НÐfTNöûÎ7,­O¿ÃÇ}‡rj;îp¨ûÀ[sy}‡aiÈŒ;s&ÒOÒ‹Wâ»'áu[!Eæ¹Ã!q÷fFí~xÛÐ{@#ôÖßu8 ᓌo\Ý9(jo¶#pjèzHüo.¬z“ åW&d)-€@ÖˆhÂ?f9´•¯Å£\‡´¡JIkä 'Tµ•z}¡ª]ƒBb’Ç¡zå^­ÑG¨^}ÝNíµ3uI1–×ÑÝŒi‚†ÉùyÞ¿û¾ÙM{ŠCK.í'·Ù¿ûB)-ºnüÓC/V9í}LÐù™¿ rðtGðé¨wxw8êìÒý~¨-Ï«ÂÀcE½’ðzxFá•Ï]jåpsôþuÆÒC¦tÈë`Bû¬j@]^˜~4i/u™†r }iµ¶SÝAdôÙölè°D&ãuò-zd9rõÚiud uÚ“¯Å‚Ü•3_gp?wxF©Œ2òäù©òë !_PϧÛÁ(â]ù`ý% µ,&ùÓHã¯/…öO•ï£þ|Ú赦ªzxóÚà›A=èòõ{-ámÝ]…Æ£$ÙÐ뮽«`¡oÐR~Ò6úÛ¦×ïoŸû7Òœ÷°,¤†>RËwò©ìгu8Ñ>råbÂ#qSèñí÷r¸ ½D¬½ñç 9S{Ñ­…ÿï{‡ä{i„¡äúwúº›0 Š/LáKö…£ó§Âað:Q:Cö 7–îÏV7‡QE'’¾1k æ`©Š€‰ðe5þ„ó0å[×5g0 3WXZ´9`®-£ÍEö|÷ÓøP_ "L½üml°XRb”Ëþo<‚U«ÂKòS×`=¨ë†uX(l~9àä³\ ¿‚†_û¯ž‡­ݧF H,ôóنʒTõJ¼eB² ·È¤9“‘\‡òŒy×=¤8ò;­äÄR96’J{-!Í{®hãŸþH§_›s@ ÈFöÍ 2^x\k€Ld‡¥Ž1ê!óëÑ6O:Udõq4ã=ƒìƒßV.ÜD×Û?sÎ0"Wôâç*£ÈÝ óýómä¥wðLs»Ž|úR¿Sæ‘?%R´¹ÞÙ­ôYßä¢P=ÿcžy))øÈgEß¹¬0þ¶B±¹/k‘n3(¡3Fv€œ%§‰¨©iQZƒGêq›Êt¾5wxö åi÷r%£Â¦£µ"¹Š–结è³ï£•6ÓgjߣhÍäîÕƒÖc½Jßö'£Mn[ƒ¿ëm¢®-U¦£¹Æ¥\Ñ^ÀH8‘6 í7ˆãsèbñJ™ÞnIÔcæ¦ÔêI½’ž/÷ÍÀB-suö·h»¦]fh 5noصÔBY¢ŠAÅm?(fý1-Å ï<793yÿ½cÀ¢L97wîV¨Â›ŽeS‹”¿ëhzn¥ÉCºÃ—“RÅàõ©Çü•qœŸ#>¸oÓGÈ‘&ÑÍ×8õM®%gÿ».ü[ìV‹çÉ“ý< ¥F -þ˜ò?¹°âÀo(o:Xýô@EòÆ+ñ˜4¨<ìvbº@ãï:» ÷ýåPeÞöÇ£$ªFÈ–R/AµÈP·BõÖy‘¼ÇñP3°Wv‹œj‰âëu ö™vnØ@C›‰CÈhz!NT»Ê-©Çç÷ ¿ƒÝ7É,„›ÿé¡/Åüûë–Ð9¼ÛbeʺomP`ºôwŸÓ»xæô_²úa¸V /U$ë/ÐÿÓÃSZ¤?t]`”}vÙg6Æ2Ö½HD`B”ÃçXLíº@e¬q¦g#~)Ùò½½7`ž3 Ôìbüˆ$*˜õe‡…ÛÌõJçyañ}©É.,›}¿ YBn:~¿ÛË_Åïé„@Ç™÷z¿A —¯ç¦5!Ç™§¨ÿåÁø¿.”TÞm×tcƒâg¶o‰\Vàývs™¤\&Á¯ö{ÇÅ¿üôù¦lûܼJ0TrͶívÕ†áK+2nPr­,BäzTýœÏ‰‡&r­â­MÐV`xAoá)|Zßnæ­†®1ùã™Ý ÇÄ#¤‘gôŽÍ‡~ùBýî= •±»á³£ÉO—ÇÔÿý s‰å /”ꛞûYè=!}àÝ· è©“äµ!†žjå7‡ŽUA¯UXÑr@ôñç9°®ƒ~:G©¤Ôø,·®÷¼©‚v… ÁWâ­ÒV6üée_#tœf«ž½0êÉ@:ºýÆÖ$‡ÕÇù0çM´V1L%°´ë¨‡™§Tò?ç½anc|;íá{¤ëÑÛ÷ª`Aö…¶K¹ ,6Qíxí#åàˆy/ÎrXutqÿ d ëç\S{˜Ãf“ãŠkAm/x ùIÁÛÙ}Z!û‘˜–ì‰z"3’dñÜ“A2¯§ eæH®vJ§÷D$R(+¥?Àp¤r‹x=Äu iʸ˜Vά#¡–NY‡2´é}žêBƳ¡9Löõ¸÷c É2?í q ^AV߯D d7«ûùs9ŽMü–Ê|Œ\w¸„7n¾BîÎ+j{Oäe 1 3G>“ð8ª2qäŽbÎ" ò<ûèÏ-B]z·œ® H!Êû¤(m¥Ô ˆÓGñÝŠmñÎÜ£‚¨Pr‘õþñÀa”Ö Íåï–F™ú®¹]®(wñ)±…ÜMTÐ#Qß] €G8òC̶PqMãt29ª¨Šfb@œü×ùêª9Í QJ£z×.îþÀJÔ4\nã*A­Nï1¥O§P'pR"[¯õŽÌ”5?CŽ+“ï¡áÍâ5ªQ4®`­ÿõjME£‚e ébš#µ Y!7iÕÃÇh~>ëYÒ%´(-ÎNK=þnmh™u÷Àt—3Z¹:QŒ­Ù?¼¢WÌDëiI}··ÐæÊ¡“Tùh+õ’s2smÇŽ2DËÒ ]|õ﨑¼rFg.ôòCÔµ×_—¬‘EÆ»qž¯DJaAø¥KGL|ŒÙ)S·ƒêOÖ×ïé Œ¹ÍØ”üeéÈì¹ä;¬·ð1äù¥™Z“C nOYàMq²f#dt«*ÄwiÂëßÕ†C^ðš1z2‰ ý©ß«ÐoqðöA㳨 Èc_Že¹ø?üÆJ\ƒâèOÉ_w¢ ´]Žç¢Ýß\¨÷¹Þà‡{Ÿ0p‹@Å6‡¿¤…:T:ä¤õü$ƒÊqÒ£ PåmË)íÝU‹oؔ륡:ˆdæ†è;¨!²‹a5l‚šnÉ*A¢<¨ý¤t­\!ê¥­ØØ„: ¡°öbs’4Å×9çº^÷ãù8ãyÂ9aÍg^³ÔP6òÂÙjx¨Èx°tvfxg&Ó>"Ñ*GLè‚*’¼¸où$ÅÉì úõ_S"…jçþc[W‘®g_acA*OM“-ÂHS±?&j=é«zç®"#Ñ·kµÉÈäslÌGX ™×ƒ\û—_ ë› ³ñddsñ׈¹ƒ¦éËêeÎÈåªJ•=*‚^èû*ü<‹<=NŠôj!gÄS¢ØýÈoÕÍÖëÉ4fü(tpɬ2<…{_ž¿ñúŠ|g°n«G±á¬ˆDI¨é²2” ’Yø&´…Ò[R<%ƒC(£; Ü£ˆ²…AÜQÞY~þ~› *rÏ'—]C¥É ǽô¨R>!xVbÕ÷Lª-¯U#¦ç|LðCwO[H´P+íë_}ÞŸ¨³ïòÅFŽ Ô} ?Ó3ƒG¸³Ÿ±¶j þOi™u…G@öa¿ §Â~È¥M™zâ S‹¤_èã È,îÊý“„,öxåd”| Ha†R‰XΨЋPzëow¥èw(#:/z:6Ê‚†”w8堜ԊúÞï)(¿[C~!†0窋7¶‰ ¢ÑÄØtË *«|oØ$ÓÂwá°ÖG¡:V9F9ï Ôúî3 ¤7Yø\¦ó áÛvä¾ïÿxxá÷ñ×.?¡uºñ©ÃÚ,´Ç|[}ö:Boog½®¨ŸÕWë¡'oæ@É¢Ä<ôtú¶œ³†ö­ñyWäÂpúôùêõ9µ,lqxk ãtç~+~…‰ß×>Z;­ÀÔ°û½Zg`fkù²âB?Ì™5Lš€…3Ɉ§žC9ÃçÅQÁ~·)kr•†eñEÕt÷‡W¼þü9¦¯sŠùRP‹¿0ØçîÔ”©›œ~‡º_R¾pŽá‘‡5ÎŒüñ¨ID׉ ’éþ MoC£÷­Þ^Ë¿ÐDÈ”›q~ M#ÿ¼ÞÜdD³€ëDìÂh.õÀYu¸Í{yV–¶ƒÐâö‰”æ¿lh)亩GK„–•§kn¢•Sú §"u´ZóðþÂ.‡ÖvÜúsÐæ@uå‰q&´Igä“VõÅ»oTÄWdÐÔÏŒöïÂod)ÒÓÍ ã‡%{Ë^Å;? )5r|í;|g™!kݨ„rá¸Øš;PÔßVAc™ð¿ç,ÈfžÿóÜá-"„¼·©ÏÌ áË ™V#zGÈÙ_fyóéA9sLz} ó–Ú©_;<ÿxãÔ•tøŠ‚uü·¯@…èÃH‰F¨m<ráó^hú»ÞÏ`òZ«ý•HR¾@{Cõ`|²tè}ï}a¿ðÏýŸ—E˳¡ûRGµïÖ'è¹ÊókuPñßewjËÐF7èò½j{nžà‰šB&ôù1(]¡Ûè$txÓiä_N"kÿ£²ôÐÙøàðçsTÐU¨p?ù»tvÔ·ê@¯àJ„ a¿ûv}C‰´à×­ˆ©»›0èúªSz]†*®É¿Ó‘…óh2£áçy0¿ ã“ù¡=·arl‘&DyÌpŸ¡¾Gûfóß$kÙÂñwម"XI!}Ç›ËOeÌÎG’Ãê™Ïò¥£c°vÖŸûŠÔ:l|¬ä7š$ð5y"Ò!vŠŠõI"Qß¡èΣÎH¸À">HfØ\ÛÖUŒ ê ýagR#ï«ùÜG¤òù»"·ij÷‹ ÿ^¤7Õóáá¶G†”į-d:+Çùc±™—>îDª#닜^] Gdsxœÿ1E9Œ#ÿ »Õ#—[»Üþ<}©ò Ÿ òŒ(×sàB>ž´„ÌoÈo/˜ ¸ñ 2²ý™šVPHJ£‹D…ÇGzvº¡H‘÷üÚÛ86¸œ+‡’¸~‹|!¥Þözù‡ÉõR8 †QædJ ª£l‡]ï ÊëÔöûô¡ÂJ?ãC›¨”‹a1®‚¨ö°c·¨ÕšXzo"^¹ôêô©fÔä_V©Ø²@­{{òêfP{`ݳq êj¯^ߨG½ïnÚç¡¾Ï î*dD­·´v„ý6:PzFïô4îw‹]­ESgµ Óhæ$ŸÆKØZsšÓ™!gcÐ<ûçjî&Z˜sô>õÒC‹ÉN¶­zM´¼öãÝ{{´"-ÏØŽKCîøÛ_¬Akr6vDhíï«©òj­¼N¸× Ú8Þ»tkmjtÛ4îÜÅ€âÛöƒ â¨7×FsžNÖq…Ùþ08%ø{¡ˆ¦è<”ÓsÚè=އoL –>«×ðÕži>‹[yöÓ…¬êèªMµnÈÈáP´#½ iÝõs’™CÂN³«ìIšá{’ÿîB¢ÅåÝÁjHf=ReÍLŸÎ|Íqz}”øRz¤äêûŸ~õ õ<ó|…¢VúºÈK‹/lo¥ÛeûßõÂ5΀¼ðç¼6”Æ2e»Þ‡²=Žžƒ¢ÿ{΢¶ö)”³lî´0ÇByx¨åíü¨ á¶LU„Нn¿Õ?PYå3Ïy¾sÖ%ÄÝ)‡êO¥IíAPëøpõDtüðôR£^P‚†,…×®Ã<¼ü‹™ºi´.dŽ *€öÄë½m’ÐñBÿp¯Y"t%Ý{©pÅ zJËYÔÒ¿þÇCoßÓ'`ˆÕ?ÿ\f gœZZZ1‡Q»ƒiÖ*¡0ÎTò^õð6LÌ.6e€©¾[ø¯ƒ™™œ¹Tñ‡0§ w‡z l>îÐ΀ŴE]©ìLXf#}‘é+ÏHܵǿÁI¹êJ¾ý°67YìAðïöžË°Õcpb’kvhô'lYzHîaÍ¡àÇH|çñ¸zŠ’l¬MŸÏA²xÙëA3Hx8xïJ9R†ê¾8Å…Tq±âT|HkGm,YŒ lZô]¿"cÚå ïÙ¡Ž ôÅ‘Uñ{¯E4 ²xïÊÊ¥Šãô13÷,që Õ ?1ä¥Ô•¢~sùj=³4oS @Þɬè¤4úI˜úõT?Øåȱƒb­iÆ»·Q²¶•ãà–/Jû–q)ÜâAÖÎÚÚa”­'‰½-À‹òÓuZííPéŒÆ·kâ‡Q•êÕHâéT¿8•nÿ5xmlõ†Q3"žóÁ¹¯¨­ºùüÑ´êÉßÿ¹¦‚zU­$Κq¨óôÅì 4;ÄæË›„†_·h)Y5ÐXòRøÈS4¹÷Ið%õ4m—À’‡£hö39é·Ñ\ v·Ž–à¹UAŸ¶Ï¢…¶CÜê ´¨ýî%6l†–z½"*ëhYu`#à˜Z©”|•—NA«Ô»!ÞýLhÍV) xbïj<v’44qc–ô«Gæ~ÁBú¦°従>sÃgÙ‹WÓ *ʼeÏÆÖ¿=2ÿ×KQ˜ Õâ&ÕùMán»³£ÿõÒð;Sˆ6äAv‡x¤tbd>ØI;õ üß¾ŒÃ&Æ{Öÿö]TuÕ×™ýÛKÑlîRPÜüoŸÌOíð>æÇÐQ~•ÊÿÚqèr˜3˽Ý[¶q2$ÐKÒQÔðï²GA×ɹãt-I>ú|:?q®.óþÛ;óëé´.qºÌ¬ù_ ý‡Ï‡Eýûû^?w[¶ûDзÔæÕÓ¿ZØRUma0Ì–£±¤†ÅN›ÀH€µxµö]Sî;“FÛL{‹Ùœ`jÏèU‹˜Qú[÷4Uf[ Mö+Á‚½ÍK¶'+°¸°ü5ž–cö’\lUGnåLX;Mµ%Æò6ž­h…r¥V¢Qû Ø ÷ “¢þŒDÚQ|qÁHâeÛ<›†dj¹Nl—‘B:¾Ê…à;”Z«•‡¾+"ÕÕ¨!%{¤iâÊ#~ŠôfÕd¦UgaûUGò¡0d:Mgv°H™×½:Ö>‚¬áñ•‹™Í6F:­³9 ý7C¨”Ë}Bè‹5àw±L9g×gÆï橾Uä¬þêÜ“üNW²dØæQà‹×ìOÿ&R³½p)…y¾Ú³Ø¡ÈØÂ=á³(Î@"5uT%l­{y^£Tá~SÝ~ çÔâÁ^¿Á_ë‚(æ™?øÑÁ%õNÏkŠ´¢´È‹R~Ã<\JìrßÇe/ 9µv²¢¼ßó/¯~”¢âP¬n¹'ªxŒ<"Žj}sýB]ˆ†‹ƒ½]ЍIô8#Î.µ^–;ššEƒR‡Ê'¨GÔŸá\ˆGRæñWh³¾òðtxˆ=¥}Û]¤á‡Â e û-S¸qô$h¼ŒÕ…ž4Á–`Oèþãt²|ºŠ„7¼f„Îðä[¥²Ðш>ÅÝüБÞUzmî(tʸ'=ë#Ì/È| ¸E ‹wmŸÿe†e‘ë|cÜÆ°²Â¼aÐ&TÆÞ>; ´üM‘ÙaËÕ;³x[þ. ³¹†D^š™I‡„²#)Ìø’6ñõ™, y®Ï ¥¢hÜSè…²5¸w°¥è—\0Òˆk= ¹@Œtï½ãOÙœF†w,7_UX"Óªµ§ Èò.«ŽGY…¾)Û#ŒûÆ;ºeL"{ór܇È9tê¤ë$`ˆ1^B«ß,iÈÛeteúPò3ÿIºIžŸ¥ ¹ÿ¢àÔVß%~[uÃõÎ Q§Äiû°áöê` JŒ¾HxÂõ ¥”Ëz]@éúT;·”!o9ÍYŒ².*¦Ñô¡(7é§¡„ BIÿ颒®ªáeã4TÙ_ ¾Üœ¨fnröâê~„™ñ†DºÔøÙÂcpҵĒÞ;ìGí !ŒŒm“¨S´®FÒ¯‹zrÂm›CCx¤%øö6»>ý-â–^Ì<}¬³ e¯ñºµ Ißµ¢#‚hºÕ6uºÍÞ÷’é¹š9½ãš7ÜN4àèC ë“ì%+¢hÑ6û:ý¬Zýdtjæ Zæš”~v@+.ÿÀ?Îqhuí†ccËa´j¤ì æ!ð0œó—Z»~¸÷v²ý«Él…ÏI¡.—Às–,døpEòÊýÃð›NËù…MÔ®½~P&ï~tLÁ ‹¾™Š˜æíuoÒ·AŽÜÑq˜e€Ìå_CIõ¾)þ#Ú„ÔÄÍîËéz4®®aAù ¹Kya ,ÃÍ… J!IŒªè‚ñ8¤§ß¿êY° ÙtÒo†ÈA.Çùƒ®©w P°J@÷ }ª¤É›%ô•åô{š dk?ƒü‘8(Ure½»” ¥×÷[‹©ª@é\­ÒGÃ(s|íQîeΡËÙ¡$i ø3 <}sÿÙÅ$¨°(J9ðœ *•ÔMßh¼‡ª ׯ—+Â÷^^_c¨ ë"«¤ƒº»Ã’W~@}!Ñõ/† Ð$FS:¹ÍÓÁ“–ZËÐê|qäZž=´Ó9kþìý?W¬NÚZAçÊà–þámÄ+øú¨úÆgÉà×óÒßtïÃ`ì«~c‹{0,áz%ŸFÚçùN…ÂØ½øÜà°0˜°÷šn;BSô£+/æ`FÛýrÂnÌú¿ðQ´‡ù¡“·Bõaš¿Ì ƒ¥xòc¦ºG`…×ö³VL7¬­¤†5_™·M'caÃìùs’?°¥«êb©b]ž~Úìã€Ý„sy–&ß‘xßúÛƒ>$¹Í>ÔäöÉøHÇ8|AŒr‹Ø÷ü.cmDªýO²+ï„#Mس»•mHoøÊ6| %ŽOȆ#SCÀKòwcÈòJ}ƒxª ÷¥˜XÿúŒœòåE&1È55ðë²òòPºù ‡N#Ÿ<·‰B« 0pìµä¾‰Bl9.Z¿½ñ ¥ÀOîŸK(ÚCªIÉ6„©…ÂOI&Q*}pŽ‚Ï{–Èó4=AY.î#=á»(Ïù¶R–U}çA›¼•&í-ßñ Úñð.âóI'ÎKÆÄ¢FÁé…Á[z¨e4p¾ŠÊµ·6_p¼¹‚ºŸšŽ®·â›ß<­ Pÿïí½ÖC5h¼+îñ" §s³$Ìç¥/A=þhr6詈š>W¬T,E3keQÊ.hÖù Ôù 'š›+ŸŒ°Aóò›ÞPP†â…XÊ·Ñ"4[ýLVZŒŸž=cXBà!µ1áøYÞÝ.6¤MÄÛ:îö67ÑðøQφ‡Èœi?ª ­ÿøÊA»LöƒòF¨Œ“È=ö9ʈ.}Ú|Ö_»xw‚¡@2—Ñ«ù7!§¾÷Úlü¹´_ÓÌ!§k§ê×¶!¯ê§ß Ì'ÎöÞ^‚ôTÔÆ!ùÿë æíi‚BƒEâP^òžLôr Ô:>)¿ í M÷•»BNîƒÖÀ¶×›B{§ÑrtPÿâx[aIç¡n=n‰‡ž#v·È ¡Ç²¬Þà:%tÇ­9¹føB×}^ׂ€wÐyIžiÏã#ÐñíÏ… tD©ø^vU€NzÚÙwÌÖÐù™Ü<þG tùD¸¾“‡îÓtdÁ"ÂÐs¿™å©yôŽ>¨}wìôÇÄPŒ˜|„òQU¿æÉ¥Ÿ¤”†á(Û/qZž0*ë$Ÿåc ¿G›ž…‰Î¹î¯îÕ0µ¢ÍrïçAø}Þº™>ÉæÅóÕš~†E¶/ñ.sÁRö”¦­×%X¹ZÛnîÿþ¸{632‘Ãú=㱉⫰Ùþªð° 5üµ: «ŠDÌXÍ˜Þ…Ä ÖÒ+SHêrÌøØ½³H®¦æÁ`Û‹{3}o‡{NõJg uÈ™K¢ƒ>H»Ä¢ã;ᎠÚáü÷íØ‘1aO̶k$2óûþP9sY¾ùd蛩â>_ú WJd·9c°çÓ rˆeñûûcÕ'oõ"wñ‰´•juä}Ô"}éöS䫹[Æ$‚<ÛKK%Pð­ÇJËÅr6ðR=·8„""7FÉŽF1Å…è#ä(qöçš&;?Êßíôn!Céë7mBjñp§©*ÊòHÍÉ7¡\¢÷µ0æ)T02~ò}Ÿ *Q,ä+âEå~§e»Ùs¨F­4Ë+Nð:–³×5Ž÷7Ë£æŸQ\Úʨ­QbŽ:wg®ýN GÝM"¦ÁH<ò’[áK˜4uN°Ý°•ACÅá°*v 4Ú<)mã:Š&Þê ÷"ÑôGÂU?4  JLtAsQÚÚ3©hþ-•Û”Ç-ô~ë zt¡Eu{³;ZbSCO5Á3h¾w¢먽-YZ]î{f¸¯­êµ³NާxøÈ-þ)‹îØÄ{sožzÌpso.êÐ\ó`zëˆ z—4ŠÅ`æj»Auµ/Ô¡ºflÁ ¥ÝÜrË…„º¦o*ï _ª¹>ÿ-sƒìDÞÅÀEuȼè×îœyÒÝ׋ÆTCJ»6Wµ<@SÿN|$x± sÝ…Ÿ&˜”®þ„Ä`焸UH§ìíšë܆¬¯]mßßpÀ—Œ'j÷=¡àƒý¡\(¢¢Ú}ôн¨¼W.‹B‰ó(mÓq(©§yÚ+ì¥ïs:% 4Ψ¿ß×ÊhäËT¨'þ—Ÿ ô@Ùà^ÆçºsP~tùÖóûÃP>¼8©KðÄ'J™«>P¹LÝ* ßeN´‰lCu/s£ÓY¨Í"!SÞÛ?ÃÂíBãAê…W73 ù„…èÑph©3ý#áöOñï©RðS;áÄjó !çÞt9Ý’ŠïýZ€^åÌõÈHèÏ<1àíƒâ’^â{”a(Çš¨!0FŒÍ›ÂèŸýQKƒ0^åuýøñE˜L~ ð¦Sí¥¢Lá÷@D…Õï]˜WñIÜGÙ cn&'uŒaI÷d¸‰&-,)U¿o¸«æ§œ6faÂEÙŠâ1¬wë¯^壇Í:á°À§‡`{t†^¬CvÅE[3V¨;ÑvÖ”I´U/Ä‘AÒ>ž®²Hþ®¬`~£÷„ÑùÛ–PãÞO©¡êO‘†Áø¶¤4"Ý×üÒ€ }BœÁKÈäks´D˜Y”55~À}èyx²ð²¿©Q¦6¦G.«=ÿüƒÜÖV1eùdÈ[ÿ;G¾´ ù_›»ÖNR¡àK+'!®«¯uÌ)AÑ£Æï=ÛŽ£#õ‰á¾”¢z~¼:Õ³úSæúŽ2ålRÓ(W˜ÐÚàxiP¡¢3•¯Ïzo0¡jLkGÁU•š¿n”¨qQG†;žµˆ·ƒŽ‹ vº\Sœ´ê:”OM†OãÒ(Y†ÔO™-0’fBùž±?hÛ^øIG惾5¿\@©·góB¾¢©ã(úU4Sß9r~&šUŒ•ÊÎ]@sM®<m 4ÿÒÞ`*Õ‹ürôJ ºhq/lõAmZŒ˜ÕéNÝCK¥öÝ`ã~´ ¹mëÿzƒô,§¨ˆ\ñh‡ëƒÀÏñÈ5Ç1ïó 2ì‰_„úã?΋:A¥ðkQ’ˆ (}X£pø¹|e8 \» ù®>Ïwà…¢së_N(GUÞí&äU¡‚gYÙQ´C i‹.Jn…½f”Í$°²â­õç<†‚—/]Å{û ü:9Og³Ô¬¼}9Ȩ ›óêŒVæÐ*tPS ÚEOð/ÂÏ̈÷’“Ði´As9 ºïŸM}ï ÝSÙÕ{ yбšþrt3srž‡.KŽfk¼tZ—Š…ÛÛCGÊxµ¬&tÜ}#q¾G:f¾Èå2;AçMÚ49ù(ÕÚÅI¡›hÉ–Íl zÈ–MsoC¯=mea”3ô“»ÈÙÁË.ž¹6ƒ•7¤ßØñÁ°°ÝfXYŒdò¾æ#Esv¼ª^} &p…j>%¦ÌÇ §Ù`¦X b¯s!Ìy¨Wò‚ ,¤2ç˜8À’ú1çÚüW°¼aÇsVÇ^Þ ÿk»×¶U¬;aS§¢âƒá0lçÚ„%ºìÀ®ïo̤H¬ÝnÛú†I–Ÿ²wM Y9¥âÄR$ŸÐv¦›BÊ"ò.‰¤ ·¬7@Z­Âч×ܾ‘-|¾;uÉôÜ󵑩Èh4%§Y,lÊš¢Ìp¹¾¦D¶Þ_‹5ÄÈÑq¼·’6¹b-ïìi@îƒMºêj.È‹ÖÓÛAÈwsý%e˜òÿÔù¢àb‚fíÚAMPhaëºs¬-{FtqEL4¯ŽBñUÖ#+,xHI½Ýß+”>¢ûîÈU<¨ÈQ'eú…Bl3¡œû“¾Ѩ@KÏñ$•kwéž¡òÛõ$Õoå¨Z%àÝༀpÔÖâIâ&jøN¼ºÜ¨YÆö R̵‰ßxGÙ*¢Žá¯/7úP·¤»Flà(qÈ>O!€G…F–]¤UÑà/Kž´]•iè׬¡‰’]’)šF_wb %A3ß+»f$hÎZX~ùš§¶EZ ¶ …’ï¢Þ´øšö’NÎ-åMÍãÂÐ2щKk× ­ eó’Ðê‚zuµƒ ZÕÝz¤ñ©žÀÃ{™{? µGÛ†q¼AÄZÛòPëoƒŠ€0 2øhåQ¯Àô»Ü†Ò ¨yág½î6¥`¸t7‹ öÉhúrxÀçåãaß.®ýóÞßíòUOÈ–|rX´ÒnäOŸS{)D9Š ZÈS5æœ,Û¨€TêmòÓ0˜ÖþFÒ\¥T.þ¡€¬3ATi2Vÿ¼'#‘ÅÍ øæ«É7^Â77óù@[(únÆÑ=xг>æÛ²@ÉA³Í4$()ÕoöW…Rëæ;¬ï•Ü±>eúËãeÞ@YÐë¯3P6ö{ðï~(¿$,KÓËF|U)•¶áo‡¥ ê™ßÃ_ Z Óôåf-Ôôóõl>›‡º>>/7=h¤æÕ:EÈ˱£GvÞƒ9IÊAfRh­»õ÷œ´?]?ö¬,:|xþú¥Þ„.'ŠÆÙ(mè9}7à‰ ôu˜&hEÊÀ÷%Æ«0$þ]H†«âÕß`„Q·ç7oÚ†qi¯=#>í0ÉLrßý´"L3<Ê—é…ßš÷]úòæžÝê•v„…GãH‹ã%ÕÆjm°lÏÑK´-+S”Õ 9YðçEX{WÊMXwùëòE`Ó”±(}š ¶OkŸ·o¾;£s4¯ï!ÑÆ>KK$!s$¾.Ò¤/U“<>Q#¹ù1ÏÑ%=B¾uÜ|Z÷jԸ得@êˆ=~Ñ)H'^·ï`Ïdx4g\zÑ ™x$¶¯¥!sÕƒ€ÖIdm¿GGaú Ùyì3Ž g±ô{¬;x N»Âaäòº~e—Ë8‰ü‚J5²^¡à¾•ê'>£°Ö•¢˜ˆÇ(’ÿЏáIŠû:DÊæ±à!#µäPºT¥ÏäñÊœa`›&kE9ûŒ!DBn}&¥“,Ê$K7Ä,(PÕ®½!Tճ嵈û †8Ï%‡ÚfÔÌãxºhÈÚŽç÷ŠUj£.Unh°( êeêéy½òD}»cg´n9àѱ\‘94´ZÞ'V/F©a•<Ö§Ðx)ÄkBMEšÜb³ÃÑL¸Â­ƒkÍ25WöxÔ¢¹ôîÙÞUR4O–”tú¸‚ìýÑ"Ð;=yH-•ùCŒ<ŒéÙKU–÷,dŠðb ¾æUû#Ô7H’ÐDæ;÷”ÿ À‚‹u:ý±×Œ¬Â "uOtêyc(eÝ_túù/(ÔçØc¡…¼/jæ/R¼0ûÂj&%|N«¸Ú«æ ÙaoƒŸ‹“@&‹אּ¤†ß¾=¿T ©¤öä¾Ç ‹M"ó«³a®%\¤6C¹ÌÕ{äPó0ÍÛd€ ½ãßGë…BK³øh{Ö8´Õ}Øúî~ZëŠÓ…ŽÅwó†{ºnkÛ×^8Ý<{Tï×¼p«¬ÝGºé“Nß?]Ü̪‹ß SÉP›ã๼ðÖagsè8ÿwØËŒ :jnÔ3ŸN}Mo“¡³%€•ʧº¢:’“¯Ö÷R¹•–Ëzš"JÚ»F ï¬à©/.Ëð˃SÛŽO5ÌÍ’˜9Óý5MaDbh÷­5Œ65SàuOÜP˜S‚ÉŒûŽ~»{`†fÑ‚”ƒfã* }bü`.”èHÀbum y,È?t,ó€U׺øuz X»Ï:«ŸçX/À6»AXÓ­rØù~ušô¢  YJ]ÖÖB’àºQËÕSHv¤²$ô8Rðþd÷¹y)÷³‰þ%G*Ì“;Wb‚4Ï\îÆì=…ô¬t}M¤šÈ0J4ÛB½ˆLvü;œ;’È<©Ön3슬Ï[«ôÍÅ*ê¾e&r˜u±_zg‡\αg„—àþà¬`äûbHc5Œ|‡¯?1£qFþ‡¡ki$÷QLãéá =(ô!÷ë&O<èµGs;Ô E(â˃QÜßêU"C*JæÝh uGi²§•JöŸð°º'Ó­|9”‰æw½hrBþBbF(ßÐo‹Ã¨øðxDåù=¨ìq‚;” Uhçü^ô ú:}$Éå¨qäâO²4 Ô eô£g¼ˆZµ‰\ýì¨ÃÔ²Tjxuƒs´´úÿâ~Ó6ÔŸá{R»W Êõ~EGé£Ñ}3UI*4!½¬ÁÖ狦^FzßF¢™sÅM×õhN¢|5€Í_oÓ3B 1¢·|¦ h‘Ê ¯.£¥SsÍ-z´ŒÍ º+8…VTÙ´ìF;hå³­ýžð¹¦ÀÛwžÀà­¡À´>sÍY(~¯¿‰¹sP£ 5çùie?"ƒMòN±ý0m°Î#–¬ÕcÓ²®@I°ÿã¤~*È×òVyó×>óD{Þ|Y¡M”d¯þyßï¥wKóvýzv3$K´½;«+EðBÎIÛ~ˆ_*p^:WñnÏŸ_Þ…5šëÚç µävô =a.E5#jÂásoýøo-%ÈÌ>®Ö _K_ \^Ê…"¾Èª£>(¦™~Tü«ŠoGžôOjû‡‹#O¸¹¡t¯•ˆq=+”z\£ù3¥UoNÜ$p£L ŒS$Ô‹ÀÅq­‘<('µ¿z÷V”wJSõ8ÑAEÏ7RKâO|Y¡Bv¾{kœ­?ÓKn=›t„Îð+Ë„9¶3üõyz·5]È‹hà×ÝtÚú!5üRÿÈŸñ, «ï¬Q´ÃÈO§«·a,Úó…Wp.LܼÑÃcÖ S7Ô^jÔ{À̇ûgžÞ2„Ùßé ‰§aAæI€r5,Þß¿c¤K6çœÅ†a%ðÓ¡K~dðG¸Øç„‰'¬Mg5TœÐ„¶í™^Í Øš‰nzX ;rïöžÑpG"'²?lø*0„ÊI-l&­ 9ñ"W÷tR´¥gä0#eëJel"5cÙÍš=û6º~?ýîudPÍ ÜŠÈEÆÒ é@dö[h)®¾ˆ¬gÞO ŒÞF¶÷³B‰GSÄNžŒ÷ÏËL¬Ó) ÏÓ¦‡ê4ò}*çÊ»!Œ/"ý^¾°C¡Bî¬_ç£ÈÁá“Òæ(Ö›^7B %+¶ J\PÚs=clVe(w‹¿¡ìfþñ¢*( ¬ò›B¥Gk"S6P¥Ó·&Èxª‹¾ãä_ìB¬^=#? šgâC¼î¢6}²VŽêä§Hñ•q¢žÃò±|ž Ô'¯âPªOÀ£‘‰Y¹lhHÅx!ÑŠ<Þþ8ÇÔÆ_ÚíyõRÑd™KFêü:šÑß¿«Îe‚fobö ñ¡9ïÿ«„n4üØg?ZЇŒýD ?C·Úõ&´èç½³+ÄLàáéÏ¡ÕÉhúøÁð—xë¹#µM’êÕ¿þ>±÷2{óÝ}ð ˜ÍíFôl ¹Œ‹CùOTH]¡³2Šƒ~€b¥P¸/0Α r×w/\)JðB••eÞ§³ml½¸< YÓ¿ÇΕ[Á§ê‰=ùZ!EêCƒŸØ'H.$áòṄdì÷?æúy™£@ÙLƒ­\H6Ôð~Õ*J[‚†A{ !¥jh¹Ú¥›ÿ9´=Ý&Õù Ú7Ö3Md+¡ã…¹Ôƒ·¿¡K`GB“º%¢9 B ÛàÕ·(¯AèúNâäyy:§hK@'gÇñ÷+AÐòDÒö±Ë?=¯Ü¿AÇGÒòaæ‹ÐIÃíâF‘A«³¹1Ð%S¸­÷º¹ÏeÖ‚]ÖÂyvwèmxÊíõú äèÕ&ÀÀ:ÕèÞ…=0¤qü®gi9 'pãHèƒQÔ|k“ãô;Þv‡`’Vœô ó˜>^´Ýr ~ÿÒo-݉…ù±œíNâ…rì'‹aX&™qs;e+#›a^ûáÏb“ÈÀÝHØâ艇-?[‹¾;û.³-û'#Ñ­Ö=¬Óç‘„4£EbIb»_ ¿FòWF³ÓT¸'øq«4îþÅÊx¨©G)³^ŸD: ‘OuÈpûo¤ªS,2.m°=Ò'ÌÑ-þñ_â×U à†‘‚;î¯ÓÔO CöVÖ²ÜÈÙ{+Ÿ# ÷of_iNA£âRîíräÛ ~ùÀùñ§ ߇\ÈKºàŽBf7ž€¤1ZXÿ8‚"Ëj–ïúPœüÙ”=Í㸛.¤ åÍ¿ÒNþ$Ø¥tîB/¿,WÍè0ôGwˆÞ7xƒŠ”fíÝ=0”{,üÇÓ0¢sWëÒþ6£ð\~ý ÆÇVÅ»”aräP\­ÛÌßù–¥·³ºÑÚn·|a>[Oêæ‚ŠúÉÊ9ÂÒ³>â’cá°Ân™©a«¹³ÔCKv°vÉ]yê†=lØMrRFUÀÖ¹©ÂÊzø[p\°j\‰ør‚ ÿ@â{øøs#)íe’iÍB$˾_ß8¢‰Wþê–#¥ {Rq R=½áÇzñ6ÒR7ò­ š }'ñi‘ѧÇeæK2surüýº,;Eß5¯_D6Á€èU}gäxèðr‰÷«ñÓÒuGiÖÜ;þFçñpñÄÆµ(ûºlçJÊ×Kñê½A%–í7Ĥ³¨bÃ^aóbÕÞS']Î×@TNÉËCM>ËܦQ«tä¦êo&ÔqÛR|Lõö¾ó&K¼Gr¤k¥8𨮷ðÖ4(‘jè#G#ÔÖoÍúh|m¾¹=AMŠÜwJ$Ñt!ZW_^Íîoÿd ;wYÛe yXÿlàS‚Rnˆ'W‹Ë¢¦|Ñ¢‡Ò¹·¼›ÀC™0†»OÐ2ŒY€Uë5Þ|¥|pm[u|{-w™mò®7í£ù‡¯÷Ò³ùüó¾êqžLPž3øWt Š-½L¯„ü|fb›$QÈU+¸ÎÌó?/\ˆà"Ìg§eªdé©é=–é¤ /~쇤Æ*v²“/ É'E¸Èu26òˆRD@>³Íq%Å(ûxøâðˆTçÖ±Kø5@ƒ%½gÁ/[xî¦I=Ó‚6:É”(khISüuL:T‰Ï×ÈDCgU"9Ãt5öߠу®%ùË(ø—DõUÚ §?]!äJ‘†‹$2(ö¸M…v¯,J|Yû‘2a‰‡vìÙåŽ ¡t^3éŸ$ʰQœŽtìAÙ¡-UÜ™(ojY0yæ*Ò~÷2•šE¥òï£b¥¨²4<ÿ“$Õ­9¦iW[˜ÞØÒg¢&?ùq’TÔ:öíšë)Ô~—öÙTê2º+\ìuG½ô òhÏ]Ô¿ò¦êÕW[40J•¼È©ŽFû¨|xV£qTêùäH[4¥Zbm(ýŠfòŽWùŽ£YQÊÐsg4¿n´yJ^-hYVO*øÙÇŒ”ÇýZû|D ¾˜ox&¢R|EwÍø빞᰺>SR»GoS@ÞÛŒo Ç¡ðªp‰£¨3|íâ;ðP(¾U\–=?§ EÊÜ2¼â"P´ê²÷óãQ(Ì»©ý–à±—³(N}ú‡‹Ï$ ¾Têqÿ[ëÔ?×?÷ÈŒÕ|±‚èRU!·ô“Ö¤CyIrŽ“ Tä¶äÒ¶B‘ ÉJý|w rB‚'ò¶îß~äu\”lW ÞLd‹¤†GîýQ&…æÑ\n¥:&h œ3ò~vÚEv¿Ÿ„Ÿ+—ŸñÕt®ìŽ´‹B¾`G(M>ôeïêj¦€ŸûÇtÂaèÐg Cîc0œ³-t.PF/œÉ¹‘áã¦çü#ø…`Ò÷“ùî-˜ö o¦¿c’¯lÃÜîygWÇ»°P×yÖ@Z–¤‹X¹ï«ÃrÆúö,E*¬áJt¡¶9è£.Áz›¾üb,l.…MoÀ_اÁ* »5•¥tH|¤*s­¡IjÂ¥Å=E²S¼þ½%HÁn4Lwj÷¬q/ñ5D"³üÕ“ýsHs£ùààQA¤WxÖ8òn9²X>ªL Ó·¢“j™÷åYwœÝIÜ—ÈĤ­×ŒìÛI7yÙëãáC®žÈmgôäää­iýH§…ü‘ƒÍ{bQ0Ùçµ£ØøÕØ¯*|¿•ŒzòÜ!_-Ôèð©ÎI3C­‹¾Tl KÄeûECÔÍ54—bÇ#Îo® QFýUýJWZÐಉȻS‹h8PÐCýø+«Qy@þMB²˜L×¢i¹Ðí}²ÝhvÅ&,¨­͉Rž0Q£ùãJ-4ßîÍ«ùãŽç(:¨D\ìXPÏwŽEKÅí3Ï5/£eÄ«oGš:Ðï ‹N”ÙÔ.o!3vbGf Óyޏ ˜›‹p¶“ZO粊P®dºœ°Ç å¸$j oþ6Õ’"| κy*à…†›sÄñíÂÄsr2ß?ú(°° i¼Ï®Py@¢÷øÉ'¯!ñ …d”ƒÁ?çušõÌé /bÆNÍæ=”9F>‰@¨6êšÖO†úªÊ&åêÐ\}˜¾ŸtZ3–n“„ÖC»¶Sɹ[ßàç@V&E5Áû<. ñB—OäþÎFrèŠjÍ8µŸàwŸÎÿ„Î[»·*·< càwhΓ$èp•ôa‚à‰2d'²z¡ÃÑœŽùt¼‘ÚÖ•Ã#oÓµT ÓõÎ{§5è’è:§{´º¥d ¼< =ï¼çnM@ß /SAN/øåÁãÞ+UƒöŽtº%‰0ôAXšÉÀF¸+Î'>†ÑAʦ҂“0^ãc|F½&7ìŸ/hÑÀŒµåKñ$Ì.² ¬ß‡Ž_Og%a‰Ê–ëb¢4,Û»Z ›t™,t ¢WebÖ‰(¶éa)ù5”—¯)”ÇC×Vœ÷¢Bé+$Û‘Ö)x¸}{äÏ({òyÓ{”g`ÕÉÜ @…¦],á.¨ôîøï›,¨’Öàs[) ÕÉ”Nô¼A¼(Êê?s5~ïÞü^ƒZ|ÕÂI':QÛ#5Ù¿€u:1óa'ê]â(°ZF}•"ß û­úíóѰo%阻Vï¹x¡Mz*ÿº´š OêJàRšÕzK‹æ§¯\8­ò Íwbùn§h£Å#±°¦ÊhIwÕZ’à‡÷“>‰¼Š–ãQ*ÊÇÐêdyTþÅwhUqf)¯É†ÀÃKýj„ãd}ÎX9Ýy¯\TÊav@õ{š4u!Ȱ·¿N©ž&CÒû‹·Ö¡ª]¹>þ‹(|{w($ÄØ ¾Ts 3&‡¬xûÍæ[3J'Ÿ—Ý/Œ² ÜŠZzãáÝ‘.·¸P¸‹ÙûÕ'm»hfÉü ÅÉ–s)gìðhßtê„VžÒEÝ¿ߎ_bB}£í·}ŸÅÑ 0õÛˆÄ-4<—QjFȵ† ~'~Z¡ÑM¿äÈ»§Ñhùqä,Q _ñÂr4nîàÚ¾Làa…*ëö)4 â2»}†ÝÃçÏÉG·¡Êñ™Dϲ‘ñðÄÅ‹¢«0gì“!²?RHïM IBqá×Ëú#PèÙ›õìH äÒ|WTƒ¬2ÍÆ{Y²„\Øâ5írÒ§¦¤9¼ ù/›ê}$Ä?i¢•q´ÉlM¶Û»!éfò’Ë™ÂÚžC½·™’›1³®'àû|¨/¥fyµ²·•îq‡†ë½7ž»†A“QÎɫㄜ8Á¨¾#­Î-¨Y­A;ÕQF¥´—ÐÁ ×›ýª:”“&¬ó }’ö¬“¼>´k­MÙ üú·ß0VѪڌŸÉW3ùCÛþGù'UA›üz?#áú¢ŒóAeO¹ ûPòä•å¥ý´„|I«G’’Jȉ‘IÇÛæ ‹B|ÿI?èN}bÂPR½añŸOÔ§@ÿÏoq—ïJ ÷ø¿š0älîÉ#štD+8`L€aé‚qLh šMnØÀ”§m7-̲6o?þóÃŒ[Ô®±°èG³ê¡ÅËçžëWPeÁ¯3†WÛµ3à·Ûû¥>F&Xàðw\l€-rÛÈÒð(ØNùÈùëû4ýLaÐl6F’³¤ºûBÒí,‡ ‰BÜUt#¯Š= )ÒŒ³×¼v#UëÙŒž5ÒˆÍÐèÄ_Dºïã,+_9±ôÎ…žC·[¹\É ÷Aã×~ÊRd¾%,ûì£ ²ÊWíŸvFvžðïO³ž"§ |z‚ûïws›¯œCR¯«×‰š÷l ñƒ§ÊÈ׸X»è´ùÍõ ’Ÿ{áÁ…ýf–¬ñPjA¯Â3ê¾A\ˆÂY7Ê«\nã‘e¿OkDó(&!õK㹊,ÆVèØà1ÞwmøQ²¡­Þ#Ð¥_‹«v.´ ¬ÍŠX|•Ê_ÎYûr{ŸN|Y|œ¸¦$ƒF ‡õ¢÷/¡ñ=j:"A4! i‹ž¡GïMŸü-4YÑsyß{ MÏø)™+¦Å1µoÞºxè(huæ(š9N>8ú‹]éŠm½RFPÁv*c:j÷ <ã¢$݆ñE¬\pè‡òñ ?Ö  žILà‡4d-Ÿä–õµ„´†÷­nS’fnÏÁDè·5*f]ø4|'»Ü(ˆÐo+ñÙ—öþ˃7¨õ* Šx³§•!²i>$jÆ¢ã/î¯à‡x‘>Š—’$o¨²õû7t¾´rh—Í­>È;’®aC ù~In¦ÛªPàñ§›çOÌ".I¾KÈ‹ÎG»þ$À×”òÛ§çéá›Âº¦Á³ð­êÈ낈:øn~Þ¼ñÄäßï*‹ÉÊ ¹ð ÑÄP6¤Ë»þÔm¢$ô§iUÄ«žâ÷Ù‚˜$ÁßÑÔÆ¥`¤=I ¥ŽwNS@ö+fcÕ Â¼Øx¦|võ T¸¯†¸Ø@MâÑ=âLh`yÖ.3ª Ëžb­ÊyÐü²â§_U´f{¸ð Ú¦ˆ­½Ïüû]4ìуÿ{^˜|ÊÚÃ/ŸuøíLKf[þíÓýZ.XmJ=“ ÊÐF9ædú1ÚX#úåßC›Öä'ÄBÛ‰F}öº6h ÖU¤Öí *ÂEŒ¡ƒóà'Kgè¼¶¢S ÝÄ‘ÄWÅ §íØd’£?ôM‹){qÀÀ‡Ï›Ôs¬0ø‡bÒçí{Nc”8¬Ô£¿¼=så Œ¯² gdÁd­õÉ=q˜ÿ}ã!ñ0÷{E~žn÷–4Þ›Þ Kùrr>ưÒvçÁá!X-ß2”Ÿ†u:ÿº Ø4¿XŸ™ð þL¾¯‘£SE"›±Þ êŸHϧh%ÝJ”e‹Fêg]ÔÒÑh0Ó£ôu$»L¾{ù7ò-9* \ôÐó®ðF“¹,ÍûÁhjõ˜$íO9š~ï»q6”ÀCû-:hæü¸:ã9º$jó6k; œNÐIš¬ã¸'ÙüS˜ãŒŸjÌàí­€²ìO7z~@~úzGÅ bÈâb;ëlF˜;¿ØØÉV@r0™wؾ¥9o®¾âÖÕ©æÏA,µâ å,5Di¾4Ëž\„HµÚ;ußú RTýdV‚-D“å“o?5†Ï>¾H8)«#Ÿ9o@FµÀ€‚û34|y ×gó•ÂæÙûyåíà ¿U%eýñu(О¶ œû …B*sR¾ð•¹¶¬Nµ¾¾U\Ù÷¾±%T:¨Á·÷g¥=´ÿÛheb>ß?ýbs Šž—…(sæáÍ6j6(î ½:%½{³ÂÖ¡tt`øaÃi(·~Èæ›ž • äp¼æ9T]K!ŸÏ€ëì%]"¨ûÓ¬:¡óo~AK'¡q°—Œ­šŸ Y˜ý–…Öûw´œ²A{Þyë#Eè\»â¶ =·æj¯üæ‡þ£Oÿ>T‚A’ ?mXa(Ãö,ã/Œ¼èr)èa€±« Óµ|0áëL%b÷¦â[ëb`fljy¬Àü™>>?õX¤°1PdR‡¥›“Ñ'c`ù÷'•{áW°ÌÔýåoðÛ!¨éëº$¬{š¥ÝN¾›µ²^°mÙÓ5ÓéŠD/ŽØÜtICâõ‘œy;v$u÷*—A²¿ ÉoÏf#y²ÛÝ—ë‘òÍGq¤®i“ ɉCZ3õ3*HÏ$%ÒX}÷†;ɯûDF“#Ò9ubÈ,ZåvD…Yxº Ž"»c_ØßÑläç›9íû¹JsªZ; ý›»ÁŸþÐ_½µ¿ì,45WÝ¿œ‡ÚN ²5‡ P çiç;xä¾Uô—€p3õa³ÑE Ž1¹]6 (ÉëH¢x¥åè{S® ¬aDJâ7~”¿›7XÐ8„ljÿжFœ~µ6ÃI‚ÊkÏ[+¤à .‡Óby¨Zhš/]^ê¶ýÏ(MYQ“¬›÷Á¼¯À¼™ƒ:ŒÃó¿+˜P÷ñ™Ä8þ]¨7,¶}€ ¤Ëd½f¡¡üU¡-lGóTMJ4:ƒ×Jì_¡Ñ@çM)fDc«ðm{GF4®ûÔ¢/·LàáÉunw4ywøiÿ­—èú)ö½—ÎM[Õ¢º¯Èð‡]]cf9n]Wq‡J)-ö¬ßæ;þ‰ÿ|_Bv‘j8·ã™IŸ}8f‚;šäËËÏ®\ÞñaDëh«QQØ©‰þ.È<øèŸ'Ñ@“À‡ÿ<ÿy*jölµ-ÏA½§ƒ[uâ 4¾Q¦„b;¾š–º§’ÇæÐ¦3ïi(ºã™i÷r1[Ù¹ÞñÑ<õmÓû5ôïüìw,×Ó uESÀ1æÿ·þ÷¹Íÿû½-9·…C{ Ë¤ðÔcôÐ~fÞã}$¥ùûîü…v{›§8ƒžoöDÂ0Û…3^ÀÈ ¥¢…g,Œ–¶÷^¥úŸ·æ’®‚–mÌÑ´x4ÇËÃÂ÷¤'/×aéìÔdòyX1ˆ ñôp†U³Å]%çK`í"‰‘°lÄÙü‰ÎÖ„?ü!·=O‡Áß14~ò(‰íßÏÖîB’Ù–fÚ^$ ˆ".eØ…ä¶îÇ­&‘ò––££R늡ں‰´ Yad»qÏ’ÁwYÜ{5òă~d˜o)Ÿ~C‚LX|¨•Gåhñ£×ôÛÈF>UXq9¨¦e¬qŸÁ•§*rulg®¯2#O5<¹V†|²ÕöFŽð@ê;…[r<(o5zð ´'h~]óCÁ”ÏŽ Uʘÿ>þ'E&Je âòðèI™:‹1@qºž¡ß2c(ž$´yø&JŸ¥þº¥™>p]L/@™¡ƒïmP®ëåÒ³. T¤?w O)ÁÌÁ{éîoTÒ›Ý7幕=ÆÅoDýD•ÂÇ6 S¨*XËÕÔ&j¥Ã-÷Õ¢ÆóÀc²Üž¨uîF ÷>AÔÏH|ì…ºì?ë(ØQ_ûx9|ì”ëhAC¿eß/ß™ÑHj®í–Ž5ÙeˆÆ—«¨mÂÐx¡»óu4!:‡ï&ß…&“œOÌBS‹1™#',Ðô‡·Ñ¡¯^ú¢sÍnð1Jí§Ag¦nÁÑe”=þ÷Hù;Üs›_\²{ÆÊ.|0ÑOÞñ¹äå÷ß¿ôµv§ïþóÊüç¡ùÏ7G³Õªï+ 1ÒE­ÜH ÒµMö“}ñì#G…­×N²Ù³ &zo§Oÿ¿žšÿúáSl‘$¿9÷?_®Á‚Ž©)!ËXDÒƒ÷”ß~ý‚d‘/ÔË)—ÿç«aÙïòâ#R+t w'ÒdõÚ~ 'Ü7¬ßYÂíp¯’ùCÙ©È0¡ÅÐzž™Ò­¦ï( K¬æ†jˆ ²Ö•†Ò×#'Ë÷ìSö?q‡à}"Û_Èt2YiJù|˜®ž‰:‚ü¥1Þ©âxHá ›]r ^8ZÕ¼é"míÚ#F{ðèrဒäUÏŸ×ÃcÙß6ª/q£T ó{Q*ÊÔ«Äß§XF¹ÕƒN´!Ú¨è¢èÎsѬz›ÕOT–!«ƒ*µg·ëŒQÕŽ¸î¡ž9ª³èÈnGC¬ìµ_ꨥYð0µë«?|Šºš|§z¡^Úàß;4 ÔwÓ5FCJëú¼£hX$w‰¬± ßÖ5 Ä£QûC›“ÜOÑØÌ+ûù“|4®~•.ÔžGàáæh)4 Måe «B×C‡˜n…]ÅãTHݸ]áÙX0ëne˜‰î74¸ ÉóŸÄfµþí+Ò¤ÝñÑ|¡îú(»¯öß™6ûóÓ¿}K¼Õ¥/ Éúd]@8$*(ÍwLÿó$þ­…¨ÙcÚé/êvj"GÓÆù·fð¥’éü•ß;ÄòÑæˆ•w]P½˜nõ&Ò¦qöÿ|…w.>¡…¦uŸžs2óÐr_"€.`Z§ÝêÍkx m:׫¤ÚQï·ì€¶?Bz”9O[‘´0„ÀÍ87¯üË…Å:Ð:á-àŸ¦m¤$Fé mÿcv=h“ßHÉ: mÏ ºsA»È›—téÐAþè~Û*èÔó9ú}¯ tÕ¾f»d¦ =^ßí§Ù ï;IÍï"P¬É}á– ƒw…æ4WÊ`˜ABùô‡0RndWÔc_~z|q‡‰rKåM0ÍÄ*&˜@³Iwò´©Ó{ŸzòÂ3š”H6Âò²˜ï§Wà×T¸è•ƒ'à÷|ÁóËôY°q˜û¡¬dl} ÉÑÿ0M )"¯ ñ>É‘H«$ u®»,…dpÇÏљɕ<*µ‘’ÛWüŠ[>RŸ}Ðs'¼i¾¿ cˆÄ=îå·Ì—?â^AÅ·6§‘!îÊL|!+2It}$5ÓCæNaêû¢ÈeþpŸ²ÇÄ~¼øõr²ÜÌÊE®Ëj¸åÙÇïˆG¶Í4ònî¾"Ül‹œf™U³ž ÿRWæ-©û(ð¦aÄš’ íjîSxQ ÝvCŠ|08`L„G©]HÙn Ø¸é1‡Ú”pßÝÒÒ½„’‡}â?m–¢Ô»ø´šʤ*ï!Ù@¹è¸¥¹ TèØÓrùÈOî$Š,Z#T’Nm³vW@e[ºWÔ·ÕP%ý5Ñ%QIT=(@y»ÊÕÊÝ£ž]0E—ãY“ä/Q˦Ȇ“U uŽˆ_O xºU’v”6¨oâÏ÷´w bô;[ #l½ ÑÈhÒ-ä mÕ…¹,—¡qäÑ‚'åCh‚Á=g•æÑ¤Ãù˜O`š:ì›/sfGÓÍç?s·ÅÐì©ÎqõŸ™hNÿšÉîc4š‡˜L<7D§î«nÍhŠÒ¼g"%â4‘.â㧉'0v¾~­†=J-Ã[›6þïoG$m" À IÈ…$fF±^ØNÞðrË럟4À™”bßž=+CãÑEÆ®yÞ„<¨¦*XB 'hLkc&wj”8™«Ë[ç‰- @rOÐä«dÌQ¡á/FŽjÝ£“ÿ¼Äž_~Bîvć×|ó×'r Õû äŸ a^2Eȉ“@¬Á …Ò ï?“9Caâ¨bî®û3^ /,Eò¢œ­‰ƒ4ËýÒbðÍÏx^ïå+ø¾É.rpãŸÜùÍÉ›ÿã¡È#íé¾·Pº¿Ü/j¼ÊÜM7ÉÙ@_»i)!'¶ˆ·Œ/…jªã|ü µ÷ب”nsB½Ñ‚k/4Ì:}_)°†&¯Þ ­“ Å¦.…8ŸÚ?ÒáYí/à wüS½aAÊfÉ]0 N*w*R;¸r–“˜`øslÃÞ'Î0ÜÚe}ƃ‡£X«&“UÇÍ‹ã`zð<ØÚvÀœ^olI,x\¬_{!‹ƒ®¿}¼”aÙêFëÊIXùåái¶Vkò…TÁZÇE†m°)ätÂ9Éþ¤ºqt°!‘XcéæíE$~«Ô„ÝŒHJa©å覄d8©ö«BržMÙ‚êH±r/Ò¶m©™ÓÉÀ&ižU²ä†_À=ÚWxÄÙq¯@‰¹ÉUd¨3u ôD¦ [éÈòp‘š˜KÙÞÜNœëß…ã^iV8ŒûÃBŒ’ÇùÒå'Þ“Èg-~„^à"ò¿~öPo;þþ‘ŸÖ9‡‡|Š;sPä)ZÅàÑ÷ÛŠ†¦(þüŒ<ƒ¯*{ ]8îgŽR/¯Â÷ë(:¤~jWÊU~ôÛŒŠÚöƒ’û%³þø”÷¢²àåk=*Tù®æ_Á燪§‹“Š?ú¡: kÃî3ÔÈ8ii}µ ÌÞë@í—W.~¨{éqXýŸ1Ô ÓÎ×ËCƒ÷Y 4<¥—ðÎã3Q˜4$¹¢Qª[@ä®946ní =?ˆÆKcÛ vhâן-ˆ¦‡-XîÌ6¡i) ŽoÞÆqŸýœmP!ñ‹ú¾û¤ÈÀ’Ù+ù3 fŽhÈß)¿Á’IÊð½–Þ÷ʹ¨M–Wš´¦4dh‘šilAÚľï©ÇÚ!õ¸'£ŒÕ¿ýLkî'–!®ê(‡gücˆzµÏDªÕf§&Äû’ªw¾xß½ÉÛ,¿ãÉ//Œl!¾ïÕí*—¯ý»¿ FôC£é¹çŽSÐTüö|´w:´HFežŠƒÖçê©âê„üw?Ý¿Êʼnçî:ÿ°úç1¤h? ~ξ^·Úö~+ê=’ ­CÙ®þƒÖž\FÝqÂß =³i 9P½d÷ÂSý¾o¥hmÍù)å íÁo²ßÇR@‡çÙsò)\ÐYa9jüdºÕ.µv/@ÏlQB1/ô}¿®sç†7 Ø¡•ø_õœÝ꜀a]n«/»`$^û­Œ=:WÎüw&Ò™ÊVœ‚©’¡ƒÓf0«ûÓ+læ,pãMŸ«Z°Xô°ý|,ç?hfµ†_÷ô,v—Áï;®v¯2Àzbem¦l±„¬ðôJÃv©laA”<ÕýìvómE-|¸npI§Ì89(q××oy‰D¤¨lÜ_àɆÔDÂL³'„‘æRìÝ0^¤û«P,éHŠô#9RI‡:‘áô&ã•aadKß²ÍFæAþκ¿‘U5¦DË%ÙѦæ5Õ$rÞj9à|Ð÷/¿|¤z—yŒ÷9l¯É"ïWÝ‹ä2Ix@Bi®’˜ù3ivS©Å£€Éï{Û¬¡(È_ñXêo ñÒ:ñ"ðРÛËöŠd0¤hF±ï5‡Î5¡„áÕ»)~×Q’(纚¼J«9wüD/ãñ”C(çP´lRƒ áß3rRàñÉi?·ÐDTâvyÕäj‰ÊæR3‘F…¨÷!â•Ô]Tå½·ÄWøÕ*i^ßëFWŠÒïfC <¤ùÅÊ`:‚iGn…Ç¢n­é‰ünÔ?=俘5€#Üeõ h˜Þ{#üÊ 4º6\•7ºμ%’Aãî`‡¤ÐC„>3êé«'MÅ›œ×Mç¹òE ½Ð,üèeÕ5ÆÅ?F˜{;Ÿ¾Út† æc?Z& 7—Êäúä½h(ëX…|v¢S¾}Dÿçõ'æÉ‡‚ùh†ëÖPxI÷]»U?v­ž*[¿_ ƒDwÆ&|­Ð.ãÛþ—óVë ¥B"Ý#ö³éø%.G>hÿçù¿@ä&ew;®*dB©’Rq«T–<+bÿ UTi™MP«*ïæÌ¢ ?Ûß\å¦Ch(¸¸ñw4)Q‰ÎëÕBË^]uås× sï R…ÔÐq´|c|”bÇ?Õ»oZŸ­húÌdØSÀ`@볃0lõ»;ESFUЬ”, aNœ7‚I‹OOJƒ¹`Ú/Ͳö£2ÌnòŽªO¾Ïq³foX̲2ëzËG÷;~¶ƒ•Zò(‘äpX]1çP€µÜ'9·Ïü…MÒÞN})!øã:6Z¨†DÌ&o#ýØÃû»©’LŠiˆÔ"Ùå„·;§œä";½FR4 ¶ÉñÉ#Õü³›#iHcV–UŒ{8¾¿9\R…{iEÃŒù!CÆÙÖeÈtKpôŒÈ_d9ÿŽýk /²]?¤’,¹ §™Öpÿ%Ê–ù‹È£ïÌ]Ë‹|’zs% ¡ÈÖå<§> ”]:ã0ÉŒ‡/æïÉ™G™Øî¦ <ª‚!çŠÅOÖˆ†Æ? ’ãÑäÕiÖš-à ¹…fœÏO𯅠ÙÔϨ¯KÂxC@ˆÒAÕ åõ®ù”½Á½Ñ÷ùHa:;•?’|ÊÃK; óÜŒ·ÏíÂ;>šÌ…Çõ–_"!½2l¾°ØÒ.çž Œ9)Æ v'´Ü!aº:(eâwKãà'Ì¡»¬ò¹–vjÂÑ §-’Uø"7ñÌï‚×Î9jåNtvlP¹È)ÆM õDÕ{ÛCÍ Q’ý\ÚÉ hz;1ZåÍ -q)ÊÐz˜õÕîb2hcMˆw壄6+ÛÎÖQ&1Ò^W$ä? ±ÎÞ(hmk—æ¯ „ÖBK–¼ûÐÚ°ph老ñÜü4ImŽÛÛEÅ|Ðöãìw hWÔbÛ‚¢XšxkèX’W ë:]xøýèÖeèþ&hнgЧÞ®ƒþý¥ABóÊ0ðÐÏ6pð4 Fú6¦}ß Ãä—þœªSßeî]0f’V¡mJÑ*‰y=0åœÌ>. 335÷ùþ†ùj9 YC°xo2sÆ?–åUom_6…_Œ›÷ÒÔà7ůônGXýàwN¼6_½[Þÿ¶Å¿8·‰Ó!Ñ=Û&­ïÍHü›Vþ¥“0’‹~úq”wÙz?’ £A 'fê=n!UD³žõØܽñâBØÑ:¤{œ±{ï”-ÒGªˆÍ›#£Ÿ¼µ2¯øåÉÌÂÛ,çìWeì“èžÅÈÖŹþÕ<9wWŠ]}3†û½›ìJ¸‡e·Æ«GÈ{Ï×f tù–ƒ~hé=FþÇχš5 gØ!^<Ô_üìâ?<ܱOSQ1EH¯Dàg?u˜—:éÙƒb!*-«U(!X#è£4…ÇÚÇ$†<èQê£åšƒ%ÊhOúéËV¡œT×rn’-*\¹&UD‰xÞc*DýË’­^Ô¡ÁïPyöf4,5¾¤PFÏúNXI° ±Áì+ÿ†D4aþ³¿Éé*šôÒ nx¦ icíÔu4sáyô†bÍUšÿ4ÜD ö¤‰}ah±6^D£Â‰–Cß .¶£ɇÄCK¬(IñaTñÊ'¤™}8ãGüFÃ?§ç§ú@É£ÚŒÀ·_!§¶hÝñQ7WÉÉ/¥ iÿ‡–C¹1à›zâ|B/kšt͓ăr÷ˆjMyÿIÅ ""XeÊ5BøÕÖ× wjdÓ¬çjç]ÂÜ*zî†æ($-š¾n¦M„ô4¦ ¹XNÈ*;Q¥J¿9ôwý˜VHýï¢p#¿òäÜ.õ\Qƒ¼&Ûp>)ÈÿšGG´añÔ鞺PÈ•¹ºšA …Agž½#R€¯ÔvEÞÀ×{éÁ–oëàÛçGáæåðýÃj‡‰6]q}ýÔVþ<”ÕÜò=Ä¥R2Ö$ ìQþTìQe(ÿs&At, *ÓõD] *»õωgcPKuz02à ~>>„UoQïô´&X Z¦’„^G¡¯3ðÇh*cù°3ü]NhËñ·@b«¢$kas$©¡>É5bdZò¾uY~¸k¸výÍ9¤HWÝMuÍ©ªÎûæÙÖ Í±÷.iq\H7Íè}ÌÔé—ß'ö­ !ÃÛ«Òc«RÈtjÏyýRÈ¢¬|ø¶+²©·xßqBŸõ˜ÓS´¸_Ø}»áÀ>äöìÜ›bƒ|Ôöin¢MÈÄJi‚×|êîÍ=ù‡¹å3Ž  ð-¿¶þ;ô^qyrÅËq­™¾Æcìç•yÊPŠ—y€šÐ/2R¼Þ4Ä(gï¼-‚ •™Ö*ÅKÏ‹ìÞm¢2MÂà˜ÆªÄÒgå”™¡ªªòU8þÕ~§m xj$çD{‰- Öi稽ü¨CrÁŽ×ðꆇ½Í_ÊG}Ðe´Dƒ~2oA]q4,>¥"²‚Fo|>‘F¢ñ5ñ}Ç ÝÑDëv¥Åd0šŠèhŒ_¨@3Ö{E\îmhNûõѳÖhÁ`}!p¢]b\~«ƒ²ók®ª à^1ý«Nf`Z‰•®Bó”³ÄÔ€o4 }eîøh2/XÙÚ‘Ð@ººÜ+~ùOZýiuA°0_n”±ZCÕg¤ÿ<‰“ÏK }7wlíÔø¼_"𹕵lüac¹hçœír{ç4Ìjÿðúµ_ó^à“75ÐȘÝÒv"šŒ²$¡9ôÁ¯¥‚Qh©œ¸>¦° ­ÉÑ%\кb¢Ìuî ¡êOË(@kÞ¥DÕ9Ah Ot¸ä­ÁTo‹½öAk¹à!ã'Bÿž6ï£ äÉw ‹þ¿¡}ƒFϵhÿp»íñ–=tØgÏiÌ‚N[fRrMèÊÙ0ôpž‚™=†åúÐ[W*tM"ú=Ùzº®’À@Ü|Q«( f≋1Ga˜è^÷ÉÑ{0⬸¸U;c|3ôƒ*a|[Vœ+Ù¦xÜé*®-ÂÌË÷í¾„ycšûL°(#ÄßÜ`©÷çµÃ_t`åsüg3‡ X »eQTkéVŒ¿Òda“òݵò3?áO„„ÀÂ÷A$'©Ú¿Ø€Ä1]‹Üþ HªB)éêGƒ»y|d¨®!›åêŸV¤R¾ó¸áÙî~嬎FÐtþRÒŸ©;Ԛ݄{«ymFÄM‘QM {î•2µ^ù`ä³€,>ñïÚ¸ú͵ýÐd´+r„ç¿L~÷Óq˜¾"çEîÌ•OjYÇqï¡LBîäÄ]yu¨ÛôH&ÿúGÔw|z­ü|>îzaÑ3¬„†-•'ŸÑ£QlGáÖ=:4öðòáþÎ&†~®©¥¢©påÈ?4£e'&pÞìŸþu4ssZ4-†z]•¡å,#sož¢Xa iâDGû1Á‹(±QøTYÓi×…‰é\`˜é$nߎa^ßÛ\ÿÞëæu©”‚t7õ——´„!ñœ=c@‰ ÄO’zÖ= …¸÷êv±ñ@Œô“­…ç§ Êfï·¢'›q*àŠ»[.„õeíú–e½S#£‡ˆå<¼Ê"t£ ’îd]ªõ™&|ÿ‰6ýsï Ë+…z°0eWOœcܹBÝÔ£äÍ;ÎAÞßó/'g¿<ù·xV‰<)8ï¢8®"œ¦,”âP¨W–{‚ K¯ÛÍÄ…¯rl¶o4Cÿg ©0߯º~^pù÷¼„×ÛñÃÿxÂfgjNB©2Óî"({þÑž.AÊßa ]cÿwNb°w6TùFi5ßÛ„šŸ÷ôšôSà§Ä7›N/ ÁBÏÉpÆ ËlÊy§h¡ÙEëfQ´Z ©öe‚vƒ å0'×ÿTû»+aãÐ׳•”0Hy‚eã“: •„.ׄÃÈû·]4š‹0æ«uÃ+À &’Ê?"-LMêú¾!g„Y³d08 ó‹abk­Íd Ùy»(ÔSàσ÷ˆ‚?Âß”»¬½¯‘ˆÞ]3qG’¤a›ÕÐa$®6<8[Œ»Ê¥º8žd"Å«8ìÌmGªhã›3£HCùˆ')þÒ\¹ª8&ˆô]äyWËÃÁófˆÊ¡dRüÕìÈÂ=)'FƒlÜ÷/NiÅ!ÇÉœ°3ßùpßâ¤`Ñîfä¡ô °PãDÞ§.‡2©ðÀ¦*÷—YRГŠrâ÷DÁÎÀ×3(âêP½EýK êúuíÓÞ«²2(12GŠT›(9}U,ÍCeˆ‹÷üýˆrŠ»2ÝÜQ¡iuß > „F·È.2jTÚ~©ï_~UÞ®ª§½‘EUYbmO?)T›õzs¬Z5ÏǼz/ˆZçÿÆEÆÆ ã±{ŒFÝÂD&±kPÿÚU |ó u”‡…ÑpeÀwµ`/õD3Æ3É q-Ýàn·¯hRV‡$"‹hZ%ïn•ì†fÝõå'¥ÑüïB?9?ZÊÍñÛõû¢ËA‘-›)%”ñï£cˆÍEú¯œ~le0UuwmúY”ù?ëå)g¯GlîF+ìøh2Úĉ=ôoBZY‹è‹Ì-H©ì—È£ºIè§|®ék™ÿ÷^éÞÇ»!–·Ô>|ótUÝ—…Vß㟫.Úø·ò¼ŠHZ˦n§@ë9Jž,bkÂõ÷v¥l'mòÕ@Ûñ©m²Rh+q9'Y ífe øHÚ×]ì–ÒÜ [jèìÿ{EºÅjÈiƯCOê¼Õè;^fBá,ôw]?øl¬ªØ>èÁ`Ù¶gÆ0´ñ.ÏòO!Œ˜Is_æÙ†Ñ%ýÏSýÒ0^PÇý:b&ëOƦ̈”?4hä¹îfç°0ÄÄ2cK÷-Ï›‘)ÃÊq‹†£ê°Êîô'NÖ$Ùß=Í‚ûïÞ†qÀº”žÝúYð·\îÞˆÆ$6þü)#= If‹yféìkÑK)ןH^ôa(ç•$Rø¦rDwó[ Ó]CÚ€°°ÚA ¤ç°ä™ºk€{½³žFYÚ!ï©…ñ¨EdºÛȧ7ƒ,BI»>¸n"Ûžsæ9ä8–¬ms‹÷…žSÌÒæGn…&ßtOEä%ÛsêÍÅÛÈwSñŽÞØc<°±÷ûM.<èwÕ¸O]™å¼.öaÆÃêrk®—ÃPØþÞèÌß}x¤Þîvê :Ši‰»àù‚âÏ=b‡ðØ“…ûVÏ ”4­÷®@”céö½þeЭ<ðýò«ÕoÙ‡Òð¸f_¾z.b¦›ûéÆó¨Ì“´ëRª<þ~ŸVç4ª²ø¦rS¢ZMMW|ëuÔ–+HŠªF­3KY»IëP‡SÄ®¶bu[Ãý¸‹úî=Gåý¶G‹!i` Çu½FѨ¸Yúc-GÓz§$ Éójo.Ž4½ý`üMî54»‘}én5ƒîüì—ÂÑâù~‘ñÉ´L¢ßoÿOÂÕ”±´:"ø:Ïâu“Тu­,Ÿí9¬~˜wË_·såd‘žûö jŽPœ¶îÿô4K-´nR½‚´¿1W÷“óAB¬ŒñŸšjB?YdËžˆ£ùûBÚõ6DM>養…È)סΚÛÁD|ãxz„xÔ‘Þ©‘®mïèxH³Mõ%,’È™ä*ÈD ]Þ%ã+dɰé&)Rþ¯‘J²„4䌜è27®†Ü oŸ†õI!OMNP{XòÅ|úˆ’ €³ó‡Q’ ¼~.e} )„Ç•i[ Ð³ÕÞû.Ý“o\·ƒ¯QÛì‹<[ÿÎSÜ{¨ïý¿ó£Ò=nnÿ‡ª´)¢Žïÿ7À¾ÿaº7é~ÊC¼¨¨^C¥ïëû™¼Puñzð…›{ &02ŸdªêŒ»÷G¦AÃAåDch|òÒ_AVšù~ì!²¸ -ëë£i’3ÐÖ¿~q=[xÇ?ÕývÁõÜÆQè³RɈè‚’õAcC r›Ý~äk #êß¹(—À˜dQž¶ÐS˜0¯¿DÍø¦‚ßZ—÷„Y’¥KSÍ!0:JÝm‹Gè®^²…¥ø©öAJX1Ha¼ñVV%£ZBaM¯&=}Þ6üÞ×;ÁÖŠi5ü}œqÐ]¨‰y”žxÞy$~…¶?M®#Ý)†»¿pW\X~xœR\§tM ;ˆT^L¬’NqwﵟŸ“:îE¹ú­¤/r©X >… çœC&îžÓb{¢yûµæ²41²®‰êkr¹…XÉâ¾Ü“Æ©+Ï‘»Fð­\N òfÇë’QÂU"Ûo¡ÀÞe‹%GQ0p&ÊÎ8…5»¿ÈþŽ@QYÒ¤ ½(öƒ™´³%Ò<ÊO“@É/E÷ÍBéŸ6y_ÿø ­W™ÆL*$ò?ö!ŽBH³y¨}•®(¨t£ÊóîŽøËœ¨*\|8¿ÏÕ†µE”™P#ž66ŠôjÙù&Ä^'äC~²³™¢ ¨Û™ã;þõ?øÞs ù†2L'2ÄèÐHðrgÑçh|àë=cþŸh"4ÛJ‚¦JÜÙIN:hvùðçúßühÁ]Í"„Kü¼} Sxòœ×(¶:Ǥܴ:.…ÒÁ†ùÚH¯5±žØs¦ŒÇ†ék,¡ŒJÌþCaÎkðÈüTºã£ÉÐ"ÿìý ¤¸^·|Ä’ËY®ßŸ¢û?oâ÷{ ðÙ)²íÝ#Ä…?6w H–"7õÍÏ;õóÏì¨!~[Ⱥ,PÚ¤ß2ÙIÎòhW„÷ Tký‘{¦Í?Jyž M‹|¬g q°ñNRt64“\}Þ~Zv= ï„–Î7Ë®bõÐ* HïzÅ€ÀÇŠ¸*ø"8÷#Tž7›U¡Uë&++™!O†jþîmº±‡Wýóz·‹_$äÂ[–QÄû×ñR è„fº–mè2©î«±¾Ýñ¡§¡W0·&õcô%—'°« ÂÀ¾s9ûÒU``6Ã)âA ¶E Ô^wƒ¡•Ì·–Ìë0"wT·ÞF‹?†?= ãçéÈ‚ÎÁä]ÚµG /aº¶ä±”KÌ]zÏÌfРᓪ"²°Ä?dám"ËM»ÖÃ+áW‚¥}ô‹§ð;»ƒ›+YÖ7ÞÛõ]-/á×{ŸÂ_Ák×^&¾Eb¢›ÖjMHò®@¯Nw’ÙÚøErÛúÝo‘òAÍjfC0Rçg³òCZv¦‰„Ó!¸'¹âØö‹Ü+£‘bsï.2Ä¥îÈ>…L‚òŸÅË&‘¹6Åù| ²&è=RQÜFöÊ‹SãɸOð3§Ÿ r•¦¹ª$´!O…¡…ãoÈ'ãoÚplÈl´3z¢Žu_ÏÝ5ËÂCôo¨;Pðfƒÿ(ÌECYšpxÇ}´b¼ƒb¼ŸËs¿$¡xú”³åYú P¬±çï}„/}QÎ:“öœD«¥” Él[­ásxâ|´Øuˆ½×­u¥¡¢uté‹}¼ë _Þ³áï¶wuÓÂÇ2§Ùö¸=;5R­†ê룻›2ÚÛVõˆÐ¿®´ßï9xéw«ë d.<ð³1ò'ô}þ‡u#Èù˜_V ¹*4ÇŠÏCîv¤C¿ü)È›\¨ zjù­Ê&…,ÌP =Egšõ ŠY˜™"¡PZÑþÏ»óP˜8FK— _mGÞ“³Â÷=%Éqµ%P¤zHÙåî¹ÿñPkü/c”Z~̺™P eÁçm#öBùô—±›œTPygÒaüìT)È-;€Ë2cO÷@ï¸YeSÔ]­©tŒ‚FÍ\+Ó'úÐÔ¬þ ·ë¡%¤êÉýìths¢ö1ÝÚñOu³hÜ?H« ½¯K>QºÁ€WZbüº- V¢æ5`€á%GËù@í‘ùËRò&ˆ%ÈnŽoÁÔ‰‰’3×Ga&MÆÿ­EÌ« ¾™< Wjš ÒaÉË'ätòqXP“ƒUº^"wú@X’ÝÅ^ vAbE–°UÊÊé´ õ¿Ôv¸“#1‰÷Ñ…HâôÆ)ŽzI&[Úlã.ŠPþòX¤Ðµ¿mp©Ìgº±ãî Ϙ4i¤;½GÙÊéãúZ_aAußSkϵ‘‰¤¤“wW2÷ÊY¦¬"kC·Ô뱽Ⱦy,à¬î»[åÞ«‚Ü•¨ÊFù‘÷eѯ—·ðÀ[æ^ÃÓx°m×ø³C(¨¯öþ©X- ­«üIÓ=€GúÕ(cÚ‡QìíƒUÿPÂs×SÓv(éÍܸÙê‰Ò¡DæÁ›.(Û÷«•çâTxh<÷aá•ÎÅüŽÃ¨4dph´ÈU¼Ší£¤·Q•û>åºh ªu²”ê[s¢F\n\–Ã{ÔrâL‰—&äCÉ–®­ýi¨»ZFËynõkRkCœþ½GÓ«v_Ьf>W$Acûö©M©‹hò¬ÚþS‘&š¸ˆšgÆ¢9)[§ªR3Z\t¬g`_AË‹‹9ÏÒðÔ3Ú8æÚytæ?º`æ/„’-úçl{ë¡Ãý@¤Éý:išf(½ß¶Æ–¶ו\+òtw|4é=ÇÞëBê#†®/zúT¢— ]G˜We[_ÏÁ§“~|VœCq'‚¦'"rã½jÔvê瓬֢]rÅíß®ó ¾=ð7>¨XesŸ›U(¡Z qÎ(¹~Z-¾ |, ¯D5‰å¡1~€xnÆš¾4Ÿli„æ+:·¤hq|EÃ5pZ¾3¸sÛKBËôq’GÒÐÒsŸã w.áZn÷ 9!h= b^\9­‘ ‚L§éþyLM³}ÒδTMüù&´Ü•®).ÞÅeCçç¡‹ÃAÙÐUûíèsèásp>Öä ½‘¾Ïøû­ Åä¼î" è¼´ruäƒÁ½Œ?brc`pÂ$ÄÛä ÍÕ¹YˆJÃÈ~ÁÛ²U|0ê?|À•Æ¥W·n=†I‘ŽÙ‹â]0íðý)'“ÌÎÛôsˆ€3pUÖä€Åº‰Q¢°|9ëÿ¡ë½£¹þÃ÷q{o22’=3³Ò}Û{ï4”†–%”¤”R´$Ù•‘‘ÈΖ½*Ùd~yÿ>¼çó9ï¿îórœÇ9Ž×õ¼îëy_ëæÓü¨s ŒÇôÏÁ‚òÓØ ñwð×=\@•'–&許‡µ{¯öÒ?F‚÷Z,m3€DÚa^ÉŒHÂL&køÉ>Š˜B þKR‚î“He·uJi^Ý“ ü¾ŠôB]B";#Ã¸ÇÈdë¾³½²™b÷ìîÆgS»EwïE6mŠô³wK‘Ã!ô˜µrfþ¡YkGÃöMíç÷Î#+I"ܽ8¸ÇÂwC^"*xꆂT9A<.ûP¨ú]8Op$Š|´MH y‰b ÏÅí“QRf L;¥fÕ$Ÿh¡Lþ·?T('ÿZÛÖ= å»Éã™r}QᆖHMìIT²×˜3:'‚*·…ÝóyQµ[TzÿÑ3ˆ6~NEm¨¶²TH‡çZ‹˜D-zõ›§¼Q»nÎ#—îê†÷ÐC}Ç{=DKôhÈ$:bßG‰FÍdg¼Mîù|"¹f|¥wŒºœÑœRHöÄý44ÿ[!Ú번ó’j‚®¢åú³þ“eÞhͶø[zVmÔíÖ¼†BÑÖGø½ÂÞKh×pÜaýÚùdŠ >u¨|Âf~ ^yôˆ›cϼ^ù ‚R ¯š«2ÐnØ^Þ 9ìù_tÚ@a·ÐÙk»5!Sù÷ ïZÊâ,rß%šz¦ ~™@”Ž< b§³É_Hîùgn˜D±·õ}â4Vðж`GˆE¼T4=<È3º]_³¯çoð¡YmÚÂìQHÞÙÔ—³”©¥õ§î?c†ŒÄ_õ,'Ø!sžAûüÉM_—ÒxdO’ËoúkÒÜ]€OoìÞ8ÏBB~Ä›þšÓ—¨Rà³×’Ö§Û¦Ïæôm¦•MŸÍCRÈ—´ü ñSv£N3̇B¥# §JþåCÓÆCßr8¡äØ‘;Íú®4RkÔ“pÓn: áå§Ôív–ÕC%¹”ÿ%¨æ~|Œ_ j÷]«¿H¼ßnZ©Ûï#†ªA!·æ³Ðè_†¾qÂЬøBØRÉZ 8“SÕ_lçOu„EòÚC·moc¬4#ô*ð Œñ‚¾®Ç;¾0ÔÀ@”;ȃ!ÿZöÉ0’öf ïŒM)t½1_½?kvÑÀÄäÅB3¯Ï0ÙØ)z5e¦Í…ì¾O\„™• §E9˜c§h‚E²X»o¿Á’&á×ÈÞYXyÚ2 ÒÃëlŸ.)•©!Á`¡sxuYœõ¼^€ÄM^Ïö¼pDÒ“ç¯%íBr±šƒðb¤”.‹þà%‡ÔêbÑ¡H'qTâ@„2ô3)6"“øó<ËÏñÈ<”høB;>÷¹žQÿ‰l‰w¸‡>ê"GõÔ}g8‚\ÚÊqË–pÚz›1Eàî#ÝÙˆï‘ÿØZw§3 †Ÿ¥œ]}Š"QßÒ£Xb’ŽptJÞ¾},óe-JŸ-CYÝGúO¢QÞà4ßkÿ(T8ñjæÜ_GTJuZì´Ç}7lÓ o!œUÐ×PBµfé°ý¥¨q>n%‚¼µ˜ÕF}º6ð9`ï@º±Gã¿¶4¢¾GbzÊú†>Ô~k£M“ÆÌõwC6ðh²ÔÈQÚýÍ|âtÍTˆÐü¡ûË•«h‘6m«½„–£ R—‡‚ÑZ¡4Ž``CRPæÝD;‘‚fÚ% <Т±ÇÔ—Ç–Dãù7i§Lv¡œY+EÕ¤+'§¯+«ƒ1—¬l(^º7¹ñ=O=×^Õ¼G“¶'­&I. ÞßÐ㸅‹\pk˜9y?$~Z”:K q•2ÚI_ùáÍUÛÜcµÍùÅmÓµÛ5~À“ãZd´-¿“¯/n÷ø®p@iÙ2“?²PElÿf'£7ÔÉ~>­áõ%^»ýd…† wdö*CãIÍk|Ó— IS|ÂL8C'Þ¨|¸Ͷ+?ªÜØ¡ùqqØ»Q„MI4§šûµ0@óJÿÓ3úªå)…ˆÕâh=Åá;²Ü ßA*¯¡ßÚD»Nº–WA;eó\t0MåN‹ÓB'ú½ Íz ]û…Êu¡gçðÝTð#¸B´Ì:z¯f­×HAŸüeÞ„ è["¥’WʃþÑKw$ÜÃ-CÎûî|¹ó÷ÆÇ¯¹Àhç›~MwøI“w— ³~{ 'rѸÃ$Uh¬Íåe˜:=¯Þ:%ÓËG F=*a¶¨êà ˜Ï³ÓM…Å_á‡éœÖaÙ^`Ù~y¬N´èu»##cõ"$œ:{BˆŠ ‰ó£ŸKgM"é'×øŒ3$/å:ñŒÍ)(º¢1§Ô8ÖøÞaÔ¢Ž8Ãtî6j×i‹¢îKEµÛ¿¨ïÐvqmN iâvOp£QÏDt|+/š„5ŸùrÍ$njru°¢ùÎAGç_h±£øÍô)´Ü•ÊÁs­”×B†¢õñŒ?k©Ñ&AÌ`Á]íE=£üÚ7t!Õ̽Üè@}bíÐtÌ/tãëÖA§È¯ õbxºRÊyáð J>ðàI|*ŠïêjÕ‰` û«cœº«v_&Ž…O”Š|ë£þñqß¹J ç_1Âè-ˆ·±:âÙS±.˪înÎ Þ†è»Q¦‚Ïç Êï¥äΪ$ˆL’l™|V´]£G>2©Ï_ßàÑÌÛ»4ÿñ}§_þ©w×u^[†@ÆQ[Æ ïWù!ØüÑCÿM_^’ÂûMô;r.\˜' „OŽ…O TÂ!WáTÈØäæçT„:·Àg¹Ãîù|*›>›3t÷ŸnúlÒîP7CÞ´×Ïzcø’öj86‘ %¢ ü—­?æD;:BÉyáƒ2e¡P#|âã“i(û1ùÉlÿ"”Û½Ð4‡ŠÑ&of;#¨ X +¸5c †\{ùàjÖo²cP_K?È÷å5¿ûBS«k…@®*´„H|£Þг[ùSLsáÉrg¡«Uªÿµm üVõKÆ}wÛ7Þ†ï±ò0$.ôa:F¬4ÍB~ÂXÔ­ÁïÙð‹ªhhÒY&Â?EPSéÂäÛ¢@~*m˜(尔Ї™&ë[¹~‹0Wbn·Ì  ÉcŸ—¸Î9zÌ‚W† †&°Ö›Ïù‚é*”'ά#‘¼®ŸõU$΢+¡ÆÑûBÛÊP‹œ üò7Ô.ý0å:MŒº1t íHú׳RËߢ¡Íå†àå/h,ÝMFfS‰¦¢³Ç§ †Ñ,æœþ.XEóâ¿ùÜy3hñ—¦ØØ&­tfµ¨Ñ:;YÅmUT$š¥Ñî7};eG:ÚWä».ïGÇ6Iá¢\s<Ï/;h À‚²lì©âkHÛû:ŠÊè!Œþáyí÷°ŠÝ˜"àÓÞS¿º·óhR_ºäúí†w>×í,ec )/ow×½H¨ËÚ—%oçjÛþË’PG ?u• ê/T±„6Ï@ƒò*ûõehäʈÛVÏwzÐOum”»úe¬Ra`hg õm’ òw¤˜…átÞöƒgG`4èìw¾0žò)hØö"ü&špç/?¾ ßusaŠÞO‘H¦C܃• `ÖØßöɾh˜Gæ]"oaÑ%üeï+CXªxz9ˆV-(¾ßAÊ7ÖåË‘ð¥îLãI$vU‹”â3GÒÃ)aä ï‘üXi4Eî[¤¼2Qh†Ô)W¾jÔU Õ5¥dp ä ŸýŒŒ9ËZÕ“OYàŠ O²¤g~̬=‹¬W}éÊùM‘ÝßËÇ@ùî¬÷îÖ·—Anáy…ä¥e6^›»»¯Ýë³e¿…|³.çɳòP ´Èé– ©¿‰¹Ê"|—ís7øP)õ‚nŠ2J„9‘:&q ” ÷YÙ1.”ÙGÉs® eû×êM. ü-º]}Që‚u‡*þò˜g5»‹*;Ô>{„ªöúíBø{Os%ÝÕÚõ¯í]A„™®·6ð6g{$µw1UéÞAÝW÷ŠÏ#Å ’/Mš£!¥ð±Ñe4ê§T&D“(þ'­OÐLW}¯As vó½Âh!“ÏÍ·>†–úÆGÆOF«K.Z™çIÑ:óìtÔO´¥[~v™ ín‰k’ÕÚ³¸ÇÚ<‡õ ³d‡ñ`ÚÏâY©›x¨h)Äö,žšt~jÄR€â—2dØÞAòfÍ£jÆq0`#Ûu÷Â~(¤´4:;L“þcþL‹ðÞB±zafþŸû‰Óì—–1¹oÚÎú/õÁëq÷÷×Èþ™B·;¼œ¹¦õVÃ"¹iôW·ktNÂÓ[R|öÒ^Å€’¤Bï&OIBªé¾,:Èà*²ô~, ™®óÉÝï_nújØ9o¾…lCsñ)N?ÈëP‰Ä(øÄ*âCÀ‚KxenÙ²gÓ_#˜žµ¹¿*b‰’Áç#Î<#“‚Ïm;ýKób /;Q+,Ó¾Ü|àÏÿL yÛ§†þˇÏ/DBɵ_2- 4r„‡¡Ê:èøòMÏA¹ÎÔž|¡¢àUÅ“ÃP• ~ªÌ!jžÆ–}=® u¹‚¾ ePÿp_^ôùTh,>·6ÛM—•\çU ¡…wÿôñZ²íü©öGFñgunA—•Ït”¯üpHÚqxl/ôi*H,„©Aÿ¼ì3*ARl|Yïô Fôç»1Ã&Æ ^ø™CÎ9è Ú—Ë›`Ò3~°úĘZ<Ü^w¾ fû“¼ëº`îå=?8o ÞëæŽ|ðw®kÚzÚV” ¯iæ\ƒµÄÂ⿇2‘ QMå×n$bå½#)‰ÄÏ_pþ .AR ÕLH6òH÷§Ì!¤¼"&véeXn?Ï5AÚZj¢3_ž!ƒ=ߪx»>2Î}ð¥âqFædÓ™QoÜá%8dÜŠlö}ÁféÈáææwxô>r6r,{æŠ<·w»ƒ»éÙTbõ¸Ÿ Þ+çª> òƒÉ¢Ò* _7»Ù[}Ũ<¿³ê¿C‰J®ï¢Ñ(MZS›˜tezjëÛèPîÇŽ’_?úQà‡äî9YT²t"O¬œÆ}Ô=¢¾Ç$h(iXmP-{ÆmHéjØ\佟˜€šK…qw¢v®÷}‹øü”ø3AõýËr?†¢á £Öñ246»y­AM°~äÎÙ‹f_• &4Ñ|Þbäú´T–œ?!ŒVqG?1I³¡äÉ'Ö×3Ѷ}øÔkîx íPégFtÈ’TÿHŒ×óFŸ¥á9m×¼»LÌ(ý銃mpÒ,©Jh7™Ã¨çí{o4¡ègUôýÇ­ó-Äeý÷àv.F*‹ãZUÍýí\ŒÄO3ßžqÛιx»»ðãçH¹íü‹Wª•{R5¶kÜ¢rËròöy[9[9•y94Im? ¶×T$úR0Ô›Ÿüô¥ˆ¨Ã2¿æAC=þgÚÍ~çv¾E7'½•Ço¨d!ºTf ƒkí^‹–Gaø\fyý‘“0ª¬\C8W ãFY÷«ÿ’Á¯û™ûwU2Ãĸ7%á›ë0™A%!ùq£Ïü)³\ 3íªæûxWa®¨qïùøÝ°ÐÞ‘þã,)ò¸„°7ÃJæ~=ûüaýœýd6$ÔßíçSÁŠÄL½Ñ’ÁéHJ9éçðÉIÏ´]u"@J𿏍'ÃH-;ÁÀk ‰´¹WbË}‘1Lÿg=2ž,8PF¼†Lµ½ÓªÈb•ÃëÍYw¨¨ÎœgAv¶3÷´’¨qçÁ’²³QòÈ5 B袊»¢ž¤gDãnÕ¹í.$È÷ñiù‹¯( -á¸Z‡‚+QqI;†Q¸ý– Á·ý¥Æcë‡Ú"ä]&(ENCÏÃL‰2Ì®µÞ|¦(›+*%…ò¶>}q—¹pï8Wòe^TÌ ™¿¦®ŒÊÝòTWoU¢ê¾‘Ò›t^Õ YƒIwQ­ü|»­;j˜ü}ýã\j‘˜âëÔnÊèm‚dK9¯3ì  ö•2¡ª @“GCÈ·m¼lå?máh+f+ÿi+f o/p”ÊÕˆ‡ã_(ïºo×­œš­›-oá;½Uü‰tüÈäs&­Í†,Ó‘ûu‘µ8W3ûàÇvnÍÖóa+?çÿä×p¶¹íò­ü?96[y5[95Û|xØë”zl7”¦‘þêu„Òw¿ŽÞk„²fñûbÎ#Û¹YOÏìÓº UnÆ­s MP£¥¬vþ™ÔÙÐÆ0½ú÷¹±•oÅBAYžÿõß¼šÿÉŸÚΫi¬#Šˆ€ž¶$Æ}¶Ð;G)Ìz úcxlÎs·ÂàÕ¬ú|}0œ”wSL¹FÿîŸP»»3õû®™Ÿø!˜4| á^ÚSÕëáúéŒ0ãw®¹ aîJYRËP,DŒY|R4„¿ß îÈ^SªÛºæ‚ñ°v±ôÔ”?HpÏO¸BK þ†Œ #ñ5‘E­$¥MæVÖ¼‡de3ÞŸ½ß!ÅÓ‰QÒ›HMº&–ŸŒ´/~{˜é"ƒÚCªœ$RdìÈíu1ÍCæŽ5ƒ=é¸Ã¶ýÙë¦jd“{ÐÚÇ3Š ~õHþkä|$pÈyÜ^6M…¯#ïÜŠM%òÕ}² ¨ÌG‘¸ ^­©(¬˜xu-1Es{wGI£„·çÈŸ»(•tÙO™Ke’äNj£\âi–ÇHpo­OåÒ * ´²ßTß*ƒ4'¾#PÍ{û¹ ZbsÇìݨ¡o›põjNøLŸóGíTý]þ¿P÷層5FoÔ²ü’c·¡=Y+vµ× ñ©¹µ)ÃH4½£Lh €fã„FùzÓh!Ff•¶-Õ_¼º’Ö¤ê×{­Ñ&!f½ˆí|#¢^ý½€õÛ5îf„oµ‹ÊöyyÕ:¡¢tÍPÊHp3Mä Tî—¥|?µ63Eÿ€zšìlÔH…úbÔbfch9ð1·e W%ú8 qþéc:ýhÒæšÜS\M·~ ÅÕšCÓƒ¼þ& ghÊ"£™S ‚§Fe`&~ôw|ƒ ÌÝ¥÷©/„…'EÂQýð·5ŸÞ*MVöëÿѪó‡µ_ªE™ÃH°øbXLøåÌœx®~IÌWx»ì-¥÷ˆ=-R¤R)Ñ“…!U;w]ó>=¤•½ëIŸ‚ô™dÅW½#£°‰©m22ù‘øVI#óº°àèÎǸ#/J¬"Ù¾ôMÉKŒãN"×—J+Èå½FâÙpw‰\ãÐâAÞ…ª·ÓzÍÈçRX—šD‹ü¿¹dJQðyòëjivs] EÑk¾´i(ÞäÂÄUnŠ{B«I5þú t{àlŸvÊÞx¥éuËåù«Íµcpo!§‡ý#¨öC/Ê~ƒs£,e¡*ëÁC9{b%Ðíüª}”­{æ5TTa!ƒ5ǦèÊ ¹P;íÇå¢_7P÷ꢧõnWÔ—g¿ò®F bhû=zѨ¸ \óšd~ ¿…fr÷J´¢¹ ­ã ){´Ðn«§ðBË3WšTä~¢U¢í‰©ù½h½ö““ü×´=Wö”D¢íÖ ,¿îü†ö©º.C¾|ØF¢q¤‚Ú”qD ó¡ÏôIÞà)õà•½(²ß.dŽoɪw®¦Aß2º¦Î1ø"¡§À`Ò#Þ–ˆºAJD9»)Ä%Ö³J±ÁÛóbxó ¦åÝqŠˆ^WžOA½m¼EV‘ ‡B„º¿Ä +¥í}ä$IO:ÃóáØ×æ0­hÈÜ»ÐÌïYç>šHì2„¬†/Á¶ã’}SIäräè˜ø9Œþ…OT.E§CàShÂ蔤˦ïT3­prk]-. ÄÁçSú;Tsüþ›e®ìzàŶI­ûöý—³?«›­WlúKCl^@¹cÀ×ÐÃÖPÑ 'T{t}Srøþ‚_¶<"9'¨ ùíeùêONÒgèCû*uûBQh’xÌ¢§¨Í%S.%,½ÐZ£½C´ÚI’>-màÜÿ ‰–#ôÜ7³º½U¦Jó+¦Ðï~fébp. ·ˆD/°§†Ö¹2­isJpý?¥„_ª‰„ßñ?R]ÞyÂä.Ås…ƒÃ0õ«µxFftsn*R¤ÁÜ>s¶K‡R`ÁIRöûãø.²ãut?,÷ϰÛR…5#i=Yb$¸®–Ê44„ƒ…Ÿæð!ñÁ•µú†T$)mõ–“D²(Bí@ŠÈ”èÝQçjBOÛ£‘iƒ¿0®;d ƒÒ©SÊŸ¦±îºˆáwdöþR_íÚ;Ô¿^éö®C6ÎÀWa|/Cˆ©À™‘9o>]ë×´GžõÏû}‘·3ôÙÙ Rä+ynàÌB„-AÑSŸQXbß™6ŽÍa1{/¬†At­áý(õ¹ë®mƒʤ¯ÍÔŽÜE¹ kטw—poç‹Ãäþ{QIù¢ðÉÒ¨²¨éO[ñ•¸¨¤ªÕ’Kï3$ £†Q÷ÞÙ[Ô\Ü¥6\n†Ú…qwoÝËFÝ´("½rgÔË1ÿZ ×ø´ú¸Ðø©aT}Òc4­UºÐ©³€æ;*ä—}WÑâЇ"îKhùµæÏËv_´6%Ê^eG›•;šG% ЮYb5ÉÚí;)½´´ÂƒÝ‘”xè9§}Î#t»`œê%B‚ýg™>J#ÉDíÁÓÆ08]grº ƒÊM2ô¯@öÙ¨ÌÒt*Hw¾~5Þ×r‡ÿ‚¤™¨4.2CHð3ìè®y·Ñ¯ªˆéÓoâ«"˜Ïh³Om|ܼQÄ¢u¤_¶kÍê¹uÛçåíqÝ+Ó%ï?ÿûøTä§¥O‹AMís{¦4øB_{7xÔßôª©;Ô f“s£Ø¡‘¥ÍÈ*ö14~N¯]¥€&W™ýLo¡I8å™Ô»óÐÄݽdž~ôs¯ºzšB_ÈP–BóáÚÝ¿ÐréŸBñ†N í¿|¾gÏ›‰+B[ÕxvÜ9GhŸ zòc:Å¿ß_~ê]þoo=ÚÝnÕ—Âûw6½¦Ðë¤P“ô–ú¸ÛŒeC_¿[ùr%è/ˆüªs/JE]™ŸÀq©“› þˇœ/·tÃ8ù7öôƒ_œž¬“‚0aó©m®7&éø$-êÂTå½KošŸÃŒÇ“ –.0gfT!ýÀ ŒFÜ"$ேÃ'ýç°\f“ëpø&¬Ù¾‰­ÎNF‚&Ã–Û HtˆLŠhN Iô¤ù#'#™VÜn+N¤°ÄJß+šHåk|Gâ(Ò4ÎÎ\½ƒô:ƒÓE·a°[+0Ù™dEösÛÞFæÄ§âVŒ/pÇ1"Ë»#Ûq[[þ•ÈOë±·©¹X›–Níÿ„<éÆ1Oš#‘7=óµÄäco)V@þÇ-Ǧ_¡ |Nä/›«(LhœOt\E©5N®Rg¡øá=+2éf¸‡÷îò‰³(ý*žKç¤Êê3Ì”´Ó£Ü$[ÁVaÜûtO=•~,*^’ /rßàÃç㨊û&—&Œ®ë–Ï+nôŸÑ³%­-QCЧ‹~Þ5¿=îsBíjª´K³¨kÅr—+õi ,Ž¥¬£AñËE–¡khípÈ5Ñä,å îô4æB/ý³§Ï‘ÏÙG*¾y_y}õ¤HêÕ çAœÃñH£ÛþÛdwç»>Ä839›¬lê¾³‹—¶ñ)‘û¾'¬Âg8ÃËxÖ¶kô>ù‰ùåØÿæÃÕ#Ì }!©ÿŒÔÇK²Ù?VŸû YIÆŸµ¯A6ï‚6«çOž=H9~™sXBÎ|•)ƒñ› ½Ø›¹fÁ¹>‹ŠUºþðY“¡Ù“øÑó¡pBrÚ8+wiÈ·¦þˇEA\;Ýó7óKÃnöí„rÿb÷ ÕVPI# tÂ?hs~xl)Ô|šW ýfuÕµfÏë6žžÏRj} á«åUA¾Í~™§"5š¿‡f–²Ckëν)‡¡my=ºc§=tÞRö:Ó×=º÷"ŠÜ’¡7ŠUù@L,ôk}Žº´ú w?QY<ÊÃòDNoSa4úá£#vÀø_‚±hi?øíìÉ`(†ŠT¸(G`êø,³ªü~˜¡'”ß3ûf׺n1µ—Â惡ßKð×…ô Ã×{°üùÐØÑçõ°¦Ü4¯ÍŠç•Æ aCàpV 7«5ì:åtIŠzð}C2«€úkß⥹ƒ¢#Rµ>¶±AZn…;'xAZ»2·² ‹ñ¼¬S2ŸæmN›!Çâv‚õÈFìçª+d€´K$EÎ3oÚ]6úbâƒkÈÛ0#{»$ùr®¨ª8é¢@É› šræ­uJ»ë¢™z7êRÍPâYC}™Î>”ªÕ|ìq»e>WÚ¾«­E¹|ÌK¯À½c‹,̦¨dB§`®ëûh®º‹Ëê"ìfxA½ÃÕÞ}Üÿ(J5ÌŽ|:¦"€Z„ùË»¸Q»x&àÑMÔ-û~ ôº6êLyÉ ¢a~·÷ÉQ74.z ©”wÑŒ’sJA~/š«?äVΦ@‹'TGêWIÑŠÌN;rŸ:Z?/™æø>¶fV^_ŸOâ™[Ψ¹Ô/>P¦Vƃ·Ÿ–/t­¢3ÉeJ»Á<¨&vAvÅÂ% ‹™rú¯÷؆Žš–&ýDïë?€¬/‚Ýñ,\›9N=³.@ʔͻ—7‹7sf¾°ÏWýãñZ©ÛÔ}5Céªÿä&^:°Ñ§ªh–Š(m¾¯™wNÞ®›ùŠJJ6Ûç}~ä_ nU%FÔWŽBg­ÕC–êtõ·[ð,U",¬ê5RG;RÎACYLîqhèéŽðv¥‚F¿o”¿Ö.oðbÆEÃ#·¡±ê,yNp4¦¬†ýÞ¨ëž|iZÐDjÓ1  MMžš=+›>T[Yh%¦äÙ°¾+™åé¸B›™ØßZ´{”nˆe‡ŽAC{Gè\§XÜÕÝΩÃm&ÐÓ°‹E?²zykèŠù¡÷ØÄ™€ï¦v¿.x*EY¯«|„!2åúøhÉùfªªìŒ 98ɇŸýíþí*{`‚¦­ðϺLî=?–ÒSS½]…k0£­!tC&æØ‚Cn¬À m6#÷ø»céHwß, ¸0{V«+àÕþY$° JóšB"Röy·„a$1ÛOrI5ÊŠ,D|þ²ýŸ‘i¤¼n¤Êíƒ4îÎg‡Ü2®­„šÌbÎ(®“ ã árú`'2[‘JRDãêãN’Wé¶&ŸY69 ÓM^œ BÎÔÄI‡ÄäQå%wÙ‰¼nÑÞ qwKWÏygäWN ÎG±!YÊžÓz”¢ÐYb /³úåÒ€b‹Ã:/ÍPòÞÔý¯'QÚå«fòýO(Ëp3–N²åòŽí¹1œ‚{Ïhˆ~ÊFEËk¿ïoðá勃1\x«Ö{*àæƒ`sÔÌgµ Õ‚³>±žÓD †"/娙)ÿî~X4jŸUx½zu%EzóÙ)Poä¸$gp¼~ѯkÅŠFׂ³Ï¢‰¹%û}ë%4Û™¤ùµóš³s¯%^3D NEß´<”Rüy’ ­¢Îí²£CëY7—A11´=šµ$|Rí&âz/íCûç’¿NÚàã·ôfjsÐÉÊÞSDß½3ߤ(â©'ÎúE¢¯‹‹¢(Ræ¨Þ”Ôò…^òaórûJÈ#—™p-°ÿgŒjC,$‡¸»ßvÙì+ù\ ÿlÞ¿?Rp°õŸ¹¡W‹ãö}Ã-¼E<üi§@$á±)ùÆJ¶k4 A;½¶âòaꥃd” ùFYz×9O&@–¿OtÕÆs¿(E.²åª2™ðAvâE¯å¬¼ø3SÊ­ÿÿ?‘{W²óæ{ÓÐ “˜ÿæCîÅf“úïnñaµÝüÀH|•¸øüÃny(OZç5Ûk•º„oX š{()6„jf›3OÃÆóâpçð¨ ÜgÕ¼ñœèÊ¥à„¦ƒq¢úbʸW -ãÚÌ÷®Öº?mÓ’Ïö\Í…NÇ?ò-¾¸&* {q»z/t+çé-A?+•Æ·[00gmlF,Š=;#aô’N)×'/îT«; ¿9ÎK§%Ÿ`æ°,"˜’?kšyÙ ¦[õ‡…zÃlñ‰É4Ë.˜ïÿ8âo¥¤}ÉÃrhMî™FXãvÈTCç¢ÂQ $übÊDoÃŽÄüoÕˆò!IÄ¢>Ó$3  ²_vFŠljÂê êK× •e"¤=ëíÀx1Dv‹îk¯@Æ,Âöad>p³íùuYܱ#Zɪ®YzI/‰.#û\Ÿ¾}v)rÚ¤fŒTÖ!¬_ñšòVXÔFœD¾w6{Xî£@‘\ÞáFÞqÈ,è¯#Ц{§ÿQvF‰·O¼3¦ÂPª')â‰*Ê”†~þȈråb Å¿,qïBÝCÞ7®¨t$I[©ê1îãM;ª±† Á}ÞâPª}xó‡é±Y“Ú•D=cÅ#±žKð3Ie²(S~/$ŒsWÃ$õÍ‘Ã×aê¬ñŽwÇ`†y¢øw_Ìö”ÛO^Ž€ùÏò]Ű˜ù‚&æª,µœ7»¸Á‡ø«íb) 91›„¶ &‹ç·9¾AâÛç¾v{ ©ÿS{Lö"ù‹¤_ÎvÑHYýŽj’îÒ0W›™–"Ý©×æM†®ÈphYÊÀ" ŸžËÕ»(Lk—Ä×ã‘%½ê¤Zc6²fjðeÞAöI[¡UJ ä4ð6fÝ‹ÜeÓ<ç^ ¯È½ª³¸ÛksU@òõú¨6¿€gC÷<£C!Q6}â•zÓÃ΂_(æ]\ïpÇ%UÈGÒRzPZ|è4Cª,Ê|Ÿµ²»é‹rwR_^ΖŽÖþ'y„QQ¶ Éßnð¡E·ä¿pßÛ/2öÁbR>íä¥ê¨vÕó½õª¯°ªyCÍðVï?k}¨mtD>$ uÉÔY¦˜ ^-yãÙ{hà÷¤Æ«8 ¥DùÈÊ¥¿ hF§ÌU%ûÍ©ãL«^k¢ÅUïCû-èªx×I¢Õó¿_(–ÐúÏž{zsh{ðCŠ2¹+Ú _ª°Eûû»¥Þ›nð¡¹ýi#ÿ=è¤Ìùl¦OùÝKñTÁ  ׿(0xËô¥:"U 3ŸõÈ0ôpÅ}dª\ÜÀG\á¤×H "<âõ¶`ó½ ë0[¼ýšp‹ÑÞ¼ø¨úðͱÞÇ„:äÁ+Õª£4-϶ñ>ÃE|Î ÂmŒ®ø}¼³]_µ¿¸mÊýä¿ùPàùãÎ y¾Ž¤I  ²äûnR1Únêĺ+ì•ã¦Óõ ½xçÛ¹EÈZ\ôW½Ô±¡i9MðBηGáÏ.„OýìD—éaã˜x‘þß|ÈB]2ŸqŠUp§…ÿ¿|Ø"ÜðÎ!¾êÿˆìü’îšõ™ùbö®GGZY5ºQ˜rJãØIMý:H_|%2ˆÄDþF©IÆØŒ)”©uÚånr ,Çš¨KPtLHRî*]ѱâ5~ûäG-“f!(Šé¾H2@µ´ÐèÐÛ*¨aëV¨ñ7µX–/”Ôe¡öÔ—òÀ—¨;a@!2F‡úsåÒ®  q¬tÆ-š°ùœ-¿Œf&œ‡Ë­ãÑü×ï,An´$ ;?™'ˆVg§ž¯•¢„ÎqKû@´]ø5|¾ü’<Éói÷*Ÿ¿FˆNΔ»ZŒeÐù²fþÒÜS<•!ÈsXo…(ßqÕeW µñRÁ7]øaå>Ô/yÓ·â;²¾mô‡ò’§Û46ïéž³‘†¤š“µ1•òïׯô½|sŽñ4ý ݦîcâYìûgŽ¡{úâfŸ*$T÷qó}͵« ··ëæ|ñ-õ¥íórm ¯ƒâ;;ëí¡9V Ø9¡Zæ¶Ðˆ@íÌäxR¬|» êÃ÷ø¡þ‘Lȧùh¸8ª\=± Ôä7ª3L¡QVÖ•Èé'G«Œë ¡t‰Ér¡áͪ_O/4Œ¦9Ùz›@cVm²ò14µ9õ>g†æ¿3çƒbàU™ô —w5|·OöåâÚèo¨ÏZJØ@{J¥õx?+tÌ}›¼ÃÌ ]š•‰ÅKÃÐý…ˆLjGü0ü-ØTÐû¨–°µˆúîÎ?<ý¡‡ìÒ’f` ¯ÒKòÎ Q drùPþˇÎ\ c¼ cÑîãGáç…ȃIÒ¿½XCUàO‰úl„4L©}ùdò¢¦û#Ù?&‚Ù8’X³“*0Ñ^)’­³N0åÃÒÁ뜯vÞ€•Œu îÀú™ù¿ôÉÐ4!±àÃç×G9”ß›.Â@ÉU{¯×Wš å…®Ê7ë‘ú£¢Ëg™V¤]ñ½b ÜAÉq‘Q/Ô³{ý62% ì{6ÿYì÷\²wFVÇì;ÓJ'=lyÝ9v'î´î‚›—Û´lïøm6ÜõõWlr*?îsýI£|Vìk~í(@ñãi¶ æ­Ð“Tð ð§j±êä(Fu?÷¼»J|ýÚÝj…Rƒ”?ã­QæÕq’˜ïJ(gµJs20÷²»v³ß9‰ŠLìľlð¡´®cë(îó“« IiE ?ôÕN¨¼yâÀê#:v…ô¨éçidy’µÅ‚+ýèŒQçÇ}AºDÔ Ïè¿kj‡e¥È¡‘òž÷ß[ªÐ„møÚÉÈ‹hFX>W$fë{¢–«næxg–Í·¢¥E_úþM´ 6ú4‚ÖcR O ­m‚“ÁÛ×h×cìÆ'€ö~,§°Q:ø­R,:qwex1 3ý©¦5Ëj<5výTŒ¹-òUn1´Fê{cÅ>÷Œ¡KÔ¯ôiÓŸp†zÊeÁ‡ Îßgf>B’ME®v1áæÜpÏ«ÖrˆYég=4:ýÿÍ û¹!ÊüÑ(݉m¼mð_Ò`×W'êÙÐ?[õÕ={ßÔ¼ÿæC%¹F¾œ#yëä­ëWÃ!k·öhâÒM^|0d{2—è5-§Îmü\g¬ºÅhC?ú”¤óEm¾G½áöÞiC'ŽÓ ’>€O¡ñ#UsD&ÑÁâÿæCªýfìÚP,}þöî×gþåÞùoQðõ„Ñkò×Ç¡ü¯çUZJ³MßÛÞ~Q¨V,‰6…ÚS’ôÞðnî}j=ÂVõ5LR/ŽnôËô«Íh¡éA.pKC ÓÕ·åÜnÐ:ó`ÞýÏOhë¹ëÿ}Ÿ7tÒÙ'ÇåªA·—òmýObÐËqZ寓Xèˉò66‚È– ëš60T{BãîEåÌ}D9ã6ÿð9ŠÁ¯ u©å̘èÌѯR†ÉoG*âü`úX§kŽ;ÌêûTNèÑÂü Õ sX,‹2tR¼ËZq޾Ú`uð´‘Ž9hi4ì?“Š„Ïº] géhÌø†"É$1ÿ8•‰¤kÃÌç–n"ë. "¤zxþÎ:ÿ(ÒšªÄòß=Œ lßG÷ÿ}…Œ/oëÍÉ ³‚N¯Ë£@dóZ¬ámCÖϧŠe"{Ù‡&k§@ä«rým¦‰Üé#Þ(5‘Ž\íÏòYŸà.Õâ«ÝWŠ‘·óñ4?G7òíi“IÕ@þØ_wÏAA³ÞÒÐÓ(¬o²#÷Š>·‰ p²D Ë€¦e¯J”:¯ëž9¡ƒ2Ç¿ü->‰r»õ¯´ÛòÞ åÓɨ0u|ÀøûÒ¦}©üȇûìO|­¹‰ûû2Ç9̹QÍråys3ª7Þ5»@lŠšÇwrf5¹ 6Å'§sO Q'çÂõ”Qï’}¾š4;@ptÏ«4b­r_ü‹ÆËÒmøMÇß½|]BŒfÓYÓ4¿åÑ‚ÇçDºZy‹®@« 1GO5¡õ`jëžÛAhkþšÁQ³íZy3¼ÄÒÑþy–¼ÔJ’}X؉ND/¿ßcø‡æK¾ÚšŠãif/[UåLäK4 µ>4— Äïù ö-;§K¶ý›[û ·ð÷¿ñ¶åýß~ÒŸ®[©0ÿŸº…ÏÿäÃÿñ³níAÜÚWºU·~þ¿ë–õÿìMü¿éò!ÁÙ®}Þëÿî?ÜâÃѲ@5Í0øzëÓ[¹>?¨àÍúén • a>õ„P]wÝÇRÄjßžì@ø–môî­´ Ô†g?oÿïñZzUOS Eˆk®bW´®:ß§~}öß½¦ÿã'ßÂõŽc—+®Loö·{|òe`àü“[6gaè5?QùõƒÿîCä=¥c\H ¿4Úö)ÎÀ„?Ul•LFv“îVz Óü+t±Ç`––Já~±à6nC®µÕ”„enfÉ;¾ü°Z>C;Yí‚ ÄîÐ9#á­Ø0ŽJ$ªÙýºþ‘ ’ˆi_ªê³EÒï¬Ï–®2#ù¬7«z#R]É}ѧd´ø¢"%¬è3t (ï!ã#½‡5ÞY aЈ3YšŽ·“Wv 뛣÷˜÷Ò!û»gìi£ÈÉò]ó<óu䞯»vxy“/D §û"ß½â7ö(^DqÌs…fv±é3 h‡ŽYó#”¨ëõóy…Ò”ç:=ß¡LÏôˆäñn”£™1ˆA>+˨ísáÏÏì;h#Ý.„`¬9/‚ûQ-ÓøÉYÔpò,ßýɵD”FR|ÌP‡É+ânÂÔcÍËúbõ Ä%&öÐE#ýSƒk¤ïÑäìÂÉ€#hº³Ñî šO&}‘ •CËÓL¾ ×tК$àå“b=´)¨è¸ßÍv¯,­ô2Ð>®oüñR8:¶- ÆWá!åkWÒó]Ðy˜éâK²Kxò!è_¸ˆ|>M§„‘ês“œ¶*;tGwªå¼ÛæÃ-ßõÿ÷½Š-<ýo¿÷.·ð·…Ï­ßÛÂÝÞ·ðPæä©ê§G•4ü/'¡v£ÿ{õ]þ7ÔJ¥‡É>ˆø—ïë…q<ãù—ï¶øñîeÔ'~RÔúub›?·k¹èïâdÿÃóOêË(¶ïklíÞæÉ­ýÀc”ÏÉâ¡Ý dßËTñïe| ¹,ºzz&¥p†>…Þ(–®ÛQÐ×ô>l`¤´l*L`°ä»àíšyö¹r¢5VF]nš¹~;ãW³óÊÏÂ/óW¤žI‹0áâýãÜao˜´˜>e³S¦^½æ}73Œ—¨S"ßÂl¶îó´‰0ÿ<|ÙãñX¬®4TËëØÞ3º&âs<æP„;庿D"•9ûaá$^ n&ÿ¹I§ä<¿ª"“sUC RYµ^4Eš4î´Kž>HO8~Ø(/2dò$¯ /#ûðñËŸù‘¹œø)›‰:î(X.ñ¢D¶aòslŠs¸“k‘ÞþºrJ ˆl¥EžêGÙäŒ_÷ÕG€iî{4®ß‘ß$|/{ æ>7ã¥B¡¯Ôš:$ (Jâ)•Qpųž4?#Å=u‹«ÏfzPfçÊÏ}n(ÛÝ7ßüò ʧEÑMš£Â{nËå°¿¨TòíFæP î£<1¢ûïÔǽÓÉF5¾OäF„¨©Öa1цšb½fÂJ¨U^NbÍ3‡:îõìÐè@=aÍ¿‡ÿ{_UÓ¶¿ "¢ ¢" b"°°„qîƒ ¨„˜ ´t ( v vww·°v vwwÿ]ÎÌœÝÁ}Ôç{ßïÿ¹¿ßìu¦îëž>gæÌ!´}¦òìó…à½'{ÊÖìËÐ~æx·m}îƒOö&³gÃÁw—SŸ íkàwg–Í”Éë £Na¥›“B€Gê» 1yÐiÆœo-‡„ÎÅËöŸÐx]5>öõÄèfØc¦~óÞÐýˆNÏc— ç”KUœ'WƒÞYË4—¨î…¨‹»§²wÓ nÿ;§mª3PQôèZS“¹<ºûâõLÉþzŸéïýÒ÷¡ŠÆCÜ>q{Äß^ߪË}­Jï· ÆÃR¨l?=.j6§ëŸG¾œ¿e°Q£Ù) ÇÅCæóNO®ý‰Ü/+Üo‘óÂ8ïþ@÷Éäþ÷™è–ñ;–ì·ºÜëÖKßmCÉý.nÇ·B2RŠjobîÔÛÖ½¥ô}ñ³9•¯0÷—u ¿éÌ<œýåP…¡5™Ç'/ «´.”yÖØ¹ÒÑ»ü¾ ü}ï®Ã®ÊïÆ¼êæ&ªæÇÃNÖó_´0c¾¬TõÂÓÌ÷¶ÛïvšÆ_ê“årTš»˜zLoª™ïÝëÔ~ nå]Yh\ÔÇ¥ocX 4—ilŠX÷tÕÓˆx« Ìš=äå2¨¼wëͼP¥WRÑÍÁ`ð!sfŽÛd¨¶òD¦îSW0J_ëâšüŒ3ØOs^Œ“Ó­l _BíY·Çm,øf)¹¡Ý>„zþ÷>ŽŒ‹°æ+\,ƒåѯ:õžõƒFã²t¾®‰¦;¾ìmfëÖ_fož×lÏì»a»5ì.jní¸ª rx‘ӜƌݯÐÂãõÛi¯.#º{ýøê `W$Þ«Wœ ­}μû¦3ÜŒÍzEMˆwANö;Éóá3Ç=ÛµÁËpåÛÝ [C;Û37¾zêB‡N6Ðóleðvöø©f¾’öhtZÔô tìftàÊ3x«²ýèçtè¼zÚ“iIÙÐutz…ôšqÐmÂU›~‹¡G^Î;CïZƒ.çO‡ ‹-Ï[÷Ü‘-ç®}xÌ-èôØw脆5Эع¶8xYþé"Roñ÷²W-f»¼°Ô8ˆüñýb˸ɡÁLÁ[«Šæuf3…±—ÕØœËüxŠ÷5΃™Û®é“ýŒç'];]}° saös]áDsþùÑ%ªÅŽQÌ•·oŸòˆ¹öÚ½Rcocþ{Ûh"Ù—ˆö#Þï1edûKÇ™‡õ_:ûÞª }¯f¾v–+ùîö³Ï?.\JËa^ü¨tÛûíGæuõ3›WlÖfÞtðÖž×?›y»cíŽÎÚo™÷>Ëf LOàÛWûÃÍ®?oÈ|Ù³HãE¥‰Ì÷ð¨÷Þ' AР~áû\ þ»ÛÑ­—ç=Óõîßœºt³Í”þZ! ¦ ½vÅjß}g ¢à£áÜÏæ ×©ù±¦§¼@?¤Mì ¯.PåŒÏ£VîBÕaÏM–xŽÃTsó„±P}dÕþ5¨B¹ósš¯„šÏ‚›‰ÌÀ´¦H8dÏ30sO:¿îrK¨;þvåžÎ…PïÅÖÞ´^‚Å 1·MÙ`ÔnhãÆÉÐp±Ê”ó\¡IßÖ]ªœ Í&¼x6f–ØÛïô±œ Âm5"ÛÆµûˆï{{CûÅÈó68-^^P{Å.h^ôrúå>/ Ušæ3ýŠÁÀ꽺˜ ?À%Õ}ÜéêË õ›lgõ¯¯À-ëݼ®Û‚ÀݼàhÝ Àãj½‹ò í¢‡ª­ŠïØî^íÃO¨Yy6øô:}^{œøn¼ûÕÔ0üî¦X‰’‡BÇê‹NYϯÁ—¾~[:í÷[øiŸti¾zÆ­U ëÉ.WÇvné¢TÈ ­‹çHû ½šl_ø"5‚Lv]m³D þ4ö¥ÁN¨7Î>¬éû\нºe”¨ûæj¥Ì•_’öAÆCÔþèqPÑsáô{UL·¨R q8<®âû\̳ñ郡;³y_âÕoï‘y"EˆÇÍRωHÿOÚ©¯˜Iæƒð¾Ä=o3oÍX?žÉ;ÖÐbÉòdFqr[¥Ýcùñµk¼O™ì?´¶™i»ŸúÚe‘¿P•ì7¾¹yîîÐÁáÌíãz}nº”¹^Ïi­Çæ~ÝÉ7ö½ôejõ럖SMºoJkÑëåÌÓ(æyXÏ6Ì ³¸„€!óZ?fS^Ì æÍÖB1]<˜wéÚ~˜Î|žXõĤ#̧áë2G{dŠÞ÷zühÿdæÛ¨ÍQ ˜b»'WŒ¯Ûñó:Ti- /M¿²¿¨Ï7Ë=ššuª-]7 tÕÊû˜• ffŽªy*_;¨]ª„ÔXRw·.¼Ù¾õÆ Pm^â÷#O€QŸû»k€±§++‚É‚Ì4=•QP;5«}ë]}À,~+õ™P/Ä;Ô §Xäjô1óˆ‚@«mat7ÍÌi–=E>ÚW|FBÄѸÀGÀ\?0ËaðÐþ ÷|ÂÌ8æê…Žs{ñã!ºOÄûïéñoô¼ not;Äí·;Ünð¾c\ÿ‰M}iâÀ˾¶ÏüÓ&éºyç­vòωh|#ã"ßÈ8ˆÆ½‚Ñ)ÇW¿çÇC<Ÿ‚ŸñxŠöížmóTî£|ûAûù/«†Îá÷ñãçÄEï¦ßõ2d®M¿Ù`R¯#Ì#…Åm:2·>ݲl|Á‹¹Ã2»Â¼Ì¤û/æg>j͇® u¿ïXÉ<öø6³úã—üóaPȰã3n3/=^¹xÝÔ`^[Îöò­¬ êv™¾w+¨=ŽnXÇ4®wk©òú%h«w«5Û t{˜e%~ê •/®»Ã½è[/Õs]”®Ï7vláU{ztº\g<º˜·é¿4ª‹’—Ï—ÔÇ6WœÇ®š 5;íìmט…Úë_oœaTfõî;Š2C nv§ÕCÛ]zo¦Vþfl£™JËÞË>.æ›WÞ‚†9¾5G>¾M¬·óYx š5½#M}3ØœÊÛ3«æKN앲cØ÷ÞÕó‹$‘=]¢ö4§Ó=z>s‡æ/–¯®ya'´Šü¾ƒ]o`5‡³u—XëÌpMZ?Þ–Ñ' Üú›_lU½3¸×hÑÕ³c0xœO];j5´ÖrgžîSðŽÓ ZÞm´ïr©¹ÝTðñiºöƒýXð]’•¿è+ø]qúÚ5‚Ž•NÚ´6ƒ€.ÝûÕÕ|¶½õÑé–]šuI™ÜÎ ºæ[uývp?tëgvf]³lèa÷ÔáÇ ЫæÈ¨Ý›³!È@ÃWÏhDU¶fŠ‚¦A=U̓žÝ†#ž<_Ï\ñ/výÄ÷2usÞÀ|J½¾i¢j_~_~@žû¢%'˜b½ƒ&uôøñ=âùR ÝŽÃ^þZ O·t;X:ºBÅ'ÝRîÍþ •§ÌÝ/°…* ne}IÞg–%Ee„jÛµ¼7l F{_WœiÆ÷óNWƒõPÓï±Å«Â `Úxt\º8Ìò—Ïöªõ–š<87&,Žõ?ûÙy&4ðvýºw”4¶ë©š×­.X k~hîÍP°‰nòNÓ„n:yK À¾•ëñ·¹àÐóiš­kpZ~«¢h·/´¨9ùH«ÂMÐjçä×óçë2êA~áTpÙ}ͦO\!¸zß«åßÐÜ^¿È8Þþ¸¯sX´©x¦¹¯²6Ö¯®Q/­#ùyÓi+^'Lß ¾é[{~< ~ù³w÷{³:š®®“ùö1Ì8:qü£1ÐÙúC.ÃûB—‡O[Æ.Àü¨  ñaÐ}Ïé¨Ã‡§BÏ;­—·ˆ‚ïVN#­+ƒùÐø•÷rSAGíô[˜+ÇfŸÕ~EΙؘÒúåëG'Èú‚¢q>ƒqûÄápûÆí¯'0ìjÓ-–ÔûcîÅsZ4ÜȇøÜ£GuYÆÏâçC4¾áûC2îáðôø‰ã+zN<¸íñíϘ Z+ƒ%é3Û-y5зséÄáY™Ì•á™ú ‘ÃøñpU—¼­¶)ñ9xßEnÊ<ÝRmøØÃªÌó#'õ.lf^®ï=µñ¸,æuð®£;NaÞ ©dk¶Gļ=}K·í¶{üó¡É†))“Û3ŸmÌfÏh”Å|µUônìTæ»a>Û ªS¼¯ºÛº›ñ Ò&2¤ÕÖþ znò½°Ö@m–Ñ|³Õ 1d_—Âí@ktÑ¥¦ œ@gýªãÕçÛWÀ¨ÌªÁqP¥Š·Š÷”£`°ù°@µ\‡Åú·æóó/‘õ }Ú&Ù_Ÿû©Å@­KëŒZ-J€:Ñy kAݪq‡{¨7¸÷îSÁÂ`ñ3Íè]Pÿ^óÞvƒ†šk¬:¯ýg8%Ö3zVí6Úlú}÷±Ö+Öþ› vOF¹éÚDêÇÌH'Í7ɵŸ‚æNë-wÖ}6~ @‡µ÷ݵí»Û^ž ­S|­¾Žé®EáÎÚ&Ð&§Þî;º§ÀÃ9»qÐ]7ð|ß^t=”¯üÑ.=Œƒv³êÜù¸: ñoþút ðm¡«›3q&ø_t¾óíø¯aÙjÐCèxÒÿÃÀ黡“Ù9»Nè@çq³f3õgBWýUu‡µ¬ë¶^íXû.tØ‘ñݤ2ôd;Šk‹¡·ãÍíߢCdë½j÷Tw@Ýб)‘Á “°µÓÝ æÊ¶ÆžŽȸƒÛ‹²öFÏ‹–u<Äó°˜gCñ‚å&ûG‘uD<.*|üÓã!ž7ã§U\Oί:ÜöV*kEÖ ªO^¾’Œ‡d=¯âû[ÜŽ]sæÕ­:žŸçAçR]=6SggT[æ†Ñ—ù3<|˜[óã|ê5þÀܱÍ_éaséþ©ÔÚ¯¶”žß©¹¥Ã—ÓŽüýìûÚSk¤ud^½:+a¿ÎÑ=>¿ëF!óAT/rk—ÞÌ'¿Âqǽ§1Ee9æ0ߘŒª®}3˜wmuÂ?lÁ…«[kƒjû\ë¬ZàÂÓç!–õ=Ÿ/×Y—>4XWÍo¨yÿ]›¶‡@ïÑ“ÕâÝAÿ¾…yØó™`0ÝÆç¡û3¨6zÙØn5Áhîæ66_ ÀøxíöSêCMGÓ9§}ÁT÷NÛnc¬Ál½·Ñ…C5 ÞÔùëõ&zÅ®›cŒ?Cƒ–9*T.Cã:37]鱬úø\­=lz>8f[ç›­_±öà#°¯?/ «à8¸>î°#»18yàY½Ò$hþuÔ,½g¡Õ´n?m<¬UqGµñCÁeÙŒ5÷Úk«½«Mz‹ùàvÉlýž”™à>!Mèë¿<GøLU™^×glÉ® íê3Ã{óëˆ!VZzÅà·¢ú{Í˵ £ºº{—­! Sµ•…¸*t®rãFVìRèrèÁ®Å­VCàì»kv…îSoµYa=·Íuº2"bº¼tp‚z3š»Ÿ±›ZšÞ rצ1W^>/68Jê-éùREíߟbÄ÷¯8¼¢ùܾp{Àëˆxž”ž÷Äë…x\#þè¹Ï«bw²®€ÆEò|I—h|<Û·û…å½RÉ}%~>¼pñ¼ö«p/æòÌãÇ®8Í\]Sw_Þ¼åüyOh¾ôÖž'—Í©DÖîiw-.X¢ÂÜŸ>õdUëeäùðÑ‹}š»î‘ûKrn[çmºƒU¦ò뇺âAïó6ÜgUŽ(ƒyg]ÑøÓ¸×d¾ô£ZÚ®µßF1E™;¾M¼qˆùºãþà.ÕñùmLñ‹µÇûïÌÕ¶—Ãâ³ BÜ7Ûe—ú¨¼C?ÏÍî~{3Of‚ö¬÷íæÚÞ€ŠZɳm*g^ ë#º/ý9k"ª^¶ƒZ«-G;BÕ“ëjû¼Ã`ßx+í P½{”šÛ¨qÌҫͶÝP«·S†§³:˜.u¬ûAk˜Ý |íßTƒŸ‡©X°uôÈÅ`iZ°âÍfhØÕeÄ›öÐXíE·Ä¶6ÐÌäãû‰}n€µ†î’6ß_ƒíÛϪz`7²z|cC¥]ÐtN 8NôíÖÐo18¯™Ðe·!´l´0ÀÝõ<0ë¶§Nµªìš[Ë7÷ˆ†Ö†ÓRµלùžö—ì ©Sàõ5À}oºcõÓ}À3f‡Õ¡å…àeõ¦køüðþ¶Ç,'/Ú_Îí©šþ|ò“Ì¢¦ç‚o±éY·¶•À¿Þü‚O+èœÍñ ,µl1L»tÖlœbØô\§õ×õß÷eÞ¼h˜^¿˜yÏ^Ñ6©Ì|\PÁ{õùkÌçÑìØ‰Û˜¯ám´‡Ÿ~ůûw·ºQ9TvìôNï¯Ï¯sôÞ®{øf h [®>¹õhûÖwYý&t3ªïjØôj½Ú·ª›?è÷¯òr¤Mw052ì±T=»P=é9žjµeíhu0nÚ±ß7»`ÒwÀšþBí¡Ÿv³mf^íc\&CÝgYÝ’Ïg€Erá÷ s‹ þ”&‹=fM€F5Z\XMëžÏŽuëûÓM"Á6¬³E@ ع»Õˆ˜"ïÌĉV=Á±þ-Û!’ö¹X´îæyhÙfj—Œ_€yReäϧÀ^`àü}´˜ýìð¶"p½ÛÔÙøG2´²%°p9xˆ:?÷²žžE ö6y-¯“cT›mC»˜™?ÀÇÚ«ªmÜ9ð=–û`×Üõàoðepâ~oè˜²ÙøŠËSx—v;yéjèJÞ«ùþ>{÷£Ô/ øª-2<TÒútØ FtZXñícP^áâÌú§@Ó·v@±V"h7ÎÆ|ºá=Uñ€JcæF,õ= ún~¾ë^]‚*ë#£íÇŒ…ªI/{ÞÖCÏ­í"@u¿çŸÇ_ã[ÚoïŽÛ5Ïœï8@¦ö¦¬Eþ10›!|pxéW¨ÇL׿YÌ÷ëTÿ6  þ, Ç•nß¡a­öSzd˜AcËa†ÝЃ¦Ù®ãŠ|Ö킇͛ù=zµYzì:½›ÿ 6D®Ž¦vF€cÇ%U[Üç‘×={ZÜ1þp³!0½™´wÍÎ<±Ê<\Ü:9óæ&pí±I=óÇ+p{0/àÀ‹TpÏÜ(«žoåÔ=¸Ú>8q¨šIð^—×­û2h?,#sÉúùàR|àêx{ð=ÍÜ©eë~/öu²ý:šÍ Ÿ?y„ÿ¸œX-:ª]äû|tqÞòðæðáÐõHƒyLÇB·¸Êã»ì‡v>λ7z™ûŸ4…ÈsÝÇ´BÝm?ò¾´V T–Ð`®¸NTéÝÿ‡pû£ß߯íŒ~T´n¨h¾¦ÔxHÏ—Ró¤ø9?bâN?'Òã"~NÄçš*XG¤ß«Áãá‘U?ÎUwÐ&çãûfú½üžÜùÚ•ãýøZêýqòÞx—3?Çh’ùÛ ×6ö¨Á‡]û¶M[Á܇juÌ·cîßéÚ²ËÀ¥d<Äè>òØ­“-™çsoÌŒ9μþ0iEÿˆÌÛqÞÓš?eø{æÃeí±MEÌçÄ S›l~ÉÏëXÌÈØcŒ=ãŽå•LûÊ5 ìAõf~W¿±ûA½JèžÜ-é 1þ½¥îÍlÐ:¿ô_öt‡Ôɼ7*­˜ÿþf“pÐïú£In;På¡–Èâx TÍO™?l+¾ê\Ë{Ûp¨^Ð2¨é!˜OÞœyäÔzøµßîNV`fÓ527ÜêY ú^'Ì—_8æyf4Оã௠Þê7Šg¡© Ò½Ü¨°nf:,qÝ9°õ3ÜÊÖ»f“=.«‚ÈÖt–¹Þ*pì–`r0è8¯jçõÚå"´¬W3Ù[˘}'V,Œ8ìÁ·L^-†Ö-×,:ÿØ\šØFŒ³6aÑ}£hŸsá•ÀóÒéån>{Ák…VIJ(hwú“ŠûɃàSá@› ½ŠÁwS¥E­ŸÎ¿¢ˆšçÎHîSŸ Y1ù\ð|pmÊsès¿ö€=yеþÆkjäCàËßîÙÝ¿=9äýtˆ¬7¦©î¨{ÌýÅêæÞ =8õN_ÑyæJí‡=ïüZj<Äχx{ËPæZÝâÀSzòç~ãùÒßÞÄMéÿŸ6êBt€—?ó0è½íà!C¯â÷¶ã‡6Ÿöà¿.¯£WmüÖ¾Ìû^ÉÎ~æC˜Çœ?h»Žùä~¤Î úÌçÏnÍ­_Á|yb1÷ûùTæ{ó›©WçC•‰§bk<˜ :=Yàvp-¨ ?öœ: 4µ©:£%hÝÛܸh~Uй~ºeñ¨ÛP©îÌØ$çwdÿE?÷— žß/½NßÛœ1r”Ÿ7=÷U-ÿX4Ôžñ±Ï‹Û`æ²4Âðå$¨»O½–¯ãx0ïûòèžäd¨ß"öN×£ _ý¼æ»û@£õZ}:ñç{kUÌXì6ŸÌF¦ ÂQù?ô¸öíwìžœ’-¦e^{pœìY¬ÖšÏݭѤËЪêí —쫳ÌjÍUpéX?±_Mh}oŽVÆ»à69vzÀª~àÞ÷ë·»­Á³KT¼Æª–àû”¹¬>Ú­¾6H¥ øT>ôX«† øžÊ›½èuSðoµêe­mÐñ`mËäL èÔ¯»ºÕx'èâ »´¨“60]qtteèî²ôÕÚ3¡çèvƒô?>„ Á½@“%Bî2<5dk"ô9~jê#ûUÕâ˜Ò»¡`öÊë(\Õí-yn_0—­ßhÝEÖïéçò®âö7©÷¹A5;‡åàû\Ü.ñ~(̯hQḇú2_Šß§Áõeãa©÷jfúÑ,úJÆC2_úþ¼ºïFÒžÉûãÔý,~>$ëÿø=šJyM +'27[mÛ–ÿ–‡x¾½OóàTX„kÑæÑ´™ÖúÓÉý쳧γæÞ<ÆßÏâ÷LѾ‹w?Z©4›$à×9ðºÿë‰G» ™ï~ó“Â5˜âU'ÝÜnµ3§3½òÃAu±Õ“Ë+î‚Z¯Ä£éç’@ÿÅH§„7 5è@•ÅûU@gÿL½ˆ ¡PÉ>pЦWè«Åt¢Ýª¬^*¹‡¾UGľŒÃd&Ûte_¨žò½gQ¶Pã@ÜÜû§>C­¡c¦ôO)€:»S; >u™ ݉ó‘gº$T1†úë†Þq Z©ÜÌæ MŒÞ¤¯ð«/‚^¶®ƒÍ÷y_ªhƒ0À¯u˰l°go0æ5WƒCÏP·7‰/Ài°zØÃjбçù8s'hõ½ïÕ»;;D×<ÿy´nhž¶og ¸¾ :Ÿ{¶'´yp«êŠÉ-ÁSUm«¯ÑrðjûFeî‚Lh—(³vâ ð;pQ½hðs_ا¨±>ø?¯˜1ò;ì|R±]›6Ðyó4¯ /@×ÃGÕJß Ý[]|œ=ϾZe@gê<¨B £ üÀ¡á··é1ß`àä‹× ®æ½Ø&ϬAëíÅ….š3WÄó}¡ÞÈZ‡?EŠÀ"`W«˜ÃOÁ2Ô¦ÆüÔ¦Ðpûêç'N@áúGa»‚ÕIƒÑ¶ ëxÉèô=°Ývûáã†B° Œ[WsˆZøvý¸8¶\Øf}RupŸ¤z#RÒîö r­·¤ 0ívì­¶¾°#R+ˆu^AkÕÌñµ§™ƒëìâ9ëB¡M@ÅàÇ5{ƒGÓ¨ƒ»¯†¶–[[¤½aÀÛ5¶wDØ;h?ôRuõóóÀç‰NÅâEàçc0— 8þyÃÓ^M?íFfv6‡N_ŠÖ=k—]N ±,|û ï»´Mtr}¯žãzèÕÜÌ+jÓ#êy+°A‡u¼³õš:G RÇL/ºG¨ëá'º6çh»­ž<ñþ(æò§Z«#7%ãn/Šö÷*iÄáqûÅë ×½Oªd‘®#Òã¡Ï³yÍz¿W¼=#ÛíŸ5èÑqþðí}+ÿ~ò7ð{ÞgîMÐqÞÈdÓ¸/Ð)}ž_\ 3è’}&¡oA î®7yË~mèѸù„ŒÛM ×ù®‡JÚã­LÒ5U@ "1ÚS!$2F•¤—“ 1Ro*˜–$XlpLh’äº* «ÂKýÕl¬lCðµ­•#¾Vw´:a‹¦ÐÉÊÎA„­*¶øBˆ/ìð…=¾á |áˆ/œð…3ºPµµ!W¶äJH®ìÈ•=¹‘+råH®œÈá!á!á!á!á!á°#v„ÃŽpØ;ÂaG8ì‡á°#v„ÞpØ{ÂaO8ì ‡=á°'ö„ÞpØáááááÂá@8‡áp „Ãp8Âá@8 ‡#áp$Ž„Ã‘p8GÂáH8 ‡#áp"N„Ép8'ÂáD8œ‡áp"N„Ùp8gÂáL8œ ‡3áp&΄Ùp8cŽ ¶66ü¥-)ä/íøK{þRÄ_:ð—Žü¥)e“v?ªÈI-&8:]k„…"›4„ aëLj¼ôÚVæZ(sm'sm/s-’¹v¹v”¹v’¹væ¯E2¼"^‘ ¯H†W$Ã+’áÉðŠdxE2¼"^^^^^^^^^^^G^G^G^G^G^G^G^G^G^G^'^'^'^'^'^'^'^'^'^'^g^g^g^g^g^g^g^g^g^ÒÎÔ„6¤¡I¯me®…2×v2×öÔ ®.Ô#¥£±´¡T@Ž‚ÃqKªš†£§‡'RÑÕ’Óã¥Aõ¤AY÷ƒ×²úÈA>Žvß”äø¸ÈØdÙ ©€- –SÏs8an§!ûœÑ.Dî‹‹²Jpò_¹Ž³¯AáÖ÷åìQøMãJÒÀnFñ¶†‰KpŠ¿Ã…ó߉äì®ÉùïAòö~äüó‘\qDP îCò÷]XP‚ûÏÖ·9D|7˜•à!Ä{¸6ÿ0â?2†‹éqôÿÒçx(ÿÒëÄY.þI¤ß)àâŸBzž^ÃÅ/@úVÒçé]x`C žAúŸÉðå¥ã¬ÍkQzÎ>Ê)Ás(]çæÙpˆÒw> C”Î ºñ¢ô^ØÏñ_D龘Æñ_Dé¿dÍñ_BùpéÇåÇå9ÿe”/W:rüWPþ\Õæø¯¢|º*æø¯¡üºÖŸã¿†òíºÇåßõûÿ ”7fqü7P~Þôãøo¢|½¥ÅñßBù{+ã¿òùvó¢ü¾=M‹C”ï·ßt-Á;(ÿï´[Á!*‡;Ë?sˆÊãnOQ¹Üí5ƒCT>ww>æ•Ó=C'QyÝ‹%à~?ÜKqñ½þÅÅ÷?ô/.~”_\ü¤iqñÓYÅÅÏ5‹‹_$¿¼S\üÚ§¸øÍžââwŠ‹ßO+.þ¨V\ü)¡¸øóÍââ/ÞÅÅ_w·,.þ1™¯+¸¯I:‰¾ê[%æ’ç ‰NÚßÝ( âeI—ÐF ¨¼I ¨"I¼ÁxIWñE 0 Œ$•Á¸µ@Pc½@PSÒCÔ’NíOAI7fvF ¨Çæk‚ú5ËAƒwA£` ñi i ÀJ’‡ÖFÍPÀV’Ov=ût,s\*8Í%åÕâ…@Ъ›@À¡:0ͤ”˜˜àDéc”´ëRCYFcB=M=9..: YŒÓœz;Ø[¥Äö‹ëk›’f›‚ü58dÓ¦½«pÞMÍhB~à±m&3HðG6üà «.²u"£ˆŠ?º¨ío–š˜k&l&•fÖP*ÇÊÆÞJ(l„ꊴ¯M?W&ÇH›—¾lÖ¥†&‡&Çr±äs5>¸o¿’GM]b}#QL6¦VRJbjdjp4=€áGR9_,«D9"#¾_xoixä.c¨â©â:ƒF ÿÑ_éœéœ!OdëøÐ¸øèPë>‘qIÉÁÉBëäˆÐÄØÐàkÿèÈ>Ö|†ýcbÕÃ"£I`Ê>ÑÁ!!¡‰ÈªÑ78¶/±UèNžíûÆEÇÅbK„ä!G階lZÑ¡)ýBc"ƒ‘]-:%6_Ç„§àf¡Û?¹oV".5812‹¯ß§/Ž’œŒ£hKÒ—‚k±Ž4á¡iñÍbbiž'9®O$ö–¯” 3JS’©Í¢ãúòT®8Lÿ¹8dBV଴|•0AÙKHpr°[d"£i«„µ4º¢ú™$)EÕª¨>ÈÞRõB Ym)»Ò‹¯?²®¨É:‘ú$K*W¯Ô¢‚“p-R—^ÛÒ\¤âÉ>VމþI]”ˆë¤,µ\Ý”^ÛÊZèd’Ú+WŠ|-–Sˆ¯Í’ˤÐZTÇeÝä꺬´Òu¾2ïÒ?"29TQ’<]Þ….(ÒZ~^;¥ÕG¯Xð·‘éAäš”FHŸ>ÁI¸°«°<ÕÚŸk²ò3²}ãbâC“’dº$ä¯"®’4no>pi£ã‚CBqc¬õó"ü§Æþ‹EüÓÆ\ºì4Þ¬?m¬å¨ |£ü…êò“ÆW¦F÷Kþ%4¶¯ìýQ±äGµCwzTûy•ܬµ‰MûWÝÚ)’òo›Òã^‰šÚÅ ‚¨†Ùä»"§r54T«›%†à:æX®ø\+‰n_¾èá!2qÊWÚÎ~9vIÃüåds V&ºs¹¢ãf-#@T>’~à—#Kû‡_η’þD&vùªê€~¹¾Hº©_N¶´S“‰Ü¼\‘I/øËeÎw’ÍüK€¤ðü©èJºÚfÉiÉ¿Tt¨oFé§».i·YîdñYAê+ª¥“÷Y-®Ë-…Õ8$ó¡ âyQKÊ»#´‘ÇŠäåxòቻˆŠoLÙ›QzaD¿lcy¤ÓQJoþX¿fòîÄ¿¥—¥O-y$ñp>c=(w_DÅ3VÀOÅ/¥?%èK—;ÎZ.ÎÿÆTxN@ɧóÑX>=QOäÒõņŠï@ùÓåkI…ÃúYPáp~ÐùMåK)ý+°ãøÍ©xXOª¼‰¿åO—/­Î\ÿq:›Srèö‰Ãáúˆõ¥ê=):ßi½ézEûã|§ÓO祼;á¥õ¢Ús©òÅáY—ÌNk."rqºèz‰Â“r ûº ë'Gë) îßèü¢íµ¸Sé,Õÿa>œßt=Q¿ ËóÐýv§åcwª¼Y6‹C\^8ŸDòþ¥Ú–o©i9Ø]@éEµ:߉ÞT?#Àãd’ç>0žÃ‰9FqÙ±3;'˜M â0…ë‡Â…ˆ9 EöaHÞäî‡0{‡}>C¢pœ‚l‡,;#ÿÄBܡȿ?WÐlÚ³6Ò³' —†ôÅ垉ôÆå—µ‚ÃnH?/¯ÂÑ(ž]ψËDñúšq8@¥Ù‡8¡x(\"Ò/ ¹CéùÉýÌaøy}bQüáÜrë‹ôÀõ.…žÆa Jç¯ßmCv@ñB‚äåE!÷Ñ(_2Püh„cP¸$”N\/14òÏ@úà ›ºA>|ÂQ¨>àü‰@rBž£Q9' ÿaH^¤÷H3ùt CéAñq=ŠEú Aᬲãz‰ê®ÏÙHÞ@Ä“ŠÜq} Cî}‘ÞY£äә䥠rÇí> é“€ò%ÉÉBzàö•%Üq~@r"Pz³Q~y£pÙ¨þF"ÿ.È—6ª‡Y¨Þ%ú¢ð,âAå<é¿AÞNÚ!¨ޥè˧+Éë'æÐñDú§R|¸<…KAþÑ(<®×IHŸ0dÇåŒôÀ啈ôÃåÝ…KCòãq8ÄŒìé¨üâoòOFñH9¢p¸¿JAñ#ÿÛÈå["âCŽäâ~#ËCrpýOArЛDßdë‹ÒÛÉO@ñââþ4(ˆÃhŠÛc‘ªG 8’Óñ¶G<±H/\bPünÈüqûŒYÀa(Â8$§²‡#=p½Åí×£dCò¢îçpþG"÷ž(|8â ¿-¯_(’‹Úµ“È‚‘Œ³etÿ¿‚ÿ×ÒÿŸª´|e|åÕç¿-üKÝ_+ §È^Vž_åÿSz+‹¯È^VýÊÊWV½•éYÖt”WoEòÿ”Þeå)kúË«wYùxœôÏââû!9í0"w|ßI‡Ãχøy?¶£Ã äå`_ßÙ}p|á<‘½-òoõÂ<”¾ØŽõ#ú³ÈŽÐ“Ò§ßïâç,_JŽ/Ńåáû*¢ÂŽùø8½ØŽÓ…Óƒýq|œ^,ŸÖç‡õó˧ë‰ãáüÀ<8¿Iú(w¬‘‡ìnÈŽó×¶ÈŽËßÇÒùå° çN!–‡ÓË û—Ê7„¸ÜH½dåÓôÓéFáÛSáp:0Öóàç8œ_8]¸°¬w€@ÞÇÇõ×{RY?Gê>J?\Ïp:;¢ð¸Ü¾Üq<E?eý€2~E¼eí—ÊÚn饈_‘|eúbš§¬ýÈ?èÃÍ»vÎâÂ)Â@$‡<'#wz~cåÞ™ @!öÇò;Pî8Öƒ–«H_šO‘>éôÑvÚN­7OŠÒ¥(´ ‚ðt¾ÐñètÓþtþÐá±|eñ¥GQ|eú+C:?p|eúÒþÊÊOç¢ú ¨>Òú”µ<èö¦¨œéðŠø”ñ+ʯ²¶gE媨œÕslWÖþ•Õce¼tº”µEz*«G´EùH§[Q|š§¬íNgYÛ­¢tÒò(½ÉûÇŠÒ­ŒWY?£¬~)«÷ Ê]€Çɲ¦[Y¹*'•õÃtûV&_QGË£Ç7EùÞŠ¯hÜv£ÂÑýÊî+Õ?ºßWÖßýj¾Q÷ ¤>+Ч¬=)é÷ÊWÖ®Ë:Î+Ò‹.G:¿õÇŠòŸN²þ¾ïR4(jŠÆ?EõBYûUt¿¨¨éz‹Û"½µeõ¾¬ã»3 â+j§eוñ—µÞ)j‡Êê…²üR6+OËz¿¢¬½Ñù¢¨?STÊê›"yt:•¿²|QÀ#Àã䤜ûdni÷¹ŸåíYè}|üuV!‡£rÏ=¨²9jrñH|$Ÿ]âûSsˆõÂîÈNâMH“·c=pøœ“ŽíÊ!ÞÇ€ýÓXùøãËÇÃápº±œ¬ÒƒÄCïë¹XÎP´_` |>±ÓÜ~š¾Rñ1ß.ÿØhßFŠ™<Öë3 H>É„¹S8Äù‡Ë•„ãdeÉëÃÑåBÊá¤Ü›Ÿ¦³T<̃ë®O´=›Ò?ÇøçzàýÃåóï£JY ¯'‡óç+Î?º~ODõz$•^Ž®OÄž”ÃÅë*ŸD.N/.O\î8=؇Gí‹ÈåôÓxÄ_Y9àôŽð”—¹ñQý®?‹-ä—'ÖéKä¹x¨ö‚åbýq9 vÊ·O59;‘‹õÁˆó“ŠWª½ãp¸¡Û;‡Û1•ÿDo×OäN÷»ìäçòòÚÈó ½,—ŠÊ©”;–§ ßÅ'úã}.¸$õMß"jœ!ý?WÏhœ$áñ¾.j_'ý+µ_Û¡ ÂVCÈ"Äã8âc¼É'zÙ¢pf_cJ/´?TŒ÷ר#{-y{þkÊ¥SŒÓMóãsfõ:Éë_jr *>âãx ëå×üôGö7£äãüÊËã}¾fýäã•’Ãa»åÊ“ä_+ÊŸ>¯ÀNÞ_ŒÃá|ÃùPr7“ç!õ¬)eÇù…ßWPúáú‡Ó…â“ýW8??ý‘òÁzátéSáXyù8ßK•}v§ÏÀ|öâüÂõˆjŸX.þ‘ò·¥äÐ톮ß^Tx¬'~ïïÃn%'§WL•§·WW„f?N—€B\N-¨taêN¢?Î_{ùpX‰óÅŒâmMÙ±þ8?ñ~yª=0b=P>ÑçàÖ‡Ôº|P~â|Åáp=§Ã‘t›ÉóÑí ÔxÒ…õÁúùŠÒ‹ÓWMÞ^ÿ#<¨]–j'¸ßÂr)¼ç+êïÄØŽË™:gƒý¹Ú$©¯Õ~®7u„šô|ž’Ãß"c¥br×É‘1ä+Ò#‚ÈQn??4]3>¢Yhß8|D–ÄÚ/81Ÿ?§œ,ç ݬo0>ïN³r³è¸’þ$ªÉœ“œDŸ©#=•¨YXbÉ·òG®'Æõo&{Pî}äñ÷û+‚¿ß_á®Çßï¯üýþJ¿¿Â]Úòl¶<›-Ïf˳Ùòl¶<›-Ïf˳Ùòl¶<›gòlBžMȳ y6!Ï&äÙ„<›gòlv<›ÏfdzÙñlv<›ÏfdzÙñlv<›ÏfϳÙólö<›=ÏfϳÙólö<›=ÏfϳÙól"žMij‰x6Ï&âÙD<›ˆgñl"žMij9ðl<›ÏæÀ³9ðl<›ÏæÀ³9ðl<›#ÏæÈ³9òlŽ<›#ÏæÈ³9òlŽ<›#ÏæÈ³9ñlN<›ÏæÄ³9ñlN<›ÏæÄ³9ñlN<›3ÏæÌ³9ólÎ<›3ÏæÌ³9ólÎ<›3ÏÆ÷%B¾/ò}‰ïK„|_"äû!ß—ù¾DÈ÷%B¾/ò}‰ïK„|_"äû!ß—ù¾DÈ÷%B¾/ò}‰ïK„|_"äû!ß—ù¾DÈ÷%B¾/ò}‰ïK„|_",éKJ¬ŽN>.9ûQzG*“Y#Pð+õN]kE!‘?Ú §ÐŸzÁgJ• gKóþ\.~–,ån¯À]HÙÍ(;­=' –·+x–'þ~⟻Û(O§›ÊOú…¸{Pvj.„5 ìt|7Ч½ L?Eï\»èçñèzT*_iAØÝ-ëçîÔ\ÞsX*¥ÛQ^]Þ¥âSsø–Ø©úCÜÛý3O)ý©g[ü,Jìôœ­Ÿ¸{(pGg×;•t;¥ŸÕ‰;KÙµ[3ÊÞ²Óé ’·‡(àO¢ÂQýC©|£ãÓs!t}骀×õçî2¿’~–¼ãAÏÁá9<„çñ+žcÀ{ñ~Ç¢Ë!ž{Á{ñ™}}Í84Cv<…ÏÀs*S×Èë…çê’p8dÇs>xO5Ž/¤âa¾3yýp:pºF 55Ô°ý‚8Dý+;ÅŸzŸÃx´÷û£p.H.ÎOî]¼§ï=ÆzÌ@kxwF޼¾x¯9^ûGý+Ù«Œú×Rgàâ=ßn§!w\>SÐÚŠ•+)÷,yÿ‰fâ³.qúÑ\>)§\´v„óu‘ röY ¥'>ë ë…ç¼È^{$o<*‡> äõIAî¨eÇ-—‡ÏÀéÁ{ÌñZδv†ç>‡ \X.NÞkŽÓCó`ý‡¡5=\Nx.Ïõ âÏíY ¯~ÇÇõ §דl´ˆóŸ š…Ú-Î?ÔŸv:ÈŒCî±þÔ½ÔÏ–w¯Ï¯î‰)ïÞš²†û]¹¿j§Z¿òòüª]Ïïæç*ß~µüþ-½Ë«GYåÍ÷?Íÿ«züª>åÝ+¨LNy¿ÐNKúY±mƒÃ#ž]IfõÖOZéZ0Ÿ­ac½Òè=‹QlúöÚ®9ËêlápÙ9½ZÖ$ø°•¨ú Vã}ë©‹/¸`«Ûm8¹â©YþÓ‚ZOB“Y±á÷q‚Í1Yó'Ïÿ2ù¶#[íp7ý´ñ3Xï®÷×Ï"h\qîfv…Áê­÷v°8 Ÿšœ> \ˆQÜbµmËS=Ùº›Æ7ß\õ$AKˆuT¯v‘­mºòS«úÅ ®®M ÚA°Úõ#âgƒÅY‹m/—7Yƒ ^غ&§[ÎÇ(¶>!yÓ.‚¶©E7UĹb×ÙѵöŒ1`«Ÿ›Qד`…´]qÇwkm9ßkž8#[sÌ»‰ ÍXÖ&"vïtDZá¹$ך6`×>¶®ëʹ£ÔÛ¸æÍéðbÎÙOv²Ù&‚¦×‚ÜÃ(61ͬyÝÅ#«f9nd›‚póâÙy? Æ(VŸ¬ú%7Ž%ØÀÔ—ÐÝKlÛ÷vJþm±áúŠÕëÖþ,n°¨ÚZÓ•áÞèq[ÌÞ0æ?—ª#cdÍ œëô ‹-7 óúq„ àñ&9½1£Œó;: a1ŠkªLÞa¨•ÿ@ãB¿êb‚ǹŒœ#`«œ ©ägv£¸ÞШ­w†Ü'hòz_b•[sÖ­³h~1F±IšÝÚ…áõÍ⪳Y‰ø‰×rÄu½Î<Ÿt^ÜLZì c„§5ׇžº'ÀÈêöÔÕþv›`µ+_tnÔM·L|Û°} }‚U‹6ZöÞ€1D̆ķÖ?0²ÕTM<ÇÞ€1ÿîóq·¿ù 0ŠEmž<×èp#k¶rîˆP_Œb¡4{Žºcdõwj½ÚÒ¾ÕïüDwoT0FqÍñûÚ8  ‡™š‡šdad%…ûqôÎŒùŸï¬Ï¶dF¶jítc OúØu1|̶¬ñò¼A±¡™¾$§1âv/â>M»q`þ×·ënµ3ÙIÒ†²ƒ V4~õë.µ¦‹1ÿŠ„íÑ,±–¹nßͳ‚0æÜûx/ÞO£XÑlp龂 ©„ÍÉÀ‰ (pôè»yúÔ‘ÄŒ#h×ëÄìCçc„ï%áö‹Ý.H]ЀÅŸÝÖ?RŸ/6«Þ¾¨K¼¸n¯„jš[p? ùcX‰æë¡Ð{“ÝÐ7žPX\g{Í!cÞ’4’¼%Ý ÞKªSaíwðPZ?¶‡ÇŸkKrî&Á§;Ö;½¢ϤÝNUwx6Cê1Ç’|u"ø²¤GÁ+‰Ô·×vóhRR‘y”jýÞ˜„#/Žñèhûºù­×ðFÚÜÁ·%ýÑtøà%íhX¥ÕÐ#‹à§pÍ»9Z?'ŒmßæÉK‚EzcÍ¢FkðX2þ¸ðèwÞö@ïl(*©{àKIû'ømâºMA½ýᇤ6§„5ⱤKƒ“¥Ô[Œ¬J눣ÉjñAuƒŠ®ß]ã±™aÐ8«‰5õ¥à-%Ù­Ájõ·ÜÙgY*«]’ÎÁ¬Ž´8î>åÑqÇž¹sx,)uegzn_Äê–ô›šŠñŒNç‰ßVÖÿªåC†Õ“ bG¿;Ÿ8³†+ãW«Gó( ÍóÎðø¡D0A#Ýœ _Ýnð¨×«öâz<Öý!½“bÚ2ö޽N¬^RoÕy”fÿóñ<>n’,i"%¹%lû–ÇÉÖÞvjÈc!7”Œ^'VfkL­íZKô™ÇÕÜS{s›®>,e‹°lߘ®%ë`ÿm¿_½Ïÿûûù¯¬gŠ”õ9ôWŸËÿírýÓ¼ÿSzÿ©ßï>gÿ·¤ó¿-e—[¡’€ëge?­J!íþOáþÉMYXeø«ò~‡ëWå–'_”¥óO¤§¼ú•·l%þ¯êªRêKå¡©¡Ñø5fUäªbƒ/l¹ðõä>ƒ-Ü79Nº_RúmV¹• V€ÞÕ¢O!µ‹ƒþª(YùÁ+Ò_ÍSôUŦâ]"6â0!…x…”ÞU"¢ÐBG (t¦¯œáº–¢]Cd¥^ñciÌ’G¼ËŒ rÇ+¬³ä‘¬XaÌ’G¼rO0Kñ)üx…œÆŸ%xÅ‘`–<âSåñnú+³äïÊÁ+ØätwäŽOÇH¾¹“¯&`Ì’Gút úëäôr„¾ Y q8±<Ò_à §ÓcÌ¢P,x… Ka…(9í# W° Šå±Ô).,…(^ÙÇØI@! ß)‹B±<â¯þd)Ì¢ÅÃ_#ÈR˜E¡XñŠ/A–Â, Q¼R§¢ ðøM†@.µ›FÚ«–¼™”\òd‰E-)9.{HzöXì!³›F-=48]k&¥$†‡&¦#«NrbplR|t0‰§ÂñšcÞ5~hà.þnu)¹ú»ÕåïVÁß­.·º Ë¿[]Ðåÿ7[]~ú$$·K´äiH‘ß¼×'2m^¨ à¡È«Îð0¬ä'(ã­4Sú›EìäeÇ“Òß)Eñà³y—Ò£DO¶Ý•>¶m’ôÇz²\x7äî°=—ÖÝá¸}ض8>âñàxø[k$?R¸Täâ¹¾ãÜÙ͈¹{´ãÐkNo%”ŸÏ%y^S‚s]b^JÌe‰›ô!ó†%ù¯R ùIÌG‰ÝP‚$æšÄHjU¤—/JÌWêLZnÒÂÔ II îM¶+ó÷Zê )%÷RÊk¹Ç’*£š%ù+..–§B×͸ÔàÄÈ`\w4P™˜ÇbñÇ3|rÏçì@”§™7 ,ä0{‡CQø‘¨lG¡AGqÏïìhôè˜æ p¹ž—p8 ½¸;™{þ&[&iFòRΧëO¿¸øoM|–cBš+/CGïÃË_³þós¦ížq”õ[°êÔ¼a¬×º%+ögÛ6àeî¼mçXiX{•l»þ¯æn²ÒcÛ¿ˆ22õIe}ujgæ{²ÞÓN|5ܚ̶ :Ѩ׼šl[=ÍÞqgÍØö…Ëk?œužm7bµCFå\ÖuѶu™cØ6z÷<º&­ 9Àºã¶óÞ@}ÊbÖݤýÙ7+¿±íO„t«µF‹õ_àÚtiFÛ&w´Dá¬Ï½¼[KXÖ·f3ÏÇËMÙö i÷ÛEÒ…© KM)ð§QY¸²º+Ó‡zALa=TƧL>N‘ÑÁ!!¡‰hüÒ” y*‚Ÿ/Èš ”Q£Œ:e4(£I-ÊhSF‡2º”©H™J”Ñ£LeÊèS¦ e (S•2Õ(cH#ÊT§Œ1ejPÆ„25)S‹2µ)cJ™:”1£L]ÊÔ£Œ9e,(SŸ2–”i@™†”iD™Æ”iB™¦”±¢L3ÊXSƆ2¶”RÆŽ2ö”QÆ2Ž”q¢Œ3ešS¦eZR¦eÊeXʸP¦5e\)ãF™6”q§Œe<)Ó–2^”ñ¦L;Ê´§LÊøPÆ—2~”ñ§LGÊP¦eô¤ýª²ûˆ¿øÿâ_ü‹ÿ;‘üÅ¿øÿâÿN”ýùŸö§Ÿ÷éù:|y =ßPÞðÊô§ =ñ»ú—7?Ê«/ž6ÊòOÙüÍïòÿéòþ]ÿÿéú@óÑù«,¿Ë[ÚÐú—7?èù?º|ÊÛ¾•ÉÿÓõnåí_p¿ú»éü·ÛÑïö³Êê²|¥Ã+óWV”µ›ß'Ê›Þò†ÿ·ë5=_^}•ÕŸòÖ§ò柲~¥¼ùQÞüüÓýòŸî~·>+“G·Oeý¤2ùÊî“hyŠîWi9t=§×“4(;½^D¯ÿ¨Svéš“¾Œ^ߑʬ,c¯€ÂËÚ¥ë\†ÈŽu”†1FþÒøŠÖƒT(»*e—Æù§õ!eFGvýH*_º.f„ìPj ø5¸ÊH÷Ÿ­/Õ£ ½~T[ðÏëIÚ2üÆ(Ÿª¢¼“ƥפúÉ®iPv¬¿TnS„ÒuÅj(ÿ¤ej#P¼>$]“]’ê&»þƒË¯ò§×oTeòKšŸÒú¥‡ÒTSÀßsáµMeíˆî—”­§Òë¥ø=8lpÛ“ÊÒ”nô©ª@~}´ BEë¡ôz§2£!“&È.»*å—­¿&ùuJz=¯ÉJãš ¸>C_ÆÍÉÊúÙú$mTP©†t¶FvzýÑ’²ÓõQõ«ÿÑžWýOɧçwU¾¢øŠæ‘×_–õ÷»ùóo—Ÿ²ß¿•ÿ*÷÷»|¿»Þñ»ú•·~ýjùüêïßÒ¿¼ù÷oáïþþ[Ú·2ÿÿ–úûßö+kûR¿¬òËûûßRÿÿ­ô•5þê÷oß_”á'ûüç£ÔþÚÿÚÿÚÿÚÿÚÕ.·oC:ßS²‰@zµC#6%¦O(>Ã@-)2ƒœm èüµPIÁOwÙýX#(ÇÎN¹½ jQÁIÁn#ƒž¥“Váâ##ÒŸøœ¸ÏWÍ*ÁS_9w±€³Ÿ×äìgWsöÂužáæørxf6ËÉ‹CrŸsáÎíDö¥œÿÙÊæmàÜÏoäÂ]ÈDöŠHî$νÀˆ³_¨ŠôØÁé{Ôšó?ìʹ¯úˆxÒQ8î q¡%þÜ.ü%c¤§ÇInw©ø4Â3=8÷ ç8<Õ…¿ËÙÏ4@úÝäÜEñƒQþíDîÕQ~qñ #Q~Ìàô9ˆä¼Bz¾Eúd¢xÝP¼¡(ßÌ9÷㉞lŽâ åð¤ŠwŽ“VŸ‹wåï¡'È5JÇp”O6\ø¨œN¯@ḃwÄg5‘ÎwÐø\].Þþ[(ß!¹38<ÑÉéxê£ôê¢ðÇ8<0‚ wæ9NJ—*âõçðôf¤OK^—ã?ÊÙ ærñN¡tŒGò¡|«Æ¡˜;¨E¼é³o6âëÎáa=¤·Ç{ì4’»!‹ËŸ¶B5·àdÔþPû…ÚňB„¨¾3ãpÈQ³¹©â,.ˆÃÁ·ŠrôâÌx„H~&Ëá@§¥8ÅÏXÀaz‡)ú¢/˜ˆ‘}QD„ɉEñcOrEႜ dï¡s›ÄôyTøœšœ?Û¹sñÙ>ÈÍ¥›AöDäŸÌ¥“í¯¹ã/¯¤£ðé äqwGÍf"Ì6–Ç!hCðP$w(’;tB.ìðx„HÎWnìH®\ÙQhCñh$w4’7¹åò›‡øÆ›!D<Pú'pý;ÑaÂQæØ#DéÌAs¹ƒ¡ØaèKO¹NAˆÒ5©"‘üIhƒõ¤9‘ÜI(ÝS¸ÐØ©Hï©,B”?Ó^ÓÆ!Dñ§/AxÃÜyâ1èà¬2¶/úëæ¤} å6€—ò”õs÷Á¨þ£zAÜ¢ö…ÚqO÷¥ìñòöd.¼8µÓ„ Ñ´Ä¡\~‰£åå’øÁ¬¼¼^?þR>N ïÏ¢ølPB.ÙÑ=A>½¥¾OÚ*Ï }yü¥¤Ab3oËûéJÙQ}n&ï> Õë¨ Ïú©^ìH¤÷(Ô¾FÛÈËk†ÜQ}‹ÚÙ8Ô.'àv†ÛÒg¢±¼œùòd'£ö‹¾–‹ê}n|¸\$w’…¼û$î>„ŒÒGÚÊ·©?nSPûžŠ$˜ê†p‡Ó†q8C¾¾³3œ~šË<~@ãÓˆ9¹â!¯²f£| pvԎȸ3g©Øa*÷q¨Ýâñ iÈ= …ïÚcRo8Ò/æ6gÏDo,Òg Â\Ä;öµ|øqŸžîTÉxÕ•ò¯&oŸTëçr0/¦¡öŸ+–7EŸJÎß,ä®ösùÓÑ88uàÏýI8­ôáôTÞ¾J^ÿOoÈùÓÄÿSñ—÷w2ÅS&¯¼|txúWÞx¿º ¤L^yËGüKÚÏ?‡þó¿ÿô‚Ÿúýj}.«œ?¥²ðÿÖïwë隯 áKÚËH? g9žmû´¨ñ”–_XoKué7FX·’/6µdí•|jŽõnçthÙ¢lÛ8·FÝZ°îk¾%ôb=¦×-ÊšmÃvüd’/ëQòŸa¬×Œ­+WÙÌz·íWy†ÛÖ}ÉÆëqWŸ‡g²žÒïæÔ˜ÊÚK?·xÛ.zq‰H¶ýÄkÒo]±Þ‹/0–ÜeÛ¬¾¿Ê÷ݶ ÷"¶½Õî•ǃ³mFÝÝsï8ËÎYV«Ï9?–­!ýæ [w’ÏøJ—$¼Ò¢¹5bÛ:Ž]ömõÖ=yÓ.£ª’ô%H?ñ¤Åúr_Nb=“–N-ܰ€us?nÿô¤Û¦D¼뙚ä}KïÛA¬+ývë¶Ç¦Ç°þWÙ¶%_ªÀz¶Z»´eW¬û²‚ÂÎO*²­ú½úì½w ×Ýû?nïdE22B ‰$Òu½ì½g‰¤¤´ˆJJ iY¥ÐNƒdï̽S"{FV¶ø¾õ¾»ßýî÷güþú~ûçtÎóyιÎx<×u^ÇuµŠŠT¿Š¢ ¨~¨¤HPâøMCOXªÒë A}!à—(A‡Â_½~»$AÉ8çîA 6ìÒ|bš aõÙ¨ÕŒ– r¯ì²Ô·]õÅñ«wZežôp$hÉ‹±†“ ¼KÈèâ=u‚ʃ…:¬ïÈ*ß#µ4ˆ‹æìšFÐÒØ¸|xæAù¥ö AzÁä8ƒc} Aã{ä:_‚æÂðï•´m=Þ@&Î^\×0Aí{àJ‚úÜí5“•M-eMf‹wõ…/Œ)yâlNÅ tÖÙï¾mXKض°{dS ZßC‘û¥kâ?ùR‡ ò}˜’…€A­…x6²Rõ-#žä ­cVtÞ°í{BN‚ÖS¯m áUËoÉÇå~î'ãYŸ,OšÔâLñÏ´%¾‡ì"hú,„ÀQ%hQlº¶Âo?A½TÀ|\’œ æ^¡ù¦`A½…›–;àAõsTGíU‚Êš•'¿¥äv2A6|á…@‚ƒM{åňëÔ·" xñ{C\¡s{HÄ *R<ì„:‚ò³=>™s¬?ñ³d',ù‡_²Ç•Óß#_XÔO—ü‘/ù‹^òÛ½g1¿äW~Éu𳿀—à¼hÏ.ùE_ò[®´˜.ùe\Ô§—œú´üHþˆ¶t~³Xîhó#]ò¿­¸XïJÈ_Ç·äW;dQY<ßYòo¿ä/zÉ?8Óbù’óàÙiÐâ9Ð⸖¾[?çgÉÿú¢}þS.¾Åòkµ?RÿEùœŸ/ù¥¶·ùKû?ýûý0à¶‹ý_|Ïw±Eÿå?åYò_¾x>Eð\´—–üŠ/Ù…Ký/ù?_:w[ò‡ïµhÿx.ö³äç[ò·ñŸ_L—ÎK=÷Í’ÿùÓ‹y·EûeÉùÒùèÒ¼».–/-ëX,·ç[œÉ¿¦Ž?Ò½‹éÒúÿ´{Û³üMÞ¥ý¿ä÷ÍbúÓm(É¿=¿[z¸´¯ÿvþ³¸ÿV¾´ïäþúüg¼¥ßÊ·þ–—þ-¿„Û%.µ¿$ŸÉâ|lùƒ<†•iÞþ&×.7ÿö\m±ý¥¸‹ëøs¨,¦:ÿ~ž~ŸÏŸõ6ýÖÏÒ÷ci<øÛs•ßò‹íü6¾¥u_ÂÉæe)nÄâ KóúóùRƒÅú?û[ÚÇKûiÉi¬ü_Ÿÿ>?ó|‘ûg ¶ßê-áø÷ý²´–—.ɱtκ§å¯ï;ÿ8§úù^”÷ç:,Åÿøù{Hö_ë+ýUþ³®ÿÖþùOí°ÿ´Þ×®üß:ø¯¶ÿ¿Õÿÿ´ŸÿTîÿ´ÿ©<ÿÓó†ÿ¿ÎCþ·Çñ'ùÿÍ¿Ë?„ŇKxýÛs¦?”ÿÖïÏrš?”“üû÷~Ïÿ.ÏßÒßÚùcû¿—¾§hçOýþ­þú_J~ïHþ¿Ûù§ùý›\¿×[*ÿC?Z¿ûóûÇùÿÓúþ§ëù§uù½üêÿÞß?Îû?Éñ‡ùûþ¤o,ýûÓóÿ´Þ?•ÿ)ÿ_=Où'9ÿI®?=ÿS?ª÷OóòOrý¾ÿÓyþÓ|ý Ïÿ$Ï—ÿ4Žßëßq÷'¹þS>ýï®ïŸêÿWñƒÃ¹;׎åáôB’ð·çc¼)†Ÿ/¬þ[yû…ï°žn›ã£zºŸåsß³Ûðó¿úù¼P¡ÿ×÷àGPƒŸy*6·Võh6YLÇÓ/ ½ð¦àÔú84¶ ðøßäùôãýŸù)bÆ­Uãg¾k±½ãÁ÷ß›5ú%Ïb?‹rÿMþžÅyü>M;ÿÑþé]”wü7¹Æ~£ø5/•óoãšüÞèïýâ·Åq¼ý1ìüÃ:͵~ï3Ç÷Û:àÐbýéÅ÷~Äþ™ü±¾8ñ£éy—Öÿg;3?êýÌ/q|8þcáü_÷ æÇ>ÀößêOüx@¶8Ÿ‹ûtl±½¾ûìçûo¾Oói¬[œ—ÑßÚkúþºÎ.¾×³u|üóyççÿÛüó§ïã?}‡~¯÷Oxÿ¯ê«ÿÄ?êÿOãý§~þTïøù¿sÿÄ?ÿ4Ï¿÷÷Oëû'yþS{àŸæçŸÚý§÷þ«rýWÇùOrüy]ÿr}Kß# Ø8æ²_bïÒMtÚÝ{öØ9ºü* vq•ØûãÃ÷ì¾ãÿšýk¤‚ßâõýKà‚¿Æ7 =¹û€‹Ä¿ü=²Íá·w»ìÙ¿ÔÀþC»%v/yö§:ì¼ç豟³s°ÛãBò×Àö$ÿW6øñ¿ÿ ðÿE$ù¿7àö‡3ù£™åÄoØ ;WG‰ÃGNî?°ð¡[¸¹¾YÓ£#›üi Ö|Ì5ïTJOð^ s…‚唇»ã¶BÎçâû¾×ßAÎ n'™äM}%§5̈ ²዆‡Lc÷Õ/ØÂë¦jû⬷?ÓÌ¢µk?uìüíû÷4!Oç}iék(Ⱥ1·&ŠŸ­Ë˜Xçåf˲„ Cå‘÷„À÷P¥0pi4Ϊââ¶‹@Õ@°Ì #1¨¦§Ê¾ÀNö3­ª|¾,ªNV»+˜A}PbATújqÞâýÙÞÏôyª¬J¿=TvÜI¾Þ÷ág½j²hJžPÓ&šÄ3 u¡ýôR‚PÍÛ26ÊfÌI‘AãŽv¶íм×íøO´ mŸ´}n0-޼z!ŸÒqèª'maN†^ö§¥GùžÂgÏw {-á W(‡e ìv«[ñ†çßP1¤½„Ñ»´dǪÖ˜ò°œ·©L°_éñJ·‡)Ó·©7Ã`æ7õI˜ã|TË{þ’œyq•C Éu…ìu…‘<ÿÁ³é&R>üx­ßC©ïT%„¯ãEÚ—Ï" rª¾Ï-‡D~ ÷ç5¼L·hôµ 7# ¯ß¥SªþÈš§¢´ W\wÚK†îÁ–#!:ÈùhùœI¼rµS²d‰!ïÖ·!Šú´Èg¿æŽ}$ò7Íx¼3‚7Ž_6œG¡@•Gª1Þ(ò¶¦h½ÿ1\·i)!u5Š•¶ÕN>à >æåÓ,¸Q8Ì6#¥n8å~“ø€›¸ÌwúY² Le¨—ÚÜ}”ͯ¬ÐCy2K׃bgPá̉R–>OD‰ISÚÍ ¨H§zXøæATºQ–_<ôU¶•’Î]ÐB5½Ô9¶Tÿê-||~5ûØO0]ÒCíæv‹#â¨[»â¸LÒêWT^T@R4øLæöõH~Ußøa'Sóí>¥`&ü*ñׇ'ÐT)þ9'{1šöÏøvÍ\Ájõ-ºRNËîx‡;ä ¼Ö3¹ eXd,]‚Z1Š ôZz µoˆ¿´¡.JâA›ú¡ºAÜ5Ÿ=Ôu—döqÿ4•ì>ošG@ƒ`ø”ýÄUÆ™ÚåÛîxAšäEÏêe"Ü¡tNòf$-{¡¯— Qé¦zò¦¯]Áìcjq6—ìfŽ>„ؾN¯.cü™.•Ççk˜íi‡D*ѲdHòÒïö-ûÙnªªlÙêw“±¡ÊnC ¤_;oFzØ2FÜ›C^ùýÂó"Ηp5'ª+~~²õ?ß•ØõrHŽ4n=391£Çƒ]!7Ìf~Þòr/Û= |kW (tlç¾ EAg+Iã¡ôk-¥A,TŒˆ3®…ª`E“„šnf¼ƒP£¿"—å5¼sàÜ.ç¢ïϘ&(†^‡†n&ûåæ©ÐÔ©eL¡‘-QÕÏÉž@›÷ýMë-¡ã IÑpäGè2’zÜ>Å= Ã÷h=”¡7•qœ¾¿>»0óNÝ“€Sl’—CzaˆôI–Ï•~ö\¶‘/CF70¬œð‚1š¤£!.0Á3zµ/ë!L鸥N² f:æßuÀÜù{ÇvœE’„½wÈYLl§Ë!g’$¤Ð¿"ÓRý©ÎQæ&œãCšæqaÍCHÁdµ{»2zªJ&âr|t¶ªYêõéU7"[ÃDòkš5ȱBêÔËÀ7Èyë‘ÎQ äve<™äŽ«Ÿ)í:Òz ùýë©ç§)Qð„eN“/  í~¸ EÆÈŠ<¯Xãºj|æœÛPü Îªœñu(©$_Z1ŠGs7yh;¡tÅÎSd­¢(Sõ„óØå“(;­å¶ë¡#ʸù&œ¸‰ ‡§%ï">œ~P×…ŠÛý)º3.¢2¥õ³ÀuÒ¨R)ÖnQu Õ’Xo«ZïEX¦õ‡kÃPëí¦#n]‘¨KåŸúæÅșߡA««˜™k ñ ·4×Ñ¡±ÝÖmô_kÐ$íQ"Ç´ š­2Ú© “„æçZI ë«áMÛóœØ—½h«›÷ð)Zq;Û+œ<ˆÖ×>Hù:ñG¼±úÓJwÏAÍü½'«OCÅÞþÜÜW³ÿ„Sx½­ä<ÙPÍê$9¢÷:T •![eíÚѵÃ%Í”{Æê,¼.Ðþм¼^ó_yãÏDÄ…—Žyç#xæ[ñü doÙ÷HØ!7ëцªÁ(Pž¿†}K9—?«z eýAr¼ÒP©íÎ7hU… ¶†CÕ›9óAP-t±vN³ª÷´Þ<ÒÕÇé ªY¡zýá}C{Œˆ<Ÿ¸eÅY¨bMüxÝ*ÝîæˆqÈA¥êƒ{†ÕP©æ?·n÷>ÓBåÇüª·-Påbò¿ø­òìõ¾š¨;ðQÅ? ê-Ì®xùÀ‡ žç]¬¡‘=lù‘#|ú0>½ÿÐEh%iÚÀhÝmIùi•òÐq“§dr t½ŒÝÇÈl =Ÿ?¨”æ‡Ï¦ŒCþ$¤0Ðk£¾>h YéÜØr†{V]– ­…Ñ€iîÉSëÞ?üñLˆQ^¹SŠQ|Ç‚=aæ^ÔÎm’0·•ùT©’\x¤Ä§‚d Ì7O.EòTÒ ?¤ô·8_œ!ÔÞE«&MöΘ޲v¤o„‹çÉÑüôÛKþ]Èà¥jzÎYØvïš {ƒ¬‰³žÐžÇ®% &)†È±ïhQxž)r^ÔØûzÿr^ª´uE^¡ ¼JþzÈg}ÓùŒ Wï”>´y‚—JV]ÔC!÷^©2‰¬’dR}Œëè_ Ë×C±[FM»©pƒ¦¨»£²3n¤e/Y½k¥NOôØz71‘,ëèG™Ú~3ïß l ‰Ã는g(È7{¸NêÝv‘odÍé­H˜gÃÀëQÉÇN#Oâ)ªHÝŸzÌsÕ¨÷?‰z· Õ''9åw¡æXãZ¯i"¿FwŠDê¢îç'ái¬×Q¿ê”¸óF4è»;Yߊ†£ÑÕaLhLI™²J¤Mx³³V‹±¡)A3vJgš¼ê{Ö,Í2fìe D‹Î±è;"¿YÑðÓ$ùmÇJ`³E«eï›ùïhùÍ¿¹9á$oçWŸ”Ïú¼5ñ±Þ‚Œr³‰PIú9`¿Ëxë]hÇMþ²í]²j+¯@¾OÇRÓZýjZ YÛ¨ôøøNH´’¹älY ñ³ê½Ï@ÜS>~Åõ[¾ÏîT(Ä8Ï_œwbÏSŸjÚP qŒ’a­”Ÿ þ誣×ý!¡Ûoº)Æ’Lø„% 9{‚ö¾L'¤²¯ÝÐqÒv…¬˜òéƒtéÝΩæDž­sp´e„×;íìÚæàuõòâ`ça"ÏukÜ[Cä9Ñs-ƽ¥ÒkQÌÙ«&•Œe_-è³SÊEã(?)³ŒrM_…¾Œ„¼tíôåMO%ÕÝË …T2Ó'ªó¡ÈqsÛÛÜ$(5 Ö­82¥£N—c¡j½zr—)T§'ÓðZï…ÚqóÍÒLP×{}×_¨ïâÓ–N¿ êßXøeå ‰ !èdÅ-æ­%ý²Ð¶aµ=¿ft°›¾8é“›óß®ƒî¸/eë‚\¡wå#ûÕð™÷È6Á6l/¾Œú·l¾d™9,KÊ)„\©Ã×ëì ÿíÁÄ Ôס(#£e‡k-”yòûìºG€Êµ‘êŽC• ¹—{T½£>&Y¥ÕVn.>Š—¡:'”ìiaT7/_Mã/Õ>ÉLB)½DûîÖ½‘/Oê ù¸ *÷ý·L ’û üÖ·D=ùÒš]Ï,Øqw+ŸíÄê¡·ˆn¨òºÿQÎå¿eÖMNÛ u²Åî}O ž\³t5k?|XþÒæîÃ4øèË?Ã8`ŸôÏÖ§ï„–œÓ’œï@ÛùØÕ_NxBÇþ½ÇœÞ]Þ6 '‡.ûœj|Q¯M9ÉBNXFCJ±‘ak¥`¸Ò‰ÁâÌ Œžÿ,{ÖFÆ4›BvíN„ EïñkÎU0µÌu}=Ì)D»(êÁÜÎs«ÝXÊ‘Ä/ýò”¨"’Ѿ£ö©ÎFò(ÓÜq…'HyªÎéÙ$RÉ)¸‚´—”VkoCúr2ݘBdÔJ=yî“2]œÛjó²YèddON#ë3†nê³qžûª‡©‘C_KVãË.ä<èQ"Ì„\Q þ¤FùÈËüX¹ì:ò™Ñ¸E‘ïCþò+R_VîDÁ³|‘¹Úó(tðÞíÕ½¾(â_~óëì;m䟬:‹b»µ+sy8pýÌú.:”ü|MúÊ39”Ú{–¹NC7Q<ËR8öeêBbâzP¶E¢3Â]å9¤X”Vª¢Â1Κ@7D–Ï£h‚„ÑvæiþϨä:[’~(UDŽk&& ™ç5Ë[z¨>s»(Ã¥ µH4LÒCj7§¯{£¹L8nFý¯ yŸÚù­'÷›¢Ž8XUX¿Ac²ÏŒé¬ªh­P¹o#šnkNûàIŽfn|ÆWuh~?nOBÑF"Þ¤m„W|Å[Fý8HƉüFsÎÉå‘ß¼C·`#òÛ¶€ª«ˆx;¶rM–í¿à-‹›«î+2î²yg±Š¨G6ܲ,€ÂoÅÔßæõF‘»ËôI ƒ7~3‹áÑ®’߬yç6$ÅíáL7#'ÚmÙ9w)¾í¶Z÷ìü-Gæ$9K¦±qû*¼‚ ¦‡;ë•t=Ärˆ­eرéïk Â8!ž;—â“‚Ü‚Ý7xíz/$ IIÆ»ç@rxdÐò:vHé^ùh~„Ò8Ivø}"´:ÅkNJDüR‹i°@Fäý‘ÙMðÚàáñK¶D{®¶ŠM`#‘çÎ]ä ‡Ì–¸Œ‹“O!«”¶tˆ·–h‡æ±ë¬€œ žnµ4!wù*3YËNÈõµqz)o^òè_‘€‚0 æºøaxÛóA-Ç+J"e„´¾@ù©C Y£D=4GÏÞŸºªUGÓxÊB¡&ž½wÖxÔ™j–\ð.†zý¾6¡®ðáYïkiö4h UýätZH›WKòB뇻w8>­‚ö׆­{”¡ó®|«Ü­mÐmñyFûYôRÌt…´:@ßÛË‚g7¡¿,‹4¦f•7%†÷ž„a:ý1‚;Œ„ ó+‡ðÂW£øÛCÏ`\©2´Ü}&O˯;­Ó­=öÍJðÍ£>ç©K’Èßs!!i“jÓý$Ï4á¼ ”ÅAþ±§‘z~K+ Ò9˜ì’eˆË6N¨\žVC¦ÍLÛ}ý®" ÍþXñð>d£qó¯™<Žì„õ;{&CpeJa§[ëMäòl›+«GÞ›±"eiÈ÷uГ³AòSwŒl«Á5%‚¼_^آȊ&[Ïn2}Ñ:í¯pÅ|$М¦À ¯•+²qã;öÞÚ(í’x£$cÊS/`ë“BY¿ƒ3Ÿ9PžUÊiÇõ¨Pò­Oÿ bíÔÎ;ïbPÑ—Ôf•õw¿¨ CU)Ò§jC㨾á99ÛÄÔD“çí Ö¨mKòéRýkÔ½•N%Öjˆúõ<±JQhÈNÝÚ)F—?.¡Fã åÕ¸ß Mª¨Ö½¿»ÍD¨N¼þͽ•>½Ñû…7uÁË:­mò*ÎÛ„’=þPÕ;wïa;Tfc”“…îNÁT¾,¨ÑdÕÓÒÚ Õm$©–Œç zíé+/ +¡J–],Tšksr…Š®/z^x@Emöþ¦Ø»Pi¸ß覷TsE}ñþU—ÝÖ^ÙBþ‹ß^®Õñ¾?u”©Éj×öÁ»—ÍŠñàýí÷øæ#|$}JÍ + Mo†ÈÕé$¡ÅsÕJëKfЦv„»:ä#W6To‡.K&-&_蹩 Ñ‚ú¦$Тh²\œIÓK[ï¡ñA%ÊKv>~€’åô¡cC'PJ—BNÑä JO:Þ˜{2iÙùѣ쀕MÀ:”çß%î¿MdÇ·Ÿ@¤èÌÛ¥| =±{º#Qé`^ü 7TáÖêÔíEÕoaît¨>¿&t×+ÔZfïçð‰Èo\ªs9<þ¨·*ÿKniËI\("òÛÇnRb}Ãa¡^^þShLRig9—&œ)ïdïë¢éÖãy©É÷ÑÌ>L‹”–Íï™ÑSž­F‹ÇêpgO"¿]aa˜’EËФ®ÇµeD~“4¿ÎÑ„ÖWûñ i&o{˜ªU(þow‹bucpÙ‡KgU·(BE¬ƒ\¯5ž`uö{ Ù| gTËN@úàâ3DܤTܤÜmIœÎö$‰/ˆv›öôÛh6ˆ ÝÍp»¼³œz;TŸ‡ÄqASµ:Hv³á=ç¢))´ËÎûCjõþ{2ìv’ÿ•¨8¤_óöX]È Û¯Ú+®r€Œ ¾Ó†Ëáõþõ"ÚÍDž«»ö­2O”$‹ ËVp…\¹dëH›ìVƒì/Eß\dN@ÎÛ³³5t|Û´¬©þª=¼1}á—¼^ ¤vˆn æ·»O¿HSÉ€*»D †ÝPöš!—Ѫ‘ˆcS™×¡ªRòBÔlr«åtN‚Úâ6ŠH?Zx—šxŠcæ | }·_e\>Ž+¿Ñ¾P ÍÅ=4ÐêÍSÆxì´ï38ÌÍø :U`~ï>èšë=”Ñ =/F¬›-¡Ïy­›¸¸ô»²9¨8¹À—vQƒ.c5*w27° †Ci¡áI0:'q5ú9ŒÕ¡`¶•8L̤6q§4´—©Ð1øFâÜKq¾æGvªl¡BÒð蛲—”‘üx*Ç~kw¤t´ü:§õ©ý Ow˜!mÛ„—£¸2\®Ô5%ÜÇå‘Û–×”_EæûcÞqÅêÈzoëô.Ç»¸¢îi¹O”®ÔÜ!EòB ¹è*•r·{#ï²/oåZä{:“%\섇÷‹œ •À5ûøgz¢}P8Bëù cÝüªûy9+б> \ÿÚ 7¨„t5^À{|ÎSY§ ´º-¿ä: ”QºÍ½oU ÊšÍfÄT›£ÜWÿ=/ïoC…2Õ¶]7&S¿õ£âmqiTvÙ³> ÒUmûiä9ªrNžÍóFMŸ½<ϵ£»Ù*ÕP·-Gj~rЖVèCCIýŒä7¦hd•n!XI‰Æ7W†}ÐC“FY»”!4·ñnAsŸã{úä}áMåërE·)´|±vÇrª`´â0‘ŠÜÖI«÷\ù#ÞX)W–R+o„êñðIÍ”;PjÙcÜžÞØ›è÷GvÕ®}ψû.[eÝ6§‰£D}-ñÛ % ¼ŽÎ/tyJÔ7Sh‚^IBZ•}ìd¤IOÍ›o‚4ÿ´&ŠW%ðšÿ‘·˜Ê,ä¬$³h½èFäMeGµ¾=ð¶JŸüu¡”FË8D=Øy+²¼Ù×Aղ곬šŠP5µó*£Ë¨.ÛS²j\7¨ŸzNÔã8Ü}©4 f›–mà^v¨Þ³ëøþû“P¥‘#2D¾p.©±çÖ¨¸1ÛJ/îg/Íú*A¥ô$ÕÎÝD½Pí¹­íTX…Ëm¾þ‹ßBÝE /Þ‚ÚšgC¯‹À; —dM)ð~ 9ïgVgh8)ÆJ{ž šä]LÚN -J)~Tá¬ÐÆq±9:†³tDÝ kcVF“(ô<œà}%©áËôF`ÀèUþ~Сe§÷ÈûÁð^–Ǩ£ve‡.î€1ý¢ ùÃÓ0qèÅ=1˜Š~%~]f¹­ŒZé¯Àܽu%‘än¹ýV$£´î°Øä7HU„Æ#¥A…¡ÝÑ^¤Vzåd;Ž´¶o¨Ž¬DúØË »W#ãÆMuGÎ4#“ëW‘“tGy¬±Ëíá>dõm’)Ýß‹+”´Ý…ßC¾y–{*ÇSzøÖÄå&ä:ÎF«Æl<­¶1j:\ȧÒ™þ?òg­ì$?˜‹‚ûKÊ¥ÂmPHmÚZ ELŠöfÜŠ¢~ÞJ1½¸~,¡ôj°@zöõÝ´Ž²½¿7g-=$zx%¨­0"ÚmõÒ£D\Åö DH¬‚˜TV§Wêðj*¥B}¼*~:>´öÄ(¶ÌÉw^°ã¯½ „8Wf‰¬€Håx?e±  ÙN5ˆuA²9Û³AÖ“riÊ­eÙ H½ZÙžä+N´ãHK[9®õÔFº‰S×!ý=¯=ë–û¡[XrïBÆÈyw¯°Áž e—…L†.Û.Ó9©`ååž Ù+8 v¾´‚ìhfÖ¢Gs¾ýÑ‘KÈ °Üv ò¾ÙÍñ–Z@þÛC„'4¡°—1ž2 ŠwßTU]We$QÏwŽíÛd¿´®.¨Úq˜—;Šª?en¢…Z‚Û¶ª`ux'èÁDÇe ïòXÍêÒ࣡bPÊuðéeýJý£ï U"œ„’ê=´/×›¬mP„Ž>ež™½Ðõ¸ËܯŠz zê…Þ‘@Ï…±sÐ/p ú=ã|ñâé—xÍCׯ§ÚŒPDéÂhXLÞÅC/`Ìþedœ‚%Lx|Ý3~¦>ÙUˆ|³LL»Haþ¸Ÿfñ'7$µæÐ+·áArqÞeÙH¤d´do‡Ô¼§sr#­¹—þÁP>¤o}nÆç3…Œ}ÛùŸŸGf9O ÙñÇÈ*óÍ®¯‡¨w\CXcs9>ôŸõ96‡«®-x¡s#ò\XÝÑ’:€|–Ô³ÒiÁ(°‚ý+Õä%\CÊ÷¾-î oNžñR~‚ksÝkówÅàúlÉzóŸPâ£ÒÀÜ-Ü(}ìÎ%;!”^ÅtëʰôKÖ@Y>?÷¾S(W×gF½*ú5$Ÿ¡q߆ЙW¶rQEåÛ{£ÎS¡êwe˜,Qý9êè@ͲÛk-TV£öL™ÞÍ·¨'^¸«†± Ök‡* £¡Ö­[žMgÐèXü)r4ްµ?ÖÈ&ÝûïßV:ƒfÒi׎õQóKÆbŠáMéMPøó:´|Zà+½•­Vz´ÁÞ­ýo7KJ3ÿo,õÎ&Ì1§¡ú®ÆÊõ„i(±èúð„[œ¸÷SË \‚ì+oB$úÞ@Ö ån2ÆiÈ4öô §)…×âæÎ…v‰zžOn•Ü#"¯]‰,”‚TUµO¾îŒê±ú*wl+dPОê±Òô®4 ¢þöJ÷éÇRxk²âè‘Pª‚;köy訸JÆÂ=‘tK‰¯PMOÍá®&5”FƼkƒ ¦æ”ÝÇêëPËÃU–»jB‚èóS£ˆr+Ûºñq`=Aáf²p?æø ŽÇP¡È!ckøžsíë‚Jö‹i¨ •)NOÞf~¿fî‘øÅo—«©xHBíƒѹë=ðŽœN¬1Ÿê£Å瘵+àCìjªâ!h|ÕË ô´ZÈÍI£c+¡µÍY_ðÀhoÞb÷Ú#ºhŽJ,χÂ=‰[îtÐ>r^63ø\ÏœÁΜ^¿‹.0|Ù%¸˜ï(Œšg^\žÔcFÙÂO¿rÃÄùYú'˜ª¡Ø*ë0 ¡Zÿ²¬³*ä™<ÖCʯïåÏš êh¥‰ýGVÔ ’‡ˆ5+QkÍȈa.‘ß”šVî(G=Í.,ñYƒ+ŽòâŸ"ò[î:w64ü6ŸÑŠF³¡,áuÏЄ-fÀ€í"šn~Pç&ðÍö¼¤öJß‹æaáÛ#úØÑ¢ís²¯Ä}Ü!›Ì5Ù›Fä7ÖDúà-D~»XIìDä7Í{Ïô)‰xS%×Õø¼Y¯iܶÊêËW;—A…õ7SÒê3P`ov¾„#²<œå/Üï2´[ÉjÊg¾¼„î Ò«.ÓD»Íö0/‰ ÄÊ»É ö…¶½6bÞ@ta[%—Ò1m–J¥=¯ây9©BÌ›*ד֥'”’#'O´ÿTË^ÊÇCbåÝÃÞ®,]$m÷ŒÈ£½]«Âˆ¸•[«Õ®©i­}·ÉX -B:¸BŒÒÍb÷uš½ƒôÒ'vKˆCÆ™úšôz <÷˜ËÆ^?"Ù¼Âûdú›>)ßL‡gCÑHdÞNoU 9[5´èöC® ÓûGêC÷„ÿˆÏáÃo}Ö|ÄÜ ²EäBQFA’¦*ñ{ )EToèŒ.AeÇóJn.€jY¥gÆ ;¡&‰åÄDø¨óÓŠ˜Ýjõ%¯2÷'n…†(´Šö†Odí—['CËn©™Ð-Ж¢ëí÷¤:Õ·¦Q¸B—Vû…èN-èoÖ?d  ½…mIÓ›ãàsiÖÖƒQ§á‹ ËÅ“Â]0d–ÀåÈëïuÅŸuÀ¨â+ ¯uQðuš[ú1×e˜ Ð ÜÄS;{~#~fæ£d3Ã`n2S_aIy=¨íd'¬Ã‹|%Ó{¤HïËÙàNTÑS"âÐq\™›7’['мŽRýé[§(•pSûÖ"šÓÚ¸¹ÔCQ.YMÃ8ƒÊjëªÏv!aùLš-•$*Öi<­Ê™ûÄÌÒPõMv™#ªwÑò—îTA-¶tûë-$¨£ïžcz®e¾BQh -:{Ž€†{›éŽ?B£K¼F†dhœ%Å<†&#·?ž òG3y¶»yæÞh~彩Í/¼‚9§Þǡ哽™æG¢ë‹1Á}hí÷6ßìÖñÆò¼¬‘›'ªUG¶0í€bÓÎ]VÉrwãt–{uÑn-J,†,3[u¾ +áu>—V’Ϲ$ß*‘—Naê¾ì ;¤ºt? v#âe"óa¤8yËmŠŸ ÚuÓÞJõS‰mõx ¯k=MÙ&(ìlá_qJ4õ¦‡ ™·¡”öP9®2ùˆíTsÇ>êçø@´ÓTF"ø” V"Ïg`ý¨½¸ºqùY¨mJ˜Y®ÕD´>kBÕ!Úm­P) ûðÒ­,(Ô56€ò’Ê"yÅ«P1"ñ±RN*_…=ÔZMU¡k|nIÇþâ7Wá¶Ï&EP{ôâž7žAÝË{L½ ~‹E“¬÷zø`»y=[4 Ù¿‚æ‹}ý/ý8¡õ¥u‚ÌòGÐ{ŽA¢>:›yï]¾zVT>»pú×ÛÙJlƒþ/òœ¯nÁ`Œï*“Ûå0ìT²rˆô&Œj¾z|ÖÆL“ÌV¸LÜ:ÅÕS–õËæ`ö¨û„n]>̵É$m”¹$¯×ÔénA2’î’ü¦Q$?UØ8‹”"Eù»Óšý®jÀü¤ÝvœšpE郙홛¥‘#FÚŠë2ècõ³§EæVïàÓÕÈêôÒûš±8®à'§ÁFd‰»ã+gm÷¾B.àò¾R1Ž< 24Š#Ÿtå­ƒòÈÿêé­•Y"(h|bèá—\0õ®µ G‘Uo¤.~¸‡¢òœ{mÅõWxÞš.Û‰ŠY›íQò¸[l‘JQÝ|uá49J×k{ßy¸e:OæysÄ-l=UšRe(OhV| V®‰&j6&h"¡×òÁ–~TRÞQ¶W•§‹*ZØ/£ê—>{˜)jÐæ3iZû –dÊãüMD~3{öDDã êí>±_¿í-ð^átÊ`&ò[¢1c×n4°¸Jw ¦=@Îj4a›lÔ{Œ¦›d?í°8Žf»÷©²¡y¨"ÉĽd"ÞDבûn ò›îý/­3hy'—ÆŽùMÚW¾v+Z_™Ô¿sü9o2Ó UŸÿoÛªÚ‘A}þõÄý^ÐæÏ ùÍÓã/!ëp—ø§k±në䕞M´¯ŽÜ ß’è zõ*k|·Û4ž‡˜+TåO¾At³Csyм¼ÉP9h6 /óÙœïc$–ÛX゘c§^í£†Ø‹ûmŸãœ»›G÷ç‘4ÒC2“ûërò®Á0{íH™{p¨Y~¤îÌ¥1è¿MÔW¯òE| ‚´:•}ÉÇ ]k€ß„;ƒ¨¯JˆŽrD-ðÜËu‚~§U‰g€Ì5G¾°7FBÔ³5\å„l±ûŒ†¿Aöר¡6"ÏM\7£–<ÆO_=áÍ×À£†D»° ÄöÝã‹PÄ1a‡öÉPr±ÿäT/”KUëlÏ…Ê#]ž\W\ *%âºcm4Ô¬Ãó$j'Vlûb©õbaÜ‚2 z›ÌõÜ&h2œq9ç-^¡v#nAÛn&5'èØ¬½½£æ*tµÄWLƒî°ºÂ¸â›Ð{ ëlÒíáóI¶Góšó0G-˜C«êÜÏÍÁ°óŽ£OÃH×˳=/á«ï•ôàéd¿ 0âô&Ú7WMÌÃÌ9]\Ï•`îô ÉË­{‘äSOÖìò*$ ížì£BŠÝq"ìRHµ-óaÕ9ÒîtÛžø‚t×,í“Y_â²ý×…É{zÉ!t^Iß Yöxl#ýÀl7yw½UDöfŽëöÈîoË0{¹ýŸÞŽX«= ý¡2©È¿×Å+PP–÷ú»´Q2¿äu_E8£g={q]úe°w JP=O¯AÉ­2_§y”ʲ/²:Ö‚›âýVo|Ž›ã£çÉȶ¡\ÈCùˆ¢vT(žbÿ$pŸ>À €ŠÃ4f›žw r»}t€ØQT‹Ü ¸!5jÞ%¬}‚Z–Me§[Q'øvæãÔÔË«I=¥à‚i¬Z†úÐðœˆ.çòb4 ׯ°/!AケÙý"4ùV¾½ë>šŽD1_sCs?Ú†pÕ_x»{fÔþhùˆlèôcQ´bîU\ÛLä7߉۟«yþˆ7–3k÷ _]Uaæ_ ӠȨ½“ÝH òDNæx>#òÏø)}È|}@ôeˆû™ëùä^EÈXçVz¨7ÒN®‘ÕÒ»))LOïíÉ#âòÙÓˆ·o Ù¶§['i¤VŸJl¼\ ™-yäm yÖid†A¡ãº-Õ Pâh¬È·Æ“ÈG4º>PÙwV¦oz'TKì”ÜÑë5Çe¯Kp Aí•eOç¼÷AíðzCy"Ïé‹XKî%ò½Ìm±c ÷!½_ûOB%Ó£ŠPŽ(Kò:–D€rgÞíÜ’P‘¬§lù•*# ï¾þF U¨gÔ¤¿ømß鬔­¶P»íDXRШ³YIÙÙ ïr½2¬2.ÁûútYRëXøx§m…ª74ÓŠÔ}Kº­®Zœj‰Ÿ¡Ý+?î¾Dt&–³ 0ÎA÷g2§æ¡Oi¯¦ox;ô§Ï+?îÙƒ§Í´¹¢ÖÂðþ°m¯atëý .5³ˆºàEbQãœñÙaš+Nâ­ÌÞfÔô–o‡yòG–ŸæT$å ½¯™ ’ÎE)ÔÝÏFòÃd—Ù_h"%«?‰à#¤& Ž_Vvi×åäl{§‡ô>uâB ÈH7Ðfâð™líhC‚Í‘¹ÞâÐÌ3dÝ»"rô­® /QüzÙ?ú[ÞÏ÷Ç•-wJ6¾(B.{J┑'ü¦’mò­WºJ“†üO¿ îo@AÒó9œQˆ‰l>Ç•…¿åÜ>X…¢L{Fóp½‰‘š\Ñao` ’Ù7‡’æcÛ1ƒÜï†/&¢tÉW™ ”é.ß2Š[8t6¶lAym­ëí^£‚…ÞN™ ö¥e¤"&>HH ›‹B%™‘AùQyðná…èLTí£XŸou5–é·ö‡|E-…³æ["ˆüæà²ñ+#ê¢ ¬ÅCh ò„ÅÁl‘ßždÕFÃ~6öíbšh4i—ÕuM–?§Ø)M7–uËû³ Ù®×ùT‡óÐüvÇ3 S ´hÝæÖ›‰;6wu4(ùÍÑë˜é#´¢oÖpf¥$ò[ļo˜o£—Ìüþo<Ôb÷>W„gêÕv¨`t\¥6ùw™q&¢Ý§ï¾p‘C˜û$MNŽ=³âÂØ7ií6×fg£`¢}&œŸIh\àµy»³µwsUžC;Dåê¾Ð*Ž€húˆÓ:I›!†¿KéªÒÂ}Êk¾ïó ™m¾庻ƒ¤ïBÒ€”ƶa:Hž)~-)¹ð;\¬ó5'"Ï=¦©ç#Úqm‘g¯±Cšqý¾Ë7” ­è\©:½ö÷ßåÖÕÑyN\£Þ:‹Œ)ø¬ƒâß]T› Ìº9ò¹h1Q/UÛ/3j UòtgüÆH¡úR¯r4ÔÚ4×¶œ–…w~—Ù|]ˆúçVÊùzYhŒš=LM Ð"±*¤'oÚè´Bç·1B{£¾ëò‡w¡ófSëð²èV*Å)•Vèå-$pÁg~†ëÝž0`;“I=í ƒ¯²VgÖÁ°¨ýfÒ¹åé _y§”/~è‚q†wòr¯T`r;_HÀô\‡kÒø6Ï·ƒ÷À{$¹gèz½;ÉŒu¾æmB êS~6¯v!e™ «²•?R' Ç[u¼CÚÞMŽž;¨ái´gfØNd"_[]x: ™§Ÿùmýí ogÄýꣽe\È)X–SyI¹ú hƹ· oÏö}‰ì¤ÈÏ ¸=‚)ªyŽVfqàšîùVòòP±ë[Éæƒë4Š¥ ¡xØÎUË”|´¹›ÑõJ]g3Ëi9‹›¼>ÑH…›Ý1«0£åN͉MÕç£ÂÛ§v¹we‘°v†#%*Ñ©ïÉ9Ë*46­A5áàk«Ð5vݽ:F‰Zç»h8rQ§íÜxdÜÔgüÐ¥c¼ ®teIÖ¡áí£"k‡Ñ(î¼ÎYç94®ÇBmšÒÌÊo‰C3µ·©»JÑ<`«ÂJV¯_xSؼ&n³ZÞ pï} Z1‰ …ÚùMø¤æƒÇÄ‹žg¿ää¨ V;YÑÀ oõÛÊŒîóKRJ%õÊžíéë‰ú™#ëΣFD»­gKó ¤?gp͊̆ÔÞ‘aŸ÷߈8©´¿-EI^Û}¥¿Üƒ$Ýà½)Dܰ;qhï#Ú}æEýÔ ·m¤³8† ¦µ¼÷´B ••´Ú“(vúÄEºÇPù)ÙYªe9T¨2ŠÒE æåºIËïBm§yÑò&¨³\+}Üj_8QÛ¥…½A>Ê; ¿§•ÜáÔ‚Šo£S–ú \K35XNʹ6Òž¯…Š3‡_tGBeøHèÆLPñùè˜dÅ/~3© »À? µôvïX‚×AY±ßúðN':Ôëk¼×%(Í=»™¬o;YÀ§+{³t[ UYæ¬ÅNih7cØÎ|a:}Ý/0Þîlé°W § -\ fji¨ÐF­$òÛYÌÌN߀z7S-ˆv›d6½í‘ßBC9R¦Ñ°¯âÁÓ/sh4®Íî€&Ë„–ÍmBÓ {†VåE£™U`aÁ(š‡\Ò¿M7-GÔuˆüæö¢¡ÚÈo¼}O?ùí*E¤a9‘ßL_Í"â£_æã†_x‹ü&öØéôò»ÊVCyûöoç¥àM•b½Þ¾‡Dý/6"ûêH§0xèjEÄ‘‘AŒÒ Äs¿éµ¼1D´Û(V†Ï$B´~‘,É„(W¼i,®)eS}ël-Dfyw±Fnƒ—²}öòG àUh¼Ò,ÄJ¼òôÜ,¿ð÷r‘oÛÜ!‘îƒÈUIH*êª~(É¥ékï+KAŠ“Ó8uÏEH¹ï©œñýþ¦™òM¢~ê°Ï2ÖÒ4XNÊM5@:WÃ;ÛëDžóÊR~2ìt_=bxmðàì‹­·¾ß«œI? Yf»V¬Ù Ùö'¶59©.ÜÃ6¨mË‚Ü0ã¬×Qð†/á.ãô$äg<½{“¶ Ë-.È=ƒâåbj75¡ôÊÝdð!¨ Ê*ÝT• •éGöE®0jöˆzkMñ‹ƒÕ¡Dœ~9žu×8Þ¿ÞòîÐç@h$£ùt;€šc¥oeÆDBk¼¦H ÍSh÷4¾)ï ™Xø{ «'7³Ä §`šEüMô•d¾@dCÛ öšq*é…ç²a¨ÙE^$õŒHFrœØû FcíKN¥²ÁØ5CþO0Qos×ÍÛ¦Ïz¼¬Ëó†ož†»Bt‘dWÚFÑ,o$£‹í¸f‹ä/ £§4üÒŠÔ+qk#R‹ÆÛn”´ŠËNéÞ·Aú®âæSú•¸\ö Îmqd¾Fζ²h?²æ°S…J²#;·LLxp®|±Œ?ˆµ¹vs0ºn„¼zN-åãÈ_-,yxsëçmÅ5®‡Å½fûQx8ñ²t„оk2n¤½ŽâRLJ;g¬PÒY$ðì”Ú›Ý!¬Çƒ›L6”Ì2ìÅÍj•~'=QÎrÇjÎg¨PÀô2íe3dÜ N< F%A'== Twˆñ¯ÿ€jæçF™õ…Q#ìÁ+ǵú„”â¢î†}q/FP{ ëj*4x‘Ï瑉hŸLöJi%ÆsÏ¢q÷åkòÑ”MÜwÕ•<4Ó(ˆ¦Ü‡æ×Å ^Îù…7yªŒvjhyMÞ,“E+F»fó½h}ÕÔä©Óº?â…¯Ée«Ýc¨Rºk{ uZ†.^²†\Ó¤¶‚ï÷ƒÞ†Áë«'¦Ü!ƒ7¡Ð8+ ÒEî ±.†Ô9‘y{!En˜@bc›$ªÆ)ŸIÛÉáñbFËȈ¼Ø0Ac ¹öÖyëé¡8Ì^è°p”ï®íP/Ìýq??ª c5·øV@M'wPëBáAÿf9¨Ë;Pwê(ÓbB™¨ ær =ã§\¢=FÃrâQè½kcI>(›hLÎ ¦¹­¢7«\¡BÑÅùÑ> æ¹õ$•ª¢Ê;o°þÅoÊB§Œh‚¡æƒéò#ÁŸ¡ö¥ ùªD¨{÷Áí…8êsóõfFœ áÂGž¶vqøDÿàÙ]Zheäkw ; íkæ=›VC籉³Œ²ÐlÇž\ðz»Þ1·ºK@¿ùå[eÛ/à S_ÔóÊ&^Wþp÷:]í9,\·ƬC:¬žñ2$’sJ ¦÷Z’÷}³ýÔ«a^íìSù€$) ~?½_I§exÇ„w ¹q4ýZ ¤è#d„GªZoZJ¤¥´”ÙrÄ év¦­ŠÕÃÿCÜ›GSÕ¿Üæ12+ó’2Ëx]¦Ìó<Ó¨¤IR†$Ñ@ƒ(¤Q%³ÌSÊ,™ç™ˆ™DÏq¯ßÝý¬õ]ÏßÏ_{}ö9ûœµö{¿ß×þ\Ÿ×‡nᔚ&u2XØÿJ_ø‰ŒDœ¬½Y÷ûZAÖcd¿T×ë´lišò+œq¸+‰6†” 9¦Þ}¼0ÓÜçÞöD>áÔµ·^ÈsõµB1¢ Ìå·'ÛQhÊé×ÓL1®(v“Û’GÑâU–«Â_Pl⃩z8îwÉ[à¯GI®IzŸ¸q”*Ý»çKÜ.”É=§ÍŒrsßG^'}C/ Rå¨äÊ/ževUŒV“~u¸ d”¾¢u"BµøÐÖÁPqÔàäw á^BÍݼ毢xhHT`h7겑¡kA}KêbÞP‚¿EÐÜó2AãDááþßyhªØGæjOð·û%\:hö#L|93 Í—$ÏðßBKš8ÎÀ?kh%¾ú{mU­ƒîð  Mô¾å¥ÏÐöëðõSoûþÆ(Ñ\û ž4˜õ°ÑüM¾t$<†àod.çDôF5:âÒúÿÒ[Ÿ4;ÒäyŸi›üE‚aÒÎA¥Þ÷¾ûžÃ§Xö£R·½¡¨CMu­}äÑ¥${P'BÖùW?cþW{¨— © vlo¬_C ©H”xB2QhÛéˆÏ\\ÌßÖA¨çXˆŠºCúòh"繋ÑoÃîùõ;d3l¼ò¯bƒœ¾ZÉ¢oå÷î©·Äéæ 0˜m?§Tz†/ @ãÈá+©û¡ éaÙþå]PhºqbS:VÇ ˆOœ‚¢wRâ±&5Ÿë©±ö(„ënA­ÐÅcEÕ?§èàã@[Å‹wð©}–ã¨Õv_è¨o ”1±5*•‘A¹ÖNI¡…2¨È«Î=Ó/Urú$÷k›àóÞ;¬òôðż~ÿh¯ Ôå­¹ëÓò©¢¤LÆhVi5­‘K€–»| €6ÙK$•¸¡Ãäˆ{†Œ tï’׳5 }6¬¦âl½ð•Bj#¥ç ;êdŸoW#èÓ&ç¸Z5Œ~ëyn»_(ySñ2L\°‘|¤w&½YŠ9ôõèRL£Ì6^ÔqQK†ùÇ×Á¨D:’R=2à§f§Åé øÅzìd˜¬X½47®½«›\ÒÍoaƒ<è8Quñ ü(vFâr¶ÞË—tJe÷p¸ɦ_/‰pˆ!Å 0°,}‹T1cA»Æ ‘vϸŠ,é!¤?üóMâ2ŠT^{KÐ ³uÄ…Q"dMý¼!è|wI^·yù9zdëÞSFîœPû‰Îpäs\,ª3B*ν•Ÿ]Pˆª×—l ðÕÞ&Kpï1ß-ƒ-ÿ\íõT%5wr0…h£ôÁ3B?£¬Ðu#•2”g8¬n!-„Š*yŒ/†¢J…ϯ¬·PM}í’èÞ5ÔP© èã‹@-s©ûi÷Pû¶»•3!WöÈü:}¯ ¤´Ž]5¹Fþ†' ßø¡ÉÛ1\ lDÓÚ7×›ßN¢YÃæ€¯¢yÿ¨vJÕ:Z¬´9^¼ª‰V¼ÎÕkýhmáßöÑ&ò…è­­ÿôv°KÂý‘ :<¸9¡þv¼=ÀË|” ·`irµÌÿO½1þ<øçPëì_®GµÞWo5A÷¿Ü’¹ÿò þåüË;(h ?¾53ú—‹ð/_!ç Û£bã¿ûÿýÜ¿ßû/¥Vú{dBá:>`N< ÍŸ÷Δ§xCëQ‡ŽgohçÞy¾jsý/W¤SBò ÛÙ è8Lf×|ü´Uæýú’?õ—wÒ”o˜/i ñç÷ô"4¸?ÔËMå&äÀTŸ-¡Cù&ÿÿ×ßþ“Ж ¥[•ú—çòïyºôíWUÌþòLî>œ•Þ‚¡žº~‰ðmíIü’¬|Wì?•J×ãîQá%=gÿr¦÷— GHÀl_ÎÀù‘ƒ°À¼gÏ­›Rð“Ñ‹6õY<ü:ü€¹ŽpŸZ™9s ÆÊÖnçÊý { ¬%j"å6m8akƒD­%™ ñÊj8Ù$Õ!­¯(øŠdÅó±]ÎHñÉoý–‚2R-&ÌÉÝ@ZG¾@Ï[¤}7§dÛ‚ SE|êÈXÐöÓ?-™•8Eò‘¥Ã€¾Ž) Ù¢¸ß¾ÖNÆ]÷47O×å!G³˜‹‘(rÛ}i7úœ‡|Œ19bžæÈ^¹®y|Eän< A¡Úó·ÒQø]a#GÞIÔ|Û#ýŲæJÆ7âp¿Àí—Ñ¿+Pb³fæK».J%¾7’K«E™¤Y½ß-+(÷ÓeÖ&˜ô^OM\F%÷»>Ï?ŸFÝ&ÓNQ„×Eç®VEµX»x]“iÔ +¹ê0ˆšÜéÞ—ðPŸÙlKœê²Ï[wê"ê»4øCð·øÎnûýçи$Ÿæ°Á×4~mX yüíú_Ø#4›Ð.8äuÍY+ÊÏ®¢%·ÝAv´ 'î;0ŽÖvÃú>æhó¸é…’·*AoêõÿÒ[ñíÏ‘mH#ù‚µŒ*Ãgv}ú•ú—ëñ¯ÎþÕUn 9™Hóá¿<’ts9Ö±§“ÃÁTøÓ)Þã $6æ´ã»Â¿¯ÿ}ÿßã3îl¦½YûK&¼ÿ«Ï¿“³O·ÕÿòLþÝØý0$O|òWß9'ÿÇUùÞÉÿqW>žÙàšêøûþòMþcô¯þÿåU¦MŸ¸ÖJÈ»…ñ2ÃPåÁ*£u jG(¿D{¼‡‰¬GÒ÷ ¹ÕßãRmÊÒ_NQë |’×Çð—kÒuQö²‹ô¦ò—}z‘ƒ’GŸˆÂ0¹6î}Ù ßZÃwöÀèíK}s“}ŸXŸ]œ9Æka’Ï]Û¨”¦5z ü ³Öy´×ˆüÁ}‹•Ý°à›¨~f(;µúœ.Ÿ‡¥');O2ÃrÛ … C1¬^-øó`  þÜÎv¿ðK¶ÊصÒÏ#±_³‰ñG$ùÝT’ì;dþZÇb~_A ¦ZvÓaH9_:ºŸiŽ“ÆÌæ"Ý“±éįáSÈ”Da!2MÓÿ茞DVU{½$‘ƒÈž6ÌIð-Sý´ÁÈÍñºƒ$¯ùH%J‰&"ÿŠOu7Pð‰Nu°]8 S›uÿ∢ e¯î{ ¸“÷ô]¥q”dZ—y–Ò4»7çöÚ£ÌRë•gxQ®g‘d×t5*ò0‘þ$ö@•ÒÕä T3öìÉ"èΚì®Ûb=j]>“uÿ†jZæm C=ÖÍd û=hpåÄ›ü©chôEúEé324™›ö¨ŠŽþÿMoŒ‘KbVåiÐü|)ˆºwª6œëržÈBYylá{%ÿä1’®¶¿<ƒ"ÏcE ×ùíõyn‚&DOÎý“7<`†Þ‡ W¬y1iæ•UäOȱµðÃý@¬ÃúŸ¿é nœÔþ¯~K¼úzuè#´0Pí­6†ÖÔµ¼@ëßÐ~´„É+i:¦¼¢ ꃠózê ùîAèHNŒíÚ€¶ÕÂA™zhiÖf˜Õ‡¦“‡ !ˆ»/%䢤NŽ¥Ac)¥ý[6 h>ùK¦n¯7´Ü¼#öÄ}Zmÿnm$N+ÜÖžÐvEä,kÇghí½jS²]a&͓еƒ³Ø‘?z*œ2ª\a@Sš-Oµ†O:ÿ1.„o%/í.½Ó€ÑÿØÅ~%ß'sÑ~ܘT{ì÷¦6Î^rò$‚Ù’¸pϺݰ@æ(ûûe6,®…¼ÉsW€_*aK?óÂJîÀ{Ïʰ¦Y–ä ‘ š6wŸn„-ù1÷ûò‡‘¨Žéª‘’P\5eøÍƒ¤'÷+x÷EròÞˆ¤øÑ"ð2Ü©YÌk˜ô öÜú«½ÊµH7“x1û2Y<~ø ‹)_†¦ïGf­û,6eâ`šÞ @¶¬G*‡)TqWz¹|“¤.rL+žo7kDn·#Œ<îÈGÇJ!b—‡ü®mí(È8Àsè 0 UˆÑk~ÐGᘋF?ö èƒ›M¾ß"P,£œÎàþ=æ3×5¡Ä²Úµ$þã(õä×㊃s(s·ŸEflåª3»Û|QþÍøÕ¥%T~¾[¿ P…ä£ÀÀi~»sTWv2¢Úå«iOY¿ úärPP=j¾éOZY^ÁCqÅ'Ä,¢N½ë“èSÕ¨O3Bá׆šº[t4§ÐØÁF†Ô¡M™»$%5Ñ´|Mw*–Í>ÝT¸auÍ+¥“¨JТýmsÒæQ´œ òìxp­Y:œÙöŽ£–_—ìÆ$Ú^£#áP_F»ÊcïN¾\C––õLt<7ë`$‚N,ölCè¢óÑS§²]¨ë¿©²ÄÃÂÖÕ/Yé‘ú‰“¶yÆ84x»± ‚ ç ’¶ ø˜^¹3š/ [O©“C®â„¿æ d~èêßßéJé·zF¹ åÊÔW#ÿ :6%¹SÏ?sÏò¼Iâ%fJl„ºÎp—f,P†OzùüK|û¯~{ôQX´© ûMt… Ð±¤*3IŠT BÓÂ_BQæÏ¡•ÿõ·‹ÓJë¯àSÐý‚ÑJP²¾{4.•J×îí¹öì(”󕊲[~† ¿é'Õk ²ÎŽ;@©ªCŸé84¾‚šÐwQ¿¿}‚Ú: nV hà¡öšuüÏßh†Þ_øКLÑͳP í…B“;å~C×SuÂÍz=8ñFŸ Å©ô™zfahÄyaar ¾Å3e‘ÒfÀ¨Wí»f##â-‹Î‡ñ±HWë²ðc¬‹âå{˜¦0ì¹mX³|;Ô÷Ì0¼̄Óc X8•@Éпd¾>.ÕK¡|]ÌS`9¿`Ð|)¬ÖOߨ¼ (ç~ïUØril »L…Äø5AZ#IŠÒ%}5ÉT–t¿7Œ#yÛIúê F¤ÌØ»hˆ4¼Þ®Ý¼H§G¸¼]î™^¯KÝLéI=eUg‘eq^ÁFÙ]o š÷„ãîŸZç,²¥‘«0Ó,\ë:ò¾õÖq4GþóGc¹¥PÐuäã¹/¸§²¡DùQô©«©L!î›ë·zr…¿F²Ú0Öø¿„Ü÷œy?P@~îA ±h(x\×:ô›P§E4ÉH1þ+hzHd°Üý¿+Ü> ½t>iú„"ÝwXØ÷üùЋdÿëo×2®ôèyÀ§Ì—co“˜¶ŸOÚÝ42‰ë[ÎOüÿáu½f+„Š¢ei¡r¢ªÝëǨøqãC)Ô ïútj³ç‹^Ì€†®VñJÜÿùsníÒRhÍmd»EuÚËÖ>9ÚÉB—›íÁ 4èõß ½õë& ü’¼›™Ce—ßÉÓÁ7¿/½)oÂèþ£I¤&ð½Í4*Ï/Æ3=CuD%áGšfõYL})”8| f~ñ®ï¾ó\WòÊøaÁ6¾>»Û ?òœò˜V€%ßÇ'Þ«jÀòÓȼ늣°Ê9Ô…—Ö—Ì6SJ*`só¦íÉ/‘˜B3€Uþ=’øßÓ§~„¤o÷K¿±FòG*]ŒHyî‘].‰ R÷ü-»;~ˆv½«EGÎ]n/œ‘é‚Ö—ÔÅEdI ùsù?²ÓÜè ½5Œ»£¸2ŸŽE.Û’FîV1äÕ¸Çp@;ùUÓöû6¢ è5›7‰qO >ǯœ1=´ãöÕJܤ^¬ä!€ì)T4g;QÊý‹TÐXÊh®Ï¸H« ¯é¡ôrTWìQGå'¤º‘Çã_^üy Õ¯ÖŠÙf‘¡æ…ë:ÆôÎxèæ«·eOSQ§ä‚o¶ê“Wp<©zGÈ•GsÍBÑø7Ý…WÏÑTXsgÁÊšñ|v8†æ<5û0ßÃíýWgnAgö§küÓ'¡ÇâTCʧ(èßçmaC:|a±ïà›^ré•ÐC0êY¼èC c©ªñÏxáÇîW‰OÄ`*-­ßJ*foù.{ÇÀüX­Eok,Ö'¶Õ’;ÃÒbH¾ÆF#¬hÊø·V {ÃÅ2Úà–ßÄÆÙ9ØlïïYôF¢â…ß!lH"ܼ—m’&îàsGrûiI{jE¤Tó;߬ö©m}êN "mtæ×Ç•éH·•âÏíŒ ¦¶e†^…ÈX~våÝ{id¶¡[+S EVz†Im:kd›º~z0äîZø”ðxÕ9…l-8‘;("l°®ùÈè’ùE¯"ÿaz1ª÷1(0òg®&è E»éF]Ga¿1£Š—(z›Nè“úg+‚ŸÜd¸_lüŽqýJ|—xãöá%Jù * Œy~b2Ã'”;cÖÑçyƼSyM7ƒŠ/>Ù¾Aeï˜ê©ùtT­îV:ã^€jRUŸÔân zü…Ò{©7QÓÄþÕÚ²8:ðвOŸuäÔyw§v žå¾ºß)4ØoèYéˆFÑÁ:'3Ñä´ÌW>£éÅ)–ëùThvš\a‹½ÍO–0x«¢Å™Á‘»ÐÒÇͲִ­B[mÒÂÐ:®]OŒNm>ýà¹[¶m‡ï¤ÚÍ£=MsÝáîPtP¢8¤©›E¨ãüxt…¤Ð)Åt(²¸g*Orh~FWÎ0 cö¤rïKð&v:uæÝ·‡¡<ê¢jÚëm¾Ý\ Ía(Ùøä 9WÓ+Yĺ Ó©øØ 5¤ÓÆ.õ ‚½¯¼ÅYÛõÅ!ËHÈ9C½Bð¹ ÄŽôƒDágúsH;84áö2Òt‰Æ½ ;iº™ÅmäI®¹íçÆb¥êÝ PÑGê€å?z– Àÿú'iŽõžã^!¼ïÅß>…aMyF¯úo‹=}оøq³<ÿ_ óÊË¿¢°Ý/£¿§hr›;VZ3 eǼ{T£¢ <4å xw2Tô²È|¸8UÒa§cîBuÔá°”Ä!¨¹ÈÌk µÏŽËˆž„úµ|yçƒÿùWè褀´–=¶œãe†öÚæˆ$ÿóÐuøV‡Dá7è=îžáñê 4°SÜñ…¡—á6A×…á›yàù¤Ž0²•\ãëÒß_ÇÑ¥¸ÄÁ¸¿YÂŒ•8ü¸˜qKãÝLÝ“ñèôÖ‚™¬‚Ô‹•40Oö,²µaAçm„u»,f„Iz½{K§iì¶DzaùÚÙûÆ>ð{r©Þa=™_·&ž6Ÿ«~ÓJ<ˆDõ –¿Å‘N3ínžÜæu1^»R…ä–o6²÷Y!åÍ2n¤~ä0Ÿ¼‡;fæ¦}:ûþž4<™d¾Æðy#‹3gÇ`>²å\ë.——ÆÝ K?G¹î!çJ£~ŠK;ò ½é!RgA~šåÒî¶»(0ÏÉ·ï™îQ®öþ –‚"³™ë)–¸%Aˆ1¸„‡ØúqÜG)N¾Úúª§(=žPò¨e˶Ü|…ð`¶É‡Uâ¨LÁv‚Ý›paÜÚ4°T’FuÆÔU3ÆW¨1ƘÖ6Z3ÚO_g©¢‹Ø·®û¨gxÁš&« ®‹jK¤¡ÑS9§Ïñ¼hr[ÇÆ‘ M —/Í>$ø\.¿!Qe šç&éÍ‘}@‹b¡Õ}ihYÓû"z=­úB¿B€0Z/=x ¦¶Ì]üüih§xbLòö´?î3Ð`ƒ1t‰éñ}èØ¢UJ¾,C(ÅÑÅÖýÇo5dðÍ yøš‘–X…*öþzƵH(Ó¡¥ ðÿgž‹`“ð_žAáű±á„|÷ˆ´Tr ò~¯ºÇõÒCNê§Å‘æËu…åùrâãm>^ê™G‚N3¹sºßCÑe¾e¢vòÿümí¼6„ºýÿÖ,*W;¶FðÕ#6.lÝÐv§ËŽVítlX L~½ô£ƒŸC—oC ûI輸&!í1åÑžÇ.C+WÖ} hjìy¥ºì »„7ô?ƒú¨·AQÛç?Ssд±¼§¤ŒZHè^xðAKßtÀ[•`h½DÕ»ë® ´N6¬D§×D’ŸäC{†o#S'á|†‚[ª ‡ó\?ÏŒ'ôŸ§wÞ Clþ¯ÔžÁ7úŒ‹«´aTÒéÀg³0æQ›¡N£þ|Ò"‹0Òð …ÑfO»%] *ùn⣄¸˜[tº­u–‚bÜë…ae§„”çΰê5Ná6ë›EusçaóM{ŵæ@$ÊÎãûÝŒ$ÒTár{H‘´.É/\cÉo™zR„”§›£'b¾ µ…yºÒ¦ÜÝ2¤æCzª4bê{È`Ö“-2&ŒU¯ßä¼ÏEfW£™7ZÍÈÊ·q»Òõ²Sþ&}®µ‚»™U¾®:m"§Ú s—2äû²Sà&òn1ÿVá-G~3áû¿¸}Q aGw¨(% ]÷ æFá3/,¹Q4ØÔÄeá0Š•È}ŠóoÆýû3»äO.¡D‚œU1J©9wºÓeäâ=¿ïIA9µ×WËÚñ S‚›,‡;*Ú7þ6[ÎCeÉ›1?e¼Põêݶ“»ÂPþGu¤›+ª½w‰äú„š{v?ó¿TZKëõ2]ù¨ý#ópo£#ê®ýRs]»„œ//B£SœD“¿ë&œDS㾆UÁh¦®U¨ò{Í•V®9v2£…’rC ¥.Z‚]Ue¦Zjn÷Q~‹Ö¦ABæÑƹK¼šèÚžêg,?…v!3Âmo–Ñþ%“t|õet(fD±^tàˆ.,HFg’ãf7^Y£‹¸‘T R¶ÅvS&=…Záƃ…‹P.}ºÜáCiÌW¯Å<È·áÛÑîrx#Èè^X@æô õ¼H[âÈ.ÎiÛ’%#B½¶QÙ´ò¬-ÐtäCBˆë¨å›„º®8bqR{À:2 whQ”Bö|Ð}W³ðmžå‰å(çqj»IÐs뙘=GßCaѯ\Ý™¿óß )äDb'¼.¥åÛæ6¼±yïÆÛ E.ѯÕ8É·û`v¾Ï“û_‹VXzœ@ %;ËÓÞÕ—C©Û%Q²ÙQ({r:›+ò=”—¼ú”û *©E‰r3ª¡Jð¥Æ/¨~¦µ¿-Šjܬ¹ô>>Úàž”túL¨ÿAÏÎþýåþ¶ÇeëZ%´68FÚÈ•@{Û»Ë/S ËáÍK‘6è5ê¶23</lVƒ¡«I4úõjðmïAS‡-iY—*6„ï‚_iFÁ¸ŽdUdÐ^ø¡Ä­óHO¦Lßå] £úoïoÂ\ÏŒÓOßXc³Ú´ ‹oVVüÃ’Ã5#&OX>ªwâëQ:ø]DUìêêë.óßCÍ`Sù—¿ZEÒîëã‰C"½ 1jHzôW½±,’³ªÿÚ{-)fO‰†tGjãÆ‚Gö¹¸ÃÆ)ŸØà'îüé†ȸì›tÃh Y˜3ß§E6g¿ŽÕÑ`ÜÕ× /n¦„œQýŸf®"O@Q\«r òu«YÊw*¢@~ÞDA–' ­h´¾:"16g{_£XÂŽ°‰çŽx€Ú%qªZ%ë®c{ëˆÒ‘{Žð®Ø¡ì±ô;Z=ZxPé˜Ñ·L6Tòô¯³‹AÕNëÙÀ5T‹P£wAy›×O÷ĵ,¦#nÔ4¢öÉQf* Ôä}è÷òêòM¦d@Ãv£tî¤×h\IuåuÍ 4µ’¢ñ×>‚f¾>îÃÊh®ÕM¤¤§ƒĆÌb±h©)%C)"ŽVºËéÿ±TóüË/i>;pÑ%‘ ZÍ_¨–Ê€¶6oÅQEè¸úDÂê&tí*Ï(û]uOc ®GAgËAýʸÐÞPu­ÿA,´”Z¥œ+†¦öâþQ‘ßÛëmdï9 P8©Y®™PÏÅ{è¦|“ãß3¡¹cô\–ó´dåHד©Î;íÇö3Z?äj´ÊB›ÇáÈcöÎÐ,df ‡ÕÇó {äÒd—1ôsîí<Ù_ë2/l4Áp[žˆßY]™Ùp¥gm‡±ý6Ù7{'aâ½û‡¼B˜ršÑ4\ЀYKG_÷8˜ÿ¢àù‘¼_Ô“¼ †¥Äk>±°Üõ2‹+ÀV%~z¯ñlÀzÁjú°'lz6ïIE¢ô–‹Å½H¢v×T]f I§[Ç5{"yvž}ϵz¤ŒÞSB4?Ô±›û‚Ãî#mÉI^£¤gL‰úU¼Ý?Oj“ƒŒ5&å È|êÞPˆ!²ÊLŠhäœEv¡³OÉBq·ì‡]ŽåáÈéxI§ý˜r?¯ @%ä]ÞñLn6ùÕáçp˜# äq•4$¢Ðé·! PØÅÎHàò. Œô%~µŽbå:`Òÿ ÷Kúïá!'B‰vžÏÂVÝ(åBŸ¿êzevEªò)Å óOŽAF]<(ø°ZxÙ¾Fo‘ÜA¥I/÷5o%T•U÷8Rí‚ØÍù¼ÞÞÕ]¦7cî.£ÆZwôÆ/WÔúpBóîqcÔ~ÁÿöéÜêÆ}Êi÷ÙBýBÆ'Ý&hø¥«#ìR§ç¯Ï—Р©Pß{;š±+úö ¡9Ĭæ¹´ ¨¸ÌÍù-I åK˜xЊ¨ÉLkZ­¶Ä?ðr=GëÍ*¥-5Q´Ù²õ˜gC;bê C玢=•– :P¥ˆÊK¡#ý¨ñל§èÄN¡4ýg ù¯ù­É,#¥Š=Õ³Nø²Ãd§Õ”5»âä²—¿6º¿=¿Æ‘3ä²Ë.iQ26CÆVœºíƒ1Hëd’@6)­®û&„ ±1—ß>ÖÞýVª¼çMتh/ ìÏ•>h©~ÙuÙBFeSR‰ð‚>w¦ªPA¾LÃi»? 3®— ‡¢Ó‡7ãi·çsëÜõì/¿¤0pÇý\y–véäÆMÂþ¯ ¿Ç;$s¾¸ EÃd¼6ÿëo¯7¤/v4ÁyEc¡4´üÚÊÂÿt½»þÊ—¢´5ïA¥‚rÃû™·P¥N»j»ìÕ¯…›ÖoðACø‹ß öŒ±L¾Ôw[iQvýÏßÈqqцÖÁ–ÃÓÐÞﯬY>]V%§øDõ¡WþКÜJ; œsYS¤€!‡²Ììö0¼5+ѸN#oÉ쪅ïêêëEÂ`œiçˆâ{aøAJù}ê>L±p ½l†™CÇÙËž’À\rÜ#ùË<°°ç5÷³ºZXŒ>ö)˜ó8,,©½Ø e!¤¿¿ïp¥Z^ß ë¬%$çb`£7õ×FÏ $²;Gò‰)‰ËE~ºD!)f÷î&ø YÇ×é—m}H‘Öö`V} ©©Õ¯VäàІ6žÛ¸3üW÷d*2²>9ë€Ì_tn§±!Õå6©ˆNÜÅ+d9å¦4«ý„‘‡«Åý­¼òÅßÔ=Ž×Ì3Õ¤ PVÉm6yQÿó¢fÅ/ü,É]ÃýAŠ«Ž6QÒ¯iÃõJ+qÚM ,¥SÊÎö£(_veÈù91*ŽìËUÉBU¡÷oX>W š¢g´zsªO¼xwÆ… 5‹"îûꌇ>>ÏÖøy uzê§ïÑ Þz=­Ðáv4¤;¾y1}m}‰W–ÍE“¡.`þUަÃmïÚ7êѬ"/ä^šx·÷Z|¼õÙÛü5Zæd¹Lu—£UZˆ¤m¬JCGÕ~¥A¢sÛu›Ñ—-{èæaw랪„.Ž>–Ì£ÐARsR»èÓö|‚ïU»·çìµ»õ-A§ˆÕ†¡^<=ÿ­487|ìoª€¦ó‰ooÒæ¤j¿ŸöBË¿±ŸuËÐÊÖÌÅj_­w£Eï3š@›ªæ!§¶Û¬Õÿð„N é7úÏ ;Åo•Ü`/ôµ¶¥¿¸ _ÝO½ØÚÍÃ1zÿ éÃHÊó‡–£Êð}f´ÈÏ«&|¥s „)ùË5-Ì0+£¾ÓßüÌ :À« ‹·žŽŸÝK7üí‹NvÀò+éêé¯ð{rkòp'¬_TxØv16Õk6ëî>F¢„2)ÛbB4a`0ÒX@2º?æË·–‘|ˆóüneV¤,÷Û­Ç€ÔÅId‡9+¶M "È‘ž#ƒfçð$2XöE ~‘BÆzêõæŒ"dölhs³pBVíΪ2Ê5dWû.ÃQÉ»­MKE‘óJ°ê£StÈâ y®iygÈÕ$>š#¿”mÃ* ÀKñdº‹º(dU!0éb‰ÂfLõóÍ(êßùöéc«²ý¸©â†ûe´/’ÇÑ¡D]”ùë_(ejEtsòJ¯_§t0¹‰²ãš]\‹Q~<©ØVæ*´ÎL¥wD¥·GT‰ ‰Qer™ÜÉK1âhú…19TW{þ|ãj4GG^óZG­ëÍE£‹£¨mþp܉ÑuuoïÕY½ƒúÖ7JÃôÐðøåÍýþThlê{$¦3MÖê*WÑt¦¢°àƒš}=é$FýÍÛ ø ¿ ¡E $zV å‡†F†gu/í~«Ö/²õ®å Í}Vç‡óßÐÖÏ@¾R3Ú£s/'A{+Íýûö¡ƒZ:yÆ]tÜ— dmçNlÙ+Iòg"~·¢Ø|^øBçÅÅeîN¬¹ÌPtY¸Äy¯+ä¾»+ð²Ú…AFsd칓íVEWk´{»„èè FH|™˜þ]UÞÅ ØÏŸ‘ƒw§ÙRÁÕ#¤[üò!•ûÉܤ,åöçw¾ ø\ÎÕtšßÂÛãnçÊ‹OAÑnò˜åo÷à™õ±ËcnÛùõœÂñÚ¿üÉ¢¡“Ý‘VÛuÛ@¦L¡Îs{Z©Cø¼™+ü¿ hºæ@ûªÛÿú[zÙys²X(9óÝ+qÞJÓw.ûoù@9©Õ‰Þ{#P±7¬¡'*ÝLb“Å®o¯—aª©=Ï«CIoûÚ>¨µ*VèB¨ßª>t‡>øÏßÒÈñ\ý­£S "GÙ¡ý»ùÎð1è2é“Ùår zy²_r¼Ñ„5¬õb¨‡!ÅÞs}G.ÂpgüNõå!ñzuI¾3Dx†ÒÂXÿÑ×~˜¨^®9x¯&»îéÕåDÁ ý£“7`ÎϪÖçâIX`y5‚ŸŸÁbhKHXÇXR9%3;Ëb[ìªwåà·­ä3ݸ.X«{ Âk9>Ç®õüƒˆ·µ'Ž"‰ƒ)ŽÝžBR¢£ŸŽ¯2"YÔÕQ6Q¤8£ºbú©*rtÝij‘6OõØ7î<ú=½É ý2îØ%‘ sÀÀíedͺØ|J]wñ_-*.ŠBŽöÕRKª&ä.šñ–ìA>Ï6ßžG( =ëÓ:8ŠBg•CiÃÖQx굕ìYÜ[×åÕâë,z Üù().•|¸¥¾w_F™äõ½MA(¯{ϳ÷ö4*èÝæm@•‹TÆ¿èï"Æ&×{¹Ž¡úñÄó BQ¨É¯xè×b²áôü:|þTN¾ûPO‘{æEq4hÉJD#åЖ¶824a7t“p.CÓ+âíÈáˆfGå<š[]ÐܨpABÔ -:|Ø–üÚ|7w¢ µªù>´Z¤¸ÿ=Э{Ì™-Ÿ’¢M)# R í»“êkÊhw÷qÅ¢ú´÷ìÞØ7¤ŽvŽcïjz‘>£èª{¹>4Ló\·<•Òöõç ´ÎÏœýœ!WöEkUûüå¤êÈÐùŸúgÜ-øBäÆPà48Ù­ÊÇûzŠ~äµ~2ì„·Š@Öª/…ù*A'¶éD®Kÿù[ ÑÉk©·ÿ[?àx¥é"´~Ýñî‘Ûh?‘½%F´ýýwSl¡«&¡{D¥ºÏGÊ®Z-B×±€A»›ÚÐq°Î4ãû0´zHµÞ8 š†nXK†Q|S’g€-êV²_>Ó¦!œGôÅõ²th#ù!Á ÍW#˜ » Z\5ŸrF¾‡–ñìÌåÛëñõ˜ºC½°{W_*´ï«Í1´,ƒNR2²´ë}Ð}1ˆòŠ/9ôÝ‘9à_©ù뉎m°SÙýs9¥Zþ4ë5|O¿Pvž>&Ñm5¶ë½x²VÆ(̲ï/P:ó)ï¾Tø‹ççd”S`ÉÙGùÛÁý°|20û‚Ã9øýn‡Ó—{"°¾ÿ²WÕ%iØd¨àºàþ‰^eÌ.½E§Èf;³ $“Øó))Ù)v†Ø.;#åL3Tœ,RZ ¨¦¬"í8óÙÎËH/˜yÿvãnd°v¼]¤”ŽŒMª ƒ™ªÈìOW­gÿYí*Ž¡ÃMdwr¨í¹‘»/çß ËEÎG‘2–g²‘»àgà+á8äÙÝ{w7òóŸûX䯇·U"švM —ÆΣ°F½a‰XŠúìn:D$b5Ïdp¿´Rasš|9‰Öô¯.ç._AëAÍœ®Þh“"Åûi×´½”QånŒv@"wôê"Ú“ÙÊv-@ºEéžÐ¸B}ÓÊö¨‹*¾÷ò´CékƒÓ^îü„\'­¿kò/Ï @ãèe™Ï™Ûü yNi>È8rN³Z²:÷o*µšoóËÇ^²uÀû§1÷UÚ!ë “Eÿ·òíçøQDJ±ÿùÛ÷>žÇG"ÿ®ÿÖ¢Ñîñ%sÚöÅ–®)A{†yÿ9ÛÐi~7@ÙZºwIýãËÝ;Iv—òAWº@¢àïèðhTäd$Ô}73KÔ%[¡iìDDgñ=¨h/˜ºò‚[&±„zîÆ5áìhq4´¨Ü^/g$„µiM4Ü ¥ Ú™i¯þv?åE¡7 >KÕùð™´m=ÿ^Wº::;N0¸)@·òmÖ©ÄaèCß‚wxa0¡g•8c†>«oUÁˆðÀÏoשáûÙÍšc‰Oabç}Û£;arÌЩ4<fÖ9. èÀü³8s„E{¢ºuj€%EonòX–\å¬%‚߇wµHi)ÃÚxžM¾wl|ý¨Ä®›„DO^¥†z"‰{øqׯHfj"æX¿ ßcc2j÷Y½U¤ÞÚÌãP9€´k«{8~ ½x63q‘.2Ø ø>N݇Œ­,¦3K‘9ĨpßáÈê‘Aòñ=²û´‰ÑÔÂÝOäÊnÎë gÆë½,žöÈ]ö±ùªòö.ß´F~ÆÛï:HPà‚¡_ø§;($Fäìªt…eCêï› è%ÇÕ±÷µ(Vç÷Ñ~} ÷+ôÑ·÷ìA‰¢ÿJU ”’~·9‡ÒeŽO[|\QöÞw’Ã[oPþB*£¿¡*0^¦&F¥]ªÃ pUl]š7õ"Rül*ÞHBµ•ø¾Ä¨áÏä%ZôñIQôLx胧ÇWKÔ‰âd:ðûê=û Z >8ž¼Ù€Ö)Ô;’»ÑÆÓùQlKÚʨOH|¼‹¶³ròG»·ê^f‰„ºqéCsÈß)L3Ck…‚ßeÕ£.‡ 0Ì(?(èBÑù×û´¶ZpßtÍdú%”ØáE(%uÅ îú#”ž;Íp¾ª å¬ó¢[]ËPAˆÍ’Œ•w¾Ë}ƪ“òTV_­QÍ¥˜üîû,TŸ >K%È€š¾LµæNà¡´îkhyu²Ë²"=¹Q¯ }îÒ‡44ȦÛGæó‚å OH~D½OÎÂïÖÑôzD›Ãe4³óŒ26Bsñ°àŠ 4_;ójn(-Ê} (gCË›Ââs%>h¥9ÛÏn6ƒV¼÷¾ü6ŽÖïW7®/Å s]õ·S±hK½ïÌSηhûÞâéæc¤“I‘K¾®M¸ÿ‡¤å«CŻ˶:7¡Ô‡3¸å8í?ãfä§þò ò»¯}°gZ…¼f™‹î“Zœá¡® Y®wÊ"jà½Çí Çj!ýºö¾µGÈ\NŽ;ñ]{»Ï8€ˆÿú[ _n{Ê®×ïn¹ãϪáÒmWDûŽKÆõÓ¥°¨ÉtŒ¬³–X<7¥Úúá×êáœj…pøÍµG+fÝÖâ~)%]±€ô|›Â9Hî.~ÄI|Ÿš^8׉d®D휻ŽÇõßw{Ì#•–¯Ã3_¤áOº¿\ wз?ïsq@z¹œhßw—‘ÁÎÙãw2v¼ÝZËÒ@æû÷RéNI"ëµgq*$/=Jw͆¤ we¯NZö g}&Oûe"än\'j§FÞ–9qÈOüêV¤Æ( Ø;3?&†B,<8EEPX w è…WB·]Ž¡Xã=Uj‡ZܯïX|I%27èß„¢”H¯Äf”~«bÇ}#”=,÷vÄ„ ååÓTèm{QR€-[ª[÷e³_·D•ÌÍFWÓòБ;GQ­}Ä7ݨ5ÌÚo’þ¬E;٧IxèÜÿCÖ{‡cù‡ÿÿöŠì™½²²"Dç‰ìì½Ú”‘´“RRQ(’–Jfö–½22²7É*#d¯„~·÷ñýô9ŽÏï¯ë¸/×徯Ãýt>Ÿçëõz¼F3¢®¡®Úÿ/QÔ—îWˆ]TÄ#ʶ^Ì!ÇÑHñÓ‹§ÉᙡJ‚_t‹ÏYŠ„´ÛÞìWD2!¹¤`„Æ¢™îÙ<ñø;}¼z­W ®*©1#Á~§_)•¡×ºÓÇtœ!bƒÌ¥oª±jðÑÌSûëØO(T¹zksë2¾˜2ã¿Ã|±”ì %”’jFVÿöï.rwXӤʅ":Õ׺;¡ˆão úÁ9(Šp¯”êÅ ,Gƒeÿÿõ­}·øLèæÎº·÷R"ð‰}]JŽÃ*cö/_ Uuçœ<œ, z9ˆøÕÔä•uíÏچυY¶o”¢¡NÊ;^¤‹èù) ìöCch…w^˜óÿÖ7³GÌûL ƒJýËñ.èüÛìeÒ = ý”˜œúz6×ŇÁ@v­ÏÈ+^JÔgp ¸#öõiSN0úËÊ÷s`5ü¸}¯Ç%ÿ8Œ ¶k?Í“\‡¨¥®T”À„hÅâ̸•ÎÕ[‚Ù_—<4]´`~ä‚X1,:³÷Ç’­Â2]Ý`‚ ¬t‡¼o¨_„µšsçDH‡à÷Ìù5²lØ´ â¹ ZÛc›%Ϫ·¨>:oýÖ/$Ѽìnxã’öÚ4ÜžDòŸ}!3†H¹D%t6iªƒ#\]¾ânJkud¨ºk·Gî 2½œÕïîB–̳u †ÜÈNCq÷¹m$rf‹(nf}AîçRgøß!ßÇnœµQàݧoÇ(•ùÎá—ŠL†¥—A±{”ϸ9:†‡ÿ|Љ|ŒXô¾wê.jXI8ýT‰ÆÃ ÷,I¢O Ö²Q¦Fê,0?§kF½…ZüR!‹³Í6oÇѰ(eôîëb4¾v¡`c˜M>…dG‘¢éñdŸÈA4û0ʹŸèšŸ%>©óx-ø.<›ö¦E‹VÅ?ËCÔhyãºWÚ)´âJ|/ëlƒV¹•_µÛ„ÑZw¼ÎÞ'­;-ãÛ ­ÑÆÞ¸Mþõe¤½*ô´áD4j©¨_xéñ+Pa2)}|yg\ù®siã?žÁÿðòè’Mô²sÿqJþ/g!íÌÒ£³ß!ëªæìüÄ—„õÍ!4°¢9 š­éb˜Þ‰B[«Pi1]É?I×áÐHæÇÐ=4ýéU*ôækxK/wBŸ Å[ûð[Ыá¦ZðÍ«:©çÎrWqÖ«’¬}öÑ ñ½ù÷<_l+­?®þ{ý?ï÷?×ý·¥ÍþXÔîµ¶Ÿã×äÿñLºn{‰uË-BÏ—pùvŠ×ÿ8 ߯ ,Èôª`8µ=‰$ÑFcx­/ðtÃyŽn£30ÿ~ÈÝ7¦^U˜Zïʥ¯÷wÁü#å‹4aQJhswœ,ÍyÙÅ(ÀʧvÅ °Ö#W@”í Ž »ûà²~ãýJW\µ…ÍO¾ @ïiX×2Q¯ÑÐÁºE :Ö£>>oCÞ{[*Š2h\æ^ī܇¦:ŸŠ2ñ;š‰pT_(Dsr×Ú}DŒh>@å³b… µŸð8?Zº<ê%­ÜB+^w?Ÿq´ª§átjFk——Gn»ø¡õ–]år!;ÚÄjP›7£-eçè»(c´½-QÿçK.ÚN?wV×E2=Ö?ß@U@ṵ̀ã?¾ÇÿpC²OüÇCù=ÿ_ÞÈÿ¼þ¿|¢¼’ÿË-ÙÓëÁw§þÿ_ß¾µŸm(¯ƒ VÖ»Öo~ìˆö3þ¨ìk¶öÓ‡ª­“Óë¢ZP#ÙüŒg j:%ƒ%NLÀçògUŸv|¯èôe¨_ ¿ÝJœwØ<'Eÿ·¾ý?v°PQ×–~†.ª¤5—þ—/Tá&ŸfF ÷Þ;9xþ€!'ï«¿y²aDÎ÷1ÝøA-ûÞgB~u}c ãTî×½BY`b8©þ6;=ü³¿àB™3 ,Âd ž0[&ûÀé1Ì·¾ùý%ï,ÚÞN.·‡¥íCÄÒaD°Rävý÷ ?¬…û:r ?†ß1&a?ßeÍÃΣ°}©ÚÍ.uîuK*dòÚ¸ I_­|š»!‰änáëÖ#H™vÇ…¶5i.ož¼­zéâ†"žœCn}ŠãLÈd¥P›EÑ€,ö§kt7<-u°êÛcCä4ÛŸˆW{ÿaë²= Èg²ÚÕøœ¾¥TúŒBŽR¿NÓ£H˜²À4é%c=·±$G‹ƒ>A!/qß×]k±…%(ë]qJ' åê-)JˆPÁi¨¤AùØüÎYžzURݶ¯‰§£šÊ'-¥£59ï^]}‡êÅ­ ¦ò¨é¬V"܇Zª «Ò™¨#í|ýáÞXÔ“ ¨ÎA‰ûŠæšxdA`×;’B4Šbná<„&8ô§žM%UÏ ïE3å´É%4ëx.ðNUÍoNŸk•'F .í†ßÛ¯Ð"=(®Ÿª-•¿²TLÔ|gP!uZ&ZÉ×·?ýŒVï}il(њ܀車îR;­í_TOžþðb*Ûô]Ç + º»Ùo ÷eõ¹]èƒü¤Kšyu÷!/ûÔóX ~ð’Æ¹¿¿ Ç{¸ÍX²¤³ƒ†þÇÇ;}: Ry"ÙçÙAæäžºñ³ÿ8'ŸâeÚ™/ª@íBt°DV74U~®gу¶‹s'÷=„›ÏßM¸¡³µ0ø6µ0t;©jøð0@¯Tû벇g ·?¶ oeg<®ö°Š8tÝ?öåž´¿r¼3øàø_ù)%ÙŠgïÜ€ñÑ_¤^æÂÏܺm¿Í0SC£YÓ{æj}|öß„…ø„©!ƒ°äaä|è ¬hßt«5{úgÎÎÂïöãÙ—D>Àf7ÓÏÒKŸ‘HHîJn ’PM7HAÒ?ë<ŠlO|ŠíùÊ“6¤ü”rͯÀ©_°¶@áÜåÍPò-Cé^uSR "ÕÏj÷A%d|ðò¦ÂósȬ~FÃg. YRŸ¾¼hì†ßd‰³k‘Óc ³ÖU÷\qa˜îòA7ŠeÚÓÊÈg^?˱ËùóƒTdÈn£ …è± QP(Ÿò¶ý‹Fùdv«qÅÈ%ÂßrU£„u§€JÍäÆe~BþCù5—Q¶Uµ¿â<)Êã=6Rú›¸∈™T3*&žoʃJîQE¹ D¨"ÿ€ÌKç9ªºrÊí&ËA`Îù-C‡ê7N¯Ú;l?iúß)s7B!ÍYî ªÓPœÛöµ8þã¥dOÿËs;\IÃù:(:þŠõ·ÿ®ä´hL|ÜéS®dê]ƒ"5Ã(WQ.(z’”á!p ŠF(KŒ‹vÆñ¬[‹— d¼<±ÆŽÊÉ(@¨€ *Þz<Œï» •\?”DG› JUy‰)`ªžšÉ|s„öጢgrð™åW±»ÔÞm×s€úc!øÛÅàK¿^Uë 4WÐY-±Cë’aŽÇ ´'pÆñ®€Î°‘CÓŽ¡;ï£Ãѯ>зOðü 3*ø:£oÁÐ ƒi5±vìZ0ܪjÇZ[£Ö‹§vÙ”Â÷¡Cç»OkÂXˆ7iò7˜8ê’¥í ?íDSCœÆ`:ê"×—ö˜:`îXó¾Ì"飰HR=ϸKÑõÏ‹aÅê³ô›Ëé°¶7¡Îc3 ~Kúi·ö˜ÃŸ‹ä§ÿôŸ­ñûNg‡]è0kµäc%$.ÏQUARc±±o…#H¾gÊèı ¤4]s ú4Ô[cgvû"]´µ{Æ2ˆëžzE3ƒŒ“Š»Ï\BæÙô¢×ÈÆã+¯9‘#-û­^-îñ*—úøÁy}ØX=r.·žêk ~ex††%(B/µq£÷z: G ùKýá‘Q¸gâ‹P–Ë'‚ã+ÊÕw/ÔÂý¿*¹ØšàëW"{LP…S¿=oŸ)ªF*üVnD`#º=™MСG¼µTœj²y‹Ù>žÁÃSŠ’»T•P»t¯ÇÜ ê¶Q¯Ï‚ä§ Õ‡̒φôufE¨)4È?òxá•VE ¬^®¾A{c‡o ç)4yëTÎN^€«bé¦ ÷ùÁýOöBvYreÀwȼt›Uν RB}؆%Br§ÞïsÍi¿óu‚áçäÓ±ç¤&o 6 ûqøsh’)â»ä mÌÅÓ”F–Ð>ÕÝÚé#GG½û·K k2ðÓÉ'ôÐsºÐxÏÚÇ¿3·ÏzLÆÌ›Â'¡Kîçcö#íW)ÏÙí‡V²o4+W… ±pµó›7|É[ÉÞc7m`qZ¢§¬–ë Vv…Õ_ž†<^ð[/HW¯Í6¥¯cn¿Wkîiæx#ñÂ{à Hêl¬œ°‡ É5}δ¯¸"%s€[`í~¤šì9Ì.‹4mKÄ‹®!íâR×ÍH¤÷˜´b³ŠŒòÖ'¼Xirûíùdénù~%8ÙÆ7š~ÅŠ Çß ޝ¸‡ˆî°Œ%r/šeL¼ÉAÞTZN^uä÷9ú`ÐV ê†=›”üPȳíÔ{ЏeÚ¶œƽo/öï Dñ‘ÝÓ«þÅ(¥äZoªÒ/×xgrn£luXU]3Ê­5ý)»Y‚û#H3¡¢• J•|G%–Ý唣¶¨Ü;¯­žUi]×;¾ã¡L‹kSz[¨®EUóš5’¶^r_®ÃÃ,ËN£Ökzçï|½¨cÍ;ãÕõ±W^®YCõ¶ß—ÔòÑ®›O{¢±Èð+­oÆh2×".AŒ¦Ý¾e÷N£YFŸ"¿Á(šû‰ÅÌùˆú-.¡%EêÕDÃx´,¨Y”¯—@«ãƒ’‹Œ,hµÉ´?ïh Z?Î&έRDVêÓvahv‹ƒÉËm)tý-Ç‹]‹ï}F’²ŽÐ}©Uðiºï µØ>(ˆ~¾ÖÆy–º_~RAVð—ZΣԄÜ˰+AÒ3ÍŸLµüâaëùxH œ z|ý+|ˆ ±(¦Ø^ÏñˆÓóKQ™mÚâªRdL¯—@’TëWš× )9–¿¤¹>™ô\|ælÁóßP4Ó£þ+ Šç//z“üÛou…Š8‰ìîpІ÷dUÞ™æòäÞg(lwõôà>EÄÞ EU{¹.xîº,•VDžþF¨›s©?Ëõ 9näF„²”ÿ‘øSJÈeŸŽ½¿vØ *oÐoHýŒ€ªg÷:ß~"Ô¹ç5ÞPã¦{‹¹[>ÎÄÇë¸A]MàÉ‹1ïà‹¢·äìájh&M—o;¨­Á&¿m~pAûAIhQ*‚NÎ=ŒÚíЭÈïP齯û“ï˜ÑÃW•xiƒ¬*ø¶ÆÐZÈÐÃçWEx’ÀȬžð«:ø~áCˆ‰¤Œ±= ù&ãýt\‡ÉÆÅɼJ˜¦mf|v¡~ÎÕóY~‡y™@³§‰C°FÎ^åKÚÖâlßÍ`yA`ìU¬µ(TvvÁz®²ÚîêTØøUÇŸ<|¶ìµŽ]c¿+ÖŸxR­‘ØS–­…1II©¿l¼D²Æ_äug‘bÙÓƒ&G ©šÏyŠ´×N3émm!ý«ë‘ñ¦¢/2¿ñ*†¢YŸ¿GQÿtsNä*S%I“ëAždåòwBÈï½k•æÎy´[Þwô€+ ß`R¦ž—BÑaòÒ€Á=(îï"þúB2Jhxð!«e\²'"·PÎnå¶aÞ#Ü-ښ‹)ïÞ&qAe±«² ¨J’Xüž¹y¬Wž½…ê|Þ¦OÖ¯£Fm~øí'ÛxøµâñÝ2Ô¾ÕcQpu¯+Ø ØX þÅÐ,ÙR ~'EÓúŽi74»íÆé$™ˆæââ,Y-3h^ÿgwZœh~º§ù%ZüŠ»üÊä(Z^èÛø¦û-§jYL“Þ •C’òi´ªªõ¯ìBª†Î ÕDv¨ÏJfzA Ÿú%â8ÝßÏ!†»Ô¤PÀlýèr÷äéÝ—å¤r€*swôÇÊ §DäcÄ È&¹ZAÐIƹðâ´yH–‰fw\üNÐá’"éŒd¼‰øÓäôsò‰}CQ`ß6ÔòT=Ùö’Æ·¡ÇBëLJÄiÐî3^f7TÓ»\•–- ë\¿†ö_:èa_îý–ñzÊ¿Íü‡îÁA›ùèºbœrHÚͲ÷Q@+M…ÑÙ³MиÜ1ëë _Æi\Å„^ìŒÇ9ªü>Mñ‹½*Z 9tÉÓD*ZêIs4ŽUC«^Ä‚h­•£ãå‚6ߔșš)hÔ)áæk…Nõ wš¢ü:¹xÒÞ„ £I^1;Oøj£hµÛÌ ÝDßÙ)GÁðŠ‚BYŒnj/EÔDÁØÑ@ùôša˜X8|‰Ý·¦ê•õãàWï^¯´.0ÿ*ã³ÿ™6Xt¥²X¸âˆ?ïÒÁêñ/ İ^–7Àç#ü»Ú/÷Àv¼Ïžºe$¦zpèi¬’4ÇYrò½G²ˆJ™ŽÄHá|úXX×R©¾Šd½riÄÚ‹¯Ö|AZ#&—}ó\¸»~KÀr=†X9ƒí‹‘ééKÿˆÙ½Èⵤh–P„lw9Üûžù!ÇS¹´¢ºgÈT@ñH ¹½ßºWvJ#ï™'\râÈX¦ýqÊ EÜo>†BbotO¿BZ>53Ü»¿ÏWÞÅ]jIŸÏ£dÎx]ü\JÓ+ˆ~ˆ²Ú=µB¯P.“cäÆÄ6î7ì‘зAEâ¹”K“Ix äV•Cn&*ßq$•<ˆ3Š*^çsà!ƒb•¯§fPrÈרÿ)j¸ËGO,ýFÍ6µûôWPËÂã}ìbÔ¡·q–8’º+>×H4Q â[C úZú•{üó4ê{Ci+óM¢VHixÑ4í-ã©4{À0Ã9Í­”Ë^¬Ž¡WóÒqž´è*­ö~z-,I:ß„V’zÇÎ*£UMKB»-Z[ °Ôi¡õ×’OŽËhcÓ’ÝøCm¬¾¼dE[–”3ñþH¼!oô1>ñüzÁnÈ Å½ò]ƒÓ¿àc°Á%mõ«%øëz¢E)dwóš@Ú2oÏs×MHñš]hÎ&‡Ä­ªÐµ3¿rãÇÕ\ˆÝ_{'ðÖMˆÑ{¬¶üÁâøHÍkQ9“ȬCÔóÊËLК¾ïâ3hÒ5a´¼ë­"zñ›ömMj}æË9Б`º®$#]9AŽ{9;¡—"lÜÅŒúÙ5ͳ᛺Ã6'+ Óe.ˆæÂȃw¢Ê¹ð„>£v‰~¼{ÉM½Æ­Ë/°Ç˜Ã¤z¹ð Ÿ%Lݤr §„™K‡£Ý0WF²ñóp,Øh›·äËÁâähé®]¿a9ØçÍà»MX5ZfO]‡uµYyCÂßiÃ-ð›Ý¸7l6“ÙÞ×µ‡¿NwÿDb‘Þ®KŒHR²¯HÕÉ|æÊd=†âå„ç£8'¤f̿ػ†„›E ©p÷ªäµ dÜ­ø¦óa%2’ù™ÐÎ#ËvåQf6²Ÿy?_áp¹HózãJÅ‘»Ç÷ÝWä';-D+¤‰½=µ7ùÙPhæÎaÎ3QTÛœ·]!ÅúzEjæ dlRcâ¾(ýÖí -eÊÎ=óÌiBù¿~‚q/+QñÜëG÷ù¢ÒåÍ axð,÷uóÙ~Të¯b~$1ƒù|ôµjx­Å…2âa=Åð!KmÔ–†__Ôå9V.Àƒú¬Œºé—"Ñ /ˆèÁS4|¼ë¬eéKSþº¥ð MJWœïGÓH.ãQ4³è|®Û‡fk'c" ~3ºkpä'´P™§T˺ŽuÔBd¤bhixwO-KZÖ&†|ôG+•²±¯%´Š¥8x‹)Ãè[òÊ¡Ž¿¯IïùK‚Ïz£!ÿ‡Jƒm”i!ß_lÄ+>¾¢mÓ!øÀ•Jà ‡»½oâ5déÀo /‚÷þ¦ú ‰Í5ì}EA¨ñ®ê±{¤ß5L{ܘôsRÑvüL"íIøœÓêâàÔgÞÎÜH…VK‹à_Ðΰš2¤é^ 6á?§ sAü“ÄZè~ü`Ýã0ôÈߤ£ç=ÝÇN¸ŽžÎà¿ (³gÞ£¢QÿÎþ¦!}ö^„\˜{ºJÈY=凡ñ²™³94¹=ãÔ¾îÍ{C~M€^½+ïG ¥?ëÖcWB} VF“D_> „u…Ñ «b3°1¤¤ÓX2[y˃ç‘è¡ÙÚa2BPÚð CÒ¥#T+_!y>šGü¨EÊ'öä¾mŸ‘Úwѧ6~Ü• '¥áôw‹2þzö¢‚:& B¦ƒ²ââ§e[€tå²í9@]so9¤¼Ž û ò‚¯â&™ØI›„<óôöh1"?5 ŸÚß0¶ ówVAÁgÉIÎI.=;v%ø6Š~!rZV›¿³GÑl%eO·1Ä}þ7ÞY‡oîg;¦Ñ~2b½í™ŸÒ‡„\O…ŽÃö„ügeyJrN;¸¸’Ðt¸5—7¹—P¯$µÔ{ èºØT•ô¿õp;ïV( jýž¼ï?ŽÉÓëÇ¡MÜØ`à ¶cé®RÍCÁÏ5kŠsPÐòêáÌâ+Âù´}ü‡ ¹.Àý£p?}9 Ü¥wZ£¨Ê¥¾ËReAÅrÕÉÔy¨$u飨•߉¸Ü%è šecË) jâ……s>€Úg¾Ż ~z`óÜûSÐ87×SwZnùÅkDÕ@Û ãËÝYiÐ!÷°ë(Ç3è’[u *$‚žËGÍÌx¡o.y]e)%HÞ Ý„!ÛoûÞÓ0½””“$£Ï¼ZÕ‹wÃáøD"ë¸hnÐZ ¹A'ÉoÃϱ€¿‚i*0cS0[¨­sÜ æŠaøÁãêfX ö4ý«Ë{ûÌ}÷ ÃÊ/ݹܰV'øþ8üž0Š š¹›ÕæaÞ°Ý‘&ŸÜº‰¾ª ’uú ‰G¥Yâ-R$“^¼öêê#¤P=~‰êyR½‰º®…»tÇI#Xp·ÈÓcܫȭ8Dí†LN!î›ÌÜÈâédu²ä²UqÚ·;•#§¾|ú[½›ÈMæ«è‘‰|½zÔ©z$Ìÿsõ žJ‰1¡Á-ÛÍÖ4”µí½kpÚË HéÑÄéNnÚár4½ä|_áÒ(š‰#£$7 šuŸ)yæÑ‹æ72CD¶ 3£ÏÑp´ˆñh÷Ì[DK11µÌ‚ïŒ >ÖüéZ±w†÷½D+?­YqF;¤°3’뉃ZâùÛÏ¡âû®AÃÀ (¡Ty7Âò__ÒçäN¼à¹©'9ŸnÝy~²O„Qo+t|&•éþBMH½Y3¨%` ?9¬ýàúž9Û´,¤Ù>éõeöç¤"°ú•ôT5|Ö{ÚáË5µ¯1óÍв¨~tÉî´ÅQ¼vЧ„ò£Ú C~Ðyõ†Æîè&egßìN…îxq5‚.»ÆÞ­œK…΃Ú:öAû.*nçÎtheóo?b$´SǤ^~†F­ƒW¯hJìŒSŸ¯8˜ºS÷®3í›"ä¸ëYç“¡ùòcÅm§Vh–JÒ=A¸?DGà¶-´vMaÀBŽûõ.'¡5 :ι;] ….QÙy÷0èQ>çP-©}±{­Ýldaàê^oÕJFqÈF2vg5ïú ß{EÝd—ia|_éXê‘4˜l:=Ôtq¦Ë.뱆9vOíÔ›aÁ©ÊÐ:E–,M˜¬aÅ9€ä°ÀZÊnòFÂsm(½*Þ+1 [¤¿L›–&‘ˆŠ7sù áÿe´ià–X’Éês—Erê=:1KaH1²çÜŒ Rõ„Ñ \Á]Ä&ó–}LHç'ZØ®ð”ÃòðS#ãhÝ—ü™·È\ªÒ·k Y?Z$d]QCö\ÿät12䌥"ÑWëÇ=Arßoh #·ùnëßÈ÷.±§÷C 0îŽÝdEÁ†s (|¦*¥V EOé&;éE1/%ùÎüK(‘:< Ý‚R[EA>ïPæÚØÕ7ÙŽ(ÇŸž:_á‹òÙÇÞ‘:¡‚Ã…{Õñm#©Ý)TjÜO)Lƒè©kY®%)õ÷N&÷ ÄteæYØ„ô'ó_ùBÚ@„Õ/¯2Èî–#¾a| òô‘/ƒ‚ðÚlí©s„zöST¾EëϤP[õ©÷¶á¨"fðpžÆ™˜Ølö¿ùÓ⊹„ûʼþ*ÿ×çŒMd…>^«„ŽQÂQè*Fáú+©e» Èý²äV! ”hr3ÞØ„2‹ØÜÖú¨Nk÷ø¤êàÓO¨s¢ýq´*7Ë2Úa+¨¡Àï%¤Lðy㦢ó&ÔëîvJi’…/Í¡Íô1М~y;ðñnh]:±Ûþ|´Wdß#^†ÎÌ“Žó¶RÐ=Lzì ™ô¤}“ ƒ¯í‹¶½[O`ð…“Û›c0œ«Nƒ3F…Gæ ý6á{J®Â&U3Rðù+b6ï¤úªá§Ù•©#ç:å¹ëÈ fo%_Ý› ó)5u“Žn°(óçÇÒQXªR¤¨3„•+Ú7>U‡5c{ÞºàøíÄ`¿ >Þ7ŒJ7„mÍ^÷‹§è‘èéa÷"‰àÊÕ%…H:´"C䣀äƒ=WüJäÑ­ûJ¤ÉLÔžäG:÷‰§¥W’‘AးF 2NpY%qê!ó0yae!²ñ…Uÿ8Ï„OÒï?ëhÄ=rQ{›¬ÖW^pƒ·«ù»#ÿä-y¢`‡²…&ŠÐOœÉƽ~zl ãŸQBÜÅ>E$ ÷‘öÛAYÆüÒéÖ(Ïx¸x¸GN–ñì¹—k„òT»QE–t¶(nUß?HmDp]¼ãþ³ Õ¿zö·@ÍŠÛ9æ¨uïÆ9#ê¸æ<âÏÑD=G¦+IßüÑ€Oô7ÄC<ÒñA?­û]¿~¯­-Mvщ¼ESÌÂ>²H4ÝNz¼;ü 𥑇ÞWkFs³ u9ý4Ÿ‘_º}]-nµßýêV‚–¤-G#ÒÏ¡å]O2º&´\™ÂɃªhuò^ôWª$ç0›:çŸé%˜sÃS¡âì­›g8·¡xÏøxs^Èsozzwž r•øX¤\x!‡\ó/³‡*dÅÙm úÑBÆ›¸ãýÀ)¤ÇøÎzåÁâ.A;ò‡ïZ–L´xëÒÛƒ©¬ü‹×ˆŒ`–æºÀ¦ù-˜ª{¤ó{‰öëÙoÀò¤Ó·Ñ¥w°& ëë ¿C6d‡Íý&S Ô§á/}wב$¦ùðÁsßA$‰ãHÖ ©C2ûµƒ)$ŽH± Ó&øRI¯ÚMÔ#Ãù\£¤TõgèàFú†ØÝüä'‘1èäy÷d>~Tbò·>²š\Ô~çì:râ{{JS:c;êEîa¾õÞÒ' yÈîó¸Ž Ÿ¸ÑE%cYäï¨ûˆa (h¾¾¼°g…ø.qæõ¡Èï'sÝVѸ÷ïœÁr‚ŽŽŽ'4›£Ô•Í=î (=ã;sve{‡¥J}Byg!ï¹e!T ’×vÿƒŠ?·ÚH¡Ò¥È»Jí¨ò:}Z3Õ8ýÒ3KŠU›Ú#ÅQ=½­ÒàJjRl{öL&ãa¿‡†ã²7Q[•¨Þ—¡.ozhZÎCÔ纟<è‡GlÜýmfÑp¹Á2LT sÏ„4_GSùYb}Y4#+`¤~€fÍ®ŒëWÐüq·ÈíWh¡¹¹p|-~y·=žƒ–ªÄºF£•™¨îk´J?uG¡çZËtkfœåDë„¿M±ÃhÃv·íÍ©khs+ö‚ M üµ}wùHT0”ºXÞó| ùÆŸ‡«9CŽ‚¹³ÝÔÈhˆgJöÌ„ts5ÓÔÈ$H)¿—õüà$NÙ+®ŠA‚µe”p %ÄQÞúþ5ŠÛ´› ì« Êþù‘ǽ4#GùÝÒ°>¨Ñ=yóë¤Q Ùx%Ý„l}ŠˆG#ÿõa†|,!¿±FE7œ˜»’N¾{áðG¹Ã•ý|÷á|ô¹¡ƒö;uÍè¢Ô-(x¸¡@y‘ $Rn­Ü!Ü_¾± _ùçãO}Üùøº$ó´0Ÿü•lcH¨Æg¯¨XB1™[¼ßt”|åò¿Xeýû¤†ƒ¡â²Ñ FXÁ'7·“AO¢ ²<ýöGZ¨>ëË1ŸO^Íýqêb°qê|a’K´®†¦ag›Ž6Ðê1qP2Úµ×C§­ Sºñ¨åš tG™IAo|f¥¸â]øÊ}È_y¾uT&{³‡á#拏Ea$Á4|—¯¬øpüühìbËÚão…ú,lÂd™ÆLÖÏ÷0-ìÇb¡¿ú›IT9SaÞš‚t9ƒ:zMÔ’f1Õçˆ:ôßfùo…¡îßg»ö†m þ§å•ó|†xäÚRQÿ©4â–1ÞŒ¢@ã¬p‹·Ÿ¼Ð”ëÏœÏ ‚ßìd¡Z¤uG³{qNWwU¡¹¸ó‚šíš×Xº¾·ç@ %9]Þ´ºvûþÍchyÌ,ɰO-;þþb‰B+õJ«à¿H:Q±-F)OðMæ¦Â• ¼§_ö¯ù0ɆK:²ŠÂÇÁ÷?ýc÷BŽ"Í·‡ÔPß!Ö¦‡ÌIYHRæ3¢N'èo¨ß܇à+å8MïCœó;Oñ)nH‘?YoÜàõsRþ‡gõ† Ô\£ ÿ@6 &?í臠ùÆ…‚ Þ8hm½âé¦ í|‚) ™Ðq›C½§Š:gUlölµ@W ùÉê s™“öîñ8èxþÞò¹ ´UJºe ­‚F62ò^;óHÊ4h_CcŒÛÝŠeÐø"ö®ÎÍuh,\“OyCN8þYz4™ M¼lÅnД/–4BÆÍ_úŸ,•†VI±aЦ¶œ²ÿ®%´¿:£ÏþH:ÙÕ<¯ÖÏ@W÷üëã[ÂÐ3ÇióÙqú}Ålx=–áÛ¥˜¢›=!0|²ç±Ù ŒšWFç*Ã7e; ¶w0ÞšnŸàgÜ„çnq-˜Ùº½m(î ó'nk’¢CöCØHXv×,|˜$ «/e¼¥ä`}Ò!v€bþœÊ¾ïËû ¶é/襛ßC¢Ë›ŸFä*‘xóÎM‘ÕHšÚ!Ã(Šäþ‘ª•?“Òߺ ~‰©?>öºY¹ išš›\¤æO½Þ‰Œû7:§™™Ø÷Ög–ÈòóÉ÷G–kÈÖó5!:à rÙl¬Í?D®7µ2y{û [ ï­äÍ]¤y¾ü/ü>¿oBÁ]•}Wß‘¢PáMås†(Ú­¨wò,î H8Åì÷ Åãûç=ºQræýŠ]6JŸâ¶Jâ @Ù§Z^g_“ <û¤í€Ÿ îÏ]p zù )t&o3¡' ÑS«+¨¢¶5”µgUsíµß7¤"è•ü¸!‰ê(šX—?¡FU¸+WŽV÷ÿ°‹ZË*ºóŒl¨Óâ¥&=ƒz­ 6úßVРÚ;ñÙþÝh˜x¿*Þë;»½ýަ4ԎÕhÚ7WGôàšEߨ'‡æÇzƒ]¼‹L$«âêТ(‘Æ…Ü -mKéË«Ñr¦ÆKS“­®=‰ ¸‹V«<¢Ün¢µ/O^¸Z÷‡šÐ@ôÉf–º[-T”ÙÆPÜx+õ}¾äe_ïºi£ Ùç¬vå ~ôÕÉËuGÙ!íÌï¶ÅL‚o4ÙÀ®nH‡½iùàC½L¼qÄžêPÞz½ QÑ"ßõ6¿Â{÷þ¨Ñs|œÁJ“âï ûj~TEa;Ò~eÈ*ØÿzÂr'Â+/×NC>Ye…þ¡^ƒë;ª£ÿí'P`7s‰¬ ¨³.?½¸³ÜëJš}á>Œ¥Z@~ÅooßéÏ„zöôLŸ…¼2ñ9ž×!ïÙáûÏ ŸÛD:n>˜çÄc¦ áðsQÈòÑP|ê©-Ó‘:(uôW=; åÖ¾Ÿ r®BE:߃=SPÉJ…GŸVCUÝõSW»¥¡¦î"yîÔn]T”k†«ö “ï& ‰É:‹Û9 Z9ï>9ùÚZeÏ5wæAGšÒÑÂ÷ «ðr7­™ôå$ðw_€~ç‡9¯æwx&ÊšÖ0Tós4å0ŒèæÊ¥0õÃha›aojüÐ÷_ÿáHãôó¬7Ã`R"òû€6˜ OàßW¿TJì‰w%À<‘ÇÁÂuVÚ¯éo`ñ÷Ì5§ÅXŽ7w§{«Ç?ïÒ:ëëGŸ·²nGÁÆ;2õRÖC°% =°ý`þnD1&/ÜEâÇ¥—AI56Þ·?Dr‘Ú\­RÅ! †h©ý$KßÜWÆ]ëÏÖÉ„™qw®‡ÇQ¶*d˜‘'‰B¦Oz~÷‡Æ‘¥÷Ák*ß|dçü ë7u9ƒÆd·D ·ÃtgäKl<>ö8 Šö{Úà& õß‹ýó¤E%äs³ûP,Ct2' %Ýõ]*RˆPÚ£¶eͬå„È­xØp¿õèý¾¾¨øÁ¯ñÀaTZŸ>#h¿ŠôÔ¸SðœŽËÊA}T?4•¥d\üOàáÞ:­ÑP;Ê‚_{uc™£ãõΣ¾SåBq=áÒ¢˜ÝDÊ ®½çýÐØ–7–Å×MŸh™J iÒíû]÷ߢ™ýaeçþB4'VÑ=»Žæ‘W‡*ÔÖÑBîÍpv»9Zó¼?¯gŠ–ª Æ„GGËlÖõ?÷ЊŸ˜Öx‘$}“.–wª¶¼íùŸ}"ø«bwù‹Õÿ¥q‰©Wæ¿uAÖ ÿõ7ž$,CVpËö÷wÈ”ü9ìÎ}R“¿ –w†D¦0÷Ý{ÄŠ³±O b¹u·r\ŸB2³J}É?ÎI¹Ç¤]À®¨¡åxA?¹êKÊ_~¶–‡¦Rª±^5h=zàärãhKíe¤ªÐ%䦳% äÐyºáZÊÓй9çS•H¨37­_í*> ‡%Øž¬qC[4UÛSqh“}-"Í{dX3Þïp(Æ/Pîp½ÜY7êeFÈuZÀòêe!ß9xsþ…&¢ø‚ÍQwÂ}*2 <Ð’zâ[I ´þˆým—íÒ§‹ÝH/BGH¥ôÆX;t‰Êìg€V&Û+Ç{ ïAbMH± XfçÅeÀе둊?·a$”¡ôŠŠ|xõÙ®[,0Ö‘ÀúLD&_±Ÿû`~¦çýûß‚¹’ü»í°0äôíàŒ&,Íò?—>–«t¡Õ¦=K°®^Êqà*l„±òÚ)ØÀ=u #«/ü-Š/úpî[Ž’:®#Év†`1ß<’õ•ÖYFŠoùù Û—š>U-ä@îzáÌû„þ$îö‰4h ÙF†°œÓnFÈÛ+“„,÷Rï^Õ¶D6WÑ=JRÈ¡ýwÍÄàr 3±Ó̾@nöõ ¾läU»üŽ ùíÔzÖ+’P :¸tƒ—…Nuo—ìG1ï")ÜKüFö¤Š Úl$ÑÙ¡äéÛº%¸¯'¿&£ƒeõCŽ ‘k¡\CÚz†TÜ2«®­*)bz·[Ýñ@ùÕ;S=s¨$ërkN]h„D§+"^? ¹Œ&ù3Ä”‰?á{f ï×7Ö/ªŒC¤``rØ[BŽ›È;d­ä´3rÒéå<¤œtµ¡YþoýNòþ<ÈUâ>FÅäyjº’ï/Þ{Y8àß~p;óœ>”‚‹W)'vֽޛ޼"äÌ/œU1oæ”î¿uBÁé!.£¸÷ú'øèë¤û)䉨y¾Ôy¿ÉOªtBÁ#oç%(œ¢Ï¡m…â^ýg•¡ô³óçEy€òïIf6ãg9îù”7T‰?h//†™sÔÖì¡Ö6æ«õ;¨}ÙõÇ A·z4Îf“Ðb'^üåˆ8´Ù»è/—±A‡dÌ1^Åt‚.¥WÌBC^0ÇõЗÞ1¼>¥ Æiý¯˜Ãàòó³íÝa¸Ç\¨¿FÕ‡úh~ÏÀ÷ÏÊ_ÙÝaì^‰YÌ‚;L¼øÓ¯!J?ލüƒ™WÉYŸ·b`Ίv™+Ѹ¢£ZzÃâs½}•°¼¯¸¬|,V&oûYVÃZ¿mp Yl0Tü¢—›¡ö¬5ìuð÷Ljïp®k>9yû+’,nÙ*‡Ë#Y=Þ"³ŠUZY:‘jê× ¯öø ¦$÷Þ¬Eº~&d8_Ì”)ù™ä~ŽMg#‹Â_R®@d;.Há}µ9 )?>ßý÷\¦KmBÞÏÂNÌ À¾(ï3y¶(„­äÚc(r[Œ4Ïq÷.RÈd¼NF‰÷ÄÜû:šq_ØÒÆ!Cm”M(>påÛ®’m¢"óɨk:E¨tô¸@·â&ª|N à²ãBµðù£…£OMÖH|Eœï.†Màá{æ«y®s¨í(_ÀO/‡º:áÊ‚[£¨O:K-ÖŽ¥OÞí¿©††Î‘cæjÛh´¸Œ²h üññå4½ý“¨`³ ÍdGi­“2Ьo…ùu 4¿þÒj„¯-h‰Byý¹ÑâEu…-«&Zr°-5,ìAËPI]ǽhùKægO ñ5%ÞåP¹ãcýÌJ¾wMü¯¿^ÿFr«µ^‡Rk@¶Há:ÖCfë‹ÛË>t:ž·ã!Ÿ”ê…bBîECôÔ .ý¨Nˆn_ VûI ‰[5rÂõÄÿ8'eÃÙ%áuo¡:Füå/¡^üy÷­&KB}‰f¶é„–áàÈåã”Ðvè¯óÇ!h÷ÏWŠ'ƒŽ>‡ªÂ‹Ðéüz¾¡:IÂ…^° Bû|n¯àO6hóÉ)ø%´­ÒÌÌìMÐl½ÛdÊüá÷F<¡½cé2ÏKµvÖyÏjSñ×f_Å<øR¥êua ë8*ª¡Éíå_Ÿh!j!ö’Z…Öà#1åòì„|xr_‰O4tPéÎúAçÕ+Ýš·¡[u—à©"è W ¦Š„¯ÊG.¶Â ¸º£«¬0 7ëS4= £²»F‚áGWé­®]0¤ë7å`êû“¼ék˜-¸ä­)õÿ‘u–aY­O¸§¤Kº”¤C‰D:¤[ÅD,DDQÄ QZî–n’”’RéÏË>ÿí¾®s>͇w]Ïz?¬Y÷üÖ3Ï=0wj—¬¼ü8,¸l¹t°€¥;…¶"w²`%”7øÉ¬•]Ë•Õk…MºwRW»oÂÖCÓ¹ ÷x$J1âªx‚$&Ø*$ LB€:4V)Øs€ª)ÏÕÅ$‘"õüõ©†”;HÛÅ5)biˆ â–s“÷I«šßW¿YxêøN(¤#ë†tì‹ñQÜÕ·ð!V 9’´‹0"WtïLò4¨¨™kú!?×ò¤¥u x[ÅøD²ánÙÏŒ'pOÉéR Ïy~µë©™Q!оýB–܇{{KzªpŸ‘¹Bö½t”&î=4à€2ÞWX²ÞþB9º+ »~¡|*C¿YL%*žú£iw÷߉¼¬Ú¤Ê“bõ·· šcîJ))ÄÑá϶SºxPµÔ>J×5>Ŭ‹Þ\DMÇÄf ŸQ[g0ÉV( uÕêÒ(ü¿¢¾D™`ò¤Ò¯E7pO Qu7ïé¥4vñ%ÇóÝhâï °ÐRŒ¦:¡§}ÖÉÐŒ¨ßúí 4KŠp’tAsK/o“04ÿY]ñýâN´ðU?-)Æû-Æ#ÑòÎÏ ?WDË•Ïæâ­N¦X^Ø×…VŸeιÂza§Ì­/-.j¼Ø­C™|a–zÁ}Öggãü|RŒ/+’)CÒ·%î°â΄ì,‡¸ð±òe}xÿq,5¤+b8n ··©‰{M¼Ø ¼G9ó03D¢=o-‘±;n°êβAâÃG8UÒ -:ðL¾¯dÜ}·~”²ÜŠš Ÿ”BvísÿôÛ¿ó¼sœ#ìÏgîÞ^åp|Kà06ÙÒBa‚þÉÇÎCö¬û½nK‚ŽÉ•õJæBÖQ…{ÏmwBfÔæósÕW!KR,-l7dµ qtå³@ŽÝé.õe‚^úl˜7vA¾ù7ëú½ PQ¿Aü_`~í©¿¥%¥ÉByÐÇa©]¨|é¸gټ݇FïjúYt$$ ÞãÀ© M· µ}æÞ®?E8ã®A["í1¿Ž^èx6j4`ªÙy® ì&нÊGg3Ò _ƒw_Iˆ3‚Ýê3³9Ò0tþî•ð˜l®þqFMGF4¦Šôs;atåÞúÃï`‚Ô¦•²R&­Sòy`z%”*øç[ø‘ðð¼œQ!ü*–”¯ø3óÒïFâ¹§`¡tDœ¥¨–\eVJ5xaÅšKÏ:Öüýå{nJÁÆŸ€ÃÒ`«-/Šíš ­”“—ž…š—Yå4Ýÿú‹4S$_ÿ³G ZªÅfË“< m'Ei ´žÉªË½íßIJyfe¡ý~k*ÿ•ah{‹¡Ù_þú´È­ðP{BÓ“Yöâ¥í~’ô²Þ@CÞR}ÿ4PÙ2üëkRÏ3ºÙ¹ðêŸGvÉGòü½î_Ÿ“×ûwý¿¾&ÿú¬œ ³m9]Ob®X.B/{Ó5 ø}}Tœá0D¼¸|ö: Wþ¢ˆÒ!ƒ‘®öQ¶¹)÷¹v¶3V&;_>Wù¨³™û®Ä9]‚_ãµû¼ƒ`~¡Ïðz-¡.\›˜Ý’ºËs¾DÓ¦Œ°ú«×l§ƒlЩÆt¾¹¿¶D}òˆîQ™C0r‰n dù!õ`ç§éä+:±tª¯ ùIùùP0ëуÓ;á«dò·þ¹(ÌÛ?—aŒ¢\ft‰†Çq¯1û‹Ý­(Qþã}袔›¾¹\A0ÊH+ÎÇì“GÙÙèc\Ì(oí3üÁi }"úFq¿ð/%’8OT¾óôÑUö¨º¤ÄI-ƒxÖ ÷×T2ª¯YYŒ0Í¡ÆEb¹Éƒ<¨É÷-ì ji›ˆæ@E¥#§»£ÞШC^å%4¨Èd¼,€Fž¿WõHfј7;qëhšØ¾œ™õASÎ_¡÷òhÚ:X” ­†f·g”¯j¢¹ Tzà²*šÈêX”K¢…Þ2ñhÑÌ´™9ý-õc~\u‹&pûìX·*Z t¦¹Æ Õ=šk,O+`Mîy•ÊÄSÈ^]jX|4éâ!Òq©ó*ùFú+A!£¿þ#ñM™ø!>¬x²‘ô7Ï¢Ý:÷ŸºË ޶G¤òµáÍã©bªûW!RµV2íм3Ðר§¡ò7?ÿ]ÿßûe&XV¶üëC”}Çx"¸)üïÿú×Oåoü×·è×ýë‡òï:eWz7¿üÍóÿoýÿ­û¯Ñ¿>(ÿ¾oþúüï=ô¯ÏI=ù¥‰t•¿¾&uñ{ WözAã{¡^d€çn2õ'ÝÐ&êylÂÚÿ„߯6ÕƒNÆB«yø×稷6”:ï¼ô›Î\¡|-ƒ+ƒÂ{¿èÃðá͹Ë/NþçkrÁ¢v.é+Œ_Ϊn9y¾³8²©À´Uà ³áP˜Ý¢2˜_†_'òkrE_À\ýŸ0½tFX° ~u.dá¿|m ØÍP««š’©ñ°áœh¦ò¶ ê¬¡2%/‰‘’X’ÔR¾»‹d R·IÔ¨‘¼‹íÕº3î(t™“s¶Gªú—OÅ¥‘Fø2k,g:ÒMD7°¸h##³Dj<©!2ÍÎuAx*²Lq¤Ÿ >ƒ»˜ ‹+m‘#4¨PJ•¹c6CwD"?‹×“€wL(HíTÝûÌ÷3E—-¡ðµÊ¢ô§PŒÂå¥Ð9ï7<\{Ð ¥n»ÿ˜þ€2U¡ü×ì%Qn–OwÙè7* k‡ªÍPãþ¹ò[×Q刦Î!9„wZÓG¥„ð £Ò­¤%Ôè~du{Ý5ãýø¢[QûaÈ^§:Ô5’•µA}2±[ÁÏ Ñ •îF\Ï2=½Q8‰‡»MƒùòŸ¢‰XÑê­Y)4àËܯ†¦÷Z²œBÐL¬WCpì5š•ÖþXšý„懙IÄSEмµô(¡…¡UtÏe´(*ÚcÑ7 NÍñz]&‚²ƒûk=º &=Œ—Ë ûøÎJ•Æ«aÐá&}o»o9}y³ŸÀm'É\Ÿ8A²aÂó™¶ÿÌ›Íý±gž}.Ÿ¥…(⫪^¯ÿÆ8ú²~}¥ØêÆ›ö-Ptä\qØ./¨H"½ïVnտϯ.zõ‘XšWC“‰ ùÞ¼ŸÐbwg8 Zƒ¯¼|Ú òÙâ^ÇC»6Áµ½*n+8êVz Ú˜Gîé_[øç;¢Ÿç~h~ŸxúÈsrhœûí2o#AЫU CåÓÐ`_Mœy–Àm>tP¯bÞr&?ç?}{¯[#@VòçMÝ}M×vx&A‹V4˨ïÐê(tÿâ:ɶ³–÷`$´—}Þ•&ÿ¾ÛD6-Ï@×Ù‹©1Ü|ÐÓÆQÚ!}ÞE›Ï>jÀ`è2‘ý‹0|3ËjŠÁFüS‹<ß„qFc¥ m˜T§]|ùf&ÇmHûmá×Ù¢~Ë÷Ö0âî¥c4`ÑNÜæmw%,zUïïÔƒUCwÖüáuX¿ðëGd‘l–i)ÏvjßK:¤1:¿‘ØDçÓMc$Z9]—Û„äb¤¼D´ƒ¸Ã<Ú1_«©šj((y§‘&²ŒHÏv éóµ«E}FƇÖï5%Ù\è²Ý:² ¸\a# F¶9º™?«_‘½‘zæ[×2rÖGvæqk#÷1kV-ä» yØ¡ o´…î Eú(ø€‰Â5*÷ˆžK8Ýq …f¬ì/0£Èhü„ø îåPíbÛ[ˆ<(Ö…QJãäÓ`9”áRÉÎ:Ö…²‰rdå‰â(¯K%{þ¦9*l\eœ¼Y‚ûûNßlHAå‹»h™ÚQµâ°ýôUD )ùã¶æ¨¾À´)h†çvFÎkG &ïs:wÔÚØíæ©:?)Ï¥Tz¢^w™¬Iägq <F#¥+ó4xN'w¢MŽ-ù(yp¡©pÂˈ,v48W¯Ž¢Yôåüg—«ÑÜÂ'#›_-ˆoO³Ô E\Êc9UM´ÔMmT¿s-G¿š´jA+ï±ÜYÑhÍÄggãÖïMšgÛaUN0âòN7·¥°ŠBºEÕý‹ßd •îR_úŸ&HÒ‰5” ßî'ÉN|H¨3oäU˜Ãû%/¦Ô[M«8*[^Q-R·)5!üÆîƒW:BX{IïQ:xë±½ï Ä|IÉ8fš·=ç»ÄT™ Rýi¥«; ýNûKN…ëé(´ðäÝÈ*0éÔ6‡léåýÁ¢ÃýáÍÄÒédȹ»'(š¨ÿ¿H–Ÿé»¼Là6Ùd^È:ð]SàÐdÚôÑÓ ¯þ/ Ÿz!c´ØäÞý|È<ÉzC„ ƒ8­Ù×~þ`™¹Š‘S¹DžÛsx\´>B¡+ã}‘!M(Éôuuꬲ$ÆÕ¼ùY¨(Œ«Ú5± Ÿ›ôj¿+vCõà*Yçq¨;qŸNMhÍ)LÅZƒ Eµ‘lWñ9h}ùÔ ÒÚ³þúûÏ—B‹¾Œ )Ð5"Zg? ½÷×Í•¡o¨ä}:? Þ^7°Ùb€aV_cÞ…Tøtï_ÑRÑ\NŒ0®üºw9°¾ß¡ž\l¥iF„‡uë0›•T]ÙðKýzýCC^˜ËÕ—½' Z›t…‹`qy®çç ,w=ÓÎ2nƒÕ5âÕ'éaãdë³à"OØâí6 ØE‰DW‹‡fO"‰„ÇȘI’ÒûãÌä‰9ÚƒòI¸ãõŠOVšReRøo"Í®+žqa~H×SLøZ2RÌ×ô:Õ#S×·…>%d©qz*Јlãk×>êÑ“µgÆ—gûÄ;-Ž!äºvV' †ƒìÝÅtp÷êJ6þã(¬5{h×µ&m™ô9äõÅcõ^¼åí@)åžû®Er(=X¸.·å*•}l/¢Âo¡…”¸ÿ”< [7ª(p?ZçˆB¸)òÅñÝTÿ2ÃEŽŸoи/$¡æ+¥ŒÄ+è}K@Xûá2êjh«š@}â÷A§eùÑ SºÄS5ìÙ ÆÃËç4-NžF“C•s vqhJ‘'{âsšVÌ-µñ>B³3GÊÐ\2øÝí¡hÞ7}@MÑ"°ÞÖ\«-¥…e´vСe—\¦§°/lþ)W_„Ò—O¼·‚ _ww[Ôd™Þ¸·¢úþŸsõ±º–;»¾{R^Rï4 ÌÝþ.±´ÞLà³ÙÉäjð.átgÚwBýXÿê Ç™¿ñƒ*Íÿ8’öïNm˜Aá«£OžCOƒyça¨>í˜|D1êÖÃ'¥lÍ 14뉕q4÷¦…óŽC+m¿_e¡> ¼æïÙm‰»J[ÎjB›õûµ´cÐz?Ðc€å ¡ŽN^Z„¦Ç©Œ„矶V(?éŸ}¶3ãû þ÷ëóTA£N‹¦‘>@е1}Ö[Ùÿé[ö\íùãÛsQÍžK4…gt½´†kªJ6Ðê:s¥nSÚ‚_ô-GG@{qû—Ÿ´à‹ZòXã=Uè©ÂÖoúÐóà ÍŽsg¡ïÊÀ½ñ0ÈSÅ{"f†å½âäzÒ`D¢Fz¿"-ŒÅŠì/? ßî+¦Üz3|‘jÇÍ´àgëiÇ0m˜k–~¤aÌ ùÎèÇKqîïÚ߆•xúžó§a­ö½²Ó l ]Ú×[ÍÂâ5*‰H´%j³ïÄ$i¤±~âˆdŸƒ<(Þ!ň-eñE¤2&¶$›'Ôƒt߯3‡žDzÒ}^.dÈ([Ù-ág‚L?Nño®Œ#K6yÏù›Èæß8E¼ìgœ ÈLŽ#ç¿O&w¨‘ûÃÇ›ðùØ×ãìÃK‘Ìæ¯ ¹}Káî^1ߣˆByý´å](’ò§$Å5ņZè(Þ>{ -Éêµh¡e¹aD-ZÝbÎa&BkÕ-;óçhýÛo€èZ(Ú”‡ˆV³RŠá=Ÿ­å»ÖÕ}æ+|j?áÀ)‹-ŽþÈÎòÉC¶â]bf˜%äÿùî/78±óù!ú¤s.3fN˜Ô„(ñ{µ)¤ n›ÎÝsÂv½ˆ—“N„ˆJWñÛîæsœqÏ㊙Èÿ{G1¡.u`¨_~ é27^‘VhB&ß“§í4<å´rôÐHd „ßpG²-4H–•¯CŽ(ÅKÆ cvXϹۓ «šÕ%åyd~­÷ç̱ƒLJª8•' C^·%Ü«2žI½i-¼=/¼¡`íŸ}w¥Ýyµ,Tlà 9©—]˜i!ïA;½0·<ü|/›Ã°%ðeh{1”·5™0'ƒŠ³QJ qðùá#¡ïîP©ü-¨Ìê8'îƒF¢ð›×O“A u9ééí9Åi,½Ð~ñäσ¹Oà‹ãì)^.èòŸ°Ùú]=K^“3¹„ºó"+­F; Ò°º wÂPir“Ù&|³tSÞêM‚ÑÔžZ{´aœüÆî6ø®KìvòÁLÕÌÕdÚÃì%½l™Âkð‹oÓÉËq0÷Fvë˜,(ÍÌçÂâÂÆçáØ(Xþª¨óè;AßJÊÝY<`Ãj>¢­á÷Ö”JÑâ$º²iq§×IH§¶Ê¶d‘4Ýši3RÉ}¬—*~jàw¥y=¤ ÷ps¨ëDJ÷¯Iœ¦HWßkvÏgÖs¿Ñ4é#SYÃ-®bdùÐú(7GÙòUä÷&V"‡%ýÛOÈ-З3'!ˆ|¯ÖRî|FÈMúI_*»ë*œ/b ³°ÖÜ<º„¢oN^÷iÞî3Ñî½²i†’ë§Â•¢Œ·°­÷΃(^+Aqº:H{•….â~åÍ;ö9¨B»Øù‰HÁ’ªÄe× TŸâÜó,ˆ5ÒÏ–Ÿ›ù‚šc ‡æPÛ%yÂ9CuU\Û_C½?!*-wÐ ÏœT7RÎÑ‹«I¡1‹ïæ}j%49Z?U˦Rfi]FÓ?ÌŠ*q›hÖ8òjWûš‡Ûж3¢Å… §f/´TËw;ê‹VÌK¥¿’ÑêW¢æø­H@Í4%]ã=—w>‚Üùå’_{!ScÚ(=©>¹‡L°oÁÇÛf‡I Ü–Î$v‹Š‡¬ñjµàÃ*ŽH™¾‡˜YʼnÖGåðÖô…®©ZØßøÞɺ±„í8¤ S„u@³Ææ}•$(rþå”TÕŸZù•Z u–MEòµ aD™ëö˜&4 l"É´<Ö÷…Ö‹î6 ÐFGW%{Z‹åóØ;AËñ3®Ÿr¡y^Û±mžš3ˆKBC&¯€-8×îtÜzºLàÃs-wOœ€z›ŠLþõDŸ53ÿÓ·š°„µhd:”bÑõšJ)æm¤2¡ÅiñOLcÍöþÞÑ-G>ót»|* Ú †í½÷Íà bñŽ`èb™¯#Ž÷€Í4+·­øZ&Rz‹W¢7÷‡Ó¬ÂÐ ®x¿£€o]VºÝ|0¦¸Õõ£õ|g&¹çè¤Ó–6‚…Õ2ð£ÝˆÙ`ræT{S…ŸÃgú¾!ñ£°DâbÞ>”+¤‰¶°&b{W–˜6î´;̇-róRVH”L¹5p•IÜô­D,ì$ëÌôG¤¸ÊXþ,)›•c‰?»ãNú¹þîÏHw­´>!C·ìN7dºÕÖ1ÔŠ,úoöt\D6öà Ò¸ëwk‚PMrî<ó`b² ¹³6EŸBÞªj"ë›}+\4_‰DAEöÍàº+¸»~ô×ÜÕ% l¶ë jœâ÷$3 GM®é®B¨µÖüH8uFý1¼õj¼:7n A|ŽL_³ ¹$Mº,£±ˆAGÄy!49—UïAЦPñ°c͘¼³Ïy‹fcÃÖNMhžÇÍtu-ÂÊ®¼ßhyŠ–ì’éM´R™t¶N/DkCÖÇߤÑzsŽ!Èå<Ú|?ïõm{nGY%Á²-É×ã>D}¡f€W >™nºŽY+@rÏ—°ÙFeˆ×*Ž:àqVö7êfàý­áº”Í»ðîçÆ®Ñ{,Ýb;$@ß‘´•{ÞèBXlÎÎMÛj.B—L"(µÄ¦è¨!zÊ-¤Ô%â™üâ¿8ÂGåOag… àÓàîW³¦tѽ!ÍêNYl£oÊBV°FoPÿEÈ&þγ CàËïÑ·¿Hý¥WyýDôpUAmYp{^j%ËÞ.ÈÈJk‘U‘€ôÙŒú‹‡j C–ÎeÏs‚Î9ºnÖCfrñ-É#*Ûû}9T#' X¼(* ¹[±)ž&¼ö÷iL‡âZñÃC,YP&•À’Á *r^Ölá³aÛíIu&¨>æ]jçtj;´Òj¹›¡á£l\Ͳ4rÞ,Ǿí»ÌrFÚ«¶fäá cÀuË}è~?C»B¨;/ôÜuà†¯kbvƒã 0ðR:$Ë8†b©½ÓõàÿqÊ‹ã0Ò1f줻ÑnâŽÇ÷ŽÒZø3Tq7µ7j­dºrþnFÁOïÝ[F½RëÇ¢4ˆÌUýúS.úh§¢ñ^Q¥6g#4¹Ì(Ë꾂¦ú ½ä£åh&º#c«¡Í)9…ÂΘ¢ùô…õˆQ´h?$ó‘æ5Z––PaG«¬»0¢­3ü¯ýøä€6…#UeFÍhÛ¡7fâýí~óô^}–Îx+•“{A¶jÞHg<Éö<`G§9Trßk§Ü/ˆËô¨¯SßT™­FÏÅBì™7WD9ø õ£JØ-%ˆvë¿òZÑ’ÀiY³$«šç;§þ¨ ^G³#]ÃnÈ!jö¡œp]7Åœ¯ ĵxG)/ÉöXìØö·|Ý9нÍYMÖýãÛýc%²LŠ)bê£ý+ã,ý­KžG›¿1«š­Ý¯ŒuÛ§y­É&–ÀÏÈhßšCÆ)ë5Ñq^Hã­R¥)!è\æ¥Ssï}vµÈTâ|,ô¢2,S ÖkòÚ¯¹®†”!?ÓðåÊ9(6÷-z“‘¥ß¿k0Ë@ùúu‹kû¨á3 [ÿde*Tó”¾ÝKÿjƒг„ûšvö{«Ðld³Cù‘2´ªÎ‰zf}„¶1‹B¾nè(äuÉö[ΚÓZ¼ÂÐ#p¤–Ÿ‚¾ÞU‹÷&¢‚~ƒüÏ“0tŽÚic OtÞjb‘˜Rá×áŸ`ì‹ÓJL´þ¸Ä·aJý—òÀL_@ð®€Uø™w)žã’;Ìw´¥‚®·=\°øƒ>â3G-,K¥¼HË «*V$×ÁÊgªbƒß%t R)èé­ÚÌL$No—¿aÓ¤>±+Ä'‘\(ÔüáiWÜ!ù@DÕC ©.YY›ž¿;§nF§süBº4¦î͇‘ᛋk¨.2…dë•„5!Ë©ûv|#C¶b O<$‘½8àˆ rE¿ºªgòù(¶ljÚ£€ãO6Í ¸û²› »Ï;Jsl×Þ«…¢Ò——®ÏáÞµƒŽÜ9‡QòúµŠu™^”‘ËxæW`r†WéÙn Âí2.Åc¨4õ ß`ì**—¼wämTB h|~‡ˆÕËtR@àƒ’¦ƒ"¨yJWWœµÍ+Ïÿ¾ºbÙrÇüCPo#qšÖ *^Mž~†FwW®rGc“´‹ó¶ýhò´§·ìšú$ š¹[æÐåw¡¹ëÌ M¹ Z\ë(ÊuEËûWÞÿlD«„׉®›ÖhýõÅ`YÅ$Úî~uú©lPî‘ Ín†¢}W^˜­ïlÝðâE—gÞªë R…Tµ#ª$cô$à.oŒ!$ä“ð _J„¸‘Û…õM2ð^ FUaÊO_©†¤= ü†P¢Ø]Š¿ñ]él±Ü7¤UvspötCž/·²§C”yˆä1‰ÁçÐc9¢ç¡–FÖÑèæ^¨÷8$ÓÃðT óëú yžì„çŸa\jÐòe³¥÷8´Ÿdû¼|š;Æd:ס‰ËÐ7j!Âßx`MXZK¾í%è½Eúä=B¤‰ð¥;BˆTÉÇ ª îshΞ¦OÿéÛỗ¶7Z„û žìS%èeŠÈ×ñbÂ}t§w®Çøñ äIÕv?Ë»ã/¡=ëw¤¾ˆáö|ŽœZU è,÷=!ßÝ9ÚÓ&bìð•ì©ÿ…G¼ÐŸèt‰ah†ìë^3*dÀ7 ŘóÜ0zäãµ­Ÿ%0>¶$Þ}À&KkgÝ?QÂLï…™ûLð‹]ò×Ï]Œ0÷õ©€÷2,äžržÞ KùC{2–r+ú§Ç#óÕÏåâ ‘UzxÇ(î’|}Æ)ï-rˆd˱B®ðÑ«Óù‘WÐ!ÝX• o%öéÈ,¢À0ß|ÜâqBþßïÃ|…]…lzQ„šLOV,ÅÅ[yµP<åùƒŸ‰oQÒ«"è#?JW®&:…²ŽòÏ,ƃQžZÖÝ+TrivmŠ£RÑCÎ3'|Q™NºÜmSU}?8œrñBÜQÆË<‹ê½äï;t¢ÆQ¶#~ãÐîÚ›ñµ–ž(Äul¢N¯áÛVw~ÔË•¿Ñº©Ž/ót¾Ò« Ñ¾Á¯¢ÐxÙ›W<ÐÄír…ƒû šZ%üò^ü€fjO?½@ó½f ó':Ђ'?×¹º -9t.j:È¢ïÅëqŽ™h-µ›ÞùÚöù>BÛ›7WÅåÑ®$ÅiÕ÷á<)gÖó ]W^ì^Ú|,iiƲvGöCâ{5[ѰðÁ/®{Q{ŒWõÁF”öÿœ·‰QÈ#ðØMŽ'vu?6†þÈ!¿Ät}ãáu“~§Å8¼atMºùÏ÷Êô‚¨?*ž‚Sáêª6‘K.$ÏY¥Dø•Ã'å«¡¥ßA†AËÅ0í:ÈtšWM‘…̉Ç7ï?]¬£òæ«Äÿ—œ{2áö߸Ý'yÇ–p}rQñ±aÈ(s×ÜÁØ ùž2ü3ÿqç—lHcß`]ráŸyÇò>΄ë<"¹Õ ÷™¿äJYmý¿_ë×AΨÎtµÏgÈ¿p\êô;~(êï”°6‹†Òx)ÛËWu¡<¯DòåB¨¬·?0¥gUý¤7Ö²ßC­‰¹%s4PÛrâºM“ß'ê_z»~ vh{xÝü&t8f¯œT8çw“}—´"ä¥Qn^ò|e6]æä$!ä¥ãc´( I=üÜçò †³•bÕujaÄ%ìw¹ÍŒ9 ‰<æá„‰—cÌI…f0¹TwÝÅf^ì¾F}>~zL§TdÞƒ93"ª/Cñ°@[6]zn§9îgwƒåñSº[†â}[ã«O„ 9ËÊeøÉ·îô€‰¸w°^ŒnAâ c 9$zviQÖÉÖ”I®ðQᎌ™Å¹’HeY¥[§|w~õ%Ë -Eº)nÅJdèܹïÝV52ÝŠstèµB–ƒc¶R×."›|Ôè!ûQd¬¤ùüM¹N¸µ,GÞb_ßýw¾£0$ úoáîƒ ÑÎoéPÈç±MÃEYo¼ ÝÌ„{K!iîEJ îò&uBºu)òçÇPn.ÑÛÀ½¨`YyÕÊÍÓöš¡òkï”/Ũ6˜ê즌ê©Î-]š¨q}ߨV· jš®^ÿ’#ˆÚZ&{'ãP—¿ûîCŽ=¨·’Užô jòšòª QÕÓž1v4v®!NC“¤¡v3†»hšÀ'¾£HÍÒŸå\_ú9_ -:ìÏCóø« ùÀ>h*•MÇõ‘—¹4 ªžŒ #O‹ÄNn‚¾ªá²hæCìw—²î[ò‡§Oó” îù¹ÒÕ><ŽÂíö…(úûìã(#3÷àœ•X¸’º_kµï½Gétó¬Ö(k)¿”ç‹rë…UuâäæO…S RüuÁ»™x`9‡ˆnù!ªž³ÚýÞ´a1rá¡ß To/76¸¬ˆvXÈw’ 5w-zÊ݇Z Öå“÷¢ÎždÎú¨—ÆxXM9 æ›õJ©£‘“mïëïOÐXjhKÕ#M®7Y$¸¡éÉDŠe² 43çJ ÛBsÃåúǬÑ„*/|.- >ô~j@«›Üòš®&h_2¸~½m&–T“‹ÑNµVNŠ2í“f¢îÜÅ£J‚%¸žÃ‚믖Ë5ðæ§Næ‹mÿñyZwHšÛ¯M|Þ›tóE@ìŽ;eYqÌ#CWÆmfQ…‘™,Þ©ÚõŽ–¢ÞÈ|í¾µeCÐ5–‚š5 x%üýæÝ[ìÖ^&´»[– ƒo'Í# xaW\2ßîSÚÿÉg{^UXÂú%È W¡îKýHధוµ>AfˆBã½Õm>s^{úq›ã*Ú™ÿÆÌ µ;ó>y8Êï'OÛlü‰Œ ½øƒ›* ¤s,|}› éz¤âÁ]îɘR¢NÐѶTòòmçh>¥1'Îú™=9ÖAq‹K7¹Ð½ÐÞ ENâo˜æ^AéYOáíP~œ2;Í*ƒ"C¦¿CU¸¶míÑl¨%:û†^ê³—ÄrR¦¡)@‚╦/´Tï{xÐê´iýÌW äƒFcÚ‰vèÜ9üü@·Ä››ÿ„Þ(ÇC9rëÐ/ïG޵?apFPgWÔ1¾½û Q™Œ€{Ç +ŒÉP*gi¸À„ý×·×vÃd¶S¦íÌ”ømì ?åÏe\ÚxsB󊳫`þOÆ)²õ4Xü¾Û=þÑ,OíºÿäÁAXÝÌž|<À"uŸ´¿Àï[2£ÑNdðg‹ÕЧ‰Oœ—ϸ$ËçN=F²†ÞOV=ÑH1ñZC}}R¢ò!jÜÙpçlÇc¤{aàà~j[µ;Ok#Ó¹×wˆ8ú‘…ÿ Ž5M²Q²ó. EvS’lïðeä0“ŒvA^’*· `•UóâšÆÝ,B;*‹¼PÈ0g€~ùŠ|Öv¼¸G÷ªÌU]ÑÇ}«¦þt,¡ô”zý½ N(»É=/ü, ¤|ŽÈ¾–B¥0®b–ËÒ¨ìj¹¦j¹Oφ|wEõ7CÚ¨áÈv Õ 5Õ㓎ߓEmů3ô:í¨»k½§ %õ–KÃô™Ð ±)Hlޥ˻›ÆÁƒýö þhòy¢þ›÷o4mÍMQáD³é×JÃéhÁº|8í!ZšÛíU‹hA«¸=†# hÃê1t×c'Ú¾“Û*y%ŒöVþ¢/Áj¯Z±˜]äûm+~χŒ·ŸÆ~­Ceâñ(§(·=Nho†øÈÜ—ŠË!'b¶/ôÈö9œÙV;{B}i¢zG "W·¬ûc­¶÷hˆ›\þÆŽ;ã ÏïÇêQÓ™<ÈyºäyÖŸJíó$¾*3@åY[aBP]þ’ž衎ÊPWZeêC_{+>ë„F±™––æcÐô”%'!ƒšU¾äðHCSæùñ~ÁahLò@_}hÐ4 s^Òzvfƒ@_¨Ûïà'·jWÂ:ÆrBpêùëV½3¾xs*ê‚ ým>þÕ·z¸”äxï³Û{=†f#ó£5Ð’Þé.rZ“›‰£wˆA["E›Ù…gОÆ1²‹ÿ=tŒ™ª:Ö—·¶JË¡[¬ßWÜü1ôêlèZoÑB_zGt@ ¾LtÕ{RLÈ£½?w„¿cìô´›0–;ѬO] ß%Tr%Æ`ª÷gåCQM˜ 8sy üzSÒ}²å1Ì;Û_¢Z ƒÅ+¬I]ó»`ùíÞ,«{þ°Fù3ar$6ro/lQÃVõ.9-·5$fpg½Dæ€$*Aß®"Ù§Ÿ²íLHÑÂ7¸Ät©”S¾PtšáÎùÍÄ›€t_SN?#E†åWaòþ·)ü—¡ÆMqd9{ïëΞDdspHJtEö _ËZþýÈ9Á¡H+<Þ«§µ7‹‘Ÿë4ÕÉ‹ú(,ü½‚×w3}õ¥Nÿ…{Š|‚O&¢pø§u…ÃM(Zùî° …ŠCÌáóª¶(ɶ¸#’H ¥#lŒEï% ¬¦‚-/Y!ÊMëì"‰{‹ /íRU†QéùwÛ¤I<0¨¦ÿk UÍÈ'&d# ß1Üœ#Ô>k;|^¡†µ¾¨èÙ7¨ÉáL€Zs:ÅewlP§ùGÞ)#7Ô‹ûÁï¹þG{ŒµÐèÌó/ø¢±lÑ W¯j4¹%1¶àަΉ<Î,—ÐÌé}r /š;W>‰±ëE ?›üz®h´|o<ìJv­zê ߘAÁ#iw½ÐÖo_äNÞh·Y0yfl ¼tsѽë‡Ç,¦ÆÈ*Â|ÐŒáöü߬Ôà%"å~øøº% Ùï:¡n|¿ë”ÄòQ-èZXn\áÕª‡h¶§—γY»®!s"‰7Ì ôû£sµ7)áåÊ@4­¤0¼<_äúÆ,Âv…ìv™‚(;ÿ‰ð¾ʆ† ’â6JÛ‡”·çÁ½y~o?A—Þ'ð±?ƒŒÆýDJÒ©ÑÇ{OÛ2s}ﯟ½ Y¦.͇¼ÛÿÆÌûWæýÈtêŽÎ-_±m}¬X[­…ô‹Fn4ã‹ð©Ý!;éì»íùuAh§Iл%·"&ÇI–ìMȸÌÚ²U™Q¿wP^ž‡ìÚi±{„!Ï)ÐïÙc(œ¿MÕôä” SÕÉy@ù~ýoWø RodnÚ,ªìï+…’UCM^ñÒ0õ+¨?ýÀr‡úh`£ÐŠŒÙ>§ZòþÃ,´nšÍ,äW@{Áw²¼°ƒð%%JPòp&tH½šY‚Þ}^O-Œæ¡¯8bï¥ù,¼.Fâql† é±LEFhL’Y[Í`tcÚèëÙ*˜ào ”Y‚I—»Ÿ‡fH É€syìaN@î\\ð(ÌÏÅöê=ð‚ű}'Âôay¶¾N°ÎV»eX=ñ„ ®W_¦k2á·Ã¡¦^-RøSÉ}ûô*- `<>ˆ$Õy~¼xdQ7Ž«¼Ø•¼Íª\H%üA¼ gwÝ­+ =†twÏÆ$¾+D†¢WÇ)k?#“ùýÈ(Û3ȼ¡à£Š¬_Û&ˆFv΃Ǭ?#ç°T°Qà9äÝ@ó‰” ùgÌ[YÒ{PpÜÛ™³o…8‡N¹õ¢Èã’q×­pïáJoŒkp_.ë–ß?”þè´ó+ʶÏ||K G6Ó2‡PÉuä£g&*ëKªòmBµ§WM#$ ú=ÓÙ¶¬Ó¨a¾Ãd_Ý;Ô”²é›¶Õ@mѳ)Jÿ .=Ä”Æ!Ô›XïúÝ Ú¶¬ì‚Ñ(Ý^ßJEìP¡æ¶®ñâÏ#ä˺¼»yľx÷5:ú9Äg)+iN·¾1×g´©!4ïVáìü‡¿1Z«éÃÈ!Å|–ß¹BvØ·]l{ $‚L¹;œ*š¬\»hªyªG.D’@­YK»}ë+ÇUõ»y´CƒDÅsÆrAW¥^?uÕšH+ÊšW¡ÑAÃËbî8]¨·ùg¹Ì u–Í×Þ’‡@mèå^;ªD¨½=ôâÏÌÔÖø¶,?m‡:Ù+†ÁÊ”;¼Gž'ù?}ãŸ8öªaïö¹š«<»@³ËOë†ÐR¦Ý V´­¹!S¨ì -uê\ý¾ÇО"Jö; :Ú?ܯüÅÚå;þ§ùõ÷:è™LÛ{X‘ ú”¨ JùH`B±÷yÏNâ #¬ ßwñp˜à ’g˜8±ñ€¾&ûº]ú¯õÃL¹]œR1'ü\¥¸K± sµVrÖ^m°Ð˜v9Э –6‰ùÃêyÅÂs*‚°Áy‚;i.¶»}ã^§ QN¥lA ’¸TÎøªˆd?tt™¹ …‰­äÜXR&|1jeͦnß& ¼‘N8žYî£>2<ýbHsé<2ɼÑçBæ nÚOd])¿Õ(øÙ÷¶·ôMG#çsÎ _ï"xe‘F_1òýù1Ä4‰'Oµdh¢àhèîV’TÜó€\µvî ;R¶e» èƒAšX{Qß!– „ûÆh´V®¥ ô½£Ì*ù†(+§Pè÷¼å¾¶ž­ú؉ ·Ù׿¢’iL{+/¨®¸¡1­†ªÊYÉb,ˆÐäøá¢Ã0ªW«ÄT<\E ‹2—üÆjœ0µµ~î¾sØ.uTò·¨Y·Àœcƒ7 º/ ÑÉÆ–Œ W4V­üàó Mü\Ss¯AS¯D¹ñ=¦hæ#•hqÑüÉïcëÖÇÐ"3®`MëZN[pï°&Fk¥ >wï¡Í›µÛŒ bhÇíH|îÚ~Œ ¯8‰Gow-üºïW¤ÔÞvÃÜ+ÇÝݧ!Ë×{=·LR^R8ïY¹ qV6˾MÕÛºFtÅž¢Û%îŽúAä¿À.ˆ¨¼¡×v Þ<^<>õ^Gªo0­åÀK𸻦§´!äã3Ç4aãí>%ªèú—/Ü4!ð`ÓÒžODzXÿ2Ý2¤ºîÍFà-O·¯A†×@ ‚GÚv¿ãåÀ¨jÈtŽÝPÓ%ðšYoÆÄæ‘ïi\ÀîïÛ}#4ôß’ =Žtá¢ö6·E>‡OÁuŸnH‡Á§XÓI9‘4øTÐ#m£> éǃšI<ª }6ë™ßçgIIó Ôï.d«Ö8 [Aî—“‚6Þìô ’ò‡’ꊟ"P¶ò¤pY®*©D¦»¡Š³Õ)©j®òèˆPPB=“€¯UÛ 4‘/¬†ÓsóT+ÖКžÍ—öã,´;ú>8c˜ _Œ¥µ®ëüw£[¡'ÿÑÀÜÙièÓÍ5çºAÈK©¢/<Ï`˜mƒövæ|뿃ïÁhUÓ â‡<0>V“¾8S“â¶ïöJDÁtþo¥ö5ø‘z¦¸ü~9ÌQ6r<9E ó#/MqkÂâ7%£`2îíóªÔYTΰº%ëÍ£1ô¹ÃÂú/à7Zæ1 ô-hwrb›&¡}ùƒNI^7ÖL.@27nzþÃ-Hõþ¨–RѳñÈ'TáÎAûJ9$‘ÎóÎØh2|:z]EJ™Ô¼ Õ_Ï"sßÚîJï5•âÅZRÜ5âK5Ëô9£i&×B‘gl‡Ù¬Æ3ä¯p?ªW‚¹_:\6MqÏ$å÷žòQ±Ûä[“'îeUôvÙ§û®E1‘{Ò£tTÙ#®(›TõE1á:Ê÷«DH¶>A%½¤·}«3¨¼›Âý¬%ª·T‰oBu·Š›×÷¸¡.RT4 &/Ç^3ÔæXpvx&ºèËÞÙ†z¿¨:õîö¢A7ØÞŸ’hTFºêYG„ƵôF•k{Ñdm©ˆÔfͧßû+þBó½¹™CÑ"ïOÈòÛ´’|nyG­+; ø¥ÝÐö¶%[ØÏb´?ßÎStþ½ïúþ²X ,'[ëæn½…œî jÎ.H[øÛåz’º˜;$Qmï³íŠ%èÚÇéÆúÛý"*¦úSÛs8ƾ§ÚÞÇþqˆ›"(µß¬OÃë&ÝK½l}cÔŽHžôLÁîz%µ ²Þ5e[Å«B±ßK®õŠP¡`^tá®5Ty®åÔýY„špËcg5 vlI-Ý®ê¥Â,Í2 :¦©¯Â# õŸü„¨¡Á„1JÒf êýÈךŠéze“©º< µÏ¼ýu'jÖ}e^ƒZaKŽ×ÉP{ÞL骓 çB$ÞÜwƒ:ãÁˆBžþÕ7Ù'Æþ2õÐøÈzë ª¾°íŸQ¤KRZPP¤û&D¤ ;@Â@AAE,:¥E%*ÒÒHw È{žw~~ß™ßóך3ëìufÏìk}®ë̬k—ô×õÕCƒÿ†ßð¹&hlž/î;wšÊ.´ZPû@sN=©êÜhI•aßÖâ? ¥îw û9;¯Ž3tD8±P>¡ü¿çÔîlBORÛo è»÷#ón. ´8ù1Þ†’iòˆ€Ÿn‡‚^FøüéÙ˜gŒ÷s‹WÞ‡©é$ªw`6ï¹æ)=XÔÛH Ù K¢ÚR²k÷aÅÔÿÌøOEX«u­÷û \”ÔšFm`ûIhýžBk$¹ùgÝaÉhΚ~nùƒäMï÷<¬Ç]~®³ùLHµ«"܈÷fHº r"ý3g²_¤gÜ16F† ŽI}=jdzGÙF× Ž,ï…~͆«#[Mnš“Áä8éþÝͲ ¹º|Y¡ð<^³©ƒÈw\¼¯Oó$òçWØ¢ ‰ÏœÚÆ1>j™v‰E 9j…­™ðP~¤žų=u玡„㥎³Ss(É/{`Œ¥~šç½Ê¥A§¤§ãñŠ:«(—¡•×;„ WîÞ\qC(ÐÚé©JEëgž~0E}K’“a¨Æ´m®e*ƒêÓ;=Ç2Q£ü}ËŠPj'•?YÉEm‡oŽïôPÇjwmQ5Ꞹ°Çc?êyOR½¸Wˆú^©§/JK£~ÒD£%fÉŠâ¹ß«ßÚÝCãc&ÆR"ð|àáËœ¾‚hʤÿA© Ír#mª¾Á .1F-Ñ•xñŽüƒ×ÑJ#¿Ö|`>npWJj1Ù‘ØkØÂGã·Ýþï9šHWˆ)¸Mq©¢ó>^šó8IäÚ;=%ÿ¾¥»[¹ŽèŸýè˜%`£¥×\ƒ@>‡[!W!`ñšyÙÆ$‘{챦©ïç\~âKS~~N2ƒÔ‹Ñ­{ ðyèàÉ:È<öh4nÒ²DžÓJÑI@V¤H‘×& d3ýü±ÅÄ ÙÒ¿ „þ³r2.Bk†.K ã7h[ÖÖcЇ£–NLÊÐ%Î>{2azðþ¼ô%ü=mH¡Y n¡VßNutrôåAyc^1,s*û£uW¯«nÄÃØ@Jõ6:˜t&?TUU3ÎÌÞ!0ÿTûÇâsXô]ýýH–¿h„f9`M¬lðÚ¤lô¸½ÖE[Ýæœ·Xÿ_ŸIJG±ÐÏoH~‰u”îÒ«ý9õÚw¿Ä+.¸—vþ€ÕÄ&î+¼Ð×ô»éÞÇX>|â fr&ù›ÈÄ_ÞÌPÈ,"Þ ôLŸ‘íf¨u¾(²N0yF®v'#ˆ|S³Kb}€|û¦S¿Ž!ÿKæ[±âÿõ$­çz|¥ÌFcïäëñPÈ ~Ö%wW(I¾šƒ¯†[sû¡$ƒ¬‡v&JþÞsö›ÊØÔ£§?'º]ÅùåBVï÷Ï¢Â>éí%Éâ¦ÚŠGPéköªüǯ¨röÆmÕ‡¨Æ˜º¥OöÕ'{ö]§@¢KƒÀ°ˆZï<ƒ'¥™PûÆ÷'íŸPÇB¹¢^Ìuå |è_(£Þ;ÍЦ{¨ï—jÕ®¶ RTžІm¶›©z—ш]ý»Q¨%?¸à+µ%‹ç{xò‹Ðô¿HçK4×k¯8øÚ-öí}"†–&‘›JïœÐªÝ|­Ê—æ¬rËX/ ‡ßŠR´î ¤½w@–AÏÿÕÿóO7ÿtšïzNá4Óº ڗȰ² z]u^ÿÿæÿõšüÓí?=ÿÓ7arÔã·!þ×g’¥ÝÀàedð¿{Hþ¿ñßü¿ïÿ»þ¿^£è$Ž’—ÿkü7ÿÿÿ½÷ÿ¯¥À×ݘôÎ]ø^-"ŸäÅ÷¹ìe¾HCé ®é Pªß´{có¿ýáŸî.·îÒ#ìýO¯M\ž6JÿO§­~=õlO•¡=ìjþ\t¶~yt8¼z.̆®ßÙ }¼ŠÒ~aå0ð±9…,Y†î óßµ#^ètÔü¿“è·RQ;Ö0uHަjè̺TÃÆ6ó$ÎÖt‹•ÎJlаܫ-d笫ëã>]/Â`}'>5é#üYû³~Ö¶vœßxö’«$¹¬I~ø´%]@²“+f¿Ž!§Åí‰Ò¤4Í|xšŒ©úWI„µn#õû÷K•ûëöBám“ü¤™H‹ú£‚Œ VÜÙ¶)4cgÅÚYŒVöwÚ"Û½m‹ /!oe‚þwqä>GÝ¥øòÚçùŒ<ˆE~ ¯ny|nº÷N …çš>,DѸc{ØÇUPl5[ŽšS%Λ+ú×¾EIó»y§ŠLPúÕzßὋ(»!ÿ4áY ʵ~a¾.}¹Øé òuQI'1´‚,UØëöä@ÕÍö>?P}{ïüç^OÔXö¾·³€ZS¤Së¥1¨=t…¯6ÛuúŒüNМB=²(K6Á먄<]Pª œxb:$ËѰ“[îÅÒu4ºH59õ†˜Ûö^Þºsý=š =µÐ«D³©3Ÿo8 …¤ƒBb4Z^,ÏùFBÌÓÙô´6SoåÒH3é§ûî–ZAòÚì“Z3H¸­Zs’^â’­öÝ t†Øª¨×jCÌø}‡WèÑüW®S}M…°) í ¯ël¾—Éá9_úÞà–בQï²dBªàÑÒ¢?Ô%Ëòà‡Â©…œö½|Pb=k"Rå^{rÞÕ‘Beˆ&úÓ©BU©±²G*TÏéô·l0@Í`½…ØÙ·PkÅûšŠN j–³ïÞû5gÏ5ŸíU†êÛùéQ'¡ªœ›7%–ªÎgèG5@UØ–w®…P½Ÿä´ütT7¾K>~‚ƒxÅ…}¬^P«ìK—ÏÕ?ïwd{÷ ˆ‘‚íAÚ4¬ÜV{·ðš¨tå>ÏBÓÚ”»pí/hÛîºÍ -m/&K  u`Báë5hþòÂêÓt¼Ü2,‰Öƒ.+/:—Únè±ìÏôÓú˜Ï«“(ÓÁ@ Yò˜L7 i0yÃ;',«Ìa´~ò i ŒuÊÎÏÀ¤ E¡ò„;Ìr5lE|‡yëÀ¡aÏ_°xá9)yÊGXv·2îJƒÕ1˺yØx±Û´G“ ¶ì>w}7†gO³}$½¸Àbå2‰äÔ»ù/}^DŠÙö9é@ÜÍnzò›îñ3ì'Ã}Ç~^˜ñFÚu‹±œ·a÷]ê2ÈXEÕxd˜™ËçBïvŸDÖåÂÓ×ÝÎä‹Á'rÑœ¼ií‚<¥ínG}!ouò_s?äWii'‘iA¾é£+·Qès¯·ø±0<ØðÓ£â™9:èþ‹!Å¥âÏ].C‰ã2UÕ\f"ߨ®S'ù¶+܆ºŠêŸ¼Û†Qïý÷‘m·Q?ÀåºÊv ¤éGŠZѰ#ôb<8mÝvË¢±óui‰Yy<¿L˜îdjFÓ÷%Ù§‰:47`˾!жr¼»/£¥)ÕÊÀ´$‘oz²yhaö†kwéèGÈvÿõ-«ÿ%¤VÒÍëÓ&Cü›÷ÑÑ~*Þ}ÜéäÆ@ÎacŸ.Ò^ö‚PÁgúêž›l¥·ÂlCáL”º¼ÞàßZj̯Ýþ=ϳBùB°)'ÙÌD¬&{ŒùIAÜü4“ g¤Ü,T©7¸éãã ½*!@(–Ðcàà…L‹·òó K<ÍSxŸd¥òå×B¶p=×¶(d‘¾ß|²N‹œW$®›Ù=¾Eb¢ „šd‚çéì‡gTÀøcºÚH¬é¥‡ªN};qüë øï —½™Ô×—OdnD*Ã>IÈ9ýåžUÛnÈcw ¨Ù‚‚2?ÒþúRø¾S•ÿlË Šýåê ¦/AiB’–ò?(ÿ´6£tªÈÞó¡3Póbîb–[(ÔQÿu¿Ü ª 6”‚&!ŸÁ{9ÐümõÂcËhh}ïæ[íä í˜×N<‡Î Ü 2žS•~¹ÝÐGª¶8Ào !'ª?…!sWõßOÀˆ|ç!«@ø­ICG#ŸãöÁ?ó.HÃä’Êé-òû03Òœ•¦óYzù9?Ó`ñÂ{.3,§ $¨sðÃjEG•+¬×4DZ±ÁŸb[z•¢Uت:z^¦qþö‹|y1>…$ñNt.G%‘Œ÷ñú”ðR¦Yθ!¥²gÐn¤ªf¿+ý9©]ؼ`ÚiµÏ¨üÌ%Czßõñ/ƒ‘qŸACúº32½ôë‹G–sn´ïOýE¶û$ÉÂ4ÈÁ¿ks«í&rã, ûgäuäµµëD~Kò.ª›(è•æ¨Q€Âó”Ú¿ÚÒ‰|SÏaëB±•¤#,—ºQ˜IøÀÆ”4=wT¥_$9¹‹²ËYþï#’Q®¾Â¥± ™Ç7·®*iÝé½|£UXö^6ýž…ª«ÇKëÜWP}͇Ô#O5fÇ y´[PkÔ5H4t™˜ßŽíµy΀:]å„ÐõH>îî…Ǩ/öžÁÅ µSN'pùFkMxOFäÛȹ;—\ðü>C§©°j4Íìyöøš-4ÿÛÙ…'¾·ýx–e‘ûf\aùè¶Ù32ÃnÏȤ#ŸŒÓÝ” Yf"êc†&Ä×¾ú!ä/q''ßåÌA¬úrG•'ÄÜí]Ñ{ Q÷Ío±§CØ©†. j!§âù»A·Ýäe߇­{ó8æ)qÁÍê]Fµpšq«v í l߉ܒ½ó-ôP~|¬oEŽ*Oç:ò…AÕ£—#ù«/¡úã¡dæÓÎP“ðå›{ʨ‹Qêyj5µžã{ÖÜ æ¨y]RÑg¨>Áá* Uy‹éUl¡Ê°qÿ“Ì¨Š´e™úBÕ½«/š zäšjið9¨¹íùÜj¯(ò‡«gÂÏ™„ï/’ þôQʳ× ‘½|æ›i 4‰‘$ÙP@ó,5?ïha¼‰š1ÐJ^û:ÿó´1xØìç`„vSáq{M!è(”Ód·„.ÿE‰ï ÅÐç&³˜ð úT/ËnÉ™Á@*çƒ;QGÞ^oÒadŸàéŸ$0Úh{ôÇ««Ï/N ©IJ‚½ª0#.dgíóê»Z=¢´ 5o—4,k;³^ʆ՘}R«{`ãèÜ©ï¹;°EÑ/©;Ô>â'z‰|cÞ{#\|?’e¯.ÊCŠ×C e½-H™úVÆ ìîaùox— ©“.N-Ë#­ûçÚ¼sHŸ}´Oˆðïh×ko® ³ÑjSt²ê.¾Ú€û·ª,Üt¨‘ÓºÂDä‘GUÑTóòzEÇÖœ_F¾m¥Øê‡(\©4Y=„B×ÈÞF_pÃÏU\GÂóPtgš=dßûMý›ë JÐS¬Jp¤â±R¹×fÜwPêî‚ÍÓÉ=(Ãyþ¼cǤݩ¨×_ž¸Uäu¼«Œjät¿¹Ø=P½oÙÕbÝ5òÄâ³ÒìQëUgá- j_7ЗÛþƒ:¿¸~ 7E]<ÑH.øõü¤nˈ ~ÐY»Ï_\Ðà“eP³˜v•G1t¢w°º'-?vìPÂó«/¿®ùŒ¢©ŸÁ©[oHÑÜPÇ_Y}-@óªÌûp"ßæÝ5ˆ|S»/•F 3j öW!kÐüžâ‚ ¤lŠy-_ q–~žÃLuž¶5,é1|øxœâA‰1„¡Ý_& w/Ξý1 A¿Þ×ZûÐ@ èÓÙuíMðwÈò«n¼*²YŸkê9d k37ƒ,M÷ˆ¯DÈìO´n¤ÂŽZðM±çD~8ïdA¨_òÁ8Ä ûOÖSAÚ,17“UvÏÉ!óðô9EÈ⦟Ѿ› 9Ž´#Á½&Gtjë;jP°ûé»+ü½qö’¾6gœ?|—ï-”–~xYå­í<Í^fPu@ÍÏ„šØ÷¿í÷XB€Èvø·>hèÙ䡇¦ÃáŠE,;Ð\$Ï{’&Zß =¾‘í_\w7Î/•„º÷-Ð#Ã"N’¿†ÎÄS~ÀG¤ÆKÝG0¤Ìn&—ë #œœuŽ[ð›÷²ñiæ W‰¹Ò˜O “…Ú/ ÚÃLjC¤ÂüÛè?{^±Á¢Ó I®#z°ütKˆ£€V*…?ñƒõÀ1æÎgðçՠȤT(l½åý]> “ÜêÒFï Zz+žHFI4/4ˆäcKWÈRÄÿûÐ>¤ÊTPe)Bêk‘ûs‹„V®—])Sé}v·¤°N!ãnÍ”ÞÎdò·ªìGCj¦‹9o‰|#_œ7Ÿ@5ƒ¤éDä6qÜ*³' ¯ÓK½¿ßißÊ­£- _l-Nõ-ð&)Q ÍV RßšW~€–?R(…%×Ó$G^QBöe÷$ã×NðÑæ4?ƒG$9ŽeÈ@ü¡¦lVˆ}ÍR•Ò~bJOóvSCô†ê¸ô"“¦wq|§‚ÐX•kJntzs«Öj ‚,9$V²ÏÀ‡ gôwçëB Í£ÊúÈò÷µoe.€‚¥ç5OåCq‰R>Õc+(›¨ 2KŠmÁ)C©¨:Vð’§ù Tߨ}÷\ø‘G1ÜäõP³Á·’>çE|¾ÏýŠsÖ‚zëj’n¨V¹çøƒÈ³•m÷‡*¡Jw®ß"ªbíX5æOCõÑüØÍV¨^‡~ÝÕwPóüÐÈÖ2BíË WÛ@øY%ÕyåiÔ?åY樅Æmù’_/@“nž"!Šš5ïþ«É -Š%1$bÐ*Iw„ m*Üޥ܅Ðî¤å>t:æî+X~µƒ®Ê‹¤lÉÐÓæÙÙôTúìEÖ%ÂÀ÷_rò̦0t«§,W¶F˜v·Ê4ÄÃhÓ"“«á>+/þVÔ• “Çw÷K¦€™ÝGh‡â¯Â¼è!‘rO&X¤.eªXfñmåt‡Õ ŠB7ça½³C*Vh 6ÃøsµÃ_Ÿˆv6^$©øÖöKÉ.NÈ\irEŠC3ÒÇžÓ#¥|©„÷n¤ å¼ú3‹©¹úþL#Me[ÇŠ‘Ò[Íe¼7GFÆ·?o0ÿFf’Âs’×Î"ë¡4‘ê#øÿÕ òOé›ÈÉ~ômõÌä.‰—¼ä¯¼ª!rFÈ—g¦nÁƒz1ïÝŽ˜¢HÓ‹«xðdÞC²Í*Ÿ1Ü}ÅBüz`G›’5ŽHÞÆcòôƒ†ü(uz>ÄíJ¯q_ÎñDÙ™Ó¿NïA¹®,Fz(ŸUüi+¶ÁôäY Ñë¨b»µ7öª{åK}ûªnª˜“Æo£z©Lç¼,jd3RÚ>”@­—ÏÌ6YQûZÄÉ[ gPÇÒÁ”¶u•/º8 £žª”\À(ê‡ð>à²GƒÏwÞ=NtEÞYÁP_4:PDU`>ŠÆ®OjVÇóë–Þläëhê?Øùz ÐüÜËzþ8¢ÎøýµÙ›‰|ëâKøLäÛ‰œ'h`úñ>*ž±hÈRê«"…imƒÊëswå(é /ˆ|¸!ÈçWÃ3sæîï 4X2r?G*ÿy­}þÏO2kŒÉÈ¢†€/Ó1î€?C“U–æðÞïÌVg AùƒL.†ÍÁvªå¼ -qÝü_ú×Ë!ùçY^oHÇŸΑÚáNùøÕ¤È¼Ž ƒ¿j sáò+¨ ëñ=mÉ•ȦTpüø¤²¹+Oþµ„¬SÝCŸèBæˆMuTþ#ÈÜãÄ'Æ.„X3|Á"„ Ÿ;‚ZÍæé– œý’è™Ò „úWKé-QyÆ®í˜<ñ¾=BMÚŽCNøUÃúádÈ»­ãa!õ e™FøÞ€wö câq(®szùƸJý¦\ÿBùbˆú—Ÿ¯ J~c|×ó}PóMÒöˆŸÔ4t.VÓ„†Ù.aæYh:–¶÷¾4—<^/¨VŸvñ‹+Ðþæ$ÛšCЙ¢[é£ = *O»Ã¯Òsv=€; ÖZ:α}C[ Ã,ïzÑÀèzÁ¯YgJý;ÁÒ “ÏÏ•[«ß†™[”ÃÓÍÔ0cᤅ†,ž<Ì“ø–õ[ãçÍû`Õ:×ñê#:X¿Ä˜_=? ÎSöMT„-Ó°É௣ðמ•÷Ð'+$±Uó*è@Ò)oéÁ_H^¯¡¾.㉔ôÖG"U¸•žþãI¤6L;â߈´"·æ²Óí‘Þ“uìS‚2’çÞ|dz<À‹,úAûÅyˆ|Û­øU½9Ÿ¥’#·Y2µcuòÞ›—òTEþKÛsì±Ó(è}Öὤ /|Wøýî=‘ov4c»QléÍ~Ñ9y”0|¶vÂè0Jž«r&±‹DiwªçlôwPv*®àjc<Ê•IÜ㊔/[]+¸PIiëRÌ YT¡öÔÜhèFÕ©ÖëzÓP}ÚGà°ÝkÔN³F­n§IŸS¡¨Ý\Õã/¬€:¦^;ñ¬¨»ñ*óç¨ê‹&wŠsÞEû„¼cÞUD¾ÍçPTË ‘UøAû½,xž^âÊØŽšLÓâÏÆ>4Ûdø+rè0Z¨›¿*µFË6®ð [ñ°X”¨XÙ YŸýBk !Ín´tpx ýƒYËü÷jòV(;ÄlÏ'–®þ‚¥¨;6—6!Ú=e#ÄÁ>t¹DEY[CÈvöò¶«–,/<(é‚À_•ƒ’ÔQÈOhðá†ä˜m­ÒÈ’élL³x Î^ÙÅÎï#­â3 ,<þXƒ.T|öW–:¥ •㳇=ÒU¡ZhäDÀÌi¨õDì¢-ԔǦ>³LgÁ…MF¨^¸QfêíÕâ†Èr¡*úÃÑ_ ê ]¢B°2T%Þeõ…jù ]žÕv¨ÙÇú¨×¨j¢¿‹Þ³á‚Úš–Áàç²$õ¾ƒPŸM;¼MäzãùQÏdN6hr¼ý£ ç(4Û©ÔÞ¤T–ÜqÅý_ ÕÊ 1¹ZÚ¾dØò]ƒöï»Rý\¡“?K"]ã tý©qTU|½»½g˜dC_ÌŽeÑEh‰t¿BC.⣻a„kÖ“·EF[î­ÉszÂXau~Çs˜äÛs÷“v9Lʦ%ËÃü>'í\ϰ0l}Àü„*,õ§úêµ…UÖ¯=$¤í°îþƒ÷yÀlµÑ0e…¿ÔéOV.ùöD‘‘þ#’QuÃØ*’W­GŠŸÅ]?×cÔRq=U_eŽ~VëI#‘æ¨ Ï¢ÒÓ|~¶˜"„ _šj¼Ä;‘)’Ÿà/Þ…,E”–TÆ|¸Ÿ-)¥‡~ 9¢xU7²ò‘[…šúQ°8ò’¼›JnÎE>GF–±ù2Øóx€Å¯ÛΆzež!úÄ#¼õ(z-ª˜–t ÅN÷ì mÖÀ£§b¶Žzâ±{ ×_¶£”àÖ7kJ7Øù¿BÙZošô~”ÛÉiÅÍåƒôÜÙŽ ÀÄ鼇¨ô†ÌP7ÊUN.Æ]¿†ªKN cQ½…fÖpl5>¯]U›C-/±Á¶q6Ô¾ú[àÙ(Ô±—¾Õm…ºª˜WŠ«P/—r9õCŸxêGÂã—B¼MhØ·ÿ…‹6ñÿ”Œø†ÆO}¸–ÆódÏ˺ìEÓ@û›éÊÐܨÌÊlǜȷ­"«¸½hivj´„Ì–È7±åÈr˜òháùy=2£G ˜7 ùæYÄ6þ½{Vt >„ ;&»nBg|7*UAÈì p­|ƒà+¾š—d²!°“þÁßž=pÊ´ètl?øe˜\Ù¬h¿ÚK=üÑdÜD—Où “7éØ7!Ž—·ä,ÝHve]Ô¢ð‡OSÀðn ‹È!úl_ÛrÈT ûÍ5×™-J½»$æ ë2où\üsÈZYŠÚó#²™üò¹™¯@–º´Èå/9Á'aw“ 2™âRgwÕ`½‹Ö9 g¸¿µÑ„Á¼‹Í¬Ë 7˜ bB€0­5Ë/™·—Nq8 Y6«êWå¾BNé·Ò]“ëšèY|­ -Êv=Z,„q©¹7ù¡xÒw4ÕgJw^1)YHC½ÙíŒùPeòY¡6ó綬LsLA¾ëÊïšLh$ñÐy> MÇóWϵëAsYYÒ—ýÐêmÛÛî´??“ÌÆ¸avÏk¡‡Âðý¥#øA¼*Ÿ Ìnü¦Y‡¡}úÏ®QÁpWSH½8Ñw2Ž”V^‡±‘ìR‡T˜<{‘,ðêu˜9¬8í}å6Ì«+?"PÌÁ"Ÿ© O, |,ß*€U‰„¢À{>°.&­ñtÏyøÃ]i&·¶øõ²5?÷Âß“ûZ]èÂÄP¬ðÏ3Q$m¢kþ#ÐäÙ'¶)w᮵{»±J#•§G‘å3RCšbâ5¤eݬIMFzW¡ÍL©dØ–¾x"t™ÞÐi‡Y\@=>®M¿t"ßö¾ô<ƒBE;9 ÷…^¾ä}`ÂÊNáŒü6÷ì¿{_E-J_ÏEáÅ0™›Óψ|óÈÛþ‰b‹n4Ò2S(ap²¿Ìc/J¼·ß±Bi×¼kW §PvŒ³A2å¾´~MB…“¢ï÷¨QI¡4£n£ e±¼V§-ªþŽýêëŠêcÔR;Ú'Q£Ñ•í¨3jµÎrM£v]@Ù¡Õ&Ô©ÍŸn7GÝ•xŠóìWQ_DFþùõX4¸Ûü#~‹‡È·Á¾#WêÑèÒý÷‹}—ð<#OßÕc÷ÑdN~aaÓÍvo~|G íøp!õ³hùË|( Ð Ù†ÏèìJÌ?±M¿¨) ÕÅýJ¦ÚH˜ë!ûD±–ƹ˜!ƮֳíD[Γ˜˜‚è½ô=á5ÍIAÊB1=øìë#<9ó[P°/Üó»ù¹†¨Ó Æà ÓÌw´¶6_ 2;y>K4‚¯›ûC¾zèC±Ä€G^= ”é<3r½ Ö¤ÁL&‘PÍ[?þŠª†ŠIÚ “È1Ç>÷|Z¨ñ°4jø 5ºËUí.ÝPÝ|÷Gæu¨ÞìòkõûPàðJMàTi½Ãé}ªRíÙH¬;¡ZÃaóf|'Ô˜e½žC5…úwfš6 ¶QÙ\¹Èêx%³ìþBýøßæVïHh¼¿¢Çý¸š‚ù”+ØÔ 9xïÛ¶2h x˜èqXZß9DMC[œ~»´ÏߣŽcý Fó‚[­o ûˆlÀs›‡Ð+÷–ç$/ô5úì›ç·¬~»ØC¯ŸÎ°xKÁˆp›4uû4Œ¶nt¯oÁXVSÎŒ¹ LRS“ ßö‚éÔ©úHsý.¿t)`¡€TôáµXJ)Õ}øŽ VšN˜^Ýý Ö%ÑÜ ÿ ÊÎù7†mŸ¢ƒ-ɆH"e|U[1I3k5o,íAòkäÎ{Û|p×EqaYþ<Ü54õCŽ ÷îÜñ¦ÒÅ}ÁgE4ßÑ S_j¦Ö2X0WŪF&yoÚ¥åËÈbt)PÃÕ ÙØô¨zˆú¡ßu~ˆ\Åö#þáLxà1¯¯ò‰7¹½ò¼‰ü%ÆGS×N¡àk§È_’(œÆQ”Ó9 j+o«„‹‹áQé‘­±'ùÑ7&©ý,XBõ:ŽÄúþ2ÔHë­aHG-öe½¶ýD¾Iì«_Z$òmœ®E}uÕÛÑßD½ ÀÙÖs¨^ö|eÄ ²¼ÝêÃ…Ñp@eò­ô.4dx—ò+݃ŠsõðüÑ2š“l5„As㊱^&´PjŸ¹/¦Bä[Æ€µ¥‘o|R­û`òu˜ŒóDvZÃ{…k )Æ+G±Tbêú%ëlˆ~°®ôè 74/ !ƒS_úø!¨1HýæÝ4Ò;«C¾ü[+’—ê Àïlÿ-5uAð»›A~CàR1õJ~2„å¯X‡ØîI§s!Y¼¿Ã~2>}®¨-R‡¹[¡L"dò¬~5!í…Ìî³ñcíuæ§ÒGÃ;5šZsIJ÷RwÝ‚,Õ7¶\zq9ÓŒ!“»Eúó"1ݨ ì‚¥"¹G´%Ƥnö<ìyC/@æ^ú3âwÛ!óÍýÜ$F9Èzq}Á1Rr¦Y$5!¯bÇK(‚ ß·LÏÑÃní}[z´P²û³Je1+”q»*÷…#\þÏLÆ¡êÞe_ß~¢þ﹦Äe@Ýõd‡}ú\иçëK =MŠÕQ²ÎÐ\¹çÖ}¶eh}Áí«uÚ]-OØš×Cçk«]Jn§ {V¬n³O~=·™fU õ†½9\˜ –ú¡Ãy¿sÓ|*`4å¦It2Œå|?ÆÞà“l×°êÅe˜žrZZ¯ƒy^ߊsXXk⦀¥•çñáB`åo M™ýnX[3j“æ+±“>°9Eéºß ÉÉ_ÖÙ·"É æÕÍÇHštw‡¿ÉÆÅHÛ½pWßhr§RÝI’°Õ¤>t¬‘G^iwùDD$ ý}ºÜâ dX?tâê£[Èäý»Ï#‚YÎ~ä«!"ßhZβ»"‡ðŽqŠä䶤ivÎCÞ‡?Ö¦#ÿåu6Å»(è³`ßņÂK÷Åç´œˆ| ¹¼ú[ÅÉÀ%%ô«ô2&PRÏälÓ;~”~t]+ûÆ ”òV¤ˆE¹/Îe **¨°.òqGà*xõ»2UHÿЇ’¡ê ý¾VTôQkGÎûŒ?›¡VÃm˜^BÔ®²úXRyu*)[»©¢Qwá›Ñ Žê GþÊiA»Êý¶ND¾µì6!»ŽFÖúœ2Óñ<3ñˆÍg4Y¼Ï&¬†æúŒôÑBAòGåI´ ýér ®]¼ ú2Õ³®ñž>)ÞÜ»ŸÅ[C‚x»@ÁñmˆÕLôk‰¢…èÁDëõã5-šì¯}D Îl…ŠJA„‡Öä4VΣðàdþînô*„€U!ñ—þ¸ÁÑòH$E[;é¤;Aæ£koMçøáë“!}ǘP4.qƒ~M—èÓ¬<_MxB³ÝôÒͨT³¼ç€ê è`J‡š‹ñÌ¢85ÂeßíþÆAuÖƒjƒ«PMNg°ûåq¨z©1’ZÚUjºR¿†ªt¶ ¼Pm¨›”kÐ 5Rå„f$ú¿‘Ì«JíP»4c+0!ugŽÜ5»Î ܳÅiÅTд‹Ó¯÷ 4}m+m44…æïÍÊÏ<¡åGKŠÚýzh-ìfòƒ¶–©í2‘èà¹E8ÞâìÚ—" ûRlN¥·%ôÞò+¶¥ö¾à¿ÓÞ0°ÍÇ7Ä Cͤ꒖0"Q0¼Úí £í®ßs?:ÁXjW&y„2L¬ìóz£ Ó3òŸG`.‘AàØ®Xð‹– Ö‡%Ç^šhÍ XñÉéøIk-oÿœqèƒ?¦c–;,°MÝÔH˜>;ý÷¨{‘ôB¡þ4y’ï¢ àZ…sÖ|b ¾¸[àTú ÜãjCÉLG‚Ô;{õ¯ mýEݧÊ÷ì|E‘e2vÎÝ>÷u™' k½â#‘MÒkU¦µÙÓHÓMQ¹”:¿}ŒÜ@žå¿7hÑwüý”ø~zä,ñTÉü ª†Åß¡"êȜ㪩8Š4^÷X³ÆÃŸSÅ_Òã‘ßçCDsðØAŒ¼¿¬„’]TlCª~(ýêÆ‹ãºÁ(ëu«>ãák<ÑyHZº1åÍmž±AØ«¡jLJJ÷EI»B¢Q…9-§Åê-ªwågjØ¢zåAK­ÔH*9Ùg‹Zî%D¾]q^% WCK±S' WQWc—é¸å>Ô ¡þ“ÍŽú‚ÞXGª¢AN°óC–Çh8tGçeÀ4>iÈÇs =bmø’ðüöðÒ½ÛGÑ4äýÛù•e47‘wÊ«E e‡OŽ3ZšsôLùÆâ¸§Ÿ&/9xó=‘†WO„Cb§àÕ‡óf“üá,ÛÝïËcÌ=ça>Œ”?ä!8òÑá´!’WqL󄀿÷‚á‡w¿”oÄFøÎ\½ûºôø±,ÇÕ‰Aà{•äí«’>ý¤º bf_k\x I}É…a ðI…ÏewT:dÌ'žù½õ Þú3BfÐF±ZäwÈ’yPpƒÌ²Ú.­”)ÛB6icÄ ’=d)ÿÎj™ÐÌÅð¿´µj)BöݾŸ öï¹U‰\» %)%ïÝG#[@øPŸ0ëË ™›×oÿp€ÌOJï:Y©=ÜÉ~@.óíÇPÈ[6yáÝx KG-ŽçØ@Õ½ÆW(ü5²ª eò÷C'9£¡B»ï뵨ýPåÇþÛ„F jiØv]»uO›gu<‡FÆ…rC—RhRíLŸŠ…æê³G¥+¨ Õ3ï²*¤@û=ûªòøètçcxTÝ”ÔA7á×Õ[©6U¢0pLÆÓÿC= vx§½`óƒáC¦ßµ,aÔ½ØQÑyƼ*k$ÅÎÂÄÀ]—ÍdK˜NKÛ.y}æÖJW3vÞÀB½«0ÂRåEA{¸+?=Í–n{ÀZÉ뉓ⰑñÉçÅ_aØÌ*ÔH ­†íÚ?Š<­tHµ-O[¤Üt|­Hî~ìeîª0>ÃÈ«ƒTÆ5Ñ.éך%Ï:ˆ¶i–9éãö#½š@ Ã/c€Ù82y}·¿ÝÝ„,:Ç…Cìˆ|càÙOnŠ"ŠŸÎi…"÷%µãîM_‘÷ÑÁ©Ó#Èõž"ë^ ¾QYpi{ˆÂËÂŽ‰Wˆ|ûD?7Z‹bóWÖ4KuPB75£·¯ %ÏòÊ}åEéûÅÈYÆP¶zM„'åÑÙÃ}±¨0¿ŠÍã¨$m êr§ •·åLL¢j/Z²Ò¥£^¨{ªSêG]h8¿‚_øÞ5 áHpƳg4±a ±TEãçé”îk¹fz¹ïñ/È””Ô”ªÍBŸ³Q|ž?~;P>„ò[¯&¿!ïpBáË¢oŒÿHK ™õ®ÁEëÇ!«þ¬ÖÛk*«PFŸF2ù|„—³Ÿ puu+k. Tp^pY‚å‰m²Ãw¡Ìôö/øÓ7¢ÏŒjýjƒ<ÏÉ×P{0ï`ŽÔmßK"ߘ¯ÎÈ ‡¨ ùŽUrÛ<ê.^9³NnLä[qáø¨5ŠÍ™LîÙF‰³Ö¡7ö%¡¤ö¸`lzJ;VÑê—\AÙ.Oj¦Q(—V¿þ˜&ªž»¹w ’ûmŽ6:T^È(·(Šª4„ó¥»P½Ýçê£>Ô¨Sz&Ô*³ñ?5ŠÚßÖ]¯û¢Î·lþì;‡Pwbn}%§õº¯¼Õ“E[Ó¶C^"ßrLÓäÃÑè2UÝCóSxžmRc ŠMÖË©.=NFsš÷JÔø’¤5ß„ÐñæÐYç ˆÝSh¬”’IJ'½¦#òàc(Åï§=!ãƒü¨Â+m ößsœÈ-³ÌDÏ}¯!‹²>cÈJ²²Ï<)€¬yZ·„Û.¡¨Ê ™[¬µéÄÜ”©p©i%Ï3îœbÔ>@ k}gò‰È5†aóÐü2á•[2dÚ˜èM]H…Ì9R*)ÈZ(Z ¯-‡Ü+œO‡¨ä _wï+ƒ+ðíeöòÃ=P„¬Wj“®C‰ån³V‘PöÀ†×¿{T¼°¡’<üªj=ß\›†Z3æ! ƒºôƒ¿]FL¡QÈ Ÿ8FC“îÚµ'ÆKÐ\ב›É/ ­n¤SI¢|Ð~%hÂIV:­ÖM²ì¾BwZžWMA,üy,ü·C¨HØÕÖa0¶¸U¼Š† QÃeïaTò.é±k0&Ò½e=ºÏÖ.Môífæ<Ì]¡XÛYQ„…ÇrT‡v¾ÀÒ-:A.·c°rÝzE߯ÖL¾>1¼Eªþ…íÃw`Sý`Ï>/°m1–Pâ;uášá²#Hzí6í:$שÛE¯I»û¥º†T¢ ÿ‡®óŽÇêÿømÏì"’$#Ù¤ëB”½²Ê¨”´£¤Rš’”‘…–Ò’ì-{ï=²7‘Uvüî¾|îüóóxÜëºÏ9ïó<ç}Îë¼Þ×í~¼ï›ÏÜÒ~$ÿÞÝDq\W½sãlÈ=d9l¯—wzY‡ Æùºýæ™ ´ÒBN]iyý1¢¿qÇ}ndEÞm©>]fC~‡H³Î[ßPðÚÌ+ƒGƸéÄ9göwr¸Ù¯o<_C·LÓs_;¤Mô·†ã]½s(ñS¯ËÖ8 ¥ xoýVñDhæpåœ.¿:Û‹ u‹K¦a¨üîgà:îèÚìœ-]€jâQ)ªZ¨>•@y>wW·R<»ì…ZÕ — SÆqoá«ÚäƒÔùֱ놫"ê%çóVÞ‚)ôů_:¡Q?î¡4Ùt¬¨?ÌMÏx¶Ï¿Dô·ˆÛuMòDëU˜ß€–¹ô jD¦tÆhžC­‚X¨GúÔ©÷Æxí?õ{ßÎ_÷ºKÔMªs¾l4¦ÊøõÝ„æM=ÓŠ´%в0Àã³Àmd¯>ÑU@ÇÃou§¡kçÑa…¯uÐ]^´«N²z ÏM%õ]‡¾ït'Æ©ZaÀo"b]3 űjÓ=„áÃ¥OÒìáç\ù¯FŠ30Áãn2*2“5[´¹]áWš”éÒFK˜®Ø“5™s<_nkG›ÃÂCêRýd¢¿1_:{jñ3ø=¯@'vH“g¸ÞBŠp26© Äq¢óï·ZŠHÓÖÏ“ Ìô P4F…Œ-m»ú·Ê#s¯úííæÈ³Éã“/²ª8«Ék…œ-õƒó!¸^µ¶|Ðÿòf®Óžù‰üöe»?1¨ àúJqŽZ2z—y–p_…O0Æx£ÈÕkfIš(:ÜÙ?ð ·tvï þ†’ vÇɉãÅXM Æ@[”qäÙ(A‚rüIâtCQaãäÍï™PéRã>öbÜIÁû*Tå7îJ3SºEÖ†jûzh>ø>Fõ™²–³"Ü]æ¼Þ~¨µ ¼R¿uáÞóÆ•š‘¨ãvTïó¡Þ˘ÀGªD«uwšš@#ýöl3Ý 4=sŒ®ÖMÂÃ?”Z¡iZö˸opÄõRâšKćyvN¢Åý|ýõ×¢ÅqÛÚºc¸? µÌ)?¢µõèa>ÃI´ÕÄ×e&‚xÐÆª¤§f Ñß&¥?yÒCÿ×Dÿí£7!jáé/ÚÖõðöê·m—òáe·¼wâx¦y‰LkË(:Å,@à»;ù]¼ðäÚ'e?›]ðX×èýÝ7ËÀä¦|ü‹nƒsðˆªªøq9ž8Þ\¼kÏ®eG˜„Áë<Ọõ!b‚â4çíˆÔt IÛ5Ñä&þ’™âÞÞ»ÔJñâ$Ïß»}âT'=æãt!ŽP'ãÞÆ ±{Y'½]€˜E¯Î8'â¸õs*ï®ÎS¹1üÓ´ÄŒô½³º|boñþÙöyÄ1{\ni—ƒxNÙ×¥÷B¢Ïù1foH¹väAâ™]aË}¨¦Ù²‹V©n‡œk›“TK!/pÿ‘Á (x'²Û9Š•{?êC©ÍBèË‚"(Ï7µs2ª¤‚®=c‚j êÜ¢$>¨©ÚàæTµêÜ#¾î°…ÛWJo¬‡&Ó›.'å¡%ðQž•+´ÑÝîÓ·ûýå½É;x ËËvð– =ôÈÿÖ˜¾-}ô…Ú¿¨ÞCÿ\oàU<ÚùP.Á~Œ˜pï11¾w—ß;q„I ™ä¥)˜RŸb‚ß ¦7ífaF¸•‘öΘ[G]ֵ݂½Í¸üþli?À6ÿ –ƒN¥kæz!™~ô»@ RÈñÎr®ˆT×>ÈOowAZ6¹ñ÷}HߤÏóç~#®Ë,™>ô Y8,Z؉¬½d¯ùm‘Ý]kÛÖ rj7ÉS¤Ý úïÒŬQäŸ1÷3D~ÇÎ#~ÍY(xýàí–ø»¸éäȶó®¸ù‘¢Ó‰ËωþÖÏBÈP!úÛŽ–^”…Ç ;”Ö«9yë¸Êì½4¡BåΈ֌ñÙ£B•‹ÌÐéPT~ÅyW(. U¿·#ëH@5Ñ÷–áWQýçÄt …1Q‡ŸÌ¿KÑ£Vé¯M/poå‰CŸºQ'åÀ@´¾7êÅùm¢KGƒøÃN¾Õ hÔ½)”^@M§$`øšžŽëíW,'ú[Ðç%— 4wÈ|u¢-ù •Ú$ÑjYðôçj&´æª<Ìó9 mO ?«ÄCúÂæÓ0*—Ï)s‘¨·ž—ø6û´¾y"ÒðF$ç}qf,¼|àQNa> a¦Â‰Ùt!ô0ØœV„㛘OÄXÀ3òNHSu‡ÀMG÷ž:E%_n3y@@} 5óæÛ¼°uúúÅßð.L’ËüÎcˆ¡¸K—ÄfɾtrV¼åéݱأ,9ÍÛŸ@^»˜æ…x(ã_…Y†¹êÓ (Ÿñî ²Œõf=…:—ÅŒ“\7 žýëç…J¾hpÔ×ªé  V÷”?{ŒÆa=»eéÑämûgλôhšQiÿDÔ×®æóÜh.ÕvhI¨-Tq:@+*ŸÒ¿‚û_T4ºÚõ£µíÖÂâŸÐV‹§¢«Å’èo9ùR½Dë?×ÉN}ÉîmgÝàKjÀ‘ Ë©"îðv^|Mdß÷g ;l\CæA'~Z¦ “?K*ë%àñÀW…à (ôs¯-ùBäÔß?ÇÛw£ÒsÀÿ3¯Ä¾M{à ¥Óœˆ’<£s²ó†××½•§»!‚è¶p¼> Ïð€hj÷€õLã˜ÝqÆbN=Èu³†Ø&‘댈{áÊšM×®ôD°ÄíÔ+‰ÕûqTžøë2㬾+^@,ƒ„JÝUˆùÖ|pä:Ñ×ú%3S~A,Æ»-NÒû¢í㉹nˆ“¦r¬–…x¥ƒÄ'%˜PbgÌÕ)Ó}x_ß… ÿ­Œ1*_!ëbâüÇE3Èy¦Á±¿dòâŒÞï ¾ 9æ‹I PL9¦ì” ¥—Â6ÛùByëµö I¨R¬HùxÒªm¸Ùÿ~¬æ0l¿nuWõ.0°?ƒótúM…ÃФIëo~4Znœ"[6ÒÖaïðgý§¡#ãÙ]M»¡ëÌÜbÝå÷ÐéÔO‡9ÐÛ#à/Ý>ý%?¥¿| C”ÎED]Fz Ê5ÇÂÏ­9ÃáQ˜Øºu~ûøv˜bÊ.Ìß”ê~_¸ÀôÏ?Ï2Bš`¶áÊO‘¢8˜ox|öì™·°8VÏtÔf–«nÿi,‰d2Š÷Öó"ïñD©ÏœŽñ"¯× x¾|„ ÄNçŸ!6%« „M@ G"<ÂOAtŠ@;L l« $v’ľÙÎE Hß!vÏ_Ö–@+%þ:†â;A™måùBC|¾Pþ}¶P®3Ýòêùþ}æð¯lC #¬úëy—Ãöý•‚ø$¯Cö´¨Ïm¾´PgùûrnTÉÎÜ¥'V–ÌTgbwBö’WH3gfygw…šr@fü .±‹ðmŸûÆ7wí!£­Æ±$³ˆÿ›q3+?XïÕKÈÕo*+Ë0†‚ÌÇó¹›³¡äöô™mnPa±î [ب:ÛD©æßÕª£^S± :6vÿ>ÉQ¨ ”w5•€ꬻ\䤸ß̹՗*KÜU- šœáIBAúß_n~Â$@Ú)~LQÔqü7óîj»š›åQTü¯¡¶[,‘TêCFdü„¡1J ó÷” 4/X¾1•!‡Ö×Ü=ûý¡ã¨ÛÅ9þXèRƒÔcº‚Ðý£Å¼TlzsE<˦¡¿‘lª“5 †¸Þ—|?n70ñ·†Ÿ|!üÜÖñ0~iÀ­ž³ &–ó¨S¿ÀÔ :òóÕ[á÷î esg˜áò¼“æsüæE)OCaÁu=Í%6Xâ ¯¸u ×^§?àÞƒäL"Ž["ÿ“Úí¶§Hõæû£‘›ZHó¼:>l›Ò}ùðÙ8»†Ý² *sÈtxácd ¢5Ò5Q@6¯+Z¾Èž«¬ñä rzO;¥En÷@ëÉ`}ä g^2‹ÓG¾*¶ô'(°³(X݈7ŸÖwüŒBm 7›òfPøñÅû&óa(⯮í¢EµÅâ¾çq›Ü2µ”(QÖ]7ûö7n÷´¬˜gkÅ[Bí㹿¡Ìcçœ?RÍ(Çgiãc͆òU!wö,½BÅüê±úM†¨Bn}ù”Ä5T½æZÆ6|QjÖœN¡ÕéµÎlyz 5—ç—Œg æ®2²¥»º¸‡Ñ0e‰£ôßL Ã\®,^†¨×Ñcu`R ê8/Ê'Ž¢QeÕ=U$Cãän¿Î>A“_S{w4Ûà>ÁÃWT­ÐLH3.`bÍ5â>òp• Å™;’é\Ñ2ü'»íO´êO¾ªx (ÜgqAëÐÏ1ôñhË$¹iÈú&Ú=ìÜ>êC¬´ü·'òã¡Æñú1nF<¹œ5Þ ™O±zº·A—gÑËÒ Òÿé*ýZó®çw UúÞíšu¢¤™l×}2ÒŽ'ÍtMš9÷—ù7ÿfÎýoFÜÕåÿͼûß̸ÿÍÈýß~S´Ë76ÌBjÄöj‡íµ¤sÓ'Ý;‚¿úüÓóªÎÿÓu&ÔºVûW’žYF?^HüÙ„³­;¯-CvôÔÅÀËc¤™¸ssî;¼†|»Ëª7µ)¡ÐI¸‡{âÏ<¹^E÷oÆÝIIFá!yB¥¡v‡›…À˜iF솓<û•] ¡éšy¼zH´ °82[¦ü› û¿w=^ÉmK°†ÞSfÅŸ¿C¿©ÌÛž9.dœxIws7 ¥0M3ŒÔÁV¹—RÿfÄ&{—éé=·×íL—‡©íŒëgŽÃoÚÄsÁ.0Ã?õ`8ó Ìí ÚKX }'—›L{aéÖËó£®#!þès 63$·q9}ˆ”FÞò5¿úUNü Á3ïÞ5ÛèÞs™nkÉL!êrzj±º…Ùè…µv GËLRífäæ”¹òÅ?y‚ÂõÏKá†Ël—Ýqãƒg»¼PÈ·‘fyž …]­³Ûî H ÈÑù7¼(ú›¼ø¶·n{¼GÐ’gJžÒçÍžÞ†Ò*e•K!¸c*G3ÊVÚ\!ïCùêw<çï_BÅy]·ƒoœPÅØía¼ëSTm)83/ýñÍüëºØ~TßïK9~wSÙ}ðß&‹šU=VÕpO"û3-»£¨Ã"~¦.u‹äκõF²¨Hϸ,4²|+AÁÚ€Æ]—%,.w¢©À–ÎŽzzÜç°sïZ4K Oàž—G ^SUùD´¼ÑEVØXóOoz·oHü<ŠÖ1¹oN¡í† Žª—N¡Ý£öÉ¿fþ_½±ûÒÉ,˜BíòËàw¯BåÑ‘œœ¯‹ÿŽg‹¸½.d¿Û^Ãî¬ Ùb/{5 wC–æÖ­S[' S–%çšíuÈ(Ðkî`®‡ ¡Ñ<_¢.îè[ö…CFêÚôÞ¥tì#¸ '3|{õX ì^nÚ̵ Jd.T|¨~å#ÜO”d¡JÏ#Mp̪wl)ؼåï/›ôe/?‘{uK:-Ps̸ëéÙ¨¹H?VPÃ5âgŽ1%ú`\‚çu¨fOø`ÂUn/²%¸•¡JëõK“c¨Òˆæ0ZÚö×÷~ÐBÕ÷üxê¢N¨vC •¾þVu=`¸¶êO|×ôõë…F«x ï;žÐ|÷ö­Ë'í •+tŒ5Ü Ú›§çŸ¾]„¶íLvmИŸÚZ¥½OùKg7sCÿ—˜cL¬û`ðGóžK²BðÜiÜ—@£C‡öŠ?9ã¶ú•^RÁÄ ï}™:˜ò›ß0¿÷ Ÿøþf$hƼƒBaN=Rð|àmXxi³«U–v²^)ÓåGÂ]©p AM$gd}z‰y )Rdé>üHåku«$] i<Š7ñΚ!Ýó߆ëz¡îYÝ"G&Ë«E^¾ýÈâwGËü†²q>8š‡ì o3ßÑÝBÎË¥ñfÉ&È}ì\qX®9òÜÓ>šq|ù ½ªì¦.£€Èv _C´{zášwÝ­õh-ÑçÄntî‚LÝà!«&ÈâÕØ§øõïxvnw17d;©ÌʯÛ9æ _C¾|†\Wú†Š6ȧ–âBjùyך|(vRè.ÊI„2cÊ@ƒÊkP)_6å|?ªÅ÷&õ›’AMZ­€ÝQ¨Û»Ü!Ëõ1¯=„Æ~A=Ù´»Ð²÷›¢2´©i ;ÛòC§¥UWéˆ"toßè(¤“ ½ûÌ?]r¾Y…\¦¢m0û³|Û“Ë0tꬩãFø!pv×!án>ñŒ_á Œ…«ñšßø{¤K‡~‚É>~þ€{ðëKå™÷0mÏÞú¢ f§Ä£|w½‡…#ã"Ÿ±ÃÙÎ纎/‘pñφ´¾H¾¡M„®É)©ÇÄ“#7!õæò­åë ‘ö\S°h*ÒÏuü¸br×µìü7Š,Oo‹o—­F¶sßÛ>8û"‡sаöM}äúõCO3 y¶Êÿ^ß|K&wvjlïd•Z¾F!õ-Æa1 ÷u켂"Ò›ÕEŠÞuâ"OŽÆm9ÖÿÔ‚³ßÔû”‚PzÇ%ë wÜ‘eHnÒŽ²oè¨ ‹Pþõô¹Ü”zTÌý¾SÄ– Uv4Ü ñ¸ŒªMÛ&Õö ¦Í-+æ•¢úùr6ßlܽí˜ßò j‘qkuiâžÉvCÉmy¨=+sòƒ“;êqòFÜêCýøæý%÷Ñ(èÐÏ7âh¼E#@ƒ¦²V¶Ó×iqßÅ»4ï¢Þ£Y~ÝØ¬þ^´Œ<á}.-o³]ª5dù§7}”÷ùÕÐ:êÇï‚0/´åIT–=Iô·u{¹òÜÿ_½±:’6y<jßë${¾ÿ‡h_Õ£…üíçØ@¶] Ås£Èj¶Ü×h™ :/‡Á·˜7g+7zA†2ï¾ç:î`2ðÍf?¤ÓÈI÷1\‡ôÏ7ë>¿‚L‹“ë³ !GæL@ã&È/:•p—&ŠÓÓ;\®ƒòÛBž_ªAÕÖçÉ ¦¡ZäÅ÷¨n 9/]­ 5¶n.žê÷¡&;„ü}a=Ôt0o¤õÏ$‘ä!b}ôròçû¿í6ý~ªŽÿú AfPµá”Êö¢ÿýÂÐÁS×þÖq/ªÞoùûË`ÑnP}çiÜwe—þö­ö\RêQ¨W,qö:eÙG ™ùË¡oRáûC¡¦Qch7ºp½1Í:³¯†'^xÝ·b6þt½ ½Çž?qU ú=5©²ŒÃ`6}Ö ­Pø!º·.ù\Œ&‰8c9Æ5b>‡n•‰*gF«k 0uë‡âõCêð[§-øàá˜Q÷˜~t¡掑_â ŪQ.ꆰdsc£[|Òîω©#9]gMRDšçL«¾Cª+õÎf¹‘æl–_p7ÒyilÔÛÿ*È ¢ ‘I7åÒved¹·´óЗd£—W¼25ìh®¿EÎc¯f´ÎÐ ·‘®¢öσÈsêfé'ÈÉèKfšLow—  ­[$Å1ªð–ù¹Þ…¯ ~ÎÑ[F‘S/Ÿmzˆ¢¾O-6 Xç˜Ðlõu”8¬W•ÃÏÛÄû]èQúÇ#YïÊ(sô:k½¶*ÊQ~ÈT=ߌò~Áѱƒ¨Ø)Õá.Œ*Ü2lëµPõü&žZ7D¶S£h†jS=¬óB?PãòbiÚiÔ½8¡“{Èo?²2Ľ ÏŠÓ]ÊQ—ìuËÌiCÔ§qsþuô RÈ€“ýR=%ðÞžèoƒ9Ô¥Ðdò¤m¥]î#ÿÁ”Æ®…fTK>Ûæ»:R›oS Å‰Ç?ðk=Z¾Š=_¼ƒ¨7ÙC[8á¥)nÂ4Ñßho8»ýÍ#d3ÑßvùUç%êíüúÍ™ökô–¹¯þ2<Ô`ÅKG¶¼Š´.€Â?§$öåÇ¢/Ö ] NÍdŽXW©(è<‰±GxÒ,(ˆu[Vö Ê_ĺ­Î=+_ bÉ¥Éõ!F*öXå'=¸!ó«l#Äp‹«cÜ1iMµ¡<·!‡²]UùoÝ7ö(`Ed¤ãܳ!)ìóæz.HX¾íH­Wä¬AÔ/Äom?Ho™u<»(Æo.zÙß$ÖsuÕ›v}î†×7|ëŒM¿7û2ËèÊÆêˆuh.—>§do¿íV÷Zr˜y-­û Wìá!ç/› ïË¿‘·„j³ÖÇM@Ñ`óžì;)PúY^Dwô'T\9Ý’9E‡f:úÒÔ@ÖT*yÔÆq -î;õæ:¥w=J Ñh¸[äûehþ0”!Ë• ­!ZSíήÐIÖÁ(-]Í/žs·óBO†IבÝLÐ÷B¥K9h XýXÐû C” ýÁ]'a¸¨ÔºàÃS)Ï$‹®‡±Ýr aC—`‚Þè·š» L†lÚ,¿Lãžù€iª ÷%˜½t>¿þª:Ìw :ÞìЀ?7³ß»d"A1ð¥‹Z1’µiµ½Š>‰ßÌ„üï!UÉß¾˜«H³¬ÔEOM‹ôgO%¹dšàº3š÷ç÷ ‹Ëþ‡>öxŒdØ0rкùÖÎ^D.5q›ÁÙ`\Ÿ\ØçÖõù<ü»—ÊQàiÌ&ÑòTü5v›§Å7姘ÜU‹›K…~~²GQÎ6ûÛä(ö©kÞWÕ%<¥OÐ^¥Äí_[wWªfáŽÇÏ=¬ôPÖ%áqiú1”?¿·€cX}N-üàŽGvçPµ>øÏ°ÑOĺ9›ç ѨþQìÐî6:üɯ>µdÈÞïŸÆ½Û?RpÌlF4ûØÃh‡zö„v¯Æ 4J£–è2A£FþÈ›hÂEÓ5òY Mµï_'JƒûîVÔàq74«¦ÞÖôb?ZˆR»6}|ƒ–»GÛó ÿém¯ðý>}~´Žô®aBÛõãTòc'ÐÎ_u³ËÊÿWoì2ßv/Ö@­áTÓqi(·ªí焼jµ§±9'!›X4 I.A–D‰ýÀ!È<3tNý‹%|“–ãêzDô±!C«Íéö‘õ}ò¶×Æ×Çy~Ç~H ³ù’ßT©‹&îØë@Åß­fc¯ßÐ7;Å{©&‰=åëÅ/{ @eûÖ®HO¨æy¦VzĪb–^¾éšÀ3LÊŠP»¡O8E0juØ uu ¦›bÍtj¶^õþRXÕŠô'ôb ÊRÏ„ûTöŸø4øé&TÖeo‹yU&ÇMŸzxBU _äOŸP}ßm«·Å?û²UßãÕÔS¥$íyt ¾t¨Ç½5…¦gM˜÷¾“½§aT”†¶¼qнôÒÐy›w½—të‡Þ%l(^•Ïë[jöC¿5‹îI–‡0øTUåž ÏÝLîîЇQá3ªßÊ•»s&’ ý®Û ýÑ”R´~ë×– 4ÁŒ%ÙŸ`c˜ó¶ªì}l “å„O§aÉ}$_ôû $<ÕvŠ–A$§y~ùi R¼üì±}D©Lß“# UÒ®'YHwÑ·Y;‚2 8èÖq!Ó®™-—$"ËÍ+;Nµ "Ži^OÙŸ_d:1U€œf®* BéÈ­´ùí/žãÈc”²Ã·NùüýõÂcY‘fq›c ¸Ýú^­‰BÕfs˜Qø‚«ê¼ËWÙoÀ±ŽšEKÿ4ÇÓ¡X’r¬YÛ”¿iÑ,†Re-6o_£tCÈïqW”1 TV7û‰²s&N—2Pþ;]¸ƒ*ŽÚòÛæ‹*Ba%}·¡ê陬¸WDʾ܃»¡Ú`Lï‘Ϩqê‰aÜ'7ÔÜ ;8f0„ZBÝé­üqïòæƒ_UQw£ÏÉv¢¿ñi-eóû¢!oþÏœ²4¦P–º7ZLô·ïdÄö&"CBWp¡ÊÁz)Íx’_ ù΋¹)I¯ÐÂ1T—ŒŽ -_Z0P]¯A«^§š° ·‰þæÍÆ8§ˆÖ!‰ýoëʉþ&mÀ݆vF0ìQoGXj4)×èíEqŒA4®köº®¥¤•1'•‡ìȡЕiê‚Od ¶\Ó*w…´±7Å׈ºI®|ìOuØy.8>ë6½ù¢(ˆ 9Ì:OÿblsÔ›J :°ÐQÙÇ¢%Û>~+ Ñ_ôJûÿÖsô‘2f[râYE_/Ä;õj]¼ ÓÂæ{êÍ ÉíÀ —=œLÇ|nÙRjŽ¿Ü$Ï©—„¾ŠIRBÚ#› Y!}ÿGuÞ“^)/xÕ„2Ž‹‹êuø}®þÑŸ~øæZæ—$ª™öœÊºe4*kvxdý,þã"ï ÙE×ké!§m][ãGÈ3ÿä“$¾ dˆÉòCÑ᫟R5Ó¡”Ú!A›ñ0”g0æ0Ù¶ulÎ-Ÿñª«¤Oó«VC­œ[Ï…D¨+é¦üìC ) W¸NA3]ÃqÍiø>½;OïntX•D¤ÒB—9Óù+ÐsÌøÌ¦Ч ËGCÿÒØéô.ü4i×aöºIJêÂÈeŽ“šÎ.ð³G̸߯p¶4¶ „IY‘a‰05+õ ê#ü®Gá,[I˜YHiÛÜ ó‡DùÌEÎÃÂ…!Ê[°â¸pÏG{ûßPíg¦¬ðÏßBÜÅLîA]í‡ñ7¢Ð`å’¤-— M›)~°_€–Kìt·X Måô½ÄýTЩ‘ìCÆÝÜ÷†YsÙ —q"S_Ì úwd¦³ˆÁà©3ñ~'ýa¸4%lá$Œš~Í?ú0¾îjë˜x-ÀöµaÊ¡üôý±ðÛ¨xFåÌ<Ìœ¾AùRBæ¢¾Š†ÈÂâ[Ó.oXzyàdý[i$¼¨·tÜ©ŠäTvCVâ1™Ö‰8¤2®4q87„4_O>±ŸF:ûïµ&×#CÌý–Ãû6"Ó¹ú³×:åò/ÑKôg‘õwk¿Û›cÈþ°M¾ìørjè¹oi8Ü‚Ël/5Ï#ìDÐÌý6ä»ÈA·‡Õù»ì£÷èó¡ æÈ}†—B(”¹¾âT /­ ;„"{æ}ç7¢¨YñÑo§v¢˜‡FôP+ŠÿŽ/{¨‚R^xÕÔ+¡t´'åa¥x”‘÷>ôåð+”%VcÞ¯Q¾ýN¯ãvTœðýRã,ªHÐFòP2£ªc¹ã…óÓÕ{+P­ífbù—S¨a'^ýyð(j²nì|Ò/€Z³‰ÇSxªQ›üëò¡O¨Ë#‰íÆDÛÎúÀ§! e,¸÷ŠÝ@cF3Ñ›÷ºˆþVÉ@Í^ˆ&c].§{%Ñt)!m‘8~5ã–ëÊm9‰æÊ eÏFUÑÂáí‡{9´ ›‹µsA«ž(-ŽjV¢¿ ‘O\&ú›ƒA“<9Ú2¶²—X¼'úÛó±Ä­‡‰z3£»Q±Fo×/íÉ?ëöEo‘± J/Ù‹Á?cS_š&CæoübP3¤eØ Ð9Aòþ‘¡ì­ póNüNSbÝÖ(;EÔUÌðh„¬1/D§°;Ý _ç’+÷ng„¯%ï§Ç·¶Côî3ÖÙùþÖqêŠü!ö2«T¦ŸÄ+r7ÍYñBB!Ç•‰~H²äø0Æ~ ’½æÜ:×qBʃªžÄ‡’Ä:ެ¬‹ûqœÚJ?s%ÒšÙ•^AºAaéY|é“·Üýõ¹Æ0EøÆØoßoÞ ßfUmï¸gA'·±Í[ÈŠb½É^Ù·zÂÏ&[CŽ¿Ÿõ®‘ûÇaI Ì ò‹N«=vÕ¦8ªL(9üTKk[-”"?ÚNëÛ$ŸÔþ~¨>pF`C$ Ô$´SxBujn»ª÷BƒðMz>;h:™ËnQŸ ßMÔŸ$?ßí_×k€.©0uô0ÎÖµ¨Cïðn~?ù£Ðÿ¶ßÒ§š E0Ì7øqÌŒl:QÓÄ4?ïðHepÁx@@Ê! ‡I51¿p˜ νwúüvüò9VÕfn~™v…¹v‡J±S¯añ& Ëa?2X¾è£SÒî†dv܆‡ø‘BR`šâg¤â¶æêi@«Ù£9o‘ÎòŽÑ©Adèúh!è9‡LÃûOûÞZFVåÛVŠÓo‘]þÃð qÜyj³ÚæCg»yäºçù%ä}Äüænßä¿»±·3e­ieSq'×/êY/ÜL&ØÔû ·($-ÜÙý·æ¸×åŒFñ×I†Ëí(õ]ctéš.î=ÿÜËAey™¢ž…£<Ûðù$ÝQTôqw}8‡ÊõĨ¨Zuʧå Å5†h÷]¨þÍ{'Õ9Üýìh¤ß-jÔzÜpJžÅ÷~„Ó½~½¨Sþl«•æFÔ[(7|ZT‰†’…k™zÐX\/D]8Mtƒ‚n·]CÓó#õ*W(p_„½ãùV!48þê™Æ5´M}t‘8µôŠhþ]òOoyOÂ>Ö£õû‚‡²;7 -çéð펎hçû¬CZ–õÿÕ[ã3Öè«PóB{½¸Ú<”Zõ7¿Û I¼Ó¨lò‚,ï¼`©á<ȳ>LÎ4ßöÝ~F[’– ÞÇyž9ÕÊáD_{$ú¹PR´ö´?tg‚”›lˆé‚t5t¤"î§,­?•’8~ûjðþ{Ysž;{Ê4ѦvÓq¢“Ô¬Õþû=‘4k©_PÃ@Ãí¾Gj©L÷ l}µµW¾×@?_dyÎ~¨ ~ŸI<î݇„݉:°›¡t3ûûý˜‹®Üo¡RÝŸ[2úTnüÈw¬ª¸î¥nB-¨Jv~× ¶ø÷ÿ׬ƒò›þùÛýj³ÒP÷:Xl)`(è%Zó 1Jr‰U¯šGb6R—ŒCë×!F‘÷-ÐIaIS]ÝŒ„O€ž%‡Œ›!ÐO{Nª€9Õ^J¹ÓÃpØä-ÅoÁ0*è2vÍuÆú²‡|î¹ÀÄ}—ÀÁs0eùísb?ü6ÍÚòþט¹µÈAë sµT7,­‡E­ÑwÜN`)«ˆåÎén$¼·M$§>È·)<-©ÃõTJU_}÷礑ˆhbÛ!tÆ[ vÃ6d7x•òl+2m>ïáP&ƒ,Îçøo+¥#ëèŒù¹Pd¿¡e+Õ΀œÒ‚YÜD‘›®ù¦î¹«È³^léÞB¾ýAƒNŸ¿¤XaïƒC(¨jç¿óô %&N¶:x²€Eä>Ð Š*”†o:ƒbG*j§Qó$ÿï÷;ÆMÄ!©wÏîk?ó ~à Ù—ybÝfF€`1*n*§B4ÇÑCwó ª°»ŠOc‰;dRè\ákœ@¸³&DçU_¾dW±"ɹÑÊ*ÄúO«ü‹J$S½8ãq’d‹eíb?}t¨ß˜7”¨[å­º=š’Ú5üŒœ R#d+%Ø!Í"æXŸE¤•p=,% é×kÓþõ¹·|‡L #œ Àéq¾ùš¿+—T žÇí–âÉaÈ:³Ÿ<ζ ²wjëÒ „M–¦ð½ãûNè¬ç™3owÝrÒÒ M³Î.Èâô‚z -âó@ßÿ³ -TRçÒ›zAUïǪ |5ŠöÅÛ@m"›ëLØ]¨÷ÑXÜi¥_¿OØ -‘è åíä=÷»fÞBgó™…%èN6ððy×½þ{w¦R^†~Ýž»Q}º00ÝatÚX† »çbáGYæÎS‘Wá§0Û½K[úaÜ"žÏIà&LdH&|è…)õ¯Vw¶E¯ù ²oùîà ¥6Ãg9˜³y{ûñ9°°©ø-t,Í~3ReC2›4Š3HÞ{‡b=KR¦ gow§Dê(~™ƒ¯&‘¶Œ™Ï‘Ž”oÛý(߇Lb‹uä]¥†Yy½”qg,òr‘ë8µã†EË=»¤©Q‰CYíS2 ¥çg=2ïAáÐ ­‰(R(¢Má<ƒ[ õ²;'ôQ\ʳý¼Q‡NÓ»¹60á »xúS(3’¶sŽJåzvÓ^ÕC…aPÝ‚ÊI{´÷¥“¡jy]}Íõ~Tc^Hµ§–FõÆbí÷Ÿp÷·c’~©¨•—U.!È„{ûé„Êl4Q—#Í1 “€úÚqžaÙVhx¹üa„H$ë‰->¢WC“£ç ¦ÃÑÔKÀÔd9îKŒ”a üf“Ͼ_yâ‹*/r-=ÐһɘL”öŸÞÔyæšbÑúÝÑo–çÞ -û§I áchçS”îKôÿêícyëþ|¨ÑšTb,>%æ}m“”!÷ñÕL÷š`bÝ&VœP¼2-ì÷ n_F|º‰ž?—äýi”æîëò¸ ÅÅxà] Q/3¹‹Ž'êJ'¿Í—#ú›Å‡w¢Ú?Ñð°ëq£î"4ðæqNg%ú[Â>¦þÃh2jõ€þD šÎß<ÙÑŒMb¶Õð-šË)¶°ºˆ‡Žiqˆ>å«Íoˆ©=dõª»â²/äæ×‡„ãÓ¹„VHbqϨ0i†äm-&Y[7CòÒëÓ* Å&‡Öxäq¼ú@0¢ý ¤ÖkKºÈiº£BfÒ‰ãU)±)îÈ¿>÷e›ð©¿ÿ§ÓŒc„o›Ïþäjý ™ÐÈÑò€²$^Ý5øY¿"Ç#8ˆ>7`A·Erõ[Û2݆¼_þçLˆuaA©}ÃÛ{PÌ=『IPzoäÒ\¨TÈÔèïÿ&Ugûoóy»@urD€S]ÔŠŠæ— u3œ»~ZkB£Dè ·'éÐB³K> § Ú,Ç\n8Cç‡ÉíAÐ}XŸe³ ô*èíï­}}ãq•s7` ´¾0¶ä) ÇúCiŽðãGø²Î2ŒæÒlÊôO€qÞz÷K¿aâÂߣ®Âdÿ—Ñëƒ_à×Cï´Àù$˜öòÛ4éü f[zªg–aá†öA¾°tuÌìËΣHhÌ\d®FòÙ±aj¤<+ªÍ%ƒÔ»¾½©¾¡†´jÏì]"ý#kÇ$ö/¸îxÀŠÁAd9²¬ad‹lGnî"kæFާü‹ԑk•;€ÒyÂ|í½qÃß÷Ï"6âÆÛj#!ò)(tÔŇ/PXQ  !u E,žÜy¥ƒ¢-ûö6¥ skS®¨º ñ)ZÛÎña4¹!jÀÃ\‚¦aF•Ž¥ÜWð{1¨Íþ”ríï…jg#Y¹¡¥]c˜Ö?½ÃK _´'¿úV mY‡Ô·výíá̳5üÿ¯ÞØ®m=ºå3T†Zþ4I…bÓž>.Ó=+æw)ûö¢/ñO_1bƒo'$ÁHžx?ó}œ=ªéÛÜÊN%Aê¥ÍŠº†ÞœÌòþå‘\¢.?¼(ʃ$ûÁýÄuRs%¡õ~|ëÌ¥èȵ»G{â[(:mSªI‡R§}ê‚›oýhm'T _—ž·)éC7¡ö¢b€÷8Ôy¯{¿äq ê&63˜¨}ÎHÔNú(Ñÿ8äŸIœÿû}H ßY¨b ¯ ánŠÐÄ;çÕ âÂÍ¡ý¤¡2Ép·õ/Z¨Š(|‘ñ‡ª_Ó,ì‘QûçoÇ®f&ï´‡º]®¡‰ON@½•üzª¾!hȹ“n›îMiŠdv1ðýy7'¥–tЉÖÿI €®Ëº<{~@ÏüØWRÿGÜ{‡cý‡ïÿ²GDDÙ›DVFÒuYÙ{oÚ¥”‘d4PI…JJeì•l²÷^‘™QV„𽽟wïïq|Žïß¿¿îã^^îã¸Ïûqž¯çõ<_0–ÝÈÈK· ÓļqÑ[0¥rJ+øõÌ|ÚRý0y~zšë°% ü5OäžoŸañÈÛB6u"øe™àGä+ÉËû2Kú`-ã`õ£Qøó‚NË_a¶HÞÛ nª!QžM°¹ îØLVìx[‚$—ˆ0'j!£:ß{¤ zžIÛð©DJKvê#ÍÝ1'E¤£žýfêÔô'NSE<·@†.Ë‹ëñ”ÈxjOÒbµ!î¡©;ö{)™ûÙ¼­|„{‡^ÖI&Ö ï1qUäx¦¢M5‰ÜT¢^Q OÜ9¾s½È§Y§Å艷Joò àFé‹‹!-¸Ÿþäb%05V?\s ÅzŸÉœÙD ‹_Gß3\DÉŸìó÷²QºnIfá .ÊLÎ7Êÿz…ò¼,º’Cò¨ £Úwú3*ZêÛʈŸE¨oØ!dz•>ͤÞLF™ëgïxPõgÔ—€Ô"<6Ez Ò. 5i †g"–P[ÑÛB>–À7'WÉ%b:Ôw'iÇ‹h(³ÛÉ\„À·øŒ©h4ÃÄl%ª…Æ¿OMCÓ] ¤çî—¢™dÄ£Ýhîð¹’üR9Z¼×4³DËá3‘ß‹ÐZvT`´W‘À·«~Îfïю櫦 #o±[Á‘º½ .Þ7øéÇBô Ò” q¬~´‚¦!WYÕ—¡Ò2Šž7š 3!)ƒG¯·çYÙ£!ç÷ïå%/YÈÌΊd>¤IÈm7¿º?'ä3ÁÊ"¥þm®m©ö†äS²-åN#\¦—¨] ©4±žº9²ÆÍ3®¤²=O9û$ø1dÞÎeÚƒìÃϤ£ gVJóè<5ä®×~–IÚ^‡KwyrÀ¹w‚”]Ü„÷-Éû 3˜tyðT jn×kÐèü³.'ÒACàœXŽüx6I÷» ÇÀç‚û²ŽÁO®\N¥IŒmWH= l)ÂÀŒàó*ßÝMÛWUjd ¯5k ÚEè—â]¨­ vÐ$? ö_“ö×|©ú9™E3hQ özøk´F«|WM%†vǯíCžrÐùðS°«"Á!Ûê’ƒþä?—((†²FL–/Ã7jíW[Gé`¤ßàæ®wQ060Š.¨éôêj¿Û§d)CÝo·—“2Σ]ϸ®ÉQ4 /–èz†F/® Ï£qÆ]o—M4éÂ/JT‘hFùÇHA>ÍÕ«óꮢÅã#Š{ýþÓ›¢,†¬Ú¼æ}Ó†vôB!¯Nø&xC+úÃÿSo»õ}g$~[CËsõM½û Úà›Sã¬ñö<¿™T.ÁWNZÝ£é"ø³«Œ¶WŒ ¹mR>–a>%ì¼YœTùßæïvotÒ|ö… äøYKÿx9¦zÏOåñtÃ|Eç !÷YÔ̼W‡²o cµidPµæ¢írêÈí¤ÕcCc¼§Û=êÐ<˜ë"5´ Z•È k> A[Šˆdή(h³¨Ù5`6 ¡Ë¡=ñÅé‚+Цÿ“Û”ìåözZÝË}Úд±¸j#>ÚZùÏK@#›$Õåvhòº”8‘úš_/¼’죇–Øé+¿$šþã›isdÏh§9ݹû¹t×>|éfº©¯ü–F¡[OIe3þ ôÑÛ¿pyf ƒ§Šõ†v°ªŒ·¥­4Œ˜ï´bØ„±à[†t/a¢D:òc–;L1½Ë¥}3*.åsÀOunÏoE¾0¯r“ëHù,Š=錦 †_6ï²æîn_™Á…»à5¬)QÞ½ó°þ”E½*x»¶vWýB¢RÊUiÜñçjÊØÉ‡Hâh©mhÕŒ¤›åôƒÁH>õ;èURí¥ó<Âë€47#Ë.CÚÍOt  ½õ×­÷å1ÈÐ ªæ¹Èh~«Íü¨32­¾½ÿ=A™+ÎÕèÌ…ãÞŠ6É—Ž—‘‚w«‰àß8îþ©ÖÑCnÞ¬˜ÚÎpäy+«up÷Cä^FF%>äßHóH}‹‚CÅDFš¤(<žêÏÃúpÞ­à ¸MÐÕ˜JRøs”P¸Kg•€’]ï6ížEé’È ­ ”™¹Ð=ÛT„òâǾ°>|‡ fÙ_«úÞ¡¢ ?Éß“EUÍ/,ËQ)­Èµâ"ª?¢AÕQ—âê‹xlŒ“Å•³5é{3·DmM5ªä½¾ycQÉ'qÔ Ë·Œ$ä6‰šs*¾½zÅ’·†FSMÑq?6ÑxY‡Ù'b'šÒ ìZ*@3ñ“s¬å©hnWÅ÷¥j-"î¼ ^CË¡t º¾y$ö¶Þ"ðs*&»‡À· Ò$£FßÌ>z—ÍôÆ2#Ó'þŸÞ’6Ä"?Ø ¢~åx4ŽXmÄÞ‘‚Šå.ý3ïþ/=¶$(>‘&†Ü´#èÈØ0Må'd²W|·y:GÈm¤{_¯gCª1þ¨+‚dÝœ»oB’”ck¸w;$û3&…¹©³ W áã«g †* ýàG__Y…íýrIÕßnA6uÏN¡ È©oM Q€ÜúOÂoU¥ ïÚµeŠÉ{÷ÖWµ¤4öŸùMsÕ0‚?u:c“~ 4wß8¼Ú ŸØz;O„:8çW¬‘ð Oë}|OðÙ0Ú;ñHø?s•ëŸn@±¹ÃþC©PrÖíèÀµcÛs؆íߊ¡,Ò¤ø³a2TpgEÑ­ý†Ê¸¨0ªzøÒhpøC$ÔîU»ªõ³Ô¿Ÿ_„&òâúC-eÐüéò™¤=¦ÐÊÜÇ{žà[Ûj/´¾"èôÇõâ(“Bèþ,ßyq:ú‰)_<Þ _ӥËҒ`8SK(2F|M®eÀ˜býnžIŸ,ËÃbQ˜¬ZÛ-VÑSuÅ×Èa–xÎÊð”)A§~x»澺*å'‚D‹Û©°˜~¶Î=Ÿ ~=á0ò™„•.Ç({XóöIé(÷‡ _#‡ÇzHäP ¹¿Ø‰©ÓGßDž@’¥Ew͇Hf·Ã/ûH?Rìϸme\ˆTÊ´îzo‘f¼ö«»A3î’» ûÂP ž0í­9‡Œ¥Ìä¯$˜‘™]&íõó&Ü›HËóŒqÙŽo8‡#§þµ¡ÆeMäÎl”¸´ˆ¼—ØukË ÿÍKb~fPp>ût,/îï0é§ E1)zÛu;”pá# ñîA©S%£‚úxÈT¼îÏÎS(«Þìóð†/¶±æÒ‹GÅ*ú”‚”¯¨$ãQåM*|w¯éë[¢š˜SÚ£®T·¸½È` ˆš‘ç•üJO£ö¯€Jæ;Ô?s0#q ¬†¹œÈÑ0±’Û§Ñ(3—ø£Ê^4þ’QóœýšL E ì¾T ù¶¥Iå'¯B.†=%2…ìê«¥L},CÕ«@r_gŠÓ¸8óx…ÒÊÎÚ—ˆ¾ Uþ…¶²4tPyVà’`$4oÕøRöÿÍïW–A«Qº–|p´±·…t(EOñȆŽJÑóío ƒ¬ í=9´=h£¡,#ä1ÊÝnï_mÏ]›øHpCÃJÿÏÜBÂmÙðþ°–›Ð¤üp·ËyB>|ΓÏ-É£Äíÿ㛪€»1åshë1Ûuùù4´§8’ˆ“gCGgG¢˜t•Uê¯/\ƒÞ€>Žo#b0HEu†é¸Gœˆ=aDØxËw€ ÆNŠ®xÓÉÁÄóÓ̹UKð}¼“aøÖA˜±xÞ`õ~ÒO%'4À¼Hã»ã1Ô°Èå;/ر~ÙGŒÚïŒ!èeN¨Ô]ÖNÙ*„?3gçɸ`KÝ;Náq,Õ<ï^;';Öd8 Z#‰I*M •%’N)¾ºŽäíþ¨œÈŠÌFFþò5¤q²-`M×GÚùóJªTÞHobý+u~ʉØþ˜»"£æØZ^Æsdš¸^×k·Ì)ª²+lѸ7&‚„y7²NÇ~¾6ÛWOêëÞwFnVÁäµ®Èqë|!"Ÿô:çÚ‘Úî׫t,/<+³%‹Â…«L·kPdò“ ‰rŠ9äÌóT‹¢û{ôJ–쨉ދÒÙÇ"¢ÕQæçØÈ»„o(¯èªBrd'*8òˆfÝEE½Õ„_i%oiìˆP)&°u0PUØx8–PµG3§ù«0’'áÚ‡šL§¤i[PÛ”ª+À·P‡®¨/8Üÿ; ÷‘ÚZøö¨˜W¾‰.§g ñ’Ä%?_4¥Žfóù³†f¢«¿×VµÑܦÑ÷7/Z„Xþ^‰–_‡ïœÿÐGàƒxsmÚ¼h0êa¦#ðM¶d$$‚À7R‡«ßŸôF9:âÐúé­¿OŠ©sÜ.µMAc_Ô§+P¡5åqà5½d9)yß :”Ž®µï…ڤċTñá×!ûv1âŸuµ'Z‰|Z?ëÛ ]P8^Pµ8M ŸÚÊwšÄBQûÖ“v”Ûs¡£î\>Pº›¹Q¡”ÊÔvIðÏ—ByNUö¥~~¨”Ñ&~TÛ_ö?Ø#[N5Æõb£½P—³æ¤¹ƒŽàOKH§„fÅVÃj™8h v±y\ä m‡®W| NiҊнW6¶gk ú,öŠ2÷ÂWrɤžó0l«‘yµ]‰ O‹¬ÓJU0ú­¤e9`üQ¾ÄÝÃ7`òš…Ä3-?˜rc(à%øÐw£Küð£ÑEÃA)æžß½be˜ïHHvæMƒEÕN“ *ðkÏ©sÕ"î°böÆX¿ö!¬n²wH5€ 2ßÓDU¥HÄÍû½$Üw”1÷Þ¸þ˜ SJ«'ÃH:ónIˆUÉ£@Ç´äRFŒûîÏG ÅC$Çîœÿâûøqdª¸ý FóЛò£D¸'ùËŸ½î•нcñæ ²öªûHy9²­';CÛÆg¡ Ny)ÙöW|q@~Ê^Ò€‚·üÜ äŠqÿ)--ýRîúŠ%Tw±îPG)¹K×?â!þ;¼zŠ¥(K\ÙDŠ+æ0¼Y DÅr÷_Ÿï¡’òÚuáýk¨¢XéÝÇŠjÆ%¥—Ó%¯m½ªhô·Ïd xðÉ©-êÙ©é_ßÖ^Ä,R‚±Ãýç“iÛ`Â),¤¸çòßž„ýÒáPqøÑ—5puDæîÝ•„EWšäÈøuü1cáwjeöÒÁj3OX»Ÿ-³ô6ö+ 9…­³3º“–HÔÊ[œ~í îXY !ý.„$$õåy_‘´£pîe—=’y®ß“?‚” q?e üÆ–ÛÇyÅiGc*X¶ ½NÇt·22äµ-z¥$"ã}^6¡\dêС«ÛÌaÞ©'âÞ‡ª›êrµYÄ“IO9¬jÚõ¾ä 7CD–ˆ³1ò„T¬«ž^@>á¿'ÈßCsõ~C* Ææ7²æœC᧪z¤~£HÆÏâ‰hã½ÿ&üw9ŠoVÏÖ´k¢düG=™”Z”Nø¡õ»ee~Xø“£¼Ö»ÂéɨàìþúËTÔl2ì¶AxWpåÖñ£¨ôÒ*FÓ`UhóFnÙ ¢j#GªÛ ;<Ögô£%Ú5YæÌ;5µš.‡þ!ð-¦³ÛZì êçR7#pMå׆)¿3owF¸ƒž¡Ñ¤zÞ1×3h¼°§¼ìò*šRrÑʱ ™HÈŽ¾ƒhnÕ9¬ínŒÏ›¢ÜŽôƹ8Lì„Ö2qYUe[¾yæïtSC;êo·ØšN ýƒÌ]'éåБzu dÔõÿÒ[áý/OÛZ"jO)e4†Ìî-ú•ü·×ã_ý««ì2R¡æãûHReöŒ¿š‚¤gÜ3ì ù‹v1/A|cV;Ææÿ½ÿïóÿ¾>íÁfòÉ!Ñyà¹DÜÇ¿úüÛcrùÕ–½òß>“oó¬¾ë’Å¿ø«ï¿='ÿÓ«ò¿úNþ§wåó9¡ ö鎿Ÿço¿Éÿôý«ÿûŒ*RfÎÜ j%øÝüéa¨»¸GZí:ÔŽPÔ„_ü âϤ|«×ÅëµIK{ŠZ£ðENýß^“.—C74˜Ô 7™§´(*m$N¾ †a2uÜÿ¦¾5Ú´†ìêÑûMµaüÐÇøúÌ2˜dËÒ_k”ƒ)n'u½Z˜Qéy<,·?Ìshnïxs:T\*ºaÞ#^ùÒP:,tªõÙݸ K/’|vc„å¶Ar>úBX½•÷çñ@%ü¹Ÿétí—,l•²¨¥^ÅîžÍúŸ‘øwSq¢Ç$’z©Šø}Éwײ BйâÀQ1F¤>Mñ#'i_D½LÝñ鋦Åóóq÷ Ý÷Îð)ÜsÔZ+AHYr‚†ÙÜb5¼(Fãí¬ï:ˆsº‘›D¼ø˜p<ò¼g%/ªóC¾UþV!(HetÒ©Æ…çKßñ>ºˆ¢vn3Á (±{]úYP&JQïÛü¹ß¥—ZoF]âB™žâ½3Ux˜s7É⎋¨X2r4±A•ô=øz2º3' >»Pj7.e<òSAõ|Óœ­ËA¨µg3ÑÄZunžyŸ;} õj¤¢J"IÑàçÌÅÊððÿßôÆðtIĬ,š_/ùRõþ„Ê ûº¬‡ ´ìeþ÷‡Åÿø1â®¶¿}Ÿ„^¿ÊË#|Ïï¬Ïqt0'|îç?~óÄcFÈâzwÓŒàfßêP†BÎŽIæø¤ývhìùoZC|çÔÿËoñ·Þ­}†Æ<ÊýUúК¼–ãcþÚOïvM…Ži×0z_輓|F¶{:ãù^vm@Ûjþ ô´4H©ÓÿІ¦ÏâS:Ç Á—£/‡)dÂ$ϧ@c …õfqh>÷Kºn¿´Ü} òÂi Z-žŸ³†6b»sgh»)tyOÇhî½eQ‡º‚ š§ k'[¡-O*ô”Û¥U:€ªsÎÑrŠ?gÿG?¾¿±º«£+^/ú`â_x–‹)|÷›Rzîù ¦7._·s&‚ÅÑ!Îuû`žÔöÐï7™°°ð>ÇI~)--ç‚•ìÎ;aMµ4ÁF%þ4mòí»Ð[²ãNd#QÝî[z&ÌHL~Ëþ7'’œ“wë;‰dļd½¡qHþ½…÷Mˆ=R1WïÖÎCš+ëo÷©EÚÙx—ÌcDH¯g>ðüI2R¼ LCFµGL7iR.Eë sÆ3ÅãäGqoj™l“„&²Î¾ÚnÔˆgO0xs:!7ír!«äñpokG>†Îcr‚È_.B§úI#\ô¾Ó  ðã»MßBQ$g†Ù]B1ãÙÛÏšP|YévÏi”|ñëy¹ÜO”îg’E™ª´î6”§{?qki _ïÓnT$þÌ;pÁê åÍ] ¨tãVÊ«=5¨<µìë[ÏŒªïûV–WðXtᑃ ¨Qïø"ü|j“GŒ{v£®ªæ-õyÔ·±&±©GCÆ.‰x U4,[Óœ~IFEwåý̂ѸB_*²MÚ?4'lžDÓŸ¾ÎƒÑœ©ÃžyÿZ¨yvÚ˜BË۴ĬÊËhUq*öÜ›5´aŠ`òVNGÛ+?œ¼õ„ЮS‡Éšy4>;kTô ã eí÷•¦x\мêÍ:¤za§nœ6 ng™‹ ÜÞ—x£Í>§Vì çÎ…üÖ CÊ$; ûð¤ê ¤êê߃T…Ô{=£ìtsú«ž×5HŸ–Ø% ñW#2œoB‚hc±‘3!×éNd |ƒ”%6ÅÀ—ö¼l·J|?d²—šx… AöA«œµTÈeW àôz y!§Þ¼þGÇ’#ßþËoÏ> 7åAÞB¿&¿<äÛW¦'H@¢N`JÈ(H_Zi¹ø¿ùæ2£°ÎúŠ|ùñ…+@ñú¾Ñèdj(Y{(p;ò$”q—³˜~rÏ™Uk¡¢ÎŠÃ[¡ª#5lßBu`lØïoEP[ç¥Â±G8©\§GmÿãõÐÇkc­‰äÝœó…ОÏ?µKæ7t]2T&ü¸Bï³ÇgÞk3à(¥öîž04b???µßbvgФÁ¨ {Ml3Œ“q•†çÂÄøÓ‹o×Á÷ñ.ò7'¬a†\·ç¾n>üàÞ©,0ËsÒ“vÏU6`þ|}¿,Ô¤¿;-ÙKÜ.ÆI°œ›7h,`«ºõ3~›.ð|³õÞ‚-‡ÆÆ ”¸¿ÆI©ä qAª„‡ ©â’æXÃ’µ£«òmDŠ´ý ºáñHÍåæØÍ•‡´ZtÇËÚeþ¡áºä¸;5¡§´ò22-ÌñÑ[ˆ"‹ã½AãžÜ·¨vÅ$S ÙóÓBÔî ×7 [ ä¹zò%{£$ò9Žÿ,Œ„)‘«ƒ_,÷;ÙÄÝ`Å[_‚C*V7ÙËV‹áOàQršKö°E6tÿÓE$ª Vø~h ‰÷ì ç Øƒ$Ì?ìß èåà=6V½9¤Ø#¨•;é‹T2oôþ@¿Îâ¯>ŠH»œð[TRéõ»š®‘("CñˆÔȨ×HñUV ™Öùu¾ä×"su;We3îmVº"õ¬Ù(5Ï<3wE×5ýÔëœÈMùÄ’_™• £_•ìÊ9È»6KAÞ_…ü)šI›ñc((Ú{þ Õ¬±‡Š¢HþîbµÏò(&Ìqä¬D?ŠÏæ1ܿ싒AG ËQúJÍàåõ”y¹CÚšwåÚbvÓ︆‡¿¥ä­|eÀ#…‰K£)“ô|ÑÎ…£¨d=Û‹ÊÕ‡xî+꣪û«w·Pï´ ‚¿|;Oê¦TÄìn£V|Po3Ì´nÑekoÍr.Âgjꭠ·÷“½S²úêüYs­!=¸V–‰Œ Rù„VD éøéÛ,ö`Ó{³ } âwûïyˆ“ȽAQ‰nB!¦NJ¶Ã>Ô] Ò¢nòT CæI †j¯7ýñ±=×'rÈÍ–ãM ‡¼·þu­C¿ 9-´IZ’áïú[^Ó"åîú»B¬ƒ ßU£HÕóhî }ÀÄLøÈ t!ýß|»v³Gë"¥¿ÿ°{ûü¤U𠔊ßÙ²áõO_×;æ|(O":S*G“•í®ßßCUˆ÷w¿O%Pí=¼·èÔfμûœ ;Íb8þãcvíA’hÍnd¾GyÚK׊l­A×Ù‹4r¾)ÐëµAsï×]ø3$œžC¥7>ÈÒÂ7Ïš'nwaTìdB9‰Œµ†åx&ÃDºs †°|OQ­š'ƒéš>ñã·aöû ·úî‡0Ç®r3§”æ-cê3»Í`á3çù‹3ò°äñüÌÇ£*°üêiÎã°Ê6Ô“‹Ö—Œ6“ŠËasó®å¹]op¹ª÷ÙHì­Mõ I6>ˆI½7G²gääŠ] Hqå™U6±"Rõ|.½Ÿ‹;{ wÅÖ"½-ÛÞ³Qö¸ûšZMòÂ2Åü¹qY¨ýúï ã¾0öôW_‘ݲ¸‘£U¹TÒT÷Fž£)b̓È'|Û"êîðÑfý•5ŽÂÇvÞ¿ÕC|• .ò¢8K%õåN”tª‘ôïCiÕõY)E”á2<–:B†ò¢G„§z”ñÈ‹4ͧ§ǼqY<Ê·jE,3HQõÚ }:{çã(«øŽvGY~˜Á#nUÓs©x´ª[á’S*IV)Eû¡r̵’‡ÉwQÕÀúíÚ²(;øÄ´O›5d”¹ö%w –éºYiÔñÓu®°E½ÐpséhpAúàÛÇ_ÐÐešéN.%] “ßbéFãsÅônŠºhr)‹o$øšºŸ5­5ì@³ÀV‹” z4nסUE‹¢ïœÁ¥»ÐrxðA²ÕZS7×ïDòcªš„çÉ©É/‰vI†CO Ð~¶â«êtd RÑg @J§¾8·P§løïþ0”…¹Mya¾Ýo÷ÓÇâ8䉌lüN°‡¬[©L"]nWøXd Ri¢/”¸úB’ÖW®Â í¼F~Ì´â².„GP­8àßq¾Ë/¦¾†¹ ßɳ¿!-EshÂ2fš™Îî…‰Õ÷= O_¤D¹ûäv—üÃkúžyãó“Ô§z¯p¬žwåé„ü ¦œ½·ý·½<þ j¼ˆ8˜^ÿo¾¹æäޔߞ—Ñ(˜Úî+© ¥§Üzކ…AY`’œhw"”÷2Ir™„Jr©  ÁPv<()~ª]Ø^úsBmäiiá3P¿–+kÿxð?¾±ŽNñª@késÓŸ\ŒÐ^ÛšàuºŽßëÏÿ½§Ò.¾ý y,ä<`èMˆ…ïAøfìsõ;‰-Œl%V{84ÀØ»hÚ$‡h˜ð2Š›5…ï.i÷Tb—`ú¡ôÅN75˜ÍÈKv© †9ÒȧݨóBÍÛ…`!-HÂ5ö5,] ¶Úê…åÛ—¥é»Ãï©¥ú‹Øë‰<šÕ1Ô°ùúè7µx9$ª3ý-Äpa÷¾æ©í¾.†Û7+‘ÌôýFæ3¤8¸YÊq±©žÙÌ%Îçàθٟ3}$ IÇÝÒ_ó#¸ÝÉž­c0—™³nw—ÉJá>ù¥ÅQö‡È¶Ò¨äÐŽœCï{ˆ”™‡z¹¤»-yçØ¸D* À‘*·O"I(ô#}=ÉÛ0Åñ3øãÁKOÖG(ÉÆ][_ù ¥¦C⊟Õá¡Ò­³îƒü(—iðéL¥(!g>ÃâFøbÜÛÔ1UBe†äU#†·¨ò#ˆ!¥mÕfÕ_½Ë8ŠL"ßFØ¡–î5sêŒjÔ¹#¬.ž¬‡z¯dì¾Äp¡Á} ]j[f4Ì_¾þã sÙ<ºDIhœ õ“ôšò¯HAÓêÞ¨ðõ4ë ü Þ‚h¾ôØû±ˆ6Z2vñtò¤ Õá3ã÷ߣõi÷ó,´‰ OéCÛµâ@²hÏ`g@!Š–NX+!½›pVÀ“ïÐôŒ¤lÀü$T²ô×3¬=…R r"o¯ö¹ð5 þí3Èÿ¾pæåË‚¿{FR"±9¿W¢{é +¹ha¤ùdÜdz½ÿ|»/ùÒ¬-A§éYÝ¡à÷2Q;Ù|[»ªgÿîk–©_#pgõ„…s7´=è²¢Qº r +>‰QÐÙK'îÿºÜ¹"XÎAç½§>kâ¼ÐQî|ê´²gìÔw€¦Æž·G—¡a¯à†ÖóH¨û0éf³Ýãü'tú'4m, —A 1mÔEž!hé›ñþ è­×){÷+BëTóÀÊÓ}Ðöê¶Pâ‹\hO3âjÜÝI8Þã!ÿ–Jèa»ÒÏ9ë ýgDcèìwÁ³×;o¥HøF—æ²úGF%ì~Ѹ ãk“C•IarÔ‹[Jh¦·08à gnúÃ\÷Ž“:¸]p¡­u–|#œêae—¸¤ó®Ï°ê:AzvÖ7 ê~^õÍ÷íå·›}({8‡ûw3KQ†È I]‚gˆÊ$»gtìE…/R\hŸŒ¨A*wã¸u+¤I ÞÒ¥âF:Ê”œT‘Þ¨'Sh\*ß½Ïú˜ŒŽz³ïÕšq÷Æý ÇoÈBñ›äµÚ îcTüºj·‰lJgŒ™J‘#D¤fï]äÚbü­ÈU†•D|u]È\ îí^ÁÅJJeéS'•sV e‰u ³0 ’¤,‹—ôym£ôiÓʈ379HÝ‘ qŽ£¦ïwr]ÿpèÂ:$÷4Äy<¤åïT"/Ì9ßGŽF!Û}–gæäAÞ§]Û]‚ž[/Eœüù#\2u—þîË?&ŸzŒ…p¿„†{»·á½Ådz\ÝPàþN‰l{f×Ç™ÿÍ·pù¥çq$P¼«,%¶¾ JÎ^&ý1 ¥/.d²?ýeÅo‹r_FB•0QvZTò½Q9¡á U‘jbmaDP}Öœ]ëó ¨õïIJ¥K‡úït,,coþ㛀ÃV¬R´6Ø>µ)†ö¶Ø3Ëo’ Ëæý¡yfèÕë63Ò= Q7|Ž+ÁЭjíz%ø¶_ÎÐfKFZæ'$ uaìÚ¼ÿ[Õ0˜Ð¨|껾+ph<ÓR†iÃØœ[AÔ0ëE÷õã]øÙ3k·èÑó²ï-V›VaáýÊŠ×ñ°ds[Ï`·3,ŸÔ:óõ$-ü. ,ttô„u‡¹Ð±@#Ø<òë"R=¥9ÐÇÄDZ¾ß#”ä6ɯzýCH¶Gù×þÛHþã¼p@×i¤ÒoÌ{f;-ìrwè,â®EÇ $'G†e?½5dbLÿxl‚™í=;VGýqo_ƒ¬¨‘²…õ‘Ÿ½…œÞÑ­G’»[ÉT¶ó0òææLæe8#ÿŠJ«ÏÛ«(aq¹‡ïŠÄí š|m‹©â§«ÔQ¢îæ)æ¶(õTà׊:•ú@­G åNé}KgFg¯êq«<Úi¾rÈg •bWk];²Ý×OûÂÕLfBýªQýܨ#%9j>åzâùæ;jy'RУn»^*GÂ;Ô¯ ¼ù®ú šIR{©Ÿ@#w§á#h¬ÖM¤ ¥&*;tE^¢©ª¤4…(šir¾y~Àͨ-\¾)¡…Ýø\Ÿ²"Z^ÎÖ~2º­ü’MìzøÑ:²ó®³ïY´)p(%Þ§Œ¶ýwƒ ? ¢=Ém­§œHÏ<úíùxh"ϵ¨“ Š:[ž»Wí ”£ËÎýôö~NCžð¿}ù>t‰%cÝg5Ko>¯9~&Iµf„œ&ÇšòÒ§¦æ¸+ ýº¦ ÿá·ùGëôÚ^¯$JIüË·/2ÚL•œÿö—4_pqˆg†Öcsת$Ó ­Ííðèabèð¾õBÜì.tí+K+ý]u¯^æÝ ƒÎ9íŠh?ho¨¼Ýÿø%´ÊI®Rü,„¦öÂþQ¡ßÛ×ÛȸP<¡Y¦™çbœäz )×à4ïX:4wŒ^ɰ€–Œ,©zÒ£„ãÎx2/ú@ë§,•ÖCÐvñøÓSÖöÐîÏo¤çÍ6Ç•'ry¡{äúT(»>ô³ûìï<×_ëÒ¯m4ðÃp[ŽçeM™Ýp¤ÛÓãb™w{§`ò£…Ó§œ|˜¶›UÕW¦¶NÑ0W#ïü™¬¢êIÞúÃRüm÷ЗްÜõ&ƒÝ[VÅÝÖ87`=o5uØ™6‡R‘(µÅ¥°7‰•‚ •¥×d¦uBõ´3’eæX÷Ü®GŠpb¢¹¤z¹yÀ?èÒŸã"Ö E:†Ô€°_…Ûó“1$YÈPÝiP–&Œçè2ãé)!•¬ËÈÂù•-i î;ôi¯mY²Ù^×h?u9^—“£»r-ùˆ<ʰ8d‹¼9ì¥: ñÈáC€øµ<t°Òã½±…}žzìx»Ž"e`Љb^œdD(ÞÎùEЬ%èrWï ôÞ§G¹"P†q‘uAåxãŸT Nàaæ¯á[ÄPaÊÕiÍMR¾x¢Ê±›íu½µ*;ÌlF/£ÊZwøÆ/GTûtF5ø´>ªGñ|xõs5£‹²ÚÝ·P;ŸáE··êÖtu]CýÔÜõ¹bj4äïbi'B#ÿýühL+þCõJš—ßà`{†¦$º²Å»9ÑŒ¨ÉHmæ0šm‰~âb曕 [JÂh±eyqŽq­vPù»r­I9¹)Õц2IXVÒméFõ¿f½B;r…™?ãhÏsÛsMz)­)_uBÍNƒ]fÓ¤PÚxê¦Ã~‚üÚtþÑöþ[¶4ƒÈ,½®FÁÐ i[ÑÊ–Ç!¥ÿ™Aé8$±¶:˜ä‡øÆlë—Öû[¡âV´áVQèjáñÜ@)9+Hö̬{vÌÒ*šŠwô¹W7Y‘r¥.Xý¡'è˜aýô¡(¸p|S?†f{?·†ÜÞÈ¿ý%ù>;e_”%ÜÒ,Û¸KxükÜïzÂëmÙ¢nAÁ0)—ŠEãÿæÛ» )—ŽF ‡Ahð%”–=VZ9HøœŽÁë!5P¶¦®bû*ä4|œý•Ê4«–˶PõN°iݪmB¢vÙ|ƒÚKúüÒ¹’Pßuj¥åˆã|;(ÃzÂAZ{øZŠu/@{¿×Õ²qè2+>Ï-¬ ½²ÇÖdVÚaàÊÙ‘µÃä0dSú™‘ņ·~ˆ7®SÃȇÃdß÷Ö˜²òús¡ ˜Ø½käðGAøNH16ý¦™X‰øß4Ãì±Ó,¥¯ˆágbô3Ùœ0/ðŽ#²®ÂOù³†%%¥¨ jXÖàGºÑðû{²éý°¾§x€øJlô&ÿÚè9ƒDVWˆ‹v¿ÄeB‹ùaH‚™½ûœ!íø:ó¦­ÉSÚÿP^C**eŸ[åY¸“¼Á›™ó>î ù•ç6•Œ ù{^\öôFÆû),DÈLy£M2´÷zÇÜt ˜D6™iÕ*OAädoqú «‚Ü1&žuO ïmãôcÕIÈŸQ|PY…”ÿ|#¯þ‰"‡¯-g¯¡˜ïáUÛk›(áÙ´áèÇŽR ì>V“xˆÂ.iWûI”-½9dÿz9­˜Gù?¾gúRŽJ‡Ã•›»Py2*ö’%ª„>8õÕ}~©²x5zêgR£Öz= ÿñvÔ¥=½é’*€z[51Ge£ÁP0þ*CÃá¶Øöz4j'Ï x؆Æ_¢¢÷ßC“Ï÷¾¸¿CÓ¬ ‡éî24K °|)‹æqËÅCŽõhñ$u¡eô`õ›çhý攬ýZ¿÷(ó³@›¸K*!h›z<â¦ØÜuE…Gû[4º;Íä3B…wŸÛÚ¸”,%ɸêoŸ—\u¥üÛgÏ,òcB’g»?ˆ}6ý!äЦÌ,4óôèvó±Ñ)H·+UÑþš i“\T½#é1œW=GÈ}sUD÷ þãÛUŒóË·þöO6g°u›“Bk¨w°˜îhçØÃúYÒ:*ʼn®lç6½š-kèæd9Û=]]¬}LG¡ƒ¸úœzAÑö~‚±Ê´}ÛûöZ݆úßó;”†¡^45÷;'4Ø7|îo*‡¦«ñîRŸ€æ„OJ¿_õBËÏñźehenfßc]­Áá  í¨ê'Á§¶[¬øÖw†Nq©÷Ú‘ÞÐä¹J¦³úZÛR£\Âá«Óù¨­} 0ñ©÷Ï6Œ$½~b:zÆfG <]›aÒC*ûØ1˜–ݹ\ÝÂ?¤•wyûÁ\Á߃\G`áÞÀ«‰Ë{aÉÏ˺à\,¿¥—ªšù ¿§¶¦Žwî†uù'm.Ѱ©\½Yü‰âJ%- ~Ѐž^OeIiÿ/ß[F²!¶«ûŽìAŠ2ñ}ZôHU˜@zœ­iÚT!/” éXÓ¨w O!½i_Ø`$2ÔS­7§ £sCÛY;Ü£ÞYYJ±†,JcÒ¬¸Ïü°a‰²Ýô?úì<-r$Ù¨^išE®Y2%ñÏÆÈ#iaÙ° ÈûF4‘ÖEùÍÊy§LQÐhwý\s {u~xoøE*-?o*žE1iu²hZ¯ ³"{÷ % ͈îNÅ¢Ôú ƒ»xhBõ³ƒc!ÊN$ZJŸBùÖ9¹d:[TøpâèŽü¨8µLfç*z2õÚ¸ *yýzã ª4‡?½íºŽjwš FFQÝøÉ„ƒjjÞ߯±úµÍýJ‚l´P÷ôM1/JÔ7ô8Ñ™„kñA•«h8[žŸ÷I¾ž³¡zƒÆm:Üù߆ФâòËÑôSC#}d#AgÝKû>h£yT¦–üí,´x´ÇþÉÜ7´ôÈUhF«ó"´NeÄhm¦*&&m”RÉÉÒ‚Ñö@¿¹•Ú1g®$È^Bò˜}‡EÖv×ùZWvf(u²Û“­Â7‹í÷;BöÇ‘`ÞïyÒÒŸ(@ÒšŸ¾¼r®R*iK_êíÛž!:éÇñoâSÇŽòBìK^ë¹K2{%.!^9TªÅ3’9^üœ:D±ýþ]QÎeÝJ¥þ-è³½îv¥¬ð<ì#‹Xþö>‘šŸº1~vÛ¿^‘?]û·2¿`è\÷S³íÜ6.ÝCÈygßVhþO®ôž_P0S}°}õìÿæ[jéUcÒ—P|iÌ5~ÎJRw-{m¹C‰Ù™Þ‡#P¾?¨¡Ó*μL¹³}=¸4ƒ˜cPOåÌwUª ê¸Ú×@­YYd¾!¿U~ê|Âûßä¨e8oÕ@ëè4½ÐIh3Þ2Î ]}Ò{nC/gæÖ÷ª0 „µ®ôõ0t¸÷Jß îŒÙ¥¼<#n>üo¯kÂ}¨s Œ÷ÿ$úÚÏ “UËÕrû`ªë¡V]VÌÒ}Ò;w×~zšÕº»œƒy¦·#ø%[‚:j`Iñ¼ô‰hXÙb9,¿-%"5£»`­î±—é l¸ŸºÝKàWkO4ùSÜáO~êà4’,:½Ê€¤a·F™…9‘üÒÑóäÇHYž¥yV4irŽ™çÀ]ÇR›ŒÁ3íU12zœ`R_Æ=.Íç•…p/Ï­‚‚0dm_-1¥lBŽ‚Ii7‰ävnóXè{†¼ê?Ü[G‘ÿò‘@š uœ~gvè²"î¯ë² k GÑu&­Ž\” K<Þ’ßô;‡ o tâÎúÞ&_”Õ|èÜ{ëh=òçj@EJý_tÁˆ/ë]ÇQùtüÕþ0Tå‘÷9öë'ÛmÁæ<ø5¸½(í< ÖaŽÙ¨ÂpÔQ÷֯ɈG½#-mѤhÀ¢{Vܾ oж#«-”¹ØÜê€Æzùóâ®h"7h3æ½MyÔ¹UîîB3j"ª£ÆÐlüј š÷3š¾"A‹ð$GËØsÊkG|Ð*øyù‚ò%´vîÞ80¤Œ6V¶ã±Õ½H—VpË©Lf8ïÄ.OB…”uý›y(©ó4f¹"Oð•}ájUîû ò’5¤i½Îÿ³îæ­ ²#(²é„LÖ#§úz æº~ Ҭ兴 Aƪ¹ñÐQ‚N,S‰—þã[ѹÛÉ÷ÿ»~ÀÑ #’hýº3öÙÙÐ~&s=L„:iúûƒ“,¡«:®{D1º¯>=´j¶]§¼­îªC‡\aÚØ0´^”lõ»zš†üÌ%‚È Ü”à`ö‡º•Ì7‘êÔ„ãGÝ)M…&‘#ßÅ¡ùVènèî‚GÕWlO?BËDægÆ2òíëñèõ:A SW_2´¨ÍÒ5-…NRÒ”;}ÐíâKqÓƒ úIŸñ¶¯T<õD§6aØ®ô‘·±Œœ?Zö*㌥^+½J— “Çh·Û5ašN4Q-m~°ˆ•y+œ†¹¤Ø'oä?ÃÂÕŸÒóG’`ÉÞýÈ791X>ç“yÍæ üŽÝiWóPÖÅn¸V^—‚MúrökNoèmÚ¥¨@$¶{Úle4‰¤âE ‰H¾Ó'ÀrÙ)f›i%£!Õ ÉÀѤU¤™à´¿ÜyéøÒÝo܇ôæ¶÷ R‘¡éhÜ`úQdô¢­Ò²~{¬ÊO¡Í]d±³©íñKÃ}7rðe#Û³§Ò¦—2‘#oÑç­`4rlŒîÞ‡<œåFÞûŠ¡M{'‘‡KbtçPP¥^·X$…Ý÷ý²Î:ªª¨[ût¨4HwHII ÌIw·€X‚ ¶ˆ¢¢‚Jˆ¢¢b‘ÒÒ’"J#!  ßáßëãÞ¿ö8ûì=8g°Ÿój¢=W»Å/h¿®¥"F¡†Í _°Ž'¿É A'AgWÊ$Ò®|΀æc9:à ü)hz=JcÍàSöîÓ}PIa›K#ôw§O˜iU“yß:UÜ×X!—\µú÷¯†ù!ËAšUž~ï@Ø3H™©~޵3oRÍghÞ«Q/žP^…„ej‡=Ô˜çõíl ¤~¼Å¬!PYýv—ü]ã!¯F»A‡þ*Æ\”¸^ÅÙ¿ï? †’7F1ûªþ£s•œ_ÿö(ZÍÓÚ,¾EŸâ>ýö”$äÓù—\‘7¡øöXQ{‚ÿ ÷¨íÿëoÅ2Üàãëµ^q#¨h;|¸ˆ²ª”4Š‚ÿ@µK€­ýðY¨‰<áïîµ×ÆÝõl¤àSú,³ñácP¯Cõâ"4‚ RÉASc‘öÄØÿø™}m í 9fáW«¡snߨ õaè1\ûK™_)ùµ·Íra€5ëçT ³. Ä¥‡ÁHºÓ\ÞRŒÔiÛ·ÕÁ÷¡Ñ7æ/ÁDb3O5L…¸ÿ@F¦_=9ßà ³½ wèÎmÀ¼É.¢RŸlX$}—4Z}–n€É“GûaErp´«Àmß¼WtüÚ§ªVu6\Ω†¿l\ÔvEÂv“÷1ÊUv$ÖýÞå0§‰$•L±9Í·Ìžó[èõcH!™äò‘U©n죷x–ƒ»O¦ïÁê¤mâcœÈ]C0¾ªÓ„LòT¯]žAgŸ/Ñó¾ÈZcû—l^9.R©Dú"·ÃÖ‰³È§ÁMk1¨…´ÁZGV¹PHìÃxqŠ<Ó¯?ÎŒb.jÕÞ“ò(ya.J½Á¥ ·ŽLDÙc5yÉñ€IÉ\Bc,*´Ä’Û]ìF噋Ö$“¨Zyt£ËÖ¡ê¬çI>ÔÍQ•œB­{ƒWöí¹…:îôwbÏ¢ž»¸v³» \öieguD£›‘ïx?H Éã-/43­|dT÷-hƒˆ“¢Å’« ×Þ´ìˆÔ.Ûw ­räl’ô×7üòÀa´qóØÅ‡¶j{O¿›F;Úè+ùk¾h7¨Gwô2Ú§Êò–³íA‡KYµž¶wñ(œð_BG2ù®…H³$×SO¨oÚ÷>ía†êñ¯ücm¥Pk|ú¢'?!×ÉL`›þÇ3(Ô:qå@]ö?H‘SŽòŽûhRœîý[Û­vøåïövAæ«ÍÕ:!ǗѺ´j§It0êümüϳãOþíÿÖ¦ÕéÕ= Q›·Bg–U¿À}è¶ ½¡j§ ½l2'þø±Bo5 {ôd$ þJ„.¯fNBÝ”ýQS¦Z&Ü"ºKà)x ³pv|®*¼g~+ŠPÏݹ)’ûšÇœM¬kvöË dé$äÁÝæÚNîÐVøÜ…QÌhg>å¡ð ‚ÏRu?~m ÛoÆ?WÐAWw—½»2ôªÞg™IoèW(õ€û~gMÈ@æv)Œ‰ ,PÃø™­zפW0E÷ÃÈáLO˜®x”?79.ÝЇ…×1ôVKŽDŸ7©VT.sï"„5™EyÎF"øuŒ­MVG6& ì?\®†¿CeY ’‘èetÚ£[ç‘Ä‹õÑÉ£CHfa.îÜô)”Ç£^d'"»°¼áo{¤ÞÞ*àP“ÂÝ¿osü@ZÉ\&âb¤·ð{–& íç¬g³+)дHâØdñÊ")ËdDÖ«íÇ_hë ûK…Ê }äÌŠc>ïˆÜŸBÊZým‘÷ëÚ}AÇäg¸/ð¾‹Ι\{Tþ…ĉ\Ž EùÀ¦°»æ(zÉù÷Df#оV渹û•¿Ñvö £tqàõui”•y³5r•ίڮEù°q’cÛq¨x.áºÉaT6f¸B!BŒÙÔGè! ÕŽ(Yµ|E¤Xn)ý›Œë ß’Þ¢Öu–àÛ¥Q‡6!9’–uKÎ{ Ù\EýHNF©_¯Ñðñ)šE54~]h~ŽGMŸlKÿ ÿ‰æÎÁ‹[?Ð"J«XÞÿ9ZÞé;¸®.‰V^Ž–Ï]Ek+Áò{ÐF¶±.ñï:Úî–u£ÍCÛÁð“)[_Ð.5\³+¥íÏ»]MR‘¦¬[ ¾¶_‚i— ìƒü0ã^Xpø Ìs•+nŸÞ€…¥·g?jÀÒ¹T¦WF°ÂmÍñ°´ V—òê^å»ÀúšÃu™19ØPÚ_rˆ× þtËøi_@!R)¿ìUÙÊÆÝ,\6‘zH둯Áóâ2MxÈíBÆijæÑ÷ÃÈBqº±÷M#²ºK¡©à"r0í-ò@®u–Z ä£q‹ÉžGþn‘]?wۢ฼fä•@Ÿªá…=(º+¡³Ý†³õÓ—ö¡4{zÄu(FYY_㘀§(7šþlm%*Ø¿%ÛÑ«¿å}R~¿¸„u^lZÀFÌêÁd_kø›ñÁ¾L$‰"yJwD¿Wç|º‘ìœo$Ý|Rœ4ÊìõZ@*?§ã–~¸‹?ùáZ“+î¡í|óíˆÒ*ä=÷{é¹xãNF†®øí-dz–Fã!ƒ,7_Ǩ‘¼EÖHƒ {’d/Îý=mÓ‡œMÙþàjtŽùe˜v£–egér#jwÑóͽJF]ŸÑÌèËD¨¯ö=(ÖW ¥¾ÊÇ-) ±²ƒ/Sø4U¨Œ|ÌýÍiüæîsŽ ÅÙÛÏ»”¡¥Cò@÷2Z)r?R®EkV–¤§Yh=Û s!#mŠÞQ«Z í ÿ ßÁ횸r¯í~ú-ìyÂö/L»"T­ÐAu·£‰0:tï–¸y<ŽðÂCk–žê­H6d×8´ô jb•(›3 õ èý,nƒÂû›áª›G WÞÄž÷õEÈ1$Ñžª"äEÏ„Üå·~Ãõ¢p¤”Žì²î…$Æ;öÂþp³Ç õ"ÄW'7e&:îô+%3 Zwú˜Î³D{!ky@S5N ò-}t¿ÿ€"•K×ÿü½%cÃçRgƒvø‘‘Ë)~PJ)1©fjûoÿîâÓNëZTyPL£úòAw'³mkœ‡â¨Ó Òœ „žùpè Ìÿõ·vZ±ÙGvÖ½½“Þ€JÖ_’²lP{`åJ 4T×{¹z»ZCÍJñ‹¨-(ï:½ŸŠ²^)Å@½¤_‚p4ÒñQ:M*ü "Nþ¿Y>´ê3‡*ÏG9 s»#Ô×¼z@ê119ôõü «Oˆ€þœ:ÿ‘<0”dHêÞMqlHšv…ÑŸ¶·>×À÷wzN}8ꡇÚO°Á‡:µäÅ*˜æŸ©Xš„Yϲ¹ÃË0÷ó¼·Ö)Xy#/ZÂK'Y¿Æ‘­Á M=Í`â)Xíר°ëµ^^¤Cð{ö/ùe²øcÂ}Vµ¶Æÿ”>©ù‹D 1¿®ÿD­ §M®º i¯}ãU¡)$Oõß>k‚”Ë{¢ݳpWMh”Ç©oHKiämç¯ôÕ‡8e"ã»·s†Ý]Èœå^OoÂ…¬»(ž9¼Eöa…?ÙŸ‘ë™dØ,ßäÍ?êÉ^wùßTvøñ‡¡`tÖ§ŸF(<‘Qæ"Œ¢w(Ÿp±ù ÄÉ8U^¾!”òs®¥û£Š2}4ÄßP.‚CNµbå>‹›™¤¢Ò£ë \xÐ…ÑêÙè8ª3l¾×{†Xò®w:5mÅ]¨Ä 6ý’˜£¨³Tœ¥Žz‹LÏhš†Ñ`±?WÈ Ñ\³ýëñ%4)N xY‚f—Ïn s¡}exN4)Z ™Mõ DË÷£ìˆ®£•;ñ1½°_hÍ{öɌߴnUØ\¢F›«ê¾éÇÑ–#éÌI{´Í«ú¦Û&„vúõŽþYh×i“Ð^d‡öŽfmr//àžK‚VC“ŽŠÆÙçÞÿ¸æ3o¥Ž¬ìŒ+œ,kúÇ3ø/ï €&ÅÜ 'ï§äsÒÝ–¸{Aö%­¹…ÉÏÿ8ÿüÍéQpEs24ÛÑÄ2¾¶VÁ²šÒ<’.íGoù™Â {h¦òE#ô~Ðô“Z鄾S¯Ÿ^‡^MOÕÂOB]ÕI=ïN¨»J²_”fïì° ï«ßç³C•]þjÑ¿×ÿý{ÿ½î¿Ü–6G—hÚõ¶Ÿã×äÿóLºnøŠvË.AÏç§rí/ÿqÆ¬É ªa8­=™$ÉFcyìÎrwÃ÷¶nÓ~7˜Lx7túÖ[˜Ž ©)´°ƒŸUËßî,톅›Æ—µ`IRðm¼+,Ïû Š•‡ÕÊv…³‰ü°Þ#[H”s6œå¹_]ýý³/ +G¢[ûêu´‘ä/½ÿ/HK\ÍHŽ3kÙDUʧå¶_ò[Ü¥mãÞáü÷d–=Žû‚´PÀ¹ñ~‡?éb.‚ ½½iT9µÈôüËÛù³—‘%âÛ‰#kÎÇ–*Odï—“3-:œSÌO¯½Gî‘ReYñ2ä­ŸlÝûë,ò-ÎÞÌú„Ú^¦ÌÔ(¸¥þ²›¥EX5‰+5 ¨÷˜¹8튷½-ýP‰ûÁ'%žP:ùbôí”e9#PƤŒr÷”eô㢼|¾ëÛ`TØÌp‹&<·JÕrGÕQ%”=§†Ukë= $®ÝY¤báF~öÜrZÔÚ÷$ ð‚+j¿ò{¡/¦ºBVKÇžßC½±ôM–õ,4h2q²k‘F£Ž_ÑùÏÚФçÎ_i4+?]̣܇z•ÅY8†–¢Fl5g‹ÐŠÜ£n?ZõSŽ]µEëĺJ<‡6§ô’VýE[žÓ·¹ý'жZ˜ÝµíN=7¾qê6Úý=TµRÄŠö÷â4©­šÑ²sôM´:ÜoØüœ‡3'5ô‘Ì€eóÜ+¨¾—M>;ìüïñ_nHÎÑû'7}bÿé)‹Ùy«±)ôŸžRä4çj!qìò8]½$ðW䕼>ñŠRN—Hàýž?]faRÿ®ËÚŽN䨺ÿ‡ò_=ÿoÞÈ_ÿo>Ñ?^Éÿæ–pözóÞlø¿þ6ÐîÞø±*XXì^ý ÄÝ'Þ|±PÕ#ÚlwÛªÿ›ù%¢µ‚o›Ÿp¯Cm§D¨øÑIøôñÉFµPåN` ëOo´çBÓMV'Ÿ)‘ÿñ·ÿ¯Óf*꺲OÐE•|´öýÿð…*<åÒ-‰¡ÿÎ;W'Ÿï0äêwé7wŒÈÞ £™8£_¿óŸÔ‚ï¦f‡_ÙÀÕé+¾˜ar8¹á+üw<{Š2få™…Èä}`®\$.Ð5 Z_ýþ\ð–îÆ¤|t„å-ub©"X-ö¼ò{0ÖŸÞrf ƒß±Mæ?~ÃæŠV¢IçaØ:_ãy(mþtK*hþÒì ’¾X­œ¿*äž“OÙ eúÍS{Z3q×…?Çn¨žE𸡍‡ÇÆ‘>ˆËâ#2ÚÊ×eS4"³ã‰Zý oÜ›6X=f‚ì–r‚ñ"rж+ç”G^óåG]MÏÿäá×”JŸPÐYò7ÿ :ŽPæŸ!=¢,^˲{P|Ð?$¼é9îÿ¶{=®¨enÅT×ËF9ùŠR"”w*mTŽDÅ?cìÓQ%Ísë²Xª©Tê(®EÈ}'ÿâÒÔ(iuj´C­“j¥r¡}¨£*¿Æ/•…zR'¯Ü߇âCøkrÑH<ü®‚•/òï~CR„¦q‚L-ìêhŽÃ!› ¼hqRBÕ]hZò+§ÇJ,£eÇ3þ7ªzhumÆ«UŽ­9to½@ëŒø¯T=h£ü¹b²ž;CЍӳÐV–¸¡ýñ'´}ç?¸ËžíÈxÉ[ân5—]ß`ø¬zìÄûÈ_PÕfè1aREÐÝÝ;L·¡hvŸçÙ>ø|^« þ.ä¿yF‘…Ïkzmÿ„\¿á636È–Ên ºÿ>Þ‰Æý–uáç!Ț⬟p÷þÇ9©Lng:§u‹1¡âÙÝð¥jl¸ÙÚÎÍwÛ:ì?™sCgkQè j!èvUÕô禇^Éö—å÷Ý ÷k\ὂÕñ¸:m1èºëòùÎ!há|s0ðÈ_ù1%ÙAlÔº/Àç{_ûÉI¡É|Û0·i‡¶=d÷Ÿýæd›ßBk¸:›–¤3´nsï{=±mWÚûHʾA»„×–•mÑ¿¹æ‘ t1_z%š&=:ìKö· oóÄpè½§0°[Û:Çø= KG «šmÂ(“ptØ]7ønè¯à~ó*LŒþÜ”|ž?òê·nç7Ãlí.­ÚÞ30¿Xçïà,&$N+†²·©S¨ú+XÕ½ì1_ ëŽtOÞ»ÏÁïö#9ç…ßßnÆeç?!‘ ìż"$¡ši”6FÒÍ_Ü {"ùôÞg«Û²2õòíÂSHÉÒEßq·}é@¦Ò¼è¦¤@zª5§•!ðù5ùg^Ȥá¦é?Ÿ,òiŸŸsDV“âœ:d÷îϪóD΋§ègºü‘Û“beÏ eäµj˜cÛí€|BT¤Én …ˆËÙè÷(øò†cd WÚD\oZCQrñ§¯9jPü]§=¿(JÏæÅgU¢4_þÐ‡Ú (Óªúµâ )Êὤt×ðÀ¤±°¥d3*$iŸxÌJ§£‹óè‰PE.ÌWïªz°ËÒ’å"°Ìäþ–¦A«G«¥†£æT}¨År$j;Ð?~°ÜŽ:ƒÙ’N}½ž¸PçA…wshÇž QðIS¦:h’xPÞÍ΋ˆÉ†0¡¯Mb¬Þy´Ø>y:nZ-¿ÆÜIðC«L¶{†ÑhíÛ¸Ïp…•^‰tS¶¡ÍÌ‹bQÁûhûôBR`ñ"ÚÉÙ|Œ:Avµ£6å"hofwzÃNí¿ˆú;ì1A-cMº›èqT”þ8%’Þ¹XÛ7UEµ ›d:P~šÞ޾À>pé ‡¶íô;ö»Ùr¡Žá¯¿Ú餧!O iUúvCÒ:ÂFäžk‰ðå leÄû³™¦ªCÜã*2RpʆÄüö%HþAœ!¢= YϪ÷ ;N@þ®¯o”¹š h—;×Õ (Ék‹D¶ÿðÒïåÌü«çv¸’& õP|äËï i(ZÍmÑœÌßéS®f\†b5“h(~˜œéÍŠG(JÍŠwÆñìZK–¡tâcRí!jøHF‚…ŒPñÚû~B_Tq|WýÕªÊËŒ÷æ¡Æé±¥ô€3Ô²g?‘…OÌ?KNûûB]@ë„s?4¸„Ÿåk…Ï_ ª3YV¡¹‚ÆÆt™Z—íMr½W¡=‘=ÞØ¯:#FÔgœÏAwA¾Óáoþз_àL¤%|Œ5´¦ï„ÁôÚ¸C¬:0ܪzˆ¥®Fí–Žï¶/ƒ±!õ3Ý'´`<Ü4eÀ&;¥fëúÁC"iá®ã0}Žãs{,Ì …(Z9ÂÂ-&áŒÌQX"©éÿÀ°Ë1 ÏŠ“`Õö“Ô« °¾/±ÞûO6ü–¸­ÛÚc›çÈOl~uƒ¿w]݇O!‘6KD˜‰åÊ˨ ©™èø@Ñ’sN›u AJ‹uÏʤþ;î¶H{ ihbìNg."½˜þñ»f‘aJÖýÞydêŸËX%z‰{¹o‰ÂKvdKOçzmP‡œ¾%óß!ÿ^ï¼~ä›Ïk ú…ß蟠I) “'Hn\mÁ}~EÎê÷ž¢¸Üù¯OßFã~îHP(Fÿ(¶ol([wÄC¬Hü¬xpÊÁ¯\2}ÛcŽ*ì†íû-Põ­üöŒrÂ^¢S9z'ØIÆÛ¢Ö^?Q‡°YÔžVØ­ª„º}¤û¼çgQ¿z•ÿ!)6ÓsäR ñ“yÊé{×ÑTïìÝÍ4ûþÂ^äù+´P>{BôZ¬ÞÓ+ËDˤí7·r¯ •U°F%Ñ:Z-ñ]'§yÖ÷Ò96f>  ÖH¾Š@›gr}Ë~±hKó݇áM6Úú? Qüжßïw0I<Á]¤/³*Y@£Üƒ þÈÛPe[L±Ý/å ² º;|YÁ (øµ©â>uò×D·i¦!ïÙÁM÷ANyJ6å½1È:ƒEöt¤>òß;,qR: Þûy5C–ãJì·Iúœ“J—g¤Q毠î^NØÓg$ðEº˜÷fˆ´1•ÌPšÚ@ûtwk§¿tõûºU ]SÁ•ÇÒAOì‰"3ÎõŽg^Ÿ7ô˜[}y:]²?ÂXÚ/Qº:­d»V/ BSÑZ;û€|.X:AÂl MNµVk½EðeØâpÚ.Mh¡È6ÓðV•!ßEï!hû–¹³½7`ÅÈ{Úº÷˜½ kƒöµWvúg 3åÍ5+è.}—ÝîÐg”°íx6úeÕlŽšÝ'äÄ÷R“3F0rWr:ez ÆbGž„ŸÚÞŸ0ÕA¡lÊ©3u|)”åm0÷}ý‹óyXä•èu¹fK3â=å}İҠ8tBÙÖúú˜pûÂoƒ}ƒ¶SðGê¢ ëp l_ª½£•ë‡Ä÷ ï k6"éI3åDνH®åïÖ¾ê”L÷<ƒë ÕTb « îj;W*V|÷,-Çs\{‹tÞS¶{#ƒœÑQ_fdœÚú¾çÌ,2w·Œ] ƽ_~Æ #Ûö=iM¶oÈIDO£-mƒ\K–™“¯r‘§?m;òùtPBþúaŸ/J·QЧíøk {f9´Â}¯Ï}¥ F±Ú™µ ”Tò(O°ÐE©çë<³¹7P&„:¢º¾e׿l–_+ÅQô¤ê%Œ¨`+’¥c¨ÄLû‘rÔ•{î´5Сê_¨že}yÚà/jhDSÕ¾¤GÍä¿Ï¹.Ô£6óûà¦ò¨ó’îäo/êÙñÌú~=‚ê¬Uj×ÑH£í÷yµh‚W¬f|ÐLxø…΀šÏ·ˆ‰F£E÷­ò;'BÑ2³OÏh­n‹Æ.磵aS˜Íýe´¡H»”d’€6…µKr âh{dPb‰mÿ0(8Œva9ÄyÕ hÏB9s(í#®³1ú¡…~S5áxî GÉOHRÞñhZ5TÎô¹Q‹î‡2þ˜gëmP`£ÿùåd‡~®c?LM¨»âèw'êAF–ÕÃérB^|%d·ÉÁS!aW¾Áû(Ápë’]×ë3Õb*(ïq„øêTi‹+¥,Ùúm×KzÈ’OJXÖ<ÿ,EEBΜ+|hõŠg[¸5~¦CÉ|Ò…%?’ãq;ãí/.òA1;‘cØà§h˜3»J{g~Ø©‡w>AQ»‡7×q(&ö[,®æ$Ôu¡óWdÈ ¸¬â퉂oΧýø¸dH¨ãF®F)[ÃÇMñÍ2B]Véòîݺ¶%T]¥ÛüÕOît4½®$øÜ³£Z?¨õÔ¿ÎÔ- ŸFgô<¡¾6øØ¹Ø7ðYÁObN»šI3äÚª@k¨ùoûïÐ~PZ”Š¡“›“A·ºøœªü’ ÷å×”›–tðM%AÊ(»Öé[‹è{`øÌš0w²"ŒÌ½¨W±³ïÃÍ%da|ïýðñ`˜øJöýÂL5-MTqÃÌžf†'g;ágð|¯Í,H[>N‚ÅtrÖj×§°¬k'¶wÌVùÇÕ©º`­¸E¾ª³ ~å)«ÑÖ¤ÁÆÏz¾”á³ð×QÇå2Ól¯ÚUr§Ù!±ÌÞ†$%u¦þ¼ñÉš~’×») ÅŠ÷®\Q¤¾×ìubä1î¹|‚Ñàï_¤{ñý—µw2\S°å¡â@¦#W_ÄRÔ"ËCþÐwñȶK$(Ê9ÊUIÒe{;E9Šü òùí^Ûuó ZÙXÑ…®2*S/H¢È0yÙ½AN :%öòl Jm |Ÿ]ˆÒ§r&ßþEÙC«7L àË1v¾Ô¨°Dpƒä*‰¾_“YDU’¤’wL¨î{¬Ayî:jðúY<üu5ë><½ñp µ_*YÕ/GÝ;Ñ=օЍEÞ‘ßÞ Ï=Ê–)GcÅ22å=ˆ&?ÊÖDeËÐì…€¥+ZìÎ#E‹†¶sO´¼áÉî*‘„VbbÌÙ-³hÕ°¹Ÿž+­6?æl~ŽÖ?ã/¼0?Œ6gû6ôŸ¡Ít³Eò+´uJV>Áf޶ÕuAµ]HÕØY¡šÄ bÙ)Œ‘{ ò«x<ûéDÂó9D@M …Lv.t/CÁ]v*'ÈW™¿i8^¹}$ÂùQG!‡ä⣠‚N2½ž–¤/ÈCŠt «óÒA‡Ë ¤³:ù*êwìל“JÖ þý[PÇ]ýpËWš^?rQ$ †ÖüûÄéÞüÐî?Q~h¨:fv{(­XC—×WMÝmèa]éȼ=F6' »G`Ð~!:‡.š¥ª«@»eŽ™À~ hÝUaêîþšVº¢æn݂ϻméö°Ôyä‰ kvAAÑWú'^ ¡0ó^þuKÜw ˆ÷–œ9Šª#}6k†¹õ óõ(E§ìÂset{Œê_ lÛÈÕÉ-<`Ò9"nh Äó©ç§’Q±ôzµS^*ßt&•0ƃ™Å/?°¡ºQ‰Ê·ã³¨A9tË,è1jž–‹™\þZmjwéî¯¢Žµ÷»¸ïĨGgRÜøê¯ú_&ÑtFÃuHh}DÐ×òϼ#Ÿ¢iß+Jéûh½Jº'‘-Ò_3ï_EË@úYöc´²U.\GkŽæå#ÜýhÝUVã÷ø*Ú.‹«Ÿù‚¶™.îÊh[Û’Øî@ƒv¶ƒüÌõ:h÷­´²ßyíí[rš¾¢}£éçç,è ϜꖄÄrŽ!ùñPÉý3’Õ„Jzåºg~B~¨Ñy]K-ðóJ’udwóXAú OÏ3?ê;·ØœCI«媹AÂêÕï—ò¼!î@ÝÍàë× Ö Lmå}Äó’8[z“@ÒÙhʹ9ȼKoÕê¹ -ù1ð!ù²À›GP,Ô¯õ\ JœÊ<¬Éöo>e±LpÚØº/ÁÇN½PØ»³îÿë¶ùaß¼>²@Ñ© 8ǘ EMä©\« ï3–Ê1˜AÑ âøÚVÿŠ/ †Ò¥ùô…O ¼ÌoB*ò&TÄ:QÕÅuAeÁ$±¾±.T}¾lçá,ÕïŒÖ¾®×@ÍǵR©Iø¤Cu’DúÔKZU]`„ÆG©O÷Ÿ{_ôÍlœ¡UØ á#@Ûµ>«•\èH´ø¥$-]¹!ÎûØ;¡—"bâ”%+|}šÄ¤e•N[ì»<`˜&kQ88FßLŠ(çÁ ]fÝ2|óœ‹îJ-LØ}<ËkS…ym`úª•%Ì.Ú8ÍšÒÂ|9ÉÆíbX´×µjù KS£e»wÿ†•PÿWƒoþÀšéZ kÚ/ø¥6'gBø?mxšðƒ?ÍdwõaÛ5à÷w@$VôóXf@’ÂðýŪ¦Hæ?_.ã=„Ï'}Ä»"5ÇsÝy1¸{H¨YÄ„ i‡T%.×j"­Â«ÎûUÈ8JvÛ|Ï2oU-eå «Û»… '/ä -è/C®'ž7ÈGvBp ò÷öÔ]ãÛ‹‚³7µÙÝ®O»|Š~=ìûV+%â’›’öç£ÔkÏW{(Qf6ä‰Oî”Û¾-ÿ¼ ¼^æÞ •–)¯)ÊGàAw®+Vs_Qík5ÓñYÄØ·ÏF_Ê£¦ï*Qü#Ô6Px:d£‹ºRpOåÖ-ÔçvùÈÏ‹†, úçߢQ_Qàc4 ÛínSöͤ(^—‚æe«'Å[³ð´´î|¦ׇ–ÛuS±o y3¦kS¢­U(Õ²¯ u=µ ©(Ú˜pÖ1ßC›ºä¢ðü ´U)÷]FÛ8ŠƒW8X‘2‚®¥àã Ôóõ}1xöœ³^iÊm²CéR¨½RÓø$:âLù/hDEõ90[©,r¹Yû&_B¶œõ³ö%èp¿ÉoªOÔ\ËÚWIšoªÃNCF€IzXSò?ÎIEÛ·¤=ÇàSnç£SNNÐDíözöj´ÚÛ°Y‡þ„vúµÔ!­SÐá+oÿôÇ4t.ŠUŠŸÝÝa¿¼µ9 Gî  t»õ5qƒÎÐmz”>°3ïQA‡èëÎþ¦á}޾„º0ïDµ`4±ø˜É CÓË“þäðÅó »îghæTxtø["´ð\|7Ò-_³¯‡yüñ̰x_³ ´IÍ«l‘ììk®4òK:º½—ø8r +=Ufâ²"ôüW[Ór†¯þR¶¥6¦0à“¸»1p†ÝToŒ²Â¨mŠUÞ©øÕkæÓ¿“t·RG&àÇgSßç0ûÙþ²nÆX`<—ae ‹ã­Bt-°ÜúêmC6¬Ž&‹<¿ ¿äGÏ®‰ÎÂÆ’^Séüí .X<ƒD÷-×µÉõ€Ò†Oˆ(’.S…U½@òhõ½):’ßjû„Ô·W¶ñáîDYIM×H¤aøù$²éC:¦ Õ#ãA11ïãÈ̹÷žTÕ4îåT¤®½3„lý’ò¾ÆÈÁz”|‘«¸H&÷½Ý“ŒÜ g 8uš„Wm;ùÍl"‚Nª ÀÈI‰)ö)*s¿zE>¹. Š.ÜäT°F ™íïÃMpÐÕ7váOP†(¤ñÆ ”=u¢NÁ zzùqoÏ+”ÏÈb-böAE×éºÌ{Ô¨,\wûn3´Þ}ÿÈ##TøÓt½ñ­ú|Æ”jJº§å^”@­°v+ë{¨Có¹Ï5I uëþ•Ñ¡~ú˜Ø£Ò 4LOµ’{ƒÆIô‹^1hf{ùt@5š[qô/ä©¡ÅÝ®Zv.héhÇô,­D ¾ªÇ² ÕLÛ!œZ´Žoa2é6Ek¾ã„Ïg³–+×Xæ„¶áaœsVÑŽg—ÆÅ/Öhë3ûå¨Ús šÞ³ÐGû‡®-ô_‡Ñ~mî'C.+<úQ~*,ˆ? ûü…â‡5,Ÿ(>B^)[ÌÕ]UD'ªdt²˜¾–Ù4…t¥> ²%HéÔú¨wÁà[oŸõ-B‚E«Êóôˆ çeóãz1qãšíÇr ÎïЫ㆘ç#ß¡íH¨ÿlmŽ@î §S${:ü;_0µàW:=P|EtºžJêßz¸މ´_+…´Ž„îÿÇäñ•#P´WÌÌhã$nÅÑ\¢Z€ÂIîËv^PØòâþìÒ ÂùôýÁ|Ú„ºîÞé|¡F(yü|ážr”Ýll¦øØ$9&C• +åTÇÒ ŠôTgTqœ§æ¿®÷ÞBm‚Ðö@¨{rK‡Iû4Ìôÿñzwš‚çç{êoBËõÛ šÑµÐvÔìBwv:tÈÞï:ÌöºdלCŠˆ ç‚±€¥%ôͧüRYŽ…þ‰WIÿÀCŸcï áZNÎM–†Ñ'¾­%´ð](!I³–Æ;ÎYµ–Ád^È1òë¡ðcüÞ¶@º ÌÚÎéªÀ<×¢U¦B>,†õS7ÃR¨Å¶|,¬ì볺u@VGxh¼¹`½6^4øÝø=iE5{þhÖXEøuÀVGº\Jë>$ú¦6HÖé$ÞU–I×I‘Ljéò‹KBõÈyªgÑHõ*úŠá½lÜ­?LÅŒ´Â]¢¸Ö>Fa4„Ú]ÃOÿaâBfWÛc¥7qo5»c»ëGd7”Ëxmp ¹Èüã¼³—ûb4çCIä·/9ö’º Ìš‰ö£pÊÛ[\!(*eUÇ\Šâ˜’“sq«€ÙU™(s“?)"›åîŸ×úp å'%~¨9¼D¥#æ—•åQåó£ è¦.t¸ÏÝ‘ 8Qó*jŠËy5>D­u…††£M¨ÓÛ¯ÄkЀzÕ„ø~,¡AáãAIQF4ºî𧃥MdzŒÔÑ´ýca")š»ÞÌK×þˆçOÞ•??Š–bÈ Áµ -»ÝJŸx÷¢ÕÕ¬p‘ƒ-hÍÄàø©=ZÇz·û,¡¨¨ZÖ!wƇº4WžE[ÖΈ’¾çh{[gNŒáR2’퉇:6â…Ï bl÷ Ip”Rª¼™ gþO_Ò_ò&#¹¯ÈAnåõ›ÂÏÜ çhõ–|!gRY(Ò‚´kµƒ:üèýƒÍî¶<¼ÿe`µwFÒݶŽù~žûÇ9©®y!5]Ÿ Ç·Ã}ø|Yí[ìB3´,i^>tÚâ)^º&PBùaÝÅ¡ÛÐyéª&íÈ.è&eeýÓÝ lbj]v¿YõJ΃Ú:ø÷Cûn*®“к7¨ÝØTpÇÇ$Ÿ …&ƒ—.j‰ïŒSŸ©8˜¶ã{W÷Oê¸+Ù^Á)Ð|!La˵Z„¤B’õî×ã¿á­]ÓxO‘PÇý|“›Ø^§]Ï?….™…ÓÜУìåT#a}qûì<íe ÿÒ>?Õ#ÊŒ P7ƒ‘LÚìæÝŸ`¬WÄSfeLì/O3N‡©/'†¾œÛ€™ò{zì`žÕG7íÚ}Xt­6±KU€eÓAsF;X=yD˜a=•–¼‰ð½6”^”쟂¿¤?-¾,O!OÖŠá÷2Æ"ø¯h6’g÷“ArjN½Øå¤áôš•fDªž§Å4ýq7±ù‚M#ÒÜ)j—ÿôʾóQ#Ãhý糯‘©L¥o÷ ²ä['f_TCÖ¼ ” Q2d£"1TûŠœ!²cWu…ÛÏÊËî7ò¾Iêé}_ü ´q£XP JšÞ+j…ܪSëøÓPä¸~Š«á7õU’ëüpÅÓ†û¥ZPòoqˆÿ”¾<~éUŽ3Êòe¤-TÜB¹—7¤'‡PÞéìm T¤qh"=t•šZÜD•5g aë¨v3øcYd¢dé`Eµjü|ýáÆÊaÔ2:uùj׋3 {Y¡n@½]˜”ê» ð* ¡¡+ëV˜=;ËK]½ÞŒ¦štã7:þ¢9©FAg¥#Z8µào©CKé[Ž4ª\h¹é°úÝT­J;>^ÔÏ@ë ÃÇy9ùІÏ4ʪ˜mª¸ ”˜n í!£G»LYÐvBñHgûq´ód-?Qˆv£—§4KuÑÞê‚…O*Ú5H•þ G¢³óóúu'á#™Ä¬û¾&‚ØÚ¦óòâóò¿ÙÑe.ÓûIS2ÕoÔÜ3Š%èêééwS"Z._‰cW£¶j¯C|šÈâDYˆ™¾œíQš ÑÛêzY\ƒÛ•U`mÿˆ ?8öáÅ-Hï²ýé[9ݲÄWÍ.CA`ˆ±\9>­ËÑö"øÙ¹<“"]ÕÇ~[¢„£Š¨Ñáü®“ÄÄ–sÿ™?-¦G¸ïc¿ï¶òúœqI,PÈËc›Ø1J8 ^âÖÌ%\1-³|7Ÿ¾ ñ·ˆJµ8J®þr븼ֆB¨{FwöTª:9Å~%øœH¾ú¦³=T{ڔﶅZ +%e„O×\™þ@ƒ>­kêøÜü¨Ù‰.š3.l‡ÑBëòQZÇ3‰Ð^R‘s‡x:³Ž9/8HB÷0©ËAKaè3J}ßÚ—zÿ>„ÁHSFÏW.0œgè¥Éž £B# E·ÿÀXjžüEFU·”÷ß…IQû7’}5ðÃòûê´±A§ÜAê®0wm(åÒ¾,XH­­Ÿrö„%éÍïˇ`¹Zš¿ÞV/ê^­<¬ëfŽ<õ¡…ðÛ•ÞîÃR(læß5‰Î0-­ÞÓçŽÓ!ÑcmÍwù?D`õRdi’­JùË#ù`ÏÅ{AýH%üàz©cîÊJÒžâCšÓ“Ë.¦ ½¼¢˜f0#2LrØ&³ Ó0yQU½ îå¨ù~†ÙfÜ}ÒÑ„œ²Ñû¾Ø®#œÀOW'òu¿Ý,XöAŽC‡”­µP˜nzÄm$÷Ý6Ø›8ñ ÅÅN9¦ gã~үćÜP†áCY~†Ê1h— ÷È£ü±rnÎ;9¨ØT+X Ú*2¤sÅñS¨ú.0I,® Ácéæéå¨ñͧw!BµÞ+låz[¡Î«“ä Œ¨ç‘û€/W œ/&¡¯Èo¾¨ûhÜñÞ0½û:š^¹r§­-Íwï"&yzYE}doÑb+9ŒöéA´L'tW­­,C4d ûÑjVnùÆe´¾ÞðͳmH[Gex¡M€Í'´YÆ©ƒªh{ìNÌ7ª($ç œK›? ŸèÄ™òž¦A…ûõknì[PÂ9ñ=ÁŠ Ny°ÀyJ¼Ì’§x —\k›É[²ãý¸½2_Åù Jêâ®2 ïihØ: û!!ci8Úó=¤™d±œø|àç¤BG¬!¿Aj‡Öž¨ñçCãÇ!÷yhQè Á—»¡MioÐ4å{h’7ëb‡ŽßõËVTÐ囤ê©ݬùº', ËëÛ­uÒqè˜b~âδog}6齤ÛÐÊåÉ\†_vú«^Ou›Õµ(Ë‹;ëá|I¬vú–ƒßw껢ök*éð¥¾mÝy²š½S£j…‹ùðzU»ó´ÞzÞyFÚžJ{'-CÉ©îZ:è|[¹AÝ÷2ü"˜¡·ÏžÍ‘¾ù%N¥ó½ƒÁGOE4wª ô“}0Ær1pi·Œoò^ΆÉ?}:˜Îþpî2‘)ÌíºÂÿÇê:,ŒF×?Ð{ K½D 7`eÊu`tù ¬ó«DÅÝòßáÒ/àÏóéDê°M×Ýe„Ļ޿÷ÙIâÙR Âë‘Ìqý`*‰3Rèï´uA*©Õ¢C“ ¸ËéLÞ)ÓpÜ3¨DßÁ…tq´|äÇ!䨽Ó=Ètä°øÔoCd1?§ûÆ5 YõdÅöõ”"»TæVtd6r2]gãÜdòw¹=FWÌôœ’™ òuÔçcÄ" XýZYäÜ@!zÞóì}(üûá|·m îÛž7ºÈMNÐÑá‰Äf+”¼ø‡³ât"JÍÞu_B™Þßi’•(wRÐo~Eå©dtOo¢Bþ§V{‰bT:ÿ6@©}U^¦³Îhe¢ûÒ팬R„Õ/íoÅP#£­Êèb8jQlùôL¥ öíû&2×PW•¨Þ•£>OÆ£ôÜûhÈq7e&ø6ÓÛŸ²ŸC“•F›%4Ës{TØ|-äævÊ %Y!Åp Z6{0ÿºˆVaÝÂ7ú_ µÖbÖâ‘´þé×~øH.Ú„ÝW‰óˆA[A*2ý—h›qü¦|Ïy´“îÖÊtgG»ÄÆÛDµÑ~o@Û«ã—ÑþzÜÙS»jaÛáÍãèP(;ecrÇç5|ØœxöTí$äÊ[<4­™ Œ)>Ya¥f‘ö6R?ÞÉ~vp’ƒ§ÖD!ÑÎ&Z(˜â)¯}‹&ÔmºÍF‡œ«!Úñ™qXï.ˆ•¥³1é†÷j4_ý|éTCö¾É× Ç"êAÔÈú0Cþ6ð¡©VEÿ)1¡îJ>ö&Òér‡+û)à&á|Œ×ÐAÇ_3='y ïoÈSž#…BñþÔë«7 ÷ÜX4,‡gâŽçï‡ø²4ë„ 2û™boBðG3÷‹*6PBæ™p{¦J¿q ®€òáÞý’áPqÁt£l¡ÒÓóXÈÃh¨ú˜q#_eÔ¸;űÍFçc—Eò¾Ÿ…úXlšÞ$‚ÏŒ²iƒv5ðeø¤}ÇU{h :—4(íº¿Â“fl¡Sªé°Íº=t›”D ZJBoBV•˜B|ãRR^烎ª?G6~'Õ,#Æ rF30&WUñþÈeøÞÔµ7{ÿS˜x-Øg£ø¦Ê5g³¼ƒ¡ÛÌÖ áðók3‰*{,ØQ®dRÂbÇ}ßÉÀ%Xvyô±úðX¥ôšØ«à k=ô,~ ðk¸ƒÉíW$lŠéèeŒÔÀßÒx¥åu$Ò7L8øÜ‰{E_Þý냤Ö'OÝDòÛæW(|Ù‘2óüméýí¸‹Ý“¸\Q÷ä¨× 7¹ ÝC¯¡– Èà‘€v¼‘É"‘G°“Y®ï‘Z»âЬâŽ!GxÃ¥*í$rëµÙpû#Ÿííg\( ÷ÉU¿á ]RÒÊV>ˆ"Ýê'Ó [QÌK?&-&%•ç[ß>€Ò¯²íåQ6Jèþµ9<Ðoa±‘ñUâÖ×ð *?½’rææªO,]VEõ/¯/ÊD¡FhºîÞ'¨e«xàÍÕ>Ô‘bœîsF=º9¾ë¨¿ýd÷¾ˆ 4¬\Y=Ãk‚Æ——‹¿/ES.i³?Ñh–ýÔúu¥/ZplÎûGòf'3ÕÒžÓhy'ÞõÒîj´;¹¨æ0Vµ6ïÙÐÚ^IVŸçZ]¾q÷š Ú¸X&›ôé¡MÇöOæ¨!´Õ¨² ÝîGÒÉŠ-QJ9Bn²Rœ~ª{¾Êl[ C±ÌS gÈ|÷#(nä%í¸OMÐ õMb]:ÈšâŸÓkу Ae^Sê ‚þÖ‰¾Zùr¥,»Ã]ˆ?ùÆGlš RåŽ5˜5úþãœ|Üä^»jε—)Ÿ¾'Fó‡è†è¡ùêÙÂÞxhm¹èã© í¼™©YÐqƒM£§š:çTì9ÿ¶@W°Õ±ÈÀzè\aßp$:ž½³Ù$€¶šÃÓÉ×í UÀÔ^ZÎwgI¹æž—ÐëPq´š"ãô®ý‚¦¢u¹ÔWä„ãæòƒ©,øÂ³·°Ä³¾|M!c„æÏ_+“‚V Ñaþ÷FЦ¶’z ÀÚ_¸²>ЄNV5ŸK ³ÐÕ½ðòÈ_!è™g·ÿä¼_o‰Úóx¯ÀÀùØâk=á0|¬ç,±å,Œþ?²Þ2ª«öy㥻KABéŽD:¤KTQQA1[$DZQºS¤)éN¥A<_<ÿÇßZ缚lî kíÙ×|î=÷5–eÑéB*0rZÅaGŒ5‡|´€R˜Œ?Ï ®3›þ[&âçaÎÙ_'ˆÓîî{÷–ÎhçÞM…•W'¤}%eamÂñ]ÅüvI»Àß [Œ-o!ÑÅÒAÙ2$Þ¸~UdÅISZ¥Ò™E‘üÎ[õ²ÉD¤¼c›ó~‘©3ú\-£E:5sK3O@Fá)—k>!³ü:SÛ4 ²\ûÊvÙ& ß³^EŽŽïñÑA—3Ïn}uî.r¿©t”ÎâAÞG1ü×j?cæÅr4î|yók¸x= Ñ–uyEâ®Ü«*î&(Ò®hpô$ŠÅ»°Þ|âï»çζ£ÄŒæÁe‡4”ráµIä B™':>'CIPnÇ„}ÏM7”Ϙ?ôàU)*QhKÚbAe.¢'6—PUc³ÿ3Ï ªgÔ¬IA0Ì+¹"ZõìK¥¸¯üù)ît ܯ1v'néê,©êÏ1s ^›Ž†Ô 4ÍÛö.£Ñß„gò h’p»ü½Ï´C¿4VÉÛ°ÙèHE™vò뮥DfŸ…¬´Ë߮ک@š» m–P-|\™¸Xux|pýÕÓ¼ð‰P7š­ã·vH±Ù÷ ®Zú½Ñ§ðÎ¥Ue3t¢¢E† 6¾Cä™î¨Qwˆvb²Ñæƒ÷×{ò´k!EQØ´[>ç÷È‡Ž›@Æøó²‹•ÓMVVb¨¹í×p*‚ê°ç¿y93Þ,d“CýùâÏí9p¡ƒ”4ÛsáâFS¬ »ä—oÀôW‚ž=Ñ„é“U$þ“ïåeÈrf"ÛûdóšIÅÎxNœûò|˽ØNg¡Déñ;Û= ûÄÝô×sÛ~&*Ú÷m¡¿br({y? êgÈ&³tÃPn³IgÊg1¼³6rˆÆçد>W€‰=o‡¯5ÃÔý±ø{ó`Vµà 1m<Ìé!1q,Ì_f§ûþñ ,üšñ>¾KaÏ-ÏÐß…§¯´:'ÏÀÚáMì[Q°A¦UÈ® ›‚R=[ðg=Š9iþ?t,¼"Hºo=²)ø.’‹T^ IŠhR2Î-vî&°çÇ`f}ŒtÌ÷ y¢ÿío·ÃÑâà~•ݹhI¬£r -ßzõ—h¬¡•ì›´K´Êç‹nпã_ƒòM߃;Ÿ•ê«ü3rž_õ×¾ ,ÒÏ<°Uø»¿ñ(~ >7nýä†O“gx¯CJRm_ñ·ýÀòø©ÈxglÃU×% ïxõ7ÓO=$NV?Õê‚>'Åg'‚h; ‚Žó%ãT¿új+õ…T£œÐtXéèRÝ4§t2S•è¸édI9´«ñN~â m?ýÊ:sÕö5mþ1hÝ¿‡ãÑ*/4GS5ך‹CÓn™Ðh‘]ÐÀ#Äž¹íC4æA¹íƒpÿ^øÛís!>®Ó¶×¯Ò |çèË5;õDïs6†Î~OUzžï,4&‡8÷T@ÓH”è/‡,h‘:–šÔZ—I­¶À7Q¹œ½cLÐÁÎbÉ©º*çAõçY±©Ðï}ù­âä †0^R5…á»^Ïh¯±Áhk<û3a˜x½Ã=ÎR ¦çît‡¹ÀÏ‚ì=ë-0ß¼WmFì|!u$VèC¾˜w,šV!§’¬?eçwP°ƒMFêföø“÷>/Îý%[™ZC’­T¡|9$ërz'¥·„½ÙÙó[š1Eã±RÒ¾<Áÿˆñ(2ø½5j|¼…L=OÓ6E–wïÞI'"Û­”^ºÖÈqêŽ(²$rêþY53rFna–4?^"ÇiȯõòÝR îtÐèX+IDÁ/Á…ëüô¸Ë¥}«hPEvóùæ•D1F§+i{P\Èn=‘Þ%Žù‡íÑ/À½Ù©­|(cøøÈ.r”­ù TÍ”‚òG?W5—¡"ELçVÓT*öº>ÕñUÄ{/³¢ú2ª‘7¨Ù©š¼b½µdÕ;/-ྀ=.‡ýP{U§ädNê|ð,‰¾yõ_Y­I²@ƒ #¤ ²nhä1ŰM¡\A'p¤éL¤¡Y™¡S€š§Šù¯Kآť{ÚÐRº{þ9Åi´d;ðb2­‚¾i_HZCkáÃZר'Ð:ëæÌ(´¼Â¥Á´)Ôxýäl/ÚÊyæxi¡íÛ·m"SthGb·Xt@6ô8ÃË¿0êÇ÷tù2OQ³»\ Ï±vi,ŽG õ¢DÛ˜‘ÿß}ÉU½lHÒçÕžò¨ƒ„}O–E|& \F“äÁÊ1EâžÙCäÚúš§ê¼ºŸô4œÀqãYš¶ÊÇ·û('Ž¿šƒä£§>¦Yú{~'I> 2”yP±Ü„,}£q Ⱦ#ÞyQ8èß<¸í>çÃw¥!ÇÓs…r"hûÜë˜ùàkgÖr•Ç@¶‡áí!ú¿ç„‚2B}j~çåRÈ ¸cÜG Y"ç#ïÕCÖ/²¾£ªmsÆ÷D£2äNQ‰§Ó5A~§‚áË“*PøõÄ×9€âáD »1Wǹ‘OùB¹ø3Á–â|¨v§¶ <•ö1§l# :ôâ©‘QBÞМ°˜€FñüZcqh>èf¸TÄ­1Gø?òR’yÙB :³‚9/;C×ÇÖµ)]è9ð¡;tÄú–^œliWƒ.>Ë]ÝU0¤ÕßEók†¿ª|ßqÿ ŒÞ*°ˆ™?ã/wï%ƒÉycÕY‘Q˜yôùëf ü´¡[âN¸ óÜÑQûaá…Á^Š 2XÚ›_T<š Ë׈oZÕnûà ²DXgú¢Vk!Ù+vTÁהΠ$Ö~4xÔÿ;’,lÚ«<—C²j¼Fvð R”ë-û‘¦$Ä®V"}÷ÙõNd:—ÏòI¢Yd'ÇF§ÓMá)÷}äp¢ðõ*CÎ\ÊÌ µÈã!Eµ;¥ùãž=tžAÁ½Q¾®Yö¸ ›ÈuGPÄ7iÖ¡![ N MÂ=‘ļ{[pïÓÅuM]”IÌWrF¹æK9¹d¨Èz4Ê[/•; ¶+n ê×Ä nnÔx>w8wèbŸ¹ZÅžï¸/])Âóé8î¿e¹’uê'ê’ËÙÉ(‹úzÏU„6‡Ð2äõî4*|!UMN¼µÔØBÓ­„…`”A³ á̇cÐÜ’(g£-d†èlSÑ¢‹ÇÊò2Z^~e3(ЉVtD!üwxÑêå—{vm´æäX¬™çAë ý³‡ÄÐúÌýpFboe>~¥b([Šñ³}f…“áïO„$üÝ_¯~£_tBC¨÷AšHîVÀ§¦—þ=K9ð±ÿØ{þVŸ9½•‹ê„¸°Ý»'EiÏ«ÂáKŒø«+µjP-þ¢ýZ½5A_¢Y…íÚ q øí’%4«ý9‘Ù-w²µ¢Þ“Ak—ãKuáh;:W“ m$Ïw½Ü!-sB“Ðì—ž3»kš¤X{XwÔCƒ-ƒÙ”eaݰGt×- î£ô³ºBísÞ?t©vBmC€b¼^ÔVE©ûxL@‘Pgɨ?ýÊ5À/‰‰}$W )Ø8¦Xnî-ð‹†V*ýÄyÐæu©ýwô«Ó:¹äAçqbª·ð]©xhÞ³ úĵ’†ú#]ŠæÇ`(.­îÛ`0Œ|+¼ö6Æ蟆«²05ü(kºÃ~ä\ðÕ–‚ùc;äÆ`ÑcË£M–o8ˆÝÊ„ÕPþàajøUz9Gΰ6ÞI_ì¼[-æï{% ÑSÞò$1ÀQ.•d’‚4¡±ZH¡È™ Ô•Huª–(&™i®LÕ}¸…ô<“b6&È(a3?yYt+wu!_ÀQÅ4dÿ-û|lw4Ç/ÆÅ*#WÊ¡–11fä‰îîÛ†|uêšV:7q'Ïʤ] úÙÆøGrà.:¹ fï£(\|¼DÒgE_íxbiZ€»ß~#KLéÁ=ÝÅ]NŠ_q¯©•bÖÝ4”!îÞß爲~Ø2ßΡ<ÃÅ?MQ!•©×2æ *ø£sðªÜŠ<¯Ñ ‡j“âµ7¶ ¦kÎ*iiÄ‘Á ‡)ܧQâeà„ÚŸcÖw_[B×$UËÅ ÔÓïOvÉDÍšO·¿£‘d©Pʤ)š0þŠ®ãGÓÊNþãËMhæ@ާ;Ñü¶“âbSZè‡÷_'CK¢^»·o¬Ñò“H„›”ZÙøú™‡¡ÕÏÊò wZ´ÐröUFб^ë±H´¹õ³ü¦'¢Íj…•„; ­Ë›3{;жBNñ”)¬´Ë^ÿÖDà¢z÷N}ȳ1 îO=íÏ=[fç•È”Ç 9äÒ2oXÑßó/hË >|´,BÍÞM}Ñ‘1\ׇ÷çÁÛÔ¤=æ¾Þ¥–q€•¢öÓŸ¶“L†XÊ«ì³ôðã!nõOð):èD^€ ¤ßy÷+ü0'd^*l0 )¬ê)ÖÞé·ÿæygŸ‹8˜Hx>söt«…ã[‡qÈ•ˆôO>8Fh²få9ïvÚtL¾´[*2+Þ}æ@ QÏNU^„L)ñOa» ³e€«# ²ïÐZ!è¥ÿo«úȳ²«Ýó ”Œê$>SCÑÖ×>FëPRœ_òq¿”Ýÿ8(h_^º®ÇÛ4n÷¡‘'Üv‡ª^67II¨õV=öd¡áºÉ\Kïvý)ÆZ’èÜl놶§#¦}ÚО•ëÉÄik öÃð=x×…ÄxSèÛ¥53›-§ï\É‚ÁÊ'4õÅ`X{ªÐ(§FVï>¬9ðÆI훩¾(Á¤]j]?L¯†Rÿ| ?ž–7-€¹")…ò?3° ón8w K†%Ø aÙSvµD›Víx /ѤÁ¯Û·º®IÃï?$~‚­–Ü(ŽË²H´ZFnVüIb‰˜¢Þô ™{¶ÿ-ã¤0Ýû4(ž©¬Rmwš!M Ö£ÖkHOtî®2— 2=XË;ö™cë­^GÖ'>g%ŸBö؈ž Ĭ¸clšä* rßîq‘XòB¾ØƒU‡tqg4} ·d& ½çu QoFá>Ë|©SõþÌž€â-9aá(™³Sø–^=Ê(|ýá`Ër&3¯eØQ!@ëR–Ñ!T*mé½ÏLŠªâ4ͧ_ÜGõµ0KÿûmˆZµ¬Ì{‰pŸ‹R‹cä9Ü¿w..\NuIó tÞ@½)¶†¨t[4H:2eûá9÷08S^CÛO~´¾hú^^›t‹ͤów¶¿âFsk«kᗶЂ2`@ß'->­¾ÃŠ–æ9u$$9h9v龜¡Õ…3ƒû†ÑjéW(õ&9ZŸÓZ -¨Eë>˜Ñ$Ò¼ zàÎ?_‘‚¿þשÿù”üó'ù?’ÿ¯ŸBbîâá÷îgÿù(DÝ‹àþêæ÷/þç³ðß:EV~1.ðEEíõðÇ“PõÒ]^Ç럿H#EÊ•?ºÐT)>[–ì -´%$Ðz`&³&ç´‰gñÍÊAë½æÔ¡å-†f}gúçkÐ$¿ÊHã !³œEËÛý$iÏ彺ÜåÆÞ)¨£v4a.úøÏפ–od£}ñÔ>‹ìPˆäûwÝ>'ÿ­÷ßúÿ|MþóY9JæÐt":Bc.Ø,A7gÃeI˜ƒžjîpoc IYZ9y¿ÌQDé“ÁpGëÇüŒù_>Ù»&Û_>Sÿh³{/Ä»…¹±ê ½~÷aa±ÇäJ5¡.ü5>»%}V戦-˜am®Û’ÖÉ~3hÄ´¿¹›N[»Ïùç"yúlÑžÝHs’¦V3Iý‡.?@òà+Y÷Ùy‘²|òK–Àu¤9ð(›Ö8éeBƒ®s¿A¦½åknåÈ<>uá°+5²fšŠ2"{ð÷] /ÃqÇa³Ÿ›[ È¥ÿ¨Ô)y$ 63o"Ÿ]ûçéE(<º|¬§•v‘î@¡ÌGŽÓ>BaÛò··sP”¿1a>Ý wóX2$™8ãKbÎ绚Q²ìÇûB0@éKFVòùÁ(+£´³Wåòå¢ð°¢‚ÿ`œ/*Ñ™øGôŒ Šèœ2I¼ªÝzòè"çUÔXV¾ÏM#‹xÒ¸{n*µ~ÙZ³Ì£¶;±üä>>ÔJ#;Á‡zDºGÆ£¹PIùÐñÎh80â”ûå,—g0õŸDSŸÍ5C’Y4ãÏJÚ:œŒæ/gf½ÅЂ;1@ä½Z4÷&Êh¢åu¥µ‹:h%$´¢VùrúÖeRhm¸ÙöR6­Y62¦?¢QÌ‹—¢ Ç9;Ú©¶¢ùíŸ÷)ÈRý5lÁד&jÔH”²,DúHbCRMõî’1Iß Ñ¤AòŽW뾬HYà1/Ωk_>9'!ƒt¢çÙc¹Óa<ºŽÍC™Y%SHMev¾ÂS‘mŠ+íTð ÜÁjRô¥ß¹BïHkð!oŒàF(e$îdó |Ç‚B4n•ÝOPX”%º„hE/)L{‡â·¨DN¡D¯Éê}¶(}ÕÌëÇtÊ~ ÝyÙQ åg VL7QIT/Ts†U¸óä5¯£ú!ýýò\ït§K‹à>feJÝäeÔî|d{cý&ê$݈nF½‡/Nð»Õ ©äX¬œ=‘‰_~V€Æ© Wã»VÐÔ¸ðÉÕ‚I<Ði,÷ÍÅ ×®ÏJ£yŸ@†Š~ZÜmÊt{!ˆ–âÝÚB£¯Ñ²¤úÇòìg´:ÀJ"‘*†VÍ=`H5‡Ö&¶Ñ_ùΣua¡°uÏü96Ïï{žJ÷í}¬ûèäG˜w1Ÿ/…,gÚ/êõ!ݸÍï“ÌÝí¾å´•^·¹y†¸AŠIâ³çMíÎù±'žV”ÍÒCñE ß×ÿbÉФͦ5Í®"÷Ü×I¶}œuýú#¡µ´bÇ'…×ðÍÌ>²ae:Nº§Æð @W Wih›ôøn<ý¨ ý¡!db*K0x-ÓvŠé ß<@#ölƘ͔ûÒõ`R‹qdé嘙³'íu€¹“…½6ïí`áè³GèÎÀÒA û·_`ÅÄ·R¥ÝÖL¼Øó×aýÌÜÈB[Ø(ÕU›m׃?gõIcô7‘Ø\ÿó53'$Y=^“Ó€äâ¤üDôýHiíš§Û‹Ô UTüÓHYJdè°…Œyº5£*ù¡Ý{%Idµ9/°Žì‚8ˆ‚‘cžaæÏÚw䬧™êXAîÚÈö\^=äý£/nÇ®‹g¤8ô¾À‹©Ð…°PxF¢ðîS‰ÇÛ.£ÈŒ­£{>+Š$ŒKô“à.Ž=(ùÀ›b=H¥µ]žË3¡,zVæ‘”K’'+K’@j¹Ó׬Pñ÷EæÉkŨÂÜsüZÝTsßQGÏÒŠ½ã§/"ZK+8;X¡Ö"ËV¤%jŸ¢\Ћ@þ+:/ÔÝÜuÉGõRúðÅ ;KåDHjÐ8/“§ï±(šz+/ÚZõ¡™à3ù£mh~dÙ_Ù›-D_Fdr¢Åø©|xu-£Ïç==_‰VÖþéY;EÑšXéÆ4[ZÇx,¯¡ƒ6©õZ·£ÍÈwó–ýMhëGË›v,í/E Ý{óÆÙVWXÓbŠ8OK঱–ì»!Íúë=÷!YHe8Û“ö§’ ‡cM$É·ûI²’êÃŒsWsË­àý²/KêõˆU ‘++€¨&éT:K~u×¾‹?]!¬µ8ÿ0¼ ˆØËÙób¾}H?b‘»=ç»ØBRoÓÉT¶AÚ­Ö—ÜŠW ÃUd1ä*dæ›·ë+@–ÌŠJðîAÈŠ{3¾|<²ïß&êý_$ËËXY!p›\ Ÿ,?dªNèîƒ.þ:£úîÛû­îs7¤™ß½—"™R¯Št §• ëòÊ8›LÈQŠœÊ!òÙžÃã¡G žÌ÷Ät 8#ÀÓ­½ J“™×rf¡¼ þëŽñ¨h0¬žPê„Êþ5²|ng¨9zASdê­(,Ä›ïC“F=ÙŽ¢SÐ<ûò‰ñhÍüSpûö,|+°îI?ó:†w×8ΆB·Ó½u+f5èé'*~Ÿ¶úo¬Ûo1Á {€ÿb* Ý¿{œWaDU:šÇÆÔ^w¯•ÀÄ-šÉ¥f:˜fL|X³³™ÉµÐ‘sZWjšðÃ|z¨‘Ü]iXÔÝ`ø(ZK+ó]?§è`¥ã©^¦Y ¬ý"^ûè¿]šŸúÀ§Eà*$z¼V40ë‚$’Þãæ=H:¸ßðϹHž”­×¯Œ”¯Wý3?Y"uÅͱߤH·ã‚O|ØMdè*b)í` õ¡™!¶faÐØý)œ˜ šé{o~¥úJ¨Ï‚.ßö©‡–¤%M'u Åî=m5ý(4ß òîcûF¨ãS–×"¡aØõA*3áù§¯ÉKþûíÄØ^¨Ý|}šúþÓ¢édtmÔˆýzÖÿô-k>—þ´óö\TËgÒ áé/­e ÉΘú‡²=4{Î\¨Ù‡–àç=+ÑÐZÔªúò³.|ÓL­¿« ]ИªÍŸ¯Êp„Ù³†Y²—aN@¥Áå|<Ì¿‘[ç9b ‹Ê³ ÙA°´ø»b06 V¾ëé?š è[q™›7ü¶]ˆ¨D;ØÜšR/\¢D¢ Ö·ºý‘„tj«tKIÓìX6"•Üßn¹ü§6RzùÓ)/"u¸÷%§šv¤£òúžÌm µÝ–wýבi=gˆ®ÁYJÛ˜®ó![\ó£œlEäÈSWØ“ô¹$©ß~ÎG^ÁžìyI!xõëí DÁœäŸŒ% ¸«¦üœ;Ö¡({ÕµÃ˸ûËÿÆí>½î –(µ~,|éQ Êú‰:øÑîCùðjIŠã=¨ØFÚ­&âŽ*j¿7n9f£:ýRûg¢Ý6ÔÅ;ž£Ö·ðÓûì¨v²ìÔÌ7Ô UÜ?z)ãçÒ…Ð@Ýó!±ãe4ü#øB½éçZ‘Dê é)F u"i4c ظG£Œæ‡k'£ªøÑBÚòÓƒýçÑâ«’züZÖ¿ÚÑ:…Váföô­Ìh}¦ÜíÑm4ó.¾‹¶¬Ë%ås)h;—¤óƃ6•hiXÆ¡¸c¬ë<í#ÈYX©#™ÛÚÓ¦iÉðÙËìÅ8ç|¼aéMq€”Àmi,âש yÈž ¹OâÖpXÚâ=ÄÌ*7?*ƒ·Ï ,4ÃþÅ÷nvõÅÎ&RDmÖùç´7î©'CYÈ‹Ó/§´ákí±Õ¹Ô¨±i(|”§ uÃj<7Fu ‘¨oI†¡é±ÁðØÍÝÐüºó¨ý"´00|=ÍEòøÝ éÄ ÏÏ9и çÚ²@ ‚BéÄ¥ .ƒ_ÐO‚Àk·Ú®?Y!ðᩦ;GB­}ù ¦ÌA-‘wy…NÆÿô­*,ñW Ô³ìÿ`Ý1 % öÒÐä¶ô'¦¾jûûÞ¡Ý[Ž|êséü±0hÍtôÛ»ß$‰%긂¡ƒm¡†8Áºt>Ù^ÚŠï¥b%×ùe /zC%œn fpÕï uhÛt À¨ÒVÇæ[0ÁJr×ÕM¦mì… *eáG«)«ñd?Ìkt§Š>)‚Eî´½‡a™Äêu VI“"¸Æà—˜Ã9b&ø}«Õi&жdÉ­~HÛ"Q ÕVßE6$¹$hd+fd.ì3Ó\‘â"SRÙÓP¤jT‹%®ðBZÆùÞÎ d¸\R›û‘©SŽ–é²Ü‹jihF6£7LÂmîÈÁ¨rDq¾Ö…ûe¨ØU¢Pà…ÊãÅž;8PMSXåmÔHÕkïÿ°‚¸wy•¢Pk,ØrÇ>6Ô>¶Ó‡Ì"ux¦_ðAÝ_)DcP$é6†×¡a•oûïëhœ-ÛÓ(‹¦É“ËEh&fÜqZÍO¥EÕz“¢$–?lëGK¿¬S'Þ¢åè ¶[Zåòf±\\G뇰º#wmŽÑ“µ¸†¶ê“çìÒ ÐŽË„ýñ ÚmÌ3Ý÷8öcG|Ÿ£C×(ÛdXq ùîìOYgªúøsá³Å†ç¨"¤t} ›­Wƒ„1Ý¢(U/ˆ·u¼Z3ï¯Ö|ظï~þÞ1r— [ûë!’þ‹ð»@‹Í¦ÝpˆƒPÇSdzA¥+>Å@ÑS—^”xd@KàÍ„o®ðQísØIcøÜ¿ëÕ¬¤wþ–a÷"LŽˆ‘}×ä 3X»û~¯;dOð­2éør"úÆ7éÿE™5þ›» z¸¦¨¹"´=/õ ÛžHÏüÔ$§. i³éµîû« ]‹žÁCøAç\ƒ<7j!#¥èºÔ!õíï}ÙÔÃ.,ZÚ- 9[±)ž'¼Tzôö¥AQµÄ¶L(JdKç†ruy_»3PaÒrcR‹*ø•t»ÕmºŸªy¡î£\|ÕŠ?4~>G,Ϲí»ÂvBZ…¾ nÍ(À7æÀ)š¦{Ð!ú~<†~•PwžéºãÄ ß‰ìc‚¾—2/2ÍÂa &ˆÆ/͆v:S[;ÃpÛ¨Ù¹k0ÚäD¬Ýât«[Mz©0uÇV4¢f…Ç£…£•áçïŸ1çE]`þf]™‚;,Š÷m*lûu‘nuR„•¾õÊ®ú?°¶î¬þµ~³&—*Áf߆ÈJ3AßlþL‰V†"ñw'®·ÉAHzëô÷_M&Hn±›Ø”–d¡y ZH}{ìDgXÒþºÃv"CáÊ‹²;ÖÈôÃÿ7ÊW#KrÅ^G=d Ü{mÀ$9¦ë ¹‚œËz¹¶î÷g0=Ž¡€ÙÕŲz|\3ÉNwEð´•9© È˜êðáîß/å 7¢„œÎNý®&”ÊýtÙÙeíÝ—ÿ £ü¥·|Ë'ü”ËÈ}!K#w¸=d{°«à“lH.¾çK+?ñÞµ5Z{!NƒÕväT,Äžxsa7×NBý¨vN¢/õ^x­dCà4ïÌY’5Í ˜×zô^GÂÓCƒnµÛ2®Œp]'Å|€,Ä:X¿c„/É„I¬)·ý-_·÷unsVƒ]ïØvÿØdKƒLÕI‰ûÊ„˜úHeuŒ ¿5) hÿ/fVr´Þ,eßöiþÕ`Kà¿§dôo­ ý˜Ý¯ÝcüÏÿUƒ®˜ sgÍÿ!¼?Œ´[5# C™û±ÈóBÈX´IÍOR%¬×à3 ¢9žÞ&ãT/ o¦.ðÛ…SPdPø&=J&&ˆµYe¡lýŠõå½4PAÇÑ;ù%*ùJÞîa|ÕcI9÷AcU+çÝ5h4µ§T{¤Íó»}2?BK—¸u@'´ð{dÝ\…öªãÞºü¢Ð%x¨z'=|¿£™àGD };ó†}\`àÛ^æ08mÜ~½ †cJD_‡†Ñ7Úln«å0Þüã߀)­¹›Ä3=Á;×àgîÙ®³^0ÚÕa–æ>,ò4]ºÚÅK?#*¸ªae(•ÊžÖ~«Û’\Ù¿Q!Cc?l30I¤@"UÒëÕHœÖªpÕ¾I­ýcW‰]\$ÔêáqO¤”z ¦á­ŒÔgmí,N_EÚ©kÑi\sÈð‰¥sãáGdÒeó 5@–Y†Åa ÈvìÞÊ!2ä8*žâ-…œEÇÄì‘'úÕECó÷(@±¥º¡ãˆ‚®?9t~0á®ó—Ô9ýß¡È'×V½=º¸[Æúüò•yÜókŸ+oö”ºr¹|]¶eåÓŸÞÌ·Fy“‹ŒWóQñF)Ò‘zTžzÐkN=Pê-fœË"¡GsE/@5œ«éµ=Pë½_¶‹é5Ô›ªäÕô@£0š˜l‡¦s?Ãx¸4¡éÛFS¿34»pT¬¸CcÛX lû:4ð˜D-&A]øâ¸v„u ¹xhAÇ­Ó&ï"]DÃ!B¤N9büj*B³…>ÿOߦg´¼Ñ%ÜWÈ¥S\ƒ —ľîc0M»ž@àÇ}R£$_·ûYÞ9¿„ÖÌÍH#1“íùÙÕ’Ð^æ3rT¡:³õ¦ÍÅ9á;Ù“ÛgñCo’ÛY¦ p¬yͬ˜C’J1§yµaäÐÇË[?‹altY¢S5&Kªg½>SÁLÿ™™{,0Ç)5÷s3ÌW%ô[Åœç¦÷ÀrÞ€°ÏL¬•ñ}Ñ©u½£·§|ac©á–û¹Œ%öqdÛ"OøÈÅé¼ÇÈ/ä”f¦AÐ7wɽú²K(8(°¿äLÈgÖ ("ä)bß-„b4d†râ÷Q\I¢™_Ï %><{ð3é-Jù–ßÿÁ¼e*T’ˆC9W…§ÖcÁ¨@#çå*ƒŠ9ôN;6$P¹ð!÷‰£¨Æ SviC5✎yø"R„ñ³Î¢V7ùûÇNí¨}XÑœã߸ô:ö¤CÝåÅø¶ Ôï6yÛìµ s®6oh¡ñË\ýïŒêhzF ÿÁ«(4ÛKö†ò‚7š_:_îä5ƒ¶‰s~Kqh©ùäósù>´Úc¹¸p´ ­ùòrÎU6  —¾»Ž“Úò»_‰wÍ@;é]Œç†xÐÞ´ Û4à:\k¿¶&¡€‹?¸­<ÀCÜ.ò–]s°ä¹ú|WðòvßÅäc)cød&wðp $½×tØq7ã;—ôF yUlJåø÷¼MŒb.Ç®ñq…„¨{±1Œ‡úù%®¯¯ŒÚo—Á«aÖ Ïäk÷+óÑ¢þ¨û=N…¸5 ="H™·ýq³ >«] -qéÆMîaz5á9µ ñA2Æ_»÷d2+X­åþ¾”—ÄÿâvŸä-Âõ)…EGòE!½ÔK‡’¹ Ò™þ ówÞ±‚ÿ9ÂuÞ‘¼õZ„ûÌ&œõd„Ì–ÞÍ×F5=¢?]é_ygœ¥¿Û …½í’v–ÑP’ ípþ¢”åK½üQ_jU§ Màk/éÕ_Yï¡ZŸÄʆ5ê¨l¹ñ\€†ImЉÉÄú—ñ`/'´<ºbu Ú\³V]O@ûé]dR¶„¼4ÍÉMÙ‚ï¬+ÜÜ$„¼t`ŽÞ Ò+z<†`0K9VK¿†=Â6ËìÇaÔUDì17Œ¿eM.°„Éß÷k®xøÃÌó]—iNçÀOïéåwaÞ’ˆúÛ@,Ò—N—œƒ¥i®{YmG`eì˜Á–‰Aß~ TK$Áoy›bï/+°)°îö€‰x)ÙÝ£›ø¾¾™=…<’Š<=»$çƒd¿ÔH.P#%-sFQŽRÛ|5¨Q;€´ßÈ2CK!FšWé2µÓî}·U‰,×ã]ºm‘mߨƒôewäPˆÙï8‚œA_è*†Äç襦e—1ä/ P¹5‚À”Ø{ wí«‹>÷–EüÛ'3¹£Øzý™FÜSÉóÏëPJ•¸ÃÔ e֥ɟAya¢·A{PÑ>2?ãª?*—YÕÚc‰j¯ý>|K/BÍþT¿àKj¨•z®)°Cµ¯ìÑí4G‹µ+ß²…PO×|ϸP<ìì¼óK W_d–%DãªÜ†ÜJU4}Aý¤k”ÍÎU‰ˆBóäVK¦;h‘( AY(ƒ–Ÿ¼W}ttЪ|þeêq´}åõ9y'Úrš°¹¾ŠvNA×TeѾøøy­wñ &ý2¹Éüš—Y¼[ù*ž_æÏê¬ó9ge¾ÂçjoF~ƒ'™.]˜>Iœ {ê< Á#I´ðø:ÄU+W²²ÅC¬R°[®R/Då|à¿aNȯŽìIêÿb̬JÖ¹ã¾ð‰í„ É•ƒsªëüü¡N›Ó0å… "Çîv8Žƒáqϲ¯,ÿ\øMª‹íŠŽ§¡qìÕ}… h(‘rq`*&pWãà%‡ÙíþEí5fâ‹:IPsÖ†­î…ÔÜî#Ý邚Éà³oäQB¦íYêÿômóŵb¢q¨?ý¤0´y÷iG”퇦ˆ…’'\Ûçë8÷“­AKì.µ®ÄçКN[*ónm{~ðfe½´‡ªÏ‡¬¯B§ÇSÅÂ"kèŽÓs=ú•z¥ÞmÁë㉃¨a°ýS•/áçÃ?ïPlõ˜Ù9¿'B"0±r³¥Ed ¦»(95:“á'mÚ®¸tB}¨~íl™˜6,²:Çx¸À2‡ïÄîón°ªå}ê¡küŠOxu:˜ 6lcêœiß36£Ã:H|ÊY(Þ”I•Óv™N~Br!åFò ¤4 YC¤®½sHãåO¤{ö+îì"d YÔáU€ÌއÞÂRäâÛÏQ«µÌÌø¼jÄÔÙ±ä#wx/ê.žYWHÙƒúßøR¸¡á'æšjhü0ϲ[Z Mݺ_O„ ™ôÀ–†w$š_i°Nô@ —$вûhiÅ“*¶…V&ÊŒŽØ¡µ9unø|6Ú8Çu®CÛk¼ :žæh—PÜ¿~¥íÇ—5RÌŠð Fµ¼4U(:&3ÍDݺƒ‡•ã‚%yžÁ¢¿ç\ ÛyÈì{óS?ãù¶ÿx„½$ÞŸWÑ#>ïÍ;ÇøM# –òVif<+ÄÈ2”òZZAÔxAd›Djt¼£§¨ƒ7²ß;¯oÙt-¿ê¼¸vç:'„µ–Šìê”#èàÛI«HžÙŸÎÛý`Ê*Ÿý·çU…%®Ÿ…trušžÔ{rEM÷3d¼P¬¿;£±Ígç~=ù¸Íqå­¬IÿbÆ¢µæ­È8uó'_<ÛìošBZQœ1‡ ¤q-å›i†¤ÆÁæÃœû¢X‹ £-©läeÛþÎÑÊ£Nœ½i~ødÛqÝ_Z†ÜÉÅÎÅÖN(t“xÃ2ÿ JNzou‚²+äTYŸÎÁ—û‘/¦ê'àk¸žCõá,¨&rŽ}Ã(µYËâÙ¦¡!P’â•N4Uî}¸Ïö´èþÌS €6f3úñVh§|ö¡Ó¬¬~Bw”ëþlùuèU¸IŽÕ?¡FHGÔ¼±ëQ©? ƒWÛ ;ŒÊR©ej{À¸ã÷·\—wÁd–K†CÌ—Ý4óŸ §ÒÏþ~󪡹EY_aáOú1²õO°4±Ë+áÑ"¬Lí¸ò`¬mdM>îc†ßbÞ5Ÿõ¾ÁæuÙ‘h72ø³Ån+À׎ÄGóO+¤ßC’•8¥ùc‘¬®û³mW4RŒ¿ÖÖZFjPRÛO#ˆ´u·Nær=F†çÆNá.ÈTݬ×~\YN½¾EÄÕ‹l;÷éÛѵ!'ÿʾPä´ Éò _AAK©h/1ä÷&ùz©ÜÙå4}y¦q›å—B_1Éîc\yŽbz®î ¸'H}þë#Ü»ö`êOÛ2ÊLiÕÞ=ã†r¼ ¢OÃPQÚÿÜkiTã)b;/ƒjž6ËÁaz¨™óää‹ OÔz3à$©gÚ®ÜaªÍ¾¨£•ì|Wõ”¾Ï0ê·¢ÁŽõ®üh¸Ì\fÄ‚Æõ=,÷ÅÐô]Ú‘Ü;yhÜßë˜xÍ+Æk‡ü6Ñ¢9'QV-§_+Ö§¡5ûÊ>âOÑÆêà͈&´6fD{vï;Þ´èðN~«ø•(:ÚÞÜýò¬uk‰̆¼›[fJyþVûóèÜ:|¢JrŽr‹"pÛƒàÄÖFHˆÌy©´òâÅ,÷†Ú>‡3Û|БP_š¿Ð¸¥ ‘k[v½±¶Ûßèˆ<þÅ®[ÏïÇÊ‹6Ù\È~²ìsò6”8æJ~Wc‚/'DµŸ‹@eÙKF:`„jõY¨ }í§ô´êÅgšš@öìÄtFhTÿ–Í'3 §Çþ²Þ;ë?þÿ·2² ™Y‰¢d$áñ‘•½I´ÓÕPÙiÈ^YÉ,²ŠˆDöÞ"{“Mf¾×ûs~ï÷çœßç¯Çq^×õr®s^÷×í~¿®çóþúÅ? ÷Ð]Ó êU ¢ìVÕ n?³ž–¯;Ôž´~*”rjþDµŸæ€Z¡¬.E&ÞéÞr~µÏŠÓ½Ì2ÿã[==Üθê ~Äû-¾‚&mC+ë7ÕМÓy÷?´¼o"N ÖtòVƒ›AÐö‘}„÷´(Wžo€Î;ÿ~+ã‡n‘_îG _ÁOµ-uÓ¿´Ð—ÓžàOaé%ž °¹ #OÏ裏݆±‚‰&ͽ50)*Wp4j¦ÎW¼VßäW†…èÒî‹Í¯`ÉÎò6Õj¬8²dt-±ÁZìá|?/Ø œO› …­ï¿{áo›¤ªó3Üe¹Mf$írÏʇ,{^¢M” É› ¬2ù#Õ©äH½ôY%Ý®·áÃå RdX ’òz‚L1 g•\à¾k~½Ô=éÈjmÍ—‘î€ûŸ-¬©zýBŽ™\ö´RÈí¶~ùÌv òr^¦ºxKùž M–óh£S¯ûÞœüúø™ÎT: ÅdoJë4¢pÅ[~#<‰:7äÍñ(ë E‘Š¿1ÓöKC is²b”œQc#I‰Eé0 ï,¹!” ¾r×ÄóÕɃ®‘cá³›ƒðSìák#í%è+ysøöR> ¸ˆÜ;¿Cgé±LNFhôÞ³´ÀèÖŒvïµJ˜àmö=¾ Sö¥óʳ¤Ög¤danq¢p¿R&,òIÚ¦<…¥Å¤ŸaeLìB”†/¬ý®«å¯5…õç\“À°ÅÞ1S;ÖÊ?UIa·‚Ëãò:-Ëóf´@’ªÂݹwHÿÈF.ä$’Wð4•Ës"•Pò‘¢ž¤þêS[yé|®%¦¿-F†¯á6”5?ÉÐ?.Þü 2ÿ}#ý8QYz_™§ÒÁý§m¤M ÇбçÚ¾¶È#æM“MJ†¼³†-ûrzÜÍŽ£or ^rî+ÆC¯J/¤¸˜àaY™hÝj+`ùû4ù)Šÿð½N­ì€m‹#™±{QšüÜöǼA”qÉ|G‡§4*(»7¢Âk'ý¸si¨è§ÿ»5ÿ2*Rè‰Õ¾E•cf}3æJxFøÚ™[»¨NÏ-:­¤ŒK/ê&ï¡Vë_‹ç¾¨cy$RØ uÓÿ„ꛢÞÀ\…ãÌ]Ôß48dQ¸ … h™ò¢ÑµÍ×CŸÐ¸Œ«¿&ÐM¥·½ˆã5ѬîÊV`Î(Zø’¾÷ ¬Hi~ÎJ‚´/äIƒÅ*ÿpG™· —M©»†Qò¶7!!mö••6}€X÷D—Ù3{!²ðIñï¥äÿf‚jcòˆò‰ÿéY~k_ Ÿ¢†Ù¬Y¡ô Ù©îr(o4q袨⮹G5Ím–-á„WùËù^Ô‹–3~ç'äª,—KNæÐHZ^Ö”¾ OAÙûëÔ‡•p¥x©CYGŠñ Ô7ÝÝ 5‘w~ZP¥CÇ`ÈîìÔT»7¯½nƒZ dzÏOQ8§#(ÅýþùÆ;q>¼þð?ûjœ¸d¡É>ùŸËYh.;S/òuZ B§w©, 5kÚ¶Nì´}&ÛycímÉþ ÜÐyæ;õ‚¯Îÿ¬GŽ˜¬…ž©‡uNAŸÌÞ¢oH`€\šñ—À :òn_ Éa‘ždãæ€1þ |$Ëv0qaåM}=LõuÛÿºÿ f¿[¤È”pÀü:¹ùò'X¬1‘4}Ø Ë ïø:WÂêÖñtX¿q¢ØV޶8.pe,~‚¿üÝî)èó3J‰¢D$±¯X’í=d§“Û» ì‘\ÏüèâX#R¦uh·°ä#µ¾óð„–Ò ¥2Kfj"Ã뎳4·o ÓqïhqNdÞRë¦m€,¾?iàïÅý‡Ûšûf#˜ãfï ä>RñU©¯ŒíÎ 2M!ß…ôK͹*È?)ÐB’…‚{äko£ÐUÊÖOCö(0@“d)ŒG(D²R¼ePlŒFõÏý(îgÅ,÷å,JHJ? nGÉÞ–k•™(íí±³·eë'¶µð lUù#¥”?•ÿ^d"4^M¾e=„ŠUr‰å/ÖQÉÈ(äømB~cÑýr:ëªÎ xêXx£á‚z/‹qµËÌŸÍPëQñõî;Z¨}±¡9÷™êJGV$?^@=¯€û w«Qÿaºä¸ ><>–nd‹h¸s~Óô<å¥m¨ ñŒ…)1šÊ*ߌqc%ä­ j<îCîÃ~_é{ÿYïxÇ7¾ ò® %m)¨òšÁÏ܉íÿ^§x Lþ³n„†~8rR.ß:óOn{Cúz²Ÿ×f?‚ì$ý)ÉC!»¨GÜLqrlž5‘Ü«„œßùAOA%MÀ·§>ðI¾¸úú LpŠ2IŸb×ÎÏŽyAiýÞòùCìPö'°xM²*¨„µgºø¡’£žåzFT;q«"§„:&>w“ÖhøJ´gyó4ëÛ.Qý1…–œO>Î]ƒ¶«îWÎfA‡îù-çG Ë…××'¡z¾¼ì_¼6}êº1œº<öµƒ;†X·h=ò&`øçO|ᣕˆ_pÃøXuÎÊl%L1{X4f¾ìÈŒì߀¹¬+%ßý¿Ã"e{à%RX ÓÉ¥+Ã2ÚÏɸþÙ¯º7ŸÊÖÿJ¸q+õÃ}Áfì q!3?oÏÞ§·ª|èÛ¸9:)$‰h¨žùQ„dÎ\ô¼:ÍHÿÎJõ1RѳrK¥U"uæ3±oìG‘îçØh+2d[¹ÈSA&…‡ÅŠ¿‘¹oC Yæ-²d‘‡œ£%E¶wªßLÁÈ‘@3µ™ŽÜc¿•‚·üîå,*ä/èh·ßÖGÁ)ÊÉžï£xÈâä¹áÆx˜å„›½Ø»Ï´ç=ŠÇ7’½ä|„•'Ò\Pê—Ü›£-(£‘Û·>‹§Ƚ6¯£Â c¹ÔæATt.wutF%\!ç+¯G*º{‡ ð û²u(ª“£ûþÎVÔX êÔðù‰ZݼQ‡ç¢véúƒZ"Ô­¡×®Ø8Œz«_IÍ~£ÿÌ;¯ hxágAÞà 4*Ü ]‹íG“£ÁÆžâhZÑ^Ä+îŒæƬQó%hy£ûëKhåïðîŽH5¬½7U/ø Ÿ»Ÿíåè҃˓I]!£‹¹ýèKª~gûÊ–DàZæL²®fÿ?ëEäô5§ÿyÇØä$Õ?¿cÏ)s±ÀÊ3Ñ›%Ѩ~û'kß3Þ‘=Ž;'Þ“ïwpÌ*ƒü·ŸLRå¡äiçfù”K~½éc •6>×î®@uŒñyßkJP3¶ªcÑ uÇ¢Œ òSÑ”ã–L‚úºìú§÷B½cüQ³i¨{ºg£±„ŽÀ+³<ùµQ¨ ˆ]ðLƒêMw‰·D †FȘ=â=ÔÜ0qº®GÈs¡¢ÑþÎP«;ðf„ Óÿø&¨ëu¼¾4ô×@SðFЈQ 4·.”õÙBKŹv+êhÍo$Vžmé'„FÙw¡½l³©Üã6tr{³ój»@WŒ3 ùcòÿÙ' r{ z‹;6ýÉ ÿniŽý'6ìycŽJúÖD§£P…yÇRÃ\UgwÕzç ÚÀ¶ÕƒÕ¨žòãñê'ÔrüêÔõJµm(ê¾õ^DÝ“÷õzîG=ÿiJ߻Ũy^J ‚ S¦š­Ñ0÷„˜á¯}=tç.š73y“ƒ¦¡G.s¾Ds–0ý7Š=hñ)öRõ׳xÎ5Á¸-¾ Ïß–»ñüÚ¨ÖY.,¼Ú“–ÞOÈEwˆÔìà}°¼É›[îÿ³&VÚ ŠnÑÆ\øñï/Ì{ž"píþ=Eˆ.ÙÖ¥8Ý@ðOK»æˆ Üxùù¹û× ”ÏñfhÄUYºfY±1Màž¯U¢9Äê¹4DkBRyZ}VФŸoß« YÇN5¨E@Îñ‡co§-!WØ›N’^rc…¿ùlÑBs}é63+ä…J üoææ<ç]üæöÄ«Š|0‰òìÀwÈ®Šþ±)ßÙg¤ßkn¼†lý u¡U.ȶ*ž)ôøÙÓãýYÿ伩ç× 9Ž‹áA(ç?ß{^Ó0õ‚+™ Wv ˜âÈ„jãC(ÕŒtÖ˜8eŸ:ÃYVŸByõ­´YËRøñ“nù¤Á4T‹>ЏMüà?g$_1‡³é=‰ÆÐÌê'h§V-¶ìvü{¡‘‰a—Ú¹ÑùtÑ3ÞçUÿ=joE_zŒBok›s­Êiè>á"Ùw‡Æ{•ŽFtIna«ñF_g;Êq†ñôR…‹)09¶þ’ؽÉ4Û:/ŽBÒ ï£–wIþëEH´¡_?ú俞’ûÂGy*åÿ›ÿö)üÛÇðo¯Á×ë‡v¸f:à{Œ>åî®=üh?òü°ÍTõLålϸ$ðfyË-8%íÿôŒüûwímíÎÕ?m„œÖ[&ð'ª[Ô*•‰K Ú2ðÙ·Æ´ÿúQþ=_MqMz÷Àÿ9ï¿ý) ƒºÂñÎ/ éóÙïÄM y4êsÐt´4 |-¢-†Ö¢ìˆâìhKQQ¯ {íuY¾3}…Žu!=æ$è2.µvf> =bìs§Þ-A/Þ[Z&‡_cš†dêå0èióõ AG§ÆüÉ™ÀȪa…sÅgk¸z]y# &Ój³Ñô éáêêoðÛeŸ¸Ä,<Ñ*]òV‚%¯±µñ‡aå³ZDöü­º6-ýnýéa€íŸ–œ7Yÿ·Ï$­«ì`ýW$½Àº N÷èÕÕÏ<7C ?A¼âŠ{éØLm!Mñ¹þ–ñn¤L°~ðØ->Êšn!3ÿVÆZ&dö—g`ÎB¶‘>Ô…"È>4ÅxÐÿrùZÝ9Cà›ÊVßä£É6ŸùuùýöÝ´Jû¯'áàú'Ï/äy(ÌaâŸzý/Ž8éÈϺŒbòßS¯æ£øù«Ñ¹ƒP‚ñ„§VJSé|UAéKÄéˆ2F";Õœ¾(Ñ´vo`åiX¤Þw~GH3×R8ŠŠ_òÖäÞA%Û[Ê8P…)}[ŸäªN÷Ò\§Dµo†€q 5^y…OK1£–mÉãÎ,cÔ¶:]Ù(j€ºr ¾§Qï•zdËݨ”nÓ©² Ò”ß=‘y‚†v[éz—ј]µÄ8ÒMîŸ{-¹}M—zy ¢¿¡ù+I~án?´Ôë¬<ôÜ­äChö>Ek³Ø-ÅWÎhÓiù§ú5=ÌÛ|ª`½œ ùü6ä" Ÿ!#ÐY†¼þOÿÏ¿ºùWW‘…nFòšÌÿé*Œ&™ûÝ*„œ×ëiðùoþ{üß^“uû¯žÿÕwöô˜ç¸!þ×g’«ÕÄèclð{Hþ¿ùïñ_ÿïûÿë5ÚOáøî÷æ¿ÇÿÿÿïßÏÿo?JÑkâÛöPR#,˜”â e÷¸¤?KA¹oöžÙÁsð#R¿…bcë¿ûÿº¯_iߣ—½÷?½¶põ>bœùŸNÛƒzÙžœ†Îw$W çs¡»ýóÃ#Ñ}Ð{n.rý6ôó*HEý€Á÷­i$©z0|wèxðže=â«ÝUû¿=&ñ/%ãv/ÂÌaYÚêá‡0wøB-Û$,t¹\¤GXªrQdS膕>­ƒw\Tam}2 Ç7 Öw3ÒSÞÃæŸÍ?Ñ:†°½ëò«vY%HOød#Qi@GÊõA$9µj1~hÉ8­nM•ï"¹yÎM"¤X#Ò¸…ÔËUûî\ñ-³ÂZdˆ˜ÊˆÛTB&Æs«lÛÈùqwõ¢²¯R•uÛ!ÛÝ«s~Ç‘ƒ·ê~‰rQ\)û€¼£÷‘ßÊ'[N½Í—ß¾’D¡ùÆæŠQäíq*öI%]Ë“¥æAqSK…ທ(ai_pæ›J=[ï?²w OlÈ=y÷4 eÛ?ï».u¸Øé‹ uQQ;9²’$•Ø©>@å-[‡€zGTÝÙ»Õç…j+^BwwQc†xf½<µ†¯ðÕå¹¢v¿qÐIÚ3¨GgÍ&xõ’f Jö¡3OB—Ä4ìæ–õ]¾ŽÆç)§g^rÛÞËÛ·¯¢Ù°·ˆ•^Zü–<›eëˆVŽòE ñh]|þGþW"Bžþ˜Ç@w òm\›iG!S“©ß¾ÜRÿÌ=N£³€w·”kO1ˆÀÛTš»¡.X÷|³ö$LÞs|†žÏå:å—tˆš±Òªô¹á–{™ßBø…’&÷‚fˆ‰{õ€%Ò¯ˆ”ü¡.IÞAO~(žYÌïÜËß ãçÌ„¿ÁªüW ÄP¡ŽÁôÊP]nrÚ˜#jæµÚ6¡v¨ÑJTç%ÔÙð>§¤W„Ú•¼%û»Ÿ¡VǨU§ï4ÔÜ*ÌŒ;åÕ?¸yÓY¡Úô£~Ü·&¨ŽºÍòÊ­jöiÊÍ6AMó«T™“„÷‘£aõºÓ¯é ¹š¡Þ^ƉíÕ3 D ¶û iõ–Ê«Åzh¡Ô•}#4-f<„ê~AëM´Û+´uøN0€öÁ)ù۬נS賯͇QèòÛ6ü¯=6>ô®u?¡×áQ0s½#ôï3U%:Mƒ¡©Ò?aXK&Ædô)Œìžr¶®¶„±Æé³Äß¹`¢+ûGˆéo˜¾DV|zÊ~r5mÇ”ÀÂÅÐá¯_°tΛ˜4í=¬xؘô¤LÀÚ„’uÂløR˜÷ªSÂö7B=%&°àâe¡ÄçYl\§‘”š‚ÿBÖ’ÍuΛI…"»ù©slÆHd8ôSŒiŽÿ;÷ÛéÖ­|9o"#…=u©t%2US6!Å}?æ#ížBÖÝ•bÍÁkÈ~Çì³Á 'rÑêÞ¸èŠ<åîÇ"oMê_KŽ äWjë$’nCþÙEã+·ð`VŸ¿Øñ(<ÔTïYùÔŸeRýÅøÅ$“Œ.—E¡¸Œtu Ÿ>IÑ:Qˆ’a3Ÿ®&£4Õ=Ýx€2Çâ#Ý­PöAž «=ÊÍ%QTÙs#¼è3N˜EÅ,um÷4TR©ièuC•½¿N¼¤DÕщŠ¢T+VSZïÏFGÓ¯ðí#ðò:¥I*oÛ‰Òᨫ úÁ¿cõKîÇvÜBý×ëJ;µh¡+œÝކ]‘瓲yИÓΣþç 4q¹.%>'‡¦+Ù³ÝÌ­hø=O“ CK&Ñ<[)´R°“奸ŒÖ攫ƒ³¾éøËCs¶n?ËÇÞCžÇ¯¯¹~^E¿ O— I/âñ-Äs)ù÷sC¬³;{)DM|8OwÙ"Ÿê«zmA¸Þê¾KôÍL®ËëÁíå&üZ=ÜëÉ_áîÄœ$¿·!f-Õs"HÞ.Ì2›qF@Úb¥Fƒ{99¹Ø§ÙeâzŒ¼“ík{³°rÅ2¼„hd!7Ǹ0¼ò„¹wD ï¨ôȽV#ÈÕ6U œ7ççä6‘™:d¯²Ð¦â0d{iæ=8« ÙL¥³5ÆÚÍšY~¸ú d³ï$O~9 ÙÁ{¢ÅO\ƒêë+Ngr6bOäk~¾kÓAŒÑî!µÛPTD<ÐX%»Õ…O·m ,X¶Ñ`ö”¿KÑ8]?>3ÿù­xªI^Ëá³Pë;>×=¨ÿzÜ“q†¦š¢C͇$¡å`ÀÐÝü.hýºvî‘u<´º¿®qvƒÎ7ûþ0žô†îJ Ý3ƒÞ3U]AŸ~B?±ÊÒ ¿ Fœ| úh †-ÝTÇÏŸ„Q¹îÃ6¡—`\–žV. &Âë ÎIÁô²’æ6é=ø=Ñš›¡ ¹z…ùõ°ôfQ–ýÓ>XÉx§ÊÁk•]=”ž¬°^Ûký 6Ë씾­Ávõ1Séæuø; üÙwr‰’œé]I ï£õ¡Q$#ΰž qGòÓœa´HYÃn/•†Ô®l>0ë‚tZg•ê?‘ ÃëõÒ$¿pd¢1hÊ\wAæ0¿ þ1d1r§ <óÙî¥ 9Ò"ÿž­íŽÈm‚s,ìYÈëÄkçÕüÖ¤E\”7PÐ'ÃIíG -kýêÈ$ðM5Ÿi¢EWS޲\ø‰â&ÌB6n£„ùÉ·Çô8PÊ7õ›Ëg{<±’“вýd®Í¨°orkû: ¢Æí¾Ë¶µ¨Ä²÷²yI.*¯É”7x¬¢êŸbÏ T››(æÑjC1·0‘ÈB~;¾÷’7#j÷¼—=ˆ¨Gôž¢¡¾h £«-8i¥i¾ã$ðîbv o£F·/¸¢)¡óLT šåô>}ô-[‰îîA«“%íÆ¥ÏѺ"–æ·ë¬Û±8pVòÙ˜˜µEáƒI¦ûiH•žŠ{ÿQ’êž• €·§¦_åÏC¢êJWµ'$Ø÷­ŒêùAÜ=Ëaì™u¦©ÇÓÓ ÂÚH)yþ®BØà-w¹/áÍöÝœp‡´·á­ª=Æ»¨É´]7ÅwŠì^ ߆ï{Ú>0À™‰þUYN¨ÒüêÄÕýF ×ü æýáÔ}š.PûîóW´£P'š ØûäÔÖyMRýq‡Úc– )ß² Æ4<†ÃM ª –1(ÙAµaóþÇ9«PkÇ2ó™júÖ|£Z fôšry¸ÔÚ†Ç{y_‚º+ üѪ9P#ý®Ä75‘ëŒ\€fö¿¿š'@‹(iQÊ%2h=«ä¿ mL×bQ=ÚIëžfm@£ç¥ýLÐi.4é ~ºŠeÕŬ¡'xI¼d± zߺK/½ûýÊ—OlËZÀ`:ç"ãAv‚Žü?ú¼È„Q*>AÍz"k¶;Vú,& ËÒ#jaZ›\»h¯2ü;xçN|,¨îi÷zKR‚Ô¼=R°¢åÂB|!ÖJû%ר`ãØü™’O»°M6 ¡»Ôb'û|Û·×6Zl?’ä­ùV€dχ+úÚ<ý¥´AÔQ¤bùœdhO‰Ô)çg&VäΆ=«®ÀòŽõÌD¦ÛZZ[«¸ï’ñZK|²†ê.=£x‡û·«­Üµ©‘3z¢„åGYÁ\ý• òúÄ'Öš® ßN°bbͯRœ®ƃ×H^ÆŸsÇCÞJn£Ñ(²;ËAsEÇ©Çù¸£8Ùš8G:/—}nÁ}%í/=™¦BiNSS§®Ç(Cª|cLeÍ"ò?‰C¹‰ŸÃTœ ˆ¦£âÛޤɨ$wø˜“ýiT!¥çb÷DÕþ7«u;T+MÊÍp@gÝŃt,¨uÝ@_vgµÏv{CjŽºx²™TÐõ‚¤ŒoI £~˜Î¬Ï®hðÁ:¬UT {ÊËâ»Ñ˜;\U…“M9u‰ *¢éšß—?chdpææ b´4Ô>­:‹V ~U:0šÀ7ÇõEßTîIfÐÁoG•E‡«É;dyWaÑ Òv>$<—«ƒ·ÖA^#ÌgJוòÞ¼—!»ÿÝ¢ðÎ_fŒ€ˆW¾::¥#ö+°îb-„Š<™[×Ú‚`'M˵¢ìÞÄjö€ï¸Ù0~b|þdÄÝà…·ëFßA§÷¡?'Ý!3Ä"e??)d‡?â’"œç# ÉueKï»ö£œrƒjšïɃÈQDk!UB2ƒ`=tQt_÷SØ|6$<- Ûc¼ã?æàoê›=ZBHäNCoÕ IÈIÂ!éĆÁò$.¦á@ÊùAe–oH}-vÿ§o‘N¶]1G(ÚÒXg‰B=­¯{™ƒmV L"‹!5óùü—¾‘.-XN!‡€ŠAÊl2r›9mW8d#¯³ŸÞßZä·qïàèÈ@Aßí¥‚¾…†U.JÄøfqzÂEW¢¤>Cq£:º÷‡Î¢„)ÑK¶>”ò2K¬¾4‡'ޤ<ú”Œ²5d+¯o  mÁ¤›*ªJ i!B%5“´ª3¨¼HÎJ/sU—¨™LƒPmJK¿2<56-ìÕP«k{iË8µ;H™šQwÇá‚R6'êárnoD4p|ð±„Íà/wzE£Ñغôë¼t šÒþmÛ@Šf“ƒ/R’EÑbÍñvzHZ!õÍ…Ó÷Ѻþ—ÑAùrXvÓ$:úŒò.{¤˜¤~¦ʾ+R>²Š©JJ‹wÒP¹#8c(¹ÕÇ‹üxZ¯@-…½Qô0G Ü]¤P»Á·š9ïC¸¾~½uÑ€Z†‹5D︡Fé®h)gï«:î WAµîü€Õ;N¨N¼Ãª¶  5Ç ÛÔÛ¡ft×^A­÷áÑí„:¿)i7»P¨¯–ì¾ò¤Ÿð¬ qÔAóÉŽB‰/ç E·@!;ŽZÕQüUç…6…ï D!¢Ð.A4[>:”¸ý˹‹¡3 LÃcø*tÍß“·þrzªÎ³¥Bo‡WwËuèwt^?ž ƒ%¿dåö™ÃðÍÞŠO'za”™¢]º) ÆZ–˜Ý i`âGÙ×o=©0-C1 u„ ~S¥Nº "‡…x1Ãu9 PÁ Ëë6NX;§p.rkÖ»»$þ­(>…ü‡ð7 ¦“׉j¾vüRE’óSÒWZÜìðo©ãÞ H.W.®ÁM”‘œWës鑚k`Éqsi«:ºVå‘Áfþãs1KdbzYo»o÷I\ÓAÖÃÂ5GGpÿ³“¤2·“ýØËšß›Èý=IâB°ò*G¨É†#_A†ª è%º5ǃ<Â!̾WñЩ‚$ù,(’ûã·!ÅapºÿëþE<Ö’ªvTâ‘c2äGIÍ? î7QêçÈå|/<ñ[s\^“ eOº±wë¡\nÙ‡íÄ6óS:V"×Q1Ân{oâT:þì5õ-#TÞR²$NÚAÕbéî…¨–ÇDn÷@5üž6Yl±¢Öµ˜S7ßEmkGs2»>Ô=}dÉÕIõ‚Ó%eCÆP?‚÷>—ƒdÝ~õ(Ù {ç#_£ño”E–châöäááVU4]·ög#]Góà¡î瀖F~üo :ãÖbo%ð­o†ï]o'ómOÒÂì#Jž‰xÈUì­&4)-ƒªëóðöÊ1b[Oˆ}°!ÈT1#¿ç-=^Ad¸Dì~Žtß|®eºYaÍ s©!äólWÂýÛÌØb“«þ‚…ö»°5˜CXᳫa+İi3U¢#œ·ð—þõZ¯CËËã™XΈØ²oÿ˜¼š2 9בqèW-ä,^¶½òrÝÕ’X2ÛJ ¶‹§¬Ô¤‡õ L…5 ³°idtúu²l›GM‡ƒ¿¬¼‡?Ø ‘ŠçÁ¢.$žñ—Zü…¤jªëÒ^HÎ@vñ(ã!¤Œ¶ÑÓ4Ô†‚Gƒ›‘Nøæ|^¦2x±N|x§ˆL$൷Йe ²y‘E?l¿˜ o _T ‡àÓ4CÒä¶H¥vªÉEÞ» ’CÊÈagž=qýu%ŒQh±D~üU owhÉ&(PtùÅ~‘y97|úç¤ñ”0ªv!º‹R”Þl ·ñÄÌÛ¢«ÍI([~d4™ÊÈýÚÝ*¹PQqûBÂɨDí¥¾Ñô•gÚ¯ëÍn¢êl€À‘;ÏQm$sÂn$5~:Oœ‰D­ÖêÞ`!yÔn6÷ÙMbEÝg9õcЍ/’Ú-Æiï ŽûWø¶OV#Æ6чö² )ƒø•‰]y4›¥Ãúæ~´Øbü+|øZ©š‘>+¿ˆÖ\Ñç¶“`©6,Y¡ªr³‚ä#ëŠ!ãÎXùÐÈ$g å®ðÃÛgûS·#Ù!ag!¹|í$(ÆÝ¾ta â=Ò6"_Û׸¸‹!b'oeÇM B¿¯,ÞÿÞ¡¿ª†>JPCL1vS7¤&ìh”B®tws†•Ý×$¾²Ê\cm’>BEtÒñoŒºP™|ZòŒ:TMÎñÌT†šƒ£'C~kB­Ø³Ç¢çí öGbúSë4¨ul“ßb‚šEÛ sÿ¨3¼ÏHò ªãßûåyªÏÒ'ˇŸ†êd{Ö$‘×P#¦Ë³Ö µ4¬ûŒk¡6¾Däî%.¨{ªn~êW$¨iîAcÝÈëͦc^©œlÐât«´¨÷´ÞQª»A®m¶ÜoË>C»Asj*t8~þhÇw :Kö¤¹&C7®x¦ÚcèÙ¬uRV„> ÿGÍò ?á1ÇŠÈ ¶Åz\!€aW±î±A åšóâmS‡±¶»ä8½`¢¸¦°‹Ã¦ù¨ì?hý€Ù¡™Æ©r°@ã¬õÉ3 G.°<© Ëé¯õ:t`õK/q'¬{”òz‡ÌÃÖ±Kæ¬ð—:óñêeß+01¼pBÊŸ01¬Œ¤Õë&±b:¸§~ýÆ "%×Õ5VÜd³þ—8iñ,DÚ¬§Ki‘ñsK­X72Çòg‹õ Ë7rkJ>ÜÏ–’ÖË0ƒq¼Ê¹…È­DMý0\ y‰^ͤ¶~B>'&–‰…  z4ÈÔŒ‚:‘>9g >ñ¨#o£#Š\‹+£#ž@QÍ^ªÈV5<æy&aû˜¿+zݯ%·¿f³6 Tëàãàgx¢ŽÈŸ6se÷_p^uw@¹0=6ù£0¥Yð@_êÆ¢Ò¡é¥·×¯¡ò²sÓ„øTm£3œØAµ¬?×ETæQÃGt¨c’ µ®Ž ¼9‡Ú6bR7Ú ®ò}«eÕ¨ÊËH¾’€ú‘Ký Aö#¿ƒ¼-hØ¿ß7¸M ù&ë%b¾¢É“îCšhºyÂô„ë^4u¸‘yú!ZWØXìZø¶ýÍæí^´¶83öÄŽÀ7ѕش0ãÙÆS=râMF öm@ê)2’0fHlþk¯#² o¢FœRݶ †3é'*VCÄÜ*p­~…ð+¯Õ/HçAh7Ãý¿½TrÆü›fâ}4»²UÙAuz»øã!̤…¾ü+D§„nѳoÁ[^Þï:ôç ÕuIƒ,>Ìã«Á‡ò^Ûý€å¨q®ù6ÈiSìÛ#>¹—yÌ'yCîêrUiä1rﻹªR—?o@ΟøÌÃü6}nOdg_ÜCç’Ùg¹¿vÐFA¶e›E;dÛîK6Hˆ€ìYí¸9~9ȹµ|†ãÀ)Ƚ´¦zUö ä—-ß3½‘É^e×ڡتbÏÃ¥b(}›þ€{‹ʦ_¥LAùî3fE+)¨d°¸õ†©ªÍ²äërþÙßÚ¶:Ë1 ún«ãµ9ÐLä©%ì=-2…kFzÐZQ‘òÙvÚýíúŒŽC§÷ÙT6¦mèŽêšºã]½d†BŒáWÒ9±êBJ´°§]K‚aý§ƒ×(a¤ç-sD£Áw2–W]‡‰Ñ¼bÇt˜Ö9Ozõ:ü>¢0ëå,¨ž~˜M6K|æ‚|<Õ°"ðAðÇv¬‰¿ûz7ÖE¥ÔžP™Â&w•…ìþAØæ×ËSÏꃿ§hÚ]é£ÈP´xó©·Ð·n üDҼГ;ä{pÏŸ»{²J!¥—ç7ëá}H  É×\‘Žu«R8=ÜnåH† ãŽÔù“‘sÈü‚^+Êê²èñqmeø¶×ÏëXrü¶û—órŸëãàËGÞûf¬ìd.ÈéÕ~û£(ø,nIêú'ZŠ’¾1û”À7Ï‚ñX]r§•’žAqƒSž{Q ðÒä® J¹\»R<ƒ'&8›ô§ß¢ìWƒö/ï†Q~×ì[É]jT”/O¸[¨D^&§Ñm‡Êã‰_‚ÝPu‚ZrWëªõ3¹±sAö…b®YÔj©8¼Ö‚ÚuÅf;-Qw5‰Ì”ý*ê KËy_ODûÖÒ¤m߆ú^iDã ÷—ú/ )OÿÕã÷Ðl^nqqë5Zìf¿|WZ‚VZIÑUuÐú—åpH±,±œÕÞ“9›‰-¿¨É ÝÕãJŽÊQx7ßKò9Œ§¬ßLr탄;u^ýlFÿu¥@|jâ÷2Ôëßþݲuͧ-O3‡^q…Ü;¡qòok»,4ß[Õã~Ô -á|§+ÙT 5|ïËŽŠKÐò Ùóˆ´¿ sŒ7ž…Ž·ú]ì¡sá.õ[Öqè6^Ün?žñ¾ôúd_ò¼’à…þæš~€ÁIëš—K]0üüÉoIê¢îœ…±öŸë#Û0‘Û’ÿÛòLSS“ÝòÙ"Ô®~Ïó®¿tÉC`±ˆXäÁµZXN+×}ðŠ V[Nš_¥ø ë‡ShmIÇa³âÑkØ øv¨-Õ‰$M®j)Ä#qNºí2’^#uÙÛ€{΋ à/@Џá™RY.ÜK1´ëO©‹4á:Âê¯h‘Þ¹?=Gc­öU¿=SƒÌrþtË+—‘ÅøB¨š› ²½cÓ£tì%è‡aCÄår•9ŒG3ãGô¼â¼âÈ'ÖâþÌëò79–þç >wŽÝø%Bï’nÁD3Špº„To'á‘¥h1Q<&5º]+ƒÇu€}¾$%vInÔ|G©Ì¤­3x"õÓ‘æOxr]µ=T8åîZþ‚jÌ•£¨èš½‰J¡ZHIðÉW(?‰ò  —~°¡ü˜­„eUÖ}SlK÷bédÈáYûbFÜ9ùÜ:I{¶^ñ½ámÈK¯½F y{³Ówß„\åv\zo!çw8r¼‡î6©¬%Büf[Ú/ÙÖ ¤žñÖí8!yc ÛÓ!™7òäìe8+fß 9/î}Ja’…\ßë‹N±ò?Ë"!¿¤•»>c¸ 8°=Û܈JjÑlëÑÁwŠ,¥ª2V¨àv7 •'ÜŽr?5›„ê»—_¿ èé®[ÚÛÐp=Õ‘FŸ 𩾸±´(ÔÄpÖ*ª›÷ØV Ý—9þµÆmèt³>igÙÝÏmö(ºŸŸs¢ [ýÂðËûÒ,«b ª6åëÕÊÂÐâïpÉRm)ÿ”P ci7D‰âSa"¿ä8{“+L³]ÃjßË0;ã¼zp½x_‘}·„Å?-Üád°¼êý&Vÿ†ÒV8PÀŸ?ÆR|ßacâ”ã`€'lÍ»m$µÂ_RR¿‡v$:¹o­Iý ³ßåïBÒ¨QâNÜÓ?–ZÉ©ˆ”·SÄíÔ©oæ‘Fº=11ɃÈpOšþSÙ2®>yõáMdöï÷ŒaE÷|µÄ@àm›»r횤IœCnkáV—ä}Pºž|(ù/¯ °)Ø£`À¢Cw -ß›×p&ð-âòÚ¸Š.:‘€k Šë-Uë}œB =3–Wü(õðºFž­-žQ KDÙÏ.MJJ(¿.ü~W`O>ï'OD%â¿b‘C$¨<ä@ÓnF„ªC*mB“¨Ö}©ô‘j4Ý‚ÙeD­j›÷ß«®¢vyûOÊxÔ]üj<¤€úBe±¿òÛÐàÎnÕ~;gßÚ(ÌH®£ñEýMçœL4ÝGk2z) Í–l“Ø„TÐ’Lÿ€€±>Zé/J”VBë±h‘z×k°xåØÒ9凣š{Wó,¤ùsSÏ_ë;%¼o“â¶ t >ÍÆëÙ”Tî»3»|cªL µ¯ì@uÜ›–@¨É·‚.æL¨=Ÿ´O V¨¢äÎß·P“{¿ˆÚà*ÔÒPøÉ@µŸÚhzyT«èJn>zÕ™Žlçbx¡ÆP7å“ÁO¨•ü‘½º þo4çªb'Ô-ÿ¶˜’ƒ†³Gí-®sB÷\YF%4‡íá ê{-_:Ê› Í¡µ¤ÕàôS/h+mKS¹×íÅ?÷…AGÛÌN…ptñÞÊÍ>ÝþbìZbàç…Äü*kè»TfGý»ág9üap‡±`˜†cZ‰U%¬aT¼hdíçkët+ùôÞ&Ò{rHcNÃÔ*MpÜ u˜ Uû-—5 óÉŒÇ÷ÔÂbP¼dh¸>,;õÑÆ«OÁj@~W=müi{¹yÖ±6Í'¬wYBa‡º¥9{Öv›îR÷F#ñ¹býYÒ&$ÝC}ß­Éæ/ò‰¾{g2Oº •Û%ò}ôDH½»WÏñÈ*Ò5ž×}rú.2’˜V~³.B¦îù[F_æpßtqOR,²Iø¬I·×!{q¦9ž@.Åî¯ïc7gå¯-]ÔòNΈíg@þGâO-¿¢ rTÒmJ‚Ž,9®Z‹¡p“áuÏ?ñHVº“÷4ÉÇã‡0öÞŠ"JôP² +¡Ô3[_Ýp<ás³ñãƒçx²û°”TsÊYÚ:k7аWMùË[T¼'BÜJû2òÛl^¢òHOaŽšªV²Öè @µ”ïGbûíPÃüî;oW\Öˆ¢UPûBâ̩쫨«¶Ç|Òšõ"¨ëòØQÿMØ‹‹±Êhîò€åßÖö )Ec¡S†|<7ÑÄ3ñ_ª1šîŒ,ß½u Í#_.¬® ¥™œ˜sAZÎæÈwAkKŽ^’¢ißXœ¨¨a*ôÂÙC7 <00¼z2’»¯>X°€„Ô7:lömÝ_ÀôÉÈ¢˜>’—ÊAxìÃÇBÃ&­ä”á!­‚ÑGö@ð%É×1ðú÷Uûçå÷!ˆe%Á¸AB³)%n…@Tµ”æãšH ™{®vÎRúS‹£šßÁ%>WЏLø¸|v|Û²7 ^3ANØF™Jl äJß/²%1†ÜŽ «§í ¸9fHÑrOç¶MiCÎRô_º:È&)q`‚ì&‡@n¥h×ÎiHH B¶ÿªîÃÑmÈ~Óønî5/äj]¿Uê96Ií9 ¹é½o~ðý+[¹Åjé‘5°S·©ÀÓND\;âItHÃMOÆ×ޤǧPâžJ“³L¼ÚHiRïšy ©Y .†Ñµ#í §ÃÛýÈpGE Hq™—)Äb™}Jný?öÞËnýŒÊ,QæH¦ÌCØ—yÈ,óP ©Œ¡¨Tš•CBJ¡h”¡DQæ1B¦dŒ’’Yü¯çyò}Ï{Î÷ü×o­ïy×Ywk}Ú×uû³ï½÷µ¯½ïí¹¯¶z`7VŽõy‡÷oÌ<ë(ì€KD퉥Ap;k+ž¬ |G6}”ÝÖ{ª­¥Ë€—5Ç‚›‚@ø§…°ŠÞ¿=aù\ â£nSúÅÆ eú0½½³dLø”_ò|¹Àuâìý ÐI?%Âs”3î<ëíLÕÑé8hø êr›ƒ½>€Æ/ÞhÐj‡kלvz«TòhÐkøpß½\ *û¸+#Á°X<Ú>fŒ‹œŽ–oý ¦ÃÍ©'ÓÁ\È„\‚ïßÄkú«ðþ­Dêämr°r‘p™<7 6ìsïvGÛ‰kU³Ip }zç[8ÙªUÙÁÎá¶”ªúh438Öã¤Ê/žw¹p ÝÎy°ò:ºëQ¯ýúÃ7”˜:|þ-º³ý¤‹­º}ÅnjǠùWdƆøy¦·¤„©/ŠaZ  Œ®‰v½ þš†®ÙÕw=ÒC7>ÞkÜ„»€Ry£7¼KÎC™Ò†¥GÐKòÇtâoPÁí“}|·Qqz2j•€JË:5ß3Bå çØ35>¢ Ç#V=§PåaÝÛâÞ&¨JlcÅ|:+ªüuímˆ'ª¼v<ÛU¡U éªiTÌ Ì®›P…º÷ì¯LT‘éÏ‘ÓäŽ*wŠ[’ãÚQ•Þ-ã“þGPÕxàdçÓ,ô–#î$5 7ª O&Û¥„ÞYt¦ŸšöCuÏÙúiÔôQ}çE¶«>¨áË ö'ê)èý äcª";Ô8È}ÕG 5SžÛè#¾µXm¯)1Ü‹Zóö¯‰ûþµE3´Ð>ÕEí·b\ú&Q—äÝѯ,M¨‡ù“ݧz+ôéÉFö€a¨okRTO§<úÜJ–0„ú{Ò8;xÐ@ƒÐº4Z4äcv)-+xÞÛÇJ…CcûA-³ìWûY]Až&te­eX/¢©àCý[zšÑÌÈfÏñ&4Ö{8\ ïßTX¸OªLõo¶h OæÌï¦ʘ ‚"û:å¡oÆQ ùÑ·ú6{&Ðï) Нè†M?”é€)õFIGÐc`9dUùüd°…¬#+­n‡µo&oº(¼N©.('`}\/MêïáŒaöçÀ—lfõîÑs£«mZŠÇ1[ð [æ3…ï:ÊuDNK¨¢¼ ¶›ùuí§Ÿ ËŸ¹G¸¤q ›‚LÖš’Ù?An$«ˆÄ-Ppå?p²c7(å0D5½µùYwµ‘I¨ruõýyv<Ѻ I¿šÃÃ1oW«%c@§PŽ£äè%§^¶{$'Ä8uŸr‚¡kq—Œæu0ÞíT–´á˜êO;?I³¸“}X›ÀüöÞkïl&ÀâÅ=?þ«ï`{_Lú©‰Ã`%â²6v§XŸM£æ¹a6‹/ ¯õi‚Ý >~3)p°;t,ú48i†é%¢t¼;Ó”¼îÞ¿Ñf‚'=úš ²»qð"J]G7ºÍ¥0q<ã>ŽÙ<óæÙ´P<.i·䡸¢Êߎ ë“jWE5×í·qAu6“ï–Fõ_ée'«PCuäñnÔxÊiÓ„ìgÔìuÆnð?jõñUýt¡¶Â.©ïP‡žÿÞ Ó—¨{Cüöæ.¨'_èM›O ê ôš¯§žFŸí×½<²]õÛÔ»l»À‡~)Yi†=28sÛ˜önW|Ó ÌßÙ‡ŠT\õôuJkÎ0`7,ÛÓ¹ïߨöö Ë׿`ŠÅ]MÀí¤+½9øŽ^Í/ðæ÷6l-?`ã•køé?€ð„$Ï4…5Þ¿æ}ý¼ÄGl‡¶Óþ)“ÝqûV§‚Œá×Ii© ç_Á`^ä Îг*ÝåGåw§’ê@ÅÙ“- .ÅéÉÕÄÓA,Ê?6ƒVËš ›b*ÐiÛs¤SôjÔ+”¸D™Üþ †ùÓÇ~ìãügϼDÁt`dzây3˜ ¶¹]1S o»¦Åí|xÿöÜî‘JzÖ„«ó„‚ÓŽ‹Î?«™`ç$gŽÊ¼(YŒ ž8¾¥lŒH©hG©I–†‚®•(ùÁ[·¢£-(q•\ðÃ{”èöÅd;æƒ<¤e)¡[º¡Ôd^ÚèÆí7´ã•èºÇ ±±ëQTRkÉæmwPTëñ»76S¡*ÁrßlQJBy©yÊx=iDò åž;Xô?®X×ÉŒ b— -V…9¨ô°QlÖÕdTVš“ËÁƒ*8mì®Ë ¢Jµ-‹'+¾£Ê©{ÇêøPe£ëƒdryTyðlFþzMTÑZ-¾°&Uøüˆ ¢ AªW sv¶¢Šì€uì i¨ÒƒgœÖ…ªÖ4ÒjQY1r) ÙëÆ*ºÁèjOÝ)u+ƒ5'â8¥î“QGÑlD°°æmZ«OlìþJ”ã\°Öxfxì#X—Ý».§ê¬‡ŸÎ”2àþÚ>[|; üÀø‡>óxl|‚_žnwê¡2jû8Ù~Q\Å83Ôbl‡ÍÂÔ>€¤AÁ9ÁÏ õA3õìÎ s™ÕÕ×ä4~Üæ¤O¶íšJ ƒ¶ø~‚ _XÍÉÄ ÖtOU(‹Ôwš¦»D &™ça Ûg Õˆ’ípWêuPÎz·/°Ap“XáM.¼[Sá20Æ.ë’hC½ÀtÛ`BÛ¼)˜Ý˜ UõóÄ­7U™SÀ"7k5ÝFØþ¥ˆõ˜µ X‰†µ1Þáëó¹Ö_Ú ¶dWŸ9Z‚]üÓ,Ù‡^à`Ÿ1{Åo+8i¹d—Åû·ïuZÝaWÓ¯±€:ÔŸ2rø”çY”Îc\tÄQÝÓÙY1²- Ý Që~‘…n\Ê(G±³!gkÕž ëÆ‹¯¥v¯BÑÌ÷TÏñ[¢¨ÆzÅ£d(2dÍî³û(B4à(wÁ±ëüžÆÀ\t­*÷¨a?ŠûÚ÷ýp:J¢Í³Vð¥ªo=7t3=Ž£ürbÇ”~Kå³êEC”‘A§H«ˆ÷[ö™)gV_BYÔµéŸv £¬guFÇ‹QÖ(CÈ=Ï`”…â@‹ëÊœ_[†_7eª:×O䨢Œá“ëÅéÃPFìÆ«¶Oð~ùüÑ ”ÑzïbÈ}”ébk6èøeŽÑÊ£¬±‚‰øêR”í¶þÄ'e”kJwÑBiå‹R?ûD‹ `­[uê^T´s•=ï®TèÂÕ¶•w¡‘ DÕg®ö¹¡jM{¶OˆÕ¤múÜg‡ê„vÁqÿ;¨ÞtÊý¸õ8j¨iÉÎA!dƒ©›ùQ³ÛõÔºkÚ6Ëç%j{”s®êUê9*¼ÐbˆºiÈ9µ§QORa£D7êµÎ¸Ý[Ž>Ëø’½wGý"mó»?ÿ@~§¦œ¿Y£!q.†¹ÌQ4âF9µ8¡†ÆŽ*Óˆ.¾@ãŒ7„H£‰½»'Ì]‚Ñ”íËãÛ=ÈÑŒVT^s¯šÓÙôqõéè—Sÿ½ÂØÓh±&^?^¡ÈÜE xk€Â¸†ŠIŸ¨¢#ŠºÍfæ¸#¯€n¨ôüÙl1XS¯Â;˜œw–xöó}Ùèú` ñŠ[`ßf,%o4‚÷oYZ˜Kô¥ýà àv{lÙ}ò5ðOÝ1Ž0}>XïÉÁÆ«ŸGK5•Ax’Ž#x—>Þ¿5ïíé›ñï†=Nfi eÌurbë91Hg Cû@Î7pôŽw(4Îó,XÜå{ߣ×p8ƒê§Ø…Re .Ö–«ªãÏ)/ÌæƒV};E\P(èÔӚ䎂^ù÷9;q`ðºKíÄ!E0Ì)ÙsQYŒsé*oû‚éÚm'æÀ\`OÅ—[î`áu®Ã¯4ïßRN5¶Êãý[ŸÂì–H°ál-Ø1ë¶s?^žKq–-u7×Ò“kóÌU£yع`»èŸG¾ß«dé@韅¿ÑlD)i#Cã%ó(©¿ü\w)º³çèë›P‚·×ìeGtkOTY-« ºicE÷üª1Š«7½«+ˆ¢ËÆrÓ¼™PW}NÓãHe#ne™R‚ââ-¢³ï £îY¯÷Wðöä¬ö°«2åÒÁÑ„ù“¨€“ÓP1Ûî•D*{ù=*q•:¹ŽÊ›åø›ÏlG•dœ;[¯ Êa™äp;TùHî¶êéTi}é ûb:ª(~ÀuØìªp)+÷kJi¡1æ ¨âåÁuô&På¡Ïã{P•Ñ™Ò#¨Z4áĻ˞è펱”*—O¨¦n3©™z—÷úŠ´».ª—;öcá5j7Yu½WØ|óLÞjÔè˵óAMzwg‡žÅÛÍË¥²¯PËK™«ŸO|Ez'iªPÛ\?çå¹µ¨ƒìÎCº˜ZÔuéuc7…'êQqý¦ð´}ª©Pk”èC}&>ã/>GŸ?Òî¥jGýWÇRtw|@YÌút#—РsõµW»Ñ÷™šŸ-^hŒó˜ù°ÐWô£AXŸãúùJÒb×MÖêL¾)E3œONé§[¡¹KÔÕF9xÿÆèí1ÿpöW½ýðóÄ.)òüC'"™Œ…Ir~žx`⮎"¬êøÂù\ èPÅy¬nëPû""Œ}§¶Xñs†À™@†+Àú°Ö[›ËØÛš¾æÏÆÃ:Õ÷5_ÃýëÍý©‡{÷[­‡ô[oÝ;1¶÷dÀï7î?î[ý¬,´„Ž[¾0Ö†Mߺ{ŸõÇhY÷'½Ø× 1·c/9~¾˜©-¾:Ú dÜ9yÅ©sAŽû…íÀPà½ôbc*(¶lg­d ®;7·N€Ú+K¥“d ¾½wÕý+Q 1õ¶-Þ²´ÞX·{ tž‡¾|Ýzñ~fï´ƒÁ‘HõÂz0t±ÉˆŽPÅû·÷Ç|ÇÇÀÔ¨³ÐrÛ˜ÝôÚCûÞÌ“W'߯¶‹W….g‚`ûסCÙC`%þìÖ¹î`}¡ÔhÝqW°¥Øëô¾qØÝ+Êàà0ì¼Þä8iCâ[s>Øéh[ÕÛ Œ÷o?¤ž£C_žf‡oAis×Ò´¯Cwþ:"XŠn·rÈ_ÌÖEqÚd†kÕP¬Œ¢oÆ|?оW´C^ ] ~¨|ÕQ Em3M=›´ˆ"ŽÆs¡ðй=ÈEPÕUFÕàÐ5÷ù¯g­PÜaþÂó[(±DPEÍÌ¥ŒQx²Ÿ*BµÅ¿RAéþç2ÜÍÅQÆ¥=ôÚÛpxu•S°8ˆÝfdDYw϶SüBYý~çOBYª?ÎÌfmCY¸F™cŒ(S7ÂáÅÝ9”1Ú募·>zÉ¥6xe¼ãM~8©€2†>ß³ òD™'¹~‰>ÒEYŒg‚Ú:åÐ3vÙÄRz(û²ß'ãE”ì–í¥†ò8v5|pDö{Ò·Öw¢¢à/T£«QI´Ë×\MTvOh ÇTñU¹ïa´ªvœ»y»¬Õ”ZìNÚ_Œê$c‚£FP½5uqÅ‹õ¨¡nÃß:=Ôx,å©jvzp¸úÄ:Ôjpf¿6ÙFa†·Ã‡V%é@§:,T :ôŠ(÷ízø ríûÓ.‚aV¸©m?sö½RÿL? ܤãás¾qqô-,<³ú¾(Öàý[Ì£…€Z°r{s'˜¢ lÖ—+uˆGƒí"Ÿç£zpXûΙóÑ-pòŒØ\ûvQ ZåO¢a¹Rv™ƒx{ë ²¿òÝ{v¥DH % ¥V¾ÉD·ÃÎÔPXý@·,³ ý·¡›ÎÈqŸ "Šß+À¸/ÃÅ‘w£WªÇP´€«ž‡ŠŒ¬zrŠá Šlj£fÜx ÅΉL?8îÝ’Xku: ePœ¥}Áb…r®ÐÊÙrÕ 7»æ{•yPQÙQ¾¸!TÒ¹YÛÿ™9*“ã^ìnòGåQLßÖ6½B¥äâÇ%QexJ .Už¢¼ÖLUŽ*£î{÷ñ Š´ó«~F©  ›[&::R¨B®ÖôöFTñúÐºš“ªøë‹l?Óõ¢ªÓù„#DQµ®t‰Þ¼z{êVîãK%¨f~ÓÖIý«èÝXÚ1š„tTÏ¢m°wçaÔ`Z=£õ½wÛ9z”5zÁ3ƒ»³¨éà×›‡ØÅPóuËX_E7Ô2”8:ɵ}°æûv6úú¸eûÖ:¾_¨CöžÑÀeÔõªºìã\ê1–Ùu›}êä² haB}¶ÖÉIèsûeã'µ–¨ÿÜL’ãíhàˇ3LÆhð×~UoÐ÷‚Ý[sÉ“ÑèØG:+ôã&Pƒ7úyðQ½«B“’OÚ$óÑtÚÈåì+ÑhnõºÑykôëðõ=NÔ€£ ü©ûÈŽ©‰ò¼ÆÛ‘Âêô  ’jÊáE°j»;Ÿúæ2 ½5;¶îv ¬–sü”Å[Œ’åÇ÷'f“О–¦uÀª¾ùîiëÀ¾WçÊ cp•œ^‡_×p¡¦´HØÐþêRèÙ_À›Ê-æ]c ü;âò£‡cAOæ¥rÖS’úaÒ´'6%¼|áóÉDëPŒ'Äãù¢:v ÓIªf ž·Ú·ÖƒìÌÑìÙò§ ÿKÓn’¾”l^ef‡­Ÿøg7šZ"…\»Y<¨ëÚ1¦Œ·+ UY³ U’|Yÿ è<µ×Ͼz16™™=À h³Ù‹CÀpwâr¥+`ìzŶug˜šÔ÷ÖÄEÙ­^­Y:0¿Ûùˆý,Xä¿sä^ÀÛ×àÚ~)`%Ù±k¿¬ÃêFØ}÷-ˆ»„Ú–C;¾€ƒ“Hù}±‡à¤ÃYÛÓfƒ÷oE¥g$ûðþí‹O7+-úœs,¥Ãûzò2ÒeÃâK”œ/ævw%<ÍfÝþKÅîq6k ;†bö}wˆÎãC×~YغNEõ?UˆB‘åW½¯ú…"nöËi¤^Báî§Î¦å¡ðG\âÛtÑ5Jß!%uGë;°cóE”xü"Ÿòä'”‚÷¶ho5z4òÕlŠÛ¥Ô‹\Ç€2Ü »N8Ò¢L°â#(³U(Å! e]CÉÊÚ­(«ÓÈu_l(ÊR1¬Ê4ü‰²¨"âŒ/ã×efŸG$ Lúð”­GQÆë;‡ŽãýÚ‰7/ç~¢ÌUš÷„}¥PfBǃ}3ŸP–Ôª÷zYôL)Å8ûZÊ~^µÃlmÊ}w™+ñ,ÊY±õ)*8˜=û`ÞÅi²ÙUM£’,ÓTµØ TVÔo5ÿ¢URŽ(û=FÕ·6:™^A5íÁåc¨N±6÷Áþc¨Þq-ëÒ߃58£-ÇQãQCzÖ8Ôl•G'Pþ µjÓ„[¹f ¶d‹¦¨ýÛÅä¸/ž¨+?î´Z«êñš™o JE½ìJ_h¡õõò„Kuö£/Uß%£n_G”Ñ*Äñvù8”OîC&úÞ 3còm‰ˆÌnÝ‚Æ Ë ‡Ð¥ÆÕ§öhòû¯¸üøV4Ý|ø»PEšmŽòööº‹æGš\§Ñ¢³ê–ïf@&£x~w9Ppíä`*×u4åNª9Û]ÙzŒ@WÜðø ÓiX“fÿ¶;j˜¶‡ø×\æÎ©£†ímÀzdC>sÓK`×sQneÄû7xX \Š4¸”0àÞÏæÒEÀw¢Üͽåxî»Ç /`cdèèš´9ž*¢9Ë%†÷o‹ïû6øtõNFÚvqûšëÚ £Ÿ†An_ËsµònP¨Òmvò€!èå¹/ÒL–€AVsD@¦9*æ¶™ƒñÓŽ`ŠY0íT¾f¡ æ¼'®Æ(V…G÷ÝCA¼ maÞ V{âiµŠ ÷ó ‡µ`GnÂS7Wë½Çb4¤À逯H‰Ù{صʣlìV?GIŽÃ‘/âp¸5xPàáþNvùÅÒ9VÂ9&ÂyZ<Ȥˆ‚4Q! ²DAŽ(È¢ H”ˆ‚2A —–"IÒ$I†$É’$9’$O’H’"IR"I$‡ ‰C†Ä!Câ!qÈ8dH2$‡ ‰C–Ä!Kâ%qÈ’8dI²$Y‡,‰C–Ä!Kâ#qÈ‘8äHr$9‡‰CŽÄ!Gâ#qÈ‘8äIò$y‡<‰CžÄ!Oâ'qÈ“8äIò$‡‰CÄ¡@âP q(8H $‡‰C‘Ä¡HâP$q(’8IŠ$E‡"‰C‘Ä¡HâP"q(‘8”HJ$%‡‰C‰Ä¡DâP"q(‘8”IÊ$e‡2‰C™Ä¡LâP&q(“8”IÊD i)©ß¢ôoQæ·(û[”û-Êÿ~‹Š¿E¥ßâÛÒ˜BN8DéåìéF©÷¸´¥+ˆ\”Òʤ޸$K¯eVÈ+¯—[!˯VÈŠ+d¥²òoYa¯Â ^…¼ +xVð*¬àUXÁ«°‚Wa¯Â ^żŠ+xWð*®àU\Á«¸‚Wq¯â ^żŠ+x•Vð*­àUZÁ«´‚Wi¯Ò ^¥¼J+x•Vð*­àU^Á«¼‚Wy¯ò ^å¼Ê+x•Wð*¯àU^ÁKêï”2RK~ÉãQ.ú±£:»;%…¿[ñŽÃnÎ~ËwP-ÝÁ°tZ·üF—Su&Â?®£^éK—®÷&žU?Ô¸œž#¤¼Ëéu‚~óÂršH8ž<òGšB8ÿ mYL¸î©Ë²žA¸>óÒESÏ"Ü÷|OÁi6áþåó¹„|^­_>ŸGÈ/rùüB¾{wý‘ò/lºóGZDà)Öì^N |%é|¤¥Þ2îåûËüå—ï¯ ”£bjùþJByªÜ–ï¯&”«ºaùþ·„òÕÀòý5„rÖ>^¾ÿ¡¼uk˜–SB¹ëŠÓÿHë å¯?b¶œž£Ajt9%‡ÃmÇá6;ãp¢µ8œ¸ '¯›-kq8©Ó8œ4þùep8¹jnÉc(ÞÃá”Y–Ç—Uøñ…?¶ˆ-.ë82ÜòXEfÿoŽððÔ+OÅ-{ß•ÉU%GI—ýÅed.qzŸ—sÀ_œ¡òó9èíúWùlù‹ƒ^û¼ÿêZ‡åƒ‹„“s„ti4%#Žº„]XyŽ8Âþùø?¹O*³ø[–]!Ë­W”×yE),þßówüzýUkª©ý»SÇý’ƒ\iE](/þ¿òüéøJüÃSíÛóG)÷ùKZüÏ;¶K@º3Hæ/î\åì/©íàöû”ôŠÇ‘ÿ÷æŸU•âÿK 2hj¿}îXÓŠ£tK“Ô-RÒ[þ˜°ÿ‰ŸXªÕ}iöy¹-?ÿÿäÙÐsŸ³÷ß³Fþe¡òuö'5çÊJbæãÃÏçøüœ½ÝÝüùöøùxýc)¤W²ÿ©ûÿ9CòŸÿ-uÜ¿Ìzi*éO÷Wgø–Ö@[ÿ×üiþeþ4|Ëk©ÿ=¦™Ù_‘•Ïßoø|öð¹yøþ'5ªúÿ\£ªÿjô’®0Å?]@ë‡/t€óî?˜ÿ¬OýSëóvusq[²Š5„£T¾Â-(ÀÏ?£`]Qjol; ÛN¶“þï¶“Hv‡m+aÛJضÒÿñ¶ÑéþË×»éžéðs gÉ=~Î^„+‰[Tä!øÿ#ÿMµäàýñ×ã*`}•ö¨ã j´™*ŽYêd§xÎÒá;8#•W¦ *¬ºs)²™¢àbaÏM 6ôæÙ‰µ›ã¢×Ûñ&Ýò;Ü«ÞTRb<‰‚ÒXÃ;· P±QëÛ·ùf¨ìMÔlñÆBTu_4oJôªµ^“ÄrKÕy·Rª‡·¢zÕáÐñLTŸ™i·]bÕG˲G ôÔg×’“Rb\˜úÀwUÇT­Q=9ýµçey¨î’þ5ÎR~¤ôA®¢öûï¸2„ûBjÒ¨¸ÑûO›³¹‡ePSü½ÌUAÔ’ÆÓ=1·>ÌÙ$YÈ£öDŽ^6»pÔåzäà w&êQG/÷lãCŸÛ¬ª7ÿ@}Å-BçÞN¢/-dãÝÌxǹ6õ­_*<Õ¼Ú%Ü}_ÏÍáð öiboEc‹%Ô«_>Aã ´ä~õ"hBkLùŒÕ4µöâ×Ó¯ÜÑ ·UEîõ›hîкU,h3¹‘ç¤'à‚óÂ8tœÁXÈÝX(JªŸê¸TI#†Bt`Õúg·Dy€öÉýGf…õ@ÿíH!në 0¸"Έ¦Ómæ ÀÂs9ô°Î`-VÕ¼¶Ø/Np¥¸ Ç¢~Äg2ã‚e–¬ï¥bÉ»&<*±¦´Àç¾ÑÓÈýðwÌ…´–L`ÔÁ æ³·@(\;Y'ý lªx_)vÅDåìÉÔsyAüí§Æé»°åœMí,K;H ßÜýŒã5ÈD(ú%ùäÖÛ8^v`ùºøÓº w@±´~¤IÀ¶’;yˆƒjð¡·,ßNHN[Ñ*´Ž—ðuÐŒª)­Ímµ·d g·îj“ܶêßq.¾­=Äj†]½¶ö?$À¸‘ý |ö0˜¾«;¯ d`6H~ä§÷50ÿ9®'ýÁ¶¯âs>¬j –üÚY‘cS`¥™õ€smX{–Èûul’=ØY¾ƒí—œ£ŠöÍ`¯\vZŒ)n>Ê {&N !°ãR÷–áËxƒ§®Ìæ†]-£M#«ÁùñbÁè—``ô`>w¬sÕ­=W‘d“BŠD´«¼àFFµ§ÑK©ó§Öl"ÅiÉ^óÐT?ó)Ž).Ì®P·9Ÿ¤ßqaˆñ^ljqeˆq_ˆñ¦ˆùæê(Öð6O£—)[êݶ¼'ŃÉûq¬+öéåßöL°s¢]¿AïÕ‡¿#¦ƒ ’;¢Bœw»Jð"*L?4BŠ3U\tÁíÞh,*ݤ¢O‰Ê}{9ÆâPåÔµãudY¿ãÉüX-Èê£5hÐM)ô^úˆ5ψ$)ÞSó~N;åÔlõL#>µõ3¹3ÚäþŽóDŒ's掜èsÔçaY9öè#úb!s·wf-úºzì6mˆÈe˜¤jDƒÌ<3·%Ç{"»÷æÜÅ!4vj4_ž<ß²zÝÔ¥Q4A“킦¸Çþ½IB3Þ}1z¸5hîóþÅV‹>´pò¶ß°ýqÀ=s½AÁb 䎞þ¸l 4½(ßÝð¨OP=;Á÷;®ÌYKÞc½§á”ŽÌ¸Þ.'ÇçëÛ8€¥Å”NPGØÚ¦^äÓlv™ÃOÂK€3&ÙÈÇL6±fÞûš;½{BÿJ˪ÅY*<äPØqš„¢…\g“¸`Óyå©‹;@4J—φS $<Œ¸ 'EAJsëÛw ñ =^$bxdß9&ïÙ òõ÷8ý.‚âì¶#;“|a«Ù‘KÏ]Õ¶2¯Y©€¤ÙÄÆÌ/ aw…²?ïx“ìhHȤ°'r úþñY‡o¯"c:N3Ð?¾yM8_e­Ž,GSÃ^’VGÑlèsùñ°hô‹ýú¦ýjhÑ}Èè«­ àÞd²©™pÊM@¡Gñ¶8§ (›òFo´ìê×GçÎ+©ÍÔù—§Þ‘/ÄoÊÖô¥Œlµ­&æÁ—|Àœó~<øÉC`U1X¿é°52T³dÂÚëÜw“tºËZ ÕÙÀU'z”ÍX¸í*˳9ö™¨Ÿð‡—Ìi¹ýA‘ùÓgaãzßК4NÉ­åÊÞ "QZw?ÈLƒhæÈ›þ_‰ !z;fº¶,T W6êƒôý§ÆòOª@öÁwƒéú)ßùÝæ 5($å ~ „­û/¾UîªúïÌšE%½ô9æ¬ê7ìî雿šœÞc UËv(Ð t>š¯Ot}ŽQëf}€m;ßyGΛ€Ñ½æV{ 0yó‚ÎÙÊÌ4'~Ynô³o'{ùÂðþí«nŽNÀ°øÁ^\ä=–4Ü›Ö(r€•h8ÙGÉ~°¶kîÙvØl¢ß%l=¤¶]<ã=äûÁ^>õYYÑ"8ÄÍ]}Hœè>[ÿn7ì¸Å褻èf:Âû~ÛÛ£¼Ðò¨÷@'•À^HóÕ†¯{=ñ˜4îíŒhWÏc©(7Õ9“üUš…<û—øo¤¸lĸi÷kŸ5BJ.I'ž'^Ÿ~aá±Ë…pR|3¢}’üœwübÈ ’¿#¦9vFT÷ãHöMòƒ¿ûþà—‰ñ‰ÏCò„y.Ñþ‰óÝ’'C{·5 ²3¹÷d{BHq«zWUÆx>E5[2¯É\‰Gï‚=V=úIšÇ6$@\öG&’ßkñ— ÔcÓFmù _'¤¢N)—8rÔC¥ ›o·¡Oµ áŒP_h“§å6 ôEîéý·YEèëúg&³µŠèß~]ã‚5hHóÃÕÅqôÝ:›þY45¼¢é_‚Ÿ¹¯áÕ~4kt ôE?ã…0îeE“ï;©™òÐ̱œù«¥h>4kÿ ´XÈ¡Öé d‡Ö™šäùô»7|Ê`m×Øé  f©â0“ ƒU£o.öI°Eì÷ì\X—p#ì0½>;¸%7X†šc¾»š½ÁƒMŠÀ‘Ö³^ —™§ýñ3ÀÍ•ÔDžÝ |[ÞèˆÜþd.ê×Õ§A0N¯ìŒ]8Óš»ì¯t‘±Â$+ž îthèÒÖ~b™“½–2tœ #›íAögCP‚/ÈøA¾n¨ ”yX(ÆÉ]vIrh¢°AÞ7•MµHçà2ÝЬ£wüá>ôëXÆAÊ×€ ÙT¡­äx™Ž¯ÊÈ$²b* ŽxZºNDh|Þ:ò è´,Ýß;ÆÁj§ùÉ5À€²×G¦“}×£«ÂÀÜÚú˜&³XcjF|{äxi- ™o"Ù‹ög»ŒŒq®¬ÿZÀvíh p÷ä)I‹æoEÝÚi_àË:‘^ZžÆl´U ¸ ×Ì^ ÂÓoE+AÄ«×T”aDëò^¼ºÈçá]f[Ü9Õ²¤Ù½òY•@æ¼’”^òV“;§çšpäçÒöÜq³Åb™ì]j© |‰ó,O ¨”VìÓ; èèé1vnPÏäçÌzÍš›¢Nžõw­ø#±z›ÕAg£Åç˜ó ÛûdŽ}*ôß9X¿ÛÛÞOßy]F-§)Ëo“×/y”>€™náËtès‘mëJ|sÁ‚j_¹Ž,Úi<’&¬`{jy!ìä˽Z)Š~Ç)îc_Àª’VˆÓµ¬÷Æß{ ¬ÙýÌå›óÉ´µ`»ªñÓ­;&`{\´r®úØžý⦮”úìsâÉIñ1‰ã?ѯüy=–Îæ¸PõöÉžˆñxS{}f¬Ð%Å%Æ %Æý%^GŒçIô—D{þ³?"êž¿’üÙŸýÚúV/Þ•ÿèßë½vö“Öñ¨PE'Ê‹/ µˆÔZŸ2@Å¿œ§…µQ©`Bm÷)~0ñ½NßsáAÓßëº>_…û7‚¾g£¡-Ï/CM4v•ú3ýžì—ybN†ÚOßvuðéûý¾E:äòš/[ѧ77oë×D}Æ&Nñ–è G`P8)~ïÀg{ß½«ž !9¶”r>èûkáä³®—Ñh]üLuvúa{&ñá{4¾ F&‰C/÷ÎtžCS×B97^F3IoM#fÐÜOÍT£F'´àW²ßîñÈo»¥43©ŠØ‰Â‘Ãb@µ¿ÿÚ´u¬zrbï꺧@ç?ï|\ÅÖÜíºqÅù30Û`@½“X¬äÊ3¨«€ÍÞ¥ToÖ Ö>î,î¸lœæ²™!6ÈjY¿^/¼¦ãáMo£ßÍéæ*Å2tŸáwa¡H%þA ?a÷œ—^ ¢Ç®¾‰ôSɹy ’X°[7dä*-©óp çÚ•W¥tæ{9ß^x~%Ì•þ l´~L•p&„yÞÝË2‘õ v¨Ùºk‚õ ^ôý^>Ò‡-۶˾ºÒR ?’$ä@æ•LâŽõ¬ gs¼'åàPXmtüV{( *’§ÁÖÓWü×Õ Å‹\tÒî†m£ßƒúŒµe/Ëhz’Éhpƒ6ï§LJ7nÐÅéìèOä½ŸŠŽ®­&`ÐÝ·3·Ä ‹Ÿ1uùðƒqЯiòa0åÉ~¸àôÌì® ÂÛ׃¡{r`^ß•ÿ@J ,N( mõ׆í[2ÏOªÂöW2z–E’`ið«ñºô}°|Ç2ÿl0 ¬¶%}÷H«üuß[UÁZøUsú$°]ÈQŒfd£ÊTú#PöôÄÛŸWºHë*¢]=>š# dLòOÄõYÊ´r8ÛÜ#’%4+¹œÛ@ЇMŒ§MŒ¯MŒgO´Obþÿ°Ž#ÌS‰ë9b¹þ¼ž#Ík ×ý%é}éŸãkÿ9B¾Äy.ÑOÇ’?$®ï~øž‡è÷ˆûÄ÷3Äý‡‘ ýÔèýâÍ åæ¨™9ÏzU.æÁm•qt¹û P‡ùMì&Ô5Õ%,Ú´ õ˜Ìù\ÛýÛïßgãc¿fÛy~­ ´>ïlч†Ötþ˜D£Î/+rD®¡±êÅ™ÌhÜ’.f_ôøo{%¬ë¦§µ%ûŸÞGsÞéç-TbÑ*s7ÚJ ¸Ç›NÈ­È+i’Ï%Ó–“äjt@Õ²6fö+¬Êó“õvÚêëÞbR°Z؇ý.W&0ô'¾eóÕfVñ§÷)Œ€ex¬Ý| lß83÷]rV£×%]vÀw1o‹*7lHâŸ[uøØŽ„ŸMfº½åm‘v¤øõÂ%ù™)°™Ú÷4Ð>ë02©Ô°†-GM~Lé²8¾@IæÕŸ4þ ºqjCt æ¹þeý,¨8jëiÉrJÖtÚ"ÌŠ«tM€fë듳§@ûþùS¼‰õ {9Úgo苹+c Û(7Ÿ¸•†Oަ~˜cÃüˆ£y`Òj~‰÷e˜mΟ>1¼Ì:yŸ)éÝóкç{£ùÁbs›¦ÀçX°(¨ü>1œÛMXÉÅžn‚íõíÈ€f,¬˸}À2?£e{šˆveÀÕE’ú-1N<Ñ.ˆóAâüh7I»G¥$OüѾˆqè‰)ÑÞˆöJì÷Äy#ñýHp~ßzïú‡F審¯óBâvåëÆ£O[?‡nR±F½ÛKJ^ Ï5{öjÍÜ%Í#¿¦­¨(ø½/@\ïù<ùåˆúùB/6ã;š/éqÿ¦Šf:y=§>0¡ù6‡õÝkû»-…BoóÂyíº®Ûâ@eVývð’-¬ ÝnׂŽa”w×À¬Îsì¬ÿÒ Œá‰;ƒŽŸfûteÛÜ9`(m`®bv‘ ªL¬OcÜ9úÜÍÀÙ3À,tA 6œwòÙz¿®ÓöI]{þð¯Î°ìP6§$ ’ MgŸ~Iý D¸¬/Üß‹_×Å*ù ¬‰“ªE÷÷<©{âwsG‚ ³ÂiÃLJÍûBk’¯ ò.µ¹Þ-7ÿªX”cßMêúª«Ùå7º/ag¨& ê/ŸMª<~ š&û<µ‚¸@›åá¼9ùeÐùöqõ^šnÐ+pîAÌ?ÀàꙘor¬`¸ïõñæ§V`ì¤QV+n¦*aLç5Àìª~\ýÁt0|¸«Y{X<кwBñloòž{hæ Vœ:¯­âv‚u c„ì¼ØüøÈ“_vWeDZCÁÁ¬¹lÓ¥`pR½¶šî¸8ì´M˜S¿»š¦*"ÑȮ쒵®¤ý0âûÄ?ωvC´«¸Üc–ªÛXIvu}u ÷½ tMAÕìCÍ9RJ‡uÐäô×°ço éÅ4¦‡©ÑìÔìT¼Év4¿xøò™v2´¸V†Bá\àÞ„5¥îíò­ö_6õåz'ÏâE ¶Ë ÚFŽû½_>^¾®óH68]œ†ŸÜø«Ÿƒq§U¤Ò]0#¿½“cã^0—¤HÛ(Û<‰-2¥°½•[ùüø^°ÚAómð²ØÐ¹Î{í ÛOg7;™•ƒý°¬ÑÓ}~à$ã§úJíìÌÛQú<‡ýa'ö‡Øvbß Áþ ûƒÎÿÞ?èüËï„Pÿ«?Ât]þ‘ô7Qáp«ð¸G&yx”ýé†UK¿%q9€û¯üæ_×Q€›—oá/Õ©Õ‚þxõ„;„´‚.ÿdFýö:BJúÌ?«|²}+~ âýWì_qù_\@ûÇÏ\¼÷9{~ÈBµ|u7¡)1`Àðƒû0`ø?C& 0`Àð_ˆ< 0`À€á¿e0`À€ÃˆûpËŸ„ ¼¥}YJÂ>õÒv*4ÄOêàAÇj¦ßÒv##á³;Ìx°à–¿M͆;kñàÀci¿‘.<Öã±n0¸Êš6}ý~›È+«@/æ­òb3#èÞ|-!:wŽÊ‹žQ£ØÔ“×~§?wЛ0·ƒÞàS&A—L0<ç­õðp˜)9ºwõŠÂv2×3 \`E¸ßÆ«­ö;£ØFäí?> v*[÷fÜÐ;¿].Ã/ß_üNÝÄÞ;³‚-¡\¶ñEwÌlå#æGJ óï1ßì[ö?I÷Ùïj´–¤]G[! ‡&ØQ³-‘œvu ° /øÁîÓáš­EwÁÕík¤4 ìqR¾>{ê+ì%”sÿx¢ÑFxE¾X I›ߥ«-;áÀǪë©NàMúÈ[“8è¹ÆÇÒ0 å9·ôïí« ˆ<ÌÇ·{hZ¥!î—pÛ?5ÃÉæKû¶xŸ†Ó'”Jµ<ÂᬣE¿D¢ œwÙ°ùV(\ ܉RóLW­4\I¶“ŸW1…ð¬ˆH×õ²¹©N$±+¢rúr޶B´ÿ·,÷;3ãöëb[îćÙ†ˆ×ø†6ú$èÅ­Çw¸|æy{@â”9e“y $ŸÞän¼Vîñ䘞兔n#ªH&%¸ß)Z,°™IÞ¬>¦OÚy»W}ƒôý íº;!sý¥¹y7Èz-𹓽 žŸêˆ™ƒÖZäÏÝ!W"Púj)¼ z´:ô.ä›\M§¯þoJ_åš@¡Nvñ!úI(ª²I=W´ J<Ž$9]>e/iKCùè°ÚÙ›—¡²í(½2§0T÷Ö¿ebð€š…¶=,§Ã¡Žaÿ|í[¨çx\2š \K Mï×ëd¬ƒÆõé;Óò4 i÷æ…× ™÷Gß•‹~в‘Ï7ç]´Š~7n‚’~m,[ž@›âêöó–ãðQ+ú$õÆh·’q^ßÌeõ=´Ðyz¹Wz'tÝÜø¼¼ºí7<¦‘ÞÝ_¤§ï„KB?%Å7rp]lÓžw`=J}9çe9˜ú Ñ®´ÍFÚo?Í''Ÿ|»p4èX—zÁÕŽ½Ñ?Q *{Úªðipƒ–(섘«JÊŽŒES:ó³ìK¨ê…˜¯&UY²÷Ð"Ø£ÖQ)ü£ö݆Ì7Û3ÑÎ v­«¯ÞsàviÐÓpi×÷}‚ìwÛÅÒB¨«#B9L–zIÀq0Ë–zÀ®x,öeIà³.cHŸ;fý_·ßû€ƒcpklæ7p"<÷Î¥f:Ð Î~#tŸÌÀEÌj„·cÜåt'Øù~‚]{±õæ½·?>ãøž¶ð®–+¼ÿüLŽ‘_äÿ ÏùVuhCÀôóÐÞ½cpˆÀDŽ.u—w'áØ—⪷!ijJ"{}:œbúc ƒÓ5Þ”vpö^îîKù,p>“ߣ,àâª,‹ûZ'áR°±PåÁÕÖ'n[ÅCDL’þµ}ˆòãÞŽ¿¢ñ¬£gvC aœ¼*üðÊ3¸µýŒáŽœWi³Ò9Hñ/(œ¿ ÉKÕ¼ÿ=¤p,=è-¸ï—ÎWŸð£÷†Fxò鱌@Ĥ‡šÅÏ”‰B¦ÎÄ“ë)wà™°õ£Nȃl‰ÝJmªüƒïýê—¯ÂË¢#.þž]oÚ¥âjÏ ä?fÎÃÿÇÞ{€Gql[£Ó3­&ƒÈDÎ 09í!šŒÈÑ E9ã!g9Id9ƒÈ„2`ƒŒmlÁØØlÞªÖ®–T ùÿÿî½ï}Ãw¶WwOwõ®°ÖÞUÕp^ÑÙ‘ô£šÒyÇÞ#O2¼¤ !¿ŽßÿÏEº4ùÇÈžqteD›Ï›÷§k®µ[MÙA7‚ËÄ¥ËM·®¿Êzhø6mî]Ê4§;{jlÖi$…?x=®Ú;Efê´.æÒkŠª_«áþRô耱߬/M1{L¢P,óå^Ns$ð­gãñ#²l¡/ƽpö\Û,H»vmÅNôpýß¾OZ|,ßV5ùíilý0K÷[ò8qMŽk9Þåø—|¨³ïn©OãæZ¼‘ü«õò³ª!M­ëò9Y®ŒŸ­™ÇíF;–ãTêÀýÜ™ùÙ^}2ÖŠ;Ý×kGû|]ƒº1Oº2odƒõdæýßf¾>V˜f3…žæ¶¶MŒ¬S‰æOù[ŒH Ì!ˆ™Žm>uMËr]Ýûå½ñ´üïoÊ6zB+_î+ß;b­Éœ û­mútUç½ciý±ªkZÛ&ц*Ïçµ­|Œ6^êç11ûTÚŒÞÎùÇK °yÉ•õuhÛÈ–UFÍL;Ö þ! ¨ƒv}s°p÷Îwio«]«ÿÉ»BžEüÔzRkÚßùçžù礢/>m~êÏßèÈÚ©ÉïtäŸ[‡^¥ã‹7y´)šƒN3° j e>ŸþtH¿ú}éÜŽG+ª JE²›JƒŠµn뻃.·}>ºËdoº*²’L?ÒõÝ9»}áA·XÇÚ—*õ÷ÁtgÒñgöÊ5(|±Ù1*Íï1³2!¾U˜;hÆåy½D·u¹):mããÝ*Ťóî_*»Åf68ÝÍëiñ‹ïè^©¥wc–ÓýZ]v˜Ü™>G´Ïw´5}áÿÙ†¹Ñ=èëYébÁUèáêì"ƒ¡¸NÂ&L;œÀ·aš!Òb?í® ªÙ4µ¥û’g’Wµy¼ËxU‰ÿˆyTAÈÄ«p*Ïü“çòwyïÕl[ÔÈâ§ä¥“㌌w~ò¢R*‹ß2ʸûV<ä¸\Ÿù&ëcÅ?æµä¿Ìw›³?>ìO«™ß†ÎyŒÚpÜj{ëBÞ;óS{n?™Çvæ¸)ãž/×£'ûÕ›ßïÏ÷÷7‡$Þ^  aZbÖ¹ühØÊYú§£y¦d»”µ\&¥i4ºìÓ=mzÛiÌç÷«\]ý„Æq9¡êÛ—¤%çëSr{YÞ’[éTç.Ë‹_¿C¡"+öþ”Î í=€.6LܬÜTºÌý}-)”tÞe^Zžn#ɬs …A¥>·Â³˜GÜÿS|[Z­£gæÔ¶è?Cç̈¡¦÷Z引¢»5­8ÆãÝÙ£yø³í¯­|RƱªœïH^Uäq.¯K~4þÊ–åÏC³­qÜŽõ¾_ïÊåuû£fÚüWž“¯)´U¨?/㋲æ”ç³ò¹.Ì¿$/d½,žÈ8§Îy>Ø9í3Wê ]¨ Ç+éOw®¯ûÝ“û¹Oªö¡Ñ¿7°æs†Ž\P`y|ÔìZ{rƒô_£áM®½ôQ*)²‘r¹i´p3ý2WàûJKº_  ¢×F_ OKm¿óBK“3ˆ@‘ÏŠsÓ9žÎäúͳ”ºwi>ó,0UéÛ¾šD‹Ú‹Ä'ˆ¯)ž!Ó¹q´ä›‘Õk‡–£eÜÿ+2Œ_gdøšVpL¿4ñZ³ éÌS·´N4Ëî]Äí½ÑL/=hÓ´Q¿nÙÜŸ6ß tü¼15š¿é\ÆY´uáÏ¿ÿèºLÛÅð ÿ•všÃ¥í™úu‘=+PˆHþ™IÈÑF­;w‹Õ¢¾Þ4•­^V7^ŽT¾ùkŸãéXÚýËž|¹N?;øì§t ÙØW‹Ñæ<§Pæû¹º{îT«—ƒÎó<í"Ï_/7˜žú¯rtåÑÙ’±¹Ò5Ö½¾íE¦I·d”¯ñˆÂŠFºòô¤;ÙÌ£;ß4m±ÇŸÂ×å=MÜŽOª}ÿE§´)Ò”º{)Šóâ(_Q¦ŒÔ¡è¾éKgÚ=…¢Yïb ‰þ’b8O‹9ûíò {¶R¬9MËH± Ä §i¾Iß–¾ËyŒ+2žIÞT漦"Ç+®ñýe9ž—æqZŽÇ©Œs²ÉWÉg5ÉyšŒSõþNgŒ õ zœY¨Æ35®Éu§r§Æ³¿åzޝ—´fž·á÷µc¿$_åúÌk¥Nøq{ôàùH/Î ûìË3(Ÿç<òÙO‰J ë-ï‹s5þ(–¹¿“FeŽÈ½¦ÏKÃùÿ¸¦÷ƒ¶]žÀWs:Ø•\abáå+šR?¸`íýiêwYDËÐô€)?ÑÌ)`R0ÍæþËëf Úœ_ýñ Ýí¸»½--Ó±-_ÑÒ„pߦ-zŸ:žŸVE˜>ãx²þ~³s>ÃRÒ†ÐúÅ›¿þŠ6a–:}æRÚò «Èœh[Í_o/‹|B;0ÉÍWdzÂúÊêÕïÌ/Jû‡Í(×iöC:è?s[tóOéðȱÅã®Þ££ÈÖoŸD'†]̉šNÈ,V`è¬íñÞ_^¢só²¢GRÐ}yßCÏÎÐÅ £¾Øõó—ty†¹@BWÍfù„®¯ý(Ý\Úe`ß3t{ôÞ^Fµh 3§uNºS¿oÕ¼ÏÐÁ¶Ôa>SóÜPå7ŠÈÑvΚÛQ±ÞBÄE3 Qä‚h©Éã,ªÓ`ØDŠÚioZ‘¢y^Ýbó¶³m¯Qôâ±™¿Ë‚ø–S–à ñMá]]æ“ß8OR㛚gªñMæ2kÅåÉõI_¬qËzë‹·ø_ZB~ŸÅ|ñwAêÁy’Ÿ¹|1‰º0º¨³¸¼b£µ¾!ëÓ†×÷ä¹|Ÿ¼ÏZ¯<Õ§Ø„7âœ\·ävëÆë¾¿›¡„'dz\ϰòI…o2Ÿ”<+Î÷I”¿Ë8(ËoÆç2εá|Ñš_ñx’óº.b»Ä¹€º2?dìÌ~Ëõz9kÏy¯\÷”ë”r_òÆšßñuy¿ÿXT?¬ý^Ï}k?@dA™KQ_ÎÛú‰mÍ•¿Z¼ àyð0Ž_rÎZ·4§_=i&[cLØ—»Ó¢ÆÍJÛhª9a£iæ2]6šq?]l¹˜n4›ãÖ\ÞW[ÀóÈ@±]5o -æ÷-5—) ÑrîÏU­Ã3öã˜wAüÞ Û_};pn}Ú$šaÖW´YLëúÅP0¯“oý¹…˜!Ñö¿V|YáPwÚU9#”ª)í9›æ{D![nu©ÛàíÏc.8ÓÁssnÎF‡†¼Îˆ!@G˜ÇÇxßíDk±`ÓŠNM¥=šÅì :;_¤tžóæ S3A‘|è"fÃк¼uƹ¢®µtuHH¯ÑõUççê;}éæŽM»s„w§ÛA/S 86ÂN:ç/¸Ú•î,Z`gÓá.¶'NÄRDƧç›åšK"öG‘½ÎýYåàFð->ÎFÍ®ÒôÒÖ§FlÔ…Q4HVzülŠ«ü™·S ÔpؽLÃû0±FÏË-5¡Ø6½µG[Ò”If"aÅÉ/¹Ž(çiê:¤7ɳ’‹ò8/Ìã»—#ã|^òÖÚ—ãßeÜ{k½R®O*¨®ßÈ}=Y‰j¹oí÷qýd<”ó½&ì_s$1]kýlÅ=k^§è„\‘ë-Ÿl  M¾üœ\çéÉy®•òwþÈŽ²=Maí/ âùu€ØM:*a÷‘Gò|l´¹­;†Æš ƒÆs¼²øŠRùÇh2¿*ïóMYYßßiæÉOÅŒ†æp¼œ7Ó\p§…";ª¾Y÷×57²iÉ£h±rBËn¦9üSßÑ´’×óÌéX‘Ÿi­Ø, Sïox¡p>ˆ6§ÉÅ(˜óƒí¹FŠhçóªC}— ¤øeí«9û·ÔÅÇÐÞ/9ļ;ŠŸÛ¿9KÇ¿ªý&§ÏGtŠ÷ÉB[/+Rt.E÷!áùŸÓù—ùz ÉÚƒ.òþ÷å¿Þ(’¢]ýmßÏ…+õ£ëærO ºÉóïÛxêׯRج͓n?9Jñ²Ö‘ÂSvÁ î3 ?b÷-8…":‹ÌúñLò!Éq8Êvò´ßí7e¦¯ )êÛº«0U èf»;œüt Eo?ùwÓ§«hfêZ›bR[ó*¹Ï&ǹºþø¾ø&ù%Qæ™ÉÆ7^ï“qæ­øÆí!×åw ]8ÿ•ëð2¯“ßÇÈý krÜ—ïÉ2Þ⇜×Éu6Ì7yŸõÝ —'Ë·æurž)÷çÌeåTÔƒ×7z-¼ûwå?¾¦>wúù|”¯úÄ4°ž9A ÇÏSíØLÃz‰…é™4’÷ ǰnŽãu”‰<tñïÑÓš×YߣðzÈP^7ÎûV#ÍáטFs>6–y=Þæ)øÊñu2¯3Lk&~4ó癜‡Í†J|uçÍãùà^¿ d=Y\óŠø’+!¾ÝÞqsÝÔ¯âÛÖ5˜9M¦µ—ƒLw|hƒHÓæ\¢Í¼/ˆöñ>Ê/s¢D‡\ݺ{û†Žšr›™Ž›i\::u|ެAã)ô“ž³z/ÝKçø»­ ÙMÁ§K¦tE¤]ÞÐ51ÿݾóêÇFKºÕ}Úµ¨ñW(,ò®öO^d$‹éN䀢gfö§ðuçn5lH]ÿÚu÷»Úi¦ùÃ)òPš ÅNxSTK¿¾… _¡(^‡ö{ù4÷ŒÁ}iPŒ¹íÐböí­žåÖrÝ3'°V>)ã˜äYrë ’?*Ï ðýeœ“ù¨Ì÷dÞ֔׬ýeÿ­yß©¹üý‘§ÔïFÔuHùœ\ß°P–§Æ3®§\ç—ë V|•땬 2¾Ê|Ü@c*e;åÓ4žóÌIÜŽ.æÁ^טÆí4Sањݿ¿VdÕO4÷á½ÕiÁ’u»£ÏR 9Ý(iíÓ-S×ûõŸéi9ó`%¯c¬É*ðOØßæýèe?ù²kð5ÚÄù‚\Ùz¥é'7z>£í×ÌÚin¿5¤=ýv×=õ—„8×hé…ϧ£ƒÌ$é0¯GåùØñíKÒÄ=ˆ “üýÙѽ ×ÐY‰Ù1%¦sÓª¦œž*”.d6 ‹GÏë%è²¹<⢫ü½Íu>¿Éß™Üîe~hEa"m‚vñüiïE}ò[?Ú'>Ûº< 1ût¸Ñw[ó>^CGŽk1îH{:±éGó ËÓ»Ë\ÏÒû l…ù¢O¹;ÐEñvƒÂt™¿#¸ÚHL¤žÑõ æB=Ýd¾Ý>Ü`õЩ&äqÜ—tGÌzÛÍ ð?ÄÞÇ)’óà(±ŒYy!E/ÿfN…ßS Ïãîú-Ê‚*ѽo'OÅÔ‘>7e) =¨.~jŠ_ywõ§IP»Ðݺ5Þäø—|ëùj\Sy¨òOå™ÌG%_äx–ãX®Ã·âçäwÓòûJ5.És+ž•Û=ðP%ÍÿrÝSέò¸^j¹ÖþœŒs¼/"÷ç¬8ÇyžÜ—³Ö)›æ{T¯ý ë{?Οzöwf8>«Xºˆº^ÉyÛ^o³ö¸^ò»e¹_7®½I š(vß:= óvŠØ>¯¸›¦q<µæsiS‡ÕZ½Œæšá§ -ð6¾)0x”×Ïó3ÒbþþciS‘ðܦ忰ËD+ͰԄVóüc­YÍê$¾j¬š—6ŠƒÒ¦çæm‰œE[yrûóƒ+OÜãÿÉ‘5Ž"¾Å·ÏþÆæ‚4«…óþßQæåqþnò¤X=ÎTŲtVÐÐÿs:7]Ôit!—@wè⥉ÿTêC—yžuu{Û¡õ‹ÓõÓ£ÄÝ_evjJ·kóü@awO7‹i\œî˜Ÿw ð°ƒ£GPÄÙ%‚çæ¢HszZƒ¢Ìée/Š~>ëV@ Šu¬>óÑÌlt×L»=éÞØ»CW¦¡û7æ?ÚU¡}!>ÓßСbNR\v!¼S(î›ü?ÛH£ N˜ÚZyžÌûÔõÉÉ«‚Ïÿ–‰Uÿü´‚¿#\5B|€µ‘Öˆ¯ 7¡µ‚5GRëÂF‘6½ú–6»ÌCÁ?‹…¿E´}®ù{h×s‚öî7 Gû‚͉¸k(!¾}ã[gv•ft‚çE§EoåïrÎó¾ÌE̯^¸O—{¦^Ôfì+º*fqzº~ùh`Tp¢¿¿óÝ ­vÓæ1…ó¾u„øÊÁ¯/E~’aË¡)jYäw¯l§(šã`¬óˆ«â Kt÷Pã—ÚÝ£ûÎÂjí]@Ÿó÷eø{î¸øýG›M·Ûléߨl6Oñ—ãø/Ééc{û ãc‡__ùwçÃ{‘w óë)/kCã‹)(‹Ñø‡Ô½üFø•í3ÌoPïø; ‰;ÒÁì.üçÍ›—^€Þ6q¯UЇpf8?ø‘Í–bl=,¶¶¶ ¶¶ Û ÛÛÛÛ ÛÛ ÛÛ íƒí‡€„‚†…ƒ‡€„‚†…ÂÎÂÎÁÎÃ.À.Â.Á.îÀ®Â®Á®ÃnÀnÂnÁnÃÂÔŠŠæ*fÎÌ] ë͸Âns›Ûþ{l›ÛÜæ¶ÿ6Ûï6·¹í¿ÍNºÍmnûo³Kns›Ûþ»L®›˜ÄâŠXdqˆE+^W1`)`)å:, ,-/2¥‡e൬Œ°L0±&“–– –––– –––––&V· À  Á ÊÀŠÂŠÁŠÃJÀJÂJÁJÃÊÀÊÂÊÁ¼aåaâ¾la•`•aU`UaÕ`Õa5`5aµ`µasÂêÀêÂêÁêÃÀÂ>†5‚5†55…‰6kkk 󵂵†µµ…µƒµ‡u€u„u‚u†uu…}ëëó…ùÁzÀzÂzÁÄ¢aX_˜?¬¬?lLü¶ƒÜýæî7w¿¹ûÍÝoî~s÷›»ßÜýæî7w¿¹ûÍÝoî~s÷Ûÿ«ýæÉë&æG=ƒÜî÷ÁÿOú¸ÜîƒÿŸü/÷ûÀ}àŽÝî÷ûàþ—ˆŒûÀ}à>pÇn÷ûÀ}ðÿKDÆ}à>p¸c·ûÀ}à>øÀƒÿ%"ã>p¸ܱÛ}à>p¼ÿ ñ7k£ýûèí>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûÄ}â>qŸ¸OÜ'î÷‰ûä¿ã$½ø;~´æbÇ^ ½ ®o~HÑdt^êó?¯M^E-Gz6o—·5^˜·vT±o¨Ñ·/H»Ž>^q£Ú›è Ôð³BG–•)D ‚ÆW*9­*5h¶²»ñbiþÖ-M‹ŒŸÓÇ?ìõ,Ôs?51¸Þޱ'ɧj×¾¿*Im´^ÓÒgÏEíøùƒîÝú9C-ê¸dÄÉþ“þ¢N5ªûï[]‡:ñû;K¿ýñÏG°w©ˆó~™©#ûÕ‘=nø¬=u`ÿdyž=µ%ãgqÔ1«ùBë¹Î¾‘í˦zC];éâèEÝn6ÙÞ^€|ã f)úÏ0ê15°nì¹ÍÔ«÷w‹‹”ÏD}>©¶ü¯)ß‘?ûÙÿù†f?Öù‚->òOÊ=¯(@ÜÝö ½mù¢­ŸÐð¥åÇÝhñ’FL7¤mÓý4šý»JüYM㿾9bñØ,4©úÄ-¯Õ#Wßy¶6¢irô¼~åO¥©ŸV½Xo@ MïÚúÛ2 ÑÌžÝF”X;‹fóóóôºÓÞ*O 6uªôºFK <°hq¯ÜÑâbaÅ7<¿D´áVzT«Ú4µXßæÙ>¢-ù޶úaz~ Žkæ±Ø³*m{Pò|Á^´³ìg×'ÖA»?Ï7¶ïµï)¤øçûv§ý)ÒÌ{õº78]ð›Y¯Ñ¡™K¾Xñ¢i_Ï~0C_:Vftù…=èD¡1;ÓÎÚL§Z, Isý1¹ÐõĹc-èlƒÃçG¥yAç®uØ:ã\1º0`ÜÆOæ£K…g7«?¡<]~úS­éŸÍ§«÷Ƨ©–³(]ÿêÎ Ïôèæ?÷údšHaéû¿¾•§#ÝɾëÂ/³ŽQx.ÑÑ©)"÷ÇÓ e£ÈÜ!Ý÷œ¬CQy¾ÿµÕÙ¥ÿׯÌF1…½ŽÞ¥Øâ£nQ4Šî–v/S¹Ýt¯JÚÏg¶}N÷ë-›lþ‚>oWÁoBî{ôÅ€Œ—î|™ŠLíÓjPÈzøY ‡.ߢ¸Îyv¥,_Œâ—ÿ3(°,}9\w¬>S‘–—êÿmTŸÛÔþ©1ÿèñËԊljäU}Ÿ_>ÿ¬ùvª»{òîïg¢:©3‹@Ä|¬5ñL£O¯/¦̧ê•ý§^ªÆ<‘(¯×,ûÑ/Ï–9¨v‘i¿]z|œÜ.²Üº'µMƒËQ=æc½í%½Q5ª¿9|ÿ™ûÍø,yμnØÈùåÐõ.K>®÷¤ççP#æacæo“9ÏRX5c?ZˆQ2bùöÞžµJµf~µå÷µ¿”µ`ðóß©SÆF¯B]ºNˆ]¹ÿ{ú„ëÝ]tÓÐä7ì—ÔßTö¡ž¥Úý’ÿ‹©7ûÙ—yÞŸy=(ËW'#:Ï !Wžc¤¥£¡ùÓV<»m kYp¢}N>>ûWŠ/êÓˆ?ÍúÊÿâ÷a}/†ËíÉ4ññêÈÒ5דkàµ2‡s‡ÐOSÈhêÍ/ë¡hú–c=æÊD3÷;0¢ZÓœZo«7™æ]¢,™ “t¸Lª÷j £ýÎù éø¹q=‡|H§Z>¬Ñ«s^ µÿúr&½¢³#éG5¥ó޽GždxIB~¿ÿŸ‹tiò‘;<ãèʈ6 ž7ïO×\k·š²ƒn–‰K—›n]•õÐð(lÚܺ”iNwöÔجÓH ðz\µ)vŠÌÔi]Ì¥×U¿VÃýþ¤èÑc¿Y_šbö˜D¡XæË½œæHà[ÏÆãGdÙB_Œ{áì¹¶=XvíÚŠèáú¿}Ÿ´ø:Y¾­jòÛÓØúa–î·äq ãš×r¼Ëñ/ùPgßÝRŸÆÍµx#ùW3êågUCšZ×ås²\?[3Ûš}-Ç©>Ôû¹3ó³+¼.úd¬wº¯×ŽöùºucžteÞÈxØžùÖNŒÒrç¨-ÛöùZŽê»ÇŠò~ß$ºr¼ñ^¾ÇwÕvܲâ]ïtfÇ‘¿mëÅ]ç¦)WÛGiHƒ9Nì<ØÂçâF¤˜íW«üqÉú0¦Iú:iz¦¡ñyE`ÜK“z Éê·(+¹Þ¬(<¢ÍBš3<ÇkŸF4Û}ëÉÌû¿Í|}¬0Íg =Ímm›Y§ÍŸò·‘˜C3-Ú|êxêÖýhI»;—ß|š–庺÷Ë{ãiùßÞ”mô„V¾ÜW¾wÄ*Z“9"ö+ZÛôéªÎ{ÇÒúcU×´¶M¢ UžÏk[ùm¼ÔÏcbö©´½ó—<`ó’+ëëж‘-«Œ<š™v¬üC@Qíúæ`áîïÒÞV»Vÿ“w…<‹ø©õ¤Ö´¿óÏ=óÏIE^|ÚüÔŸ¿Ñ!µS“ßéÈ?!#¶½JÇoòhS47f`AÿÔÊ|>;ýé~õûÒ¹VT”Š.d7”.kÝÖw]nû|t—ÉÞtUd%™~¤ë»s„wûƒn±Ž‡5/Uêïƒ7èΤãÏì•kPøb³#(bTšßcfeB|«0wÐŒ)Êóz‰nërStÚÆÇ»U"ŠIçÝ¿TvŠÍlpº›×ÿÒâßѽRKï:Ç,§ûµºì,0¹3}ŽhŸïhkúÂÿ³ s£{Ð׳ÒÅ‚«ÐÃÕÙECq„M˜v8oÃ*4C¤Å<~Ú]T³ijK÷%Ï$¯jóx—ñªÿþ󨂉WáTžù'Ïåïòþ*<Þ«?ض©‘ÅOÉK'Çï,üäE;¤T¿e”q÷­xÈq¹>óMÖÇŠÌkÉ™ï6g|ØŸV3¿ œóµá¸ÕöÖ…¼w6æ§öÜ~2íÌqSÆ=_®GOö«7¿ßŸïïoÿ44H¼½@Â:2´Ä¬…3rùѰ•³þô NG#òLÉv)k!¹L$JÓhtÙ§{Úô¶Ó˜ÏïW¹ºú ãr&BÕ·/IK.Î×§ä:x¤t«•4­N¥¶#7å¢3KÙ1¤hÖåGSœG\4ç«Ç.®hAó=JTܵ÷ZèÜ8sÒ×(Ðl6-ÞVÏ5Ëç3Zú¸Ë÷i6üN+Ö¦îÛ#ýZ½ùõæ’?e¡µœ?mŠI]+`3m¼b>BmîxÓkïÆ§ÜáU¿•MÒÓ¶•¦lˆ¼³ò¸ ËOÒžI[þl¶»#íó>î§âóé€HË_Þ§C¿\Þ´Ò+/å|ö8²¼%¶Ò©Î]–¿~‡BEVìý)*Ú{]l˜:¸Y¹©t™ûûZS(é¼Ë¼´<ÝF’Yç@ ƒJ}:o…g1;Ž"¸ÿ§ø¶4çW¶,¦¶Eÿ:gFŒ5îd>V—y#Ç»äƒÌ'eSyX‘ÇyU΃$OäønÅåÉù›œÉxÕM„£Ý»Èoñ¿´„ü>‹ùâï‚ԃǑŸ™ÞM¢î̉u—Wl´ò?YŸ6<ÿ‘çò}ò>k>wªO± nX~Xó:ŽwÝx~äû»(x¤æÌ»Aw6æÛÔ©2D™‰ ûqÀ­éí£‘e µ8¶a4Ó¡Ä~×3;õxÅz‘4ç“~<³¯K"û)æ´°Mû¹ZÙ|Ô¤™üÜœ†k‡×ýéÍ/¼®þ_Yh!絋¸]Š{‰–ĉ‰G;ZÆý¿"“hÀÖ´ £±øáXZ³ÄL`i]ë¬È0rRتð¦M3³¥jC›¿Ì{r{¿\\6Õ?3¦Ö¤­Ì¿í"üOtÑÎÓ×vÎéú˜öp ‰ýqÙ™§ hÿjýºh:˜Ò¼@‡ÖMýºÈžÇtDL×^¦cf±£èçi§Dxº{šÎœÍú°w»‹úÚ$ã8yþÍ®¼%>FO¶›Q¿5Ñe3 ÈNW 8¦_šø]{¸™¾¹š—nD÷øÐ­'c"ÿ~~˜Â¸_JRø¢­]ÃN ¡ˆ±aÁ£¼~¡È—uÈrœ¢Jo‘–¢þØüíΑ½(ú YqŠéƒã ŠéÖ˜(Š]Ÿm_žAùè®I÷ twG‘ؘµ©é^~S8éÞ\Q¡"tï§EþÃB Ð}žÌÃì´ùÒ”Ôô^«œ·Rt·ô_Æu>V™çy’Oñ¸+Ï÷—åx^šyXŽy)ï«Êú.ã¥ä³乚¿ÊxöV\ã¼Ior¾×ýnÄëM8¾6á£àVjÁ~·dþÊuVõï›ÌËš0¯ãr­øÆ<íbÊeúä•À)òO®_/~ÎZo™4ÁørÍ"êÏ:1PLÛ—¤!¾ÙWܨn£Î+‡qÜ‘É è4RŒÂ²Ãi4Ççq¦œf  ƒòyâ•4éQ½öË¿™C®ÐÝŽ»ÛÛÒ”Š?@qïÓÔØNW®VHOÓ×ÔJq¥i7‹·³g?Ì8ì—44—×Í´=8¿úãÈÏ/F6Ûìœ-e¾¯hÑûtÐñü´*¢Á¼ðÔ‘©î£/Ðzžn8ýÛg¿ÞM›„Š (F[î:pbêmÚV½ÍѯN Ww?8…Ô·#…¬}–!›ö6JèàSÅBè0ÇÁ£¼þsb²w6¤ÔtúÊÁj={ÐYOSè܆C—oΞK ôk]r]ge×Òå㟧rí¯KWcfe„ÔÓT‘Ër”\E·ª—¸<6ø…y›„¤;"wJáZ˹T¡pæiÄŠbeƒ"ÍignŠ|ýgºóUÀõÁC(ºü­½Eòä¤è¹ÄŠŘÝ:‘bĨ¬ô-Åæ£!ÅŽ›/˜J±fÐLdH ¬¸#çi2ŽÉø%ã”Ê7™7Jžã|Q¢ä¡,GæuÍø\æk2þXó,Or=² Ç]¹ŽØ…õAæqrÜËõU™Ëü¯í¸/?Þ“e<µù;1&ÔÊ{m˜§m˜§ò>kݔ˓å[qOÆa9¿3Ó’TÔƒóÀ^ ïþ]ù¯©Ïýt±åbºQ?ŸòajGë™JŽŸ7¦Ú±™†õ‰ÍLÉóÍ1iÄYãuω']ç'ÉÞñQÎÞ4µRxã¯[Ñ4‘ådŽ 'V"S¬C³µ#Çÿ’æm)ºlÙZ`†“ &>¿¢ÅìßR°b× Zá½4ϔ촊ç³kBs …¥uMªV5âõ©9âUÛ Ú°³é°‹9Ç‘˜me?{š6O9=÷l¿s\´ôñáe~§­åJûÌ•švˆYlê`ÚU9#”ª)íY3þÙwSRȆ‡bfHû9?9 T½Åq:4þòã^8éHÃ{&¶DÇ|—‰•\:PÔ~5e:µê§   …æ3šÎ" »îOç3‹…Ïltê0ýþhº$ÔÆ1”.‹ð‘b6]E©¾õ3º.Ô ëIº)¢²¯ƒÂ S¤¥;)ú¦i£HºÃã,üœÏ°”]rPļIÅ §C‘\ÿÈWm×/Þ¢‚ZÕ;3l(E›²“–¢/ü®M OŠi\lQíÉ (FŒÂ¾ó(¶ÜÔñM U§Ø¥«ÛwxÕŸbÅì±HÝmÐ#nýwþ4YD›œÉÙ£yø³í¯­y•Å+Öe+Žq^èÍ÷Kžg>fâñ[”çù¥˜/’Ÿ²|u'óTk>ÇïQçsRä}2^Zë5ÌÉó·Ê—år}dœ”z#ã¡Ô!å:Œ{Ö¾óEî?tiú¡ÙÔ•ŸëÎÏÉ<¸§X¦øk*õË•ÓgQÎoý· ú¼GO+îY뙜/å¼z8Ï{FšÃ¯1æ¸3–y=Þæ)øÊñz2çaSÇþµâË ‡h:¯SÍÜ:ã\Q×Zš •øêÎ šÇñrç·¬'‹k^;´äQt¡ù›ÎѲÛ;n"E¤•<ßY½u "ËdZkN{JPéŽmxz¾Y®9—h³Q-êëMS)xÈ¡´=×È›iÿD;Å2P¡M´7“X Dû0ˆÓÿ}x™„¹ºu÷þö 5å63ÿnbéñ³ÓÑ©ãûs`J@¡ŸôœÕ{é^:Çëþ²›‚O—L90èŠsþ‚«]ÞÐ5fÇA7†ï¼ú±Ñ’nuŸv-jü Ë#„¼+…ý“wkÈÅt'r@Ñ33ûSøºs·6¤ˆ®íºû]mŠÌ`:H‘‡Òd(v›¢Zúõ-Tø E=È*ž¢ý^>Í=c0Eߟ»khÚÓ´QÄÿn³oouL!iÒî(¤º5n+3?$/d>(óCÉ›’œgÊ8&ùU€ù QòMòUŽ{™7Êõy™Çµâç侄\¿T㬬!ìcX#XcXXS˜h¯æ°°–0X+XkXX[X;X{XXGX'XgXXWØ'°n°î0_˜¬¬'¬Lt–ø§ÃûÂüaý`ýa`aâÿ4`0l,†>µ‰!0666 6&ÁXØ8Øx†‘m"lìS˜ 6666 6666 66666¶¶[[ [[ [[[[ [[ [û ¶¶¶ÛÛÛÛ Û †m…mƒm‡í€í„í‚í†íí……ÀöÁöÃÀÂÁÃŽÀŽÂŽÁŽÃNÀNÂNÁNÃÎÀBagaç`ça`a—`—aW`Wa×`×a7`7a·`·aa°;°pX,‹†ÅÀbawa÷`÷aŸÃ¾€=€=„ÅÁ¾„=‚}ûö ì1ì[Øw°'°ïa?À~„ýûö ì)ììWØsØo°ßa/`Àþ„½„ý{{ ûöìM<õ5üGÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ó‚ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿø¯ÿšþkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þkà¿þk¾0ð_ÿ5ð_ÿµÞñÿ×ø¯ÿø¯ÿø¯ÿâÿ4Dÿ5ð_ ˆ×t ü×À ü×À ü×À ü×À ü×À ü×À ü×ÀÍÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ð_ÿ5ðßìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìðÇìÂè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zd‡Ù¡Gvè‘zdŸ‚8 Gè‘'è‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 G_‘E Gè‘zä€9 Gè‘zä€9 Gè‘zä€9 Gè‘z¤ÃþèðG‡?:üÑátø£ÃþèðG‡?:üÑátø£ÃþèðG‡?:üÑátø£ÃþèðG‡?:üÑátø£ÃþèðGþˆœú¨Cuè£}Ô¡:ôQ#2hôQ‡>ê"͆>êÐGú¨Cuè£}Ô¡:ôQ‡>êÐGú¨Cuè£}Ô¡:ôQ‡>êÐGú¨Cuè£}Ô¡:ôQ‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#z¤Ctè‘=Ò¡G:ôH‡éÐ#ýMüTÈzä=ò€y@ÐGúh@ è£}4 ôÑ€>ÐGúh@ è£}4 ôÑ€>ÐGúh@ è£}4 ôÑ€>ÐGúh@ è£=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈ€Ð#zd@ è‘=2 GôÈz-((—aÄÊ\éå7¯lŸa~ƒzóì*O—yó1q Ì-+ÎíñÇåøØ;ѵò‰Ž+$:þ(ÑqÅDÇ•Wæã2òÚ+F±‚£‘˜?Æÿ‘>Z¿9GE&þí}Ï| oUþ‡|«šè½Õ>اÍö?úåzbãr«'zG7I“c©f¢{j%3®>|Ì8S&W÷ïÛw”Áܨý³ÊW«ì]λ<þ÷ößUCz_’óÍÿÎZ¾«ëÈç×<£——_ßÞ^Ãü÷í=ܫϰ!ƒÞ~±%ï–jöCÞW‚Óö¯%bµyxêwýâ5Èo`ïêï-?å¿–ŸÒ«Oï+Çó_ËÑÞåbf/ˆto¯!}¼z1ll2-Zóÿ¸Ekþ'-jzšhü¿YR ƒÓ#üz T‹¸ÙºIÔwäðø)¯µDÿŒ÷¹—öÝKûî¥ýÿo.íK.‹1)ƨ{blŠq+Æg#þMŒÍê|,ư³uù^Á‹JüŒ¸¿£àJ¾_ àAC¾Þ„Ÿ\jÌV›ŸurÙµÙ§ÆŒ‚c5¹Œú|O+~¿¿G\<+ËeUg¼®ÃeÕç{ıÐÁ=o¾^}­ÇåVäß>æãš|(«*×£_>V`kÎ~VäßK±?UÙêð3µø¸)ß_ëð1¿³#±U⾪ÀÇ¢¬*Œõ½£!·¨Sy.£·“w¢viÄ}\‡ŸïÀþ~Ä× ²/…ø=¹ 'û]ë*Ç@9¾¿9×­±-AK…ÿ­ùYg"?ÅýBÇKpjó»>âòËó59îêsÙ•ù} +ØÆoMöG¼§(·‹x¿ˆ¥ùwo¾^˨–¨-+sYµ¹žU¸žeùý5Ëñ=-ù]¢ÞÍø¸!סס:û]‰Û¡2›ð³ÿæ´%ÄwÊq9’2ÖÕ‹÷Ë!¹lgÓl 1Z{Çõw™=™û“»fÿ—ûÔ{ÞužÜûÿ“g’«[r¾%÷ü»ÎßçûÊüºýÛ»þ­?ä÷)ïß|—}Hß7æÞõ{r÷'Wçé»÷ùø¾2?t,ü[ÿ½ÏÞÓ®"ÿ."“uq->÷(¿µ1â§2ɤõF¿ž#†ˆÜNäóRdŽšëÛ’æ®ulñÚ"®9ùºŒâZ½D÷ÖPΗ!ëòqc.¯—#~“±LÞ+¯ÉgêqYõØdÙuÝ/Îk*ÏÖLô{C~–lIóô‰~O\^åZÝDu—&}p*ï‘ï­š¨|Ù>òÞÚ|\Ÿ«'ªGr¾È#ç!õëðõZJÖfK|,ëíLäS¥d²ïë'*ÓÉï“~ʾ‘åÖMdÎD}˜¸/%&gõÿ-q Þaõþå·ä,ñøVûåC-q»þÛ}'ºWúêLäKÝD1^0Ýaûp½Q5ƃ={¼ç9ñ»®<«þžÜùûÊ–õQŸIî¹äÊ~×±¬ã»žWË_›&ç“ly¬–#ŸIì‡þžw½/V¼«ýþ+ãáßêú¡õOüÛ‡Öë¿â‡ýïûß’³ä|ýOÇîûž—ï±'¼7IŒ÷ˆÈmc´×i.šùÈ£æu¬£æù]f¡)þ§f³eÿÙfKoð1f=¹¦£Ø'6[†%ÿþ\¦k6[ÚÌpõdü¹#$éïEŠ&=O'áØ³ßûýJ»Ãf+>:á<F¿g©wß›fvÂqÞî ÇFÿ„ãŒGã1Ï7h«ÂIŸO*S¥V®ÿýï>¦(ˆ{¾xûºÇ ›-søãt™Ð¦I÷œÅ¿UL¸–Çó}m_fÌS~ÿ}ù~Hyj]ß¾– ýŸ2Û»ïÏ€knd?iÊüßõ#å„ãT¯Ž=¿DßM8Ï2cî×ÿ¬lÉ\“èÿ:&Þý¼~.á8M”q†gÇcz±âÈ‘_|ä–: >„M‘uÈk…læ‡K)îáÙQ6[ñq†X™!2Å|ŸÂ*Âð|–!x>¥Í–3ÆÝ8ŒûUè'<“u~óµÙ²MD%Ð7è»\þ8ߌëxOÚßâ7Þ3Çá÷ñÇy.àð¥ˆXÎöEŸ¦ˆ†ß[¡LÕÐÇ]ñ¬ø 2#úe/úïÊrÒfó>GÙkQÆK”}Ýf+ÙËf~ˆZ *–î{øôÞÕ=¾nžx¿—ø˜aÚ"íÝ uAý ÷ÿàOh’gãø²²ƒ¿y=å‚åÕáÃ#›ùÑBšÒfSšºä9×1 A5³CAíPß¼MЖ8.ˆ2ó¢®ÞÛm¶+Ñ]P_Ô/UOøSÄf~ ›óWøüe^ˆ¯Wön(c~Ÿ†gó¡Œ#8Âï¨w´mZèqºÊh"´¥ñ8ÏŒÁ=¸/%Ú#?|ôN“}ÿác:Q¨|ñQ5ü(†º§] ¿ûâ]|ÿq^Êð<Ú¦HÇøúåÏ  Óù/ÆL(VÑS`ææîEåFd¾<J¢àÞÀøs¢­2Œ‹ÿ(ª0žÕ+á]be c#óºøúŠXrÞFœ‚‰wüßçÁŸò°bðq ÊC{äï?&DÛ 3-Çi6ôÞc5ÇMF_¢¾iÑÏÙÑG™¼ãÇm>Ì R£ÝÒ£OòµÀsð;Ú*¦ø³ˆ±&v¼ðûyÔuˇ±Ÿæð~ÇøIƒzæÅ»³_…_9[Æëqö¼ñjz -3Á×4ð5Æp6Œ«Tÿi"ÐþèÃ<¸?C†xåA{Dâ=à^¾û(_\ƒßéЇy¢ðÜ´ãü.>ny1Fòà·ìÓYÐ÷E06Ò,B½á[Î ¶¤;hÿ©É?ÿ73,íeÚþå·ÿJÿé»þÓ:&çoâöú¯”ù¡uøOÚà}÷$ö÷ß|øÐ¾úçÕ2T>ä™ñÍö/÷«¿}Hùÿiþ[ÙÉ]û7?ÄŽö3§ÑÇüãLyCü¹éÔâÿÝß„ëñ:S|dþ±Îåý©øþ´|=ŸÛøw>—÷ÉçÒ(å©¿Ëò¤ª?ªé?äïüE„35?Ÿ–ïO­”—žÏ¥_²Ù?链ÏzNí/Õ_YïLIÛSúK¯•ñðV{(ãFú™Bi×”Êx–ïWÚÉz.¹q¦òRý]EµßÔëUSÛQÞçPê¥úOå8—å[ ?'õI>§êÿ[~(õ•ïO•L¿ÈvUãCzeËú$§/VÿQ9—(Ÿ{ªŒ[ÒsYžu.Ÿ¦”#ý‘¿¿ˆ¯‡õ~yŸê—êß3å÷Ç\Ž,_žËò'}ô3YÕ÷*õ·î{ž´Ÿè+~ÿ+¥žOz™ÌuµÝÕþûMi§?“öZ«\þÝÒãäê)ÛéqÒñeÝÿø=ýõ§â—¼Om'y¿Ú_Jý­ûe{&S/µ½¬ßÕú)ý¦–çÌ©ÔS¹ï­z?UêùìÝþ«þªíªÞ÷V»Ëò¿zwýÞzNÞÿg2ãVgÉðý­úý(uS)Wég«¿~Kê÷[:ôBig¥ÿÞ‡jÞù–N©ùÒ^V?«z¨Ž3µ“û=ݰÊUù£ŽG…ïÉŽUWß7>•úÉüȺ_Õ•“ÞÿVÿ&§Jû¼—GÉ•/Q½Oõ39¿“ÇV{ªú$Çarí)ë¡òï7åyé·¼Îh“óãbñå;Kr?å¼ ,Ÿb”÷•b,ÇXœùW†ï+ÿ>g .§_—å•‘ÏñïEù~YniyŸ,GÞ¯”#Ë/#ßÏ÷åße9¥øz1y¿¼.Ïå{d=•ßåû%æ÷ÉßK&ó\¥þ²J(þH?e{f”íQ*iÜKhoé—W6éý–_ò}åõRü|I‰²žJÿËrËÉþRν]ñç­~WÞ+ÛGöô;·ì·¤|°úE¶‹7¥”ú[~«õâëV=øziå>«=e;Êþ”íËçåø¼¨R?YŽÕî²^Èþ´ž“÷Iþ1ÊqS\ÞÏåÈö“ÏËö(­ÔÃK©§lyI9¾”rJ(ï+Ä¿K,.ÛGŽoÙ.ÒÙ.Œ…”vô–õ–ÏósÒ_µ^²ÝŠÊþ”ç²dùr\ÊúÊöPôCêŒU¾^ˆÑ*Gö›â‡DÙÎ*ïd½­q è“Òo‹Œù¿å{J+<õ(­ô§Åwù^©'êxa”ý!ÛÍÒeeœ[ú ŒÛJ;IžÈöPùYZiɧb äx•ãÊâµÂ»¼Êø.+ûCÖW¶§¬‡ôCiwé—Ôo…ïÞ9“–/ë+ë)¯[ú/ë#Ç‹Ò_’?Ò/~¿MÆãøÛlrÝÈŠÛò\Æuõwyókƒ×!¬üB>÷R޵Æ÷©÷Ë÷È ú*  KÁ@ƒ Q0TÁ0ã|šGØôTÐKAo ú(è«`€‚. R0DÁPÃŒSðiRiSÐSA/½t*裠¯‚ º T0HÁC S0NÁ§Iq”MAO½ôVЩ ‚¾ (èR0PÁ C U0LÁ8Ÿ&ÅÑ6=ôRÐ[A§‚> ú*  KÁ@ƒ Q0TÁ0ã|šÇØôTÐKAo ú(è«`€‚. R0DÁPÃŒSðiRkSÐSA/½t*裠¯‚ º T0HÁC S0NÁ§IqœMAO½ôVЩ ‚¾ (èR0PÁ C U0LÁ8Ÿ&Åñ6=ôRÐ[A§‚> ú*  KÁ@ƒ Q0TÁ0ã|š'ØôTÐKAo ú(è«`€‚. R0DÁPÃŒSðiRœhSÐSA/½t*裠¯‚ º T0HÁC S0NÁ§Iq’MAO½ôVЩ ‚¾ (èR0PÁ C U0LÁ8Ÿ&ÅOm z*襠·‚N}ôU0@A—‚ )¢`¨‚a Æ)ø4)ºl ¦TÐSÁ z)X\Ao«*èT°‘‚> vTÐWA£ KÁY *¸RÁ ·)¢àQC¼¢`˜‚± þ?ì ¼U•ÿ»û-yÙ jDÀˆˆ‘I×Þ€pCö$,5e I€°7!°'¬yKâ2šq›è8út\PQq›a\£¢¢£þqT&îÿ[ésëœ{êõ»]¯z{IßÏçæ×ÕUuÏ]ÎùÞ[Uý*{™>Ïô¦ût½9ô‹é8¦ã™N`:‘é$¦SÁt&ÓL1]Ìt9Ó•L×2-2ÝÈtÓ-L»™îbº›é¦ýLŸbú Óg™îeú<Ó˜îÓu]†iÓqLÇ3Àt"ÓIL¦‚éL¦ ˜.bº˜ér¦+™®eZdº‘é&¦[˜v3ÝÅt7Ó=Lû™>Åô¦Ï2ÝËôy¦/0ݧë-¦]LÇ1ÏtÓ‰L'1 ˜ ¦3™.`ºˆéb¦Ë™®dº–i‘éF¦›˜naÚÍtÓÝL÷0ígúÓg˜>Ët/Ó百ÀtŸ®ë3L»˜Žc:žé¦™Nb0Lg2]ÀtÓÅL—3]Ét-Ó"ÓL71Ý´›é.¦»™îaÚÏô)¦Ï0}–é^¦Ï3}é>]7d˜v1Çt<Ó L'2Ä4`*˜Îdº€é"¦‹™.gº’éZ¦E¦™nbº…i7Ó]Lw3ÝôŸéSLŸaú,Ó½LŸgúÓ}ºÞšaÚÅtÓñL'0ÈtÓ€©`:“馋˜.fºœéJ¦k™™ndº‰é¦ÝLw1ÝÍtÓ~¦O1}†é³L÷2}žé L÷éz[†iÓqLÇ3Àt"ÓIL¦‚éL¦ ˜.bº˜ér¦+™®eZdº‘é&¦[˜v3ÝÅt7Ó=Lû™>Åô¦Ï2ÝËôy¦/0ݧk©™D»˜Žc:žé¦™Nb0Lg2]ÀtÓÅL—3]Ét-Ó"ÓL71Ý´›é.¦»™îaÚÏô)¦Ï0}–é^¦Ï3}é>]oÏ0íb:Žéx¦˜Nd:‰iÀT0ÉtÓEL3]Ît%ÓµL‹L72ÝÄt Ón¦»˜îfº‡i?Ó§˜>ÃôY¦{™>Ïô¦ût½#ô‹é8¦ã™N`:‘é$¦SÁt&ÓL1]Ìt9Ó•L×2-2ÝÈtÓ-L»™îbº›é¦ýLŸbú Óg™îeú<Ó˜îÓõÎ Ó.¦ã˜Žg:éD¦“˜LÓ™L0]Ät1ÓåLW2]Ë´Èt#ÓML·0ífº‹én¦{˜ö3}Šé3LŸeº—éóL_`ºO×»2L»˜Žc:žé¦™Nb0Lg2]ÀtÓÅL—3]Ét-Ó"ÓL71Ý´›é.¦»™îaÚÏô)¦Ï0}–é^¦Ï3}é>]ïÎ0íÒ4SæïB[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[ÚÒ–¶´¥-miK[Ú\šÉ´ ÿ.4üÀrr¹ídõ½Úæûùùü<þ}¹rLÇq;åìUz^¥å–kg9{åú±\{Ëõ_¥ß—«o¹þ1µ»Üq¦q6ùG¥ã\®Cõ—JËIê?¦öVjÇä/I÷—OÓøµ¿+­_¹lòÛríLI¿7W)ËÅQR®™âÊÄ«JãÞÔž¡úçPý§\®´ÝåúÓTŸ¤\0}_)*õ³Jç‘rí4Í&>Wê§•ÆÓPãØäåʯ´ž•Ž«)›Ê7õc¥ãYiŽßàN?·tÒ±ê¤,ìî¼hÉ…k._%?ý=SáëwMÉt|òÇ¿ÉìÕþqtºú$MÃíüZÚý¦4Üü?iùiS½Ë¯vyÕö¯fó—´åÕúøj÷OÒÔèöT;ÕÚêÝÿµŽÏ´i¸Å{ZûI÷§-¯Öã[ëú¤MiÛSëú×ûøzÇsÚúT»¾õæA³_ëþLZŸjÛKz~³ñ;iªv<hýÛì|­÷üÜèùì@朗Ú<0•_ëù9i}’¦Zó®Öí«÷|Yíú¥=Þt~£×ž/“îOk¯ÚÕö7Sj6>Õ»~õæw½ý§ÙæSnñÔlío¶ñOkÏ”ª]^£Ëo¶úV»~µ®O­ç/S}’Ö7éñõno³ÍIëÛl<©uª¶¿ÕÚ^Òòë=ÿ¦­OÒú&MÕŽïf[oÔ»}ië—Ô~Òú5ÛüoÚ_ëxÌÀß…ŠiÅý‘Îgº°_×E] ]çu=[è:'£ë ¦ ™Î(ê:;£ë¦±ó åÏgzv¿®“3ºÎbÊíÏezfQW~þ¢®g ]ù~^n寮~¢¨ëL¦¼¿yùg1åýÅíñöÏËèzV¿®SŠºrÿãçOeÊýw^B=3£+ïo>ggt^Ô•·Ÿ?»¨+¯ÿ¡+/>Þ¼xù\MñÍý#Ö?E]ùxÌ)êÊý‰ó€Û‹Õi,ž‹ºòxããÃÛãk/Ÿ—ÇëÏùÂý^_ÜþÉ]yûxÿqÿ:«¨«)þbñÔ÷ÿ|<¹¿™êkš8ßLñÊã…ׇ묢®±ú3Õ—µû'>>§etåñx:SÞ>î|¼9ïy}x|ðýÓ2ºrðþáç2ºrÿ;5£+÷?“ðù‘óz SÞ|~ŒñDèÊ×_üüØz £+o?žÇ?7nŸÏ/<¾bü`åÇæ'¡+ÎÇ™Ly|òþäýÅý…×wjQWî±õЕ÷W/±õS¿®¦x‹­wXýyûyÿ™Öç¼¹¿óøŠÍ'BWnOdtåýÁùÅûÇ´ÞãëEΟÜß9¯ùõŠÉ8ïLÊÛÃË7ñ”×·çl¦¼}¦øå<äãÁû“û›i=ÍLJï7­/b×K¬~œÜŸcë‹¢®üøØzµ¨+¯ŒßE]c×›Lùù<^¹¿óúóòc×Lyù<žcñÝÏ”•wFFWîÜøúß1±ý¦ëÞß±õ+×76^E]cë¡+çY}Móaì~@QW^÷>Ÿr¾Ä®“3ºÆêo¨ÏôŒ®±þê×5æÏL¹ÿóòùúˆóÐtýbŠØúŒ)ïØüŸÑ5ÖŸ¬¼Ÿ‹L3ºÆî·°ãy{cë=fÏt¿À/†ö›îwðý±õwBÍ_LM×Ü_üzÞÄߘ²ò¹òëרýZ¡kìz _W>ŸÅÖƒLyûbë팮±ù±¨+÷?~>îO\ùúÇ_ðxá|á×±x(껿ÁÚ{S±ëv¾iþàí‹­C3ºòëÝXý…®|¼bóKFWÞ_±ù‡)Oï¦ëeÞžØõ/S¿|þ0]pÿŒùoQWÓú‘Ç'¿ßÂyèet5]¯™Ö[|ýÆçOÓz9vÿ™iÒõqìþ'k/Ÿû¯¯éþ>ŸObëQ¦¦çIÆõ6S>~Üÿøú,v¿&£«éþ–©¾Ü_EFW¯yÿÄ®çØ~Ο©E]ùüÆýƒó—oº?fš¯L÷§ùü»ž,êÊǃǧéù¯¿é~6oçolý#t5­·cëyV?~šóδ^‹=_eÊûŸÏû‡ßŸŒ­çû™u=gÇóxàñdºÞ]ÿgtåü3ù—éþéy‰W<þbÏ3ºÆâ“)ç?žû_ìù)ÓØú†)¿¿ÈçoÓó}ÓýÓzÓô|:æ_¬~&~›®÷c÷Xýø|Áûßô|Ÿ×鈴ço]9¯øñ±ß]MÏ£ùõ ÷çsúu=WèÊëcZߟ§ftLMÏbó»Ð5öü„)ß S^~Œ/¬þ¦ëEîo±8*êÊíñúóßïðù€óŸó•ool¼3ºÆü)·ÇLJ¯ÿcÏKX{øüjâ ïOÎϤ¿·âõ]¹?ðñãñÁû{jQ×ÉLùx™ž¯Ç®˜&å1o·û=YQ×X|°óc둌®¦ë¿X2åãË÷›Ö?¦û¡œw¼½±ß)et5Å·7µ¨+o_¬¿SÖ¾Øó¬¢®¦çåœ|~àöøõ¸‰ç<~øù||8L¿×‹Å«/ïØïdŠƒkìù>«¯)~cñ’Ñ5v?žµßt½›2ºòúÇž¿²úqÿ­·Ùñ¼üØý„~]M÷Ûcëe¶ŸÇ¿»É4öü…)÷O¾ßô¼9¶¾eû}SxòõŸoù~>^¦ßñþã÷sM¿ïŒÅ7SÓï³÷?˜Æ~OšÑ•ûÏ›2ºòúÄxÎÊ3]ïðúÄڛѕןé~b¬>BWÓïÃb÷ï„®|¼yù±ûaE]c¿ïו¯—M¿?æ{¾Ï4ö{Ý¢®œ"£+¯_¯›ægÓñœg&¾šxËý5¶¾/êjZ›æg^_ÎÓØõ^¿®œ÷œ||xù<^LóIìùŠÐÕô<&vÿPèjúû¾~е§¨+ÿ½,_ß›~lZ¯ðþŽ=ïdûùxòú›žOšž'òøåãÁujQWÞ>~½ã-SÞ¿I×c±ùKè{nܯkìþ+ûÅÔô<7v.£«i}ÆÇÇô{~¿Æ4Ÿòñ‹ýÞ’)oŸé÷á|üc×7¬½¼þ&^ÆxVÔÕô{GÞ?üþ‚é÷ ¼½±ûýºÆž35ýŠûŸï9¿Më¹Øz1£«éïѸ=Óï}có;S^¾éïLÏçb÷Ä®±ç‰Lùï7¸}Óß÷q­ïØ~ãýRf?öü„iì÷ EM3ðw¡™„©ÑïnJµ~ŸEZ{ÕNÃí}¦Të¿ïnöö›ÒÞþFÿý~½Ë«wù­4xªÁûš*Õ›õ~IÚÔlí;ÐÞß’öø¤i¸ÇkµS£×«Ã-5Ûûjm?éñ›ÿÔá}MuMÃíz¢ÖñÐìýQïò›Ýyªw<6ºš?µ¾3ÜÓpk³­ç[©¹ÒÎÛáæÏÕ¾ÐJƒ§fï¯j¯g›ý~\½S³Õ¯Ñ×?Nµö÷fKÍV¿ƒµNýýÍVÿf«Ïžê}­Ö×'Í6_TûùBµS£û§Ñé@¿ßÜlã;ÜêSíë­á~ÿÅd¿ÑñÓâgº4Üû¯Úë‹}|kÝ^ßÔ;5zþov~hþPëñ¨wj¶ùš;hý Í§­øi®Ôèë›F?W>Ðâë@OÃm¼}¿d¸§VÕ65z¾otyÕ.ÿ`»^ªvû}?¢Ú¼nöñl¶ù©Ö÷/ZéÀNÃ}¼‡[ý[×óV:ÐÆk¸ÏÃm~®·½f럃=5zýÜèño´ý¤i¸Õw¸¥F÷ok}¦§f¿TíÔèú4Ú~ÒÔlõmôú¦Ö÷cÒÞïK[¿F÷oÚ4ÜêÛJƒ§áö¼¤•ôÔèþ®õó˜Z§fçu£í×:hüi¶ûµ¾?RƒxÞÿw¡âˆÒ†R^Ž8Šéx¦Óc™šÊÏèjj·éxÑÅtœ®±ãY{’®ccå\MÇÇöŸ kÚúÇú'Ôí7–gëj¯˜?±ýIë›Ø?x}¸?3Ÿ°ÿïçöyý’ú×Iº&¯˜ýã˜&V/bþÂÛm/ÏRÇË]ýi¿˜Ï”ïgýwÓ¾ðß½JõqtåÇóòx{ÇW±þæþÆËãöøùIùhª¿i>3ìO=¿%­ïæO5¯_ÒúN`ÊÏ7ÄoâxH8^¼>‰÷³ùœ'~~L þ´~Æx4í7Õϰ‹oÚoRÓ|˜òü´|7® ã)DQÓŸÙñ©×¯)Ë3ÅŸ1þMó?«Ÿq>5Ïö›Ö±òLñ9)·ÏË7ñÒÔ¾”óQZVûø˜x“úz´þãýÏ׳|ýfª/ßP¿´ë—˜?šêÇ×ï Ûcì_“òö$嵩¾™&¯¤ó¥¡=1âÇOÐ5u<'íOfŸ×/ñx›æC צõbj^%ÿ„÷ÇŒjò¿ì|#O ׋ÆòLã?iÒþIê¿l¼b멤ñÏûŸµ'íú˨ ïo¤^ÏOÐ5Öüú=ázÆT¿˜=Þ†û ‰yÀ懤ý—ºÿù~®ìúÒt}Ó L9Mý“pþŽÏê«_BÿH|¿ùD¦§0åý‘òùAÒþ1žŸ°bû ×[I×?‰ï¯¤½_èš8ÞLó·‰/iýÁp½`ªolB~ÛËëg:ß4šx’ÖŸ®7Ó®ÿŒþ5©éþç¡é~{RÞr{¦ù9ÏÔÔ¦çdžûû©×G ï$å-ßo¼ßlª¿á~=?ž§¤×‰×£I燄ýmZ?&¯Ä<âñdzþÄ5áýxcs5ÝOäþÃÇËt=lOS|'^ñö˜æ+C{b×»†þH<¬>Æó ó[âù,áþXûùý­„ã˜gIç^?S{ØþÄüMx¿ØÈ Vãñ¼=ì÷4‰Ÿï™®O™=ÓõsÚõXLMüâýÃ¯Ç å§^²úÛÇëkš/M÷; ë•ÄÏÓx}¹N`jâoBžëg¸žŽÙ7Õ/éó„*?/Jí¦ù>cФý“ôþjÂù9-ó8éýmÓüeº>åÇ›ì¥|že*ߨ ߘ¶þ‰ãÐÔ~Óó¦Æú›®¿“®&0Mêß ý-éú<ñï í1>Oæå¥õÃõ°Ió™û‡ézдþ2\ï$^o˜ÆƒŸÏ4©¿¤¾Ÿhú}3_ï™æ“¤÷#L¼7­®LåÇöO`šô÷IycòWÓõzZ'½Àû7¡}ãïËÙñ±ñ7ÝŸgþ›t=cŒoΟ¤÷cyy¼?M¿ïO_¦ò¸šî%äeâõ¯?ïÓz=)Ÿ“Æ+k_Òßs'¾Ÿ9‰iÂë!ãß/™xœò~Pê뙄št¾7ÝHú<2¶ßàI×»±ë«¤ãcò?~¼¡|Ó|{~–ôz$éý*Ó|–0þy{cjÿ¤þ’¿ à{-%åªI«†j¯Zõçß§m½û»Zí¯–½´å§MµöçrçW»¼jíO[^Úñ®VýÓÚ¯´\SªuÿUjßdo¨õjýkÅãrûÝÿÍiÛ]m%åe£û/©½Zõwµë“Ô~¥å'Mi9’6U»¼jÛ¯U\WË~µ¯4Õ‹iS½ySîøJËIº?i2ÙIZN­ý§Ú|J;ÿ5ÚŸ†Z^½Ïjÿ¤/«ÝÕÞŸ´•W­þ®tµ×µîÏJí×{}Ólë‡j—_/¬V}“–—6U›wÕ¯´çךÇiϯõ|¶¿ê¿ÕNµ®ÿPë“ôøzùiž-W^R{iëWm¾µ>¦òj5>Õî÷JÓPëW+7:¾kÍÕZûOÒñªõ|½ýÑT¿–¦Ó Óz÷µËoôúä@÷ç¤ñßl×ií 7þU»¾¬¼ªßOJ{þ®ƒ6º>ÏjÛ«uý“>ŸjòñmøýƒM3Lëm¿Ö×ÇöÓ|×èñîþUãòºñªõø4:ëíOÍ~=Ùèþ1•Wm{IëcҤ㟲ýÃŽGiOÕèú§Ô¦¯ Ó´ç×úþLÚú6[ÿ¦]ïšÊ¯7Í“FÇC³ùW­û«Ùx—a:ÜÎOëÿ)í'þ;¹f‹ßFûÏVßj—Ÿ´<~|µë“¶~5¾_Uóß·$¬OÓùS5qÿ7Ïš}<ùù&m²þÍÀß…f&ÓïèMÇ'-?myµ.?i4[Jú¾Œj·/­½jûWÚã«m/iûíõ¶ßlíotª5?ëê¯ÍVŸf÷ï´þÖèö4ûür°÷éüá6?šR½ë[ïòë½¾LZŸ´åÕ;5[|&-¯ÑýgJõæy½û¿Úé`›/_õ¶×lói­ç“´å5z¼›Íþp_¯U{=Ólþ4Üã§Úóo£ã¿ÙÖÍŸ®O³µ¿ÙS³¯íOiÓpŸO=þµ®OZûõnO³ÇC½ý·Þç×:èí«wjt|5ÿMö†›ÿ 7Öúz»Ú×KiS£×¾^Mšêݾz÷OÚòkmß”†[üÖûþF½y×lõ©·ýZ—Wm켫öú Ñþ]í4ÜÆ3ijôú¤ÖñTïòê]ßFûW£×ßÃýüZÏ?Í6ß5:>ê½þ¬uj6¨õùÃ-5›¿˜Ò6¾ÃÍ¿}}šôøf[Ÿ5›ÿ%MÔÿß…V›û§jöu{­S³ùk­×™ÍVŸ´üh´ÿ5zž­÷uY³×p·—65Úÿ“¦Z÷o³÷G³ñ§Ùüù`Oµæ}½SÚxl¶öÕº=Ͷþk¥d©Ñþ[ëù¯Ùüi¸Í§Õ.¯ÑãQïëŸz_¯¤­oµO›š}>>Ðì5ûøVÛÞp·fç­)¥íïzŸÉ~ÒýÕ>¾Úý1Üü'íù¯¤i¸ßhÞ7[yÍžj½ž­výÒîoô|RïÔl<ªvîëñZÛo¶þi4/ª]Ÿ´ÞõNÃm~jôøhí©vª7¯ªš}|›mýRëõľžN{}‘tÒúÔºüzûcµû;mª÷ú¨Ùâ©Öëó´ý[ëóëíõæm½ã½ÚÇ×:Õz¾L›jÝ¿^Ï4›?$Mõæ[ZûIËk6ÿ«7›m½bJØßÿw¡âáÏíß÷ˆ’î:F×{Æ—´X*H¬ZPÒ˺Kº~ ì_TÒǶ–ô.(ïZÐKàø]/0ý²®ïyV×{t}Ïwuíž]ÒÍ»à¸O–´ç²’îX¯ë{Ž*éï/é5‹Kºr´Ú·ôN°»noIß›ÓuÇ—u½Mõ×Ú’Þ:±¤Wö—ôIK×÷éúØnÐë¡™’¾C”TÓ*¨ï»aûÂî’.ç½g®®O>¤ëcSKzý3%íû|NI„~Z·ÎëÖõf8oZúáïºî|¯®7ÏÔûã.ؾi/”û~]7,/é› ÜåºîÜ©ëö¹ºîœ¤ëŽç˜~B×;Jÿ¡®Xý÷Œß-PßGÀ/î‡~Ûñ(Ó‰L/Ðõ>¨÷Uý¼ËÔóV]·ÿV×ßÕõ†bI/RåíÒõ†î’.Ýq‘®WÁøçoÿ„®;æêºúñ:Ð]ctí†q¾ú©û‰’>ö|I¯;k&Ày?Ñõ~ð»uÐ÷C¹wž¼O×Û¡7¤Ö1Pÿ[Áβn(gÓ{t}¸K×íÀ‹õ{Kº%Ðu=ŒÃR(ÿZ8N­ŸvŽÒuû^]w¼L×?Ôõ^¨Ç-ÐÞí™^«ëŽ1ý¶®¼?ûz™®Ôuóâ’Þ5Îcñ³üõj(ï*wµ>¼¸yU±¤jZ˜v^¦kß_uíž^Ò'þí¸J×Ͱ.ºüñF°{Ñ^(ç!]_®%}¸v¥ò7¡kŒï%ª¿Îѵ÷mºî<¾¤›Àßwܧë5Ð?W@ÿîºöݨëõÐ*Þ”öž©ëö»tí;YWÞ}K¡ ý*U{ïƒq¸Ʊïºîtt½ê«®¿®[ ÛPÿÞßêºn®;Ž+é­Š#/–ô (wøÃ&Çû!KÝ%÷¿¹¤Û`ýÑ{¸®}Ëuå×k;fëúÌ+¡þ}èºÃbzˆ®÷ƒÿl‚úm_ kïb]w| ¤ëa}Öw„®×Áyg@yÂúq Œã#G1…zÞ¼Ø q»^ùß"]ßÝ_ÒåÐÞ`g Ô¯o·®ë`}óîî’^ åªø»ê{”s)èÙÊañ¾ã$]χãΆz­‡8W×£½ßÕµïQ]{—ëz?ØÙóåÝ tåñÚ óã6h×u0o¨õDß‘ºö<£ëj8oÌ[÷ÀºâÚÅ%U×7Ã÷ëÔzö÷}W×KûuíéÕµï]·Bm†òï„ñº]ÍßÏ3ý”®=ÐõVh¿Z¯ntíû~I»þªëÐ.åǽuåóÏöºîØ õí›Äô]Õuõ‡ë¡¿o‚v<¦ââ­Ô¨wKºÖ{üúøf¸ïrÔgû×JºÊ_ýñþp]û¾­kÏ›uíƒyqÔsÇ ºÞ! ~ª_>¦kßx]{?¤kϵºîü‚®}ÀÛ50^›¡]ê>¿^ß±EW>Ÿòõi¬sîÿë¾C×GÇéZ,–ôí Û?Æôy]Þ=þ×{¾®7ÇÞzÀxªû`=®;˜ªëæSáx?Û€¿›a¤®£ßµ·¤w¼º¬¿¤j½¾^ùýsºÞv—Àù}ïÕU]¿M@=§ëºÎ_ û{OÔu3ÌÃw¨yY]otƒŸëz÷â’^ íê+êÊã·÷]{Æ1eþ1´ó"°sí]{×èzwI/€öÝã¹Žß ×)×Aýøzä† Ø…öªë&U‹…®}?ÑU­+ßõ¸¶WC9WC9j¼ïÿ¹®‡ûèÊ×}—èºÊ¿PÕ‡ùÃöwéÚwœ®Çéõë[£«ºUTûÕµçE]ï‡û€7·‚ÿÜ\„ó]ÕõóÕоÅJªæýÒ2,#fÂùÝ`çF(WÝß¼ øø8Œïj^ÏôT]U¾ìôž «º^x;ìßÑ©ëö£tí{V×Þµºòñ½úëÒnÝÞÐ~Þß|ýuœw%Ô¯ç"]{?¢ë6¦j~QÚÝ¥ëŽv]{ÖèZ„y1Z|Y×[a¿º/Üsœ®}¿ÕõQ˜'„ý·A;ÕýÎîÓt½µXÒëà¼Þ-ºÞ±\×Þ½ºÞqW„ú=8W×¾~]7‡oQÏÀÕ}íËDIÕý€ž_ëz§îǽêÝÿƒ®[—ëÚ÷”®kTÿ@oZ\ÒÛ@ùõý•B×­ÀŸ[ ^z™ö|M×Mêù ”ÿôßfèußxé°ËxÔ q± êý èÝPŸ¾½ºn}A×îtå÷§¶ÁýÚ+¡?îƒçj=v¬3Õú¥úe ¬¿Öü÷Ö~¨»Þ‹õOF×û ž—Ãölþ\ízìïÝ£kߺ®†z¨çk=wèÚwŒ®=³uÝå^¡îBœ\ºúÆm´çQðÓÍï—M(©ºÞy`£®êú^Ý_W÷i”ö¯ëf¸.¸ÎW×uS¡ü{ŠÐ?ÐOëaÿ-0njžyÖsj]}9œ§î»¨qÜÐ]Ruµû¯Lû™¡kÌ›Ýo½ßÖu;Ü×SÏ×îìÖU=×]Ñ­×gÌëa\®½Ús;´¿ÇÓµûy]ï©õ?_ßßþx#ÄY¯§kÏn]/€vÌ/ÂqÖõV઺Ýñ=¿‚ïo†úõpÝ¢kï3ºö½¨«šÕüø8è#0ŸwAמOèÚû]‚òîÿã×ϽpßàhG÷eºöÿR×ËŸÓUÝ/Pªžc-z߸XWu½±AëGtí;DWõ;¥ê:%º?yŸ®@¿.›ßÃuçjÜÀ¾âQO'Ó#uåë—îuݶI×û`_½I¯÷&ðÛž}ºªqºæ‰Û ŽÕ}ü¥ÐNuÿ¨÷x¦Çêªî*íy—®Âº­å>÷ ïV÷à~ÆýÀ³ ‹Kªæ—ë ]o†óï‚õ€Z/ölÔµûº>èÚ}œ®ÛöèzywI/zlû«®Áþ«¡¶ýZ×î瘾M×^xîtþ9]ï„vªçÃÔ} ÅÃuÝ qp ÔçÊÅ`OÕïeº>\¹EÕŽ[ õé…û&[Ôu Ü·ºâœ?_èþ&S6_÷téÚ}=S¸uøQ÷&]{àºôëèz7´#Zÿ@œÞT„rø|ÓóÓçtÝõ¼ìôÂ}šGá~šò«èzt’®7ƒªßã<õx@Ý…ëÖm_÷77@y[`ü¯V÷‹Ïïrn;7(Nƒ®ƒyåú߿¨ç†=ŸÓõfX§¬Z\Rå/ªÜÞñºn€r–ï…ãázD]'^Û_Ru}¨žç+½ÆUýNó¹çD]»¡|u¿„Ï×ÝŸÔµçi]Õsåä÷tU¿º¢XÒûÀÞ°·ùi]U»ÎƒvªõR¤ìþDwÀtÓ™º*.\ ý¤âVÝÍŸ‹tí9D×î‹tUσ•ªç~‹ =ÝWéªîǪõõ=0Îo-ܾ £«ú’z~~ø×5°žî¿zÆïˆsuÿPÍ—-†v2UÏo¿¿âsŒûÀ™á¸íÏêÚóS6ö^Ë”]_ÜÚ õÞÒ®kïN]{®Òõ‰tU×ñW€_n‚uÍ hïãpßæ!u?¨»¤êw]*ž¯‚m>>Û¶ëÚ}ˆ®ÿŠ{¡|ö<ó®ç|]»á¾ØCPßž¿ëª¸¸ô6°S¿¸MýNúã-¢¤S2Ð~6Ÿ?õP×Q<þ/;óŠpÞsºòûê÷Œ‘nÕõ‰SuU~²eŒ¾­”¯÷néÖ•¯ß·^¢«º®R¿ÏRÏUmU_ÿÝëtÝz½®|~Ü ~«î‹lÝÃô¯º®¿Œ~ßø}]ùýþ|õaðµ] í‹ø÷ÑÜ í] ýÓs”®*¯ƒúðõz¾£~¯³¶¯v/»ê÷uÛ¦êª~¯®ÖÇ÷A;Õó“mŽ®Oе÷׺^×_Ò•àß½÷èz+ĵºþâÏçøzé6(GÍßWA?­v>ß—ªŸøú{%ÄÓypü-°Ž[ýs|5ØY ߟ'Jºu“®üþ…Šû·Áy=oÓõ&h¿ºžºT]ÿß~~”³­]WλûÔï`¾½c±®›ÀÖ©ß“BùT<ý¯®O§ëÖ º^ýw.ôÇzà™úýZW©ëüîïêz7Ôã6Õ?ð»ä¶Z/ÀýußN©ú½ðd8ÿõû´½PO˜ç6Áõàsu-Âzt­¿íºn¯ë:è'µ>QÏn„þà×ãê÷-‹Uûá¾Ôµà¯Ás—kÔu´{ÏçÿØõâOt½©»¤C< ëaõ;ê'vëºúk1¿T=÷è9é(]7@=•ªçjRÏ÷¶¨ùã ]ÕïXÔºT=7P´Æy)ôKÜÝÇóõóàË QÏ-ùú°ûWº*Sªü[i÷ºÎÍèºæuu½yÓÞ’^ õáë¯[à¸wÃqê÷ÂJ·MÐõòÅp<ôS÷‰ºÞ~§þ΂?_á÷®€r—÷—ô¨ÿ&øþ¸Ž¼ Ö•OôëzŒßb¨×mð½âýãGèºÆS݇éþ­®O@9¿¨û^Š/¥e ê=0®êú¯÷]×Bykö–TÝŸWÏõ{áï<6©û°îÙ ýÐ}—®[·êÊëß]»ÛuUý¢î'ªëô‹ þߥëÀÃ+‹`÷ºr^]Ð_Ò·Bùê~«ºŸÙ½œ)\g«uÃÝÐêwe|=µþ)]Ù ýõ|âsºòõ‹º¿¤æ·Í§êº ¸V„yÏ?ï%} ôÇÅ‹Kºêq7œ´G­g”n=QWõwJ¿¹Ö/j½¦â䉧uÝz—®›`½]_®×U]?EÏÙ NÕsã5°}6lo€mõ<›Û{ìÝöîÿZ£î/uíþ¾®êúJi÷·u½WÝÇ螤ëÖ•šfÔ߅†úûÑÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvk»µÝÚnm·¶[Û­íÖvsmg2í¹LæÈã ”9·b)|êX½æÊ¥×¨¯W­…Om«—E—\¼ >v]~Áê5«®¼p lw®\¶ê¢ËW©­%K—/[¥Ží¸ìò¥ËV«]«×,Yse´µtÅE-‹N[¶vͲËT‘í«¯\u1-ÂQ;Ö¬¸4*{ÙškV.+µí5жgÀÎl^}°Ô[}P…e]õÁS|õ!P ªWò“¢Oùè“}²£ONôÉ>yÑ'?úDŸ"VdÊlX‘ +²aE6¬È†Ù°"VdÊlØ‘ ;²aG6ìȆÙ°#vdÃŽlØ‘ ;²áD6œÈ†Ùp"NdÉl8‘ '²áD6œÈ†Ùp#ndÃl¸‘ 7²áF6ÜȆÙp#^dËlx‘ /²áE6¼È†Ùð"^dËlø‘ ?²áG6üȆÙð#~dÃlø‘ ?²D6‚ÈFÙ"Ad#ˆl‘ ²D6‚ÈF!²Qˆl"…ÈF!²Qˆl"…ÈF!²QP6Úò“&áÇ<~´ð£üèâG?úø1Àh-Öòh-Öòh-Öòh-Öòh-ÖòhÍBkZ³Ðš…Ö,´f¡5 ­YhÍBkZ³ÑšÖl´f£5­ÙhÍFk6Z³ÑšÖ´æ 5­9hÍAkZsКƒÖ´æ 5­¹hÍEk.ZsÑš‹Ö\´æ¢5­¹hÍCkZóК‡Ö<´æ¡5­yhÍCkZóÑšÖ|´æ£5­ùhÍGk>ZóÑšÖ´ µ­h-@kZ ÐZ€Ö´ µZ+ µZ+ µZ+ µZ+ µZC–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰…,±%²ÄB–XÈ Yb!K,d‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–ØÈYb#Kld‰,±‘%6²ÄF–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰ƒ,q%²ÄA–8ÈYâ Kd‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–¸ÈYâ"K\d‰‹,q‘%.²ÄE–xÈYâ!K²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–øÈYâ#K|d‰,ñ‘%>²ÄG–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdI€, %²$@–È’Y KdIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%dIYR@–%…ˆ%íùILÂÏyòÙ"ŸmòÙ!Ÿ]òÙ#Ÿ}ò9 Ÿ‰Ý<±›'vóÄnžØÍ»yb7Oìæ‰Ý<±›'v-b×"v-b×"v-b×"v-b×"v-b×"vmb×&vmb×&vmb×&vmb×&vmb×&vb×!vb×!vb×!vb×!vb×!v]b×%v]b×%v]b×%v]b×%v]b×%v=b×#v=b×#v=b×#v=b×#v=b×#v}b×'v}b×'v}b×'v}b×'v}b×'vb7 vb7 vb7 vb7 vb7 v ÄnØ-»b·@ìˆÝ±[ v Ä.áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^å ¯ò„Wy«<áUžð*Ox•'¼Ê^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Y„Wá•Exe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^Ù„W6á•Mxe^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^9„Wá•Cxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^¹„W.á•Kxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^y„Wá•Gxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^ù„W>á•Oxå^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáU@x^„WáÕþŸ”‡/µ”åq$&óá_ùÈ/•ùe2¿\æñ_Áò‘eò+Y>Jæ£e>FæWAž ó«Èǖɯ‘ù8ȯ…|<˯+“'B~½Ì'@~äÈo”ù¤ ó?•É“*È¡§†jC½Ò%9ôÆÐ ÈÈ'Ë|ŠÌ§Êü&ȧÉ|z¦ô¦Yž'—Éá«Y§T§Êxs®•«žÛZ9m×W¯U+±ð»p}4bå%K.\vÁå°9jå5«V,½|íŠË¢—¤¯Y¾âò5ËV.)³”ë¼hÉ…kö¿4=\Æqª%]¸Dk‡Qu*ç_ÔÇò¹$uWck›NC®tÌÊ?Ðø ÔŸ”á~Ê'“Ÿ T¯$~6رÔT]U=ÛÉ~ÕÞÁ˜1˜Í|5©ÿÒãËõ%ñ¥>€lò³Z. ”“?•»ÜÇ»#ÊäJÛ9”8åujIrë§|f°±0ÅS¸¯Rn¥™*‰Jç*7N´ÞŠO%uߦþ’ÉT#±RåÊçö©-^v¹yˆW®}ÅR¹¹€ž; Vj“—1ÐñaYa.g³«‚Ìû—~Αý¼Îåú33H}Ò´3“ðØ|¶’þ mIz.Oåúi°¾(¾ùX¨dZ#ðs“Ì*óãUâeWÒ¦ÁêÎŒOaêdÛÊVx»}ù>¼(±9Ž|}奰=Ž¥ûÕyªnaRí}Åu:ŠÞº)+¬oÔïHR×ÐÆøÊ:š|IÊSdžõ ¼Ž=–¯ú³“lwf06B Qu9¾?œ”ñÐÃHa]Ã[ê69.GÚp Ô#¼¥ÞºoŸ;p\8>‡@ÃG¯'ç«~ öÂ1y5œOåx.çzûÇ©ÔÁ¶ÃcÂñûôHòÝHhOXÏW°ãGdпÂýGA{ce«õžò§pŒ^ ǽŽ7 4<öøÜö‡ï^õca"s)#|òOäÜCà¼ðœÐ/Â~T2*I4Õ|¦bC%å9²ý’2å…mSc«¶é¹aß¿†¯i…ý5¾ Ç•öcØ&õ˜)Lã¡3ª ¯$Û¡Í#H9dz}<µCÙÿD¾SçP>©õ•ê'5öáX„ãØª|çp¨ƒbSøY±I}VlzyFgÓø Æ¢zÄ—͆߫Gz*‡ÛÇdt6“˜MÇdt6©ÇÊ×Î`ÿs6MÌ Ì&õx/´÷°ù†ŒÎ¦3³éÐgÊC±Á®›‰³)üî”LelzSf`6‰Œ™Má1™ ²ˆ^_ÓØé‚ïÕµËh°§|¶ê7Êæl:630›Ô#V›è#V›N€þªØtRÆÌ¦“àüpœ¼ ²É˘Ù>F †Lé:I=Z í<$30›Ô£r¥9èÅ&õ¸œ>§l:¶_ çA6mälz]fp6…û)›T ¼>ƒlR¸b“zÔËàãlõHš²É‚>>ô”R?e°*­NÉ÷Ñÿ‚Zæ8¾_m—;ï/÷½IÅÏ+g×Tr©\»+µkl§¡ürõávxªv=+íWS¹Iíòþái¨õ¨VT«¼jÙçÉ·åÎ7ù™éørãfâEµÛ?\Ê5ùAµìVOµöKž’úKÒzñù°\ˆ2å–›yª67+-¿ZãUé|§’‰#IëSîvú Cÿ&nÂ\IƇ [ƒìŠ9[|,âm¤ªñ:±ƒùï̸׹±ƒ©;=ÞœAªQ‰‰ø®Då$ÂW¢ƒ8¬åA"7~LƒÑ ð¬„HÑ·¼Q;vÌ S@¼ëÙ7꞉“$QäÆËÄÙ¬òFÝØ®Aúyh¤­zñŽGe¢){h¡7ˆ‰A& ø7ñjT)q7dRÉ|:È¢bðÄÃõê ²ìI ½DÓvÕvuŸÌéÝJº¹’Y=÷ƒ N"üTk²J4’)KN9‚‰ÎÚ„?´¶W2•8´jT½b•”‡ñ Œ©ƒû%꺡ùs%ËŸJR-ÐV ƃ<Èr¾Z±S-—š­DáÒǪÏJL íb½Z^7ÈW®¦ìÌjÍMµ‹Ê”=Vɇ¡-†6çVk˜ %"õøÄ÷z3Éö«í²Ï”Êì¯ô{1.ÝþJë_iû+í‡Jë-à—^b|²ò’ŽŸÉþPë™´_«Ý?ÕªõðkO1~)ŽZyÕ²ŸÖO’úµøRk?­UV«\UÞPy™Ô.·W/»i•×»Òz¥åFÒú™¸W­ù5m¼ðñê|”¶Þ•ŽÓPû½\y¥¿6?DýµyøÛÌpA•[±>\³jÙ’5—.»l |ÑyÙ•—^°llµ¯^q­úöŽUË.¼RíèX½fɪ5ÑQk._©Î—;Ö\¹Zm­Zs套G'­"ŵ/“–Jõ{MX¿ðwй¢üçÿøûw+}á‘ö²£Îw¯XzÙ²k2¥W…¿”kÆÌܲ}Éû”Cî[ÜÇ8”Ëô8¦Ç3Èô¦'2=‰é$¦y¦S›©ÃÔeê1õ™L LOfz ÓS™¾‰éiLOg*¸uœaZÔõŒ Ó¢®S2šfÀÏ¢ñVý®úC›ê/Õ®ëJ)VïÕÛ£Æ]õßeÝ%Uþ¥üè¦ôófCý–ƒ*¸Pèç¯y¦¤K \UïEz»Ä°!”Wì/©GåO*N..$.]vÏ/êí¿p¯^¿óÔ÷ 7o*é5PŽêÕ  ¼) 3à{ÕoSAÏû+ œÅpü… jüUÿe2zûÃùÜOç*LSå²ò.î×{;lO€í9 Ê_OΆòæµã3exÆïS­µHµµ\ýx*w|¹ó‡Z^Òþ+·¿ÒzU»ß†Ú¿WâçðL¦‚)ç/Ÿçø|4«¨ëй°¡ÐU7öOgjâú´Œ®|^<Ó *Ε=ÅÛÙì{¥ŠK3`?Ÿ·æ+åõ+¬sAxÿÎæÊê§x£t.ÓYJ™=¥Qû`›¯xÿòú°þÌT¸>«õþ´Ç'Õj—Ÿ¶¼FŸ_ms-|¥àþ×:–{Ã`™¾†n—{_ÿ¾\ùô=ü]ƒårõâßówU”k?û{xíUgáwû/âæ¯Y]æfœ©>MVŸÚÌ™ZÑ[ÎJ³Lî¢ýIéégú{{Št/Œ®Ò}L÷H˜UûU!ƒ¯W­þ]ýùë3ø oúºnõÚí°sÔkµÕ+µé«´z…6}}vxõÞy ïÂÐW_«W[‡«tõºjõªêpê­äÔôUÓêÓ½Zú‚ ¾FZ½>Z½zE_­^ÿ|Yfà×<«W;_%óÕ2‡ ½Fæke¾Næëe¾Aæe¾Iæ¢Ì7˼Næ[d^/ó™o•ù6™7Ê|»ÌwÈ|§ÌwÉ|·Ì÷È|¯Ìá£Í2ß'óý2? óƒ2?$óÃ2o‘ù™•ù1™·Êü¸ÌOȼMæn™{dý°Oæí2ïy§ÌOʼKæ÷Èü^™ß'óûeþg™? óeÞ-ó¿Èü!™?,óGdþ¨Ì“ù_eÞ#óÇeþ„Ìÿ&ó'eþw™?%ó§eî—ù32Væÿùs2^æ/ÈüE™Ÿ’ùK2Yæ¯Èü´Ì_•ùk2]æðÆÞ7dþ¦Ìß’ùÛ2Gæÿ”ù¿d~Væÿ–ù»2OæïËü™(ódÞ+óeþ‰Ì?•ù9™&óÏeþ…ÌÏËüK™%óÿÈük™#óoeþ2‡kŠßÉü¿2ÿ^æ?ÈüG™_”ùÿd×’ùÏ2ÿEæ¿Êü·Li5øX†Ê²2þ³2þ³2þ³2þ³2þ³2þ³2þ³2þ³2þ³2þ³2þ³!7Âß†ÈøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏNYÆVÆVÆVÆöµ™ý¿jÎÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏ¾±ôû”¬Œÿì¤Ò„Îåá4ÎàáäÎÛá”ÎÖáD•ñŸ•ñŸ•ñŸ•ñŸ•ñŸ•ñŸ2ËøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏÊøÏ.–YÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVÆVƶ(³Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬Œÿ¬ŒÿýóÌGa™‡ÉrÚ¤/ué12RÆâé #åþ‘²ÏBÖò¡Òø•þÝþ™Ò”tˆìï¶Ð©e[ÆÉºwÈñ+û#Ú¾|¨Œ™qÒï:BÎÈqÌI_;ìñ’„+ÖÃ$cÆÈ±8TŽÑhyÜ!¯,ÅéÉ.iç0ÙÇÒ7“ct˜dÁX9^‡J¾tIŒ’±p˜ô‘œdÈ(鿇ÊöŒ”~Ñ&c¸C¶¿CúÊé“c$+“cÙ%Ùu˜Î(ùj§,ÿ9¾mrœGJFŒ•ñ¨´s¨äÛXk×”˜6Rök§äרpÌŸ.÷¡Ò¿s!O¤Ceœç¤"ûéYNNöùXÉ‹1a?HŸ#;yŒŒñ1r|ÇÊX%c|¤dd—Œ±‘'•úºM2p¬Œ¥±²žceŸ·Ë¸ë’œÊI¾Œø]‰»c$óÚ$OG˲Gʱh“q:Fòº]öÇÙ®9îm²ïFKÿ%ë1FŽÕsc$GFKöwHætɘ-™7Z²q¤—¶°åœ0Zöç(ÙÏ£å|Ð!}2'Ù7ZŽËhÉ„vYÞèpÝ ýx”Œ©6«£%·F¾^c#ë>ê7¥ù¢ ø×.y2JúO§‡QÒ—Ûå2JöÛ(Ù]²þ£düŽ’q?J²¤#<_òh”ì£Q²-£d›GJ掔ìí’>Ú&ç‘ÒŸ:e¿æä<0Ròpä–R†sÅHcm’#eL”m)ûn¤äÕHY‡NÙG#C›rl»dLtÉy°Kr±Kò³KúMW8þÒç»v•|«K¶½K¶·KŽk§ìƒvy~—äA›ds—ô“®ðY°œ³º$ûsç–˜6B¶o„œ{FÈö·ÉùgÄKÌ!çÎv÷#dÌŽ}×)Çs„dÚˆ5 r¬rrÎ!™Ý&ù9"‹Þ’_vʹ¦CÆ~NÎ}2†:å¼Ô)ç´9owʱë”ó]NúC§ôõN9÷wÊztÊù¿SòµSÖ·SÚí”mèóJ§¤öSKÇ„ó}’»ré”ýÐ!ç¾NÙ¾Žð{YŽ/AìÉ:wȶu„ë9ŸtÈvµË²:$×:dltÈ~j—ñÒ.瘶W•Êl Ù#DzCŽU‡,³MöW‡<7'ëÑ!Ç·cli Ð!çÁvÙ¦v9‡·Ëñmï‡,9Û.ã¬M~ß&ý«MÆj›<ïP9÷•í-ǾKÆÏh¹VèÜTª×È Kñ>B®ÆÊõÈXYïÃþ—~5N2½KÎY²Ÿ;Ãõ†Œ¯62v³r{¬ì¿QÒÏ»äv›,w¬Œ«±!#ä<1FŽ[NÎcB—ñŸ“10Júåh9N£å˜åd<çäze”äó(Ù'm²Ž#åœÜ¹|XÖw¤ô‹¶r>ê’õëúRiŽ ®1ºÂ5žì¿vé—í!—ä9]’“]2GȘéLè !ÛÕ¾©äsr-6âÝ¥5Q›ô“NY^§ìƒNÙÖNÉ’N¹~è”ó}‡Œ‹¹>ìXTZ¯u¼Æ&ì9tH^¶KN´Ë>h{²Ävy~{x_LÎu9ÉœvYïöÓJcÜ.ÏiÏ‘ìk“¶srÙ.ý­]Ž}NúUÛ38í¿6Êdô{xü;š3†Ïü¼Á¶ù9åŽÍ”9¶\ËÕy sÊÕ'SæøÁÚ5Xٕب>öý`ö:·\y¦öómn——ÃË/W^^¹sJƒíÈ:w°2²Wn¼ÊW®?Lõ1Õ‹ö%/s wÈÒ4Ø>z -g :WR^¥cÄËÈð}’±¨ƒÅZ¹ócJ¦Ì6ÕìΡßó1àZ.òZN%>0P*Ù?”sèýºÁž™ ôÿ~ çaƒk*‹–9PY¼ÜÎì;~n9›ü~\¹s*=¶\?•Û_®ÌêY©Ýrý:X9ƒa%ý>˜Ÿ Ö®rm7ùb¥ýY‰¿%‰—JꙦ-Iü£¿,>+±W.n*‰¥JǬҸÙÿ¹˜)Ý”nåæÎ™ÒßöŠyK~ßõàßîs ¯zûÒþO‹…ë·ô¶§sÞñ•Ç>ÿ­Å”±7üõµ¯}¿X°éž¿LýþSbÎç.é;óýŸ³o›¶Éz‹˜69û›K_ÿc±°ï;§ýbûOÄìï-þÙs ó6o}ÕâX1ýÙÂ5G·˜{ÔßZxíubÁëßö¦³{­˜|~gï^³_~í·÷œ»NÌþ¿qs>d¯3¿Øÿ?7}ò_ż?úÁ«¯þg±`ÌË_}ô‹wˆ¹¯Yvä•Msþò•Ë~µj‹˜~ƃÒâóbúíï|¶´/f·½î’gN8DÌ N?¢8e–˜~MX±ÄüNþÑÿ{ã'ÅÔ9û”1_¾WLý‡¿xò[ß.fÿãšKpæÙbÖö“þøž•ï³G|ô#OÞrš˜9ß~ËKqŒ˜ël~éÝß#Î k1gƒ8c¸î¼Ùk…¬ä-¿›Ù.fœûAY³c„ø†Ÿáä½B<ûëŸ?ü©cÄÌ˯hÿ·×âŒ/ßÓsÏ‘SvÇ#Ï^tŽ˜þóÓFœ÷·Ÿ‹ûÌWžêqÄÌÍ—O•]!μÿö»>2áwbögÆŸwôÖo‰Ù£½°ÇĬiÞëþu¦XðÝãå™ç‹),üvþ?Þ)¬ü¯}ö¬ó„øúQ¿\¶æD1ûK¯û­ÿóßbúÌçn¸òþ61պ꫹îï‰y“~ú²Gïü¶˜¶x³âÌ·½ëŸæ/¶ÄŒ—¾ù““Þ.æ¯9ñu¿¿`ž˜úŠß~»ýÿ3 ÿØ9ë4GöÇO.|cשBœyÓ/ï›·˜ù¾wŒ?éŒSÅ\û„“o .3/™õð¯þô!öwóabÊé“ÏÛ~Ò‹bÖÿsîè_MÓ>{ÖÛþã燉Yû¶çŽœy§˜ùËOŸð›ßRÌØ÷·=ß?âl1ãŒCûÐݧÈvý™'~#ÄüžO¼xþÿI,øÖ³cÿðd17Û¿ù©?ÿ˜ó«iÿ>nÇ 1#ôÂs¿/f/9üÓÿR|™8£o×…·~êïbꇿó¬7þRÌ?Ÿüé—»}·˜þÕòŽi£ÄÙo ûï〈ovŸ(È~ýãüÿ2xDÌȇqRÔŽ…:áSÿˆS÷7{†˜yÊo^²úUíbÖ'Ûn~Íæ¥=ºçƒÁïÅ\èß™¡Ø?ÓÖMúØ)…brÇÿ-]U,œððÒ‹GüTÌ?ã³7yÇ=-f½ñßÞóå%_Óe¥VÌïSÖløáGOz©˜ ã=ç[wʳß{ÔÈ£6‰Ù+¾³å‰‹> ¦íæ›Ä™·w¾úï§_&fÊÑùû#¯3Ãa;½˜3ás§ñ7_“ï?_zôX±âdο~í…CÚïÓ9úßßuþÄì+~Ö>â£3öïÜù×÷‹ÙOïúÕøïÓÇ.”%<%¦þiýì¿!#æí¡·È*‹éG¿L|çð1b~æÄ0‚ÅÔé_v~õô|1ÿ¬Û¶Mÿ¿qbZ8Šm7‹i×ýøûþìN‡ã³}o„˜™ö€˜µíÏKïÿé1cÉ~GsFö½úKÇŠŠ'ßÿÃÖ¹ËzÄŒ±ÇM»ð­Ï‰ÙÇ}å×?8[Ìzä}wüþÞ‰BüöÉ×ß6þ£â¬þOm?ï£Äiwgþüwk~ôàÕbòç~ïkGˆ9½fô… ¯?%ì1ó«ç¿eë5“ÅÂðë?*¦…Ñ=z”˜÷‘yO_È1Sù øëô¿¹ë³ÏþFÌýà}ï™òõ>1í¯•-œ!æ-l˾£WŠÙáÑ—îÓ>š #BÌ«õ‘o‹©ÿ¼vñÇ»^g^ûdØ1cþáy²^Ÿ pŠ˜yÎ×þÃêÛ%fŸ:ÆÉbæÿ{ÛãvO§M_yãë3·‰92è—žüF1û%û;JLýÚ¡ŸŸ0î›bŽôÊ-_9C, k·è?Å4ˆ›i3ÿµ÷ˆ~U×s×þbÎz–ˆ©¯žý_m[#Î?ŸyÒ¡ÇŠ…a/¿åE1ùgoÿÔÕÅ•bþüý‹Yg… þo1gÊ•¿þø'%÷½çñ[—ͳÏ?AGrn?–sbºõÌ-íâM÷ù—}³çˆ¹Þ0ã½7|]LîMþùaoC.¦ÊÁyó½ç qô—>°è=Oˆ©wî\q¦¬Õêo^!΀yaþìcnxå÷'‹™á¿'}Iœ)í­û¬8+¿b׿¿éfqæŽÏ¾ÿ° ³çÿüsܸ[Ìüñ?½~ûÇÄì+Ç„„3€ ³Ž|å—}õ§baÆßüŠ˜bÿ;ļžr”»Oœ1êõþÃK?!&Ï AðG1- ÏìóŸÙ½í«·Ûbî‚G­£_6Y̽qá«Å™Û¿&¦H^õÕ61å‚W½âÁ£ÅÜ·¶33?N̻拌;æ¿Äá·SŽóÝ—?¹û§ˆÉóþô–Ëß+æAÌzyøÅbÚ[¾tNÏ/¾)&OÚ? bà'›ÄàO{î5Ÿú†Xðè;ö]}üÇÅŒp{Ý£b.Ô{꽸ã±KÄ‚ãŠ]$üè”þàð‚˜¾×W‹é?Ù6!€CSÃ(q»˜óÑ·½ð™U‡‹»+óáK׋©Ëf,øâØÅÜë?ûÚ›.̈…GoýæÿýÇÇÅÙá4üêbÊo¾°·÷±ÓÅä°5Ý,N¿bÏÄ_F̃~™zÍq«Ä¼p¶?m‰8«g¿#ˆÙ'=ö¿øÑ©bÚÒý)&_þýàÛ?ÓûQ1b´sîëö²Xø‰—½äOÿr¼œ'ï>ùÃ/ùª˜~Çþ Wœ<ã5ÿóþ-ãŬSWïxð™S¬¿ÉÈÛ(f„ÓµcÊÄkþpùbÖØÿúÐÏw¥˜|ç¼i¿üu§ûÝÐs!~f¼äéï|þš§Ä‚µw?°â;±ï™\wÒåfÎÌú»¿¬3NüÀ†õ¿DL¹ôõ{¿~U·˜ñ0óêý«XðÜ{üþù‰7œõîC·ˆ©«çüèçþKÌ —CŸÚ.æ>r\X²8³4Pbòû'‰9Eœ®8'!íw¼TÆÓeO}PL ñ{ÂgÄ<¹ˆ¡#¦?º ΀ùÿÌsBß+&Ã:jÚÆ›îýÞÝŸs­ý§Ý÷Ÿ¯štûbq&øÑ¼ãßù¾çÞ»P̧)ÿãâŒ|KzÊQBÀ<¢~#¾àô?È•€ä ðû¬þý ‰ˆçÓBª­ú_1¶ÕoÓKÓü!blOÇO†ãÕoÒgÂñÊŽ:Nm«ògÃqü7ósàxU_õý ø^µg6³3 Tý-Ū>  áüY ª¾sØy Ô~(_õÏ<ø~Šªôß<دúgÛž å¨~QýªÊSýÃûY?µOÕÿ V?Þ.eGÀqªß”=õ½ª/WÕo?U/媲ö.„ñSõ™ ß«q™Åú)òCæ¼5>ʾú^õûÖÓØ¸Íbã7™m«ãT}çéþZúí{ôܨܽ“Á®Û+Ù—¤Ì¡äÁîÿTò9m=+=¯’þªä~IÒò»—¤¬¡úF¹û)¦:×Ãwª™«Q×zµ·ÒØmDß vŸ±‘cW­:Ô»+»Jî£×¸­Úß…ßíÿë…KÕ‡‹*úË!úìÉ”ø1C=g(ç %™ÊŠj×+IÓØèÜj´e¸§¡öAµú®\9µ›4ínTÜ 9[yXäLü·òy ß+ÇG~/‹3Ð6·3X9åì´=йµi ¶™Ú8P=ËÕg°:Ö¯•WɾÁÊ©t¼*Ó4~ål—««©Ž¦¹ÔÔ_寰\]+õ‘rešêXîœ$ñTÎßLã\nß`v‹¥Áü…§Jý·\ߘâ´ì[yXäÌàóM•0«RŸ,¶Ë•G·+ë+ËÄJÊ©„£Iû` }•ðªœýrå$M¦9©’ò“øÅ`cZéw•Î=•ôI¥~Uiyê8“Wâãí¯4–Msæ`6M©R–˜Ê0þ±hÃYÛÊÍGû›{OÝ^]+M•¾×δßd×XÎÞÊŽ«4 µœjÙ7•WíqJ]þÞµÚöÊgÿrþ^iªö¸–µ³·Âã ïålÖÄÛWÎ_’öCÒñ,[^•Ê©W.õmt=kÍ¿Z—ÓèTéü\¯~òü¼·¶å×;%åæPçç¤õ¨W:ØÛQ±?§Ü?T»ÕN5´Wú›¥7‚õ†Mõ&Qxæ o´ToUÜû¼9säÿgï= ªÚµFáMïUD¤‚EÅ’l@°Ñ‹¢‚X@±£¢‚½‚½ bAVš]é9‚"V{·³×ÌÖÑsνß}ÿ÷Æ?î##;+É̬IV²22ûtöHÊ•˜Ôc&Û³(õ )õ{w.Ïí õÀ“(Ń;¨³çÍŸå_S€7¨³GOŠW™åA”õàIéÓ„úÒ,O¥Ôƒ¨”£øPϯÒ,üU!ä)ð0KÛeydøãey\ýÎjG˜……÷ ÊQO¬ß¡a \¶ÇR*wŠåÜÃñ'´”ZNàêÎàeÏÕ!îÞY¸šP¾/• À£t‚ÇZ‡UzÅò”ò´¥‡ÒMõôç·wàP½§|¥|„˜+Ê£¥3ž>Q»,ysU;·ËU€ŸÔs°:IJå H¿ïÜ.Ûnù,¹qBšÒEíJêSý¤žt©~QºXzËíðØ{)ÿ(¿©ÇÝQ,¸,Ϲý`ßyDñ¥ò¡ôˆ@}ʳ°Îü¢ý•}N=@SþS»ïp!¦öKé¡|ø zÇöüLõ_ŠÅ*/ª´Ÿ¢z\ß>½¯GêS__ßwæwxXg>ê g®k íR½ä°ðcáõ§~Åoß:Ãðáwü¤øªu¶+Á¸AËÑ~@¾3\Ú¿pû\¶j ÕŸ êA.+Yò£íÀ=1»ùÒ¹¿Ø íß” MûmhOÅ/Ê÷n€…óþ7r vIÇ/I(GéméÌÇŸË¡ÕÎv!èh9jO´¢üô÷PN¶³ý³ß›žµ©Ühší×_µ³œå=ë¤çºhšm¯^€'Ø·`ÞBéíÅê(]",9R=¥||:ù“ž±é§žÇ£‚øÕƒÎtàR{}ÆŠÙó-:>²û/Ö¸ÊîúLéeÑEõ@ ?0¤zCõ­Ï‚þîgä«?uà9µ¯ÞT.,»gõó»¤r¥öÓ¿sù?y´gé¡ ÿfÏ)Þt<7aá!ÆêX|æw¶Áý4teõ»‚˜ê.Ë~(>´Ÿ¢zEË÷eÑAñaÃ5fñÊ—òÊŸêÍçt¶#A¿WÛ™.®vç~Š«È²S/¸ÊíE0?¤zJç'ªŸ”ol»keÑEñcÙç§ïUâû—h ñ«üžÃ÷áK|óýãòfçÂDâw9|_ÁıP÷0¦I⫘ø%_Ì/9|´Äß0ßO0ñKêñF QâÓšøn~ÌaüÖò¨çûù%8‹8Œ[â¯û ‡ñ³KüÔª“vhà0>qáÿã×WÒB¾bá™Öæ…8^xÇ´E|%Ÿ¾ÂÄwC¯H<‡ñ›KÊ'røþ‰iá­C˜Ã÷Lüc óÊ ^m¥pø¾…‰¿e¡L^ >Ÿ¿rÿÉ<šÄyõø>€ {xu‰\âc<Ã÷ý˹×.į3ñÍJøH|áò}8_ã0þŽ €ßÄÏ4ñßMüû>ãÁ$~˜ežóý8“²÷yt¥rø¾|…C¡Mâ“×`|¾ÙËþl_ü-ðˆÐBüÙZ-$Ÿ¿§9Œ_d^/Å÷kLäeÃ+G|3›þä¹á‡¤yô““xÄ6ñé]Ía|?ÐÙ¾Of’'ôdÔ8ÞäpĈÞ,¹Âë!øá~ð5¨œ &~ЉíÏPöÈŒü¯ãáFüÊ_å¥áæ>â'’øŽæëñ?M|µ½’‰tâ{›øà&~»‰sâ?›øQ/ýÙ&_ß= ÍçŒüˆOdÑïÌ3a/È+>ÿÎ!ÀkS°ŸF.|ÿáŒ~ ‹3òfä!FüÚÿà=»ÂØ%ñ‹ÍoÈÓ‡xúÁ÷1M|s¯ãÉ”gÏÄ·5_®Ä>ñ5N|„·Âÿ,ãËšÀ"¾™Uñ!MäOêûrÿÕÄw>±]7Ð7‚»9ðkÐIðÄÈ•øÆ&¾‡ùúGüÉŸÝÄ´.Ð0lo(Ø)ñýí̃AFÁlƲðzÐ9âcœøØž<'¾©íòÛ#~âß2ts@—„HÄÓEa.©óßðÿ~àÐu:Xâ*îð4ü½ƒ+S.=:èX9WÔñe„VºW–¤¤Gsà¹"¿˜§àýU’äÅsÅHqÇW\EHËò‹õáÊ|aš&‘áG®ÜØä®ËëÆq¥Xó9Ùåu¼œ\Ù®ü?\y€'Ço^š+Ašyñó>[¨'ÇGsWêK¾ò€?¥GêK§óp¥ ï§‚yº ÿqO®$àAù@×Oh}išðå ž,à¥x‰Cy9 —¶'|—®ã3D°žHóé¼—¾)¾òô½à+Ry~ìòâT^ÀZŽÝžÈ…ÊY&™O Wð”¥1mê+€üE)Ÿ@n2Oä¯HåO¢9~Ðö(^òÀWªR€×*B;‚úÀ?€+ú"p%!_ äCé¥ò“ü¤I(¾ÊY‘æS<¨üi}*Wà—"Õ ø¦DëÓvŠ7S{ öHù¬r¦zOñ¡vDåNõDð¤ö(z$ÏnøDñ¥ò§|ð‡eTž”¯TŽ´¼€ïÀÀGðäS»¢ü úEóž(´+À‹¶ xP¸TŸe©Ðþ‚ö+еo¶ž€<(Ÿ¨ÞQ9+C9A}ÊÀKžÕÏ)]´]Êw%ÀG–ö´Ÿ¢|¤pXý4•#å7å›@>ÔÞ©¾@Lõê í¨ÞÓõU Ÿ¶OûK?€ïT¯i9Eè¤ZÒOåCížê· «]ÊGvOõGä+Åê_d ¨GÇALåLõ“ö€?ôc¸ö'@§$KïèøGõV‚—Ž_´ÿ£z(iqª,ü©‰ÓñøDù(ÍÒCþ ò´_¥ü£ã5Ät\¤xÈÒñ‘Žƒt\û§ý‰ }x. øÑñ†öƒ}€´ÕS—$Ô—gͨÆKª_€/]¡ö®À¢‡Úåmê -Oé¢ã)µ3 ÚÏRü€>ªwT/Ùú+DíêÑ~Tè§öHåEûÁü‡ÅG*O)ÐWA¿ú!IÇJÐ)Bûhò]žâOõŸö”´ ü½ð‹5/¡ý Å_ø¡@Ç à‹"å_ Ÿö{t”¦t|:¢|Ä´8ToéüŠÎ)¾>MS}¦vBû{ª/2@77é¼UÚ¥ü—eɋʑŽWŽ8ðY0^Pû…ç´ß¤ôÓy •å7¥Ÿö#btžNõä+Jñ`Í©>ˆÐñ•e·ÔîÙótÚ¯H°ìWÐoS½£ý/¥›ÎXóoÊ:>æ¥Ào1j—T/éü‰ÎóYýå;Q<©ý Æe*/†_œŸçŒÈ=jäŽ;²NÅå0w±‘µ¹!ÈÝtd=‚¬å‘»æÈzyï'w§‘;øÈ} Äs–”#wl0k[ü;âHyr¹«ŒÜoFvæÉÝkĹŸŽàBîé#w£‘5 rßÛLh›ÜWçÅaîœÓã0÷Ê‘»pø÷ë‘ûÀø÷ñ ‡ùëäž"rÇYO$÷î ‚¶4¡Îthc´ãÏÉ]r]nr ³–Á¿£Ü“Gîr›ü#÷¶µr_ Yî0&BÐ^¾Í.dÜ+GvGȹÜwGî¿ë ü1„fm³n©¸“ÐÊȉ/I?~‘²•ÐY£›0‰\@䣾@‹6ð…¬ ]Ãж-È•ÈÉÊc€Ó—YäÓHòÉu}@æDžM l²Æç¯ àÓè$åȈä r?%YÛ"kØäÞDY¡[x¯ÿ¥€Æm@“9ðßx5`­€g€ûÀï.àKî`À¬oñïˆ ågƒü:ø7ÍY;‰Î¸\]À›Ø!Y …rºPVxNøDôe àÜÒË€dš¬Fþð•è4Ù½!z¾da 2X <$ëtäžB²6íe lâqÜáHîë!w.ö~¹ßä0vKÚ ¶áËat˜Ü{ùde |S…¼žäÚè&m_!¼úoø?pè:ݾ ~‚{í 3.†i3é´&&ÎlTq·1µ¸›2™ø³ñÆ=r—Io=ÇÄQ\&^»”‰£C˜x=ÔËÚÃß¹âFÌbÒ»ß3ñvf‚À=è ñw&ÞìÆÄ±Ð~Ú9¯ÔÉL:ü' gâ3@Ç¡j&zè d✙Ì<‘¦÷Ô2ñÀïN 3YĤ¯vaèÜ9ð¿ÄäßÚiÃo'ðÌÒgâKC™ò‘]îF&¾ÉÄû ˜ø¦'‚ÙHåFç^1ñÅj&ÿòpÞ¥ùLœSÅg÷àu[³;Ÿž4a¦ÞõÛ ý[¼@ ŸÝ?Àkàµ>ø]ÂÄû¡ýpEàk:¤ï#ÌÆ1wg/>mþÇ‚ärÊá0ñÞ&Žñ8 ÷=TßXQ&> úò9xEjC ø„òǽ _‡A¢™ VîÐË€ß)Ðïƒ.À'Àsã¼’»ô1ä´³è8{ïHЃp.¤÷0ñ&Š'ðé(зôa;”‹„z{€Î=Àï£[™x Ô§v¸è‰ò‚4à±âÐ×hC&Þi°§Ð>Õƒ8æÃî:ªÐÞZO¬ozöBýDH|vBûq¿è€?înàk2Àß ù1 × oI —}àÄë_ÚH€¿ÒÇ€þƒ@Ï.Zò—Õ2ñvÀÈsോ¸‡â€_€ßfÈ=t^o„ò‘šOí ê­øû>ªá@OÄ{¡ß\íD¨]…~; ø½ ä|ì-Ú9õ©¼“À>"2™øÀÝx¾l‚zÑÀ¿}/ú‘Øk@Ô‹„òk©=p€/NIúŸ3ñ.àS4¤£ÿзHàCðí äGƒ¾lÎìÌÇ­Po—‰§À^ÊâÔ‹„ô–8&Þ ü ‡öv?z¿@Î;@¿âC@O"ððkÀ¢ãà¹øõ÷ŒaâÃÐßì:©ýìe>læžÉ†|è¯ößeâmàà½øüZ ù›àùÚ@¿Ec g3È!øKûKèW·Aûë€_ë€O›>jGQÀ‡m ·TŸ÷]‡™~¸Ç@_˜îêg?— t¾Å3Fq×?÷Þ1/ô*èØ rÞx¬ƒüíÐÎzhç4_€?´_8uè¢üyE=ÁPo/Œ£tÜ¡öýV4ô÷´?HáB{™PèÙøEƒï» å`\‹[ÏWóÅ!qô¿ñãÿÆÿó˜Cß‹¼3ÿöôfÖOݘJ®³^è9žy˜Ì%9\;3&= Êûˆ2óp°Wçf¾î²‰)7‰ù@˜ë˜ÉäOͬ{: ežO]È”_°‘CÛ™v€I;X3å¦A»ÐîÜkÌú£Ç5¦œïUŽûrfÖÿ“vòfÖq§=³a]ÔÖ1ç:2pý ÿvbÀ¹ŽÅLý™;™˜ò‰æ{ôdbŸï <ÀËá³¾:µ è8ÞjÌ:²ÿ¦=ÿÍÌz«Ûl&Mép+eêS<)\J‡s³ž;ù† ˜ç´Üô} \Ú.•—ÕE&=øêóð:ÂäϬfž‡€Ühý9o:&jƒ\•Ï8ÀsÒ²Îpfx1å)_ÙñLÐ ŠÕ¿€ß.Õ'*/—ÞLþ”uLy*‡yyÌsW 7Æ«ï™|*7ÀÛð¤r˜t΄y/´kÏaÒ´¾WWà#´ëþ±³¼)_(~®ßÞ¬ýÌs GPî‡êååÅÓ ìb,Ø µ“™À/›3í’ÚåKð¦<-Gó½e™ôôÛé¤i'ØšJù üó{ž vçxMŸt>êÏþŒyt^šï¾³s=Wà‡'”cëåŒþ ³a>OåKñ¥|s¹Ñz”>‡ÐãDhÒ¡0v|©¼AÞ´Ÿ£ü™TÝÙ.h¿HõÀæÍÔ.i=Jo`R~yÁ84#‡%7€C퇎C~ @Ô§íS~z¸w¶‡9†åKé¥vDõòŸê1í¨Þ/;¤ü¤ý5Å›ÒKñ› ý‡ ÐGÇmZ~Jó|ĮПѻiÜ Þt Û ÊM‚ô$È÷‡x2Ô›±#”‡/‹¹.ûCLÇÿ)ÛB=WH{q:ÃõÑ?¤mž7Ô[¤Ý>ÅÓ•Û¹ü4¼©ï Ï©¼gÇ1ñt€ãË Æ¥Ýù¹7¤ íõ}(=µðé/Þ¶†8žÃî׋ۙÊW?¯Îx{3õþ·ç‘ÿÿÿÿ)æüü~,xºñÂ0^ ‹ødBHö?Kyá:ä‘=K23${þkÃ8üŸ‡Ù¿%I²8Bö7ð‚3ÀÊà…g¼`Á '8ÌY“³×J¾3ðƒx/p9üó0|˜ä¬éç‡q:ío‘ýXøœÃìç^€ôxÀq*¤Wr˜3„Ži3ð% dЊ€ÿdA'‘.s˜o4Ò1È 6œ9"Lð&ûñd¯L.¬™ÿä †ÉBéýÈB^2ð‘ð‡ÃìÕ‡:d¡Éª]S€ÇåÀ+ƒìß’oàì˜<' bd³‹ìi{@»äÿ8CÎÃLå0߇oRÈþ>ôþüýâM€ÃtxF¦¼&rl‡ùžƒì[“=à=¼PÅ d*è :á|›p‚á; H¬‚¶&/Éwy@+ÙY )@ÿ¨·dBv æ¾C Àƒó&œSGИÃÈŒâmù#€CdÏä“s4üvMPn&üÏ’=y¢{äû‹›âJ¾§!:EÏÞݯ%páYÄd„_t;Ùí“ÿ„÷DW'oú“!=~ò]Êà#´µ…9Ä—ù~|ÓPÆa¾ i"¯±ÐöR€IÚŸí@™aÀÛ0À•|gDôm-“Ï?Fê\ÜNnÔ xÏ,à)áÿS^HÚ‰ý;@ÛÄÞßA zFt{/‡Ñåàÿth—èójhÇøEp&zEz¹æ»ÒùF‰ØÑQþ.óÃÿûós<ºÁa¾u#ú‘Æ °Ç?ÓJôîsœåë6ÙP&}ùfå6‡o{Bä[¢B8/Gô|kGìš9{'DÆòœœ $ýr˜ó…¤?âŸÕ#ç㈎‘þà¯ÿ ßC=aÎÛqpøß² UÁ™@_h£‰!‰c<#÷6ñÆSþUO =üsz ¼PÌáŸåiäœ)é7yc)9ÿËo£ž÷_ŠˆÝÖÁU^…:dgåU‡3Œkà›'ÒG\bêX(ƒ9£È?3ìíDÁ÷LäLks>–Ø.™3‡cAD¼¾Xh*sîô¥üóäL0ù¶k/.æð¿câŸÓäá&Dø•Êœ$ßþ ™ßˆ “M 7þ˜'t‹Ð ¼!z° ðçÑÈ?Læ"yƒlÎo‡±×‘79'IÎ’sŒäÜ!¿ÒŸned"Ô9Ë(D¾; Ù’3¸¼>SÈŒÁ‘¯;ä¶85i«„Åy¸ñôŒþ–Œ‘äœìk9 IÆ^>VŒŒøã)'a !rŽ4d@äÏð‚ð˜ž×Ãù:K¾##ú~—æÍgøg.m!gOù<#ßÚ‘~»™Ãœ?ÀÌUÈÙn~\zI6­+8L_ÎË:ÉaΕ7½d³¿ˆ©ËÿÎðd3èGH?|tOìŠèŠ+è–3×"ö¥Îð€Ïs²öNæ¼¹•ù®u!ØÙ^ü…Ü5æÍ=„$ÜùcA=èéx³Urî•?®}y“³Ûj þ\g7ÐAÆ<2~Üç0çxM¼ùt3¾ÝÜùc<9/|h%ºDx/ ¼‘}"gcÉ7~¼99§Ë×_}Fæüy`)è£Ȓ̬É9×.@‘¥6àþt“ÈÀ#ãÖÐ'%€MÆØ' GÂs2ö¾`Êp>’Ê/dô‹?O fž 2ôñçÄ\ÉÒ§½8¤é/â ÁÈ?ßöBtƒÌŸÉü‰Ìïx}ñ tŒÃï…‰_‚sLy>µŒžñõ‘À&gÉÃ]áó„è2蔇±Ï:¦ßÎ=™2ü¹[+ƒ#LúÑEtÌ_’=#}·P4Ø™[¾!‹‘!ÿ>‚#Ø%ùn–lp“öÊ@þ|¾p˜¾Žô…) DþdÎÈìQ?búƒ™õ%fÙ‹ÃÕƒõH}XÇH\](OÓ46€˜Ö?¸ìòú>­ÇjLj–‡çÇÚ×¥åM²ê³ÛÐÅ¢WŸUžßû7ð XüÐgáIùׇâ íÿ‰¬vÇ7vyÊ7v=J'僴oÀ‚«Ç‚«÷7ð ~S­´C6]TnÊ?ÈïÃΧøP<à¹@.,½1 zÀ‚kü íéÑzëÑç,ø´œå+fëõÏÇ~ÎÆŸâ#À¶ÏÒg\¨þ°äCåJá ôàäDñ…4õDí™ò™Â£zJé¦ðý;Ÿ%w¶ý‰–ütXüÑeñ–”£ð~_*=‚v(YøêR=¥i€kDåCñ ð ­Ã‚§Có)½´Yå1KÿôÒ˜Í|JGݽYôPøýÙò¥ügÙ· OÊ'6žl=c·GñïÍâ Å›­ß¬|úœòÙE/¶Q¸+Íî?MXz#à [/!ŸÎû XôÑôŸú?€oÌ*OÛ1¢ü¡t²ôV0.C¾‹ÿ‚þšÒGùDéfÛk\aõÃtáçs–½S¸ìñ‡¥Ï8¿§Yz©Ïâ»_Ôeá-'Ä´?§íýÉþYú/Ðw‚ù'«Ý?Ù=íh;,~ äü»`—ô¿”,{û“=±Ÿ³ìš=ÎB=ÎÏu:ò¾*1ýß1wÈëXVäõDYÞ™.ÂzÖ1éP†ÝVDzìöé=€+Ô¡¼§3þBœßÓ,‚ɦÿwÏØ°:ÒÕ‘þŽ8üªÍŽ4 q~O;ûÙ¯Ò¿â'[Ž"ò:†Žù¿¨+Â*×W¡ßÀý¯ÿŠG˳a²åÀnƒM#[ÏØòfÕa?ú‹ò¿jSˆókZٺϖSÇö§Ÿ¿³É_Ùǯô…ÍÏ_éÛvØyìÿÛý•>ü•Mý-lûSãð MŸ‹t|ÆÂ§£ž ÿ Æoêþ®Oû•=±íœÍ§_ñõwö/Ä‚ÿWú"Ú¡ýßÁcãÔ±êHÏßéß_ÙE‡ÿ¢Þ¾+ôfd¥¥#Ƭ@öàû±BVÀ äCS²CÞXÈ<Ž|GBöH†BÆ d]œú¡ad‡@“ý²ŸBÖ乬`Á‚e‡0†Ãìùw Ö¬0–È÷4dŸtB‡@öQlX|·`ì¹½4²êÄ äÛd‹èÖ…?ù‰ÿoüßø¿ñ¿s9ÿÿÿû1çÏó “ùEÇùåï‚Èßä±çº»<ÇüÞ_áòw°5g¯tœ[ÑðWmüU[lüŽ'¿{öOhcóŠæ³ièHǯèú§Â—äüµ¬ÄYáWïÿ¤Þït-³ŽºEò©>ÿ•nþJ6ÿDÿ´lG9ý꽌æSZÿJï~gK¿’Ï?‘!›ßÿTþi”€ðOlñïÞ7E8ß_ÙÌ_áúOõüŸÊûŸðôïÊýjÍ…:ös¤ Õç_ýÄYi1ˆÉ{žb‡çä}Pþ‹tÈ#ÿÉû¢ ¤% lÇ|Z8q€òëþ œ4;”!ïª`|%;àH`wûŒþK±àÜÈû§Z‡2"‹wH‹C] ƒ¼ÃíHåuu!VêP—àÕ¿Ca®„Ïò€Í—šþêGñ쨣ôGÛîîÂùõOœóó]ž¦;Ö%ïÖºÊÓwt HSäüäa7¨Ó‘ïOi¿+§3ï…9þ‰ÌžÁÎL§ù)GuNœóÓ‡$‘ÄT—•¡]ªËä?ÕeúŸê²§³.wãü”+]‡‚Xbòœ®EÐ@ÒZœÎº¬Åé¬Ët]…ê]¿Ðá0ë=8?×':êr7Ng]&åúlc€oÌé¬Ë}9uYøB׈ÌÁ'(í¡£.“ç#9uY‹ÓY—58uYŽó³/ãü\[’ØòœÎº,øÐõ"a€Ku™®u\ê¨Ëê¦xôàüÔeÂÃޜκLøÜQ—58u¹ãZÕeº¾ÓQ—Éó~ðŒ®ïеœŽºLÒC¡ÀOþºço~wÿ6;Ÿýž÷»òWïïâ·Þïêÿ¼¿ûý÷ßáñïâ÷ïòëïÚùWñøþÿj»ÿ)¾ÿ§ðþ»üÿ[rø§¿ÿÛíþoñÿßŇ]îï~ÿ½ø—Ö›¹Ó.;Ÿ¦ÿtñßäÿÓç‚{KÿÍüŠÿ?¥ÿŸòáŠ7;ÿ_•ת=6üߥÿ§tÿ§Ûýß–ËJoþ]~ü«øüSùü«íþSøÿ)þÿOñùwûë{ºoýó{á`úy€ø¼E>‚í{Ñ…³—wÜ„|[ º0d~Íð[ì7OáÇðëúv»ºSßi—^zá¢àÅ~KƒúÎã½òðÆ­®Xö¨Ök_×èá3nÐî@”³á¼Ø«ZO”ykÏĸÃãÑ• ý£t÷&±S{Qêçßo¯A©r§íÇ]HA)ÃT[b®Ù¡ó‹,Þ47>@ÉN¦*Ïö¿@I3[7ùÖ ž_¶–Û³ì3Êî}äÆˆIèÁc¡¼Så®nAE†K§5†ˆ£’kçªË>B¥—½8OwGå®ÓCº˜£Ê ‘#ž3F•ß&u2Fïúm[“²•Íéînb‡Ší^]ºŸ¾lÓ}·Êp!Ædû~]4/~fY»ž*˜W!ʬ@…£^ol½°è^¿*w Y½P<¸ö.*ú[pK*B€O™ÒµI—1ªØû`Û–SyèÑîÇÑÓî Ço¢ÎhìÜ…ê¸èÊÌ Ú¨~̤»Cí§ §ÑÞï=ožAÏÍÌ }¸‰^j„x¤ÇÞDoÌ ÃŠš¬Pó•y{¬þ°C樂ÐM‹Þë"ã‘ÅèÃsÙ—j}9Ux,Q}ÝxÉ´uK,úq}w¸ˆ0çHžrt>ÒØZVEÏ\qÙº[‹ŸR¹}Ø K®]ýíÔ],ídSl÷ËõvÝ Ë£TÖèXÑ­Í8ªgV*í:Vç;î#^›2x0V‰ŒßlvC«]ȈVÉžƒ»WMˆ+-…5Þ(˦4á†k8³>àž—7§‰œ|€µŸ«mµñÑÁ:ʾOǺïb›/qÍqo9³”c×°¡¿¾SŒ6z˜稣†ûXØ}rØ*<àØüN~ð ù)Qeoa“e_³ž&>ÀCz.LŠm¬Æ¦5:VÑgð°#'æ~_é‡Ífu½|Ú <¤µ î#[ˆx4æÆU‡5–c ÎúuœýzØÒ÷Þh:©4xêzÈ[‡]¯©<€Çá#¢ƒ·~ÇÔ æ(÷Å«³Lo…öÛ¸íÊÅjßÏIÁª£ÄãØ!mÛú“zbGû˜IÇ\±ãcV]½±ÓŒòW‡Ã–c§ZyÍ•’nØÙyëý׊òØù†ÿóÕÛ˱K/Õ“±×±K¨º‘Ó–ÃØ%?÷ÇÉI~ØU3&ß?Þ»Î@.¥X4§ÛEIk”cSñðáu”iÿòPÿéïQZƒÅê»RÐùO×=sÆ»9;zɨ¸ýЙvëèÛ¢„Ë­S¢“õ‹Ÿ*ܵFÇ{e¦\=0Åè?e‘0:!û½Ôn[”ŒUêÂp¨½R{¾>Ëà‡æËRtÃyeÏ£>èúEþ÷3î¢k¡% £÷­C×$Œ¶Ê8Ý@W5*{®¾÷3ŽZ3I(0èg9ˆðîêÅŸ A™›³ê8uEYçZÅ.}+ ÷fÒ«™¶¡[{ %¾|C·3v~ÍÑËBwN·« •B÷¬ÛŒ4<ôhø^ÖúåÖÚÆ…lØk‘Œxf„ª0*¾™úá^ÚËŸýD“ÔñŸQ•ó5æ.ѨúËšâ¹U¨¦ÊwÑÝ%ïP]À²ÐX…#èÉË*×}Þ¡†¥.œÐ³!gO=¼˜žk¤Ø}͆^hϱ¶Í”C¯æÉ•+˜¢·’Ò+ÝýCQó›þ*>q.èÝÊ0ÏÕ ?íµBjK_SqôéôU»„»èËe×Â{ÑwéõcõO¢¶M²m{æoǜ践>aañGMW5wa‘=²Þ.3Æbsc>OªÃI«gÉœÅÒAÈxß³d,w¤§EV÷GXq½æxñéÊXÙuÈóâ÷qWϱ ׬°jòM³¥I–¸»³ú»ÅÃì°æ0•›Ò† ¸§ó}´ÇÓ÷òŸ8i˜Ë¬;UàÕb¬9æ˜Õ¹pl¨¸neÈilTõbOêiܯ@ñòîˆ×xà²Xéy‘Ÿ±Iß“½%¤Îà!îÛÒNßrÀC_&È™‰Í¢¿4˸€G©ŽO œ3£uÉ»VnöÄÜý!ñ·`‹aõ-M:»°åçûÞ»°Õ=ÃÇgL2ñØ£¬G †Ç;¨ºOß¹ O(Ó=?ô(¶1Ý#riˆ?¶]â—<¸‡4¶K˜¹Úîî}lïe{ór7ì"}DK;¼ìç¶õ[6vôv™x¼& ;æëÉõ·<‰†ÔŽhNOÁNÑ%ã÷ ÆNMk·&}ÁÎfæV–(açµKô»…ÍÁ2‹Þ ç@œ^äéß„rÔžîL}1eú/g9 ]‘}óûÊ$”–l‘z7¥Ž ÓºÞ?]r 6h..¯-´Óé†ÎÝ,:óÄD%I>~¸õ¢+:cª›§>Zk?*ÜKU ¥çäõˆz݈²ªôò’æMGw{ïsº‚òzixt ÞXy¡pÐ{TœuÿaÃà{¨tDfÛ·‹QYáp«‡³¯¢Š7mQßQ¥±³’Æ&TxäüÆçr¨L>7æ€w*ÖÉju)ùçøæ›œ}4^=\è˜mòí Ê5=á>&åž›çÒ- mú¶×ö3*xŒ–o߉ —ßœfÖ¿¶®˜^õÝ–Rí¾Ò?WÔY€PiÒþæðÕÚ¨B…›šx U5j¿?¹/z¬´s/ª<Žj3¥6OŸ<=9»È©åø,ÔÐ蚯6 5.Œ4Hžp½nRcy½i–üÕ{9j’™Þð£–*Óú¯&¢Ö#Ýírò÷¡ ZÝÙx}Џ3YFH}iëÕó°ÉôC¨¶ß&N5æ˜i=wëÖª=Óåp U±?¨[‡Å9Ëž‡Œ˜†%J»¤9Va©Mc\°Løæ©²Ã±ÜæölIK¬(µdÞ¯‰•ÂîŠ |Š»`¥Y“GÆ*CNO%7«Ù¶[5JÀÝÓ÷æ¬Æ ê+›âÞÃÎsCÏážcLn˜í•ÅÚÉ"Ïmì{}ÍR‡uÏôV~õÓº¼Ý‹ Ú>íý¥6rµí&%û¦½Ö¼¼)ÐØuuåx<ð^Tã7wlbr>êb.Ê~šwÔ›n=±h÷þ<æ`CÛb/t,Áçj¯Õèøêʹk©èt ·ù@Ùiý] ¹ÖïèZUt¹hn­¹ˆo\ª\ñábº¾°E³^ö ºÞ+ÚàævEtõÛ,¹Óɪèj‚t€tÖt¥”»nvš’ ¾ªQyצù.ºZ¡y¥nDW3cÍ7ÌG×6칑»]¿í°Ø~¨Êx;5ô…a Ê’~Éßh:ʎ܉’žj¡œš­®ý ›…–¶#š£[ŸÖ>«5JD·ßG»l~%Í›÷,eî/:Ïoc”Û§L:èòtT wiOÐ?TxìGùÒÓz¨xòíz{õͨÔzCÖÅÜRT>ÇwN˜šª¼=)Òñeªöð׿2gªémìÑÆAu²g–í GOÖ÷*œºx jP=ûÓ›ôôúÐTc;[ԸǷ<Ém.jŠ1U øh…^V·[ð/Goæ7Й堿ÃbEdÞ¡w}>6…Õ[ Ö{7VF; KT+ÌÔBŸ¸å±7"<ÑË‘óà Ñ· cÛÏÌ›‡~|—W[*Vˆ9êÖ^õÄB×ã¸Æ^ÙXÄþøn¡mbX¬Çítç+X©dÖü)*Xê»X—Ó'3±œÌô÷6ëŸ`Å>'û¼ìÕ€•ž›ÊûoÆ]kÛL¾ŽU{š*« ŸÆÝÎî|#ÔZ‹5B#Šõ~‚µÖhS–^е[ŠŽ ¶ŠÀ:Õf~—×íÀú¢]‹¶V©aƒÅ¶‡V=êŒÍ4-0Äýº”˹U®Â»´oºÑ%ʨ:XØÛ®Ÿ—ß]õÏuðS ÀfB?ÄLúZã‘+Ý?ÆŽÆ£›´&ªµïÀ\»Ï-ÓÏÞÃæ/u8BçÚ°åéXQ·&Øj©ö¸½7åðXî Ö¢ÑxܵR®uwñ„ˆvcÿ]«ñÄ©Ó]sæÙáÂKkªVb»€5E -°}̰‡aÃ2±ƒ«YÉÞØ¡|ÓV¯EêØÑq¸OÔ$예.ga'ãÍú]³¿c§­FÏ>_…žÍz={bo|“ÑŠü„ÿߘ(wK¿×mh„9ºï}À³ðöLžÝ©÷ {/2¥\|kÔG—C&¦ï|°÷ÞöÍÌÿyº´G:("oJ‰18ÊÖ]¸qú¼Ä†zt®«ÏœâPs”˜±qÉý,”(⢜+ëŠÎmRq3¥[øÎ-CYÛN&⣳ÐÔ»±Í½ƒQîZƒ.ôo¢B—uâWÞÖ£âñ]ú'{¤¢’k½~(åwEe6çûêeš¢ -mÑ÷òPÅÝv‘1EÞ|M¥4Rc*½¸&¿Ý.š÷¾µ»oò…ÊŸãÛÞÓ-Ê—w ‡GçX~ð G¹#—¦¯þúåHñI3Cù?>=mPí ~|/4'êû¶¥Ù7 Â#²g»-zЦ†N l释cV=8ŠJǯØ"dÅAå‘—jD¦¡ª¡Sô„í£ÑCb×ûôLAµ¸ÄÃÍwz¢¾ó3ß¡†Qã ÷e£gIÏÖyÞ„š\Zfä¥Ö¡W£'üÑ«2½¦QRÕ µ¸ø*ÏNóÆ·-~ñѼqn&Î×­Å"C¯sjÝŒ²E¦ÎøzSÝ%ZÕº  ¥ÎÍÛ8âŽ5:ßÿlä² èÜe¹[‹5ï¡dÝ> Gü$P¢·ÿ§ÔèT^ªdBµ7:qo€zhBŠ—X¡bî°ÓäŽmÜTŠâߥ†>|…Ðén Guì/åQžÌwÇe(½íä[Ͳ¼÷°Í_óV¼F×büÔuc|Ð5?ÇÆžîèê²ï²jY èê„6ÒÓЕí§ö^|ãþ3>a2bOˆ6º:ØÏmCWE'-,F¼ñPÙ¼5V]ËYzçËÍÞ{ܰ¡û“8¼yòÊÃ壬Ûs;d_¡e…0"tSjȽ€WЭ Š~ ŒE·‘~iÛQtçiíù'_£{w7ԼלǛŸz¾â¶ Bùî½Ý1¯(õþM²FE/Ê JVöA%Åg×®IBe/3F}Ù°Uo¾u1çztÕf©äWwôxiFíëpT{¦f¡¢Ú,ôD{¶ÍÛìù¨>ñ¸ÿþè]è©ÕõœMÏ+P£ŠŠ×ô'‡Q“D{õÖ¾¡èå„’°.™ËÑëòœmF[#P³ã¦®Ç–D¢–²ÀÅkÝ¡Ö ”ÿ|5ú ªgº{(úXÖZ]l^‹>çÎÝ4 Â}•J—H3E?æ+ œ| +æ—WFMÎÂB!¾»öÇ¿Æ"â®]âj°hÁñG&1XüsŸY…O—c© ¾]™U„eÅ/o}öùª¾€•V˜ºjIªã.^vµšb•Ç_\rSÆÝde›¶ ÆêÙ2‹9z5¸Gò¼[Òw`íP÷W†Zb)¶ûÒ¿ÄzKž •ˆ$âÞÕj›d¾â>«kí’Oྒྷz«÷SÀÓ¶ÛÖ~!$s…W"ÙŒ#Øü—¼:öÇ–ž®—5+ò±•Ú… ®ØúµAÜåò×xÜé×—m*ã ¦U§>zòæ™%W’Çp±ÍÛ5× v7c»Ò6\ÅöÓÖíUø`ˆF·¯y;ä<Íüfv4×HëiÙ;^*ɵð;é Qžk…Öïø°éÞìTïpߪi=v^ÒaûvÞ°vòÊ}õXªÇ¡~‡º {A¶/…§>AÙÃíÎNøQ2&ùŽqVÕGéÜÏw÷â½·…ì.9¥4ÆdÛÍë‹.V ë_Úç….‡Df¾NBÉw7mýŒN/61«xÒ%†O^ýh;oÜó2\Ò2¿ÞX¬q e™m(,[€îX,³Û½=üÃÅ"¡ª¼ØèŸT„Š ¦D.]JF¯ýfðâ*½k¤¼Rõ,*ßÐØd<)U (JÙþf7*÷Ýôe·½*u*ã32ݘ¥}Nôçø–w²çÚ|iô°pí­´[Š(×|ÊýçPžú¤6¡‹(»FIld *°[‘°Ñºˆ6 úŒ í¶¹¾]Š߯eû ¡bKã¨Æb¨äéýUÒu¨\ÖXòex"ªŒØ{tÄØO¨zòî…~7QÍjär´£ºˆ ³f9®EõQ£Ã·Šå¡§ÏGÕˆ¿Ÿ‹ž/8ûÜýNz‰=eÆ;Ö£7N÷Þ[±@ÍyÇ“ÕÒ„Ñ»ù±ÅwÕÑûeNcÃæ^A7‹zWžyˆ>?ÿüîcXúv^gÇ—j¨-?¬r¶~?,Ô#}ˆó ,üly'ÿR,zÞÂiXÐ),¾|ÜêßûXÒAG£½g,mf‘wá4o|sŸygZëJ,_RRo¦Û+Ö›˜®ŒÅʱ½—´YŠ»†¦Šs·aÕðUI›ösp·è^íxè\¬¾é ©ëÇXsá$¥ÌxC¬å¾öáñÛ°¶Yõ~‹’¸WؾXýØ`¬Û£"2÷ÉU¬/}×7¸~,6è߸??»÷™:øxÙ3l|¡Nwåþ,Ü_æ©}IÐp<·ÅŒé/ŒÅ÷d/:6óžÄU›Œ‡4&—¾LVÆC)x¡V+<|úÁÒä u»á^s•x•[:˜­Geê ÁŸ®¡ŠÄÍßT¨£GšƒîOÞÜýqçõ®0i-Tk«<ÏvÿDTwåÜݵòI¨~´œšŽòÆ<_7=‹ºÕ¤PŒž¯¸2Ñ'½HO¸lÒ]½Æq*]zJ¡f‘¡1–õýPËÚ‘uy Q«j†z4½¿4hÑ·»vèc ¢nc›/úíßÑ÷'Šã‡™LFí‹öh¯{n‹…Œ _Ë4uÃÂ×ÅÌõS®`Ñuámö¶XüàªÙCæ×b©.Ñõ÷û4`™šºÉ3¼Z°|ݲòÙöŠXIÞtÉÆl¬ü䎼Ôö$¬ÂQÍqNZ†ÕfE<í(ŠÕ%ÖÛ}Óùk>Ê4›yk‹× 36½ª¼óSÆaÝ——iôC¸·y]íÎQ·±aɆ°ôgºØxÿŒ”Ga‡qÿ¨Å+ƒxvùÌ|›ßô|lò©×µÌeØÔK¥¸aâ2<¬¦m¯®Ø</ ¡¼ ½04 ºß ½;Ï4<ŒM^Û¾´°‰,®’6»áüéùOñp‰C 7U¢sÁËGÎGg¿œöñØ”$Y­ztf-:=Vã걕èdýÒ‹É6“ÐñäFû¦Í9èèkÓEwDŸ¡£Bß¼J@Ç*‚ô&N´ãÙ_’Ø Ôû»˜Ðr÷l3Bi“µeKB·¢ËW^“CWJm»Æóág…þç?¿@W–èI¶<Ïç_§e¥äÏ¢+BKoólÄ—¯Ô÷°¼]nú¸ÃBzº¼¥¢bÖ“A¼ô÷õ÷ÑOªªPäÇöxƒžÉ/^pó“j||ÏKô¢÷ªw 2Ы“¡¾ß†Þξkp÷à.Ô2À#2bŒ,zw‡Ä„Þ» ›Ü/3ÑG‘¾s¾_yŽ>Õ=¸¶íÙ)ôåÇ6ÃãrÐwW%Õ»ûPÛÓ°mO{uܺז¡ýÆ`á…—T‡ÝÁ¢C}{L‹Å-Šgïúj…%»?¨|ñËLБ|“ûËîØkR’‚˜>Ù"5+û¶„?7ê»mØÿN¸«Þl«=Þgîn+dðQûÖ”RÖ^Ýõî©S(j7`îå¾XôBÌq¬;=ÂËgR!Ö?yFè``46ìÓÿBNÜlTc$y d îw{•‚ûPBYni÷k”Ö£ëTËI¯Fi ‹vº_âÙ_ã®+Æ™ ‹ëüj’Rѯh©¶!ÅèÜfI‡Á—-ÐqÝñv—xöqR_I l¨}ºœƒ’FÌ9$ŒRoL÷ ežÈýˆÂÑmçÉ2Ëæ¡;“ãü­ÔQêNá¥}’QáãW.²ù«Q1wÌ'—^—QI†FM¿œxT6eÉ…!A¨ìÇrUÙ¬¼t°ùñÔa¨äÈ¡µ†*ŸQQ°jèiŸãÛðá>_ÚP÷Fm7@¹¶½Þ¼ïá‰rdâ“/N¡¼äç?‡¿DùÓ¥û·¡}ëÚAµQABæ‚'Pá’2ã¹6 ¼qÓº½±é!*é¡î±d×QT&¢2Å8 Uè©:=<UÜoû å úcÕì˜ jM¨vÊÛ•@OFêàl78å¶ }*.¸m›—оzù¾›3Ì ýè›ó{£s4w\‹ÎÑÃBiña­RXÄGúê×ï°XÏ¥å-K®`ñ/$N‹cÉ×ßäÎŽ3Å2*ÕÛ¿=NÅr»®Ä¶*vÁŠæÍ%³vÅJ/G®¯Ÿê€»ÜÌJ¶_ƒU®´©ô5öÅjižV n¸ûQ¥Oë7ÌÀ!Á܇pyE”¦)àž[æÎãÎýÄC¾–M´À:‘¢ß6~IÇzÓB+OVÁ½§•Å.Ý —œ98þ66J*52Ozƒû~Ü`Üë>P¶F]¬RÙZð"¦›n¿}~« b‘6gN¦-•WxØÙ‰3íÂfU>AÛšâð(‰Á¡k÷a,e–>&çæ–VÜ÷\W- ‡e9 ËÅ–»êžo7Ú­F÷\\Ò‚ÇŠ–zå&øãq5gFdá ‡ç,5wÀ6öÚ^ý¢´°mÞŽòÚôgØ~dÏþfg ±ÃDÅ/9GÆ`‡·–ïúnºˆ7ë­›6à6vÒØ|Ï<û)v:6wç¼ï+°³Î'ìÎ{Û#ÒxkÂ<ì"ÖíÙÒMì2‡30£qv¹ý}ÇJ¼ñm¥uû§GØun×S§†7cNÃàÙŸ¼P¦ŠžuFžï=jü¸Ò•Z¼y£vì}EÞûÖtÿ«ZyãR¨üŠŠmèÌŠK¶:¹(áòÇ.û#CyvU%50Üï•3çð7tÔKù„T2:RñÆ,A›hóYzyÏW¤¦'Z ìÈI©Œ8”ºÎSãœý”žx‰[‰2ºü!Ké©ÑÞ{Y”Aâ^:]bÚwÞx7pÓ+É•gxãYmutÌ Ÿ±ÕȨåm†¼°ì}€gŸôÅ”è²ô‚Ó*¼r-×¾5 «“nÎ87b ºönÿì)µ£Qú*ë'yšgÿczo¹ze—B¢n(»ív±pB7Õk=ÜoÊ£[%£Ö¡™wл¡CW@÷í[¢Î?<‰r•'MÓˆ $ª[=ƒ ãÅ÷ú—@Å+zß|u|8*ŸŸâWPyd†_¥Mªüãû¼÷ ûPõÌñBÇUG õæ{QÖz¨ö-Î8qÒ=™þ‡×úUßQ}ÓËãéËÑÓ-½¿.Y3 5ÎðÝ”¯‹šb\ŒçÝ÷F¯ÜõÆ(£7¹ºR!E¨ùQaõ—£w32n¿Üb€Z[m”×­BöŽ» ´$}šû­lÔë*ôe¹]]÷¨ÍèÛ]åå§=Fm®³¾Y»?Áœ=ç÷¾Z„…|_ޝž‡Eš,Çl«ÅžÎþžˆ%µ/6‘ÄÒ)>W–ä`¹ùÇ=¾{¸aÅ!Cû˜oVÆJßnß:„»Ô~‰t‡U{-©øÌw‹å´hU`¡¾Ý>gc­a5û¬¿bíŠÍf笄°N±ëŽ„0_¬/·Xçä¡iØ`eiCÞ”žØH§ÙsšAîûÍYîóÈYx ì©Ð/{Gaé*c/ÿxˆÛ´É3""ðМ~]óì°û¾W*æNxäª?z¬)ñÄhD×q07]}øŒqØbÁîò¹2ñxÌàÍRØê½ä©ÅŽíxìÅ÷¨7ãñV½©¼Ç2äêV‹aÝÄ¢kã°íÒ·%'ô°Ýõ™mý¼°}ó~«q¦#±ÃƯS’=–bGÅÛ®ng±ãŽ?^‡EóÞ×ÄåÆ¿ÒÂN!}z¼P\Œª$½eWòÆ7“ŠáÛ±óŽ.º*û°ØÞUO7£Û|bcôÜPæóWn žèz¯Ý-¾¯ÓQšâºõ-Âoøïm}gi¡ eƒÓNÙ.Gçãݿ묕Eg÷ÇO¯Bêèôq‹[ƒ&x ûL}`8¥¢¿güp”4óëMݯ«Pê¨ëgzEóæq{l Þ†n}Z³{ö‹@tÿÍ£|«¯ÃP¾ûÀ39ù#PáÌ!—Ê> ¢¬ã挶B%#bn­W×B¥·ð®ô]Q™¯í퉼÷º²g~gš²P‰µÓõ-"ê¨Èê•úâ“á?Ç·Àý§‹–¢Üi'´½»ˆrńԋסܻ7¯ÑAyhC\B»ÊËìc–:åΚ~j6*˜ìøJÞ³*4”‘¹M,I3~*Nÿòþ½ç:TºjsøD¯¨|¡«æ“A†¨2óàˆ·ÃZPõÊö!ËLªÙ®©4䤪‹ìÞõªõ[T¿¿“iUFOï¼\ôõ‹&znµ#6ãÊôr`j:'÷9z³<{h^Ô¢êÍ™{­ µŠÎX«’9}PœþN!9}?êì—E×Ñ—tqÝ£áÍèû¿mg¶Ü@íÃ> ùuÂBÚ#$» ?†…¯Ìk®¶è†Eƒ•D9B&XÜö“÷–· ±ä¸Éýúí‚¥CԂﻸ`ÙOIóÄ7c…'qc33°Ò.kùõ~×p—YwÖŽÝ+‰U&¾ûcán¬6þËìa>y¸{·C g±†üº£a5XóËò'cípO5K—à­‹±ö=Ÿ~É'Œ±ÎxÏ"†b=ÉÉ]-Ü‹°þÇU›8Ûµ½°.ƒ›GΜ}o 6Ý·ÔeÍl}Ê]®ˆnéžU«8á‡îôÈ=Ñ?p>º·wéµJËÐCå¯×·N@ùæGcõ8ˆ ‡©®!qKÚE\y¤ˆJZŒÖ׌ Cå"îûGêA•¨.ËÒÌ=º´8mØñèñ”I™²¶¯Pí¦ #‡·ÎBu-VÓŒtµP½åQg1ôT¢|‘LzVn,%%65 ÷ñݦ+‚^.1=óðÿõßá\þÿÿ?no²³÷Nö(q»Ù;[„Š6-«´””‘¬¤R Ù2BfF¶²w©lYYed„ߣ÷ñ{ö:ŽÏ÷¯ëÓùx8×óz»<Îûýzö=y¹[bƦô°lð:Xà&§Ö/,Evnv| +²'×Å`uuÏÂzXûÍøÌû‹%l`yÔ»"[õ–gÓ¢²Èüò¢Sa,°V9."é ¡3©:ëHæëæ´‹ )‹&c×/p" _ÒyòtJ¤+*ã©ÚÂ]áç‡Ù¼ÉUß3Å…Y,]Êd¿A6ßÅOå‰ïp÷§ÕþYž>ä:QÄäé>ƒ¼¦ór¿þ´‹»˜½r… D+3ÖPÄKS'¯] ź|ÙûÙ^¢äé¢\þV(-'A÷TóÊbñtйï(]¤}¯,{üŸ =@•½D¿éL¤pŸ_¦dÇy;T7íâ1Š@W<#w¶‰Póœ½þÍkÔ¨-l*Û¨‰:£+Ï#P/aÛ¶PÔ œ_\u݆+†)û.w¢±·™D‰Ÿh2Tú…6¬ ªÓÜ'fAùÍf3ѼFÔŸ]±-.ЏÓÝ–D¯Ã™iÐò®ÊÁ‡TÚhùçkñ‡ÕÓhu‘¢F‚s}‹%Îñh­úÇ5ZË­Æ”´÷!Ùù_wu¼QŸÉ*\WÖU×&¼Då¡ü”uo•r4ù<¼lñÀmé4ÁÔ/ZIí>;%8¯ß®9ÂûøRç@†îq-–ÁNH±ègõÀ/B£çfÂáI˜?KfîÚ€Â2"ð¹U»”7¸RŸ@}ЛOÏA“1Õ¶Èkh 8`óÔd7tÌùî~Åë]:K~ö]Ð]DŸ9è½²”74îO@o³ OHtôZ9Ü/à5…îks‹êÐÅYm;•üæù–ø´_Gä´†·TÛ´ÚO±hÝÏ€ÖG$,»_@k[f„ã5WhäÙ1¸˜ m+ŒÚŽ3QÐA6vÀ¤·:Ë‹X•,¡sýøÅ„ô(è¶éð: =ë;Ùçl|¡¯æQ,—d|‹—`¾E_}£{Ι>‚Á“‡Å2­ÏÃðwº…Ëy0Úùä`‰Á2ŒglR¿†Iƒ„ÚI…"˜uX‡Í&˜»„>«1ð­.¼TÊ¿‚Yf~ÉÂrÔ-2X­L–ÂS ë»Òl¾t…Ãf%?‡ý3ØVi»ŸœQŠDÁíon’0w‹V h i]†CÈk$O|ÔÆ¨4Œ”‰oM3'”ºËêPÌÇE¤³O0ÌÚ"Â]ZG'2]!“–ÅÉ‹ÈB×zÈhJYúÞl+ûˆìßž„†ðè"Ç»ÀÒb­ äzÕUµÄŽ<÷„¦}5"_5ûçÛ(jîDÌŠB$ò§u P8_Þì[_&І×GRDñàÉÄjÏT”Lÿ(O^Å{fý¯Þé>Š2‡ßôg¾)C¹À ÷@Q T ˆS&¶@Å—ïŸ<êz†Ê:‹)³ßëQe{mu¹•÷‹æöšFã—Á·U;ªd½HÛ7¡æQg.®á9ÔÊ8³£¯Š:lë¼´²P·ÌÄ´›õg¬’'×Ððx¯^ŸªËÔsŠI>A“oË~̇uñàÉFJOhö¥nË¥Ë -ø3—ý~¡Å‡ŠOâéÑòìe³jåh¹/èÿZ­HE´×ÝGkÛ½Œ~ N/0xìƒÖëÏÔöF›ã5ÏJ<ЦÖõWqû!B¾y ¨O° íŃû³ÿÀö›<2n#%¨xjYö5#J´¯Lž—U‚üʬ¸ÍÈɵ(Ë¿ÙõLeÚ!ו¼žNƒôø×™-‚5êÿ¢@iY÷ﺒºo¯à•^+EÄ£óàMY_1‰òñÁýß!õ'gÃãÈþË?Qqô|íÙfu7­“\¦9)%*t§÷r‰1-þD-”ü&Û/LÈ3£,JÇd(Ý›G-NRýOKÏ^a&ûÅ›—ôަB±Ï%#»e(6qœ§S?Å…ÚÎfázžóùÄBȵwÇrÏ&¦lAùÁ±îé´Q¨ähó3#]ƒ*YWã ¨6à¹n&59”G´ä˜ ÎK0Ës‚¼øwWç‡Ñ[«C&ÐÖ&̳tÚ¼"kœôF¡£IÌû’›t=¾¢Ÿqz<Û—TCß奎Úàs|#ËÃ\ð•,ù|“= Ä–ÙæÃ°p[Q3¦Âˆ¶ck£&Œ.ºPŸ‹„ñ{Á´ß'ÄñJÄ(˜º´ý2óF*Ì 1+ºrÁÜ£~ц°¨T©z?ùüÌ­:s€–äO¿a }Ëõw4Õ½`ÕωúùMXóê²ìæ€Ö¬Â'aKyU³Q¨é·¬7 ñÓºÇɄԜ8¤V³ÉºÜ*Ÿ]AJíœ{Ô2{:xœ÷!Ònu Fœz… .a¯EÆîÓwÿ„#sõÑVRvdýâÚmxÌwózäOrõ çC*êY~bäqùÓ Üý ù³OØ™\FÁw+óªÝP¸ŸædÙÎ*Љ‹Žup£Äk_Òá¶1Üsr‹VEïÊœæëçN±By¾¨Ã»PÑ"õËèû(T~A\MV\‰ª“egÀ#ûp!@ýîA±7ø˜¢²—ô¤;´æéýqkX†ïúkçt  ³îÌÏ'ÞG¡[>0J¥‘zbÚÕ°C¯üŠ„Ñ­ÐÓÚ–]fÑÝeJG¡sîÉC‡W_þ—oŸ¢)ߪ]€Öò-³»eMŽk=‹yŽ'õ;AÛoÑ-¥C«¿˜›­Ø,´[Þµ—~í'§Ù˜ãâ #Jvólº(t¦.ŠnQ®ü½'97¥ =—½ûz@S§•à/øLñŠM¹Ž¾øJy9=jOÔw`X¨IŽ—Fig3N¾èƒqâ³·™¸öÄQÞ/†×30͹£æ+™ ³'ÊoMˆEæÝ7 àmu[’,³¾ø3è «J¬w-j›`íŒa ‹€4l”¤çÞõ&ø%•iü€;ìŒ~T zƒÄçTâ¯GÒÝZN¼¬‚Hö[N-úº)Rlí:/¥¬€Ôrajg0 iK Q/ù CoÀS¿ydœuäm¹̹ÝR¦ªÈZ!—@º‰ì—ß|ë}‡úÕ\íñúÈ%HJLщ< M‹oå;‘Of¹¢L$ÌÕsüƒ‚e±¯^( °í}ë{(*VA»RFƒâLþ·‚?FI1¥3å >ʨɖǽ<Ë÷2Pó4­ç4{ýt|el-¨×=†b ~*¼Îc’Q†–v/§.hª¢å¢äÁ$Þ_hu›†H"­Ébú’fÑÚo“+øÝ´^>èù|ÀmŽ ·)•( MurËã§W ùvIÂñˆ,Ú^šö—]!†-aÛàS«ç ×û>kê\'(f +*FÈ'ñL:l¿ Ù˼ß+þÞw‹¢õ þ¿uþÉn ùÞïq·SJ·M], tÝî]íAMH(y­Òu!âŸI‡WIc^†ÔþÞŸk>RôÏyúûµwïm„*FCzPDÿviLZ…O¼Ô©Õ.óêÌ3Z(ñð&IUå&pd(EÏ&9ù=Ô¨ÿiñ¸îÔÀÛqîÒ¿þŠž¼Š–†¢sŒ>© (ß-oçï~>ËȺFBN²«MéS}ƒw½öÅ™¿oBù͟Ϧ͠2¯ì”j°=TݦuÌ´a„êßÏ &˜¡¶&Z‹^ôÔp¿oòx>Hª ÅòBSó}“›· 5åÖ¢ÎE1è8¢â²Ü2]Bq”vú5н4¼§K’z§ÂÉ·{þÀg¢¬SÃö_à‹u—SúiøöM(›äÈ5º)}²lwŒ°²ö8wF¯N±Tsü¹/:‚ÿ3|¯¬Š7ðh‚Éî¾=†0£?_þr3f+‰4o܃…×ûìJCðÓN³bþüêðÚ²­h„eÇ—AæY¿a•&TؘÔ~/FÓì97‚lõÓàOŠŽ™ì\ÛR»~c‰Í¼Ýüvo!)ñpö°F’}?'(ü )Ú.ŸçXÐ@ª%‘v6¤ÕH°ÿ¡‡ô_.l4q0#ã…Xçã1Ȭ`¼}ÎÞYUFÛz¸‘ýøìƒQŽ.ä¨t 9Õº¹}hö^@¾Ì°<«,”iཱུŇÂÐÐ W¢7š¯Ï‡Ä ø¼Ž]¢J=ë¢î5Îý¡Þ V¢\úE÷Ö,¨ÐÒÌ“èƒÊ´$!["V¨jÖœ¢_s÷¿‘>`©¸‚ê—dXU4Ž!&¥×;îy‹Zý+¥p u4Çe8*¤Pwc:W¡…õ˼ú™É[ÑÐf'ñ½r( † =ójB»Þ© ªWhZ£­ô³.ÍxŒO^<õ ÍOíßÌ—gG 3†ß ~h1ì±þà¢ZºNÒç-_FËÙP&šU´r©Ñ7¶;Vï{¼&ä[$_l™&Z?IšÏš±@Ö9ì{« öƒ¦èL ¼Wd%cð(†w³ÍqWêM P/Û¥höÛJ×°qøï>·lé3t"…¯Û2ŽÅ)$|&Ì•¯öõSV@b¥ôŠ‹¹˜š²”[4¯S”Uõvý„‚¿WB)Ç rÖçô-^ ¨K§³Ð8 ‰N8Žr†BË€E2hwBûw‘[Ÿtê}XÈŠ˜®¬«¼¡ÐCòTW?)˜sú©ÎÔ@‘vgÍ=gèVºtém“tVŸki‰'ÿ—oít‘*Âæ“Ð:ËÖ¡/`ûw?œjøÊ*´Jv0n–GBËüÒùÊ “Ð2&Û{|HZã£êl.ŽB[íÛ°‡ï¡C/I³‚{:m|Ÿ9 ]¡1un9¢Ð½8½“©~z]Ÿ†pÙÃ'É$’Ã4üгŠÕüŽ|c춌Ɍ‡!Þi£^FÑ×!güY c|â¤/¤á»I­€Š0?L1$Ý(³ ƒöû¾È˜Áÿxå %øá]žÉšKOÊüõr¥a%Ãïü·ø]9]lrá#¬Ok-ü2‡?J¾Ï…[%aûíRm9ó8ÕÆFÜa B×þà=/Ÿ!™IKÇû¤°;÷ ýžR…àS.”BZ®€ÍÓÓsH¿ýMänL 2>=•ré612Û|Þ6 ¼‹¬âÆ=L÷";qG·l„îîÚ¼1ùç!rfä)jq…"÷3ÏßÓÂÌÈû®Õó˜{$ 0K6¨£ ÷‘Žñ\&ÿ^ÇÔqE* ‚›øGPì9¯ƒ¹fJ$U¨~ì2D©otŸï-â^ýû ‡Ÿ ìb‡€Î93”¿ ž}å *¬5®NŽ Ò“ÃÚ!rT1K zºŽûÜwÔo½ Bµ~ê¦%¾ï¨a›ËÇ2—ØÚÖ“µ¤b4Ä)^¡öcFË`¥Û¨{@-]¨Sõ©vtoŠÝGƒï›nL#½iìaúŠ&nâ º¡éƶ*™T:š]wk¹¿â…æSaƒ7ƒ¢Ð"Q¥—¨à,Zê7óÔËCË‘·×j¾B+ÏéP¹Ð§hµñΞ«‘s×ú}öCëù“Í;OÐÆ1€$w«mÞ¥z‹%ä››×µ´õhÊ¿O›O6èr“R¡ŒÌaFµ˜•0?&’Œòg_Ú ‡d«NØÙ!‡,Ó´³‘YðÚ€K\䇤Óùó– Rž;¶ÿä~HœSì½V1 {‚ΰÔCœ_œvRŠ'¼b¯2 Ü·òûznûç¿Ül÷-)x«djÇÿ …Ny–-kÿý|†bÒpŠ1Rü‘ /R³çxŽú<Åþ—P’pG¹ÌÃòŸ_|éADÈ1—7OÞÞ\÷àf3†ê+Œ~ÏÂ7H“2S ¨rñ“(n©?3}®JZc/pÀ»Ö·÷"¡B0îÈ‹ëãðþÂrDäÛe¨Ö’q–xÁµÌ½i§êy«7¾HÉAãzŽ_C ÐD­ëõµú)´|ºÀ'õÄ Ú‡Cã–QBç›ú¢7ŸC÷eµm“gºÐë(qÀ9[>¹<,‘™Š…þÄO 2Þð'8É÷”& æ–9åêÃpНÛ®{0J|IP@(Æ|§yµ^ÓÁw…n§Ó‰aÒ$c×”4aÍiÖ †Y‡ºÎ*Xà>”Ä)À‹[ÕŒº3ÛðëÒ“çóC°ô³´í¤ë X‰.W_§I†ßlk Ra=š}лxþ°Ñä6kÐÁödÀ¤#Ù$&Á¨=Ÿ¤à}Ç–Z ’ùÞ•Ín4$û*èÿØe`Û,F–±…ß ¬hp$3U„N<‘Û&Fs•º}~²h¡vNê~B‹Û7†Thy/Ô¸E åpÿe6D+Çøm·K,hÕšÖc¶‰o‡×ù¯Š¡õ3É !ŸP$šò¨½­× 5Ÿ$Úùýë3x'äßE/Ye¶fj¶ÿzrY¶›ZBÿõ“ü×§ð_Â+ï¾}'yþé?ÿ¯çä¿ó×íS‹Ë9#CªÚ2 …ÚÑ”©2çý"’s5™WþõštŸ?&Ñ,çÝ;/î7Zýë/èêÎd/N„ÎÇZTðBîù¦zòÑ)hã–±¢º7­n ³x¡e¹‡Ü"Znïv¼4%ö¯ßä¿ïÿw|û¯½tÂü±ÐIB]ØPþÿy=¹æl5ÌÐgš¼pÓB>ß´-Њ‚/“ŒgvÙ•Â@cÇŽÑ„* û=s®Ð…QµïÁâÁøÁ³#±60IjP¾p fL‹B›×)`ž;–w·cüd ”‡%Ž/”ÙL°Â7{(éüf™º[vÖÙ~t;¾—‚ME[¡GG`+€Ïf&މ¸wêŽçd!q¹Gš8’&¸²¼$Eò¼ÆL±ny¤œ#%!“jDšW{j‡O"ý‘çµ—ry‘QÛ}/sÐdZì²èÍB–ZõíècÈ}aÕß0wS :<€œ U'îÈ"·×?«5òê§Š“–Ç ÿ뽿„Ÿ£à^o §·2(ôæývÖA9”Ew¯Å„JTØÅ/£„ ê=_¢”ÍšéÀ;”._ ”¹‡²Ç5Þn– ¼Àƒ¡V9‚ŸRMÅ™ævPIóðsYÇT^õ5ŸŽjÁ}ì©—®£Ú•&÷Žú5TŸj-*áGi´ìÖH'·@«ÓiœãÐjñkÿÃ$¿yÄOÒR£õ4·¨“hÚØM¨îÕ±C›J.?Kñ B¾ªŒ19ƒ¶^B,ʼt°A¹i1 ¥ï†îœªhùç¯ÜûÛY'îGþë)ù¯ä?ÿ¤®ídÝÌüç³Dç]‹r2·!^ý£L®Î¼<}ØIö>$ÜÉÕpÖ÷_ŸÉ¿ÿŸoóƒOm^JüçÇÂr2ñ(ºk>Úö‚_±;~Gµþõý¿úÿö 9SH¶ÿû;þÓÿÏyÿÿý,ÿõý׃ò_oÊ}'5.×Ü´’ Î‹åüÙJ÷ý&ÿõšü¿×‰ÿú‡þõ µÖ¸×‚¾ß«Žÿz޾:Í?[»@ ß­É kaXå»MÉÛT¹¿*s«BF7ž¥,+iÂxê!å€Áý0Qzˈ" ¦ÙÓZ. ¤ÁŠáƒé›0ï­SÁC« ‹ÉZÇcµ—áCø³aþß°ÔÇvMtçŸ_Wgv_ٺ딻Û˜í…Í{K|Ò—û`Û–!ú@h•õ°&c’œ‰[æR=†dBç&ö¥!ù©ú>¤lLUPtAênfá«aH'v‰-™+Ɔ_àË9db‘~“NjŠÌ =õ?“ ëܨÆgÖVÜͱµtXR9_Ió^=…ŸÑ9 :ÙÓbÚ'/A—š0gÉÆGè¾*öÒúŠ;ô¼Õ"׿S„>ÍoŸƒºnÁgû?,*ã‡àK–0wÃðtè«,„ac&«~2Ù¼³/ºý<Œ“ÄÅØ_H†‰¸U%Ò{˜®^ÿ!Vz朙IEtý`ñů9ß×áW$÷»w¦Ù°üáR'ï1X½Áßú±¦ Ö®ŸbðV< Ϧ3¤æáÏüý¹þ‹Çaç•þ •=;_H—ÚkV„¤Æ—™ŸÇ“"¹¥¡ï$Ÿ'Rú1•æ*!õ¶qlŒE3Òµ;ʯ°;உ}DÒ.#SüÇ^ÍI"dq>µh}Ùö°jšu9 ûÊÙçOˆ‰£ÕÕùÇ£ äjH;ðüU òüôµeju@~GuÿkûQâÎ釖ã(hí±\ê€"-%ñd›(º(z#ûìKŸa”|&Ò€R<Äæ…OQ:àOKÊÊòÊ3ÚòtóûL%Pá©l”t®*)rÉ__<†Ê“³o_Ôã>RW‰â!eTs®ºÞ̓ê]W=hRuzòüößEÍï{ËO>NDmËÉŸc'ÝPgƒï‡`¨4ê}¸Ìµ¢º†Ia‘çKÜ¿d¦+ É^ù‹¹QÑô£’+•óšYG  ,¡yG²Y¢EÂY¿×,ÐÒrÚ‚#F-ÿ´Æy.Õ¡Õ+Ù²ÀúQ´Æ'ߎi. õgÅàGYhsžg¡Îƒm6CÚJ¶åÐ6ÈDC¿í-b|Èêú2 ÅŠš ±€u­=S¾˜¥Ú[7(r¡€ÿaÚUÛïKkXMu2¯.P/‘Ãë+’eÞ¶nŸeº[÷6¤Òí<^•&äVиBMùÿ}©y… â„‚¾›‘}7„µ¼NC+±—úõHªš²ò.õƒ é¶÷–jìðæ]\cäs.ýxy—ÞþÑStÍkÂ# ]ÆYP¤^Ú’GÅ]„ù“nñæÒèz3kÿÀ zn˜sú 3RQô/ûàs8Eùì5ø²0Ë&¡æ—Üfô»¹`è÷xLÍ…t±ü¥Iº˜ £ ÛfªnÝ0î#E1& _¬žrwÂÔ×—ªv9Âj:¾û0/î1|›È ý>ì­â ?—³å>øÅÀÒëãyß`EPä—ä©X]Qò±9O ëŒe®§ä™aóΦ¹Ùö{ØÖ0!Ó¾D‰á®§îÏ#‰6ÉÙÓwtuðÛéÕÏH^¾Í·Øî‹”iÁW­”Ãú=Êmu!÷Oã[º9ÈðeÁÇ,™¨•¼óäÝùëgÏä9ÈÚ&èªÄ9‡ìó+ÌSÝÈyî]¯¢í7äqÕJ> ßŠüûÓ˜G›<›íiPxmûª\æÓKûAÁíñ¯émþàžWÏg[Pv__{飔O¸#;¬ŒŠU-Ë×à-*/¥Ø]ŽÂ}–å’·íñ» MÆc£ùç'o?FÍOª3a¶‚¨Ú(<.ÿuõª<$µPŸnwí¬-仼²?zÌmŸ?1FãáÏÏp4=×¾5‡ÉŒß,Esó᫳û ÐÂáàëg×ÒÑ’Òº#{Ô-ß\ŽzE>VV½ŸI Õ¯óÛ®h>T$VÂ6’vì7æºÐ¦F&7¯ÃN˾|ª!¨NQÙnwyeUÉoníø@qK3ÓÑ´/?Wæ(që_?Iv=[ðÃaÈt9g°rJÒ™ƒ„¶“k!éó¥‰çßN@‚ã›[4Íÿôï:å,r ÂyŠ]¤OAùÍá M¨•Œ,'ç‚FÓ/{¿Z’AÓ|ÏCŽ(OhýÐ|g¤ ƒ1úÀV:Íh«¯8î2“¬÷Yœ–H›Mà©ê„ïÏ~ŠBçdÙ7Þ‹ýÐÑʪíscÚkyMÄÅ¡M¡Z%þ^ ´hWŸxIȱλ-N‚éÐò$Y¥óg´˜ïxËõ?ø«z}_ê¡Ux6e@›KÏaÙ÷ Cèàó¶Ç‡ s΋*!è2ií ¹ Ýw百–æÄ"…åÉtèSf­~dÓ Ÿ•=¥÷rÃå^ ¥ àÏQ ºÃ,÷E§Î]€‘rži’þ0V|vÔ—`B-9“>@ ¦U5Œî\{³ÉÁÓò{MaQ¬ªÐs±~Q¹ÐøÈ–~%oHLÁÊPÁÛ]MÊð{pÐ(4-6ˆnÉ¥®nÀ»•À ;4á&&V½HÌ7˜ pÅI–ï0ÅìF²Åi®o‘’Eàb˜r"Rßá9~¾Zé´¸®ßŒÁ]©JþŒÈ¤_ðYnÍ™·®„Å1s kõ»n'dk¿']9Î2µ]ÏD.WÅ«ì2%ÈóÂæ.ÿ òÓ™0·}F¡ð†úÅç(ä4ñ]Cô oÌ‹@­ˆªÖͨGñüiŠ‹2(9Fcø ì2J›“–10N£,±‡øÝ'WPnÝ"²åH*3ox¢’è— ¨Üï´Û̳U÷Œ,œE5ã¢{]ǵP½œßÖáˆ4¢¼5F<›GÍ¡=ñ)hÚbNªóî+š9Fþ,Fóqþºkí¯Ñ"oÀ+þÌ´¼0ö±ô;/Z‰yJ<ªŠV_ŸœÏŠ'øÉrHz°mPÛBÐ møÞI—û¢m¼ìiÝR*x'ß ¶)3ç‡ÒJ·îƒiÐ|2äi¡Â-h36ö\¼ãÒöJdÐåª=lÙç=_k´*¸¡vóe­|–šYI;_ü§#í)‡àÛúƒ;÷xÀÐí`oÂÜÉ9oy_„F_½þêŒ,ÎM‰Á„•Š“x>L%<ÚŸ½ ~ÈIØ«E/ ¤#+ɰxÄlmaº~~inÌÙKîÃï`EøHØëÒÃð›ÈÒ«áË3Xç:úõ ËoؼMêç”Û¢AÞÐr‰îVŠõH# Ï5Æ"Â\FÚ4k:YHäÖ‰3öÆ e[¥Ão¤Î<»rÉ‘ŽÉë¶ýÅ dèüê—³swÔ÷¿;ŒÌßøj; kŽst$?²×Z¯¶¤;#§ÊA!#Ekä‘J®ÝË¿†üñÉõŠw~¢`™4o9o ·Œ19TblîfyÚ¯Q"¶ö ¥ínÜã:¬FäNŒ2ëÎ.ÌQþÆüû¸žó¨ó¸‰ìö*7 —aLÄ}’z[†mPmÁ¬Bªï<‚°º*í³^Ô¬c½üi9µ¯—éõÜý…º*9¹òCe¨O<~†SÅ réÝ‹ÑÈv,î9+ß~FÍ}vyŒ¦þQÜzW-Ðl—ØÓ‰2*4êy3D[-ºWO¨  ¥ÙEõCK´hÅ`mwi"ZµI˜g§ª£u„Ón/.B®YË– Åø -wÈa›Ô´iK¬øµ¶?Ž„Ý†j:ØMÛ©ï|öù\P÷‡¢ç³Ÿ“W” ŸáÀQæw,ÿúI²é¼w ׃×GÓÄÕ› :É35ç!I±ñvÈÍÿ|=×ÿ§©k8&k™ùF”1k Pö0lûdÁ¨I<ÐxRÎ êo휒‡&—‡Š±¬¡Uò+ãðAhwë¢w°Þ†Ž‘·£2¾Ùйî¦ð‘ŒºñÂgfÃNÁËCðEÅ0T) á[ãÅ€tû=0ôŽ„qs3 Fn–<»ô;ÆÎ+J›1ëÀ÷>k¡Ý{`*Ë/í›ü˜ÚivNXpáÙè4e‡Ÿ·âzú'`ÉÉ/QXÉVtbl;_Ào»{)܃°~†¹øC-%lV»ð3¬Á¶WoVøÁGHÔÁÆÎG„$qaŒ»*‘,ÒXøLÁR¤ÊŠt–1"5鼤–ü3¤Íÿư3Ú‡ 铦ê2‘q.þ« u2?lÒ±Œ¬öÎÝ“ßk‘]°­Ú\9HÊ«Ž9!íû‚‡e3Èc°'oïü òUN2¹‚%/Ä6",PHñº}ÀÃDneàK ÷GÑD_gæPüF …Wñ”|Í¢e~Ê ¥9h2¬H¸Q¦“æ«íXÊ}ÏŸ6YA…««g+ïB%ŽXgfÎ.‚Ÿšo£ê·ùXù¯/PM)+¼[|Ñy“È×Ù·q5û%n®˜ª¡¶éímž÷ƨ³ÆÓlDº†zõ&}Œ<:hª›žL˜OEeÍ>G™YÈ*EÓv×—÷ÚVÚq|Њæ¿cÕ8›ÙТÖêtô´¼?¨ã¨ÀŽVæs‘™hͶÅÛå~­èF6næ M*KËÝœu´õ¸÷˜riwoux gV±OÚ­MVÑis£ýèû(ó“ŸàwŽþÓ;¼‚Pº7÷ŒPÌUxK&-aäC›A²FÇ^gåûï@ºVL¼†2¤žã“ØXN&pÛƒ;N?d!Ñ™]GpW+$ì ¹æBn±óäV÷áyr1íŸÃ©gmh©ñ{sž§iûæÏÿôÞÓ£–÷üŒ¨Éß~ËÕ‘Œ.xøª˜åE¿Â/F®é@á›°'›êãPä–xý*cx¡2eÀÿ”G§É×¥Šˆ§x3@áþ©ûá'|ÿòÚ¸Þå;y0»ÉêŸ$¬àò“€Â³kÊ«BPtךÉ÷îî=FïèÑì*ÍS(OÊò<ï7¹/ÊYkCu¶Ã·Ï=Pû½Ÿûøhȹs‹ñ|ðâíâw‚ff!ÀÙÓÐ:ËyÉUv :BMÞGx·@—г³ÄÐÝwö±ý6ôæ>|µwª >Õ—r¾I†/l“´A½&ðínGóäG¢ãð9y݆GfN0Ø=€ÑcqŸÖ`œ£Ò„jŽ&ø¢N•†©SÁ¹.³µ03mà™G samâº3}°¸¿·ŽKC~VÍ}¦ÕK'­ôÊ>>…Ñ'’Òôœð›ö¢û~#9X!9¸ú 6oÒæE3>„múÚqÍóöHäNÊ<ÅçŠÄ«üduuH·^qÖk?’»-»<(@Ê3É"¦œ‘úé3ÁÐj¤£X)PˆIA†ŸMäÊ ãŠÎvmX 2¿jÙýŒYÃãßÏŠH#{Ì­µÀ3óÈI4técrÏRŽò­‹ ÿaÆÌ÷¿’QðQ؇¬±s(œÐø:„ÄE¿k^ñ¹­„g&‰mq‚Ø]ÖÛÊ(S:ý«çá!”·s2?âe‰ŠîGã#ªÅPùÕРÜGÌ7ï®e‰j]5!ÕA?÷»ó ØP3÷'±Ð£Ô>ûæL]5êŠÇúÿz¢Œz¿s·Åü ²‹“|åÑÈÉ#ñ倚œp´x¦ñqOßýz‡f`øEÂ͇Èü$LåÑ¢ÚA[z#-§‘¾B« ò<WÑÚèz£Ýô´‘61˜<Ñ€¶»oWñ]íÃCô/ö;ŠvÌGO<šú[aÌ·â ÊKOÖS¥{[ÐñZ(º2ÄÿЃ<™2çíñ´ý$™.¯óB2óåÅ]§óÿqMÙ pBb?ÇužGç÷êî^¾•šJO¶+?Žò’í:êÈn»T0L¹Ä5\}ãŽ-ÐÂ+ÝGácÏÁ>þÓÊÐ2ã—» Ú ,½L¶Þ@‡wcI¬ßwèô­Uá_€Î¹´0wYè¬ éu“r†ŽUz1&è ›¼™ë˜m2.rå3xl•*3‰c^vr•Eí„<»bNòT‘ Þg­ E`2ôgÓ´8Ö›hðñþݧŸÔB mé×Éð@‡É½7žB§Í>£ûÐuIØxotÇÞܘ5Ì…žröîÍhèã½÷èõáóði~A޽2ú7u´˜ªá›ÇÚ^‰.2ýÙ'.p F4lŒäÀ˜ÀÖÃñ(øîIüL0€¦ë(T6®Á½W ·¿Ã¼wIHyQ,® w Ü…_ Ne¦¸Ë郾±ý¤°šÖñ¾"’ ÖZÓ`&‚6¥ZJR†´ak€—¾ÓЉRÔ§ ‰á³]ÎÇ"™lÓª‘Rì×øÊ®é‹TïÙ‰Vä…o€ 2ÐÀ/}dŒdâÚDfÕ\ÿõˆdY ä-¸äƒl%D݇VÏâî—'KÕ3 gŽÉOz#^äþE2t`'ù.žU̸Ά3‚de(Pˆ†¾¹u«…ãe:PÔþ‚©—4Šë>_!ÚFI5 9Ü3ÂôôñÒA”IHŽWD¹.6­æ@×þÝP‰Öñ÷‘œ¨\®6øÖ*U?æÙ\ìoG5¾Ä˜îíTp_ÚK²‘ÎÂ1Ç¡5»E4ÒQÛ¨…lÿù2ÔùÍ]æô~õj‹_†»¢ÁÓq ÏÐèØÏ•½ámhŸI^ÚŠ¦]÷Tß]|‰f—‚.Ô‡äì¾i¢EOãáûŒÄh™ò¹üÏm´ºæÌÿž­-N´­é½A›=ã¸ÂÑ–î;'±z<Ú®™mYë㡉«dÞ¡Ý耷xm4Úϱ° DÛ¡å23I7¬Žw/ÒŽBÉvÂÛ¢'È¿Êúr5ý1dSxK] éîÉïªd MÜB."R,>í>ßRDà1Ë›Þßóà•÷€gŒŠ-Ä –­>ë…çÝuÔâ•ð¬Ôï§fx¼<í´®áÒAÈA¦ÊªÀMÿ9Û$í‚ì'd"$6”§¯rón…*¼5i_›Š2$ðW¹zy8!‡~ß ³‚"QÅGÁw‡¡hÄÝáæ­ÿi¾ËrŠ1!Ã÷ýžøÛ·bõbc©ê/¯}^·çú˃5Îæiÿ´À,~‘p<Í—õ6ûd(¢Ïž½ö¯Ç¨ôA÷j|´”«m¢ýxÿb-moXT§ÈšƒÚ¢úLÚ ¨ÿQ|˜©À>Hûî?çùšÞ—ŽÙ§šAkØÔ¹‘A/è°<'ÎQ _“ü ûî KÕò0ôÚêIT2Á'—(C×ç¶ÐŸ>XœX ßè[r<`ð‰ÂéB¿rΘð~²£ÒÒÄ !06ý íþ/ø>3Épð%LIݲÌÿê3‰§ÃÄo+Áœæã´aJXdwŠÐL‚ŸiT+w¯Ã’ýLqÛ˰"ÞV—íü~³?8ŸYÜë{SÞ?í¯†Ík,£EÏ®ÃÖ*‘6.ª8“q6 ‰;­÷Z-}GR/ʽÓònH®5ßÈ9ÝŽ”²Æ9ÕMH}·Á^ÑÕiÿøLÓÕ¼B†÷}§øœç‘ñ‡ImKY-2Ç]íXÈCÖ‹ƒ? %M‘ýBƆ¦ø ähö(÷ž×BîìsÄŒ]æÈÏ|3Êô„= ž[ðz™ÁƒÂÞY:EÙó(ZȧQV‰J4Q›¶f(µq¾E‚à™ë^÷Ž3r£¼ÂÇ!feT4¿ÌMr •}:6ã©>£êçJ–ă¨öê¨â99.Ôèýä~¸Í5_¶†^P¼ˆÚ‡’îJ .G°í[Ó7¨7çûX±I 2'G<—@£ã;©¯R’Ñ„EqÜ “д<“Uî‚0š]8§€ÓЂÅDkl Z,‡­–1¡å·$– VU´ja¡½\Öu­H"ým>ª]u̾Œ¶_Û빫à¡Å! ´ß?/â:ê÷y>½áï b"¨]d¡xܙҨè áÿVîŒ|Ä$äš8ñc—ú¿~’ײñׂè!ý0ß÷&û]òôd3Ù5]x5s%ϵ<“௣ IÉÿ4EðƒºòÌeÈÝI§Šè †ÒwóQ’¡PýèáÄÉ 1¨/úx—5>Ð\MìØ Í?6YH u«£_#_”_#¨îñB' Ï~Ï=Ðé*eæ°ï,tªï´ZéoA‡o‚hÚ¬¿û>Ÿ=¥ƒÖš…ÛѨ åE¤GÞÛh¡v9{^ǘpÞ…HÛÎ5Â×NwXºnþ½OýÌJ'Z\Ò.bÓßõ'OÖ¢Z ­¹õÄtŒt8 ìŸáM$¼ÞîŒ}nêÐå×F¶ÃÖÝiº—¼ בž÷TµÚ¦4Ÿ›s³ðk9|j~AGyú«.lôÓ…o<þJׯÂàÐH¬•ž o-l˜ð×Áhoc•NÑ(|§ÏeÜso &ŸRž|ø®f<æbí&ó`.؆+]p,^ \8ÿäöõ\6ñƒez¯ã#Ãn°Jg´ÑGvÖ”4Ð9 ÀFô>Ëè7¡°¥ò«#)ø¯Ï#ñ¯†‡×´ÝäÔÉe$o1}èP?ƒTÔ‹f˜O#ͧYš7HîIŒIÎ2¸Âš ÿÀ(ñ¤»|}:g`ñÇLQqÀ°Çh G¥'Œ¬E.èêñÂØ»Ãd4¿©á{±ñÕ~ï˜\ðZ"m’‚Ç€´—Ë50»,u‘¦w,|o(øÀ?ÃÄCÝH¹aÉ4…1jÄ V¤(e bá·`º+Ÿ˜¬«ɧ—§Âæ–~¡ã°5¨^§“%D>ånùãHœYuà•0%’°Æ¹.Ä"9ÇtF—Á¤gŠlUDêKK§ôôÕöçïÄì d(êþtˆ)G¬ƒ{§GùÁ¤hˆþd5çcO–9ìÚ=›ë“ÔÈñ„Œ:¤¹¹/¶ dÞ@¾æ©«Ÿ{èPP×’/üb+ F~;€¢÷†H× Q‚HFyÅÖ¥jÞ²©¿ŒCµ-¶…ða”gP½ø%ùSubPÙ,DÁîÞ2ªæÞá/&5Fµë§æ8ýD7¢ôDSQ3äõîâ®o¨­û43åº*êR_–ð&•B½1ci-V14È OI$uA#×°×) ü&Bvì­Ìk4í/ŽL™ E³a·Ï“üF UV|9´”8Ý_u#­„+n[‰´¡µÔlJ, Úhòe¹› íiÉôöß"x(¿‰}4Úýì˜ÁÃ.¾ß±`³fŒŠç*Tú§¾÷×üû¹äR»êÒ+xKn%ã3  9CWZs¼ÿõ“¤oµÞÁ»vÈ¡žÉ:’ks²óØÜá•^›ña§ZxaÊCç,aóO“UBÏ–ª @®Ãš¦cz4Á×qbc šN}e<ÿ6ÔË^ðf´-…¥Übûסù uZõåeh5°@YKFhWlúÁð:/«7|*…Žñì]º§¡£Ì»^íÄs·Í‚Ô€!´]»|¡ëÊGhåP~Èàµ-lR_Šk¥¡9þ\g{=AÏvfÐU@óÝg=ž Ðü³=%¿üÿÖ™X$¹\VÞɃ‡C¡mAàŒ/ áu¼Ëê N¸ñm¶UÈtE]}ßñâ)t¿¥ûÓó zª7òkìnCùò©ê£tð)ýÀ<±…4ôøåžo€¯ttMü0èÙéìë&Õ^—¼ý£!Q4p_Æ3Ž)ß›€I:ãšÙ 0]tòÎÉ_˜]xýèÁGXøðÁkÝñ*ü|Î:gG KžT†â“]°â^«<5 ~?ï8(#¾Ö·’fîÞþÞdvØi´ÁN£ûÂÉA$~‘Gž`‡¤[bÓŸ!ù1Gƒx{¤ =á[éߊ4T¼‘láH×2Ögo⮺êÆÑºÈä1sÕ–>Yx¬óú!ëøi3Ýdï*²Þ‰ô@Ž¥“nd'ü‘Û ï€½ òvØ\µF:ƒ˜sž(X¬”´…uÜU_š¢È’Á§W9'Q¬Ï´³ìQ;JìÈìy>;ƒ{¼ø½y{óQ?­¶4ÿD¹×ý¿šŠÖPÁ$!#ÿ¦*.ˆ.ÚAåxÎT…$TMRi˜êSÅý¿Â³zŸw¡úñšÈ4}„Ƀ>­ÜÏP³QàeÛQ^‚¯N2z6¾@ÎØ{ª¨W6_âYˆ!Ö$Ãhd“% Ó­‚&왺WÐôséyy¡£hæ/müŒ1-8)Zof¢Åï<ô,¹‚–­¹ÈID˜_^ÓxAÖõòò:‹ûs£ÆÌ‰C¦hû%(ïÝÚkaDû½Õu^Éçððe˜gaòC‡^×j¿}mèdu„É5+ V,çã(¦ÿ]oo:ä%yGªµ‡ ÿ>o#[’œRÏñ¶DþŽƒä_ï%Mc& ‰ÇpÙâæ—ÿã¶ä/¯þ)‰ž¢bçiÈ…K4!&^ëÊñ™ðtŒµìÃ:ÄFÌ;²½>𷯋bi×á|»Ó²À²L3T÷åÝúû\‚ ÆqÈå«÷üµ! _›õ]3÷B¡ßí#¡OÞ8N‰ïäi÷Z8g¸ÂÕõ—Ç.®?Ì‚Óbµ=¬úÿÇk.¼¦ÿǃÓÿÓŒ9OzÙyÇ%ø/ð¦òóDRÓ¥kÿzŒJ‚·ôçÓ*¡ì¸×@¸È6T¿å” 9Õ´&Wéš¡Ö>®éGÔßùHµï4Æ«S‘>†&®i™?\|ÐÒùÿ#ë-£ªz¿/^º[BTBR$$Da-R DƒREA1HAB:”nénTºAJº.ßÿ¸úãÞWkœ³÷>¯ÎÜsÎñìýyžØµžðƒÆ|£i“+ZЬ÷nçòµ;ÐÒGwAQ$Ú"?Èþ:£oUã¾èjè»Ã¦à½<´¿ž4ÀÏ”‹j¦/\¡_îWHÄü( T,…øÜ΀!?k¿y”ðБšÆrOü‚Iæ¸rfG}˜N4{í\㳑mÚ’W–aÞõ$¯áùbXTlpßvƒ%ÜÈ °„‘o^.Àšâtla”/l¸ò©n¤jÃVå-)5N$à“d<¼Åƒ„OD5g …x?ÝäÇ:$™~î›@Žäd‹ÓnïÒ’ÛÅ”g©Ç¼ ¿°†"ÝÇfÍåW‘¡ÓB·m™nL¶¨DP"³˜•2ÁQ?deÛÙÊNÁ½ÖÁ&ÑÉâ¸_\þx'"çãÓç.¨~ƃ²ªïIÝEÞî ¤ÿtÆLT+ ÖtÍ{£°ïÚYý-(ºñ$ñmŠOÄm¿=µ…’+…ÜÞ-(ͯâÀ@H²O%™Û©ãq#çäÔgU¨ðÔ%9‰¨o<‰%rÑFeI¿ú :=TY·Ù*ŒCµnÖ¯zg÷£zBABŽSj^ÛŸ–,³ÛߤÚz692Pg¹‚v¿Í2êզׅ^ûï¹’Ÿj¤)ÐÀb†ßWž ;'7¤íÐèYcb™žþärÄ$;Mˆ÷v«*¶¢©ÝÕF&ö?h6lj—ÿ,ÍŸÑ&°ÔÍÂzÙ‹ˆ]-@‰~ÎìWäø„ˆ€ÌV×¥Rá0øÈíÑú¤ªæŸ$±vÃéɽH0 Ž3;ubgW_Ye†Cô{þ!Í^—èÖ_ûHüoÆjëÉ?Rƒt§WszK°ÛO ÚG:B™÷8©ÝÏ.¨HbR‹…*‘G%ɱ½P“Æ.fÓ5 u¯ã¹×¿@ƒ%Ùo,ìÐDÕ ¸ß5r·W‰5 üMr%<÷e¡qﵖ׿C}¦FµÉc¨s¾óôôÔ¾ø¦RÒÒµÒîß-• ÖpiSýÔ¦ŠdíaØõ9•¬=Òd»ç_12yÄóß~:²1Ü:ÐÈäz™Á™šž™ê%ò Có›S™N©1ÐÇëÃk­¥­f÷ÍC[ÅtŒ²í´¯‹èÕõ[@§÷ Îrèrß!DÍS”ÒéÍ$4f>@‰:½mdN^ˆz~w ‰Æ >OÅSÖh@) iÿ< Öª„ú/¬¢á²èõ:_O4Ú tøæ…Æl«3âDÐDÉlûÖHšÞL“‘vC³æ‹;çTÑ®¦Rìú£Eu›ZÝxýšƒ}þ<½½ô´‹ ®_ëZLM!njBðbd˜(â/™ßüf±]ñ©4uCD8X~zz··Å:·Kïö¶‹„F¿ Ì=“ZUÝBˆøÆ@ðʯ÷´Gþ[‡£!l¸Ñ;'Óœö€øŸ‹”f4”š¸QÖÚ2\nÑ:¬ØCV¸y©½KÅÿ­—qˆ^ƒœ éÃþžá{„_Ù?ší¿çB¦9{ÿ7ÝŠ]9ˆ%þ{õÙº½äp½êTŒ{öëxÌ;#ÿõAfѾ¸ÿMÓ@1ŠÏ½=÷HýÞ}¡ÝWzOîÒÜ?ŽQùëYV w(Zõ¯ô¦–ØÍׯ ùå?Á¶ÕyBNø*ägNÚbßtšûê¢÷Áw³ð«§R衺ê~[}Û®n]m´'Y¡QéͧK³½ÐL¾è#ѱ -ÏÃiwªžB›Ê¹ÏWîC‡4G[È“,è²ù±ªNµ=ޏw0 ÀO©ûRIÐOØ“ÌË^›cM–2 Y,–Ð,È>‘ßú{»uN:>ñ LÔýáÛ÷ ¦ÍŸä4}ƒYÓ–z˜?cWt¼Â ¥VzÖhaéÈÍ KX‘´%&~ kZ¯Xj^…k"œRR°•–wr𠍦/̨ŸCBë03…ZH4+vïòå-$©ì»ªBƒd£ÔzdH©¡{ïÒNRÿØö W Dº¸†Oß\Å¡þ<ÓøId:ÿûîæb.2Ó¿s¶<–Š,3Â;ÎùpïáC±yOqß| Õ¨U rB™Ú|®+r/1»]]Cž9szùŒ,äç{Àòêº †ºßž0ŽBa#L´+&CÑb9ªç(þÉßô÷J6>X*®¤F©Í¨ª;Â({nHë› q·«Í‹B›²“§8XPÑîü]Æ"Tæ¾7û„óªLéÕ/]ãDµ†!s‹$zT;ŸXÙÞ‚šÉYvvû›Z¼‰M>êîizöjWwzë-ì?ƒQÿn‚ºþq"4xézÝhÓ 3LâÔÖÑh©3–Üi×”ÐZVOkCmß½.¿rÚÛÉvÙ¦B§ÍÓÔ©\è¦Oàr©­Þ\Es—àghÆdkö7èǯÔ?È`à@¤”h3 õr×ô.€‘åæ{ï7¦a uïÆ°ÂäWGëi˜‰oT<â>sÎõ\sWØaá„$Sör4ü‘ËýÀeC Ëök×?Ÿ„Õ¦Ûl¸`ãáñ7?jÁös­ï U.H0WMý¬ ‰>6( ‰Ê"Éã›9‡LŽ"ÙrjûóáH1)6H‰Ô/#|m{³ÎÚ¹5¦,>ð9gø)!“f´êh°23ÔV|³ˆ¬zmm¸÷˜×àÆ1Üq.ÄM9 9$cÜ%›^!×T)±A ´³M3Fž±Ìˆ˜'­ÈçÏcÛ$žƒöÛiÁQxè…"Ï»x2àåÛ‚E¡ƒŠíÞ“°$ž)™•þV”wκ»pÁÚé£>½NƳT¢b l/0kø­Àή»¯ž ¿è—á„D_åg§…E‘$¢Ç`ì²à¼°àU¤”ò|îùã R·Ü/vw|…táu¢[¼³ÈðÕ±sÞí82Z8œ§e{f{C j‘¥Ü„P? Ù–Šæêì8q_ÖŠW—Ì6r¬ŠˆÜ@îz.ëÏ}ÈSI–šÈ÷Ç@Áäã Ú½å2þ„ÂûÏݬ=Ž¢^okB© xô}NÖ•(™¨U«A®ŠR-¹÷×ÔPV6Ðá½€"'½Ã55ä 'h—eQÑ C2…P™ÎÑ㚪ü8'Šjéó—ˆQ=†.©©QóŽl^Æ÷xÔ6qo~±QŠºâ?ÉÈLªñ”П‹óZ£¨sU“ ¶Ð |­„£x ×hÊuM^áiõÇT©ZÐ8‹#õ¸9šÊ¯iG³únÊž,4¯*™¼´¡€–]¢‚_Š `õÓ`Uê£ÿöÇПþå Y¥é‚D)eÖw•"È7 ’c<úd]üã“Ä_2Ua¾ qzt¶£'C!Fã¹qÞÄt77ÞUp‚<ÚôO{4 tÚK£Æ¾¾µ/`WÚýÿne܈„êkÇ &²jy®“èñ@]’q󭃫»>Fðâ™I4¤9Æ‘F@ƒsâ™—A¨—~OïÐ/uܳî°lAí^òo^¬¿ æî:Kõ#ÔØøªó¸ƒšÂ¶ ˜:¨ÕÙf5ŠÛíq‡kÚ.CÝ-)³âJO¨¯*ïÒþXgØ]d•â¡)Ëÿ¼Ý¡,hþäS=C -ßoΩ‰÷Ak‡­RPâh«iz |«Úßq ¯C'ˆQ {BW[Aǽã:ÐS™üæ‚üøy8:}»+ú Î^;IsãbÜ0äå0mä#NÌ?7‹5`ìQõQ?ßA˜è(77®õ‚é@ÃMb˜M7t‘\T…ù¨Ê•µmrXŒ<2&Û®K /ü aUn4÷Ô\¬O=øÃ}„¶~çg¿`+F‚°v*çY$Ò=µgâæM$ÙëãèÔdÌ]-'ÚˆÂêl·ÆI¤š[w ÒEÚ4•î‡CÇ‘ASéÒY-dì?éŸÍµ‰{¶}T޳"KL¦|y‚l½”\7¨÷á> ™oñÀP&i€rEt59½ƒÇgiŽí»…<…-ú±älÈgñ2ÉþúïŒ ¾FއNk²Œº=Bá)˜HÖDÑ´x&•Z¿F°©éI‡’œZSD_Åñh5çTgða”vµo{è²w¢ššQ.ãZM©"Êso}ŠìAH¯>¸!ßƒŠ¹{K¯¢²¼OUãoTY`½üù²ª}j"·®@õGOm,”QSå«uFa jS ÚoH Î åñ!qBÔ‹ä ê²@}Üšî¬C†½Ò‚h(QÂÁ³3Fšº¶“ö6xÚí‚jî5R4ν²P'6…¦t!îýhæ}X…¬Ð|oºkÛZ45óý!³A«Ì©ò?bñÜ—õWþ\…¥à¸ç¹qP°ÇPòn¤;=½þÆ®7LÏl“@¬m«½†u%¼‰´ÿQac÷t¨þßzvwàµÿ[oëüM!Þ;l/ @þ0yš—tÐårPèEmÐpîaýñïÁÁOWˆûš\—žh)g£Û¨2!½&ZÂRÞ²øó•­Êþë_*ç5Èws-ÅR»ä<»Nó&kt×Ϻ}¾Yþ›9Ñ«ßZg!gOÝçÍ=¬»¾e·Íþ~·¯mE{}cý¯jåMêýof $íôåB¶”FS´šè®ºÉ°YüøÇ1Ê'ù„T‡¡ˆ¯"ôjg(”TjUpeÞ€²Â6ù=ŠåP^kwËT­*‡üÇ&¡²Ö\a¬‘ªy3¸¯«w@mÞ¢¹‚º(4˜0œy±° MžÑÁ¥±kÐ<–sî.´>¯lð¯Î‚v§&Iê[ï Ó§ówB| t—]³àoƒ9{´¢} 7>Ôbï}æ}eÀeƒ[þôTC0<^µ¿~ƈµnª{¯Ã„"ãš ˜*ÜNÞ‚ßÕ[tâç½`žÿý8õãHXäÐJPb¨‚%ɉ)âHXÑ;`>ÖŸ kVkÓr°á(ßuþ$lÝ;â­uxv2Ê9ûòsÅÍ­»ï…hY;ŸE·v©ì²3H––¬IД܉oŸ #uÑü©@?¤{S-Õ– ….ïÖcÌéÄR×Õ÷4¸§i_³ò$²„< êAd+ÓUT Â}næá5 BÈ‘º‡*¸¹ã;k ¿BžP^¥`ò>ä«8xáÀ¤ ŠùiÜBCurÏ{®¢¨¯%ÿ³E¿=ó±]$%ïí‰nÕ¸‹RÉWiåÜ­Q–^í¡ª<Êýòî¥aÐEf‹{A¨TõCQióÒÑ‹å¨RÈÕã'ªy=õ8±ŠêQ…ÉSI*¨ùè[QÎ+~Ô¶?©Ó1ù uu'Þn–ÄSgXs8 ¤Q¿ò˜Öo•U4X¾O1vÇ E—íñtÂùB&q64u2¾“‹¦Ý£Žï9úðLæ¹_¤ŸÑ¢h^T)‡­vŠÇC2‰`¥ðÅYñËŒÿx 9%9 ‰C#ôU'ÿqþrbµµ”E•Oüã”üå'„ 1rVž`ü7ÿòþòþòþr ¾2©™_JïÂø£÷¡ªY½R…°jœu;–VZ¡¶áÎ]£C— ^,+PÒ?êã¯ñ÷ Ãÿx$ÿ/¤Ö¬Ü$w©ðßuÕ+w›³ÿÍ¿Ç1¼Ù±øj¢:¥¢8þqLêûOŠv{^Nî5©þñSš›xK>íöÊ¿œ”ÖA¡<ŽÉœ„öŽ6ÊY{-èÜ×îÐá Ðõ¶ö…_Rô\?êÐCS´›_¹ m9@߃?]¿(ìvsâ'I¡–k0´wïy+o?0“7õЇÑ*EöP×4Ÿ´šT’¡ƒ©‡4ç_YÀïGTq©NÃ<‰ÅOEÒ+°H”Ä£¯ù–¤yrMÏÁJ¬¡|!„õÓÕÆç}‡-¾(W{2$P |L^„Ó£É_ ‘85˜;Ó†ICRÞ2¨: y#s|š7Rêt¸è#Í&{ßyÒA¤¯+®Ü˸ˆŒÓ]÷’3p‚„j×€²(˜k& Ê"Û ‹…Pd_¤]|e'‚?­Ün@®S‹¯Úë‚ð ãÃ{÷ŸD ß"q¬ÏïyoÎlg“ ÿZ†ßµÂ«xh¿!]ŠŽ5 ‡œ »¢w¡’•Ð Å-ÏŸ®f¸Œ’ôGL8ûáÑO}x…GPÚ®Sj`ÁeM”R4’¢\ØÅ¶®V;”§¼P7o¡žA¡†Šé¬œ³µº¨,xhäÎYT™g±hTÚÕÕ »ß'tPýžÏ'Ïs¶¨ Ü–d¨M’À›A¡†:#¼íœÏQ/уDÑIõq« ¸ý(<¦>Ii»Ûªóƒ+tÑÈå§Å-ãv16Üõ—žÍñ¡\“ÿ/äï÷Ïû{]VÇá ñÄô\”ÿßüÿ=ÿïïüåšü½ï”\Ü:0Õþ?ŽÉ;ŠkPñqú¢_ËÿïþðWïuÛtt…ó1ÕÍ£V OÎÙ#ÐN‘lýÍ•áŸ>»µ^ˆÈ8 ½- Dá Cð+;âÓ…{öпüôX ñ: 6X´¼¢ï†á\Ö¤vÒÿÓåÚÁÿŸÁ”#çZÔ‘ÿé’ºìîâ(Â"£õ‡wñÿÓå9SóSÚ°fïyyë0'lØ«Q>n˜…-{Ç?ÚµM°sW¦ÊVG f¢KG¢Ë¦M3 /‘D»‰Í‰ÓÉžxÍÚ÷(#%M[BdÐ(R—<*õAºgß™œ_C†tO[â&d\{|h¥÷dœ™½¦A…,NHÂâê‘Í·Ïw•‹÷I—}š0FŽËÄk!ä/û®œ»²¶ò8‘÷”g{ _Àüó!MØì·ë4žG¡L£'bQdëÃL6EŠ›Z(Ôú£¤É³ÏòQêþ!•ùùÛ(3º66.‡rùσEøP~&ù€êåvT^m™#G¥IKcSæT)e9)ðÕ24¹=žFõ»”mF/Ôô5*-0Ûío7Y«¸ºëP×qi{^ûžz*G¨Åÿõ' uJ4ÐP˜ìt¦Â0ù(½ºQ‚ƤJwú-Ñ$©˜fç =šy´‡³Øû£¹{úaø5´üÈwmŸS5žÔIvN̆¡…‰gZo¡í¹îeêcIzt¥í‰0¤6 8ä.‰—ãõ %!®m#kxç!ÄQݹí ±òÎû ÃSàýí+©SV3á{)„mì „Rå)é„A(±ýb‹­-Dû* *t^‚T3'·#¦Sûü¬µ'”ÈMæQ°\‡òH¥q6g/ø¶Ý-v¿Çª´3=ùÓ¡z>Œ‹Õ“j-?>*Ô…º¯XÚð êO0 ? ukº/êCm[ÂÕRz¨é’º¹ždÕӳʗ)ž@õ-ÍÏÏk¡ÚsR dõTÏy›R<1š÷Oõ3% –Ëö;ƒëÔíᕱ¢Ðú›e!¿OQ@#ËêOaÅÃÐ I}EZšßbbúpZâ|S{¡5•*Ñ•ÚÿÜÿ²úÚßäŸl‚޾£¾ÃÇ¡ëL`ïÖNè”ÜèÖ?ìË—Þ ©A×÷Ñ~Ì0 4+nœƒÃæ꤉0ü±láŽ$1Œ¾”ùøÀ¤Æã)h¢GaJøÖ–Ë<𛉤OŒWæJÎfÎõoÁ‡ ä“îÄð§±FÃO/VÄ´®Ü] …µ¦à?æÂfá÷]µpØéQÙz'ƒ„¾‰ú·ËµX‘¶™ì3’òó0xµ"¹ñ´‹`ƒR6Ý7PPîAš«/ü²¬@ú£ºe‹O"‘‘çBs¨N82õ‰—<ÕBæ6ÅÛìO(‘âk£ÎûWÈî²&'|÷  Êi<:hœ%WåŸÜCîŠ@f-ž‹ÈsbÞ@õ¤ ò¶çÔ©Î_GþÊVúqi,e³¤v…v¶¶AQæ<£¶vjç')pÌÞB‰·\o<ê$’EæÒAŽ\Qy(+xáÞ«ó´(w…,ú“BžèMxíßnŒàäB@à̇ŠAŸ=§óûQù` ­ðGTé»#Óµ‡ÕrÙR}£ú‹[æÄ©gQóüG´þèÚÊ4bJ¼¨+ÛtΙé(ž"¢uúéõ õíçwäÑàLÇçÚÛ2hh)æ?&£ŒF×Sì™Õnâéð‘džý”hüScÅù‹š³q$4 @³ Ãís yh®QQßKC‹–dDþb'ÙÑjÆ–L«p­‰œeÌŽÕÀü½Ã4„·-ë;Çw>J/¦wŸ…1}Ágï½à}ËP]rlD _•S„Š%%’öUtêaÌ·J– ‚e:~ŒùC ÎêÉêûŸ!P|XM-1B>ÙƒwÚkˆœºÃü”bwñAÉ.z¸“] DŠqñÃŽ!uàuri’:¤³*§¾ïÜ‚ o÷Ug_dE&&u’gÜnÁ¯šˆ,&Ù-¹5ÑÈv­£Å®` ÷±üÎ|±ÛßÔÎñLæ¤"·Ã… 4WF‡“ ¢ˆ|w´ÕÏ¡ÀhNÓ§f =çi¬…"£ù·mªµQ\ÇîeGI]¿£_ªPêVŸMžÊŒÈ–ާF¹O ÷=zP~˜zfÊz s¦ Òb‡å~BT©t=ÿäô­]樿lCõÈ”ß:)”¨ù¸$ΟGµçß=ø1‹ºÓRvLñ”õu¤!Ô¯/<š˜‚†$’6ÇÇ¢ÐÈäê6+ž®:6ëZ¢„&Vqî: Th¶?ÿGÂ4§P§çCKáÚ¦g)Ÿñì󱎙"JX3s>nrì×Ìâ!#ˆŸ €ÏRù¢ÞùÆ ‹­Å{sBÜS™L:ÛˆílºZ»±¢/n,í…÷ɹ\¥K!"íF ³„JZ\´;Ë¡¼¥U¿žåìúÛhrîI#Hå¾0Õr[rÜ ò‚§Ÿ@q01‰Úev(~Ä2qß¾Ý\-¬o„ï•û^kFAµû¨½Ì¦(ÔLþ|&üa×ÏLݧ´én@]SáÑé¨{JÙæ¿÷ÔêgÐ\΀Ó„p4Ÿj§WûÙAÕÆSæt¶!¨ÚƆÔYE¨v-êÜÏÜ 5ü¿÷Ô\q‡šTþ¼i¨McÔ:ä õŒ’äòBÞÐþXõYÛyh:w\ˆ½tš/‰\tÔj„–«óÉNÌÐê:•’½Ðf)ðê‹!´Ûn¾½'`±oÔÐXºØÏ x÷C÷Oñb»ðƒaÉ}£ú ü* c=O[±›y<ð­ÁŒZñ'_aØS()fÌFMD•åÆ`ü\֑환láHý®63Ñ;ìÝòl0çaóä}ñ X098ãýþ8ºÚ¨õÀò'eÿÏÂâ°¦c;# ›<Žñ‚|b°s„ÂmþP º[¤[ÖƒDmX¼Þ‡$Yo}&l(ìGacòÌ2Rêe×QÈÔ#õ‰ù™ìx¤{ûëSÆ2ÔÿªÉ˜}‚LDÊcÄ,ù²À‡gýȰ œÒŠ{†ïV¼5Áý÷S(®íæ,NáÉqYž(侟¼§TNÍFŽú"ïC§?±ŸÈ¯u~_:ë Z°)ÂEjrU8‰"©¢I)(Ö«slo:JD|rº=rê^ëPäA©m£(+”Y}f¤²Â‚rJ NIGñÄÇ1“›…Ú'ÞZV BE¯Kq®¨L9ʓѹ…*õ#Œ?¶Q-}q«ÕÃWŸ¯X æ³ã/wXPûþÊŸ4OBÔõõšøæ§,ãÚ=_¸¡þËöCËhàqÈM† }¨¸bÐèÚ —téýxúéŽå¥¹j4γ_SÊC“õ¥9î(4;ssöe-ži—¾µç©Z¸wûÕƒVRÄñÏãj¡7a®Erÿõ9ÈÛ+œžu>°ç*;µñ/–n«ëÂ{µüdÞ½ ì{OvšÞ¥¾æïžO†puy±2>qÓ»|‚mÈ#-zH!ð|Ž|áÑU<ü\.óA2„”œUáa€È0ñõs¥<û=¼¤x]’O°T8ž‚t ‘BMÈì¿@ ÑÙ:Ü£,«}ýý®¹Ó>+ÈQ¤J6Ñkœ ®òÛË 1Z¾/ûä0·Þ>qâdÅÈÇA¶ÐõóÁ—‰ «U»©øêoÈòÍÞ“& Y I¹?Sí «xK·±«²•º»%ÖBvÝV«/q%äTi†&'øB>CɳP$V,^É %ëKW ,Òm걇û®¿U³oXTBÅ»µÜ :;¨4ÉçðXµ†*Ï â÷AMç:í‡[bP¯’ÙtíI4VOËŒU­@³F. óƒÐ2`¢ÊV¿m‰¦—†‡e #ôÎDãíèz«ª»Uæ=ÓÂ<¡4²ðóý´À}õ7Ðñ—àõþÏ0xÒØ¤ó# Ë7_ûü§F-ºî¬“Ãø›$K[5˜¢œm“Úíoôu kù|0WCùÇï¹ ,”Khó÷ÁŸ¶kËœæ°<:Ç.þlVÇDqÖæÁú`~VJ¦!lŽñ+½Üz;^ rw Ç$Šóé‘HF,@°Ê I˜Vf©‹ ÙÉ«ëqËcHñÕ/ã|Ïg¤¾l9®¨Íˆt%$&ÒÈðð“„íF>2n1–Ä=O,¥{”T‘EñStÓœ ²™4°7ö!û2W‚³ø6rpé€M5rŸ~c%ï ‚<Š5™…ñTÈw)—25ø& 4=“ÖiF¡»òFeó(ÒXNÉðìŠ+}ÄKañ(©*7ÏOö¥\ÝHšo‡£ÌÐÍÁ 7”+»¬×–ñH,7B.3¢â‰‡‚ºº¨´e:bsÐ Ušï·³\DµO\Wß±º zÔ¼D¸r0j>ô=?5…ÚVu׆ú™P÷Ø CGW<ž:–ÌE,¢„ú™×鯑٢Á¯³Î¯ÍшpÎ2ï*žö·!çD¦ç?C IÑ´,ŒóÑžj<öK¼áÃ9´Hr_ÏE«…G$Îuç`™ª”àe2ä7Œ¸'0B?+exù¤Übf^ƒ/"Áw®Ôwâzºlå®^r^¿¿µâ±œY5z>ðþiÔ› Ú—ÑâC3®tBÍ |^M…PU–ˆÉLRˆöó».’ 2~®\üû&äpŒÐÕßžO û –•×¼ •Oo¡bq¯ê¾Úuø®èÝQàü ªÖ%ÓMïšCÍ-Zÿà+{¡vÊ4¸SïÔ]´*Z6ºu¬•ìëzyPÓ1år…Úª;÷–xÓ>…ªiIvc¡)¨ºó /2fªy¼%ƒªåySÑXK¨N^ˆR=\52ã U× öÄñiY·°š=|‰w A“1uâR4V>ùM#š¥Ä‹¾Lœ–fIo*u¡U¾.žÈ>ÚxÍ¢ËúL¡]ˆa@ôtØÏe-VJBg· ÇâÂtˆ]¸W@½E‚Ëvi[ðËÂ)—ÿ•ô—›Š|¼¸ ƒ^}¶ûÈûaøä›³UïÊ`”ƒ·ÝˆƒÆ÷YXUÂä­/·oq§ÃŒØÚéG§WaNÌÎÂ0ôÌïüj嬣€?¤ûÈdö²úñ´³ßaµJ,åKþkØð£M"$ª€íû‡ã‡s¯ A¥ ×çËHtiK"#™ IvŒgÍ켨m:RÔl.Cj“[!íTOŽàj^ÊÕ%d¸eúµúÍ5dâ~ð¤Ã ™‰h~®«|@V.î<‚6ÜëÒXq6{ ÷-G)}H÷FŽÐ‹cõñÈ-œ¯ðóÌ Ÿ™~\ƒ¼b|O›¯h!ßB±­žÚ$N ÷“½„BÎôz¼Ó"(¢ÕÐKÅœüw,–UPâáJ*‹e•»×Xò¥jÛÂÖ™Q¦ú£ÀŠß®žXN»ÏŽ'îÚo¡ˆ@U5MÓŠç™$G_“¡Ò=WðcT)Ö™˜ÌBµ”¦c_ú„P=Ù2„UÅ5Ó*›?ëG ögYý¨ûSÈŠ§ OåË[$êçwu|7ÌDƒ7ž‡:ÎþDÃ[\÷D) Ñèªê]-¾R<íÁNŸ^ù_>Ù`IrG“_ΉT4#ŠìG¢>’nzî9,„½ÿ:éâö*DvÁòÇZÏý‰G`5éiAú;"X¾N®Àó6‰.I²ZÂv}d?›DNR’S#íwSí¾ˆÄË~==UäH&àp_å±RDŽIŸ¼#‚ÔÊÕÚ¯‘Ž4+ BÁÉâ-Ók3dì‹êÀ=×­ÎÐ+ô#‹ð×+Ó¬jÈ&ÿÍFaº Ù;ôß©µµàu÷7tãÈ¥Ír7U‘ç°‡4Gò±VÑ7È @‘U’w• ÙŸg E‘‚ºàÏZ×Q\¦CшH%§=UÓ@)gö%ÅT”("›oJq2U½zùnÏ¢2YûÇ‹‚é¨Ò)ÉÛa„je¾!IgsQý½¬ÛD//jz½¾&—ëÚú‰jjCÒ¨ËÕ¥%Œ§DûÒ£P?èÐ\õ…}hPàë©w‚ —ކ2®ÂÓz‚-a/hÐø»»ÚˆÔš^ð²–›Å3¢î¦¹Rh!°ïõ%T´2㔘ù•K·yý&$ _ôM¦çâHw¤ãÛ. ÉõÏSOØý„ñîÔˆýýá£Ðuˆ}¦èâÍ3±ôí1å‰Ëðþå1²‡#Ö±¬Þtå „>ˆO#z ¡vN.Ë]¦ý\üXŸ¤dœ }Éܸ›íI¨OÃ'½£Ñe¬—á‹ éõ ‰Pá,Û`_Ñ•E¿Ÿô½~Uwï®Ü$ðr…@"±-¨½'§|Þßj7‰”vÒßAmüÃs½' ÆJ ËYbªM|ûªÊx êr«wÍñø¾s;ÅÿȨ"™âôùQ·›û;%ú ZtE¸d|ªËSƒcd¡¦ñäw¾KeP§Ì<ÒùËê#÷hl¶@ãïÀª¸h&¦g÷„²Ï‰ÅÜO •ôc”Þ«ÇÐ:é›QføÚfù…èFl¡Cäw’ìè «øb½ÝšGèòæ÷BïysúÀnø¹†Fã{ø ßßO¸[…5æ?Ý„a¦ æß†œ02¼ÏAžà Œµ“¹Ç;ÁäÁ ÚÂî0]ºøRÔ¹f›¬ÉÆÌ`>9¼ `Å£¾ŒÐJ™ÀR«d#œ…U•Íëò÷`}1#¥gà lµèut)S!è‹ù’M$Ür$ýÂÄõO=Êï$!éäiß4ÑKH¡½ïK$gRUÿn×·¹„´ö™‡ß#_Sñ ´2~;¤xùÎܺHïsYRN ™—˜#ÛÂ'ÊÕ†û¸Ï)Ðs‚•9H1ôf r}z¯;õjOxT„<½?ìÙ>ä#_8KòŒ› $\ã꿌B”©g¸³ñp×-ÆSãxd«¹2JÁ%D4ö¨]EÉ>"¡³o­Q*¤‰²+QešÖS¹øñØø(‰í"²±ˆf¿Y½ï× ¹IƒZ|Unè±E+«÷¿W€ßs÷oä ç@ŽkÚ‘ú÷< íø_®Šm ˜{3чÏÙ²5@${OÕ9„ÏUÍz Ñ@¸Õ¦+´zýB‹ŠÖ ç:p1–.@àk‘C¡aŽÈýƒç™t„¨:D”QÒCÄJh%ý1ˆ5ËÒå¼gIsë‡.I@ZÍÔþBú Ù>ü@ìú›KQü»Ë]ÿ¬»W»r4¶mØL 'Û[,9‹r^2%†ì…W$E_!Û¡9ò[ùÈ&ß“›p? ²šœh4މBV²ÁáézÈš®&ý<¶ëƒì.ÒqÊTýñ)ÇQ=ÈÑ z‘urÏxaé«JÈ<,rø+1Å…ö ØÞ‡RÊú™²Ìi(Ë\ºÖ¹ëoJÂËû¡â99)Á«e¨<”TDPŧ¨¸£¤5¦1›ñP—È|S1t OŠóÚ‚¦M¡ƒSö@ËkII†…Ðv’ô:‰±%t° 8TyïæÏ±Uk«¦è!}¢xÝ~¼:þ:œ¾úÒ.‰Dv’Ã@©^ïÑ0TñüXëck™«h°¿o ã"bŠ5Þ0U-ŸV- 3ïü×®œ_‚¹§F޼’ñ°àòÛ§6þ¸~öV1†e·„^² OX½4Ø¢è¤ëg…-r`óbŸ›˜4l¿ùbÆ¡‹¦û Î]E"Brc›É3HÜÃU™b<ˆddçI÷ø!…§r‘W(R³K¥˜t†!mwÖr±À:2o” ¬"c³x¬×C_ÜcyÖKõøda®U —gC6žâ¤”'YÈž÷à çÍD<Ðv-r¤˜l7N©Éø¨!Ë‹§Á‘ȧpó‰éˆ“dvPÉD!ý;Ëd¡ˆ"ï;ù®P“ øá%‡b””¹<õ¨¥.~þ–Iú eú6„§Q®Z‹üï*ìgè™tÔEEm©ƒ)oP™ÁÈýò2Té{ñÉ%DÕ¾-}ÎIXEõ˜;Ç\3PóÆ† ý= ÔVyÈž8j†ºd mÅÒQo‚]¾Xr õoö·ð|mÝÍ™yO:‡›Ð°Â[}é žfäRý3Æ^ViëhJuöý\ š•ÑÜ /ÑCóØ/óvïѲ¤ðœó<ü™zÆ:]Uy¾xç¾â$¤=›“ù¶£¿›÷ˆÅ•r¾@üó³ÛqÃCû~,Õ÷yÄZͪøë™C,Áªç8¡¼jò¤*‡È}ñ Âã 4Y«¹ùM1„zÕ ÚIÚõ·v‚ýÌrÓèÊŒM d_Sø|<¬ŠR´o91CYãɨs…¹ðuÌÔ,§ä=TJ›É’åº@Q-†>Ìê ç·ÎóƒP»gßÌlƒ2Ô¾N¨¦¤.ƒZ)ˆkùtª‡lÁÉ,ªÚ·ˆh]Šáû”ÛÜn(€ïÜÝï†ïO+Ý[ßÀ÷­³¦†e¶P•¹‘'›P­wåɽÊ-¨9oÃøtj{žó,C½—°gÒkJh|Or¬@ šêäÎ_p‚æ„Ï2hiðVh­™ÅeŸ+‰ -WîZGÅ8´Ï}ô·L†N³"&ÿ†uèÚÉÿ-,½4ÏWd<óàgãBJPô›"å2„A;ꪽ0Ôw¾–™Ü FR¶ø†Æ^fï  Õ”û~tÈôîLuç“"˜Un£¼éóFf‘F¼´°xÜáëÐá@X:'<{ïû2¬|®–#øâë–Fs>ÿ†->òÚ†$ "YùZü _O¿1ï>ŒÄæ;¿Å¨W‘ôúÁqq]$¯Ö“w`ŠC*;¦Ú-¤©ýÁæŒÒG\^ÊxŽŒöü;ú|þ¸G¬[±—PY¤ï¼Ë©ŒD6—„So&X‘}ðY¿WpµšÒäMG®ƒ9 ¼{‘{cÒç»r ò< T³šâC>•+óÏcuQ@S6î Ð<” nÇqc_ ômâõÁ#/™_ z?GñÅ—t”î—P2±¦eÆm¥ìü‘õ´ Ê\þÓp6û+~ܤ:æˆ'ø’ãe.F…äýØù!Åí&ûEP©Ü†Ý¦oUÞˆ™*ßÍÑO¿­½/FõoØv¨W5×úí8? Î —4¾w¨÷ðáÓÄ£¨/#í“ÔŽäeм hPËñz@Ù ƒé¯ºô~A#§_A÷IH𴾎‡‚ +Ul—¼|ƒ&ÊÙåoW ééÉrÎ% 4óôñ,Þ¢À3y¤¨¶Û—-¨^Ÿ û±Ûã|~frÃL°H„vd_¤.ümaIK÷²~‘@,uÊcËc_ JþâÎÙ¯añp¦ 0ÂìÚΨ–B¸þ`…ÐQ \ßáñï€à¾·}N™˜8ÎùT7÷{³.Ñÿ‚ŸKîz½„ˆÔÑ¥{P!°Qi’J’Æx -McϽãnÙ§\h¸ëoáq3W$!»éåÛ7ã{S~5%rre.*¾b‡ŸÄ]‡tȸ¢™³n ÙF§òW+!kiøWšŸ/d5l ÓЬOn4{!›¾[4“”²Õ?$û{ÄAöÐ9<º 9þ_—_>sƒÜ–MÒaÈØ2ºŠúÈe;ü>@©¤¬ÃD¬ ”zž¥Øõ·;GÝ.ž„ŠG/>D¶ T2fô†¥Ã÷Õísi‡¡†µ>š¬¡Îì[Ý¡÷ÛÐðcB4Ç”šÚ‡™ÈÓ@‹·íÐÆbéÎL; ímëÉß(®BçKC‡]ß û”á›8)øAnóýk+ôq–î÷²©ƒOíãœ)zêðî§n Œl„«¼€±Õ…JÂÀ_0阯z`Ïn³£K*²‚9û›|³¤¥° ­+ÐêšT.Ö UµÂò‰‡'æ*aUH~ªY> Ö÷eØ’ŒÂ¦@q®‡l›mljžF‚K³ùáÙ‘H8ê*swC‰K²³n •"騥ßý÷¿Â8©ËI‘jŽÛÏ–£i?~œm¢·D|wézq(2~º{™ñâõsQ·d÷!óv+5Í$²Qdϱq‡#{p–ÏÊÛx ­¶JιYÙ ŸÄƒ;¿ÿˆÐ_B>ÁÕ.…,ð'»¯û™…äB[ú‰GQÄw ë@m(Š<¤,Ø¬Ž’‡N 镟F)ë Ü55(óK^oêÛ”«/<ßl•ƒ Ž5“¨h|ïg·4*³7+\“²B•±©®/f"¨Vk=V6+ê±_7|hPÓõ‰6Ô–±üM9á…:3ߨÄD¤Pï‹vhž%ê[†f{ñø¡Ó`€ÿs4 h[¸è‰Fm¬¹_,;ÐXâ$ë]‘%4IWJ æA3ÃØø‡jßÐü ƒáɘ´ä=ýü7±5ü±ø£–¤ ¹£ÓšýMàcr‹Æ yzHò¸•«AñÌ/…8!Ö˜'9{Žb%#™Coµ@ÌÂ~#ZóVxÿîÒÌ› Bˆ„œ’èª|mš]ŠŽSƒÐø}ôkýâÖ÷ÂmHQÔUÓNsßÍuŽl×…y¡ˆÌNáø—b(³ð©—¨«€¯6C„ynð-ý˘h·-|÷aó)$„jÞˆÈîì*¨IØ|ëE'µ‚ÚúLöCPS¾> ózPíwùž,T_å3ä7ï—Y-r,‰á;ñUEgUøN|å|Íø~?°ó‚Þ(TÉ(ežU=á_¡zå¬×v2 Ôº¬T¸\‡z–¼A ~EhäÌÅ—µÐª_U±Í/èÚÖ %äjÀèàh½ù»é³œ´ÝÖ-ˆŒ»í¹Ô›Œì‚й÷‘µ¥1tå¶á?¯=¹ŸoDY¿ŸRï÷>‚~EAñû00`Ýá'C©uäŠ{`Ä…ü5m¥9Œi™xZ퇉k&ý3‰aêÇ8Ƀ¨wð;§;«Æ8æ¶XÍ轂…vƘBžæÝ|ÂKÿyó¬œ&à>ikËþ‘aÓ#èØg ØÁ‚¯‡=Ž#¡€rú¼߸¦Ž$+³jYN½H®þ _Š,)3ºõÕé#O¢|ý Ò­<>+°l‹Œ”¼¥Rµ‘ÈTùÙEÛš™Ë>0Ð5"ëÂë QâTd7»;c3t÷÷w¯5ä §O窛y r§Ž·_"œDiÞ8¡WÈ;U%w¸eòOír⹎‡”‡®îSø‰‡Iiöú¯ànScþÚ /j{,Þ%¯xyéG ”„ŽÛiÂh”QØä.ƒÇ^ñ¿4hÁãË“ŽÜšÜ¨pSíe½—*²ÑüiRŠJ)¾¬ê/Ž£Š÷¹4^T ¿@ôˆ Õ[%\²uP‹ËYTWM un…§Zy¢^»ÓÒ§ï¨þÁ‰ûQ4ËþÓÜ8€cWÿ¦Îªª¥‹ã—’)E ¥SÜ›é)iQ±±Ai¥ i$$¥KºI‘0õã}ðxÞµÎÚÿsfæüföÜ™=ÃZÜÑpʶ@ã¸ûÎÁ•¼hr<9zûÂK4=à$xW³Ͷ ÜÙæ[´(tÔšÑü7]í,'ZýصtI-ù¶||¨<‚‡º_}hcÆ#á’ûiÊHà“bä›;™Iy“P™:Í ‰9A}y”£RÆÙ<Q|´2ô=3ùL˜Á žÊò÷ÙzÀS5MËè«––ç/BÔZ¡[ëžWÇCPq“'¡‚XD¨l§¿@Èó€âÜÚþÍŒPœ 1Ó5ý7IoAâÃ:/b'FxO¥îx2êýù,Å·ß*KãÈj ¶Ô¾fÙFól=g{ ;w*²ë$ dß<ð4ë呵ø§ëç-ÿ²€ˆ$º2?ä’ç†ØCfã#Ív‚ûšÕS¶‡,‰i³¾)È:?É噌Í.c'­Ùï °} 93¯'Ú®@žpXÓ•2PÈ~þ±Wv~1ÁôJ+{ÿ*‘[‹oùûç™NCÅ&ÓŒÞP¹œIóK€ªß^?jG µ®WÂor@åÅtÉLÐä;%+N ÍiÏö•Y‘B«2펭¯CÛ‡¬KüU·¡ãIÓtÙ£è’I9S4üz¸Gíý;¡ïº÷jN×4 >f¢´ü#2Z6×Í`LW“0ɼã÷È÷¼'+€ÉÊ\¹ø·0#¯ãc½¶4⣫„9È¡‚în˜çO®ì“3…EFZz¯#øF|4²ÿ|ŸÝbO=?{ž‘.^`…•‰ð~6Û)øÍEö|rôôæý&È›¨~ˆ5’$Z3êàépÜReNÏ#&†b?ßÍ:œAÊjj÷;‘æVâÔØ÷b¤Tlæ5–Gúø‰—š‹È(a[þZò 2Mõö¸_„,³É²w î"û…±kïK ÓGà«/‹ î RÛ+¯…¼3†6 Þ⸓A‘Dñü)¸ØûIB!ù²ÒòOù¡ÈÙOXÿHÅXŒTäZQ‚{q"“ЋR‡©¼ü÷¢Lÿµ3ŸN |ëžÒ]¡nx@¼¢à€«7*Y&8É»Q…ß`Ëy…lT;øŽÕ›¶5ÊÞ@ÍxJ舫Ã'ßó2x¢Ž lùñOÏQ·½£çóP?Üçðyù4T2 긥ŠFš´Ï?Êi¢ñ±]Ϻ$ÛÑ$”v·iÅ[4Øw½.qÍMùwY»#WhpiRDËçt™9þxÄûKÁIæ\Xh0Ýu¿rTÙô•CJmöaSÏ&H¨›SÐ7õ¸ÈÂŽTó·Kq4aÀÄbÉuÇcâµ!f „%²U :–Èn»{+DÙ¿ûá½Ta?‚ÙOå4AX¥Ç<‡¾¼ðã~Lé1 Iš¯V´ ³<äTBþÈ·¿úГÕJÞ|#/‰ƒò¾úiZP)$júËd-.{ßsÛÚ~«1Ný|*Ô) ªSèsCÝÛ/]dPgôF4y÷-¨ùÒCéß Õí#!]sP5ó¦ZÐ=ªî¹RY½?U~ŸXŒÊÖÆIý¡‹®P]@Ú(¼£jd=@íC·Ck%¡žÊU²ìÄ'hHS=èœï M¹•—'†Ä ÙA…zDZÐâ²/'*­¶†~’T$Ц´c¸8å5´+Ûœî,BÇ%7×ÊüÐÙ­ûZ0™º QZEZC¯õ×'¡¿õ˜èSIšÐ×ÎO; #©S9¥Ô†0æ)-ksFÆå~ïüöê"Ln!íí‘[€i6Ã¥‡µðñø( ç]ƒY!+êl˜{Ùêzÿ2ÌŸÊ¿õxû6X¼È!Fß ßÞØWc½üÜžårmmÕä¹fNÁï+Î!ïÏ®£‚¾…-[e‘øüìÅžP$½þ6VêQ’Uæˆÿ^~€[wêû–;Ñ UDV ›J,Òê_4´P”EºdÞkìgÒáh|lañ"2éôÞqéÙ…,g®öVs+![Ó9¨ƒÈ¡o¦’t©¹>¯ð‘ÀG&ù)ʈ‘w””hÆùòñÞ¬'ôà.¿ª”7çp÷Ðˬ$å(ôôÎK½ºJÜ›ÁéÊÐŽb·9w:wA y+*E”üMù*àõ{”¡U<_Ù(w¬Òeë[yT¨¥ÜË‚4O ½Ó=Œ¸rJ=Û+•³÷*ø êùOKየ¢l"î`€š#üôûÏDmµŸß®†¼EÝ77„B,Ð`wß35564ôo˜SŸG£C ¯nü|ŠÆ$¥ËŸDÐ8óMÞë[hâæ™’3vM…nô×sy£é" ±ã…44«î÷;LÓæ 2±¿'ÅðЗ–½K^háGÚ܃–áü¡nTúx8w0Ô«yfF“ºÜØã #|©åÄ…vxùè|ià6ˆ¾Ô•­~מ‘ç|!­`€§ƒm²ÇÒžADwdÉU^x*›Â­ðþ&„õï²a)PÉ[¾CÔaKéœÍAtm*Òx|uå˜<#ɾrªb2MwŠU³@¢åI{úqRx}ž˜„Êß2ªøKT_ϯŷϗ®ÑCVçSvbZÈ6?¾Xp€²ó"ÚNêA¶g›1ûd•µ¹ô²ö°eD:އ̶ûbÅkq+³Ó1üÃ#Èj=Û¬ Yf?¯=Qª…¬8Þû¡µ8i@sGZ»r¶Íç=‘‚\I-…£s¼wÆày̲!šŒMÊ ’AñJûý^סô}ö]]#×µø¶¤vª<*ÎD¬ô™¶@åXzwÆc¨Ž¹üŒÚcjC¶b*ë„ú·{ÎþÈ‚&˜ÜHEþÏïI‹\ù­l«¬[–G -›òšä fè0Ž,¦›†ÎA²+Ÿ§×ÖŸÉþgoõ†Þ¥t Õý¿aàëOÙ‹æ0<¯Î÷®Æèv:Ì *Á¸ÁsâÚ ˜ ¼Ô|*7fOŒ¿%¦Â§/™ò\Î{anWY„Wò<|]¥t]ý £9‚›`©VîW §5|õ•{~@~†š[zMjÁÊˋėÊàר jÛ•cHXr;ò‰2´äs‰!ÉݾlÎá–—!EÅ·‘‚ÚL«…l)ŸýÐóLrE³Ø‘ïüHÇ0 Û<Ù‹ôþìËû…‘‘Ën|LÌ™ZG,R#K{Ìõ9Ó“È~ˆélâZ\äi±úÖxY·Ũ.«®¾°F æØ8f´E)“ g‚é Êô‡|6êFùÎ0ÎK-këÌܯ&¦PÉé6²Ë\A‘†ÆËÊT¨ú#‘Ѻ— Õ;ee{n”£æKÃÛ;ƒnòäþý¨ÃÉÒø¥u 3Î3å@ýÓe®FEkãq'Å…}C+hÄ«rn”%.”nÛ‹&‡·„M‰D£éS–IÖÑìçgf+¹ßÞ™ßj‡WÍJóZÛ!óxyõ ÖWÈ+êÉ9A ÅSLFÉã¥Pn–óýi"TÄk^åUƒªÇé¬Ë[¡½Ê „Ú¦’¹ ?j¨3²ÞµCj‡Ïžg=š 5˜<\.Cµ©‰K߃Påf)uͪÈí;uŸ…*åb|3U÷‰ºÿÕ°Õ"nª¿¾1gT7„Zž ÝÓ–PÛ»MË´Þ§Wë¥A“~GîùhÓ°õ$¦€©#qv>Ç¡UPÜû†d7´‘Ïï …vò ZûŽ]†Ž5]—Vx 3Œyå;;3tóîÉzÒqz)vzžц~Ÿ¦y®1ºž«Iïœ #—…)_ùž€1µgŸïí…ó†ÁDÉë_rLu0U²è3ïä3Ëý—:Z¬ásØøßwm0gøØ€7>æ)íléua‘„ÑPɱ¾IŠúÙÊÀÅ\•#)°²½/âå¹+ðë ‹ÛY«H00‰=Ì«ˆÄ$N¦çr'tËBnŠ’©N¥ÔÎ^EŠÐwõÞÎ"‘ZäÂ}¤‰wUñ¹Mƒtz<û³—"ýâîJdì¢øV>êŠÌó.;ìä:MÍÕ¦7Ý·6ÏÙYŒ\¦Ü:µÍ¸ƒjÜãÕÍíÈëW8,öÒ ù/׸‰>Ç]â~âÄLp÷UŽ{Œµ‰($¾"Eå€{UúÇR, ØAû´Ð-¦(A¼ó}yJÖÕ ¾_ÿÒC¶/Ò@9YGÉOѨðôøÂ"=`‰Íç”G¬)ûõÚ¦•/÷Ó1Lv£ªƒ©äò½ Tà oTGÍ/Li%ê¨íæU’<¥†ºKŸ_½—EƒËÞ»\а˜-.Èó*Ýxj½ßw0¤# qÙÅC}¡Ãh⪞gØú MYÃýé4Ñ´‘TÞíš=8ÍäÒ9…æ†VA³Ãjxˆ§çªÓéJ<´L®ÆrÖ-†OÞý¢ê–í> †k`Ú;ï»áÍçž53°—ô1Ä{H—ìug‚é³:BRðôóÑ„ï¯åáéŽ+Vòø âŠÜíîÐ6x*L7ÚÎ÷ÂþÜ+=œ¡Ö";®y7BÐ'NÂ!g=¢º²(Å!ñåœù]´™érô¤È0Ä\:>â“ ‰»€÷ؼ¶°^n)Œ‚ŒÒ±ë‚ï KåXÅØ]ÈêyÑÿº²7?%; Ùz”[— û´+ÅMžLÈz½•m_¾d1ˆÙû=Z[§ZXV¥0@f]Ê›šdÈœ'аr]Û^Ú×: ²Z­¶è1…ì»â¥]!Çð{ð6>Ƚ2â;列OæQ÷Æ ýŒa’\Š“Ï:x’@ißV¥±(×ýìQn¼±—B¶CeCš—’âo¨¾(WXx³ jÍ¢%üI þDOõp‘24~™ØysœšÑï}ú Zæfh‰ÝÔ íŽÝ-UÙDè ¿èZ,÷lïq}å…n9¸í«(½î]ÎÍ/˜`àTÉ3ñv~­<Ö #£íD5Ë/Â8ûÁ;2¾Œ0isÀìeõE˜®¿uøØ‚|ÊeÔ?Ts¤fÆ7¨àk»+ÉL™,dZE5œÕ…%&ë,ùWðÝMGîNê5øi¶í5»Å%Xq2´ Œ€_ :“%LnH`ýqàrú>$ ÉíéÜÇ$.§ÎOù௃œVºíH>™Þ˜õJ)Ï}Øwg.i£ú¨jîà¶+HX^DúËÎeÍ4G‘‘âØV6‘9dz;sþl›=²„WYÍBvYu×[ä?óÀ€š”Ç äIS2ó BÞF™¨K•È?ñç&é‡( sçÝŠøܳ8íê®Ñ"DGiŒPŒ«6I>ÔÆvž×ÆQJï¦û2Êtÿ)˜e­Cù>rÏ&ÊN< åîïRJ§HÂX¹PEF;ø†¦=ª‘‰¥F, z_ŒÍxÍÔL©<Óñ:Æ|vŠ/FÚÏÇEçÇQ7&!éþZüÓ?øÓsŸÒ4$/=ä Yh¸â”÷ûaU-ñBÎ_”W‚¼ÐÔàÁ±ãhöl2±í2¢tÝ.=$Œ6gƒ¬.áaSæ“ v0ŸÊ£f¨Ñ Y«ó©ÁžBòVÏk£ó:ðòì€ÚáE#ˆMeKÉ© ‚˜“mqbì¿!&=é~îªÄTÞ黿ÌÑéZBœç *iô¤¥Á„kÙÑtA8w¤dä,¼xìÂõS¿^…)Ô8}†L¦/;˜º-!o{êîùò\(&„¤RõCÙûŽ[S/f¡‚û÷÷Xu)¨b\µÝþpm¿Õ´7ó§'ÔÚGGîh\†Ú)¹ñ}Wu¡ö”hÁ'“P½jÐé²ÀUm•wØ•ÓËQŒ~kãÁÇ*zÿøC¨ uÚû:ª(åE.Ÿ‚ªršÜOu¡úÚeõ/ÛY &ë6MÕÖ›P§¨Ëp”o êG­ØF&¡qbük0¼ÿìú\y'4ÿ)}0-ó ·ë­R µé¡è‰%´5>xüM•Ä»Õ÷¹s@ç¡¡ƒ÷¹¡«k·U=­ôd˜ÿèe€~Z«Ù/žÀ+·ªëZÑ8aó„ÖÆèÉb“ú—àCÕGÝšSé0qÙ’'d.¦œÃ,ã”ÍaƧ3ûÆŒ|¦œyýåò<|Y s½³¾ÆoÅý`á1•ÇÍXz½jv½c'|ÿ=”Ï© Ëî¿ë#ˆXau¾XODÁŸêë3w‘È ~D]I²–k Ã-圻ƒFO"•c3R:w„¾^FêÕ#RÕ£¸mP~©òV Ò?~t°#T£QÂc†~=:4Vt8ö}k¿ÿbðû–š¸0¼ûXþMI³T-jÉÑô¥2ÍÑ4ÓÛqñùâÚ:òö½ÁK4OÑÞ5KЇœÈ笨w¡Åîü3ç®”¡Å¼´¡ç\2Lí«f•™Œ€´§%-=;ƒ Îd©Ý\c;Œ¹¥»"LŠzºÍ`:ä”]Ëøtõܱ@Æßð¥†e‡bÜAøšR3ï% ·IÆO™×Â’ù—’(ó%ø.oÌt­~2Ô'×&Ê€¸ÁïäsðËýääÔ3ðçç7Áo‘è*_Á² ‰.‰]‡0nqR‰^îÛäï¨CtÞ ¥Nó¶Ç2óHý+¼'}·5(ÒÚ H#½CŽÃ=2,8ˆè ú"ÓËyU¿ŸRÈò"€¦Ö[ÙY/ØX&7"'›Ü o®gÈsï ƒA\ò¾:Ã>ÿù[ìÀ•ó H[ˆ¥ŒâžAR õÌ;("J5ÕÕøE?†=Œ\‹o$DÜ}o£”ƯÝCC”(Ó¥48;ðå‡.tð©ùâ“Vù){Tºx÷+ã“RTÁš£wõòQm«ƒÞê‘aT¡¹?O§…š‰e'OãA;ê6ÅSv¨CTƒß(P×çÙв†êó‰ß§|׎s§ß)t¡aG`¥·´'µ°>_ Bã¡Å«´«îkãpòM:ç0š©ÔÎÜž CógeNçX´`#k'‰8…–ùe*³ºo`žëîûWöeùüRD9$ ~œ$J ñ5&µvjÒ+­É¦@tbøhc,¯éCÌ=K‹²Ý™“û2+¦Ô¢së 3Õ!ªéë+åk$~êîÝþ8+^øS{%Á+[©¨Š‘RÈ(æðÔq.…7*/ˆ±Ù@;;Å¢øO(;¸D-^ïÂ]ê\×âT˜Vâ~š¨¶:ší]xj¾ÝÑL¥™‚Úkî²Ý§ æ×Ç?_ê¶Cõ퇊.ìEPe"›êð¥*]#·zG5B%•¹»°G,T²}¦åˆ?•¾'»Ò¼ç¡JƒáT˜Åc¨Þz@ºnvj4}§²_Bm{Ñ­¤_ÖPeê†×nh¼þMä{¼÷bп. Íõ‚1Ù ¥`áë½[Ðü†®ˆ, ÚžDuóyA{Õ½ÌÑVèdM XÚ ]¾dÄšñÐsØç’ÊbèK¥¾È¯ ƒi^XÚí # iÙWÇÃèÀ)Ëæ@øà;2œaÜÒ$ÊK —`j›¢ 9'̰5"á]OsFôðE‹A*ù§$|Å nº?5°ÀAâ_Ú4KBµœQ‡à»óíÂolð³„wÕ×7 V÷£/t+ßs^1þHÄm¦ª×æ…$Œ; >–à%ƒ›N;$¿þZCÊ-·69&½ˆGj=O4y¸Oú¬ßZܢ糣îsE†ÏŽ×Ö‰ó|Ⲋ™´š\»ƒì!5êo/×"燨2¹%äîª`íX ÆS‘EüªÈç»÷P¢¯0î¼È%FvrLžÒ˜EAëƒê1(âhFbJEÓ†ÎÔ=Bñ'¹|oJØPÒAý v®JßJ˜5GÙÁ¸úôTÀÏת»¿¡böìû¸ÂoˆÇœOÄ00¡²Î¾­ƒ*QU™†ô¨#ª{uîÞB!‚Z”ß?uÔÎyaLùÐõΠ\¢Òn›‹b½‡Flé4â> hôþvõƒ õhl‘ËvõJ:kñº½@§¡÷^™hòµáÛó¨j4½aK>btͶEìðr% YLâ…¦Öh.WÐÚ妊æ­‘Ñq“xè©ÍϬe´Ø6œ2Oðƒ‰öY.ý`H}ÿjÚB‰ bes¦oìà(YÒøøÂóRÙn{Cxl}ü /õнý^žÒ÷º“Õ,@¸Æ‡¡ë£‘šé-y¢‚ÙŸ|’i, Bý±4„˜Ÿ´’)¦ÈmÍ‚\ <Cµ´¢JÛ å–ß”ìºà5ÛŒ5}däÚÿÒäqX‹o.r¦÷é!k(Åbϰ:dëñ¶ 좌‹ãmt^7|+d=¬Œî; ™6g)^AæóåäβrȬn(þè?Y,6“ç}!«DpÉ)²÷5æ½–ÙsÓ²š¡Ó’×¶¶ßyCÃ$õËà äkœcÜ.þ Þ’ì›Õ´ô…âUÎÜ#Û ô³2£ùÊZ|{utª‘*,÷Íç²¥@eTÊï;&ƒP½#û'û‹0¨iïžöºOu-ЧÒߟ„F£ÜÖáôÐÌrOëDß$´¼¨\Y9¸Úø¶ùØ'6Aûcžé¬g/ £)ô:‘ tEÄøD{AÏ‘]­]ú W6³ÆÁ°À~Ë_0zäm­ìøúp ŠÑ&~TØ2MK´ÝQWð^Û¿©tx6¸9Á/›àví«ðõž€ò²öÚx4JaPЄ%®ú>ç”áÛÌŸãñúÌð£ùö!Pè‡å Ú /*ôà—x€hˆÆ%øón¥›;ƒ‰ìü"é )D4Tïþ«µu§†âîp×KHåÜå9»)ù ç"®É"uKP£d¸n{~u ÄèÒþl­B†AGƒÛBÈä¿LŸÄCŽ,^Þf2|BÈö=ÁL 9柄ôÕ"Ï‘ËËÃ÷I'MðqKä/,_m4uF‘§{ŒqO5?·а±\ðò»ˆ¢3¾_PügÚZå¿¢”R,…AÜv”é¸Mü%ö5ÊMˆ_¸®ŠlD<Ç\¨t“L›®üªÔØåcJ‡jôåÇu8Q}Â3í·ÅCÔLâ×¥#û­.ä3È“¢öb,ãÂOaÔõb>A½¶î\v žÔ”@ƒ([©ó?4Ñ0­¸•~H^¸èeÈ¡qÌÇÏ1Fh’ß̳Ö‡¦“)"ò9h.êB<·¿–:Ó툖^fx%GÐ^¨Pꔽ҄W¢ûWD+ £”úDim0¼y°ãë¥óP$T-6µ;ÊôZ/¸ŠÅûঀävPù—ãâZüR=nh845î>qòcP«îwð¾ãK¨i`Y ø ÕŽ±ƒížPuP:ì|¤$Tº„Ê8•QC%…µ4©ITR†ä˜GèTà¨Mß] µ†©ž: uo_žPu€†&úøØÁDhjœ*ˆ<£ÍŽéäåqIÐb%™³ÅD Z5¬Tì'.B›™¦Á‘½ Ð~¨:¾]õ=Ú ]W±ñyd/ô8Ú¼ÜÖQ}mᤎžƒ“ u¾©0²7²‚tLF?ûYïÿs>Þ=«À¯¢o™Ôî¹Âayÿ®³§a†h5Ì…”>¡}F±@ _˜LH3ÎÒÂWVm~I¦˜4mþ¾¿ûC­N×€ïT¥&ƒ‡à§yE³ù9 XÉÊþ³8F ¿Mç\«Çþ !ïNè•'wØŸ.îyc’Fe4u×&!Y©™žT»R,= ªãQG*ÛV;ϳHË´Z:‹t¾šy|ÉBÈ M¦õ Í™TÈd†/ÞD–óa­ü‰ÍÈÖNß&l1‡‡.ùÏ'ßAî­Kõ‹¸ãÆä1CóäÛ%©q•2wÒ‘|ea"Cý/²J¸ç±1Ñ5 ¯”lßÊߎ¢ÀžBpBqCyq3ŸA”¤ï ä(îF镵™}÷PöK¡ÞÞ%”Ÿ&œ(¶CÅKŠB×:lµ8ò”¾BeÝóBoíÞ¢ª®IˆôÃ@TR½W•ÿ-jm3å¾ÌäŒÚIT5xRPÏ8ЏhÁ >YõK¬íß<Ý©¾æ QP`“ËYa4¦—R3ÞOƒÆ>16¯]cЄÄñUBš\'úø{WšüôdÕ%²DÓó7ÜÓÐt^~xAØÍN™>Ê),C³OÅs¦GÌÑÜízû!›E4ŸÒâPxXzÌ·k$@ªº•ïøJ/Ä|•¾œÄo Ï&nÔô³@ø¹›ôi|Õ¾Çx!ïðN›í:Þ á%—ÅîïR„ÐÞ6ò° !¹ü}´ÿ¶éÞòŸÎBî³Ç¦y#¯!ïdû Y(´/»-0• ÅGÜ)ʘå 4.ü¹‡”ÛpÆ„=… ÞgŸŠÇG¡òê@÷w>¨æq"¿~' jú÷l¯è†ºYFŠG Ðhû:âût74SöÌê‰¬ÅÍжmê|©&#´œ9°Ø;_‹RíÍ….¯Æ6®Ý•ÐÐåwýà4ô[wuÖÚÀ0Y«iåtŒÒœ.˜1̓’?ž<Š> •B{ÚB`*2Œ}(ø|<âÂ*¡»fïМj˘…¹‘¦r˜¯Èºäjc ‹ãu×á›í –ÕüP/ªšZ€e#@5 „ÕGOIà­½Î9¥f$"ºNÑ_Ë‹Ä mGˬZ‘tÌ•c壒+™åî±QÄ­5»d³öÿ@êSûc*q›à~Æþ¹ËH¿…<÷ ‘52Ü3þüM[™Du®—<ôCAné){>dKÌãfÍAŽ<â–j×bä‘vS~R…¼^ÚòÁ¦DÈŸ,æô‹|ˆóíXO=Ç=!eR Î(Ü0YúVî%Šz,\™PCñ’7\G”"Ñj¬2a7Â’~¡|û¢5x`¯ÓªÊTô?Ÿ|gqø7ª¨O>B¦}¨& †ûYJQƒß„?Weµ>gÌ¡¶tÏûÔÕ¾,ÃÌ‚úÖƒwEÕ(ÐP×x¸›ln“_äx!ÆmψÞ^AS5û¬‰ìX4ëhÍŒÃCO~>Þ{Ð-¯™GØW°à‘ˆÖë^FJxt±íTÐï´Õa8FP ¾JÌ–Žú®B¦Aß½ 'H¼O^£}Î âÈ©·ÕhEB4o‡Þ´)<EÚjrxÎúäœq<=þðç­ckãr÷)ëwyÑtæ5ãë%Ü-Ý¡i^B~-ð|j›Ã.ÁHü­j±´¶OÏ(ûuLFØ Þ<»ê^>yŠÏ[ôs‰A™QÚpdµ7¼ ”º"•×~Åø÷FAµç'vÒÛPgVÿÔqjiP s~j"”¿KÑó@µÖÛ›v¥  %ïxi%B¥sà1µ1m¨Üâøk@f*V•ì/J³# æP¹xfÞ-Pªî¥.žuì†ê›¯’RwM­ÅK±à'ÞP‡þ¦tÉÐÀµ—ûHl>4QIÚÏ8™B3Aƽ…Êšý´°c Z¾£¨R‚6–/U?äj Ý¡òFšÀièhV³ó®z]Ç-ƒçi¡ÇCºµ?+úVˆWMÿù®ñv•¯0¢åÓæ+3£¿RíÜ“áCèî-å M0!tnâ XÂä|žîb]%LÜ£ü’t>É‹Œ”f“Âìð“öÜOS07;¶ÇLpæ£_ÒïMÌE?ÇHuEø–ö+1#¶~Ì:Çñ^…¹¥Ù}GLáWþáúû»N"Á8ãã+“HL½ê{7í’üb\1z»ˆdD«¦Ò‘‚³4dúÙ¤ñhöÐqT<àsÏ|¶Q¤™x¨!•õÞêDÜ" ê¡;*Nù%¨þ"Oòê¡cº!g‹?jG]h¸•M†z« mñ­hXvÍÛk ‹•ê&[ ‘eL‡Æ©hT›é±{ù K \Üï‡ÆÏg¾ù?sEJ™«”l49ë/›a&}ÞƒFÅÖhª\¯Ãz7Mãg5‚hÍÈÏ«‘ÓW¡™ãO­ªY4+Mó¹Ã2ã•fMç« ¥p›ÄÙ7¯ &)‡ªCÀž=N§Ôn¸áb&[o¾#ƒ°”^åGVœ¶åk€Åe;Á¾D~ë.„œ,ÍÔäƒ`Uû÷B&áÉS¢^4…ðD1ìàc%qðOþ> ÏJI;.ZCLem¨ÍUHÜÔ]~û¤ž®{FÚn éÏ)«÷Ä@†c¢t§°-d¨tµHNC†Ôì¸ä3>ÈÐÈñ^J—„ îf¿t)Hoý~{EÒ»T4ÄÚ Cò´ï¨>dÔ:ÜoN“[‹“wo¸Y¦@ÖîŸû†â• +Oëø§³¿!ûñÎÀWª£œÄø|·/äî£RÓH&ÑÏ?GÚÌŸ—_GºBj΢KÈ ¢ôÈí¤26F~lÿÁ‚ÌõÙ,1ÈfY3ñ‹[ 9®¡Î¢(GÚ)ï¦8äõ,v VLBþˆg-"…~¸kBCÀ-÷îqR³Ñ‹Cá[ù”_G×Ö›T«n¦Q>(~Ï€O‘È%+g_4-ˆ¡Ì‘kû¾gꢼ‡.­ïY]T\‰¾㪂JÖDŠÏÂPEÅ{~±]Õ¢…é2QÃvoèɽš¨öØ¤Ñ u¶m?ühÁõ^1'FC KžE1hôÔãÛ[7a4‘l[U3@ÓïWÄY¯­r^É£EK'[ÂwãVâúÖ$)vµº‚kþܯ®€šÿçkû¯çº”bÕDÄØ¨KS¿¶þ{7k‘ì«oÒ»¦¨ X¡HŸÿf3C=”™ùj7[»Ã;¿‡)Õ^¨ôèæ),[[¯Ñ{Ýd‚š+HÐ 5ís¤7™¡ÆÅ}¾S*ª…;ø¬~*iñ“!KœPéäëÏ¿Å*‰OTång‡ŠÏ3Ôñ[·B¥¼ÑXr¨¬•lK#*Õ “êhs¨æ’}˜´“j4gx¼©žAm´Ã¢ÂÀ<Ô»k…2ºÔA£Öu*¹Ãöð^úRlñ‡Ð•½sxö)´$¼Œý’Z»ÃÖæših—wbÖᆎ”@»¢]‡ÏZõjªBϪ?¯¬ ŸÏ}Q…O†Må ü†‘cƘ\1šg†ž.øð´6L§B&víÍù~&?\°®àc‡év“™[Úᫎ€ç¹_0›Ò÷Ö©檮î½Ë0ïðÝÈ9•$ûÙRŸÀ7Óý<Û\À/Ê‹SÊѰ\,ÓïÜ¿¸*÷j8 ŸžÒ¯~ ‘ÏÔ×i$9raw]:àÕØ#Ò ;‘\õsøVλ¸õJgCªå‡NÕn*H[¥ç|éI4¹3ý#ÃH°Þ…w™À×òîÏqdÕ›ÝÊŒÈ^{ÞÌÿ9rž´4ÊéKGë¢áÏW× wòoyüÊ-¶ ¢¸ëöîGý¸GHÂü+ë( ïñt͉ø‰ûû~~åŒbµ¨Àã‚î«üEï[PʆèÊ¡cc(“ºqï å÷ è%Û¿DE¢Îa1 Dú»z×_õ£²‰»ç 5TµÏœ ü„ꩯäK¢õ­¿|ÔÝZ¤eƒzŒGB¦-”Ñà„x”Tßl,-_A#îÌþÇgÖÖbOåw£Ñ0›_óK4VôxÀü„CŽwJ&¡ñçD^;s4Q¾y@Ö•M¬rö” I¿7áŒušò ‡æ×H£©ÝÙ"†7ÓhÎò½>fŒÙÎGÜ)be]Dg=!ÆAêA°žÉ¹5¼ýaŸˆŒ©C˜¸Ýc[kfÝ“)y3žB·ççßh€`â´Cõ? ˆªÁi_ò5Ì´jÙ›š½×o^hÕ€ãÑíAíðÌÓs&zè:Äò½S¾ÿÍ2µIŸA*ÿ é Hg´NáóV†ôÓ5R!¿¼!mªãLG:¤),ìð‚4–Sü¯…ÞÂëŸáõ6ƒàõg^¡¬|*H{îzÆÆç=¤_YÐ<ÚbWXê%öCf¿m}¤Ð+ȺF*lôE²­O8Û+/AöϘ´+oÈ)¾$FÆ oÂ2Cض»CïÄ« ² (bÏuïôy¥ŒÚoÄT l˜DøéIx—ye¸ù¦áÚøE3ó‰PÍY³Së`?ÔŠÆ«*]zË<òßÐx]5Hè¡3¼Ÿ$;ÇõƒZSó¿I²ÖúWîDB{ËÄE³Çй—>e——=t}•’R“ZŸöo‰£Q„þ+i "0äO°’¡/…~Ú˜ÇÓ0¦ÔòǵHÆÙªŠ Â$#ÍÞŒã0ýÈ×ä÷¶ øtÕ•¿õf;|iJÙ¢nt ¾&U0±¾: ïÐG]ø Kº‘Dµ÷s໘ÔKâ…Ÿ»skŠß…ª#Ä×_¯XQ [Y$([ž/4FbrëR®HÊõŽ´tö’Ý?{_ùneø^ù!ÿ2R%“°ÝâÑFZÅ77ÙöÝFºÓt·{¬!}Û>¡Á@md<>k1ãð ™Oü®;Ê„l”=i£÷®!‡ÊˆfTBrO95ïnOFÞãÁ»Œ‘ÿÕ~«S¼¸«¶õcÑoÜ#ÝùË&…e˜íö%Æ} [,Þ±4¡¸w¥ÂÝ~”ô:AQˆ2Ü—+VòŸ¢¼Ìõ_lŸ[P±¶Ÿ„hzìŸÿïΙ|$*(d™äŽjT“”q¨ñŒ‚J€5î0[J¿‰:SîÂÚ†/QÿÛžë¿u?£W~zO2Ï<¦~ÑxM'sÅú§ð‡¥}æ´|tè]¡'Z©Ä†Kh£–î ¿$u´35+¿®I^¸ µ­íßx&Š¥ËÃd§”Ç“K'`ºèFY!|\r=.&ý f/1Ÿ+~sQÌá¦T0¿G+ âöMX˜'æøÄ] K?<ƒOì€\©:{4`Yõb]´Î4¬Þ}¼%QþÐríÙúò+‘µÛèŸ%GâünSÕØûHzOÛ^†î9’ßþtÏÓHácOÏû)?¿éÕrGšH©~¤³™ô©:Ù† âVÝeA¤È¤îÜ$”‡,g¯\\pB¶Õ<çû${‘ãÅÃmÉÛÛOú$ãh&îø¬ÎðíŽ(òõæ±ú+Æã.Žw ¹j¯p÷;mK—,]*î Øù÷NÕP+Ü‚b÷Êî›Ù‡ÛýÜýºõPò7©`"Êh¹?à­¼‡räDãþªû·òF"|“ ãºÏƒÊ†Ú¾æ%>¨zb4¢–…Õ "ð×…“¨Åé’e¯3ŠÚ÷õ4~ôç¡îBíÊt4.›EÃc-ÞO…£Ðp¬|ŒØE^W·]4zhòÒÌ)º¿qEx¿@c¾ ìkÖ¹Tý½'\Œäb,BãáY[™^4aä-0Lß&Š#ݯǡ‰Õ¬¸ÇZ\69}݉i¯|Èétl‚AH~yîÊcQMˆáÊÞ',²‘#‚<ú©º–ðãçN’»š35 eK!:g::”‹×Ö–×)˜áIȳãɯ °]&¡«”ƵÏtè½@O‡ÜYc9“¹Ê+^Ïæ2„ù5£ öÆ]Þé|èõ¾P-fR*®$‡’@Ú6G>‘ä2x½Òì”öÜRuÉ‚®õA 5Å]f•H~~VÚc‰’sJŽGû•@Š” ŸÝ•rHõ×-•Èø i¾Í¥½[r ½nÕ0§ñ>dD¾bµ¥¬ƒLÿ‹tFÈR.™yãf Ù{s9”þù½¯§:ñw!Ç€õ1L Þl=ÃÊêBù·.¨ !…·ž}Q#ª9P¬áå4e/.x$”ùÀ»Ë3½=z^PÉêÁùuøTs\’H]óW­áøX×u]¨¿$ðk[åhô¥¥{׿Þ¿O+Ì„ϾèÐÚsV¡‡wÚÃ\ò‡ S»ðhff>t 8;t܆^éOe2³Ðÿ€“DfІŒöI?'†¾¹ë©·`´r,tvÂÏjo‡ÀR˜Ð×./†©xžùc_ׯå%–_-S0[PßG9î_¹ìN6]†ù™R»åFX,¿¢•˜Ñçøâ1´¹8ÜzìÙ3´ P˜µIµÁcÏ…æOŽp¢ã¬ò…‚ºøðú_Ë{€Ç¯á^~–})1@ü‡@ Ь]$k×ɵkí=äŸ4Æ4ºô­k‘Ø¿Bü_!ñ¯üWHý+¤ÿ2ÿ Ù…Ü¿B~C‹‹m*ñM%±©$7•Ô¦’ÞT2›JvSÉmªM†Ä&Cb“!±ÉØdHl2$6› ‰M†Ä&Cb“!¹ÉÜdHn2$7’› ÉM†ä&Cr“!¹ÉÜdHm2¤6R› ©M†Ô&Cj“!µÉÚdHm2¤6Ò› éM†ô&Cz“!½ÉÞdHo2¤7Ò› éM†Ì&Cf“!³ÉÙdÈl2d62› ™M†Ì&Cf“!»ÉÝdÈn2d7²› ÙM†ì&Cv“!»ÉÝdÈm2ä6r› ¹M†Ü&Cn“!·ÉÛdÈm2ä6ò› ùM†ü&C~“!¿ÉßdÈo2ä7ò› ù$âbbÿIñÿ¤ÄRò?)õŸ”þOÊü'eÿ“rÿÉhÿÌ)ÄHOع;nh2'Ç»rðÿ›C\~³ÿÑâi‰¿´ä_Zê/-ý—–ùKËþ¥åþÒòÿi¹¿¸rqåþâÊýÅ•û‹+÷Wî/®Ü_\¹¿¸rqåÿâÊÿÅ•ÿ‹+ÿWþ/®ü_\ù¿¸òqåÿânö?©„Øæà-þ—–øKKþ¥¥Ö#éŸ>#ù·£íœÿíT’³Žžÿ–¸ähwf½Ä–JÐþ“¬DXÿàܺU¢Ûxðÿ|dÇ¢òŸü7UéBûºõÚ°<ë6dã>ÒgÝFo<ýyóÿöåFú«×ë÷)ùÒŽ­ßgläÏ|ôÿª)em”Ëq*ý¿ÍÝ(Ÿ§¼žž¿ñžBŽõô·ï+ú¶ž^²ñÞRÛÿÛ²÷—u¼ø¿-ßà¼S^·¼Šôÿ·•Ü*®õòUüêëåk6êQó}½|íF}ê×Ë×oÔ«¾u½|ÃFýq½|ãF=›RÖ˿ߨo3 ݺݨwó»ôÿÛ–ú·\6\·íh›[·íiôÿ¿mÛhW[”غÝh_»ióºÝhg•ǺÝhoGù:¿s£ÝžëüÎöw‰®ó»6üÐ5¾ÎïÞðGwä:¿{Ã/=&ëüž ÿôn]ç÷nø©·t߷᯾‹ëü¾ ¿õï]ç÷oø¯ÿÃ:`Ãëü ­ó7ü:D±ÎÚðïPñ:xÃÏà yëvÃßÃ!ëvÃïÃ_-þoG6ü?¢›¸n7úa$áǺÝèQ­u»Ñ/£Öaëv£Fó§ÖíF?1Ë­Ûþóð&¬~Ö׊k32­:°-“@ _«ƒïÚÐ\&˜ –µNaS!ØÓNsÍI\ß aG ÀkÅS„]ì‚€°{@´#„š‘ýÂÞµ¶ˆ²bwñµúJ%¤ê „bºl< Ï@ (¬ùmÿgáÀaª×çòµùôŸ¹t½ª[ÿlÔ÷Ÿ9ƒk=ÈÚ3’µ¹ƒ› ©c¸?;˜ýwÆâ»û[> [Aie˜Î‹ç¡à¥h‹£häYNënI ‡ÜK ‹~CK“d •™½yÖÛ¿gJþ{Ö]ªÓ‚ó‰±Íçùê²õ꽑>„KÂ-er(š‘Ù^'Õ£L Y3ë¯dhà^sŽŸ—ŒK¥ŽFt ÷™7Ã3û“ÈçªLÁ¢Ž I•öÈ P„f‚VvŠÒßðIsBñþGóÏ Òð@'}ñÓào(u5ŠÆ3âÊìO¦¤~‡²6÷ ÓêMPþkêî$ŸÓ¨øxóˆbå{TfÕ+ôp;Œp;3:à®=ª¾ôI|·„ê “KsüѨñ§×øÉÉhÔjýüN¦ už(k+K+ ž «Ã“ëx´_à]¢|<È=#Í—uEC_—ÌCÜ4h”zú†Qc37}ÝYô³Fwš7<Œhòõ€uøF šž´ÐO DÓÁ]5RÐLvLi±(Í÷ë½8„fs7éÃ3ÖÐ\Qíh@/šßòbtCÚK+$m’Ðb6ßþæàÔ²}yR0/ U®—UÚ5 d—JÝf@¦z«4An OùÁDÈ7õVsßù¹~c]Füì]×ýnB†2¨>·†çZÂ;9v.ÈÞ‰'ác¥†¢ÚvîGßf zX°=ÃÓ…_X>Ê÷vᵩ¯·ÐUð¾Kzzª›[§5AŸRÕöFnôwÖj=[ ƒyÖ¼üÊ›0ä$aÎÀy=Þä„Íî‚þÝm‘1'½ ‡¿zI9?ó?¾9gÖÄ'²@ëEÓ™5h“‹J¶Ñ¬ö®lO öJ蔟Ûxnø:¿yƒßƒ'ÐåWwBñ`'t-_sÞ´î4jÖ½6Ð3KÏÙ /ãåbÐ ^dQ-H©á®¤f«ýð™áÉsJ‚±*ê»VGa"ë’ÙRÒ˜š±ì`“,„™‹"™G“aþ°Ì¨F²9|;¢˜¹~Òe«Š¦V–†ES„ôaùÍ^£ÚްzaéRCX9ün°¢%&†µm¾}q21°E, ;Œ»bwj¨4žÚ÷¾äB† ‹Ñ<¾ 2œ±:‡,²i'”wA6íšÜë>Sô¼ör^½(Ä(‡Ü'rTý³qŸ¦L…âs:äÍ$Ž F¾õ*IŸÑ×(þN˜±¥… ™~¨ÌE¡ÈöŸxá51·Ôög§¦Äý…߸ŠïD¢$gti@E"J5=šÙ°aA™œG¹>Dx¨»æK{¼-Ê…'_zú²ôyÊ]×ÉQqÏ¢­Z,ѨO|%P‚*_ÉbȧMPõÄÉT¥6zT«kmÊD ¡­€3k¨ù,÷F$oj+ÓÓ7f .QzÀ¾ýZ¨7tÊù ¿êÇYìâòú€†‡®ð:u¡Q<ß•rA4Þèy8wæš¼ùH¦»]…¦Ê¦{¬Ol£iÛ­”£{GÐÌÒ‰½rE Íz¾½Èpw@s½LƒS_ yA‰QUž=ZpþrŒG‹+×N´wI£E;Õ@0O(o!œçÐÒ9!äÕì0’nίˆù5o9ÙOU½ƒJï´oÙ—¡P”êv~æä„7çøž>9l¢¶‡n…Ìl“‹¼Sþ=|RÉcR5-rE]!å˜Ù·Ã_ 1òÕÔöeGHHu*å»áI7†Rw•@Z¯êbLÚ_ýåû”ˆ¿Å ÅÝçÆÔH‰ \º¶š;å—¸&é& œï±HÝz(Ý8³+-“JSiÜiªK ¤OõöÙB†¿±”s¨Ñ`±J¹²  J«¢ÔÂÜÏCYØ«}aª× üƒÉecyA¨üqÜ^4ªé绊;@MÄÈøÂµ£[Ë–· ®KÃPé¢Ôÿ¾5=&žV[Ü] !ä½ÞR}gŒ¡ù¢¼§KØAhë§ñ*v€Î]ùϼ4] +akàJš ôX}˜4æ¸ }Ú¡Õ¹m}0àæìȦCŽE˜~­„[W¡·g0*öXÂv›ÆéÞùv? ‚‰¾®ã—¥`ŠCåìïï>ð¥\¾@ÂÈfž9dXŸƒ¹H9÷_ZðudG{Õu¾ŸŸ!W['‚ŸQ¤´?á§Ø¯¹ÀIuXnªx,a«¾¬CNóÀoÕ¨Š`{XÓ8r>°‹6BuvÞyzÂÖæn¶+ä]Htt2~¬Ô‰Ë_«J8Ö ©qÒSâûäHÎý¡ˆÅ¼)ÍzÏœ·cAêMr¦´”*ÜEë°b2ôb)b_ù¦aVn·k¨72}öæ5°*GÖ}rŒl$iÈžõä;ñòrúG¼"ž@ž›\ Œ4Ww©ûÍ!­`äQt)¾ý…Ș»Ã‡ÙPä²aìõQ\B‘+á‚(`Øe=t¥˜vîT0¡tåð«.a'<4y¶½c/Ê»¶™¸0¸£"ñ¹Ì~m<`³*¥‚*s<úl;QÕèÏ’CVª}å'"ÎÞF´(2knԺ«û¼nê¨J/_#{Œº[·¨-ÇñhðŽ„kô ÔŸ/p.3'èðbþÍá4r¿ù‹TT#ZªÐÄRQŒŠû$š Ü w¼Ä¦¦‡‰CÓ(®F3‰»BÌ5›hþ^ålÎ0šMŸùvV¿’À7Z!âˆßh´Y¡¿+iôšØåƒÕ ùdŒ}ׇÓÝqð®PBµ…äÆŒûè=i %ø¶ E×Y/ÈFãÜ~ò¢”=2÷i9”¡“ÍìäÖã¯é•a·eš«!Ô‚±Î²ï³˜jê@‘º³'Y?TßOIÇø3ÐPе(ì m·D˜Z„ê Ëâ6EÉIèÑc:˜i[½e|[ ÌÐo³_°JyxÉ’›Úa°q‡´“œž¯±ôEp>ƒ¾Ü›;F ~ëéþÌ÷CÿñíyÚcñChwÓXu ‚¶#WŠn¬ÏB{LžSê.EèØúýeŠõ t¾úµBì–]BÎÛ…ÆSÐõ†.‹ýÒ½Û¸SÊz©]/j`:cú¶mܘ³X:Õ^0 *G?ñ eÂB¹T¬Ž°dá*Û• Ë<+Þú-ð²úæsÂúøMùeÄûØ ¬]¼O}t(6mnž¨6º ;¯Gu¶¬Í8ÖÀëЙ$ ˜úRʯ„äÖŸ¹-ûn#¥H¶yýG3¤ÚÎòªFšÏÓY²v H·!10†{.NPìÏB†CÝÓ”W³‘qžóǺ&-2fË€½3²ÎÁZ‘Y;²ï}\ð_EŽ-§^•óÈ5¯=áŽ<íÃõçï"ïÅEÕÅLqä+Ïp2TºƒÎ5v^V*(äø]lð´'ŠD/òlœ B±OãVŸüÆýr®¼œÓñà“ òT6O” j„}µá(ýõÌÉ›Êrx(„‘‡Ò~å”Ä“ªQ~åéîžËIx¸°a—ýà*͹<·<„*aõj“÷PUdñÒÊ{=T»³«»â"ªOŸwöŽÉDMkÜÛ †Z«§ò*P§:ÿMË}Ô{ñË¡pCõONO]tžEƒu'©cÎShä¡B"…Æ-ɾ~^h.z3%å$šŠíjÜu6MËÒyŒ÷C3íÍ1×A4khcÿÔyͱ£m¸àß²èò‰‡ÆÐ‚eÊÖŠ¬-|Fžè³Ö¡E«FŽÓto÷\8w;xýÇT¾\Álù.Ô?µ^GÊdÃ˶¡à\{˜Rƒ6äÌ¥º ÙÅ»ê/s5A¦€XêJH?é*iVðÞ¶P¥Žœ„ä&IÿÔ`H¤¼Æ¢f \ª:3wú ñgë@ûžx–οúËûØN»izжS~põK|ØÝõökß ,Ò…C Ò Ê\Lg*ìm ôê&[u*”7 ‘*„’oŸç~·ù/&Ë(=óá…ÒC.ÆØ}JÉŽ]ì[Õ–£ø¡¬öJÃZ]/ÁÇ)È¿Ì "äÉ«‡?v@u¯aöCº¨eÜÈß-uÔ²Mî ¡P/y¡{K* >€Pßv<-4|™ª?Ÿò šCGW¸< ù©ý‚ê¶4tØÛ a}èR^ùîÚÐ=ßßÙ ½=Y·n‘f@ÿ×JåµÐk0tx±>·v>–\¡Z·ÏW*Ǿí±w£éÙÎÀïYƒ5ça2=Éõåãhø¢U^{gvfXX&â`Žrg$|¿?|=ÚÈTåßjâéæßXê÷¸|Ëú#,{AÇìmmXe)x*¿ú—GzÔÆàOÛ¹;’ƒ°AF]DY([ç÷HYÅ0#ÉÀÐ#«j$öqŽ~™ø I),Ý™^"YgÒG“H¤ø#v¦ë‹R‡¶l”œéFºK¯“–çqÏ g–¸&wd¸&gÉCÅLŽ1OÅ—ååaÒ|¾5#²ÓÑÍÝÏöFŽÚËD‚£Èé)y¯è!òúÛ,HÊk ¿á‹ÂC_QÐ÷ q/i: °UÞ¡]G±cF™ɸßêÕIŽ{PÒ¾—‘s?/J›ïfy©±yh}Yv¯F¹É¥Ç‚™ÞxØk«CüÙ0*M¹ÉLV©£ŠÑý÷>5¨Jt•oP-Öq7º¢†½e1×`j‰°½µ´Eío"¯‹¾¡nÚ·b)F<*7œ&þËžgö–djª¢Á›eOÑHò•{ah)Ÿ¸ý|Ϫ(š¨lëœçÊF“Ú/U‡¾_@S5ÎÂ}ûÑ4¿·ÍXò#šñËî9ܦ…f!Wï4ÝD³I“f­¹4?Ü»lø ÍCoY¼˜DjîØq¯, ÉËð+Éñ ¨9l”utk*9kš³ AÑG›w{ ¾-Hj/•äÍDÖyî‡Ü!¡üŽðžÄ'¢ê[d6L™Þ ÿi—e'öCjWÕÜs½é …Uë—9ë¡ZM<´«ÿ4hÿê7zzZ?Y¨§Cç|˜mbF7twÚE\¹™½*·6Dæß@_£8ck „ÎÌI{ ƒ’Ýy¾?…ç;kOù¡ÏZžÖéHt7»ŸáÍ&ûoí)ûnuÐ@k×­úÂzzhS³kNÎÙí®EǶ‰s¡ãgoT„t]K Óî…ÎNò89¯?ÐeôjÛòÇ5è¦ÿñºÆ…z4$IJCï;¡ë4ã0@'Aõ5(†‚ŸÇ+éü†«§ƒ]ê`ôXÄO"Œ¿?sÆôL>R 'o‡/³Ê£+ç`öBÖ¬Mƒ|E{Z=ÓIønÖ´¢­©‹íI™l…$ð3¦#ª§‘V®šéž+_wÉN½k…?³~þ L…þ‡k1l°Ý8tVèsÉšK»#É´ßA3×>$ËQ7Sðz‹~º7z›‘Ê„ŸsgßA¤QToŸFà›Íé†Ë¸»·wRQ@é'eä‚ý‘1JØ÷•Ádö/ PM¿¬A×3î¼$BöÇ|;(9î¼’³lmE®‹ÇªE‘ÇæVkÒµûÈ«8òR½Wù_D Ey£÷`DÛD) Ñ4:{Oê ÈÁ™—5(vÚÿPRÿ4J¼xYi¿÷zF)ÜŽÔ†¢Yé»§»e=Ð\nÉØ4þ!š§8pªï¸ ½þ¡Â·£hqA¥¡ÁþZ4_¿§šÙJà[Hv-uZºŽ°%ÐO#‰×ëWŽ Cõ‰¸¬G § œ†3 Zþ*+2¼r4²;£èž‡§C¶í²TQ¡.dD“Œñ}Y&ø1µkn;òðÖ†íÄ~f{HV¦•;Ä! ~ÖZAâÅÿ»7#!Ñå‰I8çHµûlžVõû¯þòÔGçñ~„"ñÉ­?©'œR¾š¢±¥yClΜ>1¤‰¿’?4A}ÊP²Ð¼Ÿ:.Jté³¶…Þü¥î¾›ü}…À»Œªß޳%óÐkCIS¨ó‰9‚¯«z³7Ñ<Êè ¥Ý¡28Ä…ã:T³Sñ{ÉC&Ýâ%3¨UÖŠsiƒ:WÇJùPZlåmM rV¶Ô3÷ ‰{ÈôVv´”M¶3hB{ôøÍ¸c[ÐÙ·@îÐ}º¦+lF¡×q@ÃD1ú=8öxÿ.ƒÁô»,1ð‘KºÙêî^øÔð-:†Æ = _êÃxIvã­Ý0©²Ëå‹`ª]³söö)˜~T?Çè޳ךSÒò`¾(µXf/|Ã×,Lû¨a‘T>Rcò,Ýš"½Ý~–Y+ùáãXÉ—¾´Ñh¿<èf¶áÕËøÏ£—a=°ÅËig6'èõd¬`çÒ3ÞÛ³†H,!úvŽIÊÉÕ„òJìvж±¡R¼º~VöüR3=žl›BÚÑq«SŽK¸{üêÀYczdØ-÷²7¬'vS?È@"ÖZóŒ«Èv&øË^2ä  1Úàÿ„\«DÉO'!/Ÿ‚„1ò ŸìÈ›ÒE¯ùÒœ…ÕÆÇž(@ÑÞÐÀ¢i”xy*ïc`|”J_"MÐå´Ú}‡”ùÍ·ÅVuåYz¦ô¯¢ÂèösrOTÒ÷óŠýª‹Ê¹[ys«ˆî¤,c—PÍ4–K¦­5v³zŒGžCÍþ=\>ÓoQû·áŸt=Ôµ¶9«~Ýõ¾ˆ~?¦„úˬâ­ûÑ ýaÝ>ËShø³hòÜÌ!4íp‰{&µ.ýœ;h’­¶BéÚ„¦R;îWIÑ4õÀ‡¤4cÿ´f?pÍ=2RÇEÑlìp ¨o±ÃÔ4µh~—r1[”)—.jP埆Æë;WeVO@uãmʨxÁ+šy‚˜à»\ãj\•¾m—p³¨6ä)ìcfÛµ¹\ÍlC3Ï!G¼ü̯@†+i›;Ÿ'¤îÿ¸x}í¼Ë‹‘IûY*×}O©…@!>Ú=+“U+ÎÎ4~ða@ˆ5ƒê<´¾y!ì!á¹ ¼Ÿùà\t[%üTûz*&×€âôiÄÿéóƒþ¥¬¶¬É 0ìb8Èý£Ûܱ·2 ws^»^´ºcÕgHþå[;i‚áÏ)‚o#z©ã'öÚt`¤ÍUÚeÊôÍ~€¬Õ¡yÛDª/Ë<¯`÷‡š«lÕ]£LPgœ® |êº §wï‚AMßEÄ ñ¢x-ƒæh¡pË탶ޘÔí 6èôœä2?}—þJU?=ßvÈxÎ;@ßwŸñ×-ä’òõÑá6®ä½Cð‰Ó€1b?Œßía…0nñG“Ùö+L´_rš†©c-Ÿu¸¿ÃôîËê~SÂÌ(ïæg_a^øúO“ •°â³‘s~œmi| K’¶Áštð³}¢IÊÔ©®†_¤ûÝ6Kfá÷xKÙýé·°¶u_´B·6-œX_Àö—Àû_øØ‘hü›†ÿM$¹˜ÏªÐ‰dòÎ|îGuB½çlôºRÅÙ´ ÍÿBÚ£üTßÛþànчÏezóþÜÄ=j7dt^ šFf¯Ð—?IƵn{,I,÷‹üâýˆ\ÔŒ¼7˜Ëp™‘äuä³¹Lö>2 ‚Žu¡PÊ;âWQTìàûÚ×oP|Tœ*¦W |¸¾ÇFÞ¥®^"—ôA™Àmfefi”QbI˜½ F;§8½IPñýÝL[VCTÞëqfÆA¡‡6³­â5ª‘ˆëÛFõÊþd*Ô ¦qò]EmÆ …뉨K^¿÷pk2êE=JÉ)`C}† ){ÑÀõU‹'c?æ÷Úòj§£Ñ2§Œäù?h²',H…ÓMbbE†ÇùДw’ÿYòšÆß÷âÚf{B'ÇûÐÌOߥéOš}â½½#ÄDàÛ™¼ð†T4¿g"ÿ8RÔëZ_h» ÷s(O…„jëÂæQ†(ÿ4²‹:i S/=±É'èo&šûš® 䆤ºŒfÀ{ÇÇÔÛ²=}—ÊäP±:¼£Ð3ÊŸ"è/8EˆR<¦Æ¾øA¦™ò«ÚX(¨q°ÕK€ª·Rª¿ >¸š[Ñ^õ„–'™¯]µ8 “õ ɱLèú¼`A×qzT5[ðCo%çèÚDè·ó}/ë¥ý[~¬tÕÉ„±·ZRô¾‰½%Êòº½YGÉÒlÿãÛá8§µmh“ÌmR~ m†|ßW¸í¡m« SæßB{橜?A_¡ÃFLb[:…´Ç¤sy¡3µ*¦s"º|û%Τ¸©½33× ½Ü¶¾ÑñÐOÊb'á ƒ‚¬±i‡k`øÕKÖ¥ øtýld(ÛŒÙýø¼G„”ˆ>NÁ”ý\´C¬L§ï_:?{æ4®»>S7€eÆË5¯)àGˆ–©7,= ™¬z˹{Š`õšû[ëhøÝÓùÁ°½Öº)˜ÁÖþÚMÕŠI$âzXö¸V‰ +—©‘Ô‰¦t=ù'’ï»2°ä[‚k¶«”iHõmcW–®Ò²Œ<Øø\€»¢K¢–é™^m±÷ì+fdøz$dò¸ 2ÕUgjé"KÉ6Ë~ gd+´×šJµÆ½ñ ¿CBO!g°·jK,r{vd8±÷Ýs‹kWõD> Ÿõ~}uä Û[+BÁþC)#k(|¢?êJx0Šúæ|=”øÅ3úÄÕ2¾ãþÍCS|Í(éÞ“ƒ| ¥YÂ;ç#{Q&nçCN¸ʪÊcÕ1”[""eYF…,ýÓ¦±Ñ¨8ìäuî5*;Qò¿õ‘Z±H³vUû›ío¢º¨Bµ™BjDÏ>ŠZG’ì/÷.¡YŸc[ª+êŽf™HWãÑÔ8·+j&h`Ìëxà¶?+šFã#û*f‰¢‰>ýZíM4ù¡ñsÿ\4½+xû„ä4ã¼Û¤VóÍÎ=ñܼ†æü²fhCðoÏHgêz¢9ûô•%.´p#’ªœ¹†V`| ÐÞùý-Ï1¿}{x‰¦mÕüv„*AíÊvN‚ÒÓí à!ä¼QÍôü¿åàZú~™À%ÿÝ×ïûkù†ü3mZü‹ée„?AWÃÔRA&ÄWë·áñŽŒÉÔ£ŒðÆgð»b*#$èü¡ñë'èðZAQºÆ_ýå2¦PW¾†‚ÛöœÙÆ%PÔçkÝËÅ«Õ _Ä'¾ì‘Hzƒ a\Dyb“À;©; Tï<y)ý_Ô:òÈo[”QRC8†0Ï šÜ!$”Ç´×ßAW"Åsç$Jè¹&\·túÎwä¹  DTº øÀЧMÏ•Ô0rZ8‰U F9›i ÂØ¬LN‡ ‡OŽ!×7arÔìkR‘|¹'¼î{SfN9ªÌuÀ\¤…„góIXØc#8¥ÉßÛD¨}ºañcwHqüàbM1ƒñˆ½Ì¥Ú?`ò)þ¤]f„/ _/­¯qÁ¬Öèʒ›ðUª ˆ¨m¾ûÕ\ kƒ%Ö“Dçʆa™ìÔ-–*yX¥wø¹'Ó~ë)g­]*‡µ" ø EØ´s¹ÿî^ì(ü&va'Fb^%*öà HRâ¹8¢ÎŽdÞ dDÄ2Haøû佑J×êÀ§²HãÃæÝlat¿3ΛPÜÅ=¯u*M+‘!Z{wˆK2i¸¥óœ YŽ©4n]|ŠlzkgœÚqïAëØ©,äÜ};>pÔ¹Öü&tŒMp›†…wøeämr:™,üzœRÓÁò(HeŬnÓB¿Ú –ìQ”ÈàÃú½—0>PH†û}ìó‹Äƒ“œ†ÛQª]ãuÑq*”17Úž×îÆC‹GNŸmòE¹W,nžB+þƒ r½¨Hp—“ì4 ÙŠe ÄX™=°FÕ\ÞGÞ;¡Úlǘι[¨qÌãVkjnÜÍ:ÙK‚Úõ£Cź٨ûF]ŸôÄk<ê.£Ö!w Xi¼ö­J£áóôó©QVhLó“¥­ªMäNøòëM£ÉQÚx¤#š^5X?%§†f»˜Wä#Ñì©>Cœã-4g·ì)#ø·ˆcFç—Ñ|çFåP”Zœ>é¢|`ZÔK²S"ðíÊcÏßhéž½$Þú ‰x(pÖp *ÆÚ3â¿C±#;o¿8äúM¿0=YîwÄŸ‡„Ì›Ú_ˆw݃ôïn'œ)Ôý¹Cû ùÏ‘ïËŽBbd\ð1wxÓý£ß áõLßN®?ÄWˆµ4¥‡‘£gðõÙ¿ú{…%øÔCÈÏz&Î?…S&¿¤cx ˜UnÐ{'Âu˜Ì$Eù¢yëžU(î>ÛJ¹ð–0Öý9Òrõo,êˆ$ýpó½íÐT ELvÊÛ!ÔP$>Ýòz¦ ŠŠ8v'ìDøË}²"JS…fw4¸¡üƒç¢ˆµ6T楟~_žHXgÖKä»^CÍž·/è*C­óÅm~ôP/Å6˜ì ÜmÉ=ÎCÓó+o1\…VÆõòð£Ð¡y‰ût)°†ÌS&C•QpÉGzè]Õ „R›—GßÀŒWk(ÚÂÇüË… 1zðÙîXáŒÝ =rxù Œ/iàIס52srøB9p[œÖ ¦$¨©Éí`ŽDÌù¾)|õ•{×Ú ߥ® 5Ø‹a:ia’?iT÷ªMrÃòcÇ»§°*é¼&|èüú%ñãëê.øó›þ¹Ï°)lazÒ·Ê[õ¦gR¿C"ãK‹öù/‘xœ¹ÊnQIcø]“5Öü~€›ý¤,˜y¹æ±ixÜÉßR"]A)Wýèîyà>ºÈrÎj{'dE&Ó³†3²%`q ,¾Ù~ -põ#Ç©o¯yä6ø.¥ò“ O«XÏŒÕȯ#T±˜ú/ªjätˆ£pwëË+;]½ïšî—¥{ªz%±p.ôÜ”Ž,P*ÇC½·ž ðßCùD¿éôÅñp`ºX§»*Mtsé=D•7\ã7·‰Põœµ¶ÿUjT°—lPE‰U[•Qëõ¶e¾-ê8Æ\:«ˆº«ºI‡/uáQ#Ñ×§–P´x˜ö~)*ÓêÜ!A£Ð÷-FóMh\#t‹õК\:öðfO'š¥=`¤ FÓÛò†¨ÔÑtócaã¯ÓhæIÑO#Jà\ÿ¢J‘cš+lžTóAóGÏÊt:ú‘Ìýçm ‚ߨOg¨+í…ª«¼Ó…¤¡Ìż¯J. |ù…™4|Û[šOaÔð^¨‚Úë@ö,ßwímÈ8¼Ï:R5Ô˜>wA’Éó†$-7}voBþY4tlòK-ˆÀ÷&Tí‘[çHކúЬF»sÐ|”j[0ËÚƒX<ÕgƒÎolo¸m¡[ƒb9кz v¥öz}’”×TîLC_ ŸïÝÈ<è3³½“Çm=_)þ|[T†î½Õ–³‰Yÿñ-þé†àOh»?ÕZmÐf=ˤv'Úž4ŠšöÄ@[{úC»«g¡kGÇ3ÚWéÕíæC'Ùäý¾ èü\VÀ,k ]kNž¯ß>†‹¡‡ø¡wm'ãœEô×× }Z©»ÍŽ=kZD:ë׺ﶈpÚ‰éô³ÇAÍ$ZÌS™èÚŽéÍJ#óR€{i²ŽD‡ßåÒDö’âBµuäxú¸»j™¹‚øçT!O5ëà ä 7v %fF~éÓšôy(ð^Úh¤?…Ô?ä;vEÂf⫽“Qìm“4y'J,ܺr³ç´ÉJÏ*E©u¯!”¡ˆU 6ÁC¯*£Ÿt?G9Ť…/õ(¿ýç×J#* e÷‡Dâ‘Wa7:«$}¾î·nFÕŽcßP-ÕuG9N5XÖ¸i%ï¡f©‰Aj‡Ì›%ÎüA]§>­~…@C-BRý7Ù,\f$€ÂÎóZ'’¡ÐŠë¼žÕ êÛ}§Sv‚Â|uG£Z?ç;ÀDàZ‰Cö™ø¤-(3œì™K™€ öö@#Ò?PÅ» öè4Tëp`$ 5™”ÇÕ¤ î"ß;ïiRøpq[9q64NìXšÓ‡–ûí\ËÎÐ~1¢Æ^k:›…}λiAwÔeí;ôsÐëlc]T ý—–;c6Ûa0®éQ|$Ktoæ·†O/Ë B-ߨ@{A &Ãø­ »¶U˜XEž+õ7U•/¯@{ꛪðçb·i§;¬g2¿Ë¿ç [rŠýT-¾HD¡¤Ýºö‰ŸÖEy’ñ#©1ñÝZÕ$—év«x~)Õ3ƒ¨J uØ÷#–ÝH»Õùù¡ËÜçáqœµé¿zÍÝÞ|€ŒÕÃBm¤¬È<|¶G×ÁÙ¸/¼Ÿá莨¨ö#×ÉÍr=?q_†‹Ž•þ%ä+Yý®Ðs†hœKw~¡°ˆ îd''Ц޵O¢„ó­¼ÖsÐh/³CÓË2Þ\·GÓåà7߈2ÑÌuê8Ö£YÏ Gäöiß>h²mÛ¢y$¥ßq&$Umx–Á’ uÅõ9«æ&PEmé£"IAÈ»ö”pxAþç8Ï}±o¾MšøAÊ ä\ÒµS¼O Ùscç¸nÀ»´–Ï•}2y³ì×Em‚înµ°YiAÂà¥Ì­™4Hð¥ŸÊ½ùŠó&Ù¼ 2øB†M>@=ÛgváV7hJ’?@êìm9Z›nV Ó§þê9Mèªs]Šö9=Ò!å¡÷Y‡C k!ôI¯Šê]?½mí¥&ýÐó°TÖäÓ èúýÈöÍð|ˆ¤ÌUò€¶²-£Û¥Í×"t;ÿñs\ɬ„Øqnù-´Ýv³^€ö£¦·­÷—@‡ó cl,t>–Ü8óVº’…¶(Wÿy'ömVz/ùôßÓ½ý Ý{Íø~ Å¹:z÷9~z>)E+óíØQ³79LÐ.¤:ÇôÃñ™ Š0­—óswÚ<ÌíÝQ K‡…Se×gO=ƒEF6kù¶ð“¶Î¼=AV˜c6?ÛjÃ/YæÛ&µÍðÇU7„‰w?¬½Íò!è%™aêˆìL4ÉÜÍBâs ¶qWœ”MÍž›™É~K)Eú ÅÖwq9¤–º¯äŠ÷‘¶ø ?õ²/î~Íü4ð;Ò/Øq·^ŽDÆìq[d/—zMº¬çeô• »v5GGœ6rðÆ“St!×îæÅ\é.ä9¸R^*мÆÊOì6‘¯Tçå›°§½ãk„BÂå´«¥4(Âpëz˜MŠ kȺ6It”Z“±_tr¯¥¢樚óFéâoçíJÛñ¾Æ‡ò¥(ûíÌok'”Oî¨Ôv6ÁÃ-ÖÔ·®ïÆ#Gô„X|»Q¥æqÐkeTåO¨^ßã‹jnD<ÒÝ ¨ÞT¨÷ž5¯YZ*t ¢¶ÎJKD¨ê žQM¥w@½%«ÅîbÔOøAVn…†’qGŒ£QÚÓ©ÕÉ14¡^»0ú’ §|?.ýÔR4µz5롪€¦‹b† Ü?ÑìÍn"Ñ4'{ÖŸ°@æa%›h¾bèýâ“/Z8´ËÉ EubkÔÓ+¾µ;.‰–ççnI®Ö€e˜Ë/wÂzßoNm…ô÷ ÃÞ“x'ØXoCÆ ÷اÁòÞ»=¦õ ûßïüÝžÀ[Æ ßSn.L·M]¸ ÁϪ¤ÖP^¥Éw{<ƒ¸“blqð&aò¢.µÇ?ïçZ ïüÕ_޶¢:ÛÈ£¢×ÕhÒ‚‚]¹Ë“ûå |â¦N®>ù¯\ŸËÐBÑ’dN‚¯ §h&ø³™½û.(Qÿ…Sš³Ç ¦Ðjïm¿4(xñéM¤Øœë¢÷}DïÝrÖ¸þÙÏgQ×@à$«Ò¬6Õ”ôY¦ÿö‡2ÿ¥çsG "§ÔE!̪nÐÚ¥[ÐCõïŸó¦¡¶&Rm—Шoôº£µûCø_rCsË}ÛëЖt}QÃS:ËŸ\i‚nþXJ+íèY“è#¾ÙäÛ½›0HôÎeÌz†Í»ùíßž†‘þ ’ãWaÔ¿s)[3Œ3³Ø\8çWf™ªÃ8 ú<ùpß |©¨ŠÓ¹Ð 3Ÿzú›zua^û{Ù«XXø-+ز?jÎÞa•…%+ÕòïO¨àgçÅ-ËòX±{jüî7ü¢ 8Jê ¿#i$ÎÂ:K}_ l&ñk¸‘©ÁÎÕ-%¿kSHläãȶ…¤Äcc*ïlPäŸÀR´_rgÿ¡‚TË‚\·XVsúµõW-Ü5ì±ÞÌΈô/žQ £ÌÑísÖÈ,?ÑÞËùYîM°w#{…-‰KrúÒÚÈ÷@žôû9fï"ßÁÜ×·xPzƒ?HU£Ðµ¿ïwŸ¡Èw «xþ‹(þ¼›ºïhnÌú\Ro=½ÚN1¡Lk W¼ý3”£%¹»%h† F-IÚ5N¨˜µÿˆé¡UT>Y^Å1ám½D.ª… ­ÃUÔP:È^.ŽšësÙ2­ô¨]zqˆ‘¼ u-vâ+åÂQïóÝÑç›Qߪo6•ê Ô¨Ë.Õ…£×QgO—çh좸ñ^šMŒvÿýˆ&cÖîyÊ éÙ™]9+—Ðt^2œæš¬Ñ>jåŽf•¿/Äì'ð-‚çe©*šG'|7o‚$Ì1™¬ª ¶ÑWUh¾*1“í¾P% -±—ëõ!_+ãdÁÂ5‚o+þƒ cÿìsË Ùß™£§’xz¼ ]Æ!Væõ !¯|sxˆ²â+öß,÷̆øä¤åì‚IHK’SÐÚ³yEoø“ bÁ÷ôun¨{Kg¬ršˆNÙMì ‡ÖO&‰ Þ8 ÜúÝ K«ñÇ»‡óÐýnçÊCîpè%yª©Fà\ã•k ôê©wÕ9Bìùó¹ÍÐU}®µ5Žü/ß:è"äŒg m¥S›×òŸýp VA›X'ýFY´~_v¯ð˜ÖIÉ>§ÑýÐ÷¸ÎÂsÚ«hsï?ª„N­ÕrÎ5è²ødãjÝáÏêÜ2… gqn']ù4ô¤z—ÃÄHlhöÁ³ñM¡ï1}–£Üsz}ô¼0®­AN¿T “<"¤1â ðE¿–W^`ÌîN¸Vju¾Z+4‚û¦ú)OÉÂÒCŸ²tæ÷°]zK+{?¬¦ºoºuÂB}&X›9ªöã§1lʼhƒíÜåÚ2Æ)$ª}ùð&S(’œ “xõÉô[SØ+‹Âêܽ·A2Hõ ‘G._i9‚7NÏ}Ã]Û#‚·Ÿ… ýS—¤ó7ˆ‘ÑbpÛ(ä62‹lN]08€¬Ä=’¥­{ãÚÌæ#Ü›šsH#9Ÿ{ÿž`Dî’6o¯äeû P0…|>Ç;wD²Q€üKCçq¬È kÞ7ŽÂ/¸mUƒQ4¡\¡©[ÅGºèƒñ€ö›h”\ìäÕ8g„ÒÊ)‘— QæOO(=®[;JŽòFÉ¡O7àa¯åëY¡¨4DݼÌóU,³y˜¾e ¶µ÷¼÷'E5ñg*"oP=ŠÞ4LöjQzËß%‡ÚT;šþÂwPçËÆ‘ksˆz¹ó ½ Qß­1î7Ÿ¬o+‰¿E#?·Ö;«ÑxöþgÿÐÇh/ßG”wMµ?8xkå éxîÕÚGoÐÌ{.\*ü)š­—Xs48wÕ0°i,Í¿çéoÜŒF »`’ì­z´¨M¾èð’À7·‹W{õÑòBpóû;°½N— ¥d¶ó …Ì„ü11„dü(?‡S>_Ð …Ù#;;äðÎ åLÄ»HÓáü)Á›¹TôÔÁAÑYâ¿ê»Z>¯%B]Ytê!60V=!ÉÞ°VéP^ýß}½7Fþê/;CŸ×kKre ¬öÅø@>¿‹wéŠú?ÿÏPÌèŽC!Fˆo´@~¸¶ÌÄ—LÈÏzBšžG/ë¯Bak½ëܹ2(¶o{é©Ã%ã̹AñPÎ{<Æo *=VFä®@µÚAGÑ&¨e¤èKqA¨ç®^—‚†Ë»ØŽþ€fjÍ‹«ŸBë€x´tŒ…Ç®8PBWV}£ÿ 蹤´­ÿ\úìD8fˆÁÀÉGEg_ÂPüÎÀƒ>0–ࢠŸ³KíF³µa,)ÀŠeOLŸçãåO†É€9nµ4:ø"sÏíô[b˜ÑOÝ3»Ÿ‡f¶h†<€Û‡‚t]UðƒóXÂ^^zXܪ¦×œß†Ÿçoê¿ø> ËKÅíÎg¯Ájd™òM"üö°¬ ÉK†µHÖÏ>…ó°ÉB“Ý¢BÛ3Á3vdG˜\K I^eç–R’øŸ»©†¶Ê¢ÙÆHuÜ%0å7ÒÜÛtâ÷šÃ]Dža {pÏè½Ì[§7‘!¾t€:ã 2=±kvôB–·®ÎŸX;íkúú赃Ⱦ÷§Pôr§pÑÍ"o|´Áá÷ÈŸ›Üyò> vz¿[ÓbEêôS—ÉP¬=…ÊǸ ÷¿_(Òd—@)™£¡K§+£ûãÄɵX”õ#µX˜Fùü“Äß\Q‘ù審¢x¤_>#âønDÁñpÝ T;ü"-`Õ||Fa䋚)£Ç/Ĺ£ö™Ñ;Tl¨Kö‘ïVÔYÔ{’z„e£õ™&üþÀŒÁÇÓ“ÉÑp:Zj[€åëJ桉Ò9ñM@“NË,]^*4=Ž5nÑtlè² ¢™]ܶÛy&4kKé5R\&ðÍfmßa4.:êŽD³johõ@Í€hû±[zë”H…ÞêÞ%yïJ-”,ÿÖAÈf¶ßnn ÿ[Ÿäßz ÿÖQxãÓøT×ßøïïÿÖ9ùwþºÃJÏ&3]¡Ñ.ânU{*´RÛ0TdþW_¤Aì[Múå¿uMzÜD[¤l¡g'æNƒ©ÞßúÝC鬅ñÐ¥F1RÿñMíŸæ4'¡ó UÐ8´¹=ÒËÇ ­+½ä&94ÐzƒÍîü¬ðßú&ÿÿ÷üŽŸèö½„.ÚÈüeÿÏuôf³Ô0–C¿AⓃ0èo™§öò1 Ïлî±*†O ;zÓ 0øÜ±\û%L(} 9r ¦ {Ž¿´€R²µàó0oPÞ²Fß9_r³ÙåÁ“Phˆ4,³SfP0À*ϱ· ð›õîìíRWXcùÚcW)‡,ùÿ‡¬óŽÆúÿÿ¿½É^YÙÊ)ñxˆlÙ+T”†¢ÌRQšŠHDF2#!”P6Ù+Ù²’‘Mø]Þçûö>çóûë~®s^×ë:ι×ãqó|¾nÏ“Ç`ý.ŸÅD 'íÜ,?‘‘ŽÄ…î9*bLHçÌM.‚äYUi¢­rHù›”„lwÒ\ói+ëwBú£‘e3y‘QÃMŠ9À™–{[L¾¥#K™j„ÆÓãÈöÔuñ–nrWê!:Òƒ\òÅ'oJëâNaîM~VsäÕN#-Œ@þ7R„"q—”…ý{i|÷e#ýð^¶J'¹óE?îeóFñ]ÊötÃÝËÖ¤=ŸP²p¡÷žô”9¡ö¥wí#Ê õÕËê)ÉPŒé÷&*ª‰”±KD¥E?ãñ°:ÜÇžtñê1ª\ªqkªXFÕ±úù­¬„Ü:9Qý—¬=ŠjXDýªÌ[AÍ5þ5ú&mÔª¡.:SÑ‚:É!{JùPï’_ìò 4Ë~7~× +¹ê)¹ÐH_-‡’‹Ä;÷w´¡IÈ\pî64Ušj¿l ‰¦­j‰Ãä&hvú+ÃL šÍtw>~Mà7÷ØQZ r4ß)b/’ƒÖ#ÊRšÖhñ™ÛßT¬ˆÐßN}Ž08ƒ–ž‚,J¼t°JºfÒùŸúnž*ªÛ®¯Ìé'„n{Jþõü[?IËûCY×Ò¶ë,ÞaÇŒ¬ô ˆUý*©9ѧØË|Ò†¸ÑÜ•gý¶}&ÿ¾ÿߺÍv¸jíbüv=æF“‰59BÞmãч / ý+jÓÿØÁm_Ñÿæÿú‹r($ê?lÿÿæÿwßÿó³üë1ú׃ò¯7å_ßI©ã•sz Ü“Ååìg·m¿É¿^“ÿýø×?´íª/u+½kí¥—iü|ì¶=GÝöS/–])¡ç§¾9™nôïýiññ} ¯Ï)äIçú ØOûÒ›h¡ ÍÙª®°#(,ÂÕGºùE½7…¿>C R_J‹ÜÓižÜa2>‡A’å*Xâv=O@…q2¾ßëWq/Ϥ@§ î»J›pϨ¡ö„ïg·¸õ¢S–¨>aw\#75bZ‰çúÊñÅ7šL*Ôf7Ø·ïˆ:9yqZ Ú¨§[õ<ìökÔoœ]Ò¾€†ºWmoûááÌõà~g4¦2ò6’6CѪcù2hR¼ÿYs/šš„×·ô§¢é÷ÛGøÐÌÒ//ðÞ'4«y”µû{>¡¿ µ¾V¢Có¨w»˜É¾"цƒåòP 2#a³ ˆ†q4´Ô>n¤êsÚÞ„÷õlTœ&¶ý$‹£c?žùBº“ÌÙ ç»2¢¦nF ~œ‡ÓT n4§+S×k;S>ŠüïÃÎH8„ØBÑY)ç!l†²ÌŸ)';‹¡ªMÉÞhG;ÔúrL=…†3…ıû éüx\¿4¤ôgú¨›©ˆÔúQ&µH×h'·Àn‹;FöIV{#Sì×oê£DÈâp$lÆü²íaU7j±Eö…³‘ω‰³ÞÙáדUä®L>ùêòÌúY2ÕÛ"¿ê­[Wöã.Š›§›£à=s÷ù|[¯ûK¶†"3"×ÞžF± F‰•¸›‡Ø¸*7%ïþ­«áÏC¹ïžYz(G7µÏ°JåÃeÂ$3mQQ[îêÌqT|ÿ’¤÷‘:‹èSB‡â«ÍOyPµåÒ=wš$DͶ,ÿý·Qý§T¡Ó³xÔ0r:‡š«|¿v=”D­joîåeÔyêòPÏ­+'-E ¤ä.d†=FïŠÎT«hd*Ð3‡ÆM Fmˆ&qgýß<1ASÓqÎ4ý[ã1WŽf¯d îU ¢9>ÿq\}Í;Üî?IG žérw.´X lø¸!‹–jÚ ïÑŠñ1«sôk´Š°  4•ƒ{Æü0òƒëW)2!‡ÿq²åOȤŠÒ-¡r‚4Ÿiê9rxsI¢ÀËò¤Ä¦rºIt›A’‹’„¾0,_ZøÏÿ!Õ/qCŒ`ÀO#².xyMè çôiˆc%öT½¯‹Ç̼òý!U²á‹© ;¼»C [ÕÙ\s¿¢oSÂû¿Z ÎYu{T¾E¿'òTóë²bhàÏÁá[gRáÃZ-W1oØ9CyI*tò’¢FœÒ!·ŠõD®ÆäŒ¯ödš@Žû¸[Ø—7Û™«•V"ä`y"òé¼r|ÿp_lèîmQ#Û†ø’8yþ6)Nü Åv™×¹§5¡TþÌŽš<(×ÜüËä#•/[¬ªSá+ÑqÒ€’P›÷gvÈç94¤\»t¹¥™0ÒÍøÎ€–w“6AndÐv͘ëÖ®ýÐ~fw}t;t„Px» ]Ó“lâ*žÐsñÜ„v+7ô- G”º¦À€éuÒ™¬Ü0R>× Ã—…Â(†d`än—YøÎfëŽv-Þa¿.S«•ó=€)1÷þDž0ã_-UñÜfçßÊVûGÀܵͫÃY?`a—ð‰S°¸ xÙÂ…V œOÉ1ÃÚÍ5c£/°¡f@¦&y‰âCœO=˜B :’³§o"ébïÓ‹H^¸Á7Ó臔É÷}Ì”‚‘ú Ê®·ê!ÝÎYýë‡2¡«—á²Ñ[d¢VôÊ’;‡ÌÝ ¿2µa—³"×odŸ_`kE®óŸ¾)Xþ@çƒ ´ë‘d_o2à®ÁZZÞð‘MÛƒ¢Zq¢¿(øQ¼1ö ½Å_ÜóêÅ̤CÊìkoÌR‡rq7eú•î£BqÝüxJs‰ÖÞOpŸia‘Ä <À®CB“ú Ao*ÒéÆ3Tÿ®<l¹ 5V Ëuã!­’w‰ƒ¨MÇÑD;i‰:ÙŽ¯lŽÝ@=cËÈç†B¨ß_êñÈ# OÝi\ÁÃýdúŸ磱q¿Ïä>4±=üæÅ•4¥4oz;è…¦ï¼Ã^‘O¡™Ù·ÒhöÇed£ÉÍCúòD?r¢…„5ûµß-hQ£kWa³n_6Õ]A(IÜ»Ñèø ŠÞ]ß¼ êj™Ž%wAöï;ÑG×·ý$o+ØÚ„÷Cšãy…Sâ ¸‘P¯;.ŽDþ8 q¶O|×ij·skŸr:¹á>%NA¡oOŽÐª:”I„~&ç†*Ã.©nS2¨™j{ÌæõÕµpb741¦šçXŠA³QmÉUÇy3Éx&Ìiñ´Qo œÌ8Žzç•k$õO£x¦ÔåØD4¬3&ÕüÔFöƒ¡³9ýh<Ì_~¥ñ šdõxÆž BSס¯ù?yÑLôh8ñ 2šu?wI#Ô“iŸdo#Z †?Å.3´˜æû$Y臖±2§åS¡•ɵUUïT´&+<§2݇ÖEAÌQKOaùr>nJ;B¾}1wˆñÓ­¹o%ÞÝÞ1\¬Qc¸Hèk{Rf‹‹ Õn@¹å«1$ß=¥ò€ç1Û*îšÉ€×Æ,©ã‡¶öK.šåB4•vÔ¯Š&ˆzô[ô£ð=ÏÕ†Ÿâ:E¸ÎàRðúæÖ¾uÊÏÑwœ±vñdµwÇç.ð¾ÄÓ’ ¶ø+Îubr…²Ú]/CÞ×ѯwÜáCtúèÕ° ÿòžHœåkȳРYT¹¹þ×/íjUÙâµéÇaãäú~h㿜3»Î7s€ÐÙÜÞ>‰#pßò^†qÛ£O!o’Lüûó4Õþ¼‡2P,Ëýâ¡”’ZäšWÝ‚rÖÉ OríPi™˜:áòªó?Ÿk=œ µNá¹òסA_ßcæ¦=4UIÚP(’A‹³F¿iû-hï.=X´Úi{¹¢ËÌ C¢ob!Ùºn‡ÚPöÁ• ›ŽºCßÍû,^„¹“kÊô0' >9±xõÕ)Ö™™ …³½öbÙl0÷dãÆø%+n£ò”~wÅ‘,$ÀÌQ£åéñ:˜íª­ÊØsnâý&jŸ`Aèhð›ü#°DdêYÙõV¸u_bY‚µ¤þî)°!àuç‘èîgÑ6I$á¹Â˜G˜ËHk& Gs©‘üÉ qªTRœûìΰ„Ôi.°#“阠žåPL$+ßfÃ~w8>CÃ[a;µ|LÐh‡høH‡{øj¨¢IpëâI45º j5G‹f æÖ÷( Yƒ¸ñÛ$U4dÏáÉMèkæ2‚—Ñrgà‹¤´œhˆ/ú³6¾éßì‡:à mÖ‚O—÷]vU½y‘“ ŠÍpàó'–m?IúÍ/çB´àͱd1U‡H†fò4õ)x­Pu#Ð÷ÄFÓHííLZÆ!ÓDÈÖ£ŒX–w‡‚ÇÁN9ï 4þ@•“¬'TV\ß<&5Ž¢þšC½D7cÿah<×Bok¾Mï¥ýÞBóÊ9ù¯dÔÐâ!\w·{ZÔÙznÑ—BóEÊù¸oh:»Þ÷äW 4rž´kæú4»ØÄ§ÚP×c¸Â´Ûêî¬Ó¿³w%pÛCã°ÇcP·ÏÓÿ±™ÔYê>øë¾åïZ4&7€†»bö¥eõ[~J)NnhƾWd[ÏõydOe­rï9‘DA[¾ð°†´K¾‰ùÎ ̺ͻ¼û k¯îCÅ×?ª.ÜM±Ù}ŸH×ÖÒaÀ÷ã‹‹K0ä¢ iĬ ?ÛÍ9îî±tÿä*<ðkl³ع`Ú‘gµÙf¯Ç ´uŽÀœ½¼¢,Ò‹°l~ KšÖwwöÂÊæÕe”°VâÈÿa6<¿¥‡~‚DMlì,1DHÌø¬å3’…ê É™CŠ$áæF¤&’8(÷i³0l¶#Cʨ¡ê4düÛ½‹:™#ŸÜ7hšGV‡ÖÑŸeȾ«­ÄX 9I ìŠÛ#7í—œÇÈ£³'KjªùŠŽ¼Ö¾„_Š®>2AA…«6wÇ£P=_ù- "¢èžøb×J)ðž ê‚É!ƒ6•d™ŽR=ÏÊu†\‚”ƒ9rjJtžO|u>ÀmA7íÉ@¼»æ®õ·'ðŠ#¹ DM‘›=¸W‘ hÿI‚Ó§º¦j‘ÿ-#2YÃ0Þ½•òí4d¨dEž1Øò[.¤¶Àû{¯>°¼¬!ðWÈ…ÐeMÈ}ü|Muò.gÜ»ê0$W‰òîÉ£Yãç˜yÄc¼KŒ:»ìAÈI¿-^Öò¾¹ÅƒÉ;k̶3'nå·¿8äž]VR[„¼ÛæL~·9¶=FŸèÑȇ& ã„Ó=N—µdÍ5 ¤MÃöGG”ýìÜyâW T dܼÎxª=y[ø¡–YPíÞäi¨Ÿäºè,3M ¾<òªƒÁg9Õˆ¡µýìyb› ø–ùø•ÔX|¯Èçz×]l£´ß àÇítÎÚѯÐGÇyÙéªôLœd°‚ÁãU1Eß—a˜ó³Õoá ;õQ÷ŒºŸé8Yã:YÔIð;¸AìÐD;ÌìÿVέ¦ ³ÅÇ^ì„9'3­‚¯á° ò\B’ž –h/¸íד…a’Ã÷Àš/mÖSÆÇ°A_6¬îbƒDn¤Ìc|ÎH¼8ÂOV^ޤ1+Eg=÷#ù¹¹aÇ ¤t…âªÞ’Ø ÈOõK—o(¢ø™‘bKÜ#/z›õ†Jçÿi{l…rÖöÆG=MQÁíXì£QTzÅQ>ò°÷óM¹4E•j+RMTûû 5;‡ Õ3g‰Ÿl¢ÆÙwgÊÛ©ñXÔ­?Ï•Pk)sCÔPçí‡×~²3¨gïÝ#Œ$'w™¼DÃØ˜ðO>¡èþd·Aã>2qC94)±Õ\@Óg÷SˆH_¡™«š‰šë]­²Ž’:£'+Ñ’ãF1ŸO;Zѽ”:~ ­™|2öÖ«‚{˜¯Ç@±§–Œ.¡OåKÕ¡Ý•‡ëÌûK#²¤ 6†“·ý$iŽîo²S óåÅm{BýqY÷pAü#~ΫÊ+Þ}†³íb×a@ÍB¯'(†Ö/é‡ÁOâ»î²ÂØ®rн«Wà—Ô¥\ýŸ0åõ1°0/fú[nßJûC\†ù”^¿¨NRXLnúRÊ ËõÉ0ñˆÖv×}LìÓ€õ^úf]S$ºið¢"It_ìxâ…d25‹âDŠH±_­›]Ý©î}™a‹'BZá—~wõXþh.h#ãÓÀ^&î5dVμµò¨ Y¦)ys.^F¶D­V‹g‘#ÚéÈî äÊ0˜¥×ãÅHúlÆ!ß…³ ©WÙPàqj€Œ4 ÒÐ×Ö¯W¢P¬tsÜÝJ±q5ô”dC±C‘óD(áΣ¢&‹{˜ÂŸÍF鸄Gâr (ÛÂv°6PÞ¹³Õwé*ÒÚ-ÍB¥B•Þ÷f™¨ü5ËâBg#ªðÅG´n¢ê#·9)’ýˆt&v¶å¨Þ*Ò£–|5ôêÈö» æÒÎû/½¨UöA!:Äuƒb*#ãPïøì‚THð¿M#ϯGÖ;ÊŸ.D£ÑÅ׊ŸÐ„<ÈúÇ:š´UyÀHŒ¦‰…o0 Ù¿ûü_øÑÜädòÖ;´ØS5\Í‚–t?¹ˆUcÑrAÐhÝ\­F|žÙKBëÁ/±²§hó›…­ç©5ÚRÎ3“´ì„ÅáÖiRÚAø¸÷>oÀ²}X£SžÁ[ªÁï©?CŠ[§{iH3‘}”µM¾s¸ÔåxÌÔ×ëg¼òêñˆØk 1½˃/¾Adk9µ˜Îgx‘ï?«R ѧíWÔ›}ésñ½5ÿ9 [¼ÞoŸ“ “XPB–ö^ßÛEÊðÞ qy,L—À_…ª9ä!„>´ä+@gy" Oîß·[Ó¾×ÿËlÇùD}6B Ù·4²å[1{¹:W¼Åk+6Ü[xìü@¯'4™žãÌ} ÍÝÓÓ½¯ý¡õöKñ|?|³ÔÿÌ ßÃt#-¡3eøÞ‡øðƒ¾.“ÁCzŸËŸÎõ/„þÔ¯'›0()©CÌCãAÉþ°Àω©‹º½Ñ0¶ûºiv÷e˜ˆ?,vC~«ÿÕOî§„vûG‡B_Ãl2ÕÂÍø«0gs-MÌÒÄÊß:tÁ{KÚ‡JX‘JüÞYkWXó^\…õE" ºCHdUt&õl27›K™ÍýDROJ©q¹sH~pªŠk¼)udô3Jjúv¥‚³Òþ½„¯:4>ñMPûkU` Ô¯7uª%úe÷#ª;¼ÐL³ßã@4;ï6²ÝwšU7ëʹסÉ/N$ùfúÖsŸ /Âé ¾túÆÓ'TP÷2Ô=ë} ÔQ;žuÑÔ'Üw:Ô²y™ðÚþ&K‹ïÖ:õ 3Íx¨sL9<ƒ5[ûOž/‡ÕACmýÉñˆsÐä °‚7žðy©ûΩB‹Ù&[´&ºèEøiûBU¦aH³unŽonw!|¯}öˆŽò,t»®ö??xn)>^ñÞ¾(3-è_Ÿ^5à/‡ÁoUÅšyƒð“>“qÏu §tzü©&ÜGYfÁïûÜ)»öÁŒ«à½é3üðGv_›·?ÌÓ{žè?‹tz«ídV°¬x ’ÎNVŸî3}úî!¬k;]dÕdD"ù[Ä+¿õ‘øO"ÃcŠ+HÚhP=fàäu†m+&ŠúqÞói¤yn?IóÎéÏ?0È@ÆÞÄ’FÈÔVï—l”Ž,wßžñ_ÙƒlF ¡“ÑAȱÞêÙîbä2éºßn‚;_vÑë3ž@>þKå´9(p.oôd]!îê…ìÎ(än|ç5Šˆt…Ѹª¢½ç‘þi”Ø'"ÿ÷¼á¼Û9ìÒn ±z= e?_Ž4ˆByÛ"ϯ@… žTM-TzûFãTéTÎó¢Oë Eªˆø¶º T½ÂUÊnQŒ°Ò0^T êõBj÷HZPC‡“ÆŠ[5¹ÓÜ®•£V©àã=DPçIàòph1êÙÓdÈ‘ ÁNIçºÊ4üÓ¼P}?döêj‡szÿ"šŒ¨ú «¢iIëÞè:J4{MïŸñ6 Íküù¸çÑâê­Ñg]ÑÒ3ïÔõB_óæ:“†Ö¼’£ãah“Î Ç{. ÀùŒ‘0´“œ‡Çaq§£öÕxøØø˜È`·:dMLÌ”‹êk§×ÊÓì: ™*1B˜ )Ó§DˆÜJÞ›PMèW—öËC܃„øö½„ú:±I*ACèk·¾TPå@Dƒ~û²xyMäÄ= B=z vÖ^…¤eUm"·>kõ6úVé?ç ˆ4½„÷Œ*E•þš ùÎܹ‚ŽŽ‡º‹!·¥Ó'0X—À[f&©³—ÿK]¾zy/È=ªh¾\0°Õ3ÏÜ!Ýâµú·*ï·x°'¡íïùW;—)Ž„p]Ñçc¢AUmw‰tÛc”Os6¦Š êlæ# ¾À·þÝUPr‰7qjÍʲUrŽÍ~Šemô¡êç¸w9¡ÆÕôÁn‡½P/óá¬íÃMhb yµxšŸi_N¤Þ€V¹›…þ¡eж²œi¤Píëç/ =©ÎƒÄ£n‚Ðþ[Çäïeè0´¢à„~÷Á{œŸ=``9tú/ }:BF³D ??èûtz Àè´çiÍn˜°»›=_ “ó»/Ð|ÛÓ?+szªù`6Xìá9Ò0g˜È6` »)¤s¢`iWŠ3Ÿ¨=¬(ëÉ¥&ÁÚ%–NÁ°Þ«Z®™.‡Dj— Ïe#qZñWB”HªÃã<…äœã©-:¿RŒ)´>J©/ÎÒÒVAÚÙUÞ‘É—È×úÝŠ)Ìï@æ Q‘@í£ÈjÌÇž í‚ìmk+£ÔÈùœŒ:°ö#î¼Ð @’v ùjÇ|:Úèp×!S¾ õ(¤úcèÈ]¹Ó_Iº¬‹âDÒJ –θ»ô=›jt J«¬³M‡ô£ƒò…®,.TàŸÑOÒŒ@%£@yë;ó¨œy“ÿ©>ª\=5d?‹jïDè‰Æž¡zàŽ-?PãPxZâUe“׎— þïèá#¡aZàŒ ás¼ ÊuãN¸ñý[³Àih óùÒô2ZßÓýmûm%«Ù¥Ö7 |þTÉ1:øžr`ŠØD:ïúg™ª„î*::Š~èõhvð;· ú?½ùø¾ Ãhà< § Pº3£tú¥“×äa<Ïé¦S“LFM¿yô¦««=Wì|`6’õwŒ55ÌyP銶À‚[™Ò”O,E6–“‚•õ×·o|‡¿ïÒš¬Õ`³ÊmºK¢‰_fчÄY#é…º¨”çyH~ÜN'ÖÀ)žôû|«i¨xmCÙB®n ÌÏúâŽò’ªÁk‡É}ÂÇ’þ>²ð˜g¾Õ‡¬Ã§nÊ@ö–<óÍPwäœs:GvòîÔÉ:`³Ky›îZ\1GÍrˆó¸ëƒbö±vzÒtS~uÏ…çt¾¿ÊpBÑvÃæ‚'(¾)½'rr÷xò{ñ~ËFiü¾XW;‹²o:ÿÔä-£¼A\j¶¯*ÌôˆÌßD¥X®$NùרüzoåX»2îÿ’þ-²UO”†n&k#Œ¾\¿óªW D7ã%Ô•£GÕKÔ\àŠº£œŠZÅÁS=rQçQ 9ɽ~Ô³HnÝ‹ìiâOZа#ßENðÝ’ÔÁøM¸(ê}{ÓÐd) =>^BÓžúLä""ÌÑWÔ^R y…œœæÌ>Âܨ6qÒÊ-»²þZÿF«5‘¡ú F´‘*)÷L8G¼aŠ…Ém¿9—øïk@{³£LÎé¯aÁt*†büÄÖ~{Ã>OaÈ:Z¢¡ß—»uÞÆ[ rzH:Ï[º ¾HFŒÀkÝyß®¸-¡+bU¿¿¦§¨ƒ¨)r¡ê{ðÒ‰¯i>ÄZP½BQ¦ìØÞØòuQÌíX&Ü#9 Ý0Uy_Öõ­s ª†I »]Ò7ÌOëþZœîZmç4)Èõ¿qôáówŽSäs:í¶¹½1º Ü-[Þ_מJþ '<{B„7àsÿ{.q’£P¢CkàCW e615G9_CÅͯTûþÊCU¬*iøS¨á—þËÍuÍ÷œZAãóI« úÐlôró¼›´ô1œR—L‚¶˜tåÞ#êÐþâPbO0@GCŸ‡Zt Ò÷>¹×=oNkY{BÿþÞðèÙ(_¸– CA¢Ak›ðóvÒ­s´"0šãüÔâc/L°&”²ž3Éd›Ç®5Î0Óf af=µ…ÌNœzƒ÷†« ,HâZ&•=,IVœ }úVÔ'_çÇš§ð¡µ4X¯¼¢¨eȇDÂòL{Ö‘øž”Þ´™’îd˜x[‡dÓÓ(‘’b®}Òë%RÛ xX ."í¨¯Y {2¼mÖ .½ˆŒßí÷ql"óå‰Íhjd•9ªA¤„ì›ëï“‘Óá¹U\ª,ƒù;ÉÈw×âø©C_pOö׊×QˆÇû#9ŠXdþ®QlE±:ÓŽiË;¸;på˜Iz J­ÝKö£O@Ùñ„Æë(¿”/ w§•D4‰w ò}ù}¶N:¨bîššö Õî{¤¦|CõË÷^“x †|Pý³:#Ô\u\SÌO@­Nö2£c;Q'écRŽK$ê¹íÌHÝKà7Ŷ®¿¼™h¸XN¿ÓqjßÕE¸mí+éѺ©D…¦v¿EUIÐìÜ÷‰5%'4Ps.¹X- <¤­Þ'¢)gç!õV´vºØÈÌ56ÃÖNd¢íú$¶ºiX-Ž&Ô™äL¯ñ·A®i@¸¨l:dµz.|Þ o.µÞ«®Ùö“$×®¹Ü»ÑIæÏlŒ-àõôrèѬ(ˆ{%2¤û·¢ä:MVÞ’nçk“gª·µàKèŒÑøÐXìLÌ9(¾3FîÔÓå)ÌZÑÏG ZòvQêën¨Éà’qìºÐÆDÕh°§¸\ÁÆM4 ê;=c\%Ó$ýô4í/ôW†FN··öîUPŸ¥›Bkuê\¯Åß·øµÁšE- P«äý½Êþ Ôš 2ÿÕ¹ µi’ÙÏX }N3›E‰‚pýs«Û‚[çé(Ç B#³çyFW*hz`m”,´šÃŒ³\Òâ¡%Aèj¼-´~nµ¹Ë= må“ñ'àÛª¤Q]¿|¿3h¸1 úOΞ>ª Ýîφ®EVB/y‰HJÈôß³Õ™A´·Øow†µè6ðNŒÜ Ù¥sàŒ­vŸqÏׇ_ï‚T®/ÚÂÔƒ.–±L˜%¥¨:ñþ´GK¹Nï‡ùoªî“ °DìÔf<·VÎì¸ía"9çä"÷=„M‰žîâ+âH¬]D÷PEI_8;°„dœ×‡š*‘r‡Ò¡°9E¤¾Òó•íZÒIk Û ràæðúÒèIdÚAéIRw ™³WÖ¸éž"ëÕt¿š‘Ýõ¸©ó¢ r>›Iº1“ƒÜî;Ÿ–AÞ›O†›®þAžæäzqÜ\-’¬!‡Bl¡ñ¡(\œË»oú!ŠÆŠ‹÷(£x¥²ð±òjÜ£*z±r¥w¸¿‰å2AÙpUýÖ(¯z‡µ®¡ÂPt‘ºËGT }@]ÿ•­ø‹´Ãý?î䵟]BUÃóÇÇVêÚêÈø’¡z ï{¾3jÌdäM‡šóœAásPëóŠ¡Öp7ê=x+àGŒzÆÔJï¼Ï¡³Ä¨í5v·Q¸ø"]Ï'½‰&‚å÷ 8 )µ¨ü™‡YhºR-ÑjÍ¥Ü믢ùfø s…/Zr,ÿ–—D«ƒ6W~>Aëëb{•¼Ð¦ù´ÝæñCh Ó¨ýÑîë3Ó X=ã 9ü¡ßÕü³ŽÇºùs‹Y¥úþKë§2T_ºáýÌmþŽû|cÿÙ™mÑGÊÇÓìrÞði9¤ò­a¾5Q-€ŽåYb>^(“²%o±‚ Ãæ¾º8n¨²‰ºhüf|­öo«o#Ô­§¥£Á;4 +8;Ý Í”srí¡åaýfõ}hÓ<þåÂõBhWâm ¿— Ž?–uhV¡+]Ú»IzýSž4B?qWªÈxl‹5Ù0d7WDO´?MH‚V_9Ãè•ãJ‰Éa0^_ö#°¯&mïå4}‰„ië¶z˜=âôI¥ü(Ì).u;¯Ð´O¸œ=/,)ž %½+ú¡l5·.š›ä%>EEXÏÈMО¤A"šÉS¿uŽ#±C¤ÚM}$™–¹qþü:’Uö]Ô$DŠ‘v^Z# ¤Ö=|ãìf6Òþ؈RŠ ž2ÈX’yL›™ON]ÿ;—‹¬;^ºÚïKC¶ß»7]O #çñ×y÷’‘{öéÑ‘£-ÈÅZ³¹ž(°ÀêuqUglw¨ff£ˆðM¶P÷£(á}mÜ2w›c²S!J2î§y茲Ì!ÖS(ßxs¡°’×þÆVûíFåãC&뚨"éíT›‹jŽÅÚÆ¼l¨îtRú0Ó'Ô¸1}ïjþ2ª_pãC­†![»”¨“p2¹ò[ ê]JÍNß$ð›V¢•Ý<ÌÒô ”PwF«-\å=ÏÑäz’މ š>òt7ÿ{ͲþX%h­¢ùXŠŒ÷Ï`´Ü[žD´DDà6*ê·ÐFüKýªiLÕ Ô]fôé_eq°òyŸ‚¨û3(ä+ðÜ…ªÇY]ó óÕ³"²Ëóæx©ó8ïü¶Ÿ$)RºÞWMZnÇê/æK OšÃ•»¼ºì¾"9ž?{"—½ñ¬%_ßB†Üñ‘V·¼éžbÓ _ÖD%j˜O@¹ÀŽÝ,„ùLÝAоjvˆ 9ÐIè?Æ´Ì [çÃI Çî||F÷úù)óÚ)A¹{»¡Ñ­èÛþh£ê=)¾uã;í%ô±Ò’¥1¨¥8ž^6xjjýwx^ÛYßÒ…Z_ÖZË÷¶„ë>|ö|ûqëüS+ê„Bܰ8JGàÅÊ ¯¿[CSôŸNòÄ)hNú•{¯Î Zr”<ßð@kMd=½#%´Uõ>.½p¾M9)wœHƒïŽ÷Ó~ äBçŽ$~Úèæáÿ4sÞz"2'ZßW@?–Ñ>ùAó¢1ŠRÍ0Ô_à­wç#ü\l¾ñjmÆÐìðõxv˜(­9ç0 ¿Õ¥½ÇaƵžæü9 Ïü~1æ÷ç¦ó;RÃâ™;÷/Ú°ÜtÇQÀÖn©„õ(èÃÆCýª¤j$šùJû IÞ6”RF²»>9âV HFI{ædRýŠQޤFÚGÑ'º³‘ÁÁµ5¾ø 2¦ »fDf½¸C#Ïu‘•±öëÛ°9d§¢2j³kCÎ}¾ƒk~£È}<ÜK#yåã½å›B‘ÿ×gFRÓ"Üåd1a‰‚£YÑñ÷ZQ8DðD“lŠžÙÈx.‹âÁ꼂/q¹”IâGv”ê£=æ~:eoǼþ:”…òRgnŒ}C…öc‰=Péæ²«ó3Tö-Ü¢%÷W^-þ®¼U÷²J·š#”ê?ùõã"ªìÕlj…¦^ßZËï49ñ_ÒtEúºÅm kjºð¾cuâ­bݯ¥Uúmñ Œ‡ÓÊv¾'W™$3®†÷a§év ¦8‘Ë6åÄémч{¢ájRð)$õÝ•ÙSðYØ5Hð2<”¹¥«—Ç3Ò@GׇC$PÅ´ÚU_/q^ ž‚:6á£åñÐPÛxÍÝšJô–Õ®A‹ŠXÞ§óÐ:Æ¢½‘  ßš–HMýÅàû|—祎Óеטž3Ê~|ŽZ‰Ü}OÞz›d½€£ŸÇèǾÁd Wù$üä9ö,Hù-Œ‚Eûâsm¿7¸Dlé “ô9 9u0Í8}¹»0fñ¶©ð:ÌIìRÿvgd³ä³ßõÀ’Ææ1o~X±Pè>ek®ŠÉêE¼°þü«nÐlö>*´‰PžÊtA’2ÕéÉÝRHÝeºv¢òãÉÝb‘Zñêë?. m‹¡÷¹Pdˆª“ZšFƲsßg½TÙøÏž<}d™îŽø¨V‹l¥VÄb"Ùȱði¦Î‰¹³—|;ön ﲤäuQQ¨¿Äïð¥Ë!ÅÙ^…çMÕ¬ÞŽ¢˜Ó ¾!ËܽóîŒO­ Jù¾¨‰ •FÙ8>öàJ”OÖ¯Õ¥<„Š-¹íþ+Z¨¬üÔù•¨:ªûñÿº„jè•-QÝóó$PƒÁ‹é’›/jþ8—'ZåïfÏþ!Ex†”N¤E=?å¼ÌªD4°òn^ûŒ‡e{((¬¾¢±ÄüéYý4‰¿¨ÇëhZºRÄ[8‡f+t¥‡­BÑBçîü!š´ÌæàMS±EkÕ%É6Y´ù½£“º+m«‹&ή©¡}‡”XÉ'SX.¬N»`¾u>†ÉdïUÈþüNŒäM1dô]¤z ©ñ—ú”=y¶ý$‰g­5Y}b ÁˆáĈvÄë>´ÌŸ…˜wov›\e'Ôy—dÞ‰íÜòškò°Azgsãu5È£W0΢ Ÿ'}ukœÙ¡ì…ÝØG®ƒ„ïï:Iæåøê¶âcê V¨ÁÌHêR,›¯ìZ&ô1¢àVMÐq.|0\“üô„z¥W;œû÷BÀ4Û:ÔrRVø²÷BÍõU¶¯˜ Æ1PGÐ{Ôä·=‹/b€ZÿÃìæ ŽÛSÓ&zê®(ÚV^…úêÒƒ·õÐx„ËCù`"4e‡œtφ悀ü¸ßÔÐRå3£%Û­í'>Kæ¶š¦›WúáÛàK¡áUø24»¯BGÛÇö*†ÐU™vJ•zöĽÛè‡~¢c…Än<023&C$C¾Î“æwâá§ kÏßB]½ýU!(pÆÛKm-k}aò©Y’© L¿3óŸ;³±•K+”0#=ªüM ‚Ÿ™Áòþ‘\㙼æ(D¢Èo4®ÓHrؘeÜÇÉ8ιtÇ!kGË6¤:z,I@WifdÏ=;Œôš·†TQïàÙcú¼ÈÔ¯òžÿ/²$mhª°#[|–!”ÜCŽnjþË´ÜÈ­»·â–ù9äÊ"&*ƒüÑM.$p—Ê4Ý>î+(˜ßbòš’…í¥œqÿ¢²ß£IÝ(QÜBmÄë6înP„ñT=”ÊHdÖÌ¡GY7¢¿zWPžOÿI™,*|åûõýùTò<Ó,s+•¯ÿ‰mjÆý™n5äê¨Ê_¸^Ó…ðîë®5Õ.TÏåüœìFŒªº‚ÕS¨ù‡ýü—óv¨UÐDéðUunßw´wÖ@=Í2‡Ìü4 s[“CÃAj•!Yb4ŠzÖþÌMöâúä÷:4•dä4UC3¹"^ÁÍq4×;|bâŒ#Zx:”ëFŽ–¹þÔÉüBk†µpo¢~´¹³G“¢Жó]{ÛÚ55 ÏS8âѬ_¥ó2·ðxÉjèíù ðgáyÂÃÜøÈb&ÿòp¼s¹ïæTÉkDÖG6Èàõ‰Ö3º•ðêõϘ3óá9zÃf×?ëÙOÝþYoû>Eáw69g&~‘ªtàÙ.óÏ"NCDƒ®kûžƒƒOH(K­{—l oŽÅµÑ¨eÁ»š89{Õ-òAƒéhñižÔ¥$̵T \— ç;]Xö¡Ÿu<©°ßΜ¸å*—ÖiÈa©ûò—…зœ6¸*vxm=η‚}‹õó&ŒþËì”;\x¯¨Û§%Eè£^{9ì~l{Œ>à©CÃðI¸<ââ÷(ªÔ/çϺ Åùmª,ê¥PZëtÅZ«Ê‡BæF' ²ÖVm´‘¾ e ¸ë´CmÞœ­šŽ4X1 þóš®Æ=ÿüzšGsŽß : ­+B¾fÃ7—&yÚ+/á{À÷©¤Äè,¾˜`'Ò?tsXôã O(1ÂŽÓv …šòŸÁõK;h†`xìSìÎúY%Õ÷ѹ³ ãê$L—kÖàWñàÚóÔu˜úºÎ {ÒfE^ÑÞ9^ý¤ƒŒÕ° ?Þ HKF<¶£ý¹°r´}eRw?¬Sí8©Më7¤ïèï‚ÍÌR¾¾9HÌæåÕÙŒ$áú.cÇÌë›âûâ#H‘™ªGÔ‰ÔÉ/æïFÚj’Yã§AÈö5N±-ó=^®ÆÛ ó…Ž‹¯è¥‰»œUcÙÂo’ôÿDä(>¬6xðr{ÙFÕ0J oV Íó.Hü^cz7#„>§ìCáò]§x&Ç.|“YyŠ fj©ßÛÈQ)Ò©»¯ë*S_lŽÿe‡ð^;“潪f°ó,5飺R衉;¶¨ñ‹ÍªKTm/W'.,žÐCí{O‹}Ï; ðyI[S .Y¢`&•êMöð<EƒÏ^dªî§Ð· JzàY•'´'©öZu~XµÞ({ü°ò1íA“ô¦ÕA-4#[}×ÒˆæžÍc+ÁOñ¥$ýUŽX´,1ˆ iEëHâ‘™/æh›ÈÁYó  ín„Èì‡ Á ñþ»ç‘ntŒm2jö¿û$ÿÙ%ùOnþ“«¨Ù2êg7þ›«0ºÄƒ «*¯lØßâÿßëîÿg×äÿÝ1ÉP¹¥üÑQ²f'MîÕß½þÅäoj´ÇÓcÿÝ5ù÷HþóùÎýç¹ìÞÃo¥>güwåÿ»þßûÿ9ÿŸïùÏ®ÉÞ;¥.¢ÛÜs=ÿ»c}–jw÷T™wzÔùÿ½þ“÷ÿä¶]vç ÷wŒº¤}Íx–Ž@U²]Í Æÿæ³ÿôËï3Ã`g+ITâüÌy_|ñž3 ¯(†šnÂh«UçË}ý0žËž”váØÿæò/ÿáEÕg0w“‰çï‡#ÿ›KÚŠ»¿'~3ݤM‹Nøß\žS2·<£ }/mæ-g-ê'­K°íìöG·©vïÊ×;èE¢¥‘G‡2ä’yûBb’é¶s¸óx"…¿ß’ó€:RÓu'ƼDÚÒ'"beO‘áY³û“kȘáë@Úî…Ì¢ŸZ/@–ÌsK×NÑ ›{Yd| rnðÒáþc¥EÅ3¦xðéßpÊÈwWÉS]W Ü)ªr¼Pèͯçc"ÿPäßð…o¦¿P,ùáÇ™8”ØN[È¡*@)s+•7MÁ(cöÝËöY>ÊÝ?¤ñë×m”Ÿ\‹œšVB¥üça*B¨¼Ì­y©UÅ7Ú%–)QmÖÚÔœ55ÊØN*¼yŽZ™:¼žÌ£öû );L~¨h\V`±×ß¼Ùëyû›Qßmuç—n4ž P">-ü g‰õJO­ ‘8…I–Ê8?U{÷áV)š’«Ý6µF³¤ºÝÊ}háÕÅæŒ–žcÄQ×Ðú‹Ðµýî x^T/ùòçX[™yvúr<׿D«Yä²ëÝþâÚ.Æ}ÈS>_J0 %–øî­ìñ݇O+»{ñ£/Ä)_Þo•Ÿn_I³Y€÷.áSç ‚&Oý­^$D:ÿîtp€j£*ß\ ÕÂíæs 4=·µ³áR¥Ù<*¶ëP£6ÍqÙjvúÞ0†zÝ,/7á høÉËîKMÖ_•ÍêCó×P,k}-'XÓÆ¢d¡ù¯~ÿ¨!45ÞL¼Z¶ûä¼7“Œ¡a~Iý•?4øë”?o‚ßY‘Ò hX~lNåoŸf ³¤¡‰×¡ŽñÆ4³ÊÛPéA‹wEøâ*hcÛø!®zÚßÈÐ^9&ч˜™ÓNBg|`Š*Ó t¥Òdº±SBwèŸû•iÐ=}6ÿd;ôÉ~ ?}çBϱƒQ&­Nøî\µúzL †xën} b…±%)Óô÷0:nY MþÆ¿T¬Ü‘!…Éù/Ìš`:Šî#Ó$̉û|g»$‹ÌdCGÁr©mÖòð6¬¤]¤œõ$…?m§‚ ÞÃúÑÓWî®FÀßö0î‡_rá_a]¬¾VìNhlGË#qàgÃÛU§‘T•¾ƒÂ‡É…ý: ‘ÒtÞC´U©ÛïŸUQ@º«!Ü?­«qŸ¬~Åoÿd¸Ø¡…ÌCGIKN#k·êm.jä úÚ¦÷é%ryüU¿[‰Ü4„6ÙQ;ä)½ªì+ùªCYO 8¡À‰_g5Oz `¡Yó×uŽ¥îÚ7} E˸^¯j9¢¸ì7Eq;{”dÍ3îî¡E)a²·œm”î½9¢—ûe'IdS<ÆcÒoÝx?ä¡‚èÅ{/éQé ÅÇb•f<1˜ø*¸ÇÁ݃ˆè²ª¾-÷ÏFuþRá&ñ/¨1tG¾… µr9R¹ÃPû…%iª-ê8~A»/Á¨«ÞFwTMõÚÏ_f–Å3$ôî?ü~¢¡ó¯Ën»Êxö\oyÓmy4²><%¯ŽÆ×SœYµ¼Ñ$j"Yà5šþ8µ~¹ÒÍí݈Íß E¦ÑÎyÆ<´Y Ê­Œ{ÜYßj儜jïHƒ@8ñ²ÅóÉK ¤_šàÔBÂYZÂ-b ¯Ê;oÍANj`Я&äÿx<ØN²»‹³×‚Z!Û#ÿùÓàÈ~ñbþÅ#ȮꟈW†:’|é³'!繓G€Äs ÜËÔâ•v‡¼FŠBP®ê÷³Mþ…Ò—¯|´~XAEh”ÓËq­=¾½`ºó% ªn0·€ÚËö¦îËPŸH>¾Âô š8me¤nBËk"¿€;ñÐNçýï–/t\ÙbH.ƒ.jý¶¯ »èBÏá8èíONî\¿ }¿f5[ NÃày?æwý/à§xõòÊØNÏ‘"=` £¡ùá0îý‡.Oâ>L¾¶?Ù,?Ó5ãYÂgavEœfÒaQçRa¹] ,ÏN(•ZÀÊLg Á^VÉ2-Žÿ€u>Ö-÷pø+ÚýäÅ–l íØrªòÁ¶è©ý ãɰ«qÎ]B‰ºÇ-=†Äðk6MþC$;zôDš!R8yyR¸„"Õ´ÔáŠÆHzòY32ØTÑ޿܉Œïö7ä8±#3³¶^éÀed‰²bNÐüŒlf9¹‘ãZoç…‚UÜ϶˜úb¯¿i˜%¤"ŸëÅ´TG3ד ¡ŠBwuµm‹Pd’Ð^Üabo%ÏûšžF‰ÉüÛö º(¥÷îÊr¡Œ~le½ Êù´ Ùç¡ü„BÙtúaT*ÞZ¹ï5€ÊIiæn ª¬¯˜8O ªý8h}€5jo8ú›øìå ÒzµcRõR¨QçIi|°À)Ôuûýàûꟗ»ÀÌgì®3Š$¡aKáöÏ)hD&c|꛽ÖÜá`G“zÅ¥¥jhfï©§Bƒò¿7«¤¡%•6õ´àZ‹7µ?K)GÛçS½ EÔ°iqsTœ n*q]³H€Ì·ÂDB*ôáÓ‘6HüÝUÂyœâä³:!nh´ýjÓ ÄI†¬ÜZå„O)yo+ƒ÷é·:‰XÝ!BÆÊé‚­D–Õÿ|FØãÛdrîIcHå»8×y[7 òÂæý¡$Œ”LëT‰?b›¹5Þ…-]¢PW{@ä•Îhðœt–ÿ' ³?ž‰§íñÌÜsN—á4·Êί›Asuw0ç#h2̤»2ž æ‰Qh¹ î/¿ s)BýVkÇÔï`kê’*4Ü(úv€µ…Y¯xBcªpÞü©hú|”éô¡ÐÂ$C©,öZ3žh>ëv„öóÇŸÊ ÃEÂÉítt^ý•ìêÎ ]7æRò`º­E^VºCÿw÷DÌ 7ÊC—­˜Èãaèÿ!Urêî |g\õÜjà‡Ÿ¥‘ìŽôÕ{ž(xäA`#Œf¾§—òÿ ã¾bI±Sö0i&©æ¦4Óç³ì,”ÂlçÁÔ:­iXø¸ËÕ¯ÌË^öþŸJ^ÀŠÿÂãþðÇÍCÔ^kÖŠÕƒËť௞ÃX¬LüpK: »G¨nþ:ԉćîéW I÷“V¯ö#Yö»§3öTHñ½°-ya © rš©ä[vŒÌò\N2¼ûYܺu[~6f.ù#ó‰ªØ£VÈzI$íÙø ²¿YQOLéBΕñ»ÕïÌðÀýªk{žÅ#>;­ ðùn%$³”I!ÿÜRÌy >tO :ú…O;îÏ`ßBQ»UpB±ö¢"'Q"UräMR ÔSäÌàEé÷Åî·'î ¬þµ^ÕQ”Û1þ`£7òÏŒ5ÖÙPI-Ñý}’,žø2eæ]¨‹p"úÌš ªú¹ôÎÄß@uêIÌoÛ¨Ñ2Áô}gµ2~¯o¡v”ÛÆóu+Ôyvœ*d— uï¯ÿI÷%Fý@¿™âÈÇxÆ:¾Ç÷ÅM4 ™à:´†g½Ý4\ F#·ÏOixcÑøÚ-ŒcÐ$`×Úe¹Móœÿª¦Ðlsu™‘ïZœó^:CÝ„çzŽù°è¡•éñ›?ÐF.íwÛŠžxÒ öÀ–;e¬Òž‡qâä¼U1‡±ëŽa—H »K·½äê"d?žÌaIW€ìĤÜ© »d[¿­¯rÔúû¥7@NóvW i-êu"’!Ÿ±ôÙ÷SmPt´Dª–QJ 6W¯´Ž@EÌ͹'^ž{|kàÚ²ª…êè¿¹Õ  Ö,ÿ ×†ÔûV“~âß6éÓ|ŽB‹FVû5ÿ"hk˜—Ÿª_‡ŽS¹l¬ßù¡sÄL“£å7t6w—‡Þˆ;3m·  ï¦þv…+ Ì‹ DÐ)ÀOó"÷µ_ðÓOÑëÃå0zÒÔlˆç Œ+w\+ÿÓ“ÖoìnRÂôë$yk-˜£^ì–Ûëoûš[ÿæ Ár#õŸ ç°R%­S"<º¯­mñXÂÚä2—Ô³1ؘ>ò§)6Gó³S²Œàß”°ZÈö#Ø¥òcTº HT8%]’¿Iä¾­¿‰dÌëK´Ç%âäÕÍøµ)¤ú”é8PŽ´—¬§Uu™Aº”,Ôì2>,–vØÊG¦oYJù‘ÅßúØ€š&²©l_ö@³V®£!äZãM¼,µƒy‹ôÀ¾ùL^Û(»j €jcVa ¹äR§†y£Hû3åz(v÷”²qÅ/”h«¢f|ö¥Ô¾ KdÊh*ý¦øŠr7n’uÜŽBù1ïÑ™›Q©â’Awæ+T!³Þ ¿Ä„ª'¶Šêë£Ú¶ù„=¿j´vÜïasB­bÞ«Ñì¨ýá—t”zê< tœ›C]›ækcĄ̃¯8êÚÛ—€g“yI%ÔÐ0ëú¾kxö§íåêW–h,<ºlwM‚¤ì)…EÑŒùùˆBr4¯ˆäyÄÒ€ç"Jµ¦G«¤RÓ¹h³òˆìróyX£)#:A ù힉L)ÌNU5)>ì"ÇXÿB¢‰hô Zˆ?q=C¡v//„WŸ|ÖÝ Ž'»ñ½ÁSøðáõ}¼ï|J7­v",‰e¯¦B„&ÛûÙ,røt]" eú|•TÝ? œ`h¹½Å+ŒVTAe^µFñ›wPý›SsÓ&Ô©>î-¸üê7e2ÌïZB£}pØNhš3ûfpšlŠÖŒ} ™½–kÓ {ç<®ÐAÃ7ÎÒÇôP?/Ãe*6õw ÅÄnAýƒ·^ï&) ~í—¹dœ54$¯|Ð<\òÓ‰õ¯AÓ‰ãó §ÞÁž4{’îB«SêŒK$´V?Ys*:䤊*gÎAç ‹¤×µúХܜ@✠݂+†Ì¡GŒqD›ü;ô:/gÿ®•oý¢^¿Wf ÿÍÑ‹÷ ˆ`°HtíBú6ü´rÏ~y†«Ì%¾8mÀ¨ßÃ~Êa?ùÚ¶>º& ö$ƒéýÖgmBaÖ§ò¶_,ýkòÈd–^°2Šà†_»?»xš©àù~ NXÓ>ž¾k[õGS*ó_ÁV}1I5ìÜ?œ0ž{‰jÍ„xË/!‰Ë¶tf2/’©2îš.Y …ãSI ú ¤jž:Ù%^´f>á=4þÈ@t5/åê*2ú˜mx} ™Åø÷º#+ ÝM4dçåË#"p §G[µmÎî_û ––ñF8½™jI@>ñ|•çþ"ÔÂü“êX<*Ðqå4 ­”8hU¢(iRT‚ Š]Þg 8/j Wütð¨{ð®ÕšJ_$^OeóBYv¥{m¥•(×ÄԹɊò _DÖƒöòÄvRñ>sž¸è€Ã{UšÚótèêÈ,3ùŠÕVöñ†=ÑC½ù‘ÙlÔJiW¬Cídëpv sÔI¯í(7|ºå|?$³‡Qÿ‡ØMn<“¯ 14h˜ß×[g”…g_ûêµýF>¼÷$©‰ÑøªæÝÓBehâŵ/£6MCü·Ø’<Ѭ4Ћg&-Hb†‘dÏÙ' é‘¡åOÒÉ¢[Ñh}{}ÄTÜ mU¥…c_Y²°Òáiø¹2ü±WqR¶=MíÅ6!þý3^øXô$úEÄôÉÄÓû<ƒhIš z+ˆrã¢íy8 úí (!,HÑ6¯ù6„ú^“Iš’†P¹#­Š Üå.ñÛ—sj|Xá¶:Ľ%¿\R É?uͨBúT°àʸ9d UÎ|ÍÜã›N»Mcä4>Þ/˜³šŸ(p²!«ã/s·4Âü¼ï/[4«‚á¶äÜ+\èzN9œêÏc< »cpñ¡ d‡³õhyÙ ?Z´ {N¡²ˆþäÜÊÜ’‰÷ý[¹ A"Èe—;ùæ7­ÈÈg ÈM<ÀKÉJG.\¯p‚Š$ƒk¾T‰Põž\À‡zªCïÛ8œùµ˜Ûñ¨ êÏ<0”v‚Æð·3¢E³Ð¼ê(dÒm¾s²½±›Ð!põÀ½û¡³Pó«Št{¥»ÿ‰n€^s#±¦ÝUèS1—¬ Þ}2a0?¸X¬éS“`˜—šg8‘FIŒ¸(9`œ(½»¡” &Eâ>É–Êíyç¥â77`¶{ð˜ÚÝ/°Ð+9.Gÿ–“d˜Ÿ+‘!Ó_gýàO<§J q¬}iò=ðùl$dD“ÀæÇë”*àßgvkØi‰!sh!QÌ,5%-"ɨ›šýNHº40PO‰"®÷5ž(!UÌÔ±“w$VýTƒî+Sd ÏßN‚pdt·zÇüÊ™&G§>ô"Ëu›sûT†‘Müë•yv-äP®±W™¯G®^Ãh­îNäÞdó|Í0|XÖ¡ä­‰‡%;ØBÆìõûZåQ¤È&éq½Š9;rIJD DAsXùéë(%ß«jL"…2ǧÓ´N¡Üe® 5ÕT”)¢˜=¾•ªRÍj£Ê¾¼9ÕÙ TÕôº½„ê=_œD3PãÛÑÁ^cÔª O²ÍEíO 7gQÇïÕ5¥Ü`Ô5ü¬¥5v õyû>è|Ç3Œ#’CÐðí¡å†‹ûñlA ¯Á 64Z•*6­GÑÎÈthZç©5!·‹æýì”f&ñœ¤§y®\/Z‰ìå"’Š6<Ò ? °z[0èÍŒ4äK¾Îòý}2Ü„vª^ArËóÔ~@"™Ô¾Ôuˆ[Lûb/vâž©z<…¸}=±UŸ×àSˆ"Åà ;x¿¦Ý~å5DÑoªÒCÐ ¹.^:½ U©aq+ ÐØv²NÈ¥šÕY'¾ýô…–Ñ–Sÿ:¡m1´~$¾:ˆŽíãz: åŸKøü¡‹ü˃—O k60³Âè t/ ‹1L8@¯Ä¾$…#ð-b¹ºÒnúuŽ0äýâ„AGK¯–Ð~øñ§Y„`88H¼_CFuøóIþÁ8óë¢LŒïwU&⇩Š% î0ËQW|ËæË~‡H^öV;Š) ø•UP°¿?TNÐË™Áj—L¾±… ×G”ïÁæïÌ”‘s°ÝiÐÛ§NƒD$_ü*ý‡ÄÛ7GÉ+…‘´%À«êN’Ïš¦Kº •îþÊžl¤iXì1´wAzg¢¬'ãŸQ¨½dô˜2ÕR½tç²DüÞç{ÙRΈY–Z"ÇJ1õFë}Üïê;ÃÎŽÉG1Â;y‹_%èÏ=A~Ýùã"ŠP`ð»3GZ> E±%/Ü|Š"‰×x{¥.¡uê9¾l&<ÜǽÍtflwÔ~PqFi‰S´®¢Ì‰˜í;;” o§îû¬ò‘훩¼Â¨8=Iæp‡O¹ ë=G•þ契ï娪SµÇA.cñê4ó%à ¬ $Þ¾È>¹çü°ìe-äÇ–8ü•Šâ#†EîCuËBEÖmÊà‚?7Ê«N˜ÂÚÍÄAŠj_ØpíTu7€M[q«¶7øç4Ü}ôì¼®´`ÌÔG¢GóÃgÏ_EbJSûÙsH:À[›b:ŠŽä£,AHå«^$Æ´\r)fß"‘¾?{­DdMÿU‰l S‡TœßÃ@d±¶õÓ<~ÙX›T(•9C $)Å?¹ò¼æñþŒÜÝ×b&J(öüsNKþ© °½È ‹A!oÿ›äP$^†ÕU# Å ï¬QD J|ú&t…– ¥睊¸– Œtø¥¹G­(çT^“E^ŒòC[âsDĨÔpšòŽà6ª`˜uÓGU½9þ”רÎhìyéj ½(öWE­šÕrBâjÇÞ)P¼‘‰:·¶ÌöÝ“F]‡\Ÿ'-PŸ¢µ»äbÌp)—Èl£¡÷p§À×®=ÏÌóÿ6ÞŽFÕµ¹VùÑ„‰WSòÏ/4õ³I—ÎÜDsÛOËdhQAw7ªÔ-ã*}»ð ­K Ï_þEæž±Ï×A^ Þ¹¯: éÏ–åkv ÷|TJP ÏmwâÇÇ îÓTjàó ˆ³YÒ6°„8¢ ßiâ ðém»/M©ÄìOxXɧ;:^—@Ä㯧Hv“öøÖCtà +¤x_Y°…œk*åÇ#K (EwqÛ*ÚN~8_˜ _§Ì-¥Ÿ ö˜…E®Ô“4aÄC4¼½üîò¯QhbÙ¿°ÔªM¯¨i+ Iâ;‹¹¡aÌÜ-ÞB}Ï6 ½G ÔÍÝ\Þ“¨{àéy§ð0Ô¨Özv½†ºm[s£ ¨ÏÚÊSHh0¸â¯vb™&¡iÀßT¸` ZüÄ}“^QC›™b´7+9^q‡ŽN‘Ä ß èl ~Yh ]Y%åµ$Ыt­·zz–¿['‰Ã7‹"æàÖMèÛ•È_—‚Aºçëò¾yð#œi%åm7 [–‹‘óÁ(ÛÚNrlb¥ôƒ‰TÆm¡1€©Ì°4a¦zÿ÷^e˜×_høæ_KêÝÔ.1^ðËØ"ÆX~wý:v8VÏ‹/Ý«[ƒõò%¢JwØ´6þB(_„m!ʦÖD$"![ÿZò‰_Í¿¶ì?Œ¤–»‹Gi7ü:ÿ´”¾R6(»2Ç#Ê概ÓH×ò*ÍþœîûøæÒjæsdrÞ5 F–£ýªƒÄ†ÈvìN4¡69<ϼžaG®Ñga¿$ä¾a3§#˜¼ü„ÖANäÛš}Z§Þ‰¡Z6sB(¤qå×ó8}ÑQˆç»‚‡µ/¼µŽ‡¯„¶ >Å#!¬/F?G©ß! Ôž.(ó¹±sáæÊ]Fv”¿ô§Õ6ç*–šêy©xP¡©ZõNiÈk4SÏy¨|»ÍMf«xV©ÐÂ÷©oÉ6žË#G­½¾lEó¢ä|ä÷½÷ôgh,„‡¿×M‡'ÚÂE+Hª^½—ý“ âhSžX+VÂe§;ùã#ÛÆ÷¹¡hˆR¡7( Êd\gâ< ¢„Ø×Q’jowdo:„êG/Òb:j™rÎFf@ÝÆÎùô³‡¡‘½å#ØA³EMó¡O;Ðú}F’`Ní=c‡Ì”é Ó•OŒž›ºÙ¬=Yé— §{3¹†ê*| 1r}ÑWý¯åŒ^ÇËÁwJûºC7º`ˆ§ì€Ÿ}3Œøê¾7%ÌÁX€kôý˜(šØŠÒxS+µÄ¡?aÖ-_“›e¯¿9,3$ÙÀ²³·Ðy¬èê‹tÝȃ?Ncõ]°vâá‰_#µ°!¦<ס ›Æ2È&áŸHI®×)Ø1Ûú§c‚D.KùQ91H¹§È õ…›ØÁ6ˆ3HÎYf„8™ÖŸNˆ]9`LoÙŸ¢]^¿%† ”~¬Ï‡ˆö¥ÕñZ‘°ÿà>‘¿ðñ…OÝâHQÕ×ÒM÷Üó:7Žëâ‚PDqAåxe TX=m‘n®†¯ö^cÄy7¡&£rJ²ßêžr<-$…Á÷1ý9õИøïƒ4‰ê2;AcÕæ,ü2€×÷ñ—î)@½éU!#a3¨»ÄnE°&…:Ò«ª—55 Ž6ìŠcã9¨»úí¢Á$ÔË©e9>„úi/«¨¯Ð°në·“LMëÕ7Þ\‡¶¼Q*aUhãÉŦhPɯ¯^…Ž÷/ºÿBgøÕ7“£G Ë{±½\ɺoëÄÄ_ž\ÚL\¢ðó‘µ )ôå¶vÔ†Üò[ì^ùñW—9Á°ÈoQ©­û02’h×$c©Í”J`ƒò}­%L¶,ðµ93×Òfƒ³Haîû4ÙƒѰHèÏn4} ËÛl ^ÂJSl¡@ÇžŸî+ÿ÷ÖO„ñ´„¿Õk™Á1áŸ×[År+ ØÅ‚¯‡½Ž#±ÈõŒ_ÒH2}ëbT¤6’­/ie»"¥öƒ|9Šp¤Îì7ÔÞwéä>+·¬#Ãú[‘5d¢,“kŠAæÚr];d­HcdhEö•Wo%IS‘Ëâî‚ýØ <0Üö·•€ùƒˆ‚—ÄMNÝø¡.nBËâY½‰!”]‚Önf ÂVÙ~(/ïõ7S¢WeÙŸ v¦þû}²ÔèOâÄé¯hµ.>„¬&ÇÝ9=¾U‡à Èéz{îô³@8»ÂÙ£yÓÑß®Ðá¾JTN¢Õÿô‚”v ˆ‰RŽŠBöxe^˜d·<×î&rß»Æékv€™ÓXiÈñœ:èŠ@à’·—ûtmVæ¯ìÒ!w6}²ë6)Žh½­"Å\ž!þ„~(³ü8ɺ5»æåŠ{|+<±Âzªo³š¨°@Íf6ý¶/Ôµ”ܵ±g€†^×Ûïî€fš[|Œ©¬ÐúbZê »Ð5žã#XûzÞ´ÎT>Oƒoòi¥ÃãÐÏ3êð²÷' Þ ø—ûm~†°Òœ›#ò§ÎßQ6…1=m¢)¶U˜xBy¨¢¦jòtJ`öX‚î3Û½þ&vVpŒ±–OFõõÁŠ`jÍ ¢ üaa`òO; k$6Ñßw™a}&œÜnþö¿'ûãÅ[“ï¾sÚMÃÎAŠS£VH¤¿²Nô«<-EöÅT…GE‘ËkìN[Åä~&òë»:òkJJ%œBþYÃóŸ¤QˆY™TÙó*ŠÜ˜—9þÅr2 ¯£Äylz~¥ØÏj0+v¢ ÏŸÉl¢”³¤õ2€òßïxÌM¢Rç¡ áp7T‘®.Rq @U[¢InÊ>Tº¸ç‹Y ÄQÙ|þalq”z± §!v ËÔáS±ý~÷Nˆqøº°Úo¹®æ¶BD͵•ºð1˜'„æÚ ¤PiHÞ:ÙUaW?B¡ƒ_7ÇE(o{µFYUƒa— 2NAøáp“mã=.9¶õ?Ý·×·Zâµ<¿@ãñŸZT<ÐX²ôí@54žÍ?š*úê—úiÜú ®{ädØ·e¨Í¯sÏ…Ú'®´ÖmVP<Ï~¶r/¤Mæ·~¸B]YËa¾¨¿øÇIáZ 4¹™Ï5ÈB­«lååyhÎÐÐq.ô†Ö¼ßÉ!)h¿ NÇçx:\þ$(•N;Ã`YZRèRå.KK‡n5"§G ÇÇ͵&PzûôÒÅRé ï’8u´- ØþÚ :y¾wæK’…¡IƒÓ…–0òe:·‚ÎƼ)œ÷P„ Å¡µä[0EN6Яøf8 ‹Tƒ`îÒ(.¸‹"CÖtXNìt}êË +W „ìßn"’‚µ|‡:l:%ßå¸ÜÙËQ}köÔìÜvk»±—£¢ÁßäÔ Hâ¹øƒêP8’Ý-‰“{Š5¹Ò;›H-dð¢Ê‰i#sb9ÕãÁà–¡…²2¦òßáòÈ@f›„¸â²?Ȫ;ðÈ¥_Ù=üêxT‘³õ¦HŒN 00UOñ)ƒ [”¯UÏjJª’ùGɈg·Q0”ÿ~Q? צ帉¢C÷|9H«P<êQ¢~c JfEr»2w£ÔCn!ç+”Q²¡•PFÙšäWém(Ï ìY3¨èXãB]¢„Ç¥‘dGí«C_õ,·®jüãPíu‹äñg¨á9o±úîj…©K_8ƒÚ#‚L'nèàiÍ¿k~a%¨—_}O<Ìψ¾×ÔäD×ÍËZËÒxÖüsò½¿QhDZ±9/"FÙ»ç Ò ±›wZˆßûÞt0Mþ’\ôÊ@ÓºïÁ–ôßÑì³|ÜΔš¿qé\õG‹`ŠðöX<÷N0ÜÖ-ó~†û·OÂìhÊ77®xÈz·ÚqÙ«Ÿ{¦½ÞŸ|¾´_€÷”¹KdÕÌõ³KÁ1ã=DöEÇúñC”BÏñ¶ûñ]Øb„½Âe·ìÂÌ!´ÇŽÆ™À ¡Œ]»ÍÕDÖréß–c0¼'%Ü »Z±Ù&BRuìtîŠÓ¤{’Ò¾| Yµ‚åé+{|[ð¹ÃT9½Q\$ì @0»ô§Hå "»®èÁ»ËˆkŒrÊÅ»¬|¢ çgVôŧÝõTªl[ÙÍÜß?‡ì¡Îí§¿@Žéß;oT 'žÿ•ÃÐ'ÏÐ?:vºr÷­¼Q˜†<ÙSÇm–ù¡Àã̇ØMC(6›’ÿIehNøß…Š6Âc½³®{|[Õ¼Z Õ‘[ƒ&P3–ÙÈ“u±¾ïé®-CC¹9ke/4•º±A‡V˜}MKù?»CÇ$noC'ç?òÍè"ÐÜ‘½Ç=FqbeŒ3Ðû“âöÂÌž¦¾ô¡€ÕL;ðc×õ¯Â-3^ÑâÿZcŒB–ªÂÄ™Cf$ “0õÚ§ýj^$̽1ZKúóKÙJ%aY¸2Ò?u~ý£¹0úïü͵ i…ÕÅízn[XOþųòCþ†›óŸ:[‰·HŠ|*a{TL£ë¶#ɬ:IYí qÖéƒRHúx0s÷&’'ž +-{ˆTt¦§:(~"Íû }ïW¤7YæGFæ íSÈô’kó„x²´Ÿ“2AÖÎÑÏ_B½;öî²Éä2g½‘´ÇEnLJû‡¿Ú!o߈º•Ë+äïÎj$ÚBÁï+?~‘Dû ö<£m}þzðJX¯µøê•¡­¯Æ¿¶(Ãφv(gœåLd2‹òea gûP©7‚Û§cÏ3Ux’'gDPÕé!rÉßFu‰æ_5ZÔØHb± @­^…þ{U¨høPHÅuÜ”(_~?€ºÜÌa-KU¨WœåÉZ¨‚×+]Ï–îåQˆÊëÈОå?žvs”d¼tŽuI¢±%yÄ´Ä'4‰bŸâšCÓ¿äÎlêÖh~«ã÷3žã.`æy„–‹S‹‡èà·raÁºóÿüŸ…ÖÈK±#:æ~ ÷9|æu*7‰¿ ñB³jÒ!I[‘H‘—±ƒ[yiíUÛmOjAŸŸ’ßuÇjÆAŒ—ìuðŽï¸ Ô™CÄD]¨gôß¾š=ê†ävÕªSû!ûRUÝÏS¿  ,´?÷2”M³žM¨€*ÓÜõ¨¤PÿVÛ_jC296©•¡·“³^CCkùrh04…Ú ó‰AÃð OÔ{±^smö…:c—ÁÀ@¨u;'sÇj)zõ,o@-³ZæO@íSâo—桨-âRAݯ|3-Chàõ’é›9qqûN9Có©¶Ì:ý h5èÉSYù íR'í¼I¨ CÎ*ÞþÙ%蓸'Û]”FËe|dÐMéuꈣ/ô¨ÔóÙâ…Þ¶­u.6èã?”ó¦Ç ¨„¼¯Kœ†ïמµ®ƒ¡»yÚLÎñ=L“üâ2Œi¾_x"ã» †ž‡Âdyú¶"k#L—ÿy¶âôf7¿ûôtØÂBÄù‹¯]°lr†?!Vhìí.2éÁRCՋŰ&{4ø¼¡Ô¸`"»ù¤µ^|×¢õµ—X3ªËµð´›yê´&ê­.8'·)àßÑ×Í.hXÆêí‡gïEÙ4C#>æŒëDhTyË|0|]µ ;çÑ„ãÝK¦ßÚhÒ"@¦ä¦‚¦×Y]z§ÑÌÐ:tqXÍyûýœ®× ù&¥&û h1|åñ’Fžë~ö{w¸f Ö ï€ÌœÙ’d!píX¹¤;+|Ì\Ô-—ƒ¨›ÏëéJÅwÛZIE"o+>ì ̣݉±[ô¤Â2Âm%øî´@è<7‘¹³>„ÒÞþ£à a UÜ…ß :ÛÅæŠÄ0Äú]y– I¢ÌÀï¸ é¶›Å1U1v7Rì+äè³;V=†œþ}…éM@°œh¢°B‘> ¹Ø*®»RÝç͆œtjÎ#…ŠÃ,å5½ç©çjÓ˜!»1íÑùúTÈ^!Ž´vÝë>Ô mS §Óš\Ÿõ'ûã ¹†\o÷ @Þí‘Ó )®5æ»Aq°LQÊBYꟼI¡¢?˜Zu,ªb¤õ®ÕBµœOØ~¨iÎðWUÞº[ŠÅÅ÷» Áô“ÌKÒWÐt¹¿n¸T Z–&…îOÐ@»#Ó‰ð¨5èXže qÓ„®Gö4’ ‡é–kÑ[è}b÷äà/~èS„‡/”e`Àý›sûGVøqµ|šd? §«5ëŠÁh7qýfbLpé<’ÁSçULënÁLÓKÇß0ŸÇb`^Ëd“¦F÷háW·+éle0üζŽi¾¡«/Yms”’aÝMWñÑ—;ð×t_:—…l9™ˆY¿Ž„íϺSå¬nHı¡â›y‰Ãòú{"©ËUÏégÜ_‡ÛZ¯)§2[r’u‘ææø‘GË_^,f¶þîÛØB¢Í?Èäë\ÙNoƒ,TŽÔœËÈZ2ëy£ËÙ‹ÞÕZ/š#—‚–ëÊ äVù¡)wíòf7õEþrùŸœÜ½O6>‡"º¾nI{à¡?3®î'¿£ÄIb+1ú³(E§9E9ƒ2tFöÞw&PNß\û×&Ê÷í-r4¢Ò ¥w+M/ªœò~÷Ò¥U¯’F°_<ˆêò§ßÞÓv@Mb ©/‘¿Qk0öüD=9j«ñèG‹± N e¨Ë°péèÊêÅ~NyºÇ?¿ÞGT½Ð²Â<rÐpË©`'¨ˆkÿ„zø£1÷6ÍíP49³{Æñãš¾ŸJjæôEs×ýdž£Eäù¡Ö>hi"Âvå¸=¬|áÕ4<99ÿV¾¼õŠ‚Tjï;£+ºx㇦埳÷…3-·>b¯tÅKqí@lfÊÓ¼Ç!¶æÑàg^ø”)sJœû&ĤŒ^9wfÞiÙÓ÷ Á; žhÙèEøârð¯A9$G¯w ]€lÖ%>Ö¾sP°ÿ‹èJU”…}1¤ý•m=¦?.B5ÏÎzœ–Ô²ü³Û´×·Z%3¸ÿzCƒÃ§h¾–Mh˜Vœ8â§ W=z3uÿÎôºü>µ]5A|*ÂP3³Ã¼—‡gÖŸNLAÍÛp'Éô4¨¥Q4Þ¼ µUôã :ô{œ,®“!‚ÜWü¾a“ÖWí;â1`»¥j+7Aq}EX¸””uý°|áú *Æœƒy6îñí²é%ž_PmŸ¤ËñÍjrÓÕíë¯@îÚ€ÛË\h`Sa ÕÒƒ&Þ õµÐòq‚ÿj9´+7©“/BG}ÙN"t<1Z>ÝíºÉù\ŽÐ+Q¬ÿW¾UÏ¿ºòúÇvÞÛ’‰Á÷a¶Þ•çC0ì ¿;ßÀhø$Ù•sËt3D˜’ øÑg 3aWí;TÒ`Þï¦ãk–XªgçSŽ×_iu±+þ‡á÷CÒ‰«f °j¶Tc¶ ëâ F¬wà/sIjC*l‰HŸÙI½ ÛîW‚R¾xÀîß5±ÿÃÞ›ÇcÙu}ã§!$d•Ddˆ2ÑZ†ÌóQ"CѬI#QQÆT¢P$ !ŠPÆHf*™Ê<Ó "¿Sçå¹îû~žç÷yÿxß?®óÓÙw­½ö^ß½÷±÷^ûplŽK ¹HuRàÙr!!c68.rÕ¸ûóƒÒ¿b20xŠŒÕK¯*Œ#Óôz¾Ôn\Z©Ê²CXÙœ3œGÙ}Ây½Ñº@\v\3è‡rÝ f.÷WÇÜGl¾ÁUË•<üyo#Ÿo(»I|®yp(vÅøU¬q„Ý« °¼d[r'жÒjk¥ŸÇõ’Kz›Þ|FÉÈ˱­äøFCµúCî^”ÓžikcD…&µÖṨÜv¤A`s n²¨ PîuBµ£Æ8B P˶]0ÊÁÍ‹¦ìÚQ«ƒùâ8«.ê$~d(Ü·õ™êT=Ñ€ª ƒ¿1 aÀí¶ŸÚÚh, }‘ñU=š4Žîµ± MBŠýå½Ñ¬†;f‚­ ÍÛ¾œd™r'ÏÞ§©«ÚÑJ£¼ÿÜp$Zß&©¯r‹C›åtõ47=Ñ6§PcØð)Œó^xû@Ø žØÆ»YDIë¶^M’d„{e厛å!N^gùFª3+Àk{Êb}mm EÒ!6ëþ“Ø‚ ¸›õúyÚ-ˆ®{ ~Šnx^¸ÐÒ¿ nÞEw®1uyå%ÁƒrÑEö‚ÇÛÀ­ž)>"µÜòV¬`ø"ý õ¾2Éßx ¯nìJx½›§"uU˜# Ô~[†ÿómPöí¼N s/”Ÿúè®Øì eÓ3#¯WBé¹Ëª»VäA‰…bŠóH>ïŽZìýŠ—X»‹{ÅAñò!ž{ 8p_Ócÿq(Ñf÷Œ´¹ ¥‹7É¿þe:½OÖ݇òú¼³IÓÛ¡âDïi¿#«áÏ·õƒ+"àíµhvcE¨®X›±é#Ô<›8Åí{jß²æÑ=†ºÐè¦~P_â›Þù‘¹¯¦@a:j{ðnkÀ1Ijø,ÉtTPZûaAónè`œqòê=èüèi[bŸ;ÚÓÌß@·<ú×ÊcлTuõ*^è_^="F³†<ŸFmÌdØ`D—]îáYôլ3e0ÁCs­ ª¾Š•¯ŠÞÊßÝÎ=ÿ¶¼~䯙 |S‡Øžïɇߣ~ ±×jµ•¦QÒp¬}þl ©™Ä:4:"½Ï#m¹=i¸¸Ê5$éÎ=d²0dÎÆ¥òƒÈq‹MÀF‰éÃndrùxä yŸ8Ã=f[Ê¢ÖâÔy\Q¦•{¼Wñ;G*}ÅÕMEÜ SáÈß+•'¨‰¶$ŠãÚ£¼Ê‘t'Q8Óâ³ÕA\·½ßä’V,®wq: Ó«Ž’Û.÷¿¾‚Ò¡YO󗣬³–ž~–=ÊŸMm¿nŠ­ñ©¿¼p#*mþ†ªÃoãŸCÜéæ˾ Õ $·jEMuÁ~Úm.¨å×(²ˆa=ê2~lÚ¨ŸyÇœñ²@¥Dµ 4]:Í=ì‹fËS™¥ØÑìí¹ÒKG*ÐÜ&kùÉ hÞꦻfÏ´pm{ë—.ƒc•ßb¢KÑòôú³­hµô&¿ßnZÅ&©ªõ@k¥gµM{4Ѻ–!ên|nñ¤uøñä'Ú,mO'Aw}ƒ"¯q8¤¼}Ðg£¶â3ûNóóA4(|Õx7²SV|<ç7â*î}¶Y7 ¼ ϽU€[lïÝéÊ&à†öç6ŸÎ(¸žxWÖã9„¯Tx“a¤Š¬æòa½Ï^á3D-­^Ç›À±K¾þÒdi€„—¶ßÔ›àßòþílÏ -ËiZ‡Ï™ßv)Y^dƒ'mÉ6¢íZ±ó¿}dd䥕éúNa‡ o,†'—‹ï6(„ôn‡¢ƒ  =æçÃÆÂ—^ZùbàZ?<árè9”Oò×}u½éÕCkN%@ÆhHŸ¢ÎuȬÉþI¾ßyʼLnÚäähâX)}ri$†uláÅ”vc–ÝR(Rç£ÿEŽo¶õ¾YE¶ãYË“¡8:ù÷y‹V(åÏø±âN$”Õ7÷ù]¤‡×5ªž©o÷Á³Ï«·ß`ƒj._]=Ps§ø×/½ÅP'°4À)± ê¯òõ=¹}ê—]÷¡r€&³›±wýàPíÀ.„Ó½ÅÕÜñÐ.¬²‹Çv:írƒYùásÊåÑîÐ=Y´cYŸ,ô9nÛ þäû7ïÊ=®0âç^¯Æ|…Õê“ç£ÙÃõìuà+oÀ!uøÖ?³÷ž1'LVŸÛ[àg7Ë‘;EF0-,¡} f^ýj^¶ ©ƒ¢XMFòºÑÅä}§¶ªÈÝÇ>Ú­É{x-2 <½yJ™jÂÞÈÞÐÅ¥1'?æ›A6Óµ%ÈÞêbrNX —]ûÉ–ÄG\~þV b¸ü{‚•ŒÉ]äøðú9òÙ üÔ{‰kB™Ã÷Ú¢àó—So,ÝP˜oý­"Qs-åä[-†ë—sñ :Š’ ýG‘PúÇcråÇPN-ŽÁ$~%*4œ£‰{„ÊŸº¥øhâ&‡õÞŸv‘Pí >ë˨¡§-`ÉŠ›Ù^î5Xõµº½ÿ¶¹Œ:I‚†¬t¿QÏþH»2-ê‰ã˜ø!ކ^aœLä}çÏÝá=:2h½Cîð¤š>~QËÖ&‹fwv˜¥)¡yìÀP¬³ZäTóME~@ËžäõÊE™h-y}õèJÜþHî@³ ÚòÙ{ís`†±Då2­÷’^Ÿ—um½<°‘ÿiVñ=ÃJvC¬x,Ó­=p§=Â!Õk7Üeð »wE$TßÜ;–#Ê©Y!ª*]<)|"•¨ßøA¤}ȉÌuþpGƒÑ ð’Ç {)T=¦*vªa¢.O¶ÚÈT7<,–똋š†ò—CP+´tƒ¦`.ê.µ\}|™ê'-¹ªÍ—ŒFæÑÔyNhòÙ̾E††|ÿæí¾d,ÍÂBªvGs6¹Íæ*Ìhëðhw,Zиì\’P…>T¿…òÐâ‡7·!•-Z>íÎñ-Ç•Û'ÄýÑÊÓòJæóB´|1jigÖ{|ê·8|Aë^]ž—ó¡Ûˆó\ù¥HѲìúõbÇä' ZÂíî‰Óe-\pãжǥpCÔ|"{ëZˆü$5Ü´·nä—º(¤ ×ß×ÑG†µAÄñï-çÌ Ìðl`«G„F§s\¬ÝÛN_Þú¢üvÇ,EˆÝõî¢($Lu;{`¤”í•v ÒIJ:7Ñ@º­¿"§­2¤k,{´³Ò+Êx¨á µì™e ^ åé’Ûo'¯¨‚´Öš½öQ/ ­/kù”9Ù®·Ÿ ÒŸŽËzž]O‡ò5“‰+É~“µ‘ÿ!oT+ 2ÒŽÞ¬ÛW-³;AöSÙ°§á¹Sá9áÞ‡ðÂΡS âoĬ籗«b#/)BÑšÛƒ/º:¡øäÇ’Õß ”Ï•Þçüc(›h]YÔ ¯)r0\™€7;ÝüÞ× ÕŒï†UÆ¡&@æÉçÕP·´÷pÔØôåý(4Œ½Ð[²! šüÞÔñŠûঠ½>hÙÞÔ°®ÜÚéj-‹û¢¡“yÿ³~Ëlø,;zåîVè¶)­‹€Þ¨Èmáƒ0`·‹[ÆP†Ï3{Ö¥ Ãè/…õ—½„ñ¢'Çv;X—`¯®×>ðmG«®ý’¯0©•WÒÛ6?Í´C6#L]qu¡ù3,N‡Ôª‘ŠÊ‡¡¥| RWÖm+´¯EÚO»y~ #½šU–¨ƒ*..R|¢2‰LžÎ*±Å_pé:Ž–Ñãȶˆ>k„j;²ûš}Ó—Çe’>ù—ƒkÝjù^'\ž˜½š;«y²©kJw¿@>ù=ê¡eѸÆO_9Ü’ J¹NÓÿDaêGnÏH+”KpCñÊž‚\¥û(é5q¢[x3JçŸ>½ÉQŽF·þªÁ=Tˆ<9‘4ÊI+ø©&ËpÓ×)Þ»³?Ÿ|e³õ7jhõ\Áe¸YX U¸ P[ÐB0K£u…?¥õ‹¡¾|YÀÛˆh¨\1–“ ··^ÜÌ€¦†æíÍthvŽþ(Ïe4¯»M•{¼-7;=éΈC«æuµéñ¸%ôÇÕ z–h{Êú¦SÚݬõñ3SÃm_ê<Ã~ãöÃÖIÁ˜ÌpAgछ|˜ð s…Ä‹ôeú‡ü žžii™nÜ]Ó`Ô&o 1 º‘¥”b¸C©Q€ɽ—œÝIž—"žÛ_eÃͪ8uÀuÒYÉÑ{p=䱟XP Äô.uZW‰¿5m¾’ïÓÓ §w*ˆï§·Oº¿ì¹yª‡mZx¥ Ðìq{T©?¼ ‘;!{/ŠOMÇ^{ ¥bÞƒ+hÏAY¼UÅ-—V(gFáÆñV(»©þ]ŽJusÏ8^•ƒ”=ï§›Ån!;7Ò‡âE.ÓÆ¡hJÍéè¦õPleÇ“àe Å_Œï ‘ß”/]š¡ỗ¤¡^r¼” Mó‡×xí%ëC¨äݰÚ..ª–È:õ»ZB5IÁ½f‰7TOœàï…šïT‡JÔ Žk¤dR© ê‹O?Þ Õ›ýK®@Ó^ÛpÕqxç%_Ûò$ >ü¢ž²x£:ûwººê5Æ C7 .P¡:§SœÜÂçë"‹^NTA·Ø¡îP°…žñlÃ/¯‹¡o@Tý>í0 *¯ï(È …áöÐú¬Á^þ$jµ®ÆïÞgÛ˜ _‚\¢´TwÁ·ÇÓ‰iqï`rØ-žoëüRú:,ag Ó9[+. íC’yÚÀ‹ÓHÍ4xáñN¤™æøe–û騦,å7%"ꂈ¾ÛŸ‘qß–(Úæzd¦/â̸²—ŽîóòúÚ„l÷ï?‰NëDŽà7/ÅÏR#ç³5õ«q9›ÝŽ›ŠËqePÐùµ«ÙWU$´ Øù¹ãÖ}ºdŒkò”åMü7 `bbØšk(Ô{¹Ôñ Н´9-^ÎJ³A‹þ­f/¶£¥z…÷…x´¼7¬ÆŒhEx3=[ Z¹üÐÍ\2ŒVÎsBW±UÕá’gü|©ÌÁ§ 6)sIƒpܾšÊ¨_y nHY,>óŠ"“ß«_±_‘‹Æ‚×3‡H_éÏ^€ˆ}…’é:®é´Ö7¢BO{Jú1?‡PÕH½s5ˆp¾–,þ½n“Ô’øn‡Øâ<–ë'!qmXóËsv²ÿõmÚúíÃX*ñ,Ò\åÅw@šFSl9¤É wÉÞ€4íLÿ¯©²¶úgPª¤Ö~¿qî—¤6i²kKÕAšì~¿5ÆVî|±ú±9N^8½Ç6žˆüh»§O²u÷ü W׆<Ðl„Lõ$Ž‘@È’`¦N‚ !é¬úÞðÆówðÆ;ðö÷6ÑmŠSP³?fo•\$ÔN:ñvÙ@}hQˆÂr5h\êö3r+4åRѤ>¢…weúlâ!Ðâ¡ùùü‰•ÐVU{®}Ü:ò…?—0…OÏ\ôNHêBWvñÕ¼ŸÐÓTûúL\2ôü™)Á C§ŸëÕ­p„ÑýÚÜù­y0®ø¼¤àÂøB«U¨ÁŸ _ó„ØËàû³ ¡¾Œð#×'¥çi)üêQü.[¨¿mÊ^”W_@Ò]Ã4”‘zÏR_¬ÑH{&‚u»{-Òuj†5U[âb ŸŠŸ9ÈÄ,›Ä6ƒ,é?ô~^õAÖçL«>K~EöõjWöìSCŽ7Qõ“\ÈY‘Áu3–—Û–uO¯–CžS˜ž0¼D>–^ÿUñ¸Æû…K¸j Þ¼]³þy uk ïÉDQ×ÍúRñ(~6‡q¬“¼ß\2µÇ2:¥}MT©‚P¶xøNÕ„*Ø’øžnˆÊ^†, Qõ×ÝÓ±»5Pm;•jÌD$jhø©7ÂÍ£9ÄYÓQ{džëû6è n\äU‹7Vh°tåÖ+^htÇÌÎTšM…ÃŽyçÅ¢Ù-¯o¹{ÄÑB®µnj³ Z~?!íÆu•¼?ܶêD6 Ú2»î !}G»}¿·ý¨ÄíT —_Áçé^_x'ƒN리ìB¶£3Ý‘ô÷0š—sZ½ Ò^ÿ”:> ™­Î7¯kBœð.™_¯«áÎòU»ÒÞD=7ÈÈMQ…¨Õ4 Ñ-Åz<Úâ¬ÜðÚk/m ‘U²R'K«!¼lÆWº´"Ê*¯Åï¿b ¥J©¨!ñ~syÿ÷jØæa t"×XáZÁEPP½f»$ ùx…¡EÅd¬Ç¸;Q\Ô{wæÍ(±5pPüJ•£q0ß.”qŸÌ{[ƒrT'¶ìü„ )?oú†’PYEØè¡Ó}T¥jl—bAd»`äó ÕMhܽïlFM§ôáë=ƒ¨•ò@9k+ê.g:›”ƒú!‹ótЈÃ.¢ÏFM<¤£å>|EÓàÅë ^þB³Õé-k÷ö“÷R·¦sšÑ¬}yP™Û}4WõºÄJ‹æ{ŸË&¡ùP¢ÂGk´P?³Iq7 ZÛgª|ÌG‹Òíyh) ~=§L-æ±?íCË\ß+bù¡kSÄJêJH–*l¢:è ±Îr·Ië^Àm%÷°ÊÜÏ9(Le¾S "¥¯îØÎ ×EÓeÏÜ[×Wæä\üX áÔ·Tüp…°%•®OAHº}͆”yïsæH­6Dì½[/V·½½ûï¶ù@œÀHõ‹o!Ñ*]Ÿ&ä6¤ð±‘OƒTŽíÉþ꺿L.bÚ÷~à9Ð 7NˆøÁc.OÁGb¹ðèÇ ‡£aðhhØ“œ%ð8f÷‡€·zjý„Î0HÓ—r×r|üÅUá"£é-;*¢ÄÀ“S´âf#ë c»‡›“úWÈøû8ÖÞ2_ø†IÑqÂÓÈôˆå+ÝáÙšîÝtÝ·"˽1à pè?•Ò`‡ÂvñIÚ}ð*ýD{õSòüE+ëM‰Pºªl­®^ ”KwÞÓT;&ëm³éÃÍ0±Ënð¶‡îï$;Ôle|šäµ7Oœ‚ú®î£VW¡q[²Ÿ4ÉÉm–#ÏÏ•Eñ̪ÐrâñÄÇõ Ðvd¯ÀV^-,±Wûà“ZÍÌî<èZ^òÂ4qôp0ŸX“¶ú®Zü^Ú ƒ'w Öž©‡‘ªäEZfž0–T´ŒûL=Ï}ä|5Œ¢*¿˜ ߥäîƒØø!’U~YúüÒXbGís¦ã$eøv("Iݾóðss¤¦7ð‘+Ú´¼¯h †O#ÝéƒÕ=p1û÷âÏ9ÇqÉCšågùô‘Eõé™åçu?ë¹wÛ¯ [„Xkˆ>rì¶éw¾œûô¾v.ÃåŒïwúžBè„\ÝëZ-Rÿ×ì• 2GÁKKTì=× Pyí@Þï(*ß8íÓŽâ —¬D©Q"a‘Í+®*”6X]¼ñB Êú‡y°Ü|Ž «ýʹ…Ê >ÓˇjPµ¼…†ªïÓìïwgö\‘A ³M sÇÍ&KzxãQû6Ãaî{¨Ço!õ5õ ôº‹ë›ÞGão¢>¿ ‡ÐŒ7'õÝ;u4ï¿ÊtçÍ!´ìÉ’jéuÆ-<¶Žlé›ÐöÊ_ÖçÞh¯wCÖDt =‚’´ÐÑÒêùVGwÜ©oUPŒ.¶Aãƒ÷¡[áû‘ÞU`T×&çk8y¿¶ÜÃ˹îÎ$õ?£‚X­"kI#Zˆ^UJÏPk7•ì$÷„€k)Lg™CäX@jÊÙ$ˆÜžãfÝo&Ò»Úó!,øÔû8×NÿÝw‹Óv'D§ÒNïjV…Ä(sáµVÜfÌ?BßE²;­²’ Ï¡&±ˆÁ m÷dÒ|5W—IÆQ\P¼3=ª;á>”L† >0J€2sQ“€¼:(‹ºÓq„ÙÊ®{Û¬…R–¡";Ú$(—´M¸{ Š]üó¨DJ¡hæðÕ i5(j ͪ­ï†b^‡˜-9P[wXµ¢J¨W?ái!cN±{‚À:(}÷à¤K:”Ó(ó¬y¨¯Ÿ;/­ØÐ•¾eOüŒ½ jŸÇ‡t!¨Æ/öU¹ÆPccÐdé[ µ7Õ²dL¡žf‘á Ï hðo?}}šô×ò†w¾ëx A‹^ã ÔQ9òüY|¹'åtøì¿Ýº>ñìõÑ¡ÏwÜÅêÃÈ÷o|Ý—¥R•¡§QÎ+ô˜ôå {_wï•’ÿÃÇ8¶ß†ÑhN™–K`\T7øæ¹301NÍ3¸ú|ô÷ø“¼)i¢ÚðSóèë»}0uNòê¢D#˜aá]| ©èêŒÒ#uN³¥fÜE¤õÕwR`Aº½+o‰ÞZ‹ Nl¼qcÈ8dþô½®;2G­÷*“kAV‡ž€’}uÈ.mß\F‹Ë´ÜjÂIJ‘ëòð‰£®¸|*Ûí"Íä¹syéC᥸:H~Gg:òi±;/‰ï³¹¯©ÞC!žWY› È+}Û]O QìEÛåÆ+¸¡·ŒI]oJù¾Ü&Ñ/2+ƒÜƒšPö7ŸåºDvTÐu¿´¦Ø•>+oºÛÕŽ*%kU¯‰Bø¦É{‘ÕMõ­óPÓ£óf9× Ôzv§ìCÝU»ž8t¢þE#íÉ–l4œx¶c†1M¤ï¾èFÓ5þ·Ä£ÑôÓËOÔ»ôÑlã£Òºc€f—-î[¹&¢Yó7Þ›þwÐ\àNåŠgdt+кÂf…æ G£x9òм}˜o‡Â{´àXóÌ4U-T;Ü}âÑÂ~XÚ‹—-öû¸.Û`Ÿ3]ª Þ?t⪤ÄòfHˆ¯Ÿ€¨Žu|Æ)†™0ùc-͸žÙûQn3DhhPAÞ7Úú0lä„Јۧ»>€z…„¦îÒ?Ð`ôB¼³†Í â@úÔéB¸=š&.¨ q§/¬H€D¿·Ï7ǶArÑI“‡×iàñRõ áѯj×Ç1; Å.ìTK$31\àÔ(ƒ‡1å½¾²ÃÃÌü½wƒò!YÎBÀñÄKH¹fX “öV¼_” ©¯§L3ß\„´¨Ü;_Cúµ²]6©ðD=¿ÿéKÈØ¥Ê£6û÷¾nÜ ¾™&¤'‘ºðtñnî]tsöئâZÈõþÝ¡™ ùáÚ~ÎImPxçˆWBa¼:Þÿþ‘s{­k„Ržc2)äþ*7íúÔäcÇ„§—Ÿ‡7,¬¯Z$àíÛÇù­Ùë FÃûÃ]~a¨}wpã»5P¹{cvH4ê?ß–žžÍÂnÎMÏÎÁ{ùÁB…ánh¹´ŠF¡õ´™IÈ_–¦†`%Ÿ”³ÐY ™6ká³w©¿sHt늿̇Þ{|ã;ÇÈóò×tM¯6 ?«øÀØå c¼•ŽûªŽÃxãÏ7 ðååñÖ3Záðížgï[%˜ î’x¨?£Z_Ô¨2ÂÔPø½C#0ó22ú®¤Š¿´Ï›¼ï|6òCÐk=Ò ó™+¤û"C¶#Ó´· .Ñ­õ5ø†Ì%‡³äßø!«È²ÞªÀµÈvœ§Ü<Ë 9±x¦f-r²¶5Í0!÷ó~ïÒÁäa0=wI»Wvqk¶ÝÏ gþ‰‚Ǽï]È@¡‡46­‹(ºh÷­Ãš¼(6`³MQ%4Xt¤¯²£4ëð.e1”5بl¬¡òÞ[g úQi*m ~—5ªFrù9î¹€j*S’’kß †JEÁËK¸yç±çKh¯£ö»NŸVjÔ»Í\{=4 ó¦èSZÐTh¦ÆZÎÍ—µönüÒ†–Ê)¾Ryh}300ïÊq´EÅk¹Fh/e6tïÅ't8Ú^»óömt Þ8ìâ€;cÄÆ÷u¬B—aõ†6¸+`͵SÙ—pï)Ü È%A"ÑR“HÔ3$‰™ü¥!÷‘¿ä4’àœfÖÆAØX ûbò—Jj^ždæÙyAn^ŸæÅyAi^P&ji)Š$M‘d(’,E’£HòI")R$%ŠDá¡pÈP8d(2 ‡ …C†Â!Cá¡pÈP8d)²Y ‡,…C–Â!Ká¥pÈR8d)²9 ‡…CŽÂ!Gá£pÈQ8ä(r9 ‡…CžÂ!Oá§pÈS8ä)òy ‡<…CžÂ!OáP p(P8(  ‡…C¡@áP p(P8)ŠE ‡"…C‘¡HáP¤p(R8)Š% ‡…C‰Â¡DáP¢p(Q8”(J% ‡…C™Â¡LáP¦p(S8”)Êe ‡2…C™Â¡<ÏA#-%õ—(ý—(ó—(û—(÷—(ÿ—¨ð—¨ø—¨ô—8Ë6»¦PI´Žî.„LçêBh³9çsH+S®â¬,½@–Y Ë.åÈò d…²âYi¬ü—¬´€Wi¯Ò^¥¼J x•ð*-àUZÀ«´€Wi¯ò^å¼Ê x•ð*/àU^À«¼€Wy¯ò^Êõ§•‘¢ €YYz,³@–] ËÍE ÚâšÑÌ_hG·ù‹JsÐÅ{¾Ä1Çs%Í–`™5«‘æ>8:‡j¬DŸ|t cÑlþ}óVµ#õsèG`0ßFzTÀÞ%Òã~œùƒ÷ ûƒGsz2‘ïñÎ9=ÈŸ~åOÕÔžå2] þ`Q>[}ΞCøyÎ3gÏ%üå}›³ç~ víøƒ…„ÿ†;ð%ÁóJ£} ¾¢Tþ?XLð–ðΕ/!øK/Í•/#êQö}®|9QŸ×.så+ˆzUÔΕ¯$ê÷çÊ¿!êY•Ð’×Ë™9}Ãìzñ‹¨3‰$1«SÏÉ’„,µ Mz,³@–] Ë-åÈ „¼a>í³ëÕüzFÔå÷BÛüÚõ÷ôÿPæY7ÅÿKuSZÀ«ü¿®ƒ‰ô•çoé ¿„ß 8Tfþ«m~,©.ȳé?Œ«ÿý˜QcøOmÿß_Ûãã÷ÜÜ€ù(MEš‹ÇŒÒÊ R’RÒ’býßÿ] ‘˜ej3ÿo¶òß]Eõùò [ÎÆÏOÞ¡ðpÜçærßõ€§Ç¿S–‰7,ÿîúçÿäAôßz ›ÝdüwþÙ]ñÆÿÑ?ÃëŸnwý?ûaýoýPý»*rð;;rá÷tåwÙwèÀ±ÿУªÿÇ=ªúÿ§GÿÔtÁøøûø_|€\éCŽNn9d¤d ö®ÌD5é_FÙÉNuŒ¨™3ô í×\ÚJyÙðüËKç_z:ÿ²Ó¿¿õFÎ)KU}ÈÑ0»nâP—½cÅ+Š}>}¾üüËŠÿþRâÔ€ßÉ;®Aê2û߯+¯PÒç_–šX•Q÷s(õ›·§¸N¸y|¢¼¼uþe®ó/M÷;ÿÒãyóúüËOç_ªš5ùµòKPåå¨ÏƒÏZSyxÁ‹rѵñ ‡à¥ðÝqžÖP¤´1òÓ#7(Þî­zF—Šo˜ÕÐÿø¯?Ñ—_wLyÉñ[ºäc3kµþzI*—_Yì–ûP—°Y÷uD248ÒÚV»DA3ÝiÖÐÑ#гƒÔu} ZtÝ™Ù9ï—W]£Ðæ|üðÞthǤ{"ñtº (¥è4@OÄÇS’¬×¡ëPÛ—{Ùеjkë04¢czFÜF¸jJø/ˆÃXfµÎ¥-:0Ú]~qéKð5pÊ£ýF|;C0cƒÉûÙ/&SBà§ÑööÔC0mìx"£fŽ”Ý42)BêÖ|³Ë/‘Jp´þ«ÐA¤«®‘¨ŠÀÅE< ƒ édlz’ñsdʵo­énFÖ3ÞÛb©*‘m¿ƒi— ² s¯úT°—M\]~í~=roøèn÷áò˜Ì®4FÞ>—Ç+rÖ£À6¿»¼*¢¸VêÕzý §P¨²¸7ý˜1®ÛðÎ@n[ŠÞ[Ð×7DJˆhÚ®A©üÕž¶³P6bòrnK"Ê·%ð°·Š¢Rÿ»M¿½‡Qµ) .^J Õ¼ïw­EÍÛ¯:޽Í@¶Áû{ü¢ÎNYé P{úĹ #[Ð0¨T`ÝÖ‡hš–m¹¦ê%šŸpÕÜ-ò¯“HcfÁD£ùOóÑu÷!òÿ³÷ƒKç§Úü @¹$ôùåp~9S[FàrWÈO k &P”@1×(A ÒÊ(K ò*¨H Ên$P…@U7"jóxfÕ ]Ð5]ƒÐ5 ]“Ð7úfB×"t-B×&tmB×!tB×%t]B×#t=B×'t}B7 tB7$tCB7"t#B7&tcB7!tB7%tS5çÓ æÐŒ°›v3ÂnFØÍ »9a7'ìæ„Ý‚°[v ÂnAØ- »%a·$ì–„ÝŠ°[v+ÂnEØ­ »5a·&ìÖ„} aßBØ·ö-„݆°ÛvÂn3g'ó;UénÜØƒ†®¿…ŸVû(G1ágÞÖž_žb—7‹¤M}¹ð[:‰Ø¿©|Œ?Ø ,έ;ƒƒV*—×ø^Æ!X#·Ø£‡Ù/ôêÇᇕQip¸UÇx¿Ÿý¦Úê‹CGcŽ<ãfÂ\÷Âi±TìKí鹨†=Ï6Ö…îÅn™÷ÆÜصÃÄj™víí7v»Í‹u|‚g°gÌÏHø¡ö¯û¹Wüм(Çvøö[N:õ¬˜GŨµÚ–>ÁÑ"ΗúÒ¢8&XU\гÇDS†ý·ºáèÇËßè¸uqÔÁRåô÷Ï8’õeÙ¹o8üëÆÍýWrpøŠm]× 6žP8po¯º;>,†C“=ÓÞ­–8Tqòh¶ ÅgˆHf§Ùa5TF¡b$¤?Æå•¶ T7x6áàÁ3üË„qðúz¥¯g8h_ô¬½TšÑ-­wÓÎ;8è¥É~r59_ì&O¯-pð§ÈÛ•a£8ä–Þyõ¹ßÏ<û¼‡Å¬N*ìÀaÝÜr'‡×WçØŽCcëèN*ËâPVÜ Ë 8tÇèñ^*Õ_úîV?ó–™í8ÆÃ!wU.^ÉÂnŸdÙ¨«8rwËÖgqñ8ÊÚ(²ØdGuŸyßWMÁQ?FÁ€ß‚8úá2¯yš8âR¤wŠw'Û$W0Ù†#NIÔz“Ü8 ÿ3Uþz8þø‘¼[$‡”ä¾PãpÍÆ™Ð2¾Æ-ÆeÃ#Tzoª¦•q¤ç¢“AË0ŽžÌNÙcÞKXOóáðõ>’úZó(LkóÚ#"Æ5‡-épìSo íâã8qà~pC/~Y™!úì‰D½ð¹Õ?÷)jóxfÿ¹O!°`ÿ¹Où'®üWþ‰+ÿc\ùç>åüÿçû”Ù³üýdÕüa¤?kþGIs?mž›VëÞnÛþ֮߷Q/ß_?Ø`ƵÝËï_Ôš¼µËã¢sûüIJ¶õ©ñ*G±œÿuBs|,–mÒµZTS…eË<­ òtv§°lå†=õ<>Xâ¦ú¢Ñ™Œ•I†Jw°tòÝ¢ìA²þn]™ªY¿µd ƒ(YÒ<º¢Œç¿èŽ"§Y_c²y6Ý×tYYw) ¿JÖõusoÈaYðÇçÈ,[~3¼¦&xV?ÁyxÖÎ.ßö`6¿ŸäC«Ùò ÜKÖËrXJ‘õÓq²—Åÿ ø­é?é¤ð;ò-Z:9[Î7b­Ñ¬Ÿ¥q²,³~=uÈË™çìÒøÕsºáÊ?öØ6ñ?ùÃâÊþ”'9eÏú“ùjŸôÇ?·U,çÒíˆ|;VΕ»]=ç§¾~Î/ìžã‰ #tžGäÏ!Ê/!ü='ü;èJ¤ù†çÊÝqŸóó>oίn ÑžÅD{çìý—çòÿúü§ÞÊ?ýíû¼`rÖ«ÊbáY¿ûßU<šå 8ijV?ÿ¸a¶^ú ñT³ùó‘«y¶ü‰þó³þNx.ºAÖÏD¾Î™f‹Èéç[ZrÙÉú“Ì¥rýd=Mœ+g¶\îé"Y2:ÄjaÙ ásñ.uXº‡ûäIrÿÝðêãPìÇ2¶á}Wbs±¬±"嘺–ù»éïw&ç#Λù¿ó‘ǵ¨cc–s lçhÅŠ7÷÷8aÍïp½ƒ"‹#Ü%Í?½ Hª}"ç]BÑû’¬%‡J¾’<æZí™DÑåH/—œš4§äoµ©~I)½Y›…õ]É·MóÇÒiˆyÌü¦ò¢Ý;Ì>"”ƒ^³‡ çî>är`nVÿyX´ˆ`ÙLú>xáqãó|;Œ¡/=ÉÇ€ñ„þ˜Ð3ý©VÍ¡mÜÚ˜»¿tùæö—+>Têºyb‘¯øg=ã–S,»VÕꤩ‚Âåéî·D°’(÷–(W½Î!EíÞiœõJvµù[ÏêOïÁ:çw F¡.XOøk üÍj[Ï`Ó€)¹¤:6þßý¡—ÆÏœ*€¹ÿTg3 `Æ6‚·àm?É7›‚knçùÅŽ?Õ;€D{:‰ú|"êóéiä,#~þSÝmø™¨_—ÕŸvõì6 %—0Æî öYOØó§˜&öõî%êÝû)HéS2ö0 u¯•#oÊ/Çö>“ÄþYÖEëɛĹö íüS]>,ò;2Á¾‡ˆö åþIÀa¢Ã™ÔáäªàÈŸê/ú_üpkþ¤ÚüÈ\0<æFò½aQûëÁЉ!Ã6©hìôSÐNJ9ŸÂ¥4‡HaWË-!ìõQón?öUoÉÖaŽÅÜÇáõ þ8ø…Ã6u»jZ ŸpÄ™Øû¬‡åÌlÞö8ã€Kì!mìq¿'±o)vIž‰X"‹]6«×\?ñ{XÄìÖˆ¥bÿÀýÉÔºW8ÂpjíêO,8šøòb¥„$ŽÙØÝ í%o0©OK[=ÿNÞˆ˲§2àð㘟Vä>²9üåØÍ×8¼"(àÝÑ—8Tv^·ñC<ùYí¶Ô#÷­üµå ]8˜Qb›Ëi†ƒ'ÇÎûsãà*“˜ !8ðúçÔÅ6ì/Lþv®òöï£Ïßp€ûm§—…¬3ǾNƒâ{Ú°¯ íz‰döU8‰¹ô `ßк´ËñKÈùtÎkœÜ…ýMv?&%à€Íùé–Š`xy„*ü5rœ¼öCNãœf8pð:‹ß'YôFèÄ!ßï79|lp¨Šct²w‡wn?r·G4Õxn¬òÃÑJA;6“ûϵ‹Q]þ2ŽŽðè2: âXÿ™3œÕ8±üˇ âÁ=8ø/«Ó¢ù5hï>Ïùʼnzß^B¢ñØ=ÿÛ4‹þœ>šÏ½ÓÓÅõ?Sa;èáéyh—ÄA/÷Ýû\$þ<Ÿ?ßHÉÄü_2Íy’šù[&ªùÃ?Ø©ùkpß­ 'ó' ­ÿÍ­¿99t™«À|Ûÿtèç/N[ÒÎûàŸÚÙ•ôW ù[ )]¿”ôßîñþôÿ1JÿÓ:;rœOvw9Bˆô;H,ˆ+ ^.ûê´ ä¹ZýI]|èÜ=œœç=Óí&[<’` 9º»ÿ‡+þ÷‹¹°O6“3S“{‚Ö™bW9[ñBåfe5¯ëðêbÚ¢Áv{((Ž4¸£ÏîKÖ¸HÖA¶mŸá¢ÄÆaN2ÑMÏ€ E®±°\cH;¬1<ÚSA9ì3h>=GKñ _ã$å°NY„;§ìæÃPUQ™Åüü2ÔŠzoï9DõÏí]Éôrè¦É*ßá‡:¼Ó¿¦r;UÞýÚk.C ÍãÏfœ‡Æ=+l•Cñ`æëìóP}E7lÅõÕðF2=L&è¥]óúüáŸê}Í´jך¡FuÈ"ý ÔT æ=gÎ…Z­~ÖÛçÊ ö[xuñb_J}Ùr­ésšoT^N¬‚×[EB¶—Bëpp2Oht¨Á3W}~ø´ÙºLÁĺB¿Ø%C¯²r%Õ×"à9´5;¼†•EÏÔöiÁè³}‘ZaÜ¡_0…n¾‚¸JÝVø:¬Î4À 0I“X÷püôÏ”Ÿ¸ÓI0eút·Š=¤Í©cõL46#mò3Ë+×é.‘óIHŒ22œ;ý+±Ì Í ÷"“˜„—bž2²@ÏDÈ}dµù-Ì×l ËtxœV G]{†¬,r^‹¿¤ü‚¹ÓóC8_îÁïõÃu#Ïçav¦Œ>äm=KÚõùr.=¥I¨@þ^î+†N( X;u!ÇÃG3ÕÔQ˜Y9#Î=Eݨ„ÌÃÌQ¬òá3nܰñàŠ-g|P2Îs¿¹Ë”f± nŽF™c? »V ßÁ”ðž”oÐòIFÅ»÷÷NrAå]+Næ$í@•èÕû!i~ÅM¨VÛr¦§ 5H~çI·Ö¢¦ ûËoÆÍïDº¬¢åQû¤X^Û»(ÔÅ»´²W¦PŸ»z»ž3´ÊŸŒC#÷ ÖýQ/иiQû—=hÊE²{xMŸúõ[ó¡™I˜uœ¥šµn•ÒZæˆæ;›cÎGóv–U§lÐÂâÊë!V´xáÖ{:¨ -×p%„ç^@Ë“+ÅÌ/Ç åÛ7Ó Ö.hµ*ì­[¼ Zí¤Ë¤}µü ƒ¶)¼2l®¬Ì3…“h ‡/ðô³Æi©ˆ Hëï:ßm”y38.ùDúu!#x3±íž»åÐݽ5Ï£d!^ÁWÂî05Ügšj0” Æ›÷3?_ççsÞ.‘éU ðÂâ_¬¯ä}¬u{_¹'ë—nºyréÅ®,1Ïyš=øN—ÿ…Äá;J>)þˆCy— ;¢Ì—AaêÄápïJ{‹R]/^®¥Î+Éýùjm!”&ý®ãVX åÚ3Q*¢i”Ãz”CzÄ|­]BWàËE uEY_ËŸüµNô-ޤ»? ï-r·Žr„@˳u{ß7CÛ{çÃeGÇ¡ÃýØÉð¥w¡sà½Uźqøìßàn©oÝr+Ÿ¼„^ž ãŸUŠÐÏ¿GÛ¨€=Ř›–ÊÃã)[·“0:,ÁétÇÆO±?íñà¯ùÚ¼øòzy:øžôÜøAY3üÈYbUSÞ SŒ~:fB ð;€éw¤g’BFÜ¥VGjº}ÏWE Mäבcâ¸hOOؤuÒ§œÞÅTý½@üf÷#d¾Ë§Q¸â²ú­Ò£s`Gv+¹Ò4º×¸Ì^çG®r=*RöNÑÄ+Ç(ã*EÎ"FÑÏÈgñ"íq›µ¢e) n’z^'ƒB×6Çi¥^@QÖó§%9¡XcpdVn¨f͹î;„RÇÂ÷]›D™õ Âô‹“QÎ6ðiR±)* <`¹wÈ•C¦T”óÓQ•Kï©Ç%„ó"N]²Gµ[‡âÓcÆPCñÓXŸ@jNÖ›„:F V¹hk²Lꄪj«J+¢ž)—­C¨ê7 &Ç+Ä¢¡|$M¦œuy$ËˈÆ\O—½F“ò™Ñã6hê~ˆñîjv4Ø`så×K4s´4¸×vÍÞ®e–ÐL@s¹ö£ÙhRï«wSÍûβ^IùÊêú§êÙÐâÜQ¡ågö ©wÿ«ÓZuð²I´Êúœe¼?“ºx®–Y„r(u~žÌv_óóm~žÍ–Çyû|œ÷?µÌîÚ¥‚ªP¹ØÎíÅ£¿âO麡—Pâ^»ƒh…ÔV¨›‰ (5Ó£ŒïÚw‚¹žÆBM¸:DIQâÛ[õÙ?ÅéU<æ :àÍž`½ÌdrÿR¿È4*OsÛyö Sâß|ú|þ·ã˜ùnA õ’°Ì’Ü©G}ª çKö…Ž ¾®´Ndk‚ï\—zÏ?wƒœuvùbðKÖJ`çö˜ö]mÙ½I<3EN’‘*wÆF6¤¹³›ãö"!\”VúP¸Né‡h¨iÅJ‘ñ„wý«vgdÞvó•g*/²jxm`¿èl“­µ¦ ÉÈñJ5R#Ì9Ã<¾Ó½ÜòI¶q…LÁγºÈ³vå ß2 äÕ¾/B“‰|IÆùoâš ‡èìŸH ÀãüßÉF ¸Ö:yQ49î d+p‰FÑ5Š*â¶£˜åäšÏp}î×V?‰ (é´)¿õW6Jó_n{#U2÷ E؆fPNÍö¦¤Ý=”ÿvʤ/¤•¸î{ï ÂG^{UO¢jï›§³ù­­½Þ-Gµ)ûU­QÃòÖ@IÖÔüÅ÷‹¹Zµ^/Îs+®E„ ã—«QïÈ©ñE¿Ñ@:ýqŸï)4,YQõ†~ëoÊXMO‡&y¢ï”›ëÑ4h"03ŸÍ䇬G³ºM÷>/2Es×rÆ£Ñh>úá]p+ZìéYB·-úx„ì…2ÐrK·âÍ-hùbå3‘?Ïbw,•’8M9ô~ÛÕÖ^ò™6åú|¼›/??oÓwø»üòŒ¥ÌÇÌÈE´"ÕŽ”ÃãO÷Ýš9³]²Ÿý;þ};(}¾óø/~‰ø=¿Ï“óqu>¾t<¶G=þ#äpßõ‹ÿæãÞß׉ùý)e?úæ¥×K_kh|y”ñ”·eüÁ~øÆ¤=|ìÒ· Õ}í ]–ÙOîCGÀ7 ŸkMÏl± )Å}o@‘FBº»¤»$E¤‘P A¤îT)»E,,¾ÿ¬õÌ·Ç÷x®ï»Îñõ.oŸyVÎZ3¿û‰™=+ùŠ•ù‡ê•îKþyÈåì÷v;‘ü[´“3QÒl8–q¼*ѳ¾½µhÑsnóa±Eb|¨•(Ö©‘X»,T/8G3äþìû±’”·C¦·®”´ïAf~!éoºwùb­¢døê؎ϦD%ã9•q¿dÎòëåÚw•‘[ææ‘­g3ɺèìÐòíæJŽ››÷½¡rÉå¶ÜufbmÉsÇ-ç´«“%o·ƒWóìž"wi}­›ª_{NW[tªÊsò@Ï'’ïÍdI7ÍÉ9uúDÏþů½¥HÖÏslÊÒVŠõö Ë^W~DÊ~Úw³$wœóë¦f5$åÓº J­^)¥f\>·]­~Ü]Ñy—”½¹B±ßæ5’Ç^{}N™ïÊJùr»¦N<_?x釲ƒ‡KÅr½rÕÜOžXqá×±ç[É“v¥n•î¯*•óîªpÝRyKñfúá³R¥ò´ýGÎ/‘*'7¬}ûp©Z£ß룆­—ª{Æ¿z÷ÉuR­hî£ó 'IµËsfÐw'ÿ°nlýümÒ'¼íXàìÏú‰ó+<^ê¾R'Ú±ÀWðG"A™úBÁ~ýf†4ëËÔiŸ¼sȺ ]üÖáÇv–ŽlNÞÓî‰ßýp4yï~ý«åk™í±ãg$ݵöJµO{ŸÚVsõwëÏÛýÃäþ‡"«eâõT¿pï¤Ù'œ-ÑÎýă}ÛvÛóò·‰ö51®SíèÑ‹w½ží‹‚ GÇOs¾jñxòÉ[·:Ñ%91¾;Ó©P«3Ië“ßÙ<¡Û]¿¶J>7ðÛSgí¦ÉžÞPð®#“ßË’¥ÉÓCF'¿á©µú¾˜üá[)·<×eYòÇŸ>ýiÉ"i“?t"©É„ºÉ_v u«'_Òë¾›b´O¾]œ«rùÉßÎÕju­É?̯Z)yk$ù§ê»k4É·+ù×*yfwiaJšG/8C­ýùüà‡KÞ\'ÚËSs¼Ú(­Ó^šžîÑVb¼=ã‹Ë†\çýr˹&ÆE¹q߯YÒ_–ôÝ??õÌ’rÓ#=u¡¸dz¤Nù%w•Ìý¦Ôý湊rËå.Ohz¯düÆ”úÛŸ¼<áø¾)’3ý g ›)¹F_Öæ¯#yòýòÅo«t¹ãÊŠÑ×u|·UMûRņr÷´‡ï»ïù>r_ÿä7GÚJþzMªïN×F ÞxÍlCGK¡ çç¾û)Üô䃾.Ek–|ùƒ†¾ù±SG›J §Ù¾—JòÌÇV¸3ËHÊò›³µ÷ )U|r¾úÕ—Ò—2Õ=zgI)³þá¦_>\Q{fø†Þ KùäÝ Ô3¥‚¾0÷ »ŒTü ÷ñì#/J¥EÝõ”¶å¤²üºvãñRå‘¡^Y§±T}bÍÔí/JµÎïÖíUã¸T_¶÷»·Ë”—šú]ßMß¿GjuÛ÷Þ7ã†Kmë¾:dž'u6VÏ5oü©÷|äÂ'KkIý…™³ì¸Yv™6â¶§“¿¹7÷/i4Mäû`¾#h¿‚v+p¸JýGZÁ_M.R¢òéýÃepÐî¥nç‚ñ`0N\Uá`úa5ª&þX*h÷R·WÁíÁã‚ç­Ôí™èç-лfö¯îO>n/i¸£Kº„Ï`èí#¢3¾—|vÕÌ Íži‘|þûÅ&×ø)ùâºG&Üx:ùýÕ7/~¥iák.¯ä¼çË”‘ÉŸuMŸýÊìû¯¹ô¶ô¿ü¡$_NßÕ{å…¯¹¬]¼V'+$_iѻͯ÷dOþ¹Eg诒mÑúÛ {%_í_ä­Æ4_]œ/ír‰¶©uè‹…ãE¯p(sÛìÝÄÖ÷«gJ‰“tlá¬)Š·ihÞ»6—´#weh;´£¤[Þ»±v¨»d¸óÊÐ|?¬•›VÔþªc9W2µ}E~íܨs£~¼=In-¼iý†OjH¶6Ú•iÖXÉÑ¿x·RJJ®¶Ö™m«ºKžI—Ƽ—÷ÉûËù¦'k\’»^É0hÎ'óåÞ__ùb•½VòתûȤ½ã¤`Íwº×¹F¯ô¥K}¤È‡ß?ÿÑÇÅ¥øš1Sï}$”øbIÖGÛ—”»õ¢Tèyó[·ŸÞ'O´þî·K^'G<~ÇP©üi¤â¦rßHÕ»Íê¯>ò¾T^rúìô’ýÎר'5oLººõFyªûñ™ZŒ“:Ý–¿™ÑQê-ÍÓñÖ¶»¥Á—´[´*ü%³ð—ÌÂ_2 É,ü%³ð—Ìþ‘¿d^)8¼Rpx¥àðJÁá•‚Ã+ÿÿëJÁÆÏÕ)sÝ㬕ÄFbgü¾²Á}³+È]¿þW¤4Rf½;½(eVüd¥”ÙYT™¢ÊÊ©Î@ª<·}8ÇÎ1…sLÿ½9¦ß.wýE„Ú_»½üõ·G¯ÝþøÕëNç~ó<9kVa„ñ_‹Åa„Æ-V†Fa„Æ?06†Fa„Æ?0v†Fa„Æ?/‚ãp±Áñ:ÿÜ]§3ý‹tøWæPËðåÇ#’Ô1¾´êJ þ5ÒÒÄAÒ¿®n&âf"3áÔ¿…¸•¸ÈJd#²þOìå r¹âç-¤ñ¯·{‘—¸“ÈGÜEø×Û½‡ð¯·{á_o÷¿ޮ­Ý‚„]ÿº%þµuýëêú×Ôõ¯§ë_K׿&‹ Ý„íÜd¿f®²\ÿº¸þµpý‹Wù×¼-Kø×¶-Gø×°õ¯[ë_«Ö¿>­MZÿ:´þµgý“ükÊV%ükÃú×õ¯ñê_ÇÕ¿V«=VÿÊ+uˆºD=âi¢>Ñ€hHøW1iL4I¿‚±-’æD ¢%ÑŠhMø—·iK´ ·[¸ÝÂín·p»…Û-Ünáv ·[¸ÝÂín·p»…Û-Ünáv ·[¸ÝÂín·p»…Û-Ünáv ·Ûßv»¥SÇáb×Ïo.„ áÂ?t¡y¸.„ ÿÐ…¿I’ Â…p!l»Ã…p!\ø‹ “$.„ áBØv‡ áB¸ðþ&I&\Â…°íÂ…pá/.üM’L¸.„ aÛ.„ áÂ_\ø›$™p!\¶;\Â…¿¸ð7I2áB¸.„mw¸.„ qáo’dÂ…p!\Ûîp!\þýÂõçmôhÙªk³°VÂJX +a%¬„•°VÂJX +a%¬„•°VÂJX +aå/VºvèÚ¨mX +a%¬„•°VÂJX +aå•ÿ}o=¬„•°VÂJX +a%¬„•¿gåß[+a%¬„•°VÂJX +aåïYùß÷ÖÃJX +a%¬„•°VÂJXù{Vþ÷½õ°VÂJX +a%¬„•°Vþž•ÿ}o=¬„•°VÂJX +a%¬„•¿]%­e™±ã©¦er/õ®~f=Þ#›Ôºå‡÷ œ.OvK÷Dl­¤üølÉÇïü@Ê}Tk⸤YòØ´}ůž¸Qʾ{Í”ûrK™9} ß=¤˜”©ø\óûÉ×Êoë{•Ò¿#}¶<]î&+¥Â°ö¥_êµQ*«×âÜ{wKµHÓ!i3ß*5Ôókµ;sàË‘§&uÝØºÿORûá‡Z¾ú|I©­Þ¿öº´K>ür͵²Ù=Gßlt“<¥Öë)yöò—jJ-µ~Áë%Ê­›^LÿÂyy*Sì Ï«ÓðXÍû«Rï©;êjuKýýÏMÍ) Ïçʘ÷·ÎÒxð„R§¶-¦Í>žxG ÒüéâSô±´TëÙúòÜŠŸ—|WÚM\ó›½ìgéè?ºúYéôöž©Ï.zZºL.Ð{_¥+Ò­í ªWX)=Ôúôšîÿ{^ú¼¿¿ëÄ^¥ÿCý^zrOiÐbLšjOÈÀcZ=Ð~° ~¦ØŽÒm&ÈÐzU?ºonnÞ¤~×»fŽ‘êùcôRCÎ( ãæ×.üËÃOÊ„UÏNlz[!™xç¡|sÏ­“Ikß_Ûç¯eJ—OWµ˜sE¦5ûuÔ™ußÉôáòÓÖ#³dÆÁco8S^–YM¿]Aæô²ú"s¨¢¯rRæ¾³Å7’³¯­òÙÐÛeáùŠÆÄtÅdñÙ»ßÌuWyùþöö+ÙU–¾“½W‹=ŸÊŠÖGÞyµlYiyc~þ¥™¬Úœëƒ³™öÈêá“Þöý=²¦féèk7¶u÷õ(0~‡!r÷|9iÄÙTiü oï‡òÆöz¶­«$[˼þfwï{Ù¶§Ö¢aÛî”ímzÏ{zloÙ™gdÅGû]_ñÈÐÆÊî3}¼â·ä•½ïÞ—.mÙÿÛ™æOCi[ÿr ëSr8ó+Û¿±NŽÜêohWŽÞöØÜ7’c·­h°lcI9žõÓoªl,'nÿæýq£:ËÉ<9:®=¸ENåëñe¥¼ÇåôýÏdx`©œ)šôÎðê—åíÒSšyÞ•wjlÔ÷¶3òn›ô;_päìàæUÚ­8+ç^hìÞ³e@Û=÷½~Û ”.–ÈdðþwÛë[jËÐ×5³)ƒ _©±GU•QÖªª‹K”1;%cc£Œ?µôåÙ5fÈ³Óæ•›¼½œLꜭO•)¼ë×CË4•'Ÿ‘÷¥q¯5‘™Õb/,sl·ãD±a2÷«.[¶þ2^æû_s룲0³ÿAgÊâÎ+rlŸ±P^&{gv£,½øJÁ\Ïþ,+FTžqeçݲ²ÌwK§.œ#¯å­ùòYÙ(¯ß׸ؙ9e-{ÊØñ²~[ï&]Úž“MOž{¸il²%úÍ•áò³lívQoÛ½‚¼©-_óÉWdûŠoú¬üm‡ìøù±—Ò—·ºVwù‰Ö²gÀÌòU½$ûæL¸ïü ·É½?gZÝå394dôguï{B/Kn[±v79rö—ÞÅEåX†Ú³NîüEŽ?úHÙ•m~”=:öú`ö½rrY ŠœR^ÎÜÛ®ykR¾O׌/Ê»½¿Oi2³¾œ—4s惵åÜì_~Réý?õ6ýño¿>õè¡DÞRíA»ì×Áþìÿ‡’¯ž¾ç™ó£n%Ž_y¡ØŠ ‰Ûƒç¯´ŸU•ãÝGîɲ©¹ÔRÛ¹ŽòYµÎûI¯D»Ó`vdmó÷–úÊI=å&hk*o5ü½ômR]í·5³?Ù½Á©e‰ö/x|оŽê©ö"hïƒ÷i8F’µŽí]³bNZ¦Y´ã•Çn“6öóÑnw”•öªêøòk•*ïø,ᨫ5²Ñ#ÖK7•z>ž¶¤×Ä“>Ùü†q¹ôoÚ!S£g3É€«Óòt­6^ì’å—ÊådˆúÞ‡©|2üío‡ÿ².ŒìKô2ºjš~ÇJ–±ƒ~õ÷H™Å‡yƒ<»`Óz·j+™Tã±:l–)·î^~áL™úkΫ÷—ûDž»òjfG§ËŒ›n¦ÅþYfVøzzå½döºb3ª¦é/s‹^S½È:™·³•Ñ/ó`YÀÖ¾å‡+²°Í‚IoÍ.)‹»=Y´ÛÚ›ä¥Yí?ë˜W“W>x-Oƒ:§ey•Wžÿ-ÛbYqéèUûW••u¾lrû(GV}ÿÌ›~üVVƒµöãßÉšßVt]Ôi·¬Ÿ8ߨ–7‹lêݳm®–®lQž·ýºC«G[ȶ—.N+ÞΑí™c Tv̹³jõ†/É®ê—{Ô˜_vû½’ ŸËÞ¥YŽÔ×*zâž{~}mŸî¿þR´ÈÃrdblCÈÑîÞw'Gd }+8ºÝ°Ïåxº½wÕŸu›œH*¿¾~a‘“7äo}OfCNÝÛÁåt¶–;'~ÿ±œ¹gò锞SåíG꾜s`y‡Ö>ûÚªònËæŽ>ÑXθtï ‹Ê¹ç3û=9_{Õ¡¾C^¿æ­sÁŠ42Qí?5ö´S¢‚›Èû³ÀU²Ú߃öª°º¿rTÐO?‘Ê_Pî_Tíï]<®QÂgà2Eµ3A{—(Ÿþ¾]ª„ï  ÚÝ?´‡ª]~Ty >O¢ýS®ÿA÷ µ>•ÕúTþÑ–ö·¬“jªÝª~`{¶Ãón—šêû ú±uT»´{ Õçh¢Ö«™zÿ–êñ­c»¿'íüwÏYK:¨<Òé®ã‡ÝÚH:?7âÇF o®Yݼ3Sné6Åï( ‘÷½¬Z³¨ô|çí¢»ŸÿDz«×éGV_2)I¨þú [_[so•çdHÉÂջͿU† ¿'Ê.%#v]”²f€ŒzïÃí;¦U’±Æ]¾²üŸ2oxÿ÷/Ê„Ø×–F&..=`Dådò‡u?õæ~'Ófº-§ýDž_ðË‚»¿È(3UÿqÎü“î#ȼöÓÆÒ$É‚§öçX>ïkYXëçVÏ=žV+2Þ.K¿³Hï'LÝ(Ëú¿øcÅ¥OÉ«ù7ßðE¾±²Êï–_y[Vµkþs9²ÉZÕŸ]O/oÒªE²©NÝ©ùö–-~¯8ÿ3²ÍÏ¢ÍÚÈŽ²îŠ –]j{ïÉK”²µ»ir9H'³äª»äYê™1‹åHÆØ†“£j;þ¯¼M.þTºÆO•êyë4jØI©p¦Ê-¬‰ý1ÑŽ©ý=¥ñG.-ù%ÑŸ Ú±bª¿¸zPíçÁíòï¥Éøãꑉý¸†Ê÷µÕíõÔëÕÿ¡DÒío]–†±D[T«çíK#zÍö›ß&úsu•— ý \Ÿ+á$hçRÕx°NÒ¥nߺRWµWÁú4PŸ·‘Zï&j;7wjn9ñ]™Äx®M§nãrN­&íׯ6ˆtJX&í¯{¤Ëã{®¶+äH7¿7òÀmÒÃ_Í´S¤wÎO Oj°]úú[­Çvyæž%‡¿ô’7ú EöD;7Tµ§ÃÕçåRJ–±ÊÙçÞƒmÞë/ÏÖô;>sdâŒ|7fØÖ[&}Ðí¡ä-ȵý§ÝØg–yãû2=§6tg¿ßdƸ Ã7Ôd–ÿµ,}Eæ¨ï{^¬{iÈü!Ý¿yqAkYpf‚öå`—,ñwŸ#ßÈ˱ݥ–,üþË>,(+ünÃoÃ…>Z÷YÛÈkfñãïÏ,«ŸŸR*«^ZÖÙÿMóõåe]ÒÊ)Ÿ\xU6ä{ýTû­ÏÈ&zcïM¼(oøÍü¸Ë²EyßVjÙá⥳țjœ¶C_w•sÄýéyëâÖ»OÝzNö¨¼·¯aM¿§)èdxø¢Ê{lÀíÝÒÉá›cLP¡ÜÑe-åȬÛü--GÕ÷xô“⟾[;IŽùÝ”RËå¸êWû×ñX))'Z¤½7ÃÒArB廓¹ýüŠœTý´“[?šÚwÙ"9¦¥—Seül–±1¾w%ò{0î Ú• = ÜQýšUû‘h×ÔãïWíù½j?}@í§A;¼Nà5𜺠ÆiA;Uú×Ìž[ )­úG‰2u{–º] æmÔ~ŒãR·g•TûÌçTVó%U•ójêýj¨õ ¼ó7A¿6ÈÔ÷ÑXGšª~aóW³¶ËžnŒ´ô{?w¾6ßòïÚ¹‡¸ó¦Ö)Òý¦£·Íh~Ezªþï oÏy1ïÔk^cÃÁz2à?ñòž zta®ä,­eðÇýoF†¾ôR¾áƒŽ"i¡ŒTÛw´š7Wýµ±}8L&lYª^R]&úñߓɟù‰û L«Ôlóœõ·Ëô£±]^PíÉì·+n«ÜÙ–¹[Í÷Ä/ïÉ|F©C‡O–Ïfò{N²¸Ä7§ûD^b›ýޡ׿Wžé¡ÃcóÊÊÎè=òœ¼ÖrøâO<#¯wë•ïüî3²–^Âì%ýeCç·Ð…–Í«nòg`dkš—ï»°S¶Éıd»>µÅêKoÈŽ¹Ýß}åË ²kXl‚DvǾ–§eïs~þVöO®Û¶hÃ7ä`åMÍâ'äPlX—"‡mQ,[¿7ä°¯Í=$G†GÒÍ-ú­ÍR}ÔŒƒÇåèìX"¤]Œ5HrlÜY¾©‡å˜ÚÏŽ×~aýäøÊcÇ¢”j\r¢Ò‚Å[«ï‘{ÝôqFÚ·[ü†åõkí[*w¥”§Dû¦úI©Û·ÔýÌÔí[Ð úoUÔëó“Aû’ØoU¾mÈ»´Ü9I½p¾Û÷¿æ’ƪŸÔ(6}Ñ_äŒMtIÿæióóÁ穦æ÷‚zð~Áã󕛚ßٷ̾kí\0o©¾·újþ¯áw±Ðµ~bêy•_Ûž—}~í"ÒñxlâB:ÞæÀÐB¯J·ûrWZ7·‡ôèY뮕.I¯Sƒ×?Xú˜ôUó£ý?ãÕºµD¨~Ý Ø´ç=2äËâ÷gÿ¡„ WÏUvf—R_l±yf=ú“‘QÆ«vþYõ½Lœà wʤóþÄZkí[ÿ ¬*ÓémѤȌI± ™U5™ù–kí[öØ–ùÃýá¤,¸mã’V·ÊBZ­aƒKÈ"Õ¿\âoû —7ïyyT½e™'­8õù”7¾.#+;HÕ_òÖ’×ìØ ²zV¬”5þôèÌ]².ö²Ýeƒš‡Øä§ëӛ孙Î5«±C¶ü-ÛÔ8ðÍ«¯d»ëβccaVÙkþ3ËnÕ^ï9·@>ØMö5ç½Ú±²ø¤ç±_/¿.‡Ôv9ü\^án9òì¢z‡6t£½-ìžã+9V«ß”Z×Ëñ{_ôG’rü‡½Ü­©œx-öÁådÓ“_ñÈ09åO'ô<.§fßìgN9K÷}åôKwœ:9Ó•3·Ç¦œí ;äÌ϶ì¼"§¼­æ¹Ælضî‰Év¢Œo‚ö-õ|c5x*¤ö»Û¾©ÇSã—`<xNÝ.%ÚµTó3Áxíã¶?iï‚v²¬Zïrªß÷¸?Vô‡G¹I%µÞO*¿A;WåAýÓÇÇdº6o©^71~SNëÆ†YäéŸýؾ6¿¢>_Sõ¼Äñ„þ}Í 3ž•Ö*O´õ§¥'½&fž¶ï¡4ÒQÍ›tV㢮b ­tó÷Âû»HÕßï.Ü(}c­f²ô¿XºæÔFÉÕ> zÐo˜Þ–Á§j¿µ»`Z:ãë­ õnGŽ<—¾óWÞŸ·o´Ê´\2YyÿCû¦ÚÕÙj¾sîæo_ø®ýR™ïg‘6wÊ‹ow_µaðAYüPµµï¯í+/©qãÒ³›&dùè)Y1óRf†¤²²Gì@€¼ÖaÓ ‘äu5Î[«Žol˜ÿfºV²ù­×ülMK ²mîê]ûGŽ–í9Û6ªzwÙá×îŸ)»Ö¿ã XYJvŸ‘žT/ûœcS²Ü=]÷MÂUG5î¬Ú¯àø\bÞ26üj"ýlõzvãµãr‡+=¼¿ðb\96`“!±iº›eØÛ7œzàd}©Ú­Ñê¸Ú85Žœà®3I&ª÷››¦È-SÕöœ^õˆ/óÚñ8ånŽzß¹K~þ¨íèGe¾ÿ5ŒxOøÃºV'e¡š'_ôe%„$K~šv¡àêòJ‘ôdª ² œ²_”/¨[ªÌ÷²2klÂYV­Z8ú–7Ë꿤g5Êñ:uÜmCU¦Šlêï”–É–Œ± $[Çú RNySõ›·Î@Fª,; “d×¢aÛò˜)»ÇùÒ=²wú›£õ—Êþ—æ/Ír¤œsÅi³nŒÚ˜2vÜîzrx|§œ/Wè"GüÃNÉÑô_¿YñÖÑrÔo [Œ‘cM·ýXôµyx‹·³ÇG­°sÑ×rÂóÔ’ »·ÏH9áÏòß´DN’ ;ŸÉ 'Õq˜Sf“]O®~\NUk¹øÔ“2¨¬#‘hg_Áãhd”j/Ç M¸Ëx¿wôÐb™ òþÄR±Ù2éâ æD¦ì÷^ÿ¢EyNÍçņcw|)3ý£€9u™£ŽÏý>Ò«ï›sdëH¾SªþÁ’[»ù¯ /_.Ö©á”¶ŸvI#¯–ù­›¯§¬RÇKV+wk¹»æÕ­²þ½ä«·T.$›Ôq²-U'ù3R²ÍjÐáÈí—åÍ+Ù›vÈÔXv¨ãß»~}mßVNÙýí«_æ)ÜJöƦ{–ýjü}g}óþx94bAÿƒŸ¬•xZ{JŽØuÁ½ GÖDÞ;a­ãÈ|TŽ^òòrLµÃÇÓlÜÜèàU9ë¾–•㕚ÎPANT\Zkã3=åÄ’¿Vøzº wé8ÿ¤›WÇÙ‚ý<õüã¿kß_Aô3ÿ´}Só}A;ó‡öM}ÁücpH]Õÿ æáƒ~]p~Lp|#1OÙûÂcË2öIøÆuÁ›Vv¤±šßh:þô¯E~x_š«v§UåBÙê÷’´- HGíËyÎK ¤sSbz¸tSÇ {ª¼Ù[Í£ôSãÀj;pMæ§.ÞÒL>Rþý'çÈ–ú¦£2lƒ?!PRFFüÄ~AFÏy1ï”)Weœš/œ æO&ªõ›L¯ï•a†L=šrÖA™eºšGš±åV?ÃȬNjìÞõ°Ìi^¢ëÏÕçÈÜ—+ø3â-˼u³,´yôÖVÛdaÞ{×w¹ï;YôÓ­þÌ®¼äg wáµömFŸKhËŠ¹çü#{²R¿Wù£–JëeuŸ]?ôþ>EÖ4~}Y¿šídèhéeCǼÑÝvVÙ4ý‹¾t1eKöX”­Œ²Öím)oƦMn–í´ÂCßî!;ýÞ´ÖIvùÃ#k¤ìöÓÄ¢d¯ßÛ­·Qöû£Î†šò{™V’¶¿ø B¹crXígGüÙžºYäè˜þùòx=å˜úüÇ~®>{òÐúr|N•Òotî$'bÝê$9±sãÇÕN''ËßùlòÀ2rRµ‡§ÜçñÜÉ©ÉÏ׬õsk9åý»££œ.Óøüì[Ê@¿·tKúD;K¸ Ú·`œ¦æ=R·oAÿ1h×r«ý7hïîQ^þо¥:´S‰öM½OêãqA>´_‰ãíÊCàü¯¼®ú<‰yR•o‚ñ^‡‚q^pœ>×%ÎKS^ó–Áñ õ¼Äñ5ŸÑÄ?ÌüÓ`iêŸn2t„4Wý×–‹ÛuºÜ¸Ib\—8E͇tRóF]Ôq«n±Ý¯¼ôPý±^ÊuŸØnn]óªÚ×jžap¯XÇO†ªþópÕI–xïð>£ÆƒãÔüÍ•O&–xË?“ëZûvð¥ý³¿­}[4ƒ‘Ó@™©Úå9±Õ©,sýnÚ¨²@XØa’Ÿ1®µoþaüÜóe¹ß‹îÒN^UÇQVåˆ ”dõ€ú òtUÖÆÒíM²>Ö»A6­_™%Óœ>²åé&#šM^.ÛÔy[Û3Ǿ쌥SÞò»u¯ÊÙç]Ù×ååÝ™OÊCöïó–Êê'òzrè·lôH&Êácmò¾1¼µ™µí@Ùòäh½Ÿ^9ýq²‹uó»È±ÕÞwnÈ/ÇŸlÔ"wž·ä¸š‡=ÑèÊ×· k/'ÞýJ§¤Ur2vØ¡¾œ|uùCl—gbØD2hÇg6øIí,§z|Pí\Ð ú{A¿­‚šgH_VûUuÜ©Zlú»P¢J}ÞHêyÈàyÁüF¢ ^/u{¦>g0Ï̃$Ú×`¾Rå… } ú‰ÁühÃeGôÛ,q|®±z¿¦ª½l®^·¥j'Û¨ÏÝΟžë¶_:*”‡Äy—«ýçJÏ™FzfžôQýÌþê{  RóCÔ÷4ÜgXî ÙºuäŽé_ÈèsãËŸòŒ›4kéñ [eBl¸qwâ8Ý䞥òÿòcZ™ª<§æ1fdòW åµãÛêxô¼ûŸ¾Poᙯú Á¼È¢·*<½¯É%Y²'6a /Ç¿••e­––zs÷W×Ú¹r“·?öæ òZÁØ ’òºš^«Æcë—LòΟ=*Õùgoø›wü ÙjôËÌX¶ )fu¶Èö›b+ ;HŽéöÞ%»bÓ#d·:ßf¯ªïWç™l;ÑJùÝÆ5oÈaÿhø-ãåÈ0£eöý䨚ï<Ö6WKçØ9®Ž§žPçלŒMf’Sþi¢õGÊÕ®¾]?GÙ]M—É;j\|öÞ³HUrNõïÏÓ›™¿ô–Ä|tÐnóùAûŒË7©Û¯ÀÕí*gUƒ2¸?hçþ0ŽS·óŸÁy˜Ö>¥>_,x\âüMå"þ¡T÷'Þ'Õyg‰ùRõ9ã8ÕoýÃù'*?ÞÇçT?ôç]ó—Êgp¿‰Ê+ÍÔücË»§ûg`\;E½oÂ¥‡ç[&\ªùÄybË©±§ rzÀXäÈ`Õ>9€È0u>ÈãƒËiŸù^F©y1±]eü{w3ò¯,Ϫþì¤éß?ü¨Lao¬º©„LSûõtu|kÆj_hK™•-–˜dÎÐW‡–[²Iæ1î™%E¨þö«þÄMYûÚ»Ê+jü´|‡>°CšVòªzØ¢WeÕYÙpr¥&¯—ûxQ¶gÈÚs½+õ^SS6Ìÿ½¦ÍÃkißâßÏÊò± y-ö±3ÊëêøßZår½:or£? Ôc¨¼QÄ(Þ/[}†-ß‘mCýuˆl¿ÕßËŽý~+8§¹ìRã¬ÝKªçÜòh>Ù»¹»AöûgeÖ® }µY?“C§7WŸŽÞSŽÚÐþDŽrtë¤U Gß*ÇbÃÓ‡åxlxÙTNœ=2â@LJå”öü…†ß,§cÝîtr¦×†¥[žóäí}c/¾R0·¼ëŸ¦_•³~«µQÎgöï 9ÿAÎZ-_˜'ÝIN mý¼ ß—zþ#p¸Ê¥öÛ¿Ú¾íâ¿kßþÐþ¤jç‚2u{–ú<˜à8ßJuê÷ >ÐÎ'æ+Õù)Áß#Ç)ƒóQ‚ü¸Oü]AàµÕà*O­Í—pú´:Ͼ¡šgk¤òCb¾R§ÇÏÕù#mÕùÓ‰qœÊ#]T¾è¦ž×#vب…ôVï×Oõ{¨ù»jœ×?¡?ƒ ñJåß,ÃüÙôû釪÷©ú=cüfù“ý2^õ£O ÆSüY—–·Ë4uáô®þ Xód†ÖcžÇe¦¯ D7™£òÂ<¿ÛôóG²`@l‘…_úÏʒѱ?ì‘WºÆ&(dùÊ8yual $«NÇ]kß>hXrdÑŠ²A‹6ûÝèìóe«:/çMu\f£ÀÝÛß–]MÜg«õúYvû£8½–ì=²kí„ã ¯ûûßXe©Î› –#ê¸õQÿ,‡F-äØÓ7¾8¾S.9>åØÇ?§Ù$'T;x*eÍ€Ûí”Ó«Ë_)W㌼’gÕ#ËÇÉ;êü²³ê|îóñãiÒèÑ4iÒ^M“&M:ÿãÔÉ齚5ꬖµF-‚¿Óº4ë<¢s£&ÁÍ‘Nñ—ɼLDÝá6mÔµÑýÍ;7j×,þˆÜþ#n ¢øßÕ«WrPüÿþkÏìÒª}‹¶ÍÔj¤õŸ‘2;%ÿ/evå4êßo¿v…àÕs§Ic­$V¯«‰×‰5ÄZb±žØ@l$6›‰7ˆ-ÄVbñ&±ØAì$vo»‰=Ä^b±Ÿ8@$‡‰#ÄQâX|Ý*²nÚMê+f£“Õ:†FÿØFaü×bga„ñߊ _ûç÷¤ýµæwÖU¿Ó$,ÂzÙ„G$©~³ßϾQõáÓ¿Ïš‘ÈDÜLd&²···Y‰lDvâvÂïÕç$rÅûâiòwy‰;‰|Ä]ÄÝÄ=ĽÄ}ÄýÄD~¢Q(Db?q€8H"Gˆ£Ä1â8q‚8Iœ"Ngˆ·‰wˆw‰³Ä9âÑ€hH4"Mˆ¦;s´9Ñ‚€E´ÑšhC´%Úí‰DG‚}> ‘h¢+ÑèNô @íEô&ú}‰~Dâ™4±ß£‰$ƒ‰!ÄPb1œAŒ$F£‰1ÄXb1ž˜@ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ÒÈGùH#iä#|¤‘4ò‘F>ҮƯŒªGâ9\'éÜ “tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|¤“tò‘N>ÒÉG:ùH'éä#|d°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>ëc°>†¿>äGƒüh ò£A~4ÈùÑ ?äGƒüh ò£A~4ÈùÑ ?äGƒüh ò£A~4ÈùÑ ?äGƒüh ò£A~4ÈùÑ ?äGƒüh ò£A>2ÈGFÿ(A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGùÈ ä#ƒ|d ò‘A>2ÈGÆÕøšMò‘I>2ÉG¦þW~òí_þ´›Õ>þÛnñŽip=èß=Äð_ÄïÉÍ"R†²óð/¥\šxYD•ÅT™G•¢ÊBª¼_•ÅU™/ÕíùUù* «òá4¿¿Šª¬¢Ê§TYG•µÄËzª^?%^6Ì/k¨z õ¸Zêq5U½Îu»*«©ûk¯«Êúi~_·7R¯ßô|¼lœòû²™z\ õ¼Öêö6jýZªûÛ·«õ袾÷nêþÞ U]=®wü Lé§×_ÕŸQï3H½ÎÐ,ñrøˆx9J½ÎXõ:£?Ž—ãU}bËx95*‰—ÓŒ—Ïï‹—³ÛÆË9¯ÄËùêùóž— ÇÅË%+ãåKÛãå²ÇãåòÅñrÕ}ª|4^®þ>^®‰oè”õê}7—Š—[â;TÊÖh¼|ójì JÙy&^ßs[¼¾çb¼Ü»$^J‰­`ÊÁ=ñÇù2~û±&±/*åø[ñúéOâ÷Ÿ©¯¿ßqR.n‰?ÿl)uû¢øóôŒ?>Øï‹ª2ØŸƒý>‹*󤺿@ª2—*óªòî?©WeéTe|5q4à÷+£êeS=>ðûè€ßß^I•©Û‹¥ªWMùýë—W·WSe…T«§ÊÀ[Ý ®xm¤^¯‰z\uó†ñ²EŽxÙJ=>pÔT•íæÄËNêñÝT½—*{žW¥º@üƒ¥ T^«ýnˆº˜zÝkãåHådÌ¡x9^•3ÆËI}ãåä¬ñrZ×x9]9¡ê³ã¿”2GíïóªÆËÕç_T=^.Vë±Xy]öm¼\^3^®/_'¨”5%ãåºH¼ÜÿáΔ-ƒb$e[<Ѧìȯï</÷¤‰—û¯ÄËC¶ÄÊ#¯ÆD9>‘5~û1åàÔË)±òíBñç¾5þ¸éâeõy…R•ÁþŸ÷Oʪ̗êy…S•Á~쟩۫¤ÄË ýxj‹*Õãuõ¸Úª´5Õójªûƒö'x~°õàõ믔êy Ôã‚ö*ØßªÛƒv%h?Z6Œ—ÍÏÇËÖêñ-ÔýæÄËNjÿê¢ÊîêñÝ:ÆËžêqƒ>êö~êõƒí5Hí/ƒU}°jgF=§J;^ŽUÏß0)Ϫýx¢Ú¯'—ˆ—SU{1M½Îó+âåìãå,åinÓx9Oµ Õë.Þ/—ü/—©÷_¡öã•â媑ñòuõþkã‰,eS¦x¹YùÞ6D•½ãå›mâåÎbånõù÷m‰—vÇËÃã÷/7L‰•'¦Çï?ùn¼<½/~ÿyï|¼¬û¢RÞm{Á”w^ßßG½OÐ ý¤Ôý¦ ÿ߫ʬªÌ“êþ¬©žx)¨ÊdU–ReÉ¿¿½¤*ƒþØC©ž—¢Ê —¢ž_vÀï_7ðW2U=è×íQyUÚ« Ý©¼%^VSeà°VŠ*ÕãŸVõ§U½áyUªÛ©Û›5Œ—-Ôí­s¨úœxÙ^½O[õ¸NjÿזּSWõþŸêöî_ÇËêu¨÷¬Ú•AÊÅpõº#Ôý£T{2F­ßø·âå³jÿž¬\L/Ÿ[/g¨~ج×T©Ú«¹ÊÉÿ9?«Ûø0÷_UÏaùµœÿºÛ \·\ðºåB×-?xÝráë–‹¨åû‚Û~V¥ÿã9‰ïi®[ÇÄ})Ý]ß¿{Î_\·¢ÿ£u+vÝûÿËë’&Íÿô}RÝ~}¨×}èº÷xøêïï ö¥×=æ‘?Ù¯þú>“bÿÙgÿëÛö_¼†²‘ü‡™ƒÅ‹ä þûã;þ«O(W㯔rõïù)ÿÕV,<ÿúOž>GŽF-šåèܨ}‹f]r4ïÜ¡Ýß8‘&þÕn™ú£];ü»WHIó|Ó?ÚØÅýW÷ähרm³‡þíëÛÿÇ×·s4oö×^'Ýÿñu"ÿjoÊÑ´Q×f9:4ÏѬ}×νþä-ñÿú-ñŸ|£±5½nÿÿÃ̙ә•îÚ¨qÛë&ÎþåšÝ¶Y·6ÍÚµ¢»®ù¿‚¦©5Èœ¸ßª :\ÁÀ§„*ƒwÐ1 :NÁÄWÓñ2cšß—9R•A/˜P Þ'è¸%:lêõ‚o,¶þÁ6¸n[ü®þŸÞþgû³ò?}üµçi~ÇÀÿºÈÑÿ,ü}àŽ`§ño‹%Í'µjß•hÖTÝrC¥íÛ]wã¿Ü×ÌæštíàŸNð[ªÓ 4õ½k«vÁÉf—®ºvë¢j‘ž×ZN}îØ‚…‚ÁB¡`áÁ`¡p°P$X(, Š«…hü‰¥‰¥‚‰¥B‰¥K…KEKEKÅK‰÷(˜x‚‰÷(˜x‚…þ“yî?P5Ú÷èÚ¤ƒrÛ:iÒܸñºíþ âÿ=¿‚Øžð[¶Âéû||þp¬ÑèAø£•^„?ÚéCø³pþȦ?áO ƒˆÁ„? 0”F 'üYraáϨù£.ãÏf'üÙ‹g‰‰Ä$Ÿe˜BL%¦þìˆ?àÏ‚Ï üQÎLb1›˜CÌ%æó 4çOõ-$þ¬øâ%âeŸ]_J,#–þ¬‰?çÏŽ¬"üÑßjâub á&×ë‰ „¿¿o"6o[ˆ­Ä6âMŸÍÛAì$vþèÓŸéðgÉ÷þ(s?q€8Hø³œ‡‰#ÄQÂ!'N'‰SÄiŸ›x‡ðgDÎçˆóÄâ"ñá>? >$>"üÙMþSâ3Âm~A|Iø#P?ë_"¾!.þ,èw„ÔàÂ_!~Jï&úG/~MOWUúçüGðÁÿüGðÁĉw"øà?‚ÿˆ0 ÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿ‘þ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?’?žêý,ï'x?·ûiÝÏè~2÷ó¸ŸÂ#øà?‚ÿþ#É~7+ÞQŠà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ#øà?‚ÿþ# üGðÁÿüûWkà?‚ÿþ#øà?‚ÿw"øà?‚ÿôºþ#øà?‚ÿþ#ø÷ÛÇþ#øà?‚ÿþ#øà?2€ÀÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿüGüSûðÁÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿüGðÁÿü‡§¦ùûœjžfø>Í0Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(ö£Øb?Šý(öý d¿»¯a_þFEþ†} ûö5ìkØ×°¯a_KŠ-5ìkØ×°¯a_ËShØ×°¯a_þ†} ûö5ìk´ýþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ† ÿþ5ükø×ð¯ùó&ø×ð¯á_ÿ† ÿþ5ükø×ð¯á_ÿ†ÍÇá?<­7MxZoxZïí´^ÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿îOá_Ç¿ŽÝÿëüëø×ñ¯ã_Ç¿Žÿ:þuÿOkð¯wñgÉüëø×ñ¯ûscø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_Ç¿Žÿ:þuüëø×ñ¯ã_ǾŽ}û:öuìëØ×±¯c_ǾîO ^OñØ7°o`ßÀ¾Áö ìØ7°o`ßÀ¾}ûö ìØ7°o`ßÀ¾}ûö ìØ7°o`ßÀ¾}ûö ìØ7°o`ßÀ¾}ûö ìØO¡OžBžBÿ_;…ÞÄ¿É&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿæíþ[ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&þMü›ø7ñoâßÄ¿‰ÿ&öMì›Íüã`öMì›Ø7±obßľ‰}û&öMì›Ø7;ûGÃì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±obßľ‰}û&öMì›Ø7±o^Ö³°oaß¾…} ûö-î´°oaß¾…} ûö-ì[Ø·°oaß¾…} ûö-ì[Ø·°oaß¾…} ûö-ì[Ø·°oaß¾…} ûö-ì[Ø·poáÞ½…y óæ-Ì[˜·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-Úz ïÞ-¼[x·ðnáÝ»…w ïÞ-¼[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u ëÖ-¬[X·°naݺ…u çÎ-œ[8·pnáܹ…s çÎ-œ[8·pnáܹ…s çέ«ñC÷6ÎmœÛ8·qnãÜÆ¹s›Ø8·qnãÜÆ¹sç6ÎmœÛ8·qnãÜÆ¹sç6ÎmœÛ8·qnãÜÆ¹sçöíþiUÎmœÛ8·qnãÜÆ¹s›6ÞÆºuë6m¼wï6Þm¼Ûx·iãmÚx÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½{÷6îmÜÛ¸·qoãÞÆ½ÝÆ?•ŠÀ½M;oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömìÛØ·±oc߯¾}û6ömÚyÿ6þmüÛø·ñoã߯¿ÿ6þmüÛø·ñoã߯¿ÿ6þí«ê<>ü;øwðïàßÁ¿ƒÿþäàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øwðïàßÁ¿ƒÿþü;øw°ï`ßÁ¾ƒ}ûöì;Øw°ï`ßÁ¾ƒ}ûöì;Øw°ï`ßÁ¾ƒ}ûöì;Øw°ï`ßÁ¾ƒ}ûöì;Øw°ï`ßÁ¾ƒ}ûîÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒ{÷îÜ;¸wpïàÞÁ½ƒyóæÌ;˜w0ï`ÞÁ¼ƒyóæÌ;˜w0ï`ÞÁ¼ƒyóæÌ;˜w0ï`ÞÁ¼ƒyóÞ¼;xwðîàݹ?EÏÅ»‹wï.Þ]¼»xwñîâÝÅ»Ë]¼»xwñîâÝÅ»‹wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¬»Xw±îbÝź‹uë.Ö]¬»Xw±îbÝź‹uë.Ö]¬»Xw±îbÝÍïŸsO`Ýź‹uë.Ö]¬»Xw±îbÝź‹uë.Ö]¬»Xw±îbÝź‹uë.Ö]¬»Xw±îbÝź[Ñ?]•Àº‹uë.m½‹wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»xwñîâÝÅ»‹w·‰F&wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»xwñîâÝÅ»‹wï.m½‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞż‹yó.æ]Ì»˜w1ïbÞÅ»‹wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»xwiç]Ì»˜w1ïbÞż‹yó.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»xwñîâÝÅ»‹wï.Þ]¼»´ñ.æ]Ì»˜w1ïbÞ½?%×ü‡yóæ=Ì{˜÷0ïaÞü‡y{˜÷0ïaÞü‡yóÞ=¼{X÷°îaÝú‡u¶Ýû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïÞ=¼{x÷ðîáÝû‡wïm¼‡yóæ=Ì{˜÷0ïaÞü‡yóæ=Ì{˜÷0ïaÞü‡yóæ=Ì{˜÷0ïaÞüÿGÌæ=Ì{˜÷0ïaÞ£÷pïáÞý‡{÷î=Ü{¸÷pïáÞý‡{÷î=Ü{¸÷pïáÞý‡{÷î=Ü{¸÷®ÆOÃOÂ}î“pŸ„û$Ü'á> ÷I¸OÂ}î“pŸÄ’pŸ„û$Ü'á> ÷I´õIØOÂ~m}þ“ðŸ„ÿ$ü'á? ÿIøOÂþ“ðŸ„ÿ$ü'á? ÿIøOÂþ“ðŸ„ÿ$ü'á? ÿIøOÂþ“ðŸ„ÿ$ü'á? ÿIøOÂþ“ðŸ„ÿ$ü'á? ÿIøOÂþ“ðŸ„ÿ$ü'á? ÿIøOúØ{pë’ªL¸NN»ö¾ñë _çN>§›Î‰†nšnhº‰ M#Ý’Än’ˆTT@ATDELŒsBù˜FDó8úëÌ0#¦ÃŒú¿ï­Ug•å>÷žx¿ÖÿÞçÙÏ9gŸªZ«ÖZï»VÕ®ó}Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'Àü'À~ì'À~ì'À~ì'À~ì'À~ì'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À}Ü'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|Ì'À|¼'À{¼'À{¬'Àz¬'Àz¬'Àyœ'Àyœ'ÿì~jcs œ[àÜç8·À¹Î-pns œ[àÜç,pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹½ƒ?<Ü[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·À¹Î-pns œ[àÜç8·ÈóX·ÀºÖ-°nu ¬[`ÝëX·ÀºEž·À»Þ-ðnw ¼[àÝïx·À»Þ-ðnw ¼[àÝïx·À»Þ-ðnw ¼[àÝïx·À»Þ-ðnw ¼[àÝïx·À»Þ-ðnw ¼[àÝïx·À»Þ-ðnw ¼[àÝïx·Èó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0oy Ì[`Þó˜·À¼æ-0ow ¼[äx Ì[`Þó˜·Èñ¸·À½î-po{‹o} ì[`ßûöŸÝOìR`?åO"ýØOýØOýØOýØOýØOýS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?öS`?îSà>îSà>îSà>îSà>îSà>îSà>îSà>æS`>æS`>æS`>æS`>ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÞSà=ÖS`=ÖS`=ÖS`=ÖS`=ÖS`=ÖS`=ÖS`=ÎS`<¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾Sà;¾SäôOñOñOñOñOñOñOñOñOqþÌ™?¡Í€ñ Ï€ñ Ï€ñ Ï€ñ Ï€ñ Ï€ñ Ïø»]`3`3.3äá ØÌ€Í ¸äoÂ3à2.3à2.3à2.3à1CîÍw3`2CÞÍ€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ¸Ì€Ë ˜Ì€É ˜Ì€É ˜Ì€É ˜Ì€É ˜Ì€É ˜Ì€É ˜Ì€É ˜Ì€É xÌ€Ç xÌ€Ç x̀nj¿‡µnŽ{?~+Èk1ø^ÝËû>þnZ›ýúNkS4óë;Ï5oÿxŽyúÎ;ö"}ñÁ2¾ }1« —Ë:ý8¯­ñåaÏwݲÁïaÛbZŒîg³Yâþ ¹Íc›YdÎ2ö¢1ºJŸ¯ƒg±÷*ç2+¯­‚+ç™×2œ¶Ì˜«àóe|·N™{‹Œ9Ï,ªËA\´Š9.ªï:ætX±qrõéºì·lͺL¬4þªbiVŒ¶_+×,œ5ïÚâß«­ÕýßÚÜV]§LÃß¼uÕ¿­£ÎXVÎca}¸Šþ‡U‡N³ñ:ë„YÇ^åžÅcñZÄóÎÖöóÖUëÞ;Z÷:™šë0cn]멙NJy÷0Ñw±[e´_ûQÇ.3×/µÚ¢¾]„Çf‰ÓðÍ<\¸Ê™µ®ËËå‹®=–ሃøü±ž£OT¦ÌyâbœuuÄcéZUM³ŒŒɉ³´9‘¾_çžÖ:t[fÝs¢u?‘vy,Ød•s^×>ñ¼}WUËœ_¦ïó~÷¾Ÿ‹òÒ~ã.â«8÷¯{/fVV[‡W«à®U¯3V1ßu¬×õëaäôUÙv>ZÆ«ê7 ¯Í»6]¥½Ëuß<²×µ¾Xf?e•{ËÔ³ûÙf.9±r¢ëøuæÔÃà¼i±tXyrU²cÑšì0¹ø »-ÃAËÈ?2ª±ggÕ±½jû¯[ßEl¾ŽXe ÇŪ×wËê¢×Zëº+óZçžÇ,ïסÓ,yzUs=¬½‰u­MW©Ç*ׄ‡Í«¨åçÕw{>Öö¢Va‡yæ1ïüÙ7˜WßuàskîeÆÏ›ëºcpõáx,à{ÝcÏ“‹cž‹Ú}Ѻp•>ž—•ZtÙ5Ù¬¶ŸÇGËú~^®:,_̺–Y…n«òÛ:t[ÕÛgë°ç:ÖËÓð´,Öã6ëÞØ§†9‘ëÿUùwUùxZ~[Öw«ò÷*÷æ±Ý¢õê"ùh]5źëáYcp‘‘÷~^ŸÍ+«ª#—±ù,5è:u™÷ËðÐc!gæu¢æ»ìúvÕ²gÁé*宲F=QëïEÇXEî^E®œ·ß<2—]O/ºöœ§Ý¢uͺâd_­jMz˜yl]uýªêïuúsì¯z}±LÛýÖ&ËÎy™ÜóÆ:óìº0³Êܸ¯c΋È[f]·Žúf-»72«¾³ä¸ƒdγÿ7Þ2ç¸e|v¢kÉYä׋æÎUrÕ":¬"w-³æ^×zü°bcÝõÙ¼q·JÛ̵=,߯büÃŒŸE8óDÚòDùd™¹Í¿'2>åíuÏcÑþXˆÓÃăÿÝ{(·l”»Ëæ_Æb|•ƒïJ9ß—¢y•¢>ñø¥H–ÿ\2ÿÒ.yã¢ïKŒð½oSÚ•‚{å¨ß4ùáXÅœ6¡Ã9…rój´R0~¨CÌ %ó¯máå†òóÚ„÷C»øÏ±ŸJ9òŠQÛ8߇ú—Ìt]b¿…¾‰å„²bÝbÜ„²óôŽã$nÇÖ4{äõ=H—<ì„zæÙ(Ô#”YžÒ.ÆdÞ˜±îq씢¾¥¹Óâ8œK<§XŸ<Û•£{ñ<óðÇGyþ‹qÇOü>Ž—x~áX¡?ã¹Çx™ƒ±_§Éœ·ys*›iƒŸy<Ú,œÇ´øŽõ ûæÅe^Låͱ#÷ ±Š&_Ïi˜Š}Çtì§<ÿN‹¯0vc{îÇÓb!ΓûñØ´yN³ó~œž‡ß8Äþ¹#O¯i±Ë‹y)¶qž½òü?­Ž˜ÆUyãLÃoÜ.϶q›iy=nÇÈ´¶qÝÏoZëÖgñ§Ù)îŸWäé“g“˜Ãb>™ÖÎÏ)o^ñÜòì•Ç%ÅO(3/OŘ–“c{äñù´ø±8Íq¬ìÇûùg¾òb>Æf‡±¼˜Cýã>;ŒÛ<1‡Lkê烸‰¿;ˆ?¦a5kÓðã7¯.ÞoŒpóæ‹0¯Å¶ÌË‹Óê©8çñCÜ?Žƒì<ÍñÍã X§G!§ÄzåqÇ´õlwÆ~˜Æ£yøŠmã0~çÜ8.C[åå¯8¯ÄºNk“‡Órqh‡<c2/âØ/FãÇsœ¯Óü;MN—æåç<æåÿX·<[äa2Ž«rô]ì߸m9j›Çeyµ@Ü'ï}7ïLJ1–÷ãÍ<ÛÆy-¶a^>ŠýÚ?æž8®bÈÃcÇN³CŒÿØ_y1óºo7mͶËûœ‡ÝØnq›<œÄó õŠ¿‹y"®¯c›‡¯yvãvZæqJÜ.GãyÅñÛ&æš8NólœÓð’—òiŒƒX¿rN»Xרw1N›Cì»ýâr?ŒMÃUçO˧!æÙ#+òüš‡Ûiø‹ÇÌ›w^‰çËÙ/Nâ×øÞ4,îç‹ýj²¯â\û)O¿p]3mþÓø2Ž—<ŒÅ8ù)ÎÓÓÖ!¥œ>ûá-o¿Á·ÉÃjŒí¼ZdÚ¼J‘¼8nK9Wl§ø}öÃ8ËÃOì»iË›÷4¼åavwLãâý87æÒ˜ÛâøŠ¹2/MËÝy|ë–·6Íãæp^y6šV…sˆ1Î3«y5U^¾ËÃ|Œá06ó0PlÅuP§ÇÍóO¼÷Ç#Ól4 «y±>óäÄñrV¨‹yµTĺÆsŒsU)3ïû<¾Ê7ÆzÞË õž6<Ŷ˜&?Σñ<§Åh\äqL¬ã4{M³[WäåÒ<¿çù?Žñý°¶_ˆ12-öâøÝ¯æ‹ù4¾öãÆØ/yy&–5 {å¨oÌ-yù"Ïçq»¼ü”W;ı÷‹96¶QÌ1WÄö‰sLü]Ü.ä´xny14-¿Åýòb#®{C~sUž¦åÀ<]ú5nç©X—xÜ<ĶÇÏ[+çåÛi÷÷Ãv,ï lǶ óW^¬åù7oü¼9æÅB õˋ͂ù×6ˆuÌ㈘SòÆíã5/¯Ä6 y(S÷ãŽøµ8eÌ8oÅöLju‹c2/6ã9Nãü¼˜ 1Æqø9On¸6Œç_qlåqiÌIqÜ•Ìþúì—O㘌k°ivÊ«Góì“WÛÅcıµ¾gYëľØÏæñ¸Óø.Ž¥ØNysŽí’çqÛcyëÚBÎøyü¶‰ù!žCÌ‘y˜c%¶}Ì÷ñ<âXÌÃw{y}òø,ökl«<Êãÿ»q,仟-cŒå}ÎÃaž=ò8$+²IÜnš>Óp™ç¯Ó1Ï–Í¿žë~ù o}·Égš½c¿M[OçÅF^}ó@þòÖž±Ýóæ’gû˜«ö/±ã‰91ôÛAëòi{DÓtË‹‘˜Kã\”g£¼z3žÇ´ø?çù73÷±¼ØçùjZ^Ž9h?|æÅö~~ˆyx?›ÅŒù7ÆGhƒXŸÐ6y\8Í&yyknâ¾ÓrA,7oNyçBtÏÿî½tt]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñõïè*]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]G×Ñut]ÿf.ÿ»÷i…}¾;¨]|¯0G»ýdÄÿöÀ<:-Úî æ‘3oŸUαs/”Ûx™¿¼1V1î,r÷“5¯üxœƒâušÌØ{žÜðuÚ÷ÓÞ¤ï~sš׫ü›s³ô›e¬YÇËk? ާÅî¢1;‹ÿf?»®ç‡eþfµ×ªl0ÍöóàtVùóä›Y8cQ¿ÎƒÉYú®b܃ÆÎk· ŽWý·LÚ¯^ÈksЋȥͪj—ƒî-#c^˜¥>^DÖA¸X&îãÏ«¨§÷ë3kî›5Ï.jÃE𰟽O4ðo?›-RCæõåþaü­ZöAyü±ö7oﯛGã1–Å̺rö¬<ºèwÉ^fÌeýµHÌÏâ‡y|½ÎµÈºsټ㭒KÊ‘‹Ô€óÈ^×߬µÁ*Æ›Çvõw}6Ïߢq<+‡¬“÷Wù·HݺˆŒyí½L]?MÞ*ëÚYsý2ñ»\&¯,+{Þ¿Eòux•±«j3­ÏAëU­_–É=«æôýÖD³ØaÖzj–šq^=fé?œðÞ"øXe-¿7®‹è:O=¹Îúä ³ú,l¿Ê|µÊ5Ǫ8y‘õÓ¼yn^žœÖï0bg^9Óôu‹äûEêÄEæ´î>óþ­’DzûAõÁ:æ9kM2«óæ©yÇŽeô·ŽXŸgŒÃâi«Š¡Upý¢5Ö,ròÞ¯b¼øÞ,²ëýî-Sw­³ß¬c„‡y컈ËþdÿUØoѿðÓ2|6oüÍš7–µožOç©/WÉg'"~–©mÖ]NëûX˧'Rö*kÓYÛ-²þš'O®êoÖZ;ïÞ:bm]suç?¯#f–í¿¨½WÁ'ÓÆÉû~Õq±è|gÁÓïüoÜãõjÈ(áõËpëhs9ú<r~ŸŸ}zxÝÆýg³ù^¿î™×ãz¾¿Þµ?öCx:t{2Þúœ…16ñùGŒÙz?^_Š×Ï‚N°Ûö©øzn? òOÆøÿŒ>ïÄçróÙþQcªWc<Øgóùhs mþ;¾û\M¼ÿKÜ!ä_†Ïça¬OàÞŸa^èS#Þ?‚ë÷1Æ{1Ö{ðþO`£·¡mfL‘6ùq¹ Û±ÂA_ØøØ³qÁ–'];ýÆo¡=ô;VŽo‡>íøâ ó|‡1OÂ\v> m «~'Ú¼2÷_½/Àû‡1Ηã3t+|¾15ØøØg£/ô> ó<öø:×.Ä=øè¤wcü߯=ØÕ~äÀ_Ç »þ¹xELìrüm¿º¼¯°í1¼¶n‡gàýsðýýN–AÌl÷:øþ¯­ âkç `›·ÉX¿™o€=¡ûIƒlèTûîÿ"¾ÿñ Ûn߃1¸yoÂf»°‹E|m"¶áëtLàÏcï‚éÈ5O€<Ø»þ˸û§oA?Œaþ6‚­va‡-|·ý}˜âäØãñž>n»¼ÿn\ˆãìm0ÇÖ-ˆ+Ñí2ÚûãçÛðÝ‹¡GcÁ^»?…ïѧŒ¹C|ƒ=[?‹{ß…v}\B7ø²XÞ8ß#FŽ}×Ëpý¼óý6ð`W'ÝîbãØ -tMÐÿØwâú Œÿ)Ì>Ø…½vÿ÷¾ò¿öÁçcwwŒ×—À–Àûÿ¹_ˆÏð¯…OŽ} .øøØÛ17øm÷1ælžÙïÃg`ɗǾ¯ÀÛl» ;v+ÏÃë—`N7£Ïm°5°XÀ¸Ç~Âan¶Ý6ŽÁ.ŧã=ôßv¿c!î3Øq cì@NŠx<ÆÏˆ¿c°ËîW¸~»ø¾L_~dC÷ì) œoÂ.Õï@_Ìßb5ÄÓçŽyoüÈù9ÈD¿Â/àýÿÄ}pÊîäÂð{ ¿Ô[Ûà‰]ÄÖ1Øbr!N̽°/ìYþgƒ-ðB1kÀu»·¹x8†øÙà¿òÃÐþÜø+ôŒmpåç†>»¸¶Ðïøj:e¿íx¯Ž¾…ÿ…>ðË.xs÷R|ùf» [0¯]ÄUñkßú8Úù?˜7üY8ÍñØ6p± ß—ÿ[àŠmØhܵÛo¾ï¿]Ä^ÌÑ|˜aŽÛ߈ûˆŸôÝÀ¼7Ás»ˆA¿gm¼ò{Ìiü½ñ1ø>ÜÆœ’7á{â¼cžíìºý7±o gõÓÜxæ•ÇÕâ3|U!ÎOB;Ø|Ü—ï^í0¾…yÔÑn1¸ ¼lÂ~»À¿v£Äìι¸^éâ~±µ»‹Wèµõ™x;n#î¶ ÎހΛŸŽÏÐs6Üù{è ÛW€Ëð\z‹‹ß¶Ýù~ô‡ov`³ý+ÞcþMðÛîcŒ¿ƒ¾;”ñ_ñùR×v<´‰\Ø@ÎÚy…‹—yjòw#¶þÀÅ_ñ¿ý•¸þ¶Gž)PwÄo.Þ]v/)tÚ¤]¿uð|9eí¶Áöܽ m§°Çö;~3àøbÀ<Õù²ÿl½ÇÅâ°j—wàóèµÌaÃô ØܵÛÅ÷ˆ±mÄödî`.;_…÷cùa œ¶ógÃäèVBÜ—ÁGu`;û„Ë“uÌ£Y[àKs§³Iñi/ È-g;à±ò¯ÁÀêbf üQdLŽ7$è³;D;äõb`²’KÝXclÿ‚Ë¡µ[OS¿]Ì}ûgІñŠ8*ÀMø¶®¯"?mCÿ ü–€;¶—[÷8 Ï77œ3ÆÑ?:¼øxû7œK¨ ,ü_ø]ܶðîcà‰®ß8a®áûmÔFÛä|äÝoÀ{ÄRØÜþCÛŠ×/qµCBýѾô·ø«’€¿2b«ùj‡£òEq¾ ßmÿgè}ŸÄõ[q!G$Ÿæd¤È‘é›Ü¸ç_nÂ÷êœ È¨ý˜ãµ-ÈÞA²ÞÞ€v€…MÆ âpœ´›pNḳ3s[qß@¾ÜwT€‡ø¸¿XðD‚xÛmœ‰¶°y zn}­Ë}æ.\àéòæbr þ¯‘3¡_“ujÁí¿wy»‚9ì mÑùm xÚÄkÿ6`ß sHS–XCþÚ¢ï`£ñhj>ÙDâê€ ÄÇb·úf‡«-¬ªˆ-ÄdÆxû|ÓäßMø!ùM·–áœ6áÿsÝBnÚbÝz§[cÕÒqJm Ôëi.nî:IÀÃ[ÀÇ&òs†\\†¶P_nÂWðÒ|ØÅþV/›¨‹6Á÷ ê˜6Êð¾{g˜çÆß9¬>ìø‰5õÞÎs¹†óغË齉8* î·ÀßôÉÀ+…š[k±¾ß@¼'˜ûb£ðá<ØÊÂÆ°ßÆáóÛ\Œn`í²~Ú¸šhõÄì•1ÿ0‡3GÇ-ä’ÆkÜæjå&Æl<Õù'ãúÜ““6qv݂ϚðÍæ÷9Þ£®[ÀÈäµh ÆÖ¬[ÈO˜êž¿7aç c6Û­7¸±ª´=òÝ&¸¶žÜDn°¨ EW×m¢Î1à¨JÅqt¶ß$_÷›ðË8eãwÜZ/…ýRä± qŸ"?oãZ/v¹›¶m ¯oÁG›¬u1ÇM~×uØ`Ùäø˜‹…&òG5UóO]ìfàÀ `"e½Âø„M ãá^ÇA5äú28Ñ2&`ËÂù®6dûòg9YuÖª ‡¯­¯tõÉ&æ¾…ØJ¿ÕñB¸º˜Màï*r{ŠxÙøu´w%';<6À÷-äš:æ» Üo~Ðå„ Ôà¤-òÚ»Gæ©»Ü:kuÙÖ‹\ nÐæ?îòGö nF»n!~7É5#‰Ÿ³Ü=®y“±[ï§hŸÀïEÌ«üK˜ÿÚ`ìMäì è³^OÐo¹=»Å­¡,ójó rå?º5× ä¸ts28gÖ'Ísð1¶ *osñÍ}¾n€wÊ›n¿cœ°‰\Z~ƒ‹· õÖ&ò÷&b¿‚uöbzk›Œ8@üZÄ}¼œ!Ö6~ÎÉË€™MÔˆƒï61F†ñ6`£x.c?—á£ìë~¹g’`N›7»µ°½Àñ_öQÁ>û2?eà4C-’}™« ˆ›&rÞêÜ2d`÷ÒÉ®.ÊÀ뵿rû¼lÁ ›°YòÍ;]NkƒõÝ:s]{·Ã×Hä” X±à ûËnÍŸ¾M8uzý›¿âj;æÝø(㺋¸´ðKŠ9YÖäVȱˆ+óCb/`iã÷}7Áå79,dŒUæÅO8Œlü¼ã¸1ÞÚ@ü1^†°ÀI¹ŒöÀÅFÕÍa˜Þd¬¡¦Èзñ£®>©#vÀ›E ФÿäÚsd0‹<³xÛÀ cl`½”ÞéÆËÀ{ɇ]ÎÙàgäƒf×Ù»¿VQ[fÀ`†Ð_¶€•&óã]í¶‡¿KÝžFî°Ü¯{·Û³)À r÷×5ÈcUÖÛ̳›n-À~-Ô©ô*²fEKYWc¬õÁðo N(".’ÿŽ{ôòO NΰFÈàÓì›q9uã!·É`Ÿôûnräø9}¢ËO÷Eþ—Û÷aí±ú´À¹!Ç6‘Sšø>ÖZð ÷7àƒ ä–k“ÖFM¬ÛÄrŠ9gÐÓ¢–KÀ‡)^ ¯qûcäŒäAÿ-ô©SàÑ‚#²KÝÚz~4¸Ÿ—)óâ4AÝ[†œj®-›Û¯* ¿—tëÖ1e¡»f©ã‘ ÊEle/q9"W4k ׸ü¿÷°“Á˜ Ø3Řuø1û ó×¾ò_ ØËZn¾ú6É9ˆ‡ u“W´ pc‚Z·õ#níYçÚT‰ Äv˺zŽ{MskO+¯vó,ÃÎ)xÑ ÞpAºëæžçê‚ô4·Iá3.2ïpc6QCY¼0–ý[—§[ŒK`&ÃÛo ù—5âÂÒçàªü˜`œ*mv·“U¼ßí•5^àðOîbŽ _Ы†5©ýCçÛß[à)ã>ø%E,o çµçJq50±QOµž.n(vÝ^Wv…[÷¤°u ¾KNwûdÜC˾Óqeú~WKqÏ1{©ÃQöwÎß)æT´^ér 럾/]‡ï:u—/Âýw‰-é:l’ÝÚÉ‚ÛZàëlV@þNßè8©úI‰?ض…¸¯€s,l—B—âÏKýŠñíÏâ"Þ'÷8Ÿgw¹üP{–ØùoݺºžIý&pn€³Úý®.àœ*Ð7Ã,bÌÂf)bÞãÀмQ‡šÏt8c‹Óþƒ[K0~ Øir¾h‚CÒË{äû*ðÒBž­#šˆ…¶ÎÀ›)Öj>i‚—Rú:gŒMúmšàÌ ~jq϶¨ƒ¯ìŽãòQ_§˜S•kž»}/Æl±S6ìû]ìRÇ”yþKSîÛ¢Oƒ5 |“1Èþ|é¯]®àÞ¶½Þá¯À:à#nm™|ԓŸÉ¸¸orílXΉñŽÜ™ ­ý6Çåu`Ý~Ôq?sb5•EíaY ¿Xl[ÖÛ.~,æû-`2E-PBŽO˜¯QwÚ9ûÐGÜ#§’ÏRÌÁ@§&ò¢Ø/s>´È ¸#þ5Ô…-âjÓåÈ„ØBN¶°QƒÜÎX»ŸÿÎÍ5½Cúc¼2kÔiéÈåô"çðë®ÎL€•¸¸I>àÞ×þäò|‚×ƬðÙðb_ëô°Èsr‹ÀQ ë™ðY‡—8µ £…üP?æòR ¼Xýugÿ:jÐðÑ|¯Û3h!OX¬ÙRØ´Â}Ö¼×â|V„^-Ô)júê×»xhü «Sî‹#î«°S•\„úË{\\@ýkÿÜñ} xn!6ëÄÖ |æeˆOø¯ ì&ˆ‘*÷ˆsÌÁ"¶,jh ÞN^äø¬þCn=Vþu7F XJœ¸ïX@LµC­ºg%Ü/çšÚS=ܧ~Ä"÷j`¯Æ9®Mkn=• O`o’žÅz¤E¹ÜKVÊà´2rƒEüÔù¶*AFŠo[[à­rm‹uŸ×!VJÀZ ã&x-ã êúÇ#…sݳ¨cÔã[ˆ—*îÕÚ®þ¶È 5øÝ¾Ìå2,-MáûÜZ*AmÓ@_BŠ6i"GØ—»ú.¹Úá‰|Êçy¾+À¾MÄRuîU€5ûBǧÉ.wÔX k/âú£.þ-ò\‚øJàø¸†¸¬|Ôù¸È¸Äý*â—ϦZˆ­b †ïkðWk"þëÀby¯…5Bk‹&÷Ì‘'’žã;î9³þkó?uÜßÏYæ-¶{ÄÕKuô¯"6ŠÀb z7 «‰˜jÂ/­ª« ÆN>áÖß|fǽ³äã.~Šéx!áøä(Ä@ó§S‚üÚúFÇ?¶æj©ÖÉ/¹<Þâú‰{6hŸ _ ÷‹¨oÖˆÇ óÚÇ\ ^à#0W{óiíÇ |\ê;ZÏsœYBíÜ$'µ7»=¯ çðW7þÆåmƒ¢`ݳ:òz ú&X5˜KŸí|JŽM~ÆÅC ù·Àgÿ_﹚+A 5~ÀÍ‘8m‚ƒê¯qû ŒÙ±Ÿ€“ qEN×¾ÔqkëG0ç6ïïqë;î[0Nå{Ðç N×Âõî{b²‰µn5K øk|‹Ã[ 5J\Vèº}»âÐÍ8j kßëž’[ Üã1àžŸÀÖ?r{´MØ¡ >Là‡:¸«õ!Ç¥EÔ6­;%6 S‚z¡†º –¹š«ñ7Ÿcšù *‚‡ZÀ]ƒÏuùLà/Ýó‹2ñ‹øi`Œ"׊ÜÿƒÝ[¬U0¿:Ÿ•üO·Fkœ…±cšÀQ˜kWMଉz¬Æ³Èé ÖmŒõ†Ûç) ŽàŸ&k/ÌÁ|ÊÍ·ùݾVƒ¾„>æoÝžWÜÛd|!¯ÕþÆÅ|x8ÝéÝB,WYßcÜ&|ׄ=[œ;tk/ZïuëÓ:b²Š^¤ü‡Üsb®|5û8¤Øt²ö°‘u`¾ò;n ˵.÷À|óz§?kúx ñ8WóUѦÎ}‚†{Z#®àƒòSí ·Öl ^ŒQply§Iß}.Æg6À}Í':ß5QË6_êbµ|7hóßp6ÚÃæ{݆qGlµ{šðY“Ï)ŸèöÉšÈõ¯rkóÂÃŽãKÌ1Àu ß·¶Ý3$êÕ|Ð=+ mš7¹ñaÓ §ƒ'›xߺPö8ÀçîÿËK¿æÆãùrnýQ÷“ëÄmϹ}³Ûcm!Oø©ÌU¯t{̬y ¸³A_}›Ãuó|‡:k~Ô‹ æ-Æ8ñÿ7ï͸y:Wçø±xš['ÉkÈCåó] Ö0·jCžW¡mÇÝ>\éç\]õyõ{å.oT¹¶ø¸³Aþ©¿ÚñQu^é§Ýºƒûàõ÷»óŒ5î»KÖ$M·—Þø”ð-8¿ñ{n?¼öƒ.·ÖyÆããNæîú/B.r¦^Å+gӥ绘k@×ú§‹Üotg« ìÃx¹Ñ}®oÌYÔØ©Ä"&ê˜÷¼Ë”î+–ÜZ¨þI·GÅ©pçÝ.v·:™u1TC[ù!wVy‹û>Ìeø¶ŠÚ¿úÇ.GÐVÄQë°2Ï ÖmÀuž¿ù„«sÈQuäËêëݲÆ/ºµY ü\gÍËù¡Ž¶|nD/ÑþÀ\õYínˆ{2åßwµ5ûÖ‘ç«/uÜÉs!òôåÿQD?Uîp¸bΠŽUÖÇ\m±‡Ä!ϼUÉÏÏsg ·»5UuVkŸ:ã UýU‡©=^'Vßîb³þ¯Aïã„5!ê½*¸¡ÑrkóÆ…n˜>b|ÖÿÈù·À\]‚Ý«_ëb¦Æg2¿ëÖLUÄcí^·G˼UM\¬Õ¸÷¿ÖÈè)ʨ‘Ê\Ç¡þ«#†ÊÀm¹¢Jì³ö†} äzø¸†xª"O”ÈùÈ˵['—‘S*¿îÖ.\GrO‰6¨]çlZå³Jø§\UàÛÚ«Ý^a ÜÅó‘†ëÄSm×éP{»«ÍëÌÿÅÅwýr×–gìꌣŸry‡k<ÖÍ%ÄKòj\;ý°ó3Ïa²Îçs±ÚÿvgËàÊ}þ´Ù7¹3žŒ½ìVG~¨¢6ã~qíË\}ÆÏe¬}ËðO{Aw»=Íø¥ˆZ´Ê=<ÔÞ5žácl½ÜíùÕ~Õáœ:TnpuUyˆg‹Op}u@ýaWVá¿ÚÉ®v« \\•¯_ëâ¤öÍ+ÕŸwmk\—A·Êý¸xª!f¹wXB Z}¿û\b|nº¼Xüˆ[+Tïu¼DV¹ßÉ=ž]—÷*ûµÓ>öìDýp¯‚ù×àÿâ:ls¯¹úÍŽÓøì©ÆgL˜S™ûŽà¦ÚÏ:ÕÈí<‡ú Î6Üç®C÷ꫜ¬*x³ ;ÕÈ]øº‚Ú«úOοÄX ÜSüe¼b RCÍSÅÚ¬ ®¨ &ªàºê«ÖP—y~ë'Åor9¬Æ¼ÜtüWç¼tùŸ±KžæZ~︢ŠÚµÊý*àŠç«Àhü\ÅzÞü‰‹Û÷䑇 ¬ß¸V€ýª¬)¸¾ãþb®Š|Q¾Ïùª V¹OÆý´G\æž y˜ë òn6(³Ž&v“*ðRxŸØëÁ"°Y㙢{\ž0x®ÖÖ¹?R9ÅÙ¸ \×Àe`£ÆüÁýäž!¬¾òq•>Á<^U_èö·÷üÊø:ÇÅI•ëFÔÀµ×»¼S…+ø¾ôF7f8©Âæeä†2ù±W&—Üêä1§Ó<¯XÂZº [Ë•¿pµkõ>Ѿ(¿Þ­G*hWã ìV`}ýúts)~ÌåÚ â Âóðôãgº1ʰ=ÏWr¾ŠÇå®òÇÜÞcåAWŸs\nh‡gºx®À<7T¬2ëøO¹ßJð3÷ ÷ö£/•¦³×ð\CV‡Jœuç¹|Ä÷ÅÏ<_A½Âóë劜;Àš£|½ã³ °U¸ßísqGÜqÏ{Oæ:ÿ–‘˰Sé«ÝüˆaΗù·òw>¶üZ7×"Ï<ƒÊŒ îý~‘ó¿a-ƒx¬ðw°Yå'Ýü =ÇW¬X‡—y6ñ³}Í]}Ëß>鸌Ïötúw¾± |T€ù2± ð•~Ñå\ž7.£Ö­ N)aM[¾ÍñŸS•¿ÌåŒ2uoUþÞ£î- GðŒã£|…{.ËÁ}pÆ÷“ ¯uíÍ×¹½r‡¡^_ïêQr×"ä‘⇜}x®øù.Ÿ•§¬i+ÀB™8æ9 `»Œ|Pâ3çß΄K¿é|ºÇÈIå;\lî½â^ü_Bõ2cõ^‰gÁÇåopþ)¢¦1À‰5b¿À˜ ÆÈQÈÝ%ðyó5¯qkgžc(ü“ãÅÒ[¿ärã‘vÜËWº|̽ôòw9Ü—à/®yÞ°ø_]|þÊa”gbkå7ºø/0®¹xŸÌù±Ì³wwº³ÂU•ÔñUù%þ‹%î_Òÿåçâ9Œ—Þãâ„ûÑEŽCE —þÄù¹ðÓ.6 ϯ Š_ápVú¯ÄLác{ÏNÀ\KÿÑíKîéÃÚ±_„=ˈ­"s)±©Ûççú•~(¾Æí¹0.ŠŒ ÆøÙ ¾Šo0îmóbáˆ';Œ•Ц\sXàÖÏ!3_ÏxzÏïôïw:î"FÉÝœëAƒËáàâw;Ý÷îÃgE¬WJ<[ó•.Ç¿ÃÅr¾.òàõr{{>¯ýmgÓü\ú|‡çb渤ˆÜWü3÷ÌÅ¥¯puO¹ µ>Ÿk•x¶ê¸[ûôã“çšÏr6ä3ǽ8¢_©¹uUéqnü½šå&·ž*ñœ3Kˆz¡„þü=áï-yfëZgs~WÖŠÄÑS÷3‡ÿÄ­_J<ûõ-.Ç—Pð·XŒ ú‰gêö¸üSüU÷,§Ä ¹¹t•‹í=œÿˆ['ËaŸ~+"÷ÿÆå®©‹Ÿãø¼Èøy—³ 1Áü_"/wßîl½gKä£"¸«ˆú“¿ÓÛ‹{ÄÏ,îÉ¢¬AÎw¹xáLö)2ÿÑ×›gèYw16ö¾/¹¼L[oV,b Uï°†,~ž[£3Š\ÿW$öøLþ¹Àq4ÏŒïrsâ™–"Ö1Åpµ4Ÿïéþ°Ã çÈß/ùl5/Àýcò*qSçr}Ç3§Ä*B=È?´ÏVø|˜çûkÎöü±Ç0ÿ ~ºF^_,ë?®—ÉéÏgÙÌW¬×÷⋼D~C]Éܰ÷qQ|“èüZ™7jž?*^á^ ܳC-Æß°ÏÞy)ò?Ÿã1ÀÃŧ8?íÙ€Ï"Pûð¹6Ÿiðœ+±Xx〽9Òg/’¾ç‹|ž—À§XtqÁs|Rd ÿÔ(|ÆQ`Üÿ¨‹uâžÞç7ž«æY’½gÅßâlUø‘@gä#æž '^æb¹@¾bÝÆ3cá|صâÞ }µ÷Û¹÷Ê<ë?îøŠ¿õ0üÏiÝä0j>.:rùn­´÷º¿ãIÃß@_žGb.&×Lþ¿w¬u çIn$>ù<ù¸ãys‰‹)óÈ'»zÒ°n@½i(››BЛ¿'3¬‰€9þv}¯?×c'g¯/cä¹(ë$á‘s¤ÏDž1¼ØÅžÙ’ï9ÞgÈ{êÇXÇZ—Ï?¸–Ûz“¶ÝÙ\°÷{´G]e®”><×vÙ;óÎór\+íé͹ß'²ž)º>ÎÅ÷¯÷Ʀ}Ÿ-ãœ,ã~…ør¿RìT–ypÿ†Ï×N©ÿ eÞOþ\s>IÚЖˆWž·ÚkŠÌëliw’ØíBiOž&íyÖ‘üÝŸ–dnÇÅ®Þ/^êx¿È£2ñ)Û\%2Ï’±Ï]ž,zŸ,>ãº?‘±Î”q2î=òã…¿µ}TìÌ{U±ÍbËË¥ÏUïˆìÓć—{â×3EGúé±áŽèóÅ"kS®3Å/§‰Ìsd¾F¾»NôaŒWľÏÛ> 6ß–6´+c­&:Ý'6áu¥ŒÏ~$˜÷@dÓžàÖ~{ro™[¢;å¾GüÄ1šòí€uÏòìÇx¼Ctê‹>×ȼè;Ö—ÌÕüÍ!ãõ|ߊ¾Û(Þÿq¢#ß›|>&6ØÝn>m™CGt½Ut¥nÈS¾ =2±ýÀ¸f|ÞØ%•öçˆí(y}ÂAüܹCч×Ò†rø¬‘x¼PôPüÓ”þ—Ëû;d,Æò E^C|z¾èQ“÷lsèO_pͺÌ={4dŽÄ±û¹‡<·‡-~ÇØÙ=2Ñ#‘1ɼ2—³åµ.×¥â÷KÄvO’qï{ž.}½‡âç,ðïé2Æ–ŒyLæW“ùm‹nlÿ¥b‹3dŒsÅÇ·ŠÍ9ÿ*b;Ïm™OGæwŠèy®¼Z¹w¦ø¨ þ:.~äý²È¹XüÅ6·‰í®p.ò›‚=ÝÏû3Ç~N©ØãÑûTÑ#•±Zb3^ü­í‹¤¿çôKåzªŒS’þç‰~WË8gˆ-NÝ(ÿ&ÿÑÿ,ó™ÃI2÷“DÿÏ›^(vOe±_CäŸ"²¯›´Än›âSŸÇiƹörÏŠ>l‹zeÏççÍmÏ”ù~¹Œ¾\©èD=k2î5âó“åó¶øý£Üzšø„±z®´¹$è¿#}j2ïåµ/ö$v®´Ö^{ÜÏ9~‰´í‰}o–¹/6y•Ȣϕv7É<.].”ñé7þ‘Þ-~Ø»³ýkD¶¿J¾§ïÞ"¶Ïä}ËhýpšÌíl™'ÖPõÞ^\"óç{ÆÒ Á|Ù–>¿Hî)þ¾N|ê}µ%²/±Zb·‡¥_[Æ:Sú<_|qšØ‚qHÎdn&GX±á%2×SE?ÚýI¢ƒ¯{øJޏOd—18¿ŽÈôµä9¢ÛSe Új,c\e4ï’[kr¿-~|@>ûzþ»Zt?Iô;.csLî'Çäàºøóöü9½¯“ùÞ%¶¦ÜMñÃuâ¿'åmú„õ̶è~–\O?že´mŠ­Ûòù2£uÄ™2—cFóÏXædeîÔaKlÓ‘9|–Øð$Ñã4™çã¥ÏŸ26È/÷ü®'ŸÉ«¬›×ŠÞÇÅ?O üx¶Œ{¹Ñø­ÉœYCß*s¥~ÛŒÅv©ègD?×çŠo–ûÛb³ ±û•âêT‘yÅŸ—Š=}§}|]KýÿÄ1cíí2ÿ elk´¼Hî¿Oäˆr)c£,}hŸLúpœ¢èvµ\ž§9ÿ´÷m¶dÎMñÉ92÷“æõ‹dŽ>g D>íàë Îñ‰rÿÑ£.÷™w}/Ud.»âÓ/”¹\,6?Yæ}—øññOÝhmt³Q|œ-6"Þ&:Rcê™Ò޾ O5ZãMìM]n0º.9Ë(¿ðª‰m¿Dî{|\*ö¾NÚ´äõç yÍŒ®oë¢ç2vEÚx_жŒÏ‹Dï{ä~YüØ?MÆçE#'|»Q\ÀèÚöc.}£´»KÆj‰ »bßþ^Ñ—¾ºXƤ­Šâ/b˯ý:ë^™'?3ÉgJ{Ê:]äøœ}¡È~ø´)¶¸\úÑ'ç‰Îö—Ëxüžqå±K]ˆ“Ä~>·vD×¶èp‘Ø’ã|‘Œ{žQ¾ä$ö}½èr¹ÌoCÆìÉ8}±?¿ï3Z°ÿ¹F×ZôõsåsKdq\¿$w7{ÿÈÞÜαøþ£{CœûurÿñÕåâOêÂøz@îÝ)sf»Ÿ0®.ö{VôóÛùbŽûJ£5cI¾óöåu£èäs¬o{¹ø3›\i4W.:UŒî¥Ç¸ÿt‰ØñÒç~±ó®ØùÛ®ß9çÍ¡´qQü—ˆØv$v8CÆ >Més·È¬Êgï—ÇI;Ïc¾¦îs;M|t¦´¡½^/c[£k•ºøê­â¯ËÄ^Wˆ­h‹§‰ÎôcMæVßP·¯“ë…â£+el¿¿FÆLåû‘ø£b4_"sôõÌùb¿ÿÁ8<.¶z²|þ4ÑéZ±ÉeF׎ÄÕ¹¢«[Q÷[e¼Wˆü]‹²‡âëe^ž7.½¾À¸bûm£uÀ¦Ñu’÷Û†øâ|™»ÇîIâ§D|àë,úßïm]et߀>¹Pô¼]|{Üèz„¾cíÒø:—ú\ :Ð>7Ë\ا/ýx]/m=v+ò™}ˆãçI;¿·¶#ãS^Kô>S¾÷kJ¿Ovž´ñû*‰®6º¿ëómT—1®›oÈX%±éóä{Ê•ô¡n÷ÊçSÄ'oý}àýsæƒ+D§šèCÎ8UÚ_ktýv¦È¿EÚ¿Xl¿øàl£kgÎå£ù‹ú2ž®™¾N¹B._¯mtï¦/ó¥n#£{¤>ß&¢ÏÝòž2ÇF÷¡®3ºî§^Ï5š§ËrŸsf ø4£ø§?¸~ñÏèGÆcGìr‡ôµ¢k¡®Ø®jtoù*é“­)Î6Êy¬Â}Æ1{›Ñ=1¯Û?ó³ßÿ>×hóûž;®’{Ä ×¿Äßå¢ãéFkʺ|÷‘A™ï4.§Ü,z{¶Ñ=üs]¼þžïSñC"÷/9gsÉw>wù<Ä÷ôý3DÇ{ŒÖÞ~žû⟲èw¾ÑµR&¯Œ×çÊ{ϵž#;Ò¯itÏÈ׃gH»º|¾]ÚG×­ýü:‰ëRÆëEÆå-bÿLæÙ[0?<(~ Ÿn»=MÚ³ïÓå»-£Ï3Šb7_Gûz‹ºzL?$6)‰^ô 1é×Q~ïƒ|w­Ì—¼ñ:£ugKæÇþŒ©;î±ùç>ìóÿ Òö6‘]“¹Ÿ!}8.÷÷n•qéÇ»DÖq±Kft_ÛïPúg9MüC½ýó˜›d|_/=ÇhÞñ\±)2N[û=aæPÿ‹Ï#ÈK—‰-ž-ý/Û#þ¸Àèz’2èûûDƶèu‹øå2s×h¬rlÆåSeÜk¤qwèè÷^ž ã_/óðkÏ Œî[ž&6½[ô»JîSÆŸ¯GikÆË#¢ƒ·9?—Å;¢?uy›ÈöÏoo´§nwŠoÿpŒ±ô÷X¿Iüâë3ÆÕ%rߊ½–~ÇÄÆ]±6ë÷|6Å''ÉŸ/ó|†´û<ѧ)º+6ºÎhÜ\e4oWŒÙŦ䘇ŒÖŒ­¡Ñúuòùéòú‘u®ÈoÝO¼Lî_$¶¨ÉÕ“±/6Ê“¾n`̼ÂèÚìyâSÚüõ±éÀh¾V|ž]/ù=ººØ÷2£u„â×A·]£itMV1º¹atoærùŽm.¹Þb”ã.»ø:ïiÿDFKäq¾7]o^(~ôÏRéâà¹FŸ³î}ŽØ1y÷‰~çÉX‰Ñ<åÛ= >¿ÎhíO›!zòþíF÷»2Ï«Œ>øVã¸Å¯[.2Z?=Ùè³ì›¥u¸Iæõ™ÛÍÒÎ×Ãô¯®ÉqüÚŸ\ûyoàǧKÛãâÏ‘|ÞÝî—×{Ö\¾þõkåï—¹ûçkô×®Ñ=Ük®ç|Lûú³,öâ܃×˸¾&¾Íèó¢«¤ï ¢û?htŸà)âÓ+DVUÆô{ž×6æ=ÿ,ý"Ÿ¯¯ÿÝe”ß| dF÷üs*Æ,ãêòz™Ø€º0N•±‹o|}TyþÙ¤ß':ÓèÚîßùçÇå>å_ØÇï7pïë…bû†¼VŒ>Ç;Uúù}ìÛÅVO“±wåz¾Ü»[æâÏNÈïöæâ÷8Ïï¯7z¦è32º·ák€+Å.Ì‘Œá¢Ø¸$ú¾OlÍ|UÿúÚó âÞÿ±u¾BìqŽÜ»Çh}~™èw•È8Ùh¬}Î3ú<—¯…}}þb£ÏDý9¿ÇéךU±×9¢?e0O´E7ö{ÀhIœ\ ú§bÓ;ƒþþÌ–Ìÿ"£Ïë>Ãè³{Ê<]Ægîȼ¶Eïß*ïO—û/3š#8/_¯1~žèp¡ÑýFÿ<ŽD¬^-6j‹~M‘CÙ}/0º6{†øðçÃF9|×hÎù Ñçd£çrüù-ÚjGt¡¬GeŒŽÑ33þž_¯at¿ØïoøÚòñòý¹b·§Ê¼N•ϧËü½ý^etÏãñb'Ï_Ô¹*²㌡+ž+y¾Ñ|â×׊ü'‰ý¥ÏlsŽø2¹ÿt±Ü»D®3E¦¶V1Zg¾Åè³Í—HÛwå}^Èó»çDÿløù2/ÎÈýü9‹·]“½Ñ8 øç”ÇØb<{.{ªÈ»TtºTúú:éFéÃñ‰¹ëŒæÒšÑç( £uÀŽÑø™2.}Éú‹Üù"™‡?×çÏntŒ>Ûôëô‹Å†žû©ÓƒFÏ`ù5äíF÷“/”ïÂýr_ø=ñ«ÅwËÜ8Æ}2/ÿl¥+~»Vú?AtyºÜ¿Ýè3¬kDæ92>Ï­Ñçb~ßÏð{õ™ÑgkþyˆÇ즴9Õ(ÿn‰m©³?«x{0ÖU~ÿŒ}ý^ûDv*÷|¢]žhô›æí9’纞ø #sók8Vƒ:'îß½Kú\&öñgeü~ý5F9ì¸q¹Çï}œ|ï÷%hóg‹­ü3Çc¢÷"ó¸ø`KìçÏD<,óïIöy‡èÿ,yåý»¥?1Ä3IŒENÁŸ9ºDÆç}þùXöÏ?¼iS®µüYÊ~¦Ø–úÄÅPÆóç üL¿sºŒ¹kô½ÇÛ²–ñk·ëŒÛ×¹JìÇyŸmtŸîÉbß-Ÿå×,~½Î9¼Yôöç/‰á®Ø¿és®èä÷’Â3m¾ö9UäcŸ&÷ýù$¿¾áë"Ÿs~­èEû¼Ddôeìe¬DæÉ:Ñçs¿†ñõñõü@ÎÀè3hÆ ñüz±SÛè~ë%ÒçV¹üwþ¾ß£lÆ3ãqs»èyºøúb£Ïy_hô¬×)bËçÈw牮žkï6Š7_Sùçõ}£{«Œ¥×Ý—óg\¬G2ÆUb· ¤¿?{ªøð i®Œq¥ÑÚÝç¯Dä|žøqSú\i´–÷çn/¾!ÿ’|r›´ŽÌé+e~È=Žç¹»%úŸet¿ØŸ“õçõgôÌ“_ûnÊü.6š›èÆGUæxø’m}œdFŸsúsѾfòõ…_Ã<ÁèÚbËè¹*†‰ö¾AæBYŒ‡;å>íÿLyõuÐ-F÷yùZedôYÕ>7òç8›F×§¾&óÏ L0WÿLsÚ1ÊÇþ̳?¿uŠÑ3žþÌK]æö ±?ug^ô{y£ûaÄêKe.—ÉxÄÙPìòM¢WCôz™ÜgßëÅþ|—_Çñ=c‘œüt±½6ëc—cmr‹ÑçÅ~Ÿ£'ö8.÷Î3zâVÑÍû‹²»â‹Kd¼g=Z{=;ð íëφ+sð{|_4zvˆ>!>¾@dôŒ>K{†ôa sŒ‘ÑóLžS™/å2><‡]'÷•þ»¢uµ|w³øèZ±‰ß{>Glq‡ÑçÇ”ãÏkzªØÄslÓèÞ²ß+÷1âÏ1øß.ø=GVÕ¯·n0Êe¾óÏ8nÝŸ-ßùgqĈ•¹_!mÞ!sØý¹äžÑ½±ØÜw›ÌíZ£ç¨/–ö¾îiÈ}ÿœŠqô4ù|’Ø}Sôõúø3§×‹ÝwŒžû¯Š¼»Ä>Wû=N¿žðën¿žmʸgнü™dÏ%wÉøœãSD–ô£{%U£çÕi#ÿhÓè9QŸ—³‰¾ŒóWIߊèv§èÌ{\÷]m´ÖݹÜoôœ÷ƒ?âÏÁûóãç‹/7Œ®•KŸT|éë‹ãòꟅùº×Jß»Å?~†õ:ãém¢‘¹ž%óñ¹ÒçŸkæÒ¦Œõ]âOktOÍ?w ÈáäÞÒŸ©½_ú-¾¹Yî?jôþ¶øË碆È÷ûJ~Ý™Ýß]ßûµçJ>¸Øè™bÆÌƒâSúì£ç˜O2[F÷qΗ×ÅmÑ›÷ü¹Rÿ¼ž¿o¼Qdú=\_7–åžþxQÿŸc´–÷ÏjN6ºG–=›ãÏåû}·Såóýb¿·åÏf^+ãueÞþüûØèYhŽ’Ø¼ÔèÚà"‘q£|¹øÆÿ΂våyG~ö2ƒí}ýoE±ÀXôûi—ˆ}î0Z#p~/ÐÿFÁ?Ïöõ±?·Âvȇb«ÛÖ ”Qûp~þ,~XóûgÎÔyGÆ¡üÈ¥FŸ\-:])6óÆ×çúñžß'¥ßo|Nè‹ÎÛrÿ2£õ5óÏ=Òןs¼^úùç:7ÝŸzœÈe\øó½ ïåbsž[0"›þóÜê맇EŽ_“£¯2zVÁ¯uùù·%özØ…1ð~£g.“׋E®ßS½Àès4ΖíXOúgU/þl®çž®ô‹¿ü™+E'¿V¼Ñß6ѳ!~dl|“ȹØêùF×þ7%±c€qòñMEîS®?Ëw†ÑßЯÜûòçÑ<§s.O7ºNækøìÒc—zùšÌŸ3¹ÇèoÍ(Ûïñûß[\/ßÝg4¶ýšÔÿn†¶ñkzÚ‘þÿt£5ÔÉFsù®øgC|hÖ'É}Ñýf£Ïüyp~GÜû³ÎþŒu{…Ìöò5½ß¿ëÉ"ëÒ§#ßÑæwÊwW=r¥Ñýqêz¯|Gßž©È:KôògyØÏŠ_"sâÆÕ›ÛteÌ—ŠÍn=üo5.;Ó¾‹É\y3Œ‘S¤ýùâ£ëDg¿7Ò½2Ñý2çTùþYb£m‘×–ñvD×»þöÏ?¾TìCLäõn™ÃçÈÜn2ú»HŸƒ½Íž,²®[eô·…‹ ŸëüyÌŽØáõ¢«?'Ä i4_’ÛΑù{ú}‚ DoÞòeFóÜ52ö3DÞ¶èñ\éÿ¸Þkôœ…÷K]>S—ŠÌ›>eÌ3nýÙ{êÁ¸óÏùŽËäzÖηˆ?šò»÷ð¯`¦ÿ¦¼?èï ~ÓÆŠïÏÚn‘¿ýdö¹—÷þ h—§Ë,sŒõœÇÖ³þÍê“EÆ:¨Ý¬v˜gìUü­ZÖ<¾žµÝ¢XÚ/–æù;¨èÛøý<ãäµ]EÌıw˜ñ•÷·Ê™eŒel9 —dßY¹l¿{ëàƒtXE .:Æ<9f–ñgÑm^|äçý8aÑ$þ<®gÕ÷ þ«Ò=¾¿,_ÎR¿ÌóÝ,ùdÞš3މyj®yþÕkÖqV‰ÍÇÂß*¹uœì—3–ÁתÚÏ;ÏuúxÑœ0Ï÷³´_•ï÷wŽžÕ«öÉ4ÞX´fX–g•3ëß<1”—Ógg‘¶«˜ç²>Y¶Ï2uÛªs`Øæ ™§ÍAmýw«â×eí·,ï,ˉËÖçóþ-3Us⺹}¿¿yyq–~yü9Kî›6Æ~:åÝ?Hö=uòýÖñ7-ÎÖ©ã"<=¯n‡Y“-ÒÑXžUÖ¼¹u¿{«Ž…Eøí ]–ÑqþXVæ,¾Êã«EÇ>¨ý,þÏûî0ì}F–•µl^sþ‰ø›§öŸÖoÑ:eÖñçi· ¬­{0/ßæÅɬuÈAíæ©—æýÛÏN'²~Z…Œy¹e-[ßÎÓfUqíAýV­Ë²1vP]±ß|瑽ŽÚÿ Ø›Ö~¾?¨Ïºæ9oÞZ4/Òö°ÿæ™÷*kæyþ]c¢k•:Ìû9ï»ekyúÌšÇÑišåep¿(ï®’k­?gí»ŸÞËæŽy¾ŸU‡Uóß*óã²Ü¾î¿ExeÕòå„iŸYOäµ]gÓ¾›u®Ó0¸ªuÊA9o•¾ž—sæí·Ìß"ü¶*._7”ªk®em“—ë§Ý[FÎ2cÌ3Ö¬ëekÈeü¸HÌ­ÚþëÌé‹òó´ï–ÑuÞút^ŽX†3óxcY‹äâýdÔvÖû³~ë1k›ƒt^UþZ ÅñA|µÌ¼Ö™Cãñ÷‹Áep4¯‡1ÎóQ§ÉÙ¢ÉÞ/ñÙ¦àÿyß‚ÿwó þp(ôü›¾3ðoü?RðÿCnÁÿ÷Åö“wíÉ»Îä]wò®7yן¼LÞ 'ïF“w‰ŒÎDFg"£3‘Ñ™ÈèLdt&2:‰ŒÎDFw"£;‘ÑÈèNdt'2º݉ŒîDFw"£;‘Ñ›ÈèMdô&2z½‰ŒÞDFo"£7‘Ñ›ÈèMdô'2úý‰ŒþDF"£?‘ÑŸÈèOdô'2úƒ‰ŒÁDÆ`"c0‘1˜ÈLd &2ƒ‰ŒÁDÆp"c8‘1œÈNd '2†ÉŒáDÆp"c8‘1šÈMdŒ&2F£‰ŒÑDÆh"c4‘1šÈMdŒ'2Æ㉌ñDÆx"c<‘1žÈOdŒ'2Æ^F©}Åú¶­o;ú¶«o{ú¶¯oúv¨oGúV¥µUZ[¥µUZ[¥µUZ[¥µUZ[¥µUZ[¥uTZG¥uTZG¥uTZG¥uTZG¥uTZG¥uUZW¥uUZW¥uUZW¥uUZW¥uUZW¥õTZO¥õTZO¥õTZO¥õTZO¥õTZO¥õUZ_¥õUZ_¥õUZ_¥õUZ_¥õUZ_¥ TÚ@¥ TÚ@¥ TÚ@¥ TÚ@¥ TÚ@¥ UÚP¥ UÚP¥ UÚP¥ UÚP¥ UÚP¥TÚH¥TÚH¥TÚH¥TÚH¥TÚH¥UÚX¥UÚX¥UÚX¥UÚX¥UšrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t”K:Ê%å’ŽrIG¹¤£\ÒQ.é(—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIW¹¤«\ÒU.é*—t•KºÊ%]å’®rIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô”KzÊ%=å’žrIO¹¤§\ÒS.é)—ô•KúÊ%}å’¾rI_¹¤¯\ÒW.é+—ô•KúÊ%}å’¾rI_¹¤¯\ÒW.é+—ô•KúÊ%}å’¾rI_¹¤¯\ÒW.é+—ô•KúÊ%}å’¾rI_¹¤¯\ÒW.é+—ô•KúÊ%}å’¾rI_¹¤¯\ÒW.é+—ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%¾H_ ¤¯ÒWé+ô•@úJ }%È@ d 2P( ”@J %È@ d 2P( ”@J %È@ d 2P( ”@J %È@ d 2P( ”@J %È@ d 2P( ”@J %#å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d \2P.(— ”KÊ%å’rÉ@¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •K†Ê%Cå’¡rÉP¹d¨\2T.*— •KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¤\2R.)—Œ”KFÊ%#å’‘rÉH¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ•KÆÊ%cå’±rÉX¹d¬\2V.+—Œ'\Rn_1!¾oï;Áûnð¾¼ïïÁûað~¼ä¶¹í@n;Ûä¶¹í@n;Ûä¶¹í@n'Û äv¹@n'Û äv¹@n'Û äv¹Ý@n7Û äv¹Ý@n7Û äv¹Ý@n/Û äö¹½@n/Û äö¹½@n/Û äö¹ý@n?Ûäö¹ý@n?Ûäö¹ý@î ;乃@î ;乃@î ;ä¹Ã@î0; ä¹Ã@î0; ä¹Ã@î(; 䎹£@î(; 䎹£@î(; 䎹ã@î8;䎹ã@î8;䎹_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU;à«vÀW퀯Ú_µ¾j|ÕøªðU'à«NÀW€¯:_u¾ê|Õ øªðU'à«NÀW€¯:_u¾ê|Õ øªðU'à«NÀW€¯:_u¾ê|Õ øªðÕÿÇÞ]@Guµ} Ofp îÜCxž3†ìÜ=@pR J€Bq(ТÅÝÝ%…bÅÝÝ‚»Û¥oÎ>óÏ¡s“Ê·¾»î÷f­vslŸ}ì?“Éoö&È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È+‚¼"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(‚Œ"È(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†ŒbÈ(†÷T yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW yÅW y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•y¥A^iWä•yeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lW6È+ä• òÊyeƒ¼²A^Ù ¯lWvÈ+;ä•òÊye‡¼²C^Ù!¯ìWvÈ+;ä•òÊye‡¼²C^Ù!¯ìWvÈ+;ä•òÊye‡¼²C^Ù!¯ìWvÈ+;ä•òÊye‡¼²C^Ù!¯ìWvÈ+;ä•2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Êe‡Œ²CFÙ!£ìQvÈ(;d”2Ê倌r@F9 £QÈ(d”2Ê倌r@F9 £QÈ(d”2Ê倌r@F9 £QÈ(d”2Ê倌r@F9 £QÈ(d”2Ê倌rÀ{*ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊy值r@^9 ¯WÈ+ä•òÊ y儼rB^9!¯œWNÈ+'ä•òÊ y儼rB^9!¯œWNÈ+'ä•òÊ y儼rB^9!¯œWNÈ+'ä•òÊ y儼rB^9!¯œWNÈ+'ä•2Ê 儌rB.9!—œKNÈ%'ä’òÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'äòÇ ùã„üqBþ8!œ?NÈ'ä òÇùã‚ÌqAæ¸ s\9.ÈdŽ 2Ç™ã‚ÌqAæ¸ s\9.ÈdŽ 2Ç™ã‚ÌqAæ¸ s\9.ÈdŽ 2Ç™ã‚ÌqAæ¸ s\9.ȼGrAþ¸ \?.xä‚,rA¹ ‹\E.È"¼GrÁ{$d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q.È(d” 2Ê傌rAF¹ £\Q`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçDW`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èÀ¢Xt‹N`Ñ ,:E'°èþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ ü9?'ðçþœÀŸøsNàÏ Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜ssN`Î Ì99'0çæœÀœ˜sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONvÈ%°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–ØrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀ“xrONàÉ <9''ðäžœÀrCN`È 9!'0䆜ÀrCN`È 9!'0䆜ÀrCN`È 9!'0䆜ÀrCN`È 9!'0䆜ÀrCN`È 9!'0䆜À“r <9!''drCN`È 99!œ?`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9-'°å¶œÀ–Ør[N`Ë l9¹ —À™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™8sgNàÌ œ93'pæΜÀ™38sgÎàÌœ9ƒ3gpæ ΜÁ™38sgÎàÌœ9ƒ3gpæ ΜÁ™38sgÎàÌœ9ƒ3gpæ ΜÁ™38sgÎàÌœ9CŸç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ9g0ç æœÁœ3˜ssÎ`ÎÌ9ƒ3gpæ ΜÁ™38sgÎàÌœ9ƒ3gpæ ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'g0ä †œÁ3r7ÎàÆÜ8ƒgpã VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8C¿å nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆú-g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`È 9ƒ!g0ä †œÁ3rCÎ`Èú-gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœ¡ßr[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ë<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ“3xr†¾Êl9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3Ør[Î`Ël9ƒ-g°å ¶œÁ–3ØrOÎàÉ<9ƒ'gðä žœÁ“3xrOÎàÉ<9ƒ'gðä žœÁ3rCÎ`È 9ƒ!g0ä †œÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆú$g0ä †œÁ3rCÎ`È 9ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7Î`Å|8ƒ g0á &œ¡qÎàÃ|8ƒgðá >œÁ‡3øpÎàÃ|8ƒgðá >œÁ‡3øpÎàÃ|8ƒgðá >œÁ‡3øpÎàÃ|8ƒgðá >œÁ‡3øpÎàÃ|8ƒgðá >œÁ‡3øpÎàÃ|8Cã VœÁŠ3ô=ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒgpã nœÁ3¸q7ÎàÆÜ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â VœÁŠ3Xq+Î`Ŭ8ƒg°â >œÁ‡3˜pÎ`ÂL8ƒ g0á &œÁ38pÎàÀ8ƒgpà œÁ38pÎàÀ8ƒgpà œÁ~3ØoûÍ`¿ì7ƒýf°ß ö›Á~3ØoûÍ`¿Œ7ƒëfpÝ ®›Áu3¸n×Í`¹,7ƒÙfpÚ N›Ái38m§Íà´œ6ƒÓfpÚ N›Ái38m§Í`³l6ƒÍf°Ù 6›Áf3Øl›Í`³l6ƒÇfðØ ›Ác3xlÍà±<6ƒÇfðØ ›Ác3xlÍà±<6ƒÁf0Ø ›Á`3lƒÍ`°ܵîZw­»ÖÀ]kà®5p׸k |µ¾Z_­¯ÖÀWkà«5ðÕ8j ì´^Z/­—ÖÀKk`¤50Òi Œ´FZ#­‘ÖÀHk`¡5°Ðøg ü³þYÿ¬ÖÀ?kàŸ5ðÏøg ü³þYÿ¬ÖÀ?kàŸ5ðÏøg ü³þYÿ¬ÖÀ?kàŸ5ðϘg ̳æYó¬yÖÀ™öùòŸ·¾,QËànÁ~­ºw ù2õÉýßkkÅéнSk//kª?¶øò_²/ÿÝ HàõŸŸ¯ÊÔQ¥|¢OçÓËÌz™Ç´\ÍO¯—þÑ× °E¯ÇØÎ}}c¾Í´}zÓ´Ÿ©]ªôÑËÑKóq|ÕnËUûü¢Ï7–§6µ+—©=™£—Ævê<«vÚMóÕö6Óvé=ìß´ýWí7Õg´×|ÝÕy0׫ÎÓúj=/Sýæó˜>úvÆ|óõ2ß/þ¦íí¦åæë›Ç´žj_.Ózê|˜Ï·é¼|Õþ¦ÕöÅLÛ©vš®·±<—i¹ùúšÛ¡Î‡ºÿÕq3Õc~>Õzê~Tí5Ý÷Æõ0Ÿws»Í÷•y¹:ïæã7ŸŸ<Ñçû5·Ëô<u}ÕúQeÿ7ü?þl£^u\æûR_߸æ¼1çƒùþTë™Ûée*Íùf>_æéÌ曎ó«üSûSçÛ|Ÿx8¿¯«Ú9‡Ô|sýj¾éz„G•êz©ód‹¾ü«çMÕŸÇCi®GÍ÷2µËô<˜Ï»ÑnSÎx©×É®¾Qóû†E•cÆF•í¢6 1IŸUq@׿Qe}½öúz-#¢Ê}ú{½¾!úüZz9`cTùÞžAùôõ¢P=<ª¬§/ïr4ª Ôë¤/ïu¡zÍŒ*Ãõåôv6Ñ×륷W]÷~z»Õõ _U6ÔÛWYß®©^Óc´ÞÎúú~†œ*ûêÛµð*ûøèǧOtêÛéëuÑÛ®Ïÿ^?î"õzßD•­gFoO'}ûÁC£Êšz;Ô}×\_op¯¨²»~œ}ôíÚ_*Cô²º¾]ËæÑëk§Ï¦Ÿ—oõí;èåp}½®úqªûRí¿ƒ¾ü[½ýC¢˜€+£¯ßJ/‡ê÷ƒ:?môzZêí¦_çÎúòïõú¾ÑÛýƒoôãø^?þŽúöê>ꤷg ¾žPíЧÕ}ÝO¿ßÔý<@¯¯¯¾Ÿú|u¿´Òç·ÐÛ>4úq~«××]¿îê¹o¥·§³~^Úêõ„ëíPÏo»ðèó»èóÕùì£×ÓF?Þúùª¢¯7@¿ÛêËëëóÕõ ß‡áú}×¥¦¾~€¾ý:÷×Û¶2ú´ñêe_ý¾ëîý¸zëõµˆ*kèûï«·¿‡iêzJ}½îúòúúê¾îª·§•>­®C7½êzuÑÛ§®wO}½^zýaj=}?Áútoýú…éûm£/ï¦og\G}=•WÝõíÛèË{^Õ—ëç­‹¾ÿVzÙZ¯WåF˜ªO¯GÝÿÝõz:{Eoo7µžj¯~¼ôú;ëÛ…é¥ÊÓæÍ£Ê¦ýªéNúòÎú}ÔYm§×SGßo5}?ôv©û§£¾}C}ºƒ¾\=ŸgF•!zª×Ó^Ÿn­·CÝ·ê¹P÷Q}º•^_;}*çÔùo«Ïo¢¯ßZßOë«Ñۢ׫?×^êuRŸP¥ñ:ËùÿWÊÿkÇÿ¿u˜ëiµ=ÿ_[ÿ»üêýu ëyšŽí~þîþÿ­vÇ´½§éض/¶û‹m»cjgl㯶ÛSýÿV»c»ŸØÿ_mwl÷ç¥^'k‡Gm_K¯§ª*õùê}§y=õû¡ú}GýXÕ¼žWôzÔòšúöµôéj{ëUÔ§+éË«©v©ý˜Ú«¦UûŒöèÓzYÑÔ>uêý®ú=«¦©žš¦ý¨úÔû*£zYÇ+úöêxÕ´:.uu¾Ôzª}µ#¢§j§ÚNµu¾ã3ÍWí2êÓ§ËêÓê|«û£’>­®«zk>_Uô²œ¾^yS©êSÇ­®“ZþÕyÓKuÝŒû2 úúÆñ›[_¿ši=uj?ªýj?ê÷8u¾Ôq©ë êQíôо\m¯îuß×!üÏ·3î7óñ™Ú§î3uœuôõÕõTÇ£®cu\ª}꺫ó£ÎKÓ´ºÎj}u\æûÖ|ߨzÕ~Ôòðèǫj§ù¾Vë™sFÍ7î³€èë©ódÎà z©îCu®˜žOíŒé>2×ãé<šÛÓöæýÄöy7glŸ[OÇi®ÏÔnÃ{:î˜öSÎÄtÅtß{¸î^êu2¶ÇÓuéu2¦6?ß1Õï)ïÌõ™_ß<÷ê¦í=½n—5­gÎ/óù‹é}…§ûÏœû1åÝß=o¦÷ Æýìi»˜ž§rßcý1=×±}÷Ô.óu4Ÿ/Oyìéü›'¦<4¿ïòô:àéyðôúç龈éùõô~ÑÓu4ß·ê9ðÔ.OÏGL÷}l_ÇÌÛ©ùÂÃöžžÓؾ®Ç´ÿØÞwžžÃ˜î‹˜ÎWL¯Ãž^Ocû~%¦çÍ|^<å™§ëÓýæ©>óqzºþ1ûñR¯“ãGÍõ"ã¹ÿñMôépÝã+G~4ª¶/ª ‹úE5`lœhÛÛëõÌ­ù§ûQÛtŒˆ*U»Ô|}ÚØnt¯èÓªjý±£ÊAQ¥úƒZÞ+ úö##£o§ÖSÇ­êQǡڧ·ÃØN÷úF½ªžAú÷úF?O?—ýÓãûj{µ¿ Qç/`ˆþ½î¾Ñ÷§Ú§ÚÓ§yôíŒó£—?Nˆ*ÕùS×ÕX¯DTù]xôv¨õÌן£ïWµK}Ÿ&$ªÝQ§Å+`bXô󡾿a´Ó_/ßDo‡Zo¤¾½ù¾P×­sxôù]¼¢M»ïƒ°èë ñÿÓãüj;µu?¨ûÉ<=ÀÔþ±éÿ¼êûƒÓG?ê{TÝgFo§ZOu^Õù3ßßcôûúÓñêë™ï'cºëب킢Ÿ?£^u¼êzªë®ŽGÍWëëÏ—QÏPçŸng,é:¨ãR1úvýVêû3å–ºæäŠ^ªë©Ú­·×¨oøU}?¦çEÕ«Ú¯®ƒþœºŸÏ8ѦzU{T©Î§i»¯žwµžÊóó®ÖSϱéüíVÛ©ûSŸoÎÝ€ñ¢×7Ë?úþvþ<]§¯æ«ú<œwOÛíWßsQùhÜoö7Ûô:cäÔ}Në«ïu™¾×iþùêûÚþzYJ/‹ë¥C/Sëe€^ª×q}?j¹ú>¨^¿Ñ®¢úz¾¦ý0µKÿ~h„ú~¦OgŽ>½ã‰i¹~œê¸Íû/¡Ï÷ÑKgôöõýäâ¦íõýG¨í„i;Õ~U¯?ý1¾wìgª_?Š^o„úž¯¯^ÖŠ¾ÝWõ«õÔtÓrýz篤i¹¹¿޾uÿ©ãÒ·7¾¥Î§ËëOŒë£Ú¥ŽËÇ´^@ôúÕyÿêú˜û¡PóÍý ¨ýi¦R/u™žOU¯ú1®QS=æçÆ|W6­¯Ú©Ü­úvéðhõ¨ã0]Ïõ¼–ÑK_ÓþÕqy™JuŠ›ŽK-ÏlªGµ__-úzªc{u^|Mû-mšVíWçS}_ÞtíP¥j‡~žÌý¨Õã~1_ý|ªóªÖS÷¹y=ã¸}£ïÏü|õz¡—jj¯Q¿§ãUÇ—:ú|óßÿŒýèÏåWωÊ-U¯Ã´õ=Zu^õ¼‹PÓê:›úÙøófë÷kê?owT_=ÉTÏ>ôÏóŸ>{ÚvêÚMý»[ÛÿôïóÇ¿ãuíÜ­{W}ÊÜZ-°v é¥ÿ3~X¿¡­õÉ_&ÛwéªO' îmF‚Ž!ÁüZwPÛ÷ìæ×!´kרt:ôÇ9Õ7õ5¼UKÞ¤þ¡z?òÖÔ?T¿GÞªÓ#oÕã‘·SýCõ«d1z ³ý…YŒÞÂ,F_a£§0‹ÑO˜Åè-Ðbôh1z³ý”ZŒÄ,F£'1‹Ñ˜ÅèEÌbô!f1z ´}‰YŒžÄ,F?b£1‹Ñ‡˜ÅèAÌbôf1z³ýZŒ^-Fb£71‹Ñ_ ÅèUÌbô)f1z³ý‰YŒÞÄ,F_b£'1‹Ñ˜ÅèEÌbô!f1z³ý‡YŒÞÃ,Fßa£ç0‹Ño˜Åè5Ìbôf1z ³ý…YŒÞÂ,F_a£§0‹ÑO˜Åè9Ðbôf1z ³}…YŒžÂ,F?a£—0‹ÑG˜Åè!Ìbôf1z³}ƒYŒžÁ,F¿`£W0‹Ñ'˜ÅèÌbôh1z³ý‚YŒ^Á,FŸ`£G0‹Ñ˜ÅèÐbô hqû0ú´=ZŒ~-F¯€£O@‹Ñ# ÅèÐbôj1ú´º‡7¶ºG7¶º7¶ºÇ6¶º‡6¶ºG6¶º6¶ºÇ5¶º‡5¶ºG5¶º5¶ºÇ4¶º‡4¶ºG4¶º4¶ºÇ3¶º‡3¶ºG3¶º3¶ºÇ2¶º‡2¶ºG2¶º2¶ºÇ1¶º‡1¶ºG1¶º1¶ºÇ0¶º‡0¶ºG0¶º0¶ºÇ/¶º‡/¶ºG/¶º/¶ºÇ.¶º‡.¶ºG.¶º.¶ºÇ-¶º‡-¶ºG-¶º-¶ºÇ,¶º‡,¶ºG,¶º,¶ºÇ+¶º‡+¶ºG+¶º+¶ºÇ*¶º‡*¶ºG*¶º*¶ºÇ)¶º‡)¶ºG)¶º)¶ºÇ(¶º‡(¶ºG(¶º(¶ºÇ'¶º‡'¶ºG'¶º'¶ºÇ&¶º‡&¶ºG&¶º&¶ºÇ%¶º‡%¶ºG%¶º%¶ºÇ$¶º‡$¶ºG$¶º$¶ºÇ#¶º‡#¶ºG#¶º#¶ºÇ"¶º‡"¶ºG"¶º"þòO÷Þœî½9Ý{s¹÷ærïÍåޛ˽7—{o.÷Þ\î½¹Ü{s¹÷æÎ÷ÄV÷ˆÄV÷€ÄV÷xÄV÷pÄV÷hÄV÷`ÄV÷XÄV÷PÄV÷HÄV÷@ÄV÷8ÄV÷0ÄV÷(ÄV÷ ÄV÷ÄV÷ÄV÷ÄV÷ÄV÷øÃV÷ðÃV÷èÃV÷àÃV÷ØÃV÷ÐÃV÷ÈÃV÷ÀÃV÷¸ÃÖ¨a‡ÿ¬ƒÇø=Bº…t îäÕÇãï<ÿøìeØ_5¹ÿ-ÿo–^ÿ-ÿ[z.¿äJÒ/ÿæ9÷?ËÿÒÖXüçü÷OÎÉÿäyŠÍqÿœ›?^šr«þ˜÷ÇKT‚®»w 5>¹Hصcp‡-B:¨â· 1>‘ˆÛ!¸KëŸ?ÄkÜ¢[è†üãµð?/tµ½þóÐ{eTÙ!<ªl¦—ª³CõaN >­:ÿê:3ªTìW"øêå7z©˜k|ø£>\TÜWª/Ó«›FD•AQe‹¨Š/a7Ñç«¿¹Uª?Z¨/­«Î.Õ‡ñÁÍ£Jõ¡œúK}ÈÚùIôõÕ—×Uçê)ª½êC.õ%ò^z©Î›¯^ªÎÆJ™Ú¯:9T ¶Ôç«ÍT'†ªS@µ¼½¾|@E½Ýͣצ¯×VŸVصÒë ¾UªNÊèóCÕ´¾Ýý¤õéºúzôí;éëÖÿØÜ(j¹ñB9Pÿãf¯šQ¥ú°VÝ=ôùýüõöúFß^Ý/ªóÕ¹¡úP^Õ£ÚoêìØh¯—>mþZ}x¤¯çkšV%¨ý¨ëa 4}yIS}êþ Ô×—¦õšQ÷]í€èû/­—ªÓ ÕÞ`};Õ¹_]}¾þÇ£“Õ>UËðèë©ÎÔyT?¨N›ÏŒ*õëi\ïžú|ý^ÿògõów—ÇöK:ÿÛo ÿÿRªŸºý?­çÿkÇõ­]ÿâñE½?ñ„vÍóUž+Lë W7­gFŽžöc®×<ß¼¾¹^5]É´žZ®êÑn3†UÇm®G—ù¼Ä„ÑÍõ¨å´žxº/Íí3cݺ~\æë®ê3_wUª÷9æûÇSi>nõ>׌Ù=/óþÍÇg>/žîsõ\x:?úv^*OÔû'õÇeõþ:‡^æ1-WÇ¥Þש÷s øèe5Ó´¹>ó~¼Lëû˜ÖSû/dZß„]¾„Â<¨ŸiÚ|Üj¾z_«×£rØ@(¾¦ù¦AmŒý+Ü£Þ_ªó§Î—¹>¦ióñù›–›Ç<˜CjÓ´—©ô5µ;µi¾ypÕnÚúê<¨å6Ó´Ú¯sLûQËÕy6ŸWózæiõ{®i¾*ÕþÌ÷£¯‡vz¨Ç8^…sÔïS¾¦õÍ×Ñ|?›í1/Wõyz.Ì÷‡y0óv~¦ió}bÂ_Æqšñž©^*O̾)LetF§/7:GÔ§Ugê÷-#·¢o¯Jõû›Ñ ¨>_}ÞRÊ´¾¹³>s§˜ª]ªs@£SCí3wz©ŽWÕ£Jcziî$Oí§†i¾úJŸ¹3Ou¼æÎÕùV¯+æNUû‚Uí2:cÔKÕy¤¹3Sõ{tYÓq{šêUíQÇ©VÓúê8ÕþÍ¯×æNS«™ç›Ž÷«NõÒ|_ûQÇ¥ö¯ïGåÅW˜ê¥úüÅèÏ3®·©]ª“D…JU½ê|™Ÿ#ãþÓKõ\çQ-7µSÝßæÎ5Õ} ŽËÜÉ¥ÑÙ¬š¯¯g<j:"ú|7ÖüOž¨ 35rÎÃü˜~b[_LÛÅvyLócÛîØ¶ã¯ÖówÏÛ_]îi½¿{ü1µ?¦ýüÝóø?µ¿¿{?ÄôóWïËØ.ÿ§û÷´Þß}^<ï'Žýã]Ë(D·.ÊÆ&øãOHÝz‡)ëÙÊÆF^[¶ n «¤Ö¥í]ú‹öu˜¾Æíëþõ_ûú_ûêõ_ûú_ûªÿóÒ¾z÷8q×î]z„ô óëÞõKž[¿¼3íò›³G¦–\6IYŒ”E[5Þ1sx™xþŽÍ粆ŠñËx Ô.q¹×âôùû-§NŸLø¸ueqì§R Ä4qdCÕeóü뉃ýÓ5 Ì#~§ÎéÆ/X$ö5;<âð"±7,çtO'‹=ï:õx”Ñ.öÚ*eXûºžØ»øÓ‰tö„Æú‡Š¬ž@£§Š£ÎÆ {VÏ׫énñĉßÖ¿Ü¿á¾8§ÁÑiâô•û¿NZšRœmzåQÍõƒÄ¹ÙÏ'^¯œFœ_šý×íÓˆ M~Ó1¾¸°®^â¶IX\xR>i*éÅÅ‚öÆ“í/&Ì8­GqiïÑÏ•n;Äåcé§(^_\M´ôūֽÅÕ-cûçö//®MZûÓÇò Äõñͺµ7–Lh\iî|qóZP‰ú}ç‰Ûe:gÿ¸q‘¸sìN@ëé«ÅÝï¼ü«ì&î‡T-^©Í&ñð‡ó]z®î$ßÚ²cBÊíâiùÛ%¿¨.žm\¶±^ÂgâEý¤­º÷ÿE¼JUqZõ‡õÅëÂcs–dïR&Ì—§Öbñ¡ZmÿɥħuÓ—<|°Qz-}™_xU– =S„–¸.­ .†ú¿©!ã•jýÍ›2¾_ø¬3ïâÊ„y½Ïº¾Ù.—Šü´¬z˜LÚ¿ñ/½¨L~á·8¡²ËEZ}7£Ñ™rˆ“r‡?“©^•XØC>iôOR"ÒK¦+øzæ¯é\2ýÛ%‹‘Ÿ…'y·p½Ì’yäÂc­ÊlßÅõ¹±¸´Ìžü§âÇä”9‰‰Û¹|E¦.jÊÜ÷|‚—¦È$óÞ®p®æíd²@úTmÓ2Yh`éBëRZd‘Ü®Õ*²Wõr”O;ÏKÒð}·÷é)5zßóøåþÒö>_W¯ÏÒq;¨ÚÇM£¥ëÕ²º[†,—%Š ßz/yO)Æ÷ý¤Cs0¶u…vGûÊR÷îßëµ´›,SÿÜÕP­ƒ,{zdƒÐ.9eùü¡³ö’-÷[7$”•–¯ø\iÀMY¥õ˜Ë/å’Õ ‰))Ò~”Õß×YžjJ\Y³íœ6YCGÊZ FnßXÇKÖ®šÓþf‘²NÀéFßFÞ—öºº8³ÉºÏÝX¹p€¬—í•(±ZÖOQ=èÍá²þ§#E¦$X]´ÉQöAkqlõêu ?ǧËÉë/øˆÃŽ­ðírMœ,6°}Æ$ÄÉû¶+韈SS$¦19Åév¸ºUgêîhÖ-U)q¶T»â/µ3Žï|gn{>Éfqaâå¼ãš~ynß8ÑþüYq¹‘‹É–|âÊë+y œª,®þ^½é›BóĵߞěY!ޏ¾cÚŒ~wJ‹·ÇéP¦”¸U0Ó ÿm»Åí©ïv[e‘z&˜ýd¸û"kýª»‰û÷\S~¹¼_<Ê45KºFkÅ“{¼«6ž žLZÔw«M<»´xeí!óŒçõåË…~i8Y¼^«Î»ÀIâíÔo¼_gß Þo¼Ù¿~¸â“µq–%I¯Š“ú,=÷Dz/ëïZ2iufõ=Ó襌Ü!_¶\2Þîç}·ô]&óî¸7ð¥Lœ´O»“m®È¤ƒ2_^1pŠôI稻n|-™b| iç•©\í½f/ˆiRœl›íxI™.Oùró¦–ú¾ß°vÊR™9s]­iÖö2[¢Ëô-³ç›¾#éx™ãmà¸$*Ê\2&Éx¼§Ì[t_Î3{Èü¿”èéœ,$æÇ½Øä­,’>Õ„÷çºÈ¢i÷zÕü{I;ÊG>qúJmDßw7–ö~¹E`’ï¥s’OÛ“ÓY»4¯Çô±wdÉꑳÚ4=+åÝõÿ_–ª×¯¾ß $²ôßSÖ›,Ë–ao2[–{\*ÇîŸ×É Ó,y›‰e¥Æå¦~W±–¬RxÁ‘iYnÊjÉ_‡¯©&kX³d˶v¡¬9(ÿëEå²ËZ#oܪÓ䂬ýË‹ö{Ú¬—uÖSf×µm2ðTÏ¥ÇÜ—ußç_2âÚ¹ÕŠÊ¥÷®ë£rÖÿ -˃>5ÞIZWÔ8¹¤c]縥ף¹õe¼ŒkŽ{!ãoîb·$2!¹ÕžÿQ&z<äй²Md’îúÜ/*“mž Å;ÐFú4µwD­K2EË s³oI,S†NZ\lΙª_ÓÓ¾œ]¦ÕiÂÂÍidši^—Ûo/"Ó.Ju1¼ù)™nõŽqiv¶“é7y¥HüóZ™a£¥MŠBƒdÆÛ7· ½!3Í»sfÓ¸ 2óLgß•óÊ,³‡•úà?Lf]ß»äºÊ•d¶Ë '+úZúŽúeήÔËdvg2/—ý˜Ìþ¸IéK}–Ë«”Ž×a¦Ìù}dÃT[ÊÈ\|jí¡ÉÜa©L_øRæ™âý]`£G2ï3ŸQiÃdþ~këÏš_J,1Ù2~A,œ÷{ïwUÈ"%òøuœ3Húož° õà\²èí1³WÞ'9«s"eŽ”Z“%ÝOvÚ/m3%ÝI¤ýʦ5ãd‘ÎLwšÐ¬tÒ亾²b{YlZŠmuÛÌ”ÅÅÙ/t·,Y÷Éò=×›IqþÚÛZ¬2 Ô‘fÍR—¥ útÊÛd’,]˜k÷»"ËT™b³ô/,Ë÷Ÿ·š·,÷pÿŽsþýe…oíñ¸É3YÉ>¸ÒñÎÅd•<jLºKV«Õ+¢D¢ßdí¦Z÷¬UüTàƒ½…eíCg¬²]ö~X¦¯õ”¬×æÿÔeФ±í.Œl'¥{7þ§Â²ÉÁ¡s®ç•ÍÇ÷y{¹lbùM·!£¯n¿'[.XÚ©WÏÒrô¢oÎ×ÿl~ëØG™µ¨#Æ÷HçUC¦O•*ûây]Å«§á^÷k¥·Ze³9î‹‹ƒf´lzSœ×ßgü“âd–äa»?½G÷æ¸sIq°¬+ lbG±ïX…=e¼·»ß‡Þ:wwõ‡â÷…¶¾.ð­8švȾÙõç¯Wê}äµ]ã¶ZÄ­úM3íØÞWD† ¬ÙæHKqoçÙ#2ºÄƒÓ?^šÜñwñh椱#E/ñ$ݱ5 ×$Oö;ú§¾÷Få¦4k/ÞÍoþqÎúŠâƒ5ã7­Z5;ÙjÖñézÑD­_n^vkƒ¡ÉšKï¤3ovìî/½?·ÛðxùiMyufáQ«eœÃmW¿q—'™•-ûEãýd‚Ý¿g,,uèÿÆa#“ä M37ãj™4²u7{D™|Ûº•ñ·o”>Gº–›o…L±÷VÒMy›Ë”ûœ%Jûþ"S\ž9¸Æc™úÈôÎu‹”i•\»lX\™ö×-éÓ”ß.Ó-h¸ëfï#2ý ínÿÚAfh<£\·L{dFÛ¶Í[îÖ•™RŸ)–ªŸÌôqWÿ1ýªËÌ/š5Jz»­Ìòfã…mKë¯w¾i[õMZ³œô]YªG¯Á2{ýKG¼—9ReœÕ©ó™ãæ'ÇË»Bæ<Ùçö…ÙŸe®¿ÖªàH&óøä}³æÂm™70íÂÍ;ÖÊ| #?´>OHpõÔ¶sóeÁιDï>sd¡[{ƒîyI¿vÖ³•þqÚ¥óïü«ô¿Þ)lú‡&²è‰ ½“÷;!éìÞNFH~ÕxGxÿÒæ7ôÕˆ“$íƒîå/÷ƒt¼«üm³ÙaÒ5a|’gY¼^ðÑ.ýÊÈ’e¬q’ŸN%eÓá¿]ô­ 8ò>UÝ…²t†Á÷Ðd™oç _Wä€,ûahPÁÃßÈò³ÇhøAV ý¡Ù†uweåÖƒÞu,"«ŽÎÝkapY=’¼Ï%¨&knÛxµò²vž‡¾cÍuÖ¶õ˜˜KÖm5"g[ÏeýòžíZó­lT³ÁüÃ…e£9·“—ºTY6-úãľó›ÊѧãÅ[•¤µlœn뎴֟¤ÿ÷ÙÓlÞqR&rœþœu[*q/l‘wx¡…âRÜ[#V¯'¯ I~-ãwâè…yrÙ)·>”ů\q éº %:ûsnËûò·aboÓ¡Á 7{z{5é·Æ—÷›>¿Z$öÔ8ØrõŽGbÏ™¼ Ú>ÐÄþÚGâ=YäëÞ3«ÛfqäMÆñ²ZűS6íÉ9Aœ¨þ,wÍTûÅÉ[õߊ,»Äé n•M#Θ¶zÁæ‚â\¼ÓSlÕó‹óÞf¬œUœŸT)ÉûâüãW«æ-]'.d.T¾Îƒ/ÏyÅo(cíçâ½^™ã½-.þÞµç›5ëÅ¥ ¶5½{S\~=l`gç;q5ÏŽsuÎXÅÕo?÷éõå÷Ä4yíâF\K®Ö×^‰›®2‡{ä½(nýÔ1cÈææâŽïÍC;†‹È=&¦û$î­^Ó7böñà”ÿºñ/ß‹ÇE‹€„µÅ“Ö²-•Ïšxß-°7@<›yN¦cÃÅ˵#^ÍË™A¼ž•¯C™ËWÄÛ=ÖnéÓE¼¿û¨ùÄÔ;ħêa^¿<¯^=O Ýöp¬ô¾Qh|©Ïc¤5,„‚ßÝ–q³§?Ø~xïÕíþ—VÔ‘ Ü•ò°LœpÏÊ q dÒò-}ßmò“Ég=ï}1õr™"EøÞ鮕2e«Ö7*S Éð|–·LÓ¬kö¶OoËtñœëš´V¦?Vtñ¬ÆdÆßödá8.3ßͳàéš×2[¹ “tvIß'ß¾0pƒÌÑ,28e¾D2ç«§µ"©¬Ì½¹Þú Gó˼+vlm7?‹Ì>"GÓ´¯d¡â΄y[ûJ¿“FÕÏp^úû8±Y×"’êM_šn]'ɯO=$mŠ´~5tLîSaìÉ‘Ò5«CªÙï›Ëâ' 9½~Ú-=}°õê<Ð,îÏûï’¥6Å©õîeJY&wœâÞMªÉ²³³¿.[³£,ïW,ň"Kd…}묿WEVê2éY§ÙßÈ*ùn =9 º¬úð—/“Õ·KôytY³z¦‚û2ÊZiãýT,ƒ¬mïžúÅ:’uŠŒŽÜž5‰ ,0¬LšþieÝ<¯-øœDÖËYe†÷dý¬) ”»@¥õÊ2öí.Ù aã·o¥ ^׺?Ùw²áeÿác¶=—¶Îš>SÙøÇD—¾½½W6iØ$bÒ­Y²iõ‚ãBÏd–?l˜éÁ')þ\uqÏõ²ÀÐýwv –qGìÏgÍQHÜé^3Þ]ûq6ÉkÚ$ŽÍÿd‰c÷“×½{v*±/϶#ùÅît·Ö|Gì\˜m\ýÂOEÄ£#Ý vüÇukã±}¿Ö MÉÖbû¶°‹­|_ó#".x;.­Øyýu³.–õìíóæV½,SÄþyáÓ6´&Oé'Yfqhƒ#²cÎ8âȪK×Ù㉣‡²?^{µ8¶ôÖúd“Åñ!Fn^'NøÍ¹2íïâÄépû©DIÄÉ5&´IžMœªôS‘£ß'§ó‡ÞøTó–8“&õDéTœMÙñRñ5[Å9{º¡7zwçëI–eg¸°zÔèìã‹K‰ß©[þƒ¸<£G,ª#®f|loÀq-圵÷¦×ùiÀââF‘« 8,n†^Êß9ëqëÒ¶õ•¬wZU(;oÜ9q7_ÿQ/\³Åýô³yŸÏ"–XXmCçÍâñó· ýI<·dßœ(ž5®5=ò^'ñüàî59Æz‹—Í_ØšÄí#^§ÚòKå7£Å››ÔËú\¼o—([Eñ±\_¯Ò©r‹Ï»Mp ^"½\9s(•´üö­O!kC§ãÝ+O”ñÊ Ê³¨c™ \س´ee¢^ÍiAê™äFåëÕÉûoúíúš 2Eѵ)¥ò•)V\3zE™zríŽ%× “iç>¾þC™Ÿeúç½ûô;"3 sUÛü±ÌÚqÞ³‡cæHßëO›sQæ˜î#ë¾î.sMªcí^<̳kͳ y·Éü9ù]7IÜ4§÷¨“ó¤ßÏî‡4Ï$ý÷õùô1Û=Ií>-§¦ÔŠ^‰›Þ¯˜´g>èj\º«tú¿MŸ¬“”ÅBÛ—ÊTeœ,qvÞó,½êHÙþr‡Œ]JÉR©¶ÜÈ•d ,=ÄQöW{Y6É—ßÔ&²Üün£¶9(+Ô™ùk­xõe¥Ôƒ;ÒN)+ßá©§®N’U—È<úæYY}_‹>~[FÉšmßÌí½a§¬Õ«Q“œ›žÊÚ£3Ìöi°KÖY2=Y»íeàѽŸ¦Üð“u?&|Lµ„¬_$ùçGÏ¥ ¹1¬åôL²ÁÌ_’œšzþËó·´åÒ-×ec_GÙ•UkÊ&-\3ì lº´ÞÕ‡–Êf¯*_[7ùû{­}dÜ'ӮؒVÆ;°x{ØÕS2A‹Rqî5ž&e(·YÿÍ2ñ£T/ïm “ž‰?·ý«Þ2ùÉ»‘Í_%}Í©¼ãXq™âížU½ÃûÉT ÂO¤îz[¦Îd»PÀúåuÌïºcYŲ2m™êµ_yZ¦«·âB/¯é[>y³v9™!tÒ·-Ó<’ôxçwUf «³¬û£2sס¹fŠ+³|ç9±Èn™ur¾ô•¶—Ùv/ؼ6aZé;pÁšLÉì…G§.üå}Rö›;*Nº s,´~7ßUFæì½hèc§È\MG‡†Û.s7síuç­2ÏÀ³W{g •yOø…?±%“ù‹³Æ9[Œ·í‡1¡¥e¡ëC îÏÛNúÝÎß%[¿³ÒHXÖÇ"eÑ…_YÇÖ“trcÓšÛ‡K-á9Ÿb£§I[™å ö•ûò<†Çk37éuéØì›½Ô»3Òù2éóy>Ëbų­uÿ[YüçôùÎ%S$Oéì\Mе?üÐeû}PºE‹=¶Û²Té~‰zo©,K7ñÊ–iñlYfü†ýû†ʲ÷ÏÿaÞ¯²|Û™S–7K(+¦ð;œnÇw²Ò­‘UÖ[/«Ü+0{Ö°¼²zžÒ Ö]y/kÖ|zèÎä5²ÖªK»ûùô—u\‰êÈð2ð¾¶?ÙŽ­²Þ¡s~{¯•Awû¨þ±žlTñ`±nMÉ&ožæ:ñK6Ù|ðÍ–;§Èoª”ýfß„dËF­_mX÷“9êÀ¼‹Ÿdó–þÉ=³ÈOß½ºá_¦ý\¨¹3ïYñr×ÅKÞ7¿[ò¡ÖöÅ…q>¥Ö¤¨ðåu$AÚÀ޹ʼnÀ«~S]ÍQÿźw‘â@‘i»keœ"öžµ\swŠ=β~­{v¹ûÍò`±¿ê»D¶ˆ¾âpÇ_Û|.“_œØ7xU»¹â|Úl»ÚuY%®NmسHõ…âfŠeÍŽ7Ü)n?ïÙ´¼k½¸ë/¯¶x×CÜÏ_c`ŸV‹—ÇÏšòxÔtç›mƒŠÇ|¯,Ë-ž´_jI‰âÉÁš”ÛTI<ÍVÇ‘åuñ4ü~ïFˆ§oÕ-˜àx6·NûÞÙ"Äó ,§í/²·›ø¾bañÒë}•2ÍNˆ—ÃãvÏ}_¼Nqøaš™‡Ä›J‡œŽ„UÅÛcj¶/﮿_qSñ¡ò•µ+ú‹››‡T¼M|hðÑuó†ôêž¹Ó­Óû¥w‹W»>h$-]3¬ºTå;i]|ˆ“”øò~Ãúâ»+ Èxý†—;²|²LÔÿÔo÷ÂeÂ%—‡_üV&n<³I¹ödÒ4!ew¼J.“]ëùöÞïáÒ§I—ÐÕGì2EíG¹ÇüQ¦¬µ÷tû>A2U½.÷w¶;)S7¼»æJ¡O2M#Ë7!7ºË´µ›ÄI—­¥LW|ƤóœY¦Ï:{S­n2ý»Jûm¬+3œ}7¸ö³Å2ã¶E]цÈLK>º X$3Ï¿è¢x-Æ»™r¶x[õf©šóЉ÷C†pª¤øx§TÝõ>K¯Òë"Kž?+½—^O“zg iu>|9îc?çî“¢Ý2‡Èx;ª6¹Ñr–L°a~ÿª;¯ÈDÇæ~c=5C&MÚ~n?ùL&o“ãqÅmŤÏó©>¿¤;/SV¨V²B@V™jí¨-idš’­í?¿?$ÓÞ(¸`uÆ"2ý ‹#|y˜¤@¦=eæu Rze•Ù’®Z`s‘¾ÇºíÔ£ªÌQbÜô·JËœ‡æmrÃ!sÿ%Iâ1v™··cÆ÷­ ÉüÓŽ¥ìÔMÊ‚/ ßš¼µ¤ô”ÁÕÄúHúÏÙ¹°TSI9ºuÍù¸•ä­+Cü½HÚºÚ‚.V("U§—Êùx„tU-plxX*Y¼Ç¡5é2§–%·«<,ðž (<çæïEËR?dÞÔ¯ä*Yú™ßOã‡øÉ²7ü”Áoƒ,÷ºòšþ—e…ÑÓ­Ê2JV²¸5~ç5Yù^qº˜»—¬ºücó%{Éê¶·¾?°†¬Iãµð](kå™Q½Kv–µ}#Â|õ‘u²Vèá]§° ̺­ÑÆ=IeÝ©ÞΨ;QÖË3óÞå—9dýÿ‡çòŽúÿ¸½÷–•½·îE²ÉÊ^!I%* ¥’U"«©’”¦Q„²¾²÷Þ{dï?ýþzŸóœó¹ç|ιÏ}^xÇNy.ÚIÉQE‘ÜG{IÎSQcþè z\_î8=:òÈŽØGð)õ|á¹)ItZ×;ܺž‚ÎoMÂEÑeàŠÖlz"Fî^ï'VmB«uõ_|ÐT;ècÒ6ÄfÅP›Çywg’ªÉÌKeÒ ’¿jç»}”ýŽã- ‰„_’ŸÕ2ª¡T(<'±¾~xÍÔÙ…æùù|”\ÿ×ñOõ:Ò„râÓö¨úóÿsþ;ñíÚF)ü>¤òšˆóÀçní礠^rì¼Þp4Ò½™Nhj¤ÚPŸƒæâ˜ÖÓ‚Ðr§êÂ]¢hh=ü_§!´–ÕÆ%ðwA›W.Ï‹\Ð.Æ@v&©Úÿñ{GÇÝ…ŽéÒ•ªpeèì5í)0…®íws“üÐÃÒ´öAxzÏ=5è2ª‚¾â»çbša@/@Ûæ†# vYÞ•üÈÃn¶¯HõùaÄŽ(&ÂWFƒˆÊ¤9ÂXÞËÜ/7ý`‚ïÖ»DnS˜|Ívë½L; W¦dÿ‚Y}ª‚Óþ0çü-Ô6ê',|(³T!†¥Ã>}ê¿°Ü»{ÓLVunì7½: kõÊã –“)ÏÃaËŒéù7nØ>)pÿó=ØJÓŸv‚½m9Û@$hL}ÿ‰Ì;öo}ü‰Ä«—(ÏQ"icócçX<ô}8ð´URŒénýJCêc×&æU\¶…dÐç¾ùhÂ;dtñ ίFfþfÇäNd|zÓ{ˆ ÙÏn‰œw[AÎ §òôËÈÓ|R!½ùojJ« áaº¹Ô¤×(¤³Ý”ÞŽ"'S˲IðÈÇG½î'ìPBñÕ<[ Ôf«vãdʆG™£<»åßÇÍ·Q¡[ÛBïÍ;T*¹¬èyÍUjFþ3cT'”£zBv5Ý=t[ÈLýÌ¿£VóöEއ›¨£•zoàãQ/ sYEüB«² l7Ûødû•ÐݽdgÎ|òƒ¦ 9D¥ïçï; §<'ÍÝùAäRÿyGÞ[¯«~™fyì»ÈË!oHçÊ Ù}äËÔç°VBþ«_Ê//ð¦ íþf{è0®Ë'ÇÃ/\‹õH'QÐ×'çô YPÇÂ3C6UÐ1WC‘s9:/F—P4ÏäJÍß~“hûRYG„âÍ£Ô1*R(ù"úä¢i”Ž—üÖb'‰²–ÆûżŸQîl†¡bÇÊǨ _t“<àM²¯².¢â4áÛF;CTæ÷¼´Å–*vvÖP5a”$åbªõJ}ZðµB ¨æ(AÍz©€qR,I_[úᦷ”í jÛÏdÔ³ï¢Îýú?ß}†ñX_‡K\l7‰þL±wõ&®܉MBƒwÏ…µ(Ñ(9Éô©¼šd¶­”‘Ä ÙÕªÌ7y4/¿Ñài÷-E'•iì[Ð*û¾€ç7N<étÝ,-ä5ÚÊI®%k¢½úÅØr_mt =3ü3](z‰íØÐíšISÆZº&RÎñBO:ÁÐÕ)Œ-,Ùøz]«¸1!æ-jع_Í&"F–?Ç_V}Ô„µGªºç3âaì±êºññÞ{u›òë'èLàf!l‡Öû9å?øÒ q‡jÓ29j¹Œä^ü†ÿºûeõÿƒ* <Æ9®Uof];¡z¹ÆöŠÍM¨Ï ÐúÞEzð}û¸á(t΄œë‘†!Þ;ŠÏ`´n7}|“&nŸ?/ø†¦^dÔköæÁÌuQ«OáwÜײs×`.vôžØ+˜ïk­´ w€…µ—S“ôY}ñká“/¾°$ËÖîØhKIÞï÷À2µv€÷cXþJrÄõx¬¸=$üE «‚Ó=f¡w`õß¾î X[ÕsˆxD D'f[îͦÂ=ñ¶\)Øòû‘îUÀË»g%¨ýaGdš)~ÁvüRb©€=æL}LRØÿ.³RâW…ÕwȷĹ‘°Í²"–‰ÉèeUÞ6!‰}¿“¥æ5$­Oš©öÂCÖœ’_!ùŸ£<7/FÊÄ–7“%ãH­ãÁb(±…´¤êÇ: ®sLþן`¤Ø<êå_Ž }¶~ÉŒÈ8¬R,Q„L“Šgyý‘yºšž÷è$²Œî-±oQ"kÓø…¦¹³Èö»7fä:²?^3N²CŽë*'fv Ó+$½öäärP^Œ$En{çÏ^´ÑÈsš?^µyo,FÑEq#_†( ¨ òû–¤;y¿B–€÷tÆ£žQáá‡ðWŒxEaø›…â^+Ä¢p®œÿ« >(² â1âσGÔÞÿà¤A±x‰ä55Ÿ+W,IGI«š‰S]¯PªîÕ'¹~V”qùp¡Uieï :°JE9ÿìËÑߤQÞ‹ÅcRþMtô  bö݃üÚ¨, g¢•‡*í"&–"¨VISÁK~5æ¿ë^TÕC0½Ã€9¨uqÅó®¿jw)n<÷iÁcîÂ)õ=ßð8-Jx£^ˆˆé9u4ø]tÏ z¬dëí3Дårfž±9¾X†$C©ãÓwÐÒ5âɰÑ]´jˆ)Kب“VDkÛDÑf½OŠ],í~®döh£Ã—L! kYtê ý-5]ÕÞ%‰a,éljy6t”n%|ã·„R7­î ž”ÃC¤Q¶aJçRã,Ñ%èå“#]ä·֥Y®Ê[SÐØ²P1YÐu¿vµ|¯A Ï iGû§ž4ø¾gUž×è®^­‡Ê{ß-u%K RÕjn5;*Ã[·³,z¢Ì5/:»¨Nº7ññ³Ô¾v޼©µ’‹õô¬ï¡¹{èâ°lhõðwª ‡¶ù}UŽ8zè0Œ°©½ÂoY5N–CWãó—}6–Ðý)ÉS4 zd%Þ)³7CÏi•=Þù"è¹¼UøN:z¾<ê3ÞC6¼YÜÐÛTxÖÆ,úÒgǶõö¡¿x:7˜†¶r¹lÁPÅ›ÅBþÖc³n0R0÷Æé1ŒÎp®ä†¸Úi*¥ 0Q:SKÅS^ž]ùaÆôö~6k ü¹8 QƒækØ5oÁ’òÑ“ÌÊl°üö£Q»ö ¬Õàöð€µMÆOq/a£_4-\ï4lÍZ4L4ÂŽ È¯Ä‡ÅðïãG“û‚H@óHÿÇo$ |üh*|‰  + ˜‘äkõ£ëY‘HõUc†cÉonµv^ZDʧaLÙ*HÝ~Ytÿ¿¤“¶ôüvºé;¾Ù§l £hWy7_2%}Û]ðG.GãJÒ\d-­ú$žò Ùoý8|#9ÏZª~¶¡@î0‡ y“ãÈÛµâ¼/‚ü_=¸ÏãaÍO|¯ƒPðmi[²8 ;ŒÒ0ÓEQì°—"ê(æ2JÚ™‚åÔ Ž(-Ï>á¾@„²÷Ô–éâPnÁH]VÌnßëL B% }©,TÞ/S/ ¼‹ª3üÇÿC ΋ÒR¯Ýñ¨ÿèû“52ˆ[á^«%oPûäFž@ŸvKÂÙ´¨«f*¿@CˆÇ9ž¼lÉ€úÖ¦QŸÌ›Ñ`Û§»pÌ>Å÷Õ¤¡‰ŸÓÇÒæ|4Ó]­ì ¼‚æl”_"‚”Ñ‚æÈ£ä&hIEUgìZ‰V4ĺÑfh͸|ʼn†OòÈ‘<ú,€6bÚ*¸ôiý°ð3´Ó=õ÷ Ú›/iF[Õ÷Ž!éèè$§*hÕ‚NÎÓ?såÑÙþ¶·ÉEt1Ÿë¾x& ]Ï—,žö;„‘ZWŸs\vC«C I• (šYj•+ÒŒÄjüËC¯'`ü}QŽFtÄßJ$'Ò‚¦ÉÖ$ñPc7$ÎGLUOHp³C…X†Îk)„_Ó»Ÿ$j¸¡´ƒvž-Ä J>“Šâˆ”Ò'4ˆ‡â±Ý]«Zÿ×K¿Tr܉þÊÛ¬¨ú[ÿ?§JŸ1"¥i ªÅ¾,¼3i†ßc¢š>¿ñÀ¯qKü/ ¡þB©ÕR 4’©¾€¦d`6{h¾õÓ䇑´°ÎPõ‹¿ƒ–LæIS•th5aÍh£cƒ6òûý{é‘ÐÖÜ<ÏX/ í%gÍŒ»žCÇk1þ«EƒÐùsè¿ Ë6èjÿݧx7z„ÞÇôÄo Ío¡oþ—Ùœ…' \Ëô­m „!'Öž‹qÎ0üFéö]¹B)WW§Û›†Ñ¿nG—‡Ãø©cOz¼'ab`ªäc×Q˜JÐÞ9&<3Wªjš«àOhyÖæ‹J˜/ëÛ¦ì÷ƒ%ºiY%GX¶O½ÎòeV •M®çÔÀšÊ½ÿúµ`½sVfN+6ß÷aæ8Óµ˜ü(`§¦ø?ÓI°'CÖ°*¦§ç]ƒÞò"á€ÜB‡¥_Ûá/µ¼¤ð:8Âô’>±g9†fbå&ço#UzÆÌ}†ŸH{øÄfK}"ÒûVÓ«½`D†¯R£6  <Œ,¯q?A¶á@çߺ,ÈéAè·6±‰< ŠA È/~—Ô¶ù LE=‘iBÁæÊqê;(<ó`WÙÈadâ Šg¿t[*ÛC©3_ž¨^ðEY©‡T”/óQ®²‡ðÏó/¨šþ"€‹•œ8eÎgY¢Š{í•0ZTKÜèdþ5sü¢>"øIKÿ'#‡Zqª;t|P{çPs…5»Aðµ3÷yäÜþ‹zeFsý/Ñ ‚”>%ŸŒ ½ÑĨÊZïÐ{4³\R0©±As•‡{œ/HÑâ„‘Iå%^´ôõ¤}£1€VXÒóý~£uµw:ÞÍC‚ÚÀ›²þh‹“Í!f?Ñ.,üuöx(ÚW³›‰ûº¢#-»gúÁ;Ù½›Õ»ˆÎïV7/Y¢ËîºK{ ºI*ËíÍ¡[¿¤—öÕRLôP{j¡+§ÉlrOÜxù°Ü”' 9ì4Ärç¸l™¸V¦)&FÃ5^A_a,{ ¯ tÚ=RQ¤€¶£n"2¿ ¹Òb¿(Åuc}ÿó|õa“Twz N_§ÀâåY¨ûWq§ö/4rqXÿÂrw[i“üBЖØ]Ìr'ºÌ´bSJ¡ï[>Û"E% ËiºõÁhÆ_þ<˜0Š#¸à‘ Ó–+™,fá­ƒŽñô*,(¦‰¯À’áš.+,‹ÒEQøÂòoËcGøÀŠy ÙºW(¬t|^¥‡Õ›=êŸÀŒe¼}ÖIž\å ‡õñÑÙKßa£ä¥MBª'l5ºš1°ÀV“eð mø»¶w¹;# vTiàì>Ž!¸Sí{tÆWZÃþSÖó2ÞHAT#mƒ„ÉÍ ×É#Ñ ‘¹0$îÞîW¤rDRM3ª§E^HVuBž£ºÉϧ_9·q)îTßü€Ôt Ê&)HK¢ülý%Òç?Þù÷hÿÜT'BFÞÊúÝudR®ô–LîEfë7³²N嚯‘ŽùA?|¡’‡ÙÊÃÖx#û`e*Åtr¬êp|îC.R »…wEÈÍ) ½XŽÞ饟PŒhÌ¥ÓÙÅSÈ"÷ÙQÒîÞM¯A/”V—ù{Gþ6ÊòŠ~¦=ç‡rrϽ¸üP^»`.ްÜ“5ê â}ëåösö¨ô«ü†‰€*o$|p˲AUé|–bQT ÒHW_–BõIô<Ç\€šQÆ=\Äδ¦çe¡Ö±˜óÉA{¨íosŒ70uJžÚJð ®àʪur:ÏÌûrßV õO™™Û•ð !P\÷{„ÆÇž¸ž˜¦þ7t†o.≦Žûîy£…"Eý[ Z¾XÈx‘.„Ö"4‹wgªñdCE»¢J4Ú>N‘8Y…ö1®|JCR蘫3`Ê ]¸~?P`¢C7Ÿoê1&ñÔÖ/Ê'X†åC‡”åÕñ¾×X–~Ô>º”u»s)è¡ZiÍVÑ2w¹ XI¾‚Uïd/)‹0š]µ [7=YË»îÖ´Ð!Ï)Ðq/Z*oŸ+Ì^ƒ†/Þ¤j O †Ì£‰‘ªz%ΈÌÖ@%¨Û6ôæðèðt9ƒü÷èÉ›¯saP»(Æ~ÅgZœßxÚ_€Î®\… øÈ,³ÎÔMÂH¤ñÔf’ŒÓ &¤u€I*šB–¢˜j¬­—¸ 3w+ú ·`ö×Ã’M˜#çëó =ïÞÛÃ|§MþÛ%cX¸]8¦`L ‹FÙÄ™—a‰¿BêçowXú™7qüÜiX>K­H2æ+"g8Óòzae›à¼¯ˆ7¬ÎPg?";k+ÍÓ±‡`ƒGŸÇÏ™6OþÙ0P¹[¯ ¥i\ƒ`›hI½)ù ìøüäâz[»3 ±G`/ZôP¦Â&H<»~f b:4«Y‘蔕”&gWªý©ùI¹^Äþd+C²”=“5ø‰äâ)9‚ñHQC4Mó©.ºM´žAš#»±9£ôH»2É­zéÏœ<ʺ‚ žS”¥²õÈxv÷^ø«Tdò¿¥ºJ7ŽÌ1?:ŸØ ËåWýœÍ9Èê)¦Sõ¶ÙLk]ü#]éeãëåKÈÁw…ó6éIä¤3mÛ |Š\ä~Ÿçï¹!7Uãá'ÈÃe}çûMFäU¹Ú_q ùÜÛÙîú” ¿é¶Z£ò¯i'®²¢À'Œd–ÂÃÑŸøžn£ ‹<ç­ÞlòhªÍʯGáð|iÖ| )Ö/»RÓ‚GˆÈGo% ˜UKÜ­ÛwPüc4?£³ JRÖ=ž4»ˆR!ÛooÿVCŠdš{CÊ(ëü¯±F å´ÆüªÌûP^²´tj‘ ”ƒîù¡¢¥@”ž*-¶ÏêÒ¡J³’± !ªõÕñûù¢¦šçÞ…‡@ÕZÁ+ºŠZÍ ©PÇöù·õÓ¨Kð¾ž*÷è SFýú+±.jhØ×FZjN&´j6Ÿ"Ðìü¹Fæ3–h~m³¶Öñ:Z4è,žâü‚V²çÖ%$“ÐúU#Óùn´9<ß6çmÜ‘Û5ŽEûkÛ+¯ç„ÐÑÅñ®úc9t¾b,”¾®?¾&žÁû<7o9_D‡"§×Þ½@‰¢~YÆŸëHú‡»nñÉ>L|²ýörÞº?“Œ´ø-Ÿ ä+rAƒ+ã~Þý5¨ 2;WsG ªÿÅCÚ^)Tm¦ÙssÊ3']Ià"éaý{­KPa²^º~Ÿ*RmŽæÈƒJÇWsî´ùð_îtÇ¡Ìa¨ùýèö‚(4è;¸?dÜæ³„µîòG¡e>=wLô´]²ìŠÒS†örš>j^èdhÝÿ\uºäo¦I;xA7ÁwÝu“ÛÐøÐÆè7t§v¿qŒÓºƒŽÉWç¸y`˜õóÛ°óa„ÁøzùÓ)Õ& o»c1§ˆ_ü¼4}of{æ`2·§îr< L?›Ñ1w;¸ ¡Ó¿Ò`žy¤èØ<5,æçu?¡„eý÷û’°2ddO[}֞ؾ«<Aï4…‚ÀVÔO‡à­ØþY¾o£ÿØÇöYËô`?gÞ,ħ µLÙ‹ô#QÍ~È:Ñ'$ y¹¡õÉÌÙ9šŽ"¹±ˆ› Ržr­rpjDê”àOa¦!H»v»÷0­5Ò4½~Ö.ŒäîÞû¼yiÐ÷\%"óßnËtÏõ™kV:²ÛV|" ½†œOô5'„‘Ûlí­å“óÈûÂåhä<ÈŸdy:Ùù ¬O6‹¸~AÁÊuÃã¼(,!–ŽÆE©EOL¢ÿÊÊé”ðâxê„R#ì ÞÝ,(«o£6ðåŠø 5=N¢‚ÞiËX Tü£­t“Å •¿Ž+ G¡jÚ{S{ûk¨^JÿA÷Hep$;|ô âÃ2+.Ô4¾Ý¸€:÷ït-¢.Éhsùc<~÷Üæ]Ôçdß|Έ¥VuƒÄƒht~2òÏ…:4‘+!M÷XE3r1 Ÿ§xb“dæDzš¯~”XôB‹u[‘Ào´h¹Sõã"Z¾¿oµ'ÙÍèâˆ>£”ñãx±D´Õu~)¥†vމS—9ùÐÞ?;´fX¢$)"¾*¡cŠ\Ü<ï/tʼÇÁtS”dÞa:†.õŠžï/£ë.é‹aÍ~ÉI&s ­Äï¼#÷ñG-Õ»C·È‘ˆ®¾³aLÆtüHÊ~LA»eˆý­ßzШÀ³K´ÿáÀwéµ:|PY•6,\åò‘wcfáÅ9•G¬_à‡ùóÅŠ"^ÌìyQ†¹óS3a(ºªs6¾ßóÿúÂPãb&ø¥«mÒ}kòÿs*%SO–~ ‡ª?³o??Ž€ê¨ù7¼ÂæPî~yÛñÔÖuQ1Í‚zÕ‘Óz{yœÃzdô<4î\²rf**íKŠ ù‰+P˜ùB‹\°,ù&h™&Ï}˜« ­onû¿ h…¶x¦­DÐîCVÎRz:Â+&öF@gRYÙÏy*èÈ)Õ)ú =\£—î±AïC*§0á ÐÏIní½Ý ›¡—Úã‚`h¸y‘cÖF¨ü¿.½£ ð[^›ÆÒÉîñ … Ž‚­Ì ‹0Y•.7›²ÓéíRõ0›Ú4ô8·æz6ÒúaQŒ†²·8–ºõ|>³TÊÔÂS±2)X}ÝI÷œì0¬«ÞÒ÷ƒM‚*•ˬïakÙåÙá}:Ø¡êû`•X ÿ<Ò}šw¨aÙÖC³±ƒbï"ÑÞ¯Ç —¤D^è2\@²×?o[Ê#ùwÚ{|ìîHEÜ>/>'€4—S íUžÆè±­/²ðÏ}—dåD6›µ‡B¢ªÈÑ4!z$¹î¨þ¬E¾OŸ }ΣÀýcúòu»(xþßLx€ ]uéü‘„¢Óòa](.žÒþZõ#JvÅ/¤­±¡Ì¯c‡—¨¡œ¿+%z¢‚0éLzÂ-TÜ)¿2®Ê;¹‡‹ÝQM¨äUBHj LkVVâÑznš¨e|5Šäýjóc}ÿo)ic³~€ºõ´BèQï?¡]uÕ5—¥D#¦.‚;G¾¡ Å’ÀSz4cW~àØ¬æ¼žÿÐe …zYer£"ZºÜÕ]œE« _§WWѺ¼p=¡mˆ> ¿¢D[KÕ²ë7Ñ.¾CædT6Ú÷ž\hïzŽÂž/ÛÑéʈôaDçš«abª‡ÐUX7Û˶ÝtKï+Õ£Û6áe¥u}L˜pÕ‹ GOC³RcÔ¬“»­ÿt Ù¶¿ß· øëÉgÿl¬ìÂøx+‹´þ{è͈IäÜ4>Ø×^!¯ hlgþÛRÍÜQW=“ú¡ý¤oFò ¨S¢ †Z.æ‚É_¡öçþÛMüP_¤\'þušéN·É@ëG´6H‚N ƒÖC½ ×?Mr-o g´Ô.pÀHú—ï[Ö0n|öðÊq˜w¹nÂZ³ÄmQ±sÚµ `a­\ý3ÃUX2Œ €¥‘ÕŸ ¶aÙíoÚ÷îƒ,e½s`%4þy ¬*sì%*Ãêê mé!XûùWg¬_ÖSÞ‘ù ÃFbÚåç&!°ùʵ›¯õ€CÄ®ôÁßÞW¶oYÀÎÉO¯‚YÀnµUŽ¥¬ ìFö¯Úì¿€ï _ŒXðŸÿ+Š„³ªáÙHÌ 7äJêWiQ$ݰ1ørãzdûiE®)l4 [JòÔzÊÕÏHÃÕ¿vpô‘ŽûÖ EѧHÿļZ™µ >LÖ,+« ceLVæ@52MEóŸ¼,Œ#•R¯ºUçVþž†-²ù9òû†o#û#‡/N­²È‘àöá\7r6=–þD¢„\óWy[Ò‡A´ß[*yµß“•7#ßõˆÎ·q—‘ÿTÕˆPÍ ¥5'?;Š•Öß—Üñp4»bDª$ ÚÎ3ѧ¡2ùï¢.(,nILXO‡"G]ɦή h`´Ó;Ax¤g•jj±ÅOK°†U¡$ÓýJßÇQj\á¨uÊ´“åTøAÙ¹’ÅÇn{(·¥’û9ÝhO _`AEÙ÷G{C½QéÔ+ TN®½àGΊ*u÷D"£»á³¨+‡P=Ί°Èø+jêøhÆò²!h_“ëxöµtùüvžS¢vtù_û)Ô™ó _û㈺çé£9SjQv¸€6‡õ„%¤fß¡a{`[üøa4^I³p]•D3%‘Äýk4·<‘ÄÝŒoÖ;w>@+¦M6©¹@´~þàG6oÚ •ý«Ñv÷º˜›Ú÷›å|0gGÇéPéï·ÐEJQ×½ñº9ó¼½øfO5ݼÅâ‚ÏøÝǘ ÚÔ'ηй49üMï4ª Ëé SÅ5®álRXQ_Ššß¥…‘ß–S`Ý?Ü{¾¯_‚v­tÑGLòмkÉœ§eõg*™øZC z¹Újå T¬ðÄ6@…¶Òù[-âPá;í_þª4Î,/‡…CLÿ’ôÇ6K_ž nnÕë|3Ä wA  k¥HRÊ$Ãè ›¨ÁkŸaüÚ•ô…˜ˆö§¥ñ†©'áíO®Áôœ@&aÏÌ +ß‹¯€?2ÌgIµHàÏÚv^®Œ(Ì¥ëðdý= ó¹”vϧ§"³¹-,>«Óµ=|–üïœ {ËG&×þÈÂòFØ?â»_ae°$#PVÇ,& ]…`ý = ÙlhpŠ ÀfØcÿů·akpËV}¶u‰”š`ç‡U_ÁQ#øgËËýbS öù”ó"ARåä\ø(–<»ö‰I?³EØ2#‰Ï;¹üú“H:vš†‚Ê•óˆBjb«"÷©PØF}öþÞ<ŠDÛñ‘ŽDÑZïž^Gfc»*pž¶Ž­Yÿ¥£DYƒ‘O¢”¤ujKžJç…f‹¾„²rÝÇ0Ž =›/á\4Ê­zyŸÛlBùõTj Õ_~½!{Þ•J¯Ý\ 9‡*YD¼ºÎý¨¶ÔÙ}5#è Ï­"Ú‹ô~0EmÙT¯·üͨÓM¼"|ó/ê~¬úVwT õÒ¯ Öˆ’ AÉ…µ‹äŒh´0“°˜V‚¦Çø‡ÿy≠Ÿ&ü›h¡¨+´ü“-fÀ€ÉZZZÒ>Ç“WÄÝ“Gÿ ÍJwîY:1´‹P¾pCWï“£Ź^yt¡c»Íñï9ºñVŸÍå˜ Õèp|JW~ÐÅ6<ûõÅ‹äº"½’J:Œ3k>·êÖ„.Õá<‘? …ÈLôÝÔ ¨›SÑs瀱Æ>ËÇcð߀ôUßpO¨œž?ªJ>g~è%©BùóKúÔÖG üÔãøMï«Pþ½'þè TDÕ\Ê«…ªv½JÏl_ø½ÐtÙFêOUþå¤3ƒ¦£ÛÉÐrupììƒz½‘¸¶ç íê<ÿ¼€ŽÐ•Ä¿k¡ó‘{úm—è:9$\¾¹]%û¤®Ê„ÐÕdôDººêªÄ·ýF¡ëÃ]&·Ð-¤(Ûf Ýíÿrgò '¢òáè½¥?6Â}ÛáÜc`@„Ïhúæ .ìÿ‹˜…aŸïºÃ]ª0òd…u©=FœvIËÔaÜ©»”ê}LšËkþš†©ió°÷Œ`–òn9)‰4̹}R]{‹ uѰ´`ùùFI0¬Ü¶øCÖÔžºþÛ„ 6‘ÎÊÌYØ–œ¢?œÛN®üq¬ù°ûå›Õð†ìëeÑ„÷DBJ¢g±”ºH”"¤:H¢èÛúÉØol®…ßGrúA3môH)vô*é+/¤v“!iG¤-úr".éŸÎüyúÕ¦çþ#“%ÙÐŒ°52w ä帅¬¿r½LCvÃO|©šÈI3x²áŠrs޹4žŒ¼§£³c™‘ÿŠvލÅÿZã×¥¸§QТT¢tƒ…vü¨žÕ»¢HGªAK†#M½Ï‰|tÁKÅP*žý¨où”å`íû4‡rë¿ ^˜ Öh|-EÅwú‚Íׯ¡²×!©­HT59C¥Õ(Žê§¾ 9dí¡fN›Ç—"Dùi_ƒÊQkùô Û}Ô±Ï|kDЌǚ¯ù>®IÆã–ן:^ D½¡ñ3©Öh¦ù“XÈ ¤5eWÖÑx•Ë-jSMë\B³]JñÄϵ‹É֯ѼWC„¼ó&Z ½¾ýšÒ-gýyjWâÑjÇþíëgËx’égÄ·ÛÈ¾šŸ®¨E[ ‹Á®>hw©¢ã¬Ð4Ú'Óø Æ¡CŸOöôot앛ȟÉA§݆gŒ]6Wú³¢«~ÈÒñOt;þND*<Ã_à>>} -«ŽpÞ ÏA¡Ö¯¿¨z´`òZq‰àKé¶wÕÒ‚6æ@þ¿q¦P3ô ¡lü÷ÚÛç´æT¼5òç?Ke^Í—ÛA©Un4ÁïûP\ºàÚŠߋɘŸQLÀ÷ãöyÖVEð2ú˜aUôÿõâ„ÏcLónPZãkd¸òÿ9Œ’Â#aRP™%èôeTþ“Ì£j s‡ê؉Ívޏÿ/½uÿ8 µ~%,PDA]ã®YWf(Ô¯{±ç‘AÞÃý h:d%q¯ö 4³|::h®¿'YA-×ä6Dõ¡Õ£”Ô.ÔÚäÆ=rË  Ýð=)…¤*tX½76~àó’k[—ÞCW=톞ƒôhÚø°ý „ÞbõC‚ý‰“Ôjq`0H¹OÀ߆­?H´^'‚‘°1ËKi‹0:^âMÄã'£s¸Ò`b›Ä;‘‘¦ZõFª}aæ×½JM3üYa0ülh ò†ß©–læíu‚âa9=c¢³%Viì¿ÀÚ3öòØh/Øpà%ie‡-¼Áß±f Û¦;÷í½‰`÷‰Ö¡¬ú.Øg#,UÞ¶@‚õ™†ÐW HN}¬ÏIÔÊOI"› ±¥Åm$\µX®<†”žDK5ÿÝAêÊIñ_ýHg›ÇebF‰ l?oBƒ2Ýø¦Z„Ì©lh¾È¹ÉB±{»PÆøCu£ êìÉ4\ó&¡1A×êÖí8h){;’A "~aœìÐ#u_{Ûü<¾]áÅÄc?.­Á˜‰îq ʾƒÜ©5&éÍŸ/gáºÚ€±--ÌŸL™—¤ï‡…G¦*á÷°¸aùÄ=õÙw^_ 5|KS­{Ÿ.ÀòÍ@¦çºÖ°"uòè»×˰2úšIùD%¬fT¿8^kA¿7C4§`ÝçJÔlÔ(lø_P*÷Ȃ͢óÈ [åýù^Q°ÍT¯uñ´<ìÜtœ¡có†Ý¿›¦/Ã`ïÑQã€×IH ÕrÙÙ‰ eõõø¯!‘@Z¡ã+$þ¼lWž´…¤¢Î®† lHö‹jìÚi$¿¶ú™Ue) :¼^—"µÂÓ»^÷‘V.©ÖÏjé‡6^Y=àO%òÿ&w‘ÑÒr]F$™‚Ï<Ïõ{„Ì™'}©Y&ý Tí‘Wn\_+ÙM—*²‹Ÿ#G@:¥šx'r>#&9ã–ƒ\¿þ«˽‰ÜsMwˆ%ó‘Wp,,…ù<…•ŒO'"¿MÓÒ å ŽóØUÎ@"}{9š8<|㊿ð¹&4ÉÒ±™ØB¡#‘“×r¦Q˜}ÄÃ!¹ED”ÏRK]BQÛ s[}<ò…¨ðo˜ŠËœH ›C‰Æ¸6’‡î(§|krJeΕRøP, l–ÎïŒû(—U\wüÀ‡E‚Úh¦Q¡}çîpÂ5T" MXS\Ceb¹›§:QÅÛE庸ªØ&´m £º²W*EXjü‹NÝË$Aàâù¹wï æ¸—²¢ö+™ y&<Æ­`‘øé êæIs¿^F=?ê’ò_h`·KFÌ!‹F>ãâßg¡IÚ{«cVïÑl«iŒ/°ÍKò\ŸCK6ºÞ‡áhuw?ªTé÷A>2Ä´[ ¢MŽåýÁ]±;å,Ü݉R Ã]F‚N<ßo ¬¡ ¤þMŒÞG·‘ôVÿ˜ñT†Æ×3³;è!nÖ—5„÷&ÔâgЩØütØ=TŽy²ÍD’€ŒÙ ¾r>°Lcò!+£†‡‡†UzÄ8¯­€áQ´”²I…fÅÃ?²›S Ž—_€Ìßõ€OÅ»ÛDÍ¡b˜ý³l ”ëÊÞV¨~uðªO \Q<ÐUP–= ªýâ«gSr+²[$þ´=ÕK÷£~ÝK¾~Û×a`½}Xíq>ŒH?,݃щ¾Õ!çaüŠË%s˜è¼9É ½S\Úpac¦ §/<§Ø‚“bÉ]Ù0Ëôòë/_b˜­¬³÷þ$Ey›SxÁ\”õº9˜Ï—,ôe†…Õu«YX’ðxa  K«¤±OæþÁrûSûüI/Xik'5Ô†Õ¥2¢9ͰÎoðœJõl¸~k%½›ÅwMFäà/OótUl'¼!½»",rqENðoê)Å?ajgY5 ÙzD¿²!‘cÒ:ï¸;ÏéȺ5ƒ¤âu•‚‰{HöÁ®ã\½(’+©Í‘AЧÈñ†\¤ò»-ø÷ÒˆJ¶¶¾EÚ=aŠöò÷H é'uáF¤˜^§2F³–’­üF¦D}SdNÏã½½‹,÷¿Î}a·@Ö›BaÄ ‘Íoµ½\<ÙOÏgg³ ‡ÛŠ¢Eá'äôç¿Ä†\~}">FùÈUbHM<££%žø#o¿´äåhä§ðúzAù¿ôøòLþC/¢);<,9¾9IW‚t6š ûÍ(DÎõò” ‹þ|G¬Ÿ"'½¯§„¨£hâïå'x¤;Ïö9+Š‹ðS5¼D‰k»‹št(ÙŸõLàÕM”v=k²+—…²¤_ ÙÃPvÈøD+£\-ÃA"Ê»˜zrŠ–)§{íP)–›%½Už-:z?Õ©Žw(hCÍæ’K>æuˆÇdb]¸Q;~}êëV2³;wq–2Ÿ`)¯õÝ@ý Ÿ)i…ZÑðóÇ›¡hBDÀË¿zÍ®‘Ìž;¡ƒæ±RÝÚófh±“ãp.Z¹PÄÙߤAëVÕÌ_Kh£”Ëj·‡¶å‰™âhoÒ¦QùÀ–ˆVbæÐ© Ðê"º|ˆ'ý‚nîãz¶Ï;ñžéás߀í[ü'‚™7ñˆÜ|tÈÃP$¦1¶î4€ÑwÈ+úXUÁj^ñ/h¦"zú,jƒ¯š«T÷ž0êûüªJês4()¡¢W-hØzÊ/¸&Kó/@Ù[³2¿3CP¨æþWwÊÚúÞ}ƒòÜ忇Bh ríMnYÎ.üf!¿V·ju×e¤R¥Ac,Î9補Œ‹C<мÍêžm¹¼v5… ýI´C½?tlÛîŸHÉâÓ/B±wÞåÊgÐűÔÞ¤›Ÿ<…)¡³÷¨KË»Vè’OäX€®®Ô6÷Eè,å³Ì€÷«7 @ï {Õ`9Oè[úÈ3>Ö99Œ£7a(ÚŒcOŒ†—>ÐÝœ‡Q»µ åË0ö©£Kòj4L˜ý÷”t¹¦” *™+aÆv˜T@¿þ|¾RÒ« ÆGúUÁRbÆ€eP ¬z·:žƒÕÞþ"k™ XÏËð÷É„Íòg—ßšØÃß­ó˜.EØ5Š|ÿI"öú}ëCøÖ‘ ­Ûµ~åÀÇ D¶‡º‘„Æ´ê‘ç,’ul«õâ¡îņËwø‘b™ëHÎØR‹ÝýåHŒ´·†+w\/ ýuÅ'ñw¡æªÕ½Ò@d’“±ðŠôFæo*1Æ‚^Èj‘ô-™ŸÙÉš>¡,!r yº;!×ô‘ô$Ùbä•¢l붦Dþ“²ÛEº/P [ƒS¡—ø»+¾#E¡º|·Ú˜ (’Ê&ñÇë yËÜš=ãâݤ¶ôy(…l3W>k¢LG‘㣵?YÔ=°î:†òÍœ8{ì1*z˜2 ŒÅ£2‡¸¨Ÿc8ªl ×Ñ@uR«T%SÔ47.+ÞGhVG†n Ôújé&:"]ûýb_ðØsZêóïñ8W¬Æ‰ˆ¨÷6»ëéì#4Ðþz䀋 z·óùÐ8/¸øü=_4KêK±ÃÑcë=¤¥hþ«òÂÄ'hQ3ßßàš†–=bŸhEsÐj1¶ôøÈ4ž¤¾ž]Ž6 ºÚ.Ähë<¡U}Ýíî®U¸Ê|AûÝçTý è0ÅX×ËçˆNòš·òRÑÙL$¨˜M]"Tdgѵ¼âºœøIt‹5¾á3›‡áÔgúéÑÒµ>v9OKn¨Ê_ƒý`Cmëê èÍÜ--~Ó—¬­¡¶nÈxÿÔTéKQFÛByø*Ý-ø¥«Ïæô^J’ã_ux@‘ò" ÉêDÆèóŒ{PÑÙŸ€‚tþ#NÉì’…êhб8r^øý:µôá_z¨™3ˆŠ¼–µ‹öíyÇ>CÝdª½fü$ÔLõmQBÃVXø/+4™”E½ …fãÑà/œÐ¢âí6ØÒ-ËI-ª ö¾ã“?ÐÎ×™''~+d‰¿„ÎëÁD©‡¡k,íjfC(ô8•ò½5~ }Ò¦o_Æ}…†çÕG¶\`H£ÌðL W¿éÏ%ª„Q‘ûç³Ír`,ÁÎïë6˜>ª­SÙKëåå0=2«Õ5Ô(%–¯ï¼F¤«¨31,~:òàKQ,s¸ÊßY‡• ói\Ž“°º“C5(é ë¹­r ü°›ò.zæ!üçºèzT vêûßÜùö4ínuB‚°ã%mÖHD‘ÜaÑ…Ä×Nð™!iÌ˨4óf<×ÍQé'„•²kÞ¯&šUNWõÒm¤M®f/ìÚEú/äÅÛ;~ßnã<„̼¸²hʃ¬Ç–‰ûæo û‡Ì̘Ì0är¼{Æxy톘FD GQÚ$6xøíÅn5½(ôÂl7¦^EªÔb„òO¢ÿ£R"”(8‘@èƒÒ÷Ô~oEÙÚsœS„¾ô¯ #µGE :þs~T– ¿)ŒªJ÷–fQý®]LÆf1jn¼m­®}‚3stèêsÔ×z}Å"uÞþL=ÔLŠºêoR.Ÿ¨•$Iú™•ß'Ñ0ráÉFG$‰v¡éë ý AR<ÑMBÁ6ÓŠ 9yJ‘nh m_ÂèHÑ* †øý~Z·¦|çÚÓ¾tó(ÚÚ‹Q˜ví¡]f Á«Ft h‘ûxOª{&G S¦edκP‡>~LŽ®cj.èÐd•)qO)}Šå]²Ãø™¯Vz{ѽÑ*’î˨&佯™x Y0ßâå]6XéS–‰P›†Ñ™ ÿ¥&h@wXtçh뿲Ä_¯-Å+qQãÐðt¼ÁŸ’jNøz­¾„ßžÚ˜8–BuyaáÃÛzð[WÝ~‰7j6;>%ç7BÁ|ŸÃkyhÌ*8è› ×víõ kæª-• ômüL>Ò× ÃL3¿æ“aÔ”ïLù ?Œ÷(sT½r…)§*µ¿A“0óCÞ˜ÿüÉHýL¯óBY_£ƒ1±´sM‹$q†°8ÒTO§ K·¿©ÞR€e¡8RóÜϰܒ1qtÚVbufÛÀª‹þÕXÓÑ®Zaë:ÁIžkã°aCR¹c›Q¦º—L~ÀVËÜðiEFØVñ¨Û» ;Ï”fuÁ? û#D–°·–~?%” ­îu(f"áão$ÁwÊ‘hBgß–  Iì<†§ó‘tæçÞû{x(åSê8óa¤8mÐyº¡ ©ŒÅ2Úbï Þ;¶Ï·ÎÞþÓ™WIHÿK¼¢§  †95r.¦"ÁúÚö'dÖÿ]H¢uÐï2¼tµYëÉßóíäid'ÑÛCȡܜû{¦í€;}{H"W2«¯âCgäþ}Uì¥Ä*ò’%:ËÜdB>ãʤeÍä7îÙïx#üi¯.ûäÞç÷âÅxøRsf•ê%Ô¾èW[lŽB\¤$Ê…—Qø_E®ó ¡£©“]ZEQÕJ]9<ÍèEÅŽb¯ðÛt¡ÄÝøÝ6¿a”RU«|RveˆÜéÓQöbÛÿs(wZE{f2å½~0«I• BYÇä›hTLën…¨Ô¢±WüUHå üŒ¢ªÍé»üAE¨6¬i÷ýj¾#öÓXÁ£kflaþ¨u\?¨fø0jº˜Se•ÛD7í…Õ“†xêúßÈ¿ž û Cêê¼<Þò S'¿…Ž…Ô5…⨸B>ÙÉm„ÅŸÓ¹`h¥ç«ûÏ×|§›|VUZ¿)° %@c×dƒMY?Ô´_—ž§?U¿ãœs?†Bù9e[”éK<œ,e2Á©©áPAvß7ëT}ÑM}#r ê^ ØÞT„–mzGÿËBÐùã§Y@Âoèó¹ìíÒ¬C—nëË?Kƒ+WŽa´š¸è·³*Œ3oJT]ƒ #·¦^¦~˜ô}fþuD ¦ÎÞºèÓò:9 Ÿ‡azÒ:1ÃFf VŸ½)pƒÙ|G›nÁŸ–8•Åïßažd,•óY$,8þŽ÷ çòÿÿmÉÞ{ï‘çI¶²gD”(”‘T*ZF*TV*£RV¶Œdg‡ìÙÙ+"Š¿ßïçû?Žßñ»u¯××uã}=ÎçóqÇûªý¬”SË?èÔN%GÀJóéìë`X­³«Vp„µÁ[ºv¹°Nê¿´P’¶Ï§ÕŸBÛ§ð;î@îÃPØüe$Ùn5Îm{üÛTºA·"àoŽ›5Ã1MØM»ÄO°Ž—¹„¦Ž#aÞΟ\$>PvýÝá[Hâ+ôIIgŠÏÓÎ1 ¹Gl­'¡>ømgs‰ø%ŒÖò°#½€TºßoKó “ò‘»ÔHG3 ªñéÉ ·Þ¯š エɣÈÈ;P¿÷þ=2 ‡$óÙ¥ 3ßfõ»o‰ÈÂú­´—ÿ²Ò$Ú|kàA6 ‰]~+d§¨s (DFîÁãq¾È)hgÂâX\ê+’EfÈíB7ü#Qy⇔¶—·ïܶþp:òE¥Ú}vZE~-÷Ò9_8ÀÈc°ï•?gŒEœPpbD›E)…þhŸ½åƒ"×¥»–PÔ^¬ñÙK.K¬ÑÓwBñ‰ÂLxƒ’29—ó ÔãÑ>^£ôA‹…ÂÁY”)à°£QÓAÙì ó7'PîvÑ'? k”¿¢!;æ4*ä=Íö=s\ í˜nh¤åíçcfÈÚ]¡vtpq-is:Œ§µsUɨëÑéíì-úM|à²>m’}OžúR}ü-Š\ôWÌ%8†„‰gâ«ü`tšìÂçl2è6Œ?¾Rm¢.n,/¼¡‰,¨€‡ê,Ô‡·—¹T›@ÝK÷YòÈ\¨i ?æÓÕÞ6ÌÙo¡*[òh÷¨ RÜáw„ª¥Q‘ÔË7 z”#ÈNNꘉÞðä‡CÃQM?Õ+Л "D¾mòߊþÞiŽGVÇüU·áÛžQQÇ¥ è’|™ª[ºm)Näu°C£Xš¨ë[è%øéh<½gÃ,ä¡×¤,ŽÞ«æÿü_ù•Qôl^PŒœß…^M¡[öçɱô÷ ¡Ï-GNŽpßµß3?,‡‹â…¾Àü¤¸2¦— ß}Î<ÿÉ#GÅ3Tßä¨~ÓQiI!»c¹ÐS÷Æ·;WŒtm`ò¥Ú^û¾ÏNßã%}¸Ë3©Júêõan ”ô:`1æÌÂ,çKXQÚ¡UfƒÕ/×h ?3ï›|Lçö¹6¡"eö†M§ªþüð'&îöƒ¸|Ø™°ùwÅ)v}el‡ Á«‹‰ò‘IH$øüKØA/$î–HܤDÒtƒ‚½e1$O9Æ÷OA)J¶O‰Æ åÊ%б¤Ñ£S=O¾tŽ<·â.}@ú ÂÎD*d¤³©™øy ™žI&_;d„,‡•R¥Ä!ëˆU¦cU=²ç»û.=½†œ…5¯ nŽ#÷ú/!'!ò)ñé©È!ÿÃ)T³A=¡ø ›-zñ·hÛ0EÎGÒ’Šø£˜›pM•nJDí(ôœëD©%Ö˜€XS”¹®»+ƒrÜ)ÖÎôGQ>zýc|·*òZ\6LÃíFüÄþwQ9e2”¶8 ¤mp„~zG—ëxC´.ú÷QEDZ•ÒšPsÈî’ùSÔr{“¡"†Ú“¯†µb¿£®Ç ½hî~Ô?H“jtî$äþ &²3ÀãWY̸³ÐȬI%TuÝäþ8ˆ¦ïŠjÞæ£Yáå­Â94oW1;õ -&ª{gÅ~ Eª¬™Z+Ox¾umÜÓ˜ž«¢mŠpž•jž¤•È2B{V²C$V‹xÊVþ™aP:$úeÐL¢ãÜ¥[±xIïÓ~ƒNÝËÝi¦ËxÏôÌK_‡d4Ûö!tu8„ü›ï¥\æÂ?ÑJ9ýáå#U|0ñ9óöÐĆ¿þ=Ý„ÚùÕŒLÿ~¨vyñŒ¹h *ŒËÃ7‡¡¬çbiúî8”Ô.-§É×ÃÇfrÂuù8øxÏÓ¹Ùq>Ê1þþ×ùß~‰OLÖ[³¿ðé8‹ßÍAÿΩÖ©Uô‡ùèæò«ï Ö_cHE-ꮫí…FwÁ—iAþÒs6Pß8¯>ó)¢«k½…ÆY£g\]–ÐÔ6Qwþm4Þ ¿!üZj âÌyhá+ÕÖ¥¢ºmhCáÝ«Á· ƒ’¡üœ°"tT'½[n€o]ÞZ\¤‰ÐÙ‘uå ktèZ:;=Rnÿ*&~CoøB°VJ(ôí²“jù‘Á@SG´ÁPQl- ·É_S†Q*W×_g`켉Ãï†0þU,«È‰&ÏÔ÷ŒîÚÁ´ ÛÞÉ ˜!Úô¹º1sºÏ>ŠÀ‚v-…чFXJó÷¹« +‘¬A²k'au²xvˆ¬~9èYIwXÀU{RÍ)ø=T&}t¤¶:šŸcð†ÂëšîÞá߱ϮH@“’%º\Š„‘Ü®C †H¬8& N[¤4”+ûÞEö§|,Ž‹)„·Èoè!¥ÛM‡œ=¤þ6È"켄têS,ßm‘¾^7xeÇ_ÎFvÌ$!s™×QåŽddcyRýy… 9*LŸ@îœkÕ/-ø/Àß„6ù h¦µ?ËQG!ÕW‹ó(r|\ÙÊt¿¯¾ˆ–ùø2%E¥WGÂñÐr„îq•D”õ_TýzåÙ¤Ú¤×ëQað#çãZ<\ñÐãT¤! ÏŒAUFæáSôJ¨vík!w:"™«UsxjL¦ÍœÜ±Ác{FÔU¨µšNíj…:/“4H¡ž+‰ÍÅWÝh pNÖû×ê«-4ÒC£óoÂ&çÐ$5åLI;šþ“"÷Gsé­Ÿ­ háAÒf"s- ÓÂz«ÐúÀñ`¶¥^´qŽö‰&­@ÛOõ=º>Çö󦣛Ñí=W\"?Ô㩦‡n™[2è(M$`4} O'ÊÞ§D§›Bõwžèlq;ñâFyt‹Wâºüo ²I…2M=2Ž+c¨‡ÕÀ—f'alþS’­)ôy‡ì&Üe„®xvW¡ã¬Ì˜íø ´š”×íÈBÓ³—ôu Áûøbc'¨ /%¼ dοßëªì{c*gSI1´xÇU¬é…öÑi¬j‚ο•!þñ—¡7õŒ'éG-¼X—üL F¼ï¶‡u¸Ã˜Ñv¸b>9Lä¦ÌÜ='Ó ¬`93%÷¤`ŽÕÄ\Š6æ=ãØ¯8ùÃÂ)‚ ¿Ý X&W¤9yýIènx ËDw?¤_€ÎúÖ?un°R•Y¬s‡Vý\Ît°ý€5¥_ñ.ƒîð‹ƒ#7dw¿‡2†®’‡pÀ?Ù(a_ü>YÒò¿`3Šú¶=Ñl-{Ú…M¶]¿ôú€9ìŒ^!‰H£ƒßÊßÔ>{×j8S–¾"AŸ¾›Zæ$qG§Î_ Gâ®%Z!)£Ÿùý‰$«8û,oÅÜ)×mÞŃg:¸ ʾÌe&ëÒ\,*²cCºã'ª¾n…"ýű»r%!YÂDé~åþï¥&·µï‡ÒfO³b)åú­âldýÌ5cá‹l›»¶iÈ!=˜3Êž†œn®,ÁWÕë-OÇIäžÕT®>Œ¼ g×âÛ‘Oc’Všþ2ò-E<œèFþTÊ3¼z3(p^jî)3 *É&AT Qßã_¯ÜE¡ßBT"gQxsêU¡в«ü6b§¹ÞM¢xm|ª?O Jj?y}Â¥ÆÕ¾èBéReÅj»(+ç8+)÷å8‹)„Y¡<Ã\v’Q$*p¿æ_(ŽCÅcªï^êßÄÃ^~¥‰ÌV¨ôšì¤›™?*o\¾íù7„<x¿ò^þâ:Ö<Šêmç¿«øŸÜ÷á@9"iÔl?Pè¶xµn´ÎµI1¢Žö]ÏÅvʱ‡ñÌ“¯b ’÷ñÁÙÈËá#„hWظ ùí+*غíd’é ]hmß%Xt¡<½•#‚¿ÞZµ.AÏL‘OÕÀKøæ *}4à«^¯Ñø]h ïWëÊ…º‚ÕñèCYP=MúŒÐª E2ž†CqôÏÛ};¡ÐTr¦Bê2 ÉIaÚt:{p¾~’ †ŸßO·þ9·ï­ÿ>wF2ÁìÂr^9©#Ì3w’5ñÁ‚Ss¹Àb?³C¨›,ß>AžóXV »Ü⾱?'HžÂÚ\ªfÛX8¬›™—éúšåÃï–›!ESc°¥jT¦“êˆí½5ª`çjÐ{ŽÏuðïü¨òæÍG°÷²B›ÿÂ$’ŸÍ5ÛÝE"›:îWsôH\úü¨ùOu$cÞ.$ßE²”FJjsg< h+ºy$x¿g’©ÓË!å™Á®ž5H-apAÆÆiY §”/"ÝÚF ªÃHŸøA‹£'>$ú•±!ãG±G zÈT¤û2|™sWÓòÞ!K–þàØÙÈšå¦=,‘lù±Ñ™ÛÈ^mÍü¢ 9z®Öøö"çºc<rs?Wxöfy¬Â4}¯Y"o̘¨ÔhòyùèœD~¾z*ÝÀGÈ?¡©œ„U§„VP°TÅ3‚² …š›x—PxùrÔ «\QKB±3nƒÒ3P<¥x3ì‡ JLw:o‰ç ”ú™ˆêk_ðP<ºx_eìÚµBôBQöVÁ¹âÊ(”ÓspÙbà@ùÃt_2µrPÁ—2WrÄKœÕضBe¿–™9OTU­z¸@wÕÏ3N»Q¶¡Æå±äëg¿ã1³×³mi<¨}RÕã2ïGÔ+úÞØéD̽ñ|^xü‘xÊek4;¤t/¶M½g*ÿÜF³Iá¦ëïÖÐÂU\Ámv-—:™M\@k?RWëÂR´Yb³dדgž›ÌB»ï‚:ºÈð”ÛÁµ«ÊùèHÏ̽÷F Oÿ¸{“y¦—µ¥CÙ0ìfƒÓO;Ú΄Š¸kŽBi»£“dH s}˜Œ1F^õ&OÉõBç7µ ázhUiôÝ&쇆go–æÇÂàËQSñÅ^¨½IÁßÍl ÕåßÞF¤AÕeÍjvPy‰ÖÓ‡'*a›úÏÎk¨¢8ëóFª©´H³~ž…ÚÃNOo²A½s€Wze44å‘”§“èÂW ÒÑiè`‰òùÛßî¥XÈ7è|+³Íí)]uõôºª »†2#À½z¬L%·OAO”Ã˱fè ·W7¦Ó†“«½W-í¡‡)æ ±¬ÄþúŽð¥ô|Ÿ¬«2W€^ÛdíÀœ*èéͦ㕅~ –ÅV®¬Î ƒ!…uf¾ù ø¾Dªz•Ç~h^HxžU£]¦SêõŽ0nÐAq‰Y&f \ã-žÁTGˆ± 9ü\{ôðd[7Ìi“þj°‡…/ÿ=×» Ëomδª[èñ4Z"øÅ}Ö‹à…;¬/§¾K5P€ßÿ®¹Ê'ÞhoJÃ#'bc5gñh¤Ú-Kš ®wÄ\B û‘ðÕó¨Y-UDpõŽè³¸ø¢ö‡ì å4ZÔ•Ó-Ž0ãD½¯›ñ¾Á}hà«.C±ë‹Ç•øÎE#¦Ý˜ËG¡ G-Ñ·ßhú¸÷_ÒÃ4K¼ÿLè "šLøýdŸ»¾åG^=hERsÔãTZ«&¯¬=½Œ6W='œ?ò£m‰uÒqm<ùw Ôh íuÿDH?º§âäç˜ o¡ÃRÍà™S¤?òËâ4EÐ¥æ!‡æ–ΉI?h™%Dð¾~é]= í ÛO8ŸŸÛ÷`Ë3¾¦ð-2Õ~ñ†1tVÿHgOx]ߣïÜÛç[^ùéÀ•(èMº<7öÒúÏqõŸ˜ƒÁcç“TÈLá»S’Ô‚—)Œ¼6w Ñ õŸã0~`k¦„H&âĨÂü®Á”Ã]êÏ¥îðSˆä5å(̪:kòx>ƒy£½‚Ë»`1µÎÒ VHÚ[Fˆ>ê)±|B^¬UŒ$ùÀº«˜ç2e<üï/\êæ€-&ûÃî?ßöóˆþ1øZ®Îu×öï(ý‘VCB¦Ô­³æŸ‘(ß»U’@IünE¡É%$sII¿Ð0…î?Úyg•†K®Š>lEjšEÆ'ó£H›œ'KÚPŒôFuá4ß“‘ó™Tþ}>d–]ô}VP‹¬÷®ô¬È‡#?kL?—2r“1xÙå#¯Üß“j¹È?<Ûb,OÞœg´C‰8}}šrí:~Ö%~¦¨Ç£-.ÀZ‹úÛdrî !xœˆëtäH2Ù‹›oC“ËP~²ÐMg’ÞôÂc4çåáa^åE 'øÓ²h™³ CÄh}P²±ÒÑ m.ì *¢m3«ó“Ú‰žð ¼ùíï¿s ïeÅSSul’ˆŽ&ÁHZðtýÚûç¼ÑéÝ(Sÿ tö±zO>Có¿?‹ëí÷ÕwŸ•5ñð­ä[r©á~ËÔK?X ºÌx»F—(®Ñ ¶A¯cð5{èTt¬á–Ë€ö“zɺ7~CóvYÛ‚-4º]¹$j¾õÿR4 à‹û©Ò‹áKãƒgŽç á°ßÃݲ$hÕ˜§™€6Ú_ñøöó'ëJ_2ô8Ý¡t>$êl†að½TîWä— 5îÉzüÆÃ¨‹Q‚&‹Š/ÊŒÀt­ HósU˜¹&N’E{f sÝús}Òa0¯K=§vÈÝÙƒ–`‘Ähely–„÷Äu2´`ù´3C‚‚¬ØË×4Ý=«Ìa3éR°ººX×£Êk“1Ƭ&ðkáÀ•` lÐË3α‹Ãoci>Óo°ùÞï™Nkü¡ÔOJ&n…íà¨1N‰6ø+ò¶ßbvI†ùŸû° 5ív“!¡Õô‰ As$*ó}Cgñ IÔÅÂ)yµ‘tŒ34¨ÕÉeŠeW‘"Àÿ¤Û7RzÍàBê»ôg_"mÖðû¾ñH·“ÖX@x (Ó®Ñt#ãÍœÌÁÑdêÔš.ÔFÑ2ß.#²úåo§]F¶R)²·¯ÈþG7ø¥¿ rªUôžä@® zÿn änYa°=jƒ¼œÄfjsKÈwx‰'$°ù&i’ú^þBþ„‹®±ç˜QÀá(É·”Xýšn…1¿! …&#ï»,K ðhydÀÕ Ù!œŒÍG1%¹*é#:(þø³„ÏtJìŘúˆ~G©GÇÿú‹¡´ü$£CoÊLu_+Y:²J¢â›(WÉf@zñ ÊWo\\k~ ãñ¹+æûÏ'õ‡[¼˜P ÝOMDåè;WKƒ.áéÿMÑ$<*=Ñlq² ÕÓhÜOºk¢†®×Ýx!UÔœ¾bZv\µ2Ÿ{,˜¢Î£!ù7Ô{õ•‘ ZoÇÞ}Û†'¤¥dåÝÍÑ8û³íân4M™š­X÷E³ñzR;u´ðøÒlÚé„– l^ô•híû[kU mIĽé'(ðd!·ÚóQý¾ÊØ€§>{øyAGß vþÒŸè$RiFHZ€ÎW¤y=:ñŒNX®ï…> {]æ2TpOæÓ´\¿´?‡Úo^%çEšŸeR‡|¤aþ@ ÷˜1|·¢ŽÕT…îM¶€=èh€øÃËÐR,î£C, šŠ†eÆ¡6ÉÊÍ/wàÿ|·{‹ý;T$h>·Ú…Ïõ÷ڜ酡ò•¡Q_ÔÈg™°Ëîç×ÙŸ¥§öséª^ìÌm„ÎÇ4UŽÐk‘{-]Ó2¬Ó´fáû)¾woU `¤—‚ÔŽF£|ØÖ®ÁXZÒZ¾Q3ŒﲌFÆÂÄ›=­Ä\˜ô Ïš‘ˆ‡)q7òWù“0MвgÌhÓËC±nÆi0sÀL2ÌŒfEª.Ý-‡¹så5÷ÐÁü7ŽÀüC¥°èÑŸ×q1– 3Ë|ie`åÃæ`P¶/¬¶wEç”úÂ/"¦;×v_ú ÛÞ®5Øx“9öòÊOؤI!qt膭§ÛÞÎÄk°q[‚Gt᯸¹½õ–ìZhÃ5µJ$p8þxþÑq$|÷µ:&1‰ ö~Ï=9…$n~m…g´ïyö›$7µÚºíu  5ìñqýÁƒ"“¥H¥aÿ=¥iÄÝ9ÙÆ‘Ž£;Î]”éÅ·‚ß6ð!ÃѳÞ§ÂQ·ùšå"2Üñxª­‚ÌRâ74³‘Åà—Ã*²š¼¥õSØC6ûóuœõÈîõÄèG²8r„Ù&_ ™BÎ ›Uòäêã —^< ‡/|*C^ËS–t#Èg¬®ðײùvE?t1¤"…—mU <§òŽV¾‚OcHÇ ÐK£ï®üë(\þ<Æ*ØEægŠO ¢˜õE­u?}oÑ5è J¼êfº™ˆ’‹åè]f𓸔È!g”!º/µkÞˆ²'~žY·”£!¥¹É1‚rëc¶=«»¨ ÿˆ±‡ä*þ!ë}·2ˆÊÎ¥5âìz¨j~ë§þ57T/¨L>wÁ5>ÞyÛY¡Çòþ¸š@í”Yå CÔS#m$QÚçѰ|˜ðÄM‹¨|ÍZ49ªUS8sÍ(íÍI¼Ñ<2¥†ù^Zò_yi‚VÅ 7Û˜ãÐF{¾&%¶m«ÝÞ&|RF;ùûrö¡Òhÿ¦ÅÐ{}"Ö< cÑñsí=ÙitRº~f ÅÖkÙÿÁÐèçC÷#EÑ־ɤd&C³õ4ÿ;Çâ\ö{©Wò%òäÿó^f,À®š ¥‡Ïk@=_£DéÇP;ßñÜ® jìßÛ%'ü…ªlîL²}¯ªXˆkŠæ‡ÏAÎGt‰ÿÏwõóþ>€ ½¯×QXB•ŠÇýW‘PcUq§¿è| -ð`K-€Æ/ i‚çÐzÙžÿª´kÆi‘N¹Ã7òõõÌ‘Mèä*¹#Mrºäîöœî;ÝÂKY_/µBw†ÚŸÔNèîM­Îó>Ýu;á¹ Ð}Ñ6yã>tËURùYC·ÃЍw%tÿU÷â͇ƒ'6­ŽÆÐKA¦ý@ú®X=eï€~SÃØAÚ³0°‘Ðð~`(y´¹ ~ †W5*úˆ`TøÄpj#ŒÅˆ˜X‡ eñ¯Î÷¯Â÷bX€Ï*üÄñ6g˜ü¦ôy Äûh .›ÃҞ`•"6ÌØii¿g¾k²þvÖ kêÜü ¿&Ø]šra«­}4˜-v˜¶ßS4oÁ¿Wy9iH ÛŸÒÏô 3ÏѾBb·®õ\Nc$Ueö¦L»¸ö¾$y&TµB½sê¬Çíh°í+ãÞƒÆðŸM ÄËÐôx†.ù&5´¸´MOÄBëë¢n›äPh+ó!r¨ ‡ãÄc×FFá[ÀüÙvó»Ðù¨šTùb"t ÜMö2ˆ‡¶Ž º¶}nïû|$ƒþÁ¶‡øØÑðd/¸Ã°dçï™ý^ý£/ãÎÁó¿aL»I¨ìíoy %3“ç¹ø +`Z‡¾³„fÌoÌ, ÀœšC‚‰),¼aU¡? Ktr,"Z°òÉø†Ý©Ç°Æwåg¯Ê;ªÆz`ãêæÑ”¦S°i}ƒÍ»|þ¸çßa"†&åv7`׸€ùõ-2$H~0W*‹D²þl+‰H<³Èa\¤í!>j—ë‘|²lOGö <È•ÄæƒT'èv¯t!M½gÌpoÒõ‹ú¼€ jMѼ¯‘é½Gúrì4²¬œœ¸Bƒì÷F>ðÈž@®SYÜ YÚÈ:d¡“WŠüú·PЂá§ó& ŸTˆüx/EŸD²²‡“£Ùbb^Œ+J•V9º*)£LÒ¶[]Kʹ¬º>KDúW«K¨¸¾/½B¥ÏcùæñˆHÔízm)E†Š™ÓšåÝxx(û†ùµDTVˆô"#ÒC•Þ"&]Ǩ:Gt¨n¦Õïl8Š¡†qÈž®.jþ[¤JûŽZC®÷)²"P§û­×Veê­—w&xT¢¡¾M—-ž¨{j×2‚&uv= hF\ö¤-_Í]õžÖh Å°YÓ×'–hu†ß`Ïž­Ùê.¢íï·‰+xr}ŒïÐÈ´÷þ:w?O­½ûÒ粎÷ÛÌü~ »è§)‡t&T "×ôBç?Ú*yiâ:6e-õ€m³5嬸éPVXéŸù¤ª\H׿² 3œÏ¼‰ƒÁѰ£?C¡›4NÊ›ë ´3$‡åÝR‡f‘æZ5Á6ø²e›¼°5á% VŠdPÙ7e6|½ Ê“¼ß;6 Ÿ‚‡¿VùÑ@ùôÌŒåØ~¾]v駺`¢…¹ã¡YeˆlÞè(t\‘\ºg“]&m×Bo=&œe‹€¥¥@«­: ¼n^™> Ã/û-7éWàGNеžû0ú=çϦk ŒU«Jv)ÙÃx„tÆÆ?˜0\9Æî| &…Uä^WŸ€)†ÒKiÔaZ0?Õòå{ø©œÝ+yÉfÜKuÄ7>Ãl3oÒ÷uc˜·9g¿jT‹’Šf'.À²Ì…•Üw°RDøóÉ‘~XûõÅÞ,~Iò‹èÈÃú!µ{¤‹°1¾n&¼›Î«Áf Bð‡<×@²°¶gë?^.‡Í*í´ °çêÚwPA 6úÎ=¶8‚Dú®ùwÞDâw]¼µº‘”F"Ádý’ݲh¢HµAò½pb±j¤Hè Ð AJ다‘Zí‰çk¤5ÿ{o銴¥Kž"}“h‡ØÇ-dèótž|†ŒÃõ¡ÿBilÀÙèÛGdþùjŒàe3²¬Rª[!‰ê™Z²s·§¿ FÕ¾&Ž„uätzØ•'\Qé#;9‘»mÜÚuyún¯Oo!ß! ¾m-È7ôÀäô»Jäþê‰ïŽ/ ø$Þi1¤CÁ³ Ÿo}ÔG¡‹ñßÝ?¢ðƒôµ™*)ÔÏN¿õEg H¿©£8·É„6Ÿ JXö4²£d"Ë=g3J3gÄ SxøûBF"tvß>Fêc!YžÖ<{Ah³÷Ò=v9 ùùèŽiDÂöÓ”è×t?Z2ìík‡¨y_@£}TxU[:Ô}Xp }Ø 5/ÍìíyCõQgù^C%tù‹j‡òµ[?âsç'w¬ /zÿw-¿ÙM«žp*šÄ„Þ¾÷ƒj.ŸÃÅŠÿ×050›ÿwZž%õ)&qC›5õk†W"ÐñH?š=޾Åh€—²°?ÒŸ¢ƒ®ßrø|Ç kïåƒ3ƒý<¥õþ²»óߺ뽶~slt/5—܃.ÏÓb-²vÐ5.þ‘{QºUïzpP A÷üál¿4 §·›bÙÍú8z.ô^è/Ö ^ï†A‹r»Ƨð=…u‚édŒ,=Éâ| £?›æä×ÀøýdE‰"{˜ô°lZÍ‚éÇGèŸZ-ÁÌPBþñ»0ïGÏó'I–ô–5 GÈ`%3µn=ÖNÏ ~ [†u‰·CS¹¥ð[Å4õϯk°åáIww@ ¶?´ìtpÁ?ŸÆ€ð#°·Á¥g•c„„·i9‘X6ódÉ/$%ºaLOË€dÛG8´Î Ý“e‡µSH©ó¾ê‰?R?OLÈ&LF:’‹¶Ñ’H<åB›Z2|øÆzêà'd:Dþcï82WîÔú‹P «ËÏÌÏ—™]\-1[“#*™ÉÙ¹O¸ÜðšBÞ¼ï†TÂÃÈÏ?‘mÜöî¥{ Qµ=}ƒÂu¿KÌ]_£èÛ§«Ž¡(þé{“Ô9I”ÜayþÈ2¥Ï.L ŸËEÙ;™Á•ü(7qŸãmÈ?T°piq ZEÅ©Ôå ·Pé5_U=ó{Ty6ð°Å8UÛ©,l2¢ºŽTa¤ù#ÄåÊy^RÔ´¿ã.ôU5»Ü«â'Cml¯åm¾‚:åÃqw‹“PϪ)°¼¾õ×gî±ëÖ£aé–ÈnŸ žxqÁä¸àU4ÎzM3ß|MífŸ<¹‹f—9  Ð<šr”ã¯"ZT9òuþ«EËí¿fv‹h­yüÕ›ú ´‰ZÞY^EÛé’Êv½hwìRod¯Ú¿5½û'˜âÞÚ£cäé£õN¡e”HÓGntê[éYf¥BçÌïƒN÷’ñ—ù÷"j34mv´¯3™ENGÉW]ÒÙð[&í|}ó ô”:LÅý…¦òªŒþPûÏy~KD* ƒXÄ_\rr‰G”æPÒunoqŠâIID;œ¡P™e5ºÜòÂÎíx½þïš77uoÚÿ[Tûþ™€Ô&úù…ÿó‰³ÿ"oPÓ<ÿ·ÿsßj‚KßÞÜûãçëîšo‡¡¦O¬Íú®Ô†æ‘.Œž‚:Õø‰l·ýžœØöŒ{ê ~økºCÃaöÂMkh¢|µ_¡YÞ‡õÙûthÕ>¢áwÚ8¥Í܃…Mžàƒ×à›ÚbدüËÐyûk6)wÊ÷¼X ”„¾ËŠWõ˜´a øÏ…û0TdMyJFìeÏ>'ƒÑÛÏ>ë¾€±½¹ë}Na¢ÌR"8—¦\y‰m•çá§{lTåÀ%˜uºN^•ÌóÉN$.ýë°8z»Ù[VDž×q{Áê]j9¾òÃð‹mü=¬×3×XV$Áïø„ÿ§Ì°•dÂF<ÕÛjNò!ðÏÍ}ýxKðsúò¾AÂQ|dc4Hìp*~ؘIU=ŽÐò¾GríY¢ÉƒžHaïÁ’Ê‚”iÜ÷¾ õ‚èNÑÔ¤;UõÔ0Iئ]¯|D&îÙ±ãÈâ ¾kYüÙ¦Së\´‘³@§ùkò|3ÿÐ?Ljü,E¯g P`rWycPèÇ÷£?¹-P”êQx AŠû·nprѢπûå"b”þCvÍÓ…å(-òËl)Q¾4“êè0*>Ua©HG¥ÈÓ§o Å¢J®Qç ‡÷¨º§cË«} Õ¯ ¿zXkˆŒce·¾¢fH…õLýj±j™™=Gíjž;ç ãP7¢¶õ!eê_än.YBÃÀÔ8= }Tã}ëÿæ0k¯jeú&>ž¼›ýrOçýš)c¾†r½};"÷â‘öÙé’“§SaîÛ«‹Z‘A0µ±Ú{͈ü—“õ´¤æý×Z‚2»KP{¢¿µõ³)Ô]fô<_á õ•϶k…ª¡ù¢QïÆf´ßô¼Òœ¹‚™,ů¡—¾Üš¼aÀ0êè«\ÉÿrhTiʲ¤àŒÓ¤KNõÀD‘ߘ:òýBuLkùðÿ+I‡éõà/Q¿ÁÏ0N¡t¢\˜‘¸zàõÊ{˜%Ÿ¼jþÃæˆÆõfe`žÓÏ®$¦Ì’H¯elÁbªcÈcGXÖœp®‡•\Šå#Ÿþ{þ×^°L8OÅÀ¯f/'5¯g°ÁLR™ð ~{Šßþm[ ›í в†ðÇöOL-ìp3 „R¤¼dú±°÷˜QÁÌ­ ¹nw`¥EEýaÙE‘‡^œ÷k´%n;€ÉÓk)ÿŠ;î÷Ðõ¯ä†‘2£iÊ1äRœó²º’t¬ÊVEÏÌÞ¶ÍnGà2äÅ&åÝÞ÷CKv+ŸùÞÙâ»sge⯱a!² µ#ŸjDö9fä³¢½ÆÕ‡\ÊÌuÅ&‘û‰QÃíÍä™;Fö\ ùøˆÔ•O*!_;MC¥Ò ä–óºŸä…úšÉÚî»(Èä'‘@—‚“ÿnHï÷F¡êl–ÛÅ(œïaQ»»‰"ŸÕãùŒQtÒŒÃÌ‘ ÅÈ/øßA‰+ÏßÒŒô£ä¨n—“Ä7õ‰¦åò“+‘DÛ~lU' I3Ìž?Ò@rÁO§à=__X,àAÏ -e¥H¤2I»WiÓŠ4§Ê=mK›‘N#ò`I)Ò[/Ü£¿€ çiLE?AF_Jæ[») ä9Ÿs2ë»Ð9‰,±Üo_ëf ëja‡dk%©¾œ6Šì‹Ö«õ9$ÈÉRùÑν ¹ åsŠ,‘ûa¤Óرÿò޲‡ú¨~%ò•Ì­,¨ ¿_Kq¯’= Ê>#:•Œ‚Gt§6†¢ÚX™\Caë2Ó›JU(r3íĽUlˆs€¿ê JPö]ŽàAIcï[yêÒ(U´ë4€Òf‹{KN LobÙ°ʾ°t¾/uåüîI¿Bù‚GC¢¦·P1Ö˜ˆ@ò)*‰æôÿ‡ªw¶ õ]޳2²@Í£Õ‚/3 –ï;Ï-OÔôž¬ª¼Œú÷ß×»ãq§Ânìƒh|J×±käšæ–\|øe Í54U#ýŒÑbúy‘ú´Š§ÑaúŒ6zÛœõ×ý/‡çÉt^½w@ûΓ+–µrè ÀiyD%ŸÇ®Y{;ü?9 I®%|ÑEŒ6áoÏl–#ß óÉœ)ðg#·öÁœ+ôÅ>7•z“í 7ß³qFCC£wœSé ÔÉÛsŸ¹Ö5R!å´£PõÍq׳8*"¬R?K5C¹Q=c\Ð(#(7.ƒ²Ïå&bf÷=1¤&S·eŸÓ¾îxÉGPÍÀšøðþ¨³J»Ÿx‡R¹V·÷ó°Å1µ{¯ú)´Éb8—$·¢öçÑ_øö @ÚZß:cÜQÑ\ ºB~tO–’ÿ¯2iþ±9 ]SœÃn™@× öCÇÄ=¡‹e‡ÕÒyŽnŒÜ€®ÈÚ²bèš'ìæl°†îÄÔ6ƒ‡Ðj¯Ì?½5ìŸÒÿB¿§”÷l6 Ìf;‹råÀwa›ûïÍ´aÄsÚ%ZwFS(ߜЃqåMÉ®k“l•Qïóò`Z±¨õ—ŸÌÜm"RƒyÚÌR¡šO°8#&¡ +~ª=ür°&TÇæ¼¿Õ=r+(`ãÏYÆœ3‹°%sþþ$ l¤\´Ø™‚ Í£a¯£ðdºØ"ºïÈ Å!1[_Apú<’Œ·¶~<È‹dÔi‡¹ñÀ0Á+Ê)G¤¤úÒ÷´ž©]úUÏÊ!mïìÕî¸(¤WZÁEŠFdxiðØ•ýÒ~Ï óˆ:™óÅ»4Úx‘Õ¶&]î8²óÄxøìGN:­,úr!䆽GInÇ7)ÙÇyÃù™×z?ýŒC?úÔ›öÍ(ø‡¸sYI…3-ºÛ'¿ èã»t\(›rñWš.J¶ÿ}‘ê[€Ò iµNhвnRÎ Ý(×LkµºõÔÛØÔ íP±ÿðg=ýTJòп¯Ü‹*ñ_ûÔDQµãZ€—¬"ª«U\bñD“zΓù5-5¶mDñXmµS›Ç¾Gæ¢NÃ9eÎÔpÔsϪ2ã5AÆøFšR4üññÐS¿ xâ‹Ð/þ?hÜùù›¶'hzA'fQÇÍ;»#v„ÐüÕe­Ì²1´hïN}ņ–{®ëdh­U¹ªèk‰6O¼{Ÿý‚¶S‹Ù%çÐNS´4Æ2íSøE•£“ä ×Ì+èÑF|="uŸÃ;¼WU+ÿ~8D⑊w8ù‡i¡iÀQùÛ­…ÈáweÑŒël„«ŒèAÇi²9h"Ó;É^˵ޒGD¶H ’¶®†–¼>¹O#Å %JIs%× HÀÙYƒ³ Šrš¹\Û÷Áé‘™JÈër ® „¼ wø›kBÞÁʇ¡@ƒZ8R6Šn=lqnþˆôœ‹ü¾P~Î('V“*dåŸDKýwߪذGºb¡šêpl"#TϵáLµï‰7ù†Ëw – •f“nCmîTÚÙ*¨‹u=‘v£¾äŽ~ð‰úX×ë‰Fû}:öÝj4…ÖmóCKèèφ|µ êyŸmíÑDüªÊÛÐÑAüB>”:©ÿ}PÉ#€.O—x›ìHèá6:åzç?i¾™{ ý[ìíõ_aˆ—¹ŸŽÙ†KyÒØnŸ…QI¾cÄ" 0–º71e&öŸñÞLØËä0„émÿvÕ[03?3šóB§ºLÏ_„E'bç¤oÿ`¹EZÌ»„ V¡Eë_Z.¬µæì šÜõ‹ÆWŠïîû|`‰Ã:l &ñªUm]†…‹L<ð·1ïUG×_Ø{¶j]®…„:~š¿‘h/I”ôß0’ô•”—ŸçG²!G¯È—vx`á²Áݤd'×OKDê˳šöCžHÇEr8t¼éÛ#oTèf#c3)Y|ª2ïWÝÙÕC¶ë%·¥¿!§’ÚIvVäѪï;x ùšX»#äëQàéúÍÓrïP(´Þ—Ž›E>\ê“~3ˆâ WÒ¨âÓQ²T7D“P ¥ã¦Ï6ä ìèÛó^úk(w(õ-)*ú±Œ};Á‡J>ÇŠ›é:På¡í¿gQua¥ó…9ªß j¾<¨!à¯yû·j¾¾“ÙÁ‡Z¯9”´ûP{ëßû>2bÔíêùÍîÇ‹úæ1éÚаG¨ÕÇü*Qú.}þ0ƒ&V¡: ~·ÐtX5NX)ÍY?оéŠi¾@Ë‚¯¥Ðšù^ÝÖ°"ÚÜz .½6ÿ¿wd÷xöyJkôuÑK·Þþ%:>É Oþþs˜õãRˆ£ FPž2N§¨@ÇH™ÑXƒe”ýY“¢áÇõÿ‘uÞáX†ÿ·³÷Þd%›DñùØ"{e‹¤¡!I…Œ …J IJI!"+{KöÈÈÞÙ{þ||Ÿç8ž¿®ãþã>Ïã¸ë}¾Þ¯û>ÏëFšÎfõÑL9˜t±/'Ö—€ÞôT¿U:h½¡¹ÂúÑ~/éÝ…ýûvtìŸÃ*T¿Usʺý *]Oû[—9C¹ãý=Æ×Üû¹¦ < e·©Ó9r¡<¨{O`<*-œìX?Cí[3¹; qîà§‘FhùÖ[þrA :¨ºŒÜ3i û<Ù’Gˆ2ôöï0Î8SÀc–¼qÏ6 1¶tœ€q¯ͶßõX…±²³¯°iÆ]óækÏÂøß#/¦~„ûe©|e˜”Ëa´Ž$„©ƒNÜf`ZõÒL"¹1Ì\uY§çü+[+ØL„9óóÅ¡0i CGc á©RßùÓa1é¥Å?,5·Ïh0À ïÌ¥ù+°z£.‡ü¬õ:¯ôò ·Â©ê“V°%¦`!ÄD;Ò·RëßÁÞ½‹7/0ïóŒŽWõxÜ"Ýæ;Jë÷IˆjÍbŽ é‡+•™cxàÒ´ÃÛú ¤0cNöØYG*+¥öhç|¤¹}›m]]éJã^àÒANª€7R›ûýó´b5N!Så—ÈwÚóÈ"´nûFß Yýõ Å~´"[óæ•ð·ãànŠ$=rÞšØz·‚\ÍÃïvî!,¡Ë*ä}2ßîgš€|3qäïÛ‚‘ÿ£ˆS½† œÞ 9OyÅLI>>‰‚Ûós¹ãx°G†ó¾Ý Õµ—Gו¢ðïùˆÍuA™¡ô.×E1®[;¢ÈñuµµÏŠZ»Q¹ŒÌ%Ç{•òPò9׬ž©J³ˆºý¨þˆ2†OÉE…P–K–èkô!”ÝûõÅRJåÉÍÏ8ž'F…ƒ×GEÄ­ñˆ¼²ÇmTü:Õœªv•œ(I¯^=†ÇΪ|“ÖDU¾Ð›kçQíwbÅToÔòh¾*…šŸCåžý9‚ÚÉ]z„%$¨[¹±þÉLõ9ô)ú, "Ë«|Ïõ¢1õ ¡ñÂs4©}KÎ6?±ŸGo­ð\4¿ÁRIÔ}-zS’ Π•væ¶Lÿ8žúÐwö4Zo0úw•öîsQòýZ ]ÈböM%´oõkCkt”=e;O€NI³cJExúÙ2AËH†*ï¤ èø¢ÕŒØ£•sÏñðGjÝ(I)¤”ÿîä£]Õï½^@gá"ëYß*hQ0§ù÷3 êÛn±ó:KCõ£·LñóPÞhÍÒAzJŠ #õ@áçS±F¡[hY?4Î3E1@«tÓ•Mr(ä-ºAéB¥ÖyN ¼P9]¡27ìõ®oéd¸G¡éÊćõ# eõó#‹]h§Êçeè$¸>ϲ1]ü$ûàOù_rŸ+ÐûîD_I@¿^.ñŽ;üm®ùÔ?âoÓù˜Ù¯o_!‚¡Sôñ]»;ûG"Åþ'²&ìHRåMIÞ€$ý×õ35þ"™wÇRòu<ð+‚'Pe)<å‹^ö"•º1“©žÒ=!òœÍé´x.6eº"ý—ˆïnv„Èð=åeVk,2æ=Ëá€LE m£¹²HÇ)’ YÚj+‰ÎAÖ™'œ+³ÈNKÝ÷=’9ä• ‹6‘ÓÉÖíÑbäzq†e@”¹»næÜŠPB^ï§®C>?£ëò€üÞ?ßèw¡Þ_8)µÏ?Þºó,;‘xžù•Ÿä ÑÇIß ÔFa¾Æt†å,A×ìçPôb™»’èo‹š õ¶!ÅC5“ÿèW‰ð0õ¡D«³”(áQçõäUJ‘i…©±¢ôCHëfE™›úÛ„$¶(kËõiÏÚåÞ„ºßoG…ÛïÓG–ð(åÍí'xLNléPU;(+{[RQm:żèj’kEDt” ¶9­Ó)AÔ­¡z,ÏŃú×jÔ ¶¸ÑPšFá ™/­Ö-2_EÓsñlŒ vh6iþÓi¸-¼^­Àà´Üe 𹌧nþ° @´îðuüÛümùζPíœé9<ŒZÑþC‘àÖ©9t$ï¼xç>:]7vpóÇÓ>:×gbÈKŽ/ìBðKšÓöÞ/ägUˆ^\ý'è ä¹÷ûVkÂ…õ}OkРa{ÕGL’ Þ…òœÚ6Õ³¿ ´°ð*_™”œ»uá’ÑÛÿï}k* à­n”±äËDÌÝ”!¼ÉµÆG.Bþeoö%&(\ ½žž%_ëÍ |Ç5ñ»· T»:mjxž‚úCŸ[êFÈ¡‘îüÝïø÷sÙ%>RÓÍw²,./ûAKÄ‹‘ÌÚóÐzïßæþOð¿~Hå*´1_ ­u¯eŸ²„V£Y†¼B hùKq«ŸJxÿJY©7 ­ª:"øU¡õu*OP ´™±~ÛíG‰ø÷†AÇEÞÚ„|èœ]îÿÖÿºÎ9 ×Cϧh‘c'Ö¡oú8õ¦ Øt{Žþ‚Ái£ƒ~Çùa¸Jü”Cîí•{`ñÌ &x³}uŒWa곎¹ñØ øwý8;\y•ÆÈïOB`a2ž†ua)½}Ë·ÞVòo¹¶œµ%× mÍ×°©ëÝߺ– ÛýmœCs°—fà¾ø ­ËÌ ’7‘˜„Æìƒ’”óž´#•@²÷C ©"Hþéª~^×i¤lž}ÁJ…4‡³TU”"]ò:q×!ƒ ²Ru”"cX±‡?2ÓÒœcoâC–äSÃTÞŠÈfðæÃûƒ$ÈÁ4«¤Ù>‚\¤«?èX÷¹ùš½Üñ-ò=rÉ­˜EJ½C²(è:JûÏ Ž·É@á§ †.(zùó£‘î_xèšcþ´3Nzó3²¥ˆ]†Ùöù§H§ê‚²±Oñ¢Ê‹í¯Ì¨P÷ýRs$*Æ}oÒ°%A¥7e‚_nâ±–ò‚… –¨*VcðDZå•.¯é?T?éËö{5Š´£ }›öýP2"~µ›g:²Ÿ{£n@$áèÍ:Ô“Ó£!9€'I½D® Áìßòçü¾h´ÞUaײŒ&7ßÅÉä iX˜S’z š}ø“¢rû š7æ>¸¥r-IÕ¯÷ó¡•Ž‚V%OÕ¾þºÒ»ßKåO‹°,£-ÄŶ52× ™:J)%l61Á]œ.ß ™?×¹ÃjàÛÏ[½dQcD?ðßqò }ò˹  ‹'ày‹?E«ìÞûößyK$UùÅ%¡$ѨÐ(cJ½´o¨ì{âïë®k_¡ìþ¯scŽÉP~(2w„”*($늺S ’mp$qÛ *×îýQÔÖ†êÏ_+“¡¦ÍÇÓ êêý2¿ aÓ6:µÐXÂJzÑŽšÎp|¹fWͯ%Ó.i9ísú­ÏÓ«3ÐÿõKÜ»:èpÕ &(%®‹9O¸eoÁŸíɹ¼îÐÇF=½?é<£õƒAâauZ.åe7þ#±ÑA;ÛŸ`,ìýëp˜¸yrÅ]@¦>Ú$_ƒ™¿Áë¿íçaNO¬æ‹³-Ìo=´úóRƒ×Ä,À2sHST†¬T¹ÜÐrƒµ”9‘±+°‘§w{%¶­ëZö‚Ý /ïØv5$Xߺ?˜ØDï¿K}ù‚$7ÈÔ.d¼F²k­½Ç¥t‘ü΃IÍ f¤LÌ$1õFê­ÔPÂf&¤Ëy’f0ˆ !JOÉZŽ!“7ÿáѦxdyûënºÖ6²SO§Åq¨!gžÜ¸«-òäH²µ‹÷#¿oÏÒm?MT›*©ñ¡E!éŒÊÜ_PÄÈôà3]BûÊk»Ë¢…‡Í³i QJ±BËî°%Ê|_áRîF9¯ X•D…SdááÝQ¨h.BϪC‰Jngdï\˜Äcâ Ù´_¢j èÚ•K¨&KY³ºÌê?™Û®t¢æE‚áJ̨-ÓÛ¨]êºì|)½oQ.ÔùçWTä<®myeÅ%éV 4c׸ùÃ~ÿ|·žzóAYdÂÓú·PÑÇ\¾D5ÍVI9¯B£óŸ¾ÝyÃ}νìeÿ ¤Ë:¿?öüçwñÏšö' {H£âü b'1ƒ¡GþmU0Þ>þëXŒIõEñ¶qÀ¸º-}Ф4Œ·Íø›’lÀ„¢I\`KL¤±hʃI‡Ð/Žji0¥Ù3/Åúþ³¦ÅØ( ön—–1"!aAêQŸoHäüåéF$þ{@‘yjI½G˜ñ€ Sºæ"¤ PØÖªcF*”Ê&•_HÃQòšºé,êB’覑¾çD2[EF¾C¿zw·éšÒ·ñ×ÂÈ\±“›‚¬ìß}‰E‘íüå{t+bÈž÷—GkaßéýÆ E™‘ëRf–r‹ r75µj²- ï‘·q”ÌÈ÷J>ªwõ ò©®uB•ú7'oX¡ ¹“kæ+½üÉI6»äuZ£PëŽs= ÿ»Q%»`ˆ¢œ]ݳî~(föVÒ®‘ = Ñ pCñ~KeÊ/ (¡§œl­ó%ËGÙÆcªQ ,…}C»éo§iP–w64»§¡Ü?3B¡‹ª(¿]S(Ç…Gø†ÇHDŠQ1\I•p«•ð¤Þ‰~(îÏç–¢µg[Ïè •ÛDÑó¿ñTŒË_«aE´.¹*çýmƈM¤²ÐŽ…~u¡ýÉÇçõ~W¡Cô–ù¬¦ :.4]¨ü¡fÓŒšR^h%i^CMæ…âJ=ÏiOäà’AõZÔ0rÀÊmž:Ê¿Åý€f‹xF¦âsPw[5ŽR¾ªî·0ún¸CÙly¥ÿ?(¡°OÝm€ŠËútàÇCÒãË16w,÷¯[]¼TcSßn6³Ö%%aM’SÆösÉI´Vut'‚Lìà·zõÃk™ÐB%é~ØÚH3D þóüÃÞ%ÍÌM.èüèüŒ»ºoïþ½{ z°‘Ð5Æz;îK9&°ï‰3„?nqÂß÷µ÷DT a Þ·f¾o=\9CÚíd·>ðÇ0ÍÜ¡<y(:ù ²FëêcÓmÜ`\©ºÈäœLüûá/éâ S¿ÙÙf2nÁÌD…i ÌqžÄœ¾T˜/I쑡׃…­ܧ|`IŸªÆòÝ,,î>àŸº«‡½ÿ]õ…µÊy¼3°QbB¤/[ÿ>¾Ö¨aÝ#ÖÁ_…íदx††»—ÕÊR ÖÿCÆ[¿Sý‘ôÊ?.Bf{$Ѹ|VxÉoއR¬BJ¹•Äác?š¡uà—×Òà.‘¡ý‰ô$3ÓÉ1È@¬Öƒ“›ÈHp&û½ó82ôèÇ›_Ff29±¯c¿…鯍Kh ²J0Ðÿð"F6æµ›æÈ~›3Ø¿a9RVXK²”‘³ÿù}Ìoä>MM×>…<¾±7®¯ ïÀÒÀÕ¶]äWxfÆhÛ±ßSA¼DT–\UÅ àØ!Q“¤ˆ3Ü¿'†iR¾}@"¥ÈÃmÉ(*&¡é¿À¶Ÿ×ò#u çðPðöé8‡q¯ªÚŠG‰ÃQwB³Þ dþkæ§åY(}¼?¾Gh eÌÕ£îô¡,f·µºˆ \hÐV9©*Øï8Ô}EÅn7ç¾mT²±ø)½†ª§wož "Bµ×+oQãÍPš€? j½=º3ÛŒºTרß6ÞG½»=ë±½h@ØFá)EF!~ù oˆÐäñæÒ‹¯çÑt†¦Ö3ŒÍM…?¾ýjŽÙ2ÃROXÑŠ<·/ )OiÍߨfDë+‚R§¿k Í’þ”³dh›5lž€Ah7•î¬þ¸‹˜†âGö9y´ÏÁ¹CzX¿ô¡D«„lÛŒ}È'øèqxÒXõȶömµµÛÌ2¿Ì¹úˆ˜B *Q&#!ÒÊ+…³›ÕC©Á¢ tÒK(Þ:ø±³þ8IÚ¿p-†+ß—|ÚÈX(Ç`y¬rêÞç /æPÈ[;(˜D&IuË™MQdwà× ËU/4æ[Õw΃æk3>Á IÐr¯8LZí´nó_Ö6û?$zÀ4ù Z38ƒo·ÛA«`Ùª¬Ñ ´¤°ý9—í -1§Ÿ?Ü_Z¦ÚÊ5•Ë Õ€àŸŠÑ~΃ÓVù¡mB/»*Z:Êo‡°Ý†Î°ÆÈåVgèfyvù¥ÔôHvTÿÚçiß…'íÁ_^ÂÓc1‹3ŒÏ%´KÃ×_ò29Âè½tÍï¹0^¢îTþ&¦tþùR´Ã?ò•ÝÙD˜'uY}¹Ó _oÖwÃ’ç•p®TXñ ãT\çËnµÀ&³‰ùR lg§¸ÓéÁ^舿 ·$ê<¼>zʼn&eÎñ%!IÒÕ:‚cHæ›S*Çþ Éoô­Ÿ«?ƒ”¯‰H_º‘!õü‚¹¹ˆÒ]?@*™HŠ 4fÄ“ò‘ñö±Û>Š­È´)þþÙWd‰ ^ìà5@6µ[Éì›ÛÈA¡KÊ’0œ[Ÿ*¨¯• bò¸å6òùŠ•Ð¼ðFþ]Ñ(ö´üqËQõÇ>/Ñä?WDá[´4uƒT(jØNå}ˆçÞ•òÅ÷kÞv¦“¢d+¿ÿÙÅg(#¥fJ(-ÜÑKƒòŒ ™£÷óTt„Ûþä}TŒ%°&ð“G¥Äœ+<ÝWñX;íÎõÄ›¨*¢Ür8^±‚t…gP]óæì{wjÔÈÉáO´ñÓí6æÝçtM‡CôYèì37ºöÒº_^s …^¯-Y6¿nøK`OÁó›Åc…B`èéZôàÛ0¢ÐZjò—Æ8žLÝŽ… ¢ï—Ôœ;aJÂêšh¨Ì8‘?Ù‚Ùêƒç§(£aþ‡î…Ü XT°›Wr†¥‚âS¹Ÿô`ÅAGNâòcX“@ɯ7†„§AH§l…ÃüŸKØUY‹òæWE‚øÌ£ÆÄÕH¤%¤yÊáÎ>oÍ+X7XŒ\ïÄXÉ$§ðçás¤”Ø Þ4Fê{î )UùHw„u¤T?;T"“àqS¹­ dQgÑê ÏD¶Ú„ÄÈ©rèFz\òˆÅ´žÔB~©F×ÊR˜$•-óÆƒ±ý.Üv(<'ueJ Åôªiæžö£ø,QÏ+O”lO7{¼Ï»píÖnñ”3“ë¼Ôg‰ Ê«GtÎQ£âŠó.¾Z¨¤~y`׃ÕèV?×BÕ`C¡ÓòÕ¨†oM.-”£zw “ì/j>wú’|Ø µÏ,gûõ¢®Ã¦ž¢6ê.”T«Œ'ïT¬\vCÃooí±†¡ñpâ9JX=¾Ï£Ð BO˜Ðæû“©ö&¹[_ÊïÞï—6>u0ùÀæÄ™+˜rµ®Ð\‚é Gz¯n0“ûXeøxüÛ¥»-Ò.sÁ©÷o¶ÌÃ|Ò}?£ìÛ°?r¨o2³­ïó{i ‰ÄÒŒV$ T¼ÏÒÃêC~I7X[yñEþu7lܺŞã[àýÎîX>ì;¼Ôüö<Ø÷b®*"Á² ›¼õu$Ò?þm[/‰ ´9žÆs"©N]KiÓ$[R!§·@òK ö(i¤üšCÁ6ˆÔy T»® íøº’Ò7ç¦É!×€Ôæ¥ÈíñÌFHŒyzŒÎœ) E>ízaföä·µ|2l³‡ì1ÙRú(0qÅm¯m Mù˜ÙâÁê—6Òn9(Ô'TëØ"”›{ç§n¡¨^Š*ÇS&{X¥8HFƒ‡ª´ ïuáaÆ,â/¤PâNu›ÅC”\þª<à„Òçï~e§d¶ ?ñ¡\‹IüÔ%o”ïàIú£Ü K/ž )¡â¹à6¹<¨Äñå à]<&?ûüS¼ª¤ Û¦?eG¼2ﯸɌêj2˜£v5f¤V/|'G­ï$Í¡ßÞ¡Înå²Ë]x¢<(Xuk õ·U£Ü{‹ÑFR?ô|¼£Ç2ÆŽ&IÙI¿µMKƒ ¯OG¢Ù軥û=´ ø}cüþ'ô— ¡ÕÑ,l'‰ÄS˜QLÔR‚ÖÚLÿ"ÏH£1á oG´=[yã=á#´‹ ]5:€öEy„’eìêÞuóëe´¼@$øÞ ÅjSûC\É,5ŒlIJ`hH˜FÌý ´×˜µãaUvŒ@¶ŸÛƶ=ãÈ>ÿ¤j©ó rŠUt©S>B.ÏÎ)|qÈÝôÓF=yµcÊ/yÿD¾‘RýöTäo[s=ÂÕ»Œªx‹2zVàÁ,ááRŽ~Ê\»¤êñ …+ ‚!}EF£¸*óPŒ6j„ïa-Ò¤m­¬GñÐ×Ó ÙxxÌ%$àWJ^YˆöG©YÿÈ¡”‘V’ÙÌÖDYaý÷)”»æAÿ7Ô,©¾†¢âG£šŸÅPùÙà î¤T™^k‘¶Aµ“ž ’ß~ †|`}ŽðÔ’VtŽ D£Ž$ê )<ñ|.C¡è-ê·Åö—ZÖ£!QjŠðK4¾”ž 5…&³ÕmÁ2×ÐL2ÓžÀ—Í/è ½úˆ¯r'ÕУeÖÙ§®"—ЪÌÐí«>ž*ÜHê÷UFëRŠûÚ¢LhÓæ=ÍžÀ¶›âÕ#úhØ‚ŠKBCîT“0>B«Ã…–gèþ!/{æ?Y›Ï°6Ó”V\ñÚ£D…®†@Ãu÷›ktPEG_¤æ[ eÅ)êKX %}¢íx ŠE­ü;6” Ð¯W. òâÓŸÏIÀß1fcà‡Æåcq;®û¯ã{—O&Bרˆgn+70rô”A¹AoÿðéE¨£l/ƒºâÜFg±EøUS~Æñ=4™ž+Mߌ†f7á„7µÐâ3%–[­þ”:²á·þç¼áÆnÉ¡ÇÐúòU÷iBghYþóq0 Z®Š_p¨†Ó皯Uûžx†{Ó\Z¶éøWk¡õÇÊk0p‡¶˜*MZÓ:hÏ;ÈÝè(tŠ|ÔT¬9]Ÿï_oP"?ßX/œV: ½ã#vOOÁßWñ³—d3`Ðt›ñWÉ( óïÔ–+ÀèáÃ{¬›0îܨ¥6}&‡|C?HºÀL|Hïh|̽žý1Óå ×MWkŸÀÒ‘\ £›°¢kÿ;ÓÖ|†|¥„AÒÖ9ØS\(8ž {gü"“^þBB“—þGK¨Ù¾dæ­#’Ü+=‘¼¥‹d¦á†ù”H®aðåçY¤táçŽÊ!B깑C‡‘N¥þ("ýzo™žG62žû6w_;™ægå,U‘åA6QBÕdS0ѲŒì»o›ÇÝwsÁ¾‰¢áò°¶ÈRÚ ßù™ñ3gö‘ß„XÖ uÂK¥ãÁ¢„ÃüêE(ìè.~ÂI E%wtY2â!‘9{½kQxØèë»ù_R(™Øk7Ñ ‡2zž+’Ó({'óÀÃ"”Û¡ØÍÿ éÙõc ½¨­ÿ÷g¼3*¥¤æ=þvuÚÂJO ªrMôRºP#~÷Lâ‰Dõã^•>Ë»¨‘!’ªÑ»ï‡á/–®» öØ6ËØUCÔýdL+VŒzW?EÍÑ+áI“yuÓ·ÑPÃêhOkÓÝ®ï)D“6éŸ\FÓw7ê¡YÞ§+«“hÞG?ÏãËŽ–´$²¾™¢ÕIîé1¯Oû~8Z6X¬‰ÖÃ_µ™›ÐVÕÌå}ã´{w oý$:00‰$QWìû!§}µkà~/µ ¨ùð¿üð÷ÌüÁl bê;—¿Á†ÆLI—/…F¶\‹žº1.Xzµ©.TøZUíÙç¶©öó±nAcPZÊŸuYŠ^Vôî{œQ!ë9Èqo¼5 t²J½o=Ø¥„o²þ5¡ökF¤×èÁ‡'&ï3@z]Êê¼X?dØ®„©¾ŸƒÌ‚^RrnÈêÚS7¿Jýßq~Œóx[‘]‚ü¢"¾Ç+W¡ð¬i—Kù»ÿÎû3ó³ Å×\(~CDûÛò4”„Õò”j¿ÞïÁÞì×/3AiÐdíßÞPÆ9Ö1b×e³…U/OAùèˆ9iûq¨Èû›û\ªôæŠ_S4AuhQŸÉBÔšu–‹ÓùAÝæÇÓÁâ™ð‹]õWÝ;VøÍžÔOI Mߥ¶ºó@  ’,¦@ë»k&dÙ× ÝÒ8!ÿr%tÏÓTm\‚n{K •I‚ž–ÉQ[ÌÐoy¡þÓ“RˆQS“9ÛCÒR ©sê0ü›ž ©ôŒ¦²=Üpøãy‡Y¯Àd†¹kL÷‡‹Jy¶Á¬6/…HV;Ìß õ{qE&d'\7Ã’ËZW¢$,/šcÊÉ…ÕŒ' ã&<°þöÝöóc°9ë4;O}rLÝ!–Ûd¡k:޹5Ò qú;Ý õ'Hÿóû3³ƒxàm¶ØµÃŒHQh´0˜$†ÔT’Œ†…ÎH›vfH¿V=i‚ŒVÞ¢òÈ<äp€¢Å Ùäe_´Îò#G1?ñfß_ä~} ¨Ùýò•Õ¸$ðú¡@\§ÞkÊ~<ÌÎMŽÂ‘,…6ŒQtôjÉ í-ð¨ûãh4JzÝ¥×[|‡2§£>æJA9éï6wP{ô/Áò=Td)ùšáJÜ·"¼ã±ïKt:Y¨æijúõ5ªÉwÍ]DõwÓåéO¨Yøó¦o$j'¼æ:¨tu“¨=+¾¼E½Ìù j¿+x²+Ñ}^ÛÈŽ_[÷ýȧ¥rëÿæ0Íÿ¦Nå%|0ytÌä<Ú‰ ׈[ ÄRQîp#$/ÜöþŒ2ÇVhøÈC7ýŒ±%h~ðœÚ' ´âg¿ÍAM¢éóÅç ò¢q¸O=”»^Kÿ¼Ø¶ŸƒY%çPb¼5ê°ðŠ·NœHsè’¯’¹†.P¶ùò‘àŒ;TqÒfV¥„_ºz:ò )ТFNÃÏí›G£cnBw‰;ç$¹+ôŤ%zÝ „îß,{0Lmáó1åŒEÿÎP…ñƒýeOca5öÈ8¾ÃÄwó£ ~01Âl8“RÆïëa2K|øB"!LÝoq^\„é$«DG§˜é~œé> ³\qóöû¼ê`4_ÿó_ø¯< ……A¢žFܰX %_ K‘oÁÊÁd‡AX}öÕ/.Á:‘y? l<%„ÒKt°¥]½Ú ;Çúƒc,ÊaÏÍC3ƒÛ ÆVïK+Ê#‘\íü’ŒGßÝ¡5UBRÖ̧$éH–ãjöí’ß¼ñyÆE)­ƒ?¨/.#µM6Ã@2 ÒúŠ8ÿÙP@z‡²÷Ìæ‘!¢p›ü72Ö­yi?AfŠÍ™‹ŸÅ0¼ã± ²¾4Oc'y€l£ ^Yû~ˆÉE‚²¦È™h˜xÁ÷4rÓ›ÿ%UÒCžÀñS>[È»è’uòØoä—¿,:w¶ù—ÏÖ3žTEš;æëò(˜s#¾<ßç_õúý&3@®("¨ôv8 EoÝHÐÊIE±’jn@‡vH_?ÐÇúôd¾ú(ñM↉Ë5”¾é.$ºŽÒ‘1úè«PæÖ;²“ý(+rá©¶e8ÊþuI:5ö årU2®¤£|ºç=õDT¨ÖÆuCTÔleÊô¼†G÷ ¶·Eð­< [Ñ Ê%ªœˆ´= éßž Ú†ˆ¨Ø'ZÔˆ¸Ûe)è‰Zª\,uŸçP‡UÒÖh+O°¿Ñ˜5¶@}ÙMW6Y4 1œ;(D†Óq£&TиãÆ!u"4/ú!áõM§øì/Å>@³éueöK$h>ξ7廒Ä æÂ“hùsñäno´J ³ˆNÅSåU<—ÿ uÀö=ys´ dÞÚµNCÛ‡ï~ÞÄÐû•}ä»nh‘žª¢¬uEm}íT#Éd’Ýg,¬—ž§†¶Fâi ú×ð[ü¼Bñ]Z¨.«Iìúוè*¾ž†Ré×K=É¡HY¹O"¢òXÕ®“oVìó­çŽè•{ð]y}ôO«|¯ËÓ.IÛï©wz[*B ¨o7Ç^|Ÿ[Å•WЪ]_Ÿìl"*?hRe\¿ ÍIkW¿ˆí÷Í…©ÉÍphW’z¾¶ù:ÆÔ>­@—_׉´ì}Oœç NîR„žcr™îo_@¯g ýè£U9½7ôÎÂÀ»›a3”0øJ¬Ÿ(>†c©ÔŠÁp‹§—ÕÚ$ŒJ½ñzœ° c¥ú4Â0ñÚ ãœ}(L½;%a 3ó—¯ýƒÙe¥æl5˜OÒèÿÍA ½¿×6,Ë`I¤Ö+¿>–C…Õ²b?Ã*G³½Q0¬u^êÏì„ZÎäö°50»nvœvivbºþ~A±Q¡.A$4"“³0“D"Ÿ.GêÛ…Hœÿ(‰kYIÒi¤Öv‘ìž1KoV9’³j”àIBŠªiK  ¤zÄuq¶Öi\þäÎ\D:•±† Io¤\hþ‹zB¥·¹µ÷óÊõÏð…:2E6ŠÖt ógæmÕ0 di–óÕ®B6’~Å©>dW¿)²•„ž+y×üô‘³X&.!·ð*ÝWäIUýpÝùN_wóXIAþ·oRÙ)öù=n2àˆ‚/îdÓ³HáÁû‹FÇRQÈ—ª®U*…Ck|LÃyQ$qo™ú^ ŠÖ]|æ%xèfIf ¸¶%½vo~'õ俏3J•/[ªxþ1èË@\͈¨Òêµ¼Óú E¨™ð+ü8¯4j?S‘¼}ÄuߟkŒ@½çlú;ÁxòóƒØ ¯bhXpÁF¾™‡‰‹¥¿Ê—:Cîˆi¤óÚäËD½p;>Å$ä)•7Ù¡l9f{ÓêT%bâH¼uaE 'Ïï÷­™)ìФ‘þˆå¹74Û:Ve¯ÏA‹'s»¸Ô ´ú²»¯4ÇÿÎËëö\ÿÿçu$ÏAKÝ¥b~¯KÐrTáïöý>I-CN{Z´¼3/5@KÚÞ-ÁL-h½t6L71Úô4š­¨/B»3u·Éýè( Z¹UÄ]êÞ>ÄêðGÑŠrI†zÝ”un2ü¥ˆv͇ARsJï{50Ôxñs@-%Œ´|Ì¿ùOƉíÄ~* Á¤?«@³ÉÌ(H”7Sœ$Ûý/ßé`Aéðô‘|EX\{›Ú»¤Ë;iuÁüåûŸ¿­Õùa Øø²&o>ÍÛN*Tçö`™×è® ¡ [.¡ å=äÌ{ýI\«õV-LÔéúö$§-0/Ï<‡”‡@>†t©ï4ºOz!г´ÛH?r8æ¼êWd´:|ö±…2ý}™YÌÖ€,>CŒîWn#›˜Ô[‚Ôd_ZÏ;J‰œãt½Dm‘‡âr‘´VòYeÒæÉ?@þ1é[ú(xÔé¤ß<ø©íDð¼ kÕèTö™ (­™ãw¢š4Ïy-«ÅÆ]D+çx“ŸO=waä’gCë¡éÜ©´U¡Òþlðí¤Hªs—ÑžŠÒ¹³# +~®Ûï¥ÚD’†'þ—6d¾ ÊÇ †J?/ 4ºîÀ9P†¬¿Þ>Ѻ ‹ËsU”9 %ÞŽwñ%TIÜk¿$¥ÇM.!)úõ\KÎS„¿£k8žFÂ÷–¾ý:È:ÙÒþNÏ 23Øj‘ Bóé²r9HW½ËÜ(Ü_«ÿÊåQxCz–À‡¸s$yÌÛ¸Ëxjÿ}mw:xÿ;ΛÙIYþÿ>ñˆñý|Ûk›é§ýwÞ¢¾þ¸Y»QøY«`3aÌ %,<”ï$ëþã‰7S*÷=ñæfo”>:æ*eÎ÷µv–¤¡Ü`4ì¨z:Tpöl®‡ÊW.5¤…Ï ª¶D‹j¢Ît]„:¹! hÐ2Íܬà‡F¥Ü»,ƒýõ (•lך_š@6 ´ ¶¼;^T m¿:ÜÙ´„÷sñ‹ÛÁ ºªX(Ó”BLg>ëmcèK½ùøˆ‰+ €efU  6¸†{!5 _0ÀÕ¤*U^«ÑýùÆ9ØL “À¤‚ íîS˜v¤ùé°SÿÊÙ?q`‚yÙò'×lßÀBÜ'74`‰RæìÆkzX§ù¼¹÷Vþªð‹gÀ:csZB¬?l6hŽ,…íщEØkSV2—BŸœ>q#±Álýµ0J$=¢ ¨:Bdjk4ä¦ñÊùA‹|b& Þ™Þ( ÿë1‰¹½ªŸÍõ…eÖÉ‹d¼Q4ˆýzK8#Š n®xË¢$ÓéUu+”92ŽšÔQŽF‰Bçs2Êïµ?(ÚÄ#ëM웣?ðèÚ=5?<–_®«ªO¢ý5y š}“õ?FÔ``Ì]î@Í1uF†FÔîòQº”º¼qæ¬ü¨O,\Ñ%i„Òï8£¬ÐȦ¾®7@M”þÞBSó©÷ýG"ÑìV±¨¢šº˜šâÓ‚/=¿ü@+ùÁ˜Ïv>xê¹ÉÝøÿÉ¡znÖ/ ´ûPM[ÕèÌá[‚%èø˜obªÅöÿæ0ÍQ¨WU (û³Yd€¶$|%¸êPÜêÏu“KVH6}žuí< ¿·å$R€Î?¹äNŸÕ é´>«Ô[/ø½yÁ ÕŸù™‡(AÅ—½,á6(Ëk(á!Ú„’a†“~Wü¡VNŸÊßçå—åk¬©P|{â [”QX.ä÷[BÕ¹3 ¯­¡!ŸÀÇЪ Z8J…ŸuæCûò é­H-èîl)è÷„¾Ž§-þÛ0°3õ%íŽ% Ÿ,6Œa†Ñôõâæ©5w²®Û ‚‰ŸG½OȮäpÈók¹ßaòxB“VØ L^Õ®f³³€É¾mÿ«fi0õÞìRÀ² LW+è¶øÁÌÖ=ꀸÃ0«,~îiz6Ì}#Láu†ùœÂšÇv‰°ðW˜ì[,–W}Ð5€¥9T°¢F­õ VãŒÃLÓa¶ÚõÁceØxËNzQ ¶,~Ì>sñ‚xñW¥³öì«C«k‘ ûùÂa¦e$â²–šbÍ@âë%á$VH2c¬“Ýd·#’¥,Ö‘\*ЏDÊ )‰—£|É‘j£:ÿÞÕWHKkaìý¯ éù…ÿ!ƒfÈs^ d¼þ‡#?¶™>ë“Õ óøɳkÆÈz¸ª$ŽZ Ù|"^°!{§ÒLjœ«È©ÊKh¼‡\,®c'¼GäóÂIù«È¿©~×~ù©ïo]û‚üMrƒ+9²(ð1B­…+ãGHj-ð`ÖfôŸk×PhîzQÜâŠè^ëèmDÑ-⫽ê(¶Î¬7í½Ÿ3ùÊKÌ…xØ÷JÌ}7”{ß×àaRVj/Ê®>Eé¼¢é¤Ö&” ÙÊЦ­FYqQVALCÙ¶ú_…HP.Îð·ÇÊ?¨vy¬ç† ±+3ñF/P‘Ÿ?€B²ÐYt´×¡òÄÍ©w4PEUÏp#|­*s§DµoÜ´¯¾¢†DG¾Ú“^Ô¬”µ¥y¨>=Ï£îå<‚øÄ Ô;—ëfošˆ'i®<ÿù7 ² ƒ¶¦N£ÑéÚĬhB³’5²S‚&³qÃ]ŒhÚH—˜?ïŠf)²¼^¢¹·A³®&Z€|—›ZeoA˺ÖŠ;¶h+iw–Oy 4‘1CkË0‚ïùú&KÃ}ì4šÏ·{s ð´Þ ïׯHìOr~{‘¢¯™}v ƒÖÖìPŒ~¿.îªRÝKßÏÏcÑ_Ÿƒ ü¢;íò(1!k¥›ŽÚç–Wo›ã~ÿ´[ézx¾;èœgە죇$(žyCvZ3ÙЗdzÁdQrìjH^¹¥™\kä/ ŠN«3“½ê.§óý!h|z¶Í‘Aš–Þü­MÜï£× x³ ÚÈ^‹O>ƒö8:i"èär¸dR®²ï‰Ó½)jÐýÍŠB[ÞþtS™ñ»{Bï±<$;ûúò¦×–Bmà¯}Qâuør²ŠÚzÓ͹¹;®ÀP‡ö xµ8Œz`"Y” £ÅÚ™ŽC!0ä¶üí LÞ´5þþL¦ý4‰#Uèàß—oNól07+0h² "rtõÒi°èªþÙºg–²·/šßЂ9³7’c¦°Ú{Ê×`tŸW9­ëöa3'_6Y{¶k£N ÕÃî’Ñã*DzT]f1B¯›tâƒöH4à0~È•IÄÎ:0 #éC¢>¦}~˜4}EÇäïnÒù=ª@JÎ#‘¤¦çY#*Aš.ŽS5¶JH÷¹Y§´’éël-zT—‘¡:òwÜÔ2þöѹ;y™¦¢ X4…UÒ¨HbYõ‹‡7Z‘-¨RÔ–¯ Ù‹DbßÍ’#'5u{rù’ñ÷>þ±ï1CÏæòörVG³$#ÿÅ |ô^(pM¶Ð5ÈÝOž‹ÊÁƒ–é1”Y7QH‹ÿÐùÖ¼êêf‡"§/4˜Ì¢h¨ÕäG X‘|¼|âÄ´ Ó$~xØn3ƃP%:ã³BêPÊü§WP*J7ê×OÔX£Ì| ¥q€Êi_KK{¡ˆ HÎÝbïBE¾¨À¹o¨´8åÖnŽ*„ÝáÕäˆV¾ÝÇó Pݱˆò¯#j2ºfwïG­öw¹£9¨“ÃÆw>POøˆ=tz¯†ú@f¯Xˆ”•©xÐpâf»Å¦:ÿz|ÿÐy4é¸G«ÿµM rægèÑ,Ò¸|ÂWÍO5†{&Ø¡û°”AÊ[´è±}¼ªˆ–Iw«Ê¾ U™ôä V<åáUJ‰!n^vTË/ÑR½A¿–Ÿ¹I6Sý*`õiä›Ç–«Ð¦v?çê=f¨Ë¦ü2xÀ*&š:Í, ´æ~ú'Ž (Þñþù䊷[ÊC×@ä)¥ôÿœ÷e{úx{ß “]¾KBNì°ÄT¥ã~þH4'ËAž¶Ž–M ü<¾Gtñ*”ÅYr1œ¼ UΗ¸œ¤ Î²¼vírüºË«Eò‰å?Ï/´ö•°€f£è™†÷,Ðr}R>°Cë-‘§ùlåÿã‡4œÛ¬¡õtL„Íåý^­žs@š÷4öH2.Bs…ó®îå-h^ h:b-þƒBî%ÐÊ×~ÛdƒZ7ŠÞTSA;yæÃ#YÐáh—’Ÿ9çétç¿L ­…ž‘G»sÕÏ¡ß?ùÐbt% üH%âÓI…!ßvÏ#ƒ0â#|£›*Æ’'hއôÁ$GFÁ¶—L×¶¤~ÛS…ÙŸj&WžæÃüéß©Ô~XÌW:ö‘ –+‹T‚5b«º:öÛ°áÓÜ:6ÛG^…ؽ°…=δïúi‘®¿+½.‰Þ}÷ ‰±D-WíÙg$ÝQp«ýW†zÄ ¯%_@Šy{ýW–ëH}B‹jý$ ÒöÉÛÈËf!ýoŸ+¡”)Ȩ•Ôã2LMì7Nþ±G–Ëô‚ЬO›"Îëî(²Ošº‡±"goÙÉ]1äÞ¦Ð3WaD>­ŠtÄôÈÿú\bQPL.;“>F‘ŸŽ¯CáÃÂg4í®£È¬é½ú&ëÿGÜ[UÙ½aßtHwwHHƒ qžtw ¢""****¢ ""6* `´HKwIƒH—t§ ¼ÜÏÌíýŸyžïï§5³g_kí™}ë8~³ÖµŠõø't®Õ£Äú•÷B”A(¥mË:{ý<ÊÔæ&¬}rÆUäxQ¾¹'ò(x£bD¦Äæ÷pÜÿ¸Nu¨UÒŸZúuDÕŸó%¿ÍºQƒ\®j¢1Òd‡{Jµ$OŸ¥AíwäßnÉE]‰k½]¨·EeÇn"Œ­<û|E£ä=ü?–çÑ$ÆÇÜïöM4{VŸ$ÄŒõeÔFn¢e:CqSŸZåzÝ[vZBëz¥ë~ºh3åíPš[Švl¡Bn¢½%éSòºð]>ì;lû~j7—ŠK‡Ë “j¹XZm9:Çuе_ŠC:¢Yþ·èzom&a‰i7—*ýînžú>¬¹¨¼T„Ato“5f7Ѭ•µÚõ²#² O\©,i†E !)›¯á»õë$¼"Í™:uíO@q G>}%äw ð„‰<‚‰6nïøz`&ØÒÅ2:däs#ú!]:#“ Ùt— /ˆƒ!yû' ¹‚õtÐ$-¤¤9ª| –ügßi´÷‹È(ü”çØ ò·Ÿìú¢ß F…ûîPQ»ä+xfh5gÿ÷Û†?­”LðmÅ3аg7ÿÎI.ïr"÷· !(¾N&ùÃy?”P«¯Œd\‡’6“2&+r(ý2Vâ•"eW7œUwç•ö&ß3ÖP)TSæùU¨êÞKRw j¼/» u¾ÇžKü‡×áé‡"àû«ŸÖ„² ™ï±ÿ…ÃЯ°è[Åmú.J~œ·¡CI<ülÉaè’ië´Ñî†îÛÏÖ gç¶e 9ôWwèç6%Ãà‰—¤º`˜8ôºQˆŒÔt‘=ç«ÜÕíBS±P4Œç˜öN*ÕÂdï#»û¡¹0£ÅdôšÌ æz‰ucÈÁ‚EoWAa,¦{Íp>]‚e¯!#rXÉ•Ñ9Hª k!„‡’X`#•qæÄV7l©ÞX¥ T„¾øXiÓ$”˜šiö™F¢Iu Þ–[HR÷éÁÄ…m$+é?wš)Ú§‡c– gjlÔ‡4¤IKªó;QŽôî=câÇ‘Q4¼„óZ2 Oß;êÖ„¬/]£*ÝAö þÇîGëúàTC’òÞïê3ÓõEþõ²–S/SPðûÁ½\(ÜÜmÇ߇{¹KDËÉ“Qü›ƒÎµ!Ü—=*¹Ó(K¥¼¼öçÊMæ½UxR Còé|Q©ûËYÎTn¹?| U+“4¡FDÁgÅx]ÔG6€¶+U-ÕÑ^í2Í‹Mjt¸GúŽ},ì?¢9ç±ÃWÐ9ž÷(ù<5º0ÆÚ¾LF×ûD}¹ÿ·“4ZÎfテ˜@: ttζÔF1ÿÔ¢[f®H*M±ðÎp†ösœ)™{ žn]Žáð}?›9Õ ¨•s2½`•_f_»¦€²…ÔPߦ^(¥ÛÛô€IŠu]ŸÛ÷ÞÅ•¶+æPÄôZéûeøöâBîõPÂÎjoݤ QÇ·„ÜÞCý¾ì”ã— v}ê 028BûìQ1·PFø±éö 3ú˜/ ‡øÃ HN¸NÂg¾'Ác 3EÒaœ½ŽÄ=&4$gÂŒÕaâYî\ó’L<É—=O ij3Òïú`’¢‘ã±LV|׌]i…©_•rz0ÃsÍÏ¬Þ fí³ûª¸7`®ZJšZ—æ‹gSd”a!?_ð%,ÖM40;uÂÒjÇÇÓŠwaEgÁOJæ&¬&ÌY¦†u.÷bò°‘,]ÏØ©›nÝ,R«VðGÛDeÐéìØðÖõ_L@‚z9WÑ…¯HD<æ3ðç[l—“gŸG’¬ë4wEwõ ž¹Ò¼fŒä£YÎO!eê‘/Ÿ¸‘*êçQ5i¤y¬Á8€tÅWN=hoEúß– ›ÈÈwL`î72™ö‡ H4 s°‡¼Oÿid)=ÐÎÏ4Žl î+ñX‘ýtýÓ5µûÈÑ9ñ8J8¹,:&«_#w½SL #¯Éãï¹× ¯N%v“ù?ï¬ÿAЗRùЈ‚wÜkI·ÈP(QJYfõ; ¯¿TçU_FÑ#aŸiÚ7po£çKÂ.—Q³ñMaF ïÛºÄeP2µþ‘ÑM”ââÓ7 ÔDé;”…îÙ(34Þ÷¼eŸÙ;AuÊIDïˆ7 \MÅ"ÉáK(dê¯~^<¸µÆ­ÄQñôû¨çnTÚ:5é>ÁÊÅ–ÓROãÒ‚çÜ€j˧¤ËšøÌÓGßÝÞÕ'Ã*ÍsÔjˆ‘ü£‡:fq–q÷QwÕ⥛ꗶš¯R£a¬Ÿ¹m83Ó/Yß}FRkÙʳÇ`ìxáÅ · 7Ô‹T‡IƒgÑÒdr0}\r†¼fŸ’e‚ù3{c)ŸÃ›ÜÁÕiXj}ºdÑËf÷¼vy`epñn>Ç]X‹O6+ˆ‡ðÛÞ«–}°y[zMòÄü‰¸V@5÷v²¾R¥õ Á0W±‘á&‘IKîo#FbäJZ¯%G’Yñ©çvçˇƒ’‘<€¨&kÊ)™8Y’o‰âžoÎÏ ŸÍ!õy“¡u¤UðåGzš[ÉÑûë‘‚4}Zë+2ÒŠzSN#“û¯Ò×öÈì¨ä½Ã¬‚,Þ$ÍQÔ k­[ç}]Zd'w¸ÚâV‰ =|¶2³I‚$‘•¹ÏßññpÞF^‡R±oJo_Iˆ>‚bôêû j  6 ‰.Ÿ“¹¡‹Â4Šw¶Ï¢ãÑKZ‘(ÊÓͺ÷j3ÌñXÏ¢Ø}éš?þ(^\uðѾƒ(ÉZùÆ‹K÷Ý[ÑèòFi2WÏŸÂP&D…`È_ e;8nÝUAyõß§¢Âä!–Cƒ¿p? ùòTyÇYª¼Ãjá ÚªnY"ŸGòQsÙYÂ˵onöù›=@]rmëÔ«’è$:| Lç\Sê7Ѱ)âÊþ"s4vì©=e4™™šº¢¿€f!§õJ·Ñ‚C}øÏ¹<´4ÍzK¯dŒ–Ûå½ h•‘èu­õ;—n~kAëñV…Þj´yÜÛºŽ¶ºç‡§n íÖÇÛ{~âTGZÍŽ¶õmzb¼È¹ÉG)À +ÃóÏVâ’ õÙ—Úa#g¨is2ì > åviÃü/¡ÄtÉóY)q=îé³V‚Ïsm7e!ï¤G‘vÛ'È9{Éä‘ëdŠÏñ<¿_«xE>9CMÖš†åÈ9¸²l¶ë7ßί¸p'WB©.Ã1ÍZ¨äOúÄ6 µü5U¥¬wó`]¦ÃgFhA¡ yhÖnf±Rƒwo~]h½ ˜è2ñö+ÿºÚ ­úl팷ÏCË9ÖPº¨ih®²rp0š¯eׇ4<‡æ7´|¯:¡…fÛXTTZZõ¼œ,L¡5Qmf4è3´¥Ùà\<í¿•n½¯ÎSN!gmœà´›sð]ƒŸCÙ^WÆ¡wA´êkm ølTS†Ã¸¦ÊT[6Œ¼•S/…1}¿…œ¶]ý~5±ßI Snþ£RϬŠn™Óž0_Òßöêe,4Õž§™„åëÍ s;)°Zeƒ’fi°að–Õf( ¶¨Œ›>…íUݪ8¢$ø“ðÛðé8Ýl´‰hFá¯åO§O#i+Å’U’ˆLå{q)³¯ŽNXFjú=F·’ž!ígm¾þ¬¤Ï«ÑMè|Œ ÜVoÏÆ"SñK•®˜)dqy‚Â÷ÈF75îÔ‹ì}‰*‡‘³åÂ2aÏ*rÏÆ·_R8ƒ|ò\޶D!£YñUEä ã$|†BÁê7­Ì,Q„!øzñç(Úx•Pïþ&Šek¿Rý… îÕ÷&¢#ÝòÙƒÌ(ó¼!˜åä¬sKß6¢|ÞèŠ6T¼£Ðb9‚ûJ\Ü ™A•¬;/5,QµO8 dÕ·V6Þ'F¼›“Ìý8µ„Nm>é@íᾚ]>Œe¿;)·Íî%֪ؠ¼èÐÍÏPÌ?LC!ÿðZ͙Ǔ">À§k ™¯ÿSœY„ ÷{›Œ -ÙôÓ7»‹b­vâ {'$S E:\0€¤R³DãÚ÷œü ¶/ÖÒ”G¶úïü³uõg€ùß~²ñ±L‡\Ö}ï|ꂼè³|rŽ¥Ç-د皥 ß"ç9µ£ÿÙGP]²e³Ë‰üŒª¶P´)–›¤˜ Å·ÆIövA‰“–XJ?”rþZ挄Ò^ÖüË¡P®bØõS*îdŸ_‚*žÆ%ñP¨~«3åñjc._qÞ#õ·éH{ 1OÍ²ÇØš>UŒfì/„‘g: M>ÐZ8¼Ó4?í*MªDœ¡3*âÝo“Uø1©Ãr¡zlŸ2™6'B? ¡ø˜ ’5_'닇¡Øò¯[0r0GwäF;ŒÉ:<;*ãêFIí:0éÂ8ýé3LÑLô¤>„9ûëÞöÇÅ`øKÓ$… ,:¾ñS ßÍÉ9í"5OdaEãºñãþ.XÝ£ÎsI† úƒÏçHÉ`ó! ˰©,lŸ¤'¢¤mB‚é«¿¹ èÎÝ–zÎm$ñàÝ|F>dÖ+äJtÂ#¿çJlîyë;Œ±H#£È0;¿…ôl·7ÏÞ[D†–cÇ?1"S¥¢Ùã÷·‘e:úìÞä3ÈnHœÒóæ=r.Zð×Õw#ÏÒan¡Î“Èÿ>µc„/éd“qÑ£ð¹‚7·.V hΉ_o ¸C»ý®Ïï³â-•åy‚2% v{rQ®ZCÿò2*TdöÈÆŒ¡RÁóš§–’¨œùŠ(ôª®¤ß4tC¨Á.Q>vÔ¼¼pZwܵ Oѹ<êªt·]_7B}C•g¼ï£á9í¹3¿øÑ8áá¡3iÑ´‡ŸêŒŠ>Z¬«LýªGËë3‚³\‚hõP;Yõ+;Z2”sþ<Ú|7ù´ÿÉ,Úm|^\ßF{úà[ѱèp?G`ÐøÒ:T¹ñ'÷':¿¹kÄ.ô)ÏB¼¢Ñ5Ü©‚sŠãÿÖaÏ ò}¯ðÁòE¡ŒèPøb sP4ú6UŽ´'Ÿ»rU&:*»*‚U¡Ý—%ÛŠ´ >G{D3A µ‰ò¦ý¨(ø&ÙÇßeJn.G®ŽBÉÑà!(2NöHs…‚€ù{îOrÙäw¨!ä%ê[(ƒbÑŽ¼ÉR (¯óqHœ_†ºW§‡‚^JAsƒ‚ýcãUh¿Á_v}ºÙ¿k{}Ö‡ä¶î/àYˆww#! ×F,úoÀع¼ woê~[£¬& øÖ{\‡IÒSyé^˜˜è0äv¦‡‰y&½éã0©ð*óJ&LŽ™2ç¯éÂ4¥¦#göS˜ÁK=\[ïaÖOÏn+ `®Ï1üD Ì׬07ÕÃB…#¿ÌÕ»°ØF½L~Ú–¶5/Þ €Çæj[XM}¼|Q%Ö÷þæê½ùz»dZ ›Þ›j5aÒðGŸðÉíÞØ1ºH·&T‹…u‚ß/ á´_bçpïSÿpËû1’„««‹AÒ^ƒ Q1HÓñåxe R:µ\œ¾%ƒTæ¨jÈÊ‘F1M"¥Sé´—³Zfê‘>£Ä(Ï jç] ôdqÖOJ-¬™¹Ü Ìî#‹cÒDÝ“Ãs­ß|f¨~6‡ÔüYýcÐfÆ™@ãË ¼óý\–5ZŸ—œ¼ÆßƒBÔãÔž-îHð¾òæë|S艺*{è 4÷³%%œ…ši ³w!óP–Î!“`™%ŽÍTéßšw}©àÈÁo³ø;JãÈøzãÉæ¥DðE“[cƒ't×ïîæ§ÿþo¹ïXA¦'è¾ÜK ¹{ÂZ×ÏÇAÑÞû%+û5¡ìè‹ü…çÕP¥ôD¯òãíhs<þ%²?ÚŸ‡›êÙBÇÛ˜œu§:èLìá¾; ]#§?y5@76Û1›RÂÏŽ[Ê¡erÐûYh¾´‡þÇaW)S` w:¯åX2 ñ¹’Ÿ\ÛïÍ%îÀ¨ÙËg|®¯à—d¡f{œdk|·åƒ‚ëLmÍèQ¨Ãp,Âë( 7ñ ãnG‘Ÿý(úg4öì”%Š);¾µ;æ‚â7•l÷£Dó“Dw¥]ŸÔ4©–ùŠR9ÛŽ?…/  ˆ(õˆI l~FâÚ“-”çðl04ü… _Ê´WPixþ¨ßoT±1áy~n ÕØŠg ƳXÏqÓI æÝ cŠÒ<ÔZyJÁ:ª:‡|ê‹,n¢n<ˆÑN@}¾,…?ËYhp„Ó›Á` ?<¿¥sÝ –Ü _=}ƒ&XlC~èšÞ>ÿùØ^4Ë÷yïõåšwŸŸþñ*-~I=ù#Ü…–~õ9ym¤h¹òV¿«Â ­®>côÕFkòÁ39Chý¢¼”Ñ®C,2H_Æ ­C¦Œ¬4r,诖W•ý–ý:-“Cýto¡zùÓpíÎQ(+}4W,6ÅÍ<®KEP¨öG{›yëŸõ„ŽÖt)ÈÍ™•öÿgÝáúCéú]LÍïºT™KùTî;ÂWÁ‰‰[¥}:!^³Z~—+§²«¸ÿϹø¸Ÿ*ú%ç[¢ fê{ÅýŒ*hP»,œ˜LM¬!sú® Ь¼Geó´8DT¼ð…ÖsÚ½Úñtÿña‡‡Ö³Jh•|!¶ž~Zô¦|îìKÚÕ­[);1õîó4Ñ•RrÐls³#÷4^û²w ZB¾:Ix­N9Rê&Ðæð{‚@\vW×tÔðò-t›†¾?·òÞ é¡…ì§ßQòÇ'‹á(tÚ÷uÖf' ¯õÇ€C.ЦęĢØÂ9Ï@B”x«Ÿh§×Šûëš—å3QÆcÀ¯æ€7ʱõ®Ð3¢üûTyšV T¼ü%Œ¾E ÷‡ûmgïåF•ü«¿¬Y QuàŽîv1ª/&ß>{ñªˆ3·r'jqyñ7s× ö³ñ¦Špé]>4ð¬â¶@}ÎcÑM4ؾ疈F“Þ‰å?>¡É¯qç—EÐlB‡ÿ$[ZôŒ|ž1AˆšæõÌ´jê|Ëý+­ûõ·^/Ä¡ÍFYÑñõSh'@ÍPâôímgUÅÏìòaÇNÕƒtˆ§]40E'¥ÙüÊAtŽ÷_]@êůq¤Ð5¬=ðVÐ?÷[°M+ý”ù>,ݹ^VƒATÇ‹GúÐôI˜¥Ï-d¦ä( …SŽà”sÐTâ(ûÛ Ê)å}%¦w}Ë1±×+ò$]°ÝþçýúsÓ‚< “‚ú^qp¤Oþšž”üi’c¯.ÿy ÉžëÎÚƒ ðyöI77÷6|.º5ÆôY’+íäøCªO˜¯y"¤ÇÛ*¸JýÛOV†Çõðûtºåv[Íò¯†ó÷7þ7ßÿ(æˆ/˜Õ©)¨;À7/¹·Aõ»œRÁŸš …ëz\ [ÿì3¯eô€âÜî" Îþç½D™º;ÔP Š\y2ôPv'P¯­ Ê{÷¼£U; •‡/ê ß ‡ªnÁ3„MÁPS}ßõž”:Ô}VWjK††¶—:gå¡I‘ jjšß¯Ftÿ(†VeïZ©*}hçÙFêä|ÇþJHã*à'½‰ eT™m‡éx]áŒ*²\ÿvLçE²ÕþÖØ;4ŽœG ìjfy'þö>†ü¶iΚûQOjF†³…™âfÆ9¦QÔ–1üÁ Š-jö±îCɹ?÷© ÌÉŒâ²Â;(—PÔFs˜>Æßž’pA¥¸+ý·¡rè;ó›RQÕçäŽá,j¼"'e‹šFÍÐÛ+ycä¨}”ÝŒÞu=Äퟺƒú7Ø2ûD6Ð0ÛŽ}Ûâ5¯–‹©á,šÉ¿ÒŒº haûàÌR 7Z¾ó£ùôl ­’fÓÆ—{Ѻ$<ÇûÚ ©‹òn¢½dõ³¯oÐÞøÃÚYUtxäýúSÊÑÿt¨zå×sF tŽk  SCÚ¢¡¥Pt½ç¯ ¿æÿÿÐᩘxg|ÈeQ®B¼Šö/[?¼Ì•Aá'‡i:$Ìš&ÉŸ¦ƒ¾TÙýÁ5ЦZA[óEêx.üiÚ‚ªÑu–Då·P^q/ÛùùG(½L'o/:»[ï½ ¾ § z…òE4-äsÍòÏ0œ‡ÜC}kEZ¬Ûòˆä(­=ž”FR/q(ûãûE>X êEY_~sÚ­ëé3ô·" }(†cï¥èÞ¯÷[Ñ=úin{æ ƒlή:ÂåN¥ï kãzÑñR&h¹.‰N¤sí•R˜¼Èÿñ‰÷Y˜Ô\ ¡Ðá…Iy±.Šó0ézÖâÆÑ$˜"«ÿ±þ¦%U&I$xaæð9É}Ã0û¸¥An-æ–o•?¼ó­Ug:ÃBã OÙ1,öÈrþšeÒwÅu!°b]¬³uXVsÄC¯‰ÀºâÓKw6ªa£æÐÁ£Úΰy‘Ûã/ü1Êž–pŸ‡l°¥mšE‚tÞäýHØN4ÞXk†ÄWîò¤ ‰½i-•)’–‘Fq¨Í"¹Í•Ú“l¤$Ó:.⦃{º‰2O¶Dêª>v㺠¤­8CÀFZ‰ôLJÝ.»"ÃÅ«7îGEÆGMv“óÈ”­ò@ ? ™çø¶Å„ ‘UC`°M™Ùb—½ëõj‘ƒæöê†ä|xpõ®ÿ:r³å苪Ežøâé½eòÈ'ó‚%Ž ÿ‘hÑWkÔ(`˜K~ÿ¶ ::Ù^7öG¡çO.ü$ C‘=“ø›)Q4îÏÕj SHÝ·`F†â±w=¯òr DÉÒ`4îc¿³¶ÍV€R·Ù4£éQz<†!Vße9_O¤½„²T)òêA(§äŸ³M‹r™âši=.(obô¥úR*0Wĸ§¡"+ù’ÊE/TúJQBïV‚Ê7Z-zdáKz3ÌŠ¾¨PÃpõ·0j<•xq1wóîÈ .Ô¢iq'êbGm?юǨÓq¶'H¾õNQH©@ŠY^=‚u4¤i z°…FÔ\¶óuhLÖ-i+Ä&„jæ<òôh².¾õTàšŽ.ÒÝ$B³Æ%JëŸ:hþ1L÷…ý´¸4¿y#$-ù=ÿ;‰–7(r?÷ åxWŽÏo´²44彩…V¹:7SWñÎøÕsq&h•é ü´¨b³úfÆaû¢…À(Ž@·yÕ…þDRh©tÞwª÷ÛÍ…>‚R»¾/…Š-Iöس›îúÓÓ’Õ´@ÈQíœù, ™æ)JÅ»\xGl"YÒ¥s·Ÿ}/…t—š¹\?yøÂÝZ«¼›OsCƒ¥áÛžfëÕõPJ§%ÍÐ^Qá‹2‘ŒPCíÛŸP—uelr©ÚüÐH×wîËj+49źɿΜ¶y?[½¶lèy¡7q+ÖC…úËnX˜žè…ANÃa*¦qzÔ2òÈãŒ(|<”ùÆHç‹ß ¿úT«NDûÁD7…sQ L¬šžP„Y’´” +-0Ïh¡@Ý ôK”oþ!x›œK3bï_©ÂÊõí,›ý“°¦ÂÕíùÖ·Üí®- ÃïN9i‹O{`+õ0ééo‡`ûy¡µì9$°jübõ¥ }BMKcè‘(šW!Xv‰{z ̇¯ ©'H¾€dQìÅ>»LÁÖãÇZ`Ž”ŸÒ6è® •ýOa•J#¤Ù›Q(k¹Št }3,}ÖH/ðÃ;BdøÄÝju{ôÞ}FfîcšD³Åù™òR¸?²¾Ì-d[l¡yùˆ9®œ­×ÎD.õfyçW×Ç‚~ŠòB ò=“$Û\:ü³;j¯Žr£À\f°D\; ÖŠ^æèE¡W.z¢(üdÜìò\Q!ꬨ1*etÕ «ð¢ ñUÿå±]ß|›ùZùÍ?÷†ˆDM¾'b½iQ븄Ü×€iÔ~Õâ°hu&¬õ÷ ýBÝÅk”fܾ¨·òúàT“ê¯Zjæz£Áz[ù€%®wæV¯J¢ÑB⹆X4îõ¡HÛRB“lW>—e4 ^i¦dG3½àÍá6_4'}4»y° ÍksU^̰¢Åó¸ýCÐR~`®‰ßCtü]‚Ïðíúîˆê "û¯ã•§Æ`94-¤\–Z4–jÛ' šy#°hÓÊd|·Ü™Å¡Ø®©íÀÀ7øvÈëOg1䓨&µôK@îžã<œÔq-Fþåþv&|uQ }z 2ù"ØÝÎÚA¦2·+ûneÝ´³U45‚üá!Æó#ç äéÝ ö’NPñJIQå¾9Ôdµ—9¶k@»hÄŸTh"oÌû¼:Í’:nLŸB¡Å¼è$?U ´ž±"&›TüÂ_¦‘A+k@_ïuhá(¶Ü.òf‡ÓBo’\ iþÈ ÒðzhÚ’è½ß2ͧÿp‹¿”‡U}ŸÍAh¥RÚã·:µÛß«ÈQ·Yh—/¼ñ(:¢àIÿ±§Ðe`ûiG¡Û-\ìààèéØG¬.;ý .- ¼ï`ðK¥ÃÀQö}ùaËÓFï>ÈûâK¿¦o„]ë‚ɰr·ãt?`ælÞ—ô½ß`Þäi!e@,NRŠÝƒeÂÕ §¾(Xµ0Ý;“Ã냦6݆°™•|̔ö“ý¦uWvýô‡¢ÆÑ$rR š¹_Ä£Ž%½Cw‘40ƒúpé’k^Üsõš/Rêe=ŒÕ›BªWÏËcÏd!­[²J‰Ò?öœ>Œô4Y‰éÍë¢EÁ=È¢mõœ¶Y×¢#šö4 {s ÁuäüÖCµõär·þœ)G>š’$ú³‰ÈŸåÃ)œt~'-ßD!§øBîDWîSÏÉoïAÑ [(5Q̧2Bª„%.íK:}d ÷¥ßI½Dm‹2ÒK]7æ¿£ìö Y—i”Ì8û}¼´yŠûï…\02F•b_ùÒ9-Tj ¹±¨>nKsTé$¢ÏJ®>qÔb:ºöþ}>j?<èäyà)ꪯ:«“…¡>×[…aæ14d¨¬_AcJι>ªg>§ÝÛ)‘G­ç 7Ù~Í`@ ÎEªÁ`‹††¶”6 _ö/ Ò^„Qi%u…_º:ÆI#J0qˆ‹[üLåKØíY£K_k9˜aþ¤[‰îK"XøÀ|-0Çb^E Á²>ßAâˆXéäÓÆß k-c,nð[e]Ôü!«+|â‹ _"M!a©p·èçE$~HÙ6-Y…¤øõ=¬ yÀÁF¥¹#H™¾tú^hRïÓŽÞ9…t”i¦¤ú È Ñ'G¸ó™(Êš§æ…qà [°9²™Í×–@ŽNñ°jšxä.<û¢é;ò½õÉo ªØ…9½Q(3HÕäŠÌåcäµ O=úÙà %ô<àéCVÛ}R¯íPÏoC]T¸!“݆J7\®Hæ½Bå#ÉÛ$M¨êdúþªQjÄï“§d.AÍœ×d‡SQûzÙ½rŽû¨{/òøL_9êg¯þ4ôDíãÚá9ühb°¾L6߈fwy/”u+¡ÅGá÷›–>hYù{ý}ZÕFg¾P|‹Ö?šº¢mVö}¨øL‰v<.£*çtÑÞ}\~£ÃS¾sŒ–ÿéP…Á\Ú;c„ÔÙf¡ u³Ì$Ñ…]>{cBÜõÿÐaÏOyn6|¨-~ÜÔ·í.Ó½º4óûdâùæý`g5ŸûÑÕð³)eˆÅ¡•ÔÏSÛÝjŸ¿¾»F>•EñKÙ¡¬ÑIûµ?”‘6¯B1ÝÂA(ø®&S³ró¦².»AöWÝw77F!{žG¼Ãý6„¦R(™lïúëÚ<¦‡P{«×øk4ßH¸ó°ðŸóósî €n›‚ òýIÐ×(ú8)fcÅ(yS`ÄÈ;Êã`?üR  nï…‰ÐÃéiaÔ0™Gicù”&ÇlIßï¶‘TEË’û`ò<¯›ßµË0yW{2„Ŧöê¬ýi3¹«31‰0sÓçsŽ‹Ì¦ºë?Z ó¤SMNoa¾Wê0ͧXèzÂ娋#6Z3Ïüa™fé6RXqÚûyxW¯«¥m®Ñ%°®µ¯UþÜ:ltžç§^›…Í »éãðÇì´^ãð&ììŽ dB‚xÊΰU$üöX´"­‰FrN_kªGQCËkç‘4Dj² Oɩ؎й¶"Ezšƒ«ª=î¹ÂêB;‡ÔÎÔ¤cÅÞ»þ6aäÌSŒô¼IsTN#ƒ ÙÛÖ¹(d”‹=ïk²ësVïò·ÄëùöSÆ$‰udi%r˜fßåC¥PûýµÈþÑÆßýíUä”á ð¾+„\ߎÓp²òX9^q5FÞÁJMmzäÇã7ï €XÕòÅY(¨ÙÉ‘¿Bá,²!¿P„™!û-ŠpÚg”\A1k¸ÏÅ¢x[š/=7JÒ–×5Å…à>ÓÑÒ«Û](•sïzÁ‡u”á8¦¦vee$T Îl¡l±§£äª7Êà6.ƒrŸœÆÉ9šP^ÖÐêÁåF”Ÿåì_õòC…‰œ×¬7SPé±ý>/.T>ÌêNxÀ*,òá4ªi®uщ¢†ôy3*ÓDÉ_Ž5^Ψ™"ª5«ŒZ£a'9ÂüQG?øMqê–î]s›D}øL¦¡v hD¥üÑ Mu©jq Ÿ†9öý|‹FfŸõèj¢1¡pCPÌ4Nê~F\‚&B&©n«hòK’ôi”š^ôcŽâ¤@Óí9Euq4»É›Qr1ÍÉøî=r<Šæ¡‰šf\õh±Çaçžc†zu®…è¢åJ°k¢ë+ä—yç@é ›¯ÔïRÊA S¾ÇBøÞ63¯àù *OÙ»ª…ÉCñïx倕V(|pšayô>ä¾;Éÿô]$dq\e]ü¶Ëµ„¬å7 ]p‘©[}Ò$'uËô@šÎ9 ûÃÌÞzBñ"ÓQøÚ2®Ae^yÓ#?¶œ¡8wòµ‡ñ§‘ÏISÏÃdø_®D˜~á×'"ýf3õ¨ÞªZüÏAú–þXx¥Tï§ ‹‹e_I¿2Âò©ì¼öeZX¥ùÅ:}ö¬eo|8  þa¿ÎÞ.€MÅýR¯uâ`k‡ðƒáNlÏiÌÜ@‚›óá6 ¾jþ=õH‰o>h›Gª„uOQ$=ÄÎR£ˆdÜÃH¡;|KœF)GÊÜϺ"U ù™í÷Hó¤:=ó^%ÒEØ="ëdGúñ¸š=Zϑ׬¨¼f™< öÜTJBæx™qÑ´dä;’¹l˜wDþG5²–ƘD‰œo.²8ji wmŽ©Ì³(䓈½<:¢‡üeQô_H©Q 2X)] Q0ºŒ vì y}W4bŒÂvi:©­(bÏ uú-Š^ÿÍÅÍŒ{Ó¯ÝõQE±ÅªëïfQÂ@¢ÀÒ«%?®÷4ÜàF)^Pÿ)Œ(¼óêû]݆Œ¼z‚r3Î3Ã"Q!â•*•È“Zj£ò—Ïm6N—QUê£U†Þ1Toÿ´gë(¢÷IÆ£‚¨ÙUâlËüµ†•#³,Qû÷±?M¤b¨S¡Ó£+º¡¢F•)POrÕléêmJMJ ~úØÃ4pTUÊFƒ“Þzähah5dĆF¼…t,hôÆ'Äš ùz-õĶÐ8ÎÐTŸ. Mx.ïŒØB“—¿Õ¸Ú=0Dù;ÅŽö;´95­ôLÓÙzÃE˜âaé'ýuIÿ¦ÝúÖ¢€Ù]=qd®ÙCiŒÅÎŽ/e^gya7¶ÑiDß„<Ù;Á-4{!çà„)éǨ]þ³øÞ™IùvæªvðÅä;Cˆõßök$)ÉÞ¦ÃÏÕuŠïz ”¨é==Åÿ*\ýÕo’@MÐÏ¡µ ¨_n#µLßߥ¨…ø^B3_ÀmI»hÑ›¤ ®†V7±:Y§ÿø°dV–î sÔÕ\ꋺè²gè×wt½û…î½2ºïÙè}4â÷?¹´ ´êi+‘K©ètïúªÕŸ@m·tdÔ`Ì)¼ 0wn’êˆy*|'Kº¼#¬÷W‹×ú#SwýOO¹¯cý¯ž2:$ŸÉ~H…´×8KC!ÅZ‰eìå$|~tYoÁ§Ü%—ŸSÛÄù ­Ù Hö\º{üÔ0¤ÝÝN:r÷Ñß~þí7ûôË®Zóýѱ™‘¿ãææõûVyïešÉ´Bþã {ÂS¾ç…ç·Þ–ãµEÕP­›}‡b‹©8i·e(!8Ý£°¥¢o¹Tí¡ìN:éôÀ!(¯¸£¹›«+ú/j yC5ÿWÑ©óP³ØDøê)M S ñÑ {áJÒ_=·ˆù»þò#û;´w´QÎ7†.²ëôOç/þ“¯æ™ž@ÏFPëÉî.è/˜èªˆ¡AŸËÏèÞÀ0¥ãÎ÷xBiy‘Ä"² cÕš‘çS`ügt†If0L©ü©}òÙfJ—~Þ\¤‚y>¥Ü³#ç`ÁÜ„2îR,^»qèú©O°T{Æ]ýÌÓÖS¸‰.Àš¢ÔÓ@ž{°~ÿ©½°ef-e«;ǧMÇÐíœK벋\×zñí+’’)®/Ó 9¥XSä®ÈÌ” 4EªÄšQ×÷H{‰èâs1¤ïذ·X@ÆDªŽ+ñÈüžØìªO!²6ý:Xxìr=/ª”ÞƒÜBí9!ßáÕ"WbQà ÑÁ N6²;’ꮊ"÷Ž,ÞýÍŠb{¹oø§\DI©|?VÃ[(]¼òâE°ʹWþ’ñA…Ó7¸ë£’‡fË¥¹¨ 9É·£PÕ`͘ 5>Y¹þ~ø 5RIÞÚ£öËÓµwóQ7sM.¾…õç:ócZÐHë²ÿk4‰’*jy“‚fý±‰³’Rh±™×GLòíÿ7>fÙ|Óg$h:é”`&á8SÈ}W˜­ŸÙý»¨þúÔüXÍ©ka°0$Ã>ýq§/z |"†efß®‡7aåðC¦º'U°Zái2ÀºY¥tÏ´lŒÞûšô×ÇþX ¿:œ vö]íI$xþ‘Õ+ ßó-xq‰ŠW­~ßœ@âUN²'sÑHjd¯î~ã ’•*ç KÌ"… ŸÁ oÜÃ)Aš~âRÍX þJ‹4µGž¼RÊEºÄ–Ù?Hßð¸‚S·ZåYÃ>3#ã4íDÇ‹Idæ&Û++ÿY¼†‰7˜‘µ[ë%9²ëœY¿Ä‰½D&QÈu„¥¦!¹ÿÌS<'?‚¼OW|Ï,D~.E[Ïú% èt«'FAq›¾õÈ) ^Ë ¡ÖCžû…Žñ èW'òPì±É’A¤"JPu^¾ý€%õn½+z»þ×¥fHÆU¾ Œ±ôÑí÷QVçÅHsÊ6–‰¤Ù œzwʹK”‹ èâ ƒò,ús¶ä(_u¦ÈÒW¾0]~´‚J§Þ½(ê8ÊZvëì,Ûx@ò».Õ»7¨F*¼¬(,Žê®gn‘ú"ĶîM9ðÏ9ÃW„~dL£Ö©×©‡nV¡vVO´¹5ê²é|¥× D=¿ã³J~¢>-CmMè êÇ8=R]aB~ÕâhÍ¥¾c{øâ|Ù‚Ñð,å±ïÔ×Ñð»Íª‘Oh$RWù°‡üÖ¤Ò“ýÐèý©Ÿ üh´ýΖîY«Þ¯Ÿ¿€Æ¾eã.r¬hüVl‡ÉÊC9í5œNk åþäÈâ?AÈø¹7ÎhÖ«ZöµB{®;Áè‹-h°{íwÊ“§=ïÜkùë/ÿúOöˆöuÙç™shúäèÍ1„4æCÛµõáýñßö_ÿËTf]ˆ(0ÿëwÿêø_}þ›S«·[ÙöSþÍ«u×mļ Ñžæ-c¬è?çfðÞÞãÍê3¡Kç¡…Š¬ø6+´üúÌšýZ‡Ä³xfä¡â“{Åyzèðæ8xÀϺ„–c.o$ýÍ™?[‰^¾†>~ Å'Ñ0pögú!š]ýïL^îtÏ‚á<[‰ÛiŒ0ê׿œ ¿Ì…R_Ä¢ôÓÆ|9˜’8@S=tfd”üûœaNÑÝoÔï*ÌçŽÔë5‡…uײóãS°äì[ä˜.ËÖpVÏ?þÒèë#¯Ÿzœ‚\*»æš>Ø<®Gy»qîŸsˆ/þØk ; ?Ø{ áv©íÜ$,–à;Wé‰Dcr¯! ÛI‰HzÎÅZr(ÉXé´üAŠóá¡Üçpâ!}ŠvHÍ”qÑæ€*Ò2$7U±K#½a}}"2¸Ûõ¥ó"ãG±67Ý·È4]y°mYd¹"D 5ümN­ÐdÎ}: à‚œdœë¶C[ß-Ùîësä-¦›þÞÎü‘ò£• ›(k1¡<ü }È»K¿\D!¹"»·(Ìî¯RšŽ"LWI¿ªÙ£¨œûÇ3™qïé‡[k±{P,ûó´p]J®V'£¤‹Åq>)KÜ×Ô'ýSX¥M·Õ(¢¬O äù4c”ë´”ëœG…Kyõód|¨dù)rÌË•ôȤ7jãÿ½»ŠªjïþŽ ÝH—¤„48'ÝÝ ( ` * (¢‚Šb ˆR"Ý ÒÒ-!-¡À»}/ŽWçöùßœ«yÁkï±Çø®9?̽~;ç ‰äRª+\°?ráY¬åÃs§QÃÈáöb¢jм=Ez" µ¨}3OŒŸD­TéÓžc¨Í™¨9ôµÃ¾$[K½C’zßÔñ$»k"f:5;ÞVé£.ÿÉz͇]¨{Nî¶ÇÛ:ÔÍ»cÊG˜su7íJ¤ÔQOU,5dõÎu^–Ó@½¶Ç&ËÅP¯E.üùIE l« È$QG«Ù‰ë¤W«]i—ëª,™Ë§¥SšC³¨ãQ¨¬uôy$ï eÇœ,{OsrØÖB·Í<é©·ÕÏCî÷¹•Èáe“+~D YW?|ù2<„—E«@Úlú¢é„¹Sr˜ SeR‡_»òf¾:óÝ€R‡TÑɪ‹P‘ÒçYáª5”1§ï@ýG­ÏlÐtPC¼Sê$4_Íʲ¡ý-QòñvòUóNÚÄâ´œúØåyÚøv¾\> ­fsåMŸë EƒRšc×whöØsÖ§V šFß0{ÚfCÓ·ÊØ[·u¡ÙìEŒ´°5\hhZÿ\IîK¸îkÙðïÐöCTu„ê.t˜|Ô»#çß)çØje û—ýúÉõÐë±Ç‚/Ç$//Ví„!ÍœƒÑ&50"8Ú>ýVÆd‰Ž„ÿõ!·žÏj9Ìl]˜Ó*†y:ÅÝ5‰†°P21ÿý?,åž´•<½V¶”öð³ÁÚÃ;¾Z½á—s´HTÄWØ2Èd=ЂDqC|ž#LH,ê8zyÁ‡éTQ·d§Ôx±ÑP@Éiu,n %ÓÙϳ‘EH}ê×Ѻ’8¤c ‰±óý‚ ~:B/Dø‘q½‰3-¾™ïO¦´…>@Vé{ß-æ ÛGE}VÁ‡>ää"gv©jáT6r5^³uä%ëÒÒóªB¾1Õ&IäŸëæ6P‚‚†K»Ø‹‚P¨*Û5Š{E.D¯xïÛƒ¢–ý‹‹Š§PÜÉ},m#%Ÿ îLCiú½jWI³PfªüÚé<”ó |Àµwå-)dÊÝdQÑ3O}"•ß2qÖdïCÕŒ©ßS¯Q=1Íæ\ý,âCPyÔ˜×U㟹ˆZ§J\tº›Pgå‰ujÔ]ª7 åžGý‘:›“Í4hØÍÂæ-.‚ƽö“*î\h:¼ÎcI¼Í[BëÌäCТlÃ`úZÝQ¾eû­ÊÍäâ)ŠÑºíCSüÖ´™»éÝNøœìXÚ±‰M ½Ž§üæ:\§%æÔü‰ÊÆØ@G–,×4SÑéÌ“×L÷ s‡1ËA¶A<¬_è­_Þ.nšFÑ6è*bWÁJ‡¢3²‡bÐ(Š®ëáÇ)dìlɽW?žEð[¨ÛAÍâ­‚2ûüÓ:PôÒúL£äLìâõ:ó2u?• º˜AÚëƒF™EzRÞø³Oó0|^Þ­ªü >˜]T‡øó.Õl%ïØ“uâž)$Š(:Ëà ŸÍ½¨*Ç£ %*ðzÓoHgØŒ¼RÁ™?S{”Îú¬v#ƒÁ…*¢æ‘ÛÌõ.Wøäí":ø8®òÖ ž±oü:;€‚O£«úiDPÈ2``E;xÍSÛ’¡ØÎIyÿM(±­ïâ-_…R2Kþ’óe}x£áe”l<õׇœ-^/w0ý»FxLÌ¡EÖ4ýí»ph£ðVÓ[è‡öÝjÐEsÄ5UÍzì’âU¯@ï¶GÍ­, |͸÷2­†Îw¥ØÌ†œ|CÞ“0ÚuÍŠ˜OÆ«s‡#Þ²ÂdWLŸ25 Lï=Ø2³RïλÍÁ<{suؽ-X0z›)`‹›Üia©Õ¢g.ŒcÍ›üܰ EÞFòÀú}ÆÆòJð[98Ï´x¶Uß ¨fÄêVïÂÜ‘xý±’SîlÚëžëúÉêìÇžY ÅÏ– ¶Jv¤v=SP)„tÂ1Œ³w!^ lÏã(dºx“>íÖ6²œ ”ý&7ˆlÁ½Â<~ȱñ3Iåa"înpµ~ø5!·z?ξ[D~ ±g÷νCA¡¶áí$¶‰š}på=î ªe°û‰âÝ×XúÈ~â^["…¬Ÿ(ËýpÅ›0‡òâg¨@miy¹nÇÎ'«ßøŽ*·Ú ’•p??Cpýv5jØœ8X!AŠZÊê1ÞMoQGí©ÍYUÔs9¹@¯‰‘2Wr%Ðhò‡Mš >vÍ¥@s¥ÅåZn´ró=UZ€–·n˜´ŠŽ¡UÐéÉûÁhýܸÏÁ?mRNÙœjCÛ6©‹¹¾¢=ñ‘CÚ˺Š<ÜXèê\Én²z¥ˆ„GÐümªrçø~¨fÿ®F¿óT$¬?-©™²áÙW¿~„C)Í>O~ówPÔ6ã/,ùou/Ÿ¬€÷Ï ¤¬³µë4cædÚ?5/uipµù¡¿:VþЄêÍKGž˜@³³…}€;Óßý¡•‹]µ p~ý®ÿB†Z“JnÖ²À軟®mIC0‘Ó”oäWSÒÒ}=&_`úTJ9ÿžN˜¾ŸÑÑjt¦ºjåÞÚÁÔw ßöªêôõQ5¡i˜Iж f/‚ÙÑÅL+×0G»ÊM5Øóû$%Þ×Ã5©7óX$Ÿì-×vƒ%V±¨Ï³° ÑÇNOÁÊCÍ^#àçÆ1qÓþ“°vuõžØD'l°ÿ&ÿî¿JlN[yÖÀæ]gyÏ»§aÛ=«Ö³\‰2…{–ŸãŽÊ€û$t2H<òJélîäÔ¶DÒ3dn‘å H6™* tW )nÑ{ëÊTZªÄgN# û%Ž–Õ6¤] •ù$’ŠôÝbg9¹ëa\%a‰ë(2Î> ™jú³g—!Úp÷FUö„ï¼—‘5h"³O1Ùæ.W÷Ô‹!ÇqOW;jÜõÓ¦‚¿Ä w_þV£r ¹÷˜Ý¬…ÈS¨*õÞ_ù¤k8c›ºŸûSIKu; ¨Ò?[{Š‚/GŽØoÉ °B–™ïN-YzõóSò8Š&z±WÅ÷аôq£„ é­p/ ”|¥­×{w*1ÿ˜Ci—/üã—¬PF¦'@@eúIÎX©&£¬EóæÊAM”ÍžxÝòâÊí h(:äŒr›LÖ6p_Ÿè¡ìþ`T¸3uì—S *Yo I»žD•'÷~0 ÓÈÉàûP}ð½g”3-BÖmf&ÔpZ.é¥GÍ+ì_]”–Q«ô2[ÜõhÔ᲎;Qÿg>¼»I¢¾t-¸â·»·Û–b¹S`]<ü"5ÖüçÃÿ|øŸÿ>€ ÞG[i¸!-ôù†ÿ Hs nXiDÈøV{›3û äæ˜ò>VR®¢£§õ¡’Dë5•ß ¨yôñ~.O!g”þ¦„÷q¢ñM-9¡/ÏO\©x-¿¯kù¥\6Ñ>…ú³? -K¹ô{ '´ù[Ïæh5Ìéµb–'WEs‡b 9±Å±Lã+4ó^›{\L¸m{—Á›UWõ¹Õ Ï]´€V ͲFçãÐ:»ÈyDLÚéBHIBsþäviå'ÑŸ=Ä'ÇЯÎ `"O†Þ[ÛZ4Ñ0`2.xç{$ Y΋*§„éTßÌ^Sü¾+‚Òã¯9oêEJûÁÌÙaæëÃ0O½u»cš Žm—‡¥äÚ|ƒ X™6nÞ¶i‡µSSwS‡#à×Þ‰·1ë5°ÅªìùD‡ ‰‚+‡Ê;˜‘˜ÎbV!ŽàÃHOâ[wö#©öaÒ¸Ð>‚£'mŸ"%M¦ú§p¤ö~Yu&…験ºª§‘áRËͽ´™È¸ÂyãCæ*2~îW @V1ãsšbÕÈ6œ¿nÜ‚…n^ÒGÎü%ç“ ÈÕåÝUõî'ò2‰§Š¼rF¾ÌÎ}taÕÈ¿,;(®KŽ‚¦´vŸ…$P¨¦íÅD#ŠøYœ“8‰¢Nr‰sÇ#QüT.Wÿy.”L¸Rjáz¥mUK(£Ì|E´å¡ï]ºg°Ò…òÚÎ—Ýæ.¡¢ÅIÏד_QùBq¶kÌ^T½#Û|¨Õý]G¾E˜ !!õ@vÑØª85Ô²Èê½Y‡Ú+^·5…§Q·ôlé{ZÔüñÊÔà ƒÖäŸM¡q ñ±ä¦(4½§ûcnÍý=·ÎS¢Åk郎¸hù´"§T1 ­ÂïD…¾¿ƒÖQ’~zÉh“qa¤q^m[{†züÑnI”A:ð9:ðüb°¾q Xó¿ º²‚C3ÎÌÓr£cÛ†ý`Í,: Œì²äÂC¾wżJªñp·àµ” %t‰\Øén„ç„"y†ƒÑpžO®kF}’f2^éÃlŽÒ~±M¨÷q÷(Üm¥«?øôq@OF•Þœ2ämœË§èŒ%«k<ój„|Ý#Rmë%ô¹m½ˆ%ø¾ãÐ_CøDbâ4M±ñ’ òA÷˜!n³Œ^*F 8h~½3½ Ÿ•ƈ> )¹4Ö&oNüÙóú:|²l´?xwrLg*”„ܼ¡©ë!‚ wœµºPy—ø÷†5@¾Ñ÷9ßëÍP@.!â4j=+–ÕW Ð"böØ Rzñ=нF×eÒ¡Ä1…‰QlãÏ}L÷ä¿>ôŸ‘JÇ(ï# ȘŽüëC–¹ ½ö÷aò›ÕD%hQ¤°®†6Fý™Ê'çÿœë!ò¨Qº˜ôÓîh@ÏÈÕ@ÝcÐGj\öƒa íºªWõ+òWeþØðgïïSöªFû®Îq†Àxí¾O1•Wa²=BÈ5¦…,¦¹¼ïüÙ¯¸ºµ¶þÙ«tgJÂ޸nm=,ÞòÔѳò†¥^9Û÷Yù°âuÜC+`Õ`&·P Öï…‹³pëÃoÁ/:Ä·a›ÖôÜãu¸ƒÓDÿ„Ü!$®IìøÕƒ;_‰Œù6 "YèÝìt”H‘¾°ë7íRs°>•ê:‰´m†‰ñWÇÁßDj{Í ™4Öî?<ˆ,’ôlªëÄȦÓ4í‘ r$Ú½Iœ5#̹_›ßîý†<ŸlXò¿è#?£Eóãj!XÊd‹àGaîÉm)ܵ»} Å52£+N ÔwjEãéj”é±ý½IŒr¿ð Z6Êg¸6ù&p¢âƒ½)KÇQÅ>ø&ËTјuf5HSF.d f;ÕÍC×®£v™z÷N:oÔmÚt›zŽú[í•¢ÑH9éûÇ 49v3¢]oÍ.ˆûœ~|-D´ésVGÐ’'È¥ï(Zñ|å ãDk¿°èi´Q4ˆY•,E[³ÛT¢Mˆv§Wy&óºý – Þ¶èP]eÕL̃wrÜh”Ú‹ŽúúÇÕÛ´Ñ)L§fªÚ'ÅŠXð°ž<—é t±Ëîø¹7õß}X©{“¬~7tï>5Tùš'RÌŠŠ Z0©Kªf/TD9ø¾î‡²iÇ¢ŽB)ûê¹À~(Liøü~òã)Ÿ’ß9çÎñì¼¥Y²m‚<ÉÿÔ¼Âìùàó;¡Ì _Mí…’ƒŠ«£Ðì¢E·Ðbõ÷ùÛCv­0 v‡‘ê {ù¹› Ãh&»½-ó9˜èù²î>: SŽWÏ×$:Àtä‘Âäî0]2lb~¦íîsîñ†ižS§ù¤ü­9dì.f03æôDŠO~]³çQ¼sæD×.Ï=…yÛ©wº¸aAZ“–&þ4,J½|Ý×Kû{{üšº`ùxûÚ9 XÉYþP(VEJ~d=ƒµ$ÑŸ‘@Ø0:-¦«¿–¤(7Àfá³ê#±°ýn!ï¥ón$jý{7àîJ5 $B2w¾‹;áT”âÌ$}š4DIäôâ÷î„#ŧXÝ莤:i¸­zÆi42@Põ%Òñå5d³Ç#IÚ÷PSdØzT»´ŽLäëäF?U‘YÈ’ƒGî²8ˆí}Oc€¬q3o³.¨!;c»‹‘1r<Ìh«‘ÿ[Þ]8»C¢JŽŽS!·Eâ&ÑÅä¿6Õþ&ùtN¶!¿\@¤nU, Õš,É©£`J#“Øün6ÛçÑ} ÷І;{}EÑ*ÿÆÉrk×ÒH§‘äB ¿;G¼”î¢dº6QYû Ü»û~žBéKöýK»8PFÞ)q·÷"ÊÌêö^« BÙ£Õ¦—†eQ¶Y+N¿{å,ó’D52pE°ûË…¸ï÷ÄYIÂü)«¬8ÂŽJ,ô;QEßÖ™Bù:ªñE ‰ª/Ž+ÊH@h”ߺ¸J‹žgœû/£fpÈБ~Ôªs¢ÏJYAþ#¯¿Ølÿßù°î{5ïåûhAã.”u¹%ŽF$/œ†õ‡ž|ÿÌúŸÿóá>ü_ø®ô,ßí³h9ÛÏ8¶–¬Sûl9ãwÁ¢ÒÕ]eâ—™äg›šAÅj€Œ^R”nz<3> EֆݕŽïæý“1ÀrJ¾*f…ìø³DûµAfK×ÉPé@Hïú%Ãz‘Òd¯œþÉ~ŒÏÉÏŸ(@ÆÕ¤rñNÈ ‰ÛÏûrJ¬Y9¡Ò6¶Æ<¬j9¨Î‘òÔH–Wè>xš¹Â.¸&̯>…Vûêo'ôm¡í.—ºÙcehg¼Ó·AðbGİa$´nG'Nüt„–ÑÇø£Ð¼tUŸ6ËšºÖæˆ*eÜØ4íÐrBc)´Ì Z÷ÞëìJ·#Ì—†$ÒÞG¡}ïR£Ëwè¸åžGÈíʱH›ÃУ¨'þrå ôFq$qÙ„/k› 5rÛ³Ò°k FÔ®æ“7܆1õÆŽi¥¿>dÑ«>& 3+ÏF÷GÀHNGrCí«RrÇÜÏ:ƒÔ—•Å™N£ãh¤VdB†‹._VŒqþèdx;2_?O}}ì² Ò¬ËQA¶þÃ’C‡º£`ë‚aýä,Ð;‰\CgƒT†’‘—û*Yå"òåJ˜ü’Dþµà蛦(h±ç©Ö½ª_ô;›š"×Ï,P'¡¨Ç™Wm(~•ƆrúJïrôBi¥+ª¤%ù(ó3 …ΡÜŇ4c(¯V²ãJ!**p¾²Øs•M¼¬×ž‰¢ªM GÎET7bhh¼$‡Û·îSñ 5òhX¼Ž£–üg‡’âÔnn~wOãꆲ(gÚˆ£þ!†Aé_ãh¨ö±{uGï… ua4•¾,û ïš Ȥ¥ÞF‹óÓ,7²)ÐÒ“Ty›½ ­Ž3ø¨› µW†àȃßhã{̦Ƣmƒ[ì?ßg@»¨6CqZm´/úÎ󠔆úï%˜ÇƒTMµ®]Áè¨J¦«m†Ngüy „dÑù“Å`XAš-?Ω]….»ïk™±b€Ÿ Ç4DÙ‡B'02_%KN¨…™o”¦âì .I³‹­J?Ê.ÊæÍŒ&5z:ÈÿVNU ‘ëjœ¢&œY<í )æ{Ãݎྉ›“ÇÖ ¡ ׳ýMÄ«El°{@\Dü±…/´ßì§\w¥>ó¹ÐvqAJS˜H©p;¤ß‰ÊfÙƒ¬4WZ$ !ôɰG < O&8qŽlònŽkN‰‚¼v-šï×!ÿ­ÞÖ­Ø&(Ð"¢ Ž(‚B*²*ÇÁu(¼ž²lî;Eú¹¼×·®Bq Wž‹Ö”¸>1!Ù‚R&ÖÄW£¿>ô¡, ?ã åÜ›g0ö¯éi6–¹^ü»ÓBT›ix¡EkÍ™r)Ú8¹ßéí£€ö;ù¡[\ÐÅ!I2Ý,=Naºý–ÐG-“'pn §5]`èæsö–ù1á½,§Î¦££Æo{Y¾ÁxÝç ŸçFa²¥xCŽ*LïÚðwµ†Ùð(!+lB~)¿w´³Ã‚ö¾‘áK3°xÇ%G¬¼–Ƙ›l{Ìaů ¸ü¬Z›˜ô~Ô…õ ÎÛGžÙÁoº_õYÏ…ak`‹dÌî­æªïkv@â·2>¯fpçQ^wUõ$$³Êxñu)ÎÑ› Ù&"UC¾ç³¦H{sV?NL/¾ì:LŒL4gÈFÃ3yŽeãU±²nUä÷|‚ÎY÷ùLæq· Œùó-.ä±pU#›xƒ|×i_ñ}B²±sFð …&¸GßÈÁ=‡]]0v ÅeIìNÞú†RWÙHcøÞ ÌsÞ¶»#õ(µh¯øå½Íý¢Ü÷´ó.T^zú¾¡¨Î2|aPñHÉ• Ô<×uÈHÒ µ ÓPWóicÖc?Ôw~÷vE[ ƒÎ\ éFãO9 Ú Ð4ÅMëƒ}?šé›P9±¡EîÏ‹?œ˜ÉoBTþ ­2ã çvæ£uÐz Äg´ùÚóîù¯,´ý<×DÐn9ôZ¨¸:0wòwðÆ*ã2AÑxÐÝ·¯Þ._ÐÆ%}ü†NÍ:ÅÁ¤xˆÑÙ&\;œü͹¦ñ¯>Ü:ñRBáÃuè–¡¹ÿ-<žÐO.w¼x®ÕÒ.š„öCEƒ!—Ü©&(Û¸$¯W%¥‚A3˽P4sQ0/ž×úôî§!çª9É|8dɬ„?²ùüOÍ«¾BÄÍòÊÎóè?è€ÊýÌTéW¡ùȾÊ#øäïù¥÷sL-´Üa@Y[¹þú: -HÊë·*Àh­²Iûe"˜ø­tܦîÖRæSÏÀt[6ð-ÃôOúÒ6. ˜Þ¢35¦­_¸ªi×9õëB`hj²ËËö˜VFŽ´­»û˜g‘SWǨî"îþp,ë‘·"rûQ_RH,D^ú/qEûÏa%V¬ëòëÛ^Ï©ðEÇ»$ó×FQ°*¯ä×0 {¤\Òz€{ö<.—~uE‡ÃŠTº¡¸óEóFN”xF®­³£%ë´òbMq¯ÜkOGç9”~8óz;¯e$¯f}I@™ßA†ŸÌ=PörÉ›“(;õ6£p·2ÊY~øD ÷‰±çQ ”«¯øÿ·1ûBwl¦¢y³nøgáEäÞQJ”S!ëó\)_9]ÿóá>üχÿÞþT0Ñ¡K‰–^ÝŽvŠ2Ȫ]âνc >E™æÕAã‰9PñðÉCþéJ(ýzMò\öu‹mBþN«ÚË&ff´á٧ߌÏêíÿóÜÒtèÆUHâÞ`ß7©­.#g?{AêE½î†Áûd'í;kÈ;üBŠ÷}<”¦,}wÌŸ…Ê8šõ§‰[PqÆÎL³6ÊtAóÕÂ]¢dÐòÂݰAá>´¦³\°ï—†¶1G­ÆÔ›Ð~ªmƒ:÷!´ËÎÞ¹ Õmö›‘K¯ùSrß½yŽŸàJQ¹#ÐüL'¶ÿq 4Ÿû±†ÆýÐüm–²QI ZžøÉÍ>¹­6Í3ÔºÐ&1ý+? ÚͼÔyÅ IÌÚ¯Z@—¦§àÇ*ôXjË–ø@oÑcçÛ™v0ð°$ðÞÞNò>üåH€ŒèI=›‚1­¬§´¶ä}ÈòJoñÝÌ,ˆAÆ/'˜'µ›; ™MvF˰ô&än@Ã9XùbNž\`kiBçÄa£ùìýÔϰu¤ëG™=9êS@ðaêVJ%Á‡þå·÷Ž!©œ’÷1O:‚ìz®š‰”‚:_zJõ‘újǃÌZ¤Û5·?Y¼çöèíGÆédó edöUVnt Y¹nð¿‰G¶bO¹‚Oyp9Ð"g¹}4åðäš:­”(t yECÆ—‚¯tŸà}·Päÿ=â>ù­5›;ç¼Q¨™¬Ë?Eã§\Z@ѳ=|ßë)Qüááû G™Q²5s%¦w¥£UŒxöú`xÏZ«)ÊTot+åE)È#©HPqWMõNvT✽-€ªBV¯#î[ :G õ–qÂ-Ñ£Sw,QãÝÜ fÞjÔâˆnÛJs@í$…œfGFÔuÊNþzÆõwïazl…3?òs½¼Ñ¨SÎãMlš´gç‹×Ú£Y½Ûè<Z˜ñ3©¸–š:¹êk[h¥ºzÝ©ƒ­UÕêl ÐT”§ú¢­nS›¯Ú´³¸)ôtíuJV½A‡³¯ýͧñ@à¬HkôO<Á$÷±ò:0 ¢x:õq>ÏÍIÀCÄî–·"íð°¤1‘±l^ÏR¦_j‹Eƒ­‰3—Ÿ#CŠ1_B>ažÛ,:ÂNj µã<ÖþL&PjÕ‚³y¥o4´²w²ë ÙìT Ã,b•ÐÿÆ=u‰) ýÏy\¨p »ëcÁ.|DüJdþ¿ ©=?æBœæ£’ÌYˆž¸¸‘( Ÿ)ºCÆ,z er׆B²&¤—ž/så6l†ëy §ÞB®®ÚÉi‘ó³ï äþù^x-s ÓmÈßç6Îz¢ òç>Ð>lüÿ}ùgŽ£ª°'»¯ù@aRé/%¦,(ò$#?Æ|Šßüôm} %Ç..-ïø³gY4w ýëó{l}#„ ¼YÍO,ìó_RŠkF%8þ»3¯'œ"‡ãúÔhX2¯ ¼Nû†¤¶œ tñ0ØRrQCÏáÀ§œ—Ô¡‘55ãÃ~˜ayú*C÷~?Qû#"³™?Â蔜v5½(Œ7ìÙŸ¹Ë&k7éê`šþuŒJ9ÌÞí™ÐÌ| ó;>®¯ùEÃÞ¸±.‹÷ì$i•sóìŠ1Åæ¬1$»_e…UçO¿÷tÁúwýé´ðkCÞÞ“¸¶Ò’ÂU>i"Q»ˆÌÖYK$>3À¦²…;÷²n$¨ ë˜&ç­v¤Ø+u½òí%¤ tÕÚ—FZÙGÃïÞ#oiJE2vŽÝ¼IÌiœýzÈš“wò–ƒžÍ7,ñ rÖ?¯Rï„Ü+§]­¨/=ÂðýeYxÂP|µŸ…RûŸîÔÁ=\õ —Q¬·bUVË¥ùÚ‰lÏ Œõd9SÍ ”“õþh#'Œòd7ü:¢B÷‡ WÅFT¾)²LÓŸjAftî¥2y ‚¨I£ªÐ£ƒZ7Ž £NšX1•²êØ&…™Å A— ³¯õ~4š;ö´Àí9šÌ¼¢ëá4+§¸üþ«ZØÊR]ÑsCKc?ß“Cjh¥ÓE¤j¨ÖZ;L˜Å_¡¶ì>ò=’hkÀñLâÚYRÙŸÖ@{çñùošêèp:Óèñ(¸•híÜ-„ßtÜñ¾y ó—ïÒD§Þ; >ôã!#Ã0žõá6ÍLÛQÿaèÖITX ÍÛØ?|¦ªÕT™±präòªP¸« ¤ Ÿ×}¯“½÷ßIü°Äcœ©•7[ûûcÎÔ·. ðu©ç¾C©¸Gó¡øBhŠn~J¼7Å/ÇÏ<©Ù…ÉA‚œuì(9¦uá~yî5¬Îøl¥?Ü8@ã2B¿Í­XŽ ,uMVgwÊ>Èž¹²ÈŒr$k¥rNÁ(²÷ÔoRWÜgø‘z8LåÕÝŸî¨2@…–¢®Ïs•¨t1î¢v2ª¸}óÙIȽš:O}ˆ¨5îg?}xýÈ"QxûþŒë¨tôùÏ&jÆ^eö´ñG­a¡­Ì/PGâÖM€ÿ;^¿j3z³ Í߆þäêâv-}1 V 1w?ÿçÃÿ|øŸÿ>4È9ü€ô=Z̶}¨<‡,oM/òì€ÞŠ€fc#hønç‘'낾oŠÿÿsFÍDË ð¸D采„yQNõ]¹Õu€é°{ å<w€Í§OžbG.ïòýÝñ:6é¢ßî@â‚5ƒNõ‹?çÚÔ'('®§[ÿY5ýŸëçæèVt@þî^ÙS½o—<ÆãêùokÚüÙîàz¾ÚÙñýãlo÷Ñçgé¡°¯•˜çS…XÍ™‰šü9×J½ø¨”xÙØ ¥Ü=JmfM}xRç»SC”×:ü™HÿëCý sež÷aι'·j@‹M4™A´ITïн $Ä–7§6ÿôÛyèq»4ìÎ }lñzá0°:`2E;Cm¯[Þe‚‘½JñÒ´t0:Ïñ=‹ÒÆ£È ´Ò`²®3è»û!˜&5}=Í}f}YdE®ý9÷ŸE^ã,¨5òÑ×Áâãˆùøˆ?¿^QÀï+ÏŽ¶‘½.‚ÕãÔçw'xÂz   q ü;5Ä{ú8l}¸ý ‰Ò„«éXërcf7’¬1Íg/GÒn–Ão É§Ó>i!•:9çÑVF¤”¿Z)M„ô#F˜Ê‘1Úq‹-þ$2_äûrÐø8²z'[½k²EöŸãä4òÈyiÛøá›·ÈýÊÿî[=ä»]˜Ð˜ƒ®ê´©W¯ ïËîð;M(Ò£²rzm Å‚‚ž¯• äG‰ƒµP†!5ým*a^Þë`)å„û’ÃL¯h^A…€ÃYÜU¨Ì×¢=jD´ª»K^!ð|ô¸€Ÿ.°ñ‰ –° ÓwNÔ(Jr@ÔÛMS,{P öÌǹmG¡Ñ¾ƒôÒóÞh"Úü~ôšnWT“ÏDóÁN`^)C‹¡Ö˜¶Í:´l#Ë |ØŠVU}ï¢Äî¢uáÝ*«÷h“‘vxº« m?Ê8¼RD»ØŸÅƒ.uhÿ$i CÐ!ªÿëÇgx *â¨â¡e<íWæ'eޱÉåËZÐ)ÉõÅe©ßÿîëaÖzÐ}ÀxýfÙŸ9oBÒ@ ªõ¶“ƒwk@Å‹ètið…‰¨þqDÝŸóÖbr^qB1Éf}>—ä×Ù.¹°CÎÓ‚nâZøó|þî“%qÿÔ¼¡¼Zö PvmL%²EjX‰keÛ’ ù»‚' ´/Hÿz¤=o=T óo€maõ±$¦‹£ùf£«~Jš<0©ñ6=GÐ ¦=^ײŒڕGN&˜q9™Øöƒ¦ÞS\©¨$8qÞh«¾çŸ:ãâéÒX£³áqI*Cðc„_ýF†ÌSÒYº>œ‡ù"ÆôËT°ðzÙö›§,¾kœøÌKÙ?oÓ<‡å ¯8ÈÛðSfkO,;¬¦IÝnFX‡‚EçŽ'°1Tþ£f~?’8!AÒ [.dBI‡D–:ŒÌk¸CšŸÜ¨ôîeÛƒPB¿<ûîæ-öZT3ìÓwÆýbiNlù7wé—D…)£Fø€ü˜ßÔÌ\–~äé¤jT½ÉIÎͯsïö‡”`bö/~Äá6Q )äí¸LË›Á¥Óµ,Ö”O Ê¤ÓgÀê’TøF¼ä¡éE^áÝïТÁà–®é ­6݉¼}¡íD|Œõ´Gu›7/²A‡à99©ÙhŸ j¨²"ä-õÕˆ,´öšÒk½†–ªkól›„|R¹HÕƒæÊ®ÿ—~ׄÏohI—{ ­QnlË™üÐæoZøÔl7´?– üà3ñ™;£ ëŠÝ›Óï…¡'äÓW;€Þõ›ùU‡(` áÕÄý ô0ô¤ñPˆéK94”Æq ÆÌnÓùGÆÿõ!]š¾.-ÌLšðØùtÀ< Ÿùø~XHòÔ±k¿K÷,­sé¢aåµÕ~ÁXò+õú• ç×} \Û`S*T7pv19Oˆ¾gÃ1 £ºHb;sçvç$å§ÑÓ@rz#3ÞN¤Ü{¥årÇ*Rßc¶üåõ鸤ß?ð B†ÓÔk‰dÈ8¬Ã}jÜ™O•…Œ›!+ƒåcbú{ÈÖlPu­g†àCÃÒSsÈY—?ù¸ï"rý>þ(¾yñ©X“4ä« e¡µCÒ§ÓÏDøPð€)q^. uóÆÑŸF‘ÇÄ.Ò¾(xã騬7Š'‘®?°¥BÉßsb›î(}~A©¥[e)Ž3E”úß™f,ŒÇPž&-¼Oè*ŒHˆœ9†J­>b…nì¨òe–Ô+û+ªeñÜq\á@`-Tk!WCãY=·CÍÁGõ~/(PÛcsÖõ‘"ê¬n½ýññê=»”ÚY æ×AgÄv¹ß»–†&Š 1EhÆÓ[exjÍ7âîW”¯£Åì—Üœ|c´8î,NV­Æ|¹Ãƒhýbs½¿ M~}ÛÂ<Úµ¼ëƒÚ½K7T¾žö!¬‡Ï£Ãá¾lÕ&®ÈnŠ9ÜѶ5Àt†^d¤3´ˆÁˆÂó£qA>0º²MòÆ›9¿œâ„Éê‘ôöü˜ÚX•”ÙA³'ͼøž»ÁÜXü¥û—`A™«Hä²=,†j2·q‘ÀÒvŽÆ]*XyŸ³›ç«Þždj’[°~+6Ìróüjúx²µ‡¶èó†m2‘ègX’6³2òÚ^D’2ŠQÿ@ $}£}{ý=’Ç{,Eƒ”‹‰ uþKHÒÅÑX|éïŽä}FÆÓ-¾ÉiȬ,›ÇŒ¬’¸·Æ²‘ý\‡Ÿg'r xTyö#7}ž´yƒ|¶³¥é{dŽ9ŽqÍ£VgÛKJGy£Aå‹bòÆÊEÅr(©¢x>£˜¥_­èYÊ8£ì“)j3Ñ'¸Ï!ºî*ÏTØÃ™Ý¸³•"Àùy<3ª^ªsŸ“iÀýW8™rûQÃ@s‘V‘5{¿î)»kMègÖoSï ®ï—ÆF÷xÔ¿”ºß$‘ /™Ü‹¾f‚Æî”±º [hªÜܵÍÙMŽI*E‹Ë’mÈé„–GN5µF+ÓÜiÑ h­Ôï8vmÚðëñiÝ¡G[*"ÊýVh»H2àˆvÝVÌ6¯Iо„®õ“¡CÌqÍ µ<ðàÙ—EM/<èݵ)1¨‰ŽœÆc¾öü›‘H¬†Ù•Õº½¦ƒ.7B [ýœß¨¶ª0‰œ‡JcNà wÅ+%N(Õ§Rxw*ŠéËŠ_öC~'¹ó)ȉ (#Vƒ,¾ÎŸ1ÿÔ¼™¯{ÛÖAÙíªpÓ PÃ9«ë®ͧ(f²ïÚüÙƒ+=q zböËT^f€£~)lŽ÷`˜û¸ëÜa£}-e÷1 &Ý-´>¼ Ó4éµÄãaæŒtô÷ðH˜yöBÀÖžf„mìÍG`z‰ ‰#þ©3/ ./=ƒÙæÒWÝZ'aN ±ß4¨æÅˆ9žYÁüàè}öÏa¡D†£v‡",–ßKÞЀ¥ÁøÂd¾°Â©¬8 ?WF^Z†ÕAÞÝÞì:°~ú‡IB‹übQL:ÜT¿Ëù÷ ‰ ÁVøen±KH¤ù#?vðî0h9‘ü‰{q$yTÿ5˜á îldì¸6¦d—ø‡wýFòx‘>› ¤4ÈÖ,<à‰ÔT÷Mü¶‘f,®õ¨n ÒÕŸÐÿv\$’ŸßšBF×¢ÒVæÈtîÎdíƒ@d~Ú~áó ndi:³?°ø.²‰íß}2Ù_ë{o@îøöÎ+¥9ó?…YÜzƒ»Çr_2³å"÷O‘±v+vä½ÿxà½@ò…g­ÏŒÚ `cÑ< <Ì<ÚB4WëÏeW¢pÊž¯Â¼¸çâ­µ‹~R(¶ïÛ= ®G(žÌ,²ë8;JŒÞþêƒRìš“6¸÷²ìk¤Û(ÝœÐ*¹eX_/e¶¢¬¸ùc™¾(›sÿæŽe”SLêáB¹š÷5´¸/¸8MâÃW”÷ÉróFER_¡ßP)U“(º'U®¹†îpSE5û×yjâÇp¿ÚOyf^DÙµDÒs‹¨ññ@óD;)j–{êz´¡ÖêôÒœ*ê(DŸ»LôçC9†Áø‡"h® ý«Ëb¹nw¶jœ€_TGÜtþóá>üχÿ ÞL6[½ÓæßÂϘöÉ ³Å±gB4a0OA;1xÊê>Ù‘~E(7¯ŽN&ô=±Hrr26Bþ:UÎØŒ@î÷…ë’,£å3–h YE‚¯9ûBÆDÈé'ŽîVô1ž—ã ¤äÒi ž¹IÈ¥‹èƒº?nü²¾V ¹-Ç6y š ŒR1¾!Ū"ÔïYË#¼î;²½D5Ð4Õí™ç-Qªä—X¡5¹ÏÀù,´•/£àûè â°³frŽkÅfH^ ÚÞËtÐîJ}•íîehÛ[J1üÕZéD—ÕC‹î¬ºe}8´P°üb¾-'4½Œøþ.¼@Nü Z«$¶¨ í݇—çО¼#–›®öB_ ¡ëeCºG!ô$ ¶_''Ó\a`Ž‚ãAø'Š&—Ì Ì™'“5dE`ÌædmZ¥ï_RéŸÖ£†™‘¾=ò„Ï™hò ‹0,$ääŸzUK×ÓrÛ÷ÀÊ]ë.eûǰZÝߨ·ãlèîŸ/Íï€ßK~wzÂöä]Kñ½¸ã'Ç÷ój›LJ ")ÇÖÀô.‚o Ç?9Ž”òÝUCH’šñÎié¸ÍÛ6“‘Áó•Ýî°WÈØrûÐâkdö¸[ÒñX Y©˜9 Ö‘­núEÒräe©…­º gÓjÛžä&=¦G<мF•ßîú _Ë~½¹(@9›t¦ÿ := ~Áç„B=Çé» ù Oúö£ ECÛ͆ŽÄ xÑ™ìú@”b¼_..ˆÒAJìŠ_P–Î"é¼Ò Êy½¡»ÜeŒû¶ÉûýS;P¡ê›-¿*}¬´±dD•°^×yžTiPUŒ¦47 S¦ÑtÁ‰Ÿ“yÍZi ;¢y¦ÜœejäŽ.»úŸîäËÓWù(yÐêóщø ´¦IðÉ^ñEë^ͬ=—Ð&]š»ˆu7Ú^|_ébu‘üIÿ´#±•k;™ûŸùÐÏ{t-ˆ º‚oËYxtC3_ôjñ/-¨9vk¼*9¾0ŽíczÿÊEoþá{JÍë+n  sÜô©`yùC½’»ìJ 7C+0irH´Å_þ­Ÿ×îÀ™gÛPv+uaÅjyª¯Û.÷A“Çú+‘¥èŸ£EÙ„TøþŽQXSî9ôy+vt™8Á ¸0GÓ óðuPªÁØ’3«q0©LèxõÀTâ`œ/ƒ$LÕV>n+ž©Ãd¯y»¶aJDOö¢Ó?µñí(;çÌìþÞCb°³'”®±=s,™ ±Ù0Oü’Äo0æºv~, šÞMÓ°Dqd媜,«6W å™ÃʃólzM°FC›I?ë ¿â¹“WëfaC'¥!¹= ~¯1<6­oï3è÷!Á±;²1„n{ãøTÛ‘(¬Ê3µƒ‰?z}x±…$‹ªüC7‘ÌLXï˜2’7„ß“þ Šç6ÔË?3#•ÌÓYÚÍq¤¦^Ë »ˆ4kÑ?¾!­þ[ŒÒ]¥Ù%úéoO8îËdE†÷+·›’‘qæZ)dØ ³>©ÀR ²¾ÜÍ_A6=AÓöˆ÷È>Äo-”–Š8Œ{¹C(Ãå‘;«=¶MKyÒJö© ï³”ÍeÈ÷übßð‡›(Àï|U3õ V¿ê¿RÌÂÇÏy ¨ñ¢_f<ŠÕê–ýøÄŒû¶)ËâÅfp¿”F¢²8%JD8±Ë JŽa¤ŒÆJQOVÝAiåû[G}ÛPº:®›$|eL)F¢xQfì•r†ʦ"5Ýåbk ß“£ÂÞúÁ_Q±|L<_p •ïÄf¤¢ê);©û;þ¦f¬biæR€húØÙ4Õ?¶zLHQ Fóà±K^„¨IÔ0µùúj©|Kø^ó4)rŠzä.°S’åÿÉ™¯¾üx w$ÄžC—üVëõ`ØÎ,l|Êà>ÜU«ûØ¿ ¦ú<6WLavÓqE½* æ_“É;PÃâ%FîyãVXö¶ò¦½m«©{v©Ã:sð™*¦.ø÷93°ô=lg˜·dÓJ!áù=¼¯‰vøPê^È59$Ý3f8ÌëµÃ‡~¤$kö¦ 0O¥ U´)‹†‘1Òìõ$÷èkGZg‘ü$‘®³sÒt+ŒfÄÈ#qkÂ`°32WGHDÿÚáCý§ƒš‘½G‚­S¹hœ–/Jê!·5›Á™¤äé˜$ !àD>/ƒGoÿxZõþ›(ÐsSÉç% P¬½Í[Qy&º«÷a7ŠÕgœÒ!ÄýB"Ç,ZúQ2ì£uPšQÙ@çmʸÆ×÷ÄK¢ìüQ!väDùŒÛ×ÛvògpÓ‰Iu*T>zNõ§ªZøä=?Šj)æôyaB¨.ã/›F3„)þ•+þ¹¨Éý…w¾òj=Ð¾ÃØ]„:{ ÔbGIP7“þ`9gêû8]²ÊGCìÄ`<äùpß›ñ49r‹o~kMŸü,çÿÍ®w©¬ª‰£¹›÷¢×e´0ç+·-¥k«^o®¢•ôiš/Ѫ7âTÚVZ§Gh´¥u¢ÍùcÑÏšRÑVVcL²ðÚÎx+rh_Åï4.˜¥¸£åت(ýC´'äœ%WµCÿcLïŽ/îè·6Ì4ö¾?Òè’Óí_/ƒ±°¨ÏµùcP])¾J`%—º/Ñïèæp€Ù[å¿óE?’ªL‘˜TÃû퇜Üñ;ƒ£çoBºãi¥Gt§ÿµÏÂ%P^‡Ë“î#…ä þ/âëòÑùd¤GŸ”m‚:òzä­ ÿÒïý¨ñ¿óQó9¾·[Í$íøn»¯HçŸóÂÃÆ{  }Ä ’© •YõsÎÕþëw¥çô& pÙÞäžµóŸ¾Cæ¶‹}ú°^Ÿ!„â ¶kÆÌ®P°éžFó§Ok×IYŽøÐ6.‰ìèTä}p§úòÎDZ8ýŸù°TçÑžÉdh>ÇOò¤:Zu ˜*¬æ¡ëÆÕ‹d Ð)ÿâi€n|w5®Óú> ="©›î‰ÐOÿn‡¸aàµ7{ÄYkÒ²îŽÙùœ† ¿u_]; #­4ŽMEc0Vþ›â­33LŒåUkÀ´QéÂöe1˜­¿°Å›óRŸÖ-êaáᾂo°Dç'ßvG– 5ÆWs`5TLüŠŒ üºö^wn: 6^3ø`››ž}·¶ÄÀà;‰È”~Ö(ŠÎÑEb«_C¿N#)ß‹|Þ|IΗlh_ˆÇ×bêr‘jŠÛx°ŸiÃ]JFŽ´Øs ò ÒÉx^Ž«GÆ„»YÅ´W¹õ‘¬ã>dó‘¯  <Žœá‰ëq:È=çëšñyë ׌ì;¿Í¢ö9Íy¢h“_¸‚"·¯ØÉ±ŽFôÎ"ŸQâÉÕ¯á(µäè¥ÇM€2iµ=(w!2ÿ@ó[T(°ªâóvEå3…èð€"±«ÌÑgˆÄù5üOQ}‚º·¼^ð¿jÛ2ŠZŠÒóTW3P‡Eæ É7ÔÝ–¸õŒõ{.ãÅ"4üH½äòKhvøü#C¡Ch.~ûFyš¯»'Ìö§¡EÙr>Y3ZÞŸ-¾ŒVš3?X̦Ñê÷%ÑøÁQ´~÷k3hé ÚûúeÐùÚRìsáx…¶ï,®o=üÏ|XÇU1• ]Åz½œÆ YâEÓ¨q1½9âf_„"fêƒB¡\ÞD6èg6”{`áB·ÅÂߪZCþl¢1…-äæîŠ&dP…×ê™'^ü­ù$¦‡Ö<²È;&ÉŸ¹¡V(ý¸ÛÍ\é5eÿ̱ ÚI—tRºáû§I®âyqè ZÚŽÕDTnË: ÃÊ=¬6ß`,Ùǃұ&O¯º7ÃTûÃÇdE0µ½Hm-ÒSwrW.8qÀÔÑ Â=n¢ëô®Æ4ÔA˜Ñ]â–´ðƒÙ¼ÃÒ?`Î%öäÍW,0/äèßϲ »^?OÛÏ‹{6S¯†%ZI}=kX>MjËR™+Å‚6Qlo`í‘ÞǪXøÕ`F¨Uó6.³ÏMÿ¸ ›â»“]œ„m¢q]9ÝîMk»| oÅù^•D©"æõ q;Ó³¼ G$eqòâø†d—:Ôu»#êX—I$iw”ÖêSadv¦Ü]7üYú ®»+³!›û9bôx”1¬¿›²‘“N;ÄE’÷ê }Ç äîå°ÛuêòÞô˜âFÞwþ?V?G#ß;éGƒ[’( bµ)&x×i¡pê•Ê_NïPÔ…êÔ7½·(6|õx#гÝN¼¸¬€ûõ5Žp½¬@‰äGÚ—(Pr«õ å¡™êÃb-”Öku½E,Ò?Š…Òµ£Œ³Ó"¹cÊ’3M¼¼÷e¿~·½cŒrùŽÝ4›1¨ ø-–‚Û#l›½PùéèŽM± ª7kò˜øUTs0õ(³Cü/þ—ÿø0¸ NîÚ…‡º¯?¤i¡Ezâ l‘"S˜îψëc¿_ƒ ´ÍŽÓA¹ßÌQ®|Zœó¤xÈòïPᆼýïËó —àS¶ìD¾…í=¼÷ÕðÏùƒçƒykÈ”¾2ürÂÞýzk‘®È½£Ců«‹+©O¤9ÕWyt{¸ëÕ¡L‹ºµr‹ªµÈ†gö)@½Ö™Æráå?ód˜«¹c åä]3vch=†OëŽC›_¸A£'´<:ie¾òö7U‡ }'Îê{†@;]ñ©ýäÐÚ’ò[ȦZbj›}’ 9?séy‚<4»ÉˆújÊ@sÕåw±tüÐâðä\Yb´2ªXÓæLCë°¹Ï/Ñ–?óoÊÑ@‡’A&‰ÉSèlýÔt`v ¾Ò‰äŒCÙTÍ>{Fèg™WÓuŸ‡òKUÆAú0¬çrä„/ ;(éðÒ³ÿǤíº%Ù¾0ÕMà^ýžfתk«ÀüK‹Ô'‚–°èvñ&µÇ4,·¡+¦…Õ°ä©|š›ðkôKï·s?àw`ãÙßéK°í¹×úôŠ4Ú’ˆ9Ýßáýԩ!­¡HJn¬Aä´ >áH¦c§V¯!ÕÓ™Ë?Uß! w¤r?ýÒžþD§{¢éZùÍ5) ƒ=GШ2n>iÎïBæ ™œ6Țŭ¼[Ù»ÌM{Ú‚‹åÄ«ÑóÈíèš§Ûì<=O~—Ÿ D>úx9Aã%ä?QIZ6ù ú_Vô¹ ¡P"y‹ûCMyuµ¬»šÅz·ãiŠ®â~•е@ºç(ùþ´|ü|5J³óް >Þñ¿—O3 ìP¦{î·(_53<þ(*ÚlK’¡²¬å+¾uTåW;¿j^±ûõbQÅ!Á¤6 5î\8(牋Pî›E-‡qbᕨý­÷zàUOÔ=i/¦úü‹CS¥êhȸìª~®±yíqAê+3ahê¬rþX!šYÓ¥ö´/¢¹Ò W¤R%Z°0­§D¿G‹©-ï··Ñ2¯'žBÕ­ü½‚Z«Éú„¶ õô•¹ÝqLhóäP[”ª9ÚªR 7DÛvª}>¯Žâaçðþ¨sxxÅÌ%G­¯VS³¨l-à¸zì8dÞ]ïyoúw¿â½K%dBÁB JR;liî»»…”äüó­Pxû¢mÉã(LÜ Œ#9ü÷Y¸CÑ3®Bš´?ç(÷&üÆvøÖŒ§¨%Jnô½N›¦ƒRù“‡ÆD{ÿáCó|¥ñ»Ï¡âcýxâ·øp²à%é9Ñÿ̇Jã†÷ Ùgí°ñ"=´špWi®Ü„v!»[•Ÿ‹¡S5d"&*¾{hpDx7BĽЈø ègÙÅ%wÞ6tƒ!Ünjë`˜ücóBŒ´…‰uŽ™ÂXé.¹gÅ)0ѽøc+½¦Õ6²g&h`6Ï顆[ Ìï'}ââ[ OxÏxKÃû‘÷aï¾ÀrmöéÛÌÇ`õ~xñÓ¬\ø(—¨~*6¢LÈç9§a³‘׫Ï| DtÖoK"áï4rVvH̽p² nI†»^ŠY!Y_E€‰R0?÷%¾ô ©ž•Oy^Aš ?4Úîé m“üÔ•ú6¤R²ì:9‡ŒvÁ׸»‘9d…·õV!²ñÆiúvÅ#§Ä{žšÒäŽú2ùÄ>iÇ'Çu7 ÿùýÊ„È-l˜ä®{¶Ž"Š8è×ÃQlÙz üè]” OŠzÊRI\Úì (cѦ˜½…r†‹yI¨`0à7Úù•ú(ƃ\P5Ú—ÅcnA͆*!ðª{¾OlrÜ@az׌+5¨ùpœ05LµO/ü¦ø …ºšük[Â?PŸíÕÈ7ÌÔÛÄ/ ÑçôÁ §ùh|É3w}‡‡MiK#2ˆÑ´Ïx¬KPÍ’Ùd ®¢ùBG»khÁíù`òÊn´h”ßXê£@ËËjê¾oN {J¼Ô)´Ê*ëÖn@kÝ‘j;ÿ÷hÝj™Ôœg6vÆM2O½ÿ#î­º/lë]éµavKÿwJš.Ô\b •!‚/JŠ™¡æ¾P®ÁH lF¥.6ƒ&?B±ìÞ:Èß8íNiå ¹åßü ®EBŽÏAî'ÏÿÖ|ZÆ£wz¥ ì‘G†Å‘j¨¿§z…š.ô2P50ÿù´±ëjø^RÆxàM>ôEf Wd‡AýW&ŸÀ°ÉuGË«vÉü:“÷cïÕ?Ó”z‘wÁ´ÔoЉfê]£wCQðW<à¦3 —_Ï?<¢!ü–œ½E¸àýï“V (.·;بØ÷ŸÐ`&â@‰Âj¯„·4(E™"XVŒ’›ÌõÙ&¿PÚ–¹3ç»s…âé” .Þ¾ú’eMj©½èPvêLúS¤C¹nž©fC)T0ã‚Šã 7Æ êPùuZÖU›¨z­ƒ Þ„¢Úù2£;l«ˆW©<¤i¸P½±&ef„5f#ý»ìQ“õq;¥$jiµh*ÐüßãÃÂ3Îghü”ݑد 9É'u¦ó_Áoõw'·Ò„ÿˇÿåÃÿòáÿ~‘ÃZˆFÝW„g ‘î(Må> ˜ª^&™Q½µ)b«bWþôÝÖ¶ó€’[¥±æŒÏÑéÎXóW?ŸNî‡7§Ãϸýý{ž¶bwûÚ_ÝU?rc’Õºõ_¿åPç߆¦é‡ò—ÌÅ¡¥"g¹æÓ$´Ö•y–…XC[{+Åìè`osn÷èˆLM2·=LÞG¶Ã‡tT²½ç†÷‹æïÈÅã`y8+¹ÝŠD„·w!ÏÎ2aý ò1¯®« ÿ©>M[3]*5ÓU¤A¡×Y&û CQämãÕÎu-›5oWQXÃý&ŽQi&ö(YqWî§ÉŽŽx(ƒ¦c½QæÌk“×â«(ÛNžms8å/N¨õ’¢¢Œx,Û^T&æí¢¤ÍC•¹n?ݨ¦22þ‚‚q-Œmr÷jxº?yvÞ³ZF-‰;WTé~¡ö-‚št=¨ËøŠLw‰õ¾ž<)^ùrK½Go£QÙ¡HúÓÃh\äúy¯Ršê”~~Ch&bÀZᙇæ¤ÎUû èÐüÇ.×Äe+´x]UŠÇyÐòlx'qÙ&Zíu æòA« A6§z´>ûØ0àl0Zo.[ÊcA›Ð—æõhKÞ:—`Œ¶b5_³Ðv2d䔺.^•! d&>†ZgÓIe¹æp÷g¼Ì~ Fê~¼z¬UòúlY«Ö}êߺË~BJ"Üè™a§6Î%þõ·ë,-I&Hc¦Rê³Z19’×”#7Ò!y÷ï6ã»ðzèÒðžjÈðË|Ì'x>\<837úõOÿ0™hݧ¿ú, »Ce^…g…79'ÛþÿÚÓ|¦¶¸úo-²ðçN 9ñ÷ýþÛ·K˜˜‚¬cÆÿÞGJUµ¸ó$þÇFß¾šþõû¿|8Ü5žù{ê?óa•ètYú%h¾V÷–”ë´Ú,û–?æ€vI©§Ìï¡SÃE5·Ç¾{É:ßýzä¼y›-¡ŸÍÅ!G` G:˜zëúŸ¹Ræ÷.±Ã0õ3aë©Ué \ì§ûcEŒÇC™Ua¢)Î]3"ðŸM‹Í¾ñ© æE-‚ø#a!†yÈqø!,ñÃ>Õ;X-.O«1ãæyІðË?÷wDO%lø‡ÞO »›‰Í$²¯ã‘€ôãt‘˜?´eÉ ‰Æ'†üU¯"IÊöšÀ°êN>ü d¤ˆ»ê´L\¹¯"•þé+]HÃ&ÉvÓþ5ÒÆfœeè”Bz-PV£GFFÃ@U5dV´¦Õ8ެßHJ½Sú‘£¥P—qx¹å™N¦Rf"¯;côÇgEȦ¡ä%C8 Þµq9¥é‡ÂCVù]¹(îø+Ê÷(K9RV™ÜQú÷põµ{Q¶Œ¬(-!å»Ü¥Þ¨•£’u Óùç¨2?Už~à Õ.-”ûtó!Î$OœF «×ö¼xp)^÷Á;ÔÊù¹x4uB™½½²sPÏj¬“³" Ä"nÈ›DÃy^ª8¢<<ô’Ÿ¡M M°ÿöF 7šžÚ§zF@Íx•Þ$î[D³–‡¼qª:hî7éÖ(CˆìÚµ¿¶ž ÅÛÛ¯¾ïê@K¥nÆ’Ñj´Ì»Gñæ=ZIÖ4ßÿ‚Vñþ½”6ähMjÀí(göŸùÐÓëTQ't•¸È¼1#„fí Ú¸à꿹ðß×鿯ßÒKœ^òŸäþÉ«ÿO?ÿÖmÎu“Ñ;õ±ë¿ý´,Öl×ö¶'ÔÊx± µþ¬ÿúÐ÷š§”y.úÐ΄OlmOê_>—ò×gÆ&ŽNh(ÐÀdáÿ¡”?ÿǸÅÖ Ó.‡Ö¬×“`j‰ë!›¯LU}£õjûø·N_f” 23EB/CœîþÙë™YäséIK™Ý0ïÚë™ü̸Âm·þÉ‘×/% yîƒåÏ.>¿zoÂ*ó­±ëùg`íñðÅeŸEX—ü˜Ý[²=Û²~cº°ù`/¯t ¶ÏÙy ¤³ A–Ó3bzK$¬z¶iØwñŸëžã¨Ûxù6’ÿDõ¥÷0’•)~›Á]—+$¾("Åp•Zû¤J«>v3©oÅM2ý‚{®°3Ç;#mÁ«‹ñ¡HWÁSõ> é»køš‘ôpÓxD&½l#çóUÈœÊðãšCÛß\É–Kt–Nü:r¾õ<•»94y•ê:ãÞ˜„ú4‰.d5wQ´<„<ã4ôs‹ÈÛ%Ó“£„|©$=¯¡ÀYñ°¼À?çðœ¢~±÷P¸Ç뿉 Š&¨t¿P‡û8†Ÿ ½¢Gqã»'øÃ+q¿ŸzÃÈ%”h]³a@)¦­¡ †D”œÎñšœwAé3‡µ‡XÆP†äzÍÞO(ËA5ÿóÊj†§úXüéÓ_Ú*¬jE¹õ¾­‡O*QÁåÌg¦&TÜèù@¸K•³ÝKÖo/£ê½‡JÜGµ2½T¯hÄè‚ÃÑSí¨Þ§¥=¯ÁŒ[º©AwÃP“߯SDõjé)DÄýïñácÅ5Ù¥h|Šaÿ¹oÈÑ«ûz%â$ü~y>¨‡Åû¿|ø_>ü/þða QË“ÊA4<Î.5Gzþ>t}ÂÇCVµìã·o–µ@Ù›uN·ù(a_ñ>ΟÇÐe¿†\nÖ±ÍcîS3®yIž ²Ï®uHvnÁǨ3¢÷ìàƒÊ…zñ¦ xóã™UÝ5mÈç¯gWã„ÎA¼µå·!·ÛÊÍ=±ÊÄo´u™‰CµÕÛçôœ Þ=fΊzš²“LbHK¡åæ€EœW0´>))½W° miÉU®š¡}&yì¼…wÓ|N[÷¿ç|í¶î µé ÐÆ¿ÀÆZ¦çõ£>‚*™Ï}ì+Ðün"xвš_öÉrׇ–J³Úª}ö¦-kçAkûSðퟅ¶ºí¦ ô{ÐA¬ë(é“‘kŠ•àûýDŠ1õXøñsŒ<”(úRÌÍ3nÁ€[PkU¢ ±ÅFòj4ÂϱŠ.¿©G0²5Îöäw*Œ;ŸQy%y¦„Xõ&`v?Ç‚æ•'0/¦”@` Óú>¯é¾ÂRã­û—9h`et G.Á/I(é„7£ ×zú`+ìꌱÃs$è9ãKv‰âU¬ý,î"IiZâ¤2’…¶[÷>Å]oÈä"%rÝÖÙ•Ú©Ïm¦¢¤fÕóóÏPêG¥ºÛÅ”á—]Û=D€²‘îqÚ(/‘±¿9iÆûÉ"o¢R^gÞu TI*°o PÄãü†»yŸ=ZÙÇà…Ê¿rój ð`ÌWp¾NŠš3â:Õ¨­'—‘Õˉ:Ms¯(ס^XRÝÝŠ84¸-Ývk |_;º(×£ñy!QéÛ hÊmù:Qç<šnŸr}9Á‡fß_%]Aów¬¡ú !há[+¬¿Ã™–Š1BíäMh9ùä³ZE{§„|žGkËâg'Éк2dвHmŒ­]×­¥Ð¦NÄßv·ÚT6ÔØˆ¶oDhOã•îsÕ… ¨y–å»zä îNxmïr÷$ŒxDr_R¨­Y¡h÷Èôç>È%"ãT°‡¬FJV×?ûcnö”ìøÞ±ÃuQññ@U}¶sRÇ ß iNÀëõm7G ùeÚÙÀ@H2­wù­ç¯¡NEº4Ògî ©¸Ïýé?lâú Yݵ—Ųÿ9?ŒÓ¹U†sÿùü0ðƒé0£ù?uÙþ+QM?üiÌ`Ç_Eû-Æ¡8vÍàŽ‹”€—±æ%("eª~÷êÏÞ¯~þ·PÖCýr¬. *Ä}gI|u¡2§lôÄÒTé¬Ðé'îð¡.é§e›ãðõñYÓ¼'PW]û°Ø¡í4¸â.@3©¨ÊÝ;÷‹ºh¦½”¡-fé–ŒBG’ùnJó~èú¦6Måp~øWEÇ+(AßÿCæG`@+›éuÁ/ü.ëC“Õ?ïU?¸hŸ#Ö_¾[˜Z–Š`LUœˆ»û¦â]6j †Yú ³ûÚ‹a®ëÆŠÅP ,è4¸ûæ Áb­ßçÎQXöi•8šj«òy™ûácГ«2`ƒQÍýÈl6lp¥*”8Û¯XÔõP|øÓðÅ$ú¨lgÄùÍP±'’™6z_Å]çC…¼«Ÿ eO¦å«1 ¤Žy/Q÷»i­ö•™Y#Ý2ム_ÈPÐøòFødjÞ`(Ñy¬ÞU¥s©È¡¾7%§ø-î z0ûZ§y‰ÍV鎤!ßìUÉý!(Há6B©Ô‡ÂÖaEu*ƒ(ºpаzRÅ{òO2OßEÉß¿TMß® ô væ(ù({B‘´dû7ÊŸÿdËxÎ ÖÎï}‡*Þ4gÌDúñÀ4GÁ¨’#âÑ{sïKN e´§¢´LÞÛ9ZyµœèïGÉe© z^¨»qHÖþê×Ó²$ÏAóä¡WñŽç$4þùÄFèq š*yžä'~ˆ¦Ëׇt ß¡YÊvܵ>hn~K½”`Íx®’R'£EèöõÉOhÉ€â1QhùP¦kñJ"ZQÿ¨œwé´5‡)FL·!ÅsGÜò ‚'+®ûS6Yßña”…üTê…ŒOðIöëxt½>dí¢âåÍ9 Žš_Nù3¿m!ÑZÞ\=øJŽòo}ÏÃ{>bÕ>>sX¯˜û«×·5 ýo¶¡4%]Þl *nTHíË‚/yEEÇÊ úF£X×1¨Íôàz¬u¢âGï:Bƒär_ å;hŒ¡ICjhræõ?>ÍŠª~Wh E]­VðíehuhW+»EmÃénÑD@‡U#¹ACtn°V`‚ïîzßžœ†*ãÒ€¸"èKwnþ¦÷.¿x݃«»EÃáçËýGò"0âêéò^= Æ4 ô6*”`Bôò¦í‹R˜âe‹ £€ŽÎæ\YM˜}wÞ$÷3ÌŸpZ爛€…åƒ]?aIƒèîÖäXþú|î—?¬ºeÝ´qƒ_Aéé—la=ûlœûø¼¦—¾óyoßúª¾Øƒê —œ\Ð6xÅQЉüm%\ˆ:‘8›â}Ó‘H$%Ý0(ÑF²s}~Š’H¾©Î4[ë…Ù9RäKËHu—r¸!z©op4¹tþÆ=9?”ßi3#åÇ]¢·¹‘^—³ƒb!ð.3ïKzdü´$ï﬌ÌTŠŸìê‘%V%Úé/²ùDß&ÿňïšbƒ¢£p/wúðGq=äÑœø‘pQyu Ãiny#+³ótòõe=ÿ®ÔŽüÕLMûQà‹Ý‰rR[œ¬Ð0L\CaEïSy(òôÙ@µÀ*Š‘ž"ˆÇ}!‹§ëFá~šL«Ã/Qâ•¥–J@JxícƒÖ“ÅôvKÑÐpùÆ(· ´—ùžN¤Ü ú«-_ìFþö ¶ë”2Út@2ÕþMZhù•ÛCÕ½-Ü ¯~‰Ÿ†æÒ‡n¼*w¡…èèÞ—MÔÐrþàÒEŽ‹;ÏË LWÛ ææ¬‘H–,Ñ3‘„Ê öEW#’nSm ´°â.&îË®Gœ‘ÒŠÅxö9î.6ø9tÝ÷X†Ë[¢ Òín4eÌõGúpc{VößÈÈ7—}~c™z™f¨:¼‘¥Ü¾2«ÝÙšÄwqŶ#Ç6Í,kû*îEóÜUä1txç–,Œ¼×E­ïWU#_[YŒpL( ˜½èC!rëêëÚgQxyq|—ÅØÞîû¹Æ‹âŽ~í.×v£D9Ñ.Â×(õ4pkÒ8¥{KÎ?¥DYƒ–uýþ”u»rƽ^<$^"¦B%ψ§ŒçjPÅ.îÀÍÜ:<ðøxÆÀž¿;ÔìÊNˆÄa ¥èðàñ]BfTa¨ùæ}Eà&j“$O³G{º.÷QOËI1÷È%4÷ø–{$ eÊvBcÁþ'Z=Æh2Û *òŒMÛ¯]?yÍÞuÉó ¢y°Hâœa6Zè»k¶ˆ–dSŒ’Ð2·rA¦F ­Ž÷î[ cD«ßô²9Go¡õÝL¬ry´a¢x4y8 m¢®²Òû -™î·›;ÕKÅ9ÿú¼|£›.Ðgž¹ñΘ¸w‡îQûÜ3#{„ëø´à {S”ŸO š&å4>i„Oø Y) yÇ÷öxîò¼t”^—°þWï9E{H/^óÕ\„TñÆnʧ´ðZXÿðH ì®( ¶€WÑ iÄQüŒ?Â"ŸÒÏHšç<ØÑ%ûGC:eÈz(7ÂúñŸþCÙá—÷óýçþCÞHåí³ÝÿÔÇ*—NR÷ípâ‘þ¯¢¼;¾6ËB Å{*£\cã¡x(:“Àù”ܪx"ÉPêb7ŸDMeîýHzZ¡<Ô'È©*ù“mC¾Þácz¥~½i¨~tižýÍn¨L¿‘ u)ôÅ_‡ áûûX½#Ð~£ŠY¿Z4e>\6³¶]¹~æ{vîÊ?ì«­¡KXÏó@%tWq¤q>ã‚ÞXÁ®GkSÐߢ”ªá烎¦1÷oÀÏ=+Ûjí`¸ÝæÍ›Äy-´´œ–'€ñªI›ÕÛ}0E[x9߈f®i^N‚¹ðœ¡—0¿ºq+½òô×O_€e–çM\ñ°Ò~0¯ÖÒ äÎÆDÁzÚ¢>£õøÝN “õ¶ŒdŠ«#Áúaº•Õ@$ºZCU‡$bö ‡×'Œ¤àPÉBÜÅR³yÕ+)½¶,”z¯´SÜ$ÒR§sFˆ]Eº¤ŸáÓ]}ÈpJt#¼#™<X”g‘åû`=­’7²g\£´}êŽ\õô~IÁÛÈó4â𹽑ÏGäk¾n „;¾“ÐD¡šüÒ’‚µ?k~AKÅUTˆC.œBIcõj.YúÜjìô²Ãe‰Îu·GùÝ„~/ö¢âùo_j KPyIŠr¥Î œÔʱ~ðíÒÒñkÇQ½8(ÄC} ÚÄŇ®žG-:¯÷oòÅP»³eúR·8ê&|H$^TB}¯ÈR…bh¨PH¢´Ñh¼pEDºÃMùÌN³ )Uö­Þ!b4­i`4qA³6§})h.*Êø¡a Ík6öÓr~D ‡úûõÑbú•÷“£héÙµÞ£û-'ªMScÐÊ>Ué$« Z•WݬäkÃ;úÊ'5ùœÐܘ-¤µØyÄŸf‹»Zí|/%†–‚ÖÐIP˜ñŽHš=œ?qCµý&µ¨ÜÍåëÓÆeVŸÉ¶ÈAI Qþ€ýA( ÌZÝÑÙçãñg)ÿÕm6eoyÖdó:íä]0¹È7#¥A£ëoŸù@õ­™µ—Јõ/|>.@«s‰zÑ+NèüÁŸç' =,š çé¡¿ß«áž? ¥Œh·IÂHS ÐÆ]†ñÓ=#m Œ0ѹk]j0&…ó"¾J†ÀIJó_—DŽ /'caRÊ^2*Ð&'<¥¸Ã4{ùT¾„ûŽ/•ê•ÝÊ€Ù»ÖκÉ0wùÖ÷‘Þ˜7Ú~´Ô Z¤æ ßGaÑé”ï,ÂÒUkeÓ¶5X.sÛÿÌV%bd²¹_ÁÚÛðfÓYcX×9“P¨æ“j¡[Ó°™÷üÎ%OØN·~.} FUåêN>B"Š£Uä¤H¬zìÛòü$ t1~´IGËÛó—y‘Ü™LÔ³ß)HD¥&Ú²X„š²®wG/™iä"M`Ì\ŠHÒâÓ•ÈÆ ¤Ó;[ïÈôÇÜŒ#Hœ‘áúZÍÍvd,r?r› ™é›y¥êU‘%Èà­óOdÛåÃø1' ÙØù%zŽ!Ç‹ÕCš—y«E5áÕ~ävü Y[¿£ÿ¬ eöVä=#M«ŸyùŽºç,¦ÞBþîQƒŽç÷Qpç[ K¿†ÂÂüoßB‘ö¾¬·ô<(¦g̵zk ÷]N¾K9âÉ—¢aS÷oÉÉLj¡ä‘H±KM(yó‹>=MJóý2”²ÖCéÑà!ûQ”Ébª”taBÙ¨yjöóº(<õÚ¿@å’‰¯½ú† ý*?d ’‹¯ê¶pª°$ª™õ£ê˜ý¿Ï[¨Öá‡uˆ]<9ÕÜì¨þ+ÒnQM ²‘Þ H~šüŠG<úPëzLÒ@Ã3Ô.ž}¥ôŠuNU «ÊÙ Îöàù]yЍ{oräükÔc8ñ†¡î%êÝ=ÿD;¶õ6«õç)†QßiüªVcê—GŒÆ¢»Ñ@Ñ) 4pöùæn÷ >„•´Ï@ƒÅsOv}é@CÉ}î´×ÑðTüÀI«T¼Q=Ýx‘è>8nq<ü²Ï›ªµÂÆëׇmÂÐj¤½'<´ ¾Žpç°: Cùƒ«m’CO ð¬t"çãGð¹M·£Õr®[Yü"í…'í¦–evòh㣀¢zxk®<úŽA2ÒÊÏ9~Û÷·¾s óò­EøHªÖèP÷à¯^ ¢]ô9³nAÉRö”ÈÖÒŽ?Ñò>u‚Êê+i§Bý *´bQzÇGk’tŽ‘ñÀ×ͬœÿ¨[Ý Ž%ð€†³ÈÒÇFÐXu@*/µšZöœú ÍÏÍz÷Ÿ+– ‚„_Ÿx ‹ÏͿ۹Á2Æ~û‚§°RìqÙÈÖ¯†ðµÀ:ñé=T5°ñ•žÝŸ63 ©Ba;7ÎJ2ó&Ìu×™º~¹û´“—Â2gaŸÈ°ôä­øÅÚX|w‡T5x ¿Oæ‡fò(pY¹¸±á0 & ¾…¶ß7—8£ˆ_x‡k€ŠQu’|¹r÷½}|új|î7ºI×ÁƒùÓ+£P*2+ÅÿÄ”þÉÛú¸8eÝh‚kÛQ^Š·˜Xe ¥mÙ¯~Fe¯êÏPµyox•a‚¨]vtü(ªs³$E- ]‹'ß"¤©=ÊÏ?< Q0äÝÑi¨Öj~'¨s—A­ã.—= A­îÃiÚÆŸPÛ˜òakØ9ÔÎ¥Ô~ðæê°kd FoÓÍ~A2Ô©Œ};Œ ui‡ÃšO¡®Õoq-GÔî·öwFÝúìKµŸ0é—ôµ/¨wÚîàü«¿|8ù¾÷Èõ8u¨!8#’´xJ‡§ú x8¡èR›ÛÁÈå±˜Ú 9E¢3kÙ×®u,„,î‡#[‚Q©OþdMÆ Þ}ªô~>éìýaŸ/BjÓ ›Àî{ðþtîѯø!—<2Bå ”–żٕZ U­²ï¾ÖBݧ¡ÌФ]yý§Û´P)ìÑúùZ% O ‹ËB›¹ÑÁ2ÐþH˜¼í.t=Ÿò<êðwÎh»<ãü4k,´qé2õO6B+é5Ïà–cÐ"çÁ ÖÍ-êVÖ.FвK½ò\´œö{Ì}®Zº¬(ž ^€Öó¾¿ú Þ@›ÎÍê‹‘æà8%«‡Žú{¿µ&@Wz»§b‘ t7L n¤ZC¯­MÂðc„þç"–Ìã0(•¾r¯A †fhCÊ­¦ax*„ØwîŒ)xñ°f‰ÁDUàäí}0ýÜèÅÈ„(ÌYYŸ½±¯ˆ$î%²XÁâ÷™ã®·ay´¹æ„Ù7XãHºG4R ëþãNî°)ý’5ÏÙ 8µ;m!á—æ™S)óH|?â^Bõi$½¤‘r¡ê1’û+鄟GŠw³×eš†q7Í^ñD>¤yæÈË耴éA17ÂÕ^€â§ÒqWdÈ-T¬Ý–E&wvx“Š,zá —DÍ`»Ù>¨9Ü‚†Ãö$l} x¨+†ÜV2 M–‰w¿Þ‹¦obéNüXF³Ú)¶9C4·R*z´2Œìõ‹Ç¹~ E[aÅ•û‡eÈ¢˜šGZíÓ{wìŒZU6¼n¶¥Fk«^^Æj-´î.(ýqd ml2¿ýÔG›ZëC_3¡­cú餛è+¡c—%#„§çoØ_ªÄÝî>ª3&­•Àû<ÚêDWùßSuÖÔ÷…´ÉÚ÷o7 Å3|fF’-¢9ct5àUyFßžPaxÙéõȦ¨’–/ÿ¼˜åi­~.Û ðþÅ͇¾‚seoÍñóÃ?óKk¤حþóüÒòsül…Nk¡2o¹ù^(¼í½Î«PØÓvó;G}7h ƒbƒ8Q'£óP"sƒ=°.Jw+z9¯ƒÒMsžf(½¿½e&‘á5)á …„Ã¥™i¨S'Ê…šV‹„úwnðM)$¢Daz#Ñ4Q÷[<ï'€æÜ¦C­¡õâU·s}ÖÐîΛñFít›/zgB7í¹â§[„ÐÓXª[Þú t+ÿ„ÿC×_Uù½ÿû0ÝÝ" !)%àyÒÝ-b ¨ ¢€Eª(‚€¨ !%ˆ€tK J·ˆtw úçó›¹ßßG÷£sf?Ù3{Ö±^¯cöµ®5å^I/«£Ï„.ç£q¢ÐèËOaRuÍ4^ë Lk2›‡ÒÁ¬¿¿¯hMÌñey KǦnlŸu„åwjþua•c£ïD6!¬\ôª|6O('q¤À–S«ÕŽ쨧úVN»Wßr¶ª<€¿CïæŒ9” ¾Ù‡‰´»U¾–{#ñ,eñZp5’¦qÑþ<äe4ÁœmHE¦¿F”iâsN8"Ò?±š•¢…Œ69z2uÈLh.´Ž¬TÚœÙÿ‘ýŒy§Ó}Uääø}yçúäæ|FžX!Œ|RÍ%ÍÃÈÿ·Ë\÷ô+£n–ç;4è{LüŒ¨]­ÊöÐD’|þ¶\š–­»È¢Ù;N“ðßÂhnÙù\7±ÍÿÕO%¼A‹÷]CÅ«Ðòعêç›hÙ@)HB,‚VFþëY U}ZQø—´>V>~k­É”or²ch‘UxžVšÜa¥£GÞà)™kHP1YÎÓµ=|“?õAëséÍ…ëPßàëÏ Æ]Õ'*sªúÅ’\NJâ¿%  4^üeé*=îëž?/ý·y& kEH ÏcIq'Î ãC·%;ì÷ÏÜëÁ“ >¿·2ƒÿüÐ}s`ú×=èx^­Ày z’C¥&Ã@‡$}\' "‘&ý‰—‚m0^K?¹e.S*¶ ¯Føa:™>G1<fÈ®:ë¯Áô}ìáÑI˜áîl,Ýx3ç*:%`V×¥¥aNŸé™ ³#Ì?”Sv9 ©[iJoa‰à4-,5-þú«ñ– Mß<¾Z+U‚úç^ÛÂêO£»æÙ°Î)œEô¨6|6YVoÃæüÚp(ÙMØö}Ýã8¹Ø ¹/ÂîPe,¯% üm~«cP°‡7©Íž }@ÂWìJߦgèÛ©Û»ŽHÂ<ú*Ò“I}¥Ú« Ù.ÝIÆùP¤xþäÝÏ9¤Ò·å^ÖGNÞÞbY¤#r¢¼6ï‰ „,”º‘‘êIWƒC2ñq0dßGfM=íWlüÈâ£îñ ad­ÿ¬ùˆdÙ¥¢T²J]‘#ûóe’?eÈ©yzÜ£TZ…ª<ÛEî«””ïX!ÏŽ¢ß­YäÓ£àú)©ü’òM÷²ËQ@°•þoN8 >¯É˜ÈâC!þ4uzÙþ¶[œ”"ž2óß›Pt(ÀùÔ_K§?–¢þí6Qeô¾,+†‘m„UçPr­O›ùÝ ”²4‘KæfB‚Fãîõ[(ÓGtM3ye+™ù¿Ç£ÅIås„(W6å÷3?å+‘nˆµ¢âÁ´oaÞPé}­ÿ]ÅT6x9”ת¬à¾H(‡@ð¥äñ™Kˆ,ÆOª :Y¾‹—Š*jŸ»¿wT5½'6†£VüÍŸ¡Ó¨=ž{OY…už½ëò¢¢@]™g_ë£nVÌMéߨgFæÂCyõ~p±JŸºúZ‡K‰(^¢þ§¼ë§uЀÉcgÓÿ¸m–ÔT‰ AE†_¾¼5Rí‘U@šGL7ë£áÓSR#ŠÑ°ÞÁâ±m=þn(}ž+ŠbÁçr?#§o×gÍç´ðçÞì’§á>1÷ iš|·Ó‘¾’iÓÑûAéíá‚jÎ(&ô/ ]¾_Ú‡Ê{ANyzaCÖÈÖ¼²+Vç™ç·KZµ £‡ê ï9ÃÿfÖË–½¶äïãfŽuÝòÿñZâKí`W •`%Á•ñªÛN²ŸÔ`…š¥šl9¨ -z¢ôæ 4D|4ÎñÚ…¦¶‡j«îÝÐ\I*tnC¾g=i•̾­B6S«Éкz™K­L Úv]VäeÈ¡ƒQj³´`:u.ˆk2@WQÈàÜm-è‘ ¿'ؽ£]’~nÐÿÛcÀ&ñ3üTþœC>.¿F?®¸¹©ÁoϤë!\Þ0Jðcõ–%Œ¥Ùúø&ÎôäZwëÂÔá?ƒ·øL`†äw‰ÍˆÌÎ4>uÌ3Fu¿øaIÁÃõ2ù,?jÒï+1€•?‰”÷=aͯè¿C>l0©Èº©_‚Íô+¶N?ŒaÛüSMGºüaŸ3™_‚Ýe‘OÏeàïâùãžo‘à‘\#aF²Ðž6  /¨òw·HEûÙW„Ô3à‹oŸE(r öøôÇ#¯Â—èÔÃäÈ÷æeCI/ò_;úPó— ¨2öŒž'GÁæ BñБýžøÊ…œ_'¾÷…ÔÕ,8PÄñV ]Š14${¢xÊyñ‚(jO¬_“€Rü‚‚Ü4Q:áÇ$‘Ê8ÊÌ<¸×”k‰GÏ«HIÈ <ïÏvÆ$)Tä`SþÞ±‰ÇôG~™÷ ¡Jüõ„µô3x|UÎõ¥¢ ªQÞT8¢ƒêä·ç…î=C Ïì=¿{¨Qî'Ù¶ÖÜÞï«Úöoœ ­˜ø½UÏHË{#›^ŸšR6–D† àm–-Ù-¨ Kሷ9õQÞJY‰Ï Å’2NœÔh¿ïu½£ä» í-Îîït¡c»$wõl;t óüêƒîKϺÇïA±)穾ÿî)ì¼ð㮩3tqZóEõG@'£ñvS5tèSúH˜êBûØ,<Ž¥†ºCn/ŸC‡}Þ±ÆJè(\Šªéµ€Nµ =QbO ‹¨Is1àtSÿ::©zÜí¹ßþ²†>¥¥ÄSw(a@ûêùÝ$ø™m¯GÄ{†þD{¶iÁïÒ£iµÏ‡aôZ¾Ã$ÕK?k±ua& &ßl0µoÜ…öïõý†0×=¶]ø~¿ûÌLÀò“÷œ¼±œ°êÌgø>ª Ö]+éxKnÂæ³Þ fíØŸñ‹Éë†Ý‰…ù!øwó\ؽû-Hh§lß‘ˆÄì÷ãò÷dþ¦y)’[óçû"%U [Þ¤vw¼QÙŒ´ÃžÌO¿FÙ‰µ{ÈX^ën—8€Ì&Љü‘e×FÇ#pÙZ |Xw£¦ãÍ«óíÈù³ìQ϶"r3^¸£¿xyϯudíô#ߢ}¾àCÐN^Ùo@Á ¸±ú…Îo«÷=2ÀÃÇÙmþ¬W¡¨¥x‘Š_áË.E‰*ï^‚P”fZ÷›&Cë¾e åBÙŠxT½À‰rvóÌ{¨À°wùû*þÜ;Áó½Õ8ímˆî Êjï5~V3„û_Cgó£Ż”ðÝbT÷ßàãzŠšÔÆ ¨Ðj“©Ñ‡:ä3?© nÉoÚ'"¨ÿ^~ª„Y _çuyäKµ‰\šGH}6 Ù¬õ؋̮{°>P·ïï¾™¬œy”uìêŸþïþÃS±!“äþÿßèbB#Ǹð³³E0æ$”ñ?{³œÞeÄ_Ó¼U„²ug1B7(ÿœß %yŽåÙ@eó²ä¯üý}ࣨê_;¨Nåyèw<¾î¹+»¨µ— ÒFïû¡ü‰³.vç ‘ýOmº |»v–¦2á |?L{òöÄ0´Fª öæ…vÉ«3RÔQÐ1¢æö) ºÚ(oÿÍÕ€žŸŽÞqöÐâkûéqøIê½ÂëI ¿†ïÅO/åÂï«¡cá׿a”ɲ§5¥ ƺ2]ýäÄa¢4ÚÀÒ²¦Êid<˜YÑö}y\æõ÷Þðõ€ÅN¹Ë/u^²þ6u™T/¬TÊ](Ü~ k¶rì+dJ°A,*û–îlö¸©Ÿ8Û½§…Äa—RAàDµ(üõuXž"‘F‚çXw`'‰èeÏð qFÁ‡p†WHêö|Æö’Ÿ1Íûº€”±Ï¿~~ûi^I–Fº=Ë–íd$íOxô™¾ fD{É"Kò ¨Ã=d›®ìÿØ@Š‚5ZÞÍ —+}Ôû?FÈû„EÕZëò‡øwwøù àL»í4(”Í_ŸRƒ"WV›ÖWPlUQëL+J,ýø¯b€ÒaõcPV£~€„çÊÁóï$>¨ð´õ–uò*m\·Ð` DÓ˜4y3.<ž®ó§ŒÀÕD> :›¡zñë# ò¨é&Õz£ µÅž¹Þ&x:³¾—gùWP¯ðÙÐ&4¸m·ÛÁZ‹FÒv½þÇѸ½¢0…˜MÏÝËËÔ¬@3O— 9Ï4EFq.*4ï>_åÞ‹>ÙáÂÊ?Ð’™ñŽc´-Z&¸·_Í_A+Õìß>h•ôÄé{Õ5´fïŒ,é{ÖZ ¢Œ'ð‘èÕR!W*4Óžô²>™‡<9'ר¨ ˜^z:ät«3 gçA+ÉJukÑþz´ºh»6_¿;ø~ i…*§×ê²@};í¿ ûœ¼ëáò¶‚ÂBm¥j‚ÿ¸Ít¯x- ¹U¾{^á-ûœÕÝ>ù*]ïÎ9Êß…º‘?&ŽqVð>viÌö ´WÕˆßxºݯ‚U‹žp@8‡ic$ }«—3´O†ß³™2==a¬0þÙ0ÉLÒ·¥;®KÀ”'Û%P€©q¡ÃxîLm|Ù5˜V8ÙM Ó/gLrE `Æ¡¥Ê¯`fož–_Ú2†¹ô”#Êûžøõ´ç´èsX¤»jàm KzÃ<¾°|äÕ„.ÉX'÷O+UPßIi¹k×>-öøHÂz­Ù¢W½5lJœZøÊ[Y­´#ïÃŽf¡Ê8ì¤4ȧÂÞüè& \E²7“É$iMø^„DÎD¼‡‘8Y·¡ÛxIiMW¢+¬,œíiÎΤòa ÝŽBÊŽ¹–N¤Ž3T\•Í@Ú`>c9¼H̳H|wš³JEû ãï¯ím„ÈLô¢ª`ÊYä„Üb¾é ëí—ß*ï‘íç@Ãjt%rØÍVÙ¨µâ‰´”¼Øyá¡¶€ÕçÅ×QøZWýçFN¡ ^øä…¢^ù å(–Ôpaã Š·}´{á*Œ‡Ý[Æ&ÔQ2LàÊn#Jñvó)‰Ç£ôÐ+³;Ñ©(Óä¿'~û Ê63Üu[ü‚GG쮕=;Œr[cfÇ÷sTÈârv*ÚT´9φ RSÝž®*ß¶¼ÏÞŒªü§Ïè^F]Rd&7Dý{—î”¶£Ái+š«g=ö½0áù7n:4dý%ö Ï]Ó0僆Í5µ•¢á|äòWÒe4eûš{bvSóÏÆ£^U†íÏ¿ÈIbŠ|NþðT°šD@û/1‰›ùÐ8»¤a­ Õ>´š½*NûùÞâúÎ Š$JÞìÐ]/DªnPBΠ7bÞðÉíéªD_#|ô+jk³„ bó†Á]·ÿff-s¨˜¿#|î8÷éÈP?úè]¿ÏõØ _ Tm¦RíÌÔÂ×ñ ÿÜÞ9¨ Ë»§¡þ6ãñÛ£½Ð˜ìð œàÛCš‰Ù9h¹Ô@ÿº1~ªúÑ@«yÀ\‹-´Ùû’©®˜@ûÅ“Eß{¡#¼WÙï§9tΞÑ-³ð†n‹òjçÔnè™6lÁé‚—÷!é£Èf ùkH44eÎúœYªråEö?~Z”"iE̯#nÿ¿ó‹í¦Nû½ÕKéÝÇ}Îß]HìøEëùÁo׊ö¿Pæ©Û ”Ø|ç9}~?+Ô­Zÿß}7&çoýß¼Ý}+°?v¿ÈøC2›¥+þIÍ÷=QâåÑv(çÈâ8¸$dàÞc•¦K×Véö÷™ß'èb¡Z8ƒaKŒ¾^2ËXP®€šoç?%)/@]PÓ›Ð,hpHì3JVƒ¦TµGU´ Ðb¬RZÜ4­l/ܲiÚ’èBM}†S“Œ{5Ð%·ll&R=J®Û­Ÿˆ /‘}&º¶ •yïžÍz C{! Üîï`x#•’þSŒDh.²]1k;­Ó» 0¡*§MvV¦Ž9+‘†\™³”~d_j`îsWWƒ… ,ʽ˜=^ç K럟wîïc+®[šœ°ºñ·DÐ7Ößl‹Å(ÃæÕ$ ‰BؾÊ2¨üÄþ$üÕk†¿l›K"Hp&—Oü6ö^ç×c­Cb7—HÑ×HÊP˜ ÒHNãN…”’¿£œƒê‘:9/pÎßé®XØD"C¡Ñîï(drR ~U} YÔÃùßÔ#›‡Ôðúïȱœ$©x0÷)o¼¢'ò†24"?äÉSÏo Àvuç;ºk($hÕôZãŽñÔò6éA1Ó/cm (±§çQZfÖ¤·,eF¶8ïÜ£=¿Ö?+F®vóƒí+¨t3z}²„•k+’2ðøÉ—+­KŸ… ÿZAu·«ÂØQóÀÓamÝ'¨ÕqíÐݨ“¨1Ô:Td‚z'™¼Ò~† ¯ð6_ìC4ìø ŸÙ}oÞ lk ESj*!¢7h¦“]ÔGòÍþ¦…ÑE+£y&iDêw´0¬&£?ˆs²«wo*¡åívÿK¥hEüÃ16Ë ­ü¯’ж8 Õú N)« õéÀ÷±øàUèTÌ#'4 |Ó#n󹇖 G’äÀ@!1OQºN+~~<ß§/•°óA]‚lÔ¹5!ø*åë÷íT:TŽR…>Þ_ß…¶µ ÙPü›ñ“”T>Œ™äq·þÇmN·ü15:=Èe$­^¸ò·‰E˜K¡"/ºé" ÔÑs2p«AsÏ«ú¡ïùÐ6_Ç×Fâ ]a÷õE?>C¢ðÕØãðÓâÁÉ÷>¾ð›–Q¤ºŠF?ʨÔÆÂø˜á e˜Dn­é»0ùµeûn'Ln:¼ é6€)[©ÅE_˜j‘Ò¼Ó ÓÑ·… 1f’?õ?þ 0ÛÂÉÁ$sCççE9Ã`‹síaû{X ‹/š/s‚¥ø1i¢WÞ°=[T\È+qŠ–ú°ZLæã…xXgÈhé|- nM»»rM°9J¢wœnݧº”ÙŸ?¢$_<8Œ`^,¥¥fþIØ2ðZ>E‚ÂP.ëÜGH8ëü“B~ ‰2¶8Ek$žÀÿÐ$ੲ"¯Fò/.¡J+HéøE/LÉ©ygÔeBZò"ïf¤§`¹ºæ¾‚ O;ÇbZ® câÕèU›fdªš<Þ]\ˆÌ‹„^ üç‘UæêÃcžVÈöä’;r ³§áGË7}ò³(›-äkr}Hx-Ö‹<âš3]1È[äøÆ ý<ò•"Ìfr vt5KM, RÞ»BÒxÈÿä.õ9fÍê,ŸŒÆÃ©ÝÎtyP”jf\z\ Å”mÅ2ŒâQÜUWùëß00»úá.dYà™ðȈð©ë28éR/×c÷yúÿÍË'ØOÎÁgå[<‘“¾ÿñZ¤m|¯µá”%ò;QЈïçÔêßWQ7 z-ÏÝ0j"Êuœi¡ÎÕ[B×"|¤¬™B“˥ߡù0±)ÅÈ´|åÈa¤û?NÉûÒõP@káSN­Ühh«zô9¼ ÚûtWšoFî÷Õã9OEÒ ëê†rD;t/ÌœŽpò„Þ‚èuÉ´$èo"V=ÎV?…U—Ÿ?l_”zgxêá·ç[{“›É©s€1ÿUîOS60!|§äÅ»˜œJ›LQ=ÓEÅÄ’çö9wÉ[0j…–?i^£-°dé`Öï–ËùšÁ)w`U¼côã«EX+h‹ýßsÃИID6ç7æNN2Àvâ©ð•úLøsÕ"ê·Eìܢݺúþ]oN’ÏÜï¥}+܉hÚë/MÅ"±Õu) …$$ùD*–-¾d‡èFŸ¸!y ÏeïMo¤ôtK¬"žBj㎎íÁ£Hk®+FºÆƒôL=.Nh CõÜï<·,dlzA«6°ÏsÏ Jë’ÈBúøóÛÛÝÈê 1ÞLÒÈ~Ü·«Óü °~uB†!¾;بöp ¹ÇŸ_£?[м#S í†rÈ×`öáÕn1ò¿¼Ê0¥„!¡®µ²(ø¬§{G uY$•œü‹Â'μá"ü6…ÖRQŒ3mŒøŠ+ tyø-%É'@ê=J½]!üˆÒõd稌PfÓ¢[:vÚGÔ-O¡Ü?ë-Iï~½¹ôì äF^~¡éª92þws_^…lû-5‡Ôè}9_˜ì÷TýŽOŒ¥ðA•U襾dÖ²]ñŸ¡ûß{m®¼S„*Á²ÅY…PÏøÇÀÍ’šcR™‹Ò% õíz»ÅƒTh7ÿ"RÉx: ­wÐCgKowøÁ},w>¦øºK£®œûÝÉÄ&]wÌþwŸph|é t±xÈo2@§HØdÑcèðÔ4>í) ê ëÛ¡ãÀlñigèкÇãö:nNå:èÔ‹Ü!3€N?såQW.è:¿x±×ýt§/¹d¬•C¯¥eÖÎè—öêHODt>£"MCmZ"^W`¸C“ᨠŒxQH^_€1³…½¡ ˜¸øoK×o¿gº«äE *1ƒÇ$ù*\^ £Š™ùÑÓ¯áñšßš—åÔQM´æ OãYT¿ü§vQâ9j|4ÔìÙÏÍJܶ¨íÿôè˜÷¨{ÔqøLÉê³(œI»r ).Üâ·s@£µ&«HaE4É;Qøý&šÉ.PêK£9I!#Ùp0š¿Èºå…aÝBw_¢¥Æröò©5´œ÷mw<•‹Va%^|Ö‚$º¯Ð:ëÌ=¹O´‘êÖøäzmRšvE4Ñ–Í¿íõo´½xíU-z‰=T°pžGÕ¦kääq²H×U·Ëwƞ݉å~Û _ÇG‚(nBñ©ø¿GÇ+œŸ ÙS‡t³_C– ¬MdKÄÿrϧâ^+¤^{ßýj‰yŸ7Ú§¯ç?@bïõs“ Î'IÅ”ï'l— ‰üöè@|ÔÿÎ)žÊíã…O¯£oþm‡\ß¡´Ò¡€á^ñ²Û[(zÜdõ{v?^…¢!(žûÆÌô¿s‰¬[¡$zèB á³”ç‹àO”×à8â»ï‹=™b‚—¡´çóÔùcPv/?<|­Ê-ƒ5­;AÅè«ßçZ 2‹µçÏ T%+¾ºÁ Õ=rÜŒ4 Ps$ Gyxj;Uº²‰Ë >Îpýq¼4óUݲ€æë—¸¶²úá‡üøŠ‡V´È«ýí% Øhú:_öq¿' ÝQžwˆˆÒ¡w΀I4 ®4P¥ÌŠÃgØ¢ÙL( ß©•_°…–ìlÚ`>í"æ×«uñ’N,ò|“y$N½ƒéŽ¿’Ö0w ‹þª‡*,Dÿ¹4ªcµÏï%Þ™ÎmX¡ûôàæ“Xit²Zh‡u…F§;İIæ–ÄÉ.Û¤MË«>ðGõ‚ŒÛ«°÷¹k{ ¤…}U÷> áûªh*õ«H,Á|§FXI:¯†¬_yŽd»Nı #ÅW‰³ô—7‘ZDþ•œ€Òþ2§¦~'‚ eÄmù Y¸CC&2ÿ”âd¡GÖ­w“ú¶/‘É4b˜ÀÐd¼»þj?/)k›"ïyÿïÙO(p£×L“¸Åd'æ¼oŽ(|9/úŸ6ªs¿,¦(ßBºù%o[ÔºmòÛ_˜u4åÛÍ2P÷_õáÈÔ¯Z[¿Âk„†Þ«ÅýgJјKÊd7ž M>G[¾©º…fœïÄØ£Y' Å Íe4L:wƒú+Zˆº,«ÚÍ E­ÕÅ8{´´U”Ñåy€–¿¼ïù9¡•“yšQŸZuü›g‰ý…ÖjÕÖOþ båVòŒ4ò»T›±FˆÜÔ!wJ€ èØîN~ƒNÿsñ9ç< ¥aA±3ö=Ž¡Ìcõ*TG¼º=©•®·ýÎø eüïGøø¼ÿÛœ»Ÿöü/CŽsÄà ×ÎC¾Ðñ`Ѧy(/eO µ*j«_‹Á·¸—ïút,ö×sÛŸ®ýï÷³.zK½ds­ÜwŒ` O7hó¥ Kz5öÒçÃH\Õø'7S+¬uŸ8¨ 4‚ý†1—a"!¢<÷•LlÆ)˜úÁäõŸìWaŠÐǽ£=eßÅ­Õ{ï¯ë¿÷„«÷ó1ŒÞx†f·Ý\/Rüh½S½ä,,<òW˜8°Ä*éS.iK,×À DÉßšX宾îoÇk'É>Ê]%ƒõ Ôò†«°É®ä‡“oaëõËë‡n Áމð×;˰+íanh6uã{>!žæSΗvHèw:ŒÎ¨‰;‘¦ ‰ì û£‡ûy&¼ºÎòɵÍÿS˜#ÅäØÉ³sH•–À+øiB“,¿©ø"ÝcÊ¥ñ“sÈ`¶5û—ì,2º´e©‰Lþ EëaÄÈœöóvæÍ`d™0I–FŽlÊn£w´==ÍÙ}«iH;¢l燜nš$ºƒO‘KÃCë¥'rm<؈¼&$Æ‘m¦È¨¹ÍŸ¡…üÞ›y[âPàëÉ̃ßñŒ‚]B£ ?øÝLF‹‡µD^IÕ£È{Þï잣èñ9gÚ£(¶;ÂµØÆGÌR΋Oœ@‰²¼ê¬3(Y8¾é¤W‚ұǬ›:ÈQ&ÓŸíÊŽQ+¦ß°A9™ŠNçlr”¿2°µ—, !²´6¥¨Ø<û“¼û4ÓÍ4ÿs¾•5ž«þY€ª¯'õô:zn»Hº[ª¡šôZá õVT— Ú”?‹&þv·z;Qó…ÇzM×;ÔÚ¤ QjD› H`±ß/Ňys¢nÝŠ|’ú}Ôs3ùÔ‹ú¤Ëaª¥T¨Íñx)å&°ë0É“ä A«]~ê$üã:uX Ïmé÷t¢aµÎ£ˆÚGhÄnSteò1¹õ{2˜€F;Þ7–œÐh.챓ú9q³&µ7ãÌdcðûÙâˆØØ9érŠÜì)´yÒis‚úåÊ6 íh¨rz›s?JŠŠç·r’…  Dêl¯ÞÙÿ÷ÿü£±£ÝË¥)™ŠKSCׯ@zò±’òkæ0,`wäãó#÷»„A¥ÛýþqØÖ÷¼Š-DÒ|/€RÇ_‹ç5 R W”T… šØó‘˜|zak¬^µÜŒUPoLô«B ‘ƒ‹¶¢ç24ÂÓ»œÐ|ÏD@;á|?èÂÙ|ZÙ$¼÷ùΚ»[íÜŒÐaz€Ûéï t†IäÎi®µG)U¦Гð!€sßwûž}$´ý õAËbì,0d=íC˘ÃO*V ­ía„X·%×µFŸÝo\v„q¡òŒÃN0Ѷ÷ã¢L½—Méš?3“‰ˆS`nš‚­Æø,^ÓMýZ ËÌQÕUlû\þæó§kòžÀN–kå%iتÜP9a“îZ0ùë!Ç+{‘}žji ß"Ñ£!‹*’À÷˜(ÑÃHšY³!} ÉUoùZ²#Ål-óñSIHU©Þùö-!Òüž87Œte”Ÿb5‘áôðØ):ƒ}O9ñ˜"m™NÉõ— dâË–1A–劘UdËZ»ðÀ59*¨õN.ÞCν‚jÁáCÈ}œÔñ+ò*~5»÷²ù¼NÉ×RŸF~·™¢m( p8‰>æ jÈ‘o:ã¡ðüã/ ï£0ÿÊ×GäÂ(B¬¹5Ò…b¼±Õ—NàóÃ.ŠA(©ÙÕSè‡Rƒ'<® Ï£tBDý}”Ù+ô?û4Z•D¥Zi¢Ü"µê¯MTh6ÓžíË@¥ÑïJBT¡ö\¸]gŠÇͪî$ìö">ÿ‘Z‚jS'¯h?wC ÖËŸ1D£ÆXÁáfÊ—¨™¾§ëñ:µÎûQ¾=Úluý­Ò¨]òoIñÕMÔ±^,× ª@q}Ź·¨{qùf"ÕêŽMœrò!D=Ñ%ã ¨Wî¨@ˆú<:Lo.$¡þ ÑêÅÔ¯3üô¸o¾Œ¢”¹Ô€^ýç×,8ÿÏU.µ4ÉBÝ·3gƬf¡Òºý¢ýb(Þz‘ðÇþ&|Ù›W¾ýrw w“ g¾pàÐÀç'­d±åEðÉ-bÉd ]ê={Jt$¿p›”ì„d*íK†žð±Ú8Ý éä;“‹¨]b‡Ê…FWr¨k;Þa(6ÍJ7MvßBëyÒ¹›*ªÐÎü+$ê tØç°£Ê‚Î÷/”,.3C×xøP±1t¿R1åpÕ‚îOß»¥B×¹ 2«YÐÅxOíã‡Vè”{W!Ê~¯þç…;⢺ÿ{_Íi³¯#СèÄßP9¶_eµ–æ #]éZ¦—(t*Fç;žpÛ÷Ì›—£Úö=T¿äQÎ?OèFÕî¢З¼yfl#  Ï·-Tƒ!¢™TU‚vjúzqz ~o¥ YˆÄÂh}iRïo š%,ó…)ŠRoºËaæ~¡D£1ÌxEz–Ä·J_ \„åÅ|Ò&šX|0h»åëk×%µÎ|†-îùÊ›a'pÑ=ÜöŽîˆÍ.  ²e¦'Ú¼ ¨FbK“ÊWQHÊý¶-÷x4’äίv DJÃ&s®F¤Î f°ºhŒtFa‘^ú"ÃÓ-Š"1d¢_T§¹´ŸŸùߨÈ‘ÕQÒaý ²Ëšü±dmÆâ_=6>ÆàA£©ã‚gc‘;Nƒhñ~:ò1&ÕÞG~wßÌ.4zЇ¼Ý6û¹Pøp³oŒ¬Ó>_‘ô(FWv¯i» 0û7R”Œ4¼òãJ?¼7sõ°Ê,´H•]£Û¦ÖP^‰“ü¦ "ÿ¶¤£'*Æs[–¢òß°ç„<÷ñøù%Þ å6ĉ”#ͧ® º”Rfw>j\žeàb&CÍ27‚ä”Ô†Sã)ƒ.¨³Èef™‡z-ùqñ¯¢Ayµ¾¸¥}M¾å‰&.O.½ˆF3*Ê“ÃÕ±hÖ·Ø@<æï}$d“Щ÷ñ#´d"ú÷5©-‹S©.Z£•ÝJ}E ZÍÕÞÒÐàFkï§ï¢ýÑzã‘{ü%?´qçáÎVG›þ‹áêh‹wr˜¡gÇê½¥Q¹Qÿ³@0Ò³¥]+†1YfÚ“^ÊðÕîCååB/(J<•^Škûº®vúdó‰Þ)&H†LŠ‘—¶7t÷sO©3êÙ¤¼™~*àp’ï –Ü×ø‰$eí—ê7à}û²žcŒ7ÄOÌÓe„A♥½W—žÑÌ_ ŸèÜTí–_A®œ>¥]*ä眦EÒ§PÄ¿HýƒŠ¥C?.’ CqÀ„úì×÷PÜ¥A3}÷”ÌÔÒ‡þ߬Œú¡u–JF‡ ÄÝÍþß¹E¥‰à}_¬ß®¥d«=÷ ,q¿~1Ÿ…ŠÃÃï@%d=“â*Áþb ¨Öbzôg| {÷·-dß3éºN\„º‘5Ʀ×,ÿ{Î4°jô |{õòŽBd7|‡·Ô×v¡5ܰæ¡]ä‰ô$Ÿ,tìVί2'íï wŽÚX@¯|%kxX$ô'^‹™;\?Ÿ³¹Þ€a’  øýü—øj;ŒZeþ•zJã*^Ênþ³0)•%P𦵞ù~˜€Yÿ}¶˜_wšyX' KÆé„)¡°üež­»[V¢ CÎòÀZë+΋ͰñV^Ì›¶ÞÜJ³‡nB¥ FØ3Ž"*l`‡ަ¾ùë„„—ÈšÝ9h¾ê¢Óg>$yxÒZãÝ’YpH§Th"…­l¹¥0R¥çgù}AZ3ÓŠ%/dà$±üü̳¦r“œfùVŸHä5[d½O;ÿ"î4²7_ñïFN£ÇŠo>U"7cPèƒÖ‹ÈÇýÀõf£;ò/Ï›ÜVÂC”Å­:†(|áÖx A<ŠRä e±½Cñ±ÖÇ¡äZJ×ño”´ËÖUUÀ£T^Z÷^ìsÚM”*Á‚Š>½Ç’ã±Æç®½¨²ñ¬QÍ!Bû1Ï TûÁ~?u©?wŒþýK@-!‰w…í¨=[;Eº‰Ìï“õ® þ¹êå’$z4äT?B¶°‹F•“œ‡¯ ‰O"Ë}+4zj e&†fiwƒº‚Þ ¹½¦’KZª%èºn¡Å»¿*U·ÐRæõpN»Z–pÇ]Ñ3C+9“ÇãÒh•ëö'­ù)a“ƒ|ªÔo}'EƒëJyGž#—ÒA¨žqø—FÍ%ÝïŸ:æ'æTψs{¨õ[˜¼»R UË…w‰OBEO¿ô?‹a( šçšiÙçfó”UÂ-ÈWýÁÙåù·Ÿk„‹×žÂçìƒÆ#Î9ðÅ܇R<€ÊîUgÖ©2Aë™ÍØ ÉõÅŽ²hÕn¢~°¨7¼ÏKö‚îê]g{èw&Ï4.8CqÙ]ý¥Ÿáw”Eø >R CÓtI+담дƒqP÷Ü<ßJ‰oøÙøΤ|åÞ¹ÃûyDÚ» cªÓDÆ‚U)<0ÃD@°Tµ¿¾Y<~§îÞ9Åë¤`>øtîÐ:/,2-‡¬†ØÀÒ…<1†Î X>-ø&]-V¼ ¿ØèÂjÊéÐh!X›}àÊY.Ø0tû<)›õï?~aÛ×õZ«>‡?æ…”fm5°çö➀C+üû´ý¯|Ç ©Ïg9„D'¼lÞÿ~„ĵ \'Þ#©ž”p ÿ6’uÁHúl¤ðd¾vìÞM¤’l‰r³¿oÒžw‹&A:Ú݆ÈÀ̵ê‚vÈ(\@Aé2©…(°œ2EæsEA…-ÉÈòZ¢yúW1²N}¹VÍÝ„ìæ±5 TÈÑzᎢ&ròt­P¬_ƒCù\ô$rŸ4³—‘ÜEž‘ô¿ý[ÇÏ@/×6ÍùŸðëNB;ÊÒsŽ(ø›ZÔl…N9/ôz€Â¿|ÛM_{ ˆ¢å%ÂfuñÊbi©E±ÈÛÿ¸È8P|\KÖôü9”°Þ›+<ž’Áî#˶z(}‹ÒÎÃSeâG êQv‰"Z_IåN\‘Å›(ŸÖÓâw Xé›5†P‰‚pdÀð;»»óÂ×ûª0ZŸSËDÕ†&•RN:„ä¸tÚ­Ÿ¨¦AûcŒî4ª«˜„ß:z 5† 6]œ÷{gm²Ó¢ j3O•JY²¢ÎíêYµçQWô©­ó7ÔÝÊ~,‰z‰)6l%P_ã¹¶Îf"êwUS \âCÇä b4è6íàYFC㵊hX(8$]F¾J}ñÚht­qíoS U%y¹° 1µ¤Ò—çhl¢Ï.÷µåÜ9nP Ö£È?uÁjy¶ÿ(Ž~º­%þy‚§ n-½æx%T¶¹LwVP@á·tv"ös뢮®‹ä2’¸.l€lñé)¶wO÷ýïýh~¥:¤mg/Ô܃ԇ‡MYkZþ›úcRò,ÔMñnÏi«â?^ “ôº4 ¸1ötw¥”w±ËlÌ€*&i­4Kv¨Žy¼d%xj@fPÚ êøº/ñJVBý°`öÃ1hŒóie5o*‘¾ –ÐÜ]òæh9|Ÿv JuÓ†VÑËŸ™?A›Ûá³îY ýþÚC©.ºýê"þX±:wDæWÔ} ;¬žìÓ è=aUç©wú/4/•'ÛÃ`kcZ!Iüònzz«à9üæÿÀ’G)#ïéR±Cÿb ¸/Àx³'ófL$L>Ì MÍXƒéЋܬÅwa¶Œ%ˆ¹  Žõ~ÊW= KׯüƒÂòÀ‚£b¯¬:ž0Pb=ksºYµþ ° Ì‚3°u4êhÌûK°=û–Œ?üiz4tWcöº¤þ-Zå ² 6z"¡REAVù}âÛ^îDâï/‰¼…O ©¢·ãÐE$+k$y«)\¸ŠJN#•ì§.uÇHÃs;AÆ0ékø×M´ s’òÓ>×K\­·D‘q’ÚsmK™‰¾d•wÔ#ËEµwÛ Ïýò×È!rGñáw?ät´,!W4C®è†ÞñžÈóä]Ä‘ {äãs¯Yâ¤@¾µfýõÊKÈ?^3üÆu6cž±VÁCº¯+¢ÐLà±—]xx\.ñ*!Šz,èÞø|;7ÇQÒ?­Q66¥i¯y0 ô¬êoÖø<”%á9€GM&Nyä¿C¹¡ï<†úʨK§/=‰JùêögP¹¯<ýÐ*ç|-õ!šÑefé]Š$ªýÜÕÜY¡B v·¹%ÔX¾áᣀš_©+è»QëÑ«—)"‰¨­õvuj(µ×DýóŽFÓŽ—-QWúåŠóîÔ-g4è©oD=uÂr:Ô+6V#Ÿ•C}Ñ!ùiÔ° .ôø‚ú *q®F¡h e9>¤²ƒ7Ьµd\ôPmË÷ê:åãÿüpzˆváöó«Pü2×çµ bôÅ _(>@!æ_ŽÍ?kO‡\ÃÎ;³ÿ]o”UûE #-% ’Ò{Óݨ(ˆ¢‚HØbƒ‚"JJ—´t*!)ÒÝ4¼<¿ùŸwæ›÷¯=gîsßgΙkíµÖ¹ö½÷µÃ®KÊ%ÎØ©cœÁ~ÃDÌ'yj$Ì âê¿%Ê<ø,BºW >u\OÞM€/ßî0|QÞõ™«‚¬@qnKå™T†øž&L‚ºN­G:ÆÔÐ(ÆLún¦ šÚö9ÞûÍü™+Õ ¡åú…åzmJhÍ=}’¸¾Ú®8–^(úg~ÌâòhíÊÑ,vƒVš ýS¯B Eå¼ikœ94¿óæyõŠšE^ÚÛ’îþ^¹Š`Ì?ÇånLRÿøÄ‚+Ÿj ù¦‘ý94/vH¶ëçCKÉÎ%cY%h{”G!í—l¯lßÔ€_ì·¸t;á7­Ã»Š/tË<"º~z[>O«žFè÷2̼ޭƒðä»ÐO–ÛÛ¿óFm)S´¾Ãx®–ÍŽLÌ(ö–ƒY—¶½“0ï>:ã/S‹Ž×y˜Î„eç˜×% °r¦=£Ù Ö:üýËwÃæµ¶Ì¼ZØ1l±ÛlÇ=c:ï\m°ÖpÕ<˜‰ïevGÒëãê1Ê:HþŽÍè­â5¤X½§ÄúñÜ!G…u¤=”ñìWÖ[¤‹Î/»»/äÛV”ºr‘qàò`ê+dNä­Y¯EÖwl߬”³‘-›þçÅÞ<@v¼ð˜õ,r=õœ¦D~—¨WžºÈû6ÿmZÏM<(2訿ÎN’sP0´XtX1½©ýø…þŠÙûÙ¦Õ eñdumJhP=oéVEÉxù ”VœÑ:`Š2³õŠ÷ÒEQ6§ã—tà:Ê¿ðm¥2Ç£‰5>ò»ÔVR®9®*hõþ} ÿZZ/íæ=Wo»êüÖ]ß[Íw\Ùy )§ï„—èÂàÏ8ßwzqPZç¼Y¿9£œ—,\†!ýiÄùÉ·œzÒÑâãIø2󼓃c{—÷VÅ?SBL¤fB%|.û2æðf¢$ɱ'Ž>2—žIcÓƒH‘Çþç¿ATÑ¡`®6GoeÈù$ûid°“AútÆâ¬ȶP;øÞrr(üJ L 7¯âƒyÿnØsÕ̽ònðH¾¨‡¼©šÔÆ¿1ÿÌ5ÏϱVkƒ ’þW‡ªÿa×7¾R©#ýÇ'Þ±ÎSƒ‚…»/$¡è"eß}ŠÃð-|‡òØ]¾2I¸<ë %©ë;V)üP:“/à$:ó¿:Ûõ&¨Ûûjª”=Úo•ÝîÍðCÛâà‹Eh$ž8þÖ—š^™ðÏ9C³}\‘Æc6hÕ}x×!~ Ú}E¨æ=àWÇ~> =è²Wþ:d½,ÛÓÜí~ÐÏBýHïÌ]h±ùÍlõ†òp=3„‘„VVß-˲¼mÆåãvþ…)ä0mYê,¦E ³óÙNKG`þ¤pŒ,W ,´Ù}þÚ~–œ/Ðû¿‚?Ò[¿ZÖwyUöŠ}I5¬_y«âÓ• ›ã&äæ™û`ç5Û‰â=»|*cw+8´ “¯X¬)®"±’.ßÓ†SH²rœ,Õé’MV^:j…‡åÊÊrªÄÈP­÷½J^eù„tÖj”ñI‚ÈÀxG”âûd"»½8<*Œ,²Õ|C¸?loÙÛÉäPÞ1V·PC.=ÇN²ÊŸÈs™H;p&ùnO7_¯–Dþ¦Û²7mï£Ð9ÉJŽ92Ã8¦ŸbÊ´×Øu DÙ£•“ï^¡Ô§¢‹#u(SÙþûò1~”]_ëëˆKC¡£Á ,¨xn”â&*¯sïuUGß×Úº¼ù¨Æ“¸¯«yÕ멎?ÐDM?Ê\ÔÖz)Ï»5€ºd!3{…~¢^ap„´—œ{?l¦´†Ûq (Æf¹E¡Éq‚œÍh*1@eŸ‚¦¿ØÍÍn¢Ù7–ý\hNEÂù€Í_—Û0©¡+óbí<;Z„ˆh_:!ˆ;X07cþã]—ÆçlPç"iT’X(²;¤’¹¸ÁŽ´íªÐsJøY’ÖÉôêþ™ ›¹åc?舡Ĭ݇sBŠúò/J¹–C¾ôE¹êàߣæ,çxüd9mmŸ—Tû‹ÛÔ7S!u'†á„É6dzgV0‚‚s¶ÜÛY¡ìÅýû‚o FüµAý‘'P?º§²Æ~^p½œ›4­wJL\Õ c¤ªõSë:t Üd%±]‚¾ƒ wšó`àt®äeU{Œ‘Ô³™Ó‚¡S›V‚õahÍ0ê•j7 'ÉÇ=l#ÎäK· aš]j[‡1‰wÕ'i`\æÚ‹°"˜¿q|lÚ&5¹¥ÕG}aÊOM¢=d¦7°9§RÂl´DùÊ—0WºßÛ²Àæk[VDd—`a¡Ài‰ù+,‰P Æzj²O䘟R.üY¼´Õ«¡Î=‹`ýÚv½,lÊeêƒí—A²$xwfšÍÓ÷LtÜ.¼ˆD6NÂZæùHü[ÎVNœIÏ O"Ë>$ÛT»·ïîÍX#Se AÊ¡û'Ê[‘:di¡êhîKð[{‡´c×~[ÇK ÝNkº´ä[dàzXÎWŒ&T³jÌ—éõÖ6é^d^ke{gyYݪS‹™)p“So-&²{R1?Š@Ž¥¶g½jbÈyÃ4•‹¹é /žG¢üGzã‘WŽå™äÆò…Œ¦tØ ?±ÂƈúðüAye‚ }B¥å$qq{ „úNþä÷ž¨b.0*ѹ€ª:îak’·Qí*_ŸŸFõ¢ýkäÛ¨)Q¾Øqé4jE=¨TóDm—^¥ÙÛ¨#ìÉrä3êü& êû†º>ò[¦ÒʨÇrñ~¥h5ê}&þcMÈŠúü_´f_n ~x—™ÑÄ/4 0Ü ŒGƒËIc¼^„hP«,œä…†œL—ŒF¡¡óé ¡Îïh˜JüÈüŒÞ)àL¿¦O€êœê:Å‘Uøó~]!X{KWL£¶›ß—¯[_…Š•gv¾G¦ XCŒGhb¾ðm}™‚¬žOÂÙþ6nôÓÌÜRhÜ[ŽˆÂ"³ž§ÔŒ»º47zÃ/b3/{5Ÿ˜þ<åÍîòj¯Ã¯'„eñšõG"ûÞïï{¢m°ˆí2ò¤¤O²SB1÷Œ„‰W#”„ù>>NB e‡>oÆ|Іò¥ø%gÎu¨Œ17-ƒêãYÔ&b¾P³’K˜\Ëu/ÊOœ`…Ñ/¿Îü„†—ÜÏNXîÞWÖö"- hJoX$ø¢ ?«Â?Í”CóNµ ŸÔ&´Þ¬y·íÜ[ý‚ð‹‚qÎ\:~ëÇwžYy Ý-LòÇÊ= O…øxfB(ô§†+˹ÀàA•'Z]É0Td%ÌU#WùÎz9À˜ÓDÓ!W ˜xÆ=–§¦‰.dgOÂl¸Óí»ãl0/½¶uCR¹E(ÎÁ’Rñ‡Ã©°Üq:6¡Ô V¼’Hœ‚uaMô™³mØ=ØØcVY"ú¶Hú“¾ ±ÃÔñt¶2#A}|¨]î3ܿа‹g»Ó—mùž qÉ¡­Š@'$EÑP k$ë¹¾ Ër÷Fhæ GÊû,vZ+H}þާ"¡&îÃgÍNŸHÖ¿çA¨–5Ò]Ší9ƒô~zIÆ>*ÈÐWJ$é‰LÏFƒ¦_¤!˳Y̤¸¿`6c"Ø Ù'‡„áè¿ÕÐ)=ƒ\¡«q²¦§;j}ÄÓšy^9pHAy®Ïð8àAbï+ÆBw‘?ƒ!ÕAø Vñ…á«Ã(|È«LåŒ2ŠF]qx¹ržyç%6r¡„èK¡C_ Qbˆëí­1”¢¸Û^ãa„Òºt"Ьk(SÿDÒ6ÝeþZ<:ˆòÏlV‡’KñhÜîö»oPiȽ?AÄñhÝË 6Tiߟj‡j¬:5¨¶­¹Æ<ÖŒê¿Y ¨î ¡FbŽ`†à4j^»NM6M‚Z¢æ Eí¨õkÀìfÝ(j{%?¡’AzINª{»¸H£=µƒºœWŸN‡~DÝWû©ãΣ¡ój‹g%ê÷c¢ŒF½R›8ª©£èÖÉâ#yªaUf~Òoí?è›F™lÿ*Rûˆî¿‡oìOŽ›[Í@®ÂÍ"ëàà]>K•=_^éïNŒ dCÚgó³'3!UýóJHüÕÔpGÙb/õ¨”<͆¨Ö .٠Ľ˜š= <]­¨v¾kO;ßë“z •Æ¥ÃäîPw›©ôéÇ;Ð0e ê§QM¡2jŠp~ÎehÏÝš‚]¥ =>"h ’úíüVÚÌûº¸ÈA›A´Ö)h­£â³8-­”ï¬èï™@‹Ááaƒ_¡9•2%Æš•U¿¿zTþ/LpþgÑñÀ…ÛÐ̼®õÖ*šµXEGwÏgO2 h¹òð ÅÛph}hÚTÂØmënŸä #dÓÏ6B:BªK ë¼\lí=ôLfš'Ï|€¾­Å'aÏæ` ˦„õûeŠnôPã‘øc¥<Ó06ß“Ú$ßóO?ĹÓ'H`Æ/?V`÷ºsëûÛž$‚…Ñ3ç~Ö5ÀR߯ísÜŸáÏôìL>;¬fz³~w 6Ò2àñ3ØöRpÑ5SD‚¯‡¾=5‹„ªw?Žø?C¢éÆãÞ|F’ÆÌÆ'\O‘lÂñÉà1¤€»Ž3 åHUeV¬õÛ÷=ê9²¸ÄˆtŠ43"}Õ ß•ô9dt /"9BÌüžÖ÷V#+Å5ª¬úVdãRvÌ<^ŽW®ß¨—AÎD=®fWäî¶3ò満¼¦o{¥Ï!_Û>WI©-ä¿<;›zO·ç…Kâ!îö²á]½©üvbÌn?¾/ëTy»Åw¤<® ¼Bɳï'ßæB©¥äç7‡¢L¬ÿ•kÓQöÊÓ\ù”7(;X>ŽG“Ú¨ü€J_¨;D´ç©™ÆW‹Q%+Ý;ºŠU'…ã¢ýö¢ºBs•¾ #jäéQ9N£Öù0λçQGåëþà?gQOäú”P3ì[ûøc «:8–Ñøª/ :w É½SG‹ÐT;ÌáÎ:1št[¿gf©üNbWÑÜ’ÐÓÛä-šÏV•¹P¢…¯Š½§ÜA´$é¶ù€–wgËü]-W*ÌE\­Î$Y^<ÜŽVRG.\%«µtsO3Q¶r‹ÄƤ )›G.Mӵà›òb' ”‚¸‚”|%äwË:Ÿ{iUH·=é ‰Û±ùˆêUHhQgï+ÏÞ彯O³… zùæ–´¤|&óbÒ™f†'Í’Ì™@ä±ÞÄï9á}O¶D‹úDí÷T ̃Øû§Ô73È!i‰½‚Zã:¤ÇøC@ÿy—ùÒ:å®BŽ‘€ï­xÈÕT2Ž€ÜŽ÷¹ü¿w}"Á±gq? /8.ìkÎû¿1ŸØT€xò%äÓ2¹™ÎlA>û¯j+—ÜÿÍɸ¿åù™ Ëù»>QÐTÏØ “Ëûhºº¡¨Ð)÷ù]Þí–šü½%z®Þò ‡RãvÃퟳPVúÔ›¸*œ5ôtBÆ êa¶win4Ô /+§k„ï&”ÊŸk<¡~™nv"Ïšx‹?ÂÏòàWÛ!Ðòù9é«­ÐVí¤òÄò8ü¢*Iqì„ß®Ÿ&O@€‚,Åô…*.95ÉÀ€Ñ×-.%:š)þ32Fø7©/ù†1ý·‹CIaÂT¶ÀáÇ Leó7WÌÀ¬®àm»0+˜'v:òê‚,xÒ×ßÕÉÄ ·~Ÿ‚åÚʽÒ}°R3¥Ï*4 ë{¥uöžµ‚ÍûÔÔ3‘d°#¹‡vø>l%N™”;!á•*/*hîÈ|×N9’„GôäÒÝF2_ÚŸ<÷qoJÎ/J’ßH%e(äýŠ iV,u¾ûCÚwg“7Î!}*Ÿø]¡Qd ž\—[BæO‡²*‚¢uø‘q˜²;{2lw°!§mñ0×z-rO°–IíAÞ9î¶ z¬ÈÔç!_ú,I=pÐ…ù˜Ùо‹×[4h‹gsÁm} ¥”É_nv£ŒÅ¯L™‡²({›ýÞC%”öÑ^K<ÚÒ@#P¬…Êg×¥{O*¢Šõ¸Eª6Prå梺Ç$ë£mÔ¸êÄ\‚ZŒõ‘V¨“`7a•ô õì»hìÉnïêL«ToJO4Œ–V#Ú¦Fcñ|î¶7lhba~;Ü}MÉ|û´o~FÓÔ?Ä'0 ™IÎwBÂ49‡OFÐüÚÅ~Õ#h¾´¶w‹-.«,…Ô¡EÏ«QÃQèg»òö8}"jiyv¸nÙži|פ| Û…qÝ…‹ZÐÔßßž9sj½Ôy^B¹åÅÐy#(Îøã{´Œ ÇãυÄAž’©¿»Ô|=ӽϓI2_É»¤ÑÿÅmJmŒÐŬhH5–èhSÕƒŒø9:±¹;¯7÷Œï7”~‹£ïdœ„ª•W|ÕIüðCñÝ”w¦ 4tñRÈ>-G¿„¼ð€öK·Ö–êá7?_‘‡ž¥ÑÊ4}Ð/+Í eánò„§"¤`ÒyúÇ^ÀàÚâg|ªQ2—7€á÷&÷švùèÆ‘ì»Bµ›pY¹º cVÃC²®¯a\ïv›P L˜}ðW¯…IoÞ¤FÑ>˜bKêÿU 37×d~ ž€9¾}v{äI`žÇN@à<,‹¶ùD,:ß8Ò÷ì ,e´rñŒêÀ.§w¶-æ°R8ÅÕk/5›{baãKû ¿lÍôVþ D‘³ÜÏXqÏm¥¬CaFH8)&/•Ó‚Ä—1 CII–Yì§8/#Ù×7ÕÁ¸Wá cÐ^¤¤ŸxÏÿØ©&ÜrýÆ}bµÄŽgî#mgÄ­ý›H÷_ÌŒhé;]_6ÜBFâ)º·†ÜȤ© _ï…ÌQû®N½•DVZ7Â¥Ÿ¸ÿîÆAndgúåóT°9‚NÔYÚ#'åêé‹ZÔÈ•úÙø¡þKä®åPKFž¹¥UÊÛȧRvUöTŒg _<í…Ì\!òå(x6ægz …<—¼c4‡ ÓËܯҢwíëó^(ú DåI2Šqz籸; „tüÉ¢Ô7(é·S˜ž|¥É‰ V¡L\çx«5Êr5xxуrú¿¿ £üÃ3þ²ñ¨0ýŠ÷æknTt_‘% EeÞÞ¨¯]ÇøÑXƒUpÛ;ö.ªZ¾ýì)þÕ<Þ‹/ݹ…êM—ˆ_ð&¡¦áG¿‡I¨U­÷Q½"µCs&j¨£Î ·ÍF]z›ìçÎf¨›ß•uürêÙ2¦Îº¢Þ„ƒ“Aê_õ*JRÔŸ,:[ù= N¬¶OŒ=Eƒòþ7-‹Ð/Õ…3Ä =œŸZGÚW…þ&„hD÷mÍËÁoO·<Éø2‹ªY:4Òn#KÆ|Yw¤=¬É÷ü¿Äõá?î¿/‚òõ[ÑJb¬ðíÒTêúápÈ^ÝPø$nYÝ4üÌ!-•E)ûÉ"$UÐïxV‚øq¢_޶#;xë9„CL_V\È2ÍßuE{¤N’’r¬Éÿâ5ówSuñ€ |c8íþò/k¿rq¢S6r³ºtP~÷ôH”2ír¥áƒ²jï;Yûd¡âªÔ¶œ8¨b™3(Ô΄ꌙkƒ'Û¡ÖxqëPÈ5ø~º ¥ÊÕê5<õ­î @Ã6uÜahúr²ÉÏ š†Z8ÌT‹¡™ÄÑÓâ ´œý-Ô“õZÇbLbo-@û—³£%cð«ò %õÄmè’<5 Ôd=šÎ¡ÄÐW|hÁÊ øünµ½ûƒ©GU´ç¾Á°õ»£C÷`TÛå¾²(ŒŸ;l—P “ß× ¿¾Ÿ‡kòUâ*˜ $o+c> „cVv`1HæØÍNX¸¯xž þ”U¶õ„Õ‹1'üÒÄ`]ž=…˜ùlʾ8·q¶/±î¿Ì¢‹ïzˆÝ?㞨IúÉ$"$œîTâ.@b½î´ÃGÏ#Iñƒ˜áµU$³®?~­÷²´=êuFŠÉt¶Ÿ*_þ§o¶Òxßê(QFZ¦ÄöŒöOaW•7Òíô~éÎAµÐ‡Ï™ˆNsGÆW#óÆi¦¾@7ÜÏøc8ÙÕ-º»…ð€¬å¾‡~þÈ9ÏTx9¹MùoO¿CÛlŠ{È{ÒR_ÒrW¿õ“‹°#¿×g¦Œ§˜9†Âb.gvìÄQ´ŽÁA=µŶLI> „~UïÀ±\”ÜçfÇeÎŽR´ãþ›ÎûQZ]q…ØËeŠ´4œØ‹QÖ÷‰)y¯=Ê»\}ó.†z\;J< „JÉF/’Jƒ÷]zâîƒ*5гç¾h£›CÜÁèxT§¤>]`ãŠê‹üõ¶«·P£¡-û› jFl\A-‡ÛŸ“L¢6ûïð—¢^¨]%ïœÁA€:çMæUU}PgÙÉ[n uý_ô6ïæ/=5‘,Ôs÷¾8wwÇ=•ORIjQ_夳b¸'º®ÿÊ“<¦‡Š;׉¦æºþó‡|B}"cPÁ÷¾þÅ(2÷抺rlÇ Hâ £äZçæØ ¤M ß1GH݉Œe+})Ê7”"φ/!שÿÄPm¶‰ÁG÷6ù³8þÆøÜÅ“Ñ.— ‹×žôЯPl<ù^ìÔTî#¹4š®u†6Ûg á¶ËõÚ/KÐtÌ.’f¥ ~æ§…¤üÓoæjÅö´ž&¶m<mâaÌ—SÿÆÖ\{‚á7›ÐJo_áF -ÖË7ËÞ°CsyörÍ×ÉúÑdտ܅fIO+ÎY±úÔì­*¬„Ÿ^5Vì½ÿ÷¸Ùå”Pıÿ®'À±\ÍmƒÚ — ƒÔ‡öÅœü*v–J2Ý¿ßô>·«‚²<¯èó ³/Ôz¢Ù䕎0DÌP&qï: ¯%ð³oQØð ò¨¹X˜ˆ6áÖ|x ¦_GñXºÂœþ·…ûj°àhœrÄ–¬©Ï_÷‰€?–ó$¥gJaÕ/.I‚ü¬Ï=­¶9-[Q?‰¥c? ò5Ñ0£¸''WÿÖ Z$²Ô(OBß“—÷´!™Bf•èÍJÜ{ÏÿEÞØRgY Ü^Gš"K™Û‹HÛdQ-ÍlôÖ,ÆQaª6Ô¶e™>Ô£*¯D–[5ǘN-ãþGŽ››í¿½ý©”©'r:vǧ½ØAîWÆëå‘—Üó0Õ#3ä»ß—Ôe1ƒüô~ï%.o¢@»ÿÓÍJо¨V­7(. èö—árñý(î¡_¥ò<%9Ih¶*Q*ér/uÒA”9%ûéÔ9”Ð Sž¢@ù=ŸÕ4¼}ðè^µÜ~ÁT2äȼh䇹‡âfÅT¹9{.Èø#ª&ÔºOÈS¢Ú*YÂ{*sÔpoÙ„¤4Ôâµ~LÍŽ¨½.gõ€¯uû }?í ~Y&mï4¼¹µªK8Ɯ٠Û'¿ ‰íë©iA4e‹÷å–AÓ¦ÞÂx e4ó‘:ê¦æ¼âéÿ(¡y¾”¶E©Zènµ¼–ŒC‹úÍÌÉd´Ô‹šqsÿˆ–…¬ÓÃJh%ß–ê…V¨n0>+ÃK‘ÎÙz/fP¦¢-ÅMöRæ™(Ÿ9ƒ ÍŽ¬+÷¡$uñú«›³ðõò»_;Õ¿xûgÿâ)vÐcx_µDógæ‡KÃ'}=µÃjŠù8‚­ÒÉ>(Õˆ¥ªBÄyÛâyZñ³ª¸ñ’Î/>v¼´«&íÛ¸Y¡$Ä‚§áëšÄëLÈi9·3=´ëå~pµ­Bn^¯ß¹Âï'ñÐÿ'µà1F¼éœxóÿ3æ³w\âò©üg~V{.]…‚ïÞД (ìþéXû­¾A³GSH?)é7cü{ÿeôyšÇ.CyÒÔù‡?¡òÛ‹õ²ƒ%PCÅ1? Î¦Ô*k9~ô EºAãåb•ø)tÓnÔšwÂW™êþÓ¿ÊŠ,¡CÕY1§Û:ƒ9{ä [ÆÍ‡ç§ôíw¶ˆ#Ðßœürpô æYßO¥‡á|„UÞ'aôu··8íÏ¥ùC9ÕSÌ/ÜÒü3‘¡!pæ¾D—/¥ÿ†¦ÅI¶bbX|¿—ðZ“,Ÿ~ÊP÷¼ Vì£B˜a-ê»ñóñ5Øä®ex°Ûu£¤d¡HP¹qŠ~Ž ÅëDÅLc‘è3o±°Ä!$QíË*É E2Æëȳp¯”ÂŒYÌR~ÒïZµA›0?)A¤]ÊÏÔBzýj©ûâÈÈ|œÀߣ™÷°8:žFV)Òßo¤‘íõRGÞñE/ñ¨úWºV_uTúÑbgx;±J0žj Uo;‹¿‘¼‡ê\5­®„¨Q×R™Ì‰ZA¯Îq:բޡèÈg)Ô#>äø¢õSh¼býACýÂg^ãhÔaÈ•÷ M®úL‹£IW¦¼ö[4}Ô˜åôŠÍuªñ‡¢YqÍÌòtš1Ф¢ySè’Ï¡…ÕÇÊWТ°ð E×ú¼ä ½°Ç¢|’/HàþÂ_ÏD_mí•2s _h"¤|™UY5Z;áŠBiPVúø\ô\(31ùY½ÿ»ŽÿÅGö]ãÑÀúð¿|ù/n“ÍŽ0¼›øË—ém"¯$bSþâ§d²62ðyÛßõ]÷âC»Ì‡Ðhêø0·¨ñ/Oµ.é˜0|†_N¤áÇ^îêî½láúÐG‘¸ôÇñô«@Þy=n8à8NDëaÑK2*08ò¢ÁE]†~¾Idâ_‚á”!¶‘ R {n9£·lZé^†±sþ&N 0n‹¬8’&n WWïêÈU[¢ë0mP}à Ì’SxÛ:Þ†¹£ž­L0ÿøF‘,Ähì …Å±½–¥ÏÚ`Yß&®x¡þ $¯Þ÷‡Õ¬@öWçöÃzÝMþ˜±]£1K8a-;Õo ËquäÐ¥ëH轤Åu`‰÷hÝÒ5@’ûDmaü?Œ6òKþ’ÏŒ‘ø#Å«Äì# ¤zÝ»ÏÚŠ4ô GZ® íÅÚSoŸT#Ý÷„«m’HÿÊ0¼³‘rë¤î6#ãâSÖ˜dÖÓ­RîÕ@–<›#šÍa¸_68ß7‘--f,P9ø÷(:½À‘è3\V ’ʵg;Ä×G>y~š›°S"ï\oÁ¨4œb&ðYFþ4WïÔ‡—Q`mNRQ@…¸¥÷ÝÆCæ†Ñ‘Ñ”(ü‰íuîÈCåüåì–E„‡'I¢üçQ‚î–ý9ã*”tóÑ{{o¥ ø}ß Lcð»˜é*”=V§çÌŠrAE,¢-ã(Ÿ«¾d™ÈŒG8®D¡bLš Î…`T6vcfšJCä1æ˜Ï@•ë¶LµîŒ¨z¦8yãðkT»»Åë¶ëçÔ‡Š3+DMgï³Â×Pk†æ¶æ‰(Ô.¶§¿u‚÷¬aM5êꦽ8Œº«¾—Î%\E½·E>ïÒÙP_¢¬óæÎÔÏW="äG‹G#¯Cr¤|J›×ð@Cv&­L O4ôæ‹Mv»ˆ†m+§h.¡‘PÏï 4rórþtþzq=9³LÕ‹*O'U~W#‹²‡ëÇ•0XmO[¸Y÷~ˆ§¿” ~e[§'W4þêÓlêcôLÈ”cžY`ôo‰^éoxù !®>³crÿòÝ¿úôßøïçÿžÿï÷ÿÅï¿xþ—÷ n·ìS~{÷/_•\îR¼½¥á¦ä;;W¡\þhè`²#T´ˆ ÛBÕU¾A–ù°ÿøêÿӹߗZHLÒ( >dšµh9™TGYÇüÍ7?)I‹ï3þÕ­-©ÆL¥ô…Ðfðyö¶‰ØÿǺÖüš/vv@ï‡+ƼÂËÐÏ[záüq%ˆâ´ºz  †‘‹{ÌFøgLz‰`LAáûžår˜xPžcbÓtñn÷£%a¶ZTÎyæ-Sºh—ÎÂBwBªÙƒhXrkÞ‘€?”}ñTN¤°’o_Ýk—‡ÞhPÆ£æÞûõ³°õîÝÛÛ°³Ì¡m™bˆ{dS=y÷ÏD-Ï^ŽûËk¤MMbõ¯‘üñZ»h¤Ô ”6‰u©C…eÍ:?Ô¬çC­½¨#{hHµO¶Œ¤íCúG]"u¨“wµüúö1Ôµò*îú¾‰º#_Ås=QïbËPr;ê<í¥ E}k*‹3Ψ_tFÔÙ ]/fO¶MïÅ£ñ±§†ì¶O›[:Hƾü3͓ʟ|ÜðÉW†Â€›oI¿ý3¢•ÇÎ2T¨ùŸJ¤AÚ?žÚ²H½£4¨¿Ë{ruÃëТ2Þ–ÑoN’£ºÅ45„J‰à£fío¯MVˆ?Þ™}á±!d6’®ñÑhBñ^3†¦l[¨È.–ˆ0‚Úo9 ÝnéÐ@õœcØgš¾¦íÛ)‡Ÿö—LŸ FCs‚¢ÉT¹ ´tÝ%cŠ7€Öí]ìŒ]К8ì&œ­ZK2~´BKÂ匑ó…ÐB§Û<~‡š_…m VÙü·(gG\SêóÏ\Ä—}Ó»þQÒ+â4Ÿ då¸@͵ïÅ L>CË=ã’1Ð$«ÚüÂÚ&¿nÇ•¯BÇéÖSI§‡ “šìCnu4üþÙpê@é:ôp+ä ôyfLKënÁ]x²ÝÌ ¶Ÿ»}6† ÅÎVR݇Ѿdêeçd˜0’½KŒ30-ܰš¦ns ¹yQ$°@ÝÝþý,Ϊ‰Ãò ÂøÉÙ°J²¶!¨MëÎNŒ*Š/a‹Î…ú<Í)ØéöB¹?qϹΈ_…œHDc¡8˜ƒÄ-ÙGå¾#i_ðó-ÜËWòö9ý¤|ãþäª Ò¨i3¬Lu#m ïñÔ¤§—H”¯G†°Dr‘·%Ȥ!–«tY˜¢Èvu‡C^Ï¥dw8y§Ý L×°§æ!·¥>‹Û¥¡ÊkF£¹Fȧv7x,ÞþµçA¯Ɉ2²Ë“+ýú …/ñÎ BÑ/5FoÇPœûØÛ ×”X䤯}u¥üW4Ø›•QF†…=o´¬´O\ ‘C¹òËÁž¨Ð˜ÿ‚UD •¦ÜÎ_Bp¦):ã€*§“‰ç‹Ÿ jè½äg:Pm2ˆˆå 2j\–ùiÞ†Z\_Äg[£öʘÌÂû ¨Ûõ…O­âêçe±÷<@C¹E+ó4æyA#}ºMì–ïÈy°£©@üëˆ,V4½oN¢ÙÇ+yϯT¡¹ÅŒln´Ø#ë7ÉX±IO¥•4ÐR'å‡ÊÝ“h9ôÛä§z#ZyS~æÈúˆÖô\ÇlÜ#Ð:Ú¤aºù<ºd6Ù|¥¿Gˆ°iñ"E_ݤÂâH¼r±ËvJ´©,“}Éÿéç¤îq„éŸ9…&4/ßCÂdå›Ê&Ä3étŽ!„XøqT²$>¿ŒÚö°‡¨Ö¤ ;Ó\ˆT§²øþyÞ§ÄD«w§ù‹Ç˜·<‡ŽLCâ¼y‚7µ(¤ï_œŒ¸KYš_Jøìàë³11é ã}$žhbrº:ù¦æ!×éžÍ…è~Èíåkõ}¹ã³Ë©:gÞ¢ä8úåÿ[Uî^øJùVIb_¿HBÁ×xÅ 0ï~…• |ÔkñÝ—Åw¶\_Ùz@Ƀ°k½…Pºôrb„AÊït´»j@åööS;fPñûÊ/â¨]J¿c{<~\õhòt¢ƒF1&oëî hêØp;ð(š?[¦DÅnžH8ì7íMÞ§ŸV8@'ç×Å&è*yĪT ½³úFÃÐÿ̦…ðì7¼³SÝv†M[Ô@FµÅék¾‡qçæ{4K`²²®&G$f »óåKÓa熙Aaæ'¨b…%aÑ£9Ãä%,+ë^㈂å¦#a{,aí6&Ún_ƒ*n‰.WØ~¹¤üë²Äõs_¤GB:=áU>V$º¿Æ%2$‚$<×H^"’ÎZN7 á^ŠwGxb«‘òF÷²›œ5Òw,8Òe#mˆÖÊÖj(ÒK³ö&¡@†+nFdšî^q=Ь”Û/MÇÍѶÓÅzˆ)S§”³ ·êNü<ÞB^¹È‚ -xðŠoáµè³› ©ÅC!F±×ºQÔ‹7²çéOH‘±.?ƒ’QRi÷x ôãÔ^ìx¤Ü,²Š7åºûIJLð(]í­kÅ´¨EV™fª…øQzê5û<ª^ô?¸ð²Õ÷½>KRq 5 $\—öŽ¡–€VÐÔÑ0VøY9Šz{¢Ÿ8Hq£~¦Ä·›J»T}ì¼Ýq»@4úsAÃ⌚¨—ÏÓ‹ESÒ\©Ó÷Ñ´l~ù'g0šyõLEœ(As±ÀO~}§Ñ¼kRAMÑâa­¹f-ZJHj’Ñ e»tæM_ô) å%™mFõ’’ˆwõn¸?©­ªß3 ¶n+$X[IBcGKÜ“”«»ë­GàûŒ”Ùùn5û÷Kc¶N©—þ·.¬{ãõØY 'võê—„3O ÃúE¾ÙÞRH}B@æDG IÆWõ†š!ùhZõ'FHúz*°ç9ä]<¾¢.3¥ÇÃüsxô¡êƒúQõá[ð½.í»??4QÌðí8>„æœmÝÑ÷µÐ·‡“”útË“×s3@÷ùW·ZngCŸZz„³ô»x‘&[Áš 2íw‡Á}x{Ͼ§08½Øü©þ µÐY°W+ÀðW¬û ‚0òéqêíd}EzŒú0ö&³ä “Œ¼–ºwQ&ªô‡LaJ AMË&¦ÓTÜ:fÝþÜ°Žƒ¹¤ê¢Z˜ÏSï~q%šÚé’^<ƒ%âQ×ùàIX>aŸòDâ,ü9‹2Ÿæ`5>•ÜΈÖ“¤XÆa³Ÿ*7!IvOžŒèD‚1Ö«ÁŒHhì~¢Õ±‰ªSLä¶‹DC¸¸¹ÛIk¾>šCr×bKÒ‹¹H¡1ÆVeåƒTê·O¼Fšn}²}»þP…"kxo,Ò™——›&'!½Ëûáþ™DdxÁù’òÎdüÞôL!…™yÎIj†!KÈ[2sÜO+DÙ7ÏlÏ«·ÔÏ [¹hUxÍÒ¾”¶\Šï*ÓÂÿåÉZu"ïl´‚HžÎ®?l–Z9܃ü ±.¦(°Ãs%F¡…ÄD?°â!»£b-NeéÛ>Œ¢¢Ïé„•§ððZxy‡ƒJ°v;åäÊ¡¤§•,9¯J³J†x¸È¢LC¾µ -Êšçj9d½@9?‰káÄ(ãóþD,¥lò­Î©@ÅÀœַ˨¬|r‚éZ"÷››üe¨r÷׎¢ï=Tõ½‡V«ÕRxñ‰Á•]ÝiïÈÜgŠšwN ,ª×¡Ö¼¨ÐYyÔná¹U³É‹:IÏnœ3D]oÊ{¾¾N¨§¬âåM‰ú{EÆ¿£þ×®4‘+4p"g?´††´¢‚ÏÐ0“Äéä`"™×þö™æÔR.¢ñ“Åçl¿ ЄÞ.±¬$MNÅX÷ª~@/ñSk¦‹ˆ.¼¶w G2 ék§ŽÁª£äFâeøÞ­+—ñÊ^Üdò"j„ÂO²$~7!+%È4¨" 2†ŠL=΃Ôò†?ݪ§ ñ@˜l‡~Ä=â½´dm15‡=ù£Í º{®GVùD/{ü¤ŠÓ¸q¾«› ñ×êÞHm|¡<¡Ïr’~MA¶…*·€äúR.99ºBaYTì€oûJM© 9 x)áD…È”¾¾œh~¢ú?ØIÃ[h|ªAóóxêšKвD}÷!ý.þƒÜ^/tx\»F)þ:ÓI‡›ŒB7 Ÿuýïÿü!ûÙóì)µ0ðlxñ%Å ì‘Ø{í- ï<—Ÿ­°„1Â÷#'NGÀ„¾Êúò“0•ù>$¥Sfµ›Â<«õx5},¼µß4ÖK¢B[<°\ràÜ÷øX±O(þ¨ýÖR|³»— ,“Aª9j°¥iL7™;¯©h¦Ýïã"L3EB½ò'h Q é¯oWdÿ¶CÚM"$MÊc]×GrÇ£÷)ž†!…ÄÇO'‘r‘ýaÿÈm¤ŽõcN|†ûôÊëÒËÈÖÿPÄü™«Hwï÷Ǫʤÿéø‰#}/SU´öˆ"³IÐÃB d=ö¾»h Û[Êl‚Èq¢¦{1“9•9|ZA®¶Š¾Øäîg¬èð‘EžÍ{9mråÈwöw¶nÔMäΨ%ù˜Œ‚îÅH´ø£0)1Å™\b͛(Dqvë@§k(a6ß„(I·ò«Ú( ¥öí¥È:SÒŠN'fˆ P&õrjcäÊ^Šj{~(o:ÃÕ„Gí¯‘&¡Ò;ÚéDw~-—=¨RþDzkp ÕØéŽê–Ø£:o³²=©=jð>öS›EMžûyTP‹fBf™ãj-ñÙiÄ vc—w˜ÞêÄ(k޳ó¡®ûÉr¥ ÔSú(é4шz;o)£ï íúÀ'‰ gÐÀ³/&¶÷ùr#VyWÿκ/è{àÕ®—nûT£‚ù¯üßÇûö¬—‘ŒÉˆ“ä˜ØBÙ|´‹Í]((sgt^¿ ÙýîvÊŒâ^{•ç¦/¤Žq|K‘n‡”µt—Ó¢ìt~õø¡÷UÀº/Н®¢yJƹÊ!r4Óxzk>2»óò#Äm•î;s2½“¦8Œ~·û®ôû½ ÂÎsÑl©jO.Þ€úð¯åæ‹Ðøƒ¶”Žì+ü$9Dq.¢x××9J2<6ƒ›=™Ç¦6 •ч‹+ŸZI6Œ¤CKsŠ¿?Q´È‘sþŸÏÛç0Íç½ ï·Uü÷ÿCeÕ ÷WÐlÑĵÎá Í.§_¸Rîòo­€+ãqhê>òãê ´L*k™8˜Aë’qP/™2´½¤†ŽB‘äÊfsè´¦ªTÑ….êâ×ñ‚‡¡GùëI[=oès¯¿ohý+ô1Y,v0÷ù@µZ û¦G]eß„Ñ×à ÚCÙ0¾Â1\ºê SE;Òª’"0;VçHâýæ3/þ‰? ‹÷v¸H{ aùªÿ÷µÛ°òD_—ó~&¬ ©Ø 1­À¦/)sèu.ØQ¡çW2 Å=ôsG‰#á—‡äÒ„¥Hì=ÅBPU†¤ìêÉÿéHðñ‚b3R KÑ’X u¥|oþ‰*¤Õª–¿§éŠò鲯J¡â›ÕB™ýWà 1<}Â&šŠÂ^Å,PŸBQë¬Ä_¤Ù(V{á²€Éo”¨–cñMB)ŸøCkV(C¾ò’Yþ©¹ !ã€r¡?²b>¸£B¨Œ‹±X*vÈP¼´Eá¸Ër Uô\>[Iº¡ê“·©Dµ^úã!J‚¨áø2îXÀjql·5àÔ^c8µŸuÛ<&)Fý´lî4ts9û¹^qu:·‰&ûü<ö¡©T|Sùšn.æ4o{£YYÜ£ =4ð#sH-´ø9žEK’“ºZ}hYjñ­|2$ÐZiû˜ù R´Þòï!¸†6¥¯„ª˜Èñb\ÆØAR”j>Sbù)ÈtÈÞ«†MÊ« T ¸›ÿHÕCnÈæ—}èB9©ê×Ïgþ3·ðqc¤;Ä]}o#°õz×çñ=5ϧø_]©ªI DÙÓ|zZ>¬®ˆ.M„÷¼¾krCxÂác¡ŸZÿâ1FioFÆU1Hôʨ{žèiͧ\ñµ d.šÝáœS„¯Ô-{Or„;§ÈF· 'ÒoTˆpr™‰_ ú¹å‹ÃûÉÆ)øn%äí¹ëüáîÿ;Ç…eÌØþSwJviláŸ÷°ñÁgê¡ðoŽÅ?|+%üŠÝöÛw@IPt¡‰V ”¶bðÊ+tX¬¾B%°glþ€êã/TKj?AmHTôfIü xïiqü%4tFOõX)BÓã;Í#Œ Y—¦õ–ðwhU%ž ƒöÓ!Ï'‚_ÉÇï¾Ì†.£ð4ë2èÕÞÛúTsú‰¸{£a{­•rs†¶¢ v˜`dÅÿˆ£g Œs„ÒÓm*Áä šC+t0½Á;<¬5 s–SŽ’Å0_’ &ýÒß;ˆÌÀò~¢oçc¼a…@ã̵°f™±—‡~6ÊŒD;DaÛŽ ;±¥ žTö—·1àž…šC%K‘ȱáÎáÐÃH²‡ÈòÂÉQ$ýq‚öÍ;$Øôš‹{Ž”W©§Z#õf¶¾Á¼ÒÞ “ÑŽ Bz¶+LÛ÷†¡/ÿKÏEdj4ùüЋY†^ð<Ãd“+êâ4ßBŽE{AÛKÈ} ‰‰2:y)§?xèÇãAÅ;QËžOQ *¸Qp,éps;5W¡¨ÐÙ7ÇÂPÜ0àäÛ.”¼ò6ڹťÝÊNs«Îâ‘§ýÎæ6(+ýûm,*ôv÷ïÛéA%Eq&þD/‰ýÂC¨zì-›áª(ªíœ™óÝÿ 5Ù´Ï¥¡ÖÕÄÑË»zTÑ5hÏñ¨»ÃóJ±ñ.êçšé|Ð@à ûD ÄјÑwó1…šœ¬¬æDSq³Ôõ+hºÃ «»‰f?ß°4O y¸± u3Z\,»êÔà‰–Êyî'ý?£ÃrqÙ\"ZÍ%h¼»z}f)²Ý®Ç£ºP_þcé,ÜѼm<›©^ÏŽ[@cJDÁ©ÐƒPÍ]f—Ñ ¥ëž‘–Ùÿ›ç™zðÐ^Èö 4ðÌÿêÀ ¨K +”ì’¼ñcHŸÎ|+ ©Œ§³Nù@¢Á?ÉÃ$7eݲ麄êãL×!/Fâbæ‰H(­srL)†ê; ·Êc¤ ž¨¬¤!ašÊÝ‹5iZ¼Î—QÈìƒöãyLä:Ðyáºê—èžçùEu—úàbc®%“L/?Ü?9S¾³êÒI+0ÃTÌU–j ³½®¶ƒ-07Øõ‘×±æ'™^¹q÷ÃÂZ]¯êǰÄSÌHË—œ²Í/­ÀŸ –{_Jí`5¢yYPNÖŸ½‡Çã`3£²zˆ¶w?žœC‚2}·˜U$ä^r H C¢æþ?™âHBdú…æå’úwë¬Dr~‰üƒMó¸w‰ïN¤]=RÎòÓÒ.! Iëî#¾Œ´ŒúTâ Ý!Æ©‡BH¯>'6Nm‰ .®K=Ù1†qïÀWydZõ}§Ç?,ŽjÔ.Š4È:^râz”²¹MÒ­þ„ì[Ä9~ #+$È%ÿ¸ò|GÐÿåaÿ©ì8äI2~¤½ëóŒ§W!–Àò{S$6K¦3ÔA!y’ó”\ÛxÈùÌJï¨* R–3ýDÑ£z_û*[QŒ\zˆTÓ%¸U]©xcPÒï0Ïõb ”æ8ãQeÝŠ2õc {eŽ¡¬A¬š~7 ʹGÕS±p¡ü‹ ›QjyTX( «‹AÅkÅ™cߣ²ˆNãê/D®§/ N7£Êócc‰K¨uÑœÿÔ(ªujŸøÔRl÷´Pó¹Ú™üáˆ]½©Û⳯µS=g>:?¸—Ÿt=DÝÔsQÎT“¨ÎðÌ8õÂ:ŽðAX~Ø£x h~Sÿ@Ãá\EÚth”ãÈYë?€ÆëX2ÑD2U´C¼ M>Üx~ºMimÂÂ"Ñô~ðºKÃsþžµ×ÚP-~v†\ÿ§ýdø}(°ö³mé ßã¹cÂD ûæ_ƒ×*Q~¹ãÞ9OHÕŽÙ—FNòïy¨—²Å$¿ç™+ÿlj µÕ{>úâñÔ$dÌÝq<ûl²¯µe†@ž§}œ[íÊžÿÊËæ…WB‘¾ê‰Ó lP›§Ìh½eƒ9’†§ bb¬Ï¢» ªbîX¾ûÌ÷ÌYi®Ã÷‹“&¡†mREþÌ£ÿåßÏFu+¡¡î'±í‰hr0½k> ÍÇEÕmÎä@Ëõ¼éµF=hí8vlÌËÚUêÌAGƒøHñBgÆ…ÔÃW:¡»ÝoÜ}}‚_zåúxÿãCff¾’óiF©ç´öül¬KYõ^F Œ×“|ü “Weƒ*¼`ÆáaV7Ýq˜›½b7÷ï¹O÷ßµËr$°èÀz×Ú”–FRôǶa%ØYðÎ4¬±>—úí'¬wÉŽSXÁæXúÄG°ýG±ƒ§þ*ou\S+A‚ÆU» {2$b3‘ˆçÞÓóù‹îqd‘›)QÉÉnámb$×»(?qi )iWýÉ|ú‘ªCwtÇ#i&Xê”"^ÕC–Šd¤¿V"!Oª ·užñ~EÆvª¬;È|:¶C‘õ åÂ3udWú5Ÿœº³DÝ×Ù[ÚÓ'÷é6ò,L$þ.^B¾÷=‘F+MÈÿóΊ¯àC˜h«9c%ˆû-ÎÕÝAof_­[ AQ£}ß{Ž^BñÎ Æ}lã()d>²v ¥¿20œü eÕkJöED¡ÑQÖ.”'«×8‘Ž jÞk´Ö¨˜ÛC)QÒŒÊ;Ž—Ž£ê±´¸7¼¨&$ÖÃý–®/T=†È]¾FÖZµªoËQ‡“%ä!A1êhÎvÓËF=­‡U.L¨ożdÈM€–9I2ÒhhšpûU2+éŽ2_%ŸGcµE¯UÊ‹h"mò…¸ã2š ~K\!OB3ö°ñXJ 4gÐz—v2-èä|uåìöüÔdn“m­šÝJ…ÙÑÏÜßRñªìR³µ<ð@ºŸg3£Ã`¼äE Ç¨vL=ä¸òŠ·ÎÞkT\ÚÓ‡Ú¸cÍ&äjjΫ°€ìã…‡&í‡÷«S[Áöòeá”ïÁGæÒo›^±I™BäTt¸ß%O­ù' dNÐAêë´ Õ4/i7ÀrßÊÖ5we¿éÃW2¥äqrPKì¦=  R}§¹D¡éºw+#]:4§V«3uCË—‡ ê— µ5âh‚l´Ýñ:ªm‰É—¤ÞäC›û/]+µëÐ:»á,fÁ­Wo)™¶íq¥;ùÈoùÿæ—ê {ðy BkR“é7hM1 üN}Úˆ)~-·÷¹Ü-¨B»KÜt¸œìšºÈ ¿ò©-Ý<÷Cû•A½*è~§Úòizro~9鵜ô$㳪‚¡Þè×TØ #fL™ aLOyˆS2ÆÍ¹ÙŸ»•Ád±Ñ•ì1˜±•>ÓÌG9?þÄ ΛÜ›Oa‰kmËò^>,ïÔe?¦5ÚOX¾ 7öLºygØšKà«í€Ý7iTÚÜ ‚Gç¹±Ù*‘¨GQ¥Z‰"$“Ö‘xÕØä˜ªuì R–,.DÕ\A‰ŒKõ9HOxe>slÎÝY,ʃŒK¾ã"aÈ|gͧWŠYÍm½®ç¦ »ž›râ³Hä ßdÔ×Gî׫FS²È—îô°cÀtÔ¾Õñioüdü”º¥öß$!y%t4(ÿ ¼½à¨%Q‰S+üòáõéö¹×àåãc ÎúeðÂŒÍé'ûËÿÕcrßÂ×Èt?È$cšI5-…œ;µ«Ci-Ÿ°¡Á!gî¨]Ø…O2Tc”ÿú—¥ØG¡°ºÌ?˜ŠÔ‚ŽõžÐ‡"׊Ÿr(úÀÃxÆÁùÿ\?ÑØN„ÏC$ƒB"¹P¢C@{óàY( ˜‰8£ø|ï~ v¿Æ*N>¸qR*/îÈr†ª¢R3ѨeøÂŸS8pP¾î‰y–?ˆM™ÒcîA­TìÏÑPóm¿´%;4:ñPÏ_fuÂø›9KÐ2û}ˆí´õ^ëu6€Ž…·Yš  Zª•þ¬@OOŽÉ¹ûCÐß,t¤1 †¨³lÐå6 ·™¥[î?£YUçUÊçáwJÛƒ«²`¢Á5…øL‹ZÝRZ; ³©'éÔ Ã¼$]¡•,<¹fY¾{–ļY¨ž6Áòr‹¢¬¬®ËØJŽ´À†ŒÈ± QOØz£v9üìŠÒÈDhD"Oæ[žf$l3»ýu‰“ûd̤‘dðQ²¦Ì=$KÛ.S8BƒäŇ«ÊÓ\‘J€HäÄ/{¤ix˜$ßÔŽôîÓ~ì¯"#aÕ’rdú:Cè²Z,Ùy§Ù¢f­´KïÞkä$ßß×dE„Ü…W^ÍU!ß’>› ªG^¥ºŒBû¢£sePÄ«,¿|ßIû»ðÌõa.J¶'Ýí¤yÒó'®÷ô ÜAxþ5Ū˜*W£’Mx,MªøÝˆ¾¬þÕ²Œ²0šø> Ù,DÔ ªâ4³žÙCéŠ'Ù@bÒÞõL¥hnŸ^~˜†AOYZzȽÚ~qÛô2|Ð !uY… ãß;‚“ SId0õî0ä‰×7sü=G¹Bµ†Š!(C[b‘þñ)hÔhÿÄ#ûZ·\o.C»fãA©;è¤  9Gôz>*PÑþ‹ïUõnÞÁÜÒÙé€vú²ô<=¸†¯Š5ì3°‡ÕÛbœßÀÈv^/ãè0Œþ"I§®†±o¤SÕ Áð»1qj¦VÆ'hîK¦“ÀÄ6Ç›øg?ajÿq‚ÏL`Ú³Ïi$5fzèôˆUaîfŠ€yáq¥,Xàø¥hLŸ ‹\ãæ6-¼°$P‘Ùî Ëð6Ýò&3¬\ò+åθ«³uµ‚ƒu°þ`mhž_6sL¶NÙÃöõˆNC'>Øýr}ML ^ñG—÷"ábà-í­³Hìp) À;IªB›E#™ÆV®ß·¸¯3Q@WU )Þº ‡žàAªxíÛÕwIæƒÊÕu_¤kfý²Lûéw½Ö™ö~F.íóó’;ȤÿÕN3H™/ÛGÞ.?…,/H­ŒuMóãÆeßÈ^íœEi{9шäŒx!×Äcºk‹Ès…ÂãüAjäS:ÿ=!¡÷?>L[Ú>ug ŽS¸ìña¢ÏèÍlþD˜óX™ E)®Î„¹×£˜ÖÚëóbZ(4›ð7ú0JÔ’ØËA)#b·'–e(ÍZG¬(ò eE“CÜaE¹hê‘Ãq¨Àÿð0»nÐ27|vHBeýÇ @rUN²ˆ_,Ô@Õˆi*™ÒTëú{ªÿj8þ¸X¸S…š ±Ì—fy?4F¿E­„žžgž¨]¥•¬J‡º¤>¡…Ú¨çøº~JLõß GÝ8‹k'"â'íù$j8Ëf ÑLv@ö±ßhr:Ȧ òéŠÿþëÇ?dÐnE”^Ýg>—@ñŸ75aYP–“@zåë'¨è8Üy³ªn¥-n²ÿÇZ-G¤"àû•ºµÑëPÃSOܵþ’…ËJÓpCÃäÁõ— Ðt>ëOÚ¿ýìO•hø+¥öòî­+9’¿¡S tñ³´_¼u„q ~ÑÑr²@ç`dÿ‡6zè^/¦1f}*ÇÝõkÿLJÔÉG®;ŽÂ°О1Yê±î¾¡c¹ÝªlÂøco.5K˜ÌÏ8nâÓ›Ñu$z0!9ݧ³ ó/SI•-í`‘©þbS–:,Å‹\ bX‘Œ0—z «õ7“bô¿Ãú=º»w`3TõQÑ­ß°ýl?SÅåAØý5ýŠ{w ^d®—— áÈ w×$6šw3>…$µjU1‹ÝHv¼n|ֺɊ¿ùº"Ŭ)X]ìFª"öþq¤¹\~c@ké Â9¤LéoZ|<—ʈ Òb‚R}øÊ›Èìî7ÕÍùYňîŽÊò!;›™?Q0r²Glü¥:Š\··üÃÿ Ï7݈“­sÈw}aÿäLò&&þº€íBÝ$'/ã~5ŠÁ†^ZËÝk¬‹¢"·&¯Í…¢x† GFp-J;Wõ–±£ô-²:jN”ÌnЬ,@ÙÅZ;ú‡WPn7¬ºÒç.*¨œ~ m•ˆŠ†žQ…¨|Á¡ÛÝÚ U?š Fõ'-?ÜZÃCfª)¹ˆjT?Zîæ£Vã8Cuüêpp|ìÕbC]Õ¦ý®W QÏEwKñ ê_Jª%e-Gƒ˜~[íN4|Ëóщô}œÎðù6ˆÆ?S¢ˆ6ZÑdB¥º¶*ÍÈg¯ˆœGs¹#Õ—B ÏÕïJÿ åÛH¥þ«·Ð*4HA$Ô ­%® ãÙ°—„÷Mî r*Ô|EÚÖAgý7`\Ÿb8&°fO‘úiÿÍc£P8 'ÿú^þ 9ޤäÜ{95eu^¬²–÷ߺÓ*ò?ç¯åv»ïñ_©ÜGHR¾¾=ËœcO„BBaê—¯nTÿžãT ÌB^çV mã,”¦œ<@_²æCÊT®BM²œãüâ_øYJ̨´µÇ‡X5ÊÄöš-Âho®@‹uÙÂQ¦thµøK;è ­?“Å÷¹C@íM·‚=.üKÂ{ö´Þ/Ë2Öþ­šc.ò¥ÐJ®K§½òú¿ó)èk@ëãçÊR¹ uø€=M4´Ù°ëržš‡¶1ÆœHh/:ÍÕ]“ U‘Â+sSÐÉ®|¾H^º:Í6>ÊB÷âÀï™Ð{õ-iùŸXèÉ ˆcß‚AoâØÇvB0T¹ûäèÉ=6W q'h„±}Å|_Ô§aœØ–g¦ó+LMfhy…éñéG©Ç`®å¯…ðükX ŒÖ©zðsç=…ã°|›ù‘lܬÆÎŽÑSÂú¯ùµµÚØ:¬.¶£óvIqÁ€\ ÎÀ£Íû&HD(ä' õ ‰›šÞý€¤í\«gŸ\GrÂKØî½CJË\¡»ï‘ºû¼ÎÁx¤ 9oAP‹ "n­5†_‘1ÁÊ!aÆ™å\~¿-C–I)ÆÌadët^¾*s 9)xýpV@îË·‰M#Ÿ„¯ì»r pÎ5ÓAÁyšw¿qÿnï¶³1(œüJË"¢EÃÕ¥kXcQüÒ¡3ô‡P2»<`¶ß¥yo°PÓÑ ì9oþ9—Ç(O` r £ê»ÂQÉçvßvX)ªÈ¬‰“WžB5QÞR¥bÔ8é’—Ö-š?TÛ½¡×ä“jÍs¹‰:ßx¨#Oo£ž³]ð)=4`{'ôÀó5Îo6©Ò*ïéHÓî¦xš¾.<ØóGÍÏŒ^2ø€–b*-¾æhu–Až%h ­MÒºIG«ÐFl_Þn}Ú’s ??f¶Ó§?iíù®]«®Ü{ê§x¨¢¤œÒ•í ®Ãˆ‰:äEŸË9‚Ž%#ß*ÍÑ©ÍxÌ*<wxºCŸZ ÷"Iε|”œEhŽävz%2)’0¬{A$b*(ìW6[ÿë“§s(Š´¾ÛXÊ!uçKÛí#^"jl“z¯’­šÔžd5›Y¥ ï$¿!1°“l!B^õç“–å»ÁKr=¿ ƒÃ?GèDx÷Øÿê1ùr×/ëÑœ=ýÒ”N“BŽrØÕRÈ·x½ÎJÏŸwiô¯„O×…ÏÑ ³Â§­zBYé>(ô?×䯅ãœdÝIÿê‰&›Çö8±2ÜNþÿ\?†|3rüW/)ìö €’}’Q¬tPjõºo={ʳgoßÑ„ —lˆ³/ ò”#)³ø¿¾Ž°šuP½ðJ”{-±W$I,¡-( ¯õüt\ùØÊ¾—³%&L¡çäב»¡Ÿ¡ß¦è©ea°ÖáÁá_þ0|ãõåXã u}c”Q¿Ef_ÀD„€4;Õ[˜ú;Øg ³‡²4ÀòÌ(Pø6rÁÂw³“…°D¨ðuR6–›BN~–‡Õv·£+‡èaƒüYÕ^Ø:Öb òó;ìÌs¤Ú¾A‹X¯à’=>,ø.zèÉi$– ;(G ‹$Â+ „JHvýÆ9{‘$XNq/])WÁ°è!¤IðÛaÏöBzcÍÌÒPd˜dÙ Œz†LÜöõœÈò€U.åÞWd{Âwâ6rtFDšíCîðwÇY£Ø/}¼#bä äß8v¿Î÷·^ðåšlD‘¹gö"b(ö©šÑÊöJFŽEÙÕ¿B釺ÙüKOQŽ|8†ÿ‚*d(é7>A%އ”Ùù㨢֣äÙ…ja$Ù&—Cñ oâßÞFDÇO|Fõ¨­ZKJRÖ†:ýÛTUÙ¨wW÷Ô>Ø~ñ.Û¹ˆFâÜ¢£ñ&Gú4)šV?¹5鋿×Ý¿lîådK«lŸE§>´zÐÕ]ù–­#®eö£>ÚúDûùÚúÏÌPWi Ýù¶JáBÝRüþYÞ}:kï‡lÌ’c±o"‘ýÅÒÎk’päT–ÑðF®6‚1ä Ž-¼Â„|òÇ~t{Ký|˜6pÖFgJÃBôöø0úfÊ3.ì=s‡sEi $Æ-QÌx²sDö Š_ 1áA‰Žíi{}”:TBjÞ™‚Ò‚û~¡¬ ŕܕk(w¯{ÝjK„¾èzË0£âOí[§/Ì¢2\Uç”"A—QÓNö¨ê+XâëŒj¯ê4N¢4éÜ”GÍ}¼Ç™I yš¨è»–Qëƒkñ½äVÔfY³ŒUE]ÙqMôé¨w/·ÖK£þç}æâÚh°yMª°j ˆ¸Mþ 15.´ÿ1@6Ž~Máy4úTT²Ü‹fÊA² Ñ£hn.ª¤Ü‹§z#–„òÐ2¦…`£•ûcýæÁ´ÚÝö÷û‚ÖI ׺„×ÐÆ6å…Éa´¥6 Ð?B‚¶Íîxcæ1^t,KŃíÛ)UÈêù5d$î%¬^QŸ2oƒºßÔ‹Tû© êÐçXâPüàf¨tP1ä?VvtÝ5øW¯?ã¹ð²™ÿ*ƒ bs–ÍòHÉ?­Áèmño¿ŒªÃ£=¿¬_Ÿµ²úWGüa°©ŒW¯ýáüÇ“¥´wÿíÓ!‚Gªn]‘Eß™C»¥~Caãs?¯£P\ë:Náe‘÷J"I¡¢ÙµßC5ª¢¾Ç™6þLJÓ¿ƒ¹(àû÷i—¨7 fÿû¦À.ÿñ!µ;œï8d™³ã]óÐô”Þ†`—škuþ0¿ƒV¢£l§¢o@Ûá~Vz h/7ªï`\„_–w]Éi ‹å6 ‘ˆ!ôps-Î\ƒ>³}£k^ÿñ!©«FÖ9>z°öo‹,Œtÿð¨f¸cw.©Ô‡˜Á¸[ƒÒéd1˜ôv¡ Ó/Ã'¤Cãaޱ#²nðÌ»[1ΞI‚…¦ÈSVt§aÉ&¹äÉê2,¶¾ ñ•‚ÕÈPÖ_*a]7\%™ñ!lаªÁ¶. sú0ì†ôËÿ"…Þž/ÉÎ!av—ss•Õ»$oü#²õ‘L×QïWø)Ü·»}/ép)RÔ%P=¿ÕŽT/½VtáAšÓý!¯H‘NµL §xéï éä™ #ÃSúõOž"c·m¥Åú2;½ÙqUcFV¦RúŽï$ȶòž’gó1rÌ2/ìÉ@®æ¸[²ÃÈóü•ÛÔôò!ûzkDùß…Û=ó¶@ÊÀ©øˆ>Ü¿?a•!Ô …úå¬_Ò¢()-•ìÅ ß°lÏB)ú*íQ_QÚbûšù¼/ÊR> $CÙ…Šã(·ªœ`è”ÎHZ/¢â§¼’þåר‘ëªí,‡ªWk×Þ¢z^ÄîÕõ¿¨I\ó¶˜õ/â!‡çO_ÝE­öÇïBP‡gEøò9ÔÕmˆs®´A½ eEÚ¯.¨Ÿ¥˜PÍp š‘•È. 7Ÿ1[Fcñ%û¨&49êú–nRMßU?xŸ„f®Ò虢…‰ÎëÓhù1ód“2ZÅO^*™Dë“_æ¿G=ÿýý­èëæàÕíG‰J6~Û'¦Ûš=½6~ÿ _D ™ ®oÞe@z²:ùZˆ4¤Œ\üX˜¡ o+&¯wQ$C3a€fN¿9'úÚöíý=„#t»ò®¿ùJNJû~µ•ñuþÏ<¸€Î¨1$9»vî ü4+²j\€ÆE•)1ä…¦%WÅ:hîŸj¾±÷¾–Úð‰BJhõ œ8uZŸ¾býÍùZ/Š|™IV…V¹(‚SJ~в¯u…ã´t‰°$Rn­¨©ý×ï[D·Ç»•úé~{|ɤ@{Ú_XŸeÕ€vÝçÆVÍõÐÁIözÛ~‰ÇŸ#ô¨„Î0÷½Éº¥_0‰¾‚ ·NQ‚Þ²Gš¦¶_¡?N¨E1B=yù¨®ÃPJàýO1-0–œ1ðFëÐÍ×Ä~ÝÚ6õ0±ÐDóW ¦/ޞ߬‚9#í"0ŸÊz²Q,'óçØa™ýðùúýë°Ê¶Å´óaݺFw ™6ëÚäeÒŽÀΉŽ_&ÃSHÀydܹÝ 3–„Ø #qèÝ,×ùwHzq¥Jëˆ2îK(-˜+¤GŠaG_뤶¡Õ%ͼ†´³S•ÛJ-Hß-›ß²vÍ徟ËUE¦ö !®0d¹FRÑ‚¼ÈæãëõÕ9žÄåøžDnÒà6âýï7]"õ–¯ò—ÙtØ… *Ñ1qlyáþ¢%9™|vƒ?eQTžî ÿí«(.ýQ0ýAJÖs1Ä?O\šAYÃHá¤%r”ëbï´;ýn<Ý“ C%à‘‘âŸBå­ÛÃE-^¨:÷jmY PCF2 4œ5cõ3gb­P‹$5òj” j[žy|àæ ê½ñÚºúõló%¢œ¼Ñ€y£î\P!Îe¼OŠnDãìŸMMÑ4®È°‡NÍOó Ü~’€–Hâ÷ F«À³ÕG‚fÐÚ>m>|ùÚh>Èy¤Ð¶6K‹GÛÐŽçs¡ï÷<Äaè£wDíy}BSŽç£ƒÌ~:ßa.t4/é6¸‡Naaë’Šè\žur=â6ºrz(ØtÍ£7ÙiY»0q”iôU8³Ï÷]¨rRu…!ÚK“#ÎP™Âªm¯Âkkãµ—3ÕÆ‹‹¦ Õ‰#t¢¬ÞE¾jœt°ƒ¤L‘{E)rðÆñÅ#$üÕ¼ÿ^YdžÖP†÷klAüýiGe"©ÿÕã¿}à¼KÁÙ&Ýzë__ÿ"ᦷmPÀ@.MáÉ(ø¤P[À4 ŸÒüiõ: -?Ô@ •¥ŒXÿóÍJq¥|(¢<6qî;åÿ½~ì}$·w_q}óFŠëí;/ýɆÒ¶QD ¬FÌTŠýßܪœÐà]¨tÌ›¸°U—ƲN®¾êª[ÏÏ¯Ç[½Áw›ÏŸÂ¸½ Æ+€û÷RÔå°+óCƒ=kõ)ajhŠßW½Ò£-*ê†Ä ôÐF{ô¶Á‰è`¨YŸ>fn³±ÏßB÷JwXªç#ès ´:kƒÑ„Ma¶Ž0¬¾h×î'£T‘œ´Sá7y¨q G LÈ•p19ÀÔv“¼j ˜%Z<6Õúþ¤gpfÚ®FlgËoXìÚ þö#–3å+õh`5÷®_Ù7KX¦á ¶²ƒ-f›^uØ)%¯EÙáÆ,HøÄ@„ïd/ÓØÄËçË#ÉëOz£-½HæÉ(2»„ä>üV‡Ó‘²VÄÁõ¼Òœ³8´{éåîO åû"C“÷©«$Èc–ù$ÑYüCÖUž¾B6¿c·¢É‹‘#uvHш¹uj®ÞãðA>[™î¡@t›Ñ‹Ûâ¸ÿuˆí?žä ŸÕ›D±³në1]GIY‘ Í·PZ™¯Ë‹ØeÛ¯_»hÎò?ùøã|Pq0õªíØT!–6òÐB5v¶ñrÔXl?©•$HáµùÆ¡µE‡6ÝÅP§áÇÓ?l\¨þšI;” ô­$ÆSЈ¿óú]!4^{\P•ñM5}WCóÇºÆØÑÒ÷‡°(a6Ze ¶ÚÐ_Gë4>É}¥²h“¼¢§‡¶Õ q7Ô¼ÐnìIPN?Ú³›±|¹¼†Gn| S“CÇr¯³ZYÑè¬I³Bj¶ˆ9jýÄPÛ X¬=ÑãLÝ›ùÆ$Ø4×68ẠM ëký6Úðí„£53Éu¨lð 9|DJÂúòíc{ (‚™Í/‡>^=dG{dnO¯œE1sÿ3GñÝ]MÈ:¾Å þxÒô CºLüzÖßlÈ Ò3ÞùÌû¯·f_k;|ál§:îÂu¼ªÎÐÔ¾ÝÔËë­ äY)= £*AÚ©º‚G·ûÖzw>¿°ðƒ˜o«ùq0$X÷šOÂ0ì”…áo»…ª0r´£M1¬~ï:½²e£]kËÊÖf0ÖöŸÐÿüÞog{¿w]÷¸Ó`Lº_Šb¬ó†©d'kô…߆Í'0{ßR+ë3üÑÕü•§š óZe²,¥`áZ½sœÇ#X ¿Uʨø–üÓ w0–CŽ g]XIˆ¨ý ’«+ЇxR`ý_YØTÿåÙº¶%—WGIÙa×<û–W‡ØÖN„-$#áãÀOïïs"Ñoó'òmQH¢õÍ»¬ø=’¾W=úˆÄ÷á•Ù’8¤ lO—<_‚”K‘ñ×ö# á³ý ²óòRê!}©d;}<2ÔŒRìîå@Æuæ³§£Y®' ¯±YÂ[º\ïO#ëTBsrÇAd¾k¡Pð9Ù¬­øÝ#WùÇÆE=9ä9½9³ñùdlk™åÂþ?>D‰‘•ˆp×j”ò Ì$9ƒÒ Ú÷'PVùÌ-2·”{œ3Röô=*ˆ­ˆæÌCÅz_îWʨ¬â¯Lh{ U,67ýQÕÅùl%瘟ܥ!1A ¡N '÷<¸B¡Ît›‘{Þà+ j•ö%ýÍÕEÂic¯ wÔuù˜7nÃŒzÉò¥~‚¨ÿsr˜¥Ž vÞt¶¿x€FŒ®ÛSÇÐX²ñìVH2š˜l‹ä éÙ¯wZ¾£ÙK>.ˆCó6-ªUæ|´d­¶ûù­däÆL£Õð7rY´Žw]-2@·Ã¹N‰äh{àœŒ¼C;Jw.àE»5£a– ãxÑ÷ã¡'ÇõPC#æqg²pRÏÛuƒ•é—²û¤Í îáÛ›WŒþÍqj*¾ú>Qßàß²§7ç7 ’ïRïP~>äqžn½°—K ²CʙζïËðŽzGð¤éŸ¾GØNÈÉOܾ’2gïåÒóòÖò{>زéù£ï_>ÛGÛ!%ãÿŠáß¼¶Wg¢Böü®´"€–ìß|µyëc’ÿæã­ T|}ê6=~ëŸO©%´ˆüLJ OÃ)(À÷»\Ø6Cøƒ—,yåÿñ!£®Yåëhâ‘g"„¦ÂsùŽ)%мm•[z’ZµT²ö‰íåÒç¾4Ç™¡ƒxŸz<ã_øó^±u•ºLšÚ…\ÚË¡!fIR¡Ï=ÍÛy0úùppgÜqˆ†­INî/7€‘Ê‚€Œø6;.¨rÚ팋)X~c „I^–†‰ 0­@êø¾f“¯p|Ï€yÁuþ¤Û˰pWÕ±Ñÿ,Q ­Í/Ár½ãÏaUõdôeÝX[{¤ä‰"lt_{!wølͬ±<ÖÛ»uÌnz!2§¶Å Q$ 7%<Äù‰¦»Ù¸ž!É… ~¿3ÙHÆ7šX™Æûz5Fšó"EÚrçè³&¤º\²mùËiv6¶Ä‘N¾xrïf‹ô1 K³¹Èðêüåý7‘±w­îÈçed¶i§ëwð@–¿çÖžB¶6º·jdÈQ™:aIJŒ\oeÙÿv ï†æâ1äSá>!Ù5ƒü±+Q2Ô(õeKþ™:î§U¥²cú’Î;߉"#¤zÈýQÜ*AlY1%}Ïl0=‡ÒL‹¨ÆS…2£;cvÑ(Ûú¢Á±ÓåfçÃ~Û ‚‚ïWß—:¨ø™ŸUmϧ”oP÷ z²¢jlOʲ8)ª7Êx;ž CM‘œ-Íëfˆþo}Ü>œ@­Ý?7íÈP‡MPÁ·ì'êZü¤püûõb’|&â#P¿çËù+Ô€†ŒdT9'¤ÐȆêç7/4~±Ù¥¥S&+ UÜQhæÊPíøü#š·+/=óàCKOá‹™·Ñ**G¨)Ðú8Ãð­‡[hãÈœjG¶‡+ënî—Ã3Z ?„G£Âö[ÉB¹I¤¬l(þø~-?ùL*ïæôòþ|Eoîl^ƒ¼ù[‘Ó®]DÿüÔ“BÈúÊê{eŠ2æ¥^…´¢Í§Y6!EÔä$&xÃ[S3úÊèyxý|p—´‡æ_‘rÕ„èß<ýsÀñþß9Ü…„ÖPj•TAR¿—ûvŽíŠ_ƒ¿®ï,FC}Ÿ…ûé·óÐè5E-aIMOÙf?‹C³kÎûWDТ¿í}vë3´ pKÊÛ~‚VfNO޽•ÜâÚçzhi5¶½-Î-þħ܃–t ®]æh†ÛÏ© Õ®É—o„Z;<Íܺ¾A›Ã{ Ý-h[à?R]yÚ»,›Y¡ãÆÓ…cÒÿô˜_FÝñ:oýðmò‚n‰¨i©g4Ðs‡³b· z7®v|Z‡þ¼Wo[Æ/ÀàaÕŒ+Γ0tg*š¦È†¾´¬•mÃè½ê—ìN=ðûl¥Øƒú6˜xþѯ´Û ¦éJx"&`¶TAÉÀ3 æ-’îõ”„…Ááñ—u°”üÎåD1¬ÄÇDlEµºÄûß/_ÙóW_ z>Ø®=SjYš ï+7 ¡}e'Ý¥$¬¾%bûIï1ˆŒïùévÉס H${{„•©:hk¸Ï"­§›¼ù©¤¿Gô¸7…6™TôBXé‚Ð匇jÈÂßç»Nê„lv~.È!äµËE\·=tµÜ‘W‚I~T)ùÃLÖeÞ«¡ÀBβˆå üӅ¬¯Tå¢ÈïÑÖ‚@`5`2EIŠ ¢‚9°r©:7§ú’ÛáÒÆP¨ãiô»Ò £7÷ùÿ›Ïatiøÿ\‹{ëBá³lô)ƒ•=ŸÔ ­±†RJ²gti§¡,çuêÄÍe¨ÐÕ>wé(TRš}ú UGCû)>„BõÃóTÊ3ðUCRÄOO¾mž»é5l 5žz·œÞA¶&O‰ø[øÙö ài¹ 4™Þê”ü3Í=K>½}-КbéQä`±§ï>Þ*s#ø5·¹iN˜Ý‘XªÍ}¡ïlÀnL‹' êgÒ=ošÓ—`{ #ŸN§EÁXa]r§E5Œ¤›ðrQÁ”*[úb%Ìd«Äñ<ú ¬ûz¦šôèÃ&Ï”ŸåÌØÚý@&ï »ŒúvÄB†H ÀAy ï}Îy$Ê*[¸¹ŠÄgm7tô„}‘ð[$[\´}uªÉshI"¨òõ[Á~B¤Îp:È}ikŸÖæ‘"ý¥Cú“¬È¨dKzc›Ë£§ 3ÏÌõ_‘ÅnN×#O Y+YJ‹(ÖÝJäÕã¥È±àõëu*reÆŽ¼µFž£¹TüÈ'¥]g¾Q÷ÿñ!Kò¥Ùœú™v¨K{í«<ôAá’c¡£å(ÊÁ–Rï€b. ¦ÅÚÄ(þôî ·ÓïQb~Ž%)Ð¥üØ|žº€Ò:1¦(«Ùëôãʽˆn”RVC…"ôTd6¨XOø³2•eÈM\Ø@Úø;¨j+HÑxÕî…m÷³I¢å@“øƒU<8¼LËÄ ¾Ç‡ô‘{¨Uïvϯ,u*âC¿D£îõž¦Ý‡á¨Wi,…™ õ‡êöûŒ¢!AEÔb°( L÷å^>ƦnsD$hÞÀI«hú¹õÈÄs4'±ºybÙ-\‡Ssï eUºHgð5´z¹âéLBÖ^«Ûp¹mT Øіm¦¿››íöù½“ŸÇCäZf <ªhÏ#õ–B /^º=•áˆêb݇>óu#ËêÌvÔq^X9ô÷ì-Úê½û»:ÿÚ]¨¼ò{ŒÂáöÿÔí—ÏL@^ç¶ë¸=äKšÞù4C©lÏ‹Aº!ç¡Ô÷ör©ïÅÚgÿú)¨CVâ i±€²$gðß>ÓÊj™º=?=Õ'ÀÒÛ°‡òêøP›päõ× ÈMá¥¼Ï 'W« z© ÐiJ!èS&{ø_-®ø eB>ä6_CEQë@PÜ9¨ %ê]–¹ô®]dŠ_ç€ï©s8Ó F&ÐûcÝþÿø]Ä…Ê[µït†™1BSg;k¶µ´H»]»]­—huú´k mÌþgñt|Mug¢†_Ía½¦#;Ðu9þÊ»Úèñh{ç8ã}A&³¥Cÿñájô¡wöwaX«žÔ0ÓFRßD=„1ü¦B©7¿žžâ놉–º}ó W`jä¤`äH#̪Jf§qŸ:-×Ov7aÁxÎТµ+“Œ;`ÙÔœåÌ9lW}„µ¬šm\í°q[ŠoÖ ¶â†X/¿ÓƒÏŸ—G-â‘€¨i,n¦ W–rìé¹ÔÒagñ-’ØÄ6ºKr!é¶ßnF[;î+ Ÿr?ÑÆï¤SëÊcÌé´í=ã‘Ëp4G:©¢ìÄO¼HÿˆÌE1æ*2$õ_1EÆøv¿°Ed6Û'è”,Óò¬ÅËÈ–ëÐÎé8¾!dÊ‚2ùõ•”·Pö UØfÊ¥;Ù^TB9¿#÷ªcQ±äÎÙ‡R>¨|÷„;ñé-TMœq4̰Aõ©¾6©vZÔ4~nS¾òñîªÐF­ÁE7YÔañÿ´+†ºNuašK¨—a—\埄”vH¶ƒ†6? ÝC¡QÚcåh‘V4¡'ò“ ÈAÓèoCâß{Ñœe·šïz Z»z _^B+ ‘|õ‰ÖÒÔ,–nÃh#«nà¨Éж†^Ç+O¡Ýy†??ï¡@½¤_1Ê¿k¿ñáÄi¤—ª“ cý6Ú2ÎAÅrÕ©A(ì9âÎGµÇsVTßE![%}t{àdV™§›Ô¾ƒôéŸ* æÿÎ_+ÈûïNsp|{oÚß?O'×€×4å,16½ yý=mùî?Ö)߃ܡùôÍqPRíŸaêÕÉü\àÇ­ôW$çÌ þLà¾ëçç¡á× i)·)h:ØúîöŒ4mèôJM‡_k ”ì&¾Ð¥—?_ó:º¯õ¨P†AOg¡4I\8ô)gzpKB3{¿E :ت¬¹™ÂÐ9÷Ï¡þBC ûŠFm {ª¼mà·ðúQƒ^˜Ð}ÑÛSo§é;ÂaVÞ\¬ã¥"ü™:æÝ”uÎ÷øœ3‡%ù?H ´°Â7Ô°XN kú¡=œFa#\ïQðl yk÷4v'õÙ•«¦0æÙËU$ªÛ8þçc,’Ô¤9 Í=A²ê—”£H±ïO¹C·R9-tß¹ã„4Ío^½Ezí¶àßö¥Èðn9…uïzdâî1 \Cæ’<ÊßBm³×·8QøiKòÝé 4®su~¡ ‡æˆÌØ‚ÚGЪ¥©.e°írózÏ”tá×eò£]¡›žø}¿Š!ônÜ‹nV ƒAb{‹"S!J|ò×¥”F¼µK«ÝaÌ›—pùjŒ?öí,‡É g*¹{S0s8¬X·—æþªm<úós í )Z°xûaç•uWXöLñ9åP «>Ó÷¬SëaýÑRI¨lv=õá.ç…=‹ûÙA'nXKH#¡³Þþ<ªr$júë¤Ì­Œ$n]£w‚%‘Œ[ô^YX’s÷me^GÊ“hk6[¤a¥ØWkɈôDŠG ‘á5½ùuqdrÎrh5ãCy¸æi‰l‚ MWF…‘ÃøÓèS-äúIW®Ç†¼ÍMN|†}( çKû‘²÷«ò¶Çd£ð™K†\‘§Pt@%\ò¡3JÜJ=QG‘„.vJN þDYß?ÙÅ”w#[º$ÄŒŠ¾­Ï÷d¨ü¸ã ÃÍO¨úÆa5⪑›Ç#”†_I JAm2†1Ö‰&Ôy4o¤|õÜì='dߠ؉,Ÿ¿hDÇ#5¥£‹Æ‹wëü'‚Ñ´e×ÞùÎ 4Ïu‘|&v-Ó×ÇX; ÕÀÜ—sÓAh½i#ê\Dж"E!¦ühwbs:Lô#ªä‰ÀÿGØy‡sý¾ß^Ù{gÎSöÞDFˆJI(¥R©¤]"ŠŠ"YÙI¢”‘½C¶²²GÈÍ÷¸?}~Çý=~Çý×õÏåý~;Ž×óz>¯ë:ÏkŸÒFa’ :Öy¯GäàþëÄ÷ ô#ºœXg1ÄPÁÓ›GQ[*ÄZJ†9Ì«éƒÏí†Õ‚OÌ'Ü¡±Šé màÓ^–C*÷Óá™ý¡MxKâèñ5ŒŠwsÿŽYÚΑ¿ß©ÇÂÓ o†oÓë¸iÕð^EGžhl…Ðz÷Xœ¼”` pr~p®Šïì£çNZl?×ï‚© ¾Ï­ÈȧZhi«ÅŸ[@»Ë×dç&è2 à+â==ýSÆ“CÐ÷Ú=êªw" Øüê‰´Ž‡Áȶû{d~ÃPÐȧò¦j{6™Ä÷B-ŽÂ§ÂÔC1ŒôNêéFwÀèð {Ôe…q1é"ooøq>]n¦=~~ö¾¹I “¢fÉF£`*@×¥¥ftïv8jÁ,¯¸ø)?˜Ý|/þú®#ÌÍ}Î?óße¤š1À‡¨aa/XÌãæÚuá,UFtíMî…åøï[û¯Âªçõh¦æð›`3äÃq9Xoöa Í¥‡ÍŽžä@W$  ²º*x UOr T#QwiFâ÷Ì (Ѥ|AôE{,ͳ¤¬™)ÜÍúnìG*ÓªŒtÚ®Ò±íon¥×kå"»øû_®l}ÈÑ}ö^¿rÅtÙ·ù#ÃÔ™+QÜ-.[½’ó_>|Yw#üm. ülÿ1S{›µÞïµGárßdÿšå5ËêA1¯Šk¢Æz(þœ]ÿ"u JlŽeÔÝ¡E©‹£k'¢Œµè¥ÉWa(§¯÷Z¾ åŸ(mý´íƒËÎss¨XûùÝÅ‘ZT7—nL9…*JýÔ5_ ªRM›M(5ªù²dKP£úìxÌZ•+j6õ1>ÒBä’{—~Oµ{¾Ÿ¬D½ÑÓ÷úrQ7`†¥m õxGïBý¥¬Á´…hHÜ/OgùäÏ•Ÿ‹}‹ÆGg| n{¡I®˜ÊZÝY4#ÜóÇ7ÍR~ú:Z ”8[D+£ÑûEhMñöaÃii´í<ÕŒ6m õDRÃhÛ²kÿ#1´ôó”UC2uµê ܧË)*÷è6ž¿çq5`ÖÕX…XG–ŠÐr»¯ßa±Â#›ú—:ÔÎ4<Ø6iï(±AI(cÙ>Ÿ?ðæÜ÷RÊ‹»œq ÷ëAÖçÎŒ®Sþ“`‰GÒò)äÎÃÎyÒå›OµvÞ‹2=÷L‡Ô#6å—MFáÅ5·[3Ñžq7óm®ÖÈö½vWþüiÈ ¦›h „µññ‡ ‰Û:gM ö°†· l"ÜÉðnÉÙÏ·p*²èléï‡Ê“|¡†w:þåÃM× ™±øëPn’|j”ö)ï ÿ—y)}ì÷Ü„F·¯Õ×·ó­ÊYòß+Ðâî븟ŽZߎtÅ$…vq&^1†qè¸Æ*ÌÄ]Œ¦W«š— ;ÿYÑ÷| —' Ñ£€¾›S|C ÿòá/qçWœ0$•r,ö$ ß¿’ã£G•Ó%u`¬t3ìÜ7ðãz¨V£ã ˜wÅ×ìÛ~VÛaø¨Ø~YE ?€9²çJ Ê0VÐ/eö3,Ül±ØSKÏ }ýöæÀŠç¯ìɤ3°¦Ñò,]n[ÏÚulG)Á¦sFYS+l=¥c*OFBÖ¤Ñf}$º÷Ñ~¤& I$r.ZþGÒ®Mç‚[HSþ$‘òPâÉà¢jÜe@SÞøiä4ê HQ0I~Ò?z<¤`‡ ¯ô7IEÆ!ü~~!³¾šâPÔ&²tLÅfÝÎG¶ˆ–ö7Æ‘ãhSǾÈepPù8Cò°eýPèEÞqÃŒãŸíïØèð~[oä÷æv¯YBáCJÝVü(ô «ãÕc?Izä/'àƒbK3ªšóö(yH1æ}W=J'ô½¡ öAÙ°8é;“(—Â(™‡åû¼”ºCÙPA6`-ëç0*–¯~™6²E娃#§P5g“â‰ôSÔ »ƒ„º¬¨å{%CÇk1Cƒn¢™µÇ’¾ZR3£‹D¡ÖUÔõªùe¢‡zÕ$ýñ†_Ð@uDh“° £¢Vèî\FcBFë:hr•Šœ©<͸C”‹Ѽq²r,°-“/·å&T£UmPìcl´îlŒé•Ö¶þH«Gø5þs®ô“õÃmý×™ß*üOÿÄÈëáš ÍY›OW—S,Ž-BZŒ`µ­äÝ”°Ð"eÊ)¢ Ufß®>,«‡/R ¨æé¡n36+Ùô4˜XXÉø5mû3oyÅÇ2h ÷ Øs‚f͵ªßW9 ùãÑ\$:hÙŸ¾aQ -‰äû *¡…µuí+ý6 <¶žo8-¦b¥ 6Ðò!Ú—_ý.´º?ª ºY mlRžTý¡íëÇoA Û|!rœ :Üö¹´íÙæBûêê_ ËmÜA°\º_~çò»=ƒò7œŸ€^aû' ´üÐwâððt¨+ôÏDҬ€õñÇ=^ÃྎàR*}б—¾AFXê9}ƒka´WVA¹bƧ®*k˜ZÁ„*gdÒÇý0õâ0÷¡ŽRø¥r¹àç·h˜£m3*Ý”…ù UÄ:X|®Àýs3–ßÛˆ>R4€5¶óê¾§°~ÿÉM§“ðG÷zÙŠ%9dÀÍw„Û¾*«[ÂqIü¯ ç"ÙÅsµ§Í‘".k¾>Ñ©ú†Ó,7t‘ÆBŸñWÛÒMšelûƒmëâ&d¬œV¸R€ÌŽ÷êt4"«˜ùeø,²á#÷×SÈùH/îlòˆ>úè‹»gOŸ©çóDþ£+…°ú&Uìÿø ¢Ð ¾?4$™(bu?XdÂŰ#ñ ë^”ðl%b»{¥ª+=;o¢ìݸÃ?£¼SùžT S?¤ÎõóšÞÒø£²› ÝWR{TÕ§¼«ûÕ¯ÆåýðFÍY¿o{ht/|ùàñº÷²Çë”®¡ÎÃ=~‡êRQ[3³D:Ñ€z±‚%“ ÇË®>AãâŒÉÙ„hz¾ìH× S4÷üÚôæVZ*Å~L»8‹Va·Ïh}Aës cBÖhsQ6ÃÎÑ6bÓí÷>7´ËñvMÿ6ÚOÚq“ï#Ä}*çÓ/†ß@Ç'k—ÒÅq?÷!BŸt.Ëz_剮—;fo^À²Zý‰]xH71³¢Í ÅWY8ö,@’ 1þÉYøž{˜óóØNÿà©·Cž1ÙnYæÈäyÒ˜æiÇØŽôéÂó‡O _å2BJ¹ø½ÝQŽTœ¡Ürü1<Õ¬{t,Äj§_iûì£;\ø†Ï¨â‚Òí[ÝÃþêñ9ù9Þ{_‹!½d1’ùˆäHŽDQSÕïÔ]L‘X~ë+Ff P¨Yzqw‡æ¹+i‡WB‘™ZØŸ@(Jȼ5ý’ŠÖë8ªÆS¶× [ñs3þ÷qœcw€:åÎ9S®7ΧÿÓÇôDHäöÿyÜN$ye§žÃKÿú+¨Jôðº+ ˆf{B¥H šò‹A•’åTu‘|,Éûž¾fÕÆCÒÊ~ÂðEU7ªYø)ÔzÜ9áOŸõÍ×c—ú ÑI€Pn4 {ß+²‡–nI†§Ê/¡­’ø¨?tÒ±ÔäËCw»á??ô^Mgbx ߯reßc†A»‰¶Äá0ÌÎ/!LcÛWýÝè‹[`ÜPáD ü|ÌÿT-LQ}u7€™ôÐ'Í…0û±~å"Ãq˜÷r»y `=*U`Ù„3‹s¬úVª»û~gô¤‹€MfÓÛÉšù°•iÓZ@/‡„U!}rH”×^¬Ò¤‚$pLÑݰIWótõL‘|IO¨öµ5Rö&Õ¶Aêñ1÷™d¤k§Ñ$¾’ˆ áß“‡‘ ö½à?ƒ,ŒÉ¦ã…bȺ¡²p±¹98¹ÇL49‘+rÏ·Ån_ä½yRÖŽ™ùIKç*kÓQ8äÈ+–Í’ÌK–DÑGË»mõQBå§›á—4”æÍ·—Þså4½oúŽ{DÎõELü@E¥ÓZÁd¨|@üÇ2ÏITÍ·±x1 5NœHö°Cxtó“dØÔ^ï 4BD™½nÌ®¨g‘G­qÇ 8üDJ¡†²w´ ñ,e‡qxšvñÅIü’AóÄ«ÁµhYCgþqM­Ö–Þ;N£Àdj˜ò,ÚzôçÜE»’­èåÄ~tùeU÷}l{Ë'w .Û³Æý*G磭<ïŽD×›©'Ä¿`è¥>ö+föˆ¢Êý "7C«ÊPÓÒV&ÈÖOzRA£ÿËC¥ì¥ðñõÂl½:T¤XXy“C©Kñ×Pä—ì+|ñ;äûÐ”Ç yÒÔ[lvÞ·ô¥þꆌT[*ßZHÓ¤­¢–ƒ¢¦ñ…Þÿ¹¯†Ñ­z§WÄÙ” >Ý›û“,¾¾£Y®‘ƒ– ¶Š¦ÃzÐÁï3NݔݱÝo^÷Hºá­1Eè7~°µ<§ •©dìša°¶Gåj_ ¥ðÓk…a©÷±^…w`øégÚáš‹Ÿ‘~¯)U.øñ«_òŒƒUÆÕÛ3ð#)‚ã@’ü\aéÐnÉ#&»TSajÒhó-)Ì$dU½¾³ª®Ÿ›ÝC`N¡@äu¦Ì‹–1S¹úÂÝÖH®‘,ü¼‘SBà ‹ ¿UÔÏ…ÀRëãïTß`…ÀZøtð,¬:¼ºÜûÚÖFèoõç‚BäÔ°ùÄo+aëK4%up R.®7ûÆ#‘f˜™ßb7_ÔóòA’î¨#F5HæÔºû›Š&RÐ]o/ŒHGÊ•‘=zk6¸ëßyPÜæC®8Á>â¤[:{&Ž,;G™> cØ!ÅR—dz—Ò×õã ² ³ù­ýDÖã[Ñߎ!;±Î}ùƒ¥E÷ x$ëŠ{µGò˜yk¾‹–ÁÝ"<_?Ü:ú?øðò±{7CQàG_‡Â%ØæC¡Í‚_€Â†®‡³®¡¨à9¡¨J7;žWeŽâÙ)3 ¯GQ’ê;Î?ËG©;‰I¢f(s ü)Çr”3‹}FðDå3”ÓȬQAézofßT¬Y<+GÁ„Êâ…5¿PEÜj>½¯ U ßÈ% š‹³L¨æ4ªwür­Yüˆšå5…Œ<ÖˆœVkòµJ¨ýÓÝæ3u<Žá³ånÔíÙ%Íj‰úå¦çœ·Ð€òljœhH¶Þ'%n‡F{)Ÿêä ñ­ ¾ÌUh2rÅ~xß:šYý"cFC4o‘"<ð-}nEÐÌh Õ]1·I´>y¹Zç‡Úø8æ½ÚÍ„¶þ~x},ínïRùƨ†öo}Ïçp]Å}$Š"ñß|Ñч¨¦\¸Ï'ªpPºŒªÄÇÛÏœ:“¸>²sßw}¹"4sÄ~;ð<Z"ÏIŠyF@ëÚ[?Ëujh÷m<­t¸y§oéÕ &Qè²å”Ôx?Ý?_͘ž‚oéw><)†¾Ä»ËW‡çþåÃñZKú­—0ÄçŒ÷Ãoîœ#HóV/‡‘Ze#†h óLl·»?4•NfÑGÀ„¸K§ L¹†ö‡ÛgÁLû@TÔ×Z˜­Ú/dþXæåRÒ×gŠa¡ ú>í)[XÚ¯Üï]‘+ì,]ÄÂì°:uVÔ›ã*üž(`·;z66☺©`K5ÄÅÒ? †M»èØŸ ‘»šm]R1’5Hcê<’Ü,•B$?¹eÀÍ¥ƒ”z Ú•Õ•¸K\™)Åti¸Sú£‘ޝçíGú'’' :„!7Ó[Ìq G†FL!33ôÊ3E–ÊÜɱÈvBÑïV±#r ×ׂÔäbÛø$¹œÜS”] ¦ÈûÞÕŒ¶œùÝ-W!`ÔÊ |Ý4åšfB¡`â€)5¹Äù 'Û Å¾Dã µ*JJÛUüôJ{¶úþN}‡²vGlâèLQîy×µOx åÛy„ýFé“Qô ¨øÁ[ÀÐK•ã¸~ºPª4Å3ÙS¨! žÞžË‹Z÷~lSD¬Ì\Dí)KMù\¨Ã,3còuO|ñ´-F½¶Óé…Å^j$âõæ»r IØ\æ®^XA2ž’fS×"¤öÙëy©<Ú™|¥‘ºªøê^ÊE¤³™ XðeG‚¢ð’š,d£Ûq3aIó, ·#©äýxG#M¡O½‚$ÌÅU¼ø¡è©àLw ¼ý•–9ðxÛ÷l MæÃ+òûMb!µÚ—ÎMó-<ûµøqêl¤p\Þ{§’$¯f1üôŸºÃ¯OL ž‡Úx“>â ¯fŸ‚8¶({òvÓ¿z|–î)çF¾ézvo$ÈýwÎÕÈ»hzB^ÞàË­ï¿Å’fª >¶ÚA¡;©Ž^ü[(lEcìúöú`-9Ç E wR桨båá…4Ÿÿu,Nº¬ô6Àú?÷HùĪìômô=ê°Ê’øróxµ üé‡E²p¨àëžs}x>0²»±ÊC%¥ò‰Éî-¨"Ót½ß ƒ—Ë Z;àÓ\ìÖnEzøÜ¦ðöÔíx¨™Ëg7!‘z+¢;B÷ ¡TL<Þù14Õòö~Î -ì?Eì¡Í†5ÂT:ž Ê{mç„n ©AfªßÐKS•`½žý›ö±S×`B+}ˆ²†>º³ë³ÁHMe©ÔtŒ­®,ŠîŸFýuÙûÓa²˜œ>~äÌØ~Ú‘˜{ÉAu>æµÄյĸa‘‹›ôX0,‹¿¹m «†=‘‘)*ð;„ûÚ1w"Øès$9õ̶üy-Ë#!¿AV2=ÌdS SCž0EŽsHÚ®¹B²«É¿¦‹ÍâF*n¿ý¢m>,©*‰¡•GºÔÜü˜[Èp¼Hî&q?2 ÓÓ.7 ó ã™ Bdm¿Èzþý=dŸ¸<¾¡º¹öÙ¯ùúU"ï^VÒRÈ×CÞÖxë" 4žø¹¾5‡BsýôÆï'QÔ¢ç“,ŠJØ}¼žŽR".úë ëþ±Çµ(?3óùYn*lüñxÙ€Êb¢{reQõŠw¡Kyjè>£l\¡Aˆ¡ÈŠº‡Ú#{¿f£ÎÝß ]GjPiòïýPBƒ]—d¯9¢á¦G«á1O4ž¾6È/¬…¦}:}ëÖíhÞ˜ùÉñØ3´%]KÚDk¦‚¨êm¿3½&ø:¿mã××x}~£=5‰v+9:ħ  £mæ˜Yi4Î%z³G5» s>¯_×…Á¼ká y27XÐnŽX»‹ .šMÚÃâd—­Ï™°4’4ÚÆC +$¤G§¹ÁªI£_jŒ ¬}Œ~î—ë§ûy\)`Ó8§Ÿöl]ÄÆŒcHð-ˆñzaOöKg˜ ±–'czì’Dæ¾ï¹ôÉht¯­%œBòò*µûW‘ò9IÉjîzqŵ!ÍGΦ@Bu¤Û¸-’3ʈ ¢â2œÃ¿Ñµäa«##2ÅŒ:\žFæíËAT±ˆÙgã£S)6šÛ|HÍ™™ ŠÂ•nQGPTì•çŸöA;ý|eËé3Š¿•~ôÎ’%Ù:Û9Þ:¡TÜ>ƒþkJ(ãg_nx|ål7_Ù‘¼Aù\šA´½¨ V\¹Y#·Í‡üŠƒ’.¨Ì%.œ0wU¸«,8.ö£*½šêð  ªéšZ[6 zÙŠPœˆjfæßbxæÈ XJ¦Ú+ƒ&U#D¨sYûA =êQ˜Šéí+F}m×?·Ñ€çä ªÍ~4¤â¸iL4FVC©ôÐ8ƒæÈƒDA4eë=Y?rÍžltK0 ¡…Ç= w´œ}v¨â#Z ËZv©&¡u¿áJÑþh3yóî _ÚQ×¾¡I E{½8/ç»&è}4Z‰®‰d˜KÃÑéÞ¤¯i¾,žÏ¢Î5;ʳEÑvÈbûrïR‚,D‹È0æH@­ñæ¸Sï#¨Ð÷x0+ Å»ÙòxJB^yêÁB—wk½f$²Ä ¯{þòW„t©Å›¡2f{_#¶÷æÖfÌßçÞ„O~TßÖ£AÑæex!ª?œ×á!òx +Çò‡Ý˜ßÎ}ÀiJžÊÿ­V®t¨ŠÒOܹ¿s¯Öï6öÿÜË4${*bÂ=ŸIAåaçêÃkÿò!¥@§]ÕøœÊÍ![jör7Æ1˜ÿˇâ×~|񠮯Þâý+úЬ¶¡Ud¡-¯o.RmçÓHÊ}¯¡=çvƒÎ»Rèd$˜ËfR„®¨-o^@o¾PÂgøVÿóMÐÂè{#{nl”ù_>8fZm§ Cô½oN&Á°«O‘ù'©ÏS[‡1óÏ%²ÛúŸH¿óåÑv56n™¾JÔ?Í?Ÿhl Aör­9Û,ÉgYç©BâžÉ^Ùƒ®HÉÊrtß"’[ãÂoªçH)hDKÝþwQ»ù¿û"†4ç¾9Q‚tœoÈj³W‘>ñÝÖrð"23¹\2vDÆñë )•Ȭ’à4ÖôY²<žF•\B6óòÔñ/ÈÁ´ú±G’9ÆØN|{†Ü^EA[‘÷Qàtv~òéuè\‹ãCþ¤Z÷~ç @ñCw·[$(äöZNU›E\Þ\ÓtG±'š†×r¸Pb194÷(J‹ÖˆŒ¶2 ¬ò²™(½·ò§Ç(Ê×wÎŒÞÜÖ­dô/å0T¬j>ŸÊÊI§GÓÂ_¡j#¯Î5ÔÐúÉœ±[µòl u¯ µ¼9=ÿµç6_1äG&©¨2uÔ=WÝn½:ŠzëÞV–‹ôhCܼ¢F,üCÇsÛÑ8‡º°ÈˆM}8ÇÕ¯ ¹îXE ¥î§ÊS¦hU8¶dª­‡ƒÂ¦9ÐV‚{¬Î­íî½ ø zXlcwS^Ã}UD.±|ix¤zÝÞFòJ›t«HÐ #Åï°´­C_`äLÜYö!®ºÛÝ)×<¡ðªåø†xÈÙJzÁYy^Z¸yøøð¶ß-¸¦ú‡çGé¾@õF»Å]x®|MÆ9ˆ’Ou¨ 熄CN.²¥ðT³F&Gwøï¼Ü‰Ñ«c¶o¹ºŽï¾T³ÓŸbkË>û ³ÍÅþ«Ù¼‡{î=šÆ‘sÚ^QÞФ°Â{*šÒKTô¦C3Ñ®‡ÕeЬ9}c!ï$4ïwK¢]i†æè½/ÍÓÑJ§m¤ %c ¿ÑC«ï±:¹ýжo)¸ê´ëF$ò3Ý…Y¹XV¿èäl÷é8 Ð%¸v-º¥9fÔSç¡'@Á§‡º¾=êtû ½åAâ›>Ð7jbKbTßůM˜…†ý”>§#0(PésÈY†ÔGoˆj8Àp|Û£ŠG%0j” õ-ÆÁr½¸Ì~ºŸ%¯Hb†ÉOˆ¯éæR n•w0«YMhæòæzj~¶ÊÂBf{q×y*XzCôþÐÏ]°²Øá.ùj7ü¶«±?(ö6ViÈ‚ïE“ë¬Hø9nÓô{—+¿³Ó¤e«]×ù‘¼BÚ©P¤ )G?kuô²!µšäB$l"méCE²º#HÿúñòíûwqWÁ K <2Eh‹Þç5@+ÃÃõy…ȶ/Q.³³9žÑý±{c†Üâ.¦¶ºÈûÃaKuÄùF´‚÷%C¯ª¹Ùã(¸æé2eŠÂ¥êÜímª(š3ÛÏ¡²í“½5RÞ’(Å9@ùÎ8eÿœ;Öƒò$•ÍO$øpOöºÃlÙ8*Ú/|â´¯E¥¥“WeEMP¥u]™ÜdÛÔ÷\õ5ÃnYºPEä«‚…€s¨½H_àÿ-uÎkív\?‹z{¢kIùÑ€<ž¦³I ‡ùÈ*§Ð8Ë£ˆe®r[Wå;²íÑÜuouƒ” ZjØÜ¦¿¾­îÅ6å ud†{‡;Ú¤ë¦^R¹„¶í~ëV^hÏ¡_n{θ‰h“ÒdXååÿ] W—êï}‡"¿'[¡n{ÿ¿c±ÓO3Ò—±P¢¯òuwÇ*”¦É6{˶î܃H§wõïºSÁÂrÙáÉOø@à׫²&k“îDv@e§XƒÃãÏïÄZ7“¯­ÿ]7j.û ¸’u .ÚŠ.аAhòë"ÑŽè‚–]d×X‰ uH¼gzÏ_½wí='³ØöW×ýÁìâ2aàmÄ%a9 ”"]€‘s,¬Æƒ0–QvEB­~¬ñKÎhß„Io3 ã#%0ý‡fÀt~fý¯ªÇŒ~†y–…IÎ XXî8Þ²d Ë$Á¥¯7nÀ*¿ù/§Âømîɘs 6Ò¼.Q®Ã–ÂÏnö^$øó>”‰N_9­¡ªÄ›wÛÈ'ù‘4÷ó+‘Vy$O#Ë;’5†”sÍÛ¦J};™ñütÒv–€ d°§J¡ºZ‚LTà˜Y/ŒÌ­‡ºÉkz5ë£Zðk]d/û™¥(„\ÜgYGHØgµøÛ»L[ä˲+ê5±E'Çï«<@¡|šÅþL%{õƒ¬•Å#™î5x…R^¾/ÞØ¡lú©¤+>(_ä™[ ™ú¯vÇ Ò\³Î{½ITµ:˳öÇ5Ø:êªLSÌäS® v«Ó'qÆzÔ¹ðÙÅæ"êÉú¸ én¡þ&Õlvo.†‰mó©ñ$áäjU2šyó×几y¿}¤*Z==À&t­eˆ³„zÑæorçžOhÛÅ£v}áÚ»QLLÞ•Ã}T^ÇD ãÐ5qW«Ï¸ZÁ,Û']÷j¾ÕJÂenŸ Þ`èkËÌ_-Ûyºêý·QÈÁó.¿j­ –mnÕœ÷ö…†ºúBš··¡êz.éÔ€ ¼Ï¿Ä*þ$Š#¼·¦Gþêâ͇ÀžÃýí=ý£¯üñýæ0»ü©­¿ó÷ù¯6þ~Z[ðèÎ~BPí«Å¿ÏigDzJ ý"ôÜ˸ÃÛ¯½k—[õtÁ÷Ë‹]ý^ÿå3C¿cSµaØ?:Ìa؆k˜‰N}Ú#r¶kÞ8ÁÈ £æ¾T]ËæÚ¤ñ£1ï»ýàÇ„ëÄ^eZ˜p>WTöÙ&'Ôâúk`úzOà™³óÏø ‹j¦°DQxâñ©4XZz9•°?VÈKr¼ êaœý.„µ å–_v°nüH÷ØØz|.Ä%ñ_d.‰!ÖÙž˜:P†D76F¬#Ä´‰Í—7I+ÞL±¿BòãŽ$"¥•ðÞWs¸Ëj„EÛl›}Ÿ\ û³‚t/$÷ÆP"ýŸsgˆ!£ õâ”­2ù‹6J¬~Fæ\çÉóÚ‘ÈÊùÓ´kÙ orIr„¬åÇe"—Ó­ÄôÝÈ£âðY{·&îÞ½Ö¤ÄEù/¾˜–OæA±ba 2[PyfJÂUíöÁ¦(*Ý›ìóÓÅ.=’?æIâÕÅÚ½³\()ÔtT»e ¥^ÒMGÓò¢LÈT_¥<Ê9{¼yá'¸­«‘¡Ì#¨ õ3Ý"¿¿˜ÏFZ¤¡23ÿ5öQT¡WJ¿¹{ U6ý'¥¼G5 5wÏlTOù3ytßÔŒ{êÆ ~‘ý¿ŽŽ=î%óTTœ Dd–áq ÔSð×wŽåGýC›¹æÏ~ ¼ƒÁ¢Ò¨(y÷†¢‘»¨{Ëh\?V$zMM•¢Ù8U&Ž\@‹$rõ¸ºm.Ôèç-y¸„Öj%Öw™ÑƘV\²>m—ú°•£]º¨HÔ~K´ÿ#ÑK±÷8’UJŠNbF±I¸?¿¶[ç2ž7d¤ö>•ú í•è©‘…G“¸šTæ·nŽH†š•‡›ßüÕß?¾ò_ý¿9õÝý“;ÿÑÝÿ«Ã´UµæõWsm–ËØ“‰¿:þÇÿù¾|ëÝÑMîÉv¨¸õa0Þ†y§NþèÞç}ÿò!éññ< øœñ§•M™j ¶â5Ärÿ+ß6™½N•sø›[[–£›>Q^ƒ¶ #±•ùоh$bÅô:íß8Å´ºÚŠÎI>é…Û²ý³L‘ðmÖ€F”ú>7m©üˇÝÔFç_VÁàÖÄÙN÷BÖsø¢lé #iJ×úÕ`ŒMªõ–ÀxêÓïÇBáç ÍG’Åj0y¥ƒú`„3L„Öúîú7‡þ_½/ðiSÜ[9‹Õ,•våOaùЧØgX5¯–éƒßp“S\¦6à“îÁsðç˜Ãtùa$I.»Íf€DÄ‚–±_‘ø åíÝ|_Ô/%Ðîj:’Ë„&wþ&EJÂøo#ÅHõ+ük·ž+R/Æ1Ó×÷!sNƒôë¤O±º£Âõ ÊÏÙv¨å!ã„äæ›°qd–kô=23†,qÜáÛùŒMr%雲ÿ¸óöOÎ!ä|+;”^ƒÜßG¯œo@Þ7î[½òG>y‹@.­ Èojo!ÂDƒÏ$ƒ’E–QH_ÊåOTŠhòµËÖ¸ ØñaK Ú?(ñ¶s)Cª ¥¦Êš ¼Qf‘äÉIfj”»öŠ€&Ìå?>¨;vï=*ˆŸþBçÒ†ŠÕZmëñ¨œÖ|øúýTí—åˆÎG §G—.Þ•F­vœš¼FÄ-ÞkZÄ/P{%“c%I uUŽö®G£îµOFV*ʨÏÊrS¼V >F¥+ÏýF#û]Ü/¡ ùÍÁ€—hÚËW×´ˆæO\]¯i¢åf^¥† Z‹¯1I¬'£Í‰šáó{жõL§ å ´wüÖ~opgÿ™Ý—ÓÃèØkÚÇG‚GèkÕíçKQÒÑÿMU R¤ÍXT:ÂÐZgK‰ýOx§½°_}Ï,´t[/>䜄8)Êè_iwII–4 ^.ML‰ï¦…Ïæ éÞBª¿D6oK<«øAêòU’¶Ô–3Ñâkn‘^‚DëßãÅü—¤›ý®ÅJ/®‘†ýÞ•rŸK8 z6Y(Õ‡jo•'ݱ3‚ú4G ³úSðµ>7üÆ«„mŸnPõlºë¼Ìoíä¾O§H„ éÝþ‚*“(h‰?j–ÜM ö'OH߇fÎG)Ž©íÐìGôë×ê7há}åÚhg¿­¿ÇíßϹ@ë8óB¸@[ïÅ §NÐþ•äË£,èx!F­ÐYýZ!ðÁ¶K¤/‰CÓ¸Ê%Ù+;ý¢žÝÖ„Þë<çë¡/Ã14²Ñ¾©¯ßs‡Ðã¤vþ×·µögÖ̤ÃpÜóy­$U9Ì8ë ã¬K•ÑGá§Àò¤4ML~œm"U€ék‡âFhËa–­e„ªbÌåÇP>9/ ¡×câ´aéÌ—â‰Ëܰ’ÂÍ1EKk×¼[ÞÁÆÙ=W6 aË?Ã-”m  ã‡_Câã•bŽ7‘tCÙA”Cò:‘ò:}œ’l8îê¦-qx€´¶¾“¿‡>ð-5sç dø2𓱰„L†Ï×V± YHœÌbNº#E w³ø}ä0ë}#s^¹>Nä§Ð=F^7kƒý.È—0¯4y1.ò•m¢àS›œƒÏèPØÞäùbŠ*Dd éBq“StÕT(ù’N¡VÕeüGfŽ  \í}ñý5ÜsæÉ‘¯íZ¨( ÑÅIXJu·‡Jnü/)eT›\ú´Ã{šÎϨ„uVn„ú–nçÏA^Þ#æ¨ã£gC%]Šz<1<4å/PºWûKM8V¸ΰ£5_×^éDS§ÄkÛ‰Í-}Õg ï¡¥’%ÏM• ´ŠxNxhr­c”Ü?ÊP£MZ‹àðW´óZoVUD{YÒO®ä£èð BÆ&±é[Ÿ­$F§úÓK±ŸÑù$p¿SvBW·üÒÕGLx ˜”"íl-º­^¾…^Yš±õ1(¬lª]œŒ$ÜÞö…’Ðû`!йªJ†¸Û>z9–²JŽ Cú ø}=J)x6³´ïq¤dµ:À(ûWo‰O×P¨é¯îâ¸Ä6Bì1–¥Íy­¿c⌇²þ6g=bù.]>÷ß|xJðq˜…%äW8|á¹ñ— Z:›ŒŸoçáWô»/¼„BNbêž(̸©Ëó¾ Ë…®\vù¸ÍËw¿:êA‘B u½¯9˜ù0ýË‡Ž­¡-ÕPZR7éò¿ó!•–»þÎþ¡ËÓÆ8¨üs0R9>úèõ·ò8À§Z—Îm|~äžžÏ5O+G“AÝw#=@ƒÐÅü~h:YÁ­Î0ò/v96DP@uëð['t…8•žû«ë~V_ÁëÚá0pg˜­UO†Ì8ÍØ™~ÃȾ ³Ä»0vÃÂ…VØ ~4ZúP9Á¤„zdÉů0[ƒ›0«˜=þ ÌØõJÀBÕ.}~XjzÎò»ò¬,3+rß½ ¿eóÏѲgÀFT ]³3l Ó²\wG‚EU°³¾…DÇò…ûMxÔ!y½"I½ª! $èuÔrîEÊ!ÊÚ¢åx¤¾$ÃSHç93pÍ6ô7Jçö?DÆ¥ÆÛ_Þ\Ý©¾ríŠ,²>^9dò!Ù_ìýpÕù ríšíy¿„<Î+·u›vV…•~"’ºWØQ(mrbÜ‹E~?v©x¸‚â÷MSŒ;ÂQ*0«þ"°¢l‰,ù‘”/§÷Ø;'… µ“œS:6¨´ k}ŸŠPõ _öç}¨!YÔÜ~ò=¾ãäáT¥¨ÝA—}R u®xí™õ¥F=u‹Í%¡n4  X¨A£]O7÷΢ e­ë†~šíz¨ùT-ÅÅýíSÑJ¢øý9»´v_/öLE›¼»§º}>¢ÀÝÑÄ hŸß%ýƒŸ÷Ï™+°ÐB'ë-Ù”÷è|„ÅþÝt-ÒçóÖGw!k-‚Ò Õ™’ÖŸGõfêöWºÈ¡x„ØÂ¬{uimEeá뀽üûÕÃPÅH¤©*å³-mbº øøµ4&Ç}|Ü|&oN’®6ÎCnóuùÈúÜqÚ=k¼ 5Y¸ûýÎ_üG¿ÙZÁå{l¨àí÷…“/À§õ#» Æ'¡‘þB±‘´¾Ð§â¨ƒNÓ™bß$MèÑæ ä¿“½iD­kË™ð]&@À½T –o2Æ\Þöoj%br– *Í'ò¡aÒ1ï%v™=è ÃY ÔêB0beÓÎÀF]l{´#XaìS‡·q!ü€=æzS!ð3õxÙ¢1˜mõ:bÏSùÒšìëÊ0sd—͘'ÌrîS=ds´ÆŸiÂ>Áܦ{BA6#Ìž(¹ osŒµœ-`ñé€&M:;,%ˆ+¸ ËUíì”˰JîÎÿzå¬×<û•( ë<ÌãJóV°±J[QÌrþüyݦaj‡·†#+µN aÒ=Ú0…í\Y/\0ý‡IØ éŸi;!é̓{{ïË"¹ä±t—rH±vŒôÖ9¤ú~ðuï§sH=è5õâ,Òí:Ó°ÖLƒô¹þY–‚‘a~Ÿù³ldÒŒ Q¿ŠÌwO2¶Ã<²,›mt’»tçE b–JS¾ðT¦iÒQCæ­¯Qm+¨Ê=Yþ‹ÍÕ\z?e X¡ú÷õ[¬ŸæP³YÝàx³¢òjù׸— ýI6:¡ÎUBç]=ª¨»¥ݤ~õ¥¿Ç¦V£±üóù§úhðãÀ­³F4h$¯îÕ»|ﺥ¾ýœ…&Ä9Iî ÙµÃþì·&ÑB²…ýháQ´œL>øË~­FOî6UGëÑÁ©ê_h³&ÙÎÜ&‰v|ïˆ ¢ý¡/—íªÐ¡ð9eÈF1:JмýzÞqª<-·Ãóýr¿ªP±jšìÓþ(d¥åñàÖ‰…ùÝÇH¥% ¦úzÁA)wxm*e¼ PÄ'j0gðׯrrWÊ»?Af-¯ÔgvKxéÄ*^)pRË£î5½i†ç­û’Ÿ Ý‚çtV¹élÊðü÷^mHk&©6†˜m»‡WD +)éA7W ä1rõòxØ@þøÝ6s ïýCï‰å=}ë/¼qí'ý™:›è…»Pµõ*ŠàSÅ=“ÃdÓðù½ŽÚÒa¨!²k¢ Õ†Úa ¡_¾P¿Øæ=ôRkç}P‡'[4+ð„³ilB˧+_ç@›úU)?¥!h/7¶P=Ò 2g)™W« ëÓÝë|æÐã˜zƒÑz)ÚæÈ›Õ¡oBQèC²|ozà*òy§?Mlˆ”$ íî å©ßÖÿ¡Š…À0ùRº‹ÕÆT„³tE[`¼JX@Âëü¼æ©"¯Î“·Tï´8uÁt›Ü—ƒ¼ŸaVÂŒÍå !Ì]¡YäwSƒùÍÒ»ªyö°˜Ì”£éÑ Ëû‡ÎŠ/ÒÁ*oçQóm.XÛ¤ŒóºëcYNä ô°¹•ýDê§5hjêp_ÿ„¥,Ô–%H|ˆ­òòj’ •Ò¾¼dc+ŽQ)Š.Î#DªèÒ*+bi¤¾Ìq²ÖÛiC¤m“å^ÿÀŸW÷ÁÕöip\;2ƬlÞýq™~g;¤F–Ûߥuf"›Ná‘Yä`¶ßµHÈ9¬³¼å8ŒÜ©6µºOr‘×£þ©®'òí¦L§»Šü l–iMå(p×§n>Ž…Tû¢~H‘ ˆšó²–é2Š\¼Yà‹ŸéÃ;iûQšxë¶«¤ Êò¥, \úcEº‚«(¿¨ê?4ô¼L[H/£Å-MuTž´8LO’jÜL¥#eܨqk ¶”8 µ¶HçïF\EmŽÅÅïJ¸—÷Û×§/ôPG罞îÔ¶Nßìaû­®„úZ£Ãï‡sÐ`„åìV+#2o£§ÔFñû»ó”øÐt³‡kn›sÍW•zzËÑŠæ2]UZ+Ëešœ@›`›ïÔ®gж£:÷VÑÚÛïf*A‡~± ·At¬¤¹uÑïÝíT²„â‡(äôH? …­cU"Ï8 †°H9ÕC™ŒÏÃæ P àYæâ”9Q9«>ðJ‡j`9A^ÊËlΆ´ô¦Í‹ûd·uçxã»ñÎ~D¯›°ûNÿüüÅÕS/UšÕÿpí¸ùÄè<­äoÑÖ!Ûù—#aâT<|ø¨ðªxí¸ 5+õöæ§@=÷æÐa* øªkôO)48ÐSïÑx´·ÜÚé;nûŒkš‚Ó;šŽBÓSÓÔbÛ£™AãA hz:£˜}šfü3|žÞ‚æ§lüã_Ÿ@KhË•…nÐzZÑϨÆÚ‚|Íyž>„ö£]«ì3òÐán2¶w7:}¼k}‰ì¶ó££oÕ„t>!²1žq¦Ý½E¹ÐËïý©®ZúNÊâ·o€þžÉëM`@yÔíHŽî¿|ˆÍ¾Éë0üh‚Dp8FEä~¬\…±_On4Ý…£§bÏÂÄæL?{E"L«1Ž¿}¿Fßß4| s÷<Å=XaÁŽÆÁ®$–4ïåùqHÊ¥Gk¦ÿ4¬…î=­ãëë³îFJŸ>çãbj_‘§Eã¾ôC$&|\½qì’¬>Oz¶¢d[æìûç‘RÌi9êÊ"î ž»ÄÅ6Š4SúÏf~+"½JâîJdg9©T‚Œ[”õû‘¹ø˜¼»"%²–¤Q,·Göåô#º¬ÈåÎG½Z7…<«ÞÏl›!ß¾´…KÎ'ÿCâKË›÷PP_—¸g+…foWF‘êûígQ¬{—ò‚ÈO”T×NQu”^GËÊ×(÷@ɲ}©÷ Æ¹Œé(T˜ñ˜äxý•’îÜ#t´G•û@ð&ÕªsDSQsÏÛbÙ æƈo¡vƒ}xÝâ8êì3 {{à7êQ<´–Aý6î“%•½h˜á¹l¬Æ!lÏ/1M£©ñÅÛ"—Ñ|oL6­g-Zʚ掿D«[G6<™Ñ:¶Â=m>ÈþùÁ…¶+û×WùˆÑ^ïS½×écèðªª.`¯:ÊÔ‹+K SÓt]¥3:G§ž´PF×ÐvŸ±ìßxàÑ1Vj/>tÿ½K̽f©ãÚ¿ Óá³—-·dSÛ7¾ÑÓ­¾ù‰PüÒCÓÇô=ñf°á!”~=Bj¾$ü¿ó!ó®ªå7·w|—ý õ¨NtàzA ã,…*ïlçéÃýsZð¹<ÁßJÃjÊŠŒ÷쇺¥2f—Ufhi`óPy¶½>¼TTˆüõ/6\'Îö‰32é[ÿ‡±¿ê*j»ÇaBº»CPJ¤;öE—H—¢(  ˆb€II‰” %  ˆ”´¤twHwóÂwnpæ¾ç™÷÷מsÎæÌg˜½ÎZkǺP§âõ[ª{ǸîËöFƒVâY~f‚h˜ÃÐ!šŠÃÃåßhÜ~P|‰üšL¼Ï8ò‡MÏóLóü¿gííµÅðkô³ÖZ<õe´˜×꼡‹–C˜=õ®¢Õ¨Æ[êOÞ õÚn;´EIhKà;ˆvüÔ?w„¡}Zý5¡€1•iΉ XúZKäû7Ûùwå€ã2>.}ðîgïQâ·A+eyæØ8ßN8ÿóõ Ó}üfÐ1($æZ;…oåuñhä0P'áòcË­?ßæ¶U a毌ێ1ïÇ[¥gÏ»ÓØc¢¿êv¼ŽG­£Vëip*¢fpv*¸æšï™´ïó3›6Àï»°¨ÉÇ ‚ÕS.9w½A¸¢˜ü÷Õjm.35J#=ЯŸ³vÔ‚Ô];Å·™“ ÂîLù€.K–uß^…μ›â­Ñ ô“ ˜q TTÅbñqrAí¤4g(hpC€ÿ?äyݸ_5Ú¢£¸éAÕ £ºS’ƒz-l}|@ÿÍäIíWÎ`°°£ÞWFN{¯$¨€ QòÍÑÜa0mÚ߆ ¥žâ“g¥À|xýîW+°’£Ç£xþ¬¯¹Òq…¶‚O6…¹µ.HõÛ·kD.Ã9JýÎØ4´l¿mÈ7¹x ãúÒo‹+¡²(ÃÛØ1P1¿ú“û¯åP>ѽñ©úž´‘ùòæ Ê9A¥õ ëpþSŽœçp"‹}Sê€/‡({é/óã~3 $4î BeÊöÖ¯:è×ç…êóŸPCBðÖº™>ju²¶¢IÁBx¢ Ó¨kñqà)YZÔkàÿç ¥ÏßÌ^¹>‹ku +“ï ¡¾ïñ…hø^ƒUßß 4\sWãá.^¨Ÿ¾±‰FÎ4°ö¾ÛB#?Ï;×Vº£ÑâÝ{Nv4~’fߣ`M¼ÆÍÒˆÿ‹&ç­K1‘š¶w©³@}q¾+ï¡ÙÜ$Þnåh^0¾ñ'ZðZrvf]@‹WÈ1©2þ %}9Ö§¡вüœZÄà?”$ÄØ÷­žïÖø]æ€Ö||2%éÑzcýÄì%´©n‘Xs m-¼~†!üíÔ=Â]np@{5·-¿ß Çæµ{a€y§:· °Þ—¹¬§öh1A·­<àP†}`ÌÜÉÎ<}AFÀßzz‰ó!:¨y¢'@lÂ’Ú¢» ¤7*”ýˆ)ÜÎáÁäE HÿdNäTD¼ZXR@}õ©Œ ¼Ð4g=KR:‰F5>š ¶™I±'†]ù'ÖîíÀt]®ž+X(tèˆ\ºCª¾ŠØ«&Îîm‡ü¼Ìnž pvHª†’ÃiW‡&ƒï4À5 Ïy¿îÀR)õį¨é…^Ê-àsÓz=J3g†ñèÙW˜àì¹fÑ@{Üäz•Vö„/»=]ÖÜ‘ÕDr»,{{2«mÇÄw$ìë:$CÍbLÐ*HÉ9¼ÎŽÚéß3²–K ˼ÁS.fòxmµžºFŽ)'éú@a[›7X”ìfZ4Ÿ©ƒríÚ¸JO¨,U’Œ%Õ6âÉQ-+P+<ð<'4°Î;NœM}v»’+] •ÍN Úœ¢‹Âàü·¨5™å {)ðéu¿Õ¾¬¥µ!°ý‡#ÓKσA@´"Å4 † ~òY£v ®ü`¼;`«æ`j¸2"ˆ‰Àì—µµƒ 7ÜßóñLPìÑOùƉŒ@»·W|Û—-*h;ºÕ|A5ïîý]3%FÅ5¼tâyÐ׿Ÿ˜$ÚÇ|•e[£a»ˆ…>ŸÕhâ*+:¬[á£#~¸.¡aÐ}çp^&ø9Ïàþ˜ÅÖ ®úÅÇ £:¶qÇÛÏ¥â[w… ¢Œyb’ºÈMôEÚ@Ìè†Ý¾­ µmï o£JõÆÁǼuìåtùĺQéß:™i%„Ê3fŠ•ÜhЯÛt\ãæ±¨êdZ¶Enªþ‘ŠÝÕÑ€jSΘޞžBõk’¯K÷Åí³î«¡f¢^'6_^ÔòöF­“µjÝþÉðí.j{¢u—½øð; ò%¯_u¦ßˆN(FÝ–ÁyˆBë@‡^þ3fjú‰Ã´¤-Ñ FWõ<…6¢>1í•ùú@6ÚIÉ ‘ 3¬hô¯@íKËhÜ”õɧšLµ›Ò8‡¦Þ?î(Ü9ÜÇF¢ÈiŒf«+×®SœuJvA‹v÷ªÞ6w¢¥qŠage´âEs ­‰ÚÖž,(Gëó“‰?ôWÐfáSÆøP´<Sÿ0í¾Åט¬“Eûëé!æ‹æ€y³·ÛG/°NoŽ6µÂ‰NMÌêßw7Ö…Ì,à:à;[O>ZÔBÙ¹üÊÒ] & oõß’‘äHÏ·µ@ÖÔ?¬€ ä;ü·7€R¹|×ñÜ:PU„( 7O7Y•Õ*Й)¦}Oó)½JdBd´(Ìyc{!…oÕˆ\­ Ç؉r$†‚à$c—>wð àpOáP!!‡SüÁµÍ²À%F§Õðxî0¤m›w_‹&=öÂ$œeRŸ2Aá.úýÈ‘b¼D8h/J¼ÑXöžM#×jìÕ3’‚uáž }aÃB´ü ÈñѬÞcÄÓ$¦ï" b“Ù²BAQÈO=Ô\ÕYR{{@y0f"A^ Tí¿"ãÝ3 Ž¦U{4^iÂÖÁ­so-—J3A›íÅÃÇ CÏMj&¦zÌ<Åžs ¯ä¨zuD |{º£Ÿ°a7ͳ:‡‹`¬ÏçþZL¦bv8ÔúÁ,¹ú-Ûs6pHpU“èÿž¯?j:­Ì(A|Ъ´?/`CM.ýv#jÒ¨™íá8†ÚàN$Â'µF­8†SD¶èéÅSïÇÂQ~:wäÙEÔŽ“àžsà‰YmìqSP'çúgüçá¨ë‚NKÍJÔœ…9Ê…zZš\˜S_¡>1·žÌ^£Ãú&âÅР``Ì‚ÿ?¨¤²Ff¬ƒFÞ¹}7rBcÔ[?ÂëÑxµ¢žÎÅL4™¼ôgŠMçÒ×Ùxᣙ™Ÿb®E%hþ}—!i7ZtÐäJ&DËÜŸNSv¦¢UL£´Ål„Öv’_8µþA›4»Ù³¡hÛ;îa8wÚÓ"¾ßdÕ³ÝYë£f]Óƒóp"ô^Àh¦àÆô¥Ÿ.$üš‹í—ˆþÊ@€(XN’!S—àeóTÁíyþgD¿‡ï1\hñ» T×nvƒ¤Ó€Ë<Hpšñ``9‚?>™$×%Ra_Ò~ …îóž~¾ ¤jðÍéµ (/¿O|'œ ªß-’R‹D@=ø„ ±éhÚ.½MÄ‚sÒV¯œméá¼h)§À\ÐåÕPºÖ zo.6½êTýèsÖq˜+`ð›ëŒ³ÜU0Â×ÛaÃkã‹Z¯.‚ɯuÇ‘n0Ó®Èp[ ‚ ‹uì!Ìj`žO\3ù” ,ã×èÿTÁ¥¯Íœ·uî‚õŽðÛNõ`/ÜGEµœ^þ_4¸-§_Õµ§uó[’f¡o.‰ ¤b&(½7ÒxÖ£¥\ KÓß:ЙçâAÜ:œÊ¾Ðã-ÚášÅxÚ1î" ¼ç® +¡ðui?¥×ÇmŒþ+ÕûÔ‡xeÈ·ü_øþÂn(QÊá¼izÑ ÜþÐR¤åÜÝ€C=l£ÒSr8tCþ6ýA{¹ÖìáÆÁsñ'þj×S¯hŸC_‹9¹ížÜCßD«Çeî¾üç}‰’•jmϲ)™6üßþ¥[òN*S!ùS©É‰Êµ»ö¤o¢Šº ײ·%P%ÿí±±+T5Ö-1c}Õts9™7f zóK6Ó¨A*z‹äÕ ÔäýRû,Õö?XñußMqµU3zº^ãB¸¾NÍ„ǸîÓ_Qª”9ÀñMA'N4´ñÜCRþ)м\Æ•o“µíÈ Éû)”VÑ›¸¥SŒH]A³ XžI°]÷wÀàº×³â X(Sp›Ï°s'åG®SÎe·¿fÌÏáñÇñ¼Z ¨`*Jnb‡Ómìå€LŽ]]Ö» (¸g~âkeýâÆC 1nó]êZgËŸsÓ:@ï³~ëlj 0NÚºw Ë綠s.ÀZ&#÷qç*;òáÔóäö—%ྥ&FŸ¼AÄz×ä€?D”–šV;?¶ÕUâƒpÝõ³a;íŒxóƒª$¨«t@*èB«-~>ÈÞ{E_ž èÖ^P蹬Z$† JÏ¢ÏÔƒŠ£ãx0¨‰êÉp†4ó¢_ó ¥n®„­¹Ú=YÝ\@dzÉtpÜôîýõUøJú•ÝK$ÀPð:c_G)}»!OrLlïæ i™˜F¯*C8\Z–;ûþXXv9ŠXü«’é¹Ü—`õvQûö<èÕ2Õ>Òz<›¢8Û¥whq>'+5÷`<9ðžºÍŠÊ¸R*¶èÂ3¥ /bÑ·O7¹*­Pî碬+ØÇõÒ²enÓŸšsüç †Täë«ñã~ÓÖàzùúž$"S!†~YyÞ2º9{0nãä_dø Váû_(]Cíe£6bõ¨+~ ›k*õìŸh£0@ý¯æ9°|Aƒ1´vk¿_¡¡ŒÖÊ:M4,3YhPu ßO `­@߬œ¡áOÏ„}ð´Ðˆ€nÀuk4z2Y%è+ó2þÚÆ§ŽÆÇz¨´/\D“Ò†¿>z¶£©¸ë»ŠIè/qK£ùóJ4óVinûšã ñö÷@óÜ·ÅÐBÓO»[¡h±ß?ì9Z*áW^³×AË ¢6(vÐÊæÒÚØš6Z“#š0r/FëáÒü“2¼5Ÿ¿Ý¶*ÈjoN ˜íó‚µÐÞ[ñZ_ÞPÀP+±¯ýV˜ç°Ïåt_¬ë—x0MI;[›"à$à0–ÎÚ'In’j©ê]lÀ×¶È/z;¶4ßB¨¶¿ôÛ†ü!ÏÇW§r:‹g»=r@áøwŸòJ$P¶yµ_\jÕ6A²_€&šÑÓ|h×6²Ÿ6ýÙwxZŠÀ K»+ÇÙ Œ¹)3Á†ÌÝð2`¥i›ÆvÓQîÒëÓprêFûò®pFYØÐ¾„Óç˜~7_Ü®o³vEvÀÝGŸUðNx¦’ÆÍ®3^²ø7#8フî0ÁgµGë(?ì`¤W·xK˽Ú¯‘æ‡åóT vèåz+?ˆ/¨-º– ’¬oi>¡2rpãxÒeœ/«zÞ‚ì…•ªñ_@ݤã>)pÛH…üa<(¬]ûØÇJjšåŒA9ô»D7¨ä§Ê òÕ¤Iþ¶UP‹É‹cãõVk§EJФæü…Zv¤|;g>À¹2öõšŽJ8¯|åá fÐi½ž5‘¾zˆ¥b²*ôÙº ] ?ƒ9i:éþ’MþÈ}] F'™$ëfCÁX•ñ'Öì˜x÷Å‹ƒéï¤ þ¼-ðbm ¼ðpDB*®ŽÍºùA¦êhÁ:Ã~ÔAö@ºpùåðă{C§ýÐW=¿?túÇ|uT·0í¥«k 3ÏqÝãz‡ î»ÝDK&òUÌcPÂgŽÅgO"QÒÃ÷"ÙNúöõ¼kzÄ!3úî[}þÌÛý˜÷òÙhEJ¸(yëØÒÐ4<|PiÖ 'z9‡Ê¯©6-¶Ï _œqØöYú¨²‹ÝÇU{ìK~´DµÎ:%ã¨Þ‡ž@©ç!jÄMçî/FM£ï ¸+ŒPËe‹*¢¸žÊ×ÿ%MˆÚ ÎÅP ¡ö “òjÔQ—‹x…P·­û°=9ê•^Ê/ë{€ú¶ndÓe A†k]1dâhHØOÖ[d _*žúq¤½Óœž Gc$ïih=и;Ýå<Ö-4±'˜œÂϦ²ÓÝåäûÑßðÝ™3-h¶ðe…FB 8fÊ^õ½hQmâ=¾ª;ZÊÏ }Î#‚VLÉž¥È 5¢&Ÿ»h½dsèl@7Ú¼G9 ¹„¶%iG_QÊ ]뵄t.´w8òAK9`rÄ.鿬z1aõ”N8ñ‚Yóá’àšŽVÒ×Ä>§Ò<‹ûk æÈ÷­¢HzQùPg 9ljûÄÈhw«x¦Jü=®ÄKF XÙUÊôªàgØÊÆåÕòĸ<й8~ ¶†ëciý±sÀ$z•-5è0WzŠôôÜV5#W5`[eIºÂ '±Þ¤h³‡Í¸Ç;_8Å®€Yhq ¸D[¤Ìôù€Ç#·ÛKøúü> µ,ÁYßÁ;¾1 èj-ZZRB‹·£¨é–@ÄðQaóÓ{ Úµ:6§Û â/8ÜI7 ö¹~ëÝ}X§²³®6B ‘€ "ïÒœ›’ Öêë è,+6ŸçJÖ)~?R¢o: ¢ ú"0|†6 Ô9˜^½Slººµ× å%ªƒ@[Q»H¸ê5èH9·=+=¾&çÓS'A_'¶?È” ‚ìÚ¾TƒaŸ1u#Ÿ×ÿÂr-Lz™zÁì…tI?ØÏILÙh8Ái?I^ËíXÀûÌ?(׆‡ú0Ù~ŠÛ~Bß L1e#PNéWgFãPÆ<ÉÏìêè ‡Ô£zG”ô©9,Wb%ÞFäÊ…|´CñÖäO­ÛŠQ¬\íeâ¶0¹ÌŒùâ†ç¡^]²]Ê8î—Em%ª¢›òS.6ä]uA%3°G¯”¢_·é§Î7>@ÕÍçž}¾cŒê5…Þb z¡hÑéLôûÚ›œfŠQÔ/ÐM¿5z¹ b…~Cý/…;> ¢ÆoÌ…÷]{آߢFc‰¡ó‚ž‡£‚®æÃõÄTú.ÔüNµÙì2jyëzù÷v2jMVßa<óýIò3erÎFm‰%¶é®€Ú3Ú7¬wPG]èßó¨s+Î%¬uKhÐs“G¢žYZçàÔ»éO€l»PÿsýÓøŸYÑ ÷€ûÍ?¨òñ þ4òzÿæcAC4†]ñZ@†ÇMÐñ6£Ék7ßÝ6iDÓ×”òS%]ÑLÜu?Ù7¥h^™þY;-ZTÜrœ‘ÛCKÛ*ËMëƒh¥’1ñަ.Zû4«~Vâ;ÚH+;ˆ¶ö½LâРݯ?žzx»†gËý\ÀRÙjJx%'Äô®ž¾r p‘sõj/à_›ÉÔãmÂüo„ê„eªƒÈ”"l.Í†ÛÆuO% tn½%šà ÔìÁ[Ëy–@Ë”>„X‡€^K©¹°eCŸþHR |7u—¢ûÀÊާÇ'Ov>OçÈŽWœ©5ùÚpŠ«ýOÅ18=ó¤àgòwà^“N(˜L>ݨìÅMp–Ú÷Ô]¥ bIn /õá†û1ö¬ ØDÁ€ÓâçÃ[ƒ¬ARË·î^“3H{ñøT˃ìଌëÒi@÷;U:ßE‚BJ¤7)(‰^˜f©¸Êý1W;ÿÊ€j\T×û¼ P¿•8Öxß45¿š” ƒs¼(,6:Οþ¦¡ ºœ*–5ÈAïíW)nÐk·~ñw º˜Í(&¿ƒÚŽêüG0¾©Úíjè&ƒgp2íÁìÚÑœüz¸H™rǼï˜w»±?]Ë_´UeCÕp©]R¨SªlðXŸl‚½kú3mœ³Àû—tï;àþÔº“ž‡:‰5Ê—ÄØÑW±i¢&!ôù~>6_Í&J~GZ ef}¸¯ôÛ™ž(®÷ÞÊCÂc¼EÕ<•ú[~Œ»æŒ½ÇÆ"(Üg Giºè¸Áæ=#ïà=_Å©þ×f¬ ±N£Üm^ögêŠÿüáæQôpPÇÓ:”û(0ÑøG;ʵÒ'Y}zp¿˜(ÀLûÿ­/æªéò·XVhðá> QN^ãþðMräÇJ!ô}%NZâ»ûÿí¹SÒ²§iÏu¼ìz4ÊݨúëZQÅÎå¬×N£JKï™nZTMB»û$`óð|J¢‹3&ªgÞJu¨*E JνÓv¨é¡‡×;¼þðû8ݸ•ÀÎTÈS?À÷;GK–c\÷ÎQ ¢BƒÝC}¾ôh¨á×ûdD‚F®ÎÇ4|Gã'±§ &²Ð¤~Aו4í‰APŠfê‹ò™Y£Ðü“Óöí£Eãjl¡ÈÃy:¶´hUœ‹{ˆ:­»]Á²?ÐÍ›í_E0ôÏ¡ ¾Þ+Ü^hoã^mkÀ$`ü‰Ž`ûµXüTÎcM€•MÔ¡`8šf‰±ö€§o©Yh[y§òb+¼€øBr¶äw ãÉß }½ ô­8‰§2ßÿ4óÕ öH|°—&´¦/gJö’€ÞÉ@«Å»Í¸ kB€å§ríëîÀ<ž[n×7¥7¾UÂ)ŸGõ[+—€«Ýòûl÷+à}4“|9ø£ïÈ `éकÜ(Íkna¸1jŽ ¢7›N{= ¶'lÒŒ ¥k/hód¯˜/|¿ž èÞoiePè§`Rz{”^Ê|­ñ•‹ûCg«AMõЉ—thèn¿Á›-Î\ÜzÐ~dû7ä 褜g(½'?Ž¿ýÍú·nsû`èÜ5;HZƘ“£âoÀ¤@u­¯÷˜…½m~¾¯ 㜕\ RÁ¢³ð–¶&\RýÑ|¥W¬×¾ÖÝÖå‚xþ/ÎØ&½IÊ=`ðp˜›UõA ž×{»î¢:–¹„öŒ­ý·:@v)%mýÂBßÔí¯àûç¢\Iz.f7WôåñÏsáØÏA­H××ø¿Î!J¸%Ñ÷xŠsÌGøý|?ßkè€oõc®¶0 _äXŒæâ.‡ç)|OHG-ka¸$¶Í¨ÝûÇ–¹5ê2²óå [E=iÚ´¸P¿èØÞdã]4x«”k9 ùrº&%Σ¡mlºªøf4ÌaJüFä¦Ôr¥ sk¨œôx†£•b¹ ÐÛÌÐE£ï\„#Cиtæ]=²j4á¯V«š§ƒ&ãJ«¾Í£iéÓá;mÁèoªU“Å«@4{úÙ;Ùþ 4—ûØÍ,é)Zq©jÌ0E‹B&üfÐ’Ò5EƒéhoùÁYÍ­$%ìñ‰¢Õ nÍ|;´näædýmŒ>ü䕚¶Â^´¼ÍE;·o½¹® €ö¬6ŇçÆ©ŒF•IÀd—Ÿ>—>X2ÛCDØÏÛ;mG# NŒÜRæ.ÜË_~KÍÞjc+¥ñ[ x:ìþE(ˆ ¶3:nH¤,ùC2ÙÛÄÓ#Q2@qJ´¦¼(?Ìù_%úÔ'óŸŒÈPÍ÷OZ3öešÔþ>Ðý¨>%è ôq ·éû!–£iql˜°â-‡ÊÒ€…BÙ3µØË4æ³sàäÛ…ÛÓªÚÀiãã21D§9šêt·€ëetâ‹ôËÀ4÷¨5É xb×n÷ˆwoyEác³J8ô§ô¹›ø'žW,þAÙ "ü”V&;=ñDÒ­Ýï¥ý±K«·R±A|Ô¸×óÑHrZIç,/Ô} ‰^©\3º ²¡‹Ì” ‡pß¿æ4ÀKØŸu…ÁV~Pâ)/©˜½ ʶÞXõAåAð5ï4+Põ(¿Y÷ Ô<;ƒržæ‚z‚ëÍm¶QÐèü8\“" Zt9Ž?ù᜵waük7Šdw{¸Žº^Çån Ðý3<>DdzE5Ä‚‡ó7ï}ˆiZÀàñƒÈŸ‰1`øöÆ-é}0znÂðk<Œ“K+i{À¤m&{Y¸¼€9´‚„ïd1Ý•áúÁROÖ­]4Å^™‚óÕˆÚ\`þl~ôµF¹¦tðã0VdªÉ1_e>i5Aë(u9†vœî-JÚáÓ|÷åp°Ð”»ðpÝÐÑ"/ø@‡æÈ¾Úm=\O4çâÿ’ÌÇñÌPªI5;I¹Áá{p²" ×%GËž==Þ7sÀ_|ÅØÇ¼uäöäÅ„)¡Ò»,D¯b÷¹Œ|.Cm<c8ب2Ü,$^ÔU‹]÷œAµ"·¾¢zéî¥ÕS¨¡î·^xƒÄá9))!|_Ô"¢YAõuµµŒÞüƒÚ˜µ-:^BíùI8럴Pg wUDúÔ}ņì9ê5ð‹Š»‡úõ-ßñ´´¡Aáíéó$¬hHg÷¹÷W[4ìóõõ"Y1ù½‡›3 á]Š,FãaäÈ…1M2åN_!Û@SU­tŠ?KÐß8kRÖ¯:h6S¡L6q Íå0Œ¾F‹¼F.÷ hÑR(“øÈI´"Bâ¹|έŽ^v¢PåC믉˜£¿ýD›RϘ˜v?¡­s|ÆšÐNQA‚u9Ú[£Kü{0úCíí ¼ËGÉ1Õ·N(ο’§S\rs¯.ÒfÀë:µš»ÛOL>æBq…½^Xâ/1¿îï©%ógŠÆV —xx©X(¢ÍwY¨ð€ÊL—Æ"w h¬ICð]@0O¨-0„¥o·ÎW“¹—ÆÉ:`î ¾Â½}X¥Âö8‰¤€­)z2‘ˆا-x š€CûTÃø2œ¢Z”JXI.a’ ö­¿€ÇK¿>%ø&¿ËXæîÂYÍÒg˜i ý‡BX7„i.5Ñjòˆ?¿Fÿe£®9÷ÄoHï1µdþ/ûO—7@º­4¦¹¥äX‹>;]¡äe÷­‹7®rÅ ˜âÛk“aE ôǤɣ˜Td=7”E@5Ïdù¡¨+EÎ\¼k^US@+$¦Ûµ»´/ãGƒŽÙ—Ÿ#ª 'íÁÔjfú;ÜSjÁàÕlóϾ[`8˜£ ]®ÆšÛñʸ`ÒùÑ‹¶Ì\Q²Í† ØKá4Ø“%gpCâôZ0à™…Î= EÝ8‘·ïj©¢üµz•ð).”C¾·ý&õ$Ê8ƒáTEƒRÓGÏۢı.ɩׇûE9N}âw†i/M.æ–ÚØžýç cÔç÷·¥ŽöV¢È„ܺ1ÖÇýÏ]HF§£|¢ûWG:Q‰ïJY´_ª8x=j먴iËDµ¾syåݨ.Y‘ÞzÌÕ¯´=ŽóÆA Ú€iQ°|8ï²c])rà\—rú®3¬‹o¡† ¿ArÔpi”Æ/¼5 ¸æ¤s ¢&ö‹.ÂÒ×Q3›ÏS¡x{ÔÂùºÿåÛZÔª Þ~+ýQ1ŒJ¿Ú4* výS2=Ôá*Á÷ªuÆ|u¡úºú†ßjË“¡ñÄâbWÔ[üJ¢~Ê õÛc/ûŠÙkØößïþó‡ê‚`T¸ŒF^>åÊo“A£KÁW”z,иçó~<—4É?çÑ_ñM³;%.-õ£#ºìkehŸ¸?Ú-2U>ÉnYDK ƒe!t´håé`ÌÆ5Z3Řø€6ï$¿bºƒ¶üÌøÒ ž¢]N—öXl§éW8Ês<­Í#°[¸Î;°¹Nßîè/WÀ'– ÊðVB³ˆ&GK þÚXW§àd¤® §€üGB!Æ6.Pž–RŽúµTâ~ÜÅšúìß¶7“n…,ð]d%0 —ÐìbLóËû^"€à©Ù‘`o^}Ð3À!Iµ§Hœ¥{X%<9pÚçÑ™ "sàhê‰Ö ÞÞºžÝåHàÿÀ¼1p ;«Ê-tòA8”)KI"D ñNÑg·€øIV¹ã ÉÃ!Ô‘ý¤MF©èm±A6gUª}At ~<(¼+*ùyX؆¹a­”k>0'”‚j`ñóŒP7V·|SšB)O¢ªÎÂ9Öèæ.8OOKä’ëº,ŠW®o(ƒ^(ElQE*è'ØXßoëƒ1Ê€¯ïpÀˆOnÇ©ã6â Ý+Ý“-«øß<‹`ö$/É— .Š>ªå‰©ówª¾ß`9!y7k+.­F#ÄÉ64×R¤hÝÀ>ïöcçKp²¬ˆ³­ðât*£ö§ö¼¯.¼:ðg®fX(±-Ë‘I%5ãTiß®BñÉëUAµ(NV<ˆKïoQüßïßK <Æ]¸Ï iRõI.ìá}âõÄqyÅ?V1Ì®ù_XöÒ6¹ å‘V¼¹øÔëØÃÈ•¤ûrMºåL¦ö¸¼E9‘Hÿ ÊÙQ ~dÒxè#«0 ÐW oÅ„èñ+_¢^Æþ0Y9øDé*Ä—3<ÃÂþûCaw¶G¨ìÁ˜öLÙgTþ^çC¹•úuÖyíe5ª ~²ÜwÀ»†§>5£Z¬~?BcT/±z=3ìà;!ÿÓAùjzlíÅNKùÏfÓ™ÝsFmþ6d¸EwPÇËÔË“Á‚Ǹî}ᦷwn $\\¥ã9ð‡ {_ùµ©ùõôŠtÛ{WÝÊ…&y°0„#дüò‚gšñut!ç Dó³'ísߣÅÓñýòÚ%hi¹?þvZÅÔ÷?Û€Ö5ϪsO6¢Íb†í?´£å&0ׇö¦bñ”i£–úÛå0GÀbpOkºû°<»êh¾Ž œÂ#ƒÛ€''2ÿ’ì+$ÍË©âó¼\‰„R@F£Ÿ¾˜B¿XŸÅÊj3)½] ¶Yåñº´Š o(÷Þ,½csk§¿ÏKÖx :£/°?&Ý=ãöÎØ¡a­pêöårLÏzàªÕLŽº¼ë÷œh–?íÛ¥Osv ¸AÿqÇž„;«Øn´Fƒè¸ñ‹1Eà[m7­©L•µï¦ {_^Ï‚ê# À°•Ž´pP,^¨^饷/®p»€ŠýØC…eP3}pc^o4:5üaÀ ¡ŽhgpXah‚Î{ ¾;'èÕ±*âÑþâ–*0ü¬Q2¬Z ÆÚzo ÕXÁ”œFºû¦ ˜m Vþ]s² |÷a°4Œ¾ôé'\ª}ùõ vØÈu&†ŸÑo‘"K¹" ŽÚy B(äÝÃÑ®†²Åí=çÎþä@–ʪӘÂѵôuìb²Ö'”Sêú¾Ùnøp?Ú•ŸNB(«¢±I'í,ú¼hhiÓ­qxNÿfeYç1?á7í¥kI¢x+*pº…%6°‡*JŸ–ÛlΣßÖeeß/ö¢–Â/ ¤Q»46¹qêÂiÙåW³C=f¸í"ö[¨oÐV㺟 TýøÍ”¡ éßž ô EC…,nr†*—x5C4ôÙVÍ1g %☔OÍ abyÉõ÷÷$[—v;àÅʸ_Ÿ‹[rO¹@˜k=ï¯é ÄQŸ>Ðþ`Ò­W¢u~»@>à›åÇÖ ”.ÖB/mj·1K^Êh.÷H1娭»˜eÉA ;¿-GKtK5ãË÷‡€¾€¦ÕÊ4 ¶µý´gÞ3‰93s!°Î)Y<{ 'n‘¾ÐN®x¯ÜWpjëQuCTp]=?c Ü7+.Î~sÕý‹UšÀ{ݯ±sK øò7;–~ÓÿC±â¦.è-Ô—dZ¡Ñú8b‰‹ „Îà| 1Õ}Yßnï´&ñAâ€y,‹S@*²ƒdólÈp_3loÙÖ¹Õ>® Ÿ–81x‹ ãÌýy3>P˜þ(~o”3¥°bnƒ²€BGüÏhPQ¹^“.ªZ1¹çRAÍ£¿rMÔïÞÁ9[Qö4,@³fY¶Íà+híuûÜ?ÚwCR/×ÃùBñÝA\\Ð=ûTyáqè]øV±uô¹œÉý}j÷ÚüÛ`0B‘L膭¯Üž (QsŠä“¨T0îé.H¯Kו+p΂ùû߸™ÀÐß~GÊ Í*–>Fá‡ç÷B2;(|O1.¿Êwõ·‘·8櫌yb·盇>ï»jæáº O=ÅÍCèÖ¥séЫl¢·Äá¡Ë‚Š’¦A”$&˨¥ŒR½wõ’»ˆQf·ŽßG†—‡¾Sn3"êøÜ}þÛ¢ýwÆ+Ǽuì_¨ HQã|T—'tQÙàúš^TªÌ.²»‡* 3ö*–úPÕü§Xõþ{‡ó˜}¶‡û\>$­D /g=ó2PS(«”ögÔ‚)Q2ak…ZufÝ™åDПñó³]?tÁ«K™áÁõ¨ó¶÷ª%êvÒ’Ë!éG½×É}½§ÊP¿kÉ÷¸;hPëùÖròì{ ]ƒ†c?EHÞ÷=œßŒ»lV†Æ \SÙÖ>¢ño Ü´œhRZæìéå4Õ·ñM]ñú›\ü§ÍÆ/”xç¡ù’êA'´HTª8ÏŸ‡–\bbÈÐ Ñ•î¹x´Z4òƒ€°­_f6#™NDëDïͲmÐÖcŸ›ŒÂhGŸÒ€ùæÚ»£Uñî5`¼såë4<à7½8³QÀ^+ë\—ãœêÑ£€‚»£À O#øöœ€0‡v)«ÊˆêÿÞˆ’²/·òÆ,ûVÔ TZÜ;ëwC€ŠC`?"é*Ðè?ããYiºÈœÚg™¿¡ü&ÿ ãïôÀ<ûÞIXùzº:>òÛgVÞ’L `oa3þTå—OK9HÂ)¼øÁÏûK@ãž«É%àñù+)®Ë|ËKuOÃðଠ–ø‹vF¬VŸ‹ëÂÒçOÐß³‘/­¸%"Ú ¦»Ù©.…oPa5Hî*Ù wQ‚ U’{·+Èéá¬'÷=”$°§ þÏ~ Ý-ÅÂÈ((ÏåýŸ2ÊAÅqÌ$^MTû¸uè5c@Ý–ù§©hÌ•÷1«Vv}î+Ðøiq[?t<6šdª@Oƒ­Ö÷p=ÿ y5ʃȻ¿õñÀp‹ñÅ­ï`¬§ÊÍä&¥ôî<{`f¶™- v~rÛ‡ád”‡û¢7àá`ìô<çB„4§¿ î³ö Í=Î…:ʃ:Êyúï¨xk²!ŸÿÉ-ýïÜÄ£~GùPGyPGz°Â•Êɱøæ¿¼§ÿäUå=ÕûЙߘäB¿uyâÜ^⊾xõóq¾ÜQÎéÑóß×^iæ}f9¾>Î?=Ê=ýòå‚á_·tq¦Ñ~‹?΋ûóW|€~Yá_nÔQž©â5Ùü¾k¨ë‘7ù­«¨{‚ü ™iêõµ3¿1Šúå¿–9ü\:\ß·~ ãÿü¡fÂåõjh$˜ºwú>ûq~ܸHÂÈ&-š˜w²8'rMµ‡¾‘ÍÐD3xŒuŠåh®˜ëã#»haõuŠ»ãZŠ÷zu^âZÑM ÷ò¼uœ‡¸Þ!œñÅma‹î}›ØG;/Ù’«KÉÐþ[¿›gÞ$æ‹ní¯Ë€}cG8©ãàxc¼Ó½xé>ŽÄM™@°áÊ_“ ĺ)þ?Më4¼v>Ð÷"ßûãêU ½Ñ´8~@õäKÅc‰, ñÎ+,sºè^Ù¶iÀP¿êbxÀÓÌôÿ¥„«×rktôY`Hy‹°°NöÐϿŜ–§… [Eà4¥_ŒËpS9Rú¦ÚïÕm\Âa{à?[²yÿ~‰…‡1^aËÂ26eC’ô–­ ¶Vü1ñûuÄ){#Ò±IŽ š¹[ˆ×¶ªŽ;ð>äÖ%2G]_žPÎý¸dm¥ªÎ,à"êR“mÆeœ Içï£o¦çð÷NPwÃyBŸàÆBÐ¥“ßMq½0g<™^ÐOž±¾ú³ ñ³Í ÁHVdçÍO%0~ÿkò¡?˜R¿†¿uÞ`–”„Õ à¢ÑeÖUkl°`Súýéq*X˜Ž¥ kB¿ÄÜ"°áÀžØûuûem¯€}¬Ó"*ý.àûÝÛu@mø~!B>XÇyˆG¹jG9lÿ_óJÃF(X+e)þ§=ê„Çÿñ‡ígÞ %gú<\ÞßßþùÃÿä¸ý®œsÆÇý®úçþ'çôØ~ð³ÿQœÏúúCYÕ7ÎìñÇùÆåeAö‰ áè—•‡œ¯Æ TùóÍVù©RTmþòiIçÃyYº7ÉŸçen†9šžk¼ecýç“U4jC?£6›šì£ŽsáŽpÝ{F‚´ ¨cp¹êa¢¡û’MÙÑHÚ[ Í„$4ÖËíWüMì l4l_EÓXÃêS³‚hFŒúÔ ±h®'·ïæˆä?Ü–PyVâ¾E+¿½úïÅ¥¢u–ågÓ?ãÑæ{[Ìõ“ßÐŽ$)ÑÚWy´×]®œÕ¥?Ÿµ%;nÏT!s(`;)%ÿÅv+{ÀãÆgW”aÉ!)€X!µB­÷aNÛhÿò­o’®Ï{€2$ñ™‘”9PŸSiÞbêÚ3ëjöÜf@¯Š·‘~Ò Ô3ˆ>Ÿ–ø Å¡ `÷Ì‘XbŽ‹q${îƒpÊ^±o_¸Š%l¬Î¯Oº±[bðl¸u»ƒöÔ£§ AxÀ­a[[DgÙ4ÖÎï‚„Äåw5³AªLæI-÷?|¾¶Íâ§j…§J]cPŠp³;Ñ®*7ë8 ƒX@íê» [ÇS á»ÿáÙ,h}Ì 'i2íö¡¦[b6 KsŽ%èçCÐ[–üpMý3\ÊÆ‹Ö~ †ÓßÞŒzýãÐR…×&ŸÀÔaÔ«ã„:\°â/b®sÝq&°lðñÏÓkîìß4%`óp˧ϼ,W-'WÔ@˜ª¶ÿB60$>Çük€þ2äÿ¦X?Î,ù~..VóxÜåøû£¼Ã¬ ½Ï—ƒ^¢Ï÷¿„qœ>Ì›G¸=jòˆrPø¯~åŽ^6!jñýÃòᘇŽsvOå§€S-2•çå?—âwlÈ?”‰«öâwÏºÍ €— ò½O>ãUg©jàgîÍÏ\©ï™qûúþ6™úÑ)±gê™l bgj„-M³A¼þŽôr‘'Hò/ùÕXÎÔwý¡Âœë s‘¤ÔÙÈäȧ?ºÍ¶‚ü°B“<ý6@=]Ù(ŒÑ~ ¬éÅ™ÈL]_PÚ:ÁCg=*„&U4™  Ê칑ºVjB¼¢Ý6  ®áI&ûú_Ž©×X…â¸"hEK*֑ùñjç{º§á¼ Ó) "Ð þ€½º³ÚÛ—C@ï‹í"®b(èßM5j«Sâþ–f0d¬o<×1†ÛÁ)ÛMž`4î‰íË{¼| z=ÕÔAÀŸt÷lŒ30–ˆÖÝù‹þú§9¸BÕÍê•ʘŨhÉ{ <óÅ1/ñÏ‘®üo¾<Ò©GyÝGíÑý£~Gx<Ê>zïáôÿÇþµÞm~Ùxœ·}ÄWÿgŽiQ>qjçÀ¿œí#žúOÎvsçB’"ÔJOPõ£ýÉÒ¥)£üÚ¯1\vÓAVsz_ýóôõg(ç ½æ)4x™•(  ­ <ø!Œ†FÅzO(¡QÊÖºØíhÌm`%Q w`.R|C“&?oŒv$¢©VSírô7ó4鎦š +>i| Í¿o#ºxO-Ì ÐØÆ¡%}¶SAf{hy’À¸ìU;Z}gGŠÑô­‹}ó€åÚÈ{ÎôΞm æäõ—l£í±kû£G¸ mÝ…sFF€9fÕqSh °ÓBÙ³mHÇOcúB„$à©La:þŸâà?ä€B ®]Ú"¿w¢—Exêâ‘÷@:JTÈ×–wš»é€r†tª=lh”ñ~izÑ]V—“f0Ì£÷Å!îÀ”o‘f`Ì"%²lÀÊ–÷\–ØÇ\ë5Tö‚!S¾G"ÀÁþá~P— pn&yœ¦®3þï};7Ç?Ð>zõ%ðퟱøæIg=”1:Ap©°žÜÙí@¯ÊH1¦E‚ȰçgÉ@q{zû·ÒHáÍø˜‚”âO»ï:ü £ÿtèÆ0Èùé*ájÜ”µÃ…èÅ›¤Œ 8\«;QÛ Êâþj± jÀ¥­)j'ÖS2.3€º_Ì‹©ç I¬ö³É´Z?™¬,Ï‚v’Ççס_@'Ý<ÿZôLª<$òAßUíÓÄ0ˆß¯)à ½•vA0V¬ N[®“œP­¶ñ60C¹ßìŒÁ.èÇõž;áä…¿‰ÏÓ¸d¨ÜþjÏanê':ÖqG<÷ß¹úÿíÿ[¯éÔ#=úùÃ#Ýx„¯£üðcÝêD#ªrçÿö‰Gþï?í¯=¯'0צ(Îø×þçùq½Œ¥³ÄœlïQSïlYÚÝãz-—ÍÛ’­î¢V™¼Ú|ô‡…ìæ¯½íÝþÉú—+9j×N˜÷Ò@¬ƒ«Ëì¨Ë7êâ[/Ôc1±áŒ‡únØ{hg:£¢©Î_Q$‡ç„uoDÿAÃf—5=o¤ç9°a›IþEìuÕ³¼Lh oÔÝ`Àä˜g=©…‚§ÐÁ×âl´x½ÿfÊÃ0´´­$¶b ŽV¾©‡gÏ1¢µ_$\LÅzhSôM¥ìÄ+´=ÓÞ̲²‹öZýÆžkÆ68§O`%ÏIˆ Á‰Ç®ñhþé|âxÿè{p¤Çê[”kwÖ×ÿÐCé3Ÿµ×¨ëšú²3sü8ö‡ÿ™7:š:ö‡ÿ©osŒç]ëçD\ŽëØôä™]%Eý9Ñ…ö>WÐŽðÆv£áó–—d]h´¼ãt`ý¯V`wÍ@S´Éõ7Ø“ÑßÓœçS¶ýáâÉ9?ÖH´4Fþ”gZ­g¤Qt_DkõwÞ¤ã"´Éb\]…¶ßÔª®ìK =žñ—Æ5¢Ù}SÀ,ác»]éØü‹›|?àDŸðÄeÅ]Àý«6µ3ø/ÏxçòƒÓO€ôÁʽý /¹b‹õ<(-Šç)"5Ûï¨ÏÕ™@{‚n¨P è霞©¤£—r…h°èIÎF“ã•ã¬ð­fž8…¬ü(Ön×ÇZ]ìdàuž©_}D|œÇ/¸\4¨ v„[kÌ®• €è Vž/&Hð\¬Ød‰©ìáøñ— ë5Õ˜DÉè\µº¡(t’…ãÐð‚Ò \íµ"P±]þ jÖœÎÌ“¿Ã¾g"h½ìeÃJYíÒç9% ³7¦XÜÿ ôÆ[‰ˆ%6ÀÀ€€˜ÌõödÝõÁ‰ãÀˆ˜´T%05fbù³´Þ‰cPÝó«»w%Áò›î‡bÒ°fìÜ­T¿6n+‚uÂàþõq±õï ôDoV™çÐ×6j$IN¡i9[ŠžQTÍžÇõ×U÷8Wÿ(ßþÈfßQš[˜¨;Ö•ÿ]ê·qAÑŒ•ŽÞÇíÑó#ŸYÆõa‰IÆä¸NKm`u¼ij¾gÂ:/€Ú)ŠLð àŸO<§Ít©²¦hÐ7Âú‡Úûúز™_ÊÐ`­ŽÕâ±^äán«Ú‡íÑýÿô?ö¤ŸÇÚŽÇûÈw#¾GY”h´%ì3Íé4ŽÝÝã%õ?ÿXß@’a¦=±ùª«KþùF7åÌDJhá¢I]ð ZÌkRjªŽ–Íoþ4ËA«Ê‹Òúsýê»)´ÙÏæ´ÞE~ì÷°-XääçÐþ™·MÛ$€Q®üÿ£ì½ã±þ¿ÿq{ï½³÷ÞÄóØ{ï•(!*Q…’P(II4DT* „hiY•Ù{į÷íw=ž×í¦[·×÷ó‡Ûãùt]yþÑu®sîç~?÷“#rí8~+rN>Þ äñ†ÙÕâÔÏGåå²@ÅöŠÛA1¨³9Ö fVUgÜ©èO])¬~YKÒã}.T I¸ËÙѬÏ|¢µÊ»?ã+w‡×ÀñB&1Ô$8ŸǸ‡—sÃÇçõåÀukÉþäËÛÀ-ËÚ&öx´*Ý^ù‰ŸÖõsCJŠ (j%Ïî¢Â~;7O´ÞQ“íõÍ!W@œµD£áÄ&ØG»³ª $C™¸¿½+)“ m÷Ëb@Ú6|SÜÄ È$e}/{ròÕIyÏAÁËã›·=(Sýšm뢕3~a½AÍR¦ãÿwP_þºvîB#h¦ªßŠøÚŒ*3¹ÖÚ sU÷í³žRØluéiÜ=Ðgblàyµ°aÖä‹0]¤ i¸†¡Bü2«¹`6K.3Æ^2?K^“£ú 5Ç~i“Õ*0Ë ͹jæ7¯/d©å€ÅÓÞ°×Ï`ÙÊOXü¬Y×›hC·=ç-³oµ`{1áân’°›“¥8àJíãvé/ÀaÌ«¸{ŽeG™½ZWÁ)¡'öµÐupö¾Ë}U \0ã›ojÉàPZÔÈ aP8ÑzY‹x¿ð2'>ýó¹º6ÙùÜßö#¡øCñ†êJÔ—AuëÆ=lÿÚsò,ú{(®Q~ÃûœÿÀ‰x_3J \ã±:ÖÀFRiêµïûà8å+´ Õµ¨>%ðhªKQ=Šö±}Ú§~ÀœÃïk¢ýL¨o´qÏŠëïþíƒ9¶ÁØIžÛÞ@⾦RG“Ç}W±‘úÎÙT>ì×€{ù¡¹$l|ä0-iA,6o];}Ü›*¦’®{Á‰ÍÐ1q¤?ÜŽÍæ”afÇæÍ…ßpžöÇéÙXŸlÀ–:ÊEö`+MU£ŸïÿÀÖØ43õm[€$ÐeGžð®‡<ž¸{ÈßÊîªßDÜßbZ¤@SÏHnÁ ´ÝéÏÅC€®ô{ú5  ¯3ÑÍO‹ÆœËï’^weû`Û·åËWîÀ±64­ ܹóâ$€wþ³Ãér">l¾{s¶d$9ï€M,W]v&Ö‚ðN–—ª2Ó R¥{LäA,a¡<‰Á$.®ÿs¢©ÏágcíUAÖÿã]®µ"Pà:»R~ä6(­4æ~VÿæM¡AB :ÄõN×½ Ô36Åm¹š^¯­bOñ€¶Å¹ƒí:›A×ã÷Ö9OAïĶ5š={þ|¡—mõxe ªd&„ÀpHˆöÕÑ{`¼ÙOõª¢5˜ämOŠ: `FÿÄŸ:ÍÌOpR¿ˆäKyÞ¿NU¯ë»ãâ`S:[øLÿØí{Ñþ|ÓëïÓ€c´†«@*8ÝÜ«1¡ÃÎã66¯86«¥‡ÌX¦¸U->¹÷<Ì›œ”>²Ãö™CZÉz—A$µiÛ¶! Ú¡95þëÖ1ë6Èw¸ß áCÔÿD}—øå»áCt¢÷£øEõcõçA¯j›ñ¾ ¾×ì?âç ÑIØ'ƒêÃ×g¯|R¿"ˆ½ö¨w{8W÷kÞTÌ]bþ<‚ïGCû¦ÏðáÀ»W±znDœˆö%fö´pÇbím´AVDÞÕ•"ëžoÉ·`ŸÕ÷Å‹|tÁ¾ÄD.ýÏ?€ñ=ˆˆ‡p÷寭9Œýd£­áÙº•¸êÃ¥ÝÆ§ã±_³‚çx:÷N­.„~¡Â¦æc$°Y xH§‚Íoñ”™pÜŒ->û ïÿ[q$»sN¬ˆ‰,c›ç8€L<Ÿý~”4Pš×D[Úàù”Æs[Ôá ?¡• êáºÔÎÁ• ™”ø4ø[:h7ŽÆ¶¦öÂf噯ï< ÿòÓÊù<0°ÄÇ/Dƒá ÍÉm`ìó€2vmL}Ÿëî*³.²Öþ [`Qû] n¬ê§|s³âÁæÛâñø)°ܹ¢ô—:TngMSKÇ™_lI)\àì32Ù‹±€KëÇ ûØÝàd¢L§á ¼vïì~•‚ç/ɢðî—É~bz¶¬ygÒ%ƒŸÈ]Ι#GÁßK4T_rl»nóŒGa‡o[JRé nø±ÎZÔŸt£ü(ÇÚf‹×äÚ›ð¼„êHãì_|!ƒYšz]o“ð{ô~(¢üŠï?Ü€ûÐýFÞ½ÿ/‘ð~„NÄ÷!òl¥rrÂGF)ì)µl½S ‘GD{¥uó_üáòž¨q>M¼ôj·mÇÜB+ÞÂñ!ŠkÔ—%ÔÛõe»–æfùÂ^Ó¾Óî¼-µŸñþÚ#7(1®ãð•ï¿¢¸+Wˆ(ÞM܃(^“§²òš¸‘¾ï&C0¶Ñ~N®l[^”{Q@ý¾…­:”a@¢Iî™Âä¤ç}˜i\¤qXæÍ PLtªº]Ô<<ÝÞimêï&?ý‡À ZZ}¦à-0½õ|Àdg ,­šÖ…̉Àæ—Hç«<Ø£×*€s¸Ø8CI¸©Ã¸µæ€L™Ó“fAP(*v<0 „Mc«¯K¹€¨ÖÇÕ¼Ä' ¾Ý¾7·@òKâÛ–js¹GIž-Âò_ߊ—,=e^²«•OnƒÊp®¶î¨ýØq; Ç4lâ.õþíí“®SÝ ;®lSw¥ôïcõSîÌ–g`ø‹ž{·9 ºý¤¼f¦±ÒU;‚ùA6†¹•°<¯WQC?Ök¯:ƒ’ÀÎ0oR¶ ÚÉ{9Àqi´ñÈÕ4pÎ wüÍ<®ö¬fö±Òà®&õKDË <ìßµ¶Í€÷ A>ù2Ø29j”$ ~6WO8DëƒM[åSÙ>84ûYáÇÕ§ |jö³”áà5{TSîw1­ÍÄ^Ý}º ƒÕ¼”/,Þ÷7þâ û×PŸ¥$#8¿%þ3¢úÅâÐ>OÔ÷Dù¨u=ïD“£%qo'aÚÆºíïÄOa}õ̋ز†JCëÄÓ[y{™4¾·÷ëÂWIÙv+¬/¼ç¾c0ÞùÖz7ëûÏ.œ‡ø /âàgßÍ—qÜÌØð€×Þ`ê;Ø/®;£Mßq⹃·“Ò–°iµ~o¨~b³Œtšiõؼ¶Ãõ¥™ƒØbơη[°ån;Ò†Elõ[Ó©w½ØZ5ƒvH||<ÈHO¿ÝÃ; d_j=SëÂÄûÛ”émâ>DÄ?|/“]ºÂt¯8GŠ\&€‘Ë3r½·˜©ªóÎr«Û¯Øo¬;­KeøœñYàP þºnMœ’…-ÂÇš€³óâZLÅàr½É ¤¯\S’‘óªcÀC±«Àâ¦ßÿðËû+âA z%Å&6]XÒkÛ"¡’Óã&AŒº—×çå!ßMkZža‡×™’u’^ŸÆAšñ µvÆ$È-rc²¹rÖŸz6€BšFvà”(«ˆúˆ›IƒÊçÝ*‹dL VV’Ü|Î4‚WxCäN‚f«èdÙ‹c gøsoa#èšNãwÐã4Ø,Õ忦10 ’¤kÿ†¦îsÃ]`¤HﬣSÆÎ¤kΕI`r,íîaŽJ0­¡Üæ”jáµÑø~5Ë{m$rrº`ÍÉכߣ6 ’Oµ[ÀnÓ«»´v‚}C9ç¡P[p¨Ýþ5Z´oLuæxä€Ó‹&¦.à\ µ%Ÿý¸”ÿ6öô×öÑ{oEÂÁêÔbö?uiÖ»(¶28Ì%ž¹y¶Î=¹dÏ |[ý\e†a#w ŠÒ»‹ñ¾ Š?”_P¡ºrãÞÃéÜPÝŠîÿ '"Ý ªQþBy õ>DûzQÞz>Ô˜l`œõw_Õ·„zÕ³¨?ô×^D”·’Ó;lçú‰|"a"â¿2Ð×%ñ¡¼ä#šAÄúS§wçmÆV"Xñ5™k3VPÙRnªÁƯ^ÈHÃb“&²›ö&Á¦ì¬i/8Š×¡(¯ÍªùüÚbMì÷Ä>Y=ý¹[¹Å(¹¿¿ û½2>"àR$†29­Ý1@zùn™cÕ#b~«õ5¯ÓÉ*×ùµcº®@ýnÇõ»‰@«±à@öÆ èlkè úÁªŒµGo€qæ&›ÞÁ<`9ü°‰ K¶° ·t ª€£ò}¦Î€psóh|YÓ^g¾é(-;àW¸—ŸS ÷÷(t‡ÃyŒMÍ?ßs-îá_Cê½£ êÜsðÉÍ] .¢ÙýÞC$·…¨P€ô%>3¡Ÿ{ANçJ×PX$¥ûù”ƒ"¯^:|T¾'ò&ýµC­~,¥ ¡*eRè Z4~ á·$@{=,×læ]:ÔÌf úžJgn‰<¨ñÿ‘¡ò'>%ÝôÛÁ(álsÿ"Ͼ§¾Uœ¦q—-ÔS†À\8)÷sâV°è ÷Íô«{ëí¬u­`ó47^¤ƒì~ÌÚˆœ`‡Ücâ¡àØ-ðÍÚœUîÝþù<¸\»¨ZøÜ”ÔM/­÷®Ëv¤ ŸÀ³PD$( ¼3V‡ïÓîƒíi÷¢Å§A¸¬6Dœº(§ÚlÉÂ>bí]»M-l‰øpÃ^n_ÿгºÒ¼zÊ›¨nDýÔBõÊ“HW€ôc(ŽÎÃyCtOÀ‡ñ‡ˆw$àHœGDºSÄïë¥Ì<؇}4a¹”ÐLäÑ¾Ò ?’ìUì“¶•×’ž#Ö9­~´üO|¾Ì¡«±Äzk2"e~ï$Æ'¡‹ãÃO­?¸¦F°|1²¦Ø-ØÏ~ñ›d÷°aLJ Œ‘o°QYÆæþlL8,Ü%X›ð{ôœ+à>Q¯†ð¡;c`dü%l>fiô4϶8¼7†³´[‰M>s=å¶Æ[Ð&tt‘øõnðfC÷†)/ÚeîŽ,µâO@ÝT¤ªæ#NÔãܼlµS|˜]RbÙýÀUîÚo3ÅÓÀöã˜ÁÏA&àxnêñY‘¸îݼç#wx®¯|+ˆÿ|õïëùÙ”@§‚,Áñ%l*|ÀÃÙ"2¯ÓNh¡˜B@g2ˆƒ»«SÙA}¢É% Òþ|­Zû˜@ŽsæÓX^(jNyp]Šå6Û‹¹Êj šË%»ržÔ÷K|úB šÇFdŒ(WA»¾.‰fÿoجPóÒù™2è¿`п—H掌zr`ذd¢èðz‰„3ßüƒéêÊX$À¬ƒ!P~^ï‡ZÕžøùi0 l¾ÈñdÅœ{^åoœuQàð„1í÷pœJ¢}õœ½L(CytÀå§YA±¸èoÞ¿k<8=ôÙ žC\ÞJÞ_Žõ%l‡-˼åäMà·io_Vqø»_·Ob‚mŹmïwdBÀdEÍ£Á ¦ºd%zø*Pûìˆþ½‚µNŽÑœ—ü ¢<÷å>DçF|ˆê×áÿxC„7èLÑý_|"â ß+ÿ‡UgŽº‘îÞ‹÷‹žÙ^VÜ:Kä ú‚F±Ë-gþƇ„:}¼£º½.nú7xâC©pÄ7b–Ǹàc YnÑw¼¾î»ýñ&Ù \÷ƒ‚½A91ß/>¤£ó†tî96Òç?ñPÍ…¸Wø‡ÇkÝ@lÚÈE{˜ã6KY³+žšˆ}r8¿¶ü­â`ê5wì÷ÙEÑŠºY Ñ©¬k÷§Ò†þt1÷U¢.Ü–•<ø° Po{Å|Adæó‡VµÎ @¿tà" m(0­ÍL›˜õsFÞÕ3K€CNœ¿,gpiìÝc7<[ –:×È€¯u…¢öå.,k”Ü‘XÂ'ž§X Ñ}ÏIÔ÷ø¥ÛæûFÏ‚”§z`©<ȲÕU–€Â.k¿zyÏ?yq>ú¶€!¨Æl¼ êæ´tkoA3†òøáÀeâàÈY5æiÀL(wèïåƒ"½soïõƒ‘EÓñŠT0¡~«›› ¦ã·–¸ù\H’Z‡X Wðâa›ðZÓç Æ`7üþçtî8|}=dm’N~Û$gb¾ƒ ¹I¯éÛ+àúºH°8h/¸?u?i¨šž]'· ZÉ—Aõ®à{#;59ãøÝ¯n“m½ Û½«£{=à°²ñ&–'Ñ <7h-Ûüx‡tΙq`##uí¾:xý†>—û8Hè§nć¨ÿ²Qç†âÅ âÍQõCð±êþ…§÷S±vI¹æMاaÚ TE‹å¤_ù̵|傹=Ökq÷êúö)üsxG¤£þìø+‚æ‚®OûÚÙ|`ë»ÌÑ›‘tû¶¥JUæã^b}Ië±þ®”úœÃêŠg[ªTþÞ³í4âÆgM-ú6ìÅfºô7 ÿ&ÆCè.–„.ylùé5ª1ÆÓØêëãm4cØÚ]²jË" q®éûe§¤±êÌülê@ÖœXEà ù¯FùÞu¦×¬¼¶ämbÂÙÊ¡u\—Ƙ)¸ZØÌwïÖë_T"êÒÄÕö7•»Uʱs=Àá+wª«z8·$[üºu¸ŒÒXœþÜZ^ÑE²['ŽÉ•ùL$ð9ìR K’ûgrjÊa“gš-7Õ3aIýÌüJDO0°ìyöÄÉ„õ9Þû‚Dt»&¹Êb><²DÒM¡‘ªêäc+ úi¢Rv;|O>ï9(ùÀð•¨îf£¡Ý#碧Á„5ñSÚ™:05Uuó±6³c©ç+Z.€ùëºõXP°Ìæ‘ÞV›û“?P÷pïyìünì©o˜9óÿ©K½3¿Ù®Ì#ö%² † œoŸ¸)Îrn‡Îz%‹Æ«ü–§ÀÕ®Ÿdgª,¸E­¤­|w÷R+º†2 8ìøAÛøy)(Úéì}š |ÙÁµl^بs¼á.7zœCú6wùÂsP(Þ6Ö«èDõìéLQCøéþÅâs„¼…ðmc›\š¬ßObþ"ÌGàºSŸêT¤sCóPø|ªC‘Žís_ßǬŸx¼ãºT¢øÍkË~–] 0^”rûµ€ æ.7’ãdžÌWs¸‡Æ±á3ÛgŒÄþÒoÒ¨ÿM¹–ö²ÌnǦ4ib³:Ö}vØÜ`—ª@›¶P‘î«ÊŠ-5õ ÊþøŒ­’óm ôÃÖlóåH"i€Ä:È/úFzooW d‰2‚N§‚}ø@ûN L‹~t’S¨Ål2ÕOMç¡ûËÞm@wJYøvqàÿüº÷ð×)S©#UfŠ%°¼=ÓÈgÒlgCFÙW=ãŒfË×+ÀU°=•]ƒ xš5—áÍAàûÉD6gcë]ÍïH^À&ìëäncnòV[|ô Dé‚›º3=AlONšUd0HP½_lìÉwíæÁÕ9 ݽéíé&U¼£¨<íóÃ$”Ÿ^XÔUƒßVtr Öï¡0Fý4®E-›í­„šÂÑ„,Ð9®ÅV¤›KŽõ¤;ý…y!¯g›Á@£Ìähu &&$l£û—ë<©¾€I@@ÿëe50]¨ïJó‚…ï‡3™Á2Äê­œ»,XÛHܹ˜ølwè(É6ƒ}ž¸J¡O8¬5Oqéä‚“Ù¯gºÁ¹èûÌV!op©œÝMin¶Y‡ò"À#ìèå®bAð2k¸Z{N| ÍXÈþàöíjMär{A„÷Û­QÉ  ²ïÿžƒµ‹×æ„ôËã<:Ês(¿ý þ>Dõê¿ð!Ò•!žÕ¸^›0ÿñOàÿþ5wˆãE¤7%Ü£¸ÃçÏ@è×¼ÿqñqæH±žD|ƒûÜÁ†l~b\näï@Èˈ?Äõig·ìß~ûá“•lûé>ˆ×—¬EFùíÙ¿S(P »Äž”î‹MTìÜ噇MÑ?‹›l:öˆOüî›x\ÎÓGç.¥b‹'Ø RO?ÂV¨>tïÇ~çî`%õ< $²Ÿ•‚.hé aªoLÄy&—¤ Ãå·z+ã€MÎGп9šÌ8LfZW—]NËé„‘ ÒOÀ¦=|Á8ØÎf×­…çüðR/u5p÷¹äü¦éÞÙ¹üï°°˜ôž ¡/µ/<Ç€ðkõÅ3 v ªÙs:S-Ä^׊a ¬ QÐÌÿ©úB>–|»ù¨ }.¿ §¹þ'ÞÙ&éS¥÷"ZÀÌÖº‡½ò2XH.·ÖëƒKç.}:°‘ÓŠ­f;×v šÈýà`{ŸNÚê8ƾ4‰;N==7>.Ÿó¨¿“5àÚ²Ž=aÜ#Ê¢*®eƒ§æÇëïn÷ƒ·À;«‡ó`‹ÔͦѤ]°Õ3õ¨äðë^©®œÛ6 G¤0A€¡»ª l¢:E¿ü§¾¦ò¿ÈÆ~k+¯¿x«³ ÏK¨Ÿ‰âáÁÿ+ˆÎñˆò"ʳx¿tCü.ÜpþKgŠëK7âÃÍ#"é†~,®/E}X”_ ý#ôý€âÏ«¨ßC˜çBsÆøü!a¾ø/})‡(¿‰jß1oÿ78!y¨]÷o|¨;þ*DVà¿ña”ÉÅ4@Âcö²¦çIp}éÚ³'«×€šVº‰ý‚ÐZÜÚò…æ'ÐgÙhþÁiLd%[BùXv—\ž<µl¢N’ šÀÁåM’UMœ?t#%]µ¾çO~k”^ÁËå?*"Ä@8ÖXk>·Dƒe.fÙ•€ø™õ°;y Åílu²“dI–F,³›AÁv»òÃfPvó,(InUE…䨤8PWc'KdMÿSÃÏÚû7îûÊÚBRuÚvÏ Άkýª^#µ™3zÓ6`<«/Q!{LÛ#Û½=.ƒy׉M?®iùÄ}ÿÛ`ch&‘švϱæ)fpxõóÓ‡tàd™YazœGHLâïƒkéݘÙáYp¿Ô´çB¤ x>.ySÁý|HCû”Á7>n.¿ÙürYœxN~€mœÞ7” .Â!cjwÝÿäùafÇw€§¡ÃÒ&ŽŸÒȸL‚5iXò–/¸áu ú¼£üˆú*(¯áóOxÃó(¯¢8CŸ{Ä7¼d»4‰½ÅóÛ‡±sQNòxŸéËpÞ Prq]"ƒˆÑIг ¼„Ÿ„×q~qüÌmþ³çñ¹|Ži#xMÈm¯`qné­ýfUâˆñ‚ý ÜbŠýÊùî# Ø„DOaâ}lŠïÄûå_ŒØôç[÷œ’®c³£¼ñ“ÙMØ‚„ºæ›nliÏlsFlåþÁ÷áGŸ`¿Š÷`ÏEDlÏþEŽÿíCl8»1@Özö{ ÑP¨T$gwåá3wĵâ‘üOË”@ós¯‡µ¶ ®‹axrI4)Ę*³Ô©^ËÎa‘dÍF`Mо¸îä lÙÊöƒf£À~é@ㇹAà¸ñûFòkà<9@ž\ÇÍè€ûžŸü=Uà½äy%C÷&ð~b»EBÔN P‚pê}Žò¹X™ØbôùÐ]óÒKÈdñwö¼}® ¹C™÷J”Ÿ_ÜÀ'âóˆ„|Œø†6Ý„P>†"¯O˜ÙèkãC‚þŸëXι>«n€ýŽMÂC6Ø@é¾ÁÁxb>ÕŸ=;x }SjèI‰M˜;‘ÛŠM¨ûíØ›¬ÜsÁô³>Ç1›Ï÷ýQ6»ÛpìžO'¶ðåb÷ƒ¯ÙxßæwZªŽOÏfì÷;?¡‚_l@rqì­L79Z®ÐH.çwy¯íuÌšôGÍ%%®/­ÊW© Æøœ^ˆ¹ /§\9°Æí¿µ·CØõUL:¿éÇèï…Ø0kàzJ—•¿Yx•»tRo]þÖjsŽI’ß·úÛS„/Oˆ –Ù€¨¬µÑÝá k=æSü«$~­ùèeÒƒ´¸L÷“w> Ó»W>óÉ+Pä\Tj=ÛŠë ¬riDÿ‹¬Ñ‰ö ; ‘dq_´ßl*yžºTW—·ƒÞ‘ÍÙ¾°.OÀõ9cú+'ÅRz°?å͵¾6d+C#ŸžK_ÝÝ—ó [0Z§òü¦†-³ÑJK:ÞÂV?ŽßžšbÅÖ™žó1Ð1ç†32–¸Ï­ÅE®+öÂ%@%?8v©hüzù•¼Òˆú9Ïî\ûIÀÂéæàâ;¬íæüÛx=øbýyfmàt£­ëÜÜM‰þØà£ ÓÞë ‚ ®Ý›”@Xò©vhO4ˆ¼? s¡tÄý}Éú•®€DCÄ–darüEÛ)òÚ¥á’T eÝdyTŽÜ fMŒµ0ªï•¥ò q"‡G¾û9h}ègáNy ºÖ÷ŶWü½¥lº¹Ÿ¹g‡Ž €a¬AŠí—(0–¥‹ÝÚ&s=ÿ˜sÁ¬?nÿpðm°˜2iØ¥œÖb\lçuçÀ6Ú>?ó…$8p¨Ó×´Ç#2²ôªüàÌ@wçgT¸T‘Òå ~· M­1[ð8uM”åÒðªÜ·~Þb¶p+äÚ7âsÀþºËÅ•qJpÈW×1*€”_Ѫ·/§²¨ÿ»Hl¤óÐæ‡«¸ÏÄFß6TŸâ¼!¯¡¸B÷(¢÷!¹‘§@ùáPܯ†÷.ÄëAô¹%ø¤¡¾â ‘>º¢&Õg¨°Þ¥£­¡Ý8¿€thˆwDó¶=“fŒ²Ög¿`ÐX996pìêDÑ] öK<ЇÑ3¹äÉ–ÍØ¸¬ÿ+nî!l²Ðp[®Ñ,±o¢J&‚âÎ?,’ßøPP"@äl”s\ ±µì!ñûš’ØzÁ–Åë=xÿ„t2%|ùP<)<+ ö'5ú>Ε6ÿ‰ª7)@«Á÷åY00ÏȾ1†§>_> vc‹8M³?^_²^|p „ØB[cn¨w‡ÄÁŽýS)ÀÙñˆ÷ú©àÖ§_Ø+¯ <…œÜ*;€ß ò+ëio,¼ë=%¨!?¶D *ëcêk@ÔóFʬ»-ˆ5ç ñç9í:Îë yž»p»Ú.–a¿Ä1!²ü׎$³¼ê“ú‹æ ÄÆÒ^xv(óSJ,õ‚Êë’S÷•@ÕJ¼…ÌJÔíôvºS—‚Æý¸²¡S@ËŒî9hOä˜?þ“ï*uË?œ½£]¼&· UjˆKy ê_g?¼,CìCxêãÃ`䬿ðŽÊ Œ#8¶&-°ƒIîæ¼–Â`Ú"f´Ý`N—ÊRî ‡G>°N‚ååW"_²€ÕœhûìK°9XzQ®n ìÈ@íöq)°Ï÷ ÷ù%$4àxf5ðäð8%N7ÖçãÂü\ÔàrV2q_S'ÞÏqë¹q-ùÆ:Ädw¥Í¹Gƒb»ÏÈÝ…ÀKê¤p²E¹rZ(QížPÿÅгøocžD8Ý£×7âGŸ(nÑ=®/%ä14‰òÊ[¨¯‰øC¤ƒAù ýéLˆêVÔïDõ/‰¸žmƒ¿ZÇu²ÀЉ2|Žä_…æ‹ >¨¯ƒôkƒ³Ç3Ò?à~kÃMk›n~ÂFΧ>âà†ãÄÉN’èmÌ€MUkøo6u7TãáPàßúÒA 2«šû¸ÎtÅ2Û8´Œûmd-+*%„­¯ž|)Ú¤dÑŒû›ÇÔá̳ tm"oAеQ=¦H4Ù^´Û:‹>I{½ªüÄÖQ`dêJÎô Â}Yuï ¯ Ì[^çÞÂŒZ`§8ç-:k K›×^Øâ~S<7_ÆVrŸ1G›$}>j¹5lÒƒMr±âœ ühM2TÅD[*<Ž¿¬±ÒœÍõÁT ¾"\ÞNu¤"w„9{TgJ7ï«­ ·Î|U𷆍šdf€²Í«‡Ùõ¯@¥6>#¶šÔlj‚5r¯‚ú¼nôïŽЬÞ|r¸Á ´ „‡YV@÷æ¯-sÚ¡ ×ª#(@5 °©$Ñð\:4§‰…Q‚'U^MzG©”û“G:ü<_³(Þw—Á‚{Ò­ÃÀr”tt±!¬;bMò÷œÛAª]-_[À)<0ñX‘Ï?Þr2ý8ÿø’·Y\}µ Þ° Û̶×êš%àQtÈAfG'˜,Þ4éÒanš»½ò9@%2PÐLµûnÚ°-¯ëðøÛÐý—¾í¿ðáFcÜ!? ”Q݇óƒ×÷Jt Î÷á| ‰^ןð"Ò¥¾úNý2{W)þ{üïþ>ú{8>$£Ïzøâ)öQú ïÏýTÄùฃÒ*ÓD]Úµ™ì~+N¬·—–//ÊçNç[ñ>jr2컉[³¦½7ö#†“ËRôîoñóÃOƒ Ëpþ÷/E¼¾õ)£}Ï;±) ïÏ”{pÓ™hÚÓƒsÙØüÝÖå窯±ù€:g’|léìB…Ñé l•.ÉÜQ¢HE4¶Þ’Ý «Ìõ€,­“ü”æP„s>>0ÝTVϯr}ȦB¹”C-@çF·èÑ3 ‰"Ÿ36æÜAÇk®w‰:Ó©Ç?'ÜŽ.ö8‰{ÂÀõVÀ0ßç%ø·xw¬ÍÁWϹ[¨A8Àp`ÐWDÉîîÊ‹8b§rk6‰‰DÔÜ¢uH ë¬.öØ€Ì"—[¡1ÈÇ›Ùç§²sûMí7Aùá‰QŠøPPÛñ~Ûé*ÐðÍOp ­8šÍÌ,f óâ××ÊM” §ÛvìÄùxÀºÝœ”r€A›bjÈb&9ªÆ]zÆÃWI?ˆÓº|'0ßFJux: ,MÏ͵郵ãÝÞFz°uÍ˳5HûÄüþ¬sà0°ú6²ÿ8iî¶<“ùœÏº<à¤Wº€y2«up»tŒkí,xØõ[¸åƒ—päÞ´Õðá”)ØsÙ|•¸kEùlÁOûáâ†iðËZnÐ0€Æä3oL&@¨w™šËq ªc7=ÖꃵY׈êÏqáyñ ÿâ7âÀÿš?ÜÈn㏾t£ß Ò™¢yý þèÜøþÿáÂÿò©ùkÞÿ_ó‡„|Šâ׋oÐàü?Ò•¾ÎRýH•ƒÏvŸº•&ôE‹8?Lð/ísñ »%Ç‚ëKûG»]_ËL}:¢=Ú»,÷`£æõg8WÎãóUµYl5ØT@ôés3G°™š ;¦¿}L=ŠÄ…Î5à>ok×;è~FÉGÑÚûáŠ@&¨T¹më ØñS¸à}"P~âÊ^Žaš€ S¦aÒ@÷ÃZOïëA`øØ«Û-Ìó¬l‘U™Àšž˜JNûop·`Ññîà²3Î& ÀGÅ·èÒÿJ D®Y0{bnÍ­夵ê·Q±¢Ï[Ö“Aòw(GDHMòVgù¾yîšl”œÔ\í*›@e×Nò éŸAÍ#!¶Â…4öꊫ½>Z¸e•ß¿]ù¾Ší ×±^µUµà±áñªœN0 fhmJc¶=ÃwÜåÀ¤—±l¶€Ycý+=¡~°øa(ÉÖ$¶]zëÒ`ëTóÉ+µìרjîpôƒcð‹ãÒÒà49¯¨.—›Cž¤`à¶wôÉP„6xìr«ãZf¯¬l/Û'Àg¬ú´dl=ÔlÌSªþ"ÍÜþ×,àÐOïÔŒjA»v¸œBƒ"B°_»=êŽ5ùà<â-P¾BóÿÒ× úr#Ÿˆ~ãB‰âÅâÉQˆæp_m¤+[q¿æ¤J†~q_mÆ9Næ~¾å^3¦ÿûÜô~ÝrP áBÜWð¹GºS4'ñmJ®%>Øû>xöÝ.cC¼ÏùÝ™ËéT®SC~kCûMu zNã¼ú/Ç+”o-â<âäàËd…ؓ،°Í©…@lög½KÍlÞ.e0º¥ [–¶¤ç‰ÁVžüˆw—¥$ÎKœÖj~V³HfrK­ø€,·j(ù©8/µþb‰fªxÊúòxa l–åÜÛ4, ¹~»ˆóõ[.6„ݦ26v¶0`1SLÿÑ lŒL®´§ÛØÔ',¯8¸Rè çÖОa“­ÀµWõáYmjà.‘"9P ¼MX¥RÊ9à_ésÎ~‚Ã;Jy+äA¸òS ë:'ÞgSN8}Äë¼<Ü,A¢ßà/•!H“6íÒKþ2Ê-±š §vç¢ì^PÌŸ$OÓêå3Üyõå |ÕןóP:¨¾=u9ÜST>uúŪU‚FxÓ=£É,Ð"ãû^¶´‹¬œœ—@wëÖºáž?ñ¦°--œ0ænåð"0 Ÿ©ty. †tiÇ¥bâÁˆ_Cò¬i"k’¿=òL¼#Ss÷ƒéIë^JÀ¬þ¨“Þo° O°SäK[º×”ú`uy_IÅÙ°¨|;¶W÷È ›ƒ½.e8Ÿú$8ìšu$ÀÑƒÚ H‡œÌD&낳 ¹95E-¸ØÎ–ïÚ ®ÞϯÏé[úø‡0¿f8dÊàka» ” wL,ìvž‘¶ø–DylÔ{·™Ì;|Îo£®¨>ýNDq¹1N7ê¾Qœ¢¸Eu0Êoˆ?D'‡øÜ<"<‹òîkŠxDBCz9¼^%Ô½¨ŽEþ8H¨GqßÄ ýÎnMÓ­~LjºÂùm}$ú“ß#ì[ÿnùÎñø|ÄO¥ôÏÆÉñxÅçœúpœŸð¯l~"Eäyº©ïP±ý3M’§vâþkÈO±U9OzΥJ®™tH9YÚƒúã”Õ£çîz!ŸdÖ3WäÂ物œ6'k‡OõýÍ/.¿Úg[,BæƒÊùþÌ:¼¿ƒÏ×cwØ[Í+ñšo€cæ4OFQp)Õ%žaYîJ ÓÖâÜð7Ì¿?ó ÑŸf]½ß#ñˆ\UŸ]S(±Ñ—×£Ž'ƒMÃ}- ´¿“÷”>dÃnøPlRyýUPd^ª+d#åmãÌ[LÆþħ[œÊ§IP ì_e ¶üüÕÉVÐì%õ9sÞ´ŸŽ[:ïº-ëóÂ@oEª¤ÓŒÀª«ÿ]ÛôñJ¤­#Å\—O4ÇÀøŒJrГ0¥%°ædfç«÷´u‡€Åæ:“ÁŠ•)]Ü:¬×¿úÛ>L;¶ÉŒ‹aßÀA²É÷]…8zj*UÏ€Ó5ÛügÖàB¾z–A \㟦1;÷~22!ðx[}jV† bn'<¥ƒM7oQ|Öx”õªÏªhM±¶Ç£Ñ‘ÑUxžBq±1ÿmŒ»éJ3·¶>ñãmÀëSô>Ï(ÞP|¡:åÅùã¸í· œùD_¢þÌFŸš¿ö_8%WÔ¼ÿ{Í-¡ù¤K³ÊØ|éžÑ'ùZ|âïð‡gû–ÄT"¯Ÿ®Ãšé:Nä•]8:ÿÖ}+ünÍm&ú—>¹óÄv›Ãä6·zas%¹ôk¿c ž:î^öÖØ’QÐgNjÚ¿çϳq?2W²—4I@!™ÆŸX”‹%†ÛµB‰óÀ¯ÏŸÙ#§ ÜáÏýÅØˆyù^ Ÿ(>Ñ=^“Uø?Ò™¢¹ Ü÷‚fƨx“ˆœ\‘ÙÛ(b*Id®×ôA‚ù\èjRütä{‚#Èd{±Ù·àsJLÌ̬B BQÿ!WVT ô}äªÞúQ«KÓ® 5Ð1¶ Ú…Š·Üæ‚ÍdÑ󌯀þ¹÷›S’RÀ@9šÛ1æ–Sðäæc÷Œ‰@ÚǸ·9E.@þo°d£ÿ˜ÖÛÖ¦£o=4§À6ÿ$í³ÝFàÀ~bó÷çSàzh½u]–¸ßâ´Ôé´k•àæµÖ¹Ï <´ÕŠ^È•‚—e1cY/ø¤¨ŸÚ™¾ ¯žÚ>L¿7u•Ä<ئ_H÷0¶¿Ù>ߢ:;*b£ClZ@è6¯6…9P*ôQ’·.áþìZÉ}a[[pß niyÆëJÀ{ˆÃÛ3K׃o:Çô«£rD.îï;± bI]d/Ò@â‘ӃϖF mpóµwq8Èi_ÏÞŠ*ÓJ…-A¹:WÆE𨦻|Y‘Wõ4’,ZË9Ь®£ô_Ù:RŒlFl°ùõÝ”\éw€ÝçxÝlãúþ‘·Œn<¡mÉ«“T¦õ»ÎƒY¶ËýC¾w,Þç:ÏýÁ‡Z¹ImûÁöÝh-ýÞ}à°­´ñ\/8.‡í-ôçŠé!]áàZØ_7äîµQl;,×À‹é\__''øœ_pÌïR…­¾;2Ø4®¿ÆË3oVaÛ·)u}Þsp(cGÝ#ÛP|á:)wxY+§äzïbÃOW¬ ’â} ôyÝ英ÑO£žÅëÆ:åUÄ×£8@Ÿ÷‰×Õ¾Áóš£E}Ê~¦8ÿ½a¾ç ó¹¿sxfà|!êgnºˆ÷=‡ŽÑ(ž§ó êL ~aqæÖ籩-eÛ÷ÞÆëLhÿÅ’xmý𦶒æ[šgÚˆ­a/‚lti$õˆT³‰1ÑŸ­ïªBú PX?<ë蔯ߌ¦yý¹ÕÉÚ|ºMŒB}ÉtÆbeº“ XÒó³Rb:€õƒýOýŸýÀNöÀ7žƒ8Dî{º§F³¿ß£aàò§‰Þî ÜÕ×ÛM71G“Iç ð_Uð=l÷'Ä;¨ÄŠ@X6ËDþä1‰ªH6Lû ¢][ïq‡Ž‡b¢@â³oí¢èqŠ3äûb^2Zå²÷ @މê÷ÑüP É¯öa»Jz’Š» ŽòëJ/*‚ ²è<‘«¸jœ'(4H9@]þ½wÌ–(Ð(¤˜+Ïo-uãO§ AûësÇ+˜èU|xŸj z©'#zÅ‚;-$DÑ ž“C]qþÞèkõÒÂN0^6¾°Ú̦®GKž¾³ç_Ãy´)ÀÂb®ùeò°ì¯¼OµÖ™ã‰?Àv—^DC$Øoi»¤}§ƒ8¯ÿqP¶=´œI,®¿z .†÷~òoV× úíÂ{JÀ2èEžixœì-¿7^2£ç¬ÖdáPÕ^Žn){«•z0æ)ðïJ)ºúmO»«/<€ãC”wþ Õ£çýkÿÓÆs£Îtã¾ ×è9x^$ô‡p)aq£¯®3%øjàù‹Àüµ§‰à“Œ×»ýju)ÎW ¿*ô=@Ø¿†¾þÒ±!ýī경±"g?×;`#C½ûwÜýË_jtq­ýÓÁ "ï¯÷‚ÔÆç>L`T~ªMw˜ôl9°›•-ì¸WÍh¼Õ>…ûé/* ç^9‚û¯tœÝEs×[aåéÞJô›úøÜçœG"¼ËõvbŠ*‰û¯*éðDP·ÐÐý<õ´–ßçDo˜8OҌڼ¬Ù‚À|ôâÈ¡OéÀÊjMjÕ lÑ”Ýo ûtdÈú­ àÌ#ùZ£ܱ缧/ØïIûmWë€ÿ1÷‰;Ôõ ¸ø¤§ú¶3˱#ù¸Döµ}ØäÝ¢=öÁ§Ÿz}ÚX¤ËzAj~éÊÀªÈ*„ñ9¦žù翌EòAIŠï‡€>(÷úm–Ëz ª³…©¨=¤µ«6 ]¼kiOA‹Í&­ÿ h“ôdÙ€îØ&Ú/\2 ÏjíéapÀg0ï{Ì 0†À”´`¤+®Oã-Æ×#ÆY©ÁTÔŠ–ù«#˜=µRe½Ôñß_+¿Ýš\ÀfÇs—㻆À.³¹mkîpˆÜÜ»GÙŸðÓ ÞIgö ‚¹«÷ÀåÄ÷›ž…}à&Ð0[|l¸·¦›+Ÿ‡¥ ¹’ô ª!U)}(uRïþóÿÝÎ}~ÑéÔy<þP½¸Ñÿic¾û/|ˆNTÏ¢¼ˆú@(¾Ð¾34Ç‹üiþòw"èBÿÚH˜GÄu¤ö\ <‹ãCä[ƒp"Ò}£úùÕ ½„=‰¸Î‡°çâ¯}4„<ŒëÓš_Bú´šëzUô{°Ÿ×¯| =rÎMޝæiÂ÷ÏŒ3´E·ÞÔ%ò‡'"$Êj‚±é¦hnýNâ^)½ïäÄR±ùâЫêðEW¥+c›…±åˆýgËjW±ßTƒ¼‡â)„6T“êE > /וbò„<]_ Ä$æe‡-€jºÎå%ý9 ù@º»ÉeèÖŽÒ¸VþÆýüC}û¯ ßRÈÌë5Ýï“?€]'”äZñ3àë×ïäx \£‘™ÆýÀkÚl'^uøÇ‰Š1‰ýÁ?ý¥;É€puå|íÙ- ê0»”ëâÒ2úž iÞ“y¡B¤K߄ƘœÁé±ÏëV 0EùÅ{(ûüø‘ÓEôq»$t,Èd+h_¢Ò(­œWµç]@ç³:]LâKÐsœùAɺØÊUŸ¨3`0á¹ówÆ0Šsÿ"ùtL¼’Ó;Àô7Ýdio"˜O„|Ù<2VT$óU¥`c˜–¼sü*ØœStUøçeŠF;Á‰ê’“_ï,8§[=ë ™Wõ>>ŠÛÍàN*ôÒ6<¦+×xo€7ó–áíÇa‹gYÿq/ØÚ~:äùÒ>ð7~ás¶-¶U|’»ü\w‡ & ‚mþá¦| L»yóñÝXÛ•Ò’eŸmôÇÿ.ü¿âC¿èïÿ5ˆæ ÿµÏâ_çÿãž‹Íþ?ï•Bú4ˆæú7äSÜ·tƒ¾ŸF¾41îÜ)$ñ(¸OÇóIª«æØwl󌷈þà}ΰD¨œ1ó—¿õtÈ·ùÔæ5p_©×/­ žvc+œÅU6}Øo3þ‡Ê¿`ë_ÞM ½†ëÂÉ…ÎjPI%•ÆÀâ,#>¿ô7ô=qÌëo¯S!ãDµ¶6°ü ü•lW/õKéã¾\j©M½×žO(ýÃmïp_74?%BÎl°s‚“˜T^p¦)+®Žá 3-Üã_ò/†¿{–=¥ÁRÖñžP¹½ùÂ=QoP;ã:òë'håYsQ©‚ÖÄ«ê)RÐ !}Æuô•óê|ì’OY‹ŸÀ°rÿ’'0kI]Ó½—@˜§(UŸ¾–¯Å_š9Ò€îäZuØ“ïU5<ËŽ$Q[‚»ŸƒS¡V…ö'pÙëÔ'^uÜ‚•sYåÀ#SƒOÚd¼&ëíÆõ_Ö5ìš^:à§Ï6y¤üëôfYøÓ &l‚?Äu'È}\¹ßrx£ü>F¼ªÂF´d³”Fãx ß·Fà×Q¼ :r£þ¿ü†Q¾Dyñ‡H¯†êJÔÿ@}|'Á§íéEº2|ÞàSøDÔ÷@º³ÿ‹8>D~¨Èçôèl皢ÞÚnu*,kž×þò½ˆ;ÔQ( ñ—Û-·háócdßã#(‚‰úk¦S9}›ˆþ¾„½¾¨¾\>jå`}eå1|/"©®¡y¶4½£Ò8䕇×÷(ÆãƒêÀ©+S>ëKÿ@_`Ý»øÓ«ü¿°ËD YÚÖW~MÀª&KIE«…ûþ²'½Û3C’7_ÇÒgŸâåÜ85¢OÍŠº}7ú€ïò9× å Î['™±B™tyRŸ¬A8üIê×@äuÁ>™÷. æ¼O´o®$H§%fª@²?áôê N ¨}²˜O ³"ȪNÜë…;¢¼L_¯ÒißÌ’pâ3á+{€Jßüu…Ý :”s ­T÷…Òxêßø¦Ä ´¬j«õ§ƒöŒµlý,è>R¼±î9zç„¿,+#,PýxLÛ(O0‚á’²µ‹`Tü¥o¬¢Œ‡ÝÙuÞS}·ÃÂä%`VH[÷úXÁµ©mA`y;>fì*XûS8¨ ­œU¸ŒÌ س—ìQ:Üé³»e)Á±@:N?à8½Ú¶þâB¸P²È>±eW}ݸ†$¢O†FGUqÄ\²üP9æJ)mM]lð·¥rmùü%l8®ÿôaç.¡ºq£?ð¿ú3ùÄÿš?DýXßH÷½qï‡(¡=¾ÿ5ó‡Ÿ4‡ø×>· |î³úž'"Þ‚Ð÷Dºs䇸C܇õ97úè#ÞIDnÜà6ò{”õ¹î…¿} ¼?Ò‡Ï$wpFK¬c³½ÖaѦؼòÐE£“îØB“VGŠ-Y~˜è™dÄVÙRâŰßò”›¨,Îcëa^áâ%ÜÄ9Š Ë³|Z[üûÁ{Æáw€r{ø·ûÉI@õ3½ZC¯hQÉ \w:uyuW`š &q%&Û eÆ­òÀ`ºç³•;°>d®È²òÂýö9ÓX—%ËW¦S/C>;ÑOù³q½SqÐlÄ÷ Šð‹¯:GýÑsWJ™N[¸Æ¡Å,˜i Ú’ò¤Þ=½ÿdÞg;×Ä<ÝòBêŸü«è3¦Í××Êadq»…ÜA¥Y¨ï®4¨mö–?Ñ ê¯G”¢b‰þmhß…GïŒzèÙ)³†“TvBø4‰Z!DÉÿN¯cçK/Úw3€1ßö· `ršww’ ˜ Èe¨[.€ùGúiºI`Yp÷'UX_øÜDCà ¶å‰l©çÀ~‰Ÿ¤£·`½½8|M ÎéÚ;,èÁ5Ò[•]–ÜyÞxWËHP‘­“0þù¡øóþ燌ðcøÿ¿Nþ¿×Ù ¯³ÞCû¿{e•t­¢ …®)µTµÑ µª¶‚š¦º%UAªèB ]¨£ t¡‰.´Ð…6ºÐ!\©(ãW*ø•*~¥†_©ãWø•&~¥…_iãWø3Tñg¨âÏPÅŸ¡Š?C†*þ Uüªø3Tñg¨âÏPß¡†?C †þ 5üjÿYoUE÷=~_RRº»¥»álRR¤»TPD¤T1@ì"QTiADºéîåþy¾À•÷÷²kï;wÎ|fÎÜ3gŽ!‹cÈâ²8†,Ž!‡cÈár8†Ž!‡cÈár8†Ž!‡cÈáò8†<Ž!cÈãò8†<Ž!cÈãò8†<Ž¡€c(à 8†Ž¡€c(à 8†Ž¡€c(àŠ8†"Ž¡ˆc(âŠ8†"Ž¡ˆc(âŠ8†"Ž¡„c(áJ8†Ž¡„c(áJ8†Ž¡„c(áÊ8†2Ž¡Œc(ãÊ8†2Ž¡Œc(ãÊ8†ò.ƒ@ZJêŸ*ýO•ù§ÊþSåþ©òÿT…ªâ?UéŸúí¿îgáé#^î;:ñq÷O{÷ :ïqòìî.DG½Ž;µ³‡æî1¤•qmâ?]z.³G—Ý£ËíÑå÷è {tÅ=ºÒ]ùŸ.¿‡+¿‡+¿‡+¿‡+¿‡+¿‡+¿‡+¿‡+¿‡+¿‡«°‡«°‡«°‡«°‡«°‡«°‡«°‡«°‡«°‡«°‡«¸‡«¸‡«¸‡«¸‡«¸‡«¸‡«¸‡«¸‡«¸‡«¸‡«´‡«´‡«´‡«´‡«´‡«´‡«´‡«´‡«´‡«´‡«¼‡«¼‡«¼‡«¼‡«¼‡«¼‡«¼‡«¼‡«¼‡‹k‰„2R¸¦øŸ.½G—Ù£ËîÑ嶇yÂÿ†ùÿÚáÎwGNì¶‚3îvKø9¶»™0Ðýˆÿvi¢ÿJïÿo›fûfÿ¿Rƒzç‹ÿíO¼wRñ_9?Ìÿ)§Á¾#Õvd@Ó¶ ß‘÷¸¶åÃÏÏomËøí‰kÁÿ“I;ß¿ý°ý9mg¿ôcÛŸ?íìÿùöÿN]#c§\æñ¢ÿɬò_5·¿ÏÙ9Î7¶íïóvŽ—¿¼ý}áÎq‹<\ÿ'‹wŽ_Ü÷?Y²Ãù®Õ»-wx¥¹ÿ'Ëv¸?8¶ËÿØá—Gl—ÿ¹s?W¶ËWìœÏ/÷íò•;çUÙ°]¾jçüªa»|õÎyÖ¤m—¯Ý9ß:Jêm¹sÞuß?þOÖïœýEÓm¹s R³ÛrçzFîþO6î\WãK©m¹s}M–uÛrç:›É}·åÎõ6—ló[v®»åÂ6¿eçú[%·ù­;õÐ:´ÍoÛ©¶çÛü¶zi·Øæ·ïÔOé6¿c§ž:жù;õÕy~›ß¹So]âÛü®úëÜæwïÔc÷Óm~÷N}ö˜mó{vêõ7É6ÿ÷Nýþ.Øæ÷îÔs¯Ê×m¹Sß½I¶åN½÷ÎÙþOöíÔŸqʶܹ}É«Ûrç~ôèoËûÒïüx[îÜŸþœÑm¹sŸ”¶åÎýð½‰ÙþÛÜùÿï¡@‹8Åa±ÃæXìH!;&†ÅŽ?Åb'÷a±Sg±Øé>,vö0;—‡Å.c±‹±ØeB,vÅ‹]íÁb×°Ø,ö¯ »³ÀóÞ¦;·šÿÖ9en=&ða0$[çEúƒ!÷Ä`(Ú¶º] †ê3C³U´Q[ÝÅ:ÃpƒaÜúA0ka0,é ÛV/Á¾uƒ8V0®­‡ îz †a0|iŒ #ŽÁ-`0ÂG0‘ FLƒߪGIF F*ƒ‘Þª+Y' F®ƒùoäR|ƒÁ(Ób0*[÷Lu ƒQ·Ç`PùvŸµo«Ï"$Àl÷Y[¤Øzú¯ãØÞGlkÁVæ_ w do3>§ÇÛ25U£ }ܱ0.ÂÈ“ sÛ9}pÎc»Nh»Éòv¡vMw·vãvTwÍv_î¾\Ø5¶ÜÝ7Ãî"îeBiÖREöÄ?çÕÝ ;‹\»Îr8c”¤î8çöçtœqÊîKŧœ±ÉNP¤Þow¯ HD}¿<ø{Ðç¼¾ûrb°ÏVÍæÒ4¬íÇó÷ë[œ³ÝØŒ”IÕY4án¬jà‘ó/XÒ®ÓÎÁa5òE“Æ”»ÉèôŸ›LÙàœØqÁvØ3_¾›šü ˜´%a„Ñ|–ó4>jýÿ’:Øjž8ºÊ‹ rF*„צ|´È5G7ß›øâ’èRu–úxñäñ+¯Â6\IF xè–ÕR`B®R¨b€It%®˜I˜×Þ¥†Tºë|0ÅzJp°ßN©?Q\Wˆ¨Rµ€‡êªþ>à EjäžÀÏØüCLA`œúH  ëµ›ïæt™C9ïAìš–X&->H °ô™I–ƒ4Fñ ã ÈDü":r2çz®‚ü†ð%å XP¶=ô7'”—ß[} ÿj’yãTçÅFÏz¹‚ÆÝzžu—@s|büBÚYжiïõ‘ó–Ûv>þ|pÐ]ØççÐÇŸPÌË&ƒéXƒA0:q§§ª›‰¡§4ŒÁdÃâÝS"0=™èÁésÌìn|µÀ€¹±pN…Ë(Xh´8\K+%.°mø˜Ö\ËH¼è3ØÐ˜Ø®Ö €Íf­äÓéu°ÓÑþ våÎ×»Z'Àþ•”lÙ!8x™Ù”æG±ÚǼ­ àø›QîýëàTÐ?°D–7ÏÞphùž–…síòõçAtùð' 6a ÎäЦï«Å Ù5þÝ þµ»H¼w7iín±Ý—}ÿ×y'ÿïKÁã³Ýãà’-ì:»î8Å⌗wŒÄvƒU7¼„'YÔÿ¿ H»‹ÍÏ&Éeîðá’ôí.zí“ì^ßîâõ®‘ón0 ÜËÿÝäf;Æ’»‹a¸¤f;NEC¢l¯¤òËþ½9O’0›ŒÆ9mŒ¿‡¢‰qå§/z*þ?ÿ_£“]#Êöº´”2ùÂþ ZùÌo±nù­=;Š·Â“ýÏ©h; 6`ô¥µÏþ ²«ÄÉÝê°„£G¼pñqÙÂ¥o—Þã·É)ƒ<›<~e({Oúµ§@ͤh•c41"±Œ5B¸Åmš¦“\ êÀ$xP÷M =°\ÚÈþò4 ØÙ­äœ9OYOgrP4ðH.¤ŒÞ5ËûúÀÿ‡•‚µá<IÿäóŠ áº×ðØÚA %u9­$3]ìF»?H3Ì\Xv½2…Gg•¸A.òÒ×±ìP¸,€,)®ƒÒcê“M/eA¥ûMÀË»# n2ïáÜ0–QSšÖ—m$B)@«²Kkýt,5#œ@wF“·ìa&è=¯|·Æ& ŽºÏ®è›‘xrísŽA8Dµœaá ‡ 8¸¸¾¤€i¨ðÊ[]0»=0dáÔ æ/OýðÈ‹¬£ÚßûòÁ²ù|šþ °Ú~Ù÷l_?J)» ¶6÷ÿ¬ƒ]”‹—¨Ì#°ÿ±ìö)æ8lÆJaÀIû¤Í¡7ûÁùú¾gú†zàR¦Âû\¯½‡cX>ˆéúy/ýýe8ìzŸ¥YÅ­MÞÀá[„Wa²‰ ¦¼Ûw*v“÷íŽS»‹b»/Çw_–ï‡üß`~¸¤ñ;F»‹e»‹\»‹Z»I¾vƒ¦ìÿÚM*‹ Ö¾3Î,|K¸öÃÀ-^¢Ê|jû -qNZ§Ð´¢¥\Ïók=á¸Eß]cþUÅ[kÔ–opÆëÜ=î²²\hCÖŠ÷˜súã(räATÜnò’Î¥±wÍÄ=pNt»ÉLð?Øn:Dx㌉ ð/L¿¶bVŒ©û‹°/ÆŽ©ë,.ù;ÙLxu»ŽPT:ŒQOHÿ Îé_iÖK^»›Œî²s ·,ÐGyǦä2àŒ¯ßÒu»6ÓçÂû %žÀœƒ¡!øX¾â{Ј…kzA®§Ï°½i͹¯ìqJ— GÂ-Í?R·€3+P=ÓиzH÷K¯wÔ‹ÄïôïGi?FY¡Ì“÷s²±Wð]µ§û¦ ü^Ôf%‘r àKò2e Ÿâ]±t˜¡yê(F_ ¾üÅ&>IDÕžàÇ$Û‚¸Ðu¼u£ÉÆ—¹±Éôaü =|'¡Áð'Èr*=’a9§wçš¼+@þ•e‰Œ(üNq6%ä%¶'™x&P¶Uîÿ¨ TžÓä[yÄê4áíË>e n5ûáG¿  Ž¾5³JÐЬuq¡ëMQjo!§Ç %.k~yÜ´žÊã_©x¢Cx ;UQØ.uô.Ì8̓B˜AƒŸ  †˜_£ü‡Ì.©‘•Âá‚ÊçÍ™WÀLµÙr²\ÌKn†±€eà”ö%‚æ‹ßïzvÞöº»cû6ÅÁi$Ì'±_\c‚ÖztÈáèÙðèÞ‚qpKNó¾p^ ¢ßí°Úw\‡êÿ§4 È˜&Ìa`¦£ãI}sg´;tœ‹ÀVqç”ÚF|•:f6ç̶|awÑz×87Ý ¶“Lh7èûîxµ;Ü º°»85ê~ÍÔ£Öí_r½–{ÝONÿú—\©>#%ƒô_ÌããÝ`(»‹ÌóÞ™ –fßÑüd€Gï[´ð0áÃýµi´x8éñ¥@?\{Ý]„^Ù—óÑ-³ ­òš±ËúˆÖ\©uŸºœBëI®³ôqÉþz·–ÑÅÏ¢Í~i²KÙÿ’ÐRÆ ž>'õ/™û®1óN$¢ñ\<]ÿ‚æ–ýb×8d^WWî… ÃkÖÏ@9zâ¬B‘ Påg~ÜWð¨kÏ$:4åC”9B®@ûSIM‹ûÐU}`?rxèk_úY©Þ†jõ/ïocñ7f†ƒÀ”lÿ}0°˜CK<‡‹½€Åñ•îY¶À*ŸŸûmÌ Øè[µ}é.ã‚t²/º8PŸÄQîŽwÜŒÇ/Qšê÷GÍ€ îaÀcÓ]‹ S^:Öxo¿VàÜT\CÀ×4Ü™€þb3=ÅýÿŒ,-Sr ¿À”Ñ?'Ú‰A„¤·9¿= DýøQ`P"ˆ ý¶Ç€„çdÖ}‚:"ôd’ò+©~oß—œ@ºQ/êr#È´•—84ì²caðÕk /qs9²I—ÄLqÝð¢K‚/(ÇÆP¸+ùƒªõ‘:ÿËÚ ®M@HÕB÷/yK‹ûU h±„…ÔÈöÅĈLÉJÐùsÓV´æ(L¸sDÄþèûÜpÉÎáë§UÜÁ8ZàBÊ 0•Ák'9¦ù_{ ÃÀ\pŠûÙ+°È—xÄVÇ#ù-€ÍAïùïÁÎÖÔ.©F‡©4» ÁYúÞ£KIÎÝBLü‰â82å2<©ë< ¹…M@¦Ø‚å̧Cã¾oñ‚ÅRP7QH^ÑçtÔô;œªõ ªël/A5'ª9$t)P¥sf¬Úé*TÁ—/´Tz •;ß<’âè‚~bœ.ÞšoRëÏ,¿E?W¹}.œF?Z…ìNNÊ¡ óZâÙ·Š¨ÊjâNüÙ\T»ÊúŠ˜“Õ}šóƒ/5šÌ ˜ÒU ¦!›5ÄñµØžÕ«Ã€Z+ŸNÎEíÄ-OåM„Qž^ÂmCNÔ‘hk@±ñ uÌ,z“–‰:ÙÅZLnµsý£2¬æ ¨süû¾›Ã¨ë×™ó«Y¨»„$ßylõ¬Üºæ§´Žz Û-Z Pïß5lÐ…­çÄE™'ÙC•h€ŸÿDß2TÖ® êBCN³ºçº¢îA»›§ƒÑheÀ#¦¿¶hüsÆ¥¢„D4Ù,•³´f¤Ã©9š$Ðq#SCóNxc"åha=‘­>-}‰\~ÃÇ‚Vâxi÷üFké_¾ù£±i×Gô…hÓÄóu«½bÎ7ßÌŸº xb1šØ;@àë.sd}ˆx˜«NEhñòðÕît ™±ýN[ä¤?>ÚÞ'ʃnÜë9@¿ØEÿhh‚Ë_*Úã%ÙY®7®Ùe!\Î𜜦}÷^+?þÌõÒ©ñŽ"ÀZúƒ-R±ØÇ“ç2V€K·Ó‘ÂO¸g¯‡u^Ë^—Ñ#´È€oyÎlTFr­³:ë„A(½0Ï3‰„;Šx—ALU‰Tè7H4±DÙ°t€TýßG.g$AÆúeS¦7È®4O‡øØ‚|¶¤gD/(> Ò»Ût”ã½è6\AµQL ó  ÌÜd^ïÐp!zÈzõ;hæš­/Ñ‚¶¡*žÓ!ÐIàYÑ1= %Th"%ßÞϬä¬q#0ð<ïpŒ Ýl 1ã©‹K©*`Rp‹ -¦&l¢?YÁLƒ‘ø ˜+œ£_Ì” ÉèÑN °¹¥Íp•¬—§“±`Ígn?6œ´"º•`ˈḻöìH7 р݊Ùð¾ýWÀ¾G*âNþ8佺ÉÌ&Ž÷Ⱥ/—ƒ“½SÑã¡xp6½ïÓÊ7¾^c›Ü°´hœz> DnVŒØŸ=D‘xÅÐÈ9Sâ1…NÔFIÀ ƒê“6ñ $P•ÕX[ú)˜"._+ŒÊ˜†¾Ô… ’®û6âs¨hºöÜÈNTø‚Pyèk*¨³óP? ò}»Žs/ã¶N%¯ÝgD%ý+.þ±ïpÇ)Z²æxŠ*Þh?ÏvC•Ožž&ÜÏŽª³GOó¢ÚO‘i™ Ĩ®šg&Yà3ªOÊÚ¯ñ5„WF[æf¢F‰×q¡Æ–`…f2 Ôr8ÖƒŠ 5<¬»NZ„}6M‡P+ý#ªCm´§»U3òP»ÓÍÀӨúu?G‰-êüÍs¿u“;ŽXüƒz^‰È½µ@½¬õå­¨6±Ñ¼œõ“×>I-D’•ú‡BjРO·°çS4ÔŸeXû×Óys¿¸µ¨œ€&˜Þâup )µ”CÙ~¹hæ^RêIŸhŽø]¸ìà#4ïhörtÜ-T•eðÞÅCK®‹òNDAh…îÛ ÃÕh´º˜[k͹€6ö“q飿º—0Zt[V«öð4ì~·VÓ~éEj1{ <=ö7=ìk‡ ¾=Í$º¾óŒ–:@vÁU&™¾( ûMÿÕÕœÒþ = ‘þB[MÇ ´UúÑéÚ@ÿÄü´zÆ-`|=ÓCû!0/ÎP_®¶[Êúw]g€óô›ù©;‰ÀÝ8có<± x_RƒÕÊ9àlApNU¿gÌë åƒ0ß´„tæcÍI Œjz7æ'Ü]Ù@êgÐæ_®q9U5§Ch rÒ¿‰˜%T@½JÙQë (I­1ï÷Pñ9¥ÉftÔÚÞ,p\°8ÕãÅê¯ štßø)®V¸¢Î*æèP„(Kkî} Æ5jìуm`òóXÄ·(0=¹ú:0»Ì.88ñåÌy4KµÝw°x÷r¿'Í)°¬+ß|: VIgdÌØHRa§lÝn¹½d»¸ÍÏ:¶Ú_š[Ú·~päVÔùhl Nn“ý„·ªÀ9ͺ÷pu¸,öeFØ‚kÜ5cÛKiã<†"ŽÂñ‚K’ø÷A“7WŒö°Ó!7¯¶÷h-U1<þæS4Ò=IhgŒº»î <ó Em˜´cÂ5QSêý§E)¨n„úx*ª3¹’ŠÿÕXs>“ùU¿z™ø©Õž5t´%EuÞƒêº- ¨áäaVrl0jv#¿™~ĵ6p<ápC=ê¿§2®£¾Ôº1¾P4è,»‘~ó Ñoýv+ï:ÍFÍðSMmzèµÉüA³Äh7®z¢¹1Ö{I_£ùß~×el÷¡…A*œ³¡hQÒç­W¨Zü~Õå¹Ê´ô;JÔ² -›ús¸£ÁèJ¦)]´JÍ»ò* ­bÏd¦àõ¡uò Ÿžê£ qÅB%ÎôÇy]½t¿úûážúêt„e3(z,ü0‡_„R濵TMc£®¿>õt¢aa½*Ьýø|èH‚éÏ =›|§ÁÖ8&ѯø^_µoXŸúØLÖé‡·Æ æã-MƒæºÀâóø¢Ã4°úN[—è6_‹÷ç¦K€ýÌMþ81"à¸"5úH² 8Ÿ`6H9\eɹ_HûZr[È4ðˆGÓ‹oÍ“x õ,wo Á•$emà |{óÆÝ§Àïí|¿œûVüò@ðZ[o —5JÏÊïaËR ¥%οqÇG ÄúÃE+„€¨Ñ¨Ê7]›àê&µ?tÔæÖ—ð„+æª$Ô†–¾wñ¾Û¨CƒWÞý1+¸‡:ïSkfÐèm#$Œ–¨Ñ²—¥øŽ ªãµYÌ\E•’ÏËÌXŸ¢ò&d¦üºýPÒ‘81U‰~”…ýpU¯“É]B5§‹=°Ú¨1èÈ'Ïרƒ‘ë»§ÿ'ÔûÌþ¼¤I ¤yïÒ`_‚†Î;TÎBcRÐ{l=M¾ðà3šì‰‰OjZEÓÎ%«ù¡¤h&›»O㽚=õ™îÚ54[uÃI7ÇÍqY(r¬¢¹à‰ÃWRŠÐÜß!+Q’j4ÿÚâT WZ°å¨²|^ˆy<mè‹£%̆‘¶K#ZúLtN`­ÐÔL1ÄU£Uƒj%ERc´yÇ4Lá=Zï_KÒÏñ@ OšëK¡¿¹®î&aù«a÷Wyp0çØ½‡Z*ïX@ï™IÀ?Ãò©Ûè ¤VËR¨mÍ7¯|Mâ˺µž ¥Tséx0¾ë‰è´¼äŽqNº§*’Á]§p™ ö÷_ÿ ÔNþ>Ÿk€Æ|ZàNÕ= 5+o9d tÖþ%žM@o?–ñ[lðºœFs'B&.7`R}õ¸C–˜Ù}rÌÎóºÁÏЯVÀÒ¶f>Ÿ ¬ùoϼ ¶w•íBÞ{ROhÎúàø0Yp|þ>p–žv±®±‹ø±³À}ç*Åø ZàQxûíVxf/imÍo‘˜³¿ð½§\¥Í¼ü9l…‡@ /-S„?„x„ˆM˜=à€ŸþBi`W…{¼=> ¢³r6ì| öü§L·kH°üØ7³’5,· @*KÈo¤‚ŒY‚Aæ•Ï©ßÝ×A6KÜüxÈÍÚË­oµWó¨‚éD[PQ:f[oÊi¦'Šãü@52·£ ÔjS”DôƒÕ³ '×@ÓºúBk¯hå7GW'/‚ŽN›ïºøÐ )T{^z™o¦ÒÒÀàõi}­}?Á¨ hâõ"˜ˆ­ˆSµK¦YV`ýJíG0˜¯«ùQËPfúy°–= ·¯lñæç¿þp{ÂÍ‚6 pDŸOe¹€sæówã!jª-ÄèRáÖs£ž«“ŠHjÿj};î$?§µÝ²Ñh{Ô¨V¼Ê¡„:DÙôRDú´W/GDîÑûÚq£ŽÕÿg!¨³ša69Du}^«K{쇺ÛsX—ñÑo¾ž„WQï1ðXK•C}VDó ¨ßÙg2œK Ü“tm¤Eƒ#'i«óѰOŸÿ[µÃh”yòoŸ„ǰ(vE¡IÞðÕÊ7Ñôe“¼ ìš½ÇmœÇ-Œæñ¬Y°×ï …g¢¿KøÐ’U­é†mZA)ă´ hÍxPÓô Ú¸Û¬ŽþŽhZ]³ÆF+sT½£ ðÒúèKÔ€@ijéþßË@86+}–݈ Üâ$;éªqÉo «}” ùPRžz}æÊƒwF?_¨žQ¿`êZ½Cêzœ@÷%*ÒŸÔO(<ܨÆÑäϬ’Àœ’Z4¹5Ž>¢aû1ì™EÉ´Nà¢ü”,¯$ Üõg꼌Wí~8óðU¿¾é4 /(Èï(€P â«ëÇÅ@øy=­÷YÑEñ¡'yê Ê¢ìD0 R‰ßGS4A†÷쾙㠛÷Ñ] #ògäm»ô$AÑø¥&ßL$(‹ÔGøÒj@u;=¨7xÞ² ñÄÁ_’ä yƒ=ç²ú'К—x.:§³°HdƒîŠaÆUЋ¾Æô‰# B†bJúÀp\U¦Kàø{¸Cý˜„œ˜¸vLeîOãN3ÁW&þ<²`Î]äËMœzxâ`É™ïðõ%XñÒ­½²zÖ‚qã=K¼`#Ò¼C’¶âÒä7oƒë‘~`/¤«'­K R}v7·æ§S_OˆãÒAÞ†¥gàÔüÆøºà}pî>«1žxÂÿ\ê"Pª‹%•".ŽEfÓá¯~ „GÏQ̽‰†©ˆÂsäPKMد;jÊ[ÏcÖ&ßЯ/¬·6†IQ9±iÔ¹xTÊ]¶‘cƒŠFs…£"±wÊI娀ÿúÇûU(ßCý#ŸÖ”oš•ÅEƆÛ^Ä=ÒáôN •¸ÛR!î8?Ž„e^X.@?O*&à³nµsצìSâ¨JlÀë`oª¡z=ÞƒPm-ù²Ê$ªûÙàžÄ‡êCËNß¿‰x´äþŠŠEßãnEÇ38^ÜaCMÂ4Ä'b*PÓ_n›Ñ·PóhÁ|ÙuÔÒat¥=ûj]ßHžæFí µ‹oÆPÇ©'ú­†e¨óÛ­S4‘u¨ûà9Më˨§Õü–X*-êuµ‰#ÒãF}¶ø‘aÞj¨?¿X¢È |y™ñéª/â N¾Ï~ '0§È“£QÇÞÒgéEh\<Û½ÙM:e^±¹Qˆ¦ß›« ÙHƒN=Š54×±uÕD-h]ÆÖƹ£Å*¸ÂþÑ -Ÿ3~öü:Z5¡{žÉ.‰Ö­xnçÓG ?7ž è:¢ÍuOi›À4ò½JÉa|Óflpj!,øßQ˜$¢bÚºGNQ°/§7ÀÝ"HtÖ‹âBûÂД¢3ì¯'챇I ®S¿jË´.‚Ç/f•=wŸCl 0ò=¹êñ› ˜O® z¹Îë2‡cIâà¨;M&›ø¸¯ªI\Tá^­ÊÉW1 À/¹^›Ø‚V¯ŠyÒ á@êÃŽ£‡mAT.n ŒÌ@|¥A³f¸ ¤‚HÂ~›‚ ³ùÚ£ºmÓ4;ø:äóÎȹ]8Š}?ô‘¨àI“?&Vµ£ÇtîÕš&>äkšuë>,V@KãUDwª:h—¶3ƒ®SòjöÍÐ#;×’2wô«ÊЈš8ÆzP%ƒñí௞`õ¶U½ÄkÞYnf(âfÇš÷ùÔfoXqííÔš¬`YÅl:¥àV›ã2mEm`£ä!ÏÛ³rŒdšÆ`—Ññ>˜ˆì—º%Ê€£šÏ’dL8…ûz}óŽç¦$3p¥{_Jƒ\ lìŒ×åá~Ý$Û =pÿþž©Ä¶4”þâ½5ý¬™yåÔGÑê¡)iÔh¸WáXÞ&êúýŠM×µŠÅÀ3E “Šîl_P½ûMê„ Të[æ=,‡ª%ëŠâî)¢ª3¾YC)\¨ZõÙÝÕo¨öÒ£ñ¢tqTÏ×½FÕ€š‚üyúýo 6·µt:Ç{¨k­Ìù5êê;É•F×"£2èѰÁçh,(¼§!ŠM¦öéÒR® ™O¦1UôhŽå†g © š'Ð{y<Í.H¼i`þÔâZ¨ŠÕCûÑâ¥2‰ò»¹hI/u8Í·-3ļà DËóŽ×æH%ÑJ)1{]Zíß0Ä6@k›ºì)ІðÛûèÏ™Ð7×2†ÐßvOרջk£ÚŽæ¬ùßšk€gq¶c”R ðõM/Þ–E@àÌqeX¸ŸœôPb‚?îD‡^¾X}¶k´ª@ª5£6ô È©+ëÒC€b!ˆ¾ž'öOÞÕLKê@±9§/  ]üúÕIhcêH¾¤]Ê_žwFÓ@_üýîtƒ-0t¯×Ù`¶ÚÉzÿ­Í>`¦ÿ8Ÿ²X„$ngaíU†•òÖT°©†Ê!`?ØZ5G9vmwd$ñ3¨e¾‡ø6p½×c±ë•îóŸJÎÌlÍ7ùöளž£Ê,à}áòí Ñ0ðy{fm¸Ï¿Ezj‡dµLU®à©Z/úgAè‹ñÙŠµ.müTZ‰"uý‘Šâ öâ¦ÕcüQ¸+–Yo+RæFØoœi }2É@®yd"•|\Ŷæ›Ä©q|@nïM­(p»ù¯2%¢­í™åCPº×OøÌ'”;ÄßM{[€*-gëε*ñs©U@Ñâ 1Ÿ¶žW<ÄmæAÓn,©ŠùhÝ®šÈñìíÎfçè¨ Ð5¾™Fº‡ÎŸ ŠýäçŸ&t_–¥ª¡Å‡J:^IwÑÀ#¥%#Ý­ù^\Ùçw¨…ï;^j¸ý±$Ÿ+Õl¯˜Ç~A¿Ø0†Ò/~¢­r]Rz?P™*h³j ²7²c´.-¨|®Âæ¬õUTõáœFN+ÑVù¦Aƒ~RÔ0tª]ýæ ­9&÷õWþI\!AC!^^|¯IÐÈ‹¤*µŽ/hì¥Eþ4Áý¹øÔ4Õ!‡AS ¥6×íÑôâË‘aê[Ï‹Ÿ¿>þäf¥˜šj¬Ñl G 6ÍQYjžóx„æ>pÑAó®0bEho´ÝäJ(Zø{¹³í;Z\8hö -ã¯G+²"âhÕ7?ñx6Z+i¥ðC‚£tw§}ÑŸûwŠä¾£Mú÷zÇÞ!lŽä|žo`ÊCIVEدÑü{©!SK)¾©B».Gsµ @T3V~LöY>ºç$êWýÝì~ýëá¼A Ð:Æ` º û‰T´«[ôªe@¦hâ"Pw¯¨÷+šNߨZ íUüÆ– tÃrË'9ý€~´œšS}ú7g™WÉ€±vðtíäI`Êdïˆì»Ìb,€å’âá±?`=”øËê°Ù+,ð…»SÚñý7Ã;žZ©8/ÏÜ ºÁ\IB”TBÖÀí—èè< çR xª»o<%Þgö×ó\h€/,ŽÔà'ðGŸKPè‹ i¿¸Óž 8­x¬Ï(§Lð°R‚ð]ÑØEe™Ì‘ËK1‹Š¡#­q ^÷Nº‹$ßžn_©>{¦}¯@Ú/ýÌÍL 9ÎÄplXdÏýjîâ¹ô[ù=Üš À#mùLã (6 ›cA¹”ò;'‰¨Nåèø(t(”á| Ÿy·[~A³Unù¹g=hxVÕž ºT”6 껂RýŸy·7Á°ÝBªÊ. 1œéêzÿY÷é΃áïWƒ£¡`îö¸×ðXTGß[.+ üEŸF!°^êgNÛü€ù÷íš`ÿé=?©¥8¶_ù)>u\”“cD.@QêÆÐ8H4à½öñ«¡=VÒ°ÏžèÆÛu4¢å_3Žï:¸¤‰fسPÃì8[i𪩟þ>œ]‹*VÿÑð¾€*8扚™·æ©Vú9›¨ÌíÕùóU¨4"Ç\G,•*YL.¤ßE¥×Ö?˜m='J^8Ne‰Êc"†RÓ,Я§ð««¨Fl¦Šš1Õµýö ½–ŽŽy\w¬¼Ž§°J,ÑÔ¨Ù Ìú×Y6Ôò†QÕªø"j­yþ²ÓÚµ½‹qŠ'GíR¢É Ìu¨Ý]q“s*µŸYýš,qµ_Ëé~ØiŒ:öYsv`Gµ_OZ›d¡ÎÄñõƒXÔõm4ã"%)ê^Wg;ª¿Š~ß#}=óU õf¶£qWÔ—Ýýåµã#Ô?Æ:ŸqÅ *»“ËçW£¡‚žÈ_ähäø1·Tn4v(›Î€&|ºU«.Ó£© fÕGËhVAÝŠ^ ͽI5lÒìA êªì玡ÅÚwfÑ/Ñr—Püõƒîhuܬz¨mð Ýð ý ¦M5¾Íʇzù?y/àÑÑëã@€ðµC„ŸË^úÄ7>«Ž±ÌÉÕÕ†ÿ {r.E(šÎad•„¹[¦{9P7gÚ=[&Z¡Ö’6®/@“ùgú˜0°9•ecAÙ;‘gï€9X=µ÷r°ž4WJ³&ököI2ƺÀÙ|cÞ +ÜŸ±áì^µw\ À÷Ú¦ 1Vìލšè€Vúµ—‚* ìÜ¥þ&ˆ¾ç+¡ q æ¡£Óø u.ByŽ*¤§ U¤„@6$¢³ç9ÈKªê‰Ðl±J6Ù-PZ»ã7øTY}$ÄŽ‚º_ŠU…$Àêõã y¯AÓjkV~´ mgÒ÷ƒŽò!™iJ<Ð-r°:cNz–‡n¼3­ýu϶¯.`øîn×@E<û:¦Ôe‰ÎBikÀY0e"û¨f”&1sròJ#—R° ÄèÜ4KÚ¹³Ž”`Å!Mø0¬…—5ïØoÍKŸTõ <[#k—©¬ÁÎdpö@_ØÛ(¥8%‚ƒ£´ŸE=8:©ŽfÈ€“]ˆ‡(¡8›N¶ùœ¸.^y3î¾û \ãüs–3®`qç·lL©,½/°È¬eî¹ß Ch0%÷#¿j4j¾|Ÿ_Õç[ÖÜE¶¿E¸0¨lhí°(;3ú.œ¤• ¨hôÏ;Ñ vTмŠ)è8ÊK#‚>Y”·O¸Gÿ.ú6ðçNë‚n{Á§R–Л?P1_£yWî8ez´aÏjP¹ð§édã:ôs@HÍó'lµ×èYî—¨êt0ùlª!ÄWz)rÕÆë¾Èd‡ê‚ ó Q=ãy—H2ªO?|H153&5R1¡F’Û]›‰á¨±®nжJ5å41j}Žš„¹Ïçö –Âß?–ÍQkÓÏN¹[Q;ÿ‚(üo[ω™ÓÕuoPçT‘ɤ™ê¾ðÞûWCúíÈØîí„z_ˇܒþŠúJTT¨6GQÿš«ú\¯.<¢ý¸Ýc u䥶ª£‘{šÚ¿Ñع³euehâJɇ•¥hª¸s¬ËÍRJÙË; 9»W—>m¢ù¯ Æ—>V Eň]h©e\rRã"ZIöò¤gñBk‰t¾¤Õh£âÛCù1hS’¸zAX0îS.o8¯[zºÙœ.lp˜_"”p1ìÐ_Ø'qxý‹ùš—{…ybÒØmšBØÏ{x¥¾ê>P{—S+¿ šyB–ã†[ó`“s¼À Ínxñ10õ8ýÔaÖcx¾‹C+À!+ǘ] Ü"·ˆlêNÏÈ'c’µÀWW:/B #‘wþ(Ãi¿?Æ"é/]g‹7AüħÇJ§½AJü9ÙË,.mÇ›xþ d¯$¾8ÇFòެ’^ÌAñ诳×ö[ƒòýåú¢QPä~xã _ ‰’Ò ­´AÅu47öÕ}·¤í˘\ŒM KßyìcÈ,6œÄt½ý0"êgYì`ho$gpŒ Ë,îKóYYã k0U|°Éú‚Ì—ús‚¹·ÛþתÝ`ñ!1Ë÷'X–{$­/`ùpUÊl`¸.Ȥl¯]OH¼våÌ&"Þ.à°ŸÙ-qëºmSÞ|è˜§ä… pþ³äéÜ®b Ò¹“àÚ%v\ó|Ü?¦üÄLG܉­3¬¶úôòAÉ!ŽÀb«*œ1¹5/›#ø%Yކú¯«Kø:¿F1?ãtF-¶ååHQ£º« d@ª+5Ãæ>ÓG5:QÞ?ÜRPéµaòÐvT©§•möò$ªüû=ôW²7ªa;a¿ô‰ak¼[ææG÷Û¾1„f VS¨g¨33‹i†´õJkâ¹vüFýI\Y—¿ !ÃhÌécéhÔ|þý³q4ac¯e4º€¦å ðâEæÑ¬Á¢#))š¢ºAvÎÍý4×>À}Í›˜,¿‚æ›ßJô.P£…Ë©uǪ£E4ðÁ*äZ"|ì#Äy-µ‹ôûç å¼—Ö÷^¹¡•Ü´Zk~ñ²¦Z[Ü<Ó–6”(ÑS6ôçQ$&´ÜmR``‰°O½ô%=“„_!a x±uÕ—ÊþkQ¼¾Ék@жÞ%GîDj&äOrqÙa–òv ñJ<{jù,Ɇ–^¼ú(¨¦ÃŒŸÁ~B…§KÌ/š ëÑÆßç@³Ÿ·pEh9K«þ,ùB©‡XlÐ[¾—r< ¼ µL·ž_Þ=]ÊÁL%×TxsOé+ÒÑ2`YÐËfIë6"QÛéä\`g:§9SòÁòÕ8êýÝ>Wìðyl¶%p{}«¯rôæh5úœÀSjªÕ(¼÷¿­ 5ܾc¢C>ª†À¯‘ú'¾ TKÞ€ å×%I~ ôÀÓ#±àã8·8‘gÄ÷Â±Ì fqõxÏqP‘\ ý/95§PÚþS¾ -*ýü8›/ÈhfOFãÕ€ìQ9]â-»m9×tÊä‹J.óx‚Âò½·®¬AI"‹á(ª&ªÌ‰ƒÊ0¸¢ÏµFílw„hHü’” Ú‘^±› ég­ÍZyO¾Úˆr€ßü‚el"è¾ÿòé¶è11µÍãDzÉãÚC0Ò~zøÒýt8äwY«÷ê ®m¾ý6ÂÌäH«ú˜žù‹é¤‰ü`)H9sk¬¬ª¿7É)Þ›GÏDy¬ÊÀ.Ò…Kþ·88dhuâög¶Ÿwdé¨ÀÕ3S%Ò8ެ‘=†b8Vò{Ÿ‚Œ Ü>>ðAÛ޲Éå‚ŠÕ ün ouî¶‹C ±ÇÅÍ£þô²iÊÔþaîÏQËý¨Y†•§9â:ª/ 9õ5}Uò R–}Œ*ˆÕÒr/£²Ñ‚㨩ØTwdmÍG{GKh ч_ž¼†~Í3ŸõìGõN¯÷ÛF-­²§! u›8Q9ŒúÂFVbLÐ Õ‘‹xûíÑ09åW†Üy4Ró«Jj(ÙßúÞ…·ŠÆóñÙä­ IÌõdo‡h4EÓžr.ÅMµXg½™5BÓ!_dÈÐŒa:AÒû3h–û»xáÏ£h¶ðËî)w4w’BŽpÀÍ ž`ÿÒæ×1^Þ‚haŒ"ý!ñ´8_7unZæÐãðuÚV¬&–õ/ Õ„¯”.hV¥6öÚð,dc{SþŒaîE;€6o í{/»Ñ;Ò—Ntžld³Z9#à±?F¥é¥Ê?€ˆíET!S1?Û4^D…@"òì#ß]3 ­ÀŸ¼IùÈ}\‡Žôå?Qû©aÿ|_8»Ò- >‰±Rgœ·²©* =ù'âzÜ+ ó VZ ú€Èü–ÇÖÀp&®‹µî#0º k•½É¦C¿hý€YþeMœ?°pe !²VªC+O€Ä7m*ÂØÉ»šiy›ehÎUZàT<ßõýöpmbºå™܇֕k<Å{Q3¶gxÞA8½?ðÜ|ÇõdøœeXƒ;ÒÿXí¯YU p=K‚1K¿éŸ­¨‡ø$ý—sï°E}tpH(ˆ¤Þä¦uâ1²ÊGÃ&> ´þ&ä§2H’ÆRFüV)§¿5” ­1à[fÚ 2b#3ä +_ˆ‰ð9_2žõA~¦i\¯‡ ‹éå$ñ@¹³’ó¢¯7¨)»mž~°5¡jøÎ)´x4êô‡ƒ–Íó³®Kî C|.¥Šüè¶k©_V½ª³QÎÊI`ÐÙHT`JÆû•Y¬ß…‰×©úæ`zaå×/‡K`V­5s„õXHZ‹˸NZ:¯6°æjœô;6ù¡ÒŒ¢ÀîÂú|Â$?88;ÜRy$ Ngø—n‚Kþç{xî‚p›ãr@°“Øç:&L$¿ÑÜ.)ÚÂ% š`¯œyŒECïl2_NÙ ¶4’Ôp³T¿öœ'KŽ U»Ðb¿Ü^D¿MNU„Š£ò¿wQüf*[‰·cg;€JOø »]CßcT¥hø£ïÆKK· Ñ÷WÖºi,_P©CÜäÑýYèGÆhó¾÷½¨âçÃi!&T­gôíª;‰÷먌:ªŸJÌ:‚ýÍ[oT@M%”Ýœ¨…¦›VÆ‹Ze®ÆKØGm˜%ãÔðÀÚ0Ùµ½j{í­¶%{æêjs·äP†×ÔÎPÆ­_”Úìû‰ØåQGlçŠÞÒmÔù’Ùhó< êª{V[øÐõ8ÄžŸdç@½Œio®y¥¢>£K%OFP¿&ñõá`4y„àEa(¢ì|=Þ>‰†3Ú+ÏÜ%D£OÇ4M]·úÙ+£Eñhо/W{ŠÍd}Ém{L†æ 6ƒR°bhþ·¡ÝþòÃhñ±5Qré´˜¬ÆñZ½Qhqµ­–ÌÞµVF™°ŒÅöã”IgàibÎÕ{øØ %üw@ôr#L#ˆM™Y¦kÕDßHЕdG\Êìk€âÙÅw×ÁþÅÞý–@]›ð´IhIŽê{`·æ›þ=Þ§Jè×»ÎQ=Ƨ.9Ùæû;Ìþ ÀªúXOmHØMߘ?öÎÎêá?8€;ÆÜ=Öé,ð, × º|¾ ²%]NÎ:EiBBîÙà Ì=?G¢D³¼¼r<Äû˜ïy´1€”ži¤rw>Hçr㩳كîæQÊ Ü¥ø”Cvv@¥€ú­Î Nã@Ì«~àA1¡›,hòºR3 Z·C ³[g@‡°¿®dÂtoZ9qLôX™WÞ\§ý‹Ê‚0ôŸ8] ÆÒyD‰ÇÀ„D˜ç»ç8¼B8v8ñ˜.|ê9fK6‚™ûÁ|ãëÝù>`¹Oà6Öb¬˜M¨¢ñÓÀZÜèÑ]áû`£ãôR\ëØ:Ü9ÃÊv~éW*z%Áþ†iØgypx&=ÅYŽï#Xè®*S~ÞûP:mp®’s³O9.ˆ^ôª]†°ëþŽ’ïgÁB$4™ÄÓ5”ný&|ªª–êM4 åKXœ?‚šÌƒì‚D5²ð±o·Ú]â/-®eTZß+Pr•ȄߊČ£"ÒSŠ?¡|ÓSßÞ"ôM6ìÅäÖï<7ÉÔ鉉Ê=¯uòn—n{>Ï×+FßèP‘ަq[ð0î8¥b¯¬ Þ’ ²‰ñ7iÂPù©×œ¦¨‚Yå̺ÃúUÙzBñÐT¥Ôç~esk<þÈx ß ÕløÛ ð;£:rrM¹‡¨î± "5ñFõÒ¥HîСúQ’Œ:¨áuˆß ÒÔx—îmƒ(>jò$.a(ðAÍ׿a ÃPKLqqá9jíþX •ûµ³õûG0¢ŽäŽ×N£.VKõÔ½rÅ¿):ýî­›a÷D}ä~ŸçÔÏ¢~ÄóSF“ $Gpg^AC,Ù«ï}ÐpY¢ôø³M4šØÝ$^Å‚Æ_Õþ~”ñ M¶/ÇwÞïA3”dß‚ÐlÛAÏ4†r4/>ýD¸X-$´P='æEKJ!jƒ¾hS¦x†1­Î9?åÅR¡ òη÷¡¿Ç=ë6(vÎ=à=x'£z„×nþfègݯ³@˜'Ãâ„«æ2@’³?‚‹ù(4M‰Lòå™g4vŠ@Mi=ÓO4×ûªÌËj€N²*ð‘70pOæˆ1²“õâ~!%`© :ì6” +ûëÝ»¯žÕ^Às[[O¦òðyý»~NÏ;·äÇ€Pö¨Ì÷k­ "ò¬)A)ÄZïNÇ/2d‘6/ož2Hû¹G?È %Þ ¹b‘³ƒš °‘Áû­ç((óçÅÝ ŠÕsÝ£j¥¥ >oMÍNù4ŒÎß LÍL_Æ”¿} -¯¡Ídytªö×¼ ¢†ƒ¸ñlË‹A_]§ßTŠ éZ1¡2Á˜”t–ç 5˜0+Üq¨SSN·Ô ª$0S).­‘sç[:Óbã`iæíw,K¾.ÝëZküÔ\82°Ñ7W*¾tlï6KZÝH»«é¦Öp¨q{)׎gû¬%xœ*Î_VÚ.:éÇm:ÁU§àþmù*p]Ç;#¿¤÷†\Þà¡7ŒÉ‘¨UJ‡è=Y¦õœÛ6çrÐRìɉåù?hp°AB/u$EÞg]1Úú½FròÿŽ›è×ê«Pûón1]¨šÙÊ;)ö0ú5 Opýb£ÏÎýŒ~Np‡Ôr£ª\…J‘Ï›¨ŽºÚ½QêjHKýØß¨EU¿a_Çêð‹[ü2€zÆ4e•O³ ¾J•O9«–hÐè$ïoE]4"â|ɘ14BØ÷4™­é ‰AÓ‹%*i4çѬaÏÍ€ë-äèåÎ{æÙ/ôþyiø2³ú†gÜ:|¯çÑd=p¿ŒîiqÏúgPMXþ‚q2ÖDp޼%…5•ÄÅ¿40Ú`3 6qÐûŧ¯|3:užŠ¦3*@ë_¦chö¯eäm¿õ«õ¶kÏ¡vˆûæ¨ÑU>y¹Mjün΄T…ošþ««±qÐ(;¸"sÀ±­2gg#[Û K-jþtž8ô/ñ§œô•‡GRT²i0öŸCâpÄk˜ˆ{ºôH¦ø“BhÀôƒ¸Î­0³Àÿ†°o æ„U’ãj`^–é© Ìoì–ËŠÂÂS=î‚ß'`ñ §òד°äç§ÄdåËß ÎÀJÈUÿ!²ç°*6éº1/«[±ÿˆ¯ÁÚpI^¸¬[O™x Á&9 Ýa²%ØÒä0/‚íØû!ËEW`gxLJFcvõ–‰”Ÿ/¶eÚ¦ðÏ‘‡ë¿mqØçU)!4ÓF‚»µS qcHø9]"%â%“¾fwdB’Àçò¥ÍöH:~ŒâúÌ0’ŸØÕ«/ÐBŠ}%%qO¤Êô(u½ß„‡ŽVôÜ]ìEš_}çãï#í C¿h„!Ò-…?›¾žˆôëUΓŸâ‘áï÷‹—‰Ä*(e­™Iô¸^xâÎiÖOž$È2uÞTžð&²v˜X¥"[uá$ÍKdÿÀG"tÿ r”¾<^–ò9+\ÿU¨ß¬‘{•@áü—ÈË{E€¸@ù¤U[u‚>"_Çá+ѧ(‘?)G4»PÀº´-¶sUj%Ë…ÔÅ×EnP£°ƒÆÜ½EIrâ¥H@Ѧ€¾~W&g=Ï’æJØ1/ÿ{Š’U?L¿!JKÙeµ•¸¡LIôý¤P”“ïÕ¿Å0Šòt¬§ ’P~ýx@ÐöOTØÌ:$©Vy×äx¹“V¨\³q1U ˆx ÜQ}=éBwïIÔŠ§û´n‚èD,ÒÿÂu岎çòµ¢^/ñšpÌo4xùíýwme4|z^°Q”?ŸÚ8CÁ€¦K³·—s>£¹>/áÈ??´¬ñýz”o­• „V¿²¡Í½<:ú ím¬Ãi2Ñ>LÂ'mlÖz‹OЊ£S¼¤ÊÈ©“袽vŸݨ)—‚úЃ–õ û¿Lôâ®?)Z‚ו-W9~ԣˑi…aSßò4’øˆ$QJtʪOa‚I+Ó¶W zÔF,¹¾@‘…èóiKø¾ ªëÃâ-6÷Ç¡nHæüé8?¨Y<­F5þ_ ïªAuf¨Ñ!;1¨ö¾ŸºpªËûR_Œ…šÄÆÐ’Ú1øÖiXë÷ö44,ý¼ä~Xš½ksÐZÀÏ5cÓ+5iÐv~xüÄÕ½ÞÊ ßØs‡Nt¿ùgüÐ}«–¸¡ ºÓ}ž^ñ¸=ö¿„«·w çó>©§ !ôô_0} S=ß¿IìAÏúkŒ¾¹Ð+¤Ä_Üa½ÿ\ŠgK /þÕ½gìÐÙh"v”òvã¸ÆçaH„×t&f†—öÿ]šƒ‘Àrƒ‘5}°Æ²ÒY cCnI«4`­·‚:ÿ#LZ)hUÎÀôŒUìɦ0Gu­š”D¼^©-gÃ2ÁÒ÷¤šIXY²y}és$¬]q¾÷ž] 6ÔzþÛ„-V‘îÚ7s°#,5M'pvÝ<ùRXJáï»÷¶#[°oXpøâ?$¤"ʸIe€Dw’P„TI ½n¶7 Û¥í¸HA÷ß…Ù:¤×>Oš}ùÄÊ’t"Ò||g™R:‚tgç9!ýÌÂßÃ#ÈhCökVØ™zôÃyØ/#˹ßEœsMÖäo–r¶ÿ¦ƒ\œ)ÏB'Òç˜RÒÛ›LȦ[(j}¿öÔMi®´®¬Øb@¡?ÁÔÍž(Ò•eÜ–ç„bãÄcY78P’—6råS:J§²iŸ®C9v–ÁØW (¿Ù`üßQT¤ÆFÍ¢ Tzn$Ø*ÇÉ¥‡tPí¨?µN‹jx ¹ì¡Va‡ïÒ»ˆ ÿrŠ.T£Îê±Ö¨çü&×” õ[#NßoLÃ#6Q]ÃÂÑðׄV™Çj}%rCS-¹uúM4[çôJÜ–CóïÑo=*ÐòëÆ™4»'hÕ¯)Bуֿž\yBå„6s!ÜMk©hûÇ9÷IÆ*Ú3qÄp»ƒ\öâLM:ZóŸ‰ô D§Ðš®B3èœy8p8]ÊxßÎ4 k¿üdél!ºýë5ñ7ôàß^`AO£‹+GfýÐëÈsé¸HŒk8Åudæ2Ú|ã¸DWˆBíE•Ô}:H0ñé³àcvöÔѦp¾ß)æÐýŒP®êžÓZ€š\Ó¾P¥y¼õlg:TØ'4Ü€OÍÑKž=,Pþ‰Œ)ƒrÊ8—ØÙ~„rª$}“oIÿ›ºýzœqÑ *=nšš¬ýo)áÑXi¨-t{7¦ uR%Ôm±>Psr»“=å ÿ+¹>/§ éTåí%ÊDøÞò×¢çM44og1+!ƒ{.Kt—(á'¹­drS´² •--´ÖÇ%KÕA[„ü–¨Q8´ûV:E@‡ü„oq5tšä“RJ©A—ÍÇ|3«ÌƒœŽØ ͇žfš-C èÓrdýýŸ|5ÈïLROI‡á *ü!~0b÷B²=ŠFcÇmBs–al¢ñbLØ')ræÀä.IÀ.˜n7»]f+‡èT~Þ䵉,)˜¼'<„°â°è¬w!VŸæMv·Ý†u~'×w$°‘ÁV}3é8l¹ð´3NÀ^âëÚ°]ó?7œˆàïò‚æØg%¬PÙµF‚ÍÙÑÙKHwHÿ¯’¨WµxË #«,±õ¤\·^­ÕG*?¢•ƺ«x¨vJ¢rriK8ZP!=ë×øa e¦—Þ«}D¦¬t‡Ãï2‘åGhóe*d×j|踇œ;ÕV¤¿v—èÈqáß{È/·· Ž‚”*#ÃC#(Lí+ýΠE!uxƒ+ÅsÛ¾u@)£à#Ô‘•(+ì&#§I†òÜ“ w•Pá{€r;ù*= ¾Ä!wÀ£Z7 N£Zil  Æþ#õÕaÔ»´²À‡:¥ » N¨{¦×-åy1ê-ƬÒÞ1Eƒ«zêOÂDÑPf“¥°h6ãJH^ì ÉŸæÏó'ЬÞâú?‰4ŠšìáC+REÇO±Æh-¢®uLÂm, .ªqï íå›êD­ÂhWAi4!§$ÉÿšiŸ¡£Õ—ò—€NYŸþHŒ óâÝ[1:“èªç÷Ð]Ýns½9q ÝçþáªCOó»§tøÔp«ð5z³v_–mÅÛþqëŸÿm¢ïõsQ®I¨±ýCGEYþS]ÚÝ€ ÕÖÃ,¡0>Í»8µ }iÉÔ­jÐYx™\àá%hk5&®•ê…–Ù ÃÜl#=É}:¡f ?OËžà:蹩âÉ¡1–á‹ÚV|w&Óô,™‚–Û¶—¯¤@[UîâżCÐ%ìËÁ}Ò7tw½ˆa(˜À÷PŒ,ßÑÿºãG ŽXS ôN“Å@ ÌÕúúxæ5Ô‡Ìi`ÑþÑ¢Ý , ¹2~·å-›>YܵmòV¦Û÷„_‚Õ˜pÆL;X“¶×~þdÖÆž0ªXÖÂz^ý“¡#Ͱq¡aû¢Ö4l†%Î%ŽÁVÈ)åjßؾNôb™`§Úe°ôx"ì26ëœ9¦b\giYàïïmóÇqn°—®mvîÉ]$n;ëîÆ‰„r†ä}!HäÄŸóÁ5‰_¯:UßÝARQwOEV$«¤8%ƒë¯YTw‘ʸë8Á“ <¤xÕüÚñH#·)Øvé¢[ÂÖøS™¢n2ü/2ØØlÊŠ#c¤fqp:2½±¶•^dž !Ps¾ƒ¬<òF:בÍ|¥æí§Ld?÷”J]¢92ˆIü½ ‘³²îÛxq r-ü¼J,UŠ<‚ã±øÖ×OXÙìØäsø¹rIe ù‰S|ÿªä!ÿG#gùÃ)(p),D8è' -Ðs˜ÜA!±„©ˆÂfõuIëF•‡¤CQÔ1ÏÊÑØÅÞ}ø+„²–¢SP²%¥ƒäžJ§¨\žš–@Ù  Ê@Ê%”+0ÖkÈ»òe‘ßäðã­ ‡gP±óϵ‘Û¨L}{CiU‰åc¼»Q5ÀC5JÂÕÊowl †Êñ,ÊØÔü—”µ÷†“ûëVJòŸó©`AÝlÙ‹5 Œ¨Ï¥h}çÕ4(‘áy²Š†Á‡>ߦ¨Dc§¿dÄìrh8!Q~¤æäÛêÛæ£ÅÎÏqÞðhõ¹Ä³%#mXù¿÷ß‹CÛkû‰Ê ýH½ÓV m~9ì+NÞ½Ýè" z\U$èÆ]~Yxt= ë÷¤}ô²L ³ýÇ„ÞyšEþsÐWµ¤à&Oª§Î* Û'«ç2±É¨rýÁ.#Émdx{Ûå´| ¬>ú¢ ¯FF~¨öIp^G}ztXJI;<ƒV%/o[Áw>~²Ï>•èíµ‚š¶KLrmPm wE±>ûà¨1æt0WE9¶D¨N­Ÿ½óè ·zERƒ ã¡áÓàCÏ¡·kåtðn mvލß/…QI³{ŸÆö`lràVýÅ“0fÈ<¯l“Ý1Sô2;0Í© §¶¦`ÆdæT&å̵Oëy sŒ‹*OÃ\íw'‰ç 071ÀŠò8,$êzA»‹¥R‚§™`i}S5ÐvV$}ÿ3æO€•uÒ›þÁjçCçÒ©ã°ÖÑ>Aj¢ ë+UD/ [a“Ï8“ZM¶|Âß·“&Ãö'‚kGGåá7'wëÌ·.ؽý\x”ôüa–Oùèÿ¦-I)ë¾ÂþôŸUµx7$dí-cE"×»›<>H\^ØUpyI%¾× ÞÙC²N]AÍ¢H¡|ìbS¡,R6º%Lü(Fêà+üῳñ°Ä°T{{.Òì SvVç#]ÒMé!ý¥qÃî@dHb© [k@Æ;ÌìFæªÈ”6QÂså/2ß(ZxÇf,1B±Ä·ï!kðzgµD*²[|û–Ù½Ö”¬?¼B¿ ¾+wY‘3x@$д¹?›¢NEî—II’BgPFêÒjò:w¼è”ò½ë;Í=õù5J;8¡€ÔÄöm3 Ò:h-í·¢MçcoSýúœØè-ŠØD=º¨¢wÚTþ<@±®ÇL”á£þñX%#Zþ.kÑ¢Ô`Av Êxž8úW¾åH‹>˜±Å¢Ü/³ HRAù4ô+wPáý™,ûi:T\¥šé»ã„Ê7¹˜sØêQ5c…\ûF)jPéRÔ B­ÖÏ¡V߇ôeozp¡nêætÑNê;™£JÁ#–ÌÕM§·ÐèÔk*¡v4yíïÓG‰xøÖO¢EÉ\¥ZÝ”îÕ]´@ëåÔvb´õ LqŽ9Œvíjo*WòÐÁèB1‹Ó:V×Ü™|#ÎG;4ko™¢Ë ÑÚMút+û`û·z¼J%}‡^>†Ž™Ý˜l.ôxѹ-d2’iÅä“.Þ‹FâÃfv‘ÝÆ0v‰ÿ*EÍ}+[/ùT ­ÔÒ3â é‚ëy+U€ú~KÓ×áÛçæBM**¨éW¿0b7Õ§<Ódø– *×¢*ØÿT…«ûü6˜ƒªN·ç/Ç¡ºxõ7ùÅÃP»ñ¬¸ªð/40SD|_w€ïQ²·²¨sà'ÃM(<ðQ†å_ÜйËâSÅ<󵄠sþh’KstíÉíôI…îW¤d©OÏ@q@ÉÙÚ èaf_éü©ÝÛ¯ü„¨ »_Û£íy;ô(<ä™\‚žž¬Ÿ2%è ¯à·Éƒ>Ÿó—ÈÍø¡ÿ’³Z¤¼ ¬¼äž¼B†±ø•dÁ¾'N#+C/hcaLÖiãTõ*Œ¿êê‘:Ÿ“uIW›aZ¹¬–©V fGHù>Áüë°²GýJ°d&6¨¶\ +wò†l.4š`@»k8¬÷~´“„Í’¼¿Ñ7°]q6÷¨3üÞùcu½G þš&ä¿’L½ÁÓÍy7‘ ã—góÚAŽé‰É{‘ä°ù·t¿9$ïÚUïGòÞåg¯ò!å*§Xáø¿V”èJŒ4—GjÿxžBº(¥©W‘¾ñ¼mrE82ÊËZO@¦÷ª×Í#‹õÝ÷i|\ÈFöóÊ"ûˆŸ±rΈ=½+÷ y¤©:zí¨Ï^n÷£ÁÈÿV“C±ŸÅB|”ž“¢Ð÷R¯¦ë§P$‹Urþ¸?Šå2µ¿õE‰^RG_º”FÖÙ°×Z(ÛõÑ5}ã OÖßoÙõè£B+ÎéßG%_sþñTTa— vCÕmáF1½!Ô µÍ Q6G-+³ªOÂû­Hßk:_£¼$QO¤gPüêgÒ:ù|pÞÔ´Œ†¹o{Î¥£±¡p‘Ø›,ßîß-åE³’ÈO'“O£yÊÝG6ãh™4¾ÙGZV•µ§&¿<@ëÆÅÁž9hÓ'þŠF´ðÿ®Rqdtíi’„£ÞV£ƒâ–®1:ºOêÔGÙ ÓµOÙwè\fI=¸„.Ó ßûy]Ñ]AërIº[ˆ\øÄj€ ŠjŠrsèY]%/a^7Í.Ε`Ü!ÿA7A:´ñl¾¹ZºŒ‚Ǥ¶Ô"`?ÒD×®~ FÎõ¿ù[mÝÁ3¡vvÐôý—Ù¾÷|3’¦Jrƒê¸u"NÚËPi`Äê–¯ŸÓR³»|ËࣱÊ2 É|ÐK¸nÄP6áKMël eÿµp¼ùßü#aŽÑ o|¡Ò¸<~‡ýëTmØ"qµš`7Æ2©¨m‹<÷+ êˆo_|Ó#õI”ã)<Ðð$«âÞo:h\0NLˆ¸MËÎ%ú¯áûT–³Vê4MíPÁظiøiAP•˜ ­fc‘ïâ9 M5Àk¸­Ú~W“Z» 47„ÌC'}qJw‰4tY6(H<†î(ÍH¢,èÏ9ÿæG4ô¹Uðæš=óÜÇ)E0DŸY/¶ã¿4«LüÏqÃHý³Áb¢Z¹qò­E!Œßv .úÒ“2Dzԓ4ašàíÊfu5ÌŒÎéôüê‚y*ÉÕ¨ý°¨™à)êN ˯Än½ûX«ìž WG7aíÂÉN×)XÿSH=,å›Åáíòô|°}óÑó¤Ù{ð;•󌧶:üi|vãñØÓº§ÝÛG>wÛ!eZ—]|—YGXòZ éõlj9V­HžÒË^,„”µrÙ“xˆEÞ@-ô Ò¤Õ³}èù‹t/ÿSèìBh¸ÒÁAŽL<¸¶lÎ,ú«Ä‹—íÅ›7×ßÄ"§ë5³ÃÈãô‹qTäòu}Ì©#q@Ü3½ê†·Pè?‹¿×›ùQä›úu¡R{çKÿL*B„’e–΢Lò9 íòv”k â˜&<ÈeHS,A´3*iÒòÅ[ñ¡Št\Œlt$ª+'¯Ì¡Æ5§ëyÛŸPk+·½¾éâõYí_ç3QWBçI˜u<êå~Í"o%EsÏ<2Ù$Eñ˜ÞÔÆ— N¡IÂÒƒ­n4‹¿+Úûéš?©1º0LŠ–½$”¬³íhM_X¢œà…6Ðñ.––mÏ5çïw ]¹s†ÊÕKè@¤£ŽÎâ”æ={èô&Ü8»…]ÚäÄ_Š¡«½†_Zü9t{cS&Uø =Eß¿O‡žçÆÕÏyä¡×¹Ÿ¶o$o ·ò«›<+N˜:û2{­¿}ZlhßÕ¡ºPÀ¾ÖËÈŒ¥Ö¯±ÂÚ€Šl¼ú ŒÍžªËº­ ½±Iñ/D c0l…¯YÚØ?­¥$NÀ‡?B¨( éëäéãë¡!ý¡ÃQ× ¨¯þðáÞCh0Ðp^áI€Æí®Wi¥-ðƒ`qÀ剴öJ((;ðMúû ¡gö¼#µ l}Mh…ÆÙ®ÊÅ43çõ¯‰>öoÙž0íöMý÷…)˜ý¢`ÆwŠæKÏe½¦ÓE¡ß,ÿéÓÂ’¸xNÐO€e‚”øpXýù=•VV.»¾W»¬«B)¤Vůaµ-oR{ÆÖn¦u¿u€u'£ê°¡g”©Vc›z‘wý6&`ˤvûºl'š„ý;m #Ç”`WÕÿûÞYøóҿ¤ þi:‹ ÛÀÞÆÓ¢ù Ü6¹Ké ÞOyµ‰&õö ªÄÉïÜÈL)’ÎVrìå'#ù£WYLHy̸ûØŸHm&ž×qó*6|Îúú² Ò:;¿òϾ‹t•5}eUH?¡Yx& onì¾B&£†$:~˜ìmz¾Yš)òyÿôiB7Q.9²«´7Ìvpçé~Á/RÈ™ÆrZéž;r5œ,¹ŽçC÷4åÎtÜû½ÀŽòÇTug§ŠQáø&uéϨxÁ¢kêY*å fÃTnÓÜLùtUIŒC¾Œ¡šÃ±k|>¢úp¤–Sù Ôüðœ8Xs µ7j,XcCPçˆÑ…ÆÔ-Ëdc¾õ õðÈu¡Á_³wy/ÙаM­šÑû2w¶ô}D3Bë+£Ï5ÐÜê­½H‰Z~¾ k\h-Ê.u=ž m’Þ|w´‹@;ò‹´¿v×ùwb:ògø-i¡S?ÏðD—héŽTùt³Ùh»£WŠVÔÕúÛ襻tQÍÞ½£~'XV>@Ÿaú¬õELò‹Õ ¸Œ®5~D¥5‚È_¤AHÿèåƒ?Å-°üÕ:}ø)'üZë+òùúä€ï ÒN¨©A»û{EV¡ÛÐÒ3õáj;£déôá[CŠ{ñËh¨§¥ êè‚*#É{SÌP%8=Ý&üªãÉnœ.H†oï ²ž‰DÀ÷lá!Ç%hÛ¥s 9+Ý_¾Zœ»ÝgfRèÂtQ'ÿÝA?¾JX‹ú‰î~Ç?´x)£§A¡Ùá:ôz6ûMŸ~¯ùmLYœGg·¿ UfÏPäT¸R!>öyP¶#Sd'¾A§Í­f+µÐ"êçÏò(ÉbŠyùB]òÏ~U–P›8K~«ª¿Ç'ë…4@U°csÁ3¨,Ôì¼x*“”þð{@åÒˆHîÙ‹P5Âã"¯µÌDOyŠ’¡^S7L£ï4¥gˆ‘o@‹BÛû¿W¾Cë {½p]hÛ7ßzú%tH>?{¨S:(¾ke‡.±|ÑãÏ ›@bÚÃbºMÚ}“l Ûòã}ú3Õÿÿ¾òæÅŠÐµ}BéÖütë2]Û¹v°Ÿè¥+ÿ·òò„žhðœùú{qè³-[艆þOJ+£F0jîópš †5Å_j<-„ãFMI!½b·ÐUûÆvÛWÌ a"SkÿçÏN]å%½¾Ç3¹*ÆÚ×anTŒZëañžÏÂ,g&¬¨ü¡UeƒÕo´„_˜a=Šé屃\Ûˆ"e†m/ p~ø}ïþåk÷‹àϸã¿s^ï`/Tözänüwê±Â­,$|ø-‰ê w K<Þ¦FÒ&ÅûËbHž£Ç÷OQ)?ìô»‰ÞCê•Û*”£õHcD§@¾‹t<—îŸ~ƒô/ û8BF:ÇêñéKÈtW2;BÚY¤¥ËåÜGÖaûW•uÈ^ºt'9KªŸG!÷Æ(/!'!ò©ð1W“Gþ듨qßö…¼tÜA¡GßïšF¢HÀ-ZR‘pó®®4|©»Žµ£Ôë½Èt+”½`¸'–zå¹s¼é5Q!m£ôAg*ñÚœ5ÍGåfs~âðXTÍ™H¤-KEõüMŽÄOPs¹–71ÃáTø`ˆÊêÜLÿJ‘߈ºD.§mî ¾ÿÓ—º bh0ñß~ú ]4JãîEc*š\ócÎhR8Oäb‚fçúY¬¹_£¹u£Z¢ZZwDq—ö£Õsó÷ÕϊкD]u§dmêÇÔ¬ÝÚÐv¼ª{VìÚSæÊ…‘™£ƒªYÆÃMt |ÁÀôPr„ßÙkD¢s?­Äëste%“&±_D7'…»¦1)èþ8ì%͸)z̾t+Ћ—4ŽFù)zu.wæ[-ãU+ŸÌP÷l´Þ !<î.üÛÏ¥ü®Â?Ñjyã·0ÌËGªtmZ½çp…F6\ÿwgjæW_¾ ï…*¿Gw™ßï@E†Åçäí!øØuªüÅÞ|¨YZÎW¨ƒÒ&r …ûPzõ¤w“Ç”Ê3n}ºßþ¿ù‡{¯ŸYÿ…Of,aQý:ÿ[§rèH¨R8T+¤5}>Ÿ5á:jZ P{Ak?1­¾M ò—s„ºžèy홇Ð@™V©¬ÿ fÍïruØAcËxmÀ³Zh*‰J¾(ü¾×˜Ü·á¡…‡vN¿¯Ý…Þ; Z©>V‚Öª¬¼åFhëÖç"} í­¯Ïù°V@'EÇ’ïØ=è’òÿW1¾ÝÉ ñú9‰Ð³ÇNªF}ï‘8¥^ËÅ8ÁP‹œ¦øš*Œ:~|Ý7F,Ý·™Âر×ï½Ha§®kdϦdÙö+X`†h;äüf?Ì >4/ƒJó7 °”~r8VVn±ÆÈ­9ÃêDÙìY¬»ÙË´ÚÂæ¡ŸY©4n°5ðQFs¸vÚ5mŽ1àºÁ§àŸØ—È—Ç‘€&çµèr9Þâ>>°`ŠÄJ£Ú´HJ@C½rà]d¿?Þç"EJ!ò‹FHíåþv_·õ³{/!vä$Ë Ò×Ưü±EÆÌÙ[­3YÈüñŒ¦jk6²±Ü®ú²B†VÑw(ûmDU¦-òE†[Òf_EÝüŸwßj£¬Æ÷ã¶(b6¦jouÀ«ÒdK3 PRTfu8‰¥—S ÍÔ£\øª Æ³¨À&Õ"³Q‡Šý¥œ7hQ¹î”ô ‹Tí§! u5™‡ÜèUP+âG ÷ :D²ãöMÉI¨3‘?ãüÇõÜ÷Í W¢þê Ú´ãöx$3K×ëš'q<õ_'šèD“ nC3ýžšs#4 |š41‡–¹9>~2¡Õ?1)Ò±p´‘Ù™n&`@Û ’KÙX´+y±Ô]‰fñlKÝèè’FZNŸêº Côú¦µ“1]O®øÝzS‡n×ý_íÈ¢‡ ‘€ù”z>VˆŽ£F¯Çí¼“èm{9ƒñÔ(¦µZˆ•}EºüO)HPÕš¤B•¦Ç”Jôê`5:Óú¨»ŒÎÊrR!…žà„½ŒXFè 8Ñdz· ­¾²£Nc+Ðljù¹ö4Þ}sÚHðÔ›],³ð‚ºôärÂó©POæ½õÜPíÀs9?”Á÷àû¾JÕÝðsd kR¡ýïׄðg¡;×ç$i©>ô_Š®Í¾+ÃÁ±?“ZaÔ|7Y©ˆÆ sfbÉÃ%¨*ÚÁŒŽÄ‡Í«R0Çji#E› ó'ﳟó ‡7‚—a{ã°(L®Dãì ‹ë¿3:ëïÀ2Qì›°ÂY×ü»ÖV*_•¹B «a~>­l¿`Meý_ ¬sp&ìp(câ*ylò“ö$ÁÖQ|-£°Û©‡/»­ÁÎòI—¤ÉhØué•Ùè³?#çHRòéàßÍg Q°QÍ™³ô zŒýµ^M wZîü©d$¾t#â±=’2†ÙÄÏ Y…ïÝw+¦Hqå³aÓÎRù´p—Dâ!×~3¯?!Í©÷å“.lHgv´òÇN"ÒŸ•ÿp²%,Uâ¾\/-ùåÔ?”±¾ó:Y.ü Ú)+@Ö/\3ö)¡È¶½çî”ߊ2ýoGØó‘Óÿ8Küy-äzÆÓª.)‹Ü³ºª5CÊÈ«è»vŸñ'òéLÐÊПE¾¥”ëã­2ÈŸKíÃk4ƒRsÇH™QPE. RóQèðUþ¯{(´E tHÄ…·'¥íKÂP”}LªmÅ<¹ò&P¼æAn8O5JÜyrԥƴ)Lu Ì#©eZ{('ï1+)Ÿ‡òœe”Iö¨À0We~ ¹Ÿð/”ÝG%=¼Lã(T>Vþ˜ÙUž9û[‡£êæÙË'ÿ¦¡zÂ;ç+ÏPóì·ã£M#¨Ý0¨î|à‡ÑòD2¨û“¢Äñ<ê_lžk‘bÄ#' O¾C4Rð«¿÷QM¬^|zn€f÷+ú¨(êÐ|¿Ô‘Š)­tèuߪT¢õíK¿ÏG£ãѹQt–äìtA—ŸÜLjŸ  [’œÀÞÓôð>ﻺ·Œ^JS\o¢òW梚®Œ>·ˆ)JÆá5ß[g“‡ Ñ¥¤aA·í*:ùÿyEvékz®©(¢e¿çN ®?³o^‚®™÷!•}™ÐvÔzÞèÀ£$^ó±XhHîÕêx+ µÅ«ciÒ¯¡jŠô.axTšŠ¼|X’ •ÄiÓ—{x J¼«ïÁ ¨YÝ9ïññ ·tkÇ’Ný…V—~ÆÐ¾'Й›")ν¿èWâ(Z`0%ÈgñŒ!üºbÓ²vþ'Œ²WyæÆóÁ˜M8…¡ëŒ_¸ìm(× w*6~ðɄڗ%ä¤0e5UÐ?ÿ¦É‡~ÄÃô góïcÞØüïKû-&˜]X~÷™Ôæ™ÛÉšFù`Á«‡¸ü`±—Ù=Ñ_–/%{ÓVJ:]â¾xà‰Óã$w`m.W·E6”µÓ_eÀæYºž&é"Øú•ð~rv4Ì?*Ý+‚ߵĮÁ:•ðç|ÌsŽ/µð/`Du;êìgVðŸ˜@BrßBë½=$r¬åþoމËjÚLk#©ón ù’å4P¶ñF A'ÑmõøÎ$}¯M/Ô>ý×NVãa “²ŽHËb:U¯z énÓ6PRFúÇoô9º áÍã°DlÈX*vC@Ñ™Þk¥g&/ saôjþ»òØ'é†FØ!ï½QQ©‘Zä;bäÖ?ü|u‡ £o ÿ¸®Áëø,¨ŒôtZAÁrµ“)Ô(ÔÔȻܧˆÂË×T«¥ž¢¨`Ò×­,óñï—)}‰â9eÛI¿QbªÝ{Gü-Jiû¤TE|Cé÷pãTœ0ʺüÔO0JD¹KÅÇʾ¦¢¼‘»ß*(Ó}{¥ÿC© %‡£²„¯Î®m ª†}ï“;‰•×èP;€qÊŸºuÎŽf_ðD=ë'³-ù­‡g¥ÑePðÈ2tó§Z;¯Z„ôÌÜûOUÐóWTzs?zM-È$²aRT½×ÏOtšI£ˆµA¡ü½‘‰ 2$½0DÆx †ÿëΞ”ï†ö_þZãÂuЬÖºKØ õwŸ.Í&Á7M+ñÅn¨‰¢äïdv€ªÏmÏRò¡ò¬¥î?-øzšödO:|…Ýÿÿ[|üí]˜lM0"V#‡éµ×[:a΀¼d½Þ~eþ{ht–K_Úži†UÿD³|Z"Xçö=Cð(6–sórMaëß¹?³fà·‚ba¢“ü‰·Á{]°G¾bÙÅŠþ‚|¤6G‘pA ©ü0?þ{ëùo‘†¼½Cú,ÉO¤…%Œë"ååîäAÒH]ö üFÀ ¤9Œo; ‘NG­.ýhÒ'ßðx}é 2,_Vò6¾ƒL!¼ïsU‘…¢þs¼ #²¾xm}Ùã ¯Äð§!ç·œŽFJäþ\(>ºƒ|¬´·F$ë?à–Ÿ®] tõüä£X@¡“úÔ”I(¢@¶Æ3‡bâ*E­O7QÂd•Îèã JýǪ{ù²/Êrud”ZG¹ÙóܧÌPÁù§´Ñ1STpZª}W…Ê1Áé·/£êѧÛRz¨~4ýÞcÝYÔ¼¥uÉŽæUñ…Ö{§QÇõ­HòjêVI½'¸ŠúŠêÆ,~¡hð¦ F5Ÿ å ËR¬9ÑèÇöƒÐø4 Õ–¥Ü E3UõåÂY4gÚ»wVS-9jˆ|ü·Ðêf÷¿¬ëßÑúqÜ]!*%´)Íø–vû w=›Ë7Ît¡=Iµf[:hd¯¬Ý9‹ŽçOŽ{—ò£Ó‡,³#èü·¯Ü¼ï5ºþN‘¹qÝî+Ì1™^B÷¥ê~Ÿ·Vèiñ¢zÜâz¹½0%îæDo®4ÉÉ2 ŒMswKz ‡Ö÷˜ÃˆO"_2¿OêÄ4ìvîtnô‡›ä‚3)÷¡¥¤y©¾²êý²ý©¥ƒêÔRÓ–»-Pi` Dç Ÿ_Óšºe)A9‹i{tÆ!(ü•£^á½ubDŽá ¼8÷!Þ/Jj[Ã.oýo^:šL2Îʯ¯°ÿß:_ãs¥Ò8U rãÇ•ªjW¨æ$×›˜ÕK3W½mÎA­–çDu |;äÁ:Ù+u"#;4eP7®kïóêËÄdÓ>ä@Ãuzõu[uhL¡{ñ…‡š¾¯pŽO„Á÷©DòßàÇ—HÙîUeøÉ°{›óᱶ󵂶[¹®‹- ½ê× öŒëÐ1˜v墨A¾TïôK…s£™Ð{Œ«7 eúõ²ÔȬ`Ð+Kjጠ?±óO†‘— –ÚÓc0F±3óHÆï‹J ‹€I÷ØÃ_ÊaZˆä õÌjxëòœ¼ óæûÅg÷nÀbn­)µI;¬üü>LôV­ˆ2Þ½„µŠáÇ1a&°q\ìä2õØï-Yêä€&WåÀé簫„Y¡¸k ï’É*& —ç­á·h.m¹¾-ÏÂgçG´šÉzÚ 7ц—‡‡y•m½tlzÊ¡ÝÛ] F*Ɇ¯gÐñÄ~_¿b4:5±z/0…¡‹èÑè¨7è—ç—ÜÍŠn“µl’–ñoH¾ gÝ•š¸cÁè•7ÂÔ½C쟓Ïâ­ù­/âF¼š÷EÅC•/e_’¿~âŠY&3Ã`E€è,ãåNY¢Œ ëonøWhWò¨æ– ?² /nAÓîÇ×l NÐàî´¨ÍÔ™üËÐ%€onå§Nõ÷†kw=NÍA½rØõ½YÐ8¢3O3 -´Å¡âÅÐ6=ͺғ ]ÞæW¨½} D{“Í4 Ëå×o}{ #]¯óâ3a,épJÁÄû2¥S²Ã0U#(ÒôPf"ÄI^ÓRÁl‰2—ù¥A˜Ké)Ž–I‚yÃÃ}÷µ¢`L0=f IÌWF—‡aIx_üÈK}XöôfÈP<+® Õ±°Êœ4Ó÷B VWk»4X`mâžÅKVKX_ øsH°6éçØÅaËB†Ïª­¶Ÿ‡Ý=Òœ¿©³²‰›a7>u”S¢þŠ<ëu§\…=’!þ‡a€,{dHh?u4¦ß‰>†>¥³]Gm±dj^$åLŒiEòDzer‚«Hîì?ŒÔÁ?Š„ú¸ðp,}ƒ/c&Ò¾ºúéþä7F ƒ"u~M'2F½}Õ?Ò‡LíºS%È"Z*Ûv‘5¬h7ÿü²•K‘Íøÿ@ö߆ñ™á‚È©ñ¾¢Û™¹bêÂ;%‚ûû ƒ“¦#òr[kÍ-!ŸòOBt òMÐdõd®#Æ©ãéǘQÀýš(IÛ”XýQtl…±¨> …&nÅù-K ðÈç[‘ç7QääDzŠ©ÈWʨAñ›_$B¦^£Äþ=«ÑA”º¡cö7\ e&Ý»Pv²3âÃRʵ«´ŠŠo£üW6ÒS>¨Pµyj­é9*Ž=(\±9¸?¿¹ôæ ª@ Û$ª¦]9_sÕer·E³PSf¼ÉÖ¹µóiuQÇðLì! Ô:gõÑLõ_=, Z°Â#7>ø£ÑÝÔlŒhÒ|9=öY •‘’S´A‹‚/N‹ç¸Ñ*gr¶b#­ÿŒÕ‘ºh£mз&«v/´[`ë;CÿÂÓ·ôW¥Ð‰D<˜~œS ¹CЕïÐÖyÆztûD¢Ž¡ãìüåÓè%òÕš´½%ÎÉðµ£Ï‘¤ÂÐ=˜ôä£ß@ñt~OóýÂéƒ}ègÔyr^¤™þ(%"óÕÜ£l0h8ݸ_:·E,Ù"÷¡5²(/Ã÷2ñ#ÄbP¯«dúÑ‚ j²ìýà ûþÿ»mÜ;ìƒP‘¡ýÐ~¾Ô]mñ¦†¯ÿ™š÷¼ç‚j…×–ìrýå;]îvÐKçÒg.#´ßt§‰®ô€nÛˆº.Ð÷Ò>:_Ýøòž©™Àpï%©ËaI a÷Z‹€Ñü¬µ"ó&+Ûc¹•ãO÷ƒôÂÄ™ä×3`RÜŸü¿¢ ˜"ø¾oÁhSËéþù0Ca-™dM³"•§/Š~†¹cŸ«¯^£ƒù6Žè"érX ê}×z*–M_} ¥•…•7Ûý1¡°ú³#ímy(¬1]‰Ø{–lûk°ùôÕhæ¹iئÉ!ñpï„;»ÁÞÄk°‹÷wÕ á¯¸«ÃŽìÙ@„ÖW$p7»9à ó~TÝ{œ‹Äû[s·ÝÄ?즓ð ’öü¶)x‰äVö;—ÏœFŠú}>®ßHÝ'2QÞŠ‡t\ æ4#x 'ûÂÒqtÞå@zñøgõ|È é+ì–„Œ†MBÔ‹Èdr%èŽ2›H‰_Ô-@“u÷U-dµ|F¦¸l®µ}œuÈ~æ¶ù¯lqäHrÊ>0‰œ/Wɇ-‘«‡ÿšÑ‹:äaP6[øôyíÜìèÆ‡‘ÏB[ñ¯]/ò퉾é`ÈEþŠ3A´•I(ððPpšêeL¹st”…2Íóo ðç‡÷ìãQd~¦ìè Š‰>Å¡¿âžWÇBµës—쌿PYÀýŠìÀ«*RäÓøáKLƒ·º!ñÿWÿÝ¿’kPaôã‚¥Tª]ßÿïTÛW\é}Ÿß‹ƒØr‹¡áû[‚‡Ð|Ö•ÿ¼üÔ½¯O:m䯆·¡ëÃièíòì €Ná¥×?N7CçËðüß¹íÐÙ[õ.Ø:kÿŒ'ÖCçEÚìÍ8蔯9æî+¢^Ü_¡óW¢öÞ"è2¹íØìaÝ”d†i×Ô ‡àœýöVèµ2Mï§õ…¾ÍŒúç+Ù#MÅÖ`hU§¢‡XF„å60Áèƒ)ã0®*þÃ;îÙ&¤Kåü÷ùó@?…ÜhÈûÞä˜uöá¿}ŠÍ¾}ïi ‚’óÔq BiIczüj*”¹?#?LÏåIÖñô|5P¡©®ý°æý^_”’}ÒS•ý‡èæ“ *Á+¶ô{ T»%Oy7C]W‘8Ï.ÔÞ›¨c$þu'é¯Ó@ƒ{ÓøÈsBøžô¥Ý61 š /9}#‡f“ø#W¡ÅæÄ‹[Ðú ŒTé\<´õÜJ¼`ð:4ššƒ*›ö¸½s±1‚ º6ÝLJ†ðs‰–7sö,ôh]ÜËÕ¿ºÒoRž^…!í:á·g`¸!\ëàt ŒÞŸáøü ÆuZóÆaÒ"`òLÛª9ÅššÁìF1e†ð{…^–UT 濚8{‹ü¾ÍìK%7UMöuÀÊ•µÃ¯ëŽÁšM»wÑlœý ÊL [¦Eßà¯É'–¤dH+>] ƒD2~ìóñH<9ÇüÈä’þ½¨v¹ÉG wud(‘’’12ý"RÐÿõmCÚj¯¨þÎoHßqä‘dñ,2ƪÕ=ãKBæÏ´?ÑãÈ:o?-èK‹·2xeŒûØÆÚÈÖk©“S€þœßôƒcQÈ’ñ —ÛŠØËGäÞ~†b#Ø8î‘ã~²¹øœ(”,(uöPTƒ ›§*RQÖ}ÁãþP<Ês’0¼Zø ëÞ÷û*"äŸAÑÈàjmI<|ƒÇ¯ÎƒÕ·\C³QÃþšˆÉj6¦Î‡+¡–+ÝX¨Ý-Ô¡4ð¢w¹Œº·7;ç þ§Ç;oVUѰÔuËÓ› £ß-~BS>–Í/Ьë]ì´ Ó²r–DKëë'âýB«÷†¥nè¢ 55õÉ& ´½äy„ß Ð®ûÔïÇyÉè zQ|l_sù{ ·kÓC›Ðù"…Ò׬tùݲïþdº–œöu~)‰nû?mŸ^ÂGƒJæâ#ôèmfW÷9åÓÕÔ$É-ALöZ¸¨0ÌÝk‘ø¢ÿ~ f¾~²+F·båC‹«å{.wwhœ=x£WŽêÞ‹ÄyõCÍaõóä·žA•PVõCù$¨é‚wa¨¼—#ÃUíZÛòEw¡ÖtúÒ |‡à·þÐÜF›æÉÚí<ÊÑ Ô· ëÞŽH$áô±ENÃ@‚[¼ÌGwr¹²õ»FŽËȦPöÀ˜.׳„,}ï!Ê“¤=[m|"qG`²,³(ÌY¦N1ñóEÃ4»±Š³òQ˜!ãPà< ³ÌÞù1 {ùSýlnšûø}‡H'¤k þLª]zù§æëJ{„|¯Ã—¤F©!°˜GÅÊza –:…’w]`…Ú¤T7ñ#¬º\ÿNôé ¬}- ?*­ û?§d<€Í’ä@õ?a;r],r#þ~°¿¹n”–˜õ¸pæZA*Å$öè+{ìðI‰ÛŒSc‘¬ðX”)î{Âýñ#Ñ)¤ Üu#½ŽÔÑç:¶BÀ×kYéýé Ý‘¡jްàù$2± ‰TËüDf?9þÖ`déx°”.Ý‚l’Ô[nÈîoŸ}U´9*>:ÚµÆ#-é)1Õ äv<ÏsA"y2ãgg_\E>¢¤¬áräß¿.50³‚ü=>5œÉCˆž4·Ž  Å؉·ñŽ(Ä÷q[ô° -9yÿ`Dá6ÞÑâ¼y©{ºpØEûxù¾9£8æ9lx‡¶CYç2£qaŒ¿±J*jEdš]G©ú8/oâ‘*w~aÖuŽCÕi"©ÊÉ"T¿ùiÝßY5ôˆCwuuQsghŽ:µµz=îP|x„:ío/¬—úê•{ÉfŒC³ïß·mÓ¡­Ö” éÚtVcì öGУø;Ðz½z¯Y”¤õCÿËn«5†yø•´O­ã öem¬y4ÀP™ê6EG~$¾rÆF çp¸QeÙ¤2#c,8¿M«ãB“­^¦À„Rfçó¶0y¶@Gb¥¦êùú–M`Æö¤ã‚q9ÌÈø25rþ<«œýæ¿N¶¨£éˆ6¿î·ê¨!d3Áñ¬·ÝÌ©PŠŠ©a óJ§«°ú9S¢ 7—«Ìê Ê´È'z²Ÿsf MPn‰úJà~?T¨I÷²|¤ˆJ&)ïõQõìá#oO¢ú®§Ã‰êÔØH“dZŒ¦3Ó¶)¨ã@á±­ð õ­Wk­¡atÐAnj4ö2Z« ?ˆfbRb—yXÑ<¹I×ð¨ Zæ)ÌqA«¾²‰ÜA´ Šþn¨W…v,­¼³W¿£ý³ÜnYCt$º@ß-"ƒÇ ¼ Òð[ï¦áÌ¿O®\þØ ‹.ÓBd`yÿÛÞ±ìXU6KÞXº ëž^ô·z$a3£a«€vä/ÖúßSÝn=ë,c$¼¡@ÇŨ€Ä2ïíó3. )Q€ #’mªpÅjG úÇœ!•NJéc?f¤y›I˜ˆô$çìžQ@†£¯Ï4©5!cF Û1ʯÈ,Eþk÷(²”lUø‰R ›ûÄûâËÌÈ!¡Ÿˆ ÈÅ™Èìf‹1LPFFöq߯BÈÌ35tÔŸYÔÿZå=AöñäJwm7äú¤SßØÝ¼-ÝÓL(Àjù%iÊGÿ*­L ÿê;<Ác‰bÔî5|B ¿ï+\Üt(ÉÛsöòb”Þ »êåN²T– í¨P®à=õ;±~Tx,¦Ìú- #\\nôF£r¶qk€S ªîêØñi_EußþW÷+ Qƒi¨r½5C¿ÙLV/¢›–Yù Ô.ã½yÚ0uU|¿O•€úçxêúó£a`rŒaåõå}Ùxަ´×.Äö¡YÉ£§~«h¾{ú^Z› ZæKk~{ˆVïbG‚šÃцfÖŒZ= mý;Ó=ýý/‡Æ5ÛLåÑñ3£ûMÑ|tâ=tM¢‰Ÿ<(¿ôìÆÿåðÃnéüøu|8zëý³—?Ñ%gi²å*Êvvm‰Þ~ŽtO]òí]’aºåÕ9­ˆ è3´$ѯøçÿñÓ÷ 6Ç “¢ÿl‰»\ÂiDÙ0¹ÿʾ¤ù˜"½bñ˦‰†õ¦æ —C~T%Ìš'^M_‡¹äSL¡á–ØÏ·j˜ͦø£òõßý¿Ç:â6Kõ\Õ.<…¶’’ØW°ê%¼jWk?jcd aÃî0oTlñ0IÝ ‚IR>2ýhØ}È$o~ª ¹ƒ›±D‰"#7Ø¢þ"‰èý \wÊ‘´‹5fÓŸ ÉÓ*¨¶%œ÷rè²Ñ+Ù~¤J¯s}‡4='/Xûf"=›’õ—§æÈ`×ä°%¸ƒŒ9Ñ 9Á{ýSð¼ÃüWd¹}"ïÖôqdùÅgbøÙ‡9Õ"Ç#^Y–ƒœÈ9dMw•» ¹•X*)ÅG‘ç±qMðZ,òN¡½ûBùù‰Ô•ì‘ÿmM‰b „È^¸“põ5µÏþE!fŸý±ô™(4º ½—…Ë2uXƒóP䣧eÅß5-VczÎo‚b£æœæÎd(!H¦pÆï&î÷}ñ–v  ê¶¹îoA©ã•ï)¢Pz–°*$ePÑÙýÜa”e»|£%cÊQ¾$ý;_òBVUÏOÖ£‚!éwÁgxèê­úÔr¨ØúÊÃ"D•ks¢¢¢Q5™ÿ±â•tTw´³ ßë‡1÷=N[à“p²G¨M%o–âÕ€ºÜå$ߪ¿ ¾õd7weVJI£±#ÙÀm©³hFv2Ó‹ ÍOF¨ \èD‹æ.A}Ðʤ¿ÎÐL­»% #Þž@Û±O9Üý/‡ö弉·Ð±l–Ãi%ĆٴªŠÐ9ªðö̓]ÿOÃèõn8ßBÛôËÚ”bSÛæex‡”ÏùDOø—¸…ÜÛ‘ Öù¬yEMq54üøLîÑ^Nöb‘×öý—+ÿÇOÅý­§êKj¡ðÝÁ–“Û @G©‘¯s eÂnµÒˆAñi±î™(}½È¥jÕ›çý~s*Â÷‚•WtýÓÿüó?Þù.»…–_^Ûø?Ù›®HÂ<ôÑ|;ÀBw ˆ* ¹váWÑTwÕKüКF”jC^×®Gѽ†aƒ·'Ö.èÂÈ¡½ÈÆEf>þɽ¾¨ÿJ½÷ó"Œ_øRÓ|ü Lì ò˜:²HÆ(M3ÔY9nyo`Žh$èÉiø£àê3æóemêo?Á"¹„q(ÿ%X²¢Œ9µËyzÏs~sªÌdì‘{¶ÿå2]1‰ål•šÉ‹ ûÂ_FÅ'êÆMH =DBÞˆ„ç9Æâh\öc¢[ ’h;-è|@ÒtóOz5\ÈèɸB8îËíhŒ³œEJ¯ u%Ť6M½]bûiyÙÔ#½Fe~:)2ØÌÞb8ƒŒ§iÍÄŠUéËÜ6dö}Áïö YBôÝéUì‘5šçm’n:²eЈÈ;õ ûw’²Ë©ƒÈ1g³PE‚\¬%¹g›ÛPN!ë‹òÜp:úÏwüT4‡õK?z~Vµ|ò:QÐPæ)ѱDRÑ[¹†Âhkmñé*ŠØš]W,EÑë©ïÈx÷ÿË­â½\=¥op?U×µG¼xÀÄûFŽº4J~ùëÔƒÒæs»¿ÏáÁÎøÂ^Y”‰³r»#ye}n0y…rŸôŠ™Ý@…h"‚OPépÔ3.¿T½¹O¨ïŽp”ù¸°%j.˹̈Z—Þy­ôBŸÞ£¥%—QÿÎ;âêSyxÔõsø)ŽŸhrL×¹m ͲóÏݯúšª>&h9þ⓱ú´~N«9À\Œ¶z›\Õ×–þËái2W)NèØj?oU!‹Nò\V*ʉèü"zÑÆÛéÿÉahba\1ÚÞ{{Ò`­ùãO'r½†•ìŠðièŠ~a&ù&~0^Oaçz5µÞ1®óP)çÈsüj3”K†Ñ ¶Bi‹ó_¯¼×ðí‘ur±d=W3ÅùBá¹`‘I!™Š›ŸÛ뉡åïuö8íj~à”1²Åß¿ó *­SïÄß䄚dî…Í=68'·ï–=&¹¯Œ'„ ùFäÞ<Ú†–pi}h:‹ jÐú«}´€ü¿ýY¥þ×Ú ´qõgÜ0…¶©#^ÐÆºÅfå¶7GD>¯QF¬@[D>]Y´Í¶sÕØ@{|r“ Á}è3{®$0 åþ_Ó¶¡Û’KÒ{*z¦2Ýĸ³ OÄöNй6 x»?ÓÁ×T÷éÁ°ÒÚ¶«£ì%‘)990®ðåû’,LÞ‹®#Rƒº÷Âå_a®?J\ö‘*Ìû¨îï…EáJv·µXšS÷ÌþF+'˜²ŽÏÁúÁÓwFiÝaÓÿõ9Ë­IØ¡b¬Ï}»ÍŸíÓÄçðì–̧Þ$fïú’6ƒ$Ãß¿çRò!Y3Mªà!Ü×OðŠjÌ©¨«ºžTÓ!›{·ê Y¤ëœºÒ‰ Šó8GQ‹Œ/ zpœßË™÷Þ< ‰@–mM|ÈfWž&{”9x£ É×ëQhƒ¸õâ~yoÙþc´ ÅÞ¢÷äÞѯÏ-¥êâÛqÉ—>¡´|j…+š¡Ì)I'·Øv”­§³^XÏ@yõ&v5CTè>T¬§ßŠ žúw”:Qùyc—šX ª6_õ¿ £€êª½ßγz!I¾à}‡š–›¶bx¤¢ƒÆµÉs¯æ=ûœ:5'•¸’ï¡ÞÙ¥æ|¦hÀô¼–¶@ åJ=ñ9ƒFUÂá6ФµøÌ›¦ÇhvF'jNÇÍ[Ûm £Å«ËZï ‡Ð²æ^{ò+v´Úýfà±L†6Z% —¬Ðö±wÇ¡Uh7¶2—™4Å ¢¬üÑ19Ü'²À˜x¼÷EçGMÄ×%ïqx“ïŠjÉÿê‡= $žÉx“óPß]-4ó?,üý3rúøÎ™s‡ÃÊ=µ=cèàt!›”%€:2={Ž ¨ð> "ºN%t•åtä?àëñËÏi%¹ _1a:Ÿà*|tsÓàj€O_²'ÌݯîõÁñùƒ%ÓæÚ_9gn ÔW܇œ_BYT;‡à“H„L| ºßàVÿoŸÂGiYç.AÑIã¬hMRø&#÷ø™tпsK£ï>Èh‹†2êCÑå LP6Óh+ðz¯'^çï/Ú‚ 2TœJ†Šì±Ô=¥Pía”ðª²3.F @u´Çµ•xã½<ýÎaáÔ…Um @CØàïÊEh´¤îHÉ´†ψT•6¡¹™8N.Œ Ziv2”s ÍËý¹mftð;õü tÎ|Õ|3ýº×9~T7B/K7=‹)ô𦲟€ÁüGˆEga(ywæÜë-Ù|&»0Fàxä=§!Œoúý U½“3ÃQƒé0#|¬Íìô9˜s%vKhÙ? ÒâÞùܰ Z;©Ù°ø=k÷§éMX>gbà›wk¯Ïæ;õªÃºPŸZéqØÔeœ=ÇÌ Ûµ9¯šÛ¶a÷é‚MÉ=-$´ÔiõÑ\E¢Ý1Ò~$éÊ/*:-€d½Î"^:à¾Ù˺QHÅ2L®Ÿ4—§4{½ž›äPØp;2üˆø¦›‰Lõ¤dÏ“-‘åoHéÍ¿zÈ~­6?Xº¹Õì98ÙW«d¸‹ò*ò×±µ?’«FÁ'Ë×]dß¡pXõ%zÍ8ß%ýæ'J0ú¦R?Oú¡š„â(3æ|¢& eßž¾ ¿ˆr1RÉOéHQÁ‡u¨Åˆ/É«§oFåûv7·OÔ¢êì|kœ9ªÕß^ÔôÓ ^õAͤ3Qï›ùQK#‰SQ» µ×wRºÈˆQ·­c•Çõ¿YD¥iBÃáï-® 1Õ¥ßÅ“hj¦ÃèsÍúUcDÑ‚-ƒÞåºZ:¥f„ơէFÁ‡’hÃr»r½_mo„KH/Îü—ãΞ)ÏÐ1çLA¹¾.:qëV;ÆQ¡óãt™{‰}ÿ—ÿ·:+ã#ªc&ißÐ9âà`´Á”™(­áÃ4]-šc9r0uüX±¡$ôem¤__¥ƒ6_­ÖTSø±¤ÃtgöîÛ±ñ9§U¨IÐpùtí#T¹»Ú•»A…ó]Æ8î=.‚B¨‚• üURGTÜìÙ˜ƒª&+GÖ÷P—`!w#\šþ¥6 @ëÇ¾Š¨ è¤ê6ñÌ¡žÓdKÞ¡*Ð7°Ã8ëF ƒÎŸäM{·aØ„™°µSF÷û¬0ÀØçc»cÞ«0^~ò6Ï„{À|ÝI˜øuhðÙô7˜ô<']¨"Sr¹Œv‘„0-äÂMþ;fÔ½f“ö™Âìåçt{Á\ùò½¯›IðÇÞòtIXÌGÉÐјÁÂýté/üY°øF•Òjƒ–Z:fµ¿2À ï¬WŠüyXõ­ÏÝ÷ÊÖúÜVÒüaãhºæ¨ l‰+X 3ÀÎÁ¯WÓ^Ãîí³WÎ0ïùŒŽWýðËE$ºÆ§D{ý.’ÕY”°«¦¶ó_ÂýikçT-£$séá>å”zÊõÛÀ\²ˆyäפ¢Œñã}ÎbÂ(Ë%K”#²»ÖÒ*(¿Ïò„óibTº<&ºßÉ«ˆr\CÅÌé–t _Tv¡$½pAUƒNª}<¨…ê|aWÖN1¢ÛCî¤ÊÔlÒön¹ ZïÃäžü<„:o» KIP¯jc=ÍB 9 )ú­"+ªNõ¡)õ,¡éÂS4«KØÇ6?¹Ç#‰Ÿöý<´ôe©"êqE«¾!²¡Y´ÑÉÙ–˜@ÛwÍ/O¾:ˆvŒÝe}{^”JžF+t ]üL¾©ŒÇÚ®·‹£:ËÚ:üŽ'@—7¿ƒÆ•‹ÑõÉ2Aëè SÙù0¨€6³âWN=Å©ÔzÑRÒH!,ÿÅÅÆ$»k’}žAWÑ"ëÉ€jhU°¤™û íWÙyÝBÍæ®øy¨h²cé$u…Ò’¢È‡½Pô~n:Ö$ bb‹¬K‚ ÿÈi&“h(`=Ø|~sñûR'„2»#—^¨š©Tû3rÜèd¸Ç ùüä»õCï uõýC«zÐA• n¹*ÐEpyžecºßºHõÃÏŠ_ûüõW ïµ~é=€ƒ<âO6øÕR›60êƒ Y|ÌlC—·/ ÊÁ°-}|÷ß=}Ødʾ£ÇJEÆ.ØÀØõóÝ®½‚ñêu©Þ{Â0©Þ7”cX S; l5I`fr§!ËÁ~Ó>9çëï ób×­„W‰qfGvgXLéN;²šËiÙ·kI`å¼kó`l ¬î$T´SÃzÕ…saü°Y5£ùŒÕ¶wÿD—R.®ËC£cœH0÷¢*óî~$Rˆ5cGâÐj?Ê}ß‘dà²aΑ_HfÀݹôö2’7>à V[@ŠKò%‰Q}H¥iÊdnpi”ôEŸ²G:mž³Í9îHŸñà‹‡#!2|I‰úÔ‹ŒJ'8!SqöÂ/‡hd®*Öu‰dB–öº*¢S‡u6‚seò7²ÓR÷‰ä@yå¯IÅ›Èéâàñ0‚¹ž`cDîî+¹W(#¯¨ßãwO!ßu“Ëò€ü~ß^v£ÞY8*½ç?ÞúÓ,;‘(DÏüâºÔ Ó¿Y[¾­žæÔ†¶Ûs›ÿyŸà¿~Hå.¼1_mõq²OÓ¬¡Íä7CA‘$´þ¢¸:@%²·RVü‰„6uÝ3øÕ¡-î¦:ÏÍVh·`)>Ð J¢}÷ ó,o]â`!tý^ø8pz‚O9‹4@oZŒ¨ªþ:ôÏæ¡ÞÔ€A£žKc04c"tý0?ŒTï·uÊ» c}rw­žøÀ$ïç]ÓU˜~¯ki:®s—Ó±Ãy˜WkŠü Sñ4¬“KY[ ö°Rè{Ó½õ¬-¹oèhÅÁ¦žß@ÛÚgØhwâþ»Œ<·“"¡]¹…ÑÛM$&$¡±xç$¼GI%‘,yX!]´ ÷¥]0,èvEʶ‘ßÏX©FâKueÒ½]‡›ÜõÈ ˆ¬TeÈx¯œØ[œ™iiN±7ó!Ë[Û*?Ed3zõ.Yˆ9˜~+kuŒ"éj>ëâž7ãØ+œï¡ûó¼Êß(@I¸+A ‹‚îc´sFe(4¦4o5š"³ŒO¢Ø¹÷G{Qâ¢cáŒx;üê[dJ»'ް/ìùO‘NÝ/e³cãY'”ߛ̕¨PÿÅ«¥^ _~i>â@‚ʯÊ3® jkÅ×…3Ö¨.s¯Öè§bë =^ó9Ô<ÀöcˆëÄ4ïõC©ñ3¨Ó2Ûùù©êEŽ]©G9’Ax”ÔGA”á<ýþUñ”?MÖ»+[—ÑìÊëø›2¹h~ïžËÍ`´x÷3EíZZ6åݽªv­I‚5/ð¡®‚vOõ^?œ>ß»—Kå]EY–Ñ_ƶ†£ã[¹çñW×щIÈEÄ/r¯ª5LvüÞË¥—Ø…KŽÿ¯~Øy"Õ2o²³ *+ Ù~v uƒ ä¸WªöƒÒ–G[¬V*Jÿóœ4õt= Ôšv&ßÍ8åÃË¡Öu•ðÍ“òܲľyË@zò^e]ðf¢ÏŠ¿ÙR„O †ý´Ôo ÇWÇvôOäè*§”²ÙCö$wI–|ä8}[ç¾W ¿¥x \ †ÏKVµ&ôƒÿö)$ô_(¬à‚¯ÝéðÚ8Ìþ Yÿqlþˆ×f¸9ÀüV¸ÍÏ(IX Y'¾ËÌ¡ÍÑÙ°R}ÜWGØÖRþˆúÆ®ÀFÁ‘—×Vra[b]ϺOþNúøÅvh ÁúÖ¡¤$z=ôE:#I|É4ÎdÇ!ÙŶ¾ÃÒz¸ïÆÝ)­If¤LÊ!5÷Cê­ô0Â&¤Ëȹg4„ ¡ÊÉZU‘ÉÿÀXs<²$4ÞÊÒÞFvê™/94³@nÂÝÑyr¥Ø:ö @ïÒµëZ(¨1]ZëO‹Â³SU¸3PÔÄ\è‰!Šgò:üeÑÆ–Ÿ’hQZ±RÛñ€5Ê|YáVéA9ŸlZ•B[²û÷{¢QÑR”žU—•=NÈÞ83…ª?_³éD¡z°ØÚùü%Ô¥¬]]æFÍoÌíç{ŒQë,ÁÈœ23êÈô5é”y£û$_J_Ð…¹ùrfâÑo ãÞ\@c ‘ªã*hú2Š3ü,/šmŽÇ¶ùA‹ƒ Ñ,"OÑÒû…dsgZ•îrWü@.•±ëUh|8eͦë¿ê/ï® cfñ™vtâ´czxj#kâ“2Ôþ/‡J¸¹:–ñ‘Ü¡…b¿×èäåÙ})Ð.8<ú\‡HÍ}¤áQ)L,q³Wô@¯áÊGhÍ& bÒª„¦†é‘Èù(¨ü|él[ TÏ+¼.„Ê‚¯êZïByI*£Ý ”eÿ5my·—?“áêc?>(L|ÜKŸ•ýÌKû¢¡¶Åæ)çhrûÙÿwÞxÏsQ},ûA'é²îÔÞÿü.þ^ë˜>ôs©<}Co2LXÀðÃÀöÊê¯0z¿c¢QµÆ¥û£yÛ9`BÓþæÔA˜hŸ 4'Ù€IE³—Á­ßaòà «–›0å–á¬ñ¦µzçEæéaÆådà}ÿ^˜zÚÁBuæo.œš€?Wã2”`þŤôà(,DéI„ˆÃâ{'Y}ê÷°Ô£vÕ˜,VD-Û}GX òœáú k“[æá‚é°áÓ–ãM0[òÖþ%ïúÿów›nEÙ ~ñh/ìõCúëãÆbÌÈå•óI¥U¹››Û´Ø÷PÂKoJfä{!Ý·ú ùoª¯u™H €Z룾6(¸ÏÅŸ5ç ŽûÂä"€B-Ç z†ìP¸mçsŠÌùVË.£gwÏoÏë(n‘ åØD…!£zA¸ÀZ…2ƒ% TÞÚé¾F©Š1¶‰ç5xŒÉ—î}DÇ™¶C4(Ëû;ìþMC”]Þµzïùåæ,…Ϫ£üvme°â'-AÅûÊê„[u¨ŒGí ôPÕ>mRj¼ÕF¾œ±¼Î‚øƒÔñ›f,j†©Úiu5¡ÖÑ^æ@½xÔ‘üýP»Àõ©8 Ò£Áù“5Sãðhupæå$c4¡¡NžÉ C3©ÌNó‹›hîtì!ÅÄ&ZÄ—~=~-‡„ïòù¡µXÝɶºhã1YüôW8Ú>?þËfDíJ/Èù…S¢ý8±ÙwéOèÈ€Â?LºñØÑG§ ~T£SÌ–åo­ãè¼Ð|óLU>†YÌ0jIû ”e-5™îWî}J«Ÿ‹ä¥CšäÚÔ0J8í1ÏåI_åC‹U<#SÉ)¨¿¦þ’R¾ªï´2lxBùïŠ*áÀ9(¥8–5ö÷;|õ§8w¦_òÃIStM@>·¿a½·nuóROCa‡ÅìZ·4”Þk~÷6e|KN¢…åPO§ÓÌq~hÖ„o\ÌV*)÷ûŽA;i¶è×ÿüÿaß’VÎ&t¥†<ᮇžk»c¿n]…^l"t}wRØë‰³„ùW9áWrÝmQõ"Œ· ¨/†¡«á«20ì­ÓAvõ=ŒxÐü‘(€Ñp±©»‘­0Vß›eïÊ5Åf§üar.?Pê¸;Lÿ`g›Í¾ ³“•¤Áð‡ó(æö§Ã|iR¯ ½,l½ûÊmëK†TµÖ¯ÃòûòÀô5X=à7wy,ÖSU xga£!ÔŒÈP¶æRãŽÔ²ÀßCv!™"ŽHpÔƒGk†Ý÷,o$ª“¥l˜CÆg[?Ò‘ôü!ó1$=rî¤È î»2F±"”r+I#ªùHÍÐ6Øè³…´äÜ¥2´ßžDrvæísd ÖèÅ©Md$8ñ9Ùm™z ã-Ï!3™œxæx#²0ùŽ+EVIú|bd3cÞX»b‰ì×8C¿Ï#GÊ ké'äxºFÿür‹¸RÓuL#O@¬ïåõ¯È;¸4x¡ý/ò+<±`tèÜË©°¿TL–ÜÕ%âƒPp|ؘ¨Ù…††(^ïÝŒ“#4)ß¡(‘räö·(&.©¸À¶ÇkÅ¡ú…S(²íúÒi÷× Wo=˜@ÉÑ7Â>½B©Â8æÇŸðàáø^á-”±ÔŒ¾Ñg‚²ø¹½í¸(Ê…Ýܪ •F…c;NuÅ™¨ØãáÖŸ¢ƒ*ÃöVß®¡ºëß+¶7‰P#îÁÆÊ«`<òjøƒ@ j'(íünA=ª‹ì MwÐàVïzlŸ¶S\’¦G“Ðë…ß_¡Ù£Í¥g™§Ñ|–¦îÒ=´4IMÈ´D«Ï2#Ò¬h³/¯?èM ÚjÏ_ØfD»ó‚Ò®_Ž }É@ÊI2tø4b™ˆ7Ñq:ËMóQ:)3 ÇîyR©ßÉí;††{ÛEùS¢Mâg‡lò~ä|øèþ›»°ëýÙ. :5Ö®1Ë 4Zrõ1…Bu’Lvbd TV¾ÏnÑeF‹‚ßDAÉ–PjWÃa(–r ~æ^_Il¢øt  ˜±HŽÁ Xå4ý8NAAá¬Dh‚#‘+)É…RCš %~E¨Hæ/ýî™5BoˆîIÑ@ýrNs4Ù hœeY¢zà ͮ|«†nÐrqÖ?$ñ ´Þ.¹wPã´oóŸÓ±øo?$zÄ4•mÙœ!×:¡M°|UÖdZSØ~žúì­Ï]Ÿ†ïÍ…Öéö -•rh3"xǧf²ÇyÈ…U~hŸ4ø\³:*®…²]ƒ®{M‘ËmnÐÃòä\”ô$ôJuÖ4îù´ÿLDGHF 2=· ²€¡ød.á¿D0r9Š—©ØÆngi}É΃‰RM—ŠWÙ0­;8@ÑsûWþþN‚yÒã«Q;}°y•°¡{–.¿¯Û+a85»a-ÏíÜ»«­°Éldf¹ ÛŸCR<é `7l4ÐŒ[ uÃ/wB¢)™S|IÃHòæB=*’ä–ɱ?Á}¾ýë§N ei”RÏ/XZŠZ ÝerR©$Rd y8»ÿM!2^S½æ¯Ø†L#÷'?ÉD–èÅN^#dÓ¸ú–}s9(ôHYç‘s+­’úb)ò+¾°ÞF¾ñRšg~Èÿ—A,šý; Zç_uVÏßó£MáSE¹JKS?D…bÆTÎ1PÂhÞ-éAy,ã°PFÙ‘ïúhPž11gì×OҸ½ƒŠ±v×åQ9)÷('¡#בQÍtt~øþ™Ò^.=AߪMò¿úa›¸Ñá,¼É²Y¹(ЦŸ¦J–D!ûkÊFë°Œ'h4䡭ωqîÄ>¨yqra-æ-”[·Ý‰ âj3ŽžÁÛPÈAšÍ4Ðy$ß–ÅE^Ãg%N.ï‰ðñãhkýçFÈÞ}ÓÜ“µÙôµIž‡ Û#ßéé[!ÈîsДb?Õ5 ?•šÀgý$Í%«ÿö)ð1RØç^_ çLŸX…¢¿zž—Õ¢ÿ[r¥W„0J•O õJµ@¯Ì“×^O¬š•s…rbÇç#_ïLÞÍXTpœWvƒ¥¯%¶yi°â¤+'yî¬I¢T¦o8lH^2 í2­û0ÿsÔþª­Eûñ«#A|Ž’)q i kÙ:ÝØó­e%ë ’í3Ð/=„ûhãïþ Š”ƒ6ˆ7M‘ú¶ç÷”êB¤;ĺNÒΉ jCïLªIð°¹ÜÖ$²h²h÷ÝÏA¶DÚÄ|âläT“ðÍzùyÄcÚŽj#¿t"£{U ü"•-÷C¡®ØãÜŽ(òGþÂÊ´8ŠÔÐüy<€ûó÷¾¸„Roö[<ÚóÝ}¶žý¥(g!×åÕo *«‡tOQ£â!ŠÓÇ´QYóÜà_oJT­Õ«yª+Œê!ÆÂ®ò5¨ f^ ¨ÙSÊ$;€ZO]2Þ°@Ó_—?_ïC=§ME>4p-’Òhx„Go )V-{ ñÇ„c¬÷Ðt$ižÀ¥ÍÙŸÄßÌ%C }E®ú¦´¼þýx.Z5ɪœ&»„6Â3[¹mÑöÖoÑ’ÂÊÿr¨{oü¥-:~`w×¾ØÏ¸–(Œ£sÄuaAÜÿåðC|]Žq6>²§{Ö2r¤4›|ìGi¦ƒ?'ÓõÒõ³Õ_+c¿Ý{þVI(üÔ±³?o–-Ó5t¿”¡19¦ [Ê¡îÅ­Iƒ¨M¤xþ£*éXT‡3¡œ¿ADïNã78cê¡´å¢ÉåÖP(K8%üL%*f:ÄïmC-±ÃÔƒ÷EÐØþ„‘ƒÑZ=L–Ÿ¹ß€Nz”OLƒž¤å¬r6ÐïPsé¥7 fšÌJ¦Áð÷©”Üs0›÷Hmäp6Ìý¥»&Ú! BÒï\i‡ù7w®›|¾ ñ£ýSI°ø)Øîðž¿—ßX[ЊÔW5¿“ô°NÃ/%êk+Ï2äãz`ãÖôUöÜë°~¯U aGIüÀRK>ìz³ï>¿ ˆËölòv—‘ÈððÇmƒ—HüU‡ãq<'’êÖ·–5ßE²%5‚ z+ÜWjmÅ})3Xs)؆º ñŒz÷y¤X÷A’¤R;¤»„ Kî‹ß¨‰»¢CjÞ™O $¶RÆ Ë'þÔþ(dý;ót»ÙMKžkeGŽTú–¼rÈE”Þò® ¹½ŸØ ‹ó O¯É‰åaÈ§Ó ÂÌž‹üÖ#ö»(Àþü³ô ! Lž÷ØmGÁÆi32ª‰²?è‘‹ÂýÂ¥±Î(J¹¹{zú*Ф¨sQ·[…£Ôr¦fÈ  Pq“55d0Wã<™•^=óej!i ûøußq«”ŸëFýŠ›!ê[kh¸­íÙW‚Æ4R†a§ ©Ð –qv4{óùÍm@ó²¢Ë3‘h1öziåØ.ZQ þðx·× ʆÑFévD¢-f—µ–¢Ó\䉃hoJèëçl'«|“ ¢ãƒ}b«&äx¬¸€PªœÃ<»¯¼c=‡Ö烈“mP¼.} Ô‘Ìúˆ‰½ø'¡÷Œ€ŽZK“Öðthö™bU)U‚:^b¾U2;¨ºNÊÀo9åt}.­À·äøë:¡P_x^Å‹òü½Ÿ(tA®ZmÂfmÈ öLmk€‚û“îc&Pr­™°µØ *š.Ô.ùíƒÚ‚…%ïØ¾=΋«‰®²B _Öá²zohÝM§cÉ–†ö‚K1Æ›–Ðémó Š*ºeø+´¿nCÏ€¥È¾wÅÐ{ÂÜá…$ô•ÓÙ?¡ê‚j…Za£uøõØžîÄñŒ%6>ãf Cw²64~ÂpéÊú(ŒD nº§’ÁhSùmñ—ÿøüÏ%˜P}ªT S_äÜ÷“ÁLùªsTè1˜{Õi•òæy\’H÷)À‚©Ï÷•|€Å§çźaiʸ¿êʽ S)¬n ‘ÞË-€õgï»°Ùë–P¢;TB%2j°ká>•‰ÝÊú–’uHD8‚‰us?üfäE’gÅ.´wt‘ts½ØvÉoÄ™\¶òC á—r¾’RH9ëGÞú ©ë¹Y"¼Ø‘öK¬þPŸ+Ò›ýäÿNàˆ z‘$*J—Q›ô“‘2é¦PþUCf AÒòÔ£Èâù”Ì—YŸð°ª8?@¶oÛ¦½È>Q½Ôu9Å+»5)"×%}·¾—ÈÝüÍ^3˜yužWxù}C¾Ñ句éÈß¾æ~ˆ/jþ2ªãy,Î&¿T‰BŸDFÊ8P8gÍKÝûŠT„@ÖŠŽEsUº 8mô(_xJhѶ‹U5àþ°¸™ÄÏ•x`üxhPcJ_ˆñž@éß‘½Âƒ(sPYfó³ÊŠï¦xN£ÜÅëè…¢‚†5Ufh*¦šÔ~+9ƒ*O³¹ß”¢ÚÌ¢xëA{Ô8z)Qêc>‘nÈ9ƒÚÝ^†£®Içê›Ó¨ÿôO¶Bq¶Ç”Y7 1QzŠHšze%Jÿl½eTVÑ®„tw—„ ¢ÂóÒÝ"%"¢"Ѝ¨`b!""‚AHwwHƒHw—t~ìócï3Æù~¾?Öœc¼c]󾯱æ\+t MæªÚžI»£™DªÁ=4w5Ðúð->d©¦CË´ó¯…ÝÐ*ÿ ¡ËO=<•¿ñ¥ïÞ1´.!ª)ˆ§Û<§Ù¢ØÑfS¬j¤SíÄ-(9(¢ïƒ*bÃh%žoyŽvyØRgeN'ÀÚLSbQùsh~á·ê õ7/ß^£õJZºÕ{PZôMíÖ@q¯H;>]ƒ"+ïŽ %È¿ß#û rÏD&¿?Ù¿Ãüg– [ýÊñˆçýß‘=Ëú1Ç96r#«Šê Ùÿ–B™AOß°ãTQ4—BmQVãYÑ%h¨.;çÇ M¦J’7ßC³‹PÔ'ƒhñš’JÌ*†Vo -™À;ÿ;o¸‘ýGb(Zß}øãxà,´¬¾ïþ>˜-ׄO,ÚWA‹iæÙꟕûžxŽkÓ\Z¶Òò­Ö@köÊG0¸ ma•'iLk¡=çâáËFG¡SøûIÅêЕðôf½1tÿb¹ä¨ä=ã#/v§ ÿC䜛L šn34Â0ßÞHM™<ŒŠ‹ï±4nÂøÙF ÕiW˜ºç÷U f"}{F#;`þã\öL—3,Þ4]}Xóþ)rÒnŠr¸ÝïT#X»g?Dú“6ùKZç `;@q1ïD,ì»ÿêË»< oòÎûh16ÛÏ|v@â'%:±[ÚxÐ4Ð0—ÂÉÔ ¾2žG '>®ÐLB¤ÊÎzÕï7Œ´jÌuÝ"ˆtë=¥º×Ó‘á¯ù§šÏqžaNÖR™ýÓ £*Ï!«¼ytPË2²í~n¿¼ƒ‹vMäõ/›¥E†Òð4ò^œ?wnù–øLˆd\P@+pð·d2 D‰ó© Ãe13’("±ëßeÉ€‡…çítÝCQÜègôBƒ$JÄôØNÔË¢4»î‰i”yJÚmX€²;仹?Þ£|rzÝX}*¾×ë/Œ<‹Jßâs‚]Àã6°ò7 U8'z(œ¨3n|áŽ|…j'<*¼–wQ=E8^½#|ßCþÝtBͱmæ±k†¨ýÃêhê^û:O§„ú& j†úwÑPÝêè_n4Ö¤½[÷7MBN'ÿ ½‚¦Ñ·ê£YΫö«“hÞK·À} -iˆe^ü2E+}®é1û~8Z:Xt­‡j25ӣЙS\ãW´~:³®öôŒÂ_¨Ê÷ýîÊùá~/µ ÷©þúùáï™Át|ÄØ{!wƒ¿\qóGÖ,‹¿µcœðïæڡü¯ÐªbÇ6¿M¹ÏÇ ¸<ƒ’j?¾´+šPÀÿ®¢È gßãŒòY²e!órã%H+ñ¼ã¿K¿d¼«ýìÖ %€P·ñ:3¤ˆOL>-¢‡äÚo« ¢}b³ 7©y=$Zd\Öµ§f~ê¿ãds{ZtƒÜ‚Þà•kÞ´Ë©,ú¿ó¦&(“ÿÌ‚¢O„4¿-¡8 †»Dóã~öd»y…JMÖô÷ÌB)ÇXLjm=”ÎåWN¼;e£#æ$í' <§0<ë­TêÎ}$o‚*¿‚^“Ũ1ë,£½µ›ßŸ‰¥B›JCm4 üfüÒGAM/â[/sC %3/}ƒÖhw“ƒéîÐni•{¥:¨+7Üà%H¿-“·B·˜ ÏòRÝ—%0¦ª*}¾†¤$éãçÕ`ø7Á—’ 0Ïú|ÃþŒó 3Ç\…ÉsUF–.˜î ‘¼Ñsš<äÂií°p‹ä~ÈUXœ™po†Nk}]1°¼|xž13 VS^Ê›pÃúçèí·Ùc°9ó,üÌÌ#Øyí•izù8h¸Læ;'ã1—æ‡ÔH”­]®öI" 3Þ˜ "éçtQwq$Ï7Zü"ŠT” †ùg‘&-|lÍî_@ÞûIdh´ò|("‡LCö¤ä-.È*'Ò:LJìE|D›½ýÈõ‘ôQóåÈ[ZíÅsù#:u?Rô¡ Ï3¦@.2zÅœú…1ŠŒ^+¤¹ƒbþ/þ|}ét—¢QÚ1ôy0ç7”•ʸ~ºüÊsö,?AE–™âŸ)Þ¨Äuç•§8Ïø§N«•†*7LM~DU#¹Î¡yWT[¹lº<ýOæÞ¶åy…šQ9•tPû Õò¤Ï¨›º°Auÿ*êwÅ\^Ðt@£ƒ oÛj»Ñ„ÙÚ)/>M!Fj¥Í½¢ÞžAóP/ñ3såhñ÷œE3ZÑ,Ì›ÀSÏ¢âüyRÿÇ¡¿¡|J6Úþ¸–ó,qíY¼’´îû!¡WKÅÖÿËa¢÷m­ 7 ö™<:frm…‡«ÅÆ-ðˆ‰W ’Å õÄuÃ(Sx¹º—ü¡#MÛùÍBþo©¼¡ÞG#rî×?_t*\û½ê ÌÙ=9a©mŸƒ9¥³_¡ØxkÔ~1жttíÿBñO‰,C§F(Ý|÷B`æ2T)œÑdR¡€m]-9ùoТJFÍÉ í›G޾» Š/sL’9CoXbŒÇí‡0ÐÁõ›Ùb†©,¼¾‹†Q£÷ B)*0.ØWú:Ü&P}ï {Ld˜¥¿Ï#¾LÖ1ã0)iMÀP“ib×bÀÔÓö°³KK0ýÅ*ÆáÌ"ÌüñM½<sœ vê×÷óªƒÁ|ý;,$ñÅ\õ§ÅïDnqÁRÞI¹ø7AþÊþÓ XQˆµß„Õ7?-Ñõ¬J’õ1“ÃÆëPâF [ºï«VßÃÎñ¾gae°çrýd —-Œ­>•R”CBÙš…ÒÂHôþñ©’°¤¾&VOƃ™Îf¿¾"Ùí[ 3NzHaýì«ÚÒ2RN§ˆ%Fš{Âg»7ä‘ξtö‰ÙÒåo“=àB†j£5Í—ÈDžyrÆõ2vdI#Ë;óD6bd"÷HÛ÷CŒ-1EŽØK÷‘‹Î¼ŸDI¹îˆòÚBž%§4ýã¿‘OîŠÈüùä[>_Ç ¯‚üÕÌ'ÖåP ó%½ÝÏ¿ªõ§3ŒxhÌ9¿ °èC¥ÏÃ_Päέ(Ìx-žUu:<¼C"䯇âÚtïéá‘_Gn™8¹£¤ÐíˇDÖQêU˜Þ%ºJ”¾}Pߢe„/½Ö´ D™~§/§ÆÞ l–±rÊÕd”K¾ñD^-å«4±XÍO¶2¦ÞpÇ£{ÛÛÂxœF…¬¨QÙÞ€ò+"Íߨä_/QuCXDô ª=*œÌµ ó¨Å"ac´…:lŸÔçŒ-POfÓÛ™U ¨ ç‘£átĨÿ„2·±ß:¬Fˆ&ãÙG<¾£é¯[¸?šM¯cs#Fóq¶½ge´è$Š2šDËÂ%ý—\žh{(Àâ}<ž *«ä¾ÒÖ>ÛOäÌñôC¦­]ëD´yýp±ð6ú=­è%ÛuA‹äxåcGQÄÆëð®OO~±M > ƒùuR“ãTÐÖH4MN÷~‹]”/zLU¥Õ_»f¡ü¦mùOG(‘ú(Xrƒ Žë=T9,ª7É6Ë÷óíï‘«O ãØúhw«=dT‰æh'î÷Ô=-å¾Pл›i'¶Ÿ[EÁ+hUÎÏ¿úÓ õäé„”Þ÷¡I^†aýÒ3hÞø²v-It¿o.NMnB»’äÛµÍ8èS‹ú±â]÷»tÓ÷=qçYl—"ü=.›zùsô܈¢°S€Þ1rù‘‹Ð﹡{Nè Dߘ!§€Á¢}„‘a(&œRµè8 ·Üð°Z›„QÉOÁQË0ÖX¢·A- R.ØùÁÔ'ÑSBA¦0“²à¶ä> sËJÍ骰ðE½ï7;!,öü^Û°,…Â5¹ua°ì'¤šž«¬ìÍvFÏ`­#ß­/µ6j8buÔí`k`nÝì7ìRï„uõ'!èè¡N<`tPÖÂL ½º¨îæ#Qî‹/œËÒHBŸL-¹¶‹Ÿ3÷¤•!‹z))÷$¯œ¶T—>‚”/8]çjn µSwÖ̵WH«”LÌyk\‹Š·J}î|ÇÔs)ï± ²çh·Ao âjJPÝË>T«á™Ö«/À“Q 'x¤Pó²Ä]ÔŽ#ºÐ9€ºßΦÓ=x†ú þá~Š¢aÞ¥ÓrÍDh7h9*ÿé)Û~Ÿ¤’&£q„ÍEÏT·zhIÜ»#ª­nç´c2¡MW½ÙŠÊÚÏRý1yº¥Vî0C—š§‘?t+ZQü“f‚—cZ·™ú©øƒ¤œsaÄœÂóI5 5º&øÔPÀHË÷ÜÛ³Ê0Nd+ZxìLz³ð7›ÌÃŒü‘²zc ˜—`}š”A ‹JâÓ ¹Š°´ö9¾çŸ,ï$Ö>ã+Ûÿÿm¬.«ÃFÒšœù4lŸ™P¦¼°{ȳF{°f±@œç9»‘Ø™¡JwÕŠœéVÛ&E2š<ó²Ô HqäÂHvêáƒÆË“HK{èoâ]¤»¨ò¬ÄÏ[(!cÿ»Ô"Özdöb¸|õ.²ŠJ~&ˆŸA¶ë‘9G)cœ¶‡TĹɯHid"¯U*MŽœ?òIݹ®G€GÏè«ò~EÁm:Ï4PH£Z«¢×EhDš1dàáéèÒSQ\0$–|Ú%n„»ׂR»_ucO Œ³òI‘¢«(;~Sÿ¥ÊÇÈßs7’GÅwî¿B¥ÄÏÄ{ŽxüOÜÃ;L9¨Bÿ~tRhñk¿)·ÆªÉ¹G˜x/¢ú÷rÃOööýðøó‹öŸPsñI†½j«ŒytuƒOÿy‚úž?µ&¼ÑÐÞ}‹/‹µ=àC“˜Œlå¥4MJ%` E³2ºæ¹ŒP4Qù²Àq-*‡G¸¢•ák¼ÍǧÞ:1pʱ¢õÐtVdÚ(Sj&d m”$qUÖ2ÚÓQRœíìC‡ çBŠ µû½T“PÂPçÿòÃúÔÊ\|D_Lq߃nÚsŒú”"KÃgqÖUXZÎR>Æ-‘¶çº®<„ZYß‘CÂP¯ašºYÎJYÕ¶iûëAAüÁÝûÐü΄Ò‰¡U %úDA´5t\fÕÚç6¨ËÞº*™)Â_éÎ\–»ÆÐ;XÁÄÀ2µò= Ö;z  _2ÀÕ/•0zl­Z»0ÆÙýOO ä¼ ÍîWS˜v .´ß©„Ù2¶çìaA¦ì¥»Í'XŒ xyKþQHŸßøHËÔ ›{ý°*߯Ì'–ë ͉QáÞ°iXrdù'lîtÉT„½¶cJæ’xÀÿÇ™ï\Hd0Wç@$ ò*#tH*]SMZÿÉu¹§Š òÍÜ©¿ÈFVèéÆP ÒÅË5 xô ƒ—ûdzÈô$¶æ(;'²TZøÏw½Bvu–7Z¼üÈEí‘ù;ˆyEM¢ËN0]a,¯Ýrf¶žPùDˆÃÜ1± ¯“Óaþ×(iž³°™_l‹¿x„þ …¥²Ê¯üÚðoþ:×ùIJXQ Ò ý«Æ¦Éa¦ÊÙ?øl|f#qå-‹ì¹7N°!ýʰgWåWUó þ¼]g\FBNkÉ)–$ºY<Hl…ÄcaÆZéxðnP¬¤Å:’I†Kº Ñr¨W#RnTå>¹öih,Œ=g»Ž—€AÈaéOú¾åQ>‚ 7»ÙsÃË‘1Aoü`m2¯¿q7FñÊâ*%d½÷Ð'(„Ù:•¾e^CÎðÆ{È™Âì<¦ãÜ ‹úr×'rSí±ÝòQ=-ÚrOB¾&ÙÁ•Läÿ¤ÚÂù"Gˆk,êQ0mó}·»;š¿Y±4…ÂZ$î=(òMƒèZŠ®3éN{îs&WáÆ”â÷®†=urÁ#cq½õ×­QÒJ5¤ôÚk”Ê)˜þÒÚ„Ò¾[)š4U(#&Â"€‰(Ó¢÷ó1ÊFv{^Ÿ@9ÿ*§`]”_™‰4 AE>>r‰ <:@kÑÑ^‹Ç&nOESEeiT¹Ç…ð³²Lô2ªþâ¢ùHÿÕt䪾ìÁ“26Ôq€š¾àõ÷ùEÔ¾’C“‚º²\ìLcPŸúêÛÂ~_4HË´5åˆFŽ51iþhB½’6²SŒ&såÃ] hÚH“»àŒfßdøy¾¾CsO ‚fí“hr]œ.GÐ’HA„­^-k}YÊØ U¸Ü+Ûóìxʃ¿éà93´¶ ÈÈíD¿Oiê—ÇÑ|¡ÝÓx˜…¦uoy~4F"oâ‹ÛK0ðÞÝ,Á)Z[Óýðý hpÝU¡|’¼ÏO°HCÂ#(s½Ls‰ì/›l¥ÝÏ-/Ξ6‡ýþi»Òõü)dØk]dÝ•†ô£‡¿ñ„ôÄæƒCIÁ¹ÁhQ|üä«K¸Ž†=€’TN‹5²¨¤ÕèLekÚ+ñÀôt_Ÿos —‡¦Ÿúkâöû¨;<Û,…6‚ƒÅ&ß@{„“7­!trÚ»™”)ï{âtJÏ7UøóËŠ\SκÿPšñ]¾=Çsðàù~èÍ™^ûçwú=øê bnÂ@’~%•õ &›squ\…¡ÍA±*19ìo"Q £Eš©C¾0þÈe;ö×K˜¼mcœñF ¦ïŸ$z¥L ³I¿ÎØ/°Âüÿ É¿ X–¥­“J„%gµë¿ð/}ÛÕü–¬Èš}’3…ÕžS÷ F÷ó*³õ{íÞCØÌÌ•‰ÕÜ…íšPÂCu°ûïáÃ*¼º^y…Ùüܤ´CÂûñÃÎH,zÞžqIžöj1îçé¤éZv$‹¾M{ÿE9R˜p ×#§ða1Rw±Ÿª¶QBÚ„f­’ *¤«µ±ø«²ŒôU¯~GLm!Ão/­Ç“ŽÈ8ºDÀ¬ŽÌ,FGöE¯h˜a£YUˆØðv![pxôrP±Sµ§ 罃|=ÁÙû3ôü|. OGÕ{æXäs½ÔÉKçüî2ùÎŒPà²þ…Ь0´L£H»‡4ø_œñF!½Wל]lQØñR½¹ÁŠø‚junŠÈåÁ»#¢‘Ÿ&¾â¶›aלÄ#‘i¾uF(i^èñ÷PÔhÎÈD­LVÞ‹PÇKôù™8UÔš;Å|4 d8FɆ·Û-6Õиáõ³§‡/¢IǽŸuhš—¹0C7ˆf¯ŒË&î  ù©ÆÀQ¶hÁ6,iðí3Züµ ^UDËï$»•¥Ihõè Ô¤ ž:äZB¾.¶”ËïÐR­^¯†¹ˆ7ãïW,ÂêëWŸ‚-W¡Mõiæµ'LP›N‘4HzÊ'š:šY@IõÓäìåP´%ìYøòøä‹µ[ÊAçÀ«SJÉÿ9ïËú:x{ß cœ2$ 3|øÈT…Ã>Ä'ÇÊBަ–ÆébB(<±GèÎp J#,9éõCåY7γö’PkYV³v¥óhÿ`þÏû ­ï±€f£÷3õqÌÐr ½¾}eƒÖ;¯sYËþç‡Ô/ζYC«cØ£#§¯ì÷jµ/a¤G yO}8ÅšËÏîj_Ù‚æ>M}Í ÅÞ{ðÐåbhåm¿k²qZ7ŠÎ|ª¢„v²Ôç yiÐá`û-7u:;/><®ð'÷]¢¯H üy±;_õú¼c/½¯€ìxB^­xº×~C¡qF¼„ný¡ô±Ø ê¾½0Éž’·íq¦kZâí©À\¡ªÉÕ×¹°°CÒ?ßK¹JÇ¿„Ãra‘ÒñWJ°FdU[Ëv6¼êc[ÇF`[ჯmˆ ìq$fè=§Á´}]ɵÏ0:ÃÃ7̉5œ5çÏ"ÉŽ¼KÍl)’þ;à{ Éìô>X®#•Žåº>1ÒôÊ–“ICºß^Wý(¾!ƒÆ—¿aNÓÈØÄvK¿Û™¯Ð (²¼DV.òÇ£È6iz99zJ'ˆ‹"×6¹®¹2òjÜR¤%¢C¾nˆêu GQÙôTºH %;?Y‹BâBçNÚÞDá9Ó'uMÖ(Ús'®s­ÅÖ½¾ ’?B u –9Ÿ›(U›“tº6 eôª¼oÉð lsO˜3¸¢|HºØÖï T|Ý?|¨•Rßš\w¶ÆãJ6 »Q…Tk¾j²1LkZÕįÞ&Aõ/¤OeœQCeòA¯vjnSZ²éBíVn2Å?¨›DÁ÷gyõ#ÜŒ<ž=AÃwõ‰‚Ñøµ–Ô‰‘'h’J_ÜÔ'„¦9—mþ¡Yý 4Ÿv=UšSŠ–¬þüß|Ÿ • É[Òº }?ì;kñuz¿—# ’B›ãå¢)µåhÙ)Ú~;íi çøb6Ð!pm6îã~/UØìnžþ¿ü°Æóè¿"|D“¤2·…†­,Õw­‘yxÒ«²¤–Œ%Ì3‚öï_›C+’ŒPáÉØ©au вçÑÓUB^?w€Ð+Èhãr…Œc³MìàW‡”lNH?¤JþJ'H2Ø÷BïÛ×"á§ ¬Õ›†WTÁrõÑ üL±V*x,þŸ}§á®ï³áWá\ë¥ÿŽ“U_´¹¨[9_ìŠØÎ@žœË/µæ¬ÿÎ[@A¿ÓJÎ+.qtzöûïà<¡øò¾'r<ºÅ>ÅÿØ*B •òÊÈ/(iÓ/c4%…Ò´±’‹?¡ìÞ†­Ò ÀþºÒÞtýšTÊ:«J…Þƒªnbߺ§PãzþBÙƒp¨»~>Tlü048 Ïü± ߟÿšÝÌ„fÞ@¢;·ÎBK¬ÜÒõ*vhÓ²Wðàx ‡ƒÜKÎB—T[§¹z7tWì¾[ €ž½g&%ô¤Ð_Ý¡•Ó”ƒ—>’¨wÁ0‘¿®¯ŒÔt å­Üçv±©X0&² z§jaª÷•å ÿ˜UcÔ:h ó½DÁ2°hÜÛ•_ø –R/Îr¼ýËü‹‡tIa%Gêäi Xó=`—tŸ6’f/mwÃöñ‡«däa7ö“¤Á›žmv›AÂ)ecž–§H\÷#xòÖ.,é¿y•ÉÚg†#–ÿ %Gò§ßR:%±ÎãR9Ò9öŒ&¸€ ÂA%ªéÐL ó™&d9}ûeé²UPó½v<‡œ>ƒÓ ‰Èó¢«ÏPã:ò­—µ\ùø~Ÿ~ØË鄇š»-ùZ#Q„«D¸œ4 \¤?ù`üÂ¥œ”¦<º¼¶s e¦rcäÞÔ£ÜP…¬'íuTèNs׿XÄ£-/æ=¾‚Ç_ë%ÕnB•üùX T½l7Ö8•³ß?K~{í£í}ŠJÔ¢;}‘ ™u$¤+½£PÏÂæÕ—>I4xÕ¡1n„FINFä‹’l”šº‡þºef¯‰§kµÐ<ßöðüÁ´X©zi¢l‹V'îR¿ß¢ÂS$_ØÆþÇ!qœ?ë…¶±<Τ ThÏðIÇâc:¼P9EØ—óÿr˜¨Òâž>ŒÁ Þ´bRh=h›5l¢Ž¢w’‹ž: ‰$ÙâRd¿V2 q.gº¬ƒà·"«å91¨•´1¸¥•isQŽ dP¶˜ì½©JiEš‚ù¡XCÞ!tØ z—VÚ¼Œ ?‘1Já÷](x+ÇçÒ”°±X™5©BŇ Û‚g¾Bý‘¬Ÿîìç”0Ð[Cûœ³èø³eœþú˜nò½ƒBÙA'ã`8PŒ/È߯èOÿòJ… ¶þybÇX˜TŸ ÐS†Éw9óÍÿtaòM^ôM"˜L91+ù¥¦ÈÙo‚©ŠßªŸVZaz\»RF“ f¹xÖ_„9«¬¾*® ˜¯–¤Òà€…b빟RGa1/Oà#+-,ÕM60Ùt¿ՎïWåŸÃÊÉE ©'°·Ptn™ Ö9‹åH/ÁF’d=C§léf–X5…u}¥A›?°gÎS×ï‡õ2‹HH4æ6°s ‰ŒwËI³n"q¦õsá}”ÓWš×ôt4Óöíæ#$O>ç›öƒ )?ü='|B©£«0 ÝGÚb¯+Áí­H·i"&·¥‚ ¼ç• æ7‘Ñ ?€_¬™;ɺõ_EæÒcí|ŒÈJï¸{‹Ù®Ö¿];ñÙ;'_8‰œÆSUÖQÈUo“ ÓÃÈ£ÿúwŽòÖ)¥Ðl‘!_ÂÞú~!ä÷ÿ(‘(àçXK²}ã%ŽJ­þÆCë•y”—Qø\@uûŠt0¸|<Ð…‡¥N˜_ÿÉ„b®Ï4ˆ^K¡xrý9BÝ'(ÁÉ«¥ã­Š’~ä…ŽYb(54ÔÚÒï¬l :e$FÂ÷7 LMÅñÙÛ(ûÈàŽòM”sâR›0=ŒòW¿~¸oÛ ÛW¦'Ùñh±É÷”ä«x¬4?”«ðÄòɲ&^£ÔÑhÁgû|Ò¯R{¡ZƒX„øŽ&ž4ÔŽ4‰|«Æ/²Ê Vi[¡Ñ*ê|ò0²bB=ê—J6 Qÿ^ÏÕ™§h0{· ~Ñ¢ÑEÛ3'¨SÐxLÞ&Sü(š¤ˆPÙ{”¡é]uƒ‰y4cŽ=k|ñšÜrþ–<„æNše½1hA¦sárPZ¤»~Ê @Ë‹’±4WVÑŠÿÆÃ»YGÑ/3ê[cãi4—%Hqy´…‡î³Ïev!!ÃyO£Ž¯Ð7"9õ'ŸZþœôäñº„ÍSÏL ‚Êá×!#ù{}RJ?ä~µã½vŠcÜ9Ö›»¡šÆýŠ¿óÁÍ´:ÿ=)hA³‰û,r½ÕhþºÁÿSë:ZhÜž~x-¶¿‡Xz ¯MÝAÿj6´¨oÓ¼/ʃ[¼ä÷™`exáÝJd"´¾K«Öµ…š6¢ËPn™2̧òJ þ¹<=X Eœ¯{úÌ” ?až í‰4ä^v*RoûÙî·õ_9Œ@fááyîÐÛQÅ#LÿÃ2©3×TL¼ ûôʲá~ÞÜ\±çJª„R úc¢ªµPÉ—,øƒujùjªJ_™í÷ÁºôS Ð$„‚²Ð¬ÞÌ0bzZ]ù„¸5 õ–|–·ýäÿü°ÿèø½hÕbmgxvZn°øÓ~˜æ*SÇàÓyÐü «Þ·!š£iÄx?wB õ®ž°°2´´j^´16€Öø³£ -ÅêÇÒ hßT8½ýµ:¯Øøº›ÛÀh7bç}†².zM@ï¢pUFm ¸lT“ÁÐaU¥é¶,á‘9~¸Æ´<³ÛöùÍз #‡é3w^Hv‡9ãK\RW]`¡¤¿íóG_XºßT{“z –}šãæ÷~Âj•9ЦÀ†v ‹ùP lSê5%gÀîªFU$áO$؉ÛÔy;„ÏÍCš‘øPFùÛ™«HÒJöÏ´ÿ>’~ Kæ}ɳ¼Î†Ç-#…îÓÄwH“ ÎÛŸ9‹t¹5qQÈ Çeãþ ‹?*uEL#ó%(üЬ´#&lz‘­/^É®A9Zn-èYE®¹ØöÛr×Wö˜½sËäk4¬Ñ»'ìy¼CÁÇÊOL MPˆþ±OñŸPn¼w@óÅŠf©è}>>Žb Žß•EâQ‚vÙý4J…6ÐÓ?&A³œÒ˜F”Í×=FÖ†ò~rݾ&#¨øRÌsƒx•2ý>ª˜àñ>e¢û¾[¨¼½²ñ5)ñyv×k_T¼¬»õ¦Õ#îhê«Ù÷ðf²Œ¨uðÍrê1:Ôž Э“•FݺÜó¿/¢~©˜%™€?ùi³£qÖÙçÿ&ÆÑ¤èJsÍ#4­àc%‹6A³öþt²‹h¾ôžòÝiV´äÒcO¿VfwÇ•F\| #§³5F¶Ã˜´­ö;g˜PÖMl¯8 Sö 3Œ 0SD=Ù“üæ­|\­.ˆÂ"QZÓ™,YG{œÚïÉÙíB5o¤aEÅGïu¬nÈPåÚ'ÁÝéÐy’ƒ°õ’˜yØ@v/Ó’Ó4!Aƒä½M®|$ô{ÞRϱ‹ÄN<[ïHgð Ù ©í+$sÊëñúô )b®ã§SH-%O?·°t¬Ï¶Ü—¾åÛØ… ÈX)oøúë3dž wIº†l:D?{¢¿"Ç’1_]}7rÿ;Ë%Øyù¾æ²tŒ° Àí“Y9éðÐüè§ž(œ}éOÆSm<|ª]‚n?ç˜ò”Js¿A©zËÓ=9(S­r‚ïî2ÊU¤÷HGŒ¡B~hÍ[q<šö™Ð?{¯¤>Ñ9ƒ*»„yÙPõîâU GT×¹B¤²¨¡ÔÝæ³®‹Z:JÁ×\_ Î õùkã|¨÷ÒîÚK4è᣼¦¤…ÆëJÓãõhâ3+0Ç)€¦/Õ“Žg°¡ÙÙè›7Ñü·þÅ7shI¸‘°´¾‹V'é? ÿ„§^dóêÝþ‡Ê#wrþ¢mô­]úp´§ûùÎ÷b8:ÙTpL³ÿ¿&r?$=òƒ—=ùß3à©Â÷ûb˜ÂáÏ(³%]è†×=©ðÇ0PÙUñ(þ8´_gÎ2%邆„p'âpF¨¡Ò?ºeu *ò Äûøº LáŒý¹{£Pâüx`OPŠô~;¥8@þý…@ÇŠÈ#¾«¿é¯y„}‹ù¢P,Ü‘;U*åun§â–¡îóÕ¡G% ¹AÎêµÞ*´O<ä+ó€n¶ßêg^BŸ™Ìö‹e4ôuín<õ!ÿ¼ïlÀØÜ81GW˜xëøL¥¬&óùœë|`ŠdÖ@V²&';t¸lé`rQ–QsæLÉ}N÷J'„©1¦¼5 ˜!WµæÈz ³x»‡sû+ÌyhZnÌ÷YçM\j…š¦FÂzX¬°æ“º÷–Ú¨–I¯:Á¿]yχd°¢oÝÃTm«É¯—=•Â`]d“³÷áØÈÓÜ7ÓjØrÝ:Q ;ZÞ<ë=]OÚ5ÁZ$(l¬ø} ÌxÄw—!ÑåoO]_#±ïùººBH²×«}ëC’Ft¤]¨Œ@r›Ï™§RHylž²æ`9R˧ˆýìäGZõåÌ–Ùz¤Ke/ÑÍ5DúÚmM)d˜ó8ÐL\ ŽÚ†/Ù:q2–3 Y¾©]ηÚ÷ÃŽ>g‹‘ýFÖsé 䘻þýb=rÝ ÷‘=…Üë‡í¢Ù‘÷Á…† äók ÍbCþ‹ñ±Nœ(p­÷¥úDÁ¨3ÏîA¡¥¹Ÿ}wQøíˆÆïC(J‘±÷ ê!Ò^Ÿ¡9T»âíÍꚌêíöCoßÿE Ç¼åØ Ôb]¸·§{ µ[e *}Q—:¹gÕÊõNnåÎ}G}¯Ã;5®¡h^g•È;€†óRâ±_õÐøÐýé½% 49{š¤/—4<_CSåÌE¹04-±rŒVàÙï§Ù vhÖ@_ýÍíNÜa¹æ³¶*iïÑïw0µL¦šÝŸzÀ׃‚TT.-ŽHðµòITžô|¸'m—F ÍýÌE‰qîP3#føÅwÊRÙ¥âLâ¡Äº™2µ y?—òÏ.p‚ì¥Í*ç‚!ãá›­Ûß!M•KeƒÛ?ïžg_ û¿îðÇpù™Bº h|¡‚Š€Öõ›‘P$ò¢dEQÊœßç-†VC•Bh¦ðëD¨¾"£ôÂ~ :né&ñk£[ûýôeæWUû9D«ô'†Ú¬/¤…õçB{¨/©¦tÄDd¯ÛÔAgÜ`×óèa¿úãbtc³%“9üíxzÔ¿LzàG+èp/žü' ôÎä¶œO‚!^ÒËkG`8ÊÈ[ì…6Œ~|ÇëðÆÅ UÛ#ea’aBº®É¦ùâ¼RWÀ¬Šh¦g8Ì»Ïx÷VîÁB¿,[ȶ?,Ñżó€Ëå›™÷{çø‹°Ã†gaõfjÓÝ*Xw2‰Õ±-‚MÕª¯ɤ°­ÐyK¢ÞvÅ{UŽr!«Šû…”p< ¾{Mãä,Z6o÷¦Q!Qˆ«O+^Ï_"ÁƒVÁOuV‘´ƒý2ÕàC$w ¡:̃”4ÞWÖÓr‘ªúñ3zN¤‰v úuFét¼BN„]Czƒ¿“bžÈp)¼÷™¸2†«ø©nÝ@¦Ž'žu¬ÅȰéW|ñ$²šúÛ6?A¶¸Élb-gä2œ¸ýÉ9»Ž;ì"wë¥çÖ‘È»—Â@øù9λ–A¶±Æ/Ûn(°~ÌÀÂ;tÆB.:ã¡&¾Ó\í(ôWåþ@¿7 ïŒ~rŸ6AÑ£Ö1–çíñ°÷Co Ek~侮Ÿ“ªúÕbB(‘½ký÷Ð-”!…Q1”Îû¿öfeÙ]ttÆQ®!­L-~†œOòn¢’¹>wè1<ÁZ<—?‘‰ÀÂpƒ‹V UŸoè‘•æ¢ÚÊ[2–Q;·MVb´º*lÐôÞ;Qu4#|?›=„fïËK,ûÐטÿÉÇ´8Õ£*%-‰ì‹'/TË*ÁŠ¢I°Ýø9h™ê?EÕË?†k÷œ¡¬ôÕ|±è7;q;ü+‚Â;ê»LÛÿyžÐÑš*9¹#s’×cÿóÜÁç¥dý¾&çuÝ.ô§y•^XC†À¥øøíRȺ«Z-»ï•ÓYÕº\ÿç½ø¨ÈýKâ - fúwÅ‹_UÐpâî¡ø$:hbñ×r ‡æ£J[;NÐr*¤"&ä:´ÞPïU¥ýŸv8©½«ä‡Vñ÷¢ë©w¡EsÚÍïHâ>·gJÙˆ¨ö¯§¯”f£ˆ'9vÐ\ø Md Z|3N‹ …V›ìûÊúÐvjs’à°ô>×´Tð1:‰ ‚¾Û݆®2¦ùDVèl lslÍÒÂnsÏŸÅûFH0}…sÐ^ oЈò½6DbÒ“Jã÷‘ä{¯ÏŸæ£HêAj×pɽÃî|\™GÊÎ&?za¤¹k2¤,Št_Ù8ÊcÑ?¤*ñ¡ 2¦roŸ?ã…̦nju.ùÈJTôÞN¶ÙÚˆH$ä£Ztoïhr ªžý(V¼¼;Oscø‘¯Ôn1ëío ÿã÷f)¯^ÊÜêÄCkýp*…ÆzéBÑ7æ]¼ XŒV¼¥f+¬k^–MG)§šc®(ÃÚ»BÇä²_“e©[UPþnZ]Ë T òØÍáB¥¼{ãf,:x|ÀOc·˜•—∞¹"Þ²å:Ú‰jœùš¹jPýÝDSEä¾j»Tq£=Gˆžðjož ˆGÝ)×øò??P|ÂöcQ0Nžä»Ìê‹Æ½÷GfõѤ¡¦y=½M›:c¸Æ“Ь_k;j1Í7ÊŠ.¬_AK~*ú› ´²˜;~XàÚ¾vìU‡ õ@,Í’¶Ú()ÌåU¢mÄáë «‹hOµ”yLÚ½Ÿ>úÏ÷-XgþJý_~XºçSVƒ(/,ô¡Á›A·§ÈDçÍ^úɉ Øÿ¼M%ÖÒ›ŽœPN.{]@lf?·¬ã{/Anpâ-‹Ýÿœ¯¿1#Àý Òɨ‹ûBêÔøÌ”x¤ˆ}¾»I.ë¶êƒô0÷¦›‹kŠžŽ1&¨@b|…Ý^öHv 8Ãc©±râÎÿ'ó—“OÐ ZÈöß>óì„ ä¸Ä×ßøßyóœîEœ»ù†u'ä”OAÁE™˜Gõ¿ö=Ñ·‚/9 ×59ã~mÿgŸÁáZ'(Îé.ª¬ÿœK”ªó£‚RçÌ•¢ƒ2¿¼AͶ6(ï¥øBsâ2TžõÔö ‚ªnkšCMõ ‡@ e¨KPVhK> mOºËB“ ÝZ+)ÄdEGßì"URžõ¥; HáÀï!¹…ôÞe”émvÈxá>máìqdö)8ò}²Önªˆ M ‡s¾eÍ,"·õäéM×óÈg‘b«zÄx%f¥8òñcäìû [0}#DÑ%u>–#(>¿ó]ä‘J]þU\Vè‡2qEmÔg¹Pî{ì³i1{Tˆôâïvú1zœŒÇÝ¢Å÷tæPå3) ë‡Tõ¶’;FŠêÎl†èQÃéX¤â´j=dMïÚ@,K¶]ã(Ô[-=sh(ûYõÃs@c‹àkÿ"¸Ðä‹õwÿÐ4q.eb¹ÍJ‚²Oq}Fó!eQož-´¤¯~—VzßÖÜKá©W®Q?~:ÿÃã^ã¡ *hÙ@ 'Šö4ECÙ þèxGNvíÎÿ‡;±¶ø’Ó¸\‰h­>¶~û˜#…‡òmNÍкáÌâ¼ZèK–V|\ÓmÇ+hjÒA=?÷­¦m¨]gŽ?åY¶¡ß¡ô.­¬•ðÜþýÞ»x]î R­¿§<Î9¾Yú›c×·V¤Æ9-¯ˆi¬ ð²$’\< e;×Ód«Aƒ0ËÇ›ýûzæÝÓhŠ`¹ýº57å?Aß+µ³mï®Ãà- ]÷Uk!ß«¼~É ÆÚ8ßw|<“Ôƒœ·…§`êSW)Lyò}ãêSª‹Äd'y`JVô˜½üL9¸?tN„éƒÕÖ3`F\iŠXŒfÏÞ89udæ^·4Ȭ=†ù姺å/a¡õõñÙŽ³°Ø¸Â]6@K=Ò›fa™äK±w/¬˜ŸÜ>+«9r‡ýÁºüÛÛ~Õ°QcwÚYݶ<¹œÎóPÂŽnÖŒ˜ãìaƒMÓ¤º)ñ$)âv‰ÆZC$"ó <Àý‰­ h( ¤Œäû‰9$5_£<ñ& ɪ]:s)º Ó/·œFªª>6½º ¤©¸FÀJR‰t†­ÏÜu@zÏ{_„;#Ã+²&Ë©dÌR æÏKD¦yÞ]QAdQál;ÊŒ¬Ÿ–]ë5k‘úÙÃê†äxyzõùuäbÍÖ¶«EîØâ‘2Yä•zO‡bçï\¸ðç5*ä×ùÿØ{˨(߯íŸéî’n$Ý›înQDÁ@DÄ@°1,DJº”éFºþó}Öúù}^üï·Ïýæ÷j¯5kæ¼fÍœÇuŸÙsí«ˆâVÄ1rv±¿bz…Ü9÷ƒ8E©§ÔZ©PìÙ}Þ /V(®”)½`Aމ7¼‚ø8Q²{×Òàcf”渾¶Í^Š2טص3 ìÄSÆDÃ(ÏõÐ;þÊçÐd(j…¡‚Ê¡W…Ûô¨+¡Õ늊f&¾\¸‡J,µDOݳP™bIíü1TÉ£¬d8T‰ª¡íÖGç£úƒYe?Ô¼\Ç´)‚ûb£$ïÜ»„X´ucä!7êе¹“ts ®k€X×mIÔë:Ó¦X‰§hÕ3ŸÖ å/>¢u4¦k »ÃøMh¹ìçД¼GÊ^˜͈5-yÐl]â÷]Áƒh>ñî£þ Z4-QÙþÐCË·ÑúŸ£Õ…ù­Ðȧh-àõçèñ´¾ú•²è]/ZOtúlú µ±9ßU´)Ò %Î\ÅëAþ÷ž™¡M®“ÈÝå(˜XàÔ?;Ûç­Gqz,?Ÿû™F-#µ†¥=á‹t…Ã\TèKûxX ~~ µ2?Þƒ\ÆÃ4Ì0ß6ø Œ(å ÌM1²yõ£¯öÀx¿Æçã`²‡ŸòÀǧ0=²j~\M~íÊÊ»ÔóØ|èBbŸ(óª§ Hò˜O)\~I{ûˆ,‡/!™H¥Ä"ù#Ž S²÷°•Z"UjÖÃÀ%¤qü!¢Vk‚t{rÊå­Wq7cÿ,k¿-2Ü yóÒŒSyÚòC‘©×àeøÐ;dá9ªýF,YÜW]йˆlOŠ@IOÙÛèžÄ³!ç¥3"¹È­Õªx )y­¦©Î=EþûRä[K¾(ðkG3é Îå†K>ëD¡º±@Î>Nr5sB‘;–O×™P4áÞ·µ(VTdV{5÷,ŸdøªƒZê/z.3¡dŒÛgËI”Z~aÔ6‰2'Ÿ™þqAÙU¦ÝÌ(ÿܬ³ái%*’z¸>FTʦ9#@Š*9Ý "j|¨FtqyŒà›/r“UÒýsßàòõÔæx<ñ=êxK*ä]žAݤ6'i’IÔ›´5Ü+<Žú‹ÁT<~h°’¼º% W­÷@£õöêk4^ÿVôeU MÒü¿&¡iŸeÖo4+pãw]>Œæá+­Tlha¾5Üᇖdñ¿¶öw£e}‘ÚÃY6´zðìáÏ¡x´V˜k0ÂH½‹®á§ù ¾;¢1¤†È1î]{j–£²"«å¹ M›q©¾s¾°l„|Ür…Or~¿ÝY$ Â¡¥C}  ÊKåýV%»ìÓÛ~JBµ7/í3(§øpk;ò\U¢îî§\þ{‡Î8@®*aŸå_u°W67’á!¦³#þPy÷Æ9G)¨IRQV»e uùŸœ;÷ÁW±{2é …¢©øÝê´JébN‚6Ë'hžBûiRò)åùðkÌ“ÌäÐÎÈv¹¿ï ´qVXoôV'_áçé®Ð2ïyœ,¦Z~KöÝj‚Vß?<O¡MCÔÐgkÚiT¨V§ ë%%Œú å¡ñ—¡ëÜùyô.tÙßFúQè9#¾ðôvI“jÉÁÏ×®mJ|/aðk­ÓÀö{òæ·—1ŒÞˆ-þàGã3¡ÑÁ?…a*ºú÷îï0{¦øCöž2˜7»[Nu¹ Iˆ§¨ÄoÂ2ñj˜Kÿ#Xµ2ß3[(ëƒæv=ư•ÿþ¨9ç Ø~0£¿BðÓïÊû.‰e ‰‹FØì­/H:ê\Ù7tÉBrh=ªBû\"éÇQp3}tø* »¼*çIsC‘~­Â’Î^‹ýú[JÅ}jïÉTò äét_Ïß(}=ó­=ÊÉ.u‡Î7£üöqy×HYT¼Í2”˜‚Ê>JêOìîâÞ#Ï™˜¢Z…ŸbÕœj µE†njMØÓQ9‰ès‡Š»_u˜¬¥¤” nÜ~/õ»¨¯µz@‹< ¹_( ³Œ¡1cmãÒ°(šRqÍ›“£9]Êa¹ƒwÑ’iÛ%Ž­‰ù³×¿¡u¯hOL²ÚôÆÉ(ÈR¡íD†»B–9ÚÍŸÕxX…¢MTMúMèètøVÑ×tºogELr†õ]öþj­ÔŸÇÛîmç?FWš©G®»Œ |¨o¸ôåŸû[PŽ jû¿réÇjáwMFez?üP3š-êµÔyq#³Ø—ãÚ7üÛ@ÜcÉhrÚû{ƒ>ÙV–Þf•%äÍiMZZ(š\Ô8ñ òµJÖEÖvÁ‡Ó< ïsCö9}S’ |é@,yÊì¤åYH{˜BZ‡÷@u>¤•– öÞì„wo¶æÚŸAÆñòeÝ~CÈf—8=Ðséï:ù¢J•=²…P°Õ(ÞøŠŽEl0x}ÿ{\' „1<ƒ’À?‘æ{Y¡ôò÷Ë:‰ŒN”9!'Êe+ÁI úP>sÙ¸þ8Ûÿ¹†Ý‰x¨˜˜òú½ •I6A›]”PõMv2ëú}¨Vtí¾55¯yÇnÆÃg&¢ÅäUø²ü\·ÄÚê[¹Ø:Ÿ„Æ_[iç¡ùò{3ÿEhwÛ=Ì m™ ÷¢S #f§þ̬tbÙgì& ÝoãYü5¡g;#áˆíô½wœGg ÐÿG!š0ضoŸ®Œ. ^¬ Ó]„Q#Yuƒq}=Óô˜<ÈÁÍ#Ó%¤’Ž$ÔðËäB^=' ÌŸÞ9…»©²ÌÉ ‘Q²_xç 2S~jQšCV¦ý‡Â-‘Ýb¾á¬ˆ$r~“ˆþB÷ yÊÏè…Æ¿½uc ÐÌh}™|¾ -nðûÔ£‚VoER¶¬}кvsý(Ã'´©œûPùÚ~oÚšF´[‘~SóŽ x]GÕüõÑÑ}\¿W Ó]^~&ëu¨Æh){âx*âÙ`Ÿ®´­rS$ç:¼ÿÜŒ´ûÿG‡½?yØ1NWÂÛܯw']˜MA¡~¹Wüó°³Z¤?Z2†HPÚɼtÝý þAò5Ѝýøji?G|jrÑýN{*K?R·®BÅnuâA½{PÚ¬)×xõ#O纒CAžþË«£P0Ï+Ñå¥Q™”*fÛ=G_Ìõ×úLó~FBkèëëqåÿÌÏ/d¼yú2ôØ•æPìM‡þ&±ÛéOg`0QHœŠ/FLN<:¼ÿ'Œ«\ŽïìƒÉ(ì¬hZ˜*¦²³¾KScö»S5æã²”4Lå;S7t§"Y`zAéÚ¦(ÌX(Í>MƒÙ«>ï ]áW¦»¾÷‘Z˜'sšnqyó}2t©+°Ð}‡ÛYÿ2,ŽØéÌÞ¿ËtÔæ×ÙÉ`ÅeÏ»a‚^W«:ÜWæÀºŽt»¢ÿ:l|;+@»ö ¶ÂöÝ Êž€?¾MÃ[°³W$A,„‰^Q½æŠ^Eâ²Ûb5Yßd¤Ð7¸¥w‰)[ŸE²H™©Ò~m¤ a÷ÜíÖŽ”ÙYNnŽH}‰Í•>qiÐ’Uœ øÛ¤ÉÞ dàKŸ;§æ‹ŒBä/Úç!“BâY?3‚ÏÙ¼,ù-Ñ€,w™Ò%בµÄi†ƒÀ‡*QŽ{ë'‘ã­ÝE÷AÈ%Ç}ùÄ aä.ó¦ãbäµq¾ä6jŠ|ƒµÚNº (€Þû®zì  øçåóoóQHûgÉB ÇÔ°ÊGŽ£( cÁ nf+årÌ©¼„⇃ãyüQ¢#ËÈ‘¥è«ZžE¢´ùhUÐv7ÊÞ¼Rúfå8jjÞZFy9I5£Ó¿Q¾ÂËYjõ*¨óJÜŸC…T— ÎT”7¶‰ lBÅ_\?W Òda2ÛÕ T¹í(}Œ›U=X^– ªÛD'ÄÍP¢¦öI÷n1Ü'{Ö‚Æ|QjܹîØÔÎÓùeŽ:£Ñ'9£/¢žaøºŠhÔ¯Ú³vh^ áù>ÍãhD· &óõ"uh,}^œCã»ÑÎý?^ ‰Å;ƒX}m4%ùöô4š¦÷܉&­D3+a³ÌC«h6.Ev÷‘šŸ`yÄE‰æÛsÊvZhq•/§ò|1Z’óߌw>‚–QiÚÜhEí´sÓ9£ˆ}[‹ÔGë•p·4·$|éDå[IŒZ7¨ {sÉáçÄÐÜ1;¯ä• µ§Ý4£¡bó•êå•v(õe\½E/O Ü}™ùœAl‰‹eþ«'f«…l¡Eæ­È’šÒ¯Ží…,=ÿ}Ž,Ý~\ù<óÈk›ØGcYÅ3#ßp†Š¢©dE–røô窬ڳ‹ðY% –™ªê hH<µX¡áKçôš.4Á×Ó³¦· %`té+´Þ?³9ð[ Ú²\K†AûácÿЋãíQè¼øÌüd tÕäç¨0B·ÚøK»á9ø¾•Ñèum~Tû~äò ¾†=âË”î0ÀªTž&ŸƒaN¦«j”0ÌÙ"󘚠ÿ!…•4K½åFqäLÜMx@–y¦bîøqo¤ÁÌÀ~QÙø•k@óBÃæ}ö3´ý| I*á°¸ø), –Ow.ÓÃ*Ý8ÛÌ™PX+ØØõF]6.FŸ‰(…-å½2ÉzÏà÷ñã,ØžÛ÷«Ü(‰®ÂÓ7¬HœÔº9o€$MWã:æqÍëu¯]bHvƒµá©2’×ð Z… ¥þð5 :-¤Uúä~Æ i2(Ngè¦ Ý/Ù¹7kq÷=‡xòoÈ0ñ¬ŽZç>2ñY|¬®AæÃFÔWUÒ‘å•Ü„XV²ŽNñ{æn#;{*~ÿ‚Í÷³˜Ò©ëùyVg}ÈS_h.wÿòK&ŽŽ À§G ÈhQ°6\%[Qèñ'¢Ä±T>væåÇSqÈÒËl·BQGF_z;³ÉÍÂ{²ƒoø¨ˆ¡øÆâç+/¡¤‘d©õ±”z»Þû5”eø@뇢ìû¤æëÝFŽ%¥D¢ÂìÙá? ¨tÏ“Ÿf#ª„U*’Yë¢ê‡wv.¨!óÖ&Çà(ju¦Rÿ>BŽxâ$Ó¡Ôî®<`Ïòu†Uò­QwóèŸ2qÔ«Ñëª/‰úQb&µû)ÑÀ€jÕbélËLÉ¡aöœ@t\%¹j¨äˆ ÑÎÉh|ÏØfÈ„MøÊw³Rn ÉsŸH[¢.4åï³6ÿ¦ÏŒÍ w§£oàΈÝA4{²©ÉÝy#U›)wt_¢Ý©•ûÚVÈÞ#Êü,–~0\‘ºØBØß:”ðTž ÎÜ5G¨zjC¹³ãs¯°I<9Gȃ»÷=¾ Åò×ÃÛèö@áþIs²·üg5ÓôrÓK,5àƒY3c¤ƒíßš—@¶kO‹”pwŸâ¿R•šwO ¼€·‹Z¡Æ» .ìG¥ðZ4.wYgSCó¢ ­0ÿhå¿!åmS ‰á_ Ýçxƒ¼Ë¿|X¿RW0 m«÷[j¨" uö¾Êy[鿯kI-RÕŸñþ[ÿóøž×V’Pš}Úßè×?H‡Ž¯U~UŽÐ%'ÿˆÍ7 ¾5ÜSl#ßûÚîC¯qFòŽçü¤¥©¼öê *„Þ¢Ó€aO×k7aTŠ;I¾¬Æ£¸ERI²`Jà„¡EÌ’ _ñßu æN唟 …êÇ*?`éåƒ%AŸÌÆO-g`=CèD{ lŸe>5¨í´Û §c‘¨vëÓ<#’èzý(ŽÖGÒš4íð¾Hf™0™›Ž´#…~þHEÛ‘òìþÒx»’H´!=«ïÙQçÈp9ï3 ÜBÆíˆ‰yÈ|GøjÈÈ#dUùâáž?‰lÓçîè Õ Gyî³6YcäúǾ1ë†<µ‡9#‘o+¢ÝKGÒ‚to×¢àt¨K@Ç%6ìþî<‚"Õ:ÕívPÌw-0§vÅnª­ìREIs¦Œëá¢(¦³1º‚²[Zjo•Q~dùdÚZ?*…5c@e·œIÑŽiÜ{š^éd ªÕ Qú¢Ž£¼_H¡V?)»î†¢ëv÷µ Ô¡ñhº,ŸºQu#äcë>|m•5_ƒ†bí—IÚÐX,>J3M÷4(g–Fs¥#Ǫ6ÐR«èÞ–h´fùþ•n‹ ­§Â†nÞC› ÃBƒ/´]dýTå»ö”¼{èTÙÑA2žø‡ì8:îï4 ´%ðaó3óû¹”oi亨¼É­­ÚÁ‚ŠhÏë£+õP0wóat»ña·'ƒ*ºSoôÅü_¹´4êóÝv £QÓë!øªÍŸÝCÙÈ´©°<`ΊÆÓ2šÉÓwD þê¯t1øgB&Áÿ T¿òw­ÿÕSN—Ô}ù7™µ“ü†«* 2lUXÇžLÁ»xd¼©EK®¯}Ný­i¯Ãt~ÕÀ{¯¥Þ§†!ëÆvºçø¿ëügÝß';¡n:PØtgväïq‹Š†-k„â¹Ö£ríPr;Ì‘ø”ßßóÂÞgY_›wýÇ/ðÚÏ·Æ7C…Õô3ÙCËPIäÛ«yyªÄž/rk8§ëÙd3¡z¢æº6!Wךü<¯-|¾ä‰MŸµ‚ºµ{!-Ä ‘ê€9cy4ÅÏr”¯¤ÿÕs›øE·ñò¿çƒÎ®ª9oSè&¿Âpwþü?ùÚežùôn„µŸì醟¥“Ý5Oé`Ð'ðòýÝÏa˜Êy§ù1Œ´=Lg]†±/Úœ g3`âÇã³Üp˜VûSç1ÌV-ý¸ºHóü*EgFüaÁÒŒêÙ…0X =xåT*,ÕŸv×:}÷Ÿ~ Ï É9XS–¹Â{6bn;'³~[ØÊ?²×ïó g'$>äïھ샤¢Wt–å!¹Êèú2RP‰fN ø¨zn®Bˆ9Ò¤ÕºE¦ ý’ó\Å‘¡kÃÑJr™Òhîq^z…,)¤A>åÈÖ2¾¿üè)ä4yð±V–y„7:ßÄ» ¿‡úÑK‰(xšd; ;xFfºk èMÏÅ›l(¾‡'ôbÆy”’) `3¾†²+† £‚{ Ù¸œ*ù^ýîn`Š*‡µÛ.Ì}GU(|ñx5ŒÖÖLip_ªÛfÜ}Ô~¹+eÐuŸøÖ_*AýÜ5…Wm$h8×E\ò´ Mt}¸’Ñì‘ÌǶçhñ31í—” Zm÷“î*û_ÓaÜqRj÷¾2´w4{h/Ä‹‚D»æíZ/Ãï+ÏäRùì¡{QæVXîUhm.+¡+…:ZáÄyø 5lZ)6¶ ê›x“c¸Éß}]^'.òêMÀ_]üõGº4+ãœÜ¿õ?ú©ò<¡óªïï>ouqK¦_ký»{Î(ï¡-þ×oªçÉ“vÁHT§½©-Œ?è –cxS$CF“³r÷ë´ãuÛÁG05'4À¬S¹,7kIþÖi£ªÛ¬[`æ´$Ý·Ý*0ûÚ›9ò–üjœ%|]4}j~¬îºLp4, Éq̼]…Å™óÇRIa™Å¯;îðUXñˆcn¸óVÏÝ}O°nQ+Û;#£7ÆòÞ‡ýõ±?6"Ig½ÉaG:¨¤ÔèÁ&¶cHœÂ¤ô¡’T¬Úl^DÒU.ò;s‘ÌÄQË=ô’W©KþBJW~£‡¬'šK’,ûø5¤ù¥¾žGtõžw’TŠpwÚ,ë¯?Èðõv —~2¶+²E¿cA¦úÉ®‡SÈÂC¾G^ñ-²j"Ý`A¶¯<:Ï­)COðôú.älê#1{d…Üž¬t_sçÏ<å Oä»»âwz9 p+Û{5.¡ e×à¡FR’°ë_O˜FáðµÜHZåUß+|”ž'ÅM¿FñÛfKF Ê(Ió-0"–¥ ®½üøbý¯ÊÌîšPË@9SÙ#ß;o¡¼^åHkÊ7U’ˆfÙ¡‚VO†›5*<‹‰suÒCEVÃ9û˨øùôGk?UT*;n¾¿‚*§^>üØuUuÖ9X·Q]ªYŸæåsÔ$YV‘@­&·Ó×ÈüÛ÷d¨ÿ3gø’ð÷œÔ9•œyðêgÔÍ¿ÇûØÒõÙõòôBÐ Àû—ʹhHÏX_µ‚†O]â5V˜ÑH@#òø= 4zÌ­µãè‡ÆÔüÏüØÃÑø ÕÑfÚ+hÜlK3’Š&¢ µq½Th°&“ý>MÊNýx-€&Û/íwß/ES[)†% hê÷iÂU M_ˆï0Ûø`—ã>ß}h½÷}BÅŸ0ä y×÷ÌdÖ?·;I·…@g‘;ÑèÃßðuÀJ<9àT¿Ÿñº~³í¯¿üÇ Ft¯È?Èý›C³§F¯ŽÙ!d±Ü®oŒùëÿ©ÿñ¿\U¶…{¥–ýî?:þ>ÿ“S?§m·³ï¥ú›Wš‚CìÄA“#Ý ¦D±æfðEP_„V­Ù¨¥œ³ÐFC^ÁFmß…ß±¼€ö!‰|ÞYEè¤Lu¯9Ë]'8÷«XB·ðòÓÀô¿9óG[É“”aèا|çq œù‘}Ž ÿ©Àoîù0\l/‘Å£?—_˸“pf’)L*ËÞm*Q€iIuº/C—`VNåb¿Ë"Ì)»ŒÁ|±o‚AŸ%,¬»}:;1 Kü>:g+ÂòpŽ5\ö€Õ³·?´„¸Ã:çHòÝç`£ˆÆ¡µ®¶¼ ¨"šæþ™C|þû{ØQšüÎÑ«„D¯#ªìçî q…$¿­’Œ)ä${|Ä]üì'%7b‘ÌßÕVjè=’°íÖQÿƒ”gcGñø#µòAAie¤eÎ9o§®ôŒï[>sÈ"ƒ¸@ycc2º;ô’gó!Ó[ñŽCú/y¦>aÇ0²ÊsÝ+E¶˜…õ—C¤è+r‘s­ÛýAÃCïò _Åî™æNHP­}½…‚‰V“ªÃ©(äCÑSõá< +D’8¼Ø‡"Õª2bP”9ˆ,OÓÅÜßžÎ={|ã~¯%R£xÁ»‘†”$^­1R®VÞü2Ö(ÝÒ/ûCDeÎtÔ©Ž¢¼OˆÔÙ,STøæ¢ðm•.7Γó£ŠujÂØ1/Tõì•ËnÒEõ§¤ÒK¥¨¥àä{áþ›[g|QÛÔùÚbº)êˆ%úÏG]šÀ¼ãc'P7úÃôÉQÔãJWO|‚zw?eÚÉ4T mgǯFvUNÖµƒ°d¥œóÊ ZÅõÃF ¶áÀùxe¨ò>hÓëËEÐaGý;”ò}Ⱦ¦uŠ&çV²ŒÏC!?›âÇxÈ7>vëÑ=Èõ]Š©9³"‰¦÷r§ô<0Cžú„¾ ^?Ï|qå ƒJçlñ‰Ïç &«ïd‡.ÔSi§øÞ4†¯¯Ïë¾g“‚mÉo2' 58?ßžî5´=WNá$è«þ™œ¹õ+èp™zÝ}Ò:v=Zvyí–sÕ-ï¡M›JŽƒsZ½öøoЄ–‘§Ì'  åGí›«×  Õ2!ÅÞ’ÚØš:„Ú7Þ×RÖ}¢ðàç$tüצ¾]æ¯ #áÛ$Õ[ƒ<|ßrÚ8q¨z½öX šÂOé‹‹ŸwÁ N¡ËKózéœNT†Q’ƒÞÃþåCÞó†çתaf;`N·Jæé÷r×§›ÀBÅøüä– ,pöÝ+;¦ª{Ù`ýVd n¯ l¹¾{žô¶­„òX÷·!ÑÛA“ÃLH"~`ä"'?P?¿ª0…df‹Í&BHA§o}m˜©˜üÞÏ&—#ÏÖ‘ÆŠ·HÏ›âø .é‹$ˆ "ãF WNj2ßœÈꈋAV¹“Öóû‘mœ£æk~#…â_!WA¥FÙTò4Ÿ“p8ÈOÞ­kxê3 dIh´H{¡àÜw^«˜ 6Yâd/B‘ÏÏyÇQ,à势Ò·é_\Ü냒Žæl&£ôý=¢Â9(·[V3˜,å§ú(B|‹Q10"†Gv•m(å«+àÞ“ÅZãi‚¨–ÈÄU_ „¹S¿§ŒŸ VzŽý™¯³ˆ{L„âôçQ{Þ@SpæêúT¸ëoA}Nªã4Îh°ôÕ4Žw†íO´Ò¢Éw6I14ëušP?ʃC|6$ûЪ-®ÑR9­«6§3 My¤ÚU‡´­¶TL¥üˆv¯ZR·=Ñ~.Ü¿“ð99²tº±IŒ£“~Ð7å?Sè|…Ž„Kg÷WI9–´‰XXBt²ñàé_'B,ö k—‹ Û2*ó7ªþŽî‡uL_ÖØ£‡˜cm+=†–ŠÏ(¸¥ ésúî[¯§ñû@[Ñ*øu?IÐZËš jÞÿ3T™,8–øêCiÔ#»Ó͇¡pœ“ÿÔéIÈ3xW)ìn 9O\LóÊ !«ºyµOç¼_æÖЊ~ ïL~ZÓ‚Ô³îul›z 'ÿø HsK*÷ãã‡÷V§¨kÇžCÖóˆ+ÝL¿áßäË5l·*œÝ£êGðY½ž –.T·J‚¢LžgE¢? ˜“ÈåöÛ¯P¼3pÚ© JANü'”¾Y3Ê!‚2ÁرPc'('æ·ÊîÈ„ŽÕˆáÂák7õT’ÛVœ­ú—Úµ0ÐCõÐ0éæÃÇÿò!WÛ©GÄLÿ3¦§ ó™[A›‚ŇÄg ƒÒ_Óp¡:¦qhskB7­§G¶æèqÌHÕH¿ ½;^õÛt ðóKîG9í0x¶;ËÞ¥† KLøOÀHwˆ-‰€ŒÕ %%²ÂDwJŸ L˾qù;³¹2ÏΞƒyöÖº»7¶aÁ4áMž,†žk9<@KíÖ=sw :vgÑ ä…5¸%–˜Ì7›«_«Âoµèb‹C°£ñô§Ö{F$ÖúiûìîQ$Yg¿­zð3îj‘=Zä‘€äN£÷í„rµmœ­–iü#´Ðým¥¾“UMJ‹‡ÖSµ»ÆöAû¤æî]!P“¶q¯¢~ª†foýz•´J'­žAyÇL`ª¨<”$œ#9Q…Gß«±ÎB>§/cÞDÎßZœ½4°ÖêCð×µ¿Œ" îÏ—á;æÐêjíz”éßþá%jwÇ:cøÉõeÒ(AÛ3*ÂX`äÙªGGÆ Œ¶”˜^*…)9¹¾óO0í“U-¸çLßÌíj7 ƒi¡îÅDG˜šÔìÜüü·N_±×™†™4=kör˜Y̳õøstk¼ÔÍ0¯$--ôâ+,Йò2_†EЉÞj½Ã°Ä*ñü}é,,ÃËc¡¾S°rK§Wßó¬nzKZôŸ€õàµãß`“ý7Åï‡a«ÂÞ×öd=ü¹îª|òº/ìÍo8Y-‰Dy¢½MË‘¸6ô&)½<’ ?Ödò+Ä]\zv#³Hvšüpru’Od ©^×EÊkž†‰çY‘ZWƒäÜŒ/Ò²_àh[ë@º•8ùwbÙ¸û»„o#2Œ©§-ñAÆÙû±S-ÿôÙå‰6ú#‹{Ú$ÿEdÏëÛûÙæ.Öõ|•@Žc'…<isÕ¾F°â r_üQ¯¾¿y÷X†7@ò•iȼÒ@¹z®7-Ý(Èû®¢­®…4vß_½‡Â†=¶åQT%ß2p—.Š-=^}—9†â鞃 ²z(¹‡–͸¥ÜÉ®>8¥Òõ û]3Qv—*ó¯ÙQ”sÿ$8vÁåµh{B…ÔQ¾Ÿô´­F&*X·þYqÑA…‚ñ'm ÷QQ)´©ÜÍÇÿ0¬o¢RŸ¸[A4ªDNyo¬@U»?"r'P]ýŽõ_ ¨É4|"úµj ¼8ùÜ•!ÿ3‡=j\®è¥›GËì_ÜU—Q·ò"ÛÛ+/QŸÇîíñFø_ãÃëKɤZKGК'u{ñû6ò:PÂÞ,Ø|pŽëÿˇÿåÃÿòáÿ >Å€ÃZ?9iŽ7?m  øòüøåš§ÐöûŠî¥¬ËÐ!Þ§òÕïtä«UNFœ„Ž [àûã íoL {mY íN°xÑ` ´¦·¨Òþ­ü!s·?Ö¡ë`¯â"ðfîç`C‚nµ)r­¡R§ªÙõ´Ï.ryJˆA'},i\á?º]ZY%ú§ñî@Ü—æ0‘HfBïÕ¤]Ú—ðÓ|L8r2mæÅÕ>¸À°\¶q`^/ŒîäL¢òú—¹Â “å.ÁÌ&ù!æ+C0O³}­kšÊT¼w*ŽÁRfC‰q¬6¬L›µîØwºÏÔõì¡$Ø’OLÙ¨‡mVµ“wôÙ(ºv°º‹Iè­gUÞø0ù$ÉÕÈ}H¦wˆìm\_N 8ÜC*Ú<­wÆÿÑçÓYTHÏžÑý9}.´…ËÒå!ã WØ«¼5dŽxß/©Ьfgt$êm¨dÃ)º 9ÊâÃ/èI"WÉ’kØòtûw~¶ŠüL’Ùb]Q ï›ýÝ:\V4 @a :Ç÷"R(Rßñ‹r¼Å.YŸQH:âÓçŽ%£¤OOÿY”N»\iíá‡r•žj(?_óÒf”à{n˜‡®t£²žëÅÃsp¯õ‰“O&¾ ZÀÇYÔ¸V¨Ðêö µ‚¿ÕD]ëüÞðÆ·¨·rꚎè4TúU¾P§C£ä_-¼Ñ$úàºòý)4‹0óÎlyŽ7ÔÞüš[@« “Û¡g©Ðú‰FþÛ´¹WSX¹7mD>{‰vÏ¥/Öd¢}nÀpó¼<:4·÷ ö¡ã’8ƒ\ÄCtæÛb° »Šûí¨/¯ K\îéy:^<бé4P?‹®BÜ6¼Qèx]âTEú.’5®ŠîÉ »Î?0ÅÐ3"É|CÑh2/ Ø=#‚Œç3frÁl¡½ê>‰?ðõüQ¯2nc¨\»ýê|”òå~6œSƒÂ¨Í3%”=»dÂ7¯IÐ× "Ž^‚Ïí&U¨ÂûÄnwMà©ùÁiÊH•nRŽºÁ oÿTí–I‘4Ú­gá½êXŒØ¡CUDkgþôø?}þw_†‚!ß^÷P̳+P¸:O¥ÊŠŠ§’ì \HìgPÅ•Êî6A‰éä\à•V(¥;8b ¥=+?mê.C™uÒ¬÷ ”‹ø¨&L>€§F6äsöBÅ,&F‰Ír´·Á‰Èù0hFæV@uQhîtò¿|È2—kÛù?óaæÓصtUhÓ¦´«…F£™Ú;gÿ™ë!ß,ÝLF9M‘Úг?y-ÂÀúÈ̪â1~v\Ñ8U÷ƒTG˜_7ýÓ÷?_õ¸Fú¬ÍqÅÂXƒÒ»”Ú`˜èü@)âñ¦E¬§yü#ÿé¯xnoÿ§¯ò=O. Ç ¾òÁâÕ“ú†¶þ°Ô«èð"¿VNóÆX3ž)*ʆ$Yxà·ð'}©k°Cgqæ6q#s™WtC’úÌ®­ÜõXl4°YÉã®ïb§§BÊ œ¿é‘†ƒõžL÷ ¤ë0IO E† s™u[dÒ^¿ivËY¤w³il ›~Ë|H¼MŸµ$äÜ/­‰²?ï=KÉ'#d´n½]'‚BKÅÂlI‚(Êûs"vG÷<_a}³s%#´ó^ÖG™Iš½fÓu(ßã8òû *6o¡‹f*çz´¦qáÞÙ¬¥c¨~twt8‹'j%4ç??ýµI…³†rQ§“:Ü-ä êUi}ßEï-;‡§¢Ñöø-'Õ—hª–1ùzó<š{‡'u¡e€äyßÛÁh-¦·»pmmøJÝûŽ -ß¾o¹ÐNìÒÝ—rh¿×8eMº,¯Q‹· :ú®ñòºSËë4t®ûlÛJ‡.»8šedñ€‘Ñ1­=» ªNðã)Y¨'uÙ»&4­îºô m¶ÿ^>èØñ ~JD2RÇðÁàЩKG­Da$ÝÉù Œ÷|Ú8:2S‚ÏÖ§;Ãt²gYæ«£0]1dn÷áL;ÞäŠØãÓ|>¾r1ÿÖBrvwKm˜=xGFÀ~‰„8ñí 9+¢‹s÷`Þaêud7/,ÈéÐѦú¢̋ä'}Ͱ´¯·çRK7,ë\?d +…˯ʔÃ`MܹâWþ}XÏ>?›¦¾zEº°µl*Cy» þ”ݯó|Ó;ÏŠ¹r#Qû«7×C/ ñØ`¶I’’ƒˆ¤ëuÜ>Ï÷ÎìA²{ùQƒT4H±[òÆÍáHùîÁË® ¤>a²£qÚiµsAXãÒ 7°§"iÎdœ˜2lÆ7,m Å…éª2‹Øpð)n!‹³„ì Zcd};“˜ ‰ìŒ¶,¦fÈq+·£^ù7ØòúÚøqäŽ}^qdŒy­ÓÿKB¾±©Î§( h`¢£C“ >¿A!Óó%E-Îjf’˜çFQK%¯¦—Þ¸‡îë©×â(þ9¨y¢Ú%uµ?ÐJó Ô¥HÏSª×QúƒQUçq”å>ޝE§Pî‚Sÿ'Ê+Lçö_DùYƒÞjT8RgqaHZuß}CE›â qí\T¢Œ>ú(¬ •~ûIòg²‚ÚÞavTõzem˜þ Õ\)Õ® ¦ÀóA1@­Å1ïòÜ4„fåískt¨}ò´kÿ±eÔ‰ŽôœDÝÆƒ»ó³VP_ÐóÉ'ûÿ=>lœ¬ã¿x­IRŽŠä_F^©#I™ ¾°qkÿÉ÷¬ÿåÃÿòáùðÿÒWú \óC›Ù~ÆÑõdRràJå„EÕ`Î*I—Y†–XXBÍZ¨¼aFTþñ¾oæ åv¡&ßk@ÉaÿUÆP;(:XñÃç#+¤ú)¹u@^[÷‰8¹øÐ½%ÏzŽr.û®²{ô9ñþ}Œ ägT³H~ƒâØ·ûø-@åÝ(»V.¨uxSou·8¨ÏôÔL^\vtÀZyî>ö’ÿ¤8^߃v§ºÇ ã:–åm5èdŒìÛL"ðbWÒI2´ï¼L_=m# BR@ëR°]¾#´ºx4Š*F帱e":¡í¸öR\Õah—½ñ­ûƒ#!_šÊùNÙ¥f–IèºûühÑ[‚nW¼“íAÏ^CÉG+w ÷9G†§ûøyÊÎ>ŠVïYi✂aÍàЦk0ªÕÜÕ%§ú/²ÇÖy+ÀÌÊæ‘}I0OU;;,Ñ Eá?Y#2`)EFðH†3¬|³ðùEðWsË#°E$Í’n×~X í–=€D>é/yëXxÙ`eI•À‡·<¯öÜF2-Š?ç‘‚ž4Ló‹Rñ¾-½™i.ªI28ž£™Fm€ ιÿü´bŠŒóG&T±#ó•³4WFU˜vCÁSÙúIºu#Gév€É×§ÈU!j0zyý¢Ô3‘Ÿ7X€¼vJ"¤Ì·¤Qp=úå®?(l½çžî&ùºxÉ/»Å®üžY É@q¯H¯™Ç(LkO5€Ò9%É œêe ²Š”_Í@уgPñÜ-ÚÑÒRTVÉ© ¾\†{U¸[ïñC5óSvë÷ÅQÃ>«ð j™245_PD(êÛ8_óµ‹iYN!A]å÷ÎCQ¯µõÙ í4ˆcQ˳—D#7†¹­14Ñ|ý}8Ída\LK-ä.*Äô½B+cùœìkh}vš%¬€mN’©í°w£í± çµÌÑîT®ðpÌo´ô¶¯·îD‡è6§÷7Ðñy‡‰$:•OòÅTîFçÁþéûçÑ…º¥Á£;hèçàÁÓA|Æ" èúÎzàni1ºÍVãÒûŒîÜ7u-Ù#0TLÀžÖë4š Â-'02“g¦5ÀÌ* IöWИ¡ÓÇÃV•¯Íy3·E“o7=Jþ¨¦|޹ɚ\¡â#Z÷qÚ²¬dFz÷‡Ox¯CZiÑÉΧðvL3i“Ý Þ&¥z/|¢ƒÔÖKj—Ûཀ;]D7dµÜ«í„‘Ï XvF!?ǃÉb >y7ž!À Šå£ÓçÈ 8|LgúÓs(îÔ¥ i†’DÃí«oZ T—ˆ>:©ʨÉ?Ø€²+YËV3PnTÄe;>Fœ*vׂ ;æ$±ÛPÉÄš^ïaú/ž§ª|pÚª¿ñþ9oþåÃÝ´›Ë< ÿ3æÄj´ÒòC›îº+ÕR.tpñ>3T¢„ÎGåÁ«<ÐÍ!M:Ý*=ï4÷Û@|±Ð™ø9`2­3àƒáÙÛæGa˜ÿ¢¢›ŒŒ˜%ö²ü€±Æ÷ïçF`¢í£C¡Lsny8ŠÂìƒç"¶úýRMvu²Ã‚žÒðÐ…XŒt/”¨î‚¥Qæ‡+X¹T]ív ÖìÌÍ{_ÀF×5ÏûŽð›~ëkþCQØþ¹M:ê‰DkEZJ­ÎH’h,þñ î:ÂTC+Émó…¾Ì!å™ÝæƒéHÝTrò~˜Ò…Ï} ×G‹sº‘ íiò‘¹È<Dzùø£²n×”ô¿¾ƒ®ù7Ìç‘[XÞêá6òY{h’?E¡‹tÞ¡PÕèS¸"ã¼Ã·Ã qÏ!w|³Ž’ ¤Ž'®þ@™`6²§(ÿ¿ãúðWT Í_tÚû •]ý­.]×À½Š“9iVݨ¶ôãP‹!v(`@ѳ”žôr êœév3•>ŒzÆ?Ó?¤§¡Î½æüÛ—ÐÈõY⊞šD‰KùŽfïÎqw”¢EÖaÝWNýhedN} ­‹VÏýºMàÄsN8G • ×¯võ…Â` Òù/¿ò ÞþýßZ\w™ˆ—%ªÎ „ÅtA=Õ>fêÁÐê©Të‰wþ_z³ÐÂZ÷(üTÓSûze¤•ÚU`¤Aͼó"Œÿ.P ³¢„©ë T%430ÝÁVË0½º»²ƒÇ¦£÷ÿ– Óv ‡O±ù·Îï鉸pfÎu8‹7ï#TëS4t0OõslÛæ/¦òMÓùÀ‚Ýåæßè‹vœ½¨ K^ñÌF#U°|—2êðõXysj/™+¬9š:ÅÅVÂz¿Ë¾O?•aóò€iôGaø-ŽÓåíðg©ÉíÀ€néÔÚU:$ZöצuD Ýoq[-H*>D§kꈻßÿ5u ÉŠ½2-ŒBååüN!eÛ´~šêI¤¾³k{Qi/xž >„ôfc>Ÿ(Ÿ#W+7ã¾\dd¿qþUC,2 ¼òŽû+2kûËé]C–ó9.Èúe!".» Ù•z,>û"GÎöu%æYä2Ð7m¢?‡Ü¯¼óãý÷"ï%š *éeÈ¿ûÓ[Êr%p^y#Ñý ®Ö¢Ðë¤ó!#(ü¹¸bkˆE½²®ˆèÆàž=·«åGñ¡»åê.‡QÒõœUsJݧÐÓ'nFéFÝâ7Â(«øää×9”»5ód§¸å¥Óƒó?¥¡üï(“wV^¨p±âéæ !T˜JÌ-ãVCÅÓË·îS£’{a9u(*ó¤„CàT)ºHéü4ªš]v]×CuŽã‘᳨)ëî¤úÆ÷‘Izö¸ÆG'¿wÖ¢vàäæÀ§ ÔyD¶él‰ºß´)«‡ Q_ìlOßÇÿ=>LQŠ#þ“V­Þ‹."/q%Qa lÌód}áòø/þ—ÿˇÿøðÚ»Òñ.*´9õý€ã^ydÕ«8ÊK¼ ïž[˜7Bó'©9¯]PsëÎ-ÁéZ¨ü"}¦˜ÊúºÈ>Jü’]¶ ÎBzPÄ&aiºy |ŸŽÍîû纥é¸Í`øpçh“Sßd·»û½?Ùç cŽšHA.ÿ]tÏì øP‚ ÿ‹T¨ÌZš‡™ ÈÝ:óqŽsÇÞÂB^‹£”é2,=½ÚtV>YQd–ÚúTΫ¸9IØlõ»™ý¾þ<÷ìŽáHB"Ë;GΗø°Gc;«–À‡ARÕ×dG‘LQÕßû$==9jä!•°þ§žJ#¤ îŠÉ‹m@zι}iÌêÈà?·oo?2NgZE5©!s šZg¼7²ò4… >ME¶÷“Š? |èãÅãL‡\ÕN/©†>!Ï”¯jºˆ7ò‹7ÇŽ-E¡@¥’ðÍÃq(ø{øèϼ{(l§ÓúmÎEZÈ»ƒ’P,"õXÖ…÷ë˜üJ…’·ÝT9ÂŒÒíy+)½Ó(göRÝ”oÁô¬·[ âÙø¯¦W“QYœ’"™š÷rÖ×íòbG5®ù¸kB¨!bû$é¦5jqTÐl›u \?2iƒÚÏæÂ˜ùëP—ãeÇvŽ3êe¨¶`Dƒƒ™_N£·’Óm[4žùURtÊM¿)z=}“†æ%’ NhùUäðÈ<Z›{^2¯¹†6:úEZëÛh«±vå`3Úih~ µ7F{Ø_Sˆ-š¯ÐÑ:4\äÞ<:¹}“®%zŠÎ~O‚,÷NãþˆY±ö—«è’Ĥøºö(eFÉ<ØÇõ°¨0 ÝHŽÚ\MvÄCÒfDf ¡x%_m÷RÇ4ŽÕ?}ñ!2d™ ¤•òÜŸrOv23hã³ b2‡JÛ6œ-®„ÓéŒÙa(h,csÔ(…\ˤu6V‚ÿ4 ¡"øŸë˜HÙ(¤ÿúßBÚy±Kòÿ‡ iN¾.‚·:ñy?g!5züÜfº4¼§ü;jÝYœ›*™:ð¡òl•¯90\)^ðI„"ÍÓbg¡8Eé4ýó¿ðæ¦kP¢txŒõx ”̽¢»Õü|yµð€ ”©³g]?e•[ªLùP~’œÂ›ùÿcï­¢ªl»7^:éFºAD`N:¤ T‹²@QPTT¥%¥[BBº[n¤6~{z°Ç{úýOÞ£9`±cçz®ëÇ|î9O@aøOï ¦×Ptê²ßÂ"áï>˼™=Í_>¼ bã!e ª^b!þò!¥¸æÛ„ƒÿ̇™7G\ȡѨ6­h—4ó/˜•oNËš¤¶œòhßEgCÉE GýB9ÜÕ ‡ž9íã;uè›d g^†€ŸFOT×aH¸m*óñ{ø>.§]µSFêDÔ3Ù`´þëá Ú˜ØùŽ:F¥ ¦îuþÐÌ|³„ïWW¼¢a}|Vùs`>ÀV’Fy;7O-Ql Á’?]ÊÉḛ̈|8±ï—H;¬Þ}Ó›Aãëk vg‰¾Àfzò3•DM$h–Ù¼`Dçûo±¨l"ÉnÖïk *HÆ<¬Éq»)vKÝüòÚ©üµ6Ä¥‘F6x0ïMÒñD¦–§#}Û°ïõÈȘÎѻ槇̟sîAβídñ Iz‚µÏkŸÔBî¥sŽ–ÔsÈ›aå!‹üOè ¯÷R `êÁÞPáªMh3÷@±îòeY-”¢çm!°92V£e Õa('{ñ½µœ*ùÖyŸ8ˆŠï>:*Õã_Ꮉ¨êoJk`~¡X&GƒN5icö*vê V½ï­ãûQ']¬jÏ!Ô˳I1AƒvFO+uÜ7s*4ïØs4ž|IÛ¹ дŒÂ#ªÒ Ímd©®éC #/OçU´Ôi'Øk¨VZ„ÆŒâ/ÑZ[Vž\Dm vE<•¸Œ¶Tv—5ÐîðÈl—¦ÚŸËÜ÷ø;î¿du¸C„·Þ½è{ æ-&b×ÄCÝwóÞõâb§}†!»þ‘·vL6Ÿð„$ÅÕÀLhØÂÞÁóÅP¥º—á¬M”÷ÍöM¨C)™Á¤üM(–öam¬ð†‚¬7'!·èq#é±øì¿2K|x²Ô²í7î'ü©9-;S²¡Ä“¨{QúTïä™Ù¥Ã NB×›“šþì?ì Ò®Ù\€>}‡àKÍ¿óá´¸|دeÓ£»®„gKÁøG¶ˆÄ³>0I^•Zóú6LJ¨+[û)ÃDþL¶¸jLø—Ù~|Áò§NоNˆ‡©ƒA,·eía:úéË̾×0ÓÀ¨iCó¯N…™iåœûç'-–0ïÉ"jÄ AÚÆL+`1_•Í*tü¤; ˜Üº–ï¨NË8MÁ*íƒ0ºÎ!XK±<ÛPN ¿)&\îi‡Mq²«Ñ$ È¨ñÒ¿‰„tQj+aJH$½Ñz1‰ü~PB’»…ª£HÚÏÒâøø’â¶W¤X“ÖN cAªœ3+¥W$pGÈnÿÒ"A¤ug ª[}t²3Æ!6\H/#{èz´ 2¨®Pk: 㡽“u;)ÔX‰qx™GºžžqDVÛGÇîé#[»Ê«ù^äpÉ£y§è‹œ9RI¹›Ç;úÌœóPò¨PœTíKE^×®ÎÝÈwHü¨D©ÑvÎÔÊ™úúzcÎ×6Í¡ÐM©çži(¢ùP|.(E7“fÝPÜ#~òI5;J¤ø pÔ°¢ä°Öåe=¸Û°ö9ýÓE”~ç³Ç¡h”üefÉte©«³Ú:JP6ðÓäµyF”#^)–;tåívùEêˆò†ï©C´QAíd(a…*6´˜ù‚Êæã®h§ Ê±®«$ÛºWUÛUûHÔ ÕYÏ]=>ŒHð¬EýãMÔð?ñ|h×jÆ^g¼yÝú»o;š½†ãt½“ÈÕÎíX6k–}u1÷>üˇÿòá¿|ø¿àCƒÏGI£Ð|ªùÝ—KqÈôÚä ÏV˜ã)¿Õ`´êÆlrd]¡\àÝØ†èàæŒšŠ–@þi‰âw#¶ó¢ÜÞ7e Û:z\,¦ø>ÉWÕÆJÅ@æay—àò+aôÍå¥^õ6êÚMh_†Ô­wg¹Éø·óçrúo#ãÏÍ“èÅuGo´+‚/kE;¾¨Ap³jXÇ›m}m…ï´±‡†Þ‘ü©ènhüáõsÊšiÒè]¡„í—†æ 5ÐR)ÀÂÕôZªŠ¿~Í©v¡[0MÌ{z¡ñV÷»· á›È’µŸÐýëôõ¡¼m½RНH¿„ÆÜŒ‹FÇîB“7áÄQùm¾üÙ\IÒ -ç ŸäÙ­%î¿t5V ýH¿XÏ#ètS8XõàtwS툡L¾d©¬û—~÷!‚W~PÁK¤ÀV+ ¼¶Ðñjÿˇô1úœ­o`r2'Ù/›ý÷\¸ÂõØK0—Î]«I·žÌL…O߇¥ó¨{³Œ°BÍ’¾¹tÖ¢2‰>‘7ÂÆ)Ï J¢!$ñªOaÞÖ¡ ù9›m><(tçö$‹Žûһ͇%ióôR´ÿ«X Rß¾ré†\Û”á]ŽNûøv¡¦õï¾!!×óå:Èûß;¾}}~aç<ä÷4íJ¼,gLEϵR+tÖ;T× e_íýÈøË‡Äúþf{vý3~v{rûë44ZG“]5h†f‰*BÝËrÐJLdá;¾ñÛoc„f¡ó˜ûàI:aèaY‹×k}}Ë}Æã4“0ðØæ¦Å=Ú­/MC ßgÙÆ²(¯ÂHý[²<­t­ió;y&HM^MpŸ€)O¦Yá¿çþ3)h9ÕzÞµþ0h1ñ{_xyßUXzz¢™ìU,Ÿ¦¾Ä™pVýTˆü®Âú°ËϹӰéwTôáÖ:¤‡Uí܇DZõ™tH¼Bw"—µ I;˜Ž†C’O¥'k!•9lj&zÜѯpý‹4î b°/Cú胛,ñÎÈx…·ô€Ñid¾˜bùæ› ²]=M¾C9Ü·Œ†¿Fî—Þ÷^Ëè!ï/~ õŸ‘ßQ&íú5ô|Ññìî7îTY:·²‚bþþÏWŠPò½Ä»÷µP†.-ãuÚv^Ümo!uåSBL®i^CÅ[G³Ø¸+p]z\£öT% ÙËYôa×{§kÑ·P#ñ2 Ÿ¿0jÙ Y3Œq Aòá>D=Î…²DÑ@d6îØÖ[Ü'`§ôìE4mèÚùý šlU½WUÈD³þ6`\*Aó¦˜æ´h&Ëò{Ø„–=oÞŠÝC«ü{W-£Ðúcúщö´ùà'cÿR mcö;Ô Ý;âä|„öo{+§>Åýo#N(YÄÑ^%^Rvx06¥lQ+%;†yHýúg>LÕ¸b¥ûúY}K~ç¼’ZP¥·•rŸSÊç¢ßѦC)Aí㈚ßóÖb>¿ä€BâÚ\.~È­±™#s`…Ï¡yD_á÷ù|N碸?5g€„GË®Jn «D6 C53ÑWÙædh8ê¨ÇÁ-‹ÒëÁ:ÐùÚi¯a®/ôÙäû[J†AÚø¸]¶ð}ÙK¹UsŒj¼Îø,` ãýNН¾2ä굄ƒ 0éàœÔ<ÍsQ×Ê¿lsâì¾ÍÚÎ?uÒá¬C}µL=‹³OV€é!>5ŸR0KIkáøpf „é3<¨`îÕ¢M×Y˜S?Éÿ >ý¼3¸ã9,þpÓ‰¼?ÕvlŠÄ²ÂrºÔdƒUÈ›?ÜúÖʲU‡_Ág$ˆ»aÓý)i‘@ÈB‡žq ¥ùÈÓYÈ€TkìŸë»(°ˆ$ £’#kHFúeýmK’û>+ô_y‚”|—´²VŠjÐséÞ 3îÈ;›ÌD4Ž´Qêqã‘Îæ³Ëm¤7èвõÊA; gæ‘Ñããá€W¡È”~8»¦ñ3²P¬ÛhI."«ÇøxQq:²m<ž$i¨CŽPùo^‡‚³~!êÇýGÈ]Sy6’cyŽ™ðŸ'”D^_>ùÄþsÈwa ÞïBþk›nlJ(°ôdWDy7 =OH©¿ÕŠ"ýgÜ™D1¶Ò—wo¡xHo€-;JTSÍ=;‚’›Z»/¨=ÃÝ[%µk(˶ye¸y§Nr¢,' ɉ](™¢ºåE‚rÜVµ±¯»Q.é´ñÉŸPÞYî(9*|_ÎÔƒŠ“¤qcW¹QùÍ‹æÀ m¿¼ðÆ÷6ëWT5ìÑþÕÅÒ±äú"²ë½ ÙƒÏú†½Q3sQò´¯à¨ ßB³3ÊÒ—Ÿÿ@®dÏ’j—XËŒš¼³¸ò/þˇÿòáÿ€og´i~QCs×úi¥4^düÙéÇ—ñfCÆ>ÚwUAÝÎ[½GmçµìOzí'7¡øHhøs—í»¾Ïõ€9MÈ1 ïvü}JܲV~~td¦ƒL~‡©ÇÁ®AªJÕ’ü{n~ÍÉoH½OdØ\ÊÇíïø™@¡ ÏG6(žøÊdEù*ŒÛìwõyCM² ï«| 4ÏÉ¿7tÇ24=¡Éº#‰ç½'4Ÿ‰Q¤‚–·f ó,Ð*à&'5• -£þuUþ–ÛzK  ’…¦n“𝠱âÅ,ËÆ¶>©$NèAC+e{ž÷‹ß{-ŒùD~Aã÷É´x¹ÐôöËb&4{›ä‡šrBK°”ß;oh훌Ï$y í×lÃÏE Aç£ÄËŽ¶Ý«¾¹G( ¯îå—wÂÀ“ú#L^ÀБ!†t¶0lz‡Ö;2þ/Ò¦ëëÒÚÀä¨ñ.Û«­0KÌ+F>¢sÉgul[|`!ÀÂ*›6–^Yª;4Âò¸W±ëz&¬]ZõÌsl† © ]¿)$`8üC4Š c„è$¾ë"±ÍäÝ;mHÊ·³OXOÉwòï3åiCÊÝ×=Z—‘:€ÑbÝ5i¹¤£=ý‘îõŠ_Òêp»Œx!£KIÈ­Sd¦³xL´3Y *ntNnó¡a±Ë rÔäŽ>\¿Î< ŽoC¼ìkœŽ¼5AL²¶ÈO:ñT˜ö_5!ÊÉFÁ6#ž¸çPø1‘ƒ´¹'Šúù„~—½ˆâɤ«6T(ùkFlÃþ$J_šSnìÐDYŠÓ ÅÛþw>Ì”‰þ*ìHÖ#ŒŠCÂ.C§P¹éªXþ1VT)"uýT‰ªY»î\bC`ÎWm$WEÓYŽ<·EÍþàZ¯0 ÔvÚ˜r VBåÍ×Óïo£ÞS÷´¶< 40»utVqŸûÉ€éh̯TçS€¦»º+ ]–Ñl-îAyÙ*šO•fÎ5B‹¾Ó‡Å)#вɈ7{°­*!6ûb)ZçÖÖÑ…×mçÑöEöwûÐöM†áž›Ñîó‘dzƒhM¨çÓÞo¸ÿŒ8s °Ñ–’Ú#5’ÉHSñD¬ íþ«x˜%c9^ÉoÜî‹ÿ®‚ú’'ÃÉH©pgS]¼æç4×}4ÛÿªmÚ4T(Í (™¡ëDØÈqç}îÑŸþ^FüÔÅm”…´+Ú­fÒJ¢îImå. IÜ/fÆÈ!~Œ ì©]Ä1Ü4-9 ±~G/{2ÄmÏÚQõARGÍãwZ¶Û:y»«„cÛYÛWí9þô'sdðFÞÜæÁ; ½<][šõ)Õðwß0Ì /ôˆDOÍö}áFc#5ä?W¹é%r v|§$Ó1‚‚´ˆ6Euy(¬!ߘ^"¢K{V[ È?‡w}¿nè_>FÏ5=ô€‰_Åãaò¼tôسH˜|ÆocG“BÖvõfC0±À…†DêdØCÅ…§0ÕPü²CËfø“zMüóaV,„ˆí©%ÌöÀúáÌɰ}%T‚ù²€”5ý[°ÐŸŸÂëKô‘{“&àç©Â/‘î‹°ÜÏÃy‘UVÏM'4jÀ:“RòÑoåð«ŒOd@l6Ÿyp눹#ætnl¿šD4žIù‰D§öw$®­¼OWŠ$õô­7†5LÂoý’ÇÓ ÷Xÿ@JƒOšùûÏ"5UÀ¼±×îŽk:¡›Š´µgôFºÎA"åùíq¤wô•¶4C·»£_ý1´åò‡3ÜÈôí¼º_á=d#PçtöGÖWú× Ùù»Þ¸¦"GnbˆùípäÎ~ÁÈ’Ü?…‡[,Y‘çÁã¾(þ8ä}–µ:ùÝùüšwÑÌ"ÿÃÑѽ(¸ãz­Û§/(”*R)ij†"Wn¯\ñ’B1ù® ®`Oaf?ÍŠß+bPŠUsÔú£9îö¾Aº…Ò MB‘»Q†ùÕ¼fÊŠ›=–™öDÙ1| QN)¹“«O 媣$ê«iPþ~aºÄ»JTðúäì"*‘{ þºŒÊišÑþ¨rÃ1ˆðØ^Tµ{•£*~ ÕUÚ+0ò Ê®$‘ºÍ£Æûý ?ZHQ³ì¬îżfÔZžX˜¹ºu£Ïˆyüßñ¡]üCa4S”^o7D®;­Mg`͉êø1ùð_>ü—ÿ|è›bº|·ͺž7é‘AFóSOw„À,Í~¨UL´%­D(3«ŠÎˆÞö=±Hrr2–mýµ©œ·‚ì±¹›’LðYëXÌp’d¼âøå <:÷äàIH/xÏÃöR³i5úÏûnëÒA4ðíon,]]©‚ìÆSÞ#Û¯ Í‘%BSE‚Ä&54¿yãðÜZRï/Lj%A놣`¥!´¿¨±ÏpʇÎ$þÀU>~èaYwMw„¾ ¶Àg‰0M.)ðq;g:§øiÈ Ã°µó×ô/žùº@ÿœ5Lõ<QØþž FŸ0 QÀ\Âç\——°p3=»ew,ݳjßc÷–«zë{×aMW}¶8·~-¼ûEÔv¶FïY‹ïFÂ@¶±KÛ|¨müÃO¢IÙ®1ûe´oó¡Ï`ü“ÓH©ÀÖQ1PÔÒ>¾94…´ÜfÍI)Hwö¥-gÈK¤ïytçÈü+dtºWÔúX ™©Ùæ V‘¥fb"yÙr²TC–ãÛ4µMgr“žÒ#š Bž}_ºîy o£ú½°lä§œJ>ß{ ßã=„‚§“vvlëëYr×t^е˜Añ‚ý2ŸjýPŠþA™è3”ögSfU*EYZóäKÊý(çNëÑn„ò[ä½Þi­¨XÑeÃÇO‹Ê‰_¬-èQ%$g¯u ªÞ©-½_•ƒê}䱺i¨aò¢‚ÄŠ{Ûçî~¾ßÚÚÇÈãªP§<Š™°(õ,_,çX„¡A•#Aj>6îø¾5®ŒF•WíIBÝФâóªwâE4K|awp`ÍˈU~†»¡EZd ]´ZF4VÛ&H¢ÕCY¶Œ¦Sh}µf_·É1´9ä'Å$±Œ¶ðÖέPí¸Íåk{ÐnYKEŒL íë×j±÷¿í’ÜÎ.ºçK$]ÛæDšÂ,R\öeN°Šþg>,°WzPN&)ƒ¯ç¡Yýò ÌCÐÊ”|NвÚ¥ó ~å@çéC7Ö4ס‡¿,쬪ô“çØrçÃ@$ãÈ7n„!µe^~â>ø¾>¢´I#¡aÕO}`ôËLo ß̻ü= ¦úËÐQèÁLë¢uNk0Ì)8R‚ù'J÷>=: ‹/rùO¿¥TnƒZX¾Výâ@ÊcXõ•˜z{å¬çö,úë&À¦x‘‰Rü$puÝyÄD ÄŸ[bUDâ@ªó]HzôbxÄG$?éÏvõrRÆ[V)¶àuÜ)©ôäÒÂ]¤ßk°dÇ‹ ë’ìmIÈ46vd Y”t¦¾z"{MþmÕ]Èõýi³¼ÇòŠ0mäÞ(G¾­' lyÇ·s-sý«#(lÃpÿáŠöÓtø]‡µšÚ´ ?PšéžÓÆÊj¿/0¨ByÂF××óP¡¶%Ûo" •¥ªÓ"ªÌx[—ý/¶sµ“× ?/Ú¨±G[ÚH÷ÌŸhÛ«ýq§ÃU´K”å)`ÛöWRËmîà~ R<~}Ø+´ÿüÏ|è}éÇŠtø>P°:×ü¡Ë…«:PuäþXEL$|a•`N} ¥b÷º=NC±e]Ù´ƒBλîe¬/!w¨Wšâ@|NÒI$|ÿ²H"u%£ÿÔœ•@8õr JîÇϯ±ºC5oåmû¥>h8·öNt1tÛçèP>2:S˜D´Þ@ß%å¶³0()ÂÙ0äÃyümTê0zçŽôôòk˜P!t¼v`2jðµ£4LV—?o)œ†Éýd±|[0)j ¢|âoý–üƒƒk¦wtöì[€™c{n±6Ù}¬éõá™0GMâ=sßÏvå<…ùqíK S°Hyèç5XRm¬ζ„Ÿ!ƒØ `…–.aæ¬FðÄ,×ÎÀº^\}Lë{øµÂ2òÜl6kRÓÕ ûàH <$#¡Ë®×üª­Hä_q!¾‰3ÜÒÞnú_AÚ•Ðïó5>Hçn8ÀþÊé½Äh)ľ#ÃqG‰t6dLýù A)™¦oC’²’ .Æq"k~ôº|îOd72o JEŽ![á„xä\¿Qø‹ÇÏ*@y>¶†·è!o¢q‘Ôè^ä{·q'¯ùß\éN»‹‚g®iǧ¡På»~¯B ºyTùü93]p÷N@ñjý’îO,(±EU!>R2šQ*’T¸;臠ô(Ëi.¢ ÍD™ Y ʪ<Þ<ìÑ‚²•¯»H†QΜrä ïM”í¦VI:ò‰&¤æRPáUT•Q*9*í 1tðýŠÊ¥£’¹B‹¨(øêfR<ªž< óxÛßÔM÷Z[8ç!î‹~Æ<52šÏËP¢fãà‘«n„¨MT?¹ˆ:{3Ý&(þ¿ãCê1 &hF(Z©FÙ\åît3‰Àº·˜¨f}ð¿|ø/þˇÿ >äIøòAÿ<šv=$Úm¿/âß Ÿtƒé•Jõ« æ`KmbH”w<êW‚¢ÅD³¡ËEÛ9QrÉú³d?h໘@ŸÖ¿²—þˆ‚¬Ûæ>ÍCƒ¿õ8"|Ò=èô}Ü!õÀ|æAÒHQ¿FëïEé±<Ä .zm7ÅSî¿£pŸÝ¨˜{›O¿µùD¡Ûºš_aŸ¢†&¢ãIp-šéMiݘOC /´3¨<ƒÖ#v‡šå µ“ÆÙœ¯Z#YwÕþž|Ò_¢š}‰î³@“uSß="6h¼õÍCi÷ f«$Ú@cØáG—ØŠ Iù¨Ì‰ÆüßsK5É6¡¹x=H²ÿ´|›-þ’Üm­$¡æ ý“ÓŠ„5tV½x‡K|Š ×Ùo™²œ…¢®Ùþ´¿ÏiQ—çüˇö¯ß“ž„²ì4W†¡/ùp.âÝ€Ó?óa±Þ³1Ðx^€$¬Òšõö1—ÙÌA+÷kWÈê¡]ñí‹ú‘ÐyÖ´V§szDãƒ6\£ Ÿ&"e›¸a öGÐi[Ò±ízµý= Öt][9 #Í´Ž £0Zú‹2ù Œì3ήԄ)ãâù-Oq˜©”¼¼És2ŸÖ¬j`þ©Äú+‚X¤÷Vl ”ƒ¥|Gͱå,X¾'.é%'«·Rõg§ÞÃz,Ë µ4#ØØ¸ÐwS öu’ˆN*!aŽfAh–>Û¬ýrpBRþ·¹|Ûù’œ?Æè`^R]‰|Uû©'yLû¹ö[€³`ÑÒ“^(÷B†¹ ž¯ë)òáÇB:/di~&ï(Á‰ìîŠå7¨Ž"W@ÔÚ×zÈ3ëqé^RòÕå¯lC«ê7´Q˜áF‹â¼Š>øæe¬¶cΈæàîGì$×¾ Ì¢£›Ê%”UßhïA…ËÁ¹jɨ”gSÁé,ªœÚ—ÿŒ^ Õ”‰ÏÊ~‰HœkU%ð5ÆiêùJKQëòõköM?PGYvŽúZê±Êâ%©Aý-Ù÷_¢a'¦) Q‰g4šÜQÊv’ÉG3Â#Â1ëhîó¤ãÔAw´Øñ™‘° ZJ>¸SÚB†–k®‘3ý hU²”KV‚Öw…%g =ÑF{º›Õb m~]‹ü¶)«>‹ahwäë—Á3/ÑžRÂõç;´O±ºM¼ùôŸù°–;/hò3t<ïu;1 »P6|*gó»#.ðE8hºÎç”*šÉû|Ï„â#!VÎô?¡Pd_MåS[ȉ2¥´g†ÏŸ)B U!ëlå±ôcoÿÔ\s“5µ0( 4‹ÉájáÄ£.w·s¥ÛäÁ—Ž-ÐJº¨Wן&¸ ç$¡Ïgq+\apïpËÇS÷`X¥‡Íî\ ŒÆ¸Ÿ£r,…‰“¦Ë.c0Ùúô9YPLn-ÐØŠ¶Âdà矗OpÂäáË"=.bêÅ·ÔC˜Ö_ä‘¶ò†™·e|òÝ0ë~üî;V˜v¼ÞϺó±o¤8`aç†rüµý°(G'mh` KN¤ö¬åÙð³PÈî ûXQ#2Ȩ‡Õz Bª7°îÉ1;Õ}6$w¼7¹2[Dcú ÎúÛÜ-–Ðây ï¿>ÕY‘‰Dñ¢ÏfMz¸•ùeö¸#’²žpã¨A²«mçŽE]AòŸDKwi‚ò9b5G1RˆŸh¹Õˆ4{v‘|UÅE¾›ùžH÷è4ëôÏHÿäÛÂá¶.bL:´%"c-kÔÉndÞñDgù…²œ¡ÚQ;œŒ¬ý·]UØ‘Ýõ<1ž{‹œT¯Ø~5d"½®Ÿ³4=î2êÄ"+äéå<@qò-òÞ=7Ƀ|)×»—sB‘?EöÙà¦4 `~ŽB|+ߎè H¼Wùê‰s¦>YcŒâÃ׎F’1¡$ûƒ¨+KJ(e¨yˆ;º wÇ<Ó¿J‰Ò›Í&0”2dªO ítPÖ ùì}bY”í~*~¾åΜX wl@yræñèG9(ÿµÓþr–)*ä:vÑn¼B%åNöpJTn ²ots@•?¶mŠU/±ÅŒJ^Cu‡æçJ ?à[+}5J̨'ТæÀ~w/¡Ý¨M“1B½­k,5ëðù?ãûßß¹œ{~ MkùB#JÚKöÑð÷#°Þ3©Ç«Öý/þˇÿòáÿ€}óbà8štÝ~JÛD‡ Ä>™¢æ0ÕŸôº#¾úèZ¥‡RïéÃÎÜÛù´0+¬pÈrû¦íål©ÔÒl£Eø”)Å?ÿŸ½‡¾ýî?\™³…tY¯áèñƒ²šl•¨Ì³­Cå¯Ë ?!ÝH{²¯²/ë÷ðÔi@‰Msù&Tê OK(AΩo¥"K¿çɰTò¼‚¦ã-H8L¡ù¼?¾¨= -Þnß.@kÞ³ã6–‹Ð¦xð®£ê´nÇYà ~ÐJ_xRê947Åý¶k‚¦W7ªÝßCcnúâ›HEht‘óЖƒÆ Ï”pzhr;_UÍL{m鲦 yØÒ}U¬é÷üÃ?h¡mϾt³ÐÞü©Amf:ЈfAÅd•ÄA&ègS×wƒÒ«¦>†0äkà|è˜ ;ìÑãcàøË‡¤­úE™0ÙEàZ™J3+ó•ûu÷Â\´U|˜5,¸\¹Ksn –ŽÚÑŒÒÁ²Ìd.í]Xýñ¥·æ|7üºùíô¯ÄEغ°ËÖé§,Ú“ˆŸx¼Í‡»hâýšï!)¹µ¯N%’Ói2:¸ %šž\¾…Ô/¦=¿«¦ -O°J?Ã&Ò9}¢×?VŠôÍ–ÚTÝÈx€Óç‡(2m<mÌí@–2¹c\vÈ–ý„Ge‡6rtXš÷´ø 7ë±w?. ãÙlýÆKÈÛö«ôÔMägˆP2]Dcå¤%ïP°?º¬ÏU…£È›\Ÿj£è»k%]•t(Þ»A[p ¥ö–­Ü¤ƒÒ©NŠs•(ËÁ7Â!ô|Ûÿ¢ƒ´œQ~(ÝõsÍ1T|¬šq•ÜØò•&Cyëwük×Q•_R}ô⪻…ß2G V‡H³êÔ ¼¬¥p!µ~‰«©ôÍ ŽÃ±ÈÏnÔ­é½}óÚÔ?~PÐÏ_ †&‹5Јi鬯ùv4aw;Ü/êŒf4^Óþœh~ÁwïÅ#ùhaKßÓº€–{¹ƒ÷”£+óZ\h*ZMVé\J~€ÖÙ=”ªæhsãú¥ïpm÷Ë»ßkBÛ)¯Ù¯™Ñ.̤剪%Ú«R«1 ¡}+µ„û»Ã¸ÿL@ÿ“ó‚¸ÿ§…s–ú7¼VIúwý)겿HøÑ–…4·¢u9¿ÂDýoŸâƒ bÃ%Úf½ ]HUÕ@¶ÊåyÑ:Ú?û ÓÓâ¶ú2ÿ³÷°÷â|輠🽓,Ϭ_nÃH±P#íj1g™¯Ý¿` qâ”±a“¢Â\’i‡ó8¯9 <\ëI5ÿ³_ñ·Þe"Ó!oþæÞÚ=­·©-q;|ò©Èæš!ÿÁû¢çUµqó5ÉþÿwŸ…ë0¼dåΧMøÝGy4î=ºÍ·¼MAPt§/6aŠŠ›ŒŠõþåCËÜ=cß@YFÝX­rÍ_>œÈ‹&=/öÏ|X¶GsÌè4º¯ì7]`€f3ž íŸw¡UøÀýòœBhWõõÄ:Ïir]ú=»Ý ŠX‡~F? n…Û0ÜhjšáCFYÏ™ÖÒ`˜<ÃçÌ|Œ´ø‹·šÃh1…ÂËÂ8ïZèÞLì‡)õõÌéqZ˜É>ñTÓ¥æ¤HÜ=J`>ŒïüK²°Èq(Õ?å ,Ug:=`9Ë _|ü «7¢4NfÂú3ò9®)ØøÆçÖg9Œ¢z+ ¤‘0°“VÁæóÌÏ«C’áoÑâ6HÖWvC“Ä)YÞx_}‰Ô/KÆ&/x!íånÍ–GzH× 8éUׂ >{¬;ŽÏ"Óß[²ó½Ööèˆ@®Ý©¼UůçÉ—‰°ƒï·}rLïæú>xó¸<2x…ê'xj_®¡¨r'ÃZŠ/Ù”~ˆ»É†ž¼àA™÷Ü‚ºJ(gÕbx3s8ò²ß£Ò¾ïíÏqOåØgT õ`=7»Œ nGyój\Hjp\GÍa†³I^U¨ýtŒ0Þ7ê:Íÿ¢L“A}m•M‘n4dO£ÙäÁ}ÓuváÃóhœ“8èó"M¯^ø¼¶ÍÃætÅAé‘ÄhÞg:Ú!´-bÙå ®¡å)BG½‡+hÅs!dÂkZ}S\_ì£DkOu ÇІ#.Bæ¤Ú|,éÒmD[ý‘Ê×SѶÙú}c¶-Ú0m{qéùpWÅc{gèH¬ö:°øÿîQIЇª«,‚¤rDðerú=K(Õd"± ‚bg»A³ïO¡P~'_-##ä®;¹Rٜϥ5× nC–{ OØ›?5—Žép`¯ ”<;—du¨ª%©Þb¥ƒ†Ë½ŒÔõ,¿Ÿ“6=›FE%Ljr¡/8]ä¨2 ¾ z6Ãf›4mM‡`´Ò9Ièë(L<xØðLQ?Ýa Sм¾­óa²è¬î—ie˜|É%9ý§NiÌ»ž8 Ó·i"™³Ò42˜}ÀǤžè s†<Ë:&0/9äø¢t,(f­û°|Å¥§YiÂR°ò(E#üo祿+gú¬#*`uË÷àãrcXO|È 0èD'ˆ_ÍÀ– _·äý$x^Áï¢Û…„ñ\r!VO‘¨* _©, ‰W²o7´É ©êÂËýŒH~/4j$)¸®ê<3„”·á›· Rûf¾$ÎCGmZÜi:ÉG•ì†tïïêpWó"}¢GGvÇv®)J³˜ØdÛ“ÃÖ…ÌâÑ•©~ËÈ@»+ …lD‡©c²T‘ýÉçÌ8“'È)ÇÐÕ¸R‚\"ÔÞOsÕq×E-ëËËnÈKœJýº4y›=ŒSfI¯Ì2Ïù³òN™ôïkGA Žï¡Ðj½ÐÆ)½*™»Åü¬¹z P‚ õ×qF”TØák\x¥Ži²qÞÀÝù•n‘É´(CõÑO¨¤¥7Xê2ÍVQÖž¥=ëa ÊÎzQ¾ø‚r¾…[×¢ùQ^ȬšÆå'O%¾@zTèâl4’A% Þ±¶ /TSº3–W‹*± ¯iÞEÕ[mÜðáª_,1d_F¼F}N––5¾UÅM0 æÌ·àëQ›íy•4êè4i+1hýßñaþ´œ×K4}ÁáHìÝ€\äzS¹ïà—FÊñÍ‘ùð_>ü—ÿ|¸ÿm[ewy‰Ì˜"ýaÚr ÖÝ0Y¹D2­zªãÄó–Ž~Ÿ»ÏÑ=pŠî„[2A.g»+Ïͪ?úùtîÕÖ­#š¿÷š¤ü£¯ÿê1EÝ]-ò¸|pZ8å:ôç÷ٺʵ<­+tWùÌ…Y^ç Ô}­É¢É} SO¯ZJBSYÖRÕ§ h®-¹Pâg -­Í”3§öAGË™ÖKmÁñQm¶9ѦðèeFÍ?:kæZªä þ£ÃÆŸO¿•SúAã|‘ÕE÷ç禀†Þ‹м×÷,ÇŽ®¿z¥Ï³%ÏÆ?ïÛ>/õÐçãmè"¹×ZäS =çOn…+ô‹Š´Tláï}Å¡C?:`èS÷¸7/ ;ñÛ+OüåC¢Aý±)i˜l}Üóµfo¿t+©‡¹ˆ·¼Æ_±ÖI± 2X2²+š¯‡åÓTëâL_ÿèò—îç³=ôH¨¾*Ïϻ͇Ôç¨îŠv!ÉfÑçè_QÛ|™˜;Δ:“3™DöHØ}O‚iyS],ô„î„}vH_ȲŒVmmO¸iÙ.µdÉY [žÆHmó!=õ³ÈÑsçÒÛÆNäæu°Þÿñ3ò¸ˆŠlQ ïÞaÝ ò³,/ïE“}Úöú(8Tl¡¯L‹Â±ͤï¡hò·kík:(>cÙºWi¥ÌŸ$˜D鲇 ßͶuÄKå3~ åNÅšÅJ.£|+y¦Ýþ8T¼21 ÞKŠÊr’áì»P…˜¯ƒŠ.÷Îvy³ê¿Eõ½#co)éWüÙ'v  æç‚‹ÇO£Vû#[Ÿ5_ÔÙè¥J¿Šº÷ ªŽÑ÷ >Ó;2ýEJ4¨ùqü¸dîËUXì=üKL‚œ†Ñ´àlή=h®Wœ“ŠCh!º­ìB6Z’ž©" GËnгQK6h[QŒGyÑút@;qÉÚì:ëË}}mª(…ØOÔ¡íéçF7Nû¢íÆþ’ÅlV´»­IiY‡öä̓¯#MÑþ†xÕú×h?á7rRC¯É‘Üd!>‚:§Iå¹gqGzr¨5ÃHí¯‘wÏ• BÑýã²íŸú¯î2ÃHID¾9BºƒÿÉõóQüí¿:Kx/ç£9]qu›1&bVT‚™Ö!fǯÓ‡»!vèêðÎJ=HòNÎ/diW´¦g|ý}~˜L¬öÓ}æ‘‹R[@þi‘ ®‰–ÿíi¬›*I¼ ·j“I¹ßB³Ý’GésNh•–yÁr.Ú5U?÷8C§›ü™Î9Уpé&_£5ô³;;ÜÃÈ’õ¥Ù¼ý{®”士0LóRÄvrFÚúé?ÁhÓÑ{,ª0ÞðÚU;èæ_ŸMϼó©æÄ¬”|‚aþËãðSX Õ¦°ÔNù@R‘ –_Yf+ÁêõÏ¿‚zÊaýú½ÇïýÁFT#‰|lfLˆ‹"áÑ‹‡›]hl|èºê5$‰ÛZVÝÎǃiÂÆÊHQ«cv–çR:ÝùIдìÒìwÆ"]xÒiÆvdÐqdbr7º©ªŽ,ʶtúãG‘­†¤øR\?r6åë3 Ï""óñxªtäse ÍxY€§ê‹¢Pè¡óImo²Éíø¼ŽâŽiÜOPÊG­$Î)elÚÛ´˜]Qö×på­](_BVù;\e>¨—âÛ*fµ¹7¸wn²4ñŽ ª_/uïâGœŽ™""pBM›Øƒ|I¨µÉ ’‚:YÁ ‡CQïË%·Ì,4°mç*»‰ûăî(Zj¡Ñõk¢l4‰`¬gWG3ì°^Ńæ'%TO Š ßžQ hÑô”ﵪZzO¸|“#D+ÝêÕÍ0´J~𮓢 ­÷t1ý¨DëìÙ”RÑF–°ªññ´‰¸ÞKeG޶¤ûx,þ™/¸,(ゎ"g¹„Ш;N÷Ú·òO.üïuúßë·ø*—›â'…¿yõÿÓÏu›uÛìG`]øŸú_?- · ØÚºÕrn¬!±ñÐà^_}]ÍöuV½ Êv6„¾ˆófüâK0hÜÐó|ÜŸ?<®©D ùã&që¿ÿq_À¿ ¦œMVl×ÞÃä"÷Sv+˜¬¨¡skÉøS§<™dÂÆ`º@8ÚïÄÃß{=Ó \oÀlâû²Åô.˜;Û{!Î÷9Ì[ðØoþÍ‘·¯F]€¥g÷ÕÞ»°Ìrôvî)Xy>|eÉ}Ö¤32{‹Öa½gKÞ{T6BVø³‹aëü7ÁDV$øxâ%1ƒ5V¼Ü0ê»ò÷ºç<ì2Vº…¤G?QéÝd%Ê9CâÓH¡µT¶û‹2RW¨·v³"uBÕð‘»1Hsÿõé÷æ_p§k!KĤË{w%bàÒ—ñ>I} ]LÕÆü ÈDº¿A-‘Ù ÓøÌÅ d‰gì¾åÐò'W²&:M/y9÷ßÏS\j nÅúgp׫Ⱥ„ÝÛ\Èfé¬lm‚¼c´ ³ ãÈ×!òª'kò‹'éy~ OKúgßü݇çó„"=n!×ÍäP,RUËõr-Jp¿~Ç€’¦ ”£”·FýHÈÜݼöÚv„e˜7‡’£Pz*ËmbÎeOí×bE9’ÛU|ŸP.œ“zîûe”׈w·ú}Nq3¿¢Öú6Ÿ†•£’ó©FæT^ïI#¤PA•L×¢µK¨úèiÈÓÝQÝÌ Þ-14oèd+jôéèÎi² æ¦~¼ÏCÔðnU½…:†CJA¯ˆþïøð¹òŠüÛb4=É(uJ¡9{õc‡_Ñ}zX/ýˇÿòá¿|ø?àCg¢¦°òA4:Ê!3Kzñ/ž ãç%«€jޱwKš ä×Ë\q¬‰^:Ê9G_ÑgFÃg¶Ñ#®U5¦}U‘2O¯´I·oBÆ“Sb@ÚÞËu’ ëð¡û¥Mí-]HR¨ãPç‚´3>|Õ¥às—‹kT;”HÞié°„J›ä7 \jPçújÖ†f2ß›½"-†¦»V¯Ý|¡9¬¨øQÞ´$ÄTœÕÖ‚Öé˜Q5H†6Wó\.{×?}¾V{W¥êDFh ð¦5‡¦©9ÃÐïÏ ‰Z.§ã'4¦ŒûNRÕCã/O÷®†ÐtƒÊ¢ºBâ÷Þ´%Âlhn}ý3ÐR»Õà“øÚˆõ¥Ý#¡=x%Í¿pt>Ž¢Õ‡îï£ä÷ˆr¡/ÎÒ2én ¸ø4WD)À{x0Ÿæ7ø>ZÖá=ù F6ÇØÃ~ÅÃØ™S{ßI߆IáÖƒq˜‘âœ×ö ƒ9Yÿ=‘æ0?eèKÿ¿ÝìÉI ?¬àWaUòEÛaýÃ¥[=}°émÚÔá ôœò ô7A¢ˆ½¶ÞV‘Ä4!jBÉîµÚ˜ö¾@Šd ÁÒHµfæ,Ҝ߈{y¢éÈuIJ?.!ý9’w?òñÿ¡ë-ƒªú£öoº;$¤‘n¤c-ºEQT QQBAD,TTP‘îîPºAº»ôá¾gžßýêÿjÍ>sfÎÌÞßϺÖ5û¬µ˜£[w=B–ˆC‡9·žãóò^B d©9¥{Èiï 9”Ûƒ\‘¹ïóõ{§sºØËŸùž[·)«…"#纩`änYÂÍóxèUü«kÿ °}èÃÁ›(juúð©|A¿{;˜þyJö\Î ?Ÿ‰2ºµo½ÿ¼BÙ_ÕZî×#Q^ððÍŽòèny£ŠÒߤÚQif¸…,êªä÷äÓki£Zb‘S32jÌšjÓð#¾ŠÝ”`¾‚ÚªÛyùõE¨óú¸Ý!EÝEIƒ º?¨o¤ð-k Z¯çù©6¢ÑƒÄƈª7hr-[®3| Ín~>uQµ ͽ…Åä2£¯ÍçÞhñï쥳hÙ÷>P8Ñ­RÙïLJ¢õÍã}Ÿi£üZ¸‹¼mæ^ˆ >@Û˜«I¡+h'oSúê ÚU‡ŽÚ”£½¹Ý¥;Y´o½å@c†:ª¦ÚôAèðÝE”á49úõ{֙ǣî¶>­¨Q¤‰ÿìt1â L^Ž>6æ‘5I´vìÊAPB3¹P/yäÑd\JNÕB¥ÆÎwéöÇܨ Ü׽ǣßÁ·§êZK=sðe†ð»°î,|Þù§yo>iP}½‰Mw\à34ªÉ•'Aòâ£15åÿé?låÞ ¬þ!ñìÿ{øÆ ¼—ÿßïƒÒ-&X¬þ/n8ý ª‡’“:!¹Ó&ûú*4l=¥q[&.ª@p›ëú@Ù)k]jÂÿì͹2,ø*h?N7FC•äÍ%’›†PS1uz}j 6?ìûCCÒÜ û“ðãñ‹üÐX×ð¬Ô¥Zµ¹GÞ\ƒ6R1µˆÌý|ÑÃÊsM:_¯‡+Àt'ZÑPY CïOÍj—ÓðëVmÌ;%:ëˆUÑqÑËfý\´ £}‡oÐe5Ãøãº§×’`Òî[KMŸ Lkf9ª ½Y³ªÓo"Raþݶ¯Y{,1][’è*…åÞ»›ÖcA°jÐìq3OÖü ^sMÁÆiç/Öð[1ß$CжY‚_h|ƒ?,šÇ—²aOãu±´üû|h„MË %'r'®o"Q¦ªž?„$n)c¥^ÌHfÑt5 )¼ï _­{T6 Ó”Hû:Mºq·l%*,ƒíqƒñÔÓàmd.jùx7ÌYÛþ0—|Fv뫵åË_ð ORNéwä ~ºôÙ ù‰-3ÿŠK2R¡(Dé>I¥2„"vJÕFQlõ,aÝœ>Jž9°2»Ûêß7Qî=çhE <|Z™´ìß.*zç:°x^Aåñ†EbžTT»JwÞRt5M©œBt~¼œVvµ©b¼”ålQçOÏTõÔseŠoyt d6d¯]AÃ?G4œ~¡qg&yš>]"Ÿ½€G ¼îþI¼…æã/ì…Ÿ¿F ¯3‚ÄÏÐbãΘAq*Z&ý{s;óZY…k•üF«U¾RÚOh}?…sg.m˜±Jòu4Ú<“ï]óû€¶´ãžŒoÒÑöÖÓ‡J¹Éh;þ Yâ)Fôò(µ0 µ•üã¯Jä‹=jUÆs _ô„]ˇžá§KßÅböu雦¸¥Ô-æÏÉnîBõ׿7œvõ ¢ÕØmÒ¬Êv"c†©ïŸóqY(<ìºc÷ýñÜfײ0Ñ.ì¶¾²’i(0b^áWÚ z®àn³u¨?ñ4Ñ@(Zú›wn‚NÍÓ3Ò)Ð[æê] ƒNÇ"Íèˆaäu¥sB3ŒßIð¸S‹1¢3U0íJÊXDóÔD:âŠu0oL\ÅÈÊ s땱u+wanÜ÷cøpÌí2lj߃ùôŠ˜¾dX˜¿nÄg| –tÍíy3:`YU4„%Ù–W³kÓŠ`¥™þ é§EXm.¤d1 µÇ^ºÀHØ8èâ舰éÁ"îðµ~iÙØšþ„mOž*¶žCð‡éðE®èNØ-–åö‚¿1…H O¾z¦3 í/qмE"ß[>S$N>ÁµÔW$+Õ© Ô‘Ìf¯®9ŠÉ;÷6›Û­‘òzfæ1¤VØÚVø*´ôu¿nÖ#Òíöv¹@†@^%öc$ÈxO¯f{æ2Å•=òxú™«¾qºެÄOl®ÕMáçÕ¿=Ï [;¿ur¸h”œ¾‹œ»m.scñà’ëÉ=äQ¾¬üYªyë%h¸ "_æ©KÏÈO"¼¤sÖYxNí<½¯ ‡(gê" Q¨|ãÑ!r 1‰÷ø†bêþ•7P<ƒ£?’íJŒs ö˜Ø£CSJ»=wó;Ç‹2U+ÚG©P¦œÿm¡œ¾˜QX.;ÊmÛ]ÝòBù$ö QŸõðð%M'£ÑnT0\nç® CEJBjžzTJ¼U:ùPå°øæm¶=TI½8‡ê…7ïS\«EÍïZU&lrˆ9‹¤C·Pkþ¢…7 îןŸâ8YQW©ù~…­,ê+½¬’mŽú¯$ê•i–Ð@uΗèÄ'4hâ æïBCû§J—¹‹Ð°«dƒ2š Lë:öI Qî…îajy4æä uÌÜ÷…)=äÅch\û<Ò M˜õ¸ýB‡‘ïnf¡Iìßw"å®hÒüûkΧ#hJ<î¥nŽû^gÜÆJÍvUµÎœdǃçÚLÆÁ®Žþ§|Ÿ#ÐIÔö@9r~Ž){wχ*¾¬7ý¹ÙPbòèËa(üB»ú-7rÿ˜‰i2†, j~þgHw.!´òMúŸùm«ìD %0@'A꿘ÆÇïùÛ2ߊ»ìT-ÿÇkñ÷z¥á”PžÌ–¬h9 Uw«dy–³ &¿¨¤äDÔÝmqï= ù£—¹ŸëA£Øˆ¤sÄ)h–Ùj§J…–×Ô¢_‘ZÝøo¾;9mÊêþ~tÍЮ¥Ù ôÝ:\º4+‰ 33Œ1œ.ºm[ÈMš‡ çû1%Ž1èëñ0úùâ P›—¾)¡d‘¶Ì×0bñØ÷ýç^ý]¥Ó/ãCcEaò’×Å4­,˜Ö-2úS¥³b¾{ïËažŸ#’ì%,ìiË;¬ Ëb©ÞGó `å´ëÎÁ7³°Z¦ªÓK9ëÚDçÎÃÆ·Ë±Ü·à·û7á{öî°Mœœìã;ÙÞ˜x‡Ý-£äýûý×ûï­µ$КUòq½ˆ„!›§d"Ñ-é‹D=HœM™Öz< IIÿ˜*ÓG2Ï!ÿCÊ2H¾§ÅºÔp)³sdÉ×7:‚j¢9fiïl½Ø³‹ô9¿TSõ #U&…ØC^d2äꦜdŠ8Àÿ‘ Yr×o¹©âjå\Ǧ5d‹S‹q}Ê7b’o³àÁÔÖ¸à˜häáMžÈ”4B>ÝÙ_ñ×E‘ß°8Œ.ü* °p[¨àF¡¬·}*](XÇzýhÐ0ªq<]Iê€BsUÚ¦¶PD™ïêYé|}ùj¤îÐo'›:Kø%B×Î5NÆ¢]Ÿí±(`£§X„²/ݶÂQn HÌõG¾2ñ"ÿ‚*jM^oý‚Ê&¯Ÿ DÕ§™ïÇæ¸Qƒ¸ÞëåÂÉ`â'_QK!ÕãÝ[ÔæÒÔ.¨±AZJýŠÔIhãŠê¿ƒºòù…‹¶ ¨›3 '%§QOôƒƒ¨÷Q–½®ˆõiMóEúQß½ ìâ›Ô¯¶²íø Ì%ÞÞÏ£§|]û¾ Aâý7LB÷Ð`l¦AÝ ¹^ DðÕ£¡…²C §Þö;³Â^)ˆÆ£ªËEãÿçéÔѲê#’8IA…ñZ`̃Q(µsŒéAÈÿèœcZG¹—ß_ÚiÚ„!¥ûîÔ3UexòÀ?#ÈXÈ=%érÒÂIدZäÂ7ÿâË;oœ!™Ø†©‘ÆÒ¦¹JSwCÞ•«Ô…—ó üw”¶×ž?Ô±¬/¯Cý%ñ8wh½•íVšo íÆÔéW$8 ãL)“ãz túÞâUƒ®Š›ç>Pñ@·ñïöÇÉÿú» ÊYœFº¡“O•ZjÚ·ó¨ûw ·9a[ò´•?sçW‹€v"gž­´Ðî­³~ýàõýïåyz?‡Ž„0.Ió&è ×5ûófºÒšOÏþ"€™ÊÎÈúyèã)Ò¤ÿw ~9^Œ-·£€!›ízº0BjË’|ü1Œ&Òu ¸ ã—.¦¾{¯ “×sél¾‡éæy©WºaՑCÔ°x6Ë„Uþ,WzZD ‡ÕEćb¢`ݶÈON× 6ÏOÜM˜æ‡­Do¡Õû­ð‡uúÈ¡û°7À@|g‚ :‹–—íè0Ù¯@Dj“¸÷½-HúúïH;;R°òú^:î†T¶‚›æKo‘¦Ôd|ìÎM¤· S´A!d¤i±`É»…LaæN윻È"°œíýgYY©»¯"[¥SuV—=r´JRpÇuáÁtKì]¿‘™­ò~Ë#Ÿ©Kªû'ä¿#f÷¤¶:+^‹¼¾‡,ƒÆÞO£0¹]Ýý (²±63{“Å9¾KŒoñ£ä)ÿ®‹·iPº’ˆ‚ðó,ʾ ú;gžŒrƒeñÞ/©ð°IûŽñð*L¹û÷èA¥÷ψ׉©QÅ+ò%‹g=ª9¾Ñ¸—׈ÏO~¡ß×;ĶKœ„¨Mü “P–uNR[R?@Ý”ŸLUA{¨Ov!Šü«<6¼ø-è é¹*ç÷AiÅË?óΡ™ ¡|M.š ¿Ð0Ç£KÍb¢¯Ñ¢ëvÉ3Ð2µW‘Ïd­BD?,›f£µñÏ›khCöíz’Y"ÚäU¯Ê׋£íÉA‰UF´Ýe:œãŽv„Y•ŠhÏJ;w,í£Ø™nš ™áÏ{ÖûñŠš[áô½ÛÏtc uÎßM5'nGšûôš+0I/Ò( Ç 5œ­Ñþ7’ Ø"1§åE äâÓO*mûºGïEáu Ò T>—±ûßþ^Oe'H. Ùº©»_$[ú©^2ÀgãT‘óÌÔ_UDl 1ñ_‰£áÓ+ÁHëB*H>u^Æ*çé>—œ™¦ŒªõLa’ï~æÿõ>æ+%ðÿî?äRýw¡ÿÿâs5Ÿ3´Cû>ñøð1þ}Ýz°ÄF ¥ôÕÑ—âÞAéXL[*”…W½8Ÿ ü¢ãJ"-Txèø“ t@e2é`×b¨üäZslß3© -@]¬Ï g 4D%ßÌ F‚$¦ÒcÐÜ—GÈ3 ­awkC»®|º¯¥=tRäù[ÑïçJÕ_NuvÐ+b䥑Eýµ¿r½â†Á8¡ÞØ­ynWù¢íFOY¼~¼vÆé7ÿ=mp„‰.û””+0Ulc³ H3µsö¿Á2V=£+àŠVæ¡¥nÈ'ù2[ò’íþs)3µ²ƒ‚âo©DÂÐ*tÙ-—wêœ^°jn@5 ÷ÍÜPa[@öï—”Ž8é@qP¾ðï}Î N¾»ðêäÜfS v’g@¶ÀYïÛ© íèuEq(žÚùþêÔ…çÈo} ‚lz#s:ÜÊ´J¸ '=R0ß_Øt›¯¯0Áðà•æÇ·`,iR?¤S&[ƒ€á/Ìœ˜ìlfÙŠÙјÉü! ³n/n^Ü¿&Qú8s²N2ÑAŽ07ëµM”ä œ•ó…ÒûºTnTþ –"ìÝ ?Á²oxßäଘý{¼>«z¤VÍ}S°fŒªÖ`=ÀNÕ¢s 6*߸K½2ßÒ¯å³y`ë{X›Å’9ìœ/Öt…?s=š÷ÿ.À^þÛG>^.ð/ÙîY˜ÜC$˜RWh<‹D”ε䤑XýÄÏ•§HtÑ<–I§*» 7ø‘ÜLÌkø2R’Æ‹ÉÎv"U©(-Uc1ÒĬ[˜iç!]Ðëå$Ñ*dÀ—›Q-ÁÈht¡é”@$2p7$qCæ;[õ~º]ÈRâqü6/`jã—mRG¶`“ïnãtÈAqƒ%3ç+rÆs JœÀƒïÑõåCîvõø)fä=•®ÛдÏÍ­¬*UÎä?/Ç`œñœ=rÖ¾„£`ÿ”I÷Û'(´ÿT$ßFÁ”ç„Q´k(ë;Š™sÿŸF ߯T³(ùÉ'ödPꯂâ›Iq”9uPܧeîÕ3Ñ…¢œÀ¶©¬ÊM…Œ95‹¡|kµÌEV<½BËémˆ !óŸo¡bà'âÛ ?QiøiuÁ7T¹xSýŸHª±|дFõi'?ÿ‚¿¨Ùí³Eˆ½|9u¼œ¨µ带)‹:¤?½G]Aåã—‡PïÎëÄ‘æW¨_º” ’@Œg«EÔìÑàߨ7E¾2>ž›ôþ|˜O§07~D£ïúqh´Wg¼B9Æ®3z-%h\9ô®M8ÍFJÎj£‰ÛŸŽ/Ð$=쇌Ó74Yó|AQÓ¦2 ï éÙw#gl¿àݺ…–ëDéhúô¤õɰ³È¹bI¤ÙFØ<ë3}ú°û½ðc’7‡Ýu*ŸtÊŒ½€â r¸žÇBA§a×ÁŽ›sÇÖz›t2Ï8ÎoÈï×£-±%ñLðÝJu*•Y¾}­ô<õS⿘êþàÊÍ„LRÍ—ƧÿñZsј++ÊÖ³çEÿ®ïëƒ`ñKW¨®óûzö¾?ÔÞ¯Z“Û×ÑúDCçi2>ø±—•“të54þÞ ‰#¸ Í«–QåÏÍ ¥VC6ÿK´>³68» mo-¥<Ë =Ÿ˜³4~:v·êû’Z ë~²÷ÁOŒÐÃ~´õåÛAè¤y¼6ý›.æãó`ð*Åü‚Ù9~/ï{iqFœd?üƒ±6ñ·÷Õ_ÂDXSzÚ(LYÿmÔ‚Ñ‚ùéSä0Ç´—(Í Ä–²á«Ô°$7ü"SÔ–^’­´šU!šc§ŸÃÚ[«¥Žî°!øMÊ©è%l–^ýækv ¶Žm† ´Ãñ#9zêzøóƒ‰ó!#ìe0»Q߇yole2î!ÁrãÑ—-HÄøõðÇU$6ù³y£tŸç—/õ¸ß ì×§ISä .H~o­Þþ¦'R*3ÀXv,R“±5]ÎDšÍ°6º— HÏí³QµÜƒ Ù fgy‘q’¡7ë~23}:sE Y°ñâ¹äd û™=çw Ù˜ c¦WÇ‘}†Ž»^rè>L¹ú¹ƒ cï\CÞ!ýOë¾!ߪ„脜òWÚfo©Bˆ1õÍ—QðFFú/KE<ä«ZÚÒ| …>‰ô‰<ñEái3men(êÖ})ðŠS÷Ôø=A‰ïÏϼ@)³{ŒÍ‘‚(=Z˜™\²QYI·N§£Ü8ÇóÒûxØÁ5¤¡ eùK‰ÕæQYÎA™3 U/ÿÐz5ˆêmGbCQ¯ÿØW}ó\Ô7§zÖñÀõó¨ôŸ¦x¢‡·x [\µØ"Cƒê¸´‡XÑaâÁ3«y4´Ý5Ô;…†1ÃÖr·ÜаIƒs½!ƒXùån× Ñ9G•k¶ÿùù´ÁãwÞhA=ÁyÑĵ{P>1?\ÄÇ%>î:«&7Åç82Ï9%b‹[Ù·ïü85Z Y¼Ï&ÿ EC†1ù‹-ù+š)\}õí$$s?)¸_ZïÚõ?†´syÎO!<*™Rí”W¼N¡øÒµ‡S›4@cºšëXÆ/hÕ¯¾3î> íÔJôzãiÐ!]|FDò0tZ™éœþ#]±ÆäÐ-üvÞËÙå¿9£]Š,+ ìqÐÉmÈ:<פ·½BÚO@»‚{$³f7´µkÙÚ]4ƒv ­ê§ ­Ð~Îÿ9¯g)´÷ÚR¾ºÞ7·‡ŠR Óà^Ýõ(éÿ™ƒãúIë$t7=ގ툇Þä./å5èo^úóÅìã'ž# ¿µ‰<0£²É››eal‘!´Òv&æC‰o.¿‚i¥+|ìYâ0[ôèÓ? Xxkö~rV –mí.Ü•(†U"éÇØla­oñôÌ¥‡°1ÕVÚò'lL|L4Ù;·f\#=¤aOî#{¾ÛU$àÒ#ìq8„5m‹g“VøIäãøºsHê£t­ö9’ßR)1óFÊÔ¥;ò­HC÷‚_òƒÒ½zpŠŸÅ’ƒ_ß ÓD¦C”ã*'/!s^±rÿÃÈê¡Ä )_Í(,žÒG9Lþµ9wàÁËç4n^`Cî$Íi«†Èû·ŸCO0ùUuÈ“è÷9Ë2¤×JÂC'%›»P(W”Túü1yUiž÷w8ä Äìƒ-“®(­£q%ÊF eíÔŠ/ŠÔ¢Ü«#F=G™ñ0©Wå¡@Tˆ—¬ÙU¬@¥cÆÊ#×£PEÄÀ—ã%ªÑrÃWÔУ fÒBÌúžÌ÷µÙRŒbÏ¢Ž€>J 5êz\¢Tú‹zUqW}^¡)×`µ-¼]?ŠÆSÙÉ•íh:áDÂ1Gz_“;È<À£ñÄ4ŸyÐ"%Žñô¯ ´ e˜çX6E+[•’ØÍ ´ælZ;Éý ­;‹«üöó‡M蚸æåF´•0J=q^m«›?·9Тí ?KÚõ•ÿ:¾ŽööÍ?ÇѾÁîÈç¬è À’|.ñÞ”6pÌ’Fís+w|ª‘Æãµôm˜H-Ö:Å ÕÏšsÿg„kÐä“@ȹñÌÀ¦Ö2µX·Ê!-Ÿ¡îÃ%Å}ÝSwÓ5¾ _çÒ¾ÿ$¯xòÅEEødÃõÚP*¿ Ñß=WbíKª qÃwüz–|íÐö¿øO ÒÞß{vSh²ÌßÙOÿ¿ù¥õd9mÿßóK+=9Š]ÿ‹Åªü•VU¤Ñ¯:#Óeôò*0öDèR†* LØ…Ç\z SëGãõNÃŒ.ÓËp:˜ ö«J€…Q¡Ó~,aYuúúögXy«üÉÙÖØ7{¥Âz®Ûµ²'ó°évL-ý3lé0µØìøÂŽö—C?JH`×ó g‹ú}ø;øvþ» Ôþt§ð¥D"ý.õÊ$ž£,X­@Ò¼/\´缘¦#”³©ÈŒ×‰r. M|Æ1g¤CúG6 /?ë!£]n†‘\ 2ÓÞ±ŽÚ@V*}δÊÈvÚ²ãÄm äd¹´sU¹9Ÿ,F>™Ÿ…?‡ï ÿßNKÃS·ðáóÁ$ÖX¬Ê‚T­ºô‘÷ JÄæ}\û£‰ÒïØÌäJ£ìLHF¶ïM”ÿðüéBI*$š~mû‰ÊMï"o¶¢êuËïqW¶P½×#µzuAýÄÖqÔºvùU¢ÅgÔ¡ä:'*fƒº%-eá3›¨œñÕ×c ý†Æ¬Œ†ß½ß¢IïC‚Ð'DhA}Þ¦xͥɞâÑⳊwÇÐâ-§yäˆ0ZZw<3üØ‹–ÿj§?¼E«÷ƒ%ÊÑZu™\#ýZ×Q ’‹¢YðÁZ–ûhSû%?2ûÚª–LDÞ\CÛdj78Ù0<ß&2K/-Mn±ÒÑ#ohŒœW”N•ðt®C7ß”[j_>´<“Õ]¼ µu>h²ñª<4|Ÿ¦lByŸxÇ¥ÏPF|}DYÊŠâ%þQ­AþÃÞ®…s²ÿq›eþÉ´Z”²®,+ï¼ †¼øðíC‰NûõgæÕЩ?P›ÓS–Ìš ã8f†‚ ýY…'G8t'†ËªÞI„þviúw|0,p'tçÍ4Œ6O¾l…‰jú©-K˜V·¯{9Ê3‰ôÊ‘a–ÌÓ)Ôxf6é_‰ŒMÁ,wG}Ñæ˜uýP*Ð!s‚¼Î(+ óÆLOì˜aá‚ÚÙÓ¡°˜ä´õEå ,œb £%ƒå†¥¡¿:Ͱ’w4î¡g9¬– »¾¶‡µ3Ž]Ë4ØàþNV›¾›Œ,kð{a}8œìlû½îvžZ…?lu&KÜn°;XöŠ×šþþ|c`’»‡7¨- }B—l*?ff‘èÇÉ€] g$a{íM€¤~aãIŽêH¶Kwœq!)ž=z;°ª€TÆö|Ã+ÆHÃÉ[Å[ tD'(½¼‘%—Òð/2R=ê¬sº‹L|ì iבY×Hÿå~dñ5v}x]YkÓuÃH‘Mæ©ú÷¢óÈž–~‰äO1rêžš¸R$„mÂÕIžì"·'%å{IäÙQö¿9*|F\ÒÆÈ/­Ø”V‚‚-ô3"QðYUòäw>âÿ¢M/¿ƒÂ?v ¾Æ£¨·ÜB“KŠ †¸œükôªŸµ ¤£Ï%yq”Šn%,?§„Òë½úÌoŸ£Œµ¹B"7ÊÔéÚ¸‰r½D^º‰c(_ÆÌß$9‡ Jæ sQ¡xÚ 'ËÂH7Å[Pùà—>¨ò¾:8Pùª™¼ŽÌêB VðX"T@ È.|xú"⫉Syê¨M–söšºê»ÞÞ;,ºÎ>“›"Q/þÆ@øÌ$êOd©©³¡Á“·×¨(ÐPîÓé¥×ÆhX£{Cv,ÈÎòPj¢Q3«ìÉ4Ö)"¢xÆ©YWOÔ¡ Ó•ßÁÏÑÄýwaU¹(š”&ûç(Ú¢)ÕY)$ éÑᨙŸÆhúø¤Œ[TšÖ:Y=´¯Å;#uEÏ2ÅÐD<ôœSf:rúu¦ë>£…?AsËÞ „û<äÆöÕAà ùn‡3#T’éÓÜ¥÷‡¢€áÜ Î( Î_¹ Ùmƒ%Îš× £äk^Ý÷˦{yW¼ÆRÎmÿ*lуänª8^WÓÿâ÷{­‰Mán‰5]ŠÿñZèGíäPÈ e`#Å•ü*Z³×a…ªn•Xš4¨ Ϥ— uQߎd\Û…†ÖZk]ð³ŒTÈuS š¾?j‘N» -BvÓík‰Ð²v‰K«XZwÏ®*Ê‘C;£Ìï¢ÜIè0¸"¡Ëù÷~ÍèA·â}ã=Á>èKî”öw‡¾‘+ývÓa@-=ƒ|B†Æ¾­º»kÁˆwÂÕ{\>0FмvS‡Æ“$·Òï܇ÉÓÝ™¶]†0-òç×M>s˜%)´U‚¹ÙúÇÎií°ÈpÚ¬fˆ–•®œ¿D> +a ƽ…&°úç#eP½7¬ûç_äwÊM&uywí‹ðûëeûÍG`Û2µªý«üa–˜7_X†ÝÑÔ'‚rðw霦wœ2„)$q(&' íéÓ Ñð¢À3$‘RÖ<à I#½­f墑œás@©ðe¤È»x¤ÔG ©‚ «êÌ<‘æÂï¶ ÇçHw—1½/î2œôpñD>d %ùuRÌ™™ÓÑ™k?½;º¨€¬§²„ÆñÀCgVIÈîx?í²r†dûõZ…#×àX·o_<ò*eÇ$‰#_Ü‹ºÂ2ä÷:ü@wÈ4»ÇΑ£ ÈÑE7p<$¹_'¾tB!—‡W‰ƒþ p¶‹–;Š:ßl¤ë¶@q†º„PŸ;(Ñ­~N☠J…;W}@~AAn7”ýÐr³÷ƒ2­ñð9uY3)9TähcLAeöjMí¿QÕxtȲwÕã¯~Xÿz5×οPD-ÊJgî 6yÀ‚PÐÔaðNÛóB—…þ‚Ô=ÝQQ¨»Ç{ü†DêEhGº',£>óõò‘ó¨E<´4CŒÄŽ„EÅjhàñä/I‡34çØa)Š•\Q AÃÝk췭аŽ.ßèˆ7Ñ ÕuöQ¢‘™Ö€ ßÞR2só7@ƒ™ÓU‰šÿó‡ÖÁ¾Ù îªÐoŇCP<©3–kÅÏõ·Whl÷ëÆÑ7Ô ‡6ãÊFdõ·ÒÛçGA¦ßøå Ó~HO°o©" دWõ âNÀ—â÷6Ý é]Vœü×Hmø¼¹,:¹o¾›Ø“Ý„òˆÏìñvPûÔGåûÇ'ÐhMùN‚Ôl¿Þë|KÉm.ìo ¡}»0síLt óü쇮‹ÏxÄïA7±)çÉÞÿöv ^h<Ꜷ|Oû¢ ƒÑˆx»¡Ú)}¥ŽBÛø<|E ít‡ÜC^<ƒvÇ,Õú*JhÏ[~ZÕcZºŸŠ?‚N¢Ý¥[ÐE=tîTB)t{8r¿²…^•å'oQB¿¾g¶ân ¤9ñÁ°ÒŸïV=):ü¥úÙ0Œyå8MQ½€‰3V[fŸÂTÜ&SÛf ̲5Õô™Â|×øvÞûUXjò'˜™„•Gï9y_qš Ÿéû§ °q¾ŒŽ·ðü~ҳɬ_Û³þ±Y]°{OIjqaþÝpºÝˆ„ŽïÔÛ?"1[Çíw9{H²pʼè’MXóçø!%U!Ã,I¤vžp¾^öi‡ï?Z˜y &ò“ëÇö±¤ÚÃác?2›‡#Ë®Á•;Ãx 1×ïë²WµÇ½<׆œÅaÝÛÊÈÍxá–ñÒqä=·Þþ}§ù–sØ¡€~âêæD –ÂõÝðJ:·­Ýf‚"šlv6ÊQL282ß%®ÆrЧ¡T¹O?/A8Ê2møÏŒ£œmï r¡|i'ÂZ±Ÿ>Gî öDð&×cÔ•¢N>²H€z!-v):½h@>ÛÌH͆…#4´DÑø½ât!³š¾Îê¼’†G"l}.WâQ+Î_ËYhq7×UÇîZ:}d~–‰V¢F}šXÑj®õ˜[F5Z'43›uAk¾ÓF,žh³™)ßPì„¶‘­ž •ÖµFk´ûà9ßèb‡öÜ‚Gî[¢ýc×f†¾a´ß\\`¬ÕGŸîôÓóPë܉cfrÆzˆÈ)ˆ&,¬2ß-/A5#µöï(ì4Ú —„ìÔØÉ+ô¥q“£†á¤ILÏÜ-eÜ×½*6§XøÊÎt>Œ“>gy*´ë:BbÐ`8½ÈeH ¿µ}†j>réJ’únB¢E‹êó”føjx°öŽaïÿ̾‘¨ö²((kØ´SÿoÿáÉW÷¦Èïÿ¿÷ž5§Q`\ü¿Ø÷»Q0ö8ó?‰[ùÚÅ$^Ó¼Q†â qBw(IOŒ¯“Q„Ò¬ûª9ùvPöyEz(g?|ÓøëI<ü5?C垇Ú7#¨¾hò€6fß*;sÖÁêÙþÓ~µƒ^ghÊ>„&Úã“ÃЭ>,âà mÒž³2ÔO¡}TË9"5:[)þfê@÷€³Ï;'Gè;VÙÉvjH}Vy½©ah8(~f9F<ÃÇ#½`ŒÉº»ås1Œw¦œ÷W€É¢kë*˜.QÕI~0»ªï÷BSŒ÷âø_¥…K/ ^ÀŠñ6u±L¬–)\ÈÛ~ ëö l«d*°I,&ÿ†Î ~÷ô»k; Û=' îIÀ.¥’À± 1øëç´2M"‹ϰ†c'‰è}åAâäÜO‘ /‘ÔýÙ¬í$?}49¢r)_=«Lói_N–Eº=ë,ÛQÈHÞ÷!ì2eÿJ޹&,‰³GÛ½E1Æc&æY$tž -ô§®ÙÏBžŒãëTTNHJ/;sï ti3 §ešA ÉjEK1Ñþy´q³_ï‡Ê&'¿Ô{-P~âµ¶ü(¥o£åøûaŸ“·Ý\>6—Ç¢¯RAð·™Ñ!¯…!³ÜoïZdã>g5·„?„²óóΊP3úÇÜù 4Ñ¿Z·?måU×oA×ËPüGìÐÉ~´>WÔ*˜:&ÂÈ\ÊýooÏ‹2L2Sô­_7¤`ÚûÀEЈ‚é !tu…éÍì}£Æ3j!Ç»CaæÅ¬y¦Ì:5–ûç®ÂÜSŠË[G`þëgI µ}ŸXyÊ{Fì,Ñ]7ñ±‡e£a¿¨RX‘|9iH«äÁ‘_ a ´w>7ºÁºWêR·¯4lT[,]«µ…ßR'+¹ãaë{ íßèÛ°£›ç`§>»Ä£Ÿë“`oaì7 x"Y¿»ùT.ж|jÊG"')"^$N4¬ë:2Ф´GWcJm,òÀãŒOH!ä˾ý)Ûï)ä²t õ;Så5ùd¤ åPÍàEúPž%âÀßÈð3m@&ÆG*ÛZ' ‘™èyyî´²(¹Çþ0@Ö€¨³#êïñÀ@ÝZL²;Ì•Ûiµ Çä—ÏY¯ð 9·hõ×nä¦êò~+‹<Ñř®ÈÇät=X»ù6ÔîfŸCþYy7ÊÂ<úf£8µ†¬=+¸ŠÂ^µéõœ(J{Wx1õŠ]ËùPw¥Åê.l>¢C‰ÖoÏÏ £”ˆGãø¤6JG\Þ­qFÞ.>‰x”|iq+& å‚÷$âPþ'C ûR6uð*~"‚ [ãšû:ªDäNq)­•íJ[]æî¡J/SÍž¡ ªØÞ¹Íö5tøO6¼„ ¶¬,IAŽZta{I©¨ÍüËsCÕuW^õ/-@]¯†Ø¨ˆzÅ6¬ $òh@}ÃQ½M ŠoÞù6”†g.ZwÝrBý7þ×ÐèÑWnQ½.4f¨Ž>( „Æá÷¥™ÉMÑx'èâ­¢649eCãyæÊ¾/üðì7š²‰ÿc`DSW/£üqhúÍ2¬¡µM¢W*IWÐLì@eæ± Ùù˜”s&Ê“}íþ"'‰ãIòy}øÃSÊjÝmCâ³Ïuký˜mÓž—õ| ó‹ã}϶`IkICÈ·–£™waÕaÍPK嬵×|šÜiƒ³Dy'•Ü`sg_­þ Ëe‡N#„¿¤‡êüx_3)„D>ÃÁU¡/¸¬S‰þ ’òª—4JW!Ù+'Yá¤P­:þZ© )·Yô’öëÔþ³ívÏ"íì7e£ãÈ@Ùð(šcß n Qþ`B&åŽH_Ú7È|4¨CU6YîmfÒ´:ã¢âg+îÈö“P—_Ý9æE.’ÊÝC.jq–û‰<…BœªO#Ÿ!Ù…èÀäçÒa»¼È¼Ï¯Öµó»ñ(е ¾±6ˆ‚ƒ[¾ógœQˆs‰zèr Gµv¥ŒÕ¢(ž¯È @qÖ|==ïë(ñoèlÓE”ª“Z¯Ü@™“áÑ­µP6kNè>+Ê-/ˆï}æÅÃ.gwY#¡"ýË >ZoTúÛw$£ê¡@ŠVDõËuE”d¨YLÇñºÉq*%¢hMµö¸*ú~É ÓA’^û~Ôù!j,~uýFuº2FPOsîŽx ê²'=ÌtD}Óop õ[÷Ä©Ÿ£ù—â‰F24(OoU}¢‚†’¿¿;J‚†cv£‹h8õþäë”<4Rã—tµ»ƒF¡—ÀOé 5迳’PA{º§r×kQßã•Íê˜ÊþpvóPìÔá|¨¯®>¤åÜC‰ÔæQP”›sÿ° ä^NL S°ƒlgµõ5¾hÈRæ±>å ]ò›?½… í_Ed×ø¶rf1âe$.^O€ÏãÃ~ð½vH>ŸÒçÿžÐæåwUˆ.@­U^y„¿ü\ïéõ¶/‚–Ή<©fh‹ o˜·„öª’Ò×^б½Rì˜)]šó#íîû¾p»ê´tUÜØÕ×ÚúŸ~þú{BÐÉvF Þ:¸éŸ{UC»ó§3†.yж+RVßí,ÌWv¡Ý|þ×5vhñˆ{é¡,tÐ9d¼¸¥´¥n×~Bgþù Ü”<ÐM#¤0K=5T]÷Õ{ ¯`UÏëè* P_âù\CïÞ$¥°iÀˆaáñÝãQ0ºv\žªTÆûøÆÁ±?i„JÌ\à!tÌv‡yú[–áÇ aqÕX¶å»5¬°1'¾á‚Õ +…cE°>•b5 $¿)hN&”Àö‘¹è7»·àO_½î!ocø›£Ífn1ˆ‹—îýk‹Do9¼ß&Æ ‰ë·?¯Òž!Ù‘K†»c!Hqö^ÜNìU¤J‘Y2䃴Â÷ò Jº¾µÎµÎÖÝÞ¼9mŠLãvþ¤MÈòÐÇ;Áñ°š°ÉÈþ‰ìðù\·i raéV·C.¿§LÑ—¼§™î1G/ò%^=;á òï=;>‚>Ü¥bCi(t !ðn² ¡½›¢½Ê’g9†P|÷£U= Ji¥ë¥|ëC™w.ך³PŽñÜË…Â(ïGv*;9S\„ªcP±.ý]‘c*Çd© þ=ª¾8˜2…êXZËÙ]ÚÎKH¥£–‹âûî3t¨T2HžÈ„:óÉ7¢‰PO›D€žõ«Š‚u-ÐðÆ;•CDhì˜#”¡õM­è>Õ‡zãmú‰Àö=¶ƒüœÐ7ëùû¿ÿKî±û/(´Kà9un_ï”jÛ´üï¾ós7ÿ/tݼÓ÷j?_$ÿ!™K€¢Õà„ƒ.–û>Qê"åá6(aÿÎ.ɸ¥$dàÑíeG—½ÖêéöóÌÁ1ºWP!œÌ°%Î •-’ÕJ¡êǹÔµE¨¹Ûþêœ>öš%jAC’VX9­4Q/*hø-ž»§Ñ ´&ÐÝ ›N‡ö£$æÉAUЩ°rÄB4ºUÎo·¤AïG¶Ù˜êø¥ÆxæûkÜ»WÇíñ†7“(éS?ÀèÓÝ¥—aÜÖAïÔnLj(蓆iUÒ{—aö ¥?Yv̧wvÖYÙÁ’Âó9ÍXÞHÖ±ŸÇVÏ_Ùb×儵Ϳ…‚~¡°·-K¢¿=(¤ò `Û“å—Ú#+øó¡YlÌè'ü=Àþ;dY NgòIp!aÏU~#Ö$v?-öú ’ò…ä%,’“G‡ÄzP!¥ôÈS—»µH˜ug>Ø é.[ÙE™þB†¼D»#O‘é„ñË /dÑŽäŸ[‹®È ÿ¡oBö•Nƒ™y㕽‘'/’¡.ù餟|v¶+:ÞÒy¡ MÃk[(ë­çcÞâG³Ç[)Qj_NCJÏ¡¬ÜœyOqÊn pÞº‰‡»‡6Ò•DP‰«Íò`Û*ªÜˆÙ˜*¤CµêÒ )N¨É!ýïR™ bÑ“pá¡UÔvçX»Á†º‡õ ¡^»×¡À§ÇÑ j°e0ߎ3]û2pMx…·ù^=@ÓöOÆ)]xäÆ;­­áx”šŠDˆ(- Òò{IÞ¢Åß/t1jh™BuW£ ­,jÉÿB«yùµÀ*hÐܱmˆ›_}wG›`OÚF'´Ù˜Åi5u´=uç}?Å+¼ÿ2|:6ìšß‰ë–°ËFîÁ•¼Ñ$0Qú˜¥, §ÔÆÒBÓŒÛ`!Ô|êº.•2~þ?N~…²1êA³ð‡ûç;ϾúCŒ0¦ÊÈä@î¸yæ]î–ÿ¸ÍèRTÕ¢3‚LFÒŠÅÛ³MD(Ê\¥Y1 nd$PCÏý‹[ ~v¿¬lÊÖ…¾V’óÐqÛX5æôšE®½Ò„«ûÇßûúÁ-£hE9 Œ}“S¯Í{ã¦(Ô` ¹ôfaª²ou;°¦~;½¹×eÓö2KK4~0Ý(7¨{«fb„ò0 fSûÌ5rr'3)Âüà¹1ÎXäâ\Ðö–"âóŠOÀrü¸,ÑKX‰™Ë/ÈãÕwÊýÖÆ°V«*÷íB¥ÙU‡ÌK„×>ðŸCV9ϪÞ6xàÑ!;vd'fû:Ža+n¿}³3?íÀ½Ê äúô±ÿµxòHèÎvRÄ"o¾sœÉ×sÈW„0—ÂŽüi1,U¯P2è2I#à¡à㟸´çQ˜õ{GÉT Š$u¹ÐýåA1ªÙ Ù qW³O6‹G‰ó†j•q(YDõŠâÜ”ºôbbñJÏŸ þ¡&²EnÖ‡÷ \a8³µ-Ê÷ÐöSgÜC*H¾ç¨xxøÐ›¯›¨¤w”jD©•Ÿtv+¸¬£*ƒ M€ý>ßé¯ñûU †'‰MgÄ{kñŒŸ) ¨ÅÛÿÞäºjóùù(Û¡Ž¶ÀÙ3CÔ½Ëc ^¸ˆz}²fJ1: •ìŒtMÐ|RÑ®›ZòsêeÆ@kyXzÎD´õ®þ¼½_¯jf<ýž›jURmе8{*ê„7üt½w4×ÿÿÿooeÙeoBáv³÷^ÑŠ”¨H*’Ý)ɨ„”Q${g¯RÙB²÷&{ü¼>ç|{ÿõûëvç<Ïó~s¿Ü¯×ëy<î·{Wþó¿©IÐÝH¬¬ÂR ¿”"4ÁŸJÝs\u0`Çþú„ñ$ OÜ˪= ÃþKœŸÆ­aTÀ§8*®ÆÆSÇ’•`¢°HŸXÂasÇœYß0Ë´™ê1Ôó'M»]2`!O#˜ Ù–D[‡>ÆÌÁr~#˜ï=σ“Ä Bê°:³2}jŒÖßžy²X—›WÍ#ÌïÁö©5Úµ«Ñ°{ý[Ò‘ô=_úk‘3ü> MKóx,[^—¤OB’O¤"™¢+HvhßPX€ ’s]ö\õDJw—·ÄãHmÔÚºÞ+‹´f:"¤Ë\¸ÿ¢‰Û¥Qu¤«œÈqÉ@úÆ(ZÕž=ž;m)ù®K iÈç×w:Ù ^x2H!«ŠW{›Ù<`c+M€q ªÖs$òÚþó%È=8N×b ‡<õ¦ïc¶Š7ú*Ýt„"òÝ{äP#ƒüO;;6ÄíðP»yRñ©°=gÄqÃ…x­ ¬¦ {ê0ñ' t|ð5%ÈGA2%_/îÆ² T™õ¬!J¯šwHÅN¢ì‰ðº¾…q”ÛµZ“ðÌAù‘gê Wñ(ÃÁ/õñ¨¤ÂÙn┆*÷ƾ¾q|µ§“.Y¿úPu±$:\ñ;ª3VÙäÊ)¢zçܘ Ôx6WsHP55Ny=¡CÍñ¹Š´ŠÔ ì°+&«CmÆ4QÔŽY÷Ú×Û‚:ÌÍIWŒQç?ÙÎ0ê,’Å+‡ŸD]ó]®.Ôý(ûÆïÌsÔÝÎÌáŠÐG=]ý³~=·ðö͟ߥ¯é¢ÆÂ±ÓK±©ÿˇÑu9l;+Pgóö¹eÉ3(/üE.3ÒÅ]‚ÌFª.—uQ7ÉÆr>VÝœÊÙÏ.Gi8)B–´ÿÝìè«ybMõdÊó=Ù£Œ÷|ª^ë'úx¯Ì|8ZOÒkX®øOîû¯¯ÍƒÅ8¨à/û‘Quô›ú.ÜðíE caš8ü|ý·Åü~ ´˜å •Ó߀Ö'V!­û¡­©Ú«ã Ç^,³?ªð:JžË_qø ïˆÛ}Lÿ;OøQ|É$´3¹Y/ ƒ6¡N‘±Âhu×0:ë. ­jt×@ë©¢³ö\ЪqØ—Ë%Zo ¦°Ã>híÕ}¶A¦mÞfdžœBû…¹K]®>Б6ïøa¹ º,Sfl¼‡n)Ö´·½ög^(I@_³¦GÈ"ô·jÐÉò«Â '…ÄõY6Ôí ‚ÑK»k:Þ{>³ÀU)'ü1Lé2§ž¢‚YVûœ_å0ŸpOšýM,:qÒëÀ²SlT잟_¹‹µrÄù°–;úNólJž²±]Éí¿œ œ±œ¯¿AÕÈ`Å-(:ß#äýr¸cRãÓ!süNCæKÈà—±~ÖþŸîÝþâûR®%tÄÌ3îñFöræ=¼íºî°2V‰Ox$¨!áíèqûù“D~g¨'>â¿}Šg²qç—ÏoîÔ‹B¶Wß{ÚÍ4ȧó-Zpy …!-A–S{ú+{ û hú+c0CÐûG™/ý„âT꾋ńÿj W.ÿ0×/(1ªµeóÚË‹é"ü—¡¤{ãóø…£Pê›÷äÉr ”Ykúk_†bÜš <ƒ¹óW/T¼Sˆ¹Áò *;å8éiä¡Z, óXÿ,Ô´)µg—B݃¿!ñ"ÐÐÏEVq˾]w>¸–Ñ ?ŽŒ,ºi–Bót¾ášÐR\ &›Ðý‹3!N:"Ü}ˆˆÒ kÚ…€A8z®ÔS%O‰B{èœéä#è÷©2{™23iƒy`¨˜W·ÆFŠÛ°Ðý!ŒåÜ?­MÞ7$¬`ú@Æþ«nÊ0û|ÓyHÛr_gîɶuXÜ÷éþÍÇɰô¢ÁÎr¶þÊ7ð…mÃ*™K;«8¬“6æ/,݆Må‹Ò.?®ÂöçömÎ^n$ðRÞ~„ ϩԮ"±8£Oµ€’´]½÷÷J$’}l·}Ã"€Uâç÷_^Ej¡#1r|HûÇŒš:Né.I‹?zÈžœ|C—ŽŒ¿å³3íG浸1½ãÑÈfGÞOà„4â®Çìé%eÍ7Ãpä‰ì˜úä†|7ºL5ˆÛðЋûMaÕ(°{#’7¹…?Êã“Z‹°Û4ÑDÉx´íË;”º$ÕÛ‰²n"œlNíx$­Mx€ÖòbBÏÄâ1¥ñWLT±¨ü€•˜ÝÏ QàrÎó]-TãŒ.¢([CõÚ™y>Ïã¨yÇxÀ_€µå5Ž´˜~@ÝjÁg¨W±ü÷ ·!x.uŸ+A£ƒ’Æ[ñdhüù¹Å«Š[hʾ9çóâš¶1Q,Ò\F³À$‡ÔUh.츠l3‰æ5–—Þœ`C‹ã Ò:\÷ÑâçÝ o;´´3K5ü¥–­»3L±ÐJµÒêñn/Þ£|ÏLþ! ½k>,"'õ=Ÿb A‚ýv,wǾB›¿C|–ƒ4ÕÏ*´åÀ^Ž£+u[º •á1wÆ¡ÜéŽ÷…;PÊ—è¦]E„A÷Õ&X÷8zÈûãvÂ?n³l}Úö¿ Yöá.^»y‡U‚…g ,V´„5j”T/U¾¯o¢ã~i›ïÍçz±Ûí{ã{[¾&€.²éŸœ>†ÐóK'h5Úú%<ºöçÁà›Š‘O.&0\Pã:Ê¡£4üÝ/.ÃhbxYvŒ&Œ®¾‘7ñ†±ë¿)X¯Â8ám×֖作(òp¹Îso^ïø Tîéÿ\è~£IZ˜ZwqºD 3ÂuvuS0ûÐ_~Ô`žYâv™„ÌïüX© €EñâêX⬼îoC˧È>Ê]%ƒ¿å©,Á*«¢7޽†µ—Ñ×Ýêƒ cª °%åfóÄÀv Wuâ;?!®F{´ zŸ ÝgX„Dõt‚v¤ÉH"Ó»Ïõáƒ==XúËôɵÌvG(ÌbløÔù“ÓH•šÈËÿi%Y|UòÂ}!”ó#§¦‘Îtmj‡ì<Ò;rgª‹Fù¿¡ÄȘúûNúÍ`d5N{šJŽ,Ç\†|´Ü5ÕÞu­qˆ;fãì.$:½axPÝM3Z„9«ù9Û˜ÄèY³ òj¬ó~ÐD^ÏUî¦7ÈWuêcïw<$-o“Øàˆ‡‹î|#£EAM¡IÃ:JàŒ±‰Dá#Óö´²(²5xp®™ÅL“/ˆŽÚ¢xiNåæI”(YµÓ-F©Ø£V­ä(îϲ)²e†©ÒnX£œô—6ûLrlœr«‰Û´ô9ÐÿøË¢Õ $ÖiÊvª¡§~-½ §aäpÙA;>íÝþqÉÆd’ÛgÄ`òÃØ%"âd˜ž `©6ú s×tRrUÂcDeË×<îG©a™êiµü}Ùl3ÕV«Êì~¼Áw`mêöÙêœÇ°‘átÒY ¶"ÊW”lõa'é®9ƒ¿.¸±ÅœÚã9±†–Þà5Ó 2)# |!,ˆ¤é•/V¤œ‘\ù–—y+RLÕ0ªœIBªrµ¶×¯ ‘&`”ãM?î+¥lø«tgû‡ÏìÓßË)¶!©ÈpF¨s? cO¦„12-|y±á¶„,Ëï;!ÛjÝSs¾È¾_Éß9UHOW±!·B•©otòxœ9RC}yµ_§» 7#Ÿ¼`Òþ÷‘_3_Ž|Õ=ÉS‰2ðCÞŪ‡ä(D¬•¾6ØŽ"ܱ•ζ(f&訴‚íŽ<(Ùkëv­¥~ñ&†×ù¡ôvÿù°<”µ,ŽH±Ô@¹9jå?«(ÿÍTkê×Tì.ö,'D%j÷Ù;µ&¨bZᓸՅøTôGJù=T?uE+ÒÕ™/|J÷Õ‡ó¿QF£FÚ¶ŽÛË`Ô¼àMùRصXj»~J¡Vñî¼BÌMÔ¶š+S§ú‚Ú#zò Ó¯QçÒÂÍ·T³¨3ñæ†mÙ ‘©$`!á:fQˆ„io­_T"±›qyL’r¾nÎVyŽäœy•' ‘Ò ‰Áì`RgÓY^2Â}†¡ÏîG{!]ØE¡›2ìŸS¥qÞÓϼ¯TdÈ|Zòäß+È*c¼iÁü ˆV¹­||†ã*üçc‘ó:Ñœ_òÐ'ÕÜE^W¯•ôv!äê5 zøyº¬vÏDÁo^/dìöøz°¿‡Eö•ú6®×  °~%½Ï ®|e»†R|'¯ *¢ôlËû™y” _7±ºˆGÙÉ£0ˆx×%N»£âa<§E Û $äòC• óÜŽ5#Ž&‹};sÕ$Ó;òØPýòÝAF2Ô(u!x—üµàÌHr¯#jÏÔµÈAݦ¼7ñ/ ~Y¥ÞmQM4L ªzwË;¿êyަT”§ú+cÑô×\=Að š%Ü—~›„æv]!ÝüÑ‚h·*©-ŠR¨.’Z¡¥Íbéþ/Õh9]sK]­<Ãâž÷ú£ÕÊC×xgo´våâÌ{®†ÖÝáæýïÕð8úd1Š¡{ë’ï})<Ö ÷™/˜iÈYR ¯Á° #í)cPeó¾ür¾=“ }üemÍÄ!Èäö)"xéƒÑÇoèìéžb[ÄÓYH~5Æw’Þùöû©…·$¥-Îu+в {ú…'ÄåÏìû oϵ*nǬ@Ê›´ßx+áÓ>e›…È–Óc¡´I¼¬³´H…¼³Ô?¸ HêÑÇ9²~( U›ªJ€¢vuš‰»? X².pÿ£ÿÕòˆšçÙ x¨?_ÔÕôÿö-*Žïåźõºèx(ýB¶Ôé ¥o÷ìãyø"hÞ¿©àå ŸñT’ *ø«y‰) R“5LöwT…Æí4ßÛˇüéûÚm/Aíà2}ãK¦ÿ¾3 ¬º_c¢}äŸuÀwxMpm ~>1ø‚‰3Ð"ôXjŒGZ·Êg–“öÖYksè:RÎü$ôt¿½öbZ°~+D²9Ý€~’As:qˆüsPt©†,Ów$È`DÉ㘋ÿŒIfð|‡ ͧÞCïGaÊÿžK8Ìüµ›|P+óF|i„É`!w†¥£C–äŸÞ;ÏË?cØïÏ}ƒ•×GTÉ<aí•_Dê‰xØè T\±¦‡m£¢‚zVØÝ”Myµc‡„—éȾ¹²!ÑLÅ%»ÏœÁ‡¢ÃïpÔ½A‰åäö5â”æ·ÉÔQ–GY*Mߘ€=N;ˆRÄ™Páv—Š,‰Í¢ têB¥•]²Uc„p­?Ä\Q¨úƒÕ/e¾{Ow wwßñ¡æañ¸À‚ÔêIŽ­§@·Œ ït¯ žCåBqÒ~4`W#›ÝBÃò1vÁ+hlÃõ–ÉÏMúÂô5MEÐ4õnP{Ð+4;¡¡èØ]ˆæ„ª‰:NkhwãO¹òZH¿ìÏj1G‹bÎ7WtMÑRIÎ8dD -³˜cU7ÑŠ‡V¹0èv…ڭ廊]1Gì©TäPèÝTj†©fhM;­BßRÜÃgÍN@÷ìØÝÅ"¨X(¸KL ¾tvKíš÷C‰HÐÌÁɦ=nVÏX&Þ‚<åªpöv÷Ü~>]-P´Ÿ39Œí³ ×ì6¥h'”úV¦×*3@µÓ¹ÕX Ñ)êFké,üÔj¤¾?§ ­7<Ï}ç•t:ö' Ûž<Ý(ÿô½Élï.ù æO¢xHa(MÒ$,aø—È3s ñ×~äš#kÉñ#õa4<„…—ï<Œ)ßvÜÓ#Ò®YiE˜ 2â¯Hæ‚I‚ùнùÍä6²åÓ ×HCäa&ølvß_n˜cX¸·tÏæ/æˆÐµÂÂYþWiªÏaÑà ×Z'–’Ï>z®ÞËS ®œ?+.ŸûŸ‰ÂjÝ¿¨°~]÷gåïHØ4+ 4m®†m—(_¾“?a÷ÓúnÙ† RŸSa:„D¶Ö ‘¸&ñ íl’êJ ¼à]G²v¸ñl&R¸3^;ê{©$š"\_í­›´\žá>Ú]·ƒQ¤c<¸äˆ6H/Pw@^ÑTïÉ31AF‡Â ‚¦wÈôRüÛÄŸ"dϽVÉÙˆ¬f±ÕútTÈöÓoÖGAÙ¹Ú)þ:#G¤G ŸBÎS¦'¤%¶k0m§{í(òèëfOuC^[‘Ž$äó9&Å7}ù訅MûððûÙ_‡î£À¯“—n(¤`áLø…=2˜šjPäÙ݃dl(:¢)crÁÅ­¶§ T2Q"Øupá¸.JÝ¢´qsW@é7o‡êPfžâ¹ž¢ÊÙ^¿'ƒ7ñHjgÓé7Î(_ϼÿ›z*Rö|Ç£w7¢¼<£P‰>ÐÊA5•ë•JØ÷!¼{“F»öUÕi ï;‹jJÆOnÉ^Cõ“ýú«Žö{¾³æÝœ1j1Ž—HZ0£öÊ)Õ¨ ¨£FvÜþ+ê¬å „ð?CÝ·ÉÖ‹£žz¤–öê[Ôk¯¤èqæAýÓï‚´‰Q¿»ß¤•k Œ6T;)vÑ ÿ-ŸT%²y)þŠ×BÃk Ë;hX‘äá:È„FÔŠ?#ÑÈXUÔ^ý´ä\ÙnP æ}Î,ÃSCÈ?~ÑraÖ7‡>]„ŸÅþ9üg v9•µZ¥Ê›'Ú¾P@Á×ûl„îéÖ%G;Ȧ'qšå¨‡Lщq–¸°½ü—0”W®©…뙳վò@Є¹ºé_ý 7,y„‰2)â¶íÖ¾üã5ŸîI’n»:5Äží(·²vÖ¯é  ‚AJ3Õ‚*_„Ì[ò;@5¨ÒôJ¹@-O‡3·D9Ôõóg>h†7·›†˜á«Rà3/y øÖù¼ø•l|Ÿp JqÑ‚ŸnC—?3~‚fÁó®™ Åoùdû¾=ê(¢ÐmB3‹j·¡#´`?Ù§(è²µ¬u×=ݿ͗½;½?R HágcØ­üHà}Ï”C) ƒ ûPn8Àð¡ÝØ|΋0òÍqõÅ3{ð4åÃ2L<ºÄÉ\t¦J™‚óaöh×§¿aàïŽ\ãÂÅFO kõ<;ò,Óû[#Õý¯œz‘oõÅSæŠsxHÿ@Çõc„xx2ðèJt; ŽÈ½}¥Lˆ"‡Bøý/ Øí¯£ ‚þ© 2±ñ(EkÍq¥¦”˜ãsP†„KìE@ÊžqË‹C¹¾ï\zÇP>{Ÿ^¼Ô*æ©Ï?‡Ç~Õ“§ZBö—’ïŸ3#:NÎÇ%K êï-E*Tge"p™žGõ…»án·åQ£ŠªáËþÔ|,ôµ4_/÷Å¢Ö²°Ž˜ jGMPœ¾l:RÑ‹ö[Q§Œ^¿³®uÕ òËö n‘‘*ù”ê ó>2zá6Án¹¨7«ôÆÉðêkZŒô)màB+MiG]T]óºú—2ä_>œè£½yjWúúŒ&|ŠŠ"ò‚¢R©{tæéÏø4È6hó™r‡,=2nI¦~Èç ó!† ~é3£1ÖòýËG¹gIòád”Óð¶ëFÆöX|ø4ÊøAe/gª ;² å…mµ»äÖPîwJ„(¾vk?Ð5¡…Ÿ,d/g›¡¹ƒrØ)è-´ÎY=¤nm7.ýý®C í…gO“|/ŽƒN•—Êþ;?féï ´÷hU‰\‡ö}¡g"áMõ‚Y{Š´¾ôጤ‚VÑCö¶d{ÿ—M¡*øþ¿ëêëÌ2ÿåÄ’«o õ–±ý ´.uIwC[Å®«‰¼2´ßK/¢„NWÛ«;·4áÇíEn½nè¡söPõƒßrO…ˆoœ‚?mI3jg¼rnüÖ„!xôM¨ÅF(ï6Á˜-uæEío0Q¨m³ëÏ Ó§³Ë}¤b`Î¥#\Tÿ4,xŒÍÈUÃ’Ó ^æs‡à¯óû–i°<Û™ÝzÖ»ªw–{GNQìZÄ·Ùmu"¡Ô¸îK7$j4Z³c@’à|–0çWHvcB㽊.R¼d7ŽUrGªµ e^x€´ç…Žn pöÓ_¹±Hÿ®¸*p2*v¬*÷"Óà•¡ÌHUdùÈW¾Ñˆl/Ù¿X«ä!{CËå?Ÿ“üdé‰ãsÈýÄk2Œay»$Fzé!_lqìç¾[xHäîÑÓàá¯vÒ\ €‚Ñåb#J¯Q8ª1á'Š–ÙûÛ~n@ êò©úƔҤyÖö[ ¥Sx=CQViVk˜Ó åæ¾+e‰¡|A×/ÙǨ5è×NcÇ>6¬$ð>AŠÍ©½Ž·kxOб¢êâÈ2x‰ê¢zíÉQÃûÍÐÚõvÔÜ”ŽêæFí´îœü©PÔ½/ßUÿõodÿ¹#ކ§P¦ ñ#³diúg¡I¥Þ¿[dhúIð5šy<Ôí2nCsÉî…çdÎh>Àd9ñ -î·«»§­¡å¡Óªw˜ÇÑ2/`ÚN9­ ¨ty ­J•£ŸºþFk·7U´Ž‹k;võÅí{¹·žÿ¤Šó:RÏÜ yU¡C-)~/õS ò«óÖ÷µ×P0Æåjé2YO^_”Ž…ŒÌÓN– §áÃì³îƒwötoMÒ)‰Þ7Hk¥ÕòCRÕ‡q‡¨yH”¦À¾>H`©<÷™]âE”\ø‰eÂaÜ6´:ììþšÙä5“}$÷=äYªŸyç Tþ•%¦PX40ùÆb`o ¼fîQE7yeK#¾CÑtCæk¦µøœ»WR²5ßÞd† ÿûU'õÍ^nŒT{ÿ•쿜x÷x‘:”,FH;AÙeêþ`*qøòj—úÄÝ=½O4M»2ç™»ÖŸCål±ÀE±ÙÿûÎvã3Ô ¼’ ŒÜz“_•Æ ñKíoÏVhÒ±<ñZ ~’LžŒõ£‚æHS>’yghµO)Ó|Èíz÷RסӗI”fÁ~uà—rЇ^ËW*ùÃöð‡ug†§ÓXièŸ „Á6›ë70\Ìt„û©Œ¦µ³1úmÃx®ÕsnO˜œ° (ýD3V•ÎÚt0·wQjù,œy/Ï]‹vIù—aÙùƒßH$¬Èn÷,¶mìéªüUûŠzظ«êÛ›[¦9ûa÷û©rÂ==•³»]ŽDW-וÖDYÿÉ3Hºz’<óâ$Ÿª!r=fTâ UUHSaìA¤^Žû#3ÖX[B‘þ¸:ujº 22Ý£úö™Éï,Œ‰ «|½%ö0ˆ¡¬ŠŃ*»&–êÈ­ïÔM^Û‚¼WˆuϾBþ;3­7ê¥ñpóù[¶Á(ä(]{ðµ'Š™e7(F :w]N”ªz°zúe$ʼ-»<úU åj;{®œ8Œòëý])Ÿñ¨Ð±°ÅVTr£ºE *<”n*¨ê÷BG¯Õy?îïmEï4'ïMXÛà-—Ü®¦CûÝ S7‹Ð5Z×þ¶}zo±‘Ú.Cÿ½¡£w[‹`ðl¡ô5{z/­o3¯ Ãg¶¬¿‚áu£ÄHµß0’®˜zLÜF)/ß1‚1¸oîÚ±ãR/ëO=Þrî‹1e0)yóäøŒLiñÈjŒùÁ´¿ºTgø.ÌlÞcwΤ†¹wRÕ7kŸÃ|å«CXhl[•_†ÅÅ’‹Ë,ù°,J#˜ì¥}ãÇý• aeéVèbn¬E;÷].? î;E 6ò°õX!ïíŸyØùá2¤[5‹/ÏͰ{#ádWðHée$¶¹(¢mQŒ$= ¶ ’LHæ(2…¬û‘|K=hÿ[b¤Ì^'Wc Gêp¡àSÕíH¾¼Xw¬÷§ùÿž‰tãî=ÇS¥~·=KV:¹#Wó× “)Íœ:‹+2¿ØÞ![áC–õvö—VœÈv½>³œ… 4_üÓ¸O 9¼hXž ½ÆƒËOÿ¨K ×M³L~n~ä¡7r MŽC^ââ¦#R‘Oõ©ôæ4ò‡O”~ê²ÁÃ$G7G5QÀ«‰úê$; 6Ýÿ茗Qh“øjCØqbÐUbBÑÛý;î£×PlV‡%ÁÑ%Ìþ~rì@)«_'ÙèP:ÜOª±Óe¶HçÇ6VPÎ_P‘øûnM(IBÊ5µúATÔ`ϨP£y:½ë÷N¢’†Š6éÛ·¨ü7]RÒ^á{÷áŒ8/Tµ“ê^D5]˜ué;¨~ Un,Ì FÙËuŠÔ’ª^êr=‹Ú‰÷jÕ%¼PÇår¸üÔñbu'MBݲÐþ/¨ç«¸m&«‚ú¬—ƒkÅêQ?‰då8þ =÷| ^õšOþBC*£íáÇ©hx%}œÏ› U¤Âҽш‹ÙÕxì9ŸÝêþ†F™$,ÎÙáÝ®,wÔàҨѽ/‰l"®,ô„`=–:¥|ŸúÞú¾¬tãø5¨Y}jçwdÊ5%x…&§ ØûÃ4äö½É °,ûÐs ;ø´ïºSÛ1ø@lÞ÷„–iÏ—¾Ûôƒäœ+Þ­§fþÕ4Oáâй=]ýãðëQÕ?^sW¤ò‚z¾Aᩎ¡2ö+PÊû)kŠƒÊyf¥L½BEŒßÓ¤tP%œ´õþT/§.;sm@mЍ…Yå8ÔŸÌ¥5•ðƒ†ÕB¢ŒF&øQ}êÜ 4½{þ1¶~<çyzÊjï¾r£8Ê´-¡9ëÇÁ5h©{õv¶¤Zwëåøe¶ ýVÃËŬpèäÙy< ¿¨˜æ-dã¡Ç µûÜêøÝƬx¢ÚúUINæ¤EÃ@æ+b:¤úH»7†Ë¬E¸«R`ôÿyoç¿8Ù,ìf “OyÎóÖà ñ¥¼¼™Ç0÷êâÀ vX݌پ© ‹™Âõ ©ƒYí´WöÂ]/%"-ÜO[/¾%Eº€¾{ÑÚÇ‘þÊprß±÷È௟n⫊ŒýÑÄÒ^Èüt,t&â3²>ÈI¿‰Jæ²'ü‘cjä¨4!çÀíݲ³È½–"ovy7F½Ž!oä9Ná£RÈ—vc–?Ä‘ø\5 ÄÃÙŒ™"A(Xǃ‘â("ì]¥zNů:<_ D‰/>Er£”Øs!á|#”ڶᎽ=Ž2T žÆ(«G/ªÄ¶ŽrßIÛfù |Ì7#ˇPñ©ÍÚpF%Káäé ŒBåa4Q3ÄciôÏC;PµõÕLãTg+±Ôm0Dõ­u–ñVÔèa3¤ B͂ق3¨å~ƒ–|†µÅ,¥Ê:Qû× ù­¯c¨ã=žñˆFu¤¹h‚ö¸}ý™Ž3qõ¸®=™‰N@½È´)çMPŸÈy­Í«õ/ø3S§Š ~¥M Íô1¼ÞÍê+}¦aMnaÊýùÐï3u†ý¨Éì' Žƒ/NZXÏBáÑ[eÇÃÂöô,SþBud½˜Ÿ“,ΚÊL W–KÔ†ðñWó»*.ìÚ§Zñ$Û³¹å/‘@âûwK™yCVRÂÛÒ¹Õ/ôvByP¿Ì‹ìwPkR9BÑí_ï0W>I¸ ?¦ Õü5¿Bs´œºƒ–ùlùÛÓЦ§œ­ÏO í¡2=αbÐaþÐÏÅE: ßiß“±ö¯4ü–g ú¥5C)´Š¾É‡ÖLvêOï½ UEí[äƒêÿrá'ÉçÿÞ#:q^º­,Ú±ÖÉЪÍ&Æd2¶÷{Žt3 ‚¶«÷¯Rž‚öûfÍL½Ð±qý­€#t…oùÛ¾‚î‹N!áõÐ{A!¹q’ú¦r,2fß@ÿöÒ£˜§ó0˜kSÁöí ¿ûé©Î £©'&+yg`|¡/³Y±ï¿~ˆógO‘¬q²ÀÞ¸ó:&†Å±sŽ-_Àrÿ¯Gž$X™™›-æ`‡uqæ¨ÀeØüœ ŸòÂŽ÷Q=s%$È?{f‰ÔFž"ñÌÏ“—¢’ôgÎÏGÜO|ÒéÑЉƒHN³ŒÕHSg^®Ýã…ûôYZfBz¥Z=ŒdÈP·Ê5k™œ^•‘a@–Ã+O¾ÿ©G6*wšÜïíÈέâ”s²^½q³FR¹6?ês·º!Ïo;cîKÈg;ôGÖù;ö»IËlãá+scÉ™œ((>ÓYôJ…y:«BDöü¦Jìä¸Ý–¿X{g%we–#ÙŸ £‡·h]}Åa\àÀòh2˜4“/qhš…é¼Ã­5³0§'xÇ.ÆH.‰¼d ‹žÂ:{>™ä(Õíž3ð·±öæÙ~Xm˜6`šƒ JY]ÊóÖ°LK;O»Ò„tC$ÁH°ýqÚ´ú"]åT¨ýÄóGzw«‘ôÕë¾Bú;HîG×r7)?ü¢&íA#!ŸHÜ·j¥û-àÒí¾<Ÿ±éˆ ™ü’BcÈ6µ!'±Œ,o…skB‘mäIŒ r8{1ît±#—mù÷F#òL²UI"ßÖ%¬È¼rÈEîIÈùѹÇëí}ðØ¥ÎÂõe”Q¡ˆ0Úúr–¿räîË£üŽ »FʨwþÁ9>+<ÖöcŸ@¹6ªœßýsZ Ue˜i',_£Újî"¢BÔðœb{ЧƒZ×ÒY*P{’é{|¶5ê¦ÙMZ§G¢¾}ï>{ò;{>Ó:Ó‡Ú ÞɪïТ‰d1OG;šZZÜy屃fä~ý:·’Ð,s…äô=F47-øFDT€æ£Žøèè(Z¸_P<‚Ëë1”Û¤hyEu9¦ä+Zöݽ–8’ˆþ¶«±'>¢ö¹¶§â_‡‘ý©æ7-ê°Sšò»tIš:sf…¡ÑÉ[ƒ7ä9T[}PŠ^0†òì¿cUtP:ñêcx )›xÈLBþ¹ßû½˜• 'RÑås(Ã?n?5¾ºœû2M¤º:Ôô!;už^bþ.ëÏ?åwçÊ/) ÝLSP·É_Ÿ~š”^Nûä¨Bóiï£y§ í؇ð—žÐéz{}ù;ôÆ’«ŠÐ·žfJìƒõçZέ}ɰù¡sö°=lÏþ©m }Œ¢ç_ó UA .Ÿ"V”’M=]–…Òþ»¥YÇP–‚Øè^iÊ¥tO´ÛGyîžœ}¨`ÐÔo<‚ŠwÅgäSñèL$ß­<¨ä±*ŸL*|á|‰ï½÷|S‚‰& ª:àŽOr ªYÅ&yI>DuÏ8Éå»·Q£Ù•$‚/µŒDýï§£v½~‚FMêDL6Ј£î) °ËC=›¼gÎæ¨WÜ›{òJêÛ2eι¡þ¤ÃÅ‹_Ñàš'ÐP“¡ÁTÙùÚoÙhxj­srü VD]¶*C#þL®ðëhäuï‚äô5D–˜¡1ý—uo¼3Óö(ûêå^êÒü|Y³ª~ÇÛúb_Ó+׃ðýUSpX\ToÜ~§,Á_\§37Ä_AÞÚæÑ·’¦Kõ{ßá7ð9“U9ïѤ×0ìz‹+Cêñ/'ÛQHºý’^ÁûþÜ”ð¿ûþÕÔÄ«:£_E ==¤Àš‰â¯9=Íõåƒ6?ÏèøÎã_щrq²36 szôPþêßÙѨd¦œvÛÇUõ>ws÷ËCÍ5™ö‚¨c7,ÕÉúìY÷¡ÓÐh²´-îßΖ´Õ¹ÙÃwM/ë@ø±C›ÂY!?‡?œnö¿ÍÃmÍÕÊ¡•ÔÉËò/´ïêË}íãïM“o/Bç‡Sscãð«ö5íäè•>3¨Ül }%5ZÎÑ$Ð_.¼è@sùýow¼\¡Ìcª:ó_`äøË cá Óq Vƒ Gq»¬Z˜ú¶Qš·³Ç)ÖHê`þ1EGËYX$·æÜu€¥P¹·ºEá¯@°ÒUû*X©ªí<æ-k—ߟòÿ,ŠŸHXNÁ–|ÄÉ”Í`Øqe;p…U ^ö‘x$!aâÃT:1ÍŠt+ó” ‰þïÏâÇ. iù½÷#ëkH~üûÙ,÷n¤d éxðÇ©¦²ØO½@šbÅ¿—ç÷8î«}”IÚˆª§•^y¡ÛƯ"éú¨´{ƒxz¾÷ùßR¨_tjøã¾Gp®B™…wbð9(°0$M‰ì ÷î­ñ!ø<98j¹ŸÌ^ù>©ÜTŽ?/ÂÝ‘y ÞÓlµ‡J@‚G‡âù{ÿÕÔÂ¥Óï\\!—ÏžL¸)ÊM¦â$Î,Cí~R×±,%øÊ9¼Õ±4 ?î¸Ühü° Í'ìâ÷­6CKñçè’Ï!ÿõ›¹V³³ ígIl:¾‚I©–+™ÿj{¡=ÁHÔ´S¤Ú×\§ƒ¶ãoUEq@kuÞ߆ü©ÿúÑä6B«´—5לÄ}j(ëJk¡e%òg eð¿ëV—3B_¥Nüo<ƒë¹Ã¡ã~h‡ÑßAè"󥋘÷„_åÎ2éf„ÐÕ'øÌ®úˆªŠzïB¿_Œ}©öKÈ£¨u‚aÆ*© 0²žv˜c›ÆEnR$Î'Ãä;S­ûî0ó"¾Œ×Ê æ} ¾,«Ã¢“IZø/X>N{á†ïkX±Z ­¾xŒR½p@ð*̹lìOHu«æªÞšs 5I@µ´FIEjT_#O‹£±@M¶-HÿŒÚ|ÇÒr êl(Xßãÿ‰z?JýÞî¢AUÝŸ«¼htk{MhM¸òÒvN@SÛÓ3ž‚hÆžêwøš5ÿ)M•RAs_ùéc×5Ñ‚O2ëþŠ2ZËèXVJ ¥ÞvÛ é´üÁ°•3•Vú‰³×=Ъ”mf¤K­Š;2ÝÑúÍM¦§Uè1‹r5Ÿ®ËAê"cS•s`ˆ±Õ‰m5*2—nDÞšƒü+/wýìÔþñöÿ8û<%yŽì¯×†w¼å9ůdá­¾º¸ºÄ?|Í^{ÑÞ(7Hdj Áë ¶§$‹´ÿñø~íh8ÓæH¿°ôÐÉuÏMÚ?pܼš¹Ñ¤$‚?ÏBþ°º¯Ô‹(øî¸;3¼—š¸;Ö °è¿cé7(’ºÐB+ø¿ú^²ÙQ²õÿ·st¹rû6@ñSkB×kP²èó'úS(”þnqjüR_ Õ³9ü”?ªxeÎôïþ«Š´N\êôé ÷CZ öKÄFÕ¡ h á=MðÕ¦Ò:÷o!4õ›Å{„ÂÏ+]$ªá]Ð"tËñ Zw_=¬3Óû¯•5y!B—š³RÁogèK{ÌÕ§¿å®ûò¶XBÿgûû8 ­χÆ~ÁP‘¥Hp&Œ$ðÕùœ†±¿}$é¢`¢pß õtL³<¼t[ëÌÆG‡?†[0ÿá]õrV,2/M±—“ÀR%‘{³ü=û„ñë³:XµOøÎë‰ßLžM¬ÃO#ãÃ]Øù:VCFµ›gæé‘Hò«˜„Y2'ñ•‹H #©ZnEQ4’3JøJ‘‹”2GgÍßo#õ[ƒÞµ1Üg³ã/#ˆtÉËÅ9z‰È`P/³,‰L,' }»Ÿ*7µÙÝ E¬Le¤ÙFµ;Î’QÒA¨ÁÝÐâF„š_Ûâ3¸P;4Ò‘ëb#ê‰&ÉØ >‰°ïãˆ4ø´Ï;ù× ”>õ.™@ã.³ÇÜEOÑT¸tÍwFMû¸subÑìÁÏÜ‹‘¼h.Ü­Î7æå ³g>£…1#‘è'A´hî=Šy´4´N¨å¼Š–¥¥‡,{‡Ñ÷ùáÐèK;¨y"Ñ7ãÿ£ë½£¹þÿÿ2²÷ÊÈ!{ÛÍÞ{BR(‘H$)-í¥(*›lQöˆM’•‘-3›¯Þç÷~½>çüÎû¯ûyžÇãù<çù|Þ¯ëõr¿ßî÷»—î+ùþ€+á,lë¯TYÛ„AëÊÇùŸ‹á‹ÞNŒªH6TUÞ<ž0÷Ê™™¯Ø½œø§ÿWWÍÇî4Åüã—ÿÕí{+yæ_/'ÿñËœ.±'RI™ÿè§bª>îÎîúwã×ßä^sA‹¥GdQiË?>Õ¹d dÁø¾{’Æ8<Þ͇½äì1Æ0H‘¾ôÇ#~jÀÇF<0Äå1CDCëÑ Kr0üëQ³·¶&Œ´=Kg\‚ÑÌö_¿.ï(eú‡¶³0t¨ó»á?náÙì®A{Ëã˜`2ˆH´®®|7Gfj&:3& ð›Œ"ä°Ç%˜»0v±“¹æožÌ-õ„…”!}ÉÏaqœÜ¶òA,J._h†?Cï×J®…ÃjþŽ'Ç÷ÁzC `âø®ÑX¥:Ù‹ÁN@Ý Sój$¤Ž9}^ ÷„,éqsÍ"1¡~i¡ ’\#êŠüŠ{éâÒ>M²#Y©ùì8I8R]ÒÈðÄ4¦§…‹d®¶#Óâ=¶û‰ÈbdX«> ƒ¬Éë¶Gã>…»ŸLpÙ³Çïó"§à0¡ªç#äŠ{Í뽂ÜQµ¨?‹<÷Ã.Ïç"oj›µ%òÍÅò¡€é4 ÁåeÌö ÉŠ<ƒBksÒªBš("ij(K{ X›&Ä%P¢è[ö§E¿"Q|ÿ÷“~ùDxpŠ$!5|¥èƒ\›×¢´ße£‹(ËD*ö åZî¾Lœ©E‡£»l¨UÊ*Þ1JEÚK¶é,¨"0ä±’0‚ª‰ÙÂ^wQÝÜ…y:‘לó÷D.jœ?Ì\ïÏ„šnåï7>E­«[|~»<§=RžwPíêž 9&¿†z³4—tÞ ~õÃhŸýhp—p ¿Ô¡¡aö‰§‹£h¸vúxª½(½ü2‡¥ªzwBÑø“¦¼È:4Q‰;ï£Ð$ómö¼Nšr0ëåI]DÓþ¤÷~§Ð´kÅ…ná4š‰|¹ü£FÍü‚O¾=ñƒ¹o¹-S  †É½)uȪà¿ «ß²þÀWÉœÇ2w_BÕÖÑ©U!òiuª¹ANä)²Ì?.6ûGoéÁ9ÏøM!¹)¯‹þñ»ÿæÓÿ¶ÿ½þßûÿûþÿê÷¿zþ¯ï_ê Uqõ¿ª 8Ó«zi*c,Évv| ZIåùð{¨é»#ê:µ>üìóÑÿúÕÿ—s—:H,²) éþ [ér:´°\¯{cŸøÏó¦’´üËžrkG–9s%C t™¼û}ÉBâÿçc½kWÚOõtÃÀë³æ|¢Ëð“¯Òë„£ ½ÙoçÃÕ#ÈÄÍ3n¿g•-ˆ`\Y¹‘p¹&ŸÞ®.´°ƒú¿k Òð»N\˜þäÌÛföÒ-ƒ…¾Ô,«ë °äŸÜ,»#(S¨¿í€†vÁå½›høëƒdáâE4:Õ1ò¾”~Ý º3†ÆöT6n'ѸÔMü¤É'ô=U0Õ5CŽ*)I.#BÎH7¨Ï#% ãiOóäê[ñ—?©CÉíÀ¤eãÏ£èäuv‡\ jÁ{RÙíu…·¾ê6d…ªÍ&¤ïúžbÃ(Á:¤vhÌÅt¥BÂ3Eª ˆ—ÐÕ©$‚xÝúÁ›lâØSàuÓòZH×øit¡œÜбµà0Ô”KÅZGA}Y!cŸ_4S=ä½¼­Œ²iwª¡Íõ´åáhOUµ˜®†ŽÞ«{™SL 3þ[/S/t¦¦ú‰fC§^dEî×NèH=“ûëD tжO„Cû“èMáÚCÿÎ*:©¼ü÷\Äǃ3»ü(œÃAqÚÝaãô¢‡öúW’&ï #BÈñ¶t"tF)h¶?ò€®©ÛÉÕ«Ð}´Ó%ãèôPï}]T—?Úš]¸*סŸGýöÇa€Á‹¹3²†[0DóÞyv†»ç.»£%Ç>S]ƒ±Á÷ÔË'ßä™ÂUbœ…ÑæÕlm3˜Sî.úø†¨ûZb#`q–T«TÒ–‡•'Žüþ «$kÂú°~Ò“ICõ1lÑ{SŸ q¾`T¼Ð†„Ç{b¿—ìG"‚BÕáB$î(PQtoDÒÁ»·4œ¿âÅC/¤|æ륦0Òhé3®L÷!Ýš0Ǭnd ˜JWjBÆèt2±Ȭ#‘Ÿ¤¬ÌoÈT…vs§ûÇþÓÅÈá~$´‹Ã¹f¾°ÝÍúˆ<¶:ø y׆>Ÿ3+2C~­«wÇS¼Qàës[QI^ .–Þ‰­B犕ÊPô´|HÞíO“úböb%y^äúV Ôâ~†ú'*(¾¢ÃÑ®Žrr¬Ǿ üÊ·ÉÓ÷Qñ«ÒÆòÝ‹¨Üò雘 ª1Nû«ÌŸF8IS}oÖ5޾'ž/¿…šÏ#Þ?0ðB­©("VuÔ9#×fÓ…zÜi’³íQe\ná•ö¦ñkÕ¸ ñÇ|Žþ{Bh ¸hgÝæ¼hdv …ór¨bZ ¥<ÍgCË1¯OðìZÅŸýøðl-ZÛ„æð¡ ¡Â•)¦:´Iʸ'«¦ƒ¶™_5®AÛ‘mÚ-hBùŽ3?í¸ùÇ¢}‚EóLû ôÎk•ß縎²±bìz|H1Ø0¥¼ø †ÒÏžê=< úT¶ïÃÈþîç¤ Ïü÷œB šÇ¯ uêóˆŸÆ&¤0ôü’ßIðUEº"Þ=ŽÙp…7¹Î–E§MeÓøn^e&í€Îˆ%Óvt—¤ùG‰/xãGäg }Þ:5„Zrö-NÅ^Ý ùºiü®fðáÁ¸„lÔ( ‘O!šœƒÂ…Þþéy(òŒ8ä•ðŠúÅù;îCÑÄïå,ƒøhÝ¡æ1–ö¿ÛN«^èá“]†Ä‡4i(þ¢ó覔|¼Vc§FeÂFa´©PºåûäpT¼6‰>7P•K'1êBuh÷7s_øúí›ËŽÔåþ8û¸ê—rB;ÀWŸ€Ö‹žôÐ"Ábß·­Ý~\7R¡ýmfPì>'R^Lž†o­!GïÕ¸CÏþ‹­ÊÐ[qƒM­XŠ› Âχ:ö+ƒáÓ‰;u]¡0j±§C ”aL_ò»±î+˜8ÙA³t¦>7|)K‡YÓ¾OJ•90ö……Qy æ§š©’D¥a1 =׋–Õ ½¢9ßÀŠz«|4¡-¬]ÂôÃÛç`cŠGª×¶/©?£†É?yN 3àz#ÑU~6$º¶Æ-6"†$¼çHžÜA$³iArŠ—ò¼IuHy¡oÙOÑiD»<è î¾ÞÊÖêsdåxBBŒ·íxL˜ù÷Lߊ¯ ²Qšl?¶GvÃ=Þö«È%¡NYÍŠ<š;)ó„|ŠqÅ^·;PàlX‰›Ö, :OÝÿ2†î›%gìCñ`¾¸þ{ÏP25SÎ¾Ú ¥ßüÉ¢# @Ù4ó ÔÁ(_mWËwû% -P…¾>è\9ª½Ùû9ÛR1^vú)Ç¡aÂ-w4Γ* TÛµj‡ÎŽÎwÐì—Ž›;ZhWÏÓ9$¡%i‘ÌÑškhY5¿Ü¶ÿ.Z÷OÇ:U µÄ·W¢uZ¢MdÃakÝz´•’ÖÝKƒ¶ßdó…ÂðrÅs>’ßí¨]Qû²É÷etÕþ¼˜[—”Síí¤¡¥»#ùV¦ÏnëjœÕ‚*çð=[íáPvzü°gÖéÿ̇‹^øO=vþ‚Ón^MKu»Ç¹ö>Y‘WBÖ-‚½žôÄaîc|{¤Þ«d×½åd‚œ{O§ïô?„§W´å†¡Ò1:¼×j_k«hAcCvc¸ ´RÌòïxDB{á¶áØ«zèJ&ÜOJýzˆ•Èšx¡ïÄ“ ŽK0¨•kÀ¢?½ƒI„ßÛÁZ 3ïó‡aZ¼DH{†gÛß6]€‘zŽ:eý “­Y~½u"κôÆ^a©“1Œ?Ë«ð”‰øsYä‹B0Yk<*Ìh ÓBÍZz‡ `&[cȯà·ßŸ ‡’a.cˆz£´æ?j÷=:› ­ßè3=€%â1ßù»S°ìäšyKêüùu åÞÎÁjJ™³¬¿¿-Ãæ9›?©ŠR3„aGõȱ©Ø$`ó¹Ë„{Ìý:=ª‘¨.ÓBq»ItDËÛû\‘ôK‘é9$ó-·%=U„:ãìµv—‘J›à’ÓãQ¤qòTÜåC ŠüQò$¤·®®¶|Ÿ Þ¯FΦ#ã£ý)Cÿ ScëåLRdá=.­³'Yïå=¼×÷щPÎs"ûú-íãCÈIïW-^›‹\OY¿-eo#·êËšƒtðøð†‚ýxòýNPûh°Ë‡í2+ûQ0×D9ÉÛ…vxÏ&*× ˆ$£økÖ<àÜmV®'„¢YŒbƒÛQ\ü!½¨ú4\‹©îvC)¶>ÏÂ"E”¾h§@ÆweÙ¤ïx+ \ó'ûCôt¨`]¤çžÿ¯H‹ F¥ÄàÄWNI¨BÙVWXƒªw gÙ^,£ºú‘Iæs5ˆt/Ìø> 6šT=“níT÷/öÐ𕘟…ÚÒÇÙ¾ð…ôäƒÝçÉ?|¸þñÅÅãÐôþ ¹D ´(Lˆ®ê@«Ü>vh3賃so¡ýa u7ÍièX¢¾ɰ«ÿ(¿§K ÝçÎQJÆCOéh»üôÑðÛ7ýØø—9ŽàȬ‡¡£‹)¦aDˆPŠüÜ Ýy¨ô»ÆÆ÷¼úåt4&5ÖŸ‘é¼W÷3{Œá·CW)™%̳ÙOÔ1ä ×W­ã]°$.Bqg‹–+¸Ž7¦Ü‡×Ôòxý6X (sŽx¬SQš…Z°¥kN?õ)vžRÑÌø_CB"ÌŸ³Ä=F¯•Žpé Q"©[pÙY$Ù·ížH„¤¹lëÆHæ¡râ^4RHÅ¿->2Œ”‹‘?]Bê¤+,é×£‘Ö¨º!§j/Ò…ˆwóAúˆñµŸo#C›Ç[Μd:CUÓÙ/Ž,Q‘;{lÍá&4zë ûÙÔ_‘ÓéKßbî犪r»‚Ü]5ƒI;œÈ󓩦û²ònFv)V#ÿ±†oQP4·ž$þ= {E$Ju„£()1…[1Šœéž’*AIû;žçPÊj(¥mPš~å{Y>ÊÐ’Sä»Õ¡¬ª§Ó,1ÊeÉj‰›D…Óoº^áT²{þ+‘»U\Ï—’¾Gµ—t±²_X üõ¼ Q£ú–ìÖð4jqЫV¸¢6_»º+©+êß¼¢õuy¯qÞäÕ@=šI¹eÎÔ[âwÑIDý–Þh£4HT×ààGCÿ#Õj“Þh¤/í9Ù‚F;/(BGv9ðVM:£š\LLàBSù´ Iê»ùw~ÿ‚qúô>æºt¹•­¿úá8ˆtÇ.«)øœ$I¡Åa¨šOñ>tŠ«ü™N®Ÿ‚ŸþÎêL’SšäÃYãœe™²ß s-Çû¨8dœXu<ðªRÙhßð7´@oÅ÷íjˆË3ŸÙš„x–²ÐÇç˜ y«’ö`âAÈ É˜æ4kƒ²k¾ “û‚¡Æùâ¢ÕÒÔ½-:¼xšb>TX/BËWºJú½ äÅñØò]®óf¼i‡ó¦7 “é27÷'2è: Ýü+:Ú3ÃÉ2 ÃDnï^‡“ÿá«3 '¿ãªÓz£a9o|86aìé(£þHL¬pŽV®ž„éÒYMi1ø=ÞàAòæóNýI9v#v¸IöÀ²OxãÚ¥nX¹el¸ÿZ¬h¸Š0¯Àf)Ëóóܰ£Á ¨fõ 6æTˆâž´H2Ù=•H2ÍJP[…¤‘žÎMd÷?$ˆ÷RmGJQÙR:¤þ¬4ðÉ©éôê”"Œ¾ô}þx2êE]Ør}…Lãè5ó´%ûbë´7²e ,'zìób —¤ì‘Ë//òÔ×óÈìQz{ y#NúTN!ߎÐÙ¥­I¸‘IÅÿ[…äœ~ž…2a¼w‹]< E¹ž$.P» ¸}~úwÒ”¨÷:#dñ¥êY¯ˆg Ì¡Ë)ÖìPŽlå1‹’ÊñR•sGÅç_ó_û£òs9os‰|Tí)–£x|AŽ34‰Õ5Œ¼ßÙIû¡æuW:ôµï« £ŽÇãd‡Û¨Ç¹ÝÕŒgQ9šÑe~»~“øW9gõeŠ¡©Ÿïì±wMh.pÖ×óø&Z¸ÛÐ^  EK™” æê ´Ü\,lßA«ªä“ÞFh}ýkÞˆÚè«_ñ8†¶$G [ôѶÒ$¶‘ƒí.3æJ3 ½Ú¶ƒõ#R´ß ï'¸‡*ŸˆÔ2“á©/йã¤(Ó~,±Âö:Rì5¼®QCº”>¯‹5 ¼OP¾6’ "½)' KûÔÚžyÏ-¼ÙçÉ>¯ m=Ýå<þ{ÖŸ(þSWªiqÞ¸Ò Ü;ª¯WW¸*ÓáßõæF]ˆI=èðümç?zLT#ÏÍõ‘€ôà܆‡éþÝîâ‹O½!oÑ*tÿœ*|À¨>…U(í™Þ;¶…qWÆDöLBËÉÓQ èvKu÷â(½=äy÷êgøH˜tòõÕÿÝÞMŽÎ=ü·îtïéñ…¿ã°)wÝš ä™2ßÏB›¿~+#q÷:”û)Þu톊¨„ ½;PùíÇõFN¨6¯1`µûŸ#wóûÔ9>Ò¬¨ õ÷ß$lVäÂW‚WmCsOÂt¿*´Þ ´Ž5ƒvCšÎ ÑFèÔ~N<ó<¾½y"þ¿¯q¼ú¸zÍb²í÷€>yç=ÝmøiNÄÈ3Ãv†EóWî³°¼¨ìDb¬pè¸{k¶¹ä¼ “°Qe&Þ-> ÛÎéHpëóÏê.F$\ør bé&y4‡|~I‰l½ŽŒ!é×stÏ^"ÙÐfð\òC¤4ñ¡ž‰´GêÍc“y=¤ Œ–ÓOŠBö³ÌÛ#È8ø)­”÷2·X¼‹ fDÖ‘G¼0ÙK{÷[o!碫˜ðáÓÈÃÕÊL™pù(g^§ €jè›å‹÷PèÍÝáñ$<`ÀÃãÙ^‹â"Çž9šE£¤éí#/:¸Qúì‹„“®(ëWu”Gó7ÊßöÜwÒú*&Éþx‘D…Ê}?iwúQ-@U’Yp1XCjŸèj:¼`7]G­·¹°}ßQ'ý»þñlÔóI;“»›GU}£/ áïÕ–«h\dMdðZM½hÅT $Ñœ)ló&…"Zi˜ˆ«Û–’VY·µÏ¢å£‚jÒ&Z}~ÆÚ>‰Ö1懨ÛéÑæT•góE´Uÿè$üÚ1.—WÍ¥£Ý\ªÎK.¼ü›¢Àï| j‹ ~º)›ûbÛ·Ì'a3+ø£ý´dÆ»<€:žâlçÜv¨\¿g[ðŸó<³ç×+? »ý§Ü„ºòŸï=­d~rfò^&‰ÆA“‹G¾ËeH7I»"}°2§m:6 Çpöóyø˜(u*Ï)*<݇3Ë¡.t!¨:Qšˆª*šSW¡µÚ¿\— #øD…-|süÈ\Lf=q^ç5ÓR¡ož÷;ÕU^¼X¨]C?Ÿ0Ÿ`ˆ‚¡£êÝw%Ga˜ÔŠz²†—­¿ý8œ#·6ùN†Âhi‚…À‡føUºÒôö ŒU %~ñ/cÒƒ„01ä}ÈZO ¦˜¿¾vd¦Ã~kËf¬À,s9wUÖaø=0ä{x¸æ†{ãùƒtLÆT’7®#ý¦~™HdО“˜ ¶EFoß%‹ìÈ”ÈD>ôA ™WÃ^ Î «‡µ·* ²MT8£†ì~SôëïÞ"Çqä~¡6äº}?V~…¹•n~>Ñõøö¹$#ßl†ù‹(ý]>üh>ó£ó…–_YÊ¡0±Õ{zSQ"9AɽNº­ Œi¢h esŠ«}ü܉d²#¤ºî(Å£éKÅ—ˆÒWòž/§@YN·€ZûN”kg$—s@“$-ã>TôÓDÅÊJrQ+¡òByttC"ªž+ÏotDu1ƒR¦ÕïˆÜ÷mG‡ãéw–PóÍ)kA—1ÔêÑwŠêFv‚=Ô}¨åöi4v7ov\¦mBýá,‡‡,'Ðà+Ïò­ÞH4Ì:þæ$ÕÝa|`žˆÆžÑÝò\—Ñ–#ûU¢)#Íê¶!4-R¥»@f…ûëÇÐüN$SkZHg‰wKv¡Åë CÖ£%Ý¡è踴¼vwÝ»yƒQþ~ó³½~¹v(™‚,-±ËZµ°’àzgo Tíïn‡ªwaë§ÓÖ±{†Üƒ|Kë¶î]ž;žyÉ^²nîè½.W„´²Ë]wσdÍû{SȈÿއº+˜­AÂû‘1®Ù²¿œØT_å¶ë£1‰&' möΡ³Ñ3­#=rÙ=u®_Þõ_©Ü (ÒUò8EÏ Åûs,× t0[LŸü$”ö™õ´@åý;批ÿåù" õU¨ žÐ3Ú1„/¬Š2§ÿÇ_ ^ðŽhW@SÃW"k&h±7޲L‚ÖÂ*V§³¡ízîÔJ³´w?>ênŠ ó& ß «I ¼Å÷CwÚÅä#Wº¡§óFF=ƒ>¾ê^é¾ýÿò!±[+±= ]ɸÿa•F¨f5æwýlô»‚ÒÝ´$k$&(ø;¤Î—»Ã´ý£ŒÚ0;sÅföï¸OÏÎÊeibX°g‰²4&ƒÅá$ÝÑ Xpà»3µVX^È<ÿüV¿K[ÀzÝCº­øÇ°ù[™¡‹« v6º®)#Aó›óv¤¸‡ÕHô%箞/»º=½ˆÄ œŒñògôÞ&"@2`™ñÐ ¤ ùãKêÝ”]Ú#[n9Hý(Î\w¸iu*1—' ݵbQM¤¿­½¿`:)3úo!“tÛo!‹šÅ|´ ²)M~{™‡ìÚ3{z®³"§Ä1ïœç›È5?ÿëÓ"r¿ÿn°Ü‚<_ï,Ÿá{„¼ã_N[ð!¿Ù¹†;è…‚ô2–QØ`oí£¡x ;a/ëЇ ˜¯œD‰zú O>£”Ê—â½a(½ç(Ë´/ÊVŽ©z¤¢¬²× ¨%Êåü -nE¿­C— N Òñ”§koö£Ê%‘œ¿Q­du©@é8"gÙ Õ¥Ô¨Wz3X¶€ZìÌ>¡öÁÖ,g,ÔÑxTé¸u-˜õ9 PÏ<[ô¤êÇÝ~•À‚Ú#LWÉæÐPyÁýE0IUu]Fc¾ÏñËdïЄíÒØC 34¥×HLñLF3Zé3ÚÒ6»~j4»Î:Œô­Î%‚lècêkÎljŠÛT¬mÜöëÙô —`¬8Æ_ÀíTJ¶=´\Ÿ6ÎÞm–[ÜÕ‡òØ¡/룮ù1·Ü ²NyôŽÞÿ™Ü؉ÛÍ—…“gÔCŠc¿ujù ¼Kð#£¤…¸x~Ç\åÖ¿œÈ›>N ɯS<櫨!OÌf€ùž+”®ªoK}Ö…Rù²§ÒPOä¬9é§ Mâ}§8„ åºW;m*´¦ŒT©0ö@[õ#z³Pho;'•‘BnGå #>!TüMt¸~Ó¶P¾í3k"f¬Ð~uá–¼qÇ.Wº’ ÿ’ùwÿRmA7n/>h?/ŸÆhüÚ“ôýk©"¡cÏc¹š²ÄÝÏåT3{ V6ÒÌO· ËÑóûd0|Ë£2w>ÆßÙ®l êTBO¢VPÛ‡è5Èn’»íýŽÒ+ÙÅ0xf‚”Û¢~öÞxM‰=0|qÚ˜‰·Fu~²‹ùØ)'Û çR˜ˆ#2¸’5 ÓÖ§[¹Éà÷‘Q‡'Ø`Þa3ý9,r¬l˜ß̓¥­†¬'äðBóŒ¹ÚÖœÙÒiç`c6Ž»þÇl¿I¡Ô<èŒå\Z/ Uqm±(o# ó.Ö‹!©„–è«æ.$sÃdãjHQ¼0ñå R‰¦…6f#ᕹôÑŸHîÎBÙs.dX<3Ö,t ™î¬x÷Š ‹©µûõœ$dÓqVˆGöu6>]]ä\{ýÇ`R ¹S?ê°A^-·‡¯8ôïÓKB+OЮ©âÊ<…‚Ó[÷ûŠQ¸Î½×7>|]n‘¼ÛƒâÄ*÷»†þšÑ’K(kÁ©XÍ‹2d4JÜ>(Û×Må{cå_¬ßi쾊îngí³Î¢²K†îu%T¿½Ø£¨‡êÛ¢J¥¨!+¦ýñÍÔô[ãð@­¶gyä6 ¨ãºêÌzìV!Yë´ çÅ aÛÑÙZ24Nþ ÝÛ,¦>i>Ë¥h.lÜ{R-¼râˆÐRª¢vó»CH×ñWh5úÓ^˳ ­‹8 üÖÑ& VX‹¶Ðö5ñiËKh§:qÆ>§í÷™0ß’BûÍyº[>'ñÐø¯£ÉãáïWâìÒðdß+™‚iz”îåótŸ@òƒµ¼²Ìëð³/éü±ý¯v9 ˜ï*C~ØÕ†£C%YÿV{Ú÷¤¼ñ‘<¸3 I ‹ŸÅ a9hä|Þix{ñn„hÄOú?)÷Ƀק:ä¨^ƒØ'ÇãtK!Æ„õðW¶Øô˜Ð7_žê餌ÓÉÆ%}§þÏÏ”6È‹[Se‚Ù;Ê·áCw˜$å(ÅßõˆKâl#PXÕDê@EÊç÷zèB‘Sùï2f(Êäb8mïð?Ût™Û‚ñðñ'ñ €PkÐÜT; %~Óa§å^ì>”ïõQ»A¹‡ÿƒHO¨Þþy0Ã*‹JL„#– š'»p@íÔl¹ÜÎóƒ:"cÆÔûw¡^üá×8áLh¼ÿ™_œ šó PÍ]V—7³¡m¦vÊõ.tô^ëu½¡]óo3ÔŸ†ïÐV%ÿ{~üÈ6:wï'ô· ¸4§°ÂOª +t¼ C&©æü¶0’QyA±l~%u<¸š™ãMNqÌD¿aJØâ–üÊ)˜Iö¤{Ð sb´…”0ÿìšyÙömXñb¦|ÞKK¥ÌrRöðgUÒZl¸ Ö$…Ž ƒÇ÷•/‡ØÂ¶0µd˜j8x§¿åú„„"·kÖ‘È0¡OÒD‰'¨KÞEÒ”ÍRYj$ût¤²,Å )y÷y|³Cê¦GïdZ:‘ÎuÊçÝUd ¬Œ #.CÆšiBÇ?UÈœ•{Š5bYK¾ëÜ=h‰ìdü}-{³ðªÁ«ÙJä^Ô%0aÕ@ÞA•ð«”—Q`ï¥7r$QȽ4¯l¯'ŠìÌG;=ÊA±ÎwQÝÔ1(1çq½oøJ«Á‹š$c”US4–%«By«‡Ôܨèyã²J*gÌd`ªy¸^ˆhkâ£Üš†9´)Ža¨õ[2dýŒêÄŒU¾¨F=7¥ä§Ýæh óhPrŽ 7cãÏŒ¡qéyý}„ShzáÃ*{|šË<Ï+É¡@‹ mCÒòOÑÒ~Z"¯Â­Tä¤O¢5÷¸æCc´¡` Š®©@›õ—I‘.ïÐv1åžGYÚ­ÔXhõá! žŽÒ¢" £Ÿ2Ôº·ó=àù*î‹j•÷‘’@—Ãw}èD‹\X7ÔF½T r‡Ê§~êÑvP²|Œt þÝn¦”3 1±™“C!Ÿï˜ ä\í Þ4¾ ™ê!÷Iÿ@šá¯‚®€w./4˜59{N¿Ô59ô=F8A•ª¢>-(Àçø"Ý“ЬÚùKê´…l8Ý*\‚Nõf5ñ/6ÐMëG}nÏ ø¡öâg¾¢ô¿WÒ¹ysJf¦ü:ágõâ‹Ô€*º*Ò´Wφ•n‹°>Û¹½ #C0òøÝª*ýL2Y¿šã'§ë%`lœúžX*1Œoî{ó2ú+LòŸ ˆ‰6‚©c}‡‡“ýaú­/‘Ì^Ä#Œ7ü`NpL>ƒžæ÷}“3¤Ë‚Ž1S«¶ý°È[žÞéKð6Õü&,‡ú”p¦Ý?3 õ|ƒ °ú`åç$¬ûgmœ´ƒÍëaÝú‡¹a»úúŠˆ ¼â (ëEÂÿ[šg‘È>Ô/ß+‰+ŸšÜ@RÕŸÏ1¸·;žW[IÉß:yp!åKÍÛUQÄH©xQiõ Ò¶²T/ÑÔ!ݶû*ãîÿÏÀ¡yaNl uklÔÏ«#Óe»ðÛe'‘¹9†ÄÂP YÕ Ö.Ÿ©C¶*‡ ë#ÈnK-”=ìŽãOŒh¯- ×r· jTÈ-¡6.®÷_>LYÜôˆ¸€|3ù'È]õvù0Þ{äf ~ Ì~¢ÀŠÂäW§/¹6¢ˆÆÊë "xàüLÜÎ#(ZOl'y0Å ˆœŸ™—¢K‘œÐ+”Nr¹Ã‚Ò7ˆ©†M4nŽŒ¸Ò.&Ÿiœï'AÓÒå÷ÁõhöI7vÍÑÍ‹ë:ùØÐÂíõÖ”n ZLááC©cht~“ÕŽ­h¬®Ÿ½<‰VibÖµ1ØœR¸¸—Õ]·X¸¦ò‘å…ôF‰Þ]éÎ&<íJ·~CeÙ;ϰÒñ]ÿ ukJ€¼Ek;Ö)·¿õm4> Ë\â©Ò@ÚÑ:Ïg i}+þHa $žb6Ó[‡„ËÝs$¯#!áóɆý—vùñÞ÷Ô{ -iM™ÅF²Îž:oU ¹rþ´Ÿvþ®Ç·Õë´" ÷žÓ‹áÓï7_.eäCivÉ•šPÞÇ{¤ûf!TÞJYXg½ð/޶›‹‡A핆•‘ë¼ð…«‘¨=bõ_>$ ‘’ æ„¦ µÕظh¹ñ;åo=ûsyjž ñݼ{ëJ¶Ø/è`ç-Yø¨Á·\èFá-Ýè;3t†÷gvÐAÏjY´ê¨>ô)žp}ܯù/R%¸\?4C¾çþÝË?,{"a4§Géàƒu{âÅA¯lyi'l½ajýF±N̆‰MõimÃ\l2‰‚¹ ,06·d¨ÀâK¡+°FËbOLÅߟƛïîëÖÂê]{Ú¨¨yXRz\tëlFó3–_„íoS¯8·— &}µäS1/s~çøDsΆ§#¸^¹òþB’žh›±ÌG2ÞOƒŸÏ8!ùŒ1X÷ e‘·[ÿR_.‹Ðø‰´PîXªt7Í Î%3 ýƒ”û瓾e~…udrõ™ìa¯E‘=Q#RÜÈÆjâ»§AÙÙÂÖv("ÇÎæ†oÈoäú¬æÙ>‹Ü×çù'¦#ç#‰ QÍ<òv ô{^F~eòÁ¦^GXÉáh¨ÂB·&®Íá4«}iõ(nèPÙ[ʆ·HܪØQŠ/«I½"¥êmè]AéíKUÞQ(«xê„E<Êåë‹è.D…‹ö=®–&¨äÿQßTàªù€fÇþD&–üFó·áòýWo¡EÐyY¡ ´ã½J(€g/ÅÞ3ºƒ Éà÷åóÒ´ºE÷G˜.ùÐ}ÿ/»úK ×}Dów?6rÙ³Aï¹ãã¹²?õ’è‘qîæÔ¤?s"ý±ÄëN»ÐÎ_ËéqÝå¿ é.x§p}sî–Ä9<ô<_q…ÉÕ5ΔÇqÊóf ·{£¦yJOQL\ð£…ꌹÀRÅ«ð%AúÐÜÂ|-!bߨåC¬adV³¸K47—¡Í²tþ(c*´›íÐŒú;@ûÚׄ{]¡ƒ‹·þ¦sþ.îï?ëí÷J3 5«¡]]oÔQ¦ÚÉ´i5—_ÿ{þ¡!9]½š´?y¡ ÌíCí¸¨o@‡›6ûÉ9èuÔj·΢S=_’¡«2\pyvºÙ.É(Áw¿n“µ)èYø5 ùz¯¾%)ûýúÓž²mÀ ÑÃ'6ð³bûÙQÏ]6• t%h†Ñ½Ÿ¸«U¦`ŒÈškº»& &Ò4¼ŽÂÔØ‘ä£T£0Û¶c&8÷æ ohU>x 9sߞà XºÍôXêé,üy83JMD«ßæVVê‹a㈊ȖÖkØ&Áy=2$8 ×ïáB^oHÔbßò(*I:9þœ}vÉC3Yï&"…yŽÀZÔ{¤jâ¼ ¥öi/X¦Ô#½sûýdˆ³°›v@&iǺ€·¥ÈL^õç‘”æ9C,™ŽÏ¸Þx9Kx˜0êø?zL¸üý›åHö®~ý©K¦H [OáÒÕEÈ3{½ÊBÇ /¾«ªùVÀ‡ë‚çh‡XàÃF#¡”Dúžë;ïÌ …cì¤=ïþÎ'š­ßåÄŠZ™ÿÙ~$ ülà4ñw¾¤°ÇÝŠ÷ŠùE°ÐB‰Åë¾Õ¬A(Ìš¹}GʳxÃÎÆ@ÅÉC$”Lþ®ë¸T§ÞUó¯$ñ¾„šâ·§îAmÚžÑGvAð¥.|Y;*yfÉ´b éÛËVÕ[Ÿ¡e2"³ìó3h{x…WŒØ:Îûå¶_X„®+íL!»9»iPtÜX~xÖ G}„~G˜¤£’‚ÁzûG¾ùÂPäëË ËaÄéANX#ü:$4óz$ÆÃx%Ø(ßÂä·‡“þ}Æ0c›¡ æ.0G K~¦™æO»úšxÂ"¡lÍ„T.,µž"ü(:.ÛÒÁYte¦j/loÓúZ [sû’­ß!ÙC÷€â]>̯¶}v ‰Ä.©IH!q¹à2=¡<’^‡^?Ë÷«F'xàÏ|môŽ2¬^çh*"W„uûj{ï)*Ø<¬2ö$“¶or”¾ÁËyAÝHXµ·ª«‹Dû‘øs3 ñ•×—#ˆ9d™bÏñs¸÷}äÑW.HnõBÏê¬7Rê»1^ gDêCÑú*ħ‘6è™ïMªëHמO¿¤Äô ËWŒ‘›-ÏÎç(2z¾!WƒÌ‰ÎÚy?EV&±Ñ‡o‘-fqë5q²+Hªzñ "G‡5Á(=r„ ^aDn™ãu=^âÿ‡SÎZé"ßtÉ¥@Y]>¼qó‡b4 öž¾Ã¾ŽÂ4ù¢cþæ(b8Ñ=,u\ 4â>ˆ¢]›Svº\(n[LbÚ„‚n|n{¿¡”$ù•œåk(}·gÕbƒeªµ½$™Pî«æ­SgP®ª°‹£¢cdÄÔa;T:ÃWxÆ•Ë_5¨Î{¢*´hÝ”<€ê{÷Ÿ`"!Däj¡¤û¾„™NŸî&´£æóŠùC%Ô–zð´….u.ãåöF Ôý¸×Íô€&ê­_/¬Ü@ƒ=Mœ|F¿Ñ ç;ë¡ë¾~uÁ94øPT¼Ô‹& 祚nŒ ©©°¼v@/šì [ÈEóûmCø-\Ÿè¶&¡Åö¦¯ÏN Z¾kºö]p­¬ý™cŒŽ 5•‰Ÿ® 1Z·ºbäô v,Må•B53½ÎͤJd9V8ü4þ\Q™‹0퀆_T ”ü”Piûñ!QŒ|zp3Hâü'È{¢pÈi[ïï|ýécó;Åtä—@É(¤™2¯—BRÞ)U/³¿õ2Jöo vý²quÆÂâïîGáSùC§1r( ¿;PNå­NýnJñPQûÔÔ¿ù_>œÊÿÀAµ‘ïSB©Öà ÿûÿÇùÊ.t]€fÒô™±ïsÐòœÎŠ`{?´Ök üfJ„ö=GYOÞˆ„Ž#ý,t¢ÐYfÐØÅ°ßÌ¢œÈ¨á;ómæ=Búðƒ“caNmúLöެ ¸ÿˇ$zNªç`è¨ZýN› ÷Ô¹UÑ_ƒÑ;¡Š&0æÜ$*A&¼\cd¥ü`*6d\"è%Ì2t…7 Úœ«ÃÌéw0ß~Ò‚ö,Z%?û³Kƒí1ÔgÄáOxKdu¬j‡(&0<‚õlçý•aS›š)uhlþ8žW-Ž^ÇbIyÏ!aV-+‡±w’Ò¹‡Äoì}òt‘Tûη“¸w{óî»#%HÞGùâV'RÆ<~-çÈ…Ô§úË_‘ ­R)ïO?‘îÎO­\£!¤NÙøì92ôXW˜­® Óá7[NÊLÈÂXB×UKŒ¬Ëï)¸ÖŸà¾¦yþûiÈÑúô–”Ûr½xå<95‰Ü.¤5·†•'1Ä&ÚË y+ü'_†õ!?Üú è—¶8KƒÂ$4”RÁçñ@ȯygŠÓ=P쌨A ³Ík¦sgPŠâÙ ÿžQ”–} —?†Òâô}Ï¢¬üi1³€”û[Ü¿ôÂrœ4¤QéjýJ\b!ªä†m_]ÝAu¢/o?±ì ÚÚ¿xþ* 5:Ÿ„% ײàås¨­ÝôÔ¡Â uÎ/ÉÑÔ8¢n†\\}(êµ"Ë›pÔ_ñ“ñøXÔ²‹hA££Noi'ÄÐø@Õƒ÷ïÐä7ýUjc43Òz}Š—Í Ò=ß1* ÅˉÐâÉA´ô¬žmþõ­t|“tùÛñŒ³½{Ê[ù4—oz u%[jýCøõõ‚Åfä¨zŸ)#eO¸ú,eg×ïª×îË߇l•ónóIx_; SDéóNûÞ$¦Aj‚ ÙJ $ ¦iÃÛò‰ëßÉ މÐO=;îï>}{w¯î Úl‚ÜëojÈè‚¡¤ï[G)w÷öƒóëî‚/úÄgWÎUÃWS‘"‹æyh^PœÁýвè$7L@ ­ý“­‘»÷µÕ‡‹R@»·ÿøÁŒ;ÐþüË/öhªžNP‚vé‚“ò>Ðöç¥Æ•}‰Ðö]ˆ9Ž„j÷õSáò/õ×û–§ÝåÇüí ÝTŸ]¾dTô£¹ ñ1–gYT¡Sû…¡Ek#tñ„“¾Þ´o^ž#t«€îK®»_É ¾7KÄ0 ¿‚b=•$½¥Õ­k ÿ©@›\˜ ÛÏMi~&ùßûp¿ †¹/%¤ ü‚‘t>ct ~ÝâG«FŸo¡Þá탩à;Csë•0k°O³@Tæ’Y<›EÒaÁp"wp– –ØŽ\hä_…?¬»QL3V-¿h$ì‡õ†ÉØòèúf44‰ì.c‡0mQ€Íe‰‚¢2œæ‘$x¹RÃE÷Æ•äÏÒ!ùï¡3–?ÊŠF›$ýÒÌLVlÊ·!]T^ÛÊid0•®=—£„Œåô—ùqyîGVoû÷Ü÷ìiö™³|ÈIÐAÄŸˆûSE“oEžR«.›@>yZÆ}îÈ_´(M/™‡‚Îð»”·…eh¯ðܾŠ$ øR´ Øû@#> ¢ñÓ"ý´ªhzŠ{àö³84?Hürï¹´ð?[år~-íRæB–ÑJýAöcÙ~´µZ\8Ú6\ ÏÔ6¡í>}o´Ûï”t"í%ùiÏ qà!ÓâÓ°»xøR×¥U19t(Ëð\ »Nìn²VßçЋô””Í¥(Ù|Föô^WÜ{±Ò^\É ~Ò „N ;@ ¤ *u@žâ~kÓûîÿ®³ŠÊ2Ú¾>%t7Ò!)ÍZ”¤t£ "6"˜¨¨¨ˆXˆ ¨„"*Jw‡¢t#ÝÝÝèçâœïæµï¼cì¹çü={¯µvs¦ÊXAÞ$|s`¿9^Ü ±Þ×OØYÃçDáÀ¼¯²ãÂKÑÇÑÿÔ¼^¦À{‹à›ÕRŠyû€Öu½-ˆx9e¯H$ù_=î½ççY‚Ž–ƒÍO÷êúcˆ„"!c[/‹>š² îNâY?È‘¯Êb쀜8¯ý ºmËËšySOr_2%ÞóͱC™Gyjüò/Êÿ{Íïï~-»{®{£5¶íwçR¡HÊJÍOJŠ+EU©ØöúV¥Ý¼þJìÓDÇo­AéÝ‘¤³«1PVú4üÙ~øy{̰ùÜkøe™ŸãÃå•nW¸F—â :ùê "ÔÙ²”¢††²²•.5hRRÕ'^ ƒ–ý'žé‰†6úÊõ©S¦Ðî<{:W:}¾| =#WÍ/Aaƒ•= ª.Z·zJÂ0Õ)­o0J~Ó°½ÆeG 9í`òq«QF™8Ì-žšl~ sñ‰Y‰¶° £ÜÞ4 ‹¯WüƒåD¹õº4°šþ³¸Â Öix¯›[Ö“e*ìä ó‹W!Œ·P} 3¾Ñæ=ÛÄ4–r™rHò!Gw¸©IO2{Ï,!¹;Ÿ¹Sü"RV Û»a4—Mmþ¾\F:Ù—‚™HßpîüC’«Èt$ñÍG/döò^WzûY=O= /@öo3 ÔÈ¥]ù0Ýy]Ÿ&f¥"@‹Aä31<ðÁÛ*…ß…&xîÌèN è¥×ÏtýQBFøVã³h<¨ÈÛáFl2­þn›ð¡\-/¯}˜;*ô{h5’‚JÄ \5Qňu컪-¶žÕü,ŽHá¶c÷µD6]DQ»î÷Û9VNÔ½óQëæ~Ô;l.>&ð øÚý_° ¢áZhViB2ÿΫËû¥‚&¡¯:FØÐÌã·a*š'ô7[Òù£E¯Y‘ Z¦^_óÖÕE«²…°Ç*nh=òæZZÚ²a.¿·†vÇÿôQ‘Eûïn—4“ÐQƒfeß‘EôUKSé%~ZÎÀlqò²_¬Ž™¯ÿ ›&ZzgŽ­BƒüúZ¯¥Tœ±·`"ñ‡’:7o§ãbPèÓ“iÜy¾L¬ži|ýÐÆzÿñÙ]½rdåÍþ§bì H:½E¯: qºNüä×!^:b=é_*¤]Ó5ÜÉçÙ«Ç­$kn…rŽVªÓGù š‡@YoÜZ·ºy\ y<”…ÒÚJ£z-õBÇõáíž5€î«;ù‘¦žÐT±šæ •Øx¹Ýe¤P+þæn\Q†¡m- >5»ûøÊðý-îX[V´8#-¿ø½*`tk¬QŒ5yw_w¹Ð`1L¸Üõc¨>“_,¤Ñ¦ <ê¢5ÞÀÌK³Ã[Iù0§£ñ'CùÌ¿j–N2“„…G5Ža®¯añÎÓ"…·°ä¯¿ƒW`Ùûø`ÒC€•hߪáLX]Q°¡çþ ë·°àþ"À¦êŸ“½Ê`[byuxü5I}êÖ&‹VUã> _0ôjNòK$5y#×â‡$šçŠ ’q_²ò‰×$H†÷g|I‚°5^âF!R.=ˆti·˜ÔèvùP€‡‡RéâöI´ÒE }å0ÅßÝȰîÎté‚2Év]驯Cæ;MÇ^N!Ëdtã—6ud»þÂT>+ 9X-Ìù\F‘ó{vý¢®,r_؜ިà@^i«*&YŸÿ—ÞàcB©ŸÁc¤Ú»|x)¶qä åçˆ=ø×"Œ£ç,7£¨e7ƒ™r)н Z‚ªZñ½s¬ %O^M$9„å7´^NУŒâŧ¤Î+(š6Tü6åEWÄò92P¡ÆCë½"**y)ZÝE%Ó~ÇM/T>êx©„gU¾¾¦!1B5Áv — T_¡Pe|ÆŒÈ5¯÷Ž…5‹z>ÿK×AmÂ)C·Ô9š1fÉ„º_úäŠ<ðpíÄ s5êíÄ´·F¾B†cÛ“§äÑP¢þÒ–÷42º²-œ~/ý|Þôë:‰ríá„04iѤZeÊD3–2«‘ÚH4——–y5…æƒüä2Úha?¼š§‡–ÎNéÉÑJê²4?Ä¢5¥ž 'ð õšÁ ó­1¼í‘móæ´.ª©…v¬C)ÕŒ¿Î°2%CvðT‡|zrß`¯SCÿÍ?@ýò•WÓ®ÞØcè% =âhÝ÷ ’Áq¡îæ­Ý\*Ÿ•:J _/¶·üb_†Xê³Æs{¾GØJÈ_Þ8ÿÜÇ”º›Koȉ[Èíú`ÓæÉß={|ùŽŒ ½MRÚë=ý^¿¶÷ý¼wý®èÇ•ý¤{ýÕæ-NIìõÇÉ[Køñó­óÔØÓ=ŸR‰nþ.¼½CAñ ~½xu”5¤*Å^E1g|ÿ2è)ùÐõ‚\CrŒ„Ð{9Óþk!4n›§åƒfM¥¤#¢»¹4܃æ4´“©F0üƒ?AÉ Í«û Ã(ÄØÚûînõ>òùõ7èq‰;çØð_>ìß³_ "€A ’³¾ëÁPIÖ•„ˆ9- tÁù=Œ‰Ê›UÐÂsýø-˜Ò¼²Ï>9 f¾ÜgÏ¿“óë|ŸŸ-àeûz¯HX¢\›_å0:ûÚ”lXU>pO§ÖÖ?/|£"ežÃÖôs¨îøË^-uä‰(rh™>AÂ;Æ„6áH4ÕéÈÊùIn]ã󼘊¤¼ÃKâø¬[Cý RÄ-·¿k@ª{…Ûf,‘Ængƒ~K iå &v[¤ ºR·4“ŽôïoÜ;ðxº×ªç/#“e+m¯+2ÿ»Ré4‰¬-´ŸTÄI‘½äÛ¸ó2r~’a›ÓoCn ‹Ä#È«ÄuF¢cù‚Wü¤©?©|Kî*دLeÍ(Xžz‡g¾…‡^Iv‘{¡˜y´è²BJüù•Xg|2.¢ w)JïŒXˆ£Lsd}» ÊÎÌûÔ^·DyyŸQÚ¨ÏÇ¢²ëSŠ©{N² rp××e±}¨Z/}Îþ )j§miøAôúäîœr5»tæžX“¢6«€¼Gq-ê˜ÖRØÿ‹AÝ Ïîã¾x¸«üÆ}j@}Rª´3’h`ÉFU[ᆆ‘›šÚ?Ðh…¾”«Ý£/³ÏF“VÅ¥w®¼hvRè¶]â34÷KlÈ´8M?ø4d -í™âoZS£•SIõ“²xQ“þ·PhÊo’È•@êþ’º‚ì(µþ~ö*”¼Hëæ© ‡¼O'v¶†AÆüÓ SÇ 5.üü›\HúÉâqr?$̾þ–{Kâò6ߪ3oÂW£³}>¡+ ˜‡áý÷uÑìÝ#R®íõÓ¿ ìÉ{s¸s -( Èüó’šÝÜ·sê¯Ø-Qø=þÇg1jzL].|š‡z·Ijq3hxË:“s] ÝYÒ’ßAÓáís—¶ò¡©KBÎ*š™8Nj·í®ä¦ò»è ©ÙÐê™74¥ÌE|%„¦x οLÞÐ4<Ï© ÙºÁƒwˆšÛNq» ¶JڧвÀw¼¬ÄZ?v˜5.²@Ûã· §îé1³˜ºí-´÷?ýíÑàâ~S’ïh ë9ÇužèÞxtµ-gz3Þj»ýNÊ ÷'`àùdMž .”7­oÃp`Y›CŒ^*}UÓãáÙžE0E[Èí; 3Eò‡ôN&Á¼éçÀ®ïÙ°Ð?$U K_bžI#†•ˆ ß-ßã°Výñå¯{÷wýÕCÍ•Ž¶«.™¥Â¿—DW®+Ö!¡mI;íÝi$({*l‰ûêé…Çvýt»ðçÀ-¤¸&ólˆ…©ÚöWr]Âý'åLÎw!] Qh÷Wf¤ßdTÒõfFÆ[‚÷BT™¯Çc}Ÿ²X{êGvA·¿L~tÈùÌU»_ÓyÄ内!ŸÑºt² ò/¤- › ákã”s(ÄÒ÷Aª4…G‡›½ ®¢h‹ñ8£1JPTªe- ÔåÏ µš(Ã!~ô©Ê&Ôp}U€òÇn^{xë#¢Mö=ú“ñEž:ŽÊÙ¬ꥨº²%oÈG‰ǬÎìꫳN§~Rµ4DêZÂ5P;I¤èjDêš0¢ƒÉe«Ç ¨?}ì=ñY4,]æ£q×Bãù–ÒšhrÖ¡óíxšI÷ÿU¿þÍoÖYÇ]ëG ×xÒU’§hiÅ™"þ­Ž<+5r¶CksŠ¼È…´9nÛ™Vƒ¶>\ º^æh÷í{ßæÍ*´[QO4+FGõJyiòwx4n:ú¡?:)Å>—ä Á³)t¡T÷–x°(’~ÄEÑÊÐÿõöäùP|ÞÝNd_x£sF’Âö©|rfÙ›ÛÄ«Çc _z–MÄy'à3™ÏPWt|lšk;’‡»ü÷~ÂêƒD•{Îü‰¥…ˆY"ýw!Ûjwþúòõø¹,nsÞáЮ*˜Mˆ½ÙÕ7Íã_ÃD7s™FzvoNé;íã=piµ[rìX8•ë!§¦ðÙƒ†«}Ò`´Är6ȼˆ÷úsX]»;ø®yW »«oB¾LÀy½¾•]Ÿ} q³ÒŠ(IßÑÆ]€â´߯Ÿ,Ã]­ËwO@‰>呟?¡ôÄÍ^Š”›PrƒJqÚ~ªI{ê AÅæå'nƒfPIpR÷©C,TkipŠ}‚Ú–WYo¿[BƒñÓv‰¹~hìZrïîi‚æ¯f®yv¦»úîá)51€?³››&„Ðù`޹ÌÄz.]ùÔtú'Ò†× Ìg { Cù'¦¾ùÁHnõ—vÓ2Š7âᤂIeÖøÅ8J˜NU zc×sÖKvvù×a%ó”#,¦WŠ|v¹ ËgX"kÇaõeá½¥@X/ðõÑp«…-rÛã×<açÜ»ÍK‘€ÁiL4† ½ N¨š"Ñœè¦ÂY$¹gð:÷³’bdgDÖi$×½ÿF2a7¿Ï½Ôök—u%’‡/=D:öA!“nHŸSØMâg†Œ×îºþ™-ºŠÅ\GVÄ$¿UOd¿i98ªK‡\dÓ÷%—"/‡ÔƒÉÛÙÈïÄ/gô\¸®¢í@‡B ¦®~'ãQTä4‰R³'Š÷œñ25ŠA©öm¢¯Œt(#~«mWr¡!œ4QákÑÖË T¬yðRre UØ$îùýcDµ_ä Ÿ½†¼@ètj±lÌ<òAíâPásüQ÷âfÁë–vÔS욦ÕoFÖÍŽ‚¤L4\¡ÿnÄ€ÆµÝ OÅúÑäSºsž>š=ïë9ç‡æ?Ǫïì Ecnœ¬ZN½U¨MGkæU-ÂÔhcå(®Õ€¶_ Ññ£=óõ~ÿëTèðIþï÷7ÂxÔÖï‘hX úZž÷cª¸‡š?äá÷wdwúÝ‘ùy6î?”ºàü(K©2TWàg‹b×—*6(9ïÄœ¥Í?"¨¨®CnýÛ'DÆT»z%ÿ|Û2J®{¿´pƒ4=e®­d.H¼×Ø*× ß´ÞÚUX‡8É6¿ÔÜ7v0å£ÃÁtà½õ ÊŸ~šy5 Õ;)7~Å1Bãå}Õì %þ¥ìÑ vøóˆüëë'vý'j9d!zè¥Eh;É Ÿ&.~Õô4 8ÖöûÊ á·[³ŸNÃàøDòå'ñ0tÝ'^Ëž†šÇóüÂp—W¤úÒ+éÉvô€1Êr}ð€qÛRƒ»”0œº\禓£–ÞÁ÷aÚ4'ào€(ÌL\Ò› Ú†¹¯EÌÄt¥0ÿ«R2ÉC *fЍÃbò±« #-°ôîó+qwXÕøÛík+iÞn°º> Î ë'¿u†MîIO³é[°õ7…Tî(üe8lM,¨£¹|)Ÿ áéÝ£}‰’²Dž¬"ñ¿«+Ú¸ïŒmžÐ'n$]\´z¾ÉÓè÷“øR!å‡O½„òHpl@ë"î¯z[•ANˆtwmgM° ýÇCVû¯"Cã÷«§¿"÷´ÿŸÈl=«ãš¡‰,%ÌEykÈf.üþ÷é"d_¿âöçÃ7äLÐúdÜ'®¤«Qñ!¯¤VµÉFõÿLJÌ_îÎ ¡ÀdmœM‡Ö.Ú–ýqG¡ÂW#7‡¿£;ë×¢;=Zg\ EŒbo_D:_HFñùYæÏWÏ¡¤—”eþä-<¨äRTeŒ2Á¿O¡ld@½¤¢ ÊK ÓQ‘Z¢Buam‰/*Ê—¿µJÚîjUÏQÙ X€¢þ6ªúl÷²J e_ƒØ«UT\ÞÏH/¶Ë‡tøw9P³Æ9г8µùDÜ,@ÿ®†¿!wP·æîÈW&V<W˜ÐŠuº·“‹ ­É< båæÑ†\óÈ·2ÚrËú}¢PÁÛwŸMû%Ø£ªh§M>o'2¯Noûæ›—žî/Û=ßUùÖ.ê@ÉýÑ »gÿ¹·_¾8íÛÇÆé ÝXÂøùHѸIeuCâõ9lÎPîæR=ÛUïöê)¨½WÂàóbeaZÿÞ;Ó’2éê]?=ßûüŠÄ·`åC)H©Š>þáç¤å¡xÉYgW˺© ×aRþZN"¸z=,ø1 Å‚î¯e7?À¼æ¾ka—¡ô&Q÷²ôÝÿñáÚmƈuvør•:#*¥¯žË®>ð?>d>Juî Ôk=o÷9 í­,©ÐtÐùѳ‚h¾»_»G«ZFl+8 v Mïç7FjøÓèÓm<´÷"îÇVeA—kK¬ý´=ô\3š)Hø®ØÄÚ¾€AÍš}ú‰Ç`è[ŒßkaÁ %JÝy]x{žÏ§Æ›ªÉæ£ïÃäÐYCõ0£,‘ÇIsÕšÇr¬ŸÀ‚ᬾis,–|–7\lƒecæ š°2k·y½4Ö’*·p¶ÂÆ3IÞ½^Ø `¹« ;ùùËæH@Ô06݃„†+K!ì»z.2³ÛYü„$–Áõ.œ¸oÛóoBK+’åÝ™tQ÷EŠÃ؃ߪ‘ÊuÄá‚Õã]ã–M°7AZɼÔ9î€4í1ï? #VËõQÛ"¤]þʲ»yºô(ש0››»ú5²¼¶8(²@ƒliòFùoÿ"'½IZ‘'rÇmiì"Ÿªeß!äO¶n•û|Èð} Iù‹‚õ¾¦Q'SP8"_¨þÚ)}yýšÐë Ï6åyÁÿ¥¸2^º%: ôÐYgA~”õ‹ñÌiÙÕ—ÃLoµ* `ã³sº¨œe˹ìˆÊw‚â]5£jaÕ[é*jŸàçÖGL4~𶆵ÄÉoÇG vtóLF>êê]i‹VÊ@=Úì ïÆQ’æÂ»DhXTÿ¨þÅG4ö+pê0;Œ&§BZcU|ÑL®è„×­_h~OrdéZ ZxÄs{0]D˳_¿ñ •GyPŒc'Z?°Ï¯æüˆ6_̼HΠmGõìÀíi´8–êO H} â™EÇ킉S#ñXØUOÿèl=9²ÏIÏH+K-Rk¡Äd€…î,îûUôBȱú5Y<¿1CÑÉ'6Š>“Û¥ÔØÑðì‘6|¥--ÿt§¾¨V¿Ÿ\›•wÞ­ÂÇõ¤/CÜ!š‰ÈÑ/9¢”ÌN óŒC„Kÿ`°2„7·}?Uø_=~>Lý=dï>ãµçßß}zù&͹µ3a÷Zšü{d}ª28lKÙ/™Õ8;Ô!‡¤@TÝó7äÜ:ázŸÅræÉ&&Jw9Ñ5v$Ñ rO%{+æÿŸkžT…Q ä­djïpgïÝ›Ä\Y€ÂѼÕ¡4(g%}CÞ?”]/<*ƒ’C'=?¼†RÝýûãÏšC™ÛÉÛ¡|ó1 OTĈ5ù¦ ¿#­Ü¥¾‚ªF·Š0n¨µ¢þúËá Ô¯s¶—SC£obpVÕkhÖÔP•ÔÛ‚VÙyÝw‡tàÏ=òÿŽ-B'qr¯’>to4ò~b[Óµ”>ŒŽ‰Á·]Úb6î—@ßsꌟœGMI ɃDÈnRAç}‹Ö³~2]r±‡ú2¤Ó·„ŸZ̧•^ÆA ©MàØŸ¨õc€\^®Íá•ݹY¤– ÆMj§è» µÛ~Ǽn‚¼ƒ/Õ7ÃWh~q!ܾ‰ÓÓexÝ…ÔqŽõl‘¡½{ô´É¦»ûúr}‘7%Ôô8眫†¦ýû+Ä>›B«síÇ£ ЮïÅ—Ãs:{m§ §¡'É%äá©÷Ðo9×l Á-/ånÂàµáŸ©dQ0DÙòi*š†ž øšÊŸ‡aùéú\îžÒÕ mƒ‘¡Ôa1Q©œS§Âøí8™ÙVo˜(:õdgL‰ùX2ÓQ:ÇšòaVç¥~›½Ì“òˆ‰]õ€ùïbI/ìaa¡Äw(ó,ö”l|CK%!CBÌn°œÎÅIuç,¬”µk}ì†ÕíȾŽaÝõq(c¹ lìø”\”­Æs̾it°ÓÖÉýñ²P\3xà"*_á¨@¢k\ù©x‰¿3‚xîã»F—£µ„¤±®y…ûÜåˆWO€#R—Å¿º£ÔÖÅÉ¢î¸ÿ û ƒ}ÅHgõåxXI0Ò{ûGRǾG††–'ª‘qØå”t¿,2KR$mÛõ·7Oãj4ÒMìûœk²w„Þ ì=‡œaí6­DžÈm;}ÃÇ\yŤk—rü¿Uø¤¡ÀDë÷òDÍ]>Ôèÿ®eƒBÅî=+cP„Ç`T4¹EÝ~<1ÔE±Ïl‡ïRW¢øÎh|õóý(ywdÃëÊI#/sЂ¼àuÝu)´ùsµÞ-[èkˆ$‡ÐªÑ‡Êñ(Zx¸Jˆ¨ -©ªJŶ%ÚépˆÈ¼y†·O<ôšwGAæ·]cÈü÷غ¶–œH¡žS…ªÙ†Ü‹ÁÇ¡„ugÿóC¬çËPhwî/dÜê;.©¸ i«ñOÑQ’ý‰o¿úâ&V¸åÙ!¶?“\æ6ì½']}òAcï»(ãg×8ørÖ²ø¾Ñ|}äüt6Ôâ_$¤iÜ÷G/do_‡toÚÉfߥ26öZþý®ÎYb½OX@ü Va®P´rÔÃ={~$ÓÿwØJ¯ðùê?oûî8Å_ïl‹>Þ‡ÊCvŠvƒþÿãCŠs6rO Þy»¶âÑÝ|«t“ls š\ÜíiY ¹`¸=ìýyhc䥃¶§~,BŒìÐÎ`ü°¬q:2?ålpMC—¿,.9ô<™¾Ì7X÷?>œ³<šÀƒ2!1Þ]¡—/C}R=a„ý¼bœ„6ŒæïøÝÏ€ñǾõö0éï„Il»~VÕ¦ÿ&×æÌƒ¼…^‰ÃéçCuаxó€GÌü/X&xÒd*W +ŸõÝ=´RaÍu.e*úl¨5}Š“ÙÕ³f5ëyòÓ°s4¾Ð‡±þ} e,þ"„„,Ñ#쇑(°Üf¸2IÄSïšûã¾ö£aO‘,¬ø-H Åé÷W¼s*J&Û¿¾idÔÏëfy!­pŽ7If!Ò½y;vMÞé_l<û±ï>Ì!Óa…Ádn›~—ü,YƒšZ/Œ!ûù†¶W|È©wRñ"}!r³&ozÉw#Ϙ~üÅ_ÖÈwadÈÑêòûŸâr©\A¡Ó‡:ÌùQðUr[Â[Ž~ã)#pEWf•ÕmPâ´BØ÷ö”ŠêÉ õ>‡Ò~ᥞO¡Læ%ƒÜì(Ûãv¨Ã—好6’'†P¡xý÷Œ*†” _íGåÔò©¨Fú uXPÃýA¼¶Ûb¼íd#jŽFךQ3¡6³x*¡ÆCÔq«œ³ñÑEÝ ’ÞHýߨ§<,¸CXˆú!!k´Ïï£!!à íÇÚhô’Œ±ø#áò!PÌQ@“ú©ÒÑËÝhöñ~KZTšW]{÷öD Zü9³¯ŽâZnºÆ Ê¡µb&ÍÁßhäÊDÏ?ˆý ú¢¬ÙPlä RÎ|Rí L‡‘Óß#5‹âáÇÑÜCëäBËh¿Xî´«æ×C¶‹»ºyzœÇ4¾û2&hìêo_Å0¿ÚÞ•þ´x½«¿±j“§ÙÿéŸüØÞûFIqTÅÇõÕÓ Ë[yð@…•¤¿|µÔ$…Š,¬eGº¾.¬ß’w²(é zç]òGã9¨325?èѰëÏ<Å?Ê ¡ÁßÃk[ÔÕ7Ê6²Ccùù4µhZh¢bØÚ6͇&÷dvz¥ÐÄÒ¼QK·Ë…o-ëŽC“q½h¾¼%4•„ºó«¾€f—7ÕמäB «¤ë+eOh©-ïºF¿Ë‡A¢>Ù8¡ÍÙîX‹Ü.ÚTT̽¶…vç1ÛŇ¡ã[» ÇKè}rüö$t ÙDÐï燞Kg†f| w6XŸ6kú-.–_<‘vmÞù”‡a0ÌFêå1f®áp÷®‚‘niyÅC06ýPQÍØ&•9‚£Ëaúë®Ómù0§t?k¢+ö·äïHÃâe÷Õì°üYžkb'V¿[мQЃ Ö“Šž°õ2â‰Ã5ø«ó¸pÍŒ ãNᮯJëä± ‰çC¡ôl}$½{«êª‚ ’‡'/Ö¼wEÊž¡X³m¤1=Ì0×2´SGRúw}‹Þªeœ¸Jgûd!“}`µÎ«zd5™¦ðŸG6 |ã’4oTøÂo¶!·Ø°þ›rwä¿~£†ÏùϯeÃz L)Ùüõ<€‚|iHQØü¥·ð¤-ŠbÛû“,Z(îÚLÄú"%+J]ÿ<»ŒÒ/Âϼºü eÒÉîöÌ¢<©êiUÎrTHo( ñDEg#ÚÚ}6¨|˜âá;ߨú0<}üÔªÏ{tÉÑh#Þù]r"©µØ"µò7PûµœÇéê/¨‹Í‰ÉÂPzùs"êN¬G anüÔ«B4¾]x¶ý’1š¸Ö6d<õB³CïÊcïΣ¹ß³×~£Å­xùQA ´¼+o}Ñ*hÇyÓέ3¿l~†6SÖ\dv„h§t;î®ÚGlܧCG®Ó„çf¹ñhaò‡È2WtºÿgiþÉ<~YZ£÷};žÖyŸø£åŠ­3³ÛÌB’Y5Qþ©yèK;Ãq–it¯ðtÁÉ3nHÊ+ÍÔ‰Ü=—c= öëy‚øüúƒ~BÄ‹ò†ØCtn¼bÓÅ·ðA½úÍó½~¥­óoJö¸0ƒÏ  ¯ÅÙ4»øýWŸÉnñÖæB\Þr0ÓYQH•¡¦¬Ù«»˜&1û Y—Ö ŽÔA¶zþ]Þ6ÈNw9¤é_ 9¢¤*~/CNTâÓ™o|³UÍ^6³{^X‰Ýš}õ¯cì¼^ª{ïL93Ž^ÿOÓK>Á»¿ó¢µðǵ½z·Ãà‡äûn/„ ä@ákßW(ŽAc~Q(;d6]‘ãåyé}qzPa8(¥è!¿•uB…>@Õ‰ç—<é⡦ññ»•¥¨w ”‘3‡F¡S¹96ÐÔ!AÿAñ´”Ÿ÷äÖ€?´Ì•i¹²ÐŦ¿ÁÏÝãé߆BßBΔ@&°žly?t †Øø£üÔv}ÕÓ™.· ÆôåO Š_†‰·‘<¯VÁ4eÉȼ‹ÌÆùF4¶eÃ|yÍÚ]ú‹°èæüäxÀ2ž(Uˆ‚U#Ž(N*X÷8s¨úÅwØŒïô.§ ‚&ãgÕ3á_¢es ª• øôÈ Qzk®Rƒ’Àý&Ü·ž®= kŒd+º‚UIH©ß]}Æ©ÇF]R~DÚVuâï‘Þÿäw2?dÄž$/þÈÌðÑx,[Y¶•–î6v ;ר‘:rËu-w¸#Ï“+ÒÖL$È¿/¡´*_&9ûÛ…¤“%Ò?J ÈòU^«Ã(®4á¬ÿ;¥x2m¤äž ŒÚåSÏ~ö¡œð­ž ÉqT84>£áMŠŠÇÅÆW¹¯ r¦¥é÷»É¨véÒGwk„7O~JøE æŠXçå#‚¨ýþ –3“ꚦS«=wG=ö%ãÁ’h@оlmMh8OÑfè߉Æí|áâsѤ„xÝ»ŠÍ*iMÊ7ÄÑ|c¥ˆØ~-¦¾ø)ΣՉÎÜÌþh÷/tõ}/Úì ±y(ƒvå-|2WÑá¾ Kø\1=ßÌ]tþ$:=ñúrIì7úÞëa{pÄQD±W^8Ù5ÊôÕÍœ`m’tëŠ+%Ô{~;Ï–åIKó•6ªð#ÆÔüÙä0çl'AŽÇGw¡»}ÙUëÕÎémRU›,÷¾·ô|™ë€ø/JV”îU«¾¿ŒnH¾ŠGf‹Ÿúϼ犽:\á£Æ”ðó¼VÚ©/ÔѬVŠûAS<ë†3ºÐÆynŒÄ:(:Þud$AW0íпQè5|õouAúß*R´n„ªN¥‡=e0ÃOçu¡†$¿¿sË~C~ÛÝŸÃü! ¢‘³0Üë6m¤Ì #“ãs½E0æñŸÍÂxtûñhk˜Xcn;ÚSgõŒ¨”¿Àô”ÁN·1ÌF¥”%Ý…ye§_.>° Ÿ%œ”è‹"…L”Nî°Dûo8ÍÀ–&Róœ`¹nSIõ–¬4¿í£ìó‚5 ¡ëÞó°n›p¿;É6†éžÈÍŒÁÖç‚õdÔ°á1†?.ÿߡÔÞ1HH±¼Õè‰Dê~G<–;ø®š¨Çà9$é9kÐ_‰¤ͼ]JêHNû¸5;()Ö†åt7,‘ê/ßmPØåCÎð=Ä>H'ºróF8 Ò›ýaü©‡ ~§òÍ"cQLOûøId&bõؘ@–‹ÿBû¼. ±.üþ‹ìù9zÜâÈùà[ʼn ä>rJ½(ô ò s×–<=ÿÿñáý O|Q`¼§Mþìò¡àNÖ Ð}§3ÉPäÀ-ÁRg½˜^báb)1³QI#(AÙ‡‹Ÿ2Qòùûh‘í#xðxñöçÅ(säÝ'‚]”WŒ%ó²@ùC»{^¡BåòMrFTPË®œC%1óŸž2T>@è!…*ÇŽôUŸAÕ¶9§ÊårT/®Ìfà¶@ä0ß­:„š.',ÿ£ö‰ øiµuz„©¤XÌð0{±ñ­£ÿPâÕ‰K©¨OºÕ#)fZ´3 ÐðéIw¦24~`3d·…GÌçH™PMš$ '¡Ù¹§A4³jhþBmÔyX-®Ü¯Ð7EËsöé ¼ŒhåéGÓÑú•Rƒ Ú¸ßNå|ˆv$ ‘]îhލ²X¨o¿Wâ8~è>*_l½ñf ™Ÿ2ù Іeñ²Ù—õéPyÆl(çö.1*Ó2|‚ÜúЦüÓLÁUzÂ˰Ò>[LÈ‹¤B²¥¢‘Óeˆ{:öºÙáÓæ8H íÝö‡ŒìÍâzÎèµW‡X€§þÁW[K%óÝ¿¯Ïi·V )ûÏç—É,îú.áo?/)È¢ì y&9Ã:ì·)W¡€Ä<>ÈÛ Š*Œ ¾^Ù;Œú–u ô"Ÿ<ø_>¬ ÖÔ,i‡_T. ïæÖJ5U¢õÉÿñáéï·†÷æ}—Ñ+@#û»®ãŸC¡)ø–„¨k4ox˜mQC«{ýõCg÷ú–>œdv+ µïÓÐ1‘0k|Iºâ^8”DäBÏû«‡þLJcUftÿ¾Á ßQ|éÿdïAì)Õb^Ú¯h@ £~®ï[­ŸÃ¸ú¡+ÉtA0)fyì) L;ùöúÛ$ÃlkHHmÌ—9 𼕅E™˜¸­Ù\Xʪx¹ÿª¬8*öžúklÌíÄBl°>}SäûCØœÌb³>ÿ¶·Ãý;(៲Ï13Ï $2n§e‹@"«êè\$!ª“Â/‹¸/ëI®¯$"Ù•z\œÚH¡»¤YZQŠTbŠŒ1Æ‹HÃu4¦—!iù²ðLº#ÒEH\ÉjDú´ÄS¢öÿaxh"˜`™à‚¾[º12—~åú(ðY/)x<͵Gvt«ÍúÒ„œ¬Û?%VSkš¢}Éøò|w:²¿˜ùìæVËÿr–äÚ % ÔîsÎ0¢ ÷Ab¯i¾Çñª3Å E‡â€…2JH] Êýp ¥\›Ý7¿¡´õYËpZc”ùÜþè'ÞCÙV†ã~›(/u%¤žNJN 軩¢b8×ÉŸw*PùMîlÊ4ª ¨Æµ¦ñ ÆË „å- ˆþ©˨9m¦.;ΉÚLg|úPçÒoW«üPÔíÏb½Î=zg®låoßFýÇéâhhÿòÙöX:­ÒGíøã‘TY^Øçhú¤½4ÉyÍušcÞ÷Ô¢…íYàW7´ô-Y=’)ŠVåÏh–^š 0qϵH´KW¶á7Äs#š‹d¤„(½zõžÚç>¤XfœŸ½UÃsR”‰õCðýÑõŸ•Fmcl}Qô5 ¤tqÿ£.‚äsÅËÚ½z_XÈßý¬¾‚ÍÆòˆ]W×£˜;Ÿ³}ZîuÃÇÉ+É·I!jé¶îmoøÞ7~Ïnz¯”ÏSOSH×€ †4(ˆ£ºßÞôJ—ÃôŒœyàWy ½W+ìf'›BÞ¹]Ÿ$¸,s7 ê©ÃdÎ¥.@c¸¥ÀÌ#hÈõöŠßШD¿¬©§þ/ (  1åço!ŸphÌ}wd–D»¦“ز  ‰+]šMû;4¥L>˜¦¬‡æÓ·oñ9Q@‹RÐLü™ h%/¬fx­}‰OKÊNA[aØIËeøSØ“–ù‹Ú+b¤å8¡“†BëžÆgèRÜî=cÝ^KäÜ~BO¼§÷pôQS³%s+A¿q€ùÐ):Ð8âE'÷ÏF9vÓ„¡s–- >­üDur¯ÔÛM☨1¬•IW‚i®û´ktZ0ë÷žˆ˜"æ¿Ô '¢É}RU{AXõ;? ~VUįÜÛÿ ÖýÕm7tÃæ¢Ü%gSB؉ª:‘Ó%ˆvRKËÝÔHÄsŠïÏ=$a=¶ððÎ’rç5;å ¹”‡°–ëK¤<Ñz™Ñ] ©ËrjQ,#­å¼×’;ÒäøçU&#Ëà[›WqùÌÍsöTÈ\h7ÅÕŒ¬Uõä¥]ÈÁ-øòFØ+䊫i\xÖŽ¼î§ _¾¡ð>·ª»ÇõQd^ñWô›B§µù¸q %=Î?Gi¥Ï¿SPVZÕý Êuqô†ºÛ ‚Ÿë•É´(Të­¼»ëgÊÔŽ5裪Ñù˹äõ¨žïj¨¨h¦G þòj‘ÚÞ®¬¶EíÇ9}C/QWEþA!ó6êQœ^-fCý‘K¢¶W¡a†¿…Uƒ+_.ºòç¥9š¸UÿèrE3åë"]ÙÐüÉ$ùãk…háo䬠€–!y_'Ž£U†¢”¥ÕZ®‰\º†¶²ö¶Qq‘h*áÆùJ˜Ã,¢4;Ð1ûýÉßEGð˜ÏG›æè_è|QíÆÛçÞ袟W}tð”ý>‰—‘ö(º/†îË$aÊ-ã¹Ã½ƒÌvÄ@Á\lâŒ×Û]ß³:)8• 'ÎH;½ƒ/î´Îêðin¹|úf,İßÒzžÑÏ0ëÿüOÝam„DrSîÐ…C¸Þû‡É†W!œ5ìո¿züç*ãL¶qºÖâdž{ïjd©»BzúÀ·}Yù9ÖX'²&—h¶†l—}Úº‘Ýš@0úx÷|°Xàô…œ,ùç1‹ócíõØsÿçš}ÿP—ÅæH{§´×·Ñý¼­Fó¥¥óh@ñ…}%ˤþðƒ¯cÁéõc(a`Îuf‘…R ÅKSÿ ŒTÝ)0ò”{¯ê5·ÁÏ…wÿxèàW‹|ÁÕg‘P¹ÉfDrj̉ž  @]¾¨XäÑ·ÐPÅÓ{æ3 4]f›`¶K– ý}ÊÐöာÛnNè—`¢Ü„nš²(‹­dèݱy7ý¨È5â):að¡ ÛaV®,Í—œé„Ñõµe9a˜0è­NqŒƒ©\2ºÈá{0kõ ö·Á|ø7vÊÛá°¨!¦ª!ÊËœ\lû.xêˆaÂð3CX×ï ŽQ‚M®G\ˆ`»Çžäê'{øçÉc{zU ¹µùõ’?"ÑëDV?$áÞöS`¿…ûZÕ×H¨Š‘¬öË…wÉ\HÉå‘w8g—óÊòÂöË"헴̰þ§H1Gæ q/2 í'Þ¿Z‡LK$ 7~"Kë]–Ûß‘mòþض2/rÚÙl¸{”"˾€~Iäë$k©zê/Mlý[@Á…^:ÃïS(bÚy‰‘QÅFÄ­ËÇ¡dIÐ]Om=”!gq}[…²³³¿>¥¢üößSýßêPQôNhgš4*?8•}¬ØÕt>QÔ¯Ñ x…‘'‡¢æ°V}mŠ7j¿Ø¬k?[‰ºH“8~õ¨îI?αGýÍú\ÑpæÑ¿÷h÷lY´¢I}âOû ŸÐlLdßFôZ0ng…Tìúñ£I™h¹µÁsnm¨I4½XÈÐ62FàZ?Ú[%.Â÷訤+ûEò-3IR·&Bç˜ñCÁWÜÐ7,Ì5QcåÍÂ~d—úq—@0Öx oœ,€zÕÀ““ œ£éöQŽjøz3…“÷¬™V÷šdï†Û¶*È4>]{Ò]ž”Ðÿ†Tòð¢-º½ï/«Œ_(J÷îû׳%Ù!VÐ_4wè¤:.|õƒB­á5—ÁÏNº×L¯W .²HÑpJšõø nˆ«AÛŸ¤’IÐQ£Í°m°݇weõûCïH™ûyŸZèïñQ©²'ƒA¢¾¾Öç/`°Îé¨\% ™ÆöÔìÀP¦‡ä™Ir]Zq=Ã}½¯Í;]`dq7ù3‚1ë+QÓ“aüç † ²»9Ñ@Ÿð`4LÅá¢õ³˜Qy•Ë~¦f×ê Óˆ¾Ãüù¨Ó7<…s ’ªý`Ñ)û«,©3,i6Fm<…eú;wLÙÀòT»Õ¹¯`e8z¤…{¬‘ì;?óÆÖê=¾„ƒ_B?{DÁÖõn'rØq!NíÝÿî¢Cü$èºÆð8Û ‰ˆ§z¥âXÕ!îÝ6’§}ï¼÷IitmD]E²â2•—oבâ3IÞúµHõõj^@šrŽ†Ë„ªH»ýL8u„éEÄr Í!ƒSÞëf{d ±½?ŠLš÷¯qk ‹îÕû‹_„‘µ{1Èÿ²GtÄìB΋]9?R[«—ðoç!ä=@U·e™ÿÿñ¡’i˜) ŒLÇXªïò!5Gb” • °;‡œEÑ׿­(zýóÚ?‡_(V õ¦ÈŒ%Xÿ´²8 d¸^ï£CxÐæXÿâÊXí$X“d lÍ«kû»Q^%·t§Rf—ù$Ž¡"§˜PÔÂyTâ*3e¿Û‹Êt*ÊCQEÇH߬U ×Ã…ÅQ=1ó)ý§Sˆì^ ù¤‡QsmÀ¨l˜µïk¾Ê£#B]rcQ]»\<¬éô÷ä¿eÔã¾2 léˆú”ìO ‰&ÑÀü­/e.ÆÓœ}õþ³v_©¾‹G"¶;ÄéÑT…=ДÙÍæ?þñœ͇¤ÍÚ•£Ñ¢W-ÇñZN=y1à‘ŒÖÔU4Ѿh£îvô…Ú†ž=D[ƒöD/çû£Cà”»q¦4ÞN¦N;r+çsBm­ Ùê›ÖJÔ X >È*U†;cÝoà‡®Ø‰WóÒËË6Áí*éÅ_Nf+‚4‹ á.Hê~=1è©q’u O|B¬ÕKµwÝOöx°*~ñ9|>õJèJ¹ê®õrvî³ÃW‘ÃCé!Oàïó7’SÍÆ­G=öæÇrU„Ì®*™üÁRȉ»ôüNfäÞ_ͶÿÌe”€aþ'>IB陣³úþLJ¬Ë2à×.þ;)²P©ÅUNoò?>«4þSêßuç:®†F•mSyh*¼³šM¹›OƒI(ì’ 5õYvQ>üa XHaT€öñ¦Œ¤aèäÉŒ*9 ]5×–ÎBO†ô­Ñ¦ÿñaÿã kE¤èθ CNçrL~ÒÀpMú¡Ñµ15ù•'ýcWÿó“qÏ¿ÙÍ¡ú·òÂԼĵ«Kû`ö¨L åy˜÷¥c¨Ùt……qþëé8Xº@l\—Õ +tÂæNøVkÕ 8âòa=Š7Ÿ¨ 6_`‡š8ØŽzjœð^þ6N|îWÛF‚”Õ*Öy$’M¶H¿Q†ÄSÝÒ'p_0 óy»e$³À¥MÊÏHqÀ`?ukRQ;{ýEò[]7.å!-GiUÊ:Ò½/ú·ê½Œô¹ŒÇîÚ#ÃØã¥˜ÒIdRŠrmx‹ÌÉ'>„äÝCV“0²/c¿‘q½¼S‚ 9&¬&»>!×·œk'›‘çÍå™”ÌBäÓmÓ~·üÇ÷in¦¢@îkç§$(èœ$£¬IŒÂNQÔ]P4B]ÿQ*'Š/ôM;-‚R"•Â#Íô(-¨xƒôU"Ê„j•NœAÙš?³#Ovu+qMjNÑÊoôåGÅèë#±þ ¨\ÏsN[NÕ4&˜ây…P#ÝÊA_çâ`SÆõÅZÔ\ØI`*çGmFyÉBUÔ¹UÑj±>‚º[§ÌÍ–éP/Œ¸É{M ˜ù/¦µ¢a*uvŽ?ŸcWS}€&:£`:Œf:?K¯ yö¨ãðtZ Ù_ó›aG+q®Ñjç´Ìð*Q>‹¶ÌVïþaoTåµý#HIww‡tÇZˆ4Hƒ !( * "*6"بH( H‚¨H©ˆ€´t H—twüàW¾3ï3ÏüþÚ3çìsÃÜŸûº®½ö¾y0,"tŒàMDÏ’UK‰ë¸×è·²8M’­%nz”Aÿ…È‹¬½Ûçnyâ‚@ÖM³¡Õ/!c36‰½ð¤yÌÞ=æÝ·¥w³N ^ÞÔw~€¦L)ךLJÁ¥`)‡s„ðÚ¯Yåè-Nxåaï(§1åRûûvæ}¸9h…ð…£Õ›çzùöó É67OCéi>–éˆÿø‘þøLîQTÛRÅÑ¿ÞÞ§íöÔÝj幃÷øCíÛ\e±cPGHñìsÉW¨Ó¿3ûñ,ÔrŽ¥^¬ƒºçûÈॠÔ?WKø+’”j6w° ¥ªÄl(¬#uÞ3’JO¤M _¸ÿøÒS|ž`ò•E†-‘ÇÜzÈd®¬êc²Œ–y×RŽlñ4ÖŸLSì³£±Õ~ä¶ÝTé·DÞ~Mÿƒ¯Û‘ßÕ­hzÊ–8Žé% PžgS£ ŠÄ†Mu²)oédG¹¤»J²w“3L@©žK'kÃPvwa]”8/ʽ_µú:„ 6³?Ùm*PqþìMi#TnXU"5ÚÒA:5¹Û©_P#èž™ËaDÞ"˜=s µæh?ŸnÏDíËš^’Âñj4LwÍfš.Üâ*ÿZó{<à´¯¤ZÒÍÔ-ïÓÞÞ‡æ "êÎe EhŠK³+Z¾ÝŸp]ù:Z5ù¬¦˜»¡ ›n¾MÄa´½àøD~M δsçD ýcy~ÑÖ;è`Þ\"òà :i<£ÜsMÛE¯j=öC—f‡Å²'4èæu2ÅÙÍ …¤Ig_ïG"u%gjKèdhÙ{.Úi‡ü¥$ÈÝØ7ñø 2¿¼”‡xc#í½Úê;¼ýãìwQGòÉïž…ˆÜ«ÖF ;ã¿yÿ>Ÿ\Ùˆ‰¹ð^ó‚Fìѽ[þöŽûê©8È|÷ÅÆTÍfû|þn‘ZWÈ¢J13ø˜¹s_ÈZš¯š{ÔÙ>Q›ÎûþÇû¿&ÄÉ««ü‹§y ò¥ëÜ¥¶û ÒhFÞܹï01ݰú ?|:Ô¯lÂÑŠØ¡ÍPØ"Zmhøßþ‹:ÒåÕûFùö‹¡²úê5+QO¨æ²$ êZŸÖÝZ!­POARÌL ½bY\ãr;¼·fï»$5׸Ãu§ «ØÕwÐý%二ôžë‘}J< ý—˜˜ ù{`0åk ¸j /óILhÝ…QwuCÏ\ß ê6žY€©Ó7ÕÂJa†iv”½`7Ì.4{×Ï[ÁÂnÿ¼´µ;°Äw`Ò>+V8‡ÒgøÁZ¢ÛUòUØ”ÿû›µC 6~ðÒÐ#áùÀóê*êH´þ°‘t”‰?”¦ 7È"i"ÉGÏôA$Ÿ&¨+ØUÊû¯é/ ÍyÛ½âpélöÄí¹™‹ {Àî]•26xü&-oCæôbUÿ´ýÈš7ø7]A98/2÷ïfA®¥œöoï¬7Ý:»ÃÈ ù£¼k+?AÁLª¹îÎw(B’:LÒÀ†b¡'÷饢¤›WÒ'KB”~ëØre³|ÈòAùb†Î›¿ÃPqºNû»Î(ª˜_äZÞ0Du–f_ùÊã8Ù¸@Ôj°ÿ)F_…ÚWK-¿) ŽôqÇÞý›¨»¾gê}GêÏ ŸÛʧ†£»F—Š^£q¯;_eæ<ÐiªB¥‡æ„1‡Y=ÑBŠ(]P¾-ý¸_·ÈýD«V.ÕÛ³žhãL62úPîq[óö A»Þ`1'óR<4.oòþ¸/:Éùj|ÑŒÅÃ_~þF€ifï&ë·ütÑ'›á/O‘ë[fÑr+,XÞ+¿ìîÕ•UYT_îCÑíÄcÝŽð=ó:³XÔ9È© qßïßáâÓß¶µá¾}ûÇÓ?¾þéã?~37*ªì\ÿ%†]çµNl×ÎU¤Îí\§-!oã|mæ íQÊîNeèX¾Ñp²­ºn̵v’¹ýé]‰H˜SЂ¾Óσlû| ¯œ‘ÐïçAè—±Z~ñÉú{ì4^I€å!Žu*:òý· 8ìS¢†‡KÙ_K}`tD5òUg9Œßnó½ðÑ&-&¢í`êåb>•0˜ŽyÍ›6 3O?¨|o±uï·>R sª¾ÆÝͦ0O–u*Ü/æç“Ç^Š€EÒÜ ·ÏU°³>×ÂrŠRÓ¤5¬¾ØòÓnX ¿tÅ1ú?ÞÍ‹>:5vø+®ÝYë·ÁÝÆµ,^Üç¸àÓ+Q*’zo¼ò|­äæBûR§Ï …y?“–ÉV>ôŠºJ²±ˆ4I‡|ÃÈ‘vãÒ~ÂH¯L97f¥ˆ §EjÄ—J‘ñƒÃèe­PdfÿëCÜÚ,Y÷{8$Ù®¬fgF¾Cû{ÑoyjKÙ¶T‹Gyx–k9ÈÿˇIã²ÑB\È?8AD'¤Ž‚&³ÊñÆ´(T´ÖdãoŒ"{;^ÿ뎢×_ÈžAÈg:±‹ˆ¬ONkkÛà>’# S¾¨ýš©¯ZLuäOë:Dð¡®Çú‡ñè'këJ0'‰úTÊŠîhà"âR?| Ë_f‹¢±±Ò“¬3»ÐdhÏ;쿊¦±¤j‘•[¹P½“;÷Ùíð÷OWþéÕÿëSÿq÷Ïwþãîÿå0qI5„q5uÇצ[*2 FìpüOÿýÞ?Ýúæ)²Î9Ú÷~ô¼´dÜ>'bß›?ÿåCbï¡êPš²ÑÀ¢Dåz›/ÕE?ü[“m’– c»ã[ëž×þ$†Æsž…™¦Ð4g lÎðZl¾öcحٗ$¢: Íêë¡)†PhŸÒ£7f…?¥µ›†ƒÊÿåÃß”—“‹ gsäb‹Kô騖)™9@¢­bp§* ²H6Ü“Jˆé: Oi¼ÈQ…ÑÀfÊ£!0ÞPqZŸâ?úyŸåÕ"{´ès%L…Öù1°xDTNÊ–”HuŒ‰Â Üe“ª€5¸Ä°ÿè%Ø8i;~†ô\yýõ>‹ ˜E˜þB¢Oä÷yxËØ'Î×úæ[$• xݲBŒä»^¶ÛöçàžÉ[¿~ë8!å\$#mÕ¤ąޛ†´qæ”9~"]þ%«fÕH?"±þ)hej¼<'‘)’“ÿÖ–?c‘XŒýÁ¢Š¬Ã¾ldx ûéÞ·aÈùìûóÅËÕÈ}êÎcóÔÓÈ+kêË¡yùŒmL…¨?^âÜkáÔ•tÜxZ¼MÒåŽ(êÝg&N½â_ZæS$Qr,¿»ŽÀ¥ævGe¤D™àTª c”-~RyòÑw”;_FãØˆ %š«/ÉP)±îØíÇ!¨Ò)Íö<óªÛ¿¸~íá^ÔlRÏ(ÿXƒ¸É¬I”„Z‹ïØcEQ›^ùDÇêsÜüÓÀ\Y u™™îŠU( ^ñÓ·JÓ+h`3ÜÊY|HïöœIAãÞÊÚ%_<Ðåä¬fë Õ•ÐBl™A|õ5Zž*ï»,WV ZŒÈO¡]{Ó£ží:~¼uÙù>´ë° øÃ»=i+ÔlfòPÂîô§ì=õH–8ñÙ´Ðz—[êsmþÂ7­ÙCjrSð¹þ·ÅÜ3.Èx)Iþ|ÞY_W”&‚äù‘11jHÒ‹ŸÖ§ù §Åßs×…ø‚abÇ_²»©ºð àeù=âÛ"O!ÚâI×ËɰyÎíÿ­Þ ò®-­ÈA¡Ì™Ò\¶ (™z-˜pœ*îåÇ>°6€ªD;u“*?øUõáÖÔW[:]­r¤6jV7ÜÔÜÛö}?ýv Bí·CŸ‹ŒžBmÿË&¯«¡vÖæì©½¡ŽýEœ]BÔùNN.µC=wªSµÍáM]—¡aˆqö– @cǵs1ö!ÐôkwÙ‹ôÓÐü-D”R ¥$MÞwƒ`‹ÇÜ½× Ä aHùºtàöó¢â³ïk@Ç©U®K=Uð'Å. ´ÆºUrV¹@7èpõP:ü§Ûš‡Þ•O¼…¾ÈÃf4ca@ùý”­ 1Ï>×;ùF÷RÁ¨~ñT-±<Œ{DöSçÃK}ÿž9˜Î #º,³gn‡EjÁü…²œ‘œ°ÇÉ6FMË#Áîõß³aíâp[àZlžNq`YÃ]¾Y/½o#Ñá¡BQ»»H|¨úëQ”AÒÃßÎÝ"lAòÛ´‘ŠÒ·â7u®mꤶò]é­FZß/”Œ-IHW6‘2; úo–—°™vÛ›„uA²8Î:±ÇÈfÒñIê²,rdÆÑ„#÷ig ½CŽÈûjFqôZòÓÝÈâýºŽ1–GãiPÈÆÕ¨îZ6ŠÈ?=Ê|¥ÅŒühJö D2|…Êi”ò‰îŸðæG™Š5ÚœÇË(w!ÊóW“&*ð‹·²ïªBÅJ²ð­¿CùÃË(7I%Tÿ¹÷4â÷íŸEX¼à•·å?{ä¹¹= öqË={óP‡+Œ‹*? uÇ;´ÊÊo¡~Ék·èäÐð¹FZE` ÛGo9J<`æ¥6¡ÿÍ͸î*ŸCó7»ô25ƒÌÛ2®«!;ùðs}K­aô›-?œJËs5²Ù‰(Û^AVÊÝý\ß +_0ð†cñVn\xøËN²åË)«¼¬!ÛÀä8ÉùЮ¡  ¾òr+GíC¯üïùp¦9«îvýÐ1¦& 7ކª2¾ƒâã: \¶ð³Â±em‹ƒÒ.o3™Ï@yŒáŠÒs¨ìå,·£¨ÜPÈìÔ†Ú³œjtýÿåÃV»êò3ÐLÙÐ×ÍÒ­WlÏå]ÜẓÙKà¶Ö-è~ÐÇÒ £½&ì&¬ +Ð0Å$úa Þ1u¤2‡áSã{ìaT\-4÷Ú/ÿ†ëÝ0¥{ŒõåU&˜èûlÝ!³EºÂü0_û†i¥ð,.0*p>¼ +Ò™—¨YS`íi3Ml Q3ݾå‚s*`mq f u!Ñ€íëÕ_$~‘ZNr 4¤ÃNÓ¡É{É+²^"åuþ ®@š#ÝÁVÏ‘Nw-oúÐ3¤Ÿ¯¹_öéæöùáÀà@id_ô0úñY“öý¸éð 9(¦Ú¾ÏŸC®‡¾…[Ü&^Tf¦CþØß‹,Ñ(˜8:2䯆Â+áŽÏQì±qœaó-”ôM¯ºÌ(+M%<‹²ù´®û¦%Q¾b”}LÛ×aùÏÏlT9êó¾ô ªKd×5ýŽpЛôÖž<Ôj&|VµÝ䦼(QGÍt}^ð7êÑðïáŸ-G טõ}ShD^á´¦[€&‚Ï4bøÑ”^Lì´Mš‹ç|¿d=‹.óC9GÐòãC¿ßNjњÿáÀa¢Y´ÉlÝ;ÌG޽3¦?›j¢½Å¦tÜwtðd²i§F§l]¾Aw]t´Ð$ÈKÁí±½ºb¨VGÆšºÙ<‰LM a®c?µ•ˆ4üê¶‘ý¾t Šè 5¤ ꨙeX+äx'j3ØmóQ¼/ŸÎ/ ÔÌÀ‡ ÆÃ«²q^Ú|Þ%]’Œfv=ØÑÇü¾×ôÏ—³Ü_ºfÏ’%{ÂÏUO½¡Q¨¡½šãJÈ Iº{Ø*"¡Åx"Ç+VÚ´ØÈù$BG"aÃòÂ;è’:Ãï’' Ý wéÃnlé7¥")Óôæe§é‡>.âA÷EEèsœ:ïëà}ét”j‚)ÐonÙDÇ÷­Ú´B˜aðg³§a ƒÜ±+ð7ÁûëaÂAip󴡇±Ì½¬«J0áI!I5x¦Ø*x€ijÃRª Ÿ0½îòêó{z˜é9•{56f¿dj:˜Â\L·Õ[V˜ý#òó£X(jb%_ð…%R¾´ÅTXö/ŸŒ–€U.Æ!ÅsX[¢.Èa „´Fuck$¸×Z¨y wÅ>¢’ßò•UBŸÇ7hp7«>m¼–=ß=º¯ã±4’Jœ|ë8,ƒdˉï]Â=]GÓ:~^BÊǧcI‘†âBõrÒ~8}iÞéfˆo‹ WÔ® ãóôM0ƒL &k-»Ÿ!Ëù´§q±>ÈF%U/u½ÙSÝ‚*Ó°ü]È_äfŸÕ™A^ê;¿~‹Ç!Ÿòˆ”u(%ò‡úÄz# Ò3ëúA¡¨õl2(Bò ×·Í E ½I‚Ïr¢ØUÒ_Ž7î£xÕÃk¡~gPR—"°}'J±Q‰2+× é3Ê>|wsdUåÙJ A…tåÚL¡«¨D&\«­ ŠÊŒ›¿ž6.¢ çhþ$‹#ª:vüLï6Gµ®Õ{Ì?§Q£NMϻΠQé*¥lî#hŠzö¨}s—E› îß”^«æº’=] %¨G$ûf&Fõ†ß»h@…²jn ‡Ñð¡s—Òt4ÚcØýM‚f½7Цõ¬'²N Ùè룓6Sh>pþØoc5´Xë+™DËe‰&ÆF ´æýFí-€6‡ËnX¡mÖò+k9h'N}æË¯Shÿ]9&ß/wÊtO¡BÑ8ÉÏCO‘™šË•S;fxFNï‡ò’ÛŸJºÀwhTþÊ] Ù¼{ô¦õvô*ãƒÂbþïŸð®‚[²”Õ ’í™Å ù/@BþÓGµŸêàMÃÁ×ñ½÷à ù‡·,Jðfå·­–-$Öí.1„°-}}„Âûä7‡/|¤çèàrµ„Ì¡‡LÝ·ë‡î# r;ºµ“Þ þK{aû„åóÙ‡P4û4õ)ÀÏ‚GFÇHÆ¡ÔÃF[uþ”Z×î ЂŠ>2qÁI/¨šktïMÖÜ^j>ÂR uò\·XÔסþgà¯ðÌ~hT»)飨 Mù†¦*ž Ð"u‘œq©Z0<¼Í{Úìî°Ðû@Yã4iüQüñÚºjŸœ.Ý~>MÄI èåùÀUµÅ¿GÁ¬oÐiè/+  `6…Ae¡ôý"õ0T$Ä/îæƒ(˪qÀè=•õö­0Þ(Sv”»¦ÄMXwÁt ÕŸ³*̬ç=Tùhs¯24\`áPïE±9Xân9q`+,¯“Gº÷Âê`º=é,-¬o¾’ükÚœ·WpW¥Y.y°ÞXªFbsyÔÉ‘dpÑ.U,ûZÏÄÉ]¸çy^‘9Ñ^¤¼Áv¶ÂýR_‘T°z-ƒ´º‡7RU#“UŒdÒ‡-®?>… +ïmº!Óýðo‰-ïE›K;Ës Ùm(æÙû´6íú3Á²bÔäv­:I¼ÿòòP‡¾¥¹‰|ò,f‰µùÈÿðxåL$ ªüy:,¹…U4Pô̵«¡Ÿ½P¼”öV u'î%Ú¼ï$!€ÒÜþþ½q³(ó6\æóM”S9ÝÛ›ŒònÆõ»ín "Ù=) õUT5=F»;U9òú¿r¢ú½îŠ<¢DÔÜ$žyrµØææºØqwû¯˜$ÔÖþ®³l‹ÓOr,+jЍ«9Ð÷½/õú™.n6УßFZr-4{ÌóQ‘×Û8¦·rîÅŶŽ|4§ºAS¹ç+Z(ɼ3:u-ý-»(. Usɇ{ÙýhcCÄÃÛ¶‹|¢çœ{ЮêÞ5[zwÅ<Èdtˆ ™•]Q4×ô\a’´/ϯRÇŸÕu_…ÏüG¾:ÚGC†úÓŒ¥ã¶ª½§{á•0$Ëö^zW÷ßÖ®_;(½ÅÝIVÃízD‡³Ëöóó3ç–üà¥d^zç³nxåq(än”ÁμŒÍ×íÏ-®lù_¶W#~/áGqè-_ùp(9Ãí!×}ÊÃuöeÆAçzï±=¦ðk¿Í‰<¨¶¥¥”S‡šêûOî n?wÜ*ž£jýß6×VŸ€ÚÆÆ OD·F½š£¦P3¡áûþÔNœN9sêbXø†~EA}@}àl±34œWð1(w…Æs^¸bžAÓ‰Ö%Ö Yhv1Ü·7ZŽ»WxZoùG;¯¢WøE,himC <Ù ƒÏýge‰ü9å‘ÎgS m£·S [iÀÙ3cÿùkê¼^¯Bß‹‘Ý}·`@Xfx÷âMœŒº{¦ö! øUG\ö…‘õ‰NÖ‚hW¥Jø“߆êÇÀô£#b®tÌ0kMek{æ5}ôaÛ ‹f® ïNÃrÀ¾óÚ¶‰°ºZœþ0T6þ8x‹ªþÂ]\õê÷>C¢]á%k'Oâî¥7±ñ‹ZH²yà‰Mç ’‹Ú/< œC ÿéë,H5¦?±¢€´Êø¢y ‘îÓY­s¹H¿I^åêzsNʺ(#sn"Ùüed]xëI¶Ÿ9\x)—*ÇkÉ=ÞªÖy&Î^w8…|?¢“Íî>BÝýDm›þ(8uÇÿ—˜ — '¶0º÷lk$Qù(™ Ÿ¿ƒª°|<ë ŸÇVnL#tЇ¬›œì–Û9‘õÒ1ÈÖw‹°(ªü/zúÔSý ò~y˜úßó!#EѧûÛºËú‰ò0 EÛr$QCq¤™`á½-?}¬sšLJó_6W7„ò¯9ò†r‡ rþ+£ã#TKU³¸*ÇoÝ’äC'ÿˇշ‡…2 iä‚ÔÛ‚MhÝwòŒîÆÒ×2¥¤ˆ\ ÛY1ã†4ôò[y¼â'€~E œ]úƒîÝŠ3´?a8á2{_£ŒLŠŽˆmý¿ÇÝÝMÈÊ·ü³ÑBãQ˜þÜpýÑ’Ì>Öâ¼h~æ_֜ѿó+Úü»Üla…~Ïò€nX»¡ÿ®%ü+l2ËY,ÈT5FÒË6èɼ¾èýÀ Rú¾9ð=ôÒ˜Ý~Úíùé”&ZeO!}K~d/2&’H©!óMñ§œMóÈúôe„ÝMmd_õŒŽl@®ö +‘Vwä~•eÖòùƒúSQ0¢¼{üï-žª»dÛ„b$–-=\Q2`jÚP\ ¥Ëþú|:e‹ói/CùºÂƒÖ©Ô[þõ]Æš^ ªœwÛ÷ìý0ªc„쎄£Ê…m¾Ó¨Õúù´bÃ+Ô¾½‹üûêè*Äg¢Ÿª@ ð#[ÐÍ­|(Z³YÚ…&òý$iwËÐTw­ 9šÍ-ëyþ_G‹§Ã|&O¼ÑrjM¿=¶­½6ž(1^C[ФÓý™½x°vmp]~í\TÞ«‚½‹‡I²œÑYƒ•”îÁ(ºœ8Ë"Ö€×?Ð9¨0š¡J§{³Aä ²Ó[´Æ¤Â¬ûª•øðô–û“櫨 …/­|¯ñ@¾¤þË¡Cqi0H¥jK']Õ>>} Ÿ#Œ² ·×?5hi‰·ë¼Ë*[zÙCßÁzxGÿñ›n©|­fÏ|ÙOßÜe ?ßM•xçÕo­,ÚY@ƒ—‹3S2!´ÊW{ŒÁïéÛ·Õ™¡Ãòfãúë[z3~²ïä8tW˜})I:=‚®NÅCï¥jç?£w¡·ü¼Á¹=ëÐ;U5rj:ú$ª¹;ž¯@ß÷Þ% ?ý±»/ ò1múçnÂP(I†AÜ( OºüØÎ0âî-%Wé£ÝÄyû7`üšÕU¿"˜”Ž«ù.JSWf¼½¹§`úí.†ôF˜±Ðà¾f³šzÝ[ùPyYàŸ{0 ÍàW¡,\Ÿºþ^™kª†ÆOÍÀ²¾cBùBX™ ½O k•Á$³Õ°Q~¦Æ)ïxÖ­é]z»Î•¥q’Ô"aT¡ÏbÊ#$êÏ'o;¢‰Ä–ô/^³B’áÖÏÒìH·rï°@ îñîÓ»wÒ–+¥Þl©Oï¿AI´nÁäÇ.íu¢…?2Sˆª ãñ{jÒš–ÈT—q?YFY”jôÄ™¢‘õ•ëX²; ²­kÞq¹ÐŒ'5ªØËΔ…§ yØøUªJù‘·tHbÏÆ*òkNª­ÖAU+Ý0Z:ëQk™Ç„Â]š—+·ò!ƒv{L ŠÉ´ï _Aq?£Ð~¦>”è%eåãÀ½5êäo¹£ô²ð“ÔÂ({ÔïÞ¬áÊÍ'кeX¢Â3¾Œ¦5kT\Sº¶iæ…ÊlåÓÑwçQEÃ#ôÃË%T} ¡î4ƒêœK¢E –¨IÚ$UqÑ ÑÚ3™åj­‰=ÒBm·±zÃûú¸¿baP§=ufJ¨’âQ·‰r¸ßÈõŠnw=ø‰„<‡v+¡¡¯[Á±ßhô#îq83šËOËÚãì— j¯‹Ðìð­{'oÌoée³+¹ ZŽF¦ý8€–A¯öÑP¡U¸e÷Ûë‹h]ÀK üZmÖ»ÜpÞZÍõIï´ûéââ¡&‚—7®_|³¯åß¶æØ$°#óÆF¾o-Lk™xú•„òç—FRB~¹˜½iœ(dÕ}ßEd²£WGÊ ŽL»½µÂ…_·ûV\׉VÜ®KX¶Û^—yô@´{‹?N……ÜÛyñöT¿ÿ–­Y»zã¤9åÿ~Ñ é“”T•‘ËðQÕRÁú”Ûß.V&GÎAv¿v•Í#þÝÚɇfâŠÕmðc´RmD (},_Û ~ú²:Ä@)_êÇÌb(û–Bô»¥*’%úŽü…ªÜåÐ]Ûûâ6¹7õ Ž¢Ã‹'@ ꟪ðrц†ÍWßÙ²ÏCÓ£ó¼ùÛ÷¹Ÿ;u¡5íTlï›|hszô茶|èÑÆƒŽÐIùÂHÕñtü.›¤3ÆÝ#WÞ‡nùÏÚ[n*jÐwW‚;=7úG¥*BLÂàAî;k0Ôâö×ÀþFÝnQú²¶½jŸ€ Œ—v—Ø,n×)öz%ùÀ´Û¥Ògu­0Ó®H×ë½æ®ÐE “‚ü‘ ¾Ü"XœNøf1Ë_î±Ç…©Áê£ÞèªÀXFf0\©›‹i¦p×éŽ¶ëæ HÄ&´Ü_Û€»[ w•ý:‡$1>4vA'‘ÌÛe8xÚ÷¨Oä”üXGJªð†›ëÃHÕ—yñYÒÔvöj½GÚ5ÉQöÕ%¤ß_´îi¼ˆ ŵzë†é*“]©ó<²Øi÷§æ%!›—Ã_ó-ŸJaKa=-ƒœŸ6y†ÜÖ´z¹íÖÈKYýI©ç.ò±ÿ¶yt ù/$óëPÑ¢ ä£ŠºbuVzÁbT‚¢çØRW~£x½!+ÑÔ0îåÐRükÓƒÒc¶–²¿QæW°'Ýà”èb>¤ôåcxÛ—­¥PѨùö•ö T–®ô¿XŒªöKŽòEQý«8Óü%^ÑZ %ÔR8”Ä“†ûô(nè§» öY}®”ŽvÜß=ôFSuݳÀf]õ‰^•eJ AèÇZ"SB42~æ4óã=šð<õ¢ ¼¦¬"Ôv ZhÎ)šïq-´=u÷i¡e@{Û«;éJx?§£’ ìvC¹Û™¡EƵ~7U¹Ëf~e÷àW·E÷®Äb¨äþؤ“53ÒEoŽiîô§ª•_‘ܸ µBûÙ¥#(·F湑R†ŒÚˆ­þR ‚FêC‚QáÐD–&¹wš‰ˆß\ø´•)¹]ÝI’¡U`ñÙƒpømoZ_~ŽÚ¥½ÜÕ/ íõµ>œ)Oà‚_ûûëíþ¶öq Ð-}+zÊZò¿|¨­³@cc }ÏýòlîzÁãÊ7¶ð*,Ûgnzè= 'Í|$þË #™¬•®WÈ`lì»ÂÙ¯0õÛŠ†º ¦=¥ …“öÀ¬È[!úÖ˜ße:ý`a-é¡WC#,3­ßú0«WcÃEbaÈòr­C4”·e4¾AÂxÃÊÚkpwØ¥ þ÷æHý'Mè 5’•Ç[N7E ÖF® ;¤ æv裋Dšå Óš\¤Û¯«Ø´x é³|¢+û‘Ñë*³ØÜ'd>ñvží²Æžçìݳ¶å3mÜÍîìC®¯yW½ÓÈKçðAEùÎ|‰¼fËü£ò-zQðþûIJ†Ï(lÿ@ÿ¶;Š‹ón¾]†â…Þ·û™‰qïí®ÜÙD”9èóhÖ!åHÚÜ!Aù¼ð 6ûjT<ÏÆpât*{uypM£êk¯16vÔ o¼þžêBÑa“PÔúraòâÔÖµÌö UÃý³Q Ïe3Q7Ï11å«ê?ÚíCyp ÌDÉ'¢±ªóï#¬x@þ‡€Ô543оô¢ÍŸª}Òª‡¯Œ]bwÍ¡å/a oãhMf¾ÆCÚ€6‡‚NöO£íÏEÏÇ}mhgRœî7}í§+ysê¡Ceùð=tŠ[`m,ëÁÃYu¾¦çÑeMöY«þ5t—ýÃ@ûr®Üüh âˆÄºg+‰;à÷#Ig+ê ÈöIxC­` i‘6ãþùlÏ’j ¹²å3Ï?~" ¯ýZ¼ ?2ïðöÊÃ~Áq0u‡»ˆÜ«Çzµ!¼Z_õ†vèÎmñD÷Ô&ã6¯lR9Nÿ3FÙ¯‡Q$o¯›¦}ÝMò_>t’«7>´í‡]uÚ ¶×Niú²nG+ì—¶ÞW¼sÓ¾b»¾˜rÌIJòDÜî\‚lù²Aµó!ÿåÊ$íŠ#Ûçy´Vÿïù«M¹Ñ´ u¨K  Èä7Û†êi(®´?ñaU J$};Cé@›Ò˜ ”· {9Ô¤C¹Ãa×®¨VyµBõd7Ô^ 1Ù˰ú_>,ÎÚôÛ7 MeìÏž†V’¯º =;\ÿ±˜RÓ.QÛâøŽ¡´—ô,=¸Á/Cýô¹†G‹ap¿/MÓš _N¦w~µ¼Å-˾ˆ”9g£#eø“NoÍéà"ÃrF?˜uð›­º:óž+UÏL#`1&ö[œº,ϹíþË k>¹:1.œ°IjûÈ­- º„/µÏ]DBx/½*î‰D™Ãš}'‘ø¨ßîŸcvHêq;~ðs’s|Mj’EJ¡QWw ¤ÑàÕW¿Ú„t"ÝcßÉŒ¾jz† P Û4Ì´!³·Ó÷‰Sd½¾xfoJ=²t¹Ð*‹\¿^SwLø oX¡âÝh%ä¿L2QÒ’ƒ‚’šç|fP¸÷Œžë'»Ki~hA%Ë332k tk|Se ÊVž,°ëuCùÖˆ§ß*P‰‘Ý“½ÔUîÚ7!ËAu«KOX‹> œ"¼&Õ.†ZíGu¿* öýWÝåù¨cήÅ>øõäÍ5Éù­Ñ@•ó)ÅÏI4ÒwÐ&2Ü@Çø‹Ü~>hz±ö`÷ š_ ÐÊÚƒ%m‡º ”ÐJú$ûŸ–h}J“Jˆ mœÿÔcðí :tÙÂñ̬ÆÞ¨3èèôÛSα FÆ42CÐ%ãÙ´‰ï^ë0rè79€J£ß™ýïÅ!›ëר#3ÏazòS8aJæÖõä!&èË …Âo’K¾×oùB‰oÆ@öÛÓÂ%ÎZùîkƱ ¢~iÔ|Y'<ÿÏ9¨Íþjÿôñ¿i x²hòåÔ¢‹à§óÅ3Ö§Ç·®ÛX͇éסAöòGºàÐ\èßïªP¿ãˆ„ÿÞ‚öÍÝMÌR–Ðyûøg®ÐÍì¶ðë ô¤7,•TB¯ÚðËÒ½Ð{n0%ˆ»zƒÞ:{ßþ ½oïË^'5‚>)’ “.ÐÏ—¤s7K®Ød5‰ëÃà@;ƒ‰ý!Vµú±þÆ“,®ïK„QÊú‡%0öL{bœH&Ä_È]½é“-J"¾ 0UûÝíLzLÿ »ùâ?ÌHî_p7…ÙjùÆ%º5˜[žYX0 Š!ë ù°®š.9¬¶¥[“Iq«m°RL+UqzÖ¢WHß6‚gŠbaH Wà^‘]€»Œ‰Œ?µB“‡Ew¤B¢ŠDtA|HÌþcÜ=QIuèž'B2w©œ¯÷‘|íHSvRTtþÈ^º‚T´9dúÌHËâx½]éÔöGYŠ[CýQÇR-ŠØ-ʱø©ºš,£Éd ¹kÌѶ¨ Z#*3ALÚŽ\èu§†Õ™œÂ?× SjåÔ–tèf;ñ;šFzdo¨_•[ÞÃERâ× /õ¹áÈD8 PE11Ýò‡Á ,G?s¯ÀІtR²$?üývAC³FÃ×¾«‡ñ/!ÅÕSÄvû¯TuÀ´ÞP™î˜Éyö@TæÒ< HVƒŠ>¶ëça±`¹goP,_ÚÙe8«ÊÌýOèÕ`]ÔeáMš0lžï¼V_„»Drcf,B°JAV?¹w?ä4 œqD’ƒý%¬åÑH& =Éu!É{ùÓ5ª"’U^3Ì©Œbî¸! óz©èßHE¢ô͉馯k¿¿Ž NIðy"“ϓ١AMdññ¼õšÒÙN¤vÆL ‡üqž”»×³ä¢\{ûyäÖ³æ÷9«‡<ó\Iwq#a”e² /ò»ú?àBA^­]_Ï °|½Š…8Šúg¶¹ê£øŸoeêgp¯\@÷¹€h”>ë"ÿ£ e¦}_2²Ì œUð—º{—Pþ÷üÀ„Y*>äïþv¹•ý ý8÷D¢?Ò'©Î½ãÕ‡·\ü*ÁUžø¶îE¦Ò4$œ–`dƒøˆLÇñnçB{Ï¥)b4*ŽR6½€ÈYÎ]O]Üö«3GfÒwæe0:Ëë˜å@Nò¡êÏÇ} `¬‘¨ÿØøéËú÷@Í5(«3¾ÿîœ Tî1”‰ºEU2¯äGZvÁ¯O?ÕÑõCuŽTë¦=Ô\ñé& ˆšÎÙ–wÒP“mÉ™Þõeëõ³í<¯žARÏà»[ï‡Cîïºízb k0Ô=×ýÃiwêŸ=úk5 ’ôר%¢ 1ñÆAïДPp$í,Bszó’óž5h© = U­k„±>/Ú MÉ€U„6ÚƒÔ™½ÍAÇòMr8ò:X‘½ã†nQ‘® KLÿåC<öd¯¡/tóômi+ *•Rc„ÁØ!ñþ:>qú¹¯m ŒœÐÎIQ> c±'o¨?ý“³¬÷›™azßŠç˜Æ̬êÌÖ.vÃ\ {Â9C3Xx;®¿W)–Rs_ðÝ‚•Í+¶qߘ`=ëÛ=ÿ«>HàG~±þr&ê¬Ô¾y"»Ì ;Œ$à]6ÿB ÉNŒ½7«Ç=9Ù{⧪Šk‘¡EŒi‚´#\O"ßÒÉ‹ÚHïÝpFþ72ò>Z™ýì„Ìi=À݃¬FÚu_ê{=ìÞ·D0ä"óÓ÷ùzyÂxÿ˜Š#_ëƒ4þtä"Ržc‚‚ÂÍÅ×PhìNî÷¤<YP}“;œ„âf/?Lï®Å½Œ‚絋Q†+©šöÇM”­¾íέ…ò·jéØˆkQÑ8!¼á® *T^ªõFÕ+¢×Ë4¡z÷¸ÚÙ!„Ë­:­Ï#Q+9ÒN„š µåíG¸Š/ãþÎèã­£j¨ûòwÔ绨&a æ²#þz2¬Ä‚Æbð"æU:}a±Ítg hÑüY–Šˆ©?ZÄ6»<BËßœvtÃÙhͦ·¦;6§uÛÎZ]@Ûn âô[îhwâ›ü§œ*/Ã=b™«b¼÷õ÷ý—ÿï:ê–¾Ûr¯2ƒo%Ø|k†Lg ªùþ{[¯çSÙ™üŸúb¦žé¶~+d„=ÚÞ$/ fó_>|š_"ys±ªJyþ÷|(’œúa„yû\GÈïà!(òcøÓ]ÙÅkG3^Ÿ‚§«cmÌסŒŠyýNÐòöù”ï]PŹ’âQúªµ½;rGÜ 6ÐßMø9éù0oeÐYj‹óoÅš ¶ø~îéDεÃuÇäGÝØ» ÐMÐÖó'€zªF%ôMŸŒ®ÎƒA>¢¿¹C0lÑDþûX9Œ\$Ž ÿAcU_s8¹_Âä!÷CÁ›0mó‰‘H&r{]§’'õÌ+ ‹ô0ÆÀ¢ß1B÷-߼ܜ%G`a kŽâÇD®ÀÆÒ¥Š† a$h|Áós %¼jý‘èåŠ-w$6´ãOˆqGR 'Ã/G ü³àç˜â+HiŸt—H9iDsVÂB瑎µ8¡Késn qFFÿ„k©ÒÈ|0d¬`#Y½, ë]‘½ÍNتü1r}ß_ºûò>Ì,rkAþÓªËRÙ%(x=¸jeî0 7;å·=A±à±¤£·XQòÕ9µ.Bs”vÖèg EÙz¶Sý»P¾ët­Ð•`Tâ¹Ã£jÀŽ*/ÍÜ¥]ï ú1‡©¼“)—~©ÞÜïŠZtÚÏN¡vˆZV¹êÚìÙ[ê‰zºÇl¯¨F Ù åSÒ 4òÈ$©âB“à#£3¡iZü$›TšÇïîŽ|ŽËUÏü&6ÑÊû÷x7uÚìêîW|ж¹º :ž£Ý‹gu6uñP¬·öYòtlýrÆ™Èë~«;Ö¡ˆ. Y•¾fÂxô¦cH1*,³Ú&_²D6‰qÝë0uñdG›×y¨äšxÓœ¾²åÿæ»h'À×74+¿îB¶¾û1²›™©Ì*Ìéw>ÞþnÌ»}jNµªb+ÿÍ ¼~¬½Å-•…ÿ=â}üÇï»ËŸ È®ôlé­Eôñz6øIKÈî è³}ž"@pwÔ/¼ ¡:RÍW¿­8\g„ßÖnb/æ¡=5¸é‘0tÊl ל‡î3¨…gÓ '@àlbÂ$ô¬±”ÆÕA/ÿAʧr÷¡—5¸þ$×è1Ðáó‚ÞWÚ1@k}AæËé†òÐßòÜG6ò1 ª¾?oNSC7õ*t?›ÂpwìÒìIQ _kz£)εŽOnÁ¸ÐýçêC0‘yÛÏ.ñL©ù”Ö¤„i™Æá77Æ`FûÄ>Ë‘­|èîôÚ[Ïæ)ÞlˆËÃü’ˆérŽ,º$úy¹ÂRàÛ+)`åÅÃúgò°æ{æéI-)Øp^þžƒ‚é5¦sø‹WsÄ8­ ÕV{(ˆ ÑÕÔ5ƒDbÜÝwf¿¸ì$9úñ—ê×1$¯i ·y†ä÷z/|”I@ Œi}»T*N’Ù‘æY‚PßK5¤T:X^ôé_OÜÿ¸]'ürPäËvÝÐÓñó£-úIýÉzÃv=ÑAXò$:xx’ÚAŠm/U‘åö÷gD’o×%û ïßÛÙ7³¥_âùD;ºõ/~oψ~¡ ?ÎsQ„vEoïsé{­ZÅM¢]ÄDPn÷8N>ʬOú߃ ¹3©{ú¡JGµmf^ª+™‡W+mŸ“R‘! €z9Ãb†¬Mh¸[ß:/š8M~N{†æœDâÅ·FÐzK¤4"í5´s¥y@C–7^Æ…N §ç¢õMÐ-»:r€ŠzL×\Í:½×³B§iò¡ï׿94!=™ÿ_ЂûcæÈ9F³KXö}/€ÑXjî,S¯U¨ž°“_?±õ‡Â´˜µÏå\f˜ çPìãƒ99ª‹³Æ>0ßÔ‹NWC)8_e‡e•ûëoa¥ËXÜæu-¬}Í}ãò˜6X^gJ Ag˜»»å$¼®í™Љ»÷M>ÑdÑFZ‡+¿©ëô·à|æz;’ß±ÏÄ—H!ü"tQj)?F{ü¼¼‰ÔNœïèjVµ+ðp¾&Ò½rXçb E;3&ÇÌ¿ÈäBý˜Ì;Y‚‰†93#Û‹´Õ†ÉRäp¸¢@ÀW‰œ-”ÇDVí‘[åņ… òÔ¾N GÞ'k1òZä7¬œGA†i•7sI(,KeïÞðE¯˜“U%Fñá<5§ÌuÜkøãþ®TG”~ÕH'k–„²L‡k™ ÅQ«AçQüÿ(ûÊ ¬Þ/ZBº»C@RnžM—t ˆ„ b'  ‚ *&JJ( ˆ "¡t‡ÒÝuõçeFç7ÿ{?=sÞstøpöYkíµC–±zêçþ0«?¥´ÉÑ o?¹g¸-ƒRóÇÄï Ý Êý>ûø@~‡ß´‹$‚úQAŠ'½v ñÀëWÔûÐl²ùæSÌ Ú*>¡ËZÒ óÊf.H’ô4cÇí/:€~½_åHë50¼“Øq®£ŒÜˆ)-(ÀÄîe‰ÅO0Sòáh´³sûVïgÕ}`qwâ{I÷°ìËW7*×kŸµ$-b°©‹Möcî»s(ÝeY܉êÜiÒÿv]êèâm ±{4!ùuÅž½h¨ƒÞ.ÖhG¢|Ú͵û™»Ñ 1¼€•LèùóãÃä(uð‹ ÂȽ?õ¢|â·PÒúæC¿fÿÌ-uqŸø¿ý†‰z“êÛ2º>£Ø”‚dÓDgÜsú.ž£·—°ÿlC¥WçË›QÅï¯Sw U¡=û\šsPõÕÉÖWåèkº«ó 7ª™o{âO„êŒß±pîOÞeÝù³ôo(qn6¿û÷uN¯sñT÷Â%°ŒÕ` ŒnDu½çòŸ íCßxíOJ)CßynJ&¹£þ{=‘ªQ£º\cô™DÔ¤mÿü,jÖ¿ÌûSFµ˜ß¹cKc†ZÏÉ‹ÞýN‚Ú_Ÿdx—„Ú»<0R£Ar©ÅÅçlPWñ]ùš'ÔãN8÷æ5)êãý³·±£õöUÑúySðm³2˜½}D³Ó ùFôœœG¿öNúôT\A£¼ÇSgg{иK¡Wš"¥ì!M°@3Ÿ¯ç5Ì Ùº¾²;,Ìhþf_âr¿3Z´•Á›ˆyŠ–5.¤ß帀VíDŸ[ÜDü'cx#ÀãÙÃ:ÏWøCYÍ>)t@Ø hìÁsˆº7>ÓRJyŠþê@nóÍó :P¾®ÿúUÝh¨C+ÎM 퇔"¼5b ß£¨ÿi šžô~¸H L5yµ®§Óež&ôaìg`—*eÚÀÎÈË1~Ò×€‡ì¦Ý…Ÿ1Àû}áJçŸæã+àÿ¸IP*œ{®‰-S8€Pȷ΃PéúÚ¹1 {ŸòÞ^îýÍ;+ËØMÞ‚Ô#Ž\MùG cI"Àš×r»¥¹UŸ‚0Ÿdk^(Ù 0°º‚Ê[¢Å–~@ ¡æ=I þð}iSìŸ=À.œu‹ïAëËS Δ ÚMñ¢ô¬õÞK,Ég×ã+Åa?wÂ÷vm0fe¦8Y¦\GŽ-kÙ#ºÇï+2Á<ÅÅùrs'X Ò‡¼~HV¢ªëÇ[Ï‚u(Qÿ¥`³ê”T+<v×_½½Jö2ת…+Àa½Ó›¡»+\Ì]ÍC ñó “×3Eæóàþê a؉C°»ì=s +<1 R·¾Znº‹Ü=p÷·>“8gG@†²Ø›s=9dPÚw¢J£³•(‰.}©òF5z¢"wCÐ|oñ{ß]¾ô,wÑ}ÔiU»Q´”ÿ®{ø3¡à£Uìå>”¦m¹/ïË¿ú°,Ò5ý zE]qßþ¦Nb} ,/½”ZQþ¯ÌΓdP~,2¿¯Ë…ò×uo_³©ÿ£#+ñÕ-Ðk}õ”§è åÝ«]ì;ú0]ëö®›¨ˆTÕRŒ‹÷¿õ¡”7ÏkލìÊ ÑxY6*3yZî$>‰ŸXŒlgDŸo_Ÿ«Kxòw-+2¾£j&‚ž@rkT#³p,'ê÷wB­ÄCëúæìÇËL¿£ó˜$í.@ÍÁ.4Äï/ ÖÈL·_·÷áâºëÖy³Íý¨7Å~Eø·>LÙ|½×ˆñ¾î™WÊEƒ›ËÑ/a|+©4ª67íÛÊÆ¯zž¤ ESR»Ý âÐÌž¤5£R4;דt¶ -à›‹w¤ %q=¡_õh¥’m­0­ÄŸ—˜ìF›#I4÷1^5ã·(O `óÎúv1 C|Û¿2½¢}ªê×,ΉªôT$Ík K•ÿ(ESɆÉüùÌ3& #ûÄñèSíÍ6€ÑeAØÏ-˜5*ØîÓo«ÝsÞÖ•i`ÿR?ºïÝpå…ê±_Þ0 ™ 1às?Aø(ªκ•ãûÖ€`µá¾ôøS T³yœiöf½9”1yö-³&¯»s‚T[%Ï©ÆÒ§¼5¨ò¢ -¶Uƒ ˜£½¸þÎT.«™92$ šoÍŠõ¾âéªùVÐ|ÐzëˆÐIÐv R¦º¶WNM™õ‚¾Gk(»Áo}2.Hn¢ F/øœð ŠÀ¤ß;ÔÜ›̾rk°ì‹½Åud\•`™­_úC§¬Ìérƒ--“RÇé“`·Ú÷¹i`h†õßzÿ€ƒ– —nd”À¡êÈ×÷ ;ÁEµ-5ZÌ ü¥ßT}¯2ÄgµyáÛ:5ïhO4I¬¯uØñìo>çÍ{ý7ü¨¢Çd»åŠzçeC<ÐëAûtà a”ÿñ\Ü÷Ã?þÔ£)9.‰r+ê¿™d‰£ì˃.úúôO.kÃá#¿Y‘çJSåQáñ3²½›¨âãÍr—•)Të\Vöξ 5½´'SWB-J„´Ö_Q;QÃÆ^Ýèӎ¸EÚ}u÷¹ê TF}:ÉolÙªQ¿ùÙÉЫP×yÕ+Q]åÉ$]KÔŸíªë™¿‰úS‰lÊGÆÑJ5…¥WRèÇÐꋆ.U4 ݼç6y¼ÜnùY Õ*Õ OF¿¸•ÃïI(¢w×Dn4Zz)¯†çç XZU¶FwÛ>²tô¡)}UG)4ýä©qž÷+4Sñ åœ2A³ëa'~\êEó–ÌNìh¡@óuâ 9ZaPÎjû‚–ó™²l\G«Ç(.:M]Aë:Ãíi+ôhSäCé'’vÀ#HPl”¦¼Ÿë?Oyt,c.ò-TŽÀ®lÞèÉÄ»yÚ%[€ä±²÷±Éx cs¥5É(òç©`ƒ«@Ÿñ”ùP¯Þ•ù¸´½Wsyº€þ¤³¤O¤+0lÔçª)ž&·NEŽ|W`ö–=Xza°¯©23¦Ëì—¡¹ËýÀZÈÔèdlkFF〓ʊ–“³¸§Þk:†ßƒÝêg¨ûn™¿`’_!ñ]X½VU_‚G‡%º‚Ðé û‰7ÇAØFg˾ÒDŽÖ·­*‚èÛ•ÖÙZfØ$[ü­]$ºŠÌ8Ar æ ¥¼=Hß@bD@V§×\åjȵ¹‡Ë?•ÿƒÂ‹ŸMƒbl+ÕŠx( y9ôµ´€Jã䮪€s –•:Üw†à…Øå);QPMVо4šä9ЉgAKB½5©$´µÏFy® :†‰…û3Aׯçó¢è]½@$^úñîOëôÁàËœJ³Åk0Üì¸,Föw2ÝjÀ¸Hn£˜LÅojM‡¥€Ù7Ó¹GÁ\pæ™X˜o1zW¿= ?éÒ™C=Á²ñîù뽚`õý™ÂõøL°î‘î(|þüÓª I‡¸Zâ`ûžÔrAñ<šÐø†¢ÿôïÝÉi¥ÿ­Ë„‹‰÷Ê¢WCç‚]ÔqxõbŠò|â‰ÓtÞ;2ü?¾`@ Ýé?:ð|»É¡?ºÐZ{m%Ÿ½RÜû„¥,¬(ýÖ‡Òd¥³j¡Lÿ ³ôvJ”Óa˜ÌùGwª®ÄÄãúîß>x¿õÐz‡[8}xKWB‘‘ô÷÷AgnX•õ--šÅÇ ŠÐ¼÷‡¼BŸ-_lVÌv£Ê©ŒÇz=—þä1{»]ÿÔ¹d'SFu‘¾Ö¯^ oظ²Q¾|é°«j4™ðæT•FMCÆí~󂻇r¢o× ¶³Á Òô¨ã¸¡j>Uê:F{Õ¤ õœ+}÷äÖ:ê3ŒX]§ AýÇ}»¯(}A?gÄ(\¾ú'¿ùÄÍ® ZœËäYLFCoÈØ„˜ùÑ/%eñ=suh¤{ùžÆS4–^vûVÀW4‘4]*áÿ M•Võ*ï;Žf(>jLí}…fÏ¥<îM¤AóTG:&“ÐÂûŸÈÈ{Ð’§Õh*Z^¢ˆ³ËsA«a§Ù¥<к9½çéU´yÁ(·âá=À{xN´Íò7¾I›=¡²ÂŲ¶%Ua ªj‰±Ú$wˆ×BøLâfŒèæq ÏgžÍ­ôJv¶š±S1@•¬¹újhòÎÄïB¥@g(´¾tñ0ðIlŤ&ópQáù:`‰Í¯Ï©¶ò‹û¿ÀcྣoýŽ8‡ÇçPnÑÎöÖdQàÉæ)Í‘ÞëŒÊ@à“wÛ£è¡$I§nGô€ „þ¥s6‡@8`LAΔDçf¿ÞŒ"q¹[-ì°¯JoòIÏ-R2ÞÅzɤ_6—J¬éã•6=Y§+ºŠª@aCÓ]ª”ÒÔy;ΪÑRzw 4‰•sPÿб4Šbã¡ühíêÉè}QÚžƒ6IºR Ó-dÂjz®œømσþDy7§.楰q£dzæé`â³õM¹Ìôyª¯þñóÐÐV¡°ˆ½XkNÊ–ì·Î¼k':B`ó‘Õ[x³ììVò%ôáp`S¾gËØïÁç=ã$Dx남•œiÏ©ê¬$àæBaó °9OÏJr¦™–”øgnéßs±ç°ùPØ<(ŒVœc8îY|zgÞÓö¼*lÞSM‹Ã©_‚¨¶ÏTøÉù[⊵x!7_›sŠÝ¯õºkð*› w›ŠÍ=}ùò€åo¼nhçÏb~“„›×4&×Ë:§¾37 ›gªá¥ò¶Û µ_ó§=s4u Ó¡±-D]Á;œ@=j¯ËÒ ­§Åî§þuÇ£×s@xj]*­õ ù߯{hzHžxR~˲e¿G{¿¥é³àÛ Ž®ž ½j´—šüùU]W3Ñp 0\Y&Ÿ Lþ¯ CËNKB—Ê)×,`«Y8iù§9YƒÆn<»Ü~s âÀD†¼¨vwï2{†øî‘,j”†=ô‰’'×AˆÁ“þj¦;ˆ]#&ÿá{ÅKW._î„}V²ÑQìGAê`QÉ8Èð Öƒ »Xœœúî(•?’½¯ JRÓ9ù¨^Õß…>—Ü_-øˆª"o–ÖeüÉ˲ÜOÏø“—U?u÷=ø¡ÿ€-Š{G¦këW?ÊFÍ.»|sÇÍ…ÃâºKLÞÑ‘ºõêá ž3ÃGý—¾åÙ³£ŸY RÒÐ`—PB`qÞ”X®[;ŠF ~èLìC㲌»dO¡É΂îÓ?vâ¶”Á÷3ñ4_ë×séI&Zâš -IB+q®øK»ß ujŠÅ×jh³£\+·]ðJ›Ó=û€€¸s¤ˆó×L/#"&^'w¢ "}xøýs ‹’O¿“L”ꙺ]o€ÔÅèƒЮ¾Q8Ñ ôwRí€q¿ö÷UŽv`[Òu²V’å绀½PïEö~àJº¡±.yxöåËϲŸýªMï>p×($ÙRÁbù'c xn}>µö®„œïðI‘k7í-Aª÷|Ýš‘ÈLðè/o€¼¼Û­‹y X¦|½šAh'>ï¹æîeøê?6µV͘ó‡wµèöé¯ü–7¸@÷èîž ï_°õ4| “ ¢©¾9€QKÿ·3².`Ê´ŸëFI˜Í)<õÒË‹Cy$ F‘`9úæþ€ß°~ôQýžMØz øµîÒƒN{ßsV?‡“!8XüÊLœ…òj{™JÁ%h5 ˜Äü.ü5¯ R Õ­¤ò€-µ(̱½­eè]ÂÍ,ý½ÿÉcÜ{‹ÍñÅÞ{lÞaîÍl·‘(ûòË(¾=Æ8ÜÄâ;±yÄØT ÿj曈ÌòÈQÃÕÚD\Oq8„›³»+¬¥4°uGôÜ”Þúo°÷wª£w†¼ÿÜÿO\Âæ:4Y;¦¡_w¿î…#Ü8|ã8oÿöa?%JÕJ#‡&9â¸X Ð4퉗Å'® Ý!UŠy“¹¦›jø÷ðÐÂi'nm¥\´8soÌœ->{.IzhgÞ!›ϼE´1X4|qm}dÑ~Ñsðªó….[-~‡œ%·ãM dŽ™|=R »Î×Oæë—QOÄ Ÿa è§r| KÒôþgc|e@.ÂGì9K”x‹L Ëü@í쯑¤hc¿’)^ÞzåïÝיء}¨5J¾˜}ïefZ½ÑCÀÜ#asg¥X´“ê…¥‘´¹ú‹°Î-¦¼eö²ÃÏåoÇ7Ík•/î§wl&³¥ÙøÆ>ݼ¹—”?«e€ Æ¦Ðã;¿õ¡r¬$IÕU¦|ÞaW" h@4.D­N(–2Â^ή·9ó5 á?>ä~‡ $ß=Hg¼&Òî=‘é¶ +öEê mÈÕ\Pš{ï {g¿œÅwæýEùÇ@Ùžêã +P¥M>?Ñj?Ô¿©±®Ô°°ô”M‚ú óSú/ 1›£nz4Ww ³8€6¹M%Sèpú.g.–ƒ®¤ˆL‡KèéÆøÒ¨ÜÛ™cê7X¡1¤† Z–_iaÿPÕ‰K¦{Àx‡€£4˜„J=Å#œÓ £57ë;`öÒu†Xã˜_Ì´jþª Ú”=E ßÁ’½¦~ë X®Ý~¶öͬ†| ¯Šœ¿€Â._]=¦ÞO<ìe!2_/Œ¡±àgï=¼3PÕw½ÏZøÅèý¬otÎ-.aøƒñÊ¿ñã©Ø¼nìÄ~ÇžÃâ›+Œý¿ŽaqŠÅ?N¢Æ‹ß#ëqó¶1¼úÏ9¦ïßRf¶õîÌÙÆpj{Îö÷oüŠ¨Þ£FFR²ÊŸQS®)SýÔâÅv@é¼ jsê4{Œ›§ãŸøûX PŸ7Eé4꿱(qåƒúQÿØX Ð7~}|« žïO•x‹†ZñçúèÞ _6%§ZSÑÈ<·­Qy0ËÙC½n &=)Þm}MÅ5SØ_ÒCÓ“L®O¬Ð¬9À »M4÷‹Ìºìn Zxx˜ïÛ=´$+~ÿ W8Z~ÁñÐ ­îËÕSº†Ö½¶Ú,°¸<#ÓéýVV€?èÔzZr³ñæ¹PQ þè Ñ! ?¤MÅ·Ÿ~Prxdáuh(ÊйÉåˆý…ŸqË@=% c€O ´‹sëß;X€~œz¤%j˜´H>ø1Knûñ@ü`›BqÅw¼#+É1ËÂ8—¥ëKUx€›‹çU„ 'ð„%zuYjoa¿­è5iàã}zùF»6ð¯Y¦ùìaA±à¸«m+ êž° ¢[bŽo|é@ÜG³¯ öÍÕО8ÿ›¯*+²gÅ‚ôßl…P9½y°V³ä­HÆŠ@Q£äð;“½ l~³ÿÔ2P 4ÕüiE ¨~EÅyÔfNóQ³ƒÆjÓáêï %ç¬ûØ´YH‚g/ÜØ@/0ñÖ¸bPê–|s!ÃÆ ›ù¹ 0Jóɾ÷è%˜Ä ÓÆ÷ À̦ÒWBþ-˜ŸÓÍ~I[U?¤ËÀrØl¾ŽwXkTÞΚ«›üG†ÍCÍ`‡ Þ¥¶†Ã7>ë¼»Œ¥Fd  U¹%ë)Ô’ÏÙ(ÙCãqÎý=Wÿo=ø7_Åx*ÆGÿKb¼‹/l~8Ž÷=:Î$£}á¿u"¦ÿ¶O W±û5dFtÅ/vÎíû¸}³â”üŒ7‡æt§‹¨ÑdüUõÛ`ÔÄEsúÓæÚoÞžáüé-j1J™ò3“@­/¸ûæxQ»'q¼ý?Ôé8³|‚uŸr÷1Ê9zß´}ЧúÓ'lz*¡ ý°s3ð=õ 7o|Ѓ‡ÐNa ó~­šá@#$Þ½68üðe”¼=‚¦ÉÈ^çSæ¡™c=§ŸE¡Ù5MÙy 94ÿF/:o’-~¢ä(6C+2÷?« ßEkã-ß¹æ7Ðf£uà`„à­‘ÉñÄAÚФ¼œ$ì ;'L¢Ä)öDñ£I@ºè.x5Ø(ÎÆÄºËžªq¡µWƒŸ€V‰üÖh@&Ð¥÷3—”rƒmBîSÏ|`Ò×u=œ¹³ç›^\ÄB\ Ü<äJ‹¿ñÒœx#þâ2ìNf'.þ ü´ØÐÌEÈ‘ÔðY%ÁâïŠ üOÍïnˆ%~â/N¸勺ÊÎ ÅqEç¢ËHàÇ’FÒ€l QŠÊá— _UàÕùæ(N†“ÑÈ–ŠŽ¶Ï¡ Pk%t7'uëQµï¿q°»8sE´ü+÷Ö¸€Ž·ñϤ.Ð]œ?3GÎúõOoŸÈÃwn¹–ÛƒÑ+Âå+¿ãȤh_¨S³=˜¹,ŸºÑæg5Ö rÅëöç¾×`¹~^ðŒÉ°¶Òõ¨Û6_Utäm¨ÁÎöÞzñê(ØÜLŸ}Ÿ eD} 4p0õ] ž\,ʾMGœóÖ±¼÷­Jt0~xÓ¨×ÕÚÎéÍÊM?“+¨©ãB>Ngaú ‹¿ÿW]øˆ2+uåŸ{‹GlOÍÿÒ‡Ø5öwáNæ™ïMvöaüõïþÖ‰Xþé¸ûØ÷ããØ~‹r£¶ššf¨âù¸GXxn@å`ûÈËõqÜw§·óFX>§·÷ÛàâyÃ9‚âéIÜ›ÎW6G)ePO~B‘{ÀÔO¤ ";Ð:‡†Hšv4PÞº'´f U©³EŸ{F˜ÓkNñ¦£±£üÆÏÖpúpf÷d  w,š¤½)<ª†æ‹_dÑuأŚ ÷Ÿ#´Âe]Uø(­Ý¯Ö™ß’G›ÂW<’]»*˜wíì/ÒV–ãáÚ™Àî§UI'S\f – Ô¸xå—:ó]¸S ÝâYL®6Ù L‘ã5 ×(qóø÷ͽïS—=R_ì¼>J€LŸá+|¶¯XáJżIór‘ â7RŸFÏ èŠj•ž¥*¨·ÑD1‰€æ-b£Åy Ðv’ªÝ‹ÇÅ&6@ÿx°Ç–o*Fvñ<›£ëù¥½`²9¨QÜó ̆)(å—Á‚Œ’æÜ5°ì̽@ô¬Cc³25ÁÖšƒ«i3>”Ãc8 G7.ì“'‡ƒoLŸS‡€3{ÛÆg½Sàr~¾é¶I\>Ö '»ô$¯›Mh {ku½~šÂUu¥ëL‘AU¼¯ÇΙâæêcóí1=˜wAsrzø+ŽWþ½ ‹Û'7Ø?{úãNì>¦3ËŸÎr(Ûàö´|c­J²MCR—l¸§$P Ý{’BØÑ‰Ø{ú%“NÅõ]qþ ÷g/¶JÎË2ÔWmâ´¼7Çû„…š+·`çÄ~ß~§©3$ú›qïûÏwV¢×réÑ@CT6Óžy4DzÌÛ'’qG?Ö´†P½pG£¾„¢UU¥;ºñ¼ÖN M4moÅr{ͼú¦wÓVÍ9œ.±Ë“F Z3J!¿ùçÒ‡1>Šh¥‡çøR;-N'n:r©ªM¢­—ʾ­Q^¹VÌî¤k€ßŸfv­ ÷hD}x»n¿.(búj3‰p ‰b:¼©>dÒJ“i¿y%¯l¢‚ )P8&øH{­ ¼`jiºRÇ‹—÷ƒ Uµ­ÙW`ü,rLû0½õ¼dëåÌ–å eÀœ¹bzóK6°ˆÒµœâ ¬ ïlªù]!õá¯}ÀÅg¸—ÁJ xªL5>>m·²*¯Ç @—%W~ƒöœÃ÷?ZÔ‚ǨYúë @Hûˆk^þ%6>ÃsejDBüÌ Äö~ø_âövýš,¦ I<>ßÔN RwÕOuù€ŒHËgŽŸ »Ú»ù0úȇËf^äìE*©¹¸ýŠ ôD¹¶´3T ¼ñ¯ä‚5U9kµ; º°rýQP?¶LìUž Ǹ9DÖã@óÔ<Å‘ ÐòþBS¶ÚjAÅÁã AImx—tc<øbž¨ƒ^FêÒ™Ðßuêºu74rx¤ýt[•dÇ\ÁÈ”)S·¿ŒcƒbOàu‚É‚è®G‡íÀl_ó¤É­Ï`6aŸÞ…^‚y~ }ã:XuúåNK‡,QOÀ ieÔ”À刋Þ yAüFç±DB`ëa£ yÿû½Jš®¼Pa‚ÛíGÂâ‹7ŒWbyŒ·þ½‡í¿ö\`8‹ýX\cø†Ësþ‡NÄå5/rž‘{#‹ÊéñÞéØŸÄå}p:Ã+lÆk1~ºí{`{Ü0^ŠñQl[ë9Yo=Fm\^ÛÏ„åþÞsÅõO—æ¡cO40ÍÚbÜå±³¯)Ç\ûMß4ZÖ6ή„Æm ./„¢ÉQ?2üd4°¿döš&šI'þø™ Í‘S3Þz冿c²®Òx2 E=Þ¦;.h™‚žî­'Zi)XÞ}~­Uuç  Mzù{jÆu€çaåÏ; ø[^o¦^Äá·Ö¨ö2žýM^šÇÒ$H˨÷G²YÇ­R/ Ïég¸•´(>j+?¸T1‰±/ðŸ­-ç*«é П;ØÓËr7MûªzKÜ¢ž7%°-v›Ý)pÙчÍgN¨TW݃ÑPKwà¡}bu4¤xÒ~‘™…ÝEÊÁ²°øƒ– B)u`OìÖÆ) "ê>sßßTD]^0o¦8óýµ‚«Ù°oíS\J·8H¹Tñ; ҿ˜ë•m+A6’çRÚÁ·ÿjè›õú4+©€²ÝÆ¡Ï1ïAõ†ë&éÉ“¿?èù‡ìª @=Yš@Û›4~q“Uæ‚–Š³ô‰ý ïzñ&€.Å[’лÁDòù0îek¿ †]Öõ×’Á(g>¥T­L’5Ï}n®3G[ey¤`î+§Nǧ妔(ÁrÒȨ𑬠ìD&îÉMÑòK¯Ü°Ó«´Ø×Àns—ÂTawx¥«k7»ËÏLŽ£–y›!öÜýf˜>ÄòŸXÞåo}ˆáÝéCìÄžÇâ㺎T—Táò.¸½fÿ#Žp>!vnï“Áøá×û[es¡¯ve6¯ qùššÂ…šîQÜ~4lßæ3|÷®¯öWµÙщؾÄ{u,¨¹¥‰lêˆáŽï€ñÊÝ[j ¢nÙs»¬PïáKVþÌØÆ_ÜḊ°uâ()öCÃôdŬ‡íìŸúžpBëNŸçzÈæc¹³wj}IýX1š \º´ÍëC½¹Zô:3X:ŽNw!Z°jl(4õ?6G´%É倛I½Éxô2¸T?D¢ÞêÂÁM‡{ä¡éà¼ûÓÜÕ@p±ç;¦&è®™`³HîNM×Cs>oóàÀ]´*û^(@Móé›bÍ•8\Âx$ÇÙù…˜| ¯jÖ^Š»ÆžÇâÃO _qûÿÒ}Øõß¾!öü?>âöó˜^Ät"nâ6ξ“ j BïID#(,Šw|Dl¯Ô6oþÇ?\=yq’]—Gª>aܲ°ÔˆËáô!×X^v›oÿ“—õ³.Ј»·ãÏlï5í»cËVWÒËï`{ä†öL*™õâò¯XÜNˆŸM÷<±³Q 8^jíëÎDо JOb´|¶ù¡Xþ$ZÝòûœL62ѺY>ä-¸™vÅö„ÔÖ}Б\"¨݌°··V kå H²¸<ß->ðä=HÞžïÒølº¯‹ œï¢Qݱ’ü\¨ú™èû%ÑK¨ø‹°@Júy\~ãßp{ÿ–gÉŠ< Lý!ÿ‰ŸÅâó°}žXÞãƭø•æ;{;·÷ ýÍ ±ý¸“WMö^ì'ÔCPþŽóòÖÎé éC ŒÛÛÛ»Ô+(ÚlˆúÎtæ9Ryâò"ý/ünÇùÿèÅí8îËør…… ÚŸö$yŽÆ™oõÕ©Ùщ}²C#VÐ¬Ì çâa4OE.Q¶-*š¥®Ìù åÈËmµÅÑj‡‰~ù2Zﯼ]ßÕ‚6?PÚ“£€ç4èhG€§ö$Û`ô”Ø™‡—Á.m‡þì}ˆ˜ÿð3_tå1W3¦YMó []ï€æµ´,Û<#ÐÙŒû÷Óúv©‘‡Z÷Qœ¤wk?0 ¦ÔñWS[ìæ¥ÂÇÀlAÉ­¦ Ì3‚¥'€u×ñdý ç?úåÛã—Àé»vÝ™4 x¢WT›ÎÃîc‚ ·®L?I›ã—Ë p‚L§ ÒÇ3? Ú·´N‚0Õ]ÅÈi L³¡6бýÞx„\”ÇŒHJñ9 è ƒT÷ ©ejÉÏ «zhržkl^b7A¾‘o:ÿs0(^Ñ>ò ”aSéÓOPeÒ9«Â ˆ8±j”zԉɛ¿Ç€£ŽíÂH;hJPX*)ƒ–%þ¦å»PÐŽxáÇøtŠG‰Ž_ò="™]gJ|qûÕ r›ðÄÄ”a?{×ÓN 0 |¯ Q&<ÞOŽ+Óò¦ËÇŒÁ¬Ä­×—ï˜?›i‹±‹‹Çv±Ú:V`™,tð)ÃK°*ØðÔ:àÖÍc¹µ»Ï€-ñmït)†ß¼ôAýEú|ðc¸§2_ âßæ1eöCÎV—ïB£Ï“Ónu¤ãò‚Xüaø‚ÅÆ+ÿÞ{ø_unoÅ®ÿ—NÄêf0~Œá†[XþӇؾ^ ·*~} S×zðo^ã·Û|ã³X~蟽ˆn…Ýj1^ø±ã'nïCÄüÄ^JŠ!)ž;úp¯àkÒÏGvø§R×ÑlªÆ±˜ñP³ÊŽ>,$6&â)F“O¢##šÖå9ÖƒfLö“%zâx(†k )2Žã÷ïä{üß®ßéþ„Ö2©ÏÿÈGk“£œVÉ€§!ÓØq ð_䛽ÞÁ·'½J€Øzq3XÙHêyS_„™Ü’A _öi*ï; CE‘›¯k€j.ƒ^Õ'hý^U [@*:S9¹ß}»§4¸XXXåz6Í’}ö¢‚ pˆ§^y“œy'Å;Nã| žªáoÌ˧wü—,o×ðYvú¼Í8»å;¾Ù©‚ æ²«—Ô.N`×å> bJÛ½¯3‚ø2>ÿðOüÇ?Ä|Çm‰ó±ºSÌßW¸>÷òjÐ¥MªÚñ±}¥…ÎxƒQë¨UÑÐ~EÕµÍŠß ,øŸ_bÈ ½ PWqä‘£;ñ¹ÇÅéÃÖÆæ™Q4ø”Ÿ Òÿ ¾Î!A‹FÌ_Q]¨Ac¢JTU?.¡ ÞSg¬<¥Ñ”óë æÃy;õj˜>´¥ò¸€/­ŒÝaÝDË#§/1唡5ÿ°»©×o£M¶ä&îÀã;~ÄxýPFy!¦w’\6¢8÷2é­@R™&-ã(°S“‘hxT`h¬®û3¸œ:±¤ ]‰;@?¬>vLK¯_9ž€¿¼½í‡³à¾ 78¸ÊV@XéÎ<§û¤‹mªi(Ý/pMkúæ~O¿š^âC>¿'@âè~Ösc 5ΦL>üGb8÷ÿëbú;ÿÖ‡ý/}øoˆéÄ¿êL±ëüDÌGÜþ®ü—>,ºhƒâ4._Tj:–(qh~Ç?Ü®/øÄŸXwŸké_}¸Íã±ï@=q¶ï–€Î¿þáï9¼gûwò°ÛqŒ«h¨#ˆKû‰ã×}Ù ÏÌpõx»Ê%C.àö‹ÿRRªÁ_¨@£}.S¯d¬vö ßf´ûªìf5­GëÑÌùd£˜G~Ì9`–;=vÒdX%¯´m{ãÚ®’/Ç+ÿ“ {H>ðÞ¨¸n8 |ç*.ï)ë„l½sc÷AˆŸIÖ#g/ˆÒ|W0™ âÇ÷;—í=ð}³95@Z#ÀgèSÈê‘‘o2Ö‚ü%¢k~«;{€/ÌËÐŒ Ò&rW;Íêiªks€¦~åµ¥]á M>R«:“™«Þ¶i ·à*Ó"†¼…Þö¬T`t¦D§B\ LF¾ ÏÆ-€Yï×_ûµŸ‚…³«àÜ¥Ÿ`E¨Ý¥Sû¬¿¦q¥9 ¶ïmojHGÀö›n¼­«à(øåÈÙ2kpz™ÎyšD3ÀUÂáƒo—øIjñоõÉ…¡ý¢UÀæçÕ¶ ËˆFGõ?6;)áøö^þÿÀù‡ÛùÔ¿õ!–ù»Î ‹k,N0ßËbùPÌGh(Ê‹~ŸŽš9ªx"QëY4qÚò?¸ƒÕ¯t3¯>ŽÖ3E]ú/žl¹ÍàÞkÌwÄ꨻ÍÇÏ’F áêÓz;_¨òžE}‰Œ]‘¡‰¨ÿ`‘´HÃé~If·UŸ‚¿å9Gd%î×Iý»gÛbÖˆ]Í,;•Ÿû5†æÚUÍUx7vâáØqÚ ö½hõ}ñÕ´þ5HË•tm¾"ø`x–Å}ã&j€ï/KÃA/ Už!E¤4°Kýiõ{ÍNi’¡ƒÅÁ ºÿî×®.ê×zJeмxQ¦»o§.M@öÕÊ`02¼ü°Än·X¦ƒaú㙹À¬ùŠTÿN°(Øû¦‰Ö+ÔÖ4w/»ÙqqŠ+!À™•w7¦¸xD³—ÂnÚðnšj9à»AI{²ô ðª1~s‚=¾Íò„RoAˆÆoÐk…„+]` = ¢W/ R˜Ã^ålÃ5ËJ [Qì6IFrü³[ eÅ]¦¦h Ò {~õxöl¨ º2Fòöºl··ðAaý0›uh2(å·¤.ûƒJ ´…£± ¨Žà29 àt³¢3ø¨¿áÜe¾ÌÕ|ÊÑ4š¿NŒ>ômºÖˆ»AGGÚÆq¿.è‡?*¬‹½¯·üíŽWÓnaw0ô©úöxg÷É/ž–õë`JòL×’ã7/u¸×o¼¶æ¨çBr15XHdßÈà| –b6—ïÛ‡‚•\õgƒÛ`mòïh¸(Ø\\‹Xûi ¶9†äåù{ÀÏü»¢VEH§)~ìQž%ôövhÌ2@㸠ÎÃêÛ°¸ûÛ/ü» ‹·¿ù*vb|öÕ™b8†éS¬Žà¿üC\_Ä6naúöS“X„¨óð~m÷GàêN·ý Œ§bunX?®?ã¡X[w__Ãa\¼ãê 0þ‰Åo|STiT!¤Š²_BCq«Ÿƒ9Ð/½õ–_“h”×K׸”j'¿ƒÕóÈž8‹f¬sºhçÝÐlžÞZ¥¦<šW:»¿¯Å- µKs6)¡¥Â[N’ÇèÐJe—è@7Z'dwõðpF›ÆOÅð.žÉgßÀ÷½üÍÔ䆈pYÜ] #ÞÍGI(Â÷õM&N á7º7${HÛ.ç­:4ùmIÞìt?óºOr|êsâ{× €¶öî'ví: ¿ï5ư~ïÊ'Óö>æd·p9j`­’_…`¦&X0Òέöªz¼ÏÀ#…z§Oho¥ƒÌòëqà#÷¬ì¸wøOÆD^ð„=Äß–?urƒ`}³žç‡î੽S)ËãHHÎ:hŸÉ÷Ñë¹ä@Z}Ã\,d~؉O¼¹¤‹«ºçwƒBPqÊXÐPº¦@Ÿ&yT²‚;o™‚ÚÒ"·}© ¨Ëåk~È °dWòßq¹ÅKÜÚ‡ÿøº*:Keí·F†@/yé§ß=0ð2¬³…ýF{žÇ†Ôƒ±»Ò>®*0Jqô³Íªf¥8°Ðõ¶/UnË´Ÿs‡¸Àz÷»ùD`ShüàrüY°;˜ØžÎöºåOJ ƒ£†.-ÁoÝî&SI(öª v³õgŽ bÓ?où_BÍ%1^?öâ|t ç0|û/]ø¿ô!ÆWÿKbue˜O€ñ?\½övÿÇ?>ü¶ÿ÷_}‡8½ˆÕ›n_cq‡ëcÄ|†í|Í·Ø7÷FÛwø$æ7Ø.ø”GqìÄåßþ!æ;lã2æâêÓîR?ïv 8>3n­Æõâø%]š&ÜFcçr{( ‰Ô# ¡·œÐTáá£ÇÄ£ŠÒ+sC€fý¯:œÈÀÅå"¼}ý0!-ß W¿ó­?Üá.€6âÜéðøžh÷¾#Ñr€ÿ<’—¸Ÿz§ŸÉ*4z¤ HQ QâüŠšÀ0ª±K@­«ðdÕêÐÞ ÄozÅ)ÏhJ#`¤¿õqó0-ެt‘|–>«˜ Òv`›_xêå0œúúÓ3^ÀÝSòùÀäUàý*»|wÉøä;ïÜ“ þ¯%ü¨œö$WÑ,x‚PSÜ5' z!&…º§ ÄÛ]¾.<0(,÷¤ÅEtùEŽ€Ìúµõþ*'Š`²—¾ŠòB{Aù>¡“ÿ›½ Æq»ìYô1€†¬þŒ×i á˜X{†4õw¼½¼äêÚ·frrÏÖ®ñþN†w‰ /¸DØX¦†umÇÕÈÁHLÁ¿h„L¬›÷^8fÆyä†ÏÀÜÿ‹ö•û`ÑÙù¬aõ.Xé]œs¾Y ÖuçÈ‚VÁölþŤ(8 ßZŸý8ë _-~‡ƒB•c¡ÇáÐð@Á—vàܱöáÝâKpå9žL ‡5̤ui‡ø6Åêo~MìëAÏð 5”Åf¶5áp ËgbñƒéÁÿ_ÿ;ÿŽG 1œÅåKÿâÁÿè¿Îÿª3ÅÕ—þ­ÿ««3ý+‹«/Åò°¾nç°ïÿ8\Åò=Ûý\XŸ1®ÿp»¿øŸúRLbøz„Oñ¹^óû‡S‚—û”ÿՇʓÕ^¢œÿ[^ÔŽýLjx¬ú§¾Wá|\}éféÛäõ$ !®dˆV2ý̃=¤Ã@ñÀtTþ·N£&È:xŒ€öDVâôí5 ç³0¤”Ff¼ ‹vúmøññ× ½âí†\N*p% žå^-…Ÿfàó‰}`’w·N=÷ˆ!KÛmT Š·2jUâÆn’W5ª@Òæ@rVXHKˆ‡] ½²’ !” ïr{¤´ùêNþÆö\~“×u@|ÇOåê€úý3 ãVASfî®ê¬hÍ«í)Mæ Ív‰ ×~ƒg i ñ ¦ò\²ÁÈÔCwOøu0©@U34”`V=Üúü9XÜûpV',Gñ´¯xäu΋Kó#ó`›Py2úx“USÈRŽøÇ~ îûNWžV™ƒs­ëÍïàÊäðlŸz,\– –yaòçGhÌë_ky‹Ñ<4*0#™ˆ‡*å Ø –lp<{ß1|Äò*®áúŸþò ÿîÇÀp‹3ì½Çü†/”ü Ó¨‡oß'Ê]´Ø‹ËSbõe8ßÀCpykOäŽ>ÄÎíz —pçö}œ¿8y7›ãþ#\_®éo1‰Ûæ4WËNß"VoíÀã?y~™ñÏ>Äòûkp ïÿôЄ]R…aQm@ô˜÷nƒû% ‘¸ú´u•H‡OÛíWÁÕÅP¾Mà õÚÔïÈõÚ£#»Ãä?]bì–…ÐGIšéŽC‚÷§ï CÀ˜xvãYØW`º¹xæ0a<0_£Ò=šë8ëýòäNÿá…ÆÑe4Óx¸äñC4GðŒÏÌ Í«·SH‹Ñ …Ð )4ŠhñÇ‹•ׂÐòa‚ÈR1[´Êv®Bqš­ïK"»¦6ÓÊ|Lð ÇÒ[LðÏ9ø „µ\µÑÎçèdÒY«à ¡}’U4ʤïîu$°¹çÇJK¡»@©:fŸ5É´“ /…ò2½<ÏO»g4å¸:q\âAC×EÀ±QpÇ߸¥ÛɳX€×Û¿…2Jv7:>&]l~àéØ˜¸"O·«K¤äÚ#—Aø»VHxò}Ó rklúûfovH‚d¹2ÙªHës¬½ø2£äÃ%ß@.ócÁ±¶S žZngFJwô(ZÙý@åõ'›¸§ƒ€‡?Ɩغª )Á‚h„û'&[†ætú¡mÐ>þÑgzï0èâEf“Ë€ÞóæÇ‰-¡`p~I<ïùyØo›åO(ÌÆgžÑWþŽ3ÓŒ›øü6Æ`N,Úëà &R_Ÿ·Ëœ©à,I°3 ,>\ 6å1Ö¹•^`w©F0ÜØ[̽G“ÔìGià0‹žy o¬ü×ÉŽ·@¤9Þ±O5Í^eˆãÄéC,ú¿úŸ0~ú7_Åîc'ί؎CÌ¿Çð«+Ãâë'ÂtÎØæ}¸ß1ý‡Õ•þÕ§ˆóí1ݸí'âúÿòqýˆÛxŒù MÊAÇØ);w|ýí>¿çZàôávý7®¯c5&u^V #ï/#4˜3À>4°ƒ§j³Çîe¢±š„ÐÄ”žÙU±ChJÖùüàùËhúÝÉhn\ÇüSú+?_¡ù¹Žmh©'¶ãeo.o³®äØ©‚6깓Çéß+v¢V¤Ì ÖHWwú;lž6Ò[¯«²|¼/¢Èi_>PN°[|æ·Z6VÇb@wå|æé)`P“ÒnëWƱ%ÿSûù=ùƒ§*ÜÀ&Ù®ž™ô§{ï¹õ’À›8ÅÇ•o|¢û5_ŒþÆ`Çôñ,Ø3¾é¨z„D:ÞÖ;‚H×é½÷ÞVƒÓò¾Æûí 1¸UN'±3ÿâÁØTó‘ç wïˆ~ž -(¼É¨ÌªxÊık\«n zU%ʉõ3€yÙ`J¨oÆkeH8ƒf èùÉëA›óTͺN/è|§#ÿ  ÷xß Ñ}0]¦é`ò‚ý×§Ó‹?%€q*Õâþs`Ú1GB|å4˜3κ:É€ÅÑÛ¤ŠÖ¬`ù•(èXëάÖÌûM¯á•ßzÒî‘óû‡6°?yúÞ“öp<úl|u!œÂi‚·Œ2ÁÙëéË1%apI­³/3“·F5›Ì®P஫÷›Î}ÄBäWÒo¢¦†!_ÚÊoÄð ûÿª/ý/ßþ¿êKÿžSƒñYnë=\Ýè_~!†—˜ßˆÝÿ¯¾Å¿û1}ˆù†Þâú·ëK1¼ÅòǸú‚íú¬®—oÝö qú«ÏÃú1\íûMÚøçã}W²QGöîʼné¸z=Œ/ãêí¶ûýq¾I­‡§ÖJ úÕûrA®>gBmí&ÿõNô›Þü"QÓD³GL3#å.áú†n­Ÿè‹éGKš[ÄúeÐ*=™° y&Zo˜Ìž™¡C[Ôì”äÔ;}Ñ‘+,7aW,óãÿÃÜ›€Sù}ÃÇ<ÏSæcžçYÜËLdžÉTDRJ¥4¨$ HƒJi044)T”B¥4‹ •©(„Äy}¿î½G¿žßÿ¹ÞëzŸ×u¹ö>Óý¹Ï}Î:k¯½Öç³Ü©çU³gá×£…ÀÞ"©”A¯Ÿ |›ë¾# øEü<|B¿‚ÀkaGÉùâ s¸êŸ)ˆøqÜm\ bµ…©DHp|Ž7] ÒZÕ¾oeu€ª|ËtqsÈ=[­–S<Š¡Œí:ÇAéwÂþm Æëù…£Q 4M‹—W+³‚N¢K­ÕÞ3 ·élŒ@jij~,/Ö£‡fi¾½&ÏÛùŶßs—Ë ÊÆÁbt‡ ³y.@«Ø`÷)°Þ`µÝµ-lÕY¾n«»¡æž/|¹àО¼²'æ8}·»§›. ¢‚̇À5É=o2xëm±4vÏMjê\ú’àÍÍyñSâUð¹ÉÀ™/ýüvË™|u…€]òüGŸCPù Ú§!Óʵv¯Æ<àó_§Ë“u`]¨¹gb¤èò<â0|= "ºòOWÍ">7®›]RøëLÌÔmCëSœ· ý²+tùOô<GÎÌS ‹âP¬WCú=âõ úÞ’:ih_å Q}t3$ö°-£›_.~Ûˆó ¨ åß¶¹ßGÝe=ÿ@~ÿñ> ¹ÏÙ.ýõѹ%¯ïTÐ÷KÈ<²‡Þ=¹LÛæ< ¾©G<ë&úOXÏϵ¤ï›èsíI`®Áù‡¦3Ï ÏKÑó®^º‡|¬‰‰ƒÝŠ—• ZaÈÈI§f¼Âп}ù¯+¬À¼f§jÌ 20»ï4zÿbXM%wÜ|¼8,$Úîd7߀zM¬#pßš×ö¼«xêYÙDàõ¥Àá+—;cù@pñ˵g ›@XiMÃÊïÛA¤áºøÉ]ŸAÌ’kx™¦>Ì:!"¦§’VåﲂAº¸ÿrŽè3IÛQÁ´e)ÈY•W­­ªùÀ3Ûý]AáãÑÅÉ xÿÐk3o(;±À TÕ„Ž ÷)‚ºäÉáMÛAS¿t˜ë°#èò¿>±o)èJ²(œj½ºó»ª/뀾³b=£³ºY,òg+£ËÉW»w]ÎGLÌ`ÚwÈáôIX¤óV³ä;Xln·«¾Äb{ÉÛŸ€•á»Á篂5ñ|yúõ`ãmàñ”Õl„ÃÒ†…À.wö‘úQ`_¯oÁ‘3ÿš_8y¬ÿü\ æ{t}ÕC~p’/ô|s×Ö¸;nŒ`pa« ¸ç‡.Vœ×ç#)ìR à¹ç÷Â=}à•Ú¿ÕžjÞ[© ’¢là³O9uEm#ÞÏñk>S°í ÖlÊòOí×ó>½8â ^Z;ë©ÄçãY2©5Ø ýKdÈÎfÆ3ý$ŠÑmôøÌøÙ'²[t×—’~ ñ ‘ÿB~ ík¢ü!ªƒAþ ÝêLQþ­[Ñ~'Zÿ¢8׳ÍÐWk8ɸ°¬ï*æÑcûGúUˆ_Lê` }T¿Ö5¸µzwæs¬·ÖSû¢²öìâóôû¥~8Nìo¤$Íçâûm£Po&¾-6*é^øg}i3£sÅe\g:6ç íâ«Ìĸ‹º¼Š Aû½ó!uq-00&ñ¬|ð <ö܉Î4¥ç-Ⱥ6ÖÌ©v NÇüÆSoTƒ“K_³/¬xx›¶í ŒÆ:Šæ—©¿¥AðHã²»+Aˆy°ü  Ξ¨q·ÆzS³Î>\»°\$l…_)s僴‰_­•¬ÈjlP4 êõ åÅz¡ __°õámP(>4»*†Ǩ×^³ž•UK"㽃@¥qû[ñGa AãËS¿ZݬµêŸvƒîÜG%«^åÆÝnσ¹1F¹y`øÓÕz&X=ȈVˆgÖ#™GÀvÕ¦ iIÿø¹!<°î8$Ú‰?u:Nbý~ NszzGîåƒKûü¥ûÁµ‹5®þ]=xð._˜º%™žÏßZ’%³¼;ÆcÌÖßPS«Ç‚2à70¿ÎÐø<œZç¡Õ‘v#g회*Æ^Ô¢yXå: ßǰ¯Ce—¤ÌÅë:l3öAÿVßößâÙyü™v‡ô(?Fë>œ<¹Lé-ðã|΢8‘Ì+âúS2^Du©>²=<WŒïÇÇ#އãCF®ì’š[Ä Õ5¡ŸV²ÒùÁÉkTô~ÐëÒ ¶;‹--G]pÞÕé|8½‚M‹‰‘øhç÷ÀØ=˜èX+":GþÖ·øôü“Uô±+8ˆõKQ^ße—ÍŠûÄwæàV+–¥XÇt ‰#«kè ñ³èå¯ûúuÄÏȻޔ|btßp™MVñ›3ÍÑSé4Pxä7U¿< ”%÷ÖÜt´ÆŒF¦]Æ}À¼\äÆê¯Õù~žèó½ÀÎ{Bcûºzàôã hîT¹Öݲ[/·Ë³À·ˆ^gúýFIkÊ%nJVºDÑ'RÖùl8/!üѳ2¤Ýñ«gj¤ugW¨>È3ÅIØ »r+d@)±[cÄå.¨ô˜ýiž j#¢~'lgæF÷üôS ëýú¬éó³ [²£—yãb0ˆz6?«‚ŒBóÓû<ƒI2ûl>~0«ùò®\–,Ì_mÙq`#oý¼t7 ƒÕ+íôØ‘½`ãÕ«Ÿ|ô:Øöä­S½²ìéqæ{QÀq>ëúi0Ç.uÿÐ+Kp±ò,j©æWß#G\­2À=5@2{?xtþ~²ª=¼Œ—ÌÙ³÷)xïó)•a_ÎÈŸŒÎ4ð;ºEtBp¸µ;ùùAÔUË2~7Ã<µÂ¥Ç¬!TG¬R^ÂÂMKFöpÿ€ˆøì_÷Œ¬ Ê–ià±]È´È2ølÖ»Bö[^Î#^¹TÈ[‰b?ˆò ËÎŒÿÿpfþpf|ˆëKgêÍ :SÄן¡ßˆÆ™ÏŸù:þ7š?øþã’þÙ3®ŸQ€óÿ¨®´.[ÿë!Ì;|»ë\†L› ?Lê—¾÷ ?§ÁëKÛ{ßúÖ©ý ët$¼nš³”èu¬Ú#2vó7ú:oVf Vß#“²öl"î±÷ñ»ñþ©cpJQfÿ=¬ó6Ñs¬óSP^ÈW^^® ŒÒ:åóÃsÔ'já³T`y#zð×Z!`Ìùn¯ œ.ïÖ÷‹æ æo5€ï§€àª›{A 35‰!ëoˆluŠä7 ±˜«{¼íR@‚Ubħ}¤ŠS å œ*”š[q®äE8œÛçê€Â©ÖZÚ(7Ë '”J¿øíìÐÇ )ÖÉ~ˆ/_·òZЋ[Ä”“Ù )Ê|8Àh™¹¢AÝ0¹.¦®ûì˜k6K”Õ¼‹ÚÍ0ý€Ö[ojëî—µi `+¸´ç¢¿صð\ƒp¨®zd!ÓNϬ¥öŠ‚ ŵɂ¦ ®^o‚ÒëÁ}‚µâ¢p;xÆÔlU-S¯ñ«~j[‚ϱ±¥Û ð[Ö[Ú` q~wEñBPöÁ ³ ;`Þ×ÛYÊ ¼¶îí¬b3ˆ{ Qàë>§§ U 2DÁ>õZ°¸“ø²$àî–Úy8€òÈ_!þÅßêkÐúrf>ÝãBrDv‚ìåÉÑ:ñ°®6ª+ó/ðÒgÄß_¬«ÍS¡!—€ãÀÖsþ_-¯­µÏhsºLðã(.ĺjä÷Õ"žÄ‡ïõc܈]ûžÆÙZã}ÎÞ¢^»%pÒ[ë^io^Øœ…óê_<³¬97‚óˆý]·imØI P­Øw /$kDª|*Ž?ݶw%Õß$~©Îá‘[KŒ•vlôWg¡ó%²LÜ©X”C ‹%€1÷f÷¶[ŠÀ4úò /°nd©º¶‘ ,Ô%W-{ìJüܹáqt~}Èá{ñ—¤÷ª Ñ«xàwÐ_•Ùñyr쎿NÁ¯ßßGŠA8Et;§uˆ„-nî± Ñeú%ûLÙ@ì¼ ÓÚÕÅ ^K”ëlß’coã½Þ鞨bñ2M –¿© ‰à}ÝÌîά= è}7(Ào(µ[ígµU†Ú8‹mo@M·žºAÉ4 .V_ÝÚùýL&  5 v¤êèæ…Fˆ¬Ëý'»Ž-Ô½7á ÊÁhyí%›þl0a”|òLý5˜žröò½æaaw{š'íM+r~Ær øÞê.?5¬˜Ê}îSÁš3c«ÊÚ`#i¤¼Ï>l­™žlz vÁ«¶Wç®û.;‚®Ÿ‡ªÍf¼ãàÄ”Úé¦- s\9ëX,ËÀùØŠóeûÆ`®TÕª'Ÿ—ƒk^×R ª#¸›³,—0ì•·²7ð `³Š6c/ÇM«¶Y.o;&G6æJðqýyâZ\ø. ¾rÈü2¿=ëì¹C\ã@×:ªox‰?Ìúüjc}ª&ѼdžƒÚSÌó›YWƒF´>ý[œˆìr¦Î¬ûFvŠì­ƒ‘CùC4¢øóæÉxųÈa]S”G$ýª—ÃëUr݋ֱHçÉõ(ÖMœ±ßùÖØ>,¼a ½.€?Ð>'½ ¿N|h_¢Ùøm5æG|ÒÉl]ý Û+æ9‘õá8?Qþ T5›ž?œõ–í"«]/q¯]šô÷EX éé¼4à;¢*‹ui¢úLÆi—A„ÿutûF`h.¢¦|ŽÚ¢˜OÌê5{›éòïÀvyvÍáÓãNˆSìšCÀ­wQ³}ï]¼¿ƒùÄUÄá(÷S XñHñ…ñcÈšµûÔ+Õ¹›º‡ÄÊŒì_¢ó†?í{÷Ðõih†í©×A.ÏppBë*(ô><™¸u(±ß+µ4ae÷‹GnqE‚zü¹ÂÐHVPO‹eª:®Ú|£wO2€îüo|!v_'íÓ/YïM?,lˆýö›Œóó÷¿ã†y{ƒéýÖäÑ`^_Rud97XŒ©œotàpnjúj¬Þ‹«®á¨í\Ÿ7ì[ÀvÞ¶èÒó`ÏÁteÂËÜ^úêm,8ÍŽÝaÖ_μ™Š. àB{áZ’n‚ý»ÇåÚЧe:àh¬SÞ3^®ùw\"Á‡é÷>nnðÝx+ã÷)ð[ÉÈ('OnïTc„È ÷.¤ÜâÙ³ç˜[K•þ›öÄ«½I«’nb?…ìb¦ÿ›iw«+Ýö²4\ü^Ÿ¢ç!{Fö†ì ­‘_ü›0ŽëP rœ™ODö…âB´?3S§æþžÑÛÊ*žýÉCD¼%Äÿ@uiλg½¤A× @º¤NÊ;|œ5kAHj:=¯Ÿi&°×÷=¨£í#œÒøgÝ·ÖøË|ùtýÒÒ‹¥~?ˆABcöË bè|.×ÊÄp ™» 1jÝ*ÂÆñ'ñÀiAјÀèËø½0 ˜•Óã%S«€eä¼õ“Åt>pÝ=K5t[lùýAºEºH'JB~iPÿMÌãGu¦ˆ—u/ØlNË*ÜÎ1µeÕŠ  —Æè[` J|û_aÓí7›>¦x‚ÚÁ ¡n÷zÌÃÐáåãq=æªç¹êTÐ/´œ§qó)®îu>úÃŒ;ßîþúLOh—„\…ÙŒIkŽØËýÏfoOÛVºIbžksÀúËê@1I°õßÝ·ãÖñvdÎe—ʇ9‚\/2ZšÀž÷I€ñwpÍßÉqg‰ xí˜ýñþwð\¼Žö’¦Nïo‘¥’•QP~Ù+œ!ÀÔàTF1Í9Ís5[æm7<½kQ&„?ºåZ’áõ~†z©G`¾eåBÎ’$XðxÁÏzýÏU¶!)vn=È\7e`­RíïßT‰—7Å6¼ÇëOd/Ëþ-Dö·{GœIB ¾ž‡ìÕ ý„ûG}é_âEœ'œ1âºÒ¿Õ—’ëm"þ!Ò ŸY_ŠôàfÔ—"~ò¯˜wˆâC¤3…êrfîïð^Ó 4ÉM»áÚ®O{+¹–­ùÅÕû›ÅÁóWü²â®Pð.ûñ<¶i9øžh¿›Üí þ•‰‚Qs& ˆwÿû÷"0ïÀ°g~“>„…Fí4*€£3Çÿ†ù¾ZŠï‡u»£î^wMíßÎUA\ ü»FKÑskÌå“4Þ—@ß×™º3õôgÖ“"{¹NE~åë‘ ï{MjÙIƒ›°¿B¾àû<­Ì+À¬E[?èé,u{3èú܆Œ¯V¿¹@·r}É»ÇiìG#/ðgægo_ÛÏÝ?Y~j!Æ+¡…%@Xîr 5DŒD„_ïÑöÔÞ V}Ûãäk{O=dÌkv$ó´B×»•‚tŸb«Â) ªgÛiîÜr‰eÛ¬3>|SØÅÍß@Ñ£¡DáJ"(µ†VŽÈo•dk‰6ÇS frM}á%+Ðàeßœ? Zìù·ç ž eí%…[@·®Ü¨MOôF¼ûrµGÀ@d³ƒ0j> ^’F'˜‡®å¿CÛŽ[ß­ÁôÝ}ÏㄘŸ*{þ,Ý,Òw&´(Ä‘%#ÖÐVÕýÝM7pþÞæŠïíÑáE`ûË6ç÷A°÷Ýl{þÖcp¸ÿnù,SfprzðpÛE˜Ó~Sú2ë¸ìý–ÚÑ©®q ÷Vqƒ{È«£¦è|ý]꯿oŠÓÉ'Ý7ÀÇúÒ'ÉÙÚà›cùšºô<ø³D×±ß ;[N(^ú Aj½û'ÔaÝÍeÂoåT@ã‰Iúšµ·@2nû©¼cDÏòŒ"Kj'Ž‘ßAõgh=Šâ¼¿õš9ά3ÙïÙ5ÂÁ~‘ÜÂu¦$q¦®®3%u5°ÿ"óôi"u’ñzw¦^ ¹.Åù ¤W…~Èþkèwà:6¤£Ÿšg¨^LtÈík½û̓èüÜݲ2ªè}©Þ‘‰×oÖì¦çý-jæÎ›ŒSxô¨·Œˆ vÍ!«—ƒê'š;/•C««Ï½îÌÆzú#ºÔܵMX—x¬a_{Ñ\bBN@âóÛ0ºÞÔ‹ûóö¤SäÓÜ`/`cUº\V¬iëû¢ßÚ;ç§]·*Yp?'®9 SRš€ÇT¼Pà 4ðm>üyÝ›LpapÉ~‚I,7ü/PAèǪXÚ¹29Bi[\¡böÿÈ™ â;ÝçGæÝÉb;.²UôHióí Þ@ÕÜäFy±äV¼z.ÜòÍî1Y·‚è:mü*#W›»@åçèñÎß ®/á™~4ﱕ;¢ :*R–³@·%|¶Fö ПŒ˜ìµÀàV‡Ûm;0 ˆŸÈ¸&‚s3ÚWóƒé7¦ÏÍÙsÁü«,G›¨X ¸Xí˜×uäãÚ`Õ ·gœsEKö`M°-àIø&ÀöòÎ|ï<Áá–³¾ÀÑvpÚø¼úÙÇap_b,ZxæFÝ÷Ù× n{¼ ËU³[–êF€g©$—ôÅtðÊ‘Îý} |v|<xâ=øIÝ<½%üë×s]û ‘:eç3û@¦‚AïZ]0_od[2ùy¿;0âµë¶?´^œ©ÿ4Óßý·øh=‹ü"ÚBö…ú!/Ò§ùC߉¬ ý£ÿ!ÉGÄu¤3ú\ ?‹ãC¤[ƒâDT÷Ö—H¯õ½ û$â:²ÏÅýhH?ŒëÓ~â/¡ú´Š“7¹–ŸN·xÓ1¢'wÛÆÛ³jqÿ™oܯ’^ž5§çw$(]­ˆ!~Ô&Y‰Y6ÒûJYìxª¡Nü<½8¯zr>â«süël*ñ+a徫•¿‰qÖ.ñuY€Â±Ø˜µf;0ÜZ~Í\E˜RްÏ5Bé§z°þ¸ëók?°?gXRë3œ›Ù}Ë?ÏJÉî÷+O¿Äh$¨ ‚@ùÇ~á2[L)8}„¿¶[6 ?ÑÞU{mÛ«AÜþ›âÍË ùm¼¯ÂdüÓ^u] ¨·ËVî yÁÑÜàhPTeV³ ¼ÊŽÍ{sÊôAµøq,ÃZаÊúÚJs­ïL«Ú‚£@w^‡RÇ¡&ºŽÛQ™-Ñva`´u”Õ¼³L=J¨<àf­†œkS‚…ç@‹À(cyÚɉ{Àª/pÑøî(°IöoS¾5 vZRAÛ2À~œ³¿¸%ûbÛfþ ά”Ÿ7Ša®uƶEßòÀ­p¿¶¯Ö ð8`«vª·¼Xz…· ‚w¦óæØŸàkø^‚ùÂð§,–yèº~”OÌ’>Á|!= ¶zBHàÕö­üAö:+öþè ˆ°­™·ïÕA˜_öFãRLDú.‰—Né×üõŸyó%ãìÙE;ˆWW´ôŠÏÿÂñÙL}ü¿Å…ÿ§ñ!²_tü?ø‡ˆWø·~ÿ‡}.þÆ?ü÷•Bõ9ˆˆxý3ü)Ö-Q_ŠùÃH—fíB±í”?ò(X§ã~?kž#3ñÑè ãî'Ïéúàd}NÒb [¾¶?ëén#Ò©!ùXWªî¡‹Õ­·Ä˜ÈéΛsßã{"Ö]k#hmÏÞ¥pàºp&î}F¬ªÀÂjÔ92ȃùû¨þ†«9™öä$ðžàé»mj üŠ ¿Á¼£íJÑ–XwCÔ ½¶¥àÌZÌU2?ò)ÖuCü)9&>«E}"tÿØQžãAå³ãN¾žËAí‡ZLsÄUЬéùxõ>èt |kN½ ³s.ɃÁ߯±_>Ñ©#.¢¬ú`Ò÷èö÷¹ `ž@a¸#z,uóKÌ^4Õ,Ê›ì‘7`]¾r4ˆÓlSøLT¢sÀ~ÙQíMP ŽÛung‰‡9uŠ<Ùa®yÿÄí2QpgZ¦o½<)‰!1oïƒ× “lfÓ7à³Ìë½âͳà3¢›+ {$Tí¾AP•Û7Ë:ÙQ!ddázû7†ˆ»ƒü’°6¾O2Öwh¼޽\Ÿâ‰á/Ý$>_©?È_œ„ã5Üo̯#{AëÈ™:úÓFþùU”?Dõjh]‰ö?оîãIêÔ >½¨® óH ”ODû¨îì¿Å‹8>Dz¨Hçtó`c{$½ÞÚí÷÷øìùدý¡{‘¼:ºá„*Ñ-yÍíW½ æO|eü¸19†^Í»ëÐ{Ùaº¾/Ù×­/mvöp9þù1Ü‘ÁÜZÆñ *0n¾¨W=” Låëi+vœÆöÁºz×ñïWžÓ×—dþ«Ð¥eäSðÜŒhR[üŒaÂkAÀ@…•Ãëþ ¥=]:@Iá³u¸>Û‚È{íc¹ÉtšÇÌwWœyÇöûæ¤\©Lñ»Ê»'@f/ç•7.@]^šÞ±ºäê W¨=óïòï‡ÚA‰á‡ÒÀðMP¾'™’õ»T·/OŒ¬¼ê„Ŭó|Ú Ñ«ßw©ñh]”ç}W:Y¡|Ê“qâêñÆ¥W@¯VâçI­% ß}huF± Ö…2ºQýø¼3˜8WÞ¼Þž ¦.êUƒÜ`~]û -ð'Xðñ|·ˆìÝñ 7UÀ¹º}<`½úüÕ‰3#`sºíýײ{`Ûã/d¶æ2Ø[ú­§2‡7“ë |Ÿ s.dp­ýš.Ìz62àªá¼\M­ Ü…Î/ÕYß™ƒKÔYªÁ³P5Ù22¼ͧÕä$ƒ ¿z©+7øúXš'ßK£ëd5D³žNƒµGççL}”÷•Óøfµƒø<Þ+pß<çOD2ïêö5ˆ$)шÁ¥ø${â§n÷a›þÄp¡—q—󼯹Ÿ‡“ÒÚ¾Q×d‘eu:@Ð⃖+ž£ó(¢ J˜„ÓÇ5—l—_–Ë?\Þ–¬Ÿ2oYÜöu¬êR'=Ópžœ¦¡/pËôÇP|™€×µL—'Lø#í—¶:ûƒ@ _Y¶sÖÛ¾Áîru›&ˆîõjá΢ë)"}6ѧzÆÕ¸¢œ¤âoïÄqß¼˜7ËŽ«ÉŒdƒÒÀ«èí÷Aåé•Ͻ—ƒÚ³ƒÞ…É Éùk8}ÒÿjÏûj*ñ¾tã“—ÈøƒÞ™Eª`0;XskB#Ö}ÖIÜÐE×oCý.Ì„[ ›ÁÂMW`9å6;¨Yƒ`•¨9žy—¬oÖ¼^ ¶ ÖÚùå€]–ø’´ˆµà ¥±ÛpÎ08¾àúÁù< æma—œÖZvv>p½¶0U0}?¸J Ãê ð4“®r7¯­«ê"ÍÀ»/Ó4ʉ |Wë ©s€ÿ¬ÇÁ·Õ”(fF …‘F¡Px&ÿ™'ÿ—Oþ3’ÿÖS3ýó¸ù8?ùŽnëjéE¢¹ž– š³˜h離lú¦ZÆFè&ƒšè£‰š¢‰š£‰ š˜¢‰9aÔÓÅ3=<ÓÇ3<3Ä3#<3Æ3<3Å3Œ¡1ô1†>ÆÐÇúCcèc }Œ¡1ô1†Æ0ÀÃc` Œa€1 0†Æ0À†Ãcb CŒaˆ1 1†!Æ0ĆÃca #Œa„1Œ0†Æ0ÂFÃca #ŒaŒ1Œ1†1Æ0ÆÆÃcc cŒaŒ1Œ1† Æ0Á&Ãc˜` Œa‚1L0† Æ0Á¦Ãc˜b SŒaŠ1L1†)Æ0ŦÃc˜a 3Œa†1Ì0†Æ0Ãfà c˜a 3„Á¤§«KŸêѧúô©}jHŸѧÆô© }jJŸþƒöÏÏ#y󒈸(rκ0м5ý,«c­DOa™±`1ù kt =3lÿÌõ¦Íõ§Í ¦Í §Í¦Í§ÍM¦ÍM§ÍÍès£i¸FÓp¦áMÃ5š†k4 ×h®Ñ4\£i¸FÓp§áOÃ5ž†k< ×x®ñ4\ãi¸ÆÓp§áOÃ5™†k2 ×d®É4\“i¸&ÓpM¦ášLÃ5™†k2 ×t®é4\Ói¸¦ÓpM§ášNÃ5†k: ×t®é4\³i¸fÓpͦášMÃ5›†k6 ×l®Ù4\³i¸Ø™õu±)þ3×›6ן67˜67œróÌÿ¸ùlƒ™|Œ)"ÙÓŠ¨5è  ÐÝÌIQ S¯fùçÕ¼ÿÜgE™úƒþÿu´â'ø÷ù¬Óÿ¼n9eÆë¬¤ÈÑ‚_Miä¸Gvj<@Þ>²cjÌ'ï/ÝôïxŠ|ülÑÔí äóŠLݾL>ÿJÆ¿§nu•|]ÉÂ;ÿŽ×É×—ZO=^Fç¦äÔã·ÈãÝþ9õx%yÜ;1áÿŽwÉãß}÷ïXEâܳy?5’x÷/Qÿ«IÜé©××øµ;§^ÿ€<ÃS¯HžÏ£¨©×בçU÷bêõÉó{S¯Bžgý…©×?%Ï÷ÿÔHž÷³{—þŸ“çÿ|­ÇÔH¾ºýS#ù~^|Úýïø’|_/éNäû{åólj$ßçk®eS#ù~_WMá7ï»aÍ~ùþßèLá¿!¯Ã›Î)üFòz4™Âo$¯K“÷~y}ÞrLá¿%¯ÓÛ;SøÍäõj^=…ßL^·­)üòúµtLá·’×±õð~+y=Û<§ðÛÈëúŽ} ÿy}ßULá¿'¯ó{óÒ©‘¼Þï°Oäuÿ=àßñyý?Ì=35’ŸÃ‡Ó#S#ùy´39MäçÒš35’ŸO{Y÷ÔH~NEL§Fòóú¸l;eêo‚üÿ'(°¦Ñ>®¦Ñ::i´./íS%Ö£I£}>L£}a£Ñ¾®¤Ñ¾} ÑúÝi´ï·h´5mðö“™F^N£´Ñh¿\h´±2m\™F›Ø7Á°tÒM7Ošÿä9³”L†  ûäyqŒS(\± wãäÏ‚=…Âw…B˜¼‚™“?¿(‘h Etò 1ˆB/¦P$'%¤&? éa Ev2È >§Pä Eá…¢$N¡(§Q(*ŠZ…¢^O¡hΦP´&¯£Ž(…¢»…BÑ›¼V!Ša…òç29I¡˜ R(擟Ù쯊e…BÔNýf±Mþf13Q(‘´©ÛZÿü†‘׊•BÑþç6ãÔ\‡œëN»OoÚ\ÚÜ`ÚÜpÚÜhÚܘœk¡ûÆÈñŸßNô›9í³Ã¡ßÍ™÷ÿå5ÿÃs3ù¿tn¦ÓpÍþÇç`E¡ü_Å™qÿÿ{öÿÞÏ_pþÃû™g> c6í} نŴçXþÅNþç6@__Ì<—ÿùwõ?cbÊÖ ´:b L­}8ÿYIëèêéü»ÆšøŸÞ!¿V´ÿ¾Ëÿô)Z£×OçTêäÊš±4:juaBü’?ñÏÞúZÎ< ãÊøÿv+Êÿö¬ÿ,DWpþ§G¨ÿÄ…æÿõø~ѧ…:_þ÷ãðÿoÃðŸNQˆ±2Š¿µteBÒ_®¨Åÿë+jñrEÿ㙲Qÿ ¨ÿûUúß›úo(þßCýï§8ÍDågš(GÂäu]1ÿß}iOÄO`Y³2jÉ2ŽÏêØ¿ë*ŠÕ±5ä˜GŽÈqê¢[ŸEŽºähHަähAŽVähGŽNä8—=¦¿½?NîŸs›ŒÙ'OŽ3²* 2DðÁ¼ }éúÇÝ ·p^eÞÎà:UYÞ$©èˆÒŽZ$]JeÕI^ ”ŠD© DÝ@ÏGÒN¨d§&î_zx£—.…$È’D½Ç¥­ó¾YÂF—Ê!¥np©+JQ’H\ºJJ,¾¿¹{£’®ñ!çÚþq‡@,…ƒR,üן$ºl—Ë—žÅÔýždŠ®Ûã•DoÔÜÙsbÊèÒ‹ˆìÐeÁ5èF§f ÖPBNGܾúcI,…$qJŽÿú¥(†ÔŠ#0Нˆ·h§·ˆ °Žž?"%S9TÍæW—u÷ÄE·eÀ³qÞÑ5ú±À×|Ÿ9>Nt&NÁ4S}¥M?@è§Å™Dø"›7r[tS@Lc8ﮘÌ=ns]HüØÄýëÌu–Ê8ó<úÈ&³ðÞ0ì2ãÀ!J-þ Ìpbì5¹uƒæÓælî—謶ǭŠàªI·1ï"¡Ã,౨0F&><3*J½)à5W­ìaX7x[5¯íîc_õSYðÕhúxéÌfð“ýIhݹþn#õÁâ©Îáo¿  k«}׬ MmyÓ AÇu ª˜!8ÎÓÿ¾;æi>Í‘3óÞ‰^ÚöN¥šYŠ(¤½ ùr4è1|EÑû—O1š;ŸaXî¢Ì4OÚÂòPœrÖ]hê§Kö›ÊPßswDœª¬"°V¬¿¹þ".•ãâYû*æðl‘j+N9 üb&¾%ûõÙË슿%¸%^¬½ qy‰ vÔ>…A² 4‰S,-!£ 3K({¬)ôDúÖü OýJ‡î~S*¦¯/í¹QÆ”îT0Íá_ôꘘ·žL<¶ûXºuçÇ„6ô\ßyßC°öÛ௽…lêêi‚~‡ÀÎÇ:Ý8¤ìû¬å«”€ã‘ºó£’0gž}n²“'¸h~zDº\ù†7]õ^îLÒ²²×΀ǵá³örà™ñ±Ó;¤¼Ž.®‰¹Þ×çÛÞûp|^¯¾à”Õ ¾cjçÓ?œåÏTo„ÿ½¿+~}†À̰8 ýƒTó3òò¾=0i"[…!¶‹ü]OòBh*[®“³#„U›S³³s <å", )À¾–{Š/n€h"ð}î_°áް)zµ¤—Ýaxè6[3 {D%š¨0òS¨Ä¥ÚQꕚΔF-*PI(*½A%3¨Dµ ElHJµ¨Ç­_H?3p³ ¥fŽ1¸ž¯äpÀebHæ‹ß7ÄPyìêѶ4\B†¨#&;Fù}NbjÅ/j[”,1fà+¿ ôñ{žzÄþÌ<Ô .U¡½ÛS+SòQk4Æ¢€‰àK15‰ùãšo'üUÂêê×½ƒÀ¶/P¬e%7pè›}ó:5œ}iOšìB€».¸‡¿W.õš_›îÙ ‘â'änráV£BB¨r œ¹4ûL¹.å=+Ô²)ü5ˆ]©Ü+R ³Ê(\®x)cŒ€æ(®(ÿ’'?½)ÛëRy¦ëÝU@º`‡õoÝ s=ɲÄyȶ±ÏãÕjæÑÂ{ÂAΔ—bfüKƒË_9m× ©ÝAB7mA1Žß³*Ý”– o>vf”3$û•ü™¢Ë¬@mÃ5ÿüSÖ aqˆqßéÐRIeøåò…Nå(Ï>-¼Uôº² ^8?ÓƒúRÝ`r~Õ«¥Áèx0O•¾:¿;êÁ, ¦’ŸBôóÅÀ,À¬ý’Ób0?"pÛ7&fcÎØ_ –¾ýE5ía@¼ý0êYÇVÖOÄނµÿR•°Ñ2ðÚð9 l]1nÔ»ºù,® `ÿõae“îFp\{7±/äÌ1Þ:çÅrspQÞì•Âs\=×ܱà¼îuG^—$ƒçì×>_jµÀ«jûV — ðIúj»žé5½”.gwlsF, íîa›Ð‚O[ã ÛU |ߺÑ6;.˜¿2m×ûŠÏyúÂÒ5«m`×Ùùo}Ù’!¼óù8È葾/QŒâ³„„äÎ\)@ e™Lz±ÄE#ëFþ}ý‰˜¤œP *õÄëP$1J¶&D-d¿BëH$á„J]º£RëVgÕÛspý©PH»ç'CÛþ´#±ú.€•ßÌÕ*‹øUèQ'Ÿ_Œ×mHŠÅ}ˆ‚J®5•¶!‰'DéE#*«x¨ªxâôJ:åQ:H{ET,éM–¶bªÅŒu(¢B¡ßtI-"Ê$*ñFTà?Ö¯dü‹¨ý˜Z‰Zx̤nÒŒ˜RµØçá÷óÍtj?¹Žè]) 3z\›øæØgíÜÆJôŸ?yðÊäïDØg…‹¬}˜BõÓÔãäèÀz+Æ‹uc¯(RĸÁòkwš´!)GßbWÜ‚‘I÷|`éÅx`a\ç&À'¬¿Ì$ÛÎþ=“?ëó€Ëþô=‰ÂÀsèØá"†<àg^Í¡.ù‹ê-êAðâs±yœ7AX‹íÍ…"•c÷•9@,òÓùÛ+&ןjÇò $%vç GøƒôÜHʺøN½ÜâÌ­Ô rr‹ÜêÇéq"÷Ó’Û AéþÏR¯… r1`ï÷Ðm v³å¡f”hŒ‰ÊðÙ Ú ¾t¶F]ÝÍòç·VÊÑ[{{GÖEoü†×ynÕnãê‘Ó`º¯)½Î- ÌëµL¿ ¥½æµ,¯ €¾Ê^Y–9`¼9Vñ‰!Ø<ŠÜrGŽìàé=ÙG«ÀþVëÁ”ÇÁÑ÷aò­šWà4ؽEÜ¡œËF”'ÞDÂÜÜEî. «ÁíBoï#-ðêÉ œÏ’sÖ^«¯l®÷¿ ÁûN(õÅø=ðùõÛ3ˆã+øY»-¬™Œ wnëû>–®3 j€ ›¥ Y ‹!øÔ{”Ñ“"bpðôû`Í ›Íé8¹v+?¼. áoú_÷‰qCÄù–·á[ò`³”WK 'x< ¾ïÞ’¡G_j?uÎÄÔÙÓÀf×ÃØÁÁ!çDOŠ×!ÞXZ5ð|Q+é¬þywö:7ÁY] WÝTaéž.k9@4ÄrÂçÆ^˜Õuò~¤]H^µô¤ñ5È<÷ºØøYäD½K zæ`?©ø®eö'ioPáÎØYG¹ j‰‡$¥ø@S¦)vE h²®‰‹ä=.ï+å\ _vžû”J+îQ1­8 ÆYaaš€é%×ëBNƒ9Í>@Ön X®j=š~Ϭ„>”sŽ<ë´ ¿îš`+fë±Ñs2^¼+³9Æù 8ìº÷8ë88-‘~ØZú œ“OtdØso´Ü(ÍwÞ¤øÜäð¨Üµ/ñgxÒbvž}iÞÎ¥ÚÖ™¸EÏnËð_Ûpná® ººÖ^yénÁ×#7+—BˆŒQ’Z½ „îͨJÈÞð§^ ÝéïZ™)ç³¼…°ËÝå"k@¯á͘ò–àÛVv’øüüèÛ¬XâùäŸÐ>)ò“x¿†´WTâ[¥‘¥ÞˆŒ¤¡P<‰(…ˆâ‹öOB”|\²ýP˜qeµ?Ž#Ñ>&–&¥*ð> [Çj¯w~ÄgÆvÇž¯:D¯äÊ Òý÷±”nqh«òöcD ÑŸv‰£Ïì&}ÿ$WôcDç~bàQ|¸Eü>bHl+såá£ÄÏ8µM?îÃO6Ôu¦K Î,ñÎ2ðŒ~ R›žA¥]*IÓRˆlgïqýV ŽÃƒsêµ×¹‡¡i§€§)*ÞwU}?% >hL~/8~yÓ[ºDÍ–7R>ÏÑïdÝœ¯Á¬ÙÛ‚Ø:€ø.= øàË·Fê H™ˆÜçTíé=®µ›†ƒÌgÞí‡4Je´4 4êSÞÚJãu ·U/>õx<È;YçÙÅN€‚ðJõÃüE Ð1¾N»á((Þ-²Ýt”®,ö¾71 Ê·-„r¨n Òá)áÊ jò¬†‹7ã–Šï^†«?­ù^÷Ïsìí/ ÕÉÌÇ@ŒC#—Ì=±ž_d}Î#,ý5` àSõ eYËgƒÑš” 'îÉuè‹£ ½¶ÊiòƒËû÷ó“Ô=Æ«Ïep`Z®?X9L_ã6"n;ÂXw—Çé¸:pªb®¨)'ßîF©ûÁù¾ö\-†vp fmÛ¢ ¬QEA’\à•eÖß^Ï.„mÔY >n­=œÀ·Q­<ëÄð_pxï® ít; ¬’ ÊKà»_ÄC†ŽAˆJ»˜mõ-Ý_¾e³Î›¿Úá69~Ç ¡)ànÜ»íÁeгø-#p 8sd•¬½ƒ[s¢õÙKavŽÚÛ58Ï€ö5Ѻù'ÔBQŠ¥ µ®@ÔBÔÊ S¬H 0ò?Èï`ê=’òŸU¿Z“è§K/2Þ+—ZG£·l#[ÙˆKZ¿Ÿ/Ÿ.¡(ƒ¨µŒÓQËæk?ˆ®ø’ÚgóŸ&ØÊ?=6?Xµ)D/wñ刅˜*…%Üî¾ä°7Üúë6×݈áÃÐ+=J¯zÁx'¯Áâ$0­ŽL6x`æÒ˜ôÃÀ¢IÌfÔ dâ†dê:°æŠ¨_¯[6Ûv°v€ÅI÷KýÕà ¬ör=ã pR>Ç»}v8«ÔẆ¹‘1U“ëP7‹²ì«Â;ÁC¸é Ïx|Þü‘šž žÝ¥ö+‚ב{UKGÁ‡]Z…ÇD |Õw34k¿À†ÎI^à¿ÿé1óDKx'3ð1‚ŒN_«©¢Að¡ue܉vÂÙ¾Aòé|Ýq•o¿ „s޶îîX9->¼µ½vßKØÌ¦ejûö.¸xŽ'Û„]AKÁÒŠ$‚èKøÌµÀ­KR ûCñòo˜šLR‘”ÍßZ£¢I\a œ™TGò¸Hº‘¸ˆúˆüíÌ} tžÈ/cj$)±Š©ý¤?Æñ$’€C$CR#‘t’AöøMîÒGÛäo‚ݵa½/Á¡¯áæ‘0Ç:)N’š.‡´*_ä뻣ç¾ihûXyóíÿkv˜µˆ‰3¼õ6øø¹ô‘—9 s¿÷óõÄïÇtÎÊø?´27_Û‚%%оj)ŠöEÐ÷í§ »Àþq†t"²´¿‚¾çσBóx‡Ÿãï1Ê«cCæÍ;¶¿Žóqö">hÝ Ãï‘ß×^¿m^a)ïÏׄ·¦×0âI߉WçyÃg„ãÂo¿N~\\ØO!êï÷vY_ÎüÄyAáeYó·CYBu{k±„ðˆkvËUb´sGWÉÅÍØ{*_ÍJÐ4×µ¦Xº”á”lä­ƒÀxç§ç¯-=ÀôS‚uoßa`™ãg¾i/°V™”Tÿì!²ŽEbSBåò¢Tàúf6R <ì=nT|ç¾òˆ|þ'{ª%ìêAॾèÎó ø…·§áàg–bUÑÕ?"1íõL£Âô¶rñ#«%@¼¾•Ñå;H.yÅóä H÷³`[2û†–ÅVUÒÐgáãcoøö˜ äÕ¼ÛFrzA!eøZ·=(I›+DIƒrÓñ¬žò“ ºÇeÀ1ǾδO-¬,Á~Pë+s·iÉ{ÐqÖŽlz º¶·Ù;žoÝú»ŒJ—¼AÏâmQ Ð;–‘âo ú"}>ëÙ@¿6¾Òc™ Ü^4wp÷-)K'FÄ;òöÍ_BŒ–qù>ØFŒEÛsl­ï#ÆÍÛ›T|šAOÓ¬ œÜZåÓ·î¨Ë&Ô,¤Ç•²b‹ÕGwKBˆ—FûE`}/Êgm6ì+2æo—J Sú…®$z›™¯ÀÅgµ³´_•Zñøñ9÷ma½,‚gT_…ـЗG9¯>‚ˆ®$%[ùˆf”>RX ³Ëö½7¡·Æp»è[rdîð}yúZ ¨9ú5'Ç@î¨{Édz Çö¶êj"–,VœµÆ´ª(”„Ö±”Ìöe½ð3ñ×AeiÖïᣜ zãüź:PgøYí”r4BÜ£eµ<@óY›v³¢"hG}yõФtã’5V\r¦K‰¯.ÜÏ* Fgsºb‚É‚Ëõ6`Vz„IsàX­ôÛµ ˆý§=3—/+ç€ÔœÁZùhË¢ë`ÕT²¨+lr8¯ö.î[‰ fyrÁvß½bo­c`ÇtËŽ/éØ-fÝ6WÍìÍb8Zëör±O¬3Á~¹~ê£Á¾<Í•:¹Îµ÷ã¿£eæj—wPÁaù›°A}+p8áÝ}_ ^è8k )ù×g=(€¹ÍkUúÜ@ „·ZCL›øò`ˆùÛì8ïŽö=Q=Šëý õ#òwȾ=¢VßhŠ%SÉu&²;´ïƒêpp `2îCñÚÌ<ß›Ýg |q½²3$õìÅ(ÿˆnϬ‡ÃöJîá:8r}€¤7p p$þ²(ûã§&z‹7”—@ñ!é·Q=Bßà–ÃË«žßçS/~>L Ìoø¢–ÅJ ¹œ¹óã)1Ã9¦.\‡íò·}—lÒSl –£òT{`âZÊ™¦Ú ÌwJ àÖÞv_úJ€ëxFË6 ~à¥^ŠótP¢Ki<]Rf$:BÞoœÞì•áaÿKUCs@´òõþS¬¸.G¢5uEþ‹· M ÷ ¼V ²qª*4v ~tb¨oyÑáa ›9(D½³ ðt¤çïO_s×bتEÏÖ7þ²õ>¯sãÐrØ{Î=tîgv¸OÚ•só×#+@?ú´ûiÍa0h`+ñ<F«z?X¶±€‰¾æq ˜1É5qò—yó:1Ç|°4ïêÉçÙ.ÞËý¬—ÅV$,ˆ›Æ]~›¥€vÆÚÙ£`¿“òp¾@+8 Ÿ`uä§ÇŸ,мÎ7 ÛBÒan•ënÁ…àV±¸\Æ´ <î–_‚à©ê<ëþ²2ðbYT«E¯öÅC¾à}ºö.„QÁ'fG#SÕ8øÊ,N‘Þо9”Ä#ëÁ/æ KrL øV –‰ÿ¶Bk¯z`{Õ~4Ï ’ÕŽÕ]ƒ€Þ­]QVް^Ÿy£(S(ØÅœg1îîrH’°xEt=ùÝuâ ½.ù©™yB”Dþ ÙŠÿäÍL)$i5SIà ûDùI´¿úÇHÆhDR>8߈â™u­ä¾ŽQ^…ô÷8ÏHÖü5>$ëˆP^µ’Bû:XZ•\_£RïÅc÷Añáº^ ÏÄÜ*K\½ax/p·REù#ìgÏ)I½ÑD|Wó6Þ¬°çS°´#Gº¦+1œÛãUfâBŒn(ýÕúÿ°÷çÑ\ýÑÿ(…¤A†4*C$J†Dòz*!ó”TJf%C”)³’È,‘ ™¢A†"ó,•™Œ™“Ò¤A‘në÷sÎë.ïÛú|×½kÝûÏ·µ^ë¹N¯cXu¯ýØýØ{W“£ý'µ,K»Et9“%Û·‘y"íÛ‰awi7,Iÿ3Í;* ú¸¡l¾Ã{°ôÅ~uëÍnäÕ•ë„×ù¦9.Ë‚õÕ.°ì§HImgÛiêêÅyý•X ¾y7ûéŒeÀu–-"çf xÌË’Y°5HßÊTÁüúO»žÌ`{À©ìMa'±Ãk_EºYvé¾ê”g? ‘ÙÑ:ÏpNˆUЗÜM¸…Ý]gwÝ“©„¤^=û¾ÏñØûù}e¦¯.dœ¾T^èá>¤NÒ.2ƒœnšá"®,ÈM`Q €ýù9!SÇ#pðò»¼|jýbûußÝZòPùÌÅt‹¶ªÉ<¬ëd ŽÀ™úÍÐ0”6çå‡&—ä½$Á)h¶FrÝ’>-×wgšDi ½þÀ³Ÿs7 }?ðN÷ÒNèHö°•½©ƒNA`㽇С©o ­îm÷þeú УSÞ|J\ó_ù!hlíLKª6RºÊ¬DïiÒú$Á ‰ç”x~‰zÉWÔ÷ ý”8‰xJè?DXMÄ¡îú˜eVJäHà¡#§•œÏ¥“qf|âø„œÄJÊ»â Õôrö¤•ê´Þ¯Êû¯›"×]Ô¦¼¯}ÎlמCž“Îl»n¼¥|(áKö3 ¢æƒó>ru…æfÞ€#sTéã”0l+HúU¬¹:îóÔœ2=êøíÂå—pN^Ù e¦ï˜ë¸"éK%GCåšÜ\Ì¢šÚ›¿U^;RŸû Çϼ­ü:£ÇL5ýÔ¼Pþ[ÕΚ=`­•éèå Ö¯Þ²HѨÁ*ŽÒ5·-Á\tÇñöàe¬®ÚRö0,=lÏs7ƒÎ yßm€ýPÞaKûZÒ‡JðJºÁàj|ÆæFlÜ×lW®h ÎØ„—ww^Ç–µZV{tT±åíJ–OSàêâíË—wSÆ’¾hOðZ])ð°ËF׸`ð÷Ù…»«‹B AZþìùÜ0Ãw‡BjAÆ<ÕØá*Û8.‰m¿né±bûÜpk„'óíÞ}¶‚ˆ¹ÁaŽqˆ.ñ©¿Æõ¢q˜>œ‡˜B@Æmsˆ/ÿ:W\Ûñ_¯ç"oTCÂʼ•½{fú²i–JA*ïlÙ¯ÀoŽ Ü ?úCv@D‘AÄûȾÞà³ÜÈÍ)fx]ë«mҞد4,q=–öÿ³üÐ?zÏ´Xb9ÔLYw˜‹?dž~Å´ï×OSf“í½ú8þw~ø¿óÃÿþ¿Úýê*9ª é?Ž‹ßê%ó÷<:ÛÇI?Á÷ý“ð£‘x›¯Wy 9Úqž‡£‰“¨_<”¨'~b¹zm^G%G§«×ˆ¾ÄyHœÄç;‚/y&é7ïû ü d¾8MÔ'ÉŸ7Ÿov\êPý6DÖ%Þ@ôK¾ÂÏ6$Ä—¿´Æœ::|~õáó™HÑØrà²=e2*¡„K׎òÉC¥ô‹Ÿ<勹úÝÝ.Ô~EÝÏtÆ”i¯ô{»–Q~}º^wä”(•gÊØ ŨùƒæIŠóf,Ö}›°¿êè=ÛÐt€A*·Vèb }½Ã Çÿ`ù’<}>·_XY¢{HÜm ÌÍ:ubktÀ¢Ï¡žÀGÅáígG—WՀùþ(»Ñ7¬»b>;ÛÙ… ×EwÑ»€Ó¼/#;ü¶Dªÿ2© ÷R—˯hÇoà^¯ÎleñŠße3K™š12j-ÿÛ‹Úu.,³„Ð\æh•ð:;©Ôʆ¥C„“n%ëïˆÞ³y½â/Ä$’¢÷@‚ï`ŒÌûe¤¹#¿ßÝ{å ù`ŸêÆ\k5/Ph–]¬¶„ìŦAꉻûìü„$ä§îÆ/ׯþóm³”{Ù8È­°b økž?O”‹=“ÿ@¥2—ùõ9.¨^ü=­D; uÎü»sÇ3¡aõ~Òé/]Ÿá¹5EšÍ¯‹3vÉ@ËCâý^‡ýÐæ~tùû>h?UÔ©Ø ¥ßmQ"éÐid™Í}wºÊIÎ'B·xíäè«}Ðã{ÚñÐ. zW–_` ­ÄÙ«|åð¯îxà Q ¦B5 c3Ê0k«ùÚ~d ¡Ãx[X$ò=b$øÂ•Œ Gþx$òF"¯$òM²~8gÿY?œ÷ç|=ä_'Á‡Ézã¿êŒÿÁJäó¾<¢ŽóŸâ û™ÿµ’cÁêa2o$ô"þ&òÐÖº'ë4o V~gzßF] GôŸþqö©wëË–P¦âií›·‘õ•'“øCÖP~&=W{û“2»åk@àÊ\Ûjz†T€pƒÐNÍ4,¾Ã]¶}—èäòÊ o€íXbÃÒ<²o‹‰¼òÈLŒ—(?˜Ó¾>ÍUJ‹J茟0ØÖ.òv*ÆšUææ§°V”¾'ÚD 룾¾*4œÂ&ËuŸ$wV`‹ÙNJG—)¸OkwkøÞdïÎF&5ðs ÓÿÌäƒÀ+÷^ÚSz1x÷TS$v-~ëwõ¨6D:´ óvBl¨çÚ¯\Eìþ-³çs$Ù®y5{D`¯ÂãÕíž Ø÷¢í„ª[PËŸÁºü7äܬ„£E|¡°¹¾=ÐŽûÚêîsâ`P¤)§Å3R»#zÊK<®…AåÁJ×´®ïPU)u-z µWš×6†BC xÚcRý›s%oBóJSžE$´ºå¹Go@«¬þ÷Élh«±Ò >à‡vs/Eié'èÖK¬Ùt:Åż:½#ðˆØtÃrû&yÜ·Ü…uÅ]¡›RÎQæTjëx’+é‰~_¢_—Ð=ÖõÉÝ V|”ˆ—ä*yü”¿{–p-¬ƒ|¾‰~ rÅÌ‚UÄÄ|¢¿žè§”¥š)o¡ m2ŸŒ[¼˜2ô+&嫸,u…ð||ôÁÈú±1²/‚X©MôY~ë çÅÛëêÊȾÇÉÃå)¼‹x)—.s70w£|ºðÆ¥ýÿ´ê"chÿºô”©qFÝŠÐR÷ü>tÿg±Ÿ7e:ïÚ†HÓu”_ ·¦Žÿc+µîÓ¤üqª»©ª^EöKѺ=¸yÓ,¡Q,qV: :¿Å1[_:èÒõãtÞX™•ÿAm–G½þš±¶ºÑú™ÑÍ«uX}éü]Û°DªÆu7-kAƒØŸVêªRe¥Z™×û©«J%‚ŸÆ,Ög§Ž_ãçÂÆ­Ã4ÒáØ”p›%çÌl¦Ü«œRü€-!ž*ŸsÀu·E[c¸?Ýâ.z£ ^Õ÷ky|ÃÖl;÷‡—mÀ÷ó“ˆ4Ÿ¶ñm™[åmÕ”„&lO^U0v™ôwïxG—r×û3v­v>iª^ 图Sc[tw«g4Ä›‚cS'k!q´AÙ=x-ö•pµ½…dÂWݬ5ØË;dþ#eÒ©Ùü‡,ƒ!£î°†ý}6À¥¾ñãÛÈ:°?;Ï9ã²û3;¢ ïó›Û‰ #e¹;öÇ+÷ÓÛâà‡•nŽ%A±*,†f°‡‚i~¢¾JJÙfQS£PšöAàÐ?ÿ¥“þstø‚•Ržÿ52ü_«.ÈQáÄJ"^Íó\Â'Gêó}Äç ‘W¼µí¡:{K1u…Í‚8FúcˆÕRÜ–f†û(CIœz¶›:HëØÖR¯SÆ¥¤žÓ|«¢LDV=ÑУL®ÎpðK¡|¬â_m5Cík&úñϧ7ŠýÙEùÎ4±Ü‚žòãîSµŒºW”Ÿ6#ÑŠ‹VPóÇXÓÕ4nÔ¾(‰¦‘«±$ž·…šDŒ›’q¾™¯yçË(, øÙ)” ‰eZ‡;ìÕÆ±|ÕÔö+Eräÿ*Ú%êyó¥èÕK¨¸n¬ŒaúL ¶óÝ9Ú¥ýXcÒJ·ÜúÖºñ§dïkÅú›Õl¡®>ظK]¼‘œJ¢#C³Ø<÷ K-°åk\”|O¸™ÅÓ±Ù<Îûèü±u¿ø-¨ƒ?Æ×_ÑÛÅŽÓ–å‚ÐÔãÓ_¬˜!,½õV¢4'vYÍDßœã…WpÜ!uˆr¨ï>> 1³Š}!~ŸVßÉo1$ÌéKÄARñŠ„Ö-_ìÝ_«-*;€}·È6xÊÈ6mOC~È–ôTo`…üÆå³‹¼é ÀÑöìTÝ*Ò¾ÿË· |^8ð’»q0.UB`d·=Z…C,Wøzp¨Ð¶Êqî(”ô\ËzŸÏBiì±ð“)([·Ü/Ùå±ë¯—_{ýå‹tŒ­ Rb,duø),jftµ=°C¹kÏöUÃXúË7õYeäÂMçµC¨¸›Ç‚GþkµÆÂUp%ø)q‘_.ÔO ø!>]†œ1¯Ï~Ñy<z ‘W>YÒGJô/ùß|þH®ú&ü¤D"¡çìX÷aoÊ—ÿÎϘ_ÑAô¥¼ð›8ìëD8Ôg`aOÅçÞÑ+üÒzÔUó>ó7õ™'E‡¨}Oó+j&› Õ6î)¦ÎÍè®Û>*L™ÊjòÊuå[m©Ù[&ʯ'37S~éÔëžÞVK™^AÑ\‹hfÓÞͬ!õÙÅ%Å:Ò@W4ýŠû2Êväó‘zìr)Á©0Êoê\€{7¾^¿¦¼ìö"` ‘å¿ÎyìŠæÏåƒC?~WVg=Ö%¯šÓÉ9ŒyÇT´À9®÷GrD [Fd.ê'ö€û”IåçOgÁóÓøØûƒ)ØZ¸wc{›$ø¢>õ¯ÛCö' ­`,VJÁÎÁ9ë¦(êêá3zŸŠÞ@\wªz½î3ìþæà#̯Œ=­3 Ê/±wõ^ÑË™O±Ï÷ªúI#I`K%eÊβ_™ól{r!ï*³ùÈŒ3ö‹†yFÇ…ƒ q+:›vAqx }ÅäK(Ý?õ˜ýsT,K.u<Ð…êq¹š—BZP—Ö d¾,ë‡bšB3ìîÉŽýk¡•¡â±ÇÚí63w5L »î@‰nŒô. ›•€þ—Î'±e0¸.ƽíÕjtÔð_sÃñ}Ë—]‚Ñ‘øÙëçq²ÃðG]è*˜œ±¾{ÂÄ|¬Â S‰ X,-qbÃJ-J?kçÇøã$>üuÁ…yà¿V/«mˆpÄIÜG|=IÞ:Ÿ'’+£®Š"ê ó+£ˆüõ:ÉÕRDÿÇ|ÞHö'«çy7Ùç8Ï› çý}ÿYIGè¹D#‘/ñ–è#!êŒóý,¤Î3?—‹˜[5ä8(N7Eqa_£Ä=H»[ä½]ª›\ÉAΧš[1 òå;å“­ÏÞ¨ÑZjž¸p…Îüœ_ª'ÂXž§Ì¦š¸bœ¡ê±så[˜=Y@ëäí$-)Å¿ƒÚÞq.»6“¯U ©ô,îñó¢æ2M,Ldq,Ã*'½Û)ŽX­»,i™OX—QŽd=ß ¶V³.†ún¬¹_%uñžÖ޽½/΋ ׌,á ®b¼¯ó¸WYܱg¯ï oýYà§Ï§o]0k¹ƒ™29“–£E áŒó Þ–ylœ1bU¬ý>]QØý¹Y¾tÿ;Hj8oú9§iŽ{±†PT’‹$yC¶Õ Z€å9äÝki‹c¿°å±!…?ÔÕo_}ùÿ|†Ò;šwÓ•‰P2åjÈuƒj¿n˜äŠƒÐ ½mÄÁkÍ‹ïóŠõBëMƒö¹÷FE ½2;¢‚%*Mg8AW–ó~íâL0œ»e‘¨F­r™ŸíÀ¤1Â.{Ø+ÎĺÓÏýÀª4Á£öQŒ`žs¹ÀM –=Ë¿¾×Þ V[þÆíÓµ`Ë6|ç*†5ëßÚнG~ààA 9Ѝ+nÚ£W+»y6oþÙ´{#5?L›‰ßº Üc¯Þ* ÞÃS{’U˜±µr¶]÷¢ øwô&Z¾5Å6hk㥨y"Ûûiy›¬d[þ@(}ÕdäJNìt{ßç¨.‚]†§rÒlxþâjÄm(Ëâ/Ï|›¡–Ûñ:ÕOaj©`ãàò[;Š=Ì»36ÿÄžß¶ï„BK!µ]ê$·ñìMš{g¥Š}7oŸX-à ¬ µ•—×…½±¸ø'{È'²¿ØŽýb¶ c¸pÀìw¶jò8ŠèZôUŠ+öì6íõÄ¡“ü'[Ư@©>nì1¿7TT$BóíhpøÍ²,Œ¸C-aïÍhH÷sD|ƒ¦wf´”V >öÙBKŽ%èdðó…U‡îœ@HïÒ£Ð?gq¿pÉ,Ú¶èPLŽæ¶­ì’÷‚kñÐ!©Ò ìî§ÑÝͼì›ö-®¡ã¦|ùsåldH"é#!ðGÄÒG³€§’yá<ï$p·‡DÞHðZBÇ!pLÄÁ…«‡  ¡guFrâ|݃¬7ÎëK ù-9·Š¨‡Ìë@ÄJSB"ýlós±ˆ¾ b#1‘ȉ§d½q¿^„º!e$Uo·_¿eŒC¨õª…ºêôܾhÁ'R”wÞËO‡Rûï JøÝ¶È. þaFí¿÷6Þ&ºÓô•ÿ¢¬ØùŒ2KqaU8íB™³Ö›´c0Ç"·Ä¢@Žƒ ]Ì££ö‹s7o©M’½ŽOuÞ"M\ÞÈêܪ¯7Ù˜Ÿ÷aÛ×;îE9IãÚž ÕX]â¢Ý!õ,‚¿s|߀mWã‹c`¿¹‘Ûÿ/?#ç.Ž_{:÷Ð ëŸ e”EacDiä×—àø/ùpÉ<WÛ›ƒî“§PG½ïéÙlòûÉý˜ý@ÜJ£ñÁ÷ŽÌ@c¯Õ§J§µÐ^²Ç}EË=èÌnÝ-5=nîg—“e¡ˆ÷»iå%¹±ãMÊ#CXIÌ9pÀ BŠû2s1 dÅ¿}žTÄ‘ a¯>CV±²\Äú‡¼6—)6/T†á5ýF®å8¾ËÇýS9NÌI)†¯ÅIM÷à™€2œâI2 °Â©oÏo‡E%Á¸Fõôœ®Úç5ܤ“ö«#¡ž™0yëõèê7˜Ÿ²ŠßZã‹Þ%:²®Áʺ ììGYœÙz#2SÆ 6Œé»ª£aKc·"s©8ì–¦l>.{aï!'‰›p°¾²:ß5ÂÁ§&nÀoZBbò¬ ®ˆ÷–¤¾gÂÕÚÍù_Ô èöÑß炵¢[”øf¡cµŸ*‹¶#ÜC6»¯~ ‘ì¢2_1"ê¶AˆõâRÜÊ{Ó÷7·žÚ ß=†¸@Êþƒvܸm鸡5ĉ‡6odˆ¤1yÚf»t܉(è/‰Cê¡'ûÆd®#}…¢ùtD2>Öœ)DÃJ×µHÇ}Ë;µÄ#ð0.SG–gÉnÜtÇ ¹»Ù(GþÙ‹Óï26àIëŒü°ñ'ù‚m53(;þîyÊ_›¼/¼Ã…JŸ ué"wP­Ék¶Eݵ*m[W$«¢þl’ÌO4Ü;S¶,f#^Ò>ržÍkGÓníïLwŠÑlO#-7¡Ž–'¯ë†šZÐÆX!4wèÚOkÊü®:‹ŽªgëÃýJðj§Šï«Ï6èꦽY↞+|V#èÓΖ~xD¯÷¨FžÇ@R¬¿Jó >.g[ªÁŽ¡×ô‹ŽþŠÁȺøÑÃ~3µ¸Õ51…±––'–vÿ‡®l³+¢”!—Ãb\™ÃŽÝ‰¦œ¥I¿Hü-Äÿ }3‡ˆHðþäñ8}MžÄ·ÁÔÒ=ÞGþsø$ðH|_‰Ï+îû¿Ò‡ôå·Âeo¥AiܺmÇËU€Ó™ø‹?È“øû…§ìª3Á-#cÚýq–Ñò ‹>äv<…í‹Þ²F ök]«5Ò:ŠIêÜÙæ8ø…íC]ó’ë|<ÀA~¨ôhæîÙÕö ­’¶ÖÐ`¶ùèÂ]M ݽù¡åOQûÁ#íŽ),9ÐåÛʵ¨Yz <¬F?bqd½@‰Õñ0ïÇ–™ õ09ã„PBsfíRœ4óüóæ9#Nå—ž¾|Œ§EÖ°úi~%qmV 8áÏ Kú5îÚ6°.:¥Íüc6½üb7ß‚«Ð=}ÞLØ'5™ïáÂy]ç÷/» Ç*é ·,.áBÎKýДY8Oºè,‡[ìÍETÜú³xÞ€þ­—†àG©¬l~³Ž qeΣIµÜW›Kf9[5p­ZýÆ-® l™ÕãÌDèšzº“A@Ø`Æ%zåVDôy°ÕmJCÔ„ÙÕMÇÃ|¥v»ç(b5Vü(ˆòÁ­üxw¾c%HàS^^€Ä›ío æ¶#Y³„Ùµõ RNŠÐÕ¸˜#­.¶ç{ªî: ]üÌ܇{¦ o¥ ÂðÐhÛ÷ÃÇ<ðHmZ±ÞòrƒØÞd* "Щsð <Ž0ë^^Œ"c¯BNi”Fì¨.Ö{òÐKâ †Ò¨ôsŒˆ‹ê[ý±.N›PÛ1t¤w*Ïv+µlìËÂóŠÓ»®öìA£{/Ívþ÷h–4ê»|-™Ë'-c¢Ð¶ãõ`lþ ´ç½uc’RDgÊÐ’¸Ç“èºq;¶/Ý=¹¾<φÐ7›“ÌþÎ{CÒÇ‹e0òÜ-lÑq />üöÇzzŒD(v^RêÇØÁS¹ CÇ0¾Q®…>Àñß8<1@{[¤|ž÷JîõüMñŸ¡®£UØeÜÆl1ÍHâ‘ÀËmo¾$Oýá q¹‡Ä}¾ebä’!q¡ºCØéŒ]4†Ž1êÙ¼§>÷GÄ\ú;« £³ñfî,t5x/^Ø Ý—<<µ{i¡;ÍýÖ(pºwxœK.’ïñ”¸Ö;Ì]‹TQè×ð{oX£LÆÏ£#j<Ýž8vÂfñÑLl° FZµŸ2ÊèqÒÕ÷RBìwœR½kÑ]öÆâÝúÝoƨ8Ò7øQø¦a5ž6×îÃlZ^òÇ;XØw *eÁŠ#´!¡ÖÖcö‹\ÂÙg>¼-Ñ8÷ª¦¼B›Š/›¼íu»áðgÙ[xÇfæK{pQÔèRï K¸DéìU·| ·_wìVÝ8ËAÙ#EAð¢îöP€wð•‘Ÿà{R#ruTü/kt¤¬GÀâ—.Ž_éøÔ?´Sù+9²3ddë⛼FóHÌ1A„¨vÔìIZD¶–VoÖÖ@´²v~ DŒÛâÍ™ïƒÛ_¿rt¢ñ±þS¾cHxõ[zL·Iæ‘mnƒ¸³lÖT‘õ5RJ‡m;4d‘ÑX∠·(ËæfÈLHQ¯aÅ}æ4›/7 ðÐI’SŒ1¸öæû”#w=OýÏÇÒÈ7°§;Ú°Ojf¹ ÅÓCÚ¾|úPü–éÒ¬¸Êθ¥1È ü'½ïIT¦ÿÜøÔwTûä¨UBmð“Ñé/¡¨¯r ?{ÓÏ7¹¼½>烗¡“¿:–RÐòsÝ.Õx4™íÌ1B«’rZ¹åI´=¶•ÿ#û;¾F™”ÜGgÖrµG,®èÚÕ±(x`=ô¦iLö_ÑÛçt©ïy"úÛ<7Í”>ÆÀ…åm]QË09¾bÓ/ ýýOüÖ‘U/~0[`ô¬•WG¬Ɔ~;·oĸÝóʯXþO:é·êÄ}Â土õ’ñŠàŸΈ¸ö¯8ÈµìØšáÿœÄýE9–¾×€„’”ðí‰2ìX}?6 ð«x6ð–“ñŽàÇÄïEÄMâ~⚸ˆŸ²§è~?9#Þ=f$VÐ+ÞÿêFÀ¿ãddó’hûF(ùÇs@‡EÔ®(ÒøAÍq¹}ÉŸ5OjÎÎ4œ…ÖCéI»zf:òFüÄÉ”À”‹;÷‘¸6Y«–,Ä3 zWÚ)X|^aìÅøg˜ƒÎd±lǹ]†~ÎÇEa·)Wž’¾öÁjYA²m8Ï™Ú/â1§Å®cÓ9¨¸- ¬w¯‚ûýg ɼÃ{ÿÅg©ïÏÂwKé¡×åð§m¼¶}éo\U–\oŒ€'J {%˜/±®· Ál&6÷’8â+éö{Âä<ŽêÌEbËâˆâ,nÍ8áBå±®öŸßZ îëéOQk¥q;iop‹15(5ÞtHf>SÀZ^Öí®lAZ€ÓK{Þ Ü~ÞóC#÷Ö]¤£¤à¡”ÝÓÅ߯ðhWý:ä^LØÔk€ü®ˆÉÚ#x®¶ÐXæ…"æ¡.«j”f¯ ˆ\rå×iÚù'¨ølÜ‘:i—„ºµ.Ÿ?=4Ƴ;õœ±¼˜lãûx ¿xo*¤±£ùâ=Jo¼=Zi–æNwW¢-ìY§Vù:„Bò¾˜PðJxñæåCèÞà³i‡;z­’¬ËK¦ÑÿB_êä³­È®) >‹¡µ/3-[Žá[%Ï~(bTáÜc]É#x³<ÉæW@Æg7•ü‡!×|Î?*‚ÖÒƒk[sÞ@qò]ÀØ;ò¹%øñÜK…œþn{â ™ï‰k–(2?í%ã&[â£]Çï˜þä½DüÓIç/S~xGÍœ¿òÜŸŒC§Ÿ²×—ZàôÁ æál10e‹?ØÄõŸxC<¯ä÷ÖÃÞNþ?ïÿ3.ù†ä6 ߆mÆ©Dì.Ó„Ø~¢Æ'=ÇÁ=“8èÅÉÍRpt±»œPß§w¶>Ežàbªùo¡•špñ’Œs¿9 7·€Ù¸vè~§:Ö 4ªÒpKqÂ2¢G[n¥ ü—p`3/"3#ÎÅ#úa÷Î7ã¸Ik=©e±qZwb×^šF¼†›¦¦£2n?r¸ZË‹ ¼ôt¼‰„}ÆiH˜{¿Ã§þ_0¶Êµ.C²R¸ýj&¤Ø};iøé¼I×þ,Àݹáè·ÉF¸—ûÅðl®”§õ/5×ÆÃ°Ÿ dóÿÚ9’ˆìá£Å¾£‡ñ¨>l÷²8zä®\óMÖŸye‘›{Åvãq¿x´1ýwR*« •£¨öâ‹ûg£tÝwÙ‹ÒPvÅÖ£ã`Ê“'2ÒýûQñüóÆxcTÑ…÷( üÍ/»E~AÍý?Í•Q·¾;@c¥êÃZ]õ²A«„˜!žG—©>dÁKÎAzÆ÷ahL—¨—ÖaBÓÍbéY÷×h>}ƒÃÞ‘-; ,¹¼‡Ð2k~Í<µ­o¼Ÿ®;™¶æÂ¢¦bh¯qåwIjDGMƲö¢·èlñ¹k¼ƒ¯&‚ý®.=€îUÑÇê]Ð#7›_ñ­½ÞÊ1Š"èknÛF‰åþ'5Ѥ[j Í\•ÖÆ NìOugJ:¼ŒKDüY˜7ñ’à©‚—½c)!Oâï‰û<y!ñ}‰8Fà”Œ[¿Îú*/ÆAe‘›Éø@æud¼ê’º™ô*ê?x¬üû¼¼ö˜ìsƓߩù§ì·æ®æL'uCI _¥}pÜ÷Z{ÃÖ}0ú5p7HJ§Œï{눑: É?›{ô‹µwú ‘ÏY•I‹áLrïý¦œ°¹µèòó¬qتpUÅ=Wtª7­KëÑ'¡o[Ï ;z•ݱÂ÷è5­“Sä°F¿Îºiµö ˆôqpw>Ç ÊÓ{)þ¬ÿÄ¡ç‘ÒEŒ¬H½t!îta~Hàí_ù!Á[‰ûüx#pEð>ÍþCãÞ/ %<6:A ¾ªéªn:Ï÷H=t^?=²_î½Ás2ÿ#ôUòü´-Ìù÷õÓ•š²›6·SóÃû¡™§®SuÔ}.‚=å³8µ2þ"kIŒ—5ó^>þ‚šzÙ½ˆd3‡ó§Ls!˜ç*d‹¯†å¹ýí¶ÂÚ]¶™ß­€´öÛ2ýøàpéoÜ5·ZW5ûÞÖÚLÕ8/V¸ZQüð‰´½Ãå#'öùç­‹’¸§á*žÚœN îÁ—Æb{À“óÖõdÇïðzxÒY}e|Üî°ùÃÏ'YðG¤ .?DšŸS®ÒÆ_ÿ4`ƒk~Þ7«KẚIl_H;B…6~00‚°šïÌ®*Wq¡ëåã|G’gFwXÑŽœø›¶Ð2¯ûœdq¸NsüÊã,cÓ½T}§í½·Ð²[8ë¤åÑuØrÑU5$šÃ.›‹!‡“sSŽ6!p<Ú‚6Ÿq!Cþ ßt\„¤Óû|àÖdN£0©NÕK «×9Å©Ág톶ƒv†ðÉÑ¡tþéon;+Œ+Ç-îX᪱x5Óèy\;:³œ†ã6‚K²ÒzÇ#¤¿é2Ýd5É#n©XïRô@Ô¼¢ܨ£–ÍEìòröNÜr•‹À,–IÞ/=ˆD'‰w´¾ yóéߺ­Ë‘Âg{âï¿wškqvoíîòVV°¹áÞúž]ÁŸÇñPnËš "x$ÕÝÙ^?ˆ\wÇ×»Y‡‘ß´ÿ}E> Îø½ê:Œ"UN—L”º9j÷ÛK£Üç{1]Á{T:‰*¬|ÿ ÕA»¾W°E ¶Jw"ôžm1V5<ºÏ3ÏìÌ»ü&j;hó‚ÐÌžCÆIƒ5bŒ‹uŽàè>ùšH&GÇ9)¦^bÂ ÚØ Í|+5uk=ÜdŠ“#QµÒçCa¼¹ý±2çœöÌzU “?ÑŸsS`v‡WKaÚÇùtÿ|„•Plz„÷uœÙ›üÜk}0l.O•úݪƒ­GõI aØõÈÞ¾N棎ôž [¶åÁi`tÃÄ·^\ìáçÜwF‹¬/‘q‘¢å¥~õ¼~Y«(»±Àç][ÉäÜ(ü~{¿~vJWönfmÁÕ¸„™§¤,·¼WC!|sO䘇ú²z<úì3„G$´VÙ"òÈ ºOÉ˽ÕXíWK*bvUæ±u!öFy‘¼r"âDm°ÚŒ„ÔA)gU[$í¬_=)щägŸö¨™!Åm$óÚw5¤)ÆJé»#CàëÐæÇ½È3W¥[I‹{ÞS1–Ý9xÈ9‘,sþ0²Ë¤¬{« '—Çøvò¾^ï2éà çñX~AN5*-y€â˜«ÏC‹­Q&;™[ø;åõ‚õœ£ò¼Ò´e=ª•wÐeñ½C­ZDt`NêÝ>¬‰AC£ø™»òãx¹OÇ-qé&4zÏõŒ†Í9œJÊë£uµaj\DÚ\¿m5^ƒö÷‡j¥Í¦ÐiÒÔ¾›Ç ¯FÈ9Ùš ;ǘýât,z}>=î8Ž~‡|ÑYº ü<î,T\Š¡À+V×z0üâg™¦W7FùŽ ¤Ê`,JÕYê·=Æ7/+QcÒþ' ½Ð‰xEàïU'ÝìxæÃÇÿœÄý¤^:Ïsÿ'½”¸&~/R?íŒqÛá˜I^/üºÿè¦D}q¡n:ŸŸ„>¥z·+$Ä*&Zb›ÖCÕkkÀ¡¨?c½@{šËDè«ß'?'Èø8_G!ë#óz©aþ‰ª©±"Ïd=qÞ/@øLdžL-†Å¡çx-j\$x­ìÍBùSþ°­?/Q“ ‡g†üçfQõR»Œ€Ù<¸|Óìâ‡{ÌʼpÉzx*ÄžtV‹€×D–¦éjød© ñ…Py¬µªŸô–Ð>³î„W( dÊÓ Tí`|| !1FÓ!låóðŸJ›>òâ°sŸ?"›Yö.¥ÜCô˜g-nжZE÷F\þÉ õ~ÚÔº¿pÞË®ž—HíJ‰Ä·£÷ª'¨xýð«!9ó"²guSúñï†h å)ñ2­ÈO]†\;“c½Ž¢Èo½åüÂç¬éÌ46£È°"Úò¡Jo±z]9ë‰ò[cªNù¨ŒU)uÕe/7 –£né~¤ÈÚáÙ…<uý‡x±f§kÅý`4w··ßSEóK™£‹øÑ2Ípû@ Ú|†ê_?½€ŽQ‡éë3ñj“N¬ÏŽ³èæ ´pd E¯®ˆsa úó•ï%ÉÄ@Êš‹ÛX¶`hYÛÃ;^c8ØíÜ““ÝY±{Æ~©­eÝjñqNýŒ‰ÿÄáõéš© @c²nMð6‚×ú¨äxP©¥D>™çüua½Ÿà­ÄI¼OðI‚j‰*˜¤ ‡ÞWžå·ƒah¬ß`”T#·ßJüVPuSâ9%|,.í—Ï‚3O)ýìs˜½£ñ'2‡ÔOÌW(·WÞxO= ßËüý„ŽjYi5Äm4A>ï„ælžÓØ!3œ;ç£#^JÕSçy¡ýPø©›UG-xØRÇutF˜XŽåÃyš =\ã%nÑ(7ã’¾ð;•oàYqþô«ØÓð>Ï.¸Q]Šª›Zn2\wþq~Ë,#×âòÄ Qi™-ð ¾·!ŸÔÑ‚tFÖÕZë—ÔÚ+ù,ÒÆŸ,8¸¡?χ'çŸD¸AWÀ Ž\D.›±kïÿèea†[Ò+£ó~¹à2ĦJ¾ÿÌF‡[É…E;™¢ÿÛ9H2·{ó^ØÛ€Á'ëOs"ÁßøgÌÒn$,WÛÎÙŽ¤œ/žÌé¸S%­ô iǃy+ßWâ.3«nºýfd%íöw”6Áý›ºÂöãÁ¬LOØd4ÆÉÖ-É@¶¥•crv'~ÿ´V9ÆÏù#‘û)<-»zù}ƒ÷SQÀ|7ücŒžúr¾Ï¯ŒâŸØ}Úƒ2ö•öë½¢|›e…Ez*'§rŸ òâúH:»E¨*ä U½nŒšK¸œŒ Ö†é6kêºÙråÒÒñLmïgíÝ,h¨I»U`O‡r¦´Ï4Ÿàe ³Øéa+4mÞéfÓùMƒft%ëÔÑœRXΜ$‡–3R¼Q®öhUÌq|­4ÚD&Ún45 ÓÀÚŠ¶ì)ëüêuѹ¿äÂkñŠ/%znÖ]r†ÃÛô¦Ðm)–±w…3zâƒyØõÐ;رl±‘ß?qxÕž'LqÅv(K…¹tos&ó+Þ^IÔ)Þú/¿¡ç×Dœ%¾k"¾‘ñë_º)Qç{[Åæ7œ •Ýç*’uB7%ãÕ<ÎI^KðS"¯ãäÕQÑÜEòR‚žÐKeËÞŠ“…×U£ÙÚ©õûÝ¿žõQë(×7Ö¶–ŽÂ‚1«ÆÊæ ‰kë4½sòRÙ8ûúú¹Š¨<Ø ^+44òƒ]Üjƒ”¬x8pÈÒ9<Äù[²·|`€ÓJú§ïäp! ¾5®*ÎÂA©úàº}mT«0ÜMðÉqFu3;*šáÅ7Z9" o›g]ÞýKàÓ|ìüJv²Ì$â²éŸ·?i¾#à^&í‡Ç}¸vàqŒ¹íOmùÌÛ‚ÜILgJšasŹ6 avF'û ÎøøSØÛç^zXûR+"7åW¯=õÑÊ­Âs¸©½î¾ôs~ÜúšÝµö§-:“.Ø´M"Éïiâ ß•T½4ô‚Õµj¤(\šyÉJAªÀ¾“{j´‘fùFleËd>%ýH… ™±r4¸ßjZ¤ÏZlzƒ­ûoãQËÒwvü–È-“ÖQD§~»f^}–J·ü‰î™ áEï¹’c×5w£ÿÄ(c‹x7ö¼RV²ÿè¤óï~3ÂGCæƒóu Bo%êGCQ˜y—ÃpÀØzÓ8޽²¡^'Úòý=¡úO _Ãø5c)gœþ ø“W¥¦î¥‰ì0çÖÊÙâ¯I­[¾ýÞs[þ>ήX3ÁÔÄ…s1ñ<“„a·MºŸ YwøêÙp¸ÃަÇR¹wŠÁéÛÌç;[\|øõºó¸äýî‘-HõÓú‹ž«×Nƒiø°4¯ÙÕó¾?DzüÃà2SwîÉOlPððÿóTZ—wv5N´CH‘вci¢ÓôvpÛO­;&­ªŒœBôifæ³{5c=,{•±]¦}´#žˆ‘¤ÅÌŽ„º"ùÎ’\$Xu^¡ ÉbÊbQ©HqÙֺ͗i’¾•¢N‹‘Á”9ù®¢™4itlî¸'xGf$¼ÝÎó–ÙáJ‹ž+>GNÀ±ã¬RÈ{žïÂ_&†'²›~šìx…ÂGêåé™PÌùœ>PÒe‹§÷Zl-FyxiÔ–+—PI lg)C5]Y*³V;jÞª?* BÝÄ¢m¦L40¬Ž­ÏÁ Ù™È,–ÝhŒº!®lñ_‹gUºZТT4~¡Ÿ ­±ƒ}¥G=Ðö“RRö7í8E{x±úut6¿X[—µ]¾÷Ë|ÝÐ#ãÌ +RŠ>¶+wæD¼fº4­W$ƒ°½®S0ø°FHIr††Wº.ÿ•ØjË1ê÷ñõ}ü_~ÊÃñ#FçŸ8t™¨¬[]K⃈W ý4ÿ“¯ô_~šå‡ÄÏ!â.á›Y˜.<惤~:ÿûÿÓG3Ÿø.«Àóíc~næ=´šþ¯ê¥„ÏŽÈ ½”¨KÎók².™Ú¶ÆZî,é#7¾2i"84Iúhˆ¼à×Vr)áÇ÷ãÌ“qIÞÏ $?¶“QÙ°oö¯˜xLü•p¾VÁS²£ô|¸Îv$äÀ.yÙω ³¶ø¦Újgxû|ÓÝ _Ã%¿ƒá/oñÈãØª/œÿ™”v¸:5~v)I®8§„Y®kë³o#ì¹êZ®ÂPD›”îlFT!_^Ƭn4­Ž<»…± FÇö*çá–õꋜRÜþ”òDÖ‰E—άÐBÒ\±WÕïWHY’¡¸©PiçÌîÃ]™¡_.ážþ„¡ÝÍßxh›ÐóIû1Ù2«ªÒú ÷ÓcþÕ%xÌpþýDÅ D®V(OåDÑ‘uÔQªµw/í€6ÊåSë£Nª£â§ši ª^¼<„ŠC¨).×f»º¦å‘x6÷ó ½Àv¼PîßÉû¼9[ökÇïAóU·G–¡Åб³˜I­Uvìr›Ñ~bZö"‡-:—ʉ8ƒWÏ„%jn¬Bw÷Y£èô–;èÙHv£:îî—ôS£ƒß-[hV`øHÂù 1ÒÁïm–zìß8Lb͈m ŸgB/%êD<û_­_,Ä!q?Ág üu¢.¯ÖôgÆ-Ñêæ•„Ï”ð“Î× ½„¨?j{‹$o¶&ëù:—¶u Ú= 㫞÷ë˹ï¨uÉ÷"{ÒvêQuÓy}…ˆFÛ*„Z6“¾4Rg1ÜRnÀŸêÛ&êóù¨íåY¯Ÿ–¤OÍÆþ ‡–?l)bûxaW#‚}Õ»àÀU÷E‹‚ói)7ÝUœH¿áÅ}ïª2ÆüárfÓz_ݸ}b_–®· ã.°=J…WÊó^Ëqj‘à§!gùœ¬0TÑ|äs®]mól%óËЕ}5w¦¾®d›æwfDt·ìt–JBTî³3)߸©ø’~»j¤·þ¸1IßÛÂø7RVH½0¼‰ã ’mtä"3 ‘Â<õúÕ¦Hmevùnb„ô¬§Ž¥-p7Õ½P]ã*²&Ù¢)£Ûñ =x÷¹iOd_¼5«(…§¨&…&Cä•«Ô”Œ ã‰4Ëäo™%(Ì©v§@ñök˦ÊQ¶:ßv\ó ʳD8“SQiÚ0˜%‡jÙM9gÅP»ÏÓAÀzê­pjðEC¢9½Å3xɨíxÑÐM‡¯ß1?-Žæ´aW‹B´rhÚn*jDÛÍœ‰}ûo CÔ°°éÞntöM¾ª?t ]·ö‡ÚØe 'AKœO:}%ÍF2rñú{ãÕ[Yñ*ùÞ ]Zó&,SÃ#ïzÚ"1jÁº²Bo–¬HhÎ\‹ñò®XÏ?K1‘hÖß\ýO:É«›ì’$ó4BW!ð÷¯:áB¼úèÆŽSÁGË\Ékâ~D<$òM‚/ÔA‰ë…¾Râþ…>Sâþ…uH2NõÅûý™9k“¨uÅyŸ)çu¤ÿøKMÌWDÇ:ü?„oç£äǪ±Tÿžö£­?XÀ<Á‡.LšB­w¼˜jìy˜Låµóþ·Ž»ÒßMd½Æ…ó—¥õôà<ôþëQ O¸¹NÚŠìóƒÇy{¥]SVð²}wëªê>øÈÛËÿŠ`†Ÿ@øÎÒÕýdßEçÚUTù?1œ‡ ×»ª}Ùâ´~G8¤Ü' —Ò!‚…Gç‚ÈŸ”)çcpƒÆ`u+nJŠŸ®sLD\f´‹Ú–¿ñPáÖèÖ {¨uÿã;’zf¸SÝ£%2‰´=[S[‡Ž!cnkÍ„†!î­ÉÎöÚ-Œ‡îîš±µx¤-þàÁOäÆð?Q^‚ü÷ÿ´˜Ï¢ øZÞ/AZŠg–ýˆRÇå<©O-Pn¦r<¼1•‚âå ÷P-Dsn‡v.j•úh6%¤¢>Œ¾U(#Ïé%o¬Ž—é-ÇM”ÿⳄaàÒ¯B´È¾`¡“Ek_Våù„n´Gú(éGø¢3ø®å«•ЕҚPÍŒž÷wîU›C¿ù*Ó·Z¸Ö´Ut%†–|Nr†áxÑòÏy0jø»Íˆs5ÞÈé0M0ÿÂ[M¿Åg¦…þ‰ÃÇËu‘ýo#ôþ…¾RIÙÇß­>¬ ó=±Œwž²³4ÿŒŸDžIà‚¨Ÿù Q$âѱ+öá T>Gô -ÐIL‹ÚôÝb©ç|žfjñdàNè7ê™öÂtÃv˜íý³mû˜½ ¹©ö÷yw²ô˜ÚYŠ‘ÏwÒoj¶õÇß,¯žª—N^’–¤c¡òÏC/ ª ©>ÓÔ˜C)%û¨z©û%ÎìÚj8vitp.gÆ…Ýrë U»¨ú ÁCçãÚ¥ÿC:û6PëÂB\‘ð)N¼Î-¿Ó]ꎃ¸¼kúfxRËg¡Æ·›Œ çÌp½Ÿ/5¡ð™÷…ÎnrøC‡°]–:é)„¿qVÏãGäñ/nI§¯"ÚïÃQá©ÜT}2°}=n…_eK\/ƒ„‹])6âÄüvÝžHq$Ÿ¶vZ*ƒ;Y¥§^uýÔ¬G'ZŒgÎG{ØhX wùx_(DÖ“Ó>ÙåÃxP¹|ÈŠö²¯³ôhèFŽõÿÁÚ{Gåü¿ñãˆP”(Ùd¤Pv„z(‰2"#2!¥4hí©=•¶´÷ÖÐÞ{ï½³*™Y_ßsîçëþÛé¼?ßs~½Î]‰sÜ×}]×ãzŒ¡&ÚHRd}½Aä8Rn{»(Çó!Õ¼}?“ˆ'ÒËUÏÖûW S¤Ò"æž<²Õ/šu»OGNƒð «Ó‘wî¶çûçlÈÍrú}…AÛÜ¢®\G±îG?]”*WÈËämD¹þÜʱýf¨ :köéª ]™ÇíQ“n“½Vô3êXÌŒJPo4OÍÉôY8 «î*£)îuŠwæn´è¯YúI#m×ýËÛÒÑ¡ñÛ(«b/º‚[ÎÊ~?&_û[ÕÑ+¼ö¾ úÂOÞÝýöê”uh˜©"môu6/¥Í{¤î¦âs3ê‚ÿs?¤Õ©³ø4´¹ªÚ~GñL‰Ž†‹ü“ÂIi¸(Ùû(¼”†ŸRº~¢Û%<š…wíJk¦æÇËí]N¾i¢ë”H=ž©T³påRzÐ[ÅÝFõY¢û }•ÂKóRù³MYyçäh:¿TôPÒþÙ2xðœcþ§9Ðe]ØÇ²n9½iû;µ’¾ª³ûȱI¡çÌå¹ÙÏès&¹?ÖÛþqU‡ãÉñ.¶^#8µh*‡®™3˜L p©_ý‹ϸ'EÝŸ.€çÎ ?NÈÁË¿qö>¯™ð=³Õ¹ýSüíM.,ú,„€oz_¢ÄÝh4ûŠu÷L rÆ­–DȬâåú½èág.%DÌxϦvQÇ«Š'³ f4FcÛ╈3ŸØX¯¥‘rƒèV$9–}÷ðã@ÊwÛëîÝ‘öh¾íš]ŽxÍáì6kYßëO¸Y}FŽMë[)…¯ÈÛ]’¡‘.‡üŸK-C¥n¡°3]´ø½.Š;ˆE€ ¥“JûÛûQ±ãGJ˜-ª, ß&mû‚êk ßúÑçPgî'u7¡þC)óõË‘h<×ÏyÞh*ˆ~§*…Ý-¶vJ£hÛzÏ£-T³=ºÍÑùãðžÃæ~è¶¾·þà.Aô„(¦zî@os¤”kÏ{ô¯²úüè„ ÞxhÆûƒïSŒ_-­™§¡ñûI}0ÖßT¸ ãb*^ ©Ï©öC /eØïy5ÿðeþ‹g:•î‘gJt‡´þJæJ—Oô†S탄?Cø¥ßaä—ÒöÁë™Uv›KBÉ]÷·WG×!<â¯AöByéZ‚eô{á´Ë9©zC—Ë÷Ööànèë+»v«‰âÑÇ/—ÜtýÕëÍÛzsÂ䎑Í&>˜íþ$²®° ‡ƒ8#éx)á—=¾vèð¿ŸGN–Ï+þ´ÎCÓÂdM·Ðõ÷Vf1¿‡ë^Èqáy‹oÆækøØ¯wŸÞï a(ùiy# WÐ'oû>9¹ÿ¢«<‚Ì—ãÝ×­ ‘_§#üùeQD…jíÎD¬/‡‚×ÄÇöïãb‡$¶ûj™EH¹o=œ 2„4v—–Ó^"£[ù›™åbdý”RBŽ{ŽW}6òÎÛŸ¡?»Lù-lDÑv#§²wPrBòa±Öl”YÍ*æÛÓˆŠîÌ:ÍuP½ù‚Ù®¢æ~ÅÍæ¨-“YöÆ™®Ë¯Ý'a9§MÆÁ·÷´®B‹¬ÿøÆ\s´1ŽÖªEÇõŸ¼5Bžè y§¥¼š=ó—µ\jüŠÞ«‡x¼GÑWâ˜ýn¾ NÕn>»|ÙÔuÈËÏÖÉM½¯ÉþGæQÒ÷¦š;둱ÉÏ‘=’Ô¹Ïþ…‡ÐöF¢Ã"z[J‡AÃ=)^)yMÃKÿá—Þ)­¯Ra¿”:ôyÑùÛZF6 âÓá~1ªSxâ:< M޵+~ÀóOVòí?á0{|®ó,øo-š/ÛnÊUM‡!{ß=‡¼ÌÉý¡²ÇÁÏû¸æ©[#ÔniÏýsî³ÍŒç"ûä‹: Æmî™ÍË˧t8÷½ú}$ÈZ/éeF’Ýø[ݵËðjÆÐ«†7æH³)¸V§Ý‹×¼]“™"†Èfo}jŽœ°Ñ“"ÅwÃdŸÂÕIˆ(?2™· EÂÙJk…ìQr¡MÉrÑʼ/9xîBå,ŽfsQ-­åq£Ž‡âËÔqöåÌTÊCý³_Êoߢqëç ëÐÔêyN9æZ<›T–~G›·µmú¼SèHñšóûº>š]YÔ~=‚)<ìè}¤#œ$o…¾½mœsgbàúSwÖ«14­­üî’SÎ<î•PxÓíR‡@–—”~ù®}ÍŽß~Òûâ‘Ú6¾ ho9Ýᘉ§Žt«\ˆ†®o²eÌç…x8møÛq-EQŽÀXú+_”.>Ñû·¾Ê=œ9Ù;o¡Jº@³µ 5ÜÚ+¦…£6Ðãd~2ꥃmXXŠÆù¶!ºZ&h^t"JàF8Z÷ÎdYØÇ‰v»°?’Ïüе¤2§jÉYtHŸ˜–ƒ^ê _O.B?×fµ»"½èšµ|Äí)†j<Yö#c/•¯œš_ê¨>œ2Sþ!µ—‘÷%ã=€à)ä¾Èˆ—’y“q$uMê„ì…ä^HîƒdÏ“O;¶nó^Êg‰ìeŒ}‡è”²Íž.›Þ%‹.Ï·ERïkÊgv/Wzãt¼EцÒó&¢$VëoQå_¼ïÇ´Lªþ¹ûÑ¢»´ýwt·h£×Ñú캲†OÇð®¡×ƒjÀ×#è<˜§Q§«|aÞÇ”ìz4–Ç/¾Ã뫉^ ;…T<-¿·Z¡å l„óFÒuùí¼Çn¾˜ç]ƒ…t)¯Û’Ë//Í«»ç/ÕÓ;VQ:^Ï^¾]ÆUêð²ò”;üÚ>ì—‚7IÁ÷QœDY¸ür#K9,wÀß¡ýíþIè½û®¾=^øm~q†/µØœˆ@ð±d»í½ÐfÇ[táZst´f: ÒvŽá†Òˆ~¹Hû[,bµïêóßëC\Ô¡9ʾƒˆ²—>¼‰ë.ñýzúIÚÎ*[ùÅ‘<¾ÿ–…Ó¼J•è hvEZëYwÿßë£á¶á®]Èê?núõðY䬸;츹çÞü ÷2Cž‹Åô%ÖÈÿÚªÛ!} …×j2c–D ¨å`*ÿÄ~”(_ªýwO\U;ž/‚ò€ßÅÖ'?¡òÂ$GŠàMT³?»ëÄtTO—è-Û‰š7éCZEš¨íU9ù|ÝÔu Üœ8€ú?|êóJƵª›çnEÓù™O¦h~þq@®s-ãÜÊVKÑvÑm]ÕÏ!´7d9ðªÇ£óVÙÄb· t3ìÔÞΊî»Ûß·ENY‡6<êÎ^=­_“ðQHÝ1òI}£ùmŒC|×Ƚƒæ·¡;ÍòVCÖVè¥ó¥o‹¦|·~‘͵΃!|®g”Áè]¦þiiE˜( ¬L5ƒéäx­ãÅ0÷åYzbçXJ :ŠÞ†Õ—ÁϹñ4ekUØbVØ)· XF×ÛÓß§Ô{"M¹pá’Ùô®âWµ#mcàÖW#pê¨>ž}?¾O7-Ïê *…Ï…­»6Ö±ÑýØ¢ê÷ÎAàvΫí"6x;3£;â=‚?4êûˆ1õ®Áç?öúâŸõž'‘jërç¢þ|“îF¬$“Ç1Ä¥=›÷~×°0“;ƒ”.Éæ[þöAažyȸ+cqo?2;çt›µ ÙAo™ÌFî®Ð;Q•÷zߘaìa(2MÜ~rEk|g,oGño™³ÚþÖåûÏ-¹ÇQ±äÀ˜›E*ªä2³R¿¡ºãBrèëxÔòh{û°²¡Nǵoò¿QÿíÜ[‰Ä.4ÚæéVC3̶û‹ÖEsK÷å½CûÜö´:ìèäUøuÀ& ݻǾ͋~„u›¦3›Ð›™~Ò6¢ý| ¼¶Ì]>eîž÷¸l…&5¯‘úcœO§âÕü—þ‘OCpr?'õOpâ“Cê…ñÎ@éëiüR‚’û9ÁE©ŸgÄWÉŸŸ‚gzy— ‹àüÐu‡ßŸ{ä,úü/ï›ÿe³ËoÞ©ñÒ=‚î¬-LtSšŽö¼”uùëÑjöÄþÝ‹ð ¯Í¡äá#è*ÜËì΀Þd…¬Ü' 2ÈýÎ÷~erÖ‡û=x9‰Üüõ}Ûn]C¾¡˜q÷È8 Ï•_kZâ#?Ö}]q¥çºƒ¼Öq¢ÜÐÝbÍ`Tæ ØÛ7.FµÓÍü¶+*¨©8âj{1¾'ÞÖ¾ë솼”‘ãsW£‰OÄäà´4—7rE¯A«n5O•ÈI´ãðål¦ÇèÜpæðJ“mè^<¿ôÏngôlê»Ê÷Y½²â»l¶£ÏÎõyFe ú;™.}ˆoÁàáýâÓ´§¬C2?‘½ŒÔËÕ£îþÅKI?%Á…çàÿÔþÿ…—2øÞPú r‡$>¦ÄŸ†û}||‡ªc²ž8Ñq‘ùš6OSº Ëî°ƒœiP oúV(„ùá;òŒš".Óï„ÿVVy}þmOºÑ[‰³_<=9£¯oݫޓ³r~qÂŒ—£á×Ç&XÌYq`ÞpXÍZÅu`w;ž®Ü|s–U lUÞ±>‚S'‹Â¡ãt~©§Ë=Ѻõp=Ìò‰õD6ÜY• qzÀc©‘ƒ‰'ž[ìŽús<>'”:eí&à¿ö ÓŽˆÙj÷þï¿?PE4¡Ñ;A÷„Ý? Õ#¤riïYC„›Nršš#ÊÅGf_Èßþè⺟íÚvÄ¿(›%$ów?ü"qÌÖ4)¿-=q¾©ã+Vßy/ˆŒÌÈ «‘•cÇ©L)ä˜pUÛi±!±²ƒýã(X~Ýs¦öqù[ïB‰àøÙˆ!(»uª}øÛiT¤^OQŒjîëývsQsµCÃ<òjn¬º—ïŠzNŽ4.¯“tÓëï‘©?„–íœ[Þm_Œ¶µ#…Þ)çѱÿ’fF +ºÿ*?ò¤ÝŸífH$t¢W"¯Š{ÛJôE•*dFkc`G†ñpôÎÿ¬Cò~&{Ù÷¦ÂIÿ /%u9%^Jã]…‡¿'Òù¥4~èT¾¥~JpRòšà¥„ŸJ»;P÷|rW$ø á£1LÚ±ËÑçTF~)—FÕãÂ(/8™í ½®ˆ) oÑòhÚ?ãùhw©½}"¼«BO´{B' @,ñ­ôdgÞäy,†‡Š7¾Uí¦Ï©´»¯±QqÐ1‡›0=òzóñÛ0_5ù]Ì]–|‹zíF`½·§'®Ú“®»Po™¡9Ž‘[¥%)¾šK}Þà·“+àæÑtOÈÕÏüs¬ù¿øÃóOºËä¨ ¼9[‰Âï¼õ™×‚¬˜£,/¿V/ÌÒ$ ›ñR€cŽî³%žbß5”ÏïÏ‹/AØÏüÜgf!RTëÁÜcÕˆîz¿-ÿ›⦱jÍܾñÝëÓ–§"‰e ò¿R4ö¥&ø"õÃèºÓáµÈ0cÙ<¯‚Yª*+uÆ #É\¯®…ÜÏkŽU/G~^qŠy: Â:ÚâÞ¡¸à¦³¹;ý~xÆ|Ñà TÕ?<=85ûʾ.û»ú/¾·~+ê—œÕY(° ?løJÑ$ÞÜôú§š?¥(VÖá­¼~L(_š®/Öœ÷CwÚNIgôü̘e-™„>9µk‘ì;Ñ_®j ªT€Á+bfì›Äð¼{‘ÚÓô1Ò>»6*MiÊ:|ÀÁ²{80œÚÓß…ñ>ÿÿÊ/%OÆz${"Õis0c?dÄIÿyNÑ7ÿé—ÿ£ƒ‘¯Céñ‰Ï7Ù7i÷êóäÐöLêþAóƒ$>T/€ø3êñ ^ʨ»˜Š_*ÔZ¿*5å_¼t†ÛŽcßÜþ/i;b ‹/A žiS<:JÿIO÷Þ·Žç½V•ùLsVÅßó3ŸœfYDÀ’ÙîÂõ‰k°9,Ý»÷4µßÙVJºiH×#Ñ|gò—™ý £ë&žì\’³Îno[¾û\hÃ3Õå]‹7 Àãë¡õÅ©ùÑë§êÚ_ÕÜ”NÉ7Ãlå‘á¿ ÜC.µþ?…´–÷ì ë ‡ìb-"(½d½˜²Hå[Ðê›ðiH_#ý‹àFT¿¢á¥R")—LMÊ(þ …—Òö:r!Ÿ+”#cÿ"ù4¼æŸþEö;âŸh›ÜÐ’sœ¾ß_¨¾ýFk)ý/¹wP¾2B‹’›0ÓõM´|2‡ÿR݉Åê ‰¢Òàe3ÿ~Þ”T'À¶eo¸Å ”梒çÊ ÐNTï}ãñþ’jÔ›†¬iBmó±ØwЍûº91]¬ Ò¾ñßûEÿö½ ãi,×Ќߩ†ï÷¡¥kÐ$“kÚ¼$“˜û ã¾¹ÊÚ+èRK=c«³ÝùÒŸJÝYÑÓç§àb­€¾Å§ÅêÙw£ÿŠ×­:ÝHËî—+^85N³o™ŽbI=…G’¹“Ü+þkOdÔ2â4ŒõJú#™g}j¦ÚÉÞGùÙýa¤ú)ñ1"ׂÌߌúünø¥Œ>„/Nå=1궈ÿj_î­¾ÜÐà^^ÐIï§´9™Êo£í…Äç†ò«ú½Òãô[wŠGCtT^íîø¨Ã~Ǻ¥:T%¾'žckã÷–3‹î×vž[@zï"X­¼:: Ï1¨±{œG[!+¶¿ÿ”_›ç·ÜjsàöHÉÚŠ7ÏžôžsVÌ„ç¤üà«5:ð×—ü5û%ü–sJÏy–…Q¿Ã¥7ñ¢nïÒá[íbÛûüók.ÊP…0§ƒ=ìÛ6"jk„@^Z¢KBÄ“F!nÈùо‚Häx±³"ç=‡¦Nüöùi‰H7Ì]ÍÐ™ŠŸÒã‹ô‘ýºâ GÔäÞŒÞÜ•ŒüÍëÂ,»Ë(ÿ¨bÖºÆ"ÓPºqÎ ”_ÕôÐaEelzõNY¨¶W[{ê®0jª_?£wu«ø'*]P •€F¬Ø3¾þš&žÝ™iü-%.ƒç%£-/ªhç÷cèè¿&$ªvÝÜ6Û†Â%Ðsælk¾Ä^ô†\©>¾ý\²IOgëL]‡¦ŸãeBéx)áËü—"£ £o0y’¾Há3´~HöORäýOüi(”ðÆHß#_'x(Ñá3øœR|T‚£¾)yÍÀ7¥üL _›ÆG#ú НM»?0æÄüs‡ s*'ë+æ;ÜíýóàN‚:ÔÇwíT»Lß/_i²ÿ¾®íú~óßk¸)`¢ÏÖíø-õ‡ò½x¢^øTi•‹h<[`ï¡Öê®oÎy½¦¬cæìãm±GaÕrѦøë<]û°âŽ¢7åcêð{ÙÁ‚ì«p2[b6$ë—-77-(‚ë‹—CÇ=>Ãýkåô\a/x|Žø$ðÄ^Úl6ïÉÁ§%q饸ï˜wĬœ®»hy7ØpÔ A:¢òû³’œs·{ÂÞ4u÷³™ RÊróGwUÄUYªzq{Ë/….~x­óòÍ‹¤sW`•ÌE$³Ô¹Z­Ž£òdÒÇ„ýF÷mFfO»Nírd¿½i°`dr¿È(‰g!ÿì¤I›Ñ] ßM[øÅжæQEé­’ºsgw¢<ÎÙTèCªÖX›õrM¢ºeô†ß²3¨Ý=œ|rÂu/£ÂuSѰ¡‹«Îe3S|û¦O Yammó $Z/اpí"û¼ÏÏFç%½,ηµè>Íb如ž§’vF£·m.›çŸ“è—è^Ëfù%ãO§q`èîº]ËO`D :üWƇ)ëðŠI"ûÛTù¢ŒwCÆ{áTz|ÆÜ'Òõ·eä“’ý‘â£Ò¾?•>㟻" /%¼R²Rþ¥DOÛ? ¿€òý'zF¢»`ðý§ðRšž™šoïˆ{ƒ¯­1Í Ï·4>åÏOÓ'“ü Š_ºüäeñï¡™¶_S,ˆÒ3Rº‹5ücÇÏòÒçXšï°Avýíü[ylê<Ï•ÅçgÀÌá»Ùš%žtÝñ¦éQÉÝß!ÐuÆ2s~8)-Ùò®7vaõy÷š·V\»OÍC‚Ú5ÉðŠý]¹Æs|5úÚ7ײÂÿK³L×:¼p‘íxÈ—Õ…ƒWÓ Â*ÁjØÛ‹0•è™K—r!jÑ[E!MDWûÌx ×€¸®¦í3ÌËÈ3Æg-qÉšyU35üðê[MBÚo1¤§ñ¹ª,SGfšSÔë?ký{iZÉ[äæB¤k¦=ò#L•JÈ 0}éžÅW­P<ñIVi0Êä?s¯CÅP1“Ñ| TŸ9£³qÆqÔ”­Œa[‹ºKC~3T«ÐÀ®xÖ cU/ ¹’ÑümYú}´mùpÀ×`9:\ª¨(t¢{‰¿–yz̲ïØ-sCÇ¡i»ÜÊÑŸ:Ü£kE÷v»µº²ÞzÊ:th®S:êDå¶0æ R8)þHꊼ&ý“ü™oyl¤ß’¹Ê¢õ= '%øyßÒr%ÎOíg4? jÞ‹0ѪþMñψn—ðR‰/å·Føi´÷?u7 Ýýî4ß·Ìý»©ÇÏ¿òîï¾NÝh<5Jßžµ[éÆ <0û9£4Uº™òRc³ŽÐïÄïžÆO3æñ>i§8Nõ1³m¡9oÓa^·òÇ–·b°pz3b4²”º'X{¨Gú~0„Ý®«6þý³|4®0óì¤mw‚óôª|+à:³»ùF™\ »²ï?9 ·C‚Z•LéxËʯ÷O99ãùQ¹‡ÙOnÃ;S'dò¢üøžâÖჿ˻0¥›–poqï5çË(öSú Â4&±õyþà{ðE„ ª&Ýòq@äö;`ôHu bWÔúýÿâœðe_Šx¿ï‡c\Bâʘ‚G‘¼é¸²e$’'í–¨Íä@êFsm‡yH["5ã]þ!d|ó‹;¦šŒÌ æ­vÏý\Nënåm䔼{5ûæyäÍ1雹áòÏóǘ¡ Aècãz9­Ö°šwšÅÏ»Ï|fÒ »¾²*$¢,ÓÔ騩™¨¸õ°B¿~>ªV\*e޲Gõ½û¥ÁÛPsAu%ÓÍ»¨•º|§ÐuûW7úMC½ŒÇáKºhP–?àã4:>ù!X¦b+5V— há°¾r"^­w·lZß½m-±O+6¢CþcâÃm=ôûÆIí-™G§¬CëøIžÓ¢©~@ö5R¤ÎñPÆ>Iö>òš|ŸO¥xá¤ni¯)=>­Qû­‘¾EpÂ!: Ò¿È×)> m¿#s+¥§¢Í¿סø0Œy…u­”ü(nªþiþ”?1ÍGŸÜ9(}Óµ­`3=¿PiË+ö*kh} z°û§ …›êÍŒÚÊ'}™]J)–/©½QoXÔX§e÷ÁJ¢HH«+ V‰;'òþ~NP{Ñ=í’ñh“rƒóŠu•³D6ÁÙõ@ÄÉCpMKùp¾h6•Kêáúk¯L¢ žÛ©L.3ƒ×Zé%GüæÁçGËA *¿-Àcµü¸•^d,Koz9/›žé_uÁnmød÷!-‹WœÒØŠðÍqÛ^.D.–ºfºê-¢F«LMGlÆþ‚7ˆ›Ñìïô ßêÍý½Î#é\PCýM¤ˆ9îq/Œ@ª¬^öÐ-V¤[‹† >–Æëá“ìT¾#{o¦ç÷?‘㢠[÷òfé‰Z²+ ßQÐWi• w~M–:…¢ke7žäEI¹aÌÚ.”åÝŪlT4³]þ¶j ª™W1DͶEu}¤éy4m}+¹ÂLÑpvï9Ne4{3sô¦øÞ­jw8?n±DûuÇK·ï£óqT®6_ º®É ­6B[Þ+v«#SÖ!ÁÝH$<3FŸnFŸRòdÜë•à=Œx)™C‰—Üˉ¿…sÿQ”zMpO²7~ ÁK|Ý(]>á›ÒôIÏ›†³œ`Jw;vèy.uÿ‡_JòØöÈûÊTKÐs-ˆ)ÁKiy‹ššã½Ÿ×CKfùjÎÜ?#Çñ©ô=ïéÎý¶)ÐW:©Þ$C÷Ç ñh ¸+­™.ÀøôöwJ¯´`¼B¶=iþiº_-Ì™„ïmÓv'rvªìg§~˜¾Žïrï݂ӧUAã$ábëÁùn‹f<-ü…g»õg^yÏýæü_"à•³±QæÁ0|¯üŒ8¬Íÿ5g}·Ì@@æ ½þŽ@6;.™ï· A68MöœFH¯ï¹ÈSŸ]þ§2"‘Û<S—‡"ºvúÕ7bш۹­é‹žâCÒ6ï@â¹äjÍ‚H¤GÈ=\«Ž””«ÖF­HwWlêåNCæyÞG‹ý’å°Xr…l¾ŒÚs"¯ïÑ×› Æ(,Ÿ™€"Ûì´zƒQ”X Û‡[ý¤|Û*ª½y5žÚ¢zþßœ¼¨qËØ±ø#jõíŸèênF]–Õó­ h` èéñ*AãÍ»Ÿ¹‹ iòP”¾4Z\õ¿>pmÒÇ/¨I¢cå5ùSy1èbÂô%É ÝNO£vLCOÔÂÀ%ßÃÿî‹õLíÃпjßɪ¥Ë0 ótìÄŸ©qš<‰Ï¬³¼þs?$uÆÈŸ™Ê·t*¾7#^úŠ| Ò'ó‚y¦ÿ঄?Cë«Sú˜2ä>M•«HðF_b*Ç–OCíƒ/¥å<‘\§üª^J|Úh¸ÉM%x©ÖV“^¡ç¸?®šyãÈBÜwvªLrùH× ë„7Ïz[ˆ'½¿yÞeÀpÓ#ŸêÚʯ⅗öÕ™âÓP:+š/•=ûÂÛÒ½™päžyìå/ºÀ¯µ_ÕàÆ§º‚íW2žÉ~ ­Ì€§µöîÚ× ðÞòÆ©y~üŽ¥zu1óÓù5m¿Þ>OÇKß#ËÓL²dr½o¢'Âz7°½`A$ëî‚{#öÞþbCÓÇÅÎÌS„ïÖú޵H•4? ó¯Çö6uª#ýêÎÀjšÈ5Äß‚ðÒˆ®ñÎ6¾ŽŸßžÒs‰î‚–CÕá—þa9ìƒRº å¯d"=¡wDCõYºŸÝ's'…ò 6Ôý–/ô·Î&ÖEo„ =¬—Álú ÿCa1øúᾃ"°ÔðZ?}ç <ÕÕÞ¹0¤v›.`êÛ‡J¹Êö&pâ×ì°‘…sI€‹Êa¸Êå†F=†»í3Ÿê ”.Ÿò ÎLaO?l ‰…GO'¤ÐykæúQ,Û!è`žf‹—jÿá ϳ²÷RŽy?‹=½)MLÌŽ›ÊÅäèçÖ‰Ì'‡~íxƒìÜ£~<¾?‘{Ã*}¯"ò×é¶žgþ‰ÂÆõŠ­\(ž‘®¬¢°¥ë+9ÊxÍP~‡sOJu*«köª„‹¢:ZêÇÁ^SÔ2s=4°BFîá«¢¨¿£eš÷4>ú=g¾mšW¬[¯iƒ–ŸÕ„>&£íý'¿è¿Ð9‹Óæwº×®|ùwÊDœä–Ö’ôÞ.RqéEÿz¡’ X]8•šÈ„¡õ#þý¯0Ü­}²ˆ)uÊ:$9_d/$óãTwCF\ô¿üKù¥Œx)ÁƒþÑWОŒ÷BÒ·É“ñçÿÁIÿ3÷‰Aß8¥)m¿$õLñ xâ?œà¥4ÝÔ?>5Äøó#õ¸}üú³ÓÖc”_åÏtñᕞô;bæØûµ3 ­Ã±%ûäsú1ç»Íô.èU?×XŽÇ;f.fºŠ'Ñ‚Êw¤ÁèÉëŽØóŸaÂÙanvÚʶ˜¯<ªi©Ë?Õ‹þ¼I‡uóú ‘m‚°[¦Ät:U ük~ž_§[y^«øîÂeëŠÍÍïÞÂÕám½Ðò>¸—¼/;0 OVþKççVSy6¾‰w¬ò7ÔÂ?c©-s^8>ÿuÛ^/u’~wuÞBðhóÄî’+75áVØÈ¹uç/×É"öV³öÖF9Ä- ½Üñm YlòÄe­|©3]ê7^µ½«WT¾ôçsº#x‘é~õïàJd¿‘”Ô_ø¹‘¨+ÒC¾s÷V–’Pú/_ÅüÉÅ ‰'øîò Lh^æ £/*r¿Ø±ì݈jÑz.óuV¨ITáI˜ž‚ºCÜÖK×Y þíF¥k_¢±ö™C¡03š_‹<–ß‚¶97-ç?ÑG‡ªÖ¡âèžq’¯¸özîJÝ0œ¹½_ê½VJBàRyÑŠ•|•Ññx†5µÒ/Ï+™²Š[Ú¹EîRwlÂk#ýŠì‹Sé/ÞÂÈ7¥òFi8)y’:!uAöB‚›ùNA¶h,,›™®Ã¥ùÑ“÷/•ËËt¢äÑàz~6mÞ#>jäû'¥îç´÷=Ñé_™;žrÙ GœpW)íFœÐgj>Tu>æ)•;ÒõRù…´}ð®I¾/Å3]ydÎÊ+x$•wy÷äC< ôµ1UþÃ:ÛÚé]0 ýµÜrbŇ¡üeüÏËmù“«­ö>9úxj×b&[£íÞöF)pœwå13Ó68,j\%sNÂοåÂè<Óƒ–vLp4Èÿ$‰g{7ì‡g΀áë]xMÿu\žk>|ÄŸy¼0…¯R ‰c9ü§—n)ÞÐ ÿ^½Ä6‘¼¸‘·Ú™¯ _ž&DÌàAÐyÿMsSSÒé*3*°º;D(2ëY[r!jš¡/¯õrij=½k âÖHô¹G+#î›&ëû¿ÿŽ„Û¯_-³ÉARëáfË¡sH¹ù€KC¹ )¯*C™-‘VW”î°÷ RìÖ 1™î|û³}§‡+Þß9nþù?øæ#·Z*&é¾6ò—ödTh¬D;sÕ"q¶>ñÔ¨Cñ©ã?']ÌQÒpÚMÊ÷Êîð‰G~bBëÍc¬¨Lw<ü*j5ªß;S,Ž“hΗ¨çVÇéÇÚ_RЇ¶£žm¦°ô,4¬¾.g:„FÉ|ž*Ù™hzhÏÖmðÍ™Í%—£•ÓlvÌ.y´=.‰ZœgŽöox T}Ñi%­~ºx=º¹§ðqªNY‡6b2’‡…)ŸPFÝy’ùt*Ü”‘ÇF^3êˆI’º%s0éoŒûå#C|·iø(ÙëHcÔ’>FôTÔ¼J›{ÉK|)]0áÅR†ûßõø¯"›̦ûÓžw¢¥Ï?Mdz“LöS~2O»®ŒÞLÇgˆ/ÍOƒº×Óô¿¿»i{ÓóG/ÆJÈï¡ò Mì­¾ºwÃ8)üÄŒÂÕT©…Ànµ§°R8)ñPwVòÓfßSMwÚl{cº1í^å8X%a¿èœ¯¶Ú: ×c›6ðŒS÷âGìyþFM kžg&Ìy Ÿ~ÑòëÌð yÑzã»!üßËL÷~¾u«X•÷zpÕα² ÛPgoœ€È»üÚgä™=#GÔç|bµ»Ž‰!¾mdY™ÐrÄ¿[wíCr ’´6ØäE\@Š"ïFÕ¿{rêýÇm}ä­—C24¬¸ºÖÙ'ãõÂ.õ!'BÃ$±|y«“ï|jÝ€üȈ-_ù~ P¶ FÍô$Š¥÷ÿ:4Œ’±?«ÊÊPöfèFÆ>T2­°ä‘oCõÊ£}Ûäw£'®@~Ùûœ)ëÐPZP&Zeä¼”Üý}‚§Ê›!ó'ã\JÝ+h¿Ì‹„gFðâgA銈ŸÌ'%ùN´}ŽÂMx¥”N‘<ðÔüLi¾Äç¾7Á]ˆO-ׂÊa#zü¸7ô´ÐsDï3ª:4—®Ó˜Ê¿4¿®²ÝŸäÆü5ö<Ù"¦r]üTZÚ#ŸûGï6]qõòÀ/³c`ºÃ)k÷Î8˜µ®š° ‹¦ø¥”žÉýŒË›•Þ°ýÌ¿‘ý"¥g¢üÀùóý•Í‚kñN•‰=t?ð$!mi×ÿæÈ¼™sÎw ¥Ë§| VŠæ¯ºP‰ía¢9Ò/vöåÓ mqD„Ÿ½7ØÃ‹èc¬ãk> öÅ!Ûú ÷ˆ—}r‰Gï7o¬R÷²Fòº•CL£Æxµë¿DéÒ4Ÿÿ®q.AÆ×¾ÿuWõÊL—ãWr.Íi/ýÜ»ÚƇÿ7ܬ²ªQ²aç»s÷PœÒ#`x*¥¢ Kó¡âÒEë;¶Põ‘7îØ‘Ô\¹|rSÎvÔ¶)´q¶~Aý}îƒ+MÑÈço\ø MßmwÚ oF+óã5ût×£}§ý³• 6ètPõY'’nûì’¿UÑ3n>¶Bôi&M[~é*.ܮ޺ƒ W{Ä 0œ¹zóæ­=xSÕR»fuÿ”uHüHdä—þ¯~¥¤þ–íÙ[[µ"÷Ÿý‘ür!uIîúf Ó©pQŠ/ÃÀ›!÷GF¶ÿ9÷‰è6^Jôø4\ˆÒkÐüÀï¿”ø“œ'2';©ÛFÓóÚ^JÓãSyO4FÝŸµ!êÞ¡=òUî]ýÞAüüß?”S蟇Ç­Gãªè>ÄwxO»óñyçaöfF6`ñlÏ·9°º'¼à•º;ž>;-Öõ3ìØs$,«Ç¨¾é¤¯6ü@K.Ü ï Yï‡kƒÞmác¸·žž ÄO¡ fìfËáÕÎý`Þ.Cøz/xõGþ6ikCñB‘Õˆéâ=¼”;ŸÁÿJÁË¥ íCÆçzïGv7":,Õgp"zT¤^Nô;â°üÁ-¯6$œ¾H%I×õL·¿ëEJˆ]®ìÑJ¤ñ–fGDÆ€Mèг}ÈÎÎðy²9^Æ‹z‘§¢ Ql„¥¾Ëœ‡®£Èî©RŽë4” å{³ÜŒG¹Î¯!©ù7P%º×¢\‚5k4¼šÅùQ›ýÇ®h¡êMNŠ+òt£Ñ¨§ä–`8šƒgå>øs­n¹¾Yhgd—ÕÔ£{{÷Ú¦rô¼ž}]øÔiô©}t¯”éÃÀQ6fÇWÞ’¹-½‚í F¬Ž˜OöOLY‡·”íí0¥Þ÷dNdÜÿËG˜Ô#c’Ÿ#ó-™?I]\„ÒÛ2Ü È=‘Â9I¿#O¾)•÷DÃIÉýþŸÜ'rWdÐaüãcJ|iþqÔ‘¥f½å• ú‘äÄÐrˆ /MUtMYÀ%ºj±A#ÿL;:¿4jøn:OÓ¿¾Q4`½¶«Ü•¥ç ¦ÕÞºü8 Þ.þ¢ûÆF¡Ì–cå50™ùç÷š ÿú˜†wÚë ÑsÕ²~kH2m‚ÃE9–ʺïÅ­v)Ã=pe¶J) ¢öJ*G†–Cê-å7¯T.¥Ë'¾n´™àî‚…Íã[6ûis÷ID\WÖªJEô¼G£CÙ¯k]úXöxâÅ<^h{~ |Ü’7Þ6Xí¾¯Ž²|¶hEÚ³ÛÇrn&áõ®%÷w·NC6³S*k`!r\W×ò|ìCžDàP¸þk,û©'ñEKùòͧñ£D|I¹eû”yÄX;ž»‚ÊU=Ìsß9£Z5lI˜;jz c½ž‡ºû9rWØv aÝ•¥»ÆÑ8Úeÿê5š[Ÿ‡çì©BóÍV­Oè8‡í·· «QÏõ¹¬zŽp7·¬zƒÞìÒrÞýè—wœým¹Í§¹Ž¾ÀФғ¯ÁÚxÃìyö|‘*ÞáŠ_Oü†)ëð~FªÉ\¡ot<†V/SñKÿ«r{pTñ­XC½&?Gê‘ꇴ¾û?ûÓ0òHžD1¥ŸÑ·†ôGâǨÇ'yà z|â/IöÍ|jhú)j®e¸w(vó©&ÚÐóÚ?œà¥Äÿä ÿ7Ÿi–å‘Ãtÿaõ¤MÅÚ»©¹ö?õÇ„_*3_øÚ@L¿¨^)8âF÷y£Ý7Ζ«;XÀí5ì;½ >F3ýëS#oüCo¸+Ìâ?§CåÙx©LzÒ¤rlüÍ稱õmÆ µ›áEWèþ–I‰1«¶¬Z0¾½ íVÙeAô¤â“ø½ˆ;Á³&^ü T%ÃÚ_BÒSÅ›ºêÒkîŸrè2Òô6È­k/ÅëSYuB -¼²VwÖ/äŒ?^vð:òê®ïŒù…‚sÇte¡xãȢ맣Ôo¯Ëï¹kQ!sK:&Híï>˜X¶Bw5 çžÙç§ NÅFvÒt ÚvüÖKÐt­âh([ Z‚ -¾ª¡}Ãü8©ø töY8ŸúŒî±ÏÏ»ŒGÐëa{‹ý7wzs9aðZÊÎÁ_¦¶«¬/¹°oÆ:noÉž²‰Ÿàô^Ê3CprdÔß“zeÄmH_%÷zRäýN|ÚH¿"û¹Û)<®øœi>Ÿî³DøÑŒþ¤„OzW5üª’Ýç”vßûçI{ÿSø*á™’9‘ø6ÑîT=¬J ª?SDÝ5ë'ï/á¡ëòiù1”©p~ñ¸¿1…»PuquöXHgŒG›Ž?ô”‡iÓ<æ6I˜»XIidÁÒkŽG°¶2vì®)†PôO«ß*°-õ/˜í6öw, Wsü„#ïïWb zá$±Ì¬æ÷~z¼¥R)6c¥"žù.,Ùœ*NÏù]íñKöÿêzSM8²¾â?×ø¿÷ã ×>'Ãÿ³†@è ¼Øµ§|ëߺ LZÅR-Aò©!‚<‚YÍ"aë‡Ð7/wí“<ŠðþÀÜn'ùóü¹qAÄìྦྷTt±¹šÛKk:÷ëò`–ñc$ïÉðx¾‰ÉG“*«|ôhFç—¤|‰P»tÈ©v9Þ|yHß§ÿú _)2ŠÇ$ÍÎ5#³[b‘Ã×>dÛ…ŽÎ—òBNfÀÝ4áä1m“9qóò¯°¾Ü&‹‚òæö- N è,WØ·½¿»4r¬áx?PnTý‘9T•¢QåE«P-º}pþàÔœ™wþr¾ ju¯f°ÝñD]DÜm»’_¨ÿuÞïjÌ54^“)u`¢û}«óöûL[†¶ƒçAû`jPúómè¬ÒÙ{Ö|Ýw÷ó±ï_ƒö`Èô†oŒ¹æç4eZ‡^æxCᥔoÆÈOc¬;Æ»ýTz`òçuùTî­/R2ì:ïÝ]H4ÿÁz£•ɇí'ƒ¸ðŠíîqÕ÷6GV3É;ü‡v=³Fæ}å‹Õ!ûZût¶mÈ).³Qé@Þne 6ägè\óv[ˆB…`¦ˆ"i¯)Ib‘Aéôû»8·p¡|ÆçØ}üe¨Ü¸…Gm@Õ{ª}­CÍí³2×û£¶fîpÒàSÔ‹/,‹®AC•Ž™Qâ~4=:zPÈ-GÞ¹ys·¡mOb±h¥$:Ndp(ð>F—íS×êiâèþ%rÚjï”uHt1ä}?Ÿæ¿r€ÿ×ýÔ©{‚PyÀ„—Mx0ŒyO´zùǯ”ä=Ñ~Žà£ŒOJ§Dò€i|S 7%ù¡$‡v'÷L w!¾ÂŒ¸ Íï†Ü/¨>JÓkh$Þ ›Ñþ˜â§igÕTU§áÁÍËÁkwRùk”"ínAòž ÊnN› s…Ñ…¬isuUèþñÁãÁ(ݹÏSz¦ƒzÃËÅ(½Ñi8-ÕV^X% gMk…kipùÙÖq¬~n¯¾÷ó.À³èÍJ+ÿTÑuùÜ—Ou~,†¯iÙ-å¿ûªø8¯«Ó¼8/Àou/…Ï?]¿ÁÛ8凎x tgÖ$§‘Âï®ì6hCÔôÞWë»çÓýk^Ìžu.« ·çdh‡"I,1Õ¡D)[ý"¤£‘*¹EQ+ý"ÒcvÉ?ãgBæÉ9Í’L¾Èvñ·=/õ¹«M¬ýýüÊ+Žé~>¤z{PäÒÔÝ’õ%ñóÜutPöý‹ƒM *Õ7…ÇÞ™j+öpK/P3.2×nãRÔiMzWµj¢MàT¹Uó|Ë,f§¡9žãd‹ Z3Ï|(k:‡öÏ¡¡&‹xÑuùlFP-º''¿UÉ¡÷†Ê1ô$-¹*¡Ç=R‹¯tcè⇑‰s30"å¬(á3µ?þ,ùÒ *Tü¯ýð͵`Ì·˜Ò¯a?${áí‡Sîÿåcʈ—Òî‘Sêò‰nƒäZ{ ÑÓæhJ¯AöÂúd‹ï½¯è<šJ¼”øŸ6Zªú•Ö Æ)o ¡’2«Q54ÛÄ.ù¸P¹4–»ª¾°t…Îç㮚Ð_¦]À–ý¬—ÎtÌ€b¶h´:Œ*˜ëÅ"wÃD_Áh·ü ÌÖÞ<Ú•ób³ÕnoÀ:wÈÉçEl7¯óô¬û¾r×ïûÁq\O~µÓu8[;ì­È‘†«ôÉ f¡psZÔry:=—æ«°ø¹‚ðyÐ.tü üYÅÂEÌàcûx[Ô/:LokŸ=Am¢’®‡zÉë¬Ÿì „ýÁÕ-¤€¨F))wÄÚIÉ3ÅR:áD•c‡7Z #ÙåfKžòA¼úôè‹§Ý(Òm˜<ªä‘©näT±æ-²K½e"טmµÿlä+†ŒG¯0E¡ZÂãº]KPªªw¬eœ ÌN§5Q‘${û³ª%_öþ`­CM•¦Ý›5ЍӱZem¦¹îO"¢érÉO«„a´8›&¼VÚ‚¶_rïªI 3°€'fb9ºóÎ5­ëBï¥ÕO}kß¡ŸÛV¿òf üÞßZ’xà ; ˆ'M]‡ö&‰|—©÷9ãÝ‚±ÿW¼”MöN /%º|âkJ|£HN ¿Må#2Þ#hþÞT>)¿MöRÊç‚–OJÕÑó2‹èdƒF§å–ÜÐ’p±6:\Aï‹÷†)ûzRüRò9ù°lifßOêapÜÏ·þ¸Œø·ntšã ãKKu¼O¤cðN粘%±¼O}‹ ¡†M9°>¹rßo"°i\ºª1÷ìO¾ ÓK=Ç«ï7/Û§Ä¥/ÇdÀE¬²Ü™ïÜf|º3}µÜ'ËYåO¾dód©Jx©E¼ åh‚Oûm– ›(ø_|•¥Œ³•gf~º‰À÷Óžíû‚ ^.!…ÑEéM˜fÛ0 ák§›¯æGd.÷hc±,bJ™™‡oíGœ÷«O´"AãCÔQ$]fçþœ5” V¦ ”é>5?ÞÿX|™I7m×Y­CvãâyËßù"W[ ߃'‘¿ó‹÷Ó•Q¸Llæ¯çPÌo «Ï,Ò빉+P^r9‘hªŽ´ Ž\CÍ¢ÍÞ%¨u´Ò`’OC=Ÿx÷Òñ64´i0­FSø3·£%eôsqÐ6p~(Œí:Åb_¯ÞlŽnåH•FùQô”ó™~âC߉ƫ¾¾?Ð?¶rBoK)“Rï=Í¿á€yno9ð&%\š¡SÖáýš ó‹~T?"xæTù¢Séïÿ /e¼2ú—’y•±þãgÊðüOßo¢ËŸÊ¿”à¥4}#£¿7Ù7©¾ÈÈ3`Ü/É\˨ǧùS9OÏ·‹õðßü‡gGroîÈïp_à T•…·] Z—ý\÷$ºžqì´KFÊŽõÇ´;"•ûD|jHN[ì5}¶T˜ÊÊŠ¿NÙèÉäM]C°È8“£µ~5å£a#é§9ûö\ØýYqëŠÖN*ƒè]ºîWò·s­­X“÷ò~º¯·K˜øP‡-•cãÅ*8d³ ªWö>½‰Ê &þ‹¡‡}LBÖ_¦ï‹UrÂÕ§“ëôèà÷aÄ›\+U3rF¢å-ϪŽN$猽Ü<ë8R…~Îö<1 éo—7ÿ¬4Ffßž-rÈY¤1=[¼¹.b]‰üÚšKÃ×XPø;ÝŽÏEɱ#Jy)ëPÖpj·Æ7;Tº¿÷xøÕ%–lϽ¬P{ËfW‰=ê×ÙnN[ÂÆÕu#­³-Ñ,6Ü]к­åÇF™—¢ƒ_\ô¸rº¾K,¿(*ŠË<Å“[зwš}bÞØ{?x%†–ùÚÚcäì§={~œœ²ô~t<Ÿv–šËÈû˜ð¯I½\…ô5FÜt*7ÒW ¿”Ü× ÎBîä@ö>Âk¡òqý–çS/ÿÚØá1ì=ñ{|O¥ùÁÕ®»óà F¨­ûÌ]á|Rp÷6Êîô[¶ói·^íÙ §#‡ÛJT‡NÁîÇu9âáîU(ÌÈ€$UE¯úÓp¾{Êyx1X‡Ê᜛vÞû”u(”ßoxÉL"Hs ñô¦éÖô‰EÈ Ð¾–ð*²¼æi}ƒ '®…8< ž®”eØ+B¾æÏ‹„K…PpÑ~–ù6<þ€´ dz”µ®TÁ ‰Ç§x5áex&ý ‹^x5ÂC=Å{lUHÐAɪÛBëÙ^Ü{Ö*IBv{AeJ!©±£;T5Þñ´u†rK±kðvÉè#Qj¨m):k¬u6£…Š­ P?±Qrƒ%î”’2†Fßœ ©ŠSÐd^“v÷©&´מð´ã€VÅx¡”“íÐfõê@þÍ·Ð~…¹ð…ÀèhÆ_]Ùˆ€. _‰sÏ@÷ý+땊¥Ð+NÉÅúúšu*޲3À@¬šÕÜ©)ø|ô¸ë ·3;ïÑ÷eÝâ0*øû-ù$\?òÏ\ Ö¯ù—_ñgÞô¿í_ŠéO ßæipûËü1?ˆåàþÛ~28]‡½÷ ÛÇt[ßayò?ó¸œ)Ö¯Áæž0ÛÄíÓ±½Ÿ>–/=ð0œ8Ná|¿Ãt×ç2ˆï¼¯ Ë…·/v5À)¯Ö˜{æ¿wt;Ží§í¿\c¡"fZ !Y†ŸY}…‹ßè#ŽA…±W™„;¦4¹†<‡ˆæË":ýWpû'âú4ºŽQï„BŒ‡élÚ \à×нzb~spøÃÍ€ý†¼cW¤&.Ú ñ´qÚršépû¾¼ÆèUbHð¢£;¢n ‰»ãó"Û™!Y7aŒv¨|çý¿ünêa¾;ï'Ýι¥Uz_¿ÖËxö­†'V/0GàÞk‘Ûjè¸äÙð7žSØùëÑ]^/á¹ß‘y®/û ßÞ¨;=^dQ”K>€W'÷Y|•‰‚b÷ÚW©t«PzüX§¾¼Éù9¨ÒÌ·ó>D›ºG¶œóP]´ƽuÞÊRœÐÚú¸>ˆn¨ Jί)„÷_Ügð»àƒŽÏQy¼kðÑ^‚4|õ4Gˆ\Hº-›*µW—^C›É~7î­ƒjƒ}Ôõ× SäµèÉÒÐãŸL6\5Rò—ŒI;¡êŒøéœ;'¡SM9ðÁ”ôÜ‘8z"ì%Tï©7«ütªîËE LBç¡î,òõè¥?¬½çTÛ„ZS´kA•ÿ´³AtÒr•?yå½Iú«YePõš—‡§Â ªˆÝ2Ød¿B§K¦ãAw!è}vˆùJž7T §õõ)<ê²%6ß ½;›g°öôV†^_„ªÜݪÖG Únhì*Ct¼Ê9âÔS=ŸèŸLUBUCëÝ·¿Ïkg'©“€Nž¿Nd‚Þ›øt1¨º7ªÖt*ªwýÏÚáôN.Æ‹únˆØ×ìË‚ªŠwjx£» r£©Œ¬–:hd^Õé@ïéjw¿¶ßÿ/.¿,­„¡êÍ«9!è˜[ãºèý'‰?…˜B9~Q&ã ¨4z»úÚî÷ω<½w³úwQ•WûùCù[W…S¿CÎ’?ò³tr°K_€þ¡ ¶×ðfï mà ”ÎîSæ«€.¢’¸„J`­ý¾o+”ÈU p·Â›‚Cý#g s<ïXâ'èH‚ðØËPöÞqrCŽ*÷k¹D¤SC§à,ëjK>ô}uÖg¡Ý‚yN›××~ÿûÜSM=‚.Æ£ÅPC\Þ‹¼_ ¤éÂdÈ­³¿ÏßQrQ¤‚Nõa“A-èw¿+D‰z ”‡U„Ô *-lL\ÔÏAçó³“ÄmjÐ÷úF¡ò‹6(ûp¾ÕSå TS›Mîß ]« \ÐᅦU¿ÕÊ~à ¥i†Ê€s/æ,Œ STþÝàî"èd2Í %øÞ“E©Ÿ jzŸ°%tŒ¬K8fC?m@Ì·†7ð&ß/KÁá÷ÿß•/7º­¡“©­—ÌŸàn£ˆäÖÿœ#üÏ9ÆístÛçÿƒ$|Ùí„rRÊÛ‰T¤äU0¤É«J)(+mÿ v…œššöY^VVö}–û_Ÿåÿ×g…ÿõYñ¿‚Ä‘×Üó°÷ö™#¯÷‘cÝÿßYݳ›ÿ‡g±¿÷‘ãŠÿóh ú]lvúv}¾{>ügfƒ”o½?FuS0 ½’û†^ é†ÈÞ)BÏ=®x¯{ŒžúÌE8òe,«Å2­å t¾Ê¢Ò{ (Mù’´ëiô؃vFV:=òïRõŠÜ…R|œÝdJ Pbq°¦ #nÅ®{f£Ä<’<Ž^M޺讃*™™C’ÇÐÛSŒ‡}Ë£ºìÍvVerô~)þB ~!úÀ5´Þ57CX]ýF-Eú_CM±Slå ¹¨éCãKêÒhÔä@ý˜!eçüG¿›Æ/r¹pÇÍ:1ûoy{¢–¬b½É¨õùsg[©)ÔÖ+Ãòê1jÿ&ö’kJuL(}a›ÓFóFÂVŒi¨›£ó`×)„ztü4^òC½—‚éNŒ@}£th‹Ñ@øƒý®Ç†Ðg­—Õ>¿Ð »ŸÇeA_ù«ú¸j¢oÆi^KÇ Ð÷ëLãçyÑÐW'MÇ t4b#Ÿö}…þ<ìf*/…ƺîÜÒxfŒ&Ißr|ЩAÓå©—ö_C3 qOÎø£_ƒnš+›¢yËŒ»A'Ð"ÑÙ’§ëWÐR·Ük¼çÞh•PaóÕèZåɬ¯¢E[ñ¡Ç%neþ·>³—s@xl].£û_Ç»m¹HŸ†øRµäùrÐÉ÷E@eù$¼Â±hî6üŒ¼ètç:‚ï5ý@ ñh(0^yþö²r>0¿(ެ>¬)Çöå{ãÂQÛ7»ØÂ&¢žÜî ¹ö”)à c8AQÚ |Ÿˆ¬_Ÿà=B²¥íò Äz_öè:ˆ0ú2\Ìö±ƒk$ß¼ARªråüù~±S¼›Àqäö”VóXL‚RaÛÓŠ‹å©é%‡@…¸æŽâ-MP“{áy4npçm”Þ$¶¯îƒH6h‡} 1]Z_Ë‹¢ë°»(õ—‡»!èá‚à'J`¨ú£Ó¾ZŒYÃC¬tÁ”l3ˆ©Ì)´C®7—‚%«ÖÊГ3`•p„ÔF}¬3'=V4ƒÍ,Y«Ž-ØiȯߪÐûäÚa²áàÈ‚Á)ã’3´;/îBpãÑý˜u9ÜÉ„0ÞŠÐô6’2ðä'•ݘï%‹óÕ7w¸ÇíÞÓ µ‹:·áƒ:ÉBoȆ Âª“}ë?¾£¬â¹=釠 ªõN‹kÒ8œ=Ð|/¿û;oÉ7ÆËÉ£Náð–ðžûý_+v=†Ç'MEQŒò´Îh>ô’BÏ»$nËfæ¡ü$b_¡w‰‰DZ|`Œ»o_†[ŽÆ4ÝÃÝ÷ãÃá#¶€ò£6s½¢bQîùç üBæ8ÞÄp‹­Ùéò¡:Óµèus¬÷ÖÔŽÿç;ˆ­ (PÛÅψ¹áx¨ç•Î9éùÔOt¹«2ô#úóùæ´üÖ_|ƒÝ¯¸U•ø˜ðþuþŸ¼Ä*Ù~U¡Q×{·´0ôãæI8Àã§ N—׷ߢÉcâÔÝ´Jhš3™‹ÕµÍÐy^~äš5Ѥœ·@¿n¨;æY£¹M-|é8<´pÜ[O--Î^Ä›°¦GË¡OžÊ’íE«(ŠCLº­³»ó|+XDó¡¥£«h«ŠUïÙçã€×P(rÞn ðû”l¹Ý®!Kâô˱@äß<]hTÄŸc†ùM*€Ô(³ör:e~}2Á_ bü$¾¿€ o‘9eYh<‚u T]ÒrÕó›À Þúé 30öŽt'(·3 w“_X€EÅÎp|/°|–v¸±Ò¬z©®Í"ÃÀÚNÖÙðÞØæÓ^³'ùþ§Ê×›ö¥Æ{ïãÀ;1›ÝÕiüµW¯JÒƒ`›ú;s­,ÖÙypcˆ¨'É’Ö_QªÛÑ}NŠƈ†Ä“³AÜ~áˆj%Hîx7ßÒÁ“#Þ7ØA¶$>“é’ È{s~ŽÍtE‰÷r{ @©ñ´Ú\Y ¨Hþ }¿gTK¬¿–uêª#vgA“n<Õª´¾i·h±­4²²~®žía–G ïûAg2)OÛò"讉²zŒ…Cs;èï \Î^¬Y1…>Ï&04H ¤Õˆ#¶WÀ8hø­Îˆ˜¤¨ì¶ý@¦#õGÎY ¹ § ›<%XDÊ=Â#œË)³5/û`õ|ß,‰Î°ȶëü`6zTŸKÛZÁ–£±Ù´{l×®?Yk »‘@‹b'!(¤x ÐÀ¤Ãi6¤îŽê…§'ÐDø“2Ÿ3Y¨¾ÕðÝnürTö+øËݼk8^Âø«+ÿäK¬NM55Ñ•ÒÕÀ­Ø÷±ë0<æ3¹m64Æà~/ÆcN1ü¿ñÙØ5щ*P{@kl3ªÂ;: ´…ã«·ªêw¿?;€ÞÑ}®±S×–½¦Êîù‚>4_°õÝá©£=DÚ±=¨µEàM)ujg"#¯{óuä[2W3¼A]~ìÎjþ¨ÇýË´ÕËpÔwBá`UÉNýyG`ÍX zqSæÍ ¯Q‹ÒÞÈ¡o÷ê¢!†ö®u¡aÿ/óéÒ¯ÑH7þÜ ý+ôáâØPw:›çv4« GyB4ëÆFhêÎÃr>ûègr'¥Ë9C43-ͼï¡úeÍ#å´‰æ~ÛWßìB ·÷ÓàµÄ¡%E©[¸¢Ñò‹ÎÛÞìhU¦ðÅçÊ5´6ì·Õc3„áðÌ,gLíìؽû¸ìæÜá-ð¤âP£qçD Õ#¢8 då×½Ñ ¸cã·w(Co+Ð{)Õ˜ËéïÉË@óSPÁŸèçÖ[ûXa’f¬+a˜w“Ö±k~ïáPü`ÿ‰’ËoœΜÇn96Þ°kY¾¹Rƒ¸¹x^ÄhìžË÷ýlu·ø«£ø%yàç}t>ªWÖl3Î 1°DxòÅž ôNYˆñ- ·Wô uVw¯d~•6Òñÿ]¯ª«rä$ü·À\•H%P¼zšð£îWP¶# q-UŠý%’ n}õë±oä j©ûÝŽPóІÇ01h§Ìç§áo –£ ­°[É'Üàèݱ63–¢¥'ϼØÁ0ôþµIÕ0¦2¨hñ$“ö,‡ù¹)0Ë8›wç9X$¢ãæqÆ`åP(­ü¬Od>›Ç[õßä«ÁvÔj¾‰Wìuê®çÌÕCá“ΑNpBE%iûíaÔ›Cý§/ŸóDzLŽR­±e;†º wµË~¦ÄÕqÏa¸ÁtáŸzðÏz«S±zô_ú«1|Õ\. žtÛ©ûîfVÐ;ýoˆé¿íãUì|#¹«}ù³uû<†Çæ_RT<ɨ¥Nlª:'µÞÖ!C÷dQ›—kg¦{j·˜|Ñð:upѯÝ\û]·gyÔž¢C]fi?ƒ¬¤Q÷3îÁ…9^ÔëKrÏ%>õ»M'.!EŸŽyŸ5Ë;‚¾”õÔÞ£F_™ˆ,¥t oN^ÆÇž ïï™ükѰ¡“ÊåýP?%ƉÆH‡ÎØ|qÀñïT “lÔÝ14CNþ²ªÍú|üIXúµ¦«8o£„æ_Þ-˜æ@‹µÔœåVhEáÖ;Ñ›hm²«•k~m¶Û‡ǘÞ¹’@È= È™VV’¢Ë§DIÕ£$Í…øÞøc [ô¾”'“¼ýzRdíÅp-ЩQ\Éú̯,•\À蘒ÿÈ·˜õ& öíÏVÇû²¹Ýï=híUQR.p–—²’†n µ¡Åß|iM²q/`øR9HÊ?„ƒ]<;š ÁÉzélÉ„qÂØú [ éäÌžÊoë^„®Ímòùs[Å@Wy`AL0¸•ÿ¤O¼ L<ïåÖç ‘äô×R`c=­—M A»ë蚀ËJeÊ6…‡W~)¹­¢M ˆÜCéO‚pjƒÅa&ˆ™l\¸D’Oh§nÓpƒÌ\Ù ¶â!kïäW% ƒ&/Ίჲ¨ËÛ®P-øöx^)4‚Æš3x]Ь7´ÕíÚ»ÄÌb {ÄlqžôöÍ È}çÅÃS`t8Üg+0LbxžÌYÕzLaå°ØÖ)ÿ\ V#í”TÊË`cCNE{êØöç„?ûÈÄû9ÙºàhÏÉÕ±™Î&·•ðO‚ëÁÓ2ʰç•å£ršðàèÙxgx <ýç;®[$ÀùCmJŠK% {Åjj·¨°54e¨Œ¡qÍ}ôýi ¨ž÷…ðÄ)KTQÂ"–|•dÈ´zË´ãô`ÁiÝé™Ñ¸ºë§bu*†Û‡Q)ï|ƒq+vÓ™ÕÂ~qª; ÷¡ýUKÏQ KdýcÇ Ô.wÎû§4ê¢/s -†ˆÝ§­¬ÙôŽhðb¢Çƒd48Ú ‘÷¼ 6X¸/K¦ãêÅAQ‘κ-ØY±ïo_Ó4Y҃øûý{‰ø¥|4Ô–Ë,4FÈ9Ë´£»#¨Ÿy£ñ@BñúúÊÝè¿ûÍ.J]4ãâÀz};h1¼êhˆæ\W8È£…ݳj¿ëÏ¥ ú·Ϫ¢•Ï<‡—zép:q“ÐKSkm=WoY£¼šÝ‰|/þ× ÛË—z€P(D'áÍ º>ú²¨HHX­¤£4yÿ¦ö<Ë«MÛdü®+yï«[uˆ[ÊYy? †gÌ]@_ép¾]=©­>Ó;±ˆCzqÀüÚ÷œ£Ÿ°ØÖ´½­.–ìË«ïsUœ¾ë˜ÀK`S)qhð•ôÛ?d¤‹ßD’ÑN x=jül üz^Õõ~@>G©&Š„Ná,íáC4¬_›‹@DïÀ¾‚Âs j~‚çÂÏ9‹Œÿ^Xv $$ßÔFÞ{ R.N_uY-A–dr¾£— änzh8 Æb]ï8¿ƒâê—ÍÛwkA9Z1;`×WP¥–›K6Uµ‡ê+ûó@Ã$å þ…|Т¡®akð4F¹Æh´-“øÕäƒÎ!nN±õdÐ=6OyBl vŸyO[yyôBµÂÊÃ'A?†ŠÆä& $úð'>Ôìô¥x…D0*8vÅþ·sød¾Sú­:òCûÀÌ’9Ûàk˜'…%Áë‹q¢;ûÀJ¦sÚâÚ;°šrÉ@ÏÁº0”Ö¥}lÂúƒ?p§ƒ­ë3Ö„‡*`‡vg5VÀù˜€3§‹yA*ªÿÐ}B`ÿÌNQöû¾zòt-ãð‡á «+±¾ V·bü†éÀ?ý ìãYì÷a¸Æø ×çü‡NÄõ5vPz¥ˆjðJô]Žâú>8ˆñÕ6Îqu-VŸnû- KÜ—(ÎâêR¬íÜ{Ÿñê>¥xÆI××ìOø,ç^‡ë}ypÌ’_| öRRƒÃõwÏΑDs_44ÃÖe>àƒF²ËÂÄÕúÐìÖð(m z÷¼"®P¾ö$±Ã(fÒw§ÙÁ˜I’½mò:˜ Ø7_L³¼ù´J­°HÕ=õ®³¬Ü|®Uu ’6ý®h°É:®ôS l§Í̘xÀÞØIl*N J—Ÿûå·“aL#xÍW¹¬yø¢ëöíËáoåÙéÉIÔ5ï0‘_„ã+LbýO¬ïò§>Äøî_ú[±ë1übõã›Om*êq}—ÚDëVÒ•µÿŠ#œOˆ­2Ïãå¯'ãê÷t+>àBœª^,ãú5Å )´ŸÆQ§´ YÄWÔtøtCÎ<Îgh=Óܬ鰣·ëÉŽ¸þ&ÖÔÙÕAþó€ÉŽï€Õ•|[Î ÷ OЧBøÚìЗýçN¯p=Çñï7õá+";>„£;gEye /gÛ»ýß3®£LƒÆ[SŽì¾‚&ç¹n³ŸµE?­šÜwÂC3ëKÚ‡>“ _Ë#ç„м4ûPÈ¡Å=Îb?­5ÐrÙB«–g+Z³&xz[ ctS L@ øˆ± @ˆð ËÍp|Jæ¼/ èžP¼sc Ž ü 9@¬$Œ‚GX"îƒô…ɪ k]Ëé¡x&ò¯¸OD©Äâ uR[ßPŽŒ(óýz Tk'‚Û£@Cv®Üå;h½ï^{ y´QÄôÝ@Ðy§üär‡ìv{N¼ùô©Ýߪ΃^‚öo²Á¨âû®o k`R=ëžf_—/…Ôˆ€%×Á5™ßºÔªÄ‹>F!¬ç&"¯°€­ÛøÌ¢»ö¶bËà#àp@O–BÉœØ-š-&óÀyRØÏø\¸~Kx­ {6]ã("3ÁƒïóÜÅPðtá?¤%ì û²Áa‘:¼Ý;®Dæ½ÞÎá¡-ú íV¼8T„:æ37%:ëp¼„Õ‘þþÄÙ¿üBLÆ+kZõ~ŒÄc×cxÄøãW,7ð§îÃŽÿô ±ëÿò·¯Çô"¦±ÜƳ%²—ÃÚ¨EP©x ¥MùŽxµêë=&\Ýü—¸z4`šC×Gj8bÞµ°ÔŽëáô!†k¬/»]oÿÕ— ²/ÒIŽÛñg†Ml‰ŒjÐà Gö¦ŠO¸þάíJB¡3šV³úBˆë¿b¸*’:™é{ÍW¿3\‡~ –ß“[û€æ»Ž´-Ø¢EÊÁ,*_´|²ó¶Dá4ZÝò‘x—J‰6²ÑºU!ñÀ«ü&]ÄøUÚÖïE¼A/ËäráϛނX>1aH~ù(˜³ò dÙ –¼Î¹±dUõM&PòÎÝßoÊTfR†@Õkf¿ýl¨OËšU=­§™èã™EжIž;CW :“”¬G ñ`wšÃ(ñcÐ-=®†g¨ÖjÀøŽfq9å8˜öl6ôˆ k3âu`ÖI1ÀÖ+µÆ€mÜ ë ÚY°·¤7° G‘I>pÖµlnï˜׋D\’…°gfB7þ/x˜=Œ² ÔÏòŽ’2ñA8?ÿIjèaÈ^Ÿÿ$¢Ó ì/Ë‹E“,Qõwtâí³¹‘1+è—B”G#É(š§¦PŽ©B‹ªVé+sgÑrìùž{¤ÐjŸ…~Í2ZÿZw½y  m¾¡r¡@ÏÏ}ØÍ‰ÿÆÇ£ìÃÉ@ð¹ÂÉ:ºˆô\¿Îêçqeá$aÎŽÿð½P|å#P40gØýjçÓ[%@ûR^‘}ž è&ƒ¿Ò†^¹±Û»o“é—-S<`Nkâ ¯枤ÍsÅ€Å>‹Š[KXf…O/ÊOÑáT£,ÿè—–ÏC`WàÚ²ËÀswxE³Ãø ·]»0¤ìnïσàrý¢X \)\%ìÒÕ= ¢Ô7IUcg@,4ÃÆì.H˜ž‘.s®©¥ŸY5•ãw4¹OGä– h@¡0çrýmcPò]c÷“¸ Êíü3…ïÂAõ‚Îèñ´ZP‡MµÚïž É:¢Rc ’ûõã4S M"LÑÙš:LúŽ c½ +Mi«¦V»mñ7mK"A/<æYS è—>ç†Ä D'*ÁÈþGÏ®·!`œß'!¡¦Ìú¥Á,L¸LEº ,xÎ<<¬r,kŠ˜Ï2« ¯/üµ`ýd¶'Ñ)l8%ééÛmªÈžGŒÏÁ®hÃw·³+ØwNää;Ž$×ÏdÊ1þ®Kã› !ˆE0Nc¾¤n¿Î?`É {=ìÎÇCãOS3®õeâú‚þ0~Áp„Õ•˜ÎûÓ?üWÝŠÿ7ˆåf°úã/Œ·°þ¦«»E›ÂŒq¼õöGíeíÝñ÷5±úv»^ÅêY¬?Ôêâþf©µ-Þn©%¿´Ã[—¯u™/|ÛñmË\fãp~â*ʪˆ4ß}()ü’ìÝúSmà`.uûŽXδ÷2‹ÆŽ>,&1'æ)GÓïÆÆ ³hFOœçøåÏhÖ”üþ™P\ŠñÚBš‚ÛäÓ~OðëõŸjÑZ6µ°ÿ·B´±6=¾Ë.ðtÄÛûÎþýg…Ö¥/wø­ÂݰJ-Hì7ÃÕí´™‡)ýY+-Y4:Åù³5ƒûr¤4vóe#PÏe1hž½tA/êÐ5`8v7[=µ˜JZâÔ†e€••Méó¦*°Ûrü P±N©ô 3aWÁQ©¾cá8ƒ§~´…eù8ðNþPä˜~Ûþ³¯³ƒ Ÿr_‹“&ë.ïó“#ÑîÑã ¡ö ÷Ì&ZÆ'ý²N?L *¹ïi‘ p¾Ýƒ‰.”äEôÒ¬„A…̃êD¶¨n{tø‡h°¯œ¯g0-g™›Ù|¯Ê=‡bå~ãSØA«Æ§tÃnÕ÷r.ÃîùÒìÌHпpßHñÊ0äLþ±ŒúO¸?q“ü­Núªv0+Káëb‹¡y3¾(z°Jö <Ö} »¾š:­ÜÓP¯OwÀîq’|jD#8È(ê§„–€cï} ü¥npNãã;î±ëcä§À+&?PÐúðVø ’Vñl‡9Á±6ÔÙûªOßÈ|Gnó†? _ÿÂÙŸ¹Ò?} ì:Œ7±ºë¿`ý!¬þÃxË`ù1 G˜ÎÃù†Øñ¶>üË?Ä|Çm‰ó±Ü)æïkN]™{~ µéÓ¥„Õïøˆ§}«‹,Pg±ÞpÂ:êV5qYÑ´F=¿¤®…ýÆçûDŠb?c4P{Zlãà>·û¸8}ØÝ>Ä2;ކ ÔïA£W8³òјõ‹0êÓhB\ºþÛ94Å{섯<úéñò-Ëþ‚¼¦©}N‡¤ Ås+7Ø6ÑòØñsÌyÕh-øòÍô+×Ñ&{jwèá?b²y$«¦3ŽÉ.[q²w¼Bf7ÖeÈ+¸ îäq²î›\Z»+ÁŒžÇ€^âñ†ô ` ס¦·úNŸ¤Ù%?+ßM"ØÒ×¾¦†tGuK5'ƒ p±„Y¿ž¼™çlÌuÀ§$öÁ'æ:ð§ Híï¹ ‚àhoSx„ù_+³ˆœQOŽv•S4 Á<×=u¯¤•gXR.€l‡yR²¬È'³ˆ¯ÝaÅ£™BÝŸñA9|\L—xT««"Éü7@Cªü½m¥,h½£ÒÊ mCkjMm ЩYÑ“¶ú­]øNÐfýÖ‡U…ûé„À ‹ÊGrQ×5©ˆí9fŸ%ØâÏ]KvÙ¯ÌU`õš:fã¢6XÏFjx¶.zćØÔÀ®•Ù 5Óökiøþ NÌK/_¸çgÎÜßK®Ÿ­Âü`Ï*{axðŒÏ,OÇtËHú°/3¹£Å;öÏ—¿ù ¼$)&üAÔÍû¤ïÆjÿ•6EvGø/}ˆñÜÿ«ˆéClýSbõë¿ôá_¾!¦ÿÈ™bÇù‰˜¸ý\ù—>,½ê€ä8®_Ti9q_zïüޏ/¨¸ßt‹kéo}¸]ÇcÏf’ÜÀ-Aý¿ýèÖ<Þ“_wú°Û8ÆåÚš’3¾ãêëÁܶ,‚'V¸<ÞclÄi4¼’-ĹA~¨©5â/¼Eマ?_(Ø¡©Á‹ Ç )ÑÌu&§ê>è—®êS3š'NM<Bº£Ý™ÿ33»úµølôcG´qk™¿¸jðÔJª:=I¿æÛ5Çõ\¸9=¡oîk ½Ë7äR&¶R@¹r&‰ŽüÐØœû¥§·“ŸÓu}3w6˜$9 O‹Òñ‰£sÀ¶7u¥g“8Ú׈*Þ®ÂZaïˆBàz{Åd$øO½=/T=‚)¹†§&nˆ³¢Ož$ˆ3T•M§‚ÔaSjIçß¼¸˜»KäuBÎŽÔf¢!9Å&ÓGP>G|)ÈgT?˰ß.ŠÓó ´ã¢€ôˆ½µŽs‚v†æíùß@רîÒQ4èQŒ}TO¾ úÓÙ«g3ÀpÁ/R¡K Lx‹Ï¸°QƒÙ‰ ý·R»Áb¬eôWòX}ùðÃTïØxìž;÷ìõô?>û\™Žƒc™ãUùpî½êÅÛ½ nÂ¶÷' Ñ—cSÀ£àM‡x{ì“v}8àA²»yè^‚숩xý°ùõ,0¡ñq£ªNw5\ý†Ý—ö?pþáv?õO}ˆõ_þ̹a¸Æp‚ùæXë‡b>B[iÁݲ‚hÔ)¼k¡ž'u‘ß%ÉXþ‹w°üÊ'–Õw -рѳ‡[^³¸ûó±õ'ëÉ“dwEpù´/}ûOןù…ï3 ÄFÞG_÷”Ê‹µß©/ɶšÓðwpðGŸsLQúVS©šë¢òŠuÝÁƒÍ/?#4»ì^sêÇšëÕ´ÖàÝØÁáÃta½’hµì1Éõ ´þ!l÷>²)´ùLŽàqàÙ–NZh~°"-'ƒ"ÔûF”’Ñ‘ö£† ŽÆœécW›=‰@v«äÇ.—FǵžV´ÏžUk%ÉìäÒ_|«+F3“+á·ûÉ]âzï›%`ÞsÙh2;Xt_ÝhV—À ñ½ÀvÆžöæià°:,Ey!våÜL,/çsV’J࣋þDÛ üQTtG+߃ ¯S‹;v*ʽÚ a¿­;tš-ò(ˆ_ŸN-§éܨ¬]ÀVÂáü-—H°SjxÇi|ì-¾áŒ‡€µ˜µï¶à˜gBQS(AÖ­ª»ßæ´y†Úñ²dàHð­`pqB¶!:‡(q~–oÃp÷§_øç†·?ëUlÅêÙÿ–3Åx Ó§XŽà_þ!n.b›·0}[Û!#î1ºÃ_Ûó¸Üé¶ŸÕ©XÎ ›‡ÂÍG`u(–cû48Ø?ŠÃ;.W€ÕŸ~ïu$T&£aê$‡É%4’¼ZKΉ~®'²þ˜Fã¼~æ•Ô;ý,?ΣT|bè$šµÏ ›÷B¿ ×êt•ѼÚIÓÁ. ´0Ò+¿«C -_s—=DVê¹Ä‡>¡uBŽ}>>hÓü‘Þi2À³"=à8øç[,-ÞAq„—Í b;Óy ˆc_^eÞ¤fq#ŠQ@Ös¾`Õµ(®Ëòæfú•ÐQÎ*) É³&‰»b toÖrè5Ã-¿ Æug`º©œJ÷å°¤zE3*Ñ[½ò*4žŽQ‚3#صÕ[ߌ÷xäЗ™#»·ÎUaùå$ðSøÖõÅ9ƒÀÑēӾ DÒ²\ÛÏ Â͆¾oA´çã:y\GZö—ÛÞ -»»À¥òÚ&© ðÍIjŠô9(=X5ðç•°ò´‰°xP»¤Â!{4rÂû¯Y‡‚ÖÒ"·K¥h+ꅾɈ°Ë©û(~ãr‹-‰ä3èíßÿíêè/U÷^ÃÔ¥ïAq´`ìgòQÂQLÍ„ž&E4ƒ¹·šŒW=XÞ”Ksó«ÍúYµd°18ãR©Þ¶ßçör»‚=_Éübp(6?ï$8 ½ß›É.5+n‹‚›ŽÁoÝî¥PG(ñbøØ¿fOËo߯ŸC‚‰~ß$q>:Æs¿ýKþ7}ˆÕ«ÿÒ‡X® ó °ú—×ÞžÿøË‡ßöÿþ5wˆÓ‹XÞtûÃnŽó¶û5-CI¯âÆ{wêIÌop\8[“À¹ƒË?ýCÌwØæeÌ?ÄåÓníÕö÷ºˆ†Üâ/›w7àæqõ%}†.Ä¢‰~ÿƒÜ>*h*ýcä5wô³xÿÁÃÎ÷Ð,eå…¹@¿‚/º…ÉÂár^¿¼–£´£o¼Dk$­ûû¼ÑF²7=¾sà‰’9pW ðŸÆò’|¥Ù™g²‹¼;V” ¤{©‡Í¨p~ecèeê‰s@c òpÕ. èn„ÇâwƒêOß»TfÀÄp+¡jó0/Ž­ ¾ÖA»Ä ²^`Ÿ_xäç:»ŒŒf\gý€ûsÅ;çé‹ÀûAqùæ’ð+÷߈S¸ *P =¥ÖÓ.œ1‘ŽäKî„§AütbÍç Aì7ù@öHȰ¨Ò+—3; ë—Ö¿ÆPƒ2q ³‹üiPU.z+" ê·݃_I‚çõê'w´å|Íz™:n÷?ž Kÿ·—¡çöù€ÞµÙ¼ü“M``nÚÏXrŒ„—Û«µÀ„³©ç°˜I¨—ŽÑ‚…}§Ùi°2/ 5yÖÁïõ.Üò›þþ'm«7ÁÎ0`Îãj9Ø7¢` [Ç“…ÅÀY¹-½9÷¸îj6y±Ø {D²ê&"Ã^çèPáçNàÑ·ö¦dñ9ìã;Jûu¬å 耇ä:åêïúšÄ3ɇñ ê(ªNÊîéÀñÖÏÄðƒéÁÿWÿ[ÿÄ#Æ‹Ïâú¥ÔÁéÂ?ÖåLqùÒ?õá¿æ±œéýX\¾ëÃbüºÝ?žþq¼Šõ{¶ç¹°9cÜüáö|ñ_ùRLbüz€_õ©aç¿ýßÂçÕÿÖ‡êÓ ~â»þ»> ÐKzGfxlFÇÞ—×ã|\¾t³òuêúc %­c¼« äFÙ{>“e¼å¸òoFC³ç'Ý‘œû3×׀߯J˜J˜X\ñÂÞìÌ:à㯛ÇÛ×JyéÀu¿h¨ø¤ðïVYLî~_±¤x‹¼¹uì©Ï=aµ5¹ÚC âx+ãÆ õ eî%{Q§dœSs.7¼´Ôå€È  (;ÅHAÊž×Ç*;/îôoOvø]ÄosøX¾>hß:¡2ùftænjþ2ƒÝóZBÅâ÷A¿ót§«Ó}0ìâz<&øE? ïÙ¹ù±ypL©;z÷48¿Êi,fm7üC߆ejÁ=ä£zkðH¦³a»Ú û˜]ŸÈh'Áy¥p…g¿y~ŒÖºù9°Õt›]ÀCコJ±÷ñP’1{Ñ’®ÄîwŒ±¾ Æk¸ù§?|Ã?ç10^Åp†Ý÷˜ßðžJ e}Äñ[ëÔm¥I\ŸË—á|áå-¡Ø}ˆ­ÛyŒ—pëöyœ¿8}3—óÖÜ\nŽéOñ1·Ãq®®¹E,oí1/¶ƒ?4é³GM&Ÿûî& è§P"[ZDšåˆjY¤F¿>eçÛD¦£ù ö™„:´$䦍ܨ‡VŽ%âQ£µ‚³-'B_£áÌ£è-?à õ_fJüÓ5·Öà´ßúî£; DrÅ—z|€øïÍ6ïs@*}ñQ÷*1w2UÃåb¨^§ðGú™MI¼"É_ ;8ÆwY¹èÃT“¶l\!AÖrÄ`SÎÔ¶.ŒÓý“O.æ«‹'öÞ–KÔû‡)€µYÓC2_ØSœĪgçH7C6þIà>£õc?1ðF0-ßÏ=ºŸÎ?ÍèŸqT ØlÉ^;hÂÞ²ì$A”G[,¾aÄFÍ>Ú å‘”~P€ÔK. ãàY©ñ=w³ä,®+ž6jùêžR EPÔh}5û”jxŽÖí[goÍç9‡AHV)ã×%PKi·”š‰ÑÑé{¹E¨­j~sm]7Ûݪ cn:ݲ ºû e7D`wØ#÷û]·A/÷Ög é"ÐüîYÿc ¹Ë¹§^ø€‘w¼v‚¹+¿Ø%&®É ¦Ô¤ŸŠS/ƒÙÍŸa•v^`Áu•{ïÚ °,äL «’×ÇÅ=ZÀú±Å‹Úàz°¹c!çÏslüÖ˜~ȃ]FÒ÷–(°ë>ýúh/8Ìf…*?Í„ógÍ9.Úè€TX:±¿Ñq`›QòÜKÂÆ¿^|ýjþ:®>ÄúŠî0÷çÆ«W±ïcøÄxÃ/†o,€ñÆ_¸ÜË6nqú0îôí³?wüCLnçLqù¸íç †ó¿øë•ÙÓtYÔÊtIÂþÙßüµÝßìª>C|Öõ$~¸ý¤ ×/Âð=H‘;¿x pgŽëï`¸-¾`Lré.¿6¡ºÑ—c„¦.÷<óüèÎüá{•¦ñe4Û¾¿âÁm}4Gð„ßÊ8Ík÷RÊKТ…Èø¶4ZU´øíÙÊ›Kahy?Al¥„#Ze?õVu†­Ë<&¿uÙmfÔ¸ŸµÀ+ÚŸDÈ` áßçà;~äúx×È>>iž¤tsJÇ9€¬$îA PøVÕÙŠÜ*Í —œiæ~ht%ÔTîôy¾;=¡­ÙÀåÄqsˆï…Mô_–çFMÈ` àöï¥ÈaÞ3Á]T ªÀ×îö€l± LŠÃf’fApê‚2=Q)WÈ)õÆ.ƒhëîˆèÔ[ aæÕÞQ Ò2¿®öɇ‚lˆ:ùªÈq®={ ã£Ò- ”]Ut¨ç¨D§×8YQ€Ú CÊnŽ ÐxYëüháhUR…hkZ,¨NtðýTÛË ;“¹7O_ôW‘¼ØÐ\ 0|Úùà~W$û/I<õSÇœ`BQv0?ñ„¡î7Î,³®â 8˜ƒ5‰ø—³»bÁÆ">ý¥¿#Øæ¥È…çÈ‚½„YhùþàP“hŸ_çNç…£ƒ‰ÀÅf–‚0ì9¸Y¤åÞɯ_良(ð&)˜î{ α“}2ª¨câ×EÆä]8}ˆõCÿÛüVŸþY¯bç±çWlãóï1¾Åreްy"LÇá|íº÷}Lÿa¹Ò?æq¾=¦·ýDÜüâ~"nq›1¿¡C=ìUÿޝ¿=òç¾8}¸ÿÆÍu¬&¦Ï+j£!`âáýa††ó†8FFBvøTëס[#Ùh¢1OÇ™ð4šúihuQb/ú©èá?ìÍ”½«ÿÉ7Ç1ÿˆáÂ÷—ahþˆÎT¾[ZúœÔ÷üK®o³­æÖ¯6š=¸S'ß/iê£XX¯‘ ¯îÌ!vÝ¡& žéM.ÕVŠàÕ®å= ÜMÍA3ù/€D„âBæƒMÔÑ6HW׆«1¾Âøî_ùÒùöÿÊ—þ¹O VÏâøp[ïár£ø…_b~#vþ_s‹Îbúó 1¾ÅÍnçK1¾ÅúǸ|Áv>Ë•âú­Û¾!NbùÆ·˜ÅíW³Í{˜.ÄÕƒØ}»½OÖ×À|B,Ý3Šs$h`%´ýP_Î_Àrh˜ïˆÍÛöÏP‹›²íøÛ÷?®O²ÝçüÆu`ê!!ø3ãìôK¶} 7“ /×£iqÏVÖh&Mg_²îüNßDžòæI¢w8ÿa™ðIkjήÿÁÜF6ÑNm&ü,PF[©{–Óúqýü™+'VŸ“ÑÙ«¢¾óÂ@dyUi°mHT9£J¯¹ÖIŽÏ•1@E;'þÎϨÊÜ>·Žôu“ Y½'®¾¤Oz^0ìG ‡ÚÏ=Qì&¡³]þ³W€¹ë%{úõq`Õ¢\:.)li̬rBÞÀ©]ò…þ†+påÍÜeiîȨrÂð£À§]R}®ºøŸ\™w4‚;,‡.€àÛÄN5Û-¾Ãšæ¥pDÅS˜~ ‚8gúÒÅË )ÿz‰2Édè:ÓnYNb¡åŒûs½¶@äM›Lø@ÑBó #i(\(üqý9¨P4pêÏDƒÌW¿ùð™LŸä‹YÐ íe׫ÍtH¦æÂ• ­øe¾õ}!è Öѯ‚@×VÁª™ÄvŸdÚ¹ÄzÉ÷šÒ¼A¿I ÈûÀ"š®È!Œ¬‚Æ[égþ?öþ<¬Æ÷{Æ·&³LEÆÌC”ˆöYÊT¢Ò¤„(Sh*ŠJ¡"šP4K¦&ižG•æQ ™çyzö}¿®ëîwäçx?Ï?ÏóÏ÷s¯Ï>ò®=ÜûZkç¹Îµn¬¸PvÓþÖ@h|iøéVî;'“÷«„0çê‘)X¾ÞrâºGоbÎë5ª:§~m9þü-tÝßY"­ˆ5G¤mGJö„žßdw»’fNÏ1h½áy©ŽA-ÞŸ  Û°îÅõÚ(HõÐy¼Ršÿ"ôä÷9Å\= ú%?gÝù_÷:Iy ý™þïÝù#O·ôgÎ_Jꃤõ‹Ö-ªkÒþ!õÁÐúEÿúLiÿâVªwRüKy"çgë¶_­1ZhKÚÛ$nŽž‹º¿ŠÎ“=Tסþµ'ŸŽùúÔpûÖž—Ôæ”Ä6ñ_z¦jp<ñ]3Ïa“8øï³ÖCª•ÿþº¥Bò³-ûKŸˆid'p>ÓŸ+‚Ô,“Dø¿kN?e ¿ó×ñ[Ò–%è!äÐOéôÐ>•»Õg~Wß‚øÚÄRDÜÕ7G¡÷¦æ‹MSMЧïìo7¼Dÿ-ž§×nåö(RJþ5ê‡4[Gùæ`ˆH€ÉøOšú}áŸâժܾ©á±··¤Kb„ÚÐúÉ}Ã1ZÑ Deì"Œ•98QÁHÒ7ÿL¶”_ñ•iFGneaBÜÙ…ùÛÄ0ñ§ô±hL±ßenµÆSšÞ‘*Û™Nñ°©R70ó™XÉô§¾˜µ²,9(¿ ò9.¾³†cÎÊìm Áa˜ûEÉáwãOÌËZxüy1æGJ?WøJ±¯L?Ï·Ä¢º£G‰}Æ^qW ðJ©÷Ö V¢X,!’í5ûCÙ£eõñEãÆµå÷°t¯ºTÕò X>ìAãr`ÅË/¿„C³ñ zøîh=ÛYy¯Úl¶¸»9wõóTF‹úcÍ£ß;BÊCý|•ŠÁc`ðqSùÜyW`tÑI{šE3ÌտŪ·Ìƒô°^×Ûfœ…ظǑ÷·õä7¬»ËuÓ×qñ×Mý—¿íñÃî}üîqG÷QÐzLq׌¶žt¹~פ<‘ô9ÿ)á‹Ô—Zö°ç­ qÜ¿sÏGžŸ>Ç…úú'gòk§î_ÿtX×|°óþ©òº|iƒ:4$ømm½G„ìÕäúÔ§ó Æ®çLa!þCuƒÒy«Mø%$WŒÀí·xZóTeë…D®Èí/¥}}Í‹í ›ùïELîæö˜~tè}òÉç þ—ëu? g—ó¿˜ç­á…kó¿û}M[|2ÿ«Ç2I1àõ¨¨.¼]û3–-‚w³ð‰yo!b#‘²ïCÄ4 Ã$kN£×€(™£N•ècÐç›QëGôsw×wìˆ?щпÞå3}Ÿ’|×5C[†8OŠ—†äíQªáÚ=¹¾ÄHS“‡:9þ]–«3¬²'¤ÍU?Y?ã…®ï ±=‚ '‚³ÇN˜€I{ŸÉ|ÓÌÔç ~}k]‰iß$ ¢Ô†c†ËÒÕá^1kMCìüšXÌJ>öRÄÅs,ª7̃Âúp¯·:–PtîµP|àR,(~u/}¬()Õ» tÿŽî¬ÃC¡R/ëµãÛi,Ö}9ÛùüM¨=sššxKÎÊ÷ ×åaÙ¦b>x`…º{Àçzehªè\o+ê -ý-o¬vÿ8Ò?Úݶï8Ýy»Vœ:]…5~z©£$D¡ßÇü‹F' λIþì£UË BÞÀXÚÞÚûW+ÖIL‹Ü}Aëå†åŒ¡…ó“¿ê÷fVþ? T`¡&ü±Bý-Æ´š£wbyC–¸Õ­ã×kfWþ,ÉÕAÚgøWÿ°;ü_ó‡Ýû‡Ýù!ç/í¾o†úLé¼~·ýô±ûïwÿ;Ê ÿמš¿æýÿ5Hê)gÎ/ÞÍÀõÿ©¯´Üv­ØYnîðΉËÞcÚ»æ‡ÉþÒûz­.Ë äü¥/ïè—Oûе§ÃÁ¨¡eÅnþËeù§$~róogäøÎæ¿7w8ðñÿcA¯·W ø{©Ñʼnc ¸=ož_hìóÔ¼Úñ9 6²-—¾iCD,žJGV»C´I2è‡ãô2?ó~‰ÕTôy¤¹hѽýèWÛzPéŽ Ä¿ lŸqƒ|ܽ„{Ørû7$Ž,7¸`-†mK:µFÝ#ÄF|ÓëøQqî‘ã"4 =Ä=8ûr%ÆKôÖèX)‡ ïšvz|Ãäß­ã‡Ú¦aÊ;©,ÿõ˜1ìq¯³üÓ£¿*½ò;· Ÿñ¹‹9F®ÓôzCÁZiâœr7(Þ6}Vu”f´ŽH+nÀ¢ÆÎŒ ³Û€Õ#g›¡º­_]‰Ç¨ Þýüš¡ ÔÛú'ý„)–å—-Óå5¶ª£NKB“§Õ²¨s*´t³›Œ½*±úXöµ¡ÐÙV|djÚTè~°²ÿ"« ½ ¥;Ròa`ý2õ™í|í4È“ü1ÆþAÆ ®ú×Y''ÛÀ§Rµáq `6®t˜YÄr8=5ñråÂÌŒáGøAR×r«­åcþ«]Fyn%ë¸>í[ÐzEç/þ寡ø²{?‘þ;Ç É#´ONq"?âöjS_ÙOÃÝÙBÜùåöj÷Ï–‘·åxà݈ٯ•oòï–Tw®x¢Èýï”r{Õȹ§¾S:'ñà½L¥Ë¶Uü‡Oüªvª©r:çÃ5’º'öŽà|jtßÚ³=K”"[Or}õW:¡¢û/ãúˆïžÜòœyð8ÿ£´J¯_·ð?Käëe‡ò¿¬:úÄ¡2ƒÿcꊾãù?S¹Níš—8©Xš›mÞdz[â4F@(8ã™gæD¯{5ÐaÄ\Dóo¸HC´túH{ëûè5i`¿à;»æëMÏXÅÆ€¤ÁCê­0pél{ŸG¥ÜÿŒzhƒ¿~ßĉÃPWÉ£}TAbƒeësõ ´žì7¿'†]™"ì¸/R%üt¹£ùóŽÕš \Œ~n'•6ÒéMÙƒ:%8eÂ,ŸgOžÂÄ5yÆF+0©Cå„”˜*¦ö(ٹȳ ÓfUJœ42s®›¾¯ ²á1Ëâã°ü˜¶ÞLÂɳoŸ¸`³VòMÍÎI‡‚MIüâwþPy»zzæ_ÔÐ]ó½J6ä=oÄÛLóMÞ6=À¿3ËæâJ¨L×+”†jï#S]°x¤Âd¿%îP›§*|ûPÔMìïÁ’ãšÇŒo^ÁÒüà ,úåÂîWÉŽÄ ­>å¢Êiи`w%Íï'VŽÊ·¿ýÂZaOvËH/Ãj%Q›sßA{O¦ÿ¡@Ǩ§ÊÖ¢Ð]vÈÞSy Ö¨ /ë)’=­/Q7vn€þ“ÂèÏ‹`àó¦Æjc)œ–ô[¿\k'f©Z¼ýºËÃ_Ô»TºÏà¿4Ùµné´*nί»¯†>R|ú/žHã²{œv÷}Ó8¥qKq0­o´H)?äææ ¤|–Ö1n¯)í#’:Fýr^%¸—âXºƒë#<ÊíMì¦wÞ™·dÃÆF·._y|Ðù¡iãMþƒŽ]3šßìãæ#žÊùÜÝ÷F˜‹WnΉøÃ¹þ„YziêTÿ®þáð;=¯‰ êÚ—xZÝcôûíÜþ5ºOÿ[Ýñ©c¹½‰’³…çy$ ‡ÄÀ†­.è1Ȩõzg„‹/Z&+ÉÍ‹é.ôœoó=Ÿ‹©@ï\Óå;öŸE?ùk3:Nçqú7OœÏ?g±ú"g—M¬W¡O÷½XI¹<÷Sÿ`Xº‘Â’º³]sÃøf§Ouí§éœÛaä~ãÂæ~ú33 ^ÞŠÞ{Ä“z¤*+Š`òêk!™}Í1Ýêräzs1L÷Ø!œ:²âßó¢37qz#nªþZŸÎòMï0gKãŽ7¿úCapxø¯wu˜×ÖcÝ©@Ì/¼ëü=Ð J•Éù!6ý°èç”+ÍK-Uõß¡r_jêþÞ9X,¬×ÔË j§ä=·¦^Á’Þ‰tU±40kwýX¾pDZï’¡1h€ÏDM[hvÞ3ÓJöÀªÁï|ÏY=€öä’õUirÐY;O.ýùGèFh…çjšCOø—_¿~ŠÐwÉô¾Øï" ‡í7F·³N|š&óƒW]3û`lìe‘» ÍŸ›Ñ{ ¿>奃½CW§h\t¯Ýãî_¾ÒÓêR7Jpø”þgo4¾(¤uñ_û€9^GïoA»÷i|Q^Hõ™î{jþºÿ…ÎVÏ´ìê¿çéÜÿ ¾4 ß…çãeºöнdOí;<>|³©»WW_ßgÁ Óúoºú‡r²zC]›ÿö}Ïü]>¾´kiêµTƒÞøŸ&ðeÖó?_ î›’óÿuíCãÕšüï‹·Þ•èÙûï9ÄÀ˜Á’ÛR ¤/t«W¤D&{Ytχè·+ª›-»æËOí–™…~Ãl Í& tïÝ5bünãwÜ?õ™Ò¹ nïE¯‹cÆN¸ã?§YMÄy!ýeL°Lì9Sd›=tÕÁ´ ã!ÏVWrsrÄÅYy‘üšàéÒ˜©¼N&£ s÷½Ô8ÿAóö>¾ãûúæGÉ&›^•ÄB!‡ý!j¡P¨^xÔã(Tf9 Óq<Õ¢ûÖ 5Cß·[z§p{¼—‰÷þ+÷­õnkæ’—·æ½‡VøñÞ¹»C{ȱ… ßCÇÒ©³®sz×ý-NN9é‘mÿ?Ív0š?çb±LŒWÄôOò—º£scNl÷Áú¯e™ZÉ^ØXi0WÞ=›”s¶ôIvÀæŠÍ_*g¿€EÚA‡++1æªÔ|#aˆÎL•}ÿf*¿.ûcưƒ÷9üIãå_ýÃñA¾Çvê)Úºr?ÓߣñH}T¯¡¯û—¿ô|‘ëv{ä|¥ÿò—¼ÍñC:H÷†w÷—Ò}pÝü¥t>ŒÖWnîòCºgŠúrºë;â‘7äמíÚ/Eûÿ”Ò¾ ÝÿFç7Èžpn~˜ÞŸf£¹{Ù™ûÿÛOGú‡ß/Ï‹x«Çÿùóº«ù·…]sDÏ¡ýá¸ÕõWvéC4¢õÔ«CK¸xí3øúP|ôó|°·hÉzˆ¯ë»wòåóÜþ!Šž÷­6Tr{7†MÑߢSRNCMÖú/àüàc¼jLÿ„qgtÜ?ö  ø¡b°GŠV´ª^æôµ¶ƒö¦¸¢€V)èü°²Ž{²kÒ>Ôìh±~TGžó3æìl±âŒÜ¿ß,u_uÂ[fcÃz ßÁ 0S¸tácÅ/lzð~®²Tœ|-ònj¹B¶Xÿ±½Ì5H J/Óvÿ<ó§æÓÑ=8]‚ž×î{1ºïÓïî'¥ñÚ§ÒºJûõ4èy/vO‹žóàW¯è-Õ)»ï3åúßÝæ ¹~aø°‡C×úrýBªgväü§„Or}ÄyõR“.òÈòéÿD©ëþ4F·o¿yŽÓ=Ÿ¹õ’ ìcÔå3%û.ÞZ9/Ó ØÂo:4i³õUgÒ¸ ÷¿ø>1'ÿϼQüŸÞëãB–ñÿð‹·®Tê ž×¡)¥êj]ûñߛ铑™>é@T»¼â¥·Q×~î¹Bõûš®vÅ Á—N-ÿù¡yú„ûulÄ šÕO•Ÿv`ˆPâz—¡#0t\ÂZ=isH(”šm¼ù’f½\¤7¯Á°¢,íè†%r?;oÀ‚PŒ ›¹þÀªTŒ~;±QlÂEHO÷WŸqÜ ãö¦yªz?Åø– ×?~ƒ‰ÚÉ÷bÒÝõ9߯ÁgÕíË.bšâé[âU 3@ì÷áð¯˜Ù+S2‡Ø}¯ç3%{5¸úEúݧ‰ìIæðn÷}5—rý º¯Šærÿ5šþò±Ñ=úîas§'›ðó»›÷F›ÿøÅ³¶=×ÿÚ/õòÛŸ†¦ý¾]}ÿEÅ=V®ðC×þòÒ™ üê­¦û¬ùŸ¦Gµ>ŽOãÞWt¹á±?·OÿÛ,éà3Óq{‰6úíìu}%ÿϸA#^ÜÙеoª¶p]€‘;„Í«‚MtAä§Ø¤„´rˆyx»õÎbôêóôDfŽ(w?§¾+¶¸áµ ÿ|©ÈAA£!~øÜ §& ¤ÙCÓ¿ƒDS ¯JcÈû—Ó Âk·Ì–ðƒ&ά„ÔñÕ›ÌÃò02eرk=ó1ú[jkÖÕ5žqh¯vÆÙÕ׌5iÁøÖÕÛNfwíi8å[RëLùò=ôñ/#LŸi5BÇ+3 _© ‘…Ü”F)Ǭ¶ eüS0[À˜—/™‰9™Oz¯ÊR‡‚ÑN©?Þ™P¼Ò»cß@Ì#ü¢Õ%”^íÝ.9 ʃ4ש랄´Îrüî­¡<‘ú¾)¾¤ûjè}/È}9Ÿ¹ÏÅ_÷£!u˜ó§ÑýQt~‰úÓ²£eôÝÍzÏòÐþó`O—¬á%ÜýgÞô«w¨‹Uêê³””½ÿ¡ÄAe˜rs×}¥«’™àÅÿcV$ÀáßôåB_/”æÿ°Ýã—”ó‹ÿ[쉔“‹(x½-ç‰EL›JSAØ5¤×J¥õåOú2ýùrˆ}ÈÓ»Õ7½jzì*ÑûŒ>÷ÒOŠþ{F>»¿'G|7Ç´O¡ôðÝÐG²À’“‹¡¯;”›‡Þ†äKûÓjEZRºjbFF¾q?aÀÿ鈳¸9 ÒYé_rüL1^ûÓ÷`“­˜8UdšòÚL^ÖzúLÚlL«ØÑÃq)dTN¾¾Û©™ï…íÛM,0kÝ£Iζtíq;?Æm«ú(ù.¦ô8ŠgËlsõ°àîÜ>Žî·°Hçã#ÑAßÁÿ&ë¼÷TÞ®ÝþÛ׋ Û'g~…úÌQÆž>Xò»Ï»¸6w,{»£}á‹/Ðã}ÉX‡•ªÞžÛß„aUd€¬þÌZhªM»ø²ºbçu7¶}ÂÜÖ_ ?÷þ‘«¥0äY޹¥uFÒÿ } &â¦Ï7ÑéÚ¤Ž#±¡áäŽÂïv0S+^çW„MiM2ñÛÁ\—Õh×'£~àÅ€pˆzÇÆ¦\?ƯOœ)wåÇϺïÇÿ/üÊiüÒçÿkþÎþë~ÿzü¿yŸ‹Íþß¾¯õçÐùC:×ß­žr{K»ùK¹ùaº—Æq‹á°£¼¿ú(ÜžŽÂwbaËDø‚–ùÞ®éÚNü9Ï'Yʨ‰·ÿí§£{éž2¯Áí•*¿¥©’y‡ÿS"æqÆÊûüßF§Ìœn´ó;Û«ï¹öŽà|áÂcúù)ˆM…¨˜ÂãoŸúsóûÔÓ·ÕY¼óv4Dõ›5>>š8aË«³v¾cÒVenï†ä¯’¶ˆ\ ·ì›¼É¼ŠÛëFç§Æ ‹«l+ÑU¥ŸÑ–îƒ) B—ëëØ`Ú‡iÛZÍ’0£øùõI…{7èM«-ä¯.<?ÞsNé¿Þñê).†hJŠÍ†âÛ²¬÷+{@É–×#W2ʳ“Ô¶@e8¯Éÿ[TÓ÷|7î³j®âŠS¶žÁëó²‡P‚eGå²N^²ÂŠò‰·–êôÂJ¥w²Ò$±ZØz¶ª_èðöšn»SÝ(E‘ùMгֽ?1#Û¾Í $£Ó #¦ª¿ñ»üUo”Ëaz,{È<ãØ(ï2øÝ¡˜å-ú4p¤7­ÞŽÜ¡¿2µ_w$T†AjïÆZÛ² þ‹ÄÊ q_ãî·Fúë4^(Žì¾Gÿ_û†i½¤u•ö©_âJªP݃»'ÙSCïÓK}eܼÙSAû‰T÷ ¾³ÿÅ9~H÷¡Ò=§‡?5·÷2ïò[¯úõÞÊW×þÚ{á¼okcÔTþ³‘7Vý¨Täæ'^ =t±ÙÖå¿pâìý±_»öû’ûúR|ùã°†¶fèZǸû"öPR³,h*„_“/úì áôvÇb¸øÛw"ô}bM¾$ý‡¾‘šmßž¡†Yûiö(佡lc Í™.*Ö[‘Ûû;Ä£j÷Gž†Æ–ìûB ÷e/;ÏéÚSS!’gwé>F\Ð?ãz£|¤ò&ûþÁ˜Ó}B¦4iBÚ&ÕëѾŒ+´›V­‡ kìÆßÿÜI=>Lúø5“ FºžüUŒ©Gmöšç$b:Ñð+â²y9ûm|ó̼6^jÀ½È\ÿH|²€'æJ‡6ïN„|Ɉ/Ñ3waö³³û¼ãÆp{¡2ÍŠ*®h@Q#'ãf‡æÔœžÿ©”nÊ^ê\û‹µÇèÜsßß×jËÜCP1ú°rÖÛþPÝw%éÏ¥oXÓ~ÿuZÔžY°?K” H _ÁÒ¨ÞÎå¹X>ï7mÅŠ«Þ}_‡AÓLD[~ñhÉhØL›ö«‡\Ù-w Ú>ŸvM-‚NäTgeó3Ð-ÛÔY|Æz¢§§jõƒ¾ž²’sGמ …Æ­b1p<¿¢&ýµäŽÖ—´ ðï³rn|9ÏîÜqòÀšŽQÜØ}?ð¿ô™îýÄÿ5HõXßÔ÷Ýý¾O”ÒúEïãû¿æ#¸þ!ÙSCçÿºŸ[·~·gŸêž”'Ò¾Ñ=©ïœîàýCn7Õ9»ïѧ}ÿïãdÞ¨ã¿øýrP¡Ò™¿÷$’¾?õ‡ôl”p˜ÔÉÿÔ¦9ÉÊa ÿˬgç7äÔ]ÅÏëÁÿ¾¢æmë»þüŸ£‡Ì<ê2ÿ{†èX±åüN+c›‰W†uÍQl½ði„â?ܯfs ¢›m$xz@ì©O–¢,ôr›>*Z}æ®7c®>úy·§/ŒZi³úo˜æKvßÕ0Ä dñ4 cnßþД^šIž3 yZ·­_ø®}Št?›d•¼ö¼"î>ˆãFNüµfïoŒpRB§ùæIë·š-Ä”ªÄ/*0­:hMv¤3fôùñÕKPe×½ž?â~%fY 9ïcùÒ1Q#®OÅœ…&3ŽØ6cnù ¹½Ÿtío£÷»X0´íãÜV,Z5k / ücÒ'ys¢ ²wÆoŸzÜÙùD·³óiNgçó/Îuv¾êÙÙùzOgç›ïVwv¾Ïììü8­³óS`gç‘Îί6ßÚ;;hvvþLëìü=¹³óý¿»o©UðZËy<Ñd¯ç¯—¯à-þæñúîàñú5óx–ðxâ‰<Þ ±<Þ`oÈOb+'YÏã _ÌãIÅñx#Gñx£Žñx£¿òxcÍy<éo<ŸÇ›p•Ç›$ÅãMöàñ¦|äñ¦™ñxÓ+y¼ y¼™—x<9Io–'ÿŽÇ›cÊãÍ-çñæÍçñ£y¼ƒy<¥C<ÞÂ×<ž²1Ç/ùÿx=È5Ùc±ËZðyðx}Îऋ÷%©“£qdå™ b_ü±·XbüÅŸq)RNýÔ´c°ûà{åÆ*X$Û½HÚŽ ãfî“?Y„µƒ–ÿ­³‚AÈ„”€™ §ý¶-D+º×_{ql/tL¿è¯Ó} 훳b%à½I«ö}ì/h{ƒ/l] íûã‡NþcËý¾^eá蚈±0Üu§ò¸2Ö>ä ý–| ëdç¼} Œõ[Ë^ìqÆF-ýÑÛŠý`Fž“úÅñüá;°¹‡¹û€a#°Ùe~‘š¥/6gÜ.Q†Íφ2ï æƒE3{Dî†ùÉ–ßó¾>‚…bh¬óX|ZßwÕ 6liíß,×´[oŸ;´ÛÉçÛa»ïÍ×[±Sd±û½JyìÊŒÝ>Âþ6¬’~ìz[ëÏ#nÅ=¸ÛIòƒÞ¬„y}{aù‡1w¢±7aÔ®1½±ßoOæçp ÍÞ[zc¼ø “çw.‡U(ñpX\çê¹?càêæ»¸9? îÃÔžon[Ž#e;ÃiögŒGþäCçq,Vo\®úTxÝÑ‘ªì¹'”^^]¦ß­Þ¼58uËúkÚ}3ø-~±ön>üë…4çÜùŽ@GqßÀŒ³8£´&õQêœÆ^h„H²_ .ç SŸªõë¡Â]cýúÞ¿‡ˆv^Mʰµˆ2uµ—üDÿxYØ«ù5bŠ®.¹Ì±Ù+›VL9…+¯.§FWšàºÞ ¥'ˆw> ö ø>%¼™¨°IÞ?Y¨T yy¯:" H™¸g͉ZH¿fÐØ~ ÈP½0«Çj/d¹kõP:‰œW‚Þ?Gÿð’¿Ë/øéÙ“^(0Lêë·rù›ìTܘƒ’:AËPZg|ÄÓ¯ežµ%©¾¨XZb~­Ó•üß,‹ª»êCô—£ºùLâóŠTÔÎ.3ŒñDmüž›[6¨ Ùž\R‰zO½ûL£Áê× ÁC£~¼B–Žš5‹cÞ^@³àÔÚ÷DóËQj3¢Ñ’!÷Ê ¡wÜÒçªÕ[¢UÍòdðI´~/ž>rÝA´E;îß9~î.¹Y°·ïÜmqÙtóúA´oœá{8ÿ!ª<7c+Ž®^¹{Wðc¸ôHÝòHélW?(t| "·è슿뛄—û^ØÄc9o¬7íÝ ½oÉGn{]òï«í "d;4ÉùX±Pi[Â9U, ªXÐÙ(Ž¥ïÄ|RÓK¸Gúïô÷V’x£Ï£ãù4w·TÖÈè¿{÷ô¢jsZµ`°\åMè!æeE ¹µäï塺&äõ×5|™¯ Óšˆ1‘kçaýmðBã°áãÇmÚ±1\ñ£·ž"ÌØpsæ>ߦ&»á¿´—csìôYú³Í`N^×"ÌIaºû|l!yjë˸6'b»öœ1‚#Œ7R˜F“L„M°{’û§â'w`=ü©áéýa3Wä…†·l†2Ã;ØÝ ú"³ãìwö·ÒÓLÄ>Áé*;|û™_ïwŽõ3]x•Û¯ó&ÏH·“Å¡†è´MÞYp•»¶3Y¡Ü®o3M±\ÄåQOò<ÇHñRîYªiZŸósnKÇÃw óÆžâ”Ϥ×Õº~ðà üÃÍ}Ý‚À¤ ·ªßq¦ˆ=P¸oéQGœg®’òh„©,¸òq¾ ÂïŽ÷BÁ‘—2®åžA4–Ã~m⸼^^ŸÂµ¡3ëŒq)úøã—/Hˆ(Þ.zp8’6Z:F˜ú ™ù­mvHQ‰ðt~Ô4‹º³V—ä‘hЙ'qÙS'öÝ/¼ ¹‚ Þÿü+ò*Rê) ä¾|ƒÂVÅ[çžOB±³®ÄÞce(•¯Œ›4J ·>õ™ºïÍ*”˜ QYÛšŠÖîA•{YƒSÉWT§ÕÓ¡†šäómc¶¡öÆæN{$PG~®Oýž oQ†,ÙâÌghÌ×¶íe2̳Ø5d ¹¬sלދÑr{ÃÙ/Qܹ}ÁíѤ8´–µxvzÚ ‹‚VYäàn:s!Þ ýJ¿£QÎÕ¸ðçΖÁlÙÔG÷¯®ò»¿ü®ÜZ&¶ª §ÈyôôY}o¡¹1’™pØV·Yþ£\%‹¹xÜÒÛ ·ñ󘑟i2&çÐH‚=°0°½Òm‹ôI½¡uÓh»›ŽQêT˜¬; ÈL/°¡ôã„ýWúcÓ+ËÊ#s°…üÝŽy‚Œ3»Iž¶!ïÏ®4Tð¹`/‰Çý—vÙ|Ü´¹«Î Ⱦ…'88–-4päp*ÎjÃ…Äá¡ÀŒ3%®8LâãpkYà©S¸žèwþüܵpÓî=£Êò!Ü™ê£õî¿?Þ\^Ž#‚¬ð²EŸÕ3F| ÇѾ¥­S«qlVvÿ×S}pÜ©ï©É‹áõèöžÓCás0g¹Kùiœd¾uƒ ø^_cѣçjV-¼­p ~½UKÇí4ƒ¿ »ýñpC“~eK4̨CÊbÎ<{– ¯3¼»qŸµÃ„<[³pà¾e¸°aÖÓN#%°a2*a$¿„H "1 $¿DÄKG?DäZ…_ W›!ŠýØZˆJn+ÆÐÁˆ¶ ž*>8íù©€˜¡iåÛ–ç æWÚ„ˆý¸ô9ÿ亩_+x“‹6 ãŠàï|µWuŸÎ ÿ×âïºþ¶ï@‰§xµŸÞ5#ÁçvÒ³qH¼)–ryR ’HH6d^`n– Æ_8‹ÔUQ—òôÊv/m~°.û…—¶™š#‹œÃì(ÞÏK®–ÈÆ$@sä¶1…kò¢¶7Í•mC¾Ýá¦ÖOGQ°Ê";,}, ˆÌôž¾Eó¾>¬©ˆb*øvÏ%“ûŠOE)É3eòÚw>)B9Éë·„þ6CÕº©Ì'F5 ¶ ¦i½ôÒ Ô ÎwÜl· õ;œ¿·Ž¯Ay¿M‚o[aÑ"HÖ½ŽôA« ~µwy™ÙfU¸Ç“ª\ÜgÓˆ9|XTÚ>½¿Ù<ö8Nì~i=YèwŽÜ¼;AHãÁ»I¼™üi6¶ßܼKW±N€V&?w€!©zLÙöh‚‰SŠCuú aÖüh@ê"­WGnŸ¬š4õf3v›[I˜’€ ÁŸv}ï·×­Äž#¦¶r°'xmïãã³?­‚}f'¯ÿû¨CðQá@ê¤ÃõÓ‚ / G׎Ï'~˜R= N¤n¸oÿå÷¸N.^]H¼2>i3‡'jlšðÀ®ä<»’|áÖÑ8Á'2îLº?rGž?ˆ?6L·0ÀQ9ö‹Ã1÷°GoÊë×9!ûIKáÍœzÑLœHÂ|â.ððÞ.·Û5‡_“ªóŠ«G¡‚ª+ÈÔn3 xŒò>-’ç5\NøÏRòºêÏ©|GÕI/!ur9Áaôù¸x%¿¯Oð !Á±”—þ…C]™ ÃåîgÂëLȹ0‰ßp=SåÖüØ¿Rþ»‘…c0 au<61ipç=˜3¨C~0‡ ¶‘ú¿ƒàõ]‡YØñKp”8aw,Yºç]uØ{¹º„‡Ua_Šª†–Ì 82ènC!÷íâÏ&6Jœðþp}ûóéN/¸3å)û&<®´#£)GoKÇE¼3†×l¯]¯†Ã'ïiàë1`Ù¥^;N¥Ý´òS¨…ßïÚ>?äàÿÄäEßðÏ<Ë&œ1»Øɦ8·Ê1Å@M,=º}|;ƒ&îÑC ¹[6¡p"§úG^¢,»>Ô Ǭ޻¡ù:bÚý.\k8ÕŨ-=q8®³iuâ”[\6#!…-ðHbËÒ*ܸuäÕvuÜtÑÊúöÉ©LÁœŒô &‰ÌI=¥$A¶ó›‘KÞ_^⸆Eí(PW^šhù …Lv¶^ˆb¦ õÎEé,•/;ôq‹Äo¹—È•–N¸Í”½ÂbT ²ûГwP-`_£ú£ÆA·vîËs¨ Æªç©¨#õ»>lðÚ M%h :@cùAÕ=ü@41_ÿ ÷hé5JÀ¤>àΔKÊ3gx!›ÀÐÆÂLܵ}Ûçñ—3>ÿ]xF,÷/\V€}LÖ’©Çåª}7÷9;IÓ?Ü`Aê-­´>Q”ÖIª³ÐxÕ$ùZ“Ô#­im¶(mö¸M‡.ÑE(Ÿ4&q´<ÕOhÚJâb{ ̰“Ä1呜ŽiX³æüe°Î½&Ü«Çé$6„—Ù]ÒNƒ%ÜØCâÏ~øô³³²`/ƒÊf–c‰N?!úé6¬ÆÁŸ~‚J× .ÖØã„=½¸À }¸òÖ«*èÁM…!>úpgdÄ#­ðØ´¯p]FŽ2hoú,gÓ…5¼½ý¤[üMpâUN‚‰!àËÐï§8=‚!Þgàç­7/õ{3Àxn—žb×yuô´6œcdP§»Y¾£¶-a).È"ïI<Ø´^ü aL™[sáÏžôê©àŒ¦ ÛHV{ú#êòÀûã‡NA4Ñ.’:CôëKþq«û»#v·£¤ã2“œŽáJÈMw¯—&¸JðöõÞ,DœUíØN"žù­þÆH õ(‘ýšw#iü´±I?vãC›rŠ‘Ì -¹|ÜœfX}ê RX¸8©äïÓ˜¬+5F  !2ïM¶Œò»…l¦ûæ!WßC]ÛÇÂu0òóèK(RfJ”˜Vl~¯*[aUÃy1Å(ÿ"³ãiÃVTnDL=ª]w\z¹T5mìD݂ܗ¯¨£^@ ¦­ˆFÑo›´ü{3SF3ÑWîô˜töõë®8ÜÄ(,¸ë™U%ÜóÚY~]uî}ê_qHùÌaæmÇ$p<,êÕsæ/aEêÅg&$®hŸêšWÒú´„Ä­჋I<-&uGè­+H‚ö5ôî¡õ‡Ö—f$ž6a6“zlÁÀ–ߨBêåV»·Õ2‰L;$¦ÏX}h'É'»xuí*¬ˆ~cMâÈæ'.¶ÿr„‰£=¤.ï³\æYæ0ì#ý RßœXY½'Ú1¤.Ç{ µój»âRi#À-颗T”$Ž“ԙ£äºgº N}áÍï™8!8e¹×DàËÀiŸ“8åX±ê›WPdÃл¯zˆn‹À×[÷õ¼`Œ36vö·‚àXeMQF09¿!„w]`Эðh„²2ƒ3ÂίÐqõCøæ¹®Ù^yˆ˜nÝ_“ÁC DÞzSþS"QL›h»Wï.n.Y¼â;b˜ÿ¿Ü—ˆþË´O6þÁe¢ß]%ý‘k'n™tŠê¡#4;wáÖqW˜Šˆ„W3™Ÿí‰Ä¼Úó%“ ‘Äʆ¸! QºÛ×"™9F²‰¸ÉÈ»Búi MPÏ@Fî!‡è\y¤¿QÐÁ@“ºVúaJõÔðû(g/ß:TWÕçôDu=#ÈE Vp*®„ Žü]Ó‘ÉA㟟G”¡™ä¿–ç)9cN˜uÅ¡ÃãP™ØZÜ%¯Û~á¿Äz/ú¤à#þ;¥X ‚ƒ}¶nð¬ˆNqŽð/ÊÇ(Ž£|âCÊ÷h|©’º¨Bp’2‰Ï…¤ÒGŠ_)þ¤ürÁiºßqý ÂO©ÞjLâŠã{”’úhBt‘µÓ¿£ýúHÿþÕ})¯4‰O:uÚB™ÆÎBl$ýÂM$:º…ð¥m¤ofIò׿ uÙ–ä.®IßÒô—¼ÊõIžscP˜ä¹üõä›Ï¯áÉÀ‘8Fò˜7y¿'ØvÁ6ø®þ4Wê!§/ó3xAý‡(Ým¼¿R A =•ùƒÿä› öºjÓ/1ç OÆ„÷LÒ½¤Ž¢[D1²‹¼/¢ÃËrÝ[!†¥-pé»Ù÷w£p¹d`ù´õpµlæÍ‘ñ‚º(HÆKí\Ot‹D¢—$‘úžÌ–»uHé1Qz‘Fô§ ÂÛ²ç1L¹D‡É#x¦@4¦×ëÇš(ô¼Ô¨å"ŽâIÍMçûlEÉ£IןÈÇ-†Þ‡ ÛéD%#Ýø‚j¹µÇî ²CÓ°ÔFíê21£ŽI¯’!¨É~qh`?ÆH4ö[ÁtÐÔÖ™a¢h´ý쎌‹Êþ@´*›\wØm:ŒÐ¢‹»ÛB½7¡ýÐûS.*âÞ [Ü_›T}Àýæÿ?ü"çjµ9S9Ý“íVHL€9_4þ(_£õÆ“Á“ó oS xxá9³ ¿¤sHœKê¢"Á¹ôyèóªÞEy#WWÉóÐzÛ]âx%©ËT?ZFt>Ê/i=¦|R‹œCÊ#uHýÕ%8v ©¿úD¯åú˜Ôw@òÕ{¨®Ìé<ä}ѺMëµåèÓ)zÅa‰{k‚Û9‡9E’¶°Ÿ9aUZø>ì%zóþó,‘£fkXôä@/;“:è€;#9•ø ¨?Àƒ\£Ä¯á5¾úñFÕcð)™TÐþÔ¾D'?ÕÈyn]¸˜•SG"HP6²Ã¹‹Ž}ê3çIåÏÜ4ÿ‰N—ò\ü»-¢‰.yi™Õtã¥çqùNÚgÁ‘[î…ï >6ó·æ»³H¼:{ü©Ÿ*Hf®2&R y éKÏ3H™ùßoDÌE£ÞŠW#èŽ‡ÚæØŽÖCѵÀ‹aƒQBú#efUâ÷Ä ‚èâU[?µÞ;¹ Õ¢™Žj¥—–˜_ç¡N~a‡¬éƒÿÏâð¤LlÍ—p&}‚-äû 'zÑ·×ZÈÔ˜ átU$N¨.BÏ5ÕSh\ÐúHuúHã‡ê+ôœem™r`IwŽi_ÖÚ7ßÅT¯q†°f/Wï.¾GΫÑìˆÀ޼÷¸CE<ýèìYøuÊ*/´'ýZ§ö“þ ƒÕ%YÏÚR®ïq€ð9zݱoK‡Y{^\Í™/Ô“«cîÇ{ÏáItcbÊ㌠sÞ„ß uÔw$ qŠ=&¦8ÍÊ‘’ðcé¡)zYê3Ć?·ŸÑ€ DáQ/>èâìöá\Ô¯¨é¯‡"D³÷sŸÛI¸0œ •Lµ á" :‚PC„+L!’\¯¨%¾¸he#Ú„m„á"›“é²/Î\ ±ê[ç>˜ƒ+ÃõŽW5à£&*!. þÙO^âI}äpfk¸áêà ¦ š"YüŠš%¸IêÔÍGJ1ÁémHa¿öaH€ù‡—JúiÖl¡Gú•ãëߘˆÌH{ã¿´`„\+Ž«è@žµþ>ÞŠü_“ ®´« µU­BÑÒ>Wʹ¡˜èÆ¥Ü"ý­r Ëß=ôa*ÜÏwܰ7ý"YÜfè”d•  E%áõUÄ_Pª¬Ü³UÄgT-Ū ܤU‡jÒ/®&õÏ“ðÕý$Ž6œ¢½\òØ N/Õ'u…ò@®?Aê*Ñ])¥ºè<’·i}¤´þQ]•Ö;Ç4>)Nå|p¯®©¼åûœ®¾F7]s-Ñq¨îCy¦)ÑU¨¯†ã›gn&ÏkAôÒ­D§øKçd¾vÃ'°&~¢3Û‘ë·§pY …vñI‚#Èuq"øü`âÌ"Rp!ý¥Ã¤~º¿žÉ'î9½.õ…ÓUtžlzŽc üz›ÔÅ+I=>Aúû¾³\&Tø<Ä)MVPÁiöÀà g‰0(ñsåäÎ °| ã\À‚«Ï‘úÒûPââžýpžiß ¿E(áSaì×ÎCÄB@¾B$K—#Úaêý[-uˆ™0±tžÅ0\"ýüXâg¹|Y©ÆgÒ\aÒʺ\cl>q‘¸Nü,ñ²L£ñÈyJ`èVÍ$2*¬¶)’ró~œ=7X[Œn’<˜ªôÓÙ·3é¹áõEd^ûR7Ë9l[o#òHžÍgÓÏŠo‰Ööì"Ui“ÜŠ(^ò™©¤(aº{æ(ùÜÄdJ”z?”tÜ"KìTûnF+£‰ ŒµõEY›éæ£('¾§ò£Œ°5å䎯uëóm$u‘úÕhœ­'u…Æ!å\ÿ‘üü—ŽÆ+ù<œŽàƒÍD¿1'Ÿg«H'Ã8°àKò|\_‚òCR·©a/ƪp <ÜÉ—và, sC~Å!òºô{r'>YG‰^püÏhÁ•9 âõÕõc”`œjß~mqAüÜ÷ ذ6 -e ÎJ1 ø+°Fœw’fœa%×óåT1…®¢Öeä§iù#šøŠbŽ™)˧ÝD,É—ß Í[8ÉšëßÇé² ¬½e‰Îzc ð‘LüC)¬<+‚T“҉o,“A‘³TÓ—1¨ª"—¥{Ë‘×ÌPÀ¶¾]jBAqœ@‡0> TŇQ„p‹)k™u(×elÜî=¬ñ¦q0*™î†R-ª™}qjØöÅ8Ô²ò¦ê˜¥ß¢Þð`€áÐt4Ì`¢høõôн9o<-¸Ôh2oz÷ZÙÍLûwšIœ·°pðZ.³ÄwƲîx1‚Î$ÜxÛøqheàuÕaN ùöN€ê3´NuïÒ> ­o4Î(ÿ“'¿?‹è¡rä\Ë“|7—ÔÓùD/䓸¡ñIû“T_íþHù÷Hê#×o$ñßÝ×Ju!ŽÒ¾ ©ç?$ïóŸüði®¯Øù²§Æ¾Ñœ®ÃùWÉu5'øt y?Û‰ïÇ’¼þ.ÿÖ¤oCtjÚ?â꬧Œ€º`?Ñ™¸~ ÃJöÂ…-«ëp¨è%Ã|àJún$<ëAòå‰ÇÕQÒcø_ŽoÕ^F'qê¸G“ táGôŒâ "xá,[&"„œÿPR¨þ¹Ÿ1Xº šÄÅ%Ò·¹ÌØãTïâä—!žðÙD°D7Ø17Ù¶Å ¤AzPG ‹ð­ÜE¬0‰<¢Ï>faûô‚ñÓ¤Q|Ð`§ 5 tãŸegG¡Œè¼;÷jÜû6¡«1«N,"Á5¬Mµäóײÿ¬º @¢^;˜a ¨ÿõ­¢ >‰·‘ø[3Xãšßš õ ‰Q1ž¢y8C¼–¢Ùч‰`4×5¯-½5ûŸü0bïòvÃ&¢‡P}’âBzNéù¥ý zî»÷÷©~Ji=¥úí£‘ç¡uh39[ȹÛAü'´ÎØd²#Ø‘¾ùrÎ÷¼Y ;æë"ì9gÀ0'ì!ýî‘ı= ûí8>H}.ûKXÁŽäss8’ô?©_õñ¥fº•VÓáJôP7⃥¾T-—uÍÆð$üíkóÜÊ{o¦òü6N:ÈñÂiGOzŒ0Ãiâ¿§z‰?© ŒâŽ s–ÚñàloK¦ãsì×ÌGÑγíM‡.*Å•$ßE20R²‘÷EÞv‘9®9ˆ9Å >üÁ%F•Ü$ŽËS¦nú|æ®’~þuAñÜ~M ñû‡ ¢ØFö,ÚóSI‚û¥j1nsž<Ä_Þ±b5’™ËS–‡dâN!õ#Õ~µ¢}ê°m•ƒõHgÆEΔ"cÜc‘ †À=B6ñ!ä ¿‰è}9 y’,@>ûõ÷CyÝBâC.zŸý)ä³Jˆ¿­4HÜ邨@ÜzÆgœžÿŸñC}ÖCgpøˆÎý~øøáÿá‡ÿïñCz~(Ï¡üækê£xêŸÔFãö+(¤ý ŠC§’:Giÿ‚âPÚO ~œ5äœË¸ZäÓ‘'~ý;SOðT!ñûí?Û†]LÛêúON÷ õÆÌmÑù@ªÃÐxëÞ¤|O–œSrΧº7™ÄçDï4)o¤¼’òMÊ?iýWÿúè#×Oü×#‰o®ßø¯>c·ùHªSQ<@}y´³†ÌÿRÜð×<3í+R_åd~ã/ÞHô®þ’ëÁõiNµ#øŸÎŸP\í@ð€™¿âú+,­YWÆÖ¶«n_Y(ŽÜa€æÙ.€€´ô?»^ ¬‘'F² ¾Œú0òN×Ü–§doƒÜ&9Ê FÖH+ÇYæU´ „ øÐÉhöΈ u)ê hYóÒW!0ÓÇíµÇpõ¹¯ø”ŒYh1$@ƒ€, ¾4š}' hle IhÒ\^w}Ûz4%Ä) ­ðyr.(ßÚHð_Ðf mŒHÜÑy_:¯Kuî}}Êÿh½ìÞ×§õ’öëiü¬ u•žo:oaHê$­SÔﳉèt¾žÎÓo#¸t{eÏ VµÒ°$GçsiŸ|7áwt.šä:gEýÖ¶D_¡s{¶5¤{‡°€ûH¼í_÷Z×YŽ„ï ¼‰Ó=‰îr˜Ìߺ1,iõ'¸‡2vðÐÍZ4oÐ\n^ê8ÉGÞä}Ÿ ß§ƒ X«q¡-N—ÆxäOöV}óFñÛ1ÕtÆTœ!¼è©«ÁŒÚÙóÎÛwˆìÜ«‰PmvÀ a÷J¿ñ„pâoŒ :HdBÏ=Ûýú4Äå±Ä ñì8’3lB]¼/@"[fô9÷ Aö¸½ÉÌÕÑ(G ãŠ[viŒ½­à2Ù±Þ)ÈúõPBð\ÞSä’úœGúœùË#èhì$~… Ü"¦½Yè‰bï¬ÁÈ2'rk3ˆrÙÿ„Ö â‹¿MòZ%ó“a&ªUvµ"ª%™p ª‰¾SÍÊǽP#Â~ÚP³Œ°G ÛÞ] Í{ˆò¹jG²B:j—°†$ÔZFù•†ªÂ‰œÓmDw£8ö‰©”ú¶)>íÞŸ ñFõÏ餜^JþwN/%Oã—Æ3­{”‡ÒzE}Tç¡s«I]§û8¸zEê õÉQ½“ÎAÒ|Cy%Å­¦¤þÑyå¿êõlj^Çä}Q«ñ›Ú=¶äï÷0îi­ØKü5Ü\3™Ç?Hæ¨\XÚ¯‰CÕŒ0ô®$OPþx„ìáàæ¢Èûàâ™ÔU®®MºÕk”,N1ìdž#N³cÀŽð#|ÆŸèKÇ‘|ÈÅ5ÑûCˆ¿;”|¯áŒ{d^%"˜1¥c^ˆüÂÆ͸.¶ÇE’obþ+Œˆ%{1®0ãÓªI¸FòC<#_B¢#lj ©‰¸ÌÊž3q“øåSˆ¯5••iEÎØðÔ#ƒ½ ÂÈîÇVäŸEñáæ—¤ú6D§¡ÌY1ceÉI(±ˆ½$¸ ¥Ç˜AA3Î^óö¼€Ê£üÂÚéçÇ¢Â-p¸MôÚJÒ—¨dÇF£Ê™ C5yýjv º ·ƒè3œ_”¼?ª×P^I}²œ”ö)ÿ#üq=ñƒq~R:‡Hx ûkÑaé\Ê¢Ïle¦g,„¸øÜNæê-ÉÜ(õ™[“þ7÷tu|ÁžìÑ¡{3^u"¾-gvQC#_Œ+ñºœàAæÉ)ß÷"w¢´ƒÀà˦ù½]z,“åuí ¸èì.¦ñf‚`Ö3 çYØ ‹0r^™­3:Mˆ$>öhvÒø¥ý zþi?’ê£ôœRŸÌ_8üÞ_u†àlKRçè|âNâ;äö­‘óHçoiÒ–ø±¹=Ntïù<ûÈœ&õ¿82m;§$8E?\óµ²Èë|rN€,C»ö;Ñ= Ô†]-|×5Eü°îŸr} ºŸäÅãÄ×êMžÿ‰ßÿ ùÿe ©Aðcܬ«÷Ÿ™JZÙ„@&ëÌ Gñ/ž%ûÞ‚‰N|ž]÷´¡؆2ÂÛ´uT×(ÒWŒ"|(šè•?ôg4¸ÌРGp•]ë2 ×™©àEóϬ™Ù|ñR% dn?‘qy: ð&¹n7ZØn®š\h¬=RÉÜdº€EêA™ïÊ\ϧ¹ÈfljõKöìä²ö×`ºþC~÷G~[ÂÒõÇÆ£9¾­å(fÃØ¥L·w ÜbÛí Q. ±v5¨è?ÒÆþÄxÜ&:nå6± Š]ûÑÕ¤/TCòt-™#®cÆ@¼®¢žQ5ÄÌÑ@pC#™ßlbÛ.=ÐLöHÝ!~åV#sÆÁ‹62ßFÏÏFRW(Nó%{Á¨„Æ­+´^uÇ©4î(î¤q×=)o¤¸–ê84Ži¤¯ÇÕ-ò:TϤ}Æÿ‹½·Œ®ò膃k)-¥BB‘B[(.E.Z(PŠ»»;$¸÷HÐxpB„OBHpâ§¥Fݽå»gríÉÓðu=ïÞ²ÞgXk¸gçÈœsŸÙ~í=Æ?äº&ßH½—Û¾5}«h·Jhõă žMúÖ‘O¥£ô_4þ!ël$ß8Žuï4<¿:&ñóMæýÊx“«´üjêï*x7Æk¥þ~ÁŸƒïw¼·1WX [ÕçⱤxÒºc½°”8ÅåÔ‹+W*S «?+æÿᯰN»9ƒsúãè°]ùœ¾U¬‹ó&~mó´Ui_eÅU{[t!;¶qÿíè¯#¨9}§ßÚ÷ÕÝ Ãñ¯!qÝ`~¿Mº±v1¯¶[¡ÐçnÁ>uÃŽœ€°¬/¼O}ó‰WŠ Eó°ng31/+ yްŸÀ1®w’Ÿ+Vo赈'žå´z—U ²¬ÝŽ ‘ý“Ø/æû^œ'žïbØŸ›7åànØ÷.¥´2\F#µ¬øH«ÓPYâHï£Gó‘á™-à3UÙã¾0d5­øüü¹…p|v£‰jˆ7ç¿ô-ø÷¨måyyø<þýÝÿü{ÞÿŸž©yÆýý {°à˜‰#‡LtsqÉWÊ¡ ó™Ï»¸º’k$ç)¹Fj®‘–k¤ÿ}¸”Ì^GŠÂµjÔ©WaôTWÒj×hPa¸|Ć5ê44Tá:jÔmP?‡¶/ýo/-¡ît‹èu;à~1¼ŽÌ¾Fî;Fœ}õ_ÆÇ/òñ^|ü3þýpöuÇ]>>3û³“ççµ?ÁÇ¿çß7ñºoàã^Ù×½ù8¯‘É.ü—ý="¢²ÿî“ý÷[ke_ðý"fññ¹Ù×- Ù׃»²¯á¼úøðñF|ü$_Ÿô÷÷ß2˜?ÃǯóqÒ¾+¸~=>>‘ó>ûçëþý{ìÏÊþûæ5ÙWï?²¯{yßö'ò}ø¾›"³¯{Úr½ª|}_ÏϱçU®×™çë'óýùú0®³™÷i¿çž¿øú}ü;×÷îÎÇãÿþ=vñyž|ÿ %³¯¼îâïíÉßÉ3=ûÄÏÊûàÉç{òó½Í×ÿ‘ëqÞ¯@îÃÝüÝ<ù9=¸Ýøúxþ]Þ‡¿wÐà¿¿ÑÙ_ôMöu1ßÇ7ûI-üø:w>o1ï—o‹ìëN¾ÿ"Þ‡…ü»/÷åÎO³¯óbù8¯>¿ðñ³ü;÷§;ÿîÃϱ³}öuþB®ÏçûFþý{lŸý÷Ya|?þÞ^ä'ß›|ÜïÇ}îÅ÷õéŸ}ÍÏ?Ÿ÷}#?¿ïÙ×9üû‚ |=ï“÷é ¾~aáìë~Þ9|þ|þ}ãëüJ?•PJM- ^¬Ï1d*ç'˜:f’HºBn“܆Œ¯9áïäH!gŒã6"çÑlÒ<:tüaãrÍ&Gf†—å3ˆ®,:|ˆÛš#§™0"û•DõæU÷îÁƒê=Ô²|ÌÄá3ùëAöærù§k Þƒ…y-Éë3¼>ï"ÿr¯“ošëpªñÙ«”³ yÙa‡vØa‡vØa‡vØa‡3vØa‡vØa‡vØaÇc0vÙa‡vØa‡vØaÇc0Âí°Ã;ì°Ã;ì°Ã;ì°ÃÇí°Ã;ì°Ã;ì°Ã;ì°Ã‰vØa‡vØa‡vØa‡vØ‘ÝpNõmËnW«Fmis—¿v†2/аFFBä©%“Ú2©#“º2©'“ú2i “†2i$“Æœä­]ËÌj›Y3«kfõ̬¾™50³†fÖÈÌÌuÌuÌuÌuÌuÌuÌuÌuÌuÌuÌuÍuÍuÍuÍuÍuÍuÍuÍuÍuÍõÌõÌõÌõÌõÌõÌõÌõÌõÌõÌõÍõÍõÍõÍõÍõÍõÍõÍõÍõÍ Ì Ì Ì Ì Ì Ì Ì Ì Ì Ì Í Í Í Í Í Í Í Í Í ÍÌÌÌÌÌÌÌÌÌÌÍÍÍÍÍÍÍÍÍe|µkÕÊ™ÖΙÖÉ™ÖÍ™ÖË™ÖÏ™6È™6Ì™6Ê™ÊjVàXcŽ‹8Và¸XcŽ8VàXcŽ8.VàXãbŽ8VàXcŽ8Và¸XcŽ‹8VàXcŽ™YcŽ8VàXcŽ™YcŽ8Và¸XcŽ8Và˜™8VàXcŽ‹8VàXcŽ™YcŽ8VàXcŽ™YcŽ8Và¸XcŽ8Và˜™8VàXcŽ‹8VàXcŽ™YcŽ8VàXcŽ™YcŽ8Và¸XcNÎÔ +pdfŽ8.VàÈÄ +pÌÌ +p¬À±ÇÅ +p¬À±Ç̬À±Ç +p\¬À±Ç +pÌÌ +p¬À±Ç +pÌÌ +p¬À±ÇÅ +pr¦VàX#3+p¬Àq±G&VàXcfVàXcŽ8.VàXcŽ8ffŽ8VàXãbŽ8VàXcfVàXcŽ8VàXcfVàXcŽ8.VàX“3µÇ ™YcŽ‹82±Ç 3³Ç +p¬Àq±Ç +p¬À13+p¬À±Ç +p¬À±Ç 3³Ç +p¬À±Ç 3³Ç +p¬Àq±Ç œœé£8%]\JçŸòL°;±“ÿ½ÉH;±;ùß›<&òÇNìÄN¬Æ·;±“G:yL䨉Xo'vb'tò˜È;±;±ßNìÄNéä1‘?vb'vb5¾Ø‰<ÒÉc"ìÄNìÄj|;±;y¤“ÇDþØ‰Ø‰Õøvb'vòH'‰ü±;±«ñíÄNìä‘Nùc'vb'VãÛ‰ØÉ#<&òÇNìÄN¬Æ·;±“G:yL䨉Xo'vb'tò˜È;±;±ßNìÄNéä1‘?vb'vb5¾Ø‰<ÒÉc"ìÄNìÄj|;±;y¤“ÇDþØ‰Ø‰Õøvb'vòH'‰ü±;±«ñíÄNìä‘Nùc'vb'VãÛ‰ØÉ#<&òÇNìÄN¬Æ·;±“G:yL䨉Xo'vb'tò˜È;±;±ßNìÄNéä1‘?vb'vb5¾Ø‰<ÒÉc"ìÄNìÄj|;±;yT“’..¥ÆóÜ&¹ ±„%,a KX–°„%,ñ¨‰£Ç¸°„%,a KX–°„%,ñh‰Â'M´´¥-miK[ÚÒ–¶´¥-miK[ºÀÐñC†³„%,a KX–°„%,ñˆ‰•e KX–°„%,a Kü/ €[–°„%,a KX–x4Ä¿v±´¥-miK[ÚÒ–¶´¥-miK?ôc²„%,a KX–°„%þˆ¿™á–°„%,a KX–°„% ñoÇÜ-miK[ÚÒ–¶´¥-miK[ÚÒýä†,a KX–°„%,a‰ÿâof¸%,a KX–°„%,a‰GCüÛ1wK[ÚÒ–¶´¥-miK[ÚÒ–¶ôãA?§÷Y–°„%,a KXÂÿ Ŀʲ„%,a KX–°„%þˆ¿À-a KX–°„%,a K<âßF»XÚÒ–¶´¥-miK[ÚÒ–¶´¥ú1@eY–°„%,a KXÂÿ Ŀʲ„%,a KX–°„%þˆ¿À-a KX–°„%,a K<âßF»XÚÒ–¶´¥-miK[ÚÒ–¶´¥ú1@eY–°„%,a KXÂÿ%\\J‡Ç³m[¼?e'–Ým_`CÉF˜•µøh½–験²¶Ê—ÉÝ0=Äïé>;²0j_Â×+ŽÌÄà­WÌ^;}W‰Cr2zÅ~jÛ]ôhï3¨àO^è¶ Ñ™–ã<Ðõýã£êçA—?õèßõt©—ÿ³vkJ£ËòOb'¾p]{¼„¹­ÃnôL,ýrÈ÷?¢Ï‘»?þ*ý3~ÝÖ(ì= :û}¥™{ŸÀ¶îL›4C÷/ÚÿÙÊé^õ¯)«–]ň!žÏ¬¿ö'FN­Ó~ñÈ.Uó›ÝFäŨÍ?È|£>?X²Ò°pŒî¾jë•ôSýI¯ ëŠïÀ®3vçÉég#c\Ø Ç[¬Ãx®;ñÍy{:o‰I\rìþ|×vwÇ”ã=–µê L½øÓɆ·½àúçgÆÀ4~ÿéü3ÕÚ ³_éÑ+e&æVŠö®Q©æwš>(kÿ>,ðdðú)/cѾ-• =„ÅÏh¼é·ÅX²¥g¯ßÇø`Y³„eûk\À —Ð3ûÚ¼ˆ•÷\žù%j%VýŸÅº*K~Hüø:<Ú>ûMÁµGà¹aø‹uVOĆowÿþÉøÕð½|Wf‡ðþa`±ŽOÝÄfç[oß^¾m{\¥Ú0l-ó½û/ÞÆvýñ›cg¥‹k?ØW~aßÍ ÿë ü?Ù4÷@h[6ŽýüËßZ!hïàq³"ä­ˆj‡³&!ôAï7»Å|„ÝŸÝêS¼h2öMn¶Åûm>©ôÏÒ[ ß‡¾ºð{é(7D,\Ñy믉ˆªï²Á¹uˆ.¶æ÷?F¼…#yo»¤D?‡ce[¾<{&NôzæÈ…ÑïâTÔ»¿:qe_hòæèpÄOúäù]¦âtæÐ ýãL¯B+‡4CbêÍC­®ÄÙaß¾U¡_,ÎýµlqÓÁùpa矃ïwü—œ_îö.=ðlq×}Hn¿¿×ñ3‘RášW¿ñ ‘¢·gM¤žß»ªÿG‘HÛ¾¢I`«fHŸ„®Tí…Œ?íQx22K陟¹Í[8p®&U9}ûwd9?Ò3#îãÚÎ<1#?l‚ë;â/·~×7ö}u÷û Íq3©ä…êwà–úøS¾Ç<×”x® îæüºÇ¢Ó¸ëP/üü+VܼÑðÜ–ûX2½ÂWkŸÂ´'4ß¾‚ß.›ÿí%wᦾmí£ÆýÖßýYg§…¢§óëÇMB·Ý¯ÖêQg:ß}ù™ªME‡y§Ú.¸°ínïÚÆï=Ì«öì‹Åð&?¿Rjl Œà÷q~¼_Ó÷Šb䡲Ê—\cäÍhu׆Ø˅MJ­€qå6DÿUø Æëg5Ç„·¤¿Þt'&M+Ù¡G¹1˜¼é˜OÒî˜2o܈ݻ `*×wݱ?Ã#v?ÜîOë»t¦¥ú»Ëñ“˜Aù5kÒ®šËSÏbN³Bgßpóøý´X>ÿððØÄÌ“gc¿×âƒvÅõÀ’=%;þ –5ZÖ§ÝßbùÇ“?ˆ{5 +O¶¿úî+žX}ôf‘…áoc-?§G…„­Á÷ºÃsôÔ° ۰኷óIòÀ«Ûºï;Œ}Þ4é4dT%hñU¾F·=mÎ.¸s¹6vÌþ¥ýþÞð ¼Z´Ùä h±7zû—AðùÇOow»ŠU®09&{¸÷—R7¾þ”VË«¬;)ñÞu#"rûÙo"ŠGný’³•q䥈ß&|}Çúe<ýŸ+8qâ—\óçG,åXÜÜÓŠÃÿé¼×ç¬| Ã} ^ ¸3_'úT(‡¤å'®ä+ôνº¤Qá¥EqÞ¹{³>Ú‰‹+(ZmÆW¸ÜrϬcû­§¸¿÷H¾¥)”Ë©f•úô™.H¸á§O«#½É¬®©õ¾@FÑt._WVh4™‡&wú¡®ò÷Ëz»’,¸V~Ýà…®½qréåÍÍ*ýòõË(…[ýjthQ9·„¶É*ð1î>yeáͺS±~Áš]/¬/‡¥¹äù®À7ZyV_ ×ï<öFvì‚Q®ŸEŒò7û_øCøEôZWê+у×Ú]º¡:íúíî?ÐÙÙÄù&_6|+Ïï~9¡\JÀKè5áú寞l†>”ÿýkÖýú[ï|8kÔùÏÜæc°³ëG'nľÿÐV!/7~l¿Q;ùòG·KŸÇ°OŸQŸ ß.pê[v_áÚ¯ ø·Ï g} ­¥Û,ÁnòýÞ/öÄ_î‡Ý#×¾ùñ2„ÍŸ[ðý­ž8ôá¯*׃µ[\DTÛý¸0”úøåÎ(m6ø’^ï²9 ­7Î:RnùÆÛ8¿<5)Æ#[' ßÿÀ—Kx}Õ¸æK¸r«U©mú"9Ë'üþŤ$Ô9ß+t9RÃÜB§œ›‚´mCý£’.#}y÷ïgô[„ŒI<ål1dö«¢Ë¸Úð½ÄЯw ëeˆ-GÖçZãÚ±7¾è™‘ŒëÚŒ‡-Ç­ßZêYÜø5ñÕûÏÃMGÉy4n½søôôb?áÖµC˜‡ÛJÆßÂ/oy}Vtj?qÂÖ°@©Õ7waj§—çå]U#»L»µS]{mÁÝÕª·ñè;8½gÍ"Ðý—¨÷F‹®ü{'ê…÷¸?Þu¬‰C[ÞBÚ‡­©?å*—çµ'¿ÉûˆýÚíµ_¿të tJ ?u£zj3z¡±‡ûðõb'÷ãúb×H (ا^jç_"oE úîÃu«¦ÜÆ`ÿ†ß¯éÞ0ÇÞå÷zÕõù?:·Å0êùá\w„ßœúŽˆÅHÊ)±wÇt®[ÞÙ­vÆ÷Ö‚i7N~^O` íò©JÜvþ®·6ÿôÚØO1mü“º¿ŽÎî:¿È3i/Ìv´ØŽ/bsøµAÕ׺ÖÄÂŒà#Cל€;íþÅFˆ×ÔÈÑå|Ÿ•”#«©g×n¯×ûR…0x”WìxR/m,¯^™þC>üy$6E[|åõ_ásFo(l-© lWw©Y9øµh¼÷ûFSà«LÅÏêoD R/±>ÖlY ¡úg{{ÖÇÜ/=±ÿe8|Yà£ïK,ø ‡Ç˜÷<"”™;`-¢Ô³F»"šúÿȈ4ßI»ÞÀ±ðžâJßÁÉj•‹ÍÌ×±ϼÿ3â.F§;" §ÛUr,毠͹*Hœßµôô•çq¶öåƒUʾ€sZýuÄ…ÓCJµèŒK;®&þžê†+KÎgÌIúÉG_z¯ü-‘µýfùY£‘9ìÝ9n¥‘F:=æ×CµG¤#ãD—š‰Ç?Ef|ç©…û=õ.®ÇuþÁ„ºEÞÆµKƒ|O¯.€ë—v,þ°ÊAÜp¬ÔO7TÁÍ„3›;Ž8…[GÕø ·÷_4?w¼ÿº>òé%¸Ûl²²,pw_eñbýú6µZ4Þ‡uíՙÕ?,ÿãH¥@q6µ#báu­ÅÌM¥Ò°8ð‹¼ßýºÓÔ.|#“n}~-ϺiG{q”ßã0\}ŠÚOcèÄÍk;ÝibôÖ`ê £ÇúèƒáÔ££zðˆ×Ò1V¹-oÝÂDê¿ÉÜwS߬ž4+$*ÇÎs¤\ùŸ›bšúZ‹=0]›…oa:ù\ôÐtúYÓî¿¶ÒçÓŸœArfiÍh˜õâ™ÞÃß©ŒÙÔsøýæP¾Í}ðy¡v3Êa¾²NóMɱ©§Ѿ^tãü&ÏÐpŸ=Ìuü,v´Ðœ›ß`ɹ¥ßLÓ K©‡—7ßÛ`ö“žXAûcýÉÕwÔó&Ö†×(Ñ3ÿ XOÿÃ#­ cu†çí1ûß>}¾Ö¼¨ï½µ[õ 6ѯ÷©Ñ¨Í¤WûÂ÷£¯íNù [/ÕßÕ{VslÏÔìü-.u{RUø¿  p4={ûÕÓ/#püªã‚¼ÑìàÚŸƒò¬¾aÕfDOìî›ùc«cØKûrÿ@µ¦áà‘F[»ºÌÇ¡ê+Ö/+3áÞúÆ#Òñ^',ûQ>®ý‹Ÿ¢Õ¶Ûû=b®ßûØÙ²8ªÜÛ Ëqì³bþŽ Å‰×zLyo;Ni·j(b5ÛÎFÜ wîF¾_ñã~™î¸b8­Ìk—Hx¿œ²T¨Ùñ$}2¤Zãü8=®ILÓïqöË%ú½É¯iÇ)Ôû©Tí¹÷v ¤ûªI+dx¿ÈùüÈj7´ÒûË|p}HÈï }ßÑoAc®Ö{wø¹îòwø€rá^²§ã™ìÁGSõƒOê<áë¨Âÿz]¥vYÑ,wO[2ý»,¡}x¸ó™Ïk7ù^·µc€%ʽúÅÙ×*œàþ©‘ócF¼–æˆcï Ÿ‰ý:à÷+£*ÆÀŸ›éì÷F Ÿ§^M;füæ'çì(X“iÇN%¿ºµv]tõƘÞõDÓOÕÃÌv%Þ*6¬fÿvaѹý0gÊ´u7uÃ\e^\Z…yEG -ñi æÑÿœG{rÞ·Ï)ÎÃü‚O~8Ëó8æÑšÇý>ÏUt·Ÿ&úhîá%«?ïWs×lT>æÒOžë¿ZEÌóæq?Îã÷ž¯Ýö X ¶o¯M|f!åÒ"ÚîÛ) käÝú“ËTÔ¦] ,×—·°’þêê+é§ŠxïÅZ®ãQ5ï¹Âek“÷ƒßȦn¿÷ÀFeeÔþÞmúÖ:q›v£zÑcïÀçyx~Üú[øª_sÞëØZíɧ‘mÉÚ!ÅŽMJ€Õ„c¹¿ö2üµØ}¿öÒª"Ôneyñs† Kêõî¯ØUF°çi­°±¯}·û}†Qá鎉Œ0-ÖŠáPFÓ[Ž(@¸R+oEdÉcÿ¸\QŽWxyi=~A9l_#ºþüQ%^/…Åu­ÛáÈ×®±q¬Ç±}SЇ{†JúÔ Gli%ØžFÜ¿K®Óµ>yÐûÍî8S±×èmþk4 òŽV¿À9µ«kÄEí¾€äb?^]ñT8R>¨ép^(Ò¾úñ‹…IMY_)ÂÈ:òcŸˆäy¸áÛ móâ¶RSÛp×QŽE€Ží:¦Ï`|¨Í£^øøâûgÜ{„á~ie üŒÏé—þ·ëƽçÚìxž´Ï–+îÞQÊ<„ÅÊ»~6ÑØ§#‹(Aø†¿­/í²ÞÔ+=©ozPî‹Ù{Ìâ.½cª¡_ÿ¹Ž¥þ™‰‡U·¿î!÷Ûàǯaâ–Søù\Ïît4ÐL§}:“ñãw•8y'cÞKÚñ2ü6ÿƒ–=7}´ h—.d\eíE£¿tX±w.òú•q÷°DycþÄ’?¿?üŽÿ,u¬äϯåÅ2G”ùÞ+—XYëä_V[‹UsŠ5~¡êÛXýá%· ³žÁZÊõÊ ê¹ŽöÍóAox¦tlâ¨<6aÅñCàåXû-[ oåŽÔLÂfµKF¼ú¾ÊÜôù[—ÌúÛ>íÖ¤äŒ6Ø1Ho8h³±lü(çü{¾àX¦~ ½¦5ìS_Ed¤¿vEÝLÄ3O#xªæD„8ÜÛ}§7BUزí)„êm7»~Œ_ß¿ÚOØÍxä^Æ[÷uý¤†ÿË¿cØ-÷?§}€ƒ´ÂTøè›,Z{)b®ïA„.½§Ê5DЮˆb\ôðù| a;|£ÂOÝÏãÈ­Pqlf¾Ö7 Ç îÓA.¿ïr‡Øç”C0±7•#×qJ=Õ¼‰x%¦XÓGœô;ú篱æÕwp¦ÁÏ÷R.–D¢ã%ÿrg6”QUìÉj8K»û|íÎŽF>ƒ ôs.ÍR !¸Ò¿šúÆHÖá‚‘Ha|&MI+× H;ÿ×/§ ƒŸ÷ª›ðàãC7ò©@ÓxÜr9~rÈ•¸£¶É•XÜÕfõp¼ÿ6ôw¼÷ª÷g¡ý¥îÌ't!ŸJ\ÆÄCO‘x«øoWSUÆ1‘ñï)”ï®ÅîÞN]qnKL]ö¦1~1ý£Uu~x·f¼_'jà·a¦ã…;_³è7Î:°Áñxóa6ùzöýÊß½’\sèGν;í§?+>0üº€übò ••a²îÜÏ?Ȭ´60Þè­¥´{—)¯fDO¬xCÿpXIù¼šöíµë ǺˆRêçÄW–¶Ý}üÏöØÀ¸ÕF¾¿w9¸Ã¦ÏÞx"«6GÔÝVâ|ux//¶¸Öø±@òoØzó›Z *]ÂvލƒÙ ‡?.n²®=ü|ûœù××ð¯­0ø'¦¼_øôp”þÌç[ àGŸíß>÷½¬|ë 5× AßdµJ.µÛø!”“¡ôCv½ò[Ç$Ánþn{Þ>Ò¸åóØ»lbË=³Žcß/9–Q`ü$¬¬vÜqÈ¥ùg¶ìÁ!ÇÉt: áYñAÝo/@„º»…Üy¹¬ò´‘-& âðŸç®{äG õÔ‘e>õK¯vÇÑÞ}ÚgÅ/8¾h’ËèkpR…wîF¬¶*~Gœ#ô:µŸˆxe¾„ü€ÓŒ_%P¯ùäme‰"IEOzNÀYG y-ˆóá§n´ïŒ‹«¾ßóÁf\VQºÎ7ܥ婩S"eê”A. 5D ’Hc>)ƒ~zæ©3½XYÔ×>]7ÑåUÜð{¥k÷Á{pK»‡SpG›)E=êÎß_âôþÏ+Cq›ÉWôfü@â›OiOûìÝõåšg¼òÚð÷y‡qVŒsÈUþ.Ï“üE'Ú±]i‡ö jâ§´ûñ÷ÐíÒ­‰ùcŽ¿4ê?êνWs®ÚÜéó¸#,z>ˆ3ùã·2.$~ë°Ê;™_½3¢Ç™š®Åû%öø8•eP%'.Ã÷J¾u¥|™Æx³È ‘ sã\>>xñ,`Üz‘ú™Áý}e€9zØûùW}kIJñzCcÅĸùG¿Í‹UA½ú¥£oYO¿Ì“¯ß ·ÑTxQnz{ò‡m?NÂæ€¶^ mNÃwìØûý6ž™cøOò€ÂW-©çZ0OÑœû²)ã MÈ?rmÊx&ø~oQ¶â¾‘¸ªÄS%)|ß‘ï߉v­Ñ»‚ )xw¥¾ÜØÏ&þ*üNýÕŸqá÷Á\Gìòa|_¹obW nÀä=Èo“÷4ùÆ'$?4ñ´´¯f1î9›ñ+‰I¾CüLwGÈÿ‹µxꃥÓWžþÄH,çû¬ÔêpV¯VŽÕ&¬¥ŸìñŽ ˆ´‡'¿ÏFG©MŒ[/ÊÍM”çZû,ÀîOÑ»Ùâ}ü3׌ycâbF„¬~!èY«í¿øW“ïߣÃŽ¿øë€ŽëÊã N›•á™:„ÞÇHÚÿ‡¿Ó¡Þ?ÖC;8§-ˆ}%¤aƒKß9þ`vž#þ÷ù6WF‚§Nà"Q‹›'p¶œrô›ã<ãé·]v¬ åÁ$;Vd‘SÛ2âÏU×üˆÔíúÆ íZÚ;kR‹!ãC¦—öÀÕ J@Æ#+£‹úƸN?ØðáHý·”[ûôÜVQ·pÇïÊó.¡‰ó¡kþ|[NÕÃ2uûܽ±èó¥åcº|aöÑNí>W4q}Á»H^Ðäëù»Hž¾57Ñ-ÿ—«ðÙ;ô…ŸÊçóù=È罘ç¼åCy ò§ØÉ†fÞOôh?âfú3¿;¿ñ>€rh(ãž?y4ýá±ÌçL ~šL=,~µëʨ …nµÂ4g7ûû]ÁŒè·ÚuxíuÌVÑÿA ˜ËϽÀKÞXÔ;Ʊøž€;ñ1KÈçË”ú¸Ž—*tD5Vë0äóXËý©¥N÷Ûðåc\4qN±EµfÜ¥%íVÑgÍ©7Dß½IûK®i¿ -ú2·^lI»Ðسä[Y·­÷õÌÁçðýDoJ³=?Ç?êOÑ›‚'}I}Ò‹ù¸Þ”3’?>6z’vÀPæõÄß•¼£±gƒt`ã_Њ>'?ãhÃÎg>ÇTÚ®”CÓˆkš®ÃšsM~qã›ó4 « ’ïϳ˜þôÆ»·hòŠŒ/¯cþŃrÇSeÝ'öþº¿Þ€·’¥z`3ã6¾: ÝŽ,u<¼ŠØ9v£ÒäðWo×õ‡½Zç‚Õí¾v¡—¯=»üµ¼Æŏ{õÂAeîßKÅ¡•*Aþ "Þ›zæ…ÙUõÚkF^¬‚è:w:x2GîxÑ1MpìÈšùÕ*ÃIí^çE¬†ŒEœJ/ä;ƒÓCŽøíóœ¡ß™­€cî8»â%qp^I¥w?ÄE7íâ²{‹è…õ&"yÀÚYÉ!Ó‘²â)gç-D*óŠi´2ªÞŸ™þçÈx«@õzûq•r9ë†Zð \F'rø~ü­yÏ9¿ø Ü^¯>Ž>ÌÞÿ¤×·œ~à æñö¨´@QO¸÷8‚vœèÑO‚£=™Ûo|òú=ê£Ìt¦ÿÖ•ys±û’ñ}$¿.zhùBð¡ãÉÇ*8m´ÉÁ…Ò^Â8¥ÁétA¸‘ÿ¦iw¬„É;Î ÿ˜ü:íLÁ‹J~bó© ?Û¡"_p×nW÷š‚™,½e yì4V¨èÇ«µ°J‹‹ÉXÃ|Ⱥ/Nê× ðPa³5Ÿ`CˆöÁÆ5ÝÄüzÞŒë›|»ëƒ}Žé…- &3ç¶µ«—ØÑ¬¬c ‡A‹õÄ?à§Ô\·õðÿôãÂ…êÏG€ÒÏMD  3-÷BqrÁ´‹C¨§BéŸïáu°ÓK°{bo•éÄw ,ÃÞm:¡ƒ}Œ?(¢ª88)õ¥ï>\0 ?ë‹CÔGáÄÃE¼\ý¥ˆß&"R…O%"Šñ€ÃÕ]7%{ÎC´ŸÔC _DIݞ±A/~ýî•^8®Ò›Ïá¤RÇqˆÕ°ßÑˆÓæC+œVð¤r»p¦™° iÀE…@Å9Ú‡~Rî(\SÀá˜t$»ÝõyëúH¹©o Òˆ«MWè¶wƒ‘A|ÏUâZ³˜¿®¼ì¹“søù¸[Ä×Ýv¤|ÿäc¸CüÂ?Ú¥Œï ÞXâ’A_ÜW¯48g±Ïú‘¯‡&¸±+ßH¾¿ñmòÓÛÔ;­˜7|—ø5Á½‰*úGôŽðåòÓP-èLÉཀྵ/GÑߪY}Œ¥0Žùƒç&îzã8“ÉGS~× j¸’ܨ—§×mûÚ§ÇŸÃ âßfQ¿ÍѰ«B&ÿ¸`U‘¼·]Rsø’8„ÅôŸ–2n¿œzfïÛ*åíÍ)†5ÌW®Ó°…üðPæôÚõðTiÜÕ›s⪚ §Â{â¾eF¿„M_ŽšQhG_øLì=þ‰Ip6ɤ¯4ÃVîßmŒCîPÖm¾rØÉøŽßöw»¸ïÞÿaõÜO®ŽCÀ« н G ðœl#HÁÇ,6úNò†¡êÿ¶=±‹ø¬Ý ^7ø/ìa>kñsû×ë÷ ¿£_P@¤ù9vkŽJ#âЯ8ܷ¼ J"˜—‹<¡uˆb¾ð° ø ÑÄ•!®àXleáó>qÄ¿þ@+@$R¯ýî•äjþwqAß¾þ¸¬¼ù"›œ®Ô¤:»bïΞHãë2JîµSÈüjmÉËœGåßµûѧʯ’dz`"·¸îíÙ‚õNðú)ÎWüG>\ž¯ìgßuÅ|ÚWâmWÒ|Z „†Exnnô–Øybš8 í(á7¯Ée§ÊµùNüE‰·ŠÝ(v¢±%ŽÂ牟7€ü?qz¹ÊçQž/yKã'RoËó Þ@p¬Â÷Gå÷|ê¨üZ¾µ£Æ0ž$uÜ·×ǘLíÔÀݪ>oò.Óµ˜œ„™Iú‡68ïù/êR»ïj¹pEçmX¢¢µÆbãW+™'XÝ^®rò”7%>fò7ú¬¼ó¼jÞPˆ-l"ÞÒ§¤úö`Ë+Õ†þè³Ûú&L‰œù¶Éwø-òë7/.ƒ?ñª•Ï6pL.yj  ”wywç§Ý®÷±›øÉ½O¾âpF/ìWѓݠ@fMÞƒ°“:‘pâF"hGFu8ª˜óÊZˆ!®ýXÅE} Età %†ò]G¬6«û#Žu%ñÑ{ªd]ÝŽ~¯3ô“”[;+稷/diq‰¸Îdž”*ZB!•xÛ´&Ç ¸Ÿ ]&ñ"2‹¿«i¸êÜüמ+€¬RZ±àóI×_Óˆ3ÜhÖo¯ó‘q³‹ ´tÅ­ÑÛ·~ûú+! q‡ŸãnŸˆä¹Kÿ‡˜mº3^7qÁ¿l¦ý'xëŽôc$ß |÷ýËfÜâÿ5à~©G½W—ϯKý(tÆuQäæSákñÅ}jüFÁ­2)zYôí?Å[;Ó¯»Zø¿×9 ùÎÞ'F:žùEƒ[xêCúšrKò&ŽJÃXÆ™Š£Ò~œJ<ÁU»R¹ÝPÌTfó_æÔSÑÏ”üçBî³E[>·/á›>WõÔò1_`™.?x+T4g~Vkøs¬%þË£š¿ÊLçä3‰÷òÚ©H®Ø¤ÕtøU=ƒ°•¸¶íÕñÙt²År(usç‚t:ã„0ž±K¥Ñª‹=º¤ösÝ0‡¹×á+.O~ó÷ˆ$ž2zÔ7î¬oƒ#ÄWÞLÁá8úõöð%Wsª Yâ4ïßå.&B’úXÝq¾¤„¸¨Ã§µpÅ‘òÅW!™ñÎTÞ¯´JøO¿ý¯ñá²´ãnp`ƒ˜Ç–ø´Ä'%®!ñQ‰—ˆ>{ó-îçÜߟ¹ó¢'…¿$Þ*ü.õ‹ßÍx­àqúÒÎ̭פ>Cò¢?åu¢Wå*—çþbœUôä â;%ßhpyô+G2~(ø¼q´ßM^‘úm*íicç²Îcõ‰ð•©Ã Ý¿˜8²¥{~^ÿÕ_b9들.q ?ï: ¯ N?Ô{áž <§ÜH\—·†“Äf®}í/ø2Ÿ±•ùŒíŒû=§ðw´R×]­À¼vÊbÖö@°ÿùØÅOOB¨6*a×7J–ÅÖÛí;_ãð‹aŽŸ¨Ê0]ÝƼv8óéôw£´û×ÑyJ*$PNž‚qÌ“õ6«È2b™§£º@há/?z ºÌòI$*íXt’T–®ö¯8§Â}«¾ÁEÖo^VpÈŸ¬ÍW¤(+&Og¤v:¯"HHSîÆ³Ûþ5>”ü“ÛäáÊ#4¸f+] =¹¿„ÿ$~)zJøéMê›F´WëÓ^½W‡v\E/оlȸ¼¼¯Ø·G5~&ßGüÏÜyg¥Ÿ*ùÑçbW‹¾”øjîC±%iò¢'‰ç1ö²ÔiQHþCpG&ïÁÏ%~¬ø¯Rß8|?™q,“÷`}Z %¸g<ûLæ»f¿wÃ/¸ê¦‡ëç.Q¸3ß,õTËx?_¾úå俵kYßêA»Ü“yA'Òp›±ù”.´Ã–w›*„$¶SŽûh^ch£ëðgÞ$ð`ôý'Š`âVv)ÔyëíØs]7±Ÿvg˜£}ßûÆáûê¼ìù{ DéòäˆöW•]8Úz»ò¼q<þ—†‘õpJ¡{žLFq)§u:²;ÎìW÷¾H"žà¼J¿»…â"qSW¨_“™ŸL­Ð:iø¤)Øé€÷ÿ5>ôèªÆXÂúX©£Û¥Ãâ®/"8Á‹J]¡Øybþ×ü!ùNêždÿËûIÞPì@É ŽUì>Á»Œ Ý(uøb÷ >n ùc,ããTz¦ÒOO\¶‰Ó0Þiâ«Ê}›˜iü:ÁµšºBâ&f0î6‹ñ¹ÔÇó_YÈ|„äg?¥s7–P.e^Bò¸+诬b\v-q²úv<OÖ{‘ö¦wê»vòë‚Í*ý0< >×›V|~þ\la\ÓÄQ—4­¿'®¨ºg=üKé@TN]ït|~A´ C4Lô„2^±[EYæycOØÙUm¾x€½ÄÑî¯ø fÛOCq€8Ƀ~†0îË0•s~ÊC¥µcCÄó„+x©Ï-D(+þÙoÉz½Ã¬kQÜ8´1Ž2ns\ëà„Bë4»†SŒËÆQÄ“hw'ïzîO 4ÅEÖ«\ÑáæHÑpêrHe½Î¿Å‡«hßÍ'žsã|¾ªÐ8ÍÔ§÷e~Nâ5â_ ÞEâ5¢§Ä.•üàCùCÑ{|±g…ÅÏ“|GGêÇÜ8U©ƒ4þ›äù¸ž±/©çDžŽÍàaø>Rlâ¬RWÈxÏDÆ1¿6…8MÉH?É÷ÍRÕâCŽa6¿¯Ôi<+ñlñ+￱CYo¿Jûb ßgï‹ÇºÛÝ#×61ùû Œsz©w­ÝÞWö¨ 9lVîØwÉðUёⰅ8m|¿í‘«ª° ;T9–Çfì¤ßëWK5º¸?Æüù9‰× JQÖ5a^&ôÜ–û•¿«†]Y_xŸúæìfÜ~O %öb/ñÏû<åHÖ÷p€xë°IcZjTáO놕Þퟂ(Uæ—/¯+váãrˆVfn¹8ú’ tOʼn~XØ-{ÞLY[ñº,$ÎlÑ€bœå~ô©—pA³_'\ÞóÁæÆŠ8z1;nšª¢åcöÿk|(rYê÷ÄïÚô§p÷ý%| qMÑkbŠÝ˜×&ü&v§ÄoÄOÿPøMâ$½h?J|s õ–àFM£ô¿`~A®ÒÇfùd0÷ÄA þTê©g‡0¯/¯7ý4¤˜ö€ô¿Ãû!zVò˜“ÇkÉä+]É8ë ~¿YÄÏ¡^™O½¿ŸËùƒÅY}Ξ«SKY±œõ¨«Êj5”ë™W÷dý߆¨ßþyû$¼zk61°ù{ehþfòÛtؤ8vð>û1Næ¯ÃšE¨áä}ÄŠ¬<¨ï5Ó/ÅSU£wk‚ ôo½˜w1}6‚/÷{ûŸŒ^Ý>§‚ªˆ‡_½ g2-ƒ?ë¿ (²'p?Bï²ÌΓØÅº¨½ìµŸu a´3Â_{.ÿƒ–o#’}0³N0†õGÕ6úý×å~ pŠ}˜ozšù¤%î_B"åOëOΑ¯.¨v;Wq‰~啇lœ‰dn¥çB ûa¥.W¥¦àëW2¦Ë«ó8zT;XH§]’N=•Ñç–Ãyó¡Ë/ë!“öY&ë¼2ÙÇý]_,qI/½]Š>”},û_ôXn=(ø4É?˜<†ð'?Ÿü]â.‚K\€È‰cJÔÄ/sá@k3¢ˆÑoR_1zYðÂ?ŸJú¡~ÌßH>ÂèMêWÑ»bÏš<¢Ôaðý nMâ¥Ôß&H;DðÜs«:AàšVWsqM§9‹âúK:àŽë«U¿ ®©:vUÄ nº²ÈÄÉ%ï,þßzÆÍ%_!yºÜ8RÑcbwš|!ùWôžèÁ7¸¯E?ŠÞ}*ö¬ÉæÊæ¾J<Ô\©?M^‘üŸ»/ŽÑŸ/ÜèI‰—òsþc¼”úØàN%ŸÈ<‡é#ùDâFòó˜ü?ן@þŸL¼öêWÁº1Ÿ?}ùky×3—2x;ö;g¡n«PîÄÇ-&?.-<ù)ç«aå€éËqù«'›F|„µ¬§ðÈÓiuïu áé8/ëb#ãûÞì+³™¸7_Õ¥¦XelãþßÉ|žàsuÚn‚É»ˆëÛC=º_yåÚ ŒñÝpèÀ&"õ×x‡5¬í]Ä~T·K+ƒô·b›jà ☯8Mœk‚êòQ½çõT•c8;ø¯6¾/žÁyâ€.ŽŸÞîÎ/•rðmµt )Ú<ž‚T~ÿTÆwÒ6ëÀÒUÈW_Dú¿oòÓ&öÏØL|”oyýÁ°…uŠÛT{‹KØA¿È/¯ÜðϯÚ(­ò~Ño fü#D§†"TU /ˆÝ´×öÒ>Û¯ªšÁAö‹ c>ÑàÔŽé†n&o©Ò†ý6!jð[+¾—„ÃÌëþðÍЭGo"ZÿìÏ!Zµõ¸· 1¬Ç=2Y'FqT·­ŒãŒb¿ÂXâ"â>üyÄŸ«n žqåÖ7ži]4¤ý‹‘Ⱥ“³ÔCçX/~¸Ò )qÙ_ôRi]xˆK[µ#Ëì×sYÁ òîÄe₯°OÍ•=¶6+tWØÇ.ùÍ@HnÜ»äÐiHfÿ…dæ —3î>“|$ù€m*Š\pÁyI½žè “Ÿ >ûo¸ñå*úSìWѳÂÇŸ’×7}™ß—>§¦.*W]„ô#ܸإÆÞ¤Ýeð/ÌËãûš|åýCu‚[“¾¥¬Sqåý3zKð¨Ì»Ïâ}™Cý'}q°^{ób‡.¦»Mì¹ÜùÃÜõP1qîW‰£HœTâ â—I¼c$í°±¬×¿KúkKÝ»õŽôsscl7ÆÝÌ•ví4ÖW‹þ“º@Á}š>TÂ7Œ‡H¿›…ìk±ˆõ¬îŒ.f¾QúÚ,ë° DZV_ƒç]©ÛÄŒ2vàÆÖÑ/4zP➌·H>Ý‹~Ó¦ú*A¾›·ŒëöÝ\ø§*„°E›=ͱqõí:=:+§ðíÁrŸSO…¨mà~ ¡žª‘è_Ø¥P+CŸÄâR÷±þé€û·D˜Ê:¾¸‡tze ÂÙ¯*B—ï¿HÚ}Ò?8JÝžóqˆb¿ÁhúS1Ó:5œS .;/GUXÈç,Ž1ÿzJa}ˆ“¬ÛŠUeë{‚÷¬|ˆ×?qœæº ìct†} ’˜o9˾dç>m®:ÅükúPð†/¾¼VZ}hõáÿ=}(úBúãŠ>ZþƧ‘ü€øu§‘< ð]îxÄc^ãóä*x7ñ%~"y„nÜÏR—húë3?eâ¦ÔËÏÌ75ùBéSÃø‰àhÿñ*xRÉ3ò}¥†àÔª»¢Ü1x5Æ™¥ød~“W”þß|ý ÞS§O<ÖæèEÝfý¶ùýº¦_ÕUØ!–±hûd­bâµVúLN†^¶76¨6k¿]ÍéÁüœÏºë ·®€-ä—m콓v“Ÿª.*óTõS—>dŽ`Õ®ëi_„Žêàh¤×±[qÃjØÛ]Äsú›êrÜÎ9ñSöaTÑ ‰rœéNÓ#ZiÃ<#pDEc'mÄ1ž+†ìÏvŠýŠãˆó‹W0¢«ßâ´Š"UŠÁ]2‰ÄƒŸ­ qNµûñ4.°¿Å%Õ £O$®hws’«@S¤è2%W¤²@šJC–Y´8ÝØéÃ5ÀéÊý.å… Ö3g2žÉ¸p&ûl]u”éÔëOã*í¡¬‚*PÔYº}_'LaßrÉ‹ ®Zêk$ß&8á;Ñk¨ÄC_§ž«F¾~…ûK®¯Rß ŸæÎ# Ÿ‹žÍ]¿øßp¥ÅSsÇQÿ[ü”ü#u˦ϸÄO¥~Ÿv®È ƒC'Ç䥎Šq.Á<¤wÙ§KÎÑ: “W”¼óÒ·nqOÒŸnq¦CúÇmŒY]‹õÏßÞä—3±’} Þ”ñ>“OäçØ nÏs_À‹õ bûO²…ùÑíºí÷`켬3ŸÀŸý>¤®1èºj„T¡åT±v'è?`Ÿng\aÅ]UgOb¿˜•v¿yQªÊ¶é9Ä𜤣Äêî¶ù?Ã)]îÑqîÚñÊ])ˆ½DâõΪîÙÃêã¼n_\Ø¿ôí©+ÄÍ%³¿E åXjaÕkRY÷ŸF<|íÞtž’AÜPÆLUØØÄÙfò¼›LâjÅž’<˜äÅüšìc‰Jþ^ô à»%O(úNøRøQ®’ß—<…äçÿI>ToDüˆàNžãUìi±C%Ÿ`ê1¤_Ô#’ïL>Qh©çõøyE¿wîúDê7Óƒü=žvã$ê]ÓoŸûRpq3föª¾ðÛœú&é¯/z‘çÅ,fÿ3ÉC,×ié¢XY1ßÒÄyaµ‚WÞ‘Óo˜¸± E{£@hl¼Ð¼W·À/àÍ<ófúǾ.~eTΆöŸë\ý œ£¢/Ì:ŽPòóîçt¡tNž^·;ƒÌ“RaÍ÷®Å@Dè¶RåEœL4ïÏ‘Ž÷Íå†cÊ ¨…ÚœžƒXžÇþåñ”‹ ìtFÁÞ6Ä Q‰ÉÙïã¬vÃÆà<óùÔi3?Ãeå¿R Éì#¢ª8Ú¦#…ö_*ûé§é¶3‘ξ÷é쯘Am&ýÑL {( Õ¬ÇQq¸Jý™õÆbçN½‰,/e€ŽEó¬×¨·'(˜ÿßMž@ü¯Å<@ð)’§~Ë]G!z­&÷©ØŸ¢ÿª’?+“ß…%¯(vì?éê·ýøžü§«Ô1Šý§ºŒ\çoýHý&}=$cð7|Ÿ‡ÎË‘: ©S”¼"íˇòŠÌ”÷Ãäi·1羉^$Í¢<7x9ɯè´W¸«¶n`ñϺ –^Wßœzª³çê<áÛ«Ù¿pë“=zhšÓyù³EzÆ^…7ñ>›UÚûÈøªUæ^Ã6͆ã±ã.À‡ß¯:„úiAÄuf— æ…B±ßN]‰}ÛUB®"2oqH›ÍsqRö1¢n–W1ú¸!8Ê~1'-í°bUú³v0âØ?1þwÝÀ ú§vH$¾ô¬:®áœW§ÍM€‹ZývÂåAªäY$ëöý‘LœDJºÖE*qèiýÛwíÓæHg]D:ë™3ÔéY•Ï"ãviõË S—UODæ U`«>7W|ó™Ë XtQ747ö©äë¼5¬aŽÑ¦þqê—øPœ&—~;67žFâ­R—+ûúë.¨—sÛw¹ÏmM=úÐy‡ì“$çÂL ý9‘ñ·I‚÷«q÷‰§LÏé—A¾›Êþ¢nÄI îl:Ï70犟'õH¬GXH|Ô"Çéš{Ú‹ù}—j΋Q]–Š·ÃjæéÖòs®ï{mìØÀ )Ä“§ÓÀ6¤¨S}Ú]@*ño©Ä禎 Úxvç[˜ß绣Û_lêߤÎÕSÚ˜:$‰_ŠþÈã6ü&8ê1±KEOþžÆÄOi7J~Ãàjø9LbÑWäç‡ð0¹p£¦/åŒÄ[¥ÏŽàc Ž”z_ò5æü7Ö[HÜÆœ—CMÎS”þ8Æ¥^˜H{YøZÎ2}ªxŽÔSIÞÄÔSI]¡ø{ìáÎ|õ’…/©on껾r姦~]Ìôi[«ºàçõŒsi+ºÁ/Ø ÚøÕ»Š:½‘^š-fÀ›}7?­€9]áC<öÚWÛYè÷”þa $€ùû@žÌ>ù!Œ„Rÿîf]Ä^ö%?Àº±0} èSÐÕ Þ4Šõ`‡Ù¯8Z·3…#Œ kõQñÏàã-±ù7ŠúöâôñMݯÍÀ×Ð@7XÙ¤K+W×ÜŽ¤ x>guÛã!¦¯ÔyÆ{.ìÐ ƒ‹ìr‰xžËÄ­]ÖmØŸÆÝNÕ ÉŵbG2Ï™L¦]š|E7¶AŠ*ç,ãgpÁ’“º„ÍKÕ¥Ì~ü³Äm$Α».8w½Sîº`áWá_±'…ŸÄÞþ7uôÛ$ž9”8-ƒ£–z%Æ[Ìù¡|Ñ{Œ³NdþOúŒN&¿Jé_8•|+õ‡n¬Ë›^J32fÐþœùmÚ—]çwÅlêí¹Ì«œ(qj‹xÞÙâ‚ú K,a¿‰eकôçÏ­%î#ñàÒ’w4ýû©_:\ü:©¯'Îà¡þ3ü\R×)öÿT¾Þy‡é¬'1çæ0ï0}‰è´ð{X¨NÃëvî”’_\ÊúdÓg˜ŸÃð3ýLãçÉùP*{Õ`66dø„´¼ˆ?0uR”‡†¯‰ÞÆ~‰R§è¯ªí\F€jû»r5Ò ¬Pˆ£·#„ò&4ÛQÄnžCºWÏóVöS>„)Aâ.„3¾‘ðG”†‰ÖÀaÖ9F³O\Œ†µæÇQÚÅÇômȇ“ÔG±ÄƱ¯]|RŒGFð$ð3<ÿøßÒ‡—üœÔW,å9]¢§$N"y Óêê›Ä•ü„Äiä*ü)ï'yɘs.$îÁ}`ø…q ÁW›¸§ÔMH¿né[#u¹ú»IÜÓ䥷Ä?¥NXìNÆ3ï*ýTG³Êhú¹¢ÇRïKŸUsž7q†/i'¹ñ~ûTÎïæûÍmÖîpÁÌ'~ÃôÕWjãö§X¢Ûë72øÚpÁ¥†Zâó‘W×NÒÃHÙ©5HÕe‹ï!x×tÇ9-’î Ú÷™ÄÛ\ RŸÒÈRèà+qq†uã ÜÔâ²n¿®é¸ÃüÒ]]ö‚éë)ü'y ù}Eÿ¿H\3·ž“ø¨ÄAsÇG+ÑΔ«ÄQ%Ï(|,ü»^ßàâ¸îCqÔ\ñÓŒJ_©käz’??Øô«Sô¬ôg}–ø›¦¸à{(7 þ€ñV“O”zfÉ' އùÍÑìË&zUâH“Yçhú€òJÎUE<üœ£º1æëcNêäÔ_ÈyôÛ–ju;Ë™/Y9»òÊö­æb5û¢¬e?sž0íâìgáõéï.ÇO5ýS}Y÷»•ûbÏ£÷SǨ¼ø5üù»ò|»`uÜïO»ªa‚•±G/_Ûœ‡ælþç_õÅ!ÝnÀÑ&^Q¬ÿ‹¡<ê©Ïâ„/¿Xö¥KÖ†7N+-Yösœ ×…òH"ð¿Ï…¹º —ȇWkÉ̤¨lleHUi›ëG‘Nÿ(C•e4XÌMú2\eñóW×uZf8nj´Â3¸­áìEqW·ƒm≦ï ÷ù"ög‘ý*|`ìOòWn=øOy|áGáOyžÄMsãMM¿±'%žÂÏcÎç¦>2ý×øùß&yGáÓ—Mê%?(çÉ0bð4rÎ…Ôë3ßúÐùÝÌçKßó‘ij=Ô7‡çŽã9]Òg2ùÛôÕç¹Ç¦~Ÿßsí9Ÿx¾>(:‹X7áξjKè7/ãù}’^Í×­;«äðÐnÏôüªœ[*ç02Nà;AîõÃVîÿí<‡ÉÔí«Sï»\Í9gF·ÿî€ÐUª‘Í—Øí¡”ÅÞ˜¦ß]ñ΀ÚÔÎ CûÊ™ó/T8@4¢TÛÓÑáˆV]¥\[âHWÆ1uÜFó8Á<ã©LÝqÌ3Æ_yë»§¯º"a¨>H‰ªûð ¯Ä|Ã9Õ6àÀ\ x‰¸Û+Œ%_SuTCІ UD*ÏeM#þ!]Ú #éçÙ?µŽLž¯‘•oË©ºËŸÅ5ž·xx·×~°¯N%ÜRm5ò?Àmú›wU¸é¼;î*s}[€éW/r^òesX7)üaò†Ô3¹ó„ÿM¾L~•«B¾<Æ:f©Ó?ÉúàØ¼úÀ_Ä/šÜl‹7âu;ïWqÚ‘fŽkˆ3쳓t@HqNJ?½8.ðœ‹JM§ß7çB]¦½¬®Žd­æJ %Y5y©Ì×§ÑÏLß©úÈÐ_ç5dê°ls\½Û°øÍå=pí)µqpqª>Êfü†[Ê;‰<‡[âb×­e­Ô9ˆ¾½$~\øb— ߉¾Z¨.ÿŸâ¦¹úzü‹è-9÷éŸô?—øu¦ß›ô)üqQ&n*¸rê!ÓwŸï7„y2Ó¯JúJ >FìSæÑMß úùÿ-ýJ¥ŽKúJɹ7ÒÿMúgŸå¼§÷t¼·1§™ñŠ%ªJªöH,c?ýü~FÏIÝñJ¸„åÁËHü†çÐl¢¾4¸­–žÃVâQ·kXv줟eð3¬¯ä9hA<.„uð»ˆóÜ£ÚêDFcŸ>^eê6¿-sú,ÞVŽfDò{E–Q–¢uع Ž8ÚÆÙ8¦ÚsßÚÌ Æ2ß§ázpÚuüªã‚‘ª y$²?\ÒÌ¿¾x¶8ÎñÜäóÚí ïÌV'Ôá2ãÊÉÔ³)ü¾©5iºMø$¤³Þ>C·W WÕ1oW!K…ÉúÅâÚu] *j5o¡Á9à¹|¢w¼Ï¿QâÒ¯Iâ#&‘ o*ü)ö¦ðسÂÇÂßRÏ(ü v›É—çÎäêSoü.Æåœ ‰Sо“}/øN9ïLöÿTÖQ¸ÞR >ÍéWÊï3ƒçb™óA•›?'s‚ïuûùò5s¾ï<ÆE0mÎù–~÷rî ùÍÔMH]Sè2…TÁJÚ!«Ø` ßßôKÌnøœ-½^x 5Œd:¼Ô)0í¯b“²Âkøc3q¾»»WŒmU [)_ »„nУÿnérε`¾,˜zÙäÙ¯~JZŠ}ÔÔ)lM ŒñÒ°cæ8.ñœÄpU?§"xß"¯é@îX5}áKÓÃsªŽO}ŒçéœpîŽ#JqRßÖ±Úmï€Xö9U ¨R>x¶| jûÞ¸€DöW<«Â@•Öáq*^=PÝuS .>ñ¢#_Æ%â^/»iCWô±ãE‘̺²ú-©<·-Mµ_½éì÷ŸA?:“çe]e)Ke]«ÄuÖcÜÐ0ü¸Éó„dÿˆ|—¸…í1éW!ü'~–øo¹ã6¹ë.„ïró¡èM‰óHž_øXüBYÏøq¹úK|ÓäG#õŠÒ-W¼§'¿ÄqLÿ5ê=Á ˜þP¹ëg”ófLþv—Ô+½§Û]V7}ÙÄÞÊx›+õ¨œwhâ2¢ÿoý¯ú¯xÒºc½°”}¿–Óþ4xšÏtaÖGåÁ:fO s,Ž=^<‡È›ý 6«2¤áKà«¢ÕÞÇâ[¶qÿíè¯5(üZUëðǽÒp”Ѿ¯îþ€æc‰ƒ æ÷ a\qëòv«®s·ää5x~vûŸ†3?A¿%ŠùÈÞ61/+àNŽðüÆc\ï$?W¬ÞÐkÏþ0§NPUšÝŽ ‘öBÏ«?ÇsFϳ?ÖE]æÑ(§ ãÆ)¥•#?©eµÀGZ†*2…ô> h?žÙ>S3µ/ Yº½s!\#ŸÝhRÒÑ|s\\òçuq)ñÀÅÅ¥¤3 8£ˆ3òÏ1d*çù†Œ!S×3åS‡ “?ç™’ý6•ÔÛ<ጼ ]Ô¿Rã³ÿþ²¼}¾ èð!nCjŽœ:d‚z‹¿²ŸÕ_^í<£ÐEg\rÆeg\qF²3Rœ‘êŒ4g¤g¿N†zý{òú·œÇ½œáíŒMÎØì gø:c‹3¶:c›3¶;c‡3v:ÃÏþÎpF 3‚œìŒg„:c—3v;c3ö:cŸ3ö;ã€3:#̇œîŒgD:#ʇíŒgqÆQgsÆqgœpÆIgœrF¬3âœïŒÓÎHpÆg$:#ÉgqÎçqáÿüþ¨û2@î‹Ü}þÝÐf¹˜0Ù™”pÆ“XÛsóܲïAì7ÏEëÔ:›6m`®ïúúØöÇÐûf×…·*äGïæ‘Gƒ†÷@ï6I]ßH-…~Ü?TÄýßj]2ïõÙèß1øÃ}›BÐ;dNçê#²ÐûÒ‰u?TÏ‹^—çý1ò'ôzþb¿Õ«üý•YC}øßù·‚..57{þçµþãoµÿc^ç?æuÿc^ï?æõÿcÞ€óò·ßyU÷%¾q‘òÍc-¦§ÿçcÿí5ÿ‡Ÿ­á¿ôÙþ?f¾,«mk÷£‘é.IAT :%DÅ»[Œma¡X؉AI HwHŠ¢¨ Š€"wr\‹»uŸsîï÷ì÷Ì5çsŒ1ÇŒ÷[ßvä߯5ûc,Y¬ÿ§ã0žÿçãðþ/ù3À8?ñ‡1žù߯Õû}½6,þÖfôëäÏ×€¥È@¾ÿy®þDµÖá‡cÂÈl„¡¡ùïÇæ!ö~ÓdÙûÿ§—?›E+ºÿß=¬¤DN]¥Åþóg„(.^0ïÇû·½Ÿ¥%S!{è‚ßi°dýRwß• „ÿg5JóüƒÌ«ÿÇDÿ»^¥À€?Ó#òK=l?3QL‰Ü8”*Ì]6@D-þãˆZü“ˆþÔR¥eA³Cÿ r¿ÔÍ£4=ØÆÜß«Qú½‰[¢?Üäø“¸†úOø¾áÿ\.æÍZB–/‡)°þuaÓ£°Sà àd}»~r÷Yß—”~ªŸà·¾ß.)”}LA´oš)ˆS  IAŠ@š@†@–`A_å ((RP¢ L BA•@ ï²¢AA“@‹@›@‡‚.AŸC)èômeÆ&ÆŒ `Ja$9…Q,FX²þuÛýÆXØP°%K0ŽÂx;‚ &L¢0™‚=#g ®nî< ¼¦xøøøL¥0‚?Át‚3 (R˜EDa6…9s —`Í£0Ÿ`…¾‹+á¬ÅBB –,%XFÐGZÂVÊÃZEa5…5k Ö¬'Ø@°‘`Áf‚-¶„Dl#ØN°ƒ`'Á.‚Ý{öDì# L„E˜‹0a",ÂDX„‰°ŽP l„EØ‹°a#¬NR ¬„uš€°a%,ÂJX1ÎQ8O0a(,ÂPX„¡°Ca†Â" …E ‹0a(,ÂPX„¡°Ca†Â" …E ‹0a(,ÂPX)–Â",…•@!‘€0V2…ÇcaÆÂJ£N0a.,Â\X„¹°saæÂ"Ì…E˜ ‹0a.,Â\X…ún¬Å%¥eå** ª(T<#¨!¨%¨#¨'h h$xNÐDð‚à%A3…W¯ Z(¼¡ð–à…¾ ‚0µo}÷±ˆ\&D‰. rWŸF¨"á|„Ï !‹V–l´’äÞ-J.4âÄAòLð±ƒÜ¹É][¬–ÐWr·æ;ý­,×÷Í=±Y\xÔû8á’ò„Óñ»,?ášä¾®B¸Ý ’"äò!Nîèl„¿ò’;º±EœèU"<ƒ­ŒK8\è$wg-Âetob#œH³ï- áÍBä+B. ¢än+6õ[|DˆJ$á¹ G \E8‘PkÂÑHœäI5µ“‰#qà&ºDì¾#EóÁä —#©ù&wx1©Ù·“yO|í ã8rýõ-.bäP•"|’$‰±[Œð2Yr€Ê“MD…p±“¤=‰¡9,ÅÉ< (Dr@ˆpjE2wÜäþ/GDEÂý¤#}xIœ‰"ëH_2/Ò$×äˆ?B}q$‡œáü„ |aý‹³i’X ¾6ˆpTv’SÊdÃáèkCx¼"9´I¼Õ=¿ÅFQžØFrM‘ð!޾ïúÞ<þ.@ø/‰õ2çb„ !üKJ‡øBø‹"ñ‡ƒÌ› ‰ó ’›œ$?8ILÕHÎád̾ïH~Š’ü#<‘Wò[Œ¸É¼Ê¥DæC©o<Âa‘|!¼t9”H»¡Djû ×T$qTôø–—}sÐç‹(ᦊDŸ±i0Y[Ò„Ÿ)ß¹HLI¾H‘y5ü¶ÎÈ…šŸÄ˜—p]a2¿ ö¤/ñ‰ÄIˆÄU‘̱U„Ø.Þ—û$Nò}œW‰´%p‘9%~?¤É:“$¹ÍGÖ,™WÂåHNÈ‘~|$ÿ ú¶/È‘±¸È5„ì) „;Š÷=#¾ ‘>r„gò¾/;—Ô“yâ$¶qäI.Êõé$ëCЬ9>Â}ÅI^©¯ù¶¦ÇLì&Ïeú¾Kø?¤¿3üß’ÌÏÿöøÿ¯åŸúý»vªçÿ–}ÿ-;þ´ÿïÚ TÿŸÚ÷¿×ßÙϬÿÿůÿöºþwûÿ®ßÿk½Ìv}þÛqû§zÿS;þÓ¼`¶¨üOõ ÔþOõüÓ|øo÷ïæá{üÿí|øwýùÓvÿ­üü}{úý!ý’‘ýßÛOt0Ÿý¬ÍŸêû™ÞŸµÿÕ³?õo ;ÿÄþßù«ò¯ôÿ;þÿ©_ÿdÿÔ¦’Kÿ$/þ©îjÿcÛ¿“ßÿ©}š›ÿ‰MÿIžÿ§~ ”[Ôß}?ŠP§=Ñ÷¼ï·  /˜µ8 $dö‚ùÔ#®™þ¡AüÚ‚;ÐF肾·Ü}?µ w¾7öô›ü>fïÈúö†¿ïíýxª¾ï­½9õwß[þ¾·ùc¨¶£) §0ê×ú—t(ØRzúdß+±Ôó ”Î> »¿îß§Ó’²~6‰jÓ'é_1XPºéòʯÑÔöÔ³¾_/èÿÍ^ó¿¡ï›úW´6TŸ¾¿ûbØ÷Ú΂ ÕÏ–‚55Þ0 }mÆý­lñ·±èqFR±°ÿ[]Ÿ_ÆÃdÊ7ZO_[]Êþ‘¬¨¾ôÑú'Re¤ú©PЧ0޲Ëâoójñ·9¡ŸyüÍžáÔXô¯PÌ©XÐvŒ¥bÞ#J£1 ÿï¾|q£þ¶¡Æ¢çÇ„ªS¡lU¥ÆœB鱤ü3¦bCç¤Õ~2;Ö÷¿’éóÍ™êOƒŽýk˜¾WqÚ”¿tîÑ>›P0¦|Öÿ¬jÌ”c)iÌúŸõfAÙÙ7®;:æ}¿ÒûÛ¼RõtüÇR0û[üéüAT\L©¸èSv¢¤ÕÆ‚%«IÔßcÿæ£å#=ÎD*&γÏMªÎ’õ?¿2¢…Ô˜&¬ÿÉú9ýë(Mªãïnpß%ÚIŸìþ̶¿»•ü;}~5î¿cóïüþU|~¥ÿw:Ù~¡s ùù™îßùÿOæôgºòëw§ôïúüì4ÿ•î_ÅëOæôŸ<ÿÏ¿ŠñŸæú:gZÿ»û™ÝÍçïæâ?±›¡ó§78ξŸ,³èÛYÀ·ÒÜÝþõ¶À’;ð_KÞì¾ÏÓ~Év¸ïsäêYÿúXò˜üëÓ_¦ÛóQí©Y QeUÏE•évt?JŸS/%™íiICKÚÎ$S/­Gˆa7]/BI~ª?-éñ)ɬ§û SeZÒþÑãñ3êiÉ9€^æ¸t\8ˆ==L?ÐO=ü­ßÚýøç4-ÿØ>†´ìŸGF|h;)=ŽKõû¡žÖKç%ÓÚïAßó%3ÿ8)?èü¦%s0çƒîÏ3@œ1ü`æ ­§ßƺ¤¥Ãæúcγ=;5-i»™ëй~™ùͬ§%£39ñ(˜ë–žöïõöÇ—¶›ƒ²ƒcû˜ë›¹Ðý(;ñ+ÕŽÎæºfêaæÓ:>¢ßÏí~aä!3^Ì|¥íåaÄŸÑ®?N´Œ8þg¿Ësæ¾Ã¬ÿSÉœæsZ2ó“wº#LÿØwú×­ÿ‡}±o0ãÅ÷ýZ2í`ÖÓã0Ÿ·|ï÷oýÈ~F|ûÛ3ìê¯o žþyÜúÛuýüù€v2ç…™?tÞ|`ØßÉhÇðû‡q©výç3?˜qbƽéûuñCÆü ï~;;~0û1âÿCü˜y3@\ûûÑóƈÏö 0ÿýí™v2ó†—öF¼ZÌyè­%ÓŽ¶Ÿû?P>0çsÀöL{þ,>Öwþ|üÖÅoöÇâÐBŸS ýŒ¼êχßûóþÝñóù(NI&ÿhbž›LûûËtÞ0÷‡ßœ?Äãwõ­gZ?sݰ¸ßþ&ÿtÿ]è{tÿöUf>1Ï9æ¾Ã´g }q€ýá·ãÐ’é?Ý~ û™ÏŠs^˜û1ÿ¿›ÚÏöf¼XW?´£$‹þ>Kó›–:Ô¼ÒRƒº×éSeUJÒí©ÿéÍÒ€’ZßôZ¥Ú©ßR›Ò£M=§õ ¥ûQõ´Ô úÑ’GêGË¡T=-µ©v´Ô¢Úi1Æ¥íJÛKÇЯK=פÚÓ’ö–ºÔsZöÛCGÇÑži/-Õ(»èv:ŒþL=êŒøÒ’žm†´t¼Õhûéyýþ^Ò¿Òúúç™¶“ÒOýO«ý’ÚÚZÒÏu)}:LILJ‘ý’ÎC:¨²ã¹áºoe¦ìÏO†]LIÇÎÚï! ©ñýúïtÜéü§ýÖeĵßof|è8ÓþSϙ냩·Þèù¢ó‰nG÷£¤õœ–ĉ9Nÿ|ÓóËØú× íSÝÞ—’Îw-º?¥ŸžZ_=†ÿJÄ‹¹~èþ:t3ôj3ÆW¥êi©Åúôz¥ãKÛGÇ•’ýû,¥×Ž ÝŸêGûÁô—ޝ/t™Ž­Ÿ^/Œ¼¦ý¡í¦ý¤÷c:ýq¡ó˜’ª”ì×OÏ?Ã>¦dî/t¼™û ']†ý높´Ÿýû %þÑãé1Ö'í¯#ú÷1z\†TcæCÒóIÇ»?.”dÆ^wýû!Ãfœ´ñ¦×3?æþ¢Çˆ½Þ5뛹>èüíߟûƒüëJŸžO:NŒ}Ž¿} i{é}ϱ_Ê|?‡þó‰1~ÿyJûËØ™ë˜žGz=Ó’¶[ï{É¢ïYߺ±èï¡ûåF™¾Ç1ëéçÔ}²ÿûNêûÇþ{&Ýï+C]îÿÞÒCÛÅìÏŸ–4ÏûdøÕÿ½8Ãî/ ûiɰ÷‡û1ÝŽ9n×7ýýãRß+þ jßvF¼˜’ÖÏü>®oû~\¦þþþ´Ý´Ìø2Æ¥¿w§ïó?Ì“ÿt}ïoÿ÷álŒ¸0ÞW1óä‡qüˆùï‡xu1ìhžó×ï/ß~»Û™q`F<ûígÎ?c½õ¿ßè_W û¨ø2ß7ü¦ÝÌuÍô—©o ¼fúOljٞßÖ¿?ýïe’k€8Pú~à¡Ì|(¿™ö~¯·¿?í/#Ÿ™ëü»é|fögö£ãÎΈ s_`¼·îßéø3âÒ_O÷gcŒOë§¿?¡ýæaÖSÏ9ˆgô:¦ó—^—ô¸íWíÌ|bîLÉ\pf~_ÁØ?ûû0Ïýý˜û)-™Ï™q`|_óÃ÷øg?ì/ì3Ìñ:¿·û‡÷éÌþŒý»¾™ùÀ‡!Ø×è¸Òú™ûWÚŽ?ŸßÎ :Ÿ™í˜ùÀ´‹™·´¤ý¤íigèùáû¬ïþQì¾}ð_ÿâãì™Ô_\!¡þ‹C©gHè‚…Ôßܤ"tIÝ,`iÀ|ºÙÏÿmÿàétë‹üû[Í¢ŸsÏ Y8{nÀÏþ‰íÞÞ¯EÿäŸØf#­8ÈÒl:îÑ5ó†Ûô‹us!kˆ&IJc"MÊpÐ=(ÝÓïJÚ߆t„ÚãâU{¶‡RÙâÙ%!ÇñðJÈ»gjÓ2 žÖ:h Ý YžÉî·?ÆÁ“õ•IªŸn@†÷žm‰9 Cé¶ÆëÈ8,ab³ž|Ú¿&í&d¯•ò^ðRrßTU< ܉›%Ù¡È|Ã\YÁJ(™tæÝ*G}(³š3ê^õ¨a;uZÉF¨ådïŒ6xÆþø¾ÜÊ^¨Yÿ¡ìïL¨µ=]í´jw{Èä$TCm¦½_§ÞY¨çtXUu+Mó®O‘…zá úµÏ‹¡þZõRùw· aáMî ó¡Å•^N‚çfU³c… ¡ÉÙøLC—$¼¢vý¼|åûÊj„0¼š-¾¶Ð"1;Ì6Þ¼ç¸Nw*´²Mز­ڔ߮·W8ïg9\Ü3|´g.˜f±`|tN”«c_Ÿxâ®Í¼ õÜ縎…åÛO{@w“Ìʵ\Ðë%"6l§²Z´OK Bö1•÷·Ù ÇÇ6‚k<ËÎÝbÚºHäa»Ö⟆¼e+¯w{!ÿÓ"ŸÍì|(øüÈÑ ocpœÉ¢÷TPä.Knƒ¦ns»ÚÁB1ËO»Ôx%QüÜ‹™'g«¡¤Ž¨ö(¡q(õ$".GevŒx"5«‡,—p[=Så÷™¶¸ÄÜ@ÅçwLjì•Be³Ï«$^z ÊÕ¡3˶¢Úˆ¿>9p%£úÇŒ…—Ô§ æ k³²ÌϨ3(‹×þA;ê-ðÉ\ñ  xW\ö·BÃ7<ÙbCåÑx±GòýsÅ8LâÚþÕ¼px©×ÛâÓcÐ4.ù„Ê\k4‹ï¼Q»— G½”Ÿ$õö ˆ.ÝwéhéšgT~A­Ž¯¶â¸øÇ|š4—Úø†}Øn놶¹GäËïžÂq“¬bî-yŒãóž„Êã¿úT·Ôý8±m‘‹…Úuœ¼ó¡£Sð|è©8,¨õÚjô ’8ZV6§<©WøòÕ}àjÙÞkŸÑ÷žÉs—WãT·w'ã´q½¾’Ä0‚#gÅ’Üè>aCì¶»ûÐP‚ÃÏøÚ"ä·Ð¿4Mg´¨Û'ÕE@•P¼®Ä ÅPØ{,"ÝÉò$·dœöˆ¬}'J‡‡ô‹_ ¥FðAÊå–À­Û ¹T+Ç}ƒ$-•[4üî0He»ožÒéþrò1'ÞÞÞ…ýíÓŸ—7ßøÒY9«×¸hAîÝI—ϺCþ+^8fòRù6CÑ’ Ãɷ졸¤ˆïݬ ýû½Nˇʼ5?ûÊ?{œv6f‡ŠX夸ãç rE ‡T8 ª5:{Õ÷@õŽg{ß÷³ƒªµÒvjPS1sIƲ÷ýûMý—¶û§C½ü¬7Ç88 Ï³7÷ 4|tI M¹ _ËÃ:]f@“”^á6=€“EU¯ž˜/÷féá,xÅ^?®ù¼Þ{”c«]¼ñÞú¤&ÛÞ^z@gÇ:h½É¯7t÷`h«7n9ßíš1ÂÎËÒáÃÆ¥'êÂÇ‘|“³òði§Ô¿ìÐ5fVµ|–¾Vzäu+ª–@¯Ó©žG<È*ôO¬†ìZ^)B º‘cï†ZæVÈ%wàÂ*Wä~rÕJÄHy]<yJù£´= {P0ý˜Jü~.¤¹,ìÍ2=©õç¹² Evš.míA±fŽYQ 5(qã¾È;Ϋ(ug{Ý]@™ÎT6Ög”Û¸mÿ²%¹¨èµc²w"*9?ÁvëTåjQJŒ2GµÏJ·Š¹Ï¢¦Š¬¯úœ9¨½ênyòéû¨'øiwÇ„U¨_ÔåÞ$††¯Ößl¿ŒÆá î}qظ7Ëw+ã­š¼gÛp¤~ÿöuŠhîãþ6^-®pÙ¼+CT]sXMa?ZVÙ’+.ˆcÌ&)êŽEëè”ðû碭œÁ¾î=¹8öT¬¿VrŽ79¤«³þ8Úå{RíVˆ™LòÀÉ’§Ÿ|äDûD¾£‰ëÑqø×tÃJtئüXEêå÷‡æíGç¸èhÛÌ è²Ÿ?&FK]CW/_WfnS7ž\…îÞÚMÔ? ¸v ó9Œž§×5u*z¡WÏã̉ê‰è.&u×%}Ç[Á58ÕFbÌŽO³q÷N]ô²[]bo?^Îét…†.Hjî·Âmeyý æüVeëg P"«4ÿÉMȯìÚÈ—d ¹_ùýWŽm‡ì+V/ò/f@&Œ]¶Ädd„ ò±KgCº_¸ÿyŸ©®½û[ë.H?«¹µêâxšÃÖ. Y8¶Ì¼;rÞé=´øäoÙ=nS^Sž>>©Šï,-íp? ¥ÇVeÄI,†òÍösçÉCåÆI Cc¡Z÷ŠÖñërðì°îuiÓ¡fO決h¨å¿¯‘°j-*‚¿TÔBíê{ž‰o–BmõöZ͵P¤ònåÖ<¨gÿ¨$P&õqyñú®Ð°aæHñNи$Í=km.<×ç}{x"4ݹ,¿gúLx)Ú³TvN34ïšqmñkexmzñ‚AìGx#\µ}oá1x§Xæ2Z*Z¯o½ì?ÚÚǵ½”íî !><;TÞdâ‘Üüm sNJT³y:t5}¶`;_v~ÙÀ5Z¾Ö8DÖ ¸#ëŠB‚ÏDvdçØõQèÐväX&Ë*@.¾¨´ÍÈlãù`Íä="ÝÅ6ˆù÷¬xRj‰‚ç›_áËBáZ‡çÔ> ÈaÅY_Åpp!Ï)g™S®ðakDñ­]®Ází(Éç4Âèú”:ÇÆ•÷Þe²6û¯Ä!“²c}ØüQ~ziÄѨxí@}Z×T–ŠÖžú?ªl°°éö„jƒ¸×šqD£zzö'Oñ&Ô¼±*wU³jª‰MŒD=m³G¡íGý»¢—KtÐ0~ßàczh‡z$Þìl-9øÃT"}”£!ñŠ”Æã*/HJ޾œ¾ÀßÉ +5ë‚´„3m<«\ 3d¼gŽFÈÑëàx y·÷­[[p ØÇ×䎅³#d,ì¢;.ûlG‚â—gÊ|s¡¤'⢅9?”VÏ3jOÚe§óbžj@ùäGå¼c¡BLnMUB Tt¾øzL7TIfÛú¶@µ¬úí®¿Ìà™Û–å;™PsÝÇÅ0ä*Ôn9îþv,Ô9‹ _õ×`¨q´)s˜4(}¹[s 4ªšL^9–ž›³?ŽhMÁÍç¥kîËkr7.î™Í Æ+«o«Kþ%s³§@ËÂÜ øòÞä¼qXó ZGÎwòŸ>ÚfH;~–ïψ†ž2„öîsƒ§–‚óÂG$>_ =g°„ü ódûú‚É·¡{‘¦ÎÕƒwàKÐç­9{Êàkä[±â¨-È Ýokü¹ÙT¾‡Š7vÜÊ{‹œgn–æÏGîIâÉ/[òWÚ8ç³`#ò¼uWl š,*ˆ_ý…·…n3æ¾"Û}͸7[áàÖ} õF[Α3¼% ß;%¼¥ÌÙë> ½Œ2¡çvXÈ{âŽcƒ*ý"P!y›NõÓ­¨Ô6öMe›=ª\a©xev£ÚQŽéÞ£Pãî«HίP›g¶ƒë‡8ÔÝ¡».êïµ/ܫކÇ?±— ¶ ±©­Æëfhò.ý’ïö8<óaÞ”“uhš9‹Oü54k]ùZôñ8´>ˆÏPá¢Û²ã*Ñhy|}Ï0žhÕ%¹ûuz8ZÏë/!65:šiÛqlP‹àlû­8žõbÉ=éûhw¶¤#Äq¢S󡝯8™ûà••#G¡}òäÏ:šèhÅk¤xË”?œ-¬Îœ'‰ì @çæÂp•°MèRðÐÜÓ¡]“¢fŸ?ÈnqÅ·ýª¢{âŒËQþ®èQµOœ„^³^O»SB$owî?†Þ_Ò£dU¡o÷.þ856öêø?Ük·øÝÅ8ôÝ–ñá#ßf4JÍÍŒ\° 7vŒµæ†–±ï‹Jn†gâ׬ XvC)’6äËÜ ö99çO»ß–¬UÃαױÃÕÌCÃùAúB ÇZ³çæ¿näîWf<ÇAzÞSH»Yß½EädðáfIñ¹éÝu¯-Að´E/Á#ÍòÙ¥rVî¼…ê‰^Ë–~…b¥ñ‰ }å¡äÓn·¸É+ ¬èéè¥w âÑ«jÓÙ³ ÊõÐ_JMfP}múÕMÓÈ=Óá¯Ð÷ä\´HéYc5~٢á&§,Em…!Ôb­ÔɰsP›t욤v&Ô¹ >éJ †zNîËfPŸzËI÷¼;4\¿fýª-㵃»ÌÖÁóç(¹,Y ^ŒŠbk?‘/o„‡®]þ^yf½lW€ýEiÑ++à­E€æ<™Ðjñj—àkh[;zûgqx_o¹çŒÐø0gÑÁ×3µ¡CÎyfzQ|jß°?«bˆ9ƒ£GF…~êF¬»žºâÌ1´2–œæn³ošBc©Zwµ|ÒQ AÛ`óùOvãØW®z‡$§áøE³ÓDàsŸRµ—8ñÌb¡ 'œìR»O3¸ŽÈH§J ãt?³%ùrè46Ðõ/G't6ð³ºß>]Ålg@Wi4xÜ2ÝäU6ZÞ%ç¡á– …5èáúTKÓ;=w´Î*Ê F¯JÛÁ¯Ð÷ò-y|}n;îñÙ‰~FG6çé§JëâØÇà´Ýº<7LÇ­G–Gý`ŠnÓîñõ¹†z Óô…‘çä K/³6xù¥3±eÊ((wÌœ˜zz ›©7» žŽº¨Ú(O¤§oÕ0‡ÔÃsFç6Œ€äúÞב5W É`ÏÙ«OŽ@BÃÅÚ·o&C¼^•ƒ¶ã ![õÞöH|ÙἫF’níÔx©Dú/èôþ|Ÿè3™n7 ²VÂtÛ È½”Õ¶ATò…>¿-_)¾ms†ÆBáB¹£¼Eä\ û\/Šã.H'¼%;—ÄVKBé»K–úvPúÎ:t…ôu(»TsêÓ%(ß¿®tÚL¨8×™s¤*kn¯xîUyç­ø Ï4v½Á ¨Yµvyù;'r.Þ[í««u3ž=MI¹õv‚"r aÜN¥¯#o@£OÙ͈Xx¾#Ioƒ’4åÌ7“ˆ{/Í6\Í_ú šSµ.8Ú¿×[¤¸=-oWò¼V…w+ϼљ+­l"Ò‡á½ÒšŠwMо¨t}•åDøPÝ{Ì^=:|{Ÿúë>|ê]«±{¿?t¥ŽÒ,¹2>ßÊ–8ÝzŠÌµßšH"kÐA­SM%ÈæX°Ñìórd¿\ÂjÖQäÔ+àVÙ\9+l’ίBž}÷;f,ˆ|›ZEQÆ NŒÛ?+ …Þ,:²ôàHá5ç‹7Žœ§V<%Å$O| —DñÌ—1G„¡ä£°ú§†r(ݺMjY–™}y;޳FÝpÞ3ûíQéÄ󽋽QÅ]\õ}çT3oTÏ:€“ï§5$ÍE­HÙ”cï¡®€ãMž^úˆ“åhaІ.ÁNñþÁhô<³£îÑ.49Ä~º.\‡Î-¹Ëަü—†öX¢Ùf§µ#2Öã¨'üåŠ1ãŒôÜ¥5n åÔÙ!‡>{ Õ½:Vìƒ,´VËV$mN«=èø2ÇêÞ½PaŠãrF$Ú^-A»•Ò7\ºáDãÃóÞ á¤Žm»ÓV.Fû”H)•gOÑqÒR¡å¦[ÐiäTmán#tÖT>;iºÈr÷jMC×Áórø—D7áGSò\î¡»¸f–/÷yôPzün??zÚu¶ çö¢W¸ý»cEƒqÊËÙ§b8–£Ï¢U*«ÞOB?í'KÒ‹£pÿiav ¸3+4ý¶Nút³§­7@à Í]‘;§ sjþÁc"ðjÍ›Ëþj±PµÞª-b} Ézí<±ô8äyÌs»ºó @ë åaûÙÐÖ|Û_VK plœï‚­MWq¼Í³wƒf ]e Ö_E{qâê÷sÖ¬=…“ ŸhÝâš…ö­Ér‚ î £Ÿæ¸¡üëÐÉÎ"9´RÍnhJ[?B£%òk™¡«‰}ؾeÑèfY»}Þ+rúŒÆÑ J豕?2Âû2z>Ι5¶zNÿôYéîxô^´æ†ÆLôi=ÙböåúÍ·:œå€SËöœ{„Ón$å†m؆[ž?½÷8Ô]_Ü=c€¨;fÈ ÑmävþÀ_œ” /‚ž&.Š·‚Ò÷Ž>ò-2oŸÞó@ì4d%wÎüºÒ+Æœ7œ ){N´ë™eCROÄéˆ(HôNÙë3o ÄOQk›šã”BLJx ÕÄÉã!qÐøOîB”qèäÒƒã!%ļ÷ÚJ²>>j.¯‚,îÓ[óŽ‘ûêÂzÎ5m's¥§áfä;e )|5 –8Ÿm-? …›¬oGèCÑ’Œ­cÚ'@1ˆÅtž‡â¶ä«wõæAÉž¢É¥Åw Ô&¦V¸Z«HÂ'?þ¬I•BÐ%ã Ö“Z ŸYGên˜œ€‰-–Ü1Ðëf(­틬çÂõyV#‘ÝBôúÓ¯ã‘ãVë°%QBÈåX“vd¦ò ²–Ïj<€¼… Ò=P@@Þ8Þé 9οø@=%_2k 8„ƒMª‡lÚ}Ec÷ZèW¡¸{oÛ‚½ˆ’–­ 1ÆÛQ:8øôͰ([•ÀmЃòû77Íõ–@%ûMÃySÞ¡òW-õ+Qµ"ãQÐê]¨Þ°ísrK jÉe›‡º‹:ÛVML¡F¼×³_Ø¡!OÝÉèÛ-h´ß1­ôMšXŽ>°é^šëüT MÙbr{G壙¬¦Ø”§q”ûÎèrkpô=Ža‘:Ih©äØÜó¶­–_Z4y>A€GiÚ(:|4æDÛOs§¿ hÂqG·ðŽmC;G']ïâD1» }5Í8©iÍ¢ÁñÐ>­ÄdÈùtôhhλ_‰NöObo5™¡³]ô‰ñü¼è2ÁÄ©²tº:*ú?<1ݼ«'vž‡îóÃê<ôÑc[i…UZzÞ]³Šïz}ÞþN>zû>Ü鲨}š ÏÕ$mC¿#;Vm;‹ÓàMÊ4v5Üñê¤ÜŠ‹’è3ÛbÿÍÎr4Øÿõ¬º»?ò9D²¯û(/ï¼þëšqT¼9áë2 e mLŸŠC®ŽýÎ)ÕS`î°w&¹OʪL^io Êo¼á„T¿cµÇVICʬMS¹6@Šâ2Û I’ïUSV=Òd_m÷Ñ „Œm<Öj!«IoñŽàÈ}ït¯‘? D¾Ž+ªS‚Âö–½3çBñå[ âm‡Ò·!Û ›¡ÜDSfÁügP3/TÝ1*Ÿåál–7Tò 5«…ª®,wßUéP­Õ£/‘¥ÕKÎ7Ͼwžñ‰©Œ…’}%ÁK‡Ç@ÞlÝ\A×¹ä:öîYé*H;¶üMùeOx¼µø|øž$-¶ð¸ÉvHÔÚ-ª\+åj›ÀÀg» ð0VxWÕ_‘?×tä¨Õ嘿êfšÍx<™›‡³;ÒüÏ¿\4Çž,«èíÙC% ¶¼Î†œ )2•åªgûº­ü仌½èÎbªíÏ‹ ÐèÅÇ4‡WPø6°GQÜŠº¥¯³Š4‡G>€âjèЛå%×Ýy¯5ƒÒ[¦^e¿ e1º¦QëW@yZËã¡!®P)òAÙè¯òæ.ºý ªkEe-b–¾øÄ4$³ j‡º>³o„º=–œ lPzÓ¾=)hÈ^8Üõ<lÛ"èM U¥âXgàEçWQh¾cí³[^=¸f¥]¼Ù;|}×Ýð.ëÜûY"Ë¡MÆãÜN7.x?mäôb‘¹ÐÿvVPÌ:ø¨?dRcìbèxxp!¿rtΛ¡¢V=º'ÍÙu†‹¾¸¸+ú€¯ÛMžõ† BVØÄ1[>![昔)û;c´ðôŒÛ¾ÈYh;>D±¹·eÙsœ‡¼Áwž:¨/Aþ¥ïëãGíEÁ›sï÷Þ¶ÁAÒÑQyb*(R¹½…­ƒ[÷u-A/µ¯fëœ2_®Ó*<½ …±ê ÑwÊó¡cödàT‡“ç>Äi!f+•Ç|Y‹,µSÑ»léÌ} 7áÐÃW†ªG #Ïm?“¹öªÐô¶y}¯e'”{}IÊU{²ÜÝyªðôl²ãÅ[úy~ÎPËÞHol<¸ñ¤¹w/Ö)·§Z+¢ŒáñœYÆî 'u x?ßk%œ7XïƒTÕDᵆE¾ó–±i{d6èMÒH´‚œæéiCþ–(ä8h …¾ñaÏn„bÞ´/q÷AIòaþé„'ÎK5ݾè0”çzÔŸÛ¼*:xdâ½ 2æÙôCj'¡²5ò„Ð*~¨–˜ ¨yªìD¾z|†ª§qœÂN ú˜XC¶b(<;xn÷–vÔslTF Ô̰—õÓ†–‘U€†…Ù†ÖCãí•gׂ+4)t$žUÙ /œç³;Í;–•…FÂëHèÅð&óÉ×rrïTUñ²é ´mîöÂ~Ð.ôùiwÇøðÀ4Ägctì|r/:7Gqõ¬)…î‹îjz×ÎÉ{qBïܦ ŠUÈÆk0Å+?Ù—o0ÒáODNþï¤Ã[‘+%’ßxÕyä9¹KbHå3ä;*!f §€w4¥:–NC¡Vž—®e€"ÊǼI ÂÁ^¹ãÓÎ?BѳûÂŽ7¢8¯•ØQ }”Ø2yŽ õb”’{fºêÄ.”ÎVXºdMÊž¸ÈS²ÔåN¹}} Ú„ U"'MB¥]V϶8• r÷ÉhG'CT]˜x¼ö³ªª(GšBþî¨UnÅV(Œº"—#|x ß‹‹ìò;ªpšÉδÇa^8­ëí*póªeõ“Bס+;¯øÔ"ÔüRäV)z9î.;‘“ –ËVm™ÅÃ}V嬹<Š–tÈ}2ÿÛhKrÿ¼j²¡õ$¯ðmlê‚Dã(Õê0xä™?w¶

ô±Ý2":†¨ˆ§§¤eŠ?² ëù¨%ðùB°i{ô$™ž­¶OB–ÐÖS_²ï ›çâ’ºÑÈžª·wÈÓËÈév1sbåäæv >öÐyjè”êîC¾úŒÈxõP” ʲ±ÉDáùÇØ$Æ¢ÈÒG¼õ]ëqpù‘ÐÝk‚PlþVÙÃ'Pbܳ瞦A(5Ó(N§xÊ$×õÌe[ˆrëOu¾ŒFE s”PÙÞQ+~ýT•Ý%½dÙ²>uÖ[· ¦±úq‹Û¨^ÒÙªfƒz2¢âi£ç£~Ý¡UUÊѰÍ'0Å~'ï¹1oõ!æl˜3¡+GŒ¬‘9uï=Ž´.uºÌÍí3Áû-R¯µm1®E´ÁË»c÷ e§¥ó»C‘8&pò]GÝ,´~-¾¬-j5Ún~7wµ-Ž¡vâímß<ˆË]'o”I:“ÒMÎÛJ£}–ß\¡Qñè8?áôú¤-è´òø-V¼:ï:³vóEt¹¸oÈaïÑõé†%ÂGÝ:äõÇí(EU!û&sôt›Äݤ§ƒ^‘Bî«¥qJÇ—} è3~úÌ+Ž‚è›{øÆÙiÜ8u•è6‘Y8íÜf£!×¥q»VG×…PôwmØâ—ú¨{Ü÷¶úÓûÈ=Dg¦Û9=x®ñ"ØÞAJÓ¦h·uL‚|™êëAa½­{ýI¨9Û*»Å‘Ï÷Áq6w[:ShïºÙkbi;E8Ô;)ŠƒG­7ŽÎ^…¢[Æû{DT¢ØKŽÀÞ”رŒýE=Jv»m|±­¥ÏÞ4~°qʆúûÝ;‚r ?ãà‹G…˜wÜœm¨4·y{+•Ëâî?Ͻ‡ªv™]ü¨V×pè^òIÔ¸xìÁõµ³Q+ºn’gi,êdn[åÖ`‡CUŒ.G/CƒØƒ‘‹ÛÑhø„mzìh|I1YëL#Ñ|aç’r^·‹‘§=šÞhÛ1®žÍnμÈ!™„£šS׭鞊àTÁg§‰–‹ÖLVÊA«¬”òü J´6.ø2«þ Úœ])ra;ŽÕéáqÇÝ)º8‡k Úy4o/oØùCÖ'VìÅI…%l’ÇýÑþ†¹Ú‹‡ÑÑiÃBç;±èä”((´Ò=/—{‚.AÕëqs£ë¦U첞è+ýq#9küwÜi@Oùäü«ÞNèÅo‡S’*¦œ>:oŸÍ²Ÿ‹¾§÷öÖoÇ©jÞ6A.*8ÍuþŒ€l)ô—ßdUµ‘ 7¥gùp]vª$?ŠyS Ï%Ì^ˆì³ãæLô²„úØI l¯œ¡°¤~§FF"<ÕÔà{©Ò=&O>)JÏÃô†-…¤¥VbV£÷@ü‘ɯ]„sáAbâù”kK ®`WõÇ×ÒçóBVxò$ˆÛÞ ¡§8×gyv‡Gi·²¬V@ë˜Äå.wx| ¢Ömw‚4·Î©U ë!ãÊÔ›Ûɹ’¶Çó¡!dûxÊ´ÌÍÓÿʃÜÊ™ì\wQ£à)Ÿ.ä;mÚ;q*ä¿úHN£ –l)oLƒBáK¶wJÙ‰oöl-†¢Œ…i»çþxE¿ašü(ydØ\ÓÆ¥Ÿ½ÜV‡´B¹Ù¶3¹«æBÅ›fñ±PeTTè'hÕM|æšPsýå3qî“PûÒ0$<–êASù%4Œ»åœì÷ c¯=¹m*4éY¨÷/*‡¯“m]ÍÁ×ÛUÏÁë˜4É'œ5ð&wRÁåwñÐ*jªóø ´Ù¶~Z»ÞŸ¨6Þ¨ií_£@3I>†]ŽÛtU>Il— Ü|/·žË„îr²mDxÀ—O£‚¨…@¯u8‡†Ú+d5nIvDöÉæ?ïAŽ2‘Ijé ×úÊý\ZòÈãº:|ªÑuä³?>&~Y·•áEŽ(”²?àÜ|S‘ÖëÔ'÷Ø…Š" yQô³m~«|.Šg¹5™•¢d­¸MàÆ%(3¼}yÎ+’»vmÂÕ9¨p+q&¢RÇL«QåþÂÔÏ–¢Ú™ÏG¦-˜‡kڼçÂ7r£îNÉ´1/סþ„²çÉyx¢~Äý5õh<òíº¶RhòAèiú ^´Úˆ¦eçrÓªƒÑ¬W’-l­:ZLÚšÇñ!Møˆ×¾J´¼ÇÿøÚŽb£‘»3ôúr´¾2Hà]¡ÚNÖ™ËûÇq„¼Z%h‹ãs»7ZFxá„Ûž§ªn¬ÀIw¦ˆ·‡; ýÓÓ[¹¹f¢ãŠ¡k*l*Ñ)’Ï4æ¢:_×¾8Äz%ºTâõ|øÂ™½§u ¾†Ü¯~½ÀYÛÖîè}; Ùº†uÌt9¶O䘮^\¶÷êß|ÎCÕŒhí1vȧ4][°d ›Zì%ZÖ±F÷z#ʺ㕅a8XÞDn]«ŠÎÒ5šb3Ûg£„MÖÆ®N(Yšç¼T¥×Ž{ÇβvÒ·÷]+G9ËfÞ@…EOÏ Í;€Jv{Íd®Z ò‚‚•ö–¨ªuåVÇkT»ùR1à†j,)qr‹Z‰Z³Ì=®VMDM³8$˜£^¾imöªh4p¸pf™(±BØØþšŠÆóoØ¥ªi IÛúãlóãðÈz½›Åhê}”«\VÍÜl]ßgਭ‘fV+ºpô›j³–÷ÐOT—OU@«ÃÏxwì‡Öl‚«ï:F›,3e¥ ´ý¸aå8Íf·õòŠ}ììh§'áqdŸNhhVÒY‡“nUØ™®âGû¨é𕲿è8yâ¾Vtr=öš[ g°=ܹY]Vtæ\>‰®'¶ŸÌ?ÛŒn¹e±*w"ÑC@Úi„’z:²,{£×‰±ƒd,qÊç•jŵÉèæMî¾ÕõÞ©‚m8Õ_pËM™ œ¶ÄÅ«P„þV²³z6jã&1âÐè"=zѰ0T»»£+:ÑÙ†|–^'uÒÊñ[í=¡`yPÔZQ{ÈZxe kh8¤ÖíŸÄ É Câ‹—sAâ¼¥Qû5àa®ÿB¿é±p_Fkü§4¸g?®Úôâv¸[vJªÅÁî~~›g´t1Ä9iŠ¿”‡eN•s¢žž˜cÎÙ¶’ù–óy )© çÍ?Ÿ´wœg„ý cOÇÿ!ë,ƒªjÛ6ŒH(Ý‚4’"Jx]€t—€”  *" ˆ"‚Ò¢¢"¢"-©tK7ˆtw7ß~¼Ïóͼ¿îÙ{íY³fÍ:îãÚ×ršþlÄ>b›œ…>J²óÎÐ/ºRLHÐ0p´S,@‹Óà aX¢¢¼» FæîrÝæ•…±gøíV{.L¤¾°—øB൑[¥ñ¬ÌržÈXX¹²ª mÍX 3¢sz‹zeªtüa©ã„­òòOX¹ô†í¿ ¬‰w†{y‰ÁÆÁ¬¨K\£°%†Ç¢6`ç Q±qÊ%$ºÙ´Ed±{Z/Níèá^K÷ßÛÆmH²}>Ï%tÉ*í†Geëq_î…‚uy¤è97ùû|R|VÊ^Œ´7NLù–¿Ez²÷_çøÞ Ãc޲w»È¤_;J¡¨Ž,V+õñ’ÞÈú™dIªiÙåîÚø{H"ý¶i‡=òÄyŠÎý¹ƒ|–&7|¯]E~ùšz|5ƒ‚!tÅ}é(üXÂŒ§¿E÷°ßú‚bé.Ù×i»PBÿ"YÊwY”ìñt÷Çcá¿O |Ai·çUe˜PÖåîù‰OxâYTMв5*4Û&ß–ßAÐøóø· *Ýj=~Íç*–ÀS×ö¿™Ù·‚jL—i§Â> zC‰ôþ÷ö¨ùQjÕ;ÿjGì;ûæýiÔMÐ[Ô¯‰ß©§@ÃG1VR•h”x#O7â 7ùöc.šR, ë‡$âiíá”GðÍ‚mè"÷)£y{ñ 5o5<Ã]xìÉËp´t9sSbî.Z•7‘ÊEö¢Í±}ŠûÅJÐ6‡¸àòŽÚ­Õ©qG{:Ž×ívÔè¤çœØˆ–‚‡Z$Û}QèS~™ —&äÒþˆA¥)ç+Ðò†iC6Úꎱœ”Òk†*VÎÊX¶‹Pž(EƬµ¥Ã|‡í¡ØoÐúû>)Ø1݃ÂÊÛŠ–PèXñÆFЍ4’îyƒRä˜ÞA(ðLg±? ]¢˹5uásã ¾nI‹û2Ÿ˜ãÇtŽC3ÛÎûÜ×_¡%Å¡=ó©!´q¾yù׫Úµ=Ê? _>.®CDZ<³?ÐÑ2æÅéÕ6q Š+ÐQ¢ulIº:ÙS;ºS^AgÆŒ“ó¤)t=”~zÐQºoö&a–…žt‘ºÈ†)è•2ÿh púˆ©Ëä6ò¡o+irˆHÄ-µ(z=ñ©#Uu- s‡:¹Ï#“ó"Ô.Ú06šÆó\@&Ù ._¾ Ó/hìn՗ÜHo×(Â|ª?)E *,ZrøÌK…À²ÐŠ-{¤.¬² ¼ ‰-€uŒ¶©rͧTtoï5ÁöêGÖcÃH¤ó)Áxë6îÉyqf_ˆ$îÕ yî¿d¤dº:yH6ì8!¾|÷ îKßüR†”DL[  H­0]u€+ iƒ–æJX#ÝÎ0™-ã&2hRo3ž­@ÆDǯ™Ÿ"xN¢Y2U9 ìÝÕ¢ç‹Æ³8<ÈÙ@u]Ÿ9获…´^A.ů%m'GHnä¼Í,ò†MNÐÕV!ß•LÜR-ò?šJ`x—‚Ê´#ëyÏQøpèy²¾ < ªç´bDñÈ`\~ð .”ÌeyK¢DMÌ­8â|”’Rm´}e qw¢DiUñf£-”%Q½Ùø Ê­ q%Ô5 ‚€íÂÞÍ<8<Ñs•¨œt|}•Ve5R3P¥^m—¬¨UUØí£)~£ZéÓŸ2kr¨a᯷“„ZDʇ¬TP»à¦ÂôIÔ ï}iq&õž®+íFC-¦§E"7ÑÈüÊhep»œ~7ÑJÈ£ÏÏä¾9-¦¹¯&ÞöŸÓ‹ãÊÔ3gˆ;¢…ïÉ„VŒx¦ßÁGX­¤ÙgO‰ò õÛו}ŒehË&õãSÞ}]P¹ý7í"Äc“®½D{‡”\ð¡í-žè¥j4~oÁí¥»1ë°,Án¬×N@cü%Ÿ¾Ãæl3ßÚ­º ê£‡püw6(uºßKê{Š>“ÜË …ŸU›‹­^„ÞwsùRVd_ 6¿…,UMÔãui²L­˜« ËyHA¯0~¼®µõØj‚œ‰ƒÄû7@ÁEÍosšP¸ÔbôU‹[ýç¢ý¡LüsXð9ïÖ!bo¡òHohÿ±@Âu¼ý¬Ð55 ïòùRZ Vµæì'¿Vø}ö"+9)Ô]ÅgLêõ׌ØÎPÇAƒÒpeõ¯Ð°õˆÕŽ;{tÅ\›œ¡)K*ˆ/§šÓÝéB®fAËðÏÀX¥6hc¥“¯VYƒöG¦g”Õ‡ c¤ÏUS/ º‚DªƒÒø¡ç”áîF€^«ìOtæ}ЧS§¾_ ý~æg£RÃ`àÏùÑ&6:Ûøê‹­Œ–æwMÃè—ßNµä”0žoö%ʨ&›æÓ–ê÷ ƒO°Ã#˜SÜà/OI‚ù5Û¯? ÀB=·øåÖX’)#ÞU¼ Ëé,E Ù°êýYÅY;Ö­‰“}2nÁæy• ú6Øóò¹¯0ŠD{÷Ò<›MÂ=69˜Yb‘¸§Û ÆI<'ZwëLQõ”ë~.¨_0ȉxv‹~Ñ ©Ü‰Ý«ÒEš,¶Î~‰«H÷ZX‹çf.20ÞtŠGƦS£ƒ‰Ìí–þ>@V׈<¯ï¼ô×ÓUþ÷üWbäQ&ö„’[7dx÷Í\ÀC ±$\åWQ é”ÿi©4&³!ÉQ ÄÃî<"ŠðÅ8,̳»P|½û›Þ/6” tp]¿ÁˆÇd+j)Ô¢4ezóÍ(Kl¼F3:€'²‡'8‡¡‚¥ô³%¤Æ“U_oLŸ»ŽJ²o>J¹[¢ò§æùxê8ɱ)HCÕîÏyb<&¨Çp?RþjÞvvΚíFmlÎsa#”۟Р~Ó៫Q)hør¿ìSGhô³|íÑxE,Õøš*ðU‹*?ÄÓ‘ª²,ÑìÏÝ=?ËVÐBŠÉ@©¬Ïømhõ”j£eW³PÜa ´>‘³ÿÉ‚Ú$Z ežÁ³b?T~ÿþ…vNbõGt¶Ñ^õ¢©´ìc ð÷,5¾gÂm>Ј¡À×+z‚~áÝ«LiÚð÷u±™Ç"?4S^?gy™ð¼K•$Ûø ¡/c”‡› Ì¸j,› J8Ûû­¬c¡è²Ó]W>^(”z¡±È?§^HV…ñÂϾÊΙ´PȾ—S¨ÇŠ4XØ)ÍCI´C/¥„”·‹¦S‡@uYÂN†½ÔåJÉÞi™FR2öò'vÐäN/‘ۚͻ‰å…¼ãÐjqòûæh»±'6Å’ ÚîGdB{þ‡S,´×|/úíÅ íñÑW ="U=¢•Úû,¹«¡Ã©ì¸&t 9ô®H÷Cùi¶î¿ÄÐ}¢š(>xz’Dï¿àv€^7ãÇÒÐ'ª8TcGýæë-Cô…0ð=°¡ýÌ6 )e$ Œ ŒPú;p3À ,… ™wý¹C0õqKy Ÿ;OÜä¼r æOÚÓærÁÂОâð”’aº#óVâMè~nQÀZù÷r“£”°I¯¬·½¶½ƒŒ·ÎÝ@"ªú_ E¸Çk4/‘éî¥ì:ð¥¸IÊÿæ}¾½ŠdŸöJÊ$fà¾ßaâÁR”8h¬ Õî0ñÓ¤5=ö8wéú"ædóAx®z^»šªu.!ó^­×#­âÈ’wIX7Y¥®9=)l ÏNš!ûß–]yªÈEþdì6ô"I_è7ªxä½Aº'AÇùšnÿ XHF~óù#¿^žBA*­;¯;ߢÐôÔU›á@Yã=}êWÕàòóÀ£á]Þ•÷ Q"¢dFþ*»åv‹¾$ñC¹è}kIí(_"šT~瞘‰ÉP BÌ*:1øÃ•eè.·ñ ÊëûÌ$-Q•Îù ݲª…†ï¬ôG ‘³‡6P³åÊ,ûÂ3B]ŠÞ+탺^²V¨#M¬Øõªþ‰ö<×F–§N'°G£ñ­Ï5ZÐäMnBÇV š¶&w+ B³¦ùw9><÷¨K Ð"Û®Õr-©‰Ï/ÊŠ£ÕÕüœ÷ûhÝí×ÑÖô!ã“°<ÛåJ©ùq í~ÐÐ-5@ûõ |艾Ÿk¥mo ñÑë¿Û¹"÷ºÚë+.Y°M´êÓïÝ#½ùTB½Çø³Ú „üx²FM" Šgê„^Ç6òf€¶Ð.äÿ(½ðÕ7rNÜv§p€¬ªµ£ þ S-§;OÄ 2˜M !£»3á2å>Èœ¸–HÚL苟Í!}ÃÜl–~Ë£^O žÕ—[£‚7ýGŒTLPú«8E…®Ê™ypêëT8¼É½ã•ÉJÑ’—¹¡ª€B1X ºæ—¢(ÔH†«øCmÆõZ¦ÇŸáwjÌmªm¨çz—e Nôõ‘FùãŽ$äCÐt4òÐä·Ðl9èË'MhylBZ‘Ð:óaœøê.´+_8yòêèÌe(½%Oè‹{›²žCÏ;‡Ç´íµÐÛ®QVî}ñ‰#{ žî_Ë`÷îXA¯(— ¾f§²’ó}#5V|<2Ö0VÁµ¼»v&ZJ±BLÓLê‹5óÃìÅÓ‡ý%ÝanþÔùÃc ° k'Le°nø)ÿ! ,ÓW0Ž™èÀJ…ç¬ û¬½Ö/}¥MÑêE¦`+AéKÅØå» rŒ*‰†?j_<:…Äg•“«àÞ¥›àœ€¤)BoWU)‘<„Û¸à7î·¡2óíGÊò»//{zâÚz¾RDºc_‡ ËÛþÍÓ4·¤³Èx:w,_™™MòFØžûã ¡‰x‹iï/ä¨õdæºÜ”lÕ¯­#ï+Öº ç³xè¹íù°^ÚõKE¡GéFo´Qdœ¿+{< ÓÄ,\Í”15*¥×uúÚôaEäA¨«•9¬é•홆ȧ´›Þ[»^ÏýzöÀnÌlê‚î‘WæÒˆFâZÞw /õ‘äñ arù§Hvñ­çI«qÜwñtU§Ùq¤ðO»´ÃŽT%=¶ÓkVHËz+­ÆéŠ5»X­n#Ãþæ¨zétd¼øª8‚V™’þxÅ!‹›Tf¨9²R‘j¯E¶ ÈGöï7éùÞ±#gÓ¥@‰O¾È]Fç3íñy¸Øª:‘/i\ýú\ò‹,§_yPí”WÓP(UÍ+cBERw4OšP£h¯öm€³xTnž¬ë¡+JØO9´uz dÎ¥¾Wï¦ð¯–·ÅI<žPã óeNG^VüåäÇCµý{QÞní|ÀV*­êFN§!û¬mgGe‘°‚ç¨âœ³çãÛq<Õ›—ð7 ÕÎÅ©²Dµ¢ú†ì$õê:j&FåöÄË ¶ÛÍ6“FÔ5å}ÅL‚ú¶]Y9ÑP)á“x¬"Ù°•/€ÜoNõ¾ X€ GáÛóœÐñCùöÑ%BÍY~Mûgâþþ;gæ¿\å&ˆ7:Š7ë1»l•!“ÏŽLä÷H/ºÞµ56é:õôþ§ÿY3dYæÃòõ!‹:É@3-²ëƒw§‡ ÿN íÉW¡™ÙÇ,fŠRÝÃ=gÿ™wó_îËÅÕ…r®B¹Ö_¥C—¡BZ‹-cÕìŸyTUT‡^ÏÁïæQU’WE^ùNè‘'”Ü"]þ™kUçŠ{õ—¥æ]ý—÷æá˜/{–D²ÿÃ{»œ¶åº¢t´ü¸-Ó]£tN´æ9ð‡eãM”†Á?óªþ;—j ô¬ÒÍó`HÂd=2ý OŒõÜtL†ÑÆQ%§¸4o‹UHÖ‚)òÒƒ5Ê%0c–EyO8ÌÙçVf ‡Áü›w<ß&^ÁÂb®WU1,9vÓÓ^î…FÍXýisXí}Õ•ö7Ö뵤¼åasøòn‡ñì@¹“®ü~$r)ñÌÓPÄ=ý &uà^×-É„ö{HÊO±s_[ɶ—‰ ²q?©%G ƒRr¸/›ñÛ#µBgF %Ò†ßK,9€ô¢ñÛêGƒ¡Sr<üT(2ÕÇ];­ð Ÿ¸`ûœÙnx¬tUû#‡H¦µŽÉ)ä»hïùûö¿ó¤¶]þpº"ÿhÂ5¯¶"ɲß¡B‘gÒa•(:#ö²‰¢ ¦/L >pA‰Î®³{®Ð¡”·OÍ‹tx\e²ñÇ|9ʈîð˜-˜£œ4¯H³Å^”wÔÓEÅßÚz>’û¯^¢¾˜W€ÊÂm#d•¢¨’³n5–$Šª—Eè“eÓP].•)íbjò÷˜Š™Ù ¶Ü¨1÷QB>yÅ.Šúmi+ã¢[høihNÑX&ÓÕÁòš(Hf?‘/DÓßС7òhF¶á“9ƒæ·ò‡h”íþåÐઇH&Ze–¿ð|6ÜÄÑ6”h—þ¶ùÿrøu·pnä>NºÌ¯f/„¡uÇåiñ1Í¢¦ì”ú|¸@Ïݦü;1ŠÐH1¬WUª¢?øÒäCÙó µ9F(ñi _RÉ„"ŸAɪ·ðk0ìÀ{÷ËPðžH†ÔÇòe âF­+ä3,†Ei»@a;ù“>ø•X˜*—·ŠøçÓöz¦o7Ñî}¨rp•h¶3ßz‡®ˆ0†¿<žYii)šíÇ¡Y4RxÙ›é?ó¢æJèV¡ÕÏú uæ´ÊWñôB[<]aΓ h vØä\ÿmZŽ…6’ÐÆéŸ[IÎmºŸv’\ -?âüiã%h¿lßÊY¤ r‘×èT óŒA›ÅAèÊØv8*Ö=çå•¥^ïƒÞãÃvSßV Oá´@=3ôø±‰=ƒ´ª÷üÎJÂPÓëúÛô”0R|vY5à ŒÍ4[´€I³ Â±_˜aR¾¤asÏÞ¼ïºo GÆ8^êÓÃâ|ŠÁ §Xîñ3žºU«ÛeôwùÁ†¶›ƒCÇlådoÌlˆÁîÙ²6¢¸‡›ÓßÁ‰SVÜnæ ‰Ó¾oW+N ®Ò}_Â}Ò‡iîÌÙ!…ö‘é· HåkðtIòÒ´Ï…èš!ÝG òã•_~ný 12š°Ž‰+¸j¤U aE–³£ÄZTÓx`gR=y¦ÙòWÞÛ]ãAö„žìA ;ä¬<~ŸÐ£¸ ?v%F@^ÐMá'ÈGõÕs¶ùYÃæÐ @!‡¸›Z- …j[šIº Èó„K¾oVP4}9óé{x”Ì·á‘J¨›L|³hCÉ8#Q×·2xŒ•íœñxTjÊÇ;(c¼]è͸唎¨¼x{åÏ/º„w â…œl)ˆª]9]_ØQië‰ú»Ì Tq|±ÿ÷ÖÁ‡%ûzTo¢Ú•máåW¨AÛRÕæ¸ƒš%¿Kü?¢v䟛S Q×ïÈ'óï¨ÿäHãËKhh¿¾1j‹FÎ$‰ÕòËhüüͲp"šäfžf`»¦+¾Äëˆf hÏ“+@ðáÎ#ŠÉ@´^žNÎvDK•›F³¤­h•Ðx±é Ú0 ßÎá_ øP&štR™À¡÷-ù_ÿχoHœ?¢/±eæ|T9]ûÛs¹~1ù“_‡õÚ¦N2•Rè µ{IsüÔlǪǰç@™÷8؇¢ëI—­  ¹”6’çäRŸ,ݺû ²úÝÍm„ ãíºñ‰Ñc®D-$‘ i?¼L¢è -mPËÕñ1¤‡8‰Ø½°„Ì‹kíâ;ÍxúFê[oÈûL½ðõÇ=øEhb–|é_¼„lœ¢óÃ~Èž:ºœžïŽ\ qiΞÈ3[A;·µ|_Cÿ¼ênBþ—RæË(˜p›³¿v…×ùÕ…›£hà“$1\»yJÈ”qî¬ËÍÖv¢ËðKÈpaâð$¡>Ô½x7ò¼võôÇ ïØ%­Èa=È׫g¹l ?S“™²Ø (?ß•»ø8á|¬"íqyRÐts´j—Úò£¦Ã¡AXòú½•ÿÌOŒo{ûš÷)ÇòœüHÈ‹+ŒWã¿Bë•Ú„±žDh“_ç°øí m>I˯½ ÍÖxòçí%h£ìZ–*õÖ†%³è h]Ò‹x5E8nnÝ"•íûÅo]òž…ööó¦ï² cáøò£³Ç Kcój…>tO8;U–”ÁŸ,1²=þðwŠ·uøôŸ4)i7|ùé,®në0t›{·ÄÁFÎ/ñÆŒ®ÁØÓy©ãç`bVÚÖïø~˜þèytßU˜3$¯|ËXó“ìÄuÚgañóŒ‰ÂZ ,0\ “‡Õo%K\\ý°¾}:»N&¶n¬]<ÃzvÅDTF¨ãž½ë3>«H^ƒ¬îH¢'Õf2†dô¾„íÝÅ}{vLxž(µ(MVuñ¥LûMe h.7A9µ«óM9³(aæV|†.*~›lSKF”=v;‡àÃ¥‰µGn]¨rö<ýÊ&‚«ÏŠ6@µ[Ö– ƶ¨!p-1®…5Gfú’œ™Qû—Úb¡s2ê¦\cÖúŠú™÷®íËDÃsotË¥|ÐÈsö]}0G]LUí>„&%è_£é–vMÙç>4S>±ÉŸæA…ÎŽ¢Ù 2/¡¥ò>%ϲGhõñü‹ô! ´a"Qas+GÛÑÅqwà ^gåÿuîÿù°í|¢IúÒ†ñ*£Òé'?³wXÌesÖó¾ s†v§ûŸ¦o~e•9Á3P:ö›^dŸæ´ ŠrúBÅîÓ|*ȹ©“Zó²LUõŽZSCwHœ6Q2ÁƒŸÜx=@šäÝ~ž®zHc[õ±pχ´éöGì!£Tã,Ë®&ü0?èsÞ„r[5ÚØ[< ½Ó¿4ïÎU(Lf¨ꀢʇ,öËÄP\sEò(5¡'vq|X _"\OùSåß ¬òÚ¦Bj”'2WôÜa‡ŠÇy•’fPét~Kí† TiÐ8ðSÄ@uìÍð…ƒŽP˳–"{^~§j¦yž‚ú5¥Í“bÐÐ[ôàIÝ4®;Ù7½úÍZÌõ2S!Ðâ’Ì?N¸#>¿I¸/±ÙNÔBg$È^ˆŒ†îïK™VÓ w_QtDBŸ~@Ä,Û!èÿÅ6~¥¨õTè‚‚`h…é,¥„ŒTZ|‘áò‡±L3šƒ÷a¢®†3ŸÁ4åH ë ˜½’æ]š s i·öXÂz§ ¥÷mXü¡q0@?–͵IƒŽ®Á*•¶çYáXû+›zWj6ZBfGÛÜ`›2V2Ç›vŸ\ztðÆ$î9úï+œBâ·*–' y6Â Š©Éä_­¾êÆ}2wüš~{#…²ÅÝS—ÊÜMÜô”ÒÄ÷v_¤cFºDþEÒ¯iÈ ¦¨âõ™„¯Ð® !‹†ò¼ã™"d}[b¿ŸÈÙåq$"w+ùÔ~äùò¾–>ù¬'õ,^G~ùŸo˜:åQPódggÌ0 ‡sµ† (ÿhq­á[ª»¡Eð³Äc¯°ÇPJž&5úó:§÷³&ÖAée5IYõx”÷8è¸ÿ!ÊëRz=eõDž¸vþòˆiþzD>ƨì¬LIÙ‚§èÞöÛ5DÕ?¯åÕ¹P½ÂB$û!¢fõóþȹyÔž ì¸ðõøGY–T‡ÐÀuÅ$ÌŠ G×î}x¦€Æfv™’ì;˜K¨ b2²rñÀPæ3Ç·íÈ–xÁ½XEÙÓ÷9„N"gúj2ÇîUäþÑðŒëo4ò¿ÿk¯Õ4òù™xúÍ\ÂCËÍxKä) ¼Ô¿T;@ŒBæ¦ìQ¿ÖPDuå+õ‰ µv¥K™ÍG±w 绢-ߊ¥y$J^ä.b‹­A©žÃµ÷Ý*ñø{^½Ô(£Wmmžþå´Sß±4{¢ü¥qŸïÁ´¨ø~ßš’¥ pxž®I•äÿù°YXWá;úrü~þ¢Ï ›£‹¥Ó‰c'³¥Eú¬ÅxÈsŒ0@Û¯¸ŸµPí7i¥§SªÒ¡Ð*ƒC !òf˜ÛCv÷™¯¢lvYÁz“Ôr ÒãD',†´“ʹé…úúá´æ çsHõ¡ÌQ~4iZäQkR×Þ Ù¬Wƒ˜I?C'ó7 àâñÅqùPxäãv_Œ.)KÈ;%ÛB±Ô5“*¢Z(‘¶ 5Z„R9͵((c sޥ倲ÉrÖÓD ¼ìñ£)¨x£“GtD*½èS/Ÿ †ªÊ¥´î~‚ßõnw…BmMVCC¼3ÔÅ''›µ†† tö ߃¡ñvÃá3çU¡©èã†[¤>4/Žª’ëhC«{V›å*´]/ ôçÙÓÜ!²O k³ÎÌÄ`þè ­/ÊŠÁßöpw›oaÐoºzrlnºs8(ïý†!ï$³ð_0rê¥Ýk[7ãϬyr'&¤ xS¥æ`êÂýÃüoa¦Èp=ùž7ÌÅ*ð‰Òü„ù9óS3X<»Í­Zø–ÖµsÅ *aå=Ÿ†´^¬9Uç{Ѻ‚£³QÒ°uœeŽýìôwÕM]øˆD-«™ÞÌkHl%êøˆIö[­ñßFÒŽ®LÏ:¼ž:{æ )îo®*’羊”à W¾Ÿ[@šÃ¯v"ÝÇéaç¤ßT¾ñ’àQÆ–¦¿Z±/‘y¼\f¬Ye/Eð°ßǃÅzÂ~/º3šÒJàÓMä±²•ý$¨‰|Ô*,‘xhH;sÀò Lì o¼Ü„Â’lyîx¸Pµ[ëzŠ…+ÊXdC [-q­¥—(ÅvšÿIõ [0²I¦xŽÒc9ŸÜ*P¶þmÈ·U”î—Ö>\¨8¡)n(Ñ‚Ø6UëÀ‰Ê¾Ñ3ÏÅñ”â—ýÍ"¨vÀ\?”³5h¾d(û¡ç…D=PDSÒ¾†gPïŽZ«-GäGí_D£}ÄjŸ•ÐXþ²óðÆ14ñüòl(çš–»ìIŽäE3î6á‡f hþ ØWÿGü¿j¤µn]ÓA«oZ¢hÃv:í”n2ÚË\¶Òû_¿ÆU¥ê¥àãì|µâgôh®|¶)4 y—ò-¦ÃÖàà»8"kè:s[^r›Î÷³ô³PQg¥£wjJ÷‹¿Ú/Å´<¡cQû¡PUŒ’×e ’«?ß Ÿ¼—^GÝó\=È#=ÄLÈ­¦é‘ìÔ«ÉK-c ù%wik¿ê@á¡Vô[…RU’ ¿n ¨´ {­»’µª'MÌcB} ó/?–ÃÐhyöñ±W."|Ý‹+ …“΢¬œZO~SysXŒ;þz C›ÒŸA—âHhÝ8Z»{÷´ 3å†VÙº¿7~ùC멨rïÙ{Ðú®Ó qÚÔ«ó‰æÕ¡i“ýÝ«8è8´›mãâ.«¾!t­ýÎX¦GèÉÎ7ò=ñzÛé& û) /am¡:G««…=…Á*~;3Š^þÄq[;¡ FË ²,¬ÂσèÜ_`ªbøÇSù@˜ý–ÂoæÅóNê5U6añàeËHv9XZzyþ7'¬Råœ5zëçyìß.ˆÂæØ­Sb¦©°óEÃ;nãýNû™[]ÄgoöØ‹EÒª“]Ô´l£wâ+3’Ç$ô8ýÄýá­ž^HYð™‘ìÒ°]¨[A:ýG/žÐ)!}Â¥âW/f‘‘!Ôb£ð.2E&)ÞíöGñ½c—Ÿ¾Åí¯¾+­w#[—íâá"dÎ’" /œŸs+."wºãPqò^~;÷,ùnyô>™N#pvýª€}5 xúoJp Ð ‹-yò“(Â÷ÛØ?E%¦ïåw£˜«œ˜?ÙqeZ³ÿ6ƒ’zä”31Ì(UÆ"í¹‚Ç¯:ÚÑt ŒŽD’úÞU”3X2´œ&Bù+COó›P1zxÓë$Dö¡-:‚‡¿ZëŽ ¡Š 5Ÿ=|ØIÙvƒ Õžt³£C Ã/ÄShQK@ÌóØ"êаÌ;=|Šzû¶å‚ÿuÑU=4ô‹\°®,G£°2‹×kehœÜ¼<¤ÄŠ&-¾1tžixš¼‡M‡’Í4ìe®ÿEó—ªrKÝÐb0Y©‘-A‚ó™ê ZÅ_ŠÖÞ׌6ôy ]û\>´c®‡¦ûïW&ü?ÖOÍÊ@ßÃdë6~‚hè¿x´/N 9¸üïÈ_‡5⊉DÙh;PÐúeÉ—à‘àSŠ­ÂPR–qÙ–~mò%)“ÇA^aÔß› Y­âô¦ØI2OL½ê9Óée2—%@j^©ú>H=p$%ÏNRƸçGühÞPJrÒo]Öé†ÌÞ±åÙJ#–»Asòï øWQÀ/3—/‹z¾„>úÇÖÅ)Š¢=<Î/@qð+o[ê(y\¤¯qiJ­^õ_±…2q‹¿ýF |¯ÉÂõ«+PÞ~Ì+wX*¾0í¤$úCåÒ÷ɯ­¡Ú>^¶ÎhjFøg…–Bàw+ã•wÏ ¾ë§&4lÉ-ô‰vB“¼Œ@’ 4_6[ûk-}®Ëõƒ6ýœ¨ö|èиB^Ð~º4™Æ^&2COÐ KÂþñ—-ÿ¡·õM´„8òˆò«„Ðc-Úƒýï6ÂðÏj’8K cØ`É‚ñð¦ºÄ‡¦0YtàIjÛ:Ì0Ï^+‘S†9µ’ߦe7a>ºu÷µ˜$,îKçß#K1:WëÝ¢`åÄ åu1Xå滛ø ë%¹WÏlÂfǵ-Á|1Ø9¡ËúóŒ-y_õ˜KÆ=;‹¶™,pïÞ·)ï"é-Á]É{-Hî7M„û/(JT^GÊdI¾aHj™þ{FéøÆÊ·"}ªú» ïcÈès"9†á$2?Ó¾sEÐÔÑÖÈÊÜÁƒÚ÷‹¦Ö¨“A9ù¬Úuä¡©ñ•ŸAÞ‚WGZX6ðÐó¶ÚïÝe(ð´Ì¨êÄOúÍäîI‚‡­Þ-–‡ £˜ôçõ+Æ(!ÍZMU›’k1‘Þ“ƒx¬ý†,C=Jÿ|¦ràÕÊ~«¡XÉ#Ay‰cŸÅÆËPqqA»]WÞ?‹RCå·bi îâ©‹ Ѽ¦î¨fDîïå'¨qz~¾®F-7t%~ê:±x»ÒÜõš?„V,Å¡!Ilw9 SQS•«£±é'Ïâ7Ñä¹È=‹ZhÚ’'¨lh&Ìíä-«…æKís³ÿåPk¹áX3Z%¹_ѳLD›jŠK§Ð6HS®&Kû9üzï–zÙ|\àk8¥Ff ‰Y5d„ûEtxEÀ“ ¶N%¯nTC‡K¡ÞdµÔ%{d®É)A…ôé…/ûÏ®}äk( >9”q~e<Õy®õòûýÔqƒÜ~ª —ך GMùZ®BvXõňÌÂkí}1o; ¯ö‚_â'(t± 0¸pJþn²XyõBå^"÷{üP39—µh õw\ÅT›5¡QqæöýmhŠSZ~ûÈšw_{_3¯ƒVAÎàÛ†sк¤YDÞÌm<·¢?· Bk%GÁ›h5Ö¸qZ&•‰M +Ø^qu#xñ•–ac-´GÙê~™‡¶ôò]¡ О?ÌVuót-?Okφ®û;ÝOe· G~ïêþ‹Ç¡W7িûôYÕéÃÇítôñÊhF¬7ÕF¹Ó2»oÈPÞõox•F*õ]=V•ˆHs²ÀôñATꥦ‘ˆ×GÝî.¦öi'äÑ/ïë¿¢ÆMÕ5.&ÔRuc;Ëzu¤ï»)ÐÞF=i•Ÿú¬‰h {‹8ðj†‡© 3Ž¡Q|ÊÉ¡çvhœ»?êØB(šüá°¡.ÆÓÔ ©óŠÑLûfÓ†Ãu4!ãr?Ç€“?ž}‹–Š¥T,$!hõv*.œmè"£äM•ÐöY ^<BàPXLOóÿù°65Œ2}e•Í¡¡¼nÌÓÀdóCoÝ`õ2[̱@;huí™zzܪ¨d¦*V¡DÜÓŸr?+ü‚?2.SÇþ·þÒ JøñÙ=ôL&d|-,J·(„´[ßÛz|*ïaÐa0)OHXÞS@ŠÁqÆKHÙMÜÔimR+µ× “ïR’çåÈ>³üà’èEB® vj´€Ÿ&AÖÎP(õÄ÷ï("òŽÚm‚¢ÉÎ ÁUóPÜ~¢TK!J>Šë=ªz ¥7sÔ÷ëB™2yÈ&”Ӳ˹Ǒ@ùV•Õ;PÉó;â¸mTù¬lFW©@ Q(õ}õ\¨m÷¹óÑ$êÆ­µüúƒ¡á KW¡1­Üppªš:O\š¸á-Z¦wýÒ8?xS0ÚßøÎŽßº aV·ˆ4 »i)÷ëž}^ãH¬?HB߃ԑ~è_ûtl¯U9 ú·ÆOÛ¹ÿç1jœ{ ¼îuW¿”ÒcëfÞoš³a’ýûĽgÆ0}ùùxûÂ>˜íšÛZ3ƒùS¦W.QÃBê%á‚nOêr}y–‡ú#š`õ»«‹z¬+è·ï²Á&ôôÛXgÁvpÁ¹`]V$âíxw½æ<î‰Ozÿ÷êŽjTmBRæ÷ôÝœH‚Wðî[ûª%tÍ)‰GãeZ.ê-Ñ{¤-«¯Ðj®Ez'úÚÛÈ(ööîÔãqd>ZW4û¬8déÝF¶¦4«Óö€O=÷D º!÷󉟹o„‘×e>öe«5’=7x׆Ä©‹ö}A¡K´âe‘¡(²²KuêGi;Ïñ)JPŽ9ÔAÉö ª<–NgÌ΀ÒïÅ«—1 ìÓå‡ZŠò|…»ÆWPq;–]ÜŸ •¨Î°H1u¢òÏkZ/¯ã©ÐZa×ûä¨ö¤æ–è%Ôxm\¥ü…OKX¥t4 Î²Ö %ƒ.ÔÝk,Y{ O×®B#Mæ¥A4¾:ôˆÈ!MÞ}r­ÉžBÓÁýÚGéÐLâbÂoÑÜ¿\™ßªî_Už2¿_±D«OïN£ óñŠy‡Ñö…%¹¼šÆÿrøÕ–¿ç¤0>.Ö´fÞÙ‹fgÿÚ”Ü@{ñ’ý°ùþ¶¶Ë²,´gñ¦˜…ß-ïò_ÚCy¬¡ãav(!SMa[K…¢»£} Sð³/ÿAû>È—Qµ#%Z€\‰ K.cH,œgÝ ?2•8# ëùõ Q‚䶬ûÅß!ä[AŸAîPâ=ï(ÅêAÈÄÃq¤ÑPú~‚Eû ÔKWÆ;A#÷]êò@hº=y±Fc šÇqŒ?„Všç½§o¥CkOv|äýh#¢7͂֊d¥ÐÊä.µeÿZâ>ÇóAKÈàŠõìyh¶R©Oõ…V¿ DQb¡Í8dàk¨?´Ÿ®ö–öŽ‚ŽÐÅ(&b è"— æµÍ†î<ØàˆøS´»“O±}tq¼?yë ÿº‹{6 ² Ž¥^†¡ Õ¦uÇt%º“úÆjZæµarÂD™Æø=Ìäþ¸Rsƒ#1Å °ðdÄ~áä7X²Š¤³%Û+.¿î½õ3†µÊIŠßÞÁ¦õ מ—û`GjÑÎÝ6 ‰"_,9D ±—×g¿Ü[º¤yÓÿ’ÞLdP6EriÇäEyÜÏéÄ|¿^)åƒ\ ž"u¸òþÙ†HÇô~¬xâ=ÒßÌh ð Fƒ>ˆ±ñ ÓuR}*od¡¹âÓ&}8,­~D¶À’¦›mÈî=º›àŸ‹œáš÷›[!wRp„¡f òèw¨žÙƒ|6ISªx¨(¥Æ§ItØŠÌkPpû\ÁðJnŸ¥ýÃ㋇»B?4?ˆbV…Ý‘(î™N›÷Å %YZßEE&¡T$ç»@GKó7¥*hõÆõ}Ÿ˜>ÚÐÞ8ûê-Á‡‚—ˆÊGJot5Nþ?VyÈ.þB_U¯†š<4èÖ¥»2ž‰m…jÈúç`¥ÑYçg79´Tlí½|Ó* þD˜–Cq¬µù— :øéùB´Ero þ²±ƒtýç‰g ƒ;lì‰y¾'¬{rá5¤äнº™ ß×}©y ߯­»pAŠåš’Ub¤IúÜKr…Œ G³sºÂñóšç!74;H?Aa*•¹·)¡'ºêÎgŒ@¡Óƒ¼ènS(Ò`5*y4Å‚iò:ÔÇ¡x¦ä§–à”dµïÿù< JïMq¤?¦„2 âU ³ÿx1ú¤úu¨ÐTZæ1…Êx_ŽJ¯ š}h[ÔÔÌO ÷’NÂïÕ *Óø*¨ßa&ï¾M¢fw8#}¡IgàNùuhN?dtÂ+ZÖСlü@ºùe^óêB÷¢s4ÝôüÌøú20 ú„_]K†/®"w.€òCuÖZ ”¿º˜0 Ãa£¾>ïÕa”û+-Y㌕§.J“fÂxxߎi-L¾>=DßÓ3ç|ïE~€Ù«a»Û†2á§:/Ñú²,¼b Ø‘‡% ã ­Ó°BÛ6Àç kœe®jœ°qµÝyʶY‡ŸÎ…ßkæ#QŸèšÎ:ƒÔ $¡YÖ5ú`»’î~úº×wüI›K–Ã=$ü©æ÷O ù¨DÚ._"¤Ü$‰e -EËÂ6³'H7â`YF228ô».3 cMcÖ9d:zûT¾À*2ÿái½Giެ™ú¬¤SÈvS{ÿqWd?oúî¤Ør†¶å¶Dî·‚óT§ƒ—ÖF»+ùLGéãfï#¦É–$e Êdï÷B¡á“ñœƒ4(’óDŠÏû4Šfèßž8‚b½?‚“uQB-Ý#U%·K漯 ô-âo6PV“m/ßKE”Óp{D¿òŽ–8 ‰EÅ‹Ý)c;¨|µW×—÷5Âan¸Hàakë¹}±¨®zO±Bð‡ó´í/Ÿ¢Všüƒ'|r¨ó $t‹õëPlT ‚L½Ç„mÐèþ§ÆF#S4‰àà×t’DÓ‚{3e©âhVL~.?ÍÛ/|í´¹‹?MFi{ÇÑŠMŽæ!')Z›ÆQMyŠ6ïΊ¾°FÛÑqW¾—¡ý‘Ùïn/¢Ã‹¥”ƒxhM„¶f§Ñ™úžÙ)t(´j}ÿð°Nq` Ír”}‚ÑDíÏ÷ª d»¾áëÍ~ÖJ~é‹¥Cû~½\M¨„C1PE¡zÀð(”Ö™ñ*†@Q‡w†yMä_zp%Ô´r¢OѹZ߃¬§öÇÞbƒLùÔñ_Ã!>|¼]یҋ¤šï‡þåâ‚ñ*@Ö½v²Ø²"ÈIY®h'ƒüìS::Ç¡èãÜaáU(1ê(¹­e:~ '@ù ª’S/Ò¡â•òšÚO¨ôÍ4ê†*[ ê ¨Ú ±—gÜÕ%…R]{™ \œÉ›¥+'äE®ÉÛ= ÎŸ|6o´ê«oK,Bãa}ñY h¦¶4a)ÜŸÈž´ ¸CÑäÓïT’Ðþzòº |. üôü&tr, üürº¢ˆˆÖæ)“Ì Õ}è›N0ÌL…~»Ê«îMßàkåÛ‡±­0lû…;ÇÁF™ëm+”ÁX4±`ÐÁƒ0.yANÑyLŒÉVêNÃT…SwÄ…øÞ>’ʺN s{›ú‰?‚ŸAÅ/^wJÂBŒ™ÖzA-,‘tKË…êÀr€Ñ½ŒN Xùx„ ÖBÈŸüQ‡ Q¿Ú#Ø¢ÊNkzÑ¿0aMV*þ žßßM‡D3bn•ÛHåkMwIÍ©¯…Z6 ™”ù]g^q$çŒøÌT})9<ޱ^Aj¦Ë'Š!mYgvC;#ÒwZ AÆà3‘çÕ‘é|¯ášÝEd‰&ñ¼r÷}c’yì×Ç“©¹½‘ë8­Ðbò´&^£1çC¾°« ûŽ>Bsôw2ß¡ÐKÒ¤ãO³ð€x‘ŠÊ¯1cYVyÿJœ§“MMõG©˜3 &Æ•(ÜWµv0{ù x°9¢¼®þùÏZš¨Hæ³ðŠ¿UèöÐ( ê¡“y¿×“PmK7éqîjÌüСêC­?˜Pcš†º7‡tjcQߣ±šï¼;¾sr—T}…Æ_*Úv_“EÓK‘N‰"FhzàX §5š¿vŽßß&M¯Í³ %ÑrçÍ#y"´VÝ8ÖgŒ6÷w6w¶ÿÕ¡Š›•ÁÕSèðr#úý´:1ì$?ðŽàK¯'±º©ýo¾ç¼µG,Cš»|Où¬ Õ ô\—GÈUû\‡E¾¶Üè—Uf Gu¥ñˆw´ð¬7V§IC­§ÖM²5¨ÚeW'ú±*x†•¾P´Bé®Y¯h=øXZÆ}z Ûw|<ÁôׯƜ‹yykCG@¾rÍVS–9Nñ}S~šeJ¿Õÿ0ý‚ªœ!’a«·P·uN|ž)š¾QÞêÚ€Öä«~!”IÐ>þÜü€ÝeèU5áùD ï4¿] ž‡®á3W¨j \ûðôx~ßgè.×,q¦î‚n©ÉÈ€ÀRèzZe6*é]»ƒŽ§S–r¥„ʼ.t‰ þtö'tUkI ,Aw”hJÁ½3Ðòhî6AoÑüû^øÂmbô^úKû§â—Çàëûù½Â²ƒ0´üÒé;Œ|Ò³8c'Åy¯Ú¬Á¸¸÷Ád(˜TLÒä9ËÓ·Yé…»ea–º¢égühXÔ¿Í¡ .†ʲҰ¸}씑 ,WÆGÚÂ8¬~*Õ- ‚Cy9Ú§Xaëó…îO.ðûc‚–AÕ4¹ ÿs‰—X»gpW·…FÂÜÍo¥õä—’µ®ÿ\ŽFò¤ Õ¦^¤,¸XÏ]¹€4L’Ò;.!]^Ðõ0*dPß™;>-ŠŒ%áóGÏU!“as+—Š2ÿèfW>‚¬¯%ÓfíŸ]ù-sÈîYd"Ü©…œGfßN#÷Ëbºˆ›ÈKúIä‰òi’ú½šéFþøy‘!õZä0ØÎõF¡J¯åæ(‰æ'?Z h±KÞ¢-н¹i)º‰”fÂÌÉ‹(9’$¯þPÚc=…æYÊÞy}¯c7Ê©6Žîh©¢‚ûÍi >T¼Üž3ß?ŒÊç:ò½D˜ÐÙ™Ÿ#ð°^‚ ³Õ]Õ§x¸v\€ìjåÅĥܨFw\£f¨÷rÑ|om&¤Îþa4ú¸×$x* MŠÎä¼ìº†¦ wU YËÙ…’h>z“ŸrZ¬}Mk¸ŠV\r­>½Ak‹Œd§$1´ydðùŽâ&ÚŽ$Ñ.颽ÂÍÞDtˆ»˜îgÖˆN4¢?ä‹Ñ9l®Dj6’ CÖ¹Ãÿƒ‡U;7«1Ðy.ùé^34NΪÞÓ‡û^4²uZ.Àj"íûîVÐyYQJdEêzE¿Ÿj9NàÿÖJ(¥¤öeIì†"ñÜJÈJ8Ãÿrv«È)XùB«è‰ÔgÜ ;Êeœ j™FzÂàùèË– _êèä3kȜ折ØÙ? ;ïô:ôZ4?Ж ãXÊI} (Ö_ñ*N"èœ<ÖÊN³'å@ù¶˜©Ùv?TÔ¾2μ•s m£Ç¡ äS3b®¸¤½’ÀMÈ‹Þ4ÄAÍy>™Æq'¨}#ã÷ÚêÆVµIèþͽwT¤ I¨´ÐÄSZ˜?X<­» ¬ÅÙC;‡_GÔÜ&|–’}U.q:‚+6’  syb“lïsè~WŸìe£½þRáä_áË áJ‰Ï0ðK´€…†‚6¤½‚‘$j«KŠ0Æ3Ù ßÒÞϾŒ' ³G7 oòÁÿhàš ø~Пbü½Ìzœ2gᶇ½-Ëëª/`Û²+¢^ÂbZ¸ÜÒ–,ó«™,•Çvfî¤Õ°F»#îdÏë¯(+X\aÓ—ó1˜ÁöU2-ø#jIÖR°‰îÏ„¿6“Fiõ± í¸kÁvûv înüVêÿ§÷d1Sy"EªLÂ%b¤j5w–úi•¬’Ce‘þ®ƒ]‡ 2JwÍŽLÿD&‡hÿòdINxøê<î;­/×û×÷ok•žFÎöÌÃì”Èsc}¶™>ùd÷OŒô£c—ZòÅs(¤¼ãœ÷LEY¬Š¹/âÁ×Ý9)Á“î)iߊRóÜT}òQæôVœ®L6ÛïÖ§€òBæösíÆ#AäçïE•ýÏ~ý¹Óªè³ûÛw Tß'kÎÚÓ†š,×{ìþ”¡¶üf.ý WÔ=oäÑÔ1ŒúůîöVd ÑIÇä#dh"Å}ßÇ’M_ŒÈ6­Ð£Y²ÌAE4/}žšLz -†y5…ä;Њf¹0eh­õ4Ýì “Ð&²yؽƒá_*¶]äVY@‡}>Öbt¢Ûu²Èe‚àK#}ï,©ü_tø[<.ÉCÚc¯?ùAøüÛæï;[3{ÏÇ “Lجúªófü&t‡8ÄýÞ¼Í:|펟¡æ³*ÕªŽ%á¼7“aŸ…rýÍ©†· ä¸)OþÎWøH|eñ ¯rïûÎyìàÖ}æÜ„<–—ÊyÏ@Þ‘kùíP°@m»ïa”:z¤pöV@•I—YÞ @Ý+sÖçkÐVÚ"kzZ]»Æuê« ½^T³€o:ègª "ŸCçÝ2îø$說ŸœÝ‚îrí;Ç ;~È7–Û ºvºnñ‘RB—ýô¨Å“÷ÐY¸X¯üâḮ²C>JÐÙ/ahJ8Þ¥uêjL=yá¯É3\È?ÂæÞUR$Ø7–ŸˆB¯Ïu9ýE<§ÎØ£Œ¢¶…*·Ë.¡Øyί§ÒðP'tÕ»,”¬ d¾‹Òf?ÎT¿BYáI™u’Ï(’?G£‚§ZI†W**^m.ý•ÿ•O¶òeú' ô&V´ºxXÙ]1ê²rôƒ)þ.¾tÚµÊWß”£NIq[(êUr¸o¥ò A'ïé»(‹F“¶Ýšh2^©±cЀ¦_îDY©ú¢Y¿CÇŸáf4Ÿáõ-¹;·ÓT‘¡•@½Ã¾Úhm]­ÙE[Ž6i †ûž£íðhÀ©¤´—ß[§²Ëb7r*ÝÉЉš,¯øv‡&M׿tH>þíhÇÿÐay Z+z 1(`Gc:÷¸3ÞžÈZú¬òH„¬îqqriûívºä²û¡nÿçèUJY¨ø¶§å~î6!·ÕSïáƒÂÐíH¥í£§õÁ5î*d_fp¢³´ƒÌÇÕÙsè!ýTÙŠú6¼O­ö¹ê¸Òz)|Zàý—Ïm7T¼ SÛå•ážÇÀòG9H+˜`ã:³?¼ÀG]ŽtP"zC‰†žÀEíîÓ…ÿynªà<Á3³Ç™æú‡ðŽ0?Lm‹ìÿFÈ‹¦ŽÌE¡êçG—¨~#@/cA˜G4ÕäÝþ^o¢¬›õžÁ_ ްQM˜7”ã 4ówZVOž#äĻŤâAÐ&J³ýX“ó2«ŸÁç…È»§šÐiÓÜ0zì!AïCQOÖBOíÅêFWè‰Pæ|Tr|öNÃ`¦ñ¹$Öu‘¾í²Ìœ£/Ä’÷¯ŽÂ7¡‰—äƒs0^±¢,Â0r\â:³>L3œÊr͇™Ò=M¢_xá‡è¾~:%ÌwÄ!cÇ&,Zª)exqÀR­ñêS*¬ˆIñ(ð€Õ˜Õ*’¼-Xg‹/äâ냾P?ù¡~Øúx·åü¾$»Tãå‹DRÉÖc#HÜù{öL¿ îŠxs¥J¡w»°øø«îQ?gG‚ò¾vð”#•U”i˳:¤)è×i±Azt -0G†a¥biçXÜ[8Üû[H™Ë»RKÏ"ëòµŠu2rÜÖÌø¦Ã)äTÖqz¨i<&¡­SÈ;ørŽq#î“߉šN(طŶûYЏŸ?£ NmZjòÏâ¡V£„|*ü{ÓD**eÌ8.Æ*³ãáÃ¥sºÞïPž4ÁýØZ-™yÊ£¹J†*K?%O£ª©ŸUc¿ªËqï<ÛHBMuÚãÉÌPû\Nƒã +ê~L\6DÚ«¦÷i6ÑÈîœÔ)4 ½aðåà4­‘¢Z-ï@³åEW4ïí+8|¤-–?+¸ˆVjƒêÁóhmvÿ•ùÓm´‰þTÖ–$ü¯”-]&øRSâ'Ú;.-y> èð¥êJ€âÿE‡_¤9X1¤Û%´dt-ãŠ6ßè"Gü…Cº'#a“ï]ÍÐt}ª:[uÇš¼zV×;¡FAñé·ô“PItæ«Òµz1ŠX{Ÿ…"-ùOÜ=ÿöc<ÝM*ÜîúOÆ?}«-+Ãÿô`ü·ßæ¿=ÿíÁie?dN4úooÅLJýÌÅ Óë¨H³¤=teš0W1–B÷Ë"=RKÿŒÿýÞ]œtgkÿlÿ³§Ôk®ùCÿnïÄÝ­7Óûgÿî"¢‰'¿ gx¶òé{Fèã_‰óß|ÿooÍ“!áhçz¬oßÑ›”‡áîú„Ñ|©@š?·áÛÙÇAÖßÎÀÄ…á•7‡ a*÷˜³‹:ÌH²d 샹žƒÏ¼›`¾Èý”—],ž:›ø–…ßÒš_ª‡Uñb ¾ï5°þè–#›À}ØÒZîL¥(„ßê¢|Â\H$:(qòéa$.9—«(Ì€»4ô‹´ƒäòÏ$›{jÄ°FJYG^1Y+¤~ÈcæÝ%tj•Чà320Z²DQqŸýfá1(¦új*Dß=ŸŸ·¾(™êÛ2Õ‰ÒŠßdýS©Q–­¥æïúY9¥˜ëã¨àtz9XoÔ5PDŸ@e§ÆÈ¶7/7Þ÷xX°òZŸÕQY9R†ûšÚ¢V“ev«BêtØÆiJ ÞˆÎ¥ÑˆH4غ;ãØÌÆŒñ$†hJN1sÈ5 Mg¾ñ„?B³iíB­ 'Ð|‰¹ºêÌ&Z’s Óȳ¢•hñÀ¡)´¶ëÕ÷7'ð°-^ÑO… C®åQO´?ü.·®jž]-¢öÓD'ʱëìmÇÐùn½<ºPnF_ø:, ­؉þGo¶ŸD#?Ö‹5¿eÐÓ#¡~VÎdÓÌñ¾†‘ËÎSÈ Öù²ò-]Òúþ«»‚qõ›’1¹û¾ØÊXÑ ²f&nOZ d29þij¹N,ß=éý Ò¢|‘y4RßH¨ý¬ýg;Cå’r¢›8d]Tÿ¹0Õü¿tú_]³÷ysßl„’=¢÷¨ÌË ÔCø7Çl7”5Š$½»ðOŸMEíSƒÄ½úsª„^.±+Zÿ¿ûm‚ŠÞȌނ‡¨°ŠÖ”ôßBá`ÈP–þï<ÀÜðÊæ-|¶wN¤]ÿ ÅYOK²Âÿé¯éa(±ÞS„ÿè¶ÿœÌ©~êÿôLýÓ[ÃWuꄃ2Œ½â²>ËÙó¿{kFRo°ÒÁ Ã[uŒž‚¹ëó‰ñì—`^ðÙ¾¤ ,Xøµ®zzˆ –Nš¤F¾ËÕäóôÆ´°Ê5gÌÐ k¤3Çãa–ÎÜtûGÇ¿j˜‡Jxa'µESsX‰/V?ÜÆ+¸KlqsX´I·É³Šš‘¬k,n£,Ék›ö‹«º"å×VåФ‘µ{´Gÿ ÒÍ4Ÿ¦MjC†ðWa·»´q¯ƒ;ç­{W‘Ùj•¢²ï ²^Ͱ?꣈lßC^nžDŽwCwô¼_"w¹QÌýý‘7„ÛêÖ7wä·v(zºW/¬tZÅ:£ðN{pž{Šn(‡÷è§à¡«sÝJÅWPJV]/+eäÀxŽÀÏÃû]/>iA¹±Æ¯imwðÈ×÷ ,NïQEüÇôà)ªº^“ÑÜ…ê–T8ö_FMïø•5‹o¨æý5Eîê¯\oÿቮ–¬êh”}ÕèéOh2Éß»V ù±}móAPb’­ø¤ªèöõ‰=a†º?ÉŽ±RBS¯ŽÞZO2´Î‹D¨ÑtÁgJëBè8¥åîäó:gw¿Z8Ýòm /ô ‡j„'±3ôŸ±Ûöðüîû ëiíÛ¾+èüb[|xð tFŽ AgeýR”ü}è²";½$uºùÅBé7 ‡ö“˜÷A[è=ìÀ%–i }þ'ŽY‡LC?ã\äýƒ100ã6Õ¡ö †8‚î\|#VgÞñÛÃè·imEî]ð-Þ¤ÕŸX &â%ç%sb`ªÇ*_”ìÌèUïåå¹sóç >óŽÞ¤"%,ú‹/Õ잇e™kV93°*ë¾ø"a=ŠV¡•¶d…=2 <äùöýK|1É<="±‰Ä¯ƒ³‚nÉá.ñlŽ*ù+HZÍvÜÿ ’¹í1¬˜@òCõNKïCJ¥}u½SóHýŒ{`¯mÒi¨µ>›Aº#’J6ŽÈxÿ˜ÏÇ£ÈÄ ùþ(2×r{¹òYƒ(6¼TB‘ÍÙêK©úd·y/jkœ¾læÕdÈýR„ÌÞÊyÖºˆ%Z ŸDœQÉždä˜\ûíÊ…+-E'ýŽ P”âñ…ùv±¾ù6°E5YOép…£˜q¹£¿*|’[$熒¹a6Ú[Ý(­¬ž³“É€²4KŽ” [(Ç#êÅ@ÈW Zç:?hÄ£¢! ë~ö«¨,ê­î&‚ð À޹´UŸ·“ ¡:½Ó¶˜t+jÏ+^QE­‡}©­²™¨sW.9ÀÖ õB™tYw£Áã‘Ø~q{4z+v¦cðš$|~Oº8¦Ù̱GéÐ,ùYX¯š'œn8‡9j¿Ïò¶ÛÓ0Jëôk¸³.A9]ULÇ:THßÊ·ýz *yOy¿…Êj÷›ƒ³P%}âÝÓ„ªA*éb©[P½~ó2?Ï$Ô²~©9iu—W¼üé6¡As÷‰ÙJhÚ?QvN6 š×³O×H@+Ç~Oýìyèdjsë8#tT>–_øh].ÍJ)=iÐÃã6ik}Ò7}ý%¡ÿ ¥‹u£î¿<¼IrØ¥´FyõˆNêÀØë'_ž³qÙŠ®(Ó§01oM[}t¦Æø“6‰Ä`†ªLß¶¡æ¼3Ù¬`žr';“è),Ì\³.â; Kæ±ýŸã`9£…¦qJVé\è<׆µ`ÿ1EØ£M03‚­#¶jžÜ¯àW*{Îæ´ì\+M+|oŒÄæŒÏh]ä÷‡õÌz$­"ºÓJŒd11¥4H~GôÇK/¤Œ]_|ÚÑ‡Ô "­ÂyìH÷òÙKööËÈpLýÚX< î=x´Šè<2svv´¥­"«’ªD[£²½ÜW`vf9ì|¥?É_Cn/A¢_sÆÈëøØLq}ùEU#.ü@A ß¶íD®R {ÇýE“#Ùˆ´á¡½­÷9£äÒý\ ]#”žÞ»ÜÇ0޲5áÔ'¶TQ.l.î×í+xÄË/¢ •ßú ûªô|}íJĨ¶ ÷và'j,«E°ã@íý6çªx¢QשDçüÔOšùíõ ÍOéN  MNËíÇhzÝ”YsÍ.:ï½ÃŠæ~¿þ}îéÖåÙìFËÇ⼯>õ£Uî})‘·h=äŸÂB†¶{ƒ›Ê‰+ÐÎ"ÁÉï=¢}üºÞçè°zõë¸':Ùœz.n&Î5yôþÚrèÂÕ&‰ñºŒ²PT¾¡Ç.’$Æýhr6l÷I±RŒ~‡õÄOc[éŠÐ™Ã–rõ¶24ú±ûÝPƒêª°ØÓ_Y¡"¬šÅù”>Qôs£†bR3!bžP8µïõÎÒ‚¯\;©Ô¿y”_B"·ÏAΕ¡oeû" ÇÍVØËNò•?ö§™€9%+ùÇ z9I×òÜ ê~ÅQ¦Bó5Îkö»ßA[5_“L|~¬Ÿ­RIã«5ï“@ש÷œl¬¿ »Ü´ùòMwèy¦»ïôE®Æî±oÝÇÑÐÍtõëìÝLèÒ67J „Îë§_¤GB«d‰TBWêYcN†tè¾üLÛ(.zÜmVÊò 7¬éJEn$ôµö‰È°ðCÿñ3sÞdbðUù{2£Àuòm¢¨$…‘Û‘{‚u.ÂCå $äÃÊ}Ϲ9`¢>à{‰Ây˜ÚÐãY}?3î6âe|ðƒü“°‘D=ÌϾuÙƒ¯_¾« Ëhák»@ð—J´ýÚÃzxsàOØâs ~lCð•;×R°<€m&æH${žÞúÅwíc8—÷ Is%ÍwóV#™Ó^ÍÛŸ\j÷u]ФTëöºýü.R'p»ê!fÕ±¦{óÈ@#|ô Û2ÞU~£çêˆLìoš¾Ž§G‰ÈÛxJÐÔ?ÅtIú.šyÌ9q´J¢ùÑ´½½:Óhá|¢³“’ -‹úËŠA« 6aÍ¡hþY'`(mÞ³?~Ö~m»•æÜ©/~|/|–’rþ<Ê5¯D¯rq=ƒò¸Mµëžð±´hȘ” èo:Û_„\rŠwJQG9Oãî¡ìÈðºcr#`Þ§VïÜÔ.Üõc€”ïD^÷!m—¡ã,ù*dÐz§Ÿ †¬$ÓÎ{qQsæüC!g(È“MÊ ‡ºŒ—Žlör¡Øªeá(Yº¥;íñʼÃù.±A¹µ¹Sô•¿¹uÖc*VRCäÿBå3O›&Ò—P˜D jìÇ^|AjÌ µÔLE5D%Pç-!S¾Ð v¥-ü 4í²?n\y š')Ž7fBë®—c-ôAÿòÐEv±Ã:ê/÷ TóæƒÄ?Q,JÐ#!Y^³n }jžÇ<7r¡?pÔºóNï¿<¼F÷àÛyån{âqØÆÞ08Ö1ÅÁøñŒÐ=0±\­Æh/Sß^Ü:ÿ]fv÷˜ŸûSsNgE}†Ÿ ' ‹¯Ý€…>bo²î°¤2á è]Ëq‘b·oÙÂÊNWó”@¬]ÙÒˆwë ñßdß–XI†èá{ð+Š÷®|zìgì%h‰¹&<ÔB’ÖS«nAHz—dÛnLÉlõfJ‘\3¼BÐãRÚñ¾•æhGê´ r]=Ù+ò‹Æ|ÔNv#ãvâ«]¼i"ûùàÍkȲõy·*ÿkd³Miù>~9¨v˜Û ÷¾_§½Ê(‘WšÊp•úòý:Ô…‚ Í=K£(”ñ´¿EmÇâúÈVQ<-Ceæ¢Jf~áVÚFégߟV{¡ì™Ü$fÇ\”“¡s2’÷A…_jœ¿U¯£2Wêö¹¹nÄëQ¬£–ލv/$1v¢ 5. «ùZ¡Ö]5²JÔ) b`™+@½µSe…×çÐP$"Zîä/4V S™Ù劦ôž‚N¡‰Íèã4ÛR”{Ä €æ]™)ºÑb§IÍÓëZÑiuf¡µ0Îô“V´Ñùùò13Úž{B&ôø&Ú%ÉýÞ«°ˆö#ÌÊ £ Ðz´:÷;Þ^}›¤þp°  ‹OXq¾WY” î¶Bó_CÉÕbÝÈö‡.Ü3Ö†GÉÎ1B§`­V‰$4l*=Úð‡êìçKó„óþ‡Œ¥«6”aã­6çú«KZ¢-(,¤ óÙ‰€ü2þçlÏ ·F£Q“»r8ª¾vÔe@öæå*[áG7ôÂ58³JnÈF߀*«<¯„GPïG%ôæ>´ð,§—ʾ…vƒˆÛvtƒÐAµ×çãA;ètySu0Õº¦ÍŸ¨R}‚íÏo-ò ¡W(“^Ž%÷Ÿ±Ç›c6üE&tGI3þ1¡„®ƒ*ùÇÐexx͉¡ËëYð~­èÚé—ÔYîVÒ¬½"*ÐSôîö5sFè9ÒkYØ _ô~Ò%…þ‚E=Ñ*øéãa¾Ç†´#~=Ò‚‘,õÄã§£` ­Z£]ùp³æX©ò2L,tŒë”šÂô~åðJ‹C0scî¢Cüài¦¯ƒ·;^¿ÏÃâÖµmêDX6p“‚UµÂª÷Ó°r`I¾“ ¶è#ž†ý!ðpˆù”n„>ì|ªè,,'Bâs™ïÙ¤pI@,¯L’¾Ñ($³éZm`-Cr`Q%¤Ôf”0áGê7\?];N[}m[f¨˜#ö BÆ;,ýoÄú‰¹É]DÕ™‹é*ÃZøõŠèô•§ÙÈf/8J®…ìN÷êÅÏ"ç5ŸŸ;ö;ë=Ïcbä%¦¢B>EpëÑ<€ü÷íúK:Qpש &;9zS•Õv*EüŽjìAÑs{¤5/¢Øãr²^><´8ýÒÐZ %ŠNo·ö¡4çP•p ÌšJ¸óé|<üÝ\÷ZÙ(Ê/¾æ)óÃ#“Ïò”Äß Ruî»0[G^™=Sw”PÕ-t¹ó‰ ªuï2`X@ ¯©l­§QKh$¶Uœµ×–ŠËo¡noÆ+fá%Ôo¹¿ÝµØ€†Ÿtc.8/¡q¹‘ƒ•Išª©ì #E³(z•ÍY.溠ExêáåghIr)wœìZ?>vá5Z“upFGÆg“~òÇÑVŒ\áí ÚÈÖÕ¦³hïl{÷˯9t¸ß”:}sÙŸ;oN3¬§>Æ;âQáRsîç±ðÜÊ=^+ ´žãêfd&›öxµx–ž¹M“5C{Þý¡zÖ¨ÑÜëM1åìaš´Â¢ðÑ¢‡ï‚‰.ä¿}ñéÇX$ääe¶K+‰A–¢oxÇXÅ_Îmø®9À{÷WöPCªŸP¯Žª-¤„MuàYH-):Ý—ékßÜ_†¬âÞÚ~¿%ÈŠ8íp ÷Z_+7[t¨ ¹°÷ˆå7(õ8(l½|ʣ̕‹ |^¿¢¸ç7TxÛð89YB%µÌ›]\¡2Á€é -Té-0>¤c€ªé¶í ÈŸP=û1ÊŸE”wM2}¹¡ÎõÔö.¯Yh`š_¢\„ÆÙ6j1ã£ÐÜåG×&ŸÖÒÄ{¬ÿå¡›LFCÇ'G?Ûæèòõ«ô³j•™•×1Ç ÏTPËí²ô‡™ªÿ—‡þ?OˆƒQ.ëC§øC4ö¯ÏÆ€–0ËF=lñðrÌøÁ/?aç‹vx¯$jû"1eæ×÷‘$ApÙbgIM‰Oˆ¯A2¶ß´R­‘|×ÜCÞkHÉj`ß|¢©OÖ«{„9"œp#2к&JÅ!csÙ¸æ´;2}xà>kA‚,]û3¯÷!½âX ý.d/yv¿«Y¹j;»¿FÖ!ïžÝ ^jäë r½ª ƒ#gXøQX‡5:ìñŠR~c;ÁŽâQ¾jÂ(é#¦{1¥ ÊU2Þ/£ì¾!¥êòxx@šóŬ*Ü£ŠZÈš@¥'AÅ«eýý7ø¤óQM¤6ÖÞÀÕÿWœ$Ž@Í ÑçË+¨ÃlÐ_4çŒzª+Ñç#UÑÀí©ì&4:åÒ[z[ M´KÂû$£iäíwoýÑ송ϯ14?'yñ’™9Zxžt•¦ª@ËS›íD#håó©¬<9­¯Y_¾Jtm"#Ò{n”£mÊ£–X]´kJ»¨ûí—z²+;åБO¾XDµ ìXWi÷² ó“hÖÓ‘L|"é³Nqšûª*\j­E6Î;ÍŽê°Æµr`êz(t<=;&À 7¾ïiÜÈ„jÃþç+õPÁÌjêøÁ Jú×?ñÇ8ÂÇK^©<éÃŽu%þÃÕé\(bfù÷½?DyCá¹ó–mjP¬ÿƒ×’£„ ®ß]TƒÒÁžÝµ7< l$‡M„Ä Ê¿=˜c¹~ *ª¬Ž¾ø•ü%UmŽåPYRÎæ´¿›Û@oµU?³~GAÂïèãH¢oš•°¤•oPg•T¥ûÜêgº®A]9ÚÐ\U_Ô• Ÿæ˜ÏGÆ ÿËÂ׻ÞwBG§*ó½®èºjäøØz ß<ì }Î#"ë%Ðÿ(R£åÏò¿<ôÕŸ̇Qޝµžtþ0–LêÆAu ÆÕ÷¿"b¿^¼É4Î…©Éò§ŠàûÙ±$Ïd˜S^ ;M& ?‹enu\0„…ªg‰?«a‰é*ÕÇ·¤°ìÇ|j-û¬Œ”ª ÉGÚ{e/Ë7:Ø`wñÜlþ[{ï„¥ººÀ/Gq™¥ø³ô`Œiïm$šÜI;t•I|Fã ”eç3 eîÈR Ø˜Ç=ÕF'8£î!E÷E:÷Ì ¤æ¹ºòiëH54,|‘~’Ïâ϶;2> vtD¦‹gÅ7…î"‹ÿR&O î«oÞØßvÙψ+Ù>¦A.ß4UC×äùº®yêð3ä{³ðìܳHŸ«±5pGaŠ ˜kfLx (ô×j58÷ö™5JŠû¬ZQ[£ÔïGcϯ\A™ê=¾¼/ððmËÁŸ ¶¨ÀÞ³kœß•èJ†}H…B»ï“¦ûã _÷I7ó/È•ßgø^™²]ÂZñÍC¶`цD7AoÇ|T|÷Aq‹5º/ô÷½Oùï“÷BˆhŸƒOž‰{oG'ÁgîÐòOÛ c-i&{g ºlɸ7'4¡{"!üþòèµêt¤äs>µ›;Ív#ÿŒ½»¦>ÌAOHüMþ/åÐ}IG”B7yÕÀ§Dèf=Í–diÝ©V^dkAÏåàx €^÷€þðËMо/êZ¥>|i÷o¨èǵ]Ôç`PöÊôçOt0|ŸÏÞ€Qê”þœ:‹­¶žyI ãV ëÄ—aRãÂçzj˜væwµMh™\:‹'ôðÃRˆ¹¿£؃ŽOŽÂbúémé½f°|¬7ôXÒeX5¡~˜óà#¬ßÐJ{÷6çVëü <|®=]p½ vÎï÷Ûá¢Cbìûa¼IzÍXoI¬"iÈEÖ·ƒ“Hfø*¾ _!¹|ù“¡(g¤´¦r ηEê,.âÏq¤ÓÓäv äD²ÅÙ-×÷Èx­Y˱º ™(÷>Iâ}Ìé‹Üç_ßCÖ³}¡ËCÈfµ:mÒʼnì. cŠÁÈ\~u‰¹«$Ûms‘—JÕbA«ù4Gün¥!ÿÓ–äûî¢ #Y]¥²4 ëÿþð-EbTinªè¢h|áNOk½Î•%»ÜöAɯӷ2_}EiµŠlÖa” ý˜5鋇³¯LmFù¾Žâ‘! GlG%Ç_ÔYW¤P%3=¿Èð-ªÔ˜Œ¿‚jÏ’c7XéQƒû‰û&_ jM‡µ Cm÷ôÖß)L¨+¬ï1¦0„ú»”ÛòåÐ`ŽŒy$†)ß|P@“ĆY;»e4Mï:Ú5‚fATTk÷MÐüh·õvÝs´PaCµ8´äØZ_×–¿:k½¶Ñj8&Tb·+Z×Äâ—ïÁh“¶Ø^G¶EÛÖ.£]àd\ÿ§Y´÷ùêùN(©¼ú.‹ŽNªÂÝ°Çø]Ä'êµ-³ÛOáÞOÊoh˜`ÑåŽ@¤¾2´]»_êF¡Õ!y_(G®@Ù¾÷z,ï¡èmjÅDäuô`)oú>îôYÓþ€]ý4¤›Ëé; KCZÔùóÁÞ"R´ÍØ5É­ùäi UâáÝÔ.H¯ï9 qóÞßu8©~ó!ÏIÆ+ªö"¶E·‰ ÛtjïçzHüo.¬LºŸe·Æe),€ ¬Ñú?ÌrhÓ?*’ÅÃ\‡´¡RIkä95T¶•J¾\ U®þ×ïÇ}ƒª•G5F?¡zx5¹ê ÔL×ÆÞ·ÿ»ŽîîýÆ!¨Ÿè—›sáýûÞ7»)qh΢ýìÖ/û÷½PJ‹®ÿòðëaíýLÐñ…?·b0º‚øtÔ»zw8êìÒå}ÞÏ©,/©Bÿ‹ŠzÅÿòð¼ü»Ó©`”ÝÍÑóW(Œ¥\ŸÔ!«…qí ª¾µÑ0ILÙ»£»SÓTŒãIñð}…˳4ñ)̉e“©öÂÏÄØýƒ Ò°•ñSz wjÇw« ôÛšiÐ+Ÿ¥®ÖVÚݛÄN¾4Ø`¼šþ5¶È¾ox^Ò‡_ê2ä‹$àOÍm½‘GHTmœ¹­·†$ª/~®+¤ã®Á?J=\ý¸û…ûP«N(î L?–ô)ÆWÿÞ*Dªáãù)Gã‘öZ‡¿éMv¤/©Òvª=ˆŒ§·=êÛ-‘IÅxl‹YŽÜ¾ó@Z÷Ýt¢ÉÑbAvîŠé¯ÎÓÈŘZ"£Œ<ÙÞªüúBÈçßýùäö¸îYñtý- 5/Æž¡–Ƨòí_ž@±óæý á (ÑëWQ­MŽRž!!º$(ã-uš'ˆ¼F(Ÿ–è×û(CꞘî‰D•Ç>ú¯ç=QÕUíþ–lT§h^ÊæÞAîßu‚õoQ«eÀ{gö=êôPùž5z_Òüùk5Ñ ––Ê=° ¢ïÓµ[Y ‰…mÄ“È4 >;Öƒf®S;{µ¢9¾L¬Ô{‹Ü'ξt@´$æíònAË !!¹´j¾A$g¦ŠÖ¹&ß8×Ð&aÔ‘L“ m#4𬭣ÝÍÇ{•>õƒ«‡ˆ™ÐÁkƒÝ¢‹ïdh¯Þ©F3§]6£ŒoÕ)uú^O¬¼Ø¶Ë°>_oØÿã&Ô‡ä'ÊlrAU¯è‹kò~ô2õ‡³A Wÿ8ׂO]¨# 1BÑ~³ ȧI“Ëœ…À³(È&9Ó@ÿ™²Šû²G÷CnæÚ¾Ž2(N¡=vªÊ~¾7‡†Wr½T_@+ŸÖd%‡wï ~ÔƒN/ïd OèêÌ7%É€wíݹS½ƒ–r¶9Ð×:µþxûâ¿#õ¥–ùTлË2O. º·žÀgB®\Œy.n Ý^ßN9D@kOCô%BÎÔ^tkæÿûÜ!Ù^jaè ùA_{úð # d\>:w6CˆR2f`¸¡„+CÝFN“ôŽÁXÃv–Ê \V㹓^µ3ßaúK‹€6;Ì´¦¶ºÈæÁo¹9U° ÂÔÃߺ‹‹¯ŤÀò™tÅã#ÏaÕ*ÿšÜäX÷ï µ¸ ›œN/Á/ÿáä3«—`G«^÷•Q  Í¿ÞPY@’ʉt&$½|Tš#Ét(Λw>Bò#¿‹O. ¥cÃ.éSKH]Àn<éôkrîCÒ‘ý3ˆŒ—ß\)Œ4@&R©cŒzÈœ<ÚêA§Š¬§M£xÏ#›ÅàìÊå»Èî1Ÿyž9ÿ|TEî&__"—ÞÁ#Ñ-ùô¥‚~ÇÏ!|°hS; ²Yé³~ÈB¡::þ<÷Qä­¤àóÓßQ4Ïe…ñ·ŠÍ ¬»M£„Îé26”œ"¢¢¢Di ©­b(Ó‘nîðú3~B»—óÉ0ÊŸgê7Z+Ä#Æ^¯¼&Pé@B¹;ª\öŠà“tDU:^—]㉨vu[}–žÕ7Õ=ZÒPó¶W`+oj³ßë‹+UG6ÖÝJKÔ‹¦1Úç…!KŸ'_>A£3ÁcF_iÐDbÁî Ñ4šzÊ¿m¨E3Åî\¢Íz4§ÜwÎä@óþ£'熖Ñ"ÅvðŠÏ*Z^ꊥÏxŒVÚL_¨¼Ž¢5“»CjX7Zõ(ÍrÅ¡MVkÌÓ¿ëmÂî,U¤ ¹Æµ7œŸÐ^ÀHø m"ÚoGgÒEâ­R½=’¨ÇÌM¡Õ‹{%Ox¼Ý? 5ÌU³+ÐzG»ÔÈϪDlØ5×@éƒòo(bý9%Å y›i‡<ÿ^‡1`Q&Ì»;Ë}Uáà û²©Eüßu4Ý÷å ÅaÀçr‚$Ÿ}Á_Åñ÷>âÓÇ6½„iÞtG S¯¸2¶¿ëÂg#·®C‘x¶éü9(1ªo>ƒñÿÉ…å~CYãÁªWÊã6Þ‰ßO„ ·“S¹×Ù•»s•A¥yëŸÅþP9B&°”p ª¼A†ò[ Tm]É~ Õý{e·Èè¡f(J°Nê`¿idž Ô·šØý¢Î:*‹õýú´HHKI‡„J7\ÝÝ"*!ˆb¡€*˜(ˆŠˆ"¢Hˆ¤4JƒtwI# ïs¾ëýó×½æYkfÍÌZ{>{?÷̾)A͇ÃD+Ìð+Îm†¶ÿÔµÞ#³:øó?ž¡œÉ¹c Íý{¬–Ǭ¡5|J”ùÚ?ëœ>Á‹aÐqÍfÖøot%ªIV^¥ÿ‡¾:¤³ú.0È>¹ä=ù†RVO_ ÃANï0F~•ÊTKÆ'CgÀä¬ü…¶¶»0sàB°E`(ÌÞ&Êš<ÃóX*U®ðÁÂ4ýw³>}X2‹I:=ò–K?d»úÂ_Ó{^'ô`moôbñqXß0u{(["Š}Gªã`'úRúfj"ÅÅfW³õ ÃÑ™êߌHš°•˜"CäÎlaQѼ¸G“ކ%7÷ZT_}®Ô1²ó5?÷ɼM;žRôON(~´„ŒªìŸöj› 3ùáæºçȲÎ|Œ8ÜÙägŒòã£É:ÚE¤ ¹jX‘­!ïa•^§,ä7)Øëtœ &¾w¾Ÿ@áGNoœ«,P”ãHQwp8¢–¥Ó›hE kÚèž‹§PÊUýÐE–{(ÃzŒ¤~, e[^9EîÑD¯É_ܨL$ÄûŠþ*ª.æ¥ÔU nìÒI®}Aˆ³ÚÓQK{üdœA‘÷ ñÓ–¨ÇûŒ6¡\ ˜¿q^9††‹# >Ò8÷²E^µšúÜ8"“ÁŽæT·(Ød£y—Vù w8Z¤@É݉—hp $øîY´2øÕ¯Ý5Öì·™wΠõäåä1å´ÉK ã-2FÛûõ¥•hwlÖ7ZíÉ-Æ£$¦ ¶ÆŒGÛ?qéý½‰aV/ö¤õš7Ÿ–ûƒî9þyRK=‡ŸfAãüŃí-¨¢á:ã¹nKÈq–oÔMMÿу鴀>”Éyïí¿¹,CÎæÏBI¹O¿Úá• _?Î42{°AF‚e©P°$dXlÙ㪠Yî¯-Ëœ„‚°ÂP‘;õP64+4“5:‡kk >ËøªÁ|<4-ÝÜüÉW-CònŸZcà·™{P5/´ ÍwwÓCÇ©ß %{ ÓÑlÎ%–úß±#zjäy‘(´SMœºmҢߦGà÷I>‡‡Äð»L5õȉRh³¹‘»t!Ú2Em@£Ô«¸Pè”[5H¨©†®ò½&ÐC¼ñ½Öˆ z;oŸ¶ˆút,V<Û`Гtps†þJökòazê,S‹Þ_¹±&â©äçf¼`jživ€?·]=-…yÙº.Ef°PCµ{šƒ–BgN(‚G—Sù*æ°zÙÕÚÝÖk—]_xhÕSØG vÙ.qèq#ñ>²wšÏYä3ïÓþC2Hv:~¾ð˜%RhøêµyÜFJU•/ð&R Mîã GšB.æå‹«Hg¬£Wب‚ $õc-Èx)8Ù¡™vñ’Ì#K|scÀ2²ºŸ©~nF‰ì?æ^XAÎ#ÛRŸb‘ë1×Áµ{IÈÓœeCíà‰|,~Aæ7,‘ßìf$UáaHÀ˜‡—…xß×å‘@áÆƒðã!(’Í«–óê>ŠÕîÕ¢Œ4ÄÃ{”ëEtvê~À…h*”\`}ææ×ÒÚÁ­Ò(SÙ2Eî*rñÄVr÷PÁ€DsO– *qf.Yl ò_­ó¯)PM]Ìí3âèvfs’/j8™Lõí•FÍrž¿Ô6ž¨ç*@f¯!•&_Ôó•H3ø‰J…?Þ£gÈhÎ 4ÞÑÎûK5ˆ¦Å¬•[I}h.vß_áH š/¼u¤v´Èæ!-}‹–W>¿uÍ­ä_Jf§G«Åßg³NÖ£õç'¢ã-NhsêÈ•ÀChËžŸD¯ü mëÞ¾j´v!GÎQ}uG{©Ä£Ÿ–Ð~È™á, *Û¾?0€!õ¦‚¯¿D}ÃUÉrYd|é™$òæ&º4>ü uw †BÙ/OÖä:(d©75§8¹Ÿõdh¯†¬]ÖpŒ¥„Ì3&—hÆ–!ÝìHôÉøýMüZ»š RZÕ¢Z´!y»Ì¸ïôHf …ߟˆàc¼ORðt$|yQ=kq2Ù—"ÿ§ß‰0È{Ðôºg÷>|oã ´ û'tV#øá¶w <"P¼ÉyVÒJJŽ¥¿ý=G%äËîÎ Pêe@Ú«JRÙT+¥¡ÌŸdâ®Ø7(':úÕ¸Ê[%K…ˆ2¡¢I%¬H!*¥mØØ„¡*»"ðç+ ¨yø.)þÆ&üzXH‘3*uÕeµY^ÿÇó "ËO¡ydjIdŽZŸÕÕÑYÓ@ÛÕ–jnÆHèx@ߤã§]i̵dßüÇCKßí,dó½¦-†>ûμÞ~#f&O탱½£ª±jþ0>µSÓåד¦ó4Wa†Fz½"[ f½=L—'ˆ`>ðUÏÝsXh;öôU€7,©:U \g„åLö²–~ø«å#kônÖˆ2åãuL`ýOfËÔ{ØÚ_ÏD0ÃŽ§;Û¹€\$ œùþt¼ ‰‡X¦^÷CRïM†ëzºHÎ[{Èv$÷Ÿëg¨MŽt‹‡%}Dj;2±)Ù¤:;êÕæƒô^$âœöùÈH£ðÇ0W™êhîñ¿ãF–Ü’çÇ>!ël°âˆ¤rÜœÐ^$è×_*8¦- yUx~QùùÞjX•  ‡{‡¹( ›ÿŠ·¸yE~Þ¨r<ž‰âïY´âtòðHc™•·Å(ÅÄÒä4LÒ¥¡¯‚ô7P6°{ @åÉ9]¹ƒJWo}awAUs2V¡Ï܈2$iåë¨Á ¸ çQókã¼]Œj‡zëL­£®7Ï sëcÔ7+*,#½ˆ†¼H}ó(õò^yMî|¬kh¾†f\bŸÔMz´Œ»›vÉ-˜»Ìï¸ØÚBüÛ´Œº<Å9Kˆª+-‹Û¥h½ÇÇ"Z ­‹© ò¼G›ËÌœÏ ÏC[ÑôÁ©vY´íU£¿-¿í"©‰Mæ˜Ñ^Ó‰G•m CW'Ù›Ÿå¢¹^åCÛÛÈëÑdj¾Kœê9¥¡ñ$Uâµy¨Œ UŒ]‡2×€827a‚.‚ïn\‚’l!ϱ$šHùäÔˆ¼Õ’þÇ_¾9kAñÏ|Cõµ8ÈЕÏW&ä²ô Îê%‰¯EIEÚð”’°ÿùƒBa|PN¡z<<« j¼¬VÍJNCÃQþ”Tžxh>Â>æ®­v»"0¿íÿœ¦í®Q$õƒÐ1c3óãtN挋ûÿF>§]ßÙ¾QÊç(Zm­âßy_¬A›ÀÙÏì3ÐÆúÁoéô2´¥LÇp[ö@»?Wåì'¯5Çù¥ ó{¢b=t5Üüåä}z¬ƒØ? ¥Bß!ÑÇ—ßÁ€R×Ó'’·`ðYÅëR^6Ž„û<~{4 ÁX±G ±ML”o¿+ˆÊ…iÒVcÿs;ð'.˜Ìór ÌkÈeÇ~¹MГÃÐXXº}ìÍûAX9Ie¢É«¾s%FB°žùÀNƒÀC†ƒ½#S°3Í;=ñù­°¿ ŸWA’è»VR,Ì!¯Wÿ%R¨´8ºy)Á?ó2+3Ryi˜ÞkFšòo^®… ÉîÙ+FÈ@ôýjÍGdô=:ê+¬‰Lk7Nõ.E#ËëòCæc‘Õ5Hý͵7Èn–º¤Vꂜ§T¨2GD+ZßOþ÷äéú{B¤[ù8GÅ@~ë–Önø¸NcÎÏŽBÍ+"SQ¸ûù¹ÀWQ¤’Þ¦¥Fņ2ï x™(YZŠ7¤ç¿ m¢ä¦OñÀ JëôK~ìR@™üõiÜQÎEnîn‹4*pÏ},=Šëê»éP¹l\ðÌ¡yTÛ3¡º´Z…˜šû!6ŒÕßF5‘h¢æç‚m}Þߨ½ÿÒ…zöqÔ¹§?Ý5úõ¸3Ÿ°4«£þo<«×Ó…†iûÖï CãG$ß(}CSS”­7mA³ßùÏurxÐ<Ù›Œ¥9-\µºoq¡%óz¿>{zí‡zºeJcˆ¨öƒÕÚvÔ‹óåL¯G†¿ùžÉò‘0·ßg{È©j£s ùÍOr–†üj¼|êl>4 ¹š'¯ÈTe@Ö3¹ÍVêúJ—cIA:óñ—{š¦!Ua\GIÕïŸy—/), 9¶L$eyú1LÈ‹]Ù^Áúð冦yGäcÈ” ¸áœß Ù´)“N…CÞäé7ºx(4¿|÷DÁ‹=\>‘ÅY7‚ƒS˜ äPÇËð Pr}»³B´J‰Î‰zÆ¥CéA¥Y(#µ¦¾=3 e¡UoÈÏ¿!è\e!p‹ÊëMMÌ6­¡â‡_ íGZ¨Žè?ncUqJo”rNCß~Ó}Ãðë3ŸëTÔ}ßz¶ÿ«Ï<¤>€¬‰ô¦×>d ‡là‚Ç÷«È%L9E|y¥d,îÔ#ßJ˜¿ïI˜ùæ°åÂLЉZjQ$ µ¡W¥Å•ÿ^ù‹GLT´öÕs¢dÅ´Ÿaús”öþ"-þé,Ê ô÷KGy‡Ý†Œ‰9Tœ’9Âo€*©¦Aî÷¯ xÿý{$B Õ_}-â¯KAMþü0ßп¨5 ýsbê-ê|KùÆ1:Šz÷«]øï£éãŽ~A4"ÓùžÚ‚Æïš}¼—úÐTÈŒ›anÍžý}µ±Á€æÁþDlÂÁh!qÏEe¨-ºy–·n åÍã)Û¬h%tjC—–­*k/4l¯õÒ›¾ƒæª E’”oÐZW5ðQÚt+»£æÿyÿó’hY&t^l«òÛü]WxúVþ;?5 rÕ»A‡ß»³s„œ¨!dú@—ào¤ÊwŸ€6Ÿ}ê¹çE È&È@†ÚëïIe¥‚Ž|ù»+U¡s ¤¶Yº—+®»'q×/œHú®?žl²¿§^´K®IÂ`ùU¹·Ú20lñÆ#F€ FiøyîÍmÁØDnyx×M˜] ¹£´¦¹OSߦ}r_Ô´-‚y“Já®7…°0œBú–÷+,EI›Ÿ{F+§³äJFFaõL÷e‰5XÿPÁo'¦G~ÁDÁõ/(–ÄØ°ŒBê­œ$‡PxìØ3¯v7i#ò™[ý…bëG–²eñ0®]'ŸOE‰Ønòo3(E®›Âa8„Òz.1 †2i<ìöÝÓ(§]Ó[äÛƒò˽ G6P1#ÞœDåˆûm»…ݨÚÈìØ} ñòÅž'QƒI¹|Ó5o_èÊù9Zýk^õû™QG]xŽu+ƒÿ^sÈA}ßhîÈ€†š±´ö„ë6æ*9­ëyMzÝããVjÐÌÖE•‹~ Íå>óÎÖ‚Æ3ýΙ7h‘©Ç±’½–ìÝQÞºh9ÑκY«Vþ¿Þ¾s@kÒ²´g ¾ôέ ‡ ÕhC®ýÞÖžm‚ü4”_ ¡Í¼÷q÷jD[§›£¯£mµN‹ú­P .ºé0 /Žº#ñ-oN#ÃÑÏÚ§àOo„²@¼"üŠ;_HSxÊè8lu&ÀwÆy+ß•4Èé.p`\‡,qkoÏȨŠù±¡Ú i_ÙìIoÂçÎÚÙÃéƒÂF³§ä É‘{>nïB’­å¥Ý*øÈ¢÷Ɖ¾œ.øê i@|1õ™:dëy>|ùòõé½ò@a3ÝÏg¹°µyß.ëÿæ W9‚ÿÉ…×9æ´ $Þˆ1óÔ](åêròýßw55QPƼ±ÓÄe‘áV7sï@9 ·ÍÀ'(¯öî Xù y¾sç ’ãgbü­2¨ºó¥$¹õÔ8Ý_9ó~yy«RÏ+B]†üñ«þð/õ1Q7ì‡æùL©ÇDyКäoÙÝrÚ¢õ¥ºÍ“ #ùösùËæÐURƬšZð=|Øý<à KPîÙô0”vrqqÙFì~¶Q‡1Æâw*R[0þg1¤Ñ€&{®?à¿z¦§¿Î~¿³ò²·¨ûë`ÞöÃ^­4Xø¼ #‘™K,£¤Ñ鎰ü„Ä]kì;ü=,[UÁwVg'Š<ù{½50Í«~ 6» OpîÂþ¸sÉÞ¯>ö‰o=SKQE’õÕqás_‘,AÆ¿ÿÆ4R„H…í].CÊpè“ü/‘ª>>NœŠií©-Â!=«&]ÇŒ2|¾t¾î­<29þä}qdQ¨ì¶ŒaFÖãïN±pª ûÝ›éÛVx yÀ:0@ y)u$¨_C¾¯ ›(s"#&ù3 ý&¨~튈ìpbßA±æöÏ&»7ñpM3ûÁM?”ô+唿΃Ò,í5]´C(SKwS€å¦~j~p°GÅÓê߯ŠK¡ Õ‹á$Ï&TK¸0™êðÕymíô‡Pãqǽ³¨¥²ñôÁ”.êÉÝý½ªŒº?šI\4âQÿšç…Ìóh(v„Õ7 6i)YÔÑäðÅÈá(4½ýEð9õ#4k=„Å÷GÐü>¦sÐm¡…@ÍîOZBnÌþqãËÖ´ÔrŒ_ù«Ž–5•ÞbCæhe¨Û-¢¼†V?¸ÖƒÊ¢µrqœd Z ½ãÓˈ6¬‚Çç1Tý°‘|FS7¦Ã²µÈÔ+˜O÷Ë ^ºï¤N_€úÈ?lE+ŸáÇK‹¦=ë›ÿöÈü_/E~¢D“›D ä6Dºíþù¯—†ß…B´.2ÛÄŸI&¥Aú½Ï'ïEþÛ—ñÕ¨á¶å¿}寵µ…æÿöR4ZxÜÉ+jü·Oæ·Vä>¦‡ÐVv…*èê1èpêŸ5Ͼ›vñ–Ò$ÐMv§­äFÝ¿c—¼Ž³KÛ-èX<üðFÖ-hÿÂ=°²ÄûoïÌÿ§Ý¦Øù“Æ=4tï³¤Ž¾üwÿîw;Ö»Dг~£Å»«úšXSTì` ÂŽ½¾¸Äëƒ=Ma8ØF¼J+F•zN¦mqƽElÎÎ0¹gäŠe¿-L+nÿŒú¤šòM(¼ƒísÖG˰0¿V@ Koö’\h_'nm¥tXõ¤ÚcþëOVó4#ò¾R3ɸõ8ìDDHPg!Q¹ÖK¾ø0$ñ¶kü“dªãÙά—B2á‡+!ïPj®T©T@ª+ù/¦3c”8 éÌ«ÈÌ~œFú­mD £ç>óƒ…JÈ4óóÅÑÖad‰L¸¯TĈ¬vo$?·× »QÐÆ*EätúfÈõ6Žñë™U䙸v²gù« \º2‘ßùr†4ë |óþó;¨…ä#ìÎ_LAážf{¿M#|ÅéI$& Dñ°£M7Ï+”È¿Õk¦Ó‹RtN²—²FPZÿÄ·1M”y¯AòÒåDÿZÞwåþô—qGÅ[a}_þ(¡²ã‘{OL êó“Ó?Š¥ÕŽ<Ñ$F r¡ì哨é"š{ð2j}¼v¢žQµWÉvÙi7P÷¬ÉÑ¡áèO=Ó ùN ÚÄùD?ð QêuõÙçÇÑäÌ#úó¯‹Ð´|`Ù+ Í÷Y‹}ú€æ9oOÙò …ÙSÛ÷ÖhÑç ©Ã슖'ÛgâBÐr`ß J{´² ÿù‡~ZyL?jGk¾ýIϾ‡¢õ51ËqhÝP·dë†6ž6x$˜¡ÍÉ=ÀiýƒÅc›“PW´¾d!œé ÷y¦X*ÂLBºß©Ñdø™,ö}U,JÓ—.?»:÷_/Íÿ×Û×Ò ][ÃÿöËü_ßÌçÀÌçüB&ð1oéø‡3>4ì7JW­ ‰kJ‘Ì›Ÿ ¹>«óà‹¥ËXÌdºÜuÛ<Ù·ÌÆÃë_ÿ«ïÿë¡*Î Þ/sùß~œ’û¥ƒ¯-™ÿë£QÑyâÃÿïyþÛKóÚ‚rw×ʉòuÎByÙ=·ó/ "uÆý΃f¨¤#÷ÏT*_aÖ…—P£»ûZE$~Ù—Ù~[Ƀº¾B]Hÿ¯4ü Rµ…æ¿Ïì û·§ªí‰{ÞÜWèÈÑ8²Ü ]5/©ò¼ þãá©€Ëë\™0ÈLfv.¶†Ò{ý¸æ²`Äñé“öŸ0Æz¨åþ!€ñ¹3ŽFÒ‡a²-ú‰Ê˜î¾tšÛ]f9c¸XeÁ¼¶ï>Xx¡Ê`Ú K$Éüæ7`ù–ßÛa_qøËÏö8Ä%VÇÈH ‹2`½j€Kl¤6KÌe„†.ÃvŸ²éÏÁ$¢ï¬Û´¾ˆÄÖ#tºo‘¤ä¯ÅÆ­I$sç}ߊeOAY‡”†ç ·Îd¸•EÇ|™“GŒ93‘yÕ.½lÅYY¤ýS#+ýù{ãsæGð@pNÑ´ÑòøÙ“´Oo­Òø%áqÐÙ š±¸k ÝÄ1r­3mÐs޼þ6Ç1èSèêÐÞÅS1•Ìû`0|~N&a˜ÆÖ7«lF²cDœãaìF …`»;Lœ›L¾#°¦¢ßž /Q„?¼wéXæ`n^ú[ðuZXµ{:ÎK"þ|£Ü&°¼Ì´nØýþŽ+Æ>ñ„uZþ†gÙ`ó”OzÑ–,l/ž°¹ŠDÞéìÉG„²-9Âä;’6ðÞè1GòlßhÅÂÜ“ïóeªqï@SaŸlÒˆkFÝ9OŒûÞù$œ´õDú·Ì×^”[!ãªÕ(úqdºó6ã'<²ß]ºG÷µuÊpM [ãR¼žc4r ž+:œOI˜3þýå×ïJÁÌ>M—hÛ¨Y}õ* ¡TÎÝ`TÞò ¿›‰Xôöª» éZ૬Áü¡‡ô¥¾ÁäZaß~mÀ§¤ÎK©º<¦¦nIù’ê…Rž…D«H á¼H£*É׃ä}Í~îà©O· _ð‡€ÎEf(üRA÷œ7Šé*Êèö4@ñæz9½x(Q<źøJü؈©(CÉlâ£`(uzãQæ ¥õ.áKÙ  Óò§AYêÆ3 ÉPnY˜Âõ”*ÕÌ^«¿ƒ7~ùû—)@e“z7¯Ÿ TGFtU샟¡C‡¯ÿ‚Ú|"ÿoFuÐ FS2ûl§Â&¬4— ÙåÂðÕhÝç¢ñ»û/ü^¶>a7` íË›úR9¿ìiÌ+øz¨zÆþÏCßý3’ßuîÂ@Ü‹^ËÛ0$Í®r9—†[çøëN†Ãèí„ì°ˆwðžjÑ£ƒIuº‘åèY˜Ör¿”¸[‚¢}å`nðÄ…Íp}X€Æo³°˜@~ÔLG–yí²4ßtÂJ^áZáAjXõ“Žm8ëæOïYü…MW+e Øvú²Ñû‰gs¬L+‘xÿZìŽa’Üdlp{ˆd|¤&£ŽH>/F¹Il„{fŠFÄXê‘êÀ£ÌŠ[‘Hñ$´¿¢éŒ*©ì"W‘áбqC™Hd¬ ~Nþv™_¨­O¶àþS›_BYÈÎ!WVhú9'ûû.)Í"¥[€pøòÉq›Ê7«¢=û^+îk(ÄúÕUsÆZ üæþ½ˆ¢]¤”¬ƒxèS¾pÉJ¤ÌRðY ”W±OÃ#”áäÖëŠÜE9ŽØ Tð›-òzTšŸp°z˃ªÇ";ˆÏ%#”?wøMªçyÎ\×EMãþs?¨¬Pks#šýõeÔùÒ g´ÖŒz¶3<Í÷ Qûæ^›Áj4 Û÷ˆF£©lg©ïV}^üv£+MÏÜøK*¢‰fOjCJÐÜFI”’ËÍÛï…»\æ@ EâmÑ¢ìšä•¢¥ø}!æ²-´ ÏT;Ñ…–cžNxH-DL¸V¡[EF´IxSÛÝ!Ïö3ðª»OLé6 #J0¿ídó×OêØ¤3ï•ÕCEü¡ì£Y PJtñËÆ“.(èàÝ ^€¼ÃÙ Þ3ŸúÎ{£þ/dÓ¦¤ÖYÀ׎}[Ž¿ªlߨ:AŸø§ûæ"¤FSQ›ÜÉý_o0oW䳓x,‡²âwd¢—R ÆéQÙy-h¸«ÔqçÄ~hiáòo¼­w8Œ—ª º=¶ÜÚ“ÕÏÖõ@'Ô]Ø”]zëö×ɠ˪´ÖП:ãWO¥ùAÇ]ÞSyÁo¡ý¢㞇zÐöý Ï(…4´½Tö»tJÚéhÿ¼e²ö,r‹„_)ÐáûøTì¨tzî# †®»ÌQŸ¡{ä^ÍÛ£}Ðûæ Űéè/Q h †A‘lº JIzi÷-^Ó Fdœe2|at~fDÊì Œ·Ïv¸WÁä²óíßaæœM#]²;̉çª6ôIÁë·®Kœ°˜9©aç}–¯Ô´Z=€¿î^ Œä°vÛdt¼è l´¾È—¤†mëÁ:*HÄ„U ©Hœh"¹<‰¤®GMŽÞ>ƒäªªôvݸGjô¬Ù;{Ük01Ø-™†ÔwN_ðEÚEfm¿qw¤×Šä¿kφ ‰{Þlz†Lü~¿”OßEæï¾iúæ*¸ßîFÞ)Jd³=m¸çË2rëcø…BâÔ&®w#wÑñÏËUjÈû IòâÍ(䫾WÊ(‚<[JJ¡`¬ÇrÓ…26ôV9»0ˆ""#dG=QLa>F¼ù½ªÁÆGrwÛ}šÈPòÞÚ5Û;5(UNÆÇa¦‚2<³2a (›äs5‚iåMUîDEŠù\¥/¼¨Ôë¼dÿç,ªR+þá§@xGÏÑm‰ê‰ŽÇzåPã/×KN-%ÔRz\ló,µC§¯Î|ŽD "SÆg¨÷œ[þ[„$¸$Ú­ÛI£‘ÂPÄ6K4Þ8!i{jM}Ôèo øNEÞT>Ë—D¡ØÖe„¶á×ÒDu AÉ¡·N_ÛAI¼qo¯Ÿ#”ÒÈ•*SÿÏ?ÿêï‚Ò½ Ouf¡Ì`éúÓ»CP6´0ûL!Ž)¦¯øBÅòã0êf¨”>Þ"²I U]¼LõÎg &ƒ„Lio/üª¬‹´¿õ©ç_\KƒÆãª¢‘ÐôÓìï!·úˆÏûL~€ßZ‰ÇW— >7äšëÙ—ÐyXááí¾yèVJÿªöìô¦Ÿöq†ñÃÞâ{”`ð« Q]H$ ›Ï™ÁÈß/C`쇷ÿ±c 0ññž|jpL}²‘xi3ýË­gvaNÙ7i?e ̺™žÐ6E‘¦´°T¨Xõ®î6¬Xœt^ÿã«®JÖa­Så lüމ:[#Ótbmº°+.Úœ¶œˆDIvÌØDKå|‚ž4’öð Ÿ’Aò·¥ysëÕ¸'b_]15îýò)\-Ê ièMn–DÜWPMßê¾Gˆ#Ìrýl Š…Ù‘YIPC½ï8îG/©‰ü÷ÈöºZ‰Ú„9­÷loü‹Ü6ÖoJsÉ·væ«\Iò¿²8U3A…‚Ï­™…èQøgmÓ×b50yçÕr 1PêÙA ª§Çª>Ù¢KeΑJ”.c•2œ@ÙüÄæ:§Û¨@ƒòåí7PÉÿWÑ:#ªtI7·å]AP®Þv£Dõ ÚÒÜ Ô¨I¼uã˜Xj¥Ê6ÄK:¢ŽcÙäDäꑾ”á¡ïBý”?yÆ’Œh(ו"ö×âZó¿h«£ñÜïÏçÑT"öLÎ4sºõ’nEÍÕvôÎHGóòÑ™Ùóh¡Á™Ã£u-¾µÖ™It£%¿,bZÞŽX¹Ws-‡ÍêLÞF+ÅÖÝ0“^´ºsÓ.èÕ0Þеš¤":…m§î…d% ÓËYö9ß/0Ÿæ@üªàÔû|NÔ*„_‰’<^‡’ûÕòRO¡€ž+X©frÏûKãó'äBÑÙ5 †oÇ•^V„vüª©PÞ“4H¯h+Ü¡ˆƒÏ ®ŠnùÝðÙ8“Q4x/d$ØèÏr=„¼çÏO‰w÷@™?9O{£=T/Ç>`Pú95k h ¾qPCZEúó/ÀïôÇïOÔC»ñ|Í¥è¸{æÓ;?èœÌdw¬ÚKðƒNUt—¡ó¨¹³{ÌtX±7ÚôÒnS"éàm)c Ô2Ðúúй.Ah›þ&›Íä í×h?+ËxÈG©Ú*N D‹v¬æ«ÐEO¶d–}ºh+ò_º@/¹«œ±½1ô3ïâ髃0P(ùÚž†„í7"J{`8÷)¨‹Óµª‹0ŽËTs)!0i1š2Å ÓE÷ºäì‡Z/ØÂü'¦¯¦Ž°¨vÔ¥&÷,m§ÙóøÃÊè«üÐÈ÷°º{uKÙ¦6´ËËß ÁV¶mD’ëìú~gò Eb­V»æ×,H²ÅÖV8Žde” ㋜Hññ¸–˾I¤üJDÞq¨ ©†#­jÍú‘V3äþU7¤«gœëÌE2]÷\-d,4IùÚ…Ì–¶¥ /Íq?¹¾¦DÖî¾…jbKdo;Ö]Aû9çã¬ní©Cîƒ :jª®È‹6S[m7ïÚÚsÊmäÿ­ýMÞÕÍ[µn\0E¡UúM—D‘¢Bu¢àî¾'¨»–¬ò½ U~øÔ¹Ì#ØY>JÚ@u¿ñç5JYï=³E-â×>/íPÛ¨ï[[`êwV‹õ žcöÔ9 4^r•TAÃmæIû:4.5Vׯ^ESEûd3" 4‹ñwf'As¿Ë»æ$hÁò3¤ìÒQ´øÔòÌZ° -ýÎußAË‚ÏÏ÷Éú •Ü‚™E|Z%9sj5½‘LNr?ZŸW«ªr´EëŸ×¨©%ððvzùÞ÷hãÑÃúž~ ‰Xjš>jn×) Ó ½¯fõ²L ºË®+‡êè›5·I(£ÅÐ ^ÈÛ/­áÇîYKÇ"¾_XýgÝß­²/H:üHJ4>æNU} )DœØ å! šÈ?©Ï88QºþC>P©µÊMAâ@vDëk ø|JBùÂ_ È8}ƒê³´õ?ëd$1»Y@ß\ùúsøîf1bç …•æì· (ãCΘ34ßøì‘ Å%úc*PbÓx‹å"Ûmã46'¡”«·,Aú5”Þeé;}JGg¾ó€²‹Â24ÝüPnÜÏ÷#¡ÂÃ.2vH~< ø%ü ªÒÍžoÔ@u/_ׯ“9ø9ÌÇçí¦ u‚Ô¼š' ~9nDoçíQh’=L9ÀD Í?o¾Ü>ë­QkGŸ”>ƒ6_ží€OנÙ¢þÏK-èò ~$¤ =mf‰šÏä ßçÖ"Õv”¬Òß¡ Gk`Äíéµk¶R0&é½gØ·&˜Hîº{*À“áƒ\én˜Ñx¸ÿâ·0ûäz·d˜Ì?XkC*X+®2Qm%ön¢-YXž¤¬’ÿš£#Z;R®Ášë¶k,ŠÀ†Caêlyjsh¼;i#³4¯n#Ñk[†++$!s"öiCÒç*É_¨‘Üâ¨×È¢.Áß:íóÔǽêÕnÙï!õã=Á1)¸Oüçþƒ]ï‘þÁ¬IÉkdä9´uõ“12uQÝ nž@–ÖÛû(Ìž CÚQ£`ä(’äºÍ²ƒ\ñZåŽÃ\È{ª€M6íò Þ+QÏx‚û—«g¡°æåÂ7¢Hî âºGm(îçøL&‡xèÕÜ“sDÉåÓ‡;(}šžuЬeÒ=’¾õ]„„öG9T"Y ³¤@ûֺϤº¨–)û IÜÕÅy.:Ö4¢F{Ô‚7j9Û+V¡…:TÙáa¢$¨›®«ëý õížÖ¼îˆ£Ù"³¶Êhd½´_¬öЍà±9‰&‹¹ÃÞã2h&Òà—‰æÂånmœ»hž®±¼Ç£-$wÏt¯¢ÅÇÇ?,£%[ïºcûC´ ñIý8(‚–ŠIüwŒ <|Óµ—ª­îï™OáÅ}+)ƨoœ,Ó¯L·nG(m À¼«M*ý…_qW­#æ¡üÓž˜OçL „å@¡çÓ>È×gße¦…œoª.Praæù•tJÈú\~¥[Õ2#bÃžŠ“@:³ïamøyóæÜb|"u`'÷; ¬‡Ò \Œ º>ä*±ñʤO«Ü>DÕ÷?û˜ö³B½O»Ýphji̓–Ÿï7+ƒ<á·ŽHÑ>h[x;g´ ã¦–CÍy=è|àÕ¥r·š 7K[}µ “çÝÓÐÁͤ²`øÚ´Øžý'^—r±€¶sÛCÞæŒÐVíXËtÚµ÷S4Ä~„ö¦`*ß&èxÙöñ£?áxÏ•ši9í¡«áqqkÇ0ôœ<ùÍu ú<8´ìù´aÀEÝÂ<™‘à30Hà † îÆÚ샑†F*Cô‡±¤uUYE˜H»ë°»¦i,IÙ÷ÁŸør#ß70çË‚µÁBUMˆ*y!,ËÝw*õ€•S?ÖèÔaõjcøyXÏš­c¹˜[l† ×Ë`§òÊé[$´’¸¤¥‰$a?G¬VN"™^+eqø1¤àýÍæ{Í)°Šþ%G*Ì‘=[lŠ4O\Cßì=‰t,ûzH5~„èOõ2ÚóïpìF¦ ÕVÛ¡SÈò4¤ùǼ.²ºZ¿¼k•ŽìælßÚ#§KÜi®Èbäz¬Þ–†<£ßŒh¬‡OÊÿ‘9 òß_ýLrÉÔ£¤Îw¡Ðûì‚[ ^xÐ{ÆV¸Š:÷S$”…¡xõ‹$úOx8'°)ÜQ%É¢*¾ ”šãõ\Y”Žá?uÁe…‚„ÄØQ®®1Á‡Páþ±Ççö ’ÓããÜᬨL;Ý…jktÏH.½Gu½ ¿É>S F8CÃÔ¬IâìeëGmƦÅ£ƒ¨öUS³wõøÍÞˆ±¢þ4ߣš½ÚhX¦ÛóRïÊ›«¦BSÒKê¬=~hæm¬û!öš»”_;µö-Hî(] ¦@‹W[4GÅŒÑRŒ(–ÏL-ÓÂYàÅ%´:ÄØX}­â2n„ N¢5U&-›ñZû¼ÙÒzGØ®Îsãö›#ððÆæ`H1Úœ¾ê"”0þ¯ßÜ:¨þ5æøie¾ä#½íÇÀv±0e¸Æ#öQªF§dÞûçAqXÐÃä^*ÈÕôQ~½} ²xb¼ïAFx%Ùú‹Öû½øvñߌYe¦#èRTãqu$du׎Íh*BîÛ0¶1ÕN((yÑi1 ùžý(6î"š©E}õPtóÙ‰ ä–¸8üˆ›JöZ‹˜Ô²@‰ÇUš¾7ýPòãõñkn” ”rˆ„{¸8¦9œc e¤WB¯'BY»$U—ó>(ïâz-±(NȉÏË•ÉNC¥ú×ha¨Þ«‘ñ›jV3.ë^ …Z>*³CF¨3¿•’o¨8Ÿ”t|ó“J…·Ð2dh#•)¿S‹¯?™p‚öÈ6zk«h‚Žíú²ú¡{KÕ¼úBSikUaà[íƒ †30¤¶³bLÑ Ã¿iœ¯ä݄ѯhï°l¿ØÅcÞ “ªÏÕk=`úýÝÓQ×àÏÌC:#â)˜—~l¨T wì+¾„Å¿³.bC°òåÈÅ2ø+\ä{ÜÔ V§2êÊkÀzËÖt·Æ8lNÇ4Üÿ;²o÷žVwG"g²_¿ì"øGÈcSª$µ´û¼:ŒäÄ œSåHÑ’šö•! )›GQ"m©J¯UïÙ´1µèvý‘^%'dóq62T“¦IK† SÀ|SQÕd9ýnJ`ä&²¾û#”d0"öjtdx`Nz|mŸ<ò¬3nx¨L!ß—2Μ@aˆ~ð<Ú…ò¹3úÎC‘ƒC'$ -P¬;µ6ô±.ßT0,vEI¯µ´Ñ? (MO¹»Pôe6r牌Q^¾ÿà€IT|°úàYÊ:*·ûUß0Ùj¢o9ø:«VNËõB3ò w¼CQ‹î£æW{1ÔÎM‘à+å@]Ç¥£¹<ë¨Oþƒ]±6 ž=NÊÈfE#*†óIÖìhìûë,cš|kuàÕý„¦KœÒçÖМîn¨§)š¿~s°k-x‡ù_$v¢Åë°‡¾ö¡%ÝáÁßh`äV³Ö€–½¼·v…˜<ôÌ ¯úˆVáï };Ž×Ÿ:QÛ&{ ní«Êñ½}ÈäÃCzïÌ3YØëÚ|'»ÒßP.qyŸµq<ÃÐ=+¼ý!ñNœ½¶{þrY!*//ñFÁ×-›…¥ Șš=[f _ªÆ÷ä>h†‰÷ub_àc> Ï O ¤ß'c»ûž ë§¥N²ì¥Óuv²w2¡š·@³ðó"Ô 8P)VAÓ•ý›üO¡åŒg‹D{´®¯¥›ÊT@[´…ĽØèØÑ×`ƒÎÃ1†w ÓðÅ÷—ÞÐQIâìuiÚ'i’JB;GÛ±wË7 íΣÃv]ÿé±xá^ImHˆ˜.@; ·«E2´ßXù£žý:¤ó·Tâ_A'÷ÙôŸ!G K‡%ŽÍº›Ç£¸½_AožœžnM"ô¯Qì߃êÇB½JÊ`èã{7öÄAXÛœ£Ûñ±—š‡ ZqÒËLaêX1ãVÓI˜éÓo.Ù‰ƒ¹÷qï]ì=aÁR)î‹å,‘L»¹´†åáðw¡A¤?ô¬ ±·Ý‰‰„͈X¹ ;û/±.}D¢ëÍ{X¦Î! iZÓ¡~$­‹ë|.ü É_ÿ ™¢Â=aGLš…$qoL Ñ"¤¡Ìxu÷9Rˆ|9ôéon?SqŽC†ÅuÖú]ç뿊,yÆòî¸짆~â>dûem# Ãq×sÙ3ðÀFæåzáä1.*áÞ*CÞѰç÷‚+ó½ÏFœÏƒçÝQÈÜ1ðHàAãùµÃ(²Ô¦jõ¶ÅÉŸL:ÐÌáaù¼ú˜wÅ(q«¸Ñò®:JnÄRKΤ£ôQúš­‰(3ºy%fåEIð´–¢‚1ýν†Tâc$>vU×ά­@µÄªv? cTgÝaxÁ>‡vÔ^wסæãŒŸ¼]wP«ó‡D ÇÔ1 ÿîГ‰ºãƒËÔ¨Ÿî,&9dІwü)(:ËÐØŒ×åðcn4©hÈC3ž#Ji"hnD¿^þNÍç´ÝûŠ÷o9IT¢%çý²Q´|ïýäìV ZñËZâQB>|A:þÃð,Z“³]]8€Ö^D’Åãh]¹ [È@àaîîj7Úx3''+ÎãU¥ƒª×zLP}Ào@[ÁéÕèÕã]˜a2ëþ^G1¨’s`0}xŠ”×·û 9g“eÊ]á«Q·ŽR'¤ï&2énÿ€ÔžØ£ô"òA™rõêH>Ê9TT'žŸúEæ¯ ò÷v§ÙNÀû’iŸLfÂï'N(L½†O¿·¥Ù ½¢éê§ZÈ:?#\Áù<^•!PÀÙßpú5|¿Ö}>!O =%œŸ»ß…ÂÞΟ¦“;„ó²ÿûΖŠýÓb¨ÿŒ|3»túÄ1(ñqX³îà%p1–ûBZ !/ö 6„Ò' ¶øE¡Ì5diŽ©‡“ÊHD Â&«@ó7üÈn¬— %\·ëÝ df¨¡Ñ¨á»^¿¸÷GKV@×÷bÛüZhÜ3~Dïl*4ÝNX"ê„©×æ ñmÐ:í´›*¤me, h§σClÚ¡ÐÍ/ÃY=2½1m¢w ߀¥ykg fü ÃÚ¡š´À(…—àÒ«06º"Þ¡ÃGâkÜašüÖ÷ Ýyø££åvÝæ2u%®&äCÑ9#Y'X|ÒC\|4–Ù¬ž¥«ÛÃJöêÁE{X½è®4èëö”/Ëaóì²D~E-lçü1¦‹D|_ÃòýBâÛïùøg¸‘”öÉ”F>’eÞ­­Ö@ŠËÛúùeHéF–\TTQ,n"-u=ßš )Òµ{úœBß.×éo_‘‰³}»` ™w +5ü/ «`pÌŠ¾ ²G…8>_܇Tùiž(™#HZ…;=ò)n\9¢…ûå­ßB!©ªDn<øüPÙòZSß%ýí­Ž‡y¿| 1pGI–ì[AÆçP*‡x|ýê”yUºóZ9 åj%xu_£"óÖkbÒ?¨lËVn=†ªï¨“/åª#*}šÎËA >«l®S¨Y2|Me†µÝc›ŠŽJ£îÞ·>dIwQï«dÇ<èø_fYFÃb‰:r4øÔü½QM®Î5¶& ¢i¡ûNña4›ÑÑ—SAó»Ǿ8\E úÊû4´ˆèýEȃRu „ûjyI”kŠÞ-»(]ºË: <”Ž }„VL,š¯ðÚ ¥ƒ«[ ¨£÷mô•ì5d²ÍñoØOs÷_í¥cõýg½ê1žãŒPöu(d[tЬ ßrkY…o6@ƒéÇt_6qÝO}©-tG“ÞH˜CkDŠLßI%hW v©–‡ŽÊ™]{'¡ó{Õo è\’ʳ¶» ÞVgÏYAG9ãæ.hÿsÎC°4Ú={| Ý ]£ÔÀVaÚï$S­3zC{+цKtH8W^Ûr†ŽºHƒ†:'茋yþh”º ßäRYNÂO¹¤’GžJлÛHWÊúEC¹?Ÿ„†Ó5'”$aˆÿBmèz2 §e‹}£^¤Ï<aܱᣋÀL&=kY{œ3û¿Ÿ¢·®€¹Z«Ò2/XèIüÊý–F)9ïR•ó‚7úinšãJ°¯Àz¨ZÿïòK°Ñ#ñ‚+_þ\rUößB¢F}yAï8Iñ–Á+$5uJºRÒ»öT”Y†#ùœºÛžSH¹N¶šqwK¦UT CÚè?¦¸—gC1.[é~ˆm7##åï‹È”µt°Ùû²8ã$åº3²ª75WŽ ›RdˆIt ¸ÀGÇþ9³«¥Äc£‘ûí-ï~ò7ȳÊáýZm:í r]F¾?$ a)+ÈŸ¹B˜+5¸“øJ `@‹<-Õ1þô«.iÜE¶O±JZB±¼FÒaž£(±Ü>ÔºÅzsìCöÇôâ"‘ó„^ íš4r—K4ÇÑ© ohû@#E$ò%º¾µ\ÃÃä×yWK5ðH¼‘¿º ûØ24´8¡¨ßDÚK?³ÑH,ñÉ×ÑJßg@©s¾¬ß8P&‚8Åùô ʵ~ø„Ьץ«n'"$Û^¹¦‚Jí®ß²SŒPÅãS…ïoTc ò8µ¨‹ê¹ú~Æ¢¬¨ióîÆ!9ÔZÕŠ“õlBçQ·¿ {w@ê+îÖ|¼ë<ʪ3˜ªAòCwYŽu¡‘§Åó;-hL”È@ýïIëQª ñÖϼêµKhr…¼Z€‹í Çóm"ÑTfË!DÙMƒÂŠ5´ãÍwLjoŒ(Pµ¬‰Lÿ,+2*Îï¾ssA6–ÑPãh“MVe²†Ëñº„¹P2ž]¸òöÜ¥^’O~2½Ï§¸PwcŽ8²ÂÏ1pÙ®CFÔÓ¸ƒ BÎýÒ“Ú.Ù¾õƒ„Ã&GßœÖùû\§QØò‚¦O(ZDA©uh`ÒÀ7½ÎÙp­D¨¯¬ø!Çq¿‰ïé%]€æô¥»$Ïê¡Uõì×+·‹¡­?3ƒüûì<ñB§kèŽï» óMsÖÁó|—söŠ~tÜÞ¹]±iíý3ϲ?Bû©§1ãN”8Hv&ó'´[›Ó1>„öw¢,Ï;ã¡}hø}ªŠ2DC©½ÌjÛ¾(ýãV'Ó‹s(ûÁêVÕʧ4¸Þ•}ƒÇÉdOŸé~‡è~„Ùgú.*ÍìèzWU£ Ï7þg:PÕ.9ѧ€Õ:0ãI7j\Ý_`6¸ŒZòÝçˆù¦CßkÖ{*u{VüNª¡þ…oînÕhÐ]ñç\³q%¯ø-=B£êÏç_EÒ¢±½§›½|1oGòÜMRE“§‚ÏTÝIqnsD2Æ|‘üVEÜUøî–"­"Rê¶/ßç~DM:Ê#ìy÷P+Žïàf[êq¾`ºuÿ žÒA½þ/Ç/2¡ñöë¾âhø2ýˈÄu4:—UnNàZ£…@ÕïÅVh|-05êöi4^~5K”†&v#‚X‰&-l!Û—yX¥¶oûš†pšß<È^‘óçäcÚQåøÌA¢Ç¹Èxdâ‘5˜3ñÏû ¼Õi¤¾S‚’PZüù’ÁÈ(öîÍy|´òhH¾*ªCN…V“oŽ, [}¦ÝBæÔÔ‚4›ÿr©þì ƒ¤¸ã'Mµ3!ž–"•µÙâz7$=MßBj%Sä@ûÈÛëg®äi¥û~'áÛB¨/§f¹õ²7•ïñ‚Æ+½WŸxD@³qÞI§q'N0jlÇÊ@Ûýs êVëÐA%ʨ”ñ :ô{sŸ—C§²KÊ„utLÒžu•7€íõ)›Õ¿ë ã­Š Ýä±|-Ó hç¸_xòp ´Ëÿãc$_q;¤LÈ×§<‡ý) ¹aS/ºì³¸hav_Ëöƒê0˜fÜ¢öˆƒÅ@šµÚœ°|î‰AU¬ž1rêÐÉ‚_žo—ú™à÷¶.‹°EnU ÛiïÙW¿N#Ñ÷4­$9KªÇ±z I·sœ³$ŠqWÉÕ‚šýH‘a’»î³©Ú¶Èfô­‘Fl†F7ñÒ}gYùDàÄò[ç{_Glãô '\Mç=ú)Ë‘ùºìã÷¸O¾f·ø´îçŽüú(ç²ËP@ÂCä¸ó“Ëbår“ú8]!jFž³iÄw)#oÓbý¢ënä³Ð/J}⃇8þÌ,Yãáô¢^…<aüó*q1 å\­¬q¿‰G—Ö‰æQLBjUó ?Š/ÆUéÚà1žw·møQ²±ýÇ—Ö($®ÖµÐв6+b‰5()oQôÊ=TÌZˆhJPGdÄÀØof¨”÷[a-u•'›Zg¸n£ªHĘ¿ª½ß/Ýù5d³.ÓC-¢}[ö—P»c·Vø1Ôé{AöÙõE‡" dÑ éõÈêðQývë'ôÙ§›t‹ÐØòýÄe%4^8¢ñ„&¾ÔtD/Д,¬=ff/šúm²ùn¡éоûÛÞëhvöàwÉ| 4+­}íEÈC«3¢hî2yWt•=èJí}ÒFPÁ~*k:z÷ <æ¤$݆ñEü¶àÜ•“ «×¡(‘IŒ¿Zr–OrÉœ€ŒÆ·mžSM–aáÈÆDè·Vu*f=H¾•[iBè·•ÄÜ‹ôyð*µ~Doö´1#D5χEϘBLâŽú>Hnä¥x& éá!òF*[ÿ·~C÷S+C»mì´û àhN¤¦ 5¦xšm«AÑ??¹ÿä@Ñ|qYêm/º‰vÿI‚Ïi•7OÏï…/ ¿µ ¿‡/5GƒŠ>4ÀW [‹h B~ýšLà&”¨ =>O“ý÷{õô£ºP'Uë%qÊnìŽ9½OÊ_Þ¹éO¥4e:P¥½T‘»ÕÌt/G: œHì~jÈ êŸlÕW,XÀ匙’rhâ VMôȇæØãw×NBë FæyÅBß'øp/_…ÎJ=†Ó¡{ÐØ»ðáèõ¼®œI ¸ÎëHõÁ`ðôôE*Ö´o|Ú0Ê{އdù Œ³“Už—x“Bž¹´†`Ú¨²?g…fß”–Ij߄ڠn­Ã°èè«Ð3Kýªu—^ÊWŸK/¬ÑÞ»mÕ¿Æc_D€ßëŒt²M`KóÙµŸc°ÝâkG×ø‰j©£‚v# T?IãRAÒ ÃþÔúBÜuz†É!÷R°ŸˆcœG*¦D·?9•¸ÛªõXFp#Ò®æ‡7rÞÆ½NÖ7YÕ ~ûñRá›·Èø…þÔ[YedN‘ÞîWÍÄ}ëEŇ‘wþæ…Âwx ï!¯uÆ]äJ×d<¤òyBY5)½Æð`XÙÝÞ±`<4º{¸HV\—¿ý¾‹BÒÕsFã×PD¨ëa«Š%Ï|Øo‹7{X:QòUui>JgìnP¥«FÙ¦íìv3T 00uÜ}þ¸z˜ö*)%Mõϲ£rk°@[Â7Tuõ*á¹…ê{â§@5Š®v1ìj@-³è¯RÏP»ïIÿ›«µ¨kÙ6‘H…ze*’‹ÏÐà€Î…+ß áE¹Í,q42 û%På‡Fn¿Ÿ^‘@c‡qÚÌO4ž}FO½†&çW“¥3š4~ýåöN˜‡/8Ë”ÐôUÌ\Ê”^‹©õùù+• (”³sL‘ѯ+­È fsýè&òráÛb¼Þ•Pª,Øù«tŠ‚–ýù¹6”÷®>4ýû\eñY% ÏM åB¦¼Çw½fJBšÕd!@¢Ú).qÿ-ˆMøCm±!Ñ Æ:S’Tîrë4ä>g6Q !Ì‹Mg*g×ì Êk-ÌkÄê’EÃô‰³¡‘åq‡Ì¨4-{‹µ)@˳ªï5YÐv„AìÞB´'M[ûùû\4"Ôøîÿ½/L=e‘—κóWCÓ’ùVÕý¿ët?W ü€v¥žÉåÐN9æjö>Ú÷}è—g| íÂXW˜íªMûÚ¡ý•žÆ}õ~èPP*a|ì‡N°¸A×åݲ`)øIeIì$=íÇ&S\^@ßD©˜² ¼û¸I=·ÿPLú¿~ ÃŒG”šaôÞ§×gì>Áø«5–ᬘ¬·^ ñ‡xq`<,æ~­ÈÏÓ-À"}Y“ïônX(”“ó7•ö[w 1ÀÚhå–‘ü4ü&¡{F›~d'­ÂŸÉ·urtjHd3Ö[Eý‰G~X}þI¯ +—}ÐÃ]GžôV1w#Í×ogn"tìNßiÜ}}ø„`ó,Òþ¸Xs:÷Zûè¹Õ"}[‘ئa-2Ú¼ã¸h0ŒL[´ç"Ë'uÍÁÛ²Èú"²÷ø dûàÕ>¿)&FBÞ)!—ËV¥ãfò0­©® ¸"ïkö(—×ÈG²|+ï·2 :Ÿ2¹»«f3îa“Á#üA‡¶‰PèXfYg>-½"0ßÕÒŠ¢}M?—}ÔQÜh\ͺ4%bÍRã+P2(씊Àq”6ËŒ4ÖBYžg£Ï¾ ˆ_WÔ š?qÔŠ3”³Ô­õÌú¼÷4é8¤­|dw¹ Yµüƒ‹þ®CÌÒ à†|ÿÍç ›gÿ®7æ‘wŒ€Â6•´ß®@‘δÃ˹P,xCeN*>3×W4¨ÕÂç× /W„ÂÖ¤oÎÆðå-ÛYé:÷m´2µ˜ƒ¯¿†²¹ %O*”ÇÙóðf;µ +”v „_Ÿƒ²^ú\ãˆßP>:0|¯ñ4TZßc È̆o äp¼î Ô]N#ŸÏ‚:ëÜ%a="høÓ¢6¡ûw~AÍ‹'¡i°—ŒµZYšÿ’…¶;·´Ýî±BGÁyÛ}EèZ¿æ¹ =×çêí~ñA¿è£?/ï)Á Éùï6Iû`(Ëþ,ãŒ<ív/êa€±»kÓõ¼0àF%ìð¦[¸ÆÂÌØÔòXù3}¼3°Hac¨È¤K×&%bNÆÂò¯„+wéaõ•ÌÔå/ðË9¤ùóoIøímžq3Õ6ë5ƒe}ª`ûDO÷L—==jsÍ=‰äÍ;ìGRñÑ•2H¶£úúl.’§zÞ~ö{)CÏ»èr‹#u]»\X^<Òšë’hœQÁ½LRÂMµw>ÒUþwà!d4=*× †Ì"5žGUxqŸwwQ’(îw鋨ÍEöqÞ™Ó‘³<¯–¡­‹Ð¿ù|™Oðà^§ëgñ™…Çr.nWdm CÁ—Þ§ÜnáÑ;V1Ÿ‚#QÌdÔŸÕF%ØÆävÙ( $ ‰öK'”–ÛÛ›fW„²FÒ’¿ð¡üí‚Á¢¦!Ìû†¬™/î*³-ƒõ $/J„,a#ÞæOL®üá »é}:vÆÐäáý$v|;Ûp¤IfcYZh{‘,8š¸œa’Ûð:n5‰”×µÕ]¬úKw,ÕÖ5¤Uèʉ Û{– Þð’Ez§(Õ»aýÈ0ßZ9J‚LwYü©•GE´ô~ÐÞmd%Šœ*®rB6ª)YWk<`8Gå­ÆœÛٿט‘»^®‰@^ÙZGc—ƒx0ýB™=9’H´=t ù;’´>¯¢@Z£wç† fÍÿ…Âå2Eñ(zR¦Ár Pœ®gè—ÌJD¦n¹†’&'G©?@i¦wœ2‹PfèÐ[{”ë~¶ô¸Û÷ž;X ô ÁÜÙoéö/TÒŸ=0åÍÊ7ÆÅ¯FG•â6 üS¨&PÏÙÜ.êåíw4¾£æ“—Çd¹¼QûÜÕ:®¨+ž•>ùÀõªöo Ø:Çs8)ÈÑð³@—\g+.|úÊŒÆRsí×u…ѸåxÜÈ.#4¹TCm³& ?»‚b|è9¾›|šN²:}(Í,ÇdŽªZ¢Ù6?ãß yxñK˜®š_åe”â A7ÍæŸ£Ë(#r|çhåÜs“O\òçŒUœgjúÏçRPØçâçú}÷ŸWæ?;™xš­6ƒˆ•~ båIQí²ç€ß³UÙûü«Ñ6{ÄD|ÿõéÿ×Só_Ÿç¨øŠ½Êu™®öu({t·™öð¿ûÁ¾œbïÖ=ÇßÞƒâ%Ÿþ°ô€«Ï½Ívµ_«ÿyn¾ÔðÅ&x@ 3ó‹ðÉýÿŸßæ?¯Í>›Š«ŒÎö_\ÿç±íšÌÚšù糪/X}¿§wê÷ .ïBæiÐ\”Vœùôþšÿ÷~ðŸ‡ê?ÿT÷qÉà·•0àö3ó4­= qØÍ¾#%…áUÓ Š<šèñ¸˜öÏ[3E2¤99+ 3’L|d’.0÷!ìÅ3ð‚…XeÛp•X¢ |3Àõ –µ3ß8´«œ3é;`mÑhÚ˜ÖwÒö&%¤À¦1IjèÁøÿùjô tÍÌøѨXÖ"’òU~ýù’E=Õ¨¤\þŸ¯†…Ã9ìé{¤VèD¿o¤Ééµ/~GN¸nö}e‰t@z%‹{²wÓ‘aB›¡Í–™2­¦o)#KœÖ†Z˜ ²7”‡ïýì,_sO9~GŽN;Dö«Èr2UiJyý™œÎDE¾ò±~éâxX!”Õ!5 ,ˆÖ´lú¡p{‡Îˆñ].P’tBñÂyÍÈÄ<–ûe£ö"J•1Ïí/IG™*‰w(–Qní+m˜*º+zqÅ^A4¯=Àjõ•eÈÃÇ~„¢JýÙíãdTs n¸§o,º²ÛÑçP3'÷ƒúªjk}{|/u~,þÒ¸÷õ´nðžºçƒúƒÐÒÀÓ@Äø¿Yÿ(E£¹‹dM}hlôº¡y ;îÙœäz„&æ>¹O¢IíóLÁŽBl‰‘¢AÓðt²ô8|˜éz„§úCêÉåŽ Ç^íÛ­ 31ýF÷–¡*u>AlVûïº"mAÚ>šOÔÝïeÔÿ#3f¿'ü]·ÄS[þ2˜¬O6GBÊK¥ùÎ鿞Ä*ÙQˆž=¦“ù´á_MfkÞ°}mŸ¾1Ù~RùõσX9ÚòaåM7Ôn¼ÌŒ´ü?L¥ÍnÄ;þõÞºðšû÷œ“™‡Ö;ÁtÁÐ6íùâŽÚ§ó½8˺ ãH5ljNhÿs>ܸG‰Ày:ФÅa„ÜŒ× ªþË…¥ºÐ6áÇÿ"CÚIIŒïíM‚vŽû¹õo@»üFZ¶àhrOÀÝ„:„Ú<£Ë„NòûwÚïRA—¾¿èWzè®b½h® =>_§Y ïÜ~’º_%0 X—ÿÔ3o Îi­TÀ0ƒ„òéwÕ0RiìPÒ˜cŸ¾ßøä•7.V6wÂ4Ó>1$ ˜MiºU óN¦äÍKÌhZ&ÙËËb Õv°:)bwH~Í=¹´76ŽpÝ“•¬€­÷ayï¦`Ǭ˜"ŠÑ‰HŽDYu"I¸[Ã%‘h$ƒ[.nHΨä­ðM)¹Äí< ‘úìÝž[‘MHÓøõMCîñª¼n±üé„_ÛœF†x»™Äâ}È$ÑýžÔ\™»„¨ïˆlà¾èh‹{üqlÜû Ÿ"ûð ˵œ|伤þŽKž¹Þȵ™FžÍÝvB-öxÐu–Y-ç!ò-ug_—ºƒü¡#Ö”L(àPw‡Â‡¶ÃÈ> ð;ÃSÁ&D(Jí±@ÊzÅÆÍŽ9×÷ „×îÖÖŸK(yÄ?1a³¥&ö‹O««£Lz±ò’ ”‹‰_š«£@…Î=­—Ž~GàJ¡È¡5F%éôvk/T¶§{N}SU2ƒˆ.ŠH¢Ú!~Ê›56¨^éýø¼j>Ï™$†Ú6%6ìû´Q÷¨ø•Œà Ô«‘t ´Ù@Ó¼z—Ѱ1Ö «Ñ能_ÒK#46ž4b ;„Æ[ îËh%Zô°rMñUÏY¥y4ít;æÿ2ÍœÌW¸íG³Í'ßó·ÅÐü‘îqïÙh±7ˆÉá} Z„YO<1BןNž-h†Ò{V†æÄ”L˜xøxP]M Œ>¨Ò˜ÕÇNþ«Ñâdîv„<|`] ¶Ì©=!“û!ëPžÊ^anødì¢þstò¯—¸ËûÓwÈßþð.ˆw ú„¶ù=„³aÌ¿N8qˆ5™¡XZáíG27(NUÌßõw}Æs¡…¥h/ÊÙ›Êrfx±_Z ¾šÌë?{_7÷G Úøëw =yíy(|_gºï5”sTF·@…×êt³œ TQñv˜•8±U¼u̶j©Ž_8øþ3Ôû²R)Ýd‡‚D ½lÐ8ëúu¥Èš}zÏ·í¤@«MCq!3´ß.²u<Ú‘9Ï"þù§z#B”ÍS»a@œTîT”4 vræ-§0ÁðǸFú‡n0úª5¼ÛÚÆ_ Gï«E˜LU·(‡éA[°·ï„9ýÞ"Ù²,X¸qáÇúSyXôøåï£ ËVWÛVì$aeõ†·ãÙ"X«+º[e ë=lî¶Ã¦ «ª[Š=üI÷dë4dE"±¦òÍ›‹HüZ©2")Å mO%$»ÏNÅ¡rɹ7eŸ h ÅŠo”}û4R3g’Í+¤yü%?ò<îÑqJâßôüe/*M¡ÁÌCì¥72…¨Ø;KW#˽EjbNad ½™<׿ ÙÆ}2¬p9"Â’ŒSÛí⥇~“Èk-~t/ÿä z|O;ùwþÈOëžÃ#Iþ¥‹]y(ü^ZÅ£èÛmE£ 3rFž!@ ÝÕ)´@©gNðõJÊ„iœÚÕ‡rßÞûíy…Š:Žƒ’÷ú%sþøWö¢²Àe>ª|UQňj§KSJߢ;ÆÝ3標u.ꄵ/jš¿}¥wuÊÜŸ»¢ÞÅ?þŒ¡þ™N¡~xÍÈj¢Ñ)ý¤77>¢1…icê§{GíšC“¶NRÛA4YrÛnt@ÓÀþ\þ|V4;bÉrk¶ÍÊap|ó&^ÿè¶hƒ ÉŸ4Ü!E–ì^ÉïÑ0sTSþVåe¨â%™2¬_ë÷Ø‹þç£ÉñÉÖr†,m2#sÍ-Ș8ð5ýX¤÷¦s‘±ú»žiÝKuâkDÙ¼@ôó¦Rm6ÿjRܱOéú‡à“ßík<-òÿ<ù•ÅQ­Äwn@m‡Ê%»"ã¿×÷1ýÐdvkî‰Ë4—¾¶ñË„VÉèìS" í‰Fº¸ÿîd¾¨qw%ðÜm·j«¿CŠÑ8 äàÇÜ+ ÐNÿ¥¤÷h*´ ½guZƒ¶ž|þ&½qÂÏ‹½si ¨Q¶{á‘Á_ß· ‹R´¶å}— €ŽW¡¹oã( Óûì9ù4Nèª:1jòp~ª_lë~µ=³%I¥Êúl%Øb [áírÙâ¢hy$jøþÓ3  I´ñÞoÃKH:eÎÎÞH‰»>øÉK$#Å·&Ž"oV¤&bšUBš‹q·#xdnG¡TÒ…÷ŽäI¥îB†Ó›ŒvÃBÈ8‘¹eŸ‹ÌwC^¸éýÂ}j±eÚîq¸mꂨ&‘ýzëA·C†È±üì¾Úmvä69༽.‹<Ÿõ.ˤàA ¥¹oÄäÈ—M³›J=ùMùnï G¾ªR;u(ÈCëÊ}˜‡†?}ìÿh¢HQCšÎGûZwoè\3J9ÝN ¼‚’DyWÔåPªè•º[çw”ñ1O{?„rÎ%˦Ey¨ùå#;Ÿœô OF%.÷çÍ'PÙBj&ʸUâß}x.uÕx|—x‹Ÿ ú7Z× ßŸ¨ù\QúÍl8!iV÷18¢®@ÆÑë‘q¨WoHªZø N½XÌ@îŠ?’Ð(³÷j¤ÝS4¾<\S0Ê&üg^É ÉÏWÎ)ᇠ}fÜ/Ü÷ƒl*~ä<&h6ÏY(\ìƒæ‘¢—Ô (ÑÂèÖ†¢g"Z’;ÊÏ÷£åç§ á¿BÐÕ…l¯OX.J1ù¨Ê<\GÚotil«‚0F¬h®2es¹ªo>YA¾›g“³Šßÿí¯½ú!¥ëGHq:=$KXQýjI…„g£è³—!ާt’ëiÄèê^ ø³Æs gÿLAäú/ޏ²”5šx}=ç!k˜pî‚ÔK+Ÿ»‚ sjŒ“o"‚0÷v=z¾éy–ÌǪ[' ?ŸÊ™äÊFZro˜ƒ¾<DEÞ7:Úç#û~‹Î 2[§BímTÉâý²ì†jFwžîÌV¢ñˆÝ~iÔÌ ‰vü–‡ÚæÃo™|¸<Ûy.õ{¡Áþ×cE”høÚÝû‰Š"´¬—çFcƒ+Š˻фÎÔÒEèš|0LWDÓç§÷]e#ä ©hoØu4grÒ,> ͧ¾G^«ü‚”Îjæ(¯Ù¿|(écîð™“Âtn:_ùTF*– tæ¹?ÿ›W…þùh²ü8ñ) 2¿EÌ—š@Æ¥ü3/cBš‰‚ƒª¶$M×f¥mB¼ân©cl|„9t—U!çÒ¿š$âºE²Ÿä&ž÷ù·ZåKW :V¨Í^dã"…DµôáæÐ$¹ÿ\ÆÉ*h~71Zã- ­ñiÊÐvdßóÝ¥dо/)у—Ú­ì»®ZGr0…‘öŠ"ÿ(ĺz£¡­½Cš¯æ%´›ž`)¸m ‡é'ý¡ûZÂ$u´»lo—”òB{õÙ¯7,% CY@Su :‰âh¬¡sI^)¢ûtã‘·£[—àçù—†‘Ð{¦têõ‘èç(œW†{ö/OÃ`T@SÆ×Ý0L~ñÏ© P­xéÕ c¦EY:–0¡£’\ÐSn©ûÇ…c`f¦Î÷~À/˜¯VÏcÈ‚EßÉì™!°,¯v}û’¬2núf$©Ã/Š£>™?]à·È»ÀsâŰùüØ2Ç]ØÿäÖ.N‡D¾öÍÚ_[ø­ü3W!$}%’P-J»ìýîËFÐ …+3õž»×‘êC‹¾õØCܽñô|„hÒ=ÈÚM?e{£TDJæ-1PžŸZ Cl«V½¹Yh›åœã ²Œ%ˆì©.EÖnöߟ-Ò}÷71§Ð1äðkqv(ãDn–Ý6šÏï#o€Í@øò.‡Tkë?@¾«ÇmÃÍ‘Ÿ“=â0î/}|án é< ¥¨‰Â¤vðc Š8ÏKôîA±0•V˵”¨ðWšÂccC7ö¢ÔsÆëÎ'PFg2Ð@¶夺—óSìQÁî²T %âñ!fA2T¢”ÒÖp½†ÊjgtdC•ðJŠÕ¨ÆÑõ³6=Õ«£áÍ»s¨ô"_§5µÏž|ÂHûuùDf ¥õ~8¼÷e*FƒK’;VOÐðW¸üþ:f4*7¹¢P…ÆûT­$XÐÄpöù‹Æd4eþÃÑìꄦ½4ƒÞihÏX/í7š»sߥB •–?WÑrJžÄ´\/¡QaÇC_ƒ /t +É»äÃKûP’âݨ¢]ÒÌÞ› $þ£‘3 Óý¡ì~}ÖËן!¯¾ä·ËýŸ„¼šHM}&)ïZçÇBR@ºªmR/!×´èZŠª 6…ï*åî=Ý–ö6AÅ>D´Ê–k‚H§¶ ,…ÿjTó¬÷Z×mÂÜ*rîªÖ(¤,šµÐ&CfÓ†\;äT¨Ö¨í]‡¼½·™VHýï®pµ° ä>)õèa 4“HnP†Ï@KimßÊ*BÿŸŠ¥·…ޝœª‚Ÿÿù§zÂUNýlÝýÖ±ÁéAÑ0¨ÿL/z[†÷hëRÂÈ\ [¼7Œ6ÍØ8“$Ôübޝ`ZaûªXSÌ~´ßTw&'e³WÁâ#å¬.™VX&I·˜šŒ€•è—ßæ6ïÁšŸˆ_fF3¬‡,SÉ ÀF¯®‹àçøc<•µ|Ä v–“Úó^X"±UIе’ÔQŸä±D2mù€†œ@Ü5\ÿ;ÔN)2ÕvS]öGªÛ€û:¤9öÖ=#žé¦ýŽ™yáÞå·É}+CÈðÚIzlM ™Ní±58(…,ÊÊGn îCVVïW·\‘Íÿwìé)ZäòÚnùH¯MX[è#Ì—û¬^A+ƒÝc6Ò¿žDÃÉ'儾›»ö÷ÿÕÄ‚Ua­üo³lònc¹äß>Û•Ž Ïn”µ/"¬ßýD˜÷^.< ­ƒ&ÆÜÖvÕh6Îq`„–ð»«KE£Ðúmâʘ´¥Æ”q6}€¶SeÎs ÕÈ`ZFÚ .&«Í @[ø|²óÅÐöŠêu©Ïh«8lòPðï{–ž|£°øbãtìaÐ친ïn¶?Ør„NÇÛ¼¦è²gæ&%×‚î¼ £nSÐ#³Ç¨’ÍzÊ/K¼‚~oÖžn'ˆŸ/ Û'ƒÙ¨z!V†‰|žõ…7ÅÅ­úãÝÉ2ùã۲✩Ö0ÅíEWuyfžUßq}ó&4w˜^JÀ¢Œñ„¥Þï—|Ò…•‰Í«`-xìºeI¬gZ1®fÈÂ&å›Ë•g¾ÃŸü _‘Hœ¤†c±‰c»¹^, © ¥¤G îbäö—¡ºŒ¬'ÖÎû·!•ò­p÷s7­8t4fópÃm¹ÍH_Ëc3"n†Œêü?çžû#S›Ý;cÿdñO|ÓÎÙ‡¬‡'c<-²ðø%òÛÈAÇföœœ¹²W¸õ£‘Gùû™"")äÍ™ˆª{» ùŽ])æ¼zßv _Ï·F\ØåÑÅÊ÷~%¡Ê‹ìa5Tc;Vé݇êµ24Qù¨ùrTn³¨µ­ã¹öP¦ .Wjò®‚Ôk¾/Sxå=¸<º\i[ˆF»žZö +¡Që·“÷£q\gñ–/šÜðñçúÊ…¦F翯«§£™Ð·‘j¶@4§ÝOLÈyóU^ƒ?¦h1æzZ¤-‡z=—‡à‰YFæÞK˱ڇj…ÿÝ/ÎPKa¾^õø¸àþ÷} ŸË»ÿå!™Ÿ©; åÊL»‹^AÅ“÷ŽtIâPùëcøúþ¿û$¾ò Ñš€híßM¨ûî«ßlß%¾Ø\u} –ú®F3®ÐTaSÉ3E -îÚ×J28¡ÍJP­/û0tN(G¸züóOõìc1Î}=ËÑ)ƒ”ª, 0T¾4^ #o_wÓh-ÂX€öUŸ`O˜Hq®|´05©JÎ³æ© `xæ#ĸ«Yañ”ÀÞ;¡œ°Ô®¿ï’` ¬ÜàžÍÔ¶ƒµSöÉÚü °îvÖZ‡É6r vQh¤ÁŸƒ‡|‰^½‡â´ÛûzƒˆÞ\6õB’”a›µða$ª5:4[Š»*¥ºÙf#ÅóxìÊï@ª“k3£HCyŸ;%ñÒÙ9)Ž àÞnò§Êdð¾¦r8™Wûq¿3²pMÊIgÐ +× SÚñÈv2/âÌW^<°8)P²»¹)ýƒ-ÕÙ‘gà‘ûál*<¸©Æõi–ùõ¥¢]ù¼Q ëe’®Ç óp.9€"/ÊŒîéúu=ÞIV%FæH‘j%§Ä2nÈ  qé‘÷(§¸+ÛÅÓ š×\å•@hòŒê&£F¥íg/*íPåõšFF¨,ªÉëxJ¡ú¬Oè±Z ÔL¶}þVµmwâ£âbQ—ñ؈#Æ ^q2“ØåƒhpÙICЈQWyXˆVÖŠèѸ'†1‘IMêéw{~FÓŠ$^D³y/«TO4ÿù£’í¤4Zì,ô“³òá ¹9>‡þt?$¼e3¥„2/úèâòqïgö¬@Ö ˜ª¹½>ý¸*^<îå®dÏGmnx¥ðÏG“Õ.N|ÃàdT´Š<ÍÞ‚´oýT×ýTÈ9}9ûÿ¾+¥°âxÊ#7m!ʣù,‹å_MT3½9)Åv÷Iá+´ ~Û†JešË²P{Ùà~ÉØ%øžvæ‡Ô %h\6ÓÜ:êÍûËVO­B‹Áò>¨y ­— Jr' Mõ]wÍYh 8þ±æ‚#!ÿVžÔÉC›KÅ¡ôí4h;GÉClM8þÚᬔKÈI›Ã/yë ýøÔ¶!Y9´—yœ“¬‡óŠF¼¯¿.öwBgky~•=5tõÎïØIXÃO±:ršñ+Г>B'oõúŽW8ª*œ…þî+‡•Á@­1ë;}"¬XøÄú؆6ÞœøS #æÒ\—¸·atÉàãT¿4Œ5p}€Éª¤qé3•÷ ›¸aîg‹.‹Û]X˜ ab™±„¥;'lÍÉ”aå¸eㆨ¬qFºþ‰„uÉýožˆäÀÆ7I¯#Øà]ZÏnƒØ©”óÑ|‡Ä&²2sdv°”{–É>—<“òøŽä%ï†òžK"åÐÁ/*Gp7Ÿåѽu¤ Žˆ¨”À½l'¸§n"½_ΣèȰ:µ0½ˆL·›x•ãgE0e×;MdÝlÁ¼ ‡lÇRul®³àðsŠ9:|È¥Ðé­ˆb¶§×é&[TæNÙu±“U|½C«{ÕXÒÙ^Q¢z]]wbÛÔ|%W”]‹Úg–rv“6 .»°C}Õ êµE4­VßF¯®Q9B¿íÑfHXC£1E=ŸE4.m‘~_O&1´~i©Áhú¤Ö“mÍnÞÍ¿ŒæWs/Þ®%ä ßþ‹Éhù„Cx|2O¤Ð‰s8>Á“ƒà”6ŒVG‚rñ,^1 /ù­ƒâ³=G4Žãnù+ì,0ÒsÇ1IÝJ3~¿xtšK÷,µ¯Q=‡ŒX'r^HŠ“1ùSWKè'Ë\Ù“ñ4;O¥=nBLÉäÃ.ª8ˆšòꪻ ˜ˆ¯ÏüÆÁ7êŽJÿ«Qínéòf›êSD,¤3ÉU‘ C¦¼{ÖyÎ}#ê—¢HIø»Æ*©Ò7¢ÚmaR ù!_† H¡@]N@gX ÅüûˆR“ ˆ½«Ú8EŠ‚îfJYŸ‚b ¡qeÚV(önsô䪆â%_ù¦ßð9z{ÿ"÷Ößýé÷½ý»TækÛÿËÃÿ‡®óŽæú}㿽÷.3›l²Òu™Ù;$ ¥B”RII)¥¨„„dW6…È–½’ì?>ç÷ñùó;ß¿®óþçuÎû¼ßÏûy=ïûº/mÚRç÷ß×·‹Ûykº5~Uh”τݠ¤| AOog®îª“î/NxÑCMHL>ÑD/Ôõ™wqǤAƒFÊns>h¼÷(x¿²4óWÒØxAËÊÊpšÂ$´ý\9¹’³w‡?Õõrúò±Uièµ×ÌHê넾ҕ~sSðü½q7ȆÞyuQ£FŠ?ˆÀ˜õwg*æ»0ñ"°,>øÝÖçÑWæ†ÐUñZ$äS¿çs=‰ºÔ[¸# ÓMá$Ê/äî—“I„ò »‚Ç&R÷¸Õ'¦¶#ÝÃëƒWÚ‘¡øâ·åvÈtìöÊkÛ)dáív¡EÖ§zsŠ„È¾,eìݸ[Ü3Ì1T¹ò›œDÞÑ—*¹pONœ|§’: TIlÜQ»‚ÂŒ¯ËfJ¥Q4d2öŒy(îÕëÊV^ŠF)eâÔPÆU”©d%îåìE¹4ïry9TÈ.¾m-b‡ŠõNŸŠþÞGÚeº“e¸?EÐÿ>a,BšÓ÷ݨ>íV°_³ 5»Ú“Nq¢öÞ±ü^WÔ4Ð`FÝ$Úw±ÄÇPÿLPò;÷­|(Hr4S* ;rƒß>BãÈ [®DKhªÄ¢•!C‡f¢§:Н¢¹@Ñ-sÁz´Ÿ|×A„–ê¼9©‡Ðê”Xâ÷%A´Žæ­fS E›YÁ=½ xøØal¼÷ÁËþÀ>T”{ašOn€ úc+)ÝB0a>2ÈPc e”2g# (·ú¼ïÌ„¯;<š }²ðáý¦å^7'iïËÙÜoOÐýÃMür«=bÚ^M3C\èkW"ˆa+ö<¸–¸Sësb] ë”ð×–ƒð¹/s—(Ñ‘­<Ú½wªõÿª<0à€z‘÷C9ÐÐ4SÀÏ~û¯§¾Íf¢Ê?$¡…4P*B¶Z:žÏ]–ù­Â¢ —O›lùcØ·øª-ææ¤—ߪ|Ï×^kC«¾;;‰ÙVž iþâ m†ïÄü«¶¹Þ?dÝc¶ráÛØðh·à]8ðU: ™®eà :-ª{k/AWRDA´Ã ôˆæÕ|| ½ïË“wi‹B×±\®tMèûáí×ým±Âµîž00ŸùÒ–u†T¤ ¿{2ÀpÉë7bï`Ôû8Iè1¿I»|wúüª-õßw1 ¦œÃY9Lšaú͸¶Äne˜°ñµP‚¹&Ò•¨™ XH¶=ûöa,å´óò¼—…•ÕW£gzOÃú½O…z`SÔÍíQÊK$$ðrÔiòF¢WFu†ôHââ,ß'…d.}ß©Ÿ‡ …_Íbfà ¤út€)Êþ1ÒîbKvCú÷ßä7†!£’î§[7‘)þ#i_޲ˆª&Ê–#kí‡ Çk8=Ù讦Ú¶±}\¢‰‚E‘çkÚeÍä6äûfjs¾ò3ò+[6È“¢@fã³{QÈðéÔM«,aÈx>MÕŽ¢õ‚¯¿Æ½<4_“O¢¤oük{æë(³'±3ºayй凨O¶1…‡ØRDCšаý“«,¿#ûI¼bEÓ]du>½)hº”Žó¼Ð¬§. wlõŸ¯½D’¡E¹¬¬Ö´òVÿy`ℵ!Zu¤¯ÛüFë5¡Áº ´•,)»w{Â3£/Úµž)ñU®Gó#ŒgRߢ۵ÚE£õ+(Ó·Vsˆ2©œFEòaèÚï ~†P¢K¿yú!»7ö‚÷¡qH $Òoùʱ• Iž¹´:Bâ›”?oeÜáÝ­.ýÓ -ðö!CÉýé-¿k'›£_†7á}¤]´ðºÌã÷xú£SCYt÷&¼û0ÜÓVuwK¿—i¿ÜrÝòÛ˜ÇW:ŸCæ´ßc'³à-ÝçG®˜™@îëÔ¤Š¨ÈÓ<¡7Rrò6b\ªÚÁ§ñéo&ߪaQÈÆ t–Yw  8„•% ÕÎþ}u SFhé’• ÈEt(œŒ¾Ð—¾¯-…bm‹7ýç‡úÓd›ÌðÕöu–Wr5”½8îÒÍ忲G¼8)¡âú¸ëèÑe¨Ú¯:0ç*5¶eÂòôP4jUÑÔßGÎÕTœ…F½<{Ë{ÆÐÔ|07¾CKXÕ½Û9éÐæAußrm`‡?ÕŦ{[ˆVz:ž–&PxBß´”¤S§Úcƒ³çmÛUC`¸[i“­ôŒÊ‘x®Ã„ÖXé÷a˜LS ~iÓî~>. Ó›ó5M&é0{ã~˜Ãû0/¼_eß° ,Òõ\eeqeÒ]Õï`õL¨L±­¬—G°sz,¦qvmûU2$$ò½ôðm!y<÷ˆ§ZAâéñ–6'1$½O!XþÉ Ï&]3¹„”Ö“]3 »:ÃÆ'.Méè5L)ï#C|o‹ÿi6d:d·h€,D¥{HÛ‘µGÅöcÁ"²7tí{:ˆ»ÖäŸ=c„\7«|Þôh"¯¿:eÙ° îyT¼ð¨à ¼dí1}ø …ÚHG/@Qcð™Z_Ñü›f(€’?u(â~ ¢ÌK¿EÓÍ»(çCàíp|Y×Z}P1‚ÀúÅÚETî]hå;Iûï˜OEö3#<9t2¿] ÕLD†‹MPóFÉÙXÅ Ôæ½M±"‡:l_9Q7>/>Ë5õ=8?$)nåC…–Îuî44\,£å<¶ˆÆ5kÃ<¶çhztnï£@3ûßBjDh~öÇÄÚ¾“hñ úlB±Z\”²Î|‡ÖÄÚêÍhsòüw¦]óh;ds2÷AÚ= g­ýƒ¥§­‚ÅQ¡Å8ä˜K÷Özèz;i`¼Û¸N‘¦¾Þn[æH[‡wõËß>îðhÒ ºå_ÂÇ»LÙFÆZj”lP¿Õ¯*·„<‚„ÈñÇüöœG~=š¦;¢ó’nÔèìß©‰‡Ù¥:U ‹7ø‡zÜøìl.¤VeS‰Íšä!P-œ2eö¾êíg…øK@CÆ)=BUhLê#œš¼ MÙ͇ۆ¡9ΞÎ3µZÎ?¡áé; -_˜®òžU€–_ˆî>S„–îÛ»ïñæm}V¡ÖR‡V ë’ŠIhieq Ûæ˜ZfméäËl5a¢üë'M?í.Þ}Ÿ‰'Cs ³êÝgéÀûÐÍïzA¾é2ôÄ=üi?¹¿ÿ1œ¾Cì/Ÿç‡~FæÊ¸¼8è³óµxSuž6RŠ0Ä-zM¹Š†ƒ'_\f…QÅÅõ+ƒ0.Ñþû¤l'ürýÀÉ¢¿ÿ8ýÜ-%ÓVpYCo7ÌTº¦ÄJÁÜ©,~­,=XPd<©ï1 K*/â2÷¾‡•Ká‚j<™°ö+ñ º«l<Œ±ßGÿ >h³´Ïé„vz§0" 3A¨Ü¡'HÆ(Xp4b).K ]šF*cþ¥¨³‚Hóæ¡\À¿H/Ü-,ºû2üîl剆L6—vwT}EæÆ8©=#ÈêžÖ#¶g²ëPd¸?(Ã]ö!'­lµ‘3ûÍFÓ^ä9ôØ®¹ã<òÝji I„{–‡¤ÌomåÃËDŃ/ÜPˆ*/Çy? ×¼ç ŠDÑ,›ÄÔà×(Þ¶7Ü*%e‹ÉtÒPz^]Âá¹6ÊêÿCÕ…ò 1:6—’Q¡‡<)ÿ*ÞÖ­;ƒÊvš †çEQõžÈò¥">Të“9pÂÑÚ×Q¢¤Õ×W‹CèPó|[ɱÃÓ¨M¯áwÖÇuê¼ò鮣nøÀ:qÔwxØK´J‡˜ÄFíú)Ѱ«…ÌÕÑøáÍOD£·Ñ”¿ì¾a·šQ Ë~”Žf+•b}ÎËh¾(y¡6ðZl¾8SîƒVìË¿eæ%ÐZÃvÃ{8mnŠ|PÜwmOÙoÕF;8ŸB±å§öUÏÙÍRá‘+OŸrïAט½±UQ:Q±óMK5RØŽØ)8ÂÐ.;W]ÖPÜ#ì~}d«ü~åóXÒXœDùÒA’Dccsï $¬ˆÑ‘GAÜl.ù+I©Î “)¶ô¶¹/¥¼¡ýÂlÞ ¯•LŽ ñŒíÔ¶/A›E[~hZ—¾4Rv7÷ç­¦BZYÃÙG/™!3i²å4d/ ,éxœÙž«ñVžÈƒÜiòñ …ÒíùšôKO/À§·¶œæ!ÿqâOa_âíùšs—©R¡À{U{œÓm{ÎæÜ=¦õí9›Ÿ‡ìI¡HÒâ£æ/¹­>8Ý@(Š•7žóÿúŸš4ýžÇ _O¿ß²•ïÊ"µÇqnÏŸ› JøBÅY ÛÝå PÅD.}ç¸2Ôp<;) uû¯7\$^‡ï~–vû‰¡‘jHØ­Åšî”ã­xhQz%b¡ìmœ)i¯vøS¡QAc|vÐcÓ×'Ã}ʧ½c½¡¿ûëg†ZŒ²çù“0|ç£ö ] Œ¦„¾,¼ ã3ŠÝ%oaòˆOV7 LM_,6õ.€é¦.±k©k0k&lûcê"Ì­W:.˳ÀÂ8û^zÿfX&‹³½ýî;¬j~‹ì›‡õ­ƒ¢½ü°Éþé²r¹: ;…×ô"‘¹ûÕ¥þHÜìýRê•’žñ¸ž4Ì‹äâµ½á¥H)SýÑ[©4Ä£3CNâ„Äáed2`RjB¦½a… È<œd͸N¬ý.®¿=é>÷p–.yäÇ‘KG%þIù*ò¢)Sî9Þ“‹øNnrw9¡P¸;åüß(Jõì=1Š'%‰ŽGÉ{÷Nf¿®C™SÓc_ÅQN÷©þ ÿhT08Çs' O¿™;¿â€ÊinÃË]v¸ßþ¶Mzñ]wŃ{5•Q½E&4ø@jzįG÷£6³ú˜ÒÍî-}ÚÙ'Ò£n܉„o­M¨ï•”‘º¹•uÞYëÐä¢sÃà-=¯6í*ëy¦7ãuMU‰ÐìÉ¥ ë×Ð<}Ö:Ng-Æ¥=‡ƒÐJ±,ž`‰`+RPæú¡­è—ÚUs<ܘ¬¹¥Ç´×§~}F·égMyQÞ‹¬¢Úé*ÈéëËëaÜ9'×ÞØJWoø‡ÆOoýÏÓÎwT·ìðhÒ¥Òk“åsàÃm½]wqR¾ÜaN9IŸ–¥OÇQC|•¬Nò7x{Í&ÿ«‘úöùÅ=“Çu;5aðê®ëÁTÙ¾:ñ^¡> º=俲 Êʯÿ‘ƒjb»·»} ^®àœæ­Xhx&ãö‹ /LݗݧMg´®óÏ^†f­½S¦"E[9ñ¾`Õ“ h±YÿYíÆ-Ïúá¸ý¾­ú3tF %ÍÌ7°•ZÖ^¸nå«Ö¢–Ë¡íì®[£kmð¤   ]¬ûŒKE5tP–3/ B'ÓLþì^ZèBß×!9¯¡;â€p…î2ôî>~0|ü ª+·Ê‡¾k9›µ’DЯàÉ—8åý«¤T Ê…006ja}ð îIˆå–‡a§ý¾|¾ #ó'&®;ÃX×Û­K¢ð‹¦ðAvüöI⢹ÓT!qÖžk0snQ£mFf׎ŒyUÁ|Iõ‘?Öš°Xh%®›,Ë“áÇè7aÍNpÍnM þNµêõ\ BÆW"$œq?-LÅ„ÄEÑa29ÓHúÉ%!3ÖÉ˸N¿d?„”ƒÝQWó‘F˜ÐÇÙâ8Ò½™å0Kz„ qnkÝu ÈD¡$ÝÔ#Ì×i³˜•!SïÏ3dûùµ/¼írü¹wÁ¤²9µö-²I!w³«†3%ò4™ >ÕÅ=ùÊä-µ$(  Àኂ•?…>_.Aá{‰QF GQôÜ)ßÊw‘(î/h|`%ùÒ¯ ¡ô;©:Ý/2({<É»²Èå©(ç3ÌP!ínþ+cJT4îü¨,à°~´UÚ’ø’mQ-Á%gc¦Q”´‘%ŸÕñóˆ;ú£æÉc©¢6u„+Óù{¨Ó¨]º—u_+©±YA}ûö‹ 2xˆ61~Ï7öNE'´ñ¡qh£ˆëgw4•ðÓâêdC³ÝBN“hÎZúövÆ Zðµ¿:…–*!O‹ÑêTæŸgÔh(n°tIm Å®FùvlåBª¹‡»¹Ñž<úôÆQWt(*vãï9ˆŽ‘ßÄñ\•´ÓÒ±u”|ìÅ“ôB )Þ× ÖiÁ`î7_Æx(¾¤ÖãIYÏ•Kn5Dý3Ç}ÿ$z¼a„±»`myüjo=Ä9?+¯~°}n(ø,àD?ˆ2 [€(ß×’»«“!2Y²uúeÉNÍbÒX¼±å£Ù÷vñjý3÷á9 i6ÆXCæ ÆJŸ7ý1Èìé“;Ûs5zÉŠ¶sbàïD>È»pa‘|,>9¿ P ‡|=†³Áã3_”WâÔ òÇ.ñ«nÏÙ¸Ò=z±=g“~Ÿº gozÿj0‚ÏéoFâ’˜ X"êØÉ ¡ÿüÐ*+/ÚÁ¾zˆ‘-²X‘ÓYÏg¡üçô'ÓËPaûJËxoTŽ5û0ÛBõ ÅFè— P;Îpˆk?|çQËùMvêègøŸ@“‚Ö/F· ¹Í¥R0_ Zƒ%¾SoåÙùSL á)òîÐÝ&=c“ ?GÔ|3^²@ÿ“¡=vM÷`Ðàà8 Þ+øq¶F­–´LƒÁxÔÝ%¡¹0IU2<ít¦Â?EPSéÂô»’*˜,Ûe!sÍVwó}—aá«™Uü3, 2¤Œo}^å:ïà}&Ö]>*2†¾"ÎWL× ‚À(in‰t}­®1 qåá{MÁHjÐá M†ä”ÒÌ»G’*ÊÙõ¯1R;ˆ·Gné›v=—ý„Ä(2\Êùó†"™â¸?ü&FæÊ'ÍD³:ÈþàªhE?²ûí#Sñ»»"?Þ;Æ|¹¨“n}HGž÷©.u¦m¾÷Õ™¯Ã( 7þK¼…Û$¼"ôQ¤t)°ô·*ŠÛ1>ŽZCIÉŠâo'rPFѱKñ[Ê1sSg~ø ,Å—te‹QQƒõ–tQùöðb£y<î—ÿXüí͂ёˆT]TÿÆD¨Z Œš'=i/Gmr‚ Ïï,¨SöqÆe–ucé;õo(å¤U¼ÃCÖžAkŸÑH¦‡ŒÌº MÄæOÍŒ iìy}^ø‹f¥+EÜ…sh¾BSjdŒ–ºþóÚTMh•Á΢j‡6jKª-2hû›¾ƒ²3í*‹&\Ö C»¤HI¾zÈ (² ûŽ´½HÛEeøÆþðÄø>©„ÒK‚Ì ‘ƒðé#ßÙÉn‡MZL€s¾ïxó†­…\,$îé~Ø ‰õ¹ûsDáÝÂÕ¿ò²{·ï¶zõý3ßöy§&¸µ·TêtB泓»Êéÿ™ÃæM¿²Õê¨Ø·5nÏQÛ xJB=¥¿þÌ5vh¸PÍÒ2*é.ÝXƒ&®ÌÈóÐÔ¿^/ûš#n¸Œ/ yzT×óü:´¨ú_SÜʉú@ï ^-žgš< ¥-¬Ý„ZÕEŸ?7ƒ¶«Í)»—»áÇi×±\¼Ðîñx¦¦×:œÙßš¹=ƒÎ„Œ%' +ÛÞgi zXh»íŠ| 7D‰„þø'øù·êL„ãèKꘄ~›¶«ïõ`€:çúw ÊZ9¦ÁàðîVê :0,xÇbF2ø:ޏÂX ûþ,/˜Hý8bs~M]âžø n5&þÐ͇z_%"uN˜ f¸¤bóFwlžï†EdþØ-ú–Ã_÷½9«•1iž„ðלRÑÿÇ}$ üy{S¡ _ëÎ5aAbõHi~3$=–JžøÉO–ESä¿CÊ+S%Q¦Hzå›f}%Òí¢ º®ì ö„üáóȘ·¦]3ý™¯¨ñô"KFvVv;²]»EW!`‚w¼o¨\ÆÝ >=úv²Èmo ²¨‹|´ÌF pÏõ‡ý6w‘ÞÙƒ<§HKïê¡°ÆÛØë!(Êïi;”¿å‡ÊitSUP"Ô‘Ô!yJ[s»Ës¡ì~JØóÍ(7°Qÿtz îÖóö?CE® ^Ôʨ4éµÈfúUYý×½|Šjûò"¬<ÔZÏ0Fõýë%ûÖQóða¶ËÝ-½-XåOC&^¦jÝû¨ûæa©Rlù¡ùê´¢99¶&‹†”ªƒ2„h%ð¼í¹=š*âßɵh&ÁÀa¶OÍe‹¸ù7ÇÑBßèøÄéchyÙY;Ûƒ­²Ýgk¥¡ ÝÚKO‚>´½»W‹¬ÐŽããC¶–´ohœ';†GÒ•ÎKûáÑ’Õà;óîxvÚé…!ËÜ{9ûx¦Í}$oÑ:¡nƒÖrÝ.€bJ C÷2Èœ¾3~‡i>˜+Õ,Í-þs¿"i–âÓ#³7àm»ûµá~ˆ™¸ôá:Yã?ç†Ðs ^Ï]×~§y"¹iôÿ2„ïÔè¼Äw¥·ü0ôµª?$K‡FR#‹l޽L'Âó:.…˜˜QæÓ{› `¤YU“Ði6ƒ±«÷¾kÔ‚’_ÕÑžµAÞ÷`çÍßC;\Œ4‡êÚG;\Œ¤OsGÞ¹ºíp.Þí)Î*ˆ”ßá_¼Q«’JÓÜ©ñË*Á,k);Ïû—cñ/‡¢ª0&¹ý'Ôõ™ˆF_‚³ÓŸ>7@#uhö·Bhl(* -„&ߺ¤Ü1;Üšæù.öCó€X÷o9hÞŒ|Pa¦¿å~çvÓtíp*Zo^m“ÝáÖìðgr5®IÍ·@GñY¹÷f„Ð9ÊpšÞætïUtp «€žG½O§ä6á' uÉÝ8èS¶ÌËŒ‡¾½ë‰_¡ÿºRCºÝî¾Å@'½¥×o¬b!º\nCÞË'`ä|vEÃñ30¦¢RK¸P†9jVÈ`òQöÞ*f˜šð¡$|{¦3©$$Ÿ0nõ™Ÿ#e×j`®CÍl?ß_X(iÚç‘°–zÃ;3~¾„U%ç`ŽXÏ> (µÿlž·» ˜ÂŽ„ú{|oV²!1S_´dP’RNûÚ¤ErR×ökŽHIÓõ|©å¦ø $‘ö÷z\Å-d` Õ?ÌÙ€Œg¾.'Þ@¦º¾/³j]Èb™Çç#ðÙXUÕæòµÜ=J¹EéœÕ7*ñyFf4îQû™ßáL‚üY/*^}DA ‡zZŠOfA‘Ž»²ß#PlR-”ÇÆ%t"DÉ»Qšœ†ž‡™e™]ê|øMP._2DZ: lnöÇ{rá¾ ®O>TÊ^¼®¡‚*= T×îV¡ÚþÑ2?:o„†œ¡ä¨^áÑaK{ 5Wb~ž¯GmôXóRY;ÌÔl†ëg0œïùä*†…+åɭá°1nþIé¬|7¸/w]Ö©îéš %ÀÆÅ²³‡Sÿ ÁC_‘Jm-$\Z &Aâ뢽JÚKHJ›Â­¢õÉÊç| |Þ#Åg“©1R?¤&ÝP-JAÚ׎¿½Lu‘Aý U^2)2væ÷9›"óc‡Ú¡Þ dµéxÓ\ƒìòÛúyÆpì½v¼(9Ãúò!Ûëæ™ðMä–_·®µ@þúO¶þUE(8Á»- E”’®m$%¡X~oP” Jø\ýó¥“=}U¸TQ6YáÉ¡Ê'cyö˜÷ÕݬZãBeÁ6?¨:DólÚñÕℯ3ª'µt>ÉÙƒšú6þ×¢Ö”¯ÿìù;¨“¦Ï{Çu_ŸLÞ`ôAý@‹Ïy¶[ùð*[%oG-]ؘ9‰&÷U „üÑt‚аHoÍÅÉ,Ó ¡E€Æ«7WŠÐŠTãFŸ•Z'Òl–У­WkëéÇhçùq0ÂÞ zìv­Â£"†In ™èno¯x››¥Å?œÌvKDš¸ç>oL`$;Ù“g»_Ó[z3% y¼ìA&KÛ\ k‚ûð‘M+(LÍ RrãxYk!‰›7R#2h;÷¤ž ƒ¸Tˆ;ª¾Ý§;v»¢Þ¬\@Á†ïq«ÆYuçy…5CÄèZ Œ‘À/]ÔªÈQ~ö‡:ë¹’ÃÏ~BMn.j¦AC©fZ)³4ÎÊo€&ƒ¦¿ý» iñÅ3:ýhÖáš–*ý Íw Ç×™AóãÂfM'hÎ!<Q+ Zö¬Ûx?V‡–iþî;… кÁ¯Ìf¾ýþÚdçЮ1¼Ì@ :Éžb}RÐéjtÇ”³ºRîͧ4ÅA…&µºDôÞÏ΂ŸCjü'ê /"&0Dxú ïwúÇŒAÿæDg¹?7 ÔÛOÔݶ‚ÁoQ™µ“^0Lh<)WjöŸŠ_áûµf âšM¹\Z0iÉ–L¡ÍSÑáÓk¥ 0ý@ÔÊU+f–„(ÎŽÉÂ\ÂØï„FYXx@³!¼–ž—ŠD ÀJ[½eº"¬Ðÿ£]6oQ-ËC‚åW#â"G‘(oît˜†;’$1˜­óu#Ù;JŸQ;Z¤H£R¦' Eªîú–ýzH+÷à*}R0Òg“•^óy†Œ"Æ&Â6)ÈäKr3 Z™7E„Æv?CÖÂ(=²Ê"dÿÜ?£ 1»‰\^7)¯#—ÏÉÕÆkÈ+zKP›ù–ªßÍêµ ¿sq}Z2- LrÉ~UB¡°”˜EqsY" C±ë_nѦGàÞfg&® ” ©!Õ\¹‰2óý:•(wû–÷]kT¨1 ׉Å}Åœ^v?£RèO½(»-?̲;N„jlGŽæI !ÄI ›ÇªgÉÕ¿¼s5UÕ`)“µÆgèÊ‹¹P'ý§gÉämÔ½¶|Õj ê+p\y_«Œƒ±´^}hXú%\ë~ÉþHMå=-–hC3YZ‡ÓÒvh®ÓÞ°—Â-\¯4«ÊÿBË$›Ó3‹ûÐjã'ùäs´9_þ‚Dâ ÚnX|ÛýíÒt‡omùa@;‰æñ:t lW~º+Ž>~©OòÏjå©~ìCѶÁ ü‘|è‰F×ßtè_£ÑÐ:x>Kèé6ø öハ¹Aj5D‡ Ä'5°É”°Ã;ññw€R›ä¶ý¾ý燹¦›•Ûó¥ÁÖ1Páàÿ-ä˜T6ê רÜ·Üþ·&¡ö{!‘¼#Ôÿö¶Êþ g¦é3Š@ãûj »b1h–xÆ¢§¤ -_gœ¿²ôA[­^4«X/t6&ûZÝÒùÏ$ÚÐûÈ,vÜò6ôU›(/®›ÀÀ%×Õ‹Aù0dÔ*]º #7¯jjŸ/‡±ÚvÇD—ïðKZøÙåÚHøð3ÍùýU˜æU:_<43¯¼ÛJçD`N7ÏO‰"ö›±_>š KŽ’r?žÅÂJ¸(kLô¬ ÌqØUƒ kC=9b$¸¡žÆ4<‹„CÅGòø‘øÈúFCc’Œ–µùÈK"Y” ¡Î: Edjôž¨óH5¥§ãÕă´AŸ7í3‘AùìY•O3ÈXCôÐUdöùÜPãÒƒ¬ß®ôøÔ#;gÀ›PþW¸K˜é‹#rú½ØвCžÃ aŒ·¯+ä¥{%)ò 3pb!BÁÎÖÀ虑ØïÚ¾+ÅòXL?ˆ¨£D `[øJt?°i¼‰²su£P>ÓÊ%öýeÜ×õêù}¨¬rQäLÙaT]ÖºC[ù¸äku/ª§”=bHAMÞ}ó6¨µÌ«>RaŠ:Åñî>ÌEÝô("½ 'ÔDZ£ƒ‡ÂWùµû¹ÐèÅ¡¨†äghR§|¡ëàš±V*¬Ýú‹æGïÛ—p_F‹oµ^wÜB+¢Ü¿l h½~_ë„$Ú¶HüM¶rF»®§Ê¯-,ñÈ®žHJÁA<ÆéHŸ÷Ý.¥y‹’ Ä€;SA± ÒOÕ9gC³õÆÁçj 8°Â8Sÿ äºGe—ePA†Óƒðk…Œð¸Ž;üg $ÏE¥s‘‚DßC=µï·úUUq}º‰m}Uñn÷©MÏZV!ŠX,³žôóN§ù{þv}åÎó ¥œÅöÉ6Á׿ÃVž†Ê¢T²ŒYq¨­ ³cJ—…ïÁôu‚x¡ÁÏ»¶þh/4š>IÉâ€&–vC˸gÐTQ÷7Žš \d0½ƒf‘Ô—Òï= ™»gÿOy¯­üè{©úÚhIz%KY-Çê‚euW õò{/~ÅÒ­œ2?ày~ä.š•ìU‚öê‰ÜøóÐ1ø&øç2tíýñhí…/tßyw÷™ðèù#x·¡‚~Ú½°î3>GÅÚäw ÐÏÝ1tx<úÜ*þ+ÃÀ—ÈoÂ`™˜³ós&.st“Åÿü3˳ &È¿sÄ¢/Lr^e›þ"SÖŸÚúâ`úèÁ›É˺0SõðòÛ–0˜ózÈrÛL +e;Â’á¨[¤£¬xÙÚ¥kåÖùöÇü`Ãæm\Mn $2j½—ˆDGɤ‰Ä‘DOF òY ’iÇï±´æD ¬ºuE ©nÝ—84Mósׯ"ýÁ¡YÍ’»È0Ô£bƒLr¢¸mî!sÒ‹½–Œ¯õ$‘Åî=§ý”ÀÇ*Ü•@뵯ù+r±5¯ž=ð y2ŒbŸ·D"_FöSj‰·ÈÏ6ÑZ¤ˆÏZ»NξA!…¼ÈIëk(BhTDtJŨ5Ïü¥ÎÁ½Ç¤Öe3LQŠïÁÚéŸó(ó&ëàu”Óg˜ûÚAòÓì_³‰à¾R Túq¨tY2¼äÒ–†QDL îŸnZ2jD¸¡[±¨´ÕFÏm[±@M¡›Ýô‹Æ¨õýY¯˜¸#êSS¥_žG]K†Ô\y¨OûÅüdê&”¾^f¾Ž†ÑöG]Ý)Osg|ASγ_ïJ ™àó÷÷­,ÐüÀ*åÓ2´8Åd°ÆDˆ–o:)øÑj%UÝ‹ú(Úœù¬ÇZ½Œ¶‹ÍáÜÜÑ.N5Iô|:x~ŽàÐBÇó¯lÒ¯¢Ó!ÿx/[7<{îAÜÖÏ‹"¼…×%%™‘¢Þy€$†ú ½õýM_@Ñ®"Î~Ò½Û÷ðU6ÿÞ‡TI½áÒBˆ·?ixïÄ5ÛÞÿá±N NÆëÛ¹/k~ùòŽÞ"%ò?ô†öAøgx9ÏÆNÞ¯0µ¸÷¿ýðïqæ¥þ`ÈÖIzÓ[r8²jί@Nr€QÎuÈå“"h· ƒÜ"šðÜ!:È;Á¥FÝš miÿùaI ×îKEÛüÒP¿þÝPq§ôÒ…K¨¢T>}'pûüð$+)Ô~ZT ùn õ5u¦aõŠ[ëÅÕ—©u7¡ñ›Å5¯!þí~™§2­Z~„d—±C[Ûî}©]Ç }m3ºs·tÝUñvío…^݇%n)ÐŦr86´ ¢.ÿ}C{ž«.Ÿ`€s"Çwi0ýäéñ§¬0±B0-ã ¿®2’†?Ã%ª\”£0sjžYMáÌÑ*HÍÇÀüF÷]¦Ž2Xâ»À´ð{VœIß2|{kGÇO„5À†Jó¢b.œ'T>‹„#9ÜH¬ÞÈ{Öñ ’”ô>ùÂÿɼ,ý®GŠë”föJHÕöÌÚv$ior+Þ?̓ 2:UùUÝÈXŠrŽyÈ|ޝ%}ŽY÷fÚîj@vb_]aÜE»Jrää ät}Ûá¼Õó$>Ò°|sr÷¾æ"Þ5UG]¬'y[IS"|uŽé¼P,[ïv}š)J¼ll(?¸¥ë´žyÝ+EÙ‚*›÷uu(_„…µƒ¸o¾ZºR%èû#>%%ëÿßýôž˜//a°$‹í†j “©4$DKþç‡43ÕÅdga|ØÞ1èxüè¸Ó¡*S4íÅ6•azßÉ„ñÔ˜‰˜éë.Þ€9MáÛ²á°ÀÈ|{–Hhs¹¯À ëêñžþUX3:áaÞþÖT›óH`˜î;<ƒD¤‹n‰#H÷êî‰ÂD|ÎÎJbH™§æ'©} úÈGÌ*쪠\vÊå‹Ý?ïQkŒƒ”àK—n÷•ü.‡þlß¿?þåHÛ?ç†Þ­;÷ ÿÕ[Ä“_¶ŠD—EÉ?þu§F³tÐë(ýO?ÌdøkaÓ²Èù‹~§½œg‹!çŽ)OtõÖº P’*¹òÕÙÌ?ù!7颿· ¶-_ü•-í6ðÿÞOäæMqÚÞ7 É4Žýß~ÈÝ©ÔbÜðßýÃý°Ævqp4¾I\ û¸G*’7ùL÷™C•.á[¨áNŽ æÚù–L¡s°µ^¼éù „¼ñ–-[ëDwf'4‰ÓWÙÒ½"iù.ám¾wö£!hŸ•|)u-ºþ(´ÞÚÒ5Qyè«{mÐw¡G¥Poب4¯¸Ý…Á+#SbQúSzµ+Æ.,ãúd¥]ê5¡'à÷.™ô$øÄšc@3 î&Ùž–0Û¦?"Ü÷ æKOO§[tÃâ@Ö‹¾¬HÊÜšT€µÚ|×&gØà¶/RG§’âD1s$ülÂDoÍÄïÔ‰ñ#IIJ>Ó$3ö§²[sBŠÃ§ˆB ês÷iÕ5"¤u÷±g¼ˆ ¢{ÄöwT"caLj"2ök»!‡¬¬ÑÊ–õMÈöË[fUl 9úõírËÓ:-s´ªyäùK7n"_¥y]Ô—3ÈÿÞZŠ%ñ –ÈþiBÖ£¦+(–á“ñGÅ %Þ=÷Éœ EéÞäˆçjd([R”ňòâ ¥“¸o©þ ß[T>ž¬£\ý ÷ó¥ŸÐ‹Ø@àö0?Úˆêßþaz¦€šcjͯ¦Q›Ú!ç»×êô;rï¥1êv2¤õÊ‹£~g±õ%¹»x¨¯¸íñŸ"4š\ì}âr M÷F$Ý,›B37kÒ+SLh^ýèL_÷s´ÔŒM„VþНÔÐ&Ž/Mœ¤?î*m‘eAû‹Á–¯ÜxdÎdnä9;:©uŸHú-‚ç^ËÅ9°ˆ¢f¿Ý+5úmŒ»'±>ô»±1^'³Üú¿¬dHCí{Ò^Òî©-pž¿RzûD ­CÒ¹ÀÖQ=H8gË&1õOîó³z²}ŽÑí˜xêŸ>uáº×ö~ò¯Š°únO‰ñTÌðÎóòçc­Œ®|‡Ò‰ŠòÀSG¡âs4AµÔä÷?¼û» êo˜>| Dc$#+Ðаö*_ê24&LÄíõƒ&_:ݶÐ䵑y8Úܤ¯j¸}„&5îßK¬qÐ$zÓçÚ4]!áŠ&¨Þêù—ÙÜΉ¬‹WÞBëÍ›ºæ¡]Ж—"fqìüh=§ö&(Ú×IÚ™\” SâjQ t]s½¢¸Ëº›èq‡oõ‘ÁAj‡Íàç·Ò˜LÍBè+¾¾Lämý —b5©]·Ïn(G?‚Á/PÑz¬†)>jÐsîøáè†4o2õ"Œ—ŽÆ]õ_…_ɪÓ%ÙŠð»t)q‚»¦©ýEÝ€w[#Ö÷'aŽyªôw=Ì÷VØM{FÀbw),g¿¢‰½f«­¦÷¶ü'Û/–³Kè¨"aÊÞ¢v‡·H|ïü·/$½óó¬”ñ>$•<éd”5辶é^! sÍ„©IÒ1k>ä‚ Gפ ÌÓ‘ñÅù|½‹2È´qYqïf²dTŸQoÊE¶lM¢ìûÈ1m#ü—Ò9 n±íCîòÙW^!ŸèÃÓjóš¸Ç{ûUÈßwëpÙtm ‘zI‡ÂbìúÄë (*®‡]_&Qܧ´Áþ¾Jª’¦§ö¢ÌÞás ir(ûcÞÒÖïÊßO{í™+‡û4­îœáA%¹j$·å‡æ=’‡W>âþwŸeí‚Ĥov—i úµ«¬†åQc„MÝç;j…·ùüÙèGÃã Áþ¨K¦Á2Ã|õJhɛܢïóZïÒt4´6žñ-"«VWÑ”N…«ZîšQÇ›TÇh¡¹”šÏÑýˆ‡uU}ê%Ñ2lå³"Å*Zý‘zh®·€6G>¦ª» íÈ5áA´{´GúƒÉ–šÙ3¼#…Ž*œ/çúåÑIàRY#!žýrû&…Ë Ý5y­HËÌo5:½\ñYLUË[úˆ/žö¾i„ǽß}ÙÞ—aa…wŸCï2ÚÃÛWYjOÞžüg?&ľÞ¨UŸ i}¹£·ð9. óÞnmxÅ7ëþN}Óñêž ÷óÿ퇂aϾp^€lz’fÁfÈQè÷£b´ÙΉõ·×9¶ê.?_¦ñþ÷óJ³¼|GírçV^¤å4~Ìyߟ†¿¼|> pyÒÃÖ÷ˆ6Ÿz•ñ¿ý…úëbæC(UÅÝ¢wþóÃV‘Æ÷öIðMÿgd×1#¨hÚ_'õÆ ªnwÛG”FAm錇"ÔIgJå’KÃwG›<›¡!óA›î½Ð¸ä<ùQ‚š½dä«ÞfCËF»Õ7Nh/rꆭþwÌ>ȯ‹ºöÅMWôm]ˆ&€>8Už?&Ð?j;âÝãƒU¬GÞ" ¯¤ø…0À˜îßŶþ˜x”ýrqù L6&6°É—Âç?¦Ü0C8öH*ç,̾~ÆS{æý玘Ô-ÀbÜr³\\",¯ýaì†5W•‡^Ô«°A0èÇ\²¥[s?½ÙÂ-ݲpbFb2o§×Wº‘Ä‹‰H‡ˆÉ„³_›A ¥ÒÌú3JH• |•›Þi»n…ßXDÞõû…ãYȘôuss«¿eÖê¾E~Y6ƒ8‚¯´"[ó…•÷’©ÈÑSùØ–Z9Õ‹Ÿ ¢D—ÑÌË[ùðs†KN'òÇîãšm@Á¬ãmlš=(B9£yòŒŠ}L¢/½€™‡Ec#£ô4c\æŒÊÖ9òº…[¡|#ËÉfꯨH:.,)•¯´ä3ú€ûÆ,N‘æ (‰ë¾J6@õôè{ª¨iãV¬¹RˆÚ,k¾Öç ÎÌ犀׻PwÊ€BtœõžÈ§_DCâ8™Ì%Z4æ0¼é^ቦƜÇ*¬Ðì×ï!n´  ó˜.BK÷™°2´–8xÊÂ.m–&G‚<šðð¤¼Ðž_è@»OÅã:!::Qò¶É¢“§VÑê <›)ÄsLo …)ßsÕçV"µÑê‚ïºðÓòÒp_‚4ÎÞMèÌù¾Õ*Hžk×ܾ§{zÜZ’kÏÔÅV)@b‚o;ÓŠísŒGè¶sÏrÿ?çºç.n÷©ÂÂõYÛû5ׯ%ÞÛ©Ûç‹ï¨/ï‘I³6ÃÏC¿… úžÖ¶•ðCÿƒÅ'ç‚` ä¨mzò VyKÞŸƒajÁl®›”ÿùá×Ë¡Œ~0}iâD’üºy$Yá·7[ðš#üùª1_/!3êŸ?¿ªÙˆ£vψ`>ž$ÎôŒ*,^´Ó_¤e+¥œÓLE°zäç›Ý·a=‹ñâAƒû°éº¸òQŸ ÍÀ_‹á( = »1¶ I|è" ´\­ïFC•1R^è®z»™…ÔYJβmH»~ëvD°2p ¦ÄAF½«=›÷)qpÿËÅÈb'uÙr¯²9äÞŸU>¡k›Nq»q÷U7ø]Fn“ò}÷Ø‘÷Ûd\JšîwùE£˜‰ü–¾(Hñó±I®- ®Ó“Tò È§ñšÝä(Nõ(ßã’9J|ûÖÓb‰ÒC”‚¿¬PöÍ)’ØÊ(où—æL@îãpéḕ˜8žŠÞòC]/ƶ1Üï+_œÚ†@~,8ð#ªŸV}ûÜ^5FfÚÓ£–ïUC‹3¨#TåKg„Ö¢KB½ðÌ&æhpôiNª¬!ªH}øÑZÆì#×ÏD^DSŠ…º@ 4Ý”ŠþU¡¶ÍñÎ._lC óþŒ©Zhâ/cøi­Æÿ¤Ÿþ‚66‰Žïbж×Èÿÿ°÷PU,K¸ð&çŒd$"‚ä¤0EP‚ € *‚’Ä(*pD1ç1aÖcN˜Åœs@0 ¢ED‘ãî™é³¹ýtîáÜwßú/k}Ô„ž®îꪮêžîÙ§Cd¶úèOp‚Íü­.[ JÿÙÑ3Ê­”ôàGè Hz›•´)$L.W¸ô¹¹oÏOœÛ‹zÖ)ûâŠçQ|¢»4Ýþµï‡©ý®žèq^¤å½a—ÜGW¨Mß_i ®¨¼7|¥O­Yüb½oco|ÿ·óͳËÔj©7¥ÏùãšæÎµœ´ÿÙºØß7>6„Ê›6lZVæjêˆQŠ?‡H´øÅ¥áñTÞ7%ïÐOÉüë>ooòÇ/4^ß2:yäÞ(þ8ñü‰Ôñ¥Û˯Ñmy¨ÚéÍy²?”uÖÖêA·I™a´qëŸ×ïwg=u9!p£ÔÆxêJCz¦‚Lp˺»‘ޝ:Q7–np¾°!ˆº•´x§ßfÖß}z4Dó6u÷¦ªuÎP~¼¬ô±Ã`êÁ‚TO}ª@5sëý‘Ô£Úõc>VREÏgO-ì6zª¹kÛ ª8Ãu†ÿq ê¥ö𮓗o¡JŽmŸ3¡W õzmAj³w?ªôV‚×ßÑT…n^§Å2åÔ»~ ?´ ÞÏñ´nÌ{BU==æÿîº+U}gÈÕí_²©šØ§=5Ž¡êü'^«òS êÜ…ðxÔ×K뢜§QÝ· œÔcÕôf¸Ÿ¨Oðº{Ýs±DV'ž­SÑ·½&;‹/ñàÎ'g~ʉejÉߦ€´Fû8iQS]˜2«Ù¤‚ºn1™Êš…î ¹ ²n†oég[Psòy·x&¨¿ÍøzÓ°4N&¶Z—ö=è5t-®'~öÙ5»7WfažªØí#·Àx¥ºÊ5÷v`º}ñG÷AoÁìÇT÷ÇÐiwi;›€É`yNïáb?°áê÷6Ôl™Oí•öO~{ÝÙœÔÅN»^X.s_ •º{ºù·÷žš< (/gƒ—uÁãÐÔ>ê™àYë}0b6t7XmyÌl*øˆêÙÎÚôüD§³¾z*G)¼pÍfÎ)ú½½7ô6Á)kt”Ê¢!äâ낺æê¸(·æ„]üæ»2zô'µ!E÷>ô8®WýìuGrwF³ivݘۇYµÇ0ëMï}Xî8®%uïàÁˆP«ԽȨ Š_î1¿‡FŸß7O*O“¤ô~ŸwýØTêáØÄœó‡{·|ÇWºò;ó;iE’“”—V£[iWuÝZC=IµöDþõ4¯ŸÜ0y{ªØaÌ$£ûaìz´¡r»¤«™ß?|5mƒƒEÞêõ‰0‹éûU©R…Õû½ÿÂú×^›¨©D½È³é}3ªtiº¾d—õÁÚ1ýyd õ1y¢oÀòxê“Z‡3gØR5|¯x9Qu#=?ìXDÕwÊ^š< œúR÷(ºó®öTCéì²¼=|hî$¯}÷>õãFù%I©UÀû¸ç¢.ˆ^Ü—dQâ´IûƒäK %O×&Ñ’Ÿ^}äÂ6•˜©•ƒÂ‚A†‚Ò;í:uPþvÔy̼§ ÚÇSìC¨‹,™”0M4Ä-«JNú€–ë†Â‹T èLì|ߺê è=^yj Ærhïv>³xüy0|º¤ÆD»Œ»˜Ùî÷“-ïg§n‚Á/ $–‡Žþ½ÛH3†N«úm™ –¡Ó4f\ëß1yU>`_xøCçj°7òÿcPwp¨šàt¥f8}ŠÝ«ïä_;l Ý"ÿL¸|s ¸—ä½ÓÑÐï«ÞSÏû³ƒSÅ‚À;^G÷ȃ8è!}<*yÅYð9–š!Ó~i‘§=l´ '5ÃóÜrÔ¸>æû›èÕhcY{ èÝžu/ˆApÍ‘ùÐÇ`âG˃#!4pB§s" lò¦¡I ï›ýºÌ˜á!•z?†þ eX„È4©#Ö|h%¹¯©J¢D×ÎUþƒë/\ê ÃÕ2ÂÝ\óÀøÏ@S…¾ÃA~ìÙ!Û»¦ž8EèÔ\`ÖoÒ¿OHÛŸ°½ÑëE…דæÏ së©ÖŠÒöIô‡h=+ý;ˆôï•Ò”¾.Léu¨­~7­7%úCÞ¨gÝ&4³¿HûÊK3=¼—Q—§ßj_’M]5Ì3«D]û²lâ]‘CÔÛYCÍ©[[S:<eêÎÑÀ=[múQw߬>ºäÝcÖÞ·w÷»¾b7U`¦÷ùjûEÔ£¦èyrG±¿kŠÖ“ÓvýâIìØ«ãkZâÛ.OÛR¯S–Oë÷zUºÑDôJÖ ö÷ “|zU Þ{=^Xâ\KUM•Ý2]6Œª^[,a䲂ª±4ù®¸%–ªSušw¾c·_ýQtkÕ¨¯f5k’ Õt¥V¡úFðœÄ"f)FƒÈ´-Ë´/È€èM£w÷q‹i×KÂA¢Pcå·L5ª› á¹FdÇŸÈ)q‰ȹº{Ù-PV:ä{Ff.¨,ö[xsÂ>P3Ýñ&P÷ ¨?ˆ,uí hl:WÍQ´ö¬Ôês tÕ ½SÔ²@¿þv겘&0Ü•º¾ìà$0ž[+¶yf$˜®>'›^fµíÌ4ý•¡ÓNmŸà‡‹ÁòöËìIµ¹`#“ü4}Ï[°}^Sn_ öoÞÊ×öÜNÆa>¡‡ÁeÃÄÔ|?ÓmP?û Í€êå]oîà‘7Flù(ðŠJ¿bt¼/t7w)ß=1|T3ÖÌÞ±ü4NÉ+„ž-«ºÈ…@ÿ¤7?$öBïQ_ò­¦à¥:÷û׆êùvKí!t¸ê$å?| ¯øôuËÏûA¿3WŸÌ+Ö‡þ¹ÓCÃüAä¶’wK¾­†E_Þlw»þ1þàé8ˆ.S½N2 †-žêh0ž¨ý i@G=ùÀ¾‡›õDjÃSc{H¯»¦íQx_mOÂë½i»¤í¶O:mw´½Óöp)*Ý-ÛOœº&o²®šºÅÿr rõ©[Ö—Ù-XÃúÅy~Ë´W°þŽöh_ÆÝ?;wŸÀøO†^éôáü®q­üãýúåw/ÉLgökп;ÌøIú÷ßʬ’Üö•zÜsQ·uû;³û2òíÔ4Œz^µx·îÒÔËõêÏÍXO•<Ø»ìuùcêu÷~W‚Po.v˜q³ž*›8>áÑsª"nJpâXệ݀§®Œ¢Þ‡äJ¤ïüJUÅMx‘3ªîS“ÔOÇú”»ÑpÏ»ÕT­JšÜîµ[©º£¾«TéPõ«V7Ž[ÒŽúzãZ€Ç©'ÌïŒþ0Ÿ¿iðtà­Ž:1&·ˆvýYֱĚæ?”ª4‰O÷Ò/»´jôõ{gÄA¶Oà£Ñ½+@þ€þ´ô‰ $ò.&ð´#(çìúѱTµÊâÇž4µ+b+4{{B»3ƙɀf™T²¦ógÐÑûª™5ô’öO_ûH n,>*¥’†¹¥ã¦ÑÛÅï¨ÄB0é½ÚQë6˜¾9±*ØPÌ.ËyûˆNâéÖ‡Î$@ç#Ëï—­”€.·¿6­¬}¶:ß+»¼ vÅ%õ×-‡ë«ƒÎƒÓ^ýÐÆe àráÎä¼ÒÝÐM&iMÅ¢à>ëî¶=>GÁÃø¸TŸ@ð\ëñ¤OUx[¼ìÜѺ_¹"Þ×à3ø¤Î}ÞÎë øuônˆéeþ•"•_/l„€S“—åM.‚^9ó»ò‚&Ô1¬œÁ'\†dÊ<…’Õ6Ë–î…PÙ» Ï—–A˜ï¸Úè9ùÐwÕÚïnS'@¿æmçoH–Bÿ‰×o†Èvs”»†WdÅ^+„AË U\—ªÃàìmR[DOCÒ£“Ë=^¥ƒÑø’Û6šÈ;–?ͤt©¢Ù6×0û„÷YÐ~Pø÷~…ãP’?¤í“¶Gú÷„÷¹wÛ0Ôªõ~ !ØŠþj?†°_”²˜'×'Ÿù½à3‡'itZ3–è/™¬¿½Tÿ /÷[,ú •ÿy7Û? 8™‰+_hÕz0û­Š¢^T™ÊÄ»´¿ˆÍÛ *1>?,~€[Ëzñû‹”So¶õsœþÜ•*[óí’ØT]ªâfá4…½qT¥…«ÂÕW™ì¾ ú÷½ûOÛ¸óL$Ug'·x´øeÖöµÎýÐÍúvQA­ ø6ÕäôUß…ñTsáÔ$Ïë ÒÕÓÀw¥9ˆfÕù´×¯±ù7ji‚ļŒ#sÚéÔ6Ƀ‰{Ë@¶“±db(È—®^3¥j(ÎÛ/• T¢Æ4<ŸªŸ³ru_ êÞÈ’{ç {<½Ó*@+ÓãËÚ3@ç¶—‹ÍÝ*Ð_ýrÞ;õ`8vqÜ€ e`Üçõ׈™)`ßu‡§i˜]m”5®æeË6î –Ǿîbë Öß.N>´Þlï+¶Í[ö¤òÂ#v€“ÊD§‹ƒËü™£@®ºùV׬øø(ÇWÏ®ïÊ£_7/¯ {µßeWBw-褣À‡·hònY{þø°ÒùÔQèÙîÏš“½ Ðö^q£Ÿôîkƒî+Að´û×ou æÛ£ÆmGËo©q¡V¤ÂjDŽ^ýšýv­x»bÌdè?;C,C7"\µ¾æ/ºp ¢ Ö›ø¨ùÌJˆ~äöÐzÐAæ¶nÏÛˆb0Ù´¡oEðN‹ï('?€zº9fÛ™Û ŒÞÒ¿—½s«ÝdϪK­~ûßõ‡´Ýµ¦9;Ê“‰/4ÅT~5ëN]I1y­ù)‡‰oÊ TÉßKÝÖíÒGzZ u×þ‹ÁtÙt&låÑ}:^¼ã9/vi\ u§ÆJÞ¤ýêî¨"qEE¬?¥÷5®‡œ#O•™ýŒ—<½­9É“*Xó^În¡ ;~ôLêv¬8‰z\üòåýeåÔÓj‹-ö÷¶ÑþDf_"Úøfಙ½ ¯Seª\ƒ_ˆµ¬«É•Éöf~w»òë‚ÂôEÔ‡ /jê©jÍ{‡v’¡>õY?~2Uslϱ~25T]жU2RYûêu¹Ë³÷æÔ·S›$?(,¤š’ênÜ^Çw뛲¿»=Âk{~¥H øîiRcÇKǦ€Ìž»‚ÏÝy^}»u_M@±o×k–·z‚rlQÅ=ÃAå^P¹{ù+P›ö^g‹ß\h7ÎÄ$õÂ(М©6^·£(h¯‹È]ÔõOнSÓÅÑ tí¦œªCŸ1÷¹Ñü—Jƒ\ï‚ñ‡¼ÁbÒU`:qÎKÞd0‹œja‘æ›E–=¼à ‡z…«Üš]|¨œ³ÚlšŽ™­»#ÚÃü“Á!±éôh×(pãñMÃï%¸lÞ~GÇ èÚPµ²hÈpO—ªT–ÅÃðç3ï¶æ6ðú4ÙU¢ñ#tÏ®]ßÿH4ø˜Ü¹jôaø>1ž±éL*øoºpI}çH˜pñ‚íÉÐ+~Ò]¥5uû¡Ì< >ðªÑ ]*„¼kå˜6B57ݲÎÕ†°˜Â¦Ù5jÐ÷|ÈÆ/ç4 ¼ë®U/v@ÿ›áOæÏö„È Çq ½šsg¤…¨ÎG7~—Ñ:'žôzîI“¾Ì­R=Æóâ-ëƒÜ“óüA=QÈúCR©Š±Æ"ûöƒ¤qáÊ×*—»©´¢t:Ú¯Òq.ÍçÀ»Ò©e¡@:7úÉ÷Š×Ì<‰Ò~³Õ8•ÿXÄÛ@‰9Ì|½/ñTMÖ‹UûæSù×ÌM·lO£ÎÎ9W²¶:³ÿù·ýá$ÍÉfÔí7(œœËúCd×ô>efÿ¡µMŽÆ¨ýìv¢Ì~ãç‡ÖŒ›”@½¼Þ;ê«åVêU‚±ËßêÑÒâsUÁT™ôðñé‹Ô[öMIoªÞN½K¢ÞÇêA}0LN K´£ª•GÌ9‘ú”w×wN¸/U›!³°ìóJêsÌBµK®P_¦ïÍš0j¨‹ª(?¿”ú>ëPRêg'ªÙþíc­göì¼N_ÑkÒ›g€ØÝ•Ϸ‰\ÃÅ÷ã2AêN{õ­{§lÄ5½üú±  š“5K÷ (=½(ž<ÂTbµ·”ÕOGóЧìõõ£›v̼C>œ=@´úô}|ÀUt6d¥+ŠÌýqÙ½¼N ÔóáV9`§:È LKiwË :Ê-¾à?ã t*Êê29´,÷6ï·ÓSå‹oú,Û*ËgÎûâãå‘àXûæ~ÕÛàbÛ¿âywèºêEJL|)¸wZ£_oïðÆõá6×ðŒ=ì;*ÿxUÈíÝg–øO{2|œï´ÿt¤|¿Ýýlýe+øß•ý‘hºòró3z.‚^ŸµD•ìæC°QaúvçBÜP8ggô9²GeÍ£!ÖqÑ…A Ð7¯]ˆOl„Çh¼«ôM‚«zØ ²ÎŒì’ƒƒdÍ„Ä{Žó">_åˆl§Iß@æ³âû9ÉÔ“‚ Î’ë¢XˆâDzÿ½°?¤ý›ð¼ moÂvHÛ1mw´ÝÐûŽiý¿ümÔ¸*'êÚä§çL¾lyoÞ/ïóqvœˆüã‘cü ò{w&Œ{}Wëéùz¼HûS´o÷~wÊë¦\eííç/ˆ¸·–ÝÇO7Õ®|Õ³õtåóŽK¢®PÅWî6û—9S/¾¼0³(èI•xP'â{¶ì¿ÈÍ*÷bý¡wªQÓ±?© ßï9šUìø0:vÚõU/©*ßž=ŸKRÕf9u¤’Yøç«îÚ;VQu~p'^Ö–ªw ÞÚP›N}̓çb©oÍñ/o–£švR߃QÀ³ ¨=ºDV×G+É€X‡Iž+OçxEÌóö¦ ù,ÒM¤º d$"õÖ¨&€Ü@ÃìÑ_úƒÂåÍFÇ|AÙz«¢§ã£–÷óÎÝz‚Ú ß¾EíçC;O“ã·&‚¦cÚö\¾>jÛØÞÂô÷UX¥q ß8;fÅ‚Ñä¾»¦>ãOË•¾ké€élJa»ù 0âirèÏ`¾(XwfEt¶>ò hã èÒµúXºÄ!°¹•jµnØ-ŒÊŒ=vŸô_§aƒ<“Nu—U+}Uú@×Ûwé÷¤Iu½íƒ‡Ôôi¦qžà9Ê:+AНŠ#“3‡ÄC÷ñ&Ü5ûv·þ~¡1à{ûìò=³vÿ ·ãùrï Y.z{äAè^8ÄÄ~9Yîùì0‚ÿŒÍ>³©B»4ööЀP…›ö­ !,|Àp#©2è{¤&H62»„]hýÏXõÿ~ñãíû}Ô〄;Óž=dìCØþHãBá8”¶»òÛô·~Öd(mŸt>´?¤çhxØYãÓ²S½[ù»VþïwljÂþÍŸ¶'Òûó Ío÷›âÏìËgæM,šsööŸdˆâbf|8b°ù ›ÈÖãà }{®YB=Y°sžÁsgêÙ3µãØñ¡„tÊ}¼ÿxcÚ°òPUÚ°³ƒn“ÏfˆY\½z–úàÞ8ÇdÖSªÚªéÁFã«Ô§'×Þ”ZSµ+6í]ÒPE}N‰2èú2îÙÁ…¢CÙ}ùaù>›¶Ü š/êÈË*²þéùRI¹Ði‹š@Úüæ‰GŠ[@v’³ª™³7È¿ûzÍWP½lÝyž-¨t|‘ý-í¨ÞÛ6&)k¨•84Ó4NW«OÊÒ­7ù·Õaè†T˜~¼[³“3ÎÆ€á™í×'÷lÆ[uJÌ¦×Æßÿêš¼OÏR ûA¢ù‘F`5­ë¥uÏãÀ&eºN­”Øu—MÌßrܽ¯ÇÖ,§AïÒm½‡€ËöòŽ'ƒ¡›îÒ+îw‚ûñ¥Õ¹—óÁÃsVé™»ËÁóäS›!ÉwÁ;àµ^sè^ý!óz¯Kà³sÎeǃQà—î³ÓZKzöOJ¬²ÆÎ›®ØQºògämö«¿ !gÖœþi+„ìjŸUSa«®.œ_>úYÖŸ>ÂËÞ¹¼»"Î$E/H‰‡§n']¾¼•xmnš‰î.3­•ÀdjÊŸ¯YñÛ%Ý$S¯k¬¹/ó‘ùÎı^UÕå7˜÷ $?(ü] aHÛ'޶oÚŽè÷ Ty'zDŽbôþšOóÚnæXH·FcÆÕMáÛØùOz|ˆü2~N/ì?éçIãÄ‹G>_;ZIHÿ}iŒ2õ(pËÇ Á]¨ÂËìîKæP§g)§›ÆúÃáùÜý!ý zßÅâ–`@½;¬>}îeQêý•›Ê©‡¨ª}ƒ—[Ì˦ªcN\=f¾Œú4EÁÖð”#Usû…œÿ‘×ìøPgÿ²±K{Q_m ׬ê”M}›•çX;w9ÕÔîŒGÇ$wªùœf÷½ÏS@¤Ç°X÷¼ñ ú`éëx¯R_­‘d¸ $§œ ¿«Ò³ -7¸€ì¾µù›:°ö6+K-&TTD–]Õ£Ôç<èê‹6+¿Èeç_†uP-ÿ²t&7¾ z…{5Ü7¥Bûù曥ÁH-ùrÔLãIƒOÞš¦ª›+¥Fœ€¯»ÎèT æR»­úíù«\FkÔ‚ÕBûW—›áMA¦à vÆÍ2_GMû·3§<÷îŽeñš#W¥‹Ô§4ýÍ· kv+ÂÁíä$¹Êù»zïy:ñxŠ=é_´¼Æ[5Îé Þ ®2:NÐc‘ñɹ[àë:Ù"úUwð«ëåø,Îzž™]^X– «;Ü)©ß½§ôéZ}»w““[´0Bb¹–|ÿ}æk›¹O,ƒÐ›}>OXyúþX{â†,ô›·z Õ!ú+ï4𿦠{óž„ê¿‚‰Ç2›t”`G¨c„þYìüüè÷“`˜×iñ×¢ÇÀ(nîXÉa1 ›–×÷U4õx­‡ö©ÐRÆïÐöò+{žý]HÏÃÒ|ö7oØ®s~ó‘ö‹Äñ`[ûCzÞtmˆtss óýªËþ/Æy˜$1ï¯Û¥j.Ýþ'ã™÷)ôûC:¾¥íØ{Ñz#µùì<ú.Õ“k9²Ç“ü©bo¹«|ƒ¨¹ÉAÆŸ©Ûìù e][öOÓÿx¸õüŽîáÞßn;³ñlþríôPêcÝÔÕ©çï°ï9¤œéÀŽúìh<,/|0õ%äî¼ë+¨†²Qß©L5ï¡™ÔW¶² Ÿ¯àÉÂIÖ2 Úk±u¶Þ4{÷>öóö qêk‘ñL#v|¨ºW=ä|gö¦¶‡ÿ%P,ßxS=Å”ߘšÄ¿ÏÕ•6Ae>• >{ÛÜH_]ÐXw¨‡M£*h]×/vXÖtm ÖÞ¹ÿÈ9Ö`¸/@£à’6/Ïݧ¸°'˜žx.?Rë+tt[$ò‡HX´Ï9øxà°Òçâ.™å`3¨ôšmûJ°ë²oÇž‹åàÐa}XÞ5pò®è}l²¸Ì)õÓTX]g­V¬¼î+"˾xVÍ¡âó§‚ç¶U»_÷’oo›Œn¹Ð½Ðpß©±9à³ Ý.¸Ï9ð‹˜´\dôtz¶êðd5ì@Mßì§Å¾GŒµ’VŒh†šuREz*!á~Á ²DÝMϪA?•ââìQ[!üRé‰Íî» bÍ ‹RýaÀòÝbw˜Â £ë\/€Ä‘áuÜ\ÀxUWŸ{ö @Z*`Ãâ=éÔãªÊ³›U¯2zKûCáùR’=Òñ)Méø•NOšŸ¡í‹¶ú="=O*<ïI¿/¤ýs éyUú:ó^ùEf|)ì/‘¼?t@Áö¨qL\I =”ù˜Ð“*ʹ1îŽÛÔ“ÝFçò×og¿÷„æK_œz[ti­óþáµLÿæ;[D¨7+—ßT³ÞÆŒË/˜ž“:ñš‰/™ï¶õ;"7Id9ûþPîìÄÚ2 j‚v.r̤j­åµ¾Ì«fæKëÅÓOìù>‹jÈ:ö}añ%ªñØ›Iáô÷Û¨æ{®?>Dý‹âS&;XòwÛm…AbVþ¥Ð¯k@j@È鬛Y ³º.pm1ÈK§­±QÊEmkí¶ƒòÚ݉jE6 ª·Ël¶s;P»¹W/&è#´‹ N±’)ÍIâÝ7ø‚ö5³ž=Žœ½Á.™~®`°ÕÙè³ôj0,‰¨îc)ÉÎÃÈßÉ›=s3˜ÜÙñéP˜÷÷œ±Ö¦Xˆˆío]têë)kI¹-=šªÁ¶x¤íWQE°Ÿ©™bÑN3æn°\;œGš‡l×뺩á'Û[§a>ÞÚ{tÜr+=ðØýbû¡#À«ÝŠq®ÜÁ{Q®ŸC¡=ô0p‰xVª >§3œ5o¿‘Ǭ.m¿ =­>õOȽßO.ÊÏ€^E‹‰fÔ@Й1†I+Cp³Áýîþ ÐÇ8÷ÎX?+ˆ9óÔù„m5ë6M& úIYŒmgù³;•+€µ\í©[Cä¹9^Î{ÃÀy¡_劫!jì9ƒÔë 1vë˜_?ÑÔ›WËRw‚¬ÁØa“ä©Ç§}Üm~ÜgüÍ®E£¡]É4&ž‡ö‡´ß>ž/~HûEÚîi?ØÊ/"JûAæœö‡Bó¥4¥ß# ûC:¦¿ßxnœ^ªãQv|HûC%‰‘廵 ¿Gaæ{èõ4èû­æIÑ÷ŸÂ¸j‡ZIª89.=pßHf³>Í—¾Ž)(Ëé•ÈÌ—–ß+÷HXz÷2æcž}õî|QÝ\Wv=\ß}Ï”ë†Rµ3µËèÐLÕy<–³ë¬DÕo Øõð)õu¶ªÇÜ…G¨Æ„2Óodßû°zš¨”"ÇŽdŒWfßs >*wùyHÙ¹íº™÷d‚;xîú” r™š1'Ìûƒ¢ÞÇs;#û€òx•ª™6@ÕÑbÊ´ŠP»½x£Ä˜÷Ðî–ûá=³%@Ë2tøwûc 3ôÝãÅ&€þÔ²w‘¶î`Ø3Ðy¤çR0ªÌŽL{˜ ¦iw›¬k€Ë:oö]½:i››^Üx,fÆŒòë0‡U+u†m|?Ó°;ž`ï£Õ];1²F/´Îãϼ°Ë·ÏÍŽ{Ÿ?·ËÃ3w|ê­ÊÌ ~ïÀãÉUצyඦòò‘ð~eéªõ# zL9ÑÑn;ø:ö{ßÓ:ü:žî\í=oÎírô,Ö—R3²~@uO5Ûä|mqé‰uû ê·I£Ï@èØCZ=ßAXmúË´­» ß‚”ªÎèï? ¹iq%Dš‰¾®;òÆW8¾ƒavêß´Îúðëýnöû­ #ÃËIÙ@=t¥ÞƒÇÄutœ(¼®Mø=!M…çmh»þ^¢°?¤í‰ž'¡×•ÑëÇ?F¿ÿCã?æù5Æ¢yÚÒãJš2ó¦ôxͯÐÏ3ël›ìÒ8º‰]‡Š¾Z¨S0ìÑªÕøðù “—Zþ¦ÔË›Š#GS%êâAÉë²ïi¨iù`Ž%EU¤õèºùéBêíÎ'éᯨ÷Y7¬×Ï®§fÐì\åË|±fZi}YfªÎutÀËG½©ÏšÓÅϬ^G}õxþà¾IÔ—ï°"íR³®¦©nòÉòq߀×(ãh2i-ˆdüq7¨÷i›Ñw£|MHL{”ÓáHë‡5K‹3v#?O¹„A¢îù‚œu‰[ƒ/ƒr÷འAeß°sæ‚Ú˜ªAí^ÊA;¿¼Àa£¯€fÈû ïSž‚Ö ™šWóŽî½‡¡ðÁÀÁÀÃôÌ50\eWzyk#S+eç“ó²šß—tX-éüg÷&0×ëµl`¦!X˜Më;­X,'{Ïk(ý 6¾^1ÓVMîÀ¨[ À¾omné¨hpôv6°›Ρ[ÔÜî¼×™ÏÊËç^€n%&Ó˵j°3•þáxLκ±Ó$s]xœ(ìéq"ý]SÂ{Dáu5´?¼²óÇM'æ{ÇtÜ,¼.€^'÷P_)åÒÆVëÇ™uã«r¾Ž”bæw^¦>=0P!‘õ‡ý‡úg$ï Þ€z{Ê@êMI·ð [H¯ÿ0`æµ7ݨ÷猼NU^²c|b)U3/`E—ù—¨ºšìÃÓï9PŸ‹dæZ:JR_G,ï|¨Š×1]•9aàzàiù%_Ë¿ "YJºª úüLÿ¹çAB%îÔâà 9¿ÎLîùd±s­ê³­ 䦇µÏz=väÖ=ïœÊýt^ܾTʤM¯G€Ú™±¹ÓözC»ýôŽLÍ;nÑ–ûí@'f顬+å WÖ8üd_+0´é?lq‚›N¬OhŸ&Û ®ùÝÛeÖ:…õ…Ž5Êó5R<À’§ðzqÒ°îb0môÞ`Ò.º«ö]–ú¥‰‚£­ÁjÅà™ªs1º\wö¬ö|nƺiÒ¾@»±ccâEð¸XS®óq3x¹íÞô°Â¼¯êØ&γñ‰#†N¼¾#‚$(€_áíí݃NCÏÒ‰Û’|!ðöŸ›!HìBýQÍ|Pa“×»uÒ¨ûà*?N};eÇÒr+ð+}ºì=ôùFÿSùпÃâÝÚg âÇí5§öÀï•3§Ô­„aÆs,åÖê€Ñ5Ÿ»º€Ì¤q%CRõËn>ÞØÊÒãCÚŸýj7mw˜܂ßšÁØ)žÎ¶7:N¤õŸþ~ïÕ{¾—½EòÙùRz]5íéõ3h\ÇøE¡q!?2TÈ_¶zŸˆömÜ?y`Õ©s™u§­æ[j¬æO><•zjÔqKlûÝoz¾´÷÷OɈ°ëÓfŒëÙ‡*‹®³4e ùý!½n;ej×¥WØ÷ò²Šêóó†RuQi®!&S¨Ï×\¿ÜðßK}é™p¥ýÄÔׯݻ”ïÛA}{kº®éá8ª©ëóq;†1ãC‘…·Fi—®±¾o7t¿¸$ì>_ö[> ¤d̯¨­réׇ,rÕ@öÙm·æY/AÁ(gÔ×Zfÿ…JˆOUÇ÷oZ¿§lBiL¹ÊΛ>h?sm诪²Åô%znMlWµŒÎIè;Ï“¡UWO¥¥A‡n£JÚWÏ‚Ž!ò»žöIé}Gƒý¾·´|æžO°ùb8s앹`7ëL¿Ï߀C¯c'—ŽÍ§n+²ž–¾—°Ž§6‹†®kfGvZRîj?ކmÜ p®ý6«ÝOÀ3´ÃèẾàõz­tæWÐ}騕a;‡ƒÏÐÆeÝ_y_xRŠäN7è9êU$1wÅOQ ¥KÒÚ6|+ͦjKèã¾³JïH„^Ô7KË2…¾ÃHXÍwp'¹­ }] ¢£ÁŽ«³•`€çÖ{näÀ Ù•ëË š÷:Bg‹Ä„O›7†\¿µ¼Üa'$u»ÖgLíT0üØó*<‘™ÃùÝ/Þ¡Š¬?I¿˜Ä¼¿þîûBÚþ– ~p,ZûsN§£ó¡ã\Ú.éýP4Ò{D¢ßCý3_J¯§¡×™¢÷1¿ò‡­ÖÕä„Ü“jhdü!3_Z÷P"ø€,cÏÌúq¡x–2ïÿéu4 ùÛ)¦ž»9¦†]GûCz¾­§)½Ÿèݰ…*_Qœe­¼’‰g+ß¹®^÷üÏÒëLѾ‹Úî"]–ðØ÷ô{ÿê…WûÇØQM!¦¹c$©æ7»wá"†.÷¢Î$€èf«·E;^xÔè«Æ€dŸn3]R?ôÄ *›Ï‹€ìùÅÄ‚8PpˆX&Õ3”ÅGö_&ÓTvmåÇÐo@mÆQµ¼Ð.šlðçPÐÛ4hà,[о¼îÍ­¯ 7uβñcï@û“ãúιF›LxJF€ÉÌ{á©*ZÐaïÔæÙÇ ÁÜJäùâÁÐYãÓ¸˜=Áê/ʶ÷3°Yæ³þ›Š Ø……x¹ÅObÊDw8 ŠëþiôpÙk§_¦Ýœ=L6q÷¦¡OêO†‚Ç9“3ï£ÁËÜ$ýÜñð®~¸øþ èQúBmÇR7ðÏ ÖØ=ý?‰¬Ûg.eé^As'l¯8B|6i°P†>ï"2g6AØñ·ò=z@¿C+z”@ÿâé³ô2Á·èǼú0èá×'*ôƒè~źi@L©“ù÷š HÌU°ôÑS0’z=ªs¥5H×<ÚXúÈ„z|v½oñÜNŒ^Óï×é¸Qx)i7mÂvHÛ/½¾›^¿F¯¿¦õŸÖ{f^äµÔµ•#öµ^OCû1¡÷…Ìz”Žö”~ß@¯»iµn•~ŸHÇ™áŸÓ/¬Ô¥ êü̂ն°v…Ö™¶²+z_â«‘–EUã©×k®<»ò8U:uÃÇm{*OZ±èÌãQì>Ä}¿ûû±ëK»[´O™ùœyQÓèåP×Ç‘YOóyy¬"ïîªþffIÓHêk¬è¢³Ã©o†ÏãìíÛSß{õ±É ó¤~¼]ÿH¶<ƒ'ŠîíÿcÀœQ ž>Ç<±Î $]tgŸ¼É÷‹_]¶è¥{Ì“ùgãM“@^,«Ÿ ôýÖ°·2p4„ßšbv·æ;D\>WxDy xûì´¢ó>ˆêjØ3é`9Dzѱ÷^ˆ9îµû³ì&k¨8bà0ò q|ºöÈtßµtá›YTѽ]ÃLeüm/¤ý½¿ò‡Â”NOÛ/ý„øþ´žôï…ý#ñ=¢°? ª\ßepyÞ”~ˆÖ™Òñ2óû(NfÞÒñ,Z'^¸×àåçZCf¾”èé÷ÿô|)½ŸØoûÓÃ5T¹˜ï©†éÉÔ[©7ãû¼èÇî»X±!ߨo*»>|•›~fïûTmà–´N¨Ïýwœ­¹C}™¯™5U”Ý?ì¬(WÄú1[þǪäÀ[è|õlþUìÕ!¡—ˆå;ë±$¶­X6¹±¤ÎZE1;2ß»N0æ)‚|j¬„Þ 'P²;¹ê\­,¨tÞÔäÓe!¨–Åwr³ÚêÛÄzý1"4²ÆWL5; ZbO&Œuݤc››=_ºŒ#Õ ãÕ—zƒqY•˜J‡n`ú6GbSçiÐÑþtÅœþO¡Ó³Ë%Ç®ƒåz½ .›€õÖTó¢ÔR°ÍÊ]­iñ ìCž¾mVŽa~+=ÎøƒóÌFÈÛ—®Õ½wëÕ+€Û^ƒJöm€”œL7'ðôóÛnjø¼ÍN7©ì€î÷'j™5xÏÝ£ñ•«ÁïKßóó5 ÀoX`ÜYЫ²ãÀÐÜÏ<øÅ™7iß!äýuÏ}ë!týÌ4ƒäoÐ7c}Hr7CŸ|/uè±qÒxéáó20Тë‚Ì—!êaÿ;Sùöø"Ë÷± O\”Çmæñx |ˆñ‘ÊÿÏSpO¬åžº§ŒîËð!nceKÛZ9ÓÇÎVv.ô‰”‹•½“#}*bKØÑöô}àH8ÑÎô }àŠDmm˜#[æÈŽ9²gŽ˜#GæÈ‰9rfŽ\˜#†‡ÃÃŽáaÇð°cxØ1<ìv ;†‡ÃÃŽáaÏð°gxØ3<ìö {†‡=ÃÞáaÏð°gx80< †‡ÃÃááÀðp`x80<Ž G†‡#ÃÑááÈðpdx82<Ž G†‡ÃÉááÄðpbx81<œN '†‡ÃÉááÌðpfx83<œÎ g†‡3ÃÙááÌðpfx¸0<\. †‡ ÃÃ…ááÂðpax¸0<\® W†‡+ÃÕááÊðpex¸2<\® Wš‡˜­ {hËÚ±‡öì¡{èÈ:±‡Îì¡ {ØÂ­¥ûE—ÄGÆŒˆCÇ’ñqè¬%…'ÂÖ•Ñø–c[ìØ;¶ÇްcGìØ ;vÆŽ]°cWöØãëˆñuÄø:b|1¾Ž_GŒ¯#Æ×ãëˆñuÂø:a|0¾N_'Œ¯Æ× ãë„ñuÂø:a|1¾Î_gŒ¯3Æ×ãëŒñuÆø:c|1¾Î_Œ¯ Æ×ãë‚ñuÁøº`|]0¾._Œ¯ Æ×ãëŠñuÅøºb|]1¾®_WŒ¯+Æ×ãËØ™¸ ch-Ƕرvl;œ¸x‹o± 1tO,&6±1qéôq1£OH´<¡ØrÛƒ'øƒjõPF~¦“ÄÖô©<áôzˆº!:î¡€Î@tq{]Î×ÎÐèúæ†ìŸtºÿç^Áùn”nßPÁù”þ༟Eö8„žË‹?û“AÏóÜ?Žò9©+¸ åwº^pÿ Ê÷lbôOzå®`ÃOzñ¹àõR@¿‹û ÒKˆïe}Áó—ÿ+sÏ_Eå¸úEðü5Tžëq‚ço rݸ/xþ&*ß-< •óönÁówPyï*( (*÷Ý ûÒ{¨ü÷2ƒÕã¾Mµ€¢úÜ/_ô“>@õz°ÞF@Qý†ÝPTϹEõ-8/àÿÕûQº€ÿ#TÿBkÿB$‡ÂRÿ"$¢µþEH.Cü#ù<‘ð‚äô䬀ÿS$¯§ãüŸ"¹=³ð†ä÷ì€1’cñjÿb$Ïç!þÏ‘\_H ø¿@ò}‘/àÿÉùe×cŠäýr…´€"¹¿üÔÿ'-Aò/ Ü! ¨J¶PÔ¯ÄüµË«¨UŠÚçÕñ Eíôº‹€¢öz2‹'øûÁ„õžÍͯÇ77¿)mn.ëÓÜ\~¦¹ù­esó»ÕÍÍ讀›?¤57W•47W57:ÕÜ\Û©¹¹nEss½xsó—Ôææ¯Ï››¿477onn2knþ±T½È(¾;~Êïøå•Èãû&<ž4¿L2M<ž\'_ÄïzðxJy<~åUçó»Šo<^»Oƒ¯ Z^<žö>O—ßCèñGÿ מ?T0¼ÇãS<žÉn¯ƒ6g6ƒÇëXËãuŠáñ,nóx–Ýx<+¾ ­5x<›©<ž-_Nöƒx<‡<^‹‡rÞÊ㹪òx]ùíÕíçÉãQWý•¿¿ç÷UÒÍ‚s«–þ«ÉJ’ÇëÒr.*8¶FÇ6Ø5[ìØ;¶ÇްcGìØ [Ñ×mé7Eèþ•å~î3…¯žùͲ9ÿ_*› Æ×õ·ËàÁãý_å#tÊ·+Æ£[ó_ïѺ䆥q'èÕï댇4©î¿ß¶ÿ"Û è¨A„'ˆd["Lk[럱‡ÇUC@VæÑüßYËÕŠžôóxÍU ù“áè˜Q qc ãG'l͘é&þ•Z g(š–ü«"ÿªˆj†±1iq†Éñ†q£ÒFg$êö·%êÆE¢?KŠé¿±°þËŒæ:-fÈÏ¡&–I 12aì~,.Ö·äÆ÷¾¢P¢Ä0ˆcÀ ‰A ƒ4 ²ä0ÈcPÀ ˆA ƒ2 ªÔ0¨ch‡Aƒ&- Út0èbÐàÁC{ †Œ0c0Á`Š¡3 1˜cè„ÁCg –¬0tÁ`Áƒ-; ö08bpÂàŒÁƒ+†®ºapÃàŽÂ<0xbðÂà¡;†|0øbðÃà¡'† zaè!C0† }0„bÃÐC? áúcˆÀ‰a†aˆÂ0C4† C0 Å‹!C<† ‰†aHÂ0à #1ŒÂŒ!C*†ÑÆ`HÃ0Ã8 ã1¤cÈÀ‰á 0da˜ˆa†l “1LÁ0Ã4 Ó1ÌÀ0Ã, ³1ÌÁ0Ã< ó1,À°Ã" ‹1,Á°Ã2 Ë1¬À°Ã* 9VcXƒa-†uÖcÈŰÃF ›0lưÃV Û0lǰßvbØ…a7†=öb؇a?†b8„á0†< G0Åp Ãq '0œÄp Ãi ùÎ`8‹á†ó.`¸ˆá†Ë®`¸Šá†ën`¸‰á†Ûî`¸‹á†û`xˆ¡Ã# …Š0<ÆðÃS Ï0cxŽá†—J0¼Âðà ¥Ê0”c¨ÀðÃ; •Þcø€¡ ÃG Õ>a@1µèWÒ)´ãÇ"íXHi²Õe!jÂB<Ž…Ô2ãXhɱ,gñ—8‹ &ap`¡Y%€¿Š’¨ó}];iš]ÐæÇà:™,Þ  Âï#•–  Ê/[»½,Ô£Yhd±éÄBW…N" -,T¯  Î—›¼B_^§ä÷mj/Yˆí@f“à ½‹,¤CÍSÿå›pÿ˹ðuRúÿ¿^ÿÝ?ay’äý«çHé7®|ÿÛ®·•þÝüuô÷ß–[åóÿÊuRº¶n—ÿ»#ýý§êûOåó«çÿé~•”×ëÿT¿DúûÝrrm§·<\óÿUúÿ»kk½úÝ|¸þýS~ðWõúÝòü§âÏß-Ï??W=n+¾ÿîõ¿›[éI[ÿý·Å-ÿtû¶U>ÿi?õ»|ÿ©þŸëß¿kß¿[ŽÚÿþß–ÛßÍÿ¿ÕNÛÊŸÿ*ß¿[ž:ÎÿÝòþ¿Ö¶õõÿT\ôO÷ó¤çþéöâšþ¿•ïJþÿ®}þûú,†¯£¦×MÿïøÇÿ;þßñÿŽÿwü¿ãÿÿïøÇÿ;þíã–‹„¿ "52#nDòÈt*‘œ–7š>‰‹IK$l”Œš–Ü’}tàg¸Žo§Â·_áÛ ‚1à[·ð-WøV/Ï· <¼œx]ð­÷$Çe‚×—3®·x>¸Þâ}.\ßð븜q^x|…Ë×sÜ_àiðºãº„Ë×%\¸žàºŠÛ…x»ãŸqÀóÄˉë-É/ã2Áe…׋S‘>+óÂÚ‘™%@„R®ùüü/®iÚ*ÿ¶âË•×ßIßVòÿo([[éd[µi[•ÿï¤ÿ§ÛâïÔý?©“ÿ„ ¹êÛß)ç?‘ÿ²¾ÿ„Îüü'u•«<ÛJΧüGÇþŽns•í[>ÿɺÿmúOÔ¥­Ú¥­Úôù_¦À[îýüö¿ú°üoÌyÿÜBé!ÿ󯕾Ùòw«Éiù[ÝŠóáýükE¥ìþµ¢¤ô¤òÈ þÂT='LÐ}aÊCù S t_˜’ø’Ê)‡ê'LIõåš?©¾¤váÚ^\Û…$gR9Iù+¨,*¯0•§å!DIéI|Ñ}aJjGR}I|Iù‹£ò SR>\å@Ò+_R{‘ä@ê¸êçúòQA墓~~Â\j“0å,O®r&åO´÷6ÒOB99ËPb>¤z‘úRz’þ(¡ç„)×öâJIý‰Š#y S’ßáê׈þ…P‚ü‰íH’3W=窷Äö%ȤW$ýáêß¹Ú)‰¯(ª·0%µ—$º/LÛÊ‘ò!Q’ܸöÏ$¿ÓVý!×xƒd§¤ö"Ô—hG¤vCr¦ûsÎq‰’ÊCh/ø!à#L‰ý©½¸–ŸkÈUHz®Šî S‚Ý‘ô¾#ù Q¢r¶Ž~ŠÔî¤~•$BþD» ÉŸ£ÝËÙV~“kË5®”k?É5%õc$ÊÕ®I|IúÉÕ‘ô¤­â^RÜB’1n$PyåW˯ŽÊ%LIã;R9¹ÎÿpnŽúL’?Çþ“(7R?Àu~‰à9ë[[ˆóuùsôƒœãsB¿Êu<ËYž\ãÒü)®q8Ç|ˆõU¦ËûWJLÏU?IvA°;šŸ†ZA¹[Q¡-— ý$x¾­èe+úݦïÑ}aJJO¢¤r(Q~<ºù+%Æ•„úÓ“ÊO’'©¤ö"å_/ЛV”$_®íȵ}Ir åS†ê!LIò!¥'Õ—”žPR»·™>p•3Gý'ò%È“˜Ïk”N˜6(G{'òm@é„)×|¸ö3\ûOR?IJÿ•@¹Ú)¡}‰r äOœŸh+½%ÙrÖgŽvÍYΤö"Q_®vG*×þ£þËC²÷6ÒC®öHÌŸ«~rì'¹Ö×C[p_˜röw\í‘«œ¹Ê‡£þpÕ®ýW¾œû’|þ‘«~r.'©<„~‰s}¹úÍ6Šç9ë'’ÆOœåI²GRHŠHq×q)þ'•“cÿùOS®ïÕ9û¸¾OàhÄ~›ëx™«ßäÚ?·Uþ\)×þküÆÕ_“h[§¸–“«½s•3×q ¡]HïÓˆåá:"åÃu^‚Ô¾m5.àh_ÿx\×Vò!Q®|¹¶{[éIÅD{ç:¾#ùå¶²w’^q?IòçWò!P½~º£ þ­¨êw„©™ ­ht_˜š(‰ogµ&PsA}ZQ+ÄG˜vÈ«í„êÑŠ¢ç„)©¾V¨\ÂÔå'LÍaJ’ƒ%â#L­Ðs´zN˜š£ç„)©^$ùXÑò¦„ò˜Ñz$DIõêLËCˆvDü…)©Ý;£û”(7B»w$PRþ$ù“¨)’Ÿ0%åO²S®åì@ $9ì®-?!Jj/R»“ìÅ”@IöB¨/q¼BìO”¤Ÿ]aʵIò$QR>éò Q E妤~€Ôoä`ž¦fè¾0%¥·ÉÜÿ]Jìç9Ê™d_¤þ¤'ºJQoIýÉŽH~Ф'$ý'ê ¡ß ê!!¢^¡ç„)Ñøí ORy:¢ç„igš¿µF÷…©mB”¤‡¤zûºŸ¦„~€èOi}¢Är(©ÝI”ä§ÌÑ}aÚ ÕC˜’ì—T~’½XÒz'DIzeH $½%Ù©<t; Q’üIùã‚ì…Ø.Dý!È«>åLh/®~T_Ò>\¢_æhï;"Ê™ãúB®vDì'IúCÊŸ«?"éI?¹æßVq©½ˆß½!PbüÀQå缎„«ÿâj¿$ù?@ü„)©Ÿ'•‡ ?\ý×x‰Ø¾$ĵœ\ËCê—Håi«øãwŸˆã’½sÔbþ¤ò“ú1’Hë,IzKØWNÎ='¼¿‹dG¤þŸ4~!ù5Òþ}’|8ÆÕD=äÚŸsÿ¹R®ñWÿȵ<$¾\×ÿq×Ëϱ?$–‡ëx‡kzŽë;9ï«ä:ÏÀq¼@”'©ß&Õ—ã¼×xŒÜŸp+?Qþ¤þ­úI®”븀X_®ñ3ÉŽHú@Ò;Žýçq4É?¶Uÿɵ]HöEÒ’þ“ÊC\?Ýò…jEú Õ²<ÁªE‡Å¢#ñ´a#™¯ZI‹I;IÄ‹•†NÄbèTbcâÒ™«#†Ð©‡ŽŽ‹aR'&Ð×%GŽI6})Û¤¥ -ü³ùÿ𛛬_ЖI‹é?:ægÙ~R1weøåŸ<:ÖŽÏJ­…9ïç4<ï˜GL´ ¹‡¾Ðø :謀öÉÐ(¤ýÑõè<=ߥ‹F÷céû†(DcÐõhÄ'£ç‡¢ëѹ/}ßC@Ðy?:Ä/¥EåéK§C×Cй¢qè¹!¨ÞƒQþ}Dt¢ýÑsÞ(?ot‰Îé×$Q(ßh”ït†h º‚ò톞ˆÎcÑý¡(ߘ J¡tCŠò E4¥Eùx¡tñ(¿(t?ÝB´/zž‡ÒÓ|Ðõ¾(]:·G÷CùÇ¢çèr£çú£óH”>åC/ãê‡Òџߊ@4€Î¥Bçqè~Dc_St>Q[DÃÑsaôóˆß Ú¥‹F4ñ€ÒõAç]Ñýþèz/tŽ& =bPþt9й?¢(;ºï®Ç"Ú]ŠèT¯è<Q_”~¢êtùQz@çžè~wú>âï†Îí‡êçƒÎõBχ ¨1º†òë…®FùDG£û(¢®ˆÆnPt‡ž‚òë‹Î#Q~áˆB黣û½ŠøÅ¢tCQ>ôç¬ÂiŠ®¡sODé×­q¨\q(¿žèúÄÇû óžˆ ëCÐsQ9l×™0U Q=DÑó¡ˆÆ (½,ËQ?tŸ~ þôã´Üh; õiqŠ€æöG炆÷Xê! ´þÓv:ÕcÖU¥õ‹öÙûÑýĿֶ§œYš…ò_²E@§ÐMª}Ž®£è¹tTî5¨´Üèþ›î?hÿ1ÙP@J£ú"¾©èú⯨ÞÏt«@á=¶ :<ÆnÖ£çèþ6 Ée)*?íG†£únX/ ´ýÐö™†øŽ) ÓÐó+µP=Qº,ôÜŒ"]‚ê»>•oµ€òPºÄhÍ@t±ÀÒ÷å¨üË/¢çP:ÚßÐLb{d#yOEv·ô0*‡ a<¶Då@ÏÓþb*g2*Oš #ñ˜Œèx,Dí² µS¢[ß (Ý/ÑqU ²§Ô©è<Ég2ÒËHÿæ ü×!ýÉEü·!= åÆÈÁÕµç|tY€®Bõ¦í—.mOt|BÇ£t<15Õñ§ýíiþt\™€ÎG¼DõB×—žÐÍa͇öÇt¹¦ :U '¦Ÿ¢ý-GòZŽÊIë]º¦ý)¿Ì@ò¦û ÔOóP?.<½ñ«?Rú_“òùUz®×Iéþn=Uï¿›îwåù»×…ÿ~÷þßÕƒßåÛVòøÕù¿Û¿[îß½Î5ݯþþÝzp­_[Ë‹k;ÿÝrýîó$~\Ÿo«Ïp¨Ÿ §ãÓVËX¥ý‰µÿuøu¢´ÿq¢]…h7ŽÔMˆº QŠ@é8ÛãW4û¯”¯µ¢ÙÿšÒqáïRo®4û÷(ÿ6Íþ{”Ïý6ÍæF鸹Íþ{”7üÛ4»m(¶9Íþg)s¥ÙÒó$›fÿ¥Ç5¿¤Ù¥ôü‚0¥çñèy!ßóèx¼Îüóù…;Á—BY÷”̲ͅœß½ê™%~Ì}AXÏ~þøà:ûyµÊéÇú@sËc¾{ÉŸ¿jIòžÿL¦Î\ÿþóô¨o)…¬;Ô >_þdÅ?/§2é¿ ÒµÊ¿æØÏ‚´úŒsEKê”-ðº…}åthD•ªh~?Å ¯þdì!Šž¯¦_/ WQ} Qúê–Rš×Ã;tŸ–Ç7t^‰^ÛT£r” êC7\Dõoü«ü˜ò}F¯uèñOP>x"ÈŸ]N…òù(s½ɱ•‡Îï;]NA~̶|qÔ´\*‘¼ŠPúJ!}æ÷Éã)z®Õãýz 7äÀ¶‹ _æ3DH>ôk/úõݾôkÈ”ßKtÞ€ô·ÕûÝ_ë×Pý¿ Éïâ_åÇè}Žò£õ‚Õë¿ê's½µSÕ_ëÝ*]ÁÏê·gäõéÛ;Ô´Ü>"}{ƒîÓúW‹äö Õ» å—ìn¿·Èþh=§ùѯy(}éÿÇÞw€IMuïgÚÎìl§ƒ€‹HéEÔúÒ‹]QY)ŠÒ¤(ê÷éØ+нw>+Š EY{Ê +6lØ+üŸ›œ÷&s'™$3³»øû›çy÷l’[Î=÷´{“™áúïóù\Ží7QÀòF{Ê×V­g=ùšÛÁ<®Wäö ·qÂ?`>Ÿf>¿5ìBÞÿ™Ç Šv6s;?s¿?¡_Ø'ßÿýÚ&£½´ùør†]²þÁžYΉ?æ…Ÿ€þÂn¿Ýþü¸ä‹í¦êS¾¾!U ׄ†ÇñÊøPŸë)sÅsO¯s-Opƒ z[žÇІ@¡î=ûË{›ÿöë.ÿ펫Á¾½äÅÝp±¿Y©[?ùo_Y©WO³+³~ŸnfѾ&–²}ÌÍ«}q5ØÇ¼ßÛl {/Ù­Y²·9*y»ŸÉj¿æ¿&W}zÊ¢²R9è>ýd?–1›ÃëmНºö5ÇÑÍä©{_Ù¼dYÎI°§Ù|7y»»e,³´‰îƒýÌûr"‚}dó=,¢3Kö±Ìˆ)æ^²Ñž¦˜zYÆ,eÒ×¢=äE9NËÔZØëaNCóßæHP? ›ìcé\NM?SÞ²®©Â=̉ékʫɰe>¥ìƒý-³`öÙM ´»d©Z X&[òÖÛì²—9 ‹àMš"è)×Ý”{©”ý1µÛ¬Ý×2W¦MZ,»´Ìz7Óä¥4{Hõë/{ìmQ~óß^r¼=-ýÈêÛè&g¨‡©ñqÈYÜEo³Í^7õLvÔË¢Üà.`V±tij¶n/³G©îASX½MiÈÊÒ‚Ò*ƒæäY&BŠ% Yþ6hqÂë4YìgrcþgQesÂLMµh“éû™fØÍÔ!‹HMAÊÎû˜sgÑe«/F§ÓkÊ©éoz y­§©ÈfƒOi±eâ»›%ÿ“Ê”Q$ØÛUdI9Äz È+¦’›ShêxӷȪfLÓ*jPº· ÐEoï@¥½9¬f~Mw~J<àk¼[Ì5S§ŸxdzpõY>h%S¼#±­BÕ¯^qÛ#T÷ »)ï¸8í!ºí öQh_…ª{†ýªî:í zÞL¦R§½¾´=¼d*uÛ«sÛsS÷ÆÔ=.·½'§=ŸšB“©Ôi/Äóž_Çž†¤ÉTŠwC$åëØÁ³JPÞ1iB¡(W›JñlWÒ„B“ ­M¥xÇKÒ„B“ åzxWJR.wm$­M¥xgIÒ„B¹ÞåÅ»†’ry¼“#im*Å;Œ’&šT(דï4&šThm*Å»Œ’&šT(×ûy’ry¼{ÈïöiìW5—öìx„KýŽÊy‡rÝ”óAå8\ïép½»rÞÚ…?µ>ë“<ïgß¼?!i½«CyuÜŠ<á÷Óê Søê¡Ü/N=WŸ¹Á¯ÊóÑ©í¹iíõ²/§êOš<«ê q¸¾r>>i_NáþEžÛ7oÞßY9W*¨z#¯VÎÕû*ÿåÊùnÊyÌ…/¥¾¼®è‡¼~Xeê¹"?Õ>‘O¥µ£Ì›£½ªö6FéO/¿ã&Ïù÷´vçO=Wü‚*·4}í¬œ«úÂ~8­_Î_2º_•ùòQä‡È÷ ?DûÈ;N)çr\q y)òäyx'y.òOÄIäaݬSùN!òÂùÝRùC~:T©ß]©‡þ–´NåãÀ¸N]Ïüpû‡wdÞz%׿ˆËÍlP5ïPß >óÒÔsŒyø¸”ß™C}éÙ©ü"oDEžŽü ?Äö&×#È‘§^ô§A1?.M‘ƒ¬‡y¦\?—ßYÃzã®äsÌÏy¯ò<‡å€¼ý®‡øþ™œÂ\!ÏDž¸ØxwX¾; ~r}äkg_šÚç•r<È+±Á»‘\!q"¿‹y@»òfŒGéGò2¿Ë‰u º®Á;žX7žxi*_ÈçPŸõHŽúw!ö§‰$Û+ä‡<öy߇?8–ï³ä6Õ8O{çR¾ë=Ù >Óyà3I¸¯®Á?ÖãfüIÙðû.‘ßûNå³-—k»Ùž«G¾ùóÛO¶çnýäk^Úw«—í»ynåœÚÏ7ß~ùðzdû.e¾úÏ–lùñÚ®×vüYاí>@¾õ"ßv‘ëõúö³ùæ/Ûzu?Üêç«^}õ›kýºž—º–_]Ûk}õ_ßrµº_­íÞé©S¿Z7?QzçùË¿tm¢e·–5û)QÛö‡w¼òÆJIôy«ÿÑM:%ZÕ<µcïæ7' ~tÑ ¯'@k#=—¯¹ùËÊÕ_¾Ôú‹i µMÿ:[[1+ ºú‚k~ŸzÁz 4Ñä©ýÊ-¾4ÑjÔ>ÿy¬CRÒÅW­HÜÜOÒæƒ;¾Ï¶ÏTýºÃ/>ÞãZÐÚ·vßå…Iû$Úݽxç×HÚ±jvßH“·mÚ.ûu×Nå µ®}ûöE“´É»O×~u|-hbÛû¾¹i»ÓZHZ:ýŽí_ÜåZÐÚr΂»WIÚýÈMïjϯ|ÅÌÖ«ÎlœhÑüµ³ŽiW-ihуsž{(Y»çµ¯® šØêÌÏí\™Ht;tö×ô=´j#Im‡Îhm›gïØgÙU§KÚh›»nûþÅK«¾¾òÕ__™Ml¯íðÎäO?­mÕö¿[½;°h"ÜñìÓ†.‘´êý7^½ºø×$hmä‚àïçÍIHÚ©íøÄ±û®m4{Êú…«××6½³¸y»6¿Õvº¾Éím—ýVõñgßSù½ºz£`£w-h¢²Qÿ­˜Z[Ûqù¦“Gn~FRíó» n^2´ê³c^{`Ï µ[.¸ïЦ±ÕŸ¼~øã‘ZIŸ»~àéÇ/ѯL-™Pù hí6'vï‡'~"i«ïWñÁ¡UK^²Wïk5ÐÚV‹z¾wsáË Uß½wl×9Í ‰FÔü¹ï,©m7ò•/¯™ÿZmW}º“ U_nuç´>Ö@ñ}(*üs½¤MÖý¯Ý¢Ú]æýÐyLË2I oº«ãAËAWŸ:kù¼ºi ‰&ÁVÕg>µtõGÏ^ÿçD ´¶÷Ð/6Œ}4Q¹ìªS§ ZÛCˆç™á ‰ò•±oïór¢|¯/Š>l2híV‹Úç¸dÕ†“Û¶ŸÖJÒMê/g¬\ºú·ï\}òÒZÐDãó~~ctÑç µ=ÖŒë¹wÓÏMwiùÍkáÏAk›V–“¤û¾«¾žóëÅï½.éê?~¸ãƒµ ÐDËùÂ@46Y `¼¤åí–ž~Ûá ÐÕ먷Ï6“=µ/š²âòÉ «_úø—çNÐ@k˯¿¢jàþ7KÚ–zëרŸ¤½÷=cÃa=,i¿ƒ·5%íyàóW<¹ö8Ъ¿ôrµµÛ­éÈk;%@«~rçÆÉ‘kkÛÎi>fÓÞskÛxħáè=ð«U«ÏLΫ^uwÏ“¾¯®zùï­ïßêÄ3«> ãxdé$I?ž¿uèöÍ?VmzÑyLÕ翵!‰½/é—ÜÙï‡ÃU_énfXÕW—Š—˜T—g?I¿Ñíú°ªo©ÕÞyȤ­t6©àfæ_UßW>Qõô×O›´o÷ïvþà۪=‘ôÝÿ\\õóHݱ˜TW¿ã$ýuê!Ñ–Ä$ýíˆ³Æ Kº©ô¬ÊÃÎ(0©o›tÂkÝ?èøªMº^ú2ïûÀ‰œw¥Iy7é³'ü·úþëEºŸŒ:ÓWâ{»ÇŸ’Ÿ|ÔÛ7ì–( Ôž÷Ìï_¥SC_$-›"É“^ÑhÚïŒJ”­ÝöR“þô,e%-?õ|ÒLºbÖI—œ7(Q®»å+åϾÓ+~Òá’VìóÆ¢ÏFí`Ò:ßøÒË{›ôƒC¯ OT|öÊÃO4úZÒFÆ “ÞVqÊ]ë^0é[7\òH[I÷«j’<¤ǞÞa]ËÇLú\¬ìðãLJQ⥿z˜”&wv¼K¢I\š“LÊzÒ„¢úÔ¯/7é-ãüüýLúP{ѳ¤M·yÿô‰÷+©á§'%š9©œ\„I÷Ô“úó¸•_ôO4]6÷ÖpÍL“RðkÿÈ+&ýYoXÒfEKÎùcÈ{&-=phá Û˜´Ýf‘1%šØ­Wß×HÚœõUR!þ‹Múùö È4$m!¤5â“^°Óë?ìÙÙ¤/WRŠVkþS–hyQ›Á­{ÿfÒ[©åÇWípÇ)§¬^6`ú ãgí­y| ¡l÷Eþ=ì¿û§¹î£ùÝÇÌõÈw¿õÅw¾Ž\÷#·”qniãðÞnÊg[ÄS4ýYùÙ–Èü5óðƒáù æÌÅ ë7ÆÌ«™=îÌ»ï. =­fÿ¿pÞ!Óæ­|5¡ðëâ[ñ¢­ø6¦A|KŸøIñy›‚xàµð8C3¾Zñ0BA<?œÛ›©ñ¢Œ0‹ÇVA˜M_%ÞÝCOïÄ€šÄç›«Í â3<âa·QKB+ÂV„…ñ”³ áHB[ÂÖ„£•ñe3í4FM<ý?x A<1<– ~ÜQ|Žx¢,¾¹Iü¸ýqñ„òxÂvñ¯øÂ$A¼Á+žêŠ{ß,#žhŠð:‰Ðp2A¼õp A¼ ÞŒ8• ž`Š·tO#ˆ· Ä—-‰·rÅí3âÉï™ñ„ù,‚x¢*ÞºO¦ÅËâI§øñä÷Íx´-Þn8— žþ.!ˆ§¢çÄÓûó âmŒ â­ âíÖ‹â¡úÅñ¶ƒx#á‚x#U<=ºÅ[¦—ÄS~ñeUâ-Ò+ã â ý•ñçUñæ„xkD|i–xëä‚xûñZ‚È®#ˆ‡é×Ä[†â[o ˆ'»âeñ¥Iÿ#ˆoƒ»‘ ¾•ò&‚øÖ?ñòÉdÂ2‚øñM‰B7ÅÛ+Â"¦ùymÜöõñДC¦òË㢡¬Ïóm èa0 @HTK51Í0H1÷Yê€20 ‘fP†0a8Âh„Á´ÔR Æ"ÐV3 ¨Ô ƒ±lk0a(ÂH„lgŒd†0®ši Â8º[ÐC3 ¤—‚Þ6FÓW3¦¿a8V£ØÍ‚*FB3gCŽ0a0ÃÂXª0‰0aã0 a0a {1öfC€# ”_(¾PúÉš©ìV@éYñuÇ,œ²pÄÂù Ç+®p¶ÂÉÎfÇ*œªp¨Â™ G*œ¨pœÂa¥@8Já e§(œÛÉ áØNe3 ÁfÜìÄa ¢ Â.Ȧp†ë@‰ -õ …[P¨´g½W`§±)u­;´¥B~[0óîÔžµN‘V¿·Oq>Q§ˆë ^b–ëqK]œ;Í—•§ˆ¥¼õÿ¥í¸ÒNÄ2‡øhÜ‹1ðߥ#báÁŠ ‚J‹NŠÔA úÕ¸ºfÞ‘5ó¦VŽ™6uÆ”š™•»O9”oUì>eÞŒ¹sçWŽ˜=ÁŒ  —‹ž3wÚìC)¹›ÍWBcGTó¿MFwصrð¡3fN7mv§ù•Õsæ#]l2f΂÷Jöœ=ãÈ®•s¦Wî¹ ,4–÷®™èŒÙ‡,˜ƒ.É[cfÌž=mþœ5|§LÞÙoá¼r8M÷˜¶¨f¾Cï…gÍ_0mÞÔšY(=fA×ÊÝgÌ®™!%3xÁ<‡È_0½fÊ‚9ÂÍQ¤‰ú|â¡•š8òOdú,Ó·èÛ.ô}ôú©Oú•GºÑ~­ÐŸ™þaOW}ÒÓ0ÓX–4î@˘–;Ц9ÒfL›{¤ü殺>ië<Ó6L·v •>i{º­GÚA¡\h/¦½™ðH«šPè`…s Ã™Žd:ÆŽõIÇ)tÓ=Êï,¯Ú_¡ü½´«v¡S˜NUèt:ƒéaLg*t–B(ô(¦‹è± =^¡'¸Ðs¤§0=Õ–:D ðš¼ê´M.°+ôø¿[jjœ©œ?N|eӞʇ—T>Óxìúò"çL¼{™3'Ü®¹Íq¦ñ:Õ³›/²³›‡Lÿ{™g¯ú­ÔIIÅ=}›rîÌš)ӞçóF »ã!žRÀlL0Û¡ÕœT¦>û÷cf~L0“Jzåŋܼ˜Î–¦ù±W9[ÎmM0<«f¦üR­éÓŒ3wû“;ªÖuWSEc.gÝé;ŠÍbÇd+†ueUzqY*mXw\QN¬ò±Kº5÷ÙŠÛÆ•(âÿ­;-uå߈å Úl¡¥î,y|í¹ë˜c, ìà ~[3*øš•w”í6Q îXt´ÌI3‹LJ-}·+¦¥î†DX.­xl¸'Ê6çºÍ‚ç¦ ´e¹þ·±Œ!h‘1v„,°cÓ˜vëŽUn ;ñرQwh)sÙ’çºÇ‚©Jʶâ6+ø~KvºT4ã²à;éh¯Ì"çËâyùó3*ÅGò²¥ø9ü ÓÔñötÊvƯ«MC=¦øù+üL (~îJýê ü|èÁLÕ¯êp£“_ÎLñó_5ßÙSµ=ü¼:>ü\~vdJ%·ÃÛøÙ4P´·Ï|ãë$ñÑJùIûþåWm(‰ÄϘ¡ßi}Œ¯G<¸£=ÅW’L]nÐøþ´aöô }yžWüãç¸@ñ3X¯¤Ìç´%Ìo¥==dn*­A½k3ÓIé”cøè*(~ndZ·ÌmÅ|îÇüMæ¯ÅÏ@©t–#èô–†<%­Lå v€Ÿ/Û“Q?›ŠŸáš¢Ðý¸ýyÞ(6~=?W‡ö§¯5îOomÜÝ›ÇyÀFã_‰s  çAGíÃoìk|_ª¤àt—;€åUÃú:y£Ÿiýv§²ÝNjÌüóøk¸ÿýù×÷bþœèT/~žtwnó >'Ï`}çñËú\nú.,Ý™/¾:ýc ߟr1ë÷‹þ¹–˱žíÏzºÏF='ŠŸ;„þN¯M¥òç¹ü.èô½Œ~&c¼Lù«4ÄùÕP ^›Èó<ÛÉþE¥Ã’Æý1ëŒsÐî<ÿ{°^ƒöäò]ø>h…b?Üû÷JÌ—ÏzŸXaO;q; #˜‰ì@GßãœèÉt8—ÃGÙÕ¯lëÀö„¯’Û‘Ç©~E\g¾Ú•MU¥ø*|åH·¯Ä~E¥Ûs?øŠ|UHßÑF»ÃY>*ÀòËtÛ?è ngÏ?èpögøJõç(@Õ¯ ÈãŸÀ|;QUÎNtüm»p=БLÓ¾:M¡ø*´=Þ•åêôó ]Ù‚ʯfcº;_ÇWBÈŸ‹äyÅËzÝ›ùíbÈ!Ë7Büì[d³âeßÈi_$ß{7^÷w¼ì¹ùÝ×ôÒžŸ>ܶ‘3m/çKŽ^y÷ƒß=7·ñ:•SêØî¹ÅöÙqæŒÙ‡OÃË¿…5 Ì™?gVÍLO[ozš§ý{ü{ü{ä|øýdHCbKûDÆ¿Çÿ¹#c|q³—|Ñúîï_jOqÔu½lûÉõ‡¹~"0×vüÆ|§¾ìÌéø×ÔÍxÕ£¡Ç“Î_N,ÉfùŸiIëeËÂïö‚¨(°¶§Þó³­b÷…Pžøô€GäêxBÚµ~0$`ÓVØRW-ë6Öòvÿ[ùµkGí+S^>ÄäóƒO)Û¢¬¾+°çî;›Ÿ ‰ÑÙœ‡NÃG€Ë‡.œ7gî´Õ]óu.ëãUq¼C…òÿâõ ñÊ„ødé÷„Mto¸ÊãÂ_æi`[ËÿÖïÏoÌ×Z®­¥so€¸ ü¾ƒø@ñ˜`ik,aŸ4²\o¡Ùç§_ ñ?/2½3C}ñ!;¯ þ3”‹Yþ/°áe×íf¹.6Åþæsñªºø}Åß©œå;õ…|ä!^ö³Éáøžýþ\F1h)§|o¾~l´üÿ’Í}ëÑÜå>~-vŠr½Ô¥ž8„þ‹O#¿n¹v“‡z8þ籜ÐñŠÏ2åúÍÊù"Ëÿ‡2µ~à>Ëú¿õ÷xÕÉΗ»mﮩ}¨×íúqêÃÏV©ï™Æåv/O™x·ŽS•‡SÛªlìÆ§öç&3/2õ{ßN™êeš{·z~Ë:É1Ïø6Ã1& ¾y CŒ ÜO8Ž >¼Ì1&pú&ž‹ut.^íü“è(æGœ‹\i¥QG1¯²|*y¼%,7ZÂÓZjŒyœð”1n=ƈŸ*y’ð£fĘçù\èÎÇ õ‹-4û´ê5·ÿÕkvmÛõ¥Þ·ëCåËéšÚS}?÷2ñ”‰÷L2wjÛ‰7»qy•™™ú½Ÿi™êfSÏoÙLúìtnù?囘R¿‰©`Ê´Ù ä*ª`^Íì©s°vŠ,˜7­F~õÒüi‹ð¯ù-L‡N›qÈ¡(Sp”õ,:c6­¹fà4&fsfLů˜ÇæÒÂìè™5SPüÐ9ó»N‘ý,Hùލi³šlY¾0Jþ2zÊ7?“ôçï¿ÿîíõ›uR¿M§fÖLþ6¼ÇÞ\þäŽú“¨xo ïEá'FåûKL‡% ŠŸàœZiPü4h¥Bñ3Ý”~ð¾Þ·KpûC“)ý¯o^¯;•ËÿÆzˆƒŸçä I–ؾ·S3cö‚|s¡dìœÙ³,ÝwR¬/Ä% fÌ’ý5~ix‘E‘ËyÜú ù[Äò¦ñÄò÷˜å/`§ÿÔ±üców©Í_Z6SÞü­móçåOÒm~]Ùòåò‡¸­¿É.û0íÝü¹dñ[Ú9~ïUÁ܃§ÌŸv«)Kô2ŸŠà´3g» aÕéôµ8Nˆ¹@ý—Ö¯àò ëWuùÓ×zÙ}µW&”gõkÂçõÃdM}¢™ø0X‹,ÐÒ'Z9_ƒ´ñˆ¶šýW¦yAeŽhçÛd‰ö ¶õšùÕp^ÐÉsÀvZú×Òå‚ísÀ>°£tuÁNÐͺkö_Ñ— ==BýŠ?;¨_÷§¢¯ú9@ýŠÀLØ9G È€]²À®Yb7¨² န{Ä,1TÁ°,0ܪ0"ŒT0ÊøúI|¥ã|`¼‚ 1Ñ#vÏ{X°§¨_µiž öó€ý-À×tf‹-À×|fƒg‰)Y`ª‚i>1Ý'ðõ¦^0Ã#³Áá>0Ó³q¦ÎrÀÙ>±Ø‚s<à\ –(8OÁùp .tÀE\ì—øÀ¥\æâ{㯰àJWùÀÕp‚k]p\¯ßY,ÕŒ'w*ntÁM¸Ù#–Ù@|þ­q[ÜîwxıÜwîvÀ îñˆ{ ÷Ùà~<`ƒ•6xPÁC V¹àa<¢`µf< ö‚Gñ•GÜï£ü!RaÂÞö(]â§™(œïŽP# îJE鮩ÈäÓJú¤#r=‚¿›(ùÜÖzG8i¢d7ˆgô%«L„öLEÉrï(¼ÕêGJŽpÁ™Þ!žÆ¢&Äþj¸W:JvËŒ‚ íQÒÏ«ÒQÒŽtåVo(ºÜÅ›=à{xÏ"¯Ù£øEÂ]¸Å‚k=â2Â¥éˆ=š¡ÓQ4(S|/ñQ™QÜÛJ¾ žwö†P›TwMGùÙÂ?2£èmxÅwZ°À.ðˆÿ2Fº#ÖÅíÒQ4Ó-œ:ÝDQ5cŒGtwA{Ä(ø=á~©(¬tF¨Y*Šš{DÜáµéˆ_“Á=S!¾í ÞÇ'yÀÞî}–ŠøèT&™ˆï“ŠðâTÄwQ°Éx±Øfñ Ñ+Mï3QøDf„†¦"6>3B×P÷ ¯0q@è¶T^ì‚ -hΠ<¾ð4 NÌŒà‡©(œæ€Aöˆ}e¢ppfÄ6¦£€ò€ÂìÝÙD¤y: RÞ%±o!Þ©ÂÞOô° xÛDìœo‚#LDnJGô“TĦ0ú§#º&±>`ã{ñ™±‚3RÛIA§ kc  <áí „¶NEô;g„NIGÁ&Ñ'+œ:Ü@ô)Á¦&¢wY Þ'ä÷—R®_˜Ž‚]S}­lÐ6ÇÛ :‘OÜ•ù¹æ ÃËô§NœÊç«ß|ÂïÑPýÚõËÑPòÞRäßPs˜ÍXòÕ~]ðV²Ê·ÜóÉK®cÞRtìÿ"êRvõyÔ‡L¶ÄyØt ×£>ùßÒí©¾ø¬¯1Õ•ž×çQß}û™ÇÚ5÷õF>yÊG;ÙŒaKÔy·#[ž¼È'Jü¹Î]CÏ{¾¿zR}e;'ÿ›ÍÖ¿9µá—Çlì2ŸG¾äžë|ýÓíT=Ú÷ÔWœÍÇ‘‹žx-W×óà5GËÕG;õ=w i§¹Œ+ßz/_Y_‡[¿^ï¹ñ¬êz.ÏD¬õsõ_NíåÃV³µïúð™ŽLò÷ã×üÊÏëœåC6u1ÏùàK_>ç?›üÑ-WðÓ¯S»ê‘­ßÜõ;´óú<¶>2õ)»†ž;»¶¶tú§ 1Ç~Û¨+³¥ÙN¾-­¿-e΢Ï-]oóq4Äxþ–¹_>ó9Ö†Ö÷|ÌC¾Ç—kýLí6ô¼5ÄQ—rÊ…‡|Ž!_íäkîê{îói'õm_ÙŽÏÏ‘¯ñåÒW] =ç~ÆY²ÏǑ혲m»¾äï÷ØæEí+ê‚÷|ÍGCÌ{¦úù”U}´•©ÝläåG&¹ö_—s]×6šÏ¶RNù:²™—|韾üõ9~ä‘ ß~úËel^Ë{½žOPWóøO²Uùà+W̧ÞesÔeÙØ@>tÆo½L×½¶oWÆïQWãÃQ*ž5'ÆÔü»è¯³%Ýë¦Ö®Î™N8yé½§­ÙÆ‘Ž:ðù+ž\{œo:¸ä¿vèp›g:~É9 y÷ÏtÔ3o{Û“¾éÈÓ†.é=䤬éЯgm÷¡o:á†×wûléG‰‘ïLþô“ñ—×sÞ•[Ÿ™ØFÒaoõ?ºIÇÅuFG·>xí„cŽõMÇo·ß®»sLÞé~OÞxýO¿Ô9Ùü˜×ØóÄtúkù¨»{žœ¨~ºö«ãWÝ_ïtÌïñ÷Úu»#_ܼ]›_ά3:ºý´V §wN££þx~ö—ó.•tØ ‹È">¯{zÆAoÕõ­3:2ÔiæË]J=ÓQýªš$H ;Z8†.itÜw;ð펫òF‡Üµû€âçÎM§O­xzç}¨3:òï£g½7v÷£#–výyÙÜ[óFGFï½ç¦“v«7Z=®çÞM?këHG÷:¯éâCÎÉ;+¼Ä¨S$4>qìþ#å’s<éûêpÖtøžw’§i[ï4ñJßîßí¼Þ™¾µqÃ%´m0Z=çˆðCíûÕ;ôÜ9×ÓêϼÓ!ŸžyÙ[Ó÷ș۰[tÿ¿6dM‡ß÷èóÏ\×Ë7­>oÎJIr¦c/8ãì{*¿w¤#m±›+׿N‹úˆŒ.otÄÐ>ÇxuÎtüÛi&&¥ÑÁ—Mx­ûã¹Òñsßü౉û{¦‰—Z1mÁžéÈg;=uêWëòN‡Uòß…„Œéqä ÁkßñMÇtû¸Ùåg½Vçt¨¾ Ô»ßa;›Ü#otøøY{­êv@·`‡N?<Æ3Òò›×Ÿýí™ïÿ÷#vëUçtÄÒ¦ìÛ%o41öø/η¸Þhõ­¶è:hG:ºg—Oí7³ÎiõÌ—|¹iû¼Ó„¾ ©ðLW Üi×_êŽøùÑE_s¤C›¸ßã*\éˆß–[UŸ•¨þbu—¯\%éðßþzàÝ&»;ÓAew/w:þíݽúë„#w݃¿Lz§¼Ññkß*¾jÅÀ¼ÑÑÚóžùý«z§£¾úpùÿf¤Ñáb×qÏwóNGÖ4Z}W²YbÐ 7O9õ‘Í’¹{òAwü‘ŽÄ>‘M¬úø¹XÙá9Óa/ÜWzàÐxÝ}G±QLüÙ«÷Õý×Ö;ùó¸•_ô»,ï.6&ºÖ™_s¢6u¹p—ßû9Ò]ô03¼Þhõ€¯Ïß:\çtĪРíÏûÅ‘;úòîì÷cÑÑœOÖ­» =ßO =±Û}úߘ7:0ÒdÝïñ÷òF'T^2õèÇ®tÜ ÇŽï³íš¼Ó;>´ì¹š§²¦ÃÈÙÌW7:xÁ)ïßÛµ©o:„×½ÙÒQ^ycëƒ×掸j}ëÂÖK\éȯ_zõô;òN‡êËÝã댎=£ ÝæªÙy§Õ´*Ù|Yw*–G“NöMGU>Qõô×OK:ð‚Iwí> $k:žŸ£ŒºÿÅïJÃçäû¢èáÃ&m/éÈ#> Gï½71ü’¾gÝøçm9Ó‘knþ²Å‡Ë]é°’ $©gòN‡l:yäæ§4W:îÅ‹÷&—UgtX›f‰×{¦ã´Ä“UG:dØs½¾\3Î37ñ´k†ýZž3*V/¡êŽûá»—zVÞè¨?Z¦¿Í™Ž›ÝWDÔ:£#®ù}êöL‡×èžé¨ÂÚ½2µ¤Îèp~oC¥£ÞýéÊÑӮ˙/Ùvè”}?I£#·}þçÿ\4Ò‘Ž¸ìÖ3<·³#M|sÓv§µ¸×7XûÈÒý_‰»ÒÝøy–W:vÃ÷ >¸è¨4:ð¸=ßùlQ4g:ê”}ÚM¹'ïtˆH•F«_˜´÷•GÌ º[qyb¨xÚ[O£cî³æ?¥8Òj^Wç›âýª\é°÷Š=öÖ׎tôç/üÒ YÓ¡u O><ïtÌ¡W\ú[›¹‰‘B³–§Ñ¡÷Ämb¤0Ÿ{^s¥Cn_4ye컬éØcnžÃ3¾~áê÷÷?*otà#âíWZ½Ç‹÷¸áæÄÈ=Åpç¼Ñêo÷»$¾|˜oºÛ°¹Çm§–3õÌu½¦î¼cÞèÈÆzàËòbÙ“•å¯z¦£V~ÑïÒç%&+Ýç Ït(ïÿ¦Ñêû¯orûÖy£x~6zÑg£¶¿®Æ•i7ò•/¯Y5Äû}n´ºë ±#äJ'ˆlaï_éÀOxä¨ä\G:nœ>AYÓ…ƒ]çJG ^¸qåªéÈ}–]uê´á‰‘“º<Úbÿ¶y££õ׃ŽtX—Ozbæ ®t×sZýñÛÈQ‰Ñ§l?ü–ÿ¾”5ÆïÕäJn¨ØžRì¬éJ²ö:wÏ4šhóìû,»:ïtÈYúƒˆ¼Ñ±dMó_=BÒAü¾o®tÜȶÿÝêÝYÓjñ·ë³®t,-ž÷½ï±¬éÄî3n~x×|Ó±ÿ{ì¶ŠSîN£#Çmxâ²ã–KZýá«O^z_ÞèÈ…Åâ¬ép~.敎hµÕÓ^øXÒ â1Ç«Ï;Ò!âõÓÏþçHÇ\ÔfpëÞ¿ù¦ƒâÛ­ø©éƒŽtàhñ€éç¼Ñ¡b{ý½›=Óq//¿æ…3zºÒÑã/ïѦÙÀ¬ép±ÚÐÎ‘Ž½FLÔ‹‰Áã&÷8ò…PöôàlyQ›œéè}C{Tw/Ï™Ž9úé ËÛ¾éH‰Þ·v¤ãz7¿iyËità˜M{Ϲj½#ÃûBNtDsQðjG:tïg÷¸î³W»é/NKš=>Z"é`~¿ṫ'~Òþ‘W$ù¿Õqe.^3ïtyÍv‘-rçÆÉ‘+f:ÒñÛ E›ž=ý`@m¿Fý%¦¿t”3ýHð+i‚Ÿ³y¥CÄ.Nô G:êÞý¾{t^£¬éð¿ÎÖVÌ:9™6|üÓ%ÇI:ú?u8~Š–˜ÐæÊW}|eÝ]||¥Ýÿ²¦ƒ¿~jýõWT9ÒBë§Ÿà™Vñ@ç_zÙ3Ã~H¥#D–¼í¼¬éñi©Ýj\éÄëô31²ë?ÜñÁ.žéЩú†Y8ç׋ß{íSW:¬â±áH‡³];ÑÑô AßtƒÍoº«cbÔÝ‹w^ÑøI‡© Á7Ýyxû¯n»´E±Ëüÿ]ôò]YÓÁ=þ:¨ÃÄÓ=Óáâ㻾“NìL¥ƒ;ýÓœ#"ŽtDÉ›woø~¡¤Ï3ô‹®4¡/z¹ÒѼÎ÷K‡7^óú“G?㛎_´øÂ¯_–Hüöò{ÇvãHÝžTWw/ûþy¾éðî8åäÍ3ƒgm·þ¥#¯•t ¯[hõQú‹Í’Žÿä–ñ?~þ3=¾pÄáe—ú¦Cæú ô“7]éññÐG–¦ÑÑ—m+$(éX#ú¦oûþÅ‹÷,i?Qé¨ÍU4#MÓèÈIï“Ǻ3O»<šÓô³¶´´•tØåúÒè ~ÏߎÝC,Ö;ÒüùÒ¡§î;‹u¥£{è%Ýíü7¶îvÆdW:–ó$7:¦ãA·~rËW:N¼þÔw¥+ô÷ZÊXZ;Ò¿_áB5-¤þN³ßï†ø¬ûÿ2}¦¼¡y««qyk6ßOÐÐãþÿ⟊†¶¯|ô_—|çúÝ) !¯†ò‘[*_ÿ"û9lh~züÿ͹ú»-U>ÿ"¿sßмfË6m6ô¸þÉòÛÇÓÐ|5” ¶„¶ë[þõ=÷ÙèÛ¿zú/þÅÿßø'ç]ÿ¿àß¹hx™ois°%ðð/êv~s]?ªuýúù-IßëaB‡¿éÿr¾V(Æ< ÿL7Jlƒ¾Q0½fÊ‚9óè¿Íšù[Ïâ¨ezŸÑLH¦2±C¨âZM+¤iÅ—s™åTf&uûŸ¯¢ót¾ŽzÝ ij#ZÉ÷ž¦{§Q{ÄAx$ݻӸ8†ÊH×ö§k¯QñõÓéz;ÂotFy…xxƒÛzîŸBmmE÷n!þšüÇnáûÐýT—ø Ó"ÄSlƒfblÇÑý/éþ>tÿYƒÏ+Þ“ÆxñUM+Íåïaº’êHýv¢zÔñ½€ê‘tã?’i\¥Ôvù\þ9*Om·#PábÚz‚ï}@×%ˆ×(õ;ˆdð8ñ0ŠÚ{…Ë>Le¥²PÙ›©_ê?z„1‘qj£ˆÆVÒ›ú>UÓÊXFú¼\góctíBºöµÑŸú{ˆÚèHýOíÿE‡Ñ˜©néNÔÆ•4†ï4­Ñ>\÷^ª;–ê$øœæ.0ŸÚz’ÚB×I¢¥ÔÖ ÔÖ_ÔÖ\jëmâi¥e¾5e¯4çK¨o¨ƒ© Ò(ñ;Ʀ6j¨ Ò‡ÒqÄÏc:™ä¥:RO©Î0ª³ŒÇ@ºSü Õém)/ô‹Ê…Io#Ð[q¬ {Gñÿwsm¶!ùžAíR{…ÄWÉP’Å×–:Ûóÿ÷³Œß§¶S½G¨]Žý—êm$^h~Š?"^ ,:'Ê¿Eåin#T?Ú–ÊŸKå¢ò4îâw¨üž4ÞÛ4­‚ømt“¦5©´Ì¨¿žêƉǨ>éLag’«h—{“ÊM¥r¯°‘DÉNbWQ¹ÖTŽæ¨øê‡ì¹Œú«˜aÑE©¶¤ñ‡h"ý¨¿»©’C!É)NsW<ê/à²KXçH/‚çP¿4®•+ ¶b¤ßqšû"*[|—'Ý Ð8ƒïQYS„ì$Jö{Ú"+Ò·Íc˜dP üÓÙtÿjk€aŸ%¤o¥ä#ʾ¢1ìE²â9ó¢q†çóiÔù‡ õ¦¹ŒÐFw ¶Ž¦±P¢¡,*jëVj«ƒ¦5&Õx³Ék°µI¶¡Óò‰Q’Maâ…Æ[L¶]BzWFuÊIÎdGK¸n­¡÷ú8Éî È~¢—R]òÄñ9Ô7ùŸ’½,ö0‹}'Ù=¶5•#ý-†?vDr ¾NeúúmEåŽãûÂ.¯-é>ÙVä{ê+Ê÷VÓ=ÒÍà‹tæ3r;ëŸÐ××-úMsÿ¤3Øï_ŠPÿÑ^†Ïˆ,s5õzWâ{!ë%éMQ_ß³\Žô0@þ$H󦹋мi.ãŠÈ—¿Ìe`ù _1ÊR_ÑöT–ô¹ÊѵbòÃ¥BïÉÊI&&Ó\¿f©Oü†H*[ð-ÕíÆ÷ȯ®§{£X†d³1j+~µKm•Ï)%^Ê(&TìÎuž§:¤£Á£H?¢4îØRªG:^DzTÒÂðùå©Þ'>Dì#ÿ&?^@u¢ç I¯ã$¯bÌ›ˆ)'XìQÈéCÖ‘“,sv ëd•áãôƒ|và4²£ðÔÉ=J|ÄHÏã䯋ÈW“N•Ÿå{rÇÙ/ Ù›ý_SŽÔ^ü%K¹_©mŠÃaꣀl/J²‰oqòÏE³Ø¢Ç…;hÜj#©í5T§•E/…¾Q?a’w„ü[”ä#[*$¿§9-¦q–î–l‘] ’?ïA}Ó8bé<ÙEùr²—Fë-¶Br ~Le‰—ÈãY ›§9ßÀ¹ å 1Ê'âäkŠ(V“”VÿdGågÎu…Êð×r §>YØ»@c,êAã"™–,#Ÿ[~×>žòƒ`G>’çƒÆ¦\(B²’ÝÅ(–Å)~‘>kðSö¢+õãQö¯B«©ÞÜ/¦zä3ŠÈ–O¥äWÊ‘ßÊ<ùœ|W`.çÑ#8§ìÅm_CúWJ¼•q("P¡©üÕTžÊP¾Û‘ÆLòS®_L~ª6,ŽÇÙ/oà|†ä%;Š‘ŽÄ;y¢~<Ëó)ⵡ|*º/•#¹Å›Q9¶ŸÙjüK˜æ¬ø|‹¾Qù É)|ûÚ±Fþ§öŠIßJn4õH·“x̤GQò±å|ÿ~Ãÿê9ª°#*%?{í—¶ûŽ­ÛHÆÑG-ss4Û¢ó½<7çXd"æÖácµQ”bhŒò¢øxêƒø*ÄeE̤Øë%øuaä›B¤§a²‘š³(µYHySœx."=,ÙË>aø¹P3#æQ’Qìöõ$ß’°¿dùY¼Î[`Ø\”t$F~-ž¤²ä+J§pYa_7B *‹˜½‚s¶8_£1Æ’|醴sŒ£9 Sß‘FlŒ¾iÑa'äB4ž0ù²1¤·…¤#ñÏiNa×WXê~&S»÷rÌxs²óÂ*.ó¯qIßBßõ8ÿôש :dCEä#‹ÉöK»pýZ*O~&(ä¿ÖÂ+Å€à›<ß«8VQlŒÇ,1ùöC¬ƒ"^“ME ä;ïZÚ]Ì9„Xcr®8‹®‰9}“sâ/Fyl!ŇøÛ\÷^cn‚¤‹áކŸ/ø•óf’gÍQ1µ¾òn^£"_!¿¥\&F> Nó]Œ<à6ÎC¿3l°€Æ¥~ ÉW}lá{Î)w6dªûF‘«þÁò û,9Ë_Ãë ±¶%ß)çëO±\_2ò?‘oèÇ]l_$çp'öý"·¢ÜµÆ_DsPLv\ãòË,kÅU\÷S'ñ%»‰Q)Ü…Ë,7æG?îg{˜Ä²oÅkçøþ}ÃÅ^]/ ¼4Þ™ï ¿Dñ(Tjø9/$· Å¿ðÖ†ŸÓï ž$³,<Š˜I¶#{‰Wp™›8{Žuîö!ä# )ÿ)¢õañó\öÃFõãKÖ‘“ïÔr¼¹sj¬EÄñ„!ýx”ëÝlÑ—i–u꼇p+ë<ÍuI±Ež‡ð| »\ÃkZ²¡Bìù¬æ\û<ä·”ûŸçöÉÎ È^b“-¼Êù‰X¯ÿQ±OD:'=(ZmY›‰ƒ®GO²œ_ÀõE<{•¯ÝfÌ»øüG˜t$ò5ûÊ¿ù>É:Hq(DkHµ±Ï çKìQü¤Ñ5†Ï‰SÙâR.Gö xêcøby\jؽثÓcù·(ù·ù³8Í{Ñÿ¸Üì»Ó䫪źAèÃÆ*,dÕŒ÷ÔDÞº?—yÐØË =Àû$”«DI7 )¾Ä÷£~^àr"GÜ—÷ßzñº«Ï-ùÁ"ìK=Â9ˆÈ(nE°)Ž{ŒÜBçg’±nÒ%†½èÍ}à@#‡ “}F>¢~(fÇîáØt»¥=¡wq[Èׄm&¨~­‘Ã/- 1DŸäû+9ÿ&:/ýZø6¾Fã œkØ©~¬cÛZgœêq¾Ï;­™ h÷…;YêO1ã½~Ðü–{òXfñ‹c¨-òQÊub—òžøy”má Ï¢}ÒŸ‚§,í]gÄÑð;<ß4®ùøBìÃ\Ï9Õ&*Cºù‰ÊP<.œn‰"¿ \!H:ú™ÊÏhäùRÆØw#{ P®$ ÓØ#BÈÖcä« _1ö ŠfZÚÄë¨jÎ©ÏØñ¼¿ÆkW¡ÏúñǾõ¼v{F41òýñÁ\f9ë›èÇYŠY±møþœÞcÉG(ö‹=•à×¼G-âåV±‹ùþ ök8£øc»’š:!y{8ÏÂäo#ä D޳!÷¢ž<.òAòéag„|mt7coËí`ý"Ú!›ˆ _ÛÄôv^#xÄè7ÔŸÏïey½ÇëŸå½˜ó ¬]D¼Èr©4|uù½ÅæBÒ¡¢F\î öÓ4Oa±_Ìþ3ÐÝØ÷ ;Çžö\V¬ÙwãõéW!rœ{˜¿=4ó¸ßˆ'znIóZ ü5éOá‰>©ß ÍI˜ây„|Båk1’i!ü‡ØW{1y¯ãyÎÏiýL>ÿ‘cÂ1–q¬0çDïÏB0އ-ùÍ}ð.þ_Øk ç9ƒxí=Áxn¡×O˜ë1ô£ë€Èöä=¿–„·LyêõNàµÏC|Y‰XO6¢˜nîŸ(ÐæPw¸œðUI#§ ]Åyÿ/ÆZZÎçì!AVˆ±ÂG‰½Æy ö.ïí¿g‘9åiAÞkÖõw±ŸifäÜúñ¤±6 •sîƒyºÙðuúA1+0†×Ф×Òµh'coHÊCØÞdºþ…_òqÁ3ÙwœoìYëÉ3°šÿÄПPµ±ÆÕ—9µî5.²œ?jä¡3<¤€ì(6ÂÒïR‹ H§ƒ"ÖüÊ{ìl£ÂŽ¢ˆ‹rn%âÅáÈS!»+9ˆõ†h£¯Ó˜×òÿ¤w«5fd"÷¡þ,‚<{vo2×sÂí,¼’ÿ †µ’~\Ç9Ù[ü\쿹Ñ2æé}Ë~|°˜ê\_g}ÎV°"VÝkÄÝ®·ç|ÎR¶¦ç‰¤g¡0Ÿ£>ù%ýÙü«ÐéS8þ f©ä{çð<ÛÀšô~#wÓ'˜gâ%@í†ÆkT1¯ÑÅ\æG¦Ë˜®bsÿÿ†–º÷„üCøE–»àWÏÄüFø¾xþC¶c©+|Ùwhw‹ÿÅsÈ«yî„Ïfý ”±ïýÒÈcõc¥a!Šá^ÄÜ’‘­F ¯bmºØ’Lj¼ëTÞ± Ïšñl^=ìžë{=ü”Õ’~ ûë+¾–ÿ³å%EfIõ}”uã!/G2ûòª,ü¹ŽCï7™ÊƒW^Ôry•é?øð=§Éôú¹ÈÒ©®[›ê}»¹Í‡¾y:’9vTGÒþ²;ÈÙ¾“Îm§IŸ×íúõQÖs›š_ŸÏ¾3IomäÍÇyèÏW9>ìxÌ&7Ñëùì»Î¤AÜrµ¼ãý,é “ö}4ä‘m>êu®3ňŒ¼$Ó¯¹µÕ‘ÌG#Ùõe/ÙÆw”±‹WnvÓÐGƘšôÑ¥¬[=ñ᧯|.ýùÉ|û/e”¶­¶žË‘ÏuM^Úµik‹<’þîÕÇš­!òˆŸ–‡þóû<´ý?’ö—³ž÷lëý{ØujãIÿU²Õ‹|äŒÙîc4x\sé».æØ6¶ç«¯döUóµ~HñéÉìêåíðп]Üi=פýe/¶åKú‘m¹Üwêßéú·ß¡ÙÛž“Ö§.äu/Ì¡ý†ìÿŸpÔ‰Ïwº‘ôߎþÜl/S{u±®mˆC}Öìš–:§û^Ú‘G2½žÚzøé7S[®{S>[’.÷ʪüxõQny“ŸùvzVë({…µ®/š uƒ§±9ðŸ©MÏzë“WýæVÙø×yË¢~.úãÔ·“ŽØñâglNmx™c¯6hWß‹gjïNrtâ)¥ý¤ÿþ2ñïÅVœúvä1ŸøÇ}/ë 7½qìË¡m¯‡_Û¶«ï4·)×”þÜæÌ—>$n¤¶çö~ƒ_¤´íÒž“ÏÍE÷íÆá¥Œû³+o×_62Kk?é÷L}ª|y9²™w»>œlÒ“çÈ—]yk;~/¾%_~Ãq<Éô²êÿ^ûqëÏ_^í_Ò¤æzdkó^ìÁ®é4é£-›²™žÁ¥ÍEÒ¾œÓø²9Üøu³“LcΊ¿ü%}tdÓ¦Ó5?ºâEvšÍ¹Ošæ<&7þêëÈENr¶¶­þï&Çlúuk+ÓáeÜ^êÙÝÏX7CÛn}ÚýŸíá6Ÿ¹´ïTO^O:ðL-—©¿þ*›ùrj'­]Kÿù°áLödÇ‹ÉTWíÃÍνŽÑ‹m¹Ù’ùzáÓMÜÆì&s¯~$ÿví;ñéx$ÓÛtj;›öme’Lo'‡“oð‚¼ô­¥7S^—/^2Ê?™¹¯l|¤_Þ¼ŽÑn¯Äï6m{•·“=ÚÕwk/“LíøËÈK–ã¯vÕïûáßï8´ å½ò䯧[»~úËt-Sû*u’ßCm3›ú^æËiì6ÏóaÓ¾mÉÔ~ìæÀµ ¥¬Qûò­£I—ÂZîÏšýè~ƱzàÕíÈ$7Ùùåß·}'3ói§oÙ™ìÃMßldj;®sžt.—ïÃn¬içüøiÛ«ÍÛñ°¥^íÆÉ¸Ê5éܧßçÇÇø­çÕžýžÆ“´¿gǦv½òêIn O™êúißm~¬}ÛÉ!­®RÖN¦^}T®G]¶×#i¯¶ž¯þÜŽL:êÇGø9Üž«Ô×!ûMšçÙÚ«_¿–ñ~RÛb>ƒãäKœ®©×½è¨“ïuòK^l*+ý´»˜ôÙH.}¹ô]Wöâ6Ÿ^ô9S;ùZkd3¯nñͶ|2sŸ)mØ•MúàÍRߎf¬ë¡œÞìì.ãµ mxÑ“ŒGÒÏÖ>ÕþÓê{íÛ¦};>rÑEù2 †¶´#Ùðý:Ù·›WË«×ÒÊ%SÛöÿúÎd[^êúõsÙ´­¶™ÏþìdfËÉ,w¨g7žº’[¦Xš«Ÿò˦e×—˜ä·½´9H:—s”“M¿<ä*ÿLruôÉÜøÌd—®ðÙw>§gŠùl;S›™lRSþwãÏ16€&3Tvà)—ÃMØú  å½öŸõá¡o¿q6“M8¶ïÀ‡íÅÆ¼ÊÈ>8ñeWÆ“_ðXÖ­Ï\·þ4;?/÷2õétîUçܯ²VË{iWå9×ÃIîyã%é“!›~2Æ>Ÿíû•W›ÉT7Û¾õ#™Þ¦Ú~©xÖ\õqÔí¨ZŸZÞ‘®w¨çF]ÚõÌÏz»Ò[Þá—O§ñæ‹7¹×—\Õytš_•æ­ß<Ë7_|üSô:×ÃMÿêÊÜŽ†–¿«=Ô‘þzº>œúñ_r¤ù>òΧOÿX×ã’çõÑ+?^Çå5ŽäûpâÏÑŽøÌ7?¹^¯¯ÃM=çJ¹º:ÜÚoè|"ÛÃ1wËÿý®'Üh–róÛŽ«Þå;ßv¡jyßíýÃô®¾â®ìo “KÖñÞãz­®ììŸrÔ[>ç<Æ«¼ç…>çו¥Ÿÿ+‡ç<´Žó6¯GÞö?Ö;Ð|û™|ÑlóDå<_G®y‘ï<½Žâ»ÛúÅo;^Û÷z?­|ÛÝ–îÿêkýã7oͶŸº>rÕ§¼­‡¶ýœ-åpóƒnõrµ?¿û¡ùšß|û†Z/ÖW>.ÏsÜÿËÙ/¬w u$ï|ͧWýÈ·yŽÿýºòé@Ýú÷ª·~×IuÇy÷¿GýÙÎÃ?uÞòµîóÛO¾Ëû=¶´vóÚ7öše»>ÜRôÀk¿ÿÿí×Ö{+÷ïáïø§è‡Û‘µÿZŸ™z­×PG®|Ô×¾N®õ·T½¬/>|¯+ê8þæºRWÇ–¢^º~5_G¶ûRyë×ãsÇ÷¢êéýW¯ëŽúz?×ï¼å½ßºö?i}õïZïÿøû8²^77Ôºq}nõò½ÞÌöp{®žV~½=õݯÏúù.W߇ß}¦-muî÷A³}Nâ±|]ǵ\ùÎõ='·8š­n©y¦ÓñOáÓéø§Ä}¿q¬¡Ö¿i÷sô39ǽ<½×”ëQ×yw}u×Û´£ž9±£q’è=z /T}™8.¯ú‚)ôlótýH„×èåÅLË•óFF¹ª_˜¢½ß¸?r’nâûôÔD¡qž(aZÎü5åûE5ùáû͸ÿ&L{r_(ãÙÈýn`þ>æöpŽñþÈ×U9à:(ƹ–Ï!g´Š~J¢)ãnÍã\·çóÎL»±väòÛ)|€/–w"ÀåxÒ·[¤È[™?)ÿö\÷·UÆùANèòú^‘;Î1ï8߬ðÿ…rçê|¡<ÚG¹uJ=u>¸ž”G\w –WGE_0¿*8Wå¡ò­ò¾™&"Ü#î{æÓ‰•B1~§r*u²>O4aùtåóŽÌWB‘‡J1?¤ÒDˆ)ä®Ú=®oÍå ¯˜/ÔïÎåz&Sô[ú´SÆåKù\ú/–scø¾²-—oÏ´›‹>¨ö¯ÎƒzîPOú±V|Þœù‡^n£øƒÞÌw>ïÊ÷á×UýWûUùQüüüŒ”_W>Äåñõhª]&bL1ÏÐs´‹x€y+`Šñ²ÿSËI?¿ªúAPußQ•¨j‹ê_T¿;þµ•_µ?U¯Ô~ÑžÓ8×*çàS_ˆÅLYïÓ☪'nþ åÐô¹5SÄ3Ø%ô<®Ì'(ø„¾ ÞW°ž7SæoƒÁßwJý|_/Ì'ôµ1—oÂõ+™¶UüCøoâv3Üçvzðõ­ø~ ýtS®«~ç±Tû”ò¬àóŠÿ؆¯ïÀ×;c^ø:Ë#Ñ…åââ±zîïÐîOÊ9Ó4¿qcþ‘‡²Ÿ‘åq9r¾’ØUáωB_T½QÆ-ã䋸…yÀ¼ \{¦Ý’©í;õ£Þ‡Ü”€Û…þ`žGCnð;(‡ü£‡"./ãüi%—ïžjGrÜM¾‘O 1ô#TìÅe=eއ)æë‹ìT/jþ§Rħ¸ù)r”þ\-þ‘wÁ_"NôY‘Ê'¨êOœâ¨“Ÿr[ªõÔq«T]'ªí¨~å ?Ÿcž!Øæz¦¥Ú«Œ#}Xz0eýSób9è§•âŸ0?àWõ#ð7¸®ÞWù‡Ÿ‚ÿ`ý–ö»Ež;ßFÑU!gUÿÀÖ°÷Fz¤úßuú£æ“?¥ÊWú ¬/ÿ{ñùnLy"½7¤êŸ›>ªvŠþ‘ ÄõnJ=ø%äW2¿àq©ö‹yçs©‡¬—Rn?9Èó‡ë*Uåö@U¹¨÷Aqí)÷•Š="/ëÎqr„mÎë2äO8ï´%Ëz½î¨è«÷!7Õß8øÓ´ü°5Ó|þ´SÌWD±è òbØ]3È‹¯ƒb]ÌqFÎë:…z?\WíWµcu_ÉÉO©÷ÕvÔsUïÖ)òåý9^uQ ?ÌõßáŸwd¹5q‡ª· •8»mýKÕ ©ˆØÇƒ~ôäs¬z²¾wáûM”ñ•ÛËOú øÄ£J>߉iÏF9ø½*¦ƒŒ‰O àz¸\¿ÑÆy?Gî뱟wWôÃÉ®Wî«~ßm¿N9—öÿ8ÊûVivì§ùô§Úº/¡ê-òRÕo¸é?üÊJ;j|‡^!~c=†ü¿"U_Óâµ:깓ÔëhO•/ÊãÜa?3-~+q ãJã_é×Ü·çyoÇסÿÈë;*í*y”ô·ð7ˆS¡ïLw‚?PäŒyD{ðNúïõÜiú·V)ŠëNñ@-—m}Pä·j¼D>Š8Og)wÌ#â7ìy[7®×ƒýS/Ä ¾< z!ó[nþûÍÈs±_ƒrÈ×Ð?öuÀÖG;3EÐ蕺^‡Ýªr”ùvÛ/qZ©ëeÿTæãØÿƒŸQíØ‰Â_©ûÕåóì‡ã9ž{bÜØ¿‘y;˹1Ÿcÿ ûÐ<þ4ù!ªÏïðÜû}Zç;p{ÐW5þ«ç ä®–SæK>·i¤´¯ÆoèÎÕõ äƒü ‹"̇:8Ø-Ê«þPÉßd\‘ë`æíc¿ùÖ½Ý@•þÔ8¡R5N¨y§CüzÖE¡àyÖæ·óÂõ±>ì–j·ê~}Ú>¾¿·±ÞÅzùôç耢?êxòuPu= ¿½iê ×Në8·¸²^áýªûÉjPóÔSóp§¼A±/Çuºj‡Ø÷‡UýøN,¯*ÅT¿ª®×ÔõäâÏ䨖s³5Rí~y:Þ[€½ôeªîÛC/waŠçàÝaOŠüA×)rPçz ø+™4QôBÕ%”óˆýÄqÄý–|Žu âì¯+÷§ÊÑÉ?ªñTõ Nó¥>·Ä0Ÿðú{Uç×UûÇ9SÄ'ýMôçùÇzûý”ø¦æíNû½nó 9Èç$L‘ßcŸy:òwØ7žo`]¡ìûbÿÈÑ/z´_9O˜GèòLìO@_Õ÷^À’—A?œžËu̸âì¤+—ççbiëSõù<Ö1ÊûŽùŸêÇA¡ghü«þÂ鹈¢iñvPøpšG§üN•ƒoÔþÕõ;¨².ļÊüBÉd\QŸ7«ï ¨òsÚ7QóÈO}B­¯ú1§üKÈ¡'j>T}âñ¨ûNûêüoPä§úU¹oÌýÂÿ±®Ã:³Ó|ëüíxœ£Ó>•›ŸPÞO‘ó ?uïWÈõ ÞKľïNŠü!eßAÆ œKÊ×±?Éë ´÷Þ†(vàôÀ¾E%Ÿ#OÁ¸å¼+òlêà_œø÷:>ûHò=ä]ØwQírQóãSç]úì3©Ï5ÕfsÅî¼¾Ÿå”_¨ëC§õäì¤ßP?÷‚8‰õ1Þ稄<ùãE;Ð3~_Äѯ;Qð§æõêþåŸÊø@¡oê¾zî¶Îø ÈÔç™ÍRå—öy&õ9ü ëãçƒ~L‡|ÿyüž‡â=æ|ëSè'ôµ½bNqXÕ;è ô %÷‡¸ ½A¾Ž}îFƒæ¾.×ë«è¯:? êû+êsÌä‹ÏY¨ú椇¬/æû»©ó.×â.ìÏÕ¤?çú;æM]¨v¬¼¯,õú%Ÿ0…½m§Œ×ï{›}SŸ§€_å9æ5íùê9]WÇ­îÛ8½—ûÊû" å…|½¥"'W`¿,7ì?Ë÷tÚEÑ3Èác…eßHêü)òSÄä•Xw ‚ýâý ø¼/»CþÄû rœØ7Á~ú]`ŸËi½ù*ë{§C¿ŸÐýB¢fæ/0†p1áÂG.ew±üß›@YA`kB_«Ï¾ÁÂÇWVª´Px~…ðºfèäㆠ4á²7 ¿£Ô¾(ªézœnÓ¾W¹ÃX*m®]kù¿3lè©åÂ/òõ°}ÛÚkLÿ"<Ãÿÿ`WÖRçgÞ-D,c_Ó^Ì1Ó©åƒVß3ÉÂw5a°å|+ÍðÛM Yéó%²µ·2ó£—-d_ߘÐ:=n†[þß›0Ûø?´Ôr}!Ó}GÀ‰:õ}«"%†k_j†}~k™£W•6^VÎùÍíç1ki{àèøÃ Á³xÝQi#ÍÓñ — ºôý‡Q_{UøP‡r'ghcá$ÂùL)gÐN5 ^o3ЇFžEÿßÃå*ù|¡BéSäfÂn…ßþ\3rg~µ”¡ó@­¥°ÒgËÿo9—ÓsÑÉÇÈcø:ùVíuËùc„µ„4öß\÷E£~ jéC™«´ùüŠð¡ñABÄ4.½«`z)—¿Ý Á ùú7 Í4þ_ÿ/a¨¥â-PÃ÷ŽHõµAʾƒaº¶žh_¶øè—0‘.<Ëí<¡Œçrë9_&â8Ç𽌄¥\&Éí}«Ö·´0üK €ý̦é|­ÉPOäÙ»q"†t3¨v·Rn‰åÿ]-²úD3æe3x¤ûb‡IÍõ…Ï ÷õ8ü»IE¾¯­"ºS:ÚŸtý9c,z¬žBô¥²³€˜»ÉÎã5ârMèðk†Ì´÷5#kžãÚXMXOXd¹vÖTÚ¹„3K 72_ÒÛÐã4%èCL‹¸ýµ n%\g©s$Óbµ¥îƒš±î¸\\æ²6ò×õBÄîÝ GöÐŒ58Φí4cŽÅ:EøŸwŒÿC÷Ù´}ªR÷U®K4ÛjSO|cÕÍ\çÍf­@×Z:1ºh–^Ó}»aÓˆE|}¦ç…ð¹ˆy„G„n",¶´ ^çæ•6æf0=€cóXÂ`òsßr%¦jJdút>9Sˆ0Jcÿ¯ú"t“ò7òLm‰q=8Ȇ§ZÂ)lO"_¦u¢ž—¾mõÙ~Ä>Á/†-ëýÊmÖôZ#?ªçVç*ý¨ékê@/Ât¶ÂNøžÈinQÛä{ÕLÛjFîF¹•ˆ3mìËk›4}í£ûÌ.OznµŸ¥Í™J´® !y-ú0š)Ź º^èm×·Ù¿îÿ>6üe R©K}„~M½fäÕš’«ÛÈB¬#Î:®9kšÿ§k×jFœÙ…}3ÉYä=Á‘×)u²Õ}|–¡Øå¾™ð®¥Üt¥^[åüJÆ5š¡3b]*ÖR"Ž>–Z6D¶Ÿ(úŽx»›ÎGk–øãÀ§ˆi"g;¸Bß”5jbtð}Ö§;Œ½#}N/ÒŒõ…Òþéò‹4d‰õ_è-±ï£É\]øAQWÄ#·¢xÁoÃêñAÈöA.Û‚©Ø’ùƒfÎùw'Ÿ±RðDñ 8”¯ó>˜ž_‰†ô5ø‰f®™O²È~±¦ï‰jäo ø¹OXŒo¬©¯ºßåOìŽ7úqÜÈK5cm/ÊÐzA·•_¸¬X›þh–Ñã“ð“ÂÆÄ¾…°ÿ¹lk.‡ç(‡ŠÜõ~Â-t<— h¼¯­Ï Sa_Wóÿb/Nä“ä3'kz¾¸še)Öà·>' ⣈OGq½c5=ïy´Èy<ŸzŽÅ² YÓ¿A±vžÊ×s˜Äç{±Ýïfα39Ôm^ä$ûð=^˅†m)—{ùb_^Ÿ£C4= äò‚?øi1ÇS¸Ý˜Šõì&CÄ^>òå`? ?b_WPâ%¸‹1Þ ùˆè—ŸOÉýnQöBãD¾†}˜¥b¹_á7„ž‰üZ¼ÿ½Æ±—õ£ÛëîÏk¤^Ur½+ùœ|›ž«ó~ ¾^"?)âŠ. ñ,î/áø>ö:…®œo¹~-ÓeLe.¯ñ…bžØÓÐínw¾÷Sº܆ÿ'}Ñãyb{ 󸄎± žjð&×(”bßU÷ç{ðuS÷²ÌÇLÅÞÐ8n·ë‘È™ ,·%ö;áÅ^Ô©|kHغðOºÔL»Ä>°È-Â\O샊§ÍbænÞ÷:‘0GÓ÷¬°®ú8î'øœ|râpPÄy㻳åwžÒö _’(?wRçŸ;W¾TÌÏØËôË“q¾åóÈ›ñ]çÞð¦ù>_/âëqj´íÅ\/n æëQ½XÜ|7ë‡Ð €Vý5üŽ&Ç|X(֋ƒr¹¨^}»„ÆçeÌG×/ýPÿ'QLè¿D)ó!ß!âþËÅå E ÆO”+´˜ÛÅ;)h¿œù€üJX¾1ÁõÜ¥‰¨Îþ‰·‡ë%z±­å;I¢™Î¿$¢wè '¢ÌWˆÇáórÔçr åÜÆSÌå0¿E?_3ÿò³º˜ŽMóxð®M)ß/Å9ʱ<Š ?OäÏò(b9òýb®_Áü¹¿0ÏS„ߥsý _q}ÐR–äXŽy‚>s?%;_q{ÐKÌ Þ™)aþe»<Ÿr~YÎ_ ׃¼Ë¸<ìíH=âþ ¹Þ9q=Øø—ŸQä~+XÏñÙ,Øk ì˃/Ô“óÃï¤À¾ ?ò»?˜?Ì[Œå ù”Bÿ gÐ/–#Þ­*äsØ}”ëAþÌ+ä=e½Â.X>!å1Ìì®”ë5Â<‚ò8‹¹_ø‘"–O!ähô_õ÷‹Ï†–q9¼ë9³I»`ZÈýq¿{œÛ…BŸ ”qÂ.YNUòý8ÚçùŽñ<–ò9Þ¡Å|ÆYNðóEJ¿° 鿹P”Gû<¸ÿ ø+n§ŒåP¿ÇíBn2Þð}øûrè1ìóŽú\O~' üƉña=Åu•Bè“"èüºôÈ,?Ì3(ø«`=«À8‡Y¤¿E;êø¸>âì=Ê÷70>øÅè7Úay"nÂNʸ>ú/TäXÀú.õ”Û/`¿ÿb;†Ÿ‰pyÈíÀ¡ÿa–â‘úùò³Þ7æõ9 »ÿø@<ƒžÀ¿H¿Š8€<ˆù…_ŒC_ —Êü ®ÊøÊòƒ=â]zðåû2Eždð)õDÎ'Ï_)ü ìï”"Þð¸à§1ŸÐä3ð3øBÞ‚|N‰Ã˜_/•<ö=‘¿)ƒ¼–i9ìzƒ+0J~ùÅ:ƒ÷ª~‡ŸGÞÅóÓHÑ™ÿ _E¾ ÿ€xÍü€é‡àO§±ÞPò|ÐFâ7Ï#ø@>ƒv7‘‡aßëe¬A1?hGÝqCñ—èqñ@ÍWàïÁÖ£2~£<ôQY?Êõü«2ßðï°ØìùìD/æå+ŒöÌcžÑ.üâ÷ƒý ìO bú}Fþ}æþ¥=Ë|yâü¢Ò?ìEæIðð Ð#Ì'òSä³ÈO1äß\ûM AÅ~äzWYËü yâ;òEÕïЇœå~‹²¿ƒsØøñrD¿ðŠ_Fœ“סO·‡¼=?Ïòå¼¾j³²ÞAÞ†<ùU!Ÿu“þT‰WÒ_b= ÿŠ}.@~…ýGìÁoa½ù"¾"OC¿ØCž…õ/·ƒü¹B·ÐkØ-·ƒ~‘ßÀŸÉ¸ycßóÈòAyèyûOÜ>òNø ¹~ƒßAþ¯øs¹¾QêËïØu˜P5/Gr_Œå¢æYˆKÈÛe^©ìÇAïÀ‡ÌÇà—0Nè#â6æñãCž¹*y,â4ì6®ÄAøo¹Àz’ljÏRJ;Bž‚8ƒõ”²ÿ)ã öSÀüòSø9•_Œþû­èù‡ÿ”y B很Uÿ„y‘ûr°wØ%ÛKöx‡¼që*Ø üü:òe¾ä>ö°¾F<…ÿÄ8¸2ÅŸÊu úÇúëV–/ô4m¿rB<€Ý)ñ\ÆEð ¿úÈà×±ÿ ¿<@Ùÿ‚¾aüØ×ŸCü Áß!Ž!NAoÁ?ä~°¾áö0ïÐÌ ò,ð}!ìSXΈ¼ßRµÉ¸.¿3 DÉS rGÑ#Ø=ìVÍëÕ}.¹ÿ„üz‹u:ìëž'ç1ŸˆËx<qqWyî†ëÈß1^ù› Ø/‡ßVöearŸë#ä…,Äãüìå ýƒQ÷obÜìë3P¹ï¢ì K?‹ç]\û¢að‡õ&ÖŸØC;Êsè⋺îT÷Õ"Š¿(Vâˆô/ÈÔ<‘ïKù ÿÄs ìK°<‘ï€"þÉ}Aì @¯¡/J¼ƒËõ4ôvÍz*÷]Á߇}©ú‰þ±N†?‚}a½ ?„ëEŠ>Êçȱ郎ðÙw|–\®«1.ž¹_¤®‡”}UÌ+Æ#×ÃÈ£à7G ^ò8á?äóøä<¹o‡ýeßqWÎ#ôù&òGø–W…ò\V>OƒßDÞŽsäʾ˜ÌwÔýÅ.äu¦r_õ gðûÐWÄ[Ä=ä1ÿ½÷Ž®ªèÞÇonK!ôŽôÞ¤$¡ã™P¥Iï5¡÷š€bQTl4Q±@¥ƒ€€4éM@Š€RE~{ŸÙÏÄL¸ß÷]ßߟ¬uÖÜsî93{vyv™“¹xx »´òàŽ±_à%üâo¼OaÅ%ÏÔ%ÁgøeéÇÔó gÌ~ù,âäÖú©ÿÃ~„_?à_Ï‚¨ë#žCÜnå uWðúz§ÉŸà§QWþ"D|?ƒ¸y"â+þ6ucäÐ[+Ï4u%øÄ-Vžgâ6à:Æ‘zy`^ðƒùRwÅzŠUW\ÓXô`~&¾F ò·æ ù#BýØü†ò|øSømÄU°'Äòý¨gÀYqôvbZ¬ÿo­üÜðv ýµôÖøEøä¨ûÂÎÛQ)ß?€\agÆï€Ë¨c½~Ñ¢Ëø ÑÛ4Ö÷fxhéä…uSa~Ö{æ½¼·ƒzÞŸžbþR¿¸¤ç¥nYñ2ü…ñã˜ìyœ'#þ@ø}_mù .~@¾àƒU—48 þJkÖõ€ÃV>høŽ|ÊŠGMÜbÕÁM>‰¸ÀÊo2Ùë(Â7ÔáÁ7ó[%˜/úE~cë›|€¿„ü÷Ÿï¡ßìz…8ñìTè‚Ýÿš: U/°×Û-äƒñSå“ð×ÀO G žB/D®az~ê–•/™unøO¬ëXõä1Ð_ðÍäÒ⽌ÀCÄíðKÈ[ÀÄ þQè0­•AÌ:™¿˜x ~z€¼ã`|{]ÕzoÑäKò<Þ ðZý"?0ñ ê¦ÐÈ r=àâg«nˆz’©ƒ ŸQo´×c`&Ã÷Öº«y¿ö$t N3u0äwðÐ{è-â;+~3ï[YñìþĬãH¨·¼†^Yù°yuÄ툟 Ž”ëf½\ú5ï¿PçE} 8g¿W‰¸Ê~?ü_Ï<·âÅPëæ ×¼ïb× ­|ù®]'2ïmÁA?,;Âû9æ7ÏÀøOðSóM]×çê¢È õ^âø7Øü'ì ~õä£XÏ@¾‡ºêîVhê¼Þ›÷ ð^˜´&Þ¶pÀäWˆÏÑ/ðݪ+Úõuã7-iê_;üâ=«ž`ýVGr]Áâ—©‹#®Ã¼/ öŠ<@î3ïÛbÝ q…çÚþz`pÒZw7x<öõð ø»¶ê†.Ĉÿ!?¡/ÜD\ƒë騂n¬Ëó^¬“ç…Ø p?½Å?³½´òã¯Q?ż‘§ÂÊ÷æ½Èqâ9áƒYWƒ’ù Å8ˆ#P¿2~ÏZï1ë;ÀK¿àÿŒ}ƒ¿VÈÔ#´â.S·‘ç|!/Øü)ü+êrÈñäˆ88Œù…ˆ»€§&/’û‘ÿÿ`ÕiŒ\7Èu¬sàýlƒÃèv§ù£n!€c}Fäcôv}ÎZù¢¡8? ½oÐ+ÍûwVÂ䨃£Žˆq€Çìqp~K憺‰ôo¿z/ÂoÁ®á§QwÅú°Y×A_žÇ{µ 8mòZĹV½Ø¼W*üBÜ‚øü5y+ä¾#îNнà½rè9pËÔ›A/üú¿-Ü7õð yü»¥§f/`ðö¾Xñ'p8|†D]Ò¼ÿc­Ã?ŽyC?PGƒ ßÄú…æ9Ø'êf¾˜8q&ÆþÀî¯à|Fœ`Þ´ò™4–¾EXtÀžwƯÿ¡ß°sÄŸð7Vœf¿·oóϼ‡dÕ™Ì{ÿà3pTÆó Ÿ‡`}ÌÔ?1¬C?Ì{ð2.ìÂ~/q¸y/ï ÀŸZ8?bã¸íÿ‘>ç g®#/@ {}0ÿ…õðtA¿·"OôoÖQP¾YïU¹Â".n[þöhänùá(KÍz!êXÏŠM\,ç OLü€8qâ_ØìñêÐð z:a?Ö:ªׂÎü,âaÔÅe|Ø‘Y_±Ö·Ìo¤ïºuͲ?Ôí nÿ¬xɼO$÷!Nµë¸g`7ˆ;a7rŽúü•áê™V}Ö¬çw`'–¾"î7q£ôƒ÷8ŒžÉy˜ÔIo`VÞ<0ëqÀÈÑŠã`ŸE`o¦=N!Þ±â\SÅsî<†¾[q‘ɯ¡gX¿só¾²w[uXó;îCžmÕ;ÌÿÃ!^Eþy©Uç7õf-?Ù;;ø%<žÜG=žˆjôù'ÐÊãñ·öx|Çè {}i<ï ù¿iÞëˆ÷ÎZáñd?M÷½NÇ3t¼EG):øÞAtÌ¢ãSÙŸ¡$¼ßÑhê‹ÿŸºèszjåÿ¼}•õ>¾©ÝGÇt8z€LÜoÞçh+L-Ñœ@ôS_þψnz.mu'r“Ç“¹´Ç“¾?@שß4º§)µOS[ŒŽ¶ôy}©ÿstÔÓ{(xyÓë¼Ç"Mè8L÷ÎÐscþþ¤£&uéÈN×§c"õÃs'³õž×{Uøšëùñþ îžL²ŸE¡ê‡x$$V ƒ?[z¦}¾J­Gï5â›,º¶•²ÉÔwƒŽª²?Ùktðþé%^{yoÒWoaù®8ÝGòðýEcåðx¢§“쩟44ÏÈ9DZMOº!DÏ‹æì?E´¬§–tÅ_Qéo¬õ‚û÷ÿ¬Çö‘îå@ýPá$§¢ôl°D‡ÿ:vÐñ =sB¾~Ô®’½½Žêg#ˆÁ÷éx™Ž‘“‡õ ­þ³txé 9úž¥–Ø×•î!{ÏH÷îGÐ}QqtN:à¹l}$; ']Ì£çø;š«ïQHm8ÄGïMÞŠÒm_6±Iâ]Y¢1êâñ0Ší­ =LzãîåÆ{G.åä}̉>/ñØËûd^x¯y<d&úÖk½ 'Ù1ȹvAôZÒõítl`ùŠîo‚dgAœé‰?J·Á^ús)²Ë(Òåp‡ÀY-›à\úLö$>gÉ9ñ0H6$šƒ¿Ó}Ò±BÛQé·Ÿtª`:'ZüWyÏz†è ­á$£ Ù•ÿ!:Þ¦k,Ò£(‹4ÔwZ²ËH¤¬¤ÂEÿ(¢¿Æ>æ§y¸Nä}UóÕµ¯zo7_Az†à#}õÍÕ{¸öVHï}²e[ô¿£yN÷û¸%žD°Ž‘ ‰¿áDS8ïßYßãy0?ÑD<'¹¤!݉ŒÓXä'YøH‚Ÿk[ '{ŸüItÐ…anq®p²éð÷è>²Û`º¾Œžc½ç{O³LžWéÖ_YãC€1䲦×÷¾¶SïãG¸ž°ÆW]°ƒ÷]¢>Y4~»óœ¯ýFfÒ©4ÝI¿{Ó’½DR?‘Äãéa ¯ÖñœôL°;+õ$YDö‰÷~º\Œp#¸‡î' ¦Në{+õxrг9£rÌ3ÊþQ¥+‹œ‰çáíéþïDç¨ß÷Içþ/ô½Ù½šVÿú.Blˆäég<˜Bß.ùÈN}Oзhýw÷#ÚIéxžµÚ#iŽiÂIá²×©»//ÙŽ/?泌ýš—dê¥q¼¼¿aŒ·3?Ñ1Nì³|&ç£þ}å51v²¯âÏÞè |ô’œ½ŒõäÓ—¨í%x>R°‹ñ“ìÙGà“=a²’Œ²Ð‘åtæx\ð‚t<@º`Y¦ù¥¶ƒì…rBï›NØØ®„ö¥^ÞS“÷¹'~…ñžrdkîžïâ‚Dgæø]·þ7õÞK>²!ßs'ˆ¯ã}X6Ò1^÷áÎìÝ×]ûÙ@­ïþU_l¦ï·H¿" _Ó¾A:³IûGw­Õô=ûð “ý,×.ô™øHs!܉ˆÐcGMiÉF‘FÓ\‰oAÒ ãn¬Ž-üÙ´õ“-EÞE’m¥!™F²Ò¼Ü=Ryo%ŽUx»Özÿ÷÷ˆ_î~q¼§ï7ËûÚ|èÑ{V]¾µ«‚²ßLÝ`<+B}³!>EÐsádá-4NøÉö‚!ÁtNq…ï ­¯^ž+ÑÂ%ïB:úéÏnÌRHãÛ“ŸýÏ:­ÿád£þ úð‘.ø¶pÿ2Ï•°ÌÏ:ÊûÂþ¦õ5¬ŸìçÃû°}~ŽòÐwD“ûû´á½ÌܽxÏM’kï›tXÇ\®þÎC3ÐA€N¸šŽô+-Çpl÷¤ÃDSÙRàcDƒ0?Šã Ò÷`ÑK²ÍÀÚø9."_ê}[ð–ÆHC󈤹¦'žf¤gÓó~W3©oºN÷FN×ú y ?ÂÉI7¢ˆ¶Ž çÃI¾ág„—D—ŸqŸxä?&zæÑ¶Í±ãb€ü𷬯[ÆÿR:h~>ºîÓóð.ùˆ?^’ ï-ǶÌvá%œŽ&ŒKCv“f“̃äIϤ!ÞDÎDvÖø Ýóq ”þHW½²ÿ¦q+µO$~ñv¶|p¬Ï4XÏÉv½ŒíctÜÀ2òž–8LöF r¾@ö$½ ¯ýÏS\å#|•sˆW^š£wªGïEŸÝý¬yŸ@Þ£ô„÷t÷ºâ}Ëx¯½ãz¶!Òa?ó±½ð’xåî¹úì]E´ù Ë~]¼×ÿŽK{o‡‘õŽÕ²ð~¥uÅדÚ:~g=õõÐ1;ûH¾(ÎiÈžÂIƒ_¦ ½¼á´7‡–%c¯—÷å}Ðx|þ]‹¥â{s|tÝG<ñËï ø3è\ÁOrpó…jZW\ysìÊ1c¤È¸œ¾Ÿ}Ÿ};ër´Äû$Sß»šnåK^²ï Å$¾Q:~wýÜqÅ^šŸ›/Þ¢–1p9’°. én4Û·üf’ô6@üóSÜä'Œò1¾_Ôq çŠAÆö¿µoaÝ 4×ñŸëOúÊüÚjZ šh,tiüKËßµ¯=³¥9ûŸÿ°\óÆÅŒ?tìêß®û¼£ýç¯/ã-Q$£Ò禉õ“â­à ú<œì2ÀqÔ ÏEŸÃY¶ä[ÃÕþ×=xK⟟} ñÅ͵IG|µèàøy¦¶©òGÑ$—hê'É&rŸèEO›ø8ŽøVëX`¡Î‘Ùº9ßë⠞ЇySA|,õí•=Õõ%— 4HüÎù9Ç<£çxF0‘x•Eã‚»_>ÍÓÝï÷°e½§XÛG¸àe_@1LýxG­;‰{ÚH±äšck€æž¦¢¦)pLp™1‘÷_%šÜ½²ù—89f¹‘ï çx•ñpÉGòð}¢ýcehö“üÓM—8ŠôÈ˾•u¦¦Ö{ö·\»ä|Üwžø3AǤáÅ5_|[4NúXVguìIy˜[Ã[*úW_ë çz^ÙGÔϵÚCó|Ÿùvѵò:âzçCA¢‹kAŠÍ¬7¯i] \ý¦s%Ž•]ÿ7O×ÁüdŸ~Êý‘:çàØ×û'+Ʊ=ד ïüGuLNs‹š­ñÖ;WøžEÇlì7\ìž%qåã’‡ñÖƒt.ƵWGÙ‡Ó¼üש¯5Dì×Íq¾?§Žw9fYg!™ñÞ¡Þ(é§´Æ7÷7®Šk^ºû—rç;y¯VΈրì³êç˜ï'ï¹µkƈÆ;qNßGÆ >Âh,t눌 YµÌÝßiä½ii¾œç…í¬~DâJ’¡ èO‰=XÖ»Ñ<¢®ÉÞ©nŠcNÒU?媌ÌâÇ6m—œ;³ïàú¬«;½Ä J G÷óžò^â³/“͹µ¡R‚[e5y ‹£HG")Î Þe/ìh:OG:EöÕZçœÿ’ýDEÒ|#H7Ã?l-¡ã2Î{ÝšÉ&œõŽhæú¿Ÿ1¯„öù>н8oñ?ÆõµŒ·t$iþøÚ‰¿ 9òï^ºúÏ¿ïÁxB¸àçß‚ß4µ >>–˜! ºÀXµ†§5¾°/‹$ìŠ";HÃy×):x½ì-@¸Å5®A’ÌÓ’]ç"ý÷±íÒœ|qZ&þ:tÏâúÇÿü."ùpoC­aŒÑš·æ0Xtö[#ö¯×u‹HÊ›8׌â:ñÍÓÈÅÂSâ ÿ|ׯ¹¦ÉqMÄB‰w"$~øJç¸ašn®™¥#›þskEqü@ñŸrÿ{2¶¼#ñYóº‹xEßG^®wpNÒPçŽ.¶0®µbyð>Í\+¬ãa®‘û9–yEÛ5û÷÷ªð;>ŒS”wòš‡—uâ¹ÀrñZ“+¯ª:þqcŽó ¹NìÆôµv׺H¿‚ô]€ë‡\ÏHÐ:ãï«óWÇI·½”Dyeþó5ò—Oq^ÌûÔùx8ù­4«ô³nžÝWǧì],ç=¢7H-…÷…ç:c­ÞÑwdÐ2 Žêš'n~ľ.‡Æ¦Ë­‘»kã]ëVp²–áIÎä£È2v%yÏÔñu:Òß´ä«Ò­îÞàë´Ÿ g½"YFMÒþ4‚ì1°Gûàâcðyñ?h²ÜÃ9w¤ü”÷s"¾G>¯×ì"ßÐu[æWàˆÆ6Î?ë¸ÇOñG€ó¢—ë€ìÓ\Ì&ýö¡ïc^¸5zŽéy]ì0’ý2å­¨ïGtM€k¬QkS¼ 8Ò×Jl~›æUp¥Žƒý,ïb:^JK¶‘hÎH¸—ð5}]Ϲ5gûÈ.£ ßÓσ$»´dáFÊï:ú(Î PŒîo¦õÊGØì¡câ` ]/ãx’cæ ç ŒW,oÖE®CP¬ã}Ù£s\޵˜¿ E9öX,zCXš…b¾hŠ;¢Iÿ3‘-¥á|–cæÓÚÏZhýá|ÆÇxÎñ'¯ÅDëyø™ç¼NÈël×´.±ý?¥ƒøžAÇí³sœÉöI6™Iè§gƒ¿êº ë¿ìïr">‡¬uwLÐ牃uÛUÿó”iFë5l´]+êëVêû[{tÛVîo6L¯7úL·­sêç7Òãõȯïïþ>ï~F·}^Ñ÷wo—r¼Žò\»kú¼“ôßã)ie>²C}|9oZWÓÓ@úkºP÷ÓXè5-î—ù<2>å÷M伩œ7”¶º¼cÓiîüKÜ.téâû”“ùÊ<;W> ¿úN¾œÐ烅îAÝt?I_ë¶ß—"—>úùkd|áG+áwB6}½£Ì7aðQÎÛŠÜÛÊ|Û£•yv~´ºðäØEÆ…\ OäßL®·ñ¤lÁ×vò}s9o&t´sAó} ë>´]„¯˜?æzÁ´­És2–Ee>2Úæ"§æòøÑRÆk/z}Ãxò+ñ„ÞæBgG9o/÷w–sÐÕG®'ÎÖí‰)ûXÚGäþz2Oè/æßPô¨¦Ð-¿è_OÆk„ùÿBŸ­ÿ…Žnb—v û‚¼p½€º ]å¾®ñ)ùÐn¥Eø#üï”Y‡~ú'èïaOýGlœê&v7v"v œé-ôõÑ›–Ä÷ίï,Ï ºú `'I"—:¾ÙrjÕBßBè­ ¾ƒ_˜·´ÀÃÞÓuÛWô ô úж“yöš+¸)t?›Ë¸°K|û^ƒ/žÕç['[2Ÿ>ÂÏ¡å\î(òê×"%Fêë}~ ¿Ð”ùqz ¿û ý-ç ™SÒÝZæ¹À®ýË÷¸ÞRîkiñ-pI~=#^~™Ü̧‹|ñ ί Þwé™ò:p®•´¸¯·ð zœ(r„ÝnœCÀEƒ;BWä$m[È/Zìt×Mé7š‰¿ª%t×}¨}I_ŽÀßÈ›FÁ·ÒJÅ2¾§Øøgû àøÿ =ì"ri!-ü ZÌ«èø‚¾`œžâÇ—Ñß·Ï/|•ûÀøeØäú Ð_àµüú”Áµ·Üoðï/Mè5ø&qU£ž)ùûĸðÿ «§ðoˆÈ×g çz }Ð _Ëy?¹¿©È¹ƒøè üâ¨fÑ·o›À/Yzsù•£O˜'ôzÚïDJ{28Ð_ôz|ÊùöŸ²…~ƒnèq»XÁç5)çg®‹üä×Bã…σ³èûzí=܆=A¾=ëëù –ø«—|¸vßMðñôñY/Áížûd\é§§àæëÐÄ¡]ïõOÙ=‘8 ¿;?: º‰|º z‚Ρ[ÆÅs†ÑÈç×ÌCÆIëð/Я^2no±Ä×ðëð‡ðŸhaW&þ•~zbâhá7ø ¿8ñº¹ÏjÛÈ<¡/À¡¢ýDжù!®4ùŽØð¸aò ™/ìuê}-æ ¾Ã/"¾Eü†ù#ž¿_6“ûšH‹xþ³¡œÃÀ~º¿Àoà{÷¢)[Ø;ZÐÓLø½]¶A'ðÅŽwÑb­¤5Ï ?ØöþB_M|*ýJÅÙøgœ·“qZË9âFðùZO|„NøY´ˆŸF Ž š¦[Ø=ìó€=š|^Æ…A®ˆžé­ŸSàktJ=‡¼B_ žN?`ž†/–</¡E¼$ô ܪ¯÷-!ô‹ß„_ì#ý ކC ½‚Ü‘O!o‚¾ï ·ƒÈ½´måû®ùu‹ün`‚n{É9â^àÀ€ùzœ¢ EßÏ¿ q¾‡¿D]â«E¼ž¸2e }5q»àåà—'…NèòùEÏø„Cò½ô×UΟŽ> ç¼uÔÛºÿ¤!ú~ó¨/ã€ÿÐ;“ß p?âPùÕ»Tñf]9G}¦Žu=„Ÿ‡D }6óFþ~uùÚþ-âOÔoÐâ{䥠xnê0"?øÄù²‚câzè!â?øÔ“ÏA ÏhQ×Áøh»ý ûÁ÷ˆÓÐ"Î=§ §µEŸ1.ì ×+ËøÕ¥…="o>"©!ýW“~âésùåÈøxéGÞÞ4ýÖ°®¿P?µë< ðñp³ƒð~ÌÔD?€ˆ“`?¨!ŸáoLþ*8müú³)ù ºÐ"Þ„{LLi×ÐãaˆCD‡ˆ^W‘G Í«ïC<4@o?TðuJàXR[ÑG‰{Ÿ=:OÓ7Tø=ì%Ý>ºH·£„¯ãäúiË<›¼Yø ÿÿ;~#/„|ëÈOêË}ÀÔë7 id݇zôõU´ðv?Ü;(z³ ¥ž!^BýЮG4Oyýi¡D}o'¥j…oˆSPƒ¿δ”sÄu+âkØZã/¤ô¦~|(¥|§!N¼'a||x vÿÿ ÿ/÷=$­#m-i–ûjI ¼_P€¼Ð".F\…xy p»·ðsÈlÝÂ>ÑÂÿ#âˆ]  ®0Tè"vÚOΑØuiƒ£à“œ#n†œL<L‰;ð?ÈsÏôZ™r~ðÇðsè~r<€çˆ[€gÈÇqŸÝâ~Øi3«Åz ôt@ÿ0®Y÷»„ž!¾€ïÀ3ȵÿX‰#EžhM|)ñÓ(íØ’ã»_ácáwWé§ïÊ”-ôr¼p-øh¯?˜8ò”xq!â¬?tµZè5üα^ÿ<ý!FœÜÁ:G¼oçO¦®T7¥^}’çïÁ~qúÿ»BÞŽü-ü>ü+êJÆN¥Ô3º ½ÐÄ¥È{Ìú†Øòà<â‚"o´ˆw0oØ ZØâ'Ä_¨£Á>X×1oÐ…uXÐ…ëˆ ºÉ<ëðÛ„N;B<„ç̺ŒŒqîGœ ¿ÐNø’˜_ßßEøÑQžG½»³ŒÛQ·J¿Ã´£Ž%üA]ùæõ_ÈÍðWÎá? GÈ#WÔ–þüƼ w}åëMv=õ.øQ䯦® ûÚ“×aÿÀ·Ö2Oà€Ñ[™â³þ$üìuH·¨S£Å¸6È ÷¨O¶’þàÇ‘§ÀÞa_Ýå~ ¸þ#ž¯$öþÃ/ãq;ü7â»N…q€Kˆ—ì<ñ]·„”ó¿€wèëF¦n~=)û‡?B= u Äã¨S–qSh¯×MIúÀ{è;æû‚Ý!î}ˆg1ÿÚÒÖ”qä ùƒ~“ïXuhÉ÷f=CúÁúH·ü)[ô3Hôú‡uCè5üïp©'˜”’~è-âÌÓ¼?#þÔ^Ÿ‚Þà=à.â#SH ¼Dû…BŸªJ‹ëˆ³‘¿Ô”q .ú¡G°à%ô8 þÿÁßÂ!Ží$-žÇz,Þ÷¾Ao1>ðx?H¾G±Ÿð­·ô×g¶œËuÄ À¿Ë¼aov½ñêàc5iá‡À?³®$tš÷Œò _….à@w\:;K œ„¼ÁÞ0>ÖSoc=¤›ÈçðÛc_Õ÷c=fðPì¸8HÆrH·„Nĉf]Rìrµ[ø-Ø3âø “WËýÀIÄQȯÑ_MÝ<(óú{ ýððGàêUÀÄh¡v„z'òã'à§„^ÔPGÇ|Lý]®§í÷Þ_ÃÀÏ›º‡èø »ÃxX'E¼†8ù{{¹ù=ò6Ø;üŒY‡”çMÞºärŸ‡ÿÀû)&>?Œ÷« ?„œìz3ìÓä±2ðô"@^ß_ìu>ø¯~rÿ á[O¡ëó}¥WUñÃÐè1üêÀeèüø ÿœï/ü.Ïc=z°.È™ü~ ú~™÷[„ÐËvBg3Á/àpuà=ü4p-ð õÛŠB'ü ðÏÔ;ä9à%ìõUóžè“ù¿Ì{râøè=âà4ÆG\eÖaäëŽÐSÔ//GŸ§ûÇ{/ˆ¿àgL^.rƒÇ{¡fýRähê”+SÎ׬³·Z‘#æ¹áy¬sÂ~Q/Dç°ô‡|þÅ^çƒßE²—Ô™áP?EÜ}1ïµ =f]NZàÆ…¿³[Øü=ìqÚžrüè`OJ¾àý+£o‚C¶ß1ë[ò¼gBO ß¸Ï®;ÃþQwA z Åú¬yßJèé&Ï!.‡ükË ‡ÜM+×á/‡À ÞŠºÞWÄûžˆG'\}ãS¶ð;æ½=yï…An‡O›÷ˆñ~½\G ýAœ×]úM̯[¼¯:Xèë/ãa^‰rޏý`>ÈËÌ{r¿‘—œ#O3qœŒo¯×ÿ7BÞ¨[ ž3uBégèø”-ê fLZàâ{ä þñ9ìv}.™uHy¸ºM^6[·¨ï˜÷;å>ä‡IR×D–sÔQ¿™üWÎa·ðËÐÈrènµ ò}Oéq3øŠ¸ò@¼û~B.°s».`êÒBO{¡?¹uSôÿ˱ßã„]@>¦n çÀèýÿWÈ70ŽÑ#y~¸Œ ù^Äçð#ÐsÈ ö}Ç|¡ÇæÿîÆ[tÉuøwÌò€^Aï‘¿túg`ݬ›\?Ì{c«Åu¹x:`Ÿ òF8ýí.÷#O‚ÿÇüåþî )[Èÿ„÷'1ü%ì -ú7õLi¼ä{ÄáÀ3è¥YŸòyÿɼ€—X‡>áÇjËuÄI¨/Bž˜?üŒYG“sÔ›ðþøÜO®£®Œÿohh=‡<v…yõIy€î•)ç{…~Ã>À'ðѶoÄuˆ+ÐøoþßMÎͺ®ôgâŒCº…›÷Šäû³eüÉÞÙžåt,¡c¿Gï‹ôÒñìEY“†Èç×=î~I~›f€<Çã :öÉžEëèàÿÚ™@K3dFþ®‡ž]X ½‰{È>–žM½G·ó–{FêßÃè×UîýÓ£÷ÞãÑû ó¾¶¼ŸÔ»‡¸ç[èà@k:éCGÏücÎÅ<î^caÑš×¼ŽËû/èXJÿ¹D§ž5ô}»'‹»çN:8ãÿ‰Žm2w¦í¿xÜ==‡èØ­yÃ{Yº÷îòèý£V­_È}kéx×£÷ÀZ,súBx6Gúà}Ю íd…aõ„×{õÞ2®ðÁ·°ŽŒ¤C¬Ã•ÝDé§È¥.ËÚ£÷%/ÖJÏÉ[Ñ…JtßEÑz&¬¶È-BæÉú9MæÆ/\~@Ç@‘ï1ó®ì£Å2ïÍ<zE^™O¡ë¼§cÍc—¦‚òL‘1ß;—ޱr-Ix³Jï™V]öÓÉ¥u™eáîÊó}‚ŽQzžžƒ¢‡=®^ºü[$ÇJá=ÿ£É,™3·ó=Zyߥ¿=®®…ùä^¶³×èøTîã=E2Š.°|×é1òkrûÞ,Ÿ $_Þ#;Œ÷íøŒÞ{潫sLçiú¦ïóüLGqá#Û5Ù¹‡«»Cå¾™ cïÉQCæÂ:·@?çå}õ~Ñûyyo=¦ñ´ì•RZïçáÎg¨ž“ç}Æ ¶ÆDÍSW~z4î°­± ‘k,#Þ³­ìRžÎyoÉzln]¬úJžç=t¤ö,µ¼‰0ÛÐtlÙ¿*úEºây™Ž÷vËÜÈx$sÙŽz´Ž’¾{¾O3Œô/Œì*lªè ï_ÓH°!ŸÌ5'Œ]-‹¿ÓzîÚÓ»Jh’½ =Ç#ÙöØžØfHß]l9äÑXÂûfòh|ÏÅr‘gY~¿H¿Œulƒ¬³ì#>”ƒýÞ6}¯Ë×w„·3è ~xfz´Þ2?GBÆ"7s±ŒÃ>‡ín’ðƒ÷e"ßéîcM¶FºFxéú3~ã…m ¾’tÊCºåጉ£aöñ\E—Lϵ)²7w/ÃXÍWŸØÏ°>º6êùºÇßsÂvÄ>ö<oÊ|ØŸ÷h¿ÚVdÊý0^p\À¾l˜÷´eÆû›¹zÆÏ¾(÷2¦’Nzä-xw,Þß™q~œn]^Na;â>¶ÆNŽX‡^ÚGèqÃÂ=zO­Ì¢7¼WvPχ¿sc’\¢[á26éNX6ú°ŒŽõru‹èp±}h7-7×0]ýÐÂ>жíΉñãŽØnù…ZÂgè1â©_7â¹>¡érã›eÂçDûfÊŒéœ5ý&üeÿÇo…³’îº6Ç|åJFMÑÆ©†BëG¼œóõDi[i¹¹ŸY'æÉ¼kÿêÆ!ÿ1fï¬|CäÉÕŠ¡ú9þ Ç“_‰/ã8àuÑ%æÝ32¢ÏõóSpìQ\°³ªø†=¢ÿµ´ ™ß®^d’ƒï£¸*¬±ÐÂØÇ˜u@èà8à’¦ÕÅþo„/³é;ȟý“Gð8‹èÇ®Œû¸û+f{¸"÷3þH,ãîUÊ{QÏ“{‡zÊõB‚QÜwvé'ƒèÍB޶‰NÎÙ-•¹°^¯ôèØâ;Ñ9öw[I–êÆð´ü,cÜd¡åiÑs~ŠãÞKŸ÷ØNôr·¶UŽk—]ìš#ÏLõ$cÚXݺ8òžÌí©·ôý–\‹™²­<+´Kùà·'¤Bàb5ÅEai+XW6x4>0ï—ˆ|¿Ö}òoÙðï)¹>¾©¶÷ùœÂËòZ¿]9³.1¦³Íü(sc½/|'|ªéãØÒÕ¶Ú&Ý>û‹²ÜI?k›2zÇ9ë#ûöÿ«Å–oØ/~)÷°¼Ø6÷9Î`LZ@ÇAé›ñc0Æ©Â#Ö[öõ.Ö°m-×:áÚÇ錱c„oŒÑ¬Ë‚G.îòoNÄ%Û‹§Œ!WeìÓ 7yŒ {Yú`|ì/zÁ:Ï8Ê9ã2¡ë=ÑÆÁOEîl«²?çÕY™pÓ±ú²ÆH76øV°i‰ØÈ"?¾—¿ã½K9ã<‚ótŽɼx/S¶GŽ ¿YÇËHŸ¼çè5±5–/ÇÍñ2GÞ[³œÈœ1—‚÷Ì¥1ǵ5Ž19Çâ}?«?Ø&F _ØQ¿ºÏçe=õi ïÖŽX?úʸQÂwöo¨Ç°ïÙ!raì•<Ä•cýY-WŸÓ O8—f|Y)¼`¼uc^á+é¦Sp $·è=ÉÙÃ4½ Çqó—«b#Qšï.¦ö‘ùr]Œs@Ž8¾ Z\,朆cuÆCÆ5Æï9B?Û$óö¬èÓ,±Õ6¢ÅÆÎ‰<9fâÚÆ±¶£ æ­;·D÷Ù_²ý°äØý3ѽ/E¿û_“és Ë1e{݇k?u„o,¿’"¯“Ú]ßźô‰Ös9va¬c[âùr¾Á12ãOÙOõ½g«».ïû­Þÿ7ŒýÖzÁƒ‡D×8OÙ%rg~fóÙ×qM`ŒÈ{Æw~ï 6ðíRïpc+–÷D‘%׫Ÿ'k~¸¾êGÝ«»¼_3êIå„Ï\§j+úʺÑJô3Fx”^ìís`âgaˆ›ƒpM‚ç’_ì‘ü¦ë‹8—ᜂãg®ÛÔ—£²Ð½^èùFøº^äG|sóN¶ßqb+¬»õ<ú7Q8œ)tðµ†"ÃFš>7xþ»ìÓrtãö7RñvýóùMÍ7÷}IôŽ}ã#ÿV„WîŸïIÎU÷‘°ï~W?ïÆ l[ìûÉ6]?Äù¶ü¦žkp}‰ëÙpžÇùë0cc 碜¯ŸÑÏ»rŸ£ù‹¼Í­ƒ²oà˜PV\¹•>Ñ4¸vÃ8Æñ6ã(džßy’k_ì«9cüäüˆý3ǼµåþÕ2_öA+=ºÆÓ]Áô¹6ÀvËöÆöÁ±È¡qªð†±žsföéìÛ^c„Fö5¼7ó§Ò7×F8¿^"øÂ6ȸÌ<ýwõ埔õýñp¡…Çáx7\ô¯¦¶U£ó‰-óŠû¯ÉbÏl;o³ì¥äâSÁá:‚ ŒOü]5^¸vÊÇv£n-:¨íÓͽØ×#aÅ9M:­;®>1rŒÎùÇ߈<ÞÑ|vñ­„Ø;÷Éù×õ8O®.zÃ6?@æÊ>‹íåm¹·¬àûƇç¸^ÊuƇ¦ÂÏ.ÂOæKuM·ëC¿ÚضV‰l8ödÈöÉötZóÚõ?ì“9¿âØÿ¤ètzѽ[roIéó¢Ö;7Ö&\ð>¬wý€ëwG깺zÔ¼rçõ«ÐrTßçÆ¹5Mî­Wn½ˆéçZû±óZ_Üúüá¯ApŽSZô‘yÇ~’ë(¯‹ ƒˆ^?«õÌÔ9.,àIŽÓŠéy¹:ÿ‰–Ÿ‹©Ì?¶ý_ä`ÌÙàI®3$‰†º¿GrXtïU±ûGDFOÊ}ì‹Øï÷yîÙ°Í Ð­;7® ðZNé/§¦ùåú„ãrc¶oöóŽè®GxxÈãÖvÜXƒë’“E&9„ö|rïÕäñxÇ#k.¿ÎjY»Xù§Æ#—ò;D.^yŸþ±/<,xÁ:Äñ1Ç5ä×Ýx;‹èî£ÚfÜšãZž®Kð$çÕ12×"b“GDG¸û†–+볢ìیÃÅäùt“ƒ¸z“ExxUèc{Ú©ùåÖæc/h{sû:ªu'LVÊÝØ1ž÷˜Ï/¶\X0(ƒè8×·ö‰î¡¦qMôÕ/¶ÅùcÍkl.ž3®³Ÿâˆ‡Ó{g—w`‹J[DÞý-êä‰/&çæº¼3X ߇h‹‡hKÊ󅥵Ǒ륄žr^^î+"ïô½r½¨\}¥pŽïíy¢ô+m!|oµ Ïa<Ì ãF¸ßâ«W¾½x¾„EoiyÎæoÉ»ðý‡¼¤?sŸœÃ÷Ò‚ß%eÜ2Ò–’¶´¥h1ï‚ÒoA¹^|E ù[òF[r²æQLøfZÐÖ’wa97­E§¡#„>á|6ò‚<,:Ñ/ƶžC¿8‡ÿf|ù¾8Úúœjžrã’ë…¬ûÜ,þ~bžr?øPÔâì×Ìó²ìÌôgñú}>@ÏJ.ôk=ï[N†/žÛviëA*?ˆï-ý³í Õýö¸Ð´ÐK?l|0ý[ç/¥E|PÌš*½Ã¹ÕŸÁëúZÌ’«Ñ7‹v?f<ÌKžÇü¡Ç%,ü*n—²æSÌ:‡ýÛöhë¯ÍÏ¢ýè׎RÙ›Õ}µô¹°Ý†ò÷à/ÆÁ»›_†Ï–>™øGÎËå´ðöeÙìqŒ±cØ|ùÙúbøjñt”´ñÀ 3_Û>,ý4ãÚ~ÈÖs ? YüIwÚúbÓ{±ä€ù¡Ä©ò è¯5o£ Ï’o)Ð!÷ÇZôÛq æaÛ[)ë9ÄWe,ˆ ¥OÖ|@‡?·ZÛ~ì¸|µŸ»[fú%¹ý´žOÕ¯…6Þ±è¶õÈè“m–üПñ3Ö¸¡pÄè§{ÒÚñ„ÓE-þÛ~ÎäU–þ¥Ê#¥3/Ëmeây‹ï%-\38gßïRµ ë.ò1|´äÿfòXÛN-}(hcÍ×ÐeÏËÖSØô÷ÉyiÐaë‘åO ^[úPØ’=ïPv í|ËÖcè¿mÏöüߨxiÇw¶9XúgÇSà/â à³ñ'à ìãXx–ªŽaëµ%“§à>ø;Ì϶C‹¯†ß¸ß’C*½Ä÷¶¼,\LeÏ ÓÂÛ®Ré5OÈ3UÜdÑièµðÉŽ‡ð½‰Çl†siM~nሩSÉsvÞlû[ƒ7Ð#´Ösß-~>Zzcê?Ðk[~r½„5oÛOÁ~L½Äö ×-ü´ãÜbÖüm|3õ6Ì×¶C‹»^˜* q!‹.“ß[vcòN9G>…|(>Yç…¬ë¡ü£Ésl=·õ|•s ·0Oë{;·óW#'ÈÕúÞ®'¥ª÷ÙãËuSŸ„Ü->›çmû´ðª¨Í'7lû·ðÍ®±èU2zkõ—jXò0r¶ì>Uü†^â{‹Ï&°ô®ó„jSùGعm?!ìÆŽCì:Ÿ¶Z×SÅíÖõšÖs©äa~…À7{ýÇ艭߰s|ùZòNå'-Ü0ë;à³Å§PþÌè‰%'»?ÛÿÜ·ùB?JZãÙumc WZ|oü·ô?cüe—à;âÆl=´ó/Gí¸ý–o”’ov5”½¦Â ·ŠZmªú¥e×v=ޮ˛<Ò²·¢Ÿ ?,=°×›RÅ–ŸAmò ÌrÑŸm½ðß·]·³óPõY{Þ¦è™Õ¯ØqO*'ý·ú ¥Ÿ/æe㌽XÈ¢+Õ|ÁËÞM< {D+ßW8!zfëôzÕ3å}…¬ñž€ëíx×~ïӞǃ 7”¼ gè3ôýZ8bןKXóµóír81;†Ý?hÉ5Ô:Qª÷s­ù¢ßPï3Ûö`ÛQ*ü1¾/mÓ¢µùkû±TõVÛn¬y…ÒŸTq7Î1~½·ë¥©è ‡ÆÀn,ùÛu›7í<ÒÈÓšý~„ý>cÈ:¶%Pë{vÜcôújá¬m÷özŸ-‡TöiûËÏÚëî©þß-è³Çý!ìÁ~/3Õý^ØuÒ…~Ør–y¿É/-|0òÖýö{3w³ƒóÀOàœ\Ç{§öøhCú-KŽvþlçk©ò6ôcáS*{…ó–=Ùò´õÜèÆÃý6}C½÷fë£ûvþeç³¶ÿI5o[/,9ã9È1¤ÞcžVÿ6=!ß÷ÿ,œ²ß? )gà—¥7¡üvªúQ9Ùëf©ê¢–¼ÐÚëÝv-p5äûQ!üY?—Ö®“¤ú¿«?›þPïÚþô=hƒùY8m׃ìñB½Oiâ#Œcßý-¼NUg·pàAKŽ©òû9‹6N§Êû OÿíuõTô.{<Ø?ôFî—xÇ“¼w¶_ß=þû8B=ƒë^9îgüP4 ¯öë½Ãýÿéþ×}Ù<òÞçówâç{^÷züs.aÖŠÖÞÿÏkÿlÿ oG‹­wÒ1›–û‘ë"¿»éÈ?ç`Ëúnöt;^ÿ¿²•ÿ—Çíøéó¤Ô“»ñÀÖï;c럭cwâë½|JGîew²ÉPtÚ<øOô#ÔüCáûýŒñO¾ÿ½½“ln‡÷ʯÛ6^Ý‹ ïG¦6OBáó{¼;ééÝdy'½ü§¸»¿VÞMn÷ÛÇýòòv×ïWoÿ=ÿ7´ÿ§¶ôoi …ywÂZ›¯¡|ÁýÎïNzq¿²¿®ýÓG…ÂÛ.ì¸ëßðûn÷‡Â°;ù£{Å€P~áv¸`ß*æ»|BÅ¡b×{ñ6ÿ6õoíëN|ûoÚîäøoôí¿¹w³çûñ-ÿ–Ö‹wº7Ôw÷Þîþû±ï;õõ¿àÉ¿ÑÑÿ÷oçKl› EËýÆÿ8l<»×¾ï…_wŠKm¼ÓxwšÛýÜûo1ô^ó»¿{éó~u󿑇ݯCÉçN4ÜMÞÿ-lµÇ¿›?ø76|7ûÕÿ½\»ßùÿ[yÜ«üï'º}¾“mÞ‰ÆûÑ…ûµ;w£áv´Þ çîG§ïÇŸÞ«­‡ÒÅ{‘ýÊâN5ØþtëßÄGÿƒï$÷Ûéÿtúßàë½Î#”>Ú6eë×ýâþ½ÒŠŽ{µÉƒwÿ©>ܯ]ØØ§u†P}Ýï˜w³É{™Û½ÚÑlý^íù~åx¿>å^±ùßØÛž¿ÓÆÝèù7úy¯÷ßKwº÷^è½ÆÝ¿¼ÝõÛÉ-Tìþë~jÓñ¾õ¼'7ïëÍ¿/àíÓ]>{Oê3 Ά%%& &g¾¤¡Iø˜Ø ÷ø†õ…ÝGËÇðÄaÝú$õÀ£ÁÞ='& —Ã÷éÞc(¾ ôèÞc@¢œø»öéߟ»õÔßtØ¿ëð}ÊiDbÿ~e÷dhKÚ"÷OLêÑ¿ÎÃô9hNL<3¢3‚HÂý[·®öÒ× ‚AaòPT÷ĤÄ2=‡&ºÝü|ø½ÿ¸Ë?41‰òeñóúÓÿ ýÙm¼'D¢Í¢Í¢Í¢Í¢Í¢-¢-¢ µm[ÈmÕC´¥C´eB´åB´åC´1!ÚØm\ˆ¶Bˆ¶bˆ¶Rˆ¶rˆ¶Jˆ¶jˆ¶Zˆ¶zˆö¡­¢U!ÚøPíøÛ·5ã-ºâSÞ}jcÑzqÞTú^‡²où*ó|Cë{´À è‰m_¡ì|k+-äû ‰/ú{àNî€oÿ“?;/ EG¨ûîµÿ{½~¯ýü[zþ×wãã«ßÿõ_¨ºÅ½êÉÚïÝúûoéÕ½öw¿|uý~Ÿÿ·ÿ-; }_ŠR"×»ÜJ_ÿ>II=†Ê™w(jƒþÛ•ï^k3U:¾cšÜáþµû!bñ!*àCE|¨„•ñ¡ È,_Î|*o>ŘO±æSœùTÁ|ªh>U2Ÿ*›OfŒ3FŒ#ÆŒcƈ1cĘ1bÌ1fŒ3FŒ#ÖŒkƈ5cÄš1bͱfŒX3F¬#ÖŒkƈ3cÄ™1âÌqfŒ83Fœ#ÎŒgƈ3cÄ™1*˜1*˜1*˜1*˜1*˜1*˜1*˜1*˜1*˜1*˜1*š1*š1*š1*š1*š1*š1*š1*š1*š1*š1*™1*™1*™1*™1*™1*™1*™1*™1*™1*™1*›1*›1*›1*›1*›1*›1*›1*›1*›1*›1ª˜1ª˜1ª˜1ª˜1ª˜1ª˜1ª˜1ª˜1ª˜1ª` _ùrå’?–Oþ“ü16ùc\òÇ É+&¬”ü±ròÇäÑÊ'V>y´òÉ£•O­|òhå“G+Ÿy´˜äÑb’G‹I-&y´˜äÑb’G‹I-&y´˜äÑb’G‹M-6y´ØäÑb“G‹M-6y´ØäÑb“G‹M-6y´¸äÑâ’G‹K-.y´¸äÑâ’G‹K-.y´¸äÑâ’Gs¡áoëà—<Ú[”&oᣈºJWõRÅkÝ·úXu^تUÅ\åTìwùkNÝ­Ò/Mtv½tÉ9¿óÒ¹a×+;‡ffÙ÷Ò³3]¹~î³s˜ãl›©Þüò— ÎÖË*Õ>ÓËùñP“ⳓ&;Û~ßú‹?—9ëŸøå»BW;ë:¼ôüÊÍ œuù¿(úÛ°&κiý³ÆÖòêc[Ã>s6=ž½Ã “E-Œ.”ïmg[šàÊg²yUŸì—+úgg£yçÇ6-ãì®Ñ·ÚÒý}½ëtIØù”³¯Þdzou»àðþðUî1·œƒOü±û@DwçP9ûÛõêšÒ&çæûCw¾Zú]çp“A3v8‡ÇTÚúiû\Αt Ê:þ³sdÑþyÎîüÚÓ­tŽ©,ùòŸl䯲¯Ï´ÛÍcæ½–ÍùuÂ…x9'Ow:]£b:çt_çLÏNuœ3Ù&ö]g“söüÃMÇ—êâüöîŠgŸ?ã\(pî‰Æyßr.öjòÁKu.m”P}Ð+ΟÍWæ>ìî\ _¶¨û›œ«G‚ïß®ë\?¼lÔ sÚ87ƒ'rŽy<àÜj—!sÜäüÊs¦Äœéª)oÍž¿|õ|måûò‡ÚѵQú­«'ŒŸªÂÃé´|ŠØ=æÓëv¨¨wt|Æ©¢¿õö“çæ«ô¹c‡|½´ Ê°Ä“ûÉbYUÆ ­–$]ö¨ÌñW^,‘Meyï×î³ûVÙJf*Q-íÃ*ûú‰Ëz†gS9'U\Ÿ½×1õÀ¨¬­Æu/¨ò¼RéL‹ù‹U¾ãKjfx9»*P寨¬'Û¨‚ó =Ø}÷sªpÅׯ4 |¯Šü¹nð‡EÚ«b¿Öª²{à U2ýƈÆ__R¥uÜðè×sUÙˆèG?J¬¡Ê ß”ùÁ<*fh›ï¿zïg—uÑ«ã´Svµ;÷󜚪ҲïgìWKUY~uñ¡—ÃTµ“ye?7M9¯•:y­Šo¹µüž¥T™ãjø>8¯j^yð»/6Qµ;þã…:­T-oåÙ³äõp£ó—ÿAÕûiÅú¤œýUƒÎGV·ZýªjxaH‹ê…?ULþ¦i³þT“² 2—´]5í»)oxç–ªYü²QûªnQÍs}9ÒÓ|±j~½yÛm9ú©GòO/¼ãœjùóü¡'šMU­öä¸ðÃTëß›·èÖ1RµÍ×û÷EãT»ÞYßÞ½ûAÕ~ý‘¼ýÝIut>Ùýò¢ªÓÒØ/ßûhœê’·Õ ²KQ ßê”íÉÌj¢oó£ÃÿªÖ ž\øü’WT¹¬¾Î1‹†¨¨êe>L(ÙÀ9S¤ñw‡§–uö¥]^*kú¡Îö[Ó'®mVßÙšíÙusÚÌw6¾2kWܬ<ÎÚþÞž½b¤³ê£3=Ÿ{a›óý®â›[?YßùnDî!–Ä9ßU«óÊ€üsÌù÷Ó›Eܺ5ØÜ¿öøžS‹ÿ:ãlÜ<î±Å{;[–4úèÝr­Ÿ^«áL/çl»üÚÖÕ‘Ï8;†÷~óûÏ;?ïÜy¾Wƒ °Ó=æŸs4²í­-󜣶X•´j‰sìï=£¯¶èæœÈ^zûó¥ç×G2údVçäËK«^yÓÞ#Ÿ:[Öùíå·}ÏÕ_çœíðÜúƒ›ª;ç>éòZÉIãß?‹*ýà”ŒÎ…#esœyÿ²s©ØütÍG®uþxjÄ죃K9ÞÌðSì»yœ+“³÷íü”×¹V³×þ¬á‘Î<™œðx!çfžR{÷­îÜjöÎÍoÃ/+Ïž©‰«2VÞâíV¥t]ù^~òÑâUk¨@î׌m™W×R#Cù"*búïM}'ZEÍ(ѦlôM½vzÁå¯Túb#GŸYZe8”žáã±*S“É•Fü~Se>åë5ãèA•uñWÎû?QÙ¿|áð¥•óêê0Ï æ*÷SÏ¿:rø•¯Ý¤G²Wª5ßoPç¹ÇT¡À™ü+gTU…oäÿüç໪XÁ\Šôí«JŒ]²çû9_©ÒÑW¦\n0V•Ùq}Fë™U¹ÓO|vii3aEë?;SqOúlÏ”ªbñƒ[<¿BU.³-ñ…ñùTÕó[Ÿ[­ª¨}qÙn¥ =öfἯªøýo=»%K´ªY¥Q¾’ÛëªZsWMxsJ?U'wÙW®¿´EÕ}gabñïw«z±o”*ùÄLUë{ë÷·Ú®©Z¶aÚ6ê‘ls~,û§_5^y°|Ã'TÓ oµ\[îÕ,Íó~(X@5;’çÕ¤­¯ªæËæÎ­³ajñjÔüùÅ£Uˤq£Æïn¬ZuyjVÏ!3Të%^oXf–j3(0)®ã›ªíœñ'®æk§ÚÝüaCÃ"+U‡ ™³/iñ“êT¯Vïšê1Õ¥vÖš“®ôQS&çÍ[²TÕù)E²¾¼WÅü0*8µÙÇ*]Þ}wê¦s.Ý…Ý?Dtv¾ÿ\ZV8;s-Ì?pýgÎO¿\{*ò»:Ζ¿£ÇÔ½älú¸Æ¯?}°ÎÙàÔ9_úø·CÏ;?=;åáñÍö:ÛèZï»3ÎÏ_ŽØu¹õ»Î®éc×-Ë:ÔÙóÌÄÆýäq~yªÑŠz.tö—ú¸øÌOs;Þ,µxüš®ÎÁ—6,È|x®s(ê«åEW<·ÿ_{9‡Æ-m»òìçÐþþ(~ês¸wÁócžÛêñþ™?Íî Αe[——Y×Ò9úd÷ÊEë5sŽ _Ózãã[œãÊDœ{³¡sâËò¼Ôµ»s2Ó͹úžrN½ØmÑÐß 8¿Uú`AÙ…:gÓí{áåíÓóùv·x(ûjç÷Oú?÷Qâ@çÂ¥ºo×ùp­s©u›«ƒ ÎÞØs"¶‰syj0jZÎÕ¾«fœªºÖ¹ÞûÄêaóœ¿&ÿõdà¡‚Îß›L=’¦µò|œwEdž^åõ½øgÚ7^P¾‘¹ç/ôlSÈkži’F¿¯ÝöëÇÚ«ˆ·r\ K©¢^zô×a»âUôû§~}+r£Jw¨ä7ïþCex3_¯¿3¯P·‡¿Ó<ç •¹ÀÞŽa—¢U–箵ì_ú’ÊÙ¬bùO¿TÙß l½˜¨röØøL§þcÔ6-ì–¨òtÝ5ñíIU¾E¯YsíkU ûÜ]^RŸ¬^ûz›ôªpúàãU|sU‘µ›®´ÍrB[f¨‡v¬>¶xd‚Š/¸ãù†ã¾Q5úwš=å”R5W×ý½ÄâoTíRŸ ßûù1UgÆéºÍæÕÃ48õ~ÝLªÞœ7Þœ‘-‹j›îålÃÕpÃ…EÍk…©G®|chƒlªI®¡ÙŠäJRM{¾VjùëKU³ZSN,[ý’j^$ªc§¹iT‹èÏê¤YP-n–=þGÂxÕòZ÷1™ŸžªZû·|YcÕLÕ¦à”oWw8­Ú¶ÎÔ»ëŠ$Õî/{Ö-ø‚êà_öj•Ä(ÕñQO\RÒuÕécW/ûAuIÈö~RÇu*¡ßkOUz²ššÐ¤Ð”ªIyU«Ÿg´?ù|AUæ¹néê/™§"Î ¾o1áúË…¾ëy¹‚³÷“ÚÕ´½ålÏåécjFgóçöî[¿ÈÙÐjgîýG:kæÄM6i–óú%ÒVXé|ïëø•¯ÈgågWß9í¬³rtÁ© ÌuV~œ½èûÚ9ß}?÷£µƒš8?|¹uô®*ל5+æ]ÛÂÙ0¬^Û½K9›‹e¾ìû:‡³õ‹WÆ?¾m¦³Í[ïà–šõœí}&Z7ø¦³ã˯ԩœÞùùä¼Ý¶svÞœøAõªQήýÊ_úîig÷œ­ó×—+êìydñÛ¹;,töfÎýؾ;½W/~ó[Íëξl›êtîyÆÙŸ«È×^¯âhõìÄ_np~Ú±E¹aŸ8‡žÙú\Ý€s¸y† c_Ïè©øö‰ q£ùÞ>9¥XMçX¡ØGÆÔrŽWõþ0ñLœs¢ÿ©÷süÊùuQîż4À9•7fÌþÅÑÎéwöÛÔÞ93xKo篋ÎÙÍg›<öõOÎï•6KìÚйÐ-GÓ¿Ö:çeJz§l6çÒõ÷2vÙ]ÍùsÀ„Š+p.ÿÙ¶›'mgçêìKOl{ä çúb%?™ö¥óWïÏm~i·ó÷Ôs™žñ¬ò$½Z'æÆMöÍ´Bs¿X¦|ùOú|ë9åŸ÷Ù®-ýª`£,ߟ<³UEäˆÙ|#ú˜ŠÊýù’̨èØ!Û–;©Ò=Ÿô|Lp±ÊðB§*Ágj¨Œ¿ß,S´È1•ùñ ¾Í®«¬ÎOVM8¨²WõþãÁTΤ÷&UÏÓV=pyzú_:OTy¿¾äþŸSù/Ô=û˅ƪàÇž‚í6\W…ßöuíðÔ UtÉé©þ×N«á}š´üc™*5©°óâG•©w«ñö—‹¨r3¯x÷DŸQ1•êýmr{~í‡^˜¥*løfkûÙ‡U¥ Ÿí¸|‘ªòû˜ß2ýð°ª^mBúÈr”óA«‘3 ÎUñ3Ÿ¸¾XÕ¸–mÊok'¨ZÝëv^ž5‡ª}°d±«k^Pu{Ÿ‰îÓø9UÏóëð¥9¾Rõß=ä䘘E5lvêlÚ¿cÔ#Ái©\M5þþ‘%³SMkD”Ï÷y9Õ¬ÀïöIWB5÷Ïz#Ë=TóSÛ'ý´j±í›ªm›ìT-¿›Ñçýi>ÕjÙÏ_$~SHµ^Ùí£‰-U›}«w¦5Rí²öú­Áè/TûaÙ¾¸úêtÕ᯵3rJ¯:-  DMV].ü¤ÞüÎêÅ…õ‡žÿ`™êôüº?þŒ|F•_½eÃÔAO«è§®Í^X+蜩{þ‡LÙžqd96±ç¢ÎϹžÜž¿„óSÎÏúw|o´³ùý9-û‘ÓÙ86î=ïa¯³¾Ð†7â†tvÖ®ÞôP•ãΚÄñ•§œÎꬉéÛ$Ç€5Ÿ¹þl†_u‘ê™lY&;ž>ÿi»g{;?ž)½¢ÍšÚÎOÞì›ÇL^ìl/²²ÝÈ;?篷rp§<ÎÎ+SZ-{äQg÷ŽáûÒÙûíéý•úôröµ|ãõü'ª8ûuTèéŠ3›¼žt‘übtõU7«áì¼é²/SçàæÝ« ?ZÎ9¤eŸ=ú=çÐwÓe+±Á9Ü*zýµ•ý#¾4WƒUqެþ¼Y©÷[;G?]Tëô…Mαå%ú_«2Þ9~\eù}vç×j3Â.ÍZåœ\ÕXù×oúAž§UðÉ]«s>tVEtø2>ü{¥¢ZÚuá(Ý¿Jì§…/¨tó †O}y­Ê0ÆùìQO•ñóe/”}èq•Ùûm‰ƒ¼*Kï4µ+ýþ½Êzò¥ZÅ[QÙŸÕ¼Ú~²3•PcîLêOûªÎœ¥òÄŒ,zth¢Ê7&é©q]Tþ?ì\Pç[U0¡UÇ)?¿® ý:<¼PØ3ªÈ›?=ðQÅÆ}Ù£@x~UbÚ¸¼­.¯JͶhà°gT™ž9Ýê“E•{)o›"å««˜ˆ÷?¹ôØ;ýç¹_¾·_Uh4q×·¯$©J¹;)Öý‚ª’¡á|_XU­ÔÁ•_ÏŸ§:uFÒ•ëJþtõ£ó¦«UOîbSTÍWòÛU\ÕºvæJÉÂÃTþU®ŸÔ_Õ=ݲôÙT½aCú¬Éôµj^µã®Â'UÃyCÓöNl¦i²êÐ+Å–«&·ræXU5íÚ¹ÊðŸr«fu{¶|½i3Õ¼lç_]ê®ZäËR¾N¯YªeUö‡3U«<ŸŠ_Bþ°Ü³®Ø~Pµiùcñb6©¶“~ïµcsÕî—6å>ëZuP/Gÿa‰êøEÓ—ÚT¾ª:—ŸVí¡ª‘ªË;»/óÖT SJ-hûdWõÜ[£æþýG%Õ*aidÝŽ‹Téc+ʤë©Âg/ŽoWå‚sò¯«+Ï´¯æìiº¡áê9“œmqÝK÷Ù¶×ù±Ú…“ŽeqÖçˆîú\ѪÎê7û>´åhEçû#·~›zðc绲/½ûÉú·œG?8tîì#Ί· N«ó°³bS¡¥ó¼7•'/7ñ`çûŸO.z2?=?èj‡_Q™c»ÖOp6MûÅéZg¯³åÞ̔Íù)ís{Ƥq¶uú Îæ:Û_Ûžûíˆä“ê/=ÑùyÙ‚+¦}à으røÂýÙœ] ê_¦¾³ë|­¤Gs|êìþpÛüw®|èìyaùø] ݽ³—]ÝüÖ^ç—™Å^Ø{¼‰³oëûÓýs®s è‹gÕÞ‚ÎÁ±Ús¾ùÅ¥ã:•ÊçîvàÇU«–:Gê‹^0q³sôáÉùÿ®¼Ø9Öq÷g–tŽOú®ô“ùK:'6¬’uÙ7ÎÉ*O~òÓˆÓΩÕÅ4m|ÖùíÙìÁ™¾ïœ97&rÛo…œócæ-Ù/ÚùýXØÄ 9Þt.ælïù=KCv=±/¾¡óÇþ[ÓÙì\îôf‡w^ÿʹrëñ¢S^Mt®­®VlçÇõœŸoŒ[ÙµºssGÕçb³)OúiÅß9±S…5ÝöT•£”÷£#³jôz[ùKo |z¹ l~´öwïUá¯|u¹Û¨oTäÓ¿gR9«¨4³~ry¯*íÙ!o˜VYe¨Ø®ï_ƒ*ãÔ…n¿IeÎ=ëï ÙT– 'ç¿•6½Êöíè#?–Ë­rüþ|ö‘;«ú|ô‚z¸–Ê[jBļW«ü³Ž¿<´C>U°u–B¯~­ W=‘þˆç5Uô‘¯Öý®Ÿ*>5ת_ë.U¥Ò4=[,ü–zð[¿§iõJª\‹þÍ–'öWåo¸|øÛUìÞ9‡'äUú¿¹e篪Ô3êÃoÆ«*Ï4{¼âº'TµõQ{òÍX9åK·ÎQt±ŠïÒgØ7Ú¨K{~½QÕ*¼.îÈ[«Uí9…¿¾üW=U·\©% öVRo®¸²Î';Uý19·ø¦•jóæ[ϧU.??e͘¡ªñª©Ù øQ5m4"í¨JϪf•»”Hw½¼j^¬ÀÛ UT-rùúß*ž Zf°9jälÕ*Ý·í·¶XªZg)¶±Sð}զȌο¥ÚÖ¿z)Ý–[ªÝ„Æç§ïȨڟìóÎ|ß(ÕqÈØ‚c/6RK¬¾öç*!jN:oÉjòƤµ_”ì­:^÷®ÚñDYUn]±§Nn¯¢N­þiÚô ÎéÇÎ~”Xx¡³ï‰&>±ÊÙ‘«ÝäY#f:[Û hõÉä[T}¸s!ÇÙРXϸìÕÉÏ5^zé“òΚ^ýßXÞþogu÷QéGŽÜ䬮Vûh™ß¢ÕßÏPvdgm¦¬Uò\wÖ?÷˾ƭ&:›N•~ãbÎ7œ­ãf·±~¢³íí.Mç—Nïìx>×û§»½êìlþΑeÏ<åì.œ±ôø·š:{®N¸Ñ¢!ç—qOæ>ÐÆÙ·öÍkjvrö¿½ã™¾ÎŸ?íGãžvÔ:ž)zà‡Î9ñ;׿æ¬Ý{D¹L{Cd¼hãjçÐ{…‡ýúþ:çp£‡Ë˜šÏ9éï[>¢£säì[壺­sŽþQø'ßžÏã™JߥÀFçD÷ê“7í˜åüºó£Ê“‚»S£ÏŸ>¦žó[ãð°/ gwζ½pcƳœó¯þT|׿2Î…4qc½‘è\X¢ø®}IÎ¥Ë ZüþšÏùsþÌå׺Os®ŒáãÑëœk#kz¦¯|ȹ1½D‹Ë::7Û<õ¡ð”'&âÚ±U/©°gûì‹Ê{eK¿ßë•ÿùÒ{7ÿö· ÖêW~ÄÍ'TDžÏ¥ž¤¢²®½Ùì‰E*ºì„Ç×ÅgSézçl>îï·T†Nþþól•qZ¿áщŽÊttÖ•o×]UYêå‰ó}þ’ʺvKÓ¼×_UÙÏÏœO×sæøãæ©á¨\—«µÏYí€ÊãͶªTÛÊ*_Ý…c7úUþmŸºøÇ`UИT¤e~UhÅÑ)Ñ·zª"ývú?½tS«wþúê"1ªD»%[¬ßªJM¿4·w“ªLÖ|YŠ÷¿©Ê%.˜þqlCU~ûOE§œ:£bz—jQ,Ÿª^`SùIiÿ?2Î2¨ª€]Û€ˆHIKJ—t§ÏCwwH Š‚Š*"6(a" ÒÝÝÝH#Ýß>?Þ÷œ™ïךÙ{Ïšµ÷¬kÝ×=³çF…ú/QŸ‡«Q)k•ë™rªÔý£KÅc´®¶ŠAш×r]L Pƒüˆk²× jº›$ü{ŽZU‚AÁψQG%ò…F` êf;Ÿ¿;þõµ •çy AïIámOÑèÚÒéë7Þ¡‰TµpÆ~o4](á :Ÿ‰æ.Bzâ7ÑÂ@­$ —-•Ó„X´òÑJúçCÊh-k}ùÚ¨Ý?;MÈÃãÇðØ(7ÚÝ¥ˆºçôíK¼uû=Бqý÷o}tºâs=MÐ/¼UÞ^G—sz×"ˆÌÐuôÛNR>º¥7‡EbøX}Vi€9ZÏ´üþ(‰(ªÉžFï%‚d–+íÅå0áS_t¡@:—ÌsβB³iåN.Ã{¨-Ù8Gµ•=šÉR§j ìI²˜rïÜûÆr/ŠœÊž?« ®CR©E?•Í §j¬|E&úPtHÝÖŒŠ«^¾ ŒÑ‡2•½W |æ POu÷A-Ùûð¢¦X‚¯ú^_„&Öo;£é ÐlQËÞ:í-—,?-t¿‚ÖÛZ¿ÎÜ“€¶KUw5— ¡²½6’¡}±äûo±³Ðñ¤Í¤³=:µ#¼%C¡ëH¤véµFè>š” UÐÃ+âž'ë ½F5ßÛäB ïMñ]3õèŸ)Ñê´| ƒÅ‘W¯;ÀÐSžŒB“*u€ßñ•PQG½ñíý3<ð'õå·W06[Xxs”&¬[FËU+ar`>D®ùL¿qù¸ãâ³4äeBl0~V1Šî&,xNó4†ÅØ ˆûKS–ì“`Åô2qïI!X-ï±Öâe†uŠZã^jØd5ãß)ï…D¯‡Ód`‡)\¬*öl¤Xx>8#ÑÍH“†’¨Ñÿ¬ßÕÇ} r—â¨q¿ù`ÅkOa£iE‡,{²?šÛN5åô¢…iujƸ2Z|HЧ G+CY‹ÞNU´6çrÏK8…6NýFtgÑö\ð°½…ÚEvöhT¢ý!ÓÐáŸüò¼Ä9trÎ{huÑK% G¢Ëë!‘oSÑ þ–¹‘ðãƒé·WR˜ñø)µgéÝ(ùl÷“€­;4‹"¹¹Ê “™3/~È´@Ï߇T+]he“ÒV¬g„Æ£¦×I'Ê v§EÅì©'Á'Ù¼L.CÅ\KwÚ_R(w‰Š a2ïóã®ûàL›ë²Ž1”5; võ;BÛôýãB'¡*²ê€ ÔŽ‹]|pf—,²þP”@ í®^Û07´.Ï>õÖăW`|«yûå³t˜|Cãc9r¦ÏƪUÔYÃì¥ëõœ`î{½|ŒU',¼={’æ;,nÑS~9Ëa|´«åý°*5ÃztEÖIwÞ²ÀÆîù[&\‘ðsß_†‹4tÓw6aob[Â’äëÐ +uE’,á†Èm$5 ð\ιŒdûvãÚò™ðÀÀB“ðg<Øu.BÝv)—:kT‘FŒ6î”ï¤Åk†º .ˆÓeêö.Ò—…•¿IŸCFaI¾Æ WdJ¢qfJ؇‡ñä ]†²,ÝË‘¼¢l5K?Ö»9‘£9ÓÝ!tÙ´l2rÿx-­öy9îÞÏiIF¾—–Ìk;z(p¬ç¾Ëè=byrJKE8%åSΣ¨mSÌ Qo/гð›@)Y¹Ey'”ŽßŒ³1@Y¾­´ùDw”+Î.út;®o*‹¹¼E%÷øW¾¢Êù÷ò®.é¨öMŒfsS‘µTHcmÕÛw å>J ¦dTˆ71jŸnVï{‚:ä=…Ùq2íræê3ñQ°PQ¢Á÷§+ö·UÑÈîb峸E4¡«÷k¥EÓÞi."6;4wX°þ£õ-Œ¤,½ZÑR›ŒêÒ3f´Òê8˜•ي֦ΥûäÐÆ3>ú '!ï®ìË]rG»l½í§Gh¿•z2kÖM_–?"NB§_¸Q{D9¡dÐ×AzÆn”Öìˆi8¢„2“AO§BÙPî‡UCè.*Äi®4]9Jß¶“®·¬£ÊÚ³ÆQ™x z¨$šKüŠÊËuPCéÒˆÄGsÔüT;,šxµ…YG—½‹Q§à ÍÑÅÔói·q F{¯kx;ÑpySж© ;.Z:¡i5›ãEV4÷°¨)´ð’“rlÑ@Ë3ïG+»ÑꊇBV83Z?jZ=W)‹6É;ǸyѶªvÝêMÚ-Ð-Do^BþY‡)-/t<DZ¡QƒN­‡‚‰†¥ÐÙqL\óIºRšÝ2ñÍC7å+ùÝñøà`íu‘rtê ôŒö»â¯¾‰ DÑà_.²¾¦|0>7º§¾ÝÛÅüÕÐÂF¶ÕÄwê?•˜§dH@Mòiqõ½{P9Nõ'æV”·Ùn]<Úe6õÂWâd ô´w¾n´”oQ:݇Ò!&Ë0­h(ç+¢¹!Õ•3d—k¡fTÌX°H¦NTô„]„æð8Ü#­ÎÁ~· ¼b;{5:J^Q<¯$ôijåŠ÷/¼‚îF»‘¤;O¡gí+eô&œxÉÿz¢¨C( †ÉK(ú h·©ìþA_}6)Å ôÇ2ŒÖqÀ@LÒãðe"Ì¿jÄñ»†ŒU­: Ã^ç( ôŠaÄý+Ûõ[%–?¿®~ºÖ0~d­èïc˜È“L>hð¦J_î ˆ‚™(Ê;\WàoMµ¦u7Á;ùx]ØooÃâ-¼âËÔÿê·Ö2a%WÑÿø­*X{˜ìŸUwâöï\ï„­[~1ùØÞ0yú’T ö|Ç}zBÔ˜\ÒÑ¡9I‚¤R!)EîZ¯·±ÆÈýHc Ü¢y¶Þ~0·B>¿¢ø¡Â(phž²•=©ÝmQx(Bƒ¸•Ei¿Þ;~ÀÅýE¯\ hB)Ò"¡"ÅÓ(}š—¡ª(e&ª<xr!¿X¯©£‚”îà·Þ˨D¡OŸù¹ Uènÿ©’ƒjF•yÁ¿j»Þ’¢ú‡Á#E#×QcçËfís¨åùê*‹?jw”6ŠE]»ó«ÞwQoì_yÜÊ/b•C#©eþ”< 4^jÎZ-ÊBÓJzžå² 4·É>ñæš>ÁG§â%ž¡¥©ÃbË?´²UX]r9ƒÖ§ÎRË<ª@›ˆãoªÌ y˜žR¥Büífîx›8ù¡ƒtäÉ£ÿ,Ñ1Bùñ?ãýè´pøÔÑ™töÖø°w{]&RH8ô¡›ìÊÒ`tÛœcÏ ëÁ;!—GŒn¢5 9£kÛqÚn³é¥ÿ†û~_Nhh>Ô/†„C»üñ†rÐx€K=÷(Á'›Ÿ“U'øçwÙ°… (¹Aéüg|Šdâdùúƒ!ßǾÙ÷?äÚ¶dˆ&Bΰ8‡”‘ärìÛÜß}¡5Î/ 0£ø‘蘔6Ó8ÏC9™ÉÆV;kéÃׯ¹¡ælNéAûd¨;qTqÁ0XTúA“œEVìU>h扺ªyšçx¤FBK|Žz—d=´ªgd}΀ÖÖ5¢h³"B\å\Lƒvmžº7Y n1p)|€Î£âÔ×C—&MGxšt‡ 枣ž½îV~¿'>ï4<þN³ý>3Q0ø4hqñÄ e¼~s—6†ËoEÌRÃȬ^vþp0ü9Qg'Ÿ c÷ýRËÃ`‚þÏNÇG˜,ÿ;œ[XÓ&üJÁ‡aöÍN»Ëi˜«¿r"ò&,²Štݯ_÷A†ówÙ^% “ÞÀ˜½¢ö”Î>Ú~YK†w|‰ýãdè»ÉÈe¶á+Í<¦æÂ¡ÛÈÇöˆåÒåë>E½BµfQHFR ^íŠDtl,ðk£+=cűs(1ü2¤/¿¥Ÿ,3}ˆ2OÒï½öå,¥ 7}PAiõ]Ö*iµ¨¾ï’A• ÑŠäs¨Vþc1\fQ¿>N}‚êê–ó/£Pó¤ÉosÑZÔša¼¼w uîLgû^ÓA=þ„¹K"¨¿Urh¿- ¶%ÿa-N@㊓²É:,hZëâK­Z€æç ߇‡£ÅÕø ¢´|””xãN Z¥D³?—[Bëú°{Ü„>j³Æ)¡÷ íøŽP›Ž« ½1Ù¸ØQtˆ:Im{‡÷mG âqýžß̩йñUÚ'72t ¡¤U¨E·¤;Òì?Yð¾ðZÅþÏè¤÷Cî⤊Æ;ÿ¨ÏA2ö£ž6Ib0&8qÆÔŒ :+E׌¡™µÿ§OðÔñ(ð™ßŒj+n™³ß "}ïCeö,”e÷²Kû|&äߎŸ}”œu}&Á=%6m OÔ@É\S¹èç:({þô½Š:T‘ÐXѹUÿˆÁ Ö˜¯øÙÚÁUMÄ×~Ø’>óà×rl_H§vvç›>œX­F¢j;ÿÔÁ¿H¢Ô]ýP÷Õî >fw ÷ß|¢¦†Nˆç}̈sãÛ"ep 'õÓp¤Nß“UסCÚ}¹Ã§Òé‘N5TæC]Ò‡ë»ÛÝëE†I«}'÷vÉëÁe’‰dÞ²¹5Ùƒ,ŸÒeroG¶w—_ ¯‘Ão5vßÁ<’8OfGºˆÜ¾S÷›ˆî!OWvÎXcòH ÖlR ÿðèˬ’·(˜›ûóÆ)þ0llß™ŠGk"ClF PœWú뇠l”L鉺¸ŒÒò†‘^b$(ó…«Døã”S˜úüðR7Ê? —¶7EÅ´Åz#ûP9Ý3es1ªN•ß¼¾åŠ`Ñvï ª_(¼nÂÝ€µeÝÍ>½¨%Ó²í=ò µ?]¥ý¬@‚º‚ÁJvvoP/³-åô~G4°›ºß=ú(üC‹zž¢qk1s¼;š¦©ðOœ…æa~–™©haQDE}Õ-í¿vJÉU£•Ï‹122´¾bGÂf6©lVoòpÐýAí(Ús–4w²@¯8 t,îqô‰ñÀãGç¼M}ÑùýÓ½ýetåwÒö±âE7ës^u‡Ñó¶Fß-b¼]Y·N»ò ­ò–ä'Π £TRá)?$9•}ÚÈAFR ‰§-¡µcä¡`UÔ«Ž:Ù€JûQÃv“PÆ=,&Å Çž@Ák“+šFÈ-*J.ûq ²[õ¯Î°@öñ 6cȾ¿éÅ$& ¹¤¡µÇl!¿â^F­Æ(&Šeúºi ¥_ îæ} ¨°Ùpíóªo®é÷ ¹RüÄ–4O êŽ;©Ö¨7@ƒÐû[^4Ac¯'É~hJl©?( ÍfÔ·Ÿ¹Bó¤x>‡`´\ ï^¤ª€Vš/:™$„ïñ÷ÉÝh«ò«x|šÐ¿IŒºq~„Ž|©©ÁÅýÐùÏÁæšÿt+G~l ñ…ž´t債Tè“nku¡2€þñƒ*‚Bk0øsr€‘ì- MJùG¤’ÂñL¨^†e‰Ë:üI,½ñ6ÒÆÅ´VF`¢Wþ&ÛB(Lù¹Ì—ì3‰ÌÕ¤ƒð·Ñ¸åë|,ÐK-ý‹¦¯´ê‡îÃRB¿Ì-!iXޡ⣰ü5ûöwzXgºÏáO NÞMª­nÂcãžl¯«žÉå÷‡=­ˆ}‚üÓH4þ~Ê¿ÄIL Ïíà¾.ZcþGqhï³ýœxÀúZ„«ôOÿé*Ôoܺ¥~Ï Ù¿ëK»‚Æ™ŽŒËfhZÿþ.Ù~O4¿"~½G»-¢*&¦X åO‘v­«hÕSJta´m¨ä8Z™+ÑVë×yÃhwÍÏPªísÂ9•\Ñao°¨4“Ð ­fˆÆ…áñü_æNè¢mªD.©…®³íOZ‰{Э·i"8"#õyÙjExÑ1sšÔÒµ E>PDMPûyuÍ„*w^3ðõ¶AyèÒ‰A(MpU]H mñ,…A/(ö³óeúöŠ þ1tŠw(¶Ó’A©¼/c¹Ö7(~ÁE Uý¢{N'§¡®óöHå…fhºÐ+îNd -Ò' [®]„Ö‚Yæ¯ÐîÎ=™²’ £Äþ¥—i¡‹G¬ðË|tóÔY¨Ø@÷ÏItÔ=ݤÁœ[µ´vþ0TGï.áx†X–¶ß¾‚Q²œ|éd øã­®_º0¶{Q(ðð,LÔ9ýýWN SµÆÎ|Ö0KTìÔü§æ¹oPŠÁB\í½$CXøª¿ßþ4,76 ¾ß„Õ%o¥õ:XÞ t9v6³ÂVñg3l“Ö=Y&Š…]ÿœþ™óvHyãÁÞœoÊ­:”‰ûîí;!0‚ûu²FþþkÂ|UD4 ð ÷ ªŽGHI6þÔ/H}yíºèÏ?x¨6Ó¡ƒ‘ŽS–ãæ‚Ò{‹JÊ4#C•ç²Li2i×ÞÚ|`ÌuM–ÈrCož8‚ Ù X~EÿèFõçʆóÏñÈ…ú$ê³Ï‘Ûà©2ëw5äÉli¹jªŽ|^¦Gi"bâ®(s.Í œ_eCã‰Ï]Dù¨±ôvATtz³¿›M•mtþX/U¡êÝ(e+›xìo¿òì’!ªcB·ëÔx5@þà¡jS]ûmy µýk•y¸}Pg5쪞ÐêÝýz%š„ Ęì^G+£áèò@ÇÆ0gô(†P iÜ ¡^6471º~p-¬c‚gÈ£¥qÞÃ;¼hue£Áëë[´N¸ÿ¶ùÓÚ4v¥òfF¡%‹…7?Ú›©Ñ9É C‚®×!Vutüw•¿}¨{UX} x§sÿˆS9Õ"ººS…§³ú Û%3"ò#mè®Áæ½sKo3ðR¶¤¡˱ ìÁÁÈÿûÁæ‡"$fÿ'W9ÌÃ,<wMí¡%È'î½)Ôú} '€ò!égg}È¡d”½ =h? ¬Š{&yî~.'R!‡Uø&çzd™êõ+¦Ü‡ß]ïÏš™ÂïsMÒ!ÛBˆq’“r»,zOÇUzbƒ éâ(1?DÎ^eå…É*ÿ>BÅ<éG¨z²À*ž 5Ú±í9bP;þ.扲#Ô_&dX¿ uÂÞÖÆ@ãnñ“gõ#Ðô8±8ššN7­®æ$÷³!4Ðr*ÛÚ9Z-]‚÷â¯C›ŒT¶†o´+Yî(K@‡ßýù='Þ[¾n0 º9öUßÀ¯—;s`úç#w©/^›¹4›]Ý`žS9ciuÊ«3 õßÀ­õKZïHX>9 Pm+]Ê.«°v*õ5¿ lHv?‡-¶Ì—§ŽLÀ¶8ʾÜ€Ý3D%–?N!Q@Ë6‘ý·ûü-Þ5Á}Ž—êw,;tÇ#÷lô’U¹M(6"yŽÿRþ¦ Rô˜©÷è@j¶ÕÑì%xÈ_y6¬â-Ò‘}H]àM@ú»ÎåïöÑ´n‚BM™ÖßK_E–Ϥ+2-]éšsx 4¡Û /ërGîø Ñ…þäu´ò»pùUjñõ êEÑ– ¥£ð])[îáJ%f¿\…âég³.êA)S²ßQzX$èRàs”}>]¯%ðåý{h+0¢âÙkÓ#I¨üàeíãÕã¨ÚêòíŠÊ.‚^ÿ]¾«‚¨~¹]îBèaÔ˜°ðx¥(€Z&Ì‘¯¡ãéC³Ï>¢nS©üÁî¨ÿIfýjÞk4Œ!wMø`ƒÆ‰&kPަãïw)ÐüÎ'™*)´HöÏ5Žq@ËÖ/_£5ÅÒ˜iT2ÚŽý¸OÑö‰3í r ´ë,QÖ¹ªƒ\E²‘OŸ£ãY‡©…kèTѲ_éÅ:Ë’«/E—l’üÓ»¡è&QgT3~ Ýi9â:ݨñž·Éù'ÉÍè(È×&݆BIyå‚:ôH2^Rq”à¥Ã1£ê³¾g -qKñ•94È2“1i…jΪXV¨H–!c2Ø„2³1Þ£‚îPr{ôø÷?¦P¬êBÄDtŠüÍ4vÔ¡Èû^e‚³>Sé¥\(õ“&lP”Îì.•=¢]«9µ Ïò§nBcÊ×i?oädoMA+ë¸ThûáÙùë¾9tp&< ®NÃÃü“Ð¥úåÓò†%tý>+žkÛ]m“ÁœÁmЕﯪ¶]¥²+òÐÍþ³«÷ÇkèΘóö±†ž[ò÷Ù¼„¡7` QŒIúÒE^4Í€ŒÝ'3Õ30DB]®´•CÛ)3ˆŒaDÒÑ€¢Ðï{QÕÔñÂW´·Ø fŸY¡>k“iÜ4`†=ßçÉé+ð÷ÛåÆ XÉàé™@Xü¾Ÿâž6,;r„.ÊDÁªÐš û cXgxD››øÊ¹úüü»OEûöz ì¬bñŸC"£¤DËí+HœýÈõƒk%2$ŽìÞ}LüO¤Šù—6G¾»²Ø÷}Ñ{lœMTMYƒV®#ªý QKÕ+íPFn!¥qçyäy63M[W¼[T ñ+uÈg6‘þ] jßÌ}ˆÂG£=Ȇ¦ñ(hŸ0x#Šb7aõæ”X*]ȼ*Rµo.Ǔ䡌Œv³û”MeºäM‰òÚ’­ Û¨Hª}ã´¥*­ IlhBU—¥}ÿ¢ðؽ±é¦7gPúæ1¯¸s¨á½®¨÷35uöÈŠkQ[“ÝýE=ê”Ý/PØPB=ûpÃÜÝ4 ÒàsÒAÃüÕ¿‡¥ÑøùÀS{‡t4ˆ<ÚPÖ‹æŒ÷‹EÐÂîÌDÕ“.´žú_ô¡h¥ÍòqªÁ[Ã^…{A¹äçgO”¼OPƒ*±èaÙ{„ëxûYµgjéßåñþhƒ:íZפÛíPïêÃr`?4œÃŒº,X¨ã¡Iýðs ÝThÚ¾ÃâFðÎæ>cñó-¾Ð’)ó˜7»ZÓ/ÑFË„¶±‚{±êÐÁB«R£¹w¬4tÿ@×øÐy}“Lèy,Ró8ú´ÌŸšœ²’hí†`ȨAwó} ß¶s}ùóŒô{LD0²Â׿×_\œ`|Y^·Å_˜øRï]w€¦òl¿¼´¨‚™–Å´•Æ}0ÇLúÄó,¨mñWüHÅ›µ;qI#°ÔÈ%yº}VÊIöÔNÃj~:s±j¬_ý¬ékø6“| ͸ ÿ<4·;`çYpè Õ $Ú·æÁ| ;ßå`bŽE’¾^³šP$ šißk@25m­óÿóÿpAƒ÷ù£œH®ÛtËNHåwGüzµ1Òd²vKCÚ8agð~#(‰ -Z£ð ™:»ûo" óù˜\¯ï®ªš¨m çy•úÅTäÖ} IIðÖ-ò¹“È×Kz¤â ´h…ÛȤ¡0™3i¶ú=¢8‡½­ZVJnö~5)dEé{žç7ýPV±²®Pµå)Ó#˜>¢"‰åÍÄ*³±?Oô}†ªŽòVU§úÿ=qÕ>É\rD¤Våü¼fÔ’#•…4Ôîýœ+Îm…ºñô7^¨„ þ_ßÌù^4 Ìâ<ñŒÀcŒFg$M>š¶-XùÍŸT¼ot- `)Ì-×ÄZAkUÞQ[hsï…¶âIG´í¿F\P¾†ö2Œfêåµèp{Ë ¯Ì{Z…âJáqå샑Kªèœì$”›€®â¿5ëë ÑÍ[¼QÌhݵ}¬åïbDxP™­ð txîü‘ÎLRϘÞ,DbãGëŒi†0Wb¸Ì­”O8ž&Üï2¥)'¶  2úé±(·¬Þ$Ë:¥œÃNÇc¡ø´÷µó¼’®~ÆCU¿}‚"ö}œB}AP¬ÇÌNi{J_yPJ™AE§hÚê(¨)OÜÍp—‚†Ŷ9hÞOÆ^é-—è¤rÚ3 u/¹¢ˆg Úí}ÿþı?É ÓräFL¦9tæ-qpxÇBgí÷âú`è|)GûÕèŒù©ÓNCŽ\5M‚Ðå=<ùNRº…<Öä‡¡ç€ kï ô*×½²})¢7qy€½å]ùUûSëFÃv›mèŠ`äû½¦N‡ø£ž‘82Ž0NîÉÅ“¤°’/$Ó Ù—Lø`öÓ¶ÆHÁ;}•8Ïœ‚Åc.%‡rŽÀÒŸ>Þ’£°ò#ÃzW¡ÖÞ[ÑlSÀFÅ÷ + JøGgôÄdç ì\}l¹}‰¨ )Š‘8x"7™ñ+î£ì9ü¥¤I+s?_YG²¤}Ò ÉHþé;LßÜFŠRO}Ë‘5¤Úc$¹‡¬eïæ,ñ"íPÌ‚ªb.Ò /Ô,æ#CDDKÑ)dÚg7Þ.‰ÌQ¹§„óEæ‚÷çòȺt×Ã|ÆÙÛü{rµñÈÈÉ+0€Ü¤CÑ_©Þ#ÿ~âD# äm¹R±ô ùíÅ Ÿj¡ •AH\÷[ú;{ÎyìŠlðØh]æ@1Im½#·QâyÏÕªæ(S:§rŽÀÝj§ý.o Êžü>åúåV¯g5´¡Bêß‘·Qé¹ýFJ'ª”ЦT„ˆá1¹7Ú3‹•G_E Ú“$ܨw㦭´#jÓúFҮʠNôsbÁªpÔqeûóÏõÛÎ̳/= øèÊ«}ò¡h8Ҭ脦þiâ%çCÐ\»ÿUЉ.´pÔ²Id…–—?ßÕhC«„œÄ®í6´nO‘ó˧BÛÃÖy×8yxâN $¢}ÆUÚvÇUt¤&ñXV”D§syÙ?.ÝÀã½—€#ð"ºXßbˆ|6‡®=ç)õ?­ ÛoZeã23tÐÍÀ[Aö¹NÞ%`-%.Öw‰Až‹:qgÎfÂÑNtèp ô¾¸Ê/VS!®†MÀx­/·Âq˜¬<²º·q ¦ÛÊøX þÒÌ˜Š·òüÍÑpéK°°¨åqtR–Ý„©Ü–Ÿ›'åÝb†UºJ†I+#X« šwf_8Ӳ׆`ë•n±õ§ØÎ[RÿRyöxOjÊR}C¢±O†>³HâªÞü­F ÷­P€o"îÿ!ôv]›DqYæ×áÁ÷ÎT¶aÃHYqíéé@BOÜØ¬Ä×jH+›:f^Ñt ÷ÓüR\‘Á&g2Oƒ™¬rÇY†ãáÇBÓ[í‘"˜•q_!rÔ1¹ñ¹úÕ]t/l"Ïk–†Ç¾®ÈwòŒÒÎâ _ù7qû' ݹ—n‘`ˆ"Sü=YS™(vGså€;!m'LT¸JQZ:ð@B™îuªs(—d–:ž. /„Èüå@¥¤x×¼á:T¦ò=§ÅÇŒ®ºý6ÈGœkoUB ›C#š×Qsm·r:༟ÜÝE§e¾ZC/ÏžCý>gèë;Èhéq¹^7¼aèæ+·|Ý70ÜúíÙèD7Œº·¿2ñ1Qö©ür˜à®­ú+ÂS¿Â¨/ÕÁLþ4ŸIò?˜»Eññ…5,T+Þ`œÞ€¥s¿¬-JaEù¢ÑP‡)¬‰ÜŒ>ïd&}³Â°õâþ¡W¶÷‚Þî#†½7ó?—dh‘XEãˆÞ a$©ãy×ôÔIïú P¹d>oƒŽ9M!¹Mu·­R„§?;µËŽT¥}.7œðËÅ#iµH[¢ßÃâté¶¾l”OGŸ×%1‡”q$"¥?8™ýd~E+@*2¾ÖçuÈZ 9’Ï‘ý{ï;väl9uO*) ¹ÊiCÿ~F‹#¬ªÕÝÈ›2¥{qá1ò‹¬¦Ÿ¹Ùˆ·è5ÖÓPè§NpÆ´0ŠüÜÕ?fE¢†Wô"\QBi‘¬çÖy”rŸõìèDéìSC¯ßÍ¢,ÁUgûc(—XË­jôl^\|Pò•T¦¢ ÃPÅmÃ#b»ÕŠ×_üMC´ ÝØ–cG R‘gù[QÓ7›øÓÛ)ÔÈM‹|‰:'âµ™_¶£î–â õú&ê'¿Ìé{¯€†~k¬VÍhl7Áóš‰M]z¤2Ù8Ñ\=1I2V -œYKo%:N¿sæS 8Z¥±GÓ<‡Ö Ù×ÕõãÑV•¾,ÅŒ‡Ôâ˜L ýxÖUEÇtÔòðÒD§dçÛœ…ýèÌlœnßÍ….[ÏŸµÿ‹n”O«39Ñ­s¡}þ0ºéëq»õCWy yWþ¡EöùEäJÐx±[^¡ÂW9¡ë·Æ‰‚f¯ÆêŸþ/ÿÙ™ùW9‰’Í^’­ðûÜ›½›.ð‹×L¤þ7¤_ìÙž…t£FºpËÿ3™Ÿå™B&uŠ™~Zd5>ñÚûûòBÚ{} Š˜˜BmßLAñåKσæÿ»wóîËùâ¢9סÂ`0Pï4Tʰf¬ÛþwªšŠ/nêÿ»GU3z úř¬î÷âìw­CÎ\ªù²òß½«ÿðþß=»Õ Òìÿå½SÉÐqSͺÚ~_}Ó=´Þ‡ì²¡Ÿy+ᥞÙ÷ªþ³K5íªàqþHYm¾Hw€±éɾ¯o0Ñ<¡îŸS1Ñªß `ö@[­F)ÌÙfRÞ ~ î9UYÂÏ`1á÷×é×°´Üg\]+^½t‡NÀƒ~¬é_;XxÝ“6ø6 #d®ªÀ¿±Ó{]–`*¼U"ÑÙÒ \=5$N´º{§ ÷ß–N켎ûù)voŠ ÙÎ*I~æ(ÜïÈñ†Þ)9.­Úò»#µjwF<)%zæ±DzéDßïèJéòYý×zj‘Cä×q#+-ä÷qª¿ò¿{RÛjgû9Ï#ÿDâ…àŽ(Dšé>µK…"Î"äŸU¡èœøÓŠ”H_ú+xó,Ju÷¸Ÿ¡E™«¡µ¿}hQNs¦ù÷b*ˆîrÛ.Ù¡’Tñ5˜ÔEµzC“PirÄs§¨}róQC¸cœ¬J5³7&SDQû´Ý7Å4ÔUúɘæ“úü}Öâ¶Îh¨tøåä¥ ‚ŸŠ¼æ~.ЦikS¢ÛhžôgAÍR -fÒuÁñZ©JgEª¡õ›°è? *hK¶åûæ×Ú]ÎûC£áö¿š ù…N¿*ÅÅ 3×]zÉÄet‰&Ú£»b÷ÿs˜ºW´0‚wSNóë¸ ¡}tÛ”Ê!䲤YÖw‚ݲÐ'é¡×š·û4SŒ™T7^‡êW¾Ãhò üáã3( m~¾¢ù ŠCGU¥«ßBáè³Ã.†üD ûCƒ ObüëÎCýò³—†g!ß<‘ý@ä&ýTÊ%†±ð¼CC„žéÇÓK´wª=ÏKµºYA½ ßåKh2ºË=Oà#-í‡~§´Š¾^½Êø?{Q ¥´ëÐ~û¸õ¯qè°¼§RÍ=ïi‹²#§¡ã‰ç?ÎÍ$è0Èð*r–†ÎðœªÐaœ´›r>:òb¸]àFöľ¬Q=7ä¬Òz~ƒÐ£¸Š>õ$¿úˆíÔ9³c&¼úõµUwxí¢~ei}iø'4|Ñ0;X„Æ·Å’ì¾Û£i¤X1ÃÓShnñ~skÂ-|I“kTVÑò¡,ͪp2Zåü²¡g½Öka$›ˆ¶ èÎ#@ÈÃÝ;3÷Ð~lõï·,/tÔ °˜ßߎN‰C÷|ZØÐ™IøJ6ÿ!^íŸÑ pÊuY¥ðÿäaw©ï' #qüµø²-ÎŒOöÃ#…ŒáJ.Âf]K7™ftr{J#Wµ;±ºoس¡üêF8ËAñáÍ©ÓNÿ­ìÐ îßC}¬lûÚWÈö´s‚Œ·›–ʲ®N-ðXê'¤$[½¤…´´Qƒó^w!=Ê[Äí‘#üòÙè”ìÚ…,ÿŸo¯Bîgê¥ÔßסÐÄ2¹¡Xò¶É·ÛžPbÖ¾t\õ”zzÓDd©ìzvy÷”¿²›ÕX¥‚Šëý‘AÖ/ ½Ñë²Çú/¨2˜O§­é„jÉc‘9O †‰’¹?÷ÔúÙÎHÛOA=чYß™?ÐÓóŒg—šþÎ ëÜ€f»‹ ¼söÐry÷•Q ´6Ýù°£Ñ í4ÏðóI8g•øöf€Àmœ!Ùtô ™Êúg¯Ò ¡0PrÖ†~êúK:]…µg Q)0ZP»Ÿá];Œù]+¬=&\« ±0¥³fþEßf|õ*K¢„àïOS_£Ì9X ¢ -¹s Dj;ßYÀRœc¶æûxX¡yÊTÛW«¿ Ìçi„`ýÒÅÏ÷HÕ`ÓNƒ£³“þß]bLOÒÆa=QHÄÂùQýG0Wßr{ˆû²·ÇOf iySå÷ëH«Ú“<ˆäq3æ™ó‹HñmBÙ¸Þ ©ª?.nò}ÅCìoçå¹ÿ!íÔè†Àß·HÿXñpô@22†ìWOü©„Ìó¿ /ŸBV NÑűÛÈþsb5=ïIŒÏHó BîùÊC Û;È›Ýÿº·ùŸÊDèÙ­¢`âÎáºMÞä×½óÜEïE¦˜‰ŸE k·JŒ‰(õãóf™¶*Ê8VAò½ý('Lýåà9*оK®äA%£÷R¥¨â°îžý½Õ:V´üzß!>¹üU.>5tK¢#¦‰PsìÞŒB[>jغ탺7Iä/üôò/¶‘Ihø”ç©Û9#4®°RÐh^ЪA*õ’¬­ÓEHó½äB½ÒFxËYÈ÷#]ב¯#Èpìu•w/2æ³úM‰Bf³7¦»Ã­xx±éT¤^²¦}{ðÔá ²Çë¸}øû 9 8ß5(^B®e–åɪ}^£\äKC_5QÁܓ攇*éœ[lÉžG•“s—ßg£Ú×9ÁoˆŠ²W²y y¸2½qǯ5]=èÖ yXã*ùÙ u.wTµtA= Éñm¤¨?>7”âË„†…:ËE¾ßÐøÇEq&ƒT4ýuÃâù/4?‘`\!ŠAóhŸ åKŸŸÚ½|hUjv˜îÑ´Þ6¬-ÿ<„¶Ê;/ø³Ðîq‘¯×=Q‡²n‚L+è¨A®T~>y›Ã*¹7åW`»óÉÂt÷EøC“$÷ÉŸ v¸ßã¼<‡ÄBðöá#HRxµ^dmŽgcÌâ˜;\õÑFB?R*ßjÿêÔšö7øºã^»‹2ÖGÌ.m¨ÿ, 2¤ ¯eå!“ŽšÎ¿2d÷ ß@V=Í¥SG+=µÊŠ(¹T9#¸EÒ‘—¯‹bžÊù_=yÑ̘†‚Ns&ö Š(¬Z–Âü]EõÕ¿OœDñh>¢®è)”ž®l6'FéÕ–Ë?ˆø= á>òªt¹ño·P‘V&ȉÄ•ÖtäTtÓPeÉ—óÕ]T5¦ñ{È~ ÕF’¹«"æ›X¢æÉlMšN<ÂÈy=h×µŸ©ê~ãEÝ:ûýEwõÃGc—ÐpV2´÷ô{4žf]Õž@3¯u«§Ždh>½yëeØa´ä—>f_½‚V'¥/©’¡uï ËÂ~´e ûñÐ7 ín–§O†ýË¡¡îGñìjtÌáùsü7:s{Âr] 'þгøßf}áæêZÅüuÕ¸v´Ë;|¶ÿ î+ sU…?Aì&Êd ð}dùõñ¼6h“c°>¾ÖJX×®e.cPýà=Q—/T>Èù1ïGà…ùâ (;OÛ";â %ZäÇ›ü^ÁçÒrR røl5ô7öŽ4n2~‹€ÒÏQ ‹¹òPa¼,(ó"ªÓé®e«½‡úüA?a/5hNz@±©Ý m¤Câ£YÁðíj[ôÉs‚Ð>1OoD¤ ªw•9 ¾é²Q[­Þsº…Îþ™ñÛ…n7mi¹Jè–Òpù­]-ÄA¯$Ûs¤ÛÍI©"èJ\9ðÐë!t±ÛZ1¹B÷›?ʉÐã+s›WzrÚ.6†ï}[×k¡ÿRÒ#—Ù@T¹’¹CÃÃnñ®\Æ,0²,êhAÚcç²+aB MvoaL‘Ý7YÑ;3ûé©;£²`6¤ï7Ÿ_$üTÎ^nȾ‹û¼~žJ…¥·&?htßÁÊq)Û¡aMëá·©œ,Øp®ì.+R‡­œÃ´*O/Âï*]UÊ÷àïô¥P…z7$šé¶}ögI®§Ò–ÒP"©¨B‘úH ’ýz©ƒöÔHÑÒ[-\ˆTí´Ö«SiH³«4Eôbél¿?µ0eE†ó¾nž/èñ‹öA±¢TÜ'΢àöT ™3ÜzYªö"+š%~øøÙ& ÂN¥ö GúiŸJ-eäŠÎ¦<5‡<ùÙÜ»¯°-Œw8oÙãøƒ¬®ýrG¡µ¼ºÿ!Š<6uo#A1;k®¸/›¸_{=‹öPJ:y1ä,” ôó×'úEî¢,½àºƒ],Êå«àHjBù‰æÛëQñÆ­o~C´¨lÒèd—ÿ æ>gí¸†ªî?ÞGÒ£Ú‹Qk—²÷‡Ž'øð§Q†»!‡=ÈÉ[°ŸàÃowé$P' ø”É{!Ô;LrF‹ È9V›¡áT¶wÔÆ34(ÌkŸ BÓ‘o£-wˆÑürbÊã›/Ð"¸ï²ÝI ´|‘lõ4üZ5Í|gÜómHöyßw@[m“Úc´‹ ´(mAû‰‹íÏ.¢&ÉÜSDÇSQ×cЙi„ʯ$]ÂÏ«ÐÏ8<ÁЮMú_>ì7>ü¹¿†?qDóŽøJ¥|"äþ[ÐÙ©t 6}U¹§˜ ûKrYó‡h úŸkÕ69c ù¢Pîø[¤-JNY,ܱsƒ¢þ£Y’®PPÇ~…Ìaò“%g8% O]ós~¹)ä¾´ÑÿŹ4Ÿ4ï-AžEܦü%‚7£§þŠ< Ã,doá3qÀ§K—¡ô¬â«Êä(—zõg$Ñ*4eUÏd»@¥ü%«¢f¨RrŒ”´Xêƒú½:qfPÃD~~—žjæjÙmˆî@mMÈ ÅØ<Ô¥IB½cî¹£‘ÐPÿD#¯”àw“+$}QÐÜô±­-í<´¤eg(,8AÛi·å÷‘ðíz›ÄÑÚÐ^ñjûb¬)t¬LkSB—ÏÇn‡ è!:û£ô>a~^°á{¢òúvZl­Ì&`ÐXlkEE†{¢}œß=…Që õ™Å¿0Öÿ‰›æÖW˜ðÏà·þSG»>s¹3ÂMn$¬¼™@®ü"ÌŸ¾-1) ¿*Ì·²oùÃbÒaAIº2XZ´{Òim +Çþði—?ƒÕ-ÃÏ2fõ°þBPOÉä>lž9 +ø¨ ¶E§â”à·7ë"×Kø;Ú×2úunø³l"‰£RÔ©{HJå¸)|]Ézûî0‡õ"E+mѯL2¤êh¨P廀4“‡=Þ_F: ÝÈpúÔÏÉóUȸ£yù1Á£û:Û‡ ’#ËZå™rvdWqá纜•&âAú'žÆQäÍäwtQy#ª‚´Z¥¬ö±(4aX0æpEf‰£¿kGq9Žb‘”(×î7ð~ŠÒÑjÊŽê(ëb c°úå9l„4þB…e çlêpTú>óéÍÅ:TiM}ònLUï±S/ªÍê˘Ëv"v>o>Ƀšñ¿Â¥ÓñˆZ&UÇþ1Ôa³3âéD=ºÌ‡#ƒÐ€çtº ¨¡Ñ¥@ócLhrC§Ë…»ÍJâˆ,‡¢%‰Nï[ ´T=w~r[­®e†M|zƒÖµžÄÙ±hË×-~×víîTš¦ýË¡^^×ïKFèøÎàC¯äItæ°É;bœ.‘ÊçœÅLþ7‡YÉ ¹&9RT¢SƈvšÞôíQQ tµÄ~^~?O&r‚ïbG¯‹A!·½?Ƹu-ŽF&GÖ šJ&J9*éù£f⨠\[šFÀó”f7¾½ý Їúð)&pu§˜Lˆ…[­óc¹há³¾­r¢”TݤoÎ2‚òA±. Ú€jmÒº ~=¨w|úÌx=šµÕ­ì’8¡µŠåK«|s8²¬@à"&Ч²:yìkjÙ¡KýVŠ„4!/ö_„nÁqÏÊXèÚ>м{3 º"ř٢ø K¥eøò—`è:Wë¿p ºž7û¶Ì ݺ%DKºÐüÃõl_ú€ä[ÎA¢h¼r`$Ù¹yƒo¨^X‚M’ºÙt•OÐÍVÚ•¹HðHäµ.q¨ª|úᜠ|ÙÌФH†âò¸á+Ë¡HëLJåU(84Ÿ0p´ò¹k”Ï)¼†Üâ2]JnÈe“Ê)v5†œ¾%‰‚íÚªIo@þõ‰ “FýP0ô2©¶H>I'mÓ]€’cÁ ÔðÅÖ3sÅ$0ºxžù ñ¾¾'VÍ 22Áß…vªB*LõÜ¡Ú1a ÒÃjd|Ï~ š‚Ú=VËÞÖ¡¶GÁïó¤Ôe2ÿÍI†úÕ÷s—Ÿ9A£[šJ‹Å,4M /ˆ­>¯¥ìû<ž‡Ak_Yp¿=´ý>¸<"ùÚU•E2´¨¡ãt¤íæ°tß{Ûò9ºOš~Šë)^=ŠÒžëЧÏ<ó8"9˜öa&Ž’AÂÜ62Ûùäÿ~B~• ñ4€ b}¾Ã “e¬¤É0ý”åþ6ëGøÝÞ’~׿*ØävoÁ/–…KU5aQ§ê«uÍXŠïÚ}&-+”ùÂÄò÷a5ÑèBëÅ8X?´ÍJã-  Â}u³Y°Uõyxãèìô^ú-Z" ³—uA"ÿ³’¾‹ÙHüwÅå³=îÉH;‰dWEwånu"Å™äŸÔ÷#êôa2Ùzo¤¹LžAøisò¿þÒGá×35]Ș«û<âý î 8”ȤŽ,a†7/"‰u‘kðûHö:Ùv#ôz–›Ì5ž„–l߂̓P§dÓ{8“’°þ?‹0Æ>ƒŠ¨Hõ‰×ᡇFá d9$¨Ìè"|ÝñøR;|ÒѼ4‚Pô´ñlLÁ/¹!ebj/7ŸJsÊ=ï›> UÃ;¬Ž~CP¿Ç¨Âç–04Í-&\q‚Ö^ÒÚúðMí×õÛ˜ =Yc-õž=tì>ó¿d×]¢<‘×Í¡kU¿‚¢ƒºù¯Æ¿í…®zîÒ”¿½Ðeiú"tûtΩi’XZŽàâáu‘àÅóoÍÐ5çbœ¹Ýùµ»blLÐS2ÉÑpå|'Z Ïë)‚¾Ûûªü†Õ=TgaÈøþ•`Ÿ¿0âØô@Ûó:ŒVÿ’}m ã_G¦¸ä~Îë4rm˜>aX|ÔR ~|U±%M†ù£Ü7Œkc`AñŽçèóAXÉ/¸‡ú°Üžî¢å «éQæ#?"`½´R›MÁNÚÇœXïøëÅsZ*‰ò"D{b³‘DÛ “ÌI÷LºTíÝZA²g×bˆø‘⬿ÍQ:¤:š|û·”9ÒÜàÖ eµGÚ®†ƒ fÝÈ ÷´øõÍKÈaB{ éã­¨ùdZN\%BVÞ"%~×`dk¼ÝÎm9ŒQÝúÂûŒ‘+è¶Ýäˆò¤dÖ~Ž8Ž|Ù;êÑ"P@à|à`©9 žKÐŽœßA¡ƤŽ*9~çåþ[Ê(Æå<Ðoˆâ¿uÅvÄ£$Ù@£qè6J«5lÈNM¡Ìû Rº¬÷('÷˶*Þå³Yíæ„QÑÕhÙ_¢²~̦¢µ!´QÞM!¿ª^ÃÑ z1¨1rA¡!‘N½Ô:„5†hédÓLQËxLÂËÓšàÑãû{y4ÆÏßûlêÖ×ÞäeFí‹ÇØ£‘Ò틇鯣‰’V™){:š©\% ½‡æÑOuÄ÷Í EZŽúD¸+Z~¦ŠSXŽB«AngúŸ•hCû²œ¶¸m ¯´oŸôF»'ä¼>Ç™Ð~l®0ì@*:¨Uïe%}‚Ž©óÉÑ\ ȩ̀j­.aUxVñ CiýÿòasîSšÏ¨¢qJJlÍU†>A.°JM¿ç8B]¡Ëk`þ¡¢'4ìUž¯gÚ€*™kÁ4Tìð¹•=硘k¸Õý ¾õ‰:ZÀ ²Ê+òíË!/Ñ%𺉠äªúšõšÍBÎRöû/¨!ÇLqÿ¾×³›NÑ}òºå×›½E @Ð=ãÚ¹^(:ºvÇ]ò,!×Fži·‡2«'‘“ç¡\þAàe6¨ òû;ÝsßOG6,AeÏ¡jƒÃO ê•ŒÉ½†T¨¾,úI—Êj4ù)žìA-=×AŸdR¨d7Z߸õü_c]z¡!`}'¾A šˆ¢hoë~†æž€¯¬J å‡“AÐh$´Ý|^|zõ|Ë«5Ÿ¯…öï‡Üg/Ÿ„Në›AœŒÎC¯ˆ~€ž”À…WÏÁ÷§Ž×Ûˆô ¿}õs1%×dR§—r0r'wjÔÌ F7ß(ìq¬…ñà®´Ÿ®>ÿù_Œ¦óC^÷øèºç|ƒ™-[ÿ”Ž"˜ãz?{+Ì~ž ÿѳL Cq‹¿7ma鈵ÇirZXÎu/}IàVÝXÈëñYXkû61ÓW¹¼÷n:Äi¹û'^<Â=ÆÓêÚíHÆrù–éÅ6¤ ’=Í”›Yb—|‘†èÛô©4e¤=—J4è†ô5­uÍÈxƨõ‘áÜ'zs>ä²h©X«E¶“‰L-®#G{ž£ ÷ÃkÄqã‘/|¶ìsŠ8 x.%=îrB!•ãã7©QD:”¶‚2ÅÜéejb£pÿúîÞ#…Å(ÕÍVrëÔC”¥™9Ùúr åzƨ=÷ö¢BI4ƒ%*½i ©aB•‡kw1ªUõÊw-=PíO—L09jì=Ê*Ïü5Ë.<¾d‰G¢šÅ½nS Îƒ¦«’…¨÷̲A3“O«Ø• {ÛÐhÍಆYšJî±”k¾Žæ?ßF6 …>K§Æ¸Z^˜¸Gt²­ž¿ñj*šGëq*Ãái´•=ûú2[*Ú×j ;¶üË¡ÖC–ëèøæ¹PeèOtfQ¬{U,.(Tuôþ7‡Y.ÂêâR©ïÄòwÚv®º.‚ü¾n2ˤT°óâº¡çš ô|ȱ=_;Ÿ—På¿tJžÝ—I&“Éâ¡)êÅ,«a ´*Õ'’Ežo|7=ikC¡ýúÜÙ&½è“Hæþqºè‡l®æC×@QZìíè&b4›œþ]1ÔÙwD ‹ÙGþ·[t&¿M‹œ„Î'ãëN ' sÒQ«57º‚NÅN‚nË'cYQÁÐcÓè¯ä½Q+qÌ$²ÐG!K"àRýÅFI‘1U0X±û·„zF’ÊZ`ÔÛÓ—«ˆÆÙÅËgrÏÁĬvûÖ©|˜&:•˜?,š:— EanÖJ“Îòü*6/ôh| ‹ãS‰•‡c`ùÁ”Û²ú;XuŒep!'†uÏ/·Rƒ,a³~Žúò»ç°ã´ÎKü˜þʯ¸ú¸ä!QìùG«'cd?¯ßÛ *ÜS½ª%ø9’]IgÒ´–D %³SÙ+ªHÅs†åv«8Ò¨Fxî3{ˆ´ÑšT m…ÈÀüb¦rö2^ùÐsçp2MG¼”æàGfïÃd¦{ý‘•Î# [ÉÙJO®ndŽ#Gè#9ënäòŸÞ}üy¢ïuPÙ9"_FdŒ¹~( °žlüEŒ‚Î¥1óÚ(T‘ÓЮ"FvÍ(úçxéèzïY äD‰¾¨7œvœ(MíXÞ_‹2×òé‹3/¢k×ó¸Ø ”ÝËó<Ô -å‹.z¡²Î]Á—IxÐÁ/{œŸU½ûRº…Ž¢Ú½þ~˵WˆD1·äÿ|ØíYÎì|µt åcÜ >œ.º©VŠ:©½ÁA5—PÏR¨<Œ N$JéünG#§éºDw4q»¯M“÷ÍŽ7<™šœ@óב±+ǨÑ"'QHR`-kPõî+:´šüÈG¼möyžnó8¶&÷½.¶’|XOÚºûÀá~‰=a2èpØn;¥Z S¼^ŒH›¢3ýåc ©ÌŠºÕN8TÚîû6÷_>lðUYù‚Ú~mÛMÅhÖoÌàñ£9]ÄšÈGaýÛy£²~ è¬û½çÜ{¨7Œ±2¯…Ê$'»Ì PûH²S>_ýâ²ß î/‘ü‚|OgؽþÏï„-N?ƒœO çb‹àýV!QfÓcxïY±åÉ 9›ŽéO!O.àV~œ|8} H‚Ç ƒeNôꟀϯA¿—ì ”šçîµó·&̉^ÆK¦ üÌâø~k¨Ðc·¨º·•¢yªF´ŠPù«ªÌ@t ª>öP•…çAõ­yîü¨Ñ#ÙµûãÕu½¡ÎxNcßêÓŒ¾œÒH€F.bÉî¸1hZš"›ƒ¯Û{­Ó õ/ Eÿurø&i{ƒ'6ÚÆn PxCG¾Å!¿dè’âñXa˜õÕ74ÒÌ„ù/Rnó™-° —¡g~ª çÛ¿úË|Ò×Y`eظ—>ï¬y›_X| Öw™H™ÒNÁfÞXù`l¿Ò«úƒð{æ@Öqh‡ÝÈ'å‘X÷š°| ’üÊZjÈEÒü/ÎoFò¸ ­#/ò!O8/çu¤~ôL¸¼¢÷~éS x‚^Ë¢”ÊbyÌÿ„M1!S…‰úŸhdþfEá½Â‹¬;NJ§<®!‡ÃÄjiã%ä&ͶnðBÞ­€©‹õ—Q€×`¯YÔùµm8d†Â­Æ E=q(FF5Ë'ˆûŸ8 —÷™¢”ï½Û[Õ(Sãµ+¼-„r…â7]¤›QáùÇ/…¬Q)¤ÃrcAUÎñíÜ÷BUfÇÞ¶b&T§Ðeä“MA ×Óì䈚Nû¢DñȧÂÒ·ñ¨S}ïáµ=k¨7ñÚªºÈ Y{Šr¢±©Õù"}@ÓÀã ‰/íÐÜ‚}œ¢ÅY¾D"s´ x¿ŽVEÜðà#´^²óºA‰¶s'Ïml ]h=„ôä¿â†ÒÕ¢Qt|ÙܯòI÷ [Ú.»£Køù®ôNíÿÍa–zû¥ãRGw?þF›VÒâŸÎÈ÷ý@ÇM¾%ØÙ£“ܵðº·Ã×®ŒnBóvs•DÐ(Ôî1\f÷;•A­‡VCy’×ãŸÎP¶÷çGAá(¾N|Øèãeøt‡©ÄÎý/6WÇøÃÇ®‚öè¹l(ÄÄÝS™ðièÐ×gÑð…ëñÀåA¨b›þ@YHu„@^N M‡í”|žAË‚ e­I0|#Ù{,1’Úí¢ûשý £º°¼Þÿt®ÿ•’Mö®ñ 1)MèÎ$ØÖ®›É£Ä— ³;±òB-tZ²‘ò]]„NÕÐG&^¶Ð­fúø t )¤MèC×’K­<ý6tÿz?ð<0zE—Ò|¤Òáû#7ñÂ{>Я¥ÅVcƒ*Wï/¥ÌÀðƒQo¯,-å #o˜„±/Kò¤90ñ°w×¼F¦žŸbè…™ÙKÞaï`ÎÿYëPüÒ ÚXñ‚Å4Ž€]• X>"|*¥eVéZû½a§ÌM“ç=lú·¹ Ò8Á›ÒÃóïáÏzƒåHäW$º¡·Á(w IhWôMÞÙâž—¡_ö™#™ÑßÌùte¤ ʰ ?”£2™{¼‰z‹$ã~)ÒZõ·ZÄ"ýÂiɲ÷éÈèØç¶ÂØLí´ µ:—ùØ]÷ÂkÈò—¿%ŒÚÙr Ù:H§‘ã¶.ç½ 7äºlþæŒÔ.ò<<6lÏg|¯EhΡno š2$Í…£PŽÙ¶,uŠ(ä sz èð™džAZÏ•ô<ï ïÎB©žŸ‚ÓõQF3ël†,Êî”Ì{F¯¢üâq»Ç¨¨Í±Oð¹**9ñ”aõÆ eÛ•»P´úe6J×36Ê´Q<¤hkžD3}(í¢Bç$žÙ¢ýèé¤ë?ÊÐáÐÜýw—ÐñÙò[I Cg:3Ñí¹t! }ñGU Юܷ ÿ/Öªö÷Wc E¾ZƒW0šiþýQY®‡77½=¹.ÂúÁ’߆RÙÐÉÏÅiðAꢂpèQ TRiì7>F¥µ–Dª!ð©Ýó½eu|¼úøú}ózÈâNïf¹q9îp@ŽJÆÄïáÈ>Õ¦kAÙªŸäšÂïÿÇ‹‹¦k¹amä eŸ ÿíJM9|Ìs×Ó;åŸ>Ïï[ƒ“ö’Ð'êP¦×ðôX |yFSâþ,ÊÓìÔÖ5A…¯DŽñPT´—Ù$ •Û!*LìPUR$×¹…àÅÙ‚9ú/„y‘wênôY¨õ£œ+mƒºÊÔ»²…KРäe(=GM{­ÍX‹8á+yl«ðIbh%šŠûA# m/¦ÞØi©Ã·²‡_oC÷²ð¯ï’ÐI´_¢š°O™å”kB¯ÔLŠqNô­ð?Ù8¯£Z`Øþ;_¾£)Œ²Dì,ƒ±'Ä"A’’0!{EYÕ…&Ç+ô')aºÜ¹ëÑ•øÑ6’Á¶A óûûˆO=…_AÅÏ^tÈÂbŒ…ÎFa ,“tÉ+ß׃•“°÷T°ú7 0|„ÖC(ã ÿjÁ¦„o=]Š lÓäe6>«‡ß˜²®(÷þ^æì"NB¢Y©ò;Hé`K烙{oÜ·®Gr9ËPi¤äyô¹êRsŸ=Î5z÷ê1_;]üéÊ:òêÛ˜¡ÃVòa€82_àŽ¸¬…Ì—{Œ×ú ë’s7B|}œí‘¨Â äº9‘¾—ÏyOqЉ.•!Kê ZKA|àÏÈ~ì) _b¸—ó:EŸ“¾<—‹û¥?©«ÿC)Ö­çáQæ2½bF†ÊÅ\X43­@…àÞÊuÉTòðí?Ëá„*ú†—¿éh£*¹×bšPª3,RÐõDg þl¼DÍmý—ѶñÈìO=šN)Ôù‹)Õæ™¨/¹5¤W“€†gª/ŸDã7Î'e5ÒÐô{y+Ù E4¿áœ*n‚÷÷Ëc‹–/\’9[UЪñ…e^‘,Zï¾zʨB„¶;ÂÇ{MÑ.¬awkwç_ÕOØù»£ãóÍ'Y3¾è̸orÿB.½ù’í„æÿæ0‹ç…T †4uz»{­¢Í-ãÌ®O‘·&QU¥ ¶O0¬¨Ï2B·ÆjÃ!ÏJhæßh¨Ê”‡šs:·ÉG4¡rÏÑZ‰Ï½PÎ?|ø;U ”âóxbŸKËXŸœ_ƒ¢Öèöϧ™ÿ“Wc.Å<ƒ‚uÉèàGðQ­z»1×ЦÇÕⲡìð­¿Ì¿¡2ˆdØæ5Ôn_’^`€Æqê;[ÂÐ’îïBýÚ&-÷½íŠfü_©¡ãöø•½ Ð9|á:M ÁkïâN}ìý]_´K\övB—ÜTD@`)tÆUZŒÊ:A'YЩlê2Â\)#ª¾ âƒJ_/þ‚ÎJb™þe芔x[vºC¢6´z>-duÛ Àw–'f&YòÐWÚ7¼2Y ûÄah5øù æ0òÕðÕR6Œ‘ð·[‡ iOÉtáH˜R}©ÍQfî²1ˆu)ÂÜÞòÆÇçÊàgý’á]èj|VMQ–vŽ»›¸ ÂJEr„=LÀÚ×Ò#£…4°y  _× ¶¿]Ùìúê >§èUÎ ‘›ð1é¿§‘x¹žu©k÷ñYIù‹dB6:±¿5‘¼eã׺ؤ|¹©Â܃ԅ>u|‹HË,û.»ý*ÒÝÛx@ƒŒZ»ó§f$©äá±K•ÈlÜÔ«î…,?»¸4‡!Û ÕlçYä¸òwÏÇæyä:÷ÉL¬Cyͽž6E¾çÅzµn£éWñWf'PP›Ô7m¶ …’ć´jP„ÛhçÃÙ`­ð¸õEØÅ#ÐòÌg+”"v-X²G©W·­åîo¡ µ…KúÊŽ¼TÑz (vã-mü{TTÚ}ÖN†Ê £»:xðä.íy+AT½Ö–¿Ð7Œj—Ú ›\?%L~u b’ÞÞÚA½7¼£g…Ñàù’徚4Ê,š'|ÂhòyŸYðôK4ût!ÿyç 4¯U]3C‹æ‹?‹ŒdѲ?ã¶õ´Zÿñ‚Îx mxÿ~°ùú m­Þ§;¿”B»§Fßî©n¡ýÈKºe=ct8x»'ÿH*:&ùdûZ4 3­Ä/G•bty0_"7Aàm^©_æ¿|X¹{»ª]æÓãöY izn{L/²?kàè°^„µTº¬®[6ÐqMUN|Uj{$~¸7Ÿ"øGhû'5 ”RïõfMí‚OÒRô\©¡`(å_òsÈ'SW>hã ¹l§3⟼÷à¯i¼«aNé~ï(GŸ?{· ÙBrǦâm!g†÷QÀäÈûYÔHàuè…ÄÇ@{Ç ÔS†G ØpÕ£ø­(s¡AÊ{(;ÏJœ”_v¤Ì-vú ¼æ…¨iÎU¨˜Oi=• ’ñ>æ&Á‹Aº«)|„y±ýÊ«ú$¨¾,¨Ð0á 5¯Dc|_CíØšîoýÿ̽gï©ËA£hi‘Ù9ChfQz¼*rZöí_Or€6nßöÈù-ø&§˜öE& ÚƒË7_,„Ž•É-ò}‰Ðõ¦.ÝÃNzüdäRÀ÷gbý2ß ÿ·D_?+ EÕg~Jƒ‘—{m®Š©Â[ÌTkd0ŒgfÍ=O&I‹§·Šn ´Ðhà½êIø!éG5‘å sgÝ-YùàgOóʆÆ3XÄÖ¼òÈç°”ùPyy[V„L5ÍV¾ÃjôQ‹“¤U°N·+íìÀ iÔå¬nL°åÍó˜,`'–Qƒ\XþJX“7R QøìÃòH"PkìJîY´ßy¥[d ã¥~ ‘"‡ˆ…Æ#©2R®s#M‹åÉøŒwHwØ&ý¾©"2„:}À­ŽLòs#3¿™Õñ‰ß—fdOO‰J»Œìç Å%u!ççŽNéyäiËQä¢Fþ[sM A(¨H!69Ò‡ÂLšé>—PTm×¥ ^ÅXmŠù|P²Ð,Ÿ#eø³ÏÉ:´ œOÁ ¯¨p~;I_¡•,8Oô @Ñ+óœ¼dxh3ˆòrØ1TçŒÿý÷^;j Ùø*ÔbW´dënEmÖ›ÝGÿ–¡®ÊÖ†gn¨Ùälcû0§…ö”¿G¢3Né‡ÈÑLŽ/Ü˚͟(6®2 EºÂþAUG´,MÌH'½ŠVÃÚ¢*íhC»Rôvhm ´O8½D»ˆ¦á“íŒÿr¨Úê羈Ž)†‚lÅèL¿çÌ'×IB.ð¾·¬þÿàðtÒKG iK¸%þ“ðúw-³:š‘'âó{³تÐ{5qºB“þl…A“ž`›Ó·@¨þ¦A³¦gMX÷ \sðÅðYS†ñ(9eÎÿqw>__ú.à E|ì?xŽKþßë¾óoCësµh PpèçºÌÇ6(\ÜkÏU ¥Ngßòô”C¥Y§EÁ*@mš¸%[â:4>(mV4? -nzu•ÐV'¡](¸í ³•F‰ÐÚNΗü:ë†ê¦æ¶¡+†RO÷ÞqèJòNà;»wI©¡ÓafÔ*6 :Š–êÔž…ž×YvÀë0t4&Ë›žïÚ2í“ ]ÖÃÕ)Ç ›ÿzŠ_ä8ôÿ,T H€Þ¤*ŸñöèCÁpQÂü6 xú™SC; ¹H¶ÙÞŒ„‘ëŸÔ~úÂÝ{‹êLuo%¶Û½» “=•ÜÇ[=a†«ä]g1Ì>Ûˆ°\o‚ŸfOétˆa‘÷¶©ŸìXÊOõØ^r‚•+i¦]Ž6°v³D4Ðü,lŒ&R¼¯m‡í‡Nã\ü¹å£òqméÕ6™ qOÒdôß[¸çćÞL©wHº9|ÄÜÉ›â+J‘òly7ua6Rû̬ŠìâÞŽôÛ¬ìÞ?mDöäd”ñí-óC¦7ÔÌ›ÉÌȬÊ,lóYCÏl|àC¶fçпã\ÁütÁUä:.°>ã÷y®îÛX{¯|ÏÌSrýçõ“;,(¨"6>Ë‹BáìÂüIQ„t’}ìc*о¸tÌõ|ŠŸ›¾üŽB %ì‹Ôï–]E©ËŒ<î™x ƒ_¾òM.ÊÖ< d{Šò?/T=BE±)… ’o¨ ²2?£ŸàÁsš%ï=2PÕ¿©ô÷Ǩv¦E0Ç/¡'µ¼åÁ‡_"º¾<šA-Ee#†Á·þ)¾zÞu¾¬½)îÿ‚z%ÅåÝhPÁ}r;ƒ:·¢"šLÙwil¢ÙDÅ‘]£z4ÿ~/ÒFÃ-úÛÿ7¡åÔˆ€wÉ-´Ú½{ж’m„ëÙo颭m•v'Ý´‹¦-îMDûáÑ÷—ýè ²¯V}9:&læWœ$Gç½äÅw|h6ÙxsœÀ!åÄø±öÿâðKµPf zˆ2Ÿ.äBSú“I<Ï![i|Å¡G^°FáêìÚúÚÛŽêS*rB-ç·'kÔŠP>NÑþa‡0·ÕN쥄¢û;‡wŽAÎ;·óþwÑ™Þú(äDWå}Îg€l÷²U­!]Èʨòòw¢€ÌjF¯GÈúþ­õ–ºäñdu§1þ¿÷c ’"PÒ‰¤XAá„ÞÜOÇðYŸ•ûg9=”HÜa¦e x‘_·ë|Ñÿ½oªð2!³p%™²CySC¼!ìÓ;âœã„yÑ܉ER*}Vp ðƒªW" V„}D[Sådý¾o¢®{ u猂‰°ASLà>÷uhê°®šºD˜C‹I¥ƒ U‚v'ZæaN,È©Š‡o‹ÍG㴡î©~ôx÷¡ÈØ Iè®ñ©jp£…Þ‘GjÜóèÕõʃHvüÊRô€&Rh]:J‚T*Úù¿ M¤ys|-Òö$ë5Û!º†Ü/´DÆáÃÅò. ¸¯h¸ç¨0²|1íÌ(½ˆl+7Ê7È)‘ó¢…émGwäQÓsŽÒ¶E~3Çû-Ó(0ùÇùPèCìŸTmgéÝæ ‹oEñ“—/Ô¢BI»–úêñ@‹IÊGaüÏ&r‘¨`Áí“ Æ…JJ¥óúžoP…4åäñõ<4ǯ½FŽêÂG–ÉžG s_›†>/ÔRæÛß|‰ÚZt§Ò÷[ î¥üz§Y6Ôÿœºb2"‚Ftþæá´[hrô’ ¿œ;šÝ¿eô]ò'šWËѬ}iG‹zµ%F;7´ìé-T:T„V+)Jñ…>hí9¨¼€¶ái–q;h÷äkYëK±9m½Ò#Eîä+ôÏDÀ–à›$Ú¡yèüZy±òž-4zšt¯mt@õAÕ¸ñì3PAtaàðÝz1ŠØ{^„O:*_ùº7ÿíLj##ksû§ãŸ>ŒÍµæÕGÃÿô`üO¿Íÿô\üON ×KÊ Ñ{+~F+ùZJA‡Ç1ñ&YèÌ1c©d*…®[×Ä»å–ÿ9þÏûîä¡¿XówçŸÇwÈ]·å]8ðïùnRh…Á?ïúäJ4ûº‡ç*Ⲙ Wh5Éo+ëßÞšØ!±'.u0X×¶k0¥#âb]u»£åiÿÞ…ñ‹ÑA¶ã`òÊðê«E0ýḑ‹«Ìʰfì…ùîǃñž°ð餻ÇÑ$X:?t1=0VÄ^ÓY^­ƒ5éb*ÁÕ°ñôއp8lë¬tdPÁ-# A1^$’”9§„Ä%—>¨Š1âž#†.LŸ‘´äÚ¯—vÿöÔH…žäŒÔŠNRŠ6¸7Šß³Sé5+TÝá22=­G¦Ç÷U7žBf1§˜_N ËWõïBÉíþÖSôËÈáêÙÍžL‚\G$¿åkE_¤ÚW5‘/nrK­ó ò/ÍßΩEA #­ìgPè†t°jþþyK™‘´Eïùjç ¢¸nÄÓek”öâ´xƒRiÓ!†Nx q¼^{£l†wó|dÊ«Ž+úeìEEŽæêÿ\?«|8æfÿäô :¿¬€ª·j멞œF5熈ÖW϶²ú>,\ yaH‡Zh¬*Ã#ÃdŒm’¨Óh×r0õÚí“Ç´eÐ`Dïêè£4ÚujBS¦dGnc4§¤š=àöÍgÆù>E‹Ý"+§Ñr™¥ªòÂZSòˆÑª°¡D$qÿi´=Ú=jègIðak²ª¯:CÞ•Q’sè ôæCmå.:ÆûÚë«ÎÔc7¹Z£Kh>ý t¥ÞŒœ¸ò_–ܯ‹êÀ@¿cw”¦ÐėͧúÈodÀ\Ÿf÷¾9¸¤Òm|ƒöâܸ’܇ÿô×t3–ØR|Â¸í»¤àÞ·÷ó?=SÿôÖVºŸvTƒ±4^Û‹<Ýÿ»·fämÃ-6z˜e|­…O¦aþæBj2×UX‰g” ‹¿74ΑÃò³ŒH¥ë°RE¹À`Jk¼ó¶éŒ=°þX>Gd"6aÅñÂíÿpü»še¨D`v3šµµ‡Õ‘Ø§*j¯ã©¥­a‰R$Ý!ÉýÔ„äcI›e‘²¦‘SZà ©ZÔvï ­âѧ†×‘~¶é<ÝËVd|˜öàn§.îs<És'ÌYlÖ¨*z/ ›ÿ{‡c^ªÈñ#äyÑÖä~3tÏÀó9ò}1‰ çŒB>›;ã'QÈÖñSÜ>C¹²Úa“à‚b»mÁ'“PbSía·á[<à?ßu¸ø:Ê)j榠‚2˜Îü©Äév¥(¶•Ç2[ïᡬƒ¬ÎY¨.ýsf°Ÿ5Ün¾zßÔ‰ZÖ4ï¸9¯¡¶gòêºÕ8êfz¼UþŠÄ«7Û~žC#7k6-Ç#h’ço·ÿ+š RÙÓøÿ‡!~¹ Ãq®hÍÄztf¹Ê,ÓÞ —À¦Ûç/Aº„Ü¥q–Ô–>¼Ø¤&¬ããñojð?ë}ëH”®96‘4¤B±á ÖcO¬â#se¡Bƒ‚ wXxéôOþ€üÞíÙ·ç.@þW•°;ŠácB¯UëB” ÉSíJzö^©X¨}÷‹ü854öè¬w§CË‚ø#MÚNøÆ°Za[íî:'½b¡cŽ,mYXºTZŸ0B7ÍjÇýŽ]öJ dáW 3®æuoµ t|·/V¼E#C¦CÐQQ·©6äç—åü¡KHºÿ>Ã&tÓ}•ò”´‡%G^©sèõ;}Ü6dú˜æ#Â%c öÄt»f ñ˜Ýóy #6ýÞ9ÀèøŒ®*ßO6kñ#ÖÉdÙÙü˜î¶ù(Aþf ªö ðß„ù…÷—M9óžÁ”*5,ùI/W“-ÀŠÂr›üYXS<¹ô"`#’î` •+l+ŠKÉ!øüÇ÷ä^$bVˆ;$³…Ä/‚sƒî(ãé<îJ•ëHZÅqÊoœÉOP—/ô#åº+çåÙ‘ú0{mÏôîçëßg_ŽôG4[â'G‘þìa;'d ?îõ3z™d³ŽQ!KÝÏW?#[Õ¦‡ú}äp±ù^ªu¹F²$ìÇ›Ã2°Šùž‹“;ؘ"ÿz'±LËc”I2)¡HG¡€©õ?n¼(¼ÚüéŒï!T=µ¸Ð†â¶·ßÑÖ¢„6›»ïC”r&þâäg€Šb?|R>²Øénw¡¼šVþn#*Ò.;Q/n£2¿„#a¾:¨s©ãÝ‘dT5¦eãäòG5 O­Ê! ²”¶¡Fb 1å'QÔbpÞ‘’oÁ#EÅ ª×5P'ª7£E1õB•ÓìÑà>s¿>E$ôI; Ék© íƒÐ,å[éÒ šç±$P£G‹ôø¹Fh™>r¾¹ÿZåkþ¹È‡Öuƒê÷›m¦Ù”þz¢C›÷„/Úë<Ó.u?‹Gï]j‹=i‚m®ÂrO¬ÐIôC€£Â$:ß·Ýô'YC—5[Çãvtèª¼Õøc%ºnñ€âï(ã$ðî­09ñ^.Y5žžMíª‚•v­¬½UŽð-«sh # jˆ,[Ì œ^‘Ï7±ŠÅž%ˆAáu×_5d–ðÁáø-íI=ÈÍa~p—Þ7¦¼~@ï(¿Ïy¦.A¦üQnÿÌ&ÈÐãH²£z ™æD>jð^d1Eìlä†5ŒB¥4|¸ôãbÔ’-Ú±(4'oÀ§ÚW‚R3îÿæªr‹rl×9¿Á·PZkXÍ—{¾ÐWÆpo@¹üöÇ¡|PÀÝÃô5T„T¼=8•ò§ßÄ TÒÈËݪÛׄø§ †í{µd„Ô^[õð£ß‚zm²w2sÐÈ9YvI1š6òNWË@ ýYÎs†yÿúÐÙÜîÎ)&h¯ˆVYül®M‡ßvgB7¿ïƒ){#è•ÛÛOú®S»Ú6èÿëÃÛ$J®¥Ÿ`TÀ€èŒ¾Œ½ˆýžÈq&Ë;#Íã`rÁ–®êØ"L ½Ü"’‚Yš2Cûú˜÷Ìáb´†êݼ¢8Xœ½aûIP –-û¾%ÁÊûfÚ†iX£w}¬—¨ ëÁ¬~cª°©L—baÛ‡ì5Ïñ¥Áï ®ü­+ؽQšY”eŠÄ–Lñõt®HòçÝFN?’V’FÞk!Fò˜˜#¥´ÂHyOâç~k¤NØXŠkâ-b\Hÿ<þ9WÛ5d<®uc,™÷I«$ºŒ,<í­™kÈvXC¦µÁ 9ž³Z\XDî£Þò_Un Ÿ‡ÑïySpжPÝXB! ¥GW~¢ÈïÖÝT«Tð†ï3J¤G<µoÅûZ¨¢Ø]Pv9üƒŒ¾ ÊÏì[éeœ@Åê‡{Ook òƒù¤ßw¯ã!ßLªVfT{mì-6þ5{Û£æ¢Áëþ;l¦fOäã†Éº€%/Ãô¦ÿZÖÌž´k—.„Ÿ”_ÅLdê`a+ç5X ¾ye8TVÐÊÛ~‘/Óõír€‡I,¿`[Б8ÚŽ+wo¼}Ìúvw˜IX"üøAbvËwGÜÃÎÑ© I?ÈZ’ T!¹óí»_‘RŽì¦>Õ[¤Öìò¸›Š{SøN  ½våñưd¤;&ɱˆL¡j¯ Üœ™ëUã9³d©àcŠq%äËÛœ/ŸçA'õA·©>ärRþ”9‰sT³µIh“F %º<.÷PÔƒ¬¸ë¨sDNÇ"â!êýpëM,5ˆGÜÎä*ECŸTÃ_’÷ÑÈeΖ­MíÓöiL¡‰sW93šžñ‘=f—-B²›ï¡yh‡Fà—´È`}òü“3Zö(̹ìy‰Ö{OD˜g“*ÁA1ÏΡMðæhÔJÚvm%LH£½èÿúk è î·”p†',kø×N£.§Q ç¤SîòúÛSX’éÒ;s:ޱë¥Ú‘ëÏ¡â冲_¶+|(+þ¢OBï©ýí¬.-BÙ¾· ‘GwsžêýÃy3ídp3ð'd¤Uê]êÚ´nÜõ¦Ôi‚÷î|áN¬k3K¶Ù\ ³ÎCn¢aWØËÈ¿9óç#;xÿN:&1›>hÒ^=º1¸› EWL‹‚ ôW€æ”Ë#(çóå¹Ê æ¶d¶Q¾sëìI¶!¨\N»+·MUÏ]->’¼†j8dKß(Õc?žr޼‡šÅW$ú¬÷ Ž’®¸– ê=Ä¥Êá"4Šú¶Ç‘ ÀGbë3úUW¡yr_à™¦h#~=ÖB}çT,vCgõ>¾ÞÝÿƒøíHè—¨¨]3‡~e×Ó®ë0xkÔ¼+¨ïoP=üvi?Œr¶?u‘±†±$›zº—0~T,ûÞ^˜XªQ¦µ>ß¿½ ¸4­ 3{z/l7Àœ­×“;‘0¿à¬[rã&,ôz:;Á/ʼn“ü°ô2Bôv€%,ït7ç{«¾›ª±Ž½°.&üMš?6EK³EdÂàw$÷}¹¬FØÑÏ>¸«$<|*îñ°µ[¹åxIîmYÉ!©¥ÖLàÉ2$;ZÉïrÉ­¸“%Ù>!eú:™&g)RYžíúÍ4òcžÊg{v+>ø²;ÒMä½ö¿ ›{”xß ³ejËô¸/²Qì1"´@N¦ßnîåäÈ-I¡»Byy¶„ïôF"?ÍBsï¯Q¼“öàÙ`?ŠX޽ì']A±ôlÅ™+ª(‘3À©ðe %ŸO?«ùêŽÒç ém PVŠÊVOÎå+³ÿQòÃãi[æzý"GMmP9ìn|ÌD5ª^T¾h†j÷•I‡È-P£ø ÃÜ{ÔZ=W^ä7‡ºB¢dÏþF}…ÅâShHíÊ¿oûyZŒ>iD£Íc²iäÑx½;'Uóšì|TvuoE3*Õ@Ò®\4äÍŸzÚ†ó¯_ң內¤OüÑ*QöÏAùE´¡?žñqmø¹Ö¢dÑö’÷™O5Ah×|òÈ6:0x†”êapµIÿ34þý%¥F´™·©B]ãèaõë(éúQèâ¯Së'’€ÆÍ[ ×} ænÞ ñ_?w÷?¤LO©CÙQfîcŽ¿º<@°éEEû¨Ü¨=]v| &:Ç5Ê aŠåxh•Éa˜¹9÷ðØÉzøÁ=Ó"H] DŽAî.Á⣖Õ-ÊxXÒá³=+ÊEÕS|°vWø—\)lR?x²½ËÃ/ôç4hÃNkeWQ^ÈÉ`– Db¢Àn©÷H’¤‘>HjѽÒÈXŽdRªÁ|£ H®E+nÀ‹”Ióv§:‘J]euKji(è°|¾ƒ´A ƒI¢ýHGÿÑIHÉéK¨ªBZxÑWdÊ÷Y2[󒆪!«mXƒh“²ßðœß±~ˆœ¹\O,‘›\$’\yŽcï aä ·ê¿UÚ…üÄç*é¬dQ ©:·ýÜ;ò¶VÙ‹"\öJž¸‚¢O*Šr‹ºo¸Ê½ñèäów bI¨PSð6ÄÒ[jï÷ Tr¼·ÔõÔ•{ˆuhùPÕ]8¹Í ÕFbÚÄö¡úê¯’Š€bÔìËN ü…Ú-á[Ý‹¨Ûª}ÙîêWè43ˆDCeŽC¤$h$Œâ÷®3 1Õ’Ñ4Ùš&³ôM‰®Œ“þ@3§_O_~ƒæ¤lüQÓhAû|Ò[î Zò“’É'Ï m^ÿè…Öv–÷~ÏáÉ ØiS^hSÀúâÛ%k´a<÷!ÖíËŒ9_Ä`à 3§XµxÔQ{A”ÍHO:咰连;Né“6çwá_£¡öqŒÇ¾q¨` 9q@P>˜ôò\6ЄÂäW­?Æ" ÿ]Î'IQÈ=v1´s¬ò/çÖ/®ž„ Ýälß½”æ-Ч¡d ©!ß;Ñ ÒJ‹Ýz^fAÖê7'™Ë’[ÒW7èý ¾?8uò 4%¼Q!šü[4(tvsaßQOÓoPærHÐ|é&”ߊ4æ?^²Ëç5ßc{ÿ@¥‡—­­)TQJ%gž‚ªtºC–P­µ@ûˆŠª§Ú·îDÌCÍì‡H‘ݼkms‘êOÛ"vŸ…FºŸ¿ð/BÓl;¥¨¾=4w;¼¤j?­kCéb½æÿxè(UïˆQÐÙjãmټݽ«¼ÍZ WqfùMôiè7äWs¼&ƒ¡Ez†*gþñЇmÞYè4Œr˜>Çã cÉ!_©v`\1&$0”&Ö%.2Ýõ±Œ'ç;ÏÀôμu¨tÌé¡Zköƒùž–Éúõ3°P|Û$í1ü–òÚ7 K÷u™ý`yA‹48‹V=¿ì·~s ÖùÔÖ`“‹›mFÇ~{ Ú]¹°Ãï¯~ És’NeB¢8þ%“)$1$t[=¤Ìª¥™#ñÜ#îHΨcÝì܈”gT\BlŠJv¿¬`ÒPÝ:ä%Ò6—Ÿ˜rBºÌ‡N³&DÈðáÔ`Ž_?2S»eMŒ¬¥Ïû›%‘£®«çsD=rï ÔÌæ¦DžNŠS×å¥oä|#c/ j0F…ŽcˆÑD«Y/¯h>Fë_½yU]²hÃ#W"¤Ô¶VŒ+2 ÝÓ(F·: vNìÐ(iGã‹JòWÛê™ýÇN³ ¬r, ÷»ϼƤy¡ñæôÞ¦õ¨Ñ|±¼¯*é m2m¡tp­•7Ú>\åZ%èÞE÷6|GöÓCámö5‹7Pà—E¾.ð·G¸çûåvÈ‹xn«ºJïü§NÓ\‚’Ÿ /ŽDtBµ±?å›nhd.lXœ–º¦Q.Ç]¾ }}ÅÔ:>‹É>I…®Ïf‹´ ýÐã¼Aã¼ ½mƇç(¡¯€‰Œ Hû¿µw-ò\[RÛßžC-º3ÐÃ¥vñÊxên¼È汫Ÿj"ƒ¯z‚Ðcv‚öbçAèåõbíÏÿ }TtÙÄç¡_&UjªŒ|Jòç$c`ˆüR²h‰|]z8# _Ùkèäyã`äËêÑé0æAUpÆof|µþ“,¼_ºTÃÔqñÕ° ?˜yh¨¾á?È.,ÖÁÂ¥•Ôi°øšÍ‡!b–,zí]LaEótºÄñŸ°vë}ÛÜûØØ®HÚåai–0) ìÄ{$œ¿‡„–u©Ž¡ÉH47mù)·I¢í±Æ‰#©±ÂGô$“ Œ ,jErý÷CDe»¾’ãñäI¤Ò¨ =4tiö—‘Ç1!mÀ"5Áº ÒQÍÿ™”øƒôï¼üåúñŠ‚C`½æ®¿Ü?ž.ž‹¬vcÝY„[ÈøBQ1y×gÏ‹åõ !7)?* ."FœÿvÚyÔRsÍB~ò"U×(ÜWžœ‚BÁ+ŽRPäµ¹`Ÿ+Šæq¨&=€â2>™”E‰ÎO«æòƒ(©tJòa5JÍ8ò’ Ó¡Lë¤lžM(Ê5#¼>ŠGóGHt ? B¤¥5û´*N-|5tY@% g£¤ž¨üÞliªªÜ½-O,8Eµ Ë¢zæ§:»ÇßQÓßoÏõÔ>;žal²ƒº½ijœÝ¨o@!ù禒}*¯ðGÃé(ÂeÁ-4jUrœãêC゜€gÃ,hW»ð0Q MôKÓ³—£Y`çB“šû0ÉÇ:¢ÅåO|û4åÑòÊþ3ñ“»þÔWäó©].ÞQɯÓnç# @›a“¬šF´­¾Yºé>‡FƤ ¾¢vHƒ˜çæ¤s¥uy³ƒ°ø{9ø|ÊmøttÆ®TŸ j&>—¬ˆCy «zlì,Oÿrމ‰„Â㥊·ŒÄ ßÑbCh’rv^?¸Jû²r _Ÿum€ôùpš€c,Æ´¿ “<REÛ~D[›þå£o¥„3d}µ©/õù ¹]vEBF¦ÏÅøñ0ÒŠ.\r1mW†íܦl¥»¹0Ðïþ¢2” ÷î©»éå#ùÌBD¶Pñíáƒ_T†T›Ù¿j…*ÞÒêv› ¨*-;ã»~· ‘Úlªçóì§‘÷>úÙ©[ v9$qùÛ-¨7K¬Ö|á ÎÐù©îƒ¦nÕü}CêÐ\ý¾°¸;Zçè/E¼üÇCg~¿îŒ.èìR¢ë.…îëºwlžØ@¯n’ð-è·:¹V ƒ#T[¶—þñÐ[D{j¸FÙ>×¹RùÀX ‰#ÅiWÙ{ÆWÈ&~¿JÊÑ/€ï“åΕÂô&ééDט;¾âF*ó%R—ua!Ÿâyü| ü¢»Nñ!™–¼éÏ­æ%ÀòH™²€\¬:Uõ1|£‚uV׿iØ<’vÊ~ÛˆIå+äÃö¯‡cto#ÁäNúáë4Hä9ºÈÿRIv:ˆÈqÏP®ü¾õŸ¸·FÏ™=2 ÷õ\¡rÊ©DJÞè÷–ÿàzUU“‹H=Éc²½å„´/†‚ml؆À}dðù•ÃÕ¿™š×YÚÃõ¼˜‚å“ýÈq1]I÷T r}^;qNæ9ò$-<¿ð¼ù²æj-uœPpßèFt(\|/s FE‡.$?§D 1Ï3Js<òçñØ __”ªÙÛu‘ûÊÜ6ž³DyÖ^âq^}T r9°g;AžoÜ‘†•êˆ3çîG•;ÂVa‘õxâ,ÝÊáQT¿p0Ñfnj†ú;ŽŒÚ±¯{»“EQ÷©Ü…‡Dk¨ïžØ:9v Ùž®ù½Ù͉v©gò¯¢ÑtØ15kj4î! ØHËG“Ú®šÃÔhZtµ)ºØÍrÊ ,w*Ñ<3Ø"øìZdýRfÞŠCËw¿ë Ѫ2ÂèdÐúÓµ9žœì.â §G["5fûÜL¾òÍõ¢}Ý(êõ™[.‹2{ŠÝ^=Û ñÝ\Ævó!4tm¯‡Z¢—E\±PQpk°é×Ì®_ì½v{0>HÜûB™EVÓ“†,/¡Pó&Qýo(cÒÍ8Ny!m¿y~Bñºx­Ç®ÞN{*^d‚’s8.ð÷ܧŒ”ƒÐBêÓ!ÐêðvT"tpÞ«xø¬:Wgòv6¡Û’”scâôLÄ…†/‡>³.rèWößi¶ùoí»EÅPk<ô0º1'šZCOZæò«<5è½6‹Ðç8zí#ô‡2EÞ¨Ò†v1ŸÆÊ^²Y%¦|2ÃÒ¾S­Tðõ>¿”·£”©¼ùõ0Sc>óšÆÍäׯÅäêå»J˜²ã=e×3 T&O©á‡©ý`g-,°Þ939< ‹Yn[’`étß½Ó‰×`Å€òQþðvS-ýõË Ø˜[©÷ÖÝåá õ©÷~å°s‰Å{‡ƒ q09s¼‰úŒÄWäîÆäáI$ÕMˆmÇ$“«xú%ÒÉÍ)lƒ -‘2—ƒ0Õu©´NpÚÜbGÒÅÙÍSH{£Yͦ¦éÈ>Mä~ƒôY‹œ—Þ„!£—Nÿ½¥/Èl¶2eÐÍŽ¬òcâÇ‚‘=¸>ôú/6䬖8öɲ¹)”LÔJ‘çĈÏB@:ò>kù@ÆtùiI뫎K¢@‰öŸÌo1(­´ß_QEb‹’Ø]P´Í#Wz Å ÿê-„Ÿ§r>£¤rec£ J}¹÷!wò"Êäùò}ß”A¹W‚ÕÅ£&6ø l~SæúAÅœ¬ÂbÝdT:¤:{å.*?O‰Yg¤FUΧN<Ñx¢x*¤ÿ4ª;eµýI¥CMAm—1ù/¨M¬`Ó^(‹:ËNåô#…¨÷¥‰<)S âg­¬–Ð0u¼Û¾{îPP¬† ±}ùVý 4Q¤Eå—hÊÆt¯¡~MwÕ¹oÙ Ù×è{â{N¡ym L£Eúâ§z¢hùäw”eÝZÝš|9Ø:‹ÖG>;E$ãIgÅ„ii´±UìÀÀ½úoŒ±£Vû«õl=î’´Ÿ‚ø"´Cûð2Ç}ªPs÷Ýùˆ/”3eh‰3d@qÀ´Ê‰~x×9˜ÉPÑùÔ;<œ©!§öÓCV7È2–Õ>ùUÒ#/] ö‚Ôâå(Úî Hi+$K¿"iâî§Õr@VC¯°ªØß9œ4ïC…ðÎVÊ=²î µGµ‹~µÚÕ©µ÷©ÃbsaUbx”›ŒKï3]eˆ4lïúc†Ã>· 2E,äÔu¨RPyAÉUmöGR®ÕAõ)¿ð—ß zùq­Þ¼ Ô|]Ii§8µSu1á¦ãçèî‡7}†‰AÙ9î¿ç¾Y}wƒæÜŽƒÒÏ…RX<µþ‡çeÔYè s€· røtßáÑPéÞÕûI{;‡nè÷xAazU _9Ó*¹õ‡—äÞº=¢€QVG×ß÷`,ÕoRƒ´ÆÕ/+]¬‹‚IBò¾ÍUø>EA;ž ÓË®eñÏ`N4ä‹RÌÇǰ /HÂB|dö¼ä/XÜ©ߣ"°«ßþ{9:°Üqäz]!¬Z%Åwñ¤Ã:íõ¬Ïqä°I:½îzU~«H‘- Ãvím­»#‘ F?gKk‰”^ͯÉg!ñð¶B/Ç îyåô¥Mãu”9ñ)î{[óg³)¾ž)LµÅ7:} ýY‘º´ZݶîÒºm¹4|2E:Eý5ÒMjd8z;衤 2ùÛîÏWc@VÎÊ©ÏvSÈÁy+­Tê8råy(ñj OOÇÙ-_äós­|¶–ŒÍ‹1ç)%QøÜP¡õkg½d<¸º€â}Þ•5êdx$Ûõî]M"”ò8âÆu‡e]E=”K÷î{\ˆÇîÖ?5ÜŠO<µßütE¥SÊá y¨²¯ùWçªöü©çoHFµ–!Ù Ôè%?:;…Zé>¼u'P§î…Ó­jÔ‹ §údf‚&–žF¼GÃ`¯±¶;ÑhtêûÎÁÓmhŒ¯ã«´’Ñ„ÓÙëõIDSBînBh:! »Ã€fÍ7 d”мÀà»Í*ZÄÚž @˪ÍÍdÐÊÿÙ>…]ŸzùáõÄtxÒ}Õ¤› ƒ²ÕWÂ…jÐÈ–Øb”6 mÓ¦Âz_Âò+¯nÝ^èðkdùá¿îÆKmp@uŸÈ«rÞ»:è£ e†RŽw¼ã»>u¡žà®‰ U ñAáþtÙ¼n3È_¸õÀÁ(òˆÎ7RwPAnÉà#Ö¨A(¨=±ÊÔY%©ûF…øBuEܼn†14&ÈöQ C•·ÚdÕ.ï gkA—»GЏ«twêeC¯“úž‚ï}æ²–ùÐßö}íÉÖ•+åUgÓB è#6}'Ç =›òOu c7W.F¿3„÷®oçN>€^ÆÞƨ«»9S}ѱ™÷ï{‡¤)¡äîêºû0ȇI´0”}Í~Îë ß%H¥Éž¯¥Ù*Æ0zÌÖ¨o Æ}YªîÀø­%eÞè«0é^×5£3 S -|ê¬0Ó–Öæ ý~x¨~(ȯ†!º^Þ6&X,)ñÓ‹N…¥óYÇÎŒ¼€³Â²“A°æÓuÏ<Ð6†„mÝ–Šá·Ï×”ó+WaG­A3A/ ~¾YW\@¢ª¢^ñ,:$¹F"ÉöI5ö]2îzŒdGÿÄ—œ]@r›FbÉs¿ò={¨þÏóH¥]›.,Š4$#,3|ˆ´×’|‹"tŽDþÈiZ-¤Oms¡RBF7ÃHîKÈl2<»|í>²žzð3ç-²‡.|PEÎ5~ <@nê“.ñŽw‘GûÈ?±sÈ,ò±Þ ù™Í´3sQ žŠ÷W8 %Kð¿p›F‘wË´ÌPtfh5Øq Å5ÆH„I™Qâ;å0JªryÕ&ŠRYÆ'ßt ÌÓÙŸ~E¹Ktƒz«ExTß=Á½q„ã*Ô(XQñšû T¢âv Gåë[*³Ôܨ²¡¯âÒ’Ž'n»ßjã~€ê¬aý/ËTP£±E³Êµb£öë1¹£ÎÝ_“¯Ÿ¢Þùà1½ÏûÑ@|ÁêÁºÊ%7֡ѱž‚4&ß³ËaCãAû³s_–Ð$ÕrØ×sM¯vÇPg?A3uº w{4§s:™Òƒæc½ ³/Ñ"·-úÙßy›À _•©he¬z#‰½­ùôŸˆGëu¨ª (ÓÑÚ+qµè9÷©õÄàA g—d–)X¨¥¯Îž]†¶ õ2=o}¨~°nÕ\ eOu*x@1ƒÈü÷#ÜðÎeƒ-ý°ëßç0: lj çþÎ£Š‹J¹Ìºdhûwަ',^ROy^‹…¯W¼•‘l÷Ÿ=±èÛÍ‘¡ƒäwuêþ²–”ùï\ølĦ‹åÉ’ü¼¥z Íç1ö¹°Bø”7ªNx Pñrý­Xx‰‰àǘ `ÚÐ×+\å†ÅYêRƒ/š°dðâí¹ñ°\•TxFÛ Võï»FœÖ€õ}Ñ¿*.‚M}g‘pø-$ÿåpclG_ÎÙÊLF‚¸W…LŸ‘ˆÆj®±‡‰'§IQã{¦ ¨h.Ü«BEI_”€ûŒ¯==‘‹/¤š>"ŠÏ¶M«EêG§åK,!íqæô}'ônX%]ëS¤ß ;IæL²ó¼:â¥Ã4ÚA¨Ù[¬éEIÖ‘ëî7…a»äÑ+ÙggËŽ|:S¥o¦Pà]¬}ƒ ³.òCQ i*©n7Ýýù’9¥$z‰þ>J1ž$j› B鮻Ƚ*(§à:ÝLÊÇø¹b¨¯áñ_Å9ã­µˆ›;TëY¨q>´/[UO|? ¹î¨&ôF.aÖ5¸žìO¬QD-ºw¬ÇO¢ö¯ñ$ù]©[tŨ¸Qõ=Kå2£!ùm69³B4T­u C£4¨¼7õ}Ù*ýïG­æ¯'Д9üݶ;šN_I™<€fÅ™A\åºhâ[ÒVU'7͹¢eÐòðõ’ÃEhE¤Šëth՗ή±z ƒLžíÍþ hØéôýIA42øÝ·ùñ¨ –>‹=Ö.ðƒO —Þkýù ”ìî.æ»9Î8VI_ÿ¯ôgy5¡¤òQû G&(~c™Eà° ï·>–IȤïúÕ~×ȨLÈOûDw– r«øý% ×hëðÞSêPà,xcYÊJ‚ÊîÝmƒê±yþ¹œ(h"U;+ÖÒmº×´ céÖÖGîjè“=“Þýz œý¹X wlÎhˆúzh*#ö€ÁO‡Wÿ­ýÑ3ãOË…¡ÏŸ|VËé2ôž•~7;=uÜÖá„ÐS}<óðé*è5 ,Zº˜}¼y6ÂæuÐOes$&î Ȭi%65 /ÍžÀa=øL¸YÚ¢ÃÃÁ猢ê`„ŠÍhÅ¥F]hˆG·þÀتÄW•±Ý|˜“ªV “Ñ mW>ÕÃT¹ìÏ9W˜™Æp÷Nkø|ÊþÁã*XNRw(7€Å&òs,D°ä{gî[9¬Ø88}P0„µ+§ÎP8ÃF“Íò©˜]Z^sð8;L—YÔü8ðÉk•§ôH”Áõø«¨’œKX(;iŒ¤Ê^½gƒ‘ì¸Bê3¼…äŽwR¾°‡!e;Ýò¥5¤ÒUÓ(û¤€4DmZ“]H{Ù?‡ÎºnÛ4s- }B§Ÿï22:»7>5 Cf£ºŸÏÙVõôøŸ#é¯ý!»àúý·ÈÙY`Faí‚ÜôÞ~†ÆÈcp+’¼L yñEøeD~®7­ç9ÅQàS—V˜m r)¾ A‘–}ªd‘Ú(¶÷XÛɨ]9…ø^Œ&G‰EÆ'g¼¿¢ä ÿ\ÞnI”ªïšÙsJe®'šÈÜG9-"•½|x”5oÑÏh­ª^èIŠŠJ"gBéh'þäu¾õBe;½™/û$Q¥kg¿w%žÐ]ðmc/AµN×1…/ÔðžÏÖúˆZG§Ê>ú¾AÖ€‰÷§}PwûDñ*ù(êW0Öÿ~û EB|ä§¡áb¼ …3 Q!'qÕóWh|5ãMÌ 4‘}.)ÆL&¿zÎ8¶¡iÆ#áï]vhætøêuQ^4gþð–úX:š·ÆÇô=ØD‹€ÃžäùÎhy$™m"} -ÇìiB¥)Ñ*ªúOÈÈ\Ò˜ñ¿ù5­µ×$j¤‘öQ¤Ë[¡RXLvøž­á;er߬ ºÙ…1å=”Ñ·é’†¢ ©ý7ÎAÁc¾"ƒWúø{5ŽÑÏÄ\ðFb·-š›긇«EÔ|¼÷z~¥iÉÄ}T¿Ä$¤"…‰ÈŒôîŸ9?áÚëÔ®D‡X-? -¥Üí"5<ØJyŸç5Ò%I\˜LGÆyùq !d¹5¥úí×®~}Žø¿è-@.[+¹§—”‘‡;^Ù´¼ùXœ‹Y …QÀ°9Áè–' } l°±ÍÃCoèUãÔŠñð§D÷•ø zslõ¶wÞ@v‘t%½]=ÇÝ˾¬FtC#†ww¹ØÝEmû.£®Ì°ÎíFÕ•®_ªÐt¯‡Q4o%šVPòq¾A³+t¬OwÿÍ…sFgú¤Ñ|X‘:XöZDRêý¤CK;ÎãL¿ñÎÚ4sç“"4Ô¨¿ªmŒô¯Îvè®À«Òû*6>øäÈi’|cê#ïȿڃêS¾q$gvuáo«ä*”œ¾è›øÖŠm*£ ŽÅÀû–¡xUÉ¿þ2ö¼éßý šÆq«.ûáØn.Ë™bm\χ2râöÇd»×_äâ†Òã¶aÐäj²fPyÚ­xÒ29 Ó6€yÒYº-vx…`zš2²ƒ ïTÔ&QÛ(ôϙͽ¨; Óï¿‚×ÿVn»Óƒ»¾³o³ŠÛF¸ z»•r=[‡^ÞóÌs¹Ð˘ä½tnzÓf_p†>öúù1—]¾6ÙòHŽÀ@¨s²|) ¶ßj¶s»ŸMý˜3ø3á‹(¥ðÃ+¯aäèàãG·aôIíË*.&øÆûý'Ñn>ür¶çl'?LVœm"4K„©š?¯K¢Š`–¸[×Çs~Äù“¸\©„e™ÂWYÁ»z²¼óÊ–‚Oƺ[ŽÂŠ#¹žb$ ¬yýT®Ôá‡÷‘¡Ê»<”£ŸíY®Ù©Œ$$XaŽ [P@¢è{&ҌĺxX3† Iúج¶®"øä]a¤CrWeýûHYÃû|=©ôê†æ¯ê A鵦ԳHëe5á% ‚׆—¢‘þe¨ád*2žòSн‹Ì™KŠUÈê¤@ž7.„ìÑšÞ²=îÈ9¸zZhH¹YFı!i× Æ!äMÝ 4äaF~Á_†µ‘™(0ôÔózÌ%ª§6ëj:„"c¹·yÅxŸ'KTU¡x äB)ÿJl‰sVŒŒ¢¤ÚW‰ÔA9”ú°˜ÍŽ22?ïuI¢ÇÏÔª (?µ¡>D…Ǫ¿ó¹‹. âÞ©ãKk ˆ™EI¯‚vó¡R|T‘ ªd”üÑäêÁ —/¶1Gµûš³ƒ±¨Á‘÷ˆ¾S 5{ð¼ÆçAÔÎ>°q÷$ê> z{ôêë£t›~ô|xªöž SÜHè; Ñè”ê ßmv4¦ÛøªÉœ€Æ5Á_«¡‰G¯ª%Ó^4=Hã[{ÌMó×é´^@3Ã8Zóh6}1ŽaÉÍovZò$ó¡ Y™À1&´ˆ÷Úò·ºŠ–bw\ÎäIb€ð‰Ðµ–>Ôˆóž`ÍiCšÕ.)²‘ð“ÁãϘ]#´D•ñ:íæ,eÙµk(}lo8â7E*ŽW¥r¡à‰ÌV7Åß¾ÒåWÄ'!‡ÎöùÞŽYÈ”û®vô¸÷ß}ÂçÏI %.Uä„T1¼-p³š©û¶› ]ý5!+PŰ?ò!äñ ´ÿ0 …ûÓ¦8…Añô"ñ;ª(3L¸rïtÀ® _>] þþi¡R4ŽåyØE¨¼ùg V¸ª<…]âr *pôè6‹4T›RÏMCõ†Ø=bwu®°xý7Ô´éël™Bm÷uóÔýP/ñÕÖL âŽÆ}š¼ô|ƒætܧfÞCkéï' ùÿxxaîdÌ©èœi‹²Yûݱo"V½‚Þ»á¿羄þçÍŠ+-0ø~–½bQô]íK—ò÷Á(÷GM!ŒeÎx6¬Ïøɇ›Wz0yÀøüœ\ |Ÿ»–df¿ Ócη÷©œƒÙ­¥Ër Ã0oØ:MàøÎqþêÙ €Åªd÷ †pXæf”ƒå„‡mõ`UÞp{ïh¬­µú1‡QFå´â}°µ|â^šYl³ðÿ¦Ñ„– Õ§ S0¯cÊ}ÿKnÜ0>é‡$cþ]¬R»þ¶òSÞþ‘÷HÖÆ:ÚÔÿ)h»ÎÇàþ׊¬×ˆè‘Z)Ý—çÒ|6Uăg%.'"ý5Ù=cJ#ȘL­#)Y¤¯/ž-½†ìd3)žÈ%Ù.et· ¹W‚|t¸‘wjüõo78(Ÿ¬ªŽB¾Ýíà ýxèØêÕÕ„WxXOAõ@+JÔÎzkçyÝYEÕÉS3ñ¨ö.íËÄj„4:Ðð„ –þÃþ¯D|¨C¢Ö–Ù…º¯;=Ü–¾ >¿ÍÏi4x²³¹Iƒ†þ>Lþh$~ßAa¬†8—ýDã[¶iŸþ0¢ ¿Ó¦ú~4©ui ØDSûÌëöeŠhºvÖã“4š…nŸηAsö†ZÛIZ4Ϥá–PðÆ;/ZÑ•D_Ãýæ®L]-ûü²6’»Ý íéOFö¬ÕC=Ý,IçF-T $Ä5^߆²á®J“äÿ½gA2ûøï{‡7 èÞ¿JLmøÞÙJvêRÙA>[•É(Á]²äëyAÎÍãŽ_¶9ÿæÆ´åL(A¾<·®@pÈÑbhj;}÷BÁ>hÿ³>L­ÿ:üä‰ÒÞAwkÃHbêqèU¯ж^ø;ÿyY¸:.õ6xoeÁàUÎ/+#rÿ­é£ìmg ßûªÅùŸ»9Q™_?T}×?fÊ螆^JE„ ÀÌOKŠ úÚî)8Oýdï¥Ö‡‘^¢–Î0Ä·üðØî}NÞñ#P/7NwXÞ€§g}ë0ZsM&þ„|3Š=û‚—&(y8ïÿü “SE5aƒ·`jb‘òîѽ0ËqŽ"xÿøQô2UżôêcË`ñ[q,EIz>Ù+ç d*Ç'`ÍÝãŠø:l$ÕòèNíò>õû›DØ.k- ý,†Ÿ¿èÓr@¢€ZòBOM]ýåH*«¸0qɔޗÍ'!¹×Ÿe™;•HÙÄ&„~ûÊ@Ý‹“é·ÓÞ»m!­»4Kóbü•´ýb\é£ó‡ÔŒíÑ&¼()M™õž¬ŽiAÖ3™¤Ùò‘ýÅ¥Ú+Ü’ÈùMÞS݆¹93’sJ¿#5_2ßFòfçùѶ/#¿¸R7+‘( Lž|âÚw…z <~®5£È†ÕÈR¡4ŠáúÍ= ™(þjèìžwsxdz‹öJj@Žøˆ"Jes2[̉͢¦ár¯Ï(»^ÕlXÌXTÏË%ú âèX~Dä¬:Y «bÒ µÜl°6?U~v7ot_Ïvm[‚­AmÛ)ÖئÔünA7ˆÒ!‡ƒÂ‘ðvø¤bÚq$ÚXû.à™$‰R>_g‘4àHоåj$ S‹väyŽäm q‡È¹q¿%…QX9R3ªPõÏi!MÆå ­ñ²xÐæ#+hBz¹ú!ãtÈhûÚ‰žU™'©bgƒM­sÄôº¯r‘©‰S¼<‰ÜM®¹Ê·H‘÷ýéÜ)Èß³«úõt:$ØoǼ"}z;·P¬©“YpË%¼«Xeor¢$}_Óàþ1”j!Š»ÅË…23U’¬-QþœRéµCGPüÙ··.¨˜xq:Óú*q™[hÚŒ¡òÃD–ûçKPUaóqèŒ:ªÈÜëY;†êuDÊ ¨yÃåb^àj‹fôæJA’­ýdôJ¨'v)ò[”êgñ=¥x€Ý¢X2ކ!˜ÃBõx›v>îßÍ…uY¿ÝÑXÕ&aeU ›êÝDÆ ÑD[}HèØ:šÔ±oø[I£é±Š‰44M¿s×c˜ÍkyùlðŽR(0e þ™ƒbg¥[ðà0ߪfX|î¼9{Ú"0•¯d@Ýs£Ž½[ÿõÈü_/ŇdñŽ3â]PÔyfçÇø¿^RáÖ÷×{è‰ÄÛlȹ¿áx?ò¿¾Œ|vš`3ãÿú.jtû[ZÊ ÿë¥ødtönqù§ÿúdzT#c¹†CoõUr¿k'¡ßîë¼aámزH0–$‚!’»½•­ÿ­ƒ²jö½·¡ÿ—Xx`ÁmèËâYYâú¯wæÿ>§Ï¬ÂþòAå=4To ŽX=ÿïú!_g Æ{ðy#°Ëm°¾t0¦Ñ(XÀH„s[Å0Œ.jówчoþf‡TïÀÄÑÏç2öwÁwÚ}åLöö0½wüªñWs˜•ÿó1*]~t|Ðg“+ƒk󧌖aqa#¨$‘–b÷]ì‚/;ŽGs`Í…|K„®6­«DïúJ•·ºÝ¶°é!NQ€5ªÏ¹‚ÈÍâÓä$9þ½Ðžñ2’J$ÖÚÍ;d*+µ‡ëåüê‡ç£òÖHÙΚ;A…T† $uçú÷³ÞÔÃHërÀP°ì(œûøÌªûÒG&†-§EF‹X‰Œ¾&dÖñÛ¼K.¬Îßùß™²ÇÇÑæ»¯!ç¬ï ÇÏ+ÈÍ×Pâ0˜‡<öWr%"ï;·=~íÈ/aqáR ,r–XÓY¢ÐÄB0¥€;¢&ŸÖF1 ³!Îÿp{Ø@mPÙI_.GIÍXñw“*(õæD.Ñs-”¶m^>Œ²?úHÏ9£üí /Y?Žâ1›Ã÷é×áñ§Ž³u’ˆŠ‡C¥TQyÚXá²#ª8 ^!@ÕÔ§Ûh•ðÄÉóþMT?¯g5öM 5)æÚU^k¢Vï!ná$NÔɼ©4ÿÔõÜP_xYŽú}{F–] ÐÀT$= ߇Ϙs¢‘Ácó7¦FhôÅZBî;öÍÅø¢ñÈ6?2K41 ûøƒúš”ŸòЇ¦Ü oŸ”ÞAÓ,"Æ¡qhÚÞúç­ù4c{Ü~6ÑÍ÷«iú‹\õ©ó-ª ·U.†#uÙ—4cy˜KÌñvšH)"¥k"¾P•³tåɵŸÿziþ¿Þò«.þžúö_¿ÌÿõÍd\Ï{Êï©ÅK¶Iîðö›÷U£:$¯¤ÛJ‡”¶‚nL.†,cúÉ3çpïÌ–gÞ6øÖöò?}ÿ_UE?ƒð‹+ÿõãT†T¾4¦û×G£ öȃ+á¿ïù_/ÍK#²/¨¡%ø f}jªïŸIZxµ™sÎwC;¡žjÇ÷<hðâýƸøšÔw^*åB³eµù»•bhMòâDê<¼ÚþÑï¸9t®>ùT·/连ªÞ$"çâŸùÐÿ^Ù÷ðr7 6='/vÕúÇC'ß+ìy0JGbàùªÆr†½ÙÀ¸Íã»z}a’Q´+DàûOwI1˜î~¤¥³C—Ïq8ËÁ<ë vÆ“°pB„ÓëîX|vœÝW¿–ˆRx µaù¶wü7¯C°ÊÃô0À¡Ö&Iˆ´Ësa£a„]d|¶* ¥øÇ®ÀŸ/Çô?ŽÎ!õ@ë–é%$4Õ§RG¢ÊU£ÍÛÓHræ;×›Ow£,IÉZ‘LÛSû·ûC$¹0 s¹÷¸eu©†jI<ݹ‘æFTÝa<È—2®Ëš‡tk9Õ+ZÈH/é“Y‹ÌOßèzF6ÿ÷å³:“ÈYâmzª¹_LË}KEÞ ‘ÌϦóÈÿŒ6K ¿ ‡|G¾¬ó¡ˆkÑH’3Š©»üTêD ¡èJd‹zWÏdI±“£~ª³¿~ãG4hšÝYðµDC÷Ëä¯9hÑpVÌ2l«N™ê$} @£v¾ý‡Uߢ±ôȱ…¢4ŽêÒŠ‘Bãé@ê°Ì 49ª¬í×Mƒ&·®ò3¸âm•k-.Qwä°.‹ý9Içäƒ:Ð7û{êt:ôW²„; í¹¿vLaˆð…L÷\/|öÜÓÌr¾È öûžØ†y§õt`4lá§TÂ7Js¯‚ê5/|Õ.dŸ“×øúœaÊs:å.Ž÷ «”‡\÷¨èhàç‚ä;ÿ›ûañŽÅã¹°ƒ°$äÃ=Á¡ËË7´‡baõû±‰W\`c?OûA&ØròÈ)ÿ- ~°º†nÊ9Ì)‡‘ˆ¬7%B¯‰Û¹?ë/àžB¯hù²¸÷ƒÇs”jÄ}#e_¤ƒòJÔÝ „xàµG¢£¹ RÇÓÝxVc‚´{É×¢¨¿ãÁ»ñ¹9e‘^ äeÕ^d˜ìbŸB¦OK 6ÑÈ2êxÚiŠÙ©ã„ôV¦‘ÓtŽ.عúu¯ÌN@žƒ«)>O(×µ@B‡ãòMo%s_’EWu×n³£Ð]Y2œA‘7gž  èxtòÖ/(~´jÐ.ôJ´¤[ž‰5GÉ=B.,å(uê˜Á ª0”ž:“ üPeCÈÿª¡¼š‚Îe½ <ÆÖÞ,xÜHßýâ ÂìdëÛi¨ÔÓÁ©}ZUñ¦¼¶aCÕ wihº¦ðDÙúq¢a5T—èÚEŽ [¿™Ž£Ö —Ò‡7…PçbŽ‹U_?êI]ãr"è@ýîke|h°Õ1í2Œ†¯‡H4·+Ñ踕¥Ý6µÞz«ÍüÍN3U, £qטLw{4ÑÊÒuœ}€&…ô+ lДÕ/`Õ!M¯]·kë8‚¦md}AwwyÉÌê†fNo‚_M ¢_‰…ÀyqTcå}LD—‹Ôo®ˆ]¹wæ¨8D›G@ÓZLLhC•Œ³Ö„¬)|(+52ÚÕ[ÌÐ RuA¾´Ö$ü †œ¥/£)-»|“kž Ø„ô·›—3Õ!eRQɘì¼Ô¸Cva’M"Š+!E„¼ì‚Þ$dfÞ»êZ¼y$^Žîù…Ìž‚Né·á_¯Ú%:(˪¥zÊõ*¨j«©ö¶CŵŒFTÊ;Ñßù• •>lf" Ç r¾I>IǪì¿8[íUmaKáLP bûµy²¡:s“Í}1jŒËÒØ³B­¼¢ÁK¥×PØìãS-õJC\ÞzÐÑOR{>Þ»êß -|Þé´B»eåü“mø44e¢²¿]{o Ý”{†V¡gÙô´Åˆ)ô-liy¿ë—]t¹ø^ÂgòÏ“?ö,À—w‰Rµ{0÷lXÏ8Æ$™®ñÀ·îŸ<­Ža0œXß­Ýfº4¨`Z‰j|9zfU/'ïÔÀ¿h/Y>kø9zúâV˜&,§wóZð+q•š,sY¨ÄÀJqÙz™ ¬yK¾j?†ï­ÂÖÿ£ë­ƒªúþ7^º»$”QB:Þo¤DPDÁ@PQAEQTP¥¤Cº én¤[Bâs?~gîÌýkÍ>sfÏ9{ö³žçµÏYÏÒUuµR±€-×§É=\°w6ÛÊ´‰9ÖÞnö ÉmÎÁ·GH&@j2Êí„äóTˆrºpD‚ýÒì~üµüN(Ò=yv·¿ü'2UÒØ…®"³¤ã¸áÁPd©÷I=ŠlêëÄ?‘#ÑÔæ»HrqË—~3}‡<ý}—”g‘ÊÍW4däyMšÕPˆ‰‹ÚŠ×Ev¥»jMŸÁ½VB­¼­‹(ÞEªIµk%?ç‰>%ù…RI³(ãY$Ï×ðòðêw…î <÷ÛòƒìB¨xe´É òü/«h>Ts í >÷ ¡ìȹýï¢P#÷ÔüÀM=Ô2î?WAc…Ú6^p½¹ŒºÉ úFkͨo;Í×|ß ¶nSÛ V£aàÎ>÷þh4™uT¦ÀŠ Ï‹™·ºn éé[¿IÅ´Ðì¹b]€b1šÛ(‹SíqEóöû!.—¹ÑÂB‰øX˜-Z”úÜ´Ü÷@„­t-C¾ª{¤u¡åØ©£"‚ÒŠ®ŸÕÝÍB#úx¼­sÒ!×ÖyÖ? @ÖT›ØØe˜ßr¶ù}Eê9e¿Þ/ýå$³gÄB ÑÅäg]ßÁ¿迹û³˜½§ 9õ½×ÆßEŸ’ŸToéÛ}›N„¼jàÈ0¶NÐ'Îtß^„¤4´&A9ÿ×ÌßÕy\$î{¿íƒÒ¢÷dâ—¡ÆùqémhVî:ÆÍ?÷ø4>€– nã¥j_h£íãz[f íŸ4ÎÖ÷@'Ôß|Ô ]úëö7ɠ˪¤ÎЇ :?¬=‘r:‚ùOäúGCûEyÊGúÐVðŒo”BÚ^©\¹tBÚég¢Ym =ƒÜ"ö{"tœ;ñvT:O1ЉBWp#ÛS‹/Ð=r¿&úpô¾{G1lúúKGT}ýaP,‹ñ•4 ½²Ëü å #ÊŦ‡Ñùé³Ó0Þ>Û™² &–µÙîµî…és6ŒŸNÂܾµ†>XØ•»ç,~дóºËWkZ,n<„ß'=™YÈaížÉèxáUØh‰È“¦…-ëcOüuU‘ˆ«˜“:8Î&@zyI]›¾wÉÕÔܙ캑Rfô¬Ù{{¤>ôk°[:iƒ<.ŠœGúE6+ã'‘I;T0Øž™ã(ßmžGVÁ+ßU<‚‘­à|й*r\a¼•{‚ 9m= )“—‘Û±?ŠÍ÷;îˆRÿu³y |Y®RGþ‡MÒo?Eê«Q%,b(Ä·é]\,‰ÂoÝ—›¼KQÔÐKõìÂ Š‰]!;| %ç#õÉÛQòtëª&§ ÈÙi?ÓD†Ò÷×ülƒjP¦ŒL€ÛLòIÍ l@¹ø3מ°N ‚±ÉãJaT¢˜ÏQNæGåÞ£Kö3gQVi†Âë(&înKÔˆsrìm”GÍß{^ñh+£¶rX‘Mx(êܺ6ý%u7ˆLYÂQÿ%¯Bæi<äg·n'‹FŠCO*8-Ñx㘴í‰4=£ÎtO(;Ç]½á{ÍCÄnÅÇ»¢…8}5½Çg´(øÌkÆg‹–zÓ:îhYUÏÙÛx­°¡¾«ŠÀ‡)t™ÄhÍ>â`GV‚Ö—zžq”£uvÚ±±D‚>t‹}JðÅ“ØÀï7€~Ç1ùQg¡Ý5w–·ÎÈt¿{Q£P¦®¶VU]}šk&–ÜPœÛÉ+·”GÈ¡«fòžYÅ“±Uâ_ãùÔ!ÕÛ·Å%õ6$\sW‰-ÚÔþüØK;á—GÔ(Ä…BS­ÙoI·¿}ÄÎL|”‚Ví¸#+Ë„œàçzötîW|t¯oº•SÓÕÃá7õÈ”ÿ™£0°o¿×>JeL·!ª…a“C…Æsf0ò{÷«€Å«ðòqt\€_ ÷’üŸÂägû©Wf0ÝVf=½s*çã9¨~Âü¨›é1XÔ=jªIKß”ªÞ×߃‹ãG×gœa•ÂUÙšâ¬u¬\`„ZÑ'OÀæÈ£D›ììoNYŽC¢Îx»3N$ÑV½«/‹¤=|×EODòè’ܹõj¤|Âp“?‡¨?uF:&“Ûû¥ò«™0³™zD¸-‘åŠí¡"Q.dSÖÔè;‚è)ó+/9ßT+Óš0"5åÖÖÞßÈkcý®$‡ ùë¦Óå‹PðµÅ‰š_4(üÒší¨ŠÖÖÕ8§¡ø!“÷ž?Q’™öÈPÏ6JÑ髃|L]h8“k,Í‚†ò]‰¿Ð(ª%/YGçn4¾œGS©·§³ƒòÑÌùÎ+Æ14WßÖ?·;ÍËF‹Î^@ Mžl>mI´Èl©7“êFKA9F¥z]´¼÷då~Í-´6¯Õ¸‡VJ-;&½htÛîÆëa¼¥g5ACtµ¸‹¬¯f¹æÎ'Ã|Šñëüp¨slõ?'~ÊE_‹“„­Cñƒj™çNÏ´Ç_¹fr.øÈœð!p¡øìúCæåWåw; yÕT$÷Y ¤–·}Û¦ˆ‚/ ®JnyÝðÅø+‹¸?5¤ÅÚÌîy¹/_žØ×Ý¥>ä|íöP½üöå³*üؘSg¶¶€f1ÿ[{5• E¬_øˆà´¦†½ßÿë´ϧÐ]zÁ§?¿¿_¹œª¨ yйŠñRt6?z2r:¬¸mzi·)–up€¶Ä±ÚƒšÐv÷ä¹.ah›Ê”Ëb= í~ô_Tä~(@¥Ö²:‰ív™¯BÙ’YÖmèv /Ï{å½ä®òÆöÆÐ϶ƒ×a üºô{µßxRÒéü¯HF]œ¯ªW]„q\¦™K € ‹ÑÜ€É]0U(Fí’³îêåü` óŸYÓM„`Qý°KMN,ýñO±çó•Ñ×ywCc`uçÚ¦ŠM;l蔕Å Áf–í“x×mعr½€Õ‰µ[ìšß°#ÉÒSζoãHVJ¥8¾Èƒ G´]&*ˆ¼C² i†C­êÌú‘^+oäÁ57dü±+t®3™uÉôNæh#Ë7ã‘Äô.d³´-ixeŽä˜¸wu÷-T["W›cw9}òÌGYÝ¡¬GÞ½ ºêj®È6“›m·PÀoí%ÕlÕÉTpu@aóí[Þ¦(²ÊôÇÇ%÷Ö”<#òGñïã+#pß »þ2›;PRo¹ÁÒúºÑúTQ&@‘«îNÊöŠ„|ßdA¹“—M{^D¢=#×ãϤ¨X'¼Ãð •ß®}R-(EÕ ¡3õ.ó‡ì,Ço Æ•ñˆ À‹š%»î‡KØ¢6ñ›3¯ìQǨ/³ízêuVKôB}§¬ÉsBxHdxÉUZ ·Ø²¥íëѸÄXàzM•ì?™Q Y¤ÏQæ4¿ry‡Âœ-ØkJ/F‹Ï?í…›ÐRéÊ¢î ´Ìÿò’Aî ZÉ/˜Y|x‚VñGy´vÜКÉè`ö§~´¾ ^Uåd‹Öµ7j$×üð^ju Ú¸÷ìŠaÃëDì5Mµ¶êU„Déé¼V6í²Lž”[W:Õ/|mÖÜ& Œï¦ñC.‡¬æ.wÈXr|Rà½úwßßÍÒOHÚÿXF<¾\Ï™<«ö‰:s}S€xh"ÿ¬1KðÁ_%ëBð‘Fý§¡ü$Ä d=iy£ _NH©xÿ¦€4[4_d­ÿî“Ïæf¹sUäë/¡ÀÍb.ÀξUšsuÜ´ÙcvlP´×|ã‹û'(*6hôS…b›Æ;ìÝ6NgsJöô–ÆÊ¾’[£ì}’P2:=PÀ¿J/Фë„2ã~ŠD„rw»Ð·CRPñÌ÷»Œ&T ¥š½Ü¨ê^®gsP;, àå¦õ´üZÇ y9jD;ú04Éí§`%…æÚÛ¯¶ÎºCËÓµÃÏJ¡í<ß–ïg?è8Jñcæ•6tºëÿXDzÚÌâ´Âå¡ÿÌEæ«í0(Z)b° C±‡ë®3ÈÛs??[“ö¢>ß¿XI‚OžR„IVÇ9²Ý0­ùˆãbæG˜}v³[:Ðæ®µ! ,ŒU™¨ý„%®n¢M9Xž ªRHOƒß/ž´t$úÁšë–ë[ƒ 3æoI“4°yJûœC£7l§ŒÌÒ½¾‡Dol™{¬¬„Ì™ØG¬ I_ª~rO¦Er‹Ãž#‹z„|ëÌxÊ©5ªÝ²ÞK"m¥¿od"2ì«åØÛƒLgMн­‘…OróÚgcdí¢¹ïßü Ù[î1P˜=CN>‡”ÃFþÈ](½çû6îù ]æ4¼ùOäsÊ¥CAáûÅi(̱\uýHŠj]þö.ìŠåD×?nÃ}WœÂf³áwýšûòN(]¬Òcúhe=˜vM’5£œCÊ {HesãŸÕÏ¡·oÏ-–ÃgR.ò+‡!m—dj¾‹1A×’®R TÖCõž$9T?ørÆ´ü8û>R/š÷´¤ÁÏÚ˜?•7NA«®X!Ã!h[ˆž3¢踭íPsA:zv©W¸ðOIËymèdL8z*Ø:xYU  ]ÉH›kïÙ¿\xSÆÅÚÎm y™³@[µÓõ:Vh×á hx›íMþì4ç› ãU[B‚á|/•›éyì¡«!¬¨¥czN Ït]‚>wnm{pѰ0ÿÄBÈ™xCÓ †%wÞÚ0ÀHC#!úÀXüºšÐ¬üJ vöÝ¡„)ºKR.˜ùPftþ/ÌÃ…RIX¨ª P#ÿKþòœKÜaåDm죬^‹e ™Q€õŒÙzö‹¹°Éiø¤áf)lW^$õ¶E¢A+©KÚZHX;bµrÉô[¨ŠB‘‚¿•ó¼ŸRíÞ%Þ*NŽ4˜-w¶Èéž¹Þ}G}ÙzH5‘i„h¦‰vYì·¹·÷#ë/µÛ¡Èþ< ¹b^w¹Z¿ ¶JE.ó΋ÑöÈãå±'´÷„iô¦"ßh¦õ Èø<6§sAÁ!«_H‚Q˜Lã©Ì….‰ÉÊ¿Óà‰{½(57CÜPüh?Eli î»aÏô÷g_o qÚ‡ÒdOË•’QFÝ“åfŽÊFòì;ámƒr"7D$¸ŒQ¾¾1Ö‡PñcXù9JTv;² Uýég}_t¡úc8É¥ÔÐ÷n%ûBš!̾ŒÌÞ¨UÏÓËÙ:,M‹ÅF{Q70]K«w õÍÞ‰ÝØ…Sk¨uаT¯/ò•+˜«î§ASÒK»z® ™—±ÞÇ·áhîRæwb--H‚”¯úS ÅëMºÃÆh)AôVÀL-SBØ!âZI²4VßdD«¨´[w…'Кæ+=§ñ6ZŸy·©ýžp\ëÆ{eŽà‡·þ ¡Ç5‘Ø)ôyóîÎ^ Ôœ¤?˜œ‡L¶ ×Û%väáŸD‚TNŒñÉ…¢À>õÒ@ŽÖ•7[' ƒ/Òóºð}H i "[ø»ßïÅèÅ ½ùø|írɟѧu¥\ÈýË®— bs]Ï–B¬ÛóçŽK;§FçC¡}>ÝŽ\ec$èR\3¬:2ºëƦµ” ':sL­ò‹#ú/-fÁ7ðŠ"ã(¤›|XØ÷ o‡»ñéç__~ÌË ÅÔÖb&uìPì~®ï]?W¼9âGð¡n±/‚/Ži gÛB)©ÃÕ»7ã ´]š¦ë(”uíy#µ¸À‰/ËTÈ< òŒÆC×¢PM­™VÀK 5«i—õ¼I¡N€ÆlÚ~šßILz ïhxžwró³JfÅhø9dh#óUZ“Šn>ûå í¡mLÖV/:¶7êËè‡îMMWòotÐw7‰¾nP 2ëÞ`> CêÛ+Æ-0ÜJwôjîmô|á˜ã~×»øÌ»aâºÚK:w˜Š öxzÓf¦1O¼ìcCå*XÞ½m¬ô o̺H Ár@ò‹¾dð[´ðüSOXL«/;¢ ë?7§º5ÇáÏTdÃÚ(Ø–‹¦öÐ8‰DGɾ·{‚Äa¦4¾Hji×úiuɉx:'ËâgRJ:s R5¢TÊÒ2—øUSr }dÝnÆdRÍø–…ÌÕ¤)²ÒÈê;ßTXåìï'…Fnã®÷3"ñ‡æ‘[Ì^‘ŒwÏÉŽ¯1( ß:ˆ»ê$ $—òd_E¡á¾/_Ø£HoZß9GÛ;tLÚÐ%º“êî†Iàþ²?ІE®(í¹–2:£ˆ²LT; …xp#gžxÁú÷VøN ÒÃÕ‡á‰ë¨Ò~¥ú– ª‹Gs .t V­xÈ÷K¢æi…Ø ¯»¨Í˜ •n/:9‰R%ܨç´t8‡o È+¸”êâðPxX|ZÖ.4¢a¾oÍ…Æîo¿ŸeiC“Ì~½ÏhºÄ#+un ̓ïªó˜¢ù›w{»ЂX0"®-Þ>:¿›-ƒ†[ÑÒ×È­f­-{ùï숰üðTFHUZ…<º?”yo>w¦µýäŽzu¯+Ç©ûõ ÙÝû0Ïja?¬gGÈe<\Ê¿oA™ÔekãPC÷Q¢r8>8ó@ÖÚÎ…Ë¥O \¨²¼ÄÿÒ7Ml–~AÚäôèÙRkH®§ÌyØ ‰R1õ¾ÉGÂ7x&RqÇtý¼ÄYŽ  dªÞN.è+Tóçk}û²õ"JUÐtµÃà¶àsøyúÔO©ö>hY_K5=Xm/,¤î¿†¡mƒqMNèÜ/Ém†¯¼ £’ä¨ç¥!hŸ ‹$míÜmŽï—oA[Ðãýv\ÿöXDœ¬$‚¶¤¥C¬ÞÐNÇëêFñ Úo­Ìhd½ƒÙ¼MÕ¯¡“÷ljmÀèÒeÏ›ã< ÝÍãOy½^Co®¼¾^Mô¯ÑŒPÏS †ã]ÏâRJˆqãŠëÔ|k›cŒÛgìeæáý>Òˬ¿aÒ±ˆe³é8L÷4oGÁ\LTŒ‹ý)X°TŽJ¶‚%’)7·ãÖ°<¼ñÄëQ/ü^hë¿ë"\mA‘¡ðÇž¬FÞƒmŽK»–n$ ÑÍfJöÉsHÂIšÒ$Ù¤õQ/E_#y„ñLÀ$ R0i‘FêÈ>væ…H;B%œöú28Qˆ%KÖ"Óí­pÕ£Qȼ¸¾ë¡AG7Çúö]Cv¡ÜëÆ '‘c¬VÓ Ž9¿[ÛäEîÆ¨›9\i¸{ãë墉Èg\XÌ»YŠü£/ïß@AlUˆÉB¡ì/ƒN¢ˆ¹ÓõÇ …{éŒç×>£ØR›šUtî#6á@7‡ûrD¾/B©;E–Á(½ñ–Vz:e3ÕlþŠÀƒ#wÿ\ÜFùÇO¥øZJPјiû~ƒ/*óI°;žFU¥µs{ëÊQ=®ªýЦ1jìÚfŽàšCM;ZÏàµÇ¨–VËß„ÚR‰ÜÞ¨kRàÐóõÆY .Ó¢AêQ é!S4 ò¡ è,Ec3~—ýa¼hòãIû@Κ©òPNCs#¦õ²÷:h>§½(y?-ßq–ªDKž5š¥£hãõììæu´”³ÄÃ>Œ ¯0<‹Öäœc×v£µ'‘tÑøu´®\†Md&øá ½Õn´ñbûôIi¯)ïUóë1A+:ЧIIムLd< vâ–€*yfÓG÷¡ðñq»VÈ>ûé`Ù +¤uë*·ñ@êN«ÞV$õ¼=Ì”/‰U¨V¯€O‡y† ëÅàãËßÉ|t VáþÎç1ˆ)ž²9ó•ðú±cŠ“oà3wëÖuYNH-oºvÕ¹2.L‹–óñA·ge7äóô7x<¦…¿î ±¹ ðí”ÔÑ—'ƒá[og­éÄ6ásÙÿ~oK E>)‘´3#„ÜÌ%›úËŠÏ8¬Ywð|ñ-¯wJ"û?Òš†BɳÇù›‚âPê°4ÇÚGàÄáøR1(·ÉÈ×ºÏ YUûdî¾·kðOlPC§Y#p³¾óRû¿.‡zÏ‚"Û¼:h¤? 6 šîÅ.ŠuÂO™7æ Ú eÊy'IDÚJ?ZäÓOæIN)»Ð-x§zdz#Ûă £a@‘ʼ¥³ ³‡~ú†uîj]ÜýF)<…—^GÀØèʾåø5|àCÛ"L‘ß)HÓ›‡ÝHm·›W`”ŸÅ}åäœañYqÑáPXæ´ OÕ°‡•¬ÚÁE{X½xRy⺬Ûÿâ¦zUÎ.Kå•×ÁV®£pŘ ¤æ9~Gâ{1‚Ó¼HJ‰dR3ɾ×ýÖDŠË[y¥HåFÂù©°iž^÷e÷¾ô´?Ö„M‘±øÔ™c2È|¾Ëu*3YyÚ¹¶ò7‘mû[¥¦7îö\1pA®§N/p·š Ý3esäK)wãbB¥«t2PˆCaúM‘©Šã5ý/%KŸ‘ס„Æi«—îçOþHwè$J³gݹa|e²‰Çׯ}ă¯K¶ß¨¤ |¿žñTbÛ|CL:ƒ*¶œe¶/ÆPí=í§K9ˆÊŸ§¾Jd£¦€UÖž‡“¨U<ì§:Í‚:'ß6–E=êè3dñÁ¨Ÿ.]#Å=‡tψ^f_FÃ"©z¡r4úÜ\Ðh€&׿[â„ÑôÛÉí¢ý.h6©k ¯ŠæÁŽÉ×Ђ©²ÃÆ>-žôÎ<%ð …L},áºZ^ß3Ét-»¨\ºK; ~(û„éîc´zÂ*Ä®õý"”÷®n*¢®~æèk9?dµÍöià ƒ¹¯©wÿ»_õß(M Ø_…B«‡/“ö•CN+±í'qÈR+ óaåû?.œã!è³]ŽêóyHÓSÓ=t8’ÜIë_|ß Ÿ~Tp’{ŸÎ'Š~;±)ëÙDŸ¿‡A«­£’â<”|”ñ> UYµœ’¾õPoÅè™BàË&¾»_¤žiÁO†ý‰¯dm åÝžÃЦJ|®Z6Ú+rÉ™& ãGã:t,Êg;»ß„?Ç£§]¡=ƒ‚Uœ½Ú¶\/I”<‡¶Ëƒ=7ò.@›^‰é1Õh»õ™zÕÚZDó7ÎÅC»ìéŠ+›§¡ý{”iýw/èøó$h”:ó^eQÛM@·rbqÐe è¥57’÷¦‚~©pï'0PïTc¯!C¢'jÃ×>Ãprºôq© ½Fú.ärŒ{Ö:'6‰!?ßÏi®ÌÎå0[ëXR*v æ{Šø£©`ñÞ{ÓäZ°|\§ŸÎ~{n ,Ï2¬…ëô¯—„ÙP¾œC°uò¼îpð&í6ïË{ƒÄ‰j×MO8"©•WâÙâ.$g,/ ³‹DŠYÝ HµF¶’þn?ÒÊ%—WìsEú[—ÙZ!“À†ÚÇtd~Ô ýçi3²²Q­»]z„l_…šý‘ã4NP­FNÝÆ°æŠäÖˆznù!wŸfàˆü¼éÕò2±ÿõu¿~ŠW(°²Çï¥Î( yqI…_Bá-Õˆ/Ë(šô4j™+Å‚ê½IüeQâÑOzêƒ(™¹ú=qÜ l;pÊ9Š tv#é°À~”ÝEq*ܹ Æ„4U𦢼™Uî/hT¤¯ô2“šA¥þÒʉbTYšk%IAu›®IúeÄ&–7vŒ©¨)HîHÒÿµ\;ñF µ£sN?›ø†ºÌ'¼»O¢^Ò-òHÏ4¸ü¦""ß ?ï÷æVGcšó|+2hòêó¹„p;4£Yd¯/ÎGsyç«‚‡ÆÐ¼‡(qð¹ Zøo—×DKz¶•c ÏÑò¥s”Ëm´â´ù™%OàÃP[SŠsKhµã_Ô®‚Ö']ÝÔöÓ£u…gåA‚^{Ê{vmN§.HÔýÆ+­)+ÍW[c^žL؇LBWÉϼ ‡ ^ž¥G™¡ònÍìúûÃðí¥Aßxs4d½–œ¬p†¯×xøÖ‚’ Õûï}âEøÒÙ wú !w#B}âOˆ0ÛðClYrRûyˆñux7\J 1»µ8üÏ‹x1ÇÈ#¬¯ˆ;ï˜ Îã“â´B T´'#7) ûmJA\™#ä]-rwü=DB¡ ìÒÁs³šðM™W–Ÿ|[q¥Îx4…Y£~Úo {)âxò__|¶ŸÀKÅî>‚Íÿ¤”­!ðbÑÅŠ/„ÜÔKZ¥E éÖ,ÂP–Õ”nDß D®$ËuƒPézë(Ê8‘¿y÷æÃPËC•¢±ë2Ô™‹ý!©¦€ÃÁ·~+“BãH¯R- 4ÌŸyvZĶ* Cëò¥gÕíË;#þ-âÐe ÜB—=_wtµ3( ÿ|°äaP‰–g)ÚhľØ6¯žã…VŠ›Ï5/¡UXD~Cú¾aÓyeN‰Ú¥Md&G9‘UÃlŽëÃ-˜ s±û 5ž.éde1Pªd¶GiDÈ…rq<’Õ=Ïx›fQ2ÓüŽ!p¡ÑÆ,q,|teá;¶©ï~šß€/üÏ.Ó¸Cü™±c_@ü^Ëý¯œ ÿ>×iÔ³`€ì°){5Û÷Pâþ8 é Ê¸c&Ò ê*Ê”÷¨Cc• c/é<4§,Þ& ©ƒí£Ego@kZ*EûÜ/‰äAÇùðÝí?È¡ãUóW¡ã» |—qô¬I+´ßܹYþÇÚú§CÒ‚¶òcÆ œ(+Dv$­Úœ--XïAÛ)Ž'qÐ64ü6IKÚOÜytY :$;Îêê„Né=iB/OCWô™Ùù›ãÐsÄËL˜Û úÜùNvKÕÁ€ƒ3ƒnQ< ƈJ³úÂ0¯pÙ¹øG02@ÕPœ{ ƪϛx¨·Á¯u‡çóZt0ec5èZø fv ­ü;WßÓ™ý°HcÇã/ KC',rÂʈˆÍÝtkX#?7á5¥ ªßJ›Y§/þ|n;£N½†Äj—õ¾G¶#I§âu=Gn$ R{ù¾bRØ9…×í D*ë}m¶ôo&ÐæDGÁQ¤›X.>í߃Œ·fªùéW‘Y죡W¯0²$ÌÑNû€lZuö¥±È¾ZcX‹»ŠܶÿŒ\©Q 3û‘§><î(Ó"òr7%êßB~ÓÍK¢w²P Ñ­ˆnŒ…xojž÷'è*þO OŠšv˜Îw  Oy|Ú‘û(ÁÁò ˜üJÊËWçIŽákË.hPú2Éf¸M"Ê´lRîo݃=ÿÄòÞ器uR·ýQ¡áz[¨+*E;^¯Ü`C•/õço+½Bu2%§#]o½ÅÙoLÝFé#¿ÊjÔ¨ýt¤µÝ?'ÜÈ¥BvL}ÐÅz¹r­—Ð@¥Ë•ˆào†Ì½Ö½9hÔ³ÀrXMNTQz_¨FÓ®ò-×fS4çû¼°„æÕߎ¿ˆ¢G‹S—/œR)@‹í(Û‰ÚhùPâICù}´b¸j³Ÿ‰À‡ÁŸ2ôï¢ÕúØ+åÃh}¬ôUŽw4Z—y,f7Øüðb¯á:Ùœ5QNrÙÄËÞJ)b¬N¨~O“®6,™¨{k•ê(àWPRoáŸ5¨hQ®‹Í‡‚èAA&^YÅ+Êœài±ÙÊ)µŸ¬^Ð|n§š7'äÏü½éÕ—öÅýx/'ÄÌ­…I‹„…ûÔ{¼¢áÃÕˆö°±¦ r‘J«ÐäW|笤ò Œi=Ô‚ôwq;f {·.CÅ[Èm>“bYž ùOÖœ ›„Þ\JƒËPðüØ«…ÓVðÍÉcwUEêJM¿²ýë‹›‰¿ã¼˜êDóçSÛì¥ø-Å6Í7H¡8}UgÊCJ®VrÞðƒRÙs…c‚üP¶kö³ÕÓ{P~h OTIðÃJ]Ûhõœ×Æ÷+^‹šsU‰iC-õi!¢ûYÈ*þëĉ¿aÖ28;Zrª“Iý'%ä ¤àÛIÓ‘h(ðë͸¿ÿ;d‹Ñ‘©éBF¹A“† [nLyÉÉyîÿeQo1E@âGõÃV‡Ò Žž2‰³ù|ìÝ»lõ’*Ø¢Ú@6S€ÆeK(Ùµžˆ¯ó rh¼Œ†ã.Ô)ù,c¼g{/>¸ôš-²{8ñ«Þv¬"´Þu×u\ƒvj)VÔè`1éÍzRšç¾ürÎ…ö ú£çUL¡ýÐÚ¤ËÀÊßõ†Õó¡Íò¾J-[(´í¹›wxo ´©¬G³ŽO(^Ñ$øá3‡ËÃÁTžt_ZÜCOàKz’ä'¾ÿ¢Þ6 ]”2{?†î”{V,¥õÐû.á“vC2ôÿ(Œ;ySùÆÃ¿1€¡G/lýøžÃˆÑò7Œ‰²,·¬„_b6.0Úyê+=Ììú¹Xsì›4—>ÂÂcºß>‡xaÉõi%u¬1÷j7ü «—_/ö±²Áz4wè¹…Fؤ8õ¾,êl'¿åY)šB¢É,?-‘ä(©ñž$ÝÎ8ýU¶É‹/æÖp¥"eªeÖÚ Z¤nÝ$›6qF:éi:£„ÈP4αœIàIJëÇ{ö^Elå½DA¸šŽ_ê§*Cö«û”î¿Ã]*5´2S‹?ª((#y)!þî¹ÕÍg»ìŠü¤7¼Î5£ÀÑdâÛAš(Ø´P·pž…mMò“Ü@‘ù=[Ӌθ7%¿×EuÅßYt_$.À}+j¼}qÿÒãø5¢9”–•_Ñ Š2O>V¹àA77]„YQ®±­Áç™3*„Éètη ’˲tBªœÌ^:{վοkŠ×EdÅDZUÖ¨‘½®ú;i5'šZ¦ùn¢öwcîÁÖ¨ó–KâCÇÔSúz†Á~ ˆvmž:ã‡Úý "¢QL_(Ù7;4‘ŠR5UBÓÄ—#+ÃhN½~¡?’ ³LßÝF ùha÷ö× E´˜7‰Ù³ˆ–þ4 Db¡hEÑ3Í„V¸ƒó6ÑjÙÄûuïU´>*ôC.G­KbëÂ_^#øá91Ç#Rhsnâ¶Ô 1^b(9u#yUOM~ú0„Œ÷y©H·a|«æO÷C…PB¢Yõä'°I‹V+@ÆÒa>¥GöÚøºõòd$§Úzr³ôÖ¢KíÂn ñÃ׳*,žô¶œåÆü—/Ò˜TÂâ?=­ìï›ç">L[AL‰=u Ù(H")‘ÏU̵6ÿoý†Qf C»\ÜõAîþŒ(}È{üå²õ¶äûluóoe@þ\«ˆÖ7çÔ§º6†B¥·{Çž¢È(íp¾’Š_L¨Z¿ûªgÍǯà}ý›÷Z ôçéh¦¸c(ëÛ+áÅÑr!|ÔÕ%9¨J[¯ÍPJMÛém—¬Q•ÒÔÊ“ÖÕ7\ÜKo‰“ý3<¨ÙòT¬5¾ µÏ_Ë}.puãã&Aõò/v²×£õ·"ù<Ô÷ ÿÕÅZ4²ký•@õKµäÊCÐt·á‰³n¯ÐÌMùÏW47eX« @ó ëÏÊ¢…Ç8}Úòe´˜” a¦ù–®„»Éî4Z6­^x#IðÃPÞÈ| ´z3ûeÒ¯ÄÔÞè^C\JÍô Kde èLÎ÷‚™¬†_ÙYPµgÜZ%š«%Ëv¯tõ@d¹PݹxÏêïs•…;d.!oYv-Kk'A¿9ÂiPˆÉëêžò£(bm«³g) ÷¶irþŒ@‰g~N,®ãþ[Ž1™O£PÚr4˜ÓÅe¹Ç”É]TQNàÉ¡g^¨ ÌÔ›ìžJæÑÉŸ …Qåfî`~Óª®ÄØ#N=Y›æ!AÍÀºcΪɨÍ{ÚI:÷êXç)TT¢Þ©þûTֻЀ¬[àv8!¯>KPeÿ“F¬Ãs«•lhxäsœ09šŒ½Þ"F3…r¥© 4Wñ’ØÄv4o´I1à§B‹#x¦Ôó Z t^‘gG´tŒÚö<ÇŠ–õñ-¦ÊK?<¼ÎwM­^‰õ_ ÁKñ_ß0º‚pJ§¸þÛ²lq™®ëÀ ÷Õ“2Z× JþWƪí¿þ‰ÿú*2#ÈÉö6ºþë™I›½3f…ÿzh’N.Ýw?3ü¯#ÆÈPk¿–ê¿ñs¨7²ÞýÛ“hf@ð‡ÿz.þë©ø¾gd³miüN_®ý¼ MášTðFú__MK}éùÒ@[h3Žó3?ð¯g¦=4ჷÍò¿ã}4AÚLV†þîŸýŠãl*´.ˆš³ÆþÿŽÿ½ï_Íÿ{ÞŽ–lß}‘=ÐeUà0Ïúzè ÷±3zCIYÞîë;0Àåé„c0xÚÇ/œñ= sr?r÷!Œ Ú«Ùù}„±Ñ²ö^/êÿõÖ¸«:• ³t->?T`>üÚ—{!ë°xtR(‰b–Íâ"ü|.Ào›òÒc¥°v‚$´xŸlĹlÅdÀ–p„¯ŸÓ;ØCË{wŸ#±çë™zñ.$™‰åøIß‹dO?—±#Å©èkêŽHuõ.ë9G¤)슥޼‚ôªïÈh‘qÑl¿Ï5%döz¯};¢YæZ*¦ÂIí6G0ærH•Ü cÚFN¢¨É‚J/䦞<§tÞw›ÍRûéð#oÇvúúov䯅{gjÞ¡ R­§Å9!Jy¥ZzŠEdGEP´=ÑàÛÚcKnôëØÐA‰¯sEã[ïQòW™b~\.JV¬·”aèZUCÙ¨/į œåáQšo»Qí |Ty}î *w…,Þï²C5&W¡\6§o®¢†ÉÌîI¿=¨é3.sñÃÔ*tQD±:Þæ6Ô-n¹¥÷õ<;¨Ä燇\/~çÛ-†F2_S&o q%×zJ.45TÏ।@³obÊ-hþxéQf;ZÈ϶]5’D‹ŸêGÈÍÑòd Ëü;´œïî ‹!ðá…¨qZ r´šàqÉ@k»1ÅýÚvh]È`±÷ÁÝ #ŒÜÑæ¢ «ü:¼ ßÜ-6º„ŠÔwöW¼BF_a¹î!+?þÆÊ4é_ŸKn^ÿ-·ouÿt÷_¯Ì=4ÿõÍÄÑm¶š>:± /“ÀûKmJÇïí†èûo¹+OÝø7~paœ—>àÿO§ÿßžšÿtž=¢å/ý"ruëùÚÖ O:èv3ýÞóÁ}9~-Œê¯ï@Áâþˆ”Gÿz¬¾õ6»×Uÿë¹)¬Ž¿Åìì·l#'þéÿ¿~›ÿzmþë³)¿ÈzúTáùÿõØŒvN|Ýœþ×gU—»ò–±wòóŸ_à>›dhÎO‹(H{ø¿þšÿw>ø¯‡ê¿þ©>~u¹§¯+`àBwšý)Úã>󆔆W¬Ê/•gÃè䯞KnÉÿzk&I†ô'f¤`ZŽM˜LîÌFG„†À5˜Õ<©µ ‹ _ ð­ÂRP»È¬ðNÛ~bn‡ß DS̰¶“Ì”ÿþX$… Åý¯¯ÆØlÞÈÚ‰3‰F¥¿. ©ˆ¿æËo™Höþ¡^ÕÒÿúj8öœŽxøiT;0 Çé2zO¼¡ Ü7»Š8¢/UóóOJ¶µŽX0¢ÔRÁ€†œÊäÍéG%¼ÃƒY…µn|(_Ê>ËUœ‚Š Z ·(—Pù·ÈyúCTóV»Æ{Ѧv7§ãÔT¤ˆkG­º£ÛõŸQǃ¸þމ-êq)mpEýŒ¬hÝ=‚wn IêàÖ£4£2½lzÀÍE«œr¥Ð¼XÙ¬©-Ì_Ö7$ Eû—Ã|Ahis#ëÁ½<´¬}’&ÑžKðC¡Ÿ1òth™"ÀBVƒ—öîe»úÎ Õ©·H/óy#Ëý±»h5a:¦ßÜöÎT&ÍÅKÏú»®è„ý¿>šLš®·J»ëþæÈÔ™ñ×- Ô–=„T6çÃõO£àË3U¹Ž©¿=‰;•J£ðaæ aÚÃúãgîæc/m ³ŠíX¦Öê¿ÄŠÑŸÑ˯º vãYZ”]44X)XûÄyþí+¼~â=4¯÷¸*ÎAË-Ù§ O7 uêrƒíw~h›Ê¹Æ[Ú íâÏiöØw@ÛÖñH‹ 窑D|3Î(<¬ú/v”ÈAë¯ÑÐT]h#%±¸Ã”m{¹øM| Me#9]BÚÜó¶ä…vÉ{.! iÐAq÷VÛmjè4 –*bv®º0N7Mè¹Qä9Å© }®\$ßW‹a@í{ÎÃË90xSbÖ`¹†Yd5ÞTÃH……Gqc6Œe²üðɼ¿*|Ü*š;`Šm—´X"%Ì|iºžkøæœ˜ƒü`‘­Jåš`iIúQ|µ;¬LFpцչü'™2`CœïŽ’ÜÿCÖ{Æsý¿ïÿFöÞ+[öÞ«ó4“½G¨¤%I„¦¢-E*©¬²ÊLfQ$3{ïM%;!þ¯>·ÿÛ÷ÂïÒq{¹ðråu<Žó¸=ÏçýQ›qÏß[Äζ]E"ËI$Þ­<žèÒ$1¾'d“p\ ?ëKä,:Z5&Hɦp2 ©ßè¿ò²i›?¾ˆgND†‹Õ–âI\Zû™›+2§žü‘^Æ¬Š½q¤öæÈÖ#E}]v9’’nîA®ä”¸cåÂÈ36Â~¾ ùNÆòkr¢@¸Ì¥B·ï(¸AsRªÍ…}~²ÜA‘ÅÞü *×Qôióø!JV?Uâ*JžÚz¾‹9¥c-Û¡µÿ<)§ÊOÙ)y5ô£âEšöö¾ET–I½ñU¦¹¾¢Zv™.É:j¼J]œ«§@­n†ö2MüYtÖ¨£šÝyè¢êzÐ?¤¾lˆz¹ˆŽË*£ÁQÊ˵nhX}1éÞQ;4z0U0CþÝ*Üx8ŒÑTFÁ;çñ#4«U>Eé¶Ž¶Bw–в9Ù¢§Ñ*Á#8ã‰Z[ÏXq>߃֛ñç–ªÐ&Q®ôNõ(ÚbTÿa_hÛí«ò$ í¼vÿªòåB»Ð¦¢-y´¿kºw_S>:0>b=÷ ž;<žµBŸ¾Ómh‡ª‚S÷#}‹³¯§ïÀ¤û·µz®Døìô²ýeëú¿ûÒO‰B^§ô<0ï'ôB{딫ÙEÞü`óê?>éc_RHyvø°í%xU1mã_L胆âŸH!AŸÖ®!yfG“vùŸ;IÈÃÛ‡Šå—DáMäÌ0‡$äíy¯Ç(-ï¬ÏöMÌüã÷¾k‚¢­„ØGB¿ xPZ¸#ø”~ÎöçÎ,¡'α”©jE§íò…²Ì í"²ïg<”š_L"ôE [ BÒŽ- ©ÊÇp›_æÂÇ ®—Ò{ÖÿñÀ}Ÿ8ÿy(}Ëäûà3øÌ[ž4ÕUW¾·j¸Á*¡.»Ï„žØ®Ð>éþê¨öŽ+‡† N*Ë<ðM’hÞ€šú|\.=­WŽvlgA»[ã[â6è¼Vêî)ÓÝ/ º¼´ÚáO ÄGêÚ¿é…aR çDUéæ{¿”Å ci)ÍLw|a"ª=¦÷PLE%qÔ!̼1˜r¨L…ï#îàáÑ sæ¥êŸò`þÒ±ok÷5aaÄÿwÈU]XrñëX>© Ë+—=—Âj}‰Ô/Ö°ÖÏ}ŒùF'lHúèûfyÀßìînKN$’où¼qy‰Ÿé´b ’R8Ÿ ÐÁ]·x¨xõÄ\`Cý¾ø>¤XJôèüŽÔl¹»À- iïÕ°½<Š &§3¸IôSDµíidn´ó—ˆ¬‘z^ª_‘ýæ51Ÿ4r>½œ97D†ÜSWs\p yãŸgÌ[g¡€ïñw‚gPè‚ £è1ytï¦ùV ŠnÿÕünz%2B*zÞ£ô]xâ’ò墷´­ÖíP!ô &s˜*Ý0)› w@•§á£w ªÅŒîs&Dš¸àI†(Ô6ñQ¾9„¨\ð7¤zuŧ)™S¡ÞGÈ/BáhàZ™UŽûh9ši~Ø£QÞ‘D§CAhlie&Œ&ßÎ=<ŽfÇoÇû;‰æÃ»LJÌ‹ÑÒrøâu#´r6Ïxq) ­)l›ßŒú£uvÀãD²9´±éè&uA›E¯É­æSh>T(ZĉvŽìW~¶¢Ýg™Ú¸Œ~©iá¾ n¨•ùnßîë¤ÈÌž? Ü”?dŒ4¯TŸ/¢QÊYcÚð±1ì䑤MÁÕÕý§T!Ïx—•½Ñ&äLïþ˜­ÔÙ{éϪ¹ü{Ÿií¢þ¤ÖÊq¦ß†¤‡»mU:Üv4#Eé]¶ùx|í¼`›æ'¿º,±øú%¨ëÒ;q²Ôúßï[I ¹³“|"Óñ„¹·çîà _xïȦôµ}ŠŠ¨¼H¼'¡ø~sU÷Ò”p9‡ ýëO,P¥¿^‰2{‚²ãf/º\† ¬wÕ¹ê7”‰¾¼ (ÿbRÅ3¹õ¯'¯úfÎJ­LV!¨ðÓ wëþ¿øžãÔ§ ?7„&&ö}‡y‹½ömÁ°Pàb/ß KrLQgÓNÁry’ô›—°š²ìÀ­õ֊|p6Hz,T$áopöäD™!±Ù>ËL GâKÁ-IUdFÁÚHºwȨ¸âŒä$Ǹ*¢u¤SCH©~Ýûº1žƒ´‡rª ’*‘ûãS‰OµÈD'o#"†Ìy‡;îWõ ëñ‰ƒÒÛÈîþ‚«÷)#EŸsç±àFѪã½fØPâØH Ãû_(­–Ò׺ΎrzøüH ÂzéÛ>ßQɼ3h×û0Tq¨xx|÷ªy\S"ýŠ/Ø}¸3‚ÚB/LŒZ!²¼ü#÷u¹k´ê®¡^ž”Ї%_4°º~ûg5î#?É¥j‚F9´Iž5ïÑØ~,>š•Ðÿì>òÍn<æ1¼h… ¢Ï&K)ÑòÙ¹ÀP=m´ k[=ª)€ÖÞÚK4hCoëxVJ mšÄ-ߤj£íCW?nBÚÊ• =¿€ö<¡ìRŸ£ýlSRù¢ú‰JRzØ£¦ù™Ï£O‘éÕu›Ç¤ð½0[$‘|ª_jï!Ìs?‚C.ûIíðhòçosz—¹5ñ¿Ê*m çDÑÁ'ÉÂðÖF딾ñEÈø^—CôvRµiT”¸Es(™K ßâŽfÈEúl’¬Â;é{áG¯îÜ£Výć‚þ'Ôå/ðÈó“Â7¢:¦®{hQæ:’sà ´ÞK™ž¨=mÓÌí©ou¡C‚ã!Må.èäÈH÷¢„N¿CI„Ìb¡óÖ&ô? ùž$èèìR©}e¶NìÅס£y^Œi&:οž¡~g·¶**… óëá—¡KWÜH’sº‰RhÓµA÷¢¦N|¯ô¢DôÄæ èû ùIJû% ¬œ}&ÑC¼Ÿ#%éÂðÍp'#®0’Ö’ó‘ÆÈÿunÔƒñ0ýª'{aÒ6§4ÏĦu^ée÷ìï®)éWðãG}Э°ßðë«á{æ‚QXšÉÿ Kš¶NØÁ ËFPN†!ü¦¹šÛwþÈÆ†Q(ƒ‡/ä—xoÀ–Â;ßNz$ òh5þ؆Ŀé4øH!i”ìë¯rÔHæ|K=ž)|بn\@ª„6óC“wfýþÑx¹F¤¿GÃ4댉z²¿™%\S”ZY"Ý¿¬ò#›ÔûÏedŸ|-Ëðµ9{yþ”;¼EšùÓO'‘7¸ÍëÔ'>`§q3zx ƒÂ܆cæQh)ò«±ùmñÛëcߌ¢Ì|<ñb‚(6TyïØp”èÞ½_[û%J“žLÀ´p”õú¥r °åŸëµ;®Ö¢¢x½xˆÎ,*uM*Ž^bD•‡,Nk^N¨f2n¡^‹*½KEY¨uòŒJ%âÞ)65É]¨C©b¼Ïç<ê4Q—PB½˜4bí¯hÀÛÓW—]€†_“àEì4zQdÒž‰Æ‡„²ÐÝGSÙUŒfßNű–¡Å åm—ûhù;F“«ž ­>ÛœˆÔú‚Ö÷õ]ÙÑÆòçÈæL´eûËÛêsmhGÖߢ]*KÃÍ·ÐþœÀ­§£è ×ö·Ùo¹²Þ+îŽGǵ© Z=týøØòXúÄfŠ-r 2Eì„öÉ×Hûóæpâ&˜x™–[’Ÿn5ä=yVï*þœ½ÕGÈ«é7o¨@ol»XQ2d„eë»g rm?}[éHÎñ£¤a€¤Ž·Ñ¯õ®BB¢°K¾F ¼<Ýñ(OëöŽ&¶þ \í¹F˜[eøíŸ€¬»Gmt™›Ãº®‘ÂUúµŒkðžñZ8ë2 Áÿç´üJú¡X#àxÿIC(nõx)¤Â%åÅôDënPšNØme|ù««y”PyðÞ "-(§¦8uLú)”åF9=k„œi·^:TÃÇØÕnI¨8éÿ讇æÿå¡úþÍ01Nø¬b©vˆ$ªn•̦ÈéBõ߃²“IP“3’i.{j ;þêß›„*בļ«Ðt[b¢Có½*wïT@+‰ò:eÌh«¬¼Nù…àçd:wèúØÅ§/Y¾ÃŸêÑsîk'ƒ¡Cɳ%Áˆù³¤-Uc0ž2¥š€ñ¹,îT+˜œh•ÉY? 3$Ô¢òžQð]kËO¾%~¦y0>ªßó™ù<_`á®n^Z;,‘d;ÌÎÄÃrÒ“š¹›°,œ›Ó k‘KÔ£â°>`zV²üüµžÍ[’ð…í¥ŒÎ÷ŽHìR‘uHÊIê©ð;â.cͰƂp$køóô¤RäÐP AªZ÷°bz¤UŠ>—“ʇôßY‚•ì."ãRtæàò(2?;­:¹ª‚¬Î îÂ*È®«+qY’9÷µF]ñAî?É®³tÈ+uq«Yx7 Hö0½uC!jÏœÙV‘qÑ™&ô5ÑÆ ¹;_Q‚_3O&J¥¦éDL,^ ìLðò¹;b¨ ¡Á·f÷•¸Üu‡«PEm˜šà5Á`Zâ<Ôð KŠÚ’F­õšüCz•‹¡§^l .mÆÈ¤Ñ"ê¥0¼¯²GÝÓ°÷/þÎÙ 4zóþÕUùy4võMŠAS’£§­bÑìeü³’Å´€ý,âNh9´+XÜL­*õ¤×Ÿ£õÓ4"ÒD´9£°{¯ÕE´5¾\ã8…vÒ¦FSG¿ =GPßÅNt +•9|™}2ÝçR2oœxfŠê¿öÔŸ6E&y‹Ó>öGá»ý—ýG¡šcÌ2ù›0| ¿v+‡G“ÔÅã -äîÓx(¢ù²ë^¯Î‹æËõ*—(Èàd>y›ô'Ñr&ô3ÁwsÇÀãÖŽ¦¯Hï/ª‚%›Øõ¥Š{¶«=A“Çê=ÔE¼ü¶vãaÞ{2çi=´°¶wê¿‚V낳âÌÊÐsce±tÚk¦½'µV¡ãÍ«O|- бl«ËwäA­,¾«iAGññLƒ9qèˆù•éuütDQ=«¼º:ªÅÅlîHþ{Nض›‚Ð'_h-D¬ÿ†.f£þ3ï¡+örçíMOèö,pßR=l¤äû¡÷ýºÕ%ßYèWc°ªæ¶€ÆÏ’g£`(³¿÷4 §þªxÎ! #ù¨,Yƈ‚úLÁ¸¯öÂfØÚγˆ¬©-u¾7‡`Và"ý—3 ðãÁ×ëžrà— íuÖ'а ãÄ`q éŒÄ;SXNKO³÷ú«'/8V4ÂZ® ËJŽ:lP¾8S}° þ&(ŠÎA"’ZÞ…f$Nî]à˜GR=JeÿpZ$cQ£:ƒœN«GC:J÷Êíæ{ÃHóÐw<EzZq»_Ç‘ñ`£XGa+2Õ º+Ø!‹¡hßÜÃdí8k2ì!é/:ù‘Ó¿Klæ•?r¿,Ù{‚üòÒsÛ=$Dþüå׿I(¨Ût°”H… ¦ë£ÉPDÉ»DæøoÜÓlðTÓP,º,ì ™ J<”p²~‰R…KŸÙm;P–OÚiW6Êû”ˆ4ÞÿˆŠÄ’­G–N R6ëï ÷„þäáum{Õ8ƒ•¶å¢úÖò2o'jÉñLz­\À½×CØxƒR£…É8 YõÎæ|¨3$¹1}¨¸³ ÎO^ (‚øÁ²‡v4ñÕ(±3!O³«DúUAÖ•‚ã !ß ß¯ßiqä\}K½F8˜ k§¦°ì†"É>ê ò6(šâ&êMþ×Ç È@ÉU"Bž”ºŸÓžÒÓ„Òn;vJ(3¯*:CAeŸ½OýH•ƒr N§ûcþÝg¹_k>ž™ðO›?÷ïy‰`ðÙØÿËC²?X>ë²Ò”FAUhœ'}†Tÿ¾Â³ÆõïžÄ¨àH¨ K2n Ú€ú¦ óV‹·Ð¤øÁÍÏç>4;šûXýð–*·jÁY:h;g|¾"‡:\$ óÅ ËrZ7ÞLJ?ÕÏõâdü7 ö/%eŒP곯¿Þ£Ÿb§ê_Âxô³^Úý 0fìwõqLgyUÇ!ÌΘ…=%gŸöo´Àò0üZˆ—øÊ ÎâŒ×ŸòÁb§9Ç É6X¾$ð3×ø$¬:{d‹2ÚïáC&¬6°^XLF±ï-üÞDÛeo¯q R}V,¦ºcR,hµ<¶ZÊ„Öý¯XÒYÕЦ~„& m«‘Dzíj5/º¼ @û¾oÕÜTÑa{~ˆœS4æDN …á¹=Ò›n³:¨1HÏœR„ŒåìòÒ)$䤛Ø¡zèÜ;»e¹ë3t~º8~D¹ºì«šñ–6týn<>2Ô Ýퟋ¾xPCÏÀ¯í“Ї O¾žœvÊú³Çé5]þÂàÞ*O}­Ã0Ôë½çÞä'®³æŒ5'‚‘ªùwœ÷l`týE±Óß2·Wå?!°‹i³Cª0UÚÈÿ(af¾é“¦dü®¾iÙ"s}m¦ì¾7`~6’•ý‡#,^wr·ß¥ Ë{›×åöÁ*ßKŸ¿©’°¦Ìõ"T¶Ö¯¿ÈxÏ éßöÓXÀvµFиQ,Û¤½ÎË-@’Ÿ#•?éqWyÅÿ&$¯ˆ}ÿP)G…?èÉh!ˆ£L¼ÙÒ=ŽoQDFn'Ùk–È\p7Éé2¯ÌÎO%- ëµ!ÝÔÈ.™E뿜 Øæ5[é‰ÛvÜsD»ÀDùµZÃrµQpƒóÓc—Qè¼öóÉÛ(¼”Àôñ<î ?m3¸OÅìß?ª aC‰}kþ'âQÊ3hâÇön”ùvêröÈ>”7VÐx‡ =¡Z)£¨tG¿l÷êATQ¥ &{â…ª“ì}aÞ/Pý+ýò°¯¨¹Z÷Œk4÷îÏ*ÙW„˜pѵÅu²ÈŽw  Þí×éL]Ñ€=,›;Š ëë{Ó;¼Ñ(J£4+©.Ð6¢)ô©†/#hÖß²òõZ\ìé—Ó øÁ˜9kx­&µÍ®Žk£ue›j\Ú¼¢ ~ûæ1Ú†Öóq/£ÝåSO‹Î ½_áñku„¼(Âu<>Cy¥§f£S½¯g(Óo'£‹Œø£B<ŒÞ¶1Œ Pág¿Ä> r¤Ñô>åÏÃãý×=3 ÏBeΟˆ»®„\ºéh|žê!äl'Ÿæ%‚Œ5›¿õu?9ªH¥Ý¾¯ê^UÌÜé¡JÄYÿÑžúËÀJì·77â­_ª—QÝÑDÿÎÛWL yH»Eõ.>²ÈY5¾ì’†\ÍsyGù8 @Ó,K›’ð­õÞ(ªÂûqý^›:(ŠüðzÌ‚Š 5ÄMÆ  D>dèM”òô|µÎ‡ÒG7rU9C…Ô”.];”vxð…²Å Í–?§ ÞÄhÞ£›Ém#-wDh©ËB›ÐW"ÇóÐþçÏDŽòèúsìO¡ÔªïÙ¼ÿ‘u9tÑËKî…áOFl¬a4àçÖ­0'ßwóÅ9ÝO0©\Ql"y¦¾§f¹³QѯÿhÂO’Åã³mÏáWÌuŸS2,ÈÐ{K,zÀbúl׈%,[¾eñ{f«ÊIÃ16°f^Ÿ›û+Ö#D£Ç¾ùÀæ²›)=ilßÎÛsQ²‰tî^‰F’ð2&[oÜEïÌ|MtÉRãK^¦J#…7¥Æ®=Hu••CÙgiÎ4¥eu#ýýR‡}º‘±âÜ—µ(gd>rýOœÓ²ò÷»Ê3$!ÛÖ£ýKªÄȱ&kýÈ-ðüP¤:î.:`“½ŠüõâÏ4Þ¿EÁÂd¥^5®•Þº©}E™âª>É¡ø“I§l"Qjß;õß («NšÉ´Žò_ÙHyQ1çRµ«’"*¿«¸î 挪MnÅåCPƒîj•Ñ*ÔʹBœ„ãvÓÄ›uæÏ”jéõ¡^h_wú 4ª”(ôBÃ1i]VQ4J§KI"=‚ƧÂ2R¼ ýPd×á|Ù 4ëy‘2õ-b¼H~£•«~ž<=Z‹Ÿè©¸Š6ÂåA6"Mh+ù#¥'†ítø ³|LÑþ„DÚ·ß"èÀ_Ç®‰Ž‹"‚ƒÍ³xàÈÕ ìôM~{Þe¯ ª*FY•P˜ £ñôŸÌþ=0k39ÆXïUTòž±¥T„9¯ùRþëÏ;<š…üùánÖß—Äþ±¶„÷qYé5ñ•P¤wtÿd¥;m%z i:CñÌü¢»–PÒ¡k[ÆÎ¥&³ôv7¡´â ;k"”©j{þ}áe™“tôjPî!>MÎ>½Imøbºç®ù¿<4ž'ßfé…ÏNqç3ê *ÊÝc$ ª¿¿›<ÏC5Wf¼¦¯A­–æè’—0Ô;U‰*Ýe€Æ°)ûšÖNø6yº¾æl´ì/r±»c­mû^ãÖ7h^{çza.túP‡ØmŒîð§úØ®ï¡3€žGŸ^SÀðÕœÌô?0*¡C-hÉ c‹gº5ŸÀD¿Ú6û§0M¬¸ëüÔ&ÌêO:è=?rÔ"ž9¶Â¯}ROgä`~{¹¾Õ2¯†»á¹Ï¿=šÃë*”híòsO¨6 ÚxvÍn¨CÛ{už¯+ö£]é9Y‡üt åì1ÐiCÇcg¿1s-£Ó¸ã±÷÷rÐù]*[Ã/ô‘›·Dåv‹'G<ú ç¡×õ'H 3ýª´mðùzçgÎ&”zëø)6ÛáÑä–ö+½˜2ƒì[̽ïÌ- ë“y†I]*a^Uoòh^Ç΄ ¹ð@2Å•ÚþDH(J¿Zo¨µ£i8Éöj@D—Nò=øp#Âfv)TÍ¥µéQ<:ÑÌ9ë7uÐä²ðàÉmihÎ{(»ŸXZÒ‡‰ç~\„Öwm:Ç[ -Ù…> «ÚÏ>¤å> í™/ò{*Cû÷½$·«B{ÿuî;üE„Ï4ú’Ð!,íPYó:›ÅY]éÿqLí >éb]¬#N;]c#d¹û ûí¹[/R ¡'môØXd!ôÖ¦| ~!/_¥VH »'2äC¼_f~™-À°éÿ³B0ÂÄò5¹(F¦mŸÛ¾€Ñ¹ÆGYU翬^+?¢„ýÙ`JuuóÂÝ1˜‘îþyL¡¾{}¼ËÃj ?¹ qË Ã¼=øëî熅¯^f™I²°t¢@H¿`?¬¨23ö™€ßQÉùRoà_´ˆ6_>l|OÛ§ãå [÷]T#Ñ[Öî%@ÃÈÞK™L¸‹…(RÑô!’3‰”ŽY@Ja¹=~óHm!ô;ÞSi_ÞW¼ÛõDûEŹ#ãÏÞ¾ÔÇÈìèÇÝSûYZ’eG'‘Í;g@BP9 )ó¼ïU!—Ë“cöNÈóîíV«ò™†;·õœEGv&2$(¸6.kDè‡þ$cQgpuQ(ßq-­Í‹âŽiYq(Ùò\*Ú>eÆ*È sPnYGÚõ©*„ûE݇Jʉ†Ž~¨<@ñš¹$U¯ˆ7$ŸBug½³³â¨yGlͯ\µ$ä÷õBt>$]Ù:›ëOîÒ£ÞÙÎÊ#æÑ€A÷†g +6­œ/¡¿‚FÑ£é}fÐØõþ É:š2KL9P¡Y_;¹W ¢ÅýkÅ$S×ÑJ¨*Ä¬ß ­©DO>ÈEë?_%†¯¡ÍªŒoCèe´Ý~6zª:í9Ö~Ê/K£ƒ®ÓÖ¥‰'èxMì­ªŠ?:µœpÙ>l€Îp6“’§.µO9¬ïSãÁ ñrM¢Wâ–TR¶&Ê¥©ö¾l¯Cª»N“Îʇ`œËYøEŸT ˆz_Ô‡w?_†@«›øƒ`zH—nii\€×Dôñ¼øžâ…ŒìÿžfPü¶­%Ekw^Ò}d‹°é‡85Ë#ã|Ó;šÈþ1l»œ‡V¹¿—B&wÛHÑzäT5{>xÆùé?šYOr»Uá߆>§þíÕ\RŸ-‚÷ó3³ÊŸþí×äú=ò…âWNo‰Ü–¡$si'ô»ªXƒiSâûç–cÒÁPã©ëÄ]Ý µÌr7ÝÕ¡žwòñ1ahÔºÒ|Žt¾Ý°ÓuÖ"…êqÑ3íÞÐz³ƒRÅ ]í…˜­úyè$âÉÌÑ}±ÃŸêŒ›p†ÇáÖdy&V?yž8éŒô?fûÀØcñ.œ t &nfŒ9gÂTf䫱2˜YPí¯|e?–ê÷ÓÂÜü¹ «K¥0ßÚ'q9k­EºæÎÁÒæ×CkJ¬°2Ã!Åp» ÖÈ“®§|ƒu}â/±Ã˰Õ1&>(ÛÅþêÕ:H4^á]?ˆ$6Þ¯~º¤m—žÉ¾pE²S>WÒ'ø‘B²aÐ$úRÉW'd_RBš»º’ ùO^ú¨ôud 3eVkEf©çe¶¥¯‘e"Ýi“ÙJG<¼t¿#GzïDrÕ/gÔr¹î˜[q Á[uŸ”ž:ê´ËGFì­B=ŸÔÍŠ4`Ñ™V»ÖOð瘳K%}ý¥£ϧçemú¡aŠƒ!í{4gi¾Að£Åz+WÕ@Z]K5²Ò$Aë‡~¾¶›—Ñ&wÑ!Ùpm§Óä&ÂÐ^µ*•è7¡RR…Þ@'ñítë6x 3ôu„Á9q'¾N@ŸW¹ž–cü¨tž¼“²Î ék(šª›`æxá{ ø´~õvdê<áwžs¶§®}‡G“+›Û¡To¯ï纅kùñÖ$Kæ^H/^“;™L©µ †_„áÕeÇ’Ïæ:ÿž_ܱ oÜÑ×c¹®DPC~÷úìåø ræ¾Ð.¨ª¾@nñKêH_q3B“béi½ $h~,xæ;;´øÎ…(¨h@ë)ý+B‹þЦ/5g%VNè‰!"µ· Ýqs¨î '´?w‚E.È´çX‡v0Bûæh”¡_uDQŠÛ­eC§'WÐÔF't\Y˨ tKôŸò¨©ƒªj–•Pèe^(Y”¢ƒ> Ž{Rý1{EkŒÖ`Û}_´ 5 …}•¨¶/áË…Û 2$0¢ 6wFÖɨ•ÕË`tzÊÖaŸŒ ¾NâU‚ 7­?Ü‚Éåû³³WŽÃtß«Q}?qøN[vèÝü¼$–¾›Ö機$;lÀÂéUÝÎXÜp7™>_ Ë•u9èÁj™½¤Q†¬ýˆ>Bh6œE6œ7dáï\Çþ¿0$reª÷b%Aâï“¢ÔÌHZžð\¾pÉŠ=^ç'Y!EÕî“Ï8L‘jŒ²?þb ÒŠ·uGú—‹œÖé1ùÌFãkd¦T“kF–+tÌD£ÈƘ¿ÿ×kdú<Ýy9ÝñµüÚ<ú*+"ì²ÈÛæ¥{œ’ LZ­E¡`‰:E{Ã.VVæ HD‘¯C{>øW¢è´xsÆÃ(~úDð×”X”¼-báK´ˆÒc£¹ÆQ.E¶Ñè£<*¸§_úZîˆJ4ÍTËyÖ¨œs«ä…ªZô~óB+Tv]É=܃šæéN¨ýÚ£pk¡Qœ¬…µ„u¾ ñIºzÇFÜ?¢‰41^Ìgï as¨Á')b4ŠSÓùèø]ºÏm­È£)]Zªà/š Î%¼î@‹È1¯Þh%}Cw/;ZsÇìquû6lŸ^]Ï[@[þrO—‹sh§±õäatÚŸÈÿµõ˜Ò$M~ûi¢±ÄÅøàB/¤^ºÏÍ‹. '·{¡kyÅ¡}x(ö có]I<]+çöûÈ&Ê„ŸçK’@Ê7M"º$0öþK0Sª Tøi&CÁSõÊ æøÿíq‡\¦†4Ÿ—L0} ^;ع_l‚äã«ëîÝý÷ÜPäñÝ;p/ÞrÏ󈎓á®Ë€Ø ™Žùg•;š0UÀ¬»z•£ïîpñëÿoï;/àäÜÛÞ—hùG™¾¾„wÙaÖÞü·W³?Cõí¿žú3MŠ|}W)¦ïB±kE‘f4”ìgôŒ˜Y€’ò¢¯OÜ: Téˆ_¹æ¿=/úQÿölrChÚ¡lñÚ¥ïÍæð!÷ådr:3THÇ9¶çÿòо (ÁÕ>ûˆT¨~UIb' ž.BõÐ|±ÕÞ5¨qz¡o!õ¾N·²8™AÝåVäG_h˜a4Ý­"ßø´ ’ƒæF†…»B¡UY¿ë£y´uz|)цŽéo4„>ûª—y%:SÉú;åF³`hR;8ï+Œ<tn½c&ûº’¥•aBJ44{±¦ìë[E|‡™ø[¿÷t½‡Ô•ónû`.º8††ÚæS*ï S¢H—­ÜkXj³¿U¼+Ÿ­íS7Xà÷cæ áóúî³®—N¦c¶ª©l —ó¼`¾ŒD5DæéKÛH¢ll™I ©Üi@2“W¹]Ï‘‚JŽ…{’©¨ã{ýµ@WÉîX‚¿é6ßs•žBF¿Â_/)Ù1™÷íORdùú°dÑÙ¢ï]¯AŽ*ä7®"Wlö#,p7MzsÐÛ\äûã=×§Ãüï}qáó +NE—lÇ=‡:¥ÏǣاߡŸ~j¢¤3SxüŠÊÈÔT|9Zˆòª‡úT¿4£" /MþÛ.Tf­ð3R¨@UeŒØÿÛÕ¯O¬¶Ø¤¢–RvÅ——+æ‚b1YF¨ó…™X³TõŽ>Ò]D5ßXѰ*{Ác‘’èÓz¯ªæÔ¤ ©C@KØÆ4— 'w¨EK‰å &“h•tÖ˜þ¢õ§?å¼eKhó‡ö“¹CÚÝ^6 nEû<ÞLMgtÔþ­)Ý.N?z¨zóÐùkù¬ÇÆ^tí–«,±FaÅqUVTäøÈ™#µ…tÉñÔfaú_bðïðÉO„…1v г<ô»îðhrï/ „7×®:Ù*&AFY™`ÿýHkzªU()+ÿ*)Hý{ß°ãRùðÿö»_ç|ØÑ×gºÛ¿öBþãc[\Õ ÿÛÃæÏ½@˜ 5\:[þíQ;ŽÈ@Õmã…ËÐì[Çú¤} Z4þÒû]Ý€ÖÝù±g‰8¡ud³I2)Úb®ïõÇ(h›Ÿ2 8» íš·/ëˆz¢10¸êTA{À©•VŸhï|ÞÉa)¡:âOŸZCçŶLîµ~è:é•´›º}Âê-¡ç8Ç+ë3¡÷*1QåQè{çø{4XéúËað‰Ú.÷bú[{*æÐMNIžƒÇ΋oDöÃ(Má•iÞa#V°?”cÜ4¾†0!zÓ•r&ózzOÁt¨w—PÁy˜Í*t<?Iæüxg?ï –´.£X`V#ÑáÅF¿0 X6¿éøT+V‘%»_<ÖŽGÇ ¿4…õ¯‰9¡Äð׆JõvWQ ]ßV®Aâ8£¥ÖS¬Hê¡+'ddG²")ÒÞ"űªÊ’¤º0W× i².|ÑkúŠô\”ÔWÔo#£ ±Pôr)2mÔÏ?E‘ š$|ƒÈš÷®à]£7²_¢¯¶DΛ—®™hø#wsà€±³òº˜ˆ­ª&¡‹ùÖʼrÄ‘ó -÷¡(,C‘»d•‡níGQÝWIWž£¸P€Óx !Õs|²4P:ò™kÊ9ðz+ÎìF-*ᤳm¨8ºÕôhþ7*ß"nâyŒª»}ÏÓôª£Úó«ìV÷P“íöÆgPÛ7moE0?Ÿûú›y¨Óc|¥ReõÌ¿ó¸EðÛŠ}‰{¶ò3×… ÑËûŸ|’‡{(Öç­Ñ”JìØô†šRiŽÉ£E¼ðÓΧ.h¥Št5 µ4#§µŠÚ(”ó mÏ ­±¹ûìÉ#hçÜàÚ¿ó^lûŽŽôψ†Ñé–”>y 3gö}öötinY&?‚s¿Z–»‡+×#n.{£ç¼[”ëG”òçžï‚íúGuÌSaÌA±ÿžï^¨ ²5óž$‡üù›37™×à­Zýï¥Õÿ½_‘¾È ©¹i1cï¶àU·÷͉Hœõ{{…¼åÏ aÀâ–®¤è]„X^Z㿌Ñ;šP”uKއ‘qΚ&B!÷ä^æ‚ äXÊ 2ÒCþîJÛÀÇbðÎc5sàmÜ¿½Nž)ðÞÔZj'Š${5c1ŠÙů±"”_XÙ°ü·_³'¯p J~|Mz$A¥î¾^ÏÈå ´›ûfUY”½O7ˆ|çn„ß~¦ = #ïþ_ºšS܆!ZšÊ[É0¬:aW”Ÿ Ã=´û¯¤}†‘+j͹ÎÜ;|‹Ñ»ó?a¬–•Ä¿ÚÆ·z.­Ù…ɳïjšÝOÁ´†FñJÌš>¨ÿC?¼ÛË_Ës³Tį®Â|>µ´ÌC&œù!Va£–z´­µþÂJe«ŠÏkAø=Ý›7ô ÖÕøŽGp¶Ã滽"²Z7aû¬³¯H& _ûÊŽ¤Ìà 2ayHF5ì’M‡d^Ý—!m{jüÓI¤Qœc0‘AºÇ¼›É5AÈÈi|€§™N}²³ij/ù°"'‡×}ƒ ä>ø¹Ú;^w‰Óß=®üñiOóòPP{¨¤çø.*ˆªyñe E ¥]·ž4ážÍøÔ ¶Ië¹¥@ô-%~hGò9£´aŒ8E¿ÊQÐ2ð±P¡‹Gc %*–È<‘“‹GeÇk#©»Qe–rwf€ª½‹X½¢«ÊÔ—oÕ¢¶ÖTÕ úKõŒ…ã÷P§Æ§Ç‰Îõ,þ$mBƒ]–ènß‹†mùíÇèÑ(qØ@P”}cW¿ŸCÓ]©Â9”†h6!ÜÁwo-^Ÿß¥ãµŸÐDÿ•u\Gë½·iöQ¹£ùû¨*³Q´=7àrɾíÞÖ¯ô£Ã.‰•uèÐ0¶~PÈÐåHBç2;¡¤‡MèM<2óÆ¥rpV_ÿ€n~ÏBxrzГãQ}BÍ ”p^N-çœBòõ¢Ä|g]~©A¬­¡í|KÄ·¿üÇúÏGÿñhþã?ýÇ£ùÏoq'¸Ê•샘‡³¨îùí蜚ÿ86ÿùø?çuJ=• ï„ÜÈ%ßCáMË©M±P¸¶Ò°>´Ã­ùï|øŸóÿðkxºÏðÕþ?›ÿx5ÿqjvòðÈ%OÝäø|7—ìǰ+T½ùqâîýV¨n—z é6µÃÍúåŦdugÌ;W~·Aƒ†ŽÏ³3Ðä@—Ä'úçÆ|+VJªšò/ÿÇ«ùÿùS;¼šÖ&’˜Ô1láÈ`Òr„áÕõHûuMâsðáí„ñËlìÆB#0™QvCR£¦ÿJÍéÜÛáLýüed,uæMÃõüªºa¡~;Ú8 –‚Ï6ãî X¹PÑ1 ¿cflŠÕLáÏ7“Å+š°I}ÇÈzÏkØ:Wåy ëÝûj Ä¿ÿDìšCÒ+âƒj¿‘Œ.“WCÿ>’W/–¾AÊ–sÓd7†lK9²<éâý„ƒ$D^Ïó| ©±þI&h$/Ý“s7^Ó]Iš@…û™"LVvïŰƹ‘ƒÞ0jFÆ©˜¡Œ‚—©ûjÐ+ŽýÔEîú¾ðÌSq(5“¥:«íÆ©GŽ„|FO’ÖÎ_ö|q51¨ÚUûzŸªµc„÷ý@_ip®f F•ªÒ¤¯‡ÃP­õB‰Ý½Ÿ¨¢ 4Þ¢†R·¥tF¨1Êî݇–qԤߴ)ÚÏ‚š–ÜÛ§—ˆšµØgÅKËPsè„À‹Z3Ô|çÓ@³† jÎÇõ:¯ù ýàþmã{G ý˜åé¾öi µlq¼Pd4ßÝXîúµ+$¯R¢­ôKÂ}â¨ó´Ñ5S¶vÔ•q}1£)õh«‰>A½q¼"yïÐÏ!žã¼µ¨/ñiDŒÀ8ê7¼Ñöt õow~ ã@uãµ!Vh°üInͤ7Æ5ž”.5Ûã¡Èe®‰ S4.¢ÑTÀ®‰&-ÓI±£éä„ÙRj4{SÈê´f&š[á'ñ“B icSiRhé&U`C´[Â'ød­µ~¦²Ì–G¿UõfÕ]C[ƒAd«RGg5~DDÐð NÆ©Ÿ‚WÔf¿¹ºè9©ÿ¨=%¼%S¤"ºdu?ê¥ôMªWQ@•GTêçhMl2€6˜ 0¼Jè¶ùÇXïç'ºDß?SQÿœ¬è8°â¹=nRü ìþ[>~°_è ;ß!NàZ©z>¯ûx\¿Ô½M§ÞI\v©2àËxZ-/ ‚žn+¸ c |¥8ˆ2;4»Ò²W˜€xL5¡ÆZ Hv„/ök}é$MßPkå­6MÐJ¹/lÞö?‚ÂýŸºOìwxøá‰…ôQ¡”–¬ÚßQò¶Òr&èbõö¨2‚0\0%ª_=i©ˆ=“ºôæPÿ›‡¿ÊÜ?”Œòä6Â~ð¢|¯wÆ¢û P~cQ¤Í¸*¸ª¨Fpé*Ô6v[CïÉ\KºNE¡÷1/ÇæÄ\wûN5³¿¬¢µnæ_ z *…Áÿæ¡ÔåýwãR›ôÖƒªå{<,ø¨nºý}·¿4Êú)ª8VsÄ }oÔ¨=¶½ë9‚&QMÓ'<'T5åk•÷5¸ÏRåj ¢Æ¬*uû/¨Yô½®‚úQ6çZF߇Zkt“„{QaczØûõ_+"8tõÞ6Kùe‚úªL—›  §×ÏG~@CF-BÉ¥«h$ÐGã×74VÓîôÒ­MHÜ»XóM¥ý|ëšåƒf÷+x}Asñ¾­¥ ‚hA§ðªI6Z:hÆtÑ9­8‰I·ÝKAk B O“ÐÆÀ³½³ Ú²6”Ô•Æœµ·´Ãó€;ôå£c!à;þÞjh| £_[ýeÄ€è ®Öo$2“¹ŸxÙ´®–w'PFÑl;äµ¢‡‡Òû9 © 2hS:ÿ¢†j·`P/¿Üã_LláI÷yâE€¶Ø…Ø®>ØдN»†¸š àêŠyxæ;!ð”Åé»Ðã_gKDòÜG=xº% „ éM_ ªhÄ>¾Ö„øØ}Ó¦1¤r¶jGo‚L®•[JÖEëŠ?B|M•Î ºµåUÍk”ßïbÄ/)«êµŒ¯·©ÓAðGn±Ó4W÷«T˜‚Ö—7Co€Nö<Ý Ð{Γ¸üT Öyõ³ƒÑƒ' é÷À¤Vñ\—ö ˜1|—ÝÚsç%Á¢¼fæqGX™àl2¦õïšÇÄpÀö‡èfº•+ØwE+>¶°G–žG¤|ƒàÇæDU žçŒÞú €èÀÚ_$ÂxºÖñ”š¯3Ž:U¾DTçê]Fgžä}Í!C9.7ü>Ñ ×øµ ?#Pú“lv"ô2Ø ³§&k'^UÑÛ7¾«¯ï‘<†»qjÓ½ëè ¾pnaö|A±éR÷ûzŸÄ]…多PÙ멸µ{'Ñ÷Ï™D9ó"¨¦6Ξ6[ ÕGQÕތܮúÖÔ9÷¢FÓ»ž0£&úvCËÔ{¨écNíf* jÆq“R¥}Žš3JdEÍ=ÊxïøÇà U~—QsÌ«x)Ò¯èÇ‘Ú()5Ôr1Ë›G¾tÇ'Æ,\:‰Ú –ÍJ( öªñ‚^‡QÇ\DRÔÏUÔu íöƃ`Ô}íyè=UÔ3ÃÚP‹~Úg%Z÷™ >'ùšôçÔ¨Ÿ£cÈîWêð¬˜!VDÅʵo}Bƒ_…]ùébÑ0þW'O)Øã!Û»K­8=hœ¸ž9‚Ñ$›ãl1?š¶~ß¾Ô—ŠfµÓWuÐ\å­‹Ï~Ä¡ïØúW´djø]òŽZ1õ|ä$ŠÖ¼Þ³èÅ¡oÖŽ\E[6ÏR« 2ç%­AËõ—€çL$·$º’¼îeѡܖÖl@b•A—5,È(ò†è1 hZ\ð»•TÚCó%¡@=Ôs(<Ãh¥…T9l®Ý«,iâá8ž+÷ `:acÃû¦XÒ(½åšË€±yÝCõ=pæ¥Äþx\9yÑä¢Ï€‡q¼¥"RxïµtŸO~ÙÂG“Ö~ ˆkôï„<“k¸o’çÃ#â¿¥rLAœëæÆÉŸ‹ ™”Æ®í®ÒzÔ eT 3ËT|‡Qäˆ7饂ÂE±„’ ;<Œ#I88Û´>mÔ(@§bYa'þL^,k]³ þÀnªecЬ¿×+,âZQädÙAÇ’:ó&{!èQ›ÏÜýÒÇ«ôÃWÀ0ÙÁÙí%€ñÒ“9Å`*ÈæQ* f|±Y7¬,À\u4ú+5Xœ Õß ÅËg$i<`µ–©æMî 6îEº U«`»ÜœÀÁ± ö©Ê¯„¼jáð¥×‰Ìšàäo{8Û\ Â^xÛz‚Ç©›©;ÿ½ ¸ÿÓ11: ©s xÊŒúq}õÂL Ï,ŸÙú ìÎá+moÞ@™bº ¥ŸÐ ‡ ¯_C©Í¶7f˜õPŠ u„‹ñï]ß÷nqõ"VoD?¼î½ß‡Ø¾qnaÏ䃲ÓË©ÿæáæQº•þ(”§÷0ÐW åf~Wíµ†òÓÃ>j]A\â8í–q¨à3EBÁÐ>Th7‰gÆb —«L¨žíøÅ¾¼-sNô!pU¡Jçú¨IýÃ?úß<|™‘=ΈJº5d[ßîñ°$‚õÂçÝûKï_ígE×J/œ«¶D•|Š'¯EìÖ3¢š÷Ë 1õƨ®ºÖ4®N~çóÂçafm j,·ðóâÙ—9¿¿ýŒ~´Åä}eÂG­­¬r™]GPûÆvr'«=ê Uò=Ýß‚zun%–xf ¾'ŒJv)©hàÐÇ'7“Ðw¬òê1j4b.‹çôü-K¾}4š¯áüJ– FS.>ÔQhf¸D™tÍX¤S‘UE T¸²â‹OÑâVw(mÇW´ÂuŽvij­¹>£.¿…6>:ÿ:×€¶”š—å Ç WÑ1<pÃGòÃ9_­q¿‡“;”ôÞ-æ©"o˰†+õ@r…ÔÌAá0µÞ³¶‰Ê@ù'9ZR«òCe7ДÂYi§B ;Åõ#{äÚ²ð7~°›Ž€>°P®8?l§Ÿu¸îÄÅœÚøŽMá[ÀÕ¸ }½¬x /«(;é_Á³ï ÈUë”}Ó„ótCêÞš‚èÃÆ†oÚA¢Vóž÷õRúXi“U[ 2ŸáSNÍ ÈýÊ]¥§3Eã}òf:ÞpÂïÂi@ÜÔñä ¶ –õN5ú‰h˜}\™á~ÞèÚÏZø aÑWAç[›]L€èÏùÊþƒÏ=þîcž`Tò½%½ ¦¤lsò²r`¦~—C©€ÌcÉŽ6l‚%‘­Ö£ƒê`W6ÏÒ66¦–¾åq³`'ꚪà ~þ޹8^P±Ò½ .—Hm‡ át¸šÈ9éMIûRJ[äkº¿v|ü Ä¼$á{*Tüî–A¤Þ”_Äß“FϾ{SïâÌ9”9gõøjéî=3EÌËUÿëñý]·ëûj†sTþwoâ¥a»8UYó«ân¾fùðHöܽ_QQÑûz£¯«[6 2ÃÎ7ê¿¡ïH?«ÕR™¯®EõDoEïßïE oÇ:3½P#õ·”'´QcoO¢¿j ®'ܺ´ÃÅÜóG¯£¦ª3Ä…‘Q¨)sóþ_äιíÓ}5ZwŽ…é æfÍަ߻}¨Ú6Ò¨Ÿ”“5Œµ)š~ŽÐvCí¦"ë<[¨Ã»âcc*3êÌâ7°?Œº¶I6‡öw£—·#ÞíÆ¨·q?½Þ£ÔÇU³¯”Wõ͈L‹K…í;íyZü€ –¼c P~‡†‰”Ò’ÅöxH1Wõ…ÈývpŠ<úM t\ëPGÓí_f¶ѬÜñ´_™h.q®¯ûËZÐÒ‘J@KL4Q!¿Ñ e Çe´Æ°~´§m ;·xmVGIª‹€c‘<<x„ÌËž/GÔdI•À‡4¾•˜ ñò%û™Ñy ã0Tአ.g†=sa_{9‘7P§œVÛ&š!ƒœ¡. ³$”üB’ ä'œÄüöeÍg:é—Àbc&ØÞ¾šuxµœ*\ÂÄ®¬Àå™ìÏG}¸¿Ów÷žu^¥ô„8’Ëò-ŽÒû+ÄJ_׺B ß,']AdzuDû±)ˆeéÎÝ.wI×rÍŒÛïAšújê>± ùt\f¸Ù j‘ùï½4AƒºÄWHéhæÉfݾŸ Zgä“’7O‚Ž˜PßgfÐ=!Æù ôŸÆèX2‚aÀ _dÁ06³`¾mµ¦¬éšå]!`ÆÌ±õꊘ˳i¬½ çÌÒ³L`ùäŒá~Û}`µèé:$"6Çò×Ý%ÁvúEßEn°›<å¼ÃÃc¡º µ…àdiï#¤g . þyÏ2À#6,úLð#ÀãruUÒB•«b‡‚PñˆY…}%úD,5íVlÿ¿ý1*©(#êÂ…ëžB»q%›ÁÌîüýÑbÇÖÿÕ }[cç 1zK¼;a+'ŠRß‘òü*ÞÉô8TZ ÿäa.õ¦…C`0Ês”ª§@ùºl_^¢ük¦û8“«v>xK2e"PLUÝOTðê¼Fyã'ò$<þŸó‰û3\vó¦1¹Æ)ÿæ!G§ÂㆽùC «m—GsQ¹èù¸7ܲ¨"}›ËTÎUêàèã>£GÕÃé©Qœ¨fñG.ÿ)´óy‘%Ø5R„q÷¿°ü±ó9ÑC†š_ë‰(íè^žð‹ÀîýÞÕ‡n¡öy±‡â~P×áÙ– ]ã}»½õëQú¤»ŽÉ4.{†¢Á%+#S|!4Â?SêÓõ]ÔþÊþÞ—v©Uß?†¦X¨ÏJf¿B3‘t÷óõñМ쓼K–h¾UoD ïZ,=9›mÑ–Þ-Ð뙡5‰AÉ IY´Sóát“+ÚâpøÄ¯"8.%_^ ›n‘ -•53àó>Wßà‚ÄU=<Ús@dFf¿á$v'ðî7èYQ÷Iå < <ãï@s>¨…¸…v|š|ÜBæy ³»Ú ÉŠ–uMÀ8á+¹.¼ÌKýzö_ÍúmîhepJËó”n×wóÚ'ÅîÀ“e-Nÿò6ð•È|šlAgÓˆµÃ œãŸ3£ä¢ÏcýsçîƒDozb¬ H}ùùŽd*D¨K'-@n¥î.×37P<š®¥XureÓMÜ$ÊqÖܹÔÞ<›¡½' c*Íñ³pˆüp~½÷8hõq¼þÐt:©ßöʈ€^çë Ò¡`Ð÷¥õÎÌg0š\î½ë¦_~3OkÂËÓ´`^uÛ½¯;,µãÕ@˜¨|¼ ؤr½!è»;]¥?¤èÁ!ÍbðG8.˜,ŒÄ2‹J÷±WS‚pê±tª»ê2™MùîªäÏžž¹„¯‡ú=i®Yî¼_Ã×ÞáH |Ê,›Þ(ûÒ!^/žF”ÑÛ§šÌÿ½:QÍ0ª‹Ò´SmE§ÿçû®ZÝÝ­ct;½<ñ¿8uéŠ÷n¾Fq¢"{>ç.1ž~:Œ}½‹)VF—ëQéxÅ·ˆΨ¢(ç#¹7ªþÐ+tªÕ˜ÞJzô5àŒ¬á ††øâQãËñÔWQ“Ö„V·íjòÞʵËqCMž>êžoP“ ÇÔ C*j ô÷£EM— Ø“qªvâDž•ÛŽÛ»>‘aùò3Ô¨c~¿ µf[‰Gm-§T’" Pûo‚vZ7Ô)êó9‘S uù¾,Ï⺛TuÙ%ìÄ‘Q‘*vfègyéÓ\O¨ïË•U<_ ÔŸv!EƒüônÝ!@1ù6,F-G~ a’7TlXŽnIìO'_F¿JGS}ÂÖÑDºòlIž<š*]y9ÎQfɯ6  @sgl²Ž£ºéÒ©þ:´Ø[a?{)-”Å ë.E«yñ)~h½å¬éðÂdûù¯pˆñ™Dµ”7ãÀçöÃÏÿºWy·>^ûé!n,Äñé“.¶É@ZE6»/(èªÇMM¾Â>§fÍn@í¼!¡ož 4¼>èž—Ú­‹ò¶Ó€>§Ê]­©ó4xñònó¬À&©9°é1ÊÇ·ù7>KñÀ%të¤Ê¢pûî® ˆž¾ »j³sÀwº)Füá>fÒÃÿÝB"ºÐU< "þ¥ 7Ì@L™x4;³$ Ÿ¢~+ Rm‹–¶Wƒ@æÆÛÇ— ¤ANÃêš;§ (HWñóš÷ˆÙ­½ƒÏ‹¤ì#EIvU5?Ÿ×VÃ2 þ›€QÍ¿4Zýg¶úAËð¨lTX'è©ÓÏÑÙn %qÓ™[ [ã[š †÷~å”?cÙG•ëkò`ºO‰½Jº ÌÈ_˜T=Õsqçƒv:Êþub`·V$O²V3â·Ìu—ÀÆñM¦±ØŽø TóÙ€ýmn‰×&;<4³?exMœ”Ø.ôË€ ï…¯8¸àQHâ¶|C¡&ÕÈRèx¬FGP/û‹w´•«;úxñeÖ÷z{Ô÷yñn^†q„é>z^ó2”Æ=‹§r÷Ùñÿåcb>¡$•ªc-±zKX`ÇôòE Ö†—ƒßÝÀžIñ×M8bÿÍC¾¸{ÅlçPÞÙ:‚f¾f”/Û•ŒÆf×'Ö…üfÞ9Y®ÓìøÅõ^ (uõšÊÅοHÉf|‡ ÖG'<¼èŽÞ0ã]¢B;G²ùt|οyHO^¶œ{ •*«ɵ=¶6f9¼Båz?u1BMkÅ“ÌPeH·CbéTm[:'Ë,j%rÅ ˆ%P½úímÎmmÔ{3-©ûj\q|#Jš½%e*Ÿå¡[íVålú¨õ×g—n´ÿŽ9D^íbC]r©³ˆ½õ|ÞŽ<—Œƒú´R•9gLPÿ¨íˆo¬dpN{hx-ãj M8ÓÙ\níŠÆoç=\^ÕF“/eJÑŒ¬ëŒišÃ»-žïæߋ㬹†ÃMj—Ðrêj³têK´ºþåó M7Ú8­tË›|má ^¥+ÙÑ­ùUÝùjÜÝ2ªºñÓ>‘¯ËãËÝ@àM‹§…GDyw†ýæ€D¡4·Î]ÈÒ|8¨.¥SWPBÀ2Pïÿ}ãÓ¯w@óªl{{'¾¥Óê"vúíHæ¨Ë-ÀØ|n-K,˜{¾ß±%6µÏw<"I“Óm4÷âŽ?,ÊqËÿÈ<)rû§ç€ïÝÑVF$Ó8î® ÂoÞQ•žÑ\;¡”GS 1K“š;§RµNû=¬@¦‘þx3yÈþ“ ÅËÚ–\F¯á ì˜Å Â|@ ":ñéú –“s]4l<¿h¬}‚CôçÊêòAk®¨"ü1 èLë“ýÚzKwe²¯ð!~ªdî %3ž©¸¦ÆlG*¬ÒÀ,ž}*ŸŸ,pöýÄ–agæâ¶¾‚µ¨ö ûp°Y™‰<Ûv“2üîœp˜RNéì\pr!Ýßb$.—4?¯/=\~Î#ºÓ @šÅ^WðÈÖ‹qêuÐOË Ã}ièÓ|hZg~ýN|(+vª]cwN÷ä/kI”^ã^›R)‹^¦·Ó¶UìÖ1ä8îÛõ}´œ«ýÿ«cèœ:¿§ Ô½ÛÍ×\ñ{y{îÖŸ“_ľÞ›/v u¨ôÆ»®¦{T,›ÅÌl¨Zê:¢Bµ ³ãé©N¨þ:R¹Å‹¢¥¢Þ/§ ÆócJÕÓ¨‰œ8¤:×5Ië-ºáé£&ªÙ±*£:ÔXtð‰> Æg›ÁÍœ}¨q,ÛÉÆß5å×f(ýÂGÍ퇇ûâèе…³©÷Q«a,»5j³Ïbg߉CÔ-D­QGf¥Õø#ê\ªŸ½Adžº5+_•® ž"<" †fôÓ`ŠÈP_t-nk 꿹|÷Td8ˆq¶ÍN_@ƒŸ*}Ån, ar¾<ö@Ò=Μ¾xŸæ*ú•|aüØ+4qî‘cº$ )_ƨ!'4S¦¾X'*‰æÔŠÞÇW£ùDgû{xhñAª©»2Z>o¯·þˆ­Z)䟤ýŒÖØ’XCÐïw4çµõo íÓËkoôˆ× …iR;>ÿݸ€1 äõß—¨¯ Ä*} •Æ@z®»òÙö; §àúQª(…$FÙ5G_Æ G ÑñéÙ¾´/>\~ ôöâ-¸ãá‚óŠ'ùþƶK*+°Yu£«Ãä›Üøu&Ø_>™šñ–¸EÜ&(äs'Ü’y+¸øH~Þ1)°þO¿©¾s‚àûj‘jVb!»ýáìs-/ïi±‰!R¾‰4+J:AÒ¦2–›îá… ÇìÖÃ|Ãh™£EŠvx(©ãMÓ:ƒej¢2[‘$‰Šxj'•ŸÅ:ð‚ú¨v®í*Ð ö1´pç-‘ÈÊà}F ýÓNß+ÐMȸibúÎÑù™R†`¨$þº­¥ Œ™F®¸?:¦¸Kµ¢`º-ž2ùèM.ÛÔé…w(Ýúû­RÜݺ¡xRkJù=Àè<6ÿ× 8гèŸOt¦±zÛá_úPw9J îÝñ?˜3é–”}ÐÛOÿæ¡¢LOáQ”êà—€ò¹µÆ^%Üåâa›(oJÓbÎkçëÚ¿ª[ wüc`Yϓݗ„[#t^ëW„q¿+ Ý={c›w (M”Syoj¦¶1Õµ$ y|]gxI èäµû\£Ãþ—ïj W;0~ô(u‘ræooš­œÂM¤ÊmÊT8É3ŸMW-~]~-ð<¤§©Te¾´èUÇ_ °uMVµC„3‡$ B@´„ýG´®6Hâ|²þe¡R­B׌‚ãA¦s«pð€4ÈÓãV*KÅ[Ljªá Þ~Ík^gi(pö-~µÜkæô~j a¿ ™cwq&ˆ \mÛ] özá[epøã] épª)5øJh.YÌ[„à^Égg1 ¼%§«ZÈÌ€lÕ’êv¢êªûø²Z½Ûö.󣕱²v¿áßý¤˜þ¶õ•þÝ׆Ñæõ0}i˜½gåÅ1ëe|%¨ê¥È§¿Ý÷¹^^&ª'ÊôÝæÓÚëS™º±sÛoÚ8+çm.Šsrì,ĦP£½Sò¾•Fì>4Ìã&!§Ñ‹D¨Ùx2¯ªðúqÉ-¾ôñî=n8Ãc÷¤µQÇÌz£1–iåçó¨óœŒ{'ÅÔ•gMîN!ƒzd/q7Yîõ£ã$Ï ™Åî?M–És@ƒ,EÂÞÒ¢aÊAëÉ•=ö½ª d¢B¿q$k.¢ Åͪ{ºhJBΧ×~ÍxêÄž@stüüá×¥Ð|¶ÎF…†D.m£/”hR£ãn¿Âš¾F–Ff‰fõr+>@󢼿÷¥G‹”dò·Kù±º]ºÒ^ë(†68èÄnñ¢ÍŠÊÙjWÀ‘Ç·»±ÏpCSï³”‘^ ÷Ó†hc ѺXÕo„mŒ×ýè€xÑŸQ=‘È.ˆïW´Jˆÿžy¿¨©ruŠIoM´îÝÿ7@Ç÷rÈ-è›OtWvã³c·èäösÖCfóì1`£oÓ*Áº¨’‚÷ñ,ªÝ‰ÿ’Úd“8P­DÎ}é;‰{\¼­{Ÿå!çï0|ü3—Ñðê½Â¡É“X~bÏ á©Ò ïÿð±i9¶áiv^³wËIÌ~à_¤qD/VQ‡~ÔÁÇoìÍeE]ÞtG½ÓÑ™l1Pßúî¨ëOPóëûƒ£hðõwy4TÖÆ½f^>Ùš*„Æ\¯šºÕGã~Ÿ*ΠI³$BŸôU4íêÿÓëˆ?š5Ÿ÷°fÕEsIO¹²ÆÐÍEòÌGÏÑbN\ö4+ZŽKØð¾Ç€V«+ Ô>ub÷Œn žHqœ§’øOyÉ~Dpð7#Opáœl£O¹ кT5™¹aëyã1 ÈæÈ¾èT¸ãG ?ËugÆ–àÐ2œ¸ô‘è*ð0«CñF™·)0{1),+û*•}€'°{¼ {ÔJ œÕÑÄ4EÀ•4ìfÒÜ¿¢Ç‘[ð'È17³ßЇ8S.2('×Ô&Xa‰Üâ“p ?¶iä!!ˆ×­n>\è)Öß=¿ƒtOÿòDZ ›ýd߬I)È¿æ°Ø¸¿Šeõ!yÙpÔ#q,J To4¼ÈÒ.5ž÷Äæ†¸ þH­Ó|º4EúÌMáPEçhŸ»ÕË Ñ º‚škGŒ¤Aowbµì)| ¹ŸÒFñ‘‡òIH+×D˜~P<êGÚfý ’÷c^ƒYeoÌXêx/¸DUÜ£ß*×üÁzûEi5Ñ0ØŠTnT¿{†ÃñÔÊÎàPAæx¼² ï·Ñ(ÅЃsð âT¼ÏàÑú1VmÀ¸/÷×I2! íòCl¨ý¦Ã³eNì|Ãßsþ½ï÷ï8ô_<Äè£GÌ>á7ª—U’‰ýwÞâ/þçü›Çø›‹Ä"·ÉÍ‹°û‚‹ß1 '^ú'¿ñ>©‹áXÁÆËÿœ·ˆšb.ZÊÜû|ø'cãß ¹ŸÌ jØy«v§ŸÓ¦ù×°ñ.FÇ?û]ZãÈAý<¥î'TvûÅ›¢¨:ÐÐ k¹°^%4’¸þ ÿ«i ¥|íŠ&D”(¿øíÍ]`ö{Û†>M/¶G‹ÒäÑç Ê÷xh%‘4u ­¥¤ké©C›zVwO í¶£ç<Ô«WYSç¡à,jïç˜üŸEO™™€ð¶o~;¿ Êq{=dÂêP пª ×Fß,uÍ‹cÀì§¶òhê:°Öi(J6LGBßíìúeàºíêà?<惫vágï„òKu¾1ø¾AÆ3q „o“mdz‚háúgq)uXÿ’ûD¤Kz¤ò¢@¦•8ÏÆî%ÈÓÊOE9ƒbdø ߀ƒ:³ófZÉ tWeøÚËóƒ<ÛÑ aÒ¸ð›ì!bærò¸s´q¢B2Édvüá„§RÐgx5ÿñ©J5ölè’ƒ±•$86QihSU­¸éŽëäD×Áž±lw,çq ¾¯ú‚uƃ_.„€íM_|_6/°¿S'yê.Š*˳göÀÖíâ‡àÒªòCÂ1ÜUgý²ëÞ”d«1Ót s=!HN်žyQ\·†}ßböe§?—QŸþöŸýÛÿ_yˆÑ]Á™Äí`'ulœX¶ydbUàª8Ë;È4kH iŠ^£:6qs’Ð~Ô ³ÂFæƒÿÃÁ?ßÇÄ‹õê·Ç¸Aõób¼ûQÙvµ¨ö=žbæŸ@|~5vžñǽ®:¦ uÔ’8I.}—wÏ?ª{,ìñ@=}}M÷GQ׬6¥ˆóÞ¾í?󉨹Ä?óˆC‡ï‡µU¡þi%ÓŸø»}5I¤ÁšØ½Û«[-m>Qhj‹²Ï`~Í25æ¾Ì%EsƤO.‡ ù¬BkÒy´hò"Îß÷Üž¾ŒÊÅ»'…Ðú§¢)Ê»hó¤Ç¢Auàò7,FóííÝ>­‘V4A „¿mìe€øÒe’ãç84ëe†iI#Pà,3<^å…}VÊ•¢µú@}\ëL¾ Ð4šŒªŽ]è$kªî-`ðæå=Wv˜Âé.³ âËc»¤(åWÀV?qD\Ž 8Ù䤯~š.í ?^·«wd•£RðLå9ã“L_`D+N¸^¹BÏpïÿ(Ó„Ç4lhjƒ@üÎÔDDHn˾7x Òù,îz^† ë¶ùù¼’È_P[gÔíÅgiõ/?€òÚôÃö£S êC .¬:zUÁ#håÖ4í{à¹-{Bt1È;ßÝs¸‚:)®QMcõåáýýÍÁù‡ƒ4œåiþsbž‡á*&ÎÅüžìñák#€rKÎwþÄæ‰þub¸ùŸøçß_h÷Ëðe<6„™Kü4ð3îM$*ªâKM»ˆ¾D”ô?2§ÇÎ?ÿó0ˆÉÁkLÕU×äS~¼µÇÃ?ºÆÌ)cç%$ãϼÝËmؤ˜Kãaç{st :‰úªŒVEŸ£“<ŠY:-hˆ;¦§dÚœºìE¿;7E’2›†Æ=Ðä G-4ÅåuÎÒMÍR{æy¢¹¼´àKzwdé!Z:r—®ú^Z {pÓÀ­-:–Æ ß7r=Î-É£m™_ÌÝ2{y+¼J’g׿¡íaGé~ LâŠnrõâúýôÏ_‡™]%{Ñò% ¤¸ÁVT]_ ¼NsÍq–Tîä@;W×sõ-Ð?9¿ù2¼N•œw fs«Žl% `MðÙ‡{8¼ƒ4>®³¥6b„ñÀsÜÀ•ÖQø¢‰Ž2ÔŠ yt™Þõn±XÑ×Ûo¥YA’úëy4HM‹f †‡ƒLÏûQ{B_[jZ«¨E©µ¢½‡@9îçÙ#'†AU8‘cYF`HéÇ %P?þNçLQ.hL´P\6€Cáýz¡·@[¡~ÿ\þè¬7,I¬<½²-7¾·`—Tä«FKÌxTÒ‘`ÊÝæ“¦ÐfÎkméy`žŸE“Øz,£ªïš¬Uƒ™öñi°9Â8>¡ãvbËzðÚ8Š=Å}ÀQ΄´Ã$ÜånÛ-U/µ]°|Ð:.훼ï…:[üˆ;íñðOœˆ™¿ÿ›‡¾ý·ÁèíobtŒÑF7˜¹cÌû¿|ýŒ÷4«<ª é*á]ÉÙ­›[ç-½ßó‰ø†åâ¾a9ø‡{õþ§/Ue,îñ“OÁøE OÿÌí6iS?¾ú}O?æù[ŽØ5¸>Ú›ãÇøÄ”…‡ú ¨ëa¯à=§ ÔSѰ­7¢€~®üiÑGýjèà }®Ýù‹¤€Q=jžãÞ,|…Æt~Ç3MïùC—ã¡Uq}hZgF]¿—Í Äe%öÚãá«C,/ãТ.ÔŸ “BËŠ¦Ï×|Ъp~(9ŽÖ·O(FŽ6ÓÑoÓ\8Ò ÙÏ7aÙÉ”ŠðùƒÔ~΂±#§…öóQ·½ îì4Ú³'ÒžòÃ\ÁçWl²üw¡¶!PK<ß§.׺[ŸQ8¨tŽ:Víû#AWëòs7`’»˜–´ó~d‘ìPº•þجÞ;ˈ¨Ç›Ùì8ÆàâR 8Ü!V× ;€g.–ê73+ðÝD”iB?Aà¨:o eÊ>Ö$ò›MžþqåÙBÂ\¬-ú”À6 ÒwüŽ~YçŽë;‡¼»£ºÇ'aPŒ{¨ã8¡ ÊSil-ïAÕ#hÑXÆÔˆÃBù\ÕAýŒDÀIbcù!~GOÀ¡Ë¼­ªLÖ ÍrÐV×âèÔ}‰Íº‘zTÞ‘ƒ¹Kš}Ù´啉Ѭ%Ù[`úêxpqʘu(n«1‚eÌW .°´q8ÅM<Vùó&döÁ`#ns)Æ l‹Ål-ûS\¯ÅCà°Ì¸üV`8±…{|Ì Z"Ó}Œ/ÀƒJ­¹<<âÂÀ§Õ@~\èú¯É7¨Ãàd}h÷¬>þÖß¿|áßq(Fw(^p<_bž}b^ÃCLÃÃw Œs÷?ÿ‡wÿáßÿ©Oü›‡ò§ÿñ‰˜ùü6¡:ë«zع|lÞÔ!*âKÝ«óðO\Œõ‡§…ª%íÿëý­Þ©'ÞCwÒosö* înRÖGÞ{þP„“:±õ!I‹µ‡¹vhx-Ÿm“r/žõÅùþý šR݈à½Ñ…fÅ6›Ÿò|Gs•¿Z†%Ѓ”×÷Ö¦ÑÒY'ÎCJoÑŠwwÎ]¼c{sù–EÚ)©Õh{ßWV ²}{<üã1ùR"r‹Ð§í›@"Tó¡u_*)Ð (hÅ/ûKƒ‰«@uþþãR) ü¼~±h_\ððúƒÜpQ`üšÂA¶˜ Õ†P}3›T^jê7†‹bAýc—äQ¯Ð4d7’„C³S~UFß@;=¢\.Ç t}´Ó%˜I@ßÖÃmZÂ}/oúàå칇¹`ê›÷Lw¹ÌŠ?žš{œûæÇÀ2îûÝÈѰ–Xb± ;6#ã*ž ©`Wìárçì pøTçQ^ ŽýÏÃø<ÀÍ@U1\‚ x¯}5í duý‰Þ¡Ž*ÆÄ&Òì=Ù—4¦gG«±õ…qðï{1þæ!FŸ˜çaôѦžPF‹óAËþ ö}_©½ýè Pö1÷Ö0^ÿžbób/ÿ‰ñ‡ø†‰±ÜÃ<ÿo~b~þ_>ñkþReÁj!yåòí5j5Lñ7GmÕ÷¥›ˆâQGXõ9÷Ð=¦ÛM©æÿ›‡˜{.0sÑÕ¢p’¿£»Uއ&+j¨Ïµä¢é7α"·ƒÑì‘ß …«”R\ŸäÐ|ÝOr½üÁ=Èúöþ¥#´*É•'ŒÖoäÉ-ÜŠE› Åj‚ªh»„éÐëÞ³€«å~\5ï2à5Ç žÐ‚Æ$® ºZbÓÀb$7×ÚD“ìM?]d ÿž¾,oÐñ\ƒûß¶-å ÐGË?£þ™´—qç§]y ¬!“fžÀÞöšQ5åì?]$$¸MÜt^åNá´ ¥÷ ö>­¡?« ÔC|¼'&ãö¡¤ÉˆöÐß|qË^‡ çjInÐsGì}~`Sä|Tg œäýzöÀõÆ€±å ðÄ&½ÙwWø>ôRx2¯‚ JîÜvÙŸÓq8 ÄŽšÍ IÇáJ©ý -þæeÖ×Qåbi‹S òšcÆ…!" 1¬ËDy”7n$ì›ø ªìGV²€šØ¶Aä5P—9hDš²š’¾“àP×›O—âAûŽ´©y èÚ]7‰Å}ùî¸w!t`Èžé2ïÕ‹‘ì³Û³—L‹Äíì`AH¨mS¦–xª|_èÀš¦§'øÌs°ù6üá™jØ%V‹8Û‚CìσÇ_òcãcÅŽ;àæi³è.¯ =Œ­¡“xõ‡£e|%ıñ%öÞ6ë|ò ÜØ½ú!ù—À…@ó'MÒ£äüЂóÊíYl¾t™ÀçCÖïh- ð÷Ýžoh£p(ÈF„sڞʪºü>ðôÚOœ ‘|¯ßR/ÚðFÑ7‹ÕD v0ûP¤ ‹†¥z€‚äb¢$U0ìc‘`¹îÔ2ÝèÚ%–=Cà¦ÐÕ¼f?b2 GLÏŠ‘¶“ƒÁ¡d`©Ð×ÊÿìΊ~ºJ„Àù\{‰$¸úífÍE‰öò0õy7ßgý˹\²U¿þHÒD¦ìÏëI‚8ëòâÝ£= ADžªµ9 R=žR«xû@&œé¬ÈùÞJ}ä wMí…ÌžRÛ9›  "üÔR[ó ×Þ±bì –ù3-÷ðiÐ`xà­$¬ šQIº²m2 Å©h×=ÌÚŸ}˜êŽ‚®g¡Ø·´Ð›³=™T¿?qEù‚Q{´#žï<˜_àòx ¦ÛœM‡ô(Áœ'©þ’®XØ)îRXËçCIÀšXäƒè° î¥v;º$–k©ïÀ¾$B#»Tß¶X%ï™§K%œç<·ãÏ/̬Î÷µšï#çÒŒó’»Zêø¬­*¹Õ„åMFÔy`èÅÆ£çaþæ!†{?þ;_úwÃEŒî1üÿœbcxøW¾sbêˆócîo,ñf?'W »ç1<¤"ôÍ9øøw›ïÁôÓü¹¿ñ?yÒ?÷3v÷¬ìêñrõ1|ã‰íÀöÇýÉ—i‰7rÃæKGGÕN>ÉAã}Gfòd,Ñxiûâ-V¥½~8«7ÝÔ‹ÇÐBx+ƒ/ÿ6ZTë —>@…–“ñ 2~t¡Õ›´j·î棓Z¤au3{u±.7*_À-|oà{™z¯Îá\@^ÞkÄÒ*5y]@jʯž1wÈý˜Ž|²…}ì3%éöæ@}™f:\ÒhåD®†Ž™]]ôS “ÀP«ú.ë&!0‹Zœú-S¬Ç®d^Æ÷Žk#ãöRªÀ¥o¨à©ÜÁöøßņÍ;×€ÿþg: w@˜EˆïëÓzåþáwäŒ:HXÊÆ=du©Ö|–õê £Í|ˆÅ- ä Îßs…ËÅ?¥üïèó™ÜëÞ ¢kã÷rÐ/šð2ÝqPë¼C«´y4,'Êó×@s@T‰yë"h]}g'(:rÖ“úI »&øùÀ¬è×Dà‰|Ãåat=` L$ô餼šÁ´2zøÃã7`N»t¾Ô,.å2w¨ƒå‚OßÅç`}'Ѳ¾[=¯Íè °À c„Â8—ÇäÆÁ]š~ù‹öÎß=~³Sæ9’âÄŸ}k€:O›T,«á`ã:Lœøw_ÛßuBÌùwÞ£Û¿ïKü›‡=aò$˜¾2Lÿ–c˜úßÿ‡}ü‡kXþÉ›`xˆñ•˜›7ÅøÅ?ùÌÏcûl:x3 RöúPÿÜÚÆÚâÞzýÇö>àícÖãC}Z)=vnçQ?=‰×ã{õC ™D›#D»¨¥ü¬ë.ú•n OrjMÌ$?a»¼×Oãι­0­ƒ½q>txyÄ-*7èk5FKLaŠѪöáÒœ7Ahå7<¸øÍÛW³¹òqÔ{p6Håxƒ®ï•ãÏ€Ýê)Åü†á·Æó×±)‡å6Éy )–ö\ ò“ŽxÂ:@ñØí¹i9P23}=Ó4oÜOËFܺ ÓŽ }äÀ ›gè~¾˜Ì&ý'ÏvóOÒùÛ…ÀÖøÃâ ŽpÊrªñWWœôpùó àAÉ"sñ€·”Œé÷=þ"…W‡6AˆÝèþa?.µ íÙ¢!š·×†—@RGÃðHhH&vÒzÞ2V IÃg\@NSSÿ:(X¤Ò©Ô‚Rx÷èè­28ØÏ6’ËÈYùL €ZH@u:¯¨ÿ¬‰ïÍÍÃ9„[3phø‰eÙ”7hãôÔŸÀ]‘ŸQÜ_sAo¸ú=ë0x]äeïðŒBýRß$Éñí²ÎHY0­CýìR‡ÀlªÄJàæ8Xp%œLй–'·ÚÏÓ_«ok¦“É`£ôn¤7, l+ŸàsÞ{/ªÈrõR8,seùÉ`,81òš0×p‚{³ƒB„Š4pçom|•2ÕdÜeD¨Có.®óåA,‡0úû»£³¿}à¿ê†ÿÊ×ü‡‡çKÿÊ“b|"Æb¾ýúß>ño.b|"æ^ÓÔÿî«Áð°"}«™Iž{ß1&nþ»/Ó'÷1÷æáTõßÿð1†(C‘ÌS’y;Ë:d€­jȽãZŠ5?Æ÷~Ëe…­Ÿ5Aåí^Q×K" æìê~ä\-hл뷸PfÇûUý2ÐÎ%s½ë¦ºïÔßÕ€>QµÚcÛ50( ÊPž¼†‹®L--´ëzêDhnÂ8˜´iŽö$NÙ‰–³¥å`Áó¤ïcX®®æü(} V+_"C^‡cœ¶§îŽ:õ¯÷åt€ü¼ßàÉVl˘ufÉòò±yößò¼¾K”R0èjŒ@ùÙ9Âoˆžˆ<ÿHÿÞ·ÍoTÊñþR$¯‘ƒHþ ήCåâ»ÑQºE^þOÄÕm||ù$¹ôI šwú‡¿å‡`lHa¶‡cͲ‘Èß÷ñ—ê­|÷LtÄç§Eµ¹›haÇ슞 ý{üÉÛ>&—4Z‹ËSìØ[tûÓÖGÖ+;W'»Ð •]Ðv©e;ǃýõKUhüQ.vi‚ûÖïV?ìo¹~¿.^Ô>$¸ÜèÁ8šD¦·Uk‰ØÜÍkQ°|-]² ?=°˜N½ïÖ¢€Šã†Ç)ÙY´þ‚ÆP}šwjäÏ8½–>ô-ÞoÚ²L\Qç,ÉóŽYÜÀ®”íº{:8ªH˜ ¤cëÈôÛRà‘÷d›‰^Cžr¹—ŽÀ÷ˆÌQ¯‰ßß›Œ2ࡃˆ,°GúÖÆ€XT…ÙœõHzþ2Á7¤ä“‚zFG@Æ„·4“ØäR¢óÅwÂAºÕg&w²ªØî >è%cïãL <œJ°:ª ×Mîõ#ˉªCÊ iîv’ôžh{Lb;IbA÷~§ó9Ðßùú3£4–§dÌ€ÑÁ{ÓÌÅ&`\òÏ'ˆL[‘ÆÊ€¹ÔöìES°äeÍ}½¬”²¿=l¸6Ѻç¨çÇÀ3l¹7K ÌÃýŽyƒc}ãµq‰{à&_gtj6 Ø¿i¿…nr /,W]®iÂv 'ûè†Æï·Ú‡ÿÛx!Âñv-Ïí«Ñed;ä8ˆž‹ð%R…àÿ-ŽøW¹‡{ þR$ŸÉ3ÅÅcþ›<ü#¯æ†aó¶ÅeT¢þÒŸ­$O(P~FóǷ賈}ˆÆÿ‘<ªrþÝ;½±ý‹_9WüÀçÃ!òñ—âòiF]U³°ãI}AÂÔ×Q}öˤìÍ[ýux}É3ÅÕ]Ì®$ŠÇàãHÜæò[ 1ìoCîôS.¤Øµ{ïTU?v™fÛ  ÌœèÌb[ï·þ-§€ÔH>RÆë;«¦É|E¯nìpms* ËÄmÚ@M|Â"‘œhîg¯ëÐ#@Ñä1‹‰€Ý>ØÖ¼#ÀàûÛÆ:J«=o4þæ°‹‰§}›€í¥ŸiŒC3pdpaøÝ+²ÙÜ‹fð䇭E?ï€ý‚ýW“ €Ÿþ»ŸC®6.alEõzA$Q=m‰†ÄL •œC@B±ËÅt¤lœT¿{™|1ç±] /mÓêÉ%éži Š¡Û¹*¦ìAy?×™ªKP™µo½úÑÔF?Ñå&(€&!q‘}hk}'¸u;t+^±x'ƒ~L`Æéî`¨~Çqñ5MqXDþ“’ J]550{š¤Ý6Ö}áQÌþOÁJÁ¾ 36­¿ºiΚ½Ù9"yú6p•Ú¿òÃ\ÓiÚ{€cÛ°ÿa ûÑ~g´ ÛU™¦ÑÇ>×H|Ñ·æ‘þ-ÏáÇ­|ˆð/’ßä¯!ù×Èó<÷¨_dx[Ýu÷GæÓ rlK¼ͧÁm‡È¿­‰7 y7ä­"ñDDÏ4Ÿ;S} ÛöSsŸ]ž¯py¦ðR—8tB sú4v8µõzåõìhØíowó;î–t¥¢Ë_‡øˆgÇŠ–&>¿TõÛÉÈ~4ñcYYâ§‘$šO3wíè̇xìü$CÀàïØ_G ¯Tò›c—ØûÄÅÙ°+‡ŒDn˜(aW'ÒÚ)ÆýQ;‘0ßbÕꢟ¹¸ßõç> •aŠ~ùn].þ’Éb>£ äݱ•ÎÜn@I\d¾/p vœRŠßêÔÉ÷Ó—”ñ~ÓÇ–ßNjRÀî#’$ÏV–€ÁFÿÛïÕÀX(nï¯ Ì–’êm¯Œô·x`Ÿ.:¦8#™ß,“n“O¼™„}N"Œé~°ÿÙý)î†à{4îü2ßÑF7=.Eéú÷à Z<0öy¿ˆ»Xzæ3€äÁˆ•ÞzmV¸£öèȺÄö[绪s*œYü€Õ}^¶ë‘(FøUR|e X–$.PI©®LÍw5J‡ÏLv !àVó2ó>hí+’?ó :*v®Î³p(¬ƒ¤5 ô'((×úÁPŸöÖ¤ŒÊÃÏ|»Þ&ºû™qéÒbþ]o0o Ý÷áÇ X¾©ê(¦Ž«‰Þ²ÒÀVŽ]Û­`ìm>YòêåƒC‰òƒ9ŠZ8FÁ¾ÃÝÚ84 %{R€\õ~Âå‘(lçóýcOÂPùƒðËßê{ÿ›<ÜJ‘íþEâ þ-Ÿô¿Ä·ÊǿƷÊCý/iBv?ÿî7E∸ÖRØ9‹ÜÊMØ…X7Û0B|ý°ôŽíóű«Ñ”«Éžq€¹,ý¶²üî8Äãrh•K•™¨åÉݤÄåAØV)hY¼¯ÈWä91;€Òë( s’ì{™\5K4ü¿Õ….í˜3Ÿ‚à#Øu—èÐY÷r :ý9l_%ìyrô¥‹¯$0¹=¼üü°fê* ÈœZΛ¯õ€slšˆ†G¸'ndð_^ñ²Ï-z€¯÷¤@üózHc®®~ÃÂÙ^û;½FA4(ý&Ã!7ì™ø}ì3Hšh^W¬ÐéÈe(zT²3z˜ç©@!³Hüà䩽 ¤©™ÃÍÞÊËûÊ~Óä‚êÇs{ö-*ƒú‡gþÎ_n‚æ‚é«XWzÐÑ<¦»x1 }áµ6NŸ»O#>+`8åx«_cŒÓ"}X=—ÀÔ?ÍÐSžÌCš½Ž4ù‚åK΄ÂWä`}@..`€l[-š„­óã§ ³Dþ 1!C¸†Á`¨Ö¿Dë_¯õïúo¥uDÿ¬£Ã­£Æ­'_ÿ‹ŠEÆ¢‚ÒȘDZPLYØ&&#(.%‰,ˆ"1d Ž $$2BÒÈ@Èâ„¢"èH‰¡#qt$Ž$Ñ‘:’FG2èÅC1ÄP 1C ÅC1ÄP 1C ÅC1ÄP qCÅG1ÄQ qCÅG1ÄQ qCÅ@1$P CÅ@1$P CÅ@1$P ICÅD1$Q ICÅD1$Q ICÅB1¤P )C ÅB1¤P )C ÅB1¤P iCÅF1¤Q iCÅF1¤Q iCÅA1dP CÅA1dP CÅA1dP YCÅE1dQ YCÅE1dQ YCÁ ÁEñC1üP?”À%ñC)üP?”ÁÿAûçõCˆû‰ø„ƒ»nLêì„[úg %d QYô‰ÿg,ºi,¶i,¾i,±i,¹i,µi,½i,³i,‹Kn•܄+¹ Wr®ä&\ÉM¸’›p%7áJn•܄+µ Wj®Ô&\©M¸R›p¥6áJm•ڄ+µ Wj®ô&\éM¸Ò›p¥7áJo•ބ+½ Wz®ô&\éM¸2›pe6áÊl•ل+³ Wf®Ì&\™M¸2›pe6áÊn•݄+» Wv®ì&\ÙM¸²›pe7áÊnÂEùŒXLe´Æ¢›Æb›Æâ›ÆBœø!þoáÖ9¸ ÌCtÊé ²‡¿“ƒ÷Æ$ÿì±ãŸÕŠ˜ÌlPEjÜÿnGºY=øg{/ÌÖí™qTGýZ7hŽ^eÛ I¸åÔè z÷{æbð¿ô.n}^þÆòÜvŽl,?Ám_péßSV|ŠÛ¯È¹ò_ZŒÛÿ¹ÒÆúÜq^2m¬/ůl~c}®öÿÒ*Üñ«ÚnÿK_ápª•6(¯æ1û¿ô5÷ ËÆþopøµ7ö‹;· û×áΧÞicÿÜy5|ÜØÿîüacÿFÜy¾°±î|?PQoPÜy¨~ü/mÆs€ÁÅ]ÇG‘™ Š»žãWþ¥-¸ëjIÙ ¸ëk5ù°Aq×Ù¶ýäÅ]oÛ« üvÜu·ŸÙÀoÇ]‡ð~î>tŒnàwâîGgê~'î¾toàwáîO7ù~7î>uWnà÷àîWÏé üÜ}ëÜÀïÅݿޑ ü>Ü}컹߇»Ÿý†øý¸ûú‰lÿîþ~*ßÀÀÝç¹çw¿’È6(î¾|·ø—âîÿ nîÅýƒ9¿6(îÿ"ÒÜ ¸ÿeÈ6yƒâþŸ¡’Ï÷? ï–Ù ¸ÿkødf㳊ÙPë•ÖÖ†O¯­Œ®­­­W¬­M¬­MÞ\[›Ú¶¶öÕgmmzpmmFmí{éÚÚ,ßÚÚϤµµyâµµ¯µµ_ýkkK:kkË%kk¿÷­­­&lžÀc]÷¬¿ÖÏ—¤h]ÙçÂ`ÈÖωü7³Ý ƒ¡ì\%¨a0; 0šõ‹§]U,a0»]0úõ‡a2ÃøƒaZC0¯ÿ9,  Ûº©ÀÞŒÁpb1® #³/ƒáÅ`ø0˜ï1y Fpý Óc0"aŒèú}·Á`$0˜$”t6#K‹ÁÈ­ÿ_ò_1˜ƒ‡1líÆûjÛúûŠxý]E¶¶±,øÏûkw¯H1¡– 7ÆÂ¸±È¦ßD7Å6Å7%6%7¥pcAä·eýç½I€¼/q粺yòÎÜúû_öù_ž›ôÿ¥s“Ù„+û¿>E æÿ*Ζß7qǕۄ!¿öÿ\‡žÿ튘ÿñ¤ÿ(k§(þÓö¬!¹ÿz|²ÿñødìVÕ?õÿx‚ÿtŠtìG|œØ=Ù<|¼ýÿrGþ?ßQ…ÿ7wôß3Ýôüsn}þɽ×OÚÇÁñ_SsÓ†Nˆ²¼S1í_™ŽQL;ƒ£·qô-Žn\„búÁQ •ÁQUÄQUÕÄQ]5Ø|E¸óq^?Âõó!Ñ…+R‹G LóÁö©©Ô^¯bWŽï„ϰWf±ßÚg§O-É }²ü?¤Ž ñ'#ù…ˆ¿©÷FâTH] â—Fû£ õ-HþávÒÊpzB´^©KCú–tK©ÙÙ·‡a{5óo¯ùŽí'¬~Á|v û)ägg?ÙQ4ÿpà²9ãûŠ>´’ˆö5Ù‘'40Ú†Æs‘>zHžÓ¨lï±T-Ø1#±¬áEzìxwácìçI›I%©hß„)úècþjïðý„²+".Na¿sL‡è±ÞÄçñ×{Ú+x&`çŒ*™ }± ÛJ-z‡ý5DšCš¢Ž],9“aŽýM:Æxö< vÍ’šN"–0S|{vÈ¡²sÏ‹‹ª@T\­JyÎH´ÌìƒãaÁã)›ò7@ÖyöÉ’U+P4¶Z‡’ãû 1‹{½|Î ÔÏ0Ì¡¼»æ»é3Ÿy Ð).Äq“Ñîœñ£·qýZ>y* `¨‹.qÞFŒ—¤ê\F€éÌnÓ £œÀ’ =e|·ØFŸ)S_eÙåÀÝŸÍ3‹Kðhg$pK]_Ð'yŒî m;ôÚôu"è|÷2Và~‡bK ÝOàó‹ÝÞ±n³5CÅ’3½rM`´·ø4ƨŒ–Œ,>î9ÆCì©Ü­Ó`Òv×{Ì0L»ö|¯.²³#ã#Öä`Á&á:ó8,]w§tv Âáº!Ö•U°Æ>ê¼úxlž‹ç< ;VSOág‡À^c͆>”¢‰Þøþ$3íП%€Èn"[±Ç^@¡ tßþ€6vŠG¯j0^Ç yîH~0Òɇ@ꬑø/BòŒÑøR‡†Û¾v´k¢`e Í?Fó(<{$Þ‹äãêd÷Ú_ןÉ«@ëgœÍ¢0Ø^ç}¿Öx® yRHþ0ÒßíˇäM±¸|M%"Bëk†çŒk|jžaGV»üAóŠÇÑr=J×Æ×Ÿ"ý‰®¦Ej½EóÑ:Ô§‚—i°ß‡„÷LåÎcgyïî0:]‹ýæw{ø$?vî7u³x6 >¥ìÒ·{9¾?& wo…/vÍðÎï²mó€éŠw¨¡áÂý–5TžK@t54`¿œ0_Ë 4aÒºGJÔ¢<@–z/'þ%PÜâ3¦ü ”µ©œå‰$°“÷´ÿ×Ó@=à°:?hõc¥ýf~ݑ˭áO°»àõ7âGÀP3ÈÄÆ_¯ 0žFø¼ ËK‡H+C9W[-òp‘L±WÞ’îeöÂ6ÒlàåÜkÃãæ|Ϻ^e¼Ê…ËóÚ ÔºtËlŒD&CžÎ>wÇçWi\zÚu™¤öúбd„>:ijœÕ]³érJPÈ'QýQÒ Àuî7k"¬[ËM»(AYV—í@‹:¨dÖDݸ|Ô˜…–®4úû_u‚¦x2ÿ4ÐúS×gÚ:^rÂ:Tæpˆ>£QxŽô*É?‰ê„€äM“Z‘0Ü~‘£š“ ‡X}>$‚QIf¦Z}'Rܽ»ŸL|‚Îwê©]Xº³×-0³â»®#”æž$—$¬o€EFðØ/6K°ü]]¯ÃS VQt ÏŒ›ÁFSÅUÎênåK Ç 2;‹‘)W,‰¯ŸšyÓ\O˜Úö•Ç®-Ö_¥ÏAù¯ùtS}‚šw„äç"}À| $^‹Ô½ ñ\4®‹ô›Åm‡ð#’÷ˆäù"ùŽhž/Òé—·µß—£¥¬Ú¥¸~ èúlBç’oOQþEê¿‘º7´n—׈ôU@û ß°/SOÁ÷ÏÄÕÍŒò3¥‹”½ÆŽ³7¼ýÊÇ„0, ¥ò}‡ýR6Éu(wŸ×X'}~×ä/ì")Ãj4¿ÿþ`×cv.ŸìÊšì…^ìZÊ·'?Ĩñý÷~´‘§öo“‹R×ÛgZM™«i]—™PD:×›¯’vÊÀ®¡¨{}þi@R¬(Afì¡$åj¹ök±/„£®Ó#Î;LÀòñX´HN(°½¦ùê—†¬Œòu]ÀyoBÝ{æ2póÍ=unžŽ0Z¥…à}¢æ_8¹øž¬j4¦þ~í(;’ùNÚæ "SGÛ»ü@´äØÀÍ;SøþÒwØåuéuïK¯ŒÜD‚vD?ÈÙÿ:µòªt¯-0 ùµ"ÁJÄ|‰eK± |¼„ ûö¨ô¿,ˆú” jŽiªôÉm ¾$ý…ja4s“_ôfHöIŸyÆu~Ð5縹{ÿl»EŠ÷²€âÝáT0´a¬»{xýð¿Ý³ýÁ¸€)6ËLfJÎ)j¦™Ü‡UŽ8©ï«\› ,}`C¯ûÔ¢‹ l/·xž°ø öÛ¯ì«+fûŽ™¶o ”àp¿·Û>ì6„Ìqhsþ\Ã’Ø@‰ï<À–®Ò'êvÉ)dÀw´¾ù]ÉÜ­}“(ÿ!r á+¤ÿ’oäe ùUHÇÖº7$_ÉË@ò’ѾÐg}¯ù†ÊW„ïÑ~_¸<ŒZI-ÆÂ3|h\¿\DÿEò·¾˜h’w…ô?ù[¾’·!£}xQÁðϾÑHÝ7N?Fërì}Žãû L~îõqÊÇçõ·'%Èçka§¶ÕìmPªÆ×©"}¼«Ù^]ÿºWØŸN=4;Ýú±ótš©z_ͱ ý7» >]Gû¢,º­u`Wÿ™ABŽ0'ªÏ¼ÔP‚Á»Æ‘áø:UnŠÕóÚ|@ú{ް¬xßψÙwÎŒÛ¨ä» Óˆ·ÃÎÄ#D„¢Cø<Ž.щk* °«)ÍËTþ0Ê:ÛÆÒã)¿ùîú`æ+²Ö1V6AW‡3xýuEáD‹'pßõòo ¼ÄÅ«”ÀgÃwI2ñ-ðO ^ýHÑ BO|Ý|DººíÜ©AìlHÃ3WjPþÒüìûâ_e7ûa2’|-D ç2¬õYƒµ…ˆ’x£r}YJûÛÇHßòƒrÉ¢Õç{ü êÆG“/]ê2Ov¸‚&w¯‰ ™ hË0$ö]?¾›ì×øA¯½`~‚ rFfŒDÀðËSuìa/0–}~Q®LRBFÒåÀŒtéxJÑ4˜Ÿ.Ù¡dçC}¿¾"°*zwæVذEÒ ßÛÌM€ùŸ|ø`­rf,"Ûì£JÇÀ$µd1[ƒ˜Ó|„4].c9sR©ú§ÐþËH}fŒ\òp¾ ÚçáC$¿ÉûEù©ŸÙZoŠË#Fø‘§_!r­ý[Ÿ \ß<¤~ ¡Èy#õ1ÈöH¿0t§gÿÑgÉûçú™ê¿ø/'·ôØÏÛV»ØÁbÑPªÕ0|¿vŸO?³…žcÇ ul핱“ÂB&»B;Ñú™o%G¹[¦b¿ï?™zÕKç_’sNÔ`ƒ­¹c±Kj³-÷ÈŸc+ëàäeÌ>a—dI (õ*”ã¥"m[Ú+ þHxf:Ë/¢ÇŽ k S L`7<Ñ* ;•ªäŽa›ñ}Q®FÉ]­w‚]¼ÖIÓGкR†¨Å`§?€ÑþDûž4B`²¼˜–ÇÖ,îÓΧ•€-ytQ¡ÕØ¿?Ÿ:ÿø pÐQΟp®@Á¹§*Àýõœ ñKØî§HPÐûÕ/'þ0႞{ c’@@1c!rçÞ¼%w›ºËMÙ·[·ò {/(>›‰ÌÔÞÊBÛeeËAå Mý~P«7)x/s4>Z¤ © ƒÖ€ÆéÁ¸Ë ³=iÝÀz´iV„̺`@F>)äp &C†ÙcÁð³ús5g0ú±»ú•Ç"˜±ðRI3€é+=Bã`fÙ>¨íoæ×šÒäü‚Å'ÖÙAB78,™SøæÕXÝ8[Bé§ 6CALMŽ`ýtçji°§Xì»2â³I–FÕ&´@¨¿]°äá18äÇà[£²ô}n®éµß°?= ¨¦82ѺM¤Ÿ¢O"|‡ä"yŠHÒß©GEê üdé{»µ^áÓ?꽑¾Ñ¸:o4'?+_'ëÜNÇÛ¥H½Ï_åéÖ~aH¾ÿÖþº8{ÍûÇÕí ò²¦Ôl[  |‹ö“Fú…!ré'Ájv’¥ýO9‰›'e’æ®2Ä£õ«ßxnìɺð;³² x¼Ÿïÿ©&ûF­·;Ç:e–KÓ¿*ö˜g$ û ;kåqþÊÇ+5»ûK9úðyÿ¾Õ ËDß?(âeRž'% @Ú:”ú«<È^×ïTtÀ×éàòþwN6ß‘Õ41ÃZÕÎê(K𥳰Ût޼ªÓÎ>:lç)Œ‘wž/ºsN¸Ö‰;ÀVq()vopD²™.3«’d:màñùÙbzÓx×>DM…¿bÚµó@èìT›üË•PÖâ|’âRX½)UÜëàóüú;ªë½ß²½dèmÀAÁ¯ŸûzˆAÑ!(ûQC+(›lȼ÷ ¨žHû9o< ê÷OôæI5‚ÁÏ _Ý@ÇÁ„AÙJœ=”Ì×úŸú(Tˆêÿ¯ñáÅì çD'è"ñèú¸_ìlÐÇ~¿á¼úð‹7Zg†äé#r áDEôÊ­z)RgƒÔÝl÷á¯jÝÎwïÊ P>@òûù…Ö§áì>¤:âé!Žl¯ iD)âWEêØ<`DΡó*˜VØùÐ)áåÞÎÌBQËèþ¨g1¤åxw'~"œ?vð;ÿûó®zØáó¦üµ*P?ë8-yù;;|¾02ïPó =fé2ìÌa³ë q?±ßgÃ_fmÇΦ‘z7ïÇ÷ás¥X>°«»˜°P¢|¹»üS9G·Õ»z%à²ðöBÀT«ÜàÈÂãæ¾Þ½ Ä ãÅv >@*’õÚË È”çj„ÞH£ó–P61=%ˆ‡µÄú¯õJr{Þ?z¦ëÞ2Y ›ªO¶l†ÝW².Ê–ÓƒyšÈƒŽ:`Ô ZФ&çqž"S,°ÜN§}ê¾l_ôÎwíKûîà´ó}"Æð ¸ŠŽm Z·¥.›{ºû¾³½<¼ËöÎDPîs~jBá ->´67ía» Â/ÂúôÕú@t§­„Oáˆi¦ )ƒx¦êÂZ ÉgÓðsGHÕ×ûKL9ƒLXxþWY³ŠNÐ[·¯ùòºB à PŒ¸2(‘¨ ÿ<Êö|Ïy}1 ’èøžVTˆ×©–@Ýãåа hnŸjR¾£ Zíü|Ùl óðœâôu8äGí•Zz$?Ý Á€crà~6<‹˜4cCýD³LC0ì?,¢¶ËŒŽtL¥€ÑÀæ 2 06¾Tÿ•z—»|>×&ô9×J/€IàÞF1é`ÒÔø;ÇÌ L™›\²ôÁôÈ6,“Iœ?pöÖ‡9 Î÷¾òû%" .ÛázÏH;•õØïèh.:Ob§¡rÇoˆý‡È3D¾!yþÌ'†«óAê ú¤Ï<’×ð7¢÷"}ŽyŒÔ³¡òñÇ"ö$"q~ ¤o`õ«h§ì™d|ÿL¤é7¿µŸ â/Bä!Î_…ÚH?iœÿgk¿°?úg"vãÖ~aˆß7ïb7~éñ9Æê,fJaa°*Ä×ë ùþ„¹œZ!¨ŸužsÏÕ`ûçØ…1bBíò'ØÅÚ–#}ØåJqž!_ìï~9½úÁ)ÀPw5.›œ­‘ê·°rÞp)lˆÆÙ3?\Rˆ· %k2mOí÷«@qÑ«KÇçPí芌·tÁ÷™Œ/ܹK?Ñ‚ùãWsZÀ°[Ìÿá•`¼ž©ëi Ì矕ÑÃ×ÃÝҟήð‡½&ÓÀ“L›¿ïiðö ôÿâ†nϲ­APÝõ›Òþ ²?©’Sç.ˆV8D4q»»A2 ±(¹¡¤ÓÕ^IìÅÛ‹½Ó}<:ßz:¥A û(Ã"”¯¿²šø ªÚ•ÖWÉâ@Ó“õÈþhÜÊ‚— e@oi—p´Û¹dIe€®d2Q‘„ :í”/ÎBzyÎçõÞÖƒ~Ý—µ™ 0p÷¡¸ÃJ _-.-¿CìOÁ`ØÄM%¤’Fr3Ï Á(¾5\ë¦8M„P_z¸ƲJÚA­4`zšgO°øädß Òñ¬«jecÙÐþfÔ[ÔÀ~Ý{ÚYTù êÿ@žw„?»‘k¨>Š“{ˆ>ŠðR_‡PD/Eü4_!|ÑÀ2²Ò>;úQøÚgÏ¡}/qr­ãJ^†·éOt•ˆßW7ŠÔ¹ý"Û¡rw\Ô?c\zx†.__ƒ‹c0ºÙGÂZo:¼gÏ› 1hß…±Ñꎾãäx9é¤+¯åZ‚¦lõoÉ“Ã×™ÚMr=$ý†ï§iúä•ã+|Ü]ûߙŚHßìÚE\H·›_ùºðk} UÇgTÓ’àíÆÓZjtžV@QÞE¾â‡÷ÇüÐô?#4Ç宅&Úo­¯¿\#„]¡ôQJ#@/\uá*õ*ìÁ¤O–¾9{É'=eNÚ³þ4y *;°v®>]œß ìõ؈uiÀ)SïfèÉ\nÈ¿r%±<«QžÃ°¯ãžfÙ¯8ØŸÿ!°sI|«ÿ}>WK¿¼[Â–ÒæcXÝÑ;´ =bé,óù„‘å(E3HíJe=úä%HñÜôt¥ÙîK?¢»ÍAÚ«D1 °¦îÁ?Î-€â¡¯Ì“, ä?.z*ã=(—†ÛËï›ÕýïX?¶KZõpkˆú{и˜ .ÃZ§ؘ÷ƒŽhÁ£‰ð Ð}Ãø¾q#èi,dÝF úeû»d;[Á n6¶¨b7JN·ŸÖÖƒÙ#$`ä\Ga?“F3=]W3©ÁØ+}|;) O0ñXó‚‰ù˜´ Š9˜”ï 6ä-[—‡NåÉ:.`zŠ“N’…¼4>öìi¡ƒk‚¯oÀÎnQ‰ž!ìXÍ‘Tc½‡¨ü@ê˾Cä"÷ù¶µ?üÖºV„níó·U."|þǼ{¸ºoä}€Öãü=H|Ñ›‘>šˆ\ýÞDêÄ‘~ÓHÿœüÜÚW ‰s¢v$ÎŒ¾/>*ˆýˆÈKÜûÑ{}?×?Wâ┨݈Ìχ‹KNIìâ&–ðÄNßN¾r {;“¥ä˜¢üûcGܶ´Ÿj7~Æ|1¤ÁþZ˧¾—óßú3:&&øyS>ñwbÔ_“Íâå#=‹{rÌ- ï„àÞ@ ,ìs-M%]n*èÓ¢Y˜Lè# ý¬EÛæÈ»žX}9«ôÙšKªÉÒ°§´±:…º ˜è+Š»½–Îý!×9`O´|¨8)œQ»ŽÛdwõí¬àG¢À+mϱ‡ùÀ7#\ײ íÚ#†;Ax¶t@Qâ8ˆ¾ø¦‘ž—âÅåKõNl ùj÷4cå#nRÎ Ù6 ²ó<'©’µAÁ[á [–€i=ó«÷ $Mš2Öt ”ßÙ­6ÞÕca‡Ì@^Gf5Ã4 ‹o«Í©ƒ–fmR|X&h7}_P‹]MÎÃaApèñàïØc O¦ç«'dûjm›J„Á RÖ‰¸¹ ®7~ÈÃŽ0{K¶H02 *¾ñŒê/?9ÐQÆÒ\-™’”`œòˆƒ–¸<Õ\K£ié@¤ä´•yì \RPk1Ä~>“ݬ‚Úƒˆý‡ð!bçýMÿüÛ<¶EôSäxß Ï=GDúL#óÕâúK£ñ…àÆ|–;x?(n>!¤Nõ‡âòs¶úWÑyy‡Ó+ÿ(:n¾´O åöª Y®h}é@¥æÙÜj|¾ŽŸ~bÃu»}^›ããÿHßiYÙws5ØIÝâK ‹¤ø>›“ä™Ah¼ÿwyªèrv¶KÁPžý7v®ê£äÉ4n±$ÍêteO+v¥è ·|Ò_´8‚Ͻ¥"ÚÛ‚±@B>lõˆ ¶‘¸wØb€œ²õnÚµ1Ø.6MÍ®-Tñ,+Yµçš&QË’éЫ°|M´ ´?z`—Ésö@® ß-§0{rìapQ½·öÚT¿öH“æáý;"€--÷ZÁÓÛÀ¡ðéE—1p&ÕÞx3 ÜjÖ« ïg%íî½ÝcÀÛuAÓ”|S ‰¬æÁøþ)Û(w²Ò‘ƒ(kc §>ˆ½LN ósƒw}™Ar’Œù¾/;H]™?«$ ²}ê@A~¼&t‡?`¨‹GîEƒbíÉ. ªS ¬·xç“ç{P%ÖGÓnPky:ðñèи3 Ê±ohöJÿâ :Äw¹“©î(Wkôèåø+ºk‚ü~^Ú†÷«“;‚Ñ¡gI5ºC`ìÝgåoÚ&ù s=jZ`FÌ7w£±Ì}ß ÿˆ‹Ëm‚Tž p¸Ô„3ãò{°¾I08ñÐlï2ìyRö§®G1=î7†« ·kCïðÏ»e{Ætéù§‡µ±C?Óe DÑyJ~Ù:o"ÿ¶öuØÚoek_έýþ¶Î_ûÇ|}ÿ˾ÒÈKœh«ÜücþZD:ø»)eõáç¯ÅÙ‘Hž§üÛ|}Hý8òÞ@ôé]däµeoðò§ï¢òé—‹øæ£dM—ð~U¤oü½ÒÐ²ÝØÏ‹üӊѨ^‹æÕéÄ)ŸªéDûÎ{~æJìÜé×÷ÚF± )F%Ò:øy("4 yr°«Þ5n–¾&&˜÷­ª ,,^!žà¢³ûû¥U€„ê>‹¬J ¾ž |øÈÊõ§?“„Âv’U‰Ä²û@uËö«ŸP+^¦x~hº_ 8é—]œõ»‘þØmÞuýNK0ˆÇµ²~F,€cÙ`J¾Èm¼®×ž¸Õòýæ°Š¯˜½3Î÷%áueøx£t^Àj^ð½`‹M@í3ãߢAøžo°,³ˆÞ“¸H…Òqe'ËÍaŸfñô9àÊ»ØÖ›Röº GVž ÇS>üíÉ#‰ÐM‹{h± rn3Õ~{À˜X:ˆêãC0Yá•”ý fœœõ‘™Š`®É=ïT},’dz ¬ÀÒMjõ‡%X&ÊrFR—¥¢ÄÜ÷¯`ñJ8¤ÏŠ,<ŽaèÖÐeËÏË\^hƒÕ%+9»³U`#ä°½ lWKe5ö€½aPÜrt%8pe8‰G»ÃÜ»ôø¤ pì#8tDÐŽÔ(„ 0£mG®ß§å‰‚‹sàâà–Æó&\{‰MÖ\·ã%ñ'¾)‚;OòµûýÀƒ™x‘e³Ù$ˆƒ·pèŸÔM8u<Š2cð$ø47†[¦ß?÷ ßT8c}¥¢!U î{Î.m— î!»Óp>­¸àðZa ÷°Ú7‹Æ¸.Ð Ç9L&Cø/)©¯'ø!J¢·üîÔv¸X˦ÎÛøbÓÿöŒ3‚+†)¥wÆWáêXíLuéH8¯ø¤¯n®í~”t¿“’Ò-¯'ª€dÙ¢ñ¾Ôçp“ÇI¹{ Rc°ªê^œ~Ì—©åŠÜÑd¶™²³‚Œ1eÂf¯\ÈJ,é¯M…»šÏÆ^†\* —_‰97cõÆgä<ضãìÈ…ücYµÓ‰ð8õ¾‰"#(ë6±d¹A¡än Kµ*(>qæ×—<&xÞ²¬<ì8/u° ÊÓçPÿ±kúmóÐTêç6À€¾tú, ¥™àPS‹ÌÉã`@íñ-€³ ¥L冟q‚QVoK ŒÛ…Ô±´OÁt¦ù ˜Ý梳[H‹½|ån6ñ(¿[S8 ôl÷[ÛÍ÷kÉÀÞ9xmü98W‰´¦†#¢ôtá†?Q¾v.៌༠ÇHéƒÚ=àx©ƒ1õ xôòŠ?c»^t͹ïƒw¦ÝiðáñQ¥¸¾5ò§o¹žƒÓOß›_Í^ÿ¯æv#”˜ró£‰ Ï·´–ÑÁÉf~ë½ÕùÇVW73B$µFm(D­žÿp¨*.6—¯°¶À¥×úÉC~„p™}ÅŒõ>\¥¯#±ˆÌ;GªÝ‰}çw½eɤIç‹,6pƒ:ªö@ð(¤P-”$…Á­â´ }Öåp{ŸÞ0}B ܹÙ6Q²z2 ˩϶ŒC¶½(É›Èy›Ò3Wîù œùNÝ&d-ãá±Ýþy]ëóP ÷K£îXÆî¿¯ÁŃ~ƒÓPrÞ‚Ú4’J-C^°ÊCE¢àë2³)¨ºzNb›•dÑ;{ ê%µ>2÷=€w¯Žˆ\쑆¦ ^‚¼SÐ,,cשïS~=v# Z? ¦7B[ÑDàvY èÈ"N}öº’ÓSúr/@Oá®z•!è[yš¹û‹ È]Éý\v¯¼ ŒÇØÀ0‘îÄÂ^RIÔè8§Õcê…*CÖð™Yé#i´ïßùÐv€0]´ }žådäR‚“‚ÄâÕ³£®5 âØJíú‹åG„Ÿ¾Ü?þ#svåË­|ˆl‡ð·â3MrîÙƒ(_ös÷*ƒ!kr3)üso!ÐßQ&&Èo®€©÷«Ó<`úž‹«VŽLqNØÅL‚i—y¤º‘§È²™.g-Üó7¼¡LôÚ¨ü<<¢ÇÕÍ Ö¶Dôb÷Á¦3)5Îr'ØÕÎäU’‚ýÙ çn§ÌƒÃ¡{®Ý•?ÁQ¢Û¼{| ÏGæN‰ / À)þM°Ç¥|pþ¥B*³ð\½;µ 6Æñ±Óª¢—àD}7ÃÇëàÙù¦ê• 'ž¿<м=% §Ö(ÔnA=øN;¸õÀ1»s½¶Ç ÉDNÿØ3\ÊòÚ™¬ç *Z”ÆBÁpÇðy‹Z¸ö}.Ø\£I*ˆHƒölz ´f æeÄÕퟗçØmºí\á!ºÉmñÃwž…D1ã¤{B¸ÖRñšÍØ®‡Á¡¹¸HÄv*RúëvŒNÖBZ^JÄì…1¸Ýù[~Ì´2\Ú¯µBÅŠ“Ý'È®>Ùn 9±‰Mo¯øB^`Ò±dg¸;[¿v™ò©s<~Ü´„Ç~2¬âäqPÀ!÷T8¬ ÷rÕ->“‡bKo’à ÒðüÍ ç3þÃðRÓøSX”Ml?·"! •î9Û”¢¡j‘ô½ÁW¨Î]TÕu˜‡×Ôj @mÜóÑ_?®B]ÍÙ„7àKÀÄåÕ0xõëR;>\Yd9”ÍcÎBO]’ EK;§ê˜=´>;©¼¦øÚ&-χŽ”z´g¡K¤70 =¤N9Û½BoŸß¹¾ww ¿5˜e¹â œ¦líJ¢€ÁkŸ©XÆC`hýOœkHƒ‘Û¨]aô„[H{Š%Œ ýöoûÌ Ÿ½ÞUÿÊ¿ò¡Ÿy‹IŠd$(W±~7WGå¢"|†Èµ¿ÉA kúD¾á?(²=¢¾Ÿ @JKV8}²ähòSî`yu‚¸«Py‡èÇÈy!rÙYF¶Cä§â‘Ö¦óî dÇMÕíŒêÁ*feªÉÑ—“ך‰¯{7vlDˆ®¨^”A8è‘S‘V¢ú³¡½áÊrà 0z,ÿÕ«è,˜à7O6"Eßæ©ûb…ö‚e¹¼£à`^>ß÷åN\ûì˜ì3B (_Ýcœ“)°œ HÏz΂ëw*Çò÷àNëþ€öxŠX…ûÛˆK¡26w/xÇé=ˆUlÖ»ý¢ç?ƒÑÙ±3¹ x¾­Œ© z;AùB®î¾/P»#TõLýÝ©pý“æ§*ˆ lºt€ì7DŽj+_ÛëÑϵT䤢!¦Xб»âvõx˜ÁW.ÈœÿÍñJçm4; !‘÷ÎG5* Hb-kɳ Àë±g½¿“O¸BêÏ#3I{ä!=C.Vø£#Ü1À6½ %Lj÷WÑtuÍMw  Šr¢ýÞ{sOÂ=áw= Ñðñ 6wËz½$šƒ‘¢môfo¡ðÿm–^K(îJüZ›m%Á4®!PêA=Ôåö*ž0F_#΂ª¼Ëm¼“xþl¼ûÕ+Þî ø>óØꣽôŸŽ•@#ß×Ö}ß i‰û¦JÎnh>óÛ›æ -d…¿º«¡5¾¾Ã¨jÚ®ý8Š…Na"6Ê¡WÐ-½Í„E0zÝ2ŽW•ÿ‚þFsYûzxò¦"nø íyŸW@A ÷Êë4`TÅó™©ŒŒSfx,E¿…Ï+lcU|åÃ+Ñ—Â| JÁˆL}OËÓqÐøú%zì‹ úÜ"úòÜË^92ÒVµ÷$ Ë5¨_ö¢rá[„Š2òúæÎ¡z/"ÿLry+µëÂag5Ÿ¨w¨r¸órw]…+QgrIØuœv¥©wÇrü!o祩罽”ÿXÿW¹táJá‡át8™7dé$Õ^‘WN¾˜ÁË'3ßAé¯à£ÙhÏ# ¾^‘·ëÀïËÉ0uÑçpæ®Ëo†"“ts#£¿=i‚sIñšùúl»m&§¿BΞûyéÈ"„>ïÓºC¶'¿>¿GÂmc)ï¯)BD›¬xBäœó®âEˆ>“˜ØA1òklíþ» V8ᄀθ,þ&ü‹ \I{û­n®&U[±—=€ø¼ë£oeCÂ’p,37\»íðjÊ0\Ü-4ÞþnÿjäÊ ©FY){Îý‚4ƒ@CC_mH/8uñE 7Ü6€÷Á¾7áö srÛœv¸½:%V× wÉ[”Z( S+Á›f{d{ÍÙ;Çü†\îŒKkDpouøúD¦<,üau¢PUåô“¹Ããx!lX¬+<á]ÉŒ'ÇË.ŒêBA]¼$E*)î ŸSŒ „¢Êkl½â’ð¬_âº#é<¼À¾"Òª‚ÒÚ3ù'l ‚q^ñ&*£NžoWÏ€ªÌɼ܈~xõî5sš#Ô\uìÑæ[·#c¯ý€7ù ‹.ZMðvow´ÁŽóPßRp1Äè¤<Ä­HàÝõJý˜Ç´ðžu”|*šr¥êäM¶Ã‡›eò+AŸ ùH2ƒ·/'|*9Æ:W\.¹Ü-…–ñЗŒöÅÐÚü¢ôƒš8´½9ËÑíoò(ÚJ' ãcØ=GA:蜌 ¿H¦Ý;¯[×DCÒJñ«¹Zè ÍÓ¾¡! }Í­û±)œåÃCø`Zq uZšbYAõnÐö ]T.!òg«ÝˆÈKDOåä/øB[ŽRäwd;„»9."Ç>EåÖÒ‰ˆ™ª2P×½™ Ó¨]‡Ê«.Ù›‰ fúüŽõçÕk­ß‘ÛÏãí9DNyóÒ°æ¢~+“Ó?åÃÀæÂ¥¶°[¸++ ŽùçMÄQ? ª6÷˜— ¢þÄžs«’Šs÷ÌÞ|5'V𸅉|÷à3œÔᨹñ<¼¬‡ Ån›wØžmœ‹pª¢¹éCI-ø2%ψL¿K{dŸ…3¼-Š„ ŠÀy0`^› Sºãá\ƒf)/‚co™w´@Èí0:ÁS¡ù idÇ!l,Ú§º€åëHÅAªë4re ù-\¢¾üƒ‡®âò‹4­M»áʨ’G_¿ÁÕâš“Áé_’šÃùÒ wÆi¬ë©é¿ö[žç†$KñÝ?­ YÂ<ÕÏõÜ\ÓÕàpH3j—ߣ ·Ç¢™¶EAF°.ywóÈ,3xjõ,²î]ŸPAÈ>Q>Ìwîžcs\v…œ>‰ª[n!¯o‡Sè xjʨ<9î\ ¿uí*< l©<ؤO•5õ‰Ë¡ˆd²ˆõR7×GËœ’Ù½®¯ª\¢2…ßoï™—‘‡2½˜SR@EÏQ=¢;PuhœOÞ•^½9þ”öäm¨1­àg8Þ ¯—Yø—× ¶È¢D!) ê.´ì ñc…—™Ðw¡SÐhw¦ˆÛܚܒ,ÝÞÀ‡šŽ—«™ÐÜïð+i› ´paæž ƒÖ“JíÌ´UÐÖJ{ Ïì,tè`ÌއÎÑçW'ZÜ¡ûúN/ëSÐëôVIƒá8ô›0þÒk»¢} œï`PçåÃ캿òa°E†ü´ê/ÝÊwˆŸt«}ˆðÛßìCDoE¶Cøá7„¯½Ï°_ósLh#?$“ûj~]YÆûMqúêÅùO-TùI¦,ß¡öâ_EéÌþxÿ5Î?ý§; YØÚðöaþÕûQex?ªBOÕ 8ìH;CWÞŽÍÜ‘6xû0Ä«ñÚ.p¦ž¹ï"À.E1*O$hà˜§jÛô8¤ØÌX'n»k'LÃÉí ӧέË]7úÔšYðn =n¼ý5øˆ¿ ÑXßGû$FZ¿Àim [i‹5ðçÁdpþ‚³w›sIÏAPܹ±”Þf½u9ÓwBÛûëï(°À{Ñ».ÄAxX&ÿµ£ùrÂý>ÀE´Ë3p)<ôf®,\Ö;šÒw¥ ® 0ë>ø7óÔgu.Bâé®÷ÏŠ}Q=óz»áˆíº}ø‘´ä‘?@ªô“¾‘JgH'ãûyÛ°îPúS¥Ïè@ÆíIßÙ¯/ K-UÐP²~µˆ'ƒœçœ\–…üwÞÓ7f7ܽ t_6™Ë~0,†Ç°ím_‹<éüppÛÍãðô5û„Š@(Zr¶ûaÏ]ª§]t€_h³ÝºÚ ,ÂíÕ›¨œ{@n¹ªî=ÿìø ªµ®?/q{Ÿ¿ò¡¿e•³IjWmõÏlõÇ þÒ­qŠ¿ùKþDøñ¿Ù‡[ý¥[)¢'£v ο„ð÷V¿©Êjê Þ„qPÓ³•Õ!¢õ4.>aˆ¿úMõøhMäCðöàE½îô‚b0ê–Íñ—þ‚·qþ"ÔÄÅQÐøÎïcO{ê2MÔŸrác'ª^×ö&¼§u*T€âœð3:ßúp’ƒ¤¦áŽ x=áÀ&*GÃ)ûÕY_+à{ø#4z|‡ÓyÊîû~e@€@ƒ|n_~p!Pùª÷—Z½fôKÕƒ°=L­ê^Vpa¤¸±]Ë"ž‘ÞÜB¢m\Oe¹ÁEG‰×ÛG}àÒáeJ†tˆ+ó84 ®ôˆ$ùúÕoéÑ8IYE¥O#< ù-á/ŧbBYuÊrwÜ:«”È+p›,V&¿BîøI%gý€L¶#¿M[(!›5õ¤íúýÎ9[ö¤·vîqW¿Ú]÷öˆÄ}ÿ •Øé™øD¡@¶»£­n ƒ|?IÒ CñUOÒWÅPâÞÙ!  ¥‡XÏ ܇Š@_ã~oy¨ ›/#)™‚j?1•Sð:VdþÕ®D¨­1¼úˆêÙYf†w÷Ý…Š"ŸAÓQ=A¢XhæØM™‰¡ƒ)…ÖuŒ“ÐÊBa¶ì¨m™á1‡– ãjHEÖºBêŠlCOªæ2­_ô iΞ …ýQŽ?÷Á`ÍϽjb0ôcï³oõu0âMr³ÞÆDŸ®°‹QÁg’{Z¥‰åóÇ<ìÕY‚Q¿b÷!zäÖ¸ÅVûo+?"þ™¿úKqüƒè›ÈóÿÇsÿ¿)âE)ÎÎD×cª¸÷çà)n=¢ŸšÕs9¾sGå¤%½89‘‰VP~sm{ÊGÖ~wÏm[”<Ãb7°ûÐmô˜Å ìG’jå}®‚#[Û3mVA8Î_œÔY G×®¿ú^˜ ÎYÜF*¿N«MÓ>Óµoà&’›zÜå2ß…ìÈÙŠð[oáäù×öR§À«GÉ*ý\jú’ocß_~£L“s½p¦‡—UÁÝ/¡rk¢1 B–ŽëhÒBØ—Öò¯«£þ;ôS½ƒDɦýSo/ÛvXm%Ê^ ¸²oõ¹õ\}ÿúóõõx»¥Æ;®Y$“ÌdRÀuG½¥wá†ÈëÑ¢]]’\UªüX{靖ÿ?~˜T(E”ì2 ‘P="+£Ìl‘¢¥T´(¥¡´wiïMKihï½§öF‰Œ"Üîs®×ûºÏåt>ßß}ß½Ïûº.Õ9ÞÏëù|>ÖëX|/º-Uí^ ÿЮ=ú'ï!P ˜kX¸A%£»O¦ßFÈ£Þ(›oÒ;âµç£7ˆØø¥{eò[D Ý9ÉÂÉ„˜'cžÊ- ˆ]ñ.Hìþ ÄeíQ}ûë8ùäýZ×ãÕgkûf…6¼VÒôâߌԱJ)æd+¼ñ|VæøFYÉ©¿’]¼ù£¬þ äÞ——_‚üc[Y¢×¿G¡´‹»uBŠ ý%Gz¢´r§ZäATˆž{0k9ªn~X]2Õ +¤Ž-ù…Z®+¡Þ.Þ¨3øºN~!ê?-Üw{ Uõ»øn£©÷Ð~{ hIçÑ÷Â[ÓÑäæ×Юķck4:¤m@ÏZt]Óßgû&ÝÖ–*öV­è)ŸÈ:mÒ‚¾õ—‰‡Š¡ßí¤þž_Z\É–!Í~vÊ:$xÁI¿"õ÷¿â¤+ØL/ŸùñŸ+ù<…—ÒæÜÿÂKÉ=ù»(ü´ÑóÑÖQÔ=ã¿û7%ü"#nJÛƒÉ÷Á§¤#›ŒqÜJáŒÐò%8i²Îêh¯dÊúè2=Åé9].ȼ ¾'¨þHãQ(~„†—^Iºž7ÖŸNÕ3Å'ÒôD' Øß­Xyt”Žèûò¡÷E2×J±U°2Ýa³4‰ÂÑlÏõ.¾{†öFÒõ¹ë•àPÇ´¹k!'î;%Ý€ó¥f«ë¼‰peû©YßþîlNWV…çÂól‡9›Ùà*òá7 |‚RÓØÝàûëÊ<[;øµÎX«»·þ›_/¹.¿þòž³Zp8[zÓŠz&|6ž—Žà"·qR_vÍnmî‡\DÎ[p>\k%¢wY<ا€ÏÏw .=ˆ—“b­NÃîˆõÜ"QÄ8YñZÞáaÄï ø6ºè$äKºfó»"qÔ9,.¿ Im].§^„"e^¤óGOi¤™­ùô‚ÿÞŒŒòüÝ,N­%O.#{ƒrŽR¸rŽ ÍL|\½%®,šÓ—ºÞñ¤½< 8˜Wë¿ìE¡:»mŠZ¸÷‡…£Dzï§³»æ£´ Ì'E‹åû™JN¿FEÆ<¡[=*¨Z)ðH½ñ3ªºn³d,–AuHjö¼Àý¨QÛ³Öí¡je’y¯-Ú‡ºíïê<ªJQ¿â’ª S:xB›ŸGã"üØï\ˆ¦õ!î¿'µÑ¼ÿJÏÙ1´( EìåÐG«¯ßFY¼íj`›!g>e>Óâs:± Çö8´lЧö+R¤ÞÈ\Ix 2·N¥w#x¹'}–ü6›™“È>Spvé䔦eis¨!ïÊq½ÜBL—ýÚ›,ÂôÆÅé_Pl£Öª·Õ¥÷t§ícã@¹Òª?þ¨¼/RuyU%›ü£Æ¤PÝ;ãë‡EÓQ»Q=yÓIÔé²ôÏ æF}ëî#¯ ñBý霙ÜhÜ®<\sX‹·WíOïBûõ¾Ù5;[й»éØ‘« è:³Óqú5±)ëИ¯ÿRëýªÿ= ©3F~âÿ5^J›Cõ4'Ì%¿¾/óÄ©×rž_Réóm¿;oÀ÷þÐÙRªnˆŽìœ”ö>Ñ› µÒx ‚·þâ²ÃQñykçàJ§¼êröA\»l¹´”U×ë’,&º_Óõ§DW£=h#¿G·FŽL¬=^EÃ̈žÜYs&a•Åi:o9ô­ÕïÀ ÜåXøŽ½j54<}ù BsþvnþU¯ýŸôDƒ<(^ ]# ¯??¹hÞƒ^ì{ý‹ÓaðêW«Dv,]OCðÙ‡&—Æa:¿zá¶Ö2˜MôG[8ÂSö–ģܰ’|lñ'M†ÎËëzÉhÂ!} ÛÕ°p:ýD[øÑA:ïd;7×u î·æÍ»»÷4Ö;¾ÉJtZMì| R¸_(i,ÒPOé ¨y¸è^¯Ã¡~™hÊÿÚ ãÂ7Ï¥¹ôñÄôËÎþó0»Âü«ÒÚ”â_]H×…ó—ì9ë,CïŸÍR"Rp˜\m³$ÎNe'­Nu„Ë…Ì­"ÕpK]ÿ*bÒU\®wW­€×L¹«{½‚*—ƒRB&üFC^K(É!àRº‘Çþ~c’÷« !ÌG–§ #ì~×áÎIQDŠy8~Ñ6BÌ…wW4ŸÿBì=ÿÖѳɈ¿7ïäI&S$ư'óse yæýïrÊ‘âÊ%™ºéžß&ƒÌ3{÷2užEöÐb·2È™V´~4yåG‘sY‚Ùg¹ýPT5"ü.Û%¿'^³nÜ„òcíkË*Q™°êàYßݨ^zòÔÐE6Ô¨Ìn|þµyš<ûW£þú¸„ï=42îuQCS‰ pÇ\´$¬¹+çî·ÙÚ²ê"-h÷ŽÜõÒ1²¼5èú¦\3#=ýï¿”<ŒÞþ'·C¯N]‡þÝ "¼J©ç™à¥„· ýìå/ë|žÌ³¤~@xyéª??¨ü÷\It¦DOJã ^BøÇ³O¶-µS¥øüsF7k¾¤ú«¬ÎšŽ§‰ïé¼ä‡í»Ãdé¸) _!ýQnCÎÌ-5+)]…³\YU¬ÑiA×m‚¶ª0=4™P¦tjêZ,¼g,pï¸(ßlhl‡hþ6h¯.™FœÊd zC=Ñ÷yý0P[¾Äì|ò°…˲áñàK]îøP˜„”½UP¤óˆd>u¸»^¿TVÝ9Õ?ÕÁæY­÷ëZj¿t¼žÛV<çÅN›—–ý=pK,Q ùº†^_܇æöfÀçág…ð­ð¿—ê;°G}º=Ëy?#HýÜ~׈+™7ÖÑ´¼¡µó ¾)È!<:íAfÍKD†¦Êœz†èanwñ¾Mxn·KcÜqz>“G6»!AÇ­J²ê ^e/ÈÄë}ó‡‰1#5!OÐÕ o6Ù°eoFWÒ½ÁÓÈŽÞ¾"hv(rK»¢Üö#_byBç]!ŠkoTeCñUއc¥f(õŸö[vÆOTÌ>û@NØß¹µÕa= •RQË{úÞòôJÔ=Ox'zÐ ;®¤VÅìBcÛpSñQ4ûtT׌@«ÿ™ë÷Ù¢-YävoD":¾5Ì~æàƒ®íMA1è6Z8àµ=½Âï[ë\ѧ貀3GÌþÕQ‹0˜ÝìeügÞÜn¯±ËŸ²u¼‘QØ&BíiW!õ7OÈXo]ÖpÓîrÖCêž|žÔ#é‡dß$s0#Jîu¥äóŒ:SòyF’ê“„_|Ñ•°(Î+Òt¦T¤áHÿèKîp¸{iÿïD·£”|Ò‹®ß;¿NoD wüMYœö‰ÓùŽò±ÊÖØ ú\KÓçº}°-üý»èèÞºïB©,ô»?|¹|ÊßÛ.jŽÇ÷µ¤¶©ÀäÞ{Ÿg'Eaz@ëÀ—y0ßè,ÉÕNù.¬V\d𛢠ëã}ßõ¼‚mǶ¼ÙfÜpÐw°]²ÕN{ ß]™Å—ùâ¾Þ,p_rßÓ/íV­åÃs‘·ŠÀ;ÊÝ@zÕß~(éÓ·Nw7÷O‘‘½àüÖ#nÛ‡¶{]hm÷UDü^WðîÔÄ,Œ‹3Ù%ˆØS††§½ v»óË—Hôä-%ÈŒ¤ï/þÔÜ™DŠÍ«›™~ÓwžpßGd>˜Ãš¦„ìÛǯ9W†#WxóÎlÿäo™®±õl" ¥Ú¦/÷E±kí–[”±ŠüZV\†Šðšk ÇþÖgÆÌN£©¨‘(Ÿ?]Hµmѹ÷ý[Pïj*uÁÅ v‘ÊMœÐRd•?­böJÿFû¹ŠCgÐiÓ~¹v'º™G KÐã»ãã2>ô]ùU'·‚ ûϱ¿›÷C§Íg¨o™²N±¿¿™òÿ‘¹àýŒºR‰äo*#Ô¾'ñÞXbrú”ý“왤.NöAÂ’~tÕR‹×Ù?…>Ï¿N¢˜^wá‘çú•¶§)*½î vüJ¿†•+rÌÜ„Û{ÿl8Xÿ·+„žKÿ}Þu´âËÜVQ<ÑÁP:5Fü”Ôñê‹ß6ÀýÙÎí|ªtܔƫëÏ̾²¨™îCJhžWÑîG÷!ñOòY¸ ³ç“ûg½ƒù°÷Ç4»ïxzE!cšS ¬8U›eµñ샴qÎÅ0ؼ³\´üQìgoKô”¥ëÓ²5b-Ï.…ó2ån³Á pÉâr,m û ß|çyÏ{åuÌ„#à]ŒG²¿rà»U>ˆ·—~›ó¹xʯ-ùÞ³µíð¿V4·?‡køÿü|U‡Àû\›´?$!ø”ù¼OxúL›s˜e.Â[Vó Þ¼Š¨õß u³ÖÑ ¡½‡Â]b÷YMÚš– nCŒkAª"âw,—âS@‚Þ«¾äåVxŽ.mue=’—,9ô-{1R ·°[/Aú²E F~#ÃfÖ¦T_d½^ö †ÈÖçÊMM³FNJÚ·ÒóÈc³ 9棊|U}Öãý(x›_ðÕñŠ®oþ˜éÕ‡âÑíᆊA(uðXöµåcu?[:¡’5HAŒ%Uj<æì!"¨¾*µo³y jÎæ‹ú–&¡öôˆTð·IÔÉ^ìfnËC½áޤj¡™hŸ¼P»Y }\âî.­hÞqœóhq8Z›7-x‡·ÌOßÉr]B›õ^[&^1tðëXL‹ú†Î»òKCäÑufžåc/Þ)ëðÙ£œF…þ,Š'#õGú ©#2WÜ“¿™jn%÷ÿ‰›ÒöCÊWHpSZߢt04¼ôX—Á¸Âð-ªoÿà?<™oió*凢ñ%DïFtn¤oý£ƒ)œ—3ýFéMo¯ûþwË+¦ã¥ÃFûDXæÓçÏ£å)y©©ti¨çÑ Q:^jh´"®0šO5¬˜3º»ö/¾r²™ŽÏ9”Ö׌:-Žž]Jç?OñnYí ÓtÁû5{`~«Y&çAžnžbª+¾Ðì‹×æÓûÛóÙ’·aß¾>Ô?õµ÷9N®´ÕþçÞóó‹öÁy@ßNæ?\¯}~xëÜÍG. ŽeàùÉ×›–ÀÇùwÀ1øë5‡¨ï\ŠÀyC‹w»îDЭ{‚:³Ä¬·`σ±rŠ×Ž¿^#ÿáë™NÈõ!rýÚ®_º©ˆ~}Ë4.»/sçt«0i Î~~û¡Ò$¨þ Ò§W7Ù߬;ä;ÞÎJñbþvß 1O¤•©ž«ó/G†X…ÅKËÈR¿dÖé6Ùõ{˜,OK ÷üÏáçœÈû˜i/õû$ B¶¹Æ\“G‘îgk?6]”(•_–ÉåG™þìŠÑ}f¨ˆ8göå<ª»°ŽöÛ¡:Í:køWÔ²™™È•ô¡ÎxÎ]GÓ‹h`ã.¨TSBcÜ›dïŒ]hÖ_µøË½,´Êû—µ¦õ£íÞoãÌrt„6Ÿ;3!‹®¾v·ùÕÑ},iÍ}#ôDJ«íz}Ê:|œ¡rÌøûL:^J›÷HÝM¥çfôÿç~H«;RgÿèihsU7´ýŽÒ™! %ø'…“ÒpQ²÷Qx) ?¥|ýÄ·Kt4ó%vîLm¢æÇ«o œ¥ß5Ò}J¤‰ÎTªƒuOÅbzÒ[ÁÛJõYâû }•ÂKsS6f"šg.;%½°¦ëKÅ%_í›)ƒϹ8¾Ì ‚.ûü6¾¥ôz¤íïÔ^HúªÎ®#Çÿ;g.Íi»—åNŸ3 ÿXgóÇA\ÒŸ:8»áج©¾j:œÍ<^ÎH€«XÝÊžõpçý!îöl<…$üdî‡N+µ;Ðv3Åëtº›¢¥\º†Ñ»ÂòëÓè3¸e`¶~ýÃÉ&¯WOÓÐôý¤>ëo*\†‘§˜JWCêsªýÂKö;F]Í?z™ÿÒ™Nå;dÔ™ß!­¿’¹€òå¿áTû ÑÏ})Áwõ¥´}Pþ3«Ê.óÃPpÓýíÕÖCÇuˆŽ†äk½ð²Ît-R:ßAto/8ªzš@—Ç÷ööÐNèë+¹tÞÇÃÏQ^ÎB¼tÿÕ›ÍÛºq㉲±µòŒ 0ÛõEŒ¯ üƒ!Ü Ñt¼”èK‰_;|ðçßï#GÑ¥sоì€ÓÀ´ˆ3¦[è~ 凨ÌoÁÀ/0ÿž7û¦o~i»µnÓ{½á¿ ~ZÞÈðÉÝ>‚ Cé}—\.#Ä|Ññõ{ÛÖ§µ.úûtD>Ÿç¯Ä%Ž˜p­]Yáˆõå’ÓáÙøØÞ½<ìsðŠóþÝ=…H¾o5˜ 6€Ô¹ÎÍŸ§#½SiÜìéBdNJ) Û-Ûˆ§. ¹²6'˜ô§#§éƧ?ùQ¸ÝرôÃEŸãÿ‰?Ç­GLB_hÕ M~r}µ°':Â>h)­œ‹.Ž%ÍW¾£ûúVÉÕÞÑSìõC}§j6Ÿ[ºdê:äÝÈÙÎK=×dÿ#ó(é{SÍŒõÈX‡äsd$uEøy¢ ðÚÞH|XÄoKù0h¸'¥+%÷4¼ô})ÑÒú*¥3%ý•ðŠ4ÿէʳq°d_?' 7Ks3ç‡ÝÆ­eóÊ›kÑuÛ´y•â!^*ó’kkz•wq¯XïÚL®¿}qý-)ý6ÅGÈ_¿¨ïJ÷÷ÒæTC¡vµga<ÛÀ*òëN˜ä?mã}ý¦Þj ‡“žÁ\ŸíñInº^Íy¥_sÒcذc°ïØÕ~‹‰OØ ‡oeí{¼èúm-cëúÓáv)¦}Ϙ<<?9þöæž±/Òf}¯ygj>‘/¥r­hß”?‹Q_JÓP:ZS:ñæÂÞR#Þt§ÄÙ>ˆ-˜ò/«ÙU÷EZÆ­ö–ÓmþAxpêH§ÊÅÐõMzúòë|p¬‰Üaó Ù#CçˆÑñÒ›œ¯ÄÌÎÃtëá¾Þ/a¾øEO~î/<]íGìKX]LýÙ#EÏÑPØi±¬v;^4+Ê ˆÀéÄ–¹ZéIpþÒðùHìk¸ÆŽJ$ŒÑýÆ[rýÞÁ»Àö ùïXø½Ì¼2°f-ní½¼Líïþȧ÷Yò4B¶]73Û\‚0û°ûmòˆ<Íà 1Êßæ|æ{‹X“;ÓL²’ï ]}æ Ý<ÔoÖƒd¥sÊ‹*O!•}l%ö›"}äRèÇÈšU¶íRû7d¿1¾ä¯:ˆ\w–sÃ{U‘ï~øÒÅ/‹Q˜½u4íµ/JÊFuÿ­¯2'î¹í·Qy,_³%¿Õ¼ÚË'S#Qä!¢„ºc! ÖÍ/A‡M˜®Ö4-8³õV$ZD˜Ùæ÷pã­mÄŸÃî~èXT‘]¹è:GÒÆ¦å¼D÷½ª‹ß¥ —gó]5±nôu°,r}†j…ß–bh4Xaáò©õ¥êƒÉÌ— ¨½Œ<—Œ|ÁS¿Èˆ—’y“q$uMê„ì…„/$ü Ùó.§çÛ,"Cå,‘½Œ±ïƒB–Ù³%Ó; °õ’óómÑÔsMå¬Ñør…wŽ'šoZS~^’ÏD|ƒJ?ÏŸÙü˜Â[T7.ÜûsZUÿð~?šS·_>mÿ]ŸÝ,Zéõ@üƒA>;¯­â£ÓqÒ}ÞÕôzP ø~äâVºæYLùéJ_˜÷ÌHr9‰§Ç-T^ï˱n [¹<+ÓX)×,›>á†Ü“t_þÛõÇYà´“?4HX—òñº.º|eN5Ü<©žÞ±‚òñzvoØiR©/KÏ ‡ÞØÁgMRð}w°4Ò~9Ñ%\OwÀßþíûY @ïÄú¾µÌ÷Ûxv5‚µÚŒBèµýIF¶Û~±Õv}¡"µféh1Û#ÚfÖãuŸáEðíSœ±ˆÕVÓߨу¸ÉYJ¾ýˆ°;vH‰|W6üzŒWÚN*‚ éÓ¾ÛŽgñ:å`w@“ R[ι=þûûÞ´‰té@fï Óï‡Î!{™Ú SÐäœ÷'ÒË ¹~lÓY!ï{‹nÛ±S(¸QñrQ ›%R6ŽíCñÍÙÅÚY÷Ä5ŸòÄPð»ÈJú *.þàJPDÕ\÷û·ONGÕDÚÁîR!T¿KÐ*ÔDM·Šôs>CÔv ö)މ¢îÏõ9%@ÚÕͳÑ(&óÅ4MÏ?÷]hBó'Þ(%ËÅh½äÊW99€·õ™öëÕãÑ~»tl¡k9:Y„´·³£³Omûpkô”uh½ZÝÉ««…Òk= ©;F=)cn£¾QçMæÙÿòå“>Fö<*wm })áÿ©¼5â‹Xû}ušN½‘ýŽøôñ¾Cò£¨<2‡ŸMÏFêÒÓæOª~>9mj©GÏM$þA‚ÏÜ5ÂwÐò6t§=½]Ÿ)½4“ iÛ^P¹‰Z¾É±ÊÅcøÈg”ÂøC†þéc7ñDAX¢"Å ¦?“âµNÁÜwõâ“B†xzXÀA\q–#Îý_ùyñ,Y°2b!;lB”Zöˆ.¡û MìŒëï…cІXãž8}zÉ#³é=\\׎¶y מꭧŽêÃ}âÄ^ÝÔ(<7®;Þ/YŸ‹‚;ùk9éyl1u"‡rÅ´ûú[1kÇ2§wF #TÂÄÀ¸ç3ÂFÕ;úŸO"âÍ¥?k=¥•bã¬,:1&BGu"öð ®Eû—ú|”9΀ÎkXEö™]8‹ä.žÃBæ[þöÁ=7{˜r‘~=VÆBc2ÚçH¸²lCV~ˆÄí'3‘³3\9¦â"rßì}{ù7gŒÝ1”Gá*_£¥oQô[F”¥5ào]mÎ9òE¢£®)¨¼‘Õ—2Žª¶‹IáoâQ³ZÛÛ‡µú;nŒ_þºñóï&v Á&÷ôPç]4Álûø×X´,˜]²7÷ÞÎ~›úSg.Ú×ËýµÎDç®Ññ9/¢Kݺñ3s#º3Ò¤m¢ŠÐ»až×–ÙK§¬ÃÇ»æ<*]¦IÍk¤þçÓ©t5ÿå?dÔÓœ…ðç¤þ .BrrH½0ò ”¿ž¦/%ø'áÏ .J}ž_%ÿ~ éÕ*láÿú'ž{d/øú¯î{cp“óïõS㥻ÜØ›gÐsLi98Ús’ùòþÖ£å̱}»àAO«}±ÁCèÊidtföAïGù™ _,`P–ÚòA®/µrѨÊúÛW÷Ÿ?ðGÍ O† y·iÃ,3È~íŠMTÞeðQßCìðìíõ¬í¡Ÿ`«ÐTPrsgn=X7Ž&w[Wí*£ôÝþBöÃK$=ÎÄó„Ï.óš.Óëø{U˜æ{§ÍGàà Ž!ØùªƒÏk&„ªŒñ+ZÊ üêÏ=×u©°O6~8ÑArܺðr¯6Þ_š‰¸µúB_Û?ÞxPÕѯ•HHöG²ú‚›t"¥SýÆDB$Ò•FÝu}d)>±axy.²×Fú=þœ¼µ=Ûnß@Þãý&CŸPpÞ¸ìFãZùÉ÷}™$JÎw†xñq£ì±›Å*ÑPTäöÙÙ5,D•£b^ë5T—q±¹”FßïèîQsò@}nòЉÙ+ѸAì‰Ä´.4•5ôó¼X…ݪՕbÒx‹CW³fl_ˆÖ5CÞɲhÛwE3ý%;:ý*;b؆ί¶LÚÑ}0·’wÛrôÄ”Èe¼ÐFߎt“ÁBÿY‡äy&{Ù÷¦ÂIÿ /%u9%^JÓ]…‡¼'Òõ¥4}èT¹¥~JpRrOðR¢O¥ñŸOxE‚¯=«ÑÛ¹ès*£¾”¦K£êŠ‘‡0Î MâüF¯+’_JÃ[´<÷1=ß펻ï e¢ð`ýŠð“o=¡°?ñý]èaV\ýh? nÚ߯ÜEŸSi¼¯‰qQÈq{E˜y³ÊéƝø1±ßMO¹6,è¶‚•HWW\•'Ýw¡Þ<;Js&¢¥Sz5çºÜþqéepõhÔv±ƒ»¶ÕÆoþðü“æüã£5¼ƒL¸YŠÃOÖêìvÌRº|yÍ=šÉ¥~Ü„à­\³tÝ!”%Ì®c a_‡Ï†X„ˆÉ¼÷‹,ˆ×z0ûx^t oË×BÜ4v-æí»ß¹65bi ^±õEÿÚh„ä{{S|ƒ2ò‘ïtd ÒÍØ6Ï)çB¦ªÊrQ#df-W×BÎ×UGŽª—!/·(ÙÇ<  m­qP”¯èäcîFçÏš/x÷¸•ÕE§û™P½·ôû’Ñ¿{ ÿBµ»ÙQ·èœÎü­kPôs«õ†4hj|3i†¦/ÉJîåh\ï7ã•KÓ±¬àbÛ5IíiDïÂÈÏÿŸêKÉ•±ÉžHõCÚÌØqÒ®SôÍúåÿèÃ`ÔëP~|’óMöMŸB}?sh{&ÅÐò I¥ ùÄŒ~|‚—2ú.¦Ò— ·Ô­HIþ/erÝq|Üõ¿ñRf£Ø!ó°ørÉÈ]›ÒÑQ~ü/zºã;(Ïi¶Ž•°ù8GÇ<¹ß<W¥°Çf–§à~O™Yòxv莶ýïð>Y¾!z~éU­×lÏ´ÈM7Pí.BV_(,ƒ0‹×üžõˆ”\~¥l±8b.Œ_¸>±ú3J9NÜ@¼Íç£9™H8×­o4Œd!…¼‡¦H•`Þz é“íGæH?Ó :ŸaràÝF¥ÇÈ,:ÇkëJ†^“b³à@ñ¸ßñ (ÓŸ•ë+®‡»lCõ {­©¨±ÕM0|wu;Öž/±ŒDýçÌc‹¾_Eã×ÇGõÜlÑ«ç]*U·ÊÞ])™Ñ>:;“ÿ:§úÌ£öœý£$ÉzK§$ŸÕ™?X{&·/ž3]øO|•vµ®†ÙRƒqîvy˜·%¯<èTƒ§õ7,Ó»úaõýχ°‘¸ïæ;—Ňü¸›áp_Ô÷Épƒ3ý: §†Ÿ¢ÕñKà²ö«¢]ÿÿƒôR`û1g”â=¼o¬+2ºIåjP9¦Š+FÓ~>DÐ÷·Žô<„DÝ8Ì#zaF¤Bt²¡^Ù7ðŒQ׏йÓÚt&¨hb÷®}Èvžq‰GKÔ—;"áÁ–<_xeùåöÉÁ#H~§‘l—.ŒTž›ç~Dºr4çí-Èêll·AVéŸU‹‘³¯³æì!äæ°,¸Ž|e«†ÌY(ÜÄvnDDųþp³F)KKùÁjg”¯Ý<»,z•ÛVÕ+oöûe1j6Ϋ&Z•$ÎQ?}§…¢ÃR4¸Y0·^MÇæDÉ®ÈBË WuGC¼](ÃÚ·ô7Úw².ë;N1©Ù¢Ð¥ãÙýçÀ\tª åe/Cïv¯Eû}ǧ¬C¢&ó™ýøSá5d?$WÆ:œÊÿKæH2= ÑÁ0úêɼGpOê}ÚN7=åЇóKÁ]çÂU/y«þ¸'Xþ4:%ÏÆ;ÛÒ5À…Ê¿OÀ×dóœGàáfrü~#UÁbw#÷åÇ ä­÷}Ï»’ç1wZ-vŠòùF{¹”£W€—ö Kš¹wíkFØæ¯ˆ·Ï­ )ÔÁ+ýI5áâ$­Žåœ§~„ò_¤½5ÚVÄf†Œ H#¡ȪðŽ?=C 9Šåµi»Ç»#)ßù=·üÇ7K£°$t óßï›âò¶r.Dé;Þýù ³Q±úZ¿°r;ªDÞy _ñBõÝC¦a«QÃz<öÃMÔ~ßœ˜¶¿ õÇ|ã'zÅÿö½1“il7Єß)‡÷¢¹£ÿIÏ^´zNúiûþe*k®¡ãnÊY]èÌ;ö¥Ä]=~rÎVrèYxzÝÜ]è½æu»VW}©Y½ŠæOÓì]¢s³¸ŽÂ#ÉÜIøŠÿÚ}†Œ8 c½’þHæYÆœš©öD²÷Qy6dÿcØ©~JrL§8ׂÌߌþònô¥Œ9D/N÷ÄèÛ">üë=9·{†pëÞú=KóÛéý”6'Sç·ÑöB’sCåUý^îqú½¥£!¾ ê¼6ïø°Ínßbª’Ü ϱ±ö{ÏIÏk“åÝzLd¬ŽV\ÿÈ±ŠŽçUÙ>J‡ƒ°%çßÿ*¯Ís|„÷î,¸>T°²\ÿî†ÝçnfÀóÇåþ׫tà}@ÿð¯™Áð[Ê}l–{&Äý˜KX+²xðö[„pŠ<q¿0ñéâÜ•ˆp”èš»1‚Q[sS{ð¢8ìÀ«!IÄ 8IîÍA"W Pyöú94µîÈNKDÚ㜕%Ì‘qóKZ|¡>²Þ”KxÄ\CÎm›;’·™/âig)•UÄ^+d"6 %ü»uÖ´GÙuM!vTĦU meA•ÝÝ5§Ôö ºêÍÄY=QÔ®Ø8Vá\ºˆ©Ô¸4`ÙîOko£qÌ]™ÙÄÍÅÎý²“КS(4qm½7„Åï^G'¯õ¶Èƒè:{®%îUžØŽ^ž3¯žÍÔ™ºM¿ÆË„ÓñR¢—ù¯üDÆÆÜ`r%}‘Âghý쟤.ÈóOòi(”èÆHß#¯<”øðrN)=*ÁQ‰Þ”Ü3èM©Dö·©Îeä ù©üøŒç>‘~Èè· ¸-£ž”씕öþTþŒxE^Jt¥dÿ¤òK‰Ÿ¶}•ûOüŒÄwÁûOá¥4?35ß2òˆ"¡7V™¦Óç[š€Êç§ù“Éù”¾t©ôÕ?‡¡™±Osåg¤|«6Žž8·ž>ÇÒr‡²êîä%ØÀØcóÖö#xâ¹¼è ,Ìì'ÌV-ò¤û.Hî0ÍJxû ¦%æᨰ˜¹7lœî7ÝãÌîónŸ5o/»¡OMIíê$xÅþ®Xå¹¾÷zÞn®a‡ÿ·RV™>:Ÿisڈંþë©Æ~ýqw7"T^0/^̃˜ïoúkâE•Ó½zÄu4ng2/EâêÑ V%¤™[É|ϯǫRïGZê•%êÈHuŒyóg²~/îM-~œˆu0Û!/ÊTÉà  Òï^xÝEc_Îè/Eé…ἫFQ>P4Ø㠪ΞÕág:êÒå1CœkP{eÀIµõ¬‘êAç¼Ð0Zü˜' MãKÒDõÅкeDÔ×h)Úœ+EUäÚѹÈßGË<]fYʶK\ÑÃ%9m§kz³Â»t-é9®·WVÔYMY‡öMµê1G©s[ÏA¤pR†}‘Ô¹'ý“|ŽÌ·Œ:6ÒoÉ\HÿDë{NJðòÜÒΕ 8?µŸÑò$¨y/ÊHL«ê7¥?#¾]¢K%¹LTÞѧўŠ7 ñ~ÊM÷ŸæüÝ‹Ôã9®}ø»¯SüM§Fù{#3w)Üb³I¦’uèf\–e9BçHÞ=MŸf²Ú[Úöæ'ª™m Ï~3 æµËny¿Žï†Œ‡S|‚•‡z´ïÈcØîü¹‚ÿï÷˜Í‡Ÿä˜Â>Áö˜ävG8M`¡ºa\˜;›n•:ÂåqGÖ}Ci¸J hUÌøLÇ[–¿ÊÑ Ï^0È2¼ï °—¼à·Áý¯Îø;ˆPP|Šë(“Òä¹ÙßKù C Rgl݈pŸç&B/!R@õÕm{Doÿy/èQbŒê±/B첟¿ó_œèIŸs#ÞoâPì¼+H\þRH4ô#’6IPz¤¶‹î2s!åVƒ¹¶ý¤.’bú'‰ôq¿¸ãªIÈ3b´uGÖó Zjw]üáõLEYäÎzÒüî8òdW?}i†üáÏ k/ på=Ë9§YQô¼ó¬„»JÖíüÎ.—ˆÒ SÇ㧘Q~Û \¿Ž•Ë®”°ÆØ¡J£ï~Iè6T_T]>CQ 5‡®*Ø£vßÊú¿i¨“ñ8tE÷ ê•.ËŠ~6DƒƒáOª<4YÞew^‡f.«k'ã¢Em˦µ»ÑÚû¬|Ú.N4ØÖEç7¤µ·|fý8eZÅÿX}úá ª}Ô©3F<”±O’½Ü“÷ñTJNê–vOùñi}ŒÚïhý‹ô-‚ëý ñIþE^§ô0´ýŽÌ­”ŸŠ6ÿ\‡ÒÃ0žWȯÓ`©àGåpSõOËç ò‰i9ú„ç üM7Gæa3ýüB…-¯çVZAë{ȃ]“2nª7K FpÃ~èËìTH~LííŒ~|“9:Þkœ¤(_¾©®œ¥^¢+Ì:TÚG|6À¢ÚÒ$5«– …µ:b`™(4–û÷{‚Úûˆïi§ŒG«”+œ–ñU°ˆm‚“‹h”´d9\R“Gd gRç’z¸ü‘I”Ás»#I¥fðZslÑ¿9ðùàÑ,qÏ™:¿-ÀcååO–ÆL_’Ö<ÁÏô®8ŠPÑÛë¾Ø~CXóÂe§î "²^ó“M7¢JÝ0]ñ1O¨Lû˜†Øôuóß5 Ž©Éß-ä7ÆëÌý½dñê|H}ÝOM$ïwØíV…”3zY·Ù‘f%)ðèÞ JتL K$Ãs‚iÙÎr—ïèþB.‹žøÓ¹rÈsðUX¡ƒ¡ïi{¤N¡ðóš3üÒëQ\öøåšk”æ*~ÄŠ,”7q^_± U¬+Î =@õ¶µŸ}ŽÑÏ£iíYÎaŠús"ç帕Ð0èuÜÌÁ›Ò{·ÜUæþ¼å)ÞÊë8\¹síbr´74£SîÆeá•ÆèrÍ}=×òÈ”uHp7҉Ό1§›1§”\÷FÆz%x#^JæPâÃ%|9ÉW£pN’?JÓR÷÷${#ÑǼ”!×òå½)ÍŸDé¼i8 9'˜òFÆŽJ>Ï¡ø÷ô¥ä<¶Ý—}eªÒϵ ù¥/¥·¨©ù©ûëZhÉ,PÍþ†ûg/p})¹H?ïIù§~ M2ô¤Õ»Bdèù4o…«ÜE˜œÞþAáµL–yûŠã4=¯vþ™3‰ÞÛºU9zfŠ#ìg¦ŒL„C€Å‡Ûpü²"äRÜa8Ûx§Dq¿‡ë¦g#™pߥÏ|Í=žûÌ7|‹‚W6ƒÌƒAø^›Œ:¤=þÕç|·0! c™^o[‚šqømBˆõ÷“ݧ³Þ÷|ô)DÎ,ûSŠèmý)KÃñ¢fúõwû_ Nh[ã7=?ć¥ò› î@âù¤*Íüh$`°FÉÉ×­ŒÛ[æv³±›7²ëo5-DV0Û‡=ûË‘#lýí£7r{~WL0A~dtÑ'æÚd¥Ö}D±Õ»HËI*·­¼Ê{ý½g6¨â˜ç+ʽÕb®é;~F¾¡®îfÔfZ>¬@=g@W—W1Õ¾òýDãIáýchvÑ[øFôZ µ¼”vÎ9—†èé¼*‚—’œ6®CÎM%x©–à“náç¸ÿI5ãÖ‘ù¸ïäXñÊù3Ý/¬ÙÄò¾†Ý¿WÛHÇãM}ªj¨¼6J^²ïðïN”ž†òYÑr©ìæÎ¿s¬;¼Ìǃw{Ñóû~­ù.p®T—qþJ‚û™ñ ðŠtxZiïªy“ï-ï›8àw<Å«ƒu#]PÝúëýó4ûYšj„°E?Öú&z"¢Ûig ¢Ùwåk¬[ˆX}óöLGÏ\ÎÐÕuH˜°Ôõ®A’øasQ™ïx8*ÒØ®Ž´ëBAËîj"CôÜŠY4²Â{?¾¾€™eÑ_ƒ¿#«®¾µù£†¶¬»†Âö ÜA —Ì™ƒÊ(Ûƒë¿U˜QayõQð—@TéÉ1k * úõXÙ‚¥ò¨e^ñ',Ùu¶k¿J­®Eê¹[Ý>壱föXö„(šCx‹ÆûÑê=K9Š…m)ÒûÇžíEç´µ­–ŠæèÚÓË”Ó×N?æûIsëÒô)ëèÉóMæQFÌT~àÿÂKÿñ=1Ì¡ä|²ŸRüÑ®_·zZµÕï¨R‚›Òt§>JÃO)¿À.uÁAúy0äçÑ~>ùyŒ>êœD’·Oú"ññÒrjH¾Ñ¥_£ò¶ m“ãÏèç ßí¼ª®ˆ¾ôÛ!”Ð}P({-í ½#z\ªî™0p;·WF9™Ê 6ÐÏþ[gcµÑ¡üx«¥…0›ÎÄæ/ùýo öJˆáé=¯µÓ…ná™®¶Ðü°NØnº(:£g7ì+¶ÎV²{ÇšmvbgàTà¬rk.rÂ_Œ>‚›ûÖIu9Ê—Oåg$ÏM;dÿƒóžNH¦ëÖÌõcضGˆD®f³×e„Ýbg90܇ˆø³Wu„5°ûÕôÝxáן•Îă8ά߼35oñü£ÝÈ5$ž;niç $÷v=½É3X6•"-f†ƒŸk;2 %'v¼CVÎQ¿Õ¾“ȹe™¶dýMäñé¶È²N¢€É¤îf Š˜Ò”TäøQ²¶‚«t½Ê”¹w'WE ¢ªZD%RU/¤~Jt›¢†Õ„ÇÀȵ÷r]çBÝðŽæiÞ[Ððð÷,›4-[3ºVÓÍ“üw…?'¡uø‹ß‹y¿ÐÎÂmý;ßk–ÿ2Ñuáð––âtÝ)TqîFï,:áâôY^<•’8k‡ü{Ç^c°S[ºpFÊ”uHÎù"{!™§â qÑÿÊ/eÔ—2â¥úÇ_A»2ò…¤o“+ãçÿÁIÿóÜ'ã”ù¥´ý’Ô3¥3`ЉSúp‚—Ò|SÿäÔüáS¬Õ™Æp焼ûi«QʯògúÁåžt1côÝ&hëpmÉ’~Nç³%çZOï€^Õó{Ëâ"ñhóã…3®Ãð…€’ÚƒT¾i‹•ýŠ'Ümæf§Ý¨s„-8”>j>½‰§ªüy—«¦µ£[`»DaÆé”ý°ß¸jRvEoçz­Ø gÁe››>¼‡‹ýû:á¥=p+.ýOöWdgWQçÙø&*[æ­«×bÖL:<ÿuÇNÁ:¯~w´ßFèǦ±]Å;Qfú„Wn¢g×Ê^­=ƒØÛMÚ‚ · üjÛø*$²Yç8c‰¤‹3ÚÓ¤~ãu뇺›Jwö|ÖÖΨõÈp»þe}ÿrd½;|XþgäD£|´PyN‚lÅá(ð_º‚õ‹?ŠêOnP[Rá9ý¾(ÏùfË&Â*ñ:s>KT'ª¬N˜žŒZI^«Å|¨{ϯpãû$jÜí ö°¢©áØ£Ë[Ð:Kñ)‡¡>ÚT­µ$‹n¡“IzCQÍCt©IÝz̼Ýß$ôÞ(¼BoÐâËâåËÑÿ &½íÑ4 jj¥]SÁI)þœöÜŸ>É•Qö¼%wÄj ©·â„¿Ró¡ªÓqO©œé”¯—:¿¶ïÝ59ß—Ò™.?2kyØ5<”ʽºë‡ ƒ|­M•&ð¸Ö¦¦þØN< ÿµô饇¡òeüe/lù“ KA;Ÿl}<³‰k6€ñvo;ãd8̹öˆuÆ6Ø/hX#sŽ{î9ý¾A×™~îe;®ýFy_GÃ]„ßH$>žÙ}‡ØÕà5ý׉Ë<ð9ð<Ä#оF AOÊà?½dKѺ&øwë%¶Š ðVîJ§ •úö,!Ši5Bdý7ÍNIAX»‹ÌÇ­Û)Þ!ê&«žÕSÄL{ì»Þj)^îá|þfçÄ­:ØãöB qãšìÃÿŽ„;o^/±ÎÆ«–CMOÎ#YñÏ=¥R$¿®ŸÑÿ©µ…iö"†Hy°K72nö™d"K¹‹'ÞßÙ®þy?7p çÿbíM ©ú¿÷qC"„̳ÈLJg›2g2UšH ‘$c!„RŠDÈÜ ³Œ™g™2¦Œ‘Jò뻾÷œû]·Õú|Öý×ÊzÝë·Þoç¹ÏÞÏÞûÙÚù%înPÇ:QþÞ…ê¯Ð‘·3ªÂ»ÁÇ>I.ÝÐh¨÷ëgl04õÅk?X„– ªÏ×Há=µë™ejh{}ôe.t(8›4ªB'Ïͼ=ÓO kW|÷žTaèÚÒV·S“‚ÚŠ:d'¡—wî´Eà,ôiÔñ´ï€~¯;´ã¾ 0P1pü±; í ¢È—µ‚aï¦\æÚ`øð]IÐ÷â Õ¹lÔÈã,D¾Â{.þ‡á*9G1ŸP¹ ôDãÓ馄}lèsÂ9b§(nÑ8å7Âüó‘A}·qú(š×¡þZ©êNÞW¯ŠžiW¿Ã¸wO …7®iAܾeQž¬Þú'™Ÿél¢ž„û%Ÿv~ɃÔ)¿O­§É!íiúЙ~ðpÁ@œh™ï;|_‚;N¥Û÷”¥Ì)³Ürž tß Ï,‚ç—DÜL¬H!¤I5¯„üÞncº*P8ü‰­Ež ¿ðZ,í„’+áµ9Ç¡ÌNPèâŸ<ù•»·PDª¼É»~i÷uN¨ ïÝw‚7 ªô =ŸYOBuŽËÍâÖY¨å*½°6$uÏsÄ6„7áq}¾S >42¾™ÚR›ƒ¦åmΖÖih™Ÿw)?m¤ûBx¬†¡ƒCkR2Ñ:µOó¯[@WâEb—ÆÐýÃáu3]ôº«Ž‹ÑC?µÉÙè7¨ Jº¬C÷v«Q$GÇ;/öµ._†Ñç~µ9Ÿ"`ü:;ÏbÝ8LÔ[±-Tÿ‡~: ò?ý¥—¢u?BŸàí›AãO¸«WàÞÑ>3TAý,°¹"tÿPÝï„Ëç0Ý” ¯›SDO=õ/?Sœ/êóW¿7ª» >m¸½Ø6t¿`õŒç•«ø=¢z‡üÛÕ(ñsÿò/­;|-ÕÁïîA÷¯ÑÕ7~?&˵ææ1ðOu׺Ôo ì¿QäC tL¥œL q~ }–‡õ—bóL÷Lbç9Rà²."Dg‰Í3a~à"uÕZÈ ®qïǯ x?ðy7·ï‘ù|d§Ùƒ½Ø\>æÅÔqoƒ§RÏjtxfúäv†›*äd›:ÏLBž.õjaç"¼HW‹èÉX€BckÏP¼_¨ýrr”òqÌ’.ÀKÙõæ¯ðÚõþïλMP¾1ÙùÏ*_]Û½U ÕÖ;VšCͦìðÊÜŸ<±âL|he¼{* óÅÌË&Äý s ù+r’Õ[Þ[[†]ˆ€öUÁ]ÍN贵ѭ–‚®á“Ã{†¾A;‹2Çã@è~x'óÝwèÿ!sGq? ‘{sºÆdî$p…ÃhÔÅT¾#å0~g‘Nã÷E˜X aÚw0éZBÄn}¦¤.ñÂL¯ú‰ Õz˜«àÚ¿_bæÛ»¸¹¦þ‰CÔåAÂþÒÿ֯śÂÁ®ö}5åèû u—h]ë£ù‡é¿tQ¬_† o­?ú³ý×{ŸÐ¹ T/Eçñqº6¯ó'¬w`ý¥¨ÿ0ºç “ï¾ó¿‘‡ß׆ꥸy|l߮ƥ{»Ú©H«w¸}Ú°ø2F‹¯w ~þ ^'§v÷-ï°¥‚v¼ê;¬ðá®Þ.sš'©‚¢“p+Aá»Ì£juV¤yùÜN0R™:±‘tÕê!ËoÆ\wš»zEbYœgÃC\¯§½bQÜ2"y$ÏIòAtAìüåê.Y?xÂ&þr[–‡¾æÍ‚t;jRKgxba^.òò"d>f×yw' ž Ç9oVCÎHÈe’Ý!oéHò €ýêùäa(2Aˆó ä´g Ô—Pö4²ÆX« ^ ¶.Uå*CùtxÖlÂ!¨œ«*Oõ9ÕÉê7V>B­ãIƒœF¨?;i³Gí44DÞ>[GM³u)Tç ¡ÕckV›æ ´#oµªï†Nn—äUèªÚŽl`H…ž›úªv<ãÐç?Ñtþ@6 d’Õ\ÝVƒ¡íóqó ·`ä.TµtöÀ¸Ô o+L¼¥8­hh“N«÷Ú &aZ‹–<úe ÌØëì£5O¡šÁ?§¾þ‡^b-‡³F±û óÁÿä#Œâ‘‡èuh|‹ÆŸ(.P]›·%¨ õDLçDù= úM±}O8­ßÿµ÷ ­+ÌaüåcŠúâüã°:"U'ˆí{|݃ÛCŒö¥]D¸[êÅ­ñsP̾}";"ñý¥¹s—Þðôÿí…óöâTºdŒßƒøFËM‚]|?3»6 ü³ÈC–[;áæŽíã,Üû˜f¯ŒÞñœÅïU«üí¢A* Q–TTmxß‹ó´ý¢ ŽJ1´¬%Ë+±=2¸=¤)yÚi_aÛ‹Í壾n¸=2™ãõ +ðŒâöÀ¸€äœv¸Ò“û òvÝXš­z /š½õú¡P%1Ý-ióq+²÷åºw^jQ­ßy4¯ìu«Ï•À[Ù½îrCDPEóŠúñ;¨ŽãêâY„ZõdzÙ×ßB=ÛösOõÐÀ*\L$Mª{[C>(@Kb~X´™-´qNS~¹ Ÿí}æDýòßî‚n÷j [Zièå³e•^¾¥±ˆÂ—oa`è~vµB; “Ÿ{J}e FÌ@*Ê^ Æú<ãîû„&ËÀ ç<|¬jn<| ¦¬²£)¾ï‚–`¢¸¥t˜ýyÖg#Ó æÉ“LÍ.°M›(ø'ÝË_ݤ”ÿŽ×cpxùWéâA–ÄÝíÂû¸±çèu(1>ÄñîíOCØGJp¢óÿœÇ'ô­Aùõƒ#œÇG÷ Ìã£þ’h¾ù—O n~ ‹k êvã‹ÃñûÚÐþpT/EýßÐý¨ÿ[*QHëó9¼ÿðåÑF79,®ýóÇh©â©éaüvѶ^3ïó†«o ýsLã™âñoáÎhrýjéß>5VÎ V^)pï$e®ˆ™¶Ï&ÙñgÊ·YWlÍÃàN´“û!]Üé\vƒ-Þ?#¤¤¾8Ÿž±u(ü098#+5!ï§OáÁz(8ÆÃ]¨jE5ž}8j %·íÎ)uCÙÇà‡ej6ðÚSÀ‚ïC3¼5¬ì–W®‡*EŽ®kd[P½âÍæ>sj{ÄOËäoAý´™îÕ±JhúÄxŒŸšÓÆþ¦ä…÷çuò3œþäƒÅ-û®MCg‘Yº2èv 7þ¸½ên‘"a{¡ÿÔ{­,ÚAÌŒò»µáh ´ ËatÂÿÖ]Ãu_^¿?ð >Þ{šq^¦ÎIHaŠ™Se23[0ÙÖÓt\æ—GÎ4ŠUý‡è<>ªÓcz)ÁžTWA뀄ó÷(^ u”WÑz=Šô~G}ÚP¾Bó?´nwÒûýzE0 Þg í&ô'EûI/]Ì>qö,ÞçWßûëÄÝÿ˜¾Šö™¢q"êÛ„«`xà,Îè1iÀꀮÞ×K7ñàçòqûc°y¤w4+0ÝÃÅ Šå§£%°Ô¯ç•dOIw‘k@pl¨¶KÕ,„$ïL¬¹qÂB¢Ç;!\>ïWèoGˆh~XO¿ î\¸õŽk÷/ˆüýR…þ#Ĩ³uþ>ŒÇ ª·´}QÎa šö¿RÅïùåJÜ2þŸ¹ÞW ZEš ð@õÿ’ÒwH[X9yj½®»ˆgS‡tY…V‰?¸}\Âoâ”V¯žà9O髨Ô#Ò kþ‰ì! -Èžz\s,^žÿ27[9pò¥YœÏ6(ËW©æÎ(ز™© ð†"½‰òÄûû¡¸TË®¤­Jµ´ÊG¿ÕCÙ·'kµXxõ8²:E¸ÞºþV9­Ê—5‚Ì b\1jcª"³–h´“¡ºâѥ׊½PK*ipìÜ)¨³¥~"©Sõ­ÄèAƒ)Ó³ïaýÝÍÏ—{õ¦Zý;VɳܠÍÉmmà„Dj†fætšì2·©s„®k'Êi/$AwN}dÓôl™§È?}§ Z³¢Hñ~ß—§R‰Ø`xŸPfôNMø0ó*ãÍ}Im÷8h< ã— Ó憉›t™‰ÏËàc¶Pþ©´˜â0,Ëzn÷ô<¦—b¾m¸þ3Âþ4BÜÖíÿ5Œþ<á\>¶÷ Ç‹XŸ)ª—â|LÿÚSˆÎåã|ùQþBû°ºº×õ]Cã]‚ýOh\ú×!ú9€ÛË~ü5ç„ö³¡û)Ðüî¨Cyã_ûÚ°}Üh_¸hI€• 3Þ/q\îí&Û3ð®ï/}ëq|ñÖoå¶?b`ð'~Ã|Ý™öï3P„ÀáÕù• ö5ïînÞ‰ßßæôíÙÞ=·!<'^Öª"—ƒ5»Ì z÷‹qmFˆlp‘9WwUülµú‹!öšRMÒokˆÐ^»ä÷®÷òÝJÌÀû(Zï·ïí3…”«yäŸ4LáAOÏ Žð°úÛó²¥iH?¨3Cy×ûˆ&Ð=ƒ »;c5tð”Ñ—Nç¹:dÍž±Ýj…ìY{ê%ÆØÞÃ|¯HŠ{©¾P°7ª»À• Ÿ?;-xIŠƒ7©Ï QA©âÑ;?3˜à%í%½‹ ‹ðz§±kIJ!”‹¨É6$„A…»ƒóe]¨ºêô˜Vª[§r VÎAÝ‘6 êÊ=N¥Ä3À»“™¤9 :ÐÈÝTBeœÍÄî²{Ę •dýÅ!‘hãqš> œ6:U ÓK‚¬‚ÿ!tuRΕÌ܆UÖ-yÐÛîä_|úo¨e)Ëß‚AÍ/ñ),Ã0¬P܈´iÀȱòÝ'½a,âv\‘*Œoæ1 =øO¢s1è}ÿ¯~šÿ´ø¿ÍQœ¡¸Gõl0Ú—öÁî{Âáå/¿Rtßî:T%<±9%t0®ßÓMÑý¡è6\=­gbº ê+L¨»àünÐúÆ£¸y —â ÏH>xcýin•í¯àê9›LÞ2Øþ5ÌW·@÷=ù¶œ#úù,üWQ^sÄûc<.Ì\É<ŽÍm õylžé©²ç» 6¯Îiݺ90´ëÀ]×°“§F^Cì¯áÝž9ˆù=sJòöŸåØnÇÏ峨Ž®6ƒÀ–óòÕ‡Ù+‚q1¶n..XªOͯ"ü )¹ÇjV3²d*îñ‚lç£mã¾ÃKüñ%ÿ8 Þ¿&‚̬ò#Ùï,w{æ%*ů¢š\ L"-ç‘N¼Ò³»òÆÞäËZ%ˆB…þÎ ÒPû0Â\Û j¸n†éþùüªm+|rÏêSohœˆ‚†ØþñÁÊûÐTHó螇´üØ¢Š ï„¶Ë¢Ù/.Ð@G(]v¢u:t®¡Œb…î+?SÚ‡\¡—Vܰ5ô=ôÕ>h¹Eñ wëp¡ “Å–~3ø°ž•u“QÆlLË;å»`üçÏïËíðñŒãΛ*Q0ù©dï õë0­ Íl;³–‹Ÿ¾š‘À'í»vê©ÿö§¹NfµO§ÈãÁÿ”þ·{-÷[üÓ¯ ?DóÂÿ”þ3üO>¦„z)®ùϹ|tnÝkÖSÐùa\Ík yaOé­_âûpóøÿÔKQÿaÔ§ ·õ²íÐñ€²ypq,£‘ðï×aë4ßXl/ÍUö¸‹é!qà±¾¹çê×ÙÜêi«ÖáFëãÑåàkW…Ôë\ÿ÷ä=*Ïåàæõ“þrV3ÄýîœÖX57MsùÆc}a5³1©éݱŸ/é¡·Ü™\´8ížÑ+žV\1§ánXÔÁ÷Õ:§sæ§@PÜS‰a´!Æï¥ÙPT5ÛwR¯~×;>©U²‘¨`x”á-™»£ˆ‡?PP@Æ0¢áw²¬“MÓŒI {c“i\þ$äöiÇj?¼/"µ­H csÂÅŽºG…¡4öÜ`­ƒ2¼\»ñ-)r ÞD‡“&¶[AÅeÿ˜÷ÜŸ¡ªÉ.Åø—ÔÐr…ŠP@ÝÓ•¼}ðΩȻ[v/4fÙeyêŽ@Ëž^ò#Wx_bl¿žÌO>nRwCg»kä<·t{„r†¹A¯ÅøÚ¡_ÐoÓô+´hï½=+Ã[‹5—œÔaôq=OþWv¯ûEØ=­¹n?èúS,ɦöÃôïÃCMÅ–0Ç0⫤ZòoÞ¹ÙW,lƒÝç„u B<þ·z)z¢øEóD´>ƛؼºï ·ï‰p(:Gé¤(ÏáêƒØœ=:§ëÇFóNL/EçòQ_SÔ7 ݃ë߯ö#Ö#pþÞØ~R\ÿ6š—b>¸ý¤®Ðy^råéQZp «¡WÔcÃü¾Çó¢ó,£Ãƒ$¬¿ýœôja­(þ…Õ#|õÒôèy€¿ˆ„PÌΰ®gõHч›#3FÙz!¨„zzá•Üšp‘ï­†0}ŽCg¾ð>VξM¸£ÿù™ç+sˆ>±°Ÿýó$ij>Y6Þ„X•¶Ö»ÂW!ždí1—<Üûém¬É™ IÂ¥Á¥Úmì”3Ÿµ»R?ØS•‡çÂCË—•×r3!ÂaGÅÚ9x¼@”pHåd|d’?¹ÄO?EôA6/q0÷œ<¯aYêk4†üfrò¹ó‡¡ å‚ ×±!(rYÌ¥÷E Ä†Že½ò+” ¤wÀûÔl~nÜdÖƒŠ’s|¡|PÕǼ‹ý˨qs‚™?¡Næ[ÊmŽUxǦ" ºa"¾Æ×ÉÕ¡ùtÍHqý>hm²)™å„ö“šÃ3ŸNA'ãþ”&__èŠu!µz =ªã¬+ÃÐ;|EO<°ú³Ï˜ˆÛÿ†Á²¥õÆ"€áióÙg´Ö0ªòâ-×þ`wxîØgµ­ÂkÂ0y¬ïă›0µÌñÕS¬fJ^9ß®s‡¹G«µñŸwÃ|Y&0¹fý‡îïƒ-Óa|„ê™ÿÚ/ú¯ùûÿ¤—Ö ýKÑx•ÿò3%8ÿ£ï7:—ÿ/ÿRT/ÅÍ7ú{£ù&Æ‹„}„ù%×Îããü‡±=O÷¥T&DÎýÕg‡î½¹`uøª»¸ \tg¶| NcQÎ5÷JðóŒËF±åeÒÏãêˆØÞ'Ô§ÝÓöâÔuÚWhl<£ú¶ ‚–ôKEÇfáV¹Iõ~.ÌG#\#͕ž"·÷·½"ƒíÃ@çcÇÜÛD>ìøáFWA›Ãx_ïØgª³#Ø›´\ê³á²ðè¢í¡£F¢ØÞ`Ô1ëhêͧü6ø|±ÝB±Ã¨^ÄÜPþ1¢…7O5;ùß…âóIí#£PZ½üd?™¼’ÿE‘tŒ Þ|føÕ“ß|yÄ, šÑ…¸JujFbU ‹êº:­çNQÁ»o ¯G MºšgkËø ¥×PÎå{$´Ýû¢šèµ M!´÷“C¡ë|8'SÓèá‹Øÿz¯ôqu¢•¹ñú!nJlÕ]"g…UDÏ¡Æ~¨³["L„ãiü)“‰Rï÷Â4Ùƒî™0Ëö ²þñødº¦ °©ÿOFynŽÜ'2Åâ2ô>Fû¯Q¼ º Êk„ºé¿|ÜP^EûKÑú:ª³ õ´€æ}h_ ¶Ï7‡‹ùÃàò3¬ßW@çtÿ£~Šê¥¸~˜sOߟ£¡…ó^LºrgËñóHh êßD¸G†zcÑíOÞƒùÞãüfÜ %n7—žÃü*Ðù2t®ûÄô–¹CV·ð<†î©ˆ³up{â·íƒ¬Õ^Ü„®þk‰‚x|PPq¶Jâõ\Zl™íï@zJˆ—k-ØuÑ îõ§7äøÆAâm(Êþ ÷Êži¼ú«»§ÞP‰_&ÏØbÙ†ßûôâöT¯%¤oÙiI|AàÉ a?G> Èh[ìÌøü žz(G&”½€ggš×è¼!çnY Gäþ /Ï4“…JKþ¤EPào¶Â”¦ …/3ÒX)Š¡hþ™ü‘Ðj(ÙŸîÆ­¤ ¥Ow»êBÙ g]ø2¼²59h¿Ÿ^ÿ´YïôÄö¬U’’ûy¨ªÔ" m‹PÝåggb µ´ ”"‘Pwöõì#a¨ïxæ©- ƳE² Ðøyëus4? ©¸(® ­9Þâ•nÐv¬öIb®tÖ»Ú™²A§lœ@ê•nè2,³SÝ¡LE%|-ÐÓNüóÇVôíwØïõðô?ýU%ûE)oq°¼ƒ¡v•Êˬ ð!ZÑð«ÛŒ^v±.±¹†ß;sÔ¸Tÿ‰Ã0ŸÉ¨méû˜IØWƒê5ÿªWö›þ'ÿR4ÿDñMØOƒùËÌ¢}pÿÉOËëнO¨).¿CûÉ û°>ST¯AçžÐ:>®ˆùtàüôÑþRû‡dw#XÝï}ä¥ç ¢ø}mh_xÇ+ío}Íà¦r¦3"åØß{Gq}᨟ê¿æS«¯ ¢[~Ï4GYœ'ÀÿÃÇ„ÝAÎp“JûL¹Z¤¶YûBP{ˆÊp(柈é4ªaïì â”â2Ü â=¬z;¢Í—|ØØÜ!Æ£öhs¾3Ü-¸¯(*Ü qtw•¥”2 þôáÙÛdp†žÞé$©Ååßêf‚dÕ„OtSøý¿l¼6‡n:à÷“âúÜžT»9È™¾\§7/MC–ýÏ`¦ l¯ÅóÞ³‹."oÌÛê%s^ü 'é;S …ŽNkc§¡˜ØL«?#JžQ½Kƒ²+§õ'$ÂàÕ‰ú²Çô?á‹sïÑólð6gi\¡¿ѸᑠûԿɹuRšT®G„ ¾ñîiP?4x'¿¨ñ+‚&ý±ËÄ}Тrþ²4Q$¼7ÛOø3Úà „|ï‡BÇo…úÛ/¡Kç¬ çv>tç/Èïc«‚^ÉÎG¯»¡¯}ÏŠÑ»,ˆØ>,wã& ]š7˜,Gàƒk¨RiŒ¦¾°•êã^ô+^DD;HˆˆH¶ÿ@æÏ׎?_®¾þ|ˆôÏ埯ßÿ{Í…?×Ðþy. ƒ;nžZ(ꀉÂX)#¨# ­Z„Þ'1ùëÄò0ðé0mTp0Ô¨ïq?âÕ7¥'¼ìB W摆šè Ö¥äÑAõ‹âCÂRiPe‘tåI©&ôæì¹5;ò'öÎön}R¯UÛ´B]PU~±Ð^ï.ô>8ÖÍØ•ƒÅ¢]¶PÍ\øõMÊŸÿ·û&nÚv݇^©Ô·ÃÃ0øàþ5Ç<~¨2û=u=†ª)µ’ï®p@oAuc} %·úóªÂÒéÃÒwC ç(ß?ÐsþUÜŸ?08»ÃÉDëÏëZÝÛóe· ú=’Í7ö ziVd\¢ƒAÍjáê}¨zö2šÿËG¨®½ûl±v z’ÉÊkÃ~ýÛ?¯û"?¾>…êÛåô_\…žœ@jß3O`¨G1øM @ÕQ5§}ç7 ZùqˆÇïSÐãеý3‘ †h¶·¡ênïêÆ3¨V)äkº½;‰‰¯yÎüÎHq¿7òÿý½ýÏk¤ÿó#î5zÜëÿó»$–Ä= •—Ç=Ü¡ .­€{L.}P\F^÷SäèRŠŠŠèciIIÉÿóXêÿ<–þ?eþÏcYÜû‘ ß;fwéú7jœsúßg¿ñ÷š#î^£X… M,‰–&³ÀW}©¹—ʨ¾›ÒE$E†ÛÞdµ¨”!ob̉\âçoÌô™!/Â~??<{õÕ6ã’òtÒcš®QÉÜõ«W?ò’ÁSUü&EI?E·,yÀISj:ðBmIŽš¯  sÃÎ'òÁ¬¯’`ï÷êõXÀ¹·­Hà£UöCæÈ»ÊØŸµüÕHs–hù†¨òÞ1F»ä9ÒNþüú6ÿQ¤ÃÈ>äUEÒ©´úµÐ éìà{û†¦é\ˆ—ó0C: -MÄN«i7:‘N.ïàýfyØó.aϳîäH·þ—’æ—HÏU‡¤šb}¤÷Õ)¢é„_HߨçêÄç ȹ}ì²2(κx(cr•¹0´ë52\bN}a— 2"ëæÇÓeŠŒv²dï>lŒŸá¤ÎÙ¹ŒL\ºîO÷ùôPV´Ä™|m*ü‚™¦¹/dþe™q.iè8}™ÖäËB‡|ò%’ÔouG>Üj¾›£…,HÈyŽZ­"Kξšºñç‘F[RÈjæfƒª<²æ¤²ðÂfù&âëlml¬õÚŸÃ…ü˜›)É @~ ËïbíèB~·ÌÖ“S$ц‘^n;Ôå;rZÎÁŽÚ½E;5 |œ™NEq (÷î µXnjÓô AÆY ¹c[p¦WèJ¥eY×öýÏ2·ˆa`01V!ýóÙ´‡ø®Ÿ}%0ï[œx£{ö×!À滿Kb±ö &”Û0Ç—R×ȵྻÊÇ:¼„¤^¨ß“/a.ýé `8ÞK{„´õ™^»ó‚H¢ù“à& fܽy½ $\4ÝJ5@ê|ñÂþeáѾa{üÈ.zË7¬æü‰÷ÏOëWEš‚Š¦b^8lõÌþ]ëM82Q2ÏjÄÊ&¿{:A¥+ÌЕÔÔγ±—vŸƒ£;_p¾W/]},)€–»Õ[eɽ ƒD>M›=æf·_S?àØ¦¤Øä‚Á|nê£ZR0\-]ݵ Æœ¾Kb…N`¢ç-òû¶%˜Þ¡O?ãØ fS/úܺ FèmÔáxwÑuÑB°r§(••hqòü­E68A’ÚN¿'¿Õ¾³0د[()–ï3=~³‹°ëjÕé§Š‘!y«^¶ÕZäe{ô¹í…)$ÏXŽi&yÃ!Þ†¥²½sðARÏ[ÚH¼ÖÀð–ôÊÇTI‡ñ¯Å'ŠGô}_ì±ùÝÜé¶ï¾efŒ”$’íê°CJi² ´ ‹±ý>áYh0Ñ–‚”M©úIÞ+F^UxÏÕ÷y)ÑyN¢)÷î¡;r?©0ñáJ>T11˜'Bª‰.8ì½}Ô†}·AêæêC”ÕâwA¯2d&ü‘FîÁÏnHÓFœoqÒÒæãk"쀼Ïpz¤½L/7CÒ阺_vw~÷§êZÍ÷ž#½‚ûÖ¹¢‘¾­SÔ.#=e^û“? C³ôöt¯0\ ½Úxm™ûvÀ÷­2éd>y™~ÄGÒàc‹Ì.]²Ñ•Gæ¹5ŽUÑ _T£&¾"‹TO‚©L‘å”2žƒ÷U1¾_´OÎ"k4Tò5n¿Gßxo+Žlr0Їúñ![ _i–[Α<©e(í) zÇZK $­<:bôa‡èQ÷æ ëgNøéÅkÞÌ*É»êÚ뤉ƒV@IÏãÞ=]‘f%e8ìŽÑŠjõÎFþ¬)=öBØÓ}~¢i˜Ÿ g”£…½¹ { æ€}O¿š £p|ks³Ûî×3…~Àþ•ôqˆð߯ÞyÖs ¿2 ²hÓƒH6«†aO ˆµûû}MIJçaÏÜO 5º:+~~d¦>íúª“ò¼œ¦&Åpð¡¯ëÒž9lk.ópPcjß„á(—¸‘Æ_ÖÕž <¯Ì@]øàìs_CÐ`¸ž–•ZÌ奦ý ³_lñÀ®3 §í8õ›,ô/oTˆŸÃX¶®ã×XÁh9»B:VL.2øÑßгÁ©ñ5Z`^Ù81ÂÇÓ‚MLµŠÀ*sbþîÏû`3°1õt¾N*Þ¸Vøöœša¸’JHÛõ ðú²v;Z Õ›n™£J{‘!ЇÃÊ/s1>,ìÛ/ù4Ã#ʃ(Q<=ÖÕQW=Œ —(þP|¢×¡¸CñŽâ¡þ„§’¿Ö¤i_ê2òi‰Më—Mã@ÞKÆIßIÆób„Vk'žïP~$¡Ž+yWŽt<{¥ þÅãOìlY¨Éñø‹»¾ÅwÔS#ÝEJ9¤‘ž÷5.5Áæxž¬¹Fåãiô¢L$ÏüŽ êDN}±ÃчŠè«"[ÑŘçì±÷ñ{>Dßz€LtçÅMÎ"“êæòÖÈTm¿À­ÖoÈŒï5û¾'ÂÈܹ›†íg‘yk¯²ò†ËÈ£42ÏìïÈâ9ï1g;odÙxÕÑœM YI{Ä;ùºÛúyJ²V¦™X°È†|K¼¿éq— ùÞÒ¤«\>„l‘äÆóe"¿…}ϧŸ ¢û'^»¥ É¡u«¡Y ÝŠì¡øÌd+²žï”`'éæÎÊ@e¬×wEvp¸{úñ¼Þ[9 /áÌù-´ {gÎ_}ÃŒ ¤÷XôU€©r³ÖCXf(œYÖmßw:+'Øçø"8¥8[bÊ(vWwÚ´G°Áð|Š™GúOÿ¾ÜÞnVàŸzhÈM‚ï¨Õ4vl€ÈO‰¢J{Ø_ß5“@Ú¾o%|)¶_Ÿ;5‚ôÈÄ·žÔx-x@»lPòy&›q?à`m{@Éôs8Lé˜<}Ž„vdæj”2ï+ c=bPIQ2^5ÑqcC¡ƒ ÞаÌs4\ÃG™T‡@KHí‡Ý1iÐþLüù{í#Ð-ˆ+ €cI‘ꥧ§À  ûs0¾>xÚ‹rŒ&îKÆÅæ UÍhì ˜jz|=u»ÌS~)zƒùvfM ù4÷mÚl)+&›$úC'ÁºÊölS?ØÆõïVŒÝ'ý3)ž¼Ǿ7ñÊ=çÚD›$ »äf‡½vd Ìúñ7^N¤ôûzëÚ1$÷ü×0{§I$ó»bôžÍŒQ¼¡|H‡þ‹Q|¢xÌj+îÌWHþ‘kJψc¸/á=E.ò¾ì/>üëÄñ úïEùÅ÷_¼H!Am\¼é²o®lD*‹ý˜E’¯þ“ëù´Årl`ñrã½KL2êW‘f§c}ëÝH«º¢²K‚Ò½°·bý9þó'cñïg¹±½_•‘¾È¾cë‘c‹†¥X¼‹âxì¬×Õ…ÈoÍ…óÖJÈÇ6ë®hºAd*Ó\.xT™IþYOȎ̵öÑäC>‹*Ò4~ôB–4 ý÷ŸD–ã=ŸßŠø¬z”]i…¬ISÇ\Ùñχfi ‡¹‘Ÿu4Œ½#mÈ–vÙG³¨óÈvÿiWG•f >¤Â©™ $>k\‹@:V1øh/ E\/½Í´(2É òf€J„—Üa•vMßO¾¹˜ toK^PT¼„Ý'Ü~ŒúYúOR´z,ìyÖâC=¯Ì×sUÔÜç`¯—òFÊÂ-`kS=(Ù±÷Ç# Ú¿÷Õ˜sÖÞ3Àk<ùÝ2ÄøÏÊRáŸÁÆM*ÞÏA$Ÿj󹈽üùö€” Hü¬ (z RÕ#R%Ñ ÓGQba™ò»}å¢OÂÁÈË@½ ‡5—Wï-õ"÷ñCsŽ(g]™äÝŽUƒÎ¯¿¨@}/÷ Ç;—Aƒ(:à9• h‘|V(/£¦g«o©‚žTçȦ5è›I‚mu5¿?`øÌmrb?ÁÄŠ¹ö+ñg0]%.kü~Ìsî}ºçÇî“^gw«; $’ƒMEtmÁœ9œÜçÛ·]™§ú”z$l á‚Rjî'ËàKh6g˜ TçÎ Qï²F†ÛeV¶ýÀîÛÂS¡ç6Ó‘ì é•Åz,Dyðÿ+¢¸+»œ¼íB‹k·ì>TG\ø&YV’°8°•ÒZowEÒÆ~ÀxgÐÒ!³ÁLå‰Åñ îu4^lW‰8{Îi_ßÅÇ•Œt\Ø¡=€çS\žÙõ’J‡é‘î°Î|î+HÏÝá6?¤7ù µt>Tq<ürÄE†—5hDu÷"£ÜGdïÞ¯GÆå§M_e"7: ¬Ø#ìt¦ È”M\ȱþfdF`QÑpŒ™Ûµ’F鯆̓|Ôü´ |þþ»·ß3YøM3®»ú Yfé,Ê*¢DVôu)\ @V_æ¾4§\EÖ 2½¯»âñuìÝ_„‘Ÿåéä 4QÈ–½ãšnK t¬Åðq Ñ´dá ^R}Zñ™Ȭ´°’Š«×vžuåÊܬÃêNØEô)õ;Кj{¯ôg^ѱ€Ý³Gf?cж'ZáÀäÁÇçZ{XB¯± ‘kªeZô¡gÀÞþÙî€7p²ËIß,ÿ Ün=yJÀ9Ng«Ø¼ %'Iw.¿ïíq6¢<¥(*ê‰ãzjÕ`ÿU‹ÝïýàÀ…Ï·ïsƒä¶ì+ÁT.e½ í¬²[o¯(žy7åŸÌZãpðñÓvެ×pèÇbÂÀé8âIñ™~—(Ó.õÅÀoPñЈhcÉÕ•E²Í%P÷ÿúàxé)ÐàkoäYxšC¼·Ò+]A;½¶~O¶èz×ÕJ½É‚cçýî°Ó%ƒÁ‰¶Êf0,ø¸ÉÉä F¯ŠË¹‚ Kú{‰4V0µëß [e³£GÕÌ`q('q,»Ž·Z E†©€Õu9ð? 6ªÛi·<¿Ã‰ýe<\àÛë¡c£GÀÑo#|‘áðFÈž[‹ê¡âP9ëÈÏ rºE âðGȃÿÊ &ws¾;¼û¯½åU4ÎEÿž‚ùéÀ@Šª¯ ýš›Ät¢(oþ•'âþý/-?é‘e%azÐö'.¿&¤|Õg,1?©hæòÔ©º]=‘b¼©þÜü0ânßχ~,ÖÎs‚H[Kk)Í›p<âpÝ-åeιtéyaÀTÃðé“Lb¾ü¯mZ¤K“ öËÛØÐ6 £E©oÎùÙ#ãÍú'¾‹e íyæjö"S<±#Õ‹†ÈÌ΋×<£÷ s¢×v¦/?Eæ‘/çm" Üή¦ÒÈ2½Sa…“/²RÒ¡yÛBùz2jf=Y·‹bl¹Û€lçù„éz#?ÖNÌÍÖÄ"¿B‹]×å‘m™Oƒ{?Èàu3’¦oiGÂ` ¥qÇtóŠv®=yA@eÙ´¯âÛU aHò eoºáºΗx`÷YÖ'Dõ4_Dù4^e'’"‰t§O½ÙˆÇÏ¡›Ùv #½v;,;Î¥`q%–'¦Mø¨Ã„ 'Œ Ý=Ñ€Œ4tlkÏ( cc‚¢½:È„2òú¼7ò±2%ÍgVχj®<[/Ÿ!sš¿’XæñùᩳA͉ãȢ撊Î(9²,˜TÇFáŒçÃgÕY³‘5-h?O%…|;h˜ñã«'òÝ 8_WŸE~nŸßÿî15²•ü2,B€HZ÷kIßÿvÂŽHüTÞ–ÀŽ9»KÂ\ü@þÁJ‰xy(ɬö%3ص ·ÿ•ã@óî1ÏK = —È U‘ëƒÝÏjE룭¦ÙW$0©ð½–á,rîOÓþܬ’ƒŠáÙ©Ànöꤌ¨2pä/$2w7œÏYà 0Ë ÔÞ•xº_{Ù€? ¡y*<‚§UøŠžp´!{È\/ì—(í6x4-¿ô$+É÷å÷ÙA:ê„×Ù—¯AöäkÛŸþ;ä/ت8–‹ÀÁÄMÛÏphái{ï+8âè·¦/s”)‚ƒøÏ©€Êe { TçJ¼NŸõk|}GXÌAƒõðq-;Ðl«ŠÏ Íí{J¯*¨çA×™úÔS«B8fÑšO& Är×eÃÁðÙYÿÊôM0<¸©¯Ì &4­2uÜ`ja}‘‡bÌJW ¨¬üÁâ€ÅÕX=n8^)~üW] X]äîÌ;62óò¿}ká{ˆã›¢8Å@nHËœ ŽtÈS÷€—„â¥ï£ >+|ëÓ—|dP×¾=èC†Büý+/$ŒCQÜÝÛ•É‘±Î‚(>Ñ÷AùÕP>,V`^‰+×ÿ‹ïþâ¿ÿ6O$äCœ~úWžxøh¬w:RÓ/Üf~S©Ëýr>$¼ ¯›ZGß®j{öo>ÄÅÅX~xé¤p‹¤Õßù¡·Y±Jò]dèNvç¨òá%[Ї.>?$Ó‘´!¬þ1%iò#¡È™þ‘-À¾Eƒg¯“Š66V! G6oó…#Ëâ[Ýx‘•¡¦O½ÓÈ×{éyw,"ë.'8Õ_  £HÎ ?ÚuB¥oB~™Vh¤?iA¶iëØvQÑâù—¢z)9µIУ-Ø)Üúºö Pù)0*¨Á®OVW'“¿Ý•¸Ô")Ø-4æÿÓ½:3Ý}¼aOÙNÝ¢1`~»¼ÇχöNU´í|`7šã_êèNÑ0çëUvÀ]ù´9@‡ x3ئ»o_þ¦k]ß“@HWmóm(ˆÊØ’TXñ€xСúÔÑs éÌö•‚¤Õ©*ž´ƒìµæ³«1 o;ï)¥v>Û%÷ƳÇ6é(„#¯b—ÓÞU€²JèteG<¨¼–<íÜjº“ûŒ…%A}yÁ«ùX=hdß~'Wx´<5²%öî㎋ðºé½¬eׄ"0¼^òXë[+U&¿¹¸’&œ9\>«s`šØ9{Ì%ÖY-‚Ï€Å̼’Sǰ¬tju»ï›4¶S àùpÜ@ø¡{$ÒÁ|«1Ý"¯¢ù!ŽßÐøã=ôzBþDþ_yb]ézSÙg¤wç³SõnôHŸÞ“%oÃHKœty2ìCïz!χÙ GJÿ͇^LÌÚ¼Xq.¦E ì9‘ùâ=ÁáïH/ ­ô®½EÈbþÉxÑdÙîuãKá8då&w¹²Ú6F­]:‰ÏÙ^Ä]=†|—äNNñG~†–È} G¶˜*•… ÛÕ,êy£.@|ôÂÙ#%×€¤;vò¼ê4ì¸ÏœfÀä7«-:Xõ`gØ~±‡*‚12]/ÓPF;gؽ[—X7®Êõ‚=1òéÇÒðúˆÙÀ°ùÅh‡ìëÏc>’î \—*„…¶w£ó»!¼~'ß¼÷¶~†ÇŸ).½ÉC·DV­@˜â¹¸yî/M¨ý°W¤d“‡£Ñ¼o&¨ßƒ¦b€è©ê µvLîÃ9eЩ ›íŸq½ûíßr@ÿ¦ñ¡å¶Ã`x˜š::* Œìú'~-q$«àß0i5^÷NxfÜ¿S^·PyÄýdD ŽÓgó)±‚e^É ÇG°vxéµÅF¶Ê&r–UpRa´ì×%?¸ úvÇ$ÉKà9~•ü‚P¹›–˜}<… ¦(³–›Lc¼ƒâå?áPýoùÕaÑ¿çÅöçl5¡XåÅæƒÿó!ª›¦íÜÞvAêÜ/9T¸ ï´Ç<”ù±:b³´+KìÓgbõ´~ˆÆ·(ŽÕ¢ð0Fâuž2¯k=ÈPSÕ+Gmd„ùgZ¢¦2–ælÀ+ºŽLHùGÒÌB>¾ñàX*þ[ßa/ÖÿÙ¦€g×8âY=M¥µÀû®5íø:‡µKåñid]Ž÷B‰ÅIdè#¢Y÷ò£àá…häâŨvÆ ùýQŠÊ~½ ˆz‡¢ü$(äXŒ„ÿ¾ ÿrvý©7•à áÁç‡ y{Œjöî S_j×íì£Ö=.@?ÅÏwþK0$HÌh|†=a™áVšìÀœZtTr“ö6sŒÈÆ »‚$gJ›!pROh[Ý–î|]æÞzVàO˧Òþ×£»œö~!¥hâÄ Ê•T8h“ â§ër(ãAÒvºIŠë3HÈÏÊ­›Y¦Ç‰š@^mNÿe€(¼=­ÅBsm†Þ§ý\GîYÍlô€²ø¶ÉŽÈ@PÉL|>yl'¨ÉªI^?œêýÜùåW“@㎧´¡q5hYÞ2ˆ'Nù‰ÅŒ '€?ÖÚ‹¯#žßIk¹ FY,kûÀ„ŒLâVL}HŽðW1‚ùî‘ÿË`Q?ýúñ‘°Lnµ¦8Öñc‡ÏfñƒmsêÁÁ;ààd±vAþ ð&Ò蔹;)tÆäz"ƒ‹Ÿ«34b÷-ʇ„zé¿ðˆÆ§è‰Æ¯èõÿÒgP|¡x@눨NJ¨{¢õB”×°×qy!ª«¢ßÇê 8^ÄòKB¾Äñc×ëÞ§'<°¸Í{ûz(—ìu¤–Èð¬6dè9OuѧÈpWIræ$¦—Ž•¨O¡ÁꓔǷ۟#S ñ­Œ™X~8[Ë_Mñz‹/¿D”ájbDÍK©ýˆãñõCê*߯3€¬ÚdGËy!_%ví݈XÆôÒo;<_çþ E~ø¼ü5Rl¾œò³%C¶­\ùsXí…Üæk¯€D{à¼K€<:ÿ’Êì÷²ÐŠz“ïÉ@amôÖ§Õ(ï¯é¥JÀ®îÉ’tþ@Ë*ÁzËú)Ч'ÔK ëˆ(/¢¸Gyð/^Ä(bÏQ>$ÐKÑ­#ò!¿uÚÚ÷¹©öØç*W&‹ÏQ>¤#sš-<üW~HXGÁô´Ÿf;%¬ÁHûo´ÊQ:׈eÙ¯äȈó9O½|'¬ëÃ饓v½3IÇ0½t¶sVÙþA!2?n·T"cŠÌ× ¬…³)âûáÌò?ЯA¾†ô1]ØFÖ”©¥÷Ó!ß’êæô #ßÔãJ‘Mû£”ÁmKøº¿µø°Ýu ~ùJ÷ú5z|ãdõ»QK VÊi-JC•œW öb±{-|h÷-Ug[ýµÝ‹!’ÖÀ 'z3hÎÛb‘¹}¦÷GŠsÃÈ`¯˜ÉÅ_2/íÌç×H½#pfÞJêpëè)8©ÄÏg+÷/àwïØº“úâö?Ö¼DX…ù뵃O—Ýe0•ML`»RçÍùMÛU@Fc¯:«CÈéú\‰·…k•cRÞgÿàó±\Þh(·ðÊú ȧÝ!µZó ¿Á@B‡Qʹ ›b¦_§æƒ1ÃO¿+5º`rµhï Ê<˜~õwÏÈó;[×¶vÞŠù V‚$S$ÁÌ`#DtmNn.Hïù¹·JãÏ÷|ØLPR%¹¼ÐE†.4|S&Ââ:4N$ìk#¬¢'¡nƒâ=Q<ò!Š'T'AûÊÐþ1ŒÇÐú.ÿÞãx ãCœn‚ò!šW¢'¦›¢ù"N_Aë³äËa.KÇ÷¡úz ÷I­"ýl½úÜ¿òÃÑ{|ã{µù‘ñ£é#–W‰=; œS{ðõC”Yĺo‹!ÈœûÑC‡£OÙºò;/~D¾ø,=|À~ ßOss[aQYiÙ¾¯]¸Y šþ6ã%€¬)^ÑïÓGÖY‚wTÞOE¾kØÔæû!¿àž{ý9¬¯fk-àͬÇO Ú¤”ãóKâë7: ôßé-³G»Vç€,˜´/Ià=Pr˜nï¼”¢•ÒNëÁ@moK"Ò© 4·S2 ß½º‘aÞR?ìοpIöv80º-Ú2S“V‰Þ…+ ÀbôÅû‹Ë0ì£\ýñØ;{Ln9§,§2ep'JO¿ËØ^$*²ˆøj¨X~ݸO®ðL} „÷‹³ñâQÁ ³ Z P‹ø1½’šªzvA‰ ™fsâhF/Ș}M›¾| äÔ8eHo‚ÉF¥öIP ù0;^ ‡'ø‚gŠ˜9©€x.|埖l>KP™kM-5›B2ŸßK >ýÀ´vÁ4ü‰FÚÏ“‚–èX4O]hO·ÔïaÛºyÎVÖ™p,ÈËçI~œÝ®Š”Ã6dbŸ”:-T› †Íƒ ÷}û´Ø;`jÿ{àÊw0«çøaøå!X(ÏŒÃñ¡¤œá`åLùN¥ldn|{0'˜ù ö¶rÂ…nk…ÛJÒÀSú»b³N¨Ž<$ά%GÕ¢ˆO^›ÄxÅaÿ6Š3Â<ð_uÃé5ñ!¡^J “¢y"š¢¯cß'Ì yÍq}{ÿª#öÕ |Øý»›Eži*¹+{` ‹› ûÐ>¹:—úß›õc}ã6‹Iß(0}gÜu¸À†ÆχÇÏh_wÎB¦`÷œ25q\ÉÂ;ãC´O|Á:¤i¬U ùÒ3’äÔŒ,¯ßͺæ0¬FèÞ;Y¬­úwÊ"ë”ábräÈ÷+½ñû‹ñº¢—·Í Ú«åÜTÑÄ>²tì ²@2ZyÜ(¼ÈvŸ+)¾ä‘k‚Ô£@)­¸hœ¹ÔÁ¦\>“@“•¶6ºßèÿÞÃ5»gvÊñ7[cåÕ´ <5`Z2ß§[ ,íJ§Ä^H›]l‘OÃ,ì›Ù¼øÆL¸%_ˆ±×^~ßoö\À÷´·I«ó!Q¦È›“€Ð*}$³‹2ˆÑLÆ8¶ƒÄΠ+yÝ eÄT‚tp€ÌX-דâ¼ÏG› V®lu§fA1[OgY¥”xÙÝuwjRÝ’õÈ¡”ëVgÙ–ƒªÒóôž9-Pkd“rˆ„£ç.ñmMæKÝö4 ÕßöTÝà-èdítÈtÔ½¶ bÖ:0 ­=úâÄ6Ò¤«Î§‚Ñöîn†?qê§›Y±³`Ú«5=÷̦8n”WÀq‚‘笕`ùû÷ÓÕò`ýësÈ͵¸À{[Œ:… xš4r饟ÇĹdcÆæñ«Í¿øÍQ>ûO}Þ(îâä• ßßÂpŠ^¾Š74NDïÿúžý¢§f‘ÆNÍwjÄx½í«FyíŸÁåu/ä…hüˆ|ùW=7·Ñõ¦ ±¼ ë;ýKoY (D†y¶-ß“Ú"#£1‹ÒÛx½Tÿ׊sÜi|Zhï%ScdæÔš”ßÍ›ÿ®¢}Û.‡îM7àëòT´{"KÎ k'Üøn"ëMŠ-ÚyȆŽ}—¯òý»úÙü,äç'þÔ­dëШǠ)–G½¿Ì: ¤fŸª×å™ôú;­øP  n`LT‚“E¢?ÒêC›Òvè8Ðð$]vSüŠÍ_ì6ÒXú2õwþ$Â|³¯›voî¨lº‰ßN?án• ¦Å»ÀSM¶ÏP!øÎ,6–»»ƒÀáË\Ë¡ d$PqèÍiÉßyZ¿ÄŽ& S|Љ»¼ríT@rƒ;äjC8H‡Vš¯ÛLì±—ob¯züá{>ÃÓSpÐT¨üñŽ“p(9ÌJäîaü]fú( š+Süù¨˜\¹È® ª“);½~õØË ¦ÙAãÌfœúGUвpt!ÏVËóÈY$èå œ÷%Þ tõs;Y%Áð}Erú²É^ÜWj &u‚î>ü`vÑšL<ò XÈSgü0;–BœYat`­’±”Û’¶az¾ôßfàѤ%Ûi°³ö8[rN7¿Ÿ•ÍÇÃMÆn_{I§†(²¸B}³®Xùì½uX[=üØÝÝÝ X,[TlPPQQ éDPBEÁ 0HQliP±óØ]|3Ã=Ãïx~ç{¿¼^¯ïÝçºö™{=1{fž}¯uÇVÞ±5ú÷¿æ‡ÿ_û…¢ÿy­*HXÝ)SÂâçÄóˆq®è—⟇çÿ·>â¿êñ‡T/÷ÓˆûL©óŸôðûjüæ5øòMÒC©^ú¾°Þ‚˜Æ’?KûljgÅüPêÿ‹ûhš¥i×ÂL¹bâ™ ÝóoköÉz(ÖKi?ÍÃkºúS¿„*?ö)·Ñò€Ï>6öà‘ŠK5ñ¬¸Ï”þÜÅ»Ÿk ÷’Õô9ľÿë½5´å”,ì`®W_¹êøÕiÓîLD­žJy+Ïë¡vȰ§¥á÷Pw¥ÙE›sÔ_4ÞEÉô Úg¶ ɨ…Æ~Íõ‹tÐL^Ó»ÁìÍhYw£†w£!huâ(C?@çFïdÎhg¡¼­{ÄZt°ü¡µÂu:eypí3º:íô¶¶¼IVê»´óÐ+¸lÐ]CôqÉ[jÚª#úE:U¹%”`à°Zž> 0¤ý+íðÙöU¶rÔ¼Ûé=Ãÿk«FS[8e‚î6ÈO*WîÓåÆhéL{cö”"å:è>j‹ñŠZ…Æ}”0ñÇÚ[“c’c“>ç_¬Æ”}6¥'jbê»Õ…žùZ˜þðN›ð} R»nü‚öa˜=ëM­#vP=Á®›™/æï² ¶nnˆ…3‚Ö|Ü‹^ôÒÜìòj‰O›ªNŸŽ%q>³‹A£|‡kW›8,Ÿ°ú¦ìãh~¾Õjˬ^b_g|û"h?3ðû[è´¶ÝW\†^ î y> ß=,î£|3Ífù®AÒºûëbÜøë>ÒÛç-úã¯~(ú¯¸¿[Ü¿&î¿׿¸î¥ºÈý—Fýs?¨c¿ô ¥ý4ô9Qÿ~=ŠýqßÍ?ö­ŠýD1Î\úaSæ.ÊEïUú/hZãW´Ïô~%þ¹Ä{‡–¾´V¾¸ð@ÚDå‡N¯ŽEBù±Çù›F51ª_óï³Tjö—NÜÃÄ¥BêC¼ý6Eþý"i?͇ýëšËr½”?>ë°ùîÊŸ×ÕöH²TùkÏ Ñ£{(Ÿ»h¤ŸÚdåŸOý‹?¶‘òÄÚ‘?—ï4BÝM;ê¿ïúJ]Ü’®rºøY)´ë¦)htË=M·¯šÖ_Úßö+š›Oöèh–¾OO|RS7Ö|e¢ÒíÖ*Ô;ûý+:hÍõãgst:=zµª;ºjÚ:+ä^FÎJН¼: çG·øõó×¢·K×ìOëÐWíÜÄÙÏÐ_gd§«¡xöÄ‹¾W®`ˆ\ÔcÝ$[ »ÚÚíFôŒœ¤XÒþú}Œ:SùèÉ@9ŒÖÓ4Žì …‰Îßo_ž Å AÓ£Ì;`¬žWíòõœß¥ÛOí:ʪ )m£æa’³U´Æ¯0¥¶{7Ÿ>˜z(3íp¤¦«5Õ~Òef5ÈJ 9YýãÇoz£Œ9SVéë¾Ã\§’õ ý1ÿiã¦Uå_°p~ë#ÊjYX”ºcÓ«× ¦:p½Ý’>Pÿú%ò¹ª–^sìŸûö;4³ÓKδ܃åOo§4WŒÂÊq=gÄ>Æj­;šæEB;qÊÉs°¾qÏæ†+Ö ×Ì… e‡¯ Ñ´ûö>pU.ýÔõÄú'IDù·?ßûŸôð×£øyÑÅ>È¿öÿm?éè#þªÿÚGüUç?÷¾êý¿×MÅ>"í3ãåëKš·>Ò_Š“¥þ¡ÏÒ>ñ’Èî•Þõ”ê¥ÿª‡bÿ_¬—ŠžXåÈIJÓo•×™™üe‡±òÓ¬ÝYRóç.|S{©›Öì÷Ðmó¼|åwª¡+ÆØ*˜£žöö†ò'÷+j×üùaÅæM>ž™¨üÓ­éO_ã=íU¼˜–jŽÚÍçöÓ›ÛuRǤ¨MC½c>ÞÛ¾ÝEƒ´ašgú'£Ñ÷q¶½eÍÑÔt]½®>cÐB.É7ý]c´ücÆð½hýHwЄaQh{¬ÎÜ-†©hogýÄ©:ƬKÒ³T@ƒÛ½ »Ñ=Dur¥’9zµ¶é}ðÂ<ô~ô²N«~ãÑ÷©_½à!Û1`tÊ“etÛd¨WÂe õïš™™Ý#Žš,5}ˆQv; ¾‡Ñ ËžþXÿ j*&ŸE—oˆJÅØ×óNvýØ ¼d_Z9\Lü6OÓ“UTÂúö¼‰)ßú§ühŽiùöû™‚¹gmtŸ„Ê'õ wýö˜£²^õËNWÌ}>`Åâ€X°êÎùß±ðÅš#3?b±¿‹Ewã¯P·ñ_h<¾'–nË3]{ÚI½÷Îh„ƒÇíÙ\9+ 5n vâüñŽÝÌ-ulþ/÷€éÿñ'ûþçÁ†øýÉUU÷­«ª<¬ªz´¨ªêñùªª§C«ªž¬ªzÑ ªê/‹ªª—w«ª^ϯªz“\UõnPUÕ{Ÿªªu«ª>™VU}®¨ªú:§ªê[bUÕþUU?÷Uÿ-2µŒ¸K*ãæR‘ÉêÅËd úÈd =¸Kü!“51Éš–ÊdͧËd-be²V=d²Öî2Y›¯2Y;=™¬}¡LÖqŠLÖ)J&ëÒU&ëê&“uû$“õX'“õÌ“Éz+Ëd}NÊdý:Édýe²ïd²AÚ2Ùàë2ÙÐñ2Ù°p™lD{™l¤“L6êµL6ZK&“¿"“Q’ÉÊdc[Ëdãd²ñÉd—ÉdÊ9ÿ÷g´þ„zªðh¯2é®iœ+UëyµT‚M©Ó9ù©…8žçÞï¯ÜŰ:ØZÓ¿z'³^¹&nÂjë®›Ýu°lç…täæbizÊÑV‡+¡®ê»ªþGo,Þªtaê,qÊ0^¡j}T_±è1Ê×}6{w;,tyœfÔ)‹œ<¦”f„BmÁ«Û‡çF`Iv»ÞÇÞ}€fbóˆG/ÏbEÑ—ÃJÑs°êâ»>›N4ƒ¶Ê€c{¬9µíÔ37+¬ëÿÓt§s t´=Ûî½ùºfrªNº ¡7üuäbÚÐ;pulUq è=jÙgm,ôÕvºQxú—zíiêõ4A@ªÕÅÓc±!zUdò¤=0¤yÆÙŸy*Œi~“´SunF¨Á4YÝyÚ"ÀìêÇTÅ o˜ÿhVSZ=XÒý[Ñ}lâÑ lªë±4olûœÝ?¬Ï|ØÏ·ZUzê$´ZÝkÚÛNüÙ-,NíµÆú|uÂöƒK–~[ï ç‰YΧ†]«,ìÂÉ™]àv_Öös¼v½šucɳ‹ØÓoûûìG·à¡Òþu}÷Dxz­ë"·Ë^o"¾=6Üo}—ðâ¹[±ÿýÊ&óZÝÆî®‘‡ŸŠú­×âPçwAª/¦àˆpùÊèsÕýÞI9F¿Ýûó‚ûØF†© dlÚó¿¾NCè‰Õ6¯Ä±ÉqÏ”#¬JcÜℇˆxV®Ù´q.N6ÎxpÿD®3n§íÙÑ[…"æå•oíâ-çàºàЗlÄ+ȼ¸G‡³Mvû®3‰µ+dyg; ©ëÔ¡G7oBÊÒ¶‰Wôgá|ü¬/Ü¥"½k§ñãôc‘aü¸cøF3d¯±ÎZ‘Œ s½¸iODvþí˜+Ýpqí›É=—§áÒOg§ «ëàJÀÕOç=À5îÍÞ‹[áú«•›šŸD®ê©¥É[7!¯çMï冊È–çpä_>±sÅÃÓ(8â:>dÚDcÑ÷þKQ4é£zC·~h?³°sX¹ %9ý2+;¢”û‘Úê<ÅÍ€Z ºÆã–Æõ³:X8tBŒ-[ÎUï¶&>I¾9÷aj·A'"¼Ìh~sÿSEi§`q?¹`ÙWX>ù&KN…5ñ—qøp—ü‹Ø2±ÁÅ9ZWaG÷·uR°‹ýƒ{pH2*¶6ÙŒmt_N“æ…†§«cûñ–Üg%gÍÙïßÀå‘ɽôÁ¥pKU-™5À»ÎÝnä;ît=³½¯O}³èžY‡áuc?w%µà½xÏ»¹í±ÿûøùÚz} ÐWóý𻢯’‡ÃÜ*¸s}üí~V=¥À’ÆMB,ОB„åßGïÕ5´š]‰ð&}{š$äâ8­ãSmø?Q FzwuD Oï‹ö!Ç*¾o‡xzÿlÃ=¸¥ŒÄq_7¾ºˆ¤åE­_l¸””ÏïÍëÖEñXºm&ïaÈxb7t‹[3d­ó«5¸ ^儸ö솗”u¼Å¥ÁÛ•îhŒËÜÓ³y€«nï´~‰ëSÛ$â–ïý÷[.8 òˆ—ó½lÚ]{[e¦nQঅæÇž/›+Å×Åωþ8Ÿüj‘¦Iö·|¨óª2"K7Þºþ²ÅDhê )ÈÔnƒåÎFÜ“L†ÖâkåFuÓ°òÚì æµ{aÕÛ{všV`µÒ ½;÷×›|¿ÞU£æ}N—T¥Kþ»úCçKQwoA›tz Í¿V}æ²Q¡©Xljí˜O sGGýÂpsèÑóZO׳g ­~0"}2¡óšÕK®b4橽V´» ËxógqzA°¾+¿òÍOlsïyשÃ`›.{uõ¶nÝÞio7l›zÞÌtL(ï¾2OKÿíû;ö™ gÃ÷=¸GW£tûsojcgè÷ÐÁµ…{[ÞÁ±7QéÐ"™=<éû^çËT3˜Á›âýoRßþ`ŒÁ*ÞY33ág`P«Ÿß_à¾dW8YGVÎV’³²DÀöËE[r>!°"ßõºÉx·$BVežþØÞ ¡ÅAÚ>éâØƒÇEº7Þ]ø!,óy8ñ%{p—vˆM!\ÃÑË;r‘H.b¢Ý,RCéÐi^–¬£pfɉ $•"¡-¡óx?ìwkHâÝðú¤^¥ „ ém—e™!ƒ»£“ÎÈrÈ—ìm‚ìnü‚¨‡œ ƒ”ûßq‰[š½êâ /[Û¶ã*§êSjẠW‘»¤nGN·djíÓ-ô¯=plÝa»QÀ=õï{åPHë²(øõ2“=(NÕ¾ñ3Ú%´nrßÒ=º·újÅQ$ʦîßV¿o9nóaã”[´åV|*ܬO9ˆ;þNúEr~·¬ëɆ£ òѨÏÃq×læt¥W Øv·aæ»JElv˜¥»eUW¬Kä‰ð zÔyÜè´WZ~üÑj›ïÎŒl»Eò?Q×D¿šJº3éÌȈvŠP¦u9aøèWoö×Áxòñ8â?Ðù&Çܲµr¦Ñº™Az;‹‹JbN–ôUôûytþù7.àDªŽÉõšø·ì²g˜–ÿ.¢þªGù ¨Q<+ÆÑK(žãkÉßIWT=o0Ûº›äï«imºÞµt^ñ¹‰ñ±X^kræ/=’¿é?ôœùp˜ÿ›ñáäF˜¿õ8qzÞBXîÿyK·õvXÏÓI <×6læÃÈÉå°¥ëÙººÃ«ãdp ó8ª­>¾.ã3œzÒÄ+·ËStáBçq[µéç«Uصë¤iÓX¸¿éÀGðð˜n}SÝTžt?û¸¨Ó(}+¼‰7}„´Å¾ ív9n ÆAZŸG¦µá˜Š‹c9çZ}\AÅ»×0rBHܱ]BÛã(¿ü¾Hñëñ5}î:û*àä­y{£›"JnˆÙ4Li¦ñ`ÄÑs<ýãÝ™éAWpæíÑÐõ%ò¨–©ÚHâß~~)éBÀŒ´ÇÇ\{‹ôískó AÆ7{ª}‘å)$Èè¦.vã e\nú†óàå¸ÚŽÿá?áúèÃÍSï„"—SÍFç#OçÇÎ[‰Dx0(¸Y0}w~µmÍ3*Š˜5\ÞQ%^h“_K:I÷©£aßfôêVMì?ë‰/Ä<ÓˆO–>‚ Åëf!õ÷ï‹ZÁfÎ#`å¶"7É›¸`vÒY).µï"(u ŸÊeV‡±_U# àLüäÖ_HÔ°‹W½†ËáNÏAÒÅNB`/çšnwZÁ{xYü‘ÛÝáÓÜûåØá=àÛ’ÿŽãà€k>øúãð²,ÓÓ›¦ÀŸž_àví/¯»8#è’ɧÄJm„ô½8F§}B=¹îuîŒcͦ¯© x‚°#õÖ4Šˆég2­š|ĉ’F¼àâQÞ;ê-FT?>Á[èÔøØÂÂÚˆåÙ­ñ1Äi'Õûú< §¹«éþiÎÄ7á¾9 ‡'õÚ¶¬AÜb¤ðQ}[HÊ+NypÆÙãýJKŽ ‹îëB¦b¯¢ ÈÙ¬ån“Çé¡FȈiž¸r°QáÚÍ­kÎDÚ!·ÑÐîÛ# lSϱă[Œ_Ç{ yZtæâP!­ë‚⦳έTJ¸‡?¤C=”¶7» +n ñ¾9i“Ê&.?Á]2n/|¾£{Â"”ëÚU¼o†r”;t•šq¹¶ÛÏüzXÍÛŽÄÇV–¤˜­oûÜâ{6ÄNŸ>_ÞæMª̳ѻüÌÂ^Š'E¿›LyÝDZcÉOÆÐz‘ÿïz_ÿ FÓçG“nŠx ñ­éë¯~*úõTЧRœ<Î;ê¢~ΤºŒ¨×³èúÿMOPÜ*Ö“Dÿ_Lóˆ< æÃ)ºl§_ŲÓg‡.<,éå*ÒuÉo‰·Ä¸VÒÉõN 5€tå:É‘ºÅâ½0Ê*š°˜a¾­¤ì¬^Éë;{ 6ñQj³¿jê?€}ýl<“á@ëlÛØ©Nf½®ñã ýÏ»¬g!] ׌Pµ ûcØ%¤+Êp?»a|„wðt‡sexîP‰Hþ¡ ¯ïý¹Èq¼¼91‡Ïö]Ϲ¾çøÄ*¢<ïÈO¹@Ý –¨¦åFËŸ ¡B¸:Ǹ§v%y(ÂõGg˽Áq!}m€S4o4çÜ« — –‹†Ç}ÛŠÓüênS€³zïËîì‰Dª¿$óÙFýuH {u$vXs¤7ã% ™ôü.¬8v¹NV røËëˆË-"ÄUAGâÇòM]9$=˧çUÐg^bUÅÍ·Sœf%”&cé·Ï^UÃUžBêóH×DÝý@ŒGŸò׸t­Eª_Šþ)Ƨ¢Í!,úÁRòwQ¿VRQÌû´Éo¤z Õ/Ä£XWÕÎÎã"ï÷XMº$꜔?)¾Û­¦(ù6ųâ÷¥úÅi:ä÷b=v== 4ÖMïóQª³š^v¦K4Lóæñ™Ì)¿²$µ¦û³Ùø å™QØòàš…—M[Øß9Ðu9 p*Õ¼xI®9vD+¤,žNœÍ‹’°³« øØMº¿×`á¢Z¶´áÅe£?*Rá­q­gTðkøt]àÎ>&á+U0‡·h zÎăA‚l5BȰõr¿j"ô´û¸G;ÊplzÐå4§Öä\î×âlý¢p‚~¯Sk]u¼£æ#Šêž1+/Ý,˜X_×ÏÚÇšátWžû >"¼®ÞœUêÆ1¹„Ë\†s;»ñŠ€dʇR:€4 ›‰7·®Eº].S·FLrKYæU'¹ÙÝAGÿ´åæâ’¾uÁå“Îõô»_ÃUŠ¿®w¹u²¢ý}Ë‘ûcÎk¿eQÈ{8G¥ RùþB`‚‚Y #ö5EÁÓ±|e…üÏ;% Eûv=ÐE´n‹¾­÷Ý|2Šùpô”#ŠùlaK”ôù¸unÊ”Ÿ—P\]*¸A+”Nç?Š­MåÓÃO®ê}ê´Î<)o\L~&æi¢Šþ&梞‰ú&O÷+GñÜ(ÒÍ‘ôûŽ"¿U ¸vÕ-&‘>‰þ-ƽ3¨~*êñLòIÿ¨+^§¤ƒ¤;sšëUº^±N+æ» è~‘þŠý±^$é!ݧT¿¥óˆõ1Ÿ^K}I)—òFê—’žŠuWSzNRÞHõcKÃfÆjsbaÍ©åžwsaÃW µ“°…®Cª³Òy¾j|jÝ9Žó{ÛÕÞÙN”·íÓ¶írk!œ¯,^ÀI\é¹íÜÏÈw`÷ÃÓ}W-»‰½T7òžüdñxxQ¾âMqµÏæ>I~ð=z}ù”éqò×#[zŽºÕò/+qqí+¡ð—ÁއœÂ±¾å#:¤"œ/ó¾.Å êWœ¢ºn4Õ¹b‡tà"þ)8Í—õ?sùb“%®­b‘ÐU˜pŽ_Fß#ùîGõè18O}1_̤|!‹§…§×Mü“óÁéèø=ª¸D~u…/ÿ†”àÚ)Ÿc­—á†uÔºúc‹‘+Èþ$äMÓãyÔŸÉw©Õ2Hñ= øòÓ" ‰ÈûX‹ÓÑ[ ¹T…{*ÔN»G!éT‘f9çyv(â£ò9ò(¦ús1/j—QLõY êWéRßK\g»]ÔÞY/ß&Õ?Äõ.ú‡˜÷‰º&Æ£¢î‰ñ¨èCèsâQŒKÅ:èW¢_,&ë(Rÿ‚øVÌçÄþ¢¨k«…p¢Q õO¬›RŸgÕœî÷¦.9ðïGúœ¤›t^±>#æÑb=Yêc’¿êç¯,¯7HÈÿ7&G¬ïly &t=’N’¿[Ò÷­é9lîõLaߪ,ØònÝ« [·ÕÀ¶Å6_Ü G¾š8:Û£æ7ã–œù2¦ÎO¸Æ>½z¶°vr«ÚÿEÜûúOûZ¯mMÞ(L«¯ÇÂz-©©ÇÞøî³·áŒž8x|\ž{¿—8Ì/ã« ¨^n6,þ0-©ó{+󂢉ª Ý)÷~Vk?„éÍm?Ze("4WýuL{ N¨ -NQ=6jŸBþ¬ SwØîÁÓç!vÙ˵=v6­ÅßÎF¼Aûîó­V㬢eB›Z:Hœ8*ñŒñ>$ rÓ)|õfømÎïøòv ¤ãJN9ï]Wò™O’Ûî“€ BúªlŠ_/öäx,.eì]1ðC&®P½å_Ñ<|÷°ï*ä&Orßsiòöšö:1ÇùTÏ-àeµó.𬮷…ë2>+žF!_¶nã"7Å9Ùa¯QL}›bG¡±‚â×¥ÓrÛD DýA£„ê1¥õ×æÌŸÒÅ:µîi̇©ò¶éÍ\–â<5ú]Äz€¨SH×D¿uLÔ=QßF ¥ºô&‚[þ‹}±«wîÃÕ“Ýà.ȸ]>Òuxñ§Ã xSþê£ðxŠ_évøR~tôþHgžPV#àú¨·î?FF€¯›!øã“A³Žê ô_˜oŒ°n¼ NDD–ðN^¬œÙ{¢›šŸ¸4³>b¨~LJ‘·ËÏw '\BíC9Gu°” cZq)Î ééX¤+[£tk2²ºåwA–cê®ôõÈäx:.ÊYYä­UÀå&Aïc^öÅ•—^8äŒÇ5ê'Þ0ÎïñöÁ^了Úßxš€<â±ü†|Cæ0òÏò yG,k·(| ¨?THýÖ"Yrªö*m*üñîÌ ñÉéŠi?A1—…s±ô;‹z#Æ¡;x9J9'åƒbþ'ú¡T‡ù—øs8凢¿õ§xU<Šñ©x>ÑoÄu/õÉÄ|Qƒê´ba—ÜoTUSƒUX¡!4N¥z¨¸?ç×úª¨“’ÞQ\)ÕGž­{±ÿ¯ÛO¤"¡Gõ,}Ú÷"í×!2 üØøLêÿS?SäA³³?F~ zž–|ûÂÉ£¦ßOç³UŠž}¦~ì_œY¾5}‹#ü'Ø.´/”àŠºÊ…†¶ÎZ‘”‘ˆP¹ËKÜ3p¬•p#¼ÎŒÛZk]qœïVÚšàdw!FäÆC|FŒhAî‡#†S¹²xÄ?ÄËó mÎRý#qÕôÊÓw»#ɤíK »"ňOøzà|t“+ºÉNñsùcÖ§-³ûŒÃª·æ~i´!q7.ù¾)øk‘=®„î»0)×H_oP_8—w·³\ ò…´r „²[öÖoT¸E¦–{zù,F1í‡) ÅÃKÝÚ¡”ö÷ÜâY´Ž)Êø¶ÿºHÜæXþKYTðaÊùÜ!¾¬ÂõN0 ºŒèbQü}g߈þ"Æ›¢‰ú'Æ›¢þ‰þÖ—>߇tK<Šº(ê¦èÇ¢‹õ±Þ#æbUÔI‘D½þ‡^ŠuÖ_tSÊ;i>QgoÌ!–òHÚg õ)ÅýyT•ú’t^‘7¤xšôUÒG±Þ$ê#åbHŸê¢b]u#ñ“ 塦T’âZâ+kÚ×g³`t÷qvDZ…öÕØ md98\x>jü½á5ýÆ'B;ñ“•pá¼.Öá Ü6÷uSf‹]´OÁWÍ1ðèÎÿ áIýô}üÝ· oa›Ðøü¥gÝÀü¨.sˆÖ…?í_ äÛÔ]^!ˆ~·BÁG']ÝôôcÂvËß+î‹ãÂô£¤~c4·ø;öC ¤*äÏFœ[³eË}!žò³âÁsžßø .R„pj ÒøòÏÖ«HÏ6Œ!3‰ ŸãB¬PÈDÎÁ ¢ßÚâÝÏÛ¿ø-G¸F~xãŒPB.öl¾‹<>ºPwFþ'¾áx…ùÀÇEówip!=Š}ò:JHoR¾q‹ï¦v_‡ÛBôÝã»ݕB;FÿÿeOoý¿ïIU?S‹>C¯Ïú?_¯]óúFîõ:ü¿CÖÞï(“5¸ñËÈýeäý2ò¿ŒÂ¿á_ëHû‘Ž&'ßSßÌœp½QÃÆô\'þÓcü?|&¡†ÕÿôY f_ýO_åÿ5ÀŽ“Î>ö\O L £nõñô‰êãÙGÕÇ gzÿ*½¿”ÞF¯Ÿ©>úWÒû›ª ô~]:6§÷SèýwôºÓû^ô¾wõ1@ƒÞ§ãé\ýW}qñÕ¯ûV¿>éÐÈêc$/Ά޷­>̪>FUÿãÞ“béèëKï+Ñû©ôýœ¿Ÿÿàjz¿-½‹Þ'ìçJóËÓûFô>=g¿dúþÊ¿ßÇ©Òê×ì®>îÿ^}õ«_ßöºúèDçñ«þФ@úž#}Ήž—ߤêc=zÝÖeÀ“ê£]½OGßÏôþEzÖ§#½îK× Z}´w ùéó~§ÿ~ «_·‰¦óÑïíMþäw›Þ¤óÑ:÷¦óú®¨>n¦ë·§ç¾®ßï}õq ½¾µ'}Ÿž“/­Skú¾CÃê£]/=-ôy{z}ßPº ^‘šóJÅOÀYø·(mt´ÍÈ®o¢c¶ÞXdºÆÚ†Ã7þêŠÐZ½…NÍ»ÕPzw¡öÚ 5ïVCÝêkè-^C-z»ñ:m íáºfÚuª?ÑGÔÓÚü³«ªªâiªQý"P½ dÿvœD÷=©![Ò±-;ÊÄÿèÜÆ¢^׫ž¿ ÷“7ðfƒ 6Ø`ƒ 6Ø`ƒ 6Ø`ƒ nø³Ál°Ál°ÁÆ0ÂÙ`ƒ 6Ø`ƒ 6Ø`ã±l°Ál°Ál°Ál°Ád6Ø`ƒ 6Ø`ƒ 6Ø`ƒ 6ØàF6l°Ál°Ál°ÁlTÿÍrüßÛVý·Û6Jüûìꎦ(Úõ‡É)‰ ÖHÑ%r¢1Z4äECA4ƈ†¢h(‰ÆX2j)Y£$KN²FK–¼d)HÖÉR”,%É’æ“æ“æ“æ“æ“æ“æ“æ“æ“æ“æ-Í1Zšc´4ÇhiŽÑÒ£¥9FKsŒ–æ-Í1ZšC^šC^šC^šC^šC^šC^šC^šC^šC^šC^šCAšCAšCAšCAšCAšCAšCAšCAšCAšCAšcŒ4ÇiŽ1Òc¤9ÆHsŒ‘æ#Í1FšcŒ4ÇiEiEiEiEiEiEiEiEiEiEi%i%i%i%i%i%i%i%i%i%iޱÒc¥9ÆJsŒ•æ+Í1Všc¬4ÇXiޱÒcÅ9êŒ9²ÆUcÊÕ˜£kLùS¡ÆSc*Ö˜J5¦8#F8ŒpdŒpáÈá0Âa„Ç#F82F8ŒpdŒpá0Âa„Ç##F82F8Œpá0‘,F8Œpá0Âa„ÃG²á0Âa„ÃGƇ#F8’Ň#F82F8Œpá0‘,F8Œpá0Âa„ÃG²á0Âa„ÃGƇ#F8’Ň#F82F8Œpá0‘,F8Œpá0Âa„ÃG²á0Âa„ÃGƇNɇŽh1Âa„#c„#ŒpáH#F8ŒpáÈá0Âa„ÃG²á0Âa„ÃGƇ#F8’Ň#F8ŒpáH#F8ŒpáÈá0©1á0Â-F8ŒpdŒpDƒ#Éb„Ç##F8ŒpáH#F8ŒpáÈá0Âa„ÃG²á0Âa„Ç#Éb„Ç##F85&#F8¢Å‡ŽŒŽh0Âa„#YŒpá0Âa„#c„Ç#Éb„Ç##F8ŒpáH#F8Œpá0Âa„#YŒpá0Âa„#c„çÆü„ÓR&kcH/ÕÚÈ f0ãÏÐe3˜ñ¿gü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒßjü!üà f0ƒ)>3˜ÁŒße´”ÉÚÒ õ,Œ-´`€`€`€ß ¬õ×[è0À 0À 0À¿4426b˜a†f˜a†f˜a†f˜ázk µ×n`€`€`€~3øïïÊb€`€`€ÿð·8 0À 0À 0ð{À{· à 3Ì0à 3Ì0à 3Ì0ÃþzC 0À 0À 0ð¿þ†3À 0À 0À¿ü·kî 3Ì0à 3Ì0à 3Ì0à ÿøè 1À 0À 0ÀÀÿø[Î 0À 0À üðß®¹3Ì0à 3Ì0à 3Ì0à 3ügà?à_ïc€`€`€ÿðßߕŠ0À 0Àÿ àop`€`€`à÷€ÿön†f˜a†f˜a†f˜a†ÿ üìÊb€`€`€ÿðßߕŠ0À 0Àÿ àop`€`€`à÷€ÿön†f˜a†f˜a†f˜a†ÿ üìÊb€`€`€ÿŸƒæ2YCx´W™t×4Εªõ¼Z*Á¦ÔéœüÔBÏsï÷WîbX l­é_ ½“Y¯\7aµõ ×Íî:X¶óB:rs±4=åh«Ã•PWõ]Uÿ£7oUº0uƒ8e¯P µ>ª¯Xô åë>›½»º@3±yÄ£—g±¢èËa¥è9Xuñ]ŸM'šA[åG@±=ÖœÚvê™›ÖõÿiºÓ¹:Úžm÷Þü]39U'Ý…Ðþ:r±Nmè¸:¶ª¸ôžGµì³6új;Ý(<ýÇK½ö4õÇzšÇ Õêâ鱨½*2yÒÒ¼FãìŽÏ¿<Æ4¿IÚ©:7#Ô`š¬î6ñh†6ÕõXš· ¶}ÎîÖg>ìç[­*=uZ-Žî5ím'þì§öZc}¾:aûÁ%K¿­÷…óÄ,çSîÀUváäÌ.p»/kû9Þ »^ͺ±äÙEìé·ý}ö£[ðPiÿº¾{"<½Öu‘Ûe¯7ßî‚·¾KxñÜ­Øÿ~e“y­nãw×GŽÈÃOEýÖ€kq¨ó» ÕSpD¸|eô¹ê~藍ßn‰ýyA}l#ÃT26íù__§!ôÄê ›ƒWâØä¸gJV¥1nqÂCD<+×lÚ8'çN<¸ "×·Óöl‡è­Â óòÊ·vñˆsp]pèK6âd^Ü£ÃÙ&»¿}×™ŒÄÚ²¼³ÔuêУ›7!eiÛÄ+ú³p>~ÖîR‘ÞµÓøqú±È0~Ü1|£2‹×Xg­HÆ…¹Þ Ü´'";ÿvÌŒ•n¸¸öÍäžËÓpé§³Ó„Õup%àÇê§óà÷fïÅ­pýÕÊ‹MÍO"WõÔÒä­›×ó¦÷rCEä Ës8ò/ŸØ¹âáiq2m" ±è{ÿ¥(šôÑ@½¡ ŠÛ?4ŠŸYØ9¬\…’œ~™QÊýHmužâf@­ÝãqË?ãúŒY(;ù²òÝeÜÎiyeÐJ”ó—oúwj­ÛÞ¼CgTÖ^=Ôc[&*9ÔéÓ¸Þ.S¼tð)ŽYõ|éÞ –Í"• U5qtJ׺S¶W‚¿ÛQç°–ÖÛ ÇöÜJ Ãî×O8—ƒÅƒGªËicAeï¶ýša®Ýy•­W¼0»"üÀÚ•P¡u¯²Ï"ÙÀþ«„g“‹Ÿ_ظ ïÁX\Ðåî—ÑXrXX¸Ðl%8 –7è«ù~ø]ÑWIÃanܹ> þöG?«žÒ@`HIã‰&¡hÏ!Âò÷êZÍ®Dx“¾=MrqœÖñ©6üƒŸ¨#½»:"†§÷Eû‡cß7C<½¶áŒÜRFb¸¯_]DÒò¢Ö/6Ü@JÊç÷æuë"x,Ý6“÷0d<±ºÅ­²ÖùÕ¿\† ¯rB|{vCŽKÊ: ÞâÒàíJ w4ÆeîéÙ< ÀU·÷Z¿Äõ©Çm’NqK÷~{È-yÄËù^6mž´]ˆýl¯O¡p¼Í¢|ù(jüpÌ‚îýQT²²çŒǘÌ/úýJ§4à™7»ïYí`®[ÄKeÄ7·û-¯³¼¨ Ê—›;©o*"Ãf–Ö{„Ê7n6ÃÞ­»Ã;í톿ðyxȈižƒÜ`þÖãÄéy ¡oþ,N/HZÿ¢ˆþ"êÚ"Ò+QœÑN1 ÖÌÍñ ¸E\Çäºä·âçÕ®guË î¥o]Ùb"4‰ÿW ýêÍþ:Xi£wù™…=Vs«^?{´éük¦ë­ÜÑ ÆßHצö~XÑî2Ö>iË_Öµ®—\+Äë8ñóét"ìÛ é…nY³Ò%+¡W|fÙA‡YXO÷g`y­É™¿ô`HëqcrÄúΖ×`Ì/—œP˜|è|)êî-˜õÕŠs˜Óü–uFÝ»uVäß›ˆ‡6“þÚ>{»0}ßil]ÝáÀÕq2lk±SÆîp$ÿÝÞaêÓµ·U°ƒxÂ%Ì9£¿Ã¸E¨õJ›6»n-ìÄQön{èí–-¾W ÏK&Ÿ+µ±oÊ_Išåð.¬=gô­/ðÙÜÂÃ'ɾœŠ=H°ÅÁƒÆáöÂ:Oà´s¿ßo‡ Lj}M*ï X+M„.ëz²á¨8úõyVÃÒ¿ ¨ôÌíˆ ¿?ñâxÂÑëË©vÚ}Ü#gDÛÛÖ¿{È1ïc^öUX8~9NºŠx•ZŸÖ-ÂÙ¾< ‘«M4Ùd­ù»²}n­q^!8ÿâç¡-ož"]yÛôæ?.#ƒCO5D/Ÿ¯à¯f«Ï#gèÂÁ*ûq‘c9—}¸ì’Ÿ“àQŒ«3rÖªrÄõæÞ/ÇïåÓÚ¨Ï\†ÜRߨ§W—%wyi˜ ò£-ÂL/™¢àðš øœë(tQ{g½|ŠŒ¿·â–ŠÕ£RŽ@‰âœì°Wþ(mËb.(}.8n&x±¤(·„0rʦnØ{¨M{”}ÉÜe…ns"oØ[åÓÏdZ5ùˆò›[ל‰´C/‡÷q§þ¼ë‡êÁu¾ªÑÆC±•—Õqá0›ßÛ®öÎîÑ]¸1ºÜ ¦17‡l­Ü…5Â2e« — oTµÏñ®÷õß`½>Ÿta­Y\4sp2fR|8ƒôS<Н‹ŸS%Ï#Ư‹‡¨¿êQþj¡ù±çËæb‰F;Hñ°&}_Œ“—Óüb\«•Ü=Ds V^›Ô¼v/¬zû`ÏNÓ ¬R|·[M±&Þ¥ûXSbÞñû¬%_GóênQà(ºÄSb¼»~Áèî܆Áé³üŠ€¡†@|0¢¸Ñ¤#4ƒ)Ååf<Ý.x óò‡<¥a3cµ9±°æV×åm~ØDñÂfNÅü_¤Õøë˜þCÏ™‡CÑÑÄ5»SàHq¿S¤¾ÖÙ $u¡ó¸ì"u?"¯q­g4<ºóöž¤Kûº ïâ íŸtá—ätcèø^µáŸÒÄnœ4öÄ;%S•wîõLaBxyIóÅQÁ-û LøÙZàøJØþå‰Smù@€óËzß5ßú1ÁÙëëÙuDæj¹#žÿ”¾9Î’þ'êø‡@Rì’ªôvw:°o“Muf"sâMO?!ýêÙBŽ¢9»1¿D–ÎõC¶ý¢vVn—qqÔõ¨~];á’ óp%S»Í¤yGqÍ¿$û[¾nl¿\´%çrÏõ˜ÓýÞTäŹÝÝFù§×ÎÚbÑ„ ¾ÄŒÒ)DQÊÂáÙÉOPœ±À¬áòŽàÏb^”„ÒËUG7š‚›×Vùe[×üô‹B¥>ñê‡ÛYÌÓ9òsüƒx‰ŠM]CísqgÿÏ[º­·£r¢ Y òä >âÅÞ½3GN{{f+ÉYYäa×{—ï‰}BÈ-jŽbá}sÒ&Ÿ6p yQûíXò«pDŒËŸß¬µÇ(^Ôè{ÖñW1ª5ÖpŸg¼¤[«I7$Ó Ö‘ŽêiÜë¤3¤|Ú2¹F¤&´îÌÆ ʱ9_çq,×ýÓXò·åä+!,œ +òsQ‡¬(ϲ|z7Ú-p@Íç)Ÿ´¦|rS;ÁÑ`Ó傯ºé}±™ôb Ýßâ7Ûªç f[wƒ=Ö1­‰I§·Q|½­ì²g˜36¯57¼'N…¶Ü~í—v¼6^? ;H‡]”OŒÙÜ®ì¤|r×þ‡÷ØaÍ—Ô턽”xÔç"¢ð¬XjJæex½&x“ÞïÒªð¡¼Þw˜ÒLãÁËà÷0`HDÞGº¦®a£Œ#ÅBA_ÓóäôGP'!Gð„‹ƒ3{#ÄðNÿ ¡Þ›åþé(޵„0b "–¤˜–„_žZÉ/ KD%*Z$³GÌ ×½Îµ»_xð8Íe¯_ >òѨÏ8Ë/»ïpëþ#nÉâŸÞntAÒ³&Aœ„"Åú¦ºéœ#8/¤Uk&¸íf¤¯š^yúnwdlølÅ¥bÈäÃkY ²îvã#d îØ 9Ó8¶..Ý0>aÂ;\¡øåå½¹C„Äy¤ûùßû/9Q1 …~|dеî®8v¹.Jg¯és×Ù·´5›¾æ.§o¡ëKäGà]W%ý÷ˆîçzr™Éq<4~<–kæÇIá<îäWYãcpq,Ønõö¶S|xfÁ…ç£Æßƒw…`;Ÿ^}æÖ5_Np|"ñüz!åHñžègbüªõmŸ[|φXùiBÓßI:$úé:ÒU}Šc ´Øâ_¿%L(Ž5#µ˜a¾­¤ì¬¥LÓJ›f7ŸÜdmlnúÆ¡±írl1µÜÓËg1lùðâÚNØ5Ö[ÓüIì(ÿ´£xÒîMÞó`_¿ÅÏdØQþdGëÝΜè*¤ ¶»÷ñ>l)O¶ ÚÅW¤ÏÙÑz´£û¶ÒöØÊ/_õWR}ÆxiÅŽûfó +ñÝvÊ'ùªÍìæp“áFùê®…çí?wšÇ£íK »‡'=¯@Ý ßÔ±2F=Àþ™ËF…¦œƒO5Nšߎð|4ã üø_Ón( lÑš£ ÎRøûð6\PÜè©ûu i "‚—½\Ûcg#„iew„Òu[›3?~Ö„w 8ÞZlœTÕ´x´ü"ãc ¹Ñ­5ALÑ„rŽ ËËÊä‹8ݲ¹Á÷ë]Ïe…×wÈãL'>a{…³ öz͇¶Aïu3f#ñ•yZú÷½H:iÚ4vÿ3¤ðì3<iíxbkôíZž…Ì‘«4Æ©áB¯¥ú‡ƒv#G«¯ÿ´¯õp‰_ÕÃWâªvBn“%®­b‘wo8çya(xùá…CÎx+ðB˜‚ÒÄšq¹v(ãÈv£JmTð2|•œXpîk¼h½æj<£¥xtõîGõhLª[šÒõ™_ àh+¬(>ÝDõ)ïjžz'`×CH¼$³¿7u‰ÏÃØJq©ÕU¶Q¼(é—PVÔ„Ó‚FCol¸í|66÷¶ÿxwfzÐìà¢äç7kÙƒÎï‚àJu ·‘©ÍþèŽ[šŒíÔ v=¸fáeÓîÄ{ù(hÉxpê[ëž<óæç¤ û¸EØËPÞ\´ÿÓÙ ûùtdxð«Dg(|)ïðãÃMß·8Ä{‰Í#~²x|Kë™ð_%,8ac×tÏ-éÄE¦¦x;8Z(Ô DS¯È T¸í¹¿¶­qÔLðDã¼W-`?Âø²¥Êy„ ËnÂ?dì]1ð#"¨y‚ê­'=ÔûNE—;þ°¼‡(Š¢ùòÑëRĸ_‹³õ‹Bì™úg÷»‰8Š+â©.zær¬h?$ðå'µËH¼#*’6Õ™q[kRh¦†Ê¾…;n@Z>!X‡´Û|"·é¼< ¿ ž¦ß»"sžNjà¹È[wØîÁÓqa̧ûyW["›Ë’?ßÙ >¨jÒb .RÜ}yÔN‘/à å9×løB6n¬Èß1r…r.ò¨>SÀ³•ùFØ)ë‡"ºÞ ¡Àƒ›T*«Ãš Q.KNÕ¾Q…;ü2¹‘†J!¬^‡»o…À÷§Ć=FÏM8*•Ö‘Õe¢)^ã-1NãHíMåÓÃOb—½÷jƒ¥”/©Q?a!ù©X—‘ê¡TOë­bþ&ÖUÖ÷ ã0¢ú·)ñ»y“ÊŠ|×ë°Ø¡læ<–T¿°z¸Sîý¬6°¾+¿òÍOlâ²pîVaCy£M¤—ñÖÁfòëÍOû¾;[(´­´üø£W•ä¯[É_¤¾B_>0q#­gGâ §{Å}ÜC2$ÝÚAq¯3ŸÕè,ëᇃñó.Šowó«¾^2öĵá︦¾²C%"ù‡*¼¨nµÎ¿¿›P¸ƒNû²f¥#qàtðèÃÍÏÃO(ïÕÆAóaêå~šۯGnís G„r„ü«œŽŽß£Š@¿eQ›~¾BÐ(!CPvÞ݆™ï̉þ¦Ž“üÁ÷È›#²[ý¾å#ª,"B_—NËm!åLj'Ã( ðµ3’ ‚~·ãS"óÆNíˆÎFSÛ$ãäû\d$‡HªŸDwwÄÈ”ŸõŽr!=4Å!Li,Å£Žôû‹uú Ž| xXêWhPý@¬oŠõUŠÏfíí¦\4à!fÒï3êÓ¨Î!Å×Åωý‹ùÇ.¢8TòP©~Jñïrú}µ_+7ª›öÏú‹Ò ½;÷×…pG£æ}Ž,–T¥Ký)o¥º˜·®Uç¹Ôšúêõ ÃÍ¡GÏKŒÇ7ð]­~5u:¯ù­9ñ‹%Õ›E¾yÁ6]ö(êê=l¥ºõ6þg Ç»|ÆéðþŽƒýF¦ÂÙPXÐp5J·?÷¦6v†~ü§·TÙKy™'}ßKXFfð&ÞÜÿ&õýáÆ8¬â53~µúùýîKv…“p„âéªcò¬g2Á-…‰U™§?¶7C(ÕsŽ 1x\¤{áÝ…T9Az9^HD½¼£ÚÎC¹ˆáYj‘â¨owšoãYGá —•!© BY{ï‡Ýân IEB€Šêï¥Q>—Îe!˲ÌÁÝÑIgd9äŒKö6Av7~AÔCN†ŽAÊýï¸Ä-Í^uq…oómÛŽ«Ïÿú:õ§®Sž»¤nÇ}ÉÈ[2µöézÈç³Éa»Q „½r(¤uYüz™Éž sªö3Ú%´nrßÒ=º·¨N^6U 0ܦ¼­Ü¢-·âÓPáf¥ðxÊAÜáË8‘PIuçJ!MŽ»f3§+½RÀ6ŽßU*b3'¾[VuÅ:>I=ƒƒu^7:íü`ô†Úæ»3#Ûn‘üOìŠ~5•tnõ)”i]N :èxòñ8ê™ óM&ýœFëF¬«ŠõT±)úý<:ÿ|Šk%Ý÷ Pœ)îꮤ×â¾1~–ꯢ¿“~­ ºŒèï«i1._K矛W‹û¤¾ù›1Õ=¥~Õ'Äþ%ÕÓ¬)¾²¡ºçfª_‰u"±ß!晎É®Ëø 'ž4±ÃÊírÇ]¸ÐyÜ9\…]»øÄÊî”'{Lç "ªð¤ûÙljšQúVxoúŸ ê³5i}Šº[Mï:*Þ½~„‘BâŽíêÚGùåïôEê÷Ê\¾øEkÞÞ覈ÚfݳI($!ŽžãiŠÿϼ "H$ÝOR¤¤ …¤ 8¦8æÚ[.¬îsd|³÷¨:ÐYžBÙÝ4ÃÅn|¢¯ŒËTG¸JyÛu.ªà3˜\.Šltþ0òt~켕øùG„ƒ‚›Ówç7AQ[¡†b¡\ªŽ/ž 3PZ´¿cÜ¢ù§'K˜ú~¢Ž.§}3+¨¿ðkãûˆ‡ÖPÝS¬‹y²>åÃÔÏÙHúdB:,æÕænñ=”Oƒ%·šƒoÀúìäÙs‡ Åf¾ú¿* ¶tÝ[½…ÀÛ4¸ˆ¯iÌvòsg^>Jbáz­gGÕØ%”!;ÂÖ§À:jðL#‹t;›OË¥áâHA8q‰âÙ+»êžX½a ®ñi`V6n´ã7`ÝB®]íݾy#oÓ??ˆ|×PûOP@ùl!_n.ÉAÕY‹¯ØM¶PöA ÿó·zƒ› »r ó·ØN¿’<´Æ…6„ Ê©^ZAç½#´G5ÿ]iÝ‹õzq?Šo7Þ1®JuN1^õhÕ]¦RÜ*ê™2醨wã(þc)~±¨—¿êâTŠ ¥x–üVœWE(÷-©ÙŸCçuSìcªÒuü«~Šº)î'õ’ôd)õã4ˆgÄþ©èÇ’NR°†úzb¾+ö¥x6T(ìÁ°“ ô5ýN \x3Š͉‡,i_“•PÖ´•ú‹[¨¾i'lÚ òûm´ŸÇ‰òéíTï÷-J}Eª/ï¡þ‹ñŽ'ßu7Ò„p»_ʰŸg6ê8@u?¡ ½—r^/ìã•Aüé¥ dÍš+arwp”Ü7Svýf{—!µ¥¼±z»×RDñáþý|ĸñ ò׈›cv¡Óæ~ˆ2äÇé«ýpVîέǸҿ š )q·ýÀ¾M*¤×µ‘&l0@:ß^¨s™ÚIõ¾>ÏÂÊ;³ÏòÇqÑõ3Ï8¸Ì³Ò¬¸j!$€¸î8鬃¼rµÜmrY!ϵ·òO}ÅŠŠú?ÝTøã=ŠW–×$ %ÄË¥eü„/p«­Ð`¨ñCÊãËí:p¿¸5*ö N«×Á¿é¡;íƒs¡¼ÁšúxÇù¶@cO˜Ó¾GŠãDýõIÜG#êä¯yãâë9¤Gs©°€ò·EÔ7ãÃeäG«èò ®®¯¯ÇyL!r ŸÏP@ÞmᢀöÕò»ÛfEíï)¡}­¥Ô¿ÅgÙ¶&5~Hý¸rÚ_WÁ±üŠÜ$Ü¡ý ÿ—R}OÜo,Ö%C_<å¿)ísã³åäWâ>4qß‹WJy#ù­˜7N!šBº3ú†³hÿš¸ïMÌCEýuGôKmò§5¾ÑIu$i¿7é¥å{úù<‘)À€òƒ Ô/‘ösÓ¾kcªã˜™~Ô0'?² ]¶­2äIrXÓþ7Ò·-¶«RÿqëÎFµ+dù5~Iûœ(ÚAu{ÒWzn;ùloKì¦~åaÛB]xðá´û^xòmÜ]jꪂša¿ÑIçzú=àó—žuÿeð5]½ß°™1¸Ebœ2`"Ñú=LuH>º­Ó Tß <2k¡cÄ>­•wLÝ•ŽàÁÂQ ËP!—„ Ûå·®w’ôNì†ñÿWY‚pÚŸÁo¯[ýÇ©Ÿu’öÏÚsiyU]N;ñ‘ìkâÖ^'xEDÌ‹¡Öù5@,¿½ _â¨/w:EhÔ!žú…gør‘ÎOœ¥}e‰´¯ )­3á<õ}Òiÿ[æ=A‘Mºvñí€ÜA•¸"<¾¸Îgó|[È'¨ÁÈçVʼn€%( ïñ»ä†œGñK÷–:_F)ñßͧgÏwߣ]ã‡6ü†‰|”Ó¼þÕÄzçè^SîÿÕ]bët}övì)¾ó #<›[†@+,:ÎÓk¤[bœ'Æ…R†â(ÑߤzÍ/qªxœD~'æ‹b½UŒÅ8QŠÅ: }NÌó´ÈÿWR^<Š×-æ‹â瞥”'’n‹Ÿ—öˆûXE¿ë¨tŸâþT½ºUS'ûÄzª'‰nÈßî»ôLh­YH½Åý;dH}+&±)Gø¡¥}Þö]„HÇç:¸.8Œí|Õa¤œ©~åF}‚]ªÂ†«š>¥X7¥ý1^Î#4Ýî´‚÷ð2~ÇÖÿÃÞ›‡åÔwïÃ¥dÊP!J*™‡$4Ï*Sf%TTŠR"%DD%’"CÑ 4¢45Ï“!!óœéwí}¯ÏvÿòÜßçy÷8Þçýã{÷>.R×°Ûk¯µÎužç‚KÈ·ôÆübÁÒä­}¯@À¦‚=ÉróŽàã&_ßHºBòUÃÆ—(ñJ.?Ïe€é.Ûƒº *!L<ƒäO^:‰ºpAOúêC2›?o$d³ƒ|HDÞHÖ‘ÉÈ‹JMÄûIHG^ûMYçMý’t ‹¹ ´B.[V@êJòÓb'47B~®Bì‹™¶Ö¾J1o—ýE¤… äuV³cá£P3½CA-òmëæ›2õÈûi`e’Ð(¼œa¦AïäOï Íblbœ'µNcgж`óUÞ[†»Ú вîY0 ¸ïôvú¤Hex€ï£]?©ÚáxêßúÿêÀcˆ×Ù!î@ù/>XÿQ¾õìcè¼Æö— ð: ýŸ^/s0ïÍÆïŸù‘þ] qÌ½ã”Æ5íi?Jó)×7RÞ*Î1i^¦ùöŸðV-ì«h]Mã__‡Þè¼S/k;¯3/çx Oý#_ã}‹Î=8y –ˆ3ý£býh‹|ŽG4¹¬füŠ­p)›¿ü­§Â>“Î?ð:sV]$WðæwsrG7¸²òƒéàÆ 9G#Áƒ¥ÿ8ƒü¯s“C˜Éôïy&ò½¼ƒi/\bÓôjðÍd=[à2òÚ*oWÛÝ–@(“nd@8;Θ‘ˆgD3c´)o!–•ƒôƒkøº ¼àæµÞèVi3ï›#$#Ÿ2ÍüCÛƒ³šü[L7#d ÙQ¯g \;í'øá<ã2Î37gáe¥uÑ‹!çÚáÌsÖ9ˆ¹“ë"j Qlù Ño˜:bQowgFªd¯Odd˜{Aεqžž„ýn Ûþ@ÿ0† ô{N8föY†\œÓça»oTÿ—+¡€•Y…"&;4‡bfJ7ë+”2pŸû(Gýf%CHþÕly°j˜*†_ j×Þa$¨cÚ‘ÿµ8¤ó§}6¦LGÈñbX6Û9؈×?Š_Ò˜%€A^— Ì8Œ÷«¼1‚m¬áòyz©ï=Hbªø‘o õz©¨kOg¢q«*d"ns‹¥W+AÃÖYÐ9ˆËæá} ã¸ëîbä»–þ`‰¦PŽz•*n6„–N-µ¨×ùoÅ¡;ÖwG‘Ïi8_˜#4®ãôé›p>GñÚ_Q¾ Åkhž¢u)þ1?¤yŸ‡Ö³4~iŸGçk0?öæ©R$׿Ñ9¾W_bž£÷Êcãø0øÛñŠçŸ«CQoïÎÒ×Ái|O3Ÿ›ß_@œÓ›yÖYjp±*–QÈÓŽ½«?–äéàó&—Oè'ã W9Ö9¾7X1ºè€`ÄBð}†!_3¼†ZOC$Îe¢JýŸ7¢›»/æ¼Y1ˆÛǪ37€«pùÏqJ"¼;ëJ¸Ž|ëë‹ÍU¤ Q”½Ð ‰ïÔ@ #ó¼ ©g•=‘‚4¦Ì•ò‚Liè¶…¬Í,±rýcçÕœ™ù¬,D ýYB1”­z––# elø­…ÊØG>ªVxyñ/Ü´–AËw\û¯Å!½/Sýí».ý`npϸú‹ÆÅ5i^£u!­{óÚh¼Ñº“â7´O¤ý!7Š“èbýHñM#Ì[”7Êé©ÿÎè#õ±1Á81Æë‡â ÿ”ê1Ïšà\Ÿþ<ç§AõÀXPÿ›x>hž¥sL›]lƒÄÍ+¹ºqÖøùì‘~óÊQÌûNø¾ŽáüÀ¥Y¿¤Tqœ@=ÆIÔ£ºaopïgq®~õRŒÞþ¸Ÿ Þz,.!àóž)4{¸¹F ›Ã<ÏÁˆ“…°°æcéäúŽsÐÈ%lƒø;O"å*þ¾®msc&Ž>37&·—¶Ô-…D쓱ïI‰‰¼dž i* qÄ Ò‘'”é.Å †p çXÙlÙ3 rqΑ‡sò|¬o pŽP4–½QB1s¶WC)[ÞIÂf¬3¶ʱo©ÅNPù¸š©Š7ÅCMçÊeu×- –¥ì‡:vœ# uÏT'¨g~½ 㡡ߊc|†B^· ¨Kkdèl׎A#Ö_MrŸWg}…&¬Ûš÷ÔŒõMóæ²Á‘m_ 8ZwžGÞ)Í;4ïÑú’ÆÍwï¤øçü¼´œ…yL¿³0niIó%Í£4¾iž¥¸íïh]ËᣘéûäpR¬k)Ÿ€êþ鼓òh]Lý¨ßå›sù?'ç—ƒÏÃåYœ³nCþ‡—bËñNQ§µ ñVês³ÏÇ;E¿N·~4ÏR2çkƒÏãÔ£÷YT"Ž¡¿7·T4qlÕ×2FÙnxÞÜ/2ò 8Ý™<~˦Î/å<£F×™°¿õƹ ç³Q¹yá’O\^ <,Ã(â!xŽUa£Š+„ ^0ìUè€Ø°kÉë.%‚²!uQWÑêê@°ÎHœ&.økÑBHFŒTÔ ¦£^#“¹Œ¾=…[¬ÜO rЇ‰òMoã<©€¹-<«€"¼ÿ£þ¤㪌±Û k‚ ì+«Ä› ©6B5 «C “ä@ úaÕžd¥PÇÐ׫ Ž•Wóóò(Û`A=Ö%õ˜§ôïñ"ï4°òË9ЈõY#ê¼Ñçú»P~1Å%½ÙËe ‡ô:¦×?Íc½ó å§Ñù7Ç ñ‰ï~â.”—Fyô~@qLŠƒrøe/¨1[F àòÕWa^¦|??ÿþ‡ŸÎoè<‚Ë›˜_iÞ¥õ,7G¤: |>Ž·FñRÌßÜëÊç>|ŽüÃÑÅ=}‡Ïÿü;/âïé8êî]U”ˆ¸apÿ)Å;3à êißOuû^ÇíÞE„[ÂEV6¹|¤Y üF3t\žË ç¸8 ÂóÇéÑW*Üàf~Æjoˆ@fc#–‘ 1 <ù: bßöQš?Á†Ó;ůcY¸ÁÊ×@"ö«Éè–‚ù=¥ïB:↙è;q«5&€ô[ÊeáÏe‡¾‚·‰€MEÉ~(àU¼+Š"(@‰ú†1¥¥R ãÊPWR1@¼1uÓe¨dØoój¡ó[ Ko“…ZœëÖ1•y õ¼fXwx&4LgWhøþôê~ShLf  É´éÍË®ÐÌ”# 9h$ã-lºr€vÌ9Z¥YÀZ=˜ÿh}É8vÉB7U9s89;Óþï,âæt^Açt½y¤4Ѻ“›büÒ¼GóàL¼®i~¤y“æSZÏrsÃ^óÂÞå1rsEŒÿÞ¾8\þ¤x)åÝÑæx§tžˆsÎÿ†Î‘‡´ß7ÿÇ×·Âø·A¾ö̯”_¸çùv'§õáµ.pq)Žo‡þ´Îqbmäáòã\0Oô·á}4pÅûçËQùj¨ZR'œA=Å9þµzžÊpž×¼ðB¼ß¿ˆ¾2>È{óc\j‡¼þƒpžGù9aìØÎ"0.¢‘׋yôÓ•ËhB⻉À›Ì~ŒáÊÒÚ–C:òÏ3Y»4 ÈÂ~+W%®@Î+n#ϵ€qù˜"EG62Ê1(1þ©é'YwT¾ËnŃ/r¿ùm ,5ly¼jñó×"¾SçÃsPϸ@N•„úï_ßVæÅ'úÿ4¢?NãMV( Mltš˜.gîShfªk‡¥Ð|è ÁÐ\Ç6p–M«öN³¯Ûä‹Óû?ÕQ^4í9DʯÁ럛ócGó$}¤u'̓ô:§sršïèõKçsÔ—ÉŠÉ’²º`ƒyã‹b¶yÊ{‡Ù‹¯Ç=Z2€Á$؇81å•R:×;ˆz{êoHuÄÔƒž7'ömiýÖb]Iç~ÔoÄ?ÇIäŸÂú’« ×äüI%™=çYXÝ.°r†‘à…xßEŒ§KèŸáƒü(¿±ìÔ)0öIpû¢à>ìBÙ6„Ž`ÃóŽ}Kâ‘ìa+D1ªAS!ˆÁzí*Ög×5‚BˆG¿˜œ'r<µ›¬¡77LfƆ›/AбÆ)å•ÅŠs½ÔÇó¢.gÞ…4ö×.iŒ­GG4¤£7ÆŒB&k7:n!>”ƒ~…¹È‹È{üÙì‡{ä#®\€úÆÂ¥#WÍt"Ô”`*E½xòJË:™F”£¿hÅVx—ÙF*ѯ§’¡ô ‚JäW¡OMUÐ†Ë ú•@úØUfªUõ†m]]Õè¿P󓈻Ä8¢ó€Eòäx^T¯Gó7ŸÀ|öïx5´¤4Òú•æYÇ4>é\ŸóYÄù>õ9åtQ½tÔ˜òÆi]ÊÕ›XwqüœËoÃçåæ!x¿ÿC'AykÔ·u*{ñüqy‹òQqînçå0æ?ê‹ãˆzmgœ7Ò:Ôï'Çs\Œý|Ž«GO"ŸìÖƒ/•â¨è—qqÎó+YB6ÇsóbÂyD0\œGrgÎô…KØ/ú"Áçu¿ däð¯!ù˜Áì¯BÑß3ŒM/Ë Âžd JŽ!‰s}` ú¥Å2pæ„Wpý`®1i$> ®£_Z‚<#Ü?7ðzºþV‰LvÓ2„$V¦«ɬ\ù0¤â}0}+@„Ìu;ô·¼Œ„[q¬c*ä°2ycÈÃûl>ÖC·Ghä‡BôÝ)Zò‘願‘SšÞæõ‘MÌJN?ˆO{¥øó¥¾¬1Üaiø‚p‡µ«wîns3»eX”±0Õ(Ã멜±{N¿ åëÌj(Çç-g—\ýÇ|j·w§qáx•G¥q@ë6ZÏõžöÖCQü…Ãaðz¥8 ÅI)Bû2ŠwlÇ:Ìõz´ï¢þÚT÷¾óõsÛ‡þØûwã±®Ýújšÿ¨.ò>9*7ˆ‡P¿'ôµpF=ë1Ä]pÞH}m\W;©kÞÄñyO±61æ\xñOì ¹ÈgýÂp_·}Œ Ñ— ç-%èKVÚE§˜ÿZ>¤|CŠP_Þÿ͇ÿ›ÿ7þ—i¾ þ¸4F_cŠÓÐùíë(NCç€4îzã5™†ßG)ßö‹?¡s¼ž©.‘ó×Çù‡›b^¦xfoÜ”›RŸÄO(ö)Ÿ”Îñy©å©ý¡»ÂûÇWCœ™ú€ÛàûáæŠÔÿþžN§|,GvÌ¡Î:ö=>gs~UÇUàj~pEúd¹£ñ–V:ü·ƒ}Y=¸Àجõ4ýö¿Àùœ¯§ÒÙþKeÀã%ý¯ƒ°n fÔE ”Q?iëCúpD0v]¢~e¾š—‘¦C  &“áêz– þÛß”•ãjýÆOч‘A[€&…×L¯µ3†4&ò›AƒÆZ{ÁMv<7²ÐŸ-ýŠóç—ÏЈšÞÂmE’K‡BVbEÈ/‘aªPÊØÎ}¼ eèoQÁ¸aè'CÛnnê[ Ðd5¬Li/Ô¢@3†”ð€º<ÖØêMY‚Ô3í·˜74 ž¹ñ„FÄ…Ñg«‰—Lm[E¡ ë¡f!(Zͬ}ßZ؃¾åt.NyÕT_Cçm”'CãŽæ5ŠwR\”â¡Ó1ÏMƸž„×}œŠùŽÆiï9"sšg{ëÿ¯ô<µ7ŽúïðSŒª[æ|Æ)~JõûXçÒûÇCG7_¤:*Ĺ(?་>]tÕapsE:÷Àùõ­;€¼'êOwyœƒúçy%EzH€ ûë_ÅÍOâãú2p|SÄû¸y"¾ ÌéïoÔƒÐúØù$þ8 dm¿!¨’!Ì<…ôû ºÆðVÆi DI1ÄSÀ~âX;ã) ¼—qö„è“ÄŒÝï¶A £²U+…tÜ“”‰üCÖÝVð9ä°rUÈcé®jÏ´+cà€AowäCòõJ÷ìmsák_<Êп´ë©*äÍU£¿E ÞÇjû3XP‹ºÿ:äÃ×aÝ[û@7Ôp6.…äÙ6â¾›FäÕÒzŠÎÁèüö/”¿F¯cŠÒù=̓”ßMç„4ßѸ¤ñHé|ŸÎ)è|þŸòáz#äPÞ)—çð‘ÖÓ´¥óNAýj¨㎛'Ò¿S=}=|¿4¿r<îÞúDÌoœÆ÷.¬­1ïr~ûx]R^܃ºSÞþÖ7Q}šq_Œ úŸÑ9ÄIv,=NÉ œ(:ò<zðø+¿ý†‘7va`Ï̾QÀ«Œèê„uÃEœ3û`ì×ð8Ìe’ùož ÖÁ¨s A<2ì0ÓuCê8¢0žcÄY¡ôï9=kw´ âqN~ƒ5A"{€$ÖVj,¤ O& ÏOÆš^æÚ7™2@1²Ørú0äâþ£<ô/ÏÇûbú2´· éPÄÜ&=„¶ Ûwpž_ÆlÛ8ø*™n|’T£@ £âXV5XÿÕ¢Ÿ~kËqêÑ÷¾ýGÛˆýh#K{ŒY/ÅAæÏæ™.¼35š½™ÔšqÎÚ‚yÛŠ¡ù_ÿÆÍ hÿå‚{(?…Îéi¼õÖQм&×)­?iþ›ˆñ9ãÆ#+Ò:öŸòá?é-h~ü#OþÓ#Õ1Ò<úOºŒ^û7¸üˆùúzÐ9Ç¿Áçùc_ÕaP"+b}ùÇ\çD\?Šçƒ›Ó nÃí}£yãÈïç_ŽÎWر—cl1¬ÚÀå3k 'ZàÅï·žª¤Tq°ßRð@ÿBOÔ'ŸÛÀ&Ðß>È'GؘÛ‘ïãÃŒ½3ÊÀy‡`Ãp\ùÅ ð!ø+ $A(öiáÈëüK.ØûýÚSÈ äd!ç7زÙ’²™iO åîXæ éìº ÈD¿˜,^–æ…ä2ãÏY‡þ‰ùßX+(`×8­€"ä—–0[à”ÀfÛœ•!”³éw-Tna $K šµ/0€jäIÔÔ3ÄzK¨EzAO\KzÔEÔ£ž¹Ùž5¾î`~3ÐÈʪwCc#0K‚&–>gM7âç ¯,çrÖМ«Oé¼î"Kk8ÌåGNˆ¸Ç?ù%þÓôÊ´Žíͧ¡x+ÕåÒëúu˜—{×w½÷6Y`ýcß!ú$ѽ0VXîFüÍšòýf´–Üc÷Û/ãÎýE÷!O’òÎìp¿·ÿöyT„z'äG9óš.‡ÛÁà‚Ÿ÷òO¹}1ŒË’ð ðÀ9Ý|Ÿg7µXZòO„sˆ_]@ªck}òx㾇K‹X£MðÁº•ãwã~Ë€¿ Ÿà â~Á̶°1‚¸1H¬‚ЉN˵] l “½!ùÈÜ\‘ÙBµ|D³é{:Ä2åÙ"g¸Ê+þæÛëÀµí¬®£ßhú4Þ@=L"žŸ$VŽ¡ÉÈ_MEœ6q5Ü ™È“½…zŒlÔEç _in# HCêîóÙ±íZ(`NÏ(Ä=ˆE¸®ýŠK­#r;(ýQåè«U÷µJæoº· Šaí¬U†j¦}ÝÒÕ8ÿ¯féEý¡FHÞ…MvÔ Ÿ¼æ&Klƒf«ÏŠ2¨Eþ[-òskw†{•iÀQýw™ÒÆœþê\ϳ´€ùœ‰â—4ôæqsñFy4˜Çh]Jóä?ñi8üëF:ßàx5ø>8bš¯0žÿàÃôâr¾lxŸ¡x+õÙ¡üŽGŠyŸÎk¸ýo¨· ¸ ·/û5ºO‘úãpõ(æ…ÝX/Ó¸¦ûŸ8Ÿ*܃CõTtnÂ驨®ö{èq çÕǤ™OÎé»)üÔ©"¶¯q>mg|Á]pq.¶ŠVú¿9MàÅŽ7øÁ› ‹p}}DbÎ:ðE>¶?ÖW¨ a±$ç÷a¸ÿ.}ò#ÿˆÂüƒºˆ«èK~uc ìPH eÕ8¾i êÁRѯ8µ35‡ Ä…o.î6JYˆ·ä ^2Oy›yìú&ÈgËÀiP Ä¬AaqÅ)ù@(6cˆçPÂÚ›p¾Rwï)»Â ”£oHòy*‘·VÉÚ°‹Bk§ºª…ÙÄÕ¸g²ëÒê*ÖØj9§D0Ç ¦ó8ªKð9Á,<ã®Gʦ¸ Å9zë‚{ëzë‚i¼Òø¥õ$'ZïÑøçtØ·Q¼Akwnì r<-‰èËKý“yU)…1¤`ÞHc\¦4ý ƒ‘Þ~·Ð_#ë{Ç^ 9Ìm‚ï)äb¿š‡ù9_“1V{·Y9„0ãáPÈÐ} NBÑé,ÑÅ‘ã }WK™-8ƒB¡Lþ/bê+R¾>ÕÏÒºò¦h¾ z_šzó¹i¼õÆO{óJ¹|ˆÿÎñKñçÿȇ˜o8ŸoŒcÊ££<ê_Eýù{çAŠûP<øŸò;rþý˜_ÿØNû:ª¯GžÁþ3ø¾¨®“Öÿ¶øóûpî`‡znoÎŽ /±#;^ NÌ6<8†÷ :_<údÎgßÏØgr}ÝÅL¯”Á…¿>Ö ÞÈ?àtRx?äâùÑè—HuŠ!ŒÚN©BÛßSö‰5`„†…h‘x¿‰ú«Q„ÜCz•YÏ£‘×ðþÀЊ¢!ñý¤¿Hai¢3 uŽiè—ÎÒZ!ëâ›ìi€lÌG¹È;ÌC_»üâôs P€{< qÿñ+R\œÎ稾âîé¢yŠâ$tnÁù@ýƒ¾‰Ö¡t>AqúHã“>3Ð9·ç‚âxpñ‚8åWs¸'ÕMP¿nê[Cu½üÝ(îÉÍ©7Å?©N˜ÖˆgR¾+õSµ@ ìsi>¶Ä¼O}V¹}ÞÈ#àâë¤}x>¹ú”îïÆçsPIX‘*”G‘¿Áùê3iã~gíõU8þŽî‹¢{nΰ6žpý³9ÿÖhñw_‰ñìëásPwJøã«?v0Aè;ÌÀž…Ç ” ‹ü Wd4Dа@0D3k›¶¹A,ú|ƱðÊ>¸nÅS Êæ9òm)W ç))8ßOC}mî=¼iÃÞ@ ã)‡•óAê³òq~Q€ød!úncÜ—ú²… ”1iH= *pY…¸vu8+†š vPµ¬lq%Ô!ßµž×œ¨¿ Xß7"ߦ)œ!øŒ€f†lt Zgh3b7à.{»”ƒûÓYB:<ÀùR;+ÍùzÒø£s úû¥ù‡Æ Å5{ç9ŠR´7>*‡u&}¤8*3Ò8¦ñÝ[¯ÏñâðuÿÀQ{á§ÿˆR_ªkÄ×£óCÚs~5´ß¤y–ú€£>‹ö›œ8å÷à}ƒã ÞÊÍ©ž™Î)ç›èËFó*Å‘lPçÈùPÞ¯è^U{äÃÎdÑà(»æDñ·þ‚îÃÀ¾í›nà$ÎKNjÕbð@_”3ègÀíÆºØ ý,¼»¾ñÝÊÞÊù§ú¡î÷2^Wp}0³FEò5„àï- ÷ÛE0ë~?Å@K±ìËÏâöa$ð.þQSýàk7ÀË,M| ¤ þ/çYãYÈbáå…‹¾ôyÕlá ·™,9æ&²By(F>`)~ž2V‡U©l Õ8ÿ¨a¦±\¡–Û´fB=öG Œ,Cé,4^bAÎ[p~ÕÊŽeLá.ËV÷Y:û@hgí -8<‘óÁëÜýYèõJ〫?1¾zçÁšãÓx¤ñI¿â¦½ù¦œ_ ­')ž‚ï‡ÛÏùˆó_Ã÷GùmtîHãóe£úC:¤ûdáø4tÏÕëã¼õýÝ8ϧ¾çÛ‘Ïö‡oî1܉{º¨?Ž Æ7ç«{9ý>~N{¬è~â£ì¢èFpFÝÄ1ôU;Ž}³+îï£ó`ü9Ïv@çØ¶Çî7•î-¥{'ð³b„{›á2^ÿ¸‡‰Óí3[ïµ›~ï™aí¿WC”;cdóbÎ1càjºÚ»ª‹ À\Ô¼ è+Çí¿`à|Ã4HalO-!q•Ú»2Ö±€0ÜdÖm(Þ,œ3æ4²†¬‡sÆü*w¢M{¡`+»HŠ÷á-¯ ç ¥ŒmÀõ P†ýaòn«ªnauL†–&$ µ¸—µùõ, `>4>ôIÝq¿F³€Îì“#¡÷-¶"ß­­üÌ£8E9¸ÇØjþ‚ûØo¶3pÓcÐΔ로_=½ÏÓyÙaÔMÒøàæ†˜gzÏ ÿ]‡ñJé÷ÑŸ§ý"‡ã`^¢s:O¤}iïüHç&ÿî‘ò8?Ï?æIêë†ñËù Ò½ø¾8=oþ+õA¦óDì?9ŸrªS¤:~Ê@|’Î_,ÑíßæEô ?À{“Sݸ9"Ý“L÷;r{±~8޼cN_Œ{}éžðÓ¸oÔ3%ÌÀ9ôû½À®±7oÞ‹Í[ —ðþã‹ûáü‘LjþAì:¸þ¢Ë¸@hO#Ãlåö,F¡Ÿc ï세¤ÎýóH.ÄײF4pã2+”†$ôULÁ}f錌Ìñ!d&³ÄQÈÂ9~.ÖñyoXÜÆ½’…¸ß­÷‹–â|²ŒYÓØ ØT骪} j0ï×bX‡÷µzܓـz¯F̛͌[ˆU´ /H›úø¤ñžp—q’x÷Y¸œW—¢/óRŸzÚGǺ‹^O4>)>JãÖ«½}1zóihŸHi_Iãâ@ôõhþâê8š—¨žömÔ7uœ¯>æ!Êg1G~„â žŠzhêçËíÉ þm¼f~ëé¬ß{1(¾‚|Žýè›A}˜ÿÐmн¨¨§¾ÝÜþSŒcêoÈíéF~Íñ•Ì · N „Û£F犸o†Î!Ρ/îyÔß^À=ÜDœ×û`œú²—W?ðG?´€ÛA?ŒŸ­…+_Ûf3ã›p`ѽݾ¿÷>¡n ‚É:òo! ñÙèKÇÑß>¦ØO^Ç=ï X?&àõrã!iˆãG—5HÆ9zª c`/ i¸8ãò&꘩N?õÁ¹}Ø…¿'<°zÿEÈgí¼§ÂmÞÝŒ×B!úì_g ¤PÊl¥·†2ܳQΤéúgÜ^¨J¬ªÙµÀS šMsC ¦š1‘„Zœ××aŸYÄõ¡ý8Ó ‘…e 4µ+ ß=¹ZD˜ ·Z§j ó•ÓÐ÷˜v6/ànŠgкî êl©Îæ š—h×{nHûFZ—Ò¸£ù’þþûºüÂM{ùzsüš·èÞ§ÊWø¾h_Çù½QŸbÊŸC^‡›R^9æ!ÎwŸÏçdœ_õ•¢üZŸâóÍÀ>ŸãS¿Rªã¢¾Rtï õ£þÏtß“#/zÖtxýÖ!#^qœQIÍÚ®è§ï†ŸËsTw|Š.áÞ8¾ ÅopÍ%Ì—/€MKâpù¨,-[ ‚°Ïâø3¨¯Ã=há¸.uðÑÈóŒelu’Ó Ž]¯râY›ßE¿}ï3¦>$ãçJ “`*HcaçùÁË6¼Kn2öÜ÷®AÎsqÞ“ÇÒõVÃí½»Lܵl!EèW|ðçk£a(ŽÉwض_Êb6ÔA%âÊÕ˜gkðóÖª²D@¨cm­¡õö ìº=1hbÖ¼y¹C3“mÎ…–VV  m juĉã9™á^>šw¼Ï¥}#Å-¨_ÅG¸ùE/¾)OZoÒ¸£õ,cßTÏHãÖmܼ¼÷œ —O=×w!.H÷LPœ’æ;zÝS~'ÝwF¯[ÔQì½Ç týö+ÅÏs÷bqûA™6ÿpŽèÐù\ÙÂí÷=‚¸¨#âÑÜžoêwO÷Î`¼qº ªkŠre˜*p ëwô8ÏÏù%þeøü×Ýk´x±4;ðf¶À¬j‚KL>#|à³^6wñd¸Œ÷އ=„5høÿî±ÿð{¯ÎË"0/sóCô«eƆ[N@æëÌ65%H@¼4áqúa^ 7pOb"£Š?¼’ð¼%·°@H]3±ÞIz?¤ãžªLäSßÄ}:Y¼³Ã»•B6»¾m=ä²mûjÈEŸS†%öc0ä³°å8(`.ß¶2(BÅ’ó„Rä©”M½>eï¥(,É»3Žƒ ä½Vîc m¨b׎„jÔ•Õ`ßR‹{Ûê›q8¨G¿ÿì£q_VâHÍÌÔuB<´¢£¥á;Â]Ü'D¯z§¸Å9¬Ç¨_?ÚgÑþ­7nÓ[wAã®wÒ¼Iq:ç§qLûBúz\ס˜â›Üüòh¨^‘ú¯õÂ{6âç¡8翆yò8¨Þz'Äé¾n~ˆuÕ+ryµ»œÂù²Ñz×ñ¶½˜Gé¾C—¡ùñÖ›ÿ„‹=ƒ,½áú~Äú“ãÓÏÒÇ‚KÇ6oÜCtý |’éqðcPÉÁù-xý]1`3(/ž¼ú{Çà%£¸Wí ç1aȃÀω¸b4êòb×ÿßs ÜŸ€þ§‰8ŸJ¾%ç‘©,=m¤cˆ;‘ûoâëeãûÊe/è3þ0·.`Tš:P„õB1î«/Å=£wЫœ•y¨üö±AܸfÓÈ[@íö†uŠÊ 2õú Ñþ(4œÿëßȬ™ŠK€fÖÞ¹´`œµÍÆË|‡ùøûðñ ùÅÇÇ7Œwôåx‡ ½™‰-þYÀÄÜŒþq¯ÙAú¶&Ûè—ù÷üõ4rÌÓ æ}œø˜ÿÄvýõõqôéùñššì3‘ßnkbÅ<ÅÏ¿¾Ë€þ4ï;ú•óŽ ÞQÉ;ªxG5ï¨áµ¼£ŽwÔÿõsô`~~%ýy Þ¿{󎋼ãïðá¾¼ÃwøóŽË¼#€wòŽ+¼#ˆwóŽÞÊ;ÂxG8ïˆà‘¼#ŠwDóŽÞË;®òŽ8Þqw\çñ¼#wÜà‰¼#‰w$óŽÞ‘Ê;ÒxG:ïÈà™¼ã&ï¸Å;²xG6ïÈá¹¼#wäóŽÛ¼£€wòŽ"ÞQÌ;JxG)ï¸Ã;ÊþóóÜCz^èÙǯÛ0_Â;†Â–Å#žþ¼Lâ~å¾yâ&Wo¹{ÕαF9Y‚ÎåxYQÁRØ2iUŒUT?µ ÚÆjõ3 ÁpîÚ-kO‚áÃ…'Æ[‚aÚÈ@•O0˜Äw kîsØüýÅÉc|À¤Siò÷wö`üàˆæaûÿûwHß× æ}}ï ññÉ3ïóןgâŸþöµYû³âßþ<ûožó·?ÏýÛŸ•ðÏ3è×¾á#s^øá ý¾GîßÔíêÿþoÿîgþÃ÷¦ü_zo*{]Õÿø=¨óñýW_§××ÿ~àóÎûÛkÌÿõÿ½–Ôþö= þáºúϯõþÿôÙÿóßí¿xŒ òÇt–ª’ÂL…Y¼ÿÿ|Åõ á×_ϤþëÿŸŸò_ý5èÏÿý“‹ÈÈ𓌭Éns³½2Ûm­­þ|aî6ñ¯.ËÞOØgŸõ¿{u¾ÿñ„˜¬¹wà¿ú+“]fóþíó÷ÿŸ¿¿Ìv³ÿìy†ýÏÃÿ¯Þ¢˜ /)›ÉXo—1Û½ÏÖþΨÚÿë3ªöÿ䌲ïôo×ÿ•Ä[Þ›Þg²u×ß ‰m¼oà=»mœÍž}zdÀP08­(UYÞ³¶äóº¢A‘9™-c­Iw¿‰2ÛäþÁØQSŽÄ‘†Æú¯ÍWš‹ýI€©J[u-Ba#)wßlÝ5‘ÜQÜ#îCJ6ŸsÏ­Œ!Å6râoýHQÏn»WJ¤xîòÑÉŸ7’âØŸuâJ¸ï¯˜™è­èy™TïnT?×Lj'4zºOˆÔ¤~,M{ALõ«ÍHãƒy¾q¢¤ÙèÁ+­TÒúÞçÑŠ¤5N6/ûJi3xå÷ŪiKÙ8h‡ðlÒöféà©+G‘»Ó” †“»wHØ­$÷Š«-¢Lî׈NJÌ×%íã>|2·'í7Ï9NPXJú&_ü±TŸ<òÚ¢¾o›é¸êm°<<’<~¨§¦ëAž,Ú#û#=†<­yªn~%‘<;ʧ°¦|ya¶jþr‹ òòd«íÄÝäuçÍoÑlòvéµAÖwé×Ò7xG>èÞ¾ß1|[°æ¥.ù"]vu é0y¢v,ù¾z‚ßz ò3åÊÕ—ÝéÀ÷q áÓ„>£ˆX«=¨»Ö _ÖB_= ó­_ÆA?y§¦ž¾0`³êÖl¤ÑõóÚìhxPц¶Zï’‘™ÛmvQWÅ Nï@ì“Z´tÃgGaµ.>Ÿö98O\F}½ë\f{¢SAjÌéèój>ÚwXGìBzqþ²³r0Î…¨ ²¬ƒñ2DÒÖY &<f'" “žh¶h=SGÅ+®Î¸Ó-œž"ÚfNýP{f1ÌâS^:2‚ÝK:>s¿¨½ïs¿MqP=ø ”Ÿè­þ‘á ªŸ®m¸ézÔfºßz>ô/{Ï7»ŒAýœ¹¦eµh<ñü`Ü>X¤ÛÒn=g,n<­om+KͦXûŒ=Ëú¼P¾•6–_ÿµÜù1¬4?{¿üÞxX=ø‹Œük¾é\óï Z;Â,ÆZŸmýÓÙé:|°nÕ”ŒÒ-] £Þ¸ùP× X¯´aª­Š4l˜ÖÒ‘í ¥?‘¹‰ +²FïKeèþ¬šéÿªôžœX¼äI"è¿Ûô6)Ì.|/›wi묕ƒéU¾ãšÞƒÁƒ‘s®…'€aö£Ž#ÁmßÉÍ·ãa}ÎÛ–¹5`Ú§µ7ä%§€PŠÔ¢á«HW›âÊäc¤eŹù ÓH­SÅõ¾R!¤Âòüò”8)RºôWÀüÉ7HÑоVOç“Û'oôín7 y ï÷_<øšäºç= X7œäŽá¼ñò³ßøuú}·óO™E¼ñåž§äÒ®³ï'w:ú•úìŠ'åêûjÝHª4N›z™™ê˜ åÅÝæ¤&1Q_gÆKR{üRÛ†‘ºS5ñ2¶Iý¼c;%„ÛHý‹¹F½W' —»)ž•#–£õU÷­!Mr¶ìÓ Í–óÓïYrŸ¯uÏì­Â™¤Íçþ¤ F¼¸ýê\·³µ™Ü߬°Í¯Ïdòàóƒ‰SVö;kŒ¾L  Þk ’G9AGž.$O¼ªv-Ò Ó$ƒ² É“Ë=….’¤kêþ¡o¢È³cuWÝv!/ž«úÞ/%¯$/K‰oN&oÔŠøWx“·ÇÏ’¹5—¼»›°Î5‚‹×£»7ù‘ωãuzÖû’¯—·ò–M#ßÒ;êNíK~ H©-x|Ë|ǵ¼þkçd„•±2M›?‚`—É®ÉÒãA¨ð½ÃM‡kп†ßªxýG4ø°e½Åì2æ~ü1&®¼!ÅKD¼¦z¬œbª;ùB£ra„HýéÚ >qé’ˆèá0Úá[Z²Œ³aŽÑØ =ð~[ÔaOê”3Ø Æ}]Axó2ÿ]BX¢öLšU"·+ئ.9Æ/ÙÓIdß»†_aæ(1ïo-¶0kÄ냟ŒƒbÎÒ®7*20ÇÃ!ýYZ%(™@Ö ßa;ê¯Ì†y÷"쮜{ Öt…X5'ÿ§/ÉÜ7›=[š|›½aÜ6£‡ä»ÁT“‹g‚ɯ/ã2ò>ŸÏ®4|j|òÞç´gXß{¿íñ+$€ÿö£3rºß¡Ïu½Ÿ›Ýwƒ@üÚú«V@°¶ÏÁWẠ$¡žôòÂèç¥/~wŸ0 PT}µ.ò |íZѲØ„Ë6?öb Éôž#TfÃŒBŠ=´ïˆéèpÙ›ƒ@ÔÚ7v^ØM;bÔ(3[†ŸÙí9FðÝß™=FƈÝu2nñÄœ #ò-aTŸÈ KÉ0:½…ÈtˆÏδ´îɈ§M4aL°ŠÃÚ“@*ô”Æw…S06Õ~AÊŠå }¿¿ÁYŸAæL`Øíá×@VeŸªR Ⱦ6\xïðu—µPhW0ÈïÚ$vsŒß5L;ßcL°î|%ú#Lôç?º~ó+˜ônØ™‘6ê0åH²nH¤LSóëã¥3&çïYÙ 3Õ&Ê[…¹€B¦wÔðãaÖ“³¡µ+J`öXÅ1]0ÇðêþúÝ¥07hóà|Å© ô ÚHKP T$Ÿ*†ˆƒªžê£„e;a^€HÖ‹`˜ÿJðôëBX°áÍõ¢G[€´>üª]&êU[¶ˆµ‚Æ´a»'ú³×yn‹VúÏíã8»+„ô]ÍK^–æ´(8‚æ¡<»×†ï`¹Ò‰åµ{æÁʉÎëŽ ¾ «µæª ,€µÙe )GA{~Ãúîâ°.ßí„ÄÊlXoÿr‘ƒ@l´ø¢pYdèùž³l;m ›ÅÎ=ë÷s>=aöh{þzñ غÏÕ³=û9˜FÅí>x`!xÆlmÝÐï(wÖü€±³€xxÙ‰ó­…Qbb²±{ɧ·N|/´EHçvi=åä®KéfëǤYÈq˜×;R/5Ô¦ðç7R]<åeþU;R¾XUÝÆÇŠ”Ôh-âÏþ]‡v¶¤S×H2'Ût™Ó²¨2%Ïó›?xH¨’îÆó÷ü¬îWÁ¾çN“ƒäxMRtÒò¦TÙqøó/ä­áð¤m6qämÖ\#ÝFÞ ñôk—þLÞíNQZ¯}›¼ë¶³h ï/…^¿ðõù°6Ò×Á~¯Ÿf^ûîF>÷ËH0M)'_dW¿ÖOM _‡-ñß²“ôDÿK]F¾ HlݾݘüØÝT(ò†ü|4k ùÇ4àSÐwb üƒƒ[íWþ_–i¯¯ûƒ€h{ðŒ3‰ ¸Ö}n{íè{]8DZö.WOö/¼#1CÝîrü¢¬t„'Z—H„Á]æû”r'ÂЬ”„~Ùé0¬jïÒ¾“ãA¤¸spÆ$c-QQ[(bå×ǘ¬} ë®ìÙ0ÿ4Œ¨X|íT_™wsÔˆ¥Ù µéöcû*å’où$oŒ6Z²O²$æfeÞ|¶$‡7-²;’?n;ž=²Æ|زyð“ õ%½-+N‡Ëw2#·; ÖZ2 vÍN€¬î½*¾óaœ˜DÈî=M0îñOåÏÈÕ~Òú Æwäik*‰Ã&}Ij{“ÖŒÎÌI†ÉÑ]ßÍ[„`jÿö†¬–H˜¶g<±?Ó;‹œôžó¼ewêjP´WØ“ vÛ\ùn³ê4í‡©ÅæâüÍõ¹0û“AŽ“ã1˜+ïöÉ£^”\žOYØ÷;(÷¬8´%ÔT½½„ÍTlaþF“jÛ#‹`Á"Á¡bFîm.‡@½Ñì›Ø†hX8ú„s¥ÝXt(Ì=ef,þî¦7­r+, =k2uÓwXf}rKZÊ3XaîÒc5Ï VyN8m¢kºù[ú¯­¬ôö®v°nâK™³ƒ@'iÆ\;Ÿñ°a»‡ÜÚÎ÷ »t÷»ÛI‡@_OK?²rl{2TãÞ 0šuÞÇ!Ò<…„n›ƒø­œ‘Aá¸ìˆÌœz¨Üøkl–ynÃï4=šÜëë|+71žÔ?púPâ(©n«›8Á5ŸTšWHÉ/&eF)ÞjVå¤T.kÒÇ‚S¤ØÈÍ$Ú` )²ç3tè»–Wo[öúS )Z[nš˜óŠ5MÒßÑ=‡”®«z£LÊ7¼8²/“T}‘+@jÒý3Šä¼IÝšw´ÄJI}§îW"u›4êí[³xi* HŒÊœFZ„ýç®™BZù5CO¯KZÃô–  "­¯?݈ˆK!mc¦/ÕéæÅù²­ŠëÞ“¶çÇôs{BîÞÙ{àKR*¹—ß?ËèÙcrÿó©c{TzHûÄœ&Òþãë¯Ãy}âE¿´Î2ÒÑ·Ïxó‡ŸÈcÕE•v“î’΋Vf™Æä©Ìc}7+'ÒUfç#þC„FlÙ+»ãíïw>\Õ7FÕÌŠ 1˜ E’ʵ0æÙĨ·IŸAzI›ðUysüDÛ±4·¥ËDtò@ûôV»Kq1LÈܘÚV=&ÅçܲŒ”‚)­¹ãŒF~‚éóUL2—ùúÑgtG·‚BÍŸ-{g‚âÆ+qâ)»aöç†WÎÖz07m¦¥{»8(û֊¢ ‚óù WÃâPÙÏ‹µ¬`©ü<™WA³$5*õùJXnëûnwèVX9¹Ó­Þy ¬zøácìü¥s£”?)Pw H3%e~þV‚CÆŠ4å.+9ARuÃ#.EIˆTWÈ¾ŽšHjâ:S‡¨û‘Z×2Ïõ™)¤N><8aäR×è¤Ô0P˜Ô;¯õ¶*M–_œY}\˜4N±îø©ÕIšF ÷…jÒ,juo~Ò-Ò¢$îÖaoEZ76 ‘Ê×#m‰g‡†MØ‚VÏ~ÄŸð¡E.c¬¤ ÿ›w#×/†£†ç‚pÇŠGk ¾ÃPÇŒ‚GIš 2+Y´BLDË—%yÆ/‚á~ë¬$‚‘á¯\t F½·=ìHHžR]vÎø5ŒµŠx÷òlÈÔ½Ö » ã® ƒ Ÿ÷Ãx_ýó'ÂÄÛIï4'eÁ¹Wò³R|aZF˜ý™ú?ùî…™±$(”þùCú9(î,»XP æÌzÐw”ü-´n6”Ëx ëߪ']®^‚£†~{n> úê׌¿9„Êb³mÚ ÿ6 Áç0pôü¾[3aÐ+±»÷‹õapS¿ðŸìahý³.ã;7`Ø«°95óAäkÑ {§# Öß©nøÞ'0\rnÛT^“¤|mÙb¹èäÆ ¾1¾m-/_ŒÚÞXÿxÝmí{ÈtÄ+°éÞÖ#ß’6:×ö¿Ê‡1{ÝÆOï RGº|fÂX¿É£–Gïé¨Ìä#AæXT’¤ó+á9|¯N’}œ£¹Þ· ÆE T]rö1n'ÏùÃx#Ok§ Ù0Á(ìáç=·`â±æv{ik˜T'ïôf¾@0I%¦ eÒ ®6:æ5]0+Zâ“À¹ XŸn¤•ís´ ›çs]×/Y‹G'!‹ðÁ@9SFV£§ T>~Ñö æÍ—Nyqæ_5¥<2ˆ UÙ³HòÉ“¶Ù/@}á¶mEsŸ€ÆÂ#ío®€…†|Ò’±¡°È+­´Äu=,~ay2"–îö¿¾e,‘¯Ï9 Ë;O¯L©I…•ϧ††œšk&.ÔOyð ´´ÞV<õKí÷ sÕ:à´ Ö¿˜S:$çl¬h‘/®M½gŽß×üØ›—•ÏÛgô ¿¼_( Æ'›æ¨ÃÖ•‹·–xŸÓÍæŸÒR.™®3ew‚±©Â—w}@ímϧþ)0ò×tc•IÍäãí»ã®~«&^ý®}ž´]¦‘$¢ÉË#ýG®·Ÿ@êÖ·Î;»†TÓýÒÓEÊfjKø“âz¢­žOŠTË›¿,#E…®Ï¶^7!¥«zÎÍu •Vy¿M!u·úšÜ° '­#¥o[ÚÞ í—7˜¹&š<¹¶¥vS>yòþ€ÑRÕTòLÚ·õØ‘SÖ³»˜Hºï{…DÖ!¯Œò¿d¹ ¯Ódª_›@ÞìL»ªvŒ¼)?i¸$c9y+­£,õy"yëôbíÑè\òöGç†iý+È»pöҹ佞Tùú€òAÖÒçÛ²ä#ß·•‹¶Ô‘?œúîŸð‚|©|9"¸‚|Y^¡¢<`ùêqVë„Ò5Òóèkä² ò}ŃîuËÈLc³5'²È/uýª;€oÿ˜Ý¥À¿Í®}o÷fè³wô{+‚@lÅla5^½!ðáhzüT:⾤êºô¬ÐPðÜ \½ïÞ¶þ 26\²³ 0[œói( yxàëó;N0ÌÐÖ:±J DÖ½šp¶ü<ˆj7î<¬bm_ä[ÖÃðMÏ’Lÿ #6÷ÙjÖ±F®3—6ñùA¾­³ÇÀ¨1Ö¡Úû`TÏò—ô 0º¹çĺw± ‘³7pŽ+H^ý¡ªïc"ï»dô©ëÝÙÛß]€±æ³6;oég‡úxw¼™³ŽÂÏÍEAV)æfЯ™ ûÆaá·¾oa\ît#ýñ“@îÚà/¢)ga|†¤}Žðj˜ð0.eêøh˜$;IhÍ( ˜¼gÙû{w˜Rîj³ý L“3GwŒL(Q¼g\ ò’K¾~3“²+GŸÎ…Ôe›ù¬ÐÃv«F;bõÎ÷ŽÃìÔë®Î¨…9¯lšÓË×ug²_…éòS•mz5+A5N˸jLJöNÿ´:¶”9Œ÷¶©…T‰ž˜ë!±‘Ô,Ë^'Kj[f¬o´Ž$õ;®ú^mxFª¿WeÜEšTôí~fÕ’fËŒ·u„´l´rÑÿèMZžå}‘uW!­S}t­V&­ËDœÙAZÍ<û5ûÈÖ/O¥l/;“¶Šo¢œ•ÈÝįÕq¾{Ƚ– ‰a}ȹûæ‚ßIû6°ø;‡<ÜÐ÷Ý7‹ä‘‘u·«´&é8?ÓØ»N”<~ºC´âÖHòÄú¡mŒÚZÒ5ªûÇCy5òœO"[ùîÒ=ÎõK‚ªyudÍ­¿:É›ó2«nÉL!ïø7Žþuü,yyÚƒ|9;òqC•Ö7½»ä3‰z,J¾®z¬¡1|su?çݰ€üxª±áØÆ_À·0¥kAk3ðÇ=1<_ T^~¼ðã>{3kß3ÊYeØaýÓ"Wå?€5á[‚`ðàáGà µ÷zYÖ<öþò°@ñVÕ\½@S},ˆ%ŸñØÓgŒX`®té[Œì˜•(1FE§Åævóò¨ðTÉ¢—0&%7J”o,H¾5We&ÈÔì­Þm· Æ©]pÕ¹ä*ÂÝ ;”a‚{ ð ³J0É^9èøöé0% Ft÷>€iftúÝZò.£U ^BØí®h #P·o¯Üëí0ûV‚™Ÿ"ÌÝ;Wï®æLP^uEC®šZãn#óí*’ÄÇ ‡µ–+N­ê3Âß™94NŽÉ8²à,|'ÑËU[¥]-ŸK>¯HrT¿šžÇÄoHåJÎ^ùaÅóùŠw'„U׬m]pÖ8g›¿8¶´/¼Ú& ÚƒÖØÊΆu2¹62ƒÎXM;~°~lÖæô¢Á°aœØ×  >°qbðóûÇîÔ†o^ý“@oƬA'Oƒþt ““{`Ó¤%š³– ƒÍR õÝxõ©ðËôO§ƒÁÇ¥ãj?^ƈUÇ'^£{ûÔŸ‡]×ïwTª@çã¼\i©0Er±”k…/nÝ/üv«y2´ïf׌9¤±òijjª¼~L~£Àš›äN²Ä©oOb!­l…ý!¤@¦ð[†¾É+ñ›wØ•äN¿j§YL²ÇO¸PÞJ²,$È-üN²´RS¥Jr_Ï•yÚjxUä ˜éƒJ÷7„ì|©lCjïo‚nV½à¼‹ÞhÝ2qû¡Ôb.S½g³w#Œ”ós´x £v|™¸ËøH|’2ÈÛ RÕVg‡¥ƒŒ£šü¡yãaÜÂ²î ¯P?Ç£§*¬&nÊ“„ɱ—Z·®Õƒis‚_ÂJm˜ñ¹V£òI!(îBà(ŽZ÷Õ§Úf7kh/ ‚¹·öÎ1=¸ ”K-#+aÿ¬A¾B @më¶Åµç+¼Zm£•êÕ=Ö£/~†…êAî÷bÀ¢‚ܧ[FÁè/in 9pcôÛC°¬¼Á›l¨êä4nm÷ÕîŒ<=wÕ{¤?ÍØ¢&HVqõ£VtÈGM„di–{ßZüpÃOÚquœ§3êØwQ÷VÝßþC¨×ÛîžœŽGÍâ?SìE¢ÁøÅI÷ÐèÝ3mJ4I½gþXA;´.—’$¢ÅÕŠìw´*»ZïãômÄ&Thœ›Ñ6ë–€ÏwN´w»b‘þ ¥VSµÐYã\RY€ºFJú•î”n=ÄNlèyɬ1c5½‰T²ý…чN(b}e“Ä Š×¿]CJn¼“ø5¼.f#Ëߣ/*?jÁê5ý3)0úPmÍôèjä£üö :„îp³¶A˭첟|iаCµa“š 5\&òÏÿÀJ}r†ÿA¥&êqŽiCå[ÅiF¨Zªv¼àp ê2C´t’ü¾mÌx„ºB§ÃOwËÀ ïo¥'0R»›>¶Aã‘gν!‡ÉçuZ=¹0}…FÌöçcøËü­ôô%˜MIEs½-ŽÑ.0¿úbr‚>óÀ¿<ú‹rlm® °xOÀïýþOX¢³Ó ñ{KßH{½Ëž÷ ¤~SÊÐT·EÄ Xùwµ·«œVW \bPÂ:‘åLs l(&H´æHÃfàÏtß|&Ø*ëš‘¤‚Ñ)¦”ù@ؽ{û·2K9ì11ôþB û?d—‹+‘ êù¦7¶Ú”'Q˜ 1½œêÛF$qîs³Ñº„¤u÷¦«¼¥ñÝÃ;n©¯üïžkÁ'ònó›‰â1¤Öõf1–ÜDZR ½ú#¤ëUøý÷2Ò÷oñ *C†^ÇÀÔFdR-Š)D¦ ¥õS¼AÈ(ë|üwTðecñž0DÅãí}¨”uó翪ÈÛ=ÕÎEÕ6Q3QT¯ )ç%?‡šs?ôÏ© ˜ß` ÁlÔ>·ìs3Èu:•ÖŸù7£ž—ÈÓºîïx”ŽÆ%ýР/\Ôü´ý)»µç†&ݶruÎhÎr¾¯ïK&Z†; ](Ek’Á—cS7ÐÆ#æÑÉM´­O,½³^‰ö¶D«çZÅÐa­Wš]< ~†.éÖA—¯_„)ìäЭ;âôÜyôPwOâò(&‘~ÜŸcCW™Â7‹(}ÍöÆ€½<r!û°í “ºÁ 3DÁÐÃ'OºÀ-‹3\×'¡¡y¾|"¿jÿÔïj\‚jžeÒö0öNµ7ú±g •>—è.^¬ƒŠ„6úRÅP¡f;»’•Ñ-Û™Öž({É—Î)ªî%Œül 5¯Å^ÓÞ„©…:zÖ÷ÐÔ5xîFT´xûE»ÕFCëܾG2=´Ç8Ô\à‚Ž·¬šö¥—¡³áÙ‹^èútÏG, ºå$ß©°7A÷ Õ=Þ¹Bè>¿YðNætGýèÐk=‡xû3¹¡§±à”ƒEô¦ÏŒnìC_ÑTÎe èß>Âåe´ ƒw(Þ,èÂÐ÷n3˜ñ„áüþÜ7nadšs9'ÂÆÔOP)ÿ¬‡ñ’Ä*˜ôõöùèÁÓæ‘ûY¬¡ð÷\¿fÝUf˜«f×|¸‹*Gì™UØ`éíG“6X9¢Éíâ «ŒŸ¬“_ÀzŸXZ´Á Øœ±®o$„!ÑßwïÁ¿ëŒÍn !Íß0ôáƒÉè$& -¨ `F’oU®dÆ"YÜ7ÍiŽ$¿¶ÙÒ¼€”£˜²8T‘ºí¼ØþùH'cãóýDÒ·w~ºN‚Œbe]|¹Ètïûî¼·²p¹šVæ kIå'‰§Ÿýú‘CWsó”Úg äŽrÉP0;мíqËÇöE‘ÿ›7á©!äÐúÄ÷: …Þ8–´¦J ˆËèqM }ÓÍŠz!ªâî#DÍñ(ùE¨ŒšÁeØÇ½æ‰P.$A}‰.åçM4äÄ¡bdBïÀ*T–Õ4”Î4B•ýR|Ê›¨¶>};hì?Ôä<'#ýÚ ¼·¯–EÜŒö])~ƒ:öTk‰º¿œE²hQ_Ý\až†þvµ?oÀ†væqŸ¬šÐhÛ¿«`ÔM>¥ôV§¡Y ÛÇ’¦<´Ð_©è ½€Vl”_cÂTКæðKr3´¡¢ª5õ¨@[býx ´c\ºàFCŒö<ò$>  ƒøºŽêm—.}\7$òôo]¥s@g‹±ÅÃÃMèâ¨öÞ5<]ÝäÕ„l›Ñí˜æÔ¯<æé'IrÝ­f»ÎŒB3Å 'a¬öÅgç=Ñöö â½ EûRb›#Ú„ÄêüKƒ¯Çaì}a¶°f2´§\¿KN¤ ?íHR ÚiP‚˜*Ç·,%¹Ù¡\_ßôzç~[ÌZû@ÿ¥/5-¡0èÆÚ}.ù ½Q޼)_Ãet{S0²åydiè(Œ×{Ôí7ãý“Å;Àä=‘A˜¹PYÝT #Ê27žWÀ\iï6e_ ,ÒMɹ(»Â’óË+,_÷`¹@ÅìJv5¬ª&ü×çª k3²³Ú—aãÝfŽ3°•®ÍHQ;ÕEÿ™ÿ¼{²dõ+â:HpbÎ#ì-/öËÏ·Û#ñ¥þ›«H ¯/ǘÿÃC2–Û¹6£Ha!^fv&©Ò3¦o1üBZAËæº»HPE¯þœ–I8|KL8Ø"$,TY^ëq›\~„lC¡Çþè³ §7aàêøò(* „å×#¿ÄMRǦ³(0÷xZ¶…š*–%¨o Èdâí]3<,±w§ÙP"ë…çbéJŸüúHílÊIß§¢|‘‡òÝ„Ÿ}Eňôç!\̨ìÆ){&ÓU½j.DÑ: úÝõæßS¨9Ëÿ î“1B ŒÌ²ò¨¬¶CÇwuv5•ÛÑ¡ÞU‚û|¬mx”¹×;;r JMf ú^ Q )ýÓ'EkK³Š`^´ ð¡}£Ù¶XÒóÿ ]•_:ÞÌE‚šÐkrAèˆMá¿Ð)*úuÖX:W±[Hx +-»OúÁw»9½›Ù³€ÇÞ­lÛ£ûîš¿{[(zJ©Èï΢gŸ”¯Îżë­þØZ_O9äØìxq¿Ìœ'9œ4Åsf¸l‰¸F¶1ÆG¢µB^BoAûS^wèpz ,ªD­GjÅ™vsÝG8Ù’ž›QˆÚz‰gRÃöP'ÈA74 u‹8Jò ¾ÐòŠ]j:ý’ûõ–£6·°r*æAc ¸âõMõžX^¹›…æAWu‡®- ecû­ ~h­DQ7ÌömžÏg‚j2ïFFþu¨ì‘<):S  áXß“wÀ£CSe &ð߃Go¾ÍFAÍ‚8ûÿh>öÆwÄù,ttæ(žÅÏÐOf“y²v†cM'7îYÀÝñË„´.0AESÀR¸ “ 5urã‘0ír³¼pf~qÝ/Þ€Y‚èw®É0ÇÐý>ä½3Ìu8ä½]4…ùÈ‚QESJX0É"ÎørùË¥ýñ‚Å_¹ãGOŸ€¥SÔJ$£.°,z’3-·–· ΈúÁÊ4uֲðºÜ4•rÖy yцýßu#ÕK°ùº@†Æ# ¶‰5S¿ÂŽÿ/.®·Õ°;Mp')ä0ìÅ‹ú¢¸’·å¯œìGBÅÄv­*V$:n+íM§…ÄYê«ÿCR®çI¿ØJ‘ìéžÙ*üBr‰§ÙB)ÖHQM4Oó©ÎyŽ·FšÃ»IÙ#ôH»<Ë­véOØa]FŸIʹ:d<µ›ýê%2]W[¡CæÐÄŸåü«>Φldõ×­|›læ5ŒîA1È®ü¢áõR0rð]àŒ$µGN:óÖÐÇÈEøy.Á¹©úÚ!—ÝבWõb_ù­Qäójc»é_ŒüæÛê þÒÈ¿ª“:°ÂŠŸ0–YFCã?ñ=ÞF!wÎë=Y(ìÝX“™W‡"Ñy2¬y’(ZdXz¡º‘\-¼ƒâ¶ÍÉ×#o ÄÇx~ÆcB(EYûpÂâJ‡o¿ü£Ž²©4 ƒ*(wì_C+6ÊkVZõ¢‚TIÉä**7„$¢R ¥@œ*/´ÍСj)³²©,!ª÷Öò^ @-uŸ½³÷€ª¥œWl5µ›ŒÆS¡®ã³ žk'PŸ,ä}UíÖ=-®‚†u’ÜÕ3и·•´ÄŠÍhÕ9>Å Å™Ó Ì'mÐêÒFMë´®×]8ÎùmåN¯IJÝC»W½ŒLgºÐAp®u6è0:þ¼!¿kš„Η¶—_Ï £«»ëM‡òxì‚©púZÂ0ƒé•²Ç“0¢CÝ"~F?ÿuÆizßÌtÏÂDNwíù˜z2­#jåy°#¦~§Áóp¡Þ5,äåv=¢„%ã½ð÷ûR°D¢êýð5¢OHþb'Fû’Y±sÌ7Ar#SQO!¤<îQéâÖ€ÔO/Š2GÚÕÈAZ;¤Ïo|ý¤MɽŒüöx3x àt"ó–¥çÝd}â‘™Î#ŠìŽåŸh/!§æ#C­qä¶X}kóè ò>w?ûòß³9‘zì ¬M4‰z|E¡pÊ5㣼(")žwšÆŨÅNäO£8ÿòòm™f”ôåxá†ÒÃìwüºXPÎÐ*Q½ÿ'ÊòjyÛ£¢Á ›$qTú«£|ÅU¾)™ˆÄ¡ZÚ{sgçK¨QBÿAÿp6ap%tÊŠ¨’E—8)Š˜oÊèúT>yŽ÷7º}Ià`º¦†Ç~¹Á¤‡îuJ>.ïÏ£Ç.éó!­«ì&ûem%n¼#÷BQmµ›ƒ×É‘ˆ®®£~TFuIJNB›M¸óõ?РȳK´ÿá wé5º|ëPQ™6$R–e ±7 fà7ÅiÕ¬_á§ÕsEŠcžÏ¼ó «c-D ð¢î©”>ŸÿÍ D˜1Áo}³®ëÿ»§Bê¥}Ér¨ü;óöóÃ¨Š›{Ã+bÕìç·]ÿBMmçIUóL¨S>aÀ±wÐÇÙ¬‡GÎ@ÃN°ã¨°;4QQé+=€¦G@aÍò—åÈo3AóyÎý}hyôœ¢ZS˜>´HA›?YKÉ9h.ßÿ÷JKÍQAgv‰ná7èæ N` ƒžûTnQ"g¡“ÜÎo»ú7"‚Û’Ã`p¨icƆ©‚¾-¹# ðGA‡FÓÉø¿GÀ8Gþæ—°s0Q™.?ót¦ÒûÛ¤ë8`æeãàܘí^Oë½; â4”=Eá°Øeàÿ™¥ –¥ç‹—JÃÊëºgd‚°¦¦5•ªçYßÃæ’ûÁ}:Ø¡êý`{·þy§û7íPÃþRÜ4@ÂSIâ[7‘hïÆ·£‹HR¬ |Î"Ùë_׈müm»R·ÍIÌ Íù§ ÆÎªHOc•¼0B ÑÃu6• È$[öÐ1YøgH±r"›Ãê}a15äh;ŽÜ÷wÔ~ÕŽ ß§OþõgPà–ž¡Bí. ù7""aÝ;~ÞC±ü)…ò¨N”xÚöZí#Ju¦Ì§­²¡ìo=AÁbu”ò D¿T!N¿s•vJ%.Œé ÊNŽ`Ñ€ª ¿º~5Cú§´**ðȲ=7ÍÔ6½Gò~u¾²¾ÿ7ŒzÊÚzlv·Q¿Ž¶áC8=\â'tª*E£#ú#Vr”hÂÔIpãðw4£ XxLì*·]›4ÑŠ×çã_º ´Ö(­HmPB÷›úóR3h›hàöê"Ú•¬ÝéÛ@¢…"¯(ÑÑÈF­ôÊ5tJi—µËBçûù¶Î×è*ÒàóB© Ý. ;È"«¾%®v=Dô³|{ÑS¿äî-å:ôÜ&<¯¼fˆwÆ= âèÑǘÀbšÔµjå# ¯"Ûö[Ž!?`-õÔßõå]ka‘1|=‰w97LÞk"¯°o9´Ä¶1o5×AwÜEŸ{}PÏn‘j 5£ÊT¡¡—¡†‹9ÿgê7¨ùõ—?²‘ê Uj%¾íA}ý‰V¹hùˆvF©ƒÐ¡iÔr¨ç/ô¥I­æŽÂÀ´Ž¢úY®ÕøúcÓÆLO ª…I ÷+f¬Å0CÜŠ1å 0›¯s.T‡æWË4>3\„E“øÐhXþYõ9–<·Ò~tôÐ@ ë K€åˆ”`òPXQáÙ»«++Ïudò…aõ×–îhŸ.¬=U|G ëwÓÎ?3 ‡W]|-ú› yd¹¶þù]ؾn ;öŸ^…=°†Ý*Ûl9wØ3Ží[q9xÿÓ|íH0lÍfK gÔ¢û³4‘˜=Ö¯"É…—ßdÄd‘tÝÁèëÕKxèã§eù&¤pÐ,0rmF*©ãkO/~F®¾Õƒ¥tÜׯ*‰=FúGVU*¬•Èða¢zIE+3¿ôW!Ód¬^ÐD$²0WH¿êDVÝëy{šŽÈèʽì\¾ºµÈ!Gnˆç‡Ó]ÈÙøPæ‰2rÍ]ä¹Ûœ†< b}~Ò—‘WGé=YYò]‰éx›|ùW WEZZ+òS#(P1n÷cÑ ãÙ•b^J¡ãÓ(} «ÿ)üàŽ"6Ä„ut(zălòÔ2Š…Æ»]½†‡»W¨&ÊQâ„$kTJ1Ýú üc ¥Ç(S·£lYvyàI”›-^x蹇ò›ª9ŸÓQ‘öøà™yT’{¤'•óǽ’”E•Ôš³ä¬¨Z› û ÕÙŸÄ]8„ɶ„…¦ßPK×_+‰— Aç’|û“ר­Ï¸óŒuâËÛŽ£î¬oôê_WÔ?CÏù´ h‡òi³ а_DRzæ·…¶¦Œ ¢éršµÇŠZ(‹ÞÝß±C+Ë{ YMhýö~ݱŽÛhË´Á&=ŠvÏnÿÌâ A¤rþoå:î^·`³@ç>‹ìVìè:!ó#RÝ¥•ô½¡ç1ž·çÞ âñÆ«×YÜÑ;ôÉíÀ[˜˜Aûòѱëx¬$5úMϪ‰È‡ç#Sù%®¡,RXÖXŒ›Û¥…á?6“` ]?½º¬C›vºØ&hÚµaÎÕ6‚º“L|-áPµTe»üF*ú–y’ê¡\GùÌõf (÷Iš *û•š'—–¢¢¡Z¶oQæ€c›dÎO_nj†vµ+2Ä¡g^ ùŒ·< i?•¢”M…‘çq—>ÃØ¥ éóOeaB >ˆ–Æ&E·­?ºS³_»çaFD%!F¡þÊ2Ÿ"Õ&¿«Û¹9²b0›®Ë“¹u æÎq)︞y%f+GXxR«ï(xƒnœì'{K‡Ç]WÿÊÁÒzÔ?â›ß`y •$#ÔVF­'Œ=„aí = Ù<¬kr‹ç÷ÃFÔà…o‘°9°éE«±Ûºó¡DʰóÓ¶7ÿˆ üsäå~¾!û|*¹„¦Gà^ÅÄlô?H¾ô‰I?³Å82#‰ÿ;ù¼:{$=Až85€‡NmëVej!ù¾’’¸+R>sÏs}XƒÔf%÷準v°ûbÌC¤›bì»d€ôó¡o&ãa¥Ôy¼(wkï: 3‰Õéäå*d!ÑýÃýáÀ7ØŠC‰B$¹¢D¯¬šcÌZÏl»´ SmÈ’'ö'4Yˆ½›´„ÚYUÍ(/¨oèµy8 ÿõË\ ˆöŠ©¹SjäPîò§Á=5({lHmwÊŽ?LÙð»e?ºS>Œ˜Ay\upnÅT¶TødÀŸùÆ«ÇhT îxÅ'4.™D–§BóÅÑS7ôz=#tuï´i<ðùûFÚ#nWÿ©Ž^é‘îáÐi?(R¶± Åû¤*„ÐÙfòH¦:k+%¶G ó'ÃM&ï·Ð%¬$Ój]mÿ\r¦s¡;¦žòþè¹n85̽ÛÑÜ£¡_”ÏdêÚ& ÌïÿKìŸ!ÿúCj0üh™u±-FúÝvIK5`Ì­«„ê}!LZ)hýž‚É)«¨ç &0Cy³Œ”Df=?©-Œ¾‚‚ùÚøòqXœ·ù|µø2,G:ßÿΡ«ê=þm Á:›hGÅ—Ø‘š¤¼Ûnüɬy°ûõ»íк?ìdÒ„õABJ¢'I”úHt7EIu‘D?Ô3©å’±_ÝX¾…äôÏæ[é‘RüÈEÒW¾Hí%KÒ†H[øÕ29oéOÿ}üÍ ¦f뷆ɆlpZÄ™;õBy9®#kÈÖ7®iÈ.kü‰ï¥rÒ Ø×_ÐFn®ä7Ác©È{B)>+‰ù/èd‹Yä¯%eMš{ …¬K$KÖQx'êIж¿4jÎpÂãÄ#/oq¢$ÝåÅ¢(Â~$ ì0Êq°öE}šEù˜µ?FÏÍP‘ «5¿• Ò;C¡¦+—PÅ÷t¿v,ª™¤Òn@ãß„]2÷P+»Õ{þk!¢Â¿´oae¨½tb‘„íê:ykBЄzM—V§âQ›+]/„¢ÁàØÉ—ùvh¥õ‹XØ Md´äVÖÐt…Ë3nCÍkÝ#²ÜKÐò×ê¹T»×hÕ£)JÞq ­_G¾¦tB›™ žšå´Ýq~ûúÉÚ3æŒ9àv¹WsSå5èh-p?:—·ŸžBçÔË4þÉè’ÏçŸ5õ]{äÇó¦³Ñí_—ñIS@wå^AVô0 _<:탞G߉JG_Æè?g¹N]G›ÊÜWé³Q¸åÛoªnm$˜¸TT,ô†ûDœ=´µ¡•9”+Ùê®E¼!”+‡ÿ^ûùŸÐš…ò·&Aü§È¡TÓ·é|Û(±Í‰'øs Šê"æ=:YáGóŠqøqÔ9×ζ~PÆëWÆÿo^tçó(Óœ'”T»'™/ÿïžrF)‘á(i¨Èrû:¢ ÿIåR5GyAUÒøFGòAþßz}œ€š³¿ïÌSÄAmîEç—¨[óe5Í%ƒú=—yú«ÐxÈV2¡æ4±öÿ::hªŠN*'‚æKòëb†¡Ðâ]Bêá­òcÞ9eÐfüž”BJ Úm ß›Z=;Èyñ¥Íà÷ÐYG»nàbÝZþl[¡ÐSä­qH ïîµzòSéò!»’-Wˆ`8jÔ&8mFƪÃýˆØaÌ>^)›+ Æ·Iüî2rÃd‹ÁȪ˜þÝO¯ZÝ—Œ?›Â¼‚ñwBj„E‡9gݰXJÏïh¾+´N®_I`õ {YR¼/¬»ð’´0Á&^åo_µmó[Î~D°ûHûPf]'ì³–¨l[#ÁÚt}Ä«y$Ц֫çóAõÒ†ã2HÆ&Klc‰äB+ÖKzHéC´Xýß ¤®˜ø=Þ‡t޹\f”ÈÀöëÔ[#c¾ÉÕïj…ÈüòÍ×gÈZ\wÒ9´ª;ûï!×f™éà&òõÙÚC¹½Õ+~â(D¡24Ð?„"TÞÒ_õ{P RV¹ øÛÇF¶Y(ex”êòo”q“‘Ó$CyžqÆ{ÊŒ¨Pë§Ürh•žú‰\å”;àѧMk™¨–ušVPcÿ©úÒ€6¹pu)h–µÉó·œPç\—[ò»Ô»¶Dw×õo調¾ †2k¬Ù߯Ñp-:—äÃ&×{Õÿ=…¦U‰ÿ$.¡yÿ•¿þühEªèXe„Ö¢êZ'$LÐÆ"3\gm¯'©5‰ ] …ᘜ:$ü«£{ƒŽV?Ó}tz)R´#1„Îs÷n_ÓG×#Ñ> £ÛN²/§nã±™ê!îÿÐÃüÞêÙ^í>5XáÎþŒÇÙ:®K7á“Ñ+ÅÿÖÐ;ñÃÌ×xÔØ¨×ÖeÕ@ÖçªóÛû~°ªÚDS £“|s³ÁК@õ¤I Ú²¯||š›Œˆ+¤º a:l€‡}è 'ybËסºxRnø÷AÏMˆ'ìBuãOµõV¨u&ÓôÈ€†;ú¶×#“¡¹ôí\x5´‹zGq²C·ô-mObè$ðþ*|†îêý ^…Q3ý£Ö”½½ScJÞ› SÙfZ¿^ÌÀ_ õ~SGZ˜³:'Eßóý®L•"ïaaÝæ‘×Ë'Üye5Âø,N¶ì‰|: K×B™žéÛÁ²´ý‘w¯—`yä5“Še¬dT½î?Z«a6µ&aÍÿBÜLܬU.ó΄D¢sÈ ›e.}y¾q°ÍT§}î„ì\s¦cóƒÝ­ óÑn°÷àˆiÈë{H Ý|þ˜ÊêZE"'´×WHüyÉ©ìÞ&’Šó0VdC²ßT£—ÎÊ ù¥•ϬªÛHiÔîKðº©o˜ßô½…´ò÷jm·‘^0¢áÂÊ*“ÿ7º‹Œ66k²¢9Ètù䳜ÀÈüÅØ»e"ˆ@Íù.²ñÊj'"»ùbyVÑ3äI§T—è@Î'Ä$'=³‘ë÷•£9×{¶ñ±Tò F=å_F>eÓw‘ß¡qñªÊ2 '{ïªd @¡¡³!§Ïä=6ä ìõ7ÉCî¸bcjªäù/ù(yûd¤®.Å#?uˆï·³†Èÿµ;€gâ øUK;8¡ ÔØÆ] Ñ9hÍï7¡0m4׋ã&("öë±aŠÚû]y®bwÿ4«ì<ÂýܹŽÏXQB”Ÿªþ…J^jØ]ТC©¾Ì'¯®¡ŒÇ)³]ùL”#ýV`Ê…rƒ¦§!^å i î¢Â÷s/í'éQq‰rªû®*'q³¤±W¡ê“ÅCGnå¡ÕÑvEíÓ¨ÕTìoU‹Ø¯'›äÎ:)k“ß6SQÏéô¹ÊdëÀsÁÓß›ƒÆ/3oàaù¹øðûHLcjw¹ÃF® Ü /?ð±Êü•Ü¢ßÐDE ýøI4Ô„¹^´R¨ê±4éýü*‹ë²5))¡¼G=lÈn ÊÎz¤ÊðÏCé[‹ÒÀ“ƒPªîµ¥?¥mn½ï>ŽBYÎÒÖ¡p¨X}“Sš½ XÈ/Õ®8@íÙÛ/©Ò ‘1‰³|”“qaZ.o³zåECkïß]-ahûkïRíÛGß¶y¥@Ç'R²”ôsÐIì—{¾â t²p,¶5jCÇÆ'aFJèè9âÞü®:Ò…xÇ硳óe«W¾t…–ðŒÚd@·×Å«‡L çª³Úeyè]üÈ36Ú ýÙÙŒ#×`0Þ‚cOœ†û?Ð]›ƒY§Õ³eK0ú©½Sêb<Œ[ü÷˜t©&•ó+˜+”`ÚqˆTÀ°þ~¾ÿ´G æM÷©-ü†Å»ý6aÕ°,ä×âzZVzú ídÇa-7#Ègø l”=9ÿÖ̶6w¬;•`×$öý'ÉdØë ¨ ç[C‚Önºåƒ39êBóÊ>3H:4Ú¾­Þƒ‡ºêÏßàGŠ%®ÃÙ£{H-~ó[œ+1Ò^ªØñ8‹ôWH”¥Ü@†ê‹¶ %¡È$/kíë‡ÌßUM…|‘ÕúÞ÷T~nd'kü„r„È1$ìãNì†\S‡ÓïÉ!¯4ek—%òÛËmê?G,MNÅ>:ä¥ôŽ…kó¬Û¤¶/ƒ”ÍQËÊ´´Hd¡Iº¬QûW„c¬§$êŠvî÷‰E½g´ÔgÞ àQ®$M˘hð6«óñÌ42ùvø€‹îôlçñ¡iîå¢3 hž|¯÷©Í(ZÆ®u“– Õã?¡uõ\_½GÚt‹¢ËFÛ…¤’£ÃShO/r%« ×õu܉ÑñظvÕtº¹Zî!ûóõŸQõÍ£Ë$cmŸ+ºq(h]Ï}‰Ç,DÊØôÑ=VQMQn=ÊʯÈKØ£g’éUÿ™\Œ¦>Ùç&D6uIKy (tBj]Máì_6Ö±«Z‡¡ž/»eO ¹#p*ØÎjjM÷÷C¥¡4e¼ó”E¯qÑ]‡ßú†lnïu¡85åU»w>©,ÌCnl¢!/@þ˜7³%ä?oà!Êüò¿y!aša_&ü¤Ô¸>z—ã÷”n¬“¸Ú@y S¾Ô"T4_>=xþ#¾þ¥Sªâ)F“ÉyáÏë—%÷·è¡zÖ(.öÒ]¨YpnËÕû µ/µR& ®²—h“ê7£¢'yY¡Ñ‚ 4îm4™Ž\þà ͪ~žÍ мUFj]yZio 3>ú m 9ɹÒÐnùG1Sât\ѼLôR:GÓ.~©€n·¾·¦¯¡WÆüí‹äoÐÏð¬êð¦; j–Ÿ ᡪ7}9D0"zëL–E6ŒÞq üö³ÆeN¼TׄI‚¬Åµ²2˜žÑîl‡¿”’KWö `N3ÖCì1,|:|ûka6,qx(Ü^ƒå°3i\®°²“M5 u ÖrB[äøa#é黸éû°•ÂuÎãˆ:ìÔõ½¹õâ5ìiÝ?ÒÕâ†QG‹[ìˆ"µÝ.¦‰ó­/YòY i⋸4«&<”ÜÅQ(Œr«~¯Æ‘šU^_-8iS«Ø :w‘þãs‰¶vd„?‘­œ‡™—ÌyUo‰¸wî*²øò%ñKr¹ÞC í׬cP÷í¯—‡šHQ_#äÍÓ³€GÇk¤È_P á—Š˜Bh;ÿh=“Mcî‰u ¢ùërðR´ì"¡`›nAk†ì\åXO´Ö¯Qt¤hRMü~¿í~8?Q¹qÈNë_;‚ŽÎâæ{èô%ÔèUº4ˉ<Œ®ö>©1!èöÅ&_*{Ý©#>$ŒFQõ÷ ô i´ý"y +Jâ]t”鯖{zЫÁ6–îë¨.ì·¯u÷:²`žõ‹›l°Ü«"£>#Ógÿ{yGº¢âc8?ˆBkß…Eþ:Uhæ(ZNŽƒúÇcõA”äPók<ÀwåüyðØÁ̵ªÊ îGÀ} çEÞX¨Þhÿ”š×õs½.¯ ©›:63ÿÀ7®î:@çôEG*!è]ÿ•z¸· †˜¦Ûϥˆ9ßɲ ëVá¨|å“n•ê[a0ýSÁ”ÿ,üÍ yù™^æ„·XŸëÑÁ¼¸xÚéF€‚ä˜PcXn¬M¡SÅë®ßÕ®+Â’p2©UÎgXjÎ?2å ËIˆ:²`ÅÉÝp‘ê¬ê>S+7†5ÝË÷|VÇ`ݤb#Ñ6âÌõƒÍ~ÂfóìÐ %FØVõ¨Ý;;O–gvÂ?MçÃD"6°·š~ëi?„Ú&´+}A‡ßI.ß(C¢qÝ}G‚R$qò šÊCÒéßœ{ïðÐÓO/ǘ‘â„QljúF¤2ÏhMº4ïØ>_wA:gçO'_ÝCúßåÝù¥È0Ä©™}î%2Ñ­­nBfÃ?$Ú~˜pÜ8øb²Ö‘¿çÛ=è5ÒØ¢·‡C¥)çÏtëwôý”B®TÖ¥ûÇûÏEñ’+ÈKv÷˜ì5&ä3­¸·¤•ü¦Ýûío‹öâ"±ÿAï}8ö^¢ƒ›¾Tª£Î¹Àš"+æ"%Q)8"‡ËsŽ=GQ:šZ¹ÅS«Ð—§aÁÃñŒ¾TÌÑ(¾uÿ²C'JÞLÙm Bi5õŠG¥×P–8Ä‹>=åεÞßšå@ùª:Ó9¨àû“Y]ºÃ,Ú'ÞÄ£RZß3#(@åf͵ä¢û¨Jª`ôsÕNÜä+DõËZN?> fÁ;â@Íe<²ZnÁ„ÚG ê‡Q'ÿ;ËíJÔ3#à•;ÝŽú»¦_3>²£A³ZÓñëhÔÖ2ÔeXˆ¦„Ö‘Ãï4ÐÜ*Ë^4W-‹Ãa™­Å8¤cÈÐ&þK­£Ý%´#8´iqíoÞsÞŠkFG'>³§µÐé?y÷Ä]"¤[SäÑÍfµù®nº[Qe–ém §Î|¸š½1¿²kùûz 0¼\™SÀaŸ( òëèZ@]] !‰JË—IO#ÃÓvr`á—õƒt.\îþæõëõßé§žRSƒ–cßÙ„ï@CçD½CiT·]‘™£×ƒÊ?ÉÇr>F@Ù(9ek;”JÞŸ(aR¡ÉÉf‘PCv+ 3*¿ê¿|#z j_‰ô;^S‚æmz× óÂÐñó—EÈ?ÐëÞϽIƒ# ž¤Á°­nø+×0RE\øç˜Œ1_Þ¬¼ã&ž=L}0ðÄêÛ°4Lžº9#|Φt³ë?ÁÔ„ÝÝ Q˜Î_yò&ßf>òi¼›“U~ü€9’Ñ—œObaÞµü§JV, Òk¹½J†ÅÌ*‡Xª°.Ut‡åž«k.Ù°J:?[ðþ Ÿö[¹Nwaý!yvb\l¬˜K6Ú÷ÂÖ‰í@ÊúØ^£6¸v5v³N:0êêÀÞûæG«Hpž[xÜ4 ¿†·e#1yaX†òU$ V¸Vt„I§òýèfñÐéåg|ÝÅ1€øR¦êv!=…Ô}×eør–YU=Іé ïwkhï"Ã!“ÍwK–ÈÈäE2¦‰L|Ýÿí¿{‡Ì"±¯ø]Ò…£4£ù²²5ÿèxŽl´/›«x‘Bâ²€=rPTøDtç"'OéÃ`är±duÿ¹,J~·nBúþÁjÈû¨WeSiù:Olõ@þ”·.?=—P@ÏÇÿÇL0 ’3ñx¥àä”…(£' 質$ ð–¾wXJŠ ^ê‘iG1×Ãî=ãÆÃ/Ê õÄ=Q|t–𤣤lSÖùoƒ(•Džtë5ÊPÚÎæöL£l§ ­ÖQ”˼æ.4Šò׿] p@…m…ë÷=Pñk¢YT*»þúdž'„ªqý‹*EwPCˆXîö{<Âí'å—iŽÚò:Ïø)BPgÉÌïw“3êuµÌeÞÄ£+'«_å¡‘êТ­`š<ó,Ky3ƒæ,¬mŽr£åç™–µ4´VûÐÔ_Ã6ûºµc-ÑNôLûWM´O磤jÉFGÎj⇠èt73ØËT ]˜Ãµß=èÇO±ËQÂxÌÇ· ÁS=ô³5ÈhþCÏ7Ó•z xÓguÂ2ó":»íÆ–š¾Aѳ¡JÙºHøÂëÑï ³04Avêg&´ÙÜ~dºX b>'YŸB5Ùµ^joø/¡±Ð§Ô*žùOº eµ1 ºA 4Ðñ4Kæø)©Ù~~'ô*í¸Ãïù!Ñ·çátˆóš‹¼T°¥ó~K€*M Ý!Póà‰¨ð¡UhPhþ¾Y M·ìuC5¶¡yßü{SÀGh•|wžºM Úœ(̾6q@»ûá÷b¾o ƒ@bÒÝb:Œ[¼ãm Ã²ð!ù²ÿÿ¿òsóEhß8¥tûïtè0 ßܼy°OtÍ¢s¡ód–¼<á'ê¿cIü.ݶù³ÐS¤´8l˜ }Áæ^'™a@Sü£Fz6 UkÊH Ãp¤Ýl{Å;ÙnY47p„±gZûÿGÇ{†sý†ÿÿöÈÞ{oÙ„Äy’­ì™%ʈJ’DËH…Ê(•Q)¤Ad$;2²÷ÊÎ^‘þ~7¾Ÿïÿ8~ÇïÖ븮×ëÆûzœçó}ãÕ¼ßg'oó‘ÞÛå…©te{0³Êú-µ0jnšë9,)oÓŠ©°Ãò÷@:Âo,ð'˜Ÿ9ëÌ>×¢Ô¤,~°îBA  ›ñ‰7î&æÂö˜íÎe—ؽ({/h+ ^ø$+Ĥ ‘ÐÓï‘|‘¸KD2y I3 ?ï-Š#yÚQþEy¤,Üès‹Gª¥‡Ê”#µH«O¯v–| éy¯'ž Y„½\ÉÔÈDo[9öû:2?>˜(mŒ¬ÒÒEÔi‰È6dýι¼9r=/.< D®¼Ê—ŸƒG‘gu„‹ù•ùuUåQàÞª%Ú¢àžð“,Û ~ö/Ë(EÏÆÐ‘Š ¸‡He¹^6JÆn+vžiC©¶ø 3”½ª·+òs ?ý#²7Äc—ûXÍy²ÑؼN5BŸMÍÛƒy¾ô¡Ù[ãüÊ×¹hžwXe#o-jGUÍ[Ñr¬¢kZüZS¦Ëù“£Ê±¤§GÐÖ3“‘ù©Ú¥‰äX«á‰>:Éì~ct`#“&±žGG;…ÇF¡Ñè”ìŸE;f„Î3ç¯ßÃRtá#½C{èºt,vd˜-âm³SÏ/:¥¢ùÖBw'iX+åvï숥QÉ|„!>~R¥»ÐòÄëã,§Ô±ãŸGëP5»œõ. *Üž=fÉ߀Ò$“’¨õA(îô)ÊÜ…ª…Å …øRON¸ª_n{»Ö;Ãy¦¿_ÛþÛ/¼ŸýÚü|=Æêܧùß9僺5bJP©W_rå Thö«ª‡CõUõ½ˆ¸vø>)$PtÆjºCf5¦žÂʸòCÚ¯áÇ´ñcîv+¨k«>ûºêó‚£®‰¼„†*ÃD ^:øI½q>¿z šPd÷JØuh¡b,9#¢-)o롵ÝO››4ÚZ²/Ÿb+…Šö…Ó£ñÐ)å±S:öº¢æÂ´Ó" {—ƒTÛŸ zëò‘8º ú³ˆ>Ë…ÚÁ`“܉¦vwÿs: FΚ:ý}f£?ųó]HaüTMçð®=Lʲï(e…)¢õ WÖú`FPï©ñQ˜Ó©¢4~ÿ2¼‡niÁR [¨ÜÊ X/˜î'ûœô­eZ,aº9%–ÖþöËꆶ#gý`›ðª–§Ÿìœÿ”厴iÙb‹EHÃãÞ?g„ÄJ#‚t¥HJ@Kµ´ß»È6KF¹I‘RØyƒüš>Ry;}ÜEšÖ>Vפך`°C†½°¥mKdz>Ó2•‚,žGTZR‘õaÅ·%2ä,5 yD!ˆ<+ž[ò#P€)]êmÔÊh~üQ…eÕÜ-Ï¢è±Qk³ý¼ú,NöËóxPLfy(’ ¥£õŽ©&£\À²ÚÏK¨À.Õ$³ZƒŠ}_¸„Óá¡és&á¨ÒGK@þ.Õ˜X”Q=ðgO&="™»u}T$jŽgLضţN{Æ4娽œIçnºÏS´\îJ¡¾;‰­Ï‹4Ô :#ç׊Ǵ»«òŒõÑø,å«Èñ4MO;UØÌŒf;âR¤£h!³ñ»‘€-½HšLeo¡U^æ\dW9ÚP c_èB[׸ q¤¥h÷µ¦SïÂÑ}ÿѶt0y¡ƒ÷’[Ìût¬»çñnCeˆ'âÉd…àQ*t‰¶ÍÓxã®–7’˜|F0Ö«ÅD¼  OÑï¼¢ As’RÚdUÊ;ÊXË!ÏÍ;9ÃÈì×;eRèö ßMºÅíçzyw—¡å´ìˆÝè4™–ToËAÝã÷çõ…t¡ÖïØµ¨Iˆ*"¼ µd®ßê©î÷Æt®ºÂhðK<­TÙÍÓX[mÿÊž\‚®ôSÞ¤_´¡ïzHuêc)ò»ÕÙâ #Æ[QJ¹ä0ö)mêÖy˜¤E«Q˜Ò”,\»-3l¦Rté0ëÈqÙ%æ ²üwÇ`^„\‰öÄi˜ÿ³™ÔQû‰n½Ï „%®šÆÍjX*W {“–ýÝNµ°ÿ‚å?OÜú<á'ç§ðÝýʱLÎ kdÄݑð÷8fË(üõXšD+°±èm9[ö=2«½°=|™$:ƒv¼VÖ¡€½ÀJ®´…ŸHÐmà¡þn‰xâÒg}¢ø®×ýÀdk$eò·¸36…d¥§ç,!Åͽúasd —‡;kØuä~ÍÛrø ,òLk©T B>ÅÓ+‰LÍȯ9N'Ãp ù¢ïµÈ @:Õ)>ý)<+5s†”…”åR 6…in ¬–í¢ð_ajÑÓ(²>!mçb£ªZÇPü¤ ÷›q”¨z’À[‰u‹¾<îŒR£FÏÚQæ™Tqú.ÊÉ;O”ƒò\”‘Ö¨À8ó!Å8y^ Ì$¢ÒQµ7Ï ‚ñ¯Q2‹5*¿$;áa€*k—nxÿ‹ÃÃázß.½Æ#—¾»Ô£FÓÙÕ€ûý°?DžHµš)ò<毠öµÆ™&)&ÔÕ9§çƒ¨¯àV_¬‰†f™_ßèà±ÄÒÞ5h¼÷Åösši2h}T.Gó‡×7 |_¢ÅÖáú&o´  ?;ôÃmXÄyKBÐöÑÈL¨˜ ž8È'Ôá0…öÍT-o\q.Þ÷–Võh¤Ï?h±ïcºØû:Ò£e‚$˜¡çÃÒŠ&ˆö:5ï«¿nZ4­\i†ŽŠ“éaü0j@¡ç°cWo¸êÉuÀøããŠu…§`"¼:+œ&Í&?ôÍ–ÀoòÁŸ‚$að{àdãæ™ýÞØ¸ó­-†¦çsJHa–¥¬~„æ\ÊqÛ æ{Xœ"<„añÆqò `)¯ã×užkû=Qí÷É#X™I×j:L «‚‡4Þ%ÁÚ%úîzé\øÛž?1jÆÅJñ¹°YMìà§YÛWBßr~«†³Ã*ëÁ÷aïy©ŽÀ¹q$$?ýÉ|w‰l«y^Ì0 qÑÓ#¿5Tœe+|ÉÒ~PÑX¸"…Øúá°ýœIš¯Á T§úÚïzW"¤á9Y[O¤c5š¬UñAú‡t?(©!Cò{mÎÎÈø>Ù¿ˆ™¾ˆßTÔGæ|õ„çQsÈò)d9#ç ²fôœ~lÙ:ƒ’YÈž›÷n«9*lãXž5!ggï•Ê‹]ȵ*Lëü„yxž*>~µ‡¼Ö‘Z­/~DLj¸ù}/è;ö£ µ^È}ÓÒÉKAÁò “NÂK(T¤êMU‡Âõu|‹½Š(²xW¥RêŠ E–-©§ ø)>™/Y(‘V°ùK%'Û\7$>¢”Æ©èŠÀï(÷}¬}³v¸~Ê]ÿ|¦ ,åõÜ69Qáý÷wÚQñ"Õ§ƒCîxHò´æ–e(ªø7ôÊÎx£šZù½9ú³¨q–iÒƒª 5/¤^==€GÍ_N7eð¢Î 5¯K|_P/1àG[±t=á÷Åc÷í%Ò.Ù ‰¸´òí„F4ó›r*Û¼æã"uW߬ ¥»„¢Çô Z-´H›CRw›¼"´]`·âÐÆ§žšMK£ý€îýv2tô8°rE%Xxö^)ãÉ_Á Á,}è2¹¨#ÁŽ‘Áµ.3¼Íh7!NqË…3v‡Ç½È@öê S ½èJï‚¶_êc"5ШúãâaÔ>~µ0; ߘIÌw‘@U0¥@‹ T”´&½ŽÎ€òK¦Z;êöPvžÎûo”ÁÍæöK(§<}á•:#TPk“fÿ> U‡\]Ëc‡× ß̲8¨Ë!)É$уŸ–¬^EÓÐÂ{á_B´ÞN³ÑVh{-»Åã- íÕ5 íåÐQI•äYÖf?n9Bg¬:ãó‘zèŒrÐ0¡×NÓ+]W¬ “9þ€‰œäþú¦È•0tŒW—[(B—]ªNÈÇrèíú@Ï'=’¬iJ¶ÐVV5ø.úWYøg³``Tí ¯üÒ:—ô4;†ÛÍ&4jœaÔ°…ò<‹ŒMvbù&ZÂõ‰UÉá÷Êý{'š:`F‡<ïO­Ìýz¾óTÿ ,~I¾¾>ÕËÇ2èˆàÏi_‚gž°º˜þ&ÝPþî\¾ßœ2› ŠŸ"ìä`;XÂã;a—|É(¨“ <„øI-Ž#áœ`d 'ÿ‹y»Ÿ·H/||Dú:ÉÏÅù‡i!å®ûä^ÒHUð¤èþÙûHKƒ;>E ½¦jMÂñndˆºïœ}ý%2.ÞPr5x„ÌøòÓU‘•¢¶$Lˆ Ùòûß…˜ßCŽ;Ÿn† Ä!×MÇ´ö:Jä))ð”Ù@~6º˜áƒ5(p6ÆMË*;»›ù)æPØ[›J’2EÈVx'ï ¸„rnË«5”4\¦×/^B©lZ7nœFYîöÔb©?(7}…Çç˜*œh–Ö?c„Šýv Õ9x(Ô3áá T9þJg]ê(>žŸ¬5GbÔ¯[ÑÞE8ðùjKüyÔtø(µ|µ*¤ò ®]DmÅìnQçý‡P• :Ô“×+ˆ6çBýŸëO.†u£áE YÊÝ‹xLåð…CŸ¦Ñ˜y7þÒi4å¬":åñÍtí¤Ük@óä;…(¡Å—¤ïq÷¹ë^[¼ïÛ‰Ö$•G¼»ÑF-uiåÑ%´½â=æúEí mRŽéêà‰½EƽÙè ·-sÿ:&*Ì0]G§…ʾSÍð¤Ifå˜É#tqÌ4"îâBW’x+ÎÉ1ò‰šÇ³ø÷{#”À©Øñß°Õ±ÑISçýQ|£¡)¯q¡¶¼ jÝRŸmWÓCe죦ÇMP®£#Lïè%ÙtFŽ)JPÄjÔ’D øfé¸K2ä›G¦é-A¾àåÂ0·@È«>Ñâãïû_F¢¼ï@ѽ•³¬‚îÿS–.Ç¥ å«?oVT:@%ùÑÙ§ P¹0uÛÕâ2T먟¯Œ†ïÔÎl=‚P#:¼1N[5cZæa§¶¡¶@\6®0 ~Üc8üÇò0ÔEÓg~ã%‡ú†%®±qh˜Œ ¿Oð~~ ’íZ>ÍŒ[¹žžÙïÁV§D.šAkLºÃü5h«ø•É‘tÚân^ßç[AåQïåXèJ¹43òÜzÎp÷œž¾£gSTÉÌ`À%EjÎ× †^šŽzDÁp£©ÆïQ¥Ø˜*$R±DqêHÿ@˜pºEó­È~ “¼d§†i5W-^ïÇ0k¼÷ùÒî}˜O¯6¢2lƒ%’æ†!¢o°lF¬”“+¥CÉ¡þ†°ê.î½HõþJôä-tp³Ã!ÏßoaK™eÈà¨*ü‹(Ñà¾uöm+oʨ#!súÆi‹oH”ë×x@Iü¯Ç¢éy$sKËþ\…l·/w.)D!§[|· ò1úÚ†æ"?Ÿü¿êŸP`p¤fm…*D“4û>¡H[ÔXØtг>|u–É%i hKPÚê­…]'Ê ]9ýô5Ê—Æp:y) bkü×] BëïßlKÝ©Ðéj|“Êõô’h¬±EÂ@‘üŸ˜ïY0lÒ™ý&ì9ŒFÒ $Œç(ùÈÁd•hýS5˜ ” ɦ;Óy‡¸¯ÀLt÷ç™H˜Õ£éMT†92!OŽÐ˜'1^Y‚‘= Ý,mX<éʘ¤¨ K •u·ÎÂ2KäTo¦,/ÏWwª±ÂÊx¼I›)ü™£Ø¦ª„5¦ øk"ÃoÖÚëoýë6FÂ&•AJ*q#l…ÅŽpI6Á?Ñ×=N”˰K2(ðÔ€µ¿n·ƒ ­'‡öY QñÅWô–DC<ŠŠOIG¸"B/"y²lœÐ2Rœðè÷C*¿Ÿ¹Â½ÜHs‹áÇi¦çH—=xíÎÅ'H¿ñã3a 2*ReÒv SðÇw}ýÈܦÕ?™§ƒ¬b_d[/!²ùçne\Dö")²)ŸÈ±©ö<@¹ÔòK»Np"whM@‡¤ò4,1Ú±E>.bsõ™ä?´À’‡üã´)ÝÏÿ @’{Âtº+FÒZˆB’Ë?sÏì¡0!æÖF¢ðxÌ·EI.‰ º²†¢ÛÇrQ\Y¾\æ°.J<ø&ya2%÷âÍ.ˆ  Ô}ÍcÿÄQFaœÉ©+e': ΢\›r‹˜Ä:Ê—±’úœB…Š5Ÿ•ú·¨8úäÓ’Åþý¤yý½/3*CŸ§ãÄT‰»y¥(ô<–I XKÁ#2cõ–'šP#ƒÖó„§jêùÞz"¬†Z“—ÍŠI¢ö»§^sf¨{?¼_¡ÐõÇþdg"@ÃÆ ·^7áq)9O 4ùðÍnþ2š¥ML—®^DóíÑR{ ´ôú^oÖæ‚Vsì½¾ ehðW{Y íH$üÆ(ñD, !Ïtà§þ{…©¿yù{‘Fç‹cE¿ÑE´Ìœô3ºJ^–áójÃSº‘Ÿ.žëÆÈ—ÅnýŸo≜0Ú†«ç÷çPsðr>¤ý],%}Af)*yFLØaÀš&Á O :ÖEMÙƒö %¨žZ„†‰ ºÄâP«¥dTlrªR¬=ü?õþŸï¶ñlp @i’vÈSë]øVs»É•AÊ^wçsC¥B¶)‡‰Ü¾¿Nÿ.rÜ÷Òý„©mœhCÊ¡ËòS`¦–=ôfY‡dhOÀ#ÿ›×ª†0Ô³KIjOñ8\Va$#e%׸F vY‡c`ìÕž—vò'÷Êž’|ä/rÇa’ aÏ„É&û4þ•*œ‘v©çÃ'+Ëv™|ȹ»îêgÖ /ã¡cs_‹‘ÏÊÑŠ~lùM4ÿYõ ÿ®ØûvÆt(õõ¢+DÁ§Ô~q*7P(úQ<é ?7pXE‘’§ñÖa¶(:;Up|PÅEi|8µWQâäíy÷ÐW(ù¢ƒ985ΗÜep›Bi )QiW”%º#µkñåŽÿ>µê(OKJÌ9„ò«#vË»¨¨pŸ©“ä<*m’u½YêC×¢J }T³¸þÛ Ð5>—¥ž9g†š_n¾n+ÕÁ£9›ž¡c¨Ó›6­’e„úê¤Ó?H”÷y4*ä<l›«U…¦G´+ó¦‚ÑœÊÁ‚TÐ-bÒ*YnW£•Àå1Æáh] ÜÄ’ˆ¶:³•i UhWáñ:é« Ú+Ü‘wˆA‡W F~«ãè$½âm”€ÎߪnËMª¢‹†ðÕ«´è*¾ZÅñï0FÄ=í¿#†vþ u¦…{(8ñ€½³vn MtÛÏ¥¾©çMÉSÿÏÿ2#AöõPo$2xVjøH}yU³-OfŠ¡Òá­}jÒ?(ÿÀóŽl¿W•ÎEË×Å À·Ð®‡õˆÿÏwõsvòîB©þÏ«G)­ \ÕëÞÞ‹¨´.½Ù“ÿ¾G|öbOÿ ?|-ežBã%+šDЬ•¨M:á ­ä««ï†Ö¡»ð¦ ‰4´Ëßê<Ù}:D²žo„ެ€ŒÍô6èèJ¯Èñ;ÕÛcQŸj¡ãš3]êÚè/¦ö·§%1ž2èø¡áË— †mM ‹’L/î®*t\¶~ÄÑ=fF }t§¡w-©öí@êpýç'+0¸¬YÚM,Ã"ÇÓ0ÃÈ“Ÿ‚Ñc«0¦"ñÓõÎ˜à™ º° ¿q´)€Ñ¦cZS¾-ÀœD7­á% XØÓû•vú,S&Dš¸,ìçÌ7u6­Ò°ú@DK7ø/üMãp«ûMÍÃaì`›yë-eýì¼È¡ ú˜r=i=Ì›H˜÷îØ Ý $öh_ýÄe‚¤j,~T'ª‘\~òýc ¤4ÈÓÈýPT!©´‹fHÓî+¢u‚ é-éâ"ƒ븆“*26e¤e¾AfS¦ú6-Ndw-Ûyl\ß½5E›¿±tŸ>"—ñßÔU9ä¹ó ß"cùwò&!tß5`¨AÁì{¤ñEÃ(¬~×åÄxÙéT}óÅfõ_ŽÂ~„™œ+¦§A©ã¬óáž(Sî¢}¬Îåò_ “DÞÌWk…Q1Ñy•½)ÉÐÞ{+Õ×ð tj6µ„V7ís{çÂÏ2èékº‡Œ ï-ï‡9O<Øöwj?Wÿêκyàì_Ñ©.~}FîjËÎ$Âø½Yn¼R˜Ôeh+˜„)‹kS‹B0c«î”djs¯ÅTNý<«¨6,}5¹fïøVø/ÿnaGøSvSÍ„¢Ö®¬I«s„u›kì~%C°é™{M”™¶ß›–”Ú_ƒ]“Ï,/¯“!A’ÌÝ™"9$’#`_JFâ©yæh“R$m¿ ~©ÉÇ‹÷tåàŒ±)ìú8ýîåv¤­ñŽì*EúΣÑRßæ1I½.Žï%2¿õÊ\L˜DÖ¥3‚—i‘ãöÐ{^¹ãÈí˜ÍØ­ƒ|ý–º9E(ÄYjp# …,Ïq¹®£È Ř/·ãPìa G9J’Í'çÄ»£TQ¹³»² ʦlyT7d ¼Û²û½‘dTä$ax±¼€Jpw`á*ï<'äŸÅâ±7jt¤ðÈuž€:wÔØv Ký䌚'®ú‹˜l£VSHÆÒ]Ôv¡›·»…º ½éO^B½®Û[]sÐàóÃWÕШÜeÛË ¯$8¼YùŒ¦š,[OŸ¢Y×›¤WhA§murT -­ƒO?‹þ…VïŒJ_×Cjjê3M”h{Ñë(¿+ ]ÇÂÂt´W‹bŸXB‡çŽna»èDøïGVx:_ Tùú± O.´RÜ›êD—²³—ŸK¡kŒäçgÿ`ô°Š¹ø=º$˜ÙÕå… b–ºº¹2ˆÉ_½+* óQ­ùá×ð‡´G{rÐyüör’â´ºX¾ãrsƒŸs²×ûh¡îÈ3ïÁP{Dã<ù­8ø.ô±æâK¨™¢^7a¨ŽÊ‘c‰O†ïÚÿK"á‡éÌÅih„¯c:ƒ ¥6Ó‹õ'tð¨&0P߂XÂ1è'aK(ž¡×d¹\79y5l{¡ÆNÉÉ¿=Ð z\q) `²—¨@Šö<üÞnçyv¦*>”D8KÁ´s(?_̰vV=³dJœI0ÇìW˜Ø°Ÿ?5<¿dºÝ…;Dºaݰ8¥~ñùb?,Õ•÷ ]†åü—ÙÇ3Â`¥€Š•Õwþt _Kß; kÔ&åz©¹ð÷dp#ÑçG°þµˆþ˜ŒlJf{ûþ>l•¥‡h4÷Á¿Ø ±ØÍØÍ:qsà ¢Uþ0ës!áìÕ¢ ÊGHì>PñÐ> I‰Û¯g$!Yñ+±xS"¤xÄ›Kä†÷\Iƒ‘:Á§kø§ÒV \ö^Bú»~tJFnÈð}ž°èÉ2± ‰ÔÈõ!s€‚Û dé¼ÿ'K¦Ù¤ ©·] ‘=èħ@Ñä¨Êu°kKF.ZR1µMäv8Ïã+‘<’çæž"ÑË£•4È/¹!=4»†ü½ÑŸÎ @L³Í­£(h1qúu² ñåþ=bƒB+¡N~ÍŒ(ÜÎ;þ­` Eê/§ @Ñ^¾R‡(N£åƒ oPÂvä£Ï‡”,N Ò$¶G)eí˜fÁ(]ÿÌ›ÄeÏÎòq]½r‘_už¢¼—î˜×ƒC¨ànP´1øoýýªãH†Jï&µJ~tà¡þ×,“QE1Æ—ŒHU»ò™õœŸ¡Ú ‘tõT jÜü¼ä,ŽšúÄá{zz¨µ32O1€Úýîw(³£Q·ãµïFY4ꯖ´%y•¡‘íÛv;<^ý4¾aMªí[:“М¸øaS®0Z¸ë?ªÔéEËA󺟭Ðú”€ážÚ̳WŸéC»Ë~k¯“—ðÄê¿ôÐ&:øý‹˜¹“‹Ž+o¾w»m¡ó“&sÿ_èÂ!öuÂi ] ‰Èµ|ÑuSG5'C#F&l¤î¡Ý-ykz”QÞÑ&ßDê²¹LƒËë0õ>ÈõÔ«Dè ‹<ò;:H¥ü¸OA3cjdÎu ¨­¯Rj‚ïë±v©s PU8g­DeÝæƒWË¡$õðÛ½£Óð5lðg¹?-”LNMYìûí’‹hû(-T‡ÍÍ{õªýd³ÆG åòÁ…Û¶YÐþÔ éêoèªÁ¤ÓìÑЫ¼b½Q ý!W-Ê2aðyÕ:ÃüúJ¡Þy†>n®»7ÀH…ÚÁve–ÉZ;çcFKG9\¸ˆªüËŠã0ÁXtþ­L å¦[= ¿U>tóÀò0¸çÞ:ŒÄoÚùþiw )­d’éêc$»nYG™n‹ä{Q†ÄâH™Ô¤I6„T6>?îÎÉ ú]Sï—¶Hg¨ð3áöÒçkÿ)|„ ub-â_6±ÛÛMdü12 ÖhFì„#óH¯«qëdùýb„ày=²® ÷QiX#;‰Ú™z rð4g¾CNµî:ΤUär¹×ž#ܱ™CÛ OÓ¨û?ò1vßXÜ@~iJþgM Èß×ôä›2xúâáÅí‹(x!ùfƒ= žûvý‹ û´=ðü‚"w3W¦†¨Q4ÏàCæõ7(6ý™€´U%xLÇtø…PÒª³–Í&³Þv5§Bi*¦†¡®(“u˜5ÀùÊqs¹ÈýE¹ŽìB¡4”ÏË™#H5A…?ÔWB$P©6ËÛ2ZULÞ¾4D5Ï#G_ŸA=/ûÓ5­¨¹))ÅþµMgglߢ®=¥û?¥_¨?lý·öÐ:%„¾ä¦FcïãëUô²h&&-v‰‡ÍÓ›ôŒŽDË#<Å9.‰h5PñûKŽ Ú„&4éG;–6Þ¹ÀF<÷e[Þˆ|é{DäÐÑë’©5k%: 1®¸:áIÏ S§$£KTÔÛ<&"tõÜ:JzÁó½mx÷BÑvï¹gÂb6 ðÓ6ÖŒ­Wf ‚i¡GËóHá '4¯HS ñ=ƒ±QåM™Pý~Î=â^T>7§ØÛóƒŠ#º}ø_B´´Æ6CÉÊõ_O>>€¯oÚúøý÷, î ÓHº ¥uâ¯ßúCE÷…CJÿW;Ñ;ûo§t+¥ð@“ ÍKÆ¢Ðrß Ž#‘Zãµ(à¹ìô§_úé¡ýnëGþ‹#о÷ün­¹á¾Oéü¾ïnÿ·n«cPŸ í&sùõ…·¡Ýû¤xƒœ=´J|á™W€µ[^œÔýÐ1{èûMèìê \ô0‚nÎÎs]—z ´‚dV; Ï²Ä~‰é ¤±1Ÿˆ…¡…‡Ù\`øwý‘¹•0z'UI2ßƽ¬ê–ßõÃäƒÃ ¬`ª?)÷XÞ-˜õgàÝL‘ýE-£!2Xz—^½šÛ+'g„Þ“-ªäëþ‰OEðWÕ,}óO lxyÓßê•‚­÷ Ûܰ£xáGPÔaØ[ãÖ·þhŒ„ו踕Xî݉Â÷¾HJtÍ„Žɶs%iŸBJú‡‹N+ŽH¥û¶üa3Òvq”‘áXÚ¹&õ&d|ßÊæxà+2K“ÿÚ;F€,eÛU¢”ÈæöûÝ·KÌÈ!¡žœŠ ÈÅ›Êìj‹<ÇÝ®ùN _΀µÈ Œ}0iÚAÁÛ NåÙŽÂÔÍùw#^¡HõßB ÷—(öÞîѲsJ|¨“:sn³>½o…2§ç&Ï|B¹›‚ïÂÊP~ìçëðT´tkð]F¥‰/4%µ×Qù%y Ë[T}Ü{¯Á$Õš¤UçÖ™PCW*/Æâ>âbÙ,©!j9Üôþ©„GëÝn— ¡6WñÕ_FÝ’ÁÄ[)¨o]RRÓ«S·9ôjШhCt·Û ?;gzLè šd¿¤­—F3ûé'Nì¢ù%.à¼Z´ˆ£æü§„–åÎüm;UhµõÏÜžrm´Ž½xU3†¶±«aÛ‹Ëh7YxMž ížïŠéòB‡7Ãf·6ÓщE1ñí°:Çœ–Ÿl„‰Ÿ šÐ ©š×a:¾·H®3M4~À)œ²XÓY‘ê.­çÞw¤™ÛΟøŽôŽåŒRT‘‘}ÒýòW!dæ™9D‰¬N»V}2½ÚMǹ>ëÖÿìéDÞV‹÷=3L(Àj™ÿrÚÇwUÖ¦… ùÍc‰bÔ÷£>£D@ã7Jñöz^Ê'F™M²@o7J”§²Ì-¶£B…¢wÔoÄQ願*ki&*Çœ°ñµÂKm¦jVP›MÛ,Ôü)êTðÞ¿õ.îyžÌù3Üò]ÝÛ¢·Ÿ Ýã“…'N¦ÃLë í˜Pœ0²$1¨úÏÿã§ÆP6ß)ÑÿÆ]këíÐì}¹þÝ*´õ ½c-x ] %6äE½F±G^|:øŸ‡†•'¬ ?¿QÚL™á‰N«c&òÿn ‡ÎeӴäöÂL˜\ ûû ~Gr g}‚)É+/—ÞÂ4ùø‹_60C4ª?=/ ³\þö…ñÕ0gžB˜µóéLáœaQ[¬o̵–Â?Q.þúßý_yÆ:æ:ê}]Ô}Ã[IYÒ øë-qã¯]¬7ÿHR’3‚M»7¼ñU°ÍÃ$*;R¤|d °÷€IÑÜãrßhÁ2y$ŠÝd‹ßEÑ{¾\w*‘´›5q+ˆ É3«¨þI8ïçÐÕã/ä‘*«nÂ9ü Òôžñµ¾üéÙT¬ó›#ƒ]“ý¶à2æ$¤äÜØï‡œ‚çí—¾"ËíÓ·fN!ëØ/>£ñ9¢yåYd9‘sÄš.»¹UXªˆ#ÏCãÚëIÈ;s”6ò©òói¨œPFþfÚÚ2åk(&ï{'Å ´Ruô Í>úÜû¾€šZj1þ&h9ùô³±Æ9´~B«5Äü mõ·¸j®þù_Ï’é¾xë„m'–¬ªäÑI‘Ëê°j*:?MX±ñsúržZEø¬m£^Ÿ1\/Aþg‚gS¹Ò`síSÕÝwèNxj&õ*šƒß²sÅAí¿D—¢%¨Vpà9Ø•Rá%tÃmPÞê¼ë]¥ÑÖéߤê¡Ä¸†)1ô2 –˜Cñ·SqsŸýž^ùN¯aŸÓîŽ'ïC#[ò½;¿ Ú:ãNòMN¨Mç^ÞÚ÷aƒszÇ^Å#hRøÊx&EZ®ÇîÏ£Ðz7HÆÆÀÚâ=QÉBÚÃuŒ‘ÿo?d>\ÿk}Ú'¸ß_7…ökÒG%¼¡u›ÍÊuŽˆä­ˆYƒö˜BºN²hŸ%ìવŽäô&AÃ{ÐaöDE`º*ƒ8¾fþƒK.)¿éÐ;ýÁUŒû# ˆØÞyk®CÞ“nqz³0œFõð¸>Œª¬lg/‹}›““JùüåaêºXB‘:ÌÒ½+®ü óƒñâòÑj°ä¯&Ù) +ÂÕì®ë‰ðg^ÃëS)%¬mžfúxj6dÏÞ§uƒ­ 4Ëí)Ø¡b¬ÿ{-y'2Åç‘Ðs[îs"³wËœE’ÑÆÆ/ø¬…&CðR ¼ špF*êïÝjèÆÕ­Gí´<ÒuM_éHŒEå%œ§üŒÏ ¸sœßÏ™Q¯îÓÄ K®D»f²ÙUfÊ£FÞx¯‹z‹^;›¡Dy`ï~ŠÇQäKI½àºæŽ,+]_'¢ ?Cz°C= m·-*K¢È;ËŽæñï(öཷ$J$¤ùüÉÐÃÍÿž¥_üŒ2ŠU.h†rRN®I(_Og½¼ñ5šØÕìQ©çÐ7}ƒNTNñ2¸£Ò…ªO~v«‹… ZK`¯œj¨õ—žgõF‘zÊûîjY†hnÙŠáѪN—&¯ý~R—÷ ukϨp¥G¡¾gv¹9Ÿ)2=ùA[$ˆF¿¾H?ò?‡Ç¿ ßõØD“¶oç^5=D³sºñóºöhÒÖ½-Œ/.i¿+AËÚ¨ŽôìhµWjè¾J†6ÚeËJ­Ðö¡_ç¡ÓßÑnbmþCá´×+Š· B‡ô»þ±E&èÄ|ðšû»ËèÝD|5:}ŸÃ›|WÔÊþý°7…Ä+or ˆÔF³ # 7óÓÿò¼9÷]X‹RÒ7†NΓdSòPG¦‚£JªüÝ 2ºêJ:òføzêÒZ).(TN™)$„|AWWM®øœÿé·¹[à~œZ’-ƒœv—ðÁªÈ9wS ¾êäüúHµs>kÒˆÄÈå@þÐ{ ®õÿSùÑGà"”œ1þ˜ E ¥r ãdBÿ{oyBäý÷í PA}(¡R‰ *fÚ < Þï‰Áüƒ%ÛPE†ÊÓ)7 êÓDÆéÞr¨Np?žq­¾~!fjܯ®%ïçé„7öËÑPQmH´% Ã#¼kà§%uçÛÖÐG$ ¦²--ÄÏ"Ø fç½j´{»=±ý<ÆŽO®@×ìW­W3¯¡gƒ£¹ægôó±ôг˜Â`oûÓ0|ÿ(±èŒ¤ïÍú¤mÃØþà3Ù‚ ‡£ï8`r+ ™Ví:LÍŽ†ÄgÁ¬°c»ÙY˜w!vMiÝÅq¿BnX†íŒO°Òøq¯Ïô&¬ú˜^.¸µßçC ú5`C(…O½ülé1Îù0ó¿9/ZÚÿÁÞãe›²(m$´Ômó×ú‹D{)b¤;ƒHÒ]XRrVÉú}cžÛ#ÅÜ%Ã÷zñHÅ2Jn‘Œ4—¦µú½‘ž›äPÄh24Ç\+Õû€Lõ¤dOÒ-‘e7¬üæ®>²_ýQxC¦¹”ÕOpp²!¯vÙh÷@ä¯cëˆV¨AÁG«Á'åß pDÍEz}¾[æUJ0^Π~’‰‹ôµÅQ&qÂùtíG”~}Ö×`¥ÓÓ‘¢’?ëHëq~T¾p´ ž¾UïÙÝüwúªÍ-µ=Ó%GkDõ×5´nüõG­—çâßµð£¶æKNenÔÙØyÛMFŒzí9üùРÔ">ó\u 7^°¸‚ÆT¾½ŸBSë]Fÿëh6¨–(¢‚lïéO»£¥SÆû»FÏÐêóOÁRhÃr»zcP m¯ß•Y™ý_9sx½C‡œsE•zèÄ­WãðŒ fÉE¥üßfÿ:ÑTŽ&™”¥è#;œ`¸ˆr¿+Ó4ý¹‘¦»Uk"G¦O9VIÁÀÇͬà¿tÐ~Y{5ÚW€t”îÜþ½˜œwú µ)š'?_Í…ïn.!v•®På|gñ÷>¡aT7T ò*ÕËPu³wOàw|o²:éÀúêR,®ß€¦E¡Ìñ&h˨Š_Ö„.ª¯è=KöÇ/ü0 í0ιRðógEÓþ0jÂLØÖeã’þk 0‘ç¸7á÷&+Ï<Å–9øíV$°Tw~ÿ:47S S^>2Ňå`Zá £],!Ìä&_H…Y ï¹—¦0wé©^ÿ!˜¯\úºõOXž-‹…¥Øãrt4f°|/K&Ÿÿ#¬¼R;`µÉZ;çt¾2Àïœ÷[Åóð÷rýŠ&°>àº6”›î%3µÇl`[\ÉJ˜‰vd¿f5¤ÁÞmÏ+ç˜÷}FÇ«qäù ]åS¡ ŽD¢:‹'*ÃHúfÇF}N ɽgR>!¥ók¿ ¤²QíLt-Fš«WÙ6´´‘®âùSr.=dठ}!³µŸ?]”kq™¾¿MÓ]Bá ûFþÈbd,^ØŽl­[çïÍ"‡Ðî[izä œÚM[C®Ö±¯i;·‘GžÎøy òÆ,u›§"ßÜsŠôŽ0äÏ=ÙpÔ\vÃÏ8‚âf¤£¡à¿¥Å®‚ß(Ô/ÇyÇa…ë;«ë+P¤yéþÖ† ŠÎ¨Ò@q®@¡® ”°«¡¶ úƒ’™ëçï_E)æò#ªE(ý˜kÁÐ\eYÄÜ k3PÎø!…³˜0ÊsÉ}H”@ù½Ÿï­e£"…åiç³Ä¨$tiBTÒW<,Êq•?Ì´fi^FÕ“H}}ÕP-ôŒz®¬6jðE\Y÷`DM¶Ü/«cP«IǯÕWµßE(<ê;„º¯{ ËIPÿûæF¦…>qQZ­ª¹æ1€¦Ôs„¦ËѬ.…‚mijŸG’{hy™å;Q¯ Z ¼"™CÝœrC¿ÑöMËó3/dÑn“1¤§b`ß‹Òé3h…á+yä[ªèØÜ!Žvè,ok¿L€'_-„Nª~C—G«mãÙqx'{XïÚ̉?XóxŒ3¨õ¤eRê†bþIw˜ê©M÷ƒî’Ö3×j MÉ’f¾4:Ùy]e¡öA SwòT5Ù±t‘º@yYIìƒ~øÿÈ:˨ª¢vmÓÒÝ Ò]"*<t#­ b` "*(¡‚ ¨Xˆ¨ˆ€¢H ‚ HwHwwwƒðñ~?ÞsÆ8¿ÖØ?öœcì±®yß×Þs®÷}v*Êä!伉ʳ.ð‡_šç˜L" ‡U¶áò9äñæß t%„"Û£“‹¼P6]ª:?|jÜ>ÒÉqBÃå‰/‡¾@ÓÚ÷'VazÐJ•.YG àúËæt |ñ;)Ý ]%ýä>ÇW¡çÓñÞ€>ýlâlÐßXù­oÄ >¦ò1³! ^ß¹2 GC'è£;vÿíçQÒSÚq,½b£w.w$ÝþcåÒÝ!B0¡Ö3˜n“ÿ”Ø*žÀôÄ¿šT{O˜£}qé†',ˆÞ± {\ ‹§Ç˜Øa)¡ã›æZ6¬|K»_I«—O5 D%ÀÚ¿%-”C°QvåÒøC~Ø*›ÖxÅj;{ó…”K°wò‰‘#§/̾-Ky$DÊ}ÏäÍØ‘8¸Ü›’¼Iú®¤kö#™>wÛrüu.ò’Íé´y.4¤»!}rØOwBdø™ð:£9 sŸasB¦ü´Å~ûd.Ë×=΄,-UeDg!ëÌ3ÎÕ‰9d§¥îýΊ*¿có·ó¤½û“gÄÈõê4Ë€(#rwÜ̺¦‚¼"ÞÏ¿\=‹|wLþ]Wä÷þóÉß ðÁ¢¡Ì~þñVŸcùŽé™ßÞ‘^D!ú÷²7tP˜¯>•a%EÐóûæY½Pì¡"úÅ"æzÛ‘¢xåä,ýJR‹ÇÚœ¡D)Ïj¯go£ ™vˆ:kÊ>†¤naV”»i°CHbòö\ßöl]QáÃC­¨tûsêÈJ ¦¼»µóL*ˆmžÌ“Fµ6ʲž¦DTŸN°Ì¤ZäÚaam…¨cDN{ò„ êUR=UäâAƒ«•ÛÜh,K£tÌM—ת—˜¯ ùÙh6F´˜´üsr¸­¼Þ®Âà#´Þeñ÷y„'nþ’rðG´móuîoü…öü@gŸw\è9Œ_ šßªñ6A‹K¾äÇ6h=\'â×mx«br¡}n¥ïGß#èô9ë,\ÝßÞˆ=¾½ÓÇx¨·ÔaÀÞ¨óÚh N›¼sŒ†Ë%N8e?‚Ñ…GV/¼`‚7ÓW×t ¦¾ëZšŽ‡ÙëÇèØá2,¨Ö‡ÿ| ‹“Ñ4¬Ë©­Û¾5v°š{#Эé4¬/»mêh½ƒ-=ï¾æõLØékq⚇½$#Ø'Hh[la¿…Ä„$4_Ü‘¤„×ÐT É>)%Š´#ù·+9§²yxî+ÒˆK²”—!]ürW#ƒ ²Rµ!cH1±§?2ÓÒœeoàC–øÃTÞÊÈfôáËçƒ$ÈÁ4§¢Õ:‚\¤k¿èX—ösó{‰óGä{â™]:‡”„{âò(è6J;kT„G/X¤¡ðó´Ec× ½ôýÉHgŠ_ugÌvAÉø¡Â{P†Ø-f˜}q?ÿ”éÔ¼ƒQ>-ê9^pBE±ýõ‚¹•ª^l¬–Få÷?4íIPåC±`2ÁM<ÚTò{ñ¼5ªÉ…Tu!6½Õã5ŸE C_¶¿ƒ¨™¯ó†Ð·aߥ“(¢§Q§q¦-ó¥7êù‡ŽÞ¬F}}I4$õRa¸ŒFsý%/ù}Ñd££Ô¡iÍn~Š”ËBó“qhñ¥+Aõö3´¬Ï~tKõZ“h\ïãC]%í2žò}?œºÜµßKO‰°¬ =¼jzŒñ ‘Ñ·6ЉéàIaïð}?T­™hÛï¥×Ø… \ÿ—¶þj™ˆìì‚*ÄJh&ÁN¡¦ï!…ª)Ý`e¤Ñjµ¤ð?û¤©§ªI Ò´íó£dW(Z ¶®*…?”—V¼ÄÿóýK<é-ÈþzÅ“‰2“òÿØžU† %ƒ^Zê8H¿¡sbd¾ÒuU Ùì m‚» U±Òþlp‡TÂ? ž· sÙªÒ„~à¿ãäú,æ–pÁïÿ—M~ÿFu÷þÏÿÎ[(«Æ/! …±&y&iKPä- MxCußÿ^w[OâugÇœã¡D<<{„”J)¤«ó; Œmp$vÇÊÖïw)ëè@ÅwŠ”²xS¨ìeó¹fÕ5ë,o@mÈ´n{Ô²’^p ††ÓÉW~Cã;餋Ú'÷sú£Ïó+3Ð’üþS5´¹iR‘@Ç…¬gÜò· ëQk|6¯ô²Qϼنþï&º/hïÀ ‹ñ° ­ñ²›~ƒ‘¨7ÿv¾ÁXÈ Çw¡0qÓpÕC@¦¾ÚÅ[ƒ™þ ¿Ž 0¯/V™ìb Ûmº^KÁR@ÐÆÈEXanˆHs‡Õr×:Bî°ž0/r#j6s4ßß^Í‚ñ =ëqØðòŽjUG‚탱}Hôið§Lr2’Ü S?ŸöÉ®6÷“ÑCò»&µ&˜‘26@ÄÜ©·62!]Ö³ô£AdVyNÖt™¼ù%G¢‘åcݽTíd§žNzÏ¡Žœ9 ãnöÈ“%ÍÖ*чü¾ÝË·ïh¡ úTa¥- ɦ}=ÂŒ"&æ_è¢X ¯ý.‹6JZfÄÒ£Œr©¶ƒ¤5Êý\å:Ò‰ ^ià¿&J'ÈBC;#PÙR„žU—UÜOËß=?‰G»Þ³é¼FµÑõË¿–Q]ž²rm…5þ0·\î4F­ ó*̨#×S¯Sä‰zì| =QŸî¡Ë Î4”Œ'|wOR—¹AÓ÷¯9_àE³­±¨foM´]Œ`~‰–žo¥Ú¾¡Uáw™þ_´áá:2z§ OKX·iÿ¯ï­¢CJñùtâ´ezrvÃ+¢c“Uÿ/‡IÜ\­+øTáÐb¾÷'tºèÑqÍÏeíŸfV!RskÖ\#*„ñenö’NèöùF¸çMi¤þLZ¥P_35¾ðªÃ3¯]h®€ò¡¥/B¹PšóûiUÓ#(.øÊ(M· Ei»¦_öûçg¸õÜ›ŠÃcžwÓ„Ò^æ’eò¨l´yJÊyê]ºzwŒ÷sîu‹D?´‘®èþýÚýŸßÅ¿k9‡^î!ÍÒs` 8aœÄ†žøµ”–ÿ†‘ÐÖñº£Õ0&ÓÁÛÂãöô“²0Þ2ãgN² Êfïšja"é´Uc`L:=LvVO‚)­îáz˜>yÆ/Ô§f^¿le¡º³÷ÏŽÃü¿µwɇaáí„ÌÀ,¾Ö“Òƒ¥ïNòÇ©¿Ãr§ê-c²pX±lyZ‹°æï1Í9ØëÛæaÓ«9Ý“`¶­} ¾ôþg?L“©IìÝ.úP̈„„¿ûü@"—䤧‘¸ÿ€2óÔ?$õaöwvÁrL©Z—‰‚BiG»š©ÊTë†#¬ðu'ÒYUÇÑM#}÷ñx<º†Œ|âu=»ÛÈtUåÇø;ad.û—Œ¬ì?}‰E–íÜ¥ût«bÈžÓÏ£½¸ï‡ôwÆŒE™‘ëbzÆ‘&AänhhÖb[DÞCß{R2#ß[ňžµÈ¨¶Þn"Žª5 oØ  ùIÖô·(8zé'ÓI<ØèšÓ9h‹BÍÿ49˜kPxöF¹ü¢1ŠrvtÎyÜA1‹ÒõT(þ,hDÏß%ú¬P&3 ”þ‘x[ÝO(]2Ê6Y²`|`9äÊ9Lÿ81Hƒò¼sC P~eÏê»G*ÌZ ]PCÅÊÒ.<Ä7—Î÷ê¯Ǥ º¦’‘vw­÷¯Û¼TcSÛj1³Þ!…! _âÆö¹ä$ZL*†jºãfðW£âñæÕth¢’v •t„Ò4‘ßÿyþaϲVú´õ zÁ] ·÷FûïÝ‚n¬'t‹ôƒž¶²‡Ž ì{â á¯[œÐÿ¹ê¾ˆZ D[ùV.äÃà­Çkrâ0ä©ÓJvë; {ÓÌ‹çÀÈcÑÉGáM0Z]•jçã*ùfg}`bö—Ÿ´«Lýeg›I»3¥î¤î0ÏiˆY½‰°PÛ-G¯‹Û_~sŸðeªJëOs°ò½ó€_â:¬IzÏ^õ…õ¯Grxg`³&ØŒÈ@¶g¿¾Ó¬dÝC¶A)ÂH`èΣ%‘ƒ„C=ŠÁ‰ªä)kf‘„ñÕößD?$½<ËEÈìˆd#š—ί"ùÍñ‡«âH©°;|ôR34Ôym#íîB9Ú?HO"53‰ ÄêÝ8¹…Œ§3?»Œ#A·A´å%d&SK«C¦£® ‘UŠþ—1²™1o®ß´DöÛœA~µ È‘°ÊZ˜q9û^®ÓGþEnáSÔt­SÈãuãúÆoäX¸Ò²‹üJ/,íÛö{*HŠ* À²›šx´? Ž 58àÁÁAŠ÷Æû7ãÄ0MÂ/(B¤.Ù¢bRZ~‹lû¼–ª^<‹âA;§Þ;£DÕPùvØ8JIFÜ}˜ñ¥sß1?/É@Ùc}ÑÝBÛ(g©q·Çå1³¥ÙUn—Ê ’ã?§ªüTîtwéMÐÁ#CvVd×QíÔîÍD¨þ.lsõCj~JðcAíü‡ÿÍ5¢ÕUöõPÿ^÷FT¶P\“¡G“à;¹µˆÐìéÖò«”sh>CSu-„-Í…¿~L±D«L¹a™g¬hCžÝë—€'´n ì0¢íeA™S?5ÑîI_Â2´Ï¶ŒÁ@t˜JuÑxZNÊùLCÑ#û9y¸×É¥ƒ{Ú¾ö¡D›˜Lû´½È'øäihÜ#؈òÌ´õm‚6õõÛÌru–\½DLÁP+—î%ÇTBÙ-j ÈhIP6î5lüÚ^s ò¥í‰_¹Àoß×|:À˜§À`9¬ Þg!'wF<ø£ä8¬ä¯ø h6ó+CÉgþÂZD¨8G"MÕ+é dw¡n†e™*ŽNñ­¸ä@ãÕŸ ˜8hº_"«þ švø/éXü=õgšüÍiœA·[ Y°xMÞdšغÎfzCSä©—÷×…¦©–­#ÅÐlDð…OÕdŸó $¥5~h™ÐÏ,#m %·ƒÙnC{H}øJ³ t²¼¸ôZfº¥Û*êöó´÷ü³Ö ä×0ÀôTÌÊߣ?s íÁðõ×¼LùÎ0z?UëgZ6Œjœ,ùSº³¾­0K>°º; ¤®k¯ÿõÀbÊ-šŽqX¾v9T·#V}ÎsjtÀz¶Ë¥/·š`‹ÙÈÌr9v2ƒ<èôaïሟ·4ê>¾>zÙ ‰&åÎòÅ!IÜ•j‚£Hæ›U¤ÀþÉoônœ­9”ïˆH_»“!õ¢¥¥ˆÒ]?@*KŠ 4Of$âr‘ñöÑÛ>ÊÍÈ´.ñùE ²D-µñ!›ú­xö­ä Ð#e‰Y@Îío¥ÔW ‘GH9~Üzù|Å i^y#ÿ.ƒh{- Zÿºå¬ök?/Òä¾TFá[´4ÕƒT(jÜJåüæ ŠIdß“ñEÉÛ•ÛSIQº™ßïÌÒ ”SÄ" ”#\ÖCƒŠŒ1é£ýû<åâv4|€ÊQ¶wQ%6ë2Oç<ÚJûïzìMT9Ò$-…XJºÊ³¨¡usî³5jfeñÇŠBmè‰ñ{,Š:=^ÞÆ®¨÷êºÖímÔ7Û4ßüнßGŽæ¤x¡1{Ô¿lÞhÊ»ÂÇÔn‹fAÿX¸Ðø‘q?å}ÓoCMqUÏ¡Úþ: >åG gNúŸ:•bNß?¶ÿ¦€.Çyhá§ÛmüλŸÓ•mNoÎ@{¯¥ÉÕ×× séõUgчÐãµ-Ïv§ú )xþòÂÀ’D}”P0 =_3øñŒ(5™õ“ÃçaÃÄ(˜ úyQÝ¥¦¤l®Š>< 3'ÉŸíŽÁ\ÅÁsS”o`á—ÞùìÀXRrXPråß'²¿éê“®‚Ô¥§°.…Ò)7æÔ5£àvØ……®kØU]ðæWC‚èôæÄH¤-¤uÂéî~ÞZ–²n² ¹þñ±ÂCHNý¨ëñK¤”õß$Þ2Eêûµ å¹Hwˆuƒ¤…T¿˜:•!“à1s…í dÑ`Ñî MG¶Ú˜_ÄiÈ©*~#õý{ä‹2j6ÔF~™F·²"˜$•/öƃíQ}®Ü(<¯xeuJ Åô+hæŸ÷¡Ä~÷Ûk(Ý'añt?ïBuš;% QÁB¡ýb¯5*Y;¤{–•QœsõÕFK»ž”x´R¯â¥®ª R¬@uühvq±5: ™ä‡|QëåÉäxI Ô9÷{%óNê9mé+ëð¡þ©õ9;âå?å-fÔP—Ã,õ˨9#³vþ'9jÿ$i|øãê~á>R|©—©m¯£ÁŽZ„GOÓH<<†¦ï곌±£Y\fÜ_m@ó¢ ¼ëÓáh1úiyÕq­(þÞÿ²ï‰}ÅChs8[IÂñ¦5¢­ÓløiY´3%¼áíl‡ögÊn|&|‚aä¢k&Ð1?‡Pº˜ztÜüÂz ­/û ~¶A±ªÄ¾`7D2kM;± ¦óx­•–&M¡Ák’õHáa¨â%æ[#³…²;¤ ü–cPL'ëumÃþ|ÞŒ¾£# ¹Ç…Ž\$…lÏÊ}í¥Zù1‰Y²<¾6×@Nè„Û¨‰=Ün lÊw‚’ú+•ËÞäP™³¸ìÕ³Ïy~9Ñ-VhäK=VTí M{‰t,i2Ð’síñ–%´yÚ„•Qý‚9þíß;ÐÙg)Lþ%ºO›Û_¢‚žb:»TíÐG­T)d´ýÏíèN».Â@±ñyk|Ô—º©ÞCA¤«#<0üz`Ëí+ŒÔÐEÒ­|Ú-{ìÈú‚‡õˆs²ýÙ1µïGö…gåËí—‘S¬´Cƒò r];î’À÷¹þØiÐ#¯NdÉEï?È7RKjКˆü-ën‡ø¢Q b—Q /£`~Áãk¥x0Cx¸ˆ£…Ò×/ªy¾Eá2‚ HD‘Ñ®²Ü“(F1Â÷¸ ŵh[DËjPâá»é˜ÌR”s ö¯óGéË‹ÑÞã(3çÞ-4€r²*r[™Z(/l¼—à1… Wï„Ñ÷£’º5UJp*5©üSp¼xœÆWˆªÓKbM²v¨nx-FúÇ/ÔT ¨É>Ú²Ê.ïƒPפ-Ž:p ¿œOSÊÿˆ-Q}EÖ5hÌA”˜ üM/¦ÆÈDL¡Ù\EKÜU´Nw$ðåEË FºCo¿¢ÕÛì§q•ôhqæ¹›ÈE´É#3vO1Ày›q}¾GжˆâŽ(ÚµxO³Çp ý–DÅH»:JZQqI)cðÝ Fã'h#™g}šnyÙÓgåí¾ÃúLCRAéchyòp-j¯{Ü\§ó‡r:ú|uß2(.HÐXÆ*(ìmÅëP jã×¶©ywzr ÷dtêËy)øõ7òÑÌR$üÒ¼tôý?·ý×Ñ=+†±ð›kläZv3Ô2stC‰QOßð©%(£l/†ê‚ìz±%¨«,9íü™ ÌÏ¥n½Fwá˜FUÐä3%›”]Í~”ºò¡·þç¼áæ¯Né¡§Ðüúmç)BhZ{Óõu0 š®ˆ[tª€&ó,—Ê”ò}O<ͽe©M;tükUÐükõy@Kd¹­y5´æœ÷09 í"_µ”+AÇ÷×kUH ëëùS*§ g|äÉî©)è=wQ> Íwë Ga˜o¤ªD F%%÷Xë·`Ü¥^[}úLù>ü"í 3ÑÁ=£Ñm0ÿnî×L‡,^7_ ¨zˇp1ŒnÁªj”ãßtX÷u:B ›ƒEÍóF°¢¼øûX<ì¾÷º •Ì^û.D¢FÇ™ÎHr¿èxü¶’™‡çR!¹¦Ñ–?gÒ•Ÿ;"‹©e‡÷?F: –š.QDúžb}ÏLd<ûcþN2Í3Î)X«!Ë£L¢˜òÓȦdù)¬iÙw?6Ž{üCÎEÇŠÚ'ÈÃÚ$Oel‡|çfÆOŸÞCþ%~3bywÔ ü+“Šóc$ù5òQØÙCâøI•Þ}Ôa͈â"óŽúW#PÒ$åÓB JÇö8LÔ* ‡þµUéi”¿›~ Ë8þQìæ~{ƒJ©™5cµ=¨üÆ ÿO´ ª$$æ<ýq¶ÛÃjw$ªqMôPºR#þ¼ÇŽǼÊ|VvQ3M$Q³-jßC_-_wE±–±+ƨ÷Í&ˆV¬õ¯|‹˜§WAC³ cÃÛh¬is¸›ÇMuèn×tç¡Ù+»Ôo.¡ù§Í5õGÑ"çÛe§µI´ì¥_àñeGkZù'?ÌÑÆ{zÌëÛ¾Žh¡ípŠs#Ú«Y¸~®ÿ‚Ÿ äl¢“Hué¾r:V¸ì÷R+ ÿÊ/ÿËÿÎ,ÌÄ@¦Þ³¹›lhÊwéb°$²e[uWqÁòÛ- ¡¼/ЬæÈ>¿CµÏÇ*¸ŽAQåCþŒK:/ðº¬À¨gßãLòX)@–Gý­•CQä}ëÑ.%ü÷«|è¸i!Dúõž,&91ù €R«ÖÄú Í~5Díó<¤ÿî!Õ%熌Ž= Ë+Ôÿç×8· ÙEÈÍÏç{ºzòΘw¸–|úï¼Ò¿«R¤dCÁ"Ú¿Ö§ 0¤Š§HçÝ~öf¿~‰ Š'«ú{f¡˜s¬mÄ¡ŠçòÊ'^Ÿ€’ÑKÒÖcPšó'*û¥"”ëϼ£h€Š‡ù½f‹!PeÑ^"Awª·¾ž ’H‡:vµºêO¬ð—14®’~>Ilöà&*$YJ€æOWÍÈ2¯B«µiLî¥2h7] )ß¼ŽÖ@*ÝM“7"¶™¡Ïú|Í·gE0©®.w¦†deç5`ø/=A\ÑYMd{¼éô ÆŸò³Ä^†É4Ku&Ö˜î •¹Ös:¼"­°pƒôΫ˂°8!?á*ÙË®ë}±Ò°²">Ï”• kiÏ”ÆÍx`ãã§—¿Æ`k&(êäL ü{î“eîq ´Ý'óÜR‘p̽1€Æ‰S?é•j¡Ü©ˆÇO¹PAö§§]é]Tâí'X¹Ê¬3…)i~¨Â}+Üß[þ\Ö¤ÓÍ@µkææ)ïPÝD±}hþj¬z˜¯LC­¼?7xÃQ'æ×A•ã¨G}­4ù#ê§/lRß¹Œ†± :ÎhBvèeKuš±ØºþNLCseˆ•]­A‹S>ŸnΠe„äɹR´ê>mCY‚6RÿüžÀA1Ÿñ¦ÿ‡ÚÆJi¿ÐáÛ•œ ¤tbõI^$lÞ÷C"Ÿ¦²íÿËa’ßMݲ‹øÔòð˜Ù9t®”·B©C¤¢Ü¡&HþY¸/äsŒ2G•jú(B'ý´±ËÐ(üè%µOÔúkGÏý˜‡ÊXóÇg ÎBÙÓ¾PŸ(q»šú}©eŸƒ9—Í/Phº=ê´ø ¶Orê†Âélc×z(ÞzýDpÆ*ÔaV£„:=}]E¥hR'§aŠæ‚Ö-©Ão"oBg¡ç$¹ôF&ÅzÝ €6î¿,V{0Lmåó5ጚ¼©NSƒñƒ}ÅÏ£œa5÷È8~ÂÄOËà w`b$˜Ù6v&eL?0ÖÀd†ÄðùXB˜zÐé²´Óq6±Î'a¦óQ|ºÇÌq½_pÔôÜÏ«6F˯°Ì{ù-,~ ½f K¿ãüe‹`y‚"ÜéÃX=$ïôoÖ^¤Xã…eØ ’!ïc¡€Íç„Pt‘¶õßT¬Õ¿Gû‚"­J`ÏÝS+Û ÆÖÈ*+"‘BÕ²œ¿¹÷Ö\IYÓŸ“h¦"Y–›EÈ/H~óÆ÷W¤´ ú¢±´‚Ôv™ ñ$Hë+âÒµ©„ôNų÷-!,o‡ü.72Vš¬{éÛÈ»äšaxô/ò+^?Ó†ü+gj ÕP ò®åĆ" fŰä—ûùW±ñ`† …ÆŒë늈¨|ŽCÑ[7b´³Q¬pVÝèQüé+áG(©G_Oæk€R?¤n˜¹^Eá›B¢(ipž¾ån}"3´êCy‘óÏu¬CQ¾ß5îÄØ TÈ6UM»œŠŠ©×î+iÄ¢R…j£²V3Súµ«xx¯ž`gGÒ*¢° ª^d0¢ú‰HÛ“úãªoŠˆŠ}£EͰ{Ö‚×P[‹¥úû<ê²JÛ›lÇàqöšs¦Vh ¿åçÆ&F4Æó…(Ðxúýè£ U4má¸!®A„fãù¿¤¼¾¢ùŸãŨGh1½q„ý" Z޳ï-¸©¢U;qŒ¥ð$ZÿY2|Æí6ñB!VoñDXI9Ï¥.´õß¹O h‰vÌÛ»¶IhÿøSÀ⟛øðAY/ù®;Z¥&ªÑ>Œ¢ö>â»þH2çðÆójd'Ç©¡¥žxš‚þü•8§Tp*Š+ã;f¡ôº ]iÊ)(’}w°è9ä9Ò+V9¬ê×É·J÷ó­û®èåûðóÈÆhW³ü¬ËÑ)LÚï©w{šJƒ!¿w7ËQb?· Êž®¢ T¸½;ÞÙµ™DT~w AIžqã|4nÆ­_IÛS“[¡Ðª"ór}ë3´iÄ|[õ‡Ž;Ç“2÷=q7(¾Cº*¤{||=×bhAï…¨ÒÈ9è÷ÞÔ?-LŸn†ÌPPÂà[±>¢èŠ¢R/8 ÃM×¼lÖ'aTæƒ×Ó˜«/2ؤ†‰wFigÂÔ±Âaæ0“¶pqéê,Ì­¨4f¨ÃBœfß_"Xìù»¾i] Ë"U^¹5‘°òPX=#ê;¬±q4:šÁz[ÞžôvجâŒ?®éÛsÇx`—æ_dG2ˆ p "¡ ™‚•…4ùt8SßÎCâÜ'q\+rHÊJ#³¾‹d÷MYz2JœU³øOR”O[kÊI!Õ® sUׯµ+{æJ8Ò©ŽÕ¦I{#ýÓ<Ë:êdx(»Ã­³ÏëC®YãWÈ^/"XÙ†Ìß™wÔB¤¥EXÁW§ÙHú8”§z‘ýˆÄM‘í8丶šsõŽrÈeÃyäR£;à†<‰j_γû!ß©ëîž« Èÿ1íC";Å~þ½7pFÁWw3éYdðàƒ ù£c‰(äKUÝ,†Â+}ÌCyQ$vo…ú~ŠV_xá%(‰âzÒi(¡cM¯Óƒ’ŸdžÝwAéƒÇÅñr¢Ì'Þ{ÝBe(G%Ss3ì0ÊSŽKøÞHE‡›ÜÎ9}¨t0)ç¥i5*ß<,û±ý ÑÌ¥òe AUïÑ.£Þ4ĵ´°šg}¨QÅ;mP›Z1u¡ÇxeQç…ªôíCf¨÷™øl}ôê'¸dÒß BÃï¢SÄÐø÷y;ÅFb4&.M~Žæ”jt.z“hÁ61"þ(ZJdcÅÌ-´èJübVÖS/&X õÔÜy–çû¾X/›9E'þ€×|º+ÚæGúŽÓ¥¡]¹§JÇZ _ãI}&À€ÖY¥f¬l»ÈCÛfØNWë(÷Y”FZ¦­?mçÕAÍËߌ3T· Ìw¡jà,5k³lÔK•Cá%§G¦îvðÇò±U›¨2ü®uvñxŸ 9¹cÃ/ ˜á¯uBód_þªX2èÙ#æá.ëS+ñÊýØ'”Ýd‡â•È-›£P+ÎÄ{ ªCòk¥ ÏAÝgÏæôvhÐL}ÂòÒíË37æ¡és«„Ì 4û²{¬6Fÿ. +êu_ÿÿçu¥ÏBSõÅ~¯‹ÐtXéƒïöý>I-GN{ št½Ó/ÖBSÒÞ-Átmh¾x&D/6 Zô5m¨/@« u§ÙƒUh+\½•ÏÞ>ÄÐ¥lC¹,Ç =îGto2ôS „ɺå ©%¥÷ýJª¿ðÝ¿ŠFš¾æÞœU…qb±?G„`ÒU Ñlf”¤JjM)a^šíAòO:XT‘œ>”« Kë{– `å_RuÉþçoosnX6“×-§Ùaçä„*ÕÙ=ØCæØuºËH(È–ÍF(ˆD99sÞu!‰c…þš’‰žìÒØ9€ä´¿-KÒÏ"¥8(F’þCꀻõ“^HG'ô"8é6ÒHFžSKAFÉ3O­T©ÿuz[-²ø 1z\¾lb2 g}y#:ç0%rŽÓõµGŠKù²ÚYÈg“N›£øùÇdoy àᓆê|_ðà·–ãA Ú(¬]©[Ök†¢´bdŒ?Qœ0“>3%¾Š§˜¶Eék‘G<>7¡ìîýX¦c(寧%ZpƯ>ó"@¥X%ß«&J¨üêõÙ;…ᨒô‘d/ôíüp‹9ÕÞŒN Ï ~é7çÑB Å«ïÍüQóÛÅáû{û~xôñ9§¨³xÿ§¿“ êý9RxõŸÚuÜ{…†Þ)z~h|‘áê7šzH´ÜàG³ØŸ¿T—ªÐ<9`€9-Jèç~F åˆZÜç%´f,zmŒŸãM~N<ñÒ•‘K‘ m‡¦³ÃdbÐ^•Jç»ÑOtˆ‘!©È^A'z*J—ö>t{,¬ü½z¿—êIÿ_~X›þŠ* )ïxQ Éu'ÎQÿbd­û(y¼y –Vžf«ဦhÞ¥×tP.õ ¤õ¢3»4„¤w§ûj|Ž2üúû¦’ãy8ülêýiP †M­Ÿô½ =ý6© ¤1Ÿ*.‘S€Tµ{ÌõÂRѯCá ©_ÞŸ%ô£Þ¦¦Sûïk¹ÛÆ+øßq~ÙÍýŠûÊòÿ÷‰‡ïóí¨caôßyó{ûÞÏ9ŒÂŸ*%» S^(dá¡ü$]ýO¼°•P¶ï‰7·zÞ3@ÑËçcnE Øåö¿eY(1 9¬‘ ¥œÝ[G% ì­k%iÞ (ïe -Ô&‚ÊÿÓ— Z!xDÈJjµÍÓ·Jù¡^%»Ò!Ãh=ÈO$Û½¯Í “š›>Ë/ƒ–º66má}nÃê¸Ü¡£œ…2éÈè–kÏe½m ½‰7Ÿ2sƒ°N/O`€ÁZ·P/¤†áóF¸W£GÖ+õþü‚qŽGvSÃ$0©dF»ûŦiþ8ý+‡Ùöo§˜`A¾äÙUû°øþÀÀ³š°L)wfó=¬„Ò|ßÚë‡5¥~U~‰4Ø`lLЉòƒ-ãZ­‘•Øý×&•¥ {-GT,CeðÑ·“_󹑨h®æj%’RT¡ÇrU•j_ …>ÏTR½˜;±üƒi„l…"}¢bÀW2ú\}Wgq™ïÇWæàBÖr«GóáÈ¡ÉúB—O¹i¼²þ†Ñ"Ÿ˜YŒwº7 Ô=%±ÔŃ¢W 2¹~¢°Üy¾œ7в_o eD Á­Uoù<”f:µ¦a£‹r‡ÆQ‹ú)*ШPè~GŽÖG¿·ðÐFûÖè/<¼~_àŽ‹Î ÕÓDµgoü´Ä=Qݱò¤í,#j2°†g¯´¡Ö#C=êtøh<_ÎD½ Þ÷–¬üh@,\Ú!m‚F²Ÿ8#lÐÄ®¦ºÇ_ÍTúo ¢¹åÔç¾Cáhq«@TY€-¿]HLðiB«‰××’ý¡â`äw<ñÈRÈì^ôÿp¨‘QGÁ_*hËëýЉ9t[°¾ŸòML5Ùÿ_“œ…zÔÄðé—â®­|#´'á+<ÀU6]×Í.Ú Ùä›s¬ëç`ø³='‘Ò´we“Ÿü® § x(X Ævñ·WÜPñŸyø ”&ïeø ·@qNm!Ñ3Þ¹ì°zê¸Lî~^&¯\eM„‚Û¯hت¡˜Âz1·ÏÊÏžŽyWo µ¹>Æ6-ÐÄQ$ü¢=ZWŽËn‡kCgï`Óï¾kÐÛö¼ÉÈoþM%'ݵ†aÃã¨HfMÝ(hœZ‡ñè“¶ÕÿaâÏaïãò0)üòjöO˜<Ó ² “Wt*ج`²wÇïŠEL}¶¸è¿â Óe¦JzMw`fû>µÿ{I˜;"qöyj&Ìÿ Œ‘ãu…¬¼Ê§±°øƒW˜ìG,•”Ð3‚åyOî3“T°ªB­ñÖÞ›†˜§Àm…Û£§G`ó#;éXضú5÷ÂÕ þÁ«~ÕözØs¬xXQ€/%™VˆËVfŠ5 ‰¯„’Ø ÉX¤©nf’Ý‹—±Ú@r™âBw¤$^‰ð©'GªÍŠÜûWÞ"-­•©÷lÒó0 ;Ï"ƒVðK^U)d¼ÞÅ‘UŠLß ÆÉª#óø:É‹«¦È*Y^øžZÙ|üÃ^±!{»Ê×°¬+È©ÆEhº‡\i,ncǽGäû¢¡âäÞÒ¸ç8üÔ ¶¯&#ƒÂàj–< | Soâ @Áè’*«Z<˜±õ¦ëêUš¿žÿ~i EtI¯¶õÔ£h‚6ñ• Û`ÖŸöÞçL±ì"sXJú^Ž|àêŽRcŸ{k=mQÆFýUñ•ç(›“?×Ü€rÁÛi:´(/!Ê*ˆI(ßòÊ EˆÞwy{N â£ ×§ú:mò •ùùý)¤ËððU[k5™¸9õ‰æ ªÊ¡ÚÙßÜ)å%b”¨þƒ›öC jJµåª?ëA­2y{šÏ€:ÁàÓýøê]Ê!ˆŽMCý³ÙîŽæ±hHsùåŸþ`4ÊÈ Üž:…&§ªb36¡ÍjÆÈ¿B4›{_:ÜÁˆæõt±¹ nh‘ /Àûå5Zz[4êi¡(vp¹K¡5ñ!QöZ´®f-½k6QŠág8ð„—@Ùi ´µ!ø™ÛŽ?dhzŒBË…VoÓažÖ¿áýΉýHÎí,ÀÀ›«ß]C ¹9ó!¾yuvÕ¨î§îóóT´î{ ”\ð =OÞ …fdÍtÓû¹åÃÕÓâ¼ß?V;?€ŸNºçØvå ó°¸Å oÈLj$J~ Y›LV…G@ž¦ô…ÑÈ»P”ÎeµNþ Êé´ÛÓÙ› úR"0?‚úçgZœ” aùCU¢à~½ A[ÅÐB@öNbò´¾wõ£“%‚v.§‹f%ªûž8Ö“ ?l(tÍ¡«“Ê‚ßãôÍA²3ýЛ3½¾üÐú½økòc¯Ã@²a9µí ¦Zrs·]†¡6A‰ d&Ÿ£:éÎCÁ0è¾ÿãLÞ´7ýùB¦ïh‡«ÒÁlò“N l0?'0h¶‹" t5²I°ä¦ñݶ{–3w.XÞІU‹Òcæ°ÖsÂ×ht?¯²š¿VïÀVV®|¼Î.ìTE'ªÝe!¢)Æ5$÷,¿Äb‚„)[tƒŽH4à4.îÆ‰$bgœ˜‰‘ô1Q¯.Ó~~˜4KÇäŸnÒÝyRŠ”fœ…&"µHMÏ%²NTˆ4'*íUî{£nQ5ÒWÛ[u«­ CEøß÷SÛÈø×G÷Þä)dšŠX"`ÑDVi“|©=d5(fÜlF¶À2Q{¾dωú4GŽœÔÔ­ÈåKÆßóô×¾Ç =>“ ÈÛÃYñ†%ù/œoç£÷B«òyn&(èax6";Z§FRfÜD!m~ñs3~(l~ÅÍÝEN¯µ4šCÑ`P¯Ì C±|ÅßðúJÓ*M“ÜAI‡­HOB-”jή1AË?^ÝB‰([oP3Qi‹r Ö¦þ^¨ s5)é•2* 9{‹½•uù"æ ÊÒ”{g¨%ªv†V„‘#ÚøvË5B ç|Ê~/FÔbtËìüÙ‡Ú­Ÿ²G'²P7‹ï\€ ÷{|ò³:mˆ£rQdÝDòm½v¨~F|„otRùÃXüͱà^˜äHû½ãu¦«šì©ÁÜu³ËÏsaáiÿTb,åªýJ+ Tކ«À:±Mu5ûmØô©oCoƒ^ÙÃgÒOƒÇ´HH×בZ„DŸ~zGZ#‰¶›ÎÜ¢ ’þSr¯š-ÆÝ„WãÏ#Å‚£Á[ë ¤>®MµaH‚´½ŠvŠòHÿ×çòCÊdÔŽëŽtF¦ö†]ŽÈr‰^P™õ²qS¼÷º7Šì“æ¡D¬ÈÙS¿®M:ÈyMŸëʹfé … }Âhžs.tÅ~-joò÷ÒFËé 'ŠsŠÑší‘@Bð}´1#}y &lßû\¬¾Lï÷Rq â0Y´?Z*–V]ŠÑíb­7£Ñ‰ŽhŽ?vC×g?/3í÷ÒC[]ÓÿË«¼/` ]l²ÚÜ67³V:ß¶E–áIŸò¢FX2=(mù3lÿþµZ•a‚2o¦vm›óP˜Âñ›¾~wð„‡Ã/‰î ³ðóÈì=3'gøÑ&«óªÒe~d$í{¡ßÍ+ÑrPÁæE]8$—±^œ¢…”4[•ü{’ÿÙwuáÍ/øñç[®mÊÇÉ®-ØZÔÿ9qŽ!ì'á·¢ûÆìÿΛOÉ𯙂 òWÝ?ïÙᅢóD’+ûžÈ8t ýÉ$;”¡ˆZuuä‡?µ–0™€âŒ±¢s)¡ÄwÓAePp]imð¼bå nê²¾PÑ%J\óª.œ9[r7 j<ÏDHŒ‹CóðL§ã+øû±ÛâÆÁ,hä %¾uÚâ—<+8 E×ég´»Zä²-í–š]Ðõ´l÷õzôì™1€þÊ6Ýœ†d<ÿŽTíi ?ò׆‘ª²¾ò}n FÁÄ/£Þ©CÕ0ÕnýäQÌj0éÇ™Ã|/±öû§ò°hÚÛ‘÷ç,¥Ÿ›å|¹ +‹ç†ôÀjެ–©6¬:&ßaÍTÆÙó;]°s4`œP öøâ?Èí!¡ÄôlãÅ$šR5åmz€$5ßžNÞØE²¢þ³Ñ—¼uføýJ'Rq¦~x›†4iI5^çK‘þTϘ8ÁYd +â¼[ÌB3¡n'Õîæ]ªâ=d/£á~ê4rùN×%y!>cmOäß(iºô.ÿÚôr¹¢Pc—5s4Šr‰”HFñüs Zw‡Pêö¨äÌ Õá•õ7P~*7VñE-*•)xÓyâ¡®Œ«zœ‹x¸éÉ|à½Kxô¹Ááz ¨ö*ïÿw–AYva»¦Cº»Cº‘¾.º» QET@EEEP”é.iPéîA6ï7óú~3{ÿß¿Ö Ã³nfXÇ}žÇÜ÷³Ö¥D}Ô>~`¬i*o·–~»¸M‚útÁ{jI©Ð~¿AšKËU†¼DS{—è7}2h~ãe§þ¸%Z~ô"Q¼„Ö¼U^Ú¢ÍéÇï£ÎÚ£í}âÔé:C´+tŸ'@ûÕê{Öš®è¨Dód“n“¾a»õ‡hÉyäàEtMä=L¾@nŒ ÆöÏ>¢û-'¢¾¼ÿ›ÃT­ÖÓ™Ãx·1>„NB]s†­uQìBZñu w$•¡X|c¼ Cû8N•Îß…Î$onç(ø¶Í’êÔ)æóDÂá`P$7J/é ß–àŠ ‡1†ýŸCD2`‚½žÄ3&µ$go™jÂ䣼ù–e˜|P%w†&Ó5feÞôÁEÇYb'˜ªü¦°ÚÓãFUò0Ës9ТÁæsúª¹7`¾FZ†ZŸJœç>ɪÀbAà36:XªŸldvé‚åµÎw'•nªÞb ´ì5XKZ(>´B ë\ž%ŠäÇ`ã£Lc—2lzü`‘^³?ºfªƒ.ßaÇŽ·¾ÿ\4È»‹.f!ñ˜ßÀ5$¶Ú® Ï9ƒ$ÙWhnŠîò ™¹ÚòËÉG³]þEÊ´Cá_Þs#UÜÏC¢2Hó*L‹q(éJ.ž¸Ûцô¿­%7µ‘ïˆ&Áüod2ï¿% шÌa^ ~ý'‘¥L­ƒŸiÙ“.€&Œð¬#Ý"C¡iÙµo(¼þL“WsEÝú@Ó±{;½Ÿv£¸¬†ÿ'f”ð½¡O|_%Ó™\Ci.>Cãm”‰ ,òÌ‘@١ᨾÇ?Pî‘£ Ô¤£¼ôÈÓñF”¯­\"9xBÍ/hžDE/n qT:™ìú•·NLyNr J‰õ»ô´“¨VVø˜»Pcå„Ly3‚eÆè+¡»|2¬Ñ\´DF‰xÉ?¨gaôÜúùÔ_³zæÃ&†eíE–kÔhœhiÅŒ¦4÷T]ž¡Ù¥Þ“3×Ñ|6¨>Ó¡¥«‡M:Z)¹dKª uú^j·Àr´9j¢k>1¶,‰­|ÐöëÙÃoÓ†ÐÎË ¼—à5ÚS=Õƒö™¾ ÙÐÁG&‘öÄ: \ ÊQÁˆì—o›šö£Aºwè& ³—Ìew#ã‘s–ÉÐ7"3õ½Z¿:Ç{ê?Ä8ݰ†Jj»y!¬Pvð9 Í!G(%)¬3†||—;\kýùü…wìzi™aGq§¾(íPµ‰†/‚ì¯<õ‡¬´w Óû!?ùß)2Z(y}šs½åTäÍ™´85AMËË‹<· !ÿXX½•;|û)md©ºÛKüþ ­Õ}¯Ýº¯@»í5×@ŸpèȺ·Òr º(s{˜Ï™B·DÐAÞjZøŽçžð^Lƒ÷Ž9v;ÂÏ­¾äk“ЛdúF¡úÃzH‰FÃ@!µÔúìküÁa±ý؆…ÏH6ü8 #iul9c0v4Ôxˆâ LØDÄj”ѣ§2dò0sTr–¼æ‘g1Á©˜½ ”añUÞàÚŒ*, µ=\¶ê‹Û>;!<°:¸t³€ã&üJŒúhQ˜Q7|׬û`ó†Ì/Éc“ð'ær!Õü[ØIÊ΢JïA‚a®ãM$"“‘Ü×NŒÄÈ•º^GŽ$w³Óvï—÷% î!y0Qmö´R2q²|¼.Š{¾º>&|4ÔğÂבV]|tÒŸéi®|º¯(H3ft²‘VÔ{˜r™$ØÇË^:"³³²ï³*²D¿J§¨EÖ:®;ú´ÈNît©Õ£ 9özùm}FÎf ’Vjä>áç庼Neb_•_!¿²} Å Ôß¿ ¨…‚º\<.D($º {U…i”"¶SN£óáó:±(ÊÓ–ø”uîÕe˜ç±C±‹†2µ. xIõþh©ý(ÉZõʇK¥n¯ju_C2wïŸÂÑ(®J0tAå:9® ÝTEY?Íß'ï¡âÔ–ƒã¸…üy'ª¾á,SÙáF¨F-]u„l#‘#¨½â*Èåº×6û.XÜE}r]Ûx4¨–è":xÌçÝ?5l¢qsÌÅ}Å–hêÜS*h6;=}Ñp-ÂßÍ”m£‡æðŸ€|´6Ï~M¯lŠÖÛ½IÁh“ô9åBÜa´5ìZ¾öµm'‚Y{kÐî~cdBÛ:Ú럞¾Œö[ïbìob¸K=Yd ;Ú7´‹ñ"ç&¥`03¬/Ïóø.VæÐ–¢1;úÚÓÕ8— ã·òþ­äZè:á~ÚξC‡%ßeøé5”ãsqzE«³êšaÀ`£†2 †ÄµU§Ûs`DർºxŒ.æ¶ïò›eæ¸7–¦=.ÜK; sVǸeOzÃBiû‹gá°Ü\w†f V®´$Íï|‚µj;”´H‡ £×¬vCé°EeÚü¡$ ¶×ô«Ÿ}B‚?I¿N ÑÍÁ&»˜$Ϊx8sIÛ(–múƒ‘ümlß“ãH™sñàÓ¤¤¦ßcr=õÒ~ÐåëÏžEúüZý¤®—ȨÈmóút2•èDÝø FÂ}µ»~`«À„†dV2ÔèÑh‚À¤^AMêó|;jfe‚‘hQ1fÄA‚V9o.OŒ£uñ‰Úõ¡hSÉÏFñÊm;úãèÑné Õ£ýlèÀmÊ‘é޶Ac±ç¶Ñ)Æ%ý.gÉnŸôŽ¿8Y„.j—©²ÛÐ5áªù°Š1ºÑüné?É´ë‡ÞÙ,êÿœo!ºéxçùa•úÏŸJsÃjçØ Z(4Ÿí ˜CæßŠuúi©°ø>çÍqùhñÜ/ñ«š *Šß}Ñþ%üÃ8þ¯Õžº?9áâ|ú¦ùò9Å©%øìyû^Ë€ ¤4ÿÕá|²Õ8¶ß³ >R Å:5‚Ô2‹Óºdøøñn]_‚-¤«¤ŒlõGüó>êÚÏ`Ë¿óäà}-˜> y¬R÷"ÞwCþÓÓ|òÎe¯[¸ÏÀ='[¾ Ä.pê>ýç=‚šÒ-»]OàgT·‡âM±¼T¥L(¹>Az¸·J˜tÄ>õCçøúSÎX(ëe-Šä€ UãÌP‘súùÄ2Tó4%)‹GBÍk½i¯·´PtÑu4Ü gp"í¦| ëÓ@h~_9úy_´Š<ÒKjöƒ¶¢áæ…1èx[eV-â ]q1o~›­Á÷™L=–³•Ðcÿɼ%úYÅÇìTa¬å Y_" %TÅû\‡‘ý¹ú#W;`LÎÕèÑay˜Ð4Ií¨Ôƒ)7Æ™¯L`¦˜f²'íÌ;^ñu<*‹Ä_š§(aÉùU FÔnOÎí©} «ZWLï÷wÃÚ†È-Ôuù“óY;GXQð¼^= ¾º~®Es}Ϻn„âNÒô»9/eÃ[&ÇóeKö÷ä¡|–Ð *VföÈÅ¡ráãÚ‡Ö’¨ò1öQä]TY͸fìZqƒÝ¢|쨴xRÂuOйê«þh¿²n‚†ÆªwOùÞAãÝùSãühštïÀ©{´hÞÃOuJÕ­ÖU§ÇÐúʬà— ÚÜÓý¨žÅ޶ï^9ƒvßÌÞï{0‡D–Ö·ÑQ>ìúÓtº“+0hzþ?5G®þÉû‰®¯Îæ™0“†–FEÇû¦kÐ1q•¿üÊü`ÿ¦{ðÖ=è³= ¿ug-Â}4Âp]ÌrÈ…  ÈO’ðô…‰‡ž7´Ê«`²ÿpƒ×˜"5Wé…ÉÉNcnWz˜\T`2˜9 SŠ/2/fÁÔ˜9sÁ/}˜¡ÔvæÌy³x¾‡k+æ ¶nÌ÷9Lk……ÚUæ&¢X¬tæ—½t–Ú©WÈOzÁò¶–Ò¹«°jæÜÃ\cki÷WΩÆÂúÞß\½Wã`£À`×Lk`ÓwS£ö– ü1$|p£wvLÎÑýªC‚¢¦zÁog‘p&0¥k¸‰¥4ß^÷½$áGêëãDt§×èl\<’Çw~9Z”.­çf®Ë"•Ú”„eŸ ¥–ÝÞ9~'VÜâ ¬=0Èhª†u/ëDc×bø­]lŸF[Ê]g¥œa[âv¯– 7°i>šþ %·OéëÍ"‘CËVïj$Žñõcoù€$Óûü —H‘ÌñîµEã5$ïä8N=x)}ãI©Åy‘Š6äÄú—|¤® »ÁÀ¥Œ´¯Ü©?{#½ñÅØSÈ`Þ¡'q=í½!é‡LOµ"´7¹óÚ¹z¶deüQ⣇l6‘nD-×=i2—Äð0rŠXLœOðD®n)+÷?EÈÓvìæ°ÃsäÛÉa$z‚œG¼o—K¡ ûXÓ›-?\W3·· G¡N㱟Ã(ÜÌ¿ÿwŠüÔ èAÑ?£ §§­QLÅùµÃ7¹¢l¿%Z¤x*ïæ¤¶Y„HJçn;ÿ>‹² ¢Ü#&rŸS~=ØBïFcãqTlüR®“²ŠÊà ‡õø~£ªÏã€1Ô`+™+œÈF`e ও@훦eù¨³ú‚uT õø5[]Cýn¸¯›„†|ÙŠV²Ñè§/ƒÑ¿}|]ïŠ š,{¾xø ͰĎüÀ#4¿qæÃ‘ˆ^´(ðKöùr-œ™ùþâ#ZK?ø#ÜÖ ¹ùí¤h½úÚ°»Òm.=b ÕE[òÁ'³¹Chû¤¢ŒÑ¡í>“>‹G{§mY9äXÔ;Z£  «û¬ï?­SCýNt¯¡fåýpÝÎa(/‹ž/‚’/÷åb(Òø£»Í¼õÏó„ζ iÈË™“ñOüç¹Ã•{2 »>˜VÐ}¾2—÷ó©Þq†,Ác))[es2)Q»Fa×+§sjL¸ÿg_|ÜÇ•ýK’ ­qP;ý­òÎçjhÔNùHͬáó†î Т²Guó´:ÅT¾Žñ‡¶Ý^ÝDºÿü°ÓKçQ•´I>[Ï‚Vƒi¿©Ô]n=ÊØ‰©w?Oó´JZZ,ã¯u怖¢Ë_öä@kx–Ó¡¤ÇÐæ’,­iíN¿' Äåv¹¦£†g¯¡‹Ø<êÝóÐ]Î<ŸÊ&?UN ÊhAï¡÷G.üù |N‹þ §`°3eçÕ;0\Ë;ó̘F';÷ ÚÀ„õè‚mÁc˜Zå½pÕ# fî/˜1ÀBx^­}” ,á1àÒ€í¦¤SB°vÛêKTØl0Ÿhë †Í-ÎÍY:Ønéù¾l‰Ó'¸ÝÌ(€VŒÿ¾’’멎#é»Þ+ß[T<ü@ã­“H{áÙêEF˜êÔ«ÖÈ”Á³uÄã"²ØøéÔ{"qñ“ ÈÞNL*+‹œ5b;;*åÈ=¨}ð™Dòñý¹žÿZùË,æ<ü†‚”ß#,E¡ÐIÿ—Ù›](ü«?œòPôSbH±YŠ= œ÷!D‰×†)m(5Xß²¢‰²^µj¾(ÏÖ»JÏ‚ Éi 4mZ¨ôå}«î‹ ÜÎÙ˪—ÆmYQ} B»„5—’ˆoœ¾xIÄ•[¥ u¸|ø[¸kQ÷ÑDse”Ì®yWs[¡!gŒ©è&m ßö¸•‚&S¾)ßß£Ùø„ë³â»h1©Çœ-­zƒG>Ìš¡ucmËzfÚ4w½æÿˆ¶ý†[/Ÿ£ÝFyñÑõè @ÍPê’…Žösêâ‚§vý°s§ún :$Ò.™£‹ªò\AÕ ºÆ‹û/¬-¢õRÖs5it¿Õr=ôŸó-Øf”Êþ/?,Û¹R^‹¡TG—JFúÐüÁ-!k¿ëÈLÂQ– ‹$æaŸ ¹ÔYî·'TP*ø JÌìæ–sJ¯OäßM=k¿ýÏ÷ëfyÞB&õí’°pȘŸ™’üé’c/‚þ¼„Þ뮺ƒ ðaîÁnîmøP|}Œéƒ¤¦TØÉ½i~·Úq?tdOr}’Ïßêp6¼…{$ÄFl#õÇçc”‘.Þ] PfBÊ©2Û ÓÑ`º¢Yud¹òõˆÞ“>d«û­µwh9:ÔÎ"ò8Oîÿí{ùíÓ]µ¥BPOzV–³…™žÏNpÌ ¨=cÔ[‚A[Ò5îc•BÉù?ïö†ª¢ìñÏ%åE(ŸTÜNsß%Þ˜–pCåçúoA•È7–×ºß+Éã9ÔzANÊ7ƒÚ‘7VóÇÈQ÷0»!½'ê{©=ß7†WÙ2ûD6Ð8Ç}Ûê%š®UˆiàZ(¼ÐŽ» he÷Ôr<7Z¿ ¤yÿhmRçÒ'VzѶ4*׉ûÚ iŠ…ðn¢½dÍ£¬Wèhúö×iuYtŠö}ùþÓáÿ8T¿8þ˜Q ]Ÿ7ÒÞRC7Úâ¡\åHt¿}AQá×…ÿ‡¤ã]ñ—U…*ñ:>k{û,O… ]œfèü0{†¤`†úÒäö…Õ6A»z%míahà9û§y ªG×YRT^CEåí×Çï ,ˆNÁQtnw½÷.ú+ž€"êUÊ'Oi¡€kŽ–á äèûU¬Ã y­Ñ$‡i¡è¸ ’úˆCùÿ/ a:PÏ(Êúì«Ë9E=:†â9öžÿ?öüVòL€¾hƒíüað¬½Éé5g¡Ü©ò?æcí\O:Ÿ Ã$Í ×yÑ)˜Ò`XSçøß=ð= SÚ‹$z¼0¥ ¦æ¦´Sî§­®N…i²š‰ïëY0#©:E"Á ³ô¦¤†aî~k£ü¯0˜_¹nRqï6,´ÝWŸí<‹M«<åİÔ#Çù;tVHß”„ԇêm‰ÞÖAXËS¼,ëJÏGlÔÀFíý‡u]aó·×^*øc’3#á¹;ØhOÛ<‡~ª¼÷!aÑDSS\¼MÈó IÍoÑR™#i9i‡Æ’Ûý¢Òxƒ”d:GE<ôpÏ¢Ìã­û‘ººÝ´¾i+O°‘V!ýÑag wd8wé꧇‘1š¢Ùaj™rTï ¤"ó<ß¶˜1²j ¶«° [ŠoƒArÐܸZÓØƒœ÷ö¯Ý¼°ŽÜl¹†¢ê'±dfo¹òÉ>¡G‰CÈè©è‹_Ô(`œG~ç† :»Ø_1½€BœýIx EöL©ñ·P¢èóG<—j¬PL1MjÑ‚ Ånz_âå@‰n’åÁ§L(Åñk›­¥¯3²i?¥G™‰x†CW”ã|r4…ö<Ê}¦ú¤ Šòʉ¹Û´(Ÿ)®Þㆠf&_jÎÇ "sA¼g:*±’/«žóAå,ŠRzRT¹Úf}Ø+ÕÎÌ2+ù£Fp-åߨu7RâAÌEļ͛#O¸P‡¦Õ“¨›uÝE;ïK ^çéžP…R48A­–_‰Fs¼ëhLÓú€a M¨¹œìêє쇤½7šjXò(УٺøÖCh>ñ¡X“-š–)mê¡å»[úO_¡Õù…Í«áñhÍïýçȱX´¾ÖH‘÷¡­'ºsý~û¡µ±9ï5´ÉÓ»J˜¶†—bž›¡M¦“ðÃâ"HÈqꛀísV£8?,«Ïö§BóH•ë°Ô!¨‘*q˜Œ†2‡¾/÷…J¬Iö8²›ïæÓÃÒµôÈ Wïšý ™–ŸBŒ”Kv½0Blò£dÈäm?úVnµóy ð…»­02Ae·ŸæE†ÊÀ×=jë5 PF§#ÃÐQ•qQK²±ŒPKíߟTŸõålòiºüÐD×ðe­ šî¦—gïr[TÉ ?£ um¦KsvÚ…ì®3»ïæR }Ë\$tR[pƉA—jmøÅ)}è« Nüä ?ȳfOÚEÃÏ6Ÿ-;z^èMÙJðRe…þò«VæÇzaÓx˜Ši†¢[G¢½ÞˆbÖàÌd#]P;’¸ÆûÔ«= „É|®Åñ0=²f~LUæHÒ?…^l…î+EênX„ÐqERXºö‡àõÇlXžK~ª«W¶³íöMÁ/U®Þ§Þ/a}ËÓáòŠ0üî’—±z¿¶Ò’žüz¶PË MÓ›/Hèi^ODOyÃäÖ‘¸§—Àrø"’Ês‚dò]$‹c/ñÛµ` ¶ž@ÖBK¤|Ÿ¾A?p© «V™ ÍÞÏErÖkHÇÐ7ËÒg‹ô·CÞ¾1#D†÷ÜmFÙW‘±ÇàMØÐdæ>¢ýV4Y\©,G]@Ögy ¨§ƒlK­4Ï¢Y‘ãâéáLäÒlQp}qy¬è§)ÏÆ#ß#I²Íå“È?·£ñâ07 Ìg†I<ï@ÁÚÑ Ž^záf !ê„Â&,ã×Q$ö­ÃÞwU(š—gVuí3î]9Î2Шƒâšj¯3¢D”ûgËI”\ym{©u¥?7ý)ì‚2kŒ¾ÝL(÷ʬ£>¾ˆº=BTÌ :-HŒÊŸ»ë…UyQ•øÒ…•±ÝÜ|ùRå Í?çE©Å¢6ß“c ¾´¨sTB>+xu_´:IM¢Þ¤­á>¡qÔ_ºLiÁí«/÷O7¢ášµ–qž/­·³U X£ñzW^Íš$š,¦4& i¯Eú–2šå¸ó¹­x¡yØj %ëZ„m·û£%iôÜæþn´¬ËS}2ËŠVŸ?éŠFk…ùf~# ×»àvŠo7wGÔ‡TÙÇV˜p•Èôð 9NhÕfX®ë˜„æâM7(—õßòd‡‡ævµ¯ðõ€Ï{ž®( ±Omí—€¼=Gy8©ŸCŽù—;Û™å¦ùp?dòŰ{œv€LnwöÝu–}ÍÁ^ÉÜ †‡ÏŒ@éÛg%] ò…²’êK¨Íî(wîЂFvј?i4ÐLÞ”ÿamZ$õ<˜ÞGB«eñq~ªxh;eCL6¥ôŸ6F=K+&ƒ6Öà¾Þ+ÐÊQb½]ì-N'…^¥ºAó¡c¤Q м%Ñ{§uZNþ᦭ê"†~›ƒÐF¥¼'pmzw¾±£sÐ!?Qt5::ãàAÿ‘‡ÐmdiGá‡G”ØþÁÐÓ)E¬)7ýIn­Š¼o`ð:K•ÓÀaöövËÛFoÞÍÿâOã3Wo]î‚©[Gé¾Ãìéü/{¿Â‚ÙÃ"ÊàNX""œ¢» +„k¡.}q°fe¾w6—ÖÍ7ì~ÃföÇ#æ¾°ý1pFu7O¿+i]ý„D.ꡳwjxÔ¹´wè&’†|¦>X¶ˆäÚçö\ºì”Ù÷ ¦‘êÅ㊄SÙHkäñQµÔ éï{/ ŸŽAFz­ì{DÈôêeñ’àdÑ{jó˜¶ Y=iÞӈ쵮­‡4‘ókÕÖƒhänû9[,(Ž|4¥©ô§S?ÛS8õ üN]$¾†B.‰EÜ)î(ܧ™[Ðу¢w·Ô(µQ̯*Fº”%ÎK¥ž<´…Riç©íQVf¹ûêÂ7”Û>&ç.ƒ ÷ÙC†’QÉOQí™ÝCÜw³8ü¬‰)ª–ø+”Íë úPkøÕ @Í {šÃÊÇýPrõ‰£Óá_Éɨ{o¿‹·ÚCÔ×\sÕ$»…†\¯‡™ÇИ¡ªayXM)9çóÍÉМ&ÙKöÀC´dÜv¹ÇJÖ„ |ë]hÝ#ò#ê¥*ÚôÜ“–—¡DÛ‰OžòéæhO°pFýI:ˆ4Q6é7¡£“×¼ÆtzdgEHt ûC%þD—}s-¥ú èú´5f;û)ºQMŹ‘ïú¡¾árÍ?ç[PŒ {´þ¯^Z\!ô¡ C)M…y|C³%½æZo.d­9¦}Û Üî=•fA§}[¼Pn[ZxŸEf·oNhPSCÞä’€ºïsÈÖ,XþE_NqÓ_xÄgõM‰ÎóBš¡Ä ³Çz)ËBê )¤´ëq TdCJa¡@Ïíøðvsž¿í9|:V´¢Ûglâ§~\ü;O¶ˆbé™\ÈÙl k¨<ŸôÞßÿ^w×ùCéŸCAПpó},Pü=X'a×¥}eExáëêÁË/äõ¡h&ظîëÿœ‡aç %SÞ[ËW¡ô…Í¥ßPÖ%3™ñ*ÜÚ/ßÊ$ž±û£ÑPÍH°ôrc jV^éXB]‹‹­óqh˜Û´Hy²ß‚ÏÛ›,A‹˜ÓÝ0´¦-ÆÜJH†ö¨ºÓ³VÐy‚YËØ]ºßE70hÀíO±‡mç¡÷£ã:[À€Ä…hÀ`«––®´. ]( Õ]‚Q#eu‡q}=ÓÔe˜<ÀÎÅ- ÓÄŽD{`Îä|V3,÷(ÕF‹oÃ.‡<‡¥±ø±Â¡°bÈ·Ÿ8fVû£ýÚù{áWëØä‹üVÝã7$Çê‹ø#×X³àHX&üCôÃߣlŸ‘¬FÒ³ü†^¶w‘y;2 ©¥tŸïœ@:ÊtsRCdè“'Üy‹Lå-3ŠóÈÂ8°ß#ÌÙ,êÏK G—ø­šDä.:ý¤ù'ò%¿x8†ÕìBΜ¾(”ªnö• Eæ 0ör0ŠE¥þàeƒ’w{îòŽô¡,«½”ôK”¿¦ë÷ÛXoÇÈ&å´£òU·‹’ù/PåÐÇm’˜fTw1O¾dŒZ‰R ”Ì¥¨ý„ó²Üpê^)¿]ÁqõoÇí«@ܵŸæƒÞh¼uT7*—ÍŒÖWÈšÐâ&ïÙòÊhõN8yÓÚ­«~¯¡/G›º§™O”^£í÷ ®l@»U©·•(ÑÇmT5@mÑÏí{ :=äá `´þCUKßçè/|¨Þ>ݨ[d§ˆÎîrøè•q÷ÿƒÃžŸ ÜlxOWü¨¹ÿt¢{q~6ûdùag­€;ú’ülþ4D„âÐFè­ëéu_ÞüE>UʼnËûÙ› ¼ÉE÷;õ(-,&7nYƒ:5ÂA½(ü¦!Ûp­òò§³ƒÜÈ 'Kÿ͵QÈYàïô¼…‘iÊfÛ»ùz–6ŸéÔ]ï5Íꇖ«I÷ŠþÙ??—áö©`øaWø™|_*ô5‰ÞOŸÁA1JÞO0bâ絿Æ•ƒo…uôÂdäÁŒô[Ô0•Oigý¦ÆìI–’wÇXªâI)˜:Ãëx9¦nêN…³8Áô^ƒÂ_¿E`ÆBþÒl| Ì^óûë&sižúGWÁ©Ót³ËkXè•>Hó~»p9ëÃÒˆÎì£ °B³Ç<‚V]ö~Þåu­¬ÝýiégXבjSX‡®3üÔ¿æ`3Të&AÆü±8iÐ4¼ ;û„cEC˜ ‘2‰óÖ~½/Z™Þ…D#¹'/77 ‰¨¢±õå3H.=UاäTl‡èÜÛ"#ÝÉ]Ý÷\du£M˜GjWjÒ±ßÝ|›4qå)AzÞÔù³ª'‘AìuÛ|2Ê'œñ7ÛÍ9›7[âõÈ|ã!cªÄ:²´9Ͱïú¡r¤ã¾ºIdgwÁóõ%ä”å ö½)„\_Òp²òØ8_t5EÞÁ*m']zäÇ£Z×î €XõʹwÙ(¨ÝÅQ°ø…¢*YäÂÇQ„™!ç5Šr:~.½ˆb^—£¹P¼=ÝŸÀ‘%i+ꛟ‡£”ùhÙ¥ín”ν}¥ðí:ÊrÑи³‚r²ªF§¶P®ÄÛYrÍåÕxÄMÈ¢ü{— rŽfT3¶¹Ô„ sœýk>¨8™û’õÚ'T¾ï(åÃÅ‚*™AÓÕlnÅÞ›¡@ í_DÝt¢¨%sÆ‚Ê|QrܹÖǵ?‰êÌ„¡Îè­ã·. žaØš’[¨_¶÷—Ç‚ Â2-chD³(*ÝxÚÕ—«—æÑøá-羟¯ÑÄâƒÁ]}m4%n ?…¦©?Ü".E3+!³454—$}ç€æç™ã8)Ð|{^ÉNS-®ñ~.=—–d|·££edжWZíqڹ휊‘„>]¿ÂõÑz5Ì=ÅýòË„¿q¢ô‡Í š7)å¡›„©Àë!|kŸ]Pô~U'Ý5n)@ÉïD•àÕ6(º{’aeôä½9ÎÿðM,ds\bMXúºëu„¬W!Cp‰é‡æ¤KNéWÜít½-ǃÌÑvLéÓaÈjТ²,‚ü™‘ï[®ÎP’7õR¹Êÿ\“Q}~ª•ï2Qþ€Ú*¢Cš,P_ÓË1ýKš ñÔ¬éh]Nd¡‡–G§l©Bk:¿[Á°´=öùfí¹ÞìïŽ@Ç çæÇ  ³2û³2}.t«Ž¿±ž‡ï›Ÿ¼¯ÂÏŠ“ÅœÞ!Ð[¿Wl…ÂX‹Rä’a0ÔÉtM•†9š¥ŸîÙå@P~5…ÆÞßq'?ü&Æ>&M;SQü¹6R`æI`ŸˆL2ÌeP½V·…¿ýô­ýI°øB¹!Ð ––ʳH³aåDN~Ç -¬ÑŒ³Îœ¾ ¿r6HÞªéÂÆ…[ã§o¦Ò>é—zÏak‡ð­ñN:lÏkÍ]E‚kÿö ¾hù=m€DM×î ´/ UÒº7‰(’`g©WB²Jîa«¤Ð¾.N£‰”£Šåž§Ý‘êù©OºÉHó &#óvÒÅ8D“u±#ýÄóÚ=:‘×¢¸¢v™¼Œö\SNEæDÙ ÑôOÈ2:Åw(sÙ0ÿÂ÷dÿö(1•9_cqÖÑBîº\sÙGqÈ'‘4:b€üåqô_H©Q *L9CQði9AÂØ{ò9ý¦xÄ…ÒõÒÚ¬PÄ‘Aúd-ŠžúÍÅÍŒ{3.ßôSE±¥ê+oæPÂH¢ÐÚ§%ß­÷4^åFi^Ðü)Œ(óqçÅ·ˆ]nÃG ^$‡£ü¬ëìðŸXTŒ9ÄG¥Å€Ê¡¥ ¤Öº¨òåC»KªK¿³ùlp5;ÞïÙ:L†è{œñ°`%jw—ºÚ3?Ca•ÁØlkÔý}äO3©êUêõÅëK ~¤¨IÕ~ 40 \³X¾ŽÛÆÒSÒh˜1Ïë^)¹ ª+F£ã¾ähcl3d†&¼Et,hòÊ/Ü– Mùz­ ĶÐô¹±¹!]*šñíŒØ@³g¿5¸:¼0\åÅŽî´;1£üHÛ Ùz£D˜ž‡ÀòOú+’šw×·ÄËíò`‘ùËÊâm(vvü¡8ó «ø³³»}°Nëé5È—‹k¥Ù ¹û'ÍIßÅíúŸÕxTSû"öÐ*BÕÊÃ!Š5j¨>ÊSC¬àš}ÄlºfˆnÛ9\×/¡ÕÁ¦`¹O¨Y;B6¶¾ë‡IVé •h(*ÞLÔŠÆ¢)´‘/Ñto½RZ‘š+ö)ÛˆEKͼ˜/̷Кù{#Í&%ZO…óߎA› Ã\ƒ@o´]b)/;¹ö<{iTØÐA"šð§Ì8:îï4 ²ÝõÃoÏÕÏiíöRÞåA"_tQ~›YU¶ƒ®q—ò¨Ïé£Ûž¡Ë\ß¼ÐýæºCô*è¹g£7z$ðõÒÂÈê‡mJ.­ª÷c7Wmþ„èzd £cnQÀ|ÀÕ!Ë4øF–´#lð—¿Â¥Ëý±i»ùg ÒÈ×¹þ—§Ï’äÞ¦AúÎË·œe‘ðÉV™eìÙ|ˆ@–Áëð>oÙ-ÉïÄß1%I!Tg®>z/ß²3;ò÷ºyùý¡G¾6@~²lËÙ6(¸êHxÂÿï}áß¿ókoëѺâ(†¶s-Ñß Äjú¹ŒÇ ”œìÑÞ2ÑWK\êŽP‘A:3p*&*#´w{u•Iÿ9m!_¨áÏ>cµ¿bBš ¿@¥«9CÑ'hŠže/ZMýËs«Ø÷ñ@²¿÷ƒŽÎvÊù£¦ÐMv…þá¹úµËÓèÙm;þ£ú '»+ãi`Ð/(øÝ+¦tÞù–H#­ORYDV`¬F›#öÌ'˜øùô³YfL«þ©{ðÁfË–^[¢‚>å¼Ó#°hiFùü|(,]¾zàʉ÷°\wÊSóÔÞ§p…_JÒCxnÃFÔ³g¥÷–…­\œ½ì1ŸpvBB·¶?$¹¢óäk’’)®¯Ð 9¥X5SìnŽªefʇ˜#UJí¨{x2Òž':÷ØM é;7­$‘1…*†ãb"2'[\ò+BÖæñýEGN ‡Éãâ*™=È-´Ññ6ÚùªÕˆ\L@SDûC9ÙPÈáPxš§:ŠÜ>´tó7+Šíå¾záÓ9””.d5¾Ž2%«Ož„ ¡¼g%鸬*ž¼öÝÓÀ•½´[ÏÏGÈýxãé"ª­óÿ2¥B­÷6î¿ï=Bí¤4’äAGÔ}v².B°õ3É'¶¡á|'aA|+šèùqñ¿D³8éâÖWŸÐ¢?!eNR­6óûˆI¾þãðÞ1â=ž½_ÑÞÑ쉽  ,صÃÖ•ç²ïyí¡{IúNhæ5hiúZ@SµÔB Еq6-ä›PÖ%Öäfòw]ÕŠ '¾ üËÅß|¤I±2þœùwü—Ÿ²ƒA¾:‰½×y‹‹ûKÚ_-×ñÓŠÇ~Pçÿ—7 d/H`$²ÃÏÞÔÆ÷^–¥SDCF“³²×ë´cÄAÛÁ8˜š½4À¤S™Ì7nWý§Êî³l>†™S4]tÊ0›t”)üŽ;Ì5Ìîþ»¨þæÔÂXm„ôå[°8$Ë>ón –fÎù ¼'†fÿî{^×`õà=¦úÕ°ÖpöáG2€u‹*™ž1ؽ9–õ1ôoŽý±~qæ(ìH]êIU$xü–‘Õ' “ù>9ˆD%k6¿¯M"ñ'Ùƒù§Hjâ¨éyõ’•©äKÌ!…ŸÑ_ÜÃ)Ašqì:RÍ©-‡eÑ"MÝ¡/”ó.e–†eîÒ7Þ¯äÔoB†6Ö[˜‘q†v²óÉ2s“í•Sx‡,>CMÄÌÈÚÈ­óÊšÙõN­ŸçDަ^"³8+ä:ÄÒNÓø¹ÿ,P<&?„¼WýOí/B~.%{ï†e èôh FAq»¾õØi û•Nm€":d=FÞº™oÜÈ«0Z•´‰ V“*ÃïQÐüGÙ—s($NäðZ …Ù/¨–}ŠB¦K¤YŽ(*ïùîTæ9Ü{òÞÖ¯„=(–óaF¸¾%×*Ã>¢¤›ÕQ>ik”jî“ù),Œ2GfÚkUFQÎ/DòLº)Êw9†Èw- âùü†2>T¶~;æã*‡zd3štQ-7žXj¹5•üîžExôÖæÎ铨mê|})ÕuDüHe£.UPÖ±1_ÔÝóeúø(êq¦ª½|†zËÓ줟£>q¡>]P:ê'‹0wBý:v„j#4ðmÔ¹Ó§®{'4 A~¸ÿnÏ5øãH_"­‰†êâw×ùÑðt—ÇŠ‚6&:¨GÃV…ÇO|÷áöº«YÄšh;;~…ôr5²©p(³ü:ËVJŸ¿PZA‹˜£>CèTÕ»ž‹V €²£lzNrîrØÞJ»Ã…¼_2®kž¼ÉùÕtãsËǪPMÙÆ—}îÄÅ@¦·Èrô5ø<û%œ`zÿnï”ZFcÈR›ÐÐëƒü™7¾ÈP(uΛ¨> •é½Ç+êB¥vòÉÛÆÐ˜tN÷#«$4»hKtIûBËåìl{š$h}¥ô>Šc—¯ºç²æÖ‰Ðî2•Ô}ÜÚùIâV\ÞB›å|EóÇhÕ¦”e瘄ï½þçê5 y$žé¸C4ÿ¬z{íº´XÆ&Û[ÒC+kS`» @ëàÆÇ*ò ÝyŸÉ?ö91õá=Ðižd®à ]“”ó¬õrð}ÓiÃ×£z¼÷ZóçšB¿Ô…¥Õ$0¨“ëòƼ†…F:¦”`TžèÀÑáÇÿù!Ï9Ãs¿*`f;p^·Lh÷qÕ¥šÀbÉøÂä¦,çù:HÜ «;¦*{XaýNxn lº½}õ¢¶­³Xö·"Á»AþãÃŒH$æ:rc׿ìyuM~ IÍ–¾™ 9¾õõab¤dôÿ8û²©ü67”¼CZֻɎAåHQ_8VT6š9?¿ïD¦Ûéí÷¢Eöæ¤õÂ~dg¯lÌnØõCÁèÄÜ<äÌ)Uÿ:•ƒÜßΊ;ä#ëÖ5£,ŒÆeÒl”›ê%9™ A7¢¸eVPɆB®ÂK÷Ï×O@ÕFκETϜښ2~†š©ŸíO7Î"î5¼§¿€Ú 3gQׯÄSÿ{3êsPÛ rFƒåFÓ{< h4Ü`ïÛB&ß™Y$DѬÇiBí7Z mðÚi¡Uë½K¥»h]öÛxú)=Ú…«^sˆBÛ K…÷ÅhמØü~ûÚχtüöî*ªª½û8]Ò-Ò%)Ý8'ÝÝ ( ` a` ˆ *v¢”HƒtIÒ¡tJH·À»}/ŽWçöùßœ«yÁ{ï±Çø®9?̽~‹ð=9°¶`CG½ÀÅÍ tºJGÂ¥½Œû*ÇZGgÖç¬W´ÓÑåÔ¯ãWÌw£k»)ë~ö~.îR× } Œû,ˆÎkBâY·öõRHtîÎ9vÇ’DŸæåƒ–>4U£1|µ“ù7d2nF_ªd‡ìe¡ôn•Ó„>«ÛíÌhAp¡š˜eô ä¥rGæ‰|‡üDû&ÔCþê½þSŽÍPPç2b}PøaxÅ0ƒŠÂFƒŒ¡˜˜Ï2½5J\T‰®C©}}'_Å ”QX ”ž-ÿëÃk /¾1ÒCÅàéú³—}ÈÕìó‚˜ùß}˜œ4Äkf Íræ™o"ŸB+•¯†Á\/´õ%qjíÒ€NÚCîéÎÐí’¨ž| ~l{ÕnÑÉAß—¬;/2Z`àlgšÝþ\bÌ-0æ;ÃWlHøµa´&o0ê ŒwÆõ¨î`…É=ñû;ÂÆ`:K:ò¬Ç Ìr4ÕDÜÙ‚9“çñÙ‚–0tþ›G?,´XuÏDrìÆª}]€Và¾è›h^X»ËÔXñ^~«†æ›— ¶úë>ÍLH¬Ùgá‰$«U\ª‘ìÛÏ<÷çHQç8òÄV©–›ÇØ«8p‡û©Â*™¤‰cŠš¾…Œ‰‚廯 óùë 7¶‘õµÜwù~dý!ÂЈœëË)j÷“qWƒ»íÛƒ¯¹5øu:r´ÄŸÜ9‰BÂa­ƒÛ)(b3}ïÒ[Ü=ò•Ña%º®°öP,ã{"¯¹œe”㹿äËN˜Cùðž«Ö5TÚamíW·ŽÊ#gS5¯ýDµöm…©*¸W€1´~»µìŽí¯”$GUÍ8ßooPOã%¹Ýiu4p;>Ç ›FѲ—ò¤Ñdü—]¡3š =qÏ£BK•ù$Õ¯üχÿ ž©\=þ¹m|¿œ/@d§Ïi¥óŸFÕ_4À·ï«ÃG^@•»O‹÷™0(—?ò2÷G!/÷=ò  4¶¶÷äÅ: »QõgŽÜéÑûÏÿE†¾Y|¶Ðò@Fø³õÀknáÚ°ÔˆõýëM®Oû!¿(לï¡”]à.>|ÒªHu^Ñ\ƒÚïïæñörÆDhNøÇ_¥$ôåÙ±K•¯¡ù÷U€´KÐ*Ö£Tú´æ¨–ý ö†Ö@àÝ<-ñƹ?lX¡ùÑe±¼8hJnv.×úM|Wf–^‡®•£œ[àͬêË„ÜjQfÉÍ[A •vy£ëQh™žç:$. môaä¤á¹r»°´LôgñÁ9üËŸs˜I$RáǨmÚXè3ºõ3¬gÅT3÷ÃLº‘öQþ¹3ŠÚ믹®DËÀÔ:ÅA–«ƒ0»cëfû$ Ì)Ù.= ©_ ŒÂ´`iÒ´iÛ® VOLÜNŒ‚=coâÖja‹MÕû‘;…V T´³ ½Õ´RÁ‡ÑÞ$7níEr݃ä á=ÆŽ÷Û?FjNÚlÍOq‡ï‹êSiÔHÏ‘ÒY<‰Œ›¯ï¡ËF¦%®kï²W%øc¯„z²‰›žÑ¯AöÁ‚5ÇÐfä,zpý¢®r,¸^gDîNßÎêÈeäc–H}éŠüÙ ô5(°(×/¡O‰Bæt…%Q¸¶õÕX#ŠX‘‹:Žb.òÉ3G£QâDwïYn”JºTfå~e„ìÕË©¢ìle¬õ¡ï]¼c´Ô‰Šº®~3QÙê¸÷«ñ/¨z®ä“{ÜT¿™+×t 5ÝF¿ G˜#%÷B­6±øê Ô±Êùq½.u—|nj‹L¢~Ùé²·jthý륹—?‡º¬*>™@Ó`Ó#©ßbÐüŽjü¯™9´ ôÞ :KV¯|Srî¡õãÊÜ2å ´yz+&üí-´‘ 0¨LE»¬sC³²hߨÒ=Ј bŒ2ÁÏЉwƒÑöÚ Üg+ðœæÒîÏ:5KǃέëŽýµÓè*8´Óš'øß÷)­Áƒ]BWÒÆTÐ-zŽìÂS :#Í;ŠÆ³üòSÂÈt!e*ë¥!LçÚ©ìß„ú ž^E»Œ låá» =œPÈ›Um0£ ¹!ëg ¨º!kÁæ ï¬!_wˆÔ[úܶAT© ||J|à‘¿1| 5s™¤Z‚D©Å;,°YÎ ' Iœ´‘æ~ðQeôžèÁƒ–GkköúØŸ=ÿ‡/ƒ—!ÇNçà½È«»o2[©|?äåLDÙ\H|Úæ\1ä_P(Šh€“Ÿ3þW› RRÔeØ »—ú¬k.A‘UÔô‘A(>¡òüçS(ñ^“ÍP†Rç4f&ñõ?sôýã·þú0pJ:K¡¢‡((k2ú¯Yg² ÂÚþ݇©¯ÃV’U Y]ˆÊ¶FZ™ §ªýs®‡èƒFIèd6Ìh¸¥Ýû¢W‚õ@¹iù …¾Ö«ê>5/a Pe˜å}ß½ÿ…ò—-0Ü{xe†+ F¿*|ˆ«º ãm™TÂîñ0)l5Éí{ëÏ~ÅÝ£¥åÏ^¥+[&æ Fõ¿ÖóÂü o=_Xø!oÿ6§–|Žz a%¬Må ¦ÃÚ§¬<†ð[賉äMئ3?ó¸‰¹Ì É@’ÚÔþön${):âß(„á·É8è©‘*snçoºÜÁÉöXºó8Òµ''^AÆ@3éíUdÖZ½kz?²J1°«¯‘ »Þ·Ù+d‘3Ùáuò´aÎýÒôfÏwäý`ÇZðÙ˜¬šÖ£àB¾{”Šðô‡mKãî˜%¶øí‹(¬•[y ¥îP6¬AÙn‡áß›$(߸û5>¡b–û7ÿ$.T¾·'máh?ªy2„^g=„šÏsbNm¢©PÚй,Ôn£¹~àÊUÔ-×ì"£÷EýoÛýÏÐpkì¾£J,š¨¦ü|¿~ÍŽ\j3D‹sN>¼ŒV¢º ¹+ChÍ[0àÖsmx¿ð>MàB[Ñ€ˆØR´S6Š[‘*C{‹›4bßN®ðŽæuÇç¬ï“|íÑ©¦Ú¦‰„÷“q^k”ރΆ†G5[uÑ%B¯v¢Æ]ÇÅ‹™æXñ "·ù±@tsøÔ¾¼'ýß}X¥¢~tí:1PýšÆÒ,Š‹¡F(¥SºvTÆ8ù¿ê…ò_GböCÇÊ:¥à^(îOkøøv ©“Ý ¹gÎð’ÝйV!ÞÔj~ѧÙгdPîB/'ö@-é~åÁahrÓ¡Ÿk¶ù{ÿáõ‡–àè¿ÅDs}<-E`8›ÃÑžå Œu^ó„ çËgk“`2úPQê;O˜,4³Í< “w¹‚wûÂ$ü2÷þÖ\ 7 -˜qy$Ío¿„¯8ò*_K¢+~3aÖ~âý­N˜“Ѧ£M< óÒo£_õ4ÂÂÞÝß:añhÛê™@+XÊ]|W¤x VÄœJå<Õ±¡ CÁ°nrR\7O6M¤©6ÀfÑ“šCñí°9—ÿÂuµ¼‹¿t‰GÒƒ‰”„%\o#œˆQžÚäsB¨w %ƒÄ»CO‘êC¼~l{)Ò7ÞV?å„´ZY ¤þéùó>q$"#iÆÏpQsdÜxðua ™)×(M–Õ‘EØš“W~YÄ÷¼¥5B¶„©79ç4ƒ©Í†ÕÄ9ïgµÖ*þF‚-o¯ŒÃ]a1¥‡GiÇ*y“è|òŽ^™h{ý ùõö·£€|P´~u< š|5[×D¡´FfñÙ](b¡àÕ{wÓ=uõy/†bÕã¶(¡£•I+Å’·ù¨ÜF©L]¢ò¶c¸g×1|/22{vr¢¬¢Kò.ßy”Öÿq¥’åט_”C¹&îQ”·ÎOÓÊBªPÏ׊Pá÷Øé@)Âü-§ª<Ä*^ï¬ ’;PÍÐÞ•Jõ*jðÇôˆjÎ)ÎJBhTÜ:¿B‡ZÞ§\{.¢vhØÀ¡)Ô©saÈI[B=C¯>Ûmÿßù°îg Ÿß]´"‰óι„<’‡£RçNÂÚý}~~dûχÿùð?þ/|H_všÿæi´žîeYý„l ö\‰;a^åòÎr ‚Ë,‚ >™[@åJ¬AJ”mú=1= ŶAÆ]UÎPàá»Ìd y.¥ßO”°Á§ÄÓD Z!»¹óx¸L0dvnȲ' ¹K'—9Žò9þñã=%ȺœRÁ*Ñùa {ù^ÌAYDˆm?TÙÇ×ZF”ÂWNš3¤„<5RäyöŸ„&îˆsGc óŸ$çûˡűæû1C{h½Í­iñPÚ˜nõ¬G¼Ø5hÍ -Û±ÉcËÎÐ<<'èœxš.Òå8@Ó~÷¯¹b„Š!DYD7~ƒ6h>¦µ^î-{îttf:æKcRßÃжg¡Q˜õ'´ß‹ñÌK ävéH´ÝAèV6x±ô~Äp¦rÛ„>[»Z ðؽ԰s†4.P6Ü„ÍÆöv•¿>ä3¨9"SKφ÷FÁ,uÕôx/Ìå]ïc N…8iÃ)N°Ôa~¢#„Ð_Í,žòÑk²ml~·dØãŒD'’cyjX‘xQiA…àÃû_Ýè~ˆäê_5)7/ %=é5/ÂHÍ“Px7çîðS•`&8ž³q‡j?32žwëû¼d‚L³‡ÇŸ–s ËÕ³;®ŽœC6!Ú5¹CjÈÞ{Pjà@'rn3®\¥"Aú#Ç‘{àtˆÚ@*òñ\槨šGþ‚`I³ )X %Û4G!«Ýuî4 pý|ÀéôO(zõ÷ÔÜŽóºå5õ²%.ÓÚQO>G©’ÎQÒ(£rI¼´e—SPÄå ÊŸ¿O;RXˆŠJ¥Ä—ŠPY‰ë¥ÕîÓ¨jæc»úD Õ킽rϳ¢¦ cCãEy„¼žµ •ßQ+Ÿ–Õç0 ê(~t*- Bݦ¦È;ZWP?œU5ÛN 0öËlŒ¢±Æû®â4Ýc¢š"h.ã'w¯çZ Éf¤ßD«³“¬×>Q¡µ7¹ê6G'Ú-a¼ i†¶>YBC÷~£ÿ»Z«6´mvüx—bZ%ètѱø'ï½2t轓¼o÷Ó|ûêÞŠÎêúºFèr*×HX]?XõGæã銣\ºÕè¶ë®ŽG0‰òÛÑzBc”»/LpËeŠÔ¤¯0õÚ\‚ãÔ¥h÷p³×BÙ{¹yyS¼™õMƒ—r%¾WPÅ|…¬è5 ® 1Èù”]2é i–{žz =Á}c×ǬBRažwÛëHÕˆZçð‚„¨Ä#sŸé ±)@µîR3|äw£ î䆴o¢e"my+æëöäd¸Ó!y¡OF<`<çù²¡É3ý}T{òs ä·éÐþ¼Òo ¶năB"úШb(¢¡¨vî_ƒ¢«i‹–þSPl˜Çwuë2”ûä»éL@©û#3’°-(cfK®u7ùëà ÔeOOùBEÏæ)ŒÿëCÚõEîçÿîÃŒ0õ&Z>hÖYu¥^È‚V.žH*h›sP¸Á œR¤“M¢Ðí¡ßØk =;dóÏ C_¿ñ¤v¿ \ÆÑ<;C|~òšìš0rÆž ðÏÐÃk¹¸û »Ư¢„©ÃñßQú2;yÿk”}Æ×z{¨åƒræ•¡¢«¯eÀmuT–ÿ™‘dÙ‰ª ¯ÐÐ?5ÃÏõË!*¤'½T‰Úg:˜Hy ®Q_rfrêk?nÌy€†®‘o–t…Ñ8äÔ•ð¸.4ýpžË¨µÍÓÜ:öBRéÝUè’¥½ûýi"¡Ÿøµ?¦52çΛ…÷Beƒ1·ü‰oP¾~QÑ ZÊ„ C¦@ñÔy¡ þDB^ë3ÉU]Õú«k00'¥hØ¢Ã_UÍÚüˆ`ì÷'•k–T0qû+uÁŽ)˜leÿü‹0¹ÌPÖÊm“¡{u¤‡Á¤ís㇭ÿÖÙÝÝÁŸÀ4ÿùV'±Føu˜zubÌ< îÝr‚Y¿DÞIº0g{©ñ7ºÁ¼­Ëá³~°àõ€Åp¸#¨B´*FMnÄ@›å¾z/¹ ý=©EÛe+èÔöá\nÛ"]¹Ò ð£ø¡ëÍlè»_|gO øü|((† M¤ŸŒIÈNÎc:{Ê¿>d}i09Ssâµá³”á3G`.û›ƒ¤É",¼»Ôp–>[R¦ÚÁªdÆ»ð Xo:}7ýc#lÆê¼Ç…D_($ø°[}+­ŠàÃ@ÉŠ›{F\^Å÷ˆ7=Á‡‡v>SÏFj!½ÏÝe†¸ãrû½ì°¯H¿sfo‹2úÎìÝ÷£™&S-CT‘Å_UµíÁdãn¸&ð:Ù»Iܼåû><áÅíD‡\Ž±ÔƒŸ‘{â¤J²ðäk ]Aþ2¡»á(ð{ȳ/û1 Ùj7uÌø¢p#Eg`Š'M»8‡b§»ùÖS£Äýƒw•³ TKöRÜI”1U3áÝKèƒO»W[ÌQþìƒz“Ѩ(FEMCŠÊ;kkȼ8PU˜k6ü¦ ª Û¼Šºk…šœ¥;¶L[nˆž¸eZ‘3×XøjP‡3¶u+à uS”r›œ™PßåSê—SFh¸KA”ù¡ Mý*ÈóñE“y¯×ñIhÖö©@â«#ZÔ { Ï¡•Ù¡³Ê›h­­—§¹º…6ê+W]ÚYÐV]£>ÈÎí`_eEº?Úëkõ×x‡VA×…Ï¢ã©*¢×ètúU …ò$î žm‰]ÆýQÌòï«.¢s!‹Jt£K׳¼Ü$<@âi}#ÚJ™™ÊáÕU†…Öx4 Ó;å÷ ÓLù“ óÜfñ!rSø:ÊkÈle6Í8_&“)K{†àS]»ƒz!dYD­²3²úߨ·> 5¡ÿ¹Ž @rW}<8$@ÒÑ€RÙÿïÂÞïó AûAivß4$†Ž_O–‚T]a#VÝ6¾s])U2ËΖ»ó˜Á'Æ«ùs'Þ@ž¾ÆñIѳ§p òþü.ü+K0óM(Pðe;ö fÞÑÝoüÿ}y9×YŠÔ8R=W/@QJÙ† s{SPa9 %¯—Ã[Þ@é‘sÁ ‹Äö,ó–Nt}xz·½”0T4iˆG|üëCj í˜$ç÷aöÕѤ”ÐlZŸ^Ê«­‚ –•„÷i[—Ò•W¡…N^F{jîÐ}0ø1×EMèabKÏz·ú¦X_³­ˆÂÀeÓG0$Ú1ýð= OÈëÖ0ˆÁhÃî½Ù;Ma¼ñ«ë&}L2¼Û§VÓ·»Ç´³_Á,ñûµÕ€X˜Ãk×ÖóaþŽƒ anž^2¥Ú‚¥ÆTÏËl°âú¡ï÷îNX»Ù›Iç늎Þ$U°•‘òTíƒ6µ‰Ên¶F’SýAìj[H¶‡cx=I )ØF´¹n´!Õé«Uo."M°»Î¦„ ÒÉ=,Œ|‹Œ|Qäi•ÈÔ1rýr4²dpõ® [nþmÈ_@NvÿˆäGÈUÿ¬þQ½ ò,t·Ù1‡ü™QÆoýäPðcÉå^*Nsî}L¦‡»¹ë“:¬üPüG劜Ž9J3ñ·ÙŸBYÛñ æÚç(/çûÞN^)®7vF¥®wYîʨz]t‘¶·5B,è¬<Êdóµ…P›>N]©[u¯2D½ ñU4(´O‰°ˆC£NYÛ½h2säq¡Ç34›zIßÍ hQAå÷ö‹ZÙËÑ\2ð@kÓÿãh£×I¤nlˆ¶:Äf,/ÑNWNr·ÚñF=‘<‡Ö4ŽgµÐÑutö»¶&:Ì6y8̉ûn$Ûºv ãþ×í·|¯Açüƒe$;µÑåÇ­{…ïzñ©—‰qï¿úp›vªõpà té%+­Ýˆ¦mìShM)\…2™kÍÕP¼ªæ—é ¥›É=Ê!7du–Ôur4óœ6C“þ©ùmZ ©9JPîOòcQæ Ô2ðÍðê±A“—ÈåÖä–žØý8\·nkú ‰Üœmý3þ’p4€á}:ö?Âaœ7éüëGgÜ‘Ã!Ìã¶Þäì4R{5ß‹\' éÞ)]Ç]ùÒÉ[È{lîøÐwäS£òÔèKC~Ÿ»>ÇÜA‰ƒ’ŸM s¦Nþô×û(Ôwª¾eE®J?óHÇÝÚ÷%æÂ“PlóCòÙ3(á—8õ¨v'J¦†qÕq ÔˆÎ¹»=¸Ç¸þÓ“E”ywm­K,Ê ÿ¶´a=„r;js:ºÊQîÞ§©Kó,(OºZ&ïŠòa{Nü&wGã÷;#tQQÓó1qµ*5w~œ©B•»ó çuSQÍãû2Bî54yëÃÄlq/ÇɃk‡F‰ž¶íͺŠZ!‡Ÿ ñn¢vüeo»@Ô&nayŽz’7®_ð ú¿óáÕËvÃ×;Ñò bìBîN÷²ç“°nÓ×wûã>üχÿùðáC£Üƒ÷ÈߢÕt뻪³ ÈúÆü<ßþv˜ã« j25†Ÿ^ùr>P)ôîç¦Øàÿ?gÔB¬ŠŽJ–½»E˜åÕ#+ 9zX&®”Ÿjêã¥ã ÛUáăÊóiúíÄKƒZ‚ õ'uÏAÚö;o AÂüy^’éäˆðð}pƒ²†ƒW:î7è°U©¾­%†ŽÛN'fYFq1õuýS‡p?/?•ž&:S}S–»Ž.ôÃ}Y¯Ð•ƒB}ê÷(¸¸®°ŒWùe.6ü8†=vÚùÅßÑ|§úû¯­0ÉÖŸt~¾ò¬×tÓ5@‘ïýDñ'?@)Xg‘ ŸNFmI†ÌßÙLÑ$žÆÆ_Ë ©™\Þ þ˜ùõ±¾$ÉDÜIªà…„}ìzi!~Èo§ÿïHÔ³Ëû~ ’çlõjžÿ9צ>I5…ðzúõÕ3ÿyý¼ü!ýÊÎ (ØõC.ìÄ(pHáu÷‚‚‡7J´íþì ‰¹Ÿ­4@á§ÄÞQö×„ëƒØ³Ó óPÔÓBÂû!ŠÃlf,ÄÌþœk¥YrØ J}JFö‘AO·J«Å·¿><®÷Ó¥¡*¾:-eþõ!©aˆ¥*ï¿û0÷Ì£_A³],Å£Vh•¬!Ö?'í¤$Ö×'6ÿôÛ8Q¯Yèö¸8èÉ( =ìë‰íO¡o¥Ïl‚n Ú_µ¾Í C{Teèèax–ógõmŒ¡(ÔÉ€ñºŽŸž`’ÜüÕ$Ïa˜ögM’½òçÜVE­0§ÑÈÏPó÷L£f£þÇ(¿2¢Ý, µl$_åZS é‡’+´-Él<Ѓî7^êÆסϾ(ÄöH Ò'&Ð~w€á••vm^×z“™+dý^J¯¾²Á”Æ¥$g.f˜r;žÜú‹&çÞR]ª¬"8qÖd«¾ûŸ:åæíÖX«ÓOœRÔà×€æµ,i˜¥¦·v¿? ³Å¢L™~40÷jÑþ»·ÌG6N ~d……OË7iŸÁâ˜O‚^ôMXÖ¤ÝÚÏ+Ò7›Ö pÞµý¬TÜ笆ß$I’þ€-7 1£¦G"k=&–U$– LfG#raîŸ'‘ô¤xŸ¯Ð"’%ß‘]G òª˜¶R¤¼þ´$dõR œÕÉY-EšAÿ¥Û«lH[èÂJ2ôo÷&tO2; ööšG¿,×;¯#k†k^]s.²SmØëH-"‡ßÄDiYrn>œ"kj@®Ç ß\ÂqWãÂÛ±Ð0ä©ûâ͵†|æ‚§ˆ¥ÿº€Â‡þ“(pz"ËxQðÒÖ™INeZzÄUùEž%¥6µãn燙7§PœóóËî[A(Ñ{Ça''JÖR“Ì==€R[:{Nk>Å=nÛåÕõë(SÀÎιueyÌø§=w¡Ü.:a²Ã¼(ª±@†ò<¶õño~ |òQ3ϱO¨p\þ %7*º¿¯díA¥)ò„ŸxP%òEë½pB¿<yýÇWÔ0îÑNtŽâ.ì×w–ÆD¨¢ÖÓ>Å‘bÔÎ^”:E•€:Ó”ëz?QO6â|ãŒÖÿª¹ª)·Ñò˜ŠÌ¹gcÈâ_^{" Ö³ßNÝ\\ýχÿùð?þ|xc$³C»J­|)§ó#Ërw°@fÌFüÌrú^ A½æµ¼Ož[Pvàq±Ä³„«þµËwæ´!߈éLj—ðŸû )ÏBÎêr–;#d ºM?|à™ä4?RSþœ›_çÙi¡$Æ­Ÿ9!óÓM¢`sÈ'ö£ãËℲɯ¬¶Ô Ú¬Ã‰·/êRÔø‡|áÛ=«ü¢Û?¡Y‹Ñ#SÛZ캒ùÞûCë±Ä8¥CÐÓeÙ4ÏíBg䥧Ӡm<¤¡&Ć·ô;4CrÐòÜA›è4W_¡šeß$䓯Mò°4µSw¾øó\ 3Ý¿¡yx*=Qþ´Äx°/f @k yÑc‹]Ðö@D:ø] ´÷M%f“Å@ç%‡×'ߊ@w؇sî?Ö®T ‚¾†—cwÏ1ÀÀ£Æaæ/`èÀs籸Iø×‡ô†úôö05nÆëp¡fIùÅ)G÷Â\Š·žCÛ5X¸cm›G K¯löº 5ÃÊD@™ÏF6¬Ÿ]ó/to…MépýàéÝHÄì:&ö–‰ãD%‡õ‘Ô~êÖÍŽ{H.ÀÐ'j …” ‚&|H½çR³_û î¸Ãb½ásé¹eÞÞóAÆ“;Vƒ“)iPçÄh²œ(µ@6Fë‡$ w½É¨úJ÷Á‡FÆe'f«®`üaÏyäþ}ìéƒÄäÃs'âÍ2¿.œ•JÎÉO>åG¡}ÌIòóP¸Ã”/á$Š>$q“±òG±àk‡å|Q"…|íž= JýžßtòD™³s*Í]Ú(Gu”9ªŒÐÿN=·`e:‚Š´O{„ Ò¤è‰¡#¨ÒrA¼ÈƒÕ>O“û|ú‚9¼·œ—8ØŠ4š)5PëhÎþÏP»ÿA}Às*ÔõÚœv Œz+[o~½¿O.¦wJ¢‘eÐUÐ[Cñžw®d ™ rƒW\1Zðþ¨6>±‚–ë w++ÖÐjús^n)Z÷u• ŽB›Sþ¼Á~´ýñy¾ŸÑ® ¾ñuaí\ÜùÎ"3U¯f¡cÛ‡³ƒètI¤ç“ú7ÜwL‚îx9 î·×•–V•Dg­ ò´{è"/ì°ïº²g®$*ûà•PѾÄa54”ò|MANƒ - ‰Ú¹é0¡+6Ûÿjí;´Ô¨-¡4…ùûáçÇ ÿ¢àÓØ“aÿì÷2§} ”…ôóºí–2ʺׇíE1Hæy13¡H ‰?‰ž?qì€æƒå!>ø /ìL„ͲYGš>Hîª{øNÇÓÝ1¼å\„þÈѹæÄõÏ~2_ö.tè=‚o.ôò}‡‚ÒÇ9ŸÒŒÿì Ÿû`>> ù–¯Žp]¸RÍÒ¼Šž©] Ø} Ši‡©)ôL¡8=ªCi¯”ÔQnþZ"ƒÒ³ªkíÅ”~‡wyŸþã¿><ìq·B ¢òÂÙH¼|X½y¸áweÿ¿û°à¸J¶C³«Ÿ³ÔÙCЪ™QÉZ í;:¶Dþ‚NñÎÝô6èö<.›šÈ=Ü}Ñ]T3ÐOܺÕǼžge26‹Ãҳà !`xi;<–ô3Œ6q5~>Áã5C™ms0±¾"%KL ÓÇ-|øŸyÀÌH¡âÅ»aN•»XÔÏæÃµYZ¹Iaa;Wë”> ,½Í]Ï÷‚_o ©-X»a½y 6¾½?ÞÒM [N ùƒvÙHô€ˆ+"EIؘøìý„´œj80Ø É_ë^áY{‹” ÇFJÓ"õ|ò\]àÒ†ur6–D†ÛCùŸEEéd³jF%²¨ŠÆç3…"›Ô;ž­‘OÈq¦Ý!À»¹½jB¼{‘C>ŽÛ½F~ûé²Ì݈‚²GœG¸gQX§£õµ3Š>‹Ó¢ˆñGqESÕây”RS>›UŠ2/— ¬e]QîÑÄ ±G¨à[w™÷5*íæúÔHV‚*Qàú,‘Õ/ÖyÎÈ6àÞK\Ìyʽ¨e¤=O§L…Ú?¾ì.¿mKèg¶oÒo ¢¾ÿçÆFÏD4¼˜¾×,™/šÝ‰½b†¦žÔñúI[h®ÚÔC†–fGd”¡•ŸT+r¹ õ!¥ßš¢yÞœŒØ9´Uéu¹²í øun1 = õ^I´Ÿ§ rF‡.»W¤èXÊWz)Ð)î¨öºFî»÷äó¼¶î÷íÜ”ì×Fç}.£q_ºÿ͇H$^ËâÎf]>“¡J~ÐÌ^?sÿ3ÔØTš%ôÏB©Ö—¾wð™§ò¥ -”Ò(EžÈ…†rÁ’½PÐÉÅHézr£ƒÊIT# ÇN«§c9?õeOëÚ(¿YýÔü\#ÔrÄë{BÓ ª©O·íþìÁUž=¿Ýq{e«ü¡ïp@»óä9ê>sÐ Fè^I;¼Oqs-÷ý`’6óë=‰D˜:%ûói4L=y.hïH S"vŽ–C0¹ÀÆ$QÿÔ©ç÷…žÀtSÙË.ã0#˜ÜkR³â$œOl`¶ø.ÇÇý0W*Ëù•Xæ+ÁBbQ*ÿ5XbŠVUJž„å#%UÑa¥Ÿo—/‡¬üe–Ô¬¬Ê)¿UÂï Ýâ°õÔGOü"iÿ*ˆï?ƒÄæQÍÇR—‘䨗É!7@Òõ_B?#Y#Sû•-¤¼(0¸ó7R&2ŠöØ!µÑ'í¢}Þ¸ƒæÎ¼YÀ6ÒŽ$´ÖOCúúc£ßï £›dê³Èä~MLÆÆ™ÏÜÿz/Y·ûxŒY¿Ú\rÙʼnöî:‚¯ }סw ~ôI+A®‚V7^㮑¼,ìyȳ,:ÒfÃ|wö½L@þ§9kSÃv(ÜÊËT<‹‚÷óÅÆ÷¢0íåú3ŸªP$m÷¾uÜ}þÆêùiWø~G’ûJ¤²ˆî<Ê’Ãw¿¸Å¡4‡ö¸]–îñ“ûu…|eš’ZD¢÷ ,Û«ù씓°|(ûËå²âî^'^Dyå”nî>i”¯}+ÙXK‡ ¡%’ï¾ bàЧ_T&7óþ}UÒµ‰b»CPíŠ{8±‡:j8¾Ê×8‚{5–YøåV“ÉÏÌ£Öû}Mcmä¨]á­ï[ØŠ:+“ 3ÔQO)ö˜¸Ñÿåûï‹¢¥’ÌF§Õ rßljÑ:œ^4‡<ôþóá>üχÿ ^OµX¹ÕŽ–ßŸž2ï‘E«#O„i#`–Šn¬ÿ„Ô+}p ÿ‚PaY›Kè{âÑ””ì„üu¨²‚¼ŸsW¥¨X!WÇ#n$ÙrŠ…^qýö‡¬±°“œ=!£ø}"ç#HË£×ê?uK7±{AôÜøymµòšlò~ƒrjåĆ4O¨ŽÒ¼}h5Ÿð¾‘{ˆjáÛpL—w~4ǨSúõ±AK&K‘ëih­XD¡·…ÐNÄé`ËìíWJ,² ÚuÛ}%é¡Í}ÇeöÛ~к§Œjð‹´Ð‹-jÜ …fýiMëú§ÐLÅa´Áršiû˜ð ü]dŽ’ä´T'Iní`„ÖÈwqnÏœ --t%N<Ú7Ý…¿Cç‹:§L¯"èN¼·& =ìÇÇ3Ü¡o†ŠóÞÓ0K)%”E˜3§kɉˆÝñ¯Uþ}¸£Øð¤Á˜ê Û­Høž‰Æ±ŠPÁ\RnÁ‰—Õ°p5#¯mO!,ݶíTu|+5½=İ®¿w¶¬ ~/¼ûMÒá Ûã·­I%ö ñ=.Οg >Ô5 –ìGrÎKlÁ™^L|t©9»ªqGXzV¤Ë4ÒóX¶n&§"£÷K‡]/‘©'ìæùWÈâu»´ý¡²Ñ°pέ!{ÝäUÊræçhD¬¸!×·_;컇üˆÉT8ò™T}¿ío„üÍ{i žç¡ õtÊ©Þ#(äò ô9¿ wMfè"äëiÊ÷_……(Þf1p(%Š÷É~ªFi¦»bO…P&„S…Cù3ÊÑ[¥œUéGyŸ×ô~¦¨°MÙ˜ÞŽJÕßíéQåƒP•5ªEä«Û5¡ÆÍúÏ¡5ù¸·2^¿:µÌ_T“ÙòúÜmãÜÐ~ÔÕõ LLªA½Ê·lÄ¥Ñh`ób%ßú9Õ¸¥¡q3íðö„ š~¹àDöø šWç®~ðEË/æÐª‚Tmùõ´N®cŒõB›¨æZ‡$)´½/Ç™Ùrí.Ô™ü0÷@{—`iVÉt€Ç3%ÚèÈÝl¥P߃Ž+:jâšèÔ ´^Õ¸/滼0aþØï{±H2ùÁ‰Ft%9äè²cü¬“ê0^^ä5¼9E‚–­MõîxJW>üôìy×Ó  5÷3sž“5A©s¦8?%@>ЄþçþÀLÇÇ2T%ß!]Ý÷n«~¤Tw$užë€%A¬Z‚9øÖ'þêäˆÿ:g^j•ñ» ¾’(‡AÂ'é`u5H.7`!-`…ô=)ç"LoAÖ™ŸÃúç¯ýsc~œüF æs(¤”Αnî†B ÿ«m)(4¯<óB¢ G Ô“ ¿B‘´mªöý(ª29r¦ Xc‰aº‡íÏ9Ž%Ü6\P2ô „78Jýw)|¢!\O"Ë{5þúð@“Ðu¨(‰˜u /ùëÕëlI¶±ÿîÃb» å»ÅÐìež:øfZ÷ž;¬—íí¬)'…mz¡S¦ÈÈíw>tu¹²®½=‚Ͻ5Œ Ÿ2ßј§¢YF¿ñ i®ð ’öÁðƨRø#Œ6?~^ûäŒWÍô–3ß„‰ÙèÂsÃoaÚ9D–‘ÊfÚíòÛÀœâq—½B"0ÿHùö§°£°Hõ¢@ðh$,¥ÑñuÑÃÊ¥ÚûSÂÚuÉé˜óOa£ g1D? ¶$JÍ•O#‘ÃsM$K<¹Ä¡„¤÷ˆkÌGòƒ¾¯£²BÒ3„ó¹¤N´©QnCÚ½™]–ÅÈ ¥üèìÂ-dR7Z2*äGæ ©{;;’‘uôùÏœÈ!¤¬7ýÕwÖ5ÞÐàEîá'­ ~sÈ¿›u³àJ% l?Jâ,t|ÃWä.ê’Žß'—&C½Î€~Ÿ3hP}¦YÙé ­(šÄhŸ úbö²:ôêf1¸]@Çr|Åœ´èt>­ò¸ýMÜ$J‡.Ïã~2'ŶC¹ÿîÃÀ³c«×( ëú]EÛ“ß¡YðñJÉšÔýY U¬ã’liOà³x¨×¿£PfÓPq¡d×­‹/¡`¨W†j)ä&ë} ~?9ndÑúR±ÿÔüÕ{päå6”‡&ίs\„Zþ/7œ–ú éäú;±ÅÇ„>Lj щÐʺ[W1úΪttY†A©Ý»š†.ˆP¡`Í^¿ySæ×ʘT#v¿ 0õvð‹ LÕV>k+ùSû(⺶aJÌH!Båðßú-eŒ‹{~Ñv÷™,ÀŒ‡jÐNWs˜5áÈh| s¤±dƒ‰07ìý=ÿ™ÌOèžm˜†Ej—åKŠ'`I£ù‹hž ,Gø†ï4j‚UzÆ æ™Ó°Å·R? qíïá÷*ûè3ËQتKËÐ4îC¢÷ y‰Oð¾ÔhG’êÓ‰tHšy&=fó’-h}º…Ö»¨!eã0¹t ¤>µ¡õ9ŸwÈ¿˜aÜü‰tô#«Y!ç‘~õñð|Ý5d¼h<°ó•52ˆÓS‰#óÝ wÉ NdI[¾Û¤‡¬¿‚Ê ÙÙÉ…v!GQì†BÁ2î4±jOC®!!ѤDܵq5ÞŽ /_°Í%äËjݦ'‚üÌJ¥ÇÕQàeÂæÍÂrŒ<ß7’~ ……Ž]ÒMLG‘/ïúJøpwøÕƒ*§N¢¸ÅÂÅÀŒ(”¨5,ÿñ‰%·iÊ£$~¡´¬ö[5)Ü~˜KÈPfÈk/¢,Ýd…0Å=”S{¸åê׆r_Þ|'»3‚òVÔ£ø¯¢üøÿ½óÇúoÿ¿-Dö–‘ì-çiïì©”Rh B”Œ’Eˆ”Ý"{e”ì•ìì‘MæO÷ïw÷ùãwÜÿ~¿ÿÜÇ帎븎ëz?ßÏçÃë:ÏSé•Êd"5=pec“j ß‘£üÞGúÇ¿ BŘXà"*݈õ•Ž*§ì$ìø›ª±²¥™K!¢AúgÓLTûÐz~B’Õ›^ñ DM¢†©ÍÔ»¨¥üáÈM‚²ÿ=>¤²K¹KM…kRü@N{é[?…`ýšˆ°zÃýÿòáùð¿|ø?Á‡ÜŸ_ë^@ãîp"qÛ|‘¾˜¼ï”̬:É7ü¾_íÛê3¥AÅð‚{ýòPº˜i2t©t''Š-YæAþ&^Ï ø¸þ…­b4 roš´ þÑã¯GûŽB–7­®c€¼³›Ï±' …·ª×iB®RBV*7ñÂ9Èa3Í]²““ÐÀFªçžç¦Ó­B}Ѥwäñ]ͯ²MSA ÑÉWp= ZéŒi<˜ÎBtÒ+=†ö£6GZ¥3 ý;µ‹)o3´'°ì­ÿ3ÿø”ßþRh #ºuŸZ,[ú‚‰X¡ùF#«·ü04+ ä+eZAs´Ã½‹¬¥Ð¢pLÒ©¹èÏÜRuòã[ÐZ¶!ÖÿÚgË>¿éöv’È;sÐùQƒqu¿%|¯I¾?E =²—Ø)É þäÌ—Ÿ¼€;bχ¡+×V¾é7À°YÈø”Á?|¸«N÷‰o9Lõß\™7…_›ŽËjÕÉ0—J&xœ®0rÏ·ÂÒE«‹´wÌ`%]£×h—¬1ž©fê‚üOYþeï`û•yK­$zîáM%ÚáCÉ×zA7d‘tϘá0¯Ç\#%Yý³7m€y* ©"MYÔŒ‘f¯;ùù¾v¤uþðþšÒuvNšn Ãq£±ƒrÈDÜš8èŒÌ5âïð¡ÀÓÍÈÞÆ#ÎÖ)Œ\4NK—%ôÛšÍàLròtL’p"އÁã×È,£æÀmè¹­èõp_œ½Í~*²«7ªE¿½:3¤Cˆö µhéG‰8”·ÔA)F%7…(íšð­'Aeæö±#'ʽºs³íøNþ l:1©F…J'ƒ/¨üTG _‚üg¨šfNŸ²Õ¤}e2h†P=Í·jÙ75¹?óÎUA­GÚw»‹QgO¡jÜ( êfÑkTp þ5§+Vhx>g )P¹Gí=>ƒ&GÂøæ¶ÆÑô©Æ'Yß'hv³KyEU ÍÏÙ]\ððA sþÐ7²»ÑRª®:us­¨¤NÓ|xV½§2¶êÑ:3B½-£mmJG[õ1‰¢»h;sQCû:~©~É,Í í,ÇVDè£Ðžó¹Šúez{laG¿u!¦q|‘F—œîÀZ9Œ…<üTW05U÷ÅVÌ¡”ÂâJ÷úÝö3{£ôw¾èRå)“x·ý\ýð‘“;~'oàày2O+>¦;ý¯}.þzt4_ªRjÅù?[œ‡4÷8—ÎàUFIäI™&x¯#§GÞ:ÿ/ý>x8þw>jÇ÷v«™äßm÷ö êüs^xxÂxfŽT1µ@‘«~î…ºý®ô‚Þ-ٛܳvþÓwÈÜv¹ïOV*ñB(Yg»aÌì ¥~›nù!4ú´v”áø‡mã“ɦ 2ÿ½ýÐçøp.áˆÍÀéÿ̇e:÷L¦@ó~’èš@hÕ1`ª´šƒv®[×/“5@§Üó?ÝDøîj\¯õ}z„Ó#6Ý’ Ÿ:áíqÃ@êEöˆ³Ö0¤eÝ»ó9 ~í¾¾ê#­4ŽMÅc0V±AñÆ™&F ŒòkÔaÚ¨l~ÛG~Õˆ]ÚâMƒ9Ékõ 0µ=–à+,Ò]“k»+ KEŽêã+¹°,*vUZ~ßx§;; ë©Ì~ßÂæ¦{_Ø–(|'ž’GÂOêÅ‘¹ºHlõ{hãøi$å{^À»“/ÉùR í “âØjbl}RMqös"Mc¨‹@éÒ‘–¸T]EúbiwŸøoȘž]B{™[Ë8îç@6/¹*?ÊcÈš´æ¯ƒÜ³Þƒ_E#ï·¢U#ûäo³¨{Fã‰ûèýÚ保¢ðÆ«Fwr¬£½³ð'¿ÇFrýK(J.:zèq tFe_gÊ^º_p°ù ÊZUó]tE¥3Eéäñ ±«´ÃSDâ‹ZþT› nà­¨@K¾×m[FQKAjŽêú+Ôa‘>ÃCòu·¥üž¢~¾W(FÃÔûI|^à¡[òù§%‹ÐD¿øè¾”u4 xØuÆÞ Í{>6ÜwÍÅîܪh#Có5·Ä_ýhQ¾T@VÎŒ–·÷‰ý*ñA+Í™,fÓhµqE$ap­ßþÞ XŒF›£_>:?E[Šýn±/Ñö­ÅMâ­¨ÿ̇õ\…SyÐõD´×Ãi šÅP4}†ZÓÛ#çÌà󾈙oÁP!g"ð3ÊŽ>²p¡[†!ƒ¯5QÖPð+ɘ– òòvE2¨@®k͉¬ÏÿÖÓCk£¡üþ]“”OÜP·/óعÛ;¹ÒcÊþ©c´“.ê4¤uÃ÷“\%sbа¸§‰0¨<Ü–}&†•zXmÎ…±¯ó”Ž0yÊxåÜx3LµG=!‹(†©íjkáv˜º›·|ɉ¦. õœù[§w5f ÂŒî"·„Å5øõ¼’wXê̺ļý’æö9úö³¬Âü®ÔgØaaϦBúõð(M+¡¯g K§ImYªòa¹DÐæ!ÛkX=H¤÷¡:~7˜jÕ>ƒuöÙé·aSlwò¡Ë“°M4®+뢻ÃÝ"m>'0,þÌ÷ê$J~<{¨‰Û™žæO8")‹“ßÀW$»ÒqþDÒe$_&ZºMOëØËÊ.}²íF3R+òï%ù¢‚{øK·Š|öÞYúج<¤{ظ0@¸£‹”C]w:2‘¡ž%i‰D™v?ÔZ‰BfgÊÝõÃo¥Ÿà¦›²¹] ÆóÏ‘ƒ2–u£)9é´ƒ\$èp¯ÞÐw,µ@î^»]§ž#OÉàíóSÜÈûÖ÷ÇʧHä{+õxpK”­ö3ÅÆ¡àïjãQ-J¿ZõÛé-ЏPúª÷E‡¯K$cD1¶;I——äñ€¾ú®•(žòX{â Jlµ‚¡|”$S‰*±ÑB)½V×0b)”ú%L׎ÒÎN äŽM(CÎ4ñâÞ'”ùòÝöR®1Ê8vÓlÆ¢¼Âw¶8 ~Th°mö8ŽJ1£;6Å‚*YSÆÄ®£êq?Óóåvˆ'íë%œP­Ü„jê ªöº*(ŽšÔF¨vt­…&?Bþ×øðöÏ—çÎ?¹ˆÆõ¼‘ åÈ)uoøçè¬÷Léðüñ_>ü/þ—ÿø0°0NîÚ…‡ºoFÑ´Ð"=q@Žp±)L÷¿Šïc¿ _ ´ÍŽÑAŵ®|Z’]2äß+s Aþwù†‹ð1Gænßü¿öÞûbøçüÁýÑœ5dI]~1ao¿±ÈTàÞѡ—•…eÈÒ'Òœê+‡üKº=ÜßÔ \‹ºµj‹j´È†göËÃ7­3BKæÉ0×pÇBËÉp3vch½‚1õÇ íZ¨A£;´>>ie¾rö·U† }'Îê»A;]É©äÐÚ’¶±Ï¦Zbýêš½’¡¹ kñY¢4Ÿ“ñÖ”†æjŸ·qtüÐr<úByR´2*[ÓæNCë°¹×o‘–?óïŒÊÒ@‡¢A‰I t¶~l:øk ¾Ò çŽCÙTí~{Fèg™SÕu›ƒŠ+ÕÆú0¨çrä„7 WÔá¥gÿ‡IÛuKs¼aª›À­æ5üZ¯9¬­ s/,Ò£-aáÜåÛÔç§aé˜ õX -¬„¤LÐ܆ߣŸ{¿^øþg72aÛ}¯õée)$´%uz°Ã‡{©ÓƒZƒ‘”Ü2P+‚ÉiÕŽ{…"šŽZ¹T13>?UÞ" ÷}¥~ú-¤=ý‘N÷DÒµò›kRþ@;zŽ€QadÜ|,Ü\Ð…Ì•Ò'8m5ÿ!·ÒnMdï27íi @.–/G= ÛÑ5_·ù"òôDoTœñG>úYAãEä?QEZ>ùú_Tö¹íÃ}Iä-nQš(üòzyw -Šön'Ð_ÇÊ•«þtÏPâÝi¹„¹”bça|²ã/"4N3 ÌP–[Þ×(÷@%+4Áìü¶%ÈPIÆò%ßš/ªð‰©Žyn£ªGœß½8Tc9žhR—êw/iȺßG уJ}¿Pëø8±ÐòÔþÚ{Óÿº;êž´ ÑF}þ…¡©254d\rU»Ð‰‡Ø<ú…]ЄúêLÇšº*{-B3kºôžö4W亯X…,Lki‘ïÐbªVëâ›;h™ß“@¡bŠV~¾‚Z+Éx· õôÕÙÝñLh}¨í¡Š9ÚªP¤7DÛvªý^/ð°shÿà xxÙÌ%Wµ¯×P³(¯G¡6[LÆhG.Rßx¡-ÀñF3u?¦G@õæ¹Vë­Pâ6z ö+ä+]šþFów?aVÖ@Úv_οööz®Áë‘ãƒî²ÿÚk1EÀ éLºUV  _^¸ÙBŠ+Óõ0wcH¥Hˆž‚W”˜Ê_EÀ{‡BŽëGAöhøZÏ;Ó¿ûÿè]21 çý•ëÛ¡pKsÿ͸y(¢$çŸ#h…¢;—mKŸÔBQÒ¦<Éáÿ»ÏÂmŠŸ²pÑdü9G¹7qml‡oÍxŠ[" ôV_jÆ4”É<4&Òûš(އ?ƒÊ߯ë¾þÇ“…/H/ˆüg>¬TT7¼Í^«‡è¡Õ„»Zsù6´ï³ «úT*A±ƒàûyuŽˆ‹Ð#~/8"aú‚vqÉÞ„7ÍÆÆÎÁaîƵ÷0Lþ!Ày>FÚBD;ÇLa¬l—ìÓ’4˜è^ø±•ÙÓªë934ð+ß)Jý\)Ì vñ.‡ùhÞ ~¥`‘ýÈ»·Ÿa©.çô棰ò ´$&;~ûË&©Êõ‡&äsœÓ°ÙÈëÑg>ŒÂ:« w$ðîwY+;$æž?YX?‡$Ã]/D­¬¯ÒOäR0?ó&¾ò©ž–O¹_EšK?ÔÛîé m“ÜÔÕomH hÙuríop v#sÐ2okX²ñÆkzw% §ø;žÚ²Xä~øy2Ú>yÇ'Çuü× ÿÙƒªÄû[(Ø0É]ÿt …qЯ…¢è’õ@…C8Š“g =ŒáFÉd.mvy”¶hÓ÷ÏÙBYÃÂ…üd”7¸6Úùû(Æ\P%Ò›åüì ‚ª U¢ÿQTs—Ô与êÃô®¯®Ö¢fÔ8azˆ8jŸžß x/‰ºšü«[B?PŸí=ÕÈ7Ì|³‰žG£O™ƒ1h|Å=om‡‡MiË"²‰Ñ´Ïx¬KPÍRÙd®£ùBGðU´àv4yu7Z4Ê­/öQ ¥ªš÷ëhÅž– yÊ­²Ë»µ›ÐZw¤ÆÎ÷Z·Z&7ç[£q“tÌÅÿȇ„{«Ùº@Wf]H„ÝâÿÝ£’¡ µW˜H¥‰à³¢BV°¹7T¨3™A™‹Í ÉÏ((‘ÙÃ[ÏÀë§Ý(­œ!¯â«/ÁûëuW;úÙßZ@Ëèp·WÊŸeq¤êÄî©Ü`¡…¦K½ T Ì~'mìúž¾—–3|]}÷³„Ž)°Ã þˈǓÑ0l²EÝÑrÆj\^ ~ƒÉqáúg`šRï~Ôn#˜–ã lPS¥®ÚŸg`êa"§PâÌß:­6ïæt!fnR;Ú!_&ïÉÎÁì^FÕL˜Óç^Ñ:¤óbCŽ1»aA.w=€ù3,ÚUœe>¢K÷Æv5sÃòD'åD¬:/ôY&TÃïí@ûUF°žN/;x69Çþ‚m%Þbao‘àI5ß9ín$Lç”~d…Dµ¡Eò•¯x5ÿfS‡$’ª„_êg@²¸àȤ‘,ÜÅyýÈwç!¤(¾ ×Î!U`þØSâB¤v<®iEC€{Œ§x)ßx mòm-®:¤ËôîÊïÚÉ5¥ïÍ&7Æ?±6å"“苚wA+ÈJ³˜Ê¯¬DT)¹*Èö0/'íÐCä¦ïn^-GN!ªkQª¸×SÃòÒŠò¿£Š¯ˆ@žVo£·³$È[i^è’wùJ¦õt¢€™Z×hx0 þN ÜtF¡Š+b‡çP$ˆß’³·÷¼Û8iÅ€b²»J|ñÀ uf"?/ªñH|Cƒ’”ÙA‚å%(±Éü-Çä7JÙ2wæ†ßE©Ù»W)b>£t`Éöõ|(#hRGíA‡2Sg2ce»y¦š %QÞŒg¼#â**ŒËß/¬G¥ÔŒìëê·QåF¼FUÏr£»l+ˆ×©ÎKÑp¡ZcmÚÌ=ªÿj¼ïÛeš¬OúØ)%PK«ESž^ã‹f¤Í8Ÿ¢q »#ñµ&ä$ŸÔ™.x joOneý—ÿˇÿåÃÿ ><ü<—µ& º¯ ý2&D:šªý,â0U³D2£ui¢…+¢WÿôÝÒ¶;¥aeqæŒPÀÑéÆí_ûW?ÏÇnß8ªþg^Vö_}ý[oU½&ž<¯O/„žqúû÷|m…zîöÕ¿º«y|ŽIFë2|ûò5—ºà4MGÉ]1ƒ–ÊÜ¥Ú“ÐZ_î^d mí­¿Î@{›sûE€ŽûéIžV;œhUrìƒú_µíã\ªá¾ÿW‡ÍËQUAÐü cr»iÿ>n mzËã9­Ê®ì»»ÿÑ+]¡5y>þ}ÝÎùáÙ7¡›$¸½4 z.œÚ1 7èj«ÞÆ?ûŠ#‡F»`èã‰k<0|š›ØVaò>$ÔŸ–€©ö=1nuðkñæSò˜KxÎózâ),œhŸ‰ ƒ%CÛ´ÒùX9K¹.Êøå¯.7´G¸}+þÕ#¡êo>ž>¤:Oy[¸I¶Jó^l$íðabfÁ;RhMýÊ!²Eª„»?‚÷Ó" Ï»sf:‚Hë¤}¾ÇÀéÜòå˜7‘Á¢C¯ã!'2®Ø¼+_ÒGæ’¶¨”ýd;|H@Gõ8Ù{n]|Þü¹xŽ[ÎÎCîsÅÂBÛ»gHg‰ðÛ ò1¯¬+#ÿ©>M[3]*3ÓU Á}©Ù&ƒQøMãõÎ5-ýeÞ®,¿ŠLf˜Ø£De¸ìO“ñPLÇ]Dé3©&©b+(ÓNžcs8 å.O¨ö’¢‚´XÛ^T"æí¢¤ÍGåÙîk,ºÏQUydü9âjÛäîTww)ö1ô;ˆÖ›‡ËóYÐ&ø…:…ù7´%oŒO4F[?ÑÚõ/Ùh;4rJM¯K“ø3E­³™¤2\³¸ûú°l…‘ú‘—Oä¡ZNŸ-{Åú¯Oý[w9Ѥ$BŽu<äÔú…¤¿þöoe$K¨ÏTAÚ·ìVLɇ”U¥ûŒë™²{£Í8\R‡® ï©ÑWײžð ‚÷—5ffG¿üé&©ÿøWŸ…ä¢w©Ì‹¡è¬Ð&çdÛÿ_{šÏÔ•Ôü­Å¾ÜIA'þ¾ßûv)S€uìøßûH™Šö#7ž¤øÐ¨óë×"Ó¿~ÿ—‡»Æ³6¦þ3V‹L—g^æõoH¹žC«Í’wÅh—Œa>ÿ:Õ]Tòz\໇Œó÷ÝŸ Gö¢?o³%ô³¹ÆÈ• ¤Þºùg®”ù½+ì0LýTÈzjF:úé>ÂX1ã±`f˜hŠwÓŒðÿÇg3ârn}ì‚9 ùþû0Ë<ä8‹ü°_¥Å–:)îˆÉ‘ÁJì¸y¾‚!üöÍۈ詂ußàÉ!÷`3©™D&5 H?L‹ #á1O‡–ÅsH4>1ä«rIÒ¶W†Uvòñàû}F ¸«^ËÄ•û:R韾µLÐ…4ll·íS‘6îÕY†NI¤×%%Qzddô2ôWQEfkZ݉cÈú•¤ìbZ?r´é2Ï"·ÓÉtÊ,äucŒüð´ùÏ4”¾`EÁp—Sš×PhȪ +oECßs=<Ž–§ÎAI«Î &7”Ú®¹ñh/Ê”“g$Æ£\—›äkÕ T´®e:8÷ •ç¦*2oY¡ê•ù ¯n>Ä™”i"‚Ó¨n•jOÀû 5éu½E­Ü÷"Q'˜ù¢GN.êYurVú£hÄ-9s 4œã¥Š'ÊÇC/øØTÑûï¬×r£é©ý*g„ÐŒWñuÒþ4k‰âWÑAók“ç¥ Ñ‚]»î÷V4Z¼¹óòû®´Tìf,­AËü;ù¯ß¡•amóƒÏh•àÛKiCŽÖ¤Ü޲fÿ™Ý=NWrBW©‹ôk3BhÖž ¬ù› ÿ}þûú-»Âé!÷QöŸ¼úÿôóoÝæÞ4½û-îoý·Ÿ–Ç™íÚÞv‡:i–G©éÐäÕPç{Ðú¯}¯¡Ìwч¾„ &|¢K0h{RßçBÚ_Ÿ›p˜P—§É¢ þCiëþÆÒ Ó.‡V­×’aj‘+ŠÍÛ¦ª¿Òz´}ø[§}%C£Ça¦xß‹ §ð?{=³ŠÝü`63¹r1«æ\{ÝÓŸÀ¼·@¨íÖ?9òæ•Ä!÷ý°ôÉÅëwïmXa»YpVŸ _^òZ€5‰9½¥ë°Þ³-smL6­òå—-Âö;L$ÈvzJLo‰„ÕO7 û.ÿsÝs8œ¯ØFÒc©>÷F²r…OC¢3¸Kc©Rü³R W«¶ÿ`AªŒÚ᣷S:,þl²égÜs•¥„9Ái _^NFºJž’‡ï’‘¾›±Îˆ¯ I7L@dÒË1rö¬Fæt†7Ž·ýÍ•lyDgéÄn"Çá°géÜ Èy°É£L×÷Æ&~ËßáBVsËCÈ3NC?»0¼]B±=¹ŠÈטNÒóä œ É÷ÿsÏ)r-î õx<ò5‘F‘D ·Kõ¸Ÿc8fßKz3?ÁZ…®©5Œ105¡ÂzÏ{Â]J¨”ãVºvg UîE=Š€ªAdz鑈‘…‡#§ÚQ­OK{NÕ·tÓÂCP“ÿZ§°Ê ÔÒ’ˆ%úßãÃ' «2ÏËÐøÃ3²_‘£W7u9â$l¼ð èa¹ø_>ü/þ—ÿø0À…¨%ºj ±KÎ’zþÇ®Ñ|~çvy ”¿^ã<7—¥ìk±çc±t9/†!›uló¨äÖŽk^‘c‚œ³«[ðáá‘ã÷ìà½ò¥obMëðúÇS«úÚðJŽÿ»*'¼wà­«¸yÝVçÜ’:¡\ìV[—™ÔX½yFÏy¾¹ÅÎZQ¯CSN²I,i´Ü°ˆ÷„ÖèÒ²{…KЖ‘Ríª©í3)cá t¸™pÚºý=çk·u“¯Ëd€6þý×hL¡ezN?òçch¡’þÔǾ Ío'§( yÃÇ+ÛMZü(Íêª÷ÿÙ›¶¤U’­í1àÝÿ Úê·›2ïA±®£„W"tÞ_}R¢ß$QŒ©ÅÁŸcäÁDЗfnþêv1 œ h­N’…!¶¸û¼êðs¬²ëÚÔcÙg‹ÞH‡qç3Ê/%nÂÔ¾–½ øu€c^ój4ÌI…(&˜Âü´¾W*ÝXl {àÃAË£«8êw~K@Ñ p'¬¿•¿ÑÓ[!×gŒ?C‚ž3Þ!‡(AÙúšE8’‘f$M*!Yp»•qo îzM&{_â8R®Ù:»RÛ õ…Í´§N HK®MR‘½„tç¹I^Ž!ýÌæ•ö»È. þú™—ʺ"+˜<.ÙmlUû>v"gÄÇçùÚ¸·m¬Èýò<±hVPBÞzÚ”šÑäçjü…ËgPàiâÓKwÖqŸMÐN~o6?!ã˜Ï¢·nìyŽbç³ÃÎ|@ ÍêgžëOQòG•Ú¹Ë(Í/³º{ˆeî»u4Æk£œø«ÍÉ3(?ÞßHvÿ6*æwæïQSGåäBû<8Îo¨¾›ñéãåý ¨®ô;/¿¶5b¿€óMRÔœÓñ¯YGm=ÙWÙ½œ¨Ót9ïªR=ê…$ׇWÆ£Á¥©¶°U4òNutQú†ÆžûD¤î0 )·ej’Ž'šnŸr}1Á‡fߟûíK¾ŠæoYƒõƒÐ»NH‡3-b÷µ“7¡ådô'aþ´Š¼˜ôi­¥-Kžž$C몠AËâ}hclíºf-‰6õ¾¶»ÐVCÉP}?Ú¾9.L{‚¯v_¨1(JDͳ,ßÕîâîÄT{—ð“0rþáá!·—ð9ÚšUÁŠwLJäƒ<ò‡dœòöÝH©ÌÊãúgÌíž ¿ß;z¸þa¼z¤¢ö«sÒÇ ßìÓœ€Ôµmõ·G å eÆYH6ý没wR¡^Yª, 2gî)»Íþé?lâÚ„ìîº+‚¢9ÿœÆë„•ãì>?ôo:ÌhþO]²ÿBT›ÅÇ4?Žìø«ˆ¿Å8”Ä­ÜuQ„RÐá2Ö¼¥C¤L5o_þÙ›ãÑÏÿÊ{¨_ŒÕ?„J1ï_$ÞºP•[>zbqªu–éô“vøP—ôã’Í1ørˆø¬i~4Ô×ÔE•¯F;u®øKÐL*¢þaç~QÉ´÷’´Å.†ÉÂ(t$›ï¦4¯ªÓTÇOÀßêÈyEè;ã{ȼð hå0¥þ†Áï2^4Ù ðó^Í£Ëöi0býªñówKSͶSŒ‡ £Êñáoa*á·QK ü¢¿ôk{ ÌvÝZ¶ò‡y7ï¼}°PwíS,ç(,yµŠ;¤[ÀŠ\¾AÖFøÍ}ýà+XgTu;ò+6ºR‰kÀvªÀ‹šŠ ¾¼ŒD”¬ñŒ’8¿*qg@2Óÿ‹×ãp—gð¾‹5ÑHÙ“eùrŒ©c߉×oÔ"­Õþr³k¤[¢s|ð _Ü ½ŠLÍë ¥:©Èjq±ºl69Ôö¦å–¼Á½~¥êt"/±Ù Ý‘ äûu]Gâ@ Rœ¡TìC!ëâzåA™?EX3©b='™§ÃQbã·Šé›e”zÎÎüPî ÊœP -ÝÞ@9϶ŒÏs”:-‚_³¯M~DK¬‹}ˆ–QÒ] W“Њúçºø÷håûèŽüÇL´úÒ°ÿ†wí•o& E sé{‘îÈóØÄ¼tïY$”‹î ½”ý~½‰Üñ¥Wª¢fÖP3“?)¹¼U[^öZPÞ¤ïûÞ a»;\øº“¼hõ«“Èó(ìÑ€A‹KÀÿØ2G$xg£Áã­¡2'4hXÉÈM9„†Ä?ÝU,Âp‡u~Zš+¡Ñ†’ÚÉc¬Èqæ«Õ§‘8ØÐÐNÉ¿rÚˆšC"&áëÂ…#çò¡’';¾ûc[Þõb”‚têùWýà£Ì—ñÈoú½‹Š—7×Þ;šû¤ý™ß6Ÿd-¯ý®k¼”¥ü[ßñðzF¬ØÂ‡g¢Ç×*gÿêµèM­|ÿëm(ËdÉ”3ƒÊ[•’{g³ás~aqñÑr¨¹Õh'ÚuêòÏs=Ñ‚z‘1‡pGhXêk¡| ±TÂH MÎ¼Þ Ç† YAåÚUšhQS­|ã­ÇÛUËÈ íC(]MtX5’4ôAç:ëay¶!øÞé¦÷5ú4ôP—ùÅC_æq¡æ±0`zÏçyj ®Tjt‹„ÂÏAúå„aÄÕÝåZ6Œiê­W*„ˆÏ¦íó2˜âe‹ ¡€ŽÎæ<M˜yëi’÷ æN8­qÄOÀ|©’FÅOXT' ßš<K_žÍ>æò…•s¯öݶ9¿)23¯ØÂZÎÙx·#°¸ª—¹óyoyn}Q[èAµ ù+N.Hh¸ì(ÉD¾¶â.DHœCñ®éÈ}$%]7(ÕF² }×$|SéWRääJ’/.!U8åpCä2RßâhréÜÀ=¹?”Þj3#å‡]"w¸‘^—³ƒbÁ?œ™÷=2~\”óuVBf*…vß%N9Òé/²yEÞ!ÿ͈o›â"â^îÌábzÈ£9ñ#ñ²0òê…Ò„]D>Vfçér.äëË~ö]±ùk˜.›ø÷£Àg»¤¶(8Y©n˜´ŠB <O‰ç£pÌÓ%=Eî—€ûƒN×<Æ4Y íòϨñ²™ó~÷MÔ”Î/˜±šFÍÜA8&6†ZRÄC8Pë…$kM!9jï¦3ÌêFísŸB]â_¢v•¹M¬]*ê0O&ߨ¹ê8úOþžŽ:ÉÁñô‚·Qgh¼…VÅ u9Ÿö„óÔ¢®©‚ígö+xãêÉ9Ö ~Ôü¤4[øó>¤QA³fb¨ O³c¶?åú ~‘!ƒPbíÔÙ‰ÿÂ!×°†>žîºömråƒÏQCv¥î1æm=Èšþè(vü¼ #a½hú^]+:¿ï™Ä–ôõ»­àÝgÉ[™Èó¸HUp>ÊVî«»o^ƒƹ}³‹Äðm«hÜñsÐä›ã\’o-úTï=ö³AëÉz»ÅHhóó¹5Ê­ íåÞ§“(÷B‡þJËg»‘¿}‚í:eŒöÐÆ£Du`“Z~çõPu¯A wÃËßb§¡¹,ê¯r8´9ì}ÑD -ž‹—9.ïy†=êÜÓuyåÿõ´œòŽ\Aq¹ó_óN£‘¡tùîh,Ø­ÕcŒ&¿D„Ÿ¢iûâ›'ï¢ÙÛ.9ƒA4Nš5ÌA ý¯á–! hIöêršQ2ZæUÍK׊¢Õ±ÞýótŒhµA/“ë†ÖáY„ÙrhÃDñxòðC´yx•ÞÛmÉt¿Þ¶Ø©ÊÎ7?£Ï­n:¯!Ô8së­1q îÞ£ú©gFöÕói1Àgö¦‡×¼Ò È49·1º>â£E¿æßÛã¾Ëý¼×QL-e9ü¯þÞ öY¸ê­¹ébÝ”1´*¤ÿVè $ûwW[ÀËÈD‚ â‡ò”?¢€2ÏH˜ç>ÚÑ%ûC:%ÈŽ’á þðOÿ¡Ì‰PŸ|ÿ¹ÿ÷¾ÒöÙîêå+'©ûv8ñHÿÞßöùÅB %{ªºÆ%@ÉPdó[( «Œ€(s±›K¦&ƒr7k$=­P‘IêàTUü)¶AŸïð1½b¿Þ4Ô<¾2Çþz7ÔÝϼ‘õiô%_† áû»8½#Ðz«šY¿Z4¥ßû˜Ù@Û®¼kæ{vîJ?ìk¬¡KHÏý`6%tWsdp>å‚Þ8Á®Ç«SÐߢ˜®~í< :šÆÞ[¸?÷,o?ª³ƒáv›×¯“æ`´ÈÒrZŽÆ«'mVîôÁm‘OÌ\=$ܼ” ³¡¹¹û^œ‡¹ùà•ˆ°\XðÈ×_;} –Xž5Ep%ÀrC¨F&_/¬fÈž}k úŒÖ?`£„A'û'lk Œd‰©!ÁÚaºå$º\Kø°IDí¯M Iá¡Òׄ¸‹¥vóºGRzl X4("õ^)-¦øI¤¥Î䌽ŽtÉ?C§»úá”ÈzhG2OdQLú…,ß¿Ñ*^DöW7(mcÜëýµäÀm䉉8|aïäóþR €AŽoÅ5q_mAYia-ŠØŸ5¿¤%‚bÊÊÄA—N¡„±Z — ýNn5vzÑa‡2DgŒ:ŽÙ£ÜnBÿkÏ÷¢‚ç×ϵ†¥¨´(I¹\žÔʵŽJ@øzeñØc¨Vt^m 5lâ‚WÒ;sFÛÏ{a^‹Ô¼áû(,Ø ÒÝX¼n­dÚ¶ Kñç<5€ñXéî—°ú&´Ùô—1¬éœI,Ru‚õÉNÕà­iØÌv÷ŠûqØÎ´Ž •ºƒ£*²õ'#…C59)«ýº4÷Iý]ŒïEÒÑŠö‚%^$w&qï?¤‰"’mHY"LMY_„»#MÔóÆ?v6M¸i1fù~cÒéýæÈôGÏG8#ÃÍÕÚ«šíÈXìvä72Ó7óJ~SA–ƒ7Î?im—ã‡Ü dOdçï9ŠÏWiúð W‹JâË Èíø^³îÛŽn|³+•Ø[‘÷Œ­~ÖäspË]HCþîQƒŽgPpç[ ɼBBü¯ŸÜ݇Âí}ÙoèyPTϘk%l ÷ûd ‡SN XÊ•HØ”À[²rñ#¢(qä>‡è•&”¸ýYŸž&¥ø~JZë¡Ôhà}ƒJg3UI¸0¡ÌÃ9jvO]” œJõ-ÔC9¿â/¿¢|ÿ£ªO¯PÑÅ[e[¨•Y’TÍúQeÌþêµO[¨Úq ' ë»xrk¸ÙQí÷}»UIÔ`#½ç—ò5 øŽœïC­›±É OQ»ä×KŗĨsªJHEÖu¶=wå+ î½ÉÏTÔc8ñš¡þê…{Fkǵ¢ÞfþÅ0ê;_×j,FýŠˆqÿ„"4`7(>¥ŽÎ^_Ýì¢Ñà}è ûWh°p!z×ç4”ØïFËq O% œ´JÇ[5Ó—‰Þ£á£cÇBO!ûœ±¿j+¬°~‰j‚V#í=¡Á]ðe„;—Õi*]o“І¢³RIœOç6ÝvŽVoȽieñ›´>œ´›Z’ÞÉ£ýŠéá¹Òè[9x•QqÁñëþ¿õí¹ï:„¤ªÇëýÕka¤‹>gv”.æL o-îø-QŒTÕ\Í8| ªƒ+¤v|´6Y×aŒŒ¾lfç¦ùÆBýÊf`Áyh˜7»_öÄ«Jæ§WBS”…aÏ© h~fÖ{àB)´ä³—$.BëÆjí÷´FhÎôäH¡ƒNV“¦˜g½ÐÕ»ûÞfè8t/7vĨދ»¦¦NCÿsi×™M”“µ—LÚ†¡fÑgÁ*10úíý»Á½0j¡¿%\¯㟦ÆÉa’~3Y<‡¦‰Í$Ãæ©à—Tôa˜ýê*Ya>ó‚„ŸŸx ÏÌ¿”؃%þWì c`¹äâ+#WX=¼JÄ×kÄw¥öPÕÂúzv_B:ØÌb(r¦ †í¼x+‰¬ÛH0Û]oÓˆDŒ~2/æ‘Ø`}Ù«dGÏ11Z\ñ=;ù4m”\ö8’ß^¨µñ¾€ ´0”ó©ÈNZÎ~ÀÝË¡Í41u¸‡ëÊRål'ÒæÐÊâFºÚ®ìàïÈ@oòhÜC ±Þåtf.2…~Í™¼êŠ, ‘có?‘uœ†«`(r°iÞy}ñrùëö>¾y ¹û´SC^!Ïü~áa©ƒÈ[)𛵱ù‡TŽ6œG~¯¬÷?ÌäPÀG©¤±á0 ¦}zàƒû¶ß7—:£ðµÐW?¥ê$ù|õîóäôõ„<`t›®!‚Å >dV=DÉûÙi¾'Þ£ÔOÞÖ'%Á(sŽÖ)°®å$yKˆ•§PAÊVýú'T:ÿEíé ª4ï ­6,F±Ë‰LE5n–䱇ó¨N×Ãâη€4u¼ ü¨ñ>&HÞwG§Õ R¨ù ÞULµŽ¹\9ô8µºghDmc欅 ¨G©ýèõÔaómd D‹¦›ý‚d¨S÷îNêÒ‡D™O¡®Õ†Ž˜–#êFö[Hù:£î·ƒì‹uÑŸ)ŒKêÆgÔ;m§1wÉê/N¾ë=r3^ j Î'/܆²á©þBN(¾ÒvNcÞòFyì¦öBn±ÈÌ*úCΛ_‹ ›;jdKð!dé“G¯J{ÀÛûª.>Löþ¡O—!½é–÷=xw:ÏáÑK~È#¿ŸI¡ e屯w¥×Au«ÌÛ†/uPÿ^Ùi(ë4iWÝüynZ¨ä÷hý|­âE'…Äd ÍÜHãĺ4´?Ö&o ‡Ž}ϦÜŽÿ3Ú.Ç87Ím\ºLý“ÐJzÃ=°å(´Èž‹`Pí€æ5+k#hÙ¥VõH¶ ZN_{Â}¡Zº¬(ž ^‚VOïß}…¯¡MçvÍåûâæà8¥¨ƒŽo÷~?nM„®Ìvw…beèn˜\O·†^[›Äá'ýÏ„-#˜ÇaP2sù^ƒ$ ÍÐUXMÃðT±÷ìS“÷àaÍ…‰jÿ»)Ûûaú™Ñó‘ ˜µ²>{k̉ßKb±‚…ï3'Æ]ïÀÒhsí ³¯°Ê‘|h¤Ö|Ç"ÜÄaSêk¾óE$àÔ"ì´=Š„Ÿ›gN¥Í!ñƒˆ{‰5§‘ôŠzÚ¥ê'Hî«X¬ê‰oÝ”nÆÝ4ѼbI|Hó4Ä‘—ñ8ÒfÄÞ UEzŠŸŠÇ\‘!¯H¡n[™ÜäÙáu:²è…&R\F6ƒífû€Vä8ú ÷YäJS#¼¨‹Ü[ÝlZü‘È«¤Af—¶gGgÙº{ÔÒP`×1±¹vü(L*~æ0 Å%^”D‘7ÝíŸlsqÿDȪA´Škô¸o©Œ’ÖÊE.BÕ(õô^§ Ê*sVø¡l¢Øç ¹r”?¬¯0pù>* éø°%˜¡ò?.ðpƒÚq4ôj½Ùo2y Z=Ëk½Ç§PƒO›F1’ 5Ý\û(ä·P«2î¢Ý•§¨cÈÆÙ[e…z|ÏMP4'³b¸ ‡íIØúñPW,¹­Dš$.ïNÝ‹¦¯ãèNüXB³ Ú)¶YC4·R,~¼<Œìߎqý@‹¶¢Ê«;÷Ë QÕóõhµ_ïíÑ3ŠhUÕÚlKÖV½¼Œ5ZhÝ]XöãÈ"ÚØ4d}ý©6uÖ‡¾¨1&¯–úÁ»|Úš$W¹ßSqÖÔ÷†ŒÉºwoÖ!Í=‘|fFR,"9cuÕáeÅ«¾=ÁBð¢Óã±Mq%$/ùü¼œí­ê×\¶åáÝóÛQÞ‚³mœ`Íñóý?óKkýB$Ù­þóüÒŠ ülENk‘o…ù^(ºã½Î+PÔÓö`ó;>ünÐ@%ñ"NFžP*}‹Ý¿>Êv+x8/…²M/sžf¨y°½e& •÷CkÓž4Âg ñãWf¦¡zxLT(j[-¿½=_ƒ"Jå·¡ w‰† ‰ºßâY?4ç5j¿ ­—¯Ÿ»Ðg ín¼¯^«~…Î@ó…óäYÐM{¡$f‹z¿ê–»ýŽ…»•~ÂÀ˜[éiEz 蚥´†­‰Â"]ïÁèÁE“D­0®IÈ,Œ&®ŠT¾„éAÁWïÀ¬ÒØåß'`î™Z@Šƒ.,°.w~G‹/•>˜‚eçÃÊ/YSaUƒ¾ÑrÍÖÔÓ¾“ÀÆ…xöF•`Øê}6uˆU ª¿žÛåCDÚí*ÅWx’âÓbP9’æ¥sR÷œFò¢Ý­AìMHI¦¿H”{w'fv@Üs×r:&U é¬?féI}FꛂKÈD©Íþn[YN˜µ½qÙY\×.Š!ûò%ûGâkÁ×þ›È»Õf¦ëè‹„Oz,ùÓpßÁ™Êlx‹Â•®/¸ãipÿã¼ ëª(žb9þQ%dzr|¼Q:éÉ£éâO(›l˜2Øü¾%Dx7¡Òe³7q«¨Òåö¶j¾Aåè*ÚT»tþi²i*jPpž±DÍâÆÒ°ñeÔÈÊðqA]f²•°(Ôg¢Ó}ãù ºî= B£pª3–E›h,N>}]öš-’»5„¦ÏØ#ö¡™Ek”î‹.4Û®Kz6ˆæÏÛz9ö—¡…Ò,ùÁ÷^hQCÁOB,Œ–FÕŒÁhYž‘s­”Ї#¼Ð꙲; †å[Fdk}D3_&š=ÈT#!åˆ%£Å{Û¡ƒgÔùí÷|hŒ’|¬9sªk® ÁÒ¨t;èóèõ2”}}Éæš ¥Ä—@aâþmŠÂÈ¿ÓÕ>}Zò¯n³S «„I ÛcVa-!òÃ~ $ÛïäσFס:·³4“×tãØÆûü¡%ª\ž- :’Ã$•n&Cw‹øž„Vèç»´?ƒuú#ÑüM0\µgtÕLÆTljbyaøáKdÀTI¬¾©ŽªÆ/cFòûP|–>—>8~ñÕä±"5Ô"+ðôWSGmá“·÷K£ŽkÀÔæ½G¨›på{į)ԛ̻¥ªÆŽúßôúSQ ÜûãK/РA7îŠìO4´$óä¡Ô@ï\¬²î×ÑHW¤Œˆâ9}Ì¿tL¿ ™|w¶‚Ÿ¡±÷Vi]µ(Wf(Ú¢ ÕY%$£‰Åhô¯V#4‰r—9]‚&ÎVì1ôgSÙÓ<14¿sÊ9ïrö~ÒyJ nÍ-û)îóP7t« Z~‘ïö¸2B-™M}”]-¬áÜÂà∕Kð¹k¤ÂUÃr+Ò‹š²ÏCŽÎù]ñ† È:õ{¸´C2ú©^ñž4ù¯f?oÛëLi‡\ï£ØÐ§ø¯¥ÔÎ¥¬P6R\ÿ ¦Ó…ÝE›êúUâhr !¢ø¡Ê«ÏÐi–ë¿ -÷4×|ú µŠTè䦴g?ìι Bv3Ýk)бvŽK³\:w=WåÈ¡›Qf«¬p zôO‡Hè0@oqøðüu]èW¼k´'8ã½ÒAÞ0øÓwÈîÝ'ø®ú)—|R~Œg®z{kÂO¿äKá\0Nðuíª6%L¤In ½ SÇûólû `FäÏðU>s˜%ùYj7¦s³ÍQ®9ݰÈpÜ´á?,+ùz#Ÿ•û-FßJaõÏ;Ê[Í~°T|–ß¹6™Ôä½µÎÂVúy{·¯fðûèǺît%øÃ,1o¾° »+¢ ÊÁߥS~¯”‘à¾B2ßAÂŒ¡==$]TçxŠ$RÊlAþHúÈÏjV.ÉR¯W ŸGŠ¢³f•JHu«´®ÉôÒœÞêÊuz†taŒŸ_Å#ƒ»ÇäCÆ;$ÃîbÈ”ÂÄø / sãû·‹ ÈJq,ÿºÐ²=Pêc•†NwsÎ i"gÈçÀoVÈ52Þm0y•>Ǧ‰#ß«çM¥dÈñð=Þ( ÎØ?~ŠE,ÏDà!É}ŸøÂ…<\"¾õ…›´<4­8PÔõj]¿%Š34%ß E‰~µSŽ‚(áDlT—„2ü‚‚Ü4gP6éë4‘Ú$ÊÍÞ½Õ’g‡O©ÉšJÉ¡"ï÷.ÆdTæ`SmïÞÂ#Fc?Ž~Aµ„KIëéÇQcMÁë¹² jR^Q:®Zäׄn=Fm¿œ½ [¨]á±0TR:Ç{ªîC4êìñº\‘¸Žº‘Z¼“—Q9T­ú§êEÿXúEŒúÄN„e媨ïóø/I+ê-°ÃR4«"ðU Aƒë:þ·­Ð ‰®ØÐÌ i…šz)ÑÐTó»`à6ÞP2u ÒGý_;&« ÿ—­ƒ¯}>}š. m)>øÕÁSÚã…vPþLï÷ í¾oûw•A hs}7r( ¨“Þ¾8ò'ÎOš Á§dûŽ:’ëû~Uωà•|(!N´éioó_ɧ7ÀÇ–ÔÍeÑ(|mlOvª#S9ì@ã“•ìw¡Íšò­©é¾ßë}CÉwºÚ<|8Þ@÷ïÒ¼µ]Ð+Ìóo¤_úÎ<á“°ýÄú¤œîßþÛSØ'xúëM èå´å{2 =Œ†Ä¿[j Ûˆòš”…tMÌÁƒxjè¦;äòü)t;åi®£„î¢å'uVУyºÿ‰øCè%jÑY ¹}Ô?NK®„~'î×?lá›Êò;÷”0¤wá³ân2|Ïq2$â½£Jbý:uágÙáõOGaübó4Õs˜i ¯*:ÞÒ+°õx`“Y¯~OÎÅå÷Án¸’ÔâÂü»r2òÖí6$tz«êÔý‰Ù{n¿-ØC’…+VäelH6iEÌ_ˆ”T¥ lù’Hí:éz¹ªiGï>\øõŒå§Ö÷±¢ÞÇáÝ2›ß/$ F–];}ßÐQdk+ tcÝAŽºîW/Nu!ç÷òûý¿•‘›ñô £%ä=µÞ½3ˆ|KN‚÷ìP@/eus² +áònD- ú­õí¾1Šh°ÛýÙ¨F1]ÊàGÅÆ(q)ŽS<§ ¥ª†x "P–i#èÁÊÙ~[R.”¯L@õÓœ¨à°À½—J {çÚgØPùûž#Oû7Tt<áépšÙÿ”Ó¦ÛÁ—‹'hª’B»­Ëõ©QèˆQ à….é ³2ÔO {LÓ5òc$ôvR^ÿ›§ ýß]Þ:;Á cm/û±IøN°ÊëG ?Fo%üZ΃Ÿ"&]\€q&ëþŽÔr˜èÍò R€©²Xckë:˜©8¢q`vU/ð¹†,í½â?ì K= çžë?‡£ßÔå2°Z¥pºèw¬Û+°¯’©À&±˜ükº‹°50ä­åè¿Ü B…$`—RIÀ±F þ:¯ÌÈ"ÁSl8°ƒDô‡²ïÌò qFáûG /Ôûé¬3í$?n‘Y»ˆ”ñOk?½~†4‚‡Ï'Ê"ÝžuËïhd$LºŸ‰LŸ‡3býå‘%e–GÌù²ýªÌl"Åw´Û^™Î"—ý“Ä?¦ÈûEÝV÷ò‡÷u]CÁY¿РPcjUŠJ_kÙXEñ5eÝã!¬(µüõgš1ÊF4ŒCyíÆ!ž¨OÛI®¡RTÇUÛ”1TÙ¼d¥ÍŠjq-¹P#]ÿO9jŠ9ö°D­’—’MΊ¨ã-Óqš£õÄ{]'HDý¹Àssü«hXôxDR” ¯;ìv³Ö£©¬Ã@°±šuU¥Ó£ÅÉ[ùY:•héç¦à7†GÅQ‚‹ ö*{â3€V×r «~EkfÆ®±öhäÓu¡`mDEÕs~^C›ä‡níÕÑ–½'¦ôÛ3´ Ñ]ctÄûbÊ„¼¨ÐRoÚßÖ%yr]Ö©¨œ‘à½ì¯ðÇЧÅ(œ“g $«5åDûýhsÆ~}jÛ?†w@µÛK-ù? ’¾‹öÀߤ}NÞôsØ@Q‹žJ ÁÜæÅøÔ„¼†¼êÀ=ÿGmûœ5ÜvyU^7ç]oBÃØs×·6ÐN¿æ €EÜ•f&7A£[go”u¡ñ1š '|÷saÒÓ/ÜthÂúCü#šœ¼¨mÁÿ M2Þoé,C“…˜•ZÒ4c«ÍsœÄwi'а:ãšý÷¿ÈIâäN>¯x*YÍc† ë‡8Ô•hž[Ö‰¶U‡šk´:jnûúô¨ÍëK•¾Ú¡;ŸN‹V_¦„Ü;a¯Äà£wÔšÔ·fÈ *îlŠ´† â£MûÞÿÕ¬zæñ`WøÔ}ò£äÈÀ¼–¸gúÄV'ìs=1Ì—ËÕ[iT;³õP;¹œ70õ‘ÙBxó4^gÔ¸>>Í)Î_À¾Ü£i’𛇶³Mô/›óà+Ù¸Öp t ™Of±‡N§@2õUsè:ãÒSÜ>ÝTƒ¾…ž¹ãåVÐgUQã‘Öý¿Lº{/Á·ÖÐÓÚïÓa˜ÈéŠÀ] [_%Ø ‡Ñ–CO`Læ²áÃk0Þðs­À¤Ÿçéæ0-¢ÞJþ#vŸÿúÎ[ôf0Û¥5/{á%Ì/N >݆%Í%u¡k_ay1¦Í‰yVÖ 4U®ÀZwÃû©.Øð$*rW:›;Y*A°ý¶Ûá¹aì8ëUð ¶Ã®¬Fb±êü–ú|g ø#ƒ$}T‘PûKbÔŒ×ÝyŽÄU½JôŽ4HÊ«VÑ&]‡dñβÂ=>Hq¤Îå¥RRþfQÐMÛ÷©CžÝv=‘v6ÓgÙÐ([ÆØÏ‡‚Û?(¿0!“rÏ£k´¯‘ÙâVÏÙOȾ™GÓéŠlDåOWr½‘½•P‡_ÍÌ‹œ%• G.¸Õá*׊ðÍϾ :z'‹¤z˜¡ëQD˼ýtוT¾¼=¿WÊò Ocþg··Ï~.\à®;n}5Wvõ4·ÿwž?‹>\zÙOHè7[B7ý³‹5ÂÐíúþ„GtmƉT5·@7 ³ï®t›ÏûsØ@÷ó‡ÜKd¡‡Îá³þósÐSI[yÆ¿z‹½NsSò@?Vò, 4PõÝU€Á’UÝ‹«ðÂe‰'µ ~¼}–Å®? J]v]¢alÍEžªR&ù&Â4qi¤Jü:ÍCèôÙæéop,…ÅU#ÙŽlkXagNyÍõV'­Ë(`}:Ëê»lQи &WÀo³¹˜×»7àÏ`³Î!?#ø[ Ånn9‚‹çÂ-þÚ"Ñ›~oRb‘ädæŸøœ§HfvÎ`w<)<Ã_íÄ]Bª,ù?V8¼¨¤¢é;›N6Ù"£ãíÍ«3&È4a§DÚŽ,ü’ØÍjÒ&÷s+r@ê©~“ä4céWްC®À'L1ç."ÏWº¨]¼È—rÉsþôkäß{öf$b¸+Å~ä [ËͰ Uý@–Ÿ€¢ß”%=ü@ñÝwêTÍ,(¥ùI7+seÞzøÍ'@9ÆS/Þ• £| ٱυ¨À˜å!T‹ŠMŸÞ–9½BåØü#‚ã‘k8’5jI,Õ]^RŸPÓC1±ÿj¥UŒ§0¡ö|Æ•˜"ÔÕ"Q% ç@½º²`Ëïhpåm€Ê!"4r*ÊÕÌF+º÷ÍwüÐL‹~òf÷ZkôT;¡¥sW!ÿ×F<*sÛ‰V þqؘ0ÓE«²îJƒl´¾4zœ÷ Úð™Å[•0¢M w2óM´u4ަ2cEÛ)%÷ž®ãhwF½âDÚÌh•顽Õ%K;ž ´/n–.ûEŠ—ƒ¹¸ß§#|‰s™/¶DUý‰Œ¯¯abyFx¤Ñ ê.ñ¤¨@éÁ1&¦AÈŸ~é Ÿrx¿yÄÀG–v¡}Ý+á™óɃºN¼Œw¨áý6+|~’3E·BÉ_Â; 您vHò`©ÎSÝÿ^8ª8U>o2¿Œ¾þÿÎ/vY¸íûV%¦7™ûœ¿9ý®;Š7 î¼^/Þ¿þ°\”÷0”Ú}œä9vj_C+µl:þß¾óSWÿ¯^ï»:¿?/2þÌ%CÙjpòA£û9Qê,åá.¨àÈæd܆J2ðéw‚*‹å‹kÍtûsæ'#]<Ôg0l‹sCíYËŒEÕJ¨ûrêc²ê"4„µ¼ŠÈf€&çwßLS4¡%Mó~5­´™©••´ CÛ3ï„Îdºðû3Ÿ Û‚Ä<ãVô*¬˜YŠA¿Š×ïŽDðíûll} «òÞ<‘ýFö›¸}ÞÀèf%ýÇ${¢³Äv&ltí6Á”º‚Ù a˜9â¡B~fOP‘}®ƒùO½½MVv°¤ðlN£!–7>=íÙŸc«^¾Û:œ°¶ù·T0ðl¼ú-G¢ [’)¤Šôá÷–aÕ‡Vð'髨¸a+üeãØ YE‚ãy|×¹pà¿!k{{ƈ½ôER¾¢4Erò˜8*¤”þùÄ#¬©SòC热]´É02¹íþ|‚LnJÄ/j."‹Ö#þÄÂFdó•ýCߎ+É2 œ†ó¢x”ý§è!CSòÓI»¸?½Œ¿kzÞÐ]D!A›–—Ú7P$ÎO7À¼Å->OtR,¢Ô¾œ†TžBY¹9óò|”Ûà¼q÷ÿØø¤$‚J\]Gv­¢Ê•ØéR:T­¯LJÎrFÒÿÎUÙ –=Žþ±ŠZÞÖNG²£Î¨Q=ƒ‡¨Û}ñÐÍ'.¨=Ò1Rlކ.Lþ¾‡£1¯ðo¾ø{hÒýÞ(«ï:š]¹ÚÙÔT$BD¯ÐR?§øÉ´üû!’.Vf‘F‡©·£ÕÑšrFÃh5/¿vóŠ Z_ï :[†6Ä_]ã³½Ñ&ø m›3ÚlÌ〉Ú M¢ˆÇ»/"fâyè«~ »ÏÈ=²R4–¬€ÆJïò•¥¡÷˜êø§#ÐþëÌH);4$É?9¹.µ2A_ÜÓ¡jœzÄ4âÁ~Ù×'å@ÉOÆ22P8ažÆÝñ·¹}ŠG4é !‘´fñö(øMD(Ê\•ù±-gÈH ž{˜[Zû_4Ž´@çB_'‰ôFÞ6:û¾™=Z‹×€ïVw]¯ÂOZFÑšjÏ”Sk,Š‡É “‡ª0ܺ¿nÂtíàêï›=0½åü:¼Ïfìe––ha¦MnDçFüн.T„ù0›òqðÁw€¹6Nî &E˜9µ Æ ‹\œë÷ºa)2¡x¡Ü –&d‰^ÀJì\qI¬¾U²6€µÆ#r™§`ƒ!£­ç¥4lz·ìî*´ÀÖ8‰î[ ºý¾Oó,w:ÄH>ûr˜Â½xj[Ý(ü“²gൎB‚¢.Û¼ûH8çñBq‰•2¶9Åê$ øÐy$à©¶!¯AòÏž*«HéúÙ0RÅ©d4dBZò15¢€V¤§`¹°î³Š Q=qmç‘ñÝ…Ø5»VdªžÖè+)Bæ%Bÿ$þSÈ*wáÞ?d{xÈNŇ9…ÙSïLàû+g¶®å#gq[xm.r½7ôR|y$tf{)â·Øõ•qú)ä+C˜Ëâ@þœØ–ºx¤¼už¤ ðP°Ë{.­yfÍŽE‘´>º¿<(F5;);)Žâªö⦠(áe ZûJ%˨â)N™¡´Ð¹ç“‹‡PzÞú‹ª4Ê–±>txåÊ#˜­mQ~€vˆ:7¨hñ=CÅã‡^§o¢’®ÕO¥~T~Üۯ౎G´¡0Æ‘?hZunÂNyŒªøJ>t‰¾J¹u!š"LÍNCõ\o¤—6”¼‡qZþ(rܸ}Fâ4äû®<Ê|Ÿ.[^x²­p‘LødD_kè5>é2Ï7â÷yúÿÕÌG¶P—yø¤z•'f:ð?^‹õÌnu4IBù;~7 ‰}ZûûâÉe¨YÏ÷q&Lºè¤ }Zhð 2°Ê‡¦k2¶÷ÍM Å’åìæÏLh!¶ ›‚¶ZŽ\Fºvøê®H×OEQœºy±ÐY}ÿSÁd t}3Xm½³ïW5r£D?@ï…MÕ:©.è[œ=íæ…±Ò’a°…X]ƒ­¾ «¯<½×?†( ó4ÂO7Î×Næ³06ñ+<·Á&‚׸?ÎØÁ”ðÒgo†`zæÃtªúIøU\bL,}rŸsÏüEÓXdùóÁ¼ –­-½³a¥@çAê X“èÏ|±ë…-`µ?7Mœ‰™DµakasÞeš~¿s´Ú˜.X=ùi{.۴۞ÿK­ÉŠYû¾ôÛ*wô]@"𮯳3ñHlsI†B)I>’ŠçHl"Ù!ºñ¨o$/å9°€”~Þ憎gÚ¬»û÷ða¤=j NºÎƒô§-|ÏLi#CÍüÏ|ïldlyF«9´Ïs¿#¥À%id!}ðéõõ>d „¸&Yd×ìí9zؾp”cÁƒo6kÞÛFîɧéO”!ïØ C—‰ò5Y¾±[‚üÏ/0Ì?QAð¯zy|Üß·#冇z­’K]þ¢°ãq³ƒ—P”ß®Èö~Šs~˜ >ï…’¢¡ž÷¾ä 4áÈ$¢ÌëUÂñl(ÛHv’zÑ嶬údãgñ°StãÈÊ *ü³Ý–ÈG¥Émå x„€‰«²)Õ4¸{-¼ÒQ#|úË[ÏWû:éûm5WËžG«´£6s­ÃgÔî_ª~±ƒ:1Kõ‡D£®ŽKà#qÔYªN¯®G½Ð>·R²FÔgÞôN—@ý¿醻Ѐµ3ù¼93Ü &û;A„«d êÑÎhhõÏ„gˆ 3¿½í‹†{9ùÇßMhtxkSUÅßÈå'›¡t@„ÕLÓ rO&;øA~fí•åÇ!/æÜ3/È• ¾™÷üä8mk:§ÅîsÊùÌ|ß§ud,ƒ÷ê¬BÏT «ží|ð,ÝÿÞksoõ2T –/}Í.‚FÆ?ÆÞּЗÆ\œ.¯7º¬î¦A×ÑÏ¢UŒ—¡ûQ‘íƒnzèi« ì{tp?VxQþ}e±JçO~¾bóÞ–ÿÛ'‘P6 ½,¾Š¿‹ G´_|ºøtûé˜óƒn -†ß÷ ûÀ\É1èÖºÅãýº¯Š¤qtÆìCOÐQÕq/.è=µtfÀçô¥/{f¬WÀ€µeöÎ{”õïN‡0ìá§&K#º¢þVa´[‡á° &ŒQH_Z„ ËÅ1Ñ0˜:óoÛ hßgù¨åG?„9CÖ.Œa°ÈÆî‘ÿ­ –Ãå8ß&ÃêÙ0nÆ X÷Š¿ïç7obƒq!lžÿ¢{þȸ88næÃÞw"7ß„(w鉑#Ñ!·“¯c%xh,E*…I?‡=`x€äUVb#WKr÷Vù_^¤ñ¾3q7– ééÅÿ¤nî úú­’%È(PÑMÉ…,œqoúeM‘u€Y†Üö5²ɱʶã2ïCÏàÁñòÔÁé;È£,jöá$ò9?dýl8üŸï7ÈR  ä…½Âk<Ôð²T¦M…ÃTëh/ÝBÑó'ûUýQüNŽëìO”¬œËVwCŽÞ•¶S*(Ûº¦Mé+ƒò ÞaÒK6x¸Å'³$ò&*ÞöìþÐÆÊ&çˆIÍâi¾Jϗ¨fyôð±ƒQ£î§Î9-Ô«{ÅÓ|µÎý©_’zŠÚ™&:ýûz£óO…ÛjÈõ‚£_ŽKDƒÃ®£ÇK‡ÐˆEéø‡óÑ„âôU~g4]o±‰VFóüSÑEíWÐR~‘ŠÐH’1’ÞÁ£íg#¶ýÑ*²Oèæðs´Ö^ÉYq_Gë…À.W÷<´‰¼wäÝ™D´¤ 6x¶ÙÇo)ôû¡LŸöG¯h—Ú²+ªƒölÁ/ ýõwOSÕ£¿ø=%+To¹HNþViø²vù$aâñxî×­P;9FqÕJ܆DƒîC>o¼ð‡„,È™9dМó²åíbÚ¢ÿ§{×*ou@ÚÅľËÌû¼ÑF½\xï.Üœn‚¤G|ÒTLIønÊÞcÙ’ɯ%<ùß9E÷¼o¼ðñeì•¿M8òžöO:2Ü*Yñ~ źÂl~Îíëïá P<%ó_˜ï0…ýï\âë™(ý@=rº”ð¿ZÆóYp‚ç”™58rHîçÅþ,qÁsP6¸óiæÔ(¿UðèÑzTXßÑ9´á•ã/~úžlƒªlÖþË<ÃP¢üâ2[ Ôô+p3Ò(AdH¿êè"Ô÷¨õæ—Cã[“ âÐ<ÊCV}Õ Z/åÚ΄¯Š“«¾ºåÐy K`¢þ+t•6a³Åèyþ;ñ4ô=ñ»AD”óÞLbÙ0t¾‰*uNF8#—,g#`ôF½Ô£E{cÉÉ¡½Ãã½Äü†õž0YÚƒÅ~÷a:ÿ®”{ØøÕÝtYÚædÓ_ðU‡ÅØ?gÇõmöù=Ë;ÛóVé>Þ½ò0ÖâšÝl»`C©Y j‡¶È¼“9Ù¥à7iKáÊÚ5ø£~ZÎûëØûÔ»Ç=Ì‹²Âê{ï‘0±:–JëK1ߨ6B’ž áçŸ"Yf¯ã[6a¤¨•:An ©E_(ø#í£ÔÔoD‘ጜ„ý‘kÈDöÈù-C2Wâd¡GÖí7ÓFöÏ‘Ã4z”À Ðd¼¹ôb_/)ë[M£‘ïéàϹ¾(pyÀR‡¸ÅÝm‹ª;‰Âÿ.?åO-C±L…;ø¨%Ÿ¸ý±Ð E™L<ÒS™‚r¡¢gd‡ûñ°¯87‡W/*¦÷ˆ£òvÁ‹H÷xTU›yÅBê÷؉9oE>—ûOµ¸Ÿ—PTl£vò@€=ê^7ÿ,ÌŽúJ:Š]–hðï µHÌU¯oœç5E“€µ’ÁãehÆ%c¾›@†æŸb­_U_EKÎ?K7âœÐ²‡…b•æ M>y™º­Äÿ×k‰ÿq›{`÷ã^ð9Èõˆ¾wúâ)(Ò¸#Ö²ñeì‰ÔP¯¦y¦æ¥8|yûüÍ7}«ý~nb“¼Ö»ý Ûâ×0@6ßÁ}ƾ„m=·„Qiÿæú{[=ùÑÛ&Šê}¦Àà IÜ9˜JŠ®È{¡ SÛ„o•,‚`úÒ÷P ö 0Cxͧ»+u?'Šß_o Øïë¿·„köõ)’Þl–æ~{{¡¾ bnÒs°x?XiÊ`™UúZ…ôQXþûue²!V ¤JÿÖ ÁwÍ¥`FXw!ËT¸@™JÔŠ&k°Å®„Ó¯aûåóK‡®ŽÀ޹píØ•õuxdbM· ú?"¡Nçs$ :IgZ‚DM "n¤©H"?LçsÿÞ¾ž ¯m°|Gr½£ÿ&)Ž"Åô„Ë çy¤úÄ/øi"’­¿¨"ÝÊåI—yd°ÜžûKv=Ú³4>G¦`¥âHbdþðýzÖ•;È2ežþø9²©zßÐóEö>Û-[x@ꈤªCrzë G!—¶¯îsqNä®m=ØŒ¼æ$f1Ȫó›?Cù¶ò¶½EZ—¯ÌÃíxHNÉ!©Ù…Jîþl%£E]Ñ2¦(šÈ+òÂá)Š}œ÷ =Œâ»c\KÜ(i™zJbÊ¥Êók¾²Î¢tÑä–›a)ÊÆ±mé&G¹¬`¶?ât(?A­œ~Ùä*{Gô=ß—wP÷TåK1GÔcklîE½ÒËÊ/® ¾íR…6U%êO))Ï¿Fƒ3+WÞQ-¢ÁÄ”»Û5B4´[ÐhBà W¥$ŠP4âÑgzu:.oŠÕ,Í¢Qƒ…àÇÅxåùJ¹³M¨í?xjÝŠóÿò¡ÚÙNãyhørüø„ÍTÙvqºA %ÛÏ’þ8]ϛ⠪ף o×dg‘0rŠ†Ž |zØA_Q ½£—Í7ÒeÙSc å™÷ü°t¤Pé5áðƒÌ³tã–÷PàA.ªy–ª›½œÉ¡¡S£ÛD| ZU®˜쾆ŽS¤óWÔÔ¡‹ùGø“/4ÐíTðÞ*zŸ©Xc†ÞÉG#o‰Í ï…š‡—.ôÝJô,NƒÞÅ“§åÖ²¡—ñ–fæûèQxSá–³Ýá£ÊA¡/þ— w$Ä þ÷¾šc–µcЭìÆßT5ÝöµòºËóЮr1Ë_ z”c \½÷sæ•sO:÷s¨QéýÜ~Ð?Šê}Å—á[ÊÖñ‰ÍH*:®a_¤ #D³iê©0êÖR{æ×2üÜN±‡ñƲ~•0Ù6GX3etç2aöv6¡T³,™Œ~A* ËÛeÏÎÀÊRi M8¬M߶߾뗤u‚m̭°ºäó(ˆöïˆÏ­  ªu1¦¿³{RƒÄÖæU/ž )÷ëÎv?xo .q¡°Hk`œ¼Û>_1!ôC(NW~«åw=J¢0ûÒk(crþ ÇE”½wkö‚ˆ Ê-v½O“_ÆÃ¡R¿-lO£¢ 'ù3LEe"þßÒ®~¨2N˜Àm]†ª#ŸòÜFS˼ªˆS©’­îçQKF%«¯€µÏÍ1p1“¡N¹7AJjêûdê°'ê/qGZç£a[ÁÛ„—÷и¢Æèš„.š¦…Õ¦\õCsχg_ Å¢%¥ËhM?sÁáFlѰ°á6{¯A–ÍÒ S#`åó[_Ÿ¬)ņŸàõŽœw—Zaóµ¢&Y3l¿ºýäƒSìôªlÚ1žÙ¢¢&vø÷çpÚ«¿nHxެՇ‰ªÏ¸}âC’{.¶ÚovÌŠC6µR)ì¯ÉWX #UzAvPmÒZZT†/û#'‰õ§ÇȘ=“—ì6‹ÌW¿‰Æ\´GÖÛ´ ÏÞCö–S«ÁƒÂÈiú@ùÕÇ*äf ‹¸Ûqù¸ïz]iöAþ•…/Ó¿UðeIGˆ¾ Ÿ¾:ÙF€byBÙloPbâý½ƒoQz=µw›ø'Ê :ä¨+áa*Ý[/Bö9í#J“bAåk‡I„ñH.ãÓP¯TÛüGÖ¬iŽ­÷ƒ˜çj~e¿¶<¸¯;¦ÿþ¥ ®Ô›Ð¢.ÔJ¯Ÿ¡@ƒw̉)†çÑèdÍJi2=špjI’-î¢iÕ4§Èù4wàyÇrÛ-F¢Œu-ÅÑòÃͰްWxÔIGÅs°­5“ ¼¶ÑêÍåUêÛh-÷r4·Ë ­K¹ßž7´D5ó“²h“˯ù'mù)a‹îUk]m'EãK*ù’ÍKå ö“þIø÷šé‰l't'F¹æi@kš_ôâQ'¨Zœ¾¹ZÕ+E7‰] ²PöŸÕ(”‰‡-pͶís³ån“t Ôk£9{ýþãö“kpÉz|Ê9h6æ‘ Ÿ^£”á†ò[5Y êLPçu|ëò6hñzv¹»|:ôZ¨ï.éB÷å€ãGRü¡¯¦ŸÁÀà =È³Ì ÁÈÛœÞÁ²Oðó‰Õ£g|¤0‰éÒ60ñM<ÆJÇ&ƒõ#|òòar;5a²)¦¢°ñ œ€iŪ½“"ûzD:°(g¡¿ˆÌ«Sy`–‰€`¹z¿¿Y|¦íÞ€yåKͤ”`áα¼‘ ^XbZ _ ·ƒåÓùâ =S°rLðUºf,¬ú›|¶3…µÔc±Ú#°>—xàü .Ø4ñþ4#[÷n?xæ¿/vÔ| ŽQZvÖÁž÷³[Îðïãï;HH}\ƒåD9úÛ%þ¼ÄõI\Ž‹‰Hj(#ÇÿÉzár }Rø1_“)d`æZóDdn< ¤r™4ÕXÜ-ùdqXQ[ ²¼”jýõ£Yg>_¬ánAö£ñuÆ TÈÑq{ñ†²ròô®RlœÅƒ#õT̹],ä¤w‘g,ýïàöä36̳ÿà‹üŽþâ}É(pCUV`Þ2P‹YŽ »Çâ·CwQøG`—ÅK_U¶>KØJ€bžþÙ,mõ(sýJLêÊ[œ:‰R¶{óE9(}ÇglÅÞe¯R:øú)£Ü[‚wã(¿Lk¤"Œ Ž—Âåñ *~èos}{•šXé[µGP…‚plȤÜÜyð ÕCmOjf¡zS‹Z'BÊÛtÚí﨩Mûu‚îj©™?ºzø"j;oyzìûÎú·%sÔcž)“±fEýë5sšÏN¡i”½Ç4Ø.~ ƒ†ïRíîÛK¡‘öS=ý­whÔ[C1t–]SÂô‰=ÑxpÔ¢›gMÌv4û)þ¡Iá;ÁÙ4åTù– ‡¦›×ÿ¶„¢iu²¿Ï šQK«|õ|ŠfæFìšx[OÁ‡ã2êÞåÎ5uGŽ„™Ó6+KðûJèøÇÓÐQœ/è ëØë4ª ªÓóWO%}}Mç z_·ÎxºA#‰×âÁ&È‘ø5Ãö&j?ÿ%ŽTiÁ‡âß9‹u· 히k]Û5ÃhBF‘…r(Þì¹mWþÇk!ãdÃ^m(iŽ?ÖWå½ì_²š3 šIV÷ƒ5;ÔÄ=X¶< u I3,ë |}gy¥« qT0ç^›84¿½Ö6Îj_ÔBc•¬¡µ?¶ôÕá hÿå–æ­¾ãç>1„No‘>™,Ðu{ýžL/ݾõ”x Ü =;¢ «Z× /²ˆžìã3p´ið3tÁÓ­Ë)N0ÜÑü¡ˆ$ ~´D]-| ?ùß³äSÊÀX"Ý=Ê“0qè_|!÷i˜lõcÞŠ‹é{¤iëð+â 7kÉM˜+g c.lÅ# ÔE`ùâAð{°2´èª< k®ŽÆ*¬§`}Þ »>X 6#€yJp¶?9—x~/\¿*<ZîÜÔÞ½^™K6¹H@¬‚Í~H¨RhU˜ý‰‚>òý^éAâöçDÂŽHª`ì:rÉÊ»îK_-D OîâÒcH%ÿ±WËõ+Òð\O’3IB:Æ:þ s]d 虦ü¸Ïõ2EÇU1dœ¦ö[ßEf¢ÏÙÝÈrFóÙÍÎBdãóù£zã%rˆÞP¾×„œ®Ö¥äÊ–È{Ù$ Áy¾‰–œrB>>ŸºeN ä[o5Ú¨:‹ü“u£¯¼†Q`+î1kõqåî[ðFÚyLŒTQ)Î(AvU ´œ¤ £ê·&òôCk¨ÁùRæ},+¢çìò›TiÔü¾«³³J…Úì,Þó˨½r3Ú÷šêÔR5WÒ÷¡îýÏSEß¡žîëµ™‘xÔ[ ΗAýg¿(\ÏY£ìóUÝÓhPÁhÜߨŒ†Z&…thXb¦I>§€Fb¥BŠ¿Ð(ÚáN‘ïg4ZT{ëeƺ֓#j;x¹ØVWÎÓ5·/lP>ø/þ¡]¼þô4¿Ü¥]¨öŒD4JPÈ~ðÏGw$¤CžIÏ9?)È5"ã•a…œž¨àIbÈ”sŸzaií•™ 1ÉJÑd”&óðnàröÞt:dT†M1ghìçÌCQŠ[PUÜÓðÜ¢o»ˆeÁ—Aý{†´Ð!ÍFör±:û('¼ÂÞA·PþÖ!m3è¹|f£Ý€z‹¹’´—Aßå1¯š3ÿÛ³¶±I½ÃEzµâ— —.ÒÄýi4ôhž:Ú›f Ý/oð?}JÝqãŽdû¿—G¡)òþŸë.±Êÿ/'–]x× ÝWEÍ=Æ( {m@®ß¤zªÿùX(©CïLxV •ôû8^ø{U¾¼¶Êk4C ''ü5oÃw…Ǣė]àGOò‚Ö1„ŸAfù—¿ëÂ8D´ŠvÀ¤2å÷Òm0íHsZ¿~ë;ü f…y×¼ª²/`É»/ZÂØVü§Cê`Íë2?ëñC°qö}œMu:lÝYìÏ뾿BBêþ݇]¿¾ü’jøgÐã¶Û„²3†/}¨ÅlÛ:Š Iî²E}…d—é¼×0DŠ—œæñj~Hµ¦Î÷öþ²˜×‘dË{üís<2¦”Ö†Ò!³Jß–úp1²ŒÏyª‰l™õÑ;-Èñ’³ÒN£9 ˜ºÎýøˆÜäÎåNöKÈû(p6Šiù…¼“ž¡@|iü§‘«xHÂiPè‹›Ï  Èó*ÉIµ×(ö¬%1ƒ‰%*<‚?5£4uÕ\SKÊêÒÄô|×B¹*Íü‘xXmQo‚û(*,µ«…åJ¢RÑÀ·ÃwPåÙØí^kTÍlÞLä„ÿÌíßp¼VÏï,ÉŽš«—ŸVÀKÔ–ˆöh!G ·ãÛ—zQ÷ܳA1ÔOÌ/œ‹DûJÃÏš6Ðø2gÇ_)4uAù¢–L4gËÕ½=“‹5Fî·¯’¡åG‘›;vxÔÿ¾á€yZÉ ®Ä’E«Ÿ,æO½B뻽Ú~éÛhsÈUó:ë Ú„Ì»©' -””¯ÿŒFÛrõç}¾£¼o‘/&Ú½yÓ#4KƒöDökûsÏ÷†[Siï~îmtÖ8û©n>xUmã]i·_§AÍ—³»íÛ¯¡hšÇÇÆ{r½>5&Ù9®^6‰®±3ÈÅõw_÷¶e¼’©á}³œ^zƒ $×fÌœ|¶ Ir8’&‰l5Ç?qC‚Äý²S•T!ÅûÄҘ¶&ÎzBv°.AÞAòÿ®·ªªýÂþ‘–nP”–f-º»-DEAÄDQ Q Q PP@Iiié”’$$¤¤áåùÎïñyg~óþµæÌÙgöì3ûº¯ë³ï{¯2¦3gÇAŽ¥ú©ÀwÞ›Ká[Vh yù¿&Þ[üÚö\5¿Q ù·x¤‹B!ª6-Òéo-8sÝýc¼5ÜÙ`†Tÿ­CÕI|¿Ë/ÕâêIÿáÄ»6ùêP¸pß0TÒŠ/RøQ†¯;”Çîîú}ŒiÒåY/(M[ß±þÌe3N¢3ÿ[g»þ™ª"Äö¾Ü†“eÆO îkUï ·Vhб<©ÍÄÇßøP@ËKS^â9ghµO(Ö|¼Úõüï;$®A§7“Õ¼üèÚÇ'á ?-#T¾ ÛC?ëö4w§/üb¥~¤æ> ¶Ùö°X¿‡á¦#\ÏŒàwR;£ÏŒe[yšs¹Áĸݽ¢Ïä0mUæ,¦M³ó9NKG`þ¤pœ,W),tØ}üÒy–œ/0øŒ¼„?Ò[= m뻾*{ž´Ö¯¼Qõþ™›ã¦äY´°ójÿ‰’=»~*cw'8¬ S¯X®)­"±²ßÓ¦SH²rœ,Íé’MV^R´FŠÃråå¹HUj|ƒP½i_¦®²~BzuÊÄAddº+Jñí=2“y.ŽŒ #«l%_æ0î ß[þfò7r¨ì˜hXª#—¾c7YÕwä¹L¤8|žÓ­7k$‘¿ÅSööQ?:'YÅé†"Gf˜Æ PL…î:»î”(´ròíK”úP|ñw½ÊTuö\>Ʋëk] é¨ ¤¼PÉŠJçF)nè£Ê:÷Þk*ã¨êóJG·Õy’i¶Î F#Õñ‡¨åK¹“çí€:Ú/äy·Q,df¯ÐwÔ/ Ž”öPFÃsïFÌ•·Ñh;a!%ÐÄÿ`vk šÞ'ÈÝl@3‰A*ëÄÏhöƒÝÂü)šßzmõ‹« -¨B8r Å«Š[fu´dcY¬›gGËK'Ñr çfìñÞøÏKãs¶¨{‘4&E, ÙÒÈ\®ÃŽôÑU¡ç”ð½4wó+¨÷ÿ3A6s *Æèˆ¡Ô¼Ó›sBŠ .J]«€é‹r5Á=«î,çxüd;mmŸ—Tÿ«Û´è€ÈÛŽi¶ÇxÂt²¼²‚+Cá¹£ïot²Ay¨ŸŸàk¨eØxä 4…í©ªu€ï®]ÎKƒö»¥¦×4œ ëwuû‡öuø9x›äè <R¸Ûšƒ§ó$/«ÙÃPœ¤¾íœ6 ŸÚ´l<ÃkF1/Õza$E>Qñ°-üv&\ò4‚Qð7¿Ô±cokNÒÀ¸Ìõ…Ððb˜¿u|lúLjqKkŒúÀ”¯ºDgÈLo<ÜïœF ³±·ª^À\Ù>/«BC˜¯k[‘]‚……B§%–/°$B%ï®ËÞQc¾ÊyðgñvÐBv¬†9÷],Q€õëÛù ¶²°(—ó¡¶›\†tËgàí™éýîÆ¸g¢Ëo¤è"Ù: k[ qÜQ9q&$='<‰¬´H¶©þ€öîÍ\#Sc AÊ!¿íH²´P­8€´I¾Ëaso‘nìzM¢Òï´gHK¾AF®—|•ÈdJ5«Îr ™_mm“þáE–µöýo­ ›kMZ îkqꯣÑBvw*–çB‘ȱÔñ¬_] 9o™¥ñqñ!7½Ñ¥ øwÈCTÐp¤?yåXŸInL!_ÈxÑç.[ä'VØø­±Ü(¯LìGÁÿdg¼ˆBDWjƒmPXˆõ—®ŠÜؾþû*ŠÎè°DŸsC1³åÆç:PÂêÇq6zG” ñ‘¨ëôE©M’¹Ñõ?(ã+(OÔ8ŠG†[£‹Â?¢ÜÞUµšA”×`K-UB…Ÿk£’†Š6ɇ¨¼œ".n¯‰ÐØÍŸúÎU-F%ºPM÷Føš¤'ª_@•›óÓ¨Q¼Ïr|µ$*».F혇Uêbî¨ãÒ¯"뉺Âî¬×I>¢niÐÀWÔó–ß2“VA}Ö‹~U¢5¨ÿ‘ø !ðÒž}±?Í'~ !…ÑÖp`"^Nãõ DÃ:‰à4âd¾d<úœOouC£4âGgìðn!gÆuÔàÔ¨ÔõG6áK,ûô„`í eB úîø¾¤tÓæ*T®<³ó92%šb“ÐAù¡›qïu b)qÉ™sªD,ÌÊÆ æx6µ©˜Ô®ä¦Ö1A}hʼn3ƒlÐû"ùËÌwhzÁýì„Õîue¿f/Ö¶„–Œ¦E‚Ojð½:âÃLa´îÔÈðImBûíÚ· !ÐɽøK~P0ÍYHGAAb÷™•§ÐÛÆ,¬Â T‰g%…Á¯´9":¨úDûg* [ s•'Àï«|g=œ aÌi¢åÐ5K˜xÆ=–§¦‰.ääLÂl„“çýñý0/½¾uKÒ¸›E(ÎÁ’r1ñûÃi°Üu:>©ÌV¡’A]INª»ºL§;³ƒzœWŸN‡E£ÞË}Ô gMPŸÐyµÍ½ õÏû2S& £~™mÕ”"ºv³zKžªEX•™Ÿô]û}Ò)Sí?AeÚ‘ß;øÊþ丅õ ä)Ü.¶ Þõ³4ÙóÕñö$ÁØ`¤´8{p2 Ò4.±\ 4„ä-MwU\ þRŸjéÓˆiÏä’½@ 1q±‹i9CTXÈó3 ²ÕÏÿÜÓ %¤^eÆB•IÙy÷ ¨÷d.{}š¦ Õ|5ë¡%LF] áû\¦ÎÜ)hÓSÎÔç#‚ö ©ç7¢ÐaþØÇÅE: cµJÙB{=Ÿåi9h§|kÍðÀÚ ¾ÿ­iû)?ǹC«ŠÚ·—*þáÂÏâÎÿÌ#:¸à ­,ëÚo¬ã¡U›M”Édt÷xö3 €¶+þW(ÞD@»¿YK)ÓOèXwý Ç]!›¾G#… ÛÉ1 ¤¦~ž—‹¯›`€¾É,‹Ô™÷0°µø$üÙ fÛ–²}» ÑÁÍnê<ð;ñØDÏ4ŒÍ÷¥µÈ÷ýÓqîô ˜ñ-ˆØ=ïÜú¾NÅ“D°0zæÜ÷ú&Xø±}Žû#ü™ž)`ßk‡™_¯ß_‚ôLxüŒ¶=\ôÌ•àËaÅ7§f‘Pí~ôï{Ïhºùø…ב¤9«ù ×S$›p|2tŒ)à¾ã cRU›—h÷¸#í£¾#‹KLH¯ôH«‡‘ªWø®dÌ!“cD1Édáÿ󴱿Ù(®Se7¶ã~.ǬãÈqåæ­JqäÜHÖçj½†Ü½vÆ^\×ìÍP¿ô9äë ½&)µ…ü—gGãÓ àáéÎüI<ÄÝY ¼›7UÞLŒÙíÃÃ~²NUž¿P|GÊí¦ÀK”<ûnòMþ!”ZJ}~{Deâï]¹~0e¯<Í“/¼‡ò†å§+ÆQñ8©­ÚÁ÷¨ü‰ºKDg‘jˆy|µU³3¼b«9QmR8!Öw/j(´V¨2¡f.‘>•ã4jŸç¼uU¿ì þsõEnN µ° !íZô7ŽQ4ªîâtXnF“«>$èÜ…¦NYl.F3p‡»ëÄhNÐkóî­%š§ñG:‰]E +Bw/Ó7h1[]>æB‰–>ªöîrÑŠôw¯åï÷hu¶üÞ5D«•J @ë3)Vw¢u¥Ô‘ W‰ðª@ýÜÓ,”­Ú"±5ý‰”­¿/MÓw«Êb7+”¸‚”|ä ÷Ê:Ÿ …ôê÷¤ÛîôÂíØzDí*$µi°TäìúÞ—'J9B»|{KZR>’y0ëN³@ôIóóG¦u,Ô‹ø'¼ëË‘hÓ„˜}ÞCjùïwJc3“R–Ø+©5oBFü? w}‘/½[î*ä øÜI¼yZ B&Qƒ÷™ã]Ï.'{–ÐùÁ á_rßý­ÄfÄ“/ €ŽÙÕlf ØÔX»äýoŸ ¿-(ÈjZ.ˆÜåDuB3}(J­ ùÙ ÅENyÏßkîún¯ÔdÏ”è»zÇ'ÊL:¶¿ÏByÙS_nâ2¨tÖÔ× ƒjÿ¯²¼X¨YVÉ8Ö ßL)U>ÖºCã2ýìD¾´21ò–DÃ÷Šà—Û!Ðöñ9éË­hè¨qR}bu~P•~¾Õ =×>8Lž6„>YË SZrj‘AK¢/[\Ê0¬˜%þ=* ~óoR_ò> cûÊ® ŠA¶‘G&áBTÈîìθݵ9–Œp­×!÷[¹äðäãφüŠÞþ|P°4í|ÀAcö?ÕÔ¹«×;4x%ÎæëúJ©‡mö¢Œå,Y”õdp×Håß}t†× ÛšhJ´QåìºtÿI%T•b¦·ŒDµ&J®|Â<Ôp›d{Ô§ƒZWRXJQ{‚©1*Óu“ì&¬S^¢¾ýO{2ÏÝœiæEéŽF±ÒêDÛÔh"^ÀÝñz?šZZxFÜØF32ŸÛÑ,íñɇŒhnšû0ÍŸÃ' ¿ÑâúÅ_jƒGÐbi-|ï Z^V] /¬G˾»WcFbÐ÷èÊ›ã ɨ}¦íÙáúaÜÿLó›å+Ø.Jè-ZÔ†–_¿:³fA£‡OÀ ¨°ú¤6o %™|Ëé h<"ö\Hä+›Ý»!5_ÎôÒº3+AÖKy—ô †¿ºý\'t1;ÒL$º:Ôô!3qŽ^lî.èÏ=ã»Î e_º™&¡zå%_M ?4(½òÊR…–“. 9' MñSÈ[7è¼ä¿µ¶Ô=üxL|Eú–F«hà—¬4£”¥ ºÊžŠ”‚!Èàù5 Ck‹5 h`¸ö·²…¼!Œ¼3åxвëG·Žä$ÙÁ¨Ý„ËÊÕm³–½ö Æõ=;„ú\`Âü[ä=:˜ôâMi€©áý)¿~”ÀÌí5™†¡0ÇGk·GžæyìÎðêXL„À¢ó­#Ï®ÀRf;Ϩ.üárz{´ÍVЧ˜bú`í…–sk_nŒ„“bòR¹mH|º”%d™Õ~Šó2’=¼öz¤*÷*ô` Ú‹” ïøßCjÆ ×ÜC=H+VGìxÆé’8#ïìÛDú¯übæD¿‘¡ûÚ‹¦;ÃÈD[õ- O¬vNŒ=EÊ_¯/Z£_š gˆ+¹?µŽFµ/‹î™¢1ý×5/ôœn{’ùiÕ²/ti¦{"kæ|yo”=¬É÷5D\â€Æˆ¿àwÅP±~'VYŒ ¾^šJ[?9« ÄM!›¢—†ÿ½¤§±*çK¡‘1¨ä{×z`Š-¼¸büÎ@îÑqC’„pÈ,½Þ½96é#÷[ ¤íDÅï/{ŸUn)G= ŸB®#ó¯GµÙn$Ñ7:äÏ>äø[óOƺ\‚l^{ÒC _ ÄdòØ©%¨¢%¹4š¡õ†7;g ÉÓåfݧ%h9fE³Òß Òà Óþé7sµr{ÚOm>âá,—ÓþÖö<{‚‘×›ÐNžh_éJm6Ë·Ë_³CkEÎrí—ÉúÑd×åÞ‡VIwkÎY±úÔì­.ª‚ï^6Wîõûû¹Õå”P½Ä±ÿÎ'À±\ÃþAF˃ÐEêM:ç?Jœ¥RÌö@Ïë>ÁçvÕÐGXžÏá±>áöEÚoaPT ‡¼Ê†‰Ë%Ü„‘µ$~ö-j¾E3±¦ÜZþ×aúUT1Õ5˜ó6øºà§ Ž&I!GÜaɆúüMïHøc5ORv¦ V}R$ÈOÁúÜÓÛÓR°óX:þ=¨\ 7~ˆ{ró îÜ¢C"«ñ(ÍŠ$ñ9yyO’)dU‹Þ®Â½î…æí q¶€ç:Ò[éÊx."]‹e4‹%2ذšÄ<@Æ©º°£mCÈü¾îUE²Þ©=Æ|j÷=rÜÜìüìO¥$Hݑӱ71=t¹_š¬;TF )„Öb¾™<_ƒMî:ÆÇ;°]?ZIJ†U§æè‘P¼^TÌ,‰>ò–KBµìÒü0$c:]Ož{¥fÌã¶òƒÁÏÕQ[¤±Ý÷•Dºø¥‚,½d0¨‘ÚðG&–ã÷ÜŠ…6€ÕÑñ4²I‘ö¼vÆý¯–ºò/â ûæäÅÊû¼tü8‡¼g-~œ¨ßÁƒîu6Q£ §8éÚ'<Ôå%æk±ˆ¢ ¿’N7¿D ¢q¿'Ç,P²Ã¨ÜÐfgä„X-˜ÈHµ…jžÎâ¯% Wm{À5BÔ¬o«JåDí —ç8êP×Hô÷G)[Ô'>äZˆŸi<âüA#ƒ¢g…ãhÜeÈ•ÿ M­zO‹£iW–¼Î4{Ôœíô’Íu«óŽ„¡yIíÌòt:Z3Š|D‹–Ÿ G>‡–†ÖÑU® eQÑAËŸÃèý‚?(ìÂ6j‹ñN½ ûŠ~<;{¶uVÊ-,} …òEvU!ÔjïD( ¥CyÙãs±saPÂÌìkývüï}ü¯>r6FüõËu›j~„ù÷Û‰¿~™Ñ!òR"þó_ý”NÖE>ïø{ׇ¾ï”yšÍýóŠ›ÿúTû’®€)ãGøáDqìÅn>ü¹w„› P$/ýq¼¿T!ÿ¼>7 pœŽ "‚ÁõðØ%UúÚ䢡Ãß_'3ó/ÁÈçáý¿{Ãï úçV30zǶý‡Þe;wÏÔ©ÉÆíï•D1ÁÄ"ášš’ÝùYí(ÑM˜6,=HpfÉ)¼Ž:zÂÜ­Q÷væ.˜|?³Ø 5÷%„ÁâØ^«²g°l`›P²ÐS׊üîÁjv ûËsû`½þ6ÜØ®Ñ˜'°·š7F&¸‡:jøÒM $ôZÒæ:0ƒÄ{tŠïè"‰QG8’ÑE}*˜ØäÅ&3c$÷âerÎŒ±R½ê_JdkGšLÆ#mWîbÝ©7Ojþ’ÈðÒ(¢»™óê¥î·"ÓâS¶¸6dÑ׫Vé×DÖ|Û#Z­á¸O6¸À7qzÜX  rðíQr ÅQï2]V RÊuf;ÄÇÛ`>y’¾[˜²S"ï\$oá¨>4šb!ð^Fþôk^iþ—Q`mNRI@ …¸¥i=ñ…QlT,% Øÿ*ï·?ŠrþpvÍ&ÂÓ$±I÷æQ‚þŽý9“j”tõÖó`¥™’ø}^£LsðÛ¸éj”=V¯ï̆rAŬ¢mã(Ÿ§±d•Ì‚ŠWb‡Q).]P÷B0ª˜¸²0O¥#ò˜pÌŽg¢êÍ£Ìu7˜PíLIêÆáW¨~‹×u—ç4†K²+ŸD-g¯³ÂÑk¨=Cã©u"u*ž‡ïùÕ‡ºÁ{Ö°¶õôÒÏ¿ZA½UŸKç’®¢þ›bï·ûÑ@¢¼ûöÎ]4(P;"äK‡†ŠQ7!5 ?HŸ×tC#vfí, w4òâ‹Ou½ˆF+§è.¡±P­wO¥»z88ÿ=¸žœY¦êGU瓪=5Ȫâv-z%V;Ón×ÿñŒRÁo¡|ëô䪀æß|šCd¢›‘Yr,ó/ ÿê-Ù#ã5/¿$4fµa\Þ_¿û7Ÿþ[ÿýþßãÿýý¿úýWÏÿú^¡g­Ê›ûýª”àòO%Ï(‹0#ßÙ¹ òŠaC©ŽPÙ&(l? ÕWù†XçÃÿó«ÿ/ç~[j#1M§€Æi¶âådhfyXc÷w¼ùNIZâÇBø7·¶¥™0—1A‡áÇYOS±ÿŸý\óm½ØÝýﯘð /Ã/Þ² ç+Ã` §õÕ0ŒL\Üc†ð›FÁ´ŸÆ¾íY®€‰W¹¦Ö0MŸèê+ ³5¢‚ôÎ0oõù'ÝÒYXèMJ3 K7š¤w$àå@"•)¬$'ÖtÁÚåá×:԰ᨵׯq¶Þž£ßsÔv–9t¬>áÙÇ”ÃOÞþÕ3QÛ3‡cçþúi‹@‹Xã+$¼Ö)%æ†×ÇŠvQ¸ÊY© N÷1º‰´¦ÕÄ&»9îîkú´¯Qÿ麩<œrž™ntgZ|íC‡Vª‹WÍS06]¹÷¿©dzæq9$Ldš˜‘SOjx0`¹¶ÓŒÜ ¹—"^©÷ü@^:™ärD¾;šÑd$‘_S&†P0`σ‡:QXú$añ1]ürvÁ™Å•ø#£•8QÂyãõ›íƒ(ÉñQÔ¥XMÄÄOŽ ´Êù²†Þ4”I%´qó#BYG]Ë"Ù”×y$kù5«-¤TPù¶VúÑ:„a! Ÿã‚¨ZÜSÉN6‚êT›÷HPƒµ­ît -jRJ›Æ»Ô£æÂ²V½€/j5ò¡öÞnÔŽˆ“=4,‡:'Û~¢E]†G?EêQ7ÿjÅÍíc¨gíQòóÛ&êýþ"ž»èŽúÛ†S‹ÙQÿ÷Ó~ªÀQ4°¡"°<ãŒÅgD ðÚŜɎ齨˜jXÀét¸¥ƒ¤aìÓ?»yrCÅ“è ï( ¸ý†ôëØ?ûQ´óØ9@¦*5ÿS‰tH¿àËSWiw•gb“ v}O®~„`’ÚTç":’ öõ©SrTw ZLKS¨Œ¢µêz<6Ù ñxwÎ…ÇFÕLºÆG£%{Í[rŽBeN‰D¤EÔ}ÍeìuÍ€&ªç#Þ‹ÐÂø%v§¾Û_2{& ­IJ¦S‚Ðöó>s¢!´Gwþdgú íIÉ#®ÂéЮí_šÙÐmI—3Ÿ/‚6z½Öñ»ÄÐú2|S°Úö¿ùC9;âÚ2ïöE|10½Ë’ì¡õl Çzh­{'nhúÚŒƒö YµÖPGè˜ü²P± ]§ÛO¥œ†nj²÷y5±Ðó½éÔ²uèãV ÈpÏœ–ÖÛ‚AúˆT»™!ê8>çy6FŠÄÎVQùÁè@*õ²s*LËÞ'ƘnZM×0†9…®¼üX îmŽøögHÕ‹ÅM`yHaüäl¬’¬mêPÀº³“ªÒ Ø¢w¡>Os vz=PîÖwÜs®;òG'äZ* å"q[Ž¢œÃ7$~¾¥ê†{ùJߎûÏñ; é ‘äÇÄiþê1î Oôð‘iHž·Hò¢…Œ}‹“‘÷É [ëS)Ÿ½1|y6&&t r½Ž$MÌAîÂÏn¾©yÈsz`{!öäõ‰òµû<„¼ñÙå4]7È·hSvýôÿ®íª÷/|¡‡ë±/Ÿ$¡ðK¢fèc1(Ê÷«´V¦¯‚úm>´IPrwëÚË£nPúÞ0üz”-½˜øÍ¨w»:M®iBÕÝÎÎS;æP“Ùsåq.Ô-eÜ=z<®ºµ¸;ÑC³³—Mï´tm¸x”­­>GÅî8‘tØ=a :[¼N?­t€nÎ/‹- nð³ô›r¡ôÎŽÀ¯g¶m„g¿ÂÐ¥¸šŽ»0bJئ 0ª#þÃ@ëŒ;·> Yz“Uõµ¹"É0cÔ[ _–s>µ,Œ “0?ÙD/, ‹n­™¦(aYEïB8G ¬¨´ ßckž˜|tû:l,PqKü¼Û/–T~\VF‚„_܇^_x• ‰üÖ¸D†E„ç:ÉË@DÒ¹Q«éf!ÜKñöO| RÞê]v•³Aá®Gú¤ Ñ^ÙZ Ciö^$ȸ`ÍmȄ̳ӽ+בÒpû…Ùîw<Úíb³ŠÄT¨?W°"·ÚNâ<ÞA^¹¨Â mxðŠOÑõ°› ©ÅC!Æñ7{QÔƒ7ªïékOú,cSq%cþ¤ÑíqCéO&·¨=ØñH…yT5oÊõˆå0š¢"}Ýë%t¨CV•n¦-=õŠ}Õ.Þ;¸ð¢ 5h_%©¼„š…×–öŽ¡¶7€vÐÔÕ4Qø^5Šú{bŸ8Hq£A–Ä×ÛÊ»V}ì¼Ýq»@4þsAÓòŒšjTÌÓ‹G3Ò<©Ó•~hV>¿ü3Í=ú¦"O”¢…XàßÓhñsRAÍÐÒ¿þ¨…VZIHj‘Ñ U§tÖmô. ã%™mEÒÒÈ·®¸/¥£ú—{ ly*$ÙXKBsW[“ÏWwï·>o3êPnwp«õ|½4vÔ)íÒÿæÃ…õnýo=vö‰ݼú)éÌSFÈ´ -0ß[iOÈœè‰!ÅäªAÀp+¤*¦×|à`‚Œ§¯¦ûžCþÅã+2CPv<ü^.T¿×PÔ¹ßêÓ¿Ýã§€Š¾GhÍÝÖ}W {8I©ÿ@7±aÿù‰ÄYøóû,Ê|˜ƒÕÄ4r;cnXO bs‡Í_TyI)‚°£tòìdd7Œ¹±] fBB“'Ú+¨æ³©Üv1’h —´öÚ#iíW"£GsH~­ÄŠôbRhŽí¯¶öF* Ï/Fæ„ë€ìÀ.ªRdìGz‹Š ³Ôdpy7òk&C9_PÞýƒLßZž)|&Ežs’š„áÈÒÀs”Ì÷Ñ QÌsàþç5[瑃޵B´:¼bí\JßF.¥·•‡éàÿâÃG²6cÝÈ;« ’¯»Ë‡­R+‡û?ÓP!ÞÅ vx®Ä)T¢8£è{Ö"1¼²›;íYÌPëî©ÁEzÔž:#+:mÂu«¦?ÒšCU†"µ!ó¹CnÊ)ȱT£âƒ<Ê%'ÇkPhTÿà+m™•”,%¨ƒ²W—“-NÔüLJÝ4¼E&W ÚôEúíkPË@êülw<ùˇë_߸ŸƒÆÔ“4{ÅŠ Yv\x}&´Èìs|°_¾ëöZÃõÐú<‘º‹æ´-Qß÷gØÕë«%€.·ë×)Å£¡;ƒt¤õÈ(ôÒðÙ4ölüLJìgϳ®ƒÁg#‹/(¦`X`ÄÞëo`dç¹ül¥Œ¾û}ât$L¨®¿&? SYïB>wÀ챎br3˜g³¯aÈ„…7öïZÆ:`ITˆ"p‹–Kœû–+öI%Ñ:ßaí ÅW»©°Á:¤–«[Z&ô“‘°óŠŠfú†î!Â4Ñ93$Ô/ò€&Å‘žñøzIöm;¤ß&BÒ”L1¶u$wTô£xŽÑ O!å"»ÿ¯ßžHïË’ü0iõ+ê3ÊÉîÞ¡Èù3W‘þAOtuU2|wüÀ‘±ƒL—©*ÛûD‘Å4ȇÐÙŽ=†o.š¸ÿÊQÊ‚ä8QÛ»˜EŽœ‚Ê>¬ WGå@ürÿbªìò–EžÍ¹rÈw¶'G/æ6ò gÖ‘D§¢à…qm÷P˜”˜âL1ŠæOwMJ¡8»M Óu”0LlAK”¤_ùQcœR´{)²ÏÔ ´’Ó‰b”I»œÖ5²—b:Þ¡/Ê[‡ýŽãjAEû›Å¤©c¨ü–.Rº–‘à¿¶ËT­x"½54…êìôŠz¥ö¨ÁÛªbOjšüC}ÕgQ‹Çã1*jÓLÈ,s<@í%>ûaÍ8Ôiþ鮿ƒºq*Zãì|¨wãd…ò„ ê+GK:M4£þÎÊØ»Ã»ø¤2™ñ ºÄÅ÷@£#ŸnÅ«ìæßy nxõ狞Þ5¨`ñ£ çøÒõöP&R€1q’\Ó£P>+äb{ Ëo09¯_†œ_7ìT˜Ä!£8þ*ÏmHãøúYº>¯e¸œe‡”ó«Ç½«†$6Ú¾úfˆå)ç ¨€¨Ñ,“é­ ˆfùz÷Åu&HØ*£=w²¼R¦8Œ¿ÃW¿k û< ÒÎ}Ñ|©êN-ނƈ/‹ÐÜ@WFOö¾“¢8Y²ËuŽ’ŒÍ¡ÍvOÖ±© hgòæâ* ‡vC’M¿3 ­õó½{D)Ðf(CFvÌùœGë0­ç½ý:*ÿ{ÿPE-óÆËfhµláZ縭.§C¯Qîúo&À5¦ãÐ&Ô{¤áê ´Mªh›:˜Cû’IP?™ tš ½ †®"‘ÔªV è>KU¥ª?©K^% †>•/'ê{ÁÀÆ;‰Föðk…!.›Õ†>¨QŸŒ˜«ì›0új„Qg8ÆW8FÊVaªxGZMRfÇêI¼žÃ|ÖÅ?‰gƒañÁi?!,_½÷mͳ VžèqúeÁÚ°ª½ó lú²„Ýä‚U~eó0Üð1§Ht ?ù“K–!±×+Au9’ú;Ù5’ÿÓÿ ú‚R+R KÓ‘X"u•|Á‰j¤Ó®‘`pé‹ è³ÇÜQ;èÖ–ý;d ¥WËRG–t÷û)dK;}»„è=÷”°Á®Yþn"7³üƒâ€%äyà|µlByw®,mMàÁGŸ©øfµQ@æÄ¯+ð…Ÿ>Ù/š†Â^Æ-PŸBQ›ìä¤9(Vwá²€iJÔȱúЦ ”­wâ¡5k”!_yÁ" Ô^PŠ”q@¹°†ì¸÷7P!LÆÅD,•º e(^Eá¸Ïz Uõ]>ZKº¢ÚC“·­Bõ~†ã!Ê‚¨éø"áXÀjslw4áÔYg<µõ:>&þ]‚é9½ŸEÐÈõÚÌÙhrðÊ5§s›hê`IëëF‹fR‰)Ìh¶¹˜Ûºí…æå &\ôÑâaCÖ°>Zê(ú:žE+’“zÍÚhUfù­½3% ÐFyû˜E()ÚlÝë#¸޶e/…ª™Éñb­\æØAR”j=Wjõ)ÈôÊ>¨A-Ê«ï U¡¤—ÿHµ?7äðËú»PŽCšÆÅ5ÂóYÿì[ø¸9ê$\}g+°õj—óøžZPüo]©ši ÄØÓ|zZÞ¯®ˆ-K†w¼›¾iqCDÒácaÚÿê1NyofæU1HöȬž|Ò[O]ÃW.µh~—sN ¾`P¯ì%Èîž"Ý‚Ü(ßQ! Èc9$~)èä4Wt-Ž@Þ[§àûU¿çv¼óûûÿ9sôŸu§d—Æþy›|¦Š^+ðþʵüÇo¥Ä‚B‰«Âž`û.( Š-2Õ„²ÎCì¿q@…I¥.«õ¨öÌ̓Ps¦7xGF´'aÎjÊQ²dæK“Ĥ_ØÁ¢É;1‚XÞGôõ|œ¬°hž¹ kV™{y&`£ÜX´Kt¶ír’Ûú‘àIÕ¯ŠFܳP{¨té196Ý=vIöY]89Ф ×)è^¿EòÁM¹„çHix•zÚß©7s 絑îv¸ŒN|2ì¿Â¼ý` >ó\DæfÓþŒÈ:Êó #q¿\ñON‹-äX´TvSgæßAôP•Ø'<ŒjÇÞì7ZEõ3s>û~ frà~sé¨}5yôrænUº´çø-ÔÛáy©Ô| ò,ˆtßk¢ÑZ%q4aòÙ|L!‡¦'ëÇ£j8ÑLÜ<-@ã ší0Ê*Åo¢yÃÐkÖÖ ´ˆ0±¥n¥GË‹åWšÜÑJ%ÿÆÉ{Ñšq¹¤|.­ç’4ß^=€Þ³9®7QCh à±t6î‹lÝŽ5™€Í4gÇm¡ùsdá©°ƒPÃ]˜n—Ù eëîQV9ÿÛÏ3íࡽPðìaYÁЙÿ­7¤.…ì0²Kò&!c:ëm¼p¤1rÌ>å ɆŸ|%BŠÜ”elÛdèjŒ3ß„ü8‰‹Y'¢ ¬ÞÉaès ÔÜ]¸S'Då¥MI«ÐRq£D‹ Íã|9… -tÏg.$×…î¨ 7Õ>%Aï<Ϫû<0àž«^I¿Â¾˜Œ3Áài•®`ñâ1§žèƒ¡e‹Îž£É0Üþd“×ù.ŒÇšüÒ¿óî®4~¸ £åqùýB0V;*9°‡Æ]l-´åa’ùÅ{¿“C0å3«!²3Ì%\åiGa¶ðÚÑ¡6˜úÍëXó“Ì/]¹ÁÂZ}¿ÚÝǰÄSÌHË—œr,.­ÀŸ ÖŸÊì`5²uYPNÖŸ½ƒÇãïa3³ªfˆ¶w?žœC‚rC׸U$ä^r HG¢–_²Ä‘„ÈìÍ‹%$½×{¬_° Éù% ¶ÌãÞ%¾»QvH9ËCLG·„4$í»ñe¤c2 ôé1õIù !ƒÆœØ8µ2º\[:è¾™â˜ö~‘GæUŸ·úüÓÈê¨Ní¢Dƒlã¥'nÆ(ã~×Iúõ}‹ØŸSà;‰<²B‚\ò«Îwý_|ûNå$ ïLŠÉ› ]>Ì7™î©Fþlåwf2(HlžJo¤‹Bò$ç)¹¶ñó™•þQ5.¢,%gþŽ¢Šú_ªÚQŒ\z˜TË%¸Õ®QñÆ¡¤ïaž›%(ÍqÆ­Ú¦eÇ÷ÊCYÃxuƒ^”»ÓHÅÊ…ò¡™¶£Ôò¨°P^‡J×K²Æ¾GÝb¦Õˆ\O_œnEÕçÇÆ’—P-æ¢ÿ©QTïÖ9Ô…šRû h£Ösõ3#‘»yS¯Í›¶u†ÒŽ=g9º ÜËO~ú£^Ú¹gªIÔŽ`|f‡Ná]Gx£!,û÷)F#Fšêïƒh4’§Dw‹s9ëî ¢I ?S=kšJ¦‰v‰w éû[ƒÏOסmxxT3šù¯»4Í£ izM†àšiâ‰ÈÒ¹¬^Í+Én×îŽÀ·'ýåœ]­P~(¶~øß{ìN^O!ÛìªÅ÷.¯]ž;÷ÙÓFÒïh¿/‘ƒO_½;‚¯Ÿ…µ²Dr➇:ȯAlêð虯ÿpbc]ù™]8L41Ÿfm¯„OCZ´_[òmbÈøS<Œ~#&Èipƒñ?7K`Ê&4¥›ö<ÌLûZÎüóܧ{gÅ[’lX‚Ì Èaq(^kdƒ–ÝŽñN7RaU °úCbÔm¯)¬×<§ÛŠ~›³ ¾ÂŽìF‡ŸB!4ý±¼iMŠ„¬úÂo9võ|ËÃþÌ+w$^à`Œ>rIŸ`‘’kzHÝÝ@ š?×H]ú²CcxëLR‡F™h !­fY(sI,Òù K‘¨!}€z8gÎ02´S¦¾±…LŽÐe³yY”u)æÃ‘M~¢sômîט&ì~ÈŠbg]2Â6ñÀüXôï‚EäJí¹§»ÜŒÜ Ë—yC‘g¬­ö’)/ò_¯Ä ÈO/õ§~ù6 ê’U÷œ¾‹‡º>1±Ž¢è½ƒFC+Î(VIOëeJ(Ö’ù<@IÂÓ,SרQŠ´lTÉ1 ¥.¬Ð›¡LF…pa ʺnÙzëžGùs‰¯Öb8QÑó PÇ,*­.åÈŸCäøºBåÙªuò1_P}?óíP‚Ô8Ü’f§™†šª¡eÇ÷1¢–)Ó¢j›¤ Cƒ¨€w±,¨«1ÌtŸ|õþPx ¾˜~Q‡7ðVE/“DC6ÏÑçÆhD¯—蔀ƴ’—5$-wýTfuMé[ìŠøÙðªÑ5^”Û¦býþì Ò6\I~´à £…7žyå¶ V¶Ë5P°q%¸IfqW £¶µë¡¢–ŸYb içsO†~äƒÔ?n;Q»ù2wâ²ò H<Þg‘T²“º’SÒBT4ßñL…–8‘'yŒÞ':ΗSC–ˆe?óS{(^UÙ–¨Ò‚JÒ#9ä¯þ]çŽõö;{o™Ùdo9O{o²“¤¡$2¢(•¢-…ŠJÓJvV©Œl²÷^Ù›~zþx¾¿ž¿®Ãq߇Ã}Ü×ûó~¿œ×y^²PCìª5å§õ’½g¹D¡ñ–g #]*4¥ŒTª1uAó·Ç jæW ¥%ìX¢L´F» S€Ö×ï¯H¾É…V·ß:–ª· evÃIÔœ Zn,ÞQ4iÝãJ7òá1¹ÿæ—ê¹óz @K€b“ÉwhI2ðÿA­DOªÊ>ìýÞý‡Ì©@›µ­,Kì´;ŸîœºÌ ¿s©-\€Nök;ºÐõA;¸9z ³êî…>gY©µ¬/0à=IÆkYƒ=‘¯¨° †/͘0óÀ¨®Ò §„?Œ›ígîZ “‰Ä†×2GaÆFê\/9ü92ê“Ï N›ûó6ŸÂ×ږŃ\XÞ©ÍŒ¡†5Ú„8–oGaÕý#ݼlÍ%òÖt÷Ãî›*­ƒ®Hð•[û¹‘:Ùþ"Î_‡$"üK5â…H&¥-þ²¡ÉÝ1YóÄ!¤ü²¸^} iŒÅÓ®Ôe!=áµù£ƒÈpáÞbÙSnd\òoAæ{k^=’Èjfãq+; Ùu]•^?»Žœ¡›ìzz¸ãժᔠò¦:>nï·E~m÷è—\ú(P\mEh}uª*¹3΢ÐÌNT_ÿùéÑãûú5ŠýZi”~Ð…’$jQÇ/ÜD)ÿ ­gÒË(óÂr¿ò7~”“"§UáõAùÞjßÈT|¾y¯®# •=ÜÏÛgžGÕ£éz·äTPýõÝ¥.e}ÔØW)ÕtBMy ¢7GPËoƒ¿ßSµ›ãr)lQ×mùÊé˼¨Ïišùïœö’ÑBíã4j>6÷ƒœM’óe{dÑÌ'mÒg¥-DLZ_œBKÏìÄÚ@b´‚”Êû{ùÝš1ô³ç‰—h=:h¯}ºm ÷fôÛDÛû°ÆV¸ƒ‡Óœ³ A;õIoûì/hÏaÊòpHí·èïøœA‡‰±cÉ—ž cçµD»4<ÓûRîó Êv‰ œö˜DŠƒ?øåY6a°7)à8ÏË=N–ªv£öØP dÔ¼Õ™ñ} )o|¤þ¤Ú¥¶Sß%àýJðH@î9x{ÉA/\¼^OùÇ|õÉ…WgÛe«ß„1'ôJ!Á”Íñû‹ÿÕãûÞ…ªë©>ð‘Œi&Ù¤²îÕ¬¦4Cnâ†:g˜(|ž•¿§ziò;¤©F)ÿõ#.K²@Ae=™o %ªœè9¥…._»½ÊX 0ƒ›ñœ½Óÿ¹Ògì ½†¢A’Aálø¢M@{ûÐy(ñ› ;§ð|ïy ú°—ƾžòqZ Ê/ïLw‚ŠÂS‘ðeøÆ—UÐH ªvŽF=ËõƒŸÄ&L©Q F2úW¢HÔE}? eÁ Ž|ýÔó× I0þvÖ4Ïþزf{­=7{Ü"õ¡}ámºFŒ#tBs¥âŸèîÎ2¾ðpúš6¤°Á uº5:ß…¡VÓT‹‡a$½â¢rÙ<Œ%µ>º‘‘õ.‰,Ä`ZÄòŽâÚY˜M>Mÿ¨Væ%è , ©`!î¦EÙî]Xõd¡zÚËË¥, 2ö°º.m#1Ü ÒÂ'&DŽÃÖ“(Õ«¡‡aW„F:Lý:x}|Ë]ÍŒ„­†¢w«6‘Øè}¯´©’ õÛiê¿“klCz·iŸGv7‘°""Œ¤ ™ªfW+‘%3ç,[ø ²•tê>8h…œäz-‰pÁ ×sÈ»¤G`ʦ‰üj×oP]EÁ}!‘‘ÙÒ(ìQš[¶ï4Šþ]xæò8%ÚÞÝï I@©ùS·z‡»Qö<¯J2AùCÊ&ò䕨hM΋Ê>‘WÕ~¢jú‚a:¦á!c︠ÍÄæ~ ª¨e”M—â†Ú¤C7½uQ7a¼âyùeÔwWIŽí°@C¹ÇÒódh´ý"‘É{MJ 8§Ñìbþ:çëf´{š[’M‰–›‡dcÑÊ~F*·Ü ­ÕßËžAÞ ­hF´¥d¤~VU޶›ñIGßáᥔ‡§ÊBÑn­ÊR»Ë(ùZK 1ŒÞû„–¡öÿOבã~“¢§(l”u|²çC'Â:àÇýxõù`¨ˆõÓHf%+ÇÉú_¿ÛÛÏT f¡¦ÿ"cjHLòŽËÐÒ@ö¶ËÛ&W!C#4ŠÔyÒŒÆ>·¾ƒŠÂÉ÷‡ ›è\¼ž©ÃÿèA(Ü*Õ• è@ ª]_꜂õ¶|n™1hÝr¹S° m ‡$«m¡ƒÎæÑsè>ô|0OYú.Rѽ}²Kf§ýÚ`ðÛÒóÔÀJº!Z¿O߆UîŠr~?ÃÛ9=Œ#C0ò›äu%Œ~'ªL „±†×S35R0>AóP"•&¶9ÞÄ?ûSN$<3†é㽎ÃÉþ0ÓM§ËO¬s—ðS¤Ì +¦3ÀÇo#úLXä7³næ%þ¯Û<¯Ã2¼Mµ¸Í +W|Jö§ÝƒÕÙÚZX´68Ï' ›þYÆ[gì`ûVX‡#/ì~»µ&ªFŠ/ùÂËzpÑÿŽÖÖy$¶¿â—çùI*‚bŒL#‘L}+Ûç{îëxͯ£"Šo]‡‚Oq#U¼ÖÝÊû$H“¡|IeÝéšX¿-ÓþDú]u¦½ïŸ‘Këâ¼Ä2éUÙjh óU»ëwËÎ KC©¥‘6²i|Þ¸êýÙ+Ò)mŽ çaá¬a䚈1¦»¹ˆÜ×(Ü/¢F^Å‹?{þãÔ¥íSáQ`6ï$…›þ¾ö¹‰Bù„Y1Jl(Bqc&Ä­E5×^]ÕD±€ÙÄ¿‘GP¼†ÄNú`J»ÆY”¢k-±‚ðK”y|ô+ÊF’P‰Ey¾ÇGØuöø¹¾Èþ*éÅ€<ÉmT>Í"v¹@U¦©¤KªQµ3#ûL_ª;ü¼\°SòÑÌWfy2"_Œ fbw3÷³ã¨U¡ù^-˜uH½‚ ´P×áUÝ”¨ê} 8úk§Ââ'Oìù$ª;ɤ¡áL¦_æ‰14.ûjóìw¦Œ% h¡ICDøµY4ýNëÕIŠf¥+Ÿ.× y±Þ‹ g´øò³MY€-Ý_íL률å4::¤Ž£UpÀ6›ZÓZß:u ­Ó.&lºêàe *‘/=ûQÃm‡•{:YŸËæ3IõìqHGá9¨S tz²ó*ÊÞ+Øó2÷Úà÷»dcÇ6íþï|­àcÈ´ŠuGZH;v2P÷é,$mî¼>Rβ˜ë¯¿†÷W;æI_EÀûïgj#ûBöøñaçI"HKÚPeµÕ€Ì Ag¬+ GÁП®øï¿~üÃúm¶PHéÑu®è ÿyS’ž¥Y‰¤×ªòák/ÿ‘ŽÛPq'eq“íâ|8Úb1,?®Õ®Üâ‡jî:â–ðõÿø,TFŠf?ÔOZ‘ØÓÿ¤ü;ÏþT‘†¯\r/ïÞ¹–%1­œü%‹EšÐvùÎQÆQøMG?:ÂÉ×û2Zé¡k½ì™ú¨ô*Ÿt{Ò§õR¿?zËa†|vrÀðR·UW„2Œfw©|´ ã1ž\ ª0™›vÒ6È ¦7#kItÃ`.LbºW{æ_$“*YØÂ"SÝåÆt5Xоİ"Ûo&ùVën¿‹Òûëìéîß_€Í`•'…wÆ`ûÙ¦¯W`÷÷ôËý»ËHðq½¤ø ¯ìïäúƒÄ†ó®F瑤Fµ"j± ÉNÖŽÏZå!9ñÀwo¤˜5ËË]HUè¥ÍÞ7Ž4WË"ú5‘¾:•ê ýmóÏ’‘áQJT@ ãïŒ|PÚDf7Ÿ©.ÎÈ*JtD†ÙÙL}‰j…“=lã/Õ1äú»½åú¹¿ë„n™CÞ[ &g‘¯ˆTˆ¸jùÛ»HN_Ūõ=Î(¸–} ÚHE„ïLÞœ F±4kŽ´À”4rªè)eG©;dµî•œ(#Y¯Qž‡2‹5¶ô¯¡ìnHe¹×}”W>ûHÊò5*äï(@¥Kö]nV¦¨â_d`&xÕâšgÜYÃCƒ¦*IÙˆªT?›ïç¢fÃ8Ceüjsp|îÑdC•Æ.7Q×YgKá2ê]yWCÊZ†úQA|6Zhð–û³#éM4ü<æõ}~%…m´ ñ„reME4š’Ï0^¾ˆf²G+¯Ää£ùñÕˆ%ÐâíužwÐ28@^8Ø­$øo âù„ï¡R2øUAÚ–÷g}0®G1å_½§¿´ëziÿÍc£? y§ÿúø_­‚¬âR}òý{95iu^´Ò—ܹ×"ü?÷¯ew¹íñ_:©ìgnx§tk{þŽ$:EŸ (€Ä‚äoU®Tÿþó5¯r:¶šif¡ô,åäE?:ø–>Tª|ªßË:Ì/þ…_%ÄŒŠ[{|ˆ#Ll ÐdžB{{š­JŽ1¥B‹ù_ÚQ'hÙøõ^lŸ´ró×ÜvÍÛã¿$<çBËÃÒt#­oТ¡?ê,W-ä:tZ+¯þ»ÿЈ‚¾æ´Ä ¥E¶àÆýOH]¿ÿ¢ö¡x¤ ºh•FPƒ ®-ÕUȘhiŸ8ã„̲Î?ß–"ˤ$ãÇ !dëpZ¾!}9)züt’ÇýWïž&›F^qm™eÈOà”lªA^r4ÆðÀnÏ%¶óQ(ôþ¥¦yX!Š„ªIU³F£Ø•Ãçè£Df™ßlŸ JñD°PÓÑ ÌO¾9ç”#ŽòÓ×»òÙáu¡¨èu·w;¤•¥×ÄÈËÏ ªˆ0O‰b1ªŸvÎIé’CŸÊ‹mž‡Q“kr˜—IµŽå8ß ¾ŽÚß¹©¯ŸÝF]'[ÛÀ3º¨ÏöAðÑñWh0¿Ù¨B«´§# ÛÛbahòªàP÷%4;7:rE?-ÄE•›½ÍÐò<ƒKÀZ§t‘ŽT µè¾œÝºR´!çz~ m¦Ïækîù®m‹Žì'ê§xøë—2Jv´Ë»ÃÆ^hŸ~q.ë(:|þ^nրޭF£–¡Qè´ÃÝüÔ=I²nær£ÌÀüK:3$·Õý"$ƒ:—„öá+…ÝÊFŒÕ¿>y:ûÂ(È õŽie)ƒäo­wz@’ˆ‘uòƒ2xoÙ¨—Þof~ŒÁkÿ²…0YxÙ—KZšë /Èu}&ô@ü¡#áýÿ«Ç÷W;[díéן¦dš²ô•Bn”(C®ù«uVzøü¼Sýo9äߺ@7Ä ù[u„2R½Pà{¡7À• Æ9Éú»Þý«'šožØãÄòP:¹ÿs-" únè2ù¯^RÐåá_öIø…³ÒA‰å«ÞõÌ(Èœ½{O¾:gò‡O€ò3¤TÌbÿú:B~jÔBåÂK±$þx¨úòöœÒCbø‘F4úØ.ª^_ѹÿêøfëȵ þw|“úïÐ8žQö=š£¯ñKX@k€_NËÅ%h¿ö¹…9t/g׈O˜(@÷éªáûÁEÐç SôÔ20PcÿèÈo_Šxu5Úè+Œ¸¼1Ì«ƒ1áÙW#y0Æ/ÅNõ¦~GOù÷šÀìátu°8 óòÞ \°pÎÍ×ôt,ÊWMÊäÀrcÐYÂ"9Xms=¶r˜6ÈŸUd¨÷ÀÖ‰f}á_?`gž#Ùæ0˜G{~Ùãü"‡ãÎ"±DÈ!Y$ù*´Â@¨ˆd·".Ø ‡!ùã2Š©"H¹ò†D#M¢Ï{¦Òi|ì& F†I– ÂðgÈ”¶ßn£ŽY±Ê&=¨B¶8ÞSw¹È£#²ÿºé>Üúá$k8ò¦Ž·‡ ŸBþ܈kmñ@Ë%o®Éž{f',Š¢ù•Œ–6 (q}4ܶî%J=ÖÉä[zвäCQ|—Qž$MQ¯!9Sf掣²j·âñúpT !É4¾Œ‡x^ÿíiðE”vÈç5¬C-•R’ÒVÔîÛ¦ò¯ÈDÝû:gæ¸õPßæ›géÎe4û,ïƒF›©Ó4&hRwgÒ#Ín¹}ÛÜËÉ–™^‹Ž½hù¨³«ü--Z…Ýü؇zhp8Ÿ¶è7ÚøÎÌPW¨£íÅÖr¡_<|Û¯ïýŸ_h—ò4ÕwÛí»Ÿô—WN¢ã8!b; \û“¤™ŠZkýõWAº·!÷céâpãEÈæXu±òr…OÖ‡,_­ÏBªÿÌå²û&wÌ+M¿zÏO;_ßPÿWO·ýÈ•idbjuiP–ëßEÞ@£êR»¢óhQ¾¶Þx, Ú ”ó,\ŸAG¥$©Èö twÔ™‹VtB?™Z\ÈÃ5˜ªŒÕx ƒ Ò£g»6aè=Se˜Ë=ÿ‘Ùyò¯~À؇ F:®˜JgÃ诓7ÝoóÀØè·>Êù9˜ ^¾˜à?“R¦Lörñ0åq4ytƦ³…™µ–È`VQZÖ!æúJŸw{óνré2¬°`Ñôã|áž_™’ÊÝp†%£¼Kž¶—`ÙùÚ÷+·Ä`åaPåè$¬.„Ñ=û« ë·¸ê )”aÓþ›½×45l;ªÇdpÁîm®¯ß#¹ h%7¸£ +—×õô˜gŒÔŸ—I®½ºNÂ…¤+œ”D'.à¾{ Ç^E ëçúÖç½ÊÀé"Ò8<3P#9‡tÁq¾·©o!}Kò /2,F]3~L¼ì¹v>ÇÙáé–ÂdùÞÎ+Ù˜%F£ß\Gö„¥W$¡È©$­î) „\­Â4£ Èx]¤àòÊøÙå)ùÿñaJÿyk=˜) ’×ÝãÃÈÛÝÊÏP¨ çÜ=ÎM¡Í÷·@Q£ÉŽa™k(v-ȘW,ÅÛ·§íô¸QòðR³Ž$”rpß÷e¤)®e¯ÜDÙ]ë–[(/øMÇSš~iÝ9{i•à†§$ *;G„O;Ú¡Š·@I·ª~}Y«¾pÕ¡Qû¶´jìã9ÉLJˆÈÝHEß¹Œš.ÅÞ· ÖËšE´ êÈ<Šm¤OEÝ«xµ¥N õŠö¹›‰i¡þæMÉ‚Š-4$ªß/`ü¨q¡í>³qôiÍ£‰`~á—å4U ©A33EÀ4?Ó¶$˜ƒQÍCƒ–n1zMIh¹»íëó7­ÞÕßìZCk–ã#hCmê§w”mšÜ0b&/»”¦òËà!sý¶í¤ d=^4ûV¯©Í‡›µBíõ"Õ*¨8\MœàÅnKCnŒ’ƒË®þ¿zý¹ã !“ùȘ`É(¤›±l–ARîYuFOóçeTìßîùeÝú¬¥å¿:âOýXHf¼qó§4ç?žä(¡½ÿïœ ƒR5«¯éä÷ƒ9¸Kr ™ùxƒâ¯Ñ.ã>PzýAÿ—ë¤ðµÉ¥Ï]å5T„ÿˆ5óoø§óƹ(àGħ”+ÔP}àS£‚ó|HíÛ/BÙÇÙñÎyh|JoM°ËM5Úý˜?@ Ñ1¶3‘Ðz¤•€^ÚÊ ëÚá·Eÿ}rèd¹ËB$lÝû¹ç­A¯é¾‘µ~ÿøTßE=ý ;Tó·Y†»~ºW2Ü„Ñ{W”ë‚Laܵ^ñì{Q˜ôtK—ñƒé¡RÁñ0ÇØ~½và0Ì»Y2Ξ{ ×ÏXÒ…%ë÷_âV—ay %Æ[V¯³F|+‡uPå÷ŒaS¬Ÿ=À_¶uh˜S‡ˆ`7¨ûDî7I$ð<þ‚Œÿf&Ò±qé"1ç ݇HòÆÞ7,SÉtt‡žÁ}»ÛÞ)AŠÚDªçwÚ*áÉ+gn¤9Û÷5è%)Ò©”òw"ý½Aíã!dxJQ÷»lÊÍ××ÙñÍŽ‹*3²2•зÿ A¶•O”Ü›1È1˼p * ¹šbïȸ!÷ó—®SÓSÈ{”¬êΰ ò}µ}æiŽüåþSña½xà@â*C°) öÉZ:¼ ERZ*™Ë(ºÁoÑ–Ž’ô”Û«PÊ|û¦Ù¼7ÊPÆ ø¢Ì°ü#…¼q”]UJ4ð=òŠç$ÌQ!?çKßò+T ËvÑr’E•5k‰ P-'l÷Æú_Ô ®~[Ìúñ°ýó§/ï£f[L؇ Ôfç^ºzutêcÊ­Q7`Y¶ÊõÒ+® ~²Ù^GƒÍçƒL'–ÑHlIÛ.¼¹¼¥›”@“7‚•>½CÓ? 7htMÐÜXûÕY~~´øüñô;&%´ŒŸ¼òej­N›kû„Öº¾IzZÐÛÕޣˇ­}¾nŸBš.+öÔšhûuÑr;‚*„?eÈÉØBÑÙ“7âRþîùÝ·(Å(ÈR ´è°˜‚O?úå )áã‚ Ç›iú^|-H ’†/.HÓ·_'ouR¼‡DfB?¬Äs2Nõ¶îÛ{=ˆ#x»rn½©"§¿ %½¿[Ky;þgœ_G;Tœ_»ð ~™‰Z6,@âò”(ò@ã’‹Â04õM5E콯¹&ô‹x%´xùOL¿-O_²Žq~†–ËÂßfÞ«@‹l8ÁEh^×¼Æñš;…YI©÷~ŽùZ]ó¯ß·,€nóvËõR}öø’IÙö6´¾N°:Ϫm:Ï,›ê ï:Ù«m[ø-н:BÜö>’9t6H%0‰¼„n µ™NR„žÒ'&6UÐ+ج¦Çyx©ìoÁ`’ÿÃü¨fæ yŸÖ?#µèêm|Æ ï@ë:˜Xh¤ùËß Ó—ï ÍoVÀœ!‡Ögqa˜Of=Ý ú&sæØa™ýÈźë°Ê¶Å´raݪZ§ÿ=lÖ¶ÊI§…Sí¿‡¦€óè¸S›¦- ²Bâàûé.óôòJ…æQ%Ü—X’7W@C¾ÞVÝHmM«Cúñ&ÒÎN•o+6#}—LnóÚ9d4“ýq![™Ú¾ 2p… ËM’¯Íȃl^öÙU†È›å}^÷“¶ø€<©âÉw¼Å‘¯ÔºÝ6P@‘މcË.É2Hç¢+ü)åÿŒ"rt×øîÞ@1©Ï©Q∮}±ü3qeNe ® ½["GÙNöÛ³oP>â鎨8*·´$ß*mÝ*lö@•¹—k˪€êÒ~Á¡<¨­÷q&Ú5I’¯ß—F-‹s1o vá­c¨k“+îè‰ú̵ Ð`.íÓ»È4úÉ^db<€&±…Ýtêhv–·ÿn\"Z$‰ßw!-ýÏW ˜A+»”ùÐåh­ñ(ë‰|Úˆ[/-kE[î¢ïõx˜ÃÀK÷¨Úñx'ÌE{étÞC\è`ö¥Ë,ì:†´‡¬K( SYúéõ°»èÂé.oÝ9ždgelCÄPºÁ[þÜ>7Üw©Â^RÅiû¯L;AôRi…\e³(½œ©:^\8ÉŽÁ¥=ðáúˆI{[x÷QøAa’,¼qcü@ÑωՃfÀK«èàÚƒJpù€V þÄ?œvP"’ü_=þ;Îϳiœ­R-wþõõ¿!jL€œmý<†D"ølxe O‡C¾|MS¿ä§øÒ*è¶C/[n°¾<)M6'þç›åbйPHybâÂÊÿ{-è‰~"»÷\qyõFŠëì:®üÉ„’ƒ6êá…¡´ZÔDŠýßܪ¬àÀ](wȸ´WFÓO¯¾ÊŠ;ÏÏªËãF-žOà‡uQ~È~¨öðÛ?¶”µY,þCJ|PoÇZyFˆã÷U®t«C³²šñ=´Ò»«*ڪקO˜C‡ëlôóþеÒ’|ü ôŽ:ú[ž7†HÂÆR[´m󑄪뜵’aŒ<Øè ǘýÂÅdSmÆ9•â0K´xbªå)üIý˜ÇùÑtÔ£;šÇ`±s7ðûÏXXþ(÷µA—V³ïû”~·€õ!Þ@K[ØRf¶®çQƒBa~ò$ jÈ`AÂ8}aÞÓ=HLc/—+‡$¯òuGš{ì8£pÐì’{ñYI]DÊa{—‹¶HsÁüðîÃe¤—}8)˜ë žgnø#S”éǸ׾Èâ´®üô%²ùœ¸I^Œɳƒ †Ô¸_»úÆ/äu¿ó1/ãòG¶&Üï‚l2ø}Ph’'tVwEÏ?¹«Ët %d„/5ÝMD)%ÞNb[”i»uó²ÊýâåuˆõB…ä6£¨L,5d讉ªÆìlãeú¨¾ØvZó8"…Çæû¯¨%2¸é& ŠÚõ?ŸþaãBÝÐWLZÁ´¨¯g)>.„†|·îs¢ÑZL^EÚ'4ùYX_øCÍb(u޲£…÷O!ÂL´Lh±¦¿…V)¼ûJdÐ:3p-HWm*b#T=Ðv4. +íØMY¾]]Cû£U!ª²èPæq^3=4hVHM1L=Kµø j¹‹ÕñPä8Wûf¾álšiéŸrY…Fùõµ>k-ø~ÊÁŠ™ä”×{9*_Bzsí¢»¡0Œ™Í'‹>ß8lK{tnO¯œy…Qsÿ3GñÃ} H?¹Å 3)ºGøÉ!U:~=ýo&dèíñüëÇ­Þ×Òß8Û¨N:óA-Šþ„4¶m7öð¸AËy +¥;´W$Jù.õAgàÈvï@ÿNQ‚¹ôG}_Íu‡AjxÐt†¸½d¢ }ß-ØðSácí­ !u{ûØoäÚ–Œt®-+Y™Âhë>Bßï0¶5Þ$Æöio_w»Ñ`)Lº] g¬õ„©÷ŽVÒè 3Þõ‰q0ûÐBo+½þèhüÎQI†ùG-Òé’°p³Î)Öý ,†Þ)aTx K¾©;èËAG‡Òo¬$†ÕdçÂêŠÂaî$X¿„Å×6Õ~ï;[ ÛË«#¤ì°k–yÇ£] lj&BÞ#aŒþ§‡œH4f'׎$šß=K‹?!é'•cOH q^› #‰E ¶T‰‹_réz¼øÍHC¸Å¬N¿Ç‡ì<<”ºHŸB*ÑF Õ#»{9qÝ‹ùüÙpd–íöëm¨G–ÐæN—‡ÓÈ:•Øô¾ý²Þ7—Ï{œlV–|ncÈUö¹aQW¹ÏnÎl|çD^i›fÙÿV\äcFéªèq2í=><ÿ¡iô: å‹]ÿÛƒ"LcžÖ-(jÝÃh¡Rb÷‰– æ1Н„…ºT¢äqÿ$Ç¢PJ~Cëá$Ê(»C溂²1YÃ¥O?¡¼èŠXg*Ôyëï©„JʾJ„6WPÙ|cÀiÓUœÎ—óL¡jÒýcTì pt‹ÂC+jLwY÷Ïë?c%AÍ’Þw³uP›pÚÈ#Í uœ?çŒ[3£îû~¹Ôû59ÄRˈú;o:Ú¡!£ËöÔ y4’h8¿ôý¶…³ýÐä|Õ½æhú½— bѬU“j•9-X+mF% ¥¼´ìè£i´úÎO.£Vñ#«…úhíz$Ûñ59Ú¼ ÍЖRß xÐvÍpˆåÒ8^öþ|8î¤.ª«GÅt¯GÖƒŽj9»®°2ýBfŸ”)Ô>~{ûšá¿9NÅ·‚ÿBÑ õÃG¾Í{zãpzà ÙñÎuöeg!#‡ól}ð¥½\*Ÿ—9F Iç:Zp,ÃêÓ&þùa!;¼s­"eÎÜË¥åÄ­äö|°yóøÏÞ|ùl9@v»¤´ïK†óÚ^ž Úó»’¯~´dÿæ«Í[ø7§p ¬ákÕS×éñ;ÿ|J5±Yø?>\xJAñ~ÜäÌö¸ ªÅ½`É)ûuLË_5Cƒàþa9&Bh,¸ëôš¶-³KNóA‹¦rúQѽ\úÜ›æ$3´ïS‹gü ¿£>)´¬’B§ñcÛ +{94ÈôÝ“dèuKñtˆü_>ØwX "€!+’ÓÊôa¸<Ï/-¾FO (Ÿu} ã¢òßa’‡¥™aâLkú‘:|ʃÙ÷×8ŠBÓ`^`ïÝÝeX¸¯âÐà›K”‚kóË£°Kïð+ã3¬ªœŽ¼ªóÖÖßý§]7d܃­™5–ݰËQ{Ðô¶(qj™Gˆ a¨ áaÎçH4ÝåÄÆõ I.ðùœËD2Þ‘×å)|¸¯GãÐu3¤HYîyÖˆTW¿l[ü¶Fû †-1¤“+žÜ{Ø"}”_ýÒl62¼¼xõ@Ä"2ö¬Õ-ZFfë6º>{wdù{ñqõ‘)dk¥{«*N†åɆ,ËÈõV†ýA;r{ohœ#E^åý§$:g/z%\š: ùÓ¿mÉ=Sô*T¶¬€‚ß2Cyæ;Pxø‘d7¹/ŠY&Š.+D¡ÄïëM. Ó"ªrW ôÈΨm¤8Ê´$Ô;t˜¡ìì|ȯ@k”—÷®ò~¡ E|¬ª{>¥AÝ'pœU¢»“–ÅHQ­AÚÓájgmiÜ2Eô}ëåšq 5»uþܶ%Cm6yïÒ_¨cþ‹ÂáïÔzç5†zÝß.^£4`$£Ê:%‰†ÖìT¿¾{ QÂf§¦öW4^a¨Øßަ. •Ï?£Y›ÒÒ3w^´8.tÙþã]´ ÏlÌ´:É0tçñZ;0§ÛR£Í‘òÚÛdñœ&ÃO¡˜H”ß~+Q ;‰ÔåõÅŸ_ÀØá‘²ÓÏÔ ü~VϯçPøöØÎÖÈMÈ™¿s=mÚ2 韟‰+€ô*VïkS´6÷$¹à’ ¤n>=IJ I"Ƨ1ÑÞš˜Ò—GÎë绤Ý4ÿꈔ«ÆDÿæé_ŽOÿîá. ´¢€Ëw_IêörßΉ]±K¢ðsâ÷­ÅH¨ë5w;ûv<¦¨Å-H ñ)Ûl~ 4y±f}zIÍzÛžç·Š ¹¿„œM>´0s×nß[ÉÍouÓCs‹‘Í]1nhÎøŸDðšS)¸v™ƒ ydî>§‚ÛFoÞahi?nêÚùZí7Ø«éî@ëßÑÊò[ÐöºÓ¢i‘Ú#ž.œú§ÇÜRêö§Ð1pç§w£t‰‡OK>£î{œ_×y:¡gã¦{þ:ôå¼|Û<~ ލ¤]sš„Á{S‘4…f0´ð­y­tFT¾`w솱óå¢êZaâùgŸ’.G˜¦ûÂ6‘³%òŠúÇÓaÞü݃î²Ï°00%ö¢–Þp>•E +ñQa[aGa­öõÃW¯íù«·º;=/lל+±(É„¿‰ü•ê‘Ю¼ƒîÊ  TÞ¶I@R† Âã{~ºý¥jðRÈÜf%GªvÚêýç‘ö¸«œÙ™n¤@Ó“Ä‚ ›LʺA,ÈtIðjÚcUdáëõ^'uD6[}gqäôØe§G®»îÚšnÈ#Î$7¢‹|!ÆëÒŸT‘!kYØbPLPþéD!ÖþW+²Qxl¤%ˆÀEûYM&˜LP‚¢Z(o^xW¼ðKeöÇ‹;ß AÙO¯5Ü£¼KpÀK¯Q‘îS˜s*ýàK83|U>³5Iª@µ•-y#>JÔp±9E\¼§¯®z†)ÔÒ©o}®Úé"%þñ¨kÆLmøõÍ.ØD4¢ÁŒË}âÓhT±ÌG㥅&÷‹¬»¤5Ñì´c×Ó‰(´Ø=ø -ƒëmSÐÊ=•l•äZÛpeˆ=ßEӻƮöhkIQ˜°‡š|èʪC»ý º¾–hŸ\Ö¿\ƒã+‡>Z”¢Ó¡jyiògèœF?“xãQþpO’ë1žælÊ8 «ˆ{¶Ä‹…E‘ì5.ŠV߇¤Ë7Òõ  ô,ƒ—½)äDŸÓ9¥鱤ªo]YÿÝÛÄ«Ïc ï{—ÍÄy'áݾáîÄÇðºùO»i!îñßËI›WáÅ7ŸÙßè ~Žˆ‘ÄàÄ˶ٟ \þ_=¾«LÙœwTÜóQ‹I±¸=}ÓDü!‚œ”Ù 4Òsÿî)}¦}Ô>ž_g둆|{V.§ȯûr7TÊ ´Ž•»CAä¦Â>_âó9l® ýŸk¡¿ábOm0ÉDžÑï_ÙóÙëÁÕVPBIöŒ.å,”f½Jž¸½ _uE´.\9唦UƒUPq,¸"#*_¤Rš ƒ*u a]!ø¾yá¶ÇT×½ãøjµ4¸¿ˆ½…_­òž–YC£É‰?ÐÔ½äÕÓÛ -Iî…öæ{úîå©03„ßs››f„9ÐuýK¥™7ôž÷Ûj>zéž7 Î鉳Ç>…á¢cÓÉDá0ZPû¾Ã¼ƇSy¸¨`J…-u1…f2•£âì›áí’½}Q ,p°æžÐ`‚Åìj‘wnÁ°1ËšðkV~¹©üÖ‹ÃB4<~Á¹ÝÑï°s]îÙæùsHÀxd\ô +U Nª™#ÑÑM…Ó HrÕðIÁ»p$Ä®ø¼“H®{-N2m/¿,8ßþcu%>œ¿ôCBÞÇ=!ÿKI¸2„)þD«îÒo‘Í(-=|Õ9‚­‡Ætéqÿ¾™×¤Ÿ*—óàõ©ËŸ‘ÿ¿DŠñ.8¨ªíHBiæîáÇSQTä$‰r‹Š÷žò57~ƒ;¶‰’˜èQæø¥ö=ýÉÅ<¦8{D’J¶ÖX/¢RÝõ‡’+[¨Ê.q5ü/ªÿ o|4„0ì B'sP‹uãOñÍÔ.ö<{ uÏm?ií@}¥î:ƒ4dÛì,NÏE£†¯ÏÑäWã±4{›íZx«-îõ÷:§„£eÕxíPèZ5¤Èªs¢õôSåÁ_Ùh˲ªE˜yÛ8‰k¼hD»$AÓaz~t` ¸H…ŽoåwËâ„ÑÙ.ü¦hì ³>Îüý*jV‘‡ŸeÈqäggî»AظvãàY×$h¤¬ ÊQ[ªV¥î÷5ìP~æKž6§b¤ „‚†§·‰L¨öô"Jþî²5ä”=´ò€,}•ý[ŸöÃÇË…Mmr=¬õÔþ»ÕQH‘lÏ,ˆƒ,©Œ×Ž ÿ£ƒ ­GðíÎÛÙG“1P»“qñG 4] ­½>h­©es8à÷Mò¤'·ìùÏ‹åÇ QÐË -B×µhRRWÍO SýØ@˜, %_š{{†&&?]¸ Ã!©Z\0ÂØ2QxëŒtû&Zz£="E‰NÞ0ÎHùͼa®Âð e?ÇòÎß%}ZM]B)í(·’”ÑhŠvüye"$•TQþ 0=™5*ÔÆþ*C%™£²—6PYÛK½&þªF P4\FÕ!Û}l¨NÙß(öh -Ó21ˆíñ!ýuþ=Ô¬s}àSšƒÚü_ヿE¢Î­îÆÝÇ¡¨[we4‰™ õëŽõy Á×ðÅ@4äŸî;zL\çø‰HÐ8´ž“VѤ¨åèÄs4#±¼}jÙ͆’³ï¡EEªpGàM´|±r܉„­î¯fOÕºñ‹Ð|iFÀåÀè7j¶;c -N¹ûk>Ÿ„–Õ¿ƒ(¸ 5ºo“—þ.´i¾óf÷j„vv;F7hŸËè?– ¿×ê)Ù½¡S7w¾úU0tÝìVÓ§ îŽ)’ØPèUüè¾_úšØ|Í aÀÞFyÍÕ/¸…¬†¡o4:/aÄÚ »ÂÓÆ„ÖŽè÷À„NBKHtL½v o…Y93Ñö ðgê„gcú5X¸ØíuÁ,–äþ 1Рï`ýb%¬éwg8ŽÀF ¹î“À~Ø>HòÖöi ìNê±+UL!!aÔ=²«HT»qòÏçh$©Nqœ‹C²~õÊ”#H±ïO™}—"R9.tÝ»çˆ4Mo^¾Az­ÖÀ1»dø°œÄº·™x"»õ)×ùËgnîÚ'ÈúÄJJdٳ䋞î"ƒYVÉ.äNÙ*ÑØ!D>5NÊ~EäÿdÛ&÷îá{E’±‹‚ aæ/Žg p|‘PCÀ } ôä;Š6ç¹Ï_†÷ç<ôøèˆÒç]uP6\ìO~ëž¾gûjåPa›îzê¢Rtžײª„F¥ºkê¡Ú—š%öŠUÔ>ÆÏ9b€øÑäúÓ:bÔ"¿œÚ‰é,³9E¨«ïמ¨œƒút¯9žM ÁÍÙÐ+DhTÒp³áþk4 />Òi¡‡f'·}P C ¹’c¾—~ åUÉÑ¥€V´òNåöf>‡Ö§ß,#âAïoQoœºÐöºCQ-×k<üÞbЗäÚuÖÎ ^žA—Ì[ —ÐñúÁWTö‚Df"§ðO©ðBÙâØÏÄ» EªÀó–2»—Ó_þWïNDS—=þWÏxâ³û³2/Óx®‚û'ÒäeÝ÷¶ÆPÏŽ >?dQçê<ù$Å¢‡|~Bþ¥cî×X]!~ßä”BÅ'ºýh'>•)ýŸkáÁ, ¢r(\ÉÕÞáþü¯nòÆÿÍ|+\Ê‚ÒçldqäðUŽéìÍJ(W<îóÖð TèÒÒ¦ž¶„Jã—w›áÛæybž<øþFÜÙ,Y ~&Øx,{ 5Mßc¹Eà— uÒÇ8hXçêøFAMa£ójž@‹¦†š¤þ´ÉÎë>SÔßWÉýuY„.zâO}Êгñ ²I1ˆíÌ MaðuÜ_çböÔ*¨tƒQOÂåµ0ãÝÁ/P“NT²¦`æHH±N5ÌýUÝxò3æçÚ’4añîãŽkë.°|<ÉëŒ}%¬zM?°J®ƒõ'K$Áú°ÙùÔkììYܯvzø;aE,.…„Nºr¨Ê¨ñ¯£Ò~%$qí¹(dûE”†t"ùþ¡Þ­·ÒÏ8ÒŠÍiX)öÕX0"=‘±F'dxEov‹D ™œÒí[Ly‘EnÚ´@6…Æk#BÈa”?òT“¹~Ñ•†ê²!OS£#¯A/òËzÓ~¦lÁÊþªŒÁ¸ ¢zü!Ur³x„’ÐkÉI¨EÆ0Ê:шڟŽå —-¢®«Ýñ ™7¨/z*]Ùë/ÒqKNië ÑâýZ߉@4iÞµsºfÙÎÏD EêZL”•=ZöÏ}»0€V›Ö"N…¤h#\dbχ¶§6§CD>ãáòý}ÕQah¯¸N˜hŒµ'¶¢²GÐ)‚ø„}C—ó[,&æy ÞäöÔ” ±’”"B³ïôA—xa=¯Šù¼›4T²"½5Ti±œT~˜åd‡shw ˜ÄáØ¯pF(àÝ¿ùxde/Gn–¨Å~†“fõ ÝÙã°cY¿ iQAÑçZ ZîŸ}nÉâ ô9¾W s‚sý³Èð¿:zÖÔó½}}¡¡$ˆêz]ó =k¡™–ö»Ø;shó`ùõÚ¹: |ùòyNCWŸÝŒÑôô¦»=¾qâ% Xÿ銶J€ÁèÖ‡rR›00R•¹ï S¶¾Nä‚á»aæòg`„S~f¡Fz¦uubÚat8ò0ê°Â¸èÁü'¤`ârŠÌ\[L~>q{‡¦EL_—>†™:.ÍE0§óРÝAæÉxÄÄü½a~§L,ý¾,,”‡ ç^‚Å~)ɦ8X*<,ÄâËÙû¹¨BOÃJET‡ÖëXÝNèÿëtÖÝ#b˜¾™Á&ÁNHù9Øjòd Ë¢‡ö.î׎ E€åçPÅ”Kà;ì/ÊÄSH\Æ âíHÊ@Ÿ¯µ„dÜ ¿4‘"¹›©oo¤RšT¦> ÕFjÛÒO¢^HëÇqÑ´émÞ-F† [ Ô^"ccëëcµÈ4âvBz@Y$)ÒO·ïù[Ü”:,d+ûs„­9:c‚ôy"WlÇá6"ä¶›¹b)‚¼bÒuœÿñarmä­â,˜l+ûöQs5Ê´£P©×kŸê7(Âc8&ú© E=¾Þ1ÒE±wìzW¨«Q|g,µö-J^Ýðõ;ŽRV"W§ÓÂQFOñyºl+ʾ=Zô¬¥jÏíVP¡æGÉ•‘T3;ØðÆ•ûâÔ%¡Šbu«u5ªx±dˆS£ÚüxìFå<ÔØ7ħÈ%S’ò@5»ú/Ö@m­˜Ù½Y¨“M0ÇÒ:†ºKdi\† ï›÷e/ƒìÚ‰– ~ÈU},·Í9sr€›%|scâL€KŽ>ÞyÓPšN{hK÷ ”æ Ò¿×ö¯n:§J$Be”]‰QÂ5¨V8 x` ä_?ä¦ôÝ@YKòÁä˜i¤š~©Úó †½>Åh~L…RÇ…U A(`°Ÿÿâ¶Í ÓãA»ùmnBq›¥ÂÛOÁ# o5¶ù#­âUûÏ{¥_-oó7úÍ44ï?û'†ß ¸ XIöšTx±ºœhvb’«¥ø+¬} ûáÓØ…f(V„'áBŠPnòãÆãâZ¨’¸œK5Oß6£ÒŒA½‘™…”Oãv>s—”~)†Æ¿ hR_+ÿ}ƒ š¾ÏR‹§…æ]ôëfEÐÌG~@¯ š™[Öêè¶½ï™å|ý!h6n)’³‚æÏ‘Þ¼ª÷¡Åõé·³w à;‹„û#e_ø^÷åÇÙ=Û~&x’•Ú\8}ß·í…¶¿ÛA‡Ë¨‰.t¾ùÉ&àóºúeïº4Ý‚¶Ï÷ìæ…žSG§ƒœ¡w&\Ÿ6wú,O~9éöú´QéÂÀ[ɇ—Eaˆ©–Ý; †»¥åKatꆢš±L(³‡Ç9S¯pzµÁ/¥k¹ã?"an÷wƒ¢Mi˜/Uæ‹ûÆ‹¯ä8Ç7s`ù“•ðSy=XcI4­èyëŸßq8­´o¯˜“#AÚ!ÜüH¸«ÒÚ…l·‘Ä÷†`vž>’]¹XsFÞ)¢ÓçkãÜ‘ªg0Ù|CiÌté}ŸBÚI“Œ¾íÜÚcÝÆ»‰ï/*ºÑÖ‘Ú¢².å(í*T½=æ9ˆê³>?öÑh!^®úìö®÷³Æhñ­¡Öã}>^ß’P[ÒÒ…ÚQz±”)õGK ÇWŸ£aAêä£ B4¾T|´ã”1šº×5¾õCs…¨/ÉWfÑ"øîy³Uhy1UnDÀ­®H§ÚC´Ûtù}Àmr^XÓ½‹¶“6œäñ€Ò¥”+!·ÑþùÚµ=)¢xÓ‹ðØ :§¿ˆ)wGçkí ³w.ã!iÞ¸ôÒŽK+ýn‚¢«Llç"ÉŒšïä,üÌ:Â~”qdgÿੇ@¶!Ù^iÆ>HãzÞãŸìÉ'XŽôhëÇ/ôßfÑCb‰èƒ½ö_ªØ|ò¼PÿöôD ÅÎ~¥­³O?ïxá{ƒ6ˆ>›bÛâü—ÇWä¹Ô@Jáb8ãQÈŠ ¦ªÝYw1Eb^¹–§V Lê!O½èÊÞ6ÈËvUÐ )ƒ|2•à?þ›:ý†ò׿±•&nÖ¢gýï×Q¶½~ª”;ï™r¼w<÷Ÿ}LO†oÿ'm„VvÖsxèÞz ¥qn÷á3ñã  w(JDc^(W0ŸªÈ÷€/…Ù?SÖô Âp@RÑGª”µ#š_@Û½S¾t©PÛt+ji¡øeöY@“ gÁƒ|[hîßóBñ |/#>îË¥í´LÕY²Ð˪¿ÆË Ý7Rö<‹„Ÿ× 920B¿ÍÄ÷¸ÁS0ÈÊû 6Xm;W}]è šaT_ÎK@ÌÆŸÅœð=SSTŸ‡g]õ`&%èyS[Ì~©]¹²ç$Ì{¸Ü9°ˆneJý±°lÄËÁ¾ V}Ž(|»ÿ ~§v|¡ ƒMFã» ê9°•fÕ’K'ƒ„jå|=2H”ÝZ Ô¨„$pBÞU¿IW³µúuŒ‘|IG æ%RéwÇ;b…Ô£#®™R HÛJ£N|=÷„þDL‚ ØóÎ÷<2Ñ'æ‰ ó†Ò•¦Ndcç1RgGŽð}?;½‘ûÎiiFä%-š+«IA~b’£U((.ž ŽÂO)–÷Z뢘Ҹ‹~U2JrçØJ2jþžw¿þÄ}B{Â&ÆP^alZ#€ ‰Ž-sFå+³OWÒQíÔ©o>„§w¾Š?GÍ%Ñ.ÔŠ“ÚïÂèŒ:fÙÔj÷¼QmÁçP¸ak[3ÎR¶†t¡qO´Ø/)4ýL¼PC€æÕ´¦_ÖÄÐbmé#±ý4ZñM&+΢µ[WANß}´)ÜŠ\ŽëE;©Ÿ¶7dðÀ—ïxdΠÃ5[æè_%èx¼…ëãñÃè|Ç/é”h]ía½nb‹(¬Ø+'tÙ4ÊõÕÍae‚lý´;4ø¾ñ*b-‚/ïf«mU¡4ÑÌ“|Š ˜ no¼ƒ|ŸoÁ+?!çG_c6;d·IþQm¶ÚyÞÒ“ô«R“”¬©¼k Y}w9Ý  ¼6ŽÉóüÏy5ô.;ëp…©àëñýYãdAP÷‘f¹Z,šSYJè@W̱Q’è¤ìŒê|ÿ~„ÓnÈC¯á£­å9Eè{¦H%eÓý5]J7zÊa ‘—ÎïD J|ŠòÈ»ƒ/ªìwÂoÄœHÌ õzL)sÀðÄØ¯^ñ0 ©7îÎÀX|Û¡x_ajÓl†É£zF»”“`jÒ`ó—1ÌÄf”¿»³ÊΕM®0'—+ô.Íæ…‹©œ½avk(ËÀ Æog8Ãbýo%Õ‹°Ôòì'ÕO?X!°<0 «vo¯u¿³µ!º;û¦Gaý¿@95l>÷ÅRØªŠ¤¤HDBÊÅõ&ï$R6ñYìDâ+j">Ǥ3â¨A_5’9´ìý¡¤Ž´·ZóÂRrehŸÎšîúÃs ä·ý#š¿‡8éD–.œ¦Á=æíà _õ>ØK¾Èi>&ötŒF&bAŸµqd>¹ùÓï²;à\Õd+Ê Ç%†×±V¸…!—‰§úÇH)Ü+ÄU÷9ôøÿðÃk'Ü B¾±ž6¹«°í‡›¹¿Kõ¤ßDaþ‹e.(r2»$Â2E3gbß £8ÕOœ™ƒ÷ââ…7LPêPÉ ¶{%(cõ’à¹ʦ*&“ûY¢œÂ­î´žG(_½xA†‚ùÔD󪡒¨Å|JO9*ó†<—‰E'G© õiTmûå\½øÕKªóè¹,Ù-ÖdkPsÜÕͪ=µÜNàËåNÔîÚ%ÉlŽºl%Æ·Pò‘Û©ÌÛ¨O¶Þ#!jƒû)_hå aèå oÆr4ºn;x`M,~‘1¢>š6‹z‡æÇBÃhfÔÐâ¾ÚˆË8Zž¾V¡5f†VÇì³ßîe@k_¼5’6ww)ý WAÛÞ—29này¡˜ÞhŒ¨ºD°/Å)±R¸†ÊÄ'[Ï?]@¦PÆ A…xX+ŸyØ 51GÌó/mûƒ2-ýK(hˆl.òb„÷œen~†õÊr\N8Ò­œýC %tôq‹ÃËÿœã4,>¸3_Øm0¼sç=¿uˆÐs ^ÛY)YlßT}¶˜+u±ûxQ¹ÌüvîVûIB.U'µð]ÈÒf»Dµ H,RÜác…‘ÁëÓ;ãQßíEm(; üÑRý°‚XSós«T¾P95´Ý[«ÕT‰V×&þõCþ©óqëC;ç}—Ó•ÈC[ÔC¯"¡9ü¢¸ˆ{´¬}ð1_§†Vï†s Gšvö-½1Á ÖìâjŸ¦ süíŒñ)ø‘rßáóó艻¿|cpî_?­1§Ûz<Žø0äÎÎ{Éžª%0´°[Ñ`O$Œ»ÇµÚ܃1u…Óéta0!jåÔnFSÎA½!¶é0ÓÚQW³åLŸÉ¼LbÊúL,äV<Ü}Æ–*öz–¦Â +S± +¬N]öd»¿'rYmŽß‚è`†N*ØRt2÷ C‚AãZÖçHäªbý-¾Iˆê%1iIsïI ’ŸÞÒãäÐBJͲŠ2Ü%ªÈh<4œŽ‰½ô‘HË“‹G²"ÝsñÓ¹m¸'+ÍSÄ~ é‡Çà ¦Nè{d#SÙkξ(d9%ïZ`lèQ—›ÔŒ,_Å—3sвcÁ¸¹?9›ì.¡@{ú_ËåÈëŸ+±2`…|u›ÆÓ ( Eì7¥‚BWÙuex HU$ö[*£¸ä‰]/.¢¤{‹÷ï¤(msÔ*šÖe^uÜüŠWQ¶•›þPðo”“<Ñ@'‡òŸ=ùô=TQ1šóð×˨\JS0“1…j|ª)­Yܨñðá—ïòˆ!ý™ý‹¨9e®.;ÆZŒR3F?QûT•»uQ$êô岜ãšB½#§×‹6.¡þ÷ƒS%/ÄÑÐþáÝÑl4ZÞ»Ž&™²{µ“ï¡Ù޲w.3h¡Ý’×S‡–v£ä^{ UÐçe“´þr—fá¡)Ú ÷œF»”le[^C<6¬9ONFˆÒËg®ª½ú‰”‹ ³3?ÂÐ/Iª´†AøtóÜ×j£6È7¶9)ò˜²>üàÚ¢þéÇJµzõ µ¸˜·ûn+¼ÛµÅRH^U×£ü坿ó¿_톄‰Óé—Èš vá’N’V¼ˆþ9võÀÔÎþQ¡¾f­aÓôYð!e׵޿-([|¢gä •õµ{üZa»;Ùïýµ“þ2Wr¡ú‰Ì±Ì9hdˆ¶â›¾ ~XQMJ{5õÔ )ä¡ÕhÊøZ% MQ&3$Ðôcêk®4sfK³j}‚按ëST Ðâué"3%|W ›N=R­©ÅßèïAëÏ´ÐÏåžÐVüä°­Õ"´÷dåTBGA¢ô>袡ÜUãüP¿Õ0mÝ~ \7¿BOªGÀPü¤¦fMçR‚>ãÛƒžtЯaâG·¯ŽÆüᤠƒÌ·÷Ãp¸—òÕ‰ý£žý&®…ñZÃ:™l%˜â¼F»B·f‚㈈)S`6©N8mæM¯‘©Ú À¢Hðñ~õC°¬"vúêî7°¢n·¦s~Ïï;åbF›±5nù?à€äÂb75q{ò\¼Š$,Ns7.¯ Wa“±s>RHúíwˆTn­þ Þ’H]^pc?å"ÒZÍú-x³â‚üÂêt¤?b~ñ÷idX 9îûÁÞém‹òÁî§'²bQQ´·úÊvž)S_wjÔ7FU£ãþ ¨^i¨¨h®G öð î'³»Tý͵nåWü|ˆ:*r׋™6PÊÀk¹„õ‡O‰ØV@Ã÷!–ÖîhìÿñtûC 4u%ÿVúÃÍ•Ï ÿ¸ÁŠw&(n-FËT#yy´Š(|=Þx­ß+JZY÷£ÍÈòGáSgÑNÖÞ.6%DŠ{p<@¦'–±šx0/îpÕGt L°m‰¯D—“jçŸÝ @WýÂo޳€žö¤âcìQ„4‘.éz ’0”s_æ…Þü3‰ðáWrÚ´ß³íܳ>,0™oÝŽH4Š‚¤ oZõðò×â—© ÉÈvupÿ½"ˆ¿u„IÿëÖÖ=7‚.jÃMºhˆÖ‹»‘nx¢Y"lÉ[ÿòø2Å]Æ…| RtlÞ‹‘ûî¼W#ë¤îÙÙýo¶~æBÎS²±v8äNúÌßby®¤Z:1 ¯õ-Ýí‘[ÛュøGäçÊÝK䛇üÒ•Ç—“ý¯×‚øk ü,ÿsŽÔ±(¥}½Ûíƒâxž¬ln (9Aúy‘,Jy:çœß‚ÏôL.̲PF©xj²s ÊÉÔĴ×€åb½–6ø:µµWž*¿Ë}8s7ªçrXH¤ Ö‚èžÀ>¨/q|5ܽG^1C³?ë8½-|·bÓ'U†¶ü²Û=¡SL¢Ÿ‘ê7tÓ”ÇZ®§Cï¦mÔÔÍ>è§ÐH ì‚JBWV]ª.+’˜î‚‘Õ•Eá}B0nÐû-ã` LÓÅ ]…ëJØ=Ô³Ñoب.Eü†¨ª†/,rp²’ž€ea÷Cw aU¿+<œSƒ¶v~2‰wÐY«E¿a0ݺU?rûB w¸Üûx`ôõªÔØ“ÃÑÏŸ­÷îÃ@½³ã¾F*4K}ÐS» ƒ9>G&(`HdaÉU„†~ö>¶èr…áùíæÏè £6§c²ÓaìëyÂÆ²Û=Ñ@ŸP*&SpÞæn>L«<*`;Ò3+ †YDŸ`öx¬×“;0wì-ÿ»oÁ0ïœ÷Z–Ì4›bÂÖBaqÏå+&“¶°8Ùa}ìü#XŠþÎE +$¤Ç§ŸºÀªQƒOÒ'X«ä|å ëçz乜)`Ó•8³w÷mغ‚~ô©'àÇYú[y~HD<Ù+™j„Äîô)QHžõ©ëê $£Ñ¾¹{ÉKÊU>[EÊW$…«gÝp×ëë¨yi¾°7úª"íÆ]¡ÌazÜ#,*Å>ø é ·ØÓ#Óa»k³‘ÈØ¥yí,—2뜹6Ÿ$„,Ýóa!¹‘íygâ¦ã$rœ\û‘_š\û{ ÿt)à^þ]õëVEÿÕ̞0C¾Ñá©D+õm?¤fO‹FÁ2>6—ˆ£(,òÖýOk?Šœ{µ²åP‰¢$Ÿ~4'@q–öV¶(}@¯÷¦JùØ–èŸ\AëÍ·6$ïQ6‹æÑÙÝÝ(§RP¶Y-³í‡¼òýâN¨È!*;w•8ËÍØ®ô¢2Šò`¿ªhé[š×£jñŠ@´ª§å„îyé‰Èæ'_D¦‹š+ýFåCD¨uMóQ!êP‹è(@]Mç?‡·Qët¿²ÕAÔ§b»cH4Ï‚¨êuÐ0•æè£8~4fé>];tMžotŠí@3¶fL®h>ûÒ«ô=Z J›w(Ç£e¯þJþÁSh5yç~¿O:ÚP×¼§‰B[hÇûFhyTÓgLÝ+ˆÙÖúÚךߟ›¯C͹ͷŒ_xQ‹AN"¢Xµ/V´Z®£Îº§…ù"ê=!nXQC&Þ“Y­h˜I—oÀ‹ÆÇØäGU¯£©öXD¡¹öײ3Æh‘7rph*-íÏO³¡µçÈ7—Z´yðÞï³òQ´c²ŽÚKy”9Eñ$ãÑŠu[+ñ«(iÔ©$F;ˆ¿ƒ“·¼ª`è|ôÖŽu·{oºCÞ óÑ{õ1¹ÿš½ì6¼óZ¸sääàvÞ-8'yŸ„׃ç†i«ô ™z£Õì¾¼R¼)åx–δ)á„X/'é"=x¡^-•©=ø÷¾¬‰á#Ö8:Nî½Z½³?!ÅÖ–/Túò²ÌEýËtöã}žC½M"}¬ÐÎ{ÚžnÐ(·Â}“*S •t¦Ž@ѮǹÅФ>}{!û44t‰ß½ÒM‘û) Fš¦#ÎYI@sFåý ƒïC"ßdÂ÷KåO9 U;,Ž—á>´IËD1ûdB;{ë±¶Óü‹1ÖÒ S’mF5iºüäŽuQÁ§½Âá.•Ð]vVtóô Y“”ÃOÑ›&Áç O7±Çá¨?ôó•órT‡ÕáÛÂjv0óýiéÓB6ˆÕø‘3£`¾^Plã®ÈKãaòÑsâ[†U0ÝTdÆ©ôfÕ+MœÃ\Wõxë°4,¤µt\¢‚¥÷DŸ¼ÆwÁÊb›«øÛ½ðÛ¦Úö°H%l¬ÒÙ„¯'×™‘°2zÓøçY$.Qüh£ûI‹W;ønñ"y©¤CžP1RWj´u³ µŠøB8lâî¢Çòdߎ"Ý»gËwÞCú]¹3Lþ²È¦)ü[™,ôÔfç!Ë8™´öjd{IûÇæ½ rŠæ:[k#÷˜Ý–òò iHø|nås³'‘ÍÝiJ/ ‹T9[¿+£pü“Ù^6¥íœì®–ðG ö>ʆI(Õÿçâ‰Æ'(KRÖô\Œ÷e¬ÛÍ¢¼íÂWvÛTX:}CZØ•ZÖɶspê¾[o? zp¨¹ë!eDžrX𻈚‹t¹¾?rPë’Æ^ûõ ¨³/¾†”õÈchÚeP‡¬lº ÓÝò™æÊ¶¹*¹Ò–a‹¦Îû+ê%¬Ð\Íê.Ý­ýhñÐ ªél&Z†§º¶é°¢UŠvÒU¥«hÝ곞já¶lº%¶Q‡Ðî¼Ó#¹ E<0ÿƒ»ày):<”ã鸎mÂ÷ÑYý15Õ þá/5ißµý3_!‰·4çCŒ¼46Ò’ÔRûËÛ?œýÃÝó‡%”wNCTáeu#†¿×îûçëßÔç|ÇäBÈÐ8¯Xr»ßÞö\?•9ilÍTmwÖç“7ºAMª¹AvÎßq!ou©vñÁOÈ÷y¾ä²ÿÿ{-p7!}…ºJu{ÛV¡(YºÉSºeçDZèÇR&¦kvÏÇá3O·Zà|ž¬‰¿Þeí"õv× ÿ}'ʲ‰|mýï¸Q}íÇgþ•løVùеÈQ¨ç²¢î‡FŸͰhÞEVz“™ZDó¸¦÷ýå½#ÿE©Åï¹î He½œ}® ÊèÁÀÙ~ÙÒºÈÄlÈ×#©Å×ÅTº`lW|FóLzš¨-„é?4}ÆóË0ë{CõÉp%Ì3-L²—’ÀÂrÛÉæ%kX& (z·qVyM9äeÂoS—púÌ3°‘ìqÙ€r¶äÆ;Y»åàÏgº z$:wýœš²oÞÿN>É‹¤Y•o…Zd‘<™,ûhúRÎ4•n‡*õÝúKÓ¥H{ÎNR Îâ[ªDª…È@öiµ‚ÈØâÕI^Ý…Ìé_TÞi#kÑÈxº¼rp^`"aA®Õ‚Ó¬‘'Ý&¿ÛÈùžŸlx¨ôrhûzÓP˜ìíY ІŸ0Þ¯÷%<¼_¿·"Bé”3ñ×Û¡l¾{V^@Ê}aè½Ñùæš´>éL¢²Å®µ?†¨ÆÒæ/÷­Á8ÑD6ñ:j¶8|¥¯E­Ë•NVåQGú˜Ó€öênRÍft£þb°ÐÙm?5œ$œ\-O@ãOÞo9hÚk®L£‡D/±EK)ât¹n´:Ãоï+Zwp©ÜZ8ж.“÷eð•ÇÆÉ£ah?pSÔÙ¢NË™dóGç}þê4âñP±Ë×ÜôÎ<íWóvŸ.o;ö!Ù¸>支uÀ²Uhõ%Oo¨ÿV›Góá.”ßÊ"ês‚O9W™EŸŸ…‚†0Ï­é¡¿\¼ÿìßµ16ø7ßþáé¾þÉÇøÍdtúSS{ïïç¿Âðç9Mþã;ó gkÞ.þýœ¶‡¥$úÛ.B׃Ô{ܽJнv­åDWü¼¶ØÑKáñ_93ð;*iQ^}#ƒí}`°š‘èÌ×0$c½öô½ õÛ«L‚áµTAŽM=þ$ìS§ŒM8OìWÜ Žó‹+}`rB%:¶·¦ouùŸÏö_–¿Œâì `6¦‡”Wù ̽Hày7 óYª2,·ÇÞ—6¾©•°¨âoÜ×fKy§žI†¥¥7S±£`…¼0Ó#·VaÁÑçêaXKULdúeë†OµO¼'Mgâþå mIô‘ð›õ©©CÅH´q{cÈ2 IŒY¼¹Ï"iéû)Vâ·H~òOìÑ}¤´ÜÿvÎwY 1išlû¡÷óËdVöµøAÿ'”H÷çây>¢§H¯D½8e­€ ¾Â b«•Șå8yI3™ÙÇ}H;ú%ïn?‡8 [àz~Ntr8„Æ¥ìm@.%»Jͽê¸wïZ£å¿~øzZ6N ùFfˆ÷ª¡€É‚ÒKc:,ßhµ 0FaÉî„cãž(rõ©ì w ­(Ðìžå@qÆãšÍ[(ñ†v:r77JNõœ5—EG·÷¯}ø·¹ H;Šrã)f9 (_e:n–ŒŠŒ,¼7Y‡Q‰N!åÎÞ5TÚô”xô UÄT\ùÜ3P5ñÏäñP=ú…ËÑ ˆ¬|µ´lq?™»¼ü¬?j%0 Ö‹Š¡Žœ¯®c/êzmf™¾C=Y;7‚E Ô§QRðìBWa׿±ÛhX3’/|åù¢É(U]F³xrÕèoÛ^¨ÖË]øx -Uø -ï3¢•ánQñÚ0´>YtŒ¥ÄmR„…"š£íѰnŠƒxàÔÑô"ft !0ˆŠÇƒ9ßwwj]ÃK T>¥ B/¡­52q©WòÁüÖí“‘a P½òøJ#áû¿üý“+ÿäÕÿÛSÿáîŸÞùwÿ/‡É«*aŒëoÿöÚt+¦‘ç9þ'ÿùyÿäÖǣ›œ“­Pú¹?ÆŠqgüñý¯zþõCZÒ“£ÙjP™ú§…E‘ªõ¶bÔD²þ«ß6䛼K’±ûÛ[›—#¿RÞ„ïgF•å˜Aë¢Ã+h·ýtè Ã~èøžQüy7tYœe‡³z4bƬÐSÙ¸e8¢ô¯vR\zSý[Ú]ó`PÇ®JÑ܆’ínöªÀ‹DK¨ÀhÒ‹Ÿ'‚â`ü”úSñ˜¼ÞF}8̦û‚j|õwýÛCÿ/ï <šV¼`±‚©Ì¦ä,_wÙ'嫦RÝS"ðî°‹JÕÀ\dÐ>|þœ°›ö#?‚ ÅwYôˆ˜ß<ʬ‰ßSÞÝËS…¤>‰þ67R\*(¡ý7)RÆü°*@ª_!u:ÎH½ÍHWÛƒ´Œ™õ’ïž ]¢Å=%ޝ¸§ä¢u›J6ÒOˆo¾EF™ï£3#ÈÍɲÝÏXÄWâ?³¨ ëØ½2½ýƒô@Jéä|ü)råR=rŸºýÐâ­/òÈšùsh\F^c[3!ä{)~6Aht%œþDÔ£:O«tµŠœ4ÛýÅ>´/¥J|G‰©’¾&O”Z$y~š‘en¾%  6FÙ/¾xð åDÏUÑ:}Gù ïë1¨˜ÜtäÖÃ0Tî•f‹Ì A5‡§W¯Ü—DVµÌêìÄ-î›įQs%m%^µè•Žw¯G¢öͯJЍËÌtG´Fõ¾D¤(ÎýFÛ±Î/WшüN¿ß›04îæùÖ¸ê¦íÏoª£ùfv™šZŠ®1ˆ­' Õ©êÁKûêѺå|»å)´µÿÑú gÿ¥MÕ¹A´ï¶ êá!Á£t5ª¶óE(nïû>Ÿª)’grÍÊìa`­½¹Ðv>j.TÝ7 ¹Í–‹¹ 36Z‚2ò¤Ù\U& †7KS¢{wÃk½—sú´ ÉW,ƒ»¹^–Ž‘:ÕÉBü–Êr@Lu(é-ሳ|ô3æ×“¿÷eÕÎøØì EWÖHƒïƒ2¿ÊB¶`¨˜MH:F5¡%ñ÷l  6Ù^ͤö ÔÕf…Ü~»ÓõÊîסaý‡iCèNïûz†D?Ì-7Š€Æ¡˜ã& õи`{ú”äChbšhŸÔ M>D¿~­þ€fî·Î 6¶Ûü=kýyÑ ZFBd¾w_9ûÂ! ZëHªž¦ûBÛÇ0j9€öŠwrþ¶y,”¼j ] £JW¥¯ïìõ2ÿ®:tŸZçºØ_ =©öAá ð“H¹`ý+ôW?µã¿¹­q0­z&£™ÎkÄðÒúY;Ue^*‹Ô;ã|Ë“’4å0©ÿe¶‘T¦ozEí.Y–æ!ªÒ}0—ó„òù%~Xò»õ$Z–ÎWL\ã„•DN¶©Ýd°6qÓ³ùS>l\뺾‘[¾©.A,HèŸs2ä&-±¿ƒ¤ë‹£ ’úx6„¨)oÑE+H‡à®ÎÝ…voýp·µ÷äïz¤óÿ@ÍØþ÷T g¦.,!ƒþ«µUlD&“'§]‘…"‘³Iô!²™t¿—º$‹_&riŸ!·¯‹¥ÞA'ä‰W˜¼’„|{®åño"ÿ «ÌÃ/iQÐÖͨéJ> ËEfì@Q£3´T(þ†V®FÙ¥|â†fNò¡LÍ]ÁÃ5ÜwþùѺV ”çë`'¬E…o϶¥¬˜çЍ2¹ôuÇ÷Ô_R j/ ¬Üò.ÚîŸýrÜÜGMQ똎•dêp=á¢)yºÓÝšUÕ!¨_ñùJÈž}h©þ®æz;;ÄÝÜn”hjî­:£ÿÍ̹î(E‹°W„^“hùDÁõ‹5Zåª4óÖ¡õˆÇz“²<ÚJ“~u&F»GaRVqehO×òrå01:dÒ[ŠªDÇÓÀùQÑ]rŠVŸ2à¡RŠä 5è:°z- =ÒÕù¢jŸ  ¢±fQ@’pªyÚæ‰C÷£…Æòz(àüþÅë%dšK_.ŒÒÚ¨[ûXÚ¡#ÐîlÑ…‘¿\÷2{óßÒ ¾{ƒ,-:ª0`ÂnÂÊð†¤šÄÝo‡‘ÛfN»-`¬ÁŒÿ•LŠ©†^©ƒé¬Â'¸Ù³ºGXc.3ÁÜð`®M·,”ïÒâë…¥ÆWL¿ËÀÊ2£<çýð[:çânÖT؈hc¢mr„-ÁÝL·B\‘`Ql,C‘èðHŽ`¯Û%¬—ú#éÓ·Õd!€äaÝöŽÝH9@Y“¿ƒÔWù2¹JiÝgúnZGâÝ¢¹ƒ‘~©ánÕû;뇯߼.ÌÏV¼Œ>Ç ëëýŸo8¾GŽ]³]Ÿ–Î"ׄãSÿ²mn“/(1Ó-"_X|ç K $ONŒz°¡ÐïgN¥WPô¡q¢a[Jø§×^f”.”&.´€²%tnûç$P®f’}JË 6a­çk>*öɨ<À†jâùM­§?!8IBU„šmäѧPëºÇ¾YojÔQ5Û\èD=Z>*¾…j4Øåöbsÿ,QÖ8oè–¢É.Çê/øÐŒ^TÔ×6 -Ä >]´Y@K×¥Ñ÷$´Ê¾¦óØ´á»?|ˆxms:$Çx)ñÀÉ̹\3 t°Ü’NüD€ŽG™lÐŽ¡s¾..º Xj¥bÖ”¤îŒ(ª6Q?a}«lòG‰ÍLÊ`±[{·µ°4ÔõÙÊ~Z=åôDê ARP2{ØÜêIœ¼™¬Å`¿ÃÇ—Í—ðþ4é Ã×"è[¾CÿäÚv~S+“3mÀ@QÑ1Ú!ä"ñ\Q€A§ÙsþŽ0˜¾‡ZU †,¬Z÷ð>€a'ë.Í0fùÚfÀi˜c°ÏTg*Æ“N"I᣶ô0•#©Îº®3GwIÐŒ¸Ã,ûÅ /S˜ÛmXIüæ6]cs3èa¾ÿTáåø0Xøi¨áh‹/úÔiRXa)ástîƒ X.oe¥\ö‡UrWÞw+oa- úå¯8qXçbU˜·€ÕÝ¥L×áÏŸwßÕŒm t0¼LãÆ?Ø,·Ý+ks§ÿÐ" «>ÝKM$½sx÷Ci$?‘â4&ƒkGHC/"ÕÏÃﺿ^Dê~§ˆ©×v×ùúµ&¤Ëò½Â´€{显̰Cõ¨@ÕËAÈxÿ4}+Ì#Ó²ÉF;Écd9÷."1ÞÙh¤š¥®ö"û[†ào-ÈiiX6ŽÜ¬/âuæ‘g÷íºN±DäUš² §F¾pŸä'è™uÏø†¡àó¬æïl2(LvoÀ¿ËE O’Ý<͉¢—É뜮ÝE±ÚûWÂÏø¡„î®ë?9QŠåq»Íño(#þû ¼ÎEÙûi7&Ö÷¡[¥Sgvʧ+5æ^FE ¡F-yTbܪ‹ø¾‚Êœ“%¿XœPÅ©ûkzŸªþ\eþ:‡êMªz'›</S˾cÃý­Ï3еn:îêRFí-¹ÈFÕ“¨+Ñÿ3*©õˆe_Í¿ÐE½±C¡ hÐ@VÕ£{ùÞwIúP™ŽFDÙùIî¡ÉÍ#¾¬¡“h&ÞÌz<ï8šO&þe;‹ÃçŽt«¢åðFÿTÅ/´Zoeü.Ž6<‰ãNò£í!ªk6åh—÷Š2p£íÅvû}¨;…Ù•^”Øà¥^™¾_å(_>Möõ`2ïærãÔŠ‚ù½'H%Å ºâVîa Wøߕйë!ŸG2HoNïo^efɯ”t~…´n‰JVsxãÀ,ZÆw’J"4¾o‚W-^„Â+Z‹¬Exõû·¦$7‘T“í|y€×… =>þQ'‡?dÓsts¹YAÎèýï¦fž;ó‡žËûþæÖ_?<|ûæ8ÝùuV‘ ÷¡|!âmÄ€¯¥ŒŽMC¥—­–ÊÒ¨&²i¤ Ò„šA 1_ÞP»øÝsàÆÎó 6w–zh’ã aQÛ„æ¯×ëžå ÁwÕ> ÐZbh¦|´Ú¥.P2®–CÇg†û·xL¡Ë>é6 ½tS|Ÿ#oR…ž yÏ .ð³ñ‘oPåÎþ4Qâ0°·'ˆ«v›¯Òÿ`_ª* ÚÅl#J‚éÚÂÍ0Z.È'æqÆoº+ɪrÀd¨ò½f‡˜þ.Su˜»fÅLXœ®ÂÜušE^˜ß,º¯œm ‹ ™ên-°|pà‚è"-¬r·7Ýö‚µMÊhÏë°>’î@¾@›[Ï%Æ-‘@]]‹óÖo$,"`¢6/Db/–²k«õHʶh÷›‡H6²bÿ •‘"ÿJÿÌ B¤Š,*· –Dêkl§k<Ýqw „¼u‚ ÒéúóöA=îq¶~ÝŠôOV6ïB†ßvI}Géî³ÉíiȢť•wtÙmw-²j-oÙ"g’Uöó,äv«=AªíŽ<{w‡§ÐÞ@^9óäÆä»ìÛ|41 (÷DŒI Šã²†ñ2Šø]¹žëb•t!í»{Q’x뮳8?Js $. LÊ3yÚÜ(»¨ì;0ðå<Œ›I쯡E¨”ºÚ:*Nš¡#IEN†¢¡bNT í«)"NF-Òùûa7P“mqñ'›îçþQ÷âµji}ÒÑžÚæôý>–ߪ ¨«1<øi0õ†˜.lµÐ£ÁÓït”šh$úpo¶ovqÌm{®éJ¿BWw ZÐ\£ýFUŒ–Š2iF§üÐ*Àê'µóy´n«È ÍB[[â½ …Ch·Â+rÖ¥íËhB¯ØÑ㑃Q{ —PÔ‹BF‡ô3RXÛ—ÇqB “„CuK{ÜÔwrùÜ‹â S-"sõ˜¼Õ¢ê[Ž‚7²Óš2 9¥qóÊémîìoÿb5Ü™èvtÝÙ??gqõ ÄH¥÷>îƒX¯ƒawžü½/s+áG¤eàvÿe‹8Ÿ¿„‡øË=ƒ ?n¯}}w¡zå™ÎþœD¨åÜ8BeuÚ¶Çÿ1@½õ>µ—жûî£P}Ç­_rôCc@J[cýqh¼glœôHdûj¢×pØ _̨ûg\„ÆßÔc/B¡é ïhÝshj¾¾ðÅZÎÉûT»Á÷³Þ¦\/CëñŽUÖYhs5Ù¿­íÇ«Ø8qIV_Å¿\ÑD²-ÓG¶½óH)â°q}wÌ]å`Fš)Ý—3¿å‘N©‡7noî a:­y¶é·(kÝÜ"cÁ YWyJd.L¦Xºd‹¬Ë)G)´™‘Õ‡zõÛr­z¾´nôBžÉ WO!ïç¸7æw ¿®6q×V ÌÞ¨UD¡Š‡­f}P¤s—â‚Ð8Š+€s»°*J®£¹å;”y¤`ÞºÔŒûPíbêtÊ͸M²½«C…ø{ímQé!|¦‰E•Š,áÅ$Tß÷áË„HB½éçâPÔ¬· ù¶8ŠZŒƒ?ú:Ïú-¤P÷;çé²nÔOu_6”DÃ@–WW¦ÑØðÊ]5¡khºÿIÆn÷4—6Î$}ƒ¶Žn¸3¢eT©k6+Z}–þSÆÖ+×WyˆÑVçk­Ç¹h÷¶ü›ß~3´—ªUÓC‡ÆéoeŽdè™tÚªOƒZdüÆCOO0S{ð ë¯È]"®$è1KÝ*_…G.\3ßB’MMï0bøAG»ú~¡à›ú1ÓÀíø&S8R¸xbõø#àå—ÂÂã6HÜóÆ%l€ï/oqÄ¢ïëIKþrmgÒkpÊ ¢8ÓŸ~Š>ú÷ºí‘]¥¸Ñ(¥öú¿ýðVÉ7‰ Èi™W*ýö¯>#µ4 }¼}%SΖÈÜO "(ä¹#çB>ðzm{ã;ÒëŽúwÃ’“ÝÊ`ÇÙY/|}(Ëòoÿú¡÷©÷ýõ¡¨î(©é’àÿ»Ê—ßßÝÉ]Ö÷Ô‡ \0ÎŽãõnøm.PvOo»Oé£Ð€Ê’X_ 5C¨..3Üw¾-3:­2B½T=‹›ÒËíñá¼\ø¯ý°þÖ(‘`&´Nœ—J)Ý‚Žý'ütÿ¬þåº'GJj˜Øú\2¯ÙKßµW, )àõ…Õ:ñìS˜§û cI—Ø¿‹ÂÄ/‘ Ñíÿ÷´§§‰´õìæK Œ8áÀJ©GOìÖ쌜2 ÕœÀQ=qâèîp*ÎËíä©W§¬±z8p»;7}¡žAE®ÛõûþZ¥÷Í}à7?ÓGµ :ÏGiÁ‰Žef8ùI¡E<È „7xž}(QGKÚÛ ¶’Dá˜e/eµo›€ä¶Ô=ýË ÍX»¼2 ÎϳcÖA6Ìÿ„¼Í"ȳ¬óUH"A»P·>€‰Kê1ú~PÚÒÆåUÇ™VíÇš Z·:®Ö[j‹U¤c)ï@½drTç,hT< ɉ-œÓ.G¤@Ûñô|7èd3'„EÒ.W¥ø‚è8³*÷¶ôÏ=º°²Ï—utvD¶`x÷wtzÙi0º«L9M Æ‘FCïýÖÀ¤”‹ç­ ˜î :Š˜/ÿÆF`ñÝÖÖYŽnïúy'*€øû®Ó$& ÛÝ-¹îO”t]žÔjæ)/>è[á§'y°/¡eÔsã .Â7´?‰Õ‹~ñÁŸQ¯}Û´í€ÒmJº#†PÆ< i}ôú$k$arÅq߶ƺö7Pþ¨Jƒi(ç!oúC}ÉÆTö»^nZ¡ŠŒ™Zôý:=ϸÕT}ìC¶un%ªùš†ÛÝÙˆêRO˜_ŸžB …«ÒÏËÎÅí±íi â¾Ëìþü¨õ啺˶*¨m/öcþMÔþPç&GÉÁ{@ìSÞ€:êJ¿?’X‚zlBó¥Î¾uø9fnH"td­ƒÐVwÍ<¥.¦92í“ù|_69ÊÈ¡_Á'Ø2 ï£ÑßBuOõи9ÛÃm.g41”æ8¥u M½~Ð)U´}pŽT™ËÍVU™®Þ=ا8y9Å -8Þª~ÙÒ…{%)G\UѲeÜq<´*n_w¬°­ÍO&}5\FE˜ÂåÐVèH\ÃÝ$´ó’Pk²^í­¥‡Y-XöÕ¾?ƒ$ÀeäÞmnƒ#]ÚØ5?nþ7r‹û—€ÐÕv2pAŽÊÏT•í idÛ½I ý•íý²È›F”p3b[ð7ÓÖ:P©V츜ZêÊ0¥‘–  õ¥µ¨>»ô*£¾¤£«Õ”Á¾N%6#6Y–¼±Ý°¢—ÀfB¡QØk$9RÃÁpŒ©Û7ô pz¦rª‘RÀqÁк–Jyà‘Š ×i| |7?lYuƒ@«6îŸI8ɬ$9e: ÂZ3fF¢Ý ò#Ð…r¼ĸé,¥žøŽÞ !ÔéxàÓ;ÒÂõΑޕ {fÝZ¼âÈ ЮÜâÄ×,aè&J–)ìYá ¬A ™a *i}½ :7‘¨(êNŸ‘éÎ ÐÄЩÉ=ZÏ?5ãêá€Î©—6‹e™ Ëþâ2åÝ ÇÀKf!¡,|%^Þs`¨â¢~á—ù÷öÄ>dãÚÇõΖ`j(àù\ ̦â¶95À"¥æ%{;8'º± ^÷7[°²€`¯¼eÖÛ n'/ß&AẼMB‰QJvö¿ýd„ÃCJ$ªQŠk‘È}”L²K”—ˆ¥‚ì¿ÙìëÕöôœs¾a‡}1ýÖ0ý×0ó²a\nŽF…êrRúóx¨Ì"GýÄéèûÂë”R‹#¨ÖñÚÄÍöa¿›†Â }šGèÇávr%j:Îö©]­5- W$žW<ìOÕ,¾)¸ë5s«2 G‘ìj!…ì¨ÙmÀñ—†,ja¿;Ž¥1´¯IDÉlQžñ±ý}ô“Ìòøë±HÔN˜Î}ruàâ%zæìûC6;'üTÔŵö‘0$uŸÑk­½A…zBÓc°GyPok³KÚ3Ô/áÑ›ÙgrÐßÂìL‚Šûc"ø×ª¨­’›ê¡_¯<¾˜_Fc4›_#Ðx²že&šLYü„7Ŧsêí|ÑÌÌ7 ÷âR4ÿºÛ˜œ¬-¸ió¤EK¼ï¹©ºÒÐ ¶É‡…l„V·Sž\nû‰6hw‚²gÃÑ–oüÝHÞx´«Cr»Ù*°jf{²~&Î;íúæ;§áHø­û£™€ןÎ]D„µïŒ:€˜á§Úà} d³úE ä»nzM…@©ª.Ù¾v¨>»ÅÕݚ˾tüË9@wñý ãf`ˆ¿É2rt{_gš:é?TVÿË+vWƒÒ*›ZFŽ]+оcÆœ¿ÅͬFàøãÌäš¶<à9¢ù@ øÎ'¸v<¨r×£txpòÁ`áR² ˆ˜»….Y¥~oÈC¼iÿ¹Ëx¦$o2R_¼ÚÒ—Yçñ@öíå/F&P úé—IzP5î9Ýç Tä9ïà*êFù—ŸËêÒë¤W¢¹ þÅ:9­X 4C¸‘˜¯ƒ¶ýâkñ$8%{ö™«=œ/㺃úüZ*·"úÀà…eó³. 0Œ=e½ F?xN¸*\Bƒmv‚60µ l»4ºfß×\Â~õ€…ne†ÇB0œY¨çcÑ«’ÚÉGÔ`“°Êð³fÎ}nẮwl·E_viÞ'Ñ~jŠ˜Màò¹÷I‹×ðÔÝëñúPw¨àYc²,”ï–”H&a†Òû¢Mg½JPêúZŠ›û:óÔͰg|è­GçåòOt‡x‹u>³j=þáwQ…¾sçGTPd£¦l€ÊóÃ1Îð™ú•=š¼2 Øü×¾>³NœzP7M/>‚ÿ×Úˆµžºyÿ@Û©õ–Ô®(^gØê,î®ï?—|xïLÝÁþbÚyÝSès ¯ãÃ[(_¼f\îæÓ¿þП8E¥Îþàþ!»Šyãÿö‡¬=Ò?õšQ¹éÏ*m.T¡Û͸+{UÖŸ¹˜½%…ª¯…EÕc=R3¶ÇPmÏe«¦ Ô@duÎnp5ÊÄn’>;‚š}Ÿêž¤Þúë+?ïy(Ï£ö&o÷‹<¨ ßÿrËý£‡¸î7ü#§R%·ã‡Ú—¹ÐðzH§>¥úY¨íP‰ÆU¯“·oË¡ÉÛ©Tgc7öqK¯•¶Œf)qø¨?¡ù›÷úæöh!°B»–Æ-Yy,5ø.£—͆—zQhíMü×y~´±l«sdŠm»ª½±eA{f¡Ž=_kçVï²7à Lá-ÀÍTüu‰ð<Ž|Ÿ±çïÆóꀨ’¹8¥]Hœ¹Û9)¹‡¦¼o;PòÍ|#Ôª†…E¬»J@fÚî¿Øt®6ßæ¦õ€ÁoíÚÉ´V`š²÷ìÖoÉúæÜ€#¼\28N 8oãÏUuÀñ”Že·E๦!ÁüÁ$–« &NGC§Â]ïÚë«A´þR©Åˆ#ˆwE½øJ]R4L.LÕz |¦Íž°äo=c¨ÈtçŽP/?(õ:¨KàƒÊãØ¡µ% fÀ¤Ä4 âŠDœ& %Ëò‚øû<èhZ©àjï‚®õ;o67Ðón6wƒ[¿ý•>êË¡R)0¾ÄÔßY&ùWI¹ ÁÌþfΰV1XHhõ©3F‚¥È’ÂÉ××ÀÚ¦ÛEÌú'œ-žQÈ} ¶Y/t¯oÃ>«QÝÓ õû×£`´+Ž·_|…æs"qÒr÷ד3ÿñël¨œ'1µê[ë¾.îa¿´l¹ë Çç\þïÔ°šâa5 ?bð›î¼ —*æÐ—d1¹¸J ôý¬÷5“«³ûë6^ñI†j½ý‰2ð"ê(÷µ“h@Ý «¸lsRÀÍ–Ä¥¼ ð˜Êf’¥?Y½Lý&.ê: ?¢mûAÚüp ®(Ë_÷ÒŸ„št@Aoýx§W(]~ïQªvŸŽ3A€Fý­]°ü' M§aò¶ÚºÕ?ëÙáä+Ne`Ô§ÛQàê¦@áÆ@`!Z—»Yl´bí²XÀa>Ê[viŽM]éXÚÑ®k;º¢§À}ŠùG‹å&ðäÏ:Þ/vÞ~†¬ÂWjÀ7•ïžuÀƒÁ%LçVCûzò^;X)Õ!oúCZÚÆ»O*ËZæBOçPÅEõæ…Žô+×)ËUõãpøyp¢¯=é÷Â6¨ÎUï~é¸3jðc R齋šðÓy{ÇKPóèk#ÞJÔê`]­D¿ÏS†ße¢v£SqDÔ"¨cpÆÚ¬¢uEÕç"~Ôcï9âDÁƒúd Êûï ¡Í+Ùôhˆñbw¹$ ÷ÛD#ç*BÒÞm£_^iOÏE¢1Ò×´´A^hÜ“Þ!mMì §¤ r¢©ìtOÅô;rçIæL+š-zZ©ÕˆÐ< UŸ†>´ 1ñšPÝ-d†‡ð‰¡esòâT9´Jü‹™Ñï&Z+Ý>y¿mÜ:=¨½ˆ¶¤éFŸQÉ¡>ÛÕÄt´ws$úNk`ó¾Y4| 8 ¢š©]pä ‹öÝEkÀ7­b¨B.•yVÏç@4™¡èßÄÑ âŠá®@zŠ ëÍCG §Û©æ›*Š×øR_m˜€òOØŽJ¦P‡^9Á^á´nÏ–&ÆÞÍ%è-ÉY`¼4öaàÍ0‹_`O ¾,UÞb½½7MÄÓÍ]ØWXS‚ϳÁ1œ×F©ºÀi7îõÊŸŽs(aY_ñV CàóÊíqÄÑþ€÷"­‹pRÌ膻ۊ—•–‚ÈÂõúE3,jyt Ä»WÆæô{@ò çÐ×Û# í¡åÌ17²};wëµ@~ÖÝNäO4 £è›´§¦AI¯ôŒ­æ(»ÊKÌçù€Ê»Æ5Êw@Œ8_OIÔŸEÎÐ%ƒ&#ó³Wª •M_¿ú¼t\£bÄõè*ë‹V?=™ËãíËÁ@ Ù•{êê½6§£`Ç¡O5`ÜïèLBÓ¦§ ?±^L³>–¦&¡>°xr/¶´\œæ¤¦ì´.w€4¿ÍÖ ø(8¤ÐN€ú±Ù¿IÚ¿G_ˆÌŽbËG¡œ²ÏNŒÆ£ŒyÒ94Ñ{VN™À”ü¾%"Wj%]=A;3ΈÞEå*„½sD ¶lÛKÐ…:’ö½Ä‚ýäŠ÷^]´_Ì8œ—EsV\M¿¤Z6æ]pC¥3?qGÏ—¡ïצN7ÝA5-§¼aŠêj‹¼ÂB "±âÓØèÇÅ9-”£¨±@¨‡aï jòq ÏGMOE;? £¦|#–ŒÁ¢ýïÝ{Ùc_¢&S©áÓÁûÏ#QawËÁ~bC)jy¥ÞÏbá€Z_º;üØJAm)šÛL'^£ŸÉæÌ®Ù¨=©Ô>ÝPGFÇúÙ£Û¨³>ü÷i¡&Ôµïуz¤´x)¢Qï}y:×ÐeÔ·qÙw£CnÂlhˆwÐsö¯?TûÂô‰ð-úõ|ïêac4†[ù\HŽÇOÐ Œ¶ É‹W_]7kBÓU Ò¤ÝÑLü¥ùehÞp‰áqZPÞt™QØE‹[jKÍkCh¹Š)醶>Z}?«yRê ZÿPqì~ÚÜó1KøJ‹v>}äåëXDÞ­·sGm³9ñ™0‘0¸À}þà#ך•~ ¼8“iÀß G ò¾ûÓ¤¬kÔüX@~_%ÊîÜãJ«-Ð…ã<?+ïH÷ÌÃÂo)_€wU6±p2ôc²Ž4ÃIÿã7U*A„5¥‘¢ìˆ6ÞŽsbSñ fJF¼f<•Ùl Ò:þõ·š]AևϯF1ä‡fåܹÝîRëz J©Ñ¼dŒ "~fšµò6¨Ä]èú-êñ1ݯó‚AóZÒXÓmkÐÖþñlRŠNñ£ˆ7±pšûw„^”*ès©ý\Ò¢ƒ—Ÿexõ¼À0¾ÃöÉï 0êf± œÌFmõùw`zU½ÇÝØ̆Nàe9Åůâ9 `I•zê?¬z<8îˆÿ›ïtÕåÃ5p®CZ¤K¦ìÙ m€“{úc]¼“ÀùæÍo²Ý/€ÿMç6^zê"ѪX”à@Ÿ%¦‰s˜EÐÇÛ¸µ(åY¡Ž…íÁ¹ÒüO¼Q|þ­å»GñSûHæwuä!î¢X2v˜Š¡H¿Á^•éâÃ1—ÿ4žâ­ýßñú,,Iý_˜±"Â6mŠr·ø9k*ÿõ‡ÿÔQ÷õppç£z””dúµåž5$]}´ÿ} ñ} ÝÿÛ_ÌÕÐ;ào‰¬ðЃs@â\ü¦ýá‹”èwU"èËr¼¬ÔÏÿíyS?dOÓÜëxÚ8*<¨û‡êÛPå¶CÖÛ‹Ü¨ÊÆw¦‡ÎÕÒí<¼¿qp?%ÉÍ5°l¦9W—¡F×¾ÂiGÔ|×Ë‘çÁ_øeœ~ü¬Ð>οV*RŸÞÇ÷+"ÖC\÷ÍR¦FCX=Ãýþ h¸ñûëDŠ~ý¹0×øÃ*œÈB“†íDÝçkÑ´7^Q!ši(.`a‹Aó¹,÷Ђi ®HôA]§žýô"ÉÃ;Ló­yœÇqÚ×͟Ű O¡mk¾ó¼>hwýV]ÛýIÀúÅþ}p©]Çš½7&›¸SÉ ð´-8“Þ8¡v‘})åÏ{Sé$gR‚q¥¿9_Áføó dhÃK䪂{Ü,â€Æ+éÎîa 3:Sº› —´qZ퀩ǂǸ6 X¿©Ö=?⠡㹎ÀyUvC(¿ Žû6l.Ÿž›/³=Ï€?p&Å!ˆcoÈ â€ðäY…QÚç ÚÊxeÔ Ä¯6sû‚ûCvY-&‰Ñw¶{òç­þ|¹”èÖÙ{ªv 4@ɬòò ¨<•û\ëj–{Ã'«]@Cý¼™lhéo’¼ ˜g®\üVÐ ´ÿvŽôÒßÍ3 U€Á»#CïÆ_áFÃK¹=0víž"+SìÁÉQÉ`V¨¾Úß÷ ,"^¶„쩃e¼«Š;QXw];‹« çÔ¿¶œï“ÛÕÏõ×õyàÁ=ë§÷OăYê-#`ôrž›U÷C¼/õõ\¾‰êYç;26÷õßÊ ù¹$TœH¾ù#åk:'¼—‹r¥xX<ÜѧßNEr܃Z–m¨Û÷Ë\oÃTöqKjèõï1øýx;‹ÐgxŸo ã.´2¢ï8LV’n÷)üIG­«ø¤ö-¨Ã÷릕 ê6qôçXA½ÛuxЀøØîdÓM4t­ŸŒg) ûs¹''Í£á-\úê„4ÂiNòBì1al½ÄÚFxµÔŽy½F#±*o…úuß`#C[v¾rCã²™7 ÈkÐÄ=:õ<=49_V?¦e¹#·ÛCÑï´³ÍÖÏ‚Ð,÷ãWòh.÷‡Eò#ôGέº)Ã-ˆüœL ˜A‹*•¦÷ý¡“Í[W W´œLœ¸+ ŽVÖyõ6 Ñšm²ÇeÛ»h}ôî{Ÿ´l´ñ¤õe–8Ú¾~íÅ%%!´{vã{Bd`ÏhÒ[žlÅéSé¿Gnk˜7p}?lk%ãÁ‘_×TD‹ßáÓÙâ Xij£2} DF­® \v~nÔÀÍÙ\¯¿5ïÀEæ:+›³´2·­å¤RÅ@v¹jÕäŒ$ȇ/8³PRâ£{÷j¹žÂàÞ¬3(ýne$¾ŠÒÊYPµ÷} ÄfjwB/ú~8 ê^W«ãŸ†wWpΣ\ÐLt¿ºÅ> Z]ïFjSEA‡+Çå› œ²õ-JØçqÝ&±ì/ÐÓ4ètèÁýŸ#ãÃÄ`P\K"|P¿yíÏHBÛ FîDKŠã»tW®ÉîIˆã÷ñx0M)«¢ué³ö™ì%Ñð–d•T z#‹ù¦0 •y³mî yŽªT¼¨VÜî ËG'ôµ¿-Æ=µgŸGp¢ÓÌù*óñ¯63´†Ò–âèÆé_¢äu9íWŸö ‹Ìy‹ö ]¬óB÷uhŽü³¶ƒýD+Á¯(ÙÊÙeœÀ¥™ÕpVü^V4ÑÁ¾ähùãG‡çföù+J ÷·0þð[oÁX\„ *»ÉJü|0îàœË¯·²å¨²op UEZ„%ˆû£ “K^A3¨NìÚ'²áQÔ &Û³¸r5Öÿ0ˆl”:¸'%#BèZÅ´+©?ï¡¶àÖÑ«_âP;‹î÷—s¨£ oí½ê â­ŽJ‹zÎÛ‘‡ã >£€˜x‡x4`h󊯵 ‰nMŸ&eCÃz;!¾Ÿíшßçç ä%è×½¼\éhL„à\tI( @nLah’9wú<ù:šªn£WþVŠ~ÇÛ’±}ÖC³™JåòI›h~¬8‡qô9Zà7q»]H‡Ã˜%CËb¤ÞK§ÜÐʨÃeJu´öœ˜%6ÿÚyÌ̼ómž0}ÛŒ¶‹ mÃ(Ðî*}ÒÛÜ€5îädä8~*.iþpDyþ™"½ àSXùt“µA÷ñ•Ü^ zhö.b€˜'âùšÐ"|Šsþ~{ÈlX>R6µ…ìàÝs%Š@kµÃJMÔú´Ö¹S@kKFèô÷CùÂÏÒcDúVÛ|50[ùH`«–N’ó¼[g€M&b—‹XØ›c'“ˆ€cÚÆ„Ÿ¨8u7Ž/ Àqê™Äåà%=ãÔöø| Ró@`ò‹œMîœÔ.{ŒýÁ„cRŠê§€(í¹f:m»'h§5à4uó¿N=ɦ+²»Ìí ]ðÝé½Ã:ȶ—ŵ´öƒ[ñÇËç©ù8æwóÇÒâøA P~yq2¢8 T~š5{•p€š¼Wкª¨ç™-ÝáM•èË›V ÕäS=Õ:aq=î= ë€ObD zŸ¾ýRY/æ6 c0´ìôL­£g³-ßú¯ñPŽ’n…˜j{m%¨âƒYcô;º°pG)vërà$ƒ×èDž\¡IÓ«¡@`>"Žzð¢¯ßÔQG« j‘S<(‡bwëEÚ1”qË/¸š¥M=íx%ÕòHO=?8/Êyüä”°=Bwnrñ ·ÔÎþäìÿÝ7ŒÓœ;8ß–6ÚW…¢sßéÇÙÎ;¸w!›Ž ˆoŸgúÕ…Jý—ËcÚQåþÂëÕXC5ˆ[Ø®=ÕùÏuæUô úeÛ1OÔ°Üþ Þ5ê¶uáÒAÝeÛ¶Jlß ¹/æôïδ-¹†3솈(Pã¹QÚ€È6Ô8èž“Î+Œš9,ÝDe/¡v¿G" N¨•ëùÀÓ—u¨MI²-òZú©f“~µkÝæø%®„: ÃÂÌÉ P§»”À³Ô÷ÙúKêîy©«HŽz%“JJÜÍP_É3©†©³hÀ w)ÿ3!â¨eß{½ó×j ƒIÑúõôOA»] =¯ÒkƽCܖѤàœ×@å4Íq9iqq͘Ð7â^,Gó„$„±Fh¹êavëZl*£§CˆâÖ‡mѪ¹8ÖlÔ[´®|#åó ´`!nôíp¹Eq¼A€ÅÎͰÌYØãÚ½)·•ç´3»àõïŒ~wWB)â _%8jÕìb£$Ÿ›ê땜œ,¨ÒýÏq øšX„µ…TÜ2ª1ß×€úgüà×›ø@ÛýÃþj Ð/“½Š®&ÑRÚ¬i`yz;ÊG,؉YÜø-+wzç8¥©w•iò€«l§”/¸ýO¬[ïýæÞXí àï«ïÝYŠÁ·¡ëƒûº³ºœÉZ¯DÙ³T¤ÂAܘà8Cv+HcSx;Ò|œ"Ù?AÖl”šÁä ðVd:†Å$€Ò«âÒŸÑ}€íXW‹Aµö­Kb¨õ“„d ‚¦©¦Íó¸RÐI}S}N1²Å¶t«Ái:b·\?ÐgU>i] Â)ßW¦a¢ííö^0£ºÿù˜(l_Ax÷ÊvÀlól¾°x˜W°èO–âu|q•`µÝëIÝÿl&¤ofmf¹•X„¸8ÁŽöbª 8åÝÁ}àzŽ•sµw0A¼þ]%ÓPÔñȉÿÙ™gûþLÈ݇}`jÏraGÉ-xպ׫QeÊZupŠ—— æ1Ü=Ä[Œà—Û·Rƒqé7D–\s EŠzùy>q8Ææ–™DßB ÔŒ…³kÿëËŸÚ§Ôâ <²Ê–|ý!æF®4ý§‹²(g2­×è%ʉF†/4XQζFh YÓ¬ÆV2BŸµ|•·ߢ|’gþÄ}LýaŠj葲]TD¨`|‚•ãûCQOö°@kT~gLw¦ü#ªx­÷¶â¬ú~Òuõi7 ª }¸Ô¿Ï»Æ•Çß· :Zœ€£¦¨A$jåRfÄþ{Bñ›³êwÔüÀÖ‡ƒŽê¯?̦±¸åŠÚïÙ‘ãß@OÓ&C…qÝ÷ÄÃ`÷Ô,L´\¡çÛ÷‡‰»Ÿuiø•{Y6 í¾ªˆmãA“|8Ø&¢QhZqéw'3šñwq£àBó¢³Çœr_£î„EÝR´¸4p½›­`Þ;Ù“ˆÖ´OjòN6¡Bƭ´­ã!4×v§Þ¨ÓVM¾C„ à0z~h¾ùpï{w×Ó~ë*£ïBóZ fVÖ¢èOüÛÓÙžyh¡2e^ÑC‹Û\Gn ¢e£ãÆtg™ÐJ®Ê縣hŸZîCW-ZÏ¡ý`væ!Ú¼D|óìü´­>ѼA…vù¿–~'è,œX™61 ÀZñzø+ÀpŽFFÓd®çŠ_õùÈ9÷ÿ{·n¼‘ó¼4DŒözïsáh®íü=m ‰yÿ–î+;m>¯ØŠAÿ¬ö> r³ñzjÔ;MYŠ2î@ëÐ+Üctž6¥7„þô–M"Ð/ÖŽ/݆BÚ¶³æ1À¸¥ ;óXHM(XXŠ€m¾XÅúñs8¦tlè‰!pñ$øâ?ƒã›51•Àsáôxxœ=ð^­´œÍ¿ |fê{–ÕÚÀ) ©kS 6:Ðà]‰’æn1ê+2”f^‘цx)K F'ðÞ‡ƒ„ú ¡¼Hv9=–z+µÏƒ|6%©@&º“tãd#Èñ^´êèù¶¹#5~î ø!ibè@ƉÛó 4ýN6òÖ2¨Í”Á‰»ªBJ ßbAMízdmº¨ëÄåžJ S¬ªUeмtïdhÅ8½mÔ²íÚ%ùv£Ï ³ÛãwûxèZî„¥94Àé"É!||Ð?ùHõσD08“_ù'ëò,ÈÃ=Ϻ‚ë`ô‹2….Ȍ۞y<T“–Té‡1i`: ÖS˜þ|Î]R­Ä; "V¯àßÏÆ–„Ž2hVe´ìŠ<¸¿–ÙIµïËøJð%PÞ¸û=;EëC¾Ê˜'ñˆs½zà󾨗cì ú5P^=ðÝzç|¡©ÚÖ(zwýNÉ`<J´)¬,mBÉ¢o˜tTQšïŽAJ7 ÊìÑ xÇøôÀw*lDÅÞ»/xY¼÷Êtù·ýá !Âý÷ƒúÒ„>*Z[5ˆ‰B•AÙÅŽgòP•qÆnåb?ªžÿFsàÖAsкßþàœËÇwdU¨ñ鬷i^jgd“ÑýˆZ±¥J'ìÏ¢6½YO1ôsüôl÷×}]ðì\fdhêº~¯gEŒ õ\ÖQÈ!@}—(ü}§ÊÑ€{é—ø'ÛhH'ds›˜ _öî¿#[‹FÞ¼’¾íPߌw°(GcFîiì«ïÐx>#/š”•;ɽԈ¦ú×ó5•ߢß)å¡OüêÑlŸR!ß<4_Z3('|-—)Ï æ¡E÷Ä7ƒqäh™”ø|Ï\Z)þõ•èèZs`± NBëkį-²íÐæ¿«L¢ÎhÛʈåê&Ú½¡›Uùê9`½rè2Þç71ƒxR‹QÀ]-ïZS༚Vñ(“#@†¿­{Ÿ ˆ„E ì^†£9t‹YÕ^@ÂÄØðûJ¾“º™7äÙ×bŽ R ÔáÝ^»ÔœB{QÉ€Öð±ßr#ÐGçÔ=ÎüŒ7OÕÂ`¾¦eú…X&Þ¼ö`V6ÞîÎwÀþ‘¿4S 8ZÙMßW§”·Œ³4'H¸2& £+Ý©/»”\ý›÷ôO^&ï©ÁÞêÊ$ú1¤ÏïñäW %+óå09§˜ç?.>ÓÎûÈzøù0ÿ“{úéÓã}¾níæú@—Ÿp˜÷ó·ä Ã’ÒßÜ(Lž©òEù‚þ‹¨;ЗâÚ…û¨g‚â<¹y!ê»÷ÆÑêÊ(Pü\îümñ`ßöŒÿõ‡Ú‰kW4ЯPš¾éÛ‡ùqãFb‰¿6èÐÄüeëSb'ÑTGø ù m4CPÉT¯\æJxÞ:>AVž§zºŒ¡ÅŸg§¥N¡eýäHïk‡yˆk¢XŸœÐ&®ønþÄÚ~ÊžRSFŽö^\=ñ"°Z÷è~^Ü+Û¢ÉwÏ7ë•þ9 H÷s!i΢uŸpÁÚ\ ÑO½÷ͼÈ"ëæƒü-âÖOßpŸ: ì‹¥Ã›ꇟ*He­o^aP¹+ÐÇöÉ_±ÿŒ +nÆû<ÍÂp÷wpj°ù,µÅÆžŽ»T׎5Á±þ#†×ØË†[¤¨M ¸©âDܶ—Ú…Ê?Í ø/láqÁ“¥·o÷‚°‰DdÓµ)*g×›qv$Í`Ó«%ï’¾\i¼Šp‰ +ú&Ù%È äÃØ2wŠbñÛW×ó¦ÒÝAþ»¼º Bî¢ïÏ· ª¹ïmÏj‚º++ø¦J‚¦Ìd»i9hÓßó3´PS„»‰GhzàôQ%¿Ð¦"ЧWÜMõƒW#¹>0L™±½ð­ Œ³­”ÁD^lûÅ70}ý}ò®È=0§y¿ë}Á"9‰j¥‰ÛŠ-.X³«üxÿ ÎYp¿¡Û£I­øÅ`ÇI :±3Nkz·ËÛŸÇX—uLúM 0 ¾µãŒÚ ÂDüpó1¹j˜¶ÿ×¼Òˆ_”lUò”ÿ1ó1xü?ì8ñJ$%óÀçáóÿÈÿëÿÉqÃü_9§š(ƒLçc>cææþ“szè¿ 8}m8ÌgýŸþP^ý…+GÂa¾qEy°SÒŸHôý¬—‚¿ÖTõíÅfÅñ2TcõôQiãûƒº,ý‹”÷uY¥«®‡ïƒæ­—ŒlýaŠšV]øGÔnwäL³SÌa.×}'¤¬­ÉªÑ &»6¾-ÝœmÉ„~}xi­˜ŒÆúxcJ^£‰]¡õÆ­ hgDsjVÍHÐ?"qÍõæö_ý%ý·¥ÔÞUø/ÑòŸ[ñihuéñô·´ñÚ{íX>Ú–&#^ý¬ˆv{*T³ºUëÛãö—!ÀÁï*b ÜË*)%¿ –㬞ð¾r,N¢©”°w”@¢”V©Ñ—äØÓvº_%b3_Ú=¤¨Â’›ÈXÍ)µ–Mæn ;±¦áÄk êëéÇü€©P3ƒøã)`MVÞ ?á©EFà´Œ'Ýõ‚ãNÊ…{²ÀS"•hwö4ðû¥›z$5‚`áÆ}OÁ |di ¢ƒ[º² >Ë®µzz¤¤žÜÔΙr¹‡uÔ¼ñùÜ>Kº”~]®Ò7U•(Ç#š vµžË8˜4.¼ ¶w9Z¾¹{oςλÜHÒf+Ðín¾&aú´§Xƒ¿Ýƒ%é·5?‚ѹl‚Xݧ`<ÿbÔç+˜†—)=7{æÎ£>G4áÌYÁb–º7`åg§·3Î 6~÷ò $Á–7ûÇ m)ØÝÝô+!ð››Ée ¥®ë$<“ ŒIE!Ø¿ÐoƂԃk‡¹ƒ¥ß#OÅ¿Ñ>\·˜_̺Çäfï~t~Š>ÞþÁÉ}ú71¸ÅŒ˜tBfa@¯öΪ‰w èÛÛëjµai5±€Î˜‰Ó¥B£€ü³|­ÓUàø½ÖÑn œ¿¿?z$H Ç[åªN+¾å]Þ7aûþP.Z„ ÆøH^=î±(~j4*ð: LW\eJi@¥¯ s¹„|gÆÂAäËËšÀã æÄ<ð4Å$NÔŠÚ˜gƒdà ٥bo\ ¨µ™™/†ÃE9—@Î’´ÌÕÄ (¦ßy̶âˆR³"Ã@=ý@ù(ѽ¥ªíå™èL%}PÙ<ÂGo;jGͪi3AÅ{=mµ4DøÅ{ìAS#Ê›\þùßSŸ±JåqeЉ•V5®§€Sã5®·ô¹á´0óqk1bÐ }‹…ûôgu·LÃÀà“ý¾r8ÞL3i¯×#5’¢Ö0fjh:Õ9Æ[¡©[ÍÞ`2îëÏ|ü û¼54AèÙÎÉ8W`*¿/^ã7ú}/µØÙó=ªiѬRÅ.AÅ‹¾ƒ‘™Oy Ã?]ùo¾ÄèTL^7fÄ|™‡Á#&Wó»ÃàƒÿCˆÚn¶m9ÕWÔq‘ñŒ¬‡ê:;8gðùÞažþ¡þ çbÐ>ކ؈?þAÃÁ«Bw¾Š¢‘o1o|'TÐ(U[ý›'hÌcp9I¨wb/ Qæ£I³oWF;“ÐÔ2›¹nÅ=ô;“›l[[ ͆Ǘ3½†æ_·[ÞÒDæ„híãMТ!ûñ`‹]´4IdZþ¬­¼r$Ãj~ŽÖ$N¾¸Ãú­ç…0¿rbD›Â9y¥[hkìâ^—Ñ(×€¥«ÿ甉 `í¼*2¸Â9²íÈ/@kúL”4¨Mጽ „?KBß~U€£ntñܼ§t’)Ë¿^¯Ùüqqml| X]Ún顪²©Žˆi U%ø®íCôYÝ—°ïããÿþ1¼ŠyÞ@d¥KY’ñwüçùa¿ŒÅ“$\ì¯Qs5ÿlù‡›‡ý0Z¬ÚSÎÞDmz3yu÷ÐOVò«ßw·öuû{Ûïî¨C7qÞÇ@uf° ­,q nüË—>¨×z.jÝ•õ_qòÒÍtEƒÅS]ßcHî ë_‰ý‰F,´½¯¤æ9³ãZHÿFõ5³üÌhŠ`ÔÓhÐìg½iD‚#§Ð"¢Ï9$ÙháÒÀÕÔ»hqKEbÙH-çkFfÏ1¡Õï¤<Ì%hCüE•üÄ3´5ÓѺ¼ƒvÛLÆBtk‹H’Ë/p’Çç¤$EàÈw>¹Ç€Ÿh‰3„«N<þ÷¼øzT´“„Îðnå} Ù£O¦ýÒ€2e˜î[)+P›Çf½uÉZµ {Ç´¿}.09à%Eô÷~ûQÙÑÕ}¾4Ä߉¹¹ÇÞ1á—Ôß.Š—Œhá&ÏQöÚÄž(ü–àÁRôyæ'â¾s•ĆPŪ†œ­ˆ2ßQ¿i7 b£ØÑ„OÉAâ^¢¼ã'ªÉ½Ø›dæ‘K”ƒ¼ºš×¹÷ò Ø¹ˆëdˆ J¦ÓfŠ-û<Ø_’²¡ª¾Õ‚ Çí@]”íô¯„>ÐX]¾¶t” ´šÞÒ„º~/áOì+,A7wýÎ>ŽôŠ„ƒÎ¶[‚“ßú•àŸ`x]åBã+}îδú ÆÛ<×ô^‚©ù¼†s£0˜ÕË«K™‘…ùóí’Íi8³»›²XœVå;xC­ä`“ô%K2ν¿J¦ ¶“øôyà´Wlb6#™l[±ËUïúèÝA? ™')¥s}Æ¿að÷ÿê ÃI’Y“Vèÿ3bæcðˆéSóÿç1Ÿ1ÿ×á(M·ð²Xïo?ŒýÝ¿}"¦þôŸøïcÞ=ŽéoQ¡ÛÕÐðÕU¦Ï8?xÜzØ z¬{êÓöÌá{âÐþS7ÂÔƒýá?ýmñ¼cBüÖí°Možñq4[ääw ãiÿ›4ÒhÕú”¼Vtr5¬¢ñ%ÆH÷ 4E—Òp…#ýþ:Íu:uëÐ.› Ðc‹F‹cø¦ÑrIÆÊK´ÚpãE:>B¬¦5…á1hëEúòžÚå»ãœp.°.Ïþ14ìRöëU΀+¸°1(ðŽô‹N8(ïþoíˆ©í¡¿yþ#Ûw¸Ù+¢å[{…@QzÞ'$¨¬Kæ)£ û˜5™@wDpn¸Hè/?VK#&ÕjJñF`5ž5Ž%;Ä+çIÑk-|p  \½<ïêôvpS€ßu¦a%ä0_x©xHI∶ÕZ\,ñ!RÞ ùGÙ~Õx}Ð[>óS9ªÓ;».˜t¨‡øxÛ«÷àïˆùþŸù‡¾‘ì½ÐÐXûázÿõÅD 0‹ ¶F|¤å^Fã„—<½žÒüõ ÷I3œÐ´7®@MMé_ßè¡ú•…Xý±4‹ ]F yÍšÌ5Ñ’ÕÕoÙbhEuAöþ¾þ\»£þ*ÊKm °_^ë¦8ô‰»¸Ö¬ Šshï“ÜËæ-RÀªP:–ØÃÉÆ»—ÛO9âk øœ›+øTuôB ‚ÖqWiˆÄd猒÷u%‡Dœ4>úYÇz‰] È íhÊRë›·ÛÚŽ´ÎÜ hªøï_R{´.·Ì/:qEkey.Ð¥mè?ªýô”W¸>ƒô³:[.`’Nz5),¬œ:‚Ô&²Àa{A~¾-8ÕÊk.¾ã”$+‚ÙÛÛ÷BQ7ð\"£nÊ^µóöÙ9·€ïô5ö;óKÀôòWNñ8!øõ{PL%œ´´V¡×ü™åŸÝ„ úÌVéJŸˆkówT1ÿ‰ÍÁÝW‘ßAê±DÚM–a!]z}Jdãå~”öf‚¼N¬'ö,P$#­`¨s4Eù Bk”.­ã_¬ÈåKlÌüÛ¯AåÊ2ñ5þYPõ¬%/}° jŠwKîÍ€z ™Î3|Јr挊WÍ÷Ik/Å£@«¸ïÊCÓ~ÐncÆqN©…S”{ÕD—ìAWŸ6McøœŽ¾íŠÕ z+GÂ-À@¸}NïIÌZ¦ô¡O`˜@nÙ¶ Fw{}ëÙ’ÀØ*ƒ>"^Lêû†o8p;ä¦çB8Ü{)NÉïﯫ„?Õ7*õûG`ú#að‡ÁFWbê2Ýúï>lÿ«Ï†g1¿‡Á5†ßëœÿÃ'Ö5o²\“Ì—@TX_Ô-Ýë>‡>ÃW˜~L]‹Ñ§ÿì{`ú¸at)Fbú±uºKxjÒ¨Ö51ý™0u£÷¹Ààú—]ûxÔi4ú‡¡ãtŸóß~M™†jùCñhº¼kù1“,š3Ͻ½„æ¦}ˆ°ßù¢?~§¾-ª …|¾²*Z´t”ŒæIžZŽúàOîBV59hÃìÐ:1e ÚèÈ]?æ1†¶ª‹~÷g¢]*©çЧËÙÄ)†c°÷.æÏgDnsgDw9ûßþMU.% –“âžzÊD=OJ_„£™ÃÔO¸¸LMîmÈ Š‹ÎÀŽ s–Mý1 r·¤¿4»“¼\€þõêq,O`\í7˵ûëÛ+¯¹ÊWkãËé c'`§ˆ7¹pÿp\ ¨ã_„cEr÷$@¸î®å‘¨wôÞÎ#<àí¿öÂW_ ìZ3èv“á$Ý‹­\ÿ ¼õýubÿIµ«a¿tž Ä&éšäÌ«Aâ)û­d›X²¬×ñ e­W^í²ò g±s®*ª‚íw ÝÜö_è9ç,ê´A鎚'(O²Õdª¼­XüÿGÜ—@Cù¾}ßw"{ö}lñ\ö}ß·D ÑF©(”„BIZi­*í…m*KÙ*) Ù²/™×÷gîg¼z;ßßÿ¼ç¼¯sœû~fž™ÏÌ3s͵|®EÅÌrÖ&m?`Áò(!Í,ð1<‹k^%¡úŸ‡À¦ÝýÍ~é<°+ÍbøòL¶>kzN~Áé•·Á9šhĵ,\®n!걂ë€Ý Þåàní%ߟI’É;a·êÁ˲ÆEµžÖŽìÒI68 ©5kÖˆý:íáŸ?±÷£Ý·(ó͈âŸ(î²Ô?Dúîoþ!ZÑùH~‘ýXú±>äEy-wÁçšý‹áñžþ"åe»±ïÜLe‚«WSæO½;³Ñôp<ösTô˜ÐNWÊܩ٠£ðOôدĮñîlÔ Þ3«c㫼åWb“ÇÞ¾Ãfœ©o“ºD‘ÇDÎþ•c¼@-Ës{»ÐRY–E[Ûáú”Ñ{ÍöÝ9ÆÀüÌs—eN°­l¹w––8ôßÞ® ¡|‰àÑ;ÖÖ¸ øècÌr*ÍA€åþç)-â'F•¼ôƒžŠµGÛA4¹DŠàq–ÏÝÆPÊ\™oãk"_´… qBÁ dÕ‚£{4=A.ö\övùP´ŠíæO< *q?‡Xw;ƒÚLA3óMUÐx(q@'´"ô\C‹A;)ñqè—tЭî‹mHm‡•j#e>_Áðù‡™sJ9`d%œŒãgÚW’‰`êw‡.vnÌÙüŸêo¸-Ô !×Àªüë²Î±°©öÏΊ»/“ûã«V€£èúÕy¿Ô©x-Wšf8üäNJáW¿Þ¡vŒÜê‹c7‚Gˆ™3Ѽ„Þ8ü,aÖ1uàÛ)â‘ù^X5ç›Éœt$nòìÙ >’ᆲë`Í5ðgÛ ëüS’ ŸxS×7W0|ÐÞóíÖ8zyN±©×KÈŽDò·TÎþÆ"0KÛÀ©åu~ŒÎGòˆô'Ò¯øüÃ%~:^Ê¢óÿàÉç#ù‰øãŸ,- `šïñî‚ ˜ÇÊ•¬ßg –;¹YÇfªÀú¸AQK/Ø6ϽhIãtá!…pJh¢Nlç穾ê=çÓÀ53Âù7Ç0¸;rY8ÆÊ§æŠŸ:àmâø¦¡q|÷Њ +Ý…UC}&Y!â`wþ€S´!–5?Vè€]£•¿ j‡F?®0~BÊîÁz-úʳ¤D±WO(Ä`eÏå¤ó/oÃãð†äùk(ÎR |_ÿª?‘ýŠäñhž'Š{"}Ô@Ê9PãlM™ÛIžƒ¶Ô.Dó;ñUÜP+ót5ö‰ºªxÙ.eõU[{ŠZŸÛûyⳬB“ ÖÑvÛ-‹|i¸™õõ{ ÎCüá/’åà{ÇÕçqXO—Ï–P†ØOþë£Í_QüÄc;¯'¥Ma¿4¼¢ÿ޲1k§UÊ`ãºN§Fvb“»š_¯RƦ[¼¨ª&±Ù/5‡Þ´¿ÇæJY}˜±›@ðïòó¢ :üz“PW6P*÷rN­Z3ß/Ãæ×)óÿðõ®ÂÔ9`~Á×{ÉmØø½£HíÅÀñ@CKh”¸<~Æ~áZÜ-ê=ÇL¯2Ãg’-ødóëÄ÷Õ_ó鹘¢sÀï~•UÌÐø‡e£Æ5úAvCžÕÕ€ü—·çîÄò虔ÆdX~²kÊ qH„ËÖ§Ç C»ßó] ½‘Éü^†ngÊVÈú¼ÿ0rlGt3†@~ï%v»“ h»Cå±w(§O뚺¤Ÿ´…¨ܨ>IÍšw ’kY1tF(Lñ h7HÝ}¶t㌿oɯ}˜Ó«þÝæ‘+—F¶¶—½Œèe™›Þc^sϱž0QaqÕÓ+SWª9×â$0Û—vs7o1˜—õÒmˆ K:MÚˆòh|¾šõ­F‚¢¢>Øò ·ç¶©€]‚ìc•:pX¾ãüõàXuoW¸=8•¯ý-Y ÎW†›Oy—s^§ÍÌÝÀ5oŪ\ž;àvïw¨©·/¸7õÝz-žô‡v\Vç™·K³Þlç¾ »ù¥3WŽ–ò±G·B@xu€Û®ÌÍXï¼Ké­—ñ¸ ’?¤_!»réÜÿå¹!»ÿ›Ÿˆòf}ŒôÒ[(~„üC4¯é­§?ª“L³þŒk"û–l¯"{Ňþ˜‹ˆôVrú{û±N ŸHž‡ˆøÄϬ,‰ù¡ÿPIö㳊ý©×¾þ:[…G,ã]Ì¿’âÑÛÓ-/ÃΟÌHÃvbCf Ë·$†l™ÎîØ‹Û¡H¯åkúý\eK‰÷Ä>š=ü±›¹Æ&»­ó.ö{f w™[ŒåO5´ÆÕÙ›wKPô[¹¿e…^лÏíÓw†7Ëy/ÞL&â„õ+`Þµ³±ª#XºK2æ¼¶‘«Ü;s€s÷ýj,¸7Ÿ¼¦ŸW¼Åo3õºTA@@øiN„\…m×qå‹q¹§.òۛ”[7ïÃyŒåµßßòOnñŸ?´ÄÛû@Òµm磫@ZB»õ­—ÈšL® S§¹3Âbß·€¢Þ¹–)¼ =¶Úά_|s„[¨6ðà)ÐmJ¾ò=Ù‚£âÏ`ã1S}‡ç°Éž-1|…•ØLlò‘‹)‡°9¡¼F±½(|ÄÏ7ÝW«Š€ærûfµIG Ë^—¥yù0Ô\ÒÐô“¦äã\=k³^z8ÜRby7—â…ß*‡ûÛ>£ïÝìÀûÔÜ룊ðߺzËOq7^œù’ÿ„+ßVŠp«‚¨`u‚ósX^8tG¯$ˆò/ƒÓd¾”rPs2Hƒ§»ËÝ +ùH›EÈ 7èleE¾‘ý9m ¢=ìÅ&ÔíOg«i‚F6¿ÂÌq>ÐÚtYæÃ'*ÐÞ×+oB7 º•IŒÛ~ÃJå²ç®OÔÀð«á­DF0²tf30Rãª)3§yÿÐG"‚ãê¼È\q7ˆS,Þ³+«ãñP›òß?to»OŠ‚Y1ÁQHí _ÅvpzÄ–ö{8'ц¿x®>ftá‚zàöŽÏ"ï² x®Ü¶a¼ø&Ü÷ÛÞ?vì8»š|?9íëHX «¦…îÑÔCÀò-Y—K Ðó¢c×Xs9»ñíºL**{Ð=âôgl$wŸ¿u‘¡¿g°†_ùýŒÇeÿð‘žûŸò‡È?DëRÿÙ¯óÿà ‘Ÿ¸$ÏÿÁ'"‘ü»ò7ÿ°äÈ^ª[ðxÑǾ³*«G)ü!9¿ ZêlÝQщ?ýC²~ÞÐ_&I›ÿÉxW(ù…‡%Ë1žP_G}é+n_w\¯¿J}Å ÏÇûFËS¥–…Ïÿ¡§÷Šjì)ÖÛ8x_Ó2Wø¯×Ký`ì—‰›nïl”.ïÔ†xŠèwŠïk`Ó_Šv¦^ðÄ~”,ª‚^qES PUu¦KyÎRòÂí¹hBw«Ú'%Fþ©?´)wU–©§9™Â=dnä—™%ÎÄ·tdgð*J‹Ü=µø‰[ú69Œ€à꼩æ9jn˜¡-¾DïVË®K¼ ➦ØtïÉ­OwÉTv€ô™ë–[ûŽÂ )>­àB%Pà®(¾7Êl*•¼çõâxôõeÆ a¿³»ú*hY21Ïñ¾íºý»ƒ§)s€£F59zå3£[g¸EŒ.{}«L¬jöOЦ‚sÏkýìd0¸6½ÃóXŽ…%i¾Wñ¢>‚l`QnþTÙzÞ~ÿ•=NŸ_þ°5Ë—€5²#1_ÁƬÝüõ9pyIôrÈð|ìyÐX# ¼[®ÿ0 ~²ÏC"+ÝÁÿʉÔäŒ3p»´Q¡á*¬Qñ-n÷‚Ýj¦Ë9EƒÚX·­Bí7ÚÖ¾Š5É.«]ž}èa:Iiò½ƒòW>òOŸ;i鈵[Ý-ûKz[gƒð‰Ðrn/¬Ï5Þxƒ ÎÇ¡ü6$wKùÂ¥uPHÞ–Ú«hEöì¿å™"=†üS”Gð7þ¯‹ ë-äßV7*¦)|§è/r}žwJæ3ŠòÜP=^ìP”Çö±££>ë;.ïx^²?‘üæ4žxr¢ëb;½ÂãçÖ=]M³Oûa9{JàÇÖ+faÿ„ßAùãˉEß"±a÷ÂvÎѵدۖ35&ÚØ¨^¤mÇ{l¬»EcY£6Q”î¯Î…MÕtˆ*|ûˆÍÒ¯ Àæìs QŒ@pb ˆþTÑ»Þ::”uQ¢¼¨Ëa åéÙÑ´žèÒ¢ä[ Rv™ÝZ€±y×íißF`>¤&~ýrð?ýº7‰T({¡3}fŠ5p¾>R-lVÜGÃúxf½÷ˆvççsÀŸ·6•‡È‚µÚÓðj'g§³³‚e¤–Ú7„g°\û<´ÑÄk|5'üIæÐšÖLoÚt*Í&*dèßNV·‰ì›&ËÐÒS ׺üõá <Ž£¢öËï›Y¨=>9»]”F¿m˜ó@³ÓK¹Ÿá/lŸ¶Ø&: eù} Y ·_‡û’Z¬,Ø×–î¼ 'ÆÅ|ž¬#â]³½¥`œ˜œ·†y^.I‚§é?YPPçËiM0Ÿ¨lIïé˼‰¯»39À:Ìæµ¢§ØÚÉÜ8øì×é©Ê‹Ö‚cŽ´z¾_8ÍÕóëeƒ‹ÅŸ'úõàzéëÈj1_p—(ÝHg EöY»r"ÁkóÞ³-—EÁÇ¢ê|ù19ð3¶à¤ž÷Û×jÖÐ(Þo ¡/×údC€Þ±ókzl Ö$]~*¬S çÑ‘žCúío~á¿ù‡È^ý›ˆòÊO€ì?<_›\ÿñOæÿþVwˆû‹(ß”|Œä¯cD<9^óöÛ釙½-{ñ žc;«NˆPär)ˆx²^Fü!žŸvtµÑ¶µ{°o~YÉö^àõ‡¸}ÉuÉ2¿c}mÛÖ‹ë`ýCx’Òý±Á¢ õ¼s°a–'q#Ý€ýŠÝã¿ñ*.—ãðèÁ±3©Øän£ÔðúwA­ë¤±ßÙ븨¼wAá£jÈI"PÝȧÿÂN©grK:Ùsï:0¬fë²K`Åù–W{“Ùúb€ÝBçü´Ûà<œÐ›Aõ¸uCO²Ú/÷ÑsáÀ7Þ3ÕÎP n§~3¶€ÐèXn˜ï,³²ò±Oåϼö€øK­É# ©Ýv8S3¤^–KaU\ “WË1¶ÃV4fï÷§‰…¨Sù쟚A9:"Ùjã»äˆACYÞBJ>4g÷Ï~Icmº4>(ÐÕ¾÷íé %Ð?JãûP EU^9P_ðåêƒK`ìwöuãÅyy{°7fM0˜¥ÞЬ {Û6žâ³`%;AÓPi6"uÍ ™ÁNQ'¶¤‡Ü›d£¶“ýmf9›+àûÜ,îh0¸´µ]©Ÿ>n–ÛG–{ÝVfž„iðŒ¼»½è ðÖ®¿øæz'ø.{csü¬Zqµ¦/i¬öNÝ+{Ç ZgJ‹ÇïÀšå=‘©yìdìÄ£aÁ Ëé±LÏÛ×ô§ƒ¹y®`÷*O_knÄõŠg"ùAþàÿ”?DëRyDzéY<^ºÄþÃ/\²þ-ÏÏ/]êþ­å™.‰Çâù¥(‹ô+9~„~üãzÅ{Èõ\¨Î¯?$×ÿ‘_ŠüC¤_C$uoX6ý?”ÝÕÁ£ÿ§¨?ð"LaÙ¿û‡ÛÍN?ct‚ Õæçeµ8O‚ç—Î=y”7{˜äjxN“ÕµUŸ¿K–c¯ö¼ŸÆN]°*\€scÁÙ¡C3À-éâ$˪ ¼ü¾„„í¥”úC)*ªY~úè7±ð"ˆž½÷­(R ÄcMuƳ›@2Tþt–CH!m¾œ+\m6³aª×úD-(Û¯UÛc\ jÞyÉu ¡¢œ¼=)´Ôúy¨YA;ðPÏ“¦=”øçÖ»a)€IºlØ|ËŒŽFèü,Í‘#¿ìÀtÔP¦Há,˜7E5ùzË–Ë¿]èª{ƒ·¯ƒc°…Lj 8<Åj‡9XÁéÅ÷7î3ƒ‹ufi¤ùipí%˜Åß÷›1£=£ày¦fÓÉ(Qð~XðªHà øQ…wv©Vƒ|ÜXn­3dsº|kø|¯¨†]Ä}š7=çõ|‡ó›; XõÞÚ.Ž€õJ3ΰ¢µÐ½ ÜDßw¤Q\é5¼þi o¸´éU$gè{ø†ç¬Rg†°×¸~{׌¸ÝE S¢ü2œ7–$ÉdPüC´’óY^ÂWòý8¿8päºÈÑãx]^Ç´”G¼ æ±Eô=¥nå[ŒªÇ'$Pä! û¼Êû™óÕO°A™S‚ù‰·±aáo§²a¿>^»å’tíŠ:QƒMÈøii¿jŦ6};aI`Ãfnï|±÷ö»ëò&ì©$¤6m›äýgbÕшꆣ_ƒMº€V½(ùDs0Ð?R¿.Töä~˜¦Æï[¼luåñ¼ÖGg$“Âl½8K‹þe(p®ï‘HÖ®®ÝÓ$_à>¡æØmÑNŶ(¼ç ÷«ckÁ布ú¶å‡ÁõDØ ï p»túküÛàþÔàѦð¾ºWûÆeصÓ^x‹1('\¤Ûfµ‡ˆ«é…±Þ/{==„Û‡(®ˆäùqKë0þÍ^E·#ùDzÉ/’o”€ôÒ_xÞ YnqÿðÖHÔ±ƒþù‡ä׉çíĆ‚ÊÏ3ÇF¨¯H:YïÅFZX49°±¤¬ú|]l¼óæTéþl2ˆ:㉢'6-´õ©î$6«zéh²6w©Ê§á^Ðin7 ÖÜ÷u¾ÍkÑ×'¶ݦ ‘nû®çù‚’^a`,ÎüH}F˜C+j\WVƒ>Ÿ‚>J<4õ TUZRâ<_½®pTýÆóÄñ:Äç²6æJ@äwUüáX Ðha.ñ±ïYOè‚Dƒß9Æñz²)J:= ÒýqÚ\´% [®Nlɘ¹w¦‰©yGAÑ"amCc5¨¨þ:ت±ÔªVè3M«ƒ†•ÈÌÍÒTÐìeþ®Rþˆ×*î…7oÔ‹U^NÌ wØ’åƒðnXù Ú#;· 0šï§Ë½ÀÈÀ‘zLŒScÏæ¹&ƒÉÐåÕ…æf`¶¡bçÒw° dì½Î¬ –7šÎ}ŸÖÛ&”oߨ¶ž±4rB`q…»f^ί¤’ò°gz…Ï;—e€‹CÖÅÛ<ÁµðŒú¾5pW´Û[ô<ªN¹ßª ¯˜W²©±´àã2ÌL“pü.~»~ü¬ÿ…] •ñÓÚ/ZnÆÏVU]¬±ïמìe¸ˆâ¡ÿVÿ„ìÓ¥ö*º­8_A–CÄß#}‹òÊ¡z"äÇá¼ÙîÃoGþÊ+]R§ˆóöÈo$ó‰xýâ>¯G$ëcÄ74ê'„ ³¶Qx}rÈÒ¾¸HÎÿÆë:¦O]Õ2¾ïrñvXWá7áîîxŠ>5ü~´ûÖ÷ªÐØ›& ë´tÚ£¸Ô ØÖµm6T¼é¤ùG¼Žc4—;îëƒlt£qÿ-¿flâÓéÖ;ŸOàq›ßi©z~m+±ßoÄò~rUØéþ×ò­@íl=Ã(;M©CÜà;·Å9ÓÔì”掴Õ»ÀÚ/ìòLÊ8…ý¼ïWܶk[Þ«¡ºYó=àíû=»Ùø3gå®!µ½Ôkg@¤¡Ô’·kÄ”¶ÎþöVñ³ƒ’¢wí@RÁÖäfOH5ìó»ü³d~Îùd²€œ´|ë£7~ ß¾E)óÑ Pá›Tm8Ú*]¤*.Å4Jÿ‹¬¾Á¦@Ì ±º­É :¯Ö<Í}úÓ3¢ÓkÁ`ÏÊþ‚ÏV8Wv埣¹Ó«*`²WaÛ@Êv0[¶ùÕ¬ùg0ÇEúÀòœêM…4+°Nšähå Û”¡ËeÕgÀþ"Û¸íV_pla Ûμ¿ÖøkRƒËúCŒºî‚àú’‡.a#5¸[ O¿Ý ŸmLÍû“^Çú…K›ÁgÓ–Ìó-÷Àoý•ŸÓcÁ?•cÉî„åÞéÓ“ƒÀ‹u>•NZ°¶ÁÐãZ{ˆÕ½Ù=të>Я`Ž»|nk¬ïŽæ¬©ÇíF¤¯¾û[~éßxû¿å—.íSƒìY\’ý=»±ãÔl„DïýE›æf’“u¾†ÍÖ\æÂHìO…Y™Ù)uÃSÇæ€ö4ÿ9Gñ Wê î?“Œí"ª>i”ü9ïÖlÇIÀÉçáäæß\M¼–"k„€'ôtåq]àscªhÞ5y‰Ø:fêݬ»ÅD•«Ý[—«‚¸ìcÝð¶hx»Cþdá HúSwªž™S¬‘G8“AžÝù'Só2PÒ-Œ¨–¥Õí¶5F™W@}ÏÕP®ÄhÐÜLÿµ¸P ˆN *µ>wœ)Aßö¶ÔÚ¢ß`0u€›V?à£Àè}ËÀ8Ö(ÅþÓv0U ë]]fcm=?9²Á¢3n[Oèu°6«Ú –¶RüÜÇõÇÀ>Ú17ó™,8ñªï3Ô¶ç=ò ,"àÊÊ|ãûö»àVBÅœ+ú<2´%túíÁëÐIÎ3ïÀ§x+é¸Õ¬PÎ6v¬Æë€õ§/Ç©Â.}çíA| Æö‚I«iøÔ$ßD b½Í»VÞÏ›ÅûL,íÛ†ìSœ· ë5$WèéOtò#—òHß"?ïWCÖ{È/ÄíAô½%÷ICq Ä¢üè6Ø>¤5BµOímomÆù”‡†xGToÛ6dÁ¦`+HáÈßÎÒOO5é¹’@ö¸@þZÍ 'Ïs†wPD.NìIæ%G,§-A•›³)ÿè&P¡“™¼Ôê/ UßV é:j Ðr0XïÉPÄÛqwº:Ì/„ihAwð”Åå‡óúð¦j«Òýa0ØÛ"dV}°pÕª¸”×`¤õyôÝó»`Œ½‹H}¸L\5ÞÐ[€i$ïê¤ 0Ë^™S—¿Ìë¤À‚©,™S9ïyDƒ•ÓîÞw\C`}öѨçœ`3&™ç9úìvžV¬˜jм¾8æú‡Kû}§‚ ã²÷à|d6ø`Ï ¸$í7××ýâ‘"ü àvT6qkM3Ïñh»r!ù bN´¤yFƒJ“_ïÍú|¢rQ>X'Žõž;,–¨ù ×(~‰äÉÙRÿo©žD~ :F÷/õ‘|"¹EÇx~)Y¡:H¤¿ÞBqMÄ¢<¤¿Ðí(Ïñ‡ÈnEñNdÿ"?Ïg[Ò_íýEêà¢Á»x=.ÿ¨ª/&÷Á@q”¿Ö=º¿:#ýÞo­§¦¾¼æê¬÷xêÓGN¸Ÿ8ÔLˆ^ÃØp)фڰá›áÄû?‚ÿÌ/í¦¥¶)»ç™ÎXŸ0 ¿K‹ý6±U\!†‘f>¯*êh¶mµ@åtäIHº.…· çµÑ?¤M4[›Lkš/}óf ¥ÁÕ}ÀÆÞ’œé‚÷QäÒ¿->»l¸sš·äg”í1_ÉQ[àZ9÷ÌÑï7%xõyLp1?›ò6ʲ䂨ŽGÑrX®+Môâñs²áêþ YWäµÿy)HžZYJÒ3â÷šè/ЍA›]}`EsJ«Ð‹Õ Hâ8/'t”Ð×(|Ï5»÷OT¾õòøŒØRAд+ %fŸ­qýèßïg@»tåÁž*ÐÍï1åœý«?W醃Aƒžè2úQ€å‰ÆÇÒÁ¨6-Dj3˜ðÑ甥ç€iÔž2QÅyýØû>Àûåg°Øn&ôÆê,X y¼·°î£ê›¬ÊÛ÷±f¹›Ž}7ý†ºÏuàÄœ¸/ŽÂçﯻH—þ \¿ýËY©îþºF¯¸ÅÀcdÍK-íðº´ËI~]3™M^5kÑqÆ›íJ§€^¢+¯#”kò_¾1aMn×áò·$ú·ü¶ó—òøKåõ£@úÙ}8?xq‹L+pâ|Î"?‘Ì+âù§d奾øÊðüĆBüvüùÈÏž÷©Y²î?{ŒÕËíôÿ¾žR·Sî½ú/J^Ú…‘6|X{;“pÎv[œw@y:_.oeP¦¡Æ¾šyÔj;úbßbøø­%¿àý-¾¿ûnröÎâýK¯o{ÈdëÓfl˜Ö÷£Ý&¼éH4Óáî±ØøÍ†é§/±ñ  WB®6ut¢Èäp6Ëœdé,sl’{ª®acÕÎK Nk¦9¤=´|wüjz›§çùße#{¾bÊ®:`ö`žôjÖD‰Ë÷Gv·ó÷›”<Óá‡÷?&ÜÞž8™[âÀÿz™q®ÎKˆ¬òýê\ž¢/ž8 Ô1€xqW·¿HRßܹ¤e—-—’™í?'m+`EÞìd›ÈOò{ä› ‚R¼…cnê%Psmºªûî*¨Ý?ÐGšëÞ®9\FDÿÜÔAçpЉc\ÉÁizÏ~~.^Núû¬ÕÃEm//5ª¤†Mf‚‰KŸFÜ™`Ús~—ÜX0?¥ÎœëBË5Tô»%µYâ±±FC°5r¾Ù^Íöî99öFià˜˜;"’u œºf_GužíÖG2߀ëQ·GËøèÀ9hœÚ†göñÏq/‡N+œðÚ’6Û~|òy›Îƒ¿ª@¹¤°=èÞŸ<Âú 7gMW`)ÍÈ+³Ak_N­é¶è+xÌ÷5øa¶e’†cü¸D<ÃßøÃ¥~à¿Õ.å—ú‡x~éÒ~3(ÏÕë/é߈֥ç/}ò ÿ­OÍõþ«?$ëS$Ïx¾ø’<œÿGy¥/³4êéOáu‡­‡®¥‰}Ò¡Ô“û—v¸l¾¦È‰ç—vöµº¿”ÿEéÓíÕÔb½ 볬<Â7s¯ßì*)Ïâ.ƃ¢ÙƒT1r:°ÿÙÇÔ뒴ر*¼ÏÛ\ÏÙ÷ÌߣP/Y~;B¨EU‹×¬~´ë¾‹ç½Mºü'¦cx€1èä°ùf9`þfk`ðy'°Ö·Åê·*Ç8wTI&p¥'¦ÒPEâý7øö[qêyƒ@èÝ#®f L/<éÖù–&æI\°qžÄì²ku ÉÇdÓi§ R—>®"%M‚ìï6IÞÈ"X1$Tšåÿ ”ºOa# ê¢éîP\êÖÓœLÿš^ ±EnL@Ü¢/­ùrè<PP{ûô•Ú„‹ž5Á{RÉjv€‡ÆûKN5ƒq(kCM’˜roê¹á©fílwg`XTW¾0ë«w‘ÆË2ùÁ–`ßb@’{—²>©uà8G_vƒ·œCŸí—+’—_›£ÆU ÁílmØ£ <¶ô=ú© ^<*ø§ÙÁ'ë„Þõà×_zX6’Vïª5,Ôƒ@‰ZÀ V°ë»ojÆÊå †]8 ü.á!‘á]ØÏ^ûjüpñH_¡ú‹¿å× ûr)ŸˆnÇýBòŠäÉâÉ‘ˆêð¾Ú(¯lÆó‚‹5þýÅûj³•)òqDâ~àÇkžeý†°5oIÖÝ:øýÈ/Äûª‘¿÷(ïÕI|V¬‹uÀ¾v}³ÁÔs~uåw9´]ÏSCýÖ~l3×Ïk;Œóê?ÏÑí¼6‰óˆCÝÏ“•cb#âFŒ‡&‚±Ñg|•neç°q‡”îèºlZΚ…O"›yô-ÞSŽR/qX§öIÙV Œœ .´êì’É¥fªá'g4;ÐÇÓUÞ‹ºZ‘¨-À(ÃÉš°R_¿êtÕæ[¢À~—›‡Ø¸8-4¢Ò¿Õ7ÛI³sM©ÀÝ?üË)ÞþfãoÀ·:¼­Çl5ðoѸT— VÐÄì(¡¬X5åˆÌ´nv=ñD{Ö )xñ‡2.g‘RKÿÑuøH»VøxyXƒL§Ñ!!zc£ªÙ`üäÕêÄce´AQóÆi…í ’;D“¦óÔÖäTÞµóþ|»ÒAãõ¡³Þ* þ¡9 V³ˆ5·L†²@‡Zäõ[…&нdãâ:Uú«WWô´ÍË›rК´*À8ZÕ".ÙÍH±ÛSq0fNÛ¿"&LDˆ²GÍÁTÛ˜æõžW`æ•R½ ÌÚðyP•{õØ ~ƒMb—ƒŠXÛ3¿¤3,›³[ ŠŽÎ€Ý²Ê¨×½`¾{“¢¸%8êÓEk Ó¶ÇY{rœ½ŒBôèÀÅrOT²a0¸šÑX2Жƒ›ýxþ½ «Á=Ø÷éÅ1ðHx·9 v™³ú[Ùo5ãuƒ=A°·1¾.Q ëóÝèg!ÿ¯ó[šWƒVdŸþÍODr¹TN—æ}#9Er‹ì`¤ßˆVäâuódù³Há}MHÖc(_·WÉv/²cQ œG$Û£xßÄ%ñÎVmóÕï÷QòÈëRoô‡€Ø—ÎJÍ;ðúˆïªéw ÐàòŠ×9‘óÃq~"°¸ö‘\…?le¸AÏEé—˜i–$:¼ᅥúéO6hräÈ-Çû&’ø5h´“ngSHgl`– ˆÜ¹¹³C  ÝNåwä¸/è>ý7u<ôëîWæD°‚ÁÌŠ‚f N›–Î7S`Ô!$·“©LT²Ý>0îÓ#êÉ! Àœ‰æÎœ‹1X/ÝÔØV+Ãè Ý.ötiÛH°%}´¿ŸÜC§7'Ùÿ7Eªàì­­ZÜ3.ìsŸØÍìQVVpœv‰õx l£¦–¯×¥‡Få©!(¶êzÂcfX~õíGâ+ «ÔxRÂdŽ5>싎Š.Áõ’‹¥úo©Üý-¯4suã¡*Ü>Eç!yFò†ä ÙH/þ­0îסùäu)Ÿˆä ù…(>³´OÍó/œC’‹ÊÞþY‡ˆê–PýÊK³ÉXyæ–"¥OêkAî‡x‡¯‚‚kW%¦Rxýt=®L÷ ¨ªâÆ›ÐügÞ·òï†\ÉZJÿÒG7y0ýÂF¥0Å• >ØXA6ËÃò¯Ø„·ž§£-6eò‘éÏ:Äã—¹ùCµ;õsƼ$ •MÝ,’X t“ÆkuÂ)õÀ/Ù¤¨¬O¥¸)zõ½@}¢„%7ù •àuü(ÏÕeà}/GL./—‰ƒ3ò[ª¥AJ=‰Úý‚!Èp ¿Ã  +T>ìùšà ò'|x~8Öáuªì\bV N[ù.[A4ò ýKÞ€ÖŽ>›3¿ÜA{{WkFÿ,èæ«Ü_uVRGïÌ1=†ÇÞ®LIJ#µh瘓`|n‡·€˜zf 3=Äûx[Òf3.Ëý ÖÜ,õií-`kÞ÷ÚK{ìs2=ÙhN<V~}: Îá»H $Ê|‹Ã+§](§¬¹æ­6५yé™b!øX_f»›%~)Z—­Oÿ‰íï§B@‡–zb¬1,f¾ k_­¯Óè…uE±Ñavu v]H—֋蔩 Èa e#%±¸ý‰äåoüáßüA$6¸éD&àÇè<$(/Åkîù¥ñqžpÉŠç•þ-¿”loãþ!ª?D}×既~pKòKQ}Ò¯xÝ!òQŸ)”—³4¾Ã‘wOÝû¥¿âÿ‘ˆxÔÿ Õoû„ãõÃh>M@Pâ‹“ÿžOGæ§®i_àtÃffn&M®¤Ôiã9ˆ¡)tl,ØètÚŽüÜcŽË+3÷M^çJE`Mþ²½ÚÜ8üX¶Ë^;ƒ÷ßàÑIîØ¼ºï»! §Ä¶Ž¤ B»x}½³ôð|ðåÇØ¾/‰“»;;Ì‚TR µÓ³4yàr磵 È]}é{9u/HÆŸX*ê¿Tóï[ƒZi¶¼›è-ÐHwû4£¤ Zi„,&ë1Ð.­  œÙz+Øò¸M¸aåË›)Ùro»Áû²ÖŒ £®˜\yÄT—Sf©ìë*7‹n·wÍÿîX½Èv›÷u²“Úw‚ý›¾r–-[ÁiMaõ±6!pžÞ¼¥°Û\‹~½ k‰÷üΊ¸ÎàY¾{õø°ëèhæ¿ãι-°Ú]7ñ¯œy5 k¾ k ƒ]ë*Ø'€Ê3÷®(Å ÄU<¬Ø~ëyF•ãÌ^”P{YLü¤. ê“®ƒÙ*“ Éw€–HÅ ZJo}cVmb>íØ½ÜÐÑ2ýöxØt??u>‡¹€þ¥¢woSíÁ õ`d»T(`‡ÅÄ"›ÁÈ»zèGËCœ¿7¹ã^:5±L§MOÎÖrƒ¹û^ӂǯÀâéçA]Z°²«}ž|¬;KDoÓO‚mæ@â·.e°ß`YÅ Ž«ÏèÞxO©×¯ïVh oW‚ÕÅ×?‚›ñ­ï"+UÀý¤a“ø¦ð¤ y–cž^Ûó¥oõƒ|ß1›9ØU²…·Ub(¾ÖIÝóD6¤\:ÿeë‰H»i(Þ…û‡Hï ü3d"?ïoóŸ–®KóL—λ@rpp½HŽáy¦ä:Ä¥}ÕðìÌ ðþϨìüæýÃ6uñÇDì×{³¶U;¶`£ ùm]·Š°±Õךº²ð~ú“jâÙ'å÷à}‰gÞÝÀxÓ›“àîm]Mé7UÿÔï˜W"нÉöuáÚz™ÛE/>i÷`H« 02?ô¸œŸçÄbœ8Nh6]¡<®¢À±÷tï®éÀÅeKe›U ÜÑt=¯‹ϯ¨0Òµ"àË!| /SØc¾¿NÚÐAÇ5Aç+@ä¡À • :ù¨­ôº+ˆ+íq Ôo‰­ï–û¶€d›cèáÇ>”>mœ+&ï¶uÊñ©s]³^  ¼YØ9õ8(=ýi*‘£ª+„¿-3µö€•ŠYAcÞc¶2WÍÇÝL¥f@ôÚ 4—öt¸íÒ:wp‚îMo[–è÷/gúÄ/†\¶Þ^FÇüºs¾Æ£NI»&úÒ†Œ¾J`z-r€‹Ì%m˜8>;ƒÅc ®3`ÿ®úí× ° ب͟÷ìÖ=uÛ¿á8dÖ6®Î^NQ+Û7©‚ó#Ñ©àÊsR4{ö¸øzÕ;¿<–U^Þ·<ëv3݇ Õ"Å‚ôA+£R¿÷ò0Ð>hfØ8ÿy7 Ÿt9t—?d/.íÿ´Tßý›ˆVdÏ"½ˆâ@H¾Ð¼3TÇ‹úÓüÑ߉œúÇüCr="žGºdÎÒ³¸ˆúÖ ?å}#ûõ«As/Èsñ<òœ‹?æÑõ0žŸ†úG¡ú%”ŸVvÑ „eöýâ¹Ïá{Îb=ÙÉñ¥‚5øü™ÖÆè†«úþð@¤ÌݲPìWM´‘€a3e®”Á7ŠR©ØøåðóÕóvø¤»ê¹þ•âØtä¶£wËg±ßôÝB»âé€À®Mÿ,¨GÜÓ_Á4 9Œvúþ@‡ÉŒ+ôXý¯ ·ç,Ç€ñÕÆ·1`žÛËè^üض‰üèØv8…§‚@~¸.èâýj1Ç$>ç‘ot\S€ÍœW‰Û~Œ½×ÿÎX&qžŸdO€™ò2Ÿäô÷`þ›y¨°=,Ã>­ìzÂx‰U!ا%¯8yÇTÜ•ëÁ鸩ü¥¾fp¡?ãÐ> ®é6OÚÂÆÁ]«C˜öz-xÂÅžÛ§€×¯â9AÑ+à˱ªgí~gXå}·s?§¬n:ötj+š>ó;ÚxÖ}P¼ú ‚Ü7nMè1ûÜݽì¹@—võêÛ°Æ;Êê…Ó¸¶´?þßüÂÿ©ˆä=ÿõ‡¨®ðoó,þ¶þ—s.þVø_Ï•Bù9¨þÕõ/ѧxßÒ%ù¥xý0êKì)BøƒGÁût<¢?oI‹}%ž°ÌxýŽÒœœŸÓ#®hÊñéÏ|:Ô·õ©!×kà}¥^>·5zÜŠÍð]î*±ëÀ~{ ÜuïFúôösÓÊ¿ai‹ã ½¾ìùlƒ¥ººÀùMZ*øç)à>¦S&Äï»Á¯™ZÓ~á †³Ü_ôïë†ê§$h8ŒÖòQôã·â“Nâ̰B»sÈÿ’m ¼ JÏz¾zß} ªÝ…\m‘ ~}åÉ[’¾ yĽ?ìçw ^ʱå§×Á¥ÃvT I z ÕrïëÕ·€‘ áCÖä0.Þ6åÃl¦ :+BN‚ù–3*{ ,STK_Ù Ö/¥Ÿ[83‚þÐ\i?8ÒlÑ0>Ê΄í«B[Ÿ‚K¾N­îpÛâÒ!]rC•†/aÕ2m=PçÚs+ F9EÒ fó H˜ûzP¬Ÿ»]w„¶ÔG¾(ÁzïÔà,ŒÆý5|Þ™_Gò‚ìÈ¥}ôÿÖoéK¤WˆòÕ]‰â(îÏñ$÷©AszQ^^ï@îSøD÷@ygÿæ/âþ!ꇊúœîmþÄDÉ·v˜Þœµ×kô½ˆÛò>_û!rÏaºN¯Ÿè§þIJÉ¿f?tªcù¥¿/y®/²/§÷Ú8ÙžGz Ÿ‹H¥o,fyB¨÷ÞP¯‹šâݤ­.ãòA¿ãйá;ï(ö%™`ɳmŸüîl%Ÿx䣀“:mõ‹€àÒT £gÒÁûþò$½Ù4BHÞ«/cYzM¯Cålvœ&¥OÍ+ÚŠ­W:@øì1÷“ ÷`YºP…lƈe2ç¬ø` âR¿í¸/ó¶Ê¿u)×­’c CõKfd¢d«DÏ>¹”ˆíAåw@3,àPÅ>Á[Í |CRˆýóP=ìÿCvÞO|"~®yÓP¯¿¨¼4~œÚ‘V(†÷…">¬~U`:6å%:ÓAwÄV¡r”ô¨\!yƒÁq'1çÏ»ËÊØ¬µŒ¼~Ù© ²ñŽ‚»sW&Áäò§Žþ¢*0íñäÑÛyÌ =v‹Ó€E>SIÜË'`%†×„€õõ4–˜þó`Hë¤n"öŠ6òòÝàÈS°Iuw'8¥nT «ç<¹8à“àòb éÙÉ8p£ãTxdÏ în†úqUI”>Ä÷!ô—“ æŒõ»âþ@PMi¬iáž÷¿­ÕÊïŸÁzâ:ïvmÁý/d7.íü·øÌR>ñßêQ<É7Êû^:÷ ù‡H¡9¾ÿVó‡ä>5¨ñynKø¼Ï>Š{"?ñä¸'Ê;Gý0ˆ÷áFqÎ¥}ôï?%¡8`tëýÝÇõTÿäŸ}ɼ?ÊI~Ï-CÂFÛme6G›cãj?N›ôÄ&ò\° *lÊúÝ`Û6#Ê£œ/…ýV¢[Nou#mö‰. ÔQ„œÖY 4_wÞ2¸tk#¾ÜNNúïé¥DƒR`ÜE¯°ì¢30kùI(i¹«ØP(ÁØí‹ÔØV+gù¦6žÀuŸ£(ËÆï·ÏûÑön²ðgº´³æòPú)¢þlüoÔ´«ñ9ˆ"Ò³®Ûƒä±s…ì‡m@šxN^l2 dFCV¥<…oîôöÝ~òoO¸–åÅóôDê¼þUñë×ÍÔqÅmWú^ÁƧž5mdSáµ1f'Áì°ÐƤÀ°X¦˜¡e=–õ,¿˜ß%uÞÍ=|ôA`{òc ##Øß NäN=ŽS"¼°# œ‰¢•ŽDpÙõ’*H\Óu×Y±€{”¯x ¾ò-•—!h© j@`›ÿ§ÿ˜ÿ§&ÿ/ÜOóÏý<äû9Éç0ýs¬¦¬„öêÊ:hO§£¬¡‹4t•5µ‰èJm4ÐFm´Ð†ˆ6Úh£ƒ6ºh£GÞP««á;u|§ï4ñ¾#â;m|§ƒïtñŽ¡chà8†Ž¡chà8†Ž¡chàš8†&Ž¡‰châš8†&Ž¡‰châš8†&Ž¡…cháZ8†Ž¡…cháZ8†Ž¡…cháDƒˆcq "ŽAÄ1ˆ8Ç âDƒˆchãÚ8†6Ž¡chãÚ8†6Ž¡chãÚ8†Ž¡ƒcèà:8†Ž¡ƒcèà:8†Ž¡ƒcèâº8†.Ž¡‹cèâº8†.Ž¡‹cèâº8†Ž¡‡cèáz8†Ž¡‡cèáz8†Ž¡‡0hÔÕÔ([uÊVƒ²Õ¤lµ(["e«MÙêP¶º”í?hÿüüP“o¢Ý¸ayO¼Ž|´ø º¡ë·¡SèÖl\N>Ã=‡º.ÿìÕí5í5íµ퉋öÚ‹ö:‹öº‹öz”=q.q.q.q.q.q.q.q.q.q®ö"\íE¸Ú‹pµáj/ÂÕ^„«½W{®ö"\íE¸:‹puáê,ÂÕY„«³Wg®Î"\E¸:‹puáê.ÂÕ]„«»Ww®î"\ÝE¸º‹puáê.ÂÕ]„«·Wo®Þ"\½E¸z‹põáê-ÂÕ[„«·—DZ 5\ÿÙ«/Úk,Úk.Úk-¨yÚÔü?²AK¾&0ÉÍÖu;Ñ#"×¢›i£×F.<šîŸG³ÿs›aá†þ÷Õˆ“|ÇΧ_lTüó¸Â’Ç-#¯äu{ãšD^,_X“s,¬¹äÛó¦öüg½D¾ÿêÍ…ãëäó ×.ß&Ÿ'í?/Ýè.ùq÷ƒŸüg}@~ü#ã…û‹ÈÏS"²pÿcòó•Ž/Ü_N~Þ'¡ÿY+ÈÏ_Ñtþ?k%§Ê¤ca%ã=½%þŸµšŒûLtáñÏÈø5_K~µ N~=/Ö-<þ%ùu½¬_xü+òë{ M~u×ÿ†üzß²q.¬ä×ý¶êÖÖwä×ÿ.Æia%¿zµ¡…•ü~ê¿gügm ¿¯†³j +ùý5º½]XÉﳉeËÂJ~¿M• øïÉïûýÎü÷ä÷ÿAuÿù:|èZÀo&_æœüfòuiq]Ào!_ŸV¦üVòuj}²€ßF¾^m;ðÛÈ×­]y¿|ýÚ¿-à$_ǧð?’¯ç'çüOäëú™qÿ3ùú~.[Àï _çýG +ùzwg\XÉ×½cØë?ëòõÿbwea%_.O.¬äÏ£“Æja%.þ'VòçÓYôca%N_ùtVòçõuK aáoŽüÿS`L"}ÝA"}ë"‘º]H¤ïå$R‰Ô{šDúÉ@"õo#‘¾HCŽ$ÒðciDžD=N"Ó’H$Òä'iÚ–Dš)"‘~Ë’HsG ¨6Í«é¶yñŸÍt÷çÝ)qþu1ý&XÂÖæùŸsãÀ5¸Óç.¦ ¾þ !hB "ó¿Ëæ? Ñ aù¼“!þŽ@ĩ낌 ›D ¬!ä …:Ai% <Uù µ}‚úüµÒ\E h½$þÑ\: =nAþ3[ÙO úXÍÂoÃüo- DZ8Vþç7l†|­è •Ž©öªä½Ú¢ÛÔí5í5íµ퉋öÚä½2ºm†¼þóÛI…~3}vø}èwséíyÌùÚtþ?½6ÝE¸zÿõk0"þ¿â,¹ýÿ‡ñÿÑûù Îÿáý,ÁÓ_„±’ô¿ß‡dÃ`Ñ9†‘“ÿ^(öÅÒ×òßWÿÏ1· ëÿ‹¹7çªïÚ‡Ís…L™2•™ˆµÌóGþõ^àß'YN"%xçfBºî/ÎVŸÿnø»¸ý÷f•ßÉëßÉìßÉí­ø]dñÏah¿“}¿›*þ~ùÿ{3ËŸ"¡75ûoM¯ÿ]”ð»xð÷¤ßÃ8o"ý4üçpß6cúçp¦ßIóßņ§ýn²ö{SÚ¿›þ³)ÓGU(=®Ïó§hþSäÑ%ûº—šmæð±+Ž®ƒ¯Š"]3Zš§}w˜åϤøöwñÃW5ºÏ¬ð§éÌjXÜâ­è¿‡}>ocŒ1F’xŸ©Å¾${Yjþ€)^0çEÆ+"Õµ+k/ÆÜ°ÏËxéÄöy*”+âv(äXŒLAz›Mñ{»?þÝŒ)Šb4_F™#’n*V0"ë«ÊHæ7Ü5¤­CC_éògk@ä*щeÜ]r³ˆ4µy¦Yà øþ ÿäÿ=_ ª†BÛóÝ_£ˆ+± y”9еd<3ãcŽJÞ»¬/£D♳æ.ïPjû¡{}_ãPúâjõ§Œf”Ýí=5Œr#|Z§#_¢Âó”Së.¨xb—Iº#ˆã>ûì»*‚‘ÆTAÕ¨®áÀ©~T' ¹NôD5\P(LN5…?YÅÉ¡¶¿XùÈ`,êâs2™ðuÔgípcÔsFƒáj¹zÿD4r¿C6¶ûÉG—ÜòÑ”…ÈR,#M‹n‡ÌÜf&Q-ÍÐ샤ÓQ4?Ö?è‡æ£Û9¨lÐÂ"¼é ýv´¨p¾r§-yYR£_¡¥?»˜ù­x´loÝH=è‚VœQí®I&huŒ8,{‘¬–-JÛôOséßÅw¿‹hožþ›ßMAÿ§äû?‡×ÿ³YÅo?¿ñúÏ!¿“Œ¿‡ÑüNêýÞœú{“Êû?çýÛþñ÷ï¡§ÿ-ø»™öï¡ÿnJú{³èÿÔ|ÿÏæ–ãõO“üºÂå·EŸÿ^'~¯ùÝœøßÍ¥þ4µp¿è½ãùŸbø‰°^wK}s˜”Í~Ñ’WówÒþ÷°Š3bÛúwÈÁ7*š[W˜ÿºÙé™åßÍðãu€úÖ9 ø™^fœÖ8+%´Vo?ünf›7è6ž¹ƒD‘ßÜ%¹" Å»™2Î$}¸\ýí¢8’»MEý:ø)3¯œ ëÈFO<™…ÛžïV¯ÞõéC8õ(‘ÑJ¶!—¢ ™ìuÒN¿ÖB–¬:Å ™†r*0×шLàn‹&xhy] *X6 ÿaAɲniŒÐLÔÊ Búë>éN(Öwoöaa:îí /yü%/FÓxDüBé=©B”Ô/QÖövQz½)ÊNÛžìs#×(V¾Be½¢Ónû®gÅÜ´GÕ'>I¯âP]a|a†/5~õ˜Ü?ƒZoE>¼”®BûÊÚÊR ¨gÊbëpÿ2ê÷ñ¿L’O@C¹‡¤²®häë’%ÃEƒÆiǯ76¡ÉÛÏ[ó~6hêîCóœ›M?ïµ _«A³£–É#hÖ.°mŸF*šËŽ*Íç£ydO°Þc4Ÿ¹Jž¹‚Šjú= hqÍW-Ð iÏ-‘´ž–€fóÙ¶çûf –õÓýÂY)¨r=¯Ò¦¡¥ÛTêÖ2¡(ÍK½°1 u¹Ë÷%A™—šûÖÈóí4æcƒœº®—cÒIõ¡%<Ï ^Êñ·±«pBÎV / 5×¶qÝû2ÕCm™Ð(ôØê^´ ­LÌy¾€ÎŠ€WRKÐ]ÝÔ2!óz•ª6×ò ¯s¿VËÉ2È·ááS^‡A'q Ž0púynØô6èÛÞ{Ôºùª” ²þæ7笚„$fhñ6«‘^[V¹è[ÍhëÌñ°d«„ù™µGF¿ ã‹øÝ¹~uG÷u@çâ%‡¡u[èJ§fÙ` ÝÓô|çÝz3ŸÌ]áfÕŒؚâLn²Þî?‚Ád­¢¾é`­cÙçÌ’OÀÄ”U;«DLyGgé§Àì~é ør@1kõ¨ÌËVO,!, ‰¤ ÎÀâó]Ƶíaù칆°røÜ`MKL +›¼»ã¥caƒxtï ¢a$Rä~§[ÿI¸G_îŒÏG2f“§üÃÏ‚èâ´Ò¤ìe.2Bê”õQ–Htón!Ý~Ü;¿UC¥ôÔ¾ñ '2yÇÜ;ý w"à ëñÈ,›~DyÛd5ÚÒš¢‰Å]îSÅj¯ ÇYoAF9ä:ª«êŸƒ»5¥+Ñ!Oéd\˜ò®VIøŒµ%Ø¡\xʹOÚPÁ€»Üu•wÌkѪÅáú¤§ü¥¨ò™,–|ÒUMSj¥Gµº&±‘·Y¨!دpb5æ]‰âÉFmezú¦LÔ%ÊؽG õ9âÓ@ƒxËmœžoÐHæ3Q''ð^(×@“µî»3'Þ£éówdº›Uh¦l¶ÃæÈ&šµ^KÕß5ŒæVNl•K¢hÞýåq¦»Zèeû|- K«òíÑ’# ð‡cZ^¸t¤­S -Û¨úƒ¹C üÁÂq ­œCžN!éúìòÞØ/Pó‚ƒíXÕK¨<ý•öÛ"‰P]/ÈÚ€Üð¦\ßãÇ —UÄNæš>då˜zóL°CÆ×ðq¥Óó¦i™'œä ©Í·ÞìÿIQO'6Ï;BbšSïwH¾2˜¶íu!¤÷¨ÎÇö¥ÿÁ_Ïë½ ×X ¤ëÔ¨)1—/-ç}„rïÎqº1(ç®»Cek'¶¥g±@Y;Mu)”öª^?YÄðÇ–q 6Î7BÙg6†AYU´Z˜ûxötw˜ê%(czÞD^*¿öŸ‰‚jºý®bPq2?qCíÈÆ¢Õ¾ ¨ëÔ0RòVƒúŸ×&GÅ2àÍR¤åÍ9BÜë%Ù{š¼å=\ÂöA«hg‰tl+xè©é‰ýÒ Ûú͸ ûMèÕ­Îkí…~7g·@VE|s0Âìs% Û¹ –º=„ÑHq»M"øH÷Ò·ë~Œ…ðv>/ ì*'~õOåò…âÆF0õй?ÓæÌDÉ1¸ÿЂÏÃ[ÚË®ýðõ̹Ú*ÌÇ¿‹&¥ýßEÌŽ«ÃâÛŠ€HqSXöe<{œ~ªöGWÛÊÆ3ô°ª³õÒÃ6Ö·³^ ïD"ýñ„Ñ2{$.¦*îXƒ¤&Ɉo“#9×›bf‹R¤4ï9qæ3R¯“ïLO­Âm´K†!cH/š*ú™w¦å¶»†záÎ^<†ÖåȲ[Ž‘•$Ù²ï%^Eÿˆ§äBcÈ}•S‘æò,t=—Ñ F¾aE—’ëwQŒ©+|ˆ…ÏÅ]~·ÅÄ9ÏŠàÞýÛl/£äέ;ƒPªrèi§ÊŒŸlkßE…ò®­¦. î¨H¼A.½GØ.‹D« Ê ·ëÖ]T5þµàýÕ>óçl¢Fz4™ —4j]àÑ}T· uT¥/‘E¢îÆ5j«¨¼%îs f ¤^[pè]pu(ݯþ QG“(…–@…*4µR¥â:Цý7Âϱ£™Ù~b§{ѬÆÿ4”T£¹øMA¦šu4¥r2wÍ'O|9iPIà7ZA∟h´^a°-iô޲ɫAÓÑXûÎ7Ç ¸cç \¢„*jK‰µ)A(ñ1(¾ßJÐmkŠ®ÓžPðÆ3¸íäG+ÉÜ3†Wé¹”¡ãÃääÖí¯•a×¥›ª!ƒÔ’±•Î rn3›iê@±º³YTßNÍÀ„ÐPØ=/ä­×„w6 ÖA§åuŠÒoãЭ·s_–]!ô¼æÝ`hg‚>ÃÜ=Ur0ÀÍC–ò¶ ·H;Èé ñsoÇCèͻھeIÐ[öd½ü›ß¥/0–Ü…–7eÇ h=p¡øÊê4´Åæ;¥mS„öŸŸ&XöAÇÓKÄnùÐ)è¼Yd2Ïé²ÙÎÍB×aÿC§öBwôÓÁ®æèÕ»t‹X‹ú# Fš#Š`Hþ‰I$¼—qÚ£Øcgã|ÆØï¿¹v&”õÒ:×Àdæäu»ø0c¹p¬­ð#Ì©è¿çÌ‚oGó¨XaÁÒU,®3¹—¼ š‡`iùù‡ÄÕð“òÓ°×Á)Xñ¾M­?˜ë¶WT_„­g#:6æHgè)s"I&>•ñ)!¹Í.«ÞëH)œcQÿΩւ³ý«‘æÃd¶ì!¤[ÅÞcÔÃ{²‘A¦k’òb2Îr|[Õ¤E¦i°wF–X)6oC¶­ wsþËȾáÔ£r&9gµ?F¸#wÛPý™›FÈã=¯:Ÿ%†¼å™NFJ7߹槵 :~8îÂ1óÜkGˆPôýGëß~â9WŽÈ Üw¿‚<Õ%ƒawm8J}>qôª²Ê„0rSÚ¢œ’XruÒ<Ê/=ØÞ}>÷5l³F¥÷þGV2¨V¯6~Û U…çÏ-½ÒCµÛº*~)¢úäg¯Ø,Ô´Á]]‰¢¨µ¶|¬ ¿uª ~Ð4ßF½Ç?ŠÖøÑàèä„·ó4®:Itž@ãÓ*ô!üÑhÒœâà牦á"WSS¢™è¶Æm'3Ðìu·Éîƒh®=§9ê:€æ ­lï;Ž£¶·5ô[6]ñà(Z2OØY“U£¥Ïð}–:´lÑÈušL'ðÛ-—¤HÏÇvÿQ$•/W0_¼ 5¤‡­ÖQA…2ÙÐâÙM(<Õ¦Ô  ¹û²G¨.†BNɶúóœo!‹_4í¹ %du•0/¼/Ú ©Ò†BÊ[ vÿ´`H¢¼Ä¬f‰œª:S7z!é{¡Ë@:ÛŽþæŽ?øË×F»nvŠ7S¿qöItØÍÕ¶K_àu” ;”¼v1›ª°·…²‹ët¬ÕiP¦?kH#Y¥w^<Êûjû·M‘VzèÃe2.&ØuÊÈzwƒÕ£ùàuí…†•º‚ŽS’IDˆ“–÷¿k‡ê£œ»tsP˸#¯Kê¨eߺυB½ÄÙ® Éhx‚½› ´Ðði¢þLêxÛ:²ÄéAˆOíçT7¥ ÝVÈ ëC§òÒWwІ®Ù¾ŽžQèéξv4ú>W*¯„^‚Áýóõyµsð®ÌðÕª-|¸P9ú%hŒ¾ñ¦g=c<' ¿ÕœñŒd×'‘1ðI«¼öÆôL13;:ŒÅà åÖpøø¬ß¸³Ê¾ô×Þ †y³L‰¾°Ðwúü5›w°è íÓ×µa™E@¸ð<üè[îV…_­§nH hÀu1e‘lœÙ!iË„D$ýƒ÷¬«‘ØÇ9æIÒ$¥°rßùlÉ:’ßI›F!Å/ÑŸü:´y­ôDÒ{–Ü¿8‹;;3Ç¿uG†KrVÜTì¸Ó1öØ¢2ßMž-°aD6:º™Û9^È^C{žH`¹²<$nßEÛ9 y ä;dô¸Hæ3 ø~"î!Í@¡aÖÊ´«(zeÔ8K<÷X?=ʾwJØ÷0rìáA)‹íÌO4v¡Ìi­O‹îÕ(7¾)å…û=7Úš҄›ôx•:ªßÞzåSƒª$@WùÕâ·³£+jØ[•p´£–0ë«P+;Ôþ"ü¬¤ÿ ê¦)1”dD}¹¡t±ö„8³§4KS ¿]}Ýñ`%žº…–¡É‘ëv,‹ ©Ê¦ÎÎ4­ýT%óõ,š©qíÖØƒf=­&ïМOvÇþV-4¹»|ãíU47mÒš A‹ý=[ÁFïÑ"ôšuÀãq¤æŠÛÿÔÞz}&9<5û³õ7†¡ò ³¦‹ ¿³})¾Ë‘ Û‚$wQ‚ü©¨c=7H"XðØ^‘øDT}É„¬† ³á¿ ý¼´âÀØHë ²¾òî÷E®zCQÕjÅyŽz¨V íì; Ú?úŒœ…–÷–êiC¡c6Ì.)³ º:E\¸š=*×Ö„gŸCo£cK6ô‡N͈|]ùw¾>€~ç+Lø ×FžÖé@>t5¹ŸàÉ!û›ßÚRw_k§–ÎkõEõôЪv¨)%w´¹Üü«ÉèˆÓÐa|)-L»::Èãå<A§ñÓM«o— ‹þÛ³bèÖ¿'!N=Ÿ ^¦ùýtâTŸƒ2`0øQ‚’ÎO¶~àìR#WÀ2aácð«'Ì®Áø=• pò6ø4­Ø.{-äLÌî¢Eª‡ú– ZÒȽA˳* öѲéò-Õ¬¿…äÔR'¢•ë0k"ý$’xÞ½xA_ªÄgßk8å4éÐLÐWÙQá•#ÓM÷(<rì%‹‹t!3†d”÷Ó"A©]rÛ’‡¶¬Gö0ÙCŠ2­œ »$úÙh‰•@Â]®žÌ(„$—û¦áo íЋôªŸð—¯>2{ç‹oüJ;Bà)å‹©+P–?ÈêÌ1Fà'†Ôb±çPú+‚&¨WJçšöPÇçA©.}ö¦àó¿­äÍ—ã?/ø.ó›ê×HÂÏYS¸éµ¡ôm¨ó‘‚®«z¾+É"Êéç¤Ü¡28Ä™ý2T³QñyÊ–@&/Ýü9s¨UÒŠwi…:WÇJvù+PZbíeC rÖvÔS·à-× Ùµœlh~=ÞÆ A m1¯ÆÜ€ŽÞ9r‡®Ð0Ya;Ò=Žý¦Š!Ðwš}‡×Ï×0qs9–ÞqJ5YßÜï¾ÄÒpè£‡ÑøXšÓxm{&Œ«lsAùb˜hÓ옾~ &ïÕÏ0ºwÃô¥¦Ô §|˜-N+‘ÞÅ _ðóÎÝÔ0O*¥1¾®M^oó†E–J>xw– ¤Î­5ÃÓôüS›ÎðËúI‡‘ó°Øìé´µëcôz ÒÖ°uî!Ïõi#$ùB;Æ$åäj‚ù¥Hv=hÓÄè4R<½|RöÌ(RïŒo@Ú‘ÖÇpûÇ‹ý'Mè‘a»Ü“ž°dkØN}'™‰Xj-2/"ë‰àO»N“!;eˆñß{ä|W%B~<y(>*ˆ #ïÐÑöü ]äÿ\ űPHíãè}å7(ÒX<ÉâOŽå¿ ŒÇ}÷ÒèK¥¸œT»íâÐŽÒ?y7X«.¢œ#s÷„ÁETÙ|ÄOîJ~žqŸuQ9o#ßpfÑ”Ù;nÕÌâ8¥[ÛQc;ËéQ§P³o§Ïä Ô~Ìeô+CumlOª_¶G½O…"_*¡å"‹XË4̸[·Ûê}/?5%ƒ&"í.ñ¯"ÐT¨Ö¥c MsÔ–(]ߢ™ä–û»eR4KÛ»×!y ÍÙÞ¯Ø÷ßFóÀÓ™iEÐ|t*_¨!ß↨ijÑâ&å|ŽR.xkP‡ÆË[¥—@uãufÊ\¨xÌ#’u„˜ »\ãk\•ºm›P“ˆ6ä+ìfbݶyœM¬ƒS W<ý,.@¦+i«;¯¤íy7yå!¼ˆÏ•NÙ*—}©…@ÞÛ>-UK™ÎÎ4~ð¦_%“ê ´ì¿z6ì.á¹ ¼uçTtY'~Wqº+ÆW€b/ôj$üê=- } Ù­Ùãg¡?&ÅÅh€úF6¹â®eBÏú¬v½HtÅ©%FfJüá·6ÒD£ïÝFôDÇOô ´êÀp««´I¿60ÿ±Ú¹±V‡æ%´oUpkÍ@‡_ æEÒAǺ„Ñý“´ÐYç4É2Ð ]ó%º·nž€ž«J•«5ÐúLšög{x¼Tíáß#ÇÝðár÷RÔFŒÎw,ò§Þ…±I)mÏøÄw_)OϦîx¸Oôo¬ÚiEÏ&6ø¢ã!ä%óºR±‡á;{zÉ¡:;XÜ É±¯?èø¦ÔHrá×çáálX£ª­–š)€M>·žïTHô”ÿ—ת.’˜ì8?“yɨSBwѶ"yËhÆØWu¤Lª»îü©ïÇ}•\‘GÚ¢ç{O²ãv¹†×1»b‘>ò^Ðâ=]dT•ÜRµœ@&ž²!õÕBdÙ]qÔn\Ù*ùî}ûˆìÌý1b¬È±î¢NT4Ž\bÝïÜÛp÷ֽϢKÇ‘W#´acÞùú¾¾÷PDü„ømŽ<(ôf„-I8 E¾±îp/ïCqÉCB#I¸7ˆîÑÖ[b”øq²§Sò;JÙ÷s¶Ä£ô\Î ½(-CÁ\$‹ò:¡¹ég>á~JŸ8"JT’-S¹W,ŒÊuÚ·ãÞ…!žçäîX9‡j Ñ|VÆHxL“±¡FÿAc“\¨u©ŠúVï(êh4MÑp£7§Õá—c¨?EwŸ×h†É›ØKƒÆ¤>ª¬ÃçÑ䔡vòÓh4u¬õwþ•ˆf$¡Š¾W(Ðìñ:­˜!š‹=å5‘Góìpfxèƒ{;/ï@‹øÜ«A3hIój›á&ZžŽ[×xN8n,qá>ÿÀoW×>V¢ÕÉ Ž‚IŸ‘8ýhÙÓgµPUòqM¤©Ê¤/Þ`‚üE»{š?!ÇËOI,¾²Wr~$Þ…Lªa–„㣮ÃÑp]gRÇ/äe„ä¬)“™›µðEî\Ù$$¯õŽKƒÄOcþ2É¥ ¹þà//m¡1{ Èš‡®Ç?JJ¿ü ¤#‡Ò^#¦¤D'Ÿuºæþš…R_ª…év¥ÓQoφRâ otígþØ’ÒñÝ~P2óã®:M ”Ü81&E8^i(-{ÊxL^ûϦm ô)ýLqÀa¨:Tef Õç¥U°ùCÍEÖêΑPo’¡ |ê: &·oƒ7Ao¿& ‹B£·X-ƒæ7h¦pËþÚ ­=±i›¬Ðá1Îiqü&!þü‘1¢~º¿l‘qŸq€Þ¯>ç\7`€[PÒ×G†X9Sv Â{VCƈ=02t³›ŽÀGË_šLvŸa¬íœÓØÐLlþ Ãõ&·Ÿ?[÷“¦¾E{5=ü ³B—¿›ž­„¹TÿصÜÛðíd£pãÓX°‹Ö¤ƒïåè-#K–P§ºr~îq[/†Ÿ›_ßž|+·E*tkaÝŠÁ‰¥ñ1l~ ¼ý‰— ‰>~Ñðß«‰$Þ, QH&ïÌ뮯ƒêÝ'cVµ*Þ¶ypöÒêóQ}mý…ÛEî>’îÉGúçrc·¨ÝÑy!hZL™{Žb#bT±=’¸÷Íå¶ò¦(yñ¹„Jn2)3I¡ì°sâôT0Þ:ÆáE‚НnfÙ±¡ò®Ó'¦Dºi³Z+ž¡Ù°˜ý~T¯ìçK1¥BÍ`Š'ßeÔÖa\S¸œ„ºäõ»ö·¤ ^ô½ÔÜBV4 a8›j¹ ]Ÿ6{0ö¡QAv/rHKœù…¦;‚T8ŒÑ46Nxè#/šñŒó=LD³ØàÛžœÛÑ|GèxàÇ^4÷3pyû«Íßó\ßÜIà·ùá ih~ûÆXÁa¤¨×µ9ÛznçR Õ6EM# !Pþ~xuò0¥»o[@ÀßT ×%]iÈ IsÉ,„WŽ‘Ô›²Ýs“ÊT¦D^RðëLðœ*È@ ©§'F?ùA–¹òÓÚ8(¬v°ÓK„ª’ª? Þ¸ZXÓ^ô€æûYÏ\µØ¡ƒå>ÉÑ,èü0gI×~ºU5Zò–@O%ÇÈÞÚ$è;äûJÖSú6üXèªSÇ^jÉ… Ðó<îšó/èòb!K·û›ßöÇû;­lB«DÞ[å;ÂÐjÄûu‰ËZ7ª0uö´eËýôÚhDÅ7Å CP{T*:Òªb;Æb¡Ó·Oü”a7µ·¦fZ ‡‹ÝÎ7&úH™‰»GÁ€K\úþzúĨya Þ_>Ê:£‡¾|؈…±‚JDï&`Â~&Æ!N&3ö,œ™¾ 3—]ªœ2ãùšgð-DËm'©,< ¯z ‹y»‹aùŽšû ›øÙÝñƨ­V¿»)˜ÃÆžÚuÕŠq$â¼û:²V‰‹’*©‘Ô‰¦l5å;’ï¾Ð¿à[Š+vË”éHõem[¶®Ò2ßYûPˆÛbJ£éw"½Ú|ÏɧLÈðù@ÈøaSÜYW¥¥7‚Ì¥›Ì{Ä‘µÈ^k"Íw%0ü =†Á>^ªÍqÈåÑõŽáÈÜ}Ë-¾MÕy)|Vû Ô‘/‚l-l¥Žø¦¯ Ð‘¾è áÁ(â›ûY&é ŠeöŠ©e~Å=ë2â¼M(áÞw•|¥˜Ã;f£zP:~ëMn¸%ʪÉcÕA”[ "ež F…lƒãfq1¨8ääy{æ*;QÊø_{ŒH­X¬Y;‰ª½Mö×P]D¡Ú\¡5b>Nß{€Z’íÏ÷, Y¯ckš+êŽd›JW£~Z¼Û5S44áqÜ{ÚîöO¢ÉÝû³EÐÔ€~¥ö¹&š~Óø¾çFšÝ¸~Dâ šsÜ|«Vó ÍOÝ÷X¿„|²æhKÐoI§êõ=Ð’œmòÂ'ZºIVN]BË7K°Ž ~ ÐÞúù­N1½x±‰&d6j~:B³€veAGééöpâFžè&z>‚Þrp-{µHà%ÿí—nÃËKF|S­Vòcç“®†¨%ƒL!™·Ö-~Í S¨Gá¹ÏÀWÅ4FH40üEã×GÀá¥Ââ ?øËc M¥®|…×í9rLJ¡¸=Þצ‡J–«>‰tÙ=áŒiÂq1å‘ußIÞ˜£ xIà³ÑáÈ(©¿­Ö{~›"‹B±?iòò¹ „ælšÜ}fÂy ¯W<' ì`ݱ%MxýýÉÉC£*P «t.Téük Ý*K…ê\j ³±ƒš~x`©nuì£v¶uÛ¡¾Gù:o€†FÙøA h2Y¸—Û’ ­ŒÖ]G%¡ƒrHF{õ%t&Q6ºÈÂë;,9)‚l‘þD ÜÈ!ïÌö«¦¹FkÓ¯"ÏÀMÅ-bä붺›茂ÛÎó¥ÆAá€Þ‰¶C»QŒoÞþˆp5îY³ØöëÀ ”¤{á¿òH¥i†Ä]gQÖæˆõ±à`”¯ Ú»ó‡»Ï1«™ãËï¹®öØ#(1é 0EÕböýÇfŸ¡úÙý§h“PSfcÞ”Zµ–¨^œ7ÛB<É·ìó¨§uZèóêWJ´ò“£!F×ë]4ºð­£'EËoVîuD“ù'ZºrÐ4lõP–Ý4£3`e“fwß Œ$è5 ©Ö¤9n4÷嚥?æCTŽïj ü&}—>èZÜÝÉϬþÉqútÞÄ:EG Ø@ÕôœM‡=”ó>XpþR Eô×CH¾þK·í9Á ¯údŠ^ùAn’í:ß5:È~’ä0재^/¥o)%7‹x@Š2³àC½ýy|µŽõ2*—g·8Fâ¸<®0¯ÛPÿóꃓ³§¡éë»v­Uh·•|YÛ®Çe ú~¸CWuò²›Šô(EÕ‡°sCo=Æo0AŸ³Ñ›æ ‚®ë›ty9S =Úæå·HÙ¡KkŽý|jÐßüvúIz÷Ž Ðz¤Eó{´Z³w_‡ÖÆ:™õ>hƒÐgi[bÐVµ3J±Ð ÚãÛ‰^œ„k³¹íö{¡S\ϳ²*º zŠ4õ$ »xeiÉþ¯aD7ƒ àßÛŠsLJ«ž*}SX€á€-Ù‹Ò#w8dSÍácÄ.¦2ío0þ¿Ó.2§†ÏçVW8aZënteéUø,YXLÔ: _ýjÎ’µ‰ÂËQ¢S¯‡`‘ìØ5æ*yX¦wø¾#Ë~ê)g¯œ+‡•b þ„ yX?ärûå­ ØRøIìÂFŒÄ¨Ò¸áýYõVN*8µá®}†6qÙȱýzBàˆr®øé˜˜ânV K¯ðóÈóÖioVŠ8òéqHNË£•5“ºm þhc4\°G"Ã7«· QLÜdobîñ9´ï _îרë4Ô†’mÏŠS¡´…ñæ¬vÊÌ8~ò­/Ê=¾`yõ¤ *XóíK”ëAÅ@êÄ›ìx`«aÐN4!ÖÚ¬øŽ ª>æô:ðÊ Õ¦ÛGuN]Cƒ§¯µd2 æÚÍì£=$¨]?2X¢›ƒºÏÕ HŽl<}bç7ùØDäd»ß{¾²®j"Þv 2¾Þ¹žx¢Òöt8ä _޼z­IQñÁÝIày×·>ÄgS…~[yþP!Úü6ý0¤ ëŸÀg'ÿàïÕæàcwI  û¡X8ß M˜þŠå†¹‚¯œ×a:•íG°†ä-;–¡¤ëd åÜ Â±î÷áæ‹lq{雫W øYh»¦ú3(ÞyHy3„ŠÅ&›ŸM½†âböí‰[Ñ~Œã:Z‘ ei‚Ó[\PþÆc^ØF*ó3Ž¿*O"¬3«¥òÏ fÇÞëgu•¡ÖÙû]«=Ôóg³¤¸@WkʾÓgàí£ Þ×.B‹)ãjy¸>´«%Dãz ,!³”)ÐMe\úŽzÄBFt¡ŸÔöÉç0«5íà]Áù"…X=øpè`ÑŒÞ=°ñ¯á*ZGÄø¹aÜup…Ì‚>Qö_£õ„É~qjjòC0C"ê|›Ÿ>ûʽl鋯’—…ô ·Á|˜Nz˜ÀwÕ]jã\°iëØ|è,K8¯É|ƒ?Ä¿}^Þ¿~Ò?ò2ƒµ;ï÷.óÂF½Ù‰ÔÈ—HdrnÞ¾à dª:4¯‹¤±|®)+H~;ÀÍ~RN=Y9½ i¸ÝÉ_P"]agýÈî¸ã>2Ïì 'µ½’²àN³“Fû2³‘9`¾ÿuB)²öÿœãìCöc… ^ž³ÈeøURå;ŸÖq™ËGOG°b>í x«jä¶‹¡PWË óS=^˜³û’9î‘¡{ z%°h&ôÔ'”Š*T*K@™žkøùn¡ü^¢Ÿtb¸?0C´ÃÝ•& »8õî¢ÊsÎW7‰Põ”¶ÿEjTç· —hPE±e;•ÇwQëÙ¦U ê8Æ^<©ˆºËºÉûÏu¢¾±È³c h0R2D{» ”iunÇ¢qè«fãÙ·hR#xEfMϼ{µ»͈Òï0Ò£Ùuy£{Têh¶þ®¨ñÇq4÷ è£!ð\ß¼J±c`ùÀ€:¾°>粃. ŠÅ@›.è.Ü–ñÁó!ôJP^R¹1 ½Í¼¾7£ò¡×ÜîF>—!t¦øõe^ºvU[M'eÿÍo 5¾Cë퉖jk€V›éj7Ò õ~£ˆYw,´¶eÜ=tñ$´ñrnéxd@Û2½ú¡ÙHè ?`Ð[^2ÉšA犓dz‘Ðm9x÷,Ÿ3ô¬lež² €¾šûOØEoÃÀx¼ãe"xÕsÊð>|p¶ʰp‡ÑOtßÎå®ÂXgŒQ±ÎLL¤­†P§Ã”γÚ)éB˜´[µ&ør}<…ù(eþÅvø¶sö»T,E^öõ%; ?*^yîS=Êoþú±ÔÊˆŠ‚9}!†QxàiØ…Ž* ŸÏ{lšPõˆ#;ûèTKsÝRŽW@ æ.Z‰[¨YæIbؽµCfÍ“¦~¡®S¯VŸB êï«ß%$ƒÃKŒ¶šhäÜ@éu¶‡ê6Žv£éÀï¡hÚX~,&~š8wö„Òk4ی罖®æ·Äî¶×Ý@‹í¾V{é ú-ìE¾N´/Z¬L>RR´EK§šGÅ^Ïвöä÷¢öƒ~ó~¯<¹­<Œ3×a3;—ŒCOʘ•½KKbõóSî²ðªâe\j>dåX—½:™õ e!ê7ÿÚWRÊ?“ /âÓ3Zxk åZl¾ì’æ_ûJžÑ ?‡çZ­wï»Ã3»(ÊúòIH¢ˆü)§vízñ¹#ÂÇʸ@þ»E׬P¸BrŽÆY ŠåÓèŽ+Žx‰>5þX-%Sä#ð™ÞKÊCIP²7—Z˜¤ú-¶;ÏHÖEkg´Ž¤@‘5ç=ë%(28ô•NÙ Š ÔïhôœoÿN¯•:äœHHÞ€×FãÝ3©cPÁÖhLú ªxæäOêOBµç™;ÆBP“EyXM’ê¼y_zM’ÂïݬåÄ9Ð8¶ee~Кo·ñs.:C›wD½Öt4 ùœqÓ‚®èóÚ7èg Ç+ØÖ¦¸úÎ-vÄ®·Á@|ÃÎ{oÙáY’{Ÿ ¼òº0ÔêŒò·6c |¼–y¨µAÆæR8}‰€‰ 0Úáè$˜Ô_î9 Óg6Ÿf\JÙFÙ‚“ìðåþ à-~]˜—­P¸‘Ô 9U§Ø ‹Rdzéóa©þªªÒ¼7ü´§~¼® ¿¼»Ì:øÙ`5‹éeÁ-gØSì£jöE" %í–•7Hü .ÚƒŒIMˆoÖªv#¹t—[Å£óH©žD½O©Ã&¸î1oGÚŽw]žãöüÓ§³Ô"ýgÏ™ëëw±zH°•”™†Nvë:"+×ÙWSì=¸ëõÜnbä<ºþF®û;îÎtѱ68‡¼¥Ë_ºo!ÿ sÙÖÐïà@‘ôÒѶqwÞ •×z„ûŽsr$›£w¤mÚ;”1M«ŒD¹Xâj²¢ T˜*s½õÀ•ô^‡VÔ  • ß0W6ªrÇXFSžEµÞ]ªK×ޣƽ¹:qñaÔ2Z"YüUˆ:Û©D9Š®£n²ÿZoꋺq~þ€O>ÚìCÃ-¡ &46Û|ðá…šÜÏ®˜öEÓ£¯j­ŒDÐtþŽFÛëChvþNæóËöh¶üü Qš»NÜÃz4ï`Ú=­m™e¦}Ð}·LÖôýèüsÏîùÐßüÖE™§tZ_o_/k"è¸fÁØñ—žãLid!ØöËt‹/ õš›•дé›]·ÙS íÎ3ÌŒqqÐ)±vâ… t¦Ì nP.ÿ•‡ý2­ =ç|únéž…>†®]æ¼ßa€â9³\= ˆù>> ï•b”y·‚aT‰¨I’‹ÆhçÒœcû`‚øÄvE˜ÔËý¾=}fvm)ˆfÀܱח§=„yFVKvð¶Î¢-Q–˜b×?ØiÃY¦ë¦µMðËU7d'ÏX-~‘$èCÀK ÃÄOØ{+};4‰O)ØÅ_pBRV5{.&^$û)©ågˆ;ÜÅ䤑Zò¶’+ÞFÚ’û|Ô‹¾¸ýWðƒÀ¯H?wˆ«å|2æt‹*Ø!Sx¹ä3Ò5d9#“=Ü[ŠlÚÕìíñÚÈΓ@JLщœÛ›æó¤:‘{ßRy™@(ò˜(ßw8´Ž¼e:OžÇJ#¿=í _‹ *§].£Aa†k—Ãl£QTHCÖõ­Gi5™{„qogðRPJb®ªÅ/”*ùræPYÊh¼)W†²_Nü´qrAù”öJmgSÜßlC}íòvL0EÖ»vhû¥Cáã÷Ï£DG¡ðT'½ï=*(|å–»ÂùW=ŸYD]'Y”¦µ©†¡´×¦(ã§?¼ö_x4£o ¹e. a6Pu…öP†%=Tÿ|ü!’jk¢Ô¶ >‡úFÏÑóи'Dï 45ß0ð·» ­É—ç5<„ ã°üÑ¥– èâ‹£´Ö®îÅQñ.Qè¾C¾Ù³D/]Fm†`È¢‹ÏþÅqæË$9|Fü÷8—±6ÁG&fÛ³§œ`ìÂôÎê0>÷GÝÝ=Ÿ*ªâuÎ6ÁÔûî¾·=º0«ýõõÓµ8˜û)+мßjNÞ`‘kÕò¯÷©à{‡÷†Uy,zjòò'ü  ç×'õ‚ŸóQ4â§F`•—¹>š7Ö“ù4ÜÈÔ`ë→ߥ $6öq dÝ@RâÑÌQ•—H6 |Š—)Úι³}SAªEvÎkÌH«9ùÌæ³n:½ÚÄÆˆô§Ÿ8:=¤@FiýÍS6È$?ÖÖÃñYœæn±u![…‰K++røÒÚï-8Ü·sÍ_îCÞ}o¸.op#?ô¿‘¬FÁKÍ~_o>Dá¯Ö |Þ(ö¨‹ºW?÷†7f¨@Éž­Çv¢tK3g‚ýC”£%¹¹!`Ž ÆÍÉÚ5N¨˜½ç€™Ì2*ŸÙÇ$¯â€˜ø¢þxª… .—ÀEÔPØÇV.†š«39Ò-ô¨]æ=ÈHÞŠº–[ •rá¨÷áæÈ#ï&4°îN£zކ5ê² uáhÌ©ïìáòM\×^I± ©ñöŸ"oÑtôìÊ-i4;9µ-wéšÍJ„3Ðü@ó£5ÚúÖîhÞQùólì¿Ep?)SE‹˜Ä¯/gM‘„)6‹eoÔ6úª ζ@¥ Ùö³EP:×w¾Þ ´2Î]"è¶’_Ø0úW[&ÉžFÈ9–ÌÝí ÒqÒÏqåóýƒ”åP±çj¹G$¤$/æŽCz²œ‚ÖŽÈ/|Ηìs¾Ç/s™BÝ :›`•ð–èØ¡±]áÐòÞ4 Ô;¡ýG¡[Ÿtj5~{ywº^n]¸Ë=$4µÃ<×8HåZ=zê5AŽÐ-{æL^S tVŸji‰'ÿÃoítòü&SÐ:ÇÜ¡ÍcõW=œÂåÐ*ÚA¿ö:Z¾.ºWœž‚–q‰^§‘=ÐYgé1mU´y·ïUB‡V¢j9Ç tZ¼·uµ…®ð‡unY‚Ð=?³•¡|zÒ?¸Éný¢‰$¶4»aпŠÉäª Ów›=̈‡®½^zø¨­AN¿Pãܤ±b ðÉ –Gž7LoO¼Tf}>Û(í3†o»'ú(ÉÂÂ]Ÿ×L¯`1¦ìšVÎXN t_w뀟3E§ßÂÊ”¾Ú·ï&°.ð˜¿U6ók_3N Qí“»Ww†"ÉÉÁ0ñ§Ì %•­²)¬OÝz$Tw¹å Ä–=xíøÌܶ9,pýaÒ?pI>s…-6C®#“ðúÄYýÈBÜÑ-qWY»Ö.M­ßÃ]i¹2jìáÈñÈëç ?#r•¶z9xF £è… äõ9ܱ%œƒüäŸê:£@E~XÓî(ô˜ËÎD5EËÞv風p'Ý@Ð<îÕ¾!­bƒó<§ŒQê´rjÔy#”þÕðcjâ=ÊÆ|Ô­!Gyã”Ðkwp¿ç–òåìPT¤nZäþ„*V9Ü;¿d"¶¶u¿ò'E5±‡*ÂÏQ=šÞ,Lö jPzÁ×)‡ÚT[šþB7PçÓÚK3ˆzy³ = ïÐÀ­1þ'¯®n*‰½@c?·–ËÞh2}ûƒh$š&È÷åŸ@3í7^Z¹hö1ïbí½çhî5.þÍWKmØïŽÚ'FnPY²¢.Ù;ÞkÑ'Qï~Úæµ"4Ø9þíç&4 >œ‘"@ŽFÿYçŽeã¾q{“½²²É)q]"[ö •JCQF¤¢4‘4ˆB2#!”P6Ù+Ù²’‘Mø=ÞãÍ{ßß_çñp»ù㹞ó:Ý÷ý9‡ŸÊ­ £É®²Ýþ²ÙhªzVò~CÓF«·zThv]KÎ<@³¾oe6Dsû˜õ3çYм.©ÕxÏ,Áß.óûˆ¢Å3‰€Þ‹÷‘hÔ½ôšvËVéù_žÁG¹€Íôb[å9ËþòIþòþrþBÒÿêßïÿåœü=ÿßRŽ*ûÐÀ¢ú”­r-¾H¥Ä¯’Tïÿ åÿ–qü-¹úË/ø[²ó·tgËßþ-oý%?s­ä¦™4ÿ•ýüË7ùûõ¿Ç7ü–¦â‚&ÚÇ9ÿïïhÍ0a+a.ü¯ÔÇ×*{_TØVûßÒõ>ÿgŽ…:Q0 úã®Ø^k:Ðj|(ÊFHu –oŸ‡q£Üû5Ë[å®3,""wä·Jðæù&¬“™¾Á"{àèÍüS°Ìö³Åþ³$¬*Z ?Ük·ù,Ç£9‘hûFÙ±ô´ÿÊzbY^‹ yfeªh‹­¥}N[¥èŒšnÒ̮ȴÔÓlú5 YJÕ"4A¶Ç® 7ô‘ãˆr7ÑÁnäR(:~]F· soð³Z ¯N¢iAò¿–þ- ‰;¤/P8¼“AÁ·Ÿ×ÓìBaë4òè[PTðÃ.v1/ß¡²wg×a”´\²!íþˆRó=wdn¡ì1õÏ=«P^ ¨·NŽ0O‰FbL¿6PIã`¤¬}*/ø™Œ…ÕânöÄó—û‰QÕ»Ú­±| ÕFëÞ|àG´¶rëàDŸr<ªÖ¨iõ³"wµVùWéuP»šºðTy3ê&…ì2.áC}o¿˜äëh(Ÿõvì¶UpÕ×Qr¡±z6%šŠwìioEÓÙàœÏlh¦<ÙvÑP ÍZÔ†ÈMÑüäÇéh4ŸîêxøŠßÜcFh)ÈÑbl»ˆƒH6ZÚ «HkÙ å'n3±B‚¿øax ­<Y”yé`…&:tÕ´o«àï|ý- øË)ùËÙ*“ü·´àïœýo¹Öÿ–Øýå™üýù¿sû·ŒäïÅí`cOª=/? Ç…$¯Âä­BZM˜Žßw,Jsî¿2’K þÎëÂ8Çåþ5WX¦äX´4–†Õ[³|R^m°nÅðxïý0$ÊoeÇÌ­2Á¦³Ã»“|˜Tm·ûQ¤¬LTPtFêf¡Ÿ ÁH'zž-ž; ûžã‹_ÈÄ"õ6™Ô™§ZË>!Ù*Õâà\›=(¡\/Åïòúœ@ž¤žÛú®/Q€ý¤/½©6 Òœ®ì ;ˆÂ"\½¤QÔ«aCøË” õ¥d±ìÇIíf(ës¤Ø(Q¾œ%vÇÓxT#ãûµvwñLrºàîË´ñwøìq¯:êŒû~Bp‹]+PüÃWxvJÂ1Ä OK;b”füH>ÞQ•­ÊÆÛÚ Æ—cò‰x(ÔŸ* Ž Ø gÇbóý¡)ÿ¾²åE€æéRüµ-Wø 2¯ÝØ|.íWfë]hv<_ÉÐ$/™`OÇ»~>o xõEëžy‘B]æî{¹LPÇGD5<µå|Lf–P{*èT£c;Ôðh åqA£UÂÅj¨‡ëVQSŸ¤Ïbtcƒ&ö¤ˆ†‘óЬ*Äõaå ´øˆ¾°ðvƒÖwûÈ5ù¡Mã{{@óUh·ýòkÈ:Ó„ø¶Wì…n»¶O¹=ÐgÀ¤eÞAý«×w?np!’è[×xŽ^P"½m cÅË?EóŽÃ/GfRáýþ0ýü÷/¿½—áwèöÞÀÜݪóM¼G`á Ý—’zXº|‚á‚âIXy6–r[rþLÞûÕqîl¼Ô§²eGb×dIiã\$5ðbŽŒ!Er3=¿>¤ôgú —¡„ÔëQ¦5H×`/?Ïn‡Û†wIUy!SÌ—¯#DÈâx0lÚb/²ídÕ0n¶CöùÓ‘O‰‰³ÎÙñç£ä®HÚùòòÌøY1ÕÙ!¿½Ú—öàŠë'š ¡à ÷¹<;¯ýC¶Š"Ó"WÞœ~bãŒÏ„+P’‡Ø¤2'¥nÿ©­æÏEYùož™ú(O7¹Û¨RÂeä2ìPI‘[þòôT™x÷œ¤w“:‹¿ïUFUÇ¢ËMyP­ÙûŽ;M"¢Vk¦ÿž›¨ñCºÀéIjšÌ :A­¾Ÿ;îK¡v•÷¼Êê¾ uù¨ï֙𬀆Òòç2¢Ñ%g*Ç4¶ èžE“ÆxãŽVDÓØÓþ¯™¢™Ù˜)g„šý©‹ö˜-Có—²ùwÊП~?¢1…íîŠw¥¡¥ ÏT™;Z®ÖX—C«CuúwhÍøÕùÅ+´Ž° 4…å};Gý0ò‚êV(2 ›ÿa’ÕÈ ŠÒ+¦r‚TŸ)êYrxí-‘Áê $Ǥq쿉tAR Rß R()øçÿÞÜ-ðجž_Úç9ubY‰=Õ.GÀ«¢Qó yþ"UÿÙL•ÞÞ¢ «l…,®ÙŸ/nR»?ڊΙµsH¡Ù ; rÕòj3£ià=Ͼ¡§Ràýj WoØ:Mé-º¹‰Q#óNiSÉz,Gs²G‚Wº3L!Û}Ì-ìóë-ÍÑN-r4†\…4^y¾r_L¨äÇ(Ÿ‘m]|Q =™%ü"ûŒ«ÜSZP¢pj[u®,”imüaò‚ŠÀçÍÖU)ð…èiÀ^)¨Éý=3èóꓯx_ln"ìŸtÓ¾³{¡ùí„m´^1ằc´’ £Ñí!{ß,AçÔ›¸ª'tŸ?3®Ó ½‹C%®ÉÐoö[ƒt:*ÖUδÀÐE¡0ŠAY¾Ýi¾½ F»^¸m³‡Ÿ©ÕËøîÁ¤˜{ß5"O˜ö¯’.ê3soäªü#`öÊÆå¡Ìï0¿Cø·Ä‰X˜WºhéB ËŒùÎ'ä™aõúª‰ñúgXW7$S—º†Dq!Î'îM"‰&Éé“בt¡çûÉ…v$/Xç›nðCʤ»>æÊÁHýåÖZô‘nûŒÁÕýéÈÐÙÃpÑø 2Q+]È”?ƒÌ]íñ?Ó‘µ~‡³×/dŸŸgmA®³¿*Z}Gç}ñ{uêxwOàŽj[ZZ÷‘K݉¢Ú±¢?)øQ¼!æ5½åÜùòÙô„c-ÊînkÈ{T‹ò±×eû”ï¢bQíÜ%x‡Ê³ 6^p·YA¡Ä5[ÜË®KB“òA2ÒéÚÔø¦2lµ5ïW Éwá~íânw‰}¨CÇÑH;a…ºYG_Ú¾†ú&V‘O„Р¯ÄãG¸ݰ6ŒúÈ >ÈC“>Ÿ‰Ýºhjwàõ³KÉhFiÑøfàš½õ {I>‰ææ_ÛIõ£ùo—áõFg´éÍýÀ‰–6ìW~5£e)ô¬^†ÚÝYT·¡8a×zÃÑç_ÿöêÆEx_[Ãt8©²~åÛ‹>¸ºÅ'ySÎÖ*ô°RžÕ?!ÉÌ‚ëñ¥ðªýüpä÷ãk÷Èw¦fK7ïSN#W'œçýQ)ÇPàÛ-´¢¥¡ï‡È¹¡Ò¨SºËŒ ª'[r†y@]UÍ=—„FÆ‹l+1h2Ž¥-¾LÈq^L²N ö´8Ú¨7„‰‚Ã?ÔE®¢Ðʯ³1·E Ê÷§”£XÖ'Å9”¤Ñ öB)Ò|Æ1”%v»ùÔå–MCkÝF…»Ìkº9¨$ÒYUq›•;8Œ=>¡ÊôÎþ¹îÓ¨j{«ùØ>T+à·²;$…(ožM¢FïNž´éIÔ4Q\"bêD­eÞ.e=fÔ®èó¹ñšucî¦Aý³*ÕR'ÑP4â/»ÔðM3»=cN¡™ëà—¼¼h.z(œx@Í»žº¤E‰æÉ¬Wª§-QÓŸb‡9ZNñ}”*ðC«Ù“ûó¨ÐÚôÊŠšW ÚœQêE›Â æ¨Åǰt17dŽBžCwˆÉãͽo9ÎÝ Þ2œ¯Vg8OðµÉ3E…b߯ÒüÅ’nŸP½ÇóÛÊïìŸN‡W†&,)cû7ï—\0-ÌT:Q?Ë!êÁ/ÑÂwy®&ü× Âq†ÞÁk›÷¥¬Q~‚ôÞ#Œ5  ³åˆ;>uwÅž^d¸™¿b]Ç— gA(³Íõ"ä~ùrËÞ¿H¹ôŸÞ‰µºd¹–š$ ª— Çÿª÷ŽÕͼ6õ0lŒ9‚¾í_ÿOgͯòMï%ø ›Û›G±„Ü·´‹á^ìÇècÈëDSÿ^Â>Mµ'÷¾,IÅp?»ï%¤–9•7 Œu"è£|TX%¤Œ»<„ª¼OgZ$AS`xŽÂU¨70ð˜¾î•R¶JdÐì¬ÙgÖvZÅ»Jön‡6Ú®¥æÐ.Ñ;>Ÿä7ÆBm){áûrÐõ{‡Ü¡÷úú]– „½“kÒìž0' <:¶pùå Òž …aó]bYl0ûhãú6ø)'n«ú˜~uÆ’öÏÇÃô!㥩±Z˜é¬©Ìèß³nâ}¦êa^èPð뼃°HdæYÑù –¹wy³,Âê5R÷€dX ¸µg‘èžÞ'ÑV)$á¹Ä˜KØËH«'ŒFr¨‘üÑ2qŠtRœùäΰˆÔ©.°-ƒé˜<¯ÙžKA†¦.ÿôsȸ¡¶çã‘»È\ùÿ°¦²¦;>å§CöR‹…ÚdGäÚu@P_Ñy$ãK¥ù—?&¾\ñú îÈ—â-àÍG¡ÚA¦CÖÅ(Êæfœ©ùÅ£JPZqàNç>U"7b”Yv<ªËüå¯L~ŽnuAň'Õd×~¡rõáÆ8Ü-áúµ¶ÏU§Œ %Û\„ÔThŸ}E2V¯os¡¨y9_»õæoÜ¿+=C¾7uˆ‡Nqí2@Ý º¸3•ïQßj0:’•ßfÂ~µ}‚F7¶kû˜¢ñ6Ñðá|*4 ÷ð ÔTCÓà–…ãªhf|NÍz–Í,lÎïTDózq“7‰jhñÀÓ›àk²ù‚Ñj{àAËÄ´¯+ü½Ö¿ ê_ïƒb:à mÒ†w_tU»¹‘íñóJŰ÷0óG–->IÚ#­ÏgB´áõá$15ÇjH‚&òTIx¥Xy-Ð÷ ÄDÁ£H-M\ÂAY³ÈÒ§ŒXRp‡ü‡ÁëNÙo¡$no¥“œ'T”_Ý8&ÕG*Fý±€:‰.ƾ<Ðp¦™ÞÎbûß Èø½¦å3 _Ȩ¡ÙC¸öv×4k°uß /¦ó”s±Õ^Ðxz­÷ÑÏjhà<îÛÄu©ö1 u ¶Ûh™IÒjo­Ñ¿up%ä¶û&aG¡v·§ÿCsu¨µл÷Ç}“ßµ`Bnõ·ÅJJë6ù”"ÒœÜЄ½/É6Ÿ?äóÀÊZäÜq"‰‚Ö<á!Me>h“0zýÚ™õšvxõBç.½ûJ¯¾Wž»l»z?’0®®¦A¿ï‡gç#`ÐEQʘY ~´YrÜÞ £iþIßUyàçèF°sÁÔQž•&#v˜¹#ÐÚ1 ³þqBJn0¿_?ªé9,jÙÜJØÞ˧˜ßW•RÂjñQþ÷ K°îù5-äÀ#$jdcg‰&B’è`Æ'ÍŸ,Ô@èTö,R$Ê 7å3"5é¤Ä>ùgH›õac  ’GŒÔö¦"㯘®ÔIÈùè®aã²Ú:¶Œü(EöõlÅ&ÚÈI’o_tĹi?g?ÌGݙғ=ÈWxð•®°7 |x.ºòÀ/ÛÞ~‡Bu |±ä7P$ˆˆ¢küŠ])¡ð|ÿ%^³ì39aŽRœ4)æ$ÛQ¦‰¦Ëj°å~dÎö£‚ÏÂécÊÛP‰3Ê‘™«™0O5WˆÞ£Ê÷É(ù®ç¨ª”Ò"wÕª„µÞf!ò5ÕE¯Dqßy#UÔ4º¶ÎóÙµ–xjôI—P»Ü°‘G u£îïOŽ'ì'÷¦MD¢¡°Ì¤å¡Qƒó‹ÍçaO*mØÕ¡Éb”*W š–šŸ|¼·ÍîõhÙ+°£¹É¯¡©hÁ¶ÆÛìv-ºéúW|ÓÑ2‘¥öfú2ZyÜzB9€Öš-kžýhÕö^'m–FŠè4·£íÀç0§o°˜®~wäIgœŒðwä`XLé´)$KtìâyZ¾#Ä’÷EdË«+CâY>ñ•¹xBn ºîðSâÙµvl«ƒØ—Ž’›BÔ$¹ù½;•ÿžöÏÁDˆ6{¬g¦ q_Ó#“4ýûçíÉ_OBºjfä)ÃM¾åBJ3¼»óò=ËójBþ 9º¤9oƒŸ®ª AîÅàô;—}à=†ä(SÞþOy´ªýŽæB.ñ(ï"£.äì½rÜo3¯ i{]ß̃IÛ«Í·4;vy/·¿8äœ^RV_„Ü›L~79¶8FéÑØ‡& b…Ó<ŽÃçÕíçä,4¡¸UÓî{{+”þèØ~ìg T¤_¿Êx ª•']‹ö—îCŽ3CÌOÍ`È”Éw\JÚⓤu˜BØ/)n:¸æ{Ô¦› âðs^æyÑþ/oJóÍoi"=Ù¶¬hzÈŒ·i,#ó…‰ —pžJ¸‡Û†ìk¡„ç[×Ömü'•¡6p<Â?cÔg›y®½…Æ •¢ü@“_é.Þ˜)hú•ìF# MŸ¿ž‘t„ÆKzaQ&h$ñͰO†z×”sÜYÌ„<¶@•úŠàcž6rŸr~æmB®HÐ §- V`äþLõÔÚ—ªóñn>§óª–ê“/;ZËð@£á­·A‘=Ðdy%Oÿ4Ÿ2~À-Q¾+zÐZÀÞR¢ñÚxo=z}оMNɱЇŽU{­}LÅðÝ}é¯x3ôÍ´‰ \…~uKýî tXóÖ ƒÄÏvÜf…Ñe»V.ÁOiïƒõ0yáC`AnLÏ÷µ Ü„ßùF¸sÉ=~Q¤°Ôø¹0”–ê’`ü ¬JÖ~HèÕ„µn^ú&=3$ºnxŸ¢<Iôžm{ä…d²Õ âDJH±G½‹]éî|žf‹#BZáç~·õY~kÍë ããÀ&îUdVɸ±ü Y¦(y³Ï_D¶D-Ö §‘ã…ÓAÉräJ7œ¡×çÅí¿Iz÷nÄ"ß¹ÓŠ)—ÙPàaJ€¬  ÒÐ×Ô­U PŒLSìí ±u5ò”bC±ý‘sýDë(áΣʢ.‡;û™ÂŸÌ@™ØøâòŠ(×̶¯&PÁ¹£Åwñ *ÑÚ/JBåÕžwæ¨ò%Óò\GªòÅE´¬¢Ú·Yi’=ˆt¦öéve¨Ñ"Ò­žt5õkÉö¸ä£Öâö|‡Ï=¨]ú^ñEˆ3ê†EWDÆ¢þ‘™yéz4ä“JžW‡FÍ·T>ž{Æç\Ë}DSò ›ïƒhÚZyð#1š%´ü¹Æ€æ—üîòæG ÓãõKÚoÑrgåPwZÑýà"V‹A«yAã5 ´öyâ ómº/ˆ•>FÛ_,lÝmÐŽrŽ™¤y;, µL‘ÒÀ‡õØw¹ýåÃúb!ù ¼¡ØûŽú$»Å,v$1S¹™Û!Áô‡Km.!™ù^ø‘ //t{D첂èžü¥g_!²¥ŒZL÷<ËóŸÑ©€'–Õ6|éSÑUBþs²|µ Þ<%&±¤„L]¾7 UàaÃÒh˜!¨e“‡|hÑW€ÎrEݽÙ¹ýn7¦|¯þ§YGç Ø>²{qx“·bþ|e¶h3¯µ/ÛroæÁG“¤-Í6ŽYbg$OÓ¹\o¹ôo&&|ƒ·8FyA- 1¡`ÐÆÙ>…ÏÏ—’¤ƒc¡8AŽèĸ(”æ–§Ò&¨CùÏ÷™²B•”ßž³/ úsÞ m¢1Ôžíïñ„F³³bœ9¡©kjªç•?´Üœg)šëƒ¯VÚâŸôáÛÑ0=çH+èHºó>î=|§¯Í`ð€ž§ 'sü  /eøÂ£À ’Ò%f„Á± ¤{¿YàÇøäy½ž0*yÕ,«ë"ŒÇ »¦¿4þ$õQÂ4»Ãƒý¡¯`&‰jþzÜe˜µ½’*fåóbõeo;a‘=È%õ},K'|ï(†ÕK,¹Ï.ÃÚ‘¥&Ý~$².<•r: ‰›,¤Íg ©'¥ô˜ü$ß7YÉ5Ö€”º²éÅÕH}³ÂVÑÙiÿ\£+y‰ ŸÛNð9N"ãOÃÒÚüRdŽôiœÊDÖs=?õ$ŒÝ5eECì rÖ¸\˜Ü‡Ûßœ%fl6A~fß0£ã¶¸ãì”狺¦•ûfEr£Õó?¡¸Mت•1J®¸ÔŠæEæ²ç­cŒÛQ^áK/³2*ê‰ym'9‚ÊWc¨ÚQ¥½‘’%<¬xVŽÕ¿~s;Xo…/êî»*žCMëW÷rÄp?ç]«wFoQû—ßÅjÔM=ÿ Rõm$¾LˆGCÅ¡3ø RYå\…ÐØõ¬>IBSÃ}ƒ;YÐt®/x!Ÿ ;¿bIaUAóZ†~Z¯B´(«C©´ü¢êcÿÆ ­ºʹîBëé^ Na´Ý3)ìÜ Êw{„_q€ÏÙ¨ÊÂû!GJýܧ„÷­Ü)ù#a²÷øÏmj[|’ײ1O–è!ù ßjÛmîTCvi?¼÷Ît.H%Ì×áÊŠWñ[š°£JMyÜ 26’©´Ý…¼“aÃ÷¡øÑÃa§Q(Ï­Š»Éz¾èÒøÄ5r@ÍÏ,$P·ÖØ¡žDð/Ê®T·x¡‰„gÇÞVhr–4¶Û}šÔ6êÌuÖ Ñ/V$ézÚæsŸŠÏÂé ®dêÚãGTPû<Ô=ó] ÔR=í¢e@8ïT¨UÓáµÃu–fßÍëÔÏ̵â öhòi¬Þ¼ÿäéRX-Ô×Ô‹8Ž{Æyã¿#e÷5hö¯'Û`k…–¤ýç/>GZ?S•jÑlöæøætÀ·š'è(OCG‘ëÊÿýðç†ÒÃeèéí2×6…¾µ©Cþ2øZY¤•;?è3wÞZƒ‘pJ§‡Û`ÜýW”ÍH&üºkɼc7L» Þ™:Å¿åv·zúýç±þ¾3°@§¿ÒFf KJ{+èì`åñn³ÇoïÚŽÓyV-F$R¸A¼üˉ'0<¤¸„¤ †U£†^H^kôЮ|©¨æŽ3ŸDš§4oM‘þìÓÃô~dÜëÕO,eŒL­u~IÆiÈrûÍ)ÿåÈf¬:ñ"9vÓ[?‘,B.Óvé»m¦¸ýy'½ã1äã÷.£mÈF3¹#Çk pGßKdwG!w“[/)¨QD¤3ŒÆU Åè= N¢Än…olô¸ó5çíŽ!”qÍ•«a(÷éb¤¡@*ØBx^9*®ó„¤hi£ò›×š'J^£JîúÔvRT¥Šˆk­MGµK\%ì–EËõc…ù2¨Q'¤~‡¤5u9i¬¹¥Pk;ÕíJj—>ÜIôu. …¡¾ Mº< n—r®­èG£¯ÑMóUW a¢ã»ìÂ\mÓgNë[@Óa5#¿!54+nÙõ¢–Í_Ñû§¿ C‹Àj>î9´¼|cäÉW´òÌ=qµšàk>Â\'¢SÑ&Wjd, mÓäyÏâÁ~8›>†öÒâsñ,l?ªsk%>4<$2”Ô€Ìññi2Â|m¿pŽò$$¹ök¥H öBÊ´I"Bn %"øÇÅû} {/>n›Ca¾ŽmJÐ|íÆçrªlˆ¨7h»U: ϯˆ¼–¸CA˜Ç 5—!qIM‡Èí¤ÍX¿yq£äŸ>‘ÆçðŽq]µ0 ‚¿Æi¾1wBŽàÑ£û»Š §¹Ã'0X·ÌMSf.þ§zþ|u  ç’ÅR~ÿ¦/fœºEº™×êÞ¨¾Û̃Ýñ­þÓ?:9L±$„ã ?Î…œª*{oÒ-ŽQÍéèJ> ȯµ‹€jø _û$+ס؛7arÕJ²T³O>Šäl  òÇ÷Ø÷·9¡ÚÕ잤ã.¨“}Úîþ4²†¼\8MOt.&P¯C‹üõÿÐRh]^Ê0V,ƒ¶µ³…UCÇÞâ7Aè ÿ¥kúç"ôYSpBŸûÀÎOп:µ_›?$£Y¤†ï |:.ôÃÈ”ç,iµ$ŒÛßNz1Ws’çh¾î†©ÙÝU|0,vÿ év˜5J` ë7‡yIJG™ì(XÜ‘ìÌ'êË*úòɉ°êÍÃÒ!x ÖzÔÊ´Òä‘HýbÁ™¬!$N-ÚûRˆIuY£§¢œs,¥Y÷'RŠ1…ÖE)"õùÙÚ:ªH;³Â;<ñr[¾Y3¥#c¿ÅݯcýÈ4"¨sYMøØãe\]³uuy„9Ÿ’QÖ|ÀíçêHR¯ _ͨO{+îØoÆr®…ôB¿¼"·ú*H—ôPœHFyÞÊ%KÞ±©½ˆFÕ5¶©>”gP9יɅŠüÓ‰Z¨l¨`skU2®ó¿'5@ÕË'&ƒfPý­=ÑèÔ|Íñ¾ù;jîOM¸¬‚û©½Ä/J¢ö Ô>VQÔM¡Oˆ#=ŠúÎÁ¯Îò›0Ù‘w2¯Ñ¨ã}hÂÈ}4~|Í…dMUXµ²äÐLüdGÑ•@4*¼f.\’ Q$h©ÁŸ›æfˆV'%’…Ñ:–¿š}Ïc´ù-¼£§qõûm«%ƒT<>ÜðéFâç›ÿ—œmP™} ïÈÍe.vk@z¯w]ºgÜŸ$y­î:Þt‡$k»r&‹,ˆ/M“Éæ/µë :”Âs#:GqË-ßuÿtÞ®nȰ[Ò°O~L˜ëènÑÁ?PL§6?”u Êe]/0ZåAU^åvÑ=ËPs€:©ØkêtMQÖŒ”{ª2¼†Æ;^jßò qèͶý'ã 1ÿB¹êñHBn›Én=¨¿äåÚìýê8•2x®A-›dçûR)¨‰9ËÕPNÐÓM)t…PÓ¹ÿY«GÔÌ4$düsŸ‰é«£ÞPwƒwäÀÁûP?%pÊ„ð{.ä—éÅž$äÆwo̧ 9ÌçsãóphyG;ø§õ;´¯d•Ø\ƒ6ò¹Ňéà[òÞIbS)è¸íŸyp²º*éè(ªù¡Ç£ÉÑïÌèûôúûN £{ 0”r8@ùÖ0ŒÐ”L\Q€±\§ëN~05õúQИªªò\¶÷™HÖ_Ñ6Ô0ëA¥'6Ò ón¥Ê“>a°Ùx@FL–×^ß¼ö þ¼Mm´Q¯‡J·©N‰$~žIkƒ¤çj£’Ÿæ"ù{ÝC[¤¼ÜïÓ:¤¡âµ e AºÚI°8í‹ÛÊŠ+®ìG&÷q+ú»ÈÂc‘yàF/²4¾±?Ù›s-6BÝ‘sÖé Ùñ¸]7s¯íaäm¼myNÄ´Êt#ÎzàŽ÷JY‡ÛèQHËMåå#žÕýö2Ý EÛŒšò5 ø†ÌÎȉqÜéÉ÷kÊà·…Úš”{Ýñ»:w  cS²|Pqº[dîÛ*Çp%r*¼B•W»*FÛTpÏï´¯‘ͨv¬$t#IaäÀźíÏP£RàEýa^Â\91zT>G­y®¨[*)¨]<ùÁ#uZÜéC}Ë4™–]hÈž*þ¨eÚó\ä£ñ )ƒgŒÑ”‹¢Î·'M3Ñãƒ7šu×e ao|qIý9Z”ËËkMï&ìêãǭЪ3 óÍ/´^¬gD[éâ2Ïø³xÐ &Y˜üÑî«s±ÿîzt0?Ääœö æÍ&£)ÆŽmÞooÔë) ™‡Š5 zs6û6ÞHÓCâYÞÚÐÅhˆÿýYÂ(b^ñèÍ™úvþ“Ûâ;³!FíÛ+zŠZˆš¤!ú 1û¼}I…ðAÖüªe ˆz0iÏözï&¯‹bvÛá|IiàiF)*»3¯nöT‘@V›”o˜Ÿö?ùkE@²»jtœS¥!ÇÿÚ¡ûOßrœŸÓI·-Íé‰Ö›çnÞÌcç–¦CöIÑÒVVòÚQ^£òà˜éšò˃^v’ãboLñ&üóø«êó—¶8Fî®éL&}‚ücžÝ!Âëð©ï—8É!(Ö¥5ô¡«RÛèêCœ¯ üúªÝ 2FŠ4ü1TsÉüáæƒÚ¦;N-{ƒ á½Å„õ9h2~¾qÖÍš{NhH%BktšJÏA h{¶?¡; ½¾×C=ºé{Ý©‡î×'µm‚=¡oOOø‹™aè/›¸’ƒA¢ŽA«ðãfâ3´"0’íüØòCŒ³Æ—°ž1…‰$Û‡®ÕÎ0Ýj¨pnfQ»© o]B§çMÄNEŸ¢[£äNÝíP–̬ýâé0TIÝ,LyÕÕé\²GÛ‡¡6´!A`¥ê(.–³qA#M½ÆvÏhB®’m”yü÷Œ ú«@§Û÷J¨ËÔK¦µ¾ µ®WâîZþ†šàr­Âæz¨QöúVé°j̘ÿèÞ„šT©¬',ŒŸÓÊbQ¦ ÎÂú¦àfŸŽJœ€40{žet¥‚Æ{6ÆIB’Ðf’é’ÍñB—ã„ì åS‹ímîh-›ˆÓ<6_W¤ŒkûìáÛ­£…ˆè0xtúä!=èr2x%²zÈ‹E’Cf¡ïš˜½¨î4  ƒåûë0¤M·Ž·aøzÈݽ`t¥ë”{žü|¤zuÁ&ßv²ŒfÀ é E屇ð»í…´ëÔ˜ûªæN1Q‹ÄN­&³{`ùÔ¶›¦‚ð‡sV>r÷}Øèî*º$ŽÄ:…t÷U•”áÀ¹Óý‹HöÛym°±)·)ï›UBêKÝ_Ø®Ô"Œö½ nc¯+y1LÛ(=IjO sÖò*7Ýcd½üŒŽãg²»1s^DÎ'Ӊצ³‘û÷]ç“òÃÈ{ýÑPãåß(ÀÓ”TG#Ž;‚«D’4åQˆ-t<Ž"…‹rxwOÝGÑ‘"ñn¯P>\V…;ÕD­B™mî¯c¸LQ.\ÍÐà€*¨Ýb­ƒ+¨8ø¢PÃå*‡Þ£®Ž{Ž*–ýEZÏàžï·rÛN/¢šÑÙ#£Ëµ­µäß}ÉP£˜w=Ï5÷e°?ð¢C­9ΠðÙ¨ýiÙH{¨ uƒî½ð#F}jå·^gÐYbÄ®ŸºZ)\|ƒ®æ‘Œ\GSÁ²»FßÑŒZTáÔýL4[®’è;±„æ Òîµ—Ñb#|À¹Ü­8–~ÉÍI¡õ>ÛõK?¡ÍU±ô]Êжé¤ýÆ‘ýhçS©þhÿå ‡Y ºøð!/×0Ìݽ2·]>¸»?ó, ‚ŒÆðÇVµá5Qð!’H8mgåSn ¯Ú½Réj!Ž„x Är‚Û^¹|U&ä¶“Ä«=é•I»_×ÂoIÆÁÓÅž—ô2¢›×áèˆëÝ vcïN:KoH螥¶¥£†Ô¤Õ¢–>UÈð¸Dï¼x ²¢ì>ò(ûçz¯´d?QÞr9 rdD4Cb96ï ™àë:óŸ^(ðä%•ß|ŽõÞÊ)_Èæý¦ïŸëx¬?6ó «toüjóX–ês¼›¾©{Í_‚ã>]Ûszz‹côòỼ|\ ©¸E+OدCÍEÔò¡˜ci†˜J%‚ìÈ›­¡Ü¨©·6–*m£Î›¼Þ_ªü[ëZ sëiuÔpœö…埞ê‚&ÊÙù¶?Ð|?Š~£ê.´jù|îj´)ó¶†ßÉ‚ö£ß—tiV 3MÆ«Iº•ü•’5@qgŠÈAèh‡ÕYš0h?[HO´?LI‚V^:ÃÈ¥#Ê Ia0VWú=°·&ìîd7~Ž„)›þþ¶:˜9èôQµìÌ*-v9/ÓüŒO¸¼/,ª #%½ Ë¡lÕ7Îê›”7Ÿ’¬¥çÄëLÐ Íĉ_ºGØ1ÒVýº’LÉ^;{v É*zÏk‘ "Åp/­1Rë¸vz# i¿¯DiyêÆOªøÚŒúÞ)Yi„ü¦`­M÷°4Þ %ÌñJ3WY÷S4½š¨kªJ‚f<Ý-þ\FóÌßÖñÚ+h1š,ëõ#­v•%-rõ‡Àh+þ¹•~Ŷ&…jæ.ãÅÉŸ¥±°üi·¢¨û(à+ðÜ…lªíGX]s!ãå“B²‹szÔ»ãïÜŸ$1R¦ÎWL›oÆ,öKMOš³´²ä¾g"9ž>y$Ÿµ¥qŽ¬ÅŸÞ@ºü‘ágÖ»´ážlÛŸWE%ª™A™À{ÉvÂ~¦á(M_ÕÛÄß“ìí ø -süf?œŒpÌö{ÐÀg|§Ÿ‚°¯”¿# n…gØö$C½UÏqñu¨Û>°ÁÇJгI”G¡†âHZéÀy¨®ñßva´€ðÚÞÒä†Ôø²ÖX½³#÷þ“ç››ý§ÖÔñõܰ0BGÈ‹ç^}³Æ¿;È&¡)ñgÎZkhΩWö|Í-Õ‘uôG)¡µ²çaɹ³ðuÒI¥ýX*|;z7õgtlKä÷¨©†.þÓg] ;"c¼å]9ôa)í£ïÐ?'­$݃}ù^ú·>À…¦k/W'`Í\c‡ñRÑê3Žð+¡ACÆk ¦]ëø§ÏqÁï½ ÌïbanONÿQjX8µlïþY–¯ðpÕªaÝŠ°~ß 2±Ê‰¦¿ÐÞkeF’7õû$¤Uì¶O¶¸µ"R„QÒž:…T?£U"¨‘öÁ‹Àc]YÈàèÚWôÓ„]3‚ö!³~ìþá§zÈÊXóåMØ,²SQ·Ú·"çnßU¿ä~q$ü‚f2ò*Äy)4†"ÿÏOŒ¤f…¸ÃÉ>b*Ý G2_ÄÝiAáÁcrÙ(zj=ý©h Škð >OÀäÒ¦ ØQº—ö°ûÉt”»ýêË`&*HŸº6JôÛ[Þ'ö@åëK®~ÌOPÅ·`ϱæhÜSq¹è›ÊvTÛÅ*ÓldPbðèç÷󨑿ýðÖÔ„?9qÇ$/õLÆ&µ  ñã&1—OçÐTê†O';šqG‰88N 9[É«ëY3hÁ_xÆÞg-÷¬?zY„V'ßM­‡Ñ¢u²¤Á¢§*ÚKøDûwrÍl7/ÚSÆžZ?â‚…E®‚Ý:xøycc€$Ì~¦>@S14S‘M@†–ëp ûWH^«ä‹!r€øO{1@Ü'Ajÿ,ˆùؤ$µNÈmÍM)1®„|6ç8; k¶LNmiðô̧ÁhÃxJ—|Çì¸<Ë»Vðëw"Á£MD"æ sZ–´#“ Î3,@ÆÎ¡Gt4µåxHí´æ?÷Qj¨æo^/c|g@µ™ãÄBS®n^Ëë0=öŸÒt¾Øÿ3t3·Å¯ªëÁ»ö•ñ7Jµ›y-µjÀo3Êz8-oé;rÕ 2“*xv’nÛ@*!'rÙ%;¹Å1zG4\=_>†¤¼½4s> » ^ä‚¢Ñá§²ç¡dåâXú}(çè|︟*™VÚ¢ á‹7çå˜àI¨e>Tõ5MWÜ] ±Ø€aIý 4«Šå~v: -£,:ëñ*ðµq‘ÔÌ_ ¾Íuzz·Ÿ„Î]&ôœQ¾ðýS¤ÐräNè}ôÆË4óôú4J?ú¥Š¹BÈ'àÏá'A*o`,ÛžêÀØEb+W˜ ÏnWÌ®…)Æ©‹] 0ƒ7Í„„×`Vb‡Æ×[ã0/—©õ¶57{yðò¥b@× KXuUJÒ(ä…µ§¿Yõ‚acð€×!¡=H¼—òdP† ’”ªMMHJ#Ù‹N³Õc€ŽKŠGj¥Ë÷/?‡´Íþ^gB‘!ªVzMh KÏ|›¹ ŠÌ&¿wæ8"ËTWÄõd+±&ÉBŽùÓµN|ȵèÛ¾ky—¤¤®ŠŠ¢@7¿ãç^,ƒdg}ž3S·~3‚bNÏø­òQrûíiŸU”ö}VA+ƒr±þ|ìÁ¨dP£G¹•šsÚü—µQEå±óKQ T%÷ãÿ9èê{éT¬Pà óÂ3%P“á“·›/j}ß›+ÚeogNÿ&EÝ8†ä¤E}?•ÜŒÊ4´öj ^ý„äº)(¬¿ ‰ÄÜɃa4;¯ÏkhV²\È[0‹æËt%¬CÑR÷öÜ~šf´ÊâàMUµCµEU©V9´ýµ­ƒº3 íª ÇO¯ª£C»´XñG3XʨJ=g±Ùa:Ñs²>½#y]é½ç©žFBJœw¯Š'ÏŸ$á´«O4Ä3Ö‰€8½ûV¹c3ýöµ¤éevÂÜYtJåÛÒM®¹¤u45\Uw\ú·ùc,zðiÂW¯Ú™JŸÙ~àÚGxÿ®‘d\Œ†/n»!.ºj„êÝÉŒ¡6ÙªéÒŽ%‚ß³n„úô3ñä/ Þ5éà¨S~¹Í¹oÔ Lñø±­A 'e¹/{T_]aûò› ªê zí†ê¼Ö'q… PcôgˆÝ"žãvV·Šž…ÚKJ¶—¡®ª¤ÝðM4äòPÙ—Y!Çij )? /ö54WúLkËõBKÛ±}O’x µºñºæ¥>ø:ð\`ph¾,M¿äehoýÐvMÕ:+RÂN¨±@÷ÎØ·ëíáÐGt¸€Øú“§Ge‰`Ð×yÂâVüpaíþS #7¿(ÀX[‰U/L<6OÔ'„©·æ ³ûa&¦bqyf£eFT¾jÃ|}ðÀ£ sXÚ3œc2 +?¯Ï ÈPÃÚäûwÁHù•&Àu I˜°Œùø gÀ—®X¤`moÞÛJ‚T‡' èé Íô¼Ü™'>]«ãÆ *2êï;}Ø€™útBÞñÿA–Äõ-Uvd‹Ë4‚â;ÈÑEÍ‘–¹õv•ß°8ƒ<ƒ™äODe‘ÿE{£Ë Ü¡:E·›û æ5›¾¢ä@aûɧܿ£¨Ü·¤n”(n©Ï6|á&JÖ+ÁXŠ>J§'0keÓ£œÑýË ¨Àgð“¤T¿ðýüöt'*{žj’½‚*ÎWÇ46áž ·êvr Tã/XËîDxûeǪZ'jäp~Jr#FM5=Áª†IÔúÍ~öóY{ÔÎo¤tü"Šº7ïupÖD}­RÇŒ¼f4¤s]•G£jÕA9b4ŽzÒöÄMwáÚÄ·Z4“bä4SCsùB^Á1´Ð?plüÔQ´¼pbŽ9Zåœû]+ûmVýˆúÐöÖN-ŠZ@;ηAì­óhߨ$za‡ý÷-ŽÑ{²|<±> —Eœÿ…eü™¡(¯UE£Jjœ.ÙhBÙÀ`ÈìÈ8TÔØ©4á wÝ6¨ÉµSו†zkƃÁ¿ÿ@ã娧Ÿ^-CÓHö‘kD å~E}È—,øêÒ¨@{é9| ø6™˜Ð EçãíEZá»^6‹Alô %DØsúC¿¤P¨ÿ)X ñÞF3C£c¶×ÍÀ©î­Ó aºX½ ?‹VŸ¦¬Áä—5¹ã¾0#òr”öv4Ìò$îc¬‚y…±z%ÒhX4æ±éËåCmËz{`õŒZûq X»&sË`ç ld”ðõ¾ÏFb¶ :zƒ‘$ÜÀÑeô0’]øªô®è R$F¦èu µ@Ò³ûy’H[E2cò8¾Ä*µf cžÇó•8[dÞ;ß~þ%²4r—±jŽ#[øu’¾ˆEÔö=Aî vQՌțÁBó´¾U›ÝEÁ¡}O){Q¸lÇ žq“+(î¶jC‰ÁÚ=÷;Ï£´ºƒÈ½Y”»òëÍW©×¨p%¶Eï**¥œ§ßãåˆ*Û´oìPÃ==·ºè :«ýµfOPxö˜Fà¾?§O–oëç¾,>¾ÚHünK·ÎÇ;ô§mvjÅ|SŠáÝâ˜Ôõ™ˆÇ^†_¯êÔ¹-~JS£Pa>!Wþ头 HäòþRØâ$|mk¥ž:e߸¿:·y´?« J®‡NwEçNº„=1ÔKbÍz¯Ïµ÷P9öÄÿ#묢ª ÛvM7H‡"‚ ”„Âó""-%¡”X(ˆŠ ` *%%¹èFº[B¥;E A`ó±¿ïcï£9Æšï\g÷¼®û`Ý«XF¬óŒqr:Ú< ‚ñ es¿˜¬W劸‘Ó³6³jò 0÷ —Îñ¥,>¤‰Lc¿È¬~¨’_ß$I†:÷aõ˜€k®ùyX32€JbØ4i0uÜp¦@"ÍêÇ”-H<ß6™üµISÃø²ì<<壦+R¶ñ²&¤?B#½^ƒi¤ûÇ5äH>ŠûšKj9™~#“×|ß½äLdQ‘ÖìQB6KdQä¸óÖj%B¹~Óÿ~yA¹CËß®ßnEÞ3¿_ö4¿E~¦÷üŸ¼G ß¤qOÿçwÞÿvrÈPøofе«xè€CŠžЇŸ”Œ¼’w¡–Ø¥¬M/¡Ì¾#f¡lñÐCAñ ÿϹÿ<—Ý{ø­ÔçŒÿî¢ü×ÿ{ÿ?çÿó=ÿÙ5ùÏ{§ÔEt›{®çwL¢ÏRíî^ƒê/óNOƒ:ÿ¿÷ÃòþŸÜ¶Ë®ó<¦ñþïŽQ—´¯ÏÒè¡J¶«¹Áøß|öŸ~yâ}æaìl%‰JƒŸ9ï‹/Þs†áµÅPÓMmµê|¹¯ÆsÙ“Ò.ûß\þå?¼¨ú æn2ñüýpäsI[q÷÷$Âo¦›´iÑ ÿ›ËsJæ–gt᯳ï¥íÃ<°å¬Eý¸u ¶Ýþè6µÃî]ùz½£H´4òðC’\2o_H A2ÝvwO¤xâ·ä< ŽÔt݉1o'‘¶ô±ˆXÙSdxVÇìþø2fø:¶{!³èßÇ‡Ö %óÜÒµS4ÈæžFß‚C¼t¸ÿXiQñŒ)¼Dú7œòòÝUòT×UCwʪ/zóëù˜È?ù7|á›é/Kc~ðq&%¶Ór¨ PÊÜJåMS0ʘ}÷²}–rþ‡4~ýºò“k‘SÓJ¨”ÿÜ*ESrµ;æÖh–TB·[¹-¼z¢ØœƒÑÒ3cŒ8êZº¶ß½Ï‹ê%_þœëb+3ÏN¿ƒBŽçú—h!‹\v½û‰8¤¶‹qò”†Ï— C‰e ¾{+{|÷ÄÓÊî^üè qÊ—÷E¥À§ÛWRçlà} K8ÇÔ9ˆ ÉS« ¤Î¿;àc Ú¨Ê7Hµpg»yÄMÏmílx Ti6Ší:TŨMs\öƒšþ£þÆP¯›åå&œ ¿"yÙ}‰ ÉúËòY}hþŠe­Ï åkÚX”,4ÿÕá5„¦Æ›‰WËöAcŸœ÷f’14Ì/©_¢z >Á:åÏ› ÁwV¤tÖ™S=1ƒÆO3†YÒÐÄëPÇxc šYåm¨ô Å»"|ñ ´±müW= íodh¯“ŽèCÌÌi'¡3>0E•iºRi2ÝØ)¡;ôåFôDOŸÍ?Ù½C²ßÇCß¹ÐÁsìß`@T†…I«¾;W­¾Ó‚!Þº[ƒXaDlIÊ4ý=ŒŽ[h“†ñ/+wdHa2DþË}³&˜N ¢ûÈ4 sâ>ßÙ. À"3ÙÐQÁc°\j›µ<¼ +i)g=IáO[ã© ƒ÷°~ôô•»«ð·=ŒûÁ—\øWX«¯»ƒÛÑòHøÙðvÕi$U¥ï ðaEraF¿ÎB¤4÷mUAêvÿ³*êHw5„û§u5î“Õ¯øý$™.vDèE!óÐQÒÒ€ÓÈÚ­z›ë 5rP}mÓûô¹<þ*‰ß­DnB›ì¨ò”^U~r+ùªCYO 8¡À‰_g5Oz `¡Yó×uŽ¥îÚ7} E˸^¯j9¢¸ì7Eq;{”dÍ3îî¡E)a²·œm”î½9¢—ûe%IdS<ÂcÒoÝx?ä¡‚èÅ{/éQé ÅÇb•f<1˜ø*¸ÇÁ݃ˆè²ª¾-÷ÏFuþRá&ñ/¨1tG¾… µr9R¹ÃPû…%iª-ê8~A»/Á¨«ÞFwTMõÚÏ_f–Å3$ôî?ü~¢¡ó¯Ën»Êxö\oyÓmy4²><%¯ŽÆ×SœYµ¼Ñ$j"Yà5šþ8µ~¹ÒÍí݈Íß E¦ÑÎyÆ<´Z?¬ "4Êéå¸Öß^0ÝùÕ 7„[@íå{S÷e¨O$_aúMœ¶2Ò 7¡å5‘_Àxh§óŠþwË:®l1$—ˆAµ‹þÛWÐÝGt¡çpôö''w®ß†¾_³š-§að¼ó»þðS¼zyeì §çH‘°†Ñ@‹Ðüp ÷þC—'ᓯíO6ËÂtÍx–ðYg˜ƒ]§™tXÔ¹TXnW˳Jå…°2Ó™B°W‡U²†L‹ã?`µ@Ë=þŠv?~±å[B;¶œª|°-zjÿÂx2ìjœsןG¢îqK‡!$1üšM“ÿÉŽ^ =‘fˆN^ž.¡H5-uø¾¢1Ò†^§|:RÉ¿»J8ó@|€|ƒC'Ä ¶_mZ8É•[«œðé%ïceð>ýV'«;DÈX9]°€Á²úŸÏ{|›LÎ=i ©|ç:o‹áfA^Øü( #%ÓºÄUâÙfü¡Æ{£°¥Kêjˆ¼Òù ž“Îòÿ$¡qöÇ3ñ´=ž™{Îé2Ü‚æöBÙùu3h îæ|M†™tWÆ3¡Ñ<1 - Áýå—a.E¨ß `ÍàƒúlM]R…†Eß°öC£ð"KãOhLΛ?•MŸ2>tZ˜d(•ÅAkÆcÍgÝŽÐ~þ¸WÙt¸H8¹nƒÎ«¿’]ÝY¡ëÆ\J B·µÈËJwbèqø÷ôƽÖBSyèã²y4 ý?¤JNÝž[ üð³4’Ý‘¾zÏÜl„ÑÌ÷ôRO¾Â¸¯XRì”=LšIª¹)MÁôùì#; ¥0Ûy0µNk>îrõ+sÀ²—ý“O%/`ÅŒáÑÇ'ðÇÍCÔ^kÖŠÕƒËť௞ÃX¬LüpK: »G¨nþ:ԉćîéW I÷ãûV¯ö#Yö»§3öTHñ½°-ya © rš©ä[vŒÌò\N2¼ûYܺu[~6f.=AæûU±G­õ’HÚ³ñAd³¢ž˜Ò…œ+ãw«ß™áÿªk{žÅ#>;­ ðùn%$³”I!ÿÜRÌy >pO :ú…O;îÏ`ßBQ»UpB±ö¢"'Q"UräMR ÔSäÌàEé÷Åî·'î ¬þµ^ÕQ”Û1þ`£7òÏŒ5ÖÙPI-Ñý}’,žø2eæ]¨‹p"úÌš ªú¹ôÎÄß@uêIÌoÛ¨Ñ2Áô}gµ2~¯o¡v”ÛÆóu+Ôyvœ*d— uý×ÿ¤û£~ ßLqä#„ÞS˜'ƒèÔWÂý¿’!Ê‹ ­|´"ÄOx?ªÚ„0{Ê«ru$(Ên@èáçJY÷“!üM©­†#ÄDJmž/€¸º¨Ò’M H6=ÁVív2Ȥ%$ u kø"É›SãÇ7ɶ19uw-Ý÷ÛA•&ÙÌ ™¼U·×T€ð~²jNX»nŸ8ñ rÞÆ*˜ÆCŽØuǰK$Ý¥Û^ru²Mæ°¤+@vbRîÔ ]²­ßÖ× 9jýýÒ›? §y»+´õ:ɉÏXúìû©6(:Z"UË(¥›«WZG "ææÜc/Ï=¾5pmYÕBuôßÜj† Pk–ÐkÃê}«I?ñ?„Æo›ôi>G¡E#«ýÚ“"hk˜—Ÿª_‡ŽS¹l¬ßù¡sÄL“£å7t6w—‡Þˆ;3m·  ï¦þv…+ Ì‹ DÐ)ÀOó"þÚ¯aØé§èõár=ij6ÄóÆ•;®•ÿi„Ië7v7)aúu’¼µƒÌQ/HvËíõ·}Í­ó…`¹‘úOÐsX©’Ö)‚?Ý×Ö¶x,amr™KêÙlLùÀÓ”›£ùÙ)YFðoJX-dû!ìRù1*Ý$*œ’.É߇$òG߈ÖßD2æõ%ÚãHqòêfüÚR} Êt(GÚKÖÓªºLÈ ]Jjv K;lå#Ó¿·¬F¥üÈòÄúØ€š&²©l_ö@³V®£!äZãM¼,µƒy‹ôÀ¾ùL^Û(»j €jcVa ¹äR§†y£Hû3åz(v÷”²qÅ/”h«¢f|ö¥Ô¾ KdÊh*ý¦øŠr7n’uÜŽBù1ïÑ™›Q©â’Awæ+T!³Þ ¿Ä„ª'¶ˆêë£Ú¶ù„=¿j´vø÷°9¡V1ïÕhvÔþðK:J= uî‹:ÎÍ¡®Móµ±afÔWuííKÀ3ŠÉ¼¤jh˜u}ß5 <ûÓörõ+K4]¶Î»Š&ARö”¢hÆüüGD!9šWDòënÇ“ÝøÞà)| øðz‹>Þw>¥›V»–D²WS!B“íýl9| º.‘ 2}¾JªîN0´Ü^€â•F*ª 2¯Z£øÍ;¨þÍ©¹¿iêTõ\~ õ›2æw-¡Ñ‡>8ì '4Í™‡}3¸ÍN6EkÆ>ÐÌ^˵i½sWh ágé#ú¨Ÿ—á2›ƒú;÷‡bb· þþ[¯w“P¿öË\2Î’W>h.€FùéÄúƒ× éÄñy…Sï`Oš½Iw¡U‡)uÆ%Ú«Ÿ¬9• rRE•3ç ó„EÒëZ}èRnN q΄nA‹CæÐ#Æ8¢Mþz—³×ÊÀ·~Q¯ß+3ÐÿæèÅ{D0X$ºv!}~Z¹ç ¿¼ ÃUæ_œ6`ÔoÈa?å0ŒŸ|m[]“{Œ’Áô~ë³6 ¡0ëSyÛ‡/Žþ5yh²ËG/XEpïݟ]<ÍTð‡|?…‚ '¬iOßµ­ƒú£)•ù¯`+ˆ>‰˜¤vü'Œç^A¢Z3!ÞòKHâ²-™Ì‹dªŒ»¦KHáøTR‚>©š§Nv‰W ­™OxÍd ºš—ru}Ì¿6¼¾†Ì‚â÷Ÿôº#+ ÝM4dçåË#"p §G[µmÎî_û ––ñF8½™jI@>ñ|•çþ"ÔÂüãêX<*Ðqå4 ­”8hU¢(iRT‚ Š]Þg 8/j Wütð¨{ð®ÕšJ_$^OeóBYv¥{m¥•(×ÄԹɊò _DÖƒöòÄvRÑŸ9 Oø:àð^¦öÃ<Ý0ª:2ËL¾¢@µ•}¼aõP£Do~d6µRÚ+‡ÄP;Ù:œ]ÃuÒk;Ê ß£n9ßÉìaÔÿ!v“F Ïä+G æ÷õÖeáÙ×¾‡zm ‘ï=Ijb4¾ªy÷´Pšxqí˨ @Ó'[lIžhVèÅ3“Š$1ÃH2„çì…ôÈ‚Ðò'édÑ­h´¾½>b*ªÒ±¯,aYXéð4|‡\þØ«8 )Ûž¦öb›ÿþŽ™Ç/|,zý¢búdâé}žA´$M½D¹qÑö<˜„ýö„ ‹¤h›×|B}¯É$MIC¨ˆÜ‘ƒVEîÎr—øíKˆ95>¬p[â^Š’_.©„dŸ¿ºfT!}*Xpeܲ…*g¾fîñM§Ý¦1ríÌÙÍG8YÕñ—¹[a~ÞþË@ ͪ`¸-9÷ ºž“C§úóÏÈî\|`ÂÙálýZ@vCÁÖ-ÈžS¨,¢¿9·2·dâ=@ÿV.hrGçå$A¾ùM+2ò(rðR2†Ò‘ ×Ã+œ "Éàš/U"T½'ð¡^†êP‡3_ sû#¶Aý™†ÒNÐþvF´hšW…,BÚ¡ÍwN¶7v:®¸÷o?tjž`U1ƒn¯t÷?Ñ Ðkn$Ö´» }ª1æ’õOaÀ»O& Fà‹5}j óRó 'òÃ(‰E"Œ¥w7”²Á¤HÜ'ÙR¹=ï¼Tüæ†ÌvS»ûz%Çåè_Àrò‘ óóÏa%2dúë¬ü‰çT‰!/M¾>¤€‚ŒhØüxREà!üûLâ"Ãn ;-1a-$Š™¥¦¤E$9uS³ß Iׂê)‘BÄÕ_ã±RÅL;yGiÕO5è¾2Eòüí$GFw«w̯,irtpêC/²\·9·OeÙÄ¿^™g×Bå{•ùzäê5ŒÖêîDîM6Ï× ÓȇeJÞš(pXâృÝ(dÌ^¿¯UEŠl’Õ«¡˜³#—¤DJ4‡•Ÿ¾ŽRò½ªÆ$R(s|:=@ëÊ]æšPSMEù‘"ŠÙãûP©z õЬ6ªìË›SÍ@UM¿A¡ÛK¨NÑóÅI45¾ì5F­ŠÀð$Û\Ôþ¤psfPuü^]SÊ F]ÃÏZZcÇPŸ·ïƒÎq<Ã8"9”ñ ßZn¸¸Ïúœ`C£UÙ¨bÓz41íŒ|A‡¦užZr»h~ÑÏNifÏIzšçÊõ¢•ÈþW."©hcÁ#½ð“«·ƒÞÌHC¾äë,ßßw ÃAh§ê$·…(R<˜°ƒ÷kaÚíW^CÄý„t’ûrqÁÝc­Ï>>—RŒƒ”Ìs!¬m{žÈäLFkŲ+Ø/A¥ùõ‰Ÿ¡ú²B«su+Ô->zuêïÞ]÷&z ”*¡$G·¡éž’ºc°14ý#QÛ͈†¦„çÍ@£HßeéßÐ`8T_!õ—º5u»·S‚°@=ÙÏÓïÍ{Þ'úMUz$×ÅK§· ¡*5,nEÛNÖ ¹T@³:ëÄ·Ÿ¾Ð2Ãrê_'´-†ÖÄ7Bѱ}\OG¡“¢üs ßè"ÿòÁàåcèš Ì¬0úÝKÂb Ð+q†/Iá|‹X®®´[~# y¿8aÐÑÒ«%´~üEãi!ï׆Qþ|Ò€0μźhÄãû]•‰øaª‡âFI‚;Ìò_Ôßò„ù²ß!’—ûa©½ÕŽbÊ~%G¬Áï•ôrf°Ú%Ó†olaCãßõå{°ù;3e`älwôö©Ó ÑÉ¿Jÿ!ñöÍQòJa$m 𪺓„ä³&é’.H¥»¿2†'i{ í]Þ™(ëñø'dj/=f…L5‡T/ݹƒ,¿÷¹ÅÞE¶”3b–¥–ȱRL½ÑêûÝC}gØÙñ ù(Fx§ oñ«ý¹Çȯ;\äa ~wæHËG¡(¶ä…›OQ$ño¯Ô%£N=Ç—Í„‡û¸·™ÎLã‘íŽÚ*Î(-qêÁ€ÖU”"³}g‡ráíÔ}Ÿ5P>²}3•W§'Éîà #—a½ûò¨Ò¿¼4ò½UuŠƒÇî¼CµEn?™(ÔH!ÔMö>D­„´{6{Þ˜opìCˆ êtG.Æ £î¶NMÚ\íÙ±HFCª™.KsC4jòô»„gÓ›~Xµg¡ÑÓð[gVìÑø Ó³ V{4q”©Z‹>…¦. ÿf¢ÉÐÌŸçŸdv>š§VwÙú‹ìmüÑò4Ù"F´úªÞ:à‡66ßø¤‹Ëþ·òÆ @ÐcO?Òò ’G޹ýWÅu¿Y˜z½ Ÿw`'i…®ú1JˆZ®_z$FQ6wÚ¯ôÒCÄõ‹§G!,ו—©lB_IŠˆtƒP¾ïÏŽ5C¸¦ëû ê}ð~=¢vŸ¢=ÄYdëóÜ3ƒ¤å-c.ÒÞ8w pßEÈ꿯 ÷÷øæQ”} rZžõê~©{Ž3 ä<:šœ- „æ#S’¤³ò’l¤è+ä¸vÄÔT]„J–ÜDÿlÈnw§;¥( ÙÉZD‡ç[ {¾¼|jƒ\ÇâÕi çKÀAY: H¼}‘}rÏùaÙËZÈ9,qø+)ÅG ‹8øCuËBEÖmÊà‚?7Ê©N˜ÂÚÍÄAŠj_ØpíTu7€M[q«¶7øç4Ü}ôì¼®´`ÌÔG¢‡óÃgÏ_EbJSûÙsH:À[›b:ŠŽä£,AHå«^$Æ´\r)fß"‘¾?{­DdMÿU‰l S‡Tœßƒ@d±¶õÓ<~ÙX›T(•9C $)åI6råÝÍãý¹»¯ÅL”Pìù眖üS-`{ƒB*ÞOn’ÿA‘xVW,3¼³F(ñé›ÐZ2”:¼«¦âZ‚2Òá—æ¶¢œSyMy1Êm‰Ï£RÃiÊ;‚Û¨r€q`ÖMUõväøS^£:£±ç¥#¨1ô¢Ø#\µjVË ‰¨{§@ñF&êÜÚ2ÛwOu5p}ž´@}ŠÖî’‹h0Ã¥\"³†ÞÃ_»ö<3ïÉ·ñv4ª~¤ÍµÊ&L¼š’~¡©ŸMºtæ&šÓØ~Z® C‹ º»Q¥hWùëÛ…Oh]Zxþò/ø3÷Œ}¾¾òñŽ¿ê,¤?[–¯Ù5Üó=R)5B%$<·Ý‰ƒ¸OS©Ï+ ÎfI#ØÀâˆ6|§‰/À§·í¾4¥R³?áAATr®©”,¢ÝÅmwV¨h;ùá|a.|2· ”~‚Úc ¹POÒ„Ððöò»Ë¿F¡‰eÿÂR«:4½Jl ¦­€&9ˆï,憆1p·x õ=Û$ô%P7wsyO  î¾§çÂÃP ZëÙõê¶mÍ* >k+O! ÁàÊ“{µÛÐèhË0 MOL… Ö ÅOÜ7é5´=I S,Pƒöf%Çë"îÐÑ)’8á[­Áï" -¡+«¤¢¼–ºs•®õVOCÏò—`ë$qøfQÄܺ }»ù‹âR0H÷|]Þ7~„3­¤¼í†aËr1r^#e»@»@à cCŽM¬”~0‘ʸ-40’và¾&Ì4PïÿÞ« óú ßžÁ’z7µKŒü2¶ˆ1¤‡ßÇ]¿Ž…ÕóâK÷êÖ`½¼A‰¨Ò6­¿Êa[ȃ²©5‰HÈÖ¿–¾¹´šù™œ…w …‚‘åh¿ê ±!²»M¨AÄ3¯gØ‘kôAØ/ ¹oØÌéf /?¡uDù¶fŸÖ©w¢@@¨–Íœ i\ùõõ-Ù¦Âsy䨵ח­h^”œü¾×ãžþ Í⃅°àð÷ºéãD[¸heIÕ«÷²’AmÊckÅJø ìt‡gé+¼·z°Pú¢Þ\è>§YQ†£o©Äd!"tsW ¸Â†Þ ¹g¹BèçižýX=ðˆ}ußOç./½ëïõR'WïC¿Jh›Ú<$•‰$¯L @zú)–{ÇoBV{¼jX¡Ñߢâ®È@N{È»×ã¡@Ð÷VÞH‰B®¼“êK. <ý<¡ïš9#Wt›vcò€4[Ô4ú´­ßg$ æDÐÞ3vÈL™:]ùÄè¹¹¡›ÍÚ“•~ zº7“k¨®Â·#×}5ÐÿZÎèu¼|§´¯;t£ †xÊøÙ7ȯî{SÂŒ¸FÿÐO‰¢‰­(0µ±RKúfÝò5¹Yöú›Ã2CR‘ ,;{ -‘—ÁŠ®¾H×<ø£áÔ8Vßk'œø5R bÊsÊ‘°y`,Ólþ‰”äzb€ó±­:&Hä²”•ƒÄ“7äïné!iiN¶ÏX’OZùú‰T¦1H[EŽ4Ë|A‡þË—¥ö}ÖȈÑ.×K"©øî%¦‹ˆ,Úç?ø(ìGÖ.ZûYä ÊYæà‹B®°ì§ëÜ·‘;½©^ÉUùØ9 eÃN"ÿîâ‰}.($: Ù§’"Áþúå(¦Ñ9L:‰#ÙÜM(ÅH]´Ce3¨2A9»‹‰|(ÿSÙ`®æ*µ:vØPEÄ­ñÍÌ,ªšÞ{ÁÓ Õ¹:T®ÉÙ ÆÔ\_¥…j5ÙMU,)£vÜ×­§t*¨sã≉Sݨ+o½H=ã‡z 54G%äРò¬nDž5ZGäø áY÷Ñ7öÁÏÑèM÷Šw¨/w³çVZ÷¢©ôIö»«h–¡:ö9L-ŒâhÕ %¿«ÑÉØp´4y¾Hj¬cµ’Ô!wr^çq°|Iî<õBy$yùäž"K€ÖnbÛ ÎT 9g™âdbX#|:!vå€1½e|ŠvYxý–b€Pú±>"Ú—V?ÆkADÂþƒûDþÂÇ>uoˆw EU_K7ÝsÏëÜ8®‹ BÅ•ã•%Paõ´Eº¹¾Ú{çÝ„šŒÊ)É~¨{Êñ´lßÇôçÔCcâ¿w~ RÐ$ªkÈì<U›³ðË\ßÇ_º§õ¦W…Œ„Í î»ÁšêH¯ª^ÖÔ€:Ú°+Žç Î?ôÛEƒI¨—'RËr|õÒ^VQ_¡aÝÖo'™š<Ö«o¼¹-ly£TÂªÐÆ“‹!MС’__½ ï_$2tÿ…Îð«o&G@—÷b{¹’tßÖ/ˆ‰¿=¹´ÿ˜¸DáçC;kRèË=l/ì¨ ¹å·>ؽ†rã¯.s>„a‘ߢR[þ02’h×$c©Í÷•J`ƒò}­%L¶,ðµ93×Òfƒ³Haîû4ÙýѰHèÏn4} ËÛl ^ÂJSl¡@ÇžŸî+ÿ÷ÖO„ñ´„¿Õk™Á1àŸ×[År+ ØÅ‚¯‡½Ž#±ÈõŒ_ÒH2}ëbT¤6’­/ie»"¥öý|9Šp¤Îì7ÔÞwéä>+·¬#Ãúc[‘5d¢,“kŠAæÚr];d­HcdhEö•Wo%IS‘Ëâî‚ýØ <0Üö·•€$5>!qbôW ´Z@V“ãÀߪCŽáät½=wúÎY œ]áì¿Ñ„¼éèoWè௕“hµÇ?½à¥Èb¢”£¢=žG™æÙ-ϵ»‰Ü÷®qúš‡ GfÆ4V`r<§z§"¸äíå>B›•ù+»tÈMŸìºM ‡#Zo«ÈC1—gÈB?”Y~œdÝ€Šš]órÅ=¾žXa½Õ·YMTX f3›~[„êZJîÚØ3@C¯ëíwþ ™æc*+´¾˜“&ƒöŒ÷G*­É S:è.tçøÖ>€ž7­3•ÏÓà›|šGéð8ôóŒ:¼ìý ƒwþå~›Ÿ!¬4çæ_Áˆü©ów”MaLO›hŠm&Sj£(‚©š<]Å„˜=– ûÌv¯¿‰c¬å“ÑCE}}°"˜Z3¨hX˜ž¤…5›èï»Ì°>Nî@7ûß“ýñ‭Éwß9í¦aç Ň©Q+$Ò_ ž¤lEâ¦!Žh:@ÒOÚ1:×ß!y­¯”RIýýºxÁiêèÜ/_BúûIÓcëeÈ(¦ÜÎo¤„L Í2‰ÚEÆ®*]6Y§úUž–"ûbªÂâGÈå5v§­â r?ùõ‚]ùˆ5%¥N!ÿ¬áùÏÒ(ĬLªìyEn ÌËb9…WƒQâÆ<6=?ŒRìg5˜;Q†çÏd6ÑÊYÒ>y2€òßïxÌM¢Rç¡ áp7T‘®.Rq @U[¢InÊ>Tº¸ç‹Y ÄQÙ|þalq”z± §!v ËÔáS±ý~÷Nˆqøº°Úo¹®æ¶BD͵•ºð1˜'„æÚ ¤PiHÞ:ÙUaW?B¡ƒ_7ÇE(o{µFYUƒa— 2NAøáp“mã=.9¶õ?Ý·×·Zâµ<¿@ãñŸZT<ÐX²ôí@54žÍ?š*zê—úiÜú ®{ädØ·e¨Í¯sÏ…ÚÇ®´ÖmVP<Ï~¶r/¤Mæ·~¸B]YËa¾¨¿øÇIáZ 4¹™Ï5ÈB­«lååyhÎÐÐq.ô†Ö¼ßÉ!)h¿ NÇçx:\þ$(•N;Ã`YZRèRå.KK‡n5"§‡ ÇÇ͵&PzûôÒÅRé ï’8u´- ØþÚ :y¾wæK’…¡IƒÓ…–0òe:·‚ÎƼ)œ÷P„ Å¡µä[0EN6Яøf8 ‹Tƒ`îÒ(.¸‹"CÖtXNìt}êË +W ï‡ìßn"’‚µ|‡:l:%ßå¸ÜÙËQ}köÔìÜvk»±—£¢ÁßäÔ Hâ¹øƒêP8’Ý-‰“{Š5¹Ò;›H-dð¢Ê‰i#sb9ÕãÁà–¡…²2¦òßáòÈ@f›„¸â²?Ȫ;ðÐ¥_Ù=üêxT‘³õ¦HŒN 00UOñ)ƒ [”¯UÏjJª’ùGɈg·Q0”ß¿‰¨…ƒkÓòÜDÑ¡{¾¤U(õ0Q¿±%³"¹]™»Qê·sÊ(Y‹ÐJ(£ìMò«ô6”gPö¬Tt¬q¡.QÂã ‡ÒH²£ŠöÕ¡¯z–ˆ[WµOâPíu‹äñg¨á9o±úîj…©K_8ƒÚ#‚L'nèàiÍ¿k~a%¨—_}O<Ìψ¾×ÔäD×ÍËZËÒxÖüsò½¿QhDZ±9/"FÙ»ç Ò›wZˆßûÞt0Mþ’\ôÊ@ÓºïÁ–ôßÑì³|ÜΔš¿qé\}‚Ááí±xî`¸­Zæý Ò> ³£)ßܸâ!ëÝjÇe¯nH|î™Böz|òùFÐztÞSæ.‘U3CÔÏ.ÇŒ÷ÙCêÇQ i<ÇÛü!â»°Å{9„ËnÙ „™Ch3B»v›«‰ ¬åÒ¿-Ç`xOJ¸vµb³M„¤êØ!éܦ 2H÷$!¥}ù²jË5ÒWöø¶às‡© rz£¸HØ€`véO‘Ê Dv]Ñ#‚w—×䔋wYùDAÎ!άè‹O!»ë©TÙ·²›¹/¾ÙC7ÚOÓ¿wÞ¨6@N<ÿ+‡¡=Nž¡xìtäî[)x£0 y²§ŽÛ,óCÇ™±›†Pl<6%ÿ“ÊÒ8œxr*ÚôκîñmUójU4T{Dn št@ÍXf#OVÔÅú¾§»¶ a俬•½ÐTrèÆAZaJô5-åÿ쓸½ œÿ8È7G ‹@sGöôʼn•1Î@ïOŠÛ 3{þ™úÒ#„:V3e4NìÀ]׿ ·Ì`xE‹?þk9Œ1 ]Xþ© g™‘4LÂÔkŸö«y‘0KôÆh-é Ì/e+t–„eáÊÈ'©+ðëÍ…Ñ¿à÷h®…XH+¬6(n×sÛÂzò/ž•Òð7ÜìÜ“©S°•x‹¤È§¶GÅ4ºn;"‘̪“”ÕgÞ(<(…¤9wo"yâ™°Ò²HEgzªƒâ'Ò¼ßÐ÷NqEzÓ¸‘õa~ddþ¡Ð>5€L/¹6Oˆ'!ËAû‰1)díýlñ%Ù»cï.›\A.sÖI{\äv|°ø«òö¨[¹¼Bþî¬&A¢-ü÷¾òãI±Ï°`Ï#A1ÚÖ篡„õZ‹¯^JÑújüûh‹2lqñlh‡rÆYÎD&³(?P¶p¶•z#¸}:ö1h¾áÉaC€z/Ök®Í¾Pgbì2µnçdbî¸@-¥C¯žå ¨eV+Ãü ¨}Jü-ðÒ<ÔµEüO*¨û•oÆ¢e ¼^2}3ç 1n nß)gh>Õ–Y§Ÿ­=y*+Ÿ¡]ê¤7 tÈYÅÛ?»bÒ÷dû ‹Òh¹Œ º)½Nqô…•úo>[¼ÐÁ¶µÎÅ}ü‡rÞôxÁ•÷u‰ÓðýÚ³Ö•ƒc0t7O›É™#¾‡i’_\†1Í÷ EÃa|÷OÁÐóP˜,OßVdm„éò?ÏVœ^ÂìæwŸž[Xˆ˜ ñµ – CÎð'„ ½ÝE&=øCÊb¨z±ÖdŸ7”‡0åWO¢¢ÂEÙy‘Ox<êÒï?L´¨Ây´[ ±¾r;ý|=ªù~gdžêC &²›«QëÕÁw-ZïP{‰5£º\ O»=)OÖD½Õçä6<ã úºÙ Ë8ãC½ýðì½(Û#‚fhÄÇœq=‚*o™†£±«Vaç<šp¼{Éô[MZÈ”ÜTÐ4ð:«Kï4šZ‡.k¢9o¿ŸÓõ4ߤÔd¿q-†¯»cõØ#ÈéÿØW˜Þˉö(  éÓ‹­áº+•?o6ä¤Ss)T„f)‡à¨é=Oµ8W›Æ ÙiÏ×§Bö q¤µë^ô¡Nxd›9Öäú¬7€ðh<áØWÈ5 äz»Oòn¼˜¾ˆPPHq­1ß Šƒ`ŠRÊRoü¼àM ýÁÔªc‘P#­·p­ªÝøã|ÂöCMsÆU娻¥X\ìß ¦Ÿd^’¾‚¦ËýuÃ¥jв4)ä?AíŽL'£֠cy–ÄMºÚß×PH‚¦[®EoE ÷±Ýヿø¡O¼P–÷oÎíYáÇÕòi6’ý0œ®6Ö¬+£ÝÄõ›‰A0Á¥óPþ LW1M¬»3M÷-KÀ|‹yi,“MšÝ£…_Ý®¤³•Áð;Û:¦ù†¬¾dµÍQJ†u7]Ň_îÀ_Ó}é\>°åd"fý:¶?ëN•³º!džŠoæ$Ëëï="ˆ¤.W=§_œAò':ÜÖzÝH9•Ù’“¬‹47Ç<\þ‚ôb1ƒ´õq߯mþA&_çÊvzd¡r¤æ”XFÖ’YÏ]È^ô®Özѹ´\ïSn ·ÊM¹k÷7㸸é“Pä/g‘ñ©AÁÉ]²ñ9Ñ}øuKÚý™qu?ù%N[‰ÑŸE)¢8Í)Ê”¡3²÷¾3rúæÚ׸6Q¾o·h‘£•)½[izQå”÷»—.Õ¨z•4‚ýâAT—?ýöž¶jSH}‰üZƒ±ç'êÉQûsXGÏ8ê\Œ]pJ(C]†…KGW&P/ösÊÓ=þèüõ>¢ê…†”惆[N;A¥hD\û'Ôã soÓÜ}‚&gvÏ8~œ@Ó÷SIÍœ¾hNãºÿØÐa´ˆ<#ÔÚ-MDØ®·‡•/¼š†' çßÊ—·^QJí}gtEoüдüsâ¾p¦åÖ‡Bì•®x)®ˆÍLyš÷ï8ÄÖ<¼ãÌ Ÿ2eN‰sß„˜”Ñ+çÎ,Ã;í!{ú¾!xGÁ-½C\þ5(‡äˆãõN¡ ͺÄÇÚw ö]©Êƒ2¢°/†´ß¡²­çþôÇE¨æÙYÓ’ƒZ–vûƒöúV«d÷_ohpøÍײ ÓŠGüô áêÑ¢‡o¦ îß™^—ß ¶«&ˆOEjf6cX‚÷òðÌúÓ‰‰ ¨yî$™žµ4ŠæÑ›W¡¶Š~œ'Jêîøj-íg‡úœôµÔþШ¬Çl#0M£Öœ"#SÐ29ñë-/´m¼uý &í»UqcñбÂü É: :[•¾9]-!kDÐC"ªuÄýôšéT?åoßD­›‚¡?Ëlc€Y¾3X_&||C<®| 0ròòù7 Ï`Œ‰".åû*Œ×ÎéÕ_Í„Ißs¼a˱0íq.^Í fŸõîͪÂÍlú’ï ,ý‰p½{~%Pã ª@øB{Í?ÒVÓÿ™Þí‚õ¡Bî㪰é¾ÓIÌÿVj€ãr ìÖ…}¶õx„Ä/4hÉ#éss[Åf $¯â ½‚T´JT…,íHãÜÞvxéþYÉÕ⾟J«5÷c‘)ȃ"Ú/YܸçÒÍÃøâ–±rÙ¾ã`ÿ…ûÅô"iÛª‘»HÍó±òÞªó ½{ùOÜ(óïEñnFjÛ‹(Ômš«Oœù/C^ X½Ge“)Jtó¦,}:ˆRŒÉÇké£tã‚Ï¡1” ÊJO-‰Ç2”$5@EZ²ÝÖWTxürZÁUbTž¨¾—>ωÆš›¤¸ˆjŽäto¡†Q Ãîõ·¨ôwØUdµ·) hÕñôK–€A¥&Ô—c7£bÃ3-rY#&…h8wÁbŒ¨Ï~aþ®ÏˆFÊש»Ð¨méÌÎ}U4vaþ:W‚&d9 ”h’¨FAoÓŒ¦úä‘·>üÙóÈöŸC³´Ó¢1‹dhîD¹lM'Œ¢…7oW¢ÅÊ1CïåT˜>RÇ!? QåýB¡o¼Úmvr?|h§®µùH‘]äv³vé xŸ¡ñ3Dìvþ’ó¢xívÛ ïªFSÎÁu§lÒÚr¢§lˆ7z!ûß)ÿÎS%ö 5o²”? §5B™TÅô3õ½s\-ž~A@ óQСßãdq| ä¾â÷ ›´†¼jß(Ø-U[¹ùŠë+Â¥젬ë‡å ×gP1æ|̳qo—M/ñü‚jû$]ŽoŽP“›®n_êt×Ü^æB› ­–4ñ¨¯µ€–üWË ]¹I¤x:êûÈvú¡Ë¨à±Ñò èn×MÎçr„^‰¢`ý¿ªð­zþuЕïÐ?¶óÞ–L ¾³õ®<‚a7HøÝùFÇ$É®ôÀø˜[¦›!”ÑÀ>S˜ »jß¡’öš9ïp®ß÷ï¿=2ʦTÈÞé /ÔT£ÜçˆZþ/6®¢GÚŸº÷ê%šÐ^3Eƒã¨”®6…FLñœã~hÌ•Ç ÈŠÆÏ/Õ_õlB‹b®³gÒФà¡D4uêî_ ƒ¦_Ÿ}Oˆ¯G³óvÔƒÆ{Ñœ)FÀß…€æIéž-ín¸[éa{÷!MÜÝN{'e÷¥´)œE ¦ìIB0 wv)òœç,Ôè Y±èÃy~ˆ…ií÷àVIÎÚÞKp+¹)õ…8ܪr׿ô\n³¼v¥j˜‚[Úïú}†báfAÐY·2ˆ\þI¡¹"M½Ì&òµûˆµBÄ2µnæMã‡$ºé9MÆ.H{lù];îñsÙ°<„üâý ;ø–Î7g%³+,PØŸm±i@ îx`=÷Ëó½dÞ÷Âýý'õo­†Âkµwº”< `ضƃ& f³^T?†‚úgCÆ ÃväDzVnžvŠq…û­ŸÏ¥Áý‰°Š;nBQ[É›è¥ï kä Ï@©ö ¶uÒqðˆBb|‡eTÌk¿(¶b‚ªÏêl¢ÔsKç[ƾÑf¨±”˜,æÊ†Úøì_—Mû ^àþÌÚÄhhè|ùÁÿ 5d¶Ðm“xÇ^YmhÿèŒÐct¸¶•3D¶:s[.ÀÕ£PFEx—s­7žÍ†ÖØ­ù ì÷¹@ÀÒ÷›F—÷³CNðÅß6²S÷,|õQŸÕ]ZÆYâ¬*;`š·I ð„:|[<œº‹~¶^Ú*=0;Ìè™Xc Ò¡’QÚ§`ñÉÜK¾|$³Že6¢A É›W2–îÚªo¹œBêøƒÝÞãë‘V¨l"æœ"Ò·E4ËÞÒA¦„³½•ÆžÈbHa4Ó^‡¬}ކ—DDqMÈ,K&?5rø˜+‰"×4sÃ;È=õæiò[}œÝù8Ã_0D¶Dá²ÇóÍfQ„_üvÍ&ÜTÏ/ÌÏ'Šâ\žþÁ^(Ù5äu€Ò3¹KÎE9µdÔu¨Ðu‰üKò=T~;,í飉ÛlŽß:Pí•.ó㨱S{C 3ngy|X§µ†½sY\ÙÂúÌT¿p§µg)«2%ê~Kf›šC}÷v7ú¥{ç¬KäÈ4Œ·“;ùsåV´³ôË¢q¢³žq¾š$}üœä`Œ¦¥­üóÑoÐl$[\¹¦wKÞt&ŸX‡{"ïÉ鈖üÖîGlàkºrƒÖkI(è,/W€ i±YãHÉýQçIbÁ²ô·Aâ@”mž» Ü¡ t;%TŸW%B¢Ùå¼,fˆm)ËŒœ„h%ò×»ý!Ú:ìLÑæHÔ Õ«ÎØ’[çz%k ¿ŠÞ­ª1\øzêü$”‹ÖKnŒ€jƒvO©TxÙzžÚj¯ opü¶t~i62Ꟁ†f×Àå·Ð¨¼óŠã]hxÆ1ú9ê“û:½¡n§|ôÉXY¨u¾©àTMµ46ò”¦O¡–6ªdç1C¨=æ0“u6êx5òÉNÚA]ýÉH~Q+¨Çݘ·±¼èöxÀÓÞGwÝ4àY Kjr_:´4>Œ=®­ŽyÔS2¡ÍZ¶h•©´k[kìö‚ó†V[Ò 32õÖN²èšú~6hßkè>‹Í ±¯á•£í]¦®&xÓq‹ÒÑ¡o$õiP n‰­¡|«CŸƒm¶.ž‡wa¾*Â0,ùhÍv?%ÌnÝàq ÆÈæ£)ià“cÍ-|YcJ™ïÁ_9u…e×¼‚É>³Ö[[á[ÏMëcMÛà]•ŒiߘÙ]Óºû„ÌÞ_üö–~™M¸Ô¿]DBÉå›gÂ/#ysJBs(RÆç·¼lÌDª*s¹Nq¤™xʯ…tvíöÞÚȸf¾æNÎ82í(ÊEV]*«.¸FƒJaÀërœŒnNoE®N–1‹ äÞs*d2ë2ò­žîjÒþ†çGín@¡ ²Úgi³q=3ÅWŽ5T(²ëK¾b¨núInB6Hbs•ëV w¢$¬Í&Ei#eióÀ>”eéã®x‰òµoüPñG™Á–iTþ@p«°–BÕSª¢çºìu¸KÔ¾ìAuý“¢ì¡¦¾i”üµ0Ô ¯ß¢)üu˜ÌøN¯9ˆº™t7´ù³ÑÀ$ž¼|j?¾3¶î‘¡Xú~óv¥ûZ„Æa-ÎbhÂ"·Ýd+š&ÙÞsIBS Çti-hêCöñ׆r4ñæÔ'³D³“¡ç]ÙrÑlRy`J,Íš]/*«FóOfV»q÷!ŸÎ=¶ßp÷¨·ÊµJ6`¿Ôx5 r´¬ƒÞϽ†¤¯ò§3…Í nxê|CÜ:q%W¨nm2™*Ù»¢ßJw~ ·*OK]Ù  7_wPGGôCÔéC=—Œ!BÿbPŸ[9„ǰ]iW¨}ç}¥÷ÎBlKB’ó+ßÏŠ› mÊëàÅãk!§áØ“|zÈ-RùL–Šì–ÊPª±æÞ&(xæ ÐÀM…ä²ÖèÑ@AUš–[JHÚÄe¯mü¾¶ÃÖ±ÿ¡˜kÞd)_':ÙF€ Lʽȅ¡aÂü/ pêz¶ÿÏv¸_ù¦|B«Šó½nO†AqÜ ³’Á{Pò@6bíyE(Û_}Id4 *¬\iªÙ• *åV‚8·<¶åIоª5‚qŸ*ÞAíÙÞ:¾BPÏïDís9¦z6­«y OçÙh®OA³Ý½˜^B+í«ñ]['¡-P¦ð]E+t0ž¬ÚÁ¡Ç·}{=]_+vÒm)†nÿæÞµð*´;Øgçè±éîÚÜh TífµâaˆáØÃ1³x'û3üú½0lQ+º©# Fc£×öG~‚VΜ2ú[aü2ÃÑŽüq˜˜Sï]ó&k O¹ØšÁ·P÷÷O}à»]ŸŽ5Ý4üÔ*¯ퟂYcí°í@ ó×Â)za‘q¿Þ µV$#ó¡éiDògûª­Û‘ò­ ÷ÜG¤V3/Þd«Š«6(ný‰ôG¶&Õ~C¦Í[Ùz&N#Ë*êâ/d6Èêgòù»®<®‘Ô󩼌›ùäG÷ !Wz gñr—·Õ»T ¿ü!õð†xô×UŽ4#Cá,)§êY!/µç<š€›¢ò«åҢس‘ªGJwQÒ}ḛ̂Èv”®<~›¢…Nç ½TTˆ>=•¹€Ê™kÈ~6à¶-Nó£w~ÿ~ò‰ÅÞ_¨¡5r×Hàv)ÜÊQ…Ú¦ÂÅPG¤÷sþ˜(êÊ7>RA}ÝÓŠIì¸Ë¦ÏWr; é› ¼¤BãKÔ^܉ÊhÒGöèô šmß_8|?Í_ni/HÁ=á37¶ì4CËs»cö×p UL»¿±îûÖq4âW(Ú鱞<¾Y ¾ÊŒW ÍCá›)¿'H¿BÝ {ÂR¨é™tbáŽ`—A¿¼$€ª c=5$p†ŸP# ’‡¯Í\<°´.7µyR1-Çï±Ý„›„‹’U›RáfX®¿hp$Œ29lØÜé¿4-¦—¾Óó«(ˆ‚qg]øB¹êI‹^)¨6Έ­€'argdSã¡öÜBRÈëx¨õþ´–ò4¤˜7ÝvìƒFy1Ù 1ê?äXø¡^çÑûrP‡²—ýuÒ¡ö`Øíou¡v•ãB¯Â$ÔÌ«í÷Ú&µæVÜiöÛñÉCa2Pç—óÍÃñ%Ô_ÈÈÌÙ0ºt^JE†çÀS i6c΂g¼[ø¬’K¡…Nvÿ˜“´\Û輡uÁãÓ”À(´ý ;AS§_ê~*5@§Cíù\‘cÐÕºÝ> î:t¶ŒTd„Wîòí=…ðfŽ|Þ´Yõ÷ÿézß©ñu;‚ú`h!gšk¼»¹qÕã©=1–02Y¢ÿíi-|ø¸Iý.å8|R¬ºO ãáÅŸFabüí&óÍc0yç.Ë–ô"øì«¥ê ßsÒó“_ÁÏñƒ)ü{çaNiz\ÂÊ J÷6]Ùp &ù3LÏ#9ý|oî¤X`›3~ô ©ÈæÍä·¥# OUÔ‡¸wH{dO,åËNd ®a¿ ™&ޏ»Ow#ËÝ»…ñùCÈÚüXì"9²?ìm›êäC.+»E.\|y= òªn ¯ 5AÎäÍo¯îBÁreyÀ-(œ>–!‚F¯ÕÛ73â¦Ðuf§Îz¡Xwu¿­J<3}Ʊº¥éá”Y2ÊTº¼þTo†r•o“üøï£"[@æ»Ó¨ƶï~ÿaTÝè·{¼Q¼•¼ÿY ª<Ò‹¹H@Í=—5œJ+Q+±Dö" [žO[êÆ{>»xŸ dæŸu¤¶£azõ¹ÿy4ªP{:¼×-“º´^EãÆ÷³ÇÑD®×k«s š$Œ}‰sASZ…Aëìûhê"Rœo¦oúŒ+lÐL½IÓ7ÍRǵ#Í©On§f©CsÇ"ºq4¯Ê ¼Ì1ïkÍ[NÖ=„ì2&”YD×%q7òhuŸƒ[R¦«/<¡‚èì×ê×­y zÕ×PqšÓí¶všú¢/D©–,Ø!‘šû×ûE@øù£’þ e®½sÑD ¢B²Å~Œ@A-SÀË’jËoÚž…ôõ/_²‚œcOã(;m /¶^âaä;¦Ë¿³ƒ|î6ÙÈ—/'ùÚEÓy²Ï÷=8OòÚܺ4' yÝš¬ÚR/{Ì_phä7:\iÍUZ:'}ϲ̆Â3ý©jPX¢sø“Ç/¸c}X†æ (RÏdKØÅ´'Nx AÉÆ)v®gíðð±•W”? Uiz¾ª¬&Å$$Šà1ÿ&ðdŒ~¸¾p-ÔZ1=Œq„z>Nö·NuÐÈÚï”Kž M¤‹ít½¡ùè)ÎP•Dxþkߦ}ŠóÐv,áp‹\4´/Tïç}oá5a \jð‚éàlô^zè~DF‘w^5겈…A›æ»ËgÖAKû¥I¬yWwÜÞ>tÜyFRÞ—Ô~S-Ÿ…‘îö§’³aÌc¶(Z‚>Ÿ/ÛÙ±Ö&ŽisVö•äbY]•¯|£ÔªÖxÓ/Ê7°6vÁ‡3 á~´0óÈ'gäA=Ì(þ­Ö€_ ­¾H¸ãv2CCÉ1¹Í1Ç#å…(f×v¤ÒŒèn5ÃÕnÁ>M³¥HÏ ›É2™€Œ3;goø s=Ï;ÉidW»~舲5Ç~ìüÉìM÷9b’¸‘˲axO¹Ï§/¤yŒüŒ£½-)(è]ᩚ‰Â1qmâeÁ¸aX[äPñ'Üä´ÝVW*Å.–Ò~ZºoÒÍ2‹Di?C!U²`”­Ol™’B«s? ôQÙ]Ÿ1ÈCUçîœOrÑ@52Õ„©hÔИüÖi€ÛUâÙĘ PÛnËÍ#[v Nrô ÓfsÔcZ·÷ú”;$[I“£‘HÄ)ïò$4¾íþýÑ!14•ëë˜ßnˆf?ÎH丱t?ÜÇs¦„-œì(?ÐêˆÐ¯}3ÏІŒæÚO´»LõÔ÷• îßtšŽ½[Õ Ÿ8õD¼O,ÀxvgÊ#§Ç0Qwv‹̤Ãcöø¦&ÛÕßͶò3¹XÁOZ¯Qõ;0[¡Ðs°#xk·h;õÁ⫪ó÷D~"YਧK¨m=Rk~¾µšÇWŸéy nöšSý! d¬3pÜ‹,;ø Bö"ë`¤çÕËÈNj{²x9w¥Œ¯fG\[›|Ò<$yŽX½ÉC~›òÏgÒPÐ× ¯º‡Âêmv¬’ܸáÒÆÞë4=¸ITf÷WÎ!ÛäíR3ƒ{ƒ>‰eD©FÜÊïŒ2®óÂåÏÛPΖìÌžoQ!g6Æ/œ€Ê[E ²ößEU²R42ˆ,¾>=¨nHáê¸5÷Œßù„Z9Ê•{ûP‡‹þbep)ê† ­.×±E6«¨êhè&/÷fBWo®z<‡Æ|=ë-Ý¥n/”¾Dã®à†ƒwÑDÕý*{8%šD~Q&›‰&ŸÓíw£©ú…mŠ.Œhj]´µ·M{ÇmÊÑLHìfiƒ<šÙ{”³>ø€f·8~4% ÀûmQë„ÈŸA¶Tu7™‡7$9ÈÅ6W@œ’kijGï ú“™É-ˆ–¶¿agÃ77È^Hå›ëJK¯ô>ƒHòÜ=M3NA÷ÌI"ë„X·mÉ)…°×><Ûµ!êðN‘ˆNˆóö»ÓïÉB‰ÑêWžCºy.EXäp {ZÈçC›M¶P€:äk‹Z€ÜÑ7ÜÇ»ò Wej#€?är¾'úîÍÜj²õŠ€{ŸE Ké 7Áå¸màsÈ;'>µãxäëJ¹jÂÒù8ÇÑä(³ zìšbE3 ð¥˜ñ—Íp߯íà~õi¸?“”›dE~RTìð º Šk+<Φ†òµÅ®/¯C›î) V¨ ûIyžœh½`´´~Ñ|÷¶t¨çiX¯³³¥‡R5ÕÎ@“¡¸e õ/höÑŒ½vžPàýÉ m{é…dÚBûǘ]g.ÇBgǰ—ù x±…%{ƒÿ~èþ*'·]ni}.n]•  =gr§zÅ» ?„`­ÀRƒî=ŒI7>À[µ¶E—rxÏUWa”¾FØÎæ†׃L1 ç³.Âí:áKKö*-ã£ð5³f g†Ly]f‰÷œiýX²Æ+EðCJî.ˆ ÂÌÆâÆkÒ¾0§AgEîs’%eøí‘ n=t²ÌÉ©õ|äj\’÷ eÕøy¤ò•ö¸¢î†«YÔ¾+=tY\ùu‘QõÁ.‰KÈ|ŒùÒ+›ëÈÒ!!Ú¦‹l‡Ç-ÆâýÈÎúCk‹öUîß9äÖÜŸV‡|£N­;³Pð°Lhä¾J·Õú¨ nhlÿXþK7É¿X°-@1…«æƒÈQ"m•ÅŽ”Öã«UñíAÙ€7Ƙ2Tà;]3Wz•|¸>·¡jcÙ‡·¿ÿ¾»häº j ¨ef„+n7¤á¦MAí8:ÎTÜ)`*5wõF]Åtîâ®ï›|~éFcÞÒ¼W¯ÔÑdì}bó 4)–êuÀ=Ü–ö,ÛÐòú?æ2o´ÖH¾%k¨‹¶:únÁ™Zhof^¶×Þ(¥˜×Ö¢£eðä§Ôx°úõ—Q_˜Ð±(àžŽ\º¯q¹¹;ÔÀÝwE„±‡d¤U³[Ò€âyê©iÚ!FÉJòPøq¸µþ„ÂB± D Ì˹˜ Ñ6¥-ãßCTóTÁûJˆ=÷:Ùi"}¸Ínyâó(œ_ªBz¬ ›ÈÚvȯíß%ðei~×È0/΄rÛ¶ô'¨¶4(Ü÷¶XõŒŸk¬(3¡NLÒ2íNÔ:”“m¬‡šÅ“7‚Ô ¦?¼¸½sjywØ&ì)…Ú¤Ž“ªMPGÎWÈݳdKk]Ó„6Cý“䌳ŽÐH¡Ì-˜¥O˘š¶ôÀ3¿†Bÿ]îÐrÈßíMÁhÅoÖ-vA›…^·™_+´Ç¨ËÒB'Å*ý3G› +`@ìüÍIèÞ´Þ+Ö^ù ‰{Ý=;_|’ò’[Z?«¯äœ‚AŸ“q}\ð–÷ãaŸð.ÑU´3béûøšTž2Œ¼s?åÊÏGQ•ÁÇi—ÃRòßaüû‰ê8˜ˆg—¹eF“›tBc.]€©IrîO|0ýÓ;ÒM¢~òæèåoÒ†YM¯§wô>Àü%É«Ò `‘‘wÓê»_‘ŒªÓv—5’—¾4ÓL¾‚”~ºû˜êðºÛ›n¯GšÀý,¼É_‘ö³Éƒ×:®È+îÞ ×ƒÌ¶#uG:UÚúeu%®Ñ:Ø!Z‚ׯÏxM9!×|ÉÁ+[;ñS–òËa*@ÏZ¬ß/K¢ÐëÎÕTÜÀýdªx{n|¢ké\¨¢],×^\Ç-£ ôê;W¡”ßã}c(³.Ø5ø¥Êþâ7ۜΊ :®WkýPéò¶;ïpkÝú­«c¾+Eó^áGu#Ý Ý•¨é6ÓȱµÆà‚çÔáq.ܯ7„ºW ´ö” þÔC»EÚ<4”¾S1(5ŽFÚn‹Å£ÑÛÇoÉuÑXå^}Ç)@ãk¦wÍÒÑøåwÞ˜€D4J|¶öá’=X¥uÅMÒ¼byÙÊÑd`œßNá5š² >4ÊÛŠ¦ªƒ]}RÐÔz\Ú}é\6=æã´f‹1¼+záØ}u÷Ä™’; ‰÷¾„˜øÄnæß•£Ñi?gÖSøÂÍ¢Ñ^9;ˆÒ;ÞÕ¥^±to´ô¡Qa‡ð¨¸óï³2 ¬S!­»Š Bßëï2xaÞÅã&zu¼`^Pºâ&òÅ„wÄCòy_ÁyBîÿ¼l{R?dל5̺I¹LŽBâYÕpo®Õ)7Árô©"Îõ”C6=/»Fd%xÈ»O³BVQåá;Á•-g*dæ1ä„èWÉäÏ@nPkÕëUE÷tÞ¨¨ù äÇfpÚÑ>…‚g‹<6(T¯{pÈ îo)VåVûýÿ¾n륆úB‘!¡ðc´À!¬ †wéˆ=®ƒÑTþÉ_—Öå)Ž…¶QmØô†ö½|å}f¤å4LŽUÙÏ6+À·Ç§û.hEÂ÷TwöÑçJð3lü½D–:ÌÆöU´©ÒÂüçÈÔ_>Ââãèx_ç.$K¹|Ä{éÞùðËŒ°»8R‰ð›(ø!M‰=ý‚÷V¤Ói÷ŸÐûŽ u'‹å›ý‘yãšÑ– õÈrš»Ñ¤ØÙh±v±m=²3÷w/ Ñ#gÙ˜wý§:ä¦1ºtU»ùªßs{‚ûî–G2Ì¢ð)ïÔëB÷qCŽm>‹ÖÜ´ÊåöIM^ýh•°OQ%4wHß`EiæqgeQ”ÕSQÞ¥¡òƒÞ{«ÆPi>ÿcŠónTæð·?ä‹j[ç%%×7£ÆÖ¦*;³«¸ýÀ©2:Ê›¨ýjȧœwÆ1´ß ÏFýêòyêœ4ڰض[ÎMÖôª|ëG3å!?©rÜT~ý4Z¢bÈ#ƒh-eü9µâ-Úz ´ˆ‹CûP•qÛ[< :ydÇÕiT,Ð9P0ä\ÉU<|·sH”äù"@`Xz(–ž#KÏRAx9âw1™˜¿zé!“ú#¤ÿ™?Böû#äÿ…?BñPú#”‰‚\ZФ¤IJ†¤dIJޤäIJ¤IJ‰¤H C†Ä!1dH C†Ä!1dH C†Ä%1dI YC–Ä%1dI YC–Ä%1dI 9CŽÄ#1äH 9CŽÄ#1äH 9CŽÄ'1äI yCžÄ'1äI yCžÄ'1äI CÄP 1H CÄP 1H CÄP$1I EC‘ÄP$1I EC‘ÄP$1I %C‰ÄP"1”H %C‰ÄP"1”H %C‰ÄP&1”I eC™ÄP&1”I eC™ÄP&1”ÿ0(¤¥¤þ–ÒK™¿¥ìßRîo)ÿ·Tø[*þ-•þ–¿i¿÷rb¥›½«#QS99ß~—þSBZ™4Š¿µô -³BË®Ðr+´ü ­°B+®ÐJ+´òßZiWiWiWiWiWiWiWiWiWiWyWyWyWyWyWyWyWyWy—4þ”2R¤ ð[K¯Ð2+´ì -·|bP.ÇŒâÏ@Ûü3¨ŽÞjœr´?¾\cÕ³ÕË?8±lÕ˜‰ •£Zyý.äO®šgç²õ'ÚPþeE| \¶wˆéÉ3þ²w‰ù÷–ß³‰år,¿çË\ÿË5µBb½"§ª¿l1±~‰úr~)±2îåüGÄöÊ¿/çWÛ­r¶ûËVÛ¯îJüË>&ržh ,["¯&Oà/[KäÖñ.ׯ#òë¯.×o úÑðc¹~#ÑŸ§ŽËõ›ˆ~5µ/×Fô¯—ë7ýlÉ^®ÿœèo+ó²%úÝú$ï/ÛFô¿í´Ñ²%ö£]jbÙûÓ>ò—í ö«#^jÙû×iÖºl‰ýì¢s_¶Äþv=^æ¿ öû…÷2ÿ±ÿÝ’Ëünbºß/ó_ãñ2v™ÿ’—W¦ËüWÄø¼^½ÌMŒÓëªeþb¼Þx-óßãÖ³e™ßCŒ_Ï»e~/1޽1Ëü^b<ûŒ—ù}ĸöÓ,óû‰ñí¯Xæã< R²l‰ñˆ¢Y¶Ä¸|µøËã?¨Ÿ¾l‰ã0˜ösÙÇcˆBgÙÇeÈ&zÙÇg¨ttÙÇé-»Ò²%Ž×[÷ÂòÏ/Âò]qiGfÔ"˜ –%'Xƒ––æ,À~@àX. am.ÀÍC ð,‰÷Àï@ ´B°T=›@ذ–@ñ'6N›í ÑA|+°e©/’‚ÔeAzÉ_Ù}‚\ðûLWL%”Y •¥¸mýL lÛK @ýòþ@½´?P.í f‹Ëï[~ïsDŸ©‰ßïäËZ’¨¥V¤I¯Ð2+´ì -·B˯РD½åOÚÑþÞ§ÈþìgD_~­Ìû³wýcú¿¨ó_ú¦ø?ä›Ò ®òíƒð?Êù‡ô•±]•Œ­‹ÿ{ÞŸ¹¤º¢Ì¶1¯þû9£Fó¯úþßí?iã×òÚ€?§4aù<¦•VV’”’–üë¬ÿâ?ë!W™ÚâÿŸ½üg£¨þ§þÊž³,ÝPŽÛ9èè!àtü¨Ûÿ &mÿlZþcƒä'Žþ§Ôÿ¶ªß—#Ú–#ðûV¬òÛ§ù·íÓ,ß®ÿs;Ìÿ¶²æ"›€ƒý G£NŽGN?õ/"ªúÿQÕÿ›ˆþåéŠù/ôóõñ%§OØïÿë“㯂ÿ UÔ5ˆÎ2survival/data/cgd.rda0000644000175100001440000001122512461741246014311 0ustar hornikusers‹í\{ÇyïÙ{pw€Äûed­-@6‚p€@æî@À÷ñÐ-»³w+ÝÝ»{"+>9ò£Ûå8U)Ûå²ùÇ‘UåØN”ØV¤TQ•J"J¬X¥ŠdW¹ˆâäŸ` ‹² ù!òõÌïÛíéž™½[„캩úÝoºû믿~L÷l_Ó³³CS“"&jbô·†nkéFX¢V4×$Sôw>ÝÎ!Ì&\t’Kˆ…ÀÑI¨SPO˜4iÊL` ˼̀Û5Hç*˜Hû ‹‹ KKe„÷–îÞÜ© Nøáƒ„»+¬$¬"ÜMøáà Vî!¬Öš{ ëëlZ5lòÁfÂ}ÀÂý l%l#<|DÁv h#´;; v» {€Â^Â>Â~B'áá ÐEè&ôz }„C„ÃÀ Ÿp”ð0pŒpœp‚p’ða€ œÒ$¤›& †£„ÇÄÂ(%ŒNr„<¡@'œ!<®ááOOþ„ð4ð Â'OÑðçg&ì9ñC­‚jC0™2jâ³}Шä«ÌRШéSÓê}`ªëL-¯ŠZƒ.–bc,@Ÿšg¦uÞjò©Ï¬;9ÏLä“¶4(ñMJ^›úKµ©N‘Wïë5ÝMšž:¥›´~j€}<[šž N±AEbšŒ2&=KPý°}Æ–Oå!G›;O"w&‘KÅ;íT&™Ž÷&‡4·7™ËŒåã£ùB¦0^°‘д#;f%íQ^ÞtìÁí‚ýÍmÍñC™áTν;ß“ÍqZg¶P0¤Í>4š9ÓϦ㇠6a~1òH"?”,d¹ÈyŤÎÌè¨ÏH¹½˜òðx.S¬ÎÂ>ûl"o(½±m$_°s©ÄKwšã½™ÑD¦Ø2; 9­Ië’É<·h±Ó‰d!+gÀw¨aäêz±õ…ç…¼ªÏÿþøuÿ8„‘fàÿ©ÿ/"_6ðÏ5þ%ø7þüb¬B®ׂ&ÉM¾<ÇÀ §È‹À‹#òðÒ yy•ùðŽWÈwxED^©ñÝ!¼¼ ¼5"o׸Uãï2ðnð^p§TÈ5î÷iÜ>¦ñqð©N‚S§ œ? ÖxDã‚ÆƒÏø ?¦ñS!üñ)ò'ÀŸôew9ª¹!¼+PíÎDA¾¹¿cxk¶Bà'‹xV†þÖ$çgÉ®ÉèÓíˆò–T¿²¢´síQúÌdCX\XÕהϯO¢´_?ÝGéç¨ã[Ë#Ÿ©?S1§`ßô/óü”„Ï»ÌéÃn9\_°»;$/w}qw“Dï¾05îA?·Aß^Ì/:ïšpÓ;ßpÃÌëÑÿ}×Ì-_tæû5nÇ<¼åGåVØeâ.ŒûÖçüùnèaî€=˜˜;þ×åðnÈ}ùîÓx%ž§Õ¯E=7 Ìü!Ä37¿ìöƒÎ]è§=hÏíÐ߉p;æïA9Ýà>¶s¿«w7ÚGç­hŸàn<ÿÌíÐsýϼóYÒ4pÏ„—ÛPÿnØmb½MÜõM׎mÈǼÜÕ 9¹àÕÇãøA´s §kÜŒy¹ ù™{¿òÌmèWîçõoÜßÌ[!ÏrÌ÷@ïFÔs3ä·ÀÞV´óôkÏú¼é› oÆßfÄ?ˆ|»Qþ:¤3wkÌí¶òíoƒ¾Œ›NؽñÀþ6ÜÎàî^7½å¶#ßv´{Ï_¹á£hç«ÜðúýÁõ`Þ»b\„}ÌP^×%؇0óŒëM°—yµÛ“<,RɾÅdöQ¢ì™öEª½wu'Êž[¥ûšQôURFØ6rÐörµÚ1ªíQl¨tÏ-¬¾&9-ïž[CÿÚáÌècvŠÏI$Æ Ù|v$1iëÍyÍÓ×ô5}MùÚ~5˜Mò·ê ³oúš¾¦x®/aÏKµøÝ.ošý™¯›o²å„̓ÕÖW©þÉæ¯´œ©®c7{ü˜ÊŸž¼ÌWµûù½ÂSô9™ÌÏÿ Ÿ´Q¶,*Ý^ˆÝ×@Õ§§U²­âçËPS%;£ØÅà‘ëõ© ЫúŒX>ºj•¼ºlX¨ò~÷ª½~zô²‚ʈâßT¡O”g¡–wõ>Pri P¶0dçžóÐx.;f? û`ÌB¼+[ÉQyý„„Æq/gÈ#_$ü‚ð6¥µð×âõ+ÂïJAk…r¿J‘›¸q%îU SÛXd•EvXØ«[Ñu€C`ž¿Dø__(²fâæßÁßÈ/ýï®ú'YT¾•Ñ"_Pîë}lyy×)ñrS¼payTý:á×$·\)¯QÉ#‡K?@Ÿwx ÿŒr|.ÏáþMp›"×PžW\Vîè“®^‹CÒOƒ“Züm!ùä%Ÿ•¯^S➉¯¯G”“cAñyV‹ÿ†>«ÜU_dž³Ô{öicBöaÓçò°íݘ!N/C÷+ÇTF%[¥~¶Õ+,-Ȧ ÛÕzêíaÒ­·_ýôòÂÚ,J›Všî×Aù‚ú>,_¥²¦v ²™ïù™Ác}U®1Ö÷L~ÍXc¬†…³Æ8ü˜pÖëAúÊ5& ýO#¯Ï#ó©kŒ•%Œˆâ#Ë•kŒõ(tÑa=E ¹À¢9ÀÚ@éç…3¯[Û€­[8kŒ”‘kŒ%çj¬1V‚@ó½µ›ÐA>ç¼ÆüF8s·\]EqqìêrÛD®1VŠ ýØé¹·ÖöÖˆâ#ï=[Üú¨kŒ%ç_m‘6Z+Ñ>÷ºí¡®1ÖH_ %¤o;ÖK|Óí7Yެ—ÓÎÔ·Öû¹ù‚˸'È5ƹ_…r—"þ†›_®1Ò^ \}â—Äu¨ÇiÈ'ÝþpÚK®KR~°yQÎB_¬1Nýäý úâ Ë£¿%Þ{dX¾+=ïæqÖ˜¡}â¨ïl´›Ä¿ ïóO„qëí¬1/þ™pM¸kÌE„åØùo@ÿæ…y>ÕK »×ãütû•¥§û•¡ÛeŠÓË1å¯$-Ȧ ÛƒÚܤÛd›_½¢¶Y”6­4=¨Ay'“¯RÙ ñl +÷^góÑĈÍÎæsËðfë“öh¡ø“ª>—Meù‡T]!g' 쫟·Ïòmb°èP1dg‡X¦þq54#3J?À2l¿Ò²™TžÃcô+íÜp"ÉâCÙ|s²X^}!_Hä8TkŽ—Ìʲì_OR…ñ¼VéÆ\öñf®¸ã¬0Anܸ±Éð{°)•($šÓ9Ê"œyΣ®69˜Z‡ïð,€Æ ´yÐû.›‚¿#ÁßÒQ¿•#DZ^¿oãèßÅáïáÄEùwpü¾£~÷†¿y£~ïF®ãüõ;6ü ý»5ü­šVQúFþ}þ6<gú.M—0“FÿMØwhøÛ3Q¿;å›3ríçį̈ߘáïË|J”ÆÁË®g.âŠTÙ+Ú½ç½G|?td¯¹÷Öþœ´+ZÜ5š-rm÷È^Qd¿U*ËOŸó^ ËßS^¦®WÊ8؇2¤îëÞxUNo']æ}K‘QﯕìpðV©}ø·•´ë(ÿ-—‹6ìAÚuèFyâ§^F¼RƵã%;×Ð+êoéÕùuzÍ>_§?]Ÿ^rñg;…õÙ¯¸øû¹”°þüia}ëó\Ößö 빟•ðwß,á»TÖ÷j…õýëþxþ8áB ¯}”p©„ÿœ¥`LX¯o¾RÂ/¹øñàóÂúI —† ¯Ðt¢„Õ?wqO·‹5OºXû¬‹fjÇ{iÎYg+øëi[Oã¸å.Ò{h׸è¦ù«÷³úÕGsÉ¡;J8LsÑῦ©exÃÅ™¿§Çý¿èõó¡Îvñ„P¹]ë'iŽzò]xÖ!‚ßu‚dDD™°|jÞ°wgS¹&=º]~q¦²ƒl{4݇½«›l0ÙT¾ç—ŸÇAØ~I¥0íT±ýa{sQêó)§ÒºÜ¬zO¦,½zûIð8`×aÕ=˜Ý‚Ù%˜ÝU7`ÕÍW†Ùµ×Ï­W¦«.½–0»ðÆD¹ë®ê²ËîºsEÉUWuÏU]pÙý6.¼.·ìf«º×ꮵR˜KmL˜Ýiu7Ú0Úy¢2—Ù(‹¬êË®±ì;QækrgÕÝWu·UÝ]Õäžjr/åtÝ4ªû(ç7¹rz¥n¡QÝ@MnŸeîž(Wwó4¹sšÜ2uwÌ0·KݽÒäNÉýer£ÔÝ''ë69YwÉ0÷È©ºEêîÕv{¬–»£É­ÑäΨ»1zÝ'ÊÜMn‡º›¡î^Ø‚tݰÏ¥îØ…~ÕÝþtw¿¨n~aî}º;_T÷½¨îz&÷¼vȇ¹ãõ \ÝíÎä>§»Í…¹Çénp]H×ÝÞÂÜÝt7·Éº·MÖ­-Ìmªîkº›ZµÝÓªå–fr?3¹éîfE73ÏïÚAï¦~ïŸSy—6½×›þ'ìgsÔߦwgÓï…©ü>ú½Réï€0]am¢þ>ÐÛá ~ã§’<è¸Yû“AL6…ÉEÑ )#h^z7Û.HwMA0Íú³¥Ì09üÎí‹ÕuG9“¤ÏïL­ž'ìmÌ çgOйY¿ùKSþïùMBâ¼2â2ø‡º=ŠÌ‴ӄ¤¾Í$‹ôZð3ArŠü×#ÈÌTî¿¡ÜŸy>ÿŸ^¦Èý.¼ yîOäpŸ‹b»¿1sÕG.Rß•û6Âu-ýœOž²¸jCÎÑeÅת_¾XQθÿb…u”óZwµm¯BÝåIœ_kqoµ]ì~í&è|6¢œ:~QYÖìjÛ]-È5øVÛðûáoW–×j¹ÕöOãfŒ«ýVÛ3÷Ä88y«í™Æ­òìr-Ÿp^ð{tä;=^ÈŒ˱e`½'´Ájñ„6zB›<¡ÍžÐ}QK‰:)¡—$ì¿óÿÃ\j§»usurvival/data/mgus.rda0000644000175100001440000002125412453524666014540 0ustar hornikusers‹í] pUÕ¹Þy¢ˆÜê´µsµ¹µRåaÉλí½lÔVm‹†àß„¼$yk{Ç8c­ö1¶õQly TI `4Á±ÇۙδS;ÎØq:Ø:ŽZ‹7Éÿ}keÿû¬³÷IÁ:=3'ß^ÿ¹Öú÷:gŸü«êòyþéóN÷ísÐ^…ö*´W¡½ ísÑ>ísÑ>í×¢ýZ´_‹ökÑ~Ú¯Cûuh¿í×£ýz´_öëÑ~Úo@û h¿íóÐ>íóÐ>í7¢ýF´ßˆöÑ~ÚoBûMh¿ í7£ýf´ßŒö›Ñ~ ÚoAû-h¿í·¢ýV´ßŠö[Ñ~ÚoCûmh¿ í·£ýv´ßŽöÛÑ>íó €è7>ýço’î˜`5è«A_ újÐWƒ¾ôÕ ¯ýÐ/ýÐ/ýÐ/ýÐ/} èk@_úÐ×€¾ô5 ¯}-èkA_ úZÐׂ¾ôµ ¯}èë@_ú:Ð×¾ôu ¯}=èëA_úzÐ׃¾ôõ ¯}è@ßúÐ7€¾ô  oýBÐ/ýBÐ/ýBÐ/ýBÐ/}#èAßúFÐ7‚¾ô oý"Ð/ý"Ð/ý"Ð/ý"Ð/ýbÐ/ýbÐ/ýbÐ/ýbÐ/}è›@ßú&Ð7¾ ôM o}3è›Aß úfÐ7ƒ¾ôÍ oýÐ/úþÛöàýØÄIÆUÆ;Æ¿* ã&ãð\öCýlUO¾Œ¯Œ÷ŒÃ¼ð>£ã<ï+,ó~ÂûãüêÕ•º?õçý÷3Ú{µêO;*‰­a:ò!=ËÔ‹üxÿªTH?ò>IûÑ>ÞÇésߥ=J_òÑ~b=÷ ¼ßq_0'H-‡÷gê{•¢£¿¹ˆÐƒ/÷U´Oõ£¯RhÆ™ò=÷?”kö)@Ž‘vP.ç ýÈý õá>ãÀyÊ}ñjEj红]”gÆQéÅñã~~¤8oæá~ôë9Ÿ8^f¾ ãÇu\é…éè_ògú‡~á~–öpŸD~ô‹žOœ?Uª?÷¯‘ù yœ×z?J9Ü终 ÓûLîÃ¹Ïæø1>Ð.úÁ¬ÕOÇ;êO½Éר§Êä[¥øpžqÜh?/pÞÓþKüü@ÿ\©úSõ0ñ¨ã õ¢½W«þO®ú›òéÚK>s»Âv™ý¼²WÇêK;ôú¦¿9h‡™w@~Þ"åqËèŒßãÚGŠš¿7B~»´_+=ÕˆyÉÝ=J¹˜éÛ“ÝE>®óYf]º>yCiu»«Ç›öh‹wšëœTô)tÎÑ1¥N¨Ïr«Ûµì÷ö°KLÐ/¢{*]“¾ÓØ™rœÙæ ñƒ‹o½¼cä…ý—ï%œ/id¤Í7ž©è=ë“Ê•­w·ÞùØzçΚͫ¯Wòjö,s5[èòõ–ÝlèvíòUþÝÇÛ<¼fP$Ë'ƒÎ¿Ê žÎ É :/”úw îxEð™KF¤_°§Søtʳøà[ÇÂr¿ÿrZþAÏCéÛ·¬HÙ´ÕHoÍ?xþ‹­)嬟 ímS3’lûXj½Ýžºþñí©ëwïM]¿!µ_LûÎõ¡ö`Ó91ý_‘?ƒõ÷‹ŸvHÄ'Xõ²ôo¿5¬çSò[’àÀÃéõm»WÑ9æÉÞ/È®¿ö¢ßÆÊ¬ð ¶v%óWïô¬Ú‘X¿‡Â~vÇ™ïÁ†iiû½¯´mvýHèž•ßð[o Ñ=ÿlMÇ/¿{õ`ÿ û®¬ø+8r8¬ïFǺÞüNêú¶ßŠ]‡îLÝÞ!q,xòwYÑ×iÇ£·dÄ?xä˜ÃÎ÷NªžvÞ–ZþÖ}â¯Í{BíÁ¶Ï„Ë;ñŒ?õ—lçU.kiXV·|ycË’‚-W,kY¹´ ¸à¢Úê;—_ì ¹‘ßž/Hàî¨\/ ¶ÈÓ‚{üáAÁŸ¶ nzCpß‚ë$0›¯ ø5B·ü¨!èúÂ`CpøM)w´þ|’àþ³ nä»ܳ ývzïý_è!›ƒgÏýB´oAÿO@Ï ®ÝŽúq‚›ç¡öõM=ŸÈ—rO·”ù»®øtÛäG}Á‹m‚»È§à¨ß(?n Ú`Wäu]&üïž½eãô¾*õ[äF¼à ®Úyò4ØñøIZÐ^-xÏ'!¤åÇÁZùc°SnDÁ#µÐc™´÷Iy7ßáÀç*øãEÁηÊC§à€<„ ÁoÏüJp-n|[ò7—{úÀw ú÷¶ÂŸ}²á žÚÿ xðjøa ìËì9"zݹÿOÐóï!ľ+ø¬‡{ÚùµöŸ@µ‚cÞïÿ5ô?*rzÿ(ø üÐö(ì­”ú#ÓáGÄÇ^ùqppd©´ËrîŸw¨o!¸|zΕ~ݽ‚»ë`ï/û°ÞÞÛÃ/˜7{ÀoüÖW)¸~~ëúèùÂÿèÁXgÝ÷I™ãÔñiÁC˜‡~²AžÅº|þ:øï{‚Ï] ú¿ þ&È=Kè:'ºÞ—òþk 7Ƨó<ÁûG…¿÷¬ÜŠø¸o–à^øcpÛVè)ß»=ów/ÈûPlÜ‚ÃO½ˆ›]ˆKòcêþ¸/øÔqÁÆ nF<;€ûXüùì ŸL‚]˜WðÓKïI;näÁnÄ¡ýßôä•òF>ñ+-MM-«§_·´àÚÆæ)bœÍ½cùú¨é‘m~ÿêþø¨ _§Zî©Òã_Ý'Û¾þíÇôú%Õ[÷‹£‹oÇ'é±õƒ¯»ëû?¢ç<>ðZËzóßgãŠ_¦|Ú/^¿ Æ£ÿÔŸ‰²‡ö1(³éÎPüt;ùQ­ÖãL¥‡þ¯¹Ó•º<ü©ù¡Ÿiχ<Ö'(ä+;È—þÕúFø¡þ?”Ÿ´<-WÙå´tÚ”¡çøWúŸæçÄÓæÃOé¯ýÅþoô9Ké£Ç‰ôF?5©¿žŸfªùœKÿ@Ÿ±JŽž/¬'ŽQíÔ'?<‘ñšàê£ç1õÊSò9jœ ²?ô™yý8>ô£cžX9Š/íþXØŸ¤›ù¾š‡ùj>èyC=Ç)¿ž¦øP¾ò“¡sÍ3½.u»–£ç‹Ž‹DÇ\~ÌSvi½]tò7ëL­CµŽ ý9 c|"õ ]De/å«ñ3rèW}˜¨æ1íqÅs? ›¡×ëû ?ôº¡®;W»^תÞôŸ¤èµ}ôüäø$}æÜºe+› f5-XÙ܈߳Ë{æßÃg¾)Œf¾…ü:Êo¨2ñ¸(@4Šyá2ù™2éßTü¡i[ 6í”§õÒúi¾¯…ç*³¿”N¾J¯ˆµ_…}ÿPv¾®Ú»ÊÏ.¿Ó_,¿¥üôNx|´FÚMàvÙI?iÿ±ÿk1ãõŽÒK{Äß´|•ý¦?ýé°KûË´kûÔ¸i~Á¹ÊNö×þÒr5¥¿ÖWûU÷‹øü_Mm_„N­—?=Ïë=â¿7`•z}s¼Þ ëmøR®Ò©ñ‹C½AÄ)½‘Pþ2ã¬ã¡žgz|]편aøêõ£ç£ZïÎy æYìüTöqcmú븢æOd|]q@¯—×Ôøj=]ü‰ºŸÖÓ¥·¶_ûSÇ'ÎC—?i‡^o)zêÍz ãÆ=InÜ—-««^Ѹ¤q ¾LJîÏ©¹<›ŽK0†Ñ/˜ ü<ð"¬Õéèw¡è\ >£žü¦“í“ÑŸ|§±ù°¿âCþÓ)ý'£|¦¢þsèO{¦’žrh§j§|âg!íSt*ûé§‹•>Ô“þø,þ ?úÉø›z þÕŸzQ‘õSA?…H;Õø“ï穇*Ïh•2ÑŒ»’Kÿp|¨÷Š/çÇ…~aýTe¿Ñ[Û…zcê§©~ÆŸô#Ç“þEùó(OVö‘ñ;ý¬ÖÇÓбꉜ7L‘t1øÐ¤§?¦); ”ô ûOáüR|.Vò.àøpèÎoú…ó˜~QëŒ~œA»IyÔWÛE¿Mæx²L?‘?çí¥?Tüàú1ö£þ áÃqSzég½îh·™œ·jQo³ç+½)gšZ´ƒ)³(ǬwÊe<ÑóÈñ ßL\VóÜÄ5o/V~â:¡?ôúœ¦üÁõô9µŽ8_y1ëZ­»óÔü¾„ãA{éOÚA=”ß©ãà µÞgœæO{i'ëMü§=œ/j¼¸~¨ä;îݧ_Y×ÜÒÐÔ² ü‰;fžÞð+,³Sx_µ³;ó ¾1;ÒP|ÌWqè§ûSw敞æ+±1J÷•>®)û½.ó« £?äSoÊ5ôZOí?-Oñ3tÔ—ú(ÿð+5îà =ù¿ÖÛ|õ•£ìÓúëqyýáïˆ_~3úÒ¾¼°?Œ,{ŠšF=Hþv*Â/F®úD§ýbÆía¾Ú.ãÚ§Û]ë€ú˜GÀ1ýIÿFØ/‘ù£ýGùÿæcúC=_ŒÜwÃú˜ù¢étŽû»Özdbâ ý©íg;ésÂòÍ'EÚ9.ÌÇ|£D¿pü?8øU®š§‘u£ÇÛt»ò›¶Cû3ò‰MÅC¯ÆËôwÅÍÿ8ÇEÅ+wU<ˆ¬Ï·¾ÖÇøÙø;Loä¿­ÊjÜ4ê8`øéõK¿Pß·RK$.r~¹ÆOëAÿPÿ¿«þé?qã«òÙÓç.m\ŒÛá´Kª›ë˜öaÿ¬±–)ª˜4"oyÝæ¨]sç2^/­©eýØ¥5+~ÚÆÿ5[9¤4¦¶ÿCÿBöiÉ;1ðmë6°~lórQ9¤îøe-«/ªò›hÈ)ä…Ï‹"^󢄥¼(ãE9/*èƒÂæªÐ\ùæªÈ\›«sUj®ÊÌU¹¹22|#Ã72|#Ã72|#Ã72|#Ã72|#Ã72ŠŒŒ"#£ÈÈ(22ŠŒŒ"#£ÈÈ(22ŠŒŒ"#£ØÈ(62ŠŒb#£ØÈ(62ŠŒb#£ØÈ(62JŒŒ#£ÄÈ(12JŒŒ#£ÄÈ(12JŒŒ#£ÔÈ(52JŒR#£ÔÈ(52JŒR#£ÔÈ(52ÊŒŒ2#£ÌÈ(32ÊŒŒ2#£ÌÈ(32ÊŒŒ2#£ÜÈ(72ÊŒr#£ÜÈ(72ÊŒr#£ÜÈ(72*ŒŒ #£ÂȨ02*ŒŒ #£ÂȨ02*ŒŒ ÊÈ+œ1Ã^ÚKß^ÙËb{Yb/Kíe™½,·—VZ¡•Vh¥Zi…VZ¡•Vh¥Zi…VZ¡•Vh¥ùVšo¥ùVšo¥ùVšo¥ùVšo¥ùVšo¥YiEVZ‘•Vd¥YiEVZ‘•Vd¥YiEVZ±•Vl¥[iÅVZ±•Vl¥[iÅVZ±•Vl¥•Xi%VZ‰•Vb¥•Xi%VZ‰•Vb¥•Xi%VZ©•Vj¥•Zi¥VZ©•Vj¥•Zi¥VZ©•Vj¥•YieVZ™•Vf¥•YieVZ™•Vf¥•YieVZ¹•Vn¥•[iåVZ¹•Vn¥•[iåVZ¹•Vn¥UXiVZ…•Va¥UXiVZ…•Va¥UXi6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–ø6–øý±$¼Wß²¬¹zEÿn0÷ìÁjÁ.ƒ¯!¿äÙ´hcf55®2YÑ.¯«®þc>PÇ®ªnZÙ¿éDi0ÍâÀô„êo¶Õ²Û5"‰N¯­^Q}Iý²êæ”»ò3úm<a¢¼s Íg+Oáh†@,Tè‹+,QXªp¤‡'èC"‡)[gy Q)Ê—¢l’qQïœ"ö â|…à9´‚x,ŒàÒ± ‚OâÃ,ˆà9Ô‚~'ýp "øD¹ nP¾<ôÂqø…=Cöc‘C0ˆ£}†>Ãu8Qßïu»+¿'ï³shGWj:ÚkîÓÔ«5ŒÚoúP¤‡k)‡üXÖû¸Ã6ŒŸ©ËŠ?ûÑ/‘C7ˆ^jÔ‡pµ_´ÿYï:”#îpޏC:ˆ®Ã:âí ÆÞwˆÑÌ;ê©|‡{¨‡>Ü#ÓC>ˆW(äúÓ‡~dzˆë0¢ëPb¶!êõ’­ÃB2=4„H{Gýð¢’ç©ú‡&ŽO¦:§³9•nIù'á™®.©Ï2áŸNwÝ?Mé|êò™ËW©ìvõI2qý’ÔgbW&<†3Ž™ŒeÒö¸¹—Ê—?\ã˜äd­ iå{Ž=¿%}†g|Nã ¾Üçfª¬‘í¤G9ÓóY"rIוšNß§õ}>r~‚¦sè©W¨?èÏc´³J¡± tzß’ôÜ—Øó_ˆŽzm¯á£PÓ™ïývDüäèω‰<'€¾zœ#ü:ÏeRÓÅòóÒ£ñk èÙÏ¥p¸çÌ)GË‹;wFŸâ:‡Æ¬­Ë.¥Ÿk½RŸÈs&êÝ¥ÊA˜.Óójô¹5®~zþU*}´}±ëšt¬Wr*U9âgö ÂY·¬×ú·Æ .k:m¯+ÎFâQ–0ný*ŒÕ/Æ>/%tÞG5jzW¿l£–ç_—BÅ?n^VzŠN¡ZwüœŠÒà~oà:Ó·g÷”æ,^ë¶¡}Òð3g²àᛪºº$r‡úÁEŸŽOœÌåA_¹x:쌷8݇´ ú[ÕEø+B:å÷öŸmrôOi«Ë®¤o½—Æo^ŠqñÒø1ܤú§’9¤MCJÞ™ÊTòcë’“ ÿt|¼°O­ž‰'ï\"|j+ÿ É2ÚÉuRFWrF¤Ñ•¬Qã“7:“:ÚM2¯¤É“&yœ ü­“=f;é£Nþ§Ïxeß0“B&MéL9^Ù4Y¤Æ¸ä‘®$’ “If-©¤N.™i’IlÒ•tr¬Ò'iJW2Ê„I)³žœ2Ï¡Ç}¤É*3MZyŠ“WfœÄR'³LšÔ2&¹¥KwÒK¢#éy«ç⓵d˜®¤˜qÉ1ã’dêd™I“fŽrÍa'ÓŒKª9ÚÉ5c’lêd›®¤›®ä›6 'þûàïa9§,Ç&Ñ•«zFroßû#¹8‰ÔOÛ¡íqåL{-µ¹&U}lÎN—|­¿ò¯+÷^ÖsyºÆ›r•~‘Ì1¬O˜ë36çgÂܟΠq¹@µ~®œ j~:Ç—~¢<‡? ]†¹BòµžŽÜ¡.ÿGr‰µãr7µ1¹F]91GšsÔåg»^/q9Hu½¶ß–œ¤öþ¡âµ²'Ò?&Wi$>ê8¡ý—4Wi¦9MãÖ»Ž ™æ8ue”ŠËyêÒ+ÆOüœÉ¤ã¯ž‡¯+ù:Ni}’æFnŽTÝߥÿhåL%ÒNWÑþv¬¿H?“éß©|©f]MÆþo´ò¦ê|€®<ª:ŸªÎ«Jd~ÕlåYÕùVuÞUWþUb\Vb\>Ö¸¼¬ÄlågÕyZãòµê¼­ÃÎßJ¤Ô|ŒËç—וèÊïš4Ï«3ß+‘þãü è?åÏáæÕy`3Í«óºòÃFòÄêùÊqUñÀ•76.ìÉÊ#«óɳ•W6Óü²§:Ïl¦ùfuÞÙÄùg•ž®|´:/­+?ípóÔê|µÙÎ[ëÊ_—Ç6.Ÿ­Îk—ßVç¹­|·ÄLóÞ]ùoG;n\>\וב'׿ËÅ~LÈ¢ûïQΈ«3°½4½–—17[tu&]×>Zgº%¾®dØMši—èʸ«Çi¤™wþweâ5|µŸÞ Û—™7.C¯‘ó:è#™9•¿ô8¨ñˆd4M˜Á×™É÷ݰ^ÎŒ¾YÊìÉð›a¦_ÑÓ™ùWeNœ Øâ2³!Xg ÎvÆ`½~5½ÎtKúS”A8ÓL‘̴I3 »2 ë¸åŠ®ù芋® Äz²œ‘Ø•™82ŸtÜ%º2£ý8çìÓ™‹ŒOv&ã‘f4vÆÁQÎp›éØ5¿UæcWäÈ÷cßž/ãùðÁ’F(¸£Rp½¤1 ¶HÚ±àžÇxPð§­‚›ÞÜw†àºõ‚æ¿„0oº%=JÐõ…Á† ó¯R>ü¦àÓ¹‚-‚?Ÿ$Ø) ƒýgAÏ;…þÀ]RÞñŠàîÇŸ‘n°gøì_Ø»÷¡·¤‡ž=|¢} è;A÷ èQ;?.ø­c‚k·£ß8Áï¿,¸yÚ᯾i¢÷2‚ž‡€ÝRÏ,l]‚?ý–‚Û$=\ðb›à.Èm«‡ü‚w }ãmh‡_: G×e"ïî°zô¾Ú’·e¯àó_”ú|)¯Ú=fIýzyÐìø äM…ÿ~'Ø^-xÏ'ÑŽñ|QÒÍk/‚}òƒ™`çÂÔBßeÒïægO‘ànô?ü€àãôÇUè÷¢`ç[è{^&x@þý<8„ñØ»Ÿù•àZIÿl] Èl;1¿÷ôAž¤Á 6úÁ#°¿óÁVè·óµëîÉ‚ݰ«û ±÷¥»~!õ½w nþ”àúû!·|¶Ã¿« ÷ÿMðgX§Û@·êeAΓö[¡/Ê÷Î| ~h¿~ÃøtýŽ`=µo€¿>¿ Öáö?ƒßr üÝñ[‡ø²WÒÏ}úÀOíXQîÚ.òÛîƒ\I‡ì[*Ø'郰³}° |*áø¥Æ<ú$ü…õ¶ í­Ð¯oŠÈýâŬ¿­]Ð{µàó°ŸÙ%{W ~åé÷{Á®„‚ë9^à³áG‚Ïb·ÿAðàÕðÓØŸ'ØsDô¼Ÿ”÷ÿ ö`=<¾a>¿ð]Á`ßÓ&Ȭc{°ÞÚ}¦A¿µ‚c½îÿ5ì:*ò{_þQðAømìi{va¼z+¥ßVIÏŸ~†ûPÏ?Áo2Ú—JYÂZÿú@}ë è þÝ«¥_Ϲ‚ݸt÷Jywü÷KÁ>Ä‹G`ÿö˜¯{ÀÆáÈaáÓW)å-¿ðÛÓX7GÏ—~G'nÆ|Ø8Ò}ŸÔs>´ýV°ãÓ‚‡a>è ¿óñYÄ¡'‡ž—4HÁ¶ï >w)øþõð÷ѳ„O'âg×ûRÞ ìÅüèHåzö¿¾3}ç¦à¡ëRõIÊ/ßwNžº.‰Ü‡Ì8ý]zŒ´œÎö¡u9)ê“êžtü’ê”ÔϙΉTý†ò‰óA&|ã|ã¥÷Uܘ%—$~šU, ï8™éü˜Ä–t}sbä µÃ5G¾ýzÿ›¼¦nÉòžP˜·´¦YË’ôÐ=õⳄÑ~ñY‚)ãYB´ß…á2ž%˜2ž%˜2ž%ŒôÅg VŠpÏ"tx–ËÏbûá»}g;ž%dë¥åñY‚³?ž%X}¦O.ž%8Ûñ,ÁÙŽg ÎvK8Ù/>KȘÏœíx–0Z/>Kp¶ãY‚)ãYB¤ž%Dêñ,Ás 4¢ÂÏ .qù`¾zõå®°`Ìćòb5zAA¼ Q n|‡&¥lhpï–?ð®QUÍ¡L¶Ç FòK>W ›ƒF‹ÅÉýO0RKâpý˜ƒ}ý$ïš)T4üä¼S•®È¡ëÅM3oŒÿ ¾ž^ø=‹ö³Ì¯—º4㤼3/ókWœ0OŠÓ›{ÑÎL¶¸–Ý@8•#,£ºXÂVŒú½bÄRYöUƒ{]«þ•o~š:7z¢=\ÚVUÍàçÑŠóø„UÓ%ÊúF_»âe8êƒùoh«Ó¤ßç#P3³kÁ’®ÿ4/J¤ü¬ïÃ0þ¹˜ÈØI³Ûþ5Êgñ”ÿ L ´Ó*jxÕNg•qDƒí2ψR;¹M×£}à,ñ~ǹŠçš‰VGÜ ¼:ýC#Àº¬EÃÛ£¥žRíï@ãª/Ðí»MÏ Ÿ½ë±ÔiÔTÐWå ØêŠ‹Ë~ÏK.'‰H‚ð=“øU[¡#GÈoYtU^òKk× Þ—Öi"L½iÚDš²I‘(”ÿYPŠqÓ&^s_“å¶I\#!>¡Ñµ#Xû##Í(âH5T’«("ñ®ÝnCƒÀs_P"Oö«%œ¯ÐU r0CŽg¾9WŸ_§U4n¿þ~툾ØWaE¤Ïcóÿ¨V†ö§‰,„ß ü¦¬(dÓ‚n×ëP«„.‚af&±ìÁì (+ÐpŸI‹š²ñŠE›Z&ér¨1nq¡Œ¶yÉ€¢óDâ·ß=ÇÚá²Õ}±ADêàI"(’Bëh+e¢´­] w‘­é¿våä‹^ ePf¤¤oeP# jWèœåuL¸c%êQíx`ÝdHÏ‘þãÁîDj.*–†tûYb6òu……>À^GôoÂ!éRIÿÈ z4Œƒs=µÄžŒ¿°Ú‡‘™t…D^Á!äbVý“iÓ—(sœ"jlç~#è`ûn«{89£øYC¥¯˜OñÒù[Ó$€K¿ÀXvI¦:Ô@ôÒ®d%3“ó§&¬«*1Ìz‡:ÖÊžk}ÿ5d,·qÎ"s‰†žóÜа ô:e¾Ù)ˆµðeÃh û *&ØÕ&îäµn|;GmŸŸWy»™µMÝ7¦aï=w¬E×)òP­l>’TD G£‚Š'ÎÀã~· Cˆé·ÝÆH¯ª±"òàè‰úûŽMÏ誾#åjm¾C0èæ¯Ù]0¬jÈ™S.ÂüòSftº¥‡YýŠÈÕí(O¡äh„(Dkí9!QÀ@•ä›’šZãWô#(Ì(ï-‰_ÖÚùÙSv˜AJßܸ[tú*u~¯ UÕ·ÏuQøcXœç—<„hÿX#)’(í­å”ù8¡íFâƒm(¾/.%Ó?”œ s4°€ü­õŸófo {— 'Bƒ©Pš:·Ž‰F¯k!DÄíT°OÞRˆÐÃt6¤}LÒt_ìý@ê70Of—/4$þ‡õÃþd{›³L­ÿm±EC«~ú½A|Oþ¶­â§b}ø]ηvI.rœIGâ²}-®œ!ßXØl¹nÐH-ùÔùê=kKLä‡ÿNc·Ëª¼XÍ+ðj(#½^´:èÒãs’ø§7¨&<ã+óWdÖ]ðò÷ÐJºñ…wûò&„°‘¶¥ù\c@=:“ƒAªÌ_à¶Å&¦T˜c*Ý—i—d §ZŽø+ŽhdÓfÒŒ‹¦tc®!s=`ÐüŽ}*€?ÕÐÀöˆþ( ®Érúô\}ýSw£ô)†Qpà!Æ1p"@Íìr3ÕÉÌϸwöGOv »*Ô?Nq•,eeKèö”þeÊXÓ;•êêm ?8I™ßf¾D‡šyÃilôn3E[@éF€çWW%ëÂA[ÓŒv++¥—499Óg@p=6'ŠàaÄðj@Šq–*Oc~ïŸTÉ¥€ÒªÏùÙÊ©ýU± „N~¾Iã0™îÖ‘Ûo˜×éËÛ¨§æÚæRbh?ëL«a¡X—`xMä«B*濳`Qƒ´x^>'é_0¤N¾lc:T¡¬ò*œ']Žêžï«•Äá.ƒ[Ôl¿;‚¾w´µ6§hòÙþ—åJV H±q'‚KÒC·Yñ¬ÌŽ>« ¥,àMø{ÍäÒ€\Hak¢F—3M¾š7Çý×Ĉøãjî›ôâ÷@yC*Â.SäÖ“ÇÆÀU‚‡Ï9Tߢ²è0¹ø ^#n1òlMáÂxK»€wÍJ'š7ÛÓ hÐ/_on8$³)kø>5°"He>lE€0Ca÷aG*½úýh”ú~.ó0CmÁVD98´Š/Yp¾4­ØCöêñ怬…g³{?«=•q²k;‘ñ {@A¤xÄ×[ÈHÁž/ìßy#ÖkØSg¯ïÖÒ=ëyÅPø•´[öº#'=ø•}Å¢tÝ.FB¸.ZÌœ{ù³›n˜«Øâ{RêJ<¸B¤àînZL¡îE I×y;*5îª#;Ï0Àë‡úG¸,øýbã ¹¿»¡'¸²K@û‡®£‡6ð]n[/Y†6ˆždC(èüHG¥©˜ÂeZúºõÐËê@òרõm"ê Við ÛæPR?×;µ$äóþ€³Zö‚I,e‰Q¯ýyaîܨ,.¹~F¡Ç`– ‚ ¼õc†ðH YÇ#]²ß*¦‹‡“éé\\ãûh2¿5–aƱ¥Ñ$Ž> Yis_ž–Â|‹7ƒùpÇŠÐjóÌϤMÛæ3À"]ÝS„½¤Ê(ÇÝ·/´möÿ›Þ[O`¢%=!`ð•R=üm%1YS€¤wÖ±ûÏØËx°Ã&˜ T¥å×*ÕGAÚe+ÙÚ> ’¹7>a"²^è2ýb†µÖ­¯gß‹¥Ê9’<‡÷†!A6skƒË.¬ñE^‰¹ÿ¢V¨èo°iaÅØ•ZÇ Ù(yÞFᇞbSˆÍ»-“¨5Àâ1ð¡'Oªïu³‹Ç¿š1Ç(Ã;^PitQ®Z¡»BlÚ²¹4¡u Tœ¼×K‰È1 6áµP„®#©l1§ ÅÿDÇõõ\PÞ{¹i¤—àdc±E J’‘TŸV¶¾ñGá/Õm>ÏQÇ6TVäÖHSzŸ‚èó6anl¸µ±€ºçGC±pòA}‡›½Ÿ¡þ)œ¶.£’XÓeçØi"ßÛëB³‰jsq]µó:e+IYïì¿C–¤&#Ñ5>%QÑøÿ¼aDh=ü€3uûÛ•Nl_fÊwƒ³jÅÊœ¬ÚøGÿ@ÑuÆcW~P,4ƒnmÓÀŒ% ¾²’OÜB{RYÒpµÆ¼`ïAç÷!9 ó4de5‘ß¾ úcýRÅýr»dweµƒÌˆ±æ@=ÃIMZ×0ëŠ0 û­êÍt÷¹‡ekÈÝèÉèé}Fƒtô8š´è£Kê锫FémÞ­y ÿiàZš¸¯†hÛ ýðØv1¹Hu§t+ #úÜ™º×°J¢êÞ‘y9Û}ƈz3@h¿“¢çÍù„0yó äÙˆ›^?JõÅ õ5 ® ³¾Ê:•g7®ñl8çܱxýbÜ.ßëd¡'eôÛÞ²óOúù%FþeRÁ„«x°®—L@Œ’(ˆì?à–UÈ~ø7˜0ZÇL®”ÃûÙÖâ÷šW¨É™ÍN:¦pÕ8‘×GgËôòø¡S.ËLØ8bÓÉUìiùXñýç,dPËï$?í”äÈå`?QÜŸë6 <=ä±,¼|¬Ôƒ€®èK íÒÃGí—µ4Ëë÷ëÅËœÌû»ã{_¤oœg‘MfðØýeyŸ_AÚbta£R×+ªf€[‘=ÈJeÿ8жë5Åá HdÙØDgçCàYƒá:¬¥š&èøöñ”Æ>õÆ™6 iJ ñÐï‡e/UX.Pëzb°&óŰgž•~ÄD},LvÔš~¤ÞC?;¢„;” bŽße³O Ì.¼?Tþsõhî×ûd¸B’¹`,kÑó«R;e6«÷Óç¶)UÑKD>½¡í îa"W É`“›|@Þ®…Íß_N×BÄÚÇ©RØ0Y8„Òè*cv·&ì®þ(%•Ž÷Ë,1Ú¡üPÐ]„0ÿŠÀ¿]>`ö"”4!¹Ê•ÄÌ/gE#‡Øþ§¼Ë¨bà/S“D¾åŠ©v":I4„~Mn¸àðß&§Ây£¸E ©Ú£Æ] à;;n‹°×Òwwí:P.7Ÿ dŽ¢'¸ÇR Ñ€ ‚M“ îìðꥸü“S@‘´rõ ÔÑ ج×çK£¾SÂWsëæ¼¬”.>^¤Œdš[7šm,‹ ŒeöùlqŠ[FÓ’¾†Ió'<° f?ò=ÙM‹¬)ÍQ¹³¾xÝ¢Ñ-¥s®ï^ü¬k7éî.nê¸'&b5²€õ_CñÜj0Ÿ˜-vZá²AnŸT^)–Að%#î¤k¶b*nè² 8öóAZ-vÒ“ÇéŸF Âçq|d^¦“`B¯”2ÞâÉ/ wÄf4îÓÑG²Ma ‹@ÿnâž#»Rˆ`V0Î3íL7–ÐoW'òÝIùÀfó €sRP¿VÏDZNòªq$s}â´¬èn< k¸Ý `Íb¼,ð¿N†÷&¡¾ìÜŽ–S ºUDžNÑXh‹´ÆS®_ÉlÙ*TýÆåãñÂÚöøeåOP†œôÊÉx¯ÆvÀò±™|x¿Õü 6Í»¬´Ô–aM^0“ÂcþSÉÉøB•ZÂhQM)³b0Æ“#êñó ³»î[šš¹céËxF‰î¦]T—af’¯xqÓo+sAW¦sè9⦴_ŒÅ¸»R)AAéh`¹œË„ó'·9r[WÄD åÆLÏaÄS]˹¶ÿfïRu6G‘I€7 éøESŽ&.þ­*EwZHpÅGz$é]4Œ(ïÎ27Ê ׆ JöÑ—~; Y@ Ó˜&{=$xaöÚ[=zù,Yºó´p6|ñQ…áHËþjëô0ðʤKÑc½óñbžíô›ÈeØÂ˜.Ô?¯+½P¾¿*Íô¤w›;V êpÊë¯˪ö”ýзŸÉˆ¸þ=1u{]M)ÈÙZÌøïl (•—,a•Ò7Ë µ0Gù^CG‹xÌSÈ7D•MÕMÓ©Šè€%òåu4vä^ðm!DpWÏÆmWv†T=C*Eû—•}´»Î£_…›ðºç€# ¬s(£æ¡ËŽÌéÜåœL¤%â&×½tßnqmh(«S½(‘Ép¼mjËjNØO¯·öPú÷ zi…=H¹÷ï!¦ŒIέ’ÅÆ’wœ#ëøsÓK ì§Ô ¿®ÀÜT#ÂPox>@Í»ÞÌ®)ŒDäú ÜÊêL€?êó%X•å)³ˆ?xöXù_16ë«ß"6ˆÌ£š$›uk0†ÎRÂ*kNÍ<½Ihô»ó`_šÍ>´cöÄ1Ú¤Ô{óá&P¬µ)Û á/œo0?@¢:Õ;óÓj´—娟D[ŽªÙ4VWl¹á”#óVòƒƒ2ðN‰h} Þå~U•íµŒÞÏÙšÉ5 [ÑÍ!&<Iú+ߦ·å²ÇçІ[µŸÐX"_ºÔ'5’xÑã•a1—Lâ«é¹n yÈn"Z,Š5ôøÌ†*1mcNl´þ‡&X®¢£.ý7$‡Q?ÊNò>Ž¢áó¥Ýƒ°äÄÑ}â¢xjyñúà_òP #î!Ût‹%65€$nµbŒÙ ƒÉ%©‰ÃÔÈh®åŽœ—Æz¢Á׋¹–+ÀwêóÐTK*-9P\ð}¦aMeÍ^ëf#¦1ÑËðÉ|•œ$Q>â]ûÝ Bl‡µ¶SûIç`.„éä½kV{$ cRB¥Äqƒ(Àå>Q,ÎÊu¢Ú—H[Æåù•­Zˆ˜e$oz€ÊrænãÇ~ÕˆžX|c×,Ãz™¡FÎ(+Y1?³êèÐ2bæ£h‘*'Öк­¨ï;”Ù?Xá|%Ü™5}„× pãš숒ÍHu-ÃY‘PŸ s޵K1ÿ‹€6ªq!M‚˜ß*ÉtÍf®wäï©¥wLl߯׶õXÓÿƒa*S™—îE²Í¼ÿ‡Ã¯eAÌt:…ûnl…­Oš¯ÉÁüü)%°š+:eö£‡êgbÈ-ù(c–à:5á"gæ8žœ%ëÀ>Âa¡*æSr)ž[mƒ_ÛnvÊå½ Ã¬ Æ‹¿èC5 «÷Ûmtæ©Òºáv¬3šñoïê(µëAÐ+4èÜÄݽG~šïCÜšCÝ1´Œ–!{í¸Rä‰Á·'µ'Y¾û·a­ .ˆ)B(~¹1à’ž˜M±@‰oýÄ™¹ÙÝl!øAÚÜ++ê9ã-Ô`±èhVïf‡ðgÜÉÂÒÈßV¡5t¢-má Oº‘ØJõÿëÿžÃVUH:fm…|À/ê;;ÆÃz$KžKoO?ƒ}ï)ºùé«ÉÂÙ…“—ï@Ê’²«ÜÏ™Z¥#‰ó>UÄ1™œ õ¥Ý]€HjŒjõABÕõ1B°XõPV %‘0áqBˆÎ6‘DZ&·á¦Î„`;€(áÇ´ øç¬Ö1Ÿ½ÊË© Oø2ÆÂƒÄEJ _¦`àñxQõZítÃÓŒ¯´©ï `„˜ð«ÐÀÅ*³¯eCwÔ'gó–ë¹ÁsWê¬8 n4êáPiKù.µIBëù6‹)&NóÀŽ{ÿwrúZ˨˜MÊŸ°è{ŸåŸÑ¦õ(izNq¬÷,Ü1ÅÈUãÑ&ûl!G 3smÍ—àSò¼¶!fÀ[CCgëR¾Ža@M¸y§À龞yr.W¢º Þ½ÄÇAUO¶áž%&ä!ä|¤üu-×CþÆ…ð5YšÂï0wÖèŽò³)¹õûð»¯YÝ(à» ñs¸í¾öYŸÀçýXf<,Eã×Úk¸™Pþ `‚/ƒ`èæu'Kö¡9l\ˆ`Àuùÿ"æ°"/„xO _w.PДìŽ4šC²Ç‰aÖÆäm…¹IÖ+ýa(䨩+ké€!ÕhOø,¿v±Ök200Òb•?Äq|–s=E1ªuç‘üëãh’®j;ŒÊIâhÅ^9)ÕÅ ŒÉ?±Q]‹æòŽ…›A\y7¶ÌzÚ§‡Ë«¬Ò>ÌÕØ.³Â°Šä¥³88µOtË7?Š®¹£Œò¿+ %'ÜÉ8 ®A4u-DûëÛ¸y»P¬|£aç›iw3ùãl1×4ÿ螣Hi¹š??cI»²w#…]ßž—ŸÕV¼º¸x/5T(©]Ï”vS ³’gÿ'‡©!¾k?bx_äÌÁcúK9Íïâõ\É`twi„³A‘‚¢øµÀF¶Ž"ưC¢x9oÓaÁõ'ûÄsgíïú <;žÁZÕ¦Êëˇgq;ÌoäÄTõ?®Ï(¢g3ã#¦ÓUv7ÎŒpwj/•Z›Œ©©wî…ú( lÌé;ÛÏv ¾=*êÖEúŽF÷–[Áþ¨"Ñ%–M*ñH `|ÚRëã£)5º, ËŠy¤q1£kÖ`ßžTÿAÝ2²Ñ†¶=¹u¯ÿÒÁ›œ$‡[.šËt«“C’<ÃqÆ$E—×’ÚM¨|Ñ<)ÜɳõF³Ç]suL|~’ºñc°ãÓ!hp]˴ЬrÿCÿ’™1Ÿ´oeÏÐØ N¼HÃ"ïZs‡y,ª¤Â¶4 ÜfÿoЬžn˜Ô.1ìq–P¶ &¤E†òJ*÷9ÍYöÞ½íµ¸59L”k6‹‘NÒg»ªYî/›NãrÚ8s Weéwè@æ3+—˺ÓÀêYݬuýµ<]oïÖ¤"'²¦¤|=φkdˈXäÃó9ºÅµåv½…™äüû~­®¥19¦A¸Ò:*Æ¡Ý\ú‹ä¢fòuã[O Cÿ#ëFí}Š²ÌŒìãP±>ŒûhSŽ|ºBë¼ÿ ‹Ì y‡ì‘ªmì©H«“?I5…ø¤$´ÐK`û‹œÞlgåZe¦=ÖÀŽù½*¿²=f¹‹ÞH€ÝžÓµÈéy÷¯›<pµm!:?®_ïmµÎn¸`}}ÄBE·˜ZÉ6c Îibû¤o©š¦7rN?tÚ «®k7®ºY¶s³ãbdù½E¹€Ÿ.á“ì§ÛÓÁðÒâÄ4#¿rü‰òÙ_=È/ ÿj×->Æ—N:3ðåŸtXÑc¦Û‡ŠŽ:…-åŠÇiHÞŸæDßÁC>sÿ²1ŸÐ*Já}¼ßè&²6ªX3PFëWW•Æ !ÜÉ«QKú®2[^Á$^J@B—š"_rnÚs'Ê·=žH•‡)¤÷äA&ù^+1Š`׿6þ’é Æòý÷Ha³iãÝY«ìݪ‚ötõ‡½\r‚|U^IL{ð/Š¥N{É!er@¶i5ÆúºP Lsþ*cßAqéÉWnâÊîjÎÙÖ˜–ýŒþâ†z\\a´èG)Ñ[‹j}ò‚OúKâ†âs>r–Rw‡°rcÜòãMÎ;­9Xä{1[[Cî«­\€ó›Z(váEž\a’ém¿l8Cå¾»aPýÓ…F'‡GªpµªÝì–ÕÁ/?´ßØÛ;N%wð&¢ 4#ø/]\þøáhÏ™‹°W°EÌ7ÀG’<³<9e¶Ù"Çð™ÿ27 «‡¸®÷þV¼¡›èI16âxˆæ#†‡î–Œ%Áó¼¸-Ù•S´®] Mši8ó¥UÜP­6·]»iùâ#ö¸Ó+L÷oäý _½‰ñ²Ïºìžfà8ãÑÃ¥´R§6̽ŒaÔ&[ klDšfïpѵùyÍÑ+QúÛnó®'^€%©›õyÅ>eÃ^¤®{mþì0Z¢E¿TTð˜¤5>þ™)`—îÕö¾ «ÇÅKKf‰hóèÔâ$\¹;P„÷õv›{µÎmWÇv#&à)sèØîDT¥q‰ l,ôQd9ö×í†Ä cßþÇá ,É‘­4„kο(ð@oõu0ÂÝ,/±RæˆF ÿ÷¹÷ÍuèëG–ÝÌtÉ%ÈZ9¨\*›Vœ§þ>ÉmàyT>ètº÷ôh"íØßwgó•"ôV7òt^­}),@sŸ¦î×]ßÈãÐQv¿}7Ž®E€1išùæs›JÖv •CZÐu°uÃ,û-Ñœƒ¡›%ßóéÄ©so:šJÒÔÍ&³`ÿÒͦøtHqõ{îš«º/g dôG/‚ùßc ›xÝÈo "ÏÃwÀ⸗xLèÄå©!¤gEÌM. ^h. ðs‚Ï\®B·¿|ûå{)#rŠ¥—/€Ë±èt–£WkR.ÑW~(xd@­¿{˜L£< ^€,Í"éDêÈb·pþÍ (e;¹qtI\êRÈÎ[(­Åù?¹uŸ)ÌÉ]s¨ôæ8è«Y3‚$ VXõ×µÆù øu¶Ù¬K0Q õÚ¡î(ˆJW×yzôu¬z©¼ZS`o3=|åÏ|~Ö6+/ nh{O4Ôˆì¬h‘óžÔV±`ùj}Ë[c&³Wœ]ž°B³'[EðU.ˆSQÝ€ qÿJgIÓ¤´2ÔÎÐÐõãÐ €³öY/i¢SzBB>u„xÞªÂå³&…σŸ£Jå,%$ÃýBäzç|ðWŠ˜‘‚«ÈAj¹í(EâÚ«½ D’+™œÑêÕz´=Ø!¿žc–µŽ ý#LË2èähJÁá+ïpQïhÂñúCax/‰þE³•ȉôy<²èèã‘j‘¯ôÅíRsãçËÈêÖ5ôµTó·›“Õ/Œ-e»T/¡FK¾sîALô}œÁ|}|ö\5_΀Ÿ˜à ‘}¹£á7ô!÷aVÌFuÍÏ™Œ”Ù3gJ·^É;Z9;q¶ÆŸ|â‘ÑÜZ¿©Ø¥æhÉV·¶`Ñ#âgÉ5#\‘ÏŸ[vÈÕë Üt”ÇŽÇi§Ÿ`dͳû¤|\?]Aµ«>È·º’Í>ãGŠƒÒ—•7ÃYÌæásýÇ ˜²OÓ.µo'Ÿíëš¾è½W«NãtÛ^],÷ª" ÁùXàêÇÍc™0£º¨³KrO#¯–žiINiT- ”[9ºJžXV·’CÃ…Û¾N¦"—ÁΊJçèR«b§ãöj$Ñ£Üõ†ñ¹™LÑ`@,Ý|ÎeΫí7ê“йëð¡#»­3õúo™))c•ÿ2ˆúÂùzi®±C þ|è ´ÖKv}’…Ò4¶"VýíÒXªæK|XKRc/9©CÒT»-6Ïöæ`wcvã–+C 8,{uyÇω薲Ü鿬#g6¬Œ ±¬ÇìáHº^ "¥“ÙfãHðqå!î*)›VBBU«*hEt³®¤n;c ĉڲ¿®°ÍV•z‹±ësê ^omÜÓÍHðþhWžÁl£-ZzˆàÜ®i›•æx{ ɆÅDªÇÜÛ0~¼k¹s•Ã\š]Ê#ž…Ÿ«BJ¬9€ä\U›à^Lcë—žÌð ¸ÓYíÜ91{%ìR<„×!»Ë5%\¹‡"EH7¬b¥~Nª¢—ÊÊAƒä¶D¡­¥“xxâp©&Ü™ÿPlk* 4ó½í¤#Å/ŠÔ€çîÁŒ,ém­'§&ŒÆlíÖ¼ÍùyÒÓЗ%×e«}¼žÏ¼¶«´¦aç:±IÂݪšÂàa.ë?6tò¯=÷0ŸÞîôX"†Gul(3¤@‘¤à½t=`,1+!ŒvÕrÅs„ÃÔÕÝUÜs®  ¸T°Mï®`ž„¾æÀýÀMRF–­…þÞ)©õ] ´)¢Á1#FÔyépŠz+â ðâìnQ/Ç:w½‰ Ä>ÝRß‚º9ÙT„ùÝaE¢Çl®kýïU„MbÑ(@)Ôa¹èœÐC›Fš^‚cžq®%OÜœ\dÑ‘(»%æ ðfØÉðÈæ=ÇÔxCÂènKGû¬(}â-î ší®å{l\FobDFI^ÌÊŠ‰ãzÅø2ѽ¼¹Ã”a¥mÈoFêV‹Ç+§~_—=°í±n ü0|d`Ý·WˆWdoÎ}à D𖜠÷QçBí÷Sƽì>MeHyÁd‰pÞ5UqèÈ<è„©¨Åc±ùŒvðÔaÎA5{K4˜³›Ì©>$ÐØ?¯U ¬(—ˆK \´œ7˜–ñ÷á^-9B28qXp¨¢ºCèÑý=Μùùýs.0¦1ýUüÇSçÖãWB5 U‘ 0ð»­¥noT8a—˜ßÙ‘¨g´šSàVãÁxŒfîfµ;Hªÿ]2 ¤Ê’¿ÿI²Ù¯U¢\…ù“´.Ðéø] âie·/àÎHÙ©Vvw8Ë¥yÝñ@<f–}0Ùmhy—Ñ2bë’Ÿ,‹r¸Àï¶ú‡€;Ðgíóm3òwÊ=`np¥#NBå-UŸªbå “ ˺Ð#º3p¸ u%ב´0¶§í‡cnÓâÿŠÔÇ1U!b–»(Xe2£Î„Ò‘Ô&u­Ø‰ñÄ!Ú÷ãÁR´òˆÏÝûTÝé~p_ð²=0è€ -¨| §Ó ŽÀéÃië |ð¶½›×»NÛ?ÆAï!eC‹y­v3O/®fN"³þÊ 'WL5#èÙüÂ,-’<šµz¡æÒ}°ôO¤–Ô—YH•¤JЬ™Æðx;‘~ÄýÕÆ,Ã=°s€H5TNGÎn`5Ö‰¯¢‡¬±+üÇ{ƒ:4¦ íØézY­êË@«å_UëÀÚá>|IÎæ·{Gæ ^”H¥ƒ{ÚNdTìÂ\¨†”X<è]³Ü°}á‡çìÐ^¨aš ?†úU(e©”m©^$fZv¡ï8òî‰êGÓ:†aU‚Þóã»Ýlã9pS„yŸ½c… V l„"n"û¼¿±*n¬;碞˜h†Þ˜j¦ {†–v…iÒ3´ÛÞïJ‹,„=…t(›ï:4-› Bcz;`$° ëŒÔ¾1†¿eí,ÿh…9èZJ÷±Î%"í/A› NjlëšãÕ‡;ïìM8Ü€_‚ÅM’8«§Ö{eºÎ8ÔZ¦â a;vij­7dìv Û‘´s1«IÃfa'ã2ªýÀÚ‚-ö{£"êÕø+ñª“\—ZõâÈFf]° *¦ø_œFÞ2¡/ä7 ±§ø .©Óån›ÅA+ŸÝH‡+sn;Ç8î‘~_»ì’.¾GÄæüyÛ†Û{×l=v̈úö°Ž.Ü,ÖUD‰Ì:º˜ªÉ#Ï2yLµ2–Äáìöd÷:œâÛÑRäu›¢Ä9ûjØJÔo—œH€)Ħ ·Ã"þ±~å3•šœšøþOSÞ˜œÈ(l¬C1ɇŠ;~óÆ ÍiPM Z“SAä™~?®tkkzæ‹ãŒæPØ›/bÀžêäBOŒËÃì‹VÌIaùOÎ6äß14[ •ïIÿ4GCÞ{ CÀÚ™/K‡R j4éE¸¦2áíë‚ 9ÈSg|a܂РtÑ,{¥…V&‰‘sôm ÖT¢1êø” “°Ìý#t¾·úEVÞ³ìÊæz¿ ‹Âþ°®Mô"õ‘–¯ZÞ:é#ëAÃÙœS9ß¹W¤I$g å-[ÿ[4dôøÝÀ´“j¸ßµà3ÏÏšŠäC%ï^òõQcæO‰(ƒI³Žúê®—Þœ¾lcÒçûÆ‘ìÀ³Õ^·ˆVøRÕ†ÏéÚ~0ý4à´™9ÎëKÕh‚kH²hŠiO„ŒöªûÖHuº`¬‘WØ[”¾$<ûxSf¹ý¤ZTUrcÏ·|Nê-ÇÈÜVá‹’t¡¦³¼ÄÕO$ë”HÜÇUýb’ü’˽1'€+S¡o—S6±ìa† ÃÖQ$“ñFþ>L'޽plàîl–9™¹»€8ì·¥ †@x—>½ÝoÅ܇>Ñm÷®×âòˆý ïŠþ8‹>à­à»#r§TéèÚÐl ½åÕ7†ÿ¥¹²8Ì$Þ)щáCÄ$û…di‘ø1Jøl¨§¼·ˆ÷¬vùàú T¶O PqÅõÚÜ“\sª³^VÆ$Þ¶@ëª[³DQÂVèz³¯„²p+„ 'Z L~ÚâÔ£Žáø®t7áF’ªÛc5ŠÒPçôËc–øXwY5‡S鲫ŸÄbïcp #OèËHšÚ¬Ó½9ŒoùåŠAÝ3c~ŸÃ&+rËàúŒ÷oü™›œj™ÒfÍ” V›â¹‰¼ÄÂSË›l+ò¢: ¤››FÇ4O EÛj„ GÜ#:¬¯J ÖÌû‘óBU­†hÜÆÖ|4Ä´(ê""ÍŸsÒ{btdê ˜Ï?Ž_,±àˆ–l&¦Vg«Fn#ãìˆps¶àWóT–ÏoŸðÛzô6±I‰&kð½Ú9ÅIë·‰-Îi]EüE1ÚÀåDÛð*Xûé<,IíCÆÄÃHô´´Yð>µíÈ’÷Ɔ½™´¶ÞƒŽÚ)I²eÊØ¸Å«Rƒ;øK¹l|»ðÃÿØ%Üð•™ZèQ`ÈÆ`9cn6õq&…ƒ™±L¶â ZÍù?3G…£ O»Œks"ßåµ–aÚó7D~3…t*oQ KC‡÷R•HKÃ{Vísâ?‹È"'u…-ùä£:æ”1¦.žó¨'ÝúÌ0¹˜ÊC…‹êó®¦é0}*!w÷=Ë”üÙW„¶Í=Âië±F’nÍIyɲiŽùp¸YŒ;㘢M[:±öÕ[ÜHý™²ˆó•ùYÎzíu±Â× E‡‘XœØ‹ÿ*Mq´”)ÅéÏtâá“ W°äao+ÔѹO1|·ë¸Î›ä‰@-|[ìÊ÷rÐǹ×<Фt2ùHät¶P6ÑÉ{È€G¹HøÙ×]ЭPDÌÉÐÂOÀVÃßm£®áVe›n²uvlaÏê ‚77- ÄRûŒ¦l$É걤wwÄ— tM]å6ðóûuÖOʼnXá¡nÇ5y<2°$–¹µ–U݉T–|x2OŽi-„è»DQÞ@Šƒà´0âG«—ÃÞ[¡áF|•!ÇaŸ×ÿ¿…3í¾oÛ†Eÿ•ÖD=hÕÀ­cçˆhÍO…§{¡OÖþ4@˜ù(=GrÈäq[&Ê}Èš%NÅn±L–í镦á¾  RfáÄs‰q©Ê4§–…Óh,¾ã§Zãe÷°ºùXÏÊùÓ Ø®‘ýÜ7n:ëI+JCa1­„i÷Õ‚ÂÝ×ùéd‚;såó•=kv»1xQž£¼¨ðQú‚KAùµ$Þ"Óvý‘ÔŸ\*›ÊW®»*h‹xê2Ê™WÒl" 座 "ò¤®îÆq1e¥Aq)?B¯±ˆmR¸Qnó:~4"ß”BCޏ§'ÀÇ[‚[¢,1 è™¿Íªê‹Î^’±6“ð›ë}>E÷¬Ú_\:W¤Yúççh;®;Æsx½£Ãó ëóÝʇøg“6±ÔfÛ+Ú4bý®³Zýû7%m’év¼µ±mVæû»É0ƒv ÿ°k’&ÿ‘4¯tKÑ%ñ= ¤†ºËnô’#sJ ƒÁ§ö¡Ô5œ`ÌÙÜ/*k÷^ y[Í.”žò>ý²ùÌJȺï=<ž÷rÛeŽå%«áŽáïƒJíYëlPȦqoŸN‹ÿF}‡î~‚á4kÈ)Áµ×áîwîmÿ ®_zqÿ‘äd°l3ÏäMÕÏDÎ3ã©'¦Uä´šr äù‹üY’uÈŒõqf}ª'ÏYtr&|G µ>œÜ²Üd/6vJ!õx‚—}¬ê‡×7Å瓸’¥zœÎþIÂúÑñÚ]AgTÿà+þ˜z’»µRBî„‚hèÚ½°S(ÿT7£œº‘,<¨Ý„½>ˆkÑ• liÙ;÷Ÿ^Ño<üÛ.Ò.N`Õ1±FÎÍkq1õ©u‚öYÛrŒl{‰·‚‚ÂWS‚[‡ÇyþÚõ³ñ2lk-Ü ÿ@Néìd}Ÿ=.÷Á©!‡­ayÆùý.3S1Ðíƒ:Ò“õôß×í¬{ã=û[ëqΈTÏ>ïG­Ýîá“uø§üBÂYvZ‘qÏÅ7m aÊ ›>¸Lôgƒ¶f¬ï¯§å³Ó9TóÍÔ7…øoµñ¼9‚V¾[çJ0޲t¼~Ô‚uTº@9G…Í#‰¡6ÓrÄk„}Úâ Xy¬²Û°ô×—V|®ån¯.{ ”þõ ³ÂnCaB¥!•ºáE/©S /8ŽˆÚ!([~> =Κ>èaE¬! xÔ@„Ÿ·|¤(—\€YÁ[ÖxNoƒŒü—Fã0ùRë»ó³4·kÅeñkBÉ¡S¨y ¶Å AßR¾±+Eàœä#KÕ­JvvË-‡ö {7É‚õœ9åƒÈt†RûjÁû‘Oci¸;¬„ÄĽ:ÀÍ bSTu¾%¼Ýƒ,ÔÝyú¸²Z~E€Né\‰)qà¹j¾Ú÷FÓßÈ“¿³;t ‰tuL²‰8 !c2x&HÖörÙƒÖ²‰íd ærøêAKø‚2ö®„ú\¥FìðDì×PCðü`Y_×ò*Ž¡ïÆ^é¹d-ð–ÝKä»BÙ‡Ëô¿PX Nõ¯þFŒ^w¡¢oæ©hUÊíù§köåãfÒŸ…@ÂÈtœGî'Ô4äsö^‚Ä#$«Îˆ˜Ý ½us9S’»l|[‚\cÄ•ÔûSµ‹|ðzÝ€®˜5_ä*©ú¬V*Û¹=#Œ¸Ú Ï‹›\¹%ûœÒ%•Ýö¯Ÿ¾h~:>c–•¡Oƒ5):BÖË“ªóÁÿDï¬ /š#8ä,\c¸æ˜_š]÷ÆžY®I.|*è%Ñ/Ò‡oÜ”])Ó¿Û†QHÒ(uyæʵ >pæËVò˜ÒËf‰ãÇf=íóŒ£Ý¿Ž4Ví¯\}÷¾aÚ”¨©Ö„ÄÒ®{7™œ8‘kÊ?Fh¨f­Æ–% çÆ~Y1Ó9½8ƒTUˆ§®ÿÆá#ÇL½à›570.×W±qR) ûˆšÌ{‰èúowïÀ ’¶?ãÿ­7eå´û9¶ðJˆÛÙÄ%Çí„ìþ$ÖgÒ™ÀÍ#‘`õ÷D(ªÛyÍH…;3W­Q tû8À˃Slek ë}¸‡f¢6tèa.C–€ÄLºPò =¨¾Ù»€Ç¿ºÖš£ Þö)óÍq ¦êʾ™¯žµ§!‹à‰d¹õÙïB-ô«ÞÙ#,JøCæÞüˆ±#ÿæÀ#ä̓k)Ii:…ä¸Mb®FÜ´mzÏ^öM´ ×eý¥\TPvwP ¾³ñù´oâ¼ã±$©ã‚‹rSö&Œ4'Ž ÔZ¡\ÞOiHõ4Î]dœwŒp-¹¶L»R©®Ùá[ôP*”§ˆ Þ„ÂlqÙærŸZNeç}3@š”¡ð35âbƒý¡ž#íãÑËÃÇ5Uhý=î‘ <Ø&Þî±õa„秱ͺï‹„c„Û h\ -óýL~.ÁœÅÀ3®8- ü`Œƒ8X9í/¥™Â»R--àr@3æË*K=ë0ý”¤!’Ý*”$ZéŠü¿Ðάô K”öO]AL.‰QW2Gˬº^%•¹d2°°Xeý§Ò¯’ŒK´¥A +ê®Å–竽ˡ6ÃÛ@NžqU÷[:îcGHz=T>¿cÏ/‡ë˜ô† ¼«Aj[·³“j õ~Gà.¨ºVKj»î`ÛÊåFro8J5þ±Ã¦ï#ëé9-äÑtà°êâMôRž™&æXGÇ Ñ×››³ñ¢)6gLÏT' S«ŒrWðâ9û2Æa—bãÛÝY¹#r$Ä&ê¾@¿jæÿôƒ oàqP' 1§ÚâÕÓàþ–¨†DÖ|튮Ô.{&mme Ž]ÕLy­šýÛàgþ.†ß±äו㾠%øßþcq”½@ªAÑërßtEê´+çRù.¾‹áh+ôTVç&±Ã!µä*•eNìîÀÊÌĺwbêñé#¶,[@ ‹yMCÔ#;v±ÌHþò+]¶>ìD7yNÉ¢=£÷›!D# nc‘¶»vñú¯ç\m ÓÆÛlŸäš¬?Ø ßX“¾ñÉ·~p‹Z¸ SwÓ¥k?ž¯…C%Ü&ÿŸ7< ¼%Z>ßñ¥®|íi“rqˆ¨Lá|¡ÒKäž÷½Å#œ¢ò׸zfPb‘j¶“¶êÏeÓƒ|“v:w±T:îµå{’:ⲡS—V†a¬X!y." Þ»p´{ÇwÆü[z>©þ̹öÒ5wŸ{³S¶M×.7%62ñù “ØÑâJ-g&‡×  ?EoFª(9é'±!-ýZ#¹§¢“=´õF~ïà„rØ–›Þ<÷Ft üûr «÷‰Çð§yÏÜ5áz ³ÓÜ‘•ÝhöJõQÌî®+-Ó%œ©_<rA=Ieù½£‘~gcãXEGJyëN–<üxIã_[ sp–Ò\»–Z…öÉLV¾ÃŒ.@LÚÜIvB§®WŠ`)*3~»ÈÃú)HõÙ .Uª^ÙÞà ÿ£(;ukˆ@üK­fˆ8ÉÂÄÔa»¤¸ù(íU¤¸·œxq¦L[wÑ6á`XµØuÏŽÕà •Ìo×>0 ‹YZsurvival/data/logan.rda0000644000175100001440000000357711732700061014655 0ustar hornikusers‹ÕZKrÛF€¤-‚?ðsx‘3¤r€¬¼EhÒ¥ EªHÙÙæ\9[Ê !M›fUb()^@æó¾ýú LÿöËÇŸ‹E!ùpÿ·µlçû?Ùþêî¯Îjó¹Z‡ÐšÖ3ûû`ýúøZt¯7uh Ï(,Æx¾MÏ9a_›î¼eå´—×ò<ïƒNµ½Cú!33öa ûŸ‡c;[¢WÇ8.­àíÌúÝ<ûíæœœŽÈÄÝÅQsÔ&=º&MŸØo!ç[ãàrŽõj»Æ}Ð8;¿2’Ÿòqƾ*{º^ã¦cŠ5—‡…fÜ=…QW?¼GcʹÌEÇ/—u.¾—j‹bÙåÞáUå+/©ÿ.._Zû_mšWûµ¦óšëSøj¹Î&WSm‘qªŽ²Ðäv¶55–êj+ëW9®žSœÃÝ‘w­GÍ-dgÁÛ©õÃz°Oy# M¸*g¼ÇáÇÅBm<ÅS¼¹U¬¦úcBëÓáŸó‘âÏS=Jí×^ 9v|íjŠ{×´«)í¹Ê.Ÿ®ŽÜ]ç7jwê| ¾2?Ÿâx'õ]ó£œîp¥uãpëxÉqî).T¬*¶RgΔ}.|†tøLÅCû³Ú”ª7åÍÚš…t˜³oî\âòÀã*×å§Ò1sœÒùZk§øãìxCõqüµSçÎQ´«ññônµøºXíöOe\Y¶—Õö&>›ÛŶº»þºØÅ‘î|[-ïv7‹uèìªÕ÷Ùþív³\ìv×›uµÒOµùªÚAÙw–Õün³Ý?} ‡7¤;ÈqP¸É§ ReŽóJ.®8ùÒå¦çÀÆk¸2ׯ鈋·ŽýrD 1PáÆâd+9¸bq„* W€©ââF¯…’‰,.xý`ËBÓÇT#g½zæ¦[Köi“tM†cíÆ5—úè°Íþ0ѸC±æ:£gÅH‡ö3öS¹`Ÿ»*ÛnËèr˜ÖfØ1òacCs®Ø×ÆÂ°Z›\Ë®éñ;ts•Çœý®¹il´Æ`“65®_WðƒqÐ’}êW‹ô)X77rõ‰ýuºÙ/—g•ÅyMÙâêTu+ß886Ž ›©œ»ƒÛÇÏl/ó—æ@9š©ºNq¶âEëYë‘9XëO9Lñ¥¼ë8)Õ?SxPnRÝ©ÞÀþé‡Wª—ikL5®ì¿Öó ¯ÑØ+‡;žt|â8?Õ³Ø&ÇåÊ×î]sêùÚóàó¢ñäyƽbÖõ±«3Õ‘â,Í÷$^ç8šíÖúvg ­Uí]Ê+z†Ôxi¼Õg=c¤rë0”²ß}ð(÷¤þ±Žc¤=P㣭6º:ÖúRŒqÜv¡7“5©ZÊ?§òûð¢Ÿk-È·_h_hÃÇ燕e|¯¯~|Çû(^W4?¡=“(§ýƒøŒ«ž{çút†rºQW÷OÉ–¾Ù×7ãÐw¶cuðš2úÖ7rG´¶`¾ˆvñb[ ‰k=Ö£8Á.õúƲ‡ç‘§zÝŒæ5æ=ÊÏ4Ê„%ÉÑØŽ£Ü.éd«÷¯aô­ß»2õ3N Ÿs2¤wÈ™„fÞPšøÄñ^hê‘ØÞ£u“x)V`gu ÌšahÆwÄ{šØÑ½X_ÊØ$chmžÑÜ@ä–¡Y§œWÄkL²c`}Rß±w˜ËkÊÐôcµÍ:sµ»$3£pÈ;Æg¤{®hí˜äŽÃ¡F4^À?b‚ø°Ÿ¼XGL ŸqÜ£9웒Ͱx-C³±o÷Mh]ÇKZ‡Ø#° ïÀ¿^\#ŠwŒ—t‡~^«½Eó<ˆû»4†ù‚ä ÏÃpÌ…}ÒÁc\ÏŠ#Èà}н#Ñ;‹Ð´sš¶"wÐ5%ùˆI=ÞD¾ÃÐ? ÇüÅx»Šr˜/ÏÎw®w`k¾ž'{• ˜³¯p^¸.êuÀ'âÒ Mžƒc’Éú˜/º¡‰‘1ÉèÉ:¶±`,M(æÀôHd!3‘3uàð›ãä‰{NIôÍÂ1·º8óžú\ÓècSÚÇõ~¯Ÿ·‚Ö'èň‰;ÿqìù\S„ãOÂ1ÆPû]/hAü!· õì;ê€ùq,k ‰Å0÷28„Ïoze¿XFç棇/йÖ4ž5^Š•’d¬_ÏVlëÄèAoí…ÆY's¿ž/¹v¸ÏMH÷hæ/ÖÅû•õrç®=pP<¿~ÿ+{æ•?c]nÖ»±Kè{íë)?^Óî§t¹ùsr ÷·òíúOåï9ñ<×–su_ ®.Ïõñ¹5ýR>½å•â»ý:‡[žZû_ðtI{Þ"^§Ö]‚W^ÓÞ—Ü÷£]oáç%zȹç˜ÿS^_ÒÆsùæµúø%ì|É\¾§±_gú×ø ¿Ùt×›õO¿¯ªùq _žþ=ªù ×ÕÍÃO[õòvÀObóù—Ûêîzƒ_ÀÚËýT/>}™7&·Õ|!r»ÛÍŸ »þèËÿÚÿ¹¿ÿçŧê®ú°Üî·<š¾ý ¥dŽÊ6survival/data/datalist0000644000175100001440000000044212545056257014617 0ustar hornikusersbladder: bladder bladder1 bladder2 cancer cgd: cgd cgd0 colon flchain heart: heart jasa jasa1 kidney leukemia: aml leukemia logan lung mgus: mgus mgus1 mgus2 nwtco ovarian pbc: pbc pbcseq rats: rats rats2 stanford2 survexp: survexp.mn survexp.us survexp.usr tobin transplant uspop2 veteran survival/data/cancer.rda0000644000175100001440000000664011732700061015002 0ustar hornikusers‹å\Ml\Õ~ù)U¤f‘EVE(ª,Ï÷Þ¼ ¯ $ˆ ÐReplçÇvl’”1ù¡ ¶Ë”vÑ%Ëî:Ë.X°d™ ]°ÌªªÍû¾szÏøö>;–¨ÄHpîÏ9çûιçž7?¿|à¸{øøÃI’lM¶nÛ²òÏÊpûÖ•mI’­ÈÆ:3cãóI²íg+³UÍŸ®üóM¹#ùþ3 wUòÉï0rä>³ÏõÝ#¾^™ù~Ä.÷õe=3ö»Í|Øð¢Ü ¹ß—6ŽÞ#}òö×e—áµ×ðÙãK±cžÉ37ë´ÏŒÝî¾±àoü _{î̃õËüï7úÔKŒ›Çݾ¬Ûó²õ2bìs³oÏwŸÑ#¿½Fù°ù6yà¿?0§ý£ÆŽ<ÍyËþ^³oÏ×ò`>XÿŒóQãÇÞOê±É×Ô½œ‡Í»åmëÊî3ï6~›Ÿ}þºàZ^æ>œ/õËJ¾{oõß~.~—­KèË9Ø~cûƒ­OêYž‰‘¶¿Ù|ÙùžÀº‰s ÿù¶uÈoð\‰cû×­®›ó.Ë^%y^ÌSæïÜ7úßÖ×ÃËÜ›wámúLÂçdw¨Z¿4[ÉoWòLeX~ð'¬WŽËî‰Jž‡ÞYèìWró«ðwëG!—ÿ^É·ÀçÊ#Ы–Gz•|ûó_Uòü^Áþ…ê Ë¥;•ìa<ß„ÞøòÜ/ƒ7ϯ÷·J¾~‡`÷[È[ˆã#ð| 8׿®ä%Ø UòâNćù{ì 7~=¬_EÜ7¾…ß{•œ¼ãó™ýµ›•ÖÝ è][ªä"â¼»³w+9yv'OøþÎ`ýòòì§ ß‡^q².‰?…ýwÀÿzuaÊó_øú7QÌÏ)ø9 ž·pÎsØ¿ o÷!?Ž«ˆö¬£ðyzOæ¬ë˨7Öó2ü]Îy¬³^&°>Þ½›~œïÀß"Î÷~|æ—ÓðÓÞß3=}ëÌçEø9…x—‘¯ÃÐ[FýžÆþkXçù/£{¨»ùQè—ÀÁ9¿ þ³_øs¹‡—Pw‹;ý¸Þ†¿³ýJ¾üKàÞàñ<Ÿ„Þ"ö§ Ïºî‚Ïæ<‡ðày̓Ïûô–à–zÀé`þ6Îo¸§°¿;9Gè±_-Âþö/ÜÅ>ò6ü ÈIøeߘ¥?øaý/ÂÏ\âó] ù"Þ_Ãÿìf!ÙOOœ¨ä”Áå|ûs¨£9ÚÁÏ+À}83àÅú™†ý˜OaŸ÷súN%Ç!ÏÁÏYÌ'ÁƒuË{Á::…ùüûóëoB8“w}~ãð‹{ð9‰ ¥í/ý[~ÌõÈïå¾'yÒŽù ó-ñ™uò˜Àœùf}<9Ï•ßcm¾C>½g¤?ÆÍsâþ@Þ ynR—¥¯/ñÛ¸¡ÿ¢ÑcÄ!âðç8æ‹qñ臼%þ>íY/¬{9‡ÞÚvRo6>ÃuÆ8_>Ï“ñðŸc\ô<ž;óü6sž3õ—­[[7ôsŒ8Üïùñ²hOž¶®©gû ×¥ÎJ_y²ýð9HÖ!ãáÜæå'㑼™}æ‹xOÇàñ~=K»ÄÇ'ÿCfÿ¤­æßöåQcwÄØñœ©O¿œ3χÍ>ómû¦<JGú1äر®È{à9Ó÷õX÷Ì+ïó‹ü%xNnôyNy¿z›õ}¡.N]ÿ[¹^þuãØ(õ~_ÞhëòØìsÚ¬ün”]Þ›uþ›U?1{»^o½þî·Þî·N7»~6«Ô½_uùÝo¹ß>Xw}£uãY×Obž“öc××ËÃÎùÿÕ7Z‡!^¡õ߈ēuÏKðv!¾¡xí'´ãµÞ~íë?ÆÛîו!1üÔåóóS—Ÿ]§~HÆâ‰ám”W o½z÷{Ïí:ý×Å Ù×åµÑy]¼Ð'Öbø!ܺ}©î½ ñ á‡üÇørßâÔí#ÿƒOõß]_íUz!ù:üÈÏÉX·ÿ}“ò˜YÕè3’ûôĬS<¬ß_‹âCiã³s»nã°¼mžBq…â°û¯ôm^¬ÛîÛüX}úÙ…â ÙÇøÇ¤Ííc|í~쬾ÍO¨BõhùÔ={ßBçlõCx1üP¾êÞçй†Î9TçœÇc¸6®Ø} ñŒÕ‘õÊ£;doqêÞwgÝ{ŠÓú3¼å÷CqÇpc}&V_±ºœ{Âçdݸcç{NÆú°½ß1ÿ¡~gýÙç[(ïGŒ}è¹}ÀèÙþeóû^ª?Û÷cýn£y3ߤžCv±ûéûAÿ±{]÷9âeÏÑæ+ÔCù·ñÄú¡ýÞz„îCèùª‹Øý }_ £­[Þƒ¯ÐýˆÕ}Ý瘵ãúûÐ=­û\á×­»Ð=ŒÕE,_±çpèyZ÷ûJì¾Ù¼„úYèêÙs‘søÒÇ%/¾O3^ñ.«´$ågý|ðý á9yÏçA½aoë‚ç6×ó×ço®u0ëë]Y3Î;â°XOv¾løßÞ½6¾pm·ŸG¾GµxÇçI=æyeþl}Œº¾aâ…ž­'™woWvÇýü‰_ÆËóä¹3®S÷KüÜ,Ö´“ýØ90Þë}»Ë_Ïô-ÖÏ_öú’çIÞà+þÞ¿ s_è—üy¸§z?·{sñK>”̧±¸ïÔc±÷z¼Ç&ÿ›v¬O¬Û¾[~ú/ßߟG|¼+_{ù ÓÀ:ýò²þ|Ï…ýQê-€÷¹yÎHÿ¯ê<ÁsRôù^—y¯Ó~Þ×üäc-È]%$ŸãÀésŸïƒÂ¿ðj@oÈàí7¼ð~hŸïפ˜ïñçÿøÎì#Î>ã¶øc}'dáóx?ù1cü>íž0väOÿ?OÖüÈ{ÇÃÆ?óç|¿}¾ç;yÔ·ðO=Îw˜}œ§äï—fßþ½‚¦¿ß§óÆ<ì2ëC>ŽÔÙ/Ìœùâï«%†ëqÁ^Þ¿b>ÛÉš9òb\;^éûgÞÎÇþ ®Û¿+@¼ÔHæ‹udî'ýò#çß0~ì½±õ}Èè“'ï–ïa?Õóü0Þ¾9Ï>ïëÓCŸq%Fòœ3qqñCþÌoêëÑØ3/C÷)3'æ“ïË›:”ä<Ù¿GÀùH½ØóA>™Wê±Î­žÄ=äãÙ{0ð¼@\äC¾â?/ãÛå¯Ûÿÿ'8¸—÷„}‹~[‡ïÑ2¯èw}ÎyÎæïl”kÓ=©×]kó6­ç'3éñîÊ`5«¬guqûé™îÇ §§Ç1~ »ÐYXìb¶­3ÉmÝñ% œ=5<>vnÓ+Ó³ù™s˜?4ÛYðvLw¦†Ç:S´¿°0ð˜½{Ñkò?þ9ft®|'±w®ÿ}=^u;Ð+]|Ž=}'%ÿ“·Ð3O¼ðÿqÃ;jz‘à‘·ÑŽAŠÏúÑt=Æ™¼=fóĽÈKæ>za'þ;Q¤xÔ ô ?ÿéGÒÏôÄïø-B“ãûŸùï›^q¸]‚ä yåùrØç[¤v7úc†‡¾qß þ@óø ?có?ÆUãz=õœ}Žÿ¡ÇNü€=QAª·ç›÷ß—Àã¡ó8cç$ò‹Î÷1óSȤú£o„šï Rw±ÇëóØÿWÇýc|àK¼©ßÈý=5&>Øûø‚ß7|ßçàÿ¹ÙÅ:úbÛGü¦M.xØGÞüà~›'È«x^@n€_÷ôyû»ÖÙÏèåõÚ÷cìsE*ÿ¢/rðôÇLOê_Ïæáƒ€øgº ýÐë"cüpÀø“Ï~îp:âßWÌîÃé|ow‘È‹ônú£‡Ç~œ;¦4Öù!ú›>sØè~`ø^Àó~ ž÷Qêv¡—Ÿñúaqózûü ]'ÿ^±yê?û¹lŒø'žÿ¤·Õ/ìtøTüðS<çõükìAï"µ?w®'b_@ÿŒ]Èî¥ô¾Ñ°uøx<ü<ÿÙÏ^·Ø×øerâž>Hñàÿ€1ï±Û ÷Yò’:ý&ÏÁ$¯Øø>äç,vÇ{ˆôö:I½§¿ì‘êáç öƒßW^*ÒuâHÞïññó¼ï{ïw ‚ççHüãûpÊäQoöqñü&Nèõ—Æ;ãýÑôðzäúKï§Îû^÷=®½"]'¿ü<žÞ7â=Tcò–<´>õõóAê©çù0­1ñ4yä%þ"Ÿ±ÛŸO 7û;ÉkòùØãýÿÆzü(Õ‹üÀ.×Ûë¶ß7ÈèüŒ~ñþÙ^aÐô…ç}èÝž#¶ï›ð±ùøüAcÏüÿ^1:ž≽SFï÷V·'>w(R¾ÄÛïì7ì真{õR{€Ño‹õOð{Œ±ÓÖ©QÃy'xÔ ÛÏ¥‚ø~âu}<_ðú 'úøù,ê‚ô!ò†xÂç€Í#:ö‰ß·ã}zâ‘WØïñõózàOß¿~¾ôó©Ó÷Šjyø—>=çHüçñ Ïæîû¹>ˆüøÜq‘êíùóÛðžò» ŸÃŽ^®~O™mÝïq1Ø£ùÆ÷Ôs[ðÕú5øªÑS‡<¿‘Ëý&ÞŒ¿ôàoúx]Äê~ˆ~¤ô ¢OÌGãïÏoüü÷·æÉž‹Ç{À ]w¿2zþó7þÂÏ~oÌõóx®…ŸÖ‰ ||¿øý>Þ÷ÛºË{ª›ü£6&¾öü«(ƾ³ùï/Åæ¿Ò|ý-ÁQ[«Â1|ÿ8M~}ŽW•uüëäÖù Çyäì¨ÃÙª¾mõjãë&?6Ź.æMqÌÉjãÛº¼­ÃÍùºÎæ&»êüÔÆÎ6ùU7×”Ãø¯MmxÔÕŒa>MtÃòmÊ6ùÓ$»í>mãŸ\.ÕÙW·ï«r¥MžÔéß6?êò´ìaâ™›«“Ý´êlÙ®^uyWç«:›êèÛî§œ¬:ûÛø:÷aýÞħMÕíÙœ}M9^Ç«MÞ4Ù2LÞÔÙÑ$·Mn·É¶1hãûaø·ÙŸm}ДwÃè0Ll‡‘·Õ|Ýê¾iŸ\ÌšüV“Éeö쥹ës—–7¾Mü/V˜¹Ì—³†ÿÌì¥þ2è‘ÉÙþìÊÕ¥o¿+]‘ÊªŽ—>#%Xþ°>Z¢Ùœ³õqÃ-óÍ…Ðé\fyìò]×±’œñŒ~#F[žsë|Qåt/AçQæ=®·Ë©š­ q[ëâ[¥[ÙèÕ¤6Uð(ãøœëé8m|>^A3L ÉÉ©ÊÏúœV~,ϱ¦Ü,ãø^¨+É9}ÆL^UN•ó¸J¯²¹=“Qåw·5—kUû`¬‚>Ç?ç—*[su#·ººS—7×¹n¯äò¼Í¾Íñ¨«±9y^+Ëu}´‚çˆáWÅÖkH.7šô«ŠIUÞŒUÈ©‹—ëRþ^×+ªèêâQÞ³MqlóÝùÕíi÷Ÿ×\ ʱôš4^!?§çhñ´í^£<žUüŸëßåy×ÁsÕå9^UmÉåEU.är»jär¬J×*|¯Ž•d—uÈíõ¶µ«ÎÞ»ÚÒVù¨J·¦zRž£ÿæz×€œ.e¼r,Úœ¹|?ÕíýœÏs=³,3+÷…× ·µ¢/n^yF¿*Â-g¼Ð]h_&õetßFŽ¢1ˆ*¯Fz{ç—/›z¿Ü!ØÜecàƒÏ=㽂ðßmtðëØ:ø»2p‡áwly×y/ØúnÃÛexðw{wÙ<ô; o§áåìe~‡Íï4ü½†×1èzº{m>g§ûÉóÁãý¼ñÙkyoNï~@~Sží0Ï'¿Ç×ý²Óð<Ÿ:†·#ƒß1èùÛµqtý]·ßó×ãìtžŸããúî´ùœ\Þû>vý»†çþð|íÛÌí‡ÝFŸÛϹ8¸Ü®Ñï6zæ½Î@çõÛõô¼p>îç î1ú\Ýw;nÇÆ^Oº†—ƒ®/óŸ¦: ûÝëÄnƒ{lìzw3øž/Þܞ׎ïë¾<º6ïùâyëùçr¼^8ÿÜ>÷ýëóÞGsùåñöõ¶Ðë—¯{ž¹šú™¯{~5íÇó¸x}ós£ëÝ1¼ÜºÛíÐýït¹óäï æê¡ëçu>g§Û›‹»¯{=î2<‡^ǼäèÜÞ¾×™Üz·%~6éÛwÛØõñuïûÞGsö»ßròsuÐû¼ók‚]ƒnǰ0§‡÷™t<Ï?÷oî¼âø9»ru*Gç}%G—;ï5£s°ÛÏó­#˜»?äò;ç‡9;æÎ^®O®u nÕÞÜŸwüœÝ9Ø6Žmíðó‚ãµµ¿ ßínŠGξžããô¹8åæ› ó÷óvNÿ]î~™‹Ï°v´Õ'—Ÿ®—ãçìnçøËØõ¶Ðãš‹sÓ½ªiþ›‚Ûµ¯ n×?måü¾¡ïËa׿)}‡Õû›¶c»òÝîÿ‡eXðï/ÿë0¯éÞ½/= ð³ÿ ·€VòÖô ‚ý™ÿXãÛÝgÄgy*ÀÁ£y›éEÁ“½¢ü§·8•êqQpBë¼½ÅÛUW¥÷uñŸéHÞj€sÁk›À>ÙsFð´ô<¤õÝ‚;ç¤×ÛÇ_7`]ðœ ñÇO¼…¼,ø¾äžß7‰+ößÓ¹üyä \xœêûºèy õ¦ì9)9·Äwi!Ào ¯#ÈÛ”×µ¾(ù·øSÉ;/>GE÷ÁÊÉé†ñmË·ÅwEüøµ¢s3éø¼ôï ÿ„æ‰ß«šŸ—Þ~_‘¾!¼74ÏÛ“«Ò{Ez—Ü÷„ÿŽäÎ QãûZçíÈ+‚äMÿI€û…G¾Ì?LõxÙÖÁçíÅ{!€Ñÿ³šŸDžô9-¾'‘£ù+‚Óšg?, .?ðšÖyû›ý‚ŸÞ”]äãYÆR}ÉcƧ„Ï[•wOkþÀ?îGêÚÅŽøkL~û[ùKâC½Xè¥òÑÿ¶ôä-MöÁ[šÿ± uøœÆÔÉ™™ó¾©yöÏÂCñý”ôx_cö#ú¼+H]üDõpªHõYšHç©+ÔCöÇb'@ú uè¤ðx[õ”øö¥yûÖ •ó:°ßú¢Ã?òoŒÛË»+ùäãu­³ÉWúà¤ægE·þÏò¶?ýï‚Æç¦zçD_õðéË[Áä}z9(Þ»Ó >H½œ=ý‹¾÷=ãƒÿ~7@úó%—:Ê9á-Aò ö;ñSŠ~}M°^Ћ?ñ|±Ô ¼x]ú_y’ÊS¼{½$ŸúE¾ÐwÔ7{w{)Ýëðy”ŽÉ³á³Ð ~œG:Eªÿa€k«éú„ q™“ÝôÙµ©©³Zç\…ArN²¯ Aúò¤Öéwž÷ätï âÇO%Wç´¸þTxÔSöùD_ÁÞ{è#(qsùF]š×<õŸþp?ÆsÄŸhž~D„?qbß’?ðÃßüÚÂ5ÉÿůR9ô1ü¶4•Îsn˜Ö:ûã‚øá7ú}< Æ>%¸òe€³âGßY\ò~~Æôïˆ?8?Q'á§þíëkŒÿÉwü;÷$•K_¤/Ÿ=}cQôç´Nÿâ9«õ ß‘]Ô1ú!çê4õ’¼cžó,ç¨OUOé+×ÄWxŸ=Nõ»Ù ¸/äùûÚ µ—söŒÖ9]—=ä5ö®Jï×nMý™óu•¾LÝYí¦úÜ’‹ÏÓÇs~'@ê8¿îð®ôˆ¿¦'ùË’?/:òn鉸ˆŽóõja!@êF¬‚ Òÿ ÑQ_ÉOö çˆ÷5ÿÑj*—{Èì“ÔNîœ×é·èAŸr?ûPzJÞºì½+>§%—¾²öE€×9ÏM©žgÅ:N>â‡Uѳßx”Úů ]ê!ûò”æÏ‰?ý{A|–%‡¾½ ùôûxoVþqîX×ù}V|è?äó‹Áî¯ï±EʇúÆ=›_çã|@<èƒÔ?üÌy™ºÅ~âžIßÐ<ûäšðÉßiáß$.ìò÷ŠÖ?ùm€7‚añÜÉyæ5áq®» ˆŸ8ú¯‰±ÎiL[ÿ—©×œSØ—ð£nߨ»û?÷íó¢ã~öâʽtW‘êùÁƒO‹žýâó üû|\t¢çu_z²_¹/Q§xîÖúï?(ŸéÃáç#7öïD€øéï&R=‰/uácéAÿ#oØ×ŸŠ纫Ò3>ï>÷Qì¢.ÝÒ:ûùg$ý± yt®| ù5ÙÉs,úÿõÇþL|~~(åã ñ—\ö+ÏCcy ù…?’½<7¹!þôGü½ ù·\ÿŸë¹ØUg„O}9£yê!ý›suægÊ{Î=Ô?ž¯ ‚Ëž{]“žüD烟 Eò¦e×'¢£ÿ|¬çô¿uÍ/JÎ%é½à]­ßÒ:ý}íKáiÌ9ú®äѧٟœOîhýCɹ£ùË‚<7øPÏ&5æ^ðÑO¼§ü¹¯~Å=aq&ÕgY|ïI÷µ~ìßìOêýMᯇ}ìz÷º¿5Õò™óúšê+~;/ÈóÒÁãÙ—75æ{ïµö>ûìû ¾~áàñüþã÷gY¶=Û¾cÛê?«Ã{¶¯þk[–íºom2µ43™e;~º:^ÓûÉê?_;³ï>rw)ýó ÷Bî—}®ï õŠfèÇìZ¡¾­7Å~̇…å.È¡Ô8xDöÉo8\·ýÝÂkŸðÙJ³cžÉ³%ë´oŠÝž¾ØðÆWÏyP¿Ìÿѧ^&þ5{B;[×óÒzû–ìëùî=òÛ'żæ[ò2Àÿ@dNûÅŽ<å¼mŸìëù*æƒõÏ8?z?©Çz$_©{;Í»òÖºÒ}æ]ã×üì× WyÉ}8_ê¥|ûÎÚ¿ïýÌü2.­KèÛ9h¿Ñþ õI=噉Ôþ¦ùÒùÞȺÄ9ÐÿˆÇ|kDò=Wâhâºú纜wQôJÉóbžšáþÀ}£ÿý©~¸ž /¹šwã-}&ãsra¨\¿<[Êo–òliX|ð'¬—Ž‹…¥¼½sÐ;Õ/å8æWáïÖB®ü½”¯ƒÏ• W,ŽôJùöç¿,å1ø½‚ý‹åAË·JÙÃþ x¾½eð广Þ<¿ÞßJù*ø‚Ýo o ŽÀóeà\ûª”—awr¨”—v!>ÌßéÀzóà×ÃúUÄýÞ7ð{§”“·B>3°÷z)GÁƒuwzï.—r q^‚ݹۥ‡<»S'Bg±~yyöSïCoq².‰?…ý7ÁÿZyaŠ Ÿ‡ú×QÌÏiø9ž7pÎsØ¿ ¯ƒ÷{CaWÿ4ìYG3àóô!ÌY×o¡ÞXÏ+ðw8°Îz™ÀúIðî]ã|þ–pî¼÷à3‡¼œŸxðþží…ëóXg>/ÁÏiÄ»‚|†Þ ê÷ ö_Æ:ÏuØCÝÍB¿ÎùmðŸý<œÛ=„¼Œº[ÚÆõüë—òyà_ÿ ‚Çó|zKØŸ‚>ëz|&0ç9,‚ÏküxÞ¡· ÿ³ÔÎæoàüf{û‹°³s„ûÕìOcÿâmì#oóÀŸ€œ„_öYúƒÖÿüÌe!ßEê‘/âýüÏÁn’ýôĉRN .ç3ØŸCÍÑ~^îsÀ™/ÖÏ4ì_Å| û¼ŸÓ·J9y~Îa> ¬[Þ ÖÑiÌ'àï,ðØç˜ÿ3X ú“À™¼ò‡_ÜëŒÏIL(í9[qý‡"hñ_u þSxåóÿ¦ÿ}Ëï× ½Ø¼*Îfñ·ŠwÊ>6¯Ê¯*^UÞ)žUãØ(ï˜ÿ­â]§jüå]/ãsò…^i~ž¥Ä:¿wª>äÏ;ü9ðYÕËB?Ü…ýQÌŸ§}DïiÌŸÁþsäEáË9ùÿsȧ…ãà÷]þœ5*~F‡þø½Êx@¾˜…öŒ—sÆÅx¸O{ÆKÿÊù¢ù½Ðã$OÚ1Äa¾->Y'/ó‡ùAÌ™oÖÇ3˜ó\ù=Vóuò è=)’þ7ωûyƒä¹Y]¡¾Å¯qCÿ9ÑcÄ!âðç8æ‹qñ臼eá>íY/¬{;‡ÞúvVoŸðc1ΡÏód<<ǧýçÎü0/‡eÎs¦>ãÒºÕº¡ŸcÄá~/Œ—õ@{òÔº¦žö®[¡ó¤ýð)HÖ!ãá\óò„ÄÉx,o²Ï|ïqâïד´ËB|ò?$û‡ µ˜íË£bwDìxÎÔ§_ΙçòÏ|kß´çCâX?†<;ÖyÏ)ïVo«¾/TÅ©ê³s•å_5ŽÍòØè÷åÍæ±*­>§­ÊïfùWå½Uç¿Uõ“²×õªxõw·õv·uºÕõ³Uý£êýªÊïnûËÝöÁªë›­£Ϫ~2yNêG×7ÊCçüÿꛭïØúÀï?$âHɪçeø»ßX¼ú‰í§xm´_¦p«úOñÖýª2Æ#埪¼SþR~ªòÓuêÇd*žÞfy¥ð6ªw·÷\×é¿*n̾*¯ÍΫâÅ>©>ÂáVíKUïMŒW ?æ?Å—ûŠSµü>åw}©WêÅä+ðc?'c]ÿû&å1YIô‰ä>ý‘uꑇúñU¼JO纮q(oÍS,®XºÿJD_ó¢v·îk~TŸþSv±xbö)þ)©ù }Нî§ÎAõ5?±zˆÕ£ò©zzßbç¬ú1¼~,_Uïsì\cç«sÎS÷?UÇ)\+u_bžô-ÖÏ_ö…’çIÞàkþÞ¿ ¹/ôKþ<ÜS¿Ÿ÷sóK>”Ì§Ø Üw걿è}§ï±äßxÓŽõ‰uí»Å§ÿ ýýy$Ä»òU¿Ø9 ¬Ó_$ï1{ãÏ÷\Ø­Þ"xŸÉsÆúYçž“¦Ï÷ºä½Ný ¼¯=ùKȇ Û»! H>ÇÓç>ß…ãUƒÞà^x?´Ï÷k˜ï çÿøVögŸq+þÃXßÙ ù¼ŸüØ¿O»GÄŽüéÿgÙº{ïxXü3yè·Ï÷|‡ †vþ©ÇùNÙÇyZþ~!ûú÷ êá~ŸzÌó°[Ö‡B«³ŸËœùâï«eÂõǸ`oï_1ŸÝlÝy1®]¢W„þ™÷óÑ¿CÁuý»Äkˆd¾XGr?é—;ÿšøÑ{£õ}HôÉ“¿wË÷°ë~o_γÏûú8äà3®L$Ïé!‰‹û{Åù3¿P~ÌžyÜÇdNþÌ'ß——:0”ä<éß#à‡|¬^ô|Oæ•z¬sÕ³¸‡B<½Ï ÄE>äkþcñ2¾Ýáºþÿ?ÃÁ½¸'ì[ôÛ¾G˼¢ßõ9ç9ËßÙ(Ö§mzV¯»×ç-«çG3cÓã «ƒµt¬ý±žïþ€Ï™™…EŽÏLcüã…űťÌvŒMrcÇÂø2†÷Ξ?y~Ó«Óscó3ç1¿ovl1XØ9=>65|rlŠö‡§Î/,(Ï“Sc ä¹ º÷Ÿ[ž˜[c¨ß7þâð‡ö5L¶Õ8È9¨sÐà ÉA‹ƒ6ºl¯Ø¨f£ÜFu5lÔ´QËFmuld¹a䆑Fn¹a䆑Fn¹aä†Q7ŒºaÔ £nuèFÝ0ê†Q7Œºa4 £a ÃhFÃ0†Ñ0Œ†a4 £aMÃhFÓ0š†Ñ4Œ¦a4 £iMÃhFË0Z†Ñ2Œ–a´ £e-ÃhFË0Z†Ñ6Œ¶a´ £mmÃhFÛ0Ú†Ñ6Œ¶at £cÃèFÇ0:†Ñ1ŒŽat £c]ÃèF×0º†Ñ5Œ®at £k]ÃècGmdć5æ>¬û°áæ[>lû°ãCG«9ZÍÑjŽVs´š£Õ­æh5G«9ZÍÑrGË-w´ÜÑrGË-w´ÜÑrGË­îhuG«;ZÝÑêŽVw´º£Õ­îhuGk8ZÃÑŽÖp´†£5­áh Gk8ZÃÑšŽÖt´¦£5­éhMGk:ZÓÑšŽÖt´–£µ­åh-Gk9ZËÑZŽÖr´–£µ­íhmGk;ZÛÑÚŽÖv´¶£µ­íhmGë8ZÇÑ:ŽÖq´Ž£u­ãhGë8ZÇѺŽÖu´®£u­ëh]Gë:Z×ÑºŽæ½$÷^’{/ɽ—äÞKrï%¹÷’Ü{Iî½$÷^’{/ɽ—äÞKrï%¹÷’Ü{Iî½$÷^’{/ɽ—äÞKrï%¹÷’Ü{Iî½$÷^’{/ɽ—äÞKòïzÉ¿×þùE«“ýRsurvival/data/stanford2.rda0000644000175100001440000000440611732700061015447 0ustar hornikusers‹µ™[lT׆ÏHJüP¡–‡‹þ—Þ4¸kØàë_«–“x´‘{¿_ÃÉ›ìß²Èù1Ðlpï<âÅîÆ÷_"î 5 þÊlxºn#'ûm‘¼À9G¾G¼C‡ Êú‘õYÄî΃?Eþ•[‰¾g‘Ûý¾Á—ˆ÷pÿôà{"gðà~³ÂCÈË>*çÑÀBæOáoÑàÛ'*ŒÀ'ûõô¼jðùùÄÿ#ƒƒÈRg9eŸ‘óz q÷í#þ!ƒÒç}Øë#ÿÏ£gòû˜ßÿ+ÔïçÄ)}#çœÜä\ëwþÅàè·¨¯ÔKÖµì7²ÏýŒ>z‘~ÄÿÁcÈÁ'çÝΜÁgLc‡%ò½‹uwà7øa íÀäå¼ØŠŸÛñ§h6’pÜc@¸—º¼L‚GO”}IîU½¬§^걺D~èËÝôßôÿ‚þ9‚ý_Ò’7ÍC3ñRÏ™ù¢ÁW‰[Ö¯ø%ëIî'r•ûÄ®§‰ûrŽÈy,öå^¹zÛq1~à]ƒo¬ªÖ#ç±øõp—É“îSrNÈýh| â§ôŸø!û°œ§rÙM¾eŸ`ŸöØÇÅž`½g:þzôtzêñÏt|:¾ÙÆY/îÙò5šÏFÇkŸFçgÛÚ½Qù¨G_oõ»Ññ™òÕ{®7Ž™Æw£ó5Ó:ÏÖ¯Få§³7SùÙæï:â3÷q¹ŸÞQƒr/“ó$^ƒ‰:˜¬ƒ©”ó'[ƒwÖà]3Ä»kpq Þ3 Ê=;¬‡¥j”ïkS°ô¿Qî…â’™b©1”{qÃXšÊ÷¹†±43”{ó,Íå{Ãuc鯠Üo8–þ¿¨÷ñ™biv(¿“ÌK¡|¯©‹¥j”ßjQ~Ǔ߅¸ß{r¿¸püw{;¾|~¼iqïÇ\ùú•׃–½³à™–ë¼¹Ö{Á§=•'øÄŒ+ÿXëèsßn ¾,‹-{GÆk18Wfký,8Qa[ ã_TÈe/šç±óyGÅPðaex‡ò_1|SôŸ®8¢~ÊùuªÌÝûfðqÙüèsÁU?‚3b¯’†¦à{+†Cù±?•Ÿ?×Lb‚÷ˆ÷8üce/N§™—|\MTž` ?Nšx¤Á1â¿Z?õoü`ù9¤ß×<ã_ð7£_ù/¢ç¬É›Žÿ“<žÇÑ÷…øiô 8—:H^FÉ×ûðÖôK­½òñwä&ˆã ú„¾lò`ëbô“¯SäçôòRßKÌŸGßôeú÷*qŸ®Ž/ø#ñ_©Éß±êüiÿ >é Û×Õý©ãÔéLuÜSøþZ ÿ[š¯Ïè·ÓÔAòv–~û„yé¿ äíqÿ}¿e=HýþÅú“>{ä7ôà?‰ü è ð±~Ûɷè;gòiëOŸü=RÇ‘š¼A¿Ä)ûƒÔóøyÖ¬ 'NAÑs =ãØ½(ve}2‰}í²Ñ7¥_Jže]Ò²žÉs8ÏÔE÷ é_Y·çÊfǯ~±n‚“ŒZÝ’×ÐCo¤&>‘G®æ¡7=•²»ò¥üëVùŸ åAk·¹Å­Ovó~s1_ü~?Ôœü™ð‹©Ú¯ömß¹ÈÕ<,R±D‹¾¦ìk6¦¯1õ3IŒGe°Å E³úšQ¡dš²òé¨eÍXÞ´}µ£õÓv>eÄ’jÖr¦lT:µ®fãöÕz•N(« µhÐé¬Úqb¶á¥lúâbÞÏØ8¢Ö§XFÕ«ËZ?aÕGu:æÔÁɘ£SÌûY;¯…ðÓª>î¤Îr¦ŠØ4'Ui¦)éĬ9É8-×AÓ)­ã^Ü–!m_ã6‘¨Ê´c\K“µùVYÛÂq[˜ŒÍWÚ:ìÔSsï·8U°6£šÐ˜º”M§Øê[ÊšLÚ(œÄÛÚ$4q1›÷¸6¥Ó?¶»­tÆ©•]“N‚Õ¤Sõ¨]òšÍ¸¶_‹ZL9Ío_“o±£âÎÚˆj…â¶ãth@Îv‘²:“NÇÛ>SCI§¹Å»ˆqLÚÎÖp“Ö¢¶»o“•²ÙPa] ¾®JßÏ)„¦%¢.è~ë;›°³:­‹Yë}sZÙÌvªÓMv_ÌÚeµ=ä¤Ô&R§míœ^v÷b1±»¦–¦Åî@:–°l:;¥]ÄNácvEé›6ƒ¯§ˆŸ²)p›A95ĈX‰èˆmr[BÛãi»·¨¨=´¬º©úº½ùå^¬>ò»¶åûå`ŽÀÖTÈó‹zúò•ãþZùó_kŽF%survival/data/kidney.rda0000644000175100001440000000174712545056257015053 0ustar hornikusers‹Å—Ík\UÆÏÌPºpᢈˆˆ„ÉýLýÊ­MMƦÕMÀ ½N&é`šÂÌøQ\8þnuåŸÒ­ àÖ]  ¸QŠ Dïä¾Ïƒç‰—¤m¤ÎüÞ{Ï9ïïœsOÎLÞ]Ù ç6çœsMågY\«Y~4\Ë)ùØGý­½Þm炧ʻٲ+<+|Nø¼ðá‹Â—„/ „má¢0FÂX˜Sa&\ž¾"|UøšðuáÂea®û<ï„cŸo:áØççÑÙ>ãûn±xoX/ÌëóIÅá¾?¼w¬ßÞ·±¿°¾˜øýÖl|7ŒØÝÜï?ú±bay1îM^ù‡VÕòïTÄ{Ä~š7îT‰òmãÀ¼×¬?æßÝ÷Ç÷>ž¿üªâm˃uÀ:^±|Œ«öë¶b¼fþ¾å¹ní»F¼¬Ÿsþü®[ݧïq\D^É·sÇo÷ÝŸµûucÛˆóä=Ë·1öÚ»šó ÄUWÿ¨Y7>½êÚ×õÐ|÷»~uõ'×i¯Ûƒ®ï ÚU癞Ão s¡ž¿ú=×6âܸ4ö¹nìXýÕÜ'Ú­Yý[ÂãÎõ‹Î§~/^>†ø;‡çíš<q.­Z½~om€:¾ñ³cÄù£ë»¦”ñá¼;ÂK ø@ÎÏîõw€®¯ŽGÖÓð÷Ùÿ]ÿ°íï—§ÿaó=êþ§Í£çZðDù1Ýå +M+cJSJ`EïÑ>¢Ïëòã¾e¥.Ÿ–ºqés䯛Ë/þW»½Oz»Ã2š?¬­žÎlŒnôvÓ\½Œè<¢àÊúŠdšéîC$búí¢;º5(£‡/ æöá.‹/rß^<ø‡ðׯ§×7 þãw¤ÿÏæ²þà]óÞ•q€÷¬ß½šþJßò/ÇPÇûû÷Óëò;kêxÿN,ϤfþêÓñþe^P×_©~>[=uÃ=Xøþh†^%¬¼1Pîy–îàfð¿Íâ¿©ÄpË~è–îððW€r±ðý2ÿáª`Ç&fïΧXüŽ,ü+À3Ò5e˜Ó„¥?¸‘¥ÿ®Ø?‰Ù±çK·ð@_~ø ý½ ýqæ?ð8Øw”…ïš ùÐ~›!¼:´×u–/g´G³gWo(çØ™öL|/@¾Th·F,|wÃ=­ _*Ã=ï +ÿ_–o7´ï—g þ]¨G´SK¿ ®Ó¾5®ÃoÊ€&ÿ¹…áÁ,ß¶cÐn« Ü,†_÷rúÏcPßò~Ã/ÒYºç±P¯âÀÛƒá¾ÀžÖ¾<ã?Çüû_cùöB=öχòVA»ù1Tžcែ=Ÿ¿ |}î¨v·`¼»öA¹œ¬tc«ÔÀ‘ŽÄDõÄ—š´æHŠSñ6l³`¤çBçŸáÏ0e'Ãé›:!3€á´\@Ù¢L‰„ò§†/AþÉÌde䟔ÍpbÃq¾ “ >üñ?&Ê ùGÏpƒt±PN,øûCºv`¯•¥'ˆ½X<éËâIw°üdøG²z“QàO€ø$VO2Á“!|"¤Ÿ˜­ÇÉNœ^C)‘ S¡ÜT(7u «'I‹„rÒ!:»nd&»®dÖ÷ gC¹³¡¼Ù>—µ7™|óýgÔ»I‘…рɀ³fX¡ž¬?‘Ì. gœÿK€K¡^‹*ö„òe®„rA½—Ôa¸ì^J¡}–]ËæBþåong˜å§¡2‡µlÅÿ2y_Ó´f)³ ‘¯T ]â§:݇Oƒþý‚‡¿ã ÆŸ-ùãõþ$–^I‚q:6–a<øãX{)#õåòü¢/¯ŸÄßÚ¹% '&}<ü$Ö éàþ±úúÊ(Æ\ÏI¾zžÉ0Þ¦* §äêãSb$?ô÷4}ø è×é0NÒœní"3ÁîY0¾fèË™ëáÐßçÂ8›ãrŽ3O`ÏÂúr2ôד,†ñ¹ú}f¬>]&”»¨±>|{)!‹¡~|ü@»-•øq\,ñ½´9``6Ãe3féû;É qÛ~ˆÞŽ/%žOékY¹JÊE@Âp:´‹ÃÄü0Žøsç%xžÇxÀd¸îó`Üâó ŸÉž é'ÀxLŒeùÒÀ¾Q¹Ì?xˆÚ¥5C¼¿ÁóGqøë¤³ò”!>RÊhgñd(à`°#üCrÝçÅêC`MÿD?Ký.ñ/ÝvBNA»ñ9ýkFS}¾Y`ׂX}øLèWiÐ_g…H|o.Ø30xç^Ô§ŸwÍ}½3 óçUŒï§÷/ªã¾äÅçÏ2ÿ™Š>Ý_©ؾN/é¾üåð\ú’ûxžÎç®ñÊf§çñeÒÆ—5£é?åã~å/,¯\NQåóT^~ùäôò‘ß|Få¶¼ü^Oü^Äkãçî©‹þðdßÊQÐþìm9Ee§ô÷ê(l?ýwóy‘^?¤í÷¡ýšÌ'Î^bIë¤s“R—E¼5k݈Öĺëñéç~ »„|¹zÕeÒiLX³Úô$×mÿÃ6¶‰Xòż†?EÂI·igE“ˆ[óLšA¢²¾n_©òG¤s§•²Â6ÎÅÿßÐjeHÔé˜gNm™Dn°z|Í¥ÄÚsßæ7×.#ßì¦Iº.üiþçáA¤ó›‡ÛþúÖqþî/ïDÿq†„Ræx†éÚ⓵»;H§SÃzb7!+V×p°;i;»tƒÛöѤÁ¢gæ?òÊ[üF¦j*énï ™øöÒœ Ù$¬ãnëÙ=Ïp­ø$r|bçcO’nJù µ‰Hا}gLø‘tj?tgRï$²Í{o·®yt\½?çù3Hû~nÖ‚„hI½Y’„GÕ›RûŸíHûê“mz.…DvkbÎIÝN"æmô¿T»éRrAÄw½Hûg?ý7Òt¼”tléø$òÅs=~~¾, [¹w¶ùVõ8ùâ– ÎxÕªùôw×$ú?ÖcæÊöƯ>Û^•D¯9gÇF¶áúŒ¨Û;H¤zÑ“?&Q‘•.ý™@:¼×9-¦QyBzŽ«ÿ].‰:ú³åÁóH'Zý•ûHÔ«öv}ÞzŠXÔÖûðÔ%ÒñueÐo¯”"·_n|-ç‰rvèT9æ[±±ëž—*n"­ÔÖ¼þAÒå‰!Ž—»"mhï ÞL¢ª7¨{e¶Ê[îHƒ ïu!aZ5Hä ¾Éó— 'Q¿4Úr ØL"BòRJüø#‰Jhûñ®éSHˆ=kÎÖ$QKmó£ó5ÞçÖÉŒ åý)rÜÖ£}&ô$þÃé•&[Ž©Þõz/Ò)½|Æ‚?ÃITIkfµùCIÄž†½®” §ìï´}Ç $"·NÙ: דðs벪šKÂטyíÖfÒŠö¾ºHðk4Aiý†uP«æêuR;¥×£¤ÝL­ ¹í±iM-H˜¹î£äpÒaÍÀô-·«òñƒ¿Øk¬ø=Þ0Iƒ÷Óï9ïþà¯ØpéÝà%‰ð{¾¯‘poø4`¼Owz.ÃÌ"²¿ß@x|,Ãî ”ùædéë÷,`¼‡gÁ÷&ù£Ç£°>7.ï@P/¼oñöé~ø}Îíò‡ðÌC €}c!¾7ð‰Õ•OºÎg?àÈàéæA9ÑŠÞžîØðû=~/= <ø»ù!?ÿîÖâSá÷O ð<ù¤úOÄï¥Nè7½ Ü©àŸ þIðûåYðã÷Ql÷dÇïFø]g„ñ‡vÐc¼“á @¼þüw/”×G²û[ÀÆ€0NL0~\¿K@:è×.ß »„c¿ ÕÇc»â¸àáOK~‹äÇq‹ãËGûzB{„ØÓ]o?¶›‹]8.ƒ¤øŽP¾âá:òþØÅ};ÉíÉóY%¼`}ìR|˜äoåDKõÃëŽãÁdÐ.­ $Àvåñ]Lºüœû1ö'ÀVúx¹ý¸¸¿Înì'Øy:ÇrÁ~ÐÕ©·¿³ÌÕ§Odß©ø}ìåס) ŸQôùÛëíws]Ýþþñöw˜·ù ú»²¨¾ä·ü¢â/,·v{[Naí)ì÷†{õ=¤¨ëad¿›Ãíó‡@$ŽW—x_ƒp‰—‡û„›Ü§“ý²=.(•cX¾|Äû©A9F¼.ù øùýÎt÷r<µ¯‹]r> 7à1º.õ6h_Ãö7º¾Þ^O£ë"‡{È/óylwOv´Ÿüü1zßÀÃ(ÞÛ|žÂüùýžâÉNOvÅñåóÔ.žì’û¡·ílÔ^FãÙ“=}ÕÃ=«ë¸3²ËÛçiA¯¯QþüŽû¥m û6½ü…ý…m ]â/×ÛÔý\Z}—ðiZ„ý»rmâßü®¿­yÛØÏ±ÁsŽf8/ƒûKûMú9â}?{àÀß)K½Möë,¿ýâejð{޲ôÜ]õLú9’ûOAy¿±úؿ׊í!ì°ÛÅþÓÐî´fêëñ÷ϰ÷Šd×å—éñŠh½.õº¦Ñ5“yí· ;Y=ì' ®ÓíŸ5û¨Ÿtì!ÿ H÷‡E;¸ÿ»¾ö«ŒŸcöâõçåüÉòq?Öêg¿Âú—ý/}!>¬ØOHù¯²t¤8´çIè§—¡¼³¬ŸñôÛµfžl? íò‡TÞ-ùXûMHwšzãßâñ'Y|Q?ŒîžîCr>Oã=¿ï«žž?FüFõõÄc”ÏpüxyŸóôüñÔÎ2Ÿ§ëkd·¿<µ§r=¥Ë¯]ù­§';Œ¯«^Ì:Ú1*Ŭt`Ñ@ŸÃ’†¶”þ²Žãâ“D@™¤ä–ƒ˜ Aów§·ôàq‰IŽ$`(“8.aH\ÂDð–p ¹#]Ò°Qè+;Á1,©% ÐR–KJpŒNŒé´>£†%Žr$ Š éhé0£oTâÀ1 ¼ð„¸áq“¤ê–M3¡åUÉ‹â WÌ‚'V<±áIžãIž„ÂIñÀ~ÈÏÌüÌÂϬüÌÆÏ‚øY0? ágœÃÌ9ÌœÃÌ9ÌœÃÌ9ÌœÃÌ9ÌœÃÌ9ÌœÃÂ9,œÃÂ9,œÃÂ9,œÃÂ9,œÃÂ9,œÃÊ9¬œÃÊ9¬œÃÊ9¬œÃÊ9¬œÃÊ9¬œÃÆ9lœÃÆ9lœÃÆ9lœÃÆ9lœÃÆ9lœ#ˆsqŽ ÎÄ9‚8GçâAœ#ˆsqŽ`ÎÌ9‚9G0çæÁœ#˜ssŽ`ÎÌ9B8Gçá!œ#„s„pŽÎÂ9B8Gç塜#”s„rŽPÎÊ9B9G(çå¡ÈQ"0 @œŠS³8µ¸ ¶ØÝG½¡9ZNpŒÙ–þ¶I××Âú‹™Þ:ß1úí-ÿ¶çs†5ëHèˆßÀKØDBü¦ü„„Í%Ä9œ qÎÃ,!ÌðoÕˆ6 ƒ$ –0DÂP ñ[>~3o-!Ρµ•¿ÙzĹLŽŽs¦zĹ,ŽN=â\ G§qn6ü8§‰s<& zI<ŸkFĹ#ŽN=â\k”S!¼³S8÷»š$têçT9Bø3àÆ©Çh“„DBL§èç`9 *zìa’Hè”òá\7GHÿ¬SBE8wÏ‘Hùz*zDMGHÿœSBEÏ›$$:%„|½L *zŒ1IH$tJùz›$„ô½!]o–Î÷U“‡dza|'ù›HþPƒt’¿½Aº§ Â-á’¿ŽûäüП¸?Ä=ïîtÞÒ ½\o©=ñ¾ï’¯£d—YН ÷Ë¿‰ð¾Êý]ôåy:\ʳºO'÷—ö´ä 3o(ù£îÓIöàý…ûÝ/â[I~)ƒÜoxxÉ/ÇËöûJþ¶’ßǃ]R~.õsöÜ/µŸ<>ïЖèÃ¥ëf8^åñÖUâ“ë sÿÜ—rA ÀýÒ}An7—þÚTòËýîÃ.¼ðþr—C»¯jØðýÈHˆïFš6|/Å÷|Ï“5nøþ‰ÏI|Í›=üø^(kßðýTÖÀJù5qhÖë…9|>`1ü½ÕP3éeí¶§¬¡Ãzã{Úš:|–µuøÞˆÏQ|OÇ÷'¸Ÿºh†ð}ßSQƒ‡×ÇK-GMþµyx}P£‡í‰Z=Y3‡Ï´OÖîáï |ÏÄ÷DYˇö ¦ß×PÛ‡åiüðwjý ×üáuµøÞŒõ‘x¸ý¨ ÄßòïÔâïFÔ ¢]ø>‡ù¡ñz`ÿ5„¨B-!¶¾ÇáøDm!Þd!Ößsq¼ÉšCo¨=Äß ¨AÄú¢µT¨I„û©‹6µUèÇñZEl·AÌÏÛÇþNF #¶O,ðc{ÈšFù÷àݵïÈßÌ‹JÃ$Ç¥/hºÂ–[P¿|µ}ùå)¨ßOQ]W£ò=åËo{ôºÝ+»ók‡·G~Û½¨ù jGAíñ¶\oËÉïQ€ñéö;@Q÷‹¢… ¿ß÷Ù¢¶¯ ùîÕóÃSþ¢Êw¿x ›ÿ^_—{Ý~÷z¼Þ/þûÝ®wÚ}U Ô–z&’Šë­í°?›Ô xjí£ÿ"J½¼Ÿ>^±ÚŸãÓtE¤ß㤖cG [õ5¤ô¿Ú/}ó0ATJikùü·žÝ_çL\Qªikøœˆ[¯¼1hq® ‘øí ‹³H­Î1/m{ÌɱF…×>"kB8Vo¿mZPãö«Íé¢ÑlDåéw[ïíC|8¿ÕGU÷plb\Êï{R·ÞÚ«m÷ETªdÿø^rì&Ž~ÿüJ97MA$ÿï÷¿5›YƒcÅÁï?¹¯u6¢ÒbþI~Ê1püõ£Å”EJ‡WGÖùtNUR£ú¡¹“Dr,‘üñ˜ÝŸ8Ÿ‡ú½®Ä"’ÚsþXØÔŸX‹ h?¯¶Zó¦&D¥î®÷cÖ¾6‹c•†¬»´/ËþÛŠo®~±™ ’'MÍŠ=ù ¢RK[‚€HJ6™73<ƒ£ýè·ß¼^áªQ)µ˜.š%¯M&÷é¢T=0wÜÖ\¥Úú ÕÔ½¦<¾Êï½zk¯ÙO̦‹YMˆ[ÏS3l "ñ¯ZÿÅAŠÒD[“º“£éô‡¥×dÄ"ÚeKd•ÚÚZaŸ­'KñE)…ãîUífMË0‘Ê=ÒÝ'¢Ò0uøßNù…c­‹Ÿ'T>6ÔžñMØó¶l¢R+ÙrdMÙDûÅ#“[Ž©NIµø…?e( ¢œ]™xHi©]n'¢ý¬¶dØ„HÊõ~±|Ù›¹ý~¸QîHƒd¥uB^Ó®5+q,MWÊö߀¸5}Ô†„¼"ñÓÖîn@Üzüü¼Ü›=LˆŠ-üÌùÒÝ~D$þk_K‹ˆFTÌ´yvF ßÍ>6vÍ!¾ÏŸ)ÿÙðXD¥6]Y=Õi?5£^£¸Z‰¶dwsâÖk?¯ß:ã-‘Tͼüm—ò§3]"[íWR­uÍß•<¨Tó÷U[:Ç·ý·1W—9ÌqëŸyï{Z!ˆ¤f" ›8úÄÒ DsôÕ֢ĭ?¨l¿ÞVÇS£ò?z%qëþWNÄw7!*¾«è¢æ5ë©l!UB8ÚzÏ>5üøgCÔ§99Zú}ýê—§"Úoié¥YîþñÙDûµ°õçcKe+õ´%ÉñJºÄ¼ÌF¼¯Ú·Î!Ôr{Ngmq¸=ç¯úÿ¨2Ç~L[ÞêËñDbýïÝþÃ~Šö‹¦]í§¯ÕU[ì(dz›Ö‡ä¥”´ŸÓn3íç²hÄrZ{†pü]×ÃíÔRó~úD [û.Z3ò–ý’ÿvûW¿}%08ðb«cì—è0ÛN8æi÷ŸeöËl»@­ûMåx•­ÙçxmìÜ®jæx½â\ºš[ ö¼é °û¡À/úO³_×úÅ'öÚ¸ÆñæÂ÷?ŒíßÃ~[íÅã7Ho[QÉöÛ‹µ")k÷9.ÓÇâã÷W ¤¤o=µå˜vôï-«9~vñ÷® _r,U¥B‡›Sÿ)°eµØy-r,ã«]xZs—">šÐ]HYVORŽ^ŽãgI¹àM)/f®È®»À]Ó§Dþc){*ârÏ/ìy“c…~lrª-©XLÉÜyãœ+²þ±Ò@ú 9%ðÕ*q7¶t&•6^©vÿ¢›Dœç蛾Hí!y?•º<³=ñÕnË+ˆï®Ÿ¬åRGp¬ómò¯› ÜÔtõþœ^ }Uíâ¤2Ýê Êo«°ë*§}ðÃ^ߟ?µ|K=ŽUCì~ÎNžõØ5· ÜíSiİgªO‰ý·ÌÕ‹;ºÜį}Ф „~â×OÛ4Aà;Ñœ>z\à'(3Çj Îê±q2GvŸîKª…¥úª·Ïi‡À¡—ŸÙ|&”T[ÿnIÇHêïі/ks|”n•vD`Å~áeßl(î \Š<Ú©­5¸ßŽÕ¡¿r¤Í~¾ÀÓO&©Cƒc ÚZò.~êpÞsMæ°.Ç tS‰J„nÈPÇvMà»ì©æ‰×š¿Ÿ–¶uíÓƒ§GêeòRðï> ú]äááþÈï÷ÓÂ~GËïwÌÂEÍ{¿ì.ª£°ß#”z>hõð¾\*n-X¹6安IŽô—LLqããF'¹ìß!ög¹&Æ9\–PÞ’”÷»¯Óý©Ðv0Ímbú|ºqc)Õ•Vð-ÃlÔ¼Ã(¯êè®Ôt—q:>Ê£;ìÒçE%Õ‚ºUVÝhÕUQÕîÒ]®éìZ!S5ÕUÝ}^5ÕT]utì¦Kjª®–êj«nœêè,g]ÕÑMë©®¾ê&¨Î_uɪk :º5ým¤ºIª£3†tGaº9Ýp’Î(ÓQéóSUGg(§©®™ê¨Š÷IÕ9UG¼tV·…êRTGg4ŸRÝ >@ut3Xªz ;¼SUUFÐÝþé &UéÒÝs©ª€nþIU¹tB{¶êèÌïÕÑfº«=Q¥ª[:3MÅt¦“îªNg~˜ØÔ6U7,TýÍPÍT½_¤:ªÆX¬:ª ›MSu+ÝU”Nª/SU;PEÂrÕQE*=§ÝTeú²êè,ÿ+ª£*RºS{´êè =݈š*8é®ïT9AU#¯«ŽªNVªŽª³UGßÞPL§ëa¨ÊðÕ½©::³KÅt³ë·UGw}]­:ºËòßTGw#§â“XÕ­UÝÕv€‰õMª^¡›µÆL2>4NC:Éø:“Á!Y#éé¯a¼,%jeN–"H‰Pšín$±“¥þ’_¶O–”D–û $š<¾»â><À ½\o©=QÊî’/BòKOREòËùÃ$ž®îyäÃ¥ÈåÚÜç“û‘K»Êax˜Ó}xcÉ­¸O'ÙƒÒiî—®·K~Y:G¤üRÿáá’dÔ¥_ÉöKRB ¡¯»Œ¤ˆáÉÞ/µŸ‹”²†A9Dò[ÉßMòËõuÄêýƒ ø¥tÒýÁ¥ÝäüOH~¹¿Äðvp~Ç¡Ýg¹ÄÚǼM—¼¡´%E¸´" ¤o(UC‰>J‹p©J¶ú3ô?Jq)J‹–®ÓÛ…K_1øQ’„KJ0¿Yʇ|þzû°X¯t¤á„± Q¶ò/ý…!ü—¥(ùÃöœ“­÷c½Q:‡vdíaˆK…²2ôöâÒ\*€’O”°Áý•KÖL€¸ä% p„ãõYÛÖËÛ«Àuå×%u¿ÐŸ!.}Âú£Ä¯Sæ÷ ±]d3„‚ÈØ¨—dç§Þ.”˜áÒ”·ë0 [oÏ8Ç¥)ó²õå¡ÔëƒKhPJ7'ì„ü)×ËÅz „ë#ó ý3Xƒòë$/áJe÷ ’’­· —®`~Y’Œýd:lÓŽí÷Wâ„q‹í‡KVpœNõgˆKòðß[LRôõåRe°þ·Ëÿ JVû8õõ|p%฿’!`÷8@¸¿r?Žsø·=¼Ýe34Ç.  v`ûÀ¿íáíMŒ——¾¡ý(…Ï#í>›_ —·Ÿlò[^AÓ¶Ü‚ú壨íË/OAýžx Ûžÿ®v+èõ»Wvç×oü¶{QóÔŽ‚ÚSØOvÞòz[Žé´û,Hv’@²³%;’t§>G÷Òvˆ÷Iºó"HwzHw¾- t§&G&ÙY‰’92éN&HwªHw¦yîÌE,¤t碗ÒiˆÒ(#éN%<_qdÒþˆ÷V³ò/ÄJx¦!æSÂãËÑ­„ç©ÛˆE$áéˆ(Ixˆ á™æ¥„çsD Ïi O(¢ O[D OwŽù“ðCôR³…#“îŒáȤ;ÓA³Í@³Ò@³$<ë]$ ÏŽz)O3:)O"HxJ ,R)ÏOÝJyòjÍ]ÚXÊsWIO)E"é9ïŠZ¿Áѽ¤'ªˆ$=#9­¤gŸ@MÊSŸcÁ$=:IOŠÀ|Iz¦s|°%=_s|0$=3@Òcò0ö ŠÔà¿å(¬´ÇÛøüòõñPâ£Ç{ï^Z=¼/·•ÄÐûl±;\q å𻥻[˜§´ž° å†« åæ§]<Õ³(ê“_ûò{m ’¿ ¶“D4¥GƉ›·ÒmÓÕ< 4æ”ì˜4†Šºððò™ b”ÿÁƒ¯„7mÔãÿꦈ7o”Ó)z|¸y£„Šÿk7ot+*öNJìQ8l 8Ö$Ë\TLyé>nðýpƒï‡|Ó³ÿ­ ¾ñÔ*Nmâ4Hœ‹Óq*Ø[ ` l‚-P° ¶@Á(Ø[ `3 6³`3 6³`3 6³`3 6³`3 6³`³6‹`³6‹`³6‹`³6‹`³6‹`³ 6«`³ 6«`³ 6«`³ 6«`³ 6«`³ 6›`³ 6›`³ 6›`³ 6›`³ 6›` lA‚-H° ¶ Á$Ø‚[` lA‚-X° 6zÃðn‘‚éöÿÙ¸O’survival/data/colon.rda0000644000175100001440000004366011732700061014664 0ustar hornikusers‹íˆÇ•ïÛ‰“(a_ö…E,&\B"/„éßÝÙ¼ìMâdµYg3±¬òs¯Ë–Ç˲-ÿnËúùGF²c+¶dÝ™‘eY–aY–åò‹XÂ"ùÄå1„ð†GXÄòží{êžs¾IÝžî¾·Gêç3=·»NUuÕ§ª«ï(7^¿Ã{׎w9Žóç-o½æõÿ^ÿñÚ·¼þ?×8ÎÌ;_çÛ¾qÏ]÷Üí8oý¯¯̼þß»çŸøÓ5çÍdØu€[€3À÷·¯v€ï¾øà6à~8 tÐÀc`Lþ ð£ÀÿüðO]d¦ùq˜i~ÂfšŸt€™æõ0Óü”Ì4?í3Í?s€™æv˜iþ¹Ì4?ã3Í¿p€™æ 0Óü¬Ì4ÿÒfšŸs€™æœì3à@óó° Ì€Í`˜š79À.04ov€]`h~Ávp ùEØfÀæ_9À.04w8À.04¿ä»À 8Ðü²ì3à@ó+° Ì€ͯ:À.04¿æ»À 8Ðüºì3à@ó¯`˜š=Øv=`ìÀUÍ[`Øö€°W5ou€`ØfÀ>p\Õü†ì»À0öàªæm°ì{À Ø€«š;`Øö€°W5ow€`ØfÀ>p\Õ¼Ãv€]`˜ûÀpUs—ì»À0öàªæ°ì{À Ø€«šßt€`ØfÀ>p\Õü–ì»À0öàªæ]°ì{À Ø€«šßv€`ØfÀ>p\Õ¼Ûv€]`˜ûÀpUóØv=`ìÀUÍy8ìg]à°œfÀ`¸/Wkš÷:À`8 ìç€=à<0.ûÀàx¸ \ÓÜíg€à,° œö€óÀ ¸ìW€àEà*pMó>8ìg]à°œfÀ`¸/Wkš{à °œvsÀp˜€}à p¼\®iÞïg€à,° œö€óÀ ¸ìW€àEà*pMó8ìg]à°œfÀ`¸/Wkš:À`8 ìç€=à<0.ûÀàx¸ \ÓÜëg€à,° œö€óÀ ¸ìW€àEà*pMó!8ìg]à°œfÀ`¸/Wkš;À`8 ìç€=à<0.ûÀàx¸ \Ó|ÄÎ;ÀY`8ìçpØ®À‹ÀUàšæ£pØÎ»À9`8Ì€ À>p8^®×4s€3ÀpØÎ{Ày`\ö+Àð"p¸¦ù¸œv€³À.pØÎ3à°\€«À5Í'à °œvsÀp˜€}à p¼\®ifp p¸ØnÎ`¸8Üìwç{ð0pxØž®ÏÀ À‹À×€«ÀKÀ5àeÍ'àà p+°Üœ&À.p;p¸ØîÎ÷3àaàð8°<\ž€€¯W—€kÀËšûàà p+°Üœ&À.p;p¸ØîÎ÷3àaàð8°<\ž€€¯W—€kÀËšO9À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—5÷;À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—58À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—5:À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—59À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—5‡ÕÜœnv€Û€³ÀØnÎw{À]Àyà^`< \öç€+ÀóÀðð"ð5à*ðp xYó;p p¸ØnÎ`¸8Üìwç{ð0pxØž®ÏÀ À‹À×€«ÀKÀ5àeÍ#p p¸ØnÎ`¸8Üìwç{ð0pxØž®ÏÀ À‹À×€«ÀKÀ5àeͧàà p+°Üœ&À.p;p¸ØîÎ÷3àaàð8°<\ž€€¯W—€kÀËšÏ8À-ÀàV`¸ 8 L€]àvàp°ÜœîfÀÃÀàq`x¸<//_®/×€—5Ÿu€[uþ.´eË–-[¶lÙ²eË–-[¶lÙ²eË–-[¶lÙ²eË–-[¶lÙ²eË–-[¶lÙ²eË–-[¶lÙ²eË–-[¶lÙ²e³è8oý/oü]è?À×Ðë¿…þ3¿7Çø9^×áï×ËÇvÆY/޸כïzõ\/Þzí¸^}×k¿q¿^y×k[½×;ÏvŸmýcÜû¼^=6Ú_ÆÍ§hÿ±ÕwÜ8¶þRôóõî§íþn´½Ç-ßzÿÙúízõ,:‹þÞvÞ¸\oõšm\Ù|5Õg£ýs£ýg½ÿÆ­÷zíi+OQ/Ø~?®ÆígãÎ#ëÕÓ6OØüÒã¹lyª.ï¤}дëënÏ¢å©:^Ñë›æï¢©êñt¥µoÓý:éùyÚóÙ•~¿‹¦ª}`Ë¿îù¹hyЦº}Wwý&=_V]¾²çÛ®ŸözbÚóeÑÏËÆ«Úe¯Ÿôü9íùzÚý·h~¶ó‹¦¦ùzÒý£èçuçg»~Òí_¶«¾¾iõ+›_Ñx“ïeËWw{Mu÷Ÿº}X6~ÝíYöúI÷¿¢å+ûù´çƒ²å/š¦=Ô=›>¾'íëºË_´©;UÝßêŽW4ÿI§iÏ¿eËS´¼ESÕã»iëIׯlùŠÆ/Z¾¦Íÿ¶Ïëý]h÷SÙ›#~øùæ\¦ùù®æ ™æM]Í¿p4ÿ øyàŸešŸq4瀹ë-ùxÓ@óãŽæŸ1þ À¿Ì4ñúÏgš7v5ñs,?Æ»ˆåëfšÛØÞ˜ÿ@l/Œ‡õÿ¬£yã@ó“™&ö?¼þz ößÏä_:šØÞx?nr4?ibýñüÏdšXþ¹®&Ž/¼ßØ>˜?Ò6¾±äÚ'ÓÄûñ™&ö'ôÆË•˜Ï™&Ž7¼?Xßœ/ ~˜?æ‡åG¿`Ãû‡ã˃ã ãÄÑÄúaûaÿº1Ó´¿Üøb{àýÀñ÷û›­¼¶ùýf¯8^°<È?Ï4såæÊ õÃþ‡ã ïÏÇM ÄúaÿÀû¾ÇòàøÀÏ?åh¢°}ðúÔÑÄþ÷QGûŸ­àüˆ¾þ$ÛçÇœOºš¸þÂësëGëƒçãøÇñ†ñq~Áñ•ó䟛Ÿºš8þÑÛ8>±=±½°¿`y¯Ï4±ÿåÖw]Ml$Ž—Üúi io¹õ”ëíg[ŸcûbÇñ•›Oºš¯ëhb{ ¿°}lë=\/¢_p|bG_ãóŠ­ÿ ïlÄú`þ6Ÿbù°>7±~¶ñ‹>Äûí‰ýͶžÆûƒŸÛÖ¹ç%(úûsn}‘iâù¹õj¦‰åÏù;ÓÌ=oñz¯Øß±ü˜îùˆùãxÎïòû„£‰ýû®ÿmã#ç#øÜö|‚í[ŸA~XÞÜýÊ4së™®&Žçœ¡¼¶ù0·ib~Øp>E¿äž“Í\ù-åù´£™k¯f®?±ÿcþ¸>BÚž_lã#·>âýÎÍÿŽf®=!¿œŸ3 £™Ûoó±¾¹õijíXÇ‹¥þ¶ýü<·þ.ÈÜü´=£¿ðyÞæß!ìŸøüšÛ¯íjæžš8ŸåÖƒ@¬_n½íhææÇLû^÷û×ï8žpýãý‚Ϲñiæö7 ¾ŸâxÈ=/Àõ¶ùë—[‡:šø¼›+WïWn~q4±½róï'ŽwÛó2Ö'÷ü Äñ‹ó‡íyûg®ÿfš¶õ#ŽOÜoAFަíyͶÞÂõΟ¶õrnÿXt}œÛÿ„ú`þØÿ°¼¶ý}œOrëQ í}’u½ Äû‡ý×g¹ýGÓ¶¿e+/ö×®£ió5¶Oîy>Gÿ\Ÿiâü†ýý€ùÛöÇló•mçïÜód¦‰÷ǧíý–ß¶ŸõAÿæÖ?]MÛz;·ž‡òáþ4úζ^˽_bûáõØ>¸?™[Ï€™fî}8œãÇ“íy?÷üïh¢ÿlý˶`{ßcóŽ¿ÜûDG37>è;<û_îý)0·¾âþ"Îß¶÷û¶ýÛzÓö~:׿ |6Ûž÷sûP>œ/°ýmï÷±|7sïßMôžŸûþ@WÓö>Ÿo°?ß<ÐüBWËc[ßZß§:š] íýBn~ïjæÞŸñó?bþ9¿@ùmÏ‹Øßrã(ÓÄxX~üþÎèô+Þ¬oî~;š¹þÄxxpýŸ{_õÁùÕælOôgÑï[ayºŽ&ö¼8>°½¯Ï4?Äûe{¿ž{>õ1Öãå¾O–iæÆ\Ÿ[8š¶ç¿\{ñþâç¶õm?}‡õÍ}OÉÑ´?Œw}¦‰õ˵Wõ˽ÏÊ4mïËÑ8?`<|·ùÇ^÷ý`û¾^np?8· ̽bÿÄÏmï›së[`îûa@›°=q½…ó-~Ž÷Ëöý1l?Üϵ}¿37¾¶ïg[÷?€¹ï“:šØþ»£‰åÉùò³=ï`yrõu4±|xÿlû‰¹òt5mßËíßu5ñ~cþ¹ý°L3÷}ð&®—mß?FæÞïsß×Í4Ñ]G˃ëuÛül;}fó«Í·Ø_sëûLÓ¶þ¶ÍÏX^ôiîyo ‰¾GàýÁüq¼Øæ“Üû•®¦í}Lnÿ°«iûû\?åê“iâ÷eq}oû~°m½‚í{ß ŸãýÄòÛÞOÚÞ'âøÅû¼>ÓÄúáóNη@lߢë±ÜüÕÕ̽7hæö— ?ì?Ðö>7·?çhÚÖgxlßGÀýÛ|Š÷/÷}K ÖÏöýp¼ÿ¹ç¨/–ßæËœÏ2MÛ÷±}pÁöý¬on?` ™{ ´} ûÎ÷èoÛz.·^t4m†ñlß÷ÍÍï@Ìßö÷¶÷s¹ý°®fî}"¿¿ñm߇̭ïàsë~)ÄϽ?澯):ôw¡NÁ4í¿w·¥ºÿ=‹²ñªN›íß+°¥ºÿ¾»éõ·¥+½þÓþûýIç7éüÛôûS ÿþB£Ò¤ý1é¤ljZý®´¿¥ìùEÓf¯U§i¯W7[jÚ¿oTwü¢ç_mýgÿ^ÓDÓf{ž¨{<4½=&Óû/¦IÇi·OÓüS÷þÍfO›­þM[Ï·©YéJ÷ífëÏUï´é÷§¦·WÕëÙ¦ïÇM:5­|Ó~þ™vª»¿7-5­|W»êN›½þM+ÓÊs¥§Iï¯Õý|Ò´ù¢ê÷ U§i·Ï´Ó•¾ßÜ´û»ÙÊSõóÖf߱şöøiýY.möö«z}q¥ßߺÛcÚë›I§iÏÿM÷ç•Ö꾓NM›ÏËæW´>›­N»~ußßIÏOEÓf듾~³å?íû¹ÙÊ[uªÛ¿eçëºýù;òÛÐß…¶©MMNWš·léj«o›ÚÔ¦65%MÛ¿ÓŽ_6möò·éêNÓî¿W{üºÓ•^¿6µ©ÎÔŽŸ6µéêMíø¿²S{ÛÔ¦6µis¦ÖßmjÓÕ“ÚñÞ¦6µ©M›#µ¾nS›&—šþ½ö6•KWû÷àÛtu¥iÿ\Ýå™tþmšnjïo›Útõ¤v¼×›®öö½Úëÿ;ÒX:íuÜ´ãOº<ÓŽgKM+O›ÚôûÒÕÖ_ëþ÷ ¦í¿²¾žô|Óö¿Í•¦]þiÇ/š¦íƒI§iÇ·¥ªÿý™¦×wÚiÒócÝ÷¯êùv³§v|4;]ií¥Í§íøiVšöóÍ´ß+_iãëJO›í~M{¿d³§¶½êMÓžï§_Õù_mÏKU×oÚûUûºé÷³ióSÝûmº²Óf¿ß›­üíóü••®´ûµÙç‹Í6?O:^ÓÚçjOÓ^?OûþO;~Ñ´ÙÊ»ÙÒ´Û·]ŸéÔôý£ªÓ´Ë3íøESÓÊ;íõMÝû1e÷ûÊ–oÚí[6m¶ò¶é÷§Íö¾¤M:M»½ë~Swjº¯§¿ît¥ù§iûuïÔ0žßü»Ðî{††˜O÷:àV`|Ж¿£i«·íüîàŒfî|¨OÑul.¿™ßOÛù¹Ï?¨Y¶ü¹öq€ð¹5?_Óv¿rý >/ZÞÂý˃ýúkîü‚íWøsŒå+Ú¿>¬Yt|åâ¿Xôþ@yl¾Èõ¬·å~Ù|Vz¼t4­íi¹9~ˆŸCû=qùÿ}Ç’¡µ<&žùa} ‹¯ríý óÃxx}Q?ÚÊo›Ï,Ÿ—žßŠ–ÛúSíå+ZÞ¯·ŒßÂã¡àýÂòþæsLx}Ž–þ]´|ÖñhûÜV>Ëz,w½ísmóaÉëËúݺ^´ÜÏn7SÌùÎ/½~-™ŸmüYÇ¿mþ‡òYçSÛùð¹mýËÏ6>;@ŒùÛ|i«_Éù¨¬«>?G‹oJ?OØÖØþ¸žÅõ›­|˜¿¥|e×/¹þh+®ß ÖÇÚ¾6b}ŠúÚVÞmÀ¢÷«è|i©O®?áùÍÒã¹h{B|,_áûm›mþµ)ÛŸ ®7Ë®ÿ¬ý«´ío mûíE}‹ñló³ ´µŸíý±e¿ôú¨àþ@QßâçÖýf[ù-ûõx>¦¢Ï…×£E燂ím[?½_…}„ãÉöþ Yp?ÞÚÞHÛ~"ö¼_¶çaËý´ïÂë1¬m¾²Ô'÷¼kiÂ÷Êc½Þ2¿žÏ ~ž«?îo¼ß…}Vt~ÁòÙêŸöoÁýb«/ <Öó±>ð}šÂï÷lϧÏöü\v=–£Í_Ø>ø³–Ïò<‹o+_Ñ÷ ¿/*Ýÿló½caÑö)º¿Zp~.ëßÂ>.º¿m›¿lϧx¾-^É÷Y¶ü­íWðûeË_xÚêo{ÿ´–ßöü]t½Ðíßû[ÑõyáïZêc}ŸŒù•í?–ça ûû‡íyжþ²<ï^oØî^,Ú_Jï'ھߌë=Û|Rt?Âæ{Ûú¸àúÈ–îó°è÷;‹úÆÖ_mÏëe}\tÿÛ·`|ë÷ËáüÜý·íÏCÿ-ºž±ŽoôOÑýXÌÛÓöýþ¢ãË–Ò¶TЗ…×GX~lÛz½¨Ÿ‹ŽW¨_ÑïsÞÏœ|²þý’ÍÇ%÷ƒJ?ÏdÑùÞ¶ÿPô}dîsKÿ(ºÞÍ=_½?¶þ‡ç[ò·Íg¹÷gEŸGŠîWÙæ³‚ã뛣åþí/ý]èïø½JE½jcÕi£ñª*?þ¾lý'ÝÞUÕ¿ªxeó/›êîÏë]_u~U}^6¿²÷»ªò—?n¾¶Twûßo£åÙhùëòñzŸO»ý›:ËÖ»jõå´Û¯h¼ºÚ»êò?nþESY”MUçWuüºÆuUñ«>Ü4)_”M“öÍzç›OÑÏ‹&[œ¢ùÔݪöSÙùoÚýi£ùMúº¶OÙù²êö«úó¢å÷¼ªÚ{ÜÏ«^ÔÝžãÆŸôú¦i뇪óŸT¬ª¼Eó+›ªö]Õ÷«ìõuû¸ìõuÏ×eÛkÒã·êTwù7Zž¢çOªÿØæÙõò+¯lùªöëFËc˯®ûSu»›6Z¾º|<íñ]·Wëî?EïWÝó¬íóI—³nÿŽûù´ÆsÙülçmôóõâ•?¶üÇý¼®ûU4þ¸×5e¾¨Û3EË[ôóÆ©zž¯ª_l4ÎFË[õùøûºSÕñ˶OÝí[Õù“ßo½ßWUÿ–k£íS¶½«îo“îOuù»)ãwÒã£ióí¸ç½~Rã¡êT·&íïªê3-ý¼ª4nþõޤ绪η]_UžÔ|U÷ø*{?Ê–·îùpÒ>/;ïMuùÌv\wüõRÕõ›ôý({~Ñëëêÿ“ö׸ŸOºÿU_Ýý§èüV6¿ö'[~륢õ7^]~²]·ÞyU—oÜüŠö×¢ã¾jÍÒþ¶¥ ä7üÿ ¥ƒMÊÿ¾{ÓÓö÷ÎuÇ«¸<ØKŸ_ôï±kn¿¢õ+Úž¹ü-ÿþÍ´Ûcêñ,ñ­ÿ %Ë‹í_øÿß¡i,{ÿšv}ÓûÝçÛ®Ÿöxµ”¯êù¡°¿ –·èõe瓉û¦ì¿ï3a–.OÕëK~Ó¾Ÿe×Eÿ}¾ªÇkéõ™íü¢,ºÞ®:þ„Ù´ñ,ûïg"+_7¼?Ô>¿ô…•—¯dù'¯h}ÿüRóþEÙù~Òý«l¼Âû –ë‹–¿hy7ŸŒWõüÓxV}¿*n¯Òó·Å7S_5m¹aœúzÁ¯îøM[OÙæ“i·OÝ×Wý<ÑôõHåþ-ÏÆ ¯/ _ß4¿7ìù·òõªm?vÒëi÷ºã•}Þô~}YNy<×þ¾¤j?4m~´°èzgÒå˱dûÛÞ/Ô^~[y&¿tnúóÖ„ýe{>(ü}²ówÅ×>ßrý¦ó²êõUÝû%%ýXv~³ÝϪÏ/û¾º,›Öþ•ï§4}>²°ô~WQT¼Ðø÷wE9aÿV}}ÙçÑÒë…ŠŸw­ë÷ŠŸïê~žŸ´¯šæÿ¦³è~ÚÄßÇ[Xx¼L;?<¯iãkÚëqø¼è|[x}Qq}kþœ4k~^˜x~UÇ«øüÂ~rÿtù¬ù×|ÿów¡eŸC®t:À¦ågaéuø´é7[ýK–¿éñ¦Þ¾eÛYöüißÿ íó¢çOº>mÿ-FXwÿ,{}{¿[þ>:ÀšûCé}û+l¼4Î7Çsíã½ìüh›'ݿʖ§èõ¶ò–­OÝ÷oÒ¬º¿Ly|6ÎO›½?a<[üºûSÝ÷ÏöyÓh)oåïM‹öOÛõeûOÑü§íæûæJ£œvÿ°•?Ÿt´•¯e9:ÀI·ÕùO{}r¥÷ç¢ã¿iÏeãm6ÿU]^ȯòý¤²×_ét,œvy¦}?«ŽWwù‹¾Ÿjøýú~ÈÕF8éøu?O»ÿÛæ»ißÿÍÞ¿jÎÿª»_uߟiÇI÷§¦?ON»}lùU¯hyl,zÿKÖÓù¨ì÷©¦]þ’lÜýr€e¯¯{¦ly›Ö¾e×»¶ü'íÃiûdÚã¡iý«îöjšïàf»¾lÿ/¿ðßÉ5müN»ÿ\iå­:ÿ¢ùáùU—§lùjÞ¯ªýû-ËÓ¸þT3 ·Ó|Öôû‰×ÛØ°öuèïB‚Éö=zÛùEó/›_Ýùm¦¥¢ÿ^FÕõ+¯êþUöüªã­ÿ´ûߤã7­þÓNuûsÒiÒãµiåizÿ.Ûߦ]Ÿ¦Ï/W{ûØ®ßló£-Mº¼“ÎÒëË¢å)›ß¤SÓÆgÑü¦Ý~¶4iŸOºý«NWÛ|9íñ5éxM›OëžOÊæ7íûÝ´ø›}½Võz¦iýi³Ÿªçßiÿ¦­š6>§]ž¦Õ¿é©éëßi÷§²i³Ï'Ó¾ÿu—§lüI×§éãaÒýwÒ××®ôúM:M{|5Íÿ¶x›­ÿm6Öý¼]õóRÙ4íõôŸW‹¦I×oÒíS6ÿºÓ´ãÛÒf¿“Þߘ´ïšVžIǯ;¿ª}|µû®êõÁ´ûwÕi³ÝÏ¢iÚ듺ÇÓ¤ó›ty§Ý¿¦½þÞì××=ÿ4m¾›öø˜ôú³îÔ´þP÷õ›-5­¿ØÒ•v7[ÿžöóiÑó›¶>kZÿ+šÆ(ÿ›Zµ÷¯6OÕ®öu{Ý©iýµîufÓÊSÖÓîÓžg'ý\Ö´ûµÙã•MÓîÿESÝíÛôöhššÖŸ¯öT·ï'ʎǦկîú4mýצbiÚý·îù¯iýi³Í§Uç7íû1éçŸI?¯”-oÕç—MMŸ¯´xM¿¿UÇÛìã­é¾µ¥²í=éûg‹_ôóªÏ¯º=6[ÿ){ý´ïWÑ´Ù®Ÿ¶ï›–_ÓSÝëÙªËWöóiÏ'“NMóQÕi³¯ÇëŽß´ö™¶/ª.OYŸ4m=;é4íòÔý|5m_”=ÚëçºÓ•¶>¨»?7-¿ºã7ý~O:m¶ùiÚ÷ïJ«OÕiÒ¾ª:5ýþ6mýR÷zâJ_O—}¾(úyÑòÔÿ¤ûcÕí]6Mz}Ô´ñT÷ú¼lûÖ}ý¤ûã¤};éñ^õùu§ºç˲©îööz¦iý¡hš´ßÊÆ/š_ÓúߤýØ´õŠ-ÿÍ¿ í¾ôã7ºÏu‡<÷^Íç¶™ 3êîžòîþûÓç;†|åÄÏP~ï¢óÏ­ÿUó‡¯iž=¯ùßiö?3äÑstÞ? ¹x÷gökþðº!_øÛ!ê 9ß¡zPýöŸ¦¸ûV‡ü›·hžùWÍC¦½öypÛ÷†ü§yö˜æ++ÄG©Î_ïiîÓn*ï·èøý!{ÄÏÑu?¼Aó/j¾rý^rùŸé÷7ù=j§}‡éº¾æ“tݧ2j‡ßjžýÍ'·ëöx†ŽŸX¥|ÿVóÀ®!_ |wiž=«yúͳ³šg~ü‘æ‘áÿ¡n÷j¿èþ=Eåý>õ‹ç©Ýμ ܼUó•ûG_wO—ÊyPóô¯4ÏþLó±lÈÛM~ç4빓xævÍèþÝHןþ‘æ™4S;>B<÷š}ºÏÏS;õ_ò•KC>Dqöt躟k>Oýnµßó”ïÓäƒÓü•ã ê?¸Hãì$yéÔ‡†|©KŸ_§yô†<@~:>«yî5Q}æ‰èƒþ§æƒÔÞ= ó¶k~—ê}Ø´åÚí "ö¿³ßþDs9Ð|”|så·t擆üVoÈþG†|‘¼ÿ<•ë~*çN:ïVâÁ.Å¡óž¢vœ§ßŸþ_VwÈÏ/Q½G¢rÝ™ÑçtßoÕŒͳÿ©yˆÊý$勾:tA·Ç šOž¢þ°ôõ!Ÿ¡òœYÐüÁ»5ÏnÑ|ò3>=ýKÍ»¨]Ìx>û?€_Õ|´§yænÍ—?:äwW©ÜD3¤rd—é˜úßü°CwŸ£rL†ÜÓÓ<ó^ÍÝ3C~ÚçìßkÞ9òfªßé=š÷RÜAן¡òŸ¢rž}Ï_¢v:{Aó ê_wSœü&ζèx?˜×\$ï¼D×?@ívÇê}ê÷{¨œ¾VóÌÍšß¡öy„êyŒÚõp‡ŽOù }þéLóôû4R=Ÿ¤þ¿¼ò1ýjEÇÿQêûi^ÚGåx|îwóô@Óø7£öZ~»æ‹fCå?Hqvö)Ÿàsš/mÑKÞ| ÒÌCŸw¨žwk.ÿF³ÿé!_ý5ÕãÍ£´.z–úãã÷öUÊçEMã—ºCž$¯Ýoú[Ws™îï]¦½nÖ\ú²æÙ ¹@ýýÌ1͇¨}î¥ö=ÓÕ\~\óQª‡f¼.ý¥æég4—?¢‰í·|µÕߌGSßct¾K÷qùGšgÍ穼æùë‘Sù—~¥¹ï¼æ™÷yÐxä?†|•ò=Cýaîãó4‡Íõúç_ò­?–þPsy—&>¯ùŒæ 4ÌSù—_Ð<ãß­ù<õŸ*ßé9Í¥žæ™¿r?­Ï–ߣù]÷ ÊïeZ?§ûøýë€TÎï/ŽÒ¸ÝoúßÍo †ÜEõ=Cqn¡ò-¯hî£õÍ·úCÞOùšñ÷4•÷ ÊçÛÄ›º”Œ÷3Öü*w•k?só<ºô3Íå—5—vi>Oqh¾|vNÇëͧ¨^мaÖˬ¹xQó>ºn7Í[ÏѺâáÞæùïIúý>³^¥Ï—¦ùíæâ’æòÍÔ^G)ÿ§é~}ÇÌß—€ÿ¬¹øwš©þf½z:Ñ\þ÷!ûÆ¿Ñ|êeúñÒ6MœNÖ<³Jå!.ÏŸÓ4Ï Ó8ÜOíýÕã3>i¼ ›‘ùl6äZïáóñ“´ïr/•çô¿ ¹òßOí2òýj.ÿTsñ‹šË4/î¦ržù æ‘.•Ï´Ë?j.oÕ\ú{ÍŇ5Ïþ‹æ2ùvݯ£T/³„ÏëgŽkâ|ŠëÓEZçFû–Cå ûiöÁÍ3@óÜüQ:ÇÏ)òïQZ'™çèo®ù4Í«;Cšõú~Óï¡yˆâÞB×/ÿ¦y.þr—ÊùiÍÃtý}ôùÒ‡4Ò<|ÄÌËæy£Oqþ·æ³½!¤z-gš8~—nÕ\œBÿ¿ƒêy;ÅyxNsi泃!o¥ú=C÷sžÎ?AÏ)Pùp=ò˜Cq©¾æ¹É”㎮æòÏ5ͺòkTŽÇéø>ÊçAÊÇÜïïRÿ9HÏÃËsš¸>X¾Ksåÿ Sè§¿©¹ü~ÍÃ3º|Ë{4Í>Tf>Ÿæâh>OûÈ7©ÿ<™Ñu‰¦y~~ê×ë iæýá2Ìén§ëûçqÊ×ìo>A~ßßßÕáâ4ÒsÁSt½y®»žò.£ö¡vÚOŸ?E÷ÍÌ3/Ñzά«ï¡ë̾‹¹úCš}Ôþo€à{4—iÞìÓx[ú©æiÚ×3ïמîkš÷ºwöuyöÑ|°ŸîË£Ä#TŸïPý#Íþ%ÍçÉGfýëû#Ô§q¶i.®hÞJõø\Fçý§æAòªÙ¿^¦ñ5zE¿’Ê·ˆ<®¹tQsù?4ÍZ|;ð5qýÒ\óÔ‚æ1šÇï[Ðå^ ~»xYÓܧƒ4O¢qlöño£zšý£¥ß§iö ¿©ù2­Û2Ê÷$í>köh?ãyòÙÞf~y„êõEºþZ˜õââaÍþ¯5O&šý÷kž:¯yOÈ»¨§~£ù }þ µÏ©_jöü²æ½w:FïQú”Ÿo'ÉOæ{'¦.½KÓÜ÷ÝÙOR»oã|uÒÑ|1Ð|t0¤yŸÐ'/=KýǼϻ•êÿqÊç}ÄÅ÷ߣy̼‡¦ýЧ(Ÿ'©¿/þ“æ­ Ì}1^7ûüý÷išqfž×–¨öQ;-Íhž<¯yb‡&Ž¿¥@shö³ŸO^ÖÄþ×_Ñ\ü¹æ>ªÏ=«ôûyÍš·Í~¡Ygì¤ë¦zšõâÉk>Mõ4} ã?Ô‡(ÎcÆÓÄ}4¯<ÚÒ|ÿż7\ü±æ“´NÙÝÒô“ïÒV͔ϮU:ŸžGÌsâÃ!Íó¡yŸox„î«ùž úyñCš}Êßì—à|Ýÿ'ÍÅŸhš÷ÊC=¼þûަù>нÙÇ(ÞŠwô'š¦^_¡zšõÒˆ°?ÑO€ç€Û5¤v2ãÖì¿çæÏš‹ïÖìß®iÞš÷~;¨>ý4Í~¬Y_?G÷ùKÙï®_âhšï)™÷çÏPÿzˆÖÓ‹Ô¯¾O÷ï1çfÿÐÌw÷¨ž@óþñ;Ôñy€îû«ä™Çé¼Ó¯i.¾„ùséa <_~­æÒYÍÅ4_]Ó4Ïñ÷R¿\ uÍTß“´oó¢Ùêi¾×eÆótŒ÷çÔiÍþ»5¿GþËV)xŸ€¾[üªfŸöÅ^¤ò.þVÓxñNâ!Š“Q¿8d¾§GíñWÝ!?éPýa>•Êaž£püßJq>›Ñu¿ÐÄý ó}ÆOh¾úQMÓOŽÿ>6ÄõÞS}M\¿Ÿ¸KÓÍjâüx‚ú­Ù9qøÍyê—£ï7þ»&î?àûÕ—¨Ÿ˜uè}T¿‘ßiíEòÁAªï^jŸÅë4Í8|„ʃëó~Ç|_g??Dõ¾…âšï׺^Ó|_ݬQ=Íû“SæÉŽæÒ/5 9Oý{é9̓4®Íó¾ŸÃõÒ!ÊÇÌßP;í¡z¾B¿'??àú{žÆÓWèü§h·‡ÚçýþAŠsýþ+Ý!O,hâþ…÷_¦ë¿¬ùÕß3ßÿ1ë*óœßÿ™æ³TŽC¦}è{;C¼~lÖ ´`öí Í÷…?N×?e¾Ÿ¶Jå¤ynž_½A3£õè^sÿNkžÚª¹ÚɬOÌûƒÇ©=ðyÜ|¿¥gêOûRS}‘Þ»ìÿMÓß Mÿ6ìߪyƒ£¹æuó¼ùÄêRypýõ÷-:Ï|_ØðTGóžOíÔÿæ1êwæï,ðý î/ÜKùî ù•~ÿ=G¢uå«ÍCtÿzT®Cô{ãû“ïÑGó9šÏÑŽp´€£-àhG 8ZÀÑŽp´£…-äh!G 9ZÈÑBŽr´£…-âhG‹8ZÄÑ"Žq´ˆ£E-âhG‹9ZÌÑbŽs´˜£Å-æh1G‹9ZÌÑŽ–p´„£%-áh GK8ZÂÑŽ–p´”£¥-åh)GK9ZÊÑRŽ–r´”£±K»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³K|v‰Ï.ñÙ%>»Äg—øìŸ]â³KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$`—ì’€]°KvIÀ. Ø%»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²KBvIÈ. Ù%!»$d—„ì’]²K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$b—Dì’ˆ]±K"vIÄ.‰Ø%»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KbvIÌ.‰Ù%1»$f—Äì’˜]³KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$a—$ì’„]’°KvIÂ.IØ% »$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’²KRvIÊ.IÙ%)»$e—¤ì’”]’Ž\r­;;’É?»âgOü싟ñs(~ŽÄϱø9?‹¸®ˆëЏ®ˆëЏ®ˆëЏ®ˆëЏ®ˆëЏžˆë‰¸žˆë‰¸žˆë‰¸žˆë‰¸žˆë‰¸¾ˆë‹¸¾ˆë‹¸¾ˆë‹¸¾ˆë‹¸¾ˆë‹¸ˆˆ¸ˆˆ¸ˆˆ¸ˆˆ¸ˆˆ¸¡ˆЏ¡ˆЏ¡ˆЏ¡ˆЏ¡ˆЏ‘ˆ‰¸‘ˆ‰¸‘ˆ‰¸‘ˆ‰¸‘ˆ‰¸±ˆ‹¸±ˆ‹¸±ˆ‹¸±ˆ‹¸±ˆ‹¸‰ˆ›ˆ¸‰ˆ›ˆ¸‰ˆ›ˆ¸‰ˆ›ˆ¸‰ˆ›ˆ¸©ˆ›Š¸©ˆ›Š¸©ˆ›Š¸©ˆ›Š¸©ˆ+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _¹ÂW®ð•+|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _yÂWžð•'|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|å _ùÂW¾ð•/|_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU |_ÂWðU | _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _…ÂW¡ðU(| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _EÂW‘ðU$| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,| _ÅÂW±ðU,|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂW‰ðU"|•_%ÂWo~¥üåò­ÿÏþë×Ðgïºí–=·|øöÝ·¼ùbþöÿþ?¢*n©Ùsurvival/data/uspop2.rda0000644000175100001440000003354012111703367015002 0ustar hornikusers‹]¼pÛç­ýH²lË"3Iú‘ÂT'8IÃÌi¨aff††¹afffffnÒpfl8ßöoõuîtT¿±e­ôîž=Ï™fZ©hõlaÕÃ3f·<,r 2Ë¿La“¯Á]:µo×>›<Á áãñ}­…ïì>|+»â»:eÐ ”7ŸP«~@ ÍŽ6ê#zÎ6è¿¢ €úé ê Y¨ktÔÙ™PëMF=ÑõÌ3´˜V¨Ç£´›†oh5¼—5¼­Àûb>Þ2ð~(„Ro3ZÚºèQ½P;ïCÙñåøÔƒ Ñ?lÄhèÆ¨Vcä*Œ‹?ð/6a”¹ˆQ¸%úÐJhKä}m½‰úÓ~”"|u_ã­™‚·u<µ·ãé7OþãÄvYNL䢿›‰±¿$*ï"–_&"xáÕûã¾T·é6®Y›põè³sœy›àt{q|:†ãÊÅÜ8}à!Âþ~KØä^ØÞ¾Å–ë¡'Îê:MHνÿRëÌ%=^NPÙ?°´`ù©æÕy1g Átà ¦¦Ë0E¿?âaî9èÚªÖ‚¼!ýð<€èbnÞ©]ñö†·úG¼#ÓàóÜÅ×Ó‡ïza”ß ºÏ£.”û;› åí/(&¡XÓ£Ä$ã{™ ߍиîB‰²nJS?>g[¼¹'àY˜O½Jxž5À3l)Þ_«à;‰ºù,Z…â(OV¡ÔîŠ2õê$ÐwŸÃ” cÚ\ŒóGñ—í‹ÿúŒ“—åϺô¥z—÷è@3¾£Vx(õŠáë´_×§xïÅ»ú ÞŽkˆ›Q™Ø¿[ö:±}­DoÈITúÍD©KĵÎD4oIøÚÔ„wGx»Ç¸{´ÂÝi5îF&Ü#à^twÒ/¸SÄu»®Ò»~p¾:‡3Ç'Ï¢pd_½^?Â*ÎÄÖm.¡ë–òý2!Íž×ë_¬îü5÷bÙЋ+ 󬢔C˜&ŒÇ”ü ¶„å5¤GסZ{XxÕ—E|˜ï\kÊ¡¬>Ä×Eí9-K´z½(ô¹ëÐ6¿FË”ˆºuê½ã¨çK Žïˆúqê -¹ êƒO¢·-øuÆûåÞIíñ© Þ?ãû)Ez°­žŽ^üêü™(w¢<þõF= Û(ŒvË0ê®ÆÓH´bÆ?7£ÈŒ\лÏG›¼má"Ô„Ã(ñ‰Ã›/3Þ²[ðäù§¦h&¶±[ýtÑç¯}=Qö™Dô\MøÅ™„û»ãžéÃuè!®:+peÏŠ3Ó3ϧàØþÇìÑ8:Á~½ö1}±;ÏÖga1³±µë%÷?ÐEéß”àK°ž¨Œ5á3A­³`¹8KÅ—˜•Á\ö¦?옺×Á”ú.Ü|-=8 ½—BMY\CÆ.à[Q]E'ðí½Ž÷Ðf¼KàíWïú·òWâ›+sl],¾€šbC=Yõý}é•ì·é¢™,åQ ÎBñÎÆ·³¾¿«£”8ƒò¬ÊœSø ÖÄÛaž;µñÌ+Š7[f-ñ¢*+Q ÎÝœ%q·Üm1”4GPÎBÍûX1âøé~Âwà ¾·ñ¾®…÷äx}Á{UôÓìˆül#Jª¯('¶£ÖI‹úrZ|ÔŸ›È@©Ö ¥U¡?yŒ±¼ƶaOÇ_ïþ§óE'ÆU/F±ùâ+Ñë4÷³[fêJ%ùuÎÈÄ·Ù>ñÆX¼¥6×õ1±q«F[ªÑ}zùæ ‘¥·1õ"Ô%„7(N¸>pWmÜgqË?®{;q}‰[v©khE\­Ê㊋s_œã&âxwGæ§ØÏœÃîºBXØ| ¡ù4BÚ7$xó*‚cc­R… Y—°ü©`)yÑBÌžN˜VnÃT²¼¿‚¡ßG¨1–¹P9(wR µ~S”&=då9†R^f÷ãÔ&P~BëwÝž }çD Ë}t%mB4ïX´¢ÐÒ¿D="¾";èú}Ù+¿ˆ?WŽoß |©¶á=z_Íòx÷¥àk;åB2ÚØZèmê¡î¹ƒò¡¤èE¼êFú!õÄ[J5Ã脱!+þ—0«ˆŽÚ¢û†V_ö_óy(×à{°R¼ëž·íð|+/9Jâj$n,FÌÇ¿ˆ_€èÑD·¯LäœDÈý‡7ŒÀ½"nï|\…ã<Ÿ€sfoãJ s5Å<ûŸ±Ï_FØÍ„-(LX\2¶¾e±…m$´žøÉÌ5„øÌ×u‚ö'Ȇ¥oUÌïWaþm.f›ì©ES¥?÷ӰGüdTh&ú(%”œí  vdÅ÷ê0¾'6á(¹·Ø¼ïäîÎ,Ãg Á7Þ"l%3ÿ‹pÕǨÃÒ¢%4B+\µ¹ åE:”ÁâÑ3+£ô’]%?SªÔGY¾µìB”•ð-Û…÷Þm¼5záõ­Xˆ·Øoøl¢©Ž•ÑJÝC;³µ†üy»ÌÅÃ7¨\Aÿ(¬µéÆþ.oRáo\ÿó¾W¾b\ž†Qp?zÛÃèU\hÎ#¨ÙÅãŠÃW¾)¾2-ðVÉ‚·ó¼Ilj«/lLFb-׈̓è:݉~žàð‡XÛ´¡#AA)XÊ}üG|?k#L›»cª*<üã>ìh,»¬Ô¾kÆ ®Ÿì0µn#Ô×QFöCi>Züº/J7™}ÿ@Ôq2»E[ï@/øý]ŒŸG¯*s~$­|=´v©Ðòü…zjZÔ#4Û\´ºÝÑÔPñ¦Úø~ü¯a|AóðÛ"Y%5¾9Ë¥oâÕ›w ÿ.5þì‚êt£Z¢>•Œb›ŒÑ¤FéŠíE/+jào}P¾ÿý­è6³ðW±lÒמ(Ûd¦LL´ÏÉûxÎÏ"qû"OO ±_ bŽýNtƒzDç­CtvaâFç ?ßKø+=¢“Ï'q~޳Tgù2bÖû?öÎG±goBبG„¥ïmZnB¿Y¸…‹m ¿Iðè#X/ÜÁê´TÌŠe±xJøpÌcûcöïÀ´å*¦ºfLŽT Ù†)¢™Ö[ ¼è&g9HõâGƒM‘»_Š‹â¸’ç¾ü…[óãýÜ_–¡øvÿâ“ùÿ-5µÜ×ñò¢O¤óQ§ŒF¿ÙÚFüCòÉ ñ—²ÂÁ#CPîÔ@:5Ÿ!:ì‚/M ¼3zà­%³ûPòP¯6øRΣÌx‡ÖæÚ‹g’!—¡œûŒòõ°ðÃ=ŒÝû>I>‘ÌòIÃ/{Êÿâ‹h¥˜ä•wy–£7ý½Ôb4“äØ4EQrˆNòÝ×Î7¿d•ÚãñÆ4'®d1÷óÀNlÈm¢óD¹ªß…‹;?T’pS2î­Ò£iÝpM¹(Y2®Š3pÕüŽ«ênœG~Á9õ0Îä“8NuÃ1¼ö›%±ÛÇIžü„í˜Ð»Í9ïß?LpêXÛ$èôï%¬ÂÒéæ²ÏJÞÂt¼¤ôè.&û08˜(»¬"4¨[–çHö>)s9N«¿ðî2ñ”ÛQÆ—ÝòÊBñô_Ô}ËÐ2¾@{†Þ+ ÆOƒ1ºBŽöù"ZïŸ$ç‹g—È€zKvRÖPÉ*„n£•Ö…Š üÔ[4²_Vñ˜¯¤÷1’7%Ï(eù$ÿl–^´As+¼˜*Þaˆ¯Hn,t£qMŒÕðW/$:)+Ü<=¼µ°Bq´4Q&ÉÎÕMrã1< ü’YÇ’8BÓ®‘Xì1“Åã3=%Zúõî,‘©Ê>!÷íE¸Sé¸~ǹúÎìÂ_oâ±ßŸ,Ù¤5ö¢¹±‡É½ï_+~rÛê¬ØÄOzFòe!¥¼é‚ðº‰çÃ}â2§–¶„˜CMˆ¹›šè¨†D®OÄ… "âò>~îÍq÷ñ⮞‚«Ê\?•FH—éη¹qÖ݆3Í6ÛZà(‡ýóeì%?9íU+a¯‰„æHEHºÁ—>Žu@‚èÄMP¶mX†ŽÃüì>æ*©0ÝlŒ©ÝhLQ£@4ÇÑJÓµ°÷ZÌ.É%OHVÙ”uiÔ ¢‘[â-»Ÿ£Ì&|´u`Wñ”/hM¾¡§éˆ¾Uö{Ë«’íZ£ïoÉXmºèdÊ ´ê P¿ö@«=­ñU´ÝV´þ ¨9þFizßá¢ø‹¯<ï*½/(»M˜ ¿xГ›è›¡~vK-ú‰•lÿ¾ øGwŒ:¢™¼¿aÔX‰1ª6þ¢’é¯@ß½íÍp´ ·ÿËIJû!øºÀ;x2žnðt®Eâ¯Hl0€ÄÐVÄÔ»-éLÔAD­N„0zx…û¸§'àº&ž^à Î:_qܪˆCteŠ=KA®?%lÆmÂÊNÃ6{¶Ÿ‚$5ε:EØMؽd|e;{ð”슧`µC$¦êBÂáŠÄ¤úHÔ–ÉDõ÷Õì6Ó\„‡ŽÆ]þ®‘‹p>»€3´?ŽÁ­p䜉=CnÂŽ®'¬CzÂÒÂv¾ 6}3¡+5Bà ÒòÁ÷" ΂µã‚ž„ôó7,£sc¾ƒ¹A6L¿a:\²ümxvÖkð›dùZ’U ˆ—d8‚ ($yPE9¿å ì•-ÝQFD.Ÿk÷d|G¢ñ­h‚od¾ýr—•FY“Nv½ä“MÅÏÏ¢ÝÌŠî”ܘ² u÷¹WÙQ­Ö£–ˆFyZõ'ÉéãJ¡/û,°YI|÷žá«pŸI|¥×7¼?ä¼:^ü¨"Úáhô¼KQ·ÉÌ»æ f¶ˆ«.Zhq½'†ðƒ?è øÊ4ü¤WÇe‡îNÂPKHŽ\®çB½4V4ÚIø»>s$Þ7ýñ¼ ÅkI‹gmb_œ'fhbzüF̈WDm!2}E"Ú$|UÂ5w¥h\¿â:Rçžš8GeÂY~Î|›pzÚãè9 G–³ØeÎìYÚvd,aÁõ±¥:Jèøµ„l¯Bð…aX/–ÁjZ'||ËÜ X¢$Ëÿ¶3û1 OÀôS:¸•V<ÿ(t—¡M›‰Ö13ZµŸÑŠ‹¤¼G %ûFf5‹dñü6Ô2y°xh!á¯âA–¢ýõ}a,FR(ÆŸ+ñÇH¾‹ˆC¯.³¾Ñ„¶ý ÚŒòºòÚ»ª¢ŸžvÚó{¨óꢜî&¾_2þk”‚‹ðM”{L7µýRôˆÙè—$wD“](^§ ­ÜÂ`#F¹Æ©|ÂÁ.ŒFŸðG†£/JAï7muFÔ?£®—=«¸„넹ӷÄsOÈyî×&áî_$ô½Fôý¢DumATžjD%#¢ì#Ü;;âú~W®‡8'•ý´GÚÂØEa‡Í„Õ•÷ÒÉ‘5ñÜ|gèfb×%¦þebÊõ%¦ú2¢†œ$âÝ'"r-&¼ƒì±[Sq»JâZ3W7?ÎÖïpfËãÙCç«àXt»dûÒØõ0Â&µ%ÌøŠ­Ö.B» "䦛ë‚ É’™«ÔH2äº|Xb6`?³k&¦Y=e— ‡'õa©|í² î®’=/Y|[Z´ñwÑzÖ”,(Þݤ$ZN™»¤BhõÓJÎoóvR!É3¿Éóó´”ì-;&uFôÛŒæ÷ñ»ãO…‘®z»¯hW‚Ñ^6B»(Y~Ì`´[ÑC\èEKéz¢^Ë/y½,JmaïÕQše÷©?J‘‚¨CÖÿÏëõ{YÐdÞÕŸ7 ?ØQßöAÿ”Q²ÊQaäj©…ûÊ¿+‰>Xæ¢vÚ°ú¨E_CD'¯úà}™_8ØLâÍW$ž°¨: 3­$d=MôÜD%Ý#òe‘G¯þe6îVYq-k‚ó–Ž3%GÅžØNÁÞ"auJ ïÇ6j-¶œ.BÏM&Ô»’™5 þÜ€à_Öc=èÁ+:i˜ËX Î’Œ’sÂ"L+›c*— ×%'Vƒ1ë ™h¤„<’ÄS´£/j7ñ†Š¯P“äÞ½o„?Å"‚QMBy"Lë߀’”%ö&¾÷¢ ²w¶§Fo„Ú»§ä5ñ•mrß¿ØÑûˆn·EËTuä¿K¢“­oQ‹?¬‘=Ýí <§ÏtaÝâOÂEKã«]ß)?¾+ð=»‰š{/Úß±’ѧIþl-ïM¸¸¼Cv©ì«<Ã1ÜÇø+¿¥þÆùðÿéÇØ– cÉyaé{º4hß…ÖJV=±ßÙÃxOÁ+=÷¬i‡gó9<5;ï%19Éñ>!FŸCT…`"v|#üó“æ §Åu¨®úàÒkàŒ¾%©#ÞŸGýÑ8Ò7ÄÞK¸Ë“B؈’Ø„ElÝÅ÷WÆrø$!iJŸÚ`í¼š ‘é°œŒÇ’º æI{1+?aZVS‰âðj0¬k ½;‹A’µ ôБhW¢mn] ž:s%ÚDÉàÍD+êËYîi™Ü넽ÂÁ e/ôAë;DîízÃA‰á³jâ—üë¯Z £hú„3’­’ej¢;D3Çš£'ŸF/<}Ömôú…ó®Š_t¦8/ü%»fØ%|çf Ô”Œ2½z6É‘Ï ÈÞîË?KöØ:ÉQÂZ/Rcd“=ÆA Wia±.’)ýò^£çü ­ÞGÔµP…ý}{†àÝ‘ŒgwMì!qxê§!¡Â âoýJtu•È߉œ“‡È _x wôp\åÏáì¿ ÇûóÍØë ìKsÑI2¶>«°E—"tó~BË$d’Jˆû,Á­c±Ž5‡ä“‘‹±ˆÇXê~Ç|23æ¿aÚ[S½ ˜ ßS„¿Z w•] ɇÀh )/QH®˜ùµóHÙO-QKÊ|/ZXΞÏr’ÕkUDi^S²ü”Œa(õ ¼ñI¦ÿØ¢Ê~[ˆön7ú°“è›>¢]?ŠVw“øòc4w6ñ˜J¨‹Ì¨¯¤ïÕ4ô ¼ŽzÙ#}¬%YN²Í¾'2—eîWcTøãÀhüå‹ào'yÉä¿¢g=‚^Q|'·øÊ§Í’käÏ󣟱Hmá¼|²›êI^])<žÜenU|ý‰Òr‚ôñzŠxËsMßšGúœpõžìµ{0|¢EÉ|ú›²îN«/ ÿ¼EØ\´•]2V^ñ ÿ‰ÿeRï€õx$’Øè‰…Ç3øoˆÚ’è°%DN­Nd©ƒDFýMxÉŸqmóeœêzŠb’LØëÔ„ û‚­ï-l!a„N…Ð仄ìïAHX/‚û§`½„5g,AKe_YC±4˜‰ùaÌõ®bº½S{“'äÄP˜•ÒCŵk$¤š ?ŸG‹îz.L¿wêŒs¨/£öHÚ}‹x|¦J²·jEd‘ûš+s-þÒpÊ4a«Ì’+o&Þ+\4Sz“é¹d8ñæ'ÃeÆ%7ŽÎô­´Kî¨3ê_7вJ™4 =¿|ÿùSa´6(k‰6Žà[ÒÅ=J4ÓÅèƒÚ\îXíŒ>\öåwaÒ QÍ—>ÞÃ(,=y4ãv]ŒÏuegÇXÁ˜—cdNô¿£[¦cöÏ…2y'¾I ðŽš‹·åbÉè-ä53Ês§,úŠþcú®Âywħ^‰[QÊ|BÙ+<±¥›xªªèy ?Mæ{‚š½j¸ÌÁ¥zèÇËaE?uTê÷= “ “Ý-žxG4n‰jkˆòì%¾_Šá-¶O¡8•e$|0ˆ?4„øÅ‰Oÿ™¨­ˆ,šˆ͈X[÷ý׸~]‡s¤xúVÉ*Á]°'\&ì·„E†b³u%tÔBã BV $Ç[‚û¬Åúz3Öò} Z¾“ õ>–0Ë4w»éëTL#»aJ_n ÿ.\ ÝšCÕéGD—¶6ŠGküNæNf7ü9ê£dÔ£GQ7J_)ž¾Z¸gMEÔÑ’/&B=}åªðë®k(³Ž¡\ —Ì®¢žÌ)Ü< íRyÙM«ïÇb$|A/ ñ¨pB3ÉŒCO¢uö §uBëPXö£Gv˜Ì„*œ·MüýêB”Ì™ñ‘ ™é¾YQr?Cíó3z’dÈ)â%a²;«Vš2…c" ÖO2ë.Œ7¯ð—I‡[Œñ/dÚ¡Ÿ“ÏõüƒäWù,MB¤Ç%ðu’<ß´3Þü—ñd>ƒÇ¸Eâ¦ñÄ:~=íÑm}D‹†"‡Õ!üõ Âý/p—ß$|< ç±ï8+ÁñešxÉì£/bO»Š°»Ç ¾ÛÑØJ|#tN6BÞ'¤n9‚ǺnVׂʖÆ2FöØ%æ‚›1J#ž¿“»9 l<SüzŒL„_ÑÏ^AŸ/œßC˜©ºìûBë$ ßæ CÏ!üž=æoôÊ·än3ÈlKž©ú«dÉÝŠ¢/¯øücÌTŒ­ñwj„ÆSŒ¹ Ÿ~F*Þ°J^gÆôº:ú1y­+²{2·@0m„äÐy§QþÊ- ±åPsË(”ß„Õ×I¦É-¯ñà8ZÜjɹÂÙAùPîEßè6“÷¶æ#úú.èG¥~½ É]—ÑÖOE½ ™çÂm”uƒñEUÅk9@âçã$\t°0ñíÓ_j q‡‡UïO~"bÔu"Š÷ÀÝ·Î×Oq9q”=}òT–î$Ìm²dñ^B] ™¸‡ÿU‚—èX_^ÅÚp Aûd|Ä2³![Éò»0GˆN˜ zà™VÏ€>Òêï!9H?Š}B[)s4"­A ZÊA´ù¼ŸÅO/=-Ú)‚ú´&êq¹ƒ3wÄÛM¨Ñâ-Ÿ§£\o‹*}R—'Ê÷Ÿ£µ¸.ûMz8m'F*¹›²éÑ»4=Xm¡xÎ?l°ñéçm´U½$CÊÎ\2 ½œä£{SQ¾‹^Ê÷Ã÷ $JÉ,ø6·D©0 uŒxy·è E‹1Âduå½t8„V¬¶øÊBÉõó1.%c<~¿ÀTüËb tc4›Ž¾C|íÒ ´Y§P‹ CT©¯lc¼yoàu%‘øñ-‰>‘Ø©)1çê]kÑé>>„È<÷_0÷¹*¸>zq w:[ŠNžÊ.mÂÞ}öè „-Ú@XUE8y,¶ô/ۈׯÅO¤O?°^>¶V!¨©äøÕ7±0sõ£˜.~ÅÔ"¦Èp&&žÀ”îFÉYe„)õHô¢“Sóeæ“ÐG ‹uë(™\ry‡ŠèÒ‰FlèurgÉÚýZîU43E¼út8z×§²+ŠK¾Þ.Ú8ß´ÿ„5ø÷åÇ8*žœMØkKô[²ïÿyþXñ“¯9Äb1~À° [I†RwŒ@ù$úèø]æ{ ŠSrêàî¨ëC„ñÄ7nÝ/“jYPÞç OL¿,¯ù}ŒxÛ©3o)F²d¢Ââ¿I¦ŸÕM˜äJ‡2xÏvÀ³³ ‰«/’0ð3 eS›¸·ˆ«ƒ(Sc"†""u2áW̸#æàòdž+Øÿœ‰]ßGX¶ØÆµÀ—›Ð¨BFm#Äå&xÊ‚ã|¢É'— $3n™û3u(æÉ 1˼›6I*ÿ%ü»¶ª0¸,ÔYïA&ñ“Ò¢‘—²ëÏÉ{_%Ùax^´¦×ÐJ 'Õþ|…¦\8–^2^ˆì,ûR™Oùœ•ª › —zEWÕÄSIÎ,²mþ É…™d‡´Â¨&ù±iôÕâZ¢ý•"ù47šäm¾d˜û’WóJ¹,üÚ]´dOÖ¤mf¢˜¢4(.ùތҨ2êÜ1¢'Ék„ Tñ”&„]D{Å{K^yŽqý:Æ)/Æ­‘ø³ÍÆ?]²~‡Ì•„Û4@Û&5ûHæ2 ‰oåǧ§àutÅóG%w$qQ13V fl.¢ãŽuã,Q»Äïv"¼BÜÝ÷àšQçÓJ8ƒrãè{O2ÊV즅„M|E˜·'¶½°ýb#tH´øÉ:B #x\Ö·°föª× ËáöX¼Y1÷y‹éqgL]‡`JlçNÀô±˜R:b¬kŒ1¶Fë¬ÅOa2gßeönäFߟ }]ú ÑÇÚg2ç3ÅÄ——ʼ>‘=Úíõv´+Mу…{f?>Jƒ‘Zt±á,~}þ­“ð?Šñn&†°ºþxF„ä=SuôƒVŒŒ1’Öaôn€‘AØüÀ`ÔsåQ¥Qˆ/ß8€5eè¨k§IÞ‰–ŠY$¯FE¹·L¾>zß?¤_‹ÐÛˆ¶%ƒa‘y1 ­”äú†×Qjé’y_àúO»º$Ö)FB–&Ä¿ÜMÜÆ¥ÄõÉJ쓱DάO„á[žÞà9®9ÅpÆÎÂQPÃÞ¤¦xJ8¶­±E ÷N8JÈø6„„l xÐS‚ÚašLÐeÑJ™œX6Ãâÿ€yùÌ™ScÚÖS£dL®¯âë’OFÝ…F¡hWÈRA²JmÉ„ÓÈìþ‰v¹&ÚñîéÑúI¶hi—<Þ­ÎñÑ@ó$´¢ÂÅ¥$C–?ájïF¨u„Ó&A3^ õ>º,9®ýhtɈƈ÷Ûƒ¥WµäÞ΢§6+õz²¼þŸ…Ð$C¶’Y0‹ÍöÎ$²àa”CP¼½Pº·ÁwgJ—â€ÐkÈo³ ¥é…Úâ°ä(a¹bW1òVO; G™cdÖÿ9Ëlg)'ž“Í!\&^‘åúÖWâI)åná/Þÿ­i¤ŠÚ…ß/w?AÞsÅæÊï /¿+#~b‘='s=S~öÏß׺:GòeTÏv”ÑQÂb2ß±_E+„ÑeωêÇΣ¾;‰Ê ”3y%ßÌC¯/ì÷»0Z)ñ¾B’ËdË*|X Íí”Lß%®¾Û‡ñ枇Ç;‚Ä` ñÊ/Y2®œBœÃ ¶C."“ŠN–^̆ûÁ%\iUs÷c?¢ö$аô °åIKè¸1„::ˆŸÔ$ø·LX?×ÁÚû#A÷T®–C±äœyMkÌùÏc:5S³*’åwK–ÏSw@‹uPÊ€¤Ï’UF 7^‹þ‹=E>ƒ"»Ý:í™pýyÉ’3µ%ÂH³E;sÊ¡ _hKhÿÙ ò8šõ²äšÝ»Qg˽œ­ŽV» ÚÚcèAºóúcdqÊì6ÄøÐKv“dŽÂÝÍäÞ†Jm-;'\´SZòÿÔLéÇ˾lVö‚°•ì®E~”ì²£Wã{7÷çRÝ•½ÑŸò:¢©Œ#P[?­si‚‘CØëäEŒ­¢›Ã½ñËnñwn'}þ€¡ÜAoîBë&ZOºŒrz¾³‹ñî‹gV <µSbHt·!ah&¢¿üNT_;Q9ÒeYCDæ ¸'Áµ1-ÎÓ>œQ©†`'ü_ Û`lýžúÁJè`5o%$¹•ø~5¬wÿĚ܅ )·°Ü¾€%Ë9ÌKbö…cš'™$¹Ü‘þÌ÷bªº^4žØAŒ×é0®ˆ>ödÀX´c¸ìüÖ¢Ÿ_FÈ^#;¶hG<¡ðkŒŠò¼”ñroíe’Ù\.ûG<&w5™Ñ|2Ã1\G1ªK~ûûþ6•H¶‹T…—â¯*÷sÄ%¯«aôx‚ÑD^O“z}…¿º—ÄØr£V[™‹¢ƒÔ´Ò¹ åál”xaðÁÃQWA÷Dß^BzÊs'ÊÖ2¨Íî£l‚ÞRò’.œ#}ðÈû›ÞÍÓUX>;ÊÍõ óñéç~3Ü#a™›øŽ:ñFwbo–"vB-b^å!bO‹Gâ>™w“®8—Yq¤îˆ½ä[ÂZ¼Ã¶x¡[:j_MHÿú]…õïôXÛÍ!èÕ‚ÚÇrT2JñÓ˜÷6Á\d?¦ë¯0u¯I]7Wƒä:Þ‚r+!{¨%{hºøøðN’#„‰È–o…žGØ2µä:W#´Oå%‰÷ÞŸ¸-_H–yøíð´3²§MÐ%¿ü8†zu¨°“0òØ´›Û%k®F¿4 £iŒGðçn]XöŸÔY${oßɩۄÓ w {ïŽ.·ý^)ñ²¨=Z¡ìÚŠRZ˜x^Šõ¡ôGüÿ¨øU«}ŠY$µýµ›pU>’AÿùïŽÂ÷+E£[ ág1þ:‰Â’ë¬A/*šª Œhi‚2+¾éðþ–€§ö7•,Y³!·sRm¬ôéÖÇÅ±æŠ h”†å}),¥Óc^2F<¿(¦õÂ]%ûJ† ËÒŠÿ‹NjÅ_x%þLyñÇ]Äo­Žñ¦¾°å Œƒå0–—Å'¹£_g™ka¤þ’e&üŒ1ïÆÙ7#Ä;vŠ/“ý'L¨W>€.ïI¿.ž›E´Ð[îD¹†êhRå>@ªÎ{ð—=óeü¾¼Ö ñ• !%WI®ö›(¯wZ|º­ÜÝ×sÒëu¨Y$ŸN‡òh·øÊBÉö[%sD »…$“©g¦ \Þƒ2§=jU4t½ˆ°öß27eW=¸‡^WxäúÔnékj|}îã“]噋Ä&-H(v“x™§¸µG‰mv˜ØÈGÄ4/H„dW÷鞸ËÙqݘ‡3uö…’QíÄv߆M•Œ’4Ÿ&‚ßvÅúC¸·ÍŸ=€ f±<ø§1Ÿ)„¹ÒGLw®b<Sæ³p·¬=%ÇWÝäút‘Œ(ž¹GòØ*ÑÆìèãdç.ê¼½áÑÍ#ôŸ—ˆgCOuÝ=•®¥ðZ)ÑÈ.a6ùyÅKâ#¢¼“Ñ¢š¢åû"™Ò,w&L×b†ó0Æü}’±ÁßMvàꯅÄsn/EÿxX´´@r©øñ¼ƒè÷ÿÂh'saÚŒ6Dîx|W”‹ePênBY/÷3X4“„z>¯üNÉAQâ³PÛ¼–¼’€–§‘°¸dÐ-å1fËc‘ðÄ“2w?Éû("³"YÉ„æÏ"þ'Ù´i|uþÄ›¿ žØ®$ 'aÊvêŒ&^ü9º\u"O$Ùá+‘™7þë\»îâ¼X Çý‰8TöôI„É{µ}‘=v§6¡•R²ý'BRº¼,?Öëϱæ/MÐØ XžÁRûÌ{º‰Nz`Ú+{¬Z[øVÖGbêÔÿèÍø»ËׯñWοØ]ü¹+àÏP |2~ó;Œ§1.KÆ8&lyt,Æù÷`¼¶:ý»dµ8ñ ¹Ï7ÒÃ5½…äÜZúøE|¨Ò?lþÒ™ð¹AªÖ™Hµ´$þÝðgæÚ*=»*þ¿cF+ÑÊN©±â4ÆÝaâ1“%õ”¬Z5YC™.lü°9JäY”â5cï£}oŒ>~%êÖo(»G¢ šš©,ÚG™µÄjhÇ‹ mœ¶Zr¬Òuª0]þ7(EjáK‹÷ZO<#Hd8ñv$nzâr%óx11#[}µ*áÇஸבþ¸*4Ç1#vå2aùÆ`«5W2Ê@BÝ'ø}C‚mÃÚu A÷GT÷-–»°4IÁ|aæú1½,…i„74—¬÷c†Ü¥Ìé½YèW„ýÏÈã¨ÌÒž†èJ¢/”l7A²ØñœŽÂJÍ…ƒ›Ê¹íÏèVÈ®’¬XSvõá%ÙOÚ?ÙºÄp´V;Ñ^´"œ5ï Fé_åž§âï/9{ûYŒ—Ñ} aÄK†É4#H8v†p×ùYhsŒß[aø¦ï]G]ÖQöÖ3”YPö‹VüQV\ÿê#>( ~L´™$šj µƒœ³ËÎLÎXöcô+Ù½“0Î,9Iüsã´W©Q_Kæ™r%Ý-|žÂÆÇŠ“Ø® ?¯#þÓ@âk\"ê™Èfmˆ´v%b‹÷É)%Þà¬ÑGӡا»›[Û‡ÌØî%´Ô;B6¦#$ÎGð˜6X¿ŽÇšý1AS{by} KµM˜ÏÆ\؆i_}L M˜BûÁUöÚIü{;á_Ù ÿtñÑkäÎd÷÷œ‰¿cVüMŸã¯"ÚI‘™¾ƒß(…?Q´$líO³ùžíÏ“dmü¡’×õ¶²„»ö 't“œ8ò>†zM|H´r&=þ‡ðÞCª¹ÙIuc6þwDŸM1ŠFÍ/0nÉy¸xÔ%c¯üΛCâWÂÍáÂöÛ¨9ò L“¿'ÞâȉÒVf~@*´'½Ð{nDÛeam”fâCvÉ»g¡=Ÿ!³ò­Ï´öòøCzW~¦°Z|Ïjã=)Z麌ăµI˜X˜øåˆ3Ö»é1Gýõ5Ñ… ¥à:Ù WÞ©8·_•¼¸‚° °­]Kè©­„Zfï$¸é%¬êôb?AõæaùSòIÃó˜ŸvÆÜÄ‚éKeL“þÆ”¯üí…-?`Pe¨÷ -€¡Â¸õ0R­ÇˆwÞÀ°‰Ÿ‡ˆÉ þ¿ü :yÒýÚÑO9ô-¿Š÷T–Ç*ôÍßÐw,v‹DŸ%>z67Ú¹‹åˆ^ £Í”¬iHVï%¿sSøk„dÈÔuðj‰ÿË0üK`lök‰Ôî‰ñÓ-É'=ÿ÷߯ŒtÂËÅ[råAÛ’N2P”/QúKOÎÞBÉ"µQo%HÿÿѶxZÉ‘-rˆ^¤™dÇÊ*0¦wÄè)»°‡ÌÇÆ{Âa7„×d'K¶KrÕ¶!¨Õ⻇÷Ü=?iŠùd)Ìa:HXëwLßàÈ L³ôÿçÿ2aiÜ¢|qþs”G“ÿý) æÿyZ¨<­mƒ6M:É9êÿ>7üÏÿ??7%ý{Èúï!Û¿‡äÙÿ=äø÷óßC®)ÿrÿ߃9kRà”5pÊ8%NÙ§SÎÀ)Wà”8jd ÔȨ‘-P#[ F¶@lÙ5²jd ÔȨ‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘=P#{ Fö@ìÙ5²jdÔȨ‘=P#{ FŽ@95rjäÔȨ‘#P#G FŽ@95rjä ÔȨ‘3P#g FÎ@œ95rjä ÔȨ‘+P#W F®@\¹5rjä ÔȨ‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘¨‘;P#w Fî@ܹ5rjäÔȨ‘;P#÷¿5,Y“’þ¯àÍÿ÷[Am´þWüÁM›üïOÿç1ÿ>#[RRÒÎYÿsÎöŸsòÎÙÿsÎñŸsÎÿœsýçœòŸsîÿÿœõ?u³þ§nÖÿÔÍúŸºYÿS7ëêfýOݬÿ©›õ?u³þ§n¶néû?ÿÍú`ÛßGsurvival/R/0000755000175100001440000000000012545056257012357 5ustar hornikuserssurvival/R/survreg.control.S0000644000175100001440000000065011732700061015642 0ustar hornikusers# $Id: survreg.control.S 11236 2009-02-14 11:46:53Z therneau $ survreg.control <- function(maxiter=30, rel.tolerance=1e-9, toler.chol=1e-10, iter.max, debug=0, outer.max = 10) { if (missing(iter.max)) { iter.max <- maxiter } else maxiter <- iter.max list(iter.max = iter.max, rel.tolerance = rel.tolerance, toler.chol= toler.chol, debug=debug, maxiter=maxiter, outer.max=outer.max) } survival/R/print.summary.coxph.S0000644000175100001440000000423512423460656016454 0ustar hornikusersprint.summary.coxph <- function(x, digits = max(getOption('digits')-3, 3), signif.stars = getOption("show.signif.stars"), ...) { if (!is.null(x$call)) { cat("Call:\n") dput(x$call) cat("\n") } if (!is.null(x$fail)) { cat(" Coxreg failed.", x$fail, "\n") return() } savedig <- options(digits = digits) on.exit(options(savedig)) omit <- x$na.action cat(" n=", x$n) if (!is.null(x$nevent)) cat(", number of events=", x$nevent, "\n") else cat("\n") if (length(omit)) cat(" (", naprint(omit), ")\n", sep="") if (nrow(x$coef)==0) { # Null model cat (" Null model\n") return() } if(!is.null(x$coefficients)) { cat("\n") if (is.R()) printCoefmat(x$coefficients, digits=digits, signif.stars=signif.stars, ...) else print(x$coefficients) } if(!is.null(x$conf.int)) { cat("\n") print(x$conf.int) } cat("\n") if (!is.null(x$concordance)) { cat("Concordance=", format(round(x$concordance[1],3)), " (se =", format(round(x$concordance[2], 3)),")\n") } cat("Rsquare=", format(round(x$rsq["rsq"],3)), " (max possible=", format(round(x$rsq["maxrsq"],3)), ")\n" ) cat("Likelihood ratio test= ", format(round(x$logtest["test"], 2)), " on ", x$logtest["df"], " df,", " p=", format(x$logtest["pvalue"]), "\n", sep = "") cat("Wald test = ", format(round(x$waldtest["test"], 2)), " on ", x$waldtest["df"], " df,", " p=", format(x$waldtest["pvalue"]), "\n", sep = "") cat("Score (logrank) test = ", format(round(x$sctest["test"], 2)), " on ", x$sctest["df"]," df,", " p=", format(x$sctest["pvalue"]), sep ="") if (is.null(x$robscore)) cat("\n\n") else cat(", Robust = ", format(round(x$robscore["test"], 2)), " p=", format(x$robscore["pvalue"]), "\n\n", sep="") if (x$used.robust) cat(" (Note: the likelihood ratio and score tests", "assume independence of\n observations within a cluster,", "the Wald and robust score tests do not).\n") invisible() } survival/R/print.coxph.penal.S0000644000175100001440000000730312536376005016054 0ustar hornikusersprint.coxph.penal <- function(x, terms=FALSE, maxlabel=25, digits=max(options()$digits - 4, 3), ...) { if (!inherits(x, 'coxph.penal')) stop("Invalid object") if (!is.null(x$call)) { cat("Call:\n") dput(x$call) cat("\n") } if (!is.null(x$fail)) { cat(" Coxph failed.", x$fail, "\n") return() } savedig <- options(digits = digits) on.exit(options(savedig)) coef <- x$coefficients if (length(coef)==0 && length(x$frail)==0) stop("Penalized print function can't be used for a null model") # # Map terms to special print functions, and the list of iteration histories # pterms <- x$pterms nterms <- length(pterms) npenal <- sum(pterms>0) print.map <- rep(0,nterms) if (!is.null(x$printfun)) { temp <- unlist(lapply(x$printfun, is.null)) #which ones are missing print.map[pterms>0] <- (1:npenal) * (!temp) } # Tedious, but build up the coef matrix a term at a time print1 <- NULL pname1 <- NULL if (is.null(x$assign2)) alist <- x$assign[-1] else alist <- x$assign2 print2 <- NULL for (i in 1:nterms) { kk <- alist[[i]] if (print.map[i] >0) { j <- print.map[i] if (pterms[i]==2) temp <- (x$printfun[[j]])(x$frail, x$fvar, ,x$df[i], x$history[[j]]) else temp <- (x$printfun[[j]])(coef[kk], x$var[kk,kk], x$var2[kk,kk], x$df[i], x$history[[j]]) print1 <- rbind(print1, temp$coef) if (is.matrix(temp$coef)) { xx <- dimnames(temp$coef)[[1]] if (is.null(xx)) xx <- rep(names(pterms)[i], nrow(temp$coef)) else xx <- paste(names(pterms)[i], xx, sep=', ') pname1 <- c(pname1, xx) } else pname1 <- c(pname1, names(pterms)[i]) print2 <- c(print2, temp$history) } else if (terms && length(kk)>1) { pname1 <- c(pname1, names(pterms)[i]) temp <- coxph.wtest(x$var[kk,kk], coef[kk])$test print1 <- rbind(print1, c(NA, NA, NA, temp, x$df[i], 1-pchisq(temp, 1))) } else { pname1 <- c(pname1, names(coef)[kk]) tempe<- (diag(x$var))[kk] temp <- coef[kk]^2/ tempe print1 <- rbind(print1, cbind(coef[kk], sqrt(tempe), sqrt((diag(x$var2))[kk]), temp, 1, 1-pchisq(temp, 1))) } } # Format out the NA's # temp <- cbind(format(print1[,1]), format(print1[,2]), # format(print1[,3]), # format(round(print1[,4], 2)), # format(round(print1[,5], 2)), # format(signif(print1[,6], 2))) # temp <- ifelse(is.na(print1), "", temp) # dimnames(temp) <- list(substring(pname1,1, maxlabel), # c("coef","se(coef)", "se2", "Chisq","DF","p")) # print(temp, quote=FALSE) dimnames(print1) <- list(substring(pname1,1, maxlabel), c("coef","se(coef)", "se2", "Chisq","DF","p")) printCoefmat(print1, signif.stars=FALSE, P.values=TRUE, has.Pvalue=TRUE, na.print="") # # Write out the remaider of the info # cat("\nIterations:", x$iter[1], "outer,", x$iter[2], "Newton-Raphson\n") if (length(print2)) { # cat("Penalized terms:\n") for (i in 1:length(print2)) cat(" ", print2[i], "\n") } logtest <- -2 * (x$loglik[1] - x$loglik[2]) if (is.null(x$df)) df <- sum(!is.na(coef)) else df <- round(sum(x$df),2) # cat("\n") cat("Degrees of freedom for terms=", format(round(x$df,1)), "\n") # cat("Cox PL (initial,final) = ", format(round(x$loglik,2)), # " Penalty = ", format(x$penalty), "\n") cat("Likelihood ratio test=", format(round(logtest, 2)), " on ", df, " df,", " p=", format(1 - pchisq(logtest, df)), sep="") omit <- x$na.action if (length(omit)) cat("\n n=", x$n, " (", naprint(omit), ")\n", sep="") else cat(" n=", x$n, "\n") invisible() } survival/R/survreg.old.S0000644000175100001440000000126411732700061014742 0ustar hornikusers# $Id: survreg.old.S 11166 2008-11-24 22:10:34Z therneau $ # Map the argument list of the old survreg to the new one # survreg.old <- function(formula, data=sys.frame(sys.parent()), ..., link=c('log',"identity"), dist=c("extreme", "logistic", "gaussian", "exponential", "rayleigh", "weibull"), fixed=list()) { dist <- match.arg(dist) link <- match.arg(link) if ((dist!='weibull' && dist != 'rayleigh') && link=='log') { if (dist=='extreme') dist <- 'weibull' else dist <- paste('log', dist, sep='') } if (is.null(fixed$scale)) scale <- 0 else scale <- fixed$scale survreg(formula, data, ..., dist=dist, scale=scale) } survival/R/summary.coxph.S0000644000175100001440000000556611732700061015316 0ustar hornikusers# $Id summary.coxph <- function(object, conf.int = 0.95, scale = 1, ...) { cox<-object beta <- cox$coefficients if (is.null(cox$coefficients)) { # Null model return(object) #The summary method is the same as print in this case } nabeta <- !(is.na(beta)) #non-missing coefs beta2 <- beta[nabeta] if(is.null(beta) | is.null(cox$var)) stop("Input is not valid") se <- sqrt(diag(cox$var)) if (!is.null(cox$naive.var)) nse <- sqrt(diag(cox$naive.var)) rval<-list(call=cox$call,fail=cox$fail, na.action=cox$na.action, n=cox$n, loglik=cox$loglik) if (!is.null(cox$nevent)) rval$nevent <- cox$nevent if (is.null(cox$naive.var)) { tmp <- cbind(beta, exp(beta), se, beta/se, 1 - pchisq((beta/ se)^2, 1)) dimnames(tmp) <- list(names(beta), c("coef", "exp(coef)", "se(coef)", "z", "Pr(>|z|)")) } else { tmp <- cbind(beta, exp(beta), nse, se, beta/se, 1 - pchisq((beta/ se)^2, 1)) dimnames(tmp) <- list(names(beta), c("coef", "exp(coef)", "se(coef)", "robust se", "z", "Pr(>|z|)")) } rval$coefficients <- tmp if (conf.int) { z <- qnorm((1 + conf.int)/2, 0, 1) beta <- beta * scale se <- se * scale tmp <- cbind(exp(beta), exp(-beta), exp(beta - z * se), exp(beta + z * se)) dimnames(tmp) <- list(names(beta), c("exp(coef)", "exp(-coef)", paste("lower .", round(100 * conf.int, 2), sep = ""), paste("upper .", round(100 * conf.int, 2), sep = ""))) rval$conf.int <- tmp } df <- length(beta2) logtest <- -2 * (cox$loglik[1] - cox$loglik[2]) rval$logtest <- c(test=logtest, df=df, pvalue=1 - pchisq(logtest, df)) rval$sctest <- c(test=cox$score, df=df, pvalue=1 - pchisq(cox$score, df)) rval$rsq<-c(rsq=1-exp(-logtest/cox$n), maxrsq=1-exp(2*cox$loglik[1]/cox$n)) rval$waldtest<-c(test=as.vector(round(cox$wald.test, 2)), df=df, pvalue=1 - pchisq(as.vector(cox$wald.test), df)) if (!is.null(cox$rscore)) rval$robscore<-c(test=cox$rscore, df=df, pvalue=1 - pchisq(cox$rscore, df)) rval$used.robust<-!is.null(cox$naive.var) if (!is.null(cox$concordance)) { if (is.matrix(cox$concordance)) temp <- colSums(cox$concordance) else temp <- cox$concordance rval$concordance <- c("concordance"= (temp[1] + temp[3]/2)/ sum(temp[1:3]), "se"= temp[5]/(2*sum(temp[1:3]))) } if (is.R()) class(rval) <-"summary.coxph" else oldClass(rval) <- "summary.coxph" rval } survival/R/summary.pyears.S0000644000175100001440000000170512505342523015473 0ustar hornikuserssummary.pyears <- function(object, ...) { if (!is.null(cl<- object$call)) { cat("Call:\n") dput(cl) cat("\n") } if (is.null(object$data)) { cat ( "Total number of person-years tabulated:", format(sum(object$pyears)), "\nTotal number of person-years off table:", format(object$offtable), "\n") } else { cat ( "Total number of person-years tabulated:", format(sum(object$data$pyears)), "\nTotal number of person-years off table:", format(object$offtable), "") } if (!is.null(object$summary)) { cat("Matches to the chosen rate table:\n ", object$summary) } cat("Observations in the data set:", object$observations, "\n") if (!is.null(object$na.action)) cat(" (", naprint(object$na.action), ")\n", sep='') cat("\n") } survival/R/residuals.survreg.penal.S0000644000175100001440000000063311732700061017254 0ustar hornikusers# $Id: residuals.survreg.penal.S 11166 2008-11-24 22:10:34Z therneau $ # This routine just stops disastrous arithmetic for models with sparse # terms. A placeholder until the proper sparse terms actions are inserted. residuals.survreg.penal <- function(object, ...) { pterms <- object$pterms if (any(pterms==2)) stop("Residualss not available for sparse models") NextMethod('residuals') } survival/R/frailty.gaussian.S0000644000175100001440000001041512444044245015757 0ustar hornikusers# # Defining function for gaussian frailty fits # frailty.gaussian <- function(x, sparse=(nclass >5), theta, df, method=c("reml", "aic", "df", "fixed"), ...) { # Check for consistency of the arguments if (missing(method)) { if (!missing(theta)) { method <- 'fixed' if (!missing(df)) stop("Cannot give both a df and theta argument") } else if (!missing(df)) { if (df==0) method <- "aic" else method <- 'df' } } method <- match.arg(method) if (method=='df' && missing(df)) stop("Method = df but no df argument") if (method=='fixed' && missing(theta)) stop("Method= fixed but no theta argument") if (method !='fixed' && !missing(theta)) stop("Method is not 'fixed', but have a theta argument") nclass <- length(unique(x[!is.na(x)])) if (sparse){ x <- as.numeric(factor(x)) #if there are missing levels, drop them if (is.R()) class(x) <- "coxph.penalty" else oldClass(x) <- "coxph.penalty" } else{ x <- factor(x) nclass <- length(levels(x)) if (is.R()) { class(x) <- c("coxph.penalty", class(x)) attr(x, 'contrasts') <- contr.treatment(nclass, contrasts=FALSE) } else { oldClass(x) <- "coxph.penalty" # Splus allows us to pass a function as the contrast, R doesn't # For large values of nclass the function is smaller # (But for really large ones you'ld use sparse=T.) attr(x,'contrasts') <- function(n,...) contr.treatment(n,FALSE) } } if (!missing(theta) & !missing(df)) stop("Cannot give both a df and theta argument") pfun<- function(coef, theta, ndead){ if (theta==0) list(recenter=0, penalty=0, flag=TRUE) else { recenter <- mean(coef) coef <- coef - recenter list(recenter = recenter, first= coef/theta, second= rep(1, length(coef))/theta, # penalty= -sum(log(dnorm(coef,0, sqrt(theta))), penalty= 0.5* sum(coef^2/theta + log(2*pi*theta)), flag=FALSE) } } printfun <- function(coef, var, var2, df, history) { if (!is.null(history$history)) theta <- history$history[nrow(history$history),1] else theta <- history$theta if (is.matrix(var)) test <- coxph.wtest(var, coef)$test else test <- sum(coef^2/var) df2 <- max(df, .5) # Stop silly p-values list(coef=c(NA, NA, NA, test, df, 1-pchisq(test, df2)), history=paste("Variance of random effect=", format(theta))) } # The final coxph object will contain a copy of printfun. Stop it from # also containing huge unnecessary variables, e.g. 'x', known at this # point in time. Not an issue for pfun, which does not get saved. # Setting to globalenv() will not suffice since coxph.wtest is not visible # outside the survival library's name space. temp <- new.env(parent=globalenv()) assign("cox.zph", cox.zph, envir=temp) #make a private copy environment(printfun) <- temp if (method=='reml') { temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c('coef', 'trH', 'loglik'), cfun = frailty.controlgauss, cparm= list( ...)) } else if (method=='fixed') { temp <- list(pfun=pfun, printfun = printfun, diag =TRUE, sparse= sparse, cfun = function(parms, iter, old){ list(theta=parms$theta, done=TRUE)}, cparm= list(theta=theta, ...)) } else if (method=='aic') { temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c("neff", "df", "plik"), cparm=list(lower=0, init=c(.1,1), ...), cfun = frailty.controlaic) } else { #df method temp <- list(pfun=pfun, printfun =printfun, diag =TRUE, sparse= sparse, cargs=('df'), cparm=list(df=df, thetas=0, dfs=0, guess=3*df/length(unclass(x)), ...), cfun = frailty.controldf) } # If not sparse, give shorter names to the coefficients, so that any # printout of them is readable. if (!sparse) { vname <- paste("gauss", levels(x), sep=':') temp <- c(temp, list(varname=vname)) } attributes(x) <- c(attributes(x), temp) x } survival/R/survpenal.fit.S0000644000175100001440000005311112111736417015274 0ustar hornikusers# # fit a penalized parametric model # survpenal.fit<- function(x, y, weights, offset, init, controlvals, dist, scale=0, nstrat=1, strata, pcols, pattr, assign, parms=NULL) { iter.max <- controlvals$iter.max outer.max <- controlvals$outer.max eps <- controlvals$rel.tolerance toler.chol <- controlvals$toler.chol if (!is.matrix(x)) stop("Invalid X matrix ") n <- nrow(x) nvar <- ncol(x) ny <- ncol(y) if (is.null(offset)) offset <- rep(0,n) if (missing(weights)|| is.null(weights)) weights<- rep(1.0,n) else if (any(weights<=0)) stop("Invalid weights, must be >0") # The strata() term in survreg signals one scale parameter is # to be fit per strata. Here strata contains the strata level of each # subject (variable not needed for only one strata), nstrat= # of strata. # Set nstrat2 = the number of coefficients I need to fit (which is 0 # if the scale is pre-fixed). if (scale <0) stop("Invalid scale") if (scale >0 && nstrat >1) stop("Cannot have both a fixed scale and strata") if (nstrat>1 && (missing(strata) || length(strata)!= n)) stop("Invalid strata variable") if (nstrat==1) strata <- rep(1,n) if (scale >0) nstrat2 <- 0 else nstrat2 <- nstrat if (is.character(dist)) { sd <- survreg.distributions[[dist]] if (is.null(sd)) stop ("Unrecognized distribution") } else sd <- dist dnum <- match(sd$name, c("Extreme value", "Logistic", "Gaussian")) if (is.na(dnum)) { # Not one of the three distributions built in to the C code # We need to set up a callback routine # This returns the 5 number distribution summary (see the density # functions in survreg.distributions). Interval censored obs require # 2 evals and all others 1, so the call to the routine will have n2 # values. dnum <- 4 # flag for the C routine n2 <- n + sum(y[,ny]==3) # # Create an expression that will be evaluated by the C-code, # but with knowledge of some current variables # In the R doc, this would be "body(function(z) {" # in Splus (Chambers book): "functionBody(function(z)" # same action, different name. Luckily 'quote' exists in both # We make very sure the result is the right type and length here, # rather than in the C code, for simplicity. fdensity <- quote({ if (length(parms)) temp <- sd$density(z, parms) else temp <- sd$density(z) if (!is.matrix(temp) || any(dim(temp) != c(n2,5)) || !is.numeric(temp)) stop("Density function returned an invalid matrix") as.vector(as.double(temp)) }) } else { fdensity <-1 #dummy value for the .Call n2 <- n #a dummy value for inclusion in rho } # This is a subset of residuals.survreg: define the first and second # derivatives at z=0 for the 4 censoring types # Used below for starting estimates derfun <- function(y, eta, sigma, density, parms) { ny <- ncol(y) status <- y[,ny] z <- (y[,1] - eta)/sigma dmat <- density(z,parms) dtemp<- dmat[,3] * dmat[,4] #f' if (any(status==3)) { z2 <- (y[,2] - eta)/sigma dmat2 <- density(z2) } else { dmat2 <- matrix(0,1,5) #dummy values z2 <- 0 } tdenom <- ((status==0) * dmat[,2]) + ((status==1) * 1 ) + ((status==2) * dmat[,1]) + ((status==3) * ifelse(z>0, dmat[,2]-dmat2[,2], dmat2[,1] - dmat[,1])) tdenom <- 1/(tdenom* sigma) dg <- -tdenom *(((status==0) * (0-dmat[,3])) + ((status==1) * dmat[,4]) + ((status==2) * dmat[,3]) + ((status==3) * (dmat2[,3]- dmat[,3]))) ddg <- (tdenom/sigma)*(((status==0) * (0- dtemp)) + ((status==1) * dmat[,5]) + ((status==2) * dtemp) + ((status==3) * (dmat2[,3]*dmat2[,4] - dtemp))) list(dg = dg, ddg = ddg - dg^2) } status <- y[,ny] # # are there any sparse frailty terms? # npenal <- length(pattr) #total number of penalized terms if (npenal == 0 || length(pcols) != npenal) stop("Invalid pcols or pattr arg") sparse <- sapply(pattr, function(x) !is.null(x$sparse) && x$sparse) if (sum(sparse) >1) stop("Only one sparse penalty term allowed") # # Create a marking vector for the terms, the same length as assign # with pterms == 0=ordinary term, 1=penalized, 2=sparse, # pindex = length of pcols = position in pterms # # Make sure that pcols is a strict subset of assign, so that the # df computation (and printing) can unambiguously decide which cols of # X are penalized and which are not when doing "terms" like actions. # To make some downstream things easier, order pcols and pattr to be # in the same relative order as the terms in 'assign' # pterms <- rep(0, length(assign)) names(pterms) <- names(assign) pindex <- rep(0, npenal) for (i in 1:npenal) { temp <- unlist(lapply(assign, function(x,y) (length(x) == length(y) && all(x==y)), pcols[[i]])) if (sparse[i]) pterms[temp] <- 2 else pterms[temp] <- 1 pindex[i] <- (seq(along.with=temp))[temp] } if ((sum(pterms==2) != sum(sparse)) || (sum(pterms>0) != npenal)) stop("pcols and assign arguments disagree") if (any(pindex != sort(pindex))) { temp <- order(pindex) pindex <- pindex[temp] pcols <- pcols[temp] pattr <- pattr[temp] } # ptype= 1 or 3 if a sparse term exists, 2 or 3 if a non-sparse exists ptype <- any(sparse) + 2*(any(!sparse)) if (any(sparse)) { sparse.attr <- (pattr[sparse])[[1]] #can't use [[sparse]] directly # if 'sparse' is a T/F vector fcol <- unlist(pcols[sparse]) if (length(fcol) > 1) stop("Sparse term must be single column") # Remove the sparse term from the X matrix frailx <- x[, fcol] x <- x[, -fcol, drop=FALSE] for (i in 1:length(assign)){ j <- assign[[i]] if (j[1] > fcol) assign[[i]] <- j-1 } for (i in 1:npenal) { j <- pcols[[i]] if (j[1] > fcol) pcol[[i]] <- j-1 } frailx <- match(frailx, sort(unique(frailx))) nfrail <- max(frailx) nvar <- nvar - 1 #Set up the callback for the sparse frailty term # (At most one sparse term is allowed). The calling code will # first set 'coef1' to the current value of the sparse coefficients, # then call the expression below. It uses a separate context (Splus # frame or R environment), so there is no conflict between that # variable name and the rest of the code. Thus, think of the below as # a funcion of the temporary variable coef1 (current value found # in the calling C code), theta1 (current value in the S code # below, using calls to cfun), and fixed known values of pfun1 etc. # The expression will constantly replace components of "coxlist1". By # creating it first, we assure the order of the components, again # to make it simpler for the C code (it can grab the first component # and know that that is 'coef', etc). # pfun1 <- sparse.attr$pfun coxlist1 <- list(coef=0, first=0, second=0, penalty=0, flag=F) f.expr1 <- quote({ if (is.null(extra1)) temp <- pfun1(coef1, theta1, n.eff) else temp <- pfun1(coef1, theta1, n.eff, extra1) if (!is.null(temp$recenter)) coxlist1$coef <- coef1 - as.double(temp$recenter) else coxlist1$coef <- coef1 if (!temp$flag) { coxlist1$first <- -as.double(temp$first) coxlist1$second <- as.double(temp$second) } else { coxlist1$first <- double(nfrail) coxlist1$second <- double(nfrail) } coxlist1$penalty <- -as.double(temp$penalty) coxlist1$flag <- as.logical(temp$flag) # Make sure the list has exactly the right structure, so # the the C code can be simple. The first line below is # probably unnecessary (belt AND suspenders); the second is # checking a possibly user-supplied penaly function if (any(names(coxlist1) != c('coef', 'first', 'second', 'penalty', 'flag'))) stop("Invalid coxlist1") if (any(sapply(coxlist1, length) != c(rep(nfrail,3), 1, 1))) stop("Incorrect length in coxlist1") coxlist1 }) } else { # no sparse terms frailx <- 0 nfrail <- 0 f.expr1 <- NULL #dummy value pfun1 <- NULL #dummy coxlist1 <- NULL # " } nvar2 <- nvar + nstrat2 if (nvar2 ==0) { # There are no non-sparse coefficients, and no scale parameters # A strange model, leading to an hmat with 0 columns. The # underlying C code will choke, since this case is not built in. stop("Cannot fit a model with no coefficients other than sparse ones") } # Now the non-sparse penalties # There can be multiple penalized terms if (sum(!sparse) >0) { full.imat <- !all(unlist(lapply(pattr, function(x) x$diag))) ipenal <- (1:length(pattr))[!sparse] #index for non-sparse terms if (full.imat) { coxlist2 <- list(coef=double(nvar), first=double(nvar), second= double(nvar^2), penalty=0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar*nvar, 1, nvar) } else { coxlist2 <- list(coef=double(nvar), first=double(nvar), second=double(nvar), penalty= 0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar, 1, nvar) } # The C code will set the variable coef2, containing the concatonation # of all the non-sparse penalized coefs. Think of the below as # a function of coef (from the C code), thetalist (set further # below), and unchanging variables such as pattr. f.expr2 <- quote({ pentot <- 0 newcoef <- coef2 for (i in ipenal) { pen.col <- pcols[[i]] tcoef <- coef2[pen.col] if (is.null(extralist[[i]])) temp <- ((pattr[[i]])$pfun)(tcoef, thetalist[[i]], n.eff) else temp <- ((pattr[[i]])$pfun)(tcoef, thetalist[[i]], n.eff,extralist[[i]]) if (!is.null(temp$recenter)) newcoef[pen.col] <- tcoef - temp$recenter if (temp$flag) coxlist2$flag[pen.col] <- TRUE else { coxlist2$flag[pen.col] <- FALSE coxlist2$first[pen.col] <- -temp$first if (full.imat) { tmat <- matrix(coxlist2$second, nvar, nvar) tmat[pen.col,pen.col] <- temp$second coxlist2$second <- c(tmat) } else coxlist2$second[pen.col] <- temp$second } pentot <- pentot - temp$penalty } coxlist2$penalty <- as.double(pentot) coxlist2$coef <- newcoef if (any(sapply(coxlist2, length) != length2)) stop("Length error in coxlist2") coxlist2 }) } else { full.imat <- FALSE # no non-sparse penalties length2 <- 0 #dummy value f.expr2 <- NULL coxlist2 <- NULL ipenal <- NULL } # Create the frame for penalized evaluation # In R new.env inherits everything, in Splus new.frame only has # what I specify at this time. The variable thetalist will # be iterated below, so we need to remember to update it within # the Splus rho each time we do! # The variables parms, sd, and n2 are used for fdensity evaluation # if (is.R()) rho <- new.env() #Splus else rho <- new.frame(list(pfun1=pfun1, theta1=NULL, extra1=NULL, # nfrail=nfrail, pcols=pcols, pattr=pattr, # length2=length2, full.imat=full.imat, # ipenal = ipenal, nvar=nvar, # coxlist1=coxlist1, coxlist2=coxlist2, # sd=sd, parms=parms, n2=n2)) # # "Unpack" the passed in paramter list, # and make the initial call to each of the external routines # cfun <- lapply(pattr, function(x) x$cfun) parmlist <- lapply(pattr, function(x,eps) c(x$cparm, eps2=eps), sqrt(eps)) extralist<- lapply(pattr, function(x) x$pparm) iterlist <- vector('list', length(cfun)) thetalist <- vector('list', length(cfun)) printfun <- lapply(pattr, function(x) x$printfun) extra1 <- NULL theta1 <- NULL for (i in 1:length(cfun)) { temp <- (cfun[[i]])(parmlist[[i]], iter=0) if (sparse[i]) { assign('theta1', temp$theta, rho) assign('extra1', extralist[[i]], rho) } thetalist[[i]] <- temp$theta iterlist[[i]] <- temp } # if (!is.R()) { # Splus support # assign('thetalist', thetalist, frame=rho) # assign('extralist', extralist, frame=rho) # } # # Manufacture the list of calls to cfun, with appropriate arguments # temp1 <- c('x', 'coef', 'plik', 'loglik', 'status', 'neff', 'df', 'trH') temp2 <- c('frailx', 'fcoef', 'fit$loglik-fit$penalty', 'fit$loglik', 'status', 'n.eff') temp3 <- c('x[,pen.col]', 'coef[pen.col]', 'fit$loglik-fit$penalty', 'fit$loglik', 'status', 'n.eff') calls <- vector('expression', length(cfun)) cargs <- lapply(pattr, function(x) x$cargs) for (i in 1:length(cfun)) { tempchar <- paste("(cfun[[", i, "]])(parmlist[[", i, "]], iter,", "iterlist[[", i, "]]") temp2b <- c(temp2, paste('pdf[', i, ']'), paste('trH[', i, ']')) temp3b <- c(temp3, paste('pdf[', i, ']'), paste('trH[', i, ']')) if (length(cargs[[i]])==0) calls[i] <- parse(text=paste(tempchar, ")")) else { temp <- match(cargs[[i]], temp1) if (any(is.na(temp))) stop(paste((cargs[[i]])[is.na(temp)], "not matched")) if (sparse[i]) temp4 <- paste(temp2b[temp], collapse=',') else temp4 <- paste(temp3b[temp], collapse=',') calls[i] <- parse(text=paste(paste(tempchar,temp4,sep=','),')')) } } need.df <- any(!is.na(match(c('df', 'trH'), unlist(cargs))))#do any use df? # # Last of the setup: create the vector of variable names # varnames <- dimnames(x)[[2]] for (i in 1:npenal) { if (!is.null(pattr[[i]]$varname)) varnames[pcols[[i]]] <- pattr[[i]]$varname } nvar2 <- nvar + nstrat2 nvar3 <- nvar2 + nfrail # # A good initial value of the scale turns out to be critical for successful # iteration, in a surprisingly large number of data sets. # The best way we've found to get one is to fit a model with only the # mean and the scale. We also the loglik of the mean-only model in the # result # Even this model needs starting guesses... yy <- ifelse(status !=3, y[,1], (y[,1]+y[,2])/2 ) coef <- sd$init(yy, weights,parms) # We sometimes get into trouble with a small initial estimate of sigma, # (the surface isn't SPD), but never with a large one. Double it. if (scale >0) vars <- log(scale) else vars <- log(4*coef[2])/2 # init gives \sigma^2, I need log(sigma) coef <- c(coef[1], rep(vars, nstrat)) # get a better initial value for the mean using the "glim" trick deriv <- derfun(y, yy, exp(vars), sd$density, parms) wt <- -1*deriv$ddg*weights coef[1] <- sum(weights*deriv$dg + wt*(yy -offset)) / sum(wt) fit0 <- .Call(Csurvreg6, iter = as.integer(20), nvar = as.integer(1), as.double(y), as.integer(ny), x = as.double(rep(1.0, n)), as.double(weights), as.double(offset), coef= as.double(coef), as.integer(nstrat2), as.integer(strata), as.double(eps), as.double(toler.chol), as.integer(dnum), fdensity, rho) # The "effective n" of the model temp <- mean(exp(fit0$coef[-1])) #overall sd n.eff <- sd$variance(temp^2) * (solve(matrix(fit0$var,1+nstrat2)))[1,1] #if (!is.R()) assign('n.eff', n.eff, frame=rho) # # Fit the model with all covariates # Start with initial values # if (is.numeric(init)) { if (length(init) == nvar) { if (scale >0) init <- c(init, log(scale)) else init <-c(rep(0, nfrail), init, fit0$coef[-1]) } else if (length(init) == nvar2) init <- c(rep(0,nfrail), init) else if (length(init) != nvar3) stop("Wrong length for inital values") if (scale >0) init <- c(init, log(scale)) } else { # The algebra behind the 'glim' trick just doesn't work here # Use the intercept fit + zeros # coef order = frailty, intercept, other covariates, sigmas init <- c(rep(0, nfrail), fit0$coef[1], rep(0, nvar-1), fit0$coef[-1]) } # # Tack on the sigmas to "assign", so that the df component includes # the sigmas if (nstrat2 >0) assign <- c(assign, list(sigma=(1+nvar):nvar2)) iter2 <- 0 iterfail <- NULL thetasave <- unlist(thetalist) for (iterx in 1:outer.max) { fit <- .Call(Csurvreg7, iter = as.integer(iter.max), as.integer(nvar), as.double(y), as.integer(ny), as.double(x), as.double(weights), as.double(offset), coef= as.double(init), as.integer(nstrat2), as.integer(strata), as.double(eps), as.double(toler.chol), as.integer(dnum), fdensity, rho, as.integer(ptype), as.integer(full.imat), as.integer(nfrail), as.integer(frailx), f.expr1, f.expr2) iter <- iterx iter2 <- iter2 + fit$iter if (fit$flag == 1000) iterfail <- c(iterfail, iter) if (nfrail >0) { fcoef <- fit$coef[1:nfrail] coef <- fit$coef[nfrail + 1:nvar2] } else coef <- fit$coef[1:nvar2] # We need to fetch back some of the results from the # evaluation area of f.expr1 and f.expr2 if (is.R()) { if (nfrail >0) coxlist1 <- get('coxlist1', envir=rho) if (ptype >1 ) coxlist2 <- get('coxlist2', envir=rho) } #else { # if (nfrail >0) coxlist1 <- get('coxlist1', frame=rho) # if (ptype >1 ) coxlist2 <- get('coxlist2', frame=rho) # } # If any penalties were infinite, the C code has made hdiag=1 out # of self-preservation (avoid zero divides). But such coefs are # guarranteed to be zero so the variance should be too. temp <- rep(FALSE, nvar2+nfrail) if (nfrail>0) temp[1:nfrail] <- coxlist1$flag if (ptype >1) temp[nfrail+ 1:nvar] <- coxlist2$flag hdiag <- ifelse(temp, 0, fit$hdiag) if (need.df) { #get the penalty portion of the second derive matrix if (nfrail>0) temp1 <- coxlist1$second else temp1 <- 0 if (ptype>1) { if (full.imat) { temp2 <- matrix(0., nvar2, nvar2) temp2[1:nvar, 1:nvar] <- coxlist2$second } else temp2 <- diag(c(coxlist2$second, rep(0, nstrat2))) } else temp2 <- 0 dftemp <-coxpenal.df(matrix(fit$hmat, ncol=nvar2), matrix(fit$hinv, ncol=nvar2), hdiag, assign, ptype, nvar2, temp1, temp2, pindex[sparse]) df <- dftemp$df var <- dftemp$var var2 <- dftemp$var2 pdf <- df[pterms>0] # df's for penalized terms trH <- dftemp$trH[pterms>0] # trace H } # # Call the control function(s) # done <- TRUE for (i in 1:length(cfun)) { pen.col <- pcols[[i]] temp <- eval(calls[i]) if (sparse[i]) assign('theta1', temp$theta, rho) thetalist[[i]] <- temp$theta iterlist[[i]] <- temp done <- done & temp$done } if (done) break #if (!is.R()) assign('thetalist', thetalist, frame=rho) # # Choose starting estimates for the next iteration # if (iter==1) { init <- coefsave <- fit$coef thetasave <- cbind(thetasave, unlist(thetalist)) } else { temp <- unlist(thetalist) coefsave <- cbind(coefsave, fit$coef) # temp = next guess for theta # *save = prior thetas and the resultant fits # choose as initial values the result for the closest old theta howclose <- apply((thetasave-temp)^2,2, sum) which <- min((1:iter)[howclose==min(howclose)]) init <- coefsave[,which] thetasave <- cbind(thetasave, temp) } } #end of the iteration loop if (!need.df) { #didn't need it iteration by iteration, but do it now #get the penalty portion of the second derive matrix if (nfrail>0) temp1 <- coxlist1$second else temp1 <- 0 if (ptype>1) { if (full.imat) { temp2 <- matrix(0., nvar2, nvar2) temp2[1:nvar, 1:nvar] <- coxlist2$second } else temp2 <- diag(c(coxlist2$second, rep(0, nstrat2))) } else temp2 <- 0 dftemp <-coxpenal.df(matrix(fit$hmat,ncol=nvar2), matrix(fit$hinv,ncol=nvar2), hdiag, assign, ptype, nvar2, temp1, temp2, pindex[sparse]) df <- dftemp$df trH <- dftemp$trH var <- dftemp$var var2 <- dftemp$var2 } if (iter.max >1 && length(iterfail)>0) warning(paste("Inner loop failed to coverge for iterations", paste(iterfail, collapse=' '))) which.sing <- (hdiag[nfrail + 1:nvar] ==0) coef[which.sing] <- NA names(iterlist) <- names(pterms[pterms>0]) cname <- varnames cname <- c(cname, rep("Log(scale)", nstrat2)) dimnames(var) <- list(cname, cname) names(coef) <- cname if (nfrail >0) { lp <- offset + fcoef[frailx] lp <- lp + x %*%coef[1:nvar] list(coefficients = coef, icoef = fit0$coef, var = var, var2 = var2, loglik = c(fit0$loglik, fit$loglik- fit$penalty), iter = c(iter, iter2), linear.predictors = as.vector(lp), frail = fcoef, fvar = dftemp$fvar, df = df, penalty= c(fit0$penalty, -fit$penalty), pterms = pterms, assign2=assign, history= iterlist, printfun=printfun, score = fit$u) } else { #no sparse terms list(coefficients = coef, icoef = fit0$coef, var = var, var2 = var2, loglik = c(fit0$loglik, fit$loglik- fit$penalty), iter = c(iter, iter2), linear.predictors = as.vector(x%*%coef[1:nvar]), df = df, df2=dftemp$df2, penalty= c(0, -fit$penalty), pterms = pterms, assign2=assign, history= iterlist, printfun= printfun, score = fit$u) } } survival/R/tcut.S0000644000175100001440000000252712265342374013465 0ustar hornikuserstcut <- function (x, breaks, labels, scale=1){ # avoid some problems with dates x <- as.numeric(x) breaks <- as.numeric(breaks) if(length(breaks) == 1) { if(breaks < 1) stop("Must specify at least one interval") if(missing(labels)) labels <- paste("Range", seq(length = breaks)) else if(length(labels) != breaks) stop("Number of labels must equal number of intervals") r <- range(x[!is.na(x)]) r[is.na(r)] <- 1 if((d <- diff(r)) == 0) { r[2] <- r[1] + 1 d <- 1 } breaks <- seq(r[1] - 0.01 * d, r[2] + 0.01 * d, length = breaks +1) } else { if(is.na(adb <- all(diff(breaks) >= 0)) || !adb) stop("breaks must be given in ascending order and contain no NA's") if(missing(labels)) labels <- paste(format(breaks[ - length(breaks)]), "+ thru ", format(breaks[-1]), sep = "") else if(length(labels) != length(breaks) - 1) stop("Number of labels must be 1 less than number of break points") } temp <- structure(x*scale, cutpoints=breaks*scale, labels=labels) if (is.R()) class(temp) <- 'tcut' else oldClass(temp) <- 'tcut' temp } "[.tcut" <- function(x, ..., drop=FALSE) { atts <- attributes(x) x <- unclass(x)[..1] attributes(x) <- atts if (is.R()) class(x) <- 'tcut' else oldClass(x) <- 'tcut' x } levels.tcut <- function(x) attr(x, 'labels') survival/R/print.coxph.S0000644000175100001440000000306212536375266014764 0ustar hornikusersprint.coxph <- function(x, digits=max(options()$digits - 4, 3), ...) { if (!is.null(cl<- x$call)) { cat("Call:\n") dput(cl) cat("\n") } if (!is.null(x$fail)) { cat(" Coxph failed.", x$fail, "\n") return() } savedig <- options(digits = digits) on.exit(options(savedig)) coef <- x$coefficients se <- sqrt(diag(x$var)) if(is.null(coef) | is.null(se)) stop("Input is not valid") if (is.null(x$naive.var)) { tmp <- cbind(coef, exp(coef), se, coef/se, # signif(1 - pchisq((coef/ se)^2, 1), digits -1)) 1- pchisq((coef/se)^2, 1)) dimnames(tmp) <- list(names(coef), c("coef", "exp(coef)", "se(coef)", "z", "p")) } else { nse <- sqrt(diag(x$naive.var)) tmp <- cbind(coef, exp(coef), nse, se, coef/se, # signif(1 - pchisq((coef/se)^2, 1), digits -1)) 1 - pchisq((coef/se)^2, 1)) dimnames(tmp) <- list(names(coef), c("coef", "exp(coef)", "se(coef)", "robust se", "z", "p")) } cat("\n") # print(tmp) printCoefmat(tmp, signif.stars=FALSE, P.values=TRUE, has.Pvalue=TRUE) logtest <- -2 * (x$loglik[1] - x$loglik[2]) if (is.null(x$df)) df <- sum(!is.na(coef)) else df <- round(sum(x$df),2) cat("\n") cat("Likelihood ratio test=", format(round(logtest, 2)), " on ", df, " df,", " p=", format(1 - pchisq(logtest, df)), "\n", sep="") omit <- x$na.action cat("n=", x$n) if (!is.null(x$nevent)) cat(", number of events=", x$nevent, "\n") else cat("\n") if (length(omit)) cat("\ (", naprint(omit), ")\n", sep="") invisible(x) } survival/R/survSplit.R0000644000175100001440000000163012470201064014476 0ustar hornikuserssurvSplit<-function(data, cut, end,event,start,id=NULL, zero=0,episode=NULL){ cut<-sort(cut) ntimes <- length(cut) n <- nrow(data) newdata <- lapply(data,rep,ntimes+1) endtime <- rep(c(cut, Inf) ,each=n) eventtime<-newdata[[end]] if( start %in% names(data)) starttime<-data[[start]] else starttime<-rep(zero,length.out=n) starttime<-c(starttime, pmax(starttime, rep(cut,each=n))) epi<-rep(0:ntimes,each=n) status <- ifelse( eventtime <= endtime & eventtime>starttime, newdata[[event]], 0) endtime<- pmin(endtime,eventtime) drop<-starttime>=endtime newdata<-do.call("data.frame",newdata) newdata[,start]<-starttime newdata[,end]<-endtime newdata[,event]<-status if (!is.null(id)) newdata[,id]<-rep(rownames(data),ntimes+1) if (!is.null(episode)) newdata[,episode]<-epi newdata<-newdata[!drop,] newdata } survival/R/print.aareg.S0000644000175100001440000000200411732700061014673 0ustar hornikusers# $Id: print.aareg.S 11250 2009-03-19 13:44:59Z tlumley $ print.aareg <- function(x, maxtime, test=c('aalen', 'nrisk'), scale=1, ...) { if (!inherits(x, 'aareg')) stop ("Must be an addreg object") if (!is.null(cl<- x$call)) { cat("Call:\n") dput(cl) cat("\n") } if (missing(test)) test <- x$test else test <- match.arg(test) if (missing(maxtime)) summ <- summary(x, test=test, scale=scale) else summ <- summary(x, maxtime=maxtime, test=test, scale=scale) omit <- x$na.action if (length(omit)) cat(" n=", x$n[1], " (", naprint(omit), ")\n", sep="") else cat(" n=", x$n[1], "\n") cat(" ", summ$n[2], "out of", x$n[3], "unique event times used\n\n") print(signif(summ$table,3)) chi <- summ$chisq df <- nrow(summ$table) -1 cat("\nChisq=", format(round(chi,2)), " on ", df, " df, p=", signif(1- pchisq(chi,df),2), "; test weights=", x$test, "\n", sep="") invisible(x) } survival/R/plot.aareg.S0000644000175100001440000000453711732700061014532 0ustar hornikusers# $Id: plot.aareg.S 11166 2008-11-24 22:10:34Z therneau $ plot.aareg <- function(x, se=TRUE, maxtime, type='s', ...) { if (!inherits(x, 'aareg')) stop ("Must be an aareg object") if (missing(maxtime)) keep <- 1:length(x$time) else keep <- 1:sum(x$time <= maxtime) yylab <- names(x$test.statistic) if (is.matrix(x$coefficient) && ncol(x$coefficient)>1) { yy <- apply(x$coefficient[keep,], 2,cumsum) yy <- rbind(0,yy) # make the plot start at 0,0 if (se) { if (!is.null(x$dfbeta)) { # There was a cluster term, so use the robust variance # dfbeta will be of dimension (n, nvar, n-unique-times) # The first variance increment is apply(dfbeta[,,1]^2,2,sum) # second is apply(dfbeta[,,2]^2,2,sum) # ... , apply(dfbeta[,,ndeath]..... # By being sneaky, it can be done quickly dd <- dim(x$dfbeta) keep2 <- 1:length(unique(x$time[keep])) temp <- matrix(x$dfbeta[,,keep2], nrow=dd[1]) se.increment <- matrix(apply(temp^2, 2, sum), nrow=dd[2]) se.yy <- sqrt(apply(t(se.increment), 2, cumsum)) } else se.yy <- sqrt(apply(x$coefficient[keep,]^2, 2,cumsum)) se.yy <- rbind(0, se.yy) } ncurve <- ncol(yy) } else { # this is the branch most often called, when someone has done # plot(fit[3]), so that only 1 coefficient remains yy <- cumsum(c(0, x$coefficient[keep])) if (se) { if (!is.null(x$dfbeta)) { dd <- dim(x$dfbeta) keep2 <- 1:length(unique(x$time[keep])) temp <- matrix(x$dfbeta[,,keep2], nrow=dd[1]) se.yy <- sqrt(cumsum(c(0, apply(temp^2, 2, sum)))) } else se.yy <- sqrt(cumsum(c(0, x$coefficient[keep]^2))) } ncurve <- 1 } xx <- c(0, x$time[keep]) # There may be multiplicities in x$times. Only plot the last of # each of them indx <- 1 + length(xx) - rev(match(unique(rev(xx)), rev(xx))) xx <- xx[indx] yy <- as.matrix(yy)[indx,] if (se) { if (is.null(x$dfbeta)) se.yy<- as.matrix(se.yy)[indx,] yy <- cbind(yy, yy + 1.96*se.yy, yy - 1.96*se.yy) if (ncurve >1) { for (i in 1:ncurve) { j <- c(i, i+ncurve, i+2*ncurve) matplot(xx, yy[,j], type=type, ..., col=1, lty=c(1,2,2), xlab='Time', ylab=yylab[i]) } } else matplot(xx, yy, type=type, ..., col=1, lty=c(1,2,2), xlab='Time', ylab=yylab) } else { matplot(xx, yy, type=type, ..., xlab='Time') } } survival/R/print.survexp.S0000644000175100001440000000366512464750557015370 0ustar hornikusersprint.survexp <- function(x, scale=1, digits = max(options()$digits - 4, 3), naprint=FALSE, ...) { if (!inherits(x, 'survexp')) stop("Invalid data") savedig <- options(digits=digits) on.exit(options(savedig)) if (!is.null(cl<- x$call)) { cat("Call:\n") dput(cl) cat("\n") } if (!is.null(x$summ)) cat(x$summ) omit <- x$na.action if (length(omit)) cat(naprint(omit), "\n") else cat("\n") if (is.null(x$strata)) { #print it as a matrix mat <- cbind(x$time/scale, x$n.risk, x$surv, x$std.err) if (!naprint) { miss <- (is.na(mat)) %*% rep(1,ncol(mat)) mat <- mat[miss<(ncol(mat)-2),,drop=FALSE] } if (is.matrix(x$surv)) cname <- dimnames(x$surv)[[2]] else cname <- "survival" if (is.matrix(x$n.risk)) cname <- c(paste("nrisk", 1:ncol(x$n.risk), sep=''), cname) else cname <- c("n.risk", cname) cname <- c("time", cname) if (!is.null(x$std.err)) cname <- c(cname, paste("se(", cname, ")", sep='')) dimnames(mat) <- list(rep("", nrow(mat)), cname) print(mat) } else { #print it out one strata at a time, since n's differ if (is.null(x$std.err)) tname <- 'survival' else tname <- c('survival', 'se(surv)') nstrat <- length(x$strata) levs <- names(x$strata) if (nrow(x$surv)==1) { mat <- cbind(c(x$n.risk), c(x$surv), c(x$std.err*x$surv)) dimnames(mat) <- list(levs, c("n.risk", tname)) cat(" Survival at time", x$time, "\n") print(mat) } else { for (i in 1:nstrat) { cat(" ", levs[i], "\n") mat <- cbind(x$time/scale, x$n.risk[,i], x$surv[,i]) if (!is.null(x$std.err)) mat<- cbind(mat, x$std.err[,i] * x$surv[,i]) if (!naprint) mat <- mat[!is.na(mat[,3]),,drop=FALSE] dimnames(mat) <- list(rep("",nrow(mat)), c("Time", "n.risk", tname)) print(mat) cat("\n") } } } invisible(x) } survival/R/cox.zph.S0000644000175100001440000000501511732700061014057 0ustar hornikusers# $Id: cox.zph.S 11218 2009-02-09 12:09:29Z therneau $ # Test proportional hazards # cox.zph <- function(fit, transform='km', global=TRUE) { call <- match.call() if (!inherits(fit, 'coxph')) stop ("Argument must be the result of coxph") if (inherits(fit, 'coxph.null')) stop("The are no score residuals for a Null model") sresid <- resid(fit, 'schoenfeld') varnames <- names(fit$coefficients) nvar <- length(varnames) ndead<- length(sresid)/nvar if (nvar==1) times <- as.numeric(names(sresid)) else times <- as.numeric(dimnames(sresid)[[1]]) # Next line is no longer necessary: survfit.km can handle (start,stop] data # if (missing(transform) && attr(fit$y, 'type') != 'right') # transform <- 'identity' if (is.character(transform)) { tname <- transform ttimes <- switch(transform, 'identity'= times, 'rank' = rank(times), 'log' = log(times), 'km' = { temp <- survfitKM(factor(rep(1,nrow(fit$y))), fit$y, se.fit=FALSE) # A nuisance to do left cont KM t1 <- temp$surv[temp$n.event>0] t2 <- temp$n.event[temp$n.event>0] km <- rep(c(1,t1), c(t2,0)) if (is.null(attr(sresid, 'strata'))) 1-km else (1- km[sort.list(sort.list(times))]) }, stop("Unrecognized transform")) } else { tname <- deparse(substitute(transform)) if (length(tname) >1) tname <- 'user' ttimes <- transform(times) } xx <- ttimes - mean(ttimes) r2 <- sresid %*% fit$var * ndead test <- xx %*% r2 # time weighted col sums corel <- c(cor(xx, r2)) z <- c(test^2 /(diag(fit$var)*ndead* sum(xx^2))) Z.ph <- cbind(corel, z, 1- pchisq(z,1)) if (global && nvar>1) { test <- c(xx %*% sresid) z <- c(test %*% fit$var %*% test) * ndead / sum(xx^2) Z.ph <- rbind(Z.ph, c(NA, z, 1-pchisq(z, ncol(sresid)))) dimnames(Z.ph) <- list(c(varnames, "GLOBAL"), c("rho", "chisq", "p")) } else dimnames(Z.ph) <- list(varnames, c("rho", "chisq", "p")) dimnames(r2) <- list(times, names(fit$coefficients)) temp <-list(table=Z.ph, x=ttimes, y=r2 + outer(rep(1,ndead), fit$coefficients), var=fit$var, call=call, transform=tname) if (is.R()) class(temp) <- "cox.zph" else oldClass(temp) <- "cox.zph" temp } "[.cox.zph" <- function(x, ..., drop=FALSE) { i <- ..1 z<- list(table=x$table[i,,drop=FALSE], x=x$x, y=x$y[ ,i,drop=FALSE], var=x$var[i,i, drop=FALSE], call=x$call, transform=x$transform) attributes(z) <- attributes(x) z } survival/R/survregDtest.S0000644000175100001440000000511611732700061015171 0ustar hornikusers# $Id$ # # Test out if a distribution object found in survreg is legal. Mostly called # by the survreg routine, but a user might use it when developing a new # distribution object # # Short form, returns just T or F # Long form, returns all of the issues with the object, or T if it is ok # survregDtest <- function(dlist, verbose=F) { errlist <- NULL if (is.null(dlist$name)) errlist <- c(errlist, "Missing a name") else if (length(dlist$name) !=1 || !is.character(dlist$name)) errlist <- c(errlist, "Invalid name") # # First case, the object is a reference to another distribution # if (!is.null(dlist$dist)) { if (!is.character(dlist$dist) || is.null(match(dlist$dist, names(survreg.distributions)))) errlist <- c(errlist, "Reference distribution not found") else { if (!is.function(dlist$trans)) errlist <- c(errlist, "Missing or invalid trans component") if (!is.function(dlist$itrans)) errlist <- c(errlist, "Missing or invalid itrans component") if (!is.function(dlist$dtrans)) errlist <- c(errlist, "Missing or invalid dtrans component") } if (is.null(errlist)) { if (!all.equal(dlist$itrans(dlist$trans(1:10)), 1:10)) errlist <- c(errlist, "trans and itrans must be inverses of each other") if (length(dlist$dtrans(1:10)) != 10) errlist <- c(errlist, "dtrans must be a 1-1 function") } } # Second case, the actual definition of a distribution else { # Comment out the next line, until some function uses the variance #if (!is.function(dlist$variance)) # errlist <- c(errlist, "Missing or invalid variance function") if (!is.function(dlist$init)) errlist <- c(errlist, "Missing or invalid init function") if (!is.function(dlist$deviance)) errlist <- c(errlist, "Missing or invalid deviance function") if (!is.function(dlist$density)) errlist <- c(errlist, "Missing or invalid density function") else { if (is.null(dlist$parms)) temp <- dlist$density(1:10/10) else temp <- dlist$density(1:10/10, unlist(dlist$parms)) if (!is.numeric(temp) || !is.matrix(temp) || nrow(temp) != 10 || ncol(temp) != 5) errlist <- c(errlist, "Density function must return a 5 column matrix") } if (!is.function(dlist$quantile)) errlist <- c(errlist, "Missing or invalid quantile function") } if (is.null(errlist)) T else if (verbose) errlist else F } survival/R/survdiff.S0000644000175100001440000000540411732700061014320 0ustar hornikuserssurvdiff <- function(formula, data, subset, na.action, rho=0) { call <- match.call() m <- match.call(expand.dots=FALSE) m$rho <- NULL if (!inherits(formula, 'formula')) stop("The 'formula' argument is not a formula") Terms <- if(missing(data)) terms(formula, 'strata') else terms(formula, 'strata', data=data) m$formula <- Terms m[[1]] <- as.name("model.frame") if (is.R()) m <- eval(m, parent.frame()) else m <- eval(m, sys.parent()) y <- model.extract(m, "response") if (!inherits(y, "Surv")) stop("Response must be a survival object") if (attr(y, 'type') != 'right') stop("Right censored data only") ny <- ncol(y) n <- nrow(y) offset<- attr(Terms, "offset") if (!is.null(offset)) { #one sample test offset <- as.numeric(m[[offset]]) if (length(attr(Terms,"factors"))>0) stop("Cannot have both an offset and groups") if (any(offset <0 | offset >1)) stop("The offset must be a survival probability") expected <- sum(-log(offset)) #sum of expected events observed <- sum(y[,ny]) if (rho!=0) { num <- sum(1/rho - ((1/rho + y[,ny])*offset^rho)) var <- sum(1- offset^(2*rho))/(2*rho) } else { var <- sum(-log(offset)) num <- var - observed } chi <- num*num/var rval <-list(n= n, obs = observed, exp=expected, var=var, chisq= chi) } else { #k sample test strats <- attr(Terms, "specials")$strata if (length(strats)) { temp <- untangle.specials(Terms, 'strata', 1) dropx <- temp$terms if (length(temp$vars)==1) strata.keep <- m[[temp$vars]] else strata.keep <- strata(m[,temp$vars], shortlabel=TRUE) } else strata.keep <- rep(1,nrow(m)) #Now create the group variable if (length(strats)) ll <- attr(Terms[-dropx], 'term.labels') else ll <- attr(Terms, 'term.labels') if (length(ll) == 0) stop("No groups to test") else groups <- strata(m[ll]) fit <- survdiff.fit(y, groups, strata.keep, rho) if (is.matrix(fit$observed)){ otmp <- apply(fit$observed,1,sum) etmp <- apply(fit$expected,1,sum) } else { otmp <- fit$observed etmp <- fit$expected } df <- (etmp >0) #remove groups with exp=0 if (sum(df) <2) chi <- 0 # No test, actually else { temp2 <- ((otmp - etmp)[df])[-1] vv <- (fit$var[df,df])[-1,-1, drop=FALSE] chi <- sum(solve(vv, temp2) * temp2) } rval <-list(n= table(groups), obs = fit$observed, exp = fit$expected, var=fit$var, chisq=chi) if (length(strats)) rval$strata <- table(strata.keep) } na.action <- attr(m, "na.action") if (length(na.action)) rval$na.action <- na.action rval$call <- call if (is.R()) class(rval) <- 'survdiff' else oldClass(rval) <- 'survdiff' rval } survival/R/coxph.fit.S0000644000175100001440000000726312377170071014407 0ustar hornikuserscoxph.fit <- function(x, y, strata, offset, init, control, weights, method, rownames) { n <- nrow(y) if (is.matrix(x)) nvar <- ncol(x) else { if (length(x)==0) nvar <-0 else nvar <-1 } time <- y[,1] status <- y[,2] # Sort the data (or rather, get a list of sorted indices) if (length(strata)==0) { sorted <- order(time) strata <- NULL newstrat <- as.integer(rep(0,n)) } else { sorted <- order(strata, time) strata <- strata[sorted] newstrat <- as.integer(c(1*(diff(as.numeric(strata))!=0), 1)) } if (missing(offset) || is.null(offset)) offset <- rep(0,n) if (missing(weights)|| is.null(weights))weights<- rep(1,n) else { if (any(weights<=0)) stop("Invalid weights, must be >0") weights <- weights[sorted] } stime <- as.double(time[sorted]) sstat <- as.integer(status[sorted]) if (nvar==0) { # A special case: Null model. # (This is why I need the rownames arg- can't use x' names) # Set things up for 0 iterations on a dummy variable x <- as.matrix(rep(1.0, n)) nullmodel <- TRUE nvar <- 1 init <- 0 maxiter <- 0 } else { nullmodel <- FALSE maxiter <- control$iter.max if (!missing(init) && length(init)>0) { if (length(init) != nvar) stop("Wrong length for inital values") } else init <- rep(0,nvar) } storage.mode(weights) <- storage.mode(init) <- "double" coxfit <- .Call(Ccoxfit6, as.integer(maxiter), stime, sstat, x[sorted,], as.double(offset[sorted]), weights, newstrat, as.integer(method=="efron"), as.double(control$eps), as.double(control$toler.chol), as.vector(init), as.integer(1)) # internally rescale if (nullmodel) { score <- exp(offset[sorted]) coxres <- .C(Ccoxmart, as.integer(n), as.integer(method=='efron'), stime, sstat, newstrat, as.double(score), as.double(weights), resid=double(n)) resid <- double(n) resid[sorted] <- coxres$resid names(resid) <- rownames list( loglik = coxfit$loglik[1], linear.predictors = offset, residuals = resid, method= c('coxph.null', 'coxph') ) } else { var <- matrix(coxfit$imat,nvar,nvar) coef <- coxfit$coef if (coxfit$flag < nvar) which.sing <- diag(var)==0 else which.sing <- rep(FALSE,nvar) infs <- abs(coxfit$u %*% var) if (maxiter >1) { if (coxfit$flag == 1000) warning("Ran out of iterations and did not converge") else { infs <- ((infs > control$eps) & infs > control$toler.inf*abs(coef)) if (any(infs)) warning(paste("Loglik converged before variable ", paste((1:nvar)[infs],collapse=","), "; beta may be infinite. ")) } } names(coef) <- dimnames(x)[[2]] lp <- c(x %*% coef) + offset - sum(coef*coxfit$means) score <- exp(lp[sorted]) coxres <- .C(Ccoxmart, as.integer(n), as.integer(method=='efron'), stime, sstat, newstrat, as.double(score), as.double(weights), resid=double(n)) resid <- double(n) resid[sorted] <- coxres$resid names(resid) <- rownames if (maxiter > 0) coef[which.sing] <- NA #leave it be if iter=0 is set concordance <- survConcordance.fit(Surv(stime, sstat), lp[sorted], strata, weights) list(coefficients = coef, var = var, loglik = coxfit$loglik, score = coxfit$sctest, iter = coxfit$iter, linear.predictors = as.vector(lp), residuals = resid, means = coxfit$means, concordance=concordance, method='coxph') } } survival/R/coxph.control.S0000644000175100001440000000117711732700061015273 0ustar hornikusers# # Gather all of the control parameters for coxph into one spot # coxph.control <- function(eps=1e-9, toler.chol = .Machine$double.eps ^ .75, iter.max=20, toler.inf= sqrt(eps), outer.max=10 ) { if (iter.max <0) stop("Invalid value for iterations") if (eps <=0) stop ("Invalid convergence criteria") if (eps <= toler.chol) warning("For numerical accuracy, tolerance should be < eps") if (toler.inf <=0) stop ("The inf.warn setting must be >0") list(eps=eps, toler.chol=toler.chol, iter.max=as.integer(iter.max), toler.inf=toler.inf, outer.max=as.integer(outer.max)) } survival/R/frailty.controlgam.S0000644000175100001440000000427611732700061016314 0ustar hornikusers# $Id: frailty.controlgam.S 11166 2008-11-24 22:10:34Z therneau $ # # The control function for a single Gamma frailty term. # frailty.controlgam <- function(opt, iter, old, group, status, loglik){ if (iter==0) { # initial call if (!is.null(opt$theta)) theta <- opt$theta #fixed theta case else { if (is.null(opt$init)) theta <- 0 #no initial value -- use 0 else theta <- opt$init[1] } list(theta=theta) } else { if (is.null(opt$trace)) trace <-FALSE else trace <- opt$trace theta <- old$theta #compute correction to the loglik if (theta==0) correct <- 0 else { if (is.matrix(group)) group <- c(group %*% 1:ncol(group)) d <- tapply(status,group,sum) correct <- frailty.gammacon(d, 1/theta) } if (!is.null(opt$theta)) # fixed theta case list(theta=theta, done=TRUE, c.loglik=loglik + correct) else { # save history of the iteration, and get the next theta if (iter==1) history <- c(theta=theta, loglik=loglik, c.loglik=loglik + correct) else history <- rbind(old$history, as.vector(c(theta, loglik, loglik + correct))) if (iter==1) { if (is.null(opt$init )) theta <-1 else theta <- opt$init[2] list(theta=theta, done=FALSE, history=history, c.loglik= loglik+correct) } else if (iter ==2) { if (history[2,3] < (history[1,3] +1)) theta <- mean(history[1:2,1]) else theta <- 2*history[2,1] if (trace) { print(history) cat(" new theta=", theta, "\n\n") } list(theta=theta, done=FALSE, history=history, c.loglik= loglik+correct) } else { #Now, history has iter rows, each row contains the value # of theta, the Cox PL, and the full LL done <- (abs(1- history[iter,3]/history[iter-1,3]) < opt$eps) x <- history[,1] y <- history[,3] if (y[iter]== max(y) && x[iter]==max(x)) newtheta <- 2* max(x) else newtheta <- frailty.brent(sqrt(x), y, lower=0)^2 if (trace) { print(history) cat(" new theta=", format(newtheta), "\n\n") } list(theta=newtheta, done=done, history=history, c.loglik = loglik + correct) } } } } survival/R/anova.survreglist.S0000644000175100001440000000420111732700061016156 0ustar hornikusersanova.survreglist <- function(object, ..., test = c("Chisq", "none")) { diff.term <- function(term.labels, i) { t1 <- term.labels[[1]] t2 <- term.labels[[2]] m1 <- match(t1, t2, FALSE) m2 <- match(t2, t1, FALSE) if(all(m1)) { if(all(m2)) return("=") else return(paste(c("", t2[ - m1]), collapse = "+")) } else { if(all(m2)) return(paste(c("", t1[ - m2]), collapse = "-")) else return(paste(i - 1, i, sep = " vs. ")) } } test <- match.arg(test) rt <- length(object) if(rt == 1) { object <- object[[1]] UseMethod("anova") } forms <- sapply(object, function(x) as.character(formula(x))) subs <- as.logical(match(forms[2, ], forms[2, 1], FALSE)) if(!all(subs)) warning("Some fit objects deleted because response differs from the first model") if(sum(subs) == 1) stop("The first model has a different response from the rest") forms <- forms[, subs] object <- object[subs] ## older survival objects might have df.resid: recent ones have df.residual dfres <- sapply(object, "[[", "df.resid", exact=FALSE) m2loglik <- -2 * sapply(object, "[[", "loglik")[2, ] tl <- lapply(object, labels) rt <- length(m2loglik) effects <- character(rt) for(i in 2:rt) effects[i] <- diff.term(tl[c(i - 1, i)], i) dm2loglik <- - diff(m2loglik) ddf <- - diff(dfres) heading <- c("Analysis of Deviance Table", paste("\nResponse: ", forms[2, 1], "\n", sep = "")) aod <- data.frame(Terms = forms[3, ], "Resid. Df" = dfres, "-2*LL" = m2loglik, Test = effects, Df = c(NA, ddf), Deviance = c(NA, dm2loglik), check.names = FALSE) if (is.R()) aod<-structure(aod,heading=heading,class=c("anova","data.frame")) # else aod <- as.anova(aod, heading) #code for Splus if(test != "none") { n <- length(object[[1]]$residuals) o <- order(dfres) ## R uses scale argument even for "Chisq" if (test=="Chisq") scale<-1 else scale<-sum(object$residuals^2)/dfres[o[1]] stat.anova(aod, test, scale, dfres[o[1]], n) } else aod } survival/R/survfitms.R0000644000175100001440000005303412533653612014544 0ustar hornikusers# Automatically generated from all.nw using noweb # Methods for survfitms objects summary.survfit <- function(object, times, censored=FALSE, scale=1, extend=FALSE, rmean=getOption('survfit.rmean'), ...) { fit <- object if (!inherits(fit, 'survfit')) stop("summary.survfit can only be used for survfit objects") # The print.rmean option is depreciated, it is still listened # to in print.survfit, but ignored here if (is.null(rmean)) rmean <- "none" temp <- survmean(fit, scale=scale, rmean) table <- temp$matrix #for inclusion in the output list rmean.endtime <- temp$end.time if (!missing(times)) { if (!is.numeric(times)) stop ("times must be numeric") times <- sort(times) } # The fit$surv object is sometimes a vector and sometimes a # matrix. We calculate row indices first, and then deal # with the cases at the end. nsurv <- if (is.matrix(fit$surv)) nrow(fit$surv) else length(fit$surv) if (is.null(fit$strata)) { nstrat <- 1 stemp <- rep(1L, nsurv) strata.names <- "" } else { nstrat <- length(fit$strata) stemp <- rep(1:nstrat, fit$strata) strata.names <- names(fit$strata) } if (missing(times)) { # just pick off the appropriate rows of the output if (censored) indx1 <- seq(along=fit$time) else indx1 <- which(fit$n.event>0) } else { # The one line function below might be opaque (even to me) -- # For n.event, we want to know the number since the last chosen # printout time point. Start with the curve of cumulative # events at c(0, stime) (the input time points), which is # the cumsum below; pluck off the values corresponding to our # time points, the [x] below; then get the difference since the # last chosen time point (or from 0, for the first chosen point). cfun <- function(x, data) diff(c(0, cumsum(c(0,data))[x])) # Process the curves one at a time, # adding the results for that curve onto a list, so the # number of events will be n.enter[[1]], n.enter[[2]], etc. # For the survival, stderr, and confidence limits it suffices # to create a single list 'indx1' containing a subscripting vector indx1 <- n.risk <- n.event <- newtimes <- vector('list', nstrat) n.enter <- vector('list', nstrat) n.censor<- vector('list', nstrat) n <- length(stemp) for (i in 1:nstrat) { who <- (1:n)[stemp==i] # the rows of the object for this strata stime <- fit$time[who] # First, toss any printing times that are outside our range if (is.null(fit$start.time)) mintime <- min(stime, 0) else mintime <- fit$start.time ptimes <- times[times >= mintime] if (!extend) { maxtime <- max(stime) ptimes <- ptimes[ptimes <= maxtime] } newtimes[[i]] <- ptimes # If we tack a -1 onto the front of the vector of survival # times, then indx1 is the subscript for that vector # corresponding to the list of "ptimes". If the input # data had stime=c(10,20) and ptimes was c(5,10,15,20), # the result would be 1,2,2,3. # For n.risk we want a slightly different index: 2,2,3,3. # "In between" times point to the next higher index for n.risk, # but the next lower one for survival. (Survival drops at time t, # the n.risk immediately afterwords at time t+0: you were at # risk just before you die, but not a moment after). The # extra point needs to be added at the end. # ntime <- length(stime) #number of points temp1 <- approx(c(mintime-1, stime), 0:ntime, xout=ptimes, method='constant', f=0, rule=2)$y indx1[[i]] <- ifelse(temp1==0, 1, 1+ who[pmax(1,temp1)]) # Why not just "who[temp1]" instead of who[pmax(1,temp1)] in the # line just above? When temp1 has zeros, the first expression # gives a vector that is shorter than temp1, and the ifelse # doesn't work right due to mismatched lengths. n.event[[i]] <- cfun(temp1+1, fit$n.event[who]) if (!is.null(fit$n.censor)) { n.censor[[i]] <- cfun(temp1+1, fit$n.censor[who]) j <- who[ntime] #last time point in the data last.n <- fit$n.risk[j] - (fit$n.event[j]+ fit$n.censor[j]) } else { # this is for the older survfit objects, which don't contain # n.censor. In this case, we don't know how many of the # people at the last time are censored then & how many go # on further. Assume we lose them all. Note normally # extend=FALSE, so this number isn't printed anyway. last.n <- 0 } # Compute the number at risk. If stime = 1,10, 20 and ptime=3,10, # 12, then temp1 = 2,2,3: the nrisk looking ahead # approx() doesn't work if stime is of length 1 if (ntime ==1) temp1 <- rep(1, length(ptimes)) else temp1 <- approx(stime, 1:ntime, xout=ptimes, method='constant', f=1, rule=2)$y n.risk[[i]] <- ifelse(ptimes>max(stime), last.n, fit$n.risk[who[temp1]]) } times <- unlist(newtimes) n.risk <- unlist(n.risk) n.event <- unlist(n.event) n.enter <- unlist(n.enter) #may be NULL n.censor<- unlist(n.censor) #may be NULL indx1 <- unlist(indx1) } # Create an output structure if (length(indx1)==length(fit$time) && all(indx1 == seq(along=fit$time))) { temp <- object #no change temp$time <- temp$time/scale temp$table <- table if (!is.null(temp$strata)) temp$strata <- factor(stemp, labels=strata.names) } else if (missing(times)) { #default censor=FALSE case temp <- object temp$time <- temp$time[indx1]/scale temp$table <- table for (j in c("n.risk", "n.event", "n.censor", "n.enter", "surv", "std.err", "cumhaz", "lower", "upper")) { zed <- temp[[j]] if (!is.null(zed)) { if (is.matrix(zed)) temp[[j]] <- zed[indx1,,drop=FALSE] else temp[[j]] <- zed[indx1] } } if (!is.null(temp$strata)) temp$strata <- factor(stemp[indx1], levels=1:nstrat, labels=strata.names) } else { #times argument was given temp <- list(n=object$n, time=times/scale, n.risk=n.risk, n.event=n.event, conf.int=fit$conf.int, type=fit$type, table=table) if (!is.null(n.censor)) temp$n.censor <- n.censor if (!is.null(n.enter)) temp$n.enter <- n.enter if (!is.null(fit$start.time)) temp$start.time <- fit$start.time # why the rbind? The user may have specified a time point before # the first event, and indx1=1 indicates that case if (is.matrix(fit$surv)) { temp$surv <- rbind(1, fit$surv)[indx1,,drop=FALSE] if (!is.null(fit$std.err)) temp$std.err <- rbind(0, fit$std.err)[indx1,,drop=FALSE] if (!is.null(fit$lower)) { temp$lower <- rbind(1, fit$lower)[indx1,,drop=FALSE] temp$upper <- rbind(1, fit$upper)[indx1,,drop=FALSE] } if (!is.null(fit$cumhaz)) temp$cumhaz <- rbind(0, fit$cumhaz)[indx1,,drop=FALSE] } else { temp$surv <- c(1, fit$surv)[indx1] if (!is.null(fit$std.err)) temp$std.err <- c(0,fit$std.err)[indx1] if (!is.null(fit$lower)) { temp$lower <- c(1, fit$lower)[indx1] temp$upper <- c(1, fit$upper)[indx1] } if (!is.null(fit$cumhaz)) temp$cumhaz <- c(0, fit$cumhaz)[indx1] } if (!is.null(fit$strata)) { scount <- unlist(lapply(newtimes, length)) temp$strata <- factor(rep(1:nstrat, scount), levels=1:nstrat, labels=strata.names) } if (length(rmean.endtime)>0 && !is.na(rmean.endtime)) temp$rmean.endtime <- rmean.endtime temp$call <- fit$call if (!is.null(fit$na.action)) temp$na.action <- fit$na.action } if (!is.null(temp$std.err)) temp$std.err <- temp$std.err*temp$surv #std error of the survival curve class(temp) <- 'summary.survfit' temp } summary.survfitms <- function(object, times, censored=FALSE, scale=1, extend=FALSE, rmean=getOption('survfit.rmean'), ...) { fit <- object if (!inherits(fit, 'survfitms')) stop("summary.survfitms can only be used for survfitms objects") if (is.null(rmean)) rmean <- "none" if (!missing(times)) { if (!is.numeric(times)) stop ("times must be numeric") times <- sort(times) } # add some temps to make survmean work object$surv <- 1-object$prev if (is.matrix(object$surv)) dimnames(object$surv) <- list(NULL, object$states) temp <- survmean(object, scale=scale, rmean) table <- temp$matrix #for inclusion in the output list rmean.endtime <- temp$end.time # The fit$prev object is usually a matrix but can be a vector # We calculate row indices first, and then deal # with the cases at the end. nprev <- if (is.matrix(fit$prev)) nrow(fit$prev) else length(fit$prev) if (is.null(fit$strata)) { nstrat <- 1 stemp <- rep(1L, nprev) strata.names <- "" } else { nstrat <- length(fit$strata) stemp <- rep(1:nstrat, fit$strata) strata.names <- names(fit$strata) } if (missing(times)) { # just pick off the appropriate rows of the output if (censored) indx1 <- seq(along=fit$time) else indx1 <- which(fit$n.event>0) } else { # The one line function below might be opaque (even to me) -- # For n.event, we want to know the number since the last chosen # printout time point. Start with the curve of cumulative # events at c(0, stime) (the input time points), which is # the cumsum below; pluck off the values corresponding to our # time points, the [x] below; then get the difference since the # last chosen time point (or from 0, for the first chosen point). cfun <- function(x, data) diff(c(0, cumsum(c(0,data))[x])) # Process the curves one at a time, # adding the results for that curve onto a list, so the # number of events will be n.enter[[1]], n.enter[[2]], etc. # For the survival, stderr, and confidence limits it suffices # to create a single list 'indx1' containing a subscripting vector indx1 <- n.risk <- n.event <- newtimes <- vector('list', nstrat) n.enter <- vector('list', nstrat) n.censor<- vector('list', nstrat) n <- length(stemp) for (i in 1:nstrat) { who <- (1:n)[stemp==i] # the rows of the object for this strata stime <- fit$time[who] # First, toss any printing times that are outside our range if (is.null(fit$start.time)) mintime <- min(stime, 0) else mintime <- fit$start.time ptimes <- times[times >= mintime] if (!extend) { maxtime <- max(stime) ptimes <- ptimes[ptimes <= maxtime] } newtimes[[i]] <- ptimes # If we tack a -1 onto the front of the vector of survival # times, then indx1 is the subscript for that vector # corresponding to the list of "ptimes". If the input # data had stime=c(10,20) and ptimes was c(5,10,15,20), # the result would be 1,2,2,3. # For n.risk we want a slightly different index: 2,2,3,3. # "In between" times point to the next higher index for n.risk, # but the next lower one for survival. (Survival drops at time t, # the n.risk immediately afterwords at time t+0: you were at # risk just before you die, but not a moment after). The # extra point needs to be added at the end. # ntime <- length(stime) #number of points temp1 <- approx(c(mintime-1, stime), 0:ntime, xout=ptimes, method='constant', f=0, rule=2)$y indx1[[i]] <- ifelse(temp1==0, 1, 1+ who[pmax(1,temp1)]) # Why not just "who[temp1]" instead of who[pmax(1,temp1)] in the # line just above? When temp1 has zeros, the first expression # gives a vector that is shorter than temp1, and the ifelse # doesn't work right due to mismatched lengths. n.event[[i]] <- cfun(temp1+1, fit$n.event[who]) if (!is.null(fit$n.censor)) { n.censor[[i]] <- cfun(temp1+1, fit$n.censor[who]) j <- who[ntime] #last time point in the data last.n <- fit$n.risk[j] - (fit$n.event[j]+ fit$n.censor[j]) } else { # this is for the older survfit objects, which don't contain # n.censor. In this case, we don't know how many of the # people at the last time are censored then & how many go # on further. Assume we lose them all. Note normally # extend=FALSE, so this number isn't printed anyway. last.n <- 0 } # Compute the number at risk. If stime = 1,10, 20 and ptime=3,10, # 12, then temp1 = 2,2,3: the nrisk looking ahead # approx() doesn't work if stime is of length 1 if (ntime ==1) temp1 <- rep(1, length(ptimes)) else temp1 <- approx(stime, 1:ntime, xout=ptimes, method='constant', f=1, rule=2)$y n.risk[[i]] <- ifelse(ptimes>max(stime), last.n, fit$n.risk[who[temp1]]) } times <- unlist(newtimes) n.risk <- unlist(n.risk) n.event <- unlist(n.event) n.enter <- unlist(n.enter) #may be NULL n.censor<- unlist(n.censor) #may be NULL indx1 <- unlist(indx1) } # Create an output structure if (length(indx1)== length(fit$time) && all(indx1 == seq(along=fit$time))) { temp <- object #no change temp$time <- temp$time/scale temp$table <- table if (!is.null(temp$strata)) temp$strata <- factor(stemp, levels=1:nstrat, labels=strata.names) } else if (missing(times)) { temp <- object temp$time <- temp$time[indx1]/scale temp$table <- table for (j in c("n.risk", "n.event", "n.censor", "n.enter", "prev", "std.err", "lower", "upper")) { zed <- temp[[j]] if (!is.null(zed)) { if (is.matrix(zed)) temp[[j]] <- zed[indx1,,drop=FALSE] else temp[[j]] <- zed[indx1] } } temp$cumhaz <- fit$cumhaz[,,indx1,drop=FALSE] if (!is.null(temp$strata)) temp$strata <- factor(stemp[indx1], levels=1:nstrat, labels=strata.names) } else { temp <- list(n=object$n, time=times/scale, n.risk=n.risk, n.event=n.event, conf.int=fit$conf.int, type=fit$type, table=table) if (!is.null(n.censor)) temp$n.censor <- n.censor if (!is.null(n.enter)) temp$n.enter <- n.enter if (!is.null(fit$start.time)) temp$start.time <- fit$start.time # why the rbind? The user may have specified a time point before # the first event, and indx1=1 indicates that case # the cumhaz array can't be done with a 1-liner if (is.matrix(fit$prev)) { temp$prev <- rbind(0, fit$prev)[indx1,,drop=FALSE] zz <- ifelse(indx1==1, NA, indx1-1) temp$cumhaz <- fit$cumhaz[,,zz, drop=FALSE] temp$cumhaz <- ifelse(is.na(temp$cumhaz), 0, temp$cumhaz) if (!is.null(fit$std.err)) temp$std.err <- rbind(0, fit$std.err)[indx1,,drop=FALSE] if (!is.null(fit$lower)) { temp$lower <- rbind(0, fit$lower)[indx1,,drop=FALSE] temp$upper <- rbind(0, fit$upper)[indx1,,drop=FALSE] } } else { temp$prev <- c(0, fit$prev[indx1]) temp$cumhaz <- c(0, fit$cumhaz[indx1]) if (!is.null(fit$std.err)) temp$std.err <- c(0, fit$std.err)[indx1] if (!is.null(fit$lower)) { temp$lower <- c(0, fit$lower)[indx1] temp$upper <- c(0, fit$upper)[indx1] } } if (!is.null(fit$strata)) { scount <- unlist(lapply(newtimes, length)) temp$strata <- factor(rep(1:nstrat, scount), levels=1:nstrat, labels=strata.names) } temp$call <- fit$call if (!is.null(fit$na.action)) temp$na.action <- fit$na.action } if (length(rmean.endtime)>0 && !is.na(rmean.endtime)) temp$rmean.endtime <- rmean.endtime class(temp) <- "summary.survfitms" temp } "[.survfitms" <- function(x, ..., drop=TRUE) { nmatch <- function(indx, target) { # This function lets R worry about character, negative, or logical subscripts # It always returns a set of positive integer indices temp <- 1:length(target) names(temp) <- target temp[indx] } if (missing(..1)) i<- NULL else i <- sort(..1) if (missing(..2)) j<- NULL else j <- ..2 if (is.null(x$strata)) { if (is.matrix(x$prev)) { # No strata, but a matrix of prevalence values # In this case, allow them to use a single i subscript as well if (is.null(j) && !is.null(i)) j <- i indx <- nmatch(j, x$states) if (any(is.na(indx))) stop("unmatched subscript", j[is.na(indx)]) else j <- as.vector(indx) x$states <- x$states[j] if (nrow(x$prev)==1 && length(j) > 1) drop<- FALSE x$prev <- x$prev[,j,drop=drop] x$cumhaz <- x$cumhaz[j,j,, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[,j,drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[,j,drop=drop] } else warning("Survfit object has only a single survival curve") } else { if (is.null(i)) keep <- seq(along.with=x$time) # rows to keep else { indx <- nmatch(i, names(x$strata)) #strata to keep if (any(is.na(indx))) stop(paste("strata", paste(i[is.na(indx)], collapse=' '), 'not matched')) # Now, i may not be in order: a user has curve[3:2] to reorder a plot # Hence the list/unlist construct which will reorder the data in the curves temp <- rep(1:length(x$strata), x$strata) keep <- unlist(lapply(i, function(x) which(temp==x))) if (length(i) <=1 && drop) x$strata <- NULL else x$strata <- x$strata[indx] x$n <- x$n[indx] x$time <- x$time[keep] x$n.risk <- x$n.risk[keep] x$n.event <- x$n.event[keep] x$n.censor<- x$n.censor[keep] } if (is.matrix(x$prev)) { # If [i,] selected only 1 row, don't collapse the columns if (length(keep) <2 && (is.null(j) || length(j) >1)) drop <- FALSE if (is.null(j)) { #only subscript rows (strata) x$prev <- x$prev[keep,,drop=drop] x$cumhaz <- x$cumhaz[,,keep, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,,drop=drop] if (!is.null(x$upper)) x$upper <-x$upper[keep,,drop=drop] if (!is.null(x$lower)) x$lower <-x$lower[keep,,drop=drop] } else { #subscript both rows (strata) and columns (states) indx <- nmatch(j, x$states) if (any(is.na(indx))) stop("unmatched subscript", j[indx]) else j <- as.vector(indx) x$states <- x$states[j] x$prev <- x$prev[keep,j, drop=drop] x$cumhaz <- x$cumhaz[j,j,keep, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[keep,j, drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[keep,j, drop=drop] } } else { x$prev <- x$prev[keep] x$cumhaz <- x$cumhaz[keep] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep] if (!is.null(x$upper)) x$upper <- x$upper[keep] if (!is.null(x$lower)) x$lower <- x$lower[keep] } } x } survival/R/tmerge.R0000644000175100001440000003414612533653613013771 0ustar hornikusers# Automatically generated from all.nw using noweb tmerge <- function(data1, data2, id, ..., tstart, tstop, options) { Call <- match.call() # The function wants to recognize special keywords in the # arguments, so define a set of functions which will be used to # mark objects new <- new.env(parent=parent.frame()) assign("tdc", function(...) {x <- list(...); class(x) <- "tdc"; x}, envir=new) assign("cumtdc", function(...) {x <- list(...); class(x) <-"cumtdc"; x}, envir=new) assign("event", function(...) {x <- list(...); class(x) <-"event"; x}, envir=new) assign("cumevent", function(...) {x <- list(...); class(x) <-"cumevent"; x}, envir=new) if (missing(data1)) stop("the data1 argument is required") if (missing(id)) stop("the id argument is required") if (!inherits(data1, "data.frame")) stop("data1 must be a data frame") tmerge.control <- function(id="id", tstart="tstart", tstop="tstop", defer =0) { if (length(defer) !=1 || !is.numeric(defer) || defer <0) stop("defer option must be a non-negative number") if (!is.character(id)) stop("id option must be a character string") if (!is.character(tstart)) stop("tstart option must be a character string") if (!is.character(tstop)) stop("tstop option must be a character string") list(id=id, tstart=tstart, tstop=tstop, defer=defer) } tname <- attr(data1, "tname") if (!is.null(tname) && any(is.null(match(unlist(tname), names(data1))))) stop("data1 does not match its own tname attribute") if (!missing(options)) { if (!is.list(options)) stop("options must be a list") if (!is.null(tname)) { # Changing a name partway through a set of calls? if (any(!is.na(match(names(options), names(tname))))) stop("cannot change names in mid-stream") topt <- do.call(tmerge.control, c(tname, options)) } else topt <- do.call(tmerge.control, options) } else if (length(tname)) topt <- do.call(tmerge.control, tname) else topt <- tmerge.control() # id, tstart, tstop are found in data2, if it is present if (!missing(data2)) { id <- eval(Call[["id"]], data2) if (!missing(tstart)) tstart <- eval(Call[["tstart"]], data2) if (!missing(tstop)) tstop <- eval(Call[["tstop"]], data2) } if (!missing(tstart) && length(tstart) != length(id)) stop("tstart and id must be the same length") if (!missing(tstop) && length(tstop) != length(id)) stop("tstop and id must be the same length") # grab the... arguments notdot <- c("data1", "data2", "id", "tstart", "tstop", "options") dotarg <- Call[is.na(match(names(Call), notdot))] dotarg[[1]] <- as.name("list") # The as-yet dotarg arguments if (missing(data2)) args <- eval(dotarg, envir=new) else args <- eval(dotarg, data2, enclos=new) argclass <- sapply(args, function(x) (class(x))[1]) argname <- names(args) if (any(argname== "")) stop("all additional argments must have a name") check <- match(argclass, c("tdc", "cumtdc", "event", "cumevent")) if (any(is.na(check))) stop(paste("argument(s)", argname[is.na(check)], "not a recognized type")) # The tcount matrix is useful for debugging tcount <- matrix(0L, length(argname), 8) dimnames(tcount) <- list(argname, c("early","late", "gap", "within", "boundary", "leading", "trailing", "tied")) tevent <- attr(data1, "tevent") # event type variables newdata <- data1 # make a copy if (!missing(tstop) || length(tname)==0) { # This is a first call indx <- match(c(topt$id, topt$tstart, topt$tstop), names(data1), nomatch=0) if (all(indx[1:2] >0) && missing(tstop)) { # case 1 above, just some data checks if (indx[3]==0) newdata[[topt$tstart]] <- 0 if (!is.numeric(newdata[[topt$tstart]]) || !is.numeric(newdata[[topt$tstop]])) stop("start and end variables must be numeric") if (any(newdata[[topt$tstart]] >= newdata[[topt$tstop]])) stop("stop time must be > start time for all observations") # If there are duplicated ids, then we need to ensure that each subject # is a sequential set of observations, sorted by time (just sort it) # If tstart was not supplied, we need to correct our "0" created above baseid <- data1[[topt$id]] if (any(duplicated(baseid))) { indx <- order(newdata$id, newdata$tstop) if (any(indx != seq(along.with=indx))) { # sort the data newdata <- newdata[indx,] baseid <- baseid[indx] } newid <- !duplicated(baseid) #is this row a new id value? n <- nrow(newdata) if (length(tstart) ==0) newdata$tstart <- ifelse(newid, 0, c(0, newdata$tstop[-n])) else { ok <- (newid[-1] | newdata$tstart[-1] >= newdata$tstop[-n]) if (any(!ok)) stop("overlapping time intervals for a subject") } } } else { if (missing(tstop)) { # case 3 above if (length(argclass)==0 || argclass[1] != "event") stop("neither a tstop argument nor an initial event argument was found") tstop <- args[[1]][[1]] } # case 2 and case 3 if (any(is.na(tstop))) stop("missing time value, when that variable defines the span") if (missing(tstart)) tstart <- rep(0, length(id)) if (any(tstart >= tstop)) stop("stop time must be > start time for all observations") if (indx[1] >0) { # the id variable is in data1 baseid <- data1[[topt$id]] if (any(duplicated(baseid))) stop("duplicate identifiers in data1") indx2 <- match(id, baseid) if (any(is.na(indx2))) stop("'id' has values not in data1") } else { if (nrow(data1) != nrow(data2)) stop("nrow(data1) != nrow(data2) and data1 is missing the id") indx2 <- seq.int(along.with=id) newdata[topt$id] <- id } newdata[indx2, topt$tstart] <- tstart newdata[indx2, topt$tstop] <- tstop } } else { #not a first call if (any(is.na(match(id, data1[[topt$id]])))) stop("id values found in data2 which are not in data1") } saveid <- id for (ii in seq(along.with=args)) { argi <- args[[ii]] baseid <- newdata[[topt$id]] dstart <- newdata[[topt$tstart]] dstop <- newdata[[topt$tstop]] # if an event time is missing then skip that obs etime <- argi[[1]] keep <- !is.na(etime) etime <- etime[keep] id <- saveid[keep] if (length(etime) != length(id)) stop("argument", argname[ii], "is not the same length as id") # For an event or cumevent, one of the later steps becomes much # easier if we sort the new data by id and time indx <- order(id, etime) id <- id[indx] etime <- etime[indx] if (length(argi) > 1) yinc <- (argi[[2]])[indx] # indx1 points to the closest start time in the baseline data (data1) # that is <= etime. indx2 to the closest end time that is >=etime. # If etime falls into a (tstart, tstop) interval, indx1 and indx2 # will match # If the "defer" argument is set and this event is of type tdc, then # any event times are artificially moved left by "defer" amount wrt # doing the indx2 match. This will cause an insertion that is too close # to an event to be labeled as itype=3 (or itype=2 if this was the last # interval for the subject) and so map later in time. defer <- rep(0., nrow(newdata)) if (topt$defer >0 && length(tevent) && argclass[ii] %in% c("tdc", "cumtdc")) { for (ename in tevent) { temp <- newdata[[ename]] if (is.logical(temp)) defer[temp] <- topt$defer else defer[temp!=0] <- topt$defer } } indx1 <- neardate(id, baseid, etime, dstart, best="prior") indx2 <- neardate(id, baseid, etime, dstop+defer, best="after") # The event times fall into one of 5 categories # 1. Before the first interval # 2. After the last interval # 3. Outside any interval but with time span, i.e, it falls into # a gap in follow-up # 4. Strictly inside an interval (does't touch either end) # 5. Inside an interval, but touching. itype <- ifelse(is.na(indx1), 1, ifelse(is.na(indx2), 2, ifelse(indx2 > indx1, 3, ifelse(etime== dstart[indx1] | etime== dstop[indx2], 5, 4)))) # Subdivide the events that touch on a boundary # 1: intervals of (a,b] (b,d], new count at b "tied edge" # 2: intervals of (a,b] (c,d] with c>b, new count at c, "front edge" # 3: intervals of (a,b] (c,d] with c>b, new count at b, "back edge" # subtype <- ifelse(itype!=5, 0, ifelse(indx1 == indx2+1, 1, ifelse(etime==dstart[indx1], 2, 3))) tcount[ii,1:7] <- table(factor(itype+subtype, levels=c(1:4, 6:8))) # count ties. id and etime are not necessarily sorted tcount[ii,8] <- sum(tapply(etime, id, function(x) sum(duplicated(x)))) # Look to see if this term has one or two arguments. If one arg # then the increment is 1, else it is the second arg. The myfun() # function will later compute totals by unique subject/time pair # if (length(argi) >1) { if (length(argi) > 2) stop("too many variables in an", argclass[ii], "call") if (diff(sapply(argi, length)) !=0) stop("different lengths in an", argclass[ii], "call") if (!is.numeric(yinc) && argclass[ii] != "event") stop("non numeric increment in an", argclass[ii], "call") myfun <- function(x, grp) { temp <- tapply(yinc[grp], x[grp], sum) ifelse(is.na(temp), 0, temp) } } else { myfun <- function(x, grp) table(x[grp]) yinc <- rep(1.0, length(etime)) # each counts as 1 } indx4 <- which(itype==4) n4 <- length(indx4) if (n4 > 0) { icount <- tapply(etime[indx4], indx1[indx4], function(x) sort(unique(x))) n.add <- sapply(icount, length) #number of rows to add # expand the data irep <- rep.int(1L, nrow(newdata)) erow <- unique(indx1[indx4]) # which rows in newdata to be expanded irep[erow] <- 1+ n.add # number of rows in new data jrep <- rep(1:nrow(newdata), irep) #stutter the duplicated rows newdata <- newdata[jrep,] #expand it out dstart <- dstart[jrep] dstop <- dstop[jrep] #fix up times nfix <- length(erow) temp <- vector("list", nfix) iend <- (cumsum(irep))[irep >1] #end row of each duplication set for (j in 1:nfix) temp[[j]] <- -(seq(n.add[j] -1, 0)) + iend[j] newrows <- unlist(temp) dstart[newrows] <- dstop[newrows-1] <- unlist(icount) newdata[[topt$tstart]] <- dstart newdata[[topt$tstop]] <- dstop for (ename in tevent) newdata[newrows-1, ename] <- 0 if (topt$defer > 0) { defer <- defer[jrep] defer[newrows] <- 0 } else defer <- rep(0, nrow(newdata)) # refresh indices baseid <- newdata[[topt$id]] indx1 <- neardate(id, baseid, etime, dstart, best="prior") indx2 <- neardate(id, baseid, etime, dstop+ defer , best="after") subtype[itype==4] <- 1 #all the "insides" are now on a tied edge itype[itype==4] <- 5 } # add it in if (argclass[ii] %in% c("cumtdc", "cumevent")) yinc <- unlist(tapply(yinc, id, cumsum)) newvar <- newdata[[argname[ii]]] #does the variable exist? if (argclass[ii] %in% c("event", "cumevent")) { if (is.null(newvar)) newvar <- rep(0, nrow(newdata)) keep <- (subtype==1 | subtype==3) # all other events are thrown away newvar[indx2[keep]] <- yinc[keep] tevent <- unique(c(tevent, argname[ii])) } else { keep <- itype != 2 # changes after the last interval are ignored indx <- ifelse(subtype==1, indx1, ifelse(subtype==3, indx2+1L, indx2)) if (is.null(newvar)) { if (length(argi)==1) newvar <- rep(0.0, nrow(newdata)) else newvar <- rep(NA_real_, nrow(newdata)) } # id can be any data type; feed integers to the C routine storage.mode(yinc) <- storage.mode(dstop) <- "double" storage.mode(newvar) <- storage.mode(etime) <- "double" newvar <- .Call("tmerge", match(baseid, baseid), dstop, newvar, match(id, baseid)[keep], etime[keep], yinc[keep], indx[keep]) } newdata[[argname[ii]]] <- newvar } attr(newdata, "tname") <- topt[c("id", "tstart", "tstop")] attr(newdata, "tcount") <- tcount if (length(tevent)) attr(newdata, "tevent") <- tevent row.names(newdata) <- NULL class(newdata) <- c("data.frame") newdata } survival/R/predict.survreg.S0000644000175100001440000001550612470201064015621 0ustar hornikuserspredict.survreg <- function(object, newdata, type=c('response', "link", 'lp', 'linear', 'terms', 'quantile','uquantile'), se.fit=FALSE, terms=NULL, p=c(.1, .9), na.action=na.pass, ...) { # # What do I need to do predictions ? # # linear predictor: exists # +se : X matrix # newdata : new X matrix # # response -- same as lp, +transform, from distribution # # p -- density function from distribution # scale(s) -- if multiple I need the strata # +se : variance matrix # newdata: new X type <-match.arg(type) if (type=='link') type<- 'lp' #true until their are link functions if (type=='linear') type<- 'lp' n <- length(object$linear.predictors) Terms <- object$terms if(!inherits(Terms, "terms")) stop("invalid terms component of object") strata <- attr(Terms, 'specials')$strata Terms <- delete.response(Terms) coef <- object$coefficients intercept <- attr(Terms, "intercept") nvar <- length(object$coefficients) vv <- object$var[1:nvar, 1:nvar] fixedscale <- (nvar == ncol(object$var)) if (missing(newdata) && (type=='terms' || se.fit)) need.x <- TRUE else need.x <- FALSE if (!missing(newdata)){ newframe <- model.frame(Terms, data=newdata, na.action= na.action, xlev=object$xlevels) na.action.used <- attr(newframe, 'na.action') } else na.action.used <- object$na.action if (length(strata) && (type=='quantile' || type=='uquantile') && !fixedscale) { # # We need to reconstruct the original "strata" variable # mf <- model.frame(object) temp <- untangle.specials(Terms, 'strata', 1) if (length(temp$vars)==1) strata.keep <- mf[[temp$vars]] else strata.keep <- strata(mf[,temp$vars], shortlabel=TRUE) strata <- as.numeric(strata.keep) nstrata <- max(strata) if (missing(newdata) && need.x){ #need the old x x <- object[['x']] if (is.null(x)) x <- model.matrix(object, mf) } else if (!missing(newdata)) { #need the new x if (length(temp$vars)==1) newstrat <- newframe[[temp$vars]] else newstrat <- strata(newframe[,temp$vars], shortlabel=TRUE) strata <- match(newstrat, levels(strata.keep)) x <- model.matrix(object, newframe) offset <- model.offset(newframe) } } else { # per subject strata not needed nstrata <- 1 if (missing(newdata)) { strata <- rep(1L, n) if (need.x) x <- model.matrix(object) } else { x <- model.matrix(object, newframe) strata <- rep(1L, nrow(x)) offset <- 0 } } scale <- object$scale[strata] #center x if terms are to be computed if(type=='p' || (type == "terms" && intercept)) x <- sweep(x, 2, object$means) # # Grab the distribution # if (is.character(object$dist)) dd <- survreg.distributions[[object$dist]] else dd <- object$dist if (is.null(dd$itrans)) { itrans <- function(x) x # identity transformation dtrans <- function (x) 1 # derivative of the transformation } else { itrans <- dd$itrans dtrans <- dd$dtrans } if (!is.null(dd$dist)) dd <- survreg.distributions[[dd$dist]] # # Now, lay out the code one case at a time. # There is some repetition this way, but otherwise the code just gets # too complicated. # if (type=='lp' || type=='response') { if (missing(newdata)) { pred <- object$linear.predictors # names(pred) <- names(object$residuals) } else pred <- drop(x %*% coef) + offset if (se.fit) se <- sqrt(diag(x %*% vv %*% t(x))) if (type=='response') { pred <- itrans(pred) if (se.fit) se <- se/ dtrans(pred) } } else if (type=='quantile' || type=='uquantile') { if (missing(newdata)) pred <- object$linear.predictors else pred <- x %*% coef # "pred" is the mean of the distribution, # now add quantiles and then invert qq <- dd$quantile(p, object$parm) if (length(qq)==1 || length(pred)==1) { pred <- pred + qq*scale if (se.fit && fixedscale) { var <- ((x %*% vv) * x) %*% rep(1., ncol(x)) se <- rep(sqrt(drop(var)), length(qq)) } else if (se.fit) { x.strata <- outer(strata, 1:nstrata, function(x,y) 1*(x==y)) se <- matrix(0, ncol=length(qq), nrow=nrow(x)) for (i in 1:(length(qq))) { temp <- cbind(x, (qq[i]*scale)* x.strata) var <- ((temp %*% object$var) *temp) %*% rep(1, ncol(temp)) se[,i] <- sqrt(drop(var)) } se <- drop(se) } } else { pred <- c(pred) + outer(scale, qq) if (se.fit && fixedscale) { var <- ((x %*% vv) * x) %*% rep(1., ncol(x)) if (length(qq) >1) { se <- rep(sqrt(drop(var)), length(qq)) se <- matrix(se, ncol=length(qq)) } else se <- sqrt(drop(var)) } else if (se.fit) { x.strata <- outer(strata, 1:nstrata, function(x,y) 1*(x==y)) se <- pred nc <- rep(1., ncol(object$var)) for (i in 1:length(qq)) { temp <- cbind(x, (qq[i]*scale)*x.strata) var <- ((temp %*% object$var)* temp) %*% nc se[,i] <- sqrt(drop(var)) } se <- drop(se) } } pred <- drop(pred) if (type == 'quantile') { pred <- itrans(pred) if (se.fit) se <- se/dtrans(pred) } } else { #terms if (is.R()) { # In S we can use Build.terms, in R we have to do it ourselves asgn <- attrassign(x,Terms) hasintercept<-attr(Terms,"intercept")>0 if (hasintercept) asgn$"(Intercept)"<-NULL nterms<-length(asgn) pred<-matrix(ncol=nterms,nrow=NROW(x)) dimnames(pred)<-list(rownames(x),names(asgn)) if (se.fit){ se<-matrix(ncol=nterms,nrow=NROW(x)) dimnames(se)<-list(rownames(x),names(asgn)) R<-object$var } for (i in 1:nterms){ ii<-asgn[[i]] pred[,i]<-x[,ii,drop=FALSE]%*%(coef[ii]) if (se.fit){ for(j in (1:NROW(x))){ xi<-x[j,ii,drop=FALSE]*(coef[ii]) vci<-R[ii,ii] se[j,i]<-sqrt(sum(xi%*% vci %*%t( xi))) } } } if (!is.null(terms)){ pred<-pred[,terms,drop=FALSE] if (se.fit) se<-se[,terms,drop=FALSE] } } # Splus code, commented out to stop a warning from R CMD check # else { # # Splus: use Build.terms to do the work # asgn <- attr(x, 'assign') # attr(x, 'constant') <- object$means # terms <- match.arg(Terms, labels.lm(object)) # asgn <- asgn[terms] # # if (se.fit) { # temp <- Build.terms(x, coef, vv, asgn, FALSE) # pred <- temp$fit # se <- temp$se.fit # } # else pred<- Build.terms(x, coef, NULL, asgn, FALSE) # const<- attr(pred, 'constant') # } } #Expand out the missing values in the result # if (!is.null(na.action.used)) { pred <- naresid(na.action.used, pred) if(se.fit) se <- naresid(na.action.used, se) } if (se.fit) list(fit=pred, se.fit=se) else pred } survival/R/print.survfit.S0000644000175100001440000002340712534355271015342 0ustar hornikusersprint.survfit <- function(x, scale=1, digits = max(options()$digits - 4, 3), print.rmean = getOption('survfit.print.rmean'), rmean = getOption('survfit.rmean'), ...) { if (inherits(x, "survfitms")) { x$surv <- 1- x$prev if (is.matrix(x$surv)) dimnames(x$surv) <- list(NULL, x$states) if (!is.null(x$lower)) { x$lower <- 1- x$lower x$upper <- 1- x$upper } } if (!is.null(cl<- x$call)) { cat("Call: ") dput(cl) cat("\n") } omit <- x$na.action if (length(omit)) cat(" ", naprint(omit), "\n") savedig <- options(digits=digits) on.exit(options(savedig)) # The print.rmean option is depreciated, with the more general # rmean option taking its place. But if someone specifically # uses print.rmean in the call, or has it as an option without # the rmean option, listen to them. if (!missing(print.rmean) && is.logical(print.rmean) && missing(rmean)) { if (print.rmean) rmean <- 'common' else rmean <- 'none' } else { if (is.null(rmean)) { if (is.logical(print.rmean)) { if (print.rmean) rmean <- 'common' else rmean <- 'none' } else rmean <- 'none' #no option set } # Check validity: it can be numeric or character if (is.numeric(rmean)) { if (is.null(x$start.time)) { if (rmean < min(x$time)) stop("Truncation point for the mean is < smallest survival") } else if (rmean < x$start.time) stop("Truncation point for the mean is < smallest survival") } else { rmean <- match.arg(rmean, c('none', 'common', 'individual')) if (length(rmean)==0) stop("Invalid value for rmean option") } } temp <- survmean(x, scale=scale, rmean) # If the first columns of survmean are identical, suppress duplicates # mtemp <- if (is.matrix(temp$matrix)) temp$matrix else matrix(temp$matrix, nrow=1, dimnames=list(NULL, names(temp$matrix))) if (all(mtemp[,2] == mtemp[,3])){ cname <- dimnames(mtemp)[[2]] mtemp <- mtemp[,-2, drop=FALSE] cname <-cname[-2] cname[2] <- "n" dimnames(mtemp)[[2]] <- cname } if (all(mtemp[,1] == mtemp[,2])) mtemp <- mtemp[,-1, drop=FALSE] temp$matrix <- drop(mtemp) print(temp$matrix) if (rmean != 'none') { if (rmean == 'individual') cat(" * restricted mean with variable upper limit\n") else cat(" * restricted mean with upper limit = ", format(temp$end.time[1]), "\n") } invisible(x) } # # The function that does all of the actual work -- output is a matrix # Used by both print.survfit and summary.survfit # survmean <- function(x, scale=1, rmean) { # The starting point for the integration of the AUC if (!is.null(x$start.time)) start.time <- x$start.time else start.time <- min(0, x$time) # # The function below is called once for each line of output, # i.e., once per curve. It creates the line of output # pfun <- function(nused, time, surv, n.risk, n.event, lower, upper, start.time, end.time) { # # Start by defining a small utility function # Multiple times, we need to find the x corresponding to the first # y that is <.5. (The y's are in decreasing order, but may have # duplicates). # Nuisance 1: if one of the y's is exactly .5, we want the mean of # the corresponding x and the first x for which y<.5. We need to # use the equivalent of all.equal to check for a .5 however: # survfit(Surv(1:100)~1) gives a value of .5 + 1.1e-16 due to # roundoff error. # Nuisance 2: there may by an NA in the y's # Nuisance 3: if no y's are <=.5, then we should return NA # Nuisance 4: the obs (or many) after the .5 may be censored, giving # a stretch of values = .5 +- epsilon # minmin <- function(y, x) { tolerance <- .Machine$double.eps^.5 #same as used in all.equal() keep <- (!is.na(y) & y <(.5 + tolerance)) if (!any(keep)) NA else { x <- x[keep] y <- y[keep] if (abs(y[1]-.5) 1 Y/N # Repeat the code, with minor variations, for each one if (is.null(x$strata)) { if (rmean=='none') end.time <- NA else if (is.numeric(rmean)) end.time <- rmean else end.time <- max(x$time) if (is.matrix(surv)) { out <- matrix(0, ncol(surv), ncols) for (i in 1:ncol(surv)) { if (is.null(x$conf.int)) out[i,] <- pfun(x$n, stime, surv[,i], x$n.risk, x$n.event, NULL, NULL, start.time, end.time) else out[i,] <- pfun(x$n, stime, surv[,i], x$n.risk, x$n.event, x$lower[,i], x$upper[,i], start.time, end.time) } dimnames(out) <- list(dimnames(surv)[[2]], plab) } else { out <- matrix(pfun(x$n, stime, surv, x$n.risk, x$n.event, x$lower, x$upper, start.time, end.time), nrow=1) dimnames(out) <- list(NULL, plab) } } else { #strata case nstrat <- length(x$strata) stemp <- rep(1:nstrat,x$strata) # the index vector for strata1, 2, etc last.time <- (rev(x$time))[match(1:nstrat, rev(stemp))] if (rmean=='none') end.time <- rep(NA, nstrat) else if (is.numeric(rmean)) end.time <- rep(rmean, nstrat) else if (rmean== 'common') end.time <- rep(median(last.time), nstrat) else end.time <- last.time if (is.matrix(surv)) { ns <- ncol(surv) out <- matrix(0, nstrat*ns, ncols) if (is.null(dimnames(surv)[[2]])) dimnames(out) <- list(rep(names(x$strata), rep(ns,nstrat)), plab) else { cname <- outer(dimnames(surv)[[2]], names(x$strata), paste, sep=", ") dimnames(out) <- list(c(cname), plab) } k <- 0 for (i in 1:nstrat) { who <- (stemp==i) for (j in 1:ns) { k <- k+1 if (is.null(x$lower)) out[k,] <- pfun(x$n[i], stime[who], surv[who,j], x$n.risk[who], x$n.event[who], NULL, NULL, start.time, end.time[i]) else out[k,] <- pfun(x$n[i], stime[who], surv[who,j], x$n.risk[who], x$n.event[who], x$lower[who,j], x$upper[who,j], start.time, end.time[i]) } } } else { #non matrix case out <- matrix(0, nstrat, ncols) dimnames(out) <- list(names(x$strata), plab) for (i in 1:nstrat) { who <- (stemp==i) if (is.null(x$lower)) out[i,] <- pfun(x$n[i], stime[who], surv[who], x$n.risk[who], x$n.event[who], NULL, NULL, start.time, end.time[i]) else out[i,] <- pfun(x$n[i], stime[who], surv[who], x$n.risk[who], x$n.event[who], x$lower[who], x$upper[who], start.time, end.time[i]) } } } if (is.null(x$lower)) out <- out[,1:7, drop=F] #toss away the limits if (rmean=='none') out <- out[,-(5:6), drop=F] #toss away the mean & sem list(matrix=out[,,drop=T], end.time=end.time) } survival/R/predict.coxph.R0000644000175100001440000003241312533653603015252 0ustar hornikusers# Automatically generated from all.nw using noweb predict.coxph <- function(object, newdata, type=c("lp", "risk", "expected", "terms"), se.fit=FALSE, na.action=na.pass, terms=names(object$assign), collapse, reference=c("strata", "sample"), ...) { if (!inherits(object, 'coxph')) stop("Primary argument much be a coxph object") Call <- match.call() type <-match.arg(type) n <- object$n Terms <- object$terms if (!missing(terms)) { if (is.numeric(terms)) { if (any(terms != floor(terms) | terms > length(object$assign) | terms <1)) stop("Invalid terms argument") } else if (any(is.na(match(terms, names(object$assign))))) stop("a name given in the terms argument not found in the model") } # I will never need the cluster argument, if present delete it. # Terms2 are terms I need for the newdata (if present), y is only # needed there if type == 'expected' if (length(attr(Terms, 'specials')$cluster)) { temp <- untangle.specials(Terms, 'cluster', 1) Terms <- object$terms[-temp$terms] } else Terms <- object$terms if (type != 'expected') Terms2 <- delete.response(Terms) else Terms2 <- Terms has.strata <- !is.null(attr(Terms, 'specials')$strata) has.offset <- !is.null(attr(Terms, 'offset')) has.weights <- any(names(object$call) == 'weights') na.action.used <- object$na.action n <- length(object$residuals) if (missing(reference) && type=="terms") reference <- "sample" else reference <- match.arg(reference) have.mf <- FALSE if (type == 'expected') { y <- object[['y']] if (is.null(y)) { # very rare case mf <- model.frame(object) y <- model.extract(mf, 'response') have.mf <- TRUE #for the logic a few lines below, avoid double work } } if (se.fit || type=='terms' || (!missing(newdata) && type=="expected") || (has.strata && (reference=="strata") || type=="expected")) { use.x <- TRUE if (is.null(object[['x']]) || has.weights || has.offset || (has.strata && is.null(object$strata))) { # I need the original model frame if (!have.mf) mf <- model.frame(object) if (nrow(mf) != n) stop("Data is not the same size as it was in the original fit") x <- model.matrix(object, data=mf) if (has.strata) { if (!is.null(object$strata)) oldstrat <- object$strata else { stemp <- untangle.specials(Terms, 'strata') if (length(stemp$vars)==1) oldstrat <- mf[[stemp$vars]] else oldstrat <- strata(mf[,stemp$vars], shortlabel=TRUE) } } else oldstrat <- rep(0L, n) weights <- model.weights(mf) if (is.null(weights)) weights <- rep(1.0, n) offset <- model.offset(mf) if (is.null(offset)) offset <- 0 } else { x <- object[['x']] if (has.strata) oldstrat <- object$strata else oldstrat <- rep(0L, n) weights <- rep(1.,n) offset <- 0 } } else { # I won't need strata in this case either if (has.strata) { stemp <- untangle.specials(Terms, 'strata', 1) Terms2 <- Terms2[-stemp$terms] has.strata <- FALSE #remaining routine never needs to look } oldstrat <- rep(0L, n) offset <- 0 use.x <- FALSE } if (!missing(newdata)) { use.x <- TRUE #we do use an X matrix later tcall <- Call[c(1, match(c("newdata", "collapse"), names(Call), nomatch=0))] names(tcall)[2] <- 'data' #rename newdata to data tcall$formula <- Terms2 #version with no response tcall$na.action <- na.action #always present, since there is a default tcall[[1]] <- as.name('model.frame') # change the function called if (!is.null(attr(Terms, "specials")$strata) && !has.strata) { temp.lev <- object$xlevels temp.lev[[stemp$vars]] <- NULL tcall$xlev <- temp.lev } else tcall$xlev <- object$xlevels mf2 <- eval(tcall, parent.frame()) collapse <- model.extract(mf2, "collapse") n2 <- nrow(mf2) if (has.strata) { if (length(stemp$vars)==1) newstrat <- mf2[[stemp$vars]] else newstrat <- strata(mf2[,stemp$vars], shortlabel=TRUE) if (any(is.na(match(newstrat, oldstrat)))) stop("New data has a strata not found in the original model") else newstrat <- factor(newstrat, levels=levels(oldstrat)) #give it all if (length(stemp$terms)) newx <- model.matrix(Terms2[-stemp$terms], mf2, contr=object$contrasts)[,-1,drop=FALSE] else newx <- model.matrix(Terms2, mf2, contr=object$contrasts)[,-1,drop=FALSE] } else { newx <- model.matrix(Terms2, mf2, contr=object$contrasts)[,-1,drop=FALSE] newstrat <- rep(0L, nrow(mf2)) } newoffset <- model.offset(mf2) if (is.null(newoffset)) newoffset <- 0 if (type== 'expected') { newy <- model.response(mf2) if (attr(newy, 'type') != attr(y, 'type')) stop("New data has a different survival type than the model") } na.action.used <- attr(mf2, 'na.action') } else n2 <- n if (type=="expected") { if (missing(newdata)) pred <- y[,ncol(y)] - object$residuals if (!missing(newdata) || se.fit) { ustrata <- unique(oldstrat) risk <- exp(object$linear.predictors) x <- x - rep(object$means, each=nrow(x)) #subtract from each column if (missing(newdata)) #se.fit must be true se <- double(n) else { pred <- se <- double(nrow(mf2)) newx <- newx - rep(object$means, each=nrow(newx)) newrisk <- c(exp(newx %*% object$coef)) } survtype<- ifelse(object$method=='efron', 3,2) for (i in ustrata) { indx <- which(oldstrat == i) afit <- agsurv(y[indx,,drop=F], x[indx,,drop=F], weights[indx], risk[indx], survtype, survtype) afit.n <- length(afit$time) if (missing(newdata)) { # In this case we need se.fit, nothing else j1 <- approx(afit$time, 1:afit.n, y[indx,1], method='constant', f=0, yleft=0, yright=afit.n)$y chaz <- c(0, afit$cumhaz)[j1 +1] varh <- c(0, cumsum(afit$varhaz))[j1 +1] xbar <- rbind(0, afit$xbar)[j1+1,,drop=F] if (ncol(y)==2) { dt <- (chaz * x[indx,]) - xbar se[indx] <- sqrt(varh + rowSums((dt %*% object$var) *dt)) * risk[indx] } else { j2 <- approx(afit$time, 1:afit.n, y[indx,2], method='constant', f=0, yleft=0, yright=afit.n)$y chaz2 <- c(0, afit$cumhaz)[j2 +1] varh2 <- c(0, cumsum(afit$varhaz))[j2 +1] xbar2 <- rbind(0, afit$xbar)[j2+1,,drop=F] dt <- (chaz * x[indx,]) - xbar v1 <- varh + rowSums((dt %*% object$var) *dt) dt2 <- (chaz2 * x[indx,]) - xbar2 v2 <- varh2 + rowSums((dt2 %*% object$var) *dt2) se[indx] <- sqrt(v2-v1)* risk[indx] } } else { #there is new data use.x <- TRUE indx2 <- which(newstrat == i) j1 <- approx(afit$time, 1:afit.n, newy[indx2,1], method='constant', f=0, yleft=0, yright=afit.n)$y chaz <-c(0, afit$cumhaz)[j1+1] pred[indx2] <- chaz * newrisk[indx2] if (se.fit) { varh <- c(0, cumsum(afit$varhaz))[j1+1] xbar <- rbind(0, afit$xbar)[j1+1,,drop=F] } if (ncol(y)==2) { if (se.fit) { dt <- (chaz * newx[indx2,]) - xbar se[indx2] <- sqrt(varh + rowSums((dt %*% object$var) *dt)) * newrisk[indx2] } } else { j2 <- approx(afit$time, 1:afit.n, newy[indx2,2], method='constant', f=0, yleft=0, yright=afit.n)$y chaz2 <- approx(-afit$time, afit$cumhaz, -newy[indx2,2], method="constant", rule=2, f=0)$y chaz2 <-c(0, afit$cumhaz)[j2+1] pred[indx2] <- (chaz2 - chaz) * newrisk[indx2] if (se.fit) { varh2 <- c(0, cumsum(afit$varhaz))[j1+1] xbar2 <- rbind(0, afit$xbar)[j1+1,,drop=F] dt <- (chaz * newx[indx2,]) - xbar dt2 <- (chaz2 * newx[indx2,]) - xbar2 v2 <- varh2 + rowSums((dt2 %*% object$var) *dt2) v1 <- varh + rowSums((dt %*% object$var) *dt) se[indx2] <- sqrt(v2-v1)* risk[indx2] } } } } } } else { if (is.null(object$coefficients)) coef<-numeric(0) else { # Replace any NA coefs with 0, to stop NA in the linear predictor coef <- ifelse(is.na(object$coefficients), 0, object$coefficients) } if (missing(newdata)) { offset <- offset - mean(offset) if (has.strata && reference=="strata") { # We can't use as.integer(oldstrat) as an index, if oldstrat is # a factor variable with unrepresented levels as.integer could # give 1,2,5 for instance. xmeans <- rowsum(x*weights, oldstrat)/c(rowsum(weights, oldstrat)) newx <- x - xmeans[match(oldstrat,row.names(xmeans)),] } else if (use.x) newx <- x - rep(object$means, each=nrow(x)) } else { offset <- newoffset - mean(offset) if (has.strata && reference=="strata") { xmeans <- rowsum(x*weights, oldstrat)/c(rowsum(weights, oldstrat)) newx <- newx - xmeans[match(newstrat, row.names(xmeans)),] } else newx <- newx - rep(object$means, each=nrow(newx)) } if (type=='lp' || type=='risk') { if (use.x) pred <- drop(newx %*% coef) + offset else pred <- object$linear.predictors if (se.fit) se <- sqrt(rowSums((newx %*% object$var) *newx)) if (type=='risk') { pred <- exp(pred) if (se.fit) se <- se * sqrt(pred) # standard Taylor series approx } } else if (type=='terms') { asgn <- object$assign nterms<-length(asgn) pred<-matrix(ncol=nterms,nrow=NROW(newx)) dimnames(pred) <- list(rownames(newx), names(asgn)) if (se.fit) se <- pred for (i in 1:nterms) { tt <- asgn[[i]] tt <- tt[!is.na(object$coefficients[tt])] xtt <- newx[,tt, drop=F] pred[,i] <- xtt %*% object$coefficient[tt] if (se.fit) se[,i] <- sqrt(rowSums((xtt %*% object$var[tt,tt]) *xtt)) } pred <- pred[,terms, drop=F] if (se.fit) se <- se[,terms, drop=F] attr(pred, 'constant') <- sum(object$coefficients*object$means, na.rm=T) } } if (type != 'terms') { pred <- drop(pred) if (se.fit) se <- drop(se) } if (!is.null(na.action.used)) { pred <- napredict(na.action.used, pred) if (is.matrix(pred)) n <- nrow(pred) else n <- length(pred) if(se.fit) se <- napredict(na.action.used, se) } if (!missing(collapse) && !is.null(collapse)) { if (length(collapse) != n2) stop("Collapse vector is the wrong length") pred <- rowsum(pred, collapse) # in R, rowsum is a matrix, always if (se.fit) se <- sqrt(rowsum(se^2, collapse)) if (type != 'terms') { pred <- drop(pred) if (se.fit) se <- drop(se) } } if (se.fit) list(fit=pred, se.fit=se) else pred } survival/R/model.matrix.coxph.R0000644000175100001440000002020712533653601016217 0ustar hornikusers# Automatically generated from all.nw using noweb # In internal use "data" will often be an already derived model frame. # We detect this via it having a terms attribute. model.matrix.coxph <- function(object, data=NULL, contrast.arg=object$contrasts, ...) { # # If the object has an "x" component, return it, unless a new # data set is given if (is.null(data) && !is.null(object[['x']])) return(object[['x']]) #don't match "xlevels" Terms <- delete.response(object$terms) if (is.null(data)) mf <- model.frame(object) else { if (is.null(attr(data, "terms"))) mf <- model.frame(Terms, data, xlev=object$xlevels) else mf <- data #assume "data" is already a model frame } cluster <- attr(Terms, "specials")$cluster if (length(cluster)) { temp <- untangle.specials(Terms, "cluster") dropterms <- temp$terms } else dropterms <- NULL attr(Terms, "intercept") <- 1 adrop <- 0 #levels of "assign" to be dropped; 0= intercept stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars) > 0) { #if there is a strata statement hasinteractions <- FALSE for (i in stemp$vars) { #multiple strata terms are allowed # The factors att has one row for each variable in the frame, one # col for each term in the model. Pick rows for each strata # var, and find if it participates in any interactions. if (any(attr(Terms, 'order')[attr(Terms, "factors")[i,] >0] >1)) hasinteractions <- TRUE } if (!hasinteractions) dropterms <- c(dropterms, stemp$terms) else adrop <- c(0, match(stemp$var, colnames(attr(Terms, 'factors')))) } if (length(dropterms)) { temppred <- attr(terms, "predvars") Terms2 <- Terms[ -dropterms] if (!is.null(temppred)) { # subscripting a Terms object currently drops predvars, in error attr(Terms2, "predvars") <- temppred[-(1+dropterms)] # "Call" object } X <- model.matrix(Terms2, mf, constrasts=contrast.arg) # we want to number the terms wrt the original model matrix # Do not forget the intercept, which will be a zero renumber <- match(colnames(attr(Terms2, "factors")), colnames(attr(Terms, "factors"))) attr(X, "assign") <- c(0, renumber)[1+attr(X, "assign")] } else X <- model.matrix(Terms, mf, contrasts=contrast.arg) # drop the intercept after the fact, and also drop strata if necessary Xatt <- attributes(X) xdrop <- Xatt$assign %in% adrop #columns to drop (always the intercept) X <- X[, !xdrop, drop=FALSE] attr(X, "assign") <- Xatt$assign[!xdrop] #if (any(adrop>0)) attr(X, "contrasts") <- Xatt$contrasts[-adrop] #else attr(X, "contrasts") <- Xatt$contrasts attr(X, "contrasts") <- Xatt$contrasts X } model.frame.coxph <- function(formula, ...) { dots <- list(...) nargs <- dots[match(c("data", "na.action", "subset", "weights"), names(dots), 0)] # If nothing has changed and the coxph object had a model component, # simply return it. if (length(nargs) ==0 && !is.null(formula$model)) return(formula$model) else { # Rebuild the original call to model.frame Terms <- terms(formula) fcall <- formula$call indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(fcall), nomatch=0) if (indx[1] ==0) stop("The coxph call is missing a formula!") temp <- fcall[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called temp$xlev <- formula$xlevels temp$formula <- Terms #keep the predvars attribute # Now, any arguments that were on this call overtake the ones that # were in the original call. if (length(nargs) >0) temp[names(nargs)] <- nargs # The documentation for model.frame implies that the environment arg # to eval will be ignored, but if we omit it there is a problem. if (is.null(environment(formula$terms))) mf <- eval(temp, parent.frame()) else mf <- eval(temp, environment(formula$terms), parent.frame()) if (!is.null(attr(formula$terms, "dataClasses"))) .checkMFClasses(attr(formula$terms, "dataClasses"), mf) if (!is.null(attr(Terms, "specials")$tt)) { # Do time transform tt <- eval(formula$call$tt) Y <- model.response(mf) strats <- attr(Terms, "specials")$strata if (length(strats)) { stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } timetrans <- untangle.specials(Terms, 'tt') ntrans <- length(timetrans$terms) if (is.null(tt)) { tt <- function(x, time, riskset, weights){ #default to O'Brien's logit rank obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } } if (is.function(tt)) tt <- list(tt) #single function becomes a list if (is.list(tt)) { if (any(!sapply(tt, is.function))) stop("The tt argument must contain function or list of functions") if (length(tt) != ntrans) { if (length(tt) ==1) { temp <- vector("list", ntrans) for (i in 1:ntrans) temp[[i]] <- tt[[1]] tt <- temp } else stop("Wrong length for tt argument") } } else stop("The tt argument must contain a function or list of functions") if (ncol(Y)==2) { if (length(strats)==0) { sorted <- order(-Y[,1], Y[,2]) newstrat <- rep.int(0L, nrow(Y)) newstrat[1] <- 1L } else { sorted <- order(strats, -Y[,1], Y[,2]) #newstrat marks the first obs of each strata newstrat <- as.integer(c(1, 1*(diff(strats[sorted])!=0))) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount1, Y[sorted,], as.integer(newstrat)) tindex <- sorted[counts$index] } else { if (length(strats)==0) { sort.end <- order(-Y[,2], Y[,3]) sort.start<- order(-Y[,1]) newstrat <- c(1L, rep(0, nrow(Y) -1)) } else { sort.end <- order(strats, -Y[,2], Y[,3]) sort.start<- order(strats, -Y[,1]) newstrat <- c(1L, as.integer(diff(strats[sort.end])!=0)) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount2, Y, as.integer(sort.start -1L), as.integer(sort.end -1L), as.integer(newstrat)) tindex <- counts$index } mf <- mf[tindex,] Y <- Surv(rep(counts$time, counts$nrisk), counts$status) type <- 'right' # new Y is right censored, even if the old was (start, stop] strats <- rep(1:length(counts$nrisk), counts$nrisk) weights <- model.weights(mf) if (!is.null(weights) && any(!is.finite(weights))) stop("weights must be finite") for (i in 1:ntrans) mf[[timetrans$var[i]]] <- (tt[[i]])(mf[[timetrans$var[i]]], Y[,1], strats, weights) mf[[".strata."]] <- strats } mf } } survival/R/print.survreg.S0000644000175100001440000000252611732700061015322 0ustar hornikusers# $Id: print.survreg.S 11166 2008-11-24 22:10:34Z therneau $ print.survreg <- function(x, ...) { if(!is.null(cl <- x$call)) { cat("Call:\n") dput(cl) } if (!is.null(x$fail)) { cat(" Survreg failed.", x$fail, "\n") return(invisible(x)) } coef <- x$coef if(any(nas <- is.na(coef))) { if(is.null(names(coef))) names(coef) <- paste("b", 1:length(coef), sep = "") cat("\nCoefficients: (", sum(nas), " not defined because of singularities)\n", sep = "") } else cat("\nCoefficients:\n") print(coef, ...) if (nrow(x$var)==length(coef)) cat("\nScale fixed at",format(x$scale),"\n") else if (length(x$scale)==1) cat ("\nScale=", format(x$scale), "\n") else { cat("\nScale:\n") print(x$scale, ...) } nobs <- length(x$linear) chi <- 2*diff(x$loglik) df <- sum(x$df) - x$idf # The sum is for penalized models cat("\nLoglik(model)=", format(round(x$loglik[2],1)), " Loglik(intercept only)=", format(round(x$loglik[1],1))) if (df > 0) cat("\n\tChisq=", format(round(chi,2)), "on", round(df,1), "degrees of freedom, p=", format(signif(1-pchisq(chi, df),2)), "\n") else cat("\n") omit <- x$na.action if (length(omit)) cat("n=", nobs, " (", naprint(omit), ")\n", sep="") else cat("n=", nobs, "\n") invisible(x) } survival/R/coxph.detail.S0000644000175100001440000000613612470201064015054 0ustar hornikuserscoxph.detail <- function(object, riskmat=FALSE) { method <- object$method if (method!='breslow' && method!='efron') stop(paste("Detailed output is not available for the", method, "method")) n <- length(object$residuals) weights <- object$weights #always present if there are weights x <- object[['x']] y <- object$y strat <- object$strata Terms <- object$terms if (!inherits(Terms, 'terms')) stop("invalid terms component of object") strats <- attr(Terms, "specials")$strata if (is.null(y) || is.null(x)) { mf <- model.frame(object) y <- model.response(mf) x <- model.matrix(object, data=mf) if (length(strats)) { stemp <- untangle.specials(object$terms, 'strata', 1) if (length(stemp$vars)==1) strat <- mf[[stemp$vars]] else strat <- strata(mf[,stemp$vars], shortlabel=TRUE) } } nvar <- ncol(x) if (ncol(y)==2) { mintime <- min(y[,1]) if (mintime < 0) y <- cbind( 2*mintime -1, y) else y <- cbind(-1,y) } if (is.null(strat)) { ord <- order(y[,2], -y[,3]) newstrat <- rep(0,n) } else { ord <- order(strat, y[,2], -y[,3]) newstrat <- c(diff(as.numeric(strat[ord]))!=0 ,1) } newstrat[n] <- 1 # sort the data x <- x[ord,] y <- y[ord,] storage.mode(y) <- 'double' score <- exp(object$linear.predictors)[ord] if (is.null(weights)) weights <- rep(1,n) else weights <- weights[ord] ndeath <- sum(y[,3]) if (riskmat) { rmat <- integer(ndeath*n) } else rmat <- as.integer(1) ff <- .C(Ccoxdetail, as.integer(n), as.integer(nvar), ndeath= as.integer(ndeath), y = y, as.double(x), index = as.integer(newstrat), event2 =as.double(score), weights = as.double(weights), means= c(method=='efron', double(ndeath*nvar-1)), u = double(ndeath*nvar), i = double(ndeath*nvar*nvar), rmat = rmat, nrisk2 = double(ndeath), double(nvar*(3 + 2*nvar))) keep <- 1:ff$ndeath vname<- dimnames(x)[[2]] time <- y[ff$index[keep],2] names(time) <- NULL means<- (matrix(ff$means,ndeath, nvar))[keep,] score<- matrix(ff$u, ndeath, nvar)[keep,] var <- array(ff$i, c(nvar, nvar, ndeath))[,,keep] if (riskmat) { rmat <- matrix(0, n, ff$ndeath) rmat[ord,] <- ff$rmat[1:(n*ff$ndeath)] # in the order of orig data dimnames(rmat) <- list(NULL, time) } if (nvar>1) { dimnames(means) <- list(time, vname) dimnames(score) <- list(time, vname) dimnames(var) <- list(vname, vname, time) } else { names(means) <- time names(score) <- time names(var) <- time } dimnames(ff$y) <- NULL temp <- list(time = time, means=means, nevent=ff$y[keep,1], nrisk = ff$y[keep,2], hazard= ff$y[keep,3], score= score, imat=var, varhaz=ff$weights[keep], y=y, x=x) if (length(strats)) temp$strata <- table((strat[ord])[ff$index[keep]]) if (riskmat) temp$riskmat <- rmat if (!all(weights==1)) { temp$weights <- weights temp$nevent.wt <- ff$event2[keep] temp$nrisk.wt <- ff$nrisk2[keep] } temp } survival/R/residuals.coxph.null.S0000644000175100001440000000053711732700061016556 0ustar hornikusers# $Id $ residuals.coxph.null <- function(object, type=c("martingale", "deviance", "score", "schoenfeld"), collapse=FALSE, weighted=FALSE, ...) { type <- match.arg(type) if (type=='martingale' || type=='deviance') NextMethod() else stop(paste("\'", type, "\' residuals are not defined for a null model", sep="")) } survival/R/model.frame.survreg.R0000644000175100001440000000510711741303442016356 0ustar hornikusersmodel.frame.survreg <- function (formula, ...) { dots <- list(...) nargs <- dots[match(c("data", "na.action", "subset"), names(dots), 0)] if (length(nargs) || is.null(formula$model)) { fcall <- formula$call indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(fcall), nomatch = 0) if (indx[1] == 0) stop("The coxph call is missing a formula!") temp <- fcall[c(1, indx)] temp[[1]] <- as.name("model.frame") temp$xlev <- formula$xlevels if (length(nargs) > 0) temp[names(nargs)] <- nargs if (is.null(environment(formula$terms))) eval(temp, parent.frame()) else eval(temp, environment(formula$terms), parent.frame()) } else formula$model } # model.matrix.survreg <- function(object, data, ...) { if (missing(data) && !is.null(object[["x"]])) object[["x"]] else { Terms <- delete.response(object$terms) strats <- attr(Terms, "specials")$strata cluster<- attr(Terms, "specials")$cluster dropx <- NULL if (length(cluster)) { tempc <- untangle.specials(Terms, 'cluster', 1:10) dropx <- tempc$terms } if (length(strats)) { temp <- untangle.specials(Terms, 'strata', 1) dropx <- c(dropx, temp$terms) } if (length(dropx)) { newTerms <- Terms[-dropx] # R (version 2.7.1) adds intercept=T anytime you drop something attr(newTerms, 'intercept') <- attr(Terms, 'intercept') # The predvars attribute, if present, is lost when we # subscript. The attribute is a Call, so has one more element # than term wrt subscripting, i.e., the called function "list" if (!is.null(attr(terms, "predvars"))) attr(newTerms, "predvars") <- attr(terms, "predvars")[-(dropx+1)] } else newTerms <- Terms # Grab the model frame. By using "newterms" for a new data set, # we allow the new data to be missing things we don't need: y, # strata, and cluster. For the original data we can assume they # are present. if (missing(data)) mf <- model.frame(object, ...) else { if (is.null(attr(data, "terms"))) mf <- model.frame(newTerms, data, xlev=object$xlevels) else mf <- data #assume we were given a model frame } model.matrix(newTerms, mf, contrasts.arg= object$contrasts) } } survival/R/survreg.S0000644000175100001440000002301312470201064014160 0ustar hornikusersif (!is.R()) setOldClass(c('survreg.penal', 'survreg')) survreg <- function(formula, data, weights, subset, na.action, dist='weibull', init=NULL, scale=0, control, parms=NULL, model=FALSE, x=FALSE, y=TRUE, robust=FALSE, score=FALSE, ...) { Call <- match.call() # save a copy of the call indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(Call), nomatch=0) if (indx[1] ==0) stop("A formula argument is required") temp <- Call[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called special <- c("strata", "cluster") temp$formula <- if(missing(data)) terms(formula, special) else terms(formula, special, data=data) if (is.R()) m <- eval(temp, parent.frame()) else m <- eval(temp, sys.parent()) Terms <- attr(m, 'terms') weights <- model.extract(m, 'weights') Y <- model.extract(m, "response") if (!inherits(Y, "Surv")) stop("Response must be a survival object") strats <- attr(Terms, "specials")$strata cluster<- attr(Terms, "specials")$cluster dropx <- NULL if (length(cluster)) { if (missing(robust)) robust <- TRUE tempc <- untangle.specials(Terms, 'cluster', 1:10) ord <- attr(Terms, 'order')[tempc$terms] if (any(ord>1)) stop ("Cluster can not be used in an interaction") cluster <- strata(m[,tempc$vars], shortlabel=TRUE) #allow multiples dropx <- tempc$terms } if (length(strats)) { temp <- untangle.specials(Terms, 'strata', 1) dropx <- c(dropx, temp$terms) if (length(temp$vars)==1) strata.keep <- m[[temp$vars]] else strata.keep <- strata(m[,temp$vars], shortlabel=TRUE) strata <- as.numeric(strata.keep) nstrata <- max(strata) } else { nstrata <- 1 strata <- 0 } if (length(dropx)) { newTerms <- Terms[-dropx] # R (version 2.7.1) adds intercept=T anytime you drop something if (is.R()) attr(newTerms, 'intercept') <- attr(Terms, 'intercept') } else newTerms <- Terms X <- model.matrix(newTerms, m) if (is.R()) { assign <- lapply(attrassign(X, newTerms)[-1], function(x) x-1) xlevels <- .getXlevels(newTerms, m) contr.save <- attr(X, 'contrasts') } else { assign <- lapply(attr(X, 'assign')[-1], function(x) x -1) xvars <- as.character(attr(newTerms, 'variables')) xvars <- xvars[-attr(newTerms, 'response')] if (length(xvars) >0) { xlevels <- lapply(m[xvars], levels) xlevels <- xlevels[!unlist(lapply(xlevels, is.null))] if(length(xlevels) == 0) xlevels <- NULL } else xlevels <- NULL contr.save <- attr(X, 'contrasts') } n <- nrow(X) nvar <- ncol(X) offset<- model.offset(m) # R returns NULL if no offset, Splus a zero if (length(offset)==0 || all(offset==0)) offset <- rep(0.,n) type <- attr(Y, "type") if (type== 'counting') stop ("Invalid survival type") # The user can either give a distribution name, in which the distribution # is found in the object survreg.distributions, or include a list object # of the same format as is found there. if (is.character(dist)) { # partial matching of names in [[ is on its way out in R, so # first use match.arg, e.g. turn 'exp' into 'exponential' dist <- match.arg(dist, names(survreg.distributions)) dlist <- survreg.distributions[[dist]] if (is.null(dlist)) stop(paste(dist, ": distribution not found")) } else if (is.list(dist)) dlist <- dist else stop("Invalid distribution object") # # Make sure it is legal # if (!survregDtest(dlist)) stop("Invalid distribution object") # If the distribution is a transformation of another, perform # said transform. # logcorrect <- 0 #correction to the loglik due to transformations if (!is.null(dlist$trans)) { tranfun <- dlist$trans exactsurv <- Y[,ncol(Y)] ==1 if (any(exactsurv)) { if (is.null(weights)) logcorrect <- sum(log(dlist$dtrans(Y[exactsurv, 1]))) else logcorrect <- sum(weights[exactsurv]*log(dlist$dtrans(Y[exactsurv, 1]))) } if (type=='interval') { if (any(Y[,3]==3)) Y <- cbind(tranfun(Y[,1:2]), Y[,3]) else Y <- cbind(tranfun(Y[,1]), Y[,3]) } else if (type=='left') Y <- cbind(tranfun(Y[,1]), 2-Y[,2]) else Y <- cbind(tranfun(Y[,1]), Y[,2]) if (!all(is.finite(Y))) stop("Invalid survival times for this distribution") } else { if (type=='left') Y[,2] <- 2- Y[,2] else if (type=='interval' && all(Y[,3]<3)) Y <- Y[,c(1,3)] } if (!is.null(dlist$scale)) { if (!missing(scale)) warning(paste(dlist$name, "has a fixed scale, user specified value ignored")) scale <- dlist$scale } if (!is.null(dlist$dist)) if (is.atomic(dlist$dist)) dlist <- survreg.distributions[[dlist$dist]] else dlist <- dlist$dist # check for parameters ptemp <- dlist$parms if (is.null(ptemp)) { if (!is.null(parms)) stop(paste(dlist$name, "distribution has no optional parameters")) } else { if (!is.numeric(ptemp)) stop("Default parameters must be a numeric vector") if (!missing(parms)) { temp <- unlist(parms) # just in case they gave a list object indx <- match(names(temp), names(ptemp)) if (any(is.na(indx))) stop("Invalid parameter names") ptemp[names(ptemp)] <- temp } parms <- ptemp } # An idea originally from Brian R: if the user gave a list of # control values, use it, but if they did not give an explicit control # argument assume that they mistakenly wrote control parameters as a # part of the "..." or other arguments if (missing(control)) control <- survreg.control(...) else control <- do.call('survreg.control', control) # The any() construction below is to catch a user that mistakenly # thinks that 'scale' can be used in a model with multiple strata, and # so provided a vector of scale values. # (A 'perhaps should be be added someday' feature). if (any(scale < 0)) stop("Invalid scale value") if (any(scale >0) && nstrata >1) stop("The scale argument is not valid with multiple strata") # Check for penalized terms pterms <- sapply(m, inherits, 'coxph.penalty') if (any(pterms)) { pattr <- lapply(m[pterms], attributes) # # the 'order' attribute has the same components as 'term.labels' # pterms always has 1 more (response), sometimes 2 (offset) # drop the extra parts from pterms temp <- c(attr(Terms, 'response'), attr(Terms, 'offset')) if (length(dropx)) temp <- c(temp, dropx+1) pterms <- pterms[-temp] temp <- match((names(pterms))[pterms], attr(Terms, 'term.labels')) ord <- attr(Terms, 'order')[temp] if (any(ord>1)) stop ('Penalty terms cannot be in an interaction') if (is.R()) assign <- attrassign(X, newTerms) else assign <- attr( X, 'assign') pcols <- assign[match(names(pterms[pterms]), names(assign))] fit <- survpenal.fit(X, Y, weights, offset, init=init, controlvals = control, dist= dlist, scale=scale, strata=strata, nstrat=nstrata, pcols, pattr, parms=parms, assign) } else fit <- survreg.fit(X, Y, weights, offset, init=init, controlvals=control, dist= dlist, scale=scale, nstrat=nstrata, strata, parms=parms) if (is.character(fit)) fit <- list(fail=fit) #error message else { if (scale==0) { nvar <- length(fit$coefficients) - nstrata fit$scale <- exp(fit$coefficients[-(1:nvar)]) if (nstrata==1) names(fit$scale) <- NULL else names(fit$scale) <- levels(strata.keep) fit$coefficients <- fit$coefficients[1:nvar] fit$idf <- 1 + nstrata } else { fit$scale <- scale fit$idf <- 1 } fit$loglik <- fit$loglik + logcorrect } if (!score) fit$score <- NULL #do not return the score vector fit$df.residual <- n - sum(fit$df) fit$terms <- Terms fit$contrasts <- contr.save if (length(xlevels)) fit$xlevels <- xlevels fit$means <- apply(X,2, mean) if (!is.null(weights)) fit$weights <- weights fit$call <- Call fit$dist <- dist if (model) fit$model <- m if (x) fit$x <- X if (y) fit$y <- Y if (length(parms)) fit$parms <- parms # Do this before attaching the na.action, so that residuals() won't # reinsert missing values under na.exclude if (robust) { fit$naive.var <- fit$var if (!model) fit$model <- m #temporary addition, so resid doesn't # have to reconstruct if (length(cluster)) fit$var <- crossprod(rowsum(residuals.survreg(fit, 'dfbeta'), cluster)) else fit$var <- crossprod(residuals.survreg(fit, 'dfbeta')) if (!model) fit$model <- NULL # take it back out } na.action <- attr(m, "na.action") if (length(na.action)) fit$na.action <- na.action if (is.R()) { if (any(pterms)) class(fit) <- c('survreg.penal', 'survreg') else class(fit) <- 'survreg' } else { if (any(pterms)) oldClass(fit) <- 'survreg.penal' else oldClass(fit) <- 'survreg' } fit } survival/R/summary.coxph.penal.S0000644000175100001440000000757512257335007016425 0ustar hornikuserssummary.coxph.penal <- function(object, conf.int = 0.95, scale=1, terms=FALSE, maxlabel=25, ...) { beta <- object$coefficients if (length(beta)==0 && length(object$frail)==0) stop("Penalized summary function can't be used for a null model") if (length(beta) > 0) { #has non-penalized coefs nacoef <- !(is.na(beta)) #non-missing coefs beta2 <- beta[nacoef] if(is.null(beta2) | is.null(object$var)) stop("Input is not valid") se <- sqrt(diag(object$var)) } # # Map terms to special print functions, and the list of iteration histories # pterms <- object$pterms nterms <- length(pterms) npenal <- sum(pterms>0) print.map <- rep(0,nterms) if (!is.null(object$printfun)) { temp <- unlist(lapply(object$printfun, is.null)) #which ones are missing print.map[pterms>0] <- (1:npenal) * (!temp) } # Tedious, but build up the coef matrix a term at a time print1 <- NULL pname1 <- NULL if (is.null(object$assign2)) alist <- object$assign[-1] else alist <- object$assign2 print2 <- NULL for (i in 1:nterms) { kk <- alist[[i]] if (print.map[i] >0) { j <- print.map[i] if (pterms[i]==2) temp <- (object$printfun[[j]])(object$frail, object$fvar, , object$df[i], object$history[[j]]) else temp <- (object$printfun[[j]])(beta[kk], object$var[kk,kk], object$var2[kk,kk], object$df[i], object$history[[j]]) print1 <- rbind(print1, temp$coef) if (is.matrix(temp$coef)) { xx <- dimnames(temp$coef)[[1]] if (is.null(xx)) xx <- rep(names(pterms)[i], nrow(temp$coef)) else xx <- paste(names(pterms)[i], xx, sep=', ') pname1 <- c(pname1, xx) } else pname1 <- c(pname1, names(pterms)[i]) print2 <- c(print2, temp$history) } else if (terms && length(kk)>1) { pname1 <- c(pname1, names(pterms)[i]) temp <- coxph.wtest(object$var[kk,kk], beta[kk])$test print1 <- rbind(print1, c(NA, NA, NA, temp, object$df[i], 1-pchisq(temp, 1))) } else { pname1 <- c(pname1, names(beta)[kk]) tempe<- (diag(object$var))[kk] temp <- beta[kk]^2/ tempe print1 <- rbind(print1, cbind(beta[kk], sqrt(tempe), sqrt((diag(object$var2))[kk]), temp, 1, 1-pchisq(temp, 1))) } } dimnames(print1) <- list(substring(pname1,1, maxlabel), c("coef","se(coef)", "se2", "Chisq","DF","p")) rval <- object[match(c("call", "fail", "na.action", "n", "nevent", "loglik", "iter", "df"), names(object), nomatch=0)] rval$coefficients <- print1 rval$print2 <- print2 if(conf.int & length(beta) >0 ) { z <- qnorm((1 + conf.int)/2, 0, 1) beta <- beta * scale se <- se * scale tmp <- cbind(exp(beta), exp(-beta), exp(beta - z * se), exp(beta + z * se)) dimnames(tmp) <- list(substring(names(beta),1, maxlabel), c("exp(coef)", "exp(-coef)", paste("lower .", round(100 * conf.int, 2), sep = ""), paste("upper .", round(100 * conf.int, 2), sep = ""))) rval$conf.int <- tmp } df <- sum(object$df) logtest <- -2 * (object$loglik[1] - object$loglik[2]) rval$logtest <- c(test = logtest, df=df, pvalue= pchisq(logtest,df, lower.tail=FALSE)) if (!is.null(object$waldtest)) rval$waldtest <- c(test= object$wald.test, df=df, pvalue = pchisq(object$wald.test, df, lower.tail=FALSE)) if (!is.null(object$concordance)) { if (is.matrix(object$concordance)) temp <- colSums(object$concordance) else temp <- object$concordance rval$concordance <- c((temp[1] + temp[3]/2)/ sum(temp[1:3]), temp[5]/(2*sum(temp[1:3]))) names(rval$concordance) <- c("concordance", "se") } class(rval) <- "summary.coxph.penal" rval } survival/R/frailty.brent.S0000644000175100001440000000327511732700061015257 0ustar hornikusers# $Id: frailty.brent.S 11166 2008-11-24 22:10:34Z therneau $ # # Brent's method for finding a maximum # If upper and/or lower is given, it transforms x to stay out of trouble # during the "bracketing" phase # frailty.brent <- function(x, y, lower, upper) { n <- length(x) if (length(y) != n) stop ("Length mismatch for x and y") if (n<3) return(mean(x)) # First, is the solution bracketed? # If not, take big steps until it is ord <- order(x) xx <- x[ord] yy <- y[ord] best <- (1:n)[yy==max(y)] if (length(best) >1) stop("Ties for max(y), I surrender") #fix this later if (best==1) { new <- xx[1] - 3*(xx[2] - xx[1]) if (!missing(lower) && !is.null(lower) && new < lower) new <- lower + (min(xx[xx>lower])-lower)/10 return(new) } if (best==n) { new <- xx[n] + 3*(xx[n] - xx[n-1]) if (!missing(upper) && !is.null(upper) && new > upper) new <- upper + (max(xx[xx xx[3] || ( (n>4) && (new-x[n]) > .5*abs(x[n-1]-x[n-2]))) { if ((xx[2]-xx[1]) > (xx[3]-xx[2])) return(xx[2] - .38*(xx[2]-xx[1])) else return(xx[2] + .32*(xx[3]-xx[2])) } else return(new) } survival/R/plot.cox.zph.S0000644000175100001440000000466011732700061015041 0ustar hornikusers# $Id: plot.cox.zph.S 11275 2009-04-06 16:18:00Z therneau $ plot.cox.zph <- function(x, resid=TRUE, se=TRUE, df=4, nsmo=40, var, ...) { xx <- x$x yy <- x$y d <- nrow(yy) df <- max(df) #error proofing nvar <- ncol(yy) pred.x <- seq(from=min(xx), to=max(xx), length=nsmo) temp <- c(pred.x, xx) lmat <- ns(temp, df=df, intercept=TRUE) pmat <- lmat[1:nsmo,] # for prediction xmat <- lmat[-(1:nsmo),] qmat <- qr(xmat) if (qmat$rank < df) stop("Spline fit is singular, try a smaller degrees of freedom") if (se) { bk <- backsolve(qmat$qr[1:df, 1:df], diag(df)) xtx <- bk %*% t(bk) seval <- d*((pmat%*% xtx) *pmat) %*% rep(1, df) } ylab <- paste("Beta(t) for", dimnames(yy)[[2]]) if (missing(var)) var <- 1:nvar else { if (is.character(var)) var <- match(var, dimnames(yy)[[2]]) if (any(is.na(var)) || max(var)>nvar || min(var) <1) stop("Invalid variable requested") } # # Figure out a 'good' set of x-axis labels. Find 8 equally spaced # values on the 'transformed' axis. Then adjust until they correspond # to rounded 'true time' values. Avoid the edges of the x axis, or # approx() may give a missing value if (x$transform == 'log') { xx <- exp(xx) pred.x <- exp(pred.x) } else if (x$transform != 'identity') { xtime <- as.numeric(dimnames(yy)[[1]]) indx <- !duplicated(xx) #avoid a warning message in R apr1 <- approx(xx[indx], xtime[indx], seq(min(xx), max(xx), length=17)[2*(1:8)]) temp <- signif(apr1$y,2) apr2 <- approx(xtime[indx], xx[indx], temp) xaxisval <- apr2$y xaxislab <- rep("",8) for (i in 1:8) xaxislab[i] <- format(temp[i]) } for (i in var) { y <- yy[,i] yhat <- pmat %*% qr.coef(qmat, y) if (resid) yr <-range(yhat, y) else yr <-range(yhat) if (se) { temp <- 2* sqrt(x$var[i,i]*seval) yup <- yhat + temp ylow<- yhat - temp yr <- range(yr, yup, ylow) } if (x$transform=='identity') plot(range(xx), yr, type='n', xlab="Time", ylab=ylab[i], ...) else if (x$transform=='log') plot(range(xx), yr, type='n', xlab="Time", ylab=ylab[i], log='x', ...) else { plot(range(xx), yr, type='n', xlab="Time", ylab=ylab[i], axes=FALSE,...) axis(1, xaxisval, xaxislab) axis(2) box() } if (resid) points(xx, y) lines(pred.x, yhat) if (se) { lines(pred.x, yup,lty=2) lines(pred.x, ylow, lty=2) } } } survival/R/survexp.fit.S0000644000175100001440000001104612111735600014763 0ustar hornikusers# Actually compute the expected survival for one or more cohorts # of subjects. If each subject is his/her own group, it gives individual # survival # group = groups (one curve per group) # x matrix contains the rate # table indices = starting point for each obs in the rate table. # y is the number of follow-up days for each subject # times = the time points at which survival is desired # death = T if we want the conditional estimate survexp.fit <- function(group, x, y, times, death, ratetable) { if (!is.matrix(x)) stop("x must be a matrix") if (ncol(x) != length(dim(ratetable))) stop("x matrix does not match the rate table") atts <- attributes(ratetable) ngrp <- max(group) times <- sort(unique(times)) if (any(times <0)) stop("Negative time point requested") if (missing(y)) y <- rep(max(times), nrow(x)) ntime <- length(times) if (!is.logical(death)) stop("Invalid value for death indicator") cuts <- atts$cutpoints if (is.null(atts$type)) { # old style rate table rfac <- atts$factor us.special <- (rfac >1) } else { rfac <- 1*(atts$type ==1) us.special <- (atts$type==4) } if (any(us.special)) { #special handling for US pop tables if (sum(us.special) >1) stop("Two columns marked for special handling as a US rate table") # Now, the 'entry' date on a US rate table is the number of days # since 1/1/1960, and the user data has been aligned to the # same system by match.ratetable and marked as "year". # US rate tables are odd: the entry for age (year=1970, age=55) # contains the daily rate for anyone who turns 55 in that year, # from their birthday forward for 365 days. So if your birthday # is on Oct 2, the 1970 table applies from 2Oct 1970 to 1Oct 1971. # The underlying C code wants to make the 1970 rate table apply # from 1Jan 1970 to 31Dec 1970. The easiest way to finess this is # to fudge everyone's enter-the-study date. If you were born # in March but entered in April, make it look like you entered in # Febuary; that way you get the first 11 months at the entry # year's rates, etc. This is the same as being born on Jan 1. # The birth date is entry date - age in days (based on 1/1/1960). # cols <- match(c("age", "year"), atts$dimid) if (any(is.na(cols))) stop("Ratetable does not have expected shape") if (exists("as.Date")) { # true for modern version of R bdate <- as.Date('1960/1/1') + (x[,cols[2]] - x[,cols[1]]) byear <- format(bdate, "%Y") # year of birth offset <- as.numeric(bdate - as.Date(paste(byear, '01/01', sep='/'))) } # The lines below were commented out to stop spurious warning # messages from "CMD check". They are very unlikely to ever # be needed, so no big loss. #else if (exists('month.day.year')) { # Splus, usually # bdate <- x[,cols[2]] - x[,cols[1]] # byear <- month.day.year(bdate)$year # offset <- bdate - julian(1,1,byear) # } #else if (exists('date.mdy')) { # the TMT date class is available # bdate <- as.date(x[,cols[2]] - x[,cols[1]]) # byear <- date.mdy(bdate)$year # offset <- bdate - mdy.date(1,1,byear) # } else stop("Can't find an appropriate date class\n") x[,cols[2]] <- x[,cols[2]] - offset # Doctor up "cutpoints" - only needed for old style rate tables # for which the C code does interpolation on the fly if (any(rfac >1)) { temp <- which(us.special) nyear <- length(cuts[[temp]]) nint <- rfac[temp] #intervals to interpolate over cuts[[temp]] <- round(approx(nint*(1:nyear), cuts[[temp]], nint:(nint*nyear))$y - .0001) } } storage.mode(x) <- storage.mode(y) <- "double" storage.mode(times) <- "double" temp <- .Call(Cpyears3b, as.integer(death), as.integer(rfac), as.integer(atts$dim), as.double(unlist(cuts)), ratetable, as.integer(group), x, y, times, as.integer(ngrp)) if (ntime==1) list(surv=temp$surv, n=temp$n) else if (ngrp >1) list(surv=apply(matrix(temp$surv, ntime, ngrp),2,cumprod), n= matrix(temp$n, ntime, ngrp)) else list(surv=cumprod(temp$surv), n=temp$n) } survival/R/coxph.rvar.S0000644000175100001440000000112111732700061014552 0ustar hornikusers# $Id: coxph.rvar.S 11166 2008-11-24 22:10:34Z therneau $ coxph.rvar <- function(fit, collapse) { rcall <- match.call() if (class(fit) != 'coxph') stop ("First argument must be a fitted Cox model") if (missing(collapse)) temp <- residuals.coxph(fit, type='dfbeta') else temp <- residuals.coxph(fit, type='dfbeta', collapse=collapse) if (any(is.na(temp))) if (ncol(temp)==1) temp<- temp[!is.na(temp),,drop=FALSE] else temp <- temp[!is.na(temp %*% rep(1,ncol(temp))),] fit$robust.var <- t(temp) %*% temp fit$rcall <- rcall fit } survival/R/print.coxph.null.S0000644000175100001440000000065211732700061015715 0ustar hornikusers# $Id: print.coxph.null.S 11166 2008-11-24 22:10:34Z therneau $ print.coxph.null <- function(x, digits=max(options()$digits - 4, 3), ...) { if (!is.null(cl<- x$call)) { cat("Call: ") dput(cl) cat("\n") } cat("Null model\n log likelihood=", format(x$loglik), "\n") omit <- x$na.action if (length(omit)) cat(" n=", x$n, " (", naprint(omit), ")\n", sep="") else cat(" n=", x$n, "\n") } survival/R/ratetableDate.S0000644000175100001440000000213411737410271015234 0ustar hornikusers# # survexp/pyears ratetables keep all dates as number of days since 1/1/1960 # convert other types of objects to this form # ratetableDate <- function(x) { UseMethod("ratetableDate", x) } # Normally used in R ratetableDate.Date <- function(x) as.numeric(x - as.Date("1960/01/01")) ratetableDate.POSIXt <- function(x) as.numeric(as.Date(x) - as.Date("1960/01/01")) # Normally Splus #ratetableDate.timeDate <- function(x) # as.numeric(x - timeDate('1/1/1960')) # Therneau's old "date" class (will someday wither away) ratetableDate.date <- function(x) as.numeric(x) # David James's old "chron" class (will someday wither away) # Support it without using the chron library, which may not be loaded. ratetableDate.chron <- function(x) { origin <- attr(x, "origin") x<- as.numeric(x) + as.Date(paste(origin["year"], origin["month"], origin["day"], sep='/')) ratetableDate(x) } ratetableDate.dates <- ratetableDate.chron # the routines that call this are responsible for a useful error message ratetableDate.default <- function(x) NULL survival/R/anova.survreg.S0000644000175100001440000000377411732700061015300 0ustar hornikusers# $Id: anova.survreg.S 11230 2009-02-09 23:37:55Z therneau $ anova.survreg <- function(object, ..., test = c("Chisq", "none")) { test <- match.arg(test) margs <- function(...) nargs() if(margs(...)) return(anova.survreglist(list(object, ...), test = test)) Terms <- object$terms term.labels <- attr(Terms, "term.labels") nt <- length(term.labels) m <- model.frame(object) family.obj <- object$family y <- model.extract(m, "response") if(!inherits(y, "Surv")) stop("Response must be a survival object") loglik <- numeric(nt + 1) df.res <- loglik if(nt) { loglik[nt + 1] <- -2 * object$loglik[2] df.res[nt + 1] <- object$df.residual fit <- object for(iterm in seq(from = nt, to = 1, by = -1)) { argslist <- list(object = fit, formula = eval(parse(text = paste("~ . -", term.labels[iterm])))) fit <- do.call("update", argslist) loglik[iterm] <- -2 * fit$loglik[2] df.res[iterm] <- fit$df.residual } dev <- c(NA, - diff(loglik)) df <- c(NA, -diff(df.res)) } else { loglik[1] <- -2 * object$loglik[2] df.res[1] <- object$df.residual #dim(y)[1] - attr(Terms, "intercept") dev <- df <- as.numeric(NA) } heading <- c("Analysis of Deviance Table\n", paste(family.obj[1], "distribution with", family.obj[2], "link\n"), paste("Response: ", as.character(formula(object))[2], "\n", sep = ""), if (nrow(fit$var) == length(fit$coefficients)) paste("Scale fixed at", format(object$scale, digits = getOption("digits")),"\n") else "Scale estimated\n", "Terms added sequentially (first to last)") aod <- data.frame(Df = df, Deviance = dev, "Resid. Df" = df.res, "-2*LL" = loglik, row.names = c("NULL", term.labels), check.names = FALSE) attr(aod, "heading") <- heading class(aod) <- c("anova", "data.frame") if(test == "none") return(aod) else stat.anova(aod, test, scale=1 ,n= nrow(y)) } survival/R/survexp.cfit.R0000644000175100001440000001063312257335007015136 0ustar hornikusers# Do expected survival based on a Cox model # This version relies on the survfit routine to do most of # the work. survexp.cfit <- function(group, ndata, y, method, coxfit, weights) { # If it is individual survival, call the predict method if (method=='individual') { temp <- predict(coxfit, newdata=ndata, type='expect', se=FALSE) return(list(surv= exp(-temp))) } # Get the set of survival curves on which I'll base my work # There is no id statement allowed yet, so no survexp for time-dependent # covariates sfit <- survfit.coxph(coxfit, newdata=ndata, se.fit=FALSE, censor=FALSE) # rare case: someone called survexp with a single-obs newdata # The average of n curves is just the curve, when n=1 if (length(group)==1) return(sfit) # number of curves to create & number of subjects ncurve <- max(group) #group was preset to contain integer group number n <- length(group) # matches nrow(ndata) # If the Cox model had strata then the newdata object also had to contain # the strata (needed to fully identify the new subjects), and the # n survival curves will be "strung out" as a single surv vector in # sfit, along with a strata component saying how many points for each. # If the Cox model did not have strata, sfit$surv and sfit$cumhaz will be # matrices with n columns. # The output should be a list with components time, n, and surv. # time = vector of unique time points # surv = matrix with 1 column per created curve (often just 1) # n = same shape as surv, containing the number of obs from ndata # that contribute to each row. newtime <- sort(unique(sfit$time)) # all of the unique times ntime <- length(newtime) newsurv <- list(time=newtime) # Each row of the input data is part of one and only one of the output # curves. Each column of gmat will contain the weights we need. # Each col sums to 1, and has zeros for those who belong to another curve gmat <- matrix(0., nrow=n, ncol=ncurve) for (i in 1:ncurve) { temp <- weights[group==i] gmat[group==i, i] <- temp/sum(temp) } # If the result is a set of curves with strata rather than a matrix, we # need to index into it, using a code trick taken from summary.survfit # Note that is is possible (though odd) for someone to specify a population # of subjects in survexp whose individual members come from different # strata in sfit. The result curves could have any of the times # that appear in any stratum. So we create a regular matrix of survivals. if (is.null(sfit$strata)) ssurv <- sfit$surv else { ssurv <- matrix(0., nrow=ntime, ncol=n) indx <- rep(1:length(sfit$strata), sfit$strata) for (i in 1:n) { itemp <- which(indx==i) ssurv[,i] <- approx(sfit$time[itemp], sfit$surv[itemp], newtime, yleft=0, method="constant", f=0, rule=2)$y } } if (method=="ederer") { # This is the most common call. We can work directly # with the returned survival curves, taking weighed averages. newsurv$n <- matrix(rep(table(group), each=ntime), nrow=ntime) newsurv$surv <- ssurv %*% gmat } else { # These are rarely used, so are implemented in S code rather than # C, even though it involves a loop over time points. # We need the hazard at each of the new time points, from which # a weighted average at each time point is computed # the Hakulinen also the survival at each time point. hazard <- apply(rbind(0, sfit$cumhaz), 2, diff) cmat <- matrix(0, ntime, ncurve) # Holds the result if (method== "conditional") { for (i in 1:ntime) { tmat <- ifelse(y >= newtime[i],1,0) * gmat #zero if not at risk cmat[i,] <- hazard[i,] %*% tmat / colSums(tmat) } } else { #Hakulinen method # Weights in this case are S(newtime) * I(newtime >=y) * gmat lsurv <- rbind(1.0, ssurv) #right continuous time for (i in 1:ntime) { tmat <- (ifelse(y>=newtime[i],1,0) * lsurv[i,]) * gmat cmat[i,] <- hazard[i,] %*% tmat / colSums(tmat) } } newsurv$surv <- exp(-apply(cmat, 2, cumsum)) } newsurv } survival/R/frailty.t.S0000644000175100001440000001002112444044435014402 0ustar hornikusers# $Id: frailty.t.S 11377 2009-12-14 22:59:56Z therneau $ # # Defining function for t-distribution frailty fits # frailty.t <- function(x, sparse=(nclass>5), theta, df, eps= 1e-5, tdf=5, method=c("aic", "df", "fixed"), ...) { nclass <- length(unique(x[!is.na(x)])) if (sparse){ x <-as.numeric(factor(x)) if (is.R()) class(x) <- "coxph.penalty" else oldClass(x) <- "coxph.penalty" } else{ x <- factor(x) if (is.R()) class(x) <- c("coxph.penalty",class(x)) else oldClass(x) <- "coxph.penalty" attr(x,'contrasts') <- contr.treatment(nclass, contrasts=FALSE) } if (tdf <=2) stop("Cannot have df <3 for the t-frailty") # Check for consistency of the arguments if (missing(method)) { if (!missing(theta)) { method <- 'fixed' if (!missing(df)) stop("Cannot give both a df and theta argument") } else if (!missing(df)) { if (df==0) method <- 'aic' else method <- 'df' } } method <- match.arg(method) if (method=='df' && missing(df)) stop("Method = df but no df argument") if (method=='fixed' && missing(theta)) stop("Method= fixed but no theta argument") if (method !='fixed' && !missing(theta)) stop("Method is not 'fixed', but have a theta argument") pfun<- function(coef, theta, ndead, tdf){ if (theta==0) list(recenter=0, penalty=0, flag=TRUE) else { sig <- theta* (tdf-2)/tdf #scale contant^2 in density formula # # Find the centering constant, using 1 NR step # temp <- 1 + coef^2/(tdf*sig) temp1 <- coef/temp temp2 <- 1/temp - (2/(tdf*sig))*coef^2/temp^2 recenter <- sum(temp1)/sum(temp2) #NR step towards MLE coef <- coef - recenter const <- (tdf+1)/(tdf*sig) temp <- 1 + coef^2/(tdf*sig) list(recenter=recenter, first= const*coef/temp, second= const*(1/temp - (2/(tdf*sig))*coef^2/temp^2), penalty= sum(.5*log(pi*tdf*sig) + ((tdf+1)/2)*log(temp) + lgamma(tdf/2) - lgamma((tdf+1)/2)), flag=FALSE) } } printfun <- function(coef, var, var2, df, history) { if (!is.null(history$history)) theta <- history$history[nrow(history$history),1] else theta <- history$theta if (is.matrix(var)) test <- coxph.wtest(var, coef)$test else test <- sum(coef^2/var) df2 <- max(df, .5) # Stop silly p-values list(coef=c(NA, NA, NA, test, df, 1-pchisq(test, df2)), history=paste("Variance of random effect=", format(theta))) } # The final coxph object will contain a copy of printfun. Stop it from # also containing huge unnecessary variables, e.g. 'x', known at this # point in time. Not an issue for pfun, which does not get saved. # The reason for using the survival namespace instead of globalenv() is # that we call coxph.wtest, which may not be visible outside the name space if (is.R()) environment(printfun) <- asNamespace('survival') if (method=='fixed') { temp <- list(pfun=pfun, pparm=tdf, printfun=printfun, diag =TRUE, sparse= sparse, cfun = function(parms, iter, old){ list(theta=parms$theta, done=TRUE)}, cparm= list(theta=theta, ...)) } else if (method=='aic') { temp <- list(pfun=pfun, pparm=tdf, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c("neff", "df", "plik"), cparm=list(lower=0, init=c(.1,1), eps=eps, ...), cfun = frailty.controlaic) } else { #df method if (missing(eps)) eps <- .1 temp <- list(pfun=pfun, pparm=tdf, printfun=printfun, diag =TRUE, sparse= sparse, cargs= c('df'), cparm=list(df=df, eps=eps, thetas=0, dfs=0, guess=3*df/length(unclass(x)), ...), cfun = frailty.controldf) } # If not sparse, give shorter names to the coefficients, so that any # printout of them is readable. if (!sparse) { vname <- paste("t", levels(x), sep=':') temp <- c(temp, list(varname=vname)) } attributes(x) <- c(attributes(x), temp) x } survival/R/residuals.coxph.S0000644000175100001440000001203412113164602015577 0ustar hornikusersresiduals.coxph <- function(object, type=c("martingale", "deviance", "score", "schoenfeld", "dfbeta", "dfbetas", "scaledsch","partial"), collapse=FALSE, weighted=FALSE, ...) { type <- match.arg(type) otype <- type if (type=='dfbeta' || type=='dfbetas') { type <- 'score' if (missing(weighted)) weighted <- TRUE # different default for this case } if (type=='scaledsch') type<-'schoenfeld' n <- length(object$residuals) rr <- object$residuals y <- object$y x <- object[['x']] # avoid matching object$xlevels vv <- object$naive.var if (is.null(vv)) vv <- object$var weights <- object$weights if (is.null(weights)) weights <- rep(1,n) strat <- object$strata method <- object$method if (method=='exact' && (type=='score' || type=='schoenfeld')) stop(paste(type, 'residuals are not available for the exact method')) if (type == 'martingale' || type == 'partial') rr <- object$residuals else { # I need Y, and perhaps the X matrix (and strata) Terms <- object$terms if (!inherits(Terms, 'terms')) stop("invalid terms component of object") strats <- attr(Terms, "specials")$strata if (is.null(y) || (is.null(x) && type!= 'deviance')) { temp <- coxph.getdata(object, y=TRUE, x=TRUE, stratax=TRUE) y <- temp$y x <- temp$x if (length(strats)) strat <- temp$strata } ny <- ncol(y) status <- y[,ny,drop=TRUE] if (type != 'deviance') { nstrat <- as.numeric(strat) nvar <- ncol(x) if (is.null(strat)) { ord <- order(y[,ny-1], -status) newstrat <- rep(0,n) } else { ord <- order(nstrat, y[,ny-1], -status) newstrat <- c(diff(as.numeric(nstrat[ord]))!=0 ,1) } newstrat[n] <- 1 # sort the data x <- x[ord,] y <- y[ord,] score <- exp(object$linear.predictors)[ord] } } # # Now I have gotton the data that I need-- do the work # if (type=='schoenfeld') { if (ny==2) { mintime <- min(y[,1]) if (mintime < 0) y <- cbind(2*mintime -1, y) else y <- cbind(-1,y) } temp <- .C(Ccoxscho, n=as.integer(n), as.integer(nvar), as.double(y), resid= as.double(x), as.double(score * weights[ord]), as.integer(newstrat), as.integer(method=='efron'), double(3*nvar) ) deaths <- y[,3]==1 if (nvar==1) rr <- temp$resid[deaths] else rr <- matrix(temp$resid[deaths], ncol=nvar) #pick rows if (weighted) rr <- rr * weights[deaths] if (length(strats)) attr(rr, "strata") <- table((strat[ord])[deaths]) time <- c(y[deaths,2]) # 'c' kills all of the attributes if (is.matrix(rr)) dimnames(rr)<- list(time, names(object$coefficients)) else names(rr) <- time if (otype=='scaledsch') { ndead <- sum(deaths) coef <- ifelse(is.na(object$coefficients), 0, object$coefficients) if (nvar==1) rr <- rr*vv *ndead + coef else rr <- rr %*%vv * ndead + outer(rep(1,nrow(rr)),coef) } return(rr) } if (type=='score') { if (ny==2) { resid <- .C(Ccoxscore, as.integer(n), as.integer(nvar), as.double(y), x=as.double(x), as.integer(newstrat), as.double(score), as.double(weights[ord]), as.integer(method=='efron'), resid= double(n*nvar), double(2*nvar))$resid } else { resid<- .C(Cagscore, as.integer(n), as.integer(nvar), as.double(y), as.double(x), as.integer(newstrat), as.double(score), as.double(weights[ord]), as.integer(method=='efron'), resid=double(n*nvar), double(nvar*6))$resid } if (nvar >1) { rr <- matrix(0, n, nvar) rr[ord,] <- matrix(resid, ncol=nvar) dimnames(rr) <- list(names(object$residuals), names(object$coefficients)) } else rr[ord] <- resid if (otype=='dfbeta') { if (is.matrix(rr)) rr <- rr %*% vv else rr <- rr * vv } else if (otype=='dfbetas') { if (is.matrix(rr)) rr <- (rr %*% vv) %*% diag(sqrt(1/diag(vv))) else rr <- rr * sqrt(vv) } } # # Multiply up by case weights (which will be 1 for unweighted) # if (weighted) rr <- rr * weights #Expand out the missing values in the result if (!is.null(object$na.action)) { rr <- naresid(object$na.action, rr) if (is.matrix(rr)) n <- nrow(rr) else n <- length(rr) if (type=='deviance') status <- naresid(object$na.action, status) } if (type=="partial"){ # This needs to be done after the naresid expansion, since the # predict function will have done naresid expansion, so that # the lengths match rr <- rr + predict(object,type="terms") } # Collapse if desired if (!missing(collapse)) { if (length(collapse) !=n) stop("Wrong length for 'collapse'") rr <- drop(rowsum(rr, collapse)) if (type=='deviance') status <- drop(rowsum(status, collapse)) } # Deviance residuals are computed after collapsing occurs if (type=='deviance') sign(rr) *sqrt(-2* (rr+ ifelse(status==0, 0, status*log(status-rr)))) else rr } survival/R/is.ratetable.S0000644000175100001440000000761211732700061015050 0ustar hornikusers# # $Id: is.ratetable.S 11183 2009-01-21 13:33:40Z therneau $ # is.ratetable <- function(x, verbose=FALSE) { dlist <- c("dim", "dimnames", "dimid", "cutpoints") if (!verbose) { if (!inherits(x, 'ratetable')) return(FALSE) att <- attributes(x) if (any(is.na(match(dlist, names(att))))) return(FALSE) nd <- length(att$dim) if (length(x) != prod(att$dim)) return(FALSE) if (!(is.list(att$dimnames) && is.list(att$cutpoints))) return(FALSE) if (length(att$dimnames)!=nd || length(att$cutpoints)!=nd) return(FALSE) # One of 'factor' (old style table) or 'type' (new style) should exist if (!is.null(att$factor)) { fac <- as.numeric(att$factor) if (any(is.na(fac))) return(FALSE) if (any(fac <0)) return(FALSE) if (length(att$factor)!=nd ) return(FALSE) } else if (!is.null(att$type)) { if (any(is.na(match(att$type, 1:4)))) return(FALSE) fac <- 1*(att$type==1) if (length(fac) != nd) return(FALSE) } else return(FALSE) if (length(att$dimid) != nd) return(FALSE) for (i in 1:nd) { n <- att$dim[i] if (length(att$dimnames[[i]]) !=n) return(FALSE) if (fac[i]!=1 && length(att$cutpoints[[i]])!=n) return(FALSE) if (fac[i]!=1 && any(order(att$cutpoints[[i]])!= 1:n)) return(FALSE) if (fac[i]==1 && !is.null(att$cutpoints[[i]])) return(FALSE) if (fac[i]>1 && i0) } else if (!is.null(att$type)) { if (any(is.na(match(att$type, 1:4)))) msg <- c(msg, 'type attribute must be 1, 2, 3, or 4') type <- att$type if (length(type)!=nd) msg <- c(msg, 'wrong length for type attribute') } else msg <- c(msg, "missing the 'type' attribute") for (i in 1:nd) { n <- att$dim[i] if (length(att$dimnames[[i]]) !=n) msg <- c(msg, paste('dimname', i, 'is the wrong length')) if (type[i] >1) { #continuous variable if (length(att$cutpoints[[i]]) != n) msg <- c(msg, paste('wrong length for cutpoints', i)) else if (any(order(att$cutpoints[[i]])!= 1:n)) msg <- c(msg, paste('unsorted cutpoints for dimension',i)) } if (type[i]==1 && !is.null(att$cutpoints[[i]])) msg <- c(msg, paste('type[', i, '] is 1; cutpoint should be null')) # This message only applies to old style rate table if (!is.null(att$fac) && type[i]==4 && i 0) { if (length(cluster$terms) >1) stop ("Can have only 1 cluster term") idvar <- m[[cluster$vars]] Terms2 <- Terms[-cluster$terms] } else { idvar <- 1:n Terms2 <- Terms } if (length(attr(Terms, "specials")$strata)) { stemp <- untangle.specials(Terms2, 'strata', 1) if (length(stemp$terms) >0) #beware strata by covariate interactions Terms2 <- Terms2[-stemp$terms] #not needed for model.matrix later if (length(stemp$vars)==1) strata.keep <- m[[stemp$vars]] else strata.keep <- strata(m[,stemp$vars], shortlabel=TRUE) } else strata.keep <- NULL if (any(attr(Terms2, "order") > 1)) stop("This function cannot deal with iteraction terms") # Figure out which are the continuous predictor variables myvars <- attr(Terms2, "term.labels") factors <- sapply(m[myvars], is.factor) protected <- sapply(m[myvars], function(x) inherits(x, "AsIs")) keepers <- factors | protected #variables to be left alone if (all(keepers)) stop ("No continuous variables to modify") if (ncol(y) ==3) { # counting process data if (is.null(strata.keep)) { etime <- sort(unique(y[y[,3]==1, 2])) #unique event times indx <- lapply(etime, function(x) which(y[,1]= x)) } else { temp <- unique(data.frame(y[,2], strata.keep)[y[,3]==1,]) etime <- temp[,1] indx <- lapply(1:nrow(temp), function(x) which(y[,1] < temp[x,1] & y[,2]>= temp[x,1] & !strata.keep == temp[x,2])) } } else { # Simple survival data if (is.null(strata.keep)) { etime <- sort(unique(y[y[,2]==1,1])) #unique event times indx <- lapply(etime, function(x) which(y[,1] >=x)) } else { temp <- unique(data.frame(y[,1], strata.keep)[y[,2]==1,]) etime <- temp[,1] indx <- lapply(1:nrow(temp), function(x) which(y[,2] >= temp[x,1] & strata.keep == temp[x,2])) } } # The indx list now has an entry for each event time containing the # row numbers of those at risk indx2 <- unlist(indx) # Create the new survival variables nrisk <- unlist(sapply(indx, length)) #number of obs at risk if (ncol(y)==3) { newdata <- list(y[indx2, 1], y[indx2,2]) newdata <- c(newdata, list(1L*(newdata[[2]]==rep(etime, nrisk) &y[indx2,3]==1))) } else { newdata <- list(y[indx2,1]) newdata <- c(newdata, list(1L*(newdata[[1]]==rep(etime, nrisk) &y[indx2,2]==1))) } names(newdata) <- dimnames(y)[[2]] # Add any untransformed variables if (any(keepers)) { temp <- lapply(myvars[keepers], function(x) all.vars(parse(text=x))) knames <- unlist(temp) } else knames <- NULL if (length(strata.keep)) { knames <- c(knames, unlist(lapply(names(m)[stemp$vars], function(x) all.vars(parse(text=x))))) } if (length(knames)) newdata <- c(newdata, lapply(data[knames], function(x) x[indx2])) # Add the identifier variable if (length(cluster$vars) >0) { clname <- all.vars(parse(text=names(m)[cluster$vars])) newdata <- c(newdata, lapply(data[clname], function(x) x[indx2])) } else newdata <- c(newdata, list(".id."=idvar[indx2])) # Add transformed variables tvars <- myvars[!keepers] newx <- lapply(m[tvars], function(z) unlist(lapply(indx, function(x) transform(z[x])))) data.frame(c(newdata, newx, list(".strata."=rep(1:length(indx), sapply(indx, length))))) } survival/R/print.ratetable.S0000644000175100001440000000035711732700061015570 0ustar hornikusers## $Id: print.ratetable.S 11166 2008-11-24 22:10:34Z therneau $ print.ratetable <- function(x, ...) { cat ("Rate table with dimension(s):", attr(x, 'dimid'), "\n") attributes(x) <- attributes(x)[c("dim", "dimnames")] NextMethod() } survival/R/summary.ratetable.S0000644000175100001440000000445411732700061016133 0ustar hornikusers# $Id: summary.ratetable.S 11437 2010-10-28 02:21:16Z therneau $ # # Print out information about a rate table: it's dimensions and keywords # summary.ratetable <- function(object, ...) { rtable<-object if (!inherits(rtable, 'ratetable')) stop("Argument is not a rate table") att <- attributes(rtable) ncat <- length(dim(rtable)) cat (" Rate table with", ncat, "dimensions:\n") for (i in 1:ncat) { # One of 'factor' (old style table) or "type" (new style) should exist if (!is.null(att$factor)) { if (att$factor[i]==0) { cat("\t", att$dimid[i], " ranges from ", format(min(att$cutpoints[[i]])), " to ", format(max(att$cutpoints[[i]])), "; with ", att$dim[i], " categories\n", sep='') } else if(att$factor[i]==1) { cat("\t", att$dimid[i], " has levels of: ", paste(att$dimnames[[i]], collapse=' '), "\n", sep='') } else { cat("\t", att$dimid[i], " ranges from " , format(min(att$cutpoints[[i]])), " to ", format(max(att$cutpoints[[i]])), "; with ", att$dim[i], " categories,\n\t\tlinearly interpolated in ", att$factor[i], " steps per division\n", sep='') } } else { if (att$type[i]==1) { cat("\t", att$dimid[i], " has levels of: ", paste(att$dimnames[[i]], collapse=' '), "\n", sep='') } else if (att$type[i]>2) { #date cat("\t", att$dimid[i], " ranges from " , format(as.Date(min(att$cutpoints[[i]]), origin='1960/01/01')), " to ", format(as.Date(max(att$cutpoints[[i]]), origin='1960/01/01')), "; with ", att$dim[i], " categories\n", sep='') } else { cat("\t", att$dimid[i], " ranges from ", format(min(att$cutpoints[[i]])), " to ", format(max(att$cutpoints[[i]])), "; with ", att$dim[i], " categories\n", sep='') } } } invisible(att) } survival/R/agsurv.R0000644000175100001440000000620312533653600014002 0ustar hornikusers# Automatically generated from all.nw using noweb agsurv <- function(y, x, wt, risk, survtype, vartype) { nvar <- ncol(as.matrix(x)) status <- y[,ncol(y)] dtime <- y[,ncol(y) -1] death <- (status==1) time <- sort(unique(dtime)) nevent <- as.vector(rowsum(wt*death, dtime)) ncens <- as.vector(rowsum(wt*(!death), dtime)) wrisk <- wt*risk rcumsum <- function(x) rev(cumsum(rev(x))) # sum from last to first nrisk <- rcumsum(rowsum(wrisk, dtime)) irisk <- rcumsum(rowsum(wt, dtime)) if (ncol(y) ==2) { temp2 <- rowsum(wrisk*x, dtime) xsum <- apply(temp2, 2, rcumsum) } else { delta <- min(diff(time))/2 etime <- c(sort(unique(y[,1])), max(y[,1])+delta) #unique entry times indx <- approx(etime, 1:length(etime), time, method='constant', rule=2, f=1)$y esum <- rcumsum(rowsum(wrisk, y[,1])) #not yet entered nrisk <- nrisk - c(esum,0)[indx] irisk <- irisk - c(rcumsum(rowsum(wt, y[,1])),0)[indx] xout <- apply(rowsum(wrisk*x, y[,1]), 2, rcumsum) #not yet entered xin <- apply(rowsum(wrisk*x, dtime), 2, rcumsum) # dtime or alive xsum <- xin - (rbind(xout,0))[indx,,drop=F] } ndeath <- rowsum(status, dtime) #unweighted death count ntime <- length(time) if (survtype ==1) { #Kalbfleisch-Prentice indx <- (which(status==1))[order(dtime[status==1])] #deaths km <- .C(Cagsurv4, as.integer(ndeath), as.double(risk[indx]), as.double(wt[indx]), as.integer(ntime), as.double(nrisk), inc = double(ntime)) } if (survtype==3 || vartype==3) { # Efron approx xsum2 <- rowsum((wrisk*death) *x, dtime) erisk <- rowsum(wrisk*death, dtime) #risk score sums at each death tsum <- .C(Cagsurv5, as.integer(length(nevent)), as.integer(nvar), as.integer(ndeath), as.double(nrisk), as.double(erisk), as.double(xsum), as.double(xsum2), sum1 = double(length(nevent)), sum2 = double(length(nevent)), xbar = matrix(0., length(nevent), nvar)) } haz <- switch(survtype, nevent/nrisk, nevent/nrisk, nevent* tsum$sum1) varhaz <- switch(vartype, nevent/(nrisk * ifelse(nevent>=nrisk, nrisk, nrisk-nevent)), nevent/nrisk^2, nevent* tsum$sum2) xbar <- switch(vartype, (xsum/nrisk)*haz, (xsum/nrisk)*haz, nevent * tsum$xbar) result <- list(n= nrow(y), time=time, n.event=nevent, n.risk=irisk, n.censor=ncens, hazard=haz, cumhaz=cumsum(haz), varhaz=varhaz, ndeath=ndeath, xbar=apply(matrix(xbar, ncol=nvar),2, cumsum)) if (survtype==1) result$surv <- km$inc result } survival/R/untangle.specials.S0000644000175100001440000000124611732700061016107 0ustar hornikusers# $Id: untangle.specials.S 11166 2008-11-24 22:10:34Z therneau $ # # This function takes a terms object, and extracts some aspects # of it into a "nice" list. It is simple an operation that # I do again and again in the modeling routines, so it was # made into a separate function # untangle.specials <- function(tt, special, order=1) { spc <- attr(tt, 'specials')[[special]] if (length(spc)==0) return(list(vars=character(0), terms=numeric(0))) facs <- attr(tt, 'factors') fname <- dimnames(facs) ff <- apply(facs[spc,,drop=FALSE], 2, sum) list(vars= (fname[[1]])[spc], terms= seq(ff)[ff & match(attr(tt, 'order'), order, nomatch=0)]) } survival/R/match.ratetable.S0000644000175100001440000000675511732700061015540 0ustar hornikusers# Do a set of error checks on whether any categorical vars match the # level set of the actual ratetable. If so they are mapped to the levels # found in the ratetable. Dates need to match dates, and others are set # to simple numerics with unclass(). A matrix is returned. # This is called by pyears and survexp, but not by users # # The categoricals are turned into integer subscripts # match.ratetable <- function(R, ratetable) { if (!is.ratetable(ratetable)) stop("Invalid rate table") dimid <- attr(ratetable, 'dimid') if (is.matrix(R)) { # older style call nd <- ncol(R) attR <- attributes(R) attributes(R) <- attR['dim'] #other attrs get in the way later Rnames <- attR$dimnames[[2]] isDate <- attR[["isDate"]] levlist <- attR[['levlist']] } else { # newer style is a dataframe nd <- length(R) Rnames <- names(R) isDate <- rep(FALSE, nd) levlist<- lapply(R, levels) for (i in 1:nd) { temp <- ratetableDate(R[[i]]) if (!is.null(temp)) { isDate[i] <- TRUE R[[i]] <- temp } } } ord <- match(dimid, Rnames) # This should not arise if (any(is.na(ord))) stop(paste("Argument '", dimid[is.na(ord)], "' needed by the ratetable was not found in the data", sep='')) # Neither should this -- two argments matched one of the dimids -- since # I demand an exact match if (any(duplicated(ord))) stop("A ratetable argument appears twice in the data") R <- R[,ord,drop=FALSE] #put cols in same order as the ratetable isDate <- isDate[ord] levlist <- levlist[ord] dtemp <-dimnames(ratetable) rtype <- attr(ratetable, 'type') # 1= class, 2=cont, 3=date, 4=US yr if (is.null(rtype)) { #old style ratetable, be backwards compatable temp <- attr(ratetable, 'factor') # we map 'old continuous' to 'new date'; since it might be a date rtype <- 1*(temp==1) + 3*(temp==0) + 4*(temp >1) } # Now, go through the dimensions of the ratetable 1 by 1, and # verify that the user's variable is compatable # with the rate table's dimensions # if (any(rtype<3 & isDate)) { indx <- which(rtype<1 & isDate) stop(paste("Data has a date type variable, but the reference", "ratetable is not a date for variable", dimid[indx])) } for (i in (1:nd)) { if (length(levlist[[i]]) >0) { #factor or character variable if (rtype[i]!=1) stop(paste("In ratetable(),", dimid[i], "must be a continuous variable")) temp <- charmatch(casefold(levlist[[i]]), casefold(dtemp[[i]])) if (any(is.na(temp))) stop(paste("Levels do not match for ratetable() variable", dimid[i])) if (any(temp==0)) stop(paste("Non-unique ratetable match for variable", dimid[i])) R[,i] <- temp[as.numeric(R[,i])] } else { # user's data isn't a factor or date R[,i] <- unclass(R[,i]) # get rid of difftimes & other such if (rtype[i]==1) { #ratetable is a factor: ok if data is integer temp <- R[,i] if (any(floor(temp)!=temp) || any(temp<=0) || max(temp) > length(dtemp[[i]])) stop(paste("The variable", dimid[i], "is out of range")) } } } R <- as.matrix(R) summ <- attr(ratetable, 'summary') if (is.null(summ)) list(R= R) else list(R= R, summ=summ(R)) } survival/R/ridge.S0000644000175100001440000000251211732700061013557 0ustar hornikusers# $Id: ridge.S 11166 2008-11-24 22:10:34Z therneau $ ridge <- function(..., theta, df=nvar/2, eps=.1, scale=TRUE) { x <- cbind(...) nvar <- ncol(x) xname <- as.character(parse(text=substitute(cbind(...))))[-1] vars <- apply(x, 2, function(z) var(z[!is.na(z)])) class(x) <- 'coxph.penalty' if (!missing(theta) && !missing(df)) stop("Only one of df or theta can be specified") if (scale) pfun <- function(coef,theta, ndead, scale) { list(penalty= sum(coef^2 *scale)*theta/2, first = theta*coef*scale, second = theta*scale, flag=FALSE) } else pfun <- function(coef,theta, ndead, scale) { list(penalty= sum(coef^2)*theta/2, first = theta*coef, second = theta, flag=FALSE) } if (!missing(theta)) { temp <- list(pfun=pfun, diag=TRUE, cfun=function(parms, iter, history) { list(theta=parms$theta, done=TRUE) }, cparm=list(theta= theta), pparm= vars, varname=paste('ridge(', xname, ')', sep='')) } else { temp <- list(pfun=pfun, diag=TRUE, cfun=frailty.controldf, cargs = 'df', cparm=list(df=df, eps=eps, thetas=0, dfs=nvar, guess=1), pparm= vars, varname=paste('ridge(', xname, ')', sep='')) } attributes(x) <- c(attributes(x), temp) x } survival/R/clogit.R0000644000175100001440000000427012466417353013766 0ustar hornikusers## conditional logistic regression ## ## case ~ exposure + strata(matching) ## clogit<-function(formula, data, weights, subset, na.action, method=c("exact","approximate", "efron", "breslow"), ... ) { Call <- match.call() # how we were called # Create a call to model.frame() that contains the formula (required) # and the data argument (if present). # It's only job is to find out the number of rows in the data # before subset or na.action are applied. indx <- match(c("formula", "data"), names(Call), nomatch=0) if (indx[1]==0) stop("A formula argument is required") mf <- Call[c(1,indx)] mf[[1]] <- as.name("model.frame") mf$na.action <- "na.pass" nrows<-NROW(eval(mf, parent.frame())) method <- match.arg(method) # Catch the rare case of a person asking for robust variance, and give # them a nicer warning than will occur if they fall through to the # coxph call if (missing(data)) temp <- terms(formula, special='cluster') else temp <- terms(formula, special="cluster", data=data) if (!is.null(attr(temp, 'specials')$cluster) && method=="exact") stop("robust variance plus the exact method is not supported") # Now build a call to coxph with the formula fixed up to have # our special left hand side. coxcall <- Call coxcall[[1]] <- as.name("coxph") newformula <- formula newformula[[2]] <- substitute(Surv(rep(1,nn),case), list(case=formula[[2]],nn=nrows)) environment(newformula) <- environment(formula) coxcall$formula<-newformula coxcall$method <- switch(method, exact="exact", efron="efron", "breslow") if (!is.null(coxcall$weights)) { coxcall$weights <- NULL warning("Weights are ignored in clogit") } coxcall<-eval(coxcall, sys.frame(sys.parent())) coxcall$userCall<-sys.call() class(coxcall)<-c("clogit","coxph") coxcall } print.clogit <- function(x,...){ x$call<-x$userCall NextMethod() } survfit.clogit <- function(formula, ...) stop("predicted survival curves are not defined for a clogit model") survival/R/survexp.R0000644000175100001440000002140312533653606014214 0ustar hornikusers# Automatically generated from all.nw using noweb survexp <- function(formula, data, weights, subset, na.action, rmap, times, method=c("ederer", "hakulinen", "conditional", "individual.h", "individual.s"), cohort=TRUE, conditional=FALSE, ratetable=survival::survexp.us, scale=1, se.fit, model=FALSE, x=FALSE, y=FALSE) { call <- match.call() m <- match.call(expand.dots=FALSE) # keep the first element (the call), and the following selected arguments m <- m[c(1, match(c('formula', 'data', 'weights', 'subset', 'na.action'), names(m), nomatch=0))] m[[1]] <- as.name("model.frame") Terms <- if(missing(data)) terms(formula, 'ratetable') else terms(formula, 'ratetable',data=data) rate <- attr(Terms, "specials")$ratetable if(length(rate) > 1) stop("Can have only 1 ratetable() call in a formula") if(length(rate) == 1) { if (!missing(rmap)) stop("The ratetable() call in a formula is depreciated") stemp <- untangle.specials(Terms, 'ratetable') rcall <- as.call(parse(text=stemp$var)[[1]]) # as a call object rcall[[1]] <- as.name('list') # make it a call to list(.. Terms <- Terms[-stemp$terms] # remove from the formula } else if (!missing(rmap)) { rcall <- substitute(rmap) if (!is.call(rcall) || rcall[[1]] != as.name('list')) stop ("Invalid rcall argument") } else rcall <- NULL # A ratetable, but not rcall argument # Check that there are no illegal names in rcall, then expand it # to include all the names in the ratetable if(is.ratetable(ratetable)) varlist <- attr(ratetable, "dimid") else if(inherits(ratetable, "coxph")) { ## Remove "log" and such things, to get just the list of # variable names varlist <- all.vars(delete.response(ratetable$terms)) } else stop("Invalid rate table") temp <- match(names(rcall)[-1], varlist) # 2,3,... are the argument names if (any(is.na(temp))) stop("Variable not found in the ratetable:", (names(rcall))[is.na(temp)]) if (any(!(varlist %in% names(rcall)))) { to.add <- varlist[!(varlist %in% names(rcall))] temp1 <- paste(text=paste(to.add, to.add, sep='='), collapse=',') if (is.null(rcall)) rcall <- parse(text=paste("list(", temp1, ")"))[[1]] else { temp2 <- deparse(rcall) rcall <- parse(text=paste("c(", temp2, ",list(", temp1, "))"))[[1]] } } # Create a temporary formula, used only in the call to model.frame newvar <- all.vars(rcall) if (length(newvar) > 0) { tform <- paste(paste(deparse(Terms), collapse=""), paste(newvar, collapse='+'), sep='+') m$formula <- as.formula(tform, environment(Terms)) } m <- eval(m, parent.frame()) n <- nrow(m) if (n==0) stop("Data set has 0 rows") if (!missing(se.fit) && se.fit) warning("se.fit value ignored") weights <- model.extract(m, 'weights') if (length(weights) ==0) weights <- rep(1.0, n) if (class(ratetable)=='ratetable' && any(weights !=1)) warning("weights ignored") if (any(attr(Terms, 'order') >1)) stop("Survexp cannot have interaction terms") if (!missing(times)) { if (any(times<0)) stop("Invalid time point requested") if (length(times) >1 ) if (any(diff(times)<0)) stop("Times must be in increasing order") } Y <- model.extract(m, 'response') no.Y <- is.null(Y) if (no.Y) { if (missing(times)) { if (is.ratetable(ratetable)) stop("either a times argument or a response is needed") } else newtime <- times } else { if (is.matrix(Y)) { if (is.Surv(Y) && attr(Y, 'type')=='right') Y <- Y[,1] else stop("Illegal response value") } if (any(Y<0)) stop ("Negative follow up time") # if (missing(npoints)) temp <- unique(Y) # else temp <- seq(min(Y), max(Y), length=npoints) temp <- unique(Y) if (missing(times)) newtime <- sort(temp) else newtime <- sort(unique(c(times, temp[temp1)) stop("Pyears cannot have interaction terms") rate <- attr(Terms, "specials")$ratetable if (length(rate) >0 || !missing(rmap) || !missing(ratetable)) { has.ratetable <- TRUE if(length(rate) > 1) stop("Can have only 1 ratetable() call in a formula") if (missing(ratetable)) stop("No rate table specified") if(length(rate) == 1) { if (!missing(rmap)) stop("The ratetable() call in a formula is depreciated") stemp <- untangle.specials(Terms, 'ratetable') rcall <- as.call(parse(text=stemp$var)[[1]]) # as a call object rcall[[1]] <- as.name('list') # make it a call to list(.. Terms <- Terms[-stemp$terms] # remove from the formula } else if (!missing(rmap)) { rcall <- substitute(rmap) if (!is.call(rcall) || rcall[[1]] != as.name('list')) stop ("Invalid rcall argument") } else rcall <- NULL # A ratetable, but not rcall argument # Check that there are no illegal names in rcall, then expand it # to include all the names in the ratetable if(is.ratetable(ratetable)) varlist <- attr(ratetable, "dimid") else if(inherits(ratetable, "coxph")) { ## Remove "log" and such things, to get just the list of # variable names varlist <- all.vars(delete.response(ratetable$terms)) } else stop("Invalid rate table") temp <- match(names(rcall)[-1], varlist) # 2,3,... are the argument names if (any(is.na(temp))) stop("Variable not found in the ratetable:", (names(rcall))[is.na(temp)]) if (any(!(varlist %in% names(rcall)))) { to.add <- varlist[!(varlist %in% names(rcall))] temp1 <- paste(text=paste(to.add, to.add, sep='='), collapse=',') if (is.null(rcall)) rcall <- parse(text=paste("list(", temp1, ")"))[[1]] else { temp2 <- deparse(rcall) rcall <- parse(text=paste("c(", temp2, ",list(", temp1, "))"))[[1]] } } # Create a temporary formula, used only in the call to model.frame newvar <- all.vars(rcall) if (length(newvar) > 0) { tform <- paste(paste(deparse(Terms), collapse=""), paste(newvar, collapse='+'), sep='+') m$formula <- as.formula(tform, environment(Terms)) } } else has.ratetable <- FALSE if (is.R()) m <- eval(m, parent.frame()) else m <- eval(m, sys.parent()) Y <- model.extract(m, 'response') if (is.null(Y)) stop ("Follow-up time must appear in the formula") if (!is.Surv(Y)){ if (any(Y <0)) stop ("Negative follow up time") Y <- as.matrix(Y) if (ncol(Y) >2) stop("Y has too many columns") } else { stype <- attr(Y, 'type') if (stype == 'right') { if (any(Y[,1] <0)) stop("Negative survival time") nzero <- sum(Y[,1]==0 & Y[,2] ==1) if (nzero >0) warning(paste(nzero, "observations with an event and 0 follow-up time,", "any rate calculations are statistically questionable")) } else if (stype != 'counting') stop("Only right-censored and counting process survival types are supported") } n <- nrow(Y) if (is.null(n) || n==0) stop("Data set has 0 observations") weights <- model.extract(m, 'weights') if (is.null(weights)) weights <- rep(1.0, n) # rdata contains the variables matching the ratetable if (has.ratetable) { rdata <- data.frame(eval(rcall, m), stringsAsFactors=TRUE) if (is.ratetable(ratetable)) { israte <- TRUE rtemp <- match.ratetable(rdata, ratetable) R <- rtemp$R } else if (inherits(ratetable, 'coxph')) { israte <- FALSE Terms <- ratetable$terms if (!is.null(attr(Terms, 'offset'))) stop("Cannot deal with models that contain an offset") strats <- attr(Terms, "specials")$strata if (length(strats)) stop("pyears cannot handle stratified Cox models") if (any(names(m[,rate]) != attr(ratetable$terms, 'term.labels'))) stop("Unable to match new data to old formula") R <- model.matrix.coxph(ratetable, data=rdata) } else stop("Invalid ratetable") } ovars <- attr(Terms, 'term.labels') if (length(ovars)==0) { # no categories! X <- rep(1,n) ofac <- odim <- odims <- ocut <- 1 } else { odim <- length(ovars) ocut <- NULL odims <- ofac <- double(odim) X <- matrix(0, n, odim) outdname <- vector("list", odim) for (i in 1:odim) { temp <- m[[ovars[i]]] if (inherits(temp, 'tcut')) { X[,i] <- temp temp2 <- attr(temp, 'cutpoints') odims[i] <- length(temp2) -1 ocut <- c(ocut, temp2) ofac[i] <- 0 outdname[[i]] <- attr(temp, 'labels') } else { temp2 <- as.factor(temp) X[,i] <- temp2 temp3 <- levels(temp2) odims[i] <- length(temp3) ofac[i] <- 1 outdname[[i]] <- temp3 } } } ocut <-c(ocut,0) #just in case it were of length 0 osize <- prod(odims) if (has.ratetable) { #include expected atts <- attributes(ratetable) cuts <- atts$cutpoints if (is.null(atts$type)) { #old stlye table rfac <- atts$factor us.special <- (rfac >1) } else { rfac <- 1*(atts$type ==1) us.special <- (atts$type==4) } if (any(us.special)) { #special handling for US pop tables # Now, the 'entry' date on a US rate table is the number of days # since 1/1/1960, and the user data has been aligned to the # same system by match.ratetable and marked as "year". # The birth date is entry date - age in days (based on 1/1/1960). # I don't much care which date functions I use to do the arithmetic # below. Unfortunately R and Splus don't share one. My "date" # class is simple, but is also one of the earlier date class # attempts, has less features than others, and will one day fade # away; so I don't want to depend on it alone. # cols <- match(c("age", "year"), atts$dimid) if (any(is.na(cols))) stop("Ratetable does not have expected shape") if (exists("as.Date")) { # true for modern version of R bdate <- as.Date('1960/1/1') + (R[,cols[2]] - R[,cols[1]]) byear <- format(bdate, "%Y") offset <- bdate - as.Date(paste(byear, "01/01", sep='/'), origin="1960/01/01") } #else if (exists('month.day.year')) { # Splus, usually # bdate <- R[,cols[2]] - R[,cols[1]] # byear <- month.day.year(bdate)$year # offset <- bdate - julian(1,1,byear) # } #else if (exists('date.mdy')) { # Therneau's date class is available # bdate <- as.date(R[,cols[2]] - R[,cols[1]]) # byear <- date.mdy(bdate)$year # offset <- bdate - mdy.date(1,1,byear) # } else stop("Can't find an appropriate date class\n") R[,cols[2]] <- R[,cols[2]] - offset # Doctor up "cutpoints" - only needed for old style rate tables # for which the C code does interpolation on the fly if (any(rfac >1)) { temp <- which(us.special) nyear <- length(cuts[[temp]]) nint <- rfac[temp] #intervals to interpolate over cuts[[temp]] <- round(approx(nint*(1:nyear), cuts[[temp]], nint:(nint*nyear))$y - .0001) } } docount <- is.Surv(Y) temp <- .C(Cpyears1, as.integer(n), as.integer(ncol(Y)), as.integer(is.Surv(Y)), as.double(Y), as.double(weights), as.integer(length(atts$dim)), as.integer(rfac), as.integer(atts$dim), as.double(unlist(cuts)), as.double(ratetable), as.double(R), as.integer(odim), as.integer(ofac), as.integer(odims), as.double(ocut), as.integer(expect=='event'), as.double(X), pyears=double(osize), pn =double(osize), pcount=double(if(docount) osize else 1), pexpect=double(osize), offtable=double(1))[18:22] } else { #no expected docount <- as.integer(ncol(Y) >1) temp <- .C(Cpyears2, as.integer(n), as.integer(ncol(Y)), as.integer(docount), as.double(Y), as.double(weights), as.integer(odim), as.integer(ofac), as.integer(odims), as.double(ocut), as.double(X), pyears=double(osize), pn =double(osize), pcount=double(if (docount) osize else 1), offtable=double(1)) [11:14] } if (data.frame) { # Create a data frame as the output, rather than a set of # rate tables keep <- (temp$pyears >0) # what rows to keep in the output names(outdname) <- ovars if (length(outdname) ==1) { # if there is only one variable, the call to "do.call" loses # the variable name, since expand.grid returns a factor df <- data.frame((outdname[[1]])[keep], pyears= temp$pyears[keep]/scale, n = temp$pn[keep]) names(df) <- c(names(outdname), 'pyears', 'n') } else { df <- cbind(do.call("expand.grid", outdname)[keep,], pyears= temp$pyears[keep]/scale, n = temp$pn[keep]) } row.names(df) <- 1:nrow(df) if (has.ratetable) df$expected <- temp$pexpect[keep] if (expect=='pyears') df$expected <- df$expected/scale if (docount) df$event <- temp$pcount[keep] out <- list(call=call, data= df, offtable=temp$offtable/scale) if (has.ratetable && !is.null(rtemp$summ)) out$summary <- rtemp$summ } else if (prod(odims) ==1) { #don't make it an array out <- list(call=call, pyears=temp$pyears/scale, n=temp$pn, offtable=temp$offtable/scale) if (has.ratetable) { out$expected <- temp$pexpect if (expect=='pyears') out$expected <- out$expected/scale if (!is.null(rtemp$summ)) out$summary <- rtemp$summ } if (docount) out$event <- temp$pcount } else { out <- list(call = call, pyears= array(temp$pyears/scale, dim=odims, dimnames=outdname), n = array(temp$pn, dim=odims, dimnames=outdname), offtable = temp$offtable/scale) if (has.ratetable) { out$expected <- array(temp$pexpect, dim=odims, dimnames=outdname) if (expect=='pyears') out$expected <- out$expected/scale if (!is.null(rtemp$summ)) out$summary <- rtemp$summ } if (docount) out$event <- array(temp$pcount, dim=odims, dimnames=outdname) } out$observations <- nrow(m) na.action <- attr(m, "na.action") if (length(na.action)) out$na.action <- na.action if (model) out$model <- m else { if (x) out$x <- X if (y) out$y <- Y } oldClass(out) <- 'pyears' out } survival/R/summary.survreg.S0000644000175100001440000000451712470201064015664 0ustar hornikusers# $Id: summary.survreg.S 11166 2008-11-24 22:10:34Z therneau $ summary.survreg<- function(object, correlation = FALSE,...) { if (!is.null(object$fail)) { warning(" Survreg failed.", object$fail, " No summary provided\n") return(invisible(object)) } nvar0 <- length(object$coefficients) nvar <- nrow(object$var) if (nvar > nvar0) { coef <- c(object$coefficients, log(object$scale)) if ( (nvar-nvar0)==1) cname <- c(names(object$coefficients), "Log(scale)") else cname <- c(names(object$coefficients), names(object$scale)) } else { coef <- object$coefficients cname <- names(object$coefficients) } n <- length(object$linear.predictors) p <- sum(!is.na(coef)) if(!p) { warning("This model has zero rank --- no summary is provided") return(invisible(object)) } if (is.null(object$naive.var)){ table <- matrix(rep(coef, 4), ncol = 4) dimnames(table) <- list(cname, c("Value", "Std. Error", "z", "p")) stds <- sqrt(diag(object$var)) table[, 2] <- stds table[, 3] <- table[, 1]/stds table[, 4] <- 2*pnorm(-abs(table[,3])) } else { table <- matrix(rep(coef, 5), ncol = 5) dimnames(table) <- list(cname, c("Value", "Std. Err","(Naive SE)", "z", "p")) stds <- sqrt(diag(object$var)) table[, 2] <- stds table[, 3] <- sqrt(diag(object$naive.var)) table[, 4] <- table[, 1]/stds table[, 5] <- 2*pnorm(-abs(table[,4])) } if(correlation) { nas <- is.na(coef) stds <- stds[!nas] correl <- diag(1/stds) %*% object$var[!nas, !nas] %*% diag(1/stds) dimnames(correl) <- list(cname, cname) } else correl <- NULL dist <- object$dist if (is.character(dist)) sd <- survreg.distributions[[dist]] else sd <- dist if (length(object$parms)) pprint<- paste(sd$name, 'distribution: parmameters=', object$parms) else pprint<- paste(sd$name, 'distribution') x <- object[match(c('call', 'df', 'loglik', 'iter', 'na.action', 'idf', 'scale', 'coefficients', 'var'), names(object), nomatch=0)] x <- c(x, list(table=table, correlation=correl, parms=pprint, n=n, chi=2*diff(object$loglik)), robust=!is.null(object$naive.var)) if (is.R()) class(x) <- 'summary.survreg' else oldClass(x) <- 'summary.survreg' x } survival/R/coxph.wtest.S0000644000175100001440000000252112113164602014752 0ustar hornikusers# # A Wald test routine, used by the Cox model # Why not just do sum(b * solve(var, b))? -- because the solve # function chokes on singular matrices. # coxph.wtest <- function(var, b, toler.chol=1e-9) { if (is.matrix(b)) { nvar <- nrow(b) ntest<- ncol(b) } else { nvar <- length(b) ntest<- 1 } if (length(var)==0) { #special case added by Tom Lumley if (nvar==0) return(list(test=numeric(0), df=0, solve=0)) else stop("Argument lengths do not match") } if (length(var)==1) { if (nvar ==1) return(list(test=b*b/var, df=1, solve=b/var)) else stop("Argument lengths do not match") } if (!is.matrix(var) || (nrow(var) != ncol(var))) stop("First argument must be a square matrix") if (nrow(var) != nvar) stop("Argument lengths do not match") temp <- .C(Ccoxph_wtest, df=as.integer(nvar), as.integer(ntest), as.double(var), tests= as.double(b), solve= double(nvar*ntest), as.double(toler.chol)) if (ntest==1) list(test=temp$tests[1], df=temp$df, solve=temp$solve) else list(test=temp$tests[1:ntest], df=temp$df, solve=matrix(temp$solve, nvar, ntest)) } survival/R/survcallback.S0000644000175100001440000001670711732700061015154 0ustar hornikusers# # This is common code for survpenal.fit and coxpenal.fit. # It's all the bookkeeping to set up the penalized callbacks # This code is in development, and not yet used by anything, the if(FALSE) # keeps it out of the distribution if(FALSE){ survcallback <- function(pcols, pattr, assign, x) { # # are there any sparse frailty terms? # npenal <- length(pattr) #total number of penalized terms if (npenal == 0 || length(pcols) != npenal) stop("Invalid pcols or pattr arg") sparse <- sapply(pattr, function(x) !is.null(x$sparse) && x$sparse) if (sum(sparse) >1) stop("Only one sparse penalty term allowed") # # Create a marking vector for the terms, the same length as assign # with pterms == 0=ordinary term, 1=penalized, 2=sparse, # pindex = length of pcols = position in pterms # # Make sure that pcols is a strict subset of assign, so that the # df computation (and printing) can unambiguously decide which cols of # X are penalized and which are not when doing "terms" like actions. # To make some downstream things easier, order pcols and pattr to be # in the same relative order as the terms in 'assign' # pterms <- rep(0, length(assign)) names(pterms) <- names(assign) pindex <- rep(0, npenal) for (i in 1:npenal) { temp <- unlist(lapply(assign, function(x,y) (length(x) == length(y) && all(x==y)), pcols[[i]])) if (sparse[i]) pterms[temp] <- 2 else pterms[temp] <- 1 pindex[i] <- (seq(along.with=temp))[temp] } if ((sum(pterms==2) != sum(sparse)) || (sum(pterms>0) != npenal)) stop("pcols and assign arguments disagree") if (any(pindex != sort(pindex))) { temp <- order(pindex) pindex <- pindex[temp] pcols <- pcols[temp] pattr <- pattr[temp] } # ptype= 1 or 3 if a sparse term exists, 2 or 3 if a non-sparse exists ptype <- any(sparse) + 2*(any(!sparse)) if (any(sparse)) { sparse.attr <- (pattr[sparse])[[1]] #can't use [[sparse]] directly # if 'sparse' is a T/F vector fcol <- unlist(pcols[sparse]) if (length(fcol) > 1) stop("Sparse term must be single column") # Remove the sparse term from the X matrix frailx <- x[, fcol] x <- x[, -fcol, drop=FALSE] for (i in 1:length(assign)){ j <- assign[[i]] if (j[1] > fcol) assign[[i]] <- j-1 } for (i in 1:npenal) { j <- pcols[[i]] if (j[1] > fcol) pcol[[i]] <- j-1 } frailx <- match(frailx, sort(unique(frailx))) nfrail <- max(frailx) nvar <- nvar - 1 #Set up the callback for the sparse frailty term # (At most one sparse term is allowed). The calling code will # first set 'coef' to the current value of the sparse coefficients, # then call the expression below. It uses a separate context (Splus # frame or R environment), so there is no conflict between that # variable name and the rest of the code. Thus, think of the below as # a funcion of the temporary variable coef (current value found # in the calling C code), theta1 (current value in the S code # below, using calls to cfun), and fixed known values of pfun1 etc. # The expression will constantly replace components of "coxlist1". By # creating it first, we assure the order of the components, again # to make it simpler for the C code (it can grab the first component # and know that that is 'coef', etc). # pfun1 <- sparse.attr$pfun coxlist1 <- list(coef=0, first=0, second=0, penalty=0, flag=F) f.expr1 <- quote({ if (is.null(extra1)) temp <- pfun1(coef1, theta1, n.eff) else temp <- pfun1(coef1, theta1, n.eff, extra1) if (!is.null(temp$recenter)) coxlist1$coef <- coef1 - as.double(temp$recenter) else coxlist1$coef <- coef1 if (!temp$flag) { coxlist1$first <- -as.double(temp$first) coxlist1$second <- as.double(temp$second) } else { coxlist1$first <- double(nfrail) coxlist1$second <- double(nfrail) } coxlist1$penalty <- -as.double(temp$penalty) coxlist1$flag <- as.logical(temp$flag) # Make sure the list has exactly the right structure, so # the the C code can be simple. The first line below is # probably unnecessary (belt AND suspenders); the second is # checking a possibly user-supplied penaly function if (any(names(coxlist1) != c('coef', 'first', 'second', 'penalty', 'flag'))) stop("Invalid coxlist1") if (any(sapply(coxlist1, length) != c(rep(nfrail,3), 1, 1))) stop("Incorrect length in coxlist1") coxlist1 }) } else { # no sparse terms frailx <- 0 nfrail <- 0 f.expr1 <- NULL #dummy value pfun1 <- NULL #dummy coxlist1 <- NULL # " } nvar2 <- nvar + nstrat2 if (nvar2 ==0) { # There are no non-sparse coefficients, and no scale parameters # A strange model, leading to an hmat with 0 columns. The # underlying C code will choke, since this case is not built in. stop("Cannot fit a model with no coefficients other than sparse ones") } # Now the non-sparse penalties # There can be multiple penalized terms if (sum(!sparse) >0) { full.imat <- !all(unlist(lapply(pattr, function(x) x$diag))) ipenal <- (1:length(pattr))[!sparse] #index for non-sparse terms if (full.imat) { coxlist2 <- list(coef=double(nvar), first=double(nvar), second= double(nvar^2), penalty=0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar*nvar, 1, nvar) } else { coxlist2 <- list(coef=double(nvar), first=double(nvar), second=double(nvar), penalty= 0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar, 1, nvar) } # The C code will set the variable coef, containing the concatonation # of all the non-sparse penalized coefs. Think of the below as # a function of coef (from the C code), thetalist (set further # below), and unchanging variables such as pattr. f.expr2 <- quote({ pentot <- 0 newcoef <- coef2 for (i in ipenal) { pen.col <- pcols[[i]] tcoef <- coef2[pen.col] if (is.null(extralist[[i]])) temp <- ((pattr[[i]])$pfun)(tcoef, thetalist[[i]], n.eff) else temp <- ((pattr[[i]])$pfun)(tcoef, thetalist[[i]], n.eff,extralist[[i]]) if (!is.null(temp$recenter)) newcoef[pen.col] <- tcoef - temp$recenter if (temp$flag) coxlist2$flag[pen.col] <- TRUE else { coxlist2$flag[pen.col] <- FALSE coxlist2$first[pen.col] <- -temp$first if (full.imat) { tmat <- matrix(coxlist2$second, nvar, nvar) tmat[pen.col,pen.col] <- temp$second coxlist2$second <- c(tmat) } else coxlist2$second[pen.col] <- temp$second } pentot <- pentot - temp$penalty } coxlist2$penalty <- as.double(pentot) coxlist2$coef <- newcoef if (any(sapply(coxlist2, length) != length2)) stop("Length error in coxlist2") coxlist2 }) } else { full.imat <- FALSE # no non-sparse penalties length2 <- 0 #dummy value f.expr2 <- NULL coxlist2 <- NULL ipenal <- NULL } list(f.expr1=f.expr1, f.expr2=f.expr2, coxlist1=coxlist1, coxlist2=coxlist2, full.imat=full.imat, ipenal=ipenal, length2=length2, pfun1=pfun1, pindex=pindex, pcols=pcols, pattr=pattr, sparse=sparse, frailx=frailx, nfrail=nfrail, nvar=nvar) } } survival/R/Surv.S0000644000175100001440000002316312257335007013440 0ustar hornikusers# # Package up surivival type data as a structure # Surv <- function(time, time2, event, type=c('right', 'left', 'interval', 'counting', 'interval2', "mstate"), origin=0) { if (missing(time)) stop ("Must have a time argument") if (!is.numeric(time)) stop ("Time variable is not numeric") nn <- length(time) # ng = number of the first 3 arguments that is present ng <- (!missing(time)) + (!missing(time2)) + (!missing(event)) # The logic below uses "ng" throughout; why not use "missing(time2)" # and missing(event) instead? Because we want to assume that # "Surv(a,b)" has the variable b matched to event rather than time2. # mtype <- match.arg(type) # Retain any attributes of the input arguments. Originally requested # by the rms package inputAttributes <- list() if (!is.null(attributes(time))) inputAttributes$time <-attributes(time) if (!missing(time2) && !is.null(attributes(time2))) inputAttributes$time2 <- attributes(time2) if (!missing(event) && !is.null(attributes(event))) inputAttributes$event <- attributes(event) # If type is missing or it is "mstate", I need to figure out for myself # whether I have (time, time2, status) or (time, status) data if (missing(type) || mtype=="mstate") { if (ng==1 || ng==2) type <- 'right' else if (ng==3) type <- 'counting' else stop ("No time variable!") # no time variable at all! } else { type <- mtype if (ng!=3 && (type=='interval' || type =='counting')) stop("Wrong number of args for this type of survival data") if (ng!=2 && (type=='right' || type=='left' || type=='interval2')) stop("Wrong number of args for this type of survival data") } if (ng==1) { # only a time variable given if (!is.numeric(time)) stop("Time variable is not numeric") ss <- cbind(time=time-origin, status=1) type <- "right" } else if (type=='right' || type=='left') { if (!is.numeric(time)) stop("Time variable is not numeric") if (missing(event)) event <- time2 # treat time2 as event if (length(event) != nn) stop ("Time and status are different lengths") if (mtype=="mstate" || (is.factor(event) && length(levels(event))>2)) { mstat <- as.factor(event) status <- as.numeric(mstat) -1 type <- "mright" } else { if (is.logical(event)) status <- as.numeric(event) else if (is.numeric(event)) { who2 <- !is.na(event) if (max(event[who2]) ==2) status <- event -1 else status <- event temp <- (status==0 | status==1) status <- ifelse(temp, status, NA) if (!all(temp[who2], na.rm=TRUE)) warning("Invalid status value, converted to NA") } else stop("Invalid status value, must be logical or numeric") } ss <- cbind(time=time-origin, status=status) } else if (type=='counting') { if (length(time2) !=nn) stop ("Start and stop are different lengths") if (length(event)!=nn) stop ("Start and event are different lengths") if (!is.numeric(time)) stop("Start time is not numeric") if (!is.numeric(time2)) stop("Stop time is not numeric") temp <- (time >= time2) if (any(temp & !is.na(temp))) { time[temp] <- NA warning("Stop time must be > start time, NA created") } if (mtype=="mstate" || (is.factor(event) && length(levels(event))>2)) { mstat <- as.factor(event) status <- as.numeric(mstat) -1 type <- "mcounting" } else { if (is.logical(event)) status <- as.numeric(event) else if (is.numeric(event)) { who2 <- !is.na(event) if (max(event[who2])==2) status <- event - 1 else status <- event temp <- (status==0 | status==1) status <- ifelse(temp, status, NA) if (!all(temp[who2], na.rm=TRUE)) warning("Invalid status value, converted to NA") } else stop("Invalid status value") } ss <- cbind(start=time-origin, stop=time2-origin, status=status) } else { #interval censored data if (type=='interval2') { # convert to "interval" type, infer the event code if (!is.numeric(time2)) stop("Time2 must be numeric") if (length(time2) !=nn) stop ("time and time2 are different lengths") time <- ifelse(time==Inf, NA, time) #allow Inf for upper/lower time2 <- ifelse(time2== -Inf, NA, time2) backwards <- (!is.na(time) & !is.na(time2) & time > time2) unknown <- (is.na(time) & is.na(time2)) status <- ifelse(is.na(time), 2, ifelse(is.na(time2), 0, ifelse(time==time2, 1,3))) time <- ifelse(status!=2, time, time2) if (any(backwards)) { warning("Invalid interval: start > stop, NA created") status[backwards] <- NA } if (any(unknown)) status[unknown] <- NA type <- 'interval' } else { #check legality of event code if (length(event)!=nn) stop("Time and status are different lengths") if (!is.numeric(event)) stop("Invalid status value, must be logical or numeric") temp <- (event==0 | event==1| event==2 | event==3) status <- ifelse(temp, event, NA) if (!all(temp, na.rm=TRUE)) warning("Status must be 0, 1, 2 or 3; converted to NA") if (any(event==3, na.rm=T)) { if (!is.numeric(time2)) stop("Time2 must be numeric") if (length(time2) !=nn) stop ("time and time2 are different lengths") temp <- (status==3 & time>time2) if (any(temp & !is.na(temp))) { status[temp] <- NA warning("Invalid interval: start > stop, NA created") } } else time2 <- 1 #dummy value, time2 is never used } ss <- cbind(time1=time-origin, time2=ifelse(!is.na(status) & status==3, time2-origin, 1), status=status) } dimnames(ss) <- list(NULL, dimnames(ss)[[2]]) #kill any tag-along row names attr(ss, "type") <- type if (type=="mright" || type=="mcounting") attr(ss, "states") <- levels(mstat)[-1] if (length(inputAttributes) > 0) attr(ss, "inputAttributes") <- inputAttributes class(ss) <- 'Surv' ss } print.Surv <- function(x, quote=FALSE, ...) { invisible(print(as.character.Surv(x), quote=quote, ...)) } as.character.Surv <- function(x, ...) { switch(attr(x, "type"), "right"={ temp <- x[,2] temp <- ifelse(is.na(temp), "?", ifelse(temp==0, "+"," ")) paste(format(x[,1]), temp, sep='') }, "counting"= { temp <- x[,3] temp <- ifelse(is.na(temp), "?", ifelse(temp==0, "+","")) paste('(', format(x[,1]), ',', format(x[,2]), temp, ']', sep='') }, "left" ={ temp <- x[,2] temp <- ifelse(is.na(temp), "?", ifelse(temp==0, "<"," ")) paste(temp, format(x[,1]), sep='') }, "interval"= { stat <- x[,3] temp <- c("+", "", "-", "]")[stat+1] temp2 <- ifelse(stat==3, paste("[", format(x[,1]), ", ",format(x[,2]), sep=''), format(x[,1])) ifelse(is.na(stat), "NA", paste(temp2, temp, sep='')) }, "mright" = { #multi-state temp <- x[,2] end <- c("+", paste(":", attr(x, "states"), sep='')) #endpoint temp <- ifelse(is.na(temp), "?", end[temp+1]) paste(format(x[,1]), temp, sep='') }, "mcounting"= { temp <- x[,3] end <- c("+", paste(":", attr(x, "states"), sep='')) #endpoint temp <- ifelse(is.na(temp), "?", end[temp+1]) paste('(', format(x[,1]), ',', format(x[,2]), temp, ']', sep='') }) } "[.Surv" <- function(x, i, j, drop=FALSE) { # If only 1 subscript is given, the result will still be a Surv object, # and the drop argument is ignored. # I would argue that x[3:4,,drop=FALSE] should return a matrix, since # the user has implicitly specified that they want a matrix. # However, [.dataframe calls [.Surv with the extra comma; its # behavior drives the choice of default. if (missing(j)) { xattr <- attributes(x) x <- unclass(x)[i,, drop=FALSE] # treat it as a matrix: handles dimnames attr(x, 'type') <- xattr$type if (!is.null(xattr$states)) attr(x, "states") <- xattr$states if (!is.null(xattr$inputAttributes)) { # If I see "names" subscript it, leave all else alone attr(x, 'inputAttributes') <- lapply(xattr$inputAttributes, function(z) { if (any(names(z)=="names")) z$names <- z$names[i] z }) } class(x) <- "Surv" #restore the class x } else { # return a matrix or vector if (is.R()) class(x) <- 'matrix' else oldClass(x) <- NULL NextMethod("[") } } is.na.Surv <- function(x) { as.vector(rowSums(is.na(unclass(x))) >0) } Math.Surv <- function(...) stop("Invalid operation on a survival time") Ops.Surv <- function(...) stop("Invalid operation on a survival time") Summary.Surv<-function(...) stop("Invalid operation on a survival time") is.Surv <- function(x) inherits(x, 'Surv') as.matrix.Surv <- function(x, ...) { y <- unclass(x) attr(y, "type") <- NULL attr(y, "states") <- NULL attr(y, "inputAttributes") <- NULL y } survival/R/is.na.coxph.penalty.S0000644000175100001440000000216211732700061016271 0ustar hornikusers# $Id: is.na.coxph.penalty.S 11447 2010-11-12 15:10:18Z therneau $ # The subscript function for coxph.penalty objects # without it the "subset" arg of a model statement tosses # away all of the attributes # "[.coxph.penalty" <- function(x, ..., drop=FALSE) { attlist <- attributes(x) attributes(x) <- attlist[match(c('dim', 'dimnames', 'levels', 'class'), names(attlist), 0)] x <- NextMethod('[') #let the default method do actual subscripting # Tack back on all of the old attributes, except dim and dimnames # which will have been properly modified by the standard [ method, # "levels" which may have dropped some, and "class" which is special attributes(x) <- c(attributes(x), attlist[is.na(match(names(attlist), c("dim", "dimnames", "levels", "class")))]) # The class will have lost it's first level oldClass(x) <- attlist$class return(x) } is.na.coxph.penalty <- function(x) { if (is.matrix(x)) is.na(c(unclass(x) %*% rep(1,ncol(x)))) else is.na(unclass(x)) } survival/R/neardate.R0000644000175100001440000000543612461741246014271 0ustar hornikusers# Create a "nearest date" index # date1: the trial date # date2: target to match to # # result: an index vector for data set 1, which shows the row in data set # 2 that has the same id, and the best date. # # best = "after" The closest date in #2 that is on or after the date in #1 # "prior" The closest date in #2 that is on or before the date in #1 # neardate <- function(id1, id2, y1, y2, best=c("after", "prior"), nomatch=NA_integer_) { if (missing(id1)) stop("id1 argument is required") if (missing(id2)) stop("id2 argument is required") if (missing(y1)) stop("y1 argument is required") if (missing(y2)) stop("y2 argument is required") if (length(id1) != length(y1)) stop("id1 and y1 have different lengths") if (length(id2) != length(y2)) stop("id2 and y2 have different lengths") best <- match.arg(best) # This check could be more sophisticated (though I don't see how to do it) # We want to make sure that the "alldate" line below makes sense for the # data types that the user passed in. if (is.factor(y1) || is.factor(y2)) stop("y1 and y2 must be sortable") if (inherits(y1, 'POSIXt')) if (!inherits(y2, 'POSIXt')) y2 <- as(y2, class(y1)) else if (inherits(y2, 'POSIXt')) y1 <- as(y1, class(y2)) alldate <- sort(unique(c(y1, y2))) y1 <- match(y1, alldate) y2 <- match(y2, alldate) # Throw out lines with missing y2, but remember which ones rowid <- 1:length(y2) if (any(is.na(y2))) { toss <- is.na(y2) y2 <- y2[!toss] if (!missing(id2)) id2 <- id2[!toss] rowid <- rowid[!toss] } n2 <- length(y2) if (n2 ==0) stop("No valid entries in data set 2") # Toss out any rows in id2 that are not possible targets for id1 # (id2 is usually the larger data set, thinning speeds it up) indx1 <- match(id2, id1) toss <- is.na(indx1) if (any(toss)) { id2 <- id2[!toss] y2 <- y2[!toss] indx1 <- indx1[!toss] rowid <- rowid[!toss] } n2 <- length(y2) if (n2 ==0) stop("No valid entries in data set 2") # We need to create a merging id. A minimal amount of # spread for the dates keeps numeric overflow at bay delta <- 1.0 + length(alldate) #numeric, not integer, on purpose hash1 <- match(id1, id1)*delta + y1 hash2 <- indx1*delta + y2 if (best=="prior") indx2 <- approx(hash2, 1:n2, hash1, method="constant", yleft=NA, yright=n2, rule=2, f=0)$y else indx2 <- approx(hash2, 1:n2, hash1, method="constant", yleft=1, yright=NA, rule=2, f=1)$y rowid[ifelse(id1== id2[indx2], indx2, NA)] } survival/R/pspline.S0000644000175100001440000001716312441062515014152 0ustar hornikusers# # the p-spline function for a Cox model # pspline <- function(x, df=4, theta, nterm=2.5*df, degree=3, eps=0.1, method, Boundary.knots=range(x), intercept=FALSE, penalty=TRUE, ...) { if (!missing(theta)) { method <- 'fixed' if (theta <=0 || theta >=1) stop("Invalid value for theta") } else if (df ==0 || (!missing(method) && method=='aic')) { method <- 'aic' nterm <- 15 #will be ok for up to 6-8 df if (missing(eps)) eps <- 1e-5 } else { method <- 'df' if (df <=1) stop ('Too few degrees of freedom') # The below used to say "df+1 > nterm", but we need some scope for # the smoother parameter to avoid strange conditions if (df > nterm) stop("`nterm' too small for df=",df) } xname <- deparse(substitute(x)) keepx <- !is.na(x) if (!all(keepx)) x <- x[keepx] #this is done before any reference to # Boundary.knots, so the default works nterm <- round(nterm) if (nterm < 3) stop("Too few basis functions") if (!missing(Boundary.knots)) { if (!is.numeric(Boundary.knots) || length(Boundary.knots) !=2 || Boundary.knots[1] >= Boundary.knots[2]) stop("Invalid values for Boundary.knots") # Check for data values outside the knot range outl <- (x < Boundary.knots[1]) outr<- (x > Boundary.knots[2]) outside <- outl | outr } else outside <- FALSE # Set up the evenly spaced knots dx <- (Boundary.knots[2] - Boundary.knots[1])/nterm knots <- c(Boundary.knots[1] + dx*((-degree):(nterm-1)), Boundary.knots[2]+ dx*(0:degree)) # Set up the basis. Inside the boundary knots we use spline.des. # Outside of them we use f(edge) + (x-edge)* f'(edge) if (any(outside)) { newx <- matrix(0., length(x), nterm + degree) if (any(outl)) { tt <- spline.des(knots, Boundary.knots[c(1,1)], degree+1, 0:1) newx[outl,] <- cbind(1, x[outl] - Boundary.knots[1]) %*% tt$design } if (any(outr)) { tt <- spline.des(knots, Boundary.knots[c(2,2)], degree+1, 0:1) newx[outr,] <- cbind(1, x[outr] - Boundary.knots[2]) %*% tt$design } if (any(inside <- !outside)) newx[inside,] <- spline.des(knots, x[inside], degree+1)$design } else newx <- spline.des(knots, x, degree+1, outer.ok=TRUE)$design # put missings back in so that the number of rows is right if (!all(keepx)) { temp <- matrix(NA, length(keepx), ncol(newx)) temp[keepx,] <- newx newx <- temp } nvar <- ncol(newx) #should be nterm + degree dmat <- diag(nvar) dmat <- apply(dmat, 2, diff, 1, 2) dmat <- t(dmat) %*% dmat if (intercept) xnames <-paste('ps(', xname, ')', 1:nvar, sep='') else { newx <- newx[,-1, drop=FALSE] dmat <- dmat[-1,-1, drop=FALSE] # rows corresponding to the 0 coef xnames <-paste('ps(', xname, ')', 1+ 2:nvar, sep='') } if (!penalty) { attributes(newx) <- c(attributes(newx), list(intercept=intercept, nterm=nterm, Boundary.knots=Boundary.knots)) if (is.R()) class(newx) <- "pspline" else oldClass(newx) <- "pspline" return(newx) } pfun <- function(coef, theta, n, dmat) { if (theta >=1) list(penalty= 100*(1-theta), flag=TRUE) else { if (theta <= 0) lambda <- 0 else lambda <- theta / (1-theta) list(penalty= c(coef %*% dmat %*% coef) * lambda/2, first = c(dmat %*% coef) * lambda , second = c(dmat * lambda), flag=FALSE ) } } printfun <- function(coef, var, var2, df, history, cbase) { test1 <- coxph.wtest(var, coef)$test # cbase contains the centers of the basis functions # do a weighted regression of these on the coefs to get a slope xmat <- cbind(1, cbase) xsig <- coxph.wtest(var, xmat)$solve # V X , where V = g-inverse(var) # [X' V X]^{-1} X' V cmat <- coxph.wtest(t(xmat)%*% xsig, t(xsig))$solve[2,] linear <- sum(cmat * coef) lvar1 <- c(cmat %*% var %*% cmat) lvar2 <- c(cmat %*% var2%*% cmat) test2 <- linear^2 / lvar1 # the "max(.5, df-1)" below stops silly (small) p-values for a # chisq of 0 on 0 df, when using AIC gives theta near 1 cmat <- rbind(c(linear, sqrt(lvar1), sqrt(lvar2), test2, 1, 1-pchisq(test2, 1)), c(NA, NA, NA, test1-test2, df-1, 1-pchisq(test1-test2, max(.5,df-1)))) dimnames(cmat) <- list(c("linear", "nonlin"), NULL) nn <- nrow(history$thetas) if (length(nn)) theta <- history$thetas[nn,1] else theta <- history$theta list(coef=cmat, history=paste("Theta=", format(theta))) } if (is.R()) { # The printfun needs to remember the spline's knots, # but I don't need (or want) to carry around the entire upteen # variables defined here as an environment # So fill in defaults for the cbase argument, and # force the function's environment to simplicity (amnesia) temp <- formals(printfun) temp$cbase <- knots[2:nvar] + (Boundary.knots[1] -knots[1]) formals(printfun) <- temp environment(printfun) <- .GlobalEnv } else { # Somewhat simpler in Splus, but because it depends on the # undocumented manner in which functions are stored, it might # stop working one day printfun[[6]] <- knots[2:nvar] + (Boundary.knots[1] - knots[1]) } if (method=='fixed') { temp <- list(pfun=pfun, printfun=printfun, pparm=dmat, diag =FALSE, cparm=list(theta=theta), varname=xnames, cfun = function(parms, iter, old) list(theta=parms$theta, done=TRUE)) } else if (method=='df') { temp <- list(pfun=pfun, printfun=printfun, diag =FALSE, cargs=('df'), cparm=list(df=df, eps=eps, thetas=c(1,0), dfs=c(1, nterm), guess=1 - df/nterm, ...), pparm= dmat, varname=xnames, cfun = frailty.controldf) } else { # use AIC temp <- list(pfun=pfun, printfun=printfun, pparm=dmat, diag =FALSE, cargs = c('neff', 'df', 'plik'), cparm=list(eps=eps, init=c(.5, .95), lower=0, upper=1, ...), varname=xnames, cfun = frailty.controlaic) } attributes(newx) <- c(attributes(newx), temp, list(intercept=intercept, nterm=nterm, Boundary.knots=Boundary.knots)) if (is.R()) class(newx) <- c("pspline", 'coxph.penalty') else oldClass(newx) <- 'coxph.penalty' newx } makepredictcall.pspline <- function(var, call) { if (call[[1]] != as.name("pspline")) return(call) #wrong phone number newcall <- call[1:2] #don't let the user override anything at <- attributes(var)[c("nterm", "intercept", "Boundary.knots")] newcall[names(at)] <- at newcall } predict.pspline <- function(object, newx, ...) { if (missing(newx)) return(object) a <- c(list(x=newx, penalty=FALSE), attributes(object)[c("intercept, Boundary.knots")]) do.call("pspline", a) } # Given a pspline basis, recover x psplineinverse <- function(x) { if (!inherits(x, "pspline")) stop("Argment must be the result of a call to pspline") intercept <- attr(x, "intercept") knots <- attr(x, "knots") nknot <- length(knots) if (!intercept) { indx <- 1:(ncol(x)+1) + (nknot- (ncol(x) +1))/2 as.vector(cbind(1-rowSums(x), x) %*% knots[indx]) } else { indx <- 1:ncol(x) + (nknot - ncol(x))/2 as.vector(x %*% knots) } } as.matrix.pspline <- function(x, ...) { temp <- attributes(x) attributes(x) <- temp['dim'] x } survival/R/survfitCI.R0000644000175100001440000002461512533653610014421 0ustar hornikusers# Automatically generated from all.nw using noweb survfitCI <- function(X, Y, weights, id, istate, type=c('kaplan-meier', 'fleming-harrington', 'fh2'), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified')){ method <- match.arg(type) # error <- match.arg(error) # if (error != "inf") # warning("Only the infinetesimal jackknife error is supported for CI curves") conf.type <- match.arg(conf.type) conf.lower<- match.arg(conf.lower) if (is.logical(conf.int)) { # A common error is for users to use "conf.int = FALSE" # it's illegal per documentation, but be kind if (!conf.int) conf.type <- "none" conf.int <- .95 } type <- attr(Y, "type") if (type !='mright' && type!='mcounting' && type != "right" && type != "counting") stop(paste("Cumulative incidence computation doesn't support \"", type, "\" survival data", sep='')) n <- nrow(Y) status <- Y[,ncol(Y)] ncurve <- length(levels(X)) state.names <- attr(Y, "states") if (missing(istate) || is.null(istate)) istate <- rep(0L, n) else if (is.factor(istate) || is.character(istate)) { # Match levels with the survival variable temp <- as.factor(istate) # append any starting states not found in Y, but remember that # if istate was a factor then not all its levels might appear appear <- (levels(istate))[unique(as.numeric(istate))] state.names <- unique(c(attr(Y, "states"), appear)) istate <- as.numeric(factor(as.character(istate), levels=state.names)) } else if (!is.numeric(istate) || any(istate != floor(istate))) stop("istate should be a vector of integers or a factor") if (length(id) ==0) id <- 1:n # these next two lines should be impossible, since istate came from the data frame if (length(istate) ==1) istate <- rep(istate,n) if (length(istate) !=n) stop ("wrong length for istate") states <- sort(unique(c(istate, 1:length(attr(Y, "states"))))) #list of all docurve2 <- function(entry, etime, status, istate, wt, states, id, se.fit) { # # round off error can cause trouble, if two times are within machine # precsion # solve this by using creating a factor ftime <- factor(c(entry,etime)) ltime <- levels(ftime) ftime <- matrix(as.integer(ftime), ncol=2) timeset <- as.numeric(ltime[sort(unique(ftime[,2]))]) #unique event times nstate <- length(states) uid <- sort(unique(id)) P <- as.vector(tapply(wt, factor(istate, levels=states), sum) / sum(wt)) P <- ifelse(is.na(P), 0, P) # initial probability distribution cstate <- istate[match(uid, id)] #current state for each observation storage.mode(wt) <- "double" # just in case someone had integer weights storage.mode(cstate) <- "integer" storage.mode(status) <- "integer" # C code has 0 based subscripts fit <- .Call(Csurvfitci, ftime, order(ftime[,1]) - 1L, order(ftime[,2]) - 1L, length(timeset), status, cstate - 1L, wt, match(id, uid) -1L, P, as.integer(se.fit)) prev0 <- table(factor(cstate, levels=states), exclude=NA)/length(cstate) if (se.fit) list(time=timeset, pmat=t(fit$p), std=sqrt(t(fit$var)), n.risk = colSums(fit$nrisk),n.event = fit$nevent, n.censor=fit$ncensor, prev0 = prev0, cumhaz=array(fit$cumhaz, dim=c(nstate,nstate, length(timeset)))) else list(time=timeset, pmat=t(fit$p), n.risk = colSums(fit$nrisk),n.event = fit$nevent, n.censor=fit$ncensor, prev0=prev0, cumhaz=array(fit$cumhaz, dim=c(nstate,nstate, length(timeset)))) } if (any(states==0)) { state0 <- TRUE states <- states + 1 istate <- istate + 1 status <- ifelse(status==0, 0, status+1) } else state0 <- FALSE curves <- vector("list", ncurve) names(curves) <- levels(X) if (ncol(Y)==2) { # 1 transition per subject indx <- which(status == istate & status!=0) if (length(indx)) { warning("an observation transitions to it's starting state, transition ignored") status[indx] <- 0 } if (length(id) && any(duplicated(id))) stop("Cannot have duplicate id values with (time, status) data") # dummy entry time that is < any event time entry <- rep(min(-1, 2*min(Y[,1])-1), n) for (i in levels(X)) { indx <- which(X==i) # temp <- docurve1(entry[indx], Y[indx,1], status[indx], # istate[indx], weights[indx], states, # id[indx]) curves[[i]] <- docurve2(entry[indx], Y[indx,1], status[indx], istate[indx], weights[indx], states, id[indx], se.fit) } } else { if (missing(id) || is.null(id)) stop("the id argument is required for start:stop data") indx <- order(id, Y[,2]) #ordered event times within subject indx1 <- c(NA, indx) #a pair of lagged indices indx2 <- c(indx, NA) same <- (id[indx1] == id[indx2] & !is.na(indx1) & !is.na(indx2)) #indx1, indx2= same id? if (any(same & X[indx1] != X[indx2])) { who <- 1 + min(which(same & X[indx1] != X[indx2])) stop("subject is in two different groups, id ", (id[indx1])[who]) } if (any(same & Y[indx1,2] != Y[indx2,1])) { who <- 1 + min(which(same & Y[indx1,2] != Y[indx2,1])) stop("gap in follow-up, id ", (id[indx1])[who]) } if (any(Y[,1] == Y[,2])) stop("cannot have start time == stop time") if (any(same & Y[indx1,3] == Y[indx2,3] & Y[indx1,3] !=0)) { who <- 1 + min(which(same & Y[indx1,1] != Y[indx2,2])) warning("subject changes to the same state, id ", (id[indx1])[who]) } if (any(same & weights[indx1] != weights[indx2])) { who <- 1 + min(which(same & weights[indx1] != weights[indx2])) stop("subject changes case weights, id ", (id[indx1])[who]) } # We only want to pay attention to the istate variable for the very first # observation of any given subject, but the program logic does better with # a full one. So construct one that will do this indx <- order(Y[,2]) uid <- unique(id) temp <- (istate[indx])[match(uid, id[indx])] #first istate for each subject istate <- temp[match(id, uid)] #replicate it to full length # Now to work for (i in levels(X)) { indx <- which(X==i) # temp <- docurve1(Y[indx,1], Y[indx,2], status[indx], # istate[indx], weights[indx], states, id[indx]) curves[[i]] <- docurve2(Y[indx,1], Y[indx,2], status[indx], istate[indx], weights[indx], states, id[indx], se.fit) } } # Turn the result into a survfit type object grabit <- function(clist, element) { temp <-(clist[[1]][[element]]) if (is.matrix(temp)) { nc <- ncol(temp) matrix(unlist(lapply(clist, function(x) t(x[[element]]))), byrow=T, ncol=nc) } else { xx <- as.vector(unlist(lapply(clist, function(x) x[element]))) if (class(temp)=="table") matrix(xx, byrow=T, ncol=length(temp)) else xx } } kfit <- list(n = as.vector(table(X)), time = grabit(curves, "time"), n.risk= grabit(curves, "n.risk"), n.event= grabit(curves, "n.event"), n.censor=grabit(curves, "n.censor"), prev = grabit(curves, "pmat"), prev0 = grabit(curves, "prev0")) nstate <- length(states) kfit$cumhaz <- array(unlist(lapply(curves, function(x) x$cumhaz)), dim=c(nstate, nstate, length(kfit$time))) if (length(curves) >1) kfit$strata <- unlist(lapply(curves, function(x) length(x$time))) if (se.fit) kfit$std.err <- grabit(curves, "std") # if state 0 was present, remove it if (state0) { kfit$prev <- kfit$prev[,-1] if (se.fit) kfit$std.err <- kfit$std.err[,-1] kfit$prev0 <- kfit$prev0[,-1] } # # Last bit: add in the confidence bands: # modeled on survfit.km, though for P instead of S # # if (se.fit) { std.err <- kfit$std.err zval <- qnorm(1- (1-conf.int)/2, 0,1) surv <- 1-kfit$prev if (conf.type=='plain') { temp <- zval* std.err kfit <- c(kfit, list(lower =pmax(kfit$prev-temp, 0), upper=pmin(kfit$prev+temp, 1), conf.type='plain', conf.int=conf.int)) } if (conf.type=='log') { #avoid some "log(0)" messages xx <- ifelse(kfit$prev==1, 1, 1- kfit$prev) temp1 <- ifelse(surv==0, NA, exp(log(xx) + zval* std.err/xx)) temp2 <- ifelse(surv==0, NA, exp(log(xx) - zval* std.err/xx)) kfit <- c(kfit, list(lower=pmax(1-temp1,0), upper= 1- temp2, conf.type='log', conf.int=conf.int)) } if (conf.type=='log-log') { who <- (surv==0 | surv==1) #special cases temp3 <- ifelse(surv==0, NA, 1) xx <- ifelse(who, .1,kfit$surv) #avoid some "log(0)" messages temp1 <- exp(-exp(log(-log(xx)) + zval*std.err/(xx*log(xx)))) temp1 <- ifelse(who, temp3, temp1) temp2 <- exp(-exp(log(-log(xx)) - zval*std.err/(xx*log(xx)))) temp2 <- ifelse(who, temp3, temp2) kfit <- c(kfit, list(lower=1-temp1, upper=1-temp2, conf.type='log-log', conf.int=conf.int)) } } kfit$states <- state.names kfit$type <- attr(Y, "type") kfit } survival/R/quantile.survfit.R0000644000175100001440000002644312112666037016027 0ustar hornikusers# # quantile function for survfit objects # # First a little function to find quantiles in a CDF # curve. It would be a trivial use of approx, except that # once in a while the survival curve has a flat spot exactly # at the requested quantile. Then we use the median of the # flat. findq <- function(x, y, p, tol) { # This case occurs for a survival curve whose upper limit never drops below 1 if (max(y, na.rm=T) < min(p)) return(rep(NA, length(p))) # Remove duplicate y values, i.e., the censors, since dups cause # issues for approx xmax <- x[length(x)] dups <- duplicated(y) if (any(dups)) { x <- x[!dups] y <- y[!dups] } n <- length(y) # quantile = where a horzontal line at p intercects the curve. At each # x the curve of 1-y jumps up to a new level # The most work is to check for horizontal lines in the survival # curve that match one of our quantiles within tolerance. If any # p matches, then our quantile is the average of the given x and # the x value of the next jump point, i.e., the usual midpoint rule # used for medians. # A flat at the end of the curve is a special case, as is the quantile # of 0. indx1 <- approx(y+tol, 1:n, p, method="constant", f=1)$y indx2 <- approx(y-tol, 1:n, p, method="constant", f=1)$y quant <- (x[indx1] + x[indx2])/2 quant[p==0] <- x[1] if (!is.na(y[n])) { lastpt <- (abs(p- y[n]) < tol) # end of the curve if (any(lastpt)) quant[lastpt] <- (x[indx1[lastpt]] + xmax)/2 } quant } doquant <- function(p, time, surv, upper, lower, firstx, tol) { qq <- findq(c(firstx,time), c(0, 1-surv), p, tol) # browser() if (missing(upper)) qq else rbind(qq, findq(c(firstx, time), c(0, 1-lower), p, tol), findq(c(firstx, time), c(0, 1-upper), p, tol)) } quantile.survfit <- function(x, probs=c(.25, .5, .75), conf.int=TRUE, tolerance= sqrt(.Machine$double.eps), ...) { if (!inherits(x, "survfit")) stop("Must be a survfit object") if (any(!is.numeric(probs)) || any(is.na(probs))) stop("invalid probability") if (any(probs <0 | probs >1)) stop("Invalid probability") if (is.null(x$lower)) conf.int <- FALSE nprob <- length(probs) pname <- format(probs*100) # What do we report for p=0? Use x$start.time if it exists, 0 otherwise xmin <- if (is.null(x$start.time)) 0 else x$start.time # There are 8 cases: strata yes/no # ncol(x$surv) =1 or >1 # conf.int = T/F if (is.null(x$strata)) { if (is.matrix(x$surv) && ncol(x$surv) >1) { qmat <- matrix(0., ncol=nprob, nrow=ncol(x$surv)) dimnames(qmat) <- list(dimnames(x$surv)[[2]], pname) if (conf.int) { qupper <- qlower <- qmat for (i in 1:ncol(x$surv)) { temp <- doquant(probs, x$time, x$surv[,i], x$upper[,i], x$lower[,i], xmin, tolerance) qmat[i,] <- temp[1,] qupper[i,] <- temp[3,] qlower[i,] <- temp[2,] } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (i in 1:ncol(x$surv)) qmat[i,] <- doquant(probs, x$time, x$surv[,i], firstx=xmin, tol=tolerance) return(qmat) } } else { # No strata and no matrix if (conf.int) { temp <- doquant(probs, x$time, x$surv, x$upper, x$lower, xmin, tolerance) dimnames(temp) <- list(NULL, pname) return(list(quantile=temp[1,], lower=temp[2,], upper=temp[3,])) } else { temp <- doquant(probs, x$time, x$surv, firstx=xmin, tol =tolerance) names(temp) <- pname return(temp) } } } else { nstrat <- length(x$strata) if (is.matrix(x$surv) && ncol(x$surv) >1) { # uncommon case, e.g., predicted survivals from a Cox model # return an array with strata as the first dimension, and # the probabilites as the third. qmat <- array(0., dim=c(nstrat, ncol(x$surv), nprob)) dimnames(qmat) <-list(names(x$strata), dimnames(x$surv)[[2]], pname) if (conf.int) { qupper <- qlower <- qmat for (strat in 1:nstrat) { z <- x[strat,] for (i in 1:ncol(z$surv)) { temp <- doquant(probs, z$time, z$surv[,i], z$upper[,i], z$lower[,i], xmin,tolerance) qmat[strat,i,] <- temp[1,] qupper[strat,i,] <- temp[3,] qlower[strat,i,] <- temp[2,] } } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (strat in 1:nstrat) { z <- x[strat] for (i in 1:ncol(z$surv)) qmat[strat,i,] <- doquant(probs, z$time, z$surv[,i], firstx=xmin, tol=tolerance) } return(qmat) } } else { # Only a strata, the most common case qmat <- matrix(0., nstrat, nprob) dimnames(qmat) <- list(names(x$strata), pname) if (conf.int) { qupper <- qlower <- qmat for (i in 1:nstrat) { z <- x[i] temp <- doquant(probs, z$time, z$surv, z$upper, z$lower, xmin, tolerance) qmat[i,] <- temp[1,] qupper[i,] <- temp[3,] qlower[i,] <- temp[2,] } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (i in 1:nstrat) { z <- x[i] qmat[i,] <- doquant(probs, z$time, z$surv, firstx=xmin, tol = tolerance) } return(qmat) } } } } # Why can't I just fudge the object and call quantile.survfit? Because # the code below uses subscripted objects, and the class of the chimeric # object doesn't work out for that operation. Also, we want to use the # state names in the dimnames of the result. # quantile.survfitms <- function(x, probs=c(.25, .5, .75), conf.int=TRUE, tolerance= sqrt(.Machine$double.eps), ...) { if (any(!is.numeric(probs)) || any(is.na(probs))) stop("invalid probability") if (any(probs <0 | probs >1)) stop("Invalid probability") if (is.null(x$lower)) conf.int <- FALSE nprob <- length(probs) pname <- format(probs*100) if (is.null(x$start.time)) xmin<-0 else xmin <- x$start.time # There are 8 cases: strata yes/no # ncol(x$surv) =1 or >1 # conf.int = T/F if (is.null(x$strata)) { if (is.matrix(x$prev) && ncol(x$prev) >1) { qmat <- matrix(0., ncol=nprob, nrow=ncol(x$prev)) dimnames(qmat) <- list(x$states, pname) if (conf.int) { qupper <- qlower <- qmat for (i in 1:ncol(x$prev)) { temp <- doquant(probs, x$time, 1-x$prev[,i], 1-x$lower[,i], 1-x$upper[,i], xmin, tolerance) qmat[i,] <- temp[1,] qupper[i,] <- temp[3,] qlower[i,] <- temp[2,] } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (i in 1:ncol(x$prev)) qmat[i,] <- doquant(probs, x$time, 1-x$prev[,i], firstx=xmin, tol=tolerance) return(qmat) } } else { # No strata and no matrix if (conf.int) { temp <- doquant(probs, x$time, 1-x$prev, 1-x$lower, 1-x$upper, firstx=xmin, tol=tolerance) dimnames(temp) <- list(NULL, pname) return(list(quantile=temp[1,], lower=temp[2,], upper=temp[3,])) } else { temp <- doquant(probs, x$time, 1-x$prev, firstx=xmin, tol=tolerance) names(temp) <- pname return(temp) } } } else { nstrat <- length(x$strata) if (is.matrix(x$prev) && ncol(x$prev) >1) { # the most common case # Return an array with strata as the first dimension, state as # the second, and the probabilites as the third. The reason # for this order is that then # (quantile(fit))[i,j,] = quantile(fit[i,j]) qmat <- array(0., dim=c(nstrat, ncol(x$prev), nprob)) dimnames(qmat) <-list(names(x$strata), x$states, pname) if (conf.int) { qupper <- qlower <- qmat for (strat in 1:nstrat) { z <- x[strat] for (i in 1:ncol(z$prev)) { temp <- doquant(probs, z$time, 1-z$prev[,i], 1-z$lower[,i], 1-z$upper[,i], xmin, tolerance) qmat[strat,i,] <- temp[1,] qupper[strat,i,] <- temp[3,] qlower[strat,i,] <- temp[2,] } } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (strat in 1:nstrat) { z <- x[strat] for (i in 1:ncol(z$prev)) qmat[strat,i,] <- doquant(probs, z$time, 1-z$prev[,i], firstx= xmin, tol=tolerance) } return(qmat) } } else { # Only a strata, which will be a rare case. Perhaps someone # typed "quantile(fit[,2], .4)" qmat <- matrix(0., nstrat, nprob) dimnames(qmat) <- list(names(x$strata), pname) if (conf.int) { qupper <- qlower <- qmat for (i in 1:nstrat) { z <- x[i] temp <- doquant(probs, z$time, 1-z$prev, 1-z$lower, 1-z$upper, xmin, tolerance) qmat[i,] <- temp[1,] qupper[i,] <- temp[3,] qlower[i,] <- temp[2,] } return(list(quantile=qmat, lower=qlower, upper=qupper)) } else { for (i in 1:nstrat) { z <- x[i] qmat[i,] <- doquant(probs, z$time, 1-z$prev, firstx=xmin, tol = tolerance) } return(qmat) } } } } survival/R/ratetable.S0000644000175100001440000001200211732700061014423 0ustar hornikusers# $Id: ratetable.S 11389 2010-02-08 22:53:51Z therneau $ # # This source file has two distinct parts in it. The first is the # ratetable(), which is used inside pyears and survexp only to allow # users to match the names of variables in their data set to the names # of the dimensions in a ratetable. It returns a matrix with one # column for each argument; usually that argument will be a vector but # may also be a single constant. The result has a class "ratetable2", # whose only purpose is to allow na.action functions to work properly. # # The second part of the file are the methods for actual rate tables, like # the table of US survival rates by age and sex (survexp.us). Rate tables # have the "ratetable" class. However, since each one is rather unique, # there is no function to create a rate table. Each consists of a multi-way # array of event rates along with a set of attributes. # # The ideal for this function would be # ratetable <- function(...) data.frame(...) # Then missing, subsets, etc would all be fine, yet the variables would still # be special in the terms result so I could find them. But -- the only # multi-column objects that model.frame will accept are matrices. So I # make a data frame (both factors and numerics) that looks like a matrix. # ratetable <- function(...) { args <- list(...) nargs <- length(args) ll <- sapply(args, length) n <- max(ll) # We assume this is the dimension of the user's data frame levlist <- vector("list", nargs) isDate <- rep(FALSE, nargs) x <- matrix(0,n,nargs) dimnames(x) <- list(1:n, names(args)) for (i in 1:nargs) { if (ll[i] ==1) args[[i]] <- rep(args[[i]], n) else if (ll[i] != n) stop(paste("Aguments do not all have the same length (arg ", i, ")", sep='')) # In Splus cut and tcut produce class 'category' if (inherits(args[[i]], 'cateogory') || is.character(args[[i]])) args[[i]] <- as.factor(args[[i]]) if (is.factor(args[[i]])) { levlist[[i]] <- levels(args[[i]]) x[,i] <- as.numeric(args[[i]]) # the vector of levels } else { temp <- ratetableDate(args[[i]]) if (is.null(temp)) x[,i] <- as.numeric(args[[i]]) else { x[,i] <- temp isDate[i] <- TRUE } } } attr(x, "isDate") <- isDate attr(x, "levlist") <- levlist if (is.R()) class(x) <- 'ratetable2' else oldClass(x) <- "ratetable2" x } # The two functions below should only be called internally, when missing # values cause model.frame to drop some rows is.na.ratetable2 <- function(x) { attributes(x) <- list(dim=dim(x)) as.vector((1 * is.na(x)) %*% rep(1, ncol(x)) >0) } "[.ratetable2" <- function(x, rows, cols, drop=FALSE) { if (!missing(cols)) { stop("This should never be called!") } aa <- attributes(x) attributes(x) <- aa[c("dim", "dimnames")] y <- x[rows,,drop=FALSE] attr(y,'isDate') <- aa$isDate attr(y,'levlist') <- aa$levlist if (is.R()) class(y) <- 'ratetable2' else oldClass(y) <- 'ratetable2' y } # # Functions to manipulate rate tables # "[.ratetable" <- function(x, ..., drop=TRUE) { aa <- attributes(x) attributes(x) <- aa[c("dim", "dimnames")] y <- NextMethod("[", drop=FALSE) newdim <- attr(y, 'dim') if (is.null(newdim)) return(y) #when the subscript was a single vector dropped <- (newdim==1) if (drop) change <- (newdim!=aa$dim & !dropped) else change <- (newdim!=aa$dim) if (any(change)) { #dims that got smaller, but not dropped newcut <- aa$cutpoints for (i in (1:length(change))[change]) if (!is.null(newcut[[i]])) newcut[[i]] <- (newcut[[i]])[match(dimnames(y)[[i]], aa$dimnames[[i]])] aa$cutpoints <- newcut } if (drop && any(dropped)){ if (all(dropped)) as.numeric(y) #single element else { #Note that we have to drop the summary function attributes(y) <- list( dim = dim(y)[!dropped], dimnames = dimnames(y)[!dropped], dimid = aa$dimid[!dropped], factor = aa$factor[!dropped], cutpoints =aa$cutpoints[!dropped], type = aa$type[!dropped]) if (is.R()) class(y) <- 'ratetable' else oldClass(y) <- 'ratetable' y } } else { aa$dim <- aa$dimnames <- NULL attributes(y) <- c(attributes(y), aa) y } } is.na.ratetable <- function(x) structure(is.na(as.vector(x)), dim=dim(x), dimnames=dimnames(x)) Math.ratetable <- function(x, ...) { attributes(x) <- attributes(x)[c("dim", "dimnames")] NextMethod(.Generic) } Ops.ratetable <- function(e1, e2) { #just treat it as an array if (nchar(.Method[1])) attributes(e1) <- attributes(e1)[c("dim","dimnames")] if (nchar(.Method[2])) attributes(e2) <- attributes(e2)[c("dim","dimnames")] NextMethod(.Generic) } as.matrix.ratetable <- function(x, ...) { attributes(x) <- attributes(x)[c("dim", "dimnames")] x } survival/R/cluster.S0000644000175100001440000000012011732700061014137 0ustar hornikusers# $Id: cluster.S 11166 2008-11-24 22:10:34Z therneau $ cluster <- function(x) x survival/R/aareg.taper.S0000644000175100001440000000223211732700061014655 0ustar hornikusers# SCCS $Id: aareg.taper.S 11166 2008-11-24 22:10:34Z therneau $ # # Do running averages of an information matrix # aareg.taper <- function(taper, imat, nevent) { dd <- dim(imat) if (length(taper)==0 || !is.numeric(taper) || any(taper <=0)) stop("Invalid taper vector") ntaper <- length(taper) ntime <- dd[3] if (ntaper > ntime) { taper <- taper[1:ntime] ntaper <- ntime } # # Turn imat into an array: 1 row per coef, one col per time # and then scale it by the number of events to get a variance # (coxph.detail returns imat = var(X) * nevents) # imat <- matrix(as.vector(imat), ncol=dd[3]) imat <- imat / rep(nevent, rep(dd[1]*dd[2], dd[3])) if (ntaper >1) { smoother <- matrix(0., ntime, ntime) tsum <- cumsum(rev(taper)) for (i in 1:ntaper) smoother[1:i, i] <- taper[seq(to=ntaper, length=i)]/tsum[i] if (ntaper < ntime) { for (i in (ntaper+1):ntime) smoother[seq(to=i, length=ntaper),i] <- taper/tsum[ntaper] } imat <- imat %*% smoother } array(imat, dim=dd) } survival/R/coxexact.fit.R0000644000175100001440000001024612504525636015101 0ustar hornikusers# This routine fits right censored data when the method is # "exact". The most common use for this option is matched # case-control data. coxexact.fit <- function(x, y, strata, offset, init, control, weights, method, rownames) { if (!is.matrix(x)) stop("Invalid formula for cox fitting function") if (!is.null(weights) && any(weights!=1)) stop("Case weights are not supported for the exact method") n <- nrow(x) nvar <- ncol(x) # The risk set addition in the C-code, which is the critically slow # part of the calculations, expects to have the data in sorted order: # (large to small times) within strata if (length(strata)==0) { sorted <- order(-y[,1]) newstrat <- as.integer(rep(0,n)) } else { sorted <- order(strata, -y[,1]) strata <- (as.numeric(strata))[sorted] newstrat <- as.integer(c(1, 1*(diff(strata)!=0))) } y <- y[sorted,] if (is.null(offset)) offset <- rep(0.,n) else offset <- offset[sorted] if (nvar==0) { # A special case: Null model. Trick the C code, which requires # at least one variable, by creating one and then doing 0 # iterations at beta=0 x <- matrix(1:n, ncol=1) init <- NULL maxiter <- 0 nullmodel <- TRUE nvar <- 1 } else { maxiter <- control$iter.max nullmodel <- FALSE } if (!is.null(init)) { if (length(init) != nvar) stop("Wrong length for inital values") } else init <- rep(0.,nvar) # Prescale the data set to improve numerical accuracy. # We will undo the scaling before finishing up. newx <- scale(x[sorted,]) # newx <- scale(x, scale=NULL) #debug rescale <- attr(newx, "scaled:scale") means <- attr(newx, "scaled:center") cfit <- .Call(Ccoxexact, as.integer(maxiter), as.double(y), # interger data? Just in case. newx, as.double(offset), as.integer(newstrat), as.double(init*rescale), as.double(control$eps), as.double(control$toler.chol) ) if (nullmodel) { score <- exp(offset[sorted]) cxres <- .C(Ccoxmart2, as.integer(n), as.double(y[,1]), as.integer(y[,2]), newstrat, score, rep(1.0, n), #weights resid=double(n)) resid <- double(n) resid[sorted] <- cxres$resid names(resid) <- rownames return( list(loglik = cfit$loglik[1], linear.predictors = offset, residuals = resid, method= c("coxph.null", "coxph"))) } loglik <- cfit$loglik[1:2] #these are packed into one vector sctest <- cfit$loglik[3] iter <- cfit$loglik[5] flag <- cfit$loglik[4] var <- matrix(cfit$imat,nvar,nvar) coef <- cfit$coef if (flag < nvar) which.sing <- diag(var)==0 else which.sing <- rep(FALSE,nvar) infs <- abs(cfit$u %*% var) if (control$iter.max >1) { if (flag == 1000) warning("Ran out of iterations and did not converge") else { infs <- ((infs > control$eps) & infs > control$toler.inf*abs(coef)) if (any(infs)) warning(paste("Loglik converged before variable ", paste((1:nvar)[infs],collapse=","), "; beta may be infinite. ")) } } names(coef) <- dimnames(x)[[2]] lp <- newx %*% coef + offset score <- as.double(exp(lp)) # Compute the residuals cxres <- .C(Ccoxmart2, as.integer(n), as.double(y[,1]), as.integer(y[,2]), newstrat, score, rep(1.0, n), #weights resid=double(n)) resid <- double(n) resid[sorted] <- cxres$resid names(resid) <- rownames coef[which.sing] <- NA lp.unsort <- double(n) lp.unsort[sorted] <- lp scmat <- diag(1/rescale, nvar,nvar) list(coefficients = coef/rescale, var = scmat %*% var %*% scmat, loglik = loglik, score = sctest, iter = iter, linear.predictors = lp.unsort, residuals = resid, means = means, method= 'coxph') } survival/R/anova.coxph.R0000644000175100001440000001106512534312164014717 0ustar hornikusers# The anova function for a coxph object anova.coxph <- function (object, ..., test = 'Chisq') { if (!inherits(object, "coxph")) stop ("argument must be a cox model") # All the ... args need to be coxph or coxme fits. If any of them # have a name attached, e.g., 'charlie=T' we assume a priori # that they are illegal # dotargs <- list(...) named <- if (is.null(names(dotargs))) rep(FALSE, length(dotargs)) else (names(dotargs) != "") if (any(named)) warning(paste("The following arguments to anova.coxph(..)", "are invalid and dropped:", paste(deparse(dotargs[named]), collapse = ", "))) dotargs <- dotargs[!named] if (length(dotargs) >0) { # Check that they are all cox or coxme models is.coxmodel <-unlist(lapply(dotargs, function(x) inherits(x, "coxph"))) is.coxme <- unlist(lapply(dotargs, function(x) inherits(x, "coxme"))) if (!all(is.coxmodel | is.coxme)) stop("All arguments must be Cox models") if (any(is.coxme)) { # We need the anova.coxmelist function from coxme # If coxme is not loaded the line below returns NULL temp <- getS3method("anova", "coxmelist", optional=TRUE) if (is.null(temp)) stop("a coxme model was found and library coxme is not loaded") else return(temp(c(list(object), dotargs), test = test)) } else return(anova.coxphlist(c(list(object), dotargs), test = test)) } # # The argument is a single Cox model # Show the nested list of models generated by this model. # By tradition the sequence is main effects (in the order found in # the model statement), then 2 way interactions, then 3, etc. # One does this by using the "assign" attribute of the model matrix. # (This does not work for penalized terms. if (length(object$rscore)>0) stop("Can't do anova tables with robust variances") has.strata <- !is.null(attr(terms(object), "specials")$strata) if (is.null(object[['y']]) || (has.strata && is.null(object$strata))) { # We need the model frame mf <- model.frame(object) Y <- model.response(mf) X <- model.matrix(object, mf) if (has.strata) { stemp <- untangle.specials(terms(object), "strata") if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } } else { Y <- object[['y']] X <- model.matrix(object) if (has.strata) strats <- object$strata } assign <- attr(X, 'assign') alevels <- sort(unique(assign)) #if there are strata the sequence has holes nmodel <- length(alevels) df <- integer(nmodel+1) #this will hold the df vector loglik <- double(nmodel+1) #and the loglike vector df[nmodel+1] <- if (is.null(object$df)) sum(!is.na(object$coefficients)) else sum(object$df) loglik[nmodel+1] <- object$loglik[2] df[1] <- 0 loglik[1] <- object$loglik[1] # Now refit intermediate models for (i in seq.int(1, length.out =nmodel-1)){ if (length(object$offset)) { if (has.strata) tfit <- coxph(Y ~ X[,assign <= alevels[i]] + strata(strats) + offset(object$offset)) else tfit <- coxph(Y ~ X[, assign<= alevels[i]] + offet(object$offset)) } else { if (has.strata) tfit <- coxph(Y ~ X[,assign <= alevels[i]] + strata(strats)) else tfit <- coxph(Y ~ X[,assign <= alevels[i]]) } df[i+1] <- sum(!is.na(tfit$coefficients)) loglik[i+1] <- tfit$loglik[2] } table <- data.frame(loglik=loglik, Chisq=c(NA, 2*diff(loglik)), Df=c(NA, diff(df))) if (nmodel == 0) #failsafe for a NULL model table <- table[1, , drop = FALSE] if (length(test) >0 && test[1]=='Chisq') { table[['Pr(>|Chi|)']] <- 1- pchisq(table$Chisq, table$Df) } temp <- attr(terms(object), "term.labels") if (has.strata) temp <- temp[-stemp$terms] row.names(table) <- c('NULL', temp) title <- paste("Analysis of Deviance Table\n Cox model: response is ", deparse(object$terms[[2]]), "\nTerms added sequentially (first to last)\n", sep = "") structure(table, heading = title, class = c("anova", "data.frame")) } survival/R/lines.survexp.R0000644000175100001440000000015212264766610015324 0ustar hornikuserslines.survexp <- function(x, type="l", ...) { type <- type NextMethod("lines", type=type, ...) } survival/R/plot.survfit.R0000644000175100001440000006214212533653602015160 0ustar hornikusers# Automatically generated from all.nw using noweb plot.survfit<- function(x, conf.int, mark.time=TRUE, mark=3, col=1,lty=1, lwd=1, cex=1, log=FALSE, xscale=1, yscale=1, firstx=0, firsty=1, xmax, ymin=0, fun, xlab="", ylab="", xaxs='S', ...) { dotnames <- names(list(...)) if (any(dotnames=='type')) stop("The graphical argument 'type' is not allowed") if (inherits(x, "survfitms")) { x$surv <- 1- x$prev if (is.matrix(x$surv)) dimnames(x$surv) <- list(NULL, x$states) if (!is.null(x$lower)) { x$lower <- 1- x$lower x$upper <- 1- x$upper } if (missing(fun)) fun <- "event" } if (missing(firsty) && !is.null(x$prev0)) firsty <- 1-x$prev0 if (is.logical(log)) { ylog <- log xlog <- FALSE if (ylog) logax <- 'y' else logax <- "" } else { ylog <- (log=='y' || log=='xy') xlog <- (log=='x' || log=='xy') logax <- log } if (!missing(fun)) { if (is.character(fun)) { if (fun=='log'|| fun=='logpct') ylog <- TRUE if (fun=='cloglog') { xlog <- TRUE if (ylog) logax <- 'xy' else logax <- 'x' } } } # The special x axis style only applies when firstx is not given if (missing(xaxs) && (firstx!=0 || !missing(fun) || (missing(fun) && inherits(x, "survfitms")))) xaxs <- par("xaxs") #use the default ssurv <- as.matrix(x$surv) stime <- x$time if( !is.null(x$upper)) { supper <- as.matrix(x$upper) slower <- as.matrix(x$lower) } else { conf.int <- FALSE supper <- NULL #marker for later code } # Two cases where we don't put marks at the censoring times if (inherits(x, 'survexp') || inherits(x, 'survfit.coxph')) { if (missing(mark.time)) mark.time <- FALSE } # set up strata if (is.null(x$strata)) { nstrat <- 1 stemp <- rep(1, length(x$time)) # same length as stime } else { nstrat <- length(x$strata) stemp <- rep(1:nstrat, x$strata) # same length as stime } ncurve <- nstrat * ncol(ssurv) firsty <- matrix(firsty, nrow=nstrat, ncol=ncol(ssurv)) if (!missing(xmax) && any(x$time>xmax)) { # prune back the survival curves # I need to replace x's over the limit with xmax, and y's over the # limit with either the prior y value or firsty keepx <- keepy <- NULL # lines to keep tempn <- table(stemp) offset <- cumsum(c(0, tempn)) for (i in 1:nstrat) { ttime <-stime[stemp==i] if (all(ttime <= xmax)) { keepx <- c(keepx, 1:tempn[i] + offset[i]) keepy <- c(keepy, 1:tempn[i] + offset[i]) } else { bad <- min((1:tempn[i])[ttime>xmax]) if (bad==1) { #lost them all if (!is.na(firstx)) { # and we are plotting lines keepy <- c(keepy, 1+offset[i]) ssurv[1+offset[i],] <- firsty[i,] } } else keepy<- c(keepy, c(1:(bad-1), bad-1) + offset[i]) keepx <- c(keepx, (1:bad)+offset[i]) stime[bad+offset[i]] <- xmax x$n.event[bad+offset[i]] <- 1 #don't plot a tick mark } } # ok, now actually prune it stime <- stime[keepx] stemp <- stemp[keepx] x$n.event <- x$n.event[keepx] if (!is.null(x$n.censor)) x$n.censor <- x$n.censor[keepx] ssurv <- ssurv[keepy,,drop=FALSE] if (!is.null(supper)) { supper <- supper[keepy,,drop=FALSE] slower <- slower[keepy,,drop=FALSE] } } stime <- stime/xscale #scaling is deferred until xmax processing is done if (!missing(fun)) { if (is.character(fun)) { tfun <- switch(fun, 'log' = function(x) x, 'event'=function(x) 1-x, 'cumhaz'=function(x) -log(x), 'cloglog'=function(x) log(-log(x)), 'pct' = function(x) x*100, 'logpct'= function(x) 100*x, #special case further below 'identity'= function(x) x, stop("Unrecognized function argument") ) } else if (is.function(fun)) tfun <- fun else stop("Invalid 'fun' argument") ssurv <- tfun(ssurv ) if (!is.null(supper)) { supper <- tfun(supper) slower <- tfun(slower) } firsty <- tfun(firsty) } if (missing(firstx)) { if (!is.null(x$start.time)) firstx <- x$start.time/xscale else { if (xlog) firstx <- min(stime[stime>0]) else firstx <- min(0, stime) } } # The default for plot and lines is to add confidence limits # if there is only one curve if (missing(conf.int)) conf.int <- (ncurve==1) if (is.logical(conf.int)) plot.surv <- TRUE else { temp <- match.arg(conf.int, c("both", "only", "none")) if (is.na(temp)) stop("invalid value for conf.int") if (temp=="none") conf.int <- FALSE else conf.int <- TRUE if (temp=="only") plot.surv <- FALSE else plot.surv <- TRUE } # Marks are not placed on confidence bands mark <- rep(mark, length.out=ncurve) mcol <- rep(col, length.out=ncurve) if (is.numeric(mark.time)) mark.time <- sort(mark.time) # The actual number of curves is ncurve*3 if there are confidence bands # If the number of line types is 1 and lty is an integer, then use lty # for the curve and lty+1 for the CI # If the length(lty) <= length(ncurve), use the same color for curve and CI # otherwise assume the user knows what they are about and has given a full # vector of line types. # Colors and line widths work like line types, excluding the +1 rule. if (conf.int) { if (length(lty)==1 && is.numeric(lty)) lty <- rep(c(lty, lty+1, lty+1), ncurve) else if (length(lty) <= ncurve) lty <- rep(rep(lty, each=3), length.out=(ncurve*3)) else lty <- rep(lty, length.out= ncurve*3) if (length(col) <= ncurve) col <- rep(rep(col, each=3), length.out=3*ncurve) else col <- rep(col, length.out=3*ncurve) if (length(lwd) <= ncurve) lwd <- rep(rep(lwd, each=3), length.out=3*ncurve) else lwd <- rep(lwd, length.out=3*ncurve) } else { col <- rep(col, length.out=ncurve) lty <- rep(lty, length.out=ncurve) lwd <- rep(lwd, length.out=ncurve) } #axis setting parmaters that depend on the fun argument if (!missing(fun)) { ymin <- tfun(ymin) #lines routine doesn't have it } # Do axis range computations if (xaxs=='S') { #special x- axis style for survival curves xaxs <- 'i' #what S thinks tempx <- max(stime) * 1.04 } else tempx <- max(stime) tempx <- c(firstx, tempx, firstx) if (ylog) { tempy <- range(ssurv[is.finite(ssurv)& ssurv>0]) if (tempy[2]==1) tempy[2] <- .99 if (any(ssurv==0)) { tempy[1] <- tempy[1]*.8 ssurv[ssurv==0] <- tempy[1] if (!is.null(supper)) { supper[supper==0] <- tempy[1] slower[slower==0] <- tempy[1] } } tempy <- c(tempy, firsty) } else tempy <- range(ssurv, firsty, finite=TRUE, na.rm=TRUE) if (missing(fun)) { tempx <- c(tempx, firstx) tempy <- c(tempy, ymin) } # # Draw the basic box # plot(range(tempx, finite=TRUE, na.rm=TRUE), range(tempy, finite=TRUE, na.rm=TRUE)*yscale, type='n', log=logax, xlab=xlab, ylab=ylab, xaxs=xaxs,...) if(yscale != 1) { if (ylog) par(usr =par("usr") -c(0, 0, log10(yscale), log10(yscale))) else par(usr =par("usr")/c(1, 1, yscale, yscale)) } # Create a step function, removing redundancies that sometimes occur in # curves with lots of censoring. dostep <- function(x,y) { keep <- is.finite(x) & is.finite(y) if (!any(keep)) return() #all points were infinite or NA if (!all(keep)) { # these won't plot anyway, so simplify (CI values are often NA) x <- x[keep] y <- y[keep] } n <- length(x) if (n==1) list(x=x, y=y) else if (n==2) list(x=x[c(1,2,2)], y=y[c(1,1,2)]) else { # replace verbose horizonal sequences like # (1, .2), (1.4, .2), (1.8, .2), (2.3, .2), (2.9, .2), (3, .1) # with (1, .2), (.3, .2),(3, .1). # They are slow, and can smear the looks of the line type. temp <- rle(y)$lengths drops <- 1 + cumsum(temp[-length(temp)]) # points where the curve drops #create a step function if (n %in% drops) { #the last point is a drop xrep <- c(x[1], rep(x[drops], each=2)) yrep <- rep(y[c(1,drops)], c(rep(2, length(drops)), 1)) } else { xrep <- c(x[1], rep(x[drops], each=2), x[n]) yrep <- c(rep(y[c(1,drops)], each=2)) } list(x=xrep, y=yrep) } } drawmark <- function(x, y, mark.time, censor, cex, ...) { if (!is.numeric(mark.time)) { xx <- x[censor] yy <- y[censor] } else { #interpolate xx <- mark.time yy <- approx(x, y, xx, method="constant", f=0)$y } points(xx, yy, cex=cex, ...) } plot.surv <- TRUE type <- 's' c1 <- 1 # keeps track of the curve number c2 <- 1 # keeps track of the lty, col, etc xend <- yend <- double(ncurve) for (i in unique(stemp)) { #for each strata who <- which(stemp==i) censor <- if (is.null(x$n.censor)) (x$n.event[who] ==0) else (x$n.censor[who] >0) #places with a censoring xx <- c(firstx, stime[who]) censor <- c(FALSE, censor) #no mark at firstx for (j in 1:ncol(ssurv)) { yy <- c(firsty[i,j], ssurv[who,j]) if (plot.surv) { if (type=='s') lines(dostep(xx, yy), lty=lty[c2], col=col[c2], lwd=lwd[c2]) else lines(xx, yy, type=type, lty=lty[c2], col=col[c2], lwd=lwd[c2]) if (is.numeric(mark.time) || mark.time) drawmark(xx, yy, mark.time, censor, pch=mark[c1], col=mcol[c1], cex=cex) } xend[c1] <- max(xx) yend[c1] <- yy[length(yy)] c1 <- c1 +1 c2 <- c2 +1 if (conf.int) { if (type == 's') { lines(dostep(xx, c(firsty[i,j], slower[who,j])), lty=lty[c2], col=col[c2],lwd=lwd[c2]) c2 <- c2 +1 lines(dostep(xx, c(firsty[i,j], supper[who,j])), lty=lty[c2], col=col[c2], lwd= lwd[c2]) c2 <- c2 + 1 } else { lines(xx, c(firsty[i,j], slower[who,j]), lty=lty[c2], col=col[c2],lwd=lwd[c2], type=type) c2 <- c2 +1 lines(xx, c(firsty[i,j], supper[who,j]), lty=lty[c2], col=col[c2], lwd= lwd[c2], type= type) c2 <- c2 + 1 } } } } invisible(list(x=xend, y=yend)) } lines.survfit <- function(x, type='s', mark=3, col=1, lty=1, lwd=1, cex=1, mark.time=TRUE, xscale=1, firstx=0, firsty=1, xmax, fun, conf.int=FALSE, ...) { xlog <- par("xlog") if (inherits(x, "survfitms")) { x$surv <- 1- x$prev if (is.matrix(x$surv)) dimnames(x$surv) <- list(NULL, x$states) if (!is.null(x$lower)) { x$lower <- 1- x$lower x$upper <- 1- x$upper } if (missing(fun)) fun <- "event" } if (missing(firsty) && !is.null(x$prev0)) firsty <- 1-x$prev0 ssurv <- as.matrix(x$surv) stime <- x$time if( !is.null(x$upper)) { supper <- as.matrix(x$upper) slower <- as.matrix(x$lower) } else { conf.int <- FALSE supper <- NULL #marker for later code } # Two cases where we don't put marks at the censoring times if (inherits(x, 'survexp') || inherits(x, 'survfit.coxph')) { if (missing(mark.time)) mark.time <- FALSE } # set up strata if (is.null(x$strata)) { nstrat <- 1 stemp <- rep(1, length(x$time)) # same length as stime } else { nstrat <- length(x$strata) stemp <- rep(1:nstrat, x$strata) # same length as stime } ncurve <- nstrat * ncol(ssurv) firsty <- matrix(firsty, nrow=nstrat, ncol=ncol(ssurv)) if (!missing(xmax) && any(x$time>xmax)) { # prune back the survival curves # I need to replace x's over the limit with xmax, and y's over the # limit with either the prior y value or firsty keepx <- keepy <- NULL # lines to keep tempn <- table(stemp) offset <- cumsum(c(0, tempn)) for (i in 1:nstrat) { ttime <-stime[stemp==i] if (all(ttime <= xmax)) { keepx <- c(keepx, 1:tempn[i] + offset[i]) keepy <- c(keepy, 1:tempn[i] + offset[i]) } else { bad <- min((1:tempn[i])[ttime>xmax]) if (bad==1) { #lost them all if (!is.na(firstx)) { # and we are plotting lines keepy <- c(keepy, 1+offset[i]) ssurv[1+offset[i],] <- firsty[i,] } } else keepy<- c(keepy, c(1:(bad-1), bad-1) + offset[i]) keepx <- c(keepx, (1:bad)+offset[i]) stime[bad+offset[i]] <- xmax x$n.event[bad+offset[i]] <- 1 #don't plot a tick mark } } # ok, now actually prune it stime <- stime[keepx] stemp <- stemp[keepx] x$n.event <- x$n.event[keepx] if (!is.null(x$n.censor)) x$n.censor <- x$n.censor[keepx] ssurv <- ssurv[keepy,,drop=FALSE] if (!is.null(supper)) { supper <- supper[keepy,,drop=FALSE] slower <- slower[keepy,,drop=FALSE] } } stime <- stime/xscale #scaling is deferred until xmax processing is done if (!missing(fun)) { if (is.character(fun)) { tfun <- switch(fun, 'log' = function(x) x, 'event'=function(x) 1-x, 'cumhaz'=function(x) -log(x), 'cloglog'=function(x) log(-log(x)), 'pct' = function(x) x*100, 'logpct'= function(x) 100*x, #special case further below 'identity'= function(x) x, stop("Unrecognized function argument") ) } else if (is.function(fun)) tfun <- fun else stop("Invalid 'fun' argument") ssurv <- tfun(ssurv ) if (!is.null(supper)) { supper <- tfun(supper) slower <- tfun(slower) } firsty <- tfun(firsty) } if (missing(firstx)) { if (!is.null(x$start.time)) firstx <- x$start.time/xscale else { if (xlog) firstx <- min(stime[stime>0]) else firstx <- min(0, stime) } } # The default for plot and lines is to add confidence limits # if there is only one curve if (missing(conf.int)) conf.int <- (ncurve==1) if (is.logical(conf.int)) plot.surv <- TRUE else { temp <- match.arg(conf.int, c("both", "only", "none")) if (is.na(temp)) stop("invalid value for conf.int") if (temp=="none") conf.int <- FALSE else conf.int <- TRUE if (temp=="only") plot.surv <- FALSE else plot.surv <- TRUE } # Marks are not placed on confidence bands mark <- rep(mark, length.out=ncurve) mcol <- rep(col, length.out=ncurve) if (is.numeric(mark.time)) mark.time <- sort(mark.time) # The actual number of curves is ncurve*3 if there are confidence bands # If the number of line types is 1 and lty is an integer, then use lty # for the curve and lty+1 for the CI # If the length(lty) <= length(ncurve), use the same color for curve and CI # otherwise assume the user knows what they are about and has given a full # vector of line types. # Colors and line widths work like line types, excluding the +1 rule. if (conf.int) { if (length(lty)==1 && is.numeric(lty)) lty <- rep(c(lty, lty+1, lty+1), ncurve) else if (length(lty) <= ncurve) lty <- rep(rep(lty, each=3), length.out=(ncurve*3)) else lty <- rep(lty, length.out= ncurve*3) if (length(col) <= ncurve) col <- rep(rep(col, each=3), length.out=3*ncurve) else col <- rep(col, length.out=3*ncurve) if (length(lwd) <= ncurve) lwd <- rep(rep(lwd, each=3), length.out=3*ncurve) else lwd <- rep(lwd, length.out=3*ncurve) } else { col <- rep(col, length.out=ncurve) lty <- rep(lty, length.out=ncurve) lwd <- rep(lwd, length.out=ncurve) } # Create a step function, removing redundancies that sometimes occur in # curves with lots of censoring. dostep <- function(x,y) { keep <- is.finite(x) & is.finite(y) if (!any(keep)) return() #all points were infinite or NA if (!all(keep)) { # these won't plot anyway, so simplify (CI values are often NA) x <- x[keep] y <- y[keep] } n <- length(x) if (n==1) list(x=x, y=y) else if (n==2) list(x=x[c(1,2,2)], y=y[c(1,1,2)]) else { # replace verbose horizonal sequences like # (1, .2), (1.4, .2), (1.8, .2), (2.3, .2), (2.9, .2), (3, .1) # with (1, .2), (.3, .2),(3, .1). # They are slow, and can smear the looks of the line type. temp <- rle(y)$lengths drops <- 1 + cumsum(temp[-length(temp)]) # points where the curve drops #create a step function if (n %in% drops) { #the last point is a drop xrep <- c(x[1], rep(x[drops], each=2)) yrep <- rep(y[c(1,drops)], c(rep(2, length(drops)), 1)) } else { xrep <- c(x[1], rep(x[drops], each=2), x[n]) yrep <- c(rep(y[c(1,drops)], each=2)) } list(x=xrep, y=yrep) } } drawmark <- function(x, y, mark.time, censor, cex, ...) { if (!is.numeric(mark.time)) { xx <- x[censor] yy <- y[censor] } else { #interpolate xx <- mark.time yy <- approx(x, y, xx, method="constant", f=0)$y } points(xx, yy, cex=cex, ...) } c1 <- 1 # keeps track of the curve number c2 <- 1 # keeps track of the lty, col, etc xend <- yend <- double(ncurve) for (i in unique(stemp)) { #for each strata who <- which(stemp==i) censor <- if (is.null(x$n.censor)) (x$n.event[who] ==0) else (x$n.censor[who] >0) #places with a censoring xx <- c(firstx, stime[who]) censor <- c(FALSE, censor) #no mark at firstx for (j in 1:ncol(ssurv)) { yy <- c(firsty[i,j], ssurv[who,j]) if (plot.surv) { if (type=='s') lines(dostep(xx, yy), lty=lty[c2], col=col[c2], lwd=lwd[c2]) else lines(xx, yy, type=type, lty=lty[c2], col=col[c2], lwd=lwd[c2]) if (is.numeric(mark.time) || mark.time) drawmark(xx, yy, mark.time, censor, pch=mark[c1], col=mcol[c1], cex=cex) } xend[c1] <- max(xx) yend[c1] <- yy[length(yy)] c1 <- c1 +1 c2 <- c2 +1 if (conf.int) { if (type == 's') { lines(dostep(xx, c(firsty[i,j], slower[who,j])), lty=lty[c2], col=col[c2],lwd=lwd[c2]) c2 <- c2 +1 lines(dostep(xx, c(firsty[i,j], supper[who,j])), lty=lty[c2], col=col[c2], lwd= lwd[c2]) c2 <- c2 + 1 } else { lines(xx, c(firsty[i,j], slower[who,j]), lty=lty[c2], col=col[c2],lwd=lwd[c2], type=type) c2 <- c2 +1 lines(xx, c(firsty[i,j], supper[who,j]), lty=lty[c2], col=col[c2], lwd= lwd[c2], type= type) c2 <- c2 + 1 } } } } invisible(list(x=xend, y=yend)) } points.survfit <- function(x, xscale=1, xmax, fun, ...) { if (inherits(x, "survfitms")) { x$surv <- 1- x$prev if (is.matrix(x$surv)) dimnames(x$surv) <- list(NULL, x$states) if (!is.null(x$lower)) { x$lower <- 1- x$lower x$upper <- 1- x$upper } if (missing(fun)) fun <- "event" } firstx <- NA # flag used in the common args conf.int <- FALSE ssurv <- as.matrix(x$surv) stime <- x$time if( !is.null(x$upper)) { supper <- as.matrix(x$upper) slower <- as.matrix(x$lower) } else { conf.int <- FALSE supper <- NULL #marker for later code } # Two cases where we don't put marks at the censoring times if (inherits(x, 'survexp') || inherits(x, 'survfit.coxph')) { if (missing(mark.time)) mark.time <- FALSE } # set up strata if (is.null(x$strata)) { nstrat <- 1 stemp <- rep(1, length(x$time)) # same length as stime } else { nstrat <- length(x$strata) stemp <- rep(1:nstrat, x$strata) # same length as stime } ncurve <- nstrat * ncol(ssurv) firsty <- matrix(firsty, nrow=nstrat, ncol=ncol(ssurv)) if (!missing(xmax) && any(x$time>xmax)) { # prune back the survival curves # I need to replace x's over the limit with xmax, and y's over the # limit with either the prior y value or firsty keepx <- keepy <- NULL # lines to keep tempn <- table(stemp) offset <- cumsum(c(0, tempn)) for (i in 1:nstrat) { ttime <-stime[stemp==i] if (all(ttime <= xmax)) { keepx <- c(keepx, 1:tempn[i] + offset[i]) keepy <- c(keepy, 1:tempn[i] + offset[i]) } else { bad <- min((1:tempn[i])[ttime>xmax]) if (bad==1) { #lost them all if (!is.na(firstx)) { # and we are plotting lines keepy <- c(keepy, 1+offset[i]) ssurv[1+offset[i],] <- firsty[i,] } } else keepy<- c(keepy, c(1:(bad-1), bad-1) + offset[i]) keepx <- c(keepx, (1:bad)+offset[i]) stime[bad+offset[i]] <- xmax x$n.event[bad+offset[i]] <- 1 #don't plot a tick mark } } # ok, now actually prune it stime <- stime[keepx] stemp <- stemp[keepx] x$n.event <- x$n.event[keepx] if (!is.null(x$n.censor)) x$n.censor <- x$n.censor[keepx] ssurv <- ssurv[keepy,,drop=FALSE] if (!is.null(supper)) { supper <- supper[keepy,,drop=FALSE] slower <- slower[keepy,,drop=FALSE] } } stime <- stime/xscale #scaling is deferred until xmax processing is done if (!missing(fun)) { if (is.character(fun)) { tfun <- switch(fun, 'log' = function(x) x, 'event'=function(x) 1-x, 'cumhaz'=function(x) -log(x), 'cloglog'=function(x) log(-log(x)), 'pct' = function(x) x*100, 'logpct'= function(x) 100*x, #special case further below 'identity'= function(x) x, stop("Unrecognized function argument") ) } else if (is.function(fun)) tfun <- fun else stop("Invalid 'fun' argument") ssurv <- tfun(ssurv ) if (!is.null(supper)) { supper <- tfun(supper) slower <- tfun(slower) } firsty <- tfun(firsty) } if (ncol(ssurv)==1) points(stime, ssurv, ...) else matpoints(stime, ssurv, ...) } survival/R/survfitKM.S0000644000175100001440000002066112122050337014422 0ustar hornikusers# A version that does more work in S, less in C survfitKM <- function(x, y, casewt=rep(1,length(x)), type=c('kaplan-meier', 'fleming-harrington', 'fh2'), error=c('greenwood', "tsiatis"), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified'), start.time, new.time) { type <- match.arg(type) method <- match(type, c("kaplan-meier", "fleming-harrington", "fh2")) error <- match.arg(error) error.int <- match(error, c("greenwood", "tsiatis")) conf.type <- match.arg(conf.type) conf.lower<- match.arg(conf.lower) if (is.logical(conf.int)) { # A common error is for users to use "conf.int = FALSE" # it's illegal, but allow it if (!conf.int) conf.type <- "none" conf.int <- .95 } if (!is.Surv(y)) stop("y must be a Surv object") if (!is.factor(x)) stop("x must be a factor") if (attr(y, 'type') != 'right' && attr(y, 'type') != 'counting') stop("Can only handle right censored or counting data") ny <- ncol(y) # Will be 2 for right censored, 3 for counting xlev <- levels(x) # Will supply names for the curves x <- as.numeric(x) # keep only the levels # Allow "new.time" as a synonym for start.time if (missing(start.time) && !missing(new.time)) start.time <- new.time if (!missing(start.time)) { n.all <- c(table(x)) # remember the original data size # remove any obs whose end time is <= start.time keep <- (y[,ny-1] >= start.time) if (all(keep==FALSE)) stop(paste("start.time =", start.time, "is greater than all time points.")) x <- x[keep] y <- y[keep,,drop=FALSE] #make sure y remains a matrix casewt <- casewt[keep] } n.used <- as.vector(table(x)) # This is for the printout nstrat <- length(n.used) # # Each of the necessary output objects is originally a list with one # element per strata. This doesn't use up extra S memory, the number # of curves is usually small enough that the "for" loop is no great # cost, and it's easier to see what's going on than C code. # If tapply gets fast again, this code will also be fast. # The remaining C-code is simple, but hard to do cleanly in S # Let nrisk=A and nevent=B. The terms in returned sum1 and sum2 are # If ndead= 0, sum1=sum2=1 (avoids a 0/0 in the S code) # If ndead= 1, sum1=1/A and sum2= 1/A*A # If ndead= 2, sum1= (1/2)[ 1/A + 1/(A- B/2)] # sum2= (1/2)[ 1/A^2 + (1/(A-B/2))^2] # If ndead =3, sum1 = (1/3)[1/A + 1/(A-B/3) + 1/(A -2B/3)] # If ndead =4, sum1 = (1/4)[1/A + 1/(A-B/4) + 1/(A -2B/4) + 1/(A-3B/4)] # and etc. time <- vector('list', nstrat) n.risk <- vector('list', nstrat) surv <- vector('list', nstrat) n.cens <- vector('list', nstrat) n.event<- vector('list', nstrat) strata <- integer(nstrat) if (se.fit) varhaz <- vector('list', nstrat) if (ny==3) n.enter <- vector('list', nstrat) uniquex <- sort(unique(x)) for (i in 1:nstrat) { who <- (x== uniquex[i]) if (ny==2) { # the "factor" + levels ensures 2 columns in temp even if all # are dead or all are alive temp <- tapply(casewt[who], list(factor(y[who,1]), factor(y[who,2], levels=0:1)), sum) temp <- ifelse(is.na(temp), 0, temp) # The two lines below do not always give the same answer # When two times differ by only the machine precision, unique # will give more values, and then "time" will be a different # length than the other components time[[i]] <- type.convert(dimnames(temp)[[1]], as.is=TRUE, dec=getOption("OutDec")) # time[[i]] <- sort(unique(y[who,1])) # old version of line above ntemp <- (dim(temp))[1] nevent <- as.vector(temp[,2]) ncens <- as.vector(temp[,1]) nrisk <- rev(cumsum(rev(temp %*% c(1,1)))) ndead <- as.vector(table(y[who,1], factor(y[who,2], levels=0:1)) [,2]) } else { # The counting process case # We have to be a bit more clever here -- if I did a table of the # start times and a separate one of the stop times, they wouldn't # necessarily match. So do it all at once with a fake 'status' # variable which is ==2 for start times and = status for stop # Also, the number of rows in output (ntemp) may be bigger than # the number of rows of input (n). n <- sum(who) temp <- factor(c(rep(2,n),y[who,3]), levels=0:3) temp <- tapply(rep(casewt[who],2), list(factor(y[who,1:2]), temp), sum) temp <- ifelse(is.na(temp), 0, temp) time[[i]] <- as.numeric(dimnames(temp)[[1]]) ntemp <- (dim(temp))[1] nevent <- as.vector(temp[,2]) ncens <- as.vector(temp[,1]) nenter <- as.vector(temp[,3]) nrisk <- cumsum(nenter - (nevent + ncens)) nrisk <- c(0, nrisk[-ntemp]) #risk counts change at time t+0 n.enter[[i]] <- nenter # again, a fake status to make sure that all the times appear ndead <- as.vector(table(y[who,1:2], factor(c(rep(0,n),y[who,3]), levels=0:2))[,2]) } strata[i] <- ntemp trisk <- ifelse(nrisk==0, 1, nrisk) #avoid 0/0 cases if (method==1) surv[[i]] <- cumprod((trisk-nevent)/trisk) if (method==2) { hazard <- nevent/trisk #Nelson's hazard estimate surv[[i]] <- exp(- cumsum(hazard)) } if (method==3) { tsum <- .C(Csurvfit4, as.integer(length(ncens)), as.integer(ndead), sum1 = as.double(nrisk), sum2 = as.double(nevent)) hazard <- nevent *tsum$sum1 surv[[i]] <- exp(-cumsum(hazard)) } if (se.fit) { if (error.int==1) # Greenwood varhaz[[i]] <- cumsum(nevent/(trisk*(trisk-nevent))) else { if (method <3) varhaz[[i]] <- cumsum(nevent/(trisk^2)) else varhaz[[i]] <- cumsum(nevent* tsum$sum2) } } n.event[[i]] <- nevent n.cens[[i]] <- ncens n.risk[[i]] <- nrisk } if (ny==2) { temp <- list(n=n.used, time = unlist(time), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor = unlist(n.cens), surv = unlist(surv), type='right') } else { temp <- list(n=n.used, time = unlist(time), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor = unlist(n.cens), n.enter = unlist(n.enter), surv = unlist(surv), type='counting') } if (nstrat >1) { names(strata) <- xlev[sort(unique(x))] temp$strata <- strata } if (!missing(start.time)) { temp$start.time <- start.time # user defined time to start temp$n.all <- n.all } if (se.fit) { std.err <- sqrt(unlist(varhaz)) temp$std.err <- std.err # # n.lag = the # at risk the last time there was an event (or # the first time of a strata) # events <- temp$n.event >0 if (nstrat==1) events[1] <- TRUE else events[1 + cumsum(c(0, strata[-nstrat]))] <- TRUE zz <- 1:length(events) n.lag <- rep(temp$n.risk[events], diff(c(zz[events], 1+max(zz)))) std.low <- switch(conf.lower, 'usual' = std.err, 'peto' = sqrt((1-temp$surv)/ temp$n.risk), 'modified' = std.err * sqrt(n.lag/temp$n.risk)) zval <- qnorm(1- (1-conf.int)/2, 0,1) if (conf.type=='plain') { temp1 <- temp$surv + zval* std.err * temp$surv temp2 <- temp$surv - zval* std.low * temp$surv temp <- c(temp, list(upper=pmin(temp1,1), lower=pmax(temp2,0), conf.type='plain', conf.int=conf.int)) } if (conf.type=='log') { #avoid some "log(0)" messages xx <- ifelse(temp$surv==0,1,temp$surv) temp1 <- ifelse(temp$surv==0, NA, exp(log(xx) + zval* std.err)) temp2 <- ifelse(temp$surv==0, NA, exp(log(xx) - zval* std.low)) temp <- c(temp, list(upper=pmin(temp1,1), lower=temp2, conf.type='log', conf.int=conf.int)) } if (conf.type=='log-log') { who <- (temp$surv==0 | temp$surv==1) #special cases temp3 <- ifelse(temp$surv==0, NA, 1) xx <- ifelse(who, .1,temp$surv) #avoid some "log(0)" messages temp1 <- exp(-exp(log(-log(xx)) + zval*std.err/log(xx))) temp1 <- ifelse(who, temp3, temp1) temp2 <- exp(-exp(log(-log(xx)) - zval*std.low/log(xx))) temp2 <- ifelse(who, temp3, temp2) temp <- c(temp, list(upper=temp1, lower=temp2, conf.type='log-log', conf.int=conf.int)) } } temp } survival/R/survConcordance.R0000644000175100001440000000440512533653605015640 0ustar hornikusers# Automatically generated from all.nw using noweb survConcordance <- function(formula, data, weights, subset, na.action) { Call <- match.call() # save a copy of of the call, as documentation m <- match.call(expand.dots=FALSE) m[[1]] <- as.name("model.frame") m$formula <- if(missing(data)) terms(formula, "strata") else terms(formula, "strata", data=data) m <- eval(m, sys.parent()) Terms <- attr(m, 'terms') Y <- model.extract(m, "response") if (!inherits(Y, "Surv")) { if (is.numeric(Y) && is.vector(Y)) Y <- Surv(Y) else stop("left hand side of the formula must be a numeric vector or a surival") } n <- nrow(Y) wt <- model.extract(m, 'weights') offset<- attr(Terms, "offset") if (length(offset)>0) stop("Offset terms not allowed") stemp <- untangle.specials(Terms, 'strata') if (length(stemp$vars)) { if (length(stemp$vars)==1) strat <- m[[stemp$vars]] else strat <- strata(m[,stemp$vars], shortlabel=TRUE) Terms <- Terms[-stemp$terms] } else strat <- NULL x <- model.matrix(Terms, m)[,-1, drop=FALSE] #remove the intercept if (ncol(x) > 1) stop("Only one predictor variable allowed") count <- survConcordance.fit(Y, x, strat, wt) if (is.null(strat)) { concordance <- (count[1] + count[3]/2)/sum(count[1:3]) std.err <- count[5]/(2* sum(count[1:3])) } else { temp <- colSums(count) concordance <- (temp[1] + temp[3]/2)/ sum(temp[1:3]) std.err <- temp[5]/(2*sum(temp[1:3])) } fit <- list(concordance= concordance, stats=count, n=n, std.err=std.err, call=Call) na.action <- attr(m, "na.action") if (length(na.action)) fit$na.action <- na.action oldClass(fit) <- 'survConcordance' fit } print.survConcordance <- function(x, ...) { if(!is.null(cl <- x$call)) { cat("Call:\n") dput(cl) cat("\n") } omit <- x$na.action if(length(omit)) cat(" n=", x$n, " (", naprint(omit), ")\n", sep = "") else cat(" n=", x$n, "\n") cat("Concordance= ", format(x$concordance), " se= ", format(x$std.err), '\n', sep='') print(x$stats) invisible(x) } survival/R/frailty.controlaic.S0000644000175100001440000000354212470201064016276 0ustar hornikusers# $Id: frailty.controlaic.S 11166 2008-11-24 22:10:34Z therneau $ # Control function to minimize the AIC # the optional paramater "caic" chooses corrected aic (default=FALSE) # n is the "effective" sample size # frailty.controlaic <- function(parms, iter, old, n, df, loglik) { if (iter==0) { # initial call if (is.null(parms$init)) theta <-0.005 else theta <- parms$init[1] return(list(theta=theta, done=FALSE)) } # by default, do the corrected AIC if (length(parms$caic)) correct <- parms$caic else correct <- FALSE if (n < df+2) dfc <- (df -n) + (df+1)*df/2 -1 #avoid pathology else dfc <- -1 + (df+1)/(1- ((df+2)/n)) if (iter==1) { # Second guess in series history <- c(theta=old$theta, loglik=loglik, df=df, aic=loglik-df, aicc=loglik - dfc) if (length(parms$init) <2) theta <-1 else theta <- parms$init[2] temp <- list(theta=theta, done=FALSE, history=history) return(temp) } history <- rbind(old$history,c(old$theta, loglik, df, loglik-df, loglik -dfc)) if (iter==2) { #Third guess theta <- mean(history[,1]) return(list(theta=theta, done=FALSE, history=history)) } # # Ok, now we're ready to actually use prior data # Now, history has iter rows, each row contains the # value of theta, the Cox PL, the df, aic, and corrected aic if (correct) aic <- history[,5] #use corrected aic for convergence else aic <- history[,4] done <- (abs(1- aic[iter]/aic[iter-1]) < parms$eps) x <- history[,1] if (x[iter]== max(aic) && x[iter]==max(x)) newtheta <- 2* max(x) else newtheta <- frailty.brent(x, aic, lower=parms$lower, upper=parms$upper) if (length(parms$trace) && parms$trace) { print(history) cat(" new theta=", format(newtheta), "\n\n") } list(theta=newtheta, done=done, history=history) } survival/R/strata.S0000644000175100001440000000450012465203762013774 0ustar hornikusers# Create a strata variable, possibly from many objects # strata <- function(..., na.group=FALSE, shortlabel, sep=', ') { # First, grab a copy of the call, which will be used to manufacture # labels for unlabeled arguments # Then get the arguments as a list words <- as.character((match.call())[-1]) allf <- list(...) # If there is only one argument, and it itself is a list, use # it instead if(length(allf) == 1 && is.list(ttt <- unclass(allf[[1]]))) allf <- ttt nterms <- length(allf) # Keep the names of named args as their label, what was typed otherwise if (is.null(names(allf))) { argname <- words[1:nterms] if (missing(shortlabel)) shortlabel <- all(sapply(allf, function(x) is.character(x) | inherits(x, 'factor'))) } else { argname <- ifelse(names(allf) == '', words[1:nterms], names(allf)) if (missing(shortlabel)) shortlabel <- FALSE } # If the arguments are not all the same length, stop now # Mostly this is to stop calls with an improper object arglength <- sapply(allf, length) if (any(arglength != arglength[1])) stop("all arguments must be the same length") if (!all(sapply(allf, is.atomic))) stop("all arguments must be vectors") # Process the first argument what <- allf[[1]] if(is.null(levels(what))) what <- factor(what) levs <- unclass(what) - 1 wlab <- levels(what) if (na.group && any(is.na(what))){ # add "NA" as a level levs[is.na(levs)] <- length(wlab) wlab <- c(wlab, "NA") } if (shortlabel) labs <- wlab else labs <- paste(argname[1], wlab, sep='=') # Now march through the other variables, if any for (i in (1:nterms)[-1]) { what <- allf[[i]] if(is.null(levels(what))) what <- factor(what) wlab <- levels(what) wlev <- unclass(what) - 1 if (na.group && any(is.na(wlev))){ wlev[is.na(wlev)] <- length(wlab) wlab <- c(wlab, "NA") } if (!shortlabel) wlab <- format(paste(argname[i], wlab, sep='=')) levs <- wlev + levs*(length(wlab)) labs <- paste(rep(labs, rep(length(wlab), length(labs))), rep(wlab, length(labs)), sep=sep) } levs <- levs + 1 ulevs <- sort(unique(levs[!is.na(levs)])) levs <- match(levs, ulevs) labs <- labs[ulevs] factor(levs, labels=labs) } survival/R/survfit.R0000644000175100001440000001742012533653607014207 0ustar hornikusers# Automatically generated from all.nw using noweb survfit <- function(formula, ...) { UseMethod("survfit", formula) } dim.survfit <- function(x) { if (is.null(x$strata)) { if (is.matrix(x$surv)) ncol(x$surv) else 1 } else { nr <- length(x$strata) if (is.matrix(x$xurv)) c(nr, ncol(x$surv)) else nr } } "[.survfit" <- function(x, ..., drop=TRUE) { nmatch <- function(indx, target) { # This function lets R worry about character, negative, or logical subscripts # It always returns a set of positive integer indices temp <- 1:length(target) names(temp) <- target temp[indx] } if (missing(..1)) i<- NULL else i <- ..1 if (missing(..2)) j<- NULL else j <- ..2 if (is.null(i) && is.null(j)) return (x) #no subscripts present! if (!is.matrix(x$surv) && !is.null(j)) stop("survfit object does not have 2 dimensions") if (is.null(x$strata)) { if (is.matrix(x$surv)) { if (is.null(j) && !is.null(i)) j <- i #special case noted above x$surv <- x$surv[,j,drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[,j,drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[,j,drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[,j,drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[,j,drop=drop] } else warning("survfit object has only a single survival curve") } else { if (is.null(i)) keep <- seq(along.with=x$time) else { indx <- nmatch(i, names(x$strata)) #strata to keep if (any(is.na(indx))) stop(paste("strata", paste(i[is.na(indx)], collapse=' '), 'not matched')) # Now, indx may not be in order: some can use curve[3:2] to reorder # The list/unlist construct will reorder the data temp <- rep(1:length(x$strata), x$strata) keep <- unlist(lapply(indx, function(x) which(temp==x))) if (length(indx) <=1 && drop) x$strata <- NULL else x$strata <- x$strata[i] x$n <- x$n[indx] x$time <- x$time[keep] x$n.risk <- x$n.risk[keep] x$n.event <- x$n.event[keep] x$n.censor<- x$n.censor[keep] if (!is.null(x$enter)) x$enter <- x$enter[keep] } if (is.matrix(x$surv)) { # If the curve has been selected by strata and keep has only # one row, we don't want to lose the second subscript too if (!is.null(i) && (is.null(j) ||length(j) >1)) drop <- FALSE if (is.null(j)) { x$surv <- x$surv[keep,,drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,,drop=drop] if (!is.null(x$upper)) x$upper <-x$upper[keep,,drop=drop] if (!is.null(x$lower)) x$lower <-x$lower[keep,,drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <-x$cumhaz[keep,,drop=drop] } else { x$surv <- x$surv[keep,j, drop=drop] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep,j, drop=drop] if (!is.null(x$upper)) x$upper <- x$upper[keep,j, drop=drop] if (!is.null(x$lower)) x$lower <- x$lower[keep,j, drop=drop] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[keep,j, drop=drop] } } else { x$surv <- x$surv[keep] if (!is.null(x$std.err)) x$std.err <- x$std.err[keep] if (!is.null(x$upper)) x$upper <- x$upper[keep] if (!is.null(x$lower)) x$lower <- x$lower[keep] if (!is.null(x$cumhaz)) x$cumhaz <- x$cumhaz[keep] } } x } survfit.formula <- function(formula, data, weights, subset, na.action, etype, id, istate, ...) { Call <- match.call() Call[[1]] <- as.name('survfit') #make nicer printout for the user # create a copy of the call that has only the arguments we want, # and use it to call model.frame() indx <- match(c('formula', 'data', 'weights', 'subset','na.action', 'istate', 'id', "etype"), names(Call), nomatch=0) #It's very hard to get the next error message other than malice # eg survfit(wt=Surv(time, status) ~1) if (indx[1]==0) stop("a formula argument is required") temp <- Call[c(1, indx)] temp[[1]] <- as.name("model.frame") m <- eval.parent(temp) Terms <- terms(formula, c("strata", "cluster")) ord <- attr(Terms, 'order') if (length(ord) & any(ord !=1)) stop("Interaction terms are not valid for this function") n <- nrow(m) Y <- model.extract(m, 'response') if (!is.Surv(Y)) stop("Response must be a survival object") casewt <- model.extract(m, "weights") if (is.null(casewt)) casewt <- rep(1,n) if (!is.null(attr(Terms, 'offset'))) warning("Offset term ignored") id <- model.extract(m, 'id') istate <- model.extract(m,"istate") temp <- untangle.specials(Terms, "cluster") if (length(temp$vars)>0) { if (length(temp$vars) > 1) stop("can not have two cluster terms") if (!is.null(id)) stop("can not have both a cluster term and an id variable") id <- m[[temp$vars]] Terms <- Terms[-temp$terms] } ll <- attr(Terms, 'term.labels') if (length(ll) == 0) X <- factor(rep(1,n)) # ~1 on the right else X <- strata(m[ll]) if (!is.Surv(Y)) stop("y must be a Surv object") # Backwards support for the now-depreciated etype argument etype <- model.extract(m, "etype") if (!is.null(etype)) { if (attr(Y, "type") == "mcounting" || attr(Y, "type") == "mright") stop("cannot use both the etype argument and mstate survival type") if (length(istate)) stop("cannot use both the etype and istate arguments") status <- Y[,ncol(Y)] etype <- as.factor(etype) temp <- table(etype, status==0) if (all(rowSums(temp==0) ==1)) { # The user had a unique level of etype for the censors newlev <- levels(etype)[order(-temp[,2])] #censors first } else newlev <- c(" ", levels(etype)[temp[,1] >0]) status <- factor(ifelse(status==0,0, as.numeric(etype)), labels=newlev) if (attr(Y, 'type') == "right") Y <- Surv(Y[,1], status, type="mstate") else if (attr(Y, "type") == "counting") Y <- Surv(Y[,1], Y[,2], status, type="mstate") else stop("etype argument incompatable with survival type") } # At one point there were lines here to round the survival # times to a certain number of digits. This approach worked # almost all the time, but only almost. The better logic is # now in the individual compuation routines if (attr(Y, 'type') == 'left' || attr(Y, 'type') == 'interval') temp <- survfitTurnbull(X, Y, casewt, ...) else if (attr(Y, 'type') == "right" || attr(Y, 'type')== "counting") temp <- survfitKM(X, Y, casewt, ...) else if (attr(Y, 'type') == "mright" || attr(Y, "type")== "mcounting") temp <- survfitCI(X, Y, weights=casewt, id=id, istate=istate, ...) else { # This should never happen stop("unrecognized survival type") } if (is.null(temp$states)) class(temp) <- 'survfit' else class(temp) <- c("survfitms", "survfit") if (!is.null(attr(m, 'na.action'))) temp$na.action <- attr(m, 'na.action') temp$call <- Call temp } survfit.Surv <- function(formula, ...) stop("the survfit function requires a formula as its first argument") survival/R/predict.survreg.penal.S0000644000175100001440000000063211732700061016712 0ustar hornikusers# $Id: predict.survreg.penal.S 11166 2008-11-24 22:10:34Z therneau $ # # This routine just stops disastrous arithmetic for models with sparse # terms. A placeholder until the proper sparse terms actions are inserted. # predict.survreg.penal <- function(object, ...) { pterms <- object$pterms if (any(pterms==2)) stop("Predictions not available for sparse models") NextMethod('predict') } survival/R/coxph.getdata.S0000644000175100001440000000320111732700061015212 0ustar hornikusers# # Reconstruct the Cox model data. This is done in so many routines # that I extracted it out. # Newer routines use model.matrix.coxph and model.frame.coxph methods. # # The "stratax" name is to avoid conflicts with the strata() function, but # still allow users to type "strata" as an arg. # coxph.getdata <- function(fit, y=TRUE, x=TRUE, stratax=TRUE, offset=FALSE) { ty <- fit[['y']] #avoid grabbing this by accident due to partial matching tx <- fit[['x']] # for x, fit$x will get fit$xlevels --> not good strat <- fit$strata Terms <- fit$terms if (is.null(attr(Terms, 'offset'))) offset <- FALSE if (offset) x<- TRUE if (!inherits(Terms, 'terms')) stop("invalid terms component of fit") strats <- attr(Terms, "specials")$strata if (length(strats)==0) stratax <- FALSE if ( (y && is.null(ty)) || (x && is.null(tx)) || (stratax && is.null(strat)) || offset) { # get the model frame m <- model.frame(fit) # Pull things out if (y && is.null(ty)) ty <- model.extract(m, 'response') if (offset) toff <- model.extract(m, 'offset') # strata was saved in the fit if and only if x was if ((x || stratax) && is.null(tx)) { if (stratax) { temp <- untangle.specials(Terms, 'strata', 1) strat <- strata(m[temp$vars], shortlabel=T) } if (x) tx <- model.matrix(fit, data=m) } } else if (offset) toff <- fit$linear.predictors -(c(tx %*% fit$coef) - sum(fit$means*fit$coef)) temp <- list() if (y) temp$y <- ty if (x) temp$x <- tx if (stratax) temp$strata <- strat if (offset) temp$offset <- toff temp } survival/R/basehaz.R0000644000175100001440000000167112466715763014132 0ustar hornikusers# # This function is simply an alias for "survfit". In the Cox model # case users often look for the words "baseline hazard" # basehaz <- function (fit, centered = TRUE) { if (!inherits(fit, "coxph")) stop("must be a coxph object") sfit <- survfit(fit, se.fit=FALSE) if (!centered) { # The right thing to do here is to call survfit with a vector of # all zeros for the "subject to predict". But if there is a factor # in the model, there may be no subject at all who will give all # zeros, so we post process instead zcoef <- ifelse(is.na(coef(fit)), 0, coef(fit)) offset <- sum(fit$means * zcoef) chaz <- sfit$cumhaz * exp(-offset) } else chaz <- sfit$cumhaz new <- data.frame(hazard=chaz, time=sfit$time) strata <- sfit$strata if (!is.null(strata)) new$strata <- factor(rep(names(strata), strata), levels = names(strata)) new } survival/R/frailty.controlgauss.S0000644000175100001440000000556312470201064016671 0ustar hornikusers# $Id: frailty.controlgauss.S 11166 2008-11-24 22:10:34Z therneau $ # # The control function for REML on a gaussian # frailty.controlgauss <- function(opt, iter, old, fcoef, trH, loglik){ if (iter==0) { # initial call # Because of how the iteration works, 0 is not a useful trial value if (!is.null(opt$theta)) theta <- opt$theta #fixed theta case else { if (is.null(opt$init)) theta <- 1 else theta <- opt$init[1] } list(theta=theta) } else { if (is.null(opt$trace)) trace <-FALSE else trace <- opt$trace nfrail <- length(fcoef) fsum <- sum(fcoef^2) theta <- old$theta resid <- fsum/(nfrail - trH/theta) - theta # save history of the iteration, and get the next theta if (iter==1) { history <- c(theta=theta, resid=resid, fsum=fsum, trace=trH) if (is.null(opt$init )) { if (resid>0) theta <- theta*3 else theta <- theta/3 } else theta <- opt$init[2] list(theta=theta, done=FALSE, history=history) } else { history <- rbind(old$history, as.vector(c(theta, resid, fsum, trH))) if (iter ==2) { if (all(history[,2] > 0)) theta <- history[2,1]*2 else if (all(history[,2] <0)) theta <- history[2,1]/2 else theta <- mean(history[1:2,1]) if (trace) { print(history) cat(" new theta=", theta, "\n\n") } list(theta=theta, done=FALSE, history=history) } else { done <- (abs(history[iter,2]) < opt$eps) ord <- order(history[,1]) tempy <- history[ord,2] #x & y from left to right tempx <- history[ord,1] # make sure we have one positve and one negative y value # y must be positive near 0, and negative for large x if (all(tempy>0)) newtheta <- 2*max(tempx) else if (all(tempy<0)) newtheta <- .5 * min(tempx) else{ #find the latest point, and one on each side of 0 b1 <- (1:iter)[ord==iter] if (b1==1) b1 <-2 else if (b1==iter) b1 <- iter-1 # Brent's formula, straight from Numerical Recipies # Why, you may ask, don't we use the uniroot() function # which is built into S, and implements Brent's method? # Because all we want is the next guess for x. The interal # loop of coxph is calling us, not the other way around. R <- tempy[b1]/ tempy[b1+1] S <- tempy[b1]/ tempy[b1-1] U <- R/S P <- S* (U*(R-U)*(tempx[b1+1]-tempx[b1]) - (1-R)*(tempx[b1]-tempx[b1-1])) Q <- (U-1)*(R-1)*(S-1) newtheta <- tempx[b1] + P/Q # if the new guess is outside the brackets, do a binomial # search step if (newtheta > tempx[b1+1]) newtheta <- mean(tempx[b1+0:1]) if( newtheta < tempx[b1-1]) newtheta <- mean(tempx[b1-0:1]) } if (trace) { print(history) cat(" new theta=", format(newtheta), "\n\n") } list(theta=newtheta, done=done, history=history) } } } } survival/R/print.survreg.penal.S0000644000175100001440000000715412423461670016432 0ustar hornikusersprint.survreg.penal <- function(x, terms=FALSE, maxlabel=25, digits=max(options()$digits - 4, 3), ...) { if (!inherits(x, 'survreg.penal')) stop("Invalid object") if (!is.null(x$call)) { cat("Call:\n") dput(x$call) cat("\n") } if (!is.null(x$fail)) { cat(" Survreg failed.", x$fail, "\n") return() } savedig <- options(digits = digits) on.exit(options(savedig)) coef <- x$coefficients if (length(coef)==0) stop("Penalized fits must have an intercept!") # # Map terms to special print functions, and the list of iteration histories # pterms <- x$pterms nterms <- length(pterms) npenal <- sum(pterms>0) print.map <- rep(0,nterms) if (!is.null(x$printfun)) { temp <- unlist(lapply(x$printfun, is.null)) #which ones are missing print.map[pterms>0] <- (1:npenal) * (!temp) } # Tedious, but build up the coef matrix a term at a time print1 <- NULL pname1 <- NULL if (is.null(x$assign2)) alist <- x$assign else alist <- x$assign2 print2 <- NULL for (i in 1:nterms) { kk <- alist[[i]] if (print.map[i] >0) { j <- print.map[i] if (pterms[i]==2) temp <- (x$printfun[[j]])(x$frail, x$fvar, ,x$df[i], x$history[[j]]) else temp <- (x$printfun[[j]])(coef[kk], x$var[kk,kk], x$var2[kk,kk], x$df[i], x$history[[j]]) print1 <- rbind(print1, temp$coef) if (is.matrix(temp$coef)) { xx <- dimnames(temp$coef)[[1]] if (is.null(xx)) xx <- rep(names(pterms)[i], nrow(temp$coef)) else xx <- paste(names(pterms)[i], xx, sep=', ') pname1 <- c(pname1, xx) } else pname1 <- c(pname1, names(pterms)[i]) print2 <- c(print2, temp$history) } else if (terms && length(kk)>1) { pname1 <- c(pname1, names(pterms)[i]) temp <- coxph.wtest(x$var[kk,kk], coef[kk])$test print1 <- rbind(print1, c(NA, NA, NA, temp, x$df[i], 1-pchisq(temp, 1))) } else { pname1 <- c(pname1, names(coef)[kk]) tempe<- (diag(x$var))[kk] temp <- coef[kk]^2/ tempe print1 <- rbind(print1, cbind(coef[kk], sqrt(tempe), sqrt((diag(x$var2))[kk]), temp, 1, 1-pchisq(temp, 1))) } } # Format out the NA's temp <- cbind(format(print1[,1]), format(print1[,2]), format(print1[,3]), format(round(print1[,4], 2)), format(round(print1[,5], 2)), format(signif(print1[,6], 2))) temp <- ifelse(is.na(print1), "", temp) dimnames(temp) <- list(substring(pname1,1, maxlabel), c("coef","se(coef)", "se2", "Chisq","DF","p")) print(temp, quote=FALSE) # # Write out the remaider of the info # if (nrow(x$var)==length(coef)) cat("\nScale fixed at",format(x$scale),"\n") else if (length(x$scale)==1) cat ("\nScale=", format(x$scale), "\n") else { cat("\nScale:\n") print(x$scale, ...) } cat("\nIterations:", x$iter[1], "outer,", x$iter[2], "Newton-Raphson\n") if (length(print2)) { # cat("Penalized terms:\n") for (i in 1:length(print2)) cat(" ", print2[i], "\n") } logtest <- -2 * (x$loglik[1] - x$loglik[2]) df <- sum(x$df) - x$idf # cat("\n") cat("Degrees of freedom for terms=", format(round(x$df,1)), "\n") # cat("Loglik (initial,final) = ", format(round(x$loglik,2)), # " Penalty = ", format(x$penalty), "\n") cat("Likelihood ratio test=", format(round(logtest, 2)), " on ", round(df,1), " df,", " p=", format(1 - pchisq(logtest, df)), sep="") n <- length(x$linear.predictors) omit <- x$na.action if (length(omit)) cat("\n n=", n, " (", naprint(omit), ")\n", sep="") else cat(" n=", n, "\n") invisible() } survival/R/survreg.fit.S0000644000175100001440000002000612470201064014740 0ustar hornikusers# # Do the actual fit of a survreg model. This routine is for the case # of no penalized terms (splines, etc). # survreg.fit<- function(x, y, weights, offset, init, controlvals, dist, scale=0, nstrat=1, strata, parms=NULL, assign) { iter.max <- controlvals$iter.max eps <- controlvals$rel.tolerance toler.chol <- controlvals$toler.chol if (!is.matrix(x)) stop("Invalid X matrix ") n <- nrow(x) nvar <- ncol(x) ny <- ncol(y) if (is.null(offset)) offset <- rep(0,n) if (missing(weights)|| is.null(weights)) weights<- rep(1.0,n) else if (any(weights<=0)) stop("Invalid weights, must be >0") if (scale <0) stop("Invalid scale") if (scale >0 && nstrat >1) stop("Cannot have both a fixed scale and strata") if (nstrat>1 && (missing(strata) || length(strata)!= n)) stop("Invalid strata variable") if (nstrat==1) strata <- rep(1,n) if (scale >0) nstrat2 <- 0 # number of variances to estimate else nstrat2 <- nstrat if (is.character(dist)) { sd <- survreg.distributions[[dist]] if (is.null(sd)) stop ("Unrecognized distribution") } else sd <- dist if (!is.function(sd$density)) stop("Missing density function in the definition of the distribution") dnum <- match(sd$name, c("Extreme value", "Logistic", "Gaussian")) if (is.na(dnum)) { # We need to set up a callback routine # This returns the 5 number distribution summary (see the density # functions in survreg.distributions). Interval censored obs require # 2 evals and all others 1, so the call to the routine will have n2 # values. dnum <- 4 # flag for the C routine n2 <- n + sum(y[,ny]==3) # not needed, keep for documentation # # Create an expression that will be evaluated by the C-code, # but with knowledge of some current variables # In the R doc, this would be "body(function(z) {" # in Splus (Chambers book): "functionBody(function(z)" # same action, different name. Luckily 'quote' exists in both. # We make very sure the result is the right type and length here # rather than in the C code, for simplicity. f.expr <- quote({ if (length(parms)) temp <- sd$density(z, parms) else temp <- sd$density(z) if (!is.matrix(temp) || any(dim(temp) != c(n2,5)) || !is.numeric(temp)) stop("Density function returned an invalid matrix") as.vector(as.double(temp)) }) # create an isolated sandbox (frame or environment) in which # we can do the evaluation without endangering local objects # but still with knowlege of sd, parms, and n2 if (is.R()) rho <- new.env() #inherits necessary objects # SPlus else rho <- new.frame(list(sd=sd, parms=parms, n2=n2)) } else { f.expr <- 1 #dummy values for the .Call rho <- 1 } # This is a subset of residuals.survreg: define the first and second # derivatives at z=0 for the 4 censoring types # Used below for starting estimates derfun <- function(y, eta, sigma, density, parms) { ny <- ncol(y) status <- y[,ny] z <- (y[,1] - eta)/sigma dmat <- density(z,parms) dtemp<- dmat[,3] * dmat[,4] #f' if (any(status==3)) { z2 <- (y[,2] - eta)/sigma dmat2 <- density(z2, parms) } else { dmat2 <- matrix(0,1,5) #dummy values z2 <- 0 } tdenom <- ((status==0) * dmat[,2]) + ((status==1) * 1 ) + ((status==2) * dmat[,1]) + ((status==3) * ifelse(z>0, dmat[,2]-dmat2[,2], dmat2[,1] - dmat[,1])) tdenom <- 1/(tdenom* sigma) dg <- -tdenom *(((status==0) * (0-dmat[,3])) + ((status==1) * dmat[,4]) + ((status==2) * dmat[,3]) + ((status==3) * (dmat2[,3]- dmat[,3]))) ddg <- (tdenom/sigma)*(((status==0) * (0- dtemp)) + ((status==1) * dmat[,5]) + ((status==2) * dtemp) + ((status==3) * (dmat2[,3]*dmat2[,4] - dtemp))) list(dg = dg, ddg = ddg - dg^2) } # # A good initial value of the scale turns out to be critical for successful # iteration, in a surprisingly large number of data sets. # The best way we've found to get one is to fit a model with only the # mean and the scale. We don't need to do this in 3 situations: # 1. The only covariate is a mean (this step is then just a duplicate # of the main fit). # 2. There are no scale parameters to estimate # 3. The user gave initial estimates for the scale # However, for 2 and 3 we still want the loglik for a mean only model # as a part of the returned object. # nvar2 <- nvar + nstrat2 meanonly <- (nvar==1 && all(x==1)) if (!meanonly) { yy <- ifelse(y[,ny]!=3, y[,1], (y[,1]+y[,2])/2 ) coef <- sd$init(yy, weights, parms) #starting estimate for this model #init returns \sigma^2, I need log(sigma) # We sometimes get into trouble with a small estimate of sigma, # (the surface isn't SPD), but never with a large one. Double it. if (scale >0) vars <- log(scale) else vars <- log(4*coef[2])/2 # log(2*sqrt(variance)) = log(4*var)/2 coef <- c(coef[1], rep(vars, nstrat)) # get a better initial value for the mean using the "glim" trick deriv <- derfun(y, yy, exp(vars), sd$density, parms) wt <- -1*deriv$ddg*weights coef[1] <- sum(weights*deriv$dg + wt*(yy -offset)) / sum(wt) # Now the fit proper (intercept only) fit0 <- .Call(Csurvreg6, iter = as.integer(20), nvar = as.integer(1), as.double(y), as.integer(ny), x = as.double(rep(1.0, n)), as.double(weights), as.double(offset), coef= as.double(coef), as.integer(nstrat2), as.integer(strata), as.double(eps), as.double(toler.chol), as.integer(dnum), f.expr, rho) } # # Fit the model with all covariates # if (is.numeric(init)) { if (length(init) == nvar && (nvar2 > nvar)) { # Add on the variance estimates from above init <- c(init, fit0$coef[-1]) } if (length(init) != nvar2) stop("Wrong length for initial parameters") if (scale >0) init <- c(init, log(scale)) } else { # Do the 'glim' method of finding an initial value of coef if (meanonly) { yy <- ifelse(y[,ny]!=3, y[,1], (y[,1]+y[,2])/2 ) coef <- sd$init(yy, weights, parms) if (scale >0) vars <- rep(log(scale), nstrat) else vars <- rep(log(4*coef[2])/2, nstrat) } else vars <- fit0$coef[-1] eta <- yy - offset #what would be true for a 'perfect' model deriv <- derfun(y, yy, exp(vars[strata]), sd$density, parms) wt <- -1*deriv$ddg*weights coef <- coxph.wtest(t(x)%*% (wt*x), c((wt*eta + weights*deriv$dg)%*% x), toler.chol=toler.chol)$solve init <- c(coef, vars) } # Now for the fit in earnest fit <- .Call(Csurvreg6, iter = as.integer(iter.max), as.integer(nvar), as.double(y), as.integer(ny), as.double(x), as.double(weights), as.double(offset), as.double(init), as.integer(nstrat2), as.integer(strata), as.double(eps), as.double(toler.chol), as.integer(dnum), f.expr, rho) if (iter.max >1 && fit$flag > nvar2) { warning("Ran out of iterations and did not converge") } cname <- dimnames(x)[[2]] if (is.null(cname)) cname <- paste("x", 1:ncol(x)) if (scale==0) cname <- c(cname, rep("Log(scale)", nstrat)) if (scale>0) fit$coef <- fit$coef[1:nvar2] names(fit$coef) <- cname if (meanonly) { coef0 <- fit$coef loglik <- rep(fit$loglik,2) } else { coef0 <- fit0$coef names(coef0) <- c("Intercept", rep("Log(scale)", nstrat)) loglik <- c(fit0$loglik, fit$loglik) } temp <- list(coefficients = fit$coef, icoef = coef0, var = matrix(fit$var, nvar2, dimnames=list(cname, cname)), loglik = loglik, iter = fit$iter, linear.predictors = c(x %*% fit$coef[1:nvar] + offset), df= length(fit$coef), score = fit$u ) temp } survival/R/survreg.distributions.S0000644000175100001440000001045712377203072017100 0ustar hornikusers# # Create the survreg.distributions object # survreg.distributions <- list( 'extreme' = list( name = "Extreme value", variance = function(parm) pi^2/6, init = function(x, weights, ...) { mean <- sum(x*weights)/ sum(weights) var <- sum(weights*(x-mean)^2)/ sum(weights) c(mean + .572, var/1.64) }, deviance= function(y, scale, parms) { status <- y[,ncol(y)] width <- ifelse(status==3,(y[,2] - y[,1])/scale, 1) temp <- width/(exp(width)-1) # the definition of "center" is discussed in the parametric # section of the survival document center <- ifelse(status==3, y[,1] - log(temp), y[,1]) temp3 <- (-temp) + log(1- exp(-exp(width))) loglik <- ifelse(status==1, -(1+log(scale)), ifelse(status==3, temp3, 0)) list(center=center, loglik=loglik) }, density = function(x,parms) { w <- exp(x) ww <- exp(-w) cbind(1-ww, ww, w*ww, (1-w), w*(w-3) +1) }, quantile = function(p,parms) log(-log(1-p)) ), logistic = list( name = "Logistic", variance = function(parm) pi^2/3, init = function(x, weights, ...) { mean <- sum(x*weights)/ sum(weights) var <- sum(weights*(x-mean)^2)/ sum(weights) c(mean, var/3.2) }, deviance= function(y, scale, parms) { status <- y[,ncol(y)] width <- ifelse(status==3,(y[,2] - y[,1])/scale, 0) # for the symmetric distributions "center" is obvious center <- ifelse(status==3, rowMeans(y), y[,1]) temp2 <- ifelse(status==3, exp(width/2), 2) #avoid a log(0) message temp3 <- log((temp2-1)/(temp2+1)) loglik <- ifelse(status==1, -log(4*scale), ifelse(status==3, temp3, 0)) list(center=center, loglik=loglik) }, density = function(x, parms) { w <- exp(x) cbind(w/(1+w), 1/(1+w), w/(1+w)^2, (1-w)/(1+w), (w*(w-4) +1)/(1+w)^2) }, quantile = function(p, parms) log(p/(1-p)) ), gaussian = list( name = "Gaussian", variance = function(parm) 1, init = function(x, weights, ...) { mean <- sum(x*weights)/ sum(weights) var <- sum(weights*(x-mean)^2)/ sum(weights) c(mean, var) }, deviance= function(y, scale, parms) { status <- y[,ncol(y)] width <- ifelse(status==3,(y[,2] - y[,1])/scale, 0) center <- ifelse(status==3, rowMeans(y), y[,1]) temp2 <- log(1 - 2*pnorm(width/2)) loglik <- ifelse(status==1, -log(sqrt(2*pi)*scale), ifelse(status==3, temp2, 0)) list(center=center, loglik=loglik) }, density = function(x, parms) { cbind(pnorm(x), pnorm(-x), dnorm(x), -x, x^2-1) }, quantile = function(p, parms) qnorm(p) ), weibull = list( name = "Weibull", dist = 'extreme', trans = function(y) log(y), dtrans= function(y) 1/y , itrans= function(x) exp(x) ), exponential = list( name = "Exponential", dist = 'extreme', trans = function(y) log(y), dtrans= function(y) 1/y, scale =1, itrans= function(x) exp(x) ), rayleigh = list( name = "Rayleigh", dist = 'extreme', trans = function(y) log(y), dtrans= function(y) 1/y, itrans= function(x) exp(x), scale =0.5 ), loggaussian = list( name = "Log Normal", dist = 'gaussian', trans = function(y) log(y), itrans= function(x) exp(x), dtrans= function(y) 1/y ), lognormal = list( name = "Log Normal", dist = 'gaussian', trans = function(y) log(y), itrans= function(x) exp(x), dtrans= function(y) 1/y ), loglogistic = list( name = "Log logistic", dist = 'logistic', trans = function(y) log(y), dtrans= function(y) 1/y , itrans= function(x) exp(x) ), t = list( name = "Student-t", variance = function(df) df/(df-2), parms = c(df=4), init = function(x, weights, df) { if (df <=2) stop ("Degrees of freedom must be >=3") mean <- sum(x*weights)/ sum(weights) var <- sum(weights*(x-mean)^2)/ sum(weights) c(mean, var*(df-2)/df) }, deviance= function(y, scale, parms) { status <- y[,ncol(y)] width <- ifelse(status==3,(y[,2] - y[,1])/scale, 0) center <- ifelse(status==3, rowMeans(y), y[,1]) temp2 <- log(1 - 2*pt(width/2, df=parms)) loglik <- ifelse(status==1, -log(dt(0, df=parms)*scale), ifelse(status==3, temp2, 0)) list(center=center, loglik=loglik) }, density = function(x, df) { cbind(pt(x, df), pt(-x, df), dt(x,df), -(df+1)*x/(df+x^2), (df+1)*(x^2 *(df+3)/(df+x^2) - 1)/(df +x^2)) }, quantile = function(p, df) qt(p, df) ) ) survival/R/summary.aareg.S0000644000175100001440000001033212055773252015252 0ustar hornikusers# The summary routine for aareg models. # A lot of the work below relates to one particular issue: the coeffients # of an aareg model often get "wild" near the end (at the largest times). # So, a common case is to # fit the model (very slow) # look at the printout -- Hmmm x1 is significant, x2 not, ...., why? # look at plot(fit) and # oh my gosh, I should have cut the time scale off at 520 days # # This routine allows one to do that. If maxtime is given, the overall # test statistic is re-computed. One consequence is that lots of the # intermediate material from the fit had to be included in the aareg # object. # The "variance" based weighting for a test is not allowed, because it would # have meant an awful lot more stuff to pass, lots more work, for a test # that is rarely used. # summary.aareg <- function(object, maxtime, test=c('aalen', 'nrisk'), scale=1,...) { if (!inherits(object, 'aareg')) stop ("Must be an aareg object") if (missing(test)) test <- object$test test <- match.arg(test) if (!missing(maxtime)) ntime <- sum(object$time <= maxtime) else ntime <- nrow(object$coefficient) times <- object$time[1:ntime] if (test=='aalen') { twt <- (as.matrix(object$tweight))[1:ntime,] scale <- apply(twt, 2, sum)/scale } else { twt <- object$nrisk[1:ntime] scale <- ntime/scale } # Compute a "slope" for each line, using appropriate weighting # Since this is a single variable model, no intercept, I # don't need to call lm.wfit! tx <- as.matrix(twt * object$coefficient[1:ntime,]) ctx <- apply(tx, 2, cumsum) if (is.matrix(twt) && ncol(twt) >1) tempwt <- apply(twt*times^2, 2, sum) else tempwt <- sum(twt*times^2) if (ncol(ctx) >1) slope<- apply(ctx* times, 2, sum)/ tempwt else slope <- sum(ctx*times) / tempwt if (!missing(maxtime) || object$test != test) { # Compute the test statistic test.stat <- apply(tx, 2, sum) #sum of tested coefficients test.var <- t(tx) %*% tx #std Poisson, ind coefficients variance if (!is.null(object$dfbeta)) { dd <- dim(object$dfbeta) indx <- match(unique(times), object$times) influence <- matrix(0, dd[1], dd[2]) for (i in 1:length(indx)) { if (test=='aalen') influence <- influence + object$dfbeta[,,i] %*% diag(twt[indx[i],]) else influence <- influence + object$dfbeta[,,i]* object$nrisk[indx[i]] } if (!is.null(object$cluster)) influence <- rowsum(influence, cluster) test.var2 <- t(influence) %*% influence } else test.var2 <- NULL } else { #use the value that was passed in test.stat <- object$test.statistic test.var <- object$test.var test.var2 <- object$test.var2 #NULL if dfbeta option was false } # create the matrix for printing out # The chisquare test does not include the intercept se1 <- sqrt(diag(test.var)) if (is.null(test.var2)) { mat <- cbind(slope, test.stat/scale, se1/scale, test.stat/se1, 2*pnorm(-abs(test.stat/se1))) dimnames(mat) <- list((dimnames(object$coefficient)[[2]]), c("slope", "coef", "se(coef)", "z", "p")) chi <- test.stat[-1] %*% solve(test.var[-1,-1],test.stat[-1]) } else { se2 <- sqrt(diag(test.var2)) mat <- cbind(slope, test.stat/scale, se1/scale, se2/scale, test.stat/se2, 2*pnorm(-abs(test.stat/se2))) dimnames(mat) <- list((dimnames(object$coefficient)[[2]]), c("slope", "coef", "se(coef)", "robust se", "z", "p")) chi <- test.stat[-1] %*% solve(test.var2[-1,-1], test.stat[-1]) } temp <- list(table=mat, test=test, test.statistic=test.stat, test.var=test.var, test.var2=test.var2, chisq=chi, n = c(object$n[1], length(unique(times)), object$n[3])) if (is.R()) class(temp) <- 'summary.aareg' else oldClass(temp) <- 'summary.aareg' temp } print.summary.aareg <- function(x, ...) { print(signif(x$table,3)) chi <- x$chisq df <- length(x$test.statistic) -1 cat("\nChisq=", format(round(chi,2)), " on ", df, " df, p=", signif(1- pchisq(chi, df),2), "; test weights=", x$test, "\n", sep='') invisible(x$table) } survival/R/predict.coxph.penal.S0000644000175100001440000000701212470201064016334 0ustar hornikusers# $Id: predict.coxph.penal.S 11516 2012-04-24 12:49:14Z therneau $ predict.coxph.penal <- function(object, newdata, type=c("lp", "risk", "expected", "terms"), se.fit=FALSE, terms=names(object$assign), collapse, safe=FALSE, ...) { type <- match.arg(type) n <- object$n pterms <- object$pterms # If there are no sparse terms if (!any(pterms==2) || (missing(newdata) && se.fit==FALSE && type!='terms')) NextMethod('predict',object,...) else { # treat the sparse term as an offset term # It gets picked up in the linear predictor, so all I need to # do is "X" it out of the model so that it doesn't get picked up # as a part of the X matrix and etc. # I know that the sparse term is a single column BTW # termname <- names(object$pterms) sparsename <- termname[object$pterms==2] nvar <- length(termname) na.action <- object$na.action object$na.action <- NULL if (missing(newdata) && (se.fit || type=='terms')) { # I need the X matrix x <- object[['x']] # object$x might grab object$xlevels if (is.null(x)) { temp <- coxph.getdata(object, y=TRUE, x=TRUE, stratax=TRUE) if (is.null(object$y)) object$y <- temp$y if (is.null(object$strata)) object$strata <- temp$strata x <- temp$x } xvar <- match(sparsename, dimnames(x)[[2]]) indx <- as.numeric(as.factor(x[,xvar])) object$x <- x[, -xvar, drop=FALSE] } if (nvar==1) { # Only the sparse term! if (!missing(newdata)) { n <- nrow(as.data.frame(newdata)) pred <- rep(0,n) } se <- sqrt(object$fvar[indx]) pred <- object$linear.predictor if (type=='risk') pred <- exp(pred) if (type=='expected') { pred <- object$y[,ncol(object$y)] -object$residuals se.fit=FALSE } } else { # temporarily remove the sparse term, call NextMethod, # and then put it back oldTerms <- object$terms temp <- attr(object$terms, 'term.labels') object$terms <- object$terms[-match(sparsename, temp)] pred <- NextMethod('predict',object,terms=terms,...) object$terms<- oldTerms if (se.fit) { se <- pred$se.fit pred <- pred$fit } if (type=='terms' && missing(newdata)) { # In this case (only) I add the sparse term back in spterm <- object$frail[indx] spstd <- sqrt(object$fvar[indx]) if (nvar==2) { if (xvar==2) { pred <- cbind(pred, spterm) if (se.fit) se <- cbind(se, spstd) } else { pred <- cbind(spterm, pred) if (se.fit) se <- cbind(spstd, se) } } else { first <- if (xvar==1) 0 else 1:(xvar-1) secnd <- if (xvar==nvar) 0 else (xvar+1):nvar pred <- cbind(pred[,first], spterm, pred[,secnd]) if (se.fit) se <- cbind(se[,first], spstd, se[,secnd]) } dimnames(pred) <- list(dimnames(x)[[1]], termname) if (se.fit) dimnames(se) <- dimnames(pred) } } #Expand out the missing values in the result # But only if operating on the original dataset if (missing(newdata) && !is.null(na.action)) { pred <- naresid(na.action, pred) if (is.matrix(pred)) n <- nrow(pred) else n <- length(pred) if(se.fit) se <- naresid(na.action, se) } # Collapse over subjects, if requested if (!missing(collapse)) { if (length(collapse) != n) stop("Collapse vector is the wrong length") pred <- drop(rowsum(pred, collapse)) if (se.fit) se <- sqrt(drop(rowsum(se^2, collapse))) } if (se.fit) list(fit=pred, se.fit=se) else pred } } survival/R/frailty.gamma.S0000644000175100001440000001154312444044357015236 0ustar hornikusers# # Defining function for gamma frailty fits # frailty.gamma <- function(x, sparse=(nclass >5), theta, df, eps= 1e-5, method=c("em", "aic", "df", "fixed"), ...) { nclass <- length(unique(x[!is.na(x)])) if (sparse) x <-as.numeric(factor(x)) #drop extra levels if a factor else{ x <- factor(x) attr(x,'contrasts') <- contr.treatment(nclass, contrasts=FALSE) } if (is.R()) class(x) <- c("coxph.penalty",class(x)) else oldClass(x) <- "coxph.penalty" # Check for consistency of the arguments if (missing(method)) { if (!missing(theta)) { method <- 'fixed' if (!missing(df)) stop("Cannot give both a df and theta argument") } else if (!missing(df)) method <- 'df' } method <- match.arg(method) if (method=='df' && missing(df)) stop("Method = df but no df argument") if (method=='fixed' && missing(theta)) stop("Method= fixed but no theta argument") if (method!='df' && !missing(df)) stop("Method is not df, but have a df argument") if (method !='fixed' && !missing(theta)) stop("Method is not 'fixed', but have a theta argument") pfun<- function(coef, theta, ndeath){ if (theta==0) list(recenter=0, penalty=0, flag=TRUE) else { recenter <- log(mean(exp(coef))) coef <- coef - recenter nu <- 1/theta list(recenter=recenter, first= (exp(coef) -1) * nu, second= exp(coef) * nu, penalty= -sum(coef)*nu, # The exp part sums to a constant flag=FALSE) } } printfun <- function(coef, var, var2, df, history) { if (!is.null(history$history)) theta <- history$history[nrow(history$history),1] else theta <- history$theta clog <- history$c.loglik if (is.matrix(var)) test <- coxph.wtest(var, coef)$test else test <- sum(coef^2/var) df2 <- max(df, .5) # Stop silly p-values list(coef=c(NA, NA, NA, test, df, 1-pchisq(test, df2)), history=paste("Variance of random effect=", format(theta), " I-likelihood =", format(round(clog,1), digits=10))) } # The final coxph object will contain a copy of printfun. Stop it from # also containing huge unnecessary variables, e.g. 'x', known at this # point in time. Not an issue for pfun, which does not get saved. # Setting to globalenv() will not suffice since coxph.wtest is not visible # outside the survival library's name space. temp <- new.env(parent=globalenv()) assign("cox.zph", cox.zph, envir=temp) #make a private copy environment(printfun) <- temp if (method=='fixed') { temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c("x", "status", "loglik"), cfun = frailty.controlgam, cparm= list(theta=theta, ...)) } else if (method=='em'){ temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c("x", "status", "loglik"), cfun = frailty.controlgam, cparm= c(list(eps=eps), ...)) } else if (method=='aic') { temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs = c("x", "status", "loglik", "neff","df", "plik"), cparm=list(eps=eps, lower=0, init=c(.1, 1), ...), cfun =function(opt, iter, old, group, status, loglik,...){ temp <- frailty.controlaic(opt, iter, old, ...) if (iter >0) { #compute correction to the loglik if (old$theta==0) correct <- 0 else { if (is.matrix(group)) group <-c(group %*% 1:ncol(group)) d <- tapply(status,group,sum) correct <- frailty.gammacon(d, 1/old$theta) } temp$c.loglik <- loglik + correct } temp }) } else { #df method # The initial guess is based on the observation that theta=1 often # gives about df= (#groups)/3 if (missing(eps)) eps <- .1 temp <- list(pfun=pfun, printfun=printfun, diag =TRUE, sparse= sparse, cargs= c('df', "x", "status", "loglik"), cparm=list(df=df, thetas=0, dfs=0, eps=eps, guess=3*df/length(unclass(x)), ...), cfun =function(opt, iter, old, df, group, status, loglik){ temp <- frailty.controldf(opt, iter, old, df) if (iter >0) { #compute correction to the loglik if (old$theta==0) correct <- 0 else { if (is.matrix(group)) group <-c(group %*% 1:ncol(group)) d <- tapply(status,group,sum) correct <- frailty.gammacon(d, 1/old$theta) } temp$c.loglik <- loglik + correct } temp }) } # If not sparse, give shorter names to the coefficients, so that any # printout of them is readable. if (!sparse) { vname <- paste("gamma", levels(x), sep=':') temp <- c(temp, list(varname=vname)) } attributes(x) <- c(attributes(x), temp) x } survival/R/survConcordance.fit.R0000644000175100001440000000543412533653606016425 0ustar hornikusers# Automatically generated from all.nw using noweb survConcordance.fit <- function(y, x, strata, weight) { btree <- function(n) { ranks <- rep(0L, n) #will be overwritten yet.to.do <- 1:n depth <- floor(logb(n,2)) start <- as.integer(2^depth) lastrow.length <- 1+n-start indx <- seq(1L, by=2L, length= lastrow.length) ranks[yet.to.do[indx]] <- start + 0:(length(indx)-1L) yet.to.do <- yet.to.do[-indx] while (start >1) { start <- as.integer(start/2) indx <- seq(1L, by=2L, length=start) ranks[yet.to.do[indx]] <- start + 0:(start-1L) yet.to.do <- yet.to.do[-indx] } ranks } docount <- function(stime, risk, wts) { if (attr(stime, 'type') == 'right') { ord <- order(stime[,1], -stime[,2]) ux <- sort(unique(risk)) n2 <- length(ux) index <- btree(n2)[match(risk[ord], ux)] - 1L .Call(Cconcordance1, stime[ord,], as.double(wts[ord]), as.integer(index), as.integer(length(ux))) } else if (attr(stime, 'type') == "counting") { sort.stop <- order(-stime[,2], stime[,3]) sort.start <- order(-stime[,1]) ux <- sort(unique(risk)) n2 <- length(ux) index <- btree(n2)[match(risk, ux)] - 1L .Call(Cconcordance2, stime, as.double(wts), as.integer(index), as.integer(length(ux)), as.integer(sort.stop-1L), as.integer(sort.start-1L)) } else stop("Invalid survival type for concordance") } if (missing(weight) || length(weight)==0) weight <- rep(1.0, length(x)) storage.mode(y) <- "double" if (missing(strata) || length(strata)==0) { count <- docount(y, x, weight) if (count[1]==0 && count[2]==0) count[5]<-0 else count[5] <- 2*sqrt(count[5]) names(count) <- c("concordant", "discordant", "tied.risk", "tied.time", "std(c-d)") } else { strata <- as.factor(strata) ustrat <- levels(strata)[table(strata) >0] #some strata may have 0 obs count <- matrix(0., nrow=length(ustrat), ncol=5) for (i in 1:length(ustrat)) { keep <- which(strata == ustrat[i]) count[i,] <- docount(y[keep,,drop=F], x[keep], weight[keep]) } count[,5] <- 2*sqrt(ifelse(count[,1]+count[,2]==0, 0, count[,5])) dimnames(count) <- list(ustrat, c("concordant", "discordant", "tied.risk", "tied.time", "std(c-d)")) } count } survival/R/print.survdiff.S0000644000175100001440000000236211732700061015453 0ustar hornikusers# $Date: 2006-08-28 14:31:20 $ $Id: print.survdiff.S 11166 2008-11-24 22:10:34Z therneau $ print.survdiff <- function(x, digits = max(options()$digits - 4, 3), ...) { saveopt <-options(digits=digits) on.exit(options(saveopt)) if (!inherits(x, 'survdiff')) stop("Object is not the result of survdiff") if (!is.null(cl<- x$call)) { cat("Call:\n") dput(cl) cat("\n") } omit <- x$na.action if (length(omit)) cat("n=", sum(x$n), ", ", naprint(omit), ".\n\n", sep='') if (length(x$n)==1) { z <- sign(x$exp - x$obs) * sqrt(x$chisq) temp <- c(x$obs, x$exp, z, signif(1-pchisq(x$chisq, 1),digits)) names(temp) <- c("Observed", "Expected", "Z", "p") print(temp) } else { if (is.matrix(x$obs)){ otmp <- apply(x$obs,1,sum) etmp <- apply(x$exp,1,sum) } else { otmp <- x$obs etmp <- x$exp } df <- (sum(1*(etmp>0))) -1 temp <- cbind(x$n, otmp, etmp, ((otmp-etmp)^2)/ etmp, ((otmp-etmp)^2)/ diag(x$var)) dimnames(temp) <- list(names(x$n), c("N", "Observed", "Expected", "(O-E)^2/E", "(O-E)^2/V")) print(temp) cat("\n Chisq=", format(round(x$chisq,1)), " on", df, "degrees of freedom, p=", format(signif(1-pchisq(x$chisq, df),digits)), "\n") } invisible(x) } survival/R/lines.survfit.coxph.S0000644000175100001440000000042011732700061016414 0ustar hornikusers# $Id: lines.survfit.coxph.S 11166 2008-11-24 22:10:34Z therneau $ lines.survfit.coxph <- function(x, mark.time=FALSE, ...) { if (is.logical(mark.time) & mark.time) stop("Invalid value for mark.time") invisible(NextMethod('lines', mark.time=mark.time)) } survival/R/coxph.R0000644000175100001440000003520712533653600013622 0ustar hornikusers# Automatically generated from all.nw using noweb #tt <- function(x) x coxph <- function(formula, data, weights, subset, na.action, init, control, ties= c("efron", "breslow", "exact"), singular.ok =TRUE, robust=FALSE, model=FALSE, x=FALSE, y=TRUE, tt, method=ties, ...) { ties <- match.arg(ties) Call <- match.call() # create a call to model.frame() that contains the formula (required) # and any other of the relevant optional arguments # then evaluate it in the proper frame indx <- match(c("formula", "data", "weights", "subset", "na.action"), names(Call), nomatch=0) if (indx[1] ==0) stop("A formula argument is required") temp <- Call[c(1,indx)] # only keep the arguments we wanted temp[[1]] <- as.name('model.frame') # change the function called special <- c("strata", "cluster", "tt") temp$formula <- if(missing(data)) terms(formula, special) else terms(formula, special, data=data) # Make "tt" visible for coxph formulas, without making it visible elsewhere if (!is.null(attr(temp$formula, "specials")$tt)) { coxenv <- new.env(parent= environment(formula)) assign("tt", function(x) x, env=coxenv) environment(temp$formula) <- coxenv } mf <- eval(temp, parent.frame()) if (nrow(mf) ==0) stop("No (non-missing) observations") Terms <- terms(mf) ## We want to pass any ... args to coxph.control, but not pass things ## like "dats=mydata" where someone just made a typo. The use of ... ## is simply to allow things like "eps=1e6" with easier typing extraArgs <- list(...) if (length(extraArgs)) { controlargs <- names(formals(coxph.control)) #legal arg names indx <- pmatch(names(extraArgs), controlargs, nomatch=0L) if (any(indx==0L)) stop(gettextf("Argument %s not matched", names(extraArgs)[indx==0L]), domain = NA) } if (missing(control)) control <- coxph.control(...) Y <- model.extract(mf, "response") if (!inherits(Y, "Surv")) stop("Response must be a survival object") type <- attr(Y, "type") if (type!='right' && type!='counting') stop(paste("Cox model doesn't support \"", type, "\" survival data", sep='')) data.n <- nrow(Y) #remember this before any time transforms if (length(attr(Terms, 'variables')) > 2) { # a ~1 formula has length 2 ytemp <- terms.inner(attr(Terms, 'variables')[1:2]) xtemp <- terms.inner(attr(Terms, 'variables')[-2]) if (any(!is.na(match(xtemp, ytemp)))) warning("a variable appears on both the left and right sides of the formula") } # The time transform will expand the data frame mf. To do this # it needs Y and the strata. Everything else (cluster, offset, weights) # should be extracted after the transform # strats <- attr(Terms, "specials")$strata if (length(strats)) { stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } timetrans <- attr(Terms, "specials")$tt if (missing(tt)) tt <- NULL if (length(timetrans)) { timetrans <- untangle.specials(Terms, 'tt') ntrans <- length(timetrans$terms) if (is.null(tt)) { tt <- function(x, time, riskset, weights){ #default to O'Brien's logit rank obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } } if (is.function(tt)) tt <- list(tt) #single function becomes a list if (is.list(tt)) { if (any(!sapply(tt, is.function))) stop("The tt argument must contain function or list of functions") if (length(tt) != ntrans) { if (length(tt) ==1) { temp <- vector("list", ntrans) for (i in 1:ntrans) temp[[i]] <- tt[[1]] tt <- temp } else stop("Wrong length for tt argument") } } else stop("The tt argument must contain a function or list of functions") if (ncol(Y)==2) { if (length(strats)==0) { sorted <- order(-Y[,1], Y[,2]) newstrat <- rep.int(0L, nrow(Y)) newstrat[1] <- 1L } else { sorted <- order(strats, -Y[,1], Y[,2]) #newstrat marks the first obs of each strata newstrat <- as.integer(c(1, 1*(diff(strats[sorted])!=0))) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount1, Y[sorted,], as.integer(newstrat)) tindex <- sorted[counts$index] } else { if (length(strats)==0) { sort.end <- order(-Y[,2], Y[,3]) sort.start<- order(-Y[,1]) newstrat <- c(1L, rep(0, nrow(Y) -1)) } else { sort.end <- order(strats, -Y[,2], Y[,3]) sort.start<- order(strats, -Y[,1]) newstrat <- c(1L, as.integer(diff(strats[sort.end])!=0)) } if (storage.mode(Y) != "double") storage.mode(Y) <- "double" counts <- .Call(Ccoxcount2, Y, as.integer(sort.start -1L), as.integer(sort.end -1L), as.integer(newstrat)) tindex <- counts$index } mf <- mf[tindex,] Y <- Surv(rep(counts$time, counts$nrisk), counts$status) type <- 'right' # new Y is right censored, even if the old was (start, stop] strats <- rep(1:length(counts$nrisk), counts$nrisk) weights <- model.weights(mf) if (!is.null(weights) && any(!is.finite(weights))) stop("weights must be finite") for (i in 1:ntrans) mf[[timetrans$var[i]]] <- (tt[[i]])(mf[[timetrans$var[i]]], Y[,1], strats, weights) } cluster<- attr(Terms, "specials")$cluster if (length(cluster)) { robust <- TRUE #flag to later compute a robust variance tempc <- untangle.specials(Terms, 'cluster', 1:10) ord <- attr(Terms, 'order')[tempc$terms] if (any(ord>1)) stop ("Cluster can not be used in an interaction") cluster <- strata(mf[,tempc$vars], shortlabel=TRUE) #allow multiples dropterms <- tempc$terms #we won't want this in the X matrix # Save away xlevels after removing cluster (we don't want to save upteen # levels of that variable, which we will never need). xlevels <- .getXlevels(Terms[-tempc$terms], mf) } else { dropterms <- NULL if (missing(robust)) robust <- FALSE xlevels <- .getXlevels(Terms, mf) } contrast.arg <- NULL #due to shared code with model.matrix.coxph attr(Terms, "intercept") <- 1 adrop <- 0 #levels of "assign" to be dropped; 0= intercept stemp <- untangle.specials(Terms, 'strata', 1) if (length(stemp$vars) > 0) { #if there is a strata statement hasinteractions <- FALSE for (i in stemp$vars) { #multiple strata terms are allowed # The factors att has one row for each variable in the frame, one # col for each term in the model. Pick rows for each strata # var, and find if it participates in any interactions. if (any(attr(Terms, 'order')[attr(Terms, "factors")[i,] >0] >1)) hasinteractions <- TRUE } if (!hasinteractions) dropterms <- c(dropterms, stemp$terms) else adrop <- c(0, match(stemp$var, colnames(attr(Terms, 'factors')))) } if (length(dropterms)) { temppred <- attr(terms, "predvars") Terms2 <- Terms[ -dropterms] if (!is.null(temppred)) { # subscripting a Terms object currently drops predvars, in error attr(Terms2, "predvars") <- temppred[-(1+dropterms)] # "Call" object } X <- model.matrix(Terms2, mf, constrasts=contrast.arg) # we want to number the terms wrt the original model matrix # Do not forget the intercept, which will be a zero renumber <- match(colnames(attr(Terms2, "factors")), colnames(attr(Terms, "factors"))) attr(X, "assign") <- c(0, renumber)[1+attr(X, "assign")] } else X <- model.matrix(Terms, mf, contrasts=contrast.arg) # drop the intercept after the fact, and also drop strata if necessary Xatt <- attributes(X) xdrop <- Xatt$assign %in% adrop #columns to drop (always the intercept) X <- X[, !xdrop, drop=FALSE] attr(X, "assign") <- Xatt$assign[!xdrop] #if (any(adrop>0)) attr(X, "contrasts") <- Xatt$contrasts[-adrop] #else attr(X, "contrasts") <- Xatt$contrasts attr(X, "contrasts") <- Xatt$contrasts offset <- model.offset(mf) if (is.null(offset) | all(offset==0)) offset <- rep(0., nrow(mf)) else if (any(!is.finite(offset))) stop("offsets must be finite") weights <- model.weights(mf) if (!is.null(weights) && any(!is.finite(weights))) stop("weights must be finite") assign <- attrassign(X, Terms) contr.save <- attr(X, "contrasts") if (missing(init)) init <- NULL else { if (length(init) != ncol(X)) stop("wrong length for init argument") temp <- X %*% init - sum(colMeans(X) * init) if (any(temp < .Machine$double.min.exp | temp > .Machine$double.max.exp)) stop("initial values lead to overflow or underflow of the exp function") } pterms <- sapply(mf, inherits, 'coxph.penalty') if (any(pterms)) { pattr <- lapply(mf[pterms], attributes) pname <- names(pterms)[pterms] # # Check the order of any penalty terms ord <- attr(Terms, "order")[match(pname, attr(Terms, 'term.labels'))] if (any(ord>1)) stop ('Penalty terms cannot be in an interaction') pcols <- assign[match(pname, names(assign))] fit <- coxpenal.fit(X, Y, strats, offset, init=init, control, weights=weights, method=method, row.names(mf), pcols, pattr, assign) } else { if( method=="breslow" || method =="efron") { if (type== 'right') fitter <- get("coxph.fit") else fitter <- get("agreg.fit") } else if (method=='exact') { if (type== "right") fitter <- get("coxexact.fit") else fitter <- get("agexact.fit") } else stop(paste ("Unknown method", method)) fit <- fitter(X, Y, strats, offset, init, control, weights=weights, method=method, row.names(mf)) } if (is.character(fit)) { fit <- list(fail=fit) class(fit) <- 'coxph' } else { if (!is.null(fit$coefficients) && any(is.na(fit$coefficients))) { vars <- (1:length(fit$coefficients))[is.na(fit$coefficients)] msg <-paste("X matrix deemed to be singular; variable", paste(vars, collapse=" ")) if (singular.ok) warning(msg) else stop(msg) } fit$n <- data.n fit$nevent <- sum(Y[,ncol(Y)]) fit$terms <- Terms fit$assign <- assign class(fit) <- fit$method if (robust) { fit$naive.var <- fit$var fit$method <- method # a little sneaky here: by calling resid before adding the # na.action method, I avoid having missings re-inserted # I also make sure that it doesn't have to reconstruct X and Y fit2 <- c(fit, list(x=X, y=Y, weights=weights)) if (length(strats)) fit2$strata <- strats if (length(cluster)) { temp <- residuals.coxph(fit2, type='dfbeta', collapse=cluster, weighted=TRUE) # get score for null model if (is.null(init)) fit2$linear.predictors <- 0*fit$linear.predictors else fit2$linear.predictors <- c(X %*% init) temp0 <- residuals.coxph(fit2, type='score', collapse=cluster, weighted=TRUE) } else { temp <- residuals.coxph(fit2, type='dfbeta', weighted=TRUE) fit2$linear.predictors <- 0*fit$linear.predictors temp0 <- residuals.coxph(fit2, type='score', weighted=TRUE) } fit$var <- t(temp) %*% temp u <- apply(as.matrix(temp0), 2, sum) fit$rscore <- coxph.wtest(t(temp0)%*%temp0, u, control$toler.chol)$test } #Wald test if (length(fit$coefficients) && is.null(fit$wald.test)) { #not for intercept only models, or if test is already done nabeta <- !is.na(fit$coefficients) # The init vector might be longer than the betas, for a sparse term if (is.null(init)) temp <- fit$coefficients[nabeta] else temp <- (fit$coefficients - init[1:length(fit$coefficients)])[nabeta] fit$wald.test <- coxph.wtest(fit$var[nabeta,nabeta], temp, control$toler.chol)$test } na.action <- attr(mf, "na.action") if (length(na.action)) fit$na.action <- na.action if (model) { if (length(timetrans)) { # Fix up the model frame -- still in the thinking stage mf[[".surv."]] <- Y mf[[".strata."]] <- strats stop("Time transform + model frame: code incomplete") } fit$model <- mf } if (x) { fit$x <- X if (length(strats)) { if (length(timetrans)) fit$strata <- strats else fit$strata <- strata.keep } } if (y) fit$y <- Y } if (!is.null(weights) && any(weights!=1)) fit$weights <- weights names(fit$means) <- names(fit$coefficients) fit$formula <- formula(Terms) if (length(xlevels) >0) fit$xlevels <- xlevels fit$contrasts <- contr.save if (any(offset !=0)) fit$offset <- offset fit$call <- Call fit$method <- method fit } survival/R/agexact.fit.S0000644000175100001440000000545212113164602014667 0ustar hornikusers agexact.fit <- function(x, y, strata, offset, init, control, weights, method, rownames) { if (!is.matrix(x)) stop("Invalid formula for cox fitting function") if (!is.null(weights) && any(weights!=1)) stop("Case weights are not supported for the exact method") n <- nrow(x) nvar <- ncol(x) if (ncol(y)==3) { start <- y[,1] stopp <- y[,2] event <- y[,3] } else { start <- rep(0,n) stopp <- y[,1] event <- y[,2] } # Sort the data (or rather, get a list of sorted indices) if (length(strata)==0) { sorted <- order(stopp, -event) newstrat <- as.integer(rep(0,n)) } else { sorted <- order(strata, stopp, -event) strata <- (as.numeric(strata))[sorted] newstrat <- as.integer(c(1*(diff(strata)!=0), 1)) } if (is.null(offset)) offset <- rep(0,n) sstart <- as.double(start[sorted]) sstop <- as.double(stopp[sorted]) sstat <- as.integer(event[sorted]) if (is.null(nvar)) { # A special case: Null model. Not worth coding up stop("Cannot handle a null model + exact calculation (yet)") } if (!is.null(init)) { if (length(init) != nvar) stop("Wrong length for inital values") } else init <- rep(0,nvar) agfit <- .C(Cagexact, iter= as.integer(control$iter.max), as.integer(n), as.integer(nvar), sstart, sstop, sstat, x= x[sorted,], as.double(offset[sorted]), newstrat, means = double(nvar), coef= as.double(init), u = double(nvar), imat= double(nvar*nvar), loglik=double(2), flag=integer(1), double(2*nvar*nvar +nvar*4 + n), integer(2*n), as.double(control$eps), as.double(control$toler.chol), sctest=double(1)) var <- matrix(agfit$imat,nvar,nvar) coef <- agfit$coef if (agfit$flag < nvar) which.sing <- diag(var)==0 else which.sing <- rep(FALSE,nvar) infs <- abs(agfit$u %*% var) if (control$iter.max >1) { if (agfit$flag == 1000) warning("Ran out of iterations and did not converge") else { infs <- ((infs > control$eps) & infs > control$toler.inf*abs(coef)) if (any(infs)) warning(paste("Loglik converged before variable ", paste((1:nvar)[infs],collapse=","), "; beta may be infinite. ")) } } names(coef) <- dimnames(x)[[2]] lp <- x %*% coef + offset - sum(coef *agfit$means) score <- as.double(exp(lp[sorted])) agres <- .C(Cagmart, as.integer(n), as.integer(0), sstart, sstop, sstat, score, rep(1.0, n), newstrat, resid=double(n)) resid <- double(n) resid[sorted] <- agres$resid names(resid) <- rownames coef[which.sing] <- NA list(coefficients = coef, var = var, loglik = agfit$loglik, score = agfit$sctest, iter = agfit$iter, linear.predictors = lp, residuals = resid, means = agfit$means, method= 'coxph') } survival/R/print.summary.survexp.R0000644000175100001440000000311012423461152017025 0ustar hornikusersprint.summary.survexp <- function(x, digits = max(options()$digits - 4, 3), ...) { savedig <- options(digits=digits) on.exit(options(savedig)) if (!is.null(cl<- x$call)) { cat("Call: ") dput(cl) cat("\n") } omit <- x$na.action if (length(omit)) cat(naprint(omit), "\n") mat <- cbind(x$time, x$n.risk, x$surv) if (is.matrix(x$n.risk)) cnames <- c("time", paste("nrisk", 1:ncol(x$n.risk), sep='')) else cnames <- c("time", "n.risk") if (is.matrix(x$surv)) ncurve <- ncol(x$surv) else ncurve <- 1 if (ncurve==1) { #only 1 curve cnames <- c(cnames, "survival") # if (!is.null(x$std.err)) { # if (is.null(x$lower)) { # mat <- cbind(mat, x$std.err) # cnames <- c(cnames, "std.err") # } # else { # mat <- cbind(mat, x$std.err, x$lower, x$upper) # cnames <- c(cnames, 'std.err', # paste("lower ", x$conf.int*100, "% CI", sep=''), # paste("upper ", x$conf.int*100, "% CI", sep='')) # } # } } else cnames <- c(cnames, paste("survival", seq(ncurve), sep='')) if (!is.matrix(mat)) mat <- matrix(mat, nrow=1) if (!is.null(mat)) { dimnames(mat) <- list(rep("", nrow(mat)), cnames) if (is.null(x$strata)) print(mat) else { #print it out one strata at a time strata <- x$strata for (i in levels(strata)) { who <- (strata==i) cat(" ", i, "\n") print(mat[who,]) cat("\n") } } } else stop("There are no observations to print.") invisible(x) } survival/R/summary.survexp.R0000644000175100001440000001232212205423426015677 0ustar hornikusers# # Almost identical to summary.survfit. The big differences # are no calls to the survmean function (irrelevant), and # there is no censoring, extend, or rmean argument. # And there is never an se, upper or lower component. # Because survexp objects do not contain n.event or n.censor, # subsetting is easier. summary.survexp <- function(object, times, scale=1, ...) { fit <- object if (!inherits(fit, 'survexp')) stop("Invalid data") # The fit$surv object is sometimes a vector and sometimes a matrix. # Make a copy of it that is always a matrix, to simplify the number of # cases for our subscripting work below. At the end of the routine # we'll turn it back into a vector if needed. Similar treatment is # given to the n.risk argument. surv <- as.matrix(fit$surv) n.risk <- as.matrix(fit$n.risk) if (is.null(fit$strata)) { nstrat <- 1 stemp <- rep(1, nrow(surv)) strata.names <- "" } else { nstrat <- length(fit$strata) stemp <- rep(1:nstrat, fit$strata) strata.names <- names(fit$strata) } # if (is.null(fit$std.err)) std.err <- NULL # else std.err <- fit$std.err * surv # if (!is.null(fit$lower)) { # lower <- as.matrix(fit$lower) # upper <- as.matrix(fit$upper) # } if (missing(times)) { times <- fit$time strata <- factor(stemp, labels=strata.names) } else { #this case is harder, since it involves "in between" points times <- sort(times) #just in case the user didn't # The basic idea is to process the curves one at a time, # adding the results for that curve onto a list, so the # survival surv[[1], surv[[2]], etc. # For the survival, stderr, and confidence limits it suffices # to create a single list 'indx1' containing a subscripting vector indx1 <- indx2 <- newtimes <- vector('list', nstrat) n <- length(stemp) for (i in 1:nstrat) { who <- (1:n)[stemp==i] # the rows of the object for this strata stime <- fit$time[who] # First, toss any printing times that are outside our range mintime <- min(stime, 0) ptimes <- times[times >= mintime] maxtime <- max(stime) ptimes <- ptimes[ptimes <= maxtime] newtimes[[i]] <- ptimes # If we tack a -1 onto the front of the vector of survival # times, then indx1 is the subscript for that vector # corresponding to the list of "ptimes". If the input # data had stime=c(10,20) and ptimes was c(5,10,15,20), # the result would be 1,2,2,3. # For n.risk we want a slightly different index: 2,2,3,3. # "In between" times point to the next higher index for n.risk, # but the next lower one for survival. (Survival drops at time t, # the n.risk immediately afterwords at time t+0: you were at # risk just before you die, but not a moment after). The # extra point needs to be added at the end. # ntime <- length(stime) #number of points temp1 <- approx(c(mintime-1, stime), 0:ntime, xout=ptimes, method='constant', f=0, rule=2)$y indx1[[i]] <- ifelse(temp1==0, 1, 1+ who[pmax(1,temp1)]) # Why not just "who[temp1]" instead of who[pmax(1,temp1)] in the # line just above? When temp1 has zeros, the first expression # gives a vector that is shorter than temp1, and the ifelse # doesn't work right due to mismatched lengths. # Compute the number at risk. If stime = 1,10, 20 and ptime=3,10, # 12, then temp1 = 2,2,3: the nrisk looking ahead # approx() doesn't work if stime is of length 1 if (ntime ==1) temp1 <- rep(1, length(ptimes)) else temp1 <- approx(stime, 1:ntime, xout=ptimes, method='constant', f=1, rule=2)$y indx2[[i]] <- ifelse(ptimes>max(stime), length(n.risk), who[temp1]) } # Now create the output list times <- unlist(newtimes) n.risk <- unlist(n.risk) indx1 <- unlist(indx1) surv <- (rbind(1.,surv))[indx1,,drop=FALSE] n.risk <- n.risk[unlist(indx2),, drop=FALSE] # if (!is.null(std.err)) std.err <- rbind(0.,std.err)[indx1,,drop=FALSE] # if (!is.null(fit$lower)) { # lower <- rbind(1.,lower)[indx1,,drop=FALSE] # upper <- rbind(1.,upper)[indx1,,drop=FALSE] # } if (!is.null(fit$strata)) { scount <- unlist(lapply(newtimes, length)) strata <- factor(rep(1:nstrat, scount), labels=names(fit$strata)) } } # # Final part of the routine: paste the material together into # the correct output structure # temp <- list(time=times/scale, n.risk=n.risk, surv=surv) if (ncol(surv)==1) { # Make surve & etc vectors again temp$surv <- drop(temp$surv) temp$n.risk <- drop(temp$n.risk) # if (!is.null(std.err)) temp$std.err <- drop(std.err) # if (!is.null(fit$lower)) { # temp$lower <- drop(lower) # temp$upper <- drop(upper) # } } # else { # if (!is.null(std.err)) temp$std.err <- std.err # if (!is.null(fit$lower)) { # temp$lower <- lower # temp$upper <- upper # } # } if (!is.null(fit$strata)) { temp$strata <- strata } temp$method <- fit$method temp$call <- fit$call if (!is.null(fit$na.action)) temp$na.action <- fit$na.action class(temp) <- "summary.survexp" temp } survival/R/survfitcoxph.fit.R0000644000175100001440000001720412533653611016025 0ustar hornikusers# Automatically generated from all.nw using noweb survfitcoxph.fit <- function(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat, id, y2, strata2, unlist=TRUE) { if (is.factor(strata)) ustrata <- levels(strata) else ustrata <- sort(unique(strata)) nstrata <- length(ustrata) survlist <- vector('list', nstrata) names(survlist) <- ustrata for (i in 1:nstrata) { indx <- which(strata== ustrata[i]) survlist[[i]] <- agsurv(y[indx,,drop=F], x[indx,,drop=F], wt[indx], risk[indx], survtype, vartype) } expand <- function(fit, x2, varmat, se.fit) { if (survtype==1) surv <- cumprod(fit$surv) else surv <- exp(-fit$cumhaz) if (is.matrix(x2) && nrow(x2) >1) { #more than 1 row in newdata fit$surv <- outer(surv, newrisk, '^') dimnames(fit$surv) <- list(NULL, row.names(x2)) if (se.fit) { varh <- matrix(0., nrow=length(fit$varhaz), ncol=nrow(x2)) for (i in 1:nrow(x2)) { dt <- outer(fit$cumhaz, x2[i,], '*') - fit$xbar varh[,i] <- (cumsum(fit$varhaz) + rowSums((dt %*% varmat)* dt))* newrisk[i]^2 } fit$std.err <- sqrt(varh) } fit$cumhaz <- outer(fit$cumhaz, newrisk, '*') } else { fit$surv <- surv^newrisk if (se.fit) { dt <- outer(fit$cumhaz, c(x2)) - fit$xbar varh <- (cumsum(fit$varhaz) + rowSums((dt %*% varmat)* dt)) * newrisk^2 fit$std.err <- sqrt(varh) } fit$cumhaz <- fit$cumhaz * newrisk } fit } if (missing(id) || is.null(id)) result <- lapply(survlist, expand, x2, varmat, se.fit) else { onecurve <- function(slist, x2, y2, strata2, newrisk, se.fit) { ntarget <- nrow(x2) #number of different time intervals surv <- vector('list', ntarget) n.event <- n.risk <- n.censor <- varh1 <- varh2 <- time <- surv hazard <- vector('list', ntarget) stemp <- as.integer(strata2) timeforward <- 0 for (i in 1:ntarget) { slist <- survlist[[stemp[i]]] indx <- which(slist$time > y2[i,1] & slist$time <= y2[i,2]) if (length(indx)==0) { timeforward <- timeforward + y2[i,2] - y2[i,1] # No deaths or censors in user interval. Possible # user error, but not uncommon at the tail of the curve. } else { time[[i]] <- diff(c(y2[i,1], slist$time[indx])) #time increments time[[i]][1] <- time[[i]][1] + timeforward timeforward <- y2[i,2] - max(slist$time[indx]) hazard[[i]] <- slist$hazard[indx]*newrisk[i] if (survtype==1) surv[[i]] <- slist$surv[indx]^newrisk[i] n.event[[i]] <- slist$n.event[indx] n.risk[[i]] <- slist$n.risk[indx] n.censor[[i]]<- slist$n.censor[indx] dt <- outer(slist$cumhaz[indx], x2[i,]) - slist$xbar[indx,,drop=F] varh1[[i]] <- slist$varhaz[indx] *newrisk[i]^2 varh2[[i]] <- rowSums((dt %*% varmat)* dt) * newrisk[i]^2 } } cumhaz <- cumsum(unlist(hazard)) if (survtype==1) surv <- cumprod(unlist(surv)) #increments (K-M) else surv <- exp(-cumhaz) if (se.fit) list(n=as.vector(table(strata)[stemp[1]]), time=cumsum(unlist(time)), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor= unlist(n.censor), surv = surv, cumhaz= cumhaz, std.err = sqrt(cumsum(unlist(varh1)) + unlist(varh2))) else list(n=as.vector(table(strata)[stemp[1]]), time=cumsum(unlist(time)), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor= unlist(n.censor), surv = surv, cumhaz= cumhaz) } if (all(id ==id[1])) { result <- list(onecurve(survlist, x2, y2, strata2, newrisk, se.fit)) } else { uid <- unique(id) result <- vector('list', length=length(uid)) for (i in 1:length(uid)) { indx <- which(id==uid[i]) result[[i]] <- onecurve(survlist, x2[indx,,drop=FALSE], y2[indx,,drop=FALSE], strata2[indx], newrisk[indx], se.fit) } names(result) <- uid } } if (unlist) { if (length(result)==1) { # the no strata case if (se.fit) result[[1]][c("n", "time", "n.risk", "n.event", "n.censor", "surv", "cumhaz", "std.err")] else result[[1]][c("n", "time", "n.risk", "n.event", "n.censor", "surv", "cumhaz")] } else { temp <-list(n = unlist(lapply(result, function(x) x$n), use.names=FALSE), time= unlist(lapply(result, function(x) x$time), use.names=FALSE), n.risk= unlist(lapply(result, function(x) x$n.risk), use.names=FALSE), n.event= unlist(lapply(result, function(x) x$n.event), use.names=FALSE), n.censor=unlist(lapply(result, function(x) x$n.censor), use.names=FALSE), strata = sapply(result, function(x) length(x$time))) names(temp$strata) <- names(result) if ((missing(id) || is.null(id)) && nrow(x2)>1) { temp$surv <- t(matrix(unlist(lapply(result, function(x) t(x$surv)), use.names=FALSE), nrow= nrow(x2))) dimnames(temp$surv) <- list(NULL, row.names(x2)) temp$cumhaz <- t(matrix(unlist(lapply(result, function(x) t(x$cumhaz)), use.names=FALSE), nrow= nrow(x2))) if (se.fit) temp$std.err <- t(matrix(unlist(lapply(result, function(x) t(x$std.err)), use.names=FALSE), nrow= nrow(x2))) } else { temp$surv <- unlist(lapply(result, function(x) x$surv), use.names=FALSE) temp$cumhaz <- unlist(lapply(result, function(x) x$cumhaz), use.names=FALSE) if (se.fit) temp$std.err <- unlist(lapply(result, function(x) x$std.err), use.names=FALSE) } temp } } else { names(result) <- ustrata result } } survival/R/frailty.S0000644000175100001440000000075211732700061014143 0ustar hornikusers# $Id: frailty.S 11166 2008-11-24 22:10:34Z therneau $ # # Parent function for frailty, calls the actuall working functions # frailty <- function(x, distribution = 'gamma', ...) { dlist <- c("gamma", "gaussian", "t") i <- pmatch(distribution, dlist) if (!is.na(i)) distribution <- dlist[i] temp <- paste("frailty", distribution, sep='.') if (!exists(temp)) stop(paste("Function '", temp, "' not found", sep="")) (get(temp))(x, ...) } survival/R/residuals.survreg.S0000644000175100001440000001744712533653604016203 0ustar hornikusers# Automatically generated from all.nw using noweb # $Id$ # # Residuals for survreg objects residuals.survreg <- function(object, type=c('response', 'deviance', 'dfbeta', 'dfbetas', 'working', 'ldcase', 'ldresp', 'ldshape', 'matrix'), rsigma =TRUE, collapse=FALSE, weighted=FALSE, ...) { type <-match.arg(type) n <- length(object$linear.predictors) Terms <- object$terms if(!inherits(Terms, "terms")) stop("invalid terms component of object") # If the variance wasn't estimated then it has no error if (nrow(object$var) == length(object$coefficients)) rsigma <- FALSE # If there was a cluster directive in the model statment then remove # it. It does not correspond to a coefficient, and would just confuse # things later in the code. cluster <- untangle.specials(Terms,"cluster")$terms if (length(cluster) >0 ) Terms <- Terms[-cluster] strata <- attr(Terms, 'specials')$strata coef <- object$coefficients intercept <- attr(Terms, "intercept") response <- attr(Terms, "response") weights <- object$weights if (is.null(weights)) weighted <- FALSE if (is.character(object$dist)) dd <- survreg.distributions[[object$dist]] else dd <- object$dist if (is.null(dd$itrans)) { itrans <- dtrans <-function(x)x } else { itrans <- dd$itrans dtrans <- dd$dtrans } if (!is.null(dd$dist)) dd <- survreg.distributions[[dd$dist]] deviance <- dd$deviance dens <- dd$density if (is.null(object$naive.var)) vv <- object$var else vv <- object$naive.var need.x <- is.na(match(type, c('response', 'deviance', 'working'))) if (is.null(object$y) || !is.null(strata) || (need.x & is.null(object[['x']]))) mf <- model.frame(object) y <- object$y if (is.null(y)) { y <- model.extract(mf, 'response') if (!is.null(dd$trans)) { tranfun <- dd$trans exactsurv <- y[,ncol(y)] ==1 if (any(exactsurv)) logcorrect <-sum(log(dd$dtrans(y[exactsurv,1]))) if (type=='interval') { if (any(y[,3]==3)) y <- cbind(tranfun(y[,1:2]), y[,3]) else y <- cbind(tranfun(y[,1]), y[,3]) } else if (type=='left') y <- cbind(tranfun(y[,1]), 2-y[,2]) else y <- cbind(tranfun(y[,1]), y[,2]) } else { if (type=='left') y[,2] <- 2- y[,2] else if (type=='interval' && all(y[,3]<3)) y <- y[,c(1,3)] } } if (!is.null(strata)) { temp <- untangle.specials(Terms, 'strata', 1) Terms2 <- Terms[-temp$terms] if (length(temp$vars)==1) strata.keep <- mf[[temp$vars]] else strata.keep <- strata(mf[,temp$vars], shortlabel=TRUE) strata <- as.numeric(strata.keep) nstrata <- max(strata) sigma <- object$scale[strata] } else { Terms2 <- Terms nstrata <- 1 sigma <- object$scale } if (need.x) { x <- object[['x']] #don't grab xlevels component if (is.null(x)) x <- model.matrix(Terms2, mf, contrasts.arg=object$contrasts) } if (type=='response') { yhat0 <- deviance(y, sigma, object$parms) rr <- itrans(yhat0$center) - itrans(object$linear.predictor) } else { status <- y[,ncol(y)] eta <- object$linear.predictors z <- (y[,1] - eta)/sigma dmat <- dens(z, object$parms) dtemp<- dmat[,3] * dmat[,4] #f' if (any(status==3)) { z2 <- (y[,2] - eta)/sigma dmat2 <- dens(z2, object$parms) } else { dmat2 <- dmat #dummy values z2 <- 0 } tdenom <- ((status==0) * dmat[,2]) + #right censored ((status==1) * 1 ) + #exact ((status==2) * dmat[,1]) + #left ((status==3) * ifelse(z>0, dmat[,2]-dmat2[,2], dmat2[,1] - dmat[,1])) #interval g <- log(ifelse(status==1, dmat[,3]/sigma, tdenom)) #loglik tdenom <- 1/tdenom dg <- -(tdenom/sigma) *(((status==0) * (0-dmat[,3])) + #dg/ eta ((status==1) * dmat[,4]) + ((status==2) * dmat[,3]) + ((status==3) * (dmat2[,3]- dmat[,3]))) ddg <- (tdenom/sigma^2) *(((status==0) * (0- dtemp)) + #ddg/eta^2 ((status==1) * dmat[,5]) + ((status==2) * dtemp) + ((status==3) * (dmat2[,3]*dmat2[,4] - dtemp))) ds <- ifelse(status<3, dg * sigma * z, tdenom*(z2*dmat2[,3] - z*dmat[,3])) dds <- ifelse(status<3, ddg* (sigma*z)^2, tdenom*(z2*z2*dmat2[,3]*dmat2[,4] - z * z*dmat[,3] * dmat[,4])) dsg <- ifelse(status<3, ddg* sigma*z, tdenom *(z2*dmat2[,3]*dmat2[,4] - z*dtemp)) deriv <- cbind(g, dg, ddg=ddg- dg^2, ds = ifelse(status==1, ds-1, ds), dds=dds - ds*(1+ds), dsg=dsg - dg*(1+ds)) if (type=='deviance') { yhat0 <- deviance(y, sigma, object$parms) rr <- (-1)*deriv[,2]/deriv[,3] #working residuals rr <- sign(rr)* sqrt(2*(yhat0$loglik - deriv[,1])) } else if (type=='working') rr <- (-1)*deriv[,2]/deriv[,3] else if (type=='dfbeta' || type== 'dfbetas' || type=='ldcase') { score <- deriv[,2] * x # score residuals if (rsigma) { if (nstrata > 1) { d4 <- matrix(0., nrow=n, ncol=nstrata) d4[cbind(1:n, strata)] <- deriv[,4] score <- cbind(score, d4) } else score <- cbind(score, deriv[,4]) } rr <- score %*% vv if (type=='dfbetas') rr <- rr %*% diag(1/sqrt(diag(vv))) if (type=='ldcase') rr<- rowSums(rr*score) } else if (type=='ldresp') { rscore <- deriv[,3] * (x * sigma) if (rsigma) { if (nstrata >1) { d6 <- matrix(0., nrow=n, ncol=nstrata) d6[cbind(1:n, strata)] <- deriv[,6]*sigma rscore <- cbind(rscore, d6) } else rscore <- cbind(rscore, deriv[,6] * sigma) } temp <- rscore %*% vv rr <- rowSums(rscore * temp) } else if (type=='ldshape') { sscore <- deriv[,6] *x if (rsigma) { if (nstrata >1) { d5 <- matrix(0., nrow=n, ncol=nstrata) d5[cbind(1:n, strata)] <- deriv[,5] sscore <- cbind(sscore, d5) } else sscore <- cbind(sscore, deriv[,5]) } temp <- sscore %*% vv rr <- rowSums(sscore * temp) } else { #type = matrix rr <- deriv } } #case weights if (weighted) rr <- rr * weights #Expand out the missing values in the result if (!is.null(object$na.action)) { rr <- naresid(object$na.action, rr) if (is.matrix(rr)) n <- nrow(rr) else n <- length(rr) } # Collapse if desired if (!missing(collapse)) { if (length(collapse) !=n) stop("Wrong length for 'collapse'") rr <- drop(rowsum(rr, collapse)) } rr } survival/R/format.Surv.S0000644000175100001440000000113411732700061014712 0ustar hornikusers# $Id: format.Surv.S 11273 2009-03-20 15:02:17Z tlumley $ # format.Surv <- function(x, ...) format(as.character.Surv(x), ...) # The function to "make something suitable for inclusion in a data frame" # was "as.data.frame.x" in versions <5, now it is "data.frameAux.x", # so here we have a version specific definition. # This is needed for both S-plus and R if (!is.R()) { if (version$major >= 5) { data.frameAux.Surv <- function(x, ...) data.frameAux.AsIs(x, ...) } else as.data.frame.Surv <- as.data.frame.model.matrix } else { as.data.frame.Surv <- as.data.frame.model.matrix } survival/R/survfit.formula.S0000644000175100001440000000734012055765142015651 0ustar hornikuserssurvfit.formula <- function(formula, data, weights, subset, na.action, etype, id, istate, ...) { Call <- match.call() Call[[1]] <- as.name('survfit') #make nicer printout for the user # create a copy of the call that has only the arguments we want, # and use it to call model.frame() mfnames <- c('formula', 'data', 'weights', 'subset','na.action', 'istate', 'id', "etype") #legal args for model.frame temp <- Call[c(1, match(mfnames, names(Call), nomatch=0))] temp[[1]] <- as.name("model.frame") if (is.R()) m <- eval.parent(temp) else m <- eval(temp, sys.parent()) Terms <- terms(formula, c("strata", "cluster")) ord <- attr(Terms, 'order') if (length(ord) & any(ord !=1)) stop("Interaction terms are not valid for this function") n <- nrow(m) Y <- model.extract(m, 'response') if (!is.Surv(Y)) stop("Response must be a survival object") casewt <- model.extract(m, "weights") if (is.null(casewt)) casewt <- rep(1,n) if (!is.null(attr(Terms, 'offset'))) warning("Offset term ignored") id <- model.extract(m, 'id') istate <- model.extract(m,"istate") temp <- untangle.specials(Terms, "cluster") if (length(temp$vars)>0) { if (length(temp$vars) > 1) stop("can not have two cluster terms") if (!is.null(id)) stop("can not have both a cluster term and an id variable") id <- m[[temp$vars]] Terms <- Terms[-temp$terms] } ll <- attr(Terms, 'term.labels') if (length(ll) == 0) X <- factor(rep(1,n)) # ~1 on the right else X <- strata(m[ll]) if (!is.Surv(Y)) stop("y must be a Surv object") # Backwards support for the now-depreciated etype argument etype <- model.extract(m, "etype") if (!is.null(etype)) { if (attr(Y, "type") == "mcounting" || attr(Y, "type") == "mright") stop("cannot use both the etype argument and mstate survival type") if (length(istate)) stop("cannot use both the etype and istate arguments") status <- Y[,ncol(Y)] etype <- as.factor(etype) temp <- table(etype, status==0) if (all(rowSums(temp==0) ==1)) { # The user had a unique level of etype for the censors newlev <- levels(etype)[order(-temp[,2])] #censors first } else newlev <- c(" ", levels(etype)[temp[,1] >0]) status <- factor(ifelse(status==0,0, as.numeric(etype)), labels=newlev) if (attr(Y, 'type') == "right") Y <- Surv(Y[,1], status, type="mstate") else if (attr(Y, "type") == "counting") Y <- Surv(Y[,1], Y[,2], status, type="mstate") else stop("etype argument incompatable with survival type") } # At one point there were lines here to round the survival # times to a certain number of digits. This approach worked # almost all the time, but only almost. The better logic is # now in the individual compuation routines if (attr(Y, 'type') == 'left' || attr(Y, 'type') == 'interval') temp <- survfitTurnbull(X, Y, casewt, ...) else if (attr(Y, 'type') == "right" || attr(Y, 'type')== "counting") temp <- survfitKM(X, Y, casewt, ...) else if (attr(Y, 'type') == "mright" || attr(Y, "type")== "mcounting") temp <- survfitCI(X, Y, weights=casewt, id=id, istate=istate, ...) else { # This should never happen stop("unrecognized survival type") } if (is.null(temp$states)) class(temp) <- 'survfit' else class(temp) <- c("survfitms", "survfit") if (!is.null(attr(m, 'na.action'))) temp$na.action <- attr(m, 'na.action') temp$call <- Call temp } survival/R/logLik.coxph.R0000644000175100001440000000117712475407540015046 0ustar hornikusers# # The AIC function depends on a logLik method # logLik.coxph <- function(object, ...) { out <- object$loglik[2] if (!is.null(object$df)) attr(out, "df") <- object$df[2] else attr(out, 'df') <- sum(!is.na(coefficients(object))) attr(out, "nobs") <- object$nevent class(out) <- 'logLik' out } logLik.survreg <- function(object, ...) { out <- object$loglik[2] dd <- diag(object$var) if (!is.null(object$df)) attr(out, "df") <- sum(object$df) else attr(out, 'df') <- sum(!is.na(dd) & dd > 0) # attr(out, "nobs") <- sum(object$df) + object$df.residual class(out) <- 'logLik' out } survival/R/lines.aareg.S0000644000175100001440000000440011732700061014653 0ustar hornikusers# $Id: lines.aareg.S 11166 2008-11-24 22:10:34Z therneau $ lines.aareg <- function(x, se=FALSE, maxtime, type='s', ...) { if (!inherits(x, 'aareg')) stop ("Must be an aareg object") if (missing(maxtime)) keep <- 1:length(x$time) else keep <- 1:sum(x$time <= maxtime) if (is.matrix(x$coefficient) && ncol(x$coefficient)>1) { yy <- apply(x$coefficient[keep,], 2,cumsum) yy <- rbind(0,yy) # make the plot start at 0,0 if (se) { if (!is.null(x$dfbeta)) { # There was a cluster term, so use the robust variance # dfbeta will be of dimension (n, nvar, n-unique-times) # The first variance increment is apply(dfbeta[,,1]^2,2,sum) # second is apply(dfbeta[,,2]^2,2,sum) # ... , apply(dfbeta[,,ndeath]..... # By being sneaky, it can be done quickly dd <- dim(x$dfbeta) keep2 <- 1:length(unique(x$time[keep])) temp <- matrix(x$dfbeta[,,keep2], nrow=dd[1]) se.increment <- matrix(apply(temp^2, 2, sum), nrow=dd[2]) se.yy <- sqrt(apply(t(se.increment), 2, cumsum)) } else se.yy <- sqrt(apply(x$coefficient[keep,]^2, 2,cumsum)) se.yy <- rbind(0, se.yy) } ncurve <- ncol(yy) } else { # this is the branch most often called, when someone has done # plot(fit[3]), so that only 1 coefficient remains yy <- cumsum(c(0, x$coefficient[keep])) if (se) { if (!is.null(x$dfbeta)) { dd <- dim(x$dfbeta) keep2 <- 1:length(unique(x$time[keep])) temp <- matrix(x$dfbeta[,,keep2], nrow=dd[1]) se.yy <- sqrt(cumsum(c(0, apply(temp^2, 2, sum)))) } else se.yy <- sqrt(cumsum(c(0, x$coefficient[keep]^2))) } ncurve <- 1 } xx <- c(0, x$time[keep]) # There may be multiplicities in x$times. Only plot the last of # each of them indx <- 1 + length(xx) - rev(match(unique(rev(xx)), rev(xx))) xx <- xx[indx] yy <- as.matrix(yy)[indx,] if (se) { if (is.null(x$dfbeta)) se.yy<- as.matrix(se.yy)[indx,] yy <- cbind(yy, yy + 1.96*se.yy, yy - 1.96*se.yy) if (ncurve >1) { for (i in 1:ncurve) { j <- c(i, i+ncurve, i+2*ncurve) matlines(xx, yy[,j], type=type, ..., col=1, lty=c(1,2,2)) } } else matlines(xx, yy, type=type, ..., col=1, lty=c(1,2,2),) } else { matlines(xx, yy, type=type, ..., xlab='Time') } } survival/R/print.summary.survfitms.S0000644000175100001440000000411512423461355017367 0ustar hornikusersprint.summary.survfitms <- function(x, digits = max(options()$digits - 4, 3), ...) { savedig <- options(digits=digits) on.exit(options(savedig)) if (!is.null(cl<- x$call)) { cat("Call: ") dput(cl) cat("\n") } omit <- x$na.action if (length(omit)) cat(naprint(omit), "\n") if (x$type == 'right' || is.null(x$n.enter)) { mat <- cbind(x$time, x$n.risk, x$n.event, x$prev) cnames <- c("time", "n.risk", "n.event") } else if (x$type == 'counting') { mat <- cbind(x$time, x$n.risk, x$n.event, x$n.enter, x$n.censor, x$prev) cnames <- c("time", "n.risk", "n.event", "entered", "censored") } if (is.matrix(x$prev)) ncurve <- ncol(x$prev) else ncurve <- 1 if (ncurve==1) { #only 1 curve cnames <- c(cnames, "prevalence") if (!is.null(x$std.err)) { if (is.null(x$lower)) { mat <- cbind(mat, x$std.err) cnames <- c(cnames, "std.err") } else { mat <- cbind(mat, x$std.err, x$lower, x$upper) cnames <- c(cnames, 'std.err', paste("lower ", x$conf.int*100, "% CI", sep=''), paste("upper ", x$conf.int*100, "% CI", sep='')) } } } else cnames <- c(cnames, paste("prevalence", seq(ncurve), sep='')) if (!is.null(x$start.time)) { mat.keep <- mat[,1] >= x$start.time mat <- mat[mat.keep,,drop=FALSE] if (is.null(dim(mat))) stop(paste("No information available using start.time =", x$start.time, ".")) } if (!is.matrix(mat)) mat <- matrix(mat, nrow=1) if (!is.null(mat)) { dimnames(mat) <- list(rep("", nrow(mat)), cnames) if (is.null(x$strata)) print(mat) else { #print it out one strata at a time strata <- x$strata if (!is.null(x$start.time)) strata <- strata[mat.keep] for (i in levels(strata)) { who <- (strata==i) cat(" ", i, "\n") print(mat[who,]) cat("\n") } } } else stop("There are no events to print. Please use the option ", "censored=TRUE with the summary function to see the censored ", "observations.") invisible(x) } survival/R/aareg.S0000644000175100001440000003440112153624142013551 0ustar hornikusers# Aalen's additive regression model # Originally, this tried to call coxph with certain options. # But we found the passing ... to a model method just doesn't work (for # optional things like weights). So the first portion of this is # essentially coxph, to set up for coxph.detail. # For distribution, the "variance" test is omitted. Not all aspects are # yet supported by the downstream printing. # aareg <- function(formula, data, weights, subset, na.action, qrtol=1e-7, nmin, dfbeta=FALSE, taper=1, test = c('aalen', 'variance', 'nrisk'), model=FALSE, x=FALSE, y=FALSE) { call <- match.call() m <- match.call(expand.dots=FALSE) temp <- c("", "formula", "data", "weights", "subset", "na.action") m <- m[ match(temp, names(m), nomatch=0)] special <- c("strata", "cluster") Terms <- if(missing(data)) terms(formula, special) else terms(formula, special, data=data) m$formula <- Terms m[[1]] <- as.name("model.frame") m <- eval(m, sys.parent()) test <- match.arg(test) #check for legal argument # Now grab the terms that we need Y <- model.extract(m, "response") if (!inherits(Y, "Surv")) stop("Response must be a survival object") offset<- attr(Terms, "offset") tt <- length(offset) offset <- if(tt == 0) rep(0, nrow(Y)) else if(tt == 1) m[[offset]] else { #multiple offset terms! add them ff <- m[[offset[1]]] for(i in 2:tt) ff <- ff + m[[offset[i]]] ff } # Create an X matrix, to feed to the coxdetail routine. attr(Terms,"intercept")<- 1 # force no intercept strats <- attr(Terms, "specials")$strata cluster<- attr(Terms, "specials")$cluster dropx <- NULL if (length(cluster)) { dfbeta <- TRUE tempc <- untangle.specials(Terms, 'cluster', 1:10) ord <- attr(Terms, 'order')[tempc$terms] if (any(ord>1)) stop ("Cluster can not be used in an interaction") cluster <- strata(m[,tempc$vars], shortlabel=TRUE) #allow multiples cluster <- as.numeric(cluster) #labels don't matter, and processing # is a bit faster without them dropx <- tempc$terms } else cluster <- 1:nrow(m) # Adding strata, when there is a coefficent per death, is identical # to doing a totally separate fit per group. # Using "factor(group)" to get multiple baselines is likely what the # user wants. However, because we have not processed the strata # statement (taken it out of X, and created the 'newstrat' of coxph) # it will act just like a factor. # I've changed my mind multiple times on commenting out the line below. # Computationally identical to factor() -- is an error message or not # an error message the greater source of confusion to a user? if (length(strats)) { stop("Strata terms not allowed") } if (length(dropx)) X <- model.matrix(Terms[-dropx], m)[,-1,drop=FALSE] else X <- model.matrix(Terms, m)[,-1,drop=FALSE] nvar <- ncol(X) nused<- nrow(X) weights <- model.extract(m, 'weights') if (length(weights)==0) weights <- rep(1.0, nused) type <- attr(Y, "type") if (type!='right' && type!='counting') stop(paste("Aalen model doesn't support \"", type, "\" survival data", sep='')) # Get the peices that I need from the coxdetail routine # 1. It expects a "counting process" type of Y if (ncol(Y)==2) { mintime <- min(Y[,1]) if (mintime < 0) Y <- cbind( 2*mintime -1, Y) else Y <- cbind(-1,Y) } # Because there are no strata, the number of unique death times is the # number that will be in the output structures times <- as.vector(Y[,2]) # toss the labels away status<- as.vector(Y[,3]) ndeath <- length(unique(times[status==1])) # Sort everything ord <- order(times, -status) times <- times[ord] status <- status[ord] weights <- weights[ord] if (x) saveX <- X X <- X[ord,,drop=FALSE] storage.mode(Y) <- 'double' ff <- .C(Ccoxdetail, as.integer(nused), as.integer(nvar), ndeath= as.integer(ndeath), y = Y[ord,], as.double(X), index = as.integer(rep(0,nused)), event2 = rep(1.0, nused), weights = as.double(weights), means= c(0., double(ndeath*nvar-1)), u = double(ndeath*nvar), i = double(ndeath*nvar*nvar), rmat = integer(ndeath*nused), nrisk2 = double(ndeath), double(nvar*(3 + 2*nvar)) ) # riskmat is an nused by ndeath 0/1 matrix showing who is present riskmat <- matrix(ff$rmat, nused, ndeath) # Note that imat, as returned by coxdetail, is Var(X) * nevents. dt <- list(means= (matrix(ff$means,ndeath, nvar)), var = aareg.taper(taper, array(ff$i, c(nvar, nvar, ndeath)), ff$event2[1:ndeath]), time = times[ff$index[1:ndeath]], nrisk= ff$nrisk2, #weighted # at risk nevent=ff$event2[1:ndeath]) #weighted number of events # Set the number of deaths that will be used in the analysis # This may be smaller than the curren "ndeath", due to small nrisk # The number of times may even be smaller, if imat is singular at that # time point. if (missing(nmin)) nmin <- 3*nvar if (nvar==1) ndeath <- sum(dt$nrisk>= nmin & c(dt$var)>0) else { ndeath <- sum(dt$nrisk >= nmin) if (ndeath >0) { while (1) { #we expect very few iterations of this loop qri <- qr(dt$var[,,ndeath], tol=qrtol) if (qri$rank >= nvar) break #not singular ndeath <- ndeath -1 } } } if (ndeath<=1) stop("The threshold 'nmin' is too high, no model can be fit") # This matches the death times in the data set to the # sorted list of unique death times. "0" = not a death index <- match(times, dt$time[1:ndeath], nomatch=0) * status deaths <- (status==1 & index >0) dindex <- index[deaths] #for each death, a pointer into dt objects nevent <- length(dindex) #total number of events (ndeath = #unique times) if (length(cluster)) ncluster <- length(unique(cluster)) else ncluster <- nused if (dfbeta) { dmat <- array(0.0, dim=c(ncluster, nvar+1, ndeath)) # the resid marix has a row for each death, and nused cols # each row has a "1" in it at the position of the death # the yhat part is subtracted later resid <- rep(0., nevent*nused) resid[nevent*((1:nused)[deaths]-1) + 1:nevent] <- 1.0 resid <- matrix(resid, ncol=nused) } # Coefficient is the step in Aalen's plots # If we keep one row of "coefficent" per death, then Aalen's # variance is coef *% t(coef), treating coef as a col vector. # If we kept one row per death, then the ndeath nvar by nvar variance # matrices would need to be kept too. So keep 1 row per event. # Things like plot will end up accumlating. # There is no such cheat for dfbeta: it is kept as "# unique deaths" # p by p matrices. # if (nvar==1) { # special case of only 1 covariate means <- dt$means[dindex] nrisk <- dt$nrisk[dindex] xx <- (X[deaths] - means) * weights[deaths] v.inverse <- 1/dt$var[dindex] #for all time points twt <- nrisk* 1/cbind(1+ means^2*v.inverse, v.inverse) coefficient <- v.inverse * xx / nrisk # Note that ybar is always w_i/nrisk, since we are doing the # regressions one event at a time. b0 <- weights[deaths]/nrisk - means*coefficient if (dfbeta) { # We first create the nused * #events matrix, and then # collapse it to be ncluster by n-unique-death-times xx <- c(X) * riskmat[,dindex] # X repeated in each col, if at risk predicted <- coefficient * t(xx) + b0*t(riskmat[,dindex]) resid <- resid - predicted #nused cols, nvevent rows temp1 <- (resid * (t(xx) -means)/(nrisk*dt$var[dindex])) * rep(weights, rep(nevent, nused)) # temp1[i,j] is the change in alpha at time i for subject j # the "intercept dfbeta" is resid*wt/sum(wt) - xbar*temp1 temp0 <- resid * outer(1/nrisk, weights) - temp1 * means # get the matrix, nused by 2, which is the influence of each # subject on the test statistic. # This is a bit easier before collapsing if (test=='nrisk') { test.dfbeta <- cbind(apply(temp0*nrisk, 2, sum), apply(temp1*nrisk, 2, sum)) } else { test.dfbeta <- cbind(apply(temp0*twt[,1], 2, sum), apply(temp1*twt[,2], 2, sum)) } # Now collapse dfbeta, first on the deaths, and then on the cluster if (nevent > ndeath) { temp1 <- rowsum(temp1, times[deaths], reorder=FALSE) temp0 <- rowsum(temp0, times[deaths], reorder=FALSE) } dmat[,1,] <- rowsum(t(temp0), cluster[ord], reorder= FALSE) dmat[,2,] <- rowsum(t(temp1), cluster[ord], reorder =FALSE) } # Compute the test statistic, including the intercept term # (Much of the code above was a litte easier to write without # the intercept term in thec coef matrix, that below is easier # with it in). coefficient <- cbind(b0,coefficient) if (test=='nrisk') { temp <- coefficient*nrisk test.statistic <- apply(temp,2,sum) test.var <- matrix(0.,2,2) diag(test.var) <- apply(temp^2, 2, sum) test.var[1,2] <- test.var[2,1] <- sum(temp[,1]*temp[,2]) } else { # full V^{-1} and diag(V){-1} variance (Aalen) are the same temp <- coefficient* twt test.statistic <- apply(temp,2,sum) test.var <- matrix(0.,2,2) diag(test.var) <- apply(temp^2, 2, sum) test.var[1,2] <- test.var[2,1] <- sum(temp[,1]*temp[,2]) } } else { # 2 or more covariates coefficient <- matrix(0,nevent, nvar) twt <- matrix(0, nevent, nvar+1) means <- dt$means[dindex,] # vector of means, at each deatj nrisk <- dt$nrisk[dindex] dindex2 <- (1:nused)[deaths] # row number of each death ybar <- weights[deaths]/nrisk test.var <- matrix(0.0, nvar, nvar) if (dfbeta) test.dfbeta <- matrix(0., nused, nvar+1) for (i in 1:nevent) { who <- riskmat[,dindex[i]] # 0/1 vector of who is at risk wt <- weights* who xx <- who* (X- rep(means[i,], rep(nused, nvar))) # (X-Xbar) # solve, and check for singularity # Note that the increment to imat, as returned by # the coxph.detail function, is Var(X) * #events # and qri is intended to be the qr of V-inverse if (i==1 || dindex[i] != dindex[i-1]) { #don't redo qr for ties qri <- qr(dt$var[,,dindex[i]], tol=qrtol) vmat <- qr.coef(qri, diag(nvar)) twt[i,] <- nrisk[i] /c(1+ means[i,] %*% vmat %*% means[i,], diag(vmat)) } else twt[i,] <- twt[i-1,] j <- dindex2[i] coefficient[i,] <-qr.coef(qri, wt[j]*xx[j,]) / nrisk[i] if (test=='variance') { temp <- wt[j]*xx[j,] test.var <- test.var + outer(temp,temp) } if (dfbeta) { resid[i, ] <- resid[i,]- c(ybar[i] + xx %*% c(coefficient[i,])) temp1 <- t(qr.coef(qri, t(resid[i,]* wt *xx)))/ nrisk[i] temp0 <- resid[i,]*wt/nrisk[i] - temp1%*% means[i,] if (test=='aalen') test.dfbeta <- test.dfbeta + cbind(temp0, temp1) %*% diag(twt[i,]) else if (test=='nrisk') test.dfbeta <- test.dfbeta + cbind(temp0, temp1)* nrisk[i] else { test.dfbeta[,-1] <- test.dfbeta[,-1] + resid[i,]* wt *xx # There really isn't a definition for what weight to # put on the intercept in the "variance" weighting # (and who really cares about "testing the intercept" # anyway). So use the twt one test.dfbeta[,1] <- test.dfbeta[,1] + temp0*twt[i,1] } dmat[,-1,dindex[i]] <- dmat[,-1, dindex[i]] + rowsum(temp1, cluster[ord], reorder=FALSE) dmat[,1, dindex[i]] <- dmat[,1,dindex[i]] + rowsum(temp0, cluster[ord], reorder=FALSE) } } temp <- apply(means*coefficient, 1, sum) # xbar * coef at time t b0 <- weights[deaths]/nrisk - temp coefficient <- cbind(b0,coefficient) # Note - the intercept is a part of the test statistic, even # though it will always be ignored in the overall chisquare test if (test=='aalen') { temp <- twt* coefficient test.statistic <- colSums(temp) test.var <- t(temp) %*% temp } else if (test=='nrisk') { temp <- coefficient * nrisk test.statistic <- colSums(temp) test.var <- t(temp) %*% temp } else { xx <- weights[deaths]*(X[deaths,] - means[dindex,]) test.statistic <- apply(xx, 2, sum) } } if (dfbeta) { # The model variance is sum( term[i]^2), i ranging over times, # and each term an n by p matrix (one row per person) # The dfbeta one is essentially [sum(term[i])]^2 # the test.dfbeta matrix contains this sum over death times temp <- rowsum(test.dfbeta, cluster, reorder=FALSE) test.var2 <- t(temp) %*% temp } dimnames(coefficient) <- list(times[deaths], c("Intercept", dimnames(X)[[2]])) names(test.statistic) <- c("Intercept", dimnames(X)[[2]]) dimnames(twt) <- NULL ans <- list(n= c(nused, ndeath, length(dt$time)), times=times[deaths], nrisk=dt$nrisk[dindex], coefficient=coefficient, test.statistic=test.statistic, test.var=test.var, test=test, tweight = twt, call=call) if (dfbeta) { ans$dfbeta <- dmat ans$test.var2 <- test.var2 } if (any(weights!=1)) ans$weights <- weights # if (ncluster < nused) ans$cluster <- as.numeric(cluster) na.action <- attr(m, "na.action") if (length(na.action)) ans$na.action <- na.action if (model) ans$model <- m else { if (x) ans$x <- saveX if (y) ans$y <- Y } if (is.R()) class(ans) <- 'aareg' else oldClass(ans) <- 'aareg' ans } "[.aareg" <- function(x, ..., drop=FALSE) { if (!inherits(x, 'aareg')) stop ("Must be an aareg object") i <- ..1 if (is.matrix(x$coefficient)) { x$coefficient <- x$coefficient[,i, drop=drop] x$tweight <- x$tweight[,i,drop=drop] } else stop("Subsripting impossible, coefficient component not a matrix") if (!is.null(x$dfbeta)){ x$dfbeta <- x$dfbeta[,i,,drop=drop] x$test.var2 <- x$test.var2[i,i,drop=drop] } x$test.statistic <- x$test.statistic[i, drop=drop] x$test.var <- x$test.var[i,i,drop=drop] x } survival/R/cch.R0000644000175100001440000003405612423461726013243 0ustar hornikusers### Suite of programs for case-cohort analysis ### Main program cch <- function(formula, data=sys.parent(), subcoh, id, stratum=NULL, cohort.size, method=c("Prentice", "SelfPrentice", "LinYing","I.Borgan","II.Borgan"), robust=FALSE){ call <- match.call() if (is.data.frame(data)){ if (inherits(id,"formula")) id<-model.frame(id,data,na.action=na.fail)[,1] if (inherits(subcoh,"formula")) subcoh<-model.frame(subcoh,data,na.action=na.fail)[,1] if (inherits(stratum,"formula")) stratum<-model.frame(stratum,data,na.action=na.fail)[,1] } ## Check id, subcoh and cohort.size variables if(length(id)!=length(unique(id))) stop("Multiple records per id not allowed") if (is.logical(subcoh)) subcoh <- as.numeric(subcoh) tt <- table(subcoh) if(min(charmatch(names(tt), c("0","1"), 0))==0) stop("Permissible values for subcohort indicator are 0/1 or TRUE/FALSE") if(length(id)>sum(cohort.size)) stop("Number of records greater than cohort size") nn <- cohort.size method<-match.arg(method) stratified<-method %in% c("I.Borgan","II.Borgan") if (!is.null(stratum)) stratum<-factor(stratum) if (stratified){ if (robust) warning("`robust' not implemented for stratified analysis.") if (is.null(stratum)) stop("method (",method,") requires 'stratum'") if (length(cohort.size)!=length(levels(stratum))) stop("cohort.size and stratum do not match") if (!(all(levels(stratum) %in% names(cohort.size)))) warning("stratum levels and names(cohort.size) do not agree") subcohort.sizes<-table(stratum) } else if(!stratified) { if (!(method =="LinYing") && robust) warning("`robust' ignored for method (",method,")") if (!is.null(stratum)) warning("'stratum' ignored for method (",method,")") if (length(cohort.size)!=1) stop("cohort size must be a scalar for unstratified analysis") subcohort.sizes<-length(id) } if (any(subcohort.sizes>cohort.size)) stop("Population smaller than sample in some strata") ## Evaluate model formula m <- match.call(expand.dots=FALSE) m$method <- m$cohort.size <- m$id <- m$subcoh <- m$stratum <-m$robust<- NULL m[[1]] <- as.name("model.frame") m <- eval(m,sys.parent()) Terms <- attr(m,"terms") Y <- model.extract(m, "response") if(!inherits(Y, "Surv")) stop("Response must be a survival object") type <- attr(Y, "type") itype<-charmatch(type,c("right","counting"),nomatch=0) cens<-switch(itype+1, stop(paste("Cox model doesn't support \"", type, "\" survival data", sep = "")), Y[,2], Y[,3]) if (any(!subcoh & !cens)) stop(sum(!subcoh & !cens),"censored observations not in subcohort") cc<-cens+1-subcoh texit<-switch(itype+1, stop(), Y[,1], Y[,2]) tenter<-switch(itype+1, stop(), rep(0,length(texit)), Y[,1]) X <- model.matrix(Terms, m) X <- X[,2:ncol(X)] fitter <- get(method) if (stratified) out<-fitter(tenter=tenter, texit=texit, cc=cc, id=id, X=X, stratum=as.numeric(stratum), stratum.sizes=cohort.size) else out<-fitter(tenter=tenter, texit=texit, cc=cc, id=id, X=X, ntot=nn, robust=robust) out$method <- method names(out$coefficients) <- dimnames(X)[[2]] if(!is.null(out$var)) dimnames(out$var) <- list(dimnames(X)[[2]], dimnames(X)[[2]]) if(!is.null(out$naive.var)) dimnames(out$naive.var) <- list(dimnames(X)[[2]], dimnames(X)[[2]]) out$call <- call out$cohort.size <- cohort.size out$stratified<-stratified if (stratified){ out$stratum<-stratum out$subcohort.size <-subcohort.sizes } else { out$subcohort.size <- tt[2] } class(out) <- "cch" out } ### Subprograms Prentice <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.00000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators ## Calculate Prentice estimate ent2 <- tenter ent2[cc==2] <- texit[cc==2]-eps fit1 <- coxph(Surv(ent2,texit,cens)~X,eps=eps,x=TRUE) ## Calculate Prentice estimate and variance nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort X <- as.matrix(X) aent <- c(tenter[cc>0],tenter[cc<2]) aexit <- c(texit[cc>0],texit[cc<2]) aX <- rbind(as.matrix(X[cc>0,]),as.matrix(X[cc<2,])) aid <- c(id[cc>0],id[cc<2]) dum <- rep(-100,nd) dum <- c(dum,rep(0,nc)) gp <- rep(1,nd) gp <- c(gp,rep(0,nc)) fit <- coxph(Surv(aent,aexit,gp)~aX+offset(dum)+cluster(aid),eps=eps,x=TRUE, iter.max=35,init=fit1$coefficients) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db <- db[gp==0,] fit$phase2var<-(1-(nc/ntot))*t(db)%*%(db) fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$coefficients <- fit$coef <- fit1$coefficients fit } SelfPrentice <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.00000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators ## Calculate Self-Prentice estimate and variance nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort X <- as.matrix(X) aent <- c(tenter[cc>0],tenter[cc<2]) aexit <- c(texit[cc>0],texit[cc<2]) aX <- rbind(as.matrix(X[cc>0,]),as.matrix(X[cc<2,])) aid <- c(id[cc>0],id[cc<2]) dum <- rep(-100,nd) dum <- c(dum,rep(0,nc)) gp <- rep(1,nd) gp <- c(gp,rep(0,nc)) fit <- coxph(Surv(aent,aexit,gp)~aX+offset(dum)+cluster(aid),eps=eps,x=TRUE) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db <- db[gp==0,,drop=FALSE] fit$phase2var<-(1-(nc/ntot))*t(db)%*%(db) fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit } LinYing <- function(tenter, texit, cc, id, X, ntot,robust){ eps <- 0.000000001 cens <- as.numeric(cc>0) # Censorship indicators subcoh <- as.numeric(cc<2) # Subcohort indicators nd <- sum(cens) # Number of failures nc <- sum(subcoh) # Number in subcohort ncd <- sum(cc==1) #Number of failures in subcohort ## Calculate Lin-Ying estimate and variance offs <- rep((ntot-nd)/(nc-ncd),length(texit)) offs[cc>0] <- 1 loffs <- log(offs) fit <- coxph(Surv(tenter, texit, cens)~X+offset(loffs)+cluster(id), eps=eps,x=TRUE) db <- resid(fit,type="dfbeta") db <- as.matrix(db) db0 <- db[cens==0,,drop=FALSE] dbm <- apply(db0,2,mean) db0 <- sweep(db0,2,dbm) fit$phase2var<-(1-(nc-ncd)/(ntot-nd))*crossprod(db0) fit$naive.var <- fit$naive.var+fit$phase2var if (robust) fit$var<- crossprod(db,db/offs)+fit$phase2var else fit$var<-fit$naive.var fit } I.Borgan <- function(tenter, texit, cc, id, X, stratum, stratum.sizes){ eps <- 0.00000001 nobs <- length(texit) idx <- 1:length(nobs) jj <- max(stratum) nn <- stratum.sizes ## Cohort stratum sizes n <- table(stratum) ## Sample stratum sizes d <- table(stratum[cc>0]) ## Failures in each stratum tt <- table(cc,stratum) cens <- as.numeric(cc>0) ## Failure indicators subcoh <- as.numeric(cc<2) ## Subcohort indicators nd <- sum(cens) ## Number of failures nc <- sum(subcoh) ## Number in subcohort ncd <- sum(as.numeric(cc==1)) #Number of failures in subcohort m0 <- tt[1,] ## Subcohort stratum sizes (noncases only) if (ncd>0) m <- m0+tt[2,] else m <- m0 #Subcohort stratum sizes X <- as.matrix(X) kk <- ncol(X) ## Number of variables wt <- as.vector(nn/m) ## Weights for Estimator I stratum <- c(stratum[cc>0],stratum[cc<2]) w <- wt[stratum] ent <- c(tenter[cc > 0], tenter[cc < 2]) exit <- c(texit[cc > 0], texit[cc < 2]) X <- rbind(as.matrix(X[cc > 0, ]), as.matrix(X[cc < 2, ])) id <- c(id[cc > 0], id[cc < 2]) dum <- rep(-100, nd) dum <- c(dum, rep(0, nc)) gp <- rep(1, nd) gp <- c(gp, rep(0, nc)) w[gp==1] <- 1 fit <- coxph(Surv(ent,exit,gp)~X+offset(dum)+cluster(id), weights=w, eps=eps,x=T, iter.max=25) score <- resid(fit, type = "score", weighted=F) sc <- resid(fit, type="score", collapse=id, weighted=T) score <- as.matrix(score) score <- score[gp == 0,,drop=F] st <- stratum[gp==0] sto <- st %o% rep(1,kk) Index <- col(score) tscore <- tapply(score,list(sto,Index),mean) pscore <- tapply(score,list(sto,Index)) score <- score-tscore[pscore] delta <- matrix(0,kk,kk) opt <- NULL for (j in 1:jj) { temp <- t(score[st==j,])%*%score[st==j,]/(m[j]-1) delta <- matrix(delta+(wt[j]-1)*nn[j]*temp,kk,kk) if(is.null(opt)) opt <- nn[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var)) else opt <- rbind(opt,nn[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var))) } z <- apply(opt,2,sum) fit$opt <- sweep(opt,2,z,FUN="/") fit$phase2var<-fit$naive.var%*%delta%*%fit$naive.var fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$delta <- delta fit$sc <- sc fit } II.Borgan <- function(tenter, texit, cc, id, X, stratum, stratum.sizes){ eps <- 0.00000001 jj <- max(stratum) nn <- stratum.sizes ## Cohort stratum sizes n <- table(stratum) ## Sample stratum sizes d <- table(stratum[cc>0]) ## Failures in each stratum tt <- table(cc,stratum) cens <- as.numeric(cc>0) ## Failure indicators subcoh <- as.numeric(cc<2) ## Subcohort indicators nd <- sum(cens) ## Number of failures nc <- sum(subcoh) ## Number in subcohort ncd <- sum(as.numeric(cc==1)) #Number of failures in subcohort m0 <- tt[1,] ## Subcohort stratum sizes (controls only) if (ncd>0) m <- m0+tt[2,] else m <- m0 #Subcohort stratum sizes X <- as.matrix(X) kk <- ncol(X) ## Number of variables nn0 <- nn-as.vector(d) #Noncases in cohort wt <- as.vector(nn0/m0) w <- wt[stratum] w[cens==1] <- 1 fit <- coxph(Surv(tenter,texit,cens)~X+cluster(id), weights=w,eps=eps,x=T, iter.max=25) ## Borgan Estimate II score <- resid(fit, type = "score", weighted=F) sc <- resid(fit,type="score", collapse=id, weighted=T) score <- as.matrix(score) score <- score[cens == 0,,drop=F] ## Scores for controls st <- stratum[cens==0] ## Stratum indicators for controls sto <- st %o% rep(1,kk) Index <- col(score) tscore <- tapply(score,list(sto,Index),mean) ## Within stratum control score means pscore <- tapply(score,list(sto,Index)) score <- score-tscore[pscore] ## Subtract off within stratum score means delta <- matrix(0,kk,kk) opt <- NULL for (j in 1:jj) { temp <- t(score[st==j,])%*%score[st==j,]/(m0[j]-1) ## Borgan equation (19) delta <- delta+(wt[j]-1)*nn0[j]*temp ## Borgan equation (17) if(is.null(opt)) opt <- nn0[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var)) else opt <- rbind(opt,nn0[j]*sqrt(diag(fit$naive.var %*% temp %*% fit$naive.var))) } z <- apply(opt,2,sum) fit$opt <- sweep(opt,2,z,FUN="/") fit$phase2var<-fit$naive.var %*% delta %*% fit$naive.var fit$naive.var <- fit$naive.var+fit$phase2var fit$var<-fit$naive.var fit$delta <- delta fit$sc <- sc fit } ## Methods vcov.cch<-function(object,...) object$var "print.cch"<- function(x,...) { ## produces summary from an x of the class "cch" call<-x$call coef <- coef(x) method <- x$method se <- sqrt(diag(vcov(x))) Z<- abs(coef/se) p<- pnorm(Z) cohort.size<-x$cohort.size subcohort.size<-x$subcohort.size coefficients <- matrix(0, nrow = length(coef), ncol = 4) dimnames(coefficients) <- list(names(coef), c("Value", "SE", "Z", "p")) coefficients[, 1] <- coef coefficients[, 2] <- se coefficients[, 3] <- Z coefficients[, 4] <- 2*(1-p) if (x$stratified){ cat("Exposure-stratified case-cohort analysis,", x$method, "method.\n") m<-rbind(subcohort=x$subcohort.size, cohort=x$cohort.size) print(m,quote=FALSE) } else{ cat("Case-cohort analysis,") cat("x$method,", x$method,"\n with subcohort of", x$subcohort.size,"from cohort of", x$cohort.size,"\n\n") } cat("Call: "); print(x$call) cat("\nCoefficients:\n") print(coefficients) invisible(x) } "summary.cch"<-function(object,...) { ## produces summary from an object of the class "cch" call<-object$call coef <- coef(object) method <- object$method se <- sqrt(diag(vcov(object))) Z<- abs(coef/se) p<- pnorm(Z) cohort.size<-object$cohort.size subcohort.size<-object$subcohort.size coefficients <- matrix(0, nrow = length(coef), ncol = 4) dimnames(coefficients) <- list(names(coef), c("Value", "SE", "Z", "p")) coefficients[, 1] <- coef coefficients[, 2] <- se coefficients[, 3] <- Z coefficients[, 4] <- 2*(1-p) structure(list(call=call, method=method, cohort.size=cohort.size, subcohort.size=subcohort.size, coefficients = coefficients, stratified=object$stratified), class = "summary.cch") } print.summary.cch <- function(x,digits=3,...){ if (x$stratified){ cat("Exposure-stratified case-cohort analysis,", x$method, "method.\n") m<-rbind(subcohort=x$subcohort.size, cohort=x$cohort.size) print(m,quote=FALSE) } else{ cat("Case-cohort analysis,") cat("x$method,", x$method,"\n with subcohort of", x$subcohort.size,"from cohort of", x$cohort.size,"\n\n") } cat("Call: "); print(x$call) cat("\nCoefficients:\n") output<-cbind(Coef=x$coefficients[,1],HR=exp(x$coefficients[,1]), "(95%"=exp(x$coefficients[,1]-1.96*x$coefficients[,2]), "CI)"=exp(x$coefficients[,1]+1.96*x$coefficients[,2]), "p"=x$coefficients[,4] ) print(round(output,3)) invisible(x) } survival/R/frailty.controldf.S0000644000175100001440000000602711732700061016135 0ustar hornikusers# $Id: frailty.controldf.S 11373 2009-10-28 17:12:59Z therneau $ # A function to calibrate the df # very empirical # Find the closest 3 points that span the target value # We know the function is monotone, so fit the function # dy = a * (dx)^p to the 3 points, where dx and dy are the distance # from the leftmost of the three points. # This method can fail near a boundary, so use step halving if things don't # go well # On input, parms$df = target degrees of freedom # parms$dfs, parms$thetas = known values (usually 0,0) # parms$guess = first guess # frailty.controldf <- function(parms, iter, old, df) { if (iter==0) { theta <- parms$guess return(list(theta=theta, done=FALSE, history=cbind(thetas=parms$thetas, dfs=parms$dfs))) } eps <- parms$eps if (length(eps)==0) eps <- .1 thetas <- c(old$history[,1], old$theta) dfs <- c(old$history[,2], df) nx <- length(thetas) if (nx==2) { #linear guess based on first two # but try extra hard to bracket the root theta <- thetas[1] + (thetas[2]-thetas[1])*(parms$df - dfs[1])/ (dfs[2] - dfs[1]) if (parms$df > df) theta <- theta * 1.5 return(list(theta=theta, done=FALSE, history=cbind(thetas=thetas, dfs=dfs), half=0)) } else{ # Now, thetas= our guesses at theta # dfs = the degrees of freedom for each guess done <- (iter>1 && (abs(dfs[nx]-parms$df) < eps)) # look for a new minimum x <- thetas y <- dfs target <- parms$df # How am I doing if ( abs( (y[nx]-target)/(y[nx-1]-target)) > .6) doing.well <- FALSE else doing.well <- TRUE ord <- order(x) if ((x[1]-x[2])*(y[1]-y[2]) >0) y <- y[ord] #monotone up else { #monotone down y <- -1* y[ord] target <- -target } x <- x[ord] if (all(y>target)) b1 <- 1 #points 1:3 are the closest then else if (all(y1 && ((target -y[b1]) < (y[b1+1] -target)))) b1 <- b1-1 } #now have the best 3 points # fit them with a power curve anchored at the leftmost one b2 <- b1 + 1:2 xx <- log(x[b2] - x[b1]) yy <- log(y[b2] - y[b1]) power <- diff(yy)/diff(xx) a <- yy[1] - power*xx[1] newx <- (log(target -y[b1]) - a)/power if (length(parms$trace) && parms$trace){ print(cbind(thetas=thetas, dfs=dfs)) cat(" new theta=" , format(x[b1] + exp(newx)), "\n\n") } list(theta=x[b1] + exp(newx), done=done, history=cbind(thetas=thetas, dfs=dfs), half=0) } } survival/R/agreg.fit.R0000644000175100001440000001021512533653577014354 0ustar hornikusers# Automatically generated from all.nw using noweb agreg.fit <- function(x, y, strata, offset, init, control, weights, method, rownames) { n <- nrow(y) nvar <- ncol(x) start <- y[,1] stopp <- y[,2] event <- y[,3] if (all(event==0)) stop("Can't fit a Cox model with 0 failures") # Sort the data (or rather, get a list of sorted indices) # For both stop and start times, the indices go from last to first if (length(strata)==0) { sort.end <- order(-stopp, event) -1L #indices start at 0 for C code sort.start<- order(-start) -1L newstrat <- n } else { sort.end <- order(strata, -stopp, event) -1L sort.start<- order(strata, -start) -1L newstrat <- cumsum(table(strata)) } if (missing(offset) || is.null(offset)) offset <- rep(0.0, n) if (missing(weights)|| is.null(weights))weights<- rep(1.0, n) else if (any(weights<=0)) stop("Invalid weights, must be >0") else weights <- as.vector(weights) if (is.null(nvar) || nvar==0) { # A special case: Null model. Just return obvious stuff # To keep the C code to a small set, we call the usual routines, but # with a dummy X matrix and 0 iterations nvar <- 1 x <- matrix(as.double(1:n), ncol=1) #keep the .C call happy maxiter <- 0 nullmodel <- TRUE if (length(init) !=0) stop("Wrong length for inital values") init <- 0.0 #dummy value to keep a .C call happy (doesn't like 0 length) } else { nullmodel <- FALSE maxiter <- control$iter.max if (is.null(init)) init <- rep(0., nvar) if (length(init) != nvar) stop("Wrong length for inital values") } # the returned value of agfit$coef starts as a copy of init, so make sure # is is a vector and not a matrix; as.double does so. # Solidify the storage mode of other arguments storage.mode(y) <- storage.mode(x) <- "double" storage.mode(offset) <- storage.mode(weights) <- "double" storage.mode(newstrat) <- "integer" agfit <- .Call(Cagfit4, y, x, newstrat, weights, offset, as.double(init), sort.end, sort.start, as.integer(method=="efron"), as.integer(maxiter), as.double(control$eps), as.double(control$toler.chol), as.integer(1)) # internally rescale var <- matrix(agfit$imat,nvar,nvar) coef <- agfit$coef if (agfit$flag < nvar) which.sing <- diag(var)==0 else which.sing <- rep(FALSE,nvar) infs <- abs(agfit$u %*% var) if (maxiter >1) { if (agfit$iter > maxiter) warning("Ran out of iterations and did not converge") else { infs <- ((infs > control$eps) & infs > control$toler.inf*abs(coef)) if (any(infs)) warning(paste("Loglik converged before variable ", paste((1:nvar)[infs],collapse=","), "; beta may be infinite. ")) } } lp <- as.vector(x %*% coef + offset - sum(coef *agfit$means)) score <- as.double(exp(lp)) resid <- .Call(Cagmart3, y, score, weights, newstrat, cbind(sort.end, sort.start), as.integer(method=='efron')) names(resid) <- rownames if (nullmodel) { list(loglik=agfit$loglik[2], linear.predictors = offset, residuals = resid, method= c("coxph.null", 'coxph') ) } else { names(coef) <- dimnames(x)[[2]] coef[which.sing] <- NA concordance <- survConcordance.fit(y, lp, strata, weights) list(coefficients = coef, var = var, loglik = agfit$loglik, score = agfit$sctest, iter = agfit$iter, linear.predictors = as.vector(lp), residuals = resid, means = agfit$means, concordance = concordance, method= 'coxph') } } survival/R/print.summary.survreg.S0000644000175100001440000000320712470201064017012 0ustar hornikusers# $Id: print.summary.survreg.S 11166 2008-11-24 22:10:34Z therneau $ print.summary.survreg <- function(x, digits = max(options()$digits - 4, 3), ...) { correl <- x$correlation if(is.null(digits)) digits <- options()$digits cat("\nCall:\n") dput(x$call) print(x$table, digits = digits) if (nrow(x$var)==length(x$coefficients)) cat("\nScale fixed at",format(x$scale, digits=digits),"\n") else if (length(x$scale)==1) cat ("\nScale=", format(x$scale, digits=digits), "\n") else { cat("\nScale:\n") print(x$scale, digits=digits, ...) } cat("\n", x$parms, "\n", sep='') df <- sum(x$df) - x$idf # The sum is for penalized models cat("Loglik(model)=", format(round(x$loglik[2],1)), " Loglik(intercept only)=", format(round(x$loglik[1],1))) if (df > 0) cat("\n\tChisq=", format(round(x$chi,2)), "on", round(df,1), "degrees of freedom, p=", format(signif(1-pchisq(x$chi, df),2)), "\n") else cat("\n") if (x$robust) cat("(Loglikelihood assumes independent observations)\n") cat("Number of Newton-Raphson Iterations:", format(trunc(x$iter)), "\n") omit <- x$na.action if (length(omit)) cat("n=", x$n, " (", naprint(omit), ")\n", sep="") else cat("n=", x$n, "\n") if(!is.null(correl)) { p <- dim(correl)[2] if(p > 1) { cat("\nCorrelation of Coefficients:\n") ll <- lower.tri(correl) correl[ll] <- format(round(correl[ll], digits=digits)) correl[!ll] <- "" print(correl[-1, - p, drop = FALSE], quote = FALSE) } } cat("\n") invisible(NULL) } survival/R/coxpenal.df.S0000644000175100001440000000610411732700061014667 0ustar hornikusers# $Id: coxpenal.df.S 11166 2008-11-24 22:10:34Z therneau $ # # degrees of freedom computation, based on Bob Gray's paper # # hmat = right hand slice of cholesky of H # hinv = right hand slice of cholesky of H-inverse # fdiag= diagonal of D-inverse # assign.list: terms information # ptype= 1 or 3 if a sparse term exists, 2 or 3 if a non-sparse exists # nvar = # of non-sparse terms # pen1 = the penalty matrix (diagonal thereof) for the sparse terms # pen2 = the penalty matrix for the non-sparse terms # sparse = indicates which term is the sparse one coxpenal.df <- function(hmat, hinv, fdiag, assign.list, ptype, nvar, pen1, pen2, sparse) { if (ptype ==1 & nvar==0) { #only sparse terms hdiag <- 1/fdiag list(fvar2=(hdiag-pen1)*fdiag^2, df=sum((hdiag-pen1)*fdiag), fvar = fdiag, trH=sum(fdiag)) } else if (ptype==2) { # only dense ones hmat.full <- t(hmat) %*% (ifelse(fdiag==0, 0,1/fdiag)* hmat) hinv.full <- hinv %*% (fdiag* t(hinv)) if (length(pen2)==length(hmat.full)) imat <- hmat.full - pen2 else imat <- hmat.full - diag(pen2) var <- hinv.full %*% imat %*% hinv.full if (length(assign.list)==1) list(var2=var, df=sum(imat * hinv.full), trH=sum(diag(hinv.full)), var=hinv.full) else { df <- trH <- NULL d2 <- diag(hinv.full) for (i in assign.list) { temp <- coxph.wtest(hinv.full[i,i], var[i,i])$solve if (is.matrix(temp)) df <- c(df, sum(diag(temp))) else df <- c(df, sum(temp)) trH<- c(trH, sum(d2[i])) } list(var2=var, df=df, trH=trH, var = hinv.full) } } else { # sparse terms + other vars nf <- length(fdiag) - nvar nr1 <- 1:nf nr2 <- (nf+1):(nf+nvar) d1 <- fdiag[nr1] d2 <- fdiag[nr2] temp <- t(hinv[nr1,]) temp2<- t(hinv[nr2,,drop=FALSE]) A.diag <- d1 + c(rep(1,nvar) %*% (temp^2*d2)) B <- hinv[nr1,] %*% (d2 * temp2) C <- hinv[nr2,] %*% (d2 * temp2) #see notation in paper var2 <- C - t(B) %*% (pen1 * B) if (ptype==3) { #additional work when we have penalties on both the sparse term # and on non-sparse terms hmat.22 <- t(hmat) %*%(ifelse(fdiag==0, 0,1/fdiag)* hmat) temp <- C - coxph.wtest(hmat.22, diag(nvar))$solve if (nvar==1) { var2 <- var2 - C*pen2*C # C will be 1 by 1 temp2 <- c(temp*pen2) } else if (length(pen2) == nvar) { var2 <- var2 - C %*% (pen2 * C) #diagonal penalty temp2 <- sum(diag(temp) * pen2) } else { var2 <- var2 - C %*% matrix(pen2,nvar) %*% C temp2 <- sum(diag(temp * pen2)) } } else temp2 <- 0 #temp2 contains trace[B'A^{-1}B P2], this line: P2=0 df <- trH <- NULL cdiag <- diag(C) for (i in 1:length(assign.list)) { if (sparse==i){ df <- c(df, nf - (sum(A.diag * pen1) + temp2)) trH <- c(trH, sum(A.diag)) } else { j <- assign.list[[i]] temp <- coxph.wtest(C[j,j], var2[j,j])$solve if (is.matrix(temp)) df <- c(df, sum(diag(temp))) else df <- c(df, sum(temp)) trH <- c(trH, sum(cdiag[j])) } } list(var=C, df=df, trH=trH, fvar=A.diag, var2=var2) } } survival/R/firstlib.R0000644000175100001440000000100712140532612014277 0ustar hornikusers.onLoad <- function(lib, pkg) { ## survfit.print.n=="start" is compatible with previous R ## and with MASS if (is.null(getOption("survfit.print.n"))) options(survfit.print.n="start") ## survfit.print.mean==TRUE is compatible with previous R/SPLUS if (is.null(getOption("survfit.print.mean"))) options(survfit.print.mean=FALSE) } .onUnload <- function(libpath) library.dynam.unload("survival", libpath) is.category <- function(x) inherits(x,"factor") || is.factor(x) survival/R/survfitTurnbull.S0000644000175100001440000002557412470201064015733 0ustar hornikusers# Compute the K-M for left/right/interval censored data via Turnbull's # slow EM calculation # x is a factor giving the groups, y is a survival object survfitTurnbull <- function(x, y, casewt, type=c('kaplan-meier', 'fleming-harrington', 'fh2'), error=c('greenwood', "tsiatis"), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified'), start.time) { type <- match.arg(type) error <- match.arg(error) conf.type <- match.arg(conf.type) conf.lower<- match.arg(conf.lower) if (is.logical(conf.int)) { # A common error is for users to use "conf.int = FALSE" # it's illegal, but allow it if (!conf.int) conf.type <- "none" conf.int <- .95 } if (!is.Surv(y)) stop("y must be a Surv object") if (!is.factor(x)) stop("x must be a factor") xlev <- levels(x) # Will supply names for the curves x <- as.numeric(x) # keep only the levels if (!missing(start.time)) { # The user has requested that survival be "survival given that they # made it to start.time". We do this by just tossing those who # are known to end before start.time. Now if one of the times were # interval censored (15,42) and start.time were 20, perhaps it should # be modified too, but we don't. I really don't know what the # correct action would be, actually. # ny <- ncol(y) # remove any obs whose end time is <= start.time keep <- (y[,ny-1] >= start.time) if (all(keep==FALSE)) stop(paste("start.time =", start.time, "is greater than all time points.")) x <- x[keep] y <- y[keep,,drop=FALSE] #make sure y remains a matrix casewt <- casewt[keep] } n.used <- as.vector(table(x)) # This is for the printout nstrat <- length(n.used) # Make sure that the time variable is not "counting" type, # and convert "left" to "interval" style. stype <- attr(y, 'type') if (stype=='counting') stop("survfitTurnbull not appropriate for counting process data") if (stype=='interval') status <- y[,3] if (stype=='left') status <- ifelse(y[,2]==0,2,1) if (stype=='right')status <- y[,2] # If any exact times were represented as interval censored, e.g. (x,x) # as the interval for some x, change the code to "uncensored". if (any(status==3)) { who <- (status==3 & y[,1]==y[,2]) status[who] <- 1 } # # Deal with the issue of "almost tied" times that are equal to # machine precision # Convert to the levels of a factor. We come back to times later if (ncol(y)==3) { ftime <- factor(y[,1:2]) y <- matrix(as.numeric(ftime), ncol=2) } else { ftime <- factor(y[,1]) y <- as.matrix(as.numeric(ftime)) } # the code below actually does the estimate, one curve at a time doit <- function(y,status, wt, ...) { n <- length(status) # Find all of the jump points for the KM in the data set, which are # the exact times, plus any right-followed-by-left pairs. # For this computation, an interval censored observation is considered # to be of the form (a,b], left censored is (-infinity,b] and right # censored is (a, infinity). If there are two interval censored # obs of (10,20] and (20,40], we do NOT want to create a jtimes entry # at 20. # # The algorithm puts a [ at t for each exact, a ( at t for each right # censored, a ] at t for each left censored, and ( and ] at t1/t2 # for each interval censor. In ties, order the parens as [, ], (. # Then find pairs of left-followed-immediately-by-right. The stat2 # variable is 0= [ at t, 1= ] at t, 2= ( at t. # The variables time2, stat2 are never needed after jtimes has # been created. if (any(status==3)) { #interval censored stat2 <- c(c(2,0,1,2)[status+1], rep(1, sum(status==3))) time2 <- c(y[,1], y[status==3,2]) } else { stat2 <- c(2,0,1)[status+1] time2 <- y[,1] } ord <- order(time2, stat2) time2 <- time2[ord] stat2 <- stat2[ord] n2 <- length(time2) pairs <- (stat2[-n2]!=1 & stat2[-1]==1) jtimes <- c(time2[stat2==0], .5*(time2[-n2] + time2[-1])[pairs]) # # If any of the left censored times are < min(jtime), then treat # them as though they were exact (for now). The formal MLE # algebra puts all their mass at an arbitray point between the # smallest of such times and -infinity. # mintime <- min(jtimes) who <- (status==2 & y[,1] < mintime) if (any(who)) { status[who] <- 1 jtimes <- c(y[who,1], jtimes) } # The KM is computed on a fake data set with njump points # standing in for the left and interval censored observations # So tempy contains the exact and right censored y data, followed # by the fakes jtimes <- sort(unique(jtimes)) njump <- length(jtimes) nreal <- sum(status<2) tempx <- factor(rep(1, njump + nreal)) #dummy x var for survfit.km tempy <- Surv(c(y[status<2, 1], jtimes), c(status[status<2], rep(1, njump))) # wtmat marks, for each left/interval obs, which jump points are in it # A column is a "fake" time point, a row is an observation # For a left censored obs, we assume that the true event time is # <= the time recorded, and for an interval one that (a, b] contains # the true event time. This is motivated by data that would come # from repeated visits, and agrees with Turnbull's paper. # If all status <=1, this is the unusual case of left censoring before # some minimal time, in which case I can skip this step. There are # no interval censored or left censored be "split". if (any(status>1)) { temp <- matrix(jtimes, nrow=sum(status>1), ncol=njump, byrow=TRUE) indx <- (1:n)[status>1] #the subjects of interest temp1 <- (temp <= y[indx,1]) # logical matrix for the left censored if (any(status>2)) #interval censored temp2 <- (temp > y[indx,1] & temp <= y[indx,2]) else temp2 <- FALSE & temp1 temp3 <- rep(status[indx]==2, njump) wtmat <- matrix(as.numeric((temp3&temp1) | (!temp3 & temp2)), ncol=njump) lwt <- wt[indx] # the input vector of case weights, for these } else { wtmat <- matrix(rep(1, length(jtimes)), nrow=1) lwt <- 1 } eps <- 1 # The initial "starter" KM is proportional to the number of intervals # that overlap each time point temp <- apply(wtmat, 2, sum) tfit <- list(time=jtimes, surv= 1- cumsum(temp)/sum(temp)) old <- tfit$surv iter <- 0 aitken1 <- jump1 <- 0 #dummy values for lagging while (eps > .00005) { iter <- iter +1 # partition each left/interval person out over the jumps jumps <- -diff(c(1, tfit$surv[match(jtimes, tfit$time)])) #KM jumps if (TRUE) { # add Aitken acceleration to speed things up # Given 3 points on a sequence, it guesses ahead. So we use # a set of 3 to guess ahead, generate 3 more regular EM, # guess ahead, 3 regular EM, etc. # Actually, we go every 5th below instead of every 3rd, to # give the EM a chance to restabilize the relative # sizes of elements of "jump". We also only allow it to # stop when comparing two "real EM" iterations. aitken2 <- aitken1 aitken1 <- jumps - jump1 jsave <- jumps if (iter%%5 ==0) { oldlik <- sum(log(wtmat %*% jumps)) jumps <- jump2 - (aitken2)^2/(aitken1 - aitken2) bad <- (jumps<=0 | jumps >=1 | is.na(jumps)) jumps[bad] <- jsave[bad] #failsafe newlik <- sum(log(wtmat %*% jumps)) if (newlik < oldlik) jumps <- jsave # aitkin didn't work! } jump2 <- jump1 # jumps, lagged by 2 iterations jump1 <- jsave # jumps, lagged by 1 iteration } wt2 <- wtmat %*% diag(jumps) wt2 <- (lwt/(apply(wt2,1,sum))) * wt2 wt2 <- apply(wt2, 2, sum) tfit <- survfitKM(tempx, tempy, casewt=c(wt[status<2], wt2), ...) if (FALSE) { # these lines are in for debugging: change the above to # " if (TRUE)" to turn on the printing cat("\n Iteration = ", iter, "\n") cat("survival=", format(round(tfit$surv[tfit$n.event>0],3)), "\n") cat(" weights=", format(round(wt2,3)), "\n") } stemp <- tfit$surv[match(jtimes, tfit$time)] if (iter%%5<2) eps <- 1 #only check eps for a pair of EM iters else eps <- max(abs(old-stemp)) old <- stemp } # # Now, fix up the "cheating" I did for any left censoreds which were # less than the smallest jump time who <- (tfit$time < mintime & tfit$n.event >0) if (any(who)) { indx <- match(mintime, tfit$time) # first "real" time # tfit$surv[who] <- tfit$surv[indx] tfit$n.event[who] <- 0 # if (!is.null(tfit$std.err)) { # tfit$std.err[who] <- tfit$std.err[indx] # tfit$lower[who] <- tfit$lower[indx] # tfit$upper[who] <- tfit$upper[indx] # } } tfit } # # Now to work, one curve at a time # time <- vector('list', nstrat) n.risk <- vector('list', nstrat) surv <- vector('list', nstrat) n.cens <- vector('list', nstrat) n.event<- vector('list', nstrat) fnumeric <- as.numeric(levels(ftime)) uniquex <- sort(unique(x)) for (i in 1:nstrat) { who <- (x== uniquex[i]) tfit <- doit(y[who,,drop=FALSE], status[who], casewt[who]) time[[i]] <- approx(1:length(fnumeric), fnumeric, tfit$time)$y n.risk[[i]] <- tfit$n.risk surv[[i]] <- tfit$surv n.cens[[i]] <- tfit$n.cens n.event[[i]]<- tfit$n.event if (i==1) { if (!is.null(tfit$std.err)) { std.err <- vector('list', nstrat) conf.lower <- vector('list', nstrat) conf.upper <- vector('list', nstrat) se.fit <- TRUE } else se.fit <- FALSE } if (se.fit) { std.err[[i]] <- tfit$std.err conf.lower[[i]] <- tfit$lower conf.upper[[i]] <- tfit$upper } } temp <- list(n=n.used, time = unlist(time), n.risk = unlist(n.risk), n.event= unlist(n.event), n.censor = unlist(n.cens), surv = unlist(surv), type='right') if (nstrat >1) { strata <- unlist(lapply(time, length)) names(strata) <- xlev[sort(unique(x))] temp$strata <- strata } if (se.fit) { temp$std.err <- unlist(std.err) temp$lower <- unlist(conf.lower) temp$upper <- unlist(conf.upper) temp$conf.type <- tfit$conf.type temp$conf.int <- tfit$conf.int } temp } survival/R/survdiff.fit.S0000644000175100001440000000256612113164602015106 0ustar hornikuserssurvdiff.fit <- function(y, x, strat, rho=0) { # # This routine is almost always called from survdiff # If called directly, remember that it does no error checking # n <- length(x) if (ncol(y) !=2) stop ("Invalid y matrix") if (nrow(y) !=n | length(x) !=n) stop("Data length mismatch") ngroup <- length(unique(x)) if (ngroup <2) stop ("There is only 1 group") if (is.category(x)) x <- as.numeric(x) else x <- match(x, unique(x)) if (missing(strat)) strat <- rep(1,n) else strat <- as.numeric(as.factor(strat)) nstrat <- length(unique(strat)) if (length(strat) !=n) stop("Data length mismatch") ord <- order(strat, y[,1], -y[,2]) strat2 <- c(1*(diff(strat[ord])!=0), 1) xx <- .C(Csurvdiff2, as.integer(n), as.integer(ngroup), as.integer(nstrat), as.double(rho), as.double(y[ord,1]), as.integer(y[ord,2]), as.integer(x[ord]), as.integer(strat2), observed = double(ngroup*nstrat), expected = double(ngroup*nstrat), var.e = double(ngroup * ngroup), double(ngroup), double(n)) if (nstrat==1) list(expected = xx$expected, observed = xx$observed, var = matrix(xx$var.e, ngroup, ngroup)) else list(expected = matrix(xx$expected, ngroup), observed = matrix(xx$observed, ngroup), var = matrix(xx$var.e, ngroup, ngroup)) } survival/R/print.summary.coxph.penal.S0000644000175100001440000000445612423460706017553 0ustar hornikusersprint.summary.coxph.penal <- function(x, digits = max(options()$digits - 3, 3), signif.stars = getOption("show.signif.stars"), ...) { if (!is.null(x$call)) { cat("Call:\n") dput(x$call) cat("\n") } if (!is.null(x$fail)) { cat(" Coxreg failed.", x$fail, "\n") return() } savedig <- options(digits = digits) on.exit(options(savedig)) omit <- x$na.action cat(" n=", x$n) if (!is.null(x$nevent)) cat(", number of events=", x$nevent, "\n") else cat("\n") if (length(omit)) cat(" (", naprint(omit), ")\n\n", sep="") else cat("\n") # Format out the NA in the coef matrix print1 <- x$coefficients temp <- cbind(format(print1[,1]), format(print1[,2]), format(print1[,3]), format(round(print1[,4], 2)), format(round(print1[,5], 2)), format(signif(print1[,6], 2))) temp <- ifelse(is.na(print1), "", temp) dimnames(temp) <- dimnames(print1) print(temp, quote=FALSE) if(length(x$conf.int) >0 ) { cat("\n") print(x$conf.int) } logtest <- -2 * (x$loglik[1] - x$loglik[2]) sctest <- x$score cat("\nIterations:", x$iter[1], "outer,", x$iter[2], "Newton-Raphson\n") if (length(x$print2)) { for (i in 1:length(x$print2)) cat(" ", x$print2[i], "\n") } if (is.null(x$df)) df <- sum(!is.na(coef)) else df <- round(sum(x$df),2) cat("Degrees of freedom for terms=", format(round(x$df,1)), "\n") if (!is.null(x$concordance)) { cat("Concordance=", format(round(x$concordance[1],3)), " (se =", format(round(x$concordance[2], 3)),")\n") } cat("Likelihood ratio test= ", format(round(logtest, 2)), " on ", df, " df,", " p=", format(1 - pchisq(logtest, df)), "\n", sep = "") if (!is.null(x$wald.test)) cat("Wald test = ", format(round(x$wald.test, 2)), " on ", df, " df, p=", format(1 - pchisq(x$wald.test, df)), sep = "") if (!is.null(x$score)) cat("\nScore (logrank) test = ", format(round(sctest, 2)), " on ", df, " df,", " p=", format(1 - pchisq(sctest, df)), sep ="") if (is.null(x$rscore)) cat("\n") else cat(", Robust = ", format(round(x$rscore, 2)), " p=", format(1 - pchisq(x$rscore, df)), "\n", sep="") invisible() } survival/R/print.summary.survfit.S0000644000175100001440000000410712423461263017026 0ustar hornikusersprint.summary.survfit <- function(x, digits = max(options()$digits - 4, 3), ...) { savedig <- options(digits=digits) on.exit(options(savedig)) if (!is.null(cl<- x$call)) { cat("Call: ") dput(cl) cat("\n") } omit <- x$na.action if (length(omit)) cat(naprint(omit), "\n") if (x$type == 'right' || is.null(x$n.enter)) { mat <- cbind(x$time, x$n.risk, x$n.event, x$surv) cnames <- c("time", "n.risk", "n.event") } else if (x$type == 'counting') { mat <- cbind(x$time, x$n.risk, x$n.event, x$n.enter, x$n.censor, x$surv) cnames <- c("time", "n.risk", "n.event", "entered", "censored") } if (is.matrix(x$surv)) ncurve <- ncol(x$surv) else ncurve <- 1 if (ncurve==1) { #only 1 curve cnames <- c(cnames, "survival") if (!is.null(x$std.err)) { if (is.null(x$lower)) { mat <- cbind(mat, x$std.err) cnames <- c(cnames, "std.err") } else { mat <- cbind(mat, x$std.err, x$lower, x$upper) cnames <- c(cnames, 'std.err', paste("lower ", x$conf.int*100, "% CI", sep=''), paste("upper ", x$conf.int*100, "% CI", sep='')) } } } else cnames <- c(cnames, paste("survival", seq(ncurve), sep='')) if (!is.null(x$start.time)) { mat.keep <- mat[,1] >= x$start.time mat <- mat[mat.keep,,drop=FALSE] if (is.null(dim(mat))) stop(paste("No information available using start.time =", x$start.time, ".")) } if (!is.matrix(mat)) mat <- matrix(mat, nrow=1) if (!is.null(mat)) { dimnames(mat) <- list(rep("", nrow(mat)), cnames) if (is.null(x$strata)) print(mat) else { #print it out one strata at a time strata <- x$strata if (!is.null(x$start.time)) strata <- strata[mat.keep] for (i in levels(strata)) { who <- (strata==i) cat(" ", i, "\n") print(mat[who,]) cat("\n") } } } else stop("There are no events to print. Please use the option ", "censored=TRUE with the summary function to see the censored ", "observations.") invisible(x) } survival/R/dsurvreg.S0000644000175100001440000000305012160146613014327 0ustar hornikusers# The density, quantile, and CDF functions for those distributions # supported by survreg # dsurvreg <- function(x, mean, scale=1, distribution='weibull', parms) { dist <- survreg.distributions[[casefold(distribution)]] if (is.null(dist)) stop("Distribution not found") if (!is.null(dist$trans)) { dx <- dist$dtrans(x) x <- dist$trans(x) x <- (x-mean)/scale dist <- survreg.distributions[[dist$dist]] y <- dist$density(x, parms)[,3] y *dx / scale } else { x <- (x-mean)/scale y <- dist$density(x, parms)[,3] y/ scale } } psurvreg <- function(q, mean, scale=1, distribution='weibull', parms) { dist <- survreg.distributions[[casefold(distribution)]] if (is.null(dist)) stop("Distribution not found") if (!is.null(dist$trans)) { q <- dist$trans(q) q <- (q-mean)/scale dist <- survreg.distributions[[dist$dist]] dist$density(q, parms)[,1] } else { q <- (q-mean)/scale dist$density(q, parms)[,1] } } qsurvreg <- function(p, mean, scale=1, distribution='weibull', parms) { dist <- survreg.distributions[[casefold(distribution)]] if (is.null(dist)) stop("Distribution not found") if (!is.null(dist$trans)) { d2 <- survreg.distributions[[dist$dist]] x <- d2$quantile(p, parms) dist$itrans(x*scale + mean) } else { x <- dist$quantile(p, parms) x*scale + mean } } rsurvreg <- function(n, mean, scale=1, distribution='weibull', parms) { if (missing(parms)) qsurvreg(runif(n), mean, scale, distribution) else qsurvreg(runif(n), mean, scale, distribution, parms) } survival/R/anova.coxphlist.S0000644000175100001440000000421512424205642015613 0ustar hornikusers# This is usually called from anova.coxph, not a user # It's first argument must be a list of coxph models anova.coxphlist <- function (object, test = 'Chisq' ,...) { if (!is.list(object)) stop("First argument must be a list") if (!all(unlist(lapply(object, function(x) inherits(x, 'coxph'))))) stop("Argument must be a list of coxph models") if (any(sapply(object, function(x) !is.null(x$naive.var)))) stop("Can't do anova tables with robust variances") responses <- as.character(unlist(lapply(object, function(x) deparse(formula(x)[[2]])))) sameresp <- (responses == responses[1]) if (!all(sameresp)) { object <- object[sameresp] warning(paste("Models with response", deparse(responses[!sameresp]), "removed because response differs from", "model 1")) } ns <- sapply(object, function(x) length(x$residuals)) if (any(ns != ns[1])) stop("models were not all fitted to the same size of dataset") nmodels <- length(object) if (nmodels == 1) # only one model remains return(anova.coxph(object[[1]], test = test)) loglik <- unlist(lapply(object, function(x) x$loglik[2])) df <- unlist(lapply(object, function(x) if (!is.null(x$df)) sum(x$df) else sum(!is.na(coef(x))))) table <- data.frame(loglik, Chisq= c(NA, abs(2*diff(loglik))), Df= abs(c(NA, diff(df)))) tfun <- function(x) paste(as.character(delete.response(terms(formula(x)))), collapse=' ') variables <- lapply(object, tfun) dimnames(table) <- list(1:nmodels, c("loglik", "Chisq", "Df")) title <- paste("Analysis of Deviance Table\n Cox model: response is ", responses[1]) topnote <- paste(" Model ", format(1:nmodels), ": ", variables, sep = "", collapse = "\n") if (!is.null(test)) { table[['P(>|Chi|)']] <- 1-pchisq(table$Chisq, table$Df) } if (is.R()) structure(table, heading = c(title, topnote), class = c("anova", "data.frame")) else structure(table, heading = c(title, topnote), class = "anova") } survival/R/survfit.coxph.R0000644000175100001440000003407412533653611015326 0ustar hornikusers# Automatically generated from all.nw using noweb survfit.coxph <- function(formula, newdata, se.fit=TRUE, conf.int=.95, individual=FALSE, type, vartype, conf.type=c("log", "log-log", "plain", "none"), censor=TRUE, id, na.action=na.pass, ...) { Call <- match.call() Call[[1]] <- as.name("survfit") #nicer output for the user object <- formula #'formula' because it has to match survfit if (!is.null(attr(object$terms, "specials")$tt)) stop("The survfit function can not yet process coxph models with a tt term") if (missing(type)) { # Use the appropriate one from the model temp1 <- c("exact", "breslow", "efron") survtype <- match(object$method, temp1) } else { temp1 <- c("kalbfleisch-prentice", "aalen", "efron", "kaplan-meier", "breslow", "fleming-harrington", "greenwood", "tsiatis", "exact") survtype <- match(match.arg(type, temp1), temp1) survtype <- c(1,2,3,1,2,3,2,2,1)[survtype] } if (missing(vartype)) { vartype <- survtype } else { temp2 <- c("greenwood", "aalen", "efron", "tsiatis") vartype <- match(match.arg(vartype, temp2), temp2) if (vartype==4) vartype<- 2 } if (!se.fit) conf.type <- "none" else conf.type <- match.arg(conf.type) has.strata <- !is.null(attr(object$terms, 'specials')$strata) if (is.null(object$y) || is.null(object[['x']]) || !is.null(object$call$weights) || (has.strata && is.null(object$strata)) || !is.null(attr(object$terms, 'offset'))) { mf <- model.frame(object) } else mf <- NULL #useful for if statements later if (is.null(mf)) y <- object[['y']] else { y <- model.response(mf) y2 <- object[['y']] if (!is.null(y2) && any(as.matrix(y2) != as.matrix(y))) stop("Could not reconstruct the y vector") } if (is.null(object[['x']])) x <- model.matrix.coxph(object, data=mf) else x <- object[['x']] n <- nrow(y) if (n != object$n[1] || nrow(x) !=n) stop("Failed to reconstruct the original data set") if (is.null(mf)) wt <- rep(1., n) else { wt <- model.weights(mf) if (is.null(wt)) wt <- rep(1.0, n) } type <- attr(y, 'type') if (type != 'right' && type != 'counting') stop("Cannot handle \"", type, "\" type survival data") missid <- missing(id) # I need this later, and setting id below makes # "missing(id)" always false if (!missid) individual <- TRUE else if (missid && individual) id <- rep(0,n) #dummy value else id <- NULL if (individual && missing(newdata)) { stop("the id and/or individual options only make sense with new data") } if (individual && type!= 'counting') stop("The individual option is only valid for start-stop data") if (is.null(mf)) offset <- 0 else { offset <- model.offset(mf) if (is.null(offset)) offset <- 0 } Terms <- object$terms if (!has.strata) strata <- rep(0L,n) else { stangle <- untangle.specials(Terms, 'strata') # used multiple times strata <- object$strata #try this first if (is.null(strata)){ if (length(stangle$vars) ==1) strata <- mf[[stangle$vars]] else strata <- strata(mf[, stangle$vars], shortlabel=TRUE) } } if (has.strata) { temp <- attr(Terms, "specials")$strata factors <- attr(Terms, "factors")[temp,] strata.interaction <- any(t(factors)*attr(Terms, "order") >1) } if (is.null(x) || ncol(x)==0) { # a model with ~1 on the right hand side # Give it a dummy x so the rest of the code goes through # (This case is really rare) x <- matrix(0., nrow=n) coef <- 0.0 varmat <- matrix(0.0,1,1) risk <- rep(exp(offset- mean(offset)), length=n) } else { varmat <- object$var coef <- ifelse(is.na(object$coefficients), 0, object$coefficients) xcenter <- object$means if (is.null(object$frail)) { x <- scale(x, center=xcenter, scale=FALSE) risk <- c(exp(x%*% coef + offset - mean(offset))) } else { keep <- !is.na(match(dimnames(x)[[2]], names(coef))) x <- x[,keep, drop=F] # varmat <- varmat[keep,keep] #coxph already has trimmed it risk <- exp(object$linear.predictor) x <- scale(x, center=xcenter, scale=FALSE) } } subterms <- function(tt, i) { dataClasses <- attr(tt, "dataClasses") predvars <- attr(tt, "predvars") oldnames <- dimnames(attr(tt, 'factors'))[[1]] tt <- tt[i] index <- match(dimnames(attr(tt, 'factors'))[[1]], oldnames) if (length(index) >0) { if (!is.null(predvars)) attr(tt, "predvars") <- predvars[c(1, index+1)] if (!is.null(dataClasses)) attr(tt, "dataClasses") <- dataClasses[index] } tt } temp <- untangle.specials(Terms, 'cluster') if (length(temp$vars)) Terms <- subterms(Terms, -temp$terms) if (missing(newdata)) { mf2 <- as.list(object$means) #create a dummy newdata names(mf2) <- names(object$coefficients) mf2 <- as.data.frame(mf2) found.strata <- FALSE } else { if (!is.null(object$frail)) stop("Newdata cannot be used when a model has frailty terms") Terms2 <- Terms if (!individual) Terms2 <- delete.response(Terms) if (is.vector(newdata, "numeric")) { if (individual) stop("newdata must be a data frame") if (is.null(names(newdata))) { stop("Newdata argument must be a data frame") } newdata <- data.frame(as.list(newdata)) } if (missid) { if (has.strata && !strata.interaction) { found.strata <- TRUE tempenv <- new.env(, parent=emptyenv()) assign("strata", function(..., na.group, shortlabel, sep) list(...), envir=tempenv) assign("list", list, envir=tempenv) for (svar in stangle$vars) { temp <- try(eval(parse(text=svar), newdata, tempenv), silent=TRUE) if (!is.list(temp) || any(unlist(lapply(temp, class))== "function")) found.strata <- FALSE } if (found.strata) mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=object$xlevels) else { Terms2 <- subterms(Terms2, -attr(Terms2, 'specials')$strata) if (!is.null(object$xlevels)) { myxlev <- object$xlevels[match(attr(Terms2, "term.labels"), names(object$xlevels), nomatch=0)] if (length(myxlev)==0) myxlev <- NULL } else myxlev <- NULL mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=myxlev) } } else { mf2 <- model.frame(Terms2, data=newdata, na.action=na.action, xlev=object$xlevels) found.strata <- has.strata #would have failed otherwise } } else { tcall <- Call[c(1, match(c('id', "na.action"), names(Call), nomatch=0))] tcall$data <- newdata tcall$formula <- Terms2 tcall$xlev <- object$xlevels tcall[[1]] <- as.name('model.frame') mf2 <- eval(tcall) found.strata <- has.strata # would have failed otherwise } } if (has.strata && found.strata) { #pull them off temp <- untangle.specials(Terms2, 'strata') strata2 <- strata(mf2[temp$vars], shortlabel=TRUE) strata2 <- factor(strata2, levels=levels(strata)) if (any(is.na(strata2))) stop("New data set has strata levels not found in the original") Terms2 <- Terms2[-temp$terms] } else strata2 <- factor(rep(0, nrow(mf2))) if (individual) { if (missing(newdata)) stop("The newdata argument must be present when individual=TRUE") if (!missid) { #grab the id variable id <- model.extract(mf2, "id") if (is.null(id)) stop("id=NULL is an invalid argument") } else id <- rep(1, nrow(mf2)) x2 <- model.matrix(Terms2, mf2)[,-1, drop=FALSE] #no intercept if (length(x2)==0) stop("Individual survival but no variables") x2 <- scale(x2, center=xcenter, scale=FALSE) offset2 <- model.offset(mf2) if (length(offset2) >0) offset2 <- offset2 - mean(offset) else offset2 <- 0 y2 <- model.extract(mf2, 'response') if (attr(y2,'type') != type) stop("Survival type of newdata does not match the fitted model") if (attr(y2, "type") != "counting") stop("Individual=TRUE is only valid for counting process data") y2 <- y2[,1:2, drop=F] #throw away status, it's never used newrisk <- exp(c(x2 %*% coef) + offset2) result <- survfitcoxph.fit(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat, id, y2, strata2) } else { if (missing(newdata)) { if (has.strata && strata.interaction) stop ("Models with strata by covariate interaction terms require newdata") x2 <- matrix(0.0, nrow=1, ncol=ncol(x)) offset2 <- 0 } else { offset2 <- model.offset(mf2) if (length(offset2) >0) offset2 <- offset2 - mean(offset) else offset2 <- 0 x2 <- model.matrix(Terms2, mf2)[,-1, drop=FALSE] #no intercept x2 <- scale(x2, center=xcenter, scale=FALSE) } newrisk <- exp(c(x2 %*% coef) + offset2) result <- survfitcoxph.fit(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat) if (has.strata && found.strata) { if (is.matrix(result$surv)) { nr <- nrow(result$surv) #a vector if newdata had only 1 row indx1 <- split(1:nr, rep(1:length(result$strata), result$strata)) rows <- indx1[as.numeric(strata2)] #the rows for each curve indx2 <- unlist(rows) #index for time, n.risk, n.event, n.censor indx3 <- as.integer(strata2) #index for n and strata for(i in 2:length(rows)) rows[[i]] <- rows[[i]]+ (i-1)*nr #linear subscript indx4 <- unlist(rows) #index for surv and std.err temp <- result$strata[indx3] names(temp) <- row.names(mf2) new <- list(n = result$n[indx3], time= result$time[indx2], n.risk= result$n.risk[indx2], n.event=result$n.event[indx2], n.censor=result$n.censor[indx2], strata = temp, surv= result$surv[indx4], cumhaz = result$cumhaz[indx4]) if (se.fit) new$std.err <- result$std.err[indx4] result <- new } } } if (!censor) { kfun <- function(x, keep){ if (is.matrix(x)) x[keep,,drop=F] else if (length(x)==length(keep)) x[keep] else x} keep <- (result$n.event > 0) if (!is.null(result$strata)) { temp <- factor(rep(names(result$strata), result$strata), levels=names(result$strata)) result$strata <- c(table(temp[keep])) } result <- lapply(result, kfun, keep) } if (se.fit) { zval <- qnorm(1- (1-conf.int)/2, 0,1) if (conf.type=='plain') { temp1 <- result$surv + zval* result$std.err * result$surv temp2 <- result$surv - zval* result$std.err * result$surv result <- c(result, list(upper=pmin(temp1,1), lower=pmax(temp2,0), conf.type='plain', conf.int=conf.int)) } if (conf.type=='log') { xx <- ifelse(result$surv==0,1,result$surv) #avoid some "log(0)" messages temp1 <- ifelse(result$surv==0, 0*result$std.err, exp(log(xx) + zval* result$std.err)) temp2 <- ifelse(result$surv==0, 0*result$std.err, exp(log(xx) - zval* result$std.err)) result <- c(result, list(upper=pmin(temp1,1), lower=temp2, conf.type='log', conf.int=conf.int)) } if (conf.type=='log-log') { who <- (result$surv==0 | result$surv==1) #special cases xx <- ifelse(who, .1,result$surv) #avoid some "log(0)" messages temp1 <- exp(-exp(log(-log(xx)) + zval*result$std.err/log(xx))) temp1 <- ifelse(who, result$surv + 0*result$std.err, temp1) temp2 <- exp(-exp(log(-log(xx)) - zval*result$std.err/log(xx))) temp2 <- ifelse(who, result$surv + 0*result$std.err, temp2) result <- c(result, list(upper=temp1, lower=temp2, conf.type='log-log', conf.int=conf.int)) } } result$call <- Call # The "type" component is in the middle -- match history indx <- match('surv', names(result)) result <- c(result[1:indx], type=attr(y, 'type'), result[-(1:indx)]) if (is.R()) class(result) <- c('survfit.cox', 'survfit') else oldClass(result) <- 'survfit.cox' result } survival/R/frailty.gammacon.S0000644000175100001440000000136712470201064015726 0ustar hornikusers# $Id: frailty.gammacon.S 11166 2008-11-24 22:10:34Z therneau $ # Correct the loglik for a gamma frailty # Term2 is the hard one, discussed in section 3.5 of the report # The penalty function only adds \vu \sum(w_j) to the CoxPL, so this # does a bit more than equation 15. # frailty.gammacon <- function(d, nu) { maxd <- max(d) if (nu > 1e7*maxd) term1 <- sum(d*d)/nu #second order Taylor series else term1 <- sum(d + nu*log(nu/(nu+d))) #easy part tbl <- table(factor(d[d>0], levels=1:maxd)) ctbl<- rev(cumsum(rev(tbl))) dlev<- 1:maxd term2.numerator <- nu + rep(dlev-1, ctbl) term2.denom <- nu + rep(dlev, tbl*dlev) term2 <- sum(log(term2.numerator/term2.denom)) term1 + term2 } survival/R/xtras.R0000644000175100001440000000300412466142446013636 0ustar hornikusersvcov.coxph<-function (object, ...) { rval<-object$var dimnames(rval)<-list(names(coef(object)),names(coef(object))) rval } vcov.survreg<-function (object, ...) { object$var } # The extractAIC methods for coxph and survreg objects are defined # in the stats package. Don't reprise them here. extractAIC.coxph.penal<- function(fit,scale,k=2,...){ edf<-sum(fit$df) loglik <- fit$loglik[length(fit$loglik)] c(edf, -2 * loglik + k * edf) } labels.survreg <- function(object, ...) attr(object,"term.labels") rep.Surv <- function(x, ...) { indx <- rep(1:nrow(x), ...) x[indx,] } # This function is just like all.vars -- except that it does not recur # on the $ sign, it follows both arguments of +, * and - in order to # track formulas, all arguments of Surv, and only the first of things # like ns(). # This is used to generate a warning in coxph if the same variable is used # on both sides, so perfection is not required. terms.inner <- function(x) { if (class(x) == "formula") c(terms.inner(x[[2]]), terms.inner(x[[3]])) else if (class(x)== "call" && (x[[1]] != as.name("$") && x[[1]] != as.name("["))) { if (x[[1]] == '+' || x[[1]]== '*' || x[[1]] == '-') { # terms in a model equation c(terms.inner(x[[2]]), terms.inner(x[[3]])) } else if (x[[1]] == as.name("Surv")) unlist(lapply(x[-1], terms.inner)) else terms.inner(x[[2]]) } else(deparse(x)) } survival/R/print.cox.zph.S0000644000175100001440000000026211732700061015211 0ustar hornikusers# $Id: print.cox.zph.S 11166 2008-11-24 22:10:34Z therneau $ print.cox.zph <- function(x, digits = max(options()$digits - 4, 3),...) invisible(print(x$table, digits=digits)) survival/R/labels.survreg.S0000644000175100001440000000021211732700061015416 0ustar hornikusers# $Id: labels.survreg.S 11166 2008-11-24 22:10:34Z therneau $ labels.survreg <- function(object, ...) attr(object$terms, "term.labels") survival/R/coxpenal.fit.R0000644000175100001440000004664712470400252015076 0ustar hornikusers# # General penalized likelihood # coxpenal.fit <- function(x, y, strata, offset, init, control, weights, method, rownames, pcols, pattr, assign) { eps <- control$eps n <- nrow(y) if (is.matrix(x)) nvar <- ncol(x) else if (length(x)==0) stop("Must have an X variable") else nvar <-1 if (missing(offset) || is.null(offset)) offset <- rep(0,n) if (missing(weights)|| is.null(weights))weights<- rep(1,n) else { if (any(weights<=0)) stop("Invalid weights, must be >0") } # Get the list of sort indices, but don't sort the data itself if (ncol(y) ==3) { if (length(strata) ==0) { sorted <- cbind(order(-y[,2], y[,3]), order(-y[,1])) -1L newstrat <- as.integer(n) } else { sorted <- cbind(order(strata, -y[,2], y[,3]), order(strata, -y[,1])) -1L newstrat <- as.integer(cumsum(table(strata))) } status <- y[,3] andersen <- TRUE } else { if (length(strata) ==0) { sorted <- order(-y[,1], y[,2]) -1L newstrat <- as.integer(n) } else { sorted <- order(strata, -y[,1], y[,2]) -1L newstrat <- as.integer(cumsum(table(strata))) } status <- y[,2] andersen <- FALSE } n.eff <- sum(y[,ncol(y)]) #effective n for a Cox model is #events # # are there any sparse frailty terms? # npenal <- length(pattr) if (npenal == 0 || length(pcols) != npenal) stop("Invalid pcols or pattr arg") sparse <- sapply(pattr, function(x) !is.null(x$sparse) && x$sparse) if (sum(sparse) >1) stop("Only one sparse penalty term allowed") # # Create a marking vector for the terms, the same length as assign # with pterms == 0=ordinary term, 1=penalized, 2=sparse, # pindex = length of pcols = position in pterms # # Make sure that pcols is a strict subset of assign, so that the # df computation (and printing) can unambiguously decide which cols of # X are penalized and which are not when doing "terms" like actions. # To make some downstream things easier, order pcols and pattr to be # in the same relative order as the terms in 'assign' # ## can't compute assign attribute in R without terms ## if (missing(assign)) assign <- attr(x, 'assign')[-1] ##Remove 'intercept' pterms <- rep(0, length(assign)) names(pterms) <- names(assign) pindex <- rep(0, npenal) for (i in 1:npenal) { temp <- unlist(lapply(assign, function(x,y) (length(x) == length(y) && all(x==y)), pcols[[i]])) if (sparse[i]) pterms[temp] <- 2 else pterms[temp] <- 1 pindex[i] <- (seq(along.with=temp))[temp] } if ((sum(pterms==2) != sum(sparse)) || (sum(pterms>0) != npenal)) stop("pcols and assign arguments disagree") if (any(pindex != sort(pindex))) { temp <- order(pindex) pindex <- pindex[temp] pcols <- pcols[temp] pattr <- pattr[temp] } # ptype= 1 or 3 if a sparse term exists, 2 or 3 if a non-sparse exists ptype <- any(sparse) + 2*(any(!sparse)) ## Make sure these get defined f.expr1<-function(coef) NULL f.expr2<-function(coef) NULL if (any(sparse)) { sparse.attr <- (pattr[sparse])[[1]] #can't use [[sparse]] directly # if 'sparse' is a T/F vector fcol <- unlist(pcols[sparse]) if (length(fcol) > 1) stop("Sparse term must be single column") # Remove the sparse term from the X matrix xx <- x[, -fcol, drop=FALSE] for (i in 1:length(assign)){ j <- assign[[i]] if (j[1] > fcol) assign[[i]] <- j-1 } for (i in 1:npenal) { j <- pcols[[i]] if (j[1] > fcol) pcols[[i]] <- j-1 } frailx <- x[, fcol] frailx <- match(frailx, sort(unique(frailx))) nfrail <- max(frailx) nvar <- nvar - 1 #Set up the callback for the sparse frailty term pfun1 <- sparse.attr$pfun ### In R we use a function and eval() it, not an expression f.expr1 <- function(coef){ coxlist1$coef <- coef if (is.null(extra1)) temp <- pfun1(coef, theta1, n.eff) else temp <- pfun1(coef, theta1, n.eff, extra1) if (!is.null(temp$recenter)) coxlist1$coef <- coxlist1$coef - as.double(temp$recenter) if (!temp$flag) { coxlist1$first <- -as.double(temp$first) coxlist1$second <- as.double(temp$second) } coxlist1$penalty <- -as.double(temp$penalty) coxlist1$flag <- as.logical(temp$flag) if (any(sapply(coxlist1, length) != c(rep(nfrail,3), 1, 1))) stop("Incorrect length in coxlist1") coxlist1 } if (!is.null(getOption("survdebug"))) debug(f.expr1) coxlist1 <- list(coef=double(nfrail), first=double(nfrail), second=double(nfrail), penalty=0.0, flag=FALSE) ## we pass f.expr1 in as an argument in R ##.C("init_coxcall1", as.integer(sys.nframe()), expr1) } else { xx <- x frailx <- 0 nfrail <- 0 } # Now the non-sparse penalties if (sum(!sparse) >0) { full.imat <- !all(unlist(lapply(pattr, function(x) x$diag))) ipenal <- (1:length(pattr))[!sparse] #index for non-sparse terms f.expr2 <- function(coef){ coxlist2$coef<-coef ## pentot <- 0 for (i in ipenal) { pen.col <- pcols[[i]] coef <- coxlist2$coef[pen.col] if (is.null(extralist[[i]])) temp <- ((pattr[[i]])$pfun)(coef, thetalist[[i]],n.eff) else temp <- ((pattr[[i]])$pfun)(coef, thetalist[[i]], n.eff,extralist[[i]]) if (!is.null(temp$recenter)) coxlist2$coef[pen.col] <- coxlist2$coef[pen.col]- temp$recenter if (temp$flag) coxlist2$flag[pen.col] <- TRUE else { coxlist2$flag[pen.col] <- FALSE coxlist2$first[pen.col] <- -temp$first if (full.imat) { tmat <- matrix(coxlist2$second, nvar, nvar) tmat[pen.col,pen.col] <- temp$second coxlist2$second <- c(tmat) } else coxlist2$second[pen.col] <- temp$second } pentot <- pentot - temp$penalty } coxlist2$penalty <- as.double(pentot) if (any(sapply(coxlist2, length) != length2)) stop("Length error in coxlist2") coxlist2 } if (!is.null(getOption("survdebug"))) debug(f.expr2) if (full.imat) { coxlist2 <- list(coef=double(nvar), first=double(nvar), second= double(nvar*nvar), penalty=0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar*nvar, 1, nvar) } else { coxlist2 <- list(coef=double(nvar), first=double(nvar), second=double(nvar), penalty= 0.0, flag=rep(FALSE,nvar)) length2 <- c(nvar, nvar, nvar, 1, nvar) } ## in R, f.expr2 is passed as an argument later ##.C("init_coxcall2", as.integer(sys.nframe()), expr2) } else full.imat <- FALSE # # Set up initial values for the coefficients # If there are no sparse terms, finit is set to a vector of length 1 # rather than length 0, just to stop some "zero length" errors for # later statements where fcoef is saved (but not used) # if (nfrail >0) finit <- rep(0,nfrail) else finit <- 0 if (!missing(init) && !is.null(init)) { if (length(init) != nvar) { if (length(init) == (nvar+nfrail)) { finit <- init[-(1:nvar)] init <- init[1:nvar] } else stop("Wrong length for inital values") } } else init <- double(nvar) # # "Unpack" the passed in paramter list, # and make the initial call to each of the external routines # cfun <- lapply(pattr, function(x) x$cfun) parmlist <- lapply(pattr, function(x,eps) c(x$cparm, eps2=eps), sqrt(eps)) extralist<- lapply(pattr, function(x) x$pparm) iterlist <- vector('list', length(cfun)) thetalist <- vector('list', length(cfun)) printfun <- lapply(pattr, function(x) x$printfun) for (i in 1:length(cfun)) { temp <- (cfun[[i]])(parmlist[[i]], iter=0) if (sparse[i]) { theta1 <- temp$theta extra1 <- extralist[[i]] } thetalist[[i]] <- temp$theta iterlist[[i]] <- temp } # # Manufacture the list of calls to cfun, with appropriate arguments # ## Amazingly, all this works in R, so I don't need to understand it. ## temp1 <- c('x', 'coef', 'plik', 'loglik', 'status', 'neff', 'df', 'trH') temp2 <- c('frailx', 'coxfit$fcoef', 'loglik1', 'coxfit$loglik', 'status', 'n.eff') temp3 <- c('xx[,pen.col]', 'coxfit$coef[pen.col]','loglik1', 'coxfit$loglik', 'status', 'n.eff') calls <- vector('expression', length(cfun)) cargs <- lapply(pattr, function(x) x$cargs) for (i in 1:length(cfun)) { tempchar <- paste("(cfun[[", i, "]])(parmlist[[", i, "]], iter,", "iterlist[[", i, "]]") temp2b <- c(temp2, paste('pdf[', i, ']'), paste('trH[', i, ']')) temp3b <- c(temp3, paste('pdf[', i, ']'), paste('trH[', i, ']')) if (length(cargs[[i]])==0) calls[i] <- parse(text=paste(tempchar, ")")) else { temp <- match(cargs[[i]], temp1) if (any(is.na(temp))) stop(paste((cargs[[i]])[is.na(temp)], "not matched")) if (sparse[i]) temp4 <- paste(temp2b[temp], collapse=',') else temp4 <- paste(temp3b[temp], collapse=',') calls[i] <- parse(text=paste(paste(tempchar,temp4,sep=','),')')) } } need.df <- any(!is.na(match(c('df', 'trH'), unlist(cargs))))#do any use df? # # Last of the setup: create the vector of variable names # varnames <- dimnames(xx)[[2]] for (i in 1:length(cfun)) { if (!is.null(pattr[[i]]$varname)) varnames[pcols[[i]]] <- pattr[[i]]$varname } ## need the current environment for callbacks rho<-environment() # # Have C store the data, and get the loglik for beta=initial, frailty=0 # storage.mode(y) <- storage.mode(weights) <- "double" storage.mode(xx) <- storage.mode(offset) <- "double" if (andersen) coxfit <- .C(Cagfit5a, as.integer(n), as.integer(nvar), y, xx , offset, weights, newstrat, sorted, means= double(nvar), coef= as.double(init), u = double(nvar), loglik=double(1), as.integer(method=='efron'), as.integer(ptype), as.integer(full.imat), as.integer(nfrail), as.integer(frailx), #R callback additions f.expr1,f.expr2,rho) else coxfit <- .C(Ccoxfit5a, as.integer(n), as.integer(nvar), y, xx, offset, weights, newstrat, sorted, means= double(nvar), coef= as.double(init), u = double(nvar), loglik=double(1), as.integer(method=='efron'), as.integer(ptype), as.integer(full.imat), as.integer(nfrail), as.integer(frailx), f.expr1,f.expr2,rho) loglik0 <- coxfit$loglik means <- coxfit$means # # Now for the actual fit # iter2 <- 0 iterfail <- NULL thetasave <- unlist(thetalist) for (outer in 1:control$outer.max) { if (andersen) coxfit <- .C(Cagfit5b, iter=as.integer(control$iter.max), as.integer(n), as.integer(nvar), as.integer(newstrat), coef = as.double(init), u = double(nvar+nfrail), hmat = double(nvar*(nvar+nfrail)), hinv = double(nvar*(nvar+nfrail)), loglik = double(1), flag = integer(1), as.double(control$eps), as.double(control$toler.chol), as.integer(method=='efron'), as.integer(nfrail), fcoef = as.double(finit), fdiag = double(nfrail+nvar), f.expr1,f.expr2,rho) else coxfit <- .C(Ccoxfit5b, iter=as.integer(control$iter.max), as.integer(n), as.integer(nvar), as.integer(newstrat), coef = as.double(init), u = double(nvar+nfrail), hmat = double(nvar*(nvar+nfrail)), hinv = double(nvar*(nvar+nfrail)), loglik = double(1), flag = integer(1), as.double(control$eps), as.double(control$toler.chol), as.integer(method=='efron'), as.integer(nfrail), fcoef = as.double(finit), fdiag = double(nfrail+nvar), f.expr1,f.expr2,rho) iter <- outer iter2 <- iter2 + coxfit$iter if (coxfit$iter >=control$iter.max) iterfail <- c(iterfail, iter) # If any penalties were infinite, the C code has made fdiag=1 out # of self-preservation (0 divides). But such coefs are guarranteed # zero so the variance should be too.) temp <- rep(FALSE, nvar+nfrail) if (nfrail>0) temp[1:nfrail] <- coxlist1$flag if (ptype >1) temp[nfrail+ 1:nvar] <- coxlist2$flag fdiag <- ifelse(temp, 0, coxfit$fdiag) if (need.df) { #get the penalty portion of the second derive matrix if (nfrail>0) temp1 <- coxlist1$second else temp1 <- 0 if (ptype>1) temp2 <- coxlist2$second else temp2 <- 0 dftemp <-coxpenal.df(matrix(coxfit$hmat, ncol=nvar), matrix(coxfit$hinv, ncol=nvar), fdiag, assign, ptype, nvar, temp1, temp2, pindex[sparse]) df <- dftemp$df var <- dftemp$var var2 <- dftemp$var2 pdf <- df[pterms>0] # df's for penalized terms trH <- dftemp$trH[pterms>0] # trace H } if (nfrail >0) penalty <- -coxlist1$penalty else penalty <- 0 if (ptype >1) penalty <- penalty - coxlist2$penalty loglik1 <- coxfit$loglik + penalty #C code returns PL - penalty if (iter==1) penalty0 <- penalty # # Call the control function(s) # done <- TRUE for (i in 1:length(cfun)) { pen.col <- pcols[[i]] temp <- eval(calls[i]) if (sparse[i]) theta1 <- temp$theta thetalist[[i]] <- temp$theta iterlist[[i]] <- temp done <- done & temp$done } if (done) break # # Choose starting estimates for the next iteration # if (iter==1) { init <- coefsave <- coxfit$coef finit <- fsave <- coxfit$fcoef thetasave <- cbind(thetasave, unlist(thetalist)) } else { # the "as.vector" removes names, dodging a bug in Splus5.1 temp <- as.vector(unlist(thetalist)) coefsave <- cbind(coefsave, coxfit$coef) fsave <- cbind(fsave, coxfit$fcoef) # temp = next guess for theta # *save = prior thetas and the resultant fits # choose as initial values the result for the closest old theta howclose <- apply((thetasave-temp)^2,2, sum) which <- min((1:iter)[howclose==min(howclose)]) if (nvar>0) init <- coefsave[,which] if (nfrail>0) finit<- fsave[,which] thetasave <- cbind(thetasave, temp) } } if (nfrail >0) { lp <- offset + coxfit$fcoef[frailx] if (nvar >0) #sparse frailties and covariates lp <- lp + x[,-fcol,drop=FALSE] %*%coxfit$coef - sum(means*coxfit$coef) } else lp <- offset + as.vector(x%*%coxfit$coef) - sum(means*coxfit$coef) # release the memory if (andersen) { .C(Cagfit5c, as.integer(nvar)) #release the memory resid <- .Call(Cagmart3, y, exp(lp), weights, newstrat, sorted, as.integer(method=='efron')) } else { expect <- .C(Ccoxfit5c, as.integer(n), as.integer(nvar), as.integer(newstrat), as.integer(method=='efron'), expect= double(n))$expect resid <- status - expect } names(resid) <- rownames if (!need.df) { #didn't need it iteration by iteration, but do it now #get the penalty portion of the second derive matrix if (nfrail>0) temp1 <- coxlist1$second else temp1 <- 0 if (ptype>1) temp2 <- coxlist2$second else temp2 <- 0 dftemp <-coxpenal.df(matrix(coxfit$hmat,ncol=nvar), matrix(coxfit$hinv,ncol=nvar), fdiag, assign, ptype, nvar, temp1, temp2, pindex[sparse]) df <- dftemp$df trH <- dftemp$trH var <- dftemp$var var2 <- dftemp$var2 } if (control$iter.max >1 && length(iterfail)>0) warning(paste("Inner loop failed to coverge for iterations", paste(iterfail, collapse=' '))) which.sing <- (fdiag[nfrail + 1:nvar] ==0) coef <- coxfit$coef names(coef) <- varnames coef[which.sing] <- NA names(iterlist) <- names(pterms[pterms>0]) if (nfrail >0) { if (nvar >0) { #sparse frailties and covariates list(coefficients = coef, var = var, var2 = var2, loglik = c(loglik0, loglik1), iter = c(iter, iter2), linear.predictors = as.vector(lp), residuals = resid, means = means, concordance= survConcordance.fit(y, lp, strata, weights), method= c('coxph.penal', 'coxph'), frail = coxfit$fcoef, fvar = dftemp$fvar, df = df, df2=dftemp$df2, penalty= c(penalty0, penalty), pterms = pterms, assign2=assign, history= iterlist, coxlist1=coxlist1, printfun=printfun) } else { #sparse frailties only list( loglik = c(loglik0, loglik1), iter = c(iter, iter2), linear.predictors = as.vector(lp), residuals = resid, means = means, concordance= survConcordance.fit(y, lp, strata, weights), method= c('coxph.penal', 'coxph'), frail = coxfit$fcoef, fvar = dftemp$fvar, df = df, df2=dftemp$df2, penalty = c(penalty0, penalty), pterms = pterms, assign2=assign, history= iterlist, printfun=printfun) } } else { #no sparse terms list(coefficients = coef, var = var, var2 = var2, loglik = c(loglik0, loglik1), iter = c(iter, iter2), linear.predictors = lp, residuals = resid, means = means, concordance= survConcordance.fit(y, lp, strata, weights), method= c('coxph.penal', 'coxph'), df = df, df2=dftemp$df2, penalty= c(penalty0, penalty), pterms = pterms, assign2=assign, history= iterlist, coxlist2=coxlist2, printfun= printfun) } } survival/R/anova.coxph.penal.R0000644000175100001440000001163212534312206016012 0ustar hornikusers# The first section of this is identical to anova.coxph anova.coxph.penal <- function (object, ..., test = 'Chisq') { if (!inherits(object, "coxph")) stop ("argument must be a cox model") # All the ... args need to be coxph or coxme fits. If any of them # have a name attached, e.g., 'charlie=T' we assume a priori # that they are illegal # dotargs <- list(...) named <- if (is.null(names(dotargs))) rep(FALSE, length(dotargs)) else (names(dotargs) != "") if (any(named)) warning(paste("The following arguments to anova.coxph(..)", "are invalid and dropped:", paste(deparse(dotargs[named]), collapse = ", "))) dotargs <- dotargs[!named] if (length(dotargs) >0) { # Check that they are all cox or coxme models is.coxmodel <-unlist(lapply(dotargs, function(x) inherits(x, "coxph"))) is.coxme <- unlist(lapply(dotargs, function(x) inherits(x, "coxme"))) if (!all(is.coxmodel | is.coxme)) stop("All arguments must be Cox models") if (any(is.coxme)) { # We need the anova.coxmelist function from coxme # If coxme is not loaded the line below returns NULL temp <- getS3method("anova", "coxmelist", optional=TRUE) if (is.null(temp)) stop("a coxme model was found and library coxme is not loaded") else return(temp(c(list(object), dotargs), test = test)) } else return(anova.coxphlist(c(list(object), dotargs), test = test)) } # # The argument is a single Cox model # Show the nested list of models generated by this model. # By tradition the sequence is main effects (in the order found in # the model statement), then 2 way interactions, then 3, etc. # One does this by using the "assign" attribute of the model matrix. # This does not work for penalized terms, however, so we use a mixed # strategy. The penalized terms do not participate in interactions # (which are the terms for which assign is really handy). Use # the model frame for the penalized terms, and assign for all the # others. if (length(object$rscore)>0) stop("Can't do anova tables with robust variances") has.strata <- !is.null(attr(terms(object), "specials")$strata) # The following line causes pspline terms to be re-evaluated correctly # The predvars attr for them does not retrieve the correct penalty attr(object$terms, "predvars") <- NULL mf <- model.frame(object) # we must have the model frame Y <- model.response(mf) X <- model.matrix(object, mf) assign <- attr(X, 'assign') if (has.strata) { stemp <- untangle.specials(terms(object), "strata") if (length(stemp$vars)==1) strata.keep <- mf[[stemp$vars]] else strata.keep <- strata(mf[,stemp$vars], shortlabel=TRUE) strats <- as.numeric(strata.keep) } pname <- names(object$pterms)[object$pterms >0] pindex <- match(pname, attr(terms(object), "term.labels")) alevels <- sort(unique(assign)) #if there are strata the sequence has holes nmodel <- length(alevels) df <- integer(nmodel+1) #this will hold the df vector loglik <- double(nmodel+1) #and the loglike vector df[nmodel+1] <- if (is.null(object$df)) sum(!is.na(object$coefficients)) else sum(object$df) loglik[nmodel+1] <- object$loglik[2] df[1] <- 0 loglik[1] <- object$loglik[1] # Now refit intermediate models assign2 <- assign[!(assign %in% pindex)] pform <- paste("mf[['", pname, "']]", sep='') for (i in seq.int(1, length.out=nmodel-1)){ j <- assign2[assign2 <= alevels[i]] if (length(j)) form <- "Y ~ X[,j]" else form <- "Y ~" form <- paste(c(form, pform[pindex <= i]), collapse=" +") if (length(object$offset)) form <- paste(form, " + offset(object$offset") if (has.strata) form <- paste(form, " + strata(strats)") tfit <- coxph(as.formula(form)) df[i+1] <- if (!is.null(tfit$df)) sum(tfit$df) else sum(!is.na(tfit$coefficients)) loglik[i+1] <- tfit$loglik[2] } table <- data.frame(loglik=loglik, Chisq=c(NA, 2*diff(loglik)), Df=c(NA, diff(df))) if (nmodel == 0) #failsafe for a NULL model table <- table[1, , drop = FALSE] if (length(test) >0 && test[1]=='Chisq') { table[['Pr(>|Chi|)']] <- 1- pchisq(table$Chisq, table$Df) } row.names(table) <- c('NULL', attr(terms(object), "term.labels")) title <- paste("Analysis of Deviance Table\n Cox model: response is ", deparse(object$terms[[2]]), "\nTerms added sequentially (first to last)\n", sep = "") structure(table, heading = title, class = c("anova", "data.frame")) } survival/R/residuals.coxph.penal.S0000644000175100001440000000236111755254054016713 0ustar hornikusers# $Id: residuals.coxph.penal.S 11516 2012-04-24 12:49:14Z therneau $ residuals.coxph.penal <- function(object, type=c("martingale", "deviance", "score", "schoenfeld", "dfbeta", "dfbetas", "scaledsch","partial"), collapse=FALSE, weighted=FALSE, ...) { type <- match.arg(type) # Are there any sparse terms, and if so do I need the X matrix? if (any(object$pterms==2) && !(type=='martingale' || type=='deviance')){ # treat the sparse term as an offset term # It gets picked up in the linear predictor, so all I need to # do is "X" it out of the model so that it doesn't get picked up # as a part of the X matrix and etc. # I know that the sparse term is a single column BTW # sparsename <- (names(object$pterms))[object$pterms==2] x <- object[['x']] #don't accidentally get object$xlevels if (is.null(x)) { temp <- coxph.getdata(object, y=TRUE, x=TRUE, stratax=TRUE) if (is.null(object$y)) object$y <- temp$y if (is.null(object$strata)) object$strata <- temp$strata x <- temp$x } object$x <- x[, -match(sparsename, dimnames(x)[[2]]), drop=FALSE] temp <- attr(object$terms, 'term.labels') object$terms <- object$terms[-match(sparsename, temp)] } NextMethod('residuals') } survival/R/attrassign.R0000644000175100001440000000252412257335007014655 0ustar hornikusers# $Id$ # When X is a model matrix, Splus and R have a different format # for the "assign" attribute # For instance # survreg(Surv(time, status) ~ age + sex + factor(ph.ecog), lung) # R gives the compact form, a vector (0, 1, 2, 3, 3, 3); which can be # read as "the first column of the X matrix (intercept) goes with none of # the terms', 'the second column goes with term 1', etc. # Splus gives a list # $(Intercept) 1 # $age 2 # $sex 3 # $factor(ph.ecog) 4 5 6 # # This function creates the Splus style of output from the R style. Several # of the routines in the package use this, as it is somewhat easier (more # transparent) to work with. # attrassign<-function (object, ...) UseMethod("attrassign") attrassign.lm<-function(object, ...){ attrassign(model.matrix(object), terms(object))} attrassign.default<-function(object, tt, ...){ if (!inherits(tt,"terms")) stop("need terms object") aa<-attr(object,"assign") if (is.null(aa)) stop("argument is not really a model matrix") ll<-attr(tt,"term.labels") temp <- c("(Intercept)", ll)[aa+1] #vector of term names # Don't put them in alphabetical order, retain the order we inherited split(seq(along=temp), factor(temp, levels=unique(temp))) } survival/vignettes/0000755000175100001440000000000012545056257014166 5ustar hornikuserssurvival/vignettes/adjcurve.Rnw0000644000175100001440000014753112533657035016472 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \SweaveOpts{keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{prefix.string=adjcurve,width=6,height=4} \setkeys{Gin}{width=\textwidth} %\VignetteIndexEntry{Adjusted Survival Curves} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text require(survival, quietly=TRUE) fdata <- flchain[flchain$futime > 7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) @ \title{Adjusted Survival Curves} \author{Terry M Therneau, Cynthia S Crowson, Elizabeth J Atkinson} \date{Jan 2015} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {adjcurve-#1.pdf}} \begin{document} \maketitle \section{Introduction} Suppose we want to investigate to what extent some factor influences survival, as an example we might compare the experience of diabetic patients who are using metformin versus those on injected insulin as their primary treatment modality. There is some evidence that metformin has a positive influence, particularly in cancers, but the ascertainment is confounded by the fact that it is a first line therapy: the patients on metformin will on average be younger and have had a diabetes diagnosis for a shorter amount of time than those using insulin. ``Young people live longer'' is not a particularly novel observation. The ideal way to test this is with a controlled clinical trial. This is of course not always possible, and assessments using available data that includes and adjusts for such confounders is also needed. There is extensive literature --- and debate --- on this topic in the areas of modeling and testing. The subtopic of how to create honest survival curve estimates in the presence of confounders is less well known, and is the focus of this note. Assume that we have an effect of interest, treatment say, and a set of possible confounding variables. Creation a pair of adjusted survival curves has two parts: definition of a reference population for the confounders, and then the computation of estimated curves for that population. There are important choices in both steps. The first, definition of a target, is often not explicitly stated but can be critical. If an outcome differs with age, myocardial infarction say, and two treatments also had age dependent efficacy, then the comparison will depend greatly on whether we are talking about a population of young, middle aged, or older subjects. The computational step has two main approaches. The first, sometimes known as \emph{marginal} analysis, first reweights the data such that each subgroup's weighted distribution matches that of our population target. An immediate consequence is that all subgroups will be balanced with respect to the confounding variables. We can then proceed with a simple analysis of survival using the reformulated data, ignoring the confounders. The second approach seeks to understand and model the effect of each confounder, with this we can then correct for them. From a comprehensive overall model we can obtain predicted survival curves for any configuration of variables, and from these get predicted overall curves for the reference population. This is often called the \emph{conditional} approach since we are using conditional survival curves given covariates $x$. A third but more minor choice is division of the covariates $x$ into effects of interest vs. confounders. For instance, we might want to see separate curves for two treatments, each adjusted for age and sex. The reference population will describe the age and sex distribution. For simplicity we will use $x$ to describe all the confounding variables and use $c$ for the control variable(s), e.g. treatment. The set $c$ might be empty, producing a single overall curve, but this is the uncommon case. As shown below, our two methods differ essentially in the \emph{order} in which the two necessary operations are done, balancing and survival curve creation. \begin{center} \begin{tabular}{rccc} Marginal: & balance data on $x$ & $\longrightarrow$ & form survival curves for each $c$\\ Conditional: & predicted curves for $\{x,c\}$ subset & $\longrightarrow$ & average the predictions for each $c$ \end{tabular} \end{center} We can think of them as ``balance and then model'' versus ``model then balance''. An analysis might use a combinations of these, of course, balancing on some factors and modeling others. All analyses are marginal analyses with respect to important predictors that are unknown to us, although in that case we have no assurance of balance on those factors. \begin{figure}[tb] \myfig{flc1} \caption{Survival of \Sexpr{nrow(flchain)} residents of Olmsted County, broken into three cohorts based on FLC value.} \label{flc1} \end{figure} \section{Free Light Chain} Our example data set for this comparison uses a particular assay of plasma immunoglobulins and is based on work of Dr Angela Dispenzieri and her colleagues at the Mayo Clinic \cite{Dispenzieri12}. In brief: plasma cells (PC) are responsible for the production of immunoglobulins, but PC comprise only a small portion ($<1$\%) of the total blood and marrow hematapoetic cell population in normal patients. The normal human repertoire is estimated to contain over $10^{8}$ unique immunoglobulins, conferring a broad range of immune protection. In multiple myeloma, the most common form of plasma cell malignancy, almost all of the circulating antigen will be identical, the product of a single malignant clone. An electrophoresis examination of circulating immunoglobulins will exhibit a ``spike'' corresponding to this unique molecule. This anomaly is used both as a diagnostic method and in monitoring the course of the disease under treatment. The presence of a similar, albeit much smaller, spike in normal patients has been a long term research interest of the Mayo Clinic hematology research group \cite{Kyle93}. In 1995 Dr Robert Kyle undertook a population based study of this, and collected serum samples on 19,261 of the 24,539 residents of Olmsted County, Minnesota, aged 50 years or more \cite{Kyle06}. In 2010 Dr. Angela Dispenzieri assayed a sub fraction of the immunoglobulins, the free light chain (FLC), on 15,748 of these subjects who had sufficient remaining sera from the original sample collection. All studies took place under the oversight of the appropriate Institutional Review Boards, which ensure rigorous safety and ethical standards in research. A subset of the Dispenzieri study is available in the survival package as data set \texttt{flchain}. Because the original study assayed nearly the entire population, there is concern that some portions of the anonymized data could be linked to actual subjects by a diligent searcher, and so only a subset of the study has been made available as a measure to strengthen anonymity. It was randomly selected from the whole within sex and age group strata so as to preserve the age/sex structure. The data set contains 3 subjects whose blood sample was obtained on the day of their death. It is rather odd to think of a sample obtained on the final day as ``predicting'' death, or indeed for any results obtained during a patient's final mortality cascade. There are also a few patients with no follow-up beyond the clinic visit at which the assay occurred. We have chosen in this analysis to exclude the handful of subjects with less than 7 days of follow-up, leaving \Sexpr{nrow(fdata)} observations. \begin{table} \centering \begin{tabular}{l|cccc} & 50--59 & 60--69 & 70--79 & 80+ \\ \hline <>= group3 <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) age1 <- cut(fdata$age, c(49,59,69,79, 110)) levels(age1) <- c(paste(c(50,60,70), c(59,69,79), sep='-'), '80+') temp1 <- table(group3, age1) temp2 <- round(100* temp1/rowSums(temp1)) pfun <- function(x,y) { paste(ifelse(x<1000, "\\phantom{0}", ""), x, " (", ifelse(y<10, "\\phantom{0}", ""), y, ") ", sep="") } cat(paste(c("FLC $<$ 3.38", pfun(temp1[1,], temp2[1,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC 3.38--4.71", pfun(temp1[2,], temp2[2,])), collapse=" & "), "\\\\\n") cat(paste(c("FLC $>$ 4.71", pfun(temp1[3,], temp2[3,])), collapse=" & "), "\n") @ \end{tabular} \caption{Comparison of the age distributions (percents) for each of the three groups.} \label{tflc1} \end{table} Figure \ref{flc1} shows the survival curves for three subgroups of the patients: those whose total free light chain (FLC) is in the upper 10\% of all values found in the full study, those in the 70--89th percentile, and the remainder. There is a clear survival effect. Average free light chain amounts rise with age, however, at least in part because it is eliminated through the kidneys and renal function declines with age. Table \ref{tflc1} shows the age distribution for each of the three groups. In the highest decile of FLC (group 3) over half the subjects are age 70 or older compared to only 23\% in those below the 70th percentile. How much of the survival difference is truly associated with FLC and how much is simply an artifact of age? (The cut points are arbitrary, but we have chosen to mimic the original study and retain them. Division into three groups is a convenient number to illustrate the methods in this vignette, but we do not make any claim that such a categorization is optimal or even sensible statistical practice.) The R code for figure 1 is shown below. <>= fdata <- flchain[flchain$futime >=7,] fdata$age2 <- cut(fdata$age, c(0,54, 59,64, 69,74,79, 89, 110), labels = c(paste(c(50,55,60,65,70,75,80), c(54,59,64,69,74,79,89), sep='-'), "90+")) fdata$group <- factor(1+ 1*(fdata$flc.grp >7) + 1*(fdata$flc.grp >9), levels=1:3, labels=c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71")) sfit1 <- survfit(Surv(futime, death) ~ group, fdata) plot(sfit1, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") text(c(11.1, 10.5, 7.5), c(.88, .57, .4), c("FLC < 3.38", "3.38 - 4.71", "FLC > 4.71"), col=c(1,2,4)) @ \section{Reference populations} There are a few populations that are commonly used as the reference group. \begin{enumerate} \item Empirical. The overall distribution of confounders $x$ in the data set as a whole. For this study we would use the observed age/sex distribution, ignoring FLC group. This is also called the ``sample'' or ``data'' distribution. \item External reference. The distribution from some external study or standard. \item Internal reference. A particular subset of the data is chosen as the reference, and other subsets are then aligned with it. \end{enumerate} Method 2 is common in epidemiology, using a reference population based on a large external population such as the age/sex distribution of the 2000 United States census. Method 3 most often arises in the case-control setting, where one group is small and precious (a rare disease say) and the other group (the controls) from which we can sample is much larger. In each case the final result of the computation can be thought of as the expected answer we ``would obtain'' in a study that was perfectly balanced with respect to the list of confounders $x$. Population 1 is the most frequent. \section{Marginal approach} \begin{table} \centering \begin{tabular}{crrrrrrrr} \multicolumn{3}{c}{Females} \\ & \multicolumn{8}{c}{Age} \\ FLC group & 50--54& 55--59& 60--64 & 65--69 & 70--74 & 75--79 & 80--89& 90+ \\ \hline <>= tab1 <- with(fdata, table(group, age2, sex)) cat("Low&", paste(tab1[1,,1], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,1], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,1], collapse=" &"), "\\\\\n") @ \\ \multicolumn{3}{c}{Males} \\ % & 50--54& 55--59& 60--64 & 65--69 & 70--74 & 75--79 & 80--89& 90+ \\ \hline <>= cat("Low&", paste(tab1[1,,2], collapse=" &"), "\\\\\n") cat("Med&", paste(tab1[2,,2], collapse=" &"), "\\\\\n") cat("High&", paste(tab1[3,,2], collapse=" &"), "\n") @ \end{tabular} \caption{Detailed age and sex distribution for the study population} \label{tab2} \end{table} \subsection{Selection} One approach for balancing is to select a subset of the data such that its distribution matches the referent for each level of $c$, i.e., for each curve that we wish to obtain. As an example we take a case-control like approach to the FLC data, with FLC high as the ``cases'' since it is the smallest group. Table \ref{tab2} shows a detailed distribution of the data with respect to age and sex. The balanced subset has all \Sexpr{tab1[3,1,1]} females aged 50--54 from the high FLC group, a random sample of \Sexpr{tab1[3,1,1]} out of the \Sexpr{tab1[1,1,1]} females in the age 50--54 low FLC group, and \Sexpr{tab1[3,1,1]} out of \Sexpr{tab1[2,1,1]} for the middle FLC. Continue this for all age/sex subsets. We cannot \emph{quite} compute a true case-control estimate for this data since there are not enough ``controls'' in the female 90+ category to be able to select one unique control for each case, and likewise in the male 80-89 and 90+ age groups. To get around this we will sample with replacement in these strata. \begin{figure}[tb] \myfig{flc2} \caption{Survival curves from a case-control sample are shown as solid lines, dashed lines are curves for the unweighted data set (as found in figure \ref{flc1}).} \label{flc2} \end{figure} <>= temp <- with(fdata, table(group, age2, sex)) dd <- dim(temp) # Select subjects set.seed(1978) select <- array(vector('list', length=prod(dd)), dim=dd) for (j in 1:dd[2]) { for (k in 1:dd[3]) { n <- temp[3,j,k] # how many to select for (i in 1:2) { indx <- which(as.numeric(fdata$group)==i & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[i,j,k] <- list(sample(indx, n, replace=(n> temp[i,j,k]))) } indx <- which(as.numeric(fdata$group)==3 & as.numeric(fdata$age2) ==j & as.numeric(fdata$sex) ==k) select[3,j,k] <- list(indx) #keep all the group 3 = high } } data2 <- fdata[unlist(select),] sfit2 <- survfit(Surv(futime, death) ~ group, data2) plot(sfit2, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) @ %\begin{table}[tb] \centering % \begin{tabular}{ccccccc} % &\multicolumn{2}{c}{FLC low} & \multicolumn{2}{c}{FLC med}& % \multicolumn{2}{c}{FLC high} \\ % & Total & Subset & Total & Subset & Total & Subset \\ \hline %<>= %tab3 <- with(fdata, table(age2, group)) %tab3 <- round(100*scale(tab3, center=F, scale=colSums(tab3))) %tab4 <- with(data2, table(age2, group)) %tab4 <- round(100*scale(tab4, center=F, scale=colSums(tab4))) %tab5 <- cbind(tab3[,1], tab4[,1], tab3[,2], tab4[,2], tab3[,3], tab4[,3]) %pfun <- function(x) paste(ifelse(x<10, paste("\\phantom{0}", x), x), % collapse=" &") %dtemp <- dimnames(tab5)[[1]] %for (j in 1:7) % cat(dtemp[j], " &", pfun(tab5[j,]), "\\\\\n") %cat(dtemp[8], " & ", pfun(tab5[8,]), "\n") %@ %\end{tabular} %\caption{Age distributions (\%) of the original data set along with that of % the subset, for the three FLC groups.} %\label{tflc2} %\end{table} The survival curves for the subset data are shown in figure \ref{flc2}. The curve for the high risk group is unchanged, since by definition all of those subjects were retained. We see that adjustment for age and sex has reduced the apparent survival difference between the groups by about half, but a clinically important effect for high FLC values remains. The curve for group 1 has moved more than that for group 2 since the age/sex adjustment is more severe for that group. <>= # I can't seem to put this all into an Sexpr z1 <- with(fdata,table(age, sex, group)) z2<- apply(z1, 1:2, min) ztemp <- 3*sum(z2) z1b <- with(fdata, table(age>64, sex, group)) ztemp2 <- sum(apply(z1b, 1:2, min)) @ In actual practice, case-control designs arise when matching and selection can occur \emph{before} data collection, leading to a substantial decrease in the amount of data that needs to be gathered and a consequent cost or time savings. When a data set is already in hand it has two major disadvantages. The first is that the approach wastes data; throwing away information in order to achieve balance is always a bad idea. Second is that though it returns an unbiased comparison, the result is for a very odd reference population. One advantage of matched subsets is that standard variance calculations for the curves are correct; the values provided by the usual Kaplan-Meier program need no further processing. We can also use the usual statistical tests to check for differences between the curves. <<>>= survdiff(Surv(futime, death) ~ group, data=data2) @ \subsection{Reweighting} \label{sect:logistic} A more natural way to adjust the data distribution is by weighting. Let $\pi(a,s)$, $a$ = age group, $s$ = sex be a target population age/sex distribution for our graph, and $p(a,s,i)$ the observed probability of each age/sex/group combination in the data. Both $\pi$ and $p$ sum to 1. Then if each observation in the data set is given a case weight of \begin{equation} w_{asi} = \frac{\pi(a,s)}{p(a,s,i)} \label{wt1} \end{equation} the weighted age/sex distribution for each of the groups will equal the target distribution $\pi$. An obvious advantage of this approach is that the resulting curves represent a tangible and well defined group. As an example, we will first adjust our curves to match the age/sex distribution of the 2000 US population, a common reference target in epidemiology studies. The \texttt{uspop2} data set is found in later releases of the survival package in R. It is an array of counts with dimensions of age, sex, and calendar year. We only want ages of 50 and over, and the population data set has collapsed ages of 100 and over into a single category. We create a table \texttt{tab100} of observed age/sex counts within group for our own data, using the same upper age threshold. New weights are the values $\pi/p$ = \texttt{pi.us/tab100}. <<>>= refpop <- uspop2[as.character(50:100),c("female", "male"), "2000"] pi.us <- refpop/sum(refpop) age100 <- factor(ifelse(fdata$age >100, 100, fdata$age), levels=50:100) tab100 <- with(fdata, table(age100, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab100 #new weights by age,sex, group range(us.wt) @ There are infinite weights! This is because the US population has coverage at all ages, but our data set does not have representatives in every age/sex/FLC group combination; there are for instance no 95 year old males in in the data set. Let us repeat the process, collapsing the US population from single years into the 8 age groups used previously in table \ref{tab2}. Merging the per age/sex/group weights found in the 3-dimensional array \texttt{us.wt} into the data set as per-subject weights uses matrix subscripts, a useful but less known feature of R. <<>>= temp <- as.numeric(cut(50:100, c(49, 54, 59, 64, 69, 74, 79, 89, 110)+.5)) pi.us<- tapply(refpop, list(temp[row(refpop)], col(refpop)), sum)/sum(refpop) tab2 <- with(fdata, table(age2, sex, group))/ nrow(fdata) us.wt <- rep(pi.us, 3)/ tab2 range(us.wt) index <- with(fdata, cbind(as.numeric(age2), as.numeric(sex), as.numeric(group))) fdata$uswt <- us.wt[index] sfit3a <-survfit(Surv(futime, death) ~ group, data=fdata, weight=uswt) @ \begin{figure}[tb] \myfig{flc3a} \caption{Population totals for the US reference (red) and for the observed data set (black).} \label{flc3a} \end{figure} A more common choice is to use the overall age/sex distribution of the sample itself as our target distribution $\pi$, i.e., the empirical distribution. Since FLC data set is population based and has excellent coverage of the county, this will not differ greatly from the US population in this case, as is displayed in figure \ref{flc3a}. <>= tab1 <- with(fdata, table(age2, sex))/ nrow(fdata) matplot(1:8, cbind(pi.us, tab1), pch="fmfm", col=c(2,2,1,1), xlab="Age group", ylab="Fraction of population", xaxt='n') axis(1, 1:8, levels(fdata$age2)) tab2 <- with(fdata, table(age2, sex, group))/nrow(fdata) tab3 <- with(fdata, table(group)) / nrow(fdata) rwt <- rep(tab1,3)/tab2 fdata$rwt <- rwt[index] # add per subject weights to the data set sfit3 <- survfit(Surv(futime, death) ~ group, data=fdata, weight=rwt) temp <- rwt[,1,] #show female data temp <- temp %*% diag(1/apply(temp,2,min)) round(temp, 1) #show female data @ \begin{figure}[tb] \myfig{flc3} \caption{Survival curves for the three groups using reweighted data are shown with solid lines, the original unweighted analysis as dashed lines. The heavier solid line adjusts to the Olmsted population and the lighter one to the US population.} \label{flc3} \end{figure} <>= plot(sfit3, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3a, mark.time=F, col=c(1,2,4), lty=1, lwd=1, xscale=365.25) lines(sfit1, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, levels(fdata$group), lty=1, col=c(1,2,4), bty='n', lwd=2) @ The calculation of weights is shown above, and finishes with a table of the weights for the females. The table was scaled so as to have a minimum weight of 1 in each column for simpler reading. We see that for the low FLC group there are larger weights for the older ages, whereas the high FLC group requires substantial weights for the youngest ages in order to achieve balance. The resulting survival curve is shown in figure \ref{flc3}. The distance between the adjusted curves is similar to the results from subset selection, which is as expected since both approaches are correcting for the same bias, but results are now for an overall population distribution that matches Olmsted County. The curves estimate what the results would have looked like, had each of the FLC groups contained the full distribution of ages. Estimation based on reweighted data is a common theme in survey sampling. Correct standard errors for the curves are readily computed using methods from that literature, and are available in some software packages. In R the \texttt{svykm} routine in the \texttt{survey} package handles both this simple case and more complex sampling schemes. Tests of the curves can be done using a weighted Cox model; the robust variance produced by \texttt{coxph} is identical to the standard Horvitz-Thompsen variance estimate used in survey sampling \cite{Binder92}. The robust score test from \texttt{coxph} corresponds to a log-rank test corrected for weighting. (In the example below the svykm function is only run if the survey package is already loaded, as the variance calculation is very slow for this large data set.) <<>>= id <- 1:nrow(fdata) cfit <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=rwt) summary(cfit)$robscore if (exists("svykm")) { #true if the survey package is loaded sdes <- svydesign(id = ~0, weights=~rwt, data=fdata) dfit <- svykm(Surv(futime, death) ~ group, design=sdes, se=TRUE) } @ Note: including the \texttt{cluster} term in the coxph call causes it to treat the weights as resampling values and thus use the proper survey sampling style variance. The default without that term would be to treat the case weights as replication counts. This same alternate variance estimate is also called for when there are correlated observations; many users will be more familiar with the cluster statement in that context. \paragraph{Inverse probability weighting} Notice that when using the overall population as the target distribution $\pi$ we can use Bayes rule to rewrite the weights as \begin{align*} \frac{1}{w_{asi}} &= \frac{{\rm Pr}({\rm age}=a, {\rm sex} =s, {\rm group}=i)} {{\rm Pr}({\rm age}=a, {\rm sex} =s)} \\ &= {\rm Pr}({\rm group}=i | {\rm age}=a, {\rm sex} =s) \end{align*} This last is precisely the probability estimated by a logistic regression model, leading to \emph{inverse probability weighting} as an alternate label for this approach. We can reproduce the weights calculated just above with three logistic regression models. <>= options(na.action="na.exclude") gg <- as.numeric(fdata$group) lfit1 <- glm(I(gg==1) ~ factor(age2) * sex, data=fdata, family="binomial") lfit2 <- glm(I(gg==2) ~ factor(age2) * sex, data=fdata, family="binomial") lfit3 <- glm(I(gg==3) ~ factor(age2) * sex, data=fdata, family="binomial") temp <- ifelse(gg==1, predict(lfit1, type='response'), ifelse(gg==2, predict(lfit2, type='response'), predict(lfit3, type='response'))) all.equal(1/temp, fdata$rwt) @ If there were only 2 groups then only a single regression model is needed since P(group 2) = 1 - P(group 1). Note the setting of na.action, which causes the predicted vector to have the same length as the original data even when there are missing values. This simplifies merging the derived weights with the original data set. An advantage of the regression framework is that one can easily accommodate more variables by using a model with additive terms and only a few selected interactions, and the model can contain continuous as well as categorical predictors. The disadvantage is that such models are often used without the necessary work to check their validity. For instance models with \texttt{age + sex} could have been used above. This makes the assumption that the odds of being a member of group 1 is linear in age and with the same slope for males and females; ditto for the models for group 2 and group 3. How well does this work? Since the goal of reweighting is to standardize the ages, a reasonable check is to compute and plot the reweighted age distribution for each flc group. \begin{figure}[tb] \myfig{flc4} \caption{The re-weighted age distribution using logistic regression with continuous age, for females, FLC groups 1--3. The target distribution is shown as a ``+''. The original unadjusted distribution is shown as dashed lines.} \label{flc4} \end{figure} Figure \ref{flc4} shows the result. The reweighted age distribution is not perfectly balanced, i.e., the `1', `2' and `3' symbols do no exactly overlay one another, but in this case the simple linear model has done an excellent job. We emphasize that whenever the reweighting is based on a simplified model then such a check is obligatory. It is quite common that a simple model is not sufficient and the resulting weight adjustment is inadequate. Like a duct tape auto repair, proceeding forward as though the underlying problem has been addressed is then most unwise. <>= lfit1b <-glm(I(gg==1) ~ age + sex, data=fdata, family="binomial") lfit2b <- glm(I(gg==2) ~ age +sex, data=fdata, family="binomial") lfit3b <- glm(I(gg==3) ~ age + sex, data=fdata, family="binomial") # weights for each group using simple logistic twt <- ifelse(gg==1, 1/predict(lfit1b, type="response"), ifelse(gg==2, 1/predict(lfit2b, type="response"), 1/predict(lfit3b, type="response"))) tdata <- data.frame(fdata, lwt=twt) #grouped plot for the females temp <- tdata[tdata$sex=='F',] temp$gg <- as.numeric(temp$group) c1 <- with(temp[temp$gg==1,], tapply(lwt, age2, sum)) c2 <- with(temp[temp$gg==2,], tapply(lwt, age2, sum)) c3 <- with(temp[temp$gg==3,], tapply(lwt, age2, sum)) xtemp <- outer(1:8, c(-.1, 0, .1), "+") #avoid overplotting ytemp <- 100* cbind(c1/sum(c1), c2/sum(c2), c3/sum(c3)) matplot(xtemp, ytemp, col=c(1,2,4), xlab="Age group", ylab="Weighted frequency (%)", xaxt='n') ztab <- table(fdata$age2) points(1:8, 100*ztab/sum(ztab), pch='+', cex=1.5, lty=2) # Add the unadjusted temp <- tab2[,1,] temp <- scale(temp, center=F, scale=colSums(temp)) matlines(1:8, 100*temp, pch='o', col=c(1,2,4), lty=2) axis(1, 1:8, levels(fdata$age2)) @ \paragraph{Rescaled weights} As the weights were defined in equation \ref{wt1}, the sum of weights for each of the groups is \Sexpr{nrow(fdata)}, the number of observations in the data set. Since the number of subjects in group 3 is one seventh of that in group 1, the average weight in group 3 is much larger. An alternative is to define weights in terms of the \emph{within} group distribution rather than the overall distribution, leading to the rescaled weights $w^*$ \begin{align} w^* &= \frac{\pi(a,s)}{p(a,s|i)} \label{wt2} \\ &= \frac{{\rm P}({\rm group}=i)} {{\rm P}({\rm group}=i | {\rm age}=a, {\rm sex}=s)} \label{wt2b} \end{align} Each group's weights are rescaled by the overall prevalence of the group. In its simplest form, the weights in each group are scaled to add up to the number of subjects in the group. <<>>= # compute new weights wtscale <- table(fdata$group)/ tapply(fdata$rwt, fdata$group, sum) wt2 <- c(fdata$rwt * wtscale[fdata$group]) c("rescaled cv"= sd(wt2)/mean(wt2), "rwt cv"=sd(fdata$rwt)/mean(fdata$rwt)) cfit2a <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight= rwt) cfit2b <- coxph(Surv(futime, death) ~ group + cluster(id), data=fdata, weight=wt2) round(c(cfit2a$rscore, cfit2b$rscore),1) @ The rescaling results in weights that are much less variable across groups. This operation has no impact on the individual survival curves or their standard errors, since within group we have multiplied all weights by a constant. When comparing curves across groups, however, the rescaled weights reduce the standard error of the test statistic. This results in increased power for the robust score test, although in this particular data set the improvement is not very large. \section{Conditional method} In the marginal approach we first balance the data set and then compute results on the adjusted data. In the conditional approach we first compute a predicted survival curve for each subject that accounts for flc group, age and sex, and then take a weighted average of the curves to get an overall estimate for each flc group. For both methods a central consideration is the population of interest, which drives the weights. Modeling has not removed the question of \emph{who} these curves should represent, it has simply changed the order of operation between the weighting step and the survival curves step. \subsection{Stratification} Our first approach is to subset the data into homogeneous age/sex strata, compute survival curves within each strata, and then combine results. We will use the same age/sex combinations as before. The interpretation of these groups is different, however. In the marginal approach it was important to find age/sex groups for which the probability of membership within each FLC group was constant within the strata (independent of age and sex, within strata), in this case it is important that the survival for each FLC group is constant in each age/sex stratum. Homogeneity of membership within each stratum and homogeneity of survival within each stratum may lead to different partitions for some data sets. Computing curves for all the combinations is easy. <>= allfit <- survfit(Surv(futime/365.25, death) ~ group + age2 + sex, fdata) temp <- summary(allfit)$table temp[1:6, c(1,4)] #abbrev printout to fit page @ The resultant survival object has 48 curves: 8 age groups * 2 sexes * 3 FLC groups. To get a single curve for the first FLC group we need to take a weighted average over the 16 age/sex combinations that apply to that group, and similarly for the second and third FLC subset. Combining the curves is a bit of a nuisance computationally because each of them is reported on a different set of time points. A solution is to use the \texttt{summary} function for survfit objects along with the \texttt{times} argument of that function. This feature was originally designed to allow printout of curves at selected time points (6 months, 1 year, \ldots), but can also be used to select a common set of time points for averaging. We will arbitrarily use 4 per year, which is sufficient to create a visually smooth plot over the time span of interest. By default \texttt{summary} does not return data for times beyond the end of a curve, i.e., when there are no subjects left at risk; the \texttt{extend} argument causes a full set of times to always be reported. As seen in the printout above, the computed curves are in sex within age within group order. The overall curve is a weighted average chosen to match the original age/sex distribution of the population. <>= xtime <- seq(0, 14, length=57) #four points/year for 14 years smat <- matrix(0, nrow=57, ncol=3) # survival curves serr <- smat #matrix of standard errors pi <- with(fdata, table(age2, sex))/nrow(fdata) #overall dist for (i in 1:3) { temp <- allfit[1:16 + (i-1)*16] #curves for group i for (j in 1:16) { stemp <- summary(temp[j], times=xtime, extend=T) smat[,i] <- smat[,i] + pi[j]*stemp$surv serr[,i] <- serr[,i] + pi[i]*stemp$std.err^2 } } serr <- sqrt(serr) matplot(xtime, smat, type='l', lwd=2, col=c(1,2,4), ylim=c(0,1), lty=1, xlab="Years from sample", ylab="Survival") lines(sfit1, mark.time=F, lty=2, col=c(1,2,4), xscale=365.25) @ \begin{figure}[tb] \myfig{flc5} \caption{Estimated curves from a stratified model, along with those from the uncorrected fit as dashed lines.} \label{flc5} \end{figure} Figure \ref{flc5} shows the resulting averaged curves. Overlaid are the curves for the unadjusted model. Very careful comparison of these curves with the weighted estimate shows that they have almost identical spread, with just a tiny amount of downward shift. There are two major disadvantages to the stratified curves. The first is that when the original data set is small or the number of confounders is large, it is not always feasible to stratify into a large enough set of groups that each will be homogeneous. The second is a technical aspect of the standard error estimate. Since the curves are formed from disjoint sets of observations they are independent and the variance of the weighted average is then a weighted sum of variances. However, when a Kaplan-Meier curve drops to zero the usual standard error estimate at that point involves 0/0 and becomes undefined, leading to the NaN (not a number) value in R. Thus the overall standard error becomes undefined if any of the component curves falls to zero. In the above example this happens at about the half way point of the graph. (Other software packages carry forward the se value from the last no-zero point on the curve, but the statistical validity of this is uncertain.) To test for overall difference between the curves we can use a stratified test statistic, which is a sum of the test statistics computed within each subgroup. The most common choice is the stratified log-rank statistic which is shown below. The score test from a stratified Cox model would give the same result. <<>>= survdiff(Surv(futime, death) ~ group + strata(age2, sex), fdata) @ \subsection{Modeling} The other approach for conditional estimation is to model the risks due to the confounders. Though we have left it till last, this is usually the first (and most often the only) approach used by most data analysts. Let's start with the very simplest method: a stratified Cox model. <>= cfit4a <- coxph(Surv(futime, death) ~ age + sex + strata(group), data=fdata) surv4a <- survfit(cfit4a) plot(surv4a, col=c(1,2,4), mark.time=F, xscale=365.25, xlab="Years post sample", ylab="Survival") @ This is a very fast and easy way to produce a set of curves, but it has three problems. First is the assumption that this simple model adequately accounts for the effects of age and sex on survival. That is, it assumes that the effect of age on mortality is linear, the sex difference is constant across all ages, and that the coefficients for both are identical for the three FLC groups. The second problem with this approach is that it produces the predicted curve for a single hypothetical subject of age \Sexpr{round(cfit4a[['means']][1], 1)} years and sex \Sexpr{round(cfit4a[['means']][2],2)}, the means of the covariates, under each of the 3 FLC scenarios. However, we are interested in the adjusted survival of a \emph{cohort} of subjects in each range of FLC, and the survival of an ``average'' subject is not the average survival of a cohort. The third and most serious issue is that it is not clear exactly what these ``adjusted'' curves represent --- exactly who \emph{is} this subject a sex of \Sexpr{round(cfit4a[['means']][2],2)}? Multiple authors have commented on this problem, see Thomsen et al \cite{Thomsen91}, Nieto and Coresh \cite{Nieto96} or chapter 10 of Therneau and Grambsh \cite{Therneau00} for examples. Even worse is a Cox model that treated the FLC group as a covariate, since that will impose a an additional constraint of proportional hazards across the 3 FLC groups. \begin{figure} \myfig{flc6} \caption{Curves for the three groups, adjusted for age and sex via a risk model. Dotted lines show the curves from marginal adjustment. Solid curves are for the simple risk model \texttt{cfit4a}.} \label{flc6} \end{figure} We can address this last problem problem by doing a proper average. A Cox model fit can produce the predicted curves for any age/sex combination. The key idea is to produce a predicted survival curve for every subject of some hypothetical population, and then take the average of these curves. The most straightforward approach is to retrieve the predicted individual curves for all \Sexpr{nrow(fdata)} subjects in the data set, assuming each of the three FLC strata one by one, and take a simple average for each strata. For this particular data set that is a bit slow since it involves \Sexpr{nrow(fdata)} curves. However there are only 98 unique age/sex pairs in the data, it is sufficient to obtain the 98 * 3 FLC groups unique curves and take a weighted average. We will make use of the survexp function, which is designed for just this purpose. Start by creating a data set which has one row for each age/sex combination along with its count. Then replicate it into 3 copies, assigning one copy to each of the three FLC strata. <>= tab4a <- with(fdata, table(age, sex)) uage <- as.numeric(dimnames(tab4a)[[1]]) tdata <- data.frame(age = uage[row(tab4a)], sex = c("F","M")[col(tab4a)], count= c(tab4a)) tdata3 <- tdata[rep(1:nrow(tdata), 3),] #three copies tdata3$group <- factor(rep(1:3, each=nrow(tdata)), labels=levels(fdata$group)) sfit4a <- survexp(~group, data=tdata3, weight = count, ratetable=cfit4a) plot(sfit4a, mark.time=F, col=c(1,2,4), lty=1, lwd=2, xscale=365.25, xlab="Years from Sample", ylab="Survival") lines(sfit3, mark.time=F, col=c(1,2,4), lty=2, lwd=1, xscale=365.25) legend(2,.4, c("FLC low", "FLC med", "FLC high"), lty=1, col=c(1,2,4), bty='n', lwd=2) @ Figure \ref{flc6} shows the result. Comparing this to the prior 3 adjustments shown in figures \ref{flc3}, \ref{flc4}, and \ref{flc5} we see that this result is different. Why? Part of the reason is due to the fact that $E[f(X)] \ne f(E[X])$ for any non-linear operation $f$, so that averages of survival curves and survival curves of averages will never be exactly the same. This may explain the small difference between the stratified and the marginal approaches of figures \ref{flc3} and \ref{flc5}, which were based on the same subsets. The Cox based result is systematically higher than the stratified one, however, so something more is indicated. Aside: An alternate computational approach is to create the individual survival curves using the \texttt{survfit} function and then take averages. <<>>= tfit <- survfit(cfit4a, newdata=tdata, se.fit=FALSE) curves <- vector('list', 3) twt <- c(tab4a)/sum(tab4a) for (i in 1:3) { temp <- tfit[i,] curves[[i]] <- list(time=temp$time, surv= c(temp$surv %*% twt)) } @ The above code is a bit sneaky. I know that the result from the survfit function contains a matrix \texttt{tfit\$surv} of 104 columns, one for each row in the tdata data frame, each column containing the curves for the three strata one after the other. Sub setting \texttt{tfit} results in the matrix for a single flc group. Outside of R an approach like the above may be needed, however. \begin{figure} \myfig{flc6b} \caption{Left panel: comparison of Cox model based adjustment (solid) with the curves based on marginal adjustment (dashed). The Cox model curves without (black) and with (red) an age*sex interaction term overlay. Right panel: plot of the predicted relative risks from a Cox model \texttt{crate} versus population values from the Minnesota rate table.} \label{flc6b} \end{figure} So why are the modeling results so different than either reweighting or stratification? Suspicion first falls on the use of a simple linear model for age and sex, so start by fitting a slightly more refined model that allows for a different slope for the two sexes, but is still linear in age. In this particular data set an external check on the fit is also available via the Minnesota death rate tables, which are included with the survival package as \texttt{survexp.mn}. This is an array that contains daily death rates by age, sex, and calender year. <>= par(mfrow=c(1,2)) cfit4b <- coxph(Surv(futime, death) ~ age*sex + strata(group), fdata) sfit4b <- survexp(~group, data=tdata3, ratetable=cfit4b, weights=count) plot(sfit4b, fun='event', xscale=365.25, xlab="Years from sample", ylab="Deaths") lines(sfit3, mark.time=FALSE, fun='event', xscale=365.25, lty=2) lines(sfit4a, fun='event', xscale=365.25, col=2) temp <- median(fdata$sample.yr) mrate <- survexp.mn[as.character(uage),, as.character(temp)] crate <- predict(cfit4b, newdata=tdata, reference='sample', type='lp') crate <- matrix(crate, ncol=2)[,2:1] # mrate has males then females, match it # crate contains estimated log(hazards) relative to a baseline, # and mrate absolute hazards, make both relative to a 70 year old for (i in 1:2) { mrate[,i] <- log(mrate[,i]/ mrate[21,2]) crate[,i] <- crate[,i] - crate[21,2] } matplot(mrate, crate, col=2:1, type='l') abline(0, 1, lty=2, col=4) @ The resulting curves are shown in the left panel of figure \ref{flc6b} and reveal that addition of an interaction term did not change the predictions, and that the Cox model result for the highest risk group is distinctly different predicted survival for the highest FLC group is distinctly different when using model based prediction. The right hand panel of the figure shows that though there are slight differences with the Minnesota values, linearity of the age effect is very well supported. So where exactly does the model go wrong? Since this is such a large data set we have the luxury of looking at subsets. This would be a very large number of curves to plot --- age by sex by FLC = 48 --- so an overlay of the observed and expected curves by group would be too confusing. Instead we will summarize each of the groups according to their observed and predicted number of events. <>= obs <- with(fdata, tapply(death, list(age2, sex, group), sum)) pred<- with(fdata, tapply(predict(cfit4b, type='expected'), list(age2, sex, group), sum)) excess <- matrix(obs/pred, nrow=8) #collapse 3 way array to 2 dimnames(excess) <- list(dimnames(obs)[[1]], c("low F", "low M", "med F", "med M", "high F", "high M")) round(excess, 1) @ The excess risks, defined as the observed/expected number of deaths, are mostly modest ranging from .8 to 1.2. The primary exception exception is the high FLC group for ages 50--59 which has values of 1.6 to 2.5; the Cox model fit has greatly overestimated the survival for the age 50--54 and 55--59 groups. Since this is also the age category with the highest count in the data set, this overestimation will have a large impact on the overall curve for high FLC subset, which is exactly where the the deviation in figure \ref{flc6b} is observed to lie. There is also mild evidence for a linear trend in age for the low FLC females, in the other direction. Altogether this suggests that the model might need to have a different age coefficient for each of the three FLC groups. <<>>= cfit5a <- coxph(Surv(futime, death) ~ group:age +sex + strata(group), fdata) cfit5b <- coxph(Surv(futime, death) ~ group:(age +sex) + strata(group), fdata) cfit5c <- coxph(Surv(futime, death) ~ group:(age *sex) + strata(group), fdata) options(show.signif.stars=FALSE) # see footnote anova(cfit4a, cfit5a, cfit5b, cfit5c) temp <- coef(cfit5a) names(temp) <- c("sex", "ageL", "ageM", "ageH") round(temp,3) @ The model with separate age coefficients for each FLC group gives a major improvement in goodness of fit, but adding separate sex coefficients per group or further interactions does not add importantly beyond that. \footnote{There are certain TV shows that make one dumber just by watching them; adding stars to the output has the same effect on statisticians.} \begin{figure} \myfig{flc7} \caption{Adjusted survival for the 3 FLC groups based on the improved Cox model fit. Dashed lines show the predictions from the marginal model.} \label{flc7} \end{figure} A recheck of the observed/expected values now shows a much more random pattern, though some excess remains in the upper right corner. The updated survival curves are shown in figure \ref{flc7} and now are in closer concordance with the marginal fit. <>= pred5a <- with(fdata, tapply(predict(cfit5a, type='expected'), list(age2, sex, group), sum)) excess5a <- matrix(obs/pred5a, nrow=8, dimnames=dimnames(excess)) round(excess5a, 1) sfit5 <- survexp(~group, data=tdata3, ratetable=cfit5a, weights=count) plot(sfit3, fun='event', xscale=365.25, mark.time=FALSE, lty=2, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, col=c(1,2,4)) @ One problem with the model based estimate is that standard errors for the curves are complex. Standard errors of the individual curves for each age/sex/FLC combination are a standard output of the survfit function, but the collection of curves is correlated since they all depend on a common estimate of the model's coefficient vector $\beta$. Curves with disparate ages are anti-correlated (an increase in the age coefficient of the model would raise one and lower the other) whereas those for close ages are positively correlated. A proper variance for the unweighted average has been derived by Gail and Byar \cite{Gail86}, but this has not been implemented in any of the standard packages, nor extended to the weighted case. A bootstrap estimate would appear to be the most feasible. \section{Conclusions} When two populations need to be adjusted and one is much larger than the other, the balanced subset method has been popular. It is most often seen in the context of a case-control study, with cases as the rarer group and a set of matched controls selected from the larger one. This method has the advantage that the usual standard error estimates from a standard package are appropriate, so no further work is required. However, in the general situation it leads to a correct answer but for the wrong problem, i.e., not for a population in which we are interested. The population reweighted estimate is flexible, has a readily available variance in some statistical packages (but not all), and the result is directly interpretable. It is the method we recommend in general. The approach can be extended to a large number of balancing factors by using a regression model to derive the weights. Exploration and checking of said model for adequacy is an important step in this case. The biggest downside to the method arises when there is a subset which is rare in the data sample but frequent in the adjusting population. In this case subjects in that subset will be assigned large weights, and the resulting curves will have high variance. The stratified method is closely related to reweighting (not shown). It does not do well if the sample size is small, however. Risk set modeling is a very flexible method, but is also the one where it is easiest to go wrong by using an inadequate model, and variance estimation is also difficult. To the extent that the fitted model is relevant, it allows for interpolation and extrapolation to a reference population with a different distribution of covariates than the one in the training data. It may be applicable in cases such as rare subsets where population reweighting is problematic, with the understanding that one is depending heavily on extrapolation in this case, which is always dangerous. \section{A note on type 3 tests} One particular software package (not R) and its proponents are very fond of something called ``type 3'' tests. Said tests are closely tied to a particular reference population: \begin{itemize} \item For all continuous covariates in the model, the empirical distribution is used as the reference. \item For all categorical adjusters, a uniform distribution over the categories is used. \end{itemize} Figure \ref{flc8} shows the fit from such a model. Not surprisingly, the predicted death rate is very high: 1/4 of our population is over 80 years old! The authors do not find such a prediction particularly useful since we don't ever expect to see a population like this (it's sort of like planning for the zombie apocalypse), but for those enamored of type 3 tests this shows how to create the corresponding curves. <>= # there is a spurious warning from the model below: R creates 3 unneeded # columns in the X matrix cfit6 <- coxph(Surv(futime, death) ~ group:age2 + sex + strata(group), fdata) saspop <- with(fdata, expand.grid(age2= levels(age2), sex= levels(sex), group = levels(group))) sfit6 <- survexp(~group, data=saspop, ratetable=cfit6) plot(sfit6, fun='event', xscale=365.25, mark.time=FALSE, lty=1, col=c(1,2,4), xlab="Years from sample", ylab="Deaths") lines(sfit5, fun='event', xscale=365.25, lty=2, col=c(1,2,4)) @ \begin{figure} \myfig{flc8} \caption{Adjusted survival for the 3 FLC groups based on a fit with categorical age, and predicting for a uniform age/sex population. Dashed lines show the predictions from the marginal model.} \label{flc8} \end{figure} \bibliographystyle{plain} \bibliography{refer} \end{document} survival/vignettes/refer.bib0000644000175100001440000017331312261334377015754 0ustar hornikusers@string{annals= {Annals of Stat.}} @string{applstat= {Applied Stat.}} @string{bioj = {Biometrical J.}} @string{biok = {Biometrika}} @string{commstata = {Comm. Stat. Theory Methods}} @string{biom = {Biometrics}} @string{jap = {J. Applied Probability}} @string{jasa = {J. Amer. Stat. Assoc.}} @string{jrssa= {J. Royal Stat. Soc. A}} @string{jrssb= {J. Royal Stat. Soc. B}} @string{jrssc= {J. Royal Stat. Soc. C}} @string{jscs = {J Stat. Comput. Simul.}} @string{lifetime = {Lifetime Data Analysis}} @string{NEJM = {New England J. Medicine}} @string{scand = {Scandinavian J. Stat.}} @string{statmed= {Stat. in Medicine}} @string{statsci = {Stat. Science}} @article{Aalen88, author= {Aalen, O. O.}, title= {Heterogeneity in survival analysis}, year= {1988}, journal=statmed, volume={7}, pages={1121-1137} } @article{Aalen89, author= {Aalen,O. O.} , title= {A linear regression model for the analysis of life times}, year= {1989}, journal=statmed, volume={8}, pages={907--925} } @article{Aalen93, author= {Aalen,O. O.} , title= {Further results on the non-parametric linear regression model in survival analysis}, year= {1993}, journal=statmed, volume={12}, pages={1569--1588} } @article{Andersen82, author= {Andersen, P. K. and Gill, R. D} , title= {Cox's regression model for counting processes: A large sample study}, year= {1982}, journal=annals, volume={10}, pages={1100--1120} } @article{Andersen89, author= {Andersen, P. K. and V{\ae}th, M.} , title= {Simple parametric and nonparametric models for excess and relative mortality}, year= {1989}, journal=biom, volume={45}, pages={523--535} } @book{Andersen92, author={Andersen, P. K. and Borgan, {\O}. and Gill, R. D. and Keiding, N.}, title= {Statistical Models Based on Counting Processes}, publisher={Springer-Verlag}, address={New York}, year= {1993} } @article{Andersen00, author= {Andersen, P. K. and Esbjerg, S. and S{\o}rensen, T.I.A.} , title= {Multi-state models for bleeding episodes and morality in lever cirrhosis}, year= {2000}, journal=statmed, volume={19}, pages={587--599} } @article{Anderson82, author= {J. R. Anderson and L. Bernstein and M. C. Pike}, year= {1982}, title= {Approximate confidence intervals for probabilities of survival and quantiles in life-table analysis}, journal=biom}, volume={38}, pages= {407--416}, } @book{Anderson58, author= {V. E Anderson and H. O. Goodman and S. Reed}, title= {Variables Related to Human Breast Cancer}, year = 1958, publisher={University of Minnesota Press}, address={Minneapolis} } @article{Barlow88, author= { Barlow, W. E. and Prentice, R. L.}, year = {1988}, title= {Residuals for relative risk regression}, journal= {Biometrika}, volume={75}, pages={65--74} } @article{Barlow94, author= {Barlow, W. E.}, year ={1994}, title= {Robust variance estimation for the case-cohort design}, journal= {Biometrics}, volume={50}, pages= {1064--1072} } @article{Bartolucci77, author = {Bartolucci, A. A. and Fraser, M. D.}, year = 1977, title = {Comparative step-up and composite tests for selecting prognostic indicators associated with survival}, journal = {Biometrical J.}, volume = 19, pages = {437-448} } @book{Becker84, author={Becker, R. A. and Chambers, J. M.}, title= {S: {A}n Interactive Environment for Data Analysis and Graphics}, publisher={Wadsworth}, address={Belmont, CA}, year= {1984} } @techreport{Bergstralh88, author= {Bergstralh, E. J. and Offord, K. P.}, year = {1988}, title = {Conditional probabilities used in calculating cohort expected survival}, number = {37}, institution= {Department of Health Sciences Research, Mayo Clinic} } @article{Berry83, author= {Berry, G.}, year = {1983}, title= {The analysis of mortality by the subject years method}, journal= biok, volume={39}, pages={173--184} } @book{Bickel77, author={Bickel, P. J. and Doksum,K. J. }, title= { Mathematical Statistics: Basic Ideas and Selected Topics}, publisher={Holden-Day}, address={San Francisco}, year= {1977} } @article{Binder92, author= {Binder, D. A.}, year= {1992}, title= {Fitting {C}ox's proportional hazards models from survey data}, journal={Biometrika}, volume={79}, pages={139--147} } @ARTICLE{Blackstone86, author = {Blackstone, E. H. and Naftel, D. C. and Turner, M. E.}, year = 1986, title = {The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information}, journal = JASA, volume = 81, pages = {615-624}, annote = {parametric survival models; non PH} } @article{Bonsel90, author= {Bonsel, G. J. and Klompmaker, I. J. and {van't Veer}, F. and Habbema, J. D. F. and Slooff, M. J. H.}, year= {1990}, title= {Use of prognostic models for assessment of value of liver transplantation in primary biliary cirrhosis}, journal={Lancet}, volume={335}, pages={493--497} } @article{Borgan95, author= {Borgan, \O. and Goldstein, L. and Langholz, B.}, year= {1995}, title= {Methods for the analysis of sampled cohort data in the {C}ox proportional hazards model}, journal=annals, volume={23}, pages={1749--1778} } @article{Breslow93, author= {Breslow, N. E. and Clayton, D. G.}, year= {1993}, title= {Approximate inference in generalized linear mixed models}, journal=jasa, volume={88}, pages={9--25} } @article{Cai95, author= {Cai, J. and Prentice, R. G.}, year= {1995}, title= {Estimating equations for hazard ratio parameters based on correlated failure time data}, journal={Biometrika}, volume={82}, pages={151--164} } @article{Cain84, author={Cain, K. C. and Lange, N. T.}, year= {1984}, title= {Approximate case influence for the proportional hazards regression model with censored data}, journal={Biometrics}, volume={40}, pages={493--499} } @book{Chambers83, author= {Chambers, J. M. and Cleveland, W. S. and Kleiner, B. and Tukey, P. A.}, year= {1983}, title={Graphical Methods for Data Analysis}, publisher ={Wasdworth}, address ={Belmont, CA} } @book{Chambers91, author= {Chambers, J. M. and Hastie, T. J.}, year= {1993}, title={Statistical Models in {S}}, publisher ={Chapman and Hall}, address ={New York} } @article{Chappell92, author= {Chappell, R.}, year= {1992}, title= {A note on linear rank tests and {G}ill and {S}chumacher's tests of proportionality}, journal={Biometrika}, volume={79}, pages= {199--201} } @article{Chen91, author= {Chen, C. H. and Wang, P. C.}, year= {1991}, title= {Diagnostic plots in {C}ox's regression model}, journal={Biometrics}, volume={47}, pages= {841--850} } @article{Clegg99, author= {Clegg, L. X. and Cai, J. and Sen, P. K.}, year= {1999}, title= {A marginal mixed baseline hazards model for multivariate failure time data}, journal=jasa, volume={55}, pages= {805--812} } @book{Cook82, author= {Cook, R. D. and Weisberg, S.}, year= {1982}, title= {Residuals and Influence in Regression}, publisher={Chapman and Hall}, address= {London} } @article{Cox72, author= {Cox, D. R.}, year= {1972}, title= {Regression models and life-tables (with discussion)}, journal=jrssb, volume={34}, pages= {187--220} } @book{Cox84, author= {Cox, D. R. and Oakes, D. O.}, year= {1984}, title= {Analysis of Survival Data}, publisher= {Chapman and Hall}, address= {London} } @book{Cochran76, author= {Cochran, W.G.}, year= {1976}, title={Sampling Techniques, third edition}, publisher ={Wiley}, address ={New York} } @article{Crowley77, author= {Crowley, J. and Hu, M.}, year= {1977}, title= {Covariance analysis of heart transplant survival data}, journal=jasa, volume={72}, pages= {27--36} } @article{Dempster77, author= {Dempster, A. P. and Laird, N. M. and Rubin, D. B.}, year= {1977}, title= {Maximum likelihood from incomplete data via the {EM} algorithm (with discussion)}, journal=jrssb, volume={39}, pages= {1--38} } @techreport{Deng95, author= {Deng, Y. and Quigley, J.M. and Van Order, R.}, year= {1995}, title= {Mortgage Terminations}, institution={Institute of Business and Economic Research, University of California at Berkeley}, type= {Working Paper}, number={95-230}, } @article{Dickson89, author= {Dickson, E. R. and Grambsch, P. M. and Fleming, T. R and Fisher, L. D. and Langworthy, A.}, year= {1989}, title= {Prognosis in primary biliary cirrhosis: Model for decision making}, journal ={Hepatology}, volume={10}, pages ={1--7} } @article{Dispenzieri12, author={A. Dispenzieri and J. Katzmann and R. Kyle and D. Larson and T. Therneau and C. Colby and R. Clark and G. Mead and S. Kumar and L.J. Melton III and S.V. Rajkumar}, year=2012, title ={Use of monclonal serum immunoglobulin free light chains to predict overall survival in the general population}, journal={Mayo Clinic Proc}, volume=87, pages= {512--523} } @article{Ederer61, author= {Ederer, F. and Axtell, L. M. and Cutler, S. J.}, year= {1961}, title= {The relative survival rate: A statistical methodology}, journal ={National Cancer Inst. Monographs}, volume={6}, pages ={101--121} } @techreport{Ederer77, author= {Ederer, F. and Heise, H.}, year= {1977}, title= {Instructions to {IBM} 650 programmers in processing survival computations}, institution={End Results Evaluation Section, National Cancer Institute}, type={Methodological Note}, number={No. 10}, pages ={101--121} } @article{Edmonson79, author= {Edmonson, J. H. and Fleming, T. R. and Decker, D. G. and Malkasian, G. D. and Jorgensen, E. O. and Jefferies, J. A. and Webb, M. J. and Kvols, L. K.}, year= {1979}, title= {Different chemotherapeutic sensitivities and host factors affecting prognosis in advanced ovarian carcinoma versus minimal residual disease }, journal ={Cancer Treatment Reports}, volume={63}, pages ={241--247} } @article{Efron77, author= {Efron, B.}, year= {1977}, title ={The efficiency of {C}ox's likelihood function for censored data}, journal=jasa, volume={72}, pages= {557--565} } @book{Efron82, author= {Efron, B.}, year= {1982}, title ={The Jackknife, the Bootstrap and Other Resampling Plans}, publisher={SIAM}, address ={Philadelphia} } @article{Ezekiel24, author= {Ezekiel, M.}, year= {1924}, title ={A method for handling curvilinear correlation for any number of variables}, journal=jasa, volume={19}, pages= {431--453} } @article{Fleming81, author= {Fleming, T. R. and Harrington, D. P.}, year= {1981}, title ={A class of hypothesis tests for one and two sample censored survival data}, journal=commstata, volume={10}, pages= {763--794} } @article{Fleming84, author= {Fleming, T. R. and Harrington, D. P.}, year= {1984}, title ={Nonparametric estimation of the survival distribution in censored data}, journal=commstata, volume={13}, pages= {2469--2486} } @book{Fleming91, author= {Fleming, T. R. and Harrington, D. P.}, year= {1991}, title= {Counting Processes and Survival Analysis}, publisher= {Wiley}, address= {New York} } @article{Eilers96, author= {Eilers, P. H. C. and Marx, B. D.}, year= {1996}, title ={Flexible smoothing with {B}-splines and penalties}, journal={Stat. Science}, volume={11}, pages= {89--121} } @article{Gail81, author= {Gail, M. H., Lubin, J. H., and Rubinstein, L. V.}, year = {1981}, title = {Likelihood Calculations for Matched Case-Control Studies and Survival Studies with Tied Death Times}, journal=biok, volume=68, pages={703--707} } @article{Gail86, author= {Gail, M. H. and Byar, D. P.}, year= {1986}, title= {Variance calculations for direct adjusted survival curves, with applications to testing for no treatment effect}, journal=bioj, volume={28}, pages= {587--599} } @article{Gill87, author= {Gill, R. and Schumacher, M.}, year= {1987}, title= {A simple test of the proportional hazards assumption}, journal={Biometrika}, volume={74}, pages= {289--300} } @article{Grambsch94, author= {Grambsch, P. M. and Therneau, T. M.}, year= {1994}, title ={Proportional hazards tests and diagnostics based on weighted residuals}, journal={Biometrika}, volume={81}, pages= {515--526} } @article{Grambsch95, author= {Grambsch, P. M. and Therneau, T. M. and Fleming, T. R.}, year= {1995}, title= {Diagnostic plots to reveal functional form for covariates in multiplicative intensity models}, journal={Biometrics}, volume={51}, pages= {1469-1482} } @article{Gray92, author= {Gray, R. J.}, year = {1992}, title= {Flexible methods for analyzing survival data using splines, with applications to breast cancer prognosis}, journal=jasa, volume={87}, pages = {942--951} } @article{Gray94, author= {Gray, R. J.}, year = {1994}, title= {Spline-based tests in survival analysis}, journal=biom, volume={50}, pages = {640--652} } @article{Green84, author={Green, P.J.}, year = {1984}, title = {Iteratively reweighted least squares for maximum likelihood estimation, and some robust and resistant alternatives (with discussion)}, journal=jrssb, volume=46, pages={149--192} } @article{Ha01, author={I. L. Ha and Y. Lee and J. Song}, year={2001}, title={Heirarchical likelihood approach for frailty models}, journal=biok, pages={233--243} } @techreport{Hall95, author= {Hall, C. B. and Zeger, S. L. and Bandeen-Roche, K. J.}, year= {1995}, title= {Adjusted variable plots for {C}ox's proportional hazards regression model}, institution={The John's Hopkins University, School of Hygiene and Public Health, Department of Biostatistics} } @incollection{Harrell86, author= {Harrell, F.}, year= {1986}, title= {The PHGLM procedure}, booktitle= {SAS Supplemental Library User's Guide, Version 5}, address= {Cary, NC}, publisher= {SAS Institute Inc} } @article{Hakama77, author= {Hakama, M. and Hakulinen, T.}, year= {1977}, title= {Estimating the expectation of life in cancer survival studies with incomplete follow-up information}, journal={J Chronic Diseases}, volume={30}, pages= {585--597} } @article{Hakulinen82, author= {Hakulinen, T.}, year= {1982}, title= {Cancer survival corrected for heterogeneity in patient withdrawal}, journal=biom, volume={38}, pages= {933--942} } @article{Hakulinen85, author= {Hakulinen, T. and Abeywickrama, K. H.}, year= {1985}, title= {A computer program package for relative survival analysis}, journal={Computer Programs in Biomedicine}, volume={19}, pages= {197--207} } @article{Hakulinen77, author= {Hakulinen, T.}, year= {1977}, title= {On long term relative survival rates}, journal={J Chronic Diseases}, volume={30}, pages= {431--443} } @article{Harrington82, author= {Harrington, D. P. and Fleming, T. R.}, year= {1982}, title= {A class of rank test procedures for censored survival data}, journal=biok, volume={69}, pages= {553-566} } @book{Hastie90, author= {Hastie, T. J. and Tibshirani, R. J.}, year= {1990}, title ={Generalized Additive Models}, publisher ={Chapman and Hall}, address= {London} } @article{Hastie96, author= {Hastie, T. J.}, year= {1996}, title= {Pseudosplines}, journal=jrssb, volume={58}, pages= {379--396} } @article{Heit99, author= {Heit, J. A. and M. D. Siverstein and D. N. Mohr and T. M. Petterson and W. M. O'Fallon and L. J. Melton III}, year= {1999}, title= {Predictors of survival after deep vein thrombosis and pulmonary embolism}, journal={Arch Internal Med}, volume={159}, pages= {445--453} } @techreport{Hodges99, author= {J. S. Hodges and D. J. Sargent}, year= {1998}, title= {Counting degrees of freedom in hierarchical and other richly-parameterized models}, institution={University of Minnesota, Division of Biostatistics}, type={Research Report}, number={98-004} } @book{Hougaard00, author={Hougaard, P.}, title= {Analysis of Multivariate Survival Data}, publisher={Springer-Verlag}, address={New York}, year= {2000} } @inproceedings{Huber67, author= {Huber, P. J.}, year= {1967}, title= {The behavior of maximum likelihood estimates under non-standard conditions}, booktitle= {Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability}, volume={1}, pages= {221--233} } @article{Huffer91, author= {Huffer, F.W. and McKeague, I.W}, title= {Weigthed least squares estimation for {A}alen's additive risk model}, year= {1991}, journal=jasa, volume={86}, pages={114--129} } @article{Hurvich98, author= {Hurvich, C. M. and Simonoff, J. S. and Tsai, C.-L.}, year= {1998}, title= {Smoothing parameter selection in nonparametric regression using an improved {A}kaike information criterion}, journal=jrssb, volume={60}, pages= {271--293} } @article{Islam94, author= {Islam, M. A.}, year= {1994}, title= {Multistate survival models for transitions and reverse transitions: an application to contraceptive use data}, journal=jrssa, volume={157}, pages= {441--455} } @article{Jeon12, author={J. Jeon and L. Hsu and M. Gorfine}, title={Bias correction in the heirarchical likelihood approach to the analysis of multivariate survival data}, year=2012, journal=Bioinformatics, volume=13, pages={384--397} } @article{Jones90, author={Jones, M. P. and Crowley, J.}, title={Asymptotic proporties of a general class of nonparametric tests for survival analysis}, year={1990}, journal=annals, volume={18},pages={1203--1220} } @book{Kalbfleisch80, author= {Kalbfleisch, J. D. and Prentice, R. L.}, year= {1980}, title= {The Statistical Analysis of Failure Time Data}, publisher= {Wiley}, address ={New York} } @article{Kay83, author= {Korn, Edward L. and Graubard, Barry I. and Midthune, Douglas}, year= {1997}, title ={Time-to-event analysis if longitudinal follow-up of a survey: Choice of the time scale}, journal={Am J of Epidemiology}, volume={145}, pages= {72--80} } @article{Klein91, author= {Kay, R.}, year= {1983}, title ={The analysis of transition times in a multistate stochastic process using proportional hazard regression models}, journal=commstata, volume={11}, pages= {1743--1756} } @article{Korn97, author= {Klein, J. P.}, year= {1991}, title ={Small sample moments of some estimators of the variance of the {K}aplan--{Meier} and {N}elson--{A}alen estimators}, journal=scand, volume={18}, pages= {333--340} } @article{Kyle06, year= 2006, author={Kyle, R. and Therneau, T. and Rajkumar, S.V. and Larson, D. and Pleva, M. and Offord, J. and Dispenzieri, A. and Katzman, J. and L.J. Melton III}, title={ Prevalence of monoclonal gammopathy of undetermined significance}, journal={New England J Medicine}, vol=354, pages={1362--1369} } @article{Lagakos84, author={Lagakos, S. W. and Schoenfeld, D. A.}, title={Properties of proportional-hazards score tests under misspecified regression models}, year={1984}, journal=biom, volume={40}, pages={1037--1048} } @article{Laird81, author= {Laird, N. and Olivier, D.}, year= {1981}, title= {Covariance analysis of censored survival data using log-linear analysis techniques}, journal=jasa, volume={76}, pages= {231--240} } @article{Langholz91, author= {Langholz, B. and Thomas, D.C.}, year= {1991}, title= {Efficiency of cohort sampling designs: some surprising resluts}, journal={Biometrics}, volume={47}, pages= {1563--1572} } @article{Langholz96, author= {Langholz, B. and Goldstein, L.}, year= {1996}, title= {Risk set sampling in epidemiologic cohort studies}, journal={Statistical Science}, volume={11}, pages= {35--53} } @article{Laurie89, author ={Laurie, J. A. and Moertel, C. G. and Fleming, T. R. and Wieand, H. S. and Leigh, J. E. and Rubin, J. and McCormack, G. W. and Gerstner, J. B. and Krook, J. E. and Malliard, J.}, year= {1989}, title= {Surgical adjuvant therapy of large-bowel carcinoma: {A}n evaluation of levamisole and the combination of levamisole and fluorouracil: The {N}orth {C}entral {C}ancer {T}reatment {G}roup and the {M}ayo {C}linic}, journal={J. Clinical Oncology}, volume={7}, pages= {1447--1456} } @incollection{Lee92, author= {Lee, E. W. and Wei, L. J. and Amato, D.}, year= {1992}, title= {{C}ox-type regression analysis for large number of small groups of correlated failure time observations}, editor= {Klein, J. P. and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, pages= {237--247}, publisher= {Kluwer}, address= {Netherlands} } @article{Lee96, author={Y. Lee and J. A. Nelder}, year={1996}, title={Hierarchical generalized linear models (with discussion)}, journal=jrssb, volume=58, pages={619--678} } @article{Lindor94, author= {Lindor, K. D. and Dickson, E. R. and Baldus, W. P. and Jorgensen, R. A. and Ludwig, J. and Murtaugh, P. A. and Harrison, J. M. and Wiesner, R. H. and Anderson, M. L. and Lange, S. M. and LeSage, G. and Rossi, S. S. and Hofman, A. F.}, year= {1994}, title= {Ursodeoxycholic acid in the treatment of primary biliary cirrhosis}, journal={Gastroenterology}, volume={106}, pages= {1284--1290} } @article{Liang86, author= {Zeger, S. L. and Liang, K. Y.}, year= {1986}, title= {Longitudinal data analysis for discrete and continuous outcomes}, journal={Biometrics}, volume={42}, pages= {121--130} } @article{Liang86b, author= {Liang, K. Y. and Zeger, S. L.}, year= {1986}, title= {Longitudinal data analysis using generalized linear models}, journal=biok, volume={73}, pages= {13--22} } @article{Liang88, author= {Zeger, S. L. and Liang, K. Y. and Albert, P. S.}, year= {1988}, title= {Models for longitudinal data: A generalized estimating equation approach}, journal=biom, volume={44}, pages= {1049--1060} } @article{Lin91, author= {Lin, D. Y.}, year= {1991}, title= {Goodness-of-fit analysis for the {C}ox regression model based on a class of parameter estimators}, journal=jasa, volume={86}, pages= {725--728} } @article{Lin89, author= {Lin, D. Y. and Wei, L. J.}, year= {1989}, title= {The robust inference for the {C}ox proportional hazards model}, journal=jasa, volume={84}, pages= {1074--1078} } @article{Lin91b, author= {Lin, D. Y. and Wei, L. J.}, year= {1991}, title= {Goodness-of-fit tests for the general {C}ox regression model}, journal= {Statistica Sinica}, volume={1}, pages= {1--17} } @article{Lin93, author= {Lin, D. Y. and Wei, L. J. and Ying, Z.}, year= {1993}, title= {Checking the {C}ox model with cumulative sums of martingale-based residuals}, journal={Biometrika}, volume={80}, pages= {557--572} } @article{Lin93b, author= {Lin, D. Y. and Ying, Z.}, year= {1993}, title= {Cox regression with incomplete covariate measurements}, journal= jasa, volume={88}, pages= {1341--1349} } @article{Lin94, author= {Lin, D. Y.}, year= {1994}, title= {Cox regression analysis of multivariate failure time data: the marginal approach}, journal= statmed, volume={13}, pages= {2233--2247} } @article{Link84, author= {C. L. Link}, year= {1984}, title= {Confidence intervals for the survival function using {C}ox's proportional-hazard model with covariates}, journal=biom, volume={40}, pages= {601--610} } @article{Link86, author= {C. L. Link}, year= {1986}, title= {Response to {J}. {O'Quigley}, correspondence section}, journal=biom, volume={42}, pages= {219--220} } @article{Lipsitz90, author= {Lipsitz, S. R. and Laird, N. M. and Harrington, D. P.}, year= {1990}, title= {Using the jackknife to estimate the variance of regression estimators from repeated measures studies}, journal=commstata, volume={19}, pages= {821--845} } @article{Lipsitz94, author= {Lipsitz, S. R. and Dear, K. B .G. and Zhao, L.}, year= {1994}, title= {Jackknife estimators of variance for parameter estimates from estimating equations with applications to clustered survival data}, journal= biom, volume={50}, pages= {842--846} } @article{Lunn95, author= {Lunn, M. and McNeil, D.}, year= {1995}, title= {Applying {C}ox regression to competing risks}, journal={Biometrics}, volume={51}, pages= {524--532} } @article{Mallows86, author= {Mallows, C. L.}, year= {1986}, title= {Augmented partial residuals}, journal={Technometrics}, volume={28}, pages= {313--319} } @article{Mantel66, author= {Mantel, N.}, year= {1966}, title= {Evaluation of survival data and two new rank order statistics arising in its consideration}, journal={Cancer Chemotherapy Reports}, volume={50}, pages= {163--166} } @article{Mantel77, author= {Mantel, N. and Bohidar, N. R. and Ciminera, J. L.}, year= {1977}, title= {Mantel--{H}aenszel analyses of litter-matched time-to-response data with modifications for recovery of interlitter information}, journal={Cancer Research}, volume={37}, pages= {3863--3868} } @book{glim, author= {McCullagh, P. and Nelder, J.A.}, year= {1983}, title= {Generalized Linear Models}, publisher= {Chapman and Hall} } @article{McGilchrist91, author= {McGilchrist, C. A. and Aisbett, C. W.}, year= {1991}, title= {Regression with frailty in survival analysis}, journal={Biometrics}, volume={47}, pages= {461--466} } @article{McGilchrist93, author= {McGilchrist, C. A.}, year= {1993}, title= {{REML} estimation for survival models with frailty}, journal={Biometrics}, volume={49}, pages= {221--225} } @article{McGilchrist95, author= {McGilchrist, C. A. and Yau, K. K. W.}, year= {1995}, title= {The derivation of {BLUP}, {ML} and {REML} estimation methods for generalised linear mixed models}, journal= commstata, volume={24}, pages= {2963--2980} } @book{Miller81, author= {Miller, Jr., R. G.}, year= {1981}, title ={Survival Analysis}, publisher= {Wiley}, address= {New York} } @article{Moertel90, author= {Moertel, C.G. and Fleming, T.R. and McDonald, J.S. and Haller, D.G. and Laurie, J.A. and Goodman, P.J. and Ungerleider, J.S. and Emerson, W.A. and Tormey, D.C. and Glick, J.H. and Veeder, M.H. and Mailliard, J.A.}, year= {1990}, title= {Levamisole and fluorouracil for adjucant therapy of resected colon carcinoma.}, journal= NEJM, volume={332}, pages= {352--358} } @article{Moreau85, author= {Moreau, T. and O'Quigley, J. and Mesbah, M.}, year= {1985}, title= { A global goodness-of-fit statistic for the proportional hazards model}, journal= {Applied Stat.}, volume={34}, pages= {212--218} } @article{Moss83, author= {Moss, A. J. and {the Multicenter Postinfarction Research Group}}, year= {1983}, title= {Risk stratification and survival after myocardial infarction}, journal= {New England J. Medicine}, volume={309}, pages= {331--336} } @article{Moss88, author= {Moss, A. J. and {the Multicenter Diltiazem Postinfarction Trial Research Group}}, year= {1988}, title= {The effect of diltiazem on mortality and reinfarction after myocardial infarction}, journal= {New England J. Medicine}, volume={319}, pages= {385--392} } @book{Mosteller77, author= {Mosteller, F. and Tukey, J. W.}, year= {1977}, title= {Data Analysis and Regression}, publisher= {Addison-Wesley}, address={Reading, MA} } @article{Nagelkerke84, author= {Nagelkerke, N. J. D. and Oosting, J. and Hart, A. A. M.}, year= {1984}, title ={A simple test for goodness of fit of {C}ox's proportional hazards model}, journal={Biometrics}, volume={40}, pages= {483--486} } @article{Neuberger86, author= {Neuberger, J. and Altman, D. G. and Christensen, E. and Tygstrup, N. and Williams, R.}, year= {1986}, title ={Use of a prognostic index in evaluation of liver transplantation for primary biliary cirrhosis}, journal={Transplantation}, volume={41}, pages= {713--716} } @article{Nielsen92, author= {Nielsen, G. G. and Gill, R. D. and Andersen, P. K. and S{\o}rensen, T. I.}, year= {1992}, title ={A counting process approach to maximum likelihood estimation of frailty models}, journal=scand, volume={19}, pages= {25--43} } @article{Nieto96, author ={F. Javier Nieto and Josef Coresh}, title = {Adjusting survival curves for confounders: a review and a new method}, year={1996}, journal={Am J of Epidemiology}, vol=143, pages= {1059--1068} } @article{Oakes93, author={Oakes, D. and A.J. Moss and J.T. Fleiss and J.T. Bigger, Jr. and T.M. Therneau and S.W. Eberly and M.P. McDermott and A. Manatunga and E. Carleen and J. Benhorin, and {the Multicenter Diltiazem Post-Infarction Research Group}}, year = {1993}, title = {Use of compliance measures in and analysis of the effect of {D}iltiazem on mortality and reinfarction after myocardial infarction}, journal=jasa, volume={88}, pages = {44-49} } @incollection{Oakes92, author= {Oakes, D.}, year= {1992}, title ={Frailty models for multiple event times}, editor= {Klein, J. P. and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, publisher= {Kluwer}, address= {Netherlands} } @article{Omori93, author={Omori,Y. and Johnson,R. A.}, title={The influence of random effects on the unconditional hazard rate and survival functions}, year={1993}, journal=biok,volume={80}, pages={910--914} } @phdthesis{Parner96, author={Parner, E.}, title={Inference in semiparametric frailty models}, year ={1997}, school={University of Aarhus, Denmark} } @article{Pettitt90, author={Pettitt, A. N. and Bin Daud, I.}, title={Investigating time dependence in {C}ox's proportional hazards model}, year= {1990}, journal=applstat, volume={39},pages={313--329} } @article{Prentice86, author= {Prentice, R. L.}, year= {1986}, title ={A case-cohort design for epidemilogic cohort studies and disease prevention trials}, journal={Biometrika}, volume={73}, pages ={1--11} } @article{Prentice92, author= {Prentice, R. L. and Cai, J.}, year= {1992}, title ={Covariance and survivor function estimation using censored multivariate failure time data}, journal={Biometrika}, volume={79}, pages ={495--512} } @incollection{Prentice91, author= {Prentice, R. L. and Cai, J.}, year= {1991}, title ={Marginal and conditional models for the analysis of multivariate failure time data}, pages ={393--406}, editor= {Klein, J. P and Goel, P. K.}, booktitle= {Survival Analysis, State of the Art}, publisher= {Kluwer Academic Publishers}, address= {Netherlands} } @article{Prentice81, author= {Prentice, R. L. and Williams, B. J. and Peterson, A. V.}, year= {1981}, title= {On the regression analysis of multivariate failure time data}, journal={Biometrika}, volume={68}, pages= {373--379} } @book{Press88, author= {Press, W.H. and Teukolsky, S.A. and Vetterling, W.T. and Flannery, B.P.}, year= {1988}, title= {Numerical Recipes in {C}}, publisher= {Cambridge University Press}, address ={Cambridge} } @article{Quantin96, author= {Quantin, C. and Moreau, T. and Asselaiin B. and Maccario, J. and Lelloucj, J. } , title= {A regression survival model for testing the proportional hazards hypothesis }, year= {1996}, journal=biom, volume={52}, pages={874--885} } @article{Quigley89, author= {O'Quigley, J. and Pessione, F.}, year= {1989}, title= {Score tests for homogeneity of regression effect in the proportional hazards model}, journal={Biometrics}, volume={45}, pages ={135--144} } @article{Reid85, author= {Reid, N. and Cr{\'{e}}peau, H.}, year= {1985}, title= {Influence functions for proportional hazards regression}, journal={Biometrika}, volume={72}, pages= {1--9} } @article{Ricci97, author= {Ricci, P. and Therneau, T. M. and Malinchoc, M. and Benson, J. T. and Petz, J. L. and Klintmalm, G. B. and Crippin, J. S. and Wiesner, R. H. and Steers, J. L. and Rakela, J. and Starzl, T. E. and Dickson, E. R.}, year= {1997}, title= {A prognostic model for the outcome of liver transplantation in patients with cholestatic liver disease}, journal={Hepatology}, volume={25}, pages= {672--677} } @manual{SAS96, author={{SAS Institute Inc.}}, year={1996}, title={{SAS/STAT} Software: Changes and Enhancements through Release 6.11}, organization={{SAS} Institute, Inc.}, address={Cary, N.C.}, note={Chapter 8: The PHREG procedure} } @article{Sastry97, author= {Sastry, N.}, title={A nested frailty model for survival data, with an application to the study of child survival in northesat Brazil}, year={1997}, journal=jasa, volume={92}, pages = {426--435} } @article{Schoenfeld80, author= {Schoenfeld, D.}, year= {1980}, title= {Chi-squared goodness-of-fit tests for the proportional hazards regression model}, journal={Biometrika}, volume={67}, pages= {145--153} } @article{Schoenfeld81, author={Schoenfeld, D.}, title={The asymptotic properties of nonparametric tests for comparing survival distributions}, year={1981}, journal=biok, volume={68}, pages={316--319} } @article{Schoenfeld83, author={Schoenfeld, D. A.}, title={Sample-size formula for the proportional-hazards regression model}, year={1983}, journal=biom, volume={39}, pages={499--503} } @article{Schum87, author={Schumacher, M. and Olschewski, M. and Schmoor,C.}, title={The impact of heterogeneity on the comparison of survival times}, year={1987}, journal=statmed, volume={6},pages={773-784} } @article{Segal93, author= {Segal, M.R. and Neuhaus, J. M.}, year= {1993}, title= {Robust inference for multivariate survival data}, journal=statmed, volume={12}, pages= {1019--1031} } @article{Segal94, author= {Segal, M.R. and Baccheti, P. and Jewell, N. P.}, year= {1994}, title= {Variances for maximum penalized likelihood estimates obtained via the {EM} algorithm }, journal=jrssb, volume={56}, pages= {345--352} } @article{Simon11, title= {Regularization Paths for {C}ox’s Proportional Hazards Model via Coordinate Descent}, author= {Noah Simon and Jerome Friedman and Trevor Hastie Rob Tibshirani}, journal={J Statistical Software}, volume=39, pages={1--123} } @article{Smith96, author= {Smith, P.J. and Hietjan, D.F.}, year= {1996}, title= {Testing and adjusting for overdispersion in generalized linear models}, journal=jrssc, volume={?} } @article{Solomon84, author= {Solomon, P. J. } , title= {Effect of misspecification of regression models in the analysis of survival data}, year= {1984}, journal=biok, volume={71}, pages={291--298} } @article{Solomon86, author= {Solomon, P. J. } , title= {Amendments and corrections}, year= {1986}, journal=biok, volume={73}, pages={245--245} } @book{Spector94, author={Spector, P.}, title= {An Introduction to {S} and {S-Plus}}, publisher={Wadsworth}, address={Pacific Grove, CA}, year= {1994} } @article{Stablein81, author= {Stablein, D. M. and Carter, Jr., W. H. and Novak, J. W.}, title= {Analysis of survival data with nonproportional hazard functions}, year= {1981}, journal={Controlled Clinical Trials}, volume={2}, pages={149--159} } @inproceedings{Stone85, author= {Stone, C. J. and Koo, C. Y.} , title= { Additive splines in statistics}, year= {1985}, booktitle={Computational Statistics Section}, organization={American Statistical Association}, address={Alexandria, Virginia}, pages={646--651} } @article{Stone86, author= {Stone, C. J.} , title= {Comment to paper by {H}astie and {T}ibshirani}, year= {1986}, journal=statsci, volume={1}, pages={312--314} } @article{Storer85, author= {Storer, B. E. and Crowley, J.}, year= {1985}, title ={A diagnostic for {C}ox regression and general conditional likelihoods}, journal =jasa, volume={80}, pages= {139-147} } @article{Struthers86, author= {Struthers, C. A. and Kalbfleisch, J. D. } , title= {Misspecified proportional hazard models}, year= {1986}, journal=biok, volume={73}, pages={363--369} } @article{Therneau90, author= {Therneau, T. M. and Grambsch, P. M. and Fleming, T. R.}, year= {1990}, title= {Martingale based residuals for survival models}, journal={Biometrika}, volume={77}, pages= {147--160} } @article{Therneau97, author= {Therneau, T. M. and Hamilton, S. A.}, year= {1997}, title= {{rhDNase} as an example of recurrent event analysis}, journal=statmed, volume={16}, pages= {2029--2047} } @article{Therneau99, author= {Therneau, T. M. and Li, H.}, year= {1999}, title= {Computing the {C}ox model for case-cohort designs}, journal=lifetime, volume={5}, pages= {99--112} } @book{Therneau00, author={Therneau, T. M. and Grambsch, P. M.}, title= {Modeling Survival Data: Extending the {C}ox Model}, publisher={Springer-Verlag}, address={New York}, year= {2000} } @article{Therneau03, author= {Therneau, T. M. and Grambsch, P. M. and Pankratz, V. S.}, year= {2003}, title= {Penalized survival models and frailty}, journal={J Computational Graphical Statistics}, volume={12}, pages= {156--175} } @article{Thomsen91, author= {Thomsen, B. L. and Keiding, N. and Altman, D. G.}, year= {1991}, title= {A note on the calculation of expected survival, illustrated by the survival of liver transplant patients}, journal=statmed, volume={10}, pages= {733--738} } @article{Thomsen92, author= {Thomsen, B. L. and Keiding, N. and Altman, D. G.}, year= {1992}, title= {Reply to a letter to the editor}, journal=statmed, volume={11}, pages= {1528--1530} } @article{Uitti93, author= {Uitti, R.J. and Ahlskog, J.E. and Maraganore, D.M. and Muenter, M.D. and Atkinson, E.J. and Cha, R.H. and O'Brien, P.C.}, year= {1993}, title= {Levodopa therapy and survival in idiopathic Parkinson's disease: Olmsted County Project}, journal ={Neurology}, volume={43}, pages= {1918--1926} } @book{Venables97, author= {Venables, W. N. and Ripley, B. D.}, year= {1997}, title= {Modern Applied Statistics with {S-PLUS}, second edition}, publisher= {Springer-Verlag}, address ={New York} } @article{Verhuel93, author= {Verheul, H. A. and Dekker, E. and Bossuyt, P. and Moulijn, A. C. and Dunning, A. J.}, year= {1993}, title= {Background mortality in clinical survival studies}, journal ={Lancet}, volume={341}, pages= {872--875} } @article{Wahba83, author= {Wahba, G.}, year= {1983}, title= {Bayesian ''confidence intervals'' for the cross-validated smoothing spline}, journal =JRSSB, volume={45}, pages= {133--150} } @article{Wei89, author= {Wei, L. J. and Lin, D. Y. and Weissfeld, L.}, year= {1989}, title= {Regression analysis of multivariate incomplete failure time data by modeling marginal distributions}, journal=jasa, volume={84}, pages= {1065--1073} } @article{Wei93, author= {Lin, D. Y. and Wei, L. J. and Ying, Z.}, year= {1993}, title= {Checking the {C}ox model with cumulative sums of martingale-based residuals}, journal=biok, volume={80}, pages= {557--572} } @article{White80, author= {White, H.}, year= {1980}, title ={A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity}, journal= {Econometrica}, volume={48}, pages= {817--838} } @article{White82, author= {White, H.}, year= {1982}, title= {Maximum likelihood estimation of misspecified models}, journal= {Econometrica}, volume={50}, pages= {1--26} } @article{Whitehead80, author= {Whitehead, J.}, year= {1980}, title= {Fitting {C}ox's regression model to survival data using {GLIM}}, journal= applstat, volume={29}, pages= {268--275} } @unpublished{Winemiller98, author= {Winemiller, M.H. and Stolp-Smith, K.A, and Silverstein, M.D. and Therneau, T.M.}, year= {1998}, title= {Sequential pneumatic compression or heparin is effective in preventing venous thromboembolism in spinal cord injury patients}, note={Submitted} } @article{Winkler88, author= {Winkler, H. Z. and Rainwater, L. M. and Myers, R. P. and Farrow, G. M. and Therneau, T. M. and Zincke, H. and Lieber, M. M.}, year= {1988}, title= {Stage {D}1 Prostatic Adenocarcinoma: {S}ignificance of nuclear {DNA} ploidy patterns studied by flow cytometry}, journal= {Mayo Clinic Proceedings}, volume={63}, pages= {103--112} } @article{Yau97, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1997}, title= {Use of generalised linear mixed models for the analysis of clustered survival data}, journal= {Biometrical Journal}, volume={39}, pages= {3--11} } @article{Zhen94, author= {Zhen, B. and Murphy, J.R.}, year= {1994}, title= {Sample size determination for an exponential survival model with an unrestricted covariate}, journal=statmed, volume={13}, pages= {391--397} } @article{Aitkin80, author= {Aitkin,M. and Clayton, D.} , title= {The fitting of exponential, Weibull and extreme value distributions to complex censored survival data using GLIM}, year= {1980}, journal=applstat, volume={29}, pages={156--163} } @article{Bernstein78, author= {Berstein,D. and Lagakos, S. W.} , title= {Sample size and power determination for stratified clinical trials}, year= {1978}, journal=jscs, volume={8}, pages={65--73} } @article{Bie87, author= {Bie, O. and Borgan, O. and Liestoel, K.}, year = {1987}, title= {Confidence intervals and confidence bands for the cumulative hazard rate function and their small sample properties}, journal= scand, volume={14}, pages={221--233} } @book{Billingsley68, author={Billingsley, P.}, title= {Convergence of Probability Measures}, publisher={Wiley}, address={New York}, year= {1968} } @article{Block85, author= {Block, H. W. and Borges, W. S. and Savits, T. H.}, year= {1985}, title= {Age-dependent minimal repair}, journal=jap, volume={22}, pages={370--385} } @article{Breslow72, author= {Breslow, N. E.}, year= {1972}, title= {Discussion of {P}rofessor {C}ox's paper}, journal=jrssb, volume={34}, pages={216--217} } @book{Breslow80, author={Breslow, N. E. and Day, N. E.}, title= {The Analysis of Case-Control Studies}, volume={1}, series = {Statistical Methods in Cancer Research}, publisher={IARC}, address={Lyon}, year= {1980} } @book{glim2, author= {McCullagh, P. and Nelder, J.A.}, year= {1989}, title= {Generalized Linear Models, 2nd ed.}, publisher= {Chapman and Hall} } @incollection{Hettmansperger98, author= {Hettmansperger, T.}, year= {1998}, title= {Median}, editor={Armitage, P. and Colton, T.}, booktitle={Encyclopedia of Biostatics}, volume={4}, publisher= {Wiley}, address={New York}, pages={2525--2526} } @book{Huber81, author={Huber, P. J.}, title= {Robust Statistics}, publisher={Wiley}, address={New York}, year= {1981} } @article{Johansen83, author= {Johansen, S.}, year= {1983}, title= {An extension of {C}ox's regression model}, journal={Int. Stat. Review}, volume={51}, pages={165--174} } @book{Little87, author={Little, R.J.A. and Rubin, D.B.}, title= {Statistical Analysis with Missing Data}, publisher={John Wiley \& Sons}, address={New York}, year= {1987} } @incollection{Little98, author= {Little, R. J.}, year= {1998}, title= {Missing Data}, editor={Armitage, P. and Colton, T.}, booktitle={Encyclopedia of Biostatics}, volume={4}, publisher= {Wiley}, pages={2622--2635} } @article{Nelson69, author= {Nelson, W.}, year= {1969}, title= {Hazard plotting for incomplete failure data}, journal= {J. Quality Technology}, volume={1}, pages= {27--52} } @article{Peto72, author= {Peto, R.}, year= {1972}, title= {Discussion of {P}rofessor {C}ox's paper}, journal=jrssb, volume={34}, pages={205--207} } @article{Prentice78, author= {Prentice, R. L. and Gloeckler, L. A. } , title= {Regression analysis of grouped survival data with application to breast cancer data}, year= {1978}, journal=biom, volume={34}, pages={57--67} } @article{Self88, author= {Self, S. G. and Prentice, R. L.}, year= {1988}, title= {Asymptotic distribution theory and efficiency results for case-cohort studies}, journal=annals, volume={16}, pages= {64--81} } @techreport{Therneau94, author= {Therneau, T. M. and Sicks, J. and Bergstralh, E. and Offord, J.}, year = {1994}, title = {Expected survival based on hazard rates}, number = {52}, institution= {Department of Health Sciences Research, Mayo Clinic} } @article{Turnbull76, author= {Turnbull, B.W.}, year= {1976}, title= {The empirical distribution function with arbitrarily grouped, censored and truncated data}, journal={jrssb}, volume={38}, pages= {290--295} } @book{Breiman84, author= {Breiman, L. and Friedman, J. H. and Olshen, R. A. and Stone, C. J.}, title= {Classification and Regression Trees}, year = 1984, publisher={Wadsworth}, address={Belmont, CA} } @article{Breslow84, author= {Breslow, N. E. and L. Edler and J. Berger}, year= {1984}, title= {A two-sample censored-data rank test for acceleration}, journal=biom, volume={40}, pages= {1049--1062} } @book{Broca66, author= {Broca, P. P.}, title= {Traites de Tumerus, volumes 1 and 2}, year = 1866, publisher={Asselin}, address={Paris} } @article{Breslow74, author= {Breslow, N. E.}, year= {1974}, title= {Covariance analysis of censored survival data}, journal=biom, volume={30}, pages= {89--99} } @article{Bryson81, author= {Bryson, M. C. and Johnson, M. E}, year= {1981}, title= {The incidence of monotone likelihood in the {C}ox model}, journal={Technometrics}, volume={23}, pages= {381--383} } @article{Chang82, author= {Chang, I. M. and Gelman, R. and Pagano, M.}, year= {1982}, title= {Corrected group prognostic curves and summary statistics}, journal={J. Chronic Diseases}, volume={35}, pages= {669--674} } @inproceedings{Clarkson89, author= {Clarkson, D. B.}, year= {1989}, title= {Computing extended maximum likelihood estimates in monotone likelihood {C}ox proportional-hazards models}, booktitle={Computer Science and Statistics: Proceedings of the 21st Symposium on the Interface}, publisher={American Statistical Association}, address={Alexandria, Virginia}, pages={464--469} } @article{DeLong94, author= {D. M. DeLong and G. H. Guirguis and Y. C. So}, year= {1994}, title= {Efficient computation of subset selection probabilities with application to {C}ox regression}, journal=biok, volume={81}, pages= {607--611} } @book{Delwiche98, author={Delwiche, L. D. and S. J. Slaughter}, title= {The Little {SAS} Book}, publisher={SAS Institute}, address={Cary, NC}, year= {1998} } @article{Ducrocq96, author={V. Ducrocq and G. Casella}, title= {A {B}ayesian analysis of mixed survival models}, year= {1996}, journal={Genet. Sel. Evol.}, volume={28}, pages= {505--529} } @article{Gail81, author= {M. H. Gail and J. H. Lubin and L. V. Rubinstein}, year= {1981}, title= {Likelihood calculations for matched case-control studies and survival studies with tied death times}, journal=biok, volume={68}, pages= {703--707} } @article{Guo92, author= {G. Guo and G. Rodr\'{\i}guez}, year= {1992}, title= {Estimating a multivariate proportional hazards model for clustered data using the {EM} algorithm, with an application to child survival in {G}uatemala}, journal=jasa, volume={87}, pages= {969--976} } @article{Henderson99, author= {R. Henderson and P. Oman}, year= {1999}, title= {Effect of frailty on marginal regression estimates in survival analysis}, journal=jrssb, volume={61}, pages= {367--379} } @article{Hougaard86, author= {Hougaard, P.}, year= {1986}, title= {Survival models for heterogeneous populations derived from stable distributions}, journal=biok, volume={73}, pages= {387--396} } @article{Huster89, author= {W. J. Huster and R. Brookmeyer and S. G. Self}, year= {1989}, title= {Modelling paired survival data with covariates}, journal=biom, volume={45}, pages= {145--156} } @techreport{Jaeckel72, author= {Jaeckel, L.}, year= {1972}, title= {The infinitesimal jackknife}, institution={Bell Laboratories}, type={Memorandum}, number={MM 72-1215-11} } @article{Kavanagh94, author= {Kavanagh, B. F. and Wallrichs, S. and Dewitz, M. and Berry, D. and Currier, B. and Ilstrup, D. and Coventry, M. B.}, year= {1994}, title ={Charnley low-friction arthroplasty of the hip. {T}wenty-year results with cement}, journal={J. Arthroplasty}, volume={9}, pages= {229--234} } @article{Klein92, author= {Klein, J. P.}, year= {1992}, title ={Semiparametric estimation of random effects using the {C}ox model based on the {EM} algorithm}, journal=biom, volume={48}, pages= {795--806} } @article{Kyle93, author= {R. A. Kyle}, year= {1993}, title= {``{B}enign'' monoclonal gammopathy --- after 20 to 35 years of follow-up}, journal={Mayo Clinic Proceedings}, volume={68}, pages= {26--36} } @article{Kyle97, author= {R. A. Kyle}, year= {1997}, title= {Moncolonal gammopathy of undetermined significance and solitary plasmacytoma. {I}mplications for progression to overt multiple myeloma}, journal={Hematology/Oncology Clinics N. Amer.}, volume={11}, pages= {71--87} } @article{Leblanc92, author = {LeBlanc, M. and Crowley, J.}, title = {Relative risk trees for censored survival data}, journal = {Biometrics}, year = {1992}, volume={48}, pages = {411-425} } @article{Lee83, author= {K. L. Lee and Harrell, Jr., F. E. and H. D. Tolley and R. A. Rosati}, year= {1983}, title= {A comparison of test statistics for assessing the effects of concomitant variables in survival analysis}, journal=biom, volume={39}, pages= {341--350} } @article{Loprinzi94, author= {Loprinzi, C. L. and Laurie, J. A. and Wieand, H. S. and Krook, J. E. and Novotny, P. J. and Kugler, J. W. and Bartel, J. and Law, M. and Bateman, M. and Klatt, N. E. and Dose, A. M. and Etzell, P. S. and Nelimark, R. A. and Mailliard, J. A. and Moertel, C. G.}, year= {1994}, title= {Prospective evaluation of prognostic variables from patient-completed questionnaires}, journal={J. Clinical Oncol.}, volume={12}, pages= {601--607} } @article{Mahe99, author= {C{\'{e}}dric Mah{\'{e}} and Sylvie Chevret}, year= {1999}, title= {Estimating regression parameters and degree of dependence for multivariate failure time data}, journal=biom, volume={55}, pages= {1078--1084} } @article{Makuch82, author= {Makuch, R. W.}, year= {1982}, title= {Adjusted survival curve estimation using covariates}, journal={J. Chronic Disease}, volume={35}, pages= {437--443} } @article{Markus89, author= {Markus, B. H. and Dickson, E. R. and Grambsch, P. M. and Fleming, T. R. and Mazzaferro, V. and Klintmalm, G. B .G. and Wiesner, R. H. and VanThiel, D. H. and Starzl, T. E.}, year= {1989}, title= {Efficiency of liver transplantation in patients with primary biliary cirrhosis}, journal=NEJM, volume={320}, pages= {1709--1713} } @article{Miller83, author= {Miller, Jr., R.G.}, year= {1983}, title= {What price {Kaplan--Meier}?}, journal=biom, volume={39}, pages= {1077--1081} } @article{Murphy81, author= {Murphy, V. K. and Haywood, L. J.}, year= {1981}, title= {Survival analysis by sex, age group and hemotype in sickle cell disease}, journal={J. Chronic Diseases}, volume={34}, pages= {313--319} } @techreport{Pugh92, author= {M. Pugh and J. Robbins and S. Lipsitz and D. Harrington}, year= {1992}, title= {Inference in the {C}ox proportional hazards model with missing covariates}, institution={Department of Biostatistics, Harvard School of Public Health}, address={Boston}, number={758Z} } @article{Ripatti00, author= {Ripatti, S. and Palmgren, J.}, year= {2000}, title= {Estimation of multivariate frailty models using penalized partial likelihood}, journal=biom, volume={56}, pages={1016--1022} } @book{Searle71, author = {Searle, S.R.}, year= {1971}, title = {Linear Models}, publisher={Wiley}, address={New York}} @booklet{Seer81, key = {Surveillance}, title= {Surveillance, Epidemiology, and End Results: Incidence and Mortality Data, 1973--77}, year = {1981}, howpublished= {National Cancer Institute Monograph 57, U.S. Department of Health and Human Services, Public Health Service}, address= {National Cancer Institute, Bethesda, MD}, note = {NIH Publication No. 81-2330}, } @article{Sellers95, author= {T. A Sellers and V. E. Anderson and J. D. Potter and S. A. Bartow and P. L. Chen and L. Everson and R. A. King and C. C. Kuni and L. H. Kushi and P. G. McGovern and S. S. Rich and J. F. Whitbeck and G. L. Wiesner}, year= {1995}, title= {Epidemiologic and genetic follow-up study of 544 Minnesota breast cancer families: {D}esign and methods}, journal={Genetic Epidemiology}, volume={12}, pages= {417--429} } @article{Silverstein99, author= {M. D. Silverstein and Loftus, Jr., E. V. and W. J. Sandborn and W. J. Tremaine and B. G. Feagan and P. J. Nietert and W. S. Harmsen and A. R. Zinsmeister}, year= {1999}, title= {Clinical course and costs of care for {C}rohn's disease: {M}arkov model analysis of a population-based cohort}, journal={Gastroenterology}, volume={117}, pages= {49--57} } @article{Tsiatis81, author= {A. A. Tsiatis}, year= {1981}, title= {A large sample study of {C}ox's regression model}, journal= annals, volume={9}, pages= {93--108} } @article{Verweij94, author= {P. J .M. Verweij and Van Houwlingen, H. C.}, year= {1994}, title= {Penalized likelihood in {C}ox regression}, journal= statmed, volume={13}, pages= {2427--2436} } @ARTICLE{Volinsky98bayesianinformation, author = {Chris Volinsky and Adrian E. Raftery}, title = {Bayesian Information Criterion for Censored Survival Models}, journal = {Biometrics}, year = {1998}, volume = {56}, pages = {256--262} } @article{Wei90, author= {L. J. Wei and Z. Ying and D. Y. Lin}, year= {1990}, title= {Linear regression analysis of censored survival data based on rank tests}, journal= biok, volume={77}, pages= {845--851} } @article{Yau97, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1997}, title= {Use of generalised linear mixed models for the analysis of clustered survival data}, journal= {Biometrical Journal}, volume={39}, pages= {3--11} } @article{Yau98, author= {Yau, K. K .W. and McGilchrist, C. A.}, year= {1998}, title= {{ML} and {REML} estimation in survival analysis with time dependent correlated frailty}, journal= statmed, volume={17}, pages= {1201--1213} } @article{Zahl96, author= {Zahl, Per-Henrik}, title= {A linear non-parametric model for the excess intensity}, year= {1996}, journal=scand, volume={23}, pages={353--364} } @booklet{smoke90, key={Department of Health}, title={The Health Benefits of Smoking Cessation}, year={1990}, howpublished ={Department of Health and Human Services. Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health}, note={DHHS Publication No (CDC)90-8416} } @booklet{lifeus40, title={United States Life Tables and Actuarial Tables, 1939--41}, author={Thomas N. E. Greville}, howpublished={Federal Security Agency, United States Public Health Service, National Office of Vital Statistics}, note={U.S. Government Printing Office, 1947} } @booklet{lifeus50, key={United States Lifetables 1950}, title={United States Life Tables for 1949--51}, howpublished={U.S. Department of Health, Education and Welfare, Public Health Service, National Office of Vital Statistics} } @booklet{lifeus60, key={United States Lifetables 1960}, title ={United States Lifetables 1959--61}, year = {1964}, howpublished= {Public Health Service Publication No. 1252}, note={Volume 1, Number 1} } @booklet{lifeus60b, key={Vital Statistics}, title ={Vital Statistics of the United States, 1960}, year = {1963}, howpublished= {U.S. Department of Health, Education, and Welfare, Public Health Service, National Center for Vital Statistics}, note={Volume 2A, Tabl3 3B} } @booklet{lifeus70, key={United States Lifetables 1970}, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-115}, note={Volume 1, Number 1} } @booklet{lifeus80, key={United States Lifetables 1990}, title ={U.S. Decennial Lifetables 1979--81}, year = {1985}, howpublished= {DHEW Publication No. PHS 85-1150-1}, note={Volume 1, Number 1} } @booklet{lifest60, key = {United States Lifetables 1959--61}, title = {Life tables for the geographic divisions of the {U}nited {S}tates: 1959--61}, note = {Vol. 1, number 3}, howpublished={National Center for Health Statistics, Public Health Service, Washington, U.S. Government Printing Office}, month=May, year={1965} } @booklet{lifemn60, title ={Minnesota State Lifetables 1959--61}, year = {1965}, howpublished= {Public Health Service Publication No. 1252}, note={Volume 2, Number 24} } @booklet{lifemn70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 24} } @booklet{lifefl70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 10} } @booklet{lifeaz70, title ={U.S. Decennial Lifetables 1969--71}, year = {1975}, howpublished= {DHEW Publication No. HRA 75-1151}, note={Volume 2, Number 3} } @booklet{lifemn80, title ={U.S. Decennial Lifetables 1979--81}, year = {1985}, howpublished= {DHEW Publication No. PHS 86-1151-2451}, note={Volume 2, Number 24} } @book{Lifetable, author={National {C}enter for {H}ealth {S}tatistics}, year={1965}, title={Life tables for the geographic divisions of the {U}nited {S}tates: 1959-1961}, volume={1}, number={3}, publisher={US Government Printing Office, Washington} } survival/vignettes/.install_extras0000644000175100001440000000001012472644117017207 0ustar hornikusersfigures survival/vignettes/compete.Rnw0000644000175100001440000010003012536325722016300 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} %\VignetteIndexEntry{Multi-state models and competing risks} %\VignetteDepends{cmprsk} \SweaveOpts{keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} % I had been putting figures in the figures/ directory, but the standard % R build script does not copy it and then R CMD check fails \SweaveOpts{prefix.string=compete,width=6,height=4} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {compete-#1.pdf}} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=10) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #ensure default require("survival") @ \title{Multi-state models and competing risks} \author{Terry M Therneau \\ \emph{Mayo Clinic}} \newcommand{\code}[1]{\texttt{#1}} <>= cmplib <- require("cmprsk", quietly=TRUE) if (cmplib) cat("\\newcommand{\\CMPRSK}{}%\n") @ \begin{document} \maketitle \section{Multi-state survival curves} \begin{figure} \myfig{sfig1} \caption{Three multi-state models. In the upper left is simple survival, in the upper right an example of competing risks, with the multi-state illness-death model below them.} \label{sfig1} \end{figure} <>= par(mar=c(.1, .1, .1, .1)) frame() par(usr=c(0,100,0,100)) # first figure xx <- c(0, 10, 10, 0) yy <- c(0, 0, 10, 10) polygon(xx +10, yy+70) polygon(xx +30, yy+70) arrows( 22, 75, 28, 75, length=.1) text(c(15, 35), c(75,75), c("Alive", "Dead")) # second figure polygon(xx +60, yy+70) for (j in c(55, 70, 85)) { polygon(xx +80, yy+j) arrows(72, (5*75 +j+5)/6, 78, (100+j*5)/6, length=.1) } text(c(65, 85,85,85), c(70,55,70,85)+5, c("A", "D1", "D2", "D3")) # third figure polygon(xx+20, yy+25) for (j in c(15,35)) { polygon(xx +40, yy+j) arrows(32, (5*30 +j+4)/6, 38, (54+j*5)/6, length=.1) } arrows(38, 2+(55 + 35*5)/6, 32, 2+ (150 + 40)/6, length=.1) arrows(45, 33, 45, 27, length=.1) text(c(25, 45,45), c(30, 20, 40), c("Health", "Death", "Illness")) @ Consider the three simple models in figure \ref{sfig1}. Each box is a patient state and each arrow a possible transition. The top left figure is simple survival: all patients start in the alive state and can make a single transition to death. The top right depicts classic competing risks: all subjects start on the left, and each can make a single transition to one of 3 terminal states. The bottom figure shows a simple multi-state situation known as the illness-death model. Traditionally the first case is handled by the Kaplan-Meier esimate and the second by the ``cumulative incidence'', the third case requires use of the the Aalen-Johansen estimate, which includes each of the first two as a special case. The AJ estimate is very flexible: subjects can appear in more than one state during the course of a study, subjects can start after time 0 (delayed entry), and they can start in any of the states. The \code{survfit} function implements the AJ estimate and can handle all these cases. Let $A(t)$ be a matrix of cumulative hazard functions, whose $ij$ element is the estimated cumulative hazard for transitions from state $i$ to state $j$. $$ A_{ij}(t) = \int_0^t dN_{ij}(t)/r_i(t) $$ where $dN$ counts the transitions and $r$ is the number of subjects still at risk in a state. The diagonal elements of $A$ are filled in last such that row sums of $A$ are equal to zero. Then the Aalen-Johansen transition matrix is \begin{equation} P(t) = \prod_{s \le t} [I + dA(s)] \label{AJest} \end{equation} The product is over all time points $s \le t$ at which a transition occured, and $dA$ is the change in the $A$ matrix at that time point. For the two state model it is fairly easy to show that this reduces to the Kaplan-Meier. The $i$th row of $P(t)$ estimates the fraction of subjects in each state at time $t$, given that subjects started in state $i$. The solution obeys the obvious constraint that the row sums at any time are equal to 1: each person has to be somewhere. I will refer to the resulting values as \emph{prevalence} estimates. If there is no censoring then prevalence is particularly easy: at a given time just count the fraction of subjects in each state. In the simple two state model the prevalence in the alive state is the usual KM survival estimate, and we have P(alive) = 1 - P(dead). For simple survival we have gotten used to the idea of using P(dead) and 1- P(dead) interchangeably, but that habit needs to be left behind for multi-state models, for them the values $1-P_k$ = probability(any other state than $k$) are not very useful. Plots for the 2 state case sometimes choose to show P(alive) and sometimes P(dead). Which one is used often depends on a historical whim of the disease specialty; cardiology journals for instance quite often use P(event) resulting in curves that rise starting from zero, but oncology journals invariably use P(alive) giving curves that fall downhill from 1. The survfit routine's historical default for the 2 state case is to print and plot P(alive), which reflects that the author of the routine was working primarily in cancer trials at the time said default was chosen. In the multi-state case, however, the curve for the initial state (leftmost in my diagrams) is rarely included in the final plot and curves start at 0. Here is an example using a simple competing risks problem. The \code{mgus2} data set contains the time to plasma cell malignancy (PCM) and/or death for 1384 subjects diagnosed with monoclonal gammopathy of undetermined significance (MGUS). Survival and progression time are in months. The curve below shows ordinary Kaplan-Meier survival for these subjects, the mean age at diagnosis is just over 70 years. <>= oldpar <- par(mfrow=c(1,2)) hist(mgus2$age, nclass=30, main='', xlab="Age") with(mgus2, tapply(age, sex, mean)) mfit1 <- survfit(Surv(futime, death) ~ sex, data=mgus2) mfit1 plot(mfit1, col=c(1,2), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Survival") legend(6, .8, c("female", "male"), col=1:2, lwd=2, bty='n') par(oldpar) @ A second model for these subjects is competing risks, which corresponds to our second figure above. For this model we are only interested in the first event for each subject. Formally we are treating progression to a plasma cell malignancy (PCM) as an \emph{absorbing state}, i.e., one that subjects never exit. We create a variable \code{etime} containing the time of the first progression, death, or last follow-up along with an event variable that contains the outcome. The starting data set \code{mgus2} has two pairs of variables \code{(ptime, pstat)} that contain the time to progression and \code{(futime, status)} that contain the time to death or last known alive, ignoring progression. <>= etime <- with(mgus2, ifelse(pstat==0, futime, ptime)) event <- with(mgus2, ifelse(pstat==0, 2*death, 1)) event <- factor(event, 0:2, labels=c("censor", "pcm", "death")) table(event) mfit2 <- survfit(Surv(etime, event) ~ sex, data=mgus2) mfit2 plot(mfit2, col=c(1,1,2,2), lty=c(2,1,2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") legend(20, .6, c("death:female", "death:male", "pcm:female", "pcm:male"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') @ The \code{mfit2} call is nearly identical to that for an ordinary Kaplan-Meier, with the exception of the \code{event} variable. \begin{enumerate} \item The event variable was created as a \emph{factor}; whereas for ordinary single state survival the status is either 0/1 or TRUE/FALSE. The first level of the factor must be censoring, which is the status code for those whose follow-up terminated without reaching either endpoint. Codes for the remaining states can be in any order. The labels for the states are unrestricted. \item A simple print of the \code{mfit1} object shows the order in which the curves will be displayed. This information was used to choose the line types and colors for the curves. \item Since these are prevalence estimates, the curves start at 0. \end{enumerate} A common mistake with competing risks is to use the Kaplan-Meier separately on each event type while treating other event types as censored. The next plot is an example of this for the PCM endpoint. <>= pcmbad <- survfit(Surv(etime, pstat) ~ sex, data=mgus2) plot(pcmbad[2], mark.time=FALSE, lwd=2, fun="event", conf=FALSE, xscale=12, xlab="Years post diagnosis", ylab="Fraction with PCM") lines(mfit2[2,1], lty=2, lwd=2, mark.time=FALSE, conf=FALSE, xscale=12) legend(0, .28, c("Males, PCM, incorrect curve", "Males, PCM, competing risk"), col=1, lwd=2, lty=c(1,2), bty='n') @ There are two problems with the \code{pcmbad} fit. The first is that it attempts to estimate the expected rate of plasma cell malignancy if all other causes of death were disallowed. In this hypothetical world it is indeed true that many more subjects would progress to PCM, but it is not a world that any of us will ever inhabit and so is of questionable interest. The second problem is that the computation for this hypothetical case is only correct if all of the competing endpoints are independent, a situation which is almost never true. The competing risks curve estimates the fraction of MGUS subjects who will actually experience PCM, sometimes known as the lifetime risk. The above code chose to plot only a subset of the curves, something that is often desirable in competing risks problems to avoid a ``tangle of yarn'' plot that simply has too many elements. This is done by subscripting the survfit object. For subscripting, multistate curves appear as a matrix with the outcomes as the second subscript. They are in order of the levels of \code{event}, i.e., as displayed by our earlier call to \code{table(event)}. The first subscript indexes the groups formed by the right hand side of the model formula, and will be in the same order as simple survival curves. Thus \code{mfit[2,1]} corresponds to males and the pcm endpoint. A third example using the MGUS data treats it as a multi-state model. In this version a subject can have multiple transitions and thus multiple rows in the data set, and it is necessary to identify which data rows go with which subject via the \code{id} argument of \code{survfit} (valid estimates standard errors both depend on this). Our model looks like the illness-death model of figure \ref{sfig1} but with ``plasma cell malignancy'' as the upper state and no arrow for a return from that state to health. The necessary data set will have two rows for any subject who has further follow-up after a PCM and one row for all others. The data set is created below using the \code{tmerge} function, which is discussed in detail in another vignette. We need to decide what to do with the 9 subjects who have PCM and death declared at the same time. They slipped through without comment in the earlier competing risks analysis, only when setting up this data set did I notice the ties. Looking back at the code, the prior example counted these subjects as a progression. In retrospect this is defensible: even though undetected before autopsy, the disease must have been present for some amount of time previous and so progression did occur first. For the multi-state model we need to be explicit in how this is coded since a sojourn time of 0 within a state is not allowed. Below we push the progression time back by .1 month when there is a tie, but that amount is entirely arbitrary. <>= ptemp <- with(mgus2, ifelse(ptime==futime & pstat==1, ptime-.1, ptime)) newdata <- tmerge(mgus2, mgus2, id=id, death=event(futime, death)) newdata <- tmerge(newdata, mgus2, id, pcm = event(ptemp, pstat)) newdata <- tmerge(newdata, newdata, id, enum=cumtdc(tstart)) with(newdata, table(death, pcm)) @ The table above shows that there are no observations in \code{newdata} that have both a pcm and death, i.e., the ties have been resolved. The last tmerge line above creates a variable \code{enum} which simply counts rows for each person; it will be used later. <>= temp <- with(newdata, ifelse(death==1, 2, pcm)) newdata$event <- factor(temp, 0:2, labels=c("censor", "pcm", "death")) mfit3 <- survfit(Surv(tstart, tstop, event) ~ sex, data=newdata, id=id) plot(mfit3[,1], mark.time=FALSE, col=1:2, lty=1, lwd=2, xscale=12, xlab="Years post MGUS diagnosis", ylab="Prevalence of PCM") legend(4, .04, c("female", "male"), lty=1, col=1:2, lwd=2, bty='n') @ This plot is quite different in that it shows the fraction of subjects \emph{currently} in the PCM state. Looking at the lower scenario in figure \ref{sfig1}, this is the fraction of subjects in the upper right box. The curve goes up whenever someone enters the box and down when they leave. Myeloma survival was quite short during the era of this study and the proportion in the PCM state rarely rises above 2 percent. I have often found the three curve display below useful in these cases. It combines the results from competing risk model used above along with a second fit that treats death after PCM as a separate state from death before progression. Only males are shown in the plot to minimize overlap. <>= d2 <- with(newdata, ifelse(enum==2, 4, as.numeric(event))) e2 <- factor(d2, labels=c("censor", "pcm", "death w/o pcm", "death after pcm")) mfit4 <- survfit(Surv(tstart, tstop, e2) ~ sex, data=newdata, id=id) plot(mfit2[2,], lty=c(2,1), xscale=12, mark.time=FALSE, lwd=2, xlab="Years post diagnosis", ylab="Prevalence") lines(mfit4[2,3], mark.time=FALSE, xscale=12, col=2, lty=2, lwd=2, conf=FALSE) legend(15, .5, c("male:death w/o pcm", "male: ever pcm", "male: death after pcm"), col=c(1,1,2), lty=c(1,2,2), lwd=2, bty='n') @ When using multi-state data to create Aalen-Johansen estimates individuals are not allowed to have gaps in the middle of their time line. For example a data with (0, 30, pcm) and (50,70, death) as the two observations for a subject; the time from 30-70 is not accounted for. The method also does not account for what is known as panel data, where a subject's state is recorded at some a priori time such as a physician visit but the actual times of state transitions are unknown. Such data requires further assumptions about the transition process in order to model the outcomes, see for instance the msm package. \section{Models} For simple two-state survival the Cox model leads to three relationships \begin{align} \lambda(t) &= \lambda_0(t) e^{X\beta} \label{hazard} \\ \Lambda(t) &= \Lambda_0(t) e^{X\beta} \label{cumhaz}\\ S(t) &= \exp(-\Lambda(t)) \label{surv} \end{align} where $\lambda$, $\Lambda$ and $S$ are the hazard, cumulative hazard and survival functions, respectively. There is a single linear predictor which governs both the rate $\lambda$ (the arrow in figure \ref{sfig1}) and the prevalence value of the left hand box $S$. For multi-state models this simplicity no longer holds: proportional hazards does not lead to proportional prevalence curves. \subsection{Competing risks, Cox model} The Cox model approach starts by fitting separate models to each of the transitions. We will illustrate using the MGUS example. <>= mtemp <- mgus2 mtemp$age <- mtemp$age/10 #age in decades (easier coefficients) mtemp$etime <- etime mtemp$event <- event options(show.signif.stars = FALSE) # display intelligence cfit2 <- coxph(Surv(futime, death) ~ age + sex + mspike, data=mtemp) cfit2 @ The effect of age and sex on non-PCM mortality is profound, which is not a surprise given the median starting age of \Sexpr{median(mgus2$age)}. %$ Risk rises \Sexpr{round(exp(coef(cfit2)[1]),1)} fold per decade of age and the death rate for males is \Sexpr{round(exp(coef(cfit2)[2]),1)} times as great as that for females. The size of the serum monoclonal spike is of no consequence for this endpoint either statistically or clinically. <>= cfit1 <- coxph(Surv(ptime, pstat) ~ age + sex + mspike, mtemp) cfit1 quantile(mgus2$mspike, na.rm=TRUE) @ The mspike size has a major impact on progression, however; each 1 gram change increases risk by \Sexpr{round(exp(coef(cfit1)[3]) ,1)} fold. The interquartile range of \code{mspike} is 0.9 gram so this risk increase is clinically important. The effect of age on the progression rate is much less pronounced, with a coefficient only 1/4 that for mortality, while the effect of sex on progression is negligible. Notice that we did not do anything special to the data set or event codes for the Cox model. The focus of coxph is on the event rates, for which the correct denominator is the set of all subjects still at risk. This is exactly what is encoded by the (futime, death) and (ptime, pstat) pairs. The effect of sex on the \emph{lifetime} probability of PCM is not zero, however. Because of a longer lifetime, an average female with MGUS will spend more total years at risk for PCM than the average male, and so has a larger lifetime risk of PCM. The average rate of progression is about 1\% per year, as shown below, while the average post diagnosis lifetime is 18 months longer for females. <>= pfit1 <- pyears(Surv(ptime, pstat) ~ sex, mtemp, scale=12) round(100* pfit1$event/pfit1$pyears, 1) # PCM rate per year temp <- summary(mfit1, rmean="common") #print the mean survival time round(temp$table[,1:6], 1) @ Prevalence estimates from the multi-state model involve the matrix $A(t; x)$ of cumulative hazard estimates. The $i,j$ off diagonal element of $A(t;x)$ is the the cumulative hazard $\lambda_{ij}(t;x)$ for the $i \rightarrow j$ transition, obtained from the fitted Cox model for that transition. These predicted hazards are formed for a chosen set of covariates $x$, e.g. in the model above we could for instance choose predicted transitions for a 72 year old male with an mspike value of 1.1. Predicted curves from a Cox model are \emph{always} with respect to a particular hypothetical subject. The notion of a baseline hazard, i.e. the hazard for a subject with all covariates equal to zero, is sometimes of mathematical convenience but only rarely corresponds to any patient of interest. The diagonal elements of $A$ are filled in last and are chosen such that row sums of are 0. The obvious analog to the univariate survival curve in equation \eqref{surv} is the matrix exponential. \begin{equation*} P(t;x) = e^{A(t;x)} \label{matexp} \end{equation*} However, this computational approach is valid only if the $A$ matrix is separable, i.e., $A(t;x) = A(x) g(t)$, something that holds true if there are no time dependent covariates in the model and if all the transitions share the same baseline hazard: a very unusual case. The matrix exponential formulation is fundamental to multi-state models with constant hazard however, see for instance the vignette for the \code{msm} package. For the Cox model we use the Aalen-Johansen estimator --- the same approach used by \code{survfit} for non-parametric estimates. \begin{equation} P(t;x) = \prod_{s\le t} (I + dA(s;x)) \label{ajest} \end{equation} where the term $dA$ is the increment in $A$ at time $s$, and there is an increment at each event time. As with survival curves from an ordinary Cox model, any such curve is computed for a prespecified set of covariate values $x$ which must be chosen by the user. For illustration we will compute the probabilities of PCM from the model for males and females under 4 cases: age of 60 vs 80 and a serum mspike of 0.5 vs 1.5; these last are the approximately the quartiles of age and mspike. Each of surv1 and surv2 below will contain 8 curves, for the 8 combinations of sex, age and mspike. <>= tdata <- expand.grid(mspike=c(.5, 1.5), age=c(6,8), sex=c("F", "M")) surv1 <- survfit(cfit1, newdata=tdata) # time to progression curves surv2 <- survfit(cfit2, newdata=tdata) # time to death curves @ The individual survival curves are not actually of interest, since each is a Cox model analog of the `pcmbad' curve we criticised earlier. Instead, the cumulative hazard portion of the results are used to build an Aalen-Johansen estimate. The $A$ matrix is particularly easy in the competing risk case: all rows but the first will be 0, since only the $1\rightarrow 2$ and $1 \rightarrow 3$ transitions are possible. Elements of the resulting 3 by 3 matrix $P(t)$ are the probability of going from state $i$ to state $j$, since everyone starts in state 1 we are only interested in the first row of $P$. A computational nuisance is that the \code{surv1} and \code{surv2} curves do not necessarily jump at the same time. We use the summary function to select values on a common time scale. (The \code{summary.survfit} function was original written to provide printed values at specified times, but turns out to be an easy way to pluck off values.) <>= cifun <- function(surv1, surv2) { utime <- sort(unique(surv1$time, surv2$time)) jump1 <- diff(c(0, summary(surv1, times=utime, extend=TRUE)$cumhaz)) jump2 <- diff(c(0, summary(surv2, times=utime, extend=TRUE)$cumhaz)) dA <- diag(3) prev <- matrix(0., nrow= 1+length(utime), ncol=3) prev[1,1] <- 1 #initial prevalence at time 0: all are in the left box for (i in 1:length(utime)) { dA[1,2] <- jump1[i] #fill in the first row of dA(s) dA[1,3] <- jump2[i] dA[1,1] <- 1- (jump1[i] + jump2[i]) prev[i+1,] <- prev[i,] %*% dA } list(time=c(0, utime), P = prev) } # Get curves for the 8 cases, and save them in a matrix. # Since they all come from the same pair of Cox models, the time values # for all curves will be the same # The cifun function above is only designed to handle one of the 8 covariate # patterns at a time, but survival curves can be subscripted. temp <- cifun(surv1[1], surv2[1]) coxtime <- temp$time coxdeath <- coxpcm <- matrix(0., nrow=length(temp$time), ncol=8) coxdeath[,1] <- temp$P[,3] coxpcm[,1] <- temp$P[,2] for (i in 2:8){ temp <- cifun(surv1[i], surv2[i]) coxdeath[,i] <- temp$P[,3] coxpcm[,i] <- temp$P[,2] } # Print out a M/F results at 20 years indx <- match(20*12, coxtime) progmat <- matrix(coxpcm[indx,], nrow=4) dimnames(progmat) <- list(c("a=50/ms=0.5", "a=50/ms=1.5", "a=80/ms=0.5", "a=80/ms=1.5"), c("female", "male")) round(100*t(progmat), 1) #males and females at 20 years @ The above table shows that females are modeled to have a higher risk of 20 year progression, even though their hazard at any given moment is nearly identical to males. The difference at 20 years is on the order of our ``back of the napkin'' person-years estimate of 1\% progression per year * 1.5 more years of life for the females, but the progression fraction varies substantially by group. Eighty year olds have a lower cumulative rate of PCM than 50 year olds due to a higher death rate, even though the hazard function for PCM rises with age. A plot of the calculated progression curves is shown below. The left hand panel has predicted curves for those with a serum mspike of 0.5 and the right for mspike=1.5, and in all cases females are predicted to have a higher level of observed progression than males. Although the Cox model hazards are assumed to be proportional, the prevalence curves are not, however. For those diagnosed at an older age the prevalence curves flatten out after 10 years, simply because so few living subjects remain who are available to have a PCM event. <>= par(mfrow=c(1,2)) matplot(coxtime/12, coxpcm[,c(1,3,5,7)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, ylim=range(coxpcm), xlab="Years", ylab="Progression to PCM") legend(1, .23, c("Female: 60", "Male: 60", "Female: 80", "Male: 80"), lty=c(1,1,2,2), col=c(1,2,1,2), lwd=2, bty='n') matplot(coxtime/12, coxpcm[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="Progression to PCM") @ In the competing risks case the prevalence function has an alternate form known as the \emph{cumulative incidence} function \begin{equation} CI_k(t) = \int_0^t \lambda_k(u) S(u-) du \label{cuminc} \end{equation} where $\lambda_k$ is the incidence function for outcome $k$ and $S$ is the overall survival curve for ``time to any endpoint''. Proving that $P_{1k}$ as computed by Aalen-Johansen estimate is equivalent to $CI(t)$ is straightforward. (The label ``cumulative incidence'' is one of the more unfortunate ones in the survival lexicon, since we normally use `incidence' and `hazard' as interchangeable synonyms but the CI is \emph{not} a cumulative hazard.) For the general multi state case it is simplest to use the \code{mstate} package; it was designed for this task and will also compute appropriate confidence intervals. The latter are complex since they must account for the uncertainty in the underlying Cox models. \subsection{Fine-Gray model} \ifdefined\CMPRSK For the competing risk case the Fine-Gray model provides an alternate way of looking at the data. As we saw above, the impact of a particular covariate on the final prevalence values $P$ can be complex, even if the models for the hazards are relatively simple. Start with the functions $F_k(t) = P_{1k}(t)$, which can be thought of as the distribution function for the improper random variable $T^*= I(\mbox{endpoint}=k)T + I(\mbox{endpoint}\ne k)\infty$. Fine and Gray refer to $F_k$ as a subdistribution function. In an analog to the survival probability in the two state model define \begin{equation} \gamma_k(t) = - d \log[1-F_k(t)]/dt \label{FG}I \end{equation} and assume that $\gamma_k(t;x) = \gamma_{k0}(t) \exp(X\beta)$. In a 2 state model $\gamma$ is the usual hazard function. In the same way that our multivariate Cox model \code{cfit2} made the simplifying assumption that the impact of male sex is to increase the hazard for death by a factor of \Sexpr{round(exp(coef(cfit2)['sexM']), 2)} independent of the subject's age or serum mspike value, the Fine-Gray model assumes that each covariate's effect on $\log(1-F)$ is a constant, independent of other variables. Both model's assumptions are wonderfully simplifying with respect to understanding a covariate --- assuming of course that either assumption is correct. (In a multi-state model at least one of the two must be false.) Let us look at the effect of sex on PCM using the Fine-Gray model, which can be computed using the \code{cmprsk} package. It does not use model formulas so variables need to be vectors or matrices. <>= if (cmplib) { temp <- mtemp temp$fstat <- as.numeric(event) # 1=censor, 2=pcm, 3=death temp$msex <- with(temp, 1* (sex=='M')) fgfit1 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE)) fgfit2 <- with(temp, crr(etime, fstat, cov1=cbind(age, msex, mspike), failcode=3, cencode=1, variance=TRUE)) cmat <- rbind("FineGray: PCM" = fgfit1$coef, "Cox: PCM" = coef(cfit1), "FineGray: death" = fgfit2$coef, "Cox: death" = coef(cfit2)) round(cmat,2) } @ The program has determined that female sex increases the PCM outcome by exp(.21) = 1.24 fold, for all values of age and mspike. The Cox model shows no effect of sex on the instantaneous hazard, but as shown in the last section Cox models do predict higher female prevalence. We had also seen that older subjects are less likely to experience PCM due to the competing risk of death; this is reflected in the FG model as a negative coefficient for age. The primary strength of the Fine-Gray model with respect to the Cox model approach is that if ``lifetime risk'' is a primary question then the model has given us a simple and digestible answer to that question: females have a 1.2 fold higher risk. A primary problem of the model is that we can't go backwards: there is not a simple analog to the Aalen-Johansen estimator to carry one from $F$ back to $\Lambda$. If one fits a set of Cox models to the arrows (hazards) then the boxes (prevalence) of figure \ref{sfig1} can be examined post fit. With the Fine-Gray approach we have information only on the boxes. To compare the two fits we can look at what the female/male ratios for each of our four chosen age/mspike combinations, when $P$ is computed from the Cox models. <>= cox.f <- log(1- progmat) #log(1-P) round(cox.f[,1] / cox.f[,2], 2) @ The Cox models, which assume proportional hazards, show a larger subdistribution hazard for those who are older, those with higher mspike values, and at longer follow-up times. The overall average, however, is similar to the single value that results from a Fine-Gray model. The predicted curves are however nuch different from those shown before for a Cox model; the Fine-Gray curves are displayed below with predictions for mspike=0.5 on the left and 1.5 on the right. <>= if (cmplib) { par(mfrow=c(1,2)) fdata <- model.matrix(~age + sex + mspike, data=tdata)[,-1] #remove intercept fpred <- predict(fgfit1, cov1=fdata) matplot(fpred[,1]/12, fpred[,c(2,4,6,8)], col=c(1,1,2,2), lty=c(1,2,1,2), ylim=range(fpred[,-1]), type='l', lwd=2, xlab="Years", ylab="FG predicted") legend(0, .22, c("Female, 60", "Male, 60","Female: 80", "Male, 80"), col=c(1,2,1,2), lty=c(1,1,2,2), lwd=2, bty='n') matplot(fpred[,1]/12, fpred[,c(3,5,7,9)], col=c(1,1,2,2), lty=c(1,2,1,2), type='l', lwd=2, xlab="Years", ylab="FG predicted") } @ This tells a very different story than the Cox model prevalence curves. Which is correct? Individual non-parametric prevalence curves are not as helpful as one would hope: there are simply too few progression events when separated into 8 groups. <>= if (cmplib) fgfit3 <- with(temp, crr(etime, fstat, cov1= cbind(age, msex, mspike), failcode=2, cencode=1, variance=TRUE, cov2=msex, tf = function(x) log(x))) @ A deeper analysis is called for, but will have to be left for another day. \else This section requires the cmprsk library, so was not created. \fi \section{Conclusions} When working with acute disease such as advanced cancer or end-stage liver disease there is often a single dominating endpoint. Ordinary single event Kaplan-Meier curves and Cox models are then efficient and sufficient tools for much of the analysis. Such data was the primary use case for survival analysis earlier in the author's career. Data with multiple important endpoints is now common, and multi-state methods are an important addition to the statistical toolbox. As shown above, they are now readily available and easy to use. It is sometimes assumed that the presence of competing risks \emph{requires} the use of a Fine-Gray model (I have seen it in referee reports), but this is not correct. The model may often be useful, but is one available option among many. Grasping the big picture for a multi-state data set is always a challenge and we should make use of as many tools as possible. We are often minded of the story of a centerian on his 100th birthday proclaiming that he was looking forward to many more years because ``I read the obituaries every day, and you almost never see someone over 100 there''. It is not always easy to reason correctly from cumulative deaths back to hazard rates. An advantage of the Cox model is that it has better diagnostic tools available, e.g., evaluation of the proportional hazards assumption via \code{cox.zph} or the martingale residuals, which can help to further refine our understanding. It is also easier to link hazard rates to a biologic rationale (perhaps incorrectly) which can help in explaining a data set. \end{document} survival/vignettes/tests.Rnw0000644000175100001440000025653512534310123016017 0ustar hornikusers\documentclass{article}[11pt] \usepackage{Sweave} \usepackage{amsmath} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} %\VignetteIndexEntry{Cox models and ``type 3'' Tests} \SweaveOpts{prefix.string=tests,width=6,height=4, keep.source=TRUE, fig=FALSE} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{width=6,height=4} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default @ \title{Populations, Cox models, and ``type III'' tests} \author{Terry M Therneau \\ \emph{Mayo Clinic}} \newcommand{\code}[1]{\texttt{#1}} \newcommand{\myfig}[1]{\includegraphics[height=!, width=\textwidth] {tests-#1.pdf}} \newcommand{\ybar}{\overline{y}} \begin{document} \maketitle \tableofcontents \section{Introduction} This note started with an interchange on the R-help. A user asked ``how do I do a type III test using the Cox model'', and I replied that this was not a well defined question. If he/she could define exactly what it was that they were after, then I would look into it. To which the response was that ``SAS does it''. A grant deadline was looming so the discussion did not get any further at that point, but it eventually led to a much longer investigation on my part, which is summarized in this note. There are three central ideas as it turns out: populations, computation, and the mapping linear models ideas onto the Cox model. The first idea, and perhaps the central one, is using the model fit from a current data set to predict for a new population. This plays an important role in predicted survival curves, see for instance the vignette on that topic or chapter 10 of our book \cite{Therneau00}; recognizing that ``type 3'' tests are simply another variant on that theme was a pivotal step in my understanding. This immediately leads to the important subtopic of ``prediction for \emph{which} population''. The SAS type 3 computations corresponds to a very particular and inflexible choice. The second theme is computational: given some summary measure and a population for which you wish to predict it, the result will be some sort of weighted average. There are two primary ways to set up this computation. In a linear model one of them can be reduced to a particular contrast $C \hat\beta$ in the fitted coefficients $\hat\beta$, which is an appealing choice since follow-up computations such as the variance of the estimate become particularly simple. A common, simple, but unreliable algorithm for creating $C$ has been a major source of confusion (hereafter referred to as the NSTT: not safe type three). The last theme is how the linear models formulae map to the Cox model case. In particular, there is a strong temptation to use $C \hat\beta$ with $C$ taken from linear models machinery and $\hat\beta$ from a fitted Cox model. The problem is that this implicitly requires a replacement of $E[\exp(X)]$ with $\exp(E[X])$. For a Cox model $C \beta$ is certainly a valid statistic for any $C$, we just have no clear idea of what it is testing. For the impatient readers among you I'll list the main conclusions of this report at the start. \begin{itemize} \item SAS type 3 predicts for a population with a uniform distribution across all categorical predictors. Scholarly papers discussing fundamental issues with using such an approach as a default analysis method have appeared almost biannually in the statistics literature, with little apparent effect on the usage of the method. SAS documentation of type 3 is almost entirely focused on the algorithm they use for computing $C$ and ignores the population issue. \item Population predictions very often make sense, including the question the type 3 approach is attempting to address. There are valid ways to compute these estimates for a Cox model, they are closely related the inverse probability weight (IPW) methods used in propensity scores and marginal structural models. \item The algorithm used to compute $C$ by the SAS glm procedure is sophisticated and reliable. The SAS phreg procedure uses the linear models approach of $C \hat\beta$ to compute a ``type 3'' contrast, with $C$ computed via the NSTT. The combination is a statistical disaster. (This is true for SAS version 9.4; I will update this note if things change.) \end{itemize} \section{Linear approximations and the Cox model} \label{sect:transfer} One foundation of my concern has to do with the relationship between linear models and coxph. The solution to the Cox model equations can be represented as an iteratively reweighted least-squares problem, with an updated weight matrix and adjusted dependent variable at each iteration, rather like a GLM model. This fact has been rediscovered multiple times, and leads to the notion that since the last iteration of the fit \emph{looks} just like a set of least-squares equations, then various least squares ideas could be carried over to the proportional hazards model by simply writing them out using these final terms. In practice, sometimes this works and sometimes it doesn't. The Wald statistic is one example of the former type, which is completely reliable as long as the coefficients $\beta$ are not too large\footnote{ In practice failure only occurs in the rare case that one of the coefficients is tending to infinity. However, in that case the failure is complete: the likelihood ratio and score tests behave perfectly well but the Wald test is worthless.}. A counter example is found in two ideas used to examine model adequacy: adjusted variable plots and constructed variable plots, each of which was carried over to the Cox model case by reprising the linear-model equations. After a fair bit of exploring I found neither is worth doing \cite{Therneau00}. Copying over a linear models formula simply did not work in this case. \begin{figure} \myfig{data} \caption{Average free light chain for males and females. The figure shows both a smooth and the means within deciles of age.} \label{fig:data} \end{figure} \section{Data set} We will motivate our discussion with the simple case of a two-way analysis. The \code{flchain} data frame contains the results of a small number of laboratory tests done on a large fraction of the 1995 population of Olmsted County, Minnesota aged 50 or older \cite{Kyle06, Dispenzieri12}. The R data set contains a 50\% random sample of this larger study and is included as a part of the survival package. The primary purpose of the study was to measure the amount of plasma immunoglobulins and its components. Intact immunoglobulins are composed of a heavy chain and light chain portion. In normal subjects there is overproduction of the light chain component by the immune cells leading to a small amount of \emph{free light chain} in the circulation. Excessive amounts of free light chain (FLC) are thought to be a marker of disregulation in the immune system. Free light chains have two major forms denoted as kappa and lambda, we will use the sum of the two. An important medical question is whether high levels of FLC have an impact on survival, which will be explored using a Cox model. To explore linear models we will compare FLC values between males and females. A confounding factor is that free light chain values rise with age, in part because it is eliminated by the kidneys and renal function declines with age. The age distribution of males and females differs, so we will need to adjust our simple comparison between the sexes for age effects. The impact of age on mortality is of course even greater and so correction for the age imbalance is is critical when exploring the impact of FLC on survival. Figure \ref{fig:data} shows the trend in free light chain values as a function of age. For illustration of linear models using factors, we have also created a categorical age value using deciles of age. The table of counts shows that the sex distribution becomes increasingly unbalanced at the older ages, from about 1/2 females in the youngest group to a 4:1 ratio in the oldest. <>= library(survival) library(splines) age2 <- cut(flchain$age, c(49, 59, 69, 79, 89, 120), labels=c("50-59", "60-69", "70-79", "80-89", "90+")) counts <- with(flchain, table(sex, age2)) counts # flchain$flc <- flchain$kappa + flchain$lambda male <- (flchain$sex=='M') mlow <- with(flchain[male,], smooth.spline(age, flc)) flow <- with(flchain[!male,], smooth.spline(age, flc)) plot(flow, type='l', ylim=range(flow$y, mlow$y), xlab="Age", ylab="FLC") lines(mlow, col=2) cellmean <- with(flchain, tapply(flc, list(sex, age2), mean, na.rm=T)) matpoints(c(55,65,75, 85, 95), t(cellmean), pch='fm', col=1:2) round(cellmean, 2) @ Notice that the male/female difference in FLC varies with age, \Sexpr{round(cellmean[1,1],1)} versus \Sexpr{round(cellmean[2,1],1)} at age 50--59 and \Sexpr{round(cellmean[1,5],1)} versus \Sexpr{round(cellmean[2,5],1)} at age 90. The data does not fit a simple additive model; there are ``interactions'' to use statistical parlance. An excess of free light chain is thought to be at least partly a reflection of immune senescence, and due to our hormonal backgrounds men and women simply do not age in quite the same way. \section{Population averages} The question of how to test for a main effect in the presence of interaction is an old one. At one time this author considered the phrase ``main effect in the presence of interaction'' to be an oxymoron, but long experience with clinical data sets has led me to the opposite conclusion. Real data always has interactions. The treatment effect of a drug will not be exactly the same for old and young, thin and obese, physically active and sedentary, etc. Explicit recognition of this is an underlying rationale of the current drive towards ``personalized medicine'', though that buzzword often focuses only on genetic differences. Any given data set may often be too small to explore these variations and our statistical models will of necessity smooth over the complexity, but interactions are nevertheless still present. Consider the data shown in figure \ref{fig1} below, which shows a particular laboratory test value by age and sex. We see that the sex effect varies by age. Given this, what could be meant by a ``main effect'' of sex? One sensible approach is to select a fixed \emph{population} for the ages, and then compute the average sex effect over that population. Indeed this is precisely what many computations do behind the scenes, e.g. the ``type 3'' estimates found in linear models. There are three essential components to the calculation: a reference population for the confounders, a summary measure of interest, and a computational algorithm. To understand how linear models methods may (or may not) extend to the proportional hazards model it is useful consider all three facets; each is revealing. Four possible choices for a target population of ages are given below. \begin{enumerate} \item Empirical: the age distribution of the sample at hand, also called the data distribution. In our sample this would be the age distribution of all \Sexpr{nrow(flchain)} subjects, ignoring sex. \item SAS: a uniform distribution is assumed over all categorical adjusters, and the data distribution for continuous ones. \item External reference: a fixed external population, e.g. the age distribution of the US 2010 census. \item MVUE: minimum variance unbiased; the implicit population corresponding to a multivariate least squares fit. \end{enumerate} Method 3 is common in epidemiology, method 1 is found in traditional survey sampling and in other common cases as we will see below. The type 3 estimates of SAS correspond to population 2. If there an interaction between two categorical variables x1 and x2, then the uniform distribution is taken to be over all combinations formed by the pair, and similarly for higher order interactions. \section{Linear models and populations} If we ignore the age effect, then everyone agrees on the best estimate of mean FLC: the simple average of FLC values within each sex. The male-female difference is estimated as the difference of these means. This is what is obtained from a simple linear regression of FLC on sex. Once we step beyond this and adjust for age, the relevant linear models can be looked at in several ways; we will explore three of them below: contrasts, case weights, and nesting. This ``all roads lead to Rome'' property of linear models is one of their fascinating aspects, at least mathematically. \subsection{Case weights} \begin{figure} \myfig{pop} \caption{Three possible adjusting populations for the FLC data set, a empirical reference in black, least squares based one in red, and the US 2000 reference population as `u'.} \label{fig:pop} \end{figure} How do we form a single number summary of ``the effect of sex on FLC''? Here are four common choices. \begin{enumerate} \item Unadjusted. The mean for males minus the mean for females. The major problem with this is that a difference in age distributions will bias the result. Looking at figure \ref{fig:data} imagine that this were two treatments A and B rather than male/female, and that the upper one had been given to predominantly 50-65 year olds and the lower predominantly to subjects over 80. An unadjusted difference would actually reverse the true ordering of the curves. \item Population adjusted. An average difference between the curves, weighted by age. Three common weightings are \begin{enumerate} \item External reference. It is common practice in epidemiology to use an external population as the reference age distribution, for instance the US 2000 census distribution. This aids in comparing results between studies. \item Empirical population. The overall population structure of the observed data. \item Least squares. The population structure that minimizes the variance of the estimated female-male difference. \end{enumerate} \end{enumerate} The principle idea behind case weights is to reweight the data such that confounders become balanced, i.e., ages are balanced when examining the sex effect and sex is balanced when examining age. Any fitted least squares estimate can be rewritten as a weighted sum of the data points with weight matrix $W= (X'X)^{-1}X'$. $W$ has $p$ rows, one per coefficient, each row is the weight vector for the corresponding element of $\hat\beta$. So we can backtrack and see what population assumption was underneath any given fit by looking at the weights for the relevant coefficient(s). Consider the two fits below. In both the second coefficient is an estimate of the overall difference in FLC values between the sexes. (The relationship in figure \ref{fig:data} is clearly curved so we have foregone the use of a simple linear term for age; there is no point in fitting an obviously incorrect model.) Since $\beta_2$ is a contrast the underlying weight vectors have negative values for the females and positive for the males. <<>>= us2000 <- rowSums(uspop2[51:101,,'2000']) fit1 <- lm(flc ~ sex, flchain, x=TRUE) fit2 <- lm(flc ~ sex + ns(age,4), flchain, x=TRUE) c(fit1$coef[2], fit2$coef[2]) wt1 <- solve(t(fit1$x)%*%fit1$x, t(fit1$x))[2,] # unadjusted wt2 <- solve(t(fit2$x)%*%fit2$x, t(fit2$x))[2,] # age-adjusted table(wt1, flchain$sex) @ To reconstruct the implied population density, one can use the density function with \code{wt1} or \code{wt2} as the case weights. Examination of \code{wt1} immediately shows that the values are $-1/n_f$ for females and $1/n_m$ for males where $n_f$ and $n_m$ are number of males and females, respectively. The linear model \code{fit1} is the simple difference in male and female means; the implied population structure for males and females is the unweighted density of each. Because this data set is very large and age is coded in years we can get a density estimate for fit2 by simple counting. The result is coded below and shown in figure \ref{fig:pop}. The empirical reference and least squares reference are nearly identical. This is not a surprise. Least squares fits produce minimum variance unbiased estimates (MVUE), and the variance of a weighted average is minimized by using weights proportional to the sample size, thus the MVUE estimate will give highest weights to those ages with a lot of people. The weights are not \emph{exactly} proportional to sample size for each age. As we all know, for a given sample size $n$ a study comparing two groups will have the most power with equal allocation between the groups. Because the M/F ratio is more unbalanced at the right edge of the age distribution the MVUE estimate gives just a little less weight there, but the difference between it and the overall data set population will be slight for all but those pathological cases where there is minimal overlap between M/F age distributions. (And in that case the entire discussion about what ``adjustment'' can or should mean is much more difficult.) <>= us2000 <- rowSums(uspop2[51:101,,'2000']) tab0 <- table(flchain$age) tab2 <- tapply(abs(wt2), flchain$age, sum) matplot(50:100, cbind(tab0/sum(tab0), tab2/sum(tab2)), type='l', lty=1, xlab="Age", ylab="Density") us2000 <- rowSums(uspop2[51:101,,'2000']) matpoints(50:100, us2000/sum(us2000), pch='u') legend(60, .02, c("Empirical reference", "LS reference"), lty=1, col=1:2, bty='n') @ The LS calculation does a population adjustment automatically for us behind the scenes via the matrix algebra of linear models. If we try to apply population reference adjustment directly a problem immediately arises: in the US reference \Sexpr{round(100*us2000[46]/sum(us2000),2)}\% of the population is aged 95 years, and our sample has no 95 year old males; it is not possible to re weight the sample so as to exactly match the US population reference. This occurs in any data set that is divided into small strata. The traditional epidemiology approach to this is to use wider age intervals of 5 or 10 years. Weights are chosen for each age/sex strata such that the sum of weights for females = sum of weights for males within each age group (balance), and the total sum of weights in an age group is equal to the reference population. The next section goes into this further. An increasingly popular approach for producing results that are standardized to the empirical reference population (i.e. the data distribution) is to use a smoothed age effect, obtained through inverse probability weights which are based on logistic regression, e.g. in the causal models literature and propensity score literature. This approach is illustrated in a vignette on adjusted survival curves which is also in the survival package. \subsection{Categorical predictors and contrasts} When the adjusting variable or variables are categorical --- a factor in R or a class variable in SAS --- then two more aspects come into play. The first is that any estimate of interest can be written in terms of the cell means. Formally, the cell means are a \emph{sufficient statistic} for the data. For our data set and using the categorized variable \code{age2} let $\theta_{ij}$ parameterize these means. $$ \begin{tabular}{cccccc} &50--59 & 60--69 & 70-79 & 80-89 & 90+ \\ \hline Female & $\theta_{11}$ & $\theta_{12}$ & $\theta_{13}$& $\theta_{14}$& $\theta_{15}$ \\ Male & $\theta_{21}$ & $\theta_{22}$ & $\theta_{23}$& $\theta_{24}$ & $\theta_{25}$ \\ \end{tabular} $$ For a design with three factors we will have $\theta_{ijk}$, etc. Because it is a sufficient statistic, any estimate or contrast of interest can be written as a weighted sum of the $\theta$s. Formulas for the resulting estimates along with their variances and tests were worked out by Yates in 1934 \cite{Yates34} and are often referred to as a Yates weighted means estimates. For higher order designs the computations can be rearranged in a form that is manageable on a desk calculator, and this is in fact the primary point of that paper. (Interestingly, his computational process turns out to be closely related to the fast Fourier transform.) The second facet of categorical variables is that another adjustment is added to the list of common estimates: \begin{enumerate} \item Unadjusted \item Population adjusted \begin{enumerate} \item External reference \item Empirical (data set) reference \item Least squares \item Uniform. A population in which each combination of the factors has the same frequency of occurrence. \end{enumerate} \end{enumerate} The uniform population plays a special role in the case of designed experiments, where equal allocation corresponds to the optimal study design. The Yates estimates are particularly simple in this case. For a hypothetical population with equal numbers in each age category the estimated average FLC for females turns out to be $\mu_f = \sum_j \theta_{1j} /5$ and the male - female contrast is $\sum_j(\theta_{2j}-\theta_{1j})/5$. We will refer to these as the ``Yates'' estimates and contrast for an effect. Conversely, the estimated age effects, treating sex as a confounding effect and assuming an equal distribution of females and males as the reference population, gives an estimated average FLC for the 60-69 year olds of $\mu_{60-69}= (\theta_{12} + \theta_{22})/2$, and etc for the other age groups. We can obtain the building blocks for Yates estimates by using the interaction function and omitting the intercept. <>= yatesfit <- lm(flc ~ interaction(sex, age2) -1, data=flchain) theta <- matrix(coef(yatesfit), nrow=2) dimnames(theta) <- dimnames(counts) round(theta,2) @ For a linear model fit, any particular weighted average of the coefficients along with its variance and the corresponding sums of squares can be computed using the \code{contrast} function given below. Let $C$ be a contrast matrix with $k$ rows, each containing one column per coefficient. Then $C\theta$ is a vector of length $k$ containing the weighted averages and $V = \hat\sigma^2 C (X'X)^{-1}C'$ is its variance matrix. The sums of squares is the increase in the sum of squared residuals if the fit were restricted to the subspace $C\theta =0$. Formulas are from chapter 5 of Searle \cite{Searle71}. Some authors reserve the word \emph{contrast} for the case where each row of $C$ sums to zero and use \emph{estimate} for all others; I am being less restrictive since the same computation serves for both. <<>>= qform <- function(beta, var) # quadratic form b' (V-inverse) b sum(beta * solve(var, beta)) contrast <- function(cmat, fit) { varmat <- vcov(fit) if (class(fit) == "lm") sigma2 <- summary(fit)$sigma^2 else sigma2 <- 1 # for the Cox model case beta <- coef(fit) if (!is.matrix(cmat)) cmat <- matrix(cmat, nrow=1) if (ncol(cmat) != length(beta)) stop("wrong dimension for contrast") estimate <- drop(cmat %*% beta) #vector of contrasts ss <- qform(estimate, cmat %*% varmat %*% t(cmat)) *sigma2 list(estimate=estimate, ss=ss, var=drop(cmat %*% varmat %*% t(cmat))) } yates.sex <- matrix(0, 2, 10) yates.sex[1, c(1,3,5,7,9)] <- 1/5 #females yates.sex[2, c(2,4,6,8,10)] <- 1/5 #males contrast(yates.sex, yatesfit)$estimate # the estimated "average" FLC for F/M contrast(yates.sex[2,]-yates.sex[,1], yatesfit) # male - female contrast @ <>= # Create the estimates table -- lots of fits emat <- matrix(0., 6, 3) dimnames(emat) <- list(c("Unadjusted", "MVUE: continuous age", "MVUE: categorical age", "Empirical (data) reference", "US200 reference", "Uniform (Yates)"), c("est", "se", "SS")) #unadjusted emat[1,] <- c(summary(fit1)$coef[2,1:2], anova(fit1)["sex", "Sum Sq"]) # MVUE -- do the two fits fit2 <- lm(flc ~ ns(age,4) + sex, flchain) emat[2,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) fit2 <- lm(flc ~ age2 + sex, flchain) emat[3,] <- c(summary(fit2)$coef[6, 1:2], anova(fit2)["sex", "Sum Sq"]) #Remainder, use contrasts tfun <- function(wt) { cvec <- c(matrix(c(-wt, wt), nrow=2, byrow=TRUE)) temp <- contrast(cvec, yatesfit) c(temp$est, sqrt(temp$var), temp$ss) } emat[4,] <- tfun(colSums(counts)/sum(counts)) usgroup <- tapply(us2000, rep(1:5, c(10,10,10,10,11)), sum)/sum(us2000) emat[5,]<- tfun(usgroup) emat[6,] <- tfun(rep(1/5,5)) @ \begin{table} \centering \begin{tabular}{l|ccc} & estimate & sd & SS \\ \hline <>= temp <- dimnames(emat)[[1]] for (i in 1:nrow(emat)) cat(temp[i], sprintf(" &%5.3f", emat[i,1]),sprintf(" &%6.5f", emat[i,2]), sprintf(" & %6.1f", emat[i,3]), "\\\\ \n") @ \end{tabular} \caption{Estimates of the male-female difference along with their standard errors. The last 4 rows are based on categorized age.} \label{tab:allest} \end{table} Table \ref{tab:est} shows all of the estimates of the male/female difference we have considered so far along with their standard errors. Because it gives a much larger weight to the 90+ age group than any of the other estimates, and that group has the largest M-F difference, the projected difference for a uniform population (Yates estimate) yields the largest contrast. It pays a large price for this in terms of standard error, however, and is over twice the value of the other approaches. As stated earlier, any least squares parameter estimate can be written as a weighted sum of the y values. Weighted averages have minimal variance when all of the weights are close to 1. The unadjusted estimate adheres to this precisely and the data-reference and MVUE stay as close as possible to constant weights, subject to balancing the population. The Yates estimate, by treating every cell equally, implicitly gives much larger weights to the oldest ages. Table \ref{tab:est} shows the effective observation weights used for each of the age categories. <>= casewt <- array(1, dim=c(2,5,4)) # case weights by sex, age group, estimator csum <- colSums(counts) casewt[,,2] <- counts[2:1,] / rep(csum, each=2) casewt[,,3] <- rep(csum, each=2)/counts casewt[,,4] <- 1/counts #renorm each so that the mean weight is 1 for (i in 1:4) { for (j in 1:2) { meanwt <- sum(casewt[j,,i]*counts[j,])/ sum(counts[j,]) casewt[j,,i] <- casewt[j,,i]/ meanwt } } @ \begin{table} \centering \begin{tabular}{rlrrrrr} &&50--59& 60--69 & 70--79 & 80--89 & 90+ \\ \hline <>= tname <- c("Unadjusted", "Min var", "Empirical", "Yates") for (i in 1:2) { for (j in 1:4) { cat("&",tname[j], " & ", paste(sprintf("%4.2f", casewt[i,,j]), collapse= " & "), "\\\\\n") if (j==1) cat(c("Female", "Male")[i]) } if (i==1) cat("\\hline ") } @ \end{tabular} \caption{Observation weights for each data point corresponding to four basic approaches. All weights are normed so as to have an average value of 1.} \label{tab:est} \end{table} Looking at table \ref{tab:est} notice the per observation weights for the $\ge 90$ age group, which is the one with the greatest female/male imbalance in the population. For all but the unbalanced estimate (which ignores age) the males are given a weight that is approximately 3 times that for females in order to re balance the shortage of males in that category. However, the absolute values of the weights differ considerably. \subsection{Different codings} Because the cell means are a sufficient statistic, all of the estimates based on categorical age can be written in terms of the cell means $\hat\theta$. The Yates contrast is the simplest to write down: $$ \begin{tabular} {rrrrrr} & 50--59 & 60--69 & 70--79 & 80--89 & 90+ \\ \hline Female & -1/5 & -1/5 & -1/5 & -1/5 & -1/5 \\ Male & 1/5 & 1/5 & 1/5 & 1/5 & 1/5 \end{tabular} $$ %(Note that for calculating a sum of squares we will get the exact same %result from a matrix using $\pm 1$ rather than $\pm 1/5$; %the Yates contrast is often written this way.) For the data set weighting the values of 1/5 are replaced by $n_{+j}/n_{++}$, the overall frequency of each age group, where a $+$ in the subscript stands for addition over that subscript in the table of counts. The US population weights use the population frequency of each age group. The MVUE contrast has weights of $w_j/\sum w_j$ where $w_j = 1/(1/n_{1j} + 1/n_{2j})$, which are admittedly not very intuitive. $$ \begin{tabular}{rrrrrr} & 50--59 & 60--69 & 70--79 & 80--89 & 90+ \\ \hline <>= temp <- 1/colSums(1/counts) temp <- temp/sum(temp) cat("Female", sprintf(" & %5.3f", -temp), "\\\\ \n") cat("Male", sprintf(" & %5.3f", temp), "\\\\ \n") @ \end{tabular} $$ In the alternate model \code{y \textasciitilde sex + age2} the MVUE contrast is much simpler, namely (0, 1, 0,0,0,0,0), and can be read directly off the printout as $\beta/se(\beta)$. The computer's calculation of $(X'X)^{-1}$ has derived the ``complex'' MVUE weights for us without needing to lift a pencil. The Yates contrast, however, cannot be created from the coefficients of the simpler model at all. This observation holds in general: a contrast that is simple to write down in one coding may appear complicated in another, or not even be possible. The usual and more familiar coding for a two way model is \begin{equation} y_{ij} = \mu + \alpha_i + \beta_j + \gamma_{ij} \label{std} \end{equation} What do the Yates' estimates look like in this form? Let $e_i$ be the Yates estimate for row $i$ and $k$ the number of columns in the two way table of $\theta$ values. Then \begin{align*} e_i &= (1/k)\sum_{j=1}^k \theta_{ij} \\ &= \mu + \alpha_i + \sum_j \left(\beta_j + \gamma_{ij}\right)/k \end{align*} and the Yates test for row effect is \begin{align} 0 &= e_i - e_{i'} \quad \forall i,i' \nonumber \\ &= (\alpha_i - \alpha_{i'}) + (1/k)\sum_j(\gamma_{ij} - \gamma_{i'j}) \label{ycont} \end{align} Equation \eqref{std} is over determined and all computer programs add constraints in order to guarantee a unique solution. However those constraints are applied, however, equation \eqref{ycont} holds. The default in R is treatment contrasts, which use the first level of any factor as a reference level. Under this constraint the reference coefficients are set to zero, i.e., all coefficients of equations \eqref{std} and \eqref{ycont} above where $i=1$ or $j=1$. We have been computing the male - female contrast, corresponding to $i=2$ and $i'=1$ in equation \eqref{ycont}, and the Yates contrast for sex becomes $\alpha_2 + 1/5(\gamma_{22} +\gamma_{23} +\gamma_{24} +\gamma_{25})$. The code below verifies that this contrast plus the usual R fit replicates the results in table \ref{tab:allest}. <>= fit3 <- lm(flc ~ sex * age2, flchain) coef(fit3) contrast(c(0,1, 0,0,0,0, .2,.2,.2,.2), fit3) #Yates @ The usual constraint is SAS is to use the last level of any class variable as the reference group, i.e., all coefficients with $i=2$ or $j=5$ in equations \eqref{std} and \eqref{ycont} are set to zero. <>= options(contrasts=c("contr.SAS", "contr.poly")) sfit1 <- lm(flc ~ sex, flchain) sfit2 <- lm(flc ~ sex + age2, flchain) sfit3 <- lm(flc ~ sex * age2, flchain) contrast(c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2), sfit3) # Yates for SAS coding @ The appendix contains SAS code and output for the three models \code{sfit1, sfit2} and \code{sfit3} above. The \code{E3} option was added to the SAS model statements, which causes a symbolic form of the contrasts that were used for ``type III'' results to be included in the printout. Look down the column labeled ``SEX'' and you will see exactly the coefficients used just above, after a bit of SAS to English translation. \begin{itemize} \item The SAS printout is labeled per equation \eqref{std}, so L1= column 1 of the full $X$ matrix = intercept. L2 = column 2 = females, L3 = column 3 = males, L4= column 4 = age 50--59, etc. \item In the symbolic printout they act as though sum constraints were in force: the last column of age is labeled with a symbolic value that would cause the age coefficients to sum to zero. However, in actuality these coefficients are set to zero. The table of parameter estimates at the end of the printout reveals this; forced zeros have a blank for their standard error. \item When calculating the contrast one can of course skip over the zero coefficients, and the R functions do not include them in the coefficient vector. Remove all of these aliased rows from the SAS symbolic printout to get the actual contrast that is used; this will agree with my notation. \item The SAS printout corresponds to a female-male contrast and I have been using male-female for illustration. This changes the signs of the contrast coefficients but not the result. \end{itemize} The \code{estimate} statement in the SAS code required that all of the coefficients be listed, even the aliased ones (someone more proficient in SAS may know a way to avoid this and enter only the non-aliased values.) %A general principle is that a given hypothesis may be represented as %a simple contrast in one coding but be complex in another. %The unadjusted test is a trivial contrast in the sfit1 coding, but a %complex one in the sfit3 coding. %The Yates test cannot be expressed as a contrast using the sfit1 or sfit2 %coding, is simple and obvious in the cell means coding, and has %simple but non obvious coefficients in the sfit3 coding. %Que sera sera. So, how do we actually compute the Yates contrast in a computer program? We will take it as a give that no one wants to memorize contrast formulas. Appendix \ref{sect:coding} describes three algorithms for the computation. One of these three (NSTT) is completely unreliable, but is included because it is so often found in code. If one uses the sum constraints commonly found in textbooks, which corresponds to the \code{contr.sum} constraint in R and to \code{effect} constraints in SAS, and there are no missing cells, then the last term in equation \eqref{ycont} is zero and the simple contrast $\alpha_i =0$ will be equal to the Yates contrast for sex. I often see this method recommended on R help in response to the question of ``how to obtain type III'', computed either by use of the \code{drop1} command or the \code{Anova} function found within the car package, but said advice almost never mentions the need for this particular non-default setting of the contrasts option\footnote{The Companion to Applied Regression (car) package is designed to be used with the book of the same name by John Fox, and the book does clarify the need for sum constraints.}. When applied to other codings the results of this procedure can be surprising. <>= options(contrasts = c("contr.treatment", "contr.poly")) #R default fit3a <- lm(flc ~ sex * age2, flchain) options(contrasts = c("contr.SAS", "contr.poly")) fit3b <- lm(flc~ sex * age2, flchain) options(contrasts=c("contr.sum", "contr.poly")) fit3c <- lm(flc ~ sex * age2, flchain) # nstt <- c(0,1, rep(0,8)) #test only the sex coef = the NSTT method temp <- rbind(unlist(contrast(nstt, fit3a)), unlist(contrast(nstt, fit3b)), unlist(contrast(nstt, fit3c)))[,1:2] dimnames(temp) <- list(c("R", "SAS", "sum"), c("effect", "SS")) print(temp) # drop1(fit3a, .~.) @ For the case of a two level effect such as sex, the NSTT contrast under the default R coding is a comparison of males to females in the first age group \textbf{only}, and under the default SAS coding it is a comparison of males to females within the \textbf{last} age group. Due to this easy creation of a test statistic which has no relation to the global comparison one expects from the ``type 3'' label the acronym \emph{not safe type three}(NSTT) was chosen, ``not SAS'' and ``nonsense'' are alternate mnemonics. \subsection{Sums of squares and projections} \label{sect:anova} The most classic exposition of least squares is as a set of projections, each on to a smaller space. Computationally we represent this as a series of model fits, each fit summarized by the change from the prior fit in terms of residual sum of squares. <>= options(show.signif.stars = FALSE) #exhibit intelligence sfit0 <- lm(flc ~ 1, flchain) sfit1b <- lm(flc ~ age2, flchain) anova(sfit0, sfit1b, sfit2, sfit3) @ The second row is a test for the age effect. The third row of the above table summarizes the improvement in fit for the model with sex + age2 over the model with just age2, a test of ``sex, adjusted for age''. This test is completely identical to the minimum variance contrast, and is in fact the way in which that SS is normally obtained. The test for a sex effect, unadjusted for age, is identical to an anova table that compares the intercept-only fit to one with sex, i.e., the second line from a call to \code{anova(sfit0, sfit1)}. The anova table for a nested sequence of models $A$, $A+B$, $A + B +C$, \ldots has a simple interpretation, outside of contrasts or populations, as an improvement in fit. Did the variable(s) $B$ add significantly to the goodness of fit for a model with just $A$, was $C$ an important addition to a model that already includes $A$ and $B$? The assessment of improvement is based on the likelihood ratio test (LRT), and extends naturally to all other models based on likelihoods. The tests based on a target population (external, data population, or Yates) do not fit naturally into this approach, however. %Obtaining the Yates contrast using a sequential sums of squares approach %is possible but a bit contrived. %Our final fit in the table will be \code{sfit3}, but %the one prior to it needs to be from a constrained version of \code{sfit3}, %whose solution lies in the space spanned by the Yates contrast %$\beta_2 + \beta_7/5 + \beta_8/5 + \beta_9/5 + \beta_{10}/5 = 0$. %There is no simple way to write down an ordinary LS model equation that %will do this, and instead one must use one a program for constrained %linear regression; these are far less familiar. %There are many algorithms to fit a constrained linear regression, one is %to transform the problem as $X\beta = (XQ)(Q'\beta) = Z \phi$ %where $Q$ is an orthogonal transformation matrix. %If the first column of $Q$ is chosen as a scaled version of the Yates %contrast, then setting that contrast equal to zero is the same as %the constraint $\phi_1 =0$; it suffices to fit a model using all but the %first column of $Z$. \subsection{What is SAS type 3?} We are now in a position to fully describe the SAS sums of squares. \begin{itemize} \item Type 1 is the output of the ANOVA table, where terms are entered in the order specified in the model. \item Type 2 is the result of a two stage process \begin{enumerate} \item Order the terms by level: 0= intercept, 1= main effects, 2= 2 way interactions, \ldots. \item For terms of level k, print the MVUE contrast from a model that includes all terms of levels $0-k$. Each of these will be equivalent to the corresponding line of a sequential ANOVA table where the term in question was entered as the last one of its level. \end{enumerate} \item Type 3 and 4 are also a 2 stage process \begin{enumerate} \item Segregate the terms into those for which a Yates contrast can be formed versus those for which it can not. The second group includes the intercept, any continuous variables, and any factor (class) variables that do not participate in interactions with other class variables. \item For variables in the first group compute Yates contrasts. For those in the second group compute the type 2 results. \end{enumerate} \end{itemize} SAS has two different algorithms for computing the Yates contrast, which correspond to the \code{ATT} and \code{STT} options of the \code{yates} function. SAS describes the two contrast algorithms in their document ``The four types of estimable functions'' \cite{SASguide}, one of which defines type 3 and the other type 4. I found it very challenging to recreate their algorithm from this document. Historical knowledge of the underlying linear model algorithms used by SAS is a useful and almost necessary adjunct, as many of the steps in the document are side effects of their calculation. When there are missing cells, then it is not possible to compute a contrast that corresponds to a uniform distribution over the cells, and thus the standard Yates contrast is also not defined. The SAS type 3 and 4 algorithms still produce a value, however. What exactly this result ``means'' and whether it is a good idea has been the subject of lengthy debates which I will not explore here. Sometimes the type 3 and type 4 algorithms will agree but often do not when there are missing cells, which further muddies the waters. Thus we have 3 different tests: the MVUE comparison which will be close but not exactly equal to the data set population, Yates comparisons which correspond to a uniform reference population, and the SAS type 3 (STT) which prints out a chimeric blend of uniform population weighting for those factor variables that participate in interactions and the MVUE weighting for all the other terms. \subsection{Which estimate is best?} Deciding which estimate is the best is complicated. Unfortunately a lot of statistical textbooks emphasize the peculiar situation of balanced data with exactly the same number of subjects in each cell. Such data is \emph{extremely} peculiar if you work in medicine; in 30 years work and several hundred studies I have seen 2 instances. In this peculiar case the unadjusted, MVUE, empirical reference and Yates populations are all correspond to a uniform population and so give identical results. No thinking about which estimate is best is required. This has led many to avoid the above question, instead pining for that distant Eden where the meaning of ``row effect'' is perfectly unambiguous. But we are faced with real data and need to make a choice. The question has long been debated in depth by wiser heads than mine. In a companion paper to his presentation at the joint statistical meetings in 1992, Macnaughton \cite{Macnaughton92} lists 54 references to the topic between 1952 and 1991. Several discussion points recur: \begin{enumerate} \item Many take the sequential ANOVA table as primary, i.e., a set of nested models along with likelihood ratio tests (LRT), and decry all comparisons of ``main effects in the presence of interaction.'' Population weightings other than the LS one do not fit nicely into the nested framework. \item Others are distressed by the fact that the MVUE adjusting population is data dependent, so that one is ``never sure exactly what hypothesis being tested''. \item A few look at the contrast coefficients themselves, with a preference for simple patterns since they ``are interpretable''. \item No one approach works for all problems. Any author who proposes a uniform rule is quickly presented with counterexamples. \end{enumerate} Those in group 1 argue strongly against the Yates weighting and those in group 2 argue for the Yates contrast. Group 3 is somewhat inexplicable to me since any change in the choice of constraint type will change all the patterns. I fear that an opening phrase from the 1986 overview/review of Herr \cite{Herr86} is still apropos, ``In an attempt to understand how we have arrived at our present state of ignorance \ldots''. There are some cases where the Yates approach is clearly sensible, for instance a designed experiment which has become unbalanced due to a failed assay or other misadventure that has caused a few data points to be missing. There are cases such as the FLC data where the Yates contrast makes little sense at all --- the hypothetical population with equal numbers of 50 and 90 year olds is one that will never be seen--- so it is rather like speculating on the the potential covariate effect in dryads and centaurs. The most raucous debate has circled around the case of testing for a treatment effect in the presence of multiple enrolling centers. Do we give each patient equal weight (MVUE) or each center equal weight (Yates). A tongue-in-cheek but nevertheless excellent commentary on the subject is given by the old curmudgeon, aka Guernsey McPearson \cite{Senn1, Senn2}. A modern summary with focus on the clinical trials arena is found in chapter 14 of the textbook by Senn \cite{Senn07} I have found two papers particularly useful in thinking about this. Senn \ref{Senn00} points out the strong parallels between tests for main effects when there may be interactions and meta analyses, cross connecting these two approaches is illuminating. A classic reference is the 1978 paper by Aitkin \cite{Aitkin78}. This was read before the Royal Statistical Society and includes remarks by 10 discussants forming a who's who of statistical theory (F Yates, J Nelder, DR Cox, DF Andrews, KR Gabriel, \ldots). The summary of the paper states that ``It is shown that a standard method of analysis used in many ANOVA programs, equivalent to Yates method of weighted squares of means, may lead to inappropriate models''; the paper goes on to carefully show why no one method can work in all cases. Despite the long tradition among RSS discussants of first congratulating the speaker and then skewering every one their conclusions, not one defense of the always-Yates approach is raised! This includes the discussion by Yates himself, who protests that his original paper advocated the proposed approach with reservations, it's primary advantage being that the computations could be performed on a desk calculator. I have two primary problems with the SAS type 3 approach. The first and greatest is that their documentation recommends the method with no reference to this substantial and sophisticated literature discussing strengths and weaknesses of the Yates contrast. This represents a level of narcissism which is completely unprofessional. %Recommending the type III approach as best for all cases, as they do, has %caused actual harm. The second is that their documentation explains the method is a way that is almost impenetrably opaque. If this is the only documentation one has, there will not be 1 statistician in 20 who would be able to explain the actual biological hypothesis which is being addressed by a type 3 test. \section{Cox models} \subsection{Tests and contrasts} Adapting the Yates test to a Cox model is problematic from the start. First, what do we mean by a ``balanced population''? In survival data, the variance of the hazard ratio for each particular sex/age combination is proportional to the number of deaths in that cell rather than the number of subjects. Carrying this forward to the canonical problem of adjusting a treatment effect for enrolling center, does this lead to equal numbers of subjects or equal numbers of events? Two centers might have equal numbers of patients but different number of events because one initiated the study at a later time (less follow up per subject), or it might have the same follow up time but a lower death rate. Should we reweight in one case (which one), both, or neither? The second issue is that the per-cell hazard ratio estimates are no longer a minimally sufficient statistic, so underlying arguments about a reference population no longer directly translate into a contrast of the parameters. A third but more minor issue is that the three common forms of the test statistic --- Wald, score, and LRT --- are identical in a linear model but not for the Cox model, so which should we choose? To start, take a look at the overall data and compute the relative death rates for each age/sex cell. <>= options(contrasts= c("contr.treatment", "contr.poly")) # R default cfit0 <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) cmean <- matrix(c(0, coef(cfit0)), nrow=2) cmean <- rbind(cmean, cmean[2,] - cmean[1,]) dimnames(cmean) <- list(c("F", "M", "M/F ratio"), dimnames(counts)[[2]]) signif(exp(cmean),3) @ Since the Cox model is a relative risk model all of the death rates are relative to one of the cells, in this case the 50--59 year old females has been arbitrarily chosen as the reference cell and so has a defined rate of 1.00. Death rates rise dramatically with age for both males and females (no surprise), with males always slightly ahead in the race to a coffin. The size of the disadvantage for males decreases in the last 2 decades, however. The possible ways to adjust for age in comparing the two sexes are \begin{enumerate} \item The likelihood ratio test. This is analogous to the sequential ANOVA table in a linear model, and has the strongest theoretical justification. \item A stratified Cox model, with age group as the stratification factor. This gives a more general and rigorous adjustment for age. Stratification on institution is a common approach in clinical trials. \item The Wald or score test for the sex coefficient, in a model that adjusts for age. This is analogous to Wald tests in the linear model, and is asymptotically equivalent the the LRT. \item The test from a reweighted model, using case weights. Results using this approach have been central to causal model literature, particularly adjustment for covariate imbalances in observational studies. (Also known as \emph{marginal structural models}). Adjustment to a uniform population is also possible. \item A Yates-like contrast in the Cox model coefficients. \begin{itemize} \item A reliable algorithm such as cell means coding. \item Unreliable approach such as the NSTT \end{itemize} \end{enumerate} I have listed these in order from the most to the least available justification, both in terms of practical experience and available theory. The two standard models are for sex alone, and sex after age. Likelihood ratio tests for these models are the natural analog to anova tables for the linear model, and are produced by the same R command. Here are results for the first three, along with the unadjusted model that contains sex only. <>= options(contrasts=c("contr.SAS", "contr.poly")) cfit1 <- coxph(Surv(futime, death) ~ sex, flchain) cfit2 <- coxph(Surv(futime, death) ~ age2 + sex, flchain) cfit3 <- coxph(Surv(futime, death) ~ sex + strata(age2), flchain) # Unadjusted summary(cfit1) # # LRT anova(cfit2) # # Stratified anova(cfit3) summary(cfit3) # # Wald test signif(summary(cfit2)$coefficients, 3) # anova(cfit1, cfit2) @ Without adjustment for age the LRT for sex is only \Sexpr{round(2*diff(cfit1$loglik),1)}, and after adjustment for %$ a it increases to \Sexpr{round(anova(cfit2)[3,2],2)}. Since females are older, not adjusting for age almost completely erases the evidence of their actual survival advantage. Results of the LRT are unchanged if we change to any of the other possible codings for the factor variables (not shown). Adjusting for age group using a stratified model gives almost identical results to the sequential LRT, in this case. The Wald tests for sex are equal to $[\beta/ se(\beta)]^2$ using the sex coefficient from the fits, \Sexpr{round(summary(cfit1)$coef[1,4]^2,2)} and \Sexpr{round(summary(cfit2)$coef[5,4]^2,2)} for the unadjusted and adjusted models, respectively. Unlike a linear model they are not exactly equal to the anova table results based on the log-likelihood, but tell the same story. Now consider weighted models, with both empirical and uniform distributions as the target age distribution. The fits require use of a robust variance, since we are approaching it via a survey sampling computation. The tapply function creates a per-subject index into the case weight table created earlier. <>= wtindx <- with(flchain, tapply(death, list(sex, age2))) cfitpop <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,3])[wtindx]) cfityates <- coxph(Surv(futime, death) ~ sex, flchain, robust=TRUE, weight = (casewt[,,4])[wtindx]) # # Glue it into a table for viewing # tfun <- function(fit, indx=1) { c(fit$coef[indx], sqrt(fit$var[indx,indx])) } coxp <- rbind(tfun(cfit1), tfun(cfit2,5), tfun(cfitpop), tfun(cfityates)) dimnames(coxp) <- list(c("Unadjusted", "Additive", "Empirical Population", "Uniform Population"), c("Effect", "se(effect)")) signif(coxp,3) @ The population estimates based on reweighting lie somewhere between the unadjusted and the sequential results. We expect that balancing to the empirical population will give a solution that is similar to the age + sex model, in the same way that the close but not identical to the MVUE estimate in a linear model. Balancing to a hypothetical population with equal numbers in each age group yields a substantially smaller estimate of effect. since it gives large weights to the oldest age group, where in this data set the male/female difference is smallest. Last, look at constructed contrasts from a cell means model. We can either fit this using the interaction, or apply the previous contrast matrix to the coefficients found above. Since the ``intercept'' of a Cox model is absorbed into the baseline hazard our contrast matrix will have one less column. <<>>= cfit4 <- coxph(Surv(futime, death) ~ sex * age2, flchain) # Uniform population contrast ysex <- c(0,-1, 0,0,0,0, -.2,-.2,-.2,-.2) #Yates for sex, SAS coding contrast(ysex[-1], cfit4) # Verify using cell means coding cfit4b <- coxph(Surv(futime, death) ~ interaction(sex, age2), flchain) temp <- matrix(c(0, coef(cfit4b)),2) # the female 50-59 is reference diff(rowMeans(temp)) #direct estimate of the Yates # temp2 <- rbind(temp, temp[2,] - temp[1,]) dimnames(temp2) <- list(c('female', 'male', 'difference'), levels(age2)) round(temp2, 3) # # # NSTT contrast contrast(c(1,0,0,0,0,0,0,0,0), cfit4) @ In the case of a two level covariate such as sex, the NSTT algorithm plus the SAS coding yields an estimate and test for a difference in sex for the \emph{first} age group; the proper contrast is an average. Since it gives more weight to the larger ages, where the sex effect is smallest, the Yates-like contrast is smaller than the result from an additive model \code{cfit2}. Nevertheless, this contrast and the sequential test are more similar for the survival outcome than for the linear models. This is due to the fact that the variances of the individual hazards for each sex/age combination are proportional to the number of deaths in that cell rather than the number of subjects per cell. A table of the number of deaths is not as imbalanced as the table of subject counts, and so the Yates and MLE ``populations'' are not as far apart as they were for the linear regression. There are fewer subjects at the higher ages but they die more frequently. Why is the Yates-like contrast so different than the result of creating a uniform age distribution using case weights followed by an MLE estimate? Again, the MLE estimate has death counts as the effective weights; the case-weighted uniform population has smaller weights for the youngest age group and that group also has the lowest death rate, resulting in lower influence for that group and an estimate shrunken towards the 90+ difference of \Sexpr{round(temp2[3,5], 3)}. All told, for survival models adjustment to a uniform population is a slippery target. \subsection{SAS phreg results} Now for the main event: what does SAS do? First, for the simple case of an additive model the SAS results are identical to those shown above. The coefficients, variances and log-likelihoods for cfit2 are identical to the phreg output for an additive model, as found in the appendix. As would be expected from the linear models case, the ``type III'' results for the additive model are simply the Wald tests for the fit, repackaged with a new label. Now look at the model that contains interactions. We originally surmised that a contrast calculation would be the most likely way in which the phreg code would implement type 3, as it is the easiest to integrate with existing code. Results are shown in the last SAS fit of the appendix. Comparing these results of the SAS printout labeled as ``Type III Wald'' to the contrasts calculated above shows that phreg is using the NSTT method. This is a bit of a shock. All of the SAS literature on type III emphasizes the care with which they form the calculation so as to always produce a Yates contrast (or in the case of missing cells a Yates-like one), and there was no hint in the documentation that phreg does anything different. As a double check direct contrast statements corresponding to the Yates and NSTT contrasts were added to the SAS code, and give confirmatory results. A further run which forced sum constraints by adding \code{'/ effect'} to the SAS class statement (not shown) restored the correct Yates contrast, as expected. As a final check, look at the NSTT version of the LRT, which corresponds to simply dropping the sex column from the $X$ matrix. <>= xmat4 <- model.matrix(cfit4) cfit4b <- coxph(Surv(futime, death) ~ xmat4[,-1], flchain) anova(cfit4b, cfit4) @ This agrees with the LR ``type 3'' test of the phreg printout. \subsection{Conclusion} Overall, both rebalanced estimates and coefficient contrasts are interesting exercises for the Cox model, but their actual utility is unclear. It is difficult to make a global optimality argument for either one, particularly in comparison to the sequential tests which have the entire weight of likelihood theory as a justification. Case reweighted estimates do play a key role when attempting to adjust for non-random treatment assignment, as found in the literature for causal analysis and marginal structural models; a topic and literature far too extensive and nuanced for discussion in this note. No special role is apparent, at least to this author, for regular or even sporadic use of a Yates contrast in survival models. The addition of such a feature and label to the SAS phreg package is a statistical calamity, one that knowledgeable and conscientious statistical practitioners will likely have to fight for the rest of their careers. In the common case of a treatment comparison, adjusted for enrolling center, the default ``type III'' printout from phreg corresponds to a comparison of treatments within the last center; the only contribution of the remainder of the data set is to help define the baseline hazard function and the effect of any continuous adjusters that happen to be in the model. The quadruple whammy of a third rate implementation (the NSTT), defaults that lead to a useless and misleading result, no documentation of the actual computation that is being done, and irrational reverence for the type III label conspire to make this a particularly unfortunate event. \appendix \section{Computing the Yates estimate} \label{sect:coding} We will take it as a given that no one wants to memorize contrast formulas, and so we need a way to compute Yates contrasts automatically in a computer program. The most direct method is to encode the original fit in terms of the cell means, as has been done throughout this report. The Yates contrast is then simply an average of estimates across the appropriate margin. However, we normally will want to solve the linear or Cox model fit in a more standard coding and then compute the Yates contrast after the fact. Note that any population re norming requires estimates of the cell means, whether they were explicit parameters or not, i.e., the model fit must include interaction terms. Here are three algorithms for this post-hoc computation. All of them depend, directly or indirectly, on the breakdown found earlier in equation \eqref{std}. \begin{align} y_{ij} &= \mu + \alpha_i + \beta_j + \gamma_{ij} + \epsilon \label{a1} \\ &= \theta_{ij} + \epsilon \label{a2}\\ \theta_{ij} &= \mu + \alpha_i + \beta_j + \gamma_{ij} \label{a3} \\ \end{align} Equation \eqref{a1} is the standard form from our linear models textbooks, equation \eqref{a2} is the cell means form, and \eqref{a3} is the result of matching them together. Using this equivalence a Yates test for row effects will be \begin{align} 0 &= e_i - e_{i'} \quad \forall i,i' \nonumber \\ &= (\alpha_i - \alpha_{i'}) + (1/k)\sum_j(\gamma_{ij} - \gamma_{i'j}) \label{ycont2} \end{align} where the subscripts $i$ and $i'$ range over the rows and $k$ is the number of columns. To illustrate the methods we will use 3 small data sets defined below. All are unbalanced. The second data set removes the aD observation and so has a zero cell, the third removes the diagonal and has 3 missing cells. <>= data1 <- data.frame(y = rep(1:6, length=20), x1 = factor(letters[rep(1:3, length=20)]), x2 = factor(LETTERS[rep(1:4, length=10)]), x3 = 1:20) data1$x1[19] <- 'c' data1 <- data1[order(data1$x1, data1$x2),] row.names(data1) <- NULL with(data1, table(x1,x2)) # data2 -- single missing cell indx <- with(data1, x1=='a' & x2=='D') data2 <- data1[!indx,] #data3 -- missing the diagonal data3 <- data1[as.numeric(data1$x1) != as.numeric(data1$x2),] @ \subsection{NSTT method} The first calculation method is based on a simple observation. If we impose the standard sums constraint on equation \eqref{a1} which is often found in textbooks (but nowhere else) of $\sum_i \alpha_i = \sum_j \beta_j = 0$, $\sum_i\gamma_{ij} =0 \; \forall j$ and $\sum_j \gamma_{ij} = 0 \; \forall i$, then the last term in equation \eqref{ycont2} is identically 0. Thus the Yates contrast corresponds exactly to a test of $\alpha=0$. In R we can choose this coding by using the \code{contr.sum} option. This approach has the appearance of simplicity: we can do an ordinary test for row effects within an interaction model. Here is R code that is often proposed for ``type III'' computation, which is based on the same process. <<>>= options(contrasts=c("contr.sum", "contr.poly")) fit1 <- lm(y ~ x1*x2, data1) drop1(fit1, .~.) @ The problem with this approach is that it depends critically on use of the sum constraints. If we apply the same code after fitting the data set under the more usual constraints a completely different value ensues. <<>>= options(contrasts=c("contr.SAS", "contr.poly")) fit2 <- lm(y ~ x1*x2, data1) drop1(fit2, .~.) options(contrasts=c("contr.treatment", "contr.poly")) fit3 <- lm(y ~ x1*x2, data1) drop1(fit3, .~.) @ Both common choices of contrasts give a different answer than contr.sum, and both are useless. I thus refer to this as the Not Safe Type Three (NSTT) algorithm, ``not SAS type three'' and ``nonsense type three'' are two other sensible expansions. This approach should NEVER be used in practice. \subsection{ATT} The key idea of the averaging approach (Averaged Type Three) is to directly evaluate equation \eqref{ycont2}. The first step of the computation is shown below <>= X <- model.matrix(fit2) ux <- unique(X) ux indx <- rep(1:3, c(4,4,4)) effects <- t(rowsum(ux, indx)/4) # turn sideways to fit the paper better effects yates <- effects[,-1] - effects[,1] yates @ The data set ux has 12 rows, one for each of the 12 unique x*x2 combinations. Because data1 was sorted, the first 4 rows correspond to x=1, the next 4 to x=2 and the next to x=3 which is useful for illustration but has no impact on the computation. The average of rows 1-4 (column 1 of \code{effects} above) is the estimated average response for subjects with x1=a, assuming a uniform distribution over the 12 cells. Any two differences between the three effects is an equivalent basis for computing the Yates contrast. We can verify that the resulting estimates correspond to a uniform target population by directly examining the case weights for the estimate. Each of them gives a total weight of 1/4 to each level of x2. Each element of $\beta\beta$ is a weighted average of the data, revealed by the rows of the matrix $(X'X)^{-1}X'$. The estimate are a weighted sum of the coefficients, so are also a weighted average of the $y$ values. <<>>= wt <- solve(t(X) %*% X, t(X)) # twelve rows (one per coef), n columns casewt <- t(effects) %*% wt # case weights for the three "row efffects" for (i in 1:3) print(tapply(casewt[i,], data1$x2, sum)) @ \subsection{STT} The SAS type III method takes a different approach, based on a a dependency matrix $D$. Start by writing the $X$ matrix for the problem using all of the parameters in equation \eqref{a1}. For our flc example this will have columns for intercept (1), sex (2), age group (5) and the age group by sex interaction (10) = 18 columns. Now define the lower triangular square matrix $D$ such that \begin{itemize} \item If the $i$th column of $X$ can be written as a linear combination of columns 1 through $i-1$, then row $i$ of $D$ contains that linear combination and $D_{ii}=0$. \item If the $i$th column is not linearly dependent on earlier ones then $D_{ii}=1$ and $D_{ij}=0$ for all $j \ne i$. \end{itemize} Columns of $D$ that correspond to linearly dependent columns of $X$ will be identically zero and can be discarded (or not) at this point. The result of this operation replicates table 12.2 in the SAS reference \cite{SASguide} labeled ``the form of estimable functions''. To obtain the Yates contrasts for an effect replace the appropriate columns of $D$ with the residuals from a regression on all columns to the right of it. Simple inspection shows that the columns of $D$ corresponding to any given effect will already be orthogonal to other effects in $D$ \emph{except} those for interactions that contain it; so the regression does not have to include all columns to the right. It is easy to demonstrate that this gives the uniform population contrast (Yates) for a large number of data sets, but I have not yet constructed a proof. (I suspect it could be approached using the Rao-Blackwell theorem.) \subsection{Bystanders} What about a model that has a extra predictor, such as \code{x3} in our example data and in the fit below? <<>>= fit4 <- lm(y ~ x1*x2 + x3, data=data1) @ The standard approach is to ignore this variable when setting up ``type III'' tests: the contrast for \code{x1} will be the same as it was in the prior model, with a 0 row in the middle for the x3 coefficient. \subsection{Missing cells} When there are combinations of factors with 0 subjects in that group, it is not possible to create a uniform population via reweighting of either subjects or parameters. There is thus no Yates contrast corresponding to the hypothetical population of interest. For that matter, adjustment to any fixed population is no longer possible, such as the US 2000 reference, unless groups are pooled so as to remove any counts of zero, and even then the estimate could be problematic due to extreme weights. This fact does not stop each of the above 3 algorithms from executing and producing a number. This raises two further issues. First, what does that number \emph{mean}? Much ink has been spilled on this subject, but I personally have never been able to come to grips with a satisfactory explanation and so have nothing to offer on the topic. I am reluctant to use such estimates. The second issue is that the computational algorithms become more fragile. \begin{itemize} \item The NSTT algorithm is a disaster in waiting, so no more needs to be said about situations where its behavior may be even worse. \item When fitting the original model, there will be one or more NA coefficients due to the linear dependencies that arise. A natural extension of the ATT method is to leave these out of the sums when computing each average. However, there are data sets for which the particular set of coefficients returned as missing will depend on the order in which variables were listed in the model statement, which in turn will change the ATT result. \item For the STT method, our statement that certain other columns in $D$ will be orthogonal to the chosen effect is no longer true. To match SAS, the orthogonalization step above should include only those effects further to the right that contain the chosen effect (the one we are constructing a contrast vector for). As a side effect, this makes the STT result invariant to the order of the variables in the model statement. \end{itemize} \section{SAS computations} The following code was executed in version 9.3 of SAS. \begin{verbatim} options ls=70; libname save "sasdata"; title "Sex only"; proc glm data=save.flc; class sex; model flc = sex; title "Sex only"; proc glm data=save.flc; class sex age2; model flc = age2 sex /solution E1 E2 E3; title "Second fit, no interaction"; proc glm data=save.flc; class sex age2; model flc = sex age2 sex*age2/solution E1 E2 E3; estimate 'yates' sex 1 -1 sex*age2 .2 .2 .2 .2 .2 -.2 -.2 -.2 -.2 -.2; title "Third fit, interaction"; proc phreg data=save.flc; class sex age2; model futime * death(0) = sex age2/ ties=efron; title "Phreg fit, sex and age, additive"; proc phreg data=save.flc; class sex age2; model futime * death(0) = sex age2 sex*age2 / ties=efron type3(all); estimate 'Yates sex' sex 1 sex*age2 .2 .2 .2 .2; contrast 'NSTT sex ' sex 1 ; contrast 'NSTT age' age2 1 0 0 0 , age2 0 1 0 0 , age2 0 0 1 0 , age2 0 0 0 1; title "Phreg fit, sex and age with interaction"; proc phreg data=save.flc; class sex age2/ param=effect; model futime * death(0) = sex age2 sex*age2 / ties=efron; title "Phreg, using effect coding"; \end{verbatim} The SAS output is voluminous, covering over a dozen pages. A subset is extracted below, leaving out portions that are unimportant to our comparison. First the GLM model for sex only. There are no differences between type 1 and type 3 output for this model. \small \begin{verbatim} ... Number of Observations Read 7874 Number of Observations Used 7874 ... Dependent Variable: flc Sum of Source DF Squares Mean Square F Value Model 1 142.19306 142.19306 42.27 Error 7872 26481.86345 3.36406 Corrected Total 7873 26624.05652 \end{verbatim} \normalsize The second fit with sex and then age. \small \begin{verbatim} Type I Estimable Functions -----------------Coefficients------------------ Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F -0.2571*L2-0.2576*L3-0.1941*L4-0.0844*L5 L7 sex M 0.2571*L2+0.2576*L3+0.1941*L4+0.0844*L5 -L7 Type II Estimable Functions ---Coefficients---- Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F 0 L7 sex M 0 -L7 Type III Estimable Functions ---Coefficients---- Effect age2 sex Intercept 0 0 age2 1 L2 0 age2 2 L3 0 age2 3 L4 0 age2 4 L5 0 age2 5 -L2-L3-L4-L5 0 sex F 0 L7 sex M 0 -L7 Dependent Variable: flc Sum of Source DF Squares Mean Square F Value Model 5 2212.13649 442.42730 142.60 Error 7868 24411.92003 3.10268 Corrected Total 7873 26624.05652 Source DF Type I SS Mean Square F Value age2 4 1929.642183 482.410546 155.48 sex 1 282.494304 282.494304 91.05 Source DF Type II SS Mean Square F Value age2 4 2069.943424 517.485856 166.79 sex 1 282.494304 282.494304 91.05 Source DF Type III SS Mean Square F Value age2 4 2069.943424 517.485856 166.79 sex 1 282.494304 282.494304 91.05 Standard Parameter Estimate Error t Value Pr > |t| Intercept 5.503757546 B 0.17553667 31.35 <.0001 age2 1 -2.587424744 B 0.17584961 -14.71 <.0001 age2 2 -2.249164537 B 0.17684133 -12.72 <.0001 age2 3 -1.770342603 B 0.17834253 -9.93 <.0001 age2 4 -1.082104827 B 0.18584656 -5.82 <.0001 age2 5 0.000000000 B sex F -0.383454133 B 0.04018624 -9.54 <.0001 sex M 0.000000000 B \end{verbatim} \normalsize The third linear models fit, containing interactions. For first portion I have trimmed off long printout on the right, i.e. the estimable functions for the age2*sex effect since they are not of interest. \small \begin{verbatim} Type I Estimable Functions --------------------Coefficients-------- Effect sex age2 Intercept 0 0 sex F L2 0 sex M -L2 0 age2 1 -0.0499*L2 L4 age2 2 -0.0373*L2 L5 age2 3 0.0269*L2 L6 age2 4 0.0482*L2 L7 age2 5 0.0121*L2 -L4-L5-L6-L7 sex*age2 F 1 0.3786*L2 0.6271*L4+0.1056*L5+0.0796*L6+0.0346*L7 sex*age2 F 2 0.2791*L2 0.0778*L4+0.5992*L5+0.0587*L6+0.0255*L7 sex*age2 F 3 0.2182*L2 0.0527*L4+0.0528*L5+0.6245*L6+0.0173*L7 sex*age2 F 4 0.1055*L2 0.0188*L4+0.0188*L5+0.0142*L6+0.7006*L7 sex*age2 F 5 0.0186*L2 -0.7764*L4-0.7764*L5-0.777*L6-0.7781*L7 sex*age2 M 1 -0.4285*L2 0.3729*L4-0.1056*L5-0.0796*L6-0.0346*L7 sex*age2 M 2 -0.3164*L2 -0.0778*L4+0.4008*L5-0.0587*L6-0.0255*L7 sex*age2 M 3 -0.1913*L2 -0.0527*L4-0.0528*L5+0.3755*L6-0.0173*L7 sex*age2 M 4 -0.0573*L2 -0.0188*L4-0.0188*L5-0.0142*L6+0.2994*L7 sex*age2 M 5 -0.0065*L2 -0.2236*L4-0.2236*L5-0.223*L6-0.2219*L7 Type II Estimable Functions --------------------Coefficients--------------------- Effect sex age2 Intercept 0 0 sex F L2 0 sex M -L2 0 age2 1 0 L4 age2 2 0 L5 age2 3 0 L6 age2 4 0 L7 age2 5 0 -L4-L5-L6-L7 sex*age2 F 1 0.41*L2 0.6271*L4+0.1056*L5+0.0796*L6+0.0346*L7 sex*age2 F 2 0.3025*L2 0.0778*L4+0.5992*L5+0.0587*L6+0.0255*L7 sex*age2 F 3 0.2051*L2 0.0527*L4+0.0528*L5+0.6245*L6+0.0173*L7 sex*age2 F 4 0.073*L2 0.0188*L4+0.0188*L5+0.0142*L6+0.7006*L7 sex*age2 F 5 0.0093*L2 -0.7764*L4-0.7764*L5-0.777*L6-0.7781*L7 sex*age2 M 1 -0.41*L2 0.3729*L4-0.1056*L5-0.0796*L6-0.0346*L7 sex*age2 M 2 -0.3025*L2 -0.0778*L4+0.4008*L5-0.0587*L6-0.0255*L7 sex*age2 M 3 -0.2051*L2 -0.0527*L4-0.0528*L5+0.3755*L6-0.0173*L7 sex*age2 M 4 -0.073*L2 -0.0188*L4-0.0188*L5-0.0142*L6+0.2994*L7 sex*age2 M 5 -0.0093*L2 -0.2236*L4-0.2236*L5-0.223*L6-0.2219*L7 Type III Estimable Functions ---------------------Coefficients--------------------- Effect sex age2 sex*age2 Intercept 0 0 0 sex F L2 0 0 sex M -L2 0 0 age2 1 0 L4 0 age2 2 0 L5 0 age2 3 0 L6 0 age2 4 0 L7 0 age2 5 0 -L4-L5-L6-L7 0 sex*age2 F 1 0.2*L2 0.5*L4 L9 sex*age2 F 2 0.2*L2 0.5*L5 L10 sex*age2 F 3 0.2*L2 0.5*L6 L11 sex*age2 F 4 0.2*L2 0.5*L7 L12 sex*age2 F 5 0.2*L2 -0.5*L4-0.5*L5-0.5*L6-0.5*L7 -L9-L10-L11-L12 sex*age2 M 1 -0.2*L2 0.5*L4 -L9 sex*age2 M 2 -0.2*L2 0.5*L5 -L10 sex*age2 M 3 -0.2*L2 0.5*L6 -L11 sex*age2 M 4 -0.2*L2 0.5*L7 -L12 sex*age2 M 5 -0.2*L2 -0.5*L4-0.5*L5-0.5*L6-0.5*L7 L9+L10+L11+L12 Source DF Type I SS Mean Square F Value sex 1 142.193063 142.193063 45.97 age2 4 2069.943424 517.485856 167.30 sex*age2 4 87.218363 21.804591 7.05 Source DF Type II SS Mean Square F Value sex 1 282.494304 282.494304 91.33 age2 4 2069.943424 517.485856 167.30 sex*age2 4 87.218363 21.804591 7.05 Source DF Type III SS Mean Square F Value sex 1 126.961986 126.961986 41.05 age2 4 1999.446491 499.861623 161.60 sex*age2 4 87.218363 21.804591 7.05 Standard Parameter Estimate Error t Value Pr > |t| yates -0.58972607 0.09204824 -6.41 <.0001 Standard Parameter Estimate Error t Value Pr > |t| Intercept 6.003043478 B 0.36672295 16.37 <.0001 sex F -1.024512614 B 0.41553944 -2.47 0.0137 sex M 0.000000000 B age2 1 -3.176876326 B 0.36950532 -8.60 <.0001 age2 2 -2.787597918 B 0.37048599 -7.52 <.0001 age2 3 -2.088127335 B 0.37292760 -5.60 <.0001 age2 4 -1.353746449 B 0.38703805 -3.50 0.0005 age2 5 0.000000000 B sex*age2 F 1 0.813889663 B 0.42023749 1.94 0.0528 sex*age2 F 2 0.716160958 B 0.42189464 1.70 0.0896 sex*age2 F 3 0.330651265 B 0.42487846 0.78 0.4365 sex*age2 F 4 0.313230835 B 0.44127621 0.71 0.4778 sex*age2 F 5 0.000000000 B sex*age2 M 1 0.000000000 B sex*age2 M 2 0.000000000 B sex*age2 M 3 0.000000000 B sex*age2 M 4 0.000000000 B sex*age2 M 5 0.000000000 B \end{verbatim} \normalsize The phreg printout for the additive model with age and sex. \small \begin{verbatim} Testing Global Null Hypothesis: BETA=0 Test Chi-Square DF Pr > ChiSq Likelihood Ratio 2357.5239 5 <.0001 Score 3823.3905 5 <.0001 Wald 2374.5250 5 <.0001 Type 3 Tests Wald Effect DF Chi-Square Pr > ChiSq sex 1 69.9646 <.0001 age2 4 2374.5211 <.0001 Analysis of Maximum Likelihood Estimates Parameter Standard Parameter DF Estimate Error Chi-Square Pr > ChiSq sex F 1 -0.36617 0.04378 69.9646 <.0001 age2 1 1 -4.18209 0.12180 1179.0289 <.0001 age2 2 1 -3.23859 0.11418 804.5068 <.0001 age2 3 1 -2.17521 0.10963 393.6524 <.0001 age2 4 1 -1.15226 0.11072 108.3077 <.0001 \end{verbatim} \normalsize The model with age*sex interaction. \small \begin{verbatim} Model Fit Statistics Without With Criterion Covariates Covariates -2 LOG L 37736.900 35374.050 AIC 37736.900 35392.050 SBC 37736.900 35443.188 Testing Global Null Hypothesis: BETA=0 Test Chi-Square DF Pr > ChiSq Likelihood Ratio 2362.8497 9 <.0001 Score 3873.5113 9 <.0001 Wald 2357.9498 9 <.0001 Type 3 Tests LR Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4607 0.4973 age2 4 932.1371 <.0001 sex*age2 4 5.3258 0.2555 Score Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4757 0.4904 age2 4 1506.8699 <.0001 sex*age2 4 5.2516 0.2624 Wald Statistics Effect DF Chi-Square Pr > ChiSq sex 1 0.4833 0.4869 age2 4 964.6007 <.0001 sex*age2 4 5.2322 0.2643 Analysis of Maximum Likelihood Estimates Parameter Standard Parameter DF Estimate Error Chi-Square sex F 1 -0.16537 0.23789 0.4833 age2 1 1 -4.02699 0.22585 317.9171 age2 2 1 -3.04796 0.21843 194.7187 age2 3 1 -1.99577 0.21577 85.5504 age2 4 1 -1.10659 0.22256 24.7216 sex*age2 F 1 1 -0.21121 0.26896 0.6167 sex*age2 F 2 1 -0.29334 0.25518 1.3214 sex*age2 F 3 1 -0.25663 0.24829 1.0684 sex*age2 F 4 1 -0.04339 0.25527 0.0289 Contrast DF Chi-Square Pr > ChiSq NSTT sex 1 0.4833 0.4869 NSTT age 4 964.6007 <.0001 Likelihood Ratio Statistics for Type 1 Analysis LR Source -2 Log L DF Chi-Square Pr > ChiSq (Without Covariates) 37736.8997 sex 37733.0932 1 3.8066 0.0511 age2 35379.3758 4 2353.7173 <.0001 sex*age2 35374.0501 4 5.3258 0.2555 Standard Label Estimate Error z Value Pr > |z| Yates -0.3263 0.06149 -5.31 <.0001 \end{verbatim} \normalsize \begin{thebibliography}{9} \bibitem{Aitkin78} M. Aitkin (1978). The analysis of unbalanced cross classifications (with discussion). \emph{J Royal Stat Soc A} 141:195-223. \bibitem{Dispenzieri12} A. Dispenzieri, J. Katzmann, R. Kyle, D. Larson, T. Therneau, C. Colby, R. Clark, .G Mead, S. Kumar, L..J Melton III and S.V. Rajkumar (2012). Use of monoclonal serum immunoglobulin free light chains to predict overall survival in the general population, \emph{Mayo Clinic Proc} 87:512--523. \bibitem{Herr86} D. G. Herr (1986). On the History of ANOVA in Unbalanced, Factorial Designs: The First 30 Years. \emph{Amer Statistician} 40:265-270. \bibitem{Kyle06} R. Kyle, T. Therneau, S.V. Rajkumar, D. Larson, M. Plevak, J. Offord, A. Dispenzieri, J. Katzmann, and L.J. Melton, III (2006), Prevalence of monoclonal gammopathy of undetermined significance, \emph{New England J Medicine} 354:1362--1369. \bibitem{Macnaughton92} D. B. Macnaughton (1992). Which sum of squares are best in an unbalanced analysis of variance. www.matstat.com/ss. \bibitem{Nelder77} J. Nelder (1977). A reformulation of linear models (with discussion). \emph{J Royal Stat Soc A} 140:48--76. \bibitem{SASguide} SAS Institute Inc. (2008), The four types of estimable functions. SAS/STAT 9.2 User's Guide, chapter 15. \bibitem{Searle71} S. R. Searle, \emph{Linear Models}, Wiley, New York, 1971. \bibitem{Senn1} S. Senn. Multi-centre trials and the finally decisive argument. www.senns.demon.co.uk/wprose.html\#FDA. \bibitem{Senn2} S. Senn. Good mixed centre practice. www.senns.demon.co.uk/wprose.html\#Mixed. \bibitem{Senn07} S. Senn. Statistical Issues in Drug Development, Wiley, New York, 2007. \bibitem{Senn00} S. Senn. The many modes of meta. Drug Information J 34:535-549, 2000. \bibitem{Therneau00} T. M. Therneau and P. M. Grambsch, \emph{Modeling Survival Data: Extending the Cox Model}, Springer-Verlag, New York, 2000. \bibitem{Yates34} F. Yates (1934). The analysis of multiple classifications with unequal numbers in the different classes. \emph{J Am Stat Assoc}, 29:51--66. \end{thebibliography} \end{document} survival/vignettes/timedep.Rnw0000644000175100001440000010571612533657167016323 0ustar hornikusers\documentclass{article} \usepackage{amsmath} \usepackage{Sweave} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \newcommand{\code}[1]{\texttt{#1}} %\VignetteIndexEntry{Using Time Dependent Covariates} \title{Using Time Dependent Covariates and Time Dependent Coefficients in the Cox Model} \author{Terry Therneau \and Cindy Crowson\\ Mayo Clinic} \begin{document} \maketitle \SweaveOpts{prefix.string=compete,width=6,height=4} \setkeys{Gin}{width=\textwidth} \SweaveOpts{keep.source=TRUE} <>= options(width=60, continue=" ") makefig <- function(file, top=1, right=1, left=4) { pdf(file, width=9.5, height=7, pointsize=18) par(mar=c(4, left, top, right) +.1) } library(survival) @ \section{Introduction} This vignette covers 3 different but interrelated concepts: \begin{itemize} \item An introduction to time dependent covariates, along with some of the most common mistakes. \item Tools for creating time-dependent covariates, or rather the data sets used to encode them. \item Time dependent coefficients. \end{itemize} \section{Time dependent covariates} One of the strengths of the Cox model is its ability to encompass covariates that change over time. The practical reason that time-dependent covariates work is based on the underlying way in which the Cox model works: at each event time the program compares the current covariate values of the subject who had the event to the current values of all others who were at risk at that time. One can think of it as a lottery model, where at each death time there is a drawing to decide which subject ``wins'' the event. Each subject's risk score $\exp(X\beta)$ determines how likely they are to win, e.g., how many ``tickets'' they have purchased for the drawing. The model tries to assign a risk score to each subject that best predicts the outcome of each drawing based on \begin{itemize} \item The risk set: which subjects are present for each event; the set of those able to ``win the prize''. \item The covariate values of each subject just prior to the event time. \end{itemize} The model has a theoretical foundation in martingale theory, a mathematical construct which arose out of the study of games of chance. A key underlying condition for a martingale like game is that present actions depend only on the past. The decision of whether to play (is one in the risk set or not) and the size of a bet (covariates) can depend in any way on prior bets and patterns of won/lost, but cannot look into the future. If this holds then multiple properties can be proven about the resulting process. The key rule for time dependent covariates in a Cox model is simple and essentially the same as that for gambling: \emph{you cannot look the future}. A covariate may change in any way based on past data or outcomes, but it may not reach forward in time. A good example of this is found in a recent analysis from the Mayo Clinic study of aging (MCSA), a study which enrolled a stratified random sample from the population of Olmsted County and then has followed them forward in time. The occurence of mild cognitive impairment (MCI), dementia, and death are all of interest. The paper starts out with a table comparing baseline covariates for those who never progress to MCI versus those who ever did, there is also a table of baseline covariates versus survival. Both of these are fine: if you think in terms of an R formula they could be written as \code{future ~ past}. A model that predicts survival as a function of ever versus never MCI is not correct, however; that is a model with a future occurence on both sides of the equation. One of the more well known examples of this error is analysis by treatment response: at the end of a trial a survival curve is made comparing those who had an early response to treatment (shrinkage of tumor, lowering of cholesterol, or whatever) to those who did not, and it discovered that responders have a better curve. A Cox model fit to the same data will demonstrate a strong ``significant'' effect. The problem arises because any early deaths, those that occur before response can be assessed, will all be assigned to the non-responder group, even deaths that have nothing to do with the condition under study. Below is a simple example based on the advanced lung cancer data set. Assume that subjects came in every monthly for 12 cycles of treatment, and randomly declare a ``response'' for 5\% of the subjects at each visit. <>= set.seed(1953) # a good year nvisit <- floor(pmin(lung$time/30.5, 12)) response <- rbinom(nrow(lung), nvisit, .05) > 0 badfit <- survfit(Surv(time/365.25, status) ~ response, data=lung) plot(badfit, mark.time=FALSE, lty=1:2, xlab="Years post diagnosis", ylab="Survival") legend(1.5, .85, c("Responders", "Non-responders"), lty=2:1, bty='n') @ What is most surprising about this error is the \emph{size} of the false effect that is produced. A Cox model using the above data reports a hazard ratio of 1.9 fold with a p-value of less than 1 in 1000. The alarm about this incorrect approach has been sounded often \cite{Anderson83, Buyse96, Suissa08} but the analysis is routinely re-discovered. A slightly subtler form of the error is discussed in Redmond et al \cite{Redmond83}. Breast cancer chemotherapy patients were divided into three groups based on whether the patient eventually received $>85$\%, 65--85\% or $<65$\% of the dose planned at the start of their treatment. The chemotherapy regiment spans 12 weeks of treatment and the early deaths, not surprisingly, do not get all their dose. If response is instead coded as a time-dependent covariate whose values depend only on the past, then the problem disappears. For treatment response this will be a variable that starts at 0 for all subjects and is recoded to 1 only when the response occurs. For dose it would measure cumulative dose to date. There are many variations on the error: interpolation of the values of a laboratory test linearly between observation times, removing subjects who do not finish the treatment plan, imputing the date of an adverse event as midway between observation times, etc. Using future data will often generate large positive or negative bias in the coefficients, but sometimes it generates little bias at all. It is nearly impossible to predict a priori which of these will occur in any given data set. Using such a covariate is similar to jogging across the Los Angeles freeway: disaster is not guaranteed --- but it is likely. The most common way to encode time-dependent covariates is to use the (start, stop] form of the model. <>= fit <- coxph(Surv(time1, time2, status) ~ age + creatinine, data=mydata) @ In data set \code{mydata} a patient might have the following observations \begin{center} \begin{tabular}{ccccccc} subject & time1 & time2 & status & age & creatinine & \ldots \\ \hline 1 & 0 & 15 & 0 & 25 & 1.3 \\ 1 & 15& 46 & 0 & 25 & 1.5 \\ 1 & 46& 73 & 0 & 25 & 1.4 \\ 1 & 73& 100& 1 & 25 & 1.6 \\ \end{tabular} \end{center} In this case the variable \code{age} = age at entry to the study stays the same from line to line, while the value of creatinine varies and is treated as 1.3 over the interval $(0, 15]$, 1.5 over $(15, 46]$, etc. The intervals are open on the left and closed on the right, which means that the creatinine is taken to be 1.3 on day 15. The status variable describes whether or not each interval ends in an event. One common question with this data setup is whether we need to worry about correlated data, since a given subject has multiple observations. The answer is no, we do not. The reason is that this representation is simply a programming trick. The likelihood equations at any time point use only one copy of any subject, the program picks out the correct row of data at each time. There two exceptions to this rule: \begin{itemize} \item When subjects have multiple events, then the rows for events are correlated and a cluster variance is needed. \item When a subject appears in overlapping intervals. This however is almost always a data error, since it corresponds to two copies of the subject being present in the same strata at the same time, e.g., they could meet themselves on the sidewalk. \end{itemize} A subject can be at risk in multiple strata at the same time, however. This corresponds to being simultaneously at risk for two distinct outcomes. \section{Building time-dependent sets with tmerge} \subsection{The function} A useful function for building data sets is \code{tmerge}, which is part of the survival library. The motivating case for \code{tmerge} came from a particular problem: the Rochester Epidemiology Project has tracked all subjects living in Olmsted County, Minnesota, from 1965 to the present. For an investigation of cumulative comorbidity we had three data sets \begin{itemize} \item base: demographic data such as sex and birth date \item timeline: one or more rows for each subject containing age intervals during which they were a resident of the county. The important variables are id, age1 and age2; each (age1, age2) pair marks an interval of residence. \item outcome: one row for each age/outcome pair of interest. The outcomes were 20 comorbid conditions as defined by NIH. \end{itemize} The structure for building the data is shown below. (The data for this example unfortunately cannot be included in the survival library so the code is shown but not executed.) <>= newd <- tmerge(data1=base, data2=timeline, id=repid, tstart=age1, tstop=age2, options(id="repid")) newd <- tmerge(newd, outcome, id=repid, mtype = cumevent(age)) newd <- with(subset(outcome, event='diabetes'), tmerge(newd, id=repid, diabetes= tdc(age))) newd <- with(subset(outcome, event='arthritis'), tmerge(newd, id=repid, event =tdc(age))) @ The first call to tmerge adds the timeline for each observation to the baseline data. The \code{tstart} and \code{tstop} arguments refer to the starting and ending times for each subject and are taken from data set 2 (\code{data2}). The \code{options} argument tells the routine that the identifier variable in data set 1 is called `repid', and will cause the identifier variable in the final output data set \code{newd} to have that name. By default, the names of the three key variables are ``id'', ``tstart'', and ``tstop''; these uniquely identify each row of the final data set. Each subsequent call adds a new variable to the data set. The second line creates an event variable which is a cumulative count of the number of comorbidities thus far, for each subject. The third line creates a time dependent covariate (tdc) which will be 0 until the age of diabetes and is 1 thereafter, the fourth line creates a time dependent variable for the presence of arthritis. This is the basic working approach for \code{tmerge}: first establish baseline covariates and the range over which the subject is at risk, and then add events and/or time dependent covariates to the data set one by one. These additions will often increase the number of rows in the data set. Say at some stage subject Smith has time intervals of (0,15), (15,40) and (40,100) and we add a time dependent covariate \code{sbp} (systolic blood pressure) which is evaluated at months 6, 12, and 24 with values of 134, 126, and 140, respectively. In the resulting data set Smith will have intervals of \begin{center} \begin{tabular}{rrrrrr} (0,6) &(6,12) & (12,15) & (15,24) & (24,40) & (40,100)\\ & 134 & 126 & 126 & 140 & 140 \end{tabular} \end{center} The value over the interval (0,6) will be the value of the variable \code{sbp} in data set 1 --- if the variable existed there it is assumed to contain the baseline value --- or NA otherwise. \subsubsection{CGD data set} Chronic granulomatous disease (CGD) is a heterogeneous group of uncommon inherited disorders characterized by recurrent pyogenic infections that usually begin early in life and may lead to death in childhood. In 1986, Genentech, Inc. conducted a randomized, double-blind, placebo-controlled trial in 128 CGD patients who received Genentech's humanized interferon gamma (rIFN-g) or placebo three %' times daily for a year. Data were collected on all serious infections until the end of followup, which occurred before day 400 for most patients. One patient was taken off on the day of his last infection; all others have some followup after their last episode. Below are the first 10 observations, see the help page for \texttt{cgd0} for the full list of variable names. The last few columns contain the duration of follow-up for the subject followed by infection times. Subject 1 was followed for 414 days and had 2 infections on days 219 and 373, subject 2 had 7 infections, and subject 3 had none. \small \begin{verbatim} 1 204 082888 1 2 12 147.0 62.0 2 2 2 2 414 219 373 2 204 082888 0 1 15 159.0 47.5 2 2 1 2 439 8 26 152 241 249 322 350 3 204 082988 1 1 19 171.0 72.7 1 2 1 2 382 4 204 091388 1 1 12 142.0 34.0 1 2 1 2 388 5 238 092888 0 1 17 162.5 52.7 1 2 1 1 383 246 253 6 245 093088 1 2 44 153.3 45.0 2 2 2 2 364 7 245 093088 0 1 22 175.0 59.7 1 2 1 2 364 292 8 245 093088 1 1 7 111.0 17.4 1 2 1 2 363 9 238 100488 0 1 27 176.0 82.8 2 2 1 1 349 294 10 238 100488 1 1 5 113.0 19.5 1 2 1 1 371 \end{verbatim} \normalsize The data set above is included as \code{cgd0} in the survival library. We want to turn this into a data set that has survival in a counting process form. \begin{itemize} \item Each row of the resulting data set represents a time interval (time1, time2] which is open on the left and closed on the right. Covariate values for that row are the covariate values that apply over that interval. \item The event variable for each row $i$ is 1 if that time interval ends with an event and 0 otherwise. \end{itemize} We don't want the variables etime1--etime7 in the final data set, so they are left out of the data1 argument in the first call. <>= newcgd <- tmerge(cgd0[, 1:13], cgd0, id=id, tstop=futime) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime1)) newcgd <- with(cgd0, tmerge(newcgd, id=id, infect = event(etime2))) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime3)) newcgd <- tmerge(newcgd, cgd0, id=id, infect = event(etime4), infect= event(etime5), infect=event(etime6), infect= event(etime7)) attr(newcgd, "tcount") newcgd <- tmerge(newcgd, newcgd, id, enum=cumtdc(tstart)) all.equal(newcgd[, c("id", "tstart", "tstop", "infect")], cgd [, c("id", "tstart", "tstop", "status")], check.attributes=FALSE) @ \begin{itemize} \item A standard way to build data sets is one addition at a time, as shown by the \code{event(etime1)} and \code{event(etime2)} lines above. When multiple additions are done using the same d the same data set, however, they can also be done using as single call as shown by the line that adds etime4--etime7. When there are multiple arguments they are processed sequentially. \item Additions with a missing time value are skipped. \item The result of \code{tmerge} is a data frame with a few extra attributes. One of these, tcount, is designed to help visualize the process, and was printed out after the last step above. (Printing after every step may often be useful.) Assume that a subject already had 3 intervals of (2,5), (5,10) and (14,40). A new event added at time 1 would be ``early'' while one at time 50 is after any interval and would be recorded as ``late''. An event at time 3 is within an interval, one at 5 is on the border of two intervals, one at 14 is at the leading edge of an interval and one at time 10 in on the trailing edge. In this data set all new additions fell strictly within prior intervals. We also see that etime6 and etime7 each added only a single event to the data. \item If two observations in data2 for a single person share exactly the same time, the created value will be the sum of the contributions. The ``tied'' column tells how often this happened; in some data sets this behavior might not be desired and one would need to break the ties before calling tmerge. \item Sometimes the ``where'' form of the call may be more convenient than using the data2 argument. An example is shown in the addition of etime2. \item The last addition above, after printing the tcount attribute, adds a simple time-dependent variable \code{enum} which is a running observation count for each subject. This can often be a useful variable in later models or processing, e.g. \code{enum==1} select off the first row for each subject. \item The extra attributes of the data frame are ephemeral: they will be lost as soon as any further manipulation is done. This is intentional. \end{itemize} The last line above shows that the created data set is identical to \code{cgd}, a (start, stop] version of the CGD data, also part of the survival library, which had been created by hand several years earlier. \subsection{Stanford heart transplant} The \code{jasa} data set contains information from the Stanford heart transplant study, in the form that it appeared in the paper of Crowley and Hu \cite{Crowley77}. Each row also contain also contains the age, time to transplant and time to last follow-up calculated from these dates. Patients were on medical treatment from their entry to the study until a matching heart became available, at which time they transferred to surgical treatment. As is often the case with real data, this data set contains a few anomalies that need to be dealt with when setting up an analysis data set. \begin{enumerate} \item The coefficients in table 6.1 of the definitive analysis found in Kalbfliesch and Prentice \cite{Kalbfleisch02} will only be obtained if covariates are defined in precisely the same way. For age this is (age in days)/ 365.25 - 40 years, and for year of enrollment it is the number of years since the start of the study: (entry date - 1967/10/1)/365.25. \item One subject died on the day of entry. However (0,0) is an illegal time interval for the program. It suffices to have them die on day 0.5. \item A subject transplanted on day 10 is considered to have been on medical treatment for days 1--10 and as transplanted starting on day 11. That is, except for patient 38 who died during the procedure on day 5. They should be treated as a transplant death; the problem is resolved by moving the transplant day to 4.5. \end{enumerate} Since time is in days the fractional time of 0.5 could be any chosen value $t$ with $0 < t < 1$, it will not affect the results. <>= tdata <- jasa[, -(1:4)] #leave off the dates, temporary data set tdata$futime <- pmax(.5, tdata$futime) # the death on day 0 indx <- with(tdata, which(wait.time == futime)) tdata$wait.time[indx] <- tdata$wait.time[indx] - .5 #the tied transplant sdata <- tmerge(tdata, tdata, id=1:nrow(tdata), death = event(futime, fustat), trans = tdc(wait.time)) attr(sdata, "tcount") coxph(Surv(tstart, tstop, death) ~ age + trans, sdata) @ This example shows one special case for the \code{tmerge} function that is moderately common: when the data1 and data2 arguments are the same, and the first created variable is an event code, then the range for each subject is inferred to be from 0 to the event time: an explicit \code{tstop} argument is not required. It also makes use of a two argument form of \code{event}. Each of the \code{event}, \code{cumevent}, \code{tdc} and \code{cumtdc} functions may have a second argument, which will be used as the value or increment to the event code or time dependent covariate. If not present a value of 1 is used. Also note that if the variable being created is already a part of data1, then our updates make changes to that variable. Be careful of this. This feature is what allowed for the \code{infection} indicator to be build up incrementally in the cgd example given earlier, but quite surprising results can occur when you think a new variable is being created de novo but its name already is in use. (As an example change ``trans'' to ``transplant'' in the code just above). For a variable that is not in data1, the starting point is either a vector of NA values or a vector of zeros; the first is used for a \code{tdc} (or \code{cumtdc}) call that has two arguments, and the zero vector for all other cases. The \code{tcount} table for the above fit shows all the deaths at the trailing edge of their interval, not surprising since last follow-up was used to define the interval of risk. Three of the transplants happened on day 0, one of which we moved to 0.5, the other 2 listed as occurring on the leading edge of the follow-up interval. The other 67 transplants were strictly within the (0, last follow up) interval of each subject. As a further example of time dependent covariates consider the PBC data. The \code{pbc} data set contains baseline data and follow-up status for a set of subjects with primary biliary cirrhosis, while the \code{pbcseq} data set contains repeated laboratory values. The first data set contains data on 312 subjects in a clinical trial plus 106 that agreed to be followed off protocol, the second data set has data only on the protocol subjects. <>= temp <- subset(pbc, id <= 312, select=c(id:sex, stage)) pbc2 <- tmerge(temp, temp, id=id, status = event(time, status)) pbc2 <- tmerge(pbc2, pbcseq, id=id, ascites = tdc(day, ascites), bili = tdc(day, bili), albumin = tdc(day, albumin), protime = tdc(day, protime), alkphos = tdc(day, alk.phos)) coef(coxph(Surv(time, status==2) ~ log(bili) + log(protime), pbc)) coef(coxph(Surv(tstart, tstop, status==2) ~ log(bili) + log(protime), pbc2)) @ The coefficients of bilirubin and prothrombin time are somewhat larger in the time-dependent analysis than using only baseline values. In this autoimmune disease there is steady progression of liver damage, along with a steady rise in these two markers of dysfunction. The baseline analysis captures patients' disease status at the start, the time-dependent also is able to account for those who progress more quickly. <<>>= attr(pbc2, "tcount") @ The tcount results are interesting. For the first addition of ascites we have 312 observations on a leading edge of follow up (all of the lab values at time 0) and 1495 within the subjects' follow-up interval. This causes a new break point to be added at each laboratory date, and for subsequent additions these 1495 are all on a boundary of two intervals. Another 138 lab values occurred after the last follow-up date of the pbc data set and are ignored. The data for the pbcseq data set was gathered at a later calendar time. Since having lab tests done is a certain marker of survival, would a better analysis have first used this information to extend the last follow-up date for these 138 subjects? Not necessarily. Odd things happen in survival analysis when risk sets are extended piecemeal. A basic tenet of the Cox model (or a survival curve) is that if someone is marked as being ``at risk'' over some interval $(s, t)$, this means that ``if they had had an event over that interval, we would have recorded it.'' Say someone ended their initial follow-up time at 3000 days and then had a lab test at 3350 days (subjects returned about once a year). If we only extend the time of those who had a test, then saying that this subject was at risk during the interval (3000, 3350) is false: if they had died in that interval, they would not have had the lab test and not obtained the extension. We need to extend the followup time of all subjects. In the study all subjects were actively followed up and the results of that endeavor are in the futime and status variables of the pbcseq data set. Extension needs to use those variables and not just the presence of another laboratory result. \section{Time dependent coefficients} Time dependent covariates and time dependent coefficients are two different extensions of a Cox model, as shown in the two equations below. \begin{align} \lambda(t) &= \lambda_0(t) e^{\beta X(t)} \label{tdcovar} \\ \lambda(t) &= \lambda_0(t) e^{\beta(t) X} \label{tdbeta} \end{align} Equation \eqref{tdcovar} is a time dependent covariate, a commonly used and well understood usage. Equation \eqref{tdbeta} has a time dependent coefficient. These models are much less common, but represent one way to deal with non-proportional hazards -- the proportional hazard assumption is precisely that the coefficient does not change over time: $\beta(t) = c$. The \code{cox.zph} function will plot an estimate of $\beta(t)$ for a study and is used to diagnose and understand non-proportional hazards. Here for example is a test case using the veterans cancer data. <>= options(show.signif.stars = FALSE) # display intelligence vfit <- coxph(Surv(time, status) ~ trt + prior + karno, veteran) vfit quantile(veteran$karno) zp <- cox.zph(vfit, transform= function(time) log(time +20)) zp plot(zp[3]) abline(0,0, col=2) @ Karnofsky score is a very important predictor, but its effect is not constant over time as shown by both the test and the plot. Early on it has a large negative effect: the risk of someone at the first quartile is approximately exp(35*.03377) = 3.2 fold times that of someone at the third quartile, but by 200 days this has waned. The cox.zph function does not produce a formal fit, however, only a graph and a linear test of whether the graph is ``flat''. What if we want to actually fit the model? If $\beta(t)$ is assumed to have a simple functional form we can fool an ordinary Cox model program in to doing the fit. The particular form $\beta(t) = a + b\log(t)$ has for instance often been assumed. Then $\beta(t) x = ax + b \log(t) x = ax + b z$ for the special time dependent covariate $z = \log(t) x$. The time scale for the plot above of $\log(t + 20)$ was chosen to make the first part of the plot roughly linear. The simple linear model does not fit over the entire range, but we will forge ahead. (After all, most who fit the log(t) form have not bothered to even look at a plot.) An obvious but incorrect approach is <>= vfit2 <- coxph(Surv(time, status) ~ trt + prior + karno + I(karno * log(time + 20)), data=veteran) @ This mistake has been made often enough the the \code{coxph} routine has been updated to print an error message for such attempts. The issue is that the above code does not actually create a time dependent covariate, rather it creates a time-static value for each subject based on their value for the covariate \code{time}; no differently than if we had constructed the variable outside of a \code{coxph} call. This variable most definitely breaks the rule about not looking into the future, and one would quickly find the circularity: large values of \code{time} predict long survival, because long survival leads to large values for \code{time}; the resulting model coefficient is large. A true time-dependent covariate can be constructed using the \emph{time-transform} functionality of coxph. <>= vfit3 <- coxph(Surv(time, status) ~ trt + prior + karno + tt(karno), data=veteran, tt = function(x, t, ...) x * log(t+20)) vfit3 @ The time dependent coefficient is estimated to be $\beta(t) =$ \Sexpr{round(coef(vfit3)[3], 3)} + \Sexpr{round(coef(vfit3)[4], 3)} * log(t + 20). We can add said line to the above plot via \code{abline(coef(vfit3)[3:4])}. Not surprisingly, the result is rather too high for time $>$ 200. (The same dichotomy exists in SAS phreg, by the way. A new covariate based on \code{time} will be fixed, while a programming statement within phreg that uses \code{time} as a variable will generate time-dependent objects. The error is less likely there because phreg's model statement has no equivalent to the \code{I()} function, i.e., you cannot create new variables on-the-fly.) \section{Predictable time-dependent covariates} Occasionally one has a time-dependent covariate whose values in the future are predictable. The most obvious of these is patient age, occasionally this may also be true for the cumulative dose of a drug. If age is entered as a linear term in the model, then the effect of changing age can be ignored in a Cox model, due to the structure of the partial likelihood. Assume that subject $i$ has an event at time $t_i$, with other subject $j \in R_i$ at risk at that time, with $a$ denoting age. The partial likelihood term is \begin{equation*} \frac{e^{\beta * a_i}}{\sum_{j \in R_i} e^{\beta* a_j}} = \frac{e^{\beta * (a_i + t_i)}}{\sum_{j \in R_i} e^{\beta* (a_j + t_i)}} \end{equation*} We see that using time-dependent age (the right hand version) or age at baseline (left hand), the partial likelihood term is identical since $\exp(\beta t_i)$ cancels out of the fraction. However, if the effect of age on risk is \emph{non-linear}, this cancellation does not occur. Since age changes continuously, we would in theory need a very large data set to completely capture the effect, an interval per day to match the usual resolution for death times. In practice this level of resolution is not necessary; though we all grow older, risk does not increase so rapidly that we need to know our age to the day! One method to create a time-changing covariate is to use the \emph{time-transform} feature of coxph. Below is an example using the pbc data set. The longest follow-up time in that data set is over 13 years, follow-up time is in days, and we might worry that the intermediate data set would be huge. The program only needs the value of the time dependent covariate(s) for each subject at the times of events, however, so the maximum number of rows in the intermediate data set is the number of subjects times the number of unique event times. <>= pfit1 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + age, pbc) pfit2 <- coxph(Surv(time, status==2) ~ log(bili) + ascites + tt(age), data=pbc, tt=function(x, t, ...) { age <- x + t/365.25 cbind(age=age, age2= (age-50)^2, age3= (age-50)^3) }) pfit2 anova(pfit2) # anova(pfit1, pfit2) #this fails 2*(pfit2$loglik - pfit1$loglik)[2] @ Since initial age is in years and time is in days, it was important to scale within the pspline function. The likelihood ratio of 10.8 on 2 degrees of freedom shows that the additional terms are mildly significant. When there are one or more terms on the right hand side of the equation marked with the tt() operator, the program will pre-compute the values of that variable for each unique event time. A user-defined function is called with arguments of \begin{itemize} \item the covariate: whatever is inside the tt() call \item the event time \item the event number: if there are multiple strata and the same event time occurs in two of them, they can be treated separately \item the weight for the observation, if the call used weights \end{itemize} There is a single call to the function with a large $x$ vector, it contains an element for each subject at risk at each event time. If there are multiple tt() terms in the formula, then the tt argument should be a list of functions with the requisite number of elements. There are other interesting uses for the time-transform capability. One example is O'Brien's logit-rank test procedure \cite{obrien78}. He proposed replacing the covariate at each event time with a logit transform of its ranks. This removes the influence of any outliers in the predictor $x$. For this case we ignore the event time argument and concentrate on the groupings. <<>>= function(x, t, riskset, weights){ obrien <- function(x) { r <- rank(x) (r-.5)/(.5+length(r)-r) } unlist(tapply(x, riskset, obrien)) } @ This relies on the fact that the input arguments to tt() are ordered by the event number or riskset. This function is used as a default if no tt argument is present in the coxph call, but there are tt terms in the model formula. (Doing so allowed me to depreciate the survobrien function). Another interesting usage is to replace the data by simple ranks, not rescaled to 0--1. <<>>= function(x, t, riskset, weights) unlist(tapply(x, riskset, rank)) @ The score statistic for this model is $(C-D)/2$, where $C$ and $D$ are the number of concordant and discordant pairs, see the survConcordance function. The score statistic from this fit is then a test for significance of the concordance statistics, and is in fact the basis for the standard error reported by survConcordance. The O'Brien test can be viewed as concordance statistic that gives equal %' weight to each event time, whereas the standard concordance weights each event proportionally to the size of the risk set. (The Cox score statistic depends on the mean $x$ at each event time; since ranks go from 1 to number at risk the mean also scales.) Although handy, the computational impact of the tt argument should be considered before using it. The Cox model requires computation of a weighted mean and variance of the covariates at each event time, a process that is inherently $O(ndp^2)$ where $n$ = the sample size, $d$ = the number of events and $p$= the number of covariates. Much of the algorithmic effort in coxph() is to use updating methods for the mean and variance matrices, reducing the compute time to $O((n+d) p^2)$. When a tt term appears updating is not possible; for even moderate size data sets the impact of $nd$ versus $n+d$ can be surprising. The time-transform is a new addition and still has some rough edges. At this moment the $x=TRUE$ argument is needed to get proper residuals and predicted values, and termplot is unable to properly reconstruct the data to plot a fit. Please communicate any concerns or interesting examples to the author. \begin{thebibliography}{9} \bibitem{Anderson83} Anderson JR, Cain KC, and Gelber RD. Analysis of survival by tumor response. J Clinical Oncology 1:710--719, 1983. \bibitem{Buyse96} M Buyse and P Piedbois. The relationship between response to treatment and survival time. Stat in Med 15:2797--2812, 1996. \bibitem{Crowley77} J Crowley and M Hu. Covariance analysis of heart transplant survival data. J American Statistical Assoc, 72:27--36, 1977. \bibitem{Kalbfleisch02} J Kalbfleisch and R Prentice. The statistical analysis of failure time data, second edition. Wiley, 2002. \bibitem{obrien78} O'Brien, Peter. A non-parametric test for association with censored data, Biometrics 34:243--250, 1978. \bibitem{Redmond83} Redmond C, Fisher B, Wieand HS. The methodologic dilemma in retrospectively correlating the amount of chemotherapy received in adjuvant therapy protocols with disease free survival: a commentary. Cancer Treatment Reports 67:519--526, 1983. \bibitem{Suissa08} S Suissa. Immortal time bias in pharmacoepidemiology. Am J Epi, 167:492-499, 2008. \end{thebibliography} \end{document} survival/vignettes/splines.Rnw0000644000175100001440000002114712533657730016340 0ustar hornikusers\documentclass{article} \usepackage{amsmath} \usepackage{Sweave} \addtolength{\textwidth}{1in} \addtolength{\oddsidemargin}{-.5in} \setlength{\evensidemargin}{\oddsidemargin} \SweaveOpts{keep.source=TRUE, fig=FALSE} %\VignetteIndexEntry{Splines, plots, and interactions} % Ross Ihaka suggestions \DefineVerbatimEnvironment{Sinput}{Verbatim} {xleftmargin=2em} \DefineVerbatimEnvironment{Soutput}{Verbatim}{xleftmargin=2em} \DefineVerbatimEnvironment{Scode}{Verbatim}{xleftmargin=2em} \fvset{listparameters={\setlength{\topsep}{0pt}}} \renewenvironment{Schunk}{\vspace{\topsep}}{\vspace{\topsep}} \SweaveOpts{prefix.string=splines,width=6,height=4} \setkeys{Gin}{width=\textwidth} <>= options(continue=" ", width=60) options(SweaveHooks=list(fig=function() par(mar=c(4.1, 4.1, .3, 1.1)))) pdf.options(pointsize=8) #text in graph about the same as regular text options(contrasts=c("contr.treatment", "contr.poly")) #reset default @ \title{Spline terms in a Cox model} \author{Terry Therneau} \begin{document} \maketitle This is a pair of topics that comes up just often enough in my work that I end up re-discovering how to do it correctly about once a year. A note showing how may be useful to others, it is certainly a useful reference for me. \section{Plotting smooth terms} Here is a simple example using the MGUS data. (And turning off the garish color defaults of termplot). <>= require(survival) mfit <- coxph(Surv(futime, death) ~ sex + pspline(age), data=mgus) termplot(mfit, term=2, se=TRUE, col.term=1, col.se=1) @ Two questions of the plot are ``how was it centered'' and whether we can easily plot it on the hazard as opposed to the log hazard scale. The solution to both is to use the plot=FALSE option of termplot, which returns the data points that would be plotted back to the user. <>= ptemp <- termplot(mfit, se=TRUE, plot=FALSE) attributes(ptemp) ptemp$age[1:4,] @ The termplot function depends on a call to predict with type='terms', which returns a centered set of predictions. Like a simple linear model fit, the intercept is a separate term, which is found in the ``constant'' attribute above, and each column of the result is centered at zero. Since any given $x$ value may appear multiple times in the data and thus in the result of predict, and the termplot function removes duplicates, the plot may not be exactly centered at zero however. Now suppose we want to redraw this on log scale with age 50 as the reference, i.e., the risk is 1 for a 50 year old. Since the Cox model is a relative hazards model we can choose whatever center we like. (If there were no one of exactly age 50 in the data set the first line below would need to do an interpolation, e.g. by using the approx function.) <>= center <- with(ptemp$age, y[x==50]) ytemp <- ptemp$age$y + outer(ptemp$age$se, c(0, -1.96, 1.96), '*') matplot(ptemp$age$x, exp(ytemp - center), log='y', type='l', lty=c(1,2,2), col=1, xlab="Age at diagnosis", ylab="Relative death rate") @ Voila! We now have a plot that is more interpretable. The approach is appropriate for any term, not just psplines. The above plot uses log scale for the y axis which is appropriate for the question of whether a non-linear age effect was even necessary for this data set (it is not), one could remove the log argument to emphasize the Gomperzian effect of age on mortality. \section{Splines in an interaction} As an example we will use the effect of age on survival in the \texttt{flchain} data set, a population based sample of subjects from Olmsted County, Minnesota. If we look at a simple model using age and sex we see that both are very significant. <>= options(show.signif.stars=FALSE) # display intelligence fit1 <- coxph(Surv(futime, death) ~ sex + pspline(age, 3), data=flchain) fit1 termplot(fit1, term=2, se=TRUE, col.term=1, col.se=1, ylab="log hazard") @ We used a smoothing spline because the printout then nicely segregates the linear and non-linear effects. The non-linearity is not very large, as compared to the linear portion, but still may be important. We would like to go forward and fit separate age curves for the males and the females, since the above fit makes an unwarranted assumption that the male/female ratio of death rates will be the same at all ages. The primary problem is that a formula of \texttt{sex * pspline(age)} does not work; the coxph routine is not clever enough to do the right thing automatically. (Perhaps some future version will be sufficiently bright, but don't hold your breath). If we were using regression splines instead, e.g. \texttt{ns(age, df=4)}, the coxph routine would succeed but the plotting would fail; the solution below works for both cases. We need to create our own dummy variables to handle the interaction. <>= agem <- with(flchain, ifelse(sex=="M", age, 60)) agef <- with(flchain, ifelse(sex=="F", age, 60)) fit2 <- coxph(Surv(futime, death) ~ sex + pspline(agef, df=3) + pspline(agem, df=3), data=flchain) anova(fit2, fit1) @ The gain in this particular problem is not great, but we will forge ahead. You might well ask why we used 60 as a dummy value of \texttt{agem} for the females instead of 0? There is nothing special about the choice, and any value within the range of ages would do as well, though I try to pick one where the standard errors of the curves are not outrageous. If a value of 0 is used it forces the pspline function to create a basis set that includes all the empty space between 0 and 50, and do predictions at 0; these last can become numerically unstable leading to errors or incorrect values. The Cox model deals with relative hazards, when doing a plot we will usually want to specify who our reference is. By default the termplot function uses an average reference, that is, any plot will be centered to have an average log hazard of 0. In this case, we decided to use 65 year old females as our reference, with all of the hazards relative to them. <>= # predictions pterm <- termplot(fit2, term=2:3, se=TRUE, plot=FALSE) # reference refdata <- data.frame(sex=c('F', 'M'), agef=c(65, 60), agem=c(60,65)) pred.ref <- predict(fit2, newdata=refdata, type="lp") # females tempf <- pterm$agef$y + outer(pterm$agef$se, c(0, -1.96, 1.96)) frow <- which(pterm$agef$x == 65) tempf <- tempf - tempf[frow,1] # shift curves # males tempm <- pterm$agem$y + outer(pterm$agem$se, c(0, -1.96, 1.96)) mrow <- which(pterm$agem$x == 65) tempm <- tempm + diff(pred.ref) - tempm[mrow,1] # plot matplot(pterm$agef$x, exp(tempf), log='y', col=1, lty=c(1,2,2), type='l', lwd=c(2,1,1), xlab="Age", ylab="Relative risk of death") matlines(pterm$agem$x, exp(tempm), log='y', col=2, lwd=c(2,1,1), lty=c(1,2,2)) legend(80, 1, c("Female", "Male"), lty=1, lwd=2, col=1:2, bty='n') @ \begin{enumerate} \item The termplot routine is used to get the data points for the plot, without executing a plot, by use of the \texttt{plot=FALSE} argument. The result is a list with one element per term; each element of the list contains x, y, and se components. \item We had decided to center the female curve at age 65, risk =1. The relative offset for the male curve can be derived directly from \texttt{fit2} by adding up the right coefficients, and I used to do it that way but would get it wrong one time out of two. So instead use the \texttt{predict} routine to get predicted log hazards for males and females at a particular age. This tells me how far apart the curves should be at that point. We force the females to go through 0, which is exp(0) =1 on the hazard scale. \item Get the predicted curve and confidence bands for the females as a matrix \texttt{tempf}, and then shift them by subtracting the value for a 65 year old female. Do the same for males, plus adding in the curve separation at age 65 from \texttt{pred.ref}. \item The male and female portions don't have quite the same set of age values, there are no 95 year old males in the data set for example, so the plot needs to be done in two steps. \end{enumerate} The final curves for males and female are not quite parallel. One thing the plot does not display is that the spacing between the male and female points also has a standard error. This moves the entire bundle of three red curves up and down. It is not clear how best to add this information into the plot. For questions of parallelism and shape, as here, it seemed best to ignore it, which is what the termplot function also does. If someone were reading individual male/female differences off the plot a different choice would be appropriate. \end{document} survival/MD50000644000175100001440000006347512545156350012500 0ustar hornikusers4a91e7d0309e2f898ae49354ef16d4f6 *DESCRIPTION 6305a6de42b5d0c08f626caec19a3073 *NAMESPACE 560a44987824f6e66d262cd7769043b3 *R/Surv.S 848a740b51619bdc8e5034a7ea9a7b06 *R/aareg.S a26f9318fb9a228e853b9866d32d357b *R/aareg.taper.S 72151182c9fb04b80ea80c28a4f36678 *R/agexact.fit.S 64cea368d781cf98a160231182ac66ac *R/agreg.fit.R 4ad3bc0f0af5cb3f14c8224927aff1ef *R/agsurv.R b61db6f67c36db0a1da7480c0bd41520 *R/anova.coxph.R d723dbbcecf2c1a463e3bea242f4a7bb *R/anova.coxph.penal.R fa335fbab61b63a982ccef24362ae8f1 *R/anova.coxphlist.S 8237686fa79b6bf3ff03ec782fc5dc5e *R/anova.survreg.S 7856c6137a3d3023dda747ea469b5d98 *R/anova.survreglist.S 6dc7b000fee7e32406b724d229fe0963 *R/attrassign.R 31e0ffaae991686af37f66671c745f64 *R/basehaz.R d2726a8734e5b992dedbb0008b2cbdef *R/cch.R a280a721edd5fc63c900bb5d59996342 *R/clogit.R 142abff209b16e6fb74a2962442ffc34 *R/cluster.S 4c7d0e4603ae92371c661ce81719f784 *R/cox.zph.S 67137c0f158ae835bff2d538cc898239 *R/coxexact.fit.R 94e49b45d947935252dac8935c85b14f *R/coxpenal.df.S c8119c0998352e18f74f0136f57ec95b *R/coxpenal.fit.R 07fa26f17f7c22a12659b8fabc336303 *R/coxph.R a848bd9479ee599ee235564866426856 *R/coxph.control.S 3676c59dacab41d12d02955b03a4a8e2 *R/coxph.detail.S fec4aff2461f7b5f2294ad72ef3d6a98 *R/coxph.fit.S 7c75397f3efcb4523f2c42f788909415 *R/coxph.getdata.S 20f3906a0bccd1ccdd8f5d029842b933 *R/coxph.rvar.S c9f7c0d4082bfeea68ab13390163a801 *R/coxph.wtest.S 7f0a427c669fdd31d7c578defab7bfe6 *R/dsurvreg.S 6b475e44b669665e6a61e4c5db3134b4 *R/firstlib.R ced58130057fb8cfb6e6dee4ba2a6d9c *R/format.Surv.S f205c222671ea50fd55a03bc7a037fef *R/frailty.S ef7a10d0a2ad7db3e2e224af7be6407e *R/frailty.brent.S 218a909b4a0b567afaf80fbe89a945d0 *R/frailty.controlaic.S dda96a5c22e785c06f1ba4e40f194e98 *R/frailty.controldf.S 6bc29d2e848783a072d0211a720f268f *R/frailty.controlgam.S 2c9f1894713f0c635b30e6f30c9fb5c1 *R/frailty.controlgauss.S ee9c0d0f791cf6da90835c24409a8f02 *R/frailty.gamma.S 8d5f1d4180a4bcea66a6e814115b124a *R/frailty.gammacon.S ac9eb99b372756010a9d0e209699baa1 *R/frailty.gaussian.S 4d0ee67c4d6ea4e84f49618dfdd776c3 *R/frailty.t.S 1bc2b7f9b1220c242b86cd1a678a87ed *R/is.na.coxph.penalty.S f4f940abf9c458cba75275e692dc42f9 *R/is.ratetable.S 392d2f367e5d2122cb375ec6043ab70e *R/labels.survreg.S 3b2e2ddba169ba87bc21830937a974fb *R/lines.aareg.S 612d578cc5306fa3814ecdc28a9005bd *R/lines.survexp.R 009724b24c5d4d0d76bcb05af737d768 *R/lines.survfit.coxph.S fccc80d1454a87ad21482fb576fc8f46 *R/logLik.coxph.R 6ece03d217744a11b9c54aa44678be39 *R/match.ratetable.S d2c3f236a60669287d1fb2903e09b6e4 *R/model.frame.survreg.R be8a11bb28600aa7e01918c47c9d77df *R/model.matrix.coxph.R b755d9882f5af6326cd4f39c65d0a098 *R/neardate.R 5220681f176b87fb822910dea854ccf4 *R/plot.aareg.S bbc7b13d91874c741e88b7cdb761aa5a *R/plot.cox.zph.S 4904a82ef0b3beb07600a00cbf55a763 *R/plot.survfit.R e9a3b797eced2c1e269467c1122b5d24 *R/predict.coxph.R 799bedb1e0005de29778ad6bc0fa2a1b *R/predict.coxph.penal.S 8e1e9c35707f42941bb07d7bbcb74986 *R/predict.survreg.S a2e60c0b4dd9667bcfbac613c929b42c *R/predict.survreg.penal.S 0fc5144be00eb21a280b70f20c838059 *R/print.aareg.S 8461a7c791de9aada7042baa2d38d8cb *R/print.cox.zph.S 93f642e4ad5d8af674516dea05174fe4 *R/print.coxph.S 9bc844ff207051b2c422f17cfb026981 *R/print.coxph.null.S 820d5d4b6bf481ab2c43114aecf51e79 *R/print.coxph.penal.S 25a86a66dc6c0368fed1b2e08e06077a *R/print.ratetable.S 51eb940cc69ddae5eb0ab5d5ae7b5086 *R/print.summary.coxph.S dd1c1dff20dbb4395c432b5dcfbaff9d *R/print.summary.coxph.penal.S a5fbce4916ba3ac805bf29e434eedf06 *R/print.summary.survexp.R 66ff0ff65063e27356e3d4b2207b9059 *R/print.summary.survfit.S e0f949867f2492881d133a83ef36e402 *R/print.summary.survfitms.S bdc72e747377b70635f0f9ce336276d8 *R/print.summary.survreg.S f1d3364c489cb576053e435e25e67293 *R/print.survdiff.S 435591842ba7380ea9c94033fd867165 *R/print.survexp.S c382b5d83ba22b3b441b8520c43ebc79 *R/print.survfit.S 2dcf8bc5145fc06b103501a956aa2e46 *R/print.survreg.S d4828baa1a76443e7aa03a472991a08d *R/print.survreg.penal.S 3263b3ef23790bd4a2c756b2ae591c81 *R/pspline.S 983c51ec85ab5b0ffe2980d51d790254 *R/pyears.R 641b5fefe304689647a4351ae179ecea *R/quantile.survfit.R 61397677849f2c0e28441a368eac37a0 *R/ratetable.S add8e58326eb43b15a1556072010e8b2 *R/ratetableDate.S f68ee67a9af4a171b8d61a67f0b0ca6e *R/residuals.coxph.S 871dc4bdd8fc9c0528e9c794498afcab *R/residuals.coxph.null.S 71bb8b32a975d5add7bff7d5963c8ea2 *R/residuals.coxph.penal.S 2fd47c2f8d06e28759406cdf7d3dbe11 *R/residuals.survreg.S 11fdb8e8cd941eb3aa184f718bd42e9c *R/residuals.survreg.penal.S 5a28f06bc0463f5278e17e894b5bbc78 *R/ridge.S d4965488f4b863ef65b00c93886454e9 *R/strata.S d8021596430ad6d8bade6e35211275ba *R/summary.aareg.S 25fbcd4b51c0eef104707aacb4a9e101 *R/summary.coxph.S 5f290ff689b07cdeccbbe8645d6fa9a1 *R/summary.coxph.penal.S fe9226f277106f203780f25c1c1807c7 *R/summary.pyears.S 4ebb40b14afa1c9bb3e58eacd035a7ba *R/summary.ratetable.S 5c340d79d336b2033f2287b104bfffd8 *R/summary.survexp.R 240b9d40c300c57bca8eb926107b5335 *R/summary.survreg.S c364ecc1455cee692df6721ba97bd722 *R/survConcordance.R 68de49d24e8240588682fe239676a8af *R/survConcordance.fit.R dbe63cccb564b0b26fc1ceec1b96c1b4 *R/survSplit.R bc393f42ff473330532bbdbf046ea772 *R/survcallback.S 6650fff9fe077bfcb2ac66ce65662cf2 *R/survdiff.S 5ec97f1eedfa55ca01bab8410d58911b *R/survdiff.fit.S 206d0d86d4a2f592c790ca1f6c49fed7 *R/survexp.R 64a2f062e53fc7b6f5607f6ef6b94e89 *R/survexp.cfit.R 4a93009dfabeb1b4a30b498366249447 *R/survexp.fit.S 0a07acc202e7db716ca92057c6ea2ee5 *R/survfit.R e08939bcea09cc87e1466e409cd6664e *R/survfit.coxph.R 7ace25a5329d25d67d0c515a63c7a15b *R/survfit.formula.S 66be50dc0ef3b14f494879281583bcd7 *R/survfitCI.R 09d44fb8e4534d995e6639369ab29f38 *R/survfitKM.S 5c0131516bd64d5ec5057afcc28b7b5f *R/survfitTurnbull.S 04d961b0116345a1034a9dedbdc2f41c *R/survfitcoxph.fit.R 35cf969fde8dd304fb1486064a0e811c *R/survfitms.R b1d5fc4d279b1b7c0af57b7bf175eb3f *R/survobrien.S 188bd1ef59103923afae70558b2dee74 *R/survpenal.fit.S 3bbb640ea778789f234b44558e78b2e7 *R/survreg.S 96e65ede62cf0161f2577850b8c57414 *R/survreg.control.S 43dd81b0ec98ddcea973ae8b6efd168f *R/survreg.distributions.S 767199fda7ec3af612995eab6d8e980f *R/survreg.fit.S 51bd3294bab1756e788db689df0fa2b6 *R/survreg.old.S 375b8fcc554efd161bef4d1973d7dbef *R/survregDtest.S f6e87aefd92c42187a712da10dbde894 *R/tcut.S 029b4c1681ccd01f922ad539c4256b48 *R/tmerge.R 102c30cb668d65064e71f8cf523022c7 *R/untangle.specials.S a357d9521d8d85be7d40df20e7f8bcfa *R/xtras.R d99157b20fd6f79088da7c13097c53e5 *build/vignette.rds 7eca04e8deb3fb7c9079d12c0279cced *cleanup 125a8993c88b795784a86dd5390b2a19 *cleanup.win ba8f4bf4656d05f12e3e66649318f52d *data/bladder.rda e458ee36a24bd3de07fe2c43502ca199 *data/cancer.rda 4204f35b614c1c940bcfe1c6d70def90 *data/cgd.rda 81cbcf099abf5f0cbea248e0012b483c *data/colon.rda f251bf2ba1ae1661640751a5c34eec69 *data/datalist 8506dffe31d64e1b4b7fd34f41c88b59 *data/flchain.rda e6b0243c6add347247c4e1951e4d68d8 *data/heart.rda 93be824b10668d2ff03e370999d2a874 *data/kidney.rda 020b59a7c604dd2499832e1b56a7bc33 *data/leukemia.rda 7b14acde4513dba5ddf30bf676500a1d *data/logan.rda ccec4c290e88d1d537995b013f6f7644 *data/lung.rda b132841ddec8bd83acb43f0e4471b53e *data/mgus.rda 4e593e3b10ceb3c8beac0fde494b2762 *data/mgus2.rda 6473af5d92183e8533ad71464faea559 *data/nwtco.rda 7a09b566d123860033607cf1dd7c40d5 *data/ovarian.rda 640a93d6e0bb2deb8668433bd39f4b05 *data/pbc.rda 2abe3dfb644c627ca0174fd59785b662 *data/rats.rda b75bf4f19a344d7defddeed1a48cbc1c *data/stanford2.rda 55fa4acd4893490d19ed48c260803353 *data/survexp.rda 2d8b5f44959ad35a300010890de7ac12 *data/tobin.txt.gz 1349117ec7f1b23d891a20571b878f11 *data/transplant.rda 76fb141b816ea857308105f26e26e87f *data/uspop2.rda d4b78dd46512d982683f6bb2a025aa9a *data/veteran.rda 647ad0bb040a42d58cbd341a6770e205 *inst/CITATION 3e8ebae349953a3b7a665eecce562802 *inst/COPYRIGHTS bae59c7c0f74ab1198b294ff629da8b5 *inst/NEWS.Rd ce943afce1adbc3d4b24e53db2929a1a *inst/NEWS.Rd.orig d992a31253c349db5c6b4bd022207d50 *inst/doc/adjcurve.R 2d86e07bd20c9e396e7c5f2d498a53c0 *inst/doc/adjcurve.Rnw 2b8afac6076b991469bec55af90d56e2 *inst/doc/adjcurve.pdf fb21e6ca6286dc398bac9db7d4722d37 *inst/doc/compete.R 7c0cb35205be857bc07818a625aba559 *inst/doc/compete.Rnw 2e944ca6e458063df88a8b56aa0e0abf *inst/doc/compete.pdf 83cdbe93f0a53d78b16dbffa1f1251db *inst/doc/splines.R efb4204ec6f04a7b6f4d4a4f0009594e *inst/doc/splines.Rnw 8a21f1cd8095c7030ad3c2e482470084 *inst/doc/splines.pdf bfec5d716db13f33cbee48864aaab17e *inst/doc/tests.R 4e631d7fd5b4a003c7d62245d5fdb296 *inst/doc/tests.Rnw aa6b6213f20ed2294436771edf2f52e6 *inst/doc/tests.pdf b2534c20ed48636d64e920a44a081b9c *inst/doc/timedep.R 5dc7f0f37e6fd60751f807541b16c7a8 *inst/doc/timedep.Rnw eb647c1e4146a654a6ca4914b0ead2ae *inst/doc/timedep.pdf ac297959a2f3093d13819580a6292410 *inst/doc/validate.pdf e3e0c8f7470da537df515c7e9bf89280 *man/Surv.Rd e1b4f2135554802fd7d56872fcbb0d42 *man/aareg.Rd ed668cbedff5d61d180b348ecc06fb80 *man/agreg.fit.Rd a243059ab841bde163e28f3831038eca *man/aml.Rd d2288843d4656e44c0a566e50f0b55e7 *man/anova.coxph.Rd b88c1e139fd5de07e43e0e93816513a0 *man/attrassign.Rd 28ac3b638b4f849f952e4765c662bef0 *man/basehaz.Rd fc00fffb10459047ec3136cf0fab0319 *man/bladder.Rd 7786f82290adc392f36583bb8e394696 *man/cch.Rd 5fe6050024be94e2212b43ebb09b4a0b *man/cgd.Rd f2211384ec14fa3c5cd19a029642bb07 *man/cgd0.Rd 423c29750d3dddad2cc3e4ca8461acf7 *man/clogit.Rd 45fd52791a5f219b29563b6bc0b38ea1 *man/cluster.Rd 93fc17a26a13146d486eca163f0fad67 *man/colon.Rd 609ac105f9fc0567d3412300cd64740c *man/cox.zph.Rd 6639b23fdb39dc53b0cc5817639e503f *man/coxph.Rd de8f44c3612140aa47ef9986f9ee9e0a *man/coxph.control.Rd d90b227d1ff0e2c44c2382944f05ddb6 *man/coxph.detail.Rd 96fa8a8d9eac7e8e0deebe787a08c75c *man/coxph.object.Rd 7d420744d3456dd97e9b3049efe7a9af *man/coxph.wtest.Rd 63939ba52046146001cebde4b6758c9d *man/dsurvreg.Rd 71f664a8e04457bf1fa491f2d6369a4e *man/flchain.Rd f021304bee346a024f131465c95e4378 *man/frailty.Rd a4f532be336afe5787e361ba953f02d8 *man/heart.Rd a97e4438991c438a09a2e3097d83aef8 *man/is.ratetable.Rd 3a65ab9dfd4e8b848853397add094231 *man/kidney.Rd ce9745dae2d6cad6d3c70af40a27bb4e *man/lines.survfit.Rd 914d8176b24badadfd48012d1b01cbed *man/logLik.coxph.Rd 378fb131601504618640f8beb5915b8f *man/logan.Rd 236b763733160810bd71f035739dfc24 *man/lung.Rd 3041e8c2d84e65b46b36d9f23be8800b *man/mgus.Rd 6216b7046c6e21e2dbc627d41f8dcc43 *man/mgus2.Rd d880904b50ae1b456211415705c9230d *man/model.frame.coxph.Rd 568a05b7245844573b768922bcaea30d *man/model.matrix.coxph.Rd 3a357662cadde697913475d7495f7687 *man/neardate.Rd f24a26e371a5b103b654e1d83edc0cae *man/nwtco.Rd 3fdf3afd881e4563da52cc1ce590a6ef *man/ovarian.Rd 1053436df37708f6c3cc449da9eb3371 *man/pbc.Rd 2100e45b8f34c72d36c2429b2179d08e *man/pbcseq.Rd 415495ead19d69d6b12b31247f18e610 *man/plot.aareg.Rd 130a89d6ffb707e2251c5942331b1d58 *man/plot.cox.zph.Rd 983917501f876cc57c1762f364d48e2c *man/plot.survfit.Rd 4865fbe6cc62138d8377255e97d86510 *man/predict.coxph.Rd 10e5e565d7daf52446e36cd7570eb9e7 *man/predict.survreg.Rd f70cce7373972ebe96d1a3b1a3c7bb5c *man/print.aareg.Rd c2c0676f5df40330a19a7e32d6fa38db *man/print.summary.coxph.Rd e050e3df696e9d25620f1c7527fbb08f *man/print.summary.survexp.Rd 71404c1cf6a8734cbd1705be65d83c43 *man/print.summary.survfit.Rd e9c888e43007f0c26dea41f91d0901ee *man/print.survfit.Rd 2be07e3d863bd261d33c28cefab2c9e7 *man/pspline.Rd c6cfdefccfc387f55642eeac9896a773 *man/pyears.Rd 2bf3a9dc6d5e99f3f33e725b5bda5165 *man/quantile.survfit.Rd 7db614234d4daac286fb02b330434668 *man/ratetable.Rd 3683c1dd8254d836e811dc0758851164 *man/ratetableDate.Rd da6c229aaaf231d2b43fb5f7cdf24efc *man/rats.Rd 1846c4b11dac5c278d588a92f0cdc961 *man/rats2.Rd 704103c600d7679e780a84cabb048c2e *man/residuals.coxph.Rd 16e171468f684b9c9a28ab436401d900 *man/residuals.survreg.Rd 9417958dd56e372fab59d563926ddb7e *man/ridge.Rd 2275b3b5eeed3ba4bdf48b0daa820077 *man/stanford2.Rd 7feaf093046e95fcf9e4a0fa5b47ccf4 *man/strata.Rd 02588806b65c0c69938ea45e5f7e4416 *man/summary.aareg.Rd 57410802045cff70955f2a76df0ee636 *man/summary.coxph.Rd 1885334cd827a74464217540865c7b91 *man/summary.survexp.Rd 9fff12352f654d179901089504c7363d *man/summary.survfit.Rd 67f70037f47809e1216ecdde4e315d5f *man/survConcordance.Rd b1b1fd427ab41d43834c2c0c05fbba55 *man/survSplit.Rd 8e81d4156c2d5a32faedb73a57685aaa *man/survdiff.Rd 7a04e5497d24616401b487c8a79970c4 *man/survexp.Rd 65e8c2fdd454ccb1a16f6dcdb86d77bc *man/survexp.fit.Rd f1a71fe6822737d6544f24e8e52251f8 *man/survexp.us.Rd 3750a843c9dfffe984b542067965524a *man/survfit.Rd b35fc0e21fb670ab158787f335db2b6d *man/survfit.coxph.Rd 1a4408ca2c831d6a1db532f4c1114480 *man/survfit.formula.Rd 0406dfb6e6b0a59175725ad889c99b86 *man/survfit.object.Rd dfbbb60cc92aef955a17146c7a8c0645 *man/survfitcoxph.fit.Rd 5d50adacfce724a343991f965fc4292f *man/survival-internal.Rd ee739a1671c6e23ae0b2e56dbcc32967 *man/survobrien.Rd d11500cd130c9e2876252a0953c33dad *man/survreg.Rd 3d62a4ef5a22c39d2df18b9e50ea30fc *man/survreg.control.Rd 734c4eaa18944faf86159d82636b150f *man/survreg.distributions.Rd 37c766235f7b2b0369d18aff07e56443 *man/survreg.object.Rd ec57c575435869dbb1c776a2d99c0e6a *man/survregDtest.Rd cceef705b631580e3ce011d02ae575ab *man/tcut.Rd 1a7184c822e4c88726fcd3dd1d32c8a5 *man/tmerge.Rd c7813220c476cd8a2752753e515a7c4e *man/tobin.Rd cb16d22b0e5fc48f13d722ad2a39848a *man/transplant.Rd 142cf81539d3cb828029c39c4c31393c *man/untangle.specials.Rd 13bf16a88bfd11ae3672755091140d8c *man/uspop2.Rd 2a684c666b999d0106a69fd13e466408 *man/veteran.Rd 3f69d360741c0c74b3afba8cc1d8c756 *noweb/Makefile 6583ac4eb425cfa6b2bf8576390e2040 *noweb/Readme 5ab65f2dfd5708f4e94c173905ff4fe2 *noweb/agreg.Rnw ade2cca0317838422c3df8e40c3d783c *noweb/all.pdf b7e6f05ce92c70768746530b4cb114a2 *noweb/casecohort.Rnw a3bc476ff51bea8252c7c1c5f683844f *noweb/concordance.Rnw c62d9e08f9696ec6f949f0b454636fc1 *noweb/coxph.Rnw 1a2da2658ebc807d57fe6a7fb39ed697 *noweb/coxsurv.Rnw 7311fe115ba221398b895de3242aac4f *noweb/coxsurv2.Rnw b818ec76c7bbdf84e52e1e26a3eb2173 *noweb/exact.nw b9ce27ad905c54ff991fad8f23d61fdc *noweb/figures/balance.pdf f994cd6aabbefe3f8bebe50b4c61e934 *noweb/figures/balance.r 85651b7327c0a38263ecab2c63303140 *noweb/main.Rnw a3884af002890ecf46180468c47d6c7f *noweb/msurv.nw f2c7a1da803141bca41ecc91b50b22a0 *noweb/plot.Rnw 977507b973dc5612b48b1460215583f5 *noweb/predict.coxph.Rnw 11321732cb15889975eaf15b623d05e6 *noweb/pyears.Rnw eba900d489efe016a6028f57973f8e2b *noweb/ratetable.Rnw 00153a591d69dc50d6852fb8f9f2ddb2 *noweb/refer.bib 6a9afea118340fdbb67fbd82044c8ec7 *noweb/residuals.survreg.Rnw 69fe903be23d0f6586582f62a4b422dd *noweb/survexp.Rnw 72f71f52d782c327190c553069de976a *noweb/survfit.Rnw f7c1d42c01506843723e74432e966b98 *noweb/survfitCI.Rnw 300cd327a0c57cced082892e924905e5 *noweb/survfitKM.Rnw e94672e4cc47a5d464a11eb593f4f321 *noweb/survfitms.Rnw 9534b235ab22d3c6d25f7a1821cac049 *noweb/tail 85ab08004b79f41cc03b776295299e05 *noweb/tmerge.Rnw f4956e2493c8ebe6c0d37c5a4d5c23de *src/agexact.c 5e9895762edebdd45109fa025b19d1c9 *src/agfit4.c a7fa8f474f0c9d896c4ac2eceda91a05 *src/agfit5.c 2510be13268a84154521b1b600a02b7e *src/agmart.c 594eb5924365cb5371b8d853cfde98fb *src/agmart2.c a7616401e3f02d65ed94de3ed279cb30 *src/agmart3.c b5693996124d767fd6c612b8cd4c9f24 *src/agscore.c 422e2dd61103541815331727949e97c7 *src/agsurv3.c 900680762d85537d9a1b42c2600c80ac *src/agsurv4.c 8af1aa4a7182f7c1eee810a09c821bbf *src/agsurv5.c 9d58cebadc5a235bb5860076bb15ce10 *src/chinv2.c ee4dd3ce1704790636f67fc942e86e33 *src/chinv3.c 99da13e6ef4d8be44d3acd9f0e6b4324 *src/cholesky2.c bf4dd7cff7207e17eeaeb9d37b3e4088 *src/cholesky3.c 4df7e06ba19e37b66b6c0ed6119f3219 *src/chsolve2.c 7e8e47e291ebc7308604421e099383b6 *src/chsolve3.c 13835d422c0b7db2a0916a9a0561a67f *src/concordance1.c a92ec8bacf00f9cc6b16f701283c4834 *src/cox_Rcallback.c 2ea96485136e6fd4215d294fdb4b73f0 *src/coxcount1.c af7acd4c3d735233e69d2ce5a687e988 *src/coxdetail.c 2d852f8007c3065aaf27907373f8c696 *src/coxexact.c 94d40f3dda3928807d25e1becee26bc3 *src/coxfit5.c 61ab5fd52e4b3e38ca19d5a2c9b34156 *src/coxfit6.c 6a85c40ec6742f6986c0e53932a5d1da *src/coxmart.c 040bc0dff9e64ba27ca36c1ed540747d *src/coxmart2.c 1ff88ef5614cf0138ea5315f69d5ff4d *src/coxph_wtest.c d9ee5e75ea894b8f6526ecd7e0917e63 *src/coxsafe.c dca45548f139de8e5dbf3fd1f7d8bcaf *src/coxscho.c eb162e94b31d0ec6ed0e68b767ac13d5 *src/coxscore.c c84812b76eb487d1681437f9e00aae17 *src/dmatrix.c ff7a20d8d165802ea3766d9fa75e14a1 *src/doloop.c 064da177122e444a026f36830a8b52ad *src/init.c 23bb38ccf2c88c0f47776f5b38b03ab6 *src/pyears1.c 3f85ef9d57e28eb78ac14afb2cf8e855 *src/pyears2.c 97e294af3935dca814ebd26bae16f168 *src/pyears3b.c db3ba3cf82fecd184767eb3a9ccbb2a6 *src/pystep.c 49ffabd94126ed4b7a2c46dffa646087 *src/survConcordance.c 80beea3c5d4e2c8495444dfcbe497016 *src/survS.h 4fbd9ee9033cb469f23534595aaa1695 *src/survdiff2.c a03b1a3c6764b98696b3c32718f31bd7 *src/survfit4.c 643c952406884d38b7409138cbdd7c0a *src/survfitci.c 20e4b99ef6b980c657d0f34be576cbad *src/survpenal.c 95bc8a528dd88d5968e2ecebe4f426d1 *src/survproto.h 2d7f18518c13727a50f4b9e59a6bc9e6 *src/survreg6.c 7483bb0e3da8485fae01944322021385 *src/survreg7.c 1143f87e1941fd27622fc229b13d840c *src/survregc1.c 73f646a72c2783018c329ca3b9534192 *src/survregc2.c 9348631d5f86125d99f657cc27ee4c3e *src/tmerge.c 29b0e992cd2c0c581c18c7f8987802e9 *tests/aareg.R 7c1f73136f34d9b4ccaf1a94a4e54126 *tests/aareg.Rout.save 7adcf3da61f0cb888ed3ae1787d85fd5 *tests/anova.R 5a9d96c769ebdb6aa6e411040eeeb481 *tests/anova.Rout.save b165f66d37de4e6d4ff52dba6d96b69e *tests/bladder.R bcf3a1b172c61d8e8c55d6feb3007d23 *tests/bladder.Rout.save ac713b96ed0ece83c73db9957606191a *tests/book1.R a9412254944c1c35ac5d3e151af42d9d *tests/book1.Rout.save c55188c3d14ae560672027b032818831 *tests/book2.R 73a796b980a681ea43c78c07a7986921 *tests/book2.Rout.save bd7615dda98191dcd20cba6ee7b8b982 *tests/book3.R 973cbeafa571ecb19a57e38d6bf7f3c3 *tests/book3.Rout.save 70160788066ddb894f98bb16c6ad6b77 *tests/book4.R 2f09a529bfe0601d9eebfc5ee7d0364b *tests/book4.Rout.save 075eb8504981ad1cc32021ac897a0ae1 *tests/book5.R 03229d828d3e6418cc84f02bcc456898 *tests/book5.Rout.save 6c123e7a51c707a4708de781550425a8 *tests/book6.R f57af023b975047184a7bf53e6cfd659 *tests/book6.Rout.save 7f0043933913df6795c66c8af6a278fc *tests/book7.R fa62c55a8870dd4d4992cd27833d909f *tests/book7.Rout.save 34ecdf393bd3d5934796c59f1aba27b5 *tests/cancer.R 98f7c25380af0fbe07574d49774045a5 *tests/cancer.Rout.save 9700f9df5c3fafb13be25a1a551bf3b3 *tests/clogit.R 32bef9152f59c8d27772ecd46efae75f *tests/clogit.Rout.save 0d2d614825b6e0e9ed7772c057510510 *tests/concordance.R 650a2b87e1e9279de9b84cb83aff98a0 *tests/concordance.Rout.save c82eeced77b617173ec9570c73f2a9a2 *tests/counting.R 1ea18186b135e14be8b306e6e935488d *tests/counting.Rout.save c212cf74254128290e416d4ffe65b070 *tests/coxsurv.R ba07f03683b9cbeb48d961815b6f6421 *tests/coxsurv.Rout.save b9ec1543269837e1177f1b29e3e647ce *tests/coxsurv2.R c3af1a1d5d78f2ec88319041b0f3c724 *tests/coxsurv2.Rout.save daf2f2353861e51e0a42d32fd66899b7 *tests/coxsurv3.R 37ce48c7b195e8d51462125bb577487f *tests/coxsurv3.Rout.save 18a3b389ac1df5e8d31cb54d4fc84d88 *tests/coxsurv4.R 450400f45ddcf1d8ecc8aa02c01fcfde *tests/coxsurv4.Rout.save 005886030229bdccfcd64481a0bcba70 *tests/data.capacitor 0d9df702d89f2892c3c1019c4ba4830f *tests/data.cracks 24d53e3037dc40c31e3825757010d330 *tests/data.donnell 4a1536544ca5c94f2b0a6ba121dfcb29 *tests/data.fluid e98a53ae045a2c998873558384b4496f *tests/data.interval b249fba4d50304f5abe1e8ffd5453c00 *tests/data.motor 427145fabf975124abe33aa7ef844117 *tests/data.peterson 34501965c3dcaea3558942135502bed1 *tests/data.rat2 3376a88bd91fe9e39caeda263f31b14f *tests/data.smoke c749f7cf62ac6445dc1e8e973245c916 *tests/data.turbine afcd2288df28b6f112e38b60cc85f719 *tests/data.valve 37b3b890eb0064e23dfa8ac56e53a726 *tests/detail.R 69a7172274b6f3692a938aaa2b350b9d *tests/detail.Rout.save a6ef6105689649a42cd6c80f150b9e6e *tests/difftest.R 2acf663f8bea2653cc452936127f9d3b *tests/difftest.Rout.save 991edc0be7125b4e63755574f1a3892b *tests/doaml.R 26e3cef8ee57b2b4b8d82c98e56844c4 *tests/doaml.Rout.save c248a344eed896873690a5bd20e91a70 *tests/doweight.R 4a76279dd178d42194131b108792f45e *tests/doweight.Rout.save 0c21db5785bcd9b01a30837dbfec5af2 *tests/expected.R c3293525ec99524e1c8379284b4ad864 *tests/expected.Rout.save 6b9016c3bcf4b6e4a8c4ee8b5d70ec3d *tests/expected2.R 139d605a2e7c02a9d339a35d12ae0252 *tests/expected2.Rout.save ad56241e8136d97f4c732eb444fdb46b *tests/factor.R c5da54a4d59d6824dae9ff849a705124 *tests/factor.Rout.save 02de4596b58fb4fdf0dbe5533f0b418e *tests/factor2.R 12c3b09b26f1307be02a08f4524d5a09 *tests/factor2.Rout.save 9ed932a843ad295f3712cf9616dcd6c8 *tests/fr_cancer.R b101465c733601a58bd581c819f0bf77 *tests/fr_cancer.Rout.save a3d4515cba9f78cd376c96c4020fa835 *tests/fr_kidney.R 87a0ecf02362639af295636bfb97590e *tests/fr_kidney.Rout.save cb0228f2570cedaab05550c53a7b7249 *tests/fr_lung.R 7a5619d6a9628bb594fbae58a3576fb1 *tests/fr_lung.Rout.save 9f58191cb3ec3ef7e94a34d4bfc81f5f *tests/fr_ovarian.R 4a957db5854dc2cd6f95dd208fbb4a22 *tests/fr_ovarian.Rout.save 55e3f6b9ec3eca458ba39aa6df86ed2c *tests/fr_rat1.R a4c8844dc4e04a2b6b7aef0930aa7b8f *tests/fr_rat1.Rout.save e6c4e1539a2ed476794a0bbe52ec9ac4 *tests/fr_rat2.R 3ee062e87a93f0fbc4bcae4a25cbcd41 *tests/fr_rat2.Rout.save e2b0adf5d38af756f7b0b7fb3276b54b *tests/fr_resid.R 3a3aa775c3570e09acd826f88c20fd64 *tests/fr_resid.Rout.save 1e2bc4e6a3d37c8658ca50e7ff612b52 *tests/fr_simple.R b2e68d9716510d02d5447b816b3e97c3 *tests/fr_simple.Rout.save 8a5c35fa11403beb1fd5830cee8f8678 *tests/frailty.R 3f2f6496cf052a80ce9700cead8f9568 *tests/frailty.Rout.save 18b3c00b4da6297b3b34f90925c8f17a *tests/frank.R e0fdc91c27e7d792f9d1feacd7dab6b1 *tests/frank.Rout.save d2618321d50113816fc01504b37980ad *tests/gray1.rda da2189dca094543c747c16ee2ff1e7b3 *tests/infcox.R 0ed395671cfe514e25e7a1c01102e343 *tests/infcox.Rout.save 82613d425beb20ed7ec958f7363c15ff *tests/jasa.R a7eb71ed5e3554f85859e269e85f2336 *tests/jasa.Rout.save 4e2a56269261de9bfff6d9fbf0f2fafb *tests/model.matrix.R a22ac6d5a5bd48427a73b022080ccebb *tests/model.matrix.Rout.save 32962da1fbc1bd7fe2d80f77ac23cd5a *tests/mrtest.R 45fb36c9a86dba3e3b4cdb9b7c466adf *tests/mrtest.Rout.save 47a92a7fedb1912814f5b71d55cc1470 *tests/nested.R 89322b4b48d5537d1395478bf9d6932b *tests/nested.Rout.save 56af7992839030a27e993ab69643a333 *tests/ovarian.R 43c01a4ea1140ed269f2a2a166bc1c8e *tests/ovarian.Rout.save c2e558fbdd9d32310d7e6acad43e7d6f *tests/plot.R c11f65f1efc009dadc20baeaef146999 *tests/plot1.pdf.save 90e8c53861f1af5739edc19dd2a8cdcc *tests/prednew.R f33c674fcdbf23170839778cb9d99911 *tests/prednew.Rout.save 1572abde0117b7fb5f744bb2da3f40bf *tests/pspline.R 8d6b15d29a85306a763e90fbc19ceacc *tests/pspline.Rout.save 1a59eea4913cae48454d543db3182fcf *tests/pyear.R 9da387fd993d611ec3c5caf442a9ac3a *tests/pyear.Rout.save 404bd9fcdffec5846229fd7860a88099 *tests/quantile.R 72e5b5e93d53191306786950ea2eb605 *tests/quantile.Rout.save a2d9dfdd822551e38aad4a0dcbc6164f *tests/r_capacitor.R b82aa77a6b0d72eab6663095f37554b2 *tests/r_capacitor.Rout.save 843037a999f6e0b7a437022064ee4c3f *tests/r_donnell.R e274d05ac892dd51da69d284c7a1cd3b *tests/r_donnell.Rout.save 3f721f6e1617f9ccabfd8242839866f2 *tests/r_lung.R f5f5289cc617c35399b7e0f2397bdfba *tests/r_lung.Rout.save c64871cdf6b3821b9c363f414300277c *tests/r_peterson.R 2b7b875244e745983bbc6075708d753c *tests/r_peterson.Rout.save 7540e96b30573ae4957ac72c1783dabc *tests/r_resid.R 174d928f93974d236344089dc002aad5 *tests/r_resid.Rout.save c030b53c848a6d2bee207d8bb4fb0ac6 *tests/r_sas.R 2872d8467b0df50beb1f936674b9406e *tests/r_sas.R.orig 7ba19e28f9eb153fe623ff22c3339764 *tests/r_sas.Rout.save 59912fe4c23adf729630b2a9256749bd *tests/r_scale.R 9c44d08c1462edc432d1162f6353a785 *tests/r_scale.Rout.save 3c6a7056af6fb278f774abf6a69c3968 *tests/r_stanford.R ec0f2b0eb632b8c1d0a0fd9cf4cfefc3 *tests/r_stanford.Rout.save ad42cc2df70a8719e7627dd1b96b0a74 *tests/r_strata.R e69c3f424f42f1e7db9b53b01375c723 *tests/r_strata.Rout.save 989f37c165e69d16e359eec89d0b3d71 *tests/r_tdist.R 1fc390713c29b90191c7382f7ff18875 *tests/r_tdist.Rout.save 714c0d3571c7fa78db7e1231f9d1ae9d *tests/r_user.R bcf83472af25c444668fb004755e2b52 *tests/r_user.Rout.save 01890b568b9fcb7db62832346d2421a3 *tests/ratetable.R c1b700bbf08eac95bbb9326edd964e47 *tests/ratetable.Rout.save 8d8a52c50d9cf57f7775b38989813141 *tests/rounding.R 42ed1037cc91e95a350c74734ad34461 *tests/rounding.Rout.save 5a65fb2b39182a19953db1807ec3f991 *tests/singtest.R b77bef3fb117a8c5a1743623f1d06eea *tests/singtest.Rout.save f7a452292c0f25593b8cbc15bdd572d5 *tests/strata2.R 3c47f3393460e219bacd64b6e7d9c0c9 *tests/strata2.Rout.save 29d08550132c65bfde059ff1be9338aa *tests/stratatest.R fe2734507c238020983489f7d79445f7 *tests/stratatest.Rout.save 6de28d7b9cbf046367dcde728bc3c3fc *tests/summary_survfit.R 0dacfc22dc9bc43719c4c32e4a9b6ac1 *tests/summary_survfit.Rout.save 734bc2984b926635f759b5f669b8cdcb *tests/surv.R 93df5892354ed75338eaca326f4b1889 *tests/surv.Rout.save a6b430161be16c02c6359cffde39870e *tests/survfit2.R ba758aa2a524b5cbc53e7e850ac909ef *tests/survfit2.Rout.save b0f6760a9a3c67d3161839f69783cfda *tests/survreg2.R 13b9d7c952840e299affee6a5ab8d971 *tests/survreg2.Rout.save b5b10d8809abb91c7de72191ea3632ed *tests/survtest.R ef1affd024aa021ef07b70126064ac61 *tests/survtest.Rout.save e384df09bd4eae5b4783346a234bd334 *tests/testci.R da41b087a06be779507c0322556829d2 *tests/testci.Rout.save 69e4f2e2a58c449eaaa2b921ad3e2f50 *tests/testci2.R ad92dc9bcff6f26fe0e620f01f7a4d2f *tests/testci2.Rout.save 3eb98e4f934e6845ed620ea906658989 *tests/testnull.R 6a2a25754a6c6e37ed8ae5af3f6b5e34 *tests/testnull.Rout.save 8ad1d2cf5276fef4010a607c2d83ecdb *tests/testreg.R 2645c2ac783de52f9dc5e60f61e3102d *tests/testreg.Rout.save 63a88e76e6f31cae53a96889e1faa51d *tests/tiedtime.R b371603c7ff9af6d88b71eb432ffb823 *tests/tiedtime.Rout.save cbf88f37cce8fca1e91e7ae4c1d467e5 *tests/ties.rda 80de93676a12796ccaf4bec34ffac410 *tests/tt.R 034d3e089db16bb435452262139922d8 *tests/tt.Rout.save b1dd6e9b7bf7a5a26e8879c3f7fb03d9 *tests/turnbull.R d0e0c0f2ef0cbc69aebb7800b5fee3a2 *tests/turnbull.Rout.save 2d86e07bd20c9e396e7c5f2d498a53c0 *vignettes/adjcurve.Rnw 7c0cb35205be857bc07818a625aba559 *vignettes/compete.Rnw c3162ad5bf37d039ec9c0b4298b8c6ee *vignettes/refer.bib efb4204ec6f04a7b6f4d4a4f0009594e *vignettes/splines.Rnw 4e631d7fd5b4a003c7d62245d5fdb296 *vignettes/tests.Rnw 5dc7f0f37e6fd60751f807541b16c7a8 *vignettes/timedep.Rnw survival/build/0000755000175100001440000000000012545056257013255 5ustar hornikuserssurvival/build/vignette.rds0000644000175100001440000000056312545056257015620 0ustar hornikusers‹}RAOÂ0- P@.ÞìI.ÊÅ_` \Œ‰L<Ò¬¦°µMûró›kÇFÔCûí{{í{ûÞÞ;„:‰"·»E¢žÛZn]¹Õ$ Òvõ‚ñE¼2k¼ÊǺ±J5`²:l j#X,])pШªrýÈ+‹ÀéÄi‰5Kè0µþýíó*Aqo‘!ÐTqH,e’Ó܈Ô» l:ÉÍÜQ(t%£ ‰`XŒBÉÀ»ªÏòm³n5Ї~ŸN3ïöf3…©sOG ArH‡jÍŒp~ìŸÓ|~:±&v„üÄJœ0± ªªt޹„ÓE*(2ñÀ™O¤ y¸¦w¸;× ¾Ö<±§ÚØåÉÛrýíô?µQ/ÝÞ”,-†Øò`c,„©À¢‰^FcÿX+¾+O%äuþÛ2®¯ µÚ ‚Øeöï¹m·Û}Ÿ:Šfƒ£v8C6˜wÞu?{‡ŒÄ=survival/DESCRIPTION0000644000175100001440000000176412545156347013675 0ustar hornikusersTitle: Survival Analysis Maintainer: Terry M Therneau Priority: recommended Package: survival Version: 2.38-3 Depends: R (>= 2.13.0), graphics, stats Imports: splines, methods Suggests: cmprsk LazyData: Yes LazyLoad: Yes ByteCompile: Yes Authors@R: c(person(c("Terry", "M"), "Therneau", email="therneau.terry@mayo.edu", role=c("aut", "cre")), person("Thomas", "Lumley", role=c("ctb", "trl"), comment="original S->R port and maintainer until 2009")) Description: Contains the core survival analysis routines, including definition of Surv objects, Kaplan-Meier and Aalen-Johansen (multi-state) curves, Cox models, and parametric accelerated failure time models. License: LGPL (>= 2) NeedsCompilation: yes Packaged: 2015-07-01 21:33:35 UTC; therneau Author: Terry M Therneau [aut, cre], Thomas Lumley [ctb, trl] (original S->R port and maintainer until 2009) Repository: CRAN Date/Publication: 2015-07-02 08:40:39 survival/man/0000755000175100001440000000000012534312154012716 5ustar hornikuserssurvival/man/survexp.Rd0000644000175100001440000002025512417501331014722 0ustar hornikusers\name{survexp} \alias{survexp} \alias{print.survexp} \title{ Compute Expected Survival } \description{ Returns either the expected survival of a cohort of subjects, or the individual expected survival for each subject. } \usage{ survexp(formula, data, weights, subset, na.action, rmap, times, method=c("ederer", "hakulinen", "conditional", "individual.h", "individual.s"), cohort=TRUE, conditional=FALSE, ratetable=survival::survexp.us, scale=1, se.fit, model=FALSE, x=FALSE, y=FALSE) } \arguments{ \item{formula}{ formula object. The response variable is a vector of follow-up times and is optional. The predictors consist of optional grouping variables separated by the \code{+} operator (as in \code{survfit}), and is often \code{~1}, i.e., expected survival for the entire group. } \item{data}{ data frame in which to interpret the variables named in the \code{formula}, \code{subset} and \code{weights} arguments. } \item{weights}{ case weights. This is most useful when conditional survival for a known population is desired, e.g., the data set would contain all unique age/sex combinations and the weights would be the proportion of each. } \item{subset}{ expression indicating a subset of the rows of \code{data} to be used in the fit. } \item{na.action}{ function to filter missing data. This is applied to the model frame after \code{subset} has been applied. Default is \code{options()$na.action}. } \item{rmap}{ an optional list that maps data set names to the ratetable names. See the details section below. } \item{times}{ vector of follow-up times at which the resulting survival curve is evaluated. If absent, the result will be reported for each unique value of the vector of times supplied in the response value of the \code{formula}. } \item{method}{computational method for the creating the survival curves. The \code{individual} option does not create a curve, rather it retrieves the predicted survival \code{individual.s} or cumulative hazard \code{individual.h} for each subject. The default is to use \code{method='ederer'} if the formula has no response, and \code{method='hakulinen'} otherwise.} \item{cohort}{logical value. This argument has been superseded by the \code{method} argument. To maintain backwards compatability, if is present and TRUE, it implies \code{method='individual.s'}.} \item{conditional}{logical value. This argument has been superseded by the \code{method} argument. To maintain backwards compatability, if it is present and TRUE it implies \code{method='conditional'}.} \item{ratetable}{ a table of event rates, such as \code{survexp.mn}, or a fitted Cox model. Note the \code{survival::} prefix in the default argument is present to avoid the (rare) case of a user who expects the default table but just happens to have an object named "survexp.us" in their own directory.} \item{scale}{ numeric value to scale the results. If \code{ratetable} is in units/day, \code{scale = 365.25} causes the output to be reported in years. } \item{se.fit}{ compute the standard error of the predicted survival. This argument is currently ignored. Standard errors are not a defined concept for population rate tables (they are treated as coming from a complete census), and for Cox models the calculation is hard. Despite good intentions standard errors for this latter case have not been coded and validated. } \item{model,x,y}{ flags to control what is returned. If any of these is true, then the model frame, the model matrix, and/or the vector of response times will be returned as components of the final result, with the same names as the flag arguments. }} \value{ if \code{cohort=TRUE} an object of class \code{survexp}, otherwise a vector of per-subject expected survival values. The former contains the number of subjects at risk and the expected survival for the cohort at each requested time. The cohort survival is the hypothetical survival for a cohort of subjects enrolled from the population at large, but matching the data set on the factors found in the rate table. } \details{ Individual expected survival is usually used in models or testing, to `correct' for the age and sex composition of a group of subjects. For instance, assume that birth date, entry date into the study, sex and actual survival time are all known for a group of subjects. The \code{survexp.us} population tables contain expected death rates based on calendar year, sex and age. Then \preformatted{ haz <- survexp(fu.time ~ 1, data=mydata, rmap = list(year=entry.dt, age=(birth.dt-entry.dt)), method='individual.h')) } gives for each subject the total hazard experienced up to their observed death time or last follow-up time (variable fu.time) This probability can be used as a rescaled time value in models: \preformatted{ glm(status ~ 1 + offset(log(haz)), family=poisson) glm(status ~ x + offset(log(haz)), family=poisson) } In the first model, a test for intercept=0 is the one sample log-rank test of whether the observed group of subjects has equivalent survival to the baseline population. The second model tests for an effect of variable \code{x} after adjustment for age and sex. The ratetable being used may have different variable names than the user's data set, this is dealt with by the \code{rmap} argument. The rate table for the above calculation was \code{survexp.us}, a call to \code{summary{survexp.us}} reveals that it expects to have variables \code{age} = age in days, \code{sex}, and \code{year} = the date of study entry, we create them in the \code{rmap} line. The sex variable was not mapped, therefore the function assumes that it exists in \code{mydata} in the correct format. (Note: for factors such as sex, the program will match on any unique abbreviation, ignoring case.) Cohort survival is used to produce an overall survival curve. This is then added to the Kaplan-Meier plot of the study group for visual comparison between these subjects and the population at large. There are three common methods of computing cohort survival. In the "exact method" of Ederer the cohort is not censored, for this case no response variable is required in the formula. Hakulinen recommends censoring the cohort at the anticipated censoring time of each patient, and Verheul recommends censoring the cohort at the actual observation time of each patient. The last of these is the conditional method. These are obtained by using the respective time values as the follow-up time or response in the formula. } \references{ Berry, G. (1983). The analysis of mortality by the subject-years method. \emph{Biometrics}, 39:173-84. Ederer, F., Axtell, L. and Cutler, S. (1961). The relative survival rate: a statistical methodology. \emph{Natl Cancer Inst Monogr}, 6:101-21. Hakulinen, T. (1982). Cancer survival corrected for heterogeneity in patient withdrawal. \emph{Biometrics}, 38:933-942. Therneau, T. and Grambsch, P. (2000). Modeling survival data: Extending the Cox model. Springer. Chapter 10. Verheul, H., Dekker, E., Bossuyt, P., Moulijn, A. and Dunning, A. (1993). Background mortality in clinical survival studies. \emph{Lancet}, 341: 872-875. } \seealso{ \code{\link{survfit}}, \code{\link{pyears}}, \code{\link{survexp.us}}, \code{\link{survexp.fit}}. } \examples{ # # Stanford heart transplant data # We don't have sex in the data set, but know it to be nearly all males. # Estimate of conditional survival fit1 <- survexp(futime ~ 1, rmap=list(sex="male", year=accept.dt, age=(accept.dt-birth.dt)), method='conditional', data=jasa) summary(fit1, times=1:10*182.5, scale=365) #expected survival by 1/2 years # Estimate of expected survival stratified by prior surgery survexp(~ surgery, rmap= list(sex="male", year=accept.dt, age=(accept.dt-birth.dt)), method='ederer', data=jasa, times=1:10 * 182.5) ## Compare the survival curves for the Mayo PBC data to Cox model fit ## pfit <-coxph(Surv(time,status>0) ~ trt + log(bili) + log(protime) + age + platelet, data=pbc) plot(survfit(Surv(time, status>0) ~ trt, data=pbc), mark.time=FALSE) lines(survexp( ~ trt, ratetable=pfit, data=pbc), col='purple') } \keyword{survival} survival/man/survreg.control.Rd0000644000175100001440000000161711732700061016363 0ustar hornikusers\name{survreg.control} \alias{survreg.control} %- Also NEED an `\alias' for EACH other topic documented here. \title{Package options for survreg and coxph} \description{ This functions checks and packages the fitting options for \code{\link{survreg}} } \usage{ survreg.control(maxiter=30, rel.tolerance=1e-09, toler.chol=1e-10, iter.max, debug=0, outer.max=10) } %- maybe also `usage' for other objects documented here. \arguments{ \item{maxiter}{maximum number of iterations } \item{rel.tolerance}{relative tolerance to declare convergence } \item{toler.chol}{Tolerance to declare Cholesky decomposition singular} \item{iter.max}{same as \code{maxiter}} \item{debug}{print debugging information} \item{outer.max}{maximum number of outer iterations for choosing penalty parameters} } \value{ A list with the same elements as the input } \seealso{ \code{\link{survreg}}} \keyword{survival} survival/man/mgus2.Rd0000644000175100001440000000264712463757274014274 0ustar hornikusers\name{mgus2} \alias{mgus2} \docType{data} \title{Monoclonal gammapothy data} \description{Natural history of 1341 sequential patients with monoclonal gammapothy of undetermined significance (MGUS). } \usage{data("mgus2")} \format{ A data frame with 1384 observations on the following 10 variables. \describe{ \item{\code{id}}{subject identifier} \item{\code{age}}{age at diagnosis, in years} \item{\code{sex}}{a factor with levels \code{F} \code{M}} \item{\code{hgb}}{hemoglobin} \item{\code{creat}}{creatinine} \item{\code{mspike}}{size of the monoclonal serum splike} \item{\code{ptime}}{time until progression to a plasma cell malignancy (PCM) or last contact, in months} \item{\code{pstat}}{occurence of PCM: 0=no, 1=yes } \item{\code{futime}}{time until death or last contact, in months} \item{\code{death}}{occurence of death: 0=no, 1=yes} } } \details{ This is a larger follow-on study of the condition also found in data set \code{mgus}. } \source{Mayo Clinic data courtesy of Dr. Robert Kyle. All patient identifiers have been removed, age rounded to the nearest year, and follow-up times rounded to the nearest month.} \references{ R. Kyle, T. Therneau, V. Rajkumar, J. Offord, D. Larson, M. Plevak, and L. J. Melton III, A long-terms study of prognosis in monoclonal gammopathy of undertermined significance. New Engl J Med, 346:564-569 (2002). } \keyword{datasets} survival/man/lines.survfit.Rd0000644000175100001440000001021412264766610016030 0ustar hornikusers\name{lines.survfit} \alias{lines.survfit} \alias{points.survfit} \alias{lines.survexp} \title{ Add Lines or Points to a Survival Plot } \description{ Often used to add the expected survival curve(s) to a Kaplan-Meier plot generated with \code{plot.survfit}. } \usage{ \method{lines}{survfit}(x, type="s", mark=3, col=1, lty=1, lwd=1, cex=1, mark.time=TRUE, xscale=1, firstx=0, firsty=1, xmax, fun, conf.int=FALSE, ...) \method{lines}{survexp}(x, type="l", ...) \method{points}{survfit}(x, xscale, xmax, fun, ...) } \arguments{ \item{x}{ a survival object, generated from the \code{survfit} or \code{survexp} functions. } \item{type}{ the line type, as described in \code{lines}. The default is a step function for \code{survfit} objects, and a connected line for \code{survexp} objects. All other arguments for \code{lines.survexp} are identical to those for \code{lines.survfit}. } \item{mark, col, lty, lwd, cex}{ vectors giving the mark symbol, color, line type, line width and character size for the added curves. } \item{...}{other graphical parameters} \item{mark.time}{ controls the labeling of the curves. If \code{FALSE}, no labeling is done. If \code{TRUE}, then curves are marked at each censoring time. If \code{mark.time} is a numeric vector, then curves are marked at the specified time points. } \item{xscale}{ a number used to divide the x values. If time was originally in days, a value of 365.25 would give a plotted scale in years. } \item{firstx, firsty}{ the starting point for the survival curves. If either of these is set to \code{NA} or < blank > the plot will start at the first time point of the curve. } \item{xmax}{ the maximum horizontal plot coordinate. This shortens the curve before plotting it, so unlike using the \code{xlim} graphical parameter, warning messages about out of bounds points are not generated. } \item{fun}{ an arbitrary function defining a transformation of the survival curve. For example \code{fun=log} is an alternative way to draw a log-survival curve (but with the axis labeled with log(S) values). Four often used transformations can be specified with a character argument instead: "log" is the same as using the \code{log=T} option, "event" plots cumulative events (f(y) = 1-y), "cumhaz" plots the cumulative hazard function (f(y) = -log(y)) and "cloglog" creates a complimentary log-log survival plot (f(y) = log(-log(y) along with log scale for the x-axis). } \item{conf.int}{ if \code{TRUE}, confidence bands for the curves are also plotted. If set to \code{"only"}, then only the CI bands are plotted, and the curve itself is left off. This can be useful for fine control over the colors or line types of a plot. } } \value{ a list with components \code{x} and \code{y}, containing the coordinates of the last point on each of the curves (but not of the confidence limits). This may be useful for labeling. } \section{Side Effects}{ one or more curves are added to the current plot. } \seealso{ \code{\link{lines}}, \code{\link{par}}, \code{\link{plot.survfit}}, \code{\link{survfit}}, \code{\link{survexp}}. } \details{ When the \code{survfit} function creates a multi-state survival curve the resulting object has class `survfitms'. The only difference in the plots is that that it defaults to a curve that goes from lower left to upper right (starting at 0), where survival curves default to starting at 1 and going down. All other options are identical. } \examples{ fit <- survfit(Surv(time, status==2) ~ sex, pbc,subset=1:312) plot(fit, mark.time=FALSE, xscale=365.25, xlab='Years', ylab='Survival') lines(fit[1], lwd=2, xscale=365.24) #darken the first curve and add marks # Add expected survival curves for the two groups, # based on the US census data # The data set does not have entry date, use the midpoint of the study efit <- survexp(~ ratetable(sex=sex,age=age*365.35,year=as.Date('1979/1/1')) + sex, data=pbc, times=(0:24)*182) temp <- lines(efit, lty=2, xscale=365.24, lwd=2:1) text(temp, c("Male", "Female"), adj= -.1) #labels just past the ends title(main="Primary Biliary Cirrhosis, Observed and Expected") } \keyword{survival} survival/man/print.summary.survfit.Rd0000644000175100001440000000144111732700061017533 0ustar hornikusers\name{print.summary.survfit} \alias{print.summary.survfit} \title{ Print Survfit Summary } \description{ Prints the result of \code{summary.survfit}. } \usage{ \method{print}{summary.survfit}(x, digits = max(options() $digits-4, 3), ...) } \arguments{ \item{x}{ an object of class \code{"summary.survfit"}, which is the result of the \code{summary.survfit} function. } \item{digits}{ the number of digits to use in printing the numbers. } \item{\dots}{for future methods} } \value{ \code{x}, with the invisible flag set to prevent printing. } \section{Side Effects}{ prints the summary created by \code{summary.survfit}. } \seealso{ \code{\link{options}}, \code{\link{print}}, \code{\link{summary.survfit}}. } \keyword{print} % docclass is function % Converted by Sd2Rd version 37351. survival/man/colon.Rd0000644000175100001440000000511011732700061014311 0ustar hornikusers\name{colon} \alias{colon} \title{Chemotherapy for Stage B/C colon cancer} \usage{colon} \description{These are data from one of the first successful trials of adjuvant chemotherapy for colon cancer. Levamisole is a low-toxicity compound previously used to treat worm infestations in animals; 5-FU is a moderately toxic (as these things go) chemotherapy agent. There are two records per person, one for recurrence and one for death} \format{ \tabular{ll}{ id:\tab id\cr study:\tab 1 for all patients\cr rx:\tab Treatment - Obs(ervation), Lev(amisole), Lev(amisole)+5-FU\cr sex:\tab 1=male\cr age:\tab in years\cr obstruct:\tab obstruction of colon by tumour\cr perfor:\tab perforation of colon\cr adhere:\tab adherence to nearby organs\cr nodes:\tab number of lymph nodes with detectable cancer\cr time:\tab days until event or censoring\cr status:\tab censoring status\cr differ:\tab differentiation of tumour (1=well, 2=moderate, 3=poor)\cr extent:\tab Extent of local spread (1=submucosa, 2=muscle, 3=serosa, 4=contiguous structures)\cr surg:\tab time from surgery to registration (0=short, 1=long)\cr node4:\tab more than 4 positive lymph nodes\cr etype:\tab event type: 1=recurrence,2=death\cr }} \note{The study is originally described in Laurie (1989). The main report is found in Moertel (1990). This data set is closest to that of the final report in Moertel (1991). A version of the data with less follow-up time was used in the paper by Lin (1994). } \references{ JA Laurie, CG Moertel, TR Fleming, HS Wieand, JE Leigh, J Rubin, GW McCormack, JB Gerstner, JE Krook and J Malliard. Surgical adjuvant therapy of large-bowel carcinoma: An evaluation of levamisole and the combination of levamisole and fluorouracil: The North Central Cancer Treatment Group and the Mayo Clinic. J Clinical Oncology, 7:1447-1456, 1989. DY Lin. Cox regression analysis of multivariate failure time data: the marginal approach. Statistics in Medicine, 13:2233-2247, 1994. CG Moertel, TR Fleming, JS MacDonald, DG Haller, JA Laurie, PJ Goodman, JS Ungerleider, WA Emerson, DC Tormey, JH Glick, MH Veeder and JA Maillard. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. New England J of Medicine, 332:352-358, 1990. CG Moertel, TR Fleming, JS MacDonald, DG Haller, JA Laurie, CM Tangen, JS Ungerleider, WA Emerson, DC Tormey, JH Glick, MH Veeder and JA Maillard, Fluorouracil plus Levamisole as and effective adjuvant therapy after resection of stage II colon carcinoma: a final report. Annals of Internal Med, 122:321-326, 1991. } \keyword{survival} survival/man/survreg.distributions.Rd0000644000175100001440000001032312160146127017602 0ustar hornikusers\name{survreg.distributions} \alias{survreg.distributions} \title{Parametric Survival Distributions} \usage{ survreg.distributions } \description{ List of distributions for accelerated failure models. These are location-scale families for some transformation of time. The entry describes the cdf \eqn{F} and density \eqn{f} of a canonical member of the family. } \format{ There are two basic formats, the first defines a distribution de novo, the second defines a new distribution in terms of an old one. \tabular{ll}{ name:\tab name of distribution\cr variance:\tab function(parms) returning the variance (currently unused)\cr init(x,weights,...):\tab Function returning an initial\cr \tab estimate of the mean and variance \cr \tab (used for initial values in the iteration)\cr density(x,parms):\tab Function returning a matrix with columns \eqn{F},\cr \tab \eqn{1-F},\eqn{f},\eqn{f'/f},\eqn{f''/f}\cr quantile(p,parms):\tab Quantile function\cr scale:\tab Optional fixed value for the scale parameter\cr parms:\tab Vector of default values and names for any additional parameters\cr deviance(y,scale,parms):\tab Function returning the deviance for a\cr \tab saturated model; used only for deviance residuals. } and to define one distribution in terms of another \tabular{ll}{ name:\tab name of distribution\cr dist:\tab name of parent distribution\cr trans:\tab transformation (eg log)\cr dtrans:\tab derivative of transformation\cr itrans:\tab inverse of transformation\cr scale:\tab Optional fixed value for scale parameter\cr } } \details{ There are four basic distributions:\code{extreme}, \code{gaussian}, \code{logistic} and \code{t}. The last three are parametrised in the same way as the distributions already present in \R. The extreme value cdf is \deqn{F=1-e^{-e^t}.} When the logarithm of survival time has one of the first three distributions we obtain respectively \code{weibull}, \code{lognormal}, and \code{loglogistic}. The location-scale parameterizaion of a Weibull distribution found in \code{survreg} is not the same as the parameterization of \code{\link{rweibull}}. The other predefined distributions are defined in terms of these. The \code{exponential} and \code{rayleigh} distributions are Weibull distributions with fixed \code{scale} of 1 and 0.5 respectively, and \code{loggaussian} is a synonym for \code{lognormal}. For speed parts of the three most commonly used distributions are hardcoded in C; for this reason the elements of \code{survreg.distributions} with names of "Extreme value", "Logisitic" and "Gaussian" should not be modified. (The order of these in the list is not important, recognition is by name.) As an alternative to modifying \code{survreg.distributions} a new distribution can be specified as a separate list. This is the preferred method of addition and is illustrated below. } \seealso{\code{\link{survreg}}, \code{\link{pweibull}}, \code{\link{pnorm}},\code{\link{plogis}}, \code{\link{pt}}, \code{\link{survregDtest}} } \examples{ # time transformation survreg(Surv(time, status) ~ ph.ecog + sex, dist='weibull', data=lung) # change the transformation to work in years # intercept changes by log(365), everything else stays the same my.weibull <- survreg.distributions$weibull my.weibull$trans <- function(y) log(y/365) my.weibull$itrans <- function(y) 365*exp(y) survreg(Surv(time, status) ~ ph.ecog + sex, lung, dist=my.weibull) # Weibull parametrisation y<-rweibull(1000, shape=2, scale=5) survreg(Surv(y)~1, dist="weibull") # survreg scale parameter maps to 1/shape, linear predictor to log(scale) # Cauchy fit mycauchy <- list(name='Cauchy', init= function(x, weights, ...) c(median(x), mad(x)), density= function(x, parms) { temp <- 1/(1 + x^2) cbind(.5 + atan(x)/pi, .5+ atan(-x)/pi, temp/pi, -2 *x*temp, 2*temp*(4*x^2*temp -1)) }, quantile= function(p, parms) tan((p-.5)*pi), deviance= function(...) stop('deviance residuals not defined') ) survreg(Surv(log(time), status) ~ ph.ecog + sex, lung, dist=mycauchy) } \keyword{survival} survival/man/attrassign.Rd0000644000175100001440000000333312055773252015376 0ustar hornikusers\name{attrassign} \alias{attrassign.default} \alias{attrassign} \alias{attrassign.lm} \title{Create new-style "assign" attribute} \description{ The \code{"assign"} attribute on model matrices describes which columns come from which terms in the model formula. It has two versions. R uses the original version, but the alternate version found in S-plus is sometimes useful. } \usage{ \method{attrassign}{default}(object, tt,...) \method{attrassign}{lm}(object,...) } %- maybe also `usage' for other objects documented here. \arguments{ \item{object}{model matrix or linear model object} \item{tt}{terms object} \item{...}{ignored} } \value{ A list with names corresponding to the term names and elements that are vectors indicating which columns come from which terms } \details{ For instance consider the following \preformatted{ survreg(Surv(time, status) ~ age + sex + factor(ph.ecog), lung) } R gives the compact for for assign, a vector (0, 1, 2, 3, 3, 3); which can be read as ``the first column of the X matrix (intercept) goes with none of the terms, the second column of X goes with term 1 of the model equation, the third column of X with term 2, and columns 4-6 with term 3''. The alternate (S-Plus default) form is a list \preformatted{ $(Intercept) 1 $age 2 $sex 3 $factor(ph.ecog) 4 5 6 } } \seealso{\code{\link{terms}},\code{\link{model.matrix}}} \examples{ formula <- Surv(time,status)~factor(ph.ecog) tt <- terms(formula) mf <- model.frame(tt,data=lung) mm <- model.matrix(tt,mf) ## a few rows of data mm[1:3,] ## old-style assign attribute attr(mm,"assign") ## alternate style assign attribute attrassign(mm,tt) } \keyword{models} survival/man/rats2.Rd0000644000175100001440000000137412052731313014242 0ustar hornikusers\name{rats2} \alias{rats2} \docType{data} \title{Rat data from Gail et al.} \description{48 rats were injected with a carcinogen, and then randomized to either drug or placebo. The number of tumors ranges from 0 to 13; all rats were censored at 6 months after randomization. } \usage{rats2} \format{ \tabular{ll}{ rat:\tab id\cr trt:\tab treatment,(1=drug, 0=control) \cr observation:\tab within rat\cr start:\tab entry time\cr stop:\tab exit time\cr status:\tab event status, 1=tumor, 0=censored\cr } } \source{ MH Gail, TJ Santner, and CC Brown (1980), An analysis of comparative carcinogenesis experiments based on multiple times to tumor. \emph{Biometrics} \bold{36}, 255--266. } \keyword{survival} \keyword{datasets} survival/man/rats.Rd0000644000175100001440000000221712453556344014173 0ustar hornikusers\name{rats} \alias{rats} \docType{data} \title{Rat treatment data from Mantel et al} \description{Rat treatment data from Mantel et al. Three rats were chosen from each of 100 litters, one of which was treated with a drug, and then all followed for tumor incidence. } \usage{rats} \format{ \tabular{ll}{ litter:\tab litter number from 1 to 100\cr rx:\tab treatment,(1=drug, 0=control) \cr time:\tab time to tumor or last follow-up\cr status:\tab event status, 1=tumor and 0=censored\cr sex:\tab male or female } } \source{ N. Mantel, N. R. Bohidar and J. L. Ciminera. Mantel-Haenszel analyses of litter-matched time to response data, with modifications for recovery of interlitter information. Cancer Research, 37:3863-3868, 1977. } \references{ E. W. Lee, L. J. Wei, and D. Amato, Cox-type regression analysis for large number of small groups of correlated failure time observations, in "Survival Analysis, State of the Art", Kluwer, 1992. } \note{Since only 2/150 of the male rats have a tumor, most analyses use only females (odd numbered litters), e.g. Lee et al.} \keyword{survival} \keyword{datasets} survival/man/survreg.Rd0000644000175100001440000001033712415007440014703 0ustar hornikusers\name{survreg} \alias{survreg} \alias{model.frame.survreg} \alias{labels.survreg} \alias{print.survreg.penal} \alias{print.summary.survreg} \alias{survReg} \alias{anova.survreg} \alias{vcov.survreg} \alias{anova.survreglist} \title{ Regression for a Parametric Survival Model } \description{ Fit a parametric survival regression model. These are location-scale models for an arbitrary transform of the time variable; the most common cases use a log transformation, leading to accelerated failure time models. } \usage{ survreg(formula, data, weights, subset, na.action, dist="weibull", init=NULL, scale=0, control,parms=NULL,model=FALSE, x=FALSE, y=TRUE, robust=FALSE, score=FALSE, \dots) } \arguments{ \item{formula}{ a formula expression as for other regression models. The response is usually a survival object as returned by the \code{Surv} function. See the documentation for \code{Surv}, \code{lm} and \code{formula} for details. } \item{data}{ a data frame in which to interpret the variables named in the \code{formula}, \code{weights} or the \code{subset} arguments. } \item{weights}{optional vector of case weights} \item{subset}{ subset of the observations to be used in the fit } \item{na.action}{ a missing-data filter function, applied to the model.frame, after any \code{subset} argument has been used. Default is \code{options()\$na.action}. } \item{dist}{ assumed distribution for y variable. If the argument is a character string, then it is assumed to name an element from \code{\link{survreg.distributions}}. These include \code{"weibull"}, \code{"exponential"}, \code{"gaussian"}, \code{"logistic"},\code{"lognormal"} and \code{"loglogistic"}. Otherwise, it is assumed to be a user defined list conforming to the format described in \code{\link{survreg.distributions}}. } \item{parms}{ a list of fixed parameters. For the t-distribution for instance this is the degrees of freedom; most of the distributions have no parameters. } \item{init}{ optional vector of initial values for the parameters. } \item{scale}{ optional fixed value for the scale. If set to <=0 then the scale is estimated. } \item{control}{ a list of control values, in the format produced by \code{\link{survreg.control}}. The default value is \code{survreg.control()} } \item{model,x,y}{ flags to control what is returned. If any of these is true, then the model frame, the model matrix, and/or the vector of response times will be returned as components of the final result, with the same names as the flag arguments.} \item{score}{return the score vector. (This is expected to be zero upon successful convergence.) } \item{robust}{Use robust 'sandwich' standard errors, based on independence of individuals if there is no \code{cluster()} term in the formula, based on independence of clusters if there is.} \item{\dots}{ other arguments which will be passed to \code{survreg.control}. }} \value{ an object of class \code{survreg} is returned. } \seealso{ \code{\link{survreg.object}}, \code{\link{survreg.distributions}}, \code{\link{pspline}}, \code{\link{frailty}}, \code{\link{ridge}} } \examples{ # Fit an exponential model: the two fits are the same survreg(Surv(futime, fustat) ~ ecog.ps + rx, ovarian, dist='weibull', scale=1) survreg(Surv(futime, fustat) ~ ecog.ps + rx, ovarian, dist="exponential") # # A model with different baseline survival shapes for two groups, i.e., # two different scale parameters survreg(Surv(time, status) ~ ph.ecog + age + strata(sex), lung) # There are multiple ways to parameterize a Weibull distribution. The survreg # function embeds it in a general location-scale family, which is a # different parameterization than the rweibull function, and often leads # to confusion. # survreg's scale = 1/(rweibull shape) # survreg's intercept = log(rweibull scale) # For the log-likelihood all parameterizations lead to the same value. y <- rweibull(1000, shape=2, scale=5) survreg(Surv(y)~1, dist="weibull") # Economists fit a model called `tobit regression', which is a standard # linear regression with Gaussian errors, and left censored data. tobinfit <- survreg(Surv(durable, durable>0, type='left') ~ age + quant, data=tobin, dist='gaussian') } \keyword{survival} survival/man/uspop2.Rd0000644000175100001440000000205312111727573014443 0ustar hornikusers\name{uspop2} \alias{uspop2} \docType{data} \title{Projected US Population} \description{US population by age and sex, for 2000 through 2020} \usage{data(uspop2)} \format{ The data is a matrix with dimensions age, sex, and calendar year. Age goes from 0 through 100, where the value for age 100 is the total for all ages of 100 or greater. } \details{ This data is often used as a "standardized" population for epidemiolgy studies.} \source{ NP2008_D1: Projected Population by Single Year of Age, Sex, Race, and Hispanic Origin for the United States: July 1, 2000 to July 1, 2050, www.census.gov/population/projections. } \examples{ us50 <- uspop2[51:101,, "2000"] #US 2000 population, 50 and over age <- as.integer(dimnames(us50)[[1]]) smat <- model.matrix( ~ factor(floor(age/5)) -1) ustot <- t(smat) \%*\% us50 #totals by 5 year age groups temp <- c(50,55, 60, 65, 70, 75, 80, 85, 90, 95) dimnames(ustot) <- list(c(paste(temp, temp+4, sep="-"), "100+"), c("male", "female")) } \seealso{\code{\link{uspop}}} \keyword{datasets} survival/man/survfitcoxph.fit.Rd0000644000175100001440000000567211732700061016541 0ustar hornikusers\name{survfitcoxph.fit} \alias{survfitcoxph.fit} \title{ A direct interface to the `computational engine' of survfit.coxph } \description{ This program is mainly supplied to allow other packages to invoke the survfit.coxph function at a `data' level rather than a `user' level. It does no checks on the input data that is provided, which can lead to unexpected errors if that data is wrong. } \usage{ survfitcoxph.fit(y, x, wt, x2, risk, newrisk, strata, se.fit, survtype, vartype, varmat, id, y2, strata2, unlist=TRUE) } %- maybe also 'usage' for other objects documented here. \arguments{ \item{y}{the response variable used in the Cox model. (Missing values removed of course.) } \item{x}{covariate matrix used in the Cox model } \item{wt}{weight vector for the Cox model. If the model was unweighted use a vector of 1s. } \item{x2}{matrix describing the hypothetical subjects for which a curve is desired. Must have the same number of columns as \code{x}. } \item{risk}{the risk score exp(X beta) from the fitted Cox model. If the model had an offset, include it in the argument to exp. } \item{newrisk}{risk scores for the hypothetical subjects } \item{strata}{strata variable used in the Cox model. This will be a factor. } \item{se.fit}{if \code{TRUE} the standard errors of the curve(s) are returned } \item{survtype}{1=Kalbfleish-Prentice, 2=Nelson-Aalen, 3=Efron. It is usual to match this to the approximation for ties used in the \code{coxph} model: KP for `exact', N-A for `breslow' and Efron for `efron'. } \item{vartype}{1=Greenwood, 2=Aalen, 3=Efron } \item{varmat}{the variance matrix of the coefficients } \item{id}{optional; if present and not NULL this should be a vector of identifiers of length \code{nrow(x2)}. A mon-null value signifies that \code{x2} contains time dependent covariates, in which case this identifies which rows of \code{x2} go with each subject. } \item{y2}{survival times, for time dependent prediction. It gives the time range (time1,time2] for each row of \code{x2}. Note: this must be a Surv object and thus contains a status indicator, which is never used in the routine, however. } \item{strata2}{vector of strata indicators for \code{x2}. This must be a factor. } \item{unlist}{if \code{FALSE} the result will be a list with one element for each strata. Otherwise the strata are ``unpacked'' into the form found in a \code{survfit} object.} } \value{a list containing nearly all the components of a \code{survfit} object. All that is missing is to add the confidence intervals, the type of the original model's response (as in a coxph object), and the class. } \note{The source code for for both this function and \code{survfit.coxph} is written using noweb. For complete documentation see the \code{inst/sourcecode.pdf} file. } \author{Terry Therneau} \seealso{\code{\link{survfit.coxph}} } \keyword{survival} survival/man/ratetable.Rd0000644000175100001440000000201712030336264015147 0ustar hornikusers\name{ratetable} \alias{ratetable} \alias{[.ratetable} \alias{[.ratetable2} \alias{print.ratetable} \alias{is.na.ratetable} \alias{summary.ratetable} \title{Ratetable reference in formula} \description{ This function matches variable names in data to those in a ratetable for \code{\link{survexp}} } \usage{ ratetable(...) } \arguments{ \item{\dots}{tags matching dimensions of the ratetable and variables in the data frame (see example)} } \value{ A data frame } \seealso{\code{\link{survexp}},\code{\link{survexp.us}},\code{\link{is.ratetable}}} \examples{ fit <- survfit(Surv(time, status) ~ sex, pbc,subset=1:312) # The data set does not have entry date, use the midpoint of the study efit <- survexp(~ ratetable(sex=sex,age=age*365.35,year=as.Date('1979/1/1')) + sex, data=pbc, times=(0:24)*182) \dontrun{ plot(fit, mark.time=F, xscale=365.25, xlab="Years post diagnosis", ylab="Survival") lines(efit, col=2, xscale=365.25) # Add the expected survival line } } \keyword{survival}%-- one or more ... survival/man/residuals.survreg.Rd0000644000175100001440000000531111732700061016671 0ustar hornikusers\name{residuals.survreg} \alias{residuals.survreg} \alias{residuals.survreg.penal} \title{Compute Residuals for `survreg' Objects} \description{ This is a method for the function \code{\link{residuals}} for objects inheriting from class \code{survreg}. } \usage{ \method{residuals}{survreg}(object, type=c("response", "deviance","dfbeta","dfbetas", "working","ldcase","ldresp","ldshape", "matrix"), rsigma=TRUE, collapse=FALSE, weighted=FALSE, ...) } \arguments{ \item{object}{ an object inheriting from class \code{survreg}. } \item{type}{ type of residuals, with choices of \code{"response"}, \code{"deviance"}, \code{"dfbeta"}, \code{"dfbetas"}, \code{"working"}, \code{"ldcase"}, \code{"lsresp"}, \code{"ldshape"}, and \code{"matrix"}. See the LaTeX documentation (\code{survival/doc/survival.ps.gz}) for more detail. } \item{rsigma}{ include the scale parameters in the variance matrix, when doing computations. (I can think of no good reason not to). } \item{collapse}{ optional vector of subject groups. If given, this must be of the same length as the residuals, and causes the result to be per group residuals. } \item{weighted}{ give weighted residuals? Normally residuals are unweighted. }\item{...}{other unused arguments}} \value{ A vector or matrix of residuals is returned. Response residuals are on the scale of the original data, working residuals are on the scale of the linear predictor, and deviance residuals are on log-likelihood scale. The dfbeta residuals are a matrix, where the ith row gives the approximate change in the coefficients due to the addition of subject i. The dfbetas matrix contains the dfbeta residuals, with each column scaled by the standard deviation of that coefficient. The matrix type produces a matrix based on derivatives of the log-likelihood function. Let \eqn{L} be the log-likelihood, \eqn{p} be the linear predictor \eqn{X\beta}{X \%*\% coef}, and \eqn{s} be \eqn{\log(\sigma)}. Then the 6 columns of the matrix are \eqn{L}, \eqn{dL/dp},\eqn{\partial^2L/\partial p^2}{ddL/(dp dp)}, \eqn{dL/ds}, \eqn{\partial^2L/\partial s^2}{ddL/(ds ds)} and \eqn{\partial^2L/\partial p\partial s}{ddL/(dp ds)}. Diagnostics based on these quantities are discussed in an article by Escobar and Meeker. The main ones are the likelihood displacement residuals for perturbation of a case weight (\code{ldcase}), the response value (\code{ldresp}), and the \code{shape}. } \references{ Escobar, L. A. and Meeker, W. Q. (1992). Assessing influence in regression analysis with censored data. \emph{Biometrics} \bold{48}, 507-528. } \seealso{\code{\link{predict.survreg}}} \examples{ fit <- survreg(Surv(time,status) ~x, aml) rr <- residuals(fit, type='matrix') } \keyword{survival} % Converted by Sd2Rd version 0.3-2. survival/man/print.survfit.Rd0000644000175100001440000000713012461741246016052 0ustar hornikusers\name{print.survfit} \alias{print.survfit} \title{ Print a Short Summary of a Survival Curve } \description{ Print number of observations, number of events, the restricted mean survival and its standard error, and the median survival with confidence limits for the median. } \usage{ \method{print}{survfit}(x, scale=1, digits = max(options()$digits - 4,3), print.rmean=getOption("survfit.print.rmean"), rmean = getOption('survfit.rmean'),...) } \arguments{ \item{x}{ the result of a call to the \code{survfit} function. } \item{scale}{ a numeric value to rescale the survival time, e.g., if the input data to survfit were in days, \code{scale=365} would scale the printout to years. } \item{digits}{Number of digits to print} \item{print.rmean,rmean}{Options for computation and display of the restricted mean.} \item{\dots}{for future results} } \value{ x, with the invisible flag set to prevent printing. (The default for all print functions in R is to return the object passed to them; print.survfit complies with this pattern. If you want to capture these printed results for further processing, see the \code{table} component of \code{summary.survfit}.) } \section{Side Effects}{ The number of observations, the number of events, the median survival with its confidence interval, and optionally the restricted mean survival (\code{rmean}) and its standard error, are printed. If there are multiple curves, there is one line of output for each. } \details{ The mean and its variance are based on a truncated estimator. That is, if the last observation(s) is not a death, then the survival curve estimate does not go to zero and the mean is undefined. There are four possible approaches to resolve this, which are selected by the \code{rmean} option. The first is to set the upper limit to a constant, e.g.,\code{rmean=365}. In this case the reported mean would be the expected number of days, out of the first 365, that would be experienced by each group. This is useful if interest focuses on a fixed period. Other options are \code{"none"} (no estimate), \code{"common"} and \code{"individual"}. The \code{"common"} option uses the maximum time for all curves in the object as a common upper limit for the auc calculation. For the \code{"individual"}options the mean is computed as the area under each curve, over the range from 0 to the maximum observed time for that curve. Since the end point is random, values for different curves are not comparable and the printed standard errors are an underestimate as they do not take into account this random variation. This option is provided mainly for backwards compatability, as this estimate was the default (only) one in earlier releases of the code. Note that SAS (as of version 9.3) uses the integral up to the last \emph{event} time of each individual curve; we consider this the worst of the choices and do not provide an option for that calculation. The median and its confidence interval are defined by drawing a horizontal line at 0.5 on the plot of the survival curve and its confidence bands. The intersection of the line with the lower CI band defines the lower limit for the median's interval, and similarly for the upper band. If any of the intersections is not a point, then we use the smallest point of intersection, e.g., if the survival curve were exactly equal to 0.5 over an interval. } \section{References}{ Miller, Rupert G., Jr. (1981). \emph{Survival Analysis.} New York:Wiley, p 71. } \seealso{ \code{\link{summary.survfit}}, \code{\link{quantile.survfit}} } \keyword{survival} % docclass is function % Converted by Sd2Rd version 37351. survival/man/summary.survexp.Rd0000644000175100001440000000266111732700061016417 0ustar hornikusers\name{summary.survexp} \alias{summary.survexp} \title{Summary function for a survexp object} \description{ Returns a list containing the values of the survival at specified times. } \usage{ \method{summary}{survexp}(object, times, scale = 1, ...) } \arguments{ \item{object}{ the result of a call to the \code{survexp} function } \item{times}{ vector of times; the returned matrix will contain 1 row for each time. Missing values are not allowed. } \item{scale}{ numeric value to rescale the survival time, e.g., if the input data to \code{survfit} were in days, \code{scale = 365.25} would scale the output to years. } \item{\dots}{For future methods} } \details{ A primary use of this function is to retreive survival at fixed time points, which will be properly interpolated by the function. } \value{ a list with the following components: \item{surv}{ the estimate of survival at time t. } \item{time}{ the timepoints on the curve. } \item{n.risk}{ In expected survival each subject from the data set is matched to a hypothetical person from the parent population, matched on the characteristics of the parent population. The number at risk is the number of those hypothetical subject who are still part of the calculation. } } \author{Terry Therneau} \seealso{\code{\link{survexp}} } % Add one or more standard keywords, see file 'KEYWORDS' in the % R documentation directory. \keyword{ survival } survival/man/survdiff.Rd0000644000175100001440000000710311732700061015033 0ustar hornikusers\name{survdiff} \alias{survdiff} \alias{print.survdiff} \title{ Test Survival Curve Differences } \description{ Tests if there is a difference between two or more survival curves using the \eqn{G^\rho}{G-rho} family of tests, or for a single curve against a known alternative. } \usage{ survdiff(formula, data, subset, na.action, rho=0) } \arguments{ \item{formula}{ a formula expression as for other survival models, of the form \code{Surv(time, status) ~ predictors}. For a one-sample test, the predictors must consist of a single \code{offset(sp)} term, where \code{sp} is a vector giving the survival probability of each subject. For a k-sample test, each unique combination of predictors defines a subgroup. A \code{strata} term may be used to produce a stratified test. To cause missing values in the predictors to be treated as a separate group, rather than being omitted, use the \code{strata} function with its \code{na.group=T} argument. } \item{data}{ an optional data frame in which to interpret the variables occurring in the formula. } \item{subset}{ expression indicating which subset of the rows of data should be used in the fit. This can be a logical vector (which is replicated to have length equal to the number of observations), a numeric vector indicating which observation numbers are to be included (or excluded if negative), or a character vector of row names to be included. All observations are included by default. } \item{na.action}{ a missing-data filter function. This is applied to the \code{model.frame} after any subset argument has been used. Default is \code{options()$na.action}. } \item{rho}{ a scalar parameter that controls the type of test. }} \value{ a list with components: \item{n}{ the number of subjects in each group. } \item{obs}{ the weighted observed number of events in each group. If there are strata, this will be a matrix with one column per stratum. } \item{exp}{ the weighted expected number of events in each group. If there are strata, this will be a matrix with one column per stratum. } \item{chisq}{ the chisquare statistic for a test of equality. } \item{var}{ the variance matrix of the test. } \item{strata}{ optionally, the number of subjects contained in each stratum. }} \section{METHOD}{ This function implements the G-rho family of Harrington and Fleming (1982), with weights on each death of \eqn{S(t)^\rho}{S(t)^rho}, where \eqn{S(t)}{S} is the Kaplan-Meier estimate of survival. With \code{rho = 0} this is the log-rank or Mantel-Haenszel test, and with \code{rho = 1} it is equivalent to the Peto & Peto modification of the Gehan-Wilcoxon test. If the right hand side of the formula consists only of an offset term, then a one sample test is done. To cause missing values in the predictors to be treated as a separate group, rather than being omitted, use the \code{factor} function with its \code{exclude} argument. } \references{ Harrington, D. P. and Fleming, T. R. (1982). A class of rank test procedures for censored survival data. \emph{Biometrika} \bold{69}, 553-566.} \examples{ ## Two-sample test survdiff(Surv(futime, fustat) ~ rx,data=ovarian) ## Stratified 7-sample test survdiff(Surv(time, status) ~ pat.karno + strata(inst), data=lung) ## Expected survival for heart transplant patients based on ## US mortality tables expect <- survexp(futime ~ ratetable(age=(accept.dt - birth.dt), sex=1,year=accept.dt,race="white"), jasa, cohort=FALSE, ratetable=survexp.usr) ## actual survival is much worse (no surprise) survdiff(Surv(jasa$futime, jasa$fustat) ~ offset(expect)) } \keyword{survival} % Converted by Sd2Rd version 0.3-2. and hand edited. survival/man/bladder.Rd0000644000175100001440000000604412461741246014615 0ustar hornikusers\name{bladder} \docType{data} \alias{bladder} \alias{bladder1} \alias{bladder2} \title{Bladder Cancer Recurrences} \usage{ bladder1 bladder bladder2 } \description{Data on recurrences of bladder cancer, used by many people to demonstrate methodology for recurrent event modelling. Bladder1 is the full data set from the study. It contains all three treatment arms and all recurrences for 118 subjects; the maximum observed number of recurrences is 9. Bladder is the data set that appears most commonly in the literature. It uses only the 85 subjects with nonzero follow-up who were assigned to either thiotepa or placebo, and only the first four recurrences for any patient. The status variable is 1 for recurrence and 0 for everything else (including death for any reason). The data set is laid out in the competing risks format of the paper by Wei, Lin, and Weissfeld. Bladder2 uses the same subset of subjects as bladder, but formated in the (start, stop] or Anderson-Gill style. Note that in transforming from the WLW to the AG style data set there is a quite common programming mistake that leads to extra follow-up time for 12 subjects: all those with follow-up beyond their 4th recurrence). Over this extended time these subjects are by definition not at risk for another event in the WLW data set. } \format{ bladder1 \tabular{ll}{ id:\tab Patient id\cr treatment:\tab Placebo, pyridoxine (vitamin B6), or thiotepa\cr number:\tab Initial number of tumours (8=8 or more)\cr size:\tab Size (cm) of largest initial tumour\cr recur:\tab Number of recurrences \cr start,stop:\tab The start and end time of each time interval\cr status:\tab End of interval code, 0=censored, 1=recurrence, \cr \tab 2=death from bladder disease, 3=death other/unknown cause\cr rtumor:\tab Number of tumors found at the time of a recurrence\cr rsize:\tab Size of largest tumor at a recurrence\cr enum:\tab Event number (observation number within patient)\cr } bladder \tabular{ll}{ id:\tab Patient id\cr rx:\tab Treatment 1=placebo 2=thiotepa\cr number:\tab Initial number of tumours (8=8 or more)\cr size:\tab size (cm) of largest initial tumour\cr stop:\tab recurrence or censoring time\cr enum:\tab which recurrence (up to 4)\cr } bladder2 \tabular{ll}{ id:\tab Patient id\cr rx:\tab Treatment 1=placebo 2=thiotepa\cr number:\tab Initial number of tumours (8=8 or more)\cr size:\tab size (cm) of largest initial tumour\cr start:\tab start of interval (0 or previous recurrence time)\cr stop:\tab recurrence or censoring time\cr enum:\tab which recurrence (up to 4)\cr } } \source{ Andrews DF, Hertzberg AM (1985), DATA: A Collection of Problems from Many Fields for the Student and Research Worker, New York: Springer-Verlag. LJ Wei, DY Lin, L Weissfeld (1989), Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. \emph{Journal of the American Statistical Association}, \bold{84}. } \keyword{datasets} \keyword{survival} survival/man/pbcseq.Rd0000644000175100001440000000661312334154254014473 0ustar hornikusers\name{pbcseq} \alias{pbcseq} \docType{data} \title{Mayo Clinic Primary Biliary Cirrhosis, sequential data} \description{ This data is a continuation of the PBC data set, and contains the follow-up laboratory data for each study patient. An analysis based on the data can be found in Murtagh, et. al. The primary PBC data set contains only baseline measurements of the laboratory paramters. This data set contains multiple laboratory results, but only on the 312 randomized patients. Some baseline data values in this file differ from the original PBC file, for instance, the data errors in prothrombin time and age which were discovered after the orignal analysis (see Fleming and Harrington, figure 4.6.7). One "feature" of the data deserves special comment. The last observation before death or liver transplant often has many more missing covariates than other data rows. The original clinical protocol for these patients specified visits at 6 months, 1 year, and annually thereafter. At these protocol visits lab values were obtained for a large pre-specified battery of tests. "Extra" visits, often undertaken because of worsening medical condition, did not necessarily have all this lab work. The missing values are thus potentially informative. } \usage{pbc} \format{ \tabular{ll}{ id:\tab case number\cr age:\tab in years\cr sex:\tab m/f\cr trt:\tab 1/2/NA for D-penicillmain, placebo, not randomised\cr time:\tab number of days between registration and the earlier of death,\cr \tab transplantion, or study analysis in July, 1986\cr status:\tab status at endpoint, 0/1/2 for censored, transplant, dead\cr day:\tab number of days between enrollment and this visit date\cr \tab all measurements below refer to this date\cr albumin:\tab serum albumin (mg/dl)\cr alk.phos:\tab alkaline phosphotase (U/liter)\cr ascites:\tab presence of ascites \cr ast:\tab aspartate aminotransferase, once called SGOT (U/ml)\cr bili:\tab serum bilirunbin (mg/dl)\cr chol:\tab serum cholesterol (mg/dl)\cr copper:\tab urine copper (ug/day)\cr edema:\tab 0 no edema, 0.5 untreated or successfully treated\cr \tab 1 edema despite diuretic therapy\cr hepato:\tab presence of hepatomegaly or enlarged liver\cr platelet:\tab platelet count\cr protime:\tab standardised blood clotting time\cr spiders:\tab blood vessel malformations in the skin\cr stage:\tab histologic stage of disease (needs biopsy)\cr trig:\tab triglycerides (mg/dl)\cr } } \source{ T Therneau and P Grambsch, "Modeling Survival Data: Extending the Cox Model", Springer-Verlag, New York, 2000. ISBN: 0-387-98784-3. } \examples{ # Create the start-stop-event triplet needed for coxph first <- with(pbcseq, c(TRUE, diff(id) !=0)) #first id for each subject last <- c(first[-1], TRUE) #last id time1 <- with(pbcseq, ifelse(first, 0, day)) time2 <- with(pbcseq, ifelse(last, futime, c(day[-1], 0))) event <- with(pbcseq, ifelse(last, status, 0)) fit1 <- coxph(Surv(time1, time2, event) ~ age + sex + log(bili), pbcseq) } \references{ Murtaugh PA. Dickson ER. Van Dam GM. Malinchoc M. Grambsch PM. Langworthy AL. Gips CH. "Primary biliary cirrhosis: prediction of short-term survival based on repeated patient visits." Hepatology. 20(1.1):126-34, 1994. Fleming T and Harrington D., "Counting Processes and Survival Analysis", Wiley, New York, 1991. } \keyword{datasets} survival/man/frailty.Rd0000644000175100001440000001174612453556525014704 0ustar hornikusers\name{frailty} \alias{frailty} \alias{frailty.gamma} \alias{frailty.gaussian} \alias{frailty.t} \title{ Random effects terms } \description{ The frailty function allows one to add a simple random effects term to a Cox or survreg model. } \usage{ frailty(x, distribution="gamma", ...) frailty.gamma(x, sparse = (nclass > 5), theta, df, eps = 1e-05, method = c("em","aic", "df", "fixed"), ...) frailty.gaussian(x, sparse = (nclass > 5), theta, df, method =c("reml","aic", "df", "fixed"), ...) frailty.t(x, sparse = (nclass > 5), theta, df, eps = 1e-05, tdf = 5, method = c("aic", "df", "fixed"), ...) } \arguments{ \item{x}{ the variable to be entered as a random effect. It is always treated as a factor. } \item{distribution}{ either the \code{gamma}, \code{gaussian} or \code{t} distribution may be specified. The routines \code{frailty.gamma}, \code{frailty.gaussian} and \code{frailty.t} do the actual work. } \item{\dots}{Arguments for specific distribution, including (but not limited to) } \item{sparse}{ cutoff for using a sparse coding of the data matrix. If the total number of levels of \code{x} is larger than this value, then a sparse matrix approximation is used. The correct cutoff is still a matter of exploration: if the number of levels is very large (thousands) then the non-sparse calculation may not be feasable in terms of both memory and compute time. Likewise, the accuracy of the sparse approximation appears to be related to the maximum proportion of subjects in any one class, being best when no one class has a large membership. } \item{theta}{ if specified, this fixes the variance of the random effect. If not, the variance is a parameter, and a best solution is sought. Specifying this implies \code{method='fixed'}. } \item{df}{ if specified, this fixes the degrees of freedom for the random effect. Specifying this implies \code{method='df'}. Only one of \code{theta} or \code{df} should be specified. } \item{method}{ the method used to select a solution for theta, the variance of the random effect. The \code{fixed} corresponds to a user-specified value, and no iteration is done. The \code{df} selects the variance such that the degrees of freedom for the random effect matches a user specified value. The \code{aic} method seeks to maximize Akiake's information criteria 2*(partial likelihood - df). The \code{em} and \code{reml} methods are specific to Cox models with gamma and gaussian random effects, respectively. Please see further discussion below. } \item{tdf}{ the degrees of freedom for the t-distribution. } \item{eps}{ convergence critera for the iteration on theta. } } \value{ this function is used in the model statment of either \code{coxph} or \code{survreg}. It's results are used internally. } \details{ The \code{frailty} plugs into the general penalized modeling framework provided by the \code{coxph} and \code{survreg} routines. This framework deals with likelihood, penalties, and degrees of freedom; these aspects work well with either parent routine. Therneau, Grambsch, and Pankratz show how maximum likelihood estimation for the Cox model with a gamma frailty can be accomplished using a general penalized routine, and Ripatti and Palmgren work through a similar argument for the Cox model with a gaussian frailty. Both of these are specific to the Cox model. Use of gamma/ml or gaussian/reml with \code{survreg} does not lead to valid results. The extensible structure of the penalized methods is such that the penalty function, such as \code{frailty} or \code{pspine}, is completely separate from the modeling routine. The strength of this is that a user can plug in any penalization routine they choose. A weakness is that it is very difficult for the modeling routine to know whether a sensible penalty routine has been supplied. Note that use of a frailty term implies a mixed effects model and use of a cluster term implies a GEE approach; these cannot be mixed. The \code{coxme} package has superseded this method. It is faster, more stable, and more flexible. } \section{References}{ S Ripatti and J Palmgren, Estimation of multivariate frailty models using penalized partial likelihood, Biometrics, 56:1016-1022, 2000. T Therneau, P Grambsch and VS Pankratz, Penalized survival models and frailty, J Computational and Graphical Statistics, 12:156-175, 2003. } \seealso{ \link{coxph}, \link{survreg} } \examples{ # Random institutional effect coxph(Surv(time, status) ~ age + frailty(inst, df=4), lung) # Litter effects for the rats data rfit2a <- survreg(Surv(time, status) ~ rx + frailty.gaussian(litter, df=13, sparse=FALSE), rats, subset= (sex=='f')) rfit2b <- survreg(Surv(time, status) ~ rx + frailty.gaussian(litter, df=13, sparse=TRUE), rats, subset= (sex=='f')) } \keyword{survival} survival/man/print.summary.coxph.Rd0000644000175100001440000000103511732700061017151 0ustar hornikusers\name{print.summary.coxph} \alias{print.summary.coxph} \title{ Print method for summary.coxph objects } \description{ Produces a printed summary of a fitted coxph model } \usage{ \method{print}{summary.coxph}(x, digits=max(getOption("digits") - 3, 3), signif.stars = getOption("show.signif.stars"), ...) } \arguments{ \item{x}{ the result of a call to \code{summary.coxph} } \item{digits}{significant digits to print} \item{signif.stars}{ Show stars to highlight small p-values } \item{\dots}{For future methods} } survival/man/cluster.Rd0000644000175100001440000000222712453556617014707 0ustar hornikusers\name{cluster} \alias{cluster} \title{ Identify clusters. } \description{ This is a special function used in the context of survival models. It identifies correlated groups of observations, and is used on the right hand side of a formula. Using \code{cluster()} in a formula implies that robust sandwich variance estimators are desired.} \usage{ cluster(x) } \arguments{ \item{x}{ A character, factor, or numeric variable. } } \value{ \code{x} } \details{ The function's only action is semantic, to mark a variable as the cluster indicator. The resulting variance is what is known as the ``working independence'' variance in a GEE model. Note that one cannot use both a frailty term and a cluster term in the same model, the first is a mixed-effects approach to correlation and the second a GEE approach, and these don't mix. } \seealso{ \code{\link{coxph}}, \code{\link{survreg}} } \examples{ marginal.model <- coxph(Surv(time, status) ~ rx + cluster(litter), rats, subset=(sex=='f')) frailty.model <- coxph(Surv(time, status) ~ rx + frailty(litter), rats, subset=(sex=='f')) } \keyword{survival} survival/man/tobin.Rd0000644000175100001440000000134711732700061014322 0ustar hornikusers\name{tobin} \alias{tobin} \docType{data} \title{Tobin's Tobit data} \description{ Economists fit a parametric censored data model called the \sQuote{tobit}. These data are from Tobin's original paper. } \usage{tobin} \format{ A data frame with 20 observations on the following 3 variables. \describe{ \item{durable}{Durable goods purchase} \item{age}{Age in years} \item{quant}{Liquidity ratio (x 1000)} } } \source{ J Tobin (1958), Estimation of relationships for limited dependent variables. \emph{Econometrica} \bold{26}, 24--36. } \examples{ tfit <- survreg(Surv(durable, durable>0, type='left') ~age + quant, data=tobin, dist='gaussian') predict(tfit,type="response") } \keyword{datasets} survival/man/is.ratetable.Rd0000644000175100001440000000210311732700061015553 0ustar hornikusers\name{is.ratetable} \alias{is.ratetable} \alias{Math.ratetable} \alias{Ops.ratetable} \title{ Verify that an object is of class ratetable. } \description{ The function verifies not only the \code{class} attribute, but the structure of the object. } \usage{ is.ratetable(x, verbose=FALSE) } \arguments{ \item{x}{ the object to be verified. } \item{verbose}{ if \code{TRUE} and the object is not a ratetable, then return a character string describing the way(s) in which \code{x} fails to be a proper ratetable object. } } \value{ returns \code{TRUE} if \code{x} is a ratetable, and \code{FALSE} or a description if it is not. } \details{ Rate tables are used by the \code{pyears} and \code{survexp} functions, and normally contain death rates for some population, categorized by age, sex, or other variables. They have a fairly rigid structure, and the \code{verbose} option can help in creating a new rate table. } \seealso{ \code{\link{pyears}}, \code{\link{survexp}}. } \examples{ is.ratetable(survexp.us) # True is.ratetable(cancer) # False } \keyword{survival} survival/man/survexp.fit.Rd0000644000175100001440000000420611732700061015501 0ustar hornikusers\name{survexp.fit} \alias{survexp.fit} \title{ Compute Expected Survival } \description{ Compute expected survival times. } \usage{ survexp.fit(group, x, y, times, death, ratetable) } \arguments{ \item{group}{if there are multiple survival curves this identifies the group, otherwise it is a constant. Must be an integer.} \item{x}{A matrix whose columns match the dimensions of the \code{ratetable}, in the correct order. } \item{y}{ the follow up time for each subject. } \item{times}{ the vector of times at which a result will be computed. } \item{death}{ a logical value, if \code{TRUE} the conditional survival is computed, if \code{FALSE} the cohort survival is computed. See \code{\link{survexp}} for more details. } \item{ratetable}{ a rate table, such as \code{survexp.uswhite}. } } \value{ A list containing the number of subjects and the expected survival(s) at each time point. If there are multiple groups, these will be matrices with one column per group. } \details{ For conditional survival \code{y} must be the time of last follow-up or death for each subject. For cohort survival it must be the potential censoring time for each subject, ignoring death. For an exact estimate \code{times} should be a superset of \code{y}, so that each subject at risk is at risk for the entire sub-interval of time. For a large data set, however, this can use an inordinate amount of storage and/or compute time. If the \code{times} spacing is more coarse than this, an actuarial approximation is used which should, however, be extremely accurate as long as all of the returned values are > .99. For a subgroup of size 1 and \code{times} > \code{y}, the conditional method reduces to exp(-h) where h is the expected cumulative hazard for the subject over his/her observation time. This is used to compute individual expected survival. } \section{Warning}{ Most users will call the higher level routine \code{survexp}. Consequently, this function has very few error checks on its input arguments. } \seealso{ \code{\link{survexp}}, \code{\link{survexp.us}}. } \keyword{survival } % docclass is function % Converted by Sd2Rd version 37351. survival/man/ovarian.Rd0000644000175100001440000000156112466146533014661 0ustar hornikusers\name{ovarian} \alias{ovarian} \docType{data} \title{Ovarian Cancer Survival Data} \usage{ovarian} \description{Survival in a randomised trial comparing two treatments for ovarian cancer} \format{ \tabular{ll}{ futime:\tab survival or censoring time\cr fustat:\tab censoring status\cr age: \tab in years\cr resid.ds:\tab residual disease present (1=no,2=yes)\cr rx:\tab treatment group\cr ecog.ps:\tab ECOG performance status (1 is better, see reference)\cr } } \source{Terry Therneau} \references{ Edmunson, J.H., Fleming, T.R., Decker, D.G., Malkasian, G.D., Jefferies, J.A., Webb, M.J., and Kvols, L.K., Different Chemotherapeutic Sensitivities and Host Factors Affecting Prognosis in Advanced Ovarian Carcinoma vs. Minimal Residual Disease. Cancer Treatment Reports, 63:241-47, 1979. } \keyword{datasets} \keyword{survival} survival/man/anova.coxph.Rd0000644000175100001440000000411211732700061015424 0ustar hornikusers\name{anova.coxph} \alias{anova.coxph} \alias{anova.coxphlist} \title{Analysis of Deviance for a Cox model.} \usage{ \method{anova}{coxph}(object, \dots, test = 'Chisq') } \description{ Compute an analysis of deviance table for one or more Cox model fits. } \arguments{ \item{object}{An object of class \code{coxph}} \item{\dots}{Further \code{coxph} objects} \item{test}{a character string. The appropriate test is a chisquare, all other choices result in no test being done.} } \details{ Specifying a single object gives a sequential analysis of deviance table for that fit. That is, the reductions in the model log-likelihood as each term of the formula is added in turn are given in as the rows of a table, plus the log-likelihoods themselves. A robust variance estimate is normally used in situations where the model may be mis-specified, e.g., multiple events per subject. In this case a comparison of partial-likelihood values does not make sense, and \code{anova} will refuse to print results. If more than one object is specified, the table has a row for the degrees of freedom and loglikelihood for each model. For all but the first model, the change in degrees of freedom and loglik is also given. (This only make statistical sense if the models are nested.) It is conventional to list the models from smallest to largest, but this is up to the user. The table will optionally contain test statistics (and P values) comparing the reduction in loglik for each row. } \value{ An object of class \code{"anova"} inheriting from class \code{"data.frame"}. } \section{Warning}{ The comparison between two or more models by \code{anova} or will only be valid if they are fitted to the same dataset. This may be a problem if there are missing values.} \seealso{ \code{\link{coxph}}, \code{\link{anova}}. } \examples{ fit <- coxph(Surv(futime, fustat) ~ resid.ds *rx + ecog.ps, data = ovarian) anova(fit) fit2 <- coxph(Surv(futime, fustat) ~ resid.ds +rx + ecog.ps, data=ovarian) anova(fit2,fit) } \keyword{models} \keyword{regression} \keyword{survival} survival/man/summary.survfit.Rd0000644000175100001440000000734312357770065016426 0ustar hornikusers\name{summary.survfit} \alias{summary.survfit} \title{ Summary of a Survival Curve } \description{ Returns a list containing the survival curve, confidence limits for the curve, and other information. } \usage{ \method{summary}{survfit}(object, times=, censored=FALSE, scale=1, extend=FALSE, rmean=getOption('survfit.rmean'), ...) } \arguments{ \item{object}{ the result of a call to the \code{survfit} function. } \item{times}{ vector of times; the returned matrix will contain 1 row for each time. The vector will be sorted into increasing order; missing values are not allowed. If \code{censored=T}, the default \code{times} vector contains all the unique times in \code{fit}, otherwise the default \code{times} vector uses only the event (death) times. } \item{censored}{ logical value: should the censoring times be included in the output? This is ignored if the \code{times} argument is present. } \item{scale}{ numeric value to rescale the survival time, e.g., if the input data to \code{survfit} were in days, \code{scale = 365.25} would scale the output to years. } \item{extend}{ logical value: if TRUE, prints information for all specified \code{times}, even if there are no subjects left at the end of the specified \code{times}. This is only valid if the \code{times} argument is present. } \item{rmean}{Show restricted mean: see \code{\link{print.survfit}} for details} \item{\dots}{for future methods} } \value{ a list with the following components: \item{surv}{ the estimate of survival at time t+0. } \item{time}{ the timepoints on the curve. } \item{n.risk}{ the number of subjects at risk at time t-0 (but see the comments on weights in the \code{survfit} help file). } \item{n.event}{ if the \code{times} argument is missing, then this column is the number of events that occurred at time t. Otherwise, it is the cumulative number of events that have occurred since the last time listed until time t+0. } \item{n.entered}{ This is present only for counting process survival data. If the \code{times} argument is missing, this column is the number of subjects that entered at time t. Otherwise, it is the cumulative number of subjects that have entered since the last time listed until time t. } \item{n.exit.censored}{ if the \code{times} argument is missing, this column is the number of subjects that left without an event at time t. Otherwise, it is the cumulative number of subjects that have left without an event since the last time listed until time t+0. This is only present for counting process survival data. } \item{std.err}{ the standard error of the survival value. } \item{conf.int}{ level of confidence for the confidence intervals of survival. } \item{lower}{ lower confidence limits for the curve. } \item{upper}{ upper confidence limits for the curve. } \item{strata}{ indicates stratification of curve estimation. If \code{strata} is not \code{NULL}, there are multiple curves in the result and the \code{surv}, \code{time}, \code{n.risk}, etc. vectors will contain multiple curves, pasted end to end. The levels of \code{strata} (a factor) are the labels for the curves. } \item{call}{ the statement used to create the \code{fit} object. } \item{na.action}{ same as for \code{fit}, if present. } \item{table}{ table of information that is returned from \code{print.survfit} function. } \item{type}{ type of data censoring. Passed through from the fit object. } } \seealso{ \code{\link{survfit}}, \code{\link{print.summary.survfit}} } \examples{ summary( survfit( Surv(futime, fustat)~1, data=ovarian)) summary( survfit( Surv(futime, fustat)~rx, data=ovarian)) } \keyword{survival} survival/man/quantile.survfit.Rd0000644000175100001440000000677612112666037016554 0ustar hornikusers\name{quantile.survfit} \alias{quantile.survfit} \alias{quantile.survfitms} \title{Quantiles from a survfit object} \description{Retrieve quantiles and confidence intervals for them from a survfit object. } \usage{ \method{quantile}{survfit}(x, probs = c(0.25, 0.5, 0.75), conf.int = TRUE, tolerance= sqrt(.Machine$double.eps), ...) \method{quantile}{survfitms}(x, probs = c(0.25, 0.5, 0.75), conf.int = TRUE, tolerance= sqrt(.Machine$double.eps), ...) } \arguments{ \item{x}{a result of the survfit function} \item{probs}{numeric vector of probabilities with values in [0,1]} \item{conf.int}{should lower and upper confidence limits be returned?} \item{tolerance}{tolerance for checking that the suvival curve exactly equals one of the quantiles} \item{...}{optional arguments for other methods} } \details{ The kth quantile for a survival curve S(t) is the location at which a horizontal line at height p= 1-k intersects the plot of S(t). Since S(t) is a step function, it is possible for the curve to have a horizontal segment at exactly 1-k, in which case the midpoint of the horizontal segment is returned. This mirrors the standard behavior of the median when data is uncensored. If the survival curve does not fall to 1-k, then that quantile is undefined. In order to be consistent with other quantile functions, the argument \code{prob} of this function applies to the cumulative distribution function F(t) = 1-S(t). Confidence limits for the values are based on the intersection of the horizontal line at 1-k with the upper and lower limits for the survival curve. Hence confidence limits use the same p-value as was in effect when the curve was created, and will differ depending on the \code{conf.type} option of \code{survfit}. If the survival curves have no confidence bands, confidence limits for the quantiles are not available. When a horizontal seqment of the survival curve exactly matches one of the requested quantiles the returned value will be the midpoint of the horizontal seqment; this agrees with the usual definition of a median for uncensored data. Since the survival curve is computed as a series of products, however, there may be round off error. Assume for instance a sample of size 20 with no tied times and no censoring. The survival curve after the 10th death is (19/20)(18/19)(17/18) ... (10/11) = 10/20, but the computed result will not be exactly 0.5. Any horizontal segment whose absolute difference with a requested percentile is less than \code{tolerance} is considered to be an exact match. } \value{ The quantiles will be a vector if the \code{survfit} object contains only a single curve, otherwise it will be a matrix or array. In this case the last dimension will index the quantiles. If confidence limits are requested, then result will be a list with components \code{quantile}, \code{lower}, and \code{upper}, otherwise it is the vector or matrix of quantiles. } \author{Terry Therneau} \seealso{\code{\link{survfit}}, \code{\link{print.survfit}}, \code{\link{qsurvreg}} } \examples{ fit <- survfit(Surv(time, status) ~ ph.ecog, data=lung) quantile(fit) cfit <- coxph(Surv(time, status) ~ age + strata(ph.ecog), data=lung) csurv<- survfit(cfit, newdata=data.frame(age=c(40, 60, 80)), conf.type ="none") temp <- quantile(csurv, 1:5/10) temp[2,3,] # quantiles for second level of ph.ecog, age=80 quantile(csurv[2,3], 1:5/10) # quantiles of a single curve, same result } \keyword{ survival } survival/man/transplant.Rd0000644000175100001440000000467512504015531015403 0ustar hornikusers\name{transplant} \alias{transplant} \docType{data} \title{Liver transplant waiting list} \description{ Subjects on a liver transplant waiting list from 1990-1999, and their disposition: received a transplant, died while waiting, withdrew from the list, or censored. } \usage{data("transplant")} \format{ A data frame with 815 observations on the following 6 variables. \describe{ \item{\code{age}}{age at addition to the waiting list} \item{\code{sex}}{\code{m} or \code{f}} \item{\code{abo}}{blood type: \code{A}, \code{B}, \code{AB} or \code{O}} \item{\code{year}}{year in which they entered the waiting list} \item{\code{futime}}{time from entry to final disposition} \item{\code{event}}{final disposition: \code{censored}, \code{death}, \code{ltx} or \code{withdraw}} } } \details{ This represents the transplant experience in a particular region, over a time period in which liver transplant became much more widely recognized as a viable treatment modality. The number of liver transplants rises over the period, but the number of subjects added to the liver transplant waiting list grew much faster. Important questions addressed by the data are the change in waiting time, who waits, and whether there was an consequent increase in deaths while on the list. Blood type is an important consideration. Donor livers from subjects with blood type O can be used by patients with A, B, AB or 0 blood types, whereas a recipient of type B cannot accept an A or AB liver for instance. Thus type O subjects on the waiting list are at a disadvantage, since the pool of competitors is larger for type O donor livers. This data is of historical interest but has little relevance to current practice. Liver allocation policies have evolved and now depend directly on individual patient's risk and need, assessments of which are regularly updated while a patient is on the waiting list. The overall organ shortage remains acute, however. } \examples{ period <- cut(transplant$year, c(1989, 1992, 1995, 1997, 2000), labels=c('90-92', '93-95', '96-97', '98-99')) pfit <- survfit(Surv(futime, event) ~ period, transplant) pfit[,2] #time to liver transplant plot(pfit[,2], mark.time=FALSE, col=1:4, lwd=2, xmax=735, xscale=30.5, xlab="Months", ylab="Fraction transplanted", xaxt = 'n') temp <- c(0, 6, 12, 18, 24) axis(1, temp, temp) legend(15, .35, levels(period), lty=1, col=1:4, lwd=2, bty='n') } \keyword{datasets} survival/man/flchain.Rd0000644000175100001440000000715112260346575014627 0ustar hornikusers\name{flchain} \alias{flchain} \docType{data} \title{Assay of serum free light chain for 7874 subjects.} \description{ This is a stratified random sample containing 1/2 of the subjects from a study of the relationship between serum free light chain (FLC) and mortality. The original sample contains samples on approximately 2/3 of the residents of Olmsted County aged 50 or greater. } \usage{data(flchain)} \format{ A data frame with 7874 persons containing the following variables. \describe{ \item{\code{age }}{age in years} \item{\code{sex}}{F=female, M=male} \item{\code{sample.yr}}{the calendar year in which a blood sample was obtained} \item{\code{kappa}}{serum free light chain, kappa portion} \item{\code{lambda}}{serum free light chain, lambda portion} \item{\code{flc.grp}}{the FLC group for the subject, as used in the original analysis} \item{\code{creatinine}}{serum creatinine} \item{\code{mgus}}{1 if the subject had been diagnosed with monoclonal gammapothy (MGUS)} \item{\code{futime}}{days from enrollment until death. Note that there are 3 subjects whose sample was obtained on their death date.} \item{\code{death}}{0=alive at last contact date, 1=dead} \item{\code{chapter}}{for those who died, a grouping of their primary cause of death by chapter headings of the International Code of Diseases ICD-9} } } \details{In 1995 Dr. Robert Kyle embarked on a study to determine the prevalence of monoclonal gammopathy of undetermined significance (MGUS) in Olmsted County, Minnesota, a condition which is normally only found by chance from a test (serum electrophoresis) which is ordered for other causes. Later work suggested that one component of immunoglobulin production, the serum free light chain, might be a possible marker for immune disregulation. In 2010 Dr. Angela Dispenzieri and colleagues assayed FLC levels on those samples from the original study for which they had patient permission and from which sufficient material remained for further testing. They found that elevated FLC levels were indeed associated with higher death rates. Patients were recruited when they came to the clinic for other appointments, with a final random sample of those who had not yet had a visit since the study began. An interesting side question is whether there are differences between early, mid, and late recruits. This data set contains an age and sex stratified random sample that includes 7874 of the original 15759 subjects. The original subject identifiers and dates have been removed to protect patient identity. Subsampling was done to further protect this information. } \source{The primary investigator (A Dispenzieri) and statistician (T Therneau) for the study.} \references{ A Dispenzieri, J Katzmann, R Kyle, D Larson, T Therneau, C Colby, R Clark, G Mead, S Kumar, LJ Melton III and SV Rajkumar (2012). Use of monclonal serum immunoglobulin free light chains to predict overall survival in the general population, Mayo Clinic Proceedings 87:512-523. R Kyle, T Therneau, SV Rajkumar, D Larson, M Plevak, J Offord, A Dispenzieri, J Katzmann, and LJ Melton, III, 2006, Prevalence of monoclonal gammopathy of undetermined significance, New England J Medicine 354:1362-1369. } \examples{ data(flchain) age.grp <- cut(flchain$age, c(49,54, 59,64, 69,74,79, 89, 110), labels= paste(c(50,55,60,65,70,75,80,90), c(54,59,64,69,74,79,89,109), sep='-')) table(flchain$sex, age.grp) } \keyword{datasets} survival/man/survival-internal.Rd0000644000175100001440000000462612234466366016716 0ustar hornikusers\name{survival-internal} \alias{survival-internal} \alias{agexact.fit} \alias{as.matrix.ratetable} \alias{coxpenal.df} \alias{coxpenal.fit} \alias{is.category} \alias{is.na.ratetable2} \alias{is.na.coxph.penalty} \alias{match.ratetable} \alias{survfitCI} \alias{survfitKM} \alias{survreg.fit} \alias{survpenal.fit} \alias{survdiff.fit} \title{Internal survival functions} \description{Internal survival functions} \usage{ survreg.fit(x, y, weights, offset, init, controlvals, dist, scale = 0, nstrat = 1, strata, parms = NULL,assign) survpenal.fit(x, y, weights, offset, init, controlvals, dist, scale = 0, nstrat = 1, strata, pcols, pattr, assign, parms = NULL) survdiff.fit(y, x, strat, rho = 0) is.category(x) match.ratetable(R, ratetable) \method{as.matrix}{ratetable}(x, ...) \method{is.na}{ratetable2}(x) \method{is.na}{coxph.penalty}(x) coxpenal.df(hmat, hinv, fdiag, assign.list, ptype, nvar, pen1, pen2, sparse) coxpenal.fit(x, y, strata, offset, init, control, weights, method, rownames, pcols, pattr, assign) coxph.wtest(var, b, toler.chol = 1e-09) agexact.fit(x, y, strata, offset, init, control, weights, method, rownames) survfitCI(X, Y, weights, id, istate, type=c('kaplan-meier', 'fleming-harrington', 'fh2'), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified')) survfitKM(x, y, casewt=rep(1,length(x)), type=c('kaplan-meier', 'fleming-harrington', 'fh2'), error=c('greenwood', "tsiatis"), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified'), start.time, new.time) survfitTurnbull(x, y, casewt, type=c('kaplan-meier', 'fleming-harrington', 'fh2'), error=c('greenwood', "tsiatis"), se.fit=TRUE, conf.int= .95, conf.type=c('log', 'log-log', 'plain', 'none'), conf.lower=c('usual', 'peto', 'modified'), start.time) } \details{The arguments to these routines are not guarranteed to stay the same from release to release -- call them at your own risk!} \keyword{survival} \keyword{internal} survival/man/survfit.formula.Rd0000644000175100001440000002404312461741246016365 0ustar hornikusers\name{survfit.formula} \alias{survfit.formula} \alias{[.survfit} \title{ Compute a Survival Curve for Censored Data } \description{ Computes an estimate of a survival curve for censored data using either the Kaplan-Meier or the Fleming-Harrington method. For competing risks data it computes the cumulative incidence curve. } \usage{ \method{survfit}{formula}(formula, data, weights, subset, na.action, etype, id, istate, ...) } \arguments{ \item{formula}{ a formula object, which must have a \code{Surv} object as the response on the left of the \code{~} operator and, if desired, terms separated by + operators on the right. One of the terms may be a \code{strata} object. For a single survival curve the right hand side should be \code{~ 1}. } \item{data}{ a data frame in which to interpret the variables named in the formula, \code{subset} and \code{weights} arguments. } \item{weights}{ The weights must be nonnegative and it is strongly recommended that they be strictly positive, since zero weights are ambiguous, compared to use of the \code{subset} argument. } \item{subset}{ expression saying that only a subset of the rows of the data should be used in the fit. } \item{na.action}{ a missing-data filter function, applied to the model frame, after any \code{subset} argument has been used. Default is \code{options()$na.action}. } \item{etype}{ a variable giving the type of event. This has been superseded by multi-state Surv objects; see example below. } \item{id}{ identifies individual subjects, when a given person can have multiple lines of data. } \item{istate}{for multi-state models, identifies the initial state of each subject} \item{\dots}{ The following additional arguments are passed to internal functions called by \code{survfit}. \describe{ \item{type}{ a character string specifying the type of survival curve. Possible values are \code{"kaplan-meier"}, \code{"fleming-harrington"} or \code{"fh2"} if a formula is given. This is ignored for competing risks or when the Turnbull estimator is used. } \item{error}{ a character string specifying the error. Possible values are \code{"greenwood"} for the Greenwood formula or \code{"tsiatis"} or \code{"aalen"} for the Tsiatis/Aalen formula, or \code{"robust"} for a robust variance. The last of these is assumed if non-integer case weights are provided. } \item{conf.type}{ One of \code{"none"}, \code{"plain"}, \code{"log"} (the default), or \code{"log-log"}. Only enough of the string to uniquely identify it is necessary. The first option causes confidence intervals not to be generated. The second causes the standard intervals \code{curve +- k *se(curve)}, where k is determined from \code{conf.int}. The log option calculates intervals based on the cumulative hazard or log(survival). The last option bases intervals on the log hazard or log(-log(survival)). } \item{conf.lower}{ a character string to specify modified lower limits to the curve, the upper limit remains unchanged. Possible values are \code{"usual"} (unmodified), \code{"peto"}, and \code{"modified"}. T he modified lower limit is based on an "effective n" argument. The confidence bands will agree with the usual calculation at each death time, but unlike the usual bands the confidence interval becomes wider at each censored observation. The extra width is obtained by multiplying the usual variance by a factor m/n, where n is the number currently at risk and m is the number at risk at the last death time. (The bands thus agree with the un-modified bands at each death time.) This is especially useful for survival curves with a long flat tail. The Peto lower limit is based on the same "effective n" argument as the modified limit, but also replaces the usual Greenwood variance term with a simple approximation. It is known to be conservative. } \item{start.time}{ numeric value specifying a time to start calculating survival information. The resulting curve is the survival conditional on surviving to \code{start.time}. } \item{conf.int}{ the level for a two-sided confidence interval on the survival curve(s). Default is 0.95. } \item{se.fit}{ a logical value indicating whether standard errors should be computed. Default is \code{TRUE}. } } } } \value{ an object of class \code{"survfit"}. See \code{survfit.object} for details. Methods defined for survfit objects are \code{print}, \code{plot}, \code{lines}, and \code{points}. } \details{ The estimates used are the Kalbfleisch-Prentice (Kalbfleisch and Prentice, 1980, p.86) and the Tsiatis/Link/Breslow, which reduce to the Kaplan-Meier and Fleming-Harrington estimates, respectively, when the weights are unity. The Greenwood formula for the variance is a sum of terms d/(n*(n-m)), where d is the number of deaths at a given time point, n is the sum of weights for all individuals still at risk at that time, and m is the sum of weights for the deaths at that time. The justification is based on a binomial argument when weights are all equal to one; extension to the weighted case is ad hoc. Tsiatis (1981) proposes a sum of terms d/(n*n), based on a counting process argument which includes the weighted case. The two variants of the F-H estimate have to do with how ties are handled. If there were 3 deaths out of 10 at risk, then the first increments the hazard by 3/10 and the second by 1/10 + 1/9 + 1/8. For the first method S(t) = exp(H), where H is the Nelson-Aalen cumulative hazard estimate, whereas the \code{fh2} method will give results S(t) results closer to the Kaplan-Meier. When the data set includes left censored or interval censored data (or both), then the EM approach of Turnbull is used to compute the overall curve. When the baseline method is the Kaplan-Meier, this is known to converge to the maximum likelihood estimate. The cumulative incidence curve is an alternative to the Kaplan-Meier for competing risks data. For instance, in patients with MGUS, conversion to an overt plasma cell malignancy occurs at a nearly constant rate among those still alive. A Kaplan-Meier estimate, treating death due to other causes as censored, gives a 20 year cumulate rate of 33\% for the 241 early patients of Kyle. This estimates the incidence of conversion if all other causes of death were removed, which is an unrealistic assumption given the mean starting age of 63 and a median follow up of over 21 years. The CI estimate, on the other hand, estimates the total number of conversions that will actually occur. Because the population is older, this is much smaller than the KM, 22\% at 20 years for Kyle's data. If there were no censoring, then CI(t) could very simply be computed as total number of patients with progression by time t divided by the sample size n. } \section{References}{ Dorey, F. J. and Korn, E. L. (1987). Effective sample sizes for confidence intervals for survival probabilities. \emph{Statistics in Medicine} \bold{6}, 679-87. Fleming, T. H. and Harrington, D. P. (1984). Nonparametric estimation of the survival distribution in censored data. \emph{Comm. in Statistics} \bold{13}, 2469-86. Kablfleisch, J. D. and Prentice, R. L. (1980). \emph{The Statistical Analysis of Failure Time Data.} New York:Wiley. Kyle, R. A. (1997). Moncolonal gammopathy of undetermined significance and solitary plasmacytoma. Implications for progression to overt multiple myeloma\}, \emph{Hematology/Oncology Clinics N. Amer.} \bold{11}, 71-87. Link, C. L. (1984). Confidence intervals for the survival function using Cox's proportional hazards model with covariates. \emph{Biometrics} \bold{40}, 601-610. Turnbull, B. W. (1974). Nonparametric estimation of a survivorship function with doubly censored data. \emph{J Am Stat Assoc}, \bold{69}, 169-173. } \seealso{ \code{\link{survfit.coxph}} for survival curves from Cox models, \code{\link{survfit.object}} for a description of the components of a survfit object, \code{\link{print.survfit}}, \code{\link{plot.survfit}}, \code{\link{lines.survfit}}, \code{\link{coxph}}, \code{\link{Surv}}. } \examples{ #fit a Kaplan-Meier and plot it fit <- survfit(Surv(time, status) ~ x, data = aml) plot(fit, lty = 2:3) legend(100, .8, c("Maintained", "Nonmaintained"), lty = 2:3) #fit a Cox proportional hazards model and plot the #predicted survival for a 60 year old fit <- coxph(Surv(futime, fustat) ~ age, data = ovarian) plot(survfit(fit, newdata=data.frame(age=60)), xscale=365.25, xlab = "Years", ylab="Survival") # Here is the data set from Turnbull # There are no interval censored subjects, only left-censored (status=3), # right-censored (status 0) and observed events (status 1) # # Time # 1 2 3 4 # Type of observation # death 12 6 2 3 # losses 3 2 0 3 # late entry 2 4 2 5 # tdata <- data.frame(time =c(1,1,1,2,2,2,3,3,3,4,4,4), status=rep(c(1,0,2),4), n =c(12,3,2,6,2,4,2,0,2,3,3,5)) fit <- survfit(Surv(time, time, status, type='interval') ~1, data=tdata, weight=n) # # Time to progression/death for patients with monoclonal gammopathy # Competing risk curves (cumulative incidence) fitKM <- survfit(Surv(stop, event=='progression') ~1, data=mgus1, subset=(start==0)) fitCI <- survfit(Surv(stop, status*as.numeric(event), type="mstate") ~1, data=mgus1, subset=(start==0)) # CI curves are always plotted from 0 upwards, rather than 1 down plot(fitCI, xscale=365.25, xmax=7300, mark.time=FALSE, col=2:3, xlab="Years post diagnosis of MGUS") lines(fitKM, fun='event', xscale=365.25, xmax=7300, mark.time=FALSE, conf.int=FALSE) text(10, .4, "Competing risk: death", col=3) text(16, .15,"Competing risk: progression", col=2) text(15, .30,"KM:prog") } \keyword{survival} survival/man/ridge.Rd0000644000175100001440000000303511732700061014275 0ustar hornikusers\name{ridge} \alias{ridge} \title{ Ridge regression} \usage{ ridge(..., theta, df=nvar/2, eps=0.1, scale=TRUE) } \arguments{ \item{\dots}{predictors to be ridged } \item{theta}{penalty is \code{theta}/2 time sum of squared coefficients } \item{df}{Approximate degrees of freedom } \item{eps}{ Accuracy required for \code{df} } \item{scale}{ Scale variables before applying penalty? } } \description{ When used in a \link{coxph} or \link{survreg} model formula, specifies a ridge regression term. The likelihood is penalised by \code{theta}/2 time the sum of squared coefficients. If \code{scale=T} the penalty is calculated for coefficients based on rescaling the predictors to have unit variance. If \code{df} is specified then \code{theta} is chosen based on an approximate degrees of freedom. } \value{ An object of class \code{coxph.penalty} containing the data and control functions. } \references{ Gray (1992) "Flexible methods of analysing survival data using splines, with applications to breast cancer prognosis" JASA 87:942--951 } \seealso{ \code{\link{coxph}},\code{\link{survreg}},\code{\link{pspline}},\code{\link{frailty}} } \examples{ coxph(Surv(futime, fustat) ~ rx + ridge(age, ecog.ps, theta=1), ovarian) lfit0 <- survreg(Surv(time, status) ~1, cancer) lfit1 <- survreg(Surv(time, status) ~ age + ridge(ph.ecog, theta=5), cancer) lfit2 <- survreg(Surv(time, status) ~ sex + ridge(age, ph.ecog, theta=1), cancer) lfit3 <- survreg(Surv(time, status) ~ sex + age + ph.ecog, cancer) } \keyword{survival }%-- one or more ... survival/man/survregDtest.Rd0000644000175100001440000000301211732700061015677 0ustar hornikusers\name{survregDtest} \alias{survregDtest} \title{Verify a survreg distribution} \description{ This routine is called by \code{survreg} to verify that a distribution object is valid. } \usage{ survregDtest(dlist, verbose = F) } %- maybe also 'usage' for other objects documented here. \arguments{ \item{dlist}{the list describing a survival distribution} \item{verbose}{return a simple TRUE/FALSE from the test for validity (the default), or a verbose description of any flaws.} } \details{ If the \code{survreg} function rejects your user-supplied distribution as invalid, this routine will tell you why it did so. } \value{ TRUE if the distribution object passes the tests, and either FALSE or a vector of character strings if not. } \author{Terry Therneau} \seealso{\code{\link{survreg.distributions}}, \code{\link{survreg}}} \examples{ # An invalid distribution (it should have "init =" on line 2) # surveg would give an error message mycauchy <- list(name='Cauchy', init<- function(x, weights, ...) c(median(x), mad(x)), density= function(x, parms) { temp <- 1/(1 + x^2) cbind(.5 + atan(temp)/pi, .5+ atan(-temp)/pi, temp/pi, -2 *x*temp, 2*temp^2*(4*x^2*temp -1)) }, quantile= function(p, parms) tan((p-.5)*pi), deviance= function(...) stop('deviance residuals not defined') ) survregDtest(mycauchy, TRUE) } \keyword{survival} survival/man/predict.survreg.Rd0000644000175100001440000000547712067401404016346 0ustar hornikusers\name{predict.survreg} \alias{predict.survreg} \alias{predict.survreg.penal} \title{ Predicted Values for a `survreg' Object } \description{ Predicted values for a \code{survreg} object } \usage{ \method{predict}{survreg}(object, newdata, type=c("response", "link", "lp", "linear", "terms", "quantile", "uquantile"), se.fit=FALSE, terms=NULL, p=c(0.1, 0.9), na.action=na.pass, ...) } \arguments{ \item{object}{ result of a model fit using the \code{survreg} function. } \item{newdata}{ data for prediction. If absent predictions are for the subjects used in the original fit. } \item{type}{ the type of predicted value. This can be on the original scale of the data (response), the linear predictor (\code{"linear"}, with \code{"lp"} as an allowed abbreviation), a predicted quantile on the original scale of the data (\code{"quantile"}), a quantile on the linear predictor scale (\code{"uquantile"}), or the matrix of terms for the linear predictor (\code{"terms"}). At this time \code{"link"} and linear predictor (\code{"lp"}) are identical. } \item{se.fit}{ if \code{TRUE}, include the standard errors of the prediction in the result. } \item{terms}{ subset of terms. The default for residual type \code{"terms"} is a matrix with one column for every term (excluding the intercept) in the model. } \item{p}{ vector of percentiles. This is used only for quantile predictions. } \item{na.action}{ applies only when the \code{newdata} argument is present, and defines the missing value action for the new data. The default is to include all observations.} \item{\dots}{for future methods} } \value{ a vector or matrix of predicted values. } \references{ Escobar and Meeker (1992). Assessing influence in regression analysis with censored data. \emph{Biometrics,} 48, 507-528. } \seealso{ \code{\link{survreg}}, \code{\link{residuals.survreg}} } \examples{ # Draw figure 1 from Escobar and Meeker, 1992. fit <- survreg(Surv(time,status) ~ age + I(age^2), data=stanford2, dist='lognormal') with(stanford2, plot(age, time, xlab='Age', ylab='Days', xlim=c(0,65), ylim=c(.1, 10^5), log='y', type='n')) with(stanford2, points(age, time, pch=c(2,4)[status+1], cex=.7)) pred <- predict(fit, newdata=list(age=1:65), type='quantile', p=c(.1, .5, .9)) matlines(1:65, pred, lty=c(2,1,2), col=1) # Predicted Weibull survival curve for a lung cancer subject with # ECOG score of 2 lfit <- survreg(Surv(time, status) ~ ph.ecog, data=lung) pct <- 1:98/100 # The 100th percentile of predicted survival is at +infinity ptime <- predict(lfit, newdata=data.frame(ph.ecog=2), type='quantile', p=pct, se=TRUE) matplot(cbind(ptime$fit, ptime$fit + 2*ptime$se.fit, ptime$fit - 2*ptime$se.fit)/30.5, 1-pct, xlab="Months", ylab="Survival", type='l', lty=c(1,2,2), col=1) } \keyword{survival} survival/man/survConcordance.Rd0000644000175100001440000000704712257041540016353 0ustar hornikusers\name{survConcordance} \alias{survConcordance} \alias{survConcordance.fit} \title{ Compute a concordance measure. } \description{ This function computes the concordance between a right-censored survival time and a single continuous covariate } \usage{ survConcordance(formula, data, weights, subset, na.action) survConcordance.fit(y, x, strata, weight) } \arguments{ \item{formula}{ a formula with a survival time on the left and a single covariate on the right, along with an optional \code{strata()} term. The left hand term can also be a numeric vector. } \item{data}{ a data frame } \item{weights,subset,na.action}{as for \code{coxph}} \item{x, y, strata, weight}{predictor, response, strata, and weight vectors for the direct call} } \value{ an object containing the concordance, followed by the number of pairs that agree, disagree, are tied, and are not comparable. } \details{ The \code{survConcordance.fit} function computes the result but does no data checking whatsoever. It is intended as a hook for other packages that wish to compute concordance, and the data has already been assembled and verified. Concordance is defined as Pr(agreement) for any two randomly chosen observations, where in this case agreement means that the observation with the shorter survival time of the two also has the larger risk score. The predictor (or risk score) will often be the result of a Cox model or other regression. For continuous covariates concordance is equivalent to Kendall's tau, and for logistic regression is is equivalent to the area under the ROC curve. A value of 1 signifies perfect agreement, .6-.7 is a common result for survival data, .5 is an agreement that is no better than chance, and .3-.4 is the performace of some stock market analysts. The computation involves all n(n-1)/2 pairs of data points in the sample. For survival data, however, some of the pairs are incomparable. For instance a pair of times (5+, 8), the first being a censored value. We do not know whether the first survival time is greater than or less than the second. Among observations that are comparable, pairs may also be tied on survival time (but only if both are uncensored) or on the predictor. The final concondance is (agree + tied/2)/(agree + disagree + tied). There is, unfortunately, one aspect of the formula above that is unclear. Should the count of ties include observations that are tied on survival time y, tied on the predictor x, or both? By default the concordance only counts ties in x, treating tied survival times as incomparable; this agrees with the AUC calculation used in logistic regression. The Goodman-Kruskal Gamma statistic is (agree-disagree)/(agree + disagree), ignoring ties. It ranges from -1 to +1 similar to a correlation coefficient. Kendall's tau uses ties of both types. All of the components are returned in the result, however, so people can compute other combinations if interested. (If two observations have the same survival and the same x, they are counted in the tied survival time category). The algorithm is based on a balanced binary tree, which allows the computation to be done in O(n log n) time. } \seealso{ summary.coxph } \examples{ survConcordance(Surv(time, status) ~age, data=lung) options(na.action=na.exclude) fit <- coxph(Surv(time, status) ~ ph.ecog + age + sex, lung) survConcordance(Surv(time, status) ~predict(fit), lung) \dontrun{ n=227 (1 observations deleted due to missing values) Concordance= 0.6371102 , Gamma= 0.2759638 concordant discordant tied risk tied time 12544 7117 126 28 }} \keyword{survival} survival/man/model.frame.coxph.Rd0000644000175100001440000000140611732700061016514 0ustar hornikusers\name{model.frame.coxph} \Rdversion{1.1} \alias{model.frame.coxph} \title{Model.frame method for coxph objects} \description{ Recreate the model frame of a coxph fit. } \usage{ \method{model.frame}{coxph}(formula, ...) } \arguments{ \item{formula}{the result of a \code{coxph} fit} \item{\dots}{other arguments to \code{model.frame}} } \details{ For details, see the manual page for the generic function. This function would rarely be called by a user, it is mostly used inside functions like \code{residual} that need to recreate the data set from a model in order to do further calculations. } \value{the model frame used in the original fit, or a parallel one for new data. } \author{Terry Therneau} \seealso{\code{\link{model.frame}}} \keyword{ survival } survival/man/coxph.detail.Rd0000644000175100001440000000537111732700061015572 0ustar hornikusers\name{coxph.detail} \alias{coxph.detail} \title{ Details of a Cox Model Fit } \description{ Returns the individual contributions to the first and second derivative matrix, at each unique event time. } \usage{ coxph.detail(object, riskmat=FALSE) } \arguments{ \item{object}{ a Cox model object, i.e., the result of \code{coxph}. } \item{riskmat}{ include the at-risk indicator matrix in the output? } } \value{ a list with components \item{time}{ the vector of unique event times } \item{nevent}{ the number of events at each of these time points. } \item{means}{ a matrix with one row for each event time and one column for each variable in the Cox model, containing the weighted mean of the variable at that time, over all subjects still at risk at that time. The weights are the risk weights \code{exp(x \%*\% fit$coef)}. } \item{nrisk}{ number of subjects at risk. } \item{score}{ the contribution to the score vector (first derivative of the log partial likelihood) at each time point. } \item{imat}{ the contribution to the information matrix (second derivative of the log partial likelihood) at each time point. } \item{hazard}{ the hazard increment. Note that the hazard and variance of the hazard are always for some particular future subject. This routine uses \code{object$mean} as the future subject. } \item{varhaz}{ the variance of the hazard increment. } \item{x,y}{ copies of the input data. } \item{strata}{ only present for a stratified Cox model, this is a table giving the number of time points of component \code{time} that were contributed by each of the strata. } \item{riskmat}{ a matrix with one row for each time and one column for each observation containing a 0/1 value to indicate whether that observation was (1) or was not (0) at risk at the given time point. } } \details{ This function may be useful for those who wish to investigate new methods or extensions to the Cox model. The example below shows one way to calculate the Schoenfeld residuals. } \seealso{ \code{\link{coxph}}, \code{\link{residuals.coxph}} } \examples{ fit <- coxph(Surv(futime,fustat) ~ age + rx + ecog.ps, ovarian, x=TRUE) fitd <- coxph.detail(fit) # There is one Schoenfeld residual for each unique death. It is a # vector (covariates for the subject who died) - (weighted mean covariate # vector at that time). The weighted mean is defined over the subjects # still at risk, with exp(X beta) as the weight. events <- fit$y[,2]==1 etime <- fit$y[events,1] #the event times --- may have duplicates indx <- match(etime, fitd$time) schoen <- fit$x[events,] - fitd$means[indx,] } \keyword{survival} survival/man/agreg.fit.Rd0000644000175100001440000000261512234466366015073 0ustar hornikusers\name{agreg.fit} \alias{agreg.fit} \alias{coxph.fit} \title{Cox model fitting functions} \description{ These are the the functions called by coxph that do the actual computation. In certain situations, e.g. a simulation, it may be advantagous to call these directly rather than the usual \code{coxph} call using a model formula. } \usage{ agreg.fit(x, y, strata, offset, init, control, weights, method, rownames) coxph.fit(x, y, strata, offset, init, control, weights, method, rownames) } \arguments{ \item{x}{Matix of predictors. This should \emph{not} include an intercept.} \item{y}{a \code{Surv} object containing either 2 columns (coxph.fit) or 3 columns (agreg.fit).} \item{strata}{a vector containing the stratification, or NULL} \item{offset}{optional offset vector} \item{init}{initial values for the coefficients} \item{control}{the result of a call to \code{coxph.control}} \item{weights}{optional vector of weights} \item{method}{method for hanling ties, one of "breslow" or "efron"} \item{rownames}{this is only needed for a NULL model, in which case it contains the rownames (if any) of the original data.} } \details{ This routine does no checking that arguments are the proper length or type. Only use it if you know what you are doing! } \value{ a list containing results of the fit} \author{Terry Therneau} \seealso{\code{\link{coxph}}} \keyword{ survival } survival/man/strata.Rd0000644000175100001440000000247512372440140014510 0ustar hornikusers\name{strata} \alias{strata} \title{ Identify Stratification Variables } \description{ This is a special function used in the context of the Cox survival model. It identifies stratification variables when they appear on the right hand side of a formula. } \usage{ strata(..., na.group=FALSE, shortlabel, sep=', ') } \arguments{ \item{\dots}{ any number of variables. All must be the same length. } \item{na.group}{ a logical variable, if \code{TRUE}, then missing values are treated as a distinct level of each variable. } \item{shortlabel}{if \code{TRUE} omit variable names from resulting factor labels. The default action is to omit the names if all of the arguments are factors, and none of them was named.} \item{sep}{ the character used to separate groups, in the created label } } \value{ a new factor, whose levels are all possible combinations of the factors supplied as arguments. } \details{ The result is identical to the \code{interaction} function, but for the labeling of the factors (\code{strata} is more verbose). } \seealso{ \code{\link{coxph}}, \code{\link{interaction}} } \examples{ a <- factor(rep(1:3,4), labels="low", "medium", "high") b <- factor(rep(1:4,3)) levels(strata(b)) levels(strata(a,b,shortlabel=TRUE)) coxph(Surv(futime, fustat) ~ age + strata(rx), data=ovarian) } \keyword{survival} survival/man/stanford2.Rd0000644000175100001440000000132011732700061015100 0ustar hornikusers\name{stanford2} \alias{stanford2} \docType{data} \title{More Stanford Heart Transplant data} \description{ This contains the Stanford Heart Transplant data in a different format. The main data set is in \code{\link{heart}}. } \usage{stanford2} \format{ \tabular{ll}{ id: \tab ID number\cr time:\tab survival or censoring time\cr status:\tab censoring status\cr age: \tab in years\cr t5: \tab T5 mismatch score\cr } } \seealso{ \code{\link{predict.survreg}}, \code{\link{heart}} } \source{ LA Escobar and WQ Meeker Jr (1992), Assessing influence in regression analysis with censored data. \emph{Biometrics} \bold{48}, 507--528. Page 519. } \keyword{datasets} \keyword{survival} survival/man/lung.Rd0000644000175100001440000000225211732700061014150 0ustar hornikusers\name{lung} \docType{data} \alias{cancer} \alias{lung} \title{NCCTG Lung Cancer Data} \description{ Survival in patients with advanced lung cancer from the North Central Cancer Treatment Group. Performance scores rate how well the patient can perform usual daily activities. } \usage{ lung cancer } \format{ \tabular{ll}{ inst:\tab Institution code\cr time:\tab Survival time in days\cr status:\tab censoring status 1=censored, 2=dead\cr age:\tab Age in years\cr sex:\tab Male=1 Female=2\cr ph.ecog:\tab ECOG performance score (0=good 5=dead)\cr ph.karno:\tab Karnofsky performance score (bad=0-good=100) rated by physician\cr pat.karno:\tab Karnofsky performance score as rated by patient\cr meal.cal:\tab Calories consumed at meals\cr wt.loss:\tab Weight loss in last six months\cr } } \source{Terry Therneau} \references{ Loprinzi CL. Laurie JA. Wieand HS. Krook JE. Novotny PJ. Kugler JW. Bartel J. Law M. Bateman M. Klatt NE. et al. Prospective evaluation of prognostic variables from patient-completed questionnaires. North Central Cancer Treatment Group. Journal of Clinical Oncology. 12(3):601-7, 1994. } \keyword{datasets} survival/man/cgd0.Rd0000644000175100001440000000274112461741246014035 0ustar hornikusers\name{cgd0} \docType{data} \alias{cgd0} \title{Chronic Granulotomous Disease data} \description{Data are from a placebo controlled trial of gamma interferon in chronic granulotomous disease (CGD). Contains the data on time to serious infections observed through end of study for each patient. } \usage{cgd0} \format{ \describe{ \item{id}{subject identifiction number} \item{center}{enrolling center } \item{random}{date of randomization } \item{treatment}{placebo or gamma interferon } \item{sex}{sex} \item{age}{age in years, at study entry } \item{height}{height in cm at study entry} \item{weight}{weight in kg at study entry} \item{inherit}{pattern of inheritance } \item{steroids}{use of steroids at study entry,1=yes} \item{propylac}{use of prophylactic antibiotics at study entry} \item{hos.cat}{a categorization of the centers into 4 groups} \item{futime}{days to last follow-up} \item{etime1-etime7}{up to 7 infection times for the subject} } } \details{ The \code{cgdraw} data set (this one) is in the form found in the references, with one line per patient and no recoding of the variables. The \code{cgd} data set has been further processed so as to have one line per event, with covariates such as center recoded as factors to include meaningful labels. } \source{ Fleming and Harrington, Counting Processes and Survival Analysis, appendix D.2. } \seealso{\code{\link{cgd}}} \keyword{datasets} \keyword{survival} survival/man/survreg.object.Rd0000644000175100001440000000445412421527065016162 0ustar hornikusers\name{survreg.object} \alias{survreg.object} \alias{print.survreg} \alias{summary.survreg} \title{ Parametric Survival Model Object } \description{ This class of objects is returned by the \code{survreg} function to represent a fitted parametric survival model. Objects of this class have methods for the functions \code{print}, \code{summary}, \code{predict}, and \code{residuals}. } \section{COMPONENTS}{ The following components must be included in a legitimate \code{survreg} object. \describe{ \item{coefficients}{ the coefficients of the \code{linear.predictors}, which multiply the columns of the model matrix. It does not include the estimate of error (sigma). The names of the coefficients are the names of the single-degree-of-freedom effects (the columns of the model matrix). If the model is over-determined there will be missing values in the coefficients corresponding to non-estimable coefficients. } \item{icoef}{ coefficients of the baseline model, which will contain the intercept and log(scale), or mulitple scale factors for a stratified model. } \item{var}{ the variance-covariance matrix for the parameters, including the log(scale) parameter(s). } \item{loglik}{ a vector of length 2, containing the log-likelihood for the baseline and full models. } \item{iter}{ the number of iterations required } \item{linear.predictors}{ the linear predictor for each subject. } \item{df}{ the degrees of freedom for the final model. For a penalized model this will be a vector with one element per term. } \item{scale}{ the scale factor(s), with length equal to the number of strata. } \item{idf}{ degrees of freedom for the initial model. } \item{means}{ a vector of the column means of the coefficient matrix. } \item{dist}{ the distribution used in the fit.} \item{weights}{included for a weighted fit.} } The object will also have the following components found in other model results (some are optional): \code{linear predictors}, \code{weights}, \code{x}, \code{y}, \code{model}, \code{call}, \code{terms} and \code{formula}. See \code{lm}. } \seealso{ \code{\link{survreg}}, \code{\link{lm}} } \keyword{regression} \keyword{survival} % Converted by Sd2Rd version 0.3-2. survival/man/untangle.specials.Rd0000644000175100001440000000230411732700061016620 0ustar hornikusers\name{untangle.specials} \alias{untangle.specials} \title{ Help Process the `specials' Argument of the `terms' Function. } \description{ Given a \code{terms} structure and a desired special name, this returns an index appropriate for subscripting the \code{terms} structure and another appropriate for the data frame. } \usage{ untangle.specials(tt, special, order=1) } \arguments{ \item{tt}{ a \code{terms} object. } \item{special}{ the name of a special function, presumably used in the terms object. } \item{order}{ the order of the desired terms. If set to 2, interactions with the special function will be included. }} \value{ a list with two components: \item{vars}{ a vector of variable names, as would be found in the data frame, of the specials. } \item{terms}{ a numeric vector, suitable for subscripting the terms structure, that indexes the terms in the expanded model formula which involve the special. }} \examples{ formula<-Surv(tt,ss)~x+z*strata(id) tms<-terms(formula,specials="strata") ## the specials attribute attr(tms,"specials") ## main effects untangle.specials(tms,"strata") ## and interactions untangle.specials(tms,"strata",order=1:2) } \keyword{survival} % Converted by Sd2Rd version 0.3-2. survival/man/summary.aareg.Rd0000644000175100001440000000641311732700061015761 0ustar hornikusers\name{summary.aareg} \alias{summary.aareg} \title{ Summarize an aareg fit } \description{ Creates the overall test statistics for an Aalen additive regression model } \usage{ \method{summary}{aareg}(object, maxtime, test=c("aalen", "nrisk"), scale=1,...) } \arguments{ \item{object}{ the result of a call to the \code{aareg} function } \item{maxtime}{ truncate the input to the model at time "maxtime" } \item{test}{ the relative time weights that will be used to compute the test } \item{scale}{ scales the coefficients. For some data sets, the coefficients of the Aalen model will be very small (10-4); this simply multiplies the printed values by a constant, say 1e6, to make the printout easier to read. } \item{\dots}{for future methods} } \value{ a list is returned with the following components \item{ table }{ a matrix with rows for the intercept and each covariate, and columns giving a slope estimate, the test statistic, it's standard error, the z-score and a p-value } \item{ test }{ the time weighting used for computing the test statistics } \item{ test.statistic }{ the vector of test statistics } \item{ test.var }{ the model based variance matrix for the test statistic } \item{ test.var2 }{ optionally, a robust variance matrix for the test statistic } \item{ chisq }{ the overall test (ignoring the intercept term) for significance of any variable } \item{ n }{ a vector containing the number of observations, the number of unique death times used in the computation, and the total number of unique death times } } \details{ It is not uncommon for the very right-hand tail of the plot to have large outlying values, particularly for the standard error. The \code{maxtime} parameter can then be used to truncate the range so as to avoid these. This gives an updated value for the test statistics, without refitting the model. The slope is based on a weighted linear regression to the cumulative coefficient plot, and may be a useful measure of the overall size of the effect. For instance when two models include a common variable, "age" for instance, this may help to assess how much the fit changed due to the other variables, in leiu of overlaying the two plots. (Of course the plots are often highly non-linear, so it is only a rough substitute). The slope is not directly related to the test statistic, as the latter is invariant to any monotone transformation of time. } \seealso{ aareg, plot.aareg } \examples{ afit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, data=lung, dfbeta=TRUE) summary(afit) \dontrun{ slope test se(test) robust se z p Intercept 5.05e-03 1.9 1.54 1.55 1.23 0.219000 age 4.01e-05 108.0 109.00 106.00 1.02 0.307000 sex -3.16e-03 -19.5 5.90 5.95 -3.28 0.001030 ph.ecog 3.01e-03 33.2 9.18 9.17 3.62 0.000299 Chisq=22.84 on 3 df, p=4.4e-05; test weights=aalen } summary(afit, maxtime=600) \dontrun{ slope test se(test) robust se z p Intercept 4.16e-03 2.13 1.48 1.47 1.450 0.146000 age 2.82e-05 85.80 106.00 100.00 0.857 0.392000 sex -2.54e-03 -20.60 5.61 5.63 -3.660 0.000256 ph.ecog 2.47e-03 31.60 8.91 8.67 3.640 0.000271 Chisq=27.08 on 3 df, p=5.7e-06; test weights=aalen }} \keyword{survival} survival/man/survexp.us.Rd0000644000175100001440000000607611732700061015355 0ustar hornikusers\name{ratetables} \alias{survexp.us} \alias{survexp.usr} \alias{survexp.az} \alias{survexp.azr} \alias{survexp.fl} \alias{survexp.flr} \alias{survexp.mn} \alias{survexp.mnwhite} \alias{survexp.wnc} \title{ Census Data Sets for the Expected Survival and Person Years Functions } \description{ Census data sets for the expected survival and person years functions. } \details{ \describe{ \item{us}{ total United States population, by age and sex, 1960 to 1980. } \item{uswhite}{ United States white population, by age and sex, 1950 to 1980. This is no longer included, but can be extracted from \code{survexp.usr} as shown in the examples. } \item{usr}{ United States population, by age, sex and race, 1960 to 1980. Race is white, nonwhite, or black. For 1960 and 1970 the black population values were not reported separately, so the nonwhite values were used. } \item{mn}{ total Minnesota population, by age and sex, 1970 and 1980. } \item{mnwhite}{ Minnesota white population, by age and sex, 1960 to 1980. } \item{fl}{ total Florida population, by age and sex, 1970 and 1980. } \item{flr}{ Florida population, by age, sex and race, 1970-1980. Race is white, nonwhite, or black. For 1970 the black population values were not reported separately, so the nonwhite values were used. } \item{az}{ total Arizona population, by age and sex, 1970 and 1980. } \item{azr}{ Arizona population, by age, sex and race, 1970-1980. Race is white versus nonwhite. For 1970 the nonwhite population values were not reported separately. In order to make the rate table be a matrix, the 1980 values were repeated. (White and non-white values are quite different). } } Each of these tables contains the daily hazard rate for a matched subject from the population, defined as \eqn{-\log(1-q)/365.24} where \eqn{q} is the 1 year probability of death as reported in the original tables. For age 25 in 1970, for instance, \eqn{p = 1-q} is is the probability that a subject who becomes 25 years of age in 1970 will achieve his/her 26th birthday. The tables are recast in terms of hazard per day entirely for computational convenience. (The fraction .24 in the denominator is based on 24 leap years per century.) Each table is stored as an array, with additional attributes, and can be subset and manipulated as standard S arrays. Interpolation between calendar years is done \dQuote{on the fly} by the \code{\link{survexp}} routine. Some of the deficiencies, e.g., 1970 Arizona non-white, are a result of local (Mayo Clinic) conditions. The data probably exists, but we don't have a copy it in the library. The tables have been augmented to contain extrapolated values for 1990 and 2000. The details can be found in Mayo Clinic Biostatistics technical report 63 at \url{http://www.mayo.edu/hsr/techrpt.html}. } \examples{ survexp.uswhite <- survexp.usr[,,"white",] } \keyword{survival} \keyword{datasets} survival/man/tmerge.Rd0000644000175100001440000001415712465225442014506 0ustar hornikusers\name{tmerge} \alias{tmerge} \title{Time based merge for survival data} \description{ A common task in survival analysis is the creation of start,stop data sets which have multiple intervals for each subject, along with the covariate values that apply over that interval. This function aids in the creation of such data sets. } \usage{ tmerge(data1, data2, id,\dots, tstart, tstop, options) } \arguments{ \item{data1}{the primary data set, to which new variables and/or observation will be added} \item{data2}{optional second data set in which the other arguments will be found} \item{id}{subject identifier} \item{\dots}{operations that add new variables or intervals, see below} \item{tstart}{optional variable to define the valid time range for each subject, only used on an initial call} \item{tstop}{optional variable to define the valid time range for each subject, only used on an initial call} \item{options}{a list of options. Valid ones are id, tstart, and tstop, which will be the names of the three mandatory variables in the output data. The other is defer, which sets a numeric amount of time before an event when covariate changes are disallowed (the are deferred until after the event in that case.)} } \details{ The program is usually run in multiple passes, the first of which defines the basic structure, and subsequent ones that add new variables to that structure. For a more complete explanation of how this routine works refer to the vignette on time-dependent variables. There are 4 types of optional arguments: a time dependent covariate (tdc), cumulative count (cumtdc), event (event) or cumulative event (cumevent). Time dependent covariates change their values before an event, events are outcomes. \itemize{ \item{newname = tdc(y, x)}{A new time dependent covariate variable will created. The argument \code{y} is assumed to be on the scale of the start and end time, and each instance decribes the occurent of a "condition" at that time. The second argument \code{x} is optional. In the case where \code{x} is missing the count variable starts at 0 for each subject and becomes 1 at the time of the event; if \code{x} is present the count is set to the value of \code{x}. If a given subject has multiple rows of data with the same time value the sum of those rows will be assigned. \item{newname = cumtdc(y,x)}{Similar to tdc, except that the event count is accumulated over time for each subject.} \item{newname = event(y,x)}{Mark an event at time y. In the ususal case that \code{x} is missing, the new 0/1 variable will be similar to the 0/1 status variable of a survival time, and that is in fact how it will normally be used. For multiple types of endpoints the \code{x} argument can be used encode the type of event. } \item{newname = cumevent(y,x)}{Cumulative events}. } } Say that a subject had an interval of observation from age 17 to 38, denoted as (17, 38] and that a marker occurs at age 24. A tdc variable is a predictor which is assumed to apply from the time it occured to the end of followup for the subject. The updated data set will have intervals of (17,24] and (24, 38] with a count of 0 for the first interval and 1 for the second, assuming no other occurences for this subject at exactly time 24. An event is an outcome, so if coded as an event the said occurence would be placed in the (17,24] interval, with the new variable marking that this interval finished with an event. } \value{a data frame with two extra attributes \code{tname} and \code{tcount}. The first contains the names of the key variables; it's persistence from call to call allows the user to avoid constantly reentering the \code{options} arguments. The tcount variable contains counts of the match types. New time values that occur before the first interval for a subject are "early", those after the last interval for a subject are "late", and those that fall into a gap are of type "gap". The most common type will usually be "within", for those new times that fall inside an existing interval and cause it to be split into two. Observations that fall exactly on the edge of an interval are counted as "leading" edge, "trailing" or "boundary". The first corresponds for instance to an occurence at 17 for someone with an interval (17, 35] who is not at risk just before time 17. A \code{tdc} at time 17 will affect this interval but not an \code{event}. Symmetrically an \code{event} occurence at 35 would count in the (17,35] interval, but a \code{tdc} would not. The last case is where the main data set has touching intervals for a subject, e.g. (17, 28] and (28,35] and a new occurence lands at the join. Events will go to the earlier interval and counts to the latter one. It is wise to look at \code{attr(data, 'tcount')} after each step of a data set build to avoid surprises. These extra attributes are ephemeral, and will be discarded if the dataframe is modified in any way. This is intentional. } \author{Terry Therneau} \seealso{\code{\link{neardate}}} \examples{ # The data set jasa contains the famous Stanford Heart Transplant data # set, as it appeared in Crowley and Hu, JASA 72:27-36, 1971. # Two special cases need to be dealt with: # subject 15 died on day 0 which leads to an illegal (0,0] interval, # make them die on day 0.5 instead # subject 38 dies on the day of transplant, make tx happen "earlier in # the day" (before death) by subtracting .1 from their transplant day # tdata <- jasa[, -(1:4)] #leave off the dates, temporary data set tdata$futime <- pmax(.5, tdata$futime) # the death on day 0 indx <- with(tdata, which(wait.time == futime)) tdata$wait.time[indx] <- tdata$wait.time[indx] - .5 #the tied transplant sdata <- tmerge(tdata, tdata, id=1:nrow(tdata), death = event(futime, fustat), trans = tdc(wait.time)) attr(sdata, "tcount") # Shows two subjects transplanted on the day of entry, the "front edge" of # their follow-up interval fit <- coxph(Surv(tstart, tstop, death) ~ trans + age, data=sdata) } \keyword{ survival } survival/man/ratetableDate.Rd0000644000175100001440000000152011732700061015741 0ustar hornikusers\name{ratetableDate} \alias{ratetableDate} \title{Convert date objects to ratetable form} \description{ This method converts dates from various forms into the internal form used in \code{ratetable} objects. } \usage{ ratetableDate(x) } \arguments{ \item{x}{a date. The function currently has methods for Date, date, POSIXt, timeDate, and chron objects. } } \details{ This function is useful for those who create new ratetables, but is normally invisible to users. It is used internally by the \code{survexp} and \code{pyears} functions to map the various date formats; if a new method is added then those routines will automatically be adapted to the new date type. } \value{a numeric vector, the number of days since 1/1/1960.} \author{Terry Therneau} \seealso{\code{\link{pyears}}, \code{\link{survexp}}} \keyword{survival} survival/man/predict.coxph.Rd0000644000175100001440000001154112347642052015766 0ustar hornikusers\name{predict.coxph} \alias{predict.coxph} \alias{predict.coxph.penal} \title{ Predictions for a Cox model } \description{ Compute fitted values and regression terms for a model fitted by \code{\link{coxph}} } \usage{ \method{predict}{coxph}(object, newdata, type=c("lp", "risk", "expected", "terms"), se.fit=FALSE, na.action=na.pass, terms=names(object$assign), collapse, reference=c("strata", "sample"), ...) } \arguments{ \item{object}{ the results of a coxph fit. } \item{newdata}{ Optional new data at which to do predictions. If absent predictions are for the data frame used in the original fit. When coxph has been called with a formula argument created in another context, i.e., coxph has been called within another function and the formula was passed as an argument to that function, there can be problems finding the data set. See the note below. } \item{type}{ the type of predicted value. Choices are the linear predictor (\code{"lp"}), the risk score exp(lp) (\code{"risk"}), the expected number of events given the covariates and follow-up time (\code{"expected"}), and the terms of the linear predictor (\code{"terms"}). } \item{se.fit}{ if TRUE, pointwise standard errors are produced for the predictions. } \item{na.action}{ applies only when the \code{newdata} argument is present, and defines the missing value action for the new data. The default is to include all observations. When there is no newdata, then the behavior of missing is dictated by the na.action option of the original fit.} \item{terms}{ if type="terms", this argument can be used to specify which terms should be included; the default is all. } \item{collapse}{ optional vector of subject identifiers. If specified, the output will contain one entry per subject rather than one entry per observation. } \item{reference}{reference for centering predictions, see details below} \item{\dots}{For future methods} } \value{ a vector or matrix of predictions, or a list containing the predictions (element "fit") and their standard errors (element "se.fit") if the se.fit option is TRUE. } \details{ The Cox model is a \emph{relative} risk model; predictions of type "linear predictor", "risk", and "terms" are all relative to the sample from which they came. By default, the reference value for each of these is the mean covariate within strata. The primary underlying reason is statistical: a Cox model only predicts relative risks between pairs of subjects within the same strata, and hence the addition of a constant to any covariate, either overall or only within a particular stratum, has no effect on the fitted results. Using the \code{reference="strata"} option causes this to be true for predictions as well. When the results of \code{predict} are used in further calculations it may be desirable to use a fixed reference level. Use of \code{reference="sample"} will use the overall means, and agrees with the \code{linear.predictors} component of the coxph object (which uses the overall mean for backwards compatability with older code). Predictions of \code{type="terms"} are almost invariably passed forward to further calculation, so for these we default to using the sample as the reference. Predictions of type "expected" incorporate the baseline hazard and are thus absolute instead of relative; the \code{reference} option has no effect on these. Models that contain a \code{frailty} term are a special case: due to the technical difficulty, when there is a \code{newdata} argument the predictions will always be for a random effect of zero. } \note{ Some predictions can be obtained directly from the coxph object, and for others it is necessary for the routine to have the entirety of the original data set, e.g., for type = \code{terms} or if standard errors are requested. This extra information is saved in the coxph object if \code{model=TRUE}, if not the original data is reconstructed. If it is known that such residuals will be required overall execution will be slightly faster if the model information is saved. In some cases the reconstruction can fail. The most common is when coxph has been called inside another function and the formula was passed as one of the arguments to that enclosing function. Another is when the data set has changed between the original call and the time of the prediction call. In each of these the simple solution is to add \code{model=TRUE} to the original coxph call. } \seealso{ \code{\link{predict}},\code{\link{coxph}},\code{\link{termplot}} } \examples{ options(na.action=na.exclude) # retain NA in predictions fit <- coxph(Surv(time, status) ~ age + ph.ecog + strata(inst), lung) #lung data set has status coded as 1/2 mresid <- (lung$status-1) - predict(fit, type='expected') #Martingale resid predict(fit,type="lp") predict(fit,type="expected") predict(fit,type="risk",se.fit=TRUE) predict(fit,type="terms",se.fit=TRUE) } \keyword{survival} survival/man/cgd.Rd0000644000175100001440000000306012461741246013750 0ustar hornikusers\name{cgd} \docType{data} \alias{cgd} \alias{cgd.raw} \title{Chronic Granulotomous Disease data} \description{Data are from a placebo controlled trial of gamma interferon in chronic granulotomous disease (CGD). Contains the data on time to serious infections observed through end of study for each patient. } \usage{cgd} \format{ \describe{ \item{id}{subject identifiction number} \item{center}{enrolling center } \item{random}{date of randomization } \item{treatment}{placebo or gamma interferon } \item{sex}{sex} \item{age}{age in years, at study entry } \item{height}{height in cm at study entry} \item{weight}{weight in kg at study entry} \item{inherit}{pattern of inheritance } \item{steroids}{use of steroids at study entry,1=yes} \item{propylac}{use of prophylactic antibiotics at study entry} \item{hos.cat}{a categorization of the centers into 4 groups} \item{tstart, tstop}{start and end of each time interval } \item{status}{1=the interval ends with an infection } \item{enum}{observation number within subject} } } \details{ The \code{cgd0} data set is in the form found in the references, with one line per patient and no recoding of the variables. The \code{cgd} data set (this one) has been cast into (start, stop] format with one line per event, and covariates such as center recoded as factors to include meaningful labels. } \source{ Fleming and Harrington, Counting Processes and Survival Analysis, appendix D.2. } \seealso{\code{link{cgd0}}} \keyword{datasets} \keyword{survival} survival/man/plot.survfit.Rd0000644000175100001440000001400212375216043015664 0ustar hornikusers\name{plot.survfit} \alias{plot.survfit} \title{ Plot method for \code{survfit} objects } \usage{ \method{plot}{survfit}(x, conf.int=, mark.time=TRUE, mark=3, col=1, lty=1, lwd=1, cex=1, log=FALSE, xscale=1, yscale=1, firstx=0, firsty=1, xmax, ymin=0, fun, xlab="", ylab="", xaxs="S", \dots) } \arguments{ \item{x}{ an object of class \code{survfit}, usually returned by the \code{survfit} function. } \item{conf.int}{ determines whether confidence intervals will be plotted. The default is to do so if there is only 1 curve, i.e., no strata. } \item{mark.time}{ controls the labeling of the curves. If set to \code{FALSE}, no labeling is done. If \code{TRUE}, then curves are marked at each censoring time which is not also a death time. If \code{mark.time} is a numeric vector, then curves are marked at the specified time points. } \item{mark}{ vector of mark parameters, which will be used to label the curves. The \code{lines} help file contains examples of the possible marks. The vector is reused cyclically if it is shorter than the number of curves. } \item{col}{ a vector of integers specifying colors for each curve. The default value is 1. } \item{lty}{ a vector of integers specifying line types for each curve. The default value is 1. } \item{lwd}{ a vector of numeric values for line widths. The default value is 1. } \item{cex}{ a numeric value specifying the size of the marks. This is not treated as a vector; all marks have the same size. } \item{log}{ a logical value, if TRUE the y axis wll be on a log scale. Alternately, one of the standard character strings "x", "y", or "xy" can be given to specific logarithmic horizontal and/or vertical axes. } \item{yscale}{ a numeric value used to multiply the labels on the y axis. A value of 100, for instance, would be used to give a percent scale. Only the labels are changed, not the actual plot coordinates, so that adding a curve with "\code{lines(surv.exp(...))}", say, will perform as it did without the \code{yscale} argument. } \item{xscale}{ a numeric value used like \code{yscale} for labels on the x axis. A value of 365.25 will give labels in years instead of the original days. } \item{firstx, firsty}{ the starting point for the survival curves. If either of these is set to \code{NA} the plot will start at the first time point of the curve. By default, the plot program obeys tradition by having the plot start at (0,0). If \code{start.time} argument is used in \code{survfit}, \code{firstx} is set to that value. } \item{xmax}{ the maximum horizontal plot coordinate. This can be used to shrink the range of a plot. It shortens the curve before plotting it, so that unlike using the \code{xlim} graphical parameter, warning messages about out of bounds points are not generated. } \item{ymin}{ lower boundary for y values. Survival curves are most often drawn in the range of 0-1, even if none of the curves approach zero. The parameter is ignored if the \code{fun} argument is present, or if it has been set to \code{NA}. } \item{fun}{ an arbitrary function defining a transformation of the survival curve. For example \code{fun=log} is an alternative way to draw a log-survival curve (but with the axis labeled with log(S) values), and \code{fun=sqrt} would generate a curve on square root scale. Four often used transformations can be specified with a character argument instead: \code{"log"} is the same as using the \code{log=T} option, \code{"event"} plots cumulative events (f(y) = 1-y), \code{"cumhaz"} plots the cumulative hazard function (f(y) = -log(y)), and \code{"cloglog"} creates a complimentary log-log survival plot (f(y) = log(-log(y)) along with log scale for the x-axis). } \item{xlab}{ label given to the x-axis. } \item{ylab}{ label given to the y-axis. } \item{xaxs}{ either \code{"S"} for a survival curve or a standard x axis style as listed in \code{par}. Survival curves are usually displayed with the curve touching the y-axis, but not touching the bounding box of the plot on the other 3 sides. Type \code{"S"} accomplishes this by manipulating the plot range and then using the \code{"i"} style internally. } \item{\dots}{for future methods} } \value{ a list with components \code{x} and \code{y}, containing the coordinates of the last point on each of the curves (but not the confidence limits). This may be useful for labeling. } \description{ A plot of survival curves is produced, one curve for each strata. The \code{log=T} option does extra work to avoid log(0), and to try to create a pleasing result. If there are zeros, they are plotted by default at 0.8 times the smallest non-zero value on the curve(s). Curves are plotted in the same order as they are listed by \code{print} (which gives a 1 line summary of each). This will be the order in which \code{col}, \code{lty}, etc are used. } \details{ When the \code{survfit} function creates a multi-state survival curve the resulting object also has class `survfitms'. Competing risk curves are a common case. The only difference in the plots is that multi-state defaults to a curve that goes from lower left to upper right (starting at 0), where survival curves by default start at 1 and go down. All other options are identical. } \seealso{ \code{\link{points.survfit}}, \code{\link{lines.survfit}}, \code{\link{par}}, \code{\link{survfit}} } \examples{ leukemia.surv <- survfit(Surv(time, status) ~ x, data = aml) plot(leukemia.surv, lty = 2:3) legend(100, .9, c("Maintenance", "No Maintenance"), lty = 2:3) title("Kaplan-Meier Curves\nfor AML Maintenance Study") lsurv2 <- survfit(Surv(time, status) ~ x, aml, type='fleming') plot(lsurv2, lty=2:3, fun="cumhaz", xlab="Months", ylab="Cumulative Hazard") } \keyword{survival} \keyword{hplot} survival/man/summary.coxph.Rd0000644000175100001440000000263311732700061016023 0ustar hornikusers\name{summary.coxph} \alias{summary.coxph} \title{ Summary method for Cox models } \description{ Produces a summary of a fitted coxph model } \usage{ \method{summary}{coxph}(object, conf.int=0.95, scale=1,...) } \arguments{ \item{object}{ the result of a coxph fit } \item{conf.int}{ level for computation of the confidence intervals. If set to FALSE no confidence intervals are printed } \item{scale}{ vector of scale factors for the coefficients, defaults to 1. The confidence limits are for the risk change associated with one scale unit. } \item{\dots}{for future methods} } \value{ An object of class \code{summary.coxph}. } \seealso{ coxph, print.coxph } \examples{ fit <- coxph(Surv(time, status) ~ age + sex, lung) summary(fit) \dontrun{ Call: coxph(formula = Surv(time, status) ~ age + sex, data = lung) n= 228 coef exp(coef) se(coef) z p age 0.017 1.017 0.00922 1.85 0.0650 sex -0.513 0.599 0.16745 -3.06 0.0022 exp(coef) exp(-coef) lower .95 upper .95 age 1.017 0.983 0.999 1.036 sex 0.599 1.670 0.431 0.831 Rsquare= 0.06 (max possible= 0.999 ) Likelihood ratio test= 14.1 on 2 df, p=0.000857 Wald test = 13.5 on 2 df, p=0.00119 Score (logrank) test = 13.7 on 2 df, p=0.00105 }} \keyword{survival} % docclass is function % Converted by Sd2Rd version 37351. survival/man/mgus.Rd0000644000175100001440000000675412461741246014203 0ustar hornikusers\name{mgus} \alias{mgus} \alias{mgus1} \docType{data} \title{Monoclonal gammapothy data} \description{ Natural history of 241 subjects with monoclonal gammapothy of undetermined significance (MGUS). } \usage{ mgus mgus1 } \format{ mgus: A data frame with 241 observations on the following 12 variables. \tabular{ll}{ id:\tab subject id \cr age:\tab age in years at the detection of MGUS \cr sex:\tab \code{male} or \code{female} \cr dxyr:\tab year of diagnosis \cr pcdx:\tab for subjects who progress to a plasma cell malignancy \cr \tab the subtype of malignancy: multiple myeloma (MM) is the \cr \tab most common, followed by amyloidosis (AM), macroglobulinemia (MA),\cr \tab and other lymphprolifative disorders (LP) \cr pctime:\tab days from MGUS until diagnosis of a plasma cell malignancy \cr futime:\tab days from diagnosis to last follow-up \cr death:\tab 1= follow-up is until death \cr alb:\tab albumin level at MGUS diagnosis \cr creat:\tab creatinine at MGUS diagnosis \cr hgb:\tab hemoglobin at MGUS diagnosis \cr mspike:\tab size of the monoclonal protien spike at diagnosis \cr } mgus1: The same data set in start,stop format. Contains the id, age, sex, and laboratory variable described above along with \tabular{ll}{ start, stop:\tab sequential intervals of time for each subject \cr status:\tab =1 if the interval ends in an event \cr event:\tab a factor containing the event type: censor, death, or plasma cell malignancy \cr enum: \tab event number for each subject: 1 or 2 } } \details{ Plasma cells are responsible for manufacturing immunoglobulins, an important part of the immune defense. At any given time there are estimated to be about \eqn{10^6} different immunoglobulins in the circulation at any one time. When a patient has a plasma cell malignancy the distribuion will become dominated by a single isotype, the product of the malignant clone, visible as a spike on a serum protein electrophoresis. Monoclonal gammapothy of undertermined significance (MGUS) is the presence of such a spike, but in a patient with no evidence of overt malignancy. This data set of 241 sequential subjects at Mayo Clinic was the groundbreaking study defining the natural history of such subjects. Due to the diligence of the principle investigator 0 subjects have been lost to follow-up. Three subjects had MGUS detected on the day of death. In data set \code{mgus1} these subjects have the time to MGUS coded as .5 day before the death in order to avoid tied times. These data sets were updated in Jan 2015 to correct some small errors. } \source{ Mayo Clinic data courtesy of Dr. Robert Kyle. } \examples{ # Create the competing risk curves for time to first of death or PCM sfit <- survfit(Surv(start, stop, event) ~ sex, mgus1, subset=(enum==1)) print(sfit) # the order of printout is the order in which they plot plot(sfit, xscale=365.25, lty=c(2,1,2,1), col=c(1,1,2,2), xlab="Years after MGUS detection", ylab="Proportion") legend(0, .8, c("Death/male", "Death/female", "PCM/male", "PCM/female"), lty=c(1,1,2,2), col=c(2,1,2,1), bty='n') title("Curves for the first of plasma cell malignancy or death") # The plot shows that males have a higher death rate than females (no # surprise) but their rates of conversion to PCM are essentially the same. } \references{ R Kyle, Benign monoclonal gammopathy -- after 20 to 35 years of follow-up, Mayo Clinic Proc 1993; 68:26-36. } \keyword{datasets} \keyword{survival} survival/man/kidney.Rd0000644000175100001440000000263511732700061014473 0ustar hornikusers\name{kidney} \alias{kidney} \title{Kidney catheter data} \format{ \tabular{ll}{ patient:\tab id\cr time:\tab time\cr status:\tab event status\cr age:\tab in years\cr sex:\tab 1=male, 2=female\cr disease:\tab disease type (0=GN, 1=AN, 2=PKD, 3=Other)\cr frail:\tab frailty estimate from original paper\cr }} \description{ Data on the recurrence times to infection, at the point of insertion of the catheter, for kidney patients using portable dialysis equipment. Catheters may be removed for reasons other than infection, in which case the observation is censored. Each patient has exactly 2 observations. This data has often been used to illustrate the use of random effects (frailty) in a survival model. However, one of the males (id 21) is a large outlier, with much longer survival than his peers. If this observation is removed no evidence remains for a random subject effect. } \section{Note}{ The original paper ignored the issue of tied times and so is not exactly reproduced by the survival package. } \examples{ kfit <- coxph(Surv(time, status)~ age + sex + disease + frailty(id), kidney) kfit0 <- coxph(Surv(time, status)~ age + sex + disease, kidney) kfitm1 <- coxph(Surv(time,status) ~ age + sex + disease + frailty(id, dist='gauss'), kidney) } \source{ CA McGilchrist, CW Aisbett (1991), Regression with frailty in survival analysis. \emph{Biometrics} \bold{47}, 461--66. } \keyword{survival} survival/man/survfit.object.Rd0000644000175100001440000000625511732700061016161 0ustar hornikusers\name{survfit.object} \alias{survfit.object} \title{ Survival Curve Object } \description{ This class of objects is returned by the \code{survfit} class of functions to represent a fitted survival curve. Objects of this class have methods for the functions \code{print}, \code{summary}, \code{plot}, \code{points} and \code{lines}. The \code{\link{print.survfit}} method does more computation than is typical for a print method and is documented on a separate page. Class of objects that represent a fitted survival curve. } \section{Structure}{ The following components must be included in a legitimate \code{survfit} object. } \arguments{ \item{n}{ total number of subjects in each curve. } \item{time}{ the time points at which the curve has a step. } \item{n.risk}{ the number of subjects at risk at t. } \item{n.event}{ the number of events that occur at time t. } \item{n.enter}{ for counting process data only, the number of subjects that enter at time t. } \item{n.censor}{ for counting process data only, the number of subjects who exit the risk set, without an event, at time t. (For right censored data, this number can be computed from the successive values of the number at risk). } \item{surv}{ the estimate of survival at time t+0. This may be a vector or a matrix. } \item{std.err}{ the standard error of the cumulative hazard or -log(survival). } \item{upper}{ upper confidence limit for the survival curve. } \item{lower}{ lower confidence limit for the survival curve. } \item{strata}{ if there are multiple curves, this component gives the number of elements of the \code{time} etc. vectors corresponding to the first curve, the second curve, and so on. The names of the elements are labels for the curves. } \item{start.time}{ the value specified for the \code{start.time} argument, if it was used in the call. } \item{n.all}{ for counting process data, and any time that the \code{start.time} argument was used, this contains the total number of observations that were available. Not all may have been used in creating the curve, in which case this value will be larger than \code{n} above. of observations that were available } \item{conf.type}{ the approximation used to compute the confidence limits. } \item{conf.int}{ the level of the confidence limits, e.g. 90 or 95\%. } \item{na.action}{ the returned value from the na.action function, if any. It will be used in the printout of the curve, e.g., the number of observations deleted due to missing values. } \item{call}{ an image of the call that produced the object. } \item{type}{ type of survival censoring. } } \section{Subscripts}{ Survfit objects that contain multiple survival curves can be subscripted. This is most often used to plot a subset of the curves. Usually a single subscript will be used. In one particular case -- survival curves for multiple covariate values, from a Cox model that includes a \code{strata} statement -- there is a matrix of curves and 2 subscripts may be used. (In this case \code{summary.survfit} will also print the data as a matrix). } \seealso{ \code{\link{plot.survfit}}, \code{\link{summary.survfit}}, \code{\link{print.survfit}}, \code{\link{survfit}}. } \keyword{survival} survival/man/pspline.Rd0000644000175100001440000000677711732700061014675 0ustar hornikusers\name{pspline} \alias{pspline} \alias{psplineinverse} \title{Smoothing splines using a pspline basis} \usage{ pspline(x, df=4, theta, nterm=2.5 * df, degree=3, eps=0.1, method, Boundary.knots=range(x), intercept=FALSE, penalty=TRUE, ...) psplineinverse(x)} \arguments{ \item{x}{for psline: a covariate vector. The function does not apply to factor variables. For psplineinverse x will be the result of a pspline call.} \item{df}{the desired degrees of freedom. One of the arguments \code{df} or \code{theta}' must be given, but not both. If \code{df=0}, then the AIC = (loglik -df) is used to choose an "optimal" degrees of freedom. If AIC is chosen, then an optional argument `caic=T' can be used to specify the corrected AIC of Hurvich et. al. } \item{theta}{roughness penalty for the fit. It is a monotone function of the degrees of freedom, with theta=1 corresponding to a linear fit and theta=0 to an unconstrained fit of nterm degrees of freedom. } \item{nterm}{ number of splines in the basis } \item{degree}{ degree of splines } \item{eps}{accuracy for \code{df} } \item{method}{the method for choosing the tuning parameter \code{theta}. If theta is given, then 'fixed' is assumed. If the degrees of freedom is given, then 'df' is assumed. If method='aic' then the degrees of freedom is chosen automatically using Akaike's information criterion.} \item{\dots}{optional arguments to the control function} \item{Boundary.knots}{the spline is linear beyond the boundary knots. These default to the range of the data.} \item{intercept}{if TRUE, the basis functions include the intercept.} \item{penalty}{if FALSE a large number of attributes having to do with penalized fits are excluded. Most useful for exploring the code so as to return a matrix with few added attributes.} } \description{ Specifies a penalised spline basis for the predictor. This is done by fitting a comparatively small set of splines and penalising the integrated second derivative. Traditional smoothing splines use one basis per observation, but several authors have pointed out that the final results of the fit are indistinguishable for any number of basis functions greater than about 2-3 times the degrees of freedom. Eilers and Marx point out that if the basis functions are evenly spaced, this leads to significant computational simplifications. } \value{ Object of class \code{pspline, coxph.penalty} containing the spline basis, with the appropriate attributes to be recognized as a penalized term by the coxph or survreg functions. For psplineinverse the original x vector is reconstructed. } \seealso{\code{\link{coxph}},\code{\link{survreg}},\code{\link{ridge}}, \code{\link{frailty}} } \references{ Eilers, Paul H. and Marx, Brian D. (1996). Flexible smoothing with B-splines and penalties. Statistical Science, 11, 89-121. Hurvich, C.M. and Simonoff, J.S. and Tsai, Chih-Ling (1998). Smoothing parameter selection in nonparametric regression using an improved Akaike information criterion, JRSSB, volume 60, 271--293. } \examples{ lfit6 <- survreg(Surv(time, status)~pspline(age, df=2), cancer) plot(cancer$age, predict(lfit6), xlab='Age', ylab="Spline prediction") title("Cancer Data") fit0 <- coxph(Surv(time, status) ~ ph.ecog + age, cancer) fit1 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,3), cancer) fit3 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,8), cancer) fit0 fit1 fit3 } \keyword{ survival} survival/man/pyears.Rd0000644000175100001440000001705112516205563014521 0ustar hornikusers\name{pyears} \alias{pyears} \title{ Person Years } \description{ This function computes the person-years of follow-up time contributed by a cohort of subjects, stratified into subgroups. It also computes the number of subjects who contribute to each cell of the output table, and optionally the number of events and/or expected number of events in each cell. } \usage{ pyears(formula, data, weights, subset, na.action, rmap, ratetable, scale=365.25, expect=c('event', 'pyears'), model=FALSE, x=FALSE, y=FALSE, data.frame=FALSE) } \arguments{ \item{formula}{ a formula object. The response variable will be a vector of follow-up times for each subject, or a \code{Surv} object containing the survival time and an event indicator. The predictors consist of optional grouping variables separated by + operators (exactly as in \code{survfit}), time-dependent grouping variables such as age (specified with \code{tcut}), and optionally a \code{ratetable} term. This latter matches each subject to his/her expected cohort. } \item{data}{ a data frame in which to interpret the variables named in the \code{formula}, or in the \code{subset} and the \code{weights} argument. } \item{weights}{ case weights. } \item{subset}{ expression saying that only a subset of the rows of the data should be used in the fit. } \item{na.action}{ a missing-data filter function, applied to the model.frame, after any \code{subset} argument has been used. Default is \code{options()$na.action}. } \item{rmap}{ an optional list that maps data set names to the ratetable names. See the details section below. } \item{ratetable}{ a table of event rates, such as \code{survexp.uswhite}. } \item{scale}{ a scaling for the results. As most rate tables are in units/day, the default value of 365.25 causes the output to be reported in years. } \item{expect}{ should the output table include the expected number of events, or the expected number of person-years of observation. This is only valid with a rate table. } \item{data.frame}{ return a data frame rather than a set of arrays.} \item{model, x, y}{ If any of these is true, then the model frame, the model matrix, and/or the vector of response times will be returned as components of the final result. } } \value{ a list with components: \item{pyears}{ an array containing the person-years of exposure. (Or other units, depending on the rate table and the scale). The dimension and dimmanes of the array correspond to the variables on the right hand side of the model equation. } \item{n}{ an array containing the number of subjects who contribute time to each cell of the \code{pyears} array. } \item{event}{ an array containing the observed number of events. This will be present only if the response variable is a \code{Surv} object. } \item{expected}{ an array containing the expected number of events (or person years if \code{expect ="pyears"}). This will be present only if there was a \code{ratetable} term. } \item{data}{ if the \code{data.frame} option was set, a data frame containing the variables \code{n}, \code{event}, \code{pyears} and \code{event} that supplants the four arrays listed above, along with variables corresponding to each dimension. There will be one row for each cell in the arrays.} \item{offtable}{ the number of person-years of exposure in the cohort that was not part of any cell in the \code{pyears} array. This is often useful as an error check; if there is a mismatch of units between two variables, nearly all the person years may be off table. } \item{summary}{ a summary of the rate-table matching. This is also useful as an error check. } \item{call}{ an image of the call to the function. } \item{observations}{the number of observations in the input data set, after any missings were removed.} \item{na.action}{ the \code{na.action} attribute contributed by an \code{na.action} routine, if any. } } \details{ Because \code{pyears} may have several time variables, it is necessary that all of them be in the same units. For instance, in the call \preformatted{ py <- pyears(futime ~ rx, rmap=list(age=age, sex=sex, year=entry.dt), ratetable=survexp.us) } the natural unit of the ratetable is hazard per day, it is important that \code{futime}, \code{age} and \code{entry.dt} all be in days. Given the wide range of possible inputs, it is difficult for the routine to do sanity checks of this aspect. The ratetable being used may have different variable names than the user's data set, this is dealt with by the \code{rmap} argument. The rate table for the above calculation was \code{survexp.us}, a call to \code{summary{survexp.us}} reveals that it expects to have variables \code{age} = age in days, \code{sex}, and \code{year} = the date of study entry, we create them in the \code{rmap} line. The sex variable is not mapped, therefore the code assumes that it exists in \code{mydata} in the correct format. (Note: for factors such as sex, the program will match on any unique abbreviation, ignoring case.) A special function \code{tcut} is needed to specify time-dependent cutpoints. For instance, assume that age is in years, and that the desired final arrays have as one of their margins the age groups 0-2, 2-10, 10-25, and 25+. A subject who enters the study at age 4 and remains under observation for 10 years will contribute follow-up time to both the 2-10 and 10-25 subsets. If \code{cut(age, c(0,2,10,25,100))} were used in the formula, the subject would be classified according to his starting age only. The \code{tcut} function has the same arguments as \code{cut}, but produces a different output object which allows the \code{pyears} function to correctly track the subject. The results of \code{pyears} are normally used as input to further calculations. The \code{print} routine, therefore, is designed to give only a summary of the table. } \seealso{ \code{\link{ratetable}}, \code{\link{survexp}}, \code{\link{Surv}}. } \examples{ # Look at progression rates jointly by calendar date and age # temp.yr <- tcut(mgus$dxyr, 55:92, labels=as.character(55:91)) temp.age <- tcut(mgus$age, 34:101, labels=as.character(34:100)) ptime <- ifelse(is.na(mgus$pctime), mgus$futime, mgus$pctime) pstat <- ifelse(is.na(mgus$pctime), 0, 1) pfit <- pyears(Surv(ptime/365.25, pstat) ~ temp.yr + temp.age + sex, mgus, data.frame=TRUE) # Turn the factor back into numerics for regression tdata <- pfit$data tdata$age <- as.numeric(as.character(tdata$temp.age)) tdata$year<- as.numeric(as.character(tdata$temp.yr)) fit1 <- glm(event ~ year + age+ sex +offset(log(pyears)), data=tdata, family=poisson) \dontrun{ # fit a gam model gfit.m <- gam(y ~ s(age) + s(year) + offset(log(time)), family = poisson, data = tdata) } # Example #2 Create the hearta data frame: hearta <- by(heart, heart$id, function(x)x[x$stop == max(x$stop),]) hearta <- do.call("rbind", hearta) # Produce pyears table of death rates on the surgical arm # The first is by age at randomization, the second by current age fit1 <- pyears(Surv(stop/365.25, event) ~ cut(age + 48, c(0,50,60,70,100)) + surgery, data = hearta, scale = 1) fit2 <- pyears(Surv(stop/365.25, event) ~ tcut(age + 48, c(0,50,60,70,100)) + surgery, data = hearta, scale = 1) fit1$event/fit1$pyears #death rates on the surgery and non-surg arm fit2$event/fit2$pyears #death rates on the surgery and non-surg arm } \keyword{survival} survival/man/nwtco.Rd0000644000175100001440000000237311732700061014341 0ustar hornikusers\name{nwtco} \alias{nwtco} \docType{data} \title{Data from the National Wilm's Tumor Study} \description{ Missing data/masurement error example. Tumor histology predicts survival, but prediction is stronger with central lab histology than with the local institution determination. } \usage{nwtco} \format{ A data frame with 4028 observations on the following 9 variables. \describe{ \item{\code{seqno}}{id number} \item{\code{instit}}{Histology from local institution} \item{\code{histol}}{Histology from central lab} \item{\code{stage}}{Disease stage} \item{\code{study}}{study} \item{\code{rel}}{indicator for relapse} \item{\code{edrel}}{time to relapse} \item{\code{age}}{age in months} \item{\code{in.subcohort}}{Included in the subcohort for the example in the paper} } } \source{ \url{http://faculty.washington.edu/norm/software.html} } \references{ NE Breslow and N Chatterjee (1999), Design and analysis of two-phase studies with binary outcome applied to Wilms tumour prognosis. \emph{Applied Statistics} \bold{48}, 457--68. } \examples{ with(nwtco, table(instit,histol)) anova(coxph(Surv(edrel,rel)~histol+instit,data=nwtco)) anova(coxph(Surv(edrel,rel)~instit+histol,data=nwtco)) } \keyword{datasets} survival/man/survfit.Rd0000644000175100001440000000273512312040337014711 0ustar hornikusers\name{survfit} \alias{survfit} \title{Create survival curves} \description{ This function creates survival curves from either a formula (e.g. the Kaplan-Meier), a previously fitted Cox model, or a previously fitted accelerated failure time model. } \usage{ survfit(formula, ...) } %- maybe also 'usage' for other objects documented here. \arguments{ \item{formula}{either a formula or a previously fitted model} \item{\dots}{other arguments to the specific method} } \details{ A survival curve is based on a tabulation of the number at risk and number of events at each unique death time. When time is a floating point number the definition of "unique" is subject to interpretation. The code uses factor() to define the set. For further details see the documentation for the appropriate method, i.e., \code{?survfit.formula} or \code{?survfit.coxph}. } \value{ An object of class \code{survfit} containing one or more survival curves. } \author{Terry Therneau} \note{Older releases of the code also allowed the specification for a single curve to omit the right hand of the formula, i.e., \code{~ 1}. Handling this case required some non-standard and fairly fragile manipulations, and this case is no longer supported. } \seealso{\code{\link{survfit.formula}}, \code{\link{survfit.coxph}}, \code{\link{survfit.object}}, \code{\link{print.survfit}}, \code{\link{plot.survfit}}, \code{\link{quantile.survfit}}, \code{\link{summary.survfit}}} \keyword{ survival} survival/man/model.matrix.coxph.Rd0000644000175100001440000000263311732700061016731 0ustar hornikusers\name{model.matrix.coxph} \Rdversion{1.1} \alias{model.matrix.coxph} \title{ Model.matrix method for coxph models } \description{ Reconstruct the model matrix for a cox model. } \usage{ \method{model.matrix}{coxph}(object, data=NULL, contrast.arg = object$contrasts, ...) } %- maybe also 'usage' for other objects documented here. \arguments{ \item{object}{the result of a \code{coxph} model} \item{data}{optional, a data frame from which to obtain the data} \item{contrast.arg}{optional, a contrasts object describing how factors should be coded} \item{\dots}{other possible argument to \code{model.frame}} } \details{ When there is a \code{data} argument this function differs from most of the other \code{model.matrix} methods in that the response variable for the original formual is \emph{not} required to be in the data. If the data frame contains a \code{terms} attribute then it is assumed to be the result of a call to \code{model.frame}, otherwise a call to \code{model.frame} is applied with the data as an argument. } \value{ The model matrix for the fit } \author{Terry Therneau} \seealso{\code{\link{model.matrix}}} \examples{ fit1 <- coxph(Surv(time, status) ~ age + factor(ph.ecog), data=lung) xfit <- model.matrix(fit1) fit2 <- coxph(Surv(time, status) ~ age + factor(ph.ecog), data=lung, x=TRUE) all.equal(model.matrix(fit1), fit2$x) } \keyword{ survival } survival/man/plot.aareg.Rd0000644000175100001440000000155111732700061015240 0ustar hornikusers\name{plot.aareg} \alias{plot.aareg} \title{ Plot an aareg object. } \description{ Plot the estimated coefficient function(s) from a fit of Aalen's additive regression model. } \usage{ \method{plot}{aareg}(x, se=TRUE, maxtime, type='s', ...) } \arguments{ \item{x}{ the result of a call to the \code{aareg} function } \item{se}{ if TRUE, standard error bands are included on the plot } \item{maxtime}{ upper limit for the x-axis. } \item{type}{ graphical parameter for the type of line, default is "steps". } \item{\dots }{ other graphical parameters such as line type, color, or axis labels. } } \section{Side Effects}{ A plot is produced on the current graphical device. } \section{References}{ Aalen, O.O. (1989). A linear regression model for the analysis of life times. Statistics in Medicine, 8:907-925. } \seealso{ aareg } survival/man/coxph.Rd0000644000175100001440000002510412516205523014331 0ustar hornikusers\name{coxph} \alias{coxph} \alias{vcov.coxph} \alias{print.coxph.null} \alias{print.coxph.penal} \alias{coxph.penalty} \alias{[.coxph.penalty} \alias{coxph.getdata} \alias{summary.coxph.penal} \title{ Fit Proportional Hazards Regression Model } \description{ Fits a Cox proportional hazards regression model. Time dependent variables, time dependent strata, multiple events per subject, and other extensions are incorporated using the counting process formulation of Andersen and Gill. } \usage{ coxph(formula, data=, weights, subset, na.action, init, control, ties=c("efron","breslow","exact"), singular.ok=TRUE, robust=FALSE, model=FALSE, x=FALSE, y=TRUE, tt, method, ...) } \arguments{ \item{formula}{ a formula object, with the response on the left of a \code{~} operator, and the terms on the right. The response must be a survival object as returned by the \code{Surv} function. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}, or in the \code{subset} and the \code{weights} argument. } \item{weights}{ vector of case weights. If \code{weights} is a vector of integers, then the estimated coefficients are equivalent to estimating the model from data with the individual cases replicated as many times as indicated by \code{weights}. } \item{subset}{ expression indicating which subset of the rows of data should be used in the fit. All observations are included by default. } \item{na.action}{ a missing-data filter function. This is applied to the model.frame after any subset argument has been used. Default is \code{options()\$na.action}. } \item{init}{ vector of initial values of the iteration. Default initial value is zero for all variables. } \item{control}{ Object of class \code{\link{coxph.control}} specifying iteration limit and other control options. Default is \code{coxph.control(...)}. } \item{ties}{ a character string specifying the method for tie handling. If there are no tied death times all the methods are equivalent. Nearly all Cox regression programs use the Breslow method by default, but not this one. The Efron approximation is used as the default here, it is more accurate when dealing with tied death times, and is as efficient computationally. The ``exact partial likelihood'' is equivalent to a conditional logistic model, and is appropriate when the times are a small set of discrete values. See further below. } \item{singular.ok}{ logical value indicating how to handle collinearity in the model matrix. If \code{TRUE}, the program will automatically skip over columns of the X matrix that are linear combinations of earlier columns. In this case the coefficients for such columns will be NA, and the variance matrix will contain zeros. For ancillary calculations, such as the linear predictor, the missing coefficients are treated as zeros. } \item{robust}{ this argument has been deprecated, use a cluster term in the model instead. (The two options accomplish the same goal -- creation of a robust variance -- but the second is more flexible). } \item{model}{ logical value: if \code{TRUE}, the model frame is returned in component \code{model}. } \item{x}{ logical value: if \code{TRUE}, the x matrix is returned in component \code{x}. } \item{y}{ logical value: if \code{TRUE}, the response vector is returned in component \code{y}. } \item{tt}{optional list of time-transform functions.} \item{method}{alternate name for the \code{ties} argument.} \item{...}{Other arguments will be passed to \code{\link{coxph.control}} } } \value{ an object of class \code{coxph} representing the fit. See \code{coxph.object} for details. } \section{Side Effects}{ Depending on the call, the \code{predict}, \code{residuals}, and \code{survfit} routines may need to reconstruct the x matrix created by \code{coxph}. It is possible for this to fail, as in the example below in which the predict function is unable to find \code{tform}. \preformatted{ tfun <- function(tform) coxph(tform, data=lung) fit <- tfun(Surv(time, status) ~ age) predict(fit)} In such a case add the \code{model=TRUE} option to the \code{coxph} call to obviate the need for reconstruction, at the expense of a larger \code{fit} object. } \details{ The proportional hazards model is usually expressed in terms of a single survival time value for each person, with possible censoring. Andersen and Gill reformulated the same problem as a counting process; as time marches onward we observe the events for a subject, rather like watching a Geiger counter. The data for a subject is presented as multiple rows or "observations", each of which applies to an interval of observation (start, stop]. The routine internally scales and centers data to avoid overflow in the argument to the exponential function. These actions do not change the result, but lead to more numerical stability. However, arguments to offset are not scaled since there are situations where a large offset value is a purposefully used. Users should not use normally allow large numeric offset values. } \section{Special terms}{ There are three special terms that may be used in the model equation. A \code{strata} term identifies a stratified Cox model; separate baseline hazard functions are fit for each strata. The \code{cluster} term is used to compute a robust variance for the model. The term \code{+ cluster(id)} where each value of \code{id} is unique is equivalent to specifying the \code{robust=T} argument. If the \code{id} variable is not unique, it is assumed that it identifies clusters of correlated observations. The robust estimate arises from many different arguments and thus has had many labels. It is variously known as the Huber sandwich estimator, White's estimate (linear models/econometrics), the Horvitz-Thompson estimate (survey sampling), the working independence variance (generalized estimating equations), the infinitesimal jackknife, and the Wei, Lin, Weissfeld (WLW) estimate. A time-transform term allows variables to vary dynamically in time. In this case the \code{tt} argument will be a function or a list of functions (if there are more than one tt() term in the model) giving the appropriate transform. See the examples below. } \section{Convergence}{ In certain data cases the actual MLE estimate of a coefficient is infinity, e.g., a dichotomous variable where one of the groups has no events. When this happens the associated coefficient grows at a steady pace and a race condition will exist in the fitting routine: either the log likelihood converges, the information matrix becomes effectively singular, an argument to exp becomes too large for the computer hardware, or the maximum number of interactions is exceeded. (Nearly always the first occurs.) The routine attempts to detect when this has happened, not always successfully. The primary consequence for he user is that the Wald statistic = coefficient/se(coefficient) is not valid in this case and should be ignored; the likelihood ratio and score tests remain valid however. } \section{Ties}{ There are three possible choices for handling tied event times. The Breslow approximation is the easiest to program and hence became the first option coded for almost all computer routines. It then ended up as the default option when other options were added in order to "maintain backwards compatability". The Efron option is more accurate if there are a large number of ties, and it is the default option here. In practice the number of ties is usually small, in which case all the methods are statistically indistinguishable. Using the "exact partial likelihood" approach the Cox partial likelihood is equivalent to that for matched logisitic regression. (The \code{clogit} function uses the \code{coxph} code to do the fit.) It is technically appropriate when the time scale is discrete and has only a few unique values, and some packages refer to this as the "discrete" option. There is also an "exact marginal likelihood" due to Prentice which is not implemented here. The calculation of the exact partial likelihood is numerically intense. Say for instance 15 of 180 subjects at risk had an event on day 7; then the code needs to compute sums over all 180-choose-15 > 10^43 different possible subsets of size 15. There is an efficient recursive algorithm for this task, but even with this the compuatation can be insufferably long. With (start, stop) data it is much worse since the recursion needs to start anew for each unique start time. } \section{Penalized regression}{ \code{coxph} can now maximise a penalised partial likelihood with arbitrary user-defined penalty. Supplied penalty functions include ridge regression (\link{ridge}), smoothing splines (\link{pspline}), and frailty models (\link{frailty}). } \references{ Andersen, P. and Gill, R. (1982). Cox's regression model for counting processes, a large sample study. \emph{Annals of Statistics} \bold{10}, 1100-1120. Therneau, T., Grambsch, P., Modeling Survival Data: Extending the Cox Model. Springer-Verlag, 2000. } \seealso{ \code{\link{cluster}}, \code{\link{strata}}, \code{\link{Surv}}, \code{\link{survfit}}, \code{\link{pspline}}, \code{\link{frailty}}, \code{\link{ridge}}. } \examples{ # Create the simplest test data set test1 <- list(time=c(4,3,1,1,2,2,3), status=c(1,1,1,0,1,1,0), x=c(0,2,1,1,1,0,0), sex=c(0,0,0,0,1,1,1)) # Fit a stratified model coxph(Surv(time, status) ~ x + strata(sex), test1) # Create a simple data set for a time-dependent model test2 <- list(start=c(1,2,5,2,1,7,3,4,8,8), stop=c(2,3,6,7,8,9,9,9,14,17), event=c(1,1,1,1,1,1,1,0,0,0), x=c(1,0,0,1,0,1,1,1,0,0)) summary(coxph(Surv(start, stop, event) ~ x, test2)) # # Create a simple data set for a time-dependent model # test2 <- list(start=c(1, 2, 5, 2, 1, 7, 3, 4, 8, 8), stop =c(2, 3, 6, 7, 8, 9, 9, 9,14,17), event=c(1, 1, 1, 1, 1, 1, 1, 0, 0, 0), x =c(1, 0, 0, 1, 0, 1, 1, 1, 0, 0) ) summary( coxph( Surv(start, stop, event) ~ x, test2)) # Fit a stratified model, clustered on patients bladder1 <- bladder[bladder$enum < 5, ] coxph(Surv(stop, event) ~ (rx + size + number) * strata(enum) + cluster(id), bladder1) # Fit a time transform model using current age coxph(Surv(time, status) ~ ph.ecog + tt(age), data=lung, tt=function(x,t,...) pspline(x + t/365.25)) } \keyword{survival} survival/man/logLik.coxph.Rd0000644000175100001440000000275212534312154015554 0ustar hornikusers\name{logLik.coxph} \alias{logLik.coxph} \alias{logLik.survreg} \title{logLik method for a Cox model} \description{The logLik function for survival models} \usage{ \method{logLik}{coxph}(object, ...) \method{logLik}{survreg}(object, ...) } \arguments{ \item{object}{the result of a \code{coxph} or \code{survreg} fit} \item{\dots}{optional arguments for other instances of the method} } \details{ The logLik function is used by summary functions in R such as \code{AIC}. For a Cox model, this method returns the partial likelihood. The number of degrees of freedom (df) used by the fit and the effective number of observations (nobs) are added as attributes. Per Raftery and others, the effective number of observations is the taken to be the number of events in the data set. For a \code{survreg} model the proper value for the effective number of observations is still an open question (at least to this author). For right censored data the approach of \code{logLik.coxph} is the possible the most sensible, but for interval censored observations the result is unclear. The code currently does not add a \emph{nobs} attribute. } \value{an object of class \code{logLik}} \references{ Robert E. Kass and Adrian E. Raftery (1995). "Bayes Factors". J. American Statistical Assoc. 90 (430): 791. Raftery A.E. (1995), "Bayesian Model Selection in Social Research", Sociological methodology, 111-196. } \seealso{\code{\link{logLik}}} \author{Terry Therneau} \keyword{ survival} survival/man/survfit.coxph.Rd0000644000175100001440000002420412504344712016033 0ustar hornikusers\name{survfit.coxph} \alias{survfit.coxph} \title{ Compute a Survival Curve from a Cox model } \description{ Computes the predicted survivor function for a Cox proportional hazards model. } \usage{ \method{survfit}{coxph}(formula, newdata, se.fit=TRUE, conf.int=.95, individual=FALSE, type,vartype, conf.type=c("log","log-log","plain","none"), censor=TRUE, id, na.action=na.pass, ...) } \arguments{ \item{formula}{ A \code{coxph} object. } \item{newdata}{ a data frame with the same variable names as those that appear in the \code{coxph} formula. It is also valid to use a vector, if the data frame would consist of a single row. The curve(s) produced will be representative of a cohort whose covariates correspond to the values in \code{newdata}. Default is the mean of the covariates used in the \code{coxph} fit. } \item{individual}{ This argument has been superseded by the \code{id} argument and is present only for backwards compatability. A logical value indicating whether each row of \code{newdata} represents a distinct individual (FALSE, the default), or if each row of the data frame represents different time epochs for only one individual (TRUE). In the former case the result will have one curve for each row in \code{newdata}, in the latter only a single curve will be produced. } \item{conf.int}{ the level for a two-sided confidence interval on the survival curve(s). Default is 0.95. } \item{se.fit}{ a logical value indicating whether standard errors should be computed. Default is \code{TRUE}. } \item{type,vartype}{ a character string specifying the type of survival curve. Possible values are \code{"aalen"}, \code{"efron"}, or \code{"kalbfleisch-prentice"} (only the first two characters are necessary). The default is to match the computation used in the Cox model. The Nelson-Aalen-Breslow estimate for \code{ties='breslow'}, the Efron estimate for \code{ties='efron'} and the Kalbfleisch-Prentice estimate for a discrete time model \code{ties='exact'}. Variance estimates are the Aalen-Link-Tsiatis, Efron, and Greenwood. The default will be the Efron estimate for \code{ties='efron'} and the Aalen estimate otherwise. } \item{conf.type}{ One of \code{"none"}, \code{"plain"}, \code{"log"} (the default), or \code{"log-log"}. Only enough of the string to uniquely identify it is necessary. The first option causes confidence intervals not to be generated. The second causes the standard intervals \code{curve +- k *se(curve)}, where k is determined from \code{conf.int}. The log option calculates intervals based on the cumulative hazard or log(survival). The last option bases intervals on the log hazard or log(-log(survival)). } \item{censor}{if FALSE time points at which there are no events (only censoring) are not included in the result.} \item{id}{optional variable name of subject identifiers. If this is present, then each group of rows with the same subject id represents the covariate path through time of a single subject, and the result will contain one curve per subject. If the \code{coxph} fit had strata then that must also be specified in \code{newdata}. If missing, then each individual row of \code{newdata} is presumed to represent a distinct subject and there will be \code{nrow(newdata)} times the number of strata curves in the result (one for each strata/subject combination). result.} \item{na.action}{the na.action to be used on the newdata argument} \item{\dots}{for future methods} } \value{ an object of class \code{"survfit"}. See \code{survfit.object} for details. Methods defined for survfit objects are \code{print}, \code{plot}, \code{lines}, and \code{points}. } \details{ Serious thought has been given to removing the default value for \code{newdata}, which is to use a single "psuedo" subject with covariate values equal to the means of the data set, since the resulting curve(s) almost never make sense. It remains due to an unwarranted attachment to the option shown by some users and by other packages. Two particularly egregious examples are factor variables and interactions. Suppose one were studying interspecies transmission of a virus, and the data set has a factor variable with levels ("pig", "chicken") and about equal numbers of observations for each. The ``mean'' covariate level will be 1/2 -- is this a flying pig? As to interactions assume data with sex coded as 0/1, ages ranging from 50 to 80, and a model with age*sex. The ``mean'' value for the age:sex interaction term will be about 30, a value that does not occur in the data. Users are strongly advised to use the newdata argument. When the original model contains time-dependent covariates, then the path of that covariate through time needs to be specified in order to obtain a predicted curve. This requires \code{newdata} to contain multiple lines for each hypothetical subject which gives the covariate values, time interval, and strata for each line (a subject can change strata), along with an \code{id} variable which demarks which rows belong to each subject. The time interval must have the same (start, stop, status) variables as the original model: although the status variable is not used and thus can be set to a dummy value of 0 or 1, it is necessary for the variables to be recognized as a \code{Surv} object. Last, although predictions with a time-dependent covariate path can be useful, it is very easy to create a prediction that is senseless. Users are encouraged to seek out a text that discusses the issue in detail. When a model contains strata but no time-dependent covariates the user of this routine has a choice. If newdata argument does not contain strata variables then the returned object will be a matrix of survival curves with one row for each strata in the model and one column for each row in newdata. (This is the historical behavior of the routine.) If newdata does contain strata variables, then the result will contain one curve per row of newdata, based on the indicated stratum of the original model. In the rare case of a model with strata by covariate interactions the strata variable must be included in newdata, the routine does not allow it to be omitted (predictions become too confusing). (Note that the model Surv(time, status) ~ age*strata(sex) expands internally to strata(sex) + age:sex; the sex variable is needed for the second term of the model.) When all the coefficients are zero, the Kalbfleisch-Prentice estimator reduces to the Kaplan-Meier, the Aalen estimate to the exponential of Nelson's cumulative hazard estimate, and the Efron estimate to the Fleming-Harrington estimate of survival. The variances of the curves from a Cox model are larger than these non-parametrec estimates, however, due to the variance of the coefficients. See \code{\link{survfit}} for more details about the counts (number of events, number at risk, etc.) The censor argument was fixed at FALSE in earlier versions of the code and not made available to the user. The default argument is sensible in most instances --- and causes the familiar + sign to appear on plots --- it is not sensible for time dependent covariates since it may lead to a large number of spurious marks. } \section{References}{ Fleming, T. H. and Harrington, D. P. (1984). Nonparametric estimation of the survival distribution in censored data. \emph{Comm. in Statistics} \bold{13}, 2469-86. Kablfleisch, J. D. and Prentice, R. L. (1980). \emph{The Statistical Analysis of Failure Time Data.} New York:Wiley. Link, C. L. (1984). Confidence intervals for the survival function using Cox's proportional hazards model with covariates. \emph{Biometrics} \bold{40}, 601-610. Therneau T and Grambsch P (2000), Modeling Survival Data: Extending the Cox Model, Springer-Verlag. Tsiatis, A. (1981). A large sample study of the estimate for the integrated hazard function in Cox's regression model for survival data. \emph{Annals of Statistics} \bold{9}, 93-108. } \seealso{ \code{\link{print.survfit}}, \code{\link{plot.survfit}}, \code{\link{lines.survfit}}, \code{\link{coxph}}, \code{\link{Surv}}, \code{\link{strata}}. } \examples{ #fit a Kaplan-Meier and plot it fit <- survfit(Surv(time, status) ~ x, data = aml) plot(fit, lty = 2:3) legend(100, .8, c("Maintained", "Nonmaintained"), lty = 2:3) #fit a Cox proportional hazards model and plot the #predicted survival for a 60 year old fit <- coxph(Surv(futime, fustat) ~ age, data = ovarian) plot(survfit(fit, newdata=data.frame(age=60)), xscale=365.25, xlab = "Years", ylab="Survival") # Here is the data set from Turnbull # There are no interval censored subjects, only left-censored (status=3), # right-censored (status 0) and observed events (status 1) # # Time # 1 2 3 4 # Type of observation # death 12 6 2 3 # losses 3 2 0 3 # late entry 2 4 2 5 # tdata <- data.frame(time =c(1,1,1,2,2,2,3,3,3,4,4,4), status=rep(c(1,0,2),4), n =c(12,3,2,6,2,4,2,0,2,3,3,5)) fit <- survfit(Surv(time, time, status, type='interval') ~1, data=tdata, weight=n) # # Time to progression/death for patients with monoclonal gammopathy # Competing risk curves (cumulative incidence) fit1 <- survfit(Surv(stop, event=='progression') ~1, data=mgus1, subset=(start==0)) fit2 <- survfit(Surv(stop, status) ~1, data=mgus1, subset=(start==0), etype=event) #competing risks # CI curves are always plotted from 0 upwards, rather than 1 down plot(fit2, fun='event', xscale=365.25, xmax=7300, mark.time=FALSE, col=2:3, xlab="Years post diagnosis of MGUS") lines(fit1, fun='event', xscale=365.25, xmax=7300, mark.time=FALSE, conf.int=FALSE) text(10, .4, "Competing Risk: death", col=3) text(16, .15,"Competing Risk: progression", col=2) text(15, .30,"KM:prog") } \keyword{survival} survival/man/residuals.coxph.Rd0000644000175100001440000000606311732700061016322 0ustar hornikusers\name{residuals.coxph} \alias{residuals.coxph.penal} \alias{residuals.coxph.null} \alias{residuals.coxph} \title{ Calculate Residuals for a `coxph' Fit } \description{ Calculates martingale, deviance, score or Schoenfeld residuals for a Cox proportional hazards model. } \usage{ \method{residuals}{coxph}(object, type=c("martingale", "deviance", "score", "schoenfeld", "dfbeta", "dfbetas", "scaledsch","partial"), collapse=FALSE, weighted=FALSE, ...) \method{residuals}{coxph.null}(object, type=c("martingale", "deviance","score","schoenfeld"), collapse=FALSE, weighted=FALSE, ...) } \arguments{ \item{object}{ an object inheriting from class \code{coxph}, representing a fitted Cox regression model. Typically this is the output from the \code{coxph} function. } \item{type}{ character string indicating the type of residual desired. Possible values are \code{"martingale"}, \code{"deviance"}, \code{"score"}, \code{"schoenfeld"}, "dfbeta"', \code{"dfbetas"}, and \code{"scaledsch"}. Only enough of the string to determine a unique match is required. } \item{collapse}{ vector indicating which rows to collapse (sum) over. In time-dependent models more than one row data can pertain to a single individual. If there were 4 individuals represented by 3, 1, 2 and 4 rows of data respectively, then \code{collapse=c(1,1,1, 2, 3,3, 4,4,4,4)} could be used to obtain per subject rather than per observation residuals. } \item{weighted}{ if \code{TRUE} and the model was fit with case weights, then the weighted residuals are returned. }\item{...}{other unused arguments}} \value{ For martingale and deviance residuals, the returned object is a vector with one element for each subject (without \code{collapse}). For score residuals it is a matrix with one row per subject and one column per variable. The row order will match the input data for the original fit. For Schoenfeld residuals, the returned object is a matrix with one row for each event and one column per variable. The rows are ordered by time within strata, and an attribute \code{strata} is attached that contains the number of observations in each strata. The scaled Schoenfeld residuals are used in the \code{cox.zph} function. The score residuals are each individual's contribution to the score vector. Two transformations of this are often more useful: \code{dfbeta} is the approximate change in the coefficient vector if that observation were dropped, and \code{dfbetas} is the approximate change in the coefficients, scaled by the standard error for the coefficients. } \section{NOTE}{ For deviance residuals, the status variable may need to be reconstructed. For score and Schoenfeld residuals, the X matrix will need to be reconstructed. } \references{ T. Therneau, P. Grambsch, and T. Fleming. "Martingale based residuals for survival models", \emph{Biometrika}, March 1990. } \seealso{ \code{\link{coxph}}} \examples{ fit <- coxph(Surv(start, stop, event) ~ (age + surgery)* transplant, data=heart) mresid <- resid(fit, collapse=heart$id) } \keyword{survival} % Converted by Sd2Rd version 0.3-2. survival/man/coxph.control.Rd0000644000175100001440000000227711732700061016012 0ustar hornikusers\name{coxph.control} \alias{coxph.control} \title{Ancillary arguments for controling coxph fits} \description{ This is used to set various numeric parameters controling a Cox model fit. Typically it would only be used in a call to \code{coxph}. } \usage{ coxph.control(eps = 1e-09, toler.chol = .Machine$double.eps^0.75, iter.max = 20, toler.inf = sqrt(eps), outer.max = 10) } \arguments{ \item{eps}{Iteration continues until the relative change in the log partial likelihood is less than eps. Must be positive.} \item{toler.chol}{Tolerance for detection of singularity during a Cholesky decomposion of the variance matrix, i.e., for detecting a redundant predictor variable.} \item{iter.max}{Maximum number of iterations to attempt for convergence.} \item{toler.inf}{Tolerance criteria for the warning message about a possible infinite coefficient value.} \item{outer.max}{For a penalized coxph model, e.g. with pspline terms, there is an outer loop of iteration to determine the penalty parameters; maximum number of iterations for this outer loop.} } \value{ a list containing the values of each of the above constants } \author{Terry Therneau } \seealso{\code{\link{coxph}} } \keyword{survival} survival/man/cox.zph.Rd0000644000175100001440000000464411732700061014603 0ustar hornikusers\name{cox.zph} \alias{cox.zph} \alias{[.cox.zph} \alias{print.cox.zph} \title{ Test the Proportional Hazards Assumption of a Cox Regression } \description{ Test the proportional hazards assumption for a Cox regression model fit (\code{coxph}). } \usage{ cox.zph(fit, transform="km", global=TRUE) } \arguments{ \item{fit}{ the result of fitting a Cox regression model, using the \code{coxph} function. } \item{transform}{ a character string specifying how the survival times should be transformed before the test is performed. Possible values are \code{"km"}, \code{"rank"}, \code{"identity"} or a function of one argument. } \item{global}{ should a global chi-square test be done, in addition to the per-variable tests. } } \value{ an object of class \code{"cox.zph"}, with components: \item{table}{ a matrix with one row for each variable, and optionally a last row for the global test. Columns of the matrix contain the correlation coefficient between transformed survival time and the scaled Schoenfeld residuals, a chi-square, and the two-sided p-value. For the global test there is no appropriate correlation, so an NA is entered into the matrix as a placeholder. } \item{x}{ the transformed time axis. } \item{y}{ the matrix of scaled Schoenfeld residuals. There will be one column per variable and one row per event. The row labels contain the original event times (for the identity transform, these will be the same as \code{x}). } \item{call}{ the calling sequence for the routine. The computations require the original \code{x} matrix of the Cox model fit. Thus it saves time if the \code{x=TRUE} option is used in \code{coxph}. This function would usually be followed by both a plot and a print of the result. The plot gives an estimate of the time-dependent coefficient \code{beta(t)}. If the proportional hazards assumption is true, \code{beta(t)} will be a horizontal line. The printout gives a test for \code{slope=0}. } } \references{ P. Grambsch and T. Therneau (1994), Proportional hazards tests and diagnostics based on weighted residuals. \emph{Biometrika,} \bold{81}, 515-26. } \seealso{ \code{\link{coxph}}, \code{\link{Surv}}. } \examples{ fit <- coxph(Surv(futime, fustat) ~ age + ecog.ps, data=ovarian) temp <- cox.zph(fit) print(temp) # display the results plot(temp) # plot curves } \keyword{survival} survival/man/print.summary.survexp.Rd0000644000175100001440000000113211732700061017542 0ustar hornikusers\name{print.summary.survexp} \alias{print.summary.survexp} \title{Print Survexp Summary} \description{ Prints the results of \code{summary.survexp} } \usage{ \method{print}{summary.survexp}(x, digits = max(options()$digits - 4, 3), ...) } \arguments{ \item{x}{ an object of class \code{summary.survexp}. } \item{digits}{ the number of digits to use in printing the result. } \item{\dots}{for future methods} } \value{ \code{x}, with the invisible flag set to prevent further printing. } \author{Terry Therneau} \seealso{\code{link{summary.survexp}}, \code{\link{survexp}}} \keyword{ survival } survival/man/tcut.Rd0000644000175100001440000000225112265342777014203 0ustar hornikusers\name{tcut} \alias{tcut} \alias{[.tcut} \alias{levels.tcut} \title{Factors for person-year calculations} \description{ Attaches categories for person-year calculations to a variable without losing the underlying continuous representation } \usage{ tcut(x, breaks, labels, scale=1) \method{levels}{tcut}(x) } %- maybe also `usage' for other objects documented here. \arguments{ \item{x}{numeric/date variable } \item{breaks}{breaks between categories, which are right-continuous } \item{labels}{labels for categories } \item{scale}{Multiply \code{x} and \code{breaks} by this.} } \value{ An object of class \code{tcut} } \seealso{ \code{\link{cut}}, \code{\link{pyears}} } \examples{ mdy.date <- function(m,d,y) as.Date(paste(ifelse(y<100, y+1900, y), m, d, sep='/')) temp1 <- mdy.date(6,6,36) temp2 <- mdy.date(6,6,55)# Now compare the results from person-years # temp.age <- tcut(temp2-temp1, floor(c(-1, (18:31 * 365.24))), labels=c('0-18', paste(18:30, 19:31, sep='-'))) temp.yr <- tcut(temp2, mdy.date(1,1,1954:1965), labels=1954:1964) temp.time <- 3700 #total days of fu py1 <- pyears(temp.time ~ temp.age + temp.yr, scale=1) #output in days py1 } \keyword{survival} survival/man/aml.Rd0000644000175100001440000000116111732700061013752 0ustar hornikusers\name{aml} \docType{data} \alias{aml} \alias{leukemia} \title{Acute Myelogenous Leukemia survival data} \description{Survival in patients with Acute Myelogenous Leukemia. The question at the time was whether the standard course of chemotherapy should be extended ('maintainance') for additional cycles.} \usage{ aml leukemia } \format{ \tabular{ll}{ time:\tab survival or censoring time\cr status:\tab censoring status\cr x: \tab maintenance chemotherapy given? (factor)\cr } } \source{ Rupert G. Miller (1997), \emph{Survival Analysis}. John Wiley & Sons. ISBN: 0-471-25218-2. } \keyword{datasets} survival/man/pbc.Rd0000644000175100001440000000434611732700061013755 0ustar hornikusers\name{pbc} \alias{pbc} \docType{data} \title{Mayo Clinic Primary Biliary Cirrhosis Data} \description{D This data is from the Mayo Clinic trial in primary biliary cirrhosis (PBC) of the liver conducted between 1974 and 1984. A total of 424 PBC patients, referred to Mayo Clinic during that ten-year interval, met eligibility criteria for the randomized placebo controlled trial of the drug D-penicillamine. The first 312 cases in the data set participated in the randomized trial and contain largely complete data. The additional 112 cases did not participate in the clinical trial, but consented to have basic measurements recorded and to be followed for survival. Six of those cases were lost to follow-up shortly after diagnosis, so the data here are on an additional 106 cases as well as the 312 randomized participants. A nearly identical data set found in appendix D of Fleming and Harrington; this version has fewer missing values. } \usage{pbc} \format{ \tabular{ll}{ age:\tab in years\cr albumin:\tab serum albumin (g/dl)\cr alk.phos:\tab alkaline phosphotase (U/liter)\cr ascites:\tab presence of ascites \cr ast:\tab aspartate aminotransferase, once called SGOT (U/ml)\cr bili:\tab serum bilirunbin (mg/dl)\cr chol:\tab serum cholesterol (mg/dl)\cr copper:\tab urine copper (ug/day)\cr edema:\tab 0 no edema, 0.5 untreated or successfully treated\cr \tab 1 edema despite diuretic therapy\cr hepato:\tab presence of hepatomegaly or enlarged liver\cr id:\tab case number\cr platelet:\tab platelet count\cr protime:\tab standardised blood clotting time\cr sex:\tab m/f\cr spiders:\tab blood vessel malformations in the skin\cr stage:\tab histologic stage of disease (needs biopsy)\cr status:\tab status at endpoint, 0/1/2 for censored, transplant, dead\cr time: \tab number of days between registration and the earlier of death,\cr \tab transplantion, or study analysis in July, 1986\cr trt:\tab 1/2/NA for D-penicillmain, placebo, not randomised\cr trig:\tab triglycerides (mg/dl)\cr } } \source{ T Therneau and P Grambsch (2000), \emph{Modeling Survival Data: Extending the Cox Model}, Springer-Verlag, New York. ISBN: 0-387-98784-3. } \keyword{datasets} survival/man/Surv.Rd0000644000175100001440000001334612257041540014153 0ustar hornikusers\name{Surv} \alias{Surv} \alias{is.Surv} \alias{print.Surv} \alias{Math.Surv} \alias{Summary.Surv} \alias{[.Surv} \alias{format.Surv} \alias{as.data.frame.Surv} \alias{as.character.Surv} \alias{as.matrix.Surv} \alias{is.na.Surv} \alias{Ops.Surv} \title{ Create a Survival Object } \description{ Create a survival object, usually used as a response variable in a model formula. Argument matching is special for this function, see Details below. } \usage{ Surv(time, time2, event, type=c('right', 'left', 'interval', 'counting', 'interval2', 'mstate'), origin=0) is.Surv(x) } \arguments{ \item{time}{ for right censored data, this is the follow up time. For interval data, the first argument is the starting time for the interval. } \item{event}{ The status indicator, normally 0=alive, 1=dead. Other choices are \code{TRUE}/\code{FALSE} (\code{TRUE} = death) or 1/2 (2=death). For interval censored data, the status indicator is 0=right censored, 1=event at \code{time}, 2=left censored, 3=interval censored. Although unusual, the event indicator can be omitted, in which case all subjects are assumed to have an event. } \item{time2}{ ending time of the interval for interval censored or counting process data only. Intervals are assumed to be open on the left and closed on the right, \code{(start, end]}. For counting process data, \code{event} indicates whether an event occurred at the end of the interval. } \item{type}{ character string specifying the type of censoring. Possible values are \code{"right"}, \code{"left"}, \code{"counting"}, \code{"interval"}, \code{"interval2"} or \code{"mstate"}. } \item{origin}{ for counting process data, the hazard function origin. This option was intended to be used in conjunction with a model containing time dependent strata in order to align the subjects properly when they cross over from one strata to another, but it has rarely proven useful.} \item{x}{ any R object. } } \value{ An object of class \code{Surv}. There are methods for \code{print}, \code{is.na}, and subscripting survival objects. \code{Surv} objects are implemented as a matrix of 2 or 3 columns that has further attributes. These include the type (left censored, right censored, counting process, etc.) and labels for the states for multi-state objects. Any attributes of the input arguments are also preserved in \code{inputAttributes}. This may be useful for other packages that have attached further information to data items such as labels; none of the routines in the survival package make use of these values, however. In the case of \code{is.Surv}, a logical value \code{TRUE} if \code{x} inherits from class \code{"Surv"}, otherwise an \code{FALSE}. } \details{ When the \code{type} argument is missing the code assumes a type based on the following rules: \itemize{ \item If there are two unnamed arguments, they will match \code{time} and \code{event} in that order. If there are three unnamed arguments they match \code{time}, \code{time2} and \code{event}. \item If the event variable is a factor then type \code{mstate} is assumed. Otherwise type \code{right} if there is no \code{time2} argumement, and type \code{counting} if there is. } As a consequence the \code{type} argument can usually be omitted. When the survival type is "mstate" then the status variable will be treated as a factor. The first level of the factor is taken to represent censoring and remaining ones a transition to the given state. Interval censored data can be represented in two ways. For the first use \code{type = "interval"} and the codes shown above. In that usage the value of the \code{time2} argument is ignored unless event=3. The second approach is to think of each observation as a time interval with (-infinity, t) for left censored, (t, infinity) for right censored, (t,t) for exact and (t1, t2) for an interval. This is the approach used for type = interval2. Infinite values can be represented either by actual infinity (Inf) or NA. The second form has proven to be the more useful one. Presently, the only methods allowing interval censored data are the parametric models computed by \code{survreg} and survival curves computed by \code{survfit}; for both of these, the distinction between open and closed intervals is unimportant. The distinction is important for counting process data and the Cox model. The function tries to distinguish between the use of 0/1 and 1/2 coding for censored data via the condition \code{if (max(status)==2)}. If 1/2 coding is used and all the subjects are censored, it will guess wrong. In any questionable case it is safer to use logical coding, e.g., \code{Surv(time, status==3)} would indicate that a \code{3} is the code for an event. For multi-state survival (type= "mstate") the status variable can have multiple levels. The first of these will stand for censoring, and the others for various event types, e.g., causes of death. Surv objects can be subscripted either as a vector, e.g. \code{x[1:3]} using a single subscript, in which case the \code{drop} argument is ignored and the result will be a survival object; or as a matrix by using two subscripts. If the second subscript is missing and \code{drop=F} (the default), the result of the subscripting will be a Surv object, e.g., \code{x[1:3,,drop=F]}, otherwise the result will be a matrix (or vector), in accordance with the default behavior for subscripting matrices. } \seealso{ \code{\link{coxph}}, \code{\link{survfit}}, \code{\link{survreg}}. } \examples{ with(lung, Surv(time, status)) Surv(heart$start, heart$stop, heart$event) } \keyword{survival} survival/man/coxph.object.Rd0000644000175100001440000000511711732700061015574 0ustar hornikusers\name{coxph.object} \alias{coxph.object} \alias{extractAIC.coxph.penal} \alias{print.coxph} \title{ Proportional Hazards Regression Object } \description{ This class of objects is returned by the \code{coxph} class of functions to represent a fitted proportional hazards model. Objects of this class have methods for the functions \code{print}, \code{summary}, \code{residuals}, \code{predict} and \code{survfit}. } \section{Components}{ The following components must be included in a legitimate \code{coxph} object. } \arguments{ \item{coefficients}{ the vector of coefficients. If the model is over-determined there will be missing values in the vector corresponding to the redundant columns in the model matrix. } \item{var}{ the variance matrix of the coefficients. Rows and columns corresponding to any missing coefficients are set to zero. } \item{naive.var}{ this component will be present only if the \code{robust} option was true. If so, the \code{var} component will contain the robust estimate of variance, and this component will contain the ordinary estimate. } \item{loglik}{ a vector of length 2 containing the log-likelihood with the initial values and with the final values of the coefficients. } \item{score}{ value of the efficient score test, at the initial value of the coefficients. } \item{rscore}{ the robust log-rank statistic, if a robust variance was requested. } \item{wald.test}{ the Wald test of whether the final coefficients differ from the initial values. } \item{iter}{ number of iterations used. } \item{linear.predictors}{ the vector of linear predictors, one per subject. Note that this vector has been centered, see \code{predict.coxph} for more details. } \item{residuals}{ the martingale residuals. } \item{means}{ vector of column means of the X matrix. Subsequent survival curves are adjusted to this value. } \item{n}{ the number of observations used in the fit. } \item{nevent}{ the number of events (usually deaths) used in the fit. } \item{weights}{ the vector of case weights, if one was used. } \item{method}{ the computation method used. } \item{na.action}{ the na.action attribute, if any, that was returned by the \code{na.action} routine. The object will also contain the following, for documentation see the \code{lm} object: \code{terms}, \code{assign}, \code{formula}, \code{call}, and, optionally, \code{x}, \code{y}, and/or \code{frame}. } } \seealso{ \code{\link{coxph}}, \code{\link{coxph.detail}}, \code{\link{cox.zph}}, \code{\link{residuals.coxph}}, \code{\link{survfit}}, \code{\link{survreg}}. } \keyword{survival} survival/man/veteran.Rd0000644000175100001440000000137611732700061014655 0ustar hornikusers\name{veteran} \alias{veteran} \docType{data} \title{Veterans' Administration Lung Cancer study} \description{Randomised trial of two treatment regimens for lung cancer. This is a standard survival analysis data set.} \usage{veteran} \format{ \tabular{ll}{ trt:\tab 1=standard 2=test\cr celltype:\tab 1=squamous, 2=smallcell, 3=adeno, 4=large\cr time:\tab survival time\cr status:\tab censoring status\cr karno:\tab Karnofsky performance score (100=good)\cr diagtime:\tab months from diagnosis to randomisation\cr age:\tab in years\cr prior:\tab prior therapy 0=no, 1=yes\cr } } \source{ D Kalbfleisch and RL Prentice (1980), \emph{The Statistical Analysis of Failure Time Data}. Wiley, New York. } \keyword{datasets} survival/man/coxph.wtest.Rd0000644000175100001440000000123012141031547015465 0ustar hornikusers\name{coxph.wtest} \alias{coxph.wtest} \title{Compute a quadratic form} \description{ This function is used internally by several survival routines. It computes a simple quadratic form, while properly dealing with missings. } \usage{ coxph.wtest(var, b, toler.chol = 1e-09) } \arguments{ \item{var}{variance matrix} \item{b}{vector} \item{toler.chol}{tolerance for the internal choelsky decomposition} } \details{ Compute b' V-inverse b. Equivalent to sum(b * solve(V,b)), except for the case of redundant covariates in the original model, which lead to NA values in V and b. } \value{a real number} \author{Terry Therneau} \keyword{ survival } survival/man/survobrien.Rd0000644000175100001440000000567211732700061015412 0ustar hornikusers\name{survobrien} \alias{survobrien} \title{ O'Brien's Test for Association of a Single Variable with Survival } \description{ Peter O'Brien's test for association of a single variable with survival This test is proposed in Biometrics, June 1978. } \usage{ survobrien(formula, data, subset, na.action, transform) } \arguments{ \item{formula}{ a valid formula for a cox model. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}, or in the \code{subset} and the \code{weights} argument. } \item{subset}{ expression indicating which subset of the rows of data should be used in the fit. All observations are included by default. } \item{na.action}{ a missing-data filter function. This is applied to the model.frame after any subset argument has been used. Default is \code{options()\$na.action}. } \item{transform}{the transformation function to be applied at each time point. The default is O'Brien's suggestion logit(tr) where tr = (rank(x)- 1/2)/ length(x) is the rank shifted to the range 0-1 and logit(x) = log(x/(1-x)) is the logit transform. }} \value{ a new data frame. The response variables will be column names returned by the \code{Surv} function, i.e., "time" and "status" for simple survival data, or "start", "stop", "status" for counting process data. Each individual event time is identified by the value of the variable \code{.strata.}. Other variables retain their original names. If a predictor variable is a factor or is protected with \code{I()}, it is retained as is. Other predictor variables have been replaced with time-dependent logit scores. The new data frame will have many more rows that the original data, approximately the original number of rows * number of deaths/2. } \section{Method}{ A time-dependent cox model can now be fit to the new data. The univariate statistic, as originally proposed, is equivalent to single variable score tests from the time-dependent model. This equivalence is the rationale for using the time dependent model as a multivariate extension of the original paper. In O'Brien's method, the x variables are re-ranked at each death time. A simpler method, proposed by Prentice, ranks the data only once at the start. The results are usually similar. } \references{ O'Brien, Peter, "A Nonparametric Test for Association with Censored Data", \emph{Biometrics} 34: 243-250, 1978. } \note{ A prior version of the routine returned new time variables rather than a strata. Unfortunately, that strategy does not work if the original formula has a strata statement. This new data set will be the same size, but the \code{coxph} routine will process it slightly faster. } \seealso{ \code{\link{survdiff}} } \keyword{survival} \examples{ xx <- survobrien(Surv(futime, fustat) ~ age + factor(rx) + I(ecog.ps), data=ovarian) coxph(Surv(time, status) ~ age + strata(.strata.), data=xx) } survival/man/basehaz.Rd0000644000175100001440000000263312466507346014643 0ustar hornikusers\name{basehaz} \alias{basehaz} \title{Alias for the survfit function} \description{ Compute the predicted survival curve for a Cox model. } \usage{ basehaz(fit, centered=TRUE) } \arguments{ \item{fit}{a coxph fit} \item{centered}{if TRUE return data from a predicted survival curve at the mean values of the covariates \code{fit$mean}, if FALSE return a prediction for all covariates equal to zero.} } \details{ This function is simply an alias for \code{survfit}, which is the actual function that does all the computations. See the manual page for that function for the preferred use. This function survives only for backwards support of prior usage. The function returns a data frame containing the \code{time}, \code{cumhaz} and optionally the strata (if the fitted Cox model used a strata statement), which are copied the \code{survfit} result. If there are factor variables in the model, then the default predictions at the "mean" are meaningless since they do not correspond to any possible subject; correct results require use of the \code{newdata} argument of survfit. Results for all covariates =0 are normally only of use as a building block for further calcuations. } \value{ a data frame with variable names of \code{hazard}, \code{time} and optionally \code{strata}. The first is actually the cumulative hazard. } \seealso{\code{\link{survfit.coxph}}} \keyword{survival } survival/man/aareg.Rd0000644000175100001440000001713011740632003014262 0ustar hornikusers\name{aareg} \alias{aareg} \title{ Aalen's additive regression model for censored data } \description{ Returns an object of class \code{"aareg"} that represents an Aalen model. } \usage{ aareg(formula, data, weights, subset, na.action, qrtol=1e-07, nmin, dfbeta=FALSE, taper=1, test = c('aalen', 'variance', 'nrisk'), model=FALSE, x=FALSE, y=FALSE) } \arguments{ \item{formula}{ a formula object, with the response on the left of a `~' operator and the terms, separated by \code{+} operators, on the right. The response must be a \code{Surv} object. Due to a particular computational approach that is used, the model MUST include an intercept term. If "-1" is used in the model formula the program will ignore it. } \item{data}{ data frame in which to interpret the variables named in the \code{formula}, \code{subset}, and \code{weights} arguments. This may also be a single number to handle some speci al cases -- see below for details. If \code{data} is missi ng, the variables in the model formula should be in the search path. } \item{weights}{ vector of observation weights. If supplied, the fitting algorithm minimizes the sum of the weights multiplied by the squared residuals (see below for additional technical details). The length of \code{weights} must be the same as the number of observations. The weights must be nonnegative and it i s recommended that they be strictly positive, since zero weights are ambiguous. To exclude particular observations from the model, use the \code{subset} argument instead of zero weights. } \item{subset}{ expression specifying which subset of observations should be used in the fit. Th is can be a logical vector (which is replicated to have length equal to the numb er of observations), a numeric vector indicating the observation numbers to be i ncluded, or a character vector of the observation names that should be included. All observations are included by default. } \item{na.action}{ a function to filter missing data. This is applied to the \code{model.fr ame} after any \code{subset} argument has be en applied. The default is \code{na.fail}, which returns a n error if any missing values are found. An alternative is \code{na.excl ude}, which deletes observations that contain one or more missing values. } \item{qrtol}{ tolerance for detection of singularity in the QR decomposition } \item{nmin}{ minimum number of observations for an estimate; defaults to 3 times the number of covariates. This essentially truncates the computations near the tail of the data set, when n is small and the calcualtions can become numerically unstable. } \item{dfbeta}{ should the array of dfbeta residuals be computed. This implies computation of the sandwich variance estimate. The residuals will always be computed if there is a \code{cluster} term in the model formula. } \item{taper}{ allows for a smoothed variance estimate. Var(x), where x is the set of covariates, is an important component of the calculations for the Aalen regression model. At any given time point t, it is computed over all subjects who are still at risk at time t. The tape argument allows smoothing these estimates, for example \code{taper=(1:4)/4} would cause the variance estimate used at any event time to be a weighted average of the estimated variance matrices at the last 4 death times, with a weight of 1 for the current death time and decreasing to 1/4 for prior event times. The default value gives the standard Aalen model. } \item{test}{ selects the weighting to be used, for computing an overall ``average'' coefficient vector over time and the subsequent test for equality to zero. } \item{model, x, y }{ should copies of the model frame, the x matrix of predictors, or the response vector y be included in the saved result. } } \value{ an object of class \code{"aareg"} representing the fit, with the following components: \item{n}{vector containing the number of observations in the data set, the number of event times, and the number of event times used in the computation} \item{times}{vector of sorted event times, which may contain duplicates} \item{nrisk}{vector containing the number of subjects at risk, of the same length as \code{times}} \item{coefficient}{matrix of coefficients, with one row per event and one column per covariate} \item{test.statistic}{the value of the test statistic, a vector with one element per covariate} \item{test.var}{variance-covariance matrix for the test} \item{test}{the type of test; a copy of the \code{test} argument above} \item{tweight}{matrix of weights used in the computation, one row per event} \item{call}{a copy of the call that produced this result} } \details{ The Aalen model assumes that the cumulative hazard H(t) for a subject can be expressed as a(t) + X B(t), where a(t) is a time-dependent intercept term, X is the vector of covariates for the subject (possibly time-dependent), and B(t) is a time-dependent matrix of coefficients. The estimates are inheritly non-parametric; a fit of the model will normally be followed by one or more plots of the estimates. The estimates may become unstable near the tail of a data set, since the increment to B at time t is based on the subjects still at risk at time t. The tolerance and/or nmin parameters may act to truncate the estimate before the last death. The \code{taper} argument can also be used to smooth out the tail of the curve. In practice, the addition of a taper such as 1:10 appears to have little effect on death times when n is still reasonably large, but can considerably dampen wild occilations in the tail of the plot. } \section{References}{ Aalen, O.O. (1989). A linear regression model for the analysis of life times. Statistics in Medicine, 8:907-925. Aalen, O.O (1993). Further results on the non-parametric linear model in survival analysis. Statistics in Medicine. 12:1569-1588. } \seealso{ print.aareg, summary.aareg, plot.aareg } \examples{ # Fit a model to the lung cancer data set lfit <- aareg(Surv(time, status) ~ age + sex + ph.ecog, data=lung, nmin=1) \dontrun{ lfit Call: aareg(formula = Surv(time, status) ~ age + sex + ph.ecog, data = lung, nmin = 1 ) n=227 (1 observations deleted due to missing values) 138 out of 138 unique event times used slope coef se(coef) z p Intercept 5.26e-03 5.99e-03 4.74e-03 1.26 0.207000 age 4.26e-05 7.02e-05 7.23e-05 0.97 0.332000 sex -3.29e-03 -4.02e-03 1.22e-03 -3.30 0.000976 ph.ecog 3.14e-03 3.80e-03 1.03e-03 3.70 0.000214 Chisq=26.73 on 3 df, p=6.7e-06; test weights=aalen plot(lfit[4], ylim=c(-4,4)) # Draw a plot of the function for ph.ecog } lfit2 <- aareg(Surv(time, status) ~ age + sex + ph.ecog, data=lung, nmin=1, taper=1:10) \dontrun{lines(lfit2[4], col=2) # Nearly the same, until the last point} # A fit to the mulitple-infection data set of children with # Chronic Granuomatous Disease. See section 8.5 of Therneau and Grambsch. fita2 <- aareg(Surv(tstart, tstop, status) ~ treat + age + inherit + steroids + cluster(id), data=cgd) \dontrun{ n= 203 69 out of 70 unique event times used slope coef se(coef) robust se z p Intercept 0.004670 0.017800 0.002780 0.003910 4.55 5.30e-06 treatrIFN-g -0.002520 -0.010100 0.002290 0.003020 -3.36 7.87e-04 age -0.000101 -0.000317 0.000115 0.000117 -2.70 6.84e-03 inheritautosomal 0.001330 0.003830 0.002800 0.002420 1.58 1.14e-01 steroids 0.004620 0.013200 0.010600 0.009700 1.36 1.73e-01 Chisq=16.74 on 4 df, p=0.0022; test weights=aalen } } \keyword{survival} % docclass is function survival/man/plot.cox.zph.Rd0000644000175100001440000000326711732700061015560 0ustar hornikusers\name{plot.cox.zph} \alias{plot.cox.zph} \title{ Graphical Test of Proportional Hazards } \description{ Displays a graph of the scaled Schoenfeld residuals, along with a smooth curve. } \usage{ \method{plot}{cox.zph}(x, resid=TRUE, se=TRUE, df=4, nsmo=40, var, \dots) } \arguments{ \item{x}{ result of the \code{cox.zph} function. } \item{resid}{ a logical value, if \code{TRUE} the residuals are included on the plot, as well as the smooth fit. } \item{se}{ a logical value, if \code{TRUE}, confidence bands at two standard errors will be added. } \item{df}{ the degrees of freedom for the fitted natural spline, \code{df=2} leads to a linear fit. } \item{nsmo}{ number of points used to plot the fitted spline. } \item{var}{ the set of variables for which plots are desired. By default, plots are produced in turn for each variable of a model. Selection of a single variable allows other features to be added to the plot, e.g., a horizontal line at zero or a main title. This has been superseded by a subscripting method; see the example below. } \item{\dots}{ additional graphical arguments passed to the \code{plot} function. } } \section{Side Effects}{ a plot is produced on the current graphics device. } \seealso{ \code{\link{coxph}}, \code{\link{cox.zph}}. } \examples{ vfit <- coxph(Surv(time,status) ~ trt + factor(celltype) + karno + age, data=veteran, x=TRUE) temp <- cox.zph(vfit) plot(temp, var=5) # Look at Karnofsy score, old way of doing plot plot(temp[5]) # New way with subscripting abline(0, 0, lty=3) # Add the linear fit as well abline(lm(temp$y[,5] ~ temp$x)$coefficients, lty=4, col=3) title(main="VA Lung Study") } \keyword{survival} survival/man/dsurvreg.Rd0000644000175100001440000000630312300663224015046 0ustar hornikusers\name{dsurvreg} \alias{dsurvreg} \alias{psurvreg} \alias{qsurvreg} \alias{rsurvreg} \title{ Distributions available in survreg. } \description{ Density, cumulative distribution function, quantile function and random generation for the set of distributions supported by the \code{survreg} function. } \usage{ dsurvreg(x, mean, scale=1, distribution='weibull', parms) psurvreg(q, mean, scale=1, distribution='weibull', parms) qsurvreg(p, mean, scale=1, distribution='weibull', parms) rsurvreg(n, mean, scale=1, distribution='weibull', parms) } \arguments{ \item{x}{ vector of quantiles. Missing values (\code{NA}s) are allowed. } \item{q}{ vector of quantiles. Missing values (\code{NA}s) are allowed. } \item{p}{ vector of probabilities. Missing values (\code{NA}s) are allowed. } \item{n}{number of random deviates to produce} \item{mean}{vector of linear predictors for the model. This is replicated to be the same length as \code{p}, \code{q} or \code{n}. } \item{scale}{ vector of (positive) scale factors. This is replicated to be the same length as \code{p}, \code{q} or \code{n}. } \item{distribution}{ character string giving the name of the distribution. This must be one of the elements of \code{survreg.distributions} } \item{parms}{ optional parameters, if any, of the distribution. For the t-distribution this is the degrees of freedom. } } \value{ density (\code{dsurvreg}), probability (\code{psurvreg}), quantile (\code{qsurvreg}), or for the requested distribution with mean and scale parameters \code{mean} and \code{sd}. } \details{ Elements of \code{q} or \code{p} that are missing will cause the corresponding elements of the result to be missing. The \code{location} and \code{scale} values are as they would be for \code{survreg}. The label "mean" was an unfortunate choice (made in mimicry of qnorm); since almost none of these distributions are symmetric it will not actually be a mean, but corresponds instead to the linear predictor of a fitted model. Translation to the usual parameterization found in a textbook is not always obvious. For example, the Weibull distribution is fit using the Extreme value distribution along with a log transformation. Letting \eqn{F(t) = 1 - \exp[-(at)^p]}{F(t) = 1 - exp(-(at)^p)} be the cumulative distribution of the Weibull using a standard parameterization in terms of \eqn{a} and \eqn{p}, the survreg location corresponds to \eqn{-\log(a)}{-log(a)} and the scale to \eqn{1/p} (Kalbfleish and Prentice, section 2.2.2). } \section{References}{ Kalbfleish, J. D. and Prentice, R. L. (1970). \emph{The Statistical Analysis of Failure Time Data} Wiley, New York. } \seealso{ \code{\link{survreg}}, \code{\link{Normal}} } \examples{ # List of distributions available names(survreg.distributions) \dontrun{ [1] "extreme" "logistic" "gaussian" "weibull" "exponential" [6] "rayleigh" "loggaussian" "lognormal" "loglogistic" "t" } # Compare results all.equal(dsurvreg(1:10, 2, 5, dist='lognormal'), dlnorm(1:10, 2, 5)) # Hazard function for a Weibull distribution x <- seq(.1, 3, length=30) haz <- dsurvreg(x, 2, 3)/ (1-psurvreg(x, 2, 3)) \dontrun{ plot(x, haz, log='xy', ylab="Hazard") #line with slope (1/scale -1) } } \keyword{distribution} survival/man/survSplit.Rd0000644000175100001440000000361211732700061015217 0ustar hornikusers\name{survSplit} \alias{survSplit} %- Also NEED an `\alias' for EACH other topic documented here. \title{Split a survival data set at specified times } \description{ Given a survival data set and a set of specified cut times, split each record into multiple subrecords at each cut time. The new data set will be in `counting process' format, with a start time, stop time, and event status for each record. } \usage{ survSplit(data, cut, end, event, start, id = NULL, zero = 0, episode=NULL) } %- maybe also `usage' for other objects documented here. \arguments{ \item{data}{data frame} \item{cut}{vector of timepoints to cut at} \item{end}{character string with name of event time variable } \item{event}{character string with name of censoring indicator } \item{start}{character string with name of start time variable (will be created if it does not exist) } \item{id}{character string with name of new id variable to create (optional)} \item{zero}{If \code{start} doesn't already exist, this is the time that the original records start. May be a vector or single value.} \item{episode}{character string with name of new episode variable (optional)} } \value{ New, longer, data frame. } \details{ The function also works when the original data are in counting-process format, but the \code{id} and \code{episode} options are of little use in this context. } \seealso{\code{\link{Surv}}, \code{\link{cut}}, \code{\link{reshape}} } \examples{ aml3<-survSplit(aml,cut=c(5,10,50),end="time",start="start", event="status",episode="i") summary(aml) summary(aml3) coxph(Surv(time,status)~x,data=aml) ## the same coxph(Surv(start,time,status)~x,data=aml3) aml4<-survSplit(aml3,cut=20,end="time",start="start", event="status") coxph(Surv(start,time,status)~x,data=aml4) } \keyword{survival }% at least one, from doc/KEYWORDS \keyword{utilities}% __ONLY ONE__ keyword per line survival/man/clogit.Rd0000644000175100001440000001043012514444605014471 0ustar hornikusers\name{clogit} \alias{clogit} \title{Conditional logistic regression } \description{ Estimates a logistic regression model by maximising the conditional likelihood. Uses a model formula of the form \code{case.status~exposure+strata(matched.set)}. The default is to use the exact conditional likelihood, a commonly used approximate conditional likelihood is provided for compatibility with older software. } \usage{ clogit(formula, data, weights, subset, na.action, method=c("exact", "approximate", "efron", "breslow"), \dots) } \arguments{ \item{formula}{Model formula} \item{data}{data frame } \item{weights}{optional, names the variable containing case weights} \item{subset}{optional, subset the data} \item{na.action}{optional na.action argument. By default the global option \code{na.action} is used.} \item{method}{use the correct (exact) calculation in the conditional likelihood or one of the approximations} \item{\dots}{optional arguments, which will be passed to \code{coxph.control}} } \value{ An object of class \code{"clogit"}, which is a wrapper for a \code{"coxph"} object. } \author{Thomas Lumley} \details{ It turns out that the logliklihood for a conditional logistic regresson model = loglik from a Cox model with a particular data structure. Proving this is a nice homework exercise for a PhD statistics class; not too hard, but the fact that it is true is surprising. When a well tested Cox model routine is available many packages use this `trick' rather than writing a new software routine from scratch, and this is what the clogit routine does. In detail, a stratified Cox model with each case/control group assigned to its own stratum, time set to a constant, status of 1=case 0=control, and using the exact partial likelihood has the same likelihood formula as a conditional logistic regression. The clogit routine creates the necessary dummy variable of times (all 1) and the strata, then calls coxph. The computation of the exact partial likelihood can be very slow, however. If a particular strata had say 10 events out of 20 subjects we have to add up a denominator that involves all possible ways of choosing 10 out of 20, which is 20!/(10! 10!) = 184756 terms. Gail et al describe a fast recursion method which partly ameleorates this; it was incorporated into version 2.36-11 of the survival package. The compuation remains infeasable for very large groups of ties, say 100 ties out of 500 subjects, and may even lead to integer overflow for the subscripts -- in this latter case the routine will refuse to undertake the task. The Efron approximation is normally a sufficiently accurate substitute. Most of the time conditional logistic modeling is applied data with 1 case + k controls per set, however, in which case all of the approximations for ties lead to exactly the same result. The 'appoximate' option maps to the Breslow approximation for the Cox model, for historical reasons. It is not clear how case weights should be handled. For instance if there are two deaths in a strata, one with weight=1 and one with weight=2, should the likelihood calculation consider all subsets of size 2 or all subsets of size 3? Consequently, case weights are ignored by the routine. } \seealso{\code{\link{strata}},\code{\link{coxph}},\code{\link{glm}} } \examples{ \dontrun{clogit(case ~ spontaneous + induced + strata(stratum), data=infert)} # A multinomial response recoded to use clogit # The revised data set has one copy per possible outcome level, with new # variable tocc = target occupation for this copy, and case = whether # that is the actual outcome for each subject. # See the catspec package for more details on the Logan approach. resp <- levels(logan$occupation) n <- nrow(logan) indx <- rep(1:n, length(resp)) logan2 <- data.frame(logan[indx,], id = indx, tocc = factor(rep(resp, each=n))) logan2$case <- (logan2$occupation == logan2$tocc) clogit(case ~ tocc + tocc:education + strata(id), logan2) } \section{References}{ Michell H Gail, Jay H Lubin and Lawrence V Rubinstein. Likelihood calculations for matched case-control studies and survival studies with tied death times. Biometrika 68:703-707, 1980. } \keyword{survival} \keyword{models} survival/man/print.aareg.Rd0000644000175100001440000000276211732700061015423 0ustar hornikusers\name{print.aareg} \alias{print.aareg} \title{ Print an aareg object } \description{ Print out a fit of Aalen's additive regression model } \usage{ \method{print}{aareg}(x, maxtime, test=c("aalen", "nrisk"),scale=1,...) } \arguments{ \item{x}{ the result of a call to the \code{aareg} function } \item{maxtime}{ the upper time point to be used in the test for non-zero slope } \item{test}{ the weighting to be used in the test for non-zero slope. The default weights are based on the variance of each coefficient, as a function of time. The alternative weight is proportional to the number of subjects still at risk at each time point. } \item{scale}{scales the coefficients. For some data sets, the coefficients of the Aalen model will be very small (10-4); this simply multiplies the printed values by a constant, say 1e6, to make the printout easier to read.} \item{\dots}{for future methods} } \value{ the calling argument is returned. } \section{Side Effects}{ the results of the fit are displayed. } \details{ The estimated increments in the coefficient estimates can become quite unstable near the end of follow-up, due to the small number of observations still at risk in a data set. Thus, the test for slope will sometimes be more powerful if this last `tail' is excluded. } \section{References}{ Aalen, O.O. (1989). A linear regression model for the analysis of life times. Statistics in Medicine, 8:907-925. } \seealso{ aareg } \keyword{survival} % docclass is function % Converted by Sd2Rd version 37351. survival/man/logan.Rd0000644000175100001440000000164611732700061014311 0ustar hornikusers\name{logan} \docType{data} \alias{logan} \title{Data from the 1972-78 GSS data used by Logan} \usage{data(logan)} \description{ Intergenerational occupational mobility data with covariates. } \format{ A data frame with 838 observations on the following 4 variables. \describe{ \item{occupation}{subject's occupation, a factor with levels \code{farm}, \code{operatives}, \code{craftsmen}, \code{sales}, and \code{professional}} \item{focc}{father's occupation} \item{education}{total years of schooling, 0 to 20} \item{race}{levels of \code{non-black} and \code{black}} } } \source{ General Social Survey data, see the web site for detailed information on the variables. \url{http://www3.norc.org/GSS+Website}. } \references{ Logan, John A. (1983). A Multivariate Model for Mobility Tables. \cite{American Journal of Sociology} 89: 324-349.} \keyword{datasets} survival/man/cch.Rd0000644000175100001440000001206111732700061013737 0ustar hornikusers\alias{cch} \name{cch} \title{Fits proportional hazards regression model to case-cohort data} \description{ Returns estimates and standard errors from relative risk regression fit to data from case-cohort studies. A choice is available among the Prentice, Self-Prentice and Lin-Ying methods for unstratified data. For stratified data the choice is between Borgan I, a generalization of the Self-Prentice estimator for unstratified case-cohort data, and Borgan II, a generalization of the Lin-Ying estimator. } \usage{ cch(formula, data = sys.parent(), subcoh, id, stratum=NULL, cohort.size, method =c("Prentice","SelfPrentice","LinYing","I.Borgan","II.Borgan"), robust=FALSE) } \arguments{ \item{formula}{ A formula object that must have a \code{\link{Surv}} object as the response. The Surv object must be of type \code{"right"}, or of type \code{"counting"}. } \item{subcoh}{ Vector of indicatorsfor subjects sampled as part of the sub-cohort. Code \code{1} or \code{TRUE} for members of the sub-cohort, \code{0} or \code{FALSE} for others. If \code{data} is a data frame then \code{subcoh} may be a one-sided formula. } \item{id}{ Vector of unique identifiers, or formula specifying such a vector. } \item{stratum}{A vector of stratum indicators or a formula specifying such a vector} \item{cohort.size}{ Vector with size of each stratum original cohort from which subcohort was sampled } \item{data}{ An optional data frame in which to interpret the variables occurring in the formula. } \item{method}{ Three procedures are available. The default method is "Prentice", with options for "SelfPrentice" or "LinYing". } \item{robust}{For \code{"LinYing"} only, if \code{robust=TRUE}, use design-based standard errors even for phase I} } \value{ An object of class "cch" incorporating a list of estimated regression coefficients and two estimates of their asymptotic variance-covariance matrix. \item{coef}{ regression coefficients. } \item{naive.var}{ Self-Prentice model based variance-covariance matrix. } \item{var}{ Lin-Ying empirical variance-covariance matrix. }} \details{ Implements methods for case-cohort data analysis described by Therneau and Li (1999). The three methods differ in the choice of "risk sets" used to compare the covariate values of the failure with those of others at risk at the time of failure. "Prentice" uses the sub-cohort members "at risk" plus the failure if that occurs outside the sub-cohort and is score unbiased. "SelfPren" (Self-Prentice) uses just the sub-cohort members "at risk". These two have the same asymptotic variance-covariance matrix. "LinYing" (Lin-Ying) uses the all members of the sub-cohort and all failures outside the sub-cohort who are "at risk". The methods also differ in the weights given to different score contributions. The \code{data} argument must not have missing values for any variables in the model. There must not be any censored observations outside the subcohort. } \author{Norman Breslow, modified by Thomas Lumley} \references{ Prentice, RL (1986). A case-cohort design for epidemiologic cohort studies and disease prevention trials. Biometrika 73: 1--11. Self, S and Prentice, RL (1988). Asymptotic distribution theory and efficiency results for case-cohort studies. Annals of Statistics 16: 64--81. Lin, DY and Ying, Z (1993). Cox regression with incomplete covariate measurements. Journal of the American Statistical Association 88: 1341--1349. Barlow, WE (1994). Robust variance estimation for the case-cohort design. Biometrics 50: 1064--1072 Therneau, TM and Li, H (1999). Computing the Cox model for case-cohort designs. Lifetime Data Analysis 5: 99--112. Borgan, \eqn{O}{O}, Langholz, B, Samuelsen, SO, Goldstein, L and Pogoda, J (2000) Exposure stratified case-cohort designs. Lifetime Data Analysis 6, 39-58. } \seealso{ \code{twophase} and \code{svycoxph} in the "survey" package for more general two-phase designs. \url{http://faculty.washington.edu/tlumley/survey/} } \examples{ ## The complete Wilms Tumor Data ## (Breslow and Chatterjee, Applied Statistics, 1999) ## subcohort selected by simple random sampling. ## subcoh <- nwtco$in.subcohort selccoh <- with(nwtco, rel==1|subcoh==1) ccoh.data <- nwtco[selccoh,] ccoh.data$subcohort <- subcoh[selccoh] ## central-lab histology ccoh.data$histol <- factor(ccoh.data$histol,labels=c("FH","UH")) ## tumour stage ccoh.data$stage <- factor(ccoh.data$stage,labels=c("I","II","III","IV")) ccoh.data$age <- ccoh.data$age/12 # Age in years ## ## Standard case-cohort analysis: simple random subcohort ## fit.ccP <- cch(Surv(edrel, rel) ~ stage + histol + age, data =ccoh.data, subcoh = ~subcohort, id=~seqno, cohort.size=4028) fit.ccP fit.ccSP <- cch(Surv(edrel, rel) ~ stage + histol + age, data =ccoh.data, subcoh = ~subcohort, id=~seqno, cohort.size=4028, method="SelfPren") summary(fit.ccSP) ## ## (post-)stratified on instit ## stratsizes<-table(nwtco$instit) fit.BI<- cch(Surv(edrel, rel) ~ stage + histol + age, data =ccoh.data, subcoh = ~subcohort, id=~seqno, stratum=~instit, cohort.size=stratsizes, method="I.Borgan") summary(fit.BI) } \keyword{survival} survival/man/neardate.Rd0000644000175100001440000000752412461741246015007 0ustar hornikusers\name{neardate} \alias{neardate} \title{ Find the index of the closest value in data set 2, for each entry in data set one. } \description{ A common task in medical work is to find the closest lab value to some index date, for each subject. } \usage{ neardate(id1, id2, y1, y2, best = c("after", "prior"), nomatch = NA_integer_) } \arguments{ \item{id1}{vector of subject identifiers for the index group} \item{id2}{vector of identifiers for the reference group} \item{y1}{normally a vector of dates for the index group, but any orderable data type is allowed} \item{y2}{reference set of dates} \item{best}{find the index of the first y2 value prior to or equal to the target y1 value, for each subject, or the one after?} \item{nomatch}{the value to return for items without a match} } \details{ This routine is closely related to \code{match} and to \code{findInterval}, the first of which finds exact matches and the second closest matches. This finds the closest matching date within sets of exactly matching identifiers. Closest date matching is often needed in clinical studies. For example data set 1 might contain the subject identifier and the date of some procedure and data set set 2 has the dates and values for laboratory tests, and the query is to find the first test value after the intervention but no closer than 7 days. The \code{id1} and \code{id2} arguments are similar to \code{match} in that we are searching for instances of \code{id1} that will be found in \code{id2}, and the result is the same length as \code{id1}. However, instead of returning the first match with \code{id2} this routine returns the one that best matches with respect to \code{y1}. The \code{y1} and \code{y2} arguments need not be dates, the function works for any data type such that the expression \code{c(y1, y2)} gives a sensible, sortable result. Be careful about matching Date and DateTime values and the impact of time zones, however, see \code{\link{as.POSIXct}}. If \code{y1} and \code{y2} are not of the same class the user is on their own. Since there exist pairs of unmatched data types where the result could be sensible, the routine will in this case proceed under the assumption that "the user knows what they are doing". Caveat emptor. } \value{the index of the matching observations in the second data set, or the \code{nomatch} value for no successful match} \author{Terry Therneau} \examples{ data1 <- data.frame(id = 1:10, entry.dt = as.Date(paste("2011", 1:10, "5", sep='-'))) temp1 <- c(1,4,5,1,3,6,9, 2,7,8,12,4,6,7,10,12,3) data2 <- data.frame(id = c(1,1,1,2,2,4,4,5,5,5,6,8,8,9,10,10,12), lab.dt = as.Date(paste("2011", temp1, "1", sep='-')), chol = round(runif(17, 130, 280))) #first cholesterol on or after enrollment indx1 <- neardate(data1$id, data2$id, data1$entry.dt, data2$lab.dt) data2[indx1, "chol"] # Closest one, either before or after. # indx2 <- neardate(data1$id, data2$id, data1$entry.dt, data2$lab.dt, best="prior") ifelse(is.na(indx1), indx2, # none after, take before ifelse(is.na(indx2), indx1, #none before ifelse(abs(data2$lab.dt[indx2]- data1$entry.dt) < abs(data2$lab.dt[indx1]- data1$entry.dt), indx2, indx1))) # closest date before or after, but no more than 21 days prior to index indx2 <- ifelse((data1$entry.dt - data2$lab.dt[indx2]) >21, NA, indx2) ifelse(is.na(indx1), indx2, # none after, take before ifelse(is.na(indx2), indx1, #none before ifelse(abs(data2$lab.dt[indx2]- data1$entry.dt) < abs(data2$lab.dt[indx1]- data1$entry.dt), indx2, indx1))) } \seealso{\code{\link{match}}, \code{\link{findInterval}}} \keyword{ manip } \keyword{ utilities } survival/man/heart.Rd0000644000175100001440000000261611732700061014312 0ustar hornikusers\name{heart} \docType{data} \alias{jasa1} \alias{jasa} \alias{heart} \title{Stanford Heart Transplant data} \description{Survival of patients on the waiting list for the Stanford heart transplant program.} \usage{ heart jasa jasa1 } \format{ jasa: original data \tabular{ll}{ birth.dt:\tab birth date \cr accept.dt:\tab acceptance into program \cr tx.date:\tab transplant date \cr fu.date:\tab end of followup \cr fustat:\tab dead or alive \cr surgery:\tab prior bypass surgery\cr age: \tab age (in days)\cr futime:\tab followup time\cr wait.time:\tab time before transplant\cr transplant:\tab transplant indicator\cr mismatch:\tab mismatch score\cr hla.a2:\tab particular type of mismatch\cr mscore:\tab another mismatch score\cr reject:\tab rejection occurred\cr } jasa1, heart: processed data \tabular{ll}{ start, stop, event: \tab Entry and exit time and status for this interval of time\cr age:\tab age-48 years\cr year:\tab year of acceptance (in years after 1 Nov 1967)\cr surgery:\tab prior bypass surgery 1=yes\cr transplant: \tab received transplant 1=yes\cr id:\tab patient id\cr } } \seealso{\code{\link{stanford2}}} \source{ J Crowley and M Hu (1977), Covariance analysis of heart transplant survival data. \emph{Journal of the American Statistical Association}, \bold{72}, 27--36. } \keyword{datasets} \keyword{survival} survival/cleanup0000755000175100001440000000007112334226223013515 0ustar hornikusers#!/bin/sh (cd noweb; make clean) (cd tests; rm plot1.pdf)survival/.Rinstignore0000644000175100001440000000002611741354632014453 0ustar hornikusersinst/doc/validate.tex survival/cleanup.win0000755000175100001440000000004111732700061014304 0ustar hornikusers#!/bin/sh (cd noweb; make clean)