relsurv/0000755000176200001440000000000013400227643011755 5ustar liggesusersrelsurv/inst/0000755000176200001440000000000013400221667012732 5ustar liggesusersrelsurv/inst/CITATION0000644000176200001440000000175713400172110014065 0ustar liggesusersbibentry(bibtype = "Article", title = "Nonparametric Relative Survival Analysis with the {R} Package {relsurv}", author = c(person(given = c("Maja", "Pohar"), family = "Perme", email = "maja.pohar@mf.uni-lj.si"), person(given = "Klemen", family = "Pavli\\v{c}")), journal = "Journal of Statistical Software", year = "2018", volume = "87", number = "8", pages = "1--27", doi = "10.18637/jss.v087.i08", header = "To cite relsurv in publications use:" ) bibentry(bibtype= "Article", title="Relative survival analysis in {R}", author=c(person(c("M.", "Pohar")), person(c("J.", "Stare"))), year = "2006", journal= "Computer methods and programs in biomedicine", volume = "81", issue = 3, pages= 272-278, header = "For regression models cite:" ) relsurv/inst/news.Rd0000644000176200001440000000261513377434224014211 0ustar liggesusers\name{NEWS} \title{NEWS file for the relsurv package} \section{Changes in version 2.2-3}{ \itemize{ \item 28 Nov 2018 The CITATION changed to include the paper descrbing the package published in JSS } } \section{Changes in version 2.2-2}{ \itemize{ \item 10 Oct 2018 Corrected a bug in rformulate. Strata did not work correctly. \item 16 Oct 2018 Removed package splines from Depends to Imports. Set the depends for package survival to >= 2.42 } } \section{Changes in version 2.2-1}{ \itemize{ \item 10 Aug 2018 Corrected a bug in rformulate. R in (rtable)date format is put into rform$data, the original format of the variables is not preserved } } \section{Changes in version 2.2}{ \itemize{ \item 15 Apr 2018 Multiple changes to rformulate function (by Terry Therneau) to be in line with the new survival package requirements - several date formats are now allowed (date, Date, POSIXt) \item 7 Aug 2018 Add the rmap argument to functions rs.surv, rsmul, rsadd, rstrans, nessie, rs.period, rsdiff,cmp.rel, as is the practice in the survival package, and update the manual pages and examples. The ratetable() argument in the formula is still allowed but flagged as deprecated. \item Allow all the transrate functions to work without the dimid attribute \item New Slovene population tables included (up to 2016) } } relsurv/src/0000755000176200001440000000000013377433672012562 5ustar liggesusersrelsurv/src/pystep.c0000644000176200001440000000566613400221670014243 0ustar liggesusers/* $Id: pystep.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** Returns the amount of time that will be spent in the current "cell", ** along with the index of the cell (treating a multi-way array as linear). ** This is a basic calculation in all of the person-years work. ** ** Input ** nc: number of categories ** data[nc] start points, for the data values ** fac[nc] 1: category is a factor, 0: it is continuous ** >=2: special handling for "years" dim of US rate tables ** dims[nc] the extent of each category ** cuts[nc,dims+1] ragged array, containing the start for each interval ** step the amount of time remaining for the subject. ** edge if =0, then the cuts contain +1 obs, and we are strict ** about out-of-range cells. If it is a 1, then the ** table is assummed to extend infinitly at the edges. ** ** Output ** *index linear index into the array ** if *index == -1, then the returned amount of time is "off table"; ** if one of the dimensions has fac >1 -- ** *index2 second index for linear interpolation ** *wt a number between 0 and 1, amount of wt for the first index ** this will be 1 if none of the dims have fac >1 ** ** Return value amount of time in indexed cell. */ #include "survprotomoj.h" double pystep(int nc, int *index, int *index2, double *wt, double *data, Sint *fac, Sint *dims, double **cuts, double step, int edge) { int i,j; double maxtime; double shortfall; double temp; int kk, dtemp; kk=1; *index =0; *index2=0; *wt =1; shortfall =0; maxtime = step; for (i=0; i1) dtemp = 1 + (fac[i]-1)*dims[i]; else dtemp = dims[i]; for (j=0; j shortfall) { if (temp > step) shortfall = step; else shortfall = temp; } if (temp < maxtime) maxtime = temp; } else if (j==dtemp){ /*bigger than last cutpoint */ if (edge==0) { temp = cuts[i][j] - data[i]; /* time to upper limit */ if (temp <=0) shortfall = step; else if (temp < maxtime) maxtime = temp; } if (fac[i] >1) j = dims[i] -1; /*back to normal indices */ else j--; } else { temp = cuts[i][j] - data[i]; /* time to next cutpoint */ if (temp < maxtime) maxtime = temp; j--; if (fac[i] >1) { /*interpolate the year index */ *wt = 1.0 - (j%fac[i])/ (double)fac[i]; j /= fac[i]; *index2 = kk; } } *index += j*kk; } kk *= dims[i]; } *index2 += *index; if (shortfall ==0) return(maxtime); else { *index = -1; return(shortfall); } } relsurv/src/pystep2.c0000644000176200001440000000410213400221670014305 0ustar liggesusers/* trying to make a faster version of pystep used in net survival calculation - I do not care about changes within a small interval. /*$Id: pystep.c 11166 2008-11-24 22:10:34Z therneau $ */ /* ** Returns the amount of time that will be spent in the current "cell", ** along with the index of the cell (treating a multi-way array as linear). ** This is a basic calculation in all of the person-years work. ** ** Input ** nc: number of categories ** data[nc] start points, for the data values ** fac[nc] 1: category is a factor, 0: it is continuous ** >=2: special handling for "years" dim of US rate tables ** dims[nc] the extent of each category ** cuts[nc,dims+1] ragged array, containing the start for each interval ** step the amount of time remaining for the subject. ** edge if =0, then the cuts contain +1 obs, and we are strict ** about out-of-range cells. If it is a 1, then the ** table is assummed to extend infinitly at the edges. ** ** Output ** *index linear index into the array ** if *index == -1, then the returned amount of time is "off table"; ** if one of the dimensions has fac >1 -- ** *index2 second index for linear interpolation ** *wt a number between 0 and 1, amount of wt for the first index ** this will be 1 if none of the dims have fac >1 ** ** Return value amount of time in indexed cell. */ #include "survprotomoj.h" double pystep2(int nc, int *index, int *index2, double *wt, double *data, Sint *fac, Sint *dims, double **cuts, double step, int edge) { int i,j; double shortfall; int kk, dtemp; kk=1; *index =0; *index2=0; *wt =1; shortfall =0; for (i=0; i #include "survprotomoj.h" /* my habit is to name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP expc(SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2) { int i,k; int n, edim; double **x; double *data2; double **ecut, *etemp; double hazard; /*cum hazard over an interval */ double etime, et2; int indx, indx2; double wt; int *efac, *edims; double *expect, *y ; SEXP rlist, rlistnames; /*my declarations*/ SEXP si2; double *si; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); /*si2 = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - a je to prav???*/ /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(si2 = allocVector(REALSXP, n)); /* Si for each individual*/ si = REAL(si2); /*initialize Si values*/ for (i=0; i0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k #include "survprotomoj.h" /* using thernau's habit: name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP netfastpinter2( SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2,SEXP ys2, SEXP status2, SEXP times2, SEXP myprec2) { int i,j,k,jfine; int n, edim, ntime, nprec; double **x; double *data2, *si, *sitt; double **ecut, *etemp; double hazard; /*cum hazard over an interval */ double thiscell, time, et2, fyisi, /* fyisi and fyidlisi are the values in the finer division of the interval, ftime is the tiny time in those intervals */ fyidlisi, fyidlisi2, fyisi2, ftime, fthiscell, fint, sisum, sisumtt, lambdapi, lambdapi2, timestart; int indx, indx2; double wt; int *efac, *edims, *status; double *expect, *y,*ys, *times, *myprec; SEXP rlist, rlistnames; /*my declarations*/ SEXP yidli2, dnisi2,yisi2,yidlisi2,yi2,dni2,dnisisq2,yisitt2,yidlisitt2,yidlisiw2; double *yidli, *dnisi,*yisi,*yidlisi,*yi,*dni,*dnisisq, *yisitt,*yidlisitt,*yidlisiw; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ ys = REAL(ys2); status = INTEGER(status2); /* status */ times = REAL(times2); ntime = LENGTH(times2); /*length of times for reportint */ myprec = REAL(myprec2); //nprec = LENGTH(myprec); /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); si = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - this is a pointer, the values are called using s[i]*/ sitt = (double *)ALLOC(n, sizeof(double)); /*Si at the beg. of the interval for each individual */ /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(yidli2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ yidli = REAL(yidli2); PROTECT(dnisi2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ dnisi = REAL(dnisi2); PROTECT(yisi2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisi = REAL(yisi2); PROTECT(yisitt2 = allocVector(REALSXP, ntime)); /*add tt*/ yisitt = REAL(yisitt2); PROTECT(yidlisi2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ yidlisi = REAL(yidlisi2); PROTECT(yidlisitt2 = allocVector(REALSXP, ntime)); /*add tt*/ yidlisitt = REAL(yidlisitt2); PROTECT(yidlisiw2 = allocVector(REALSXP, ntime)); /*add w*/ yidlisiw = REAL(yidlisiw2); PROTECT(yi2 = allocVector(REALSXP, ntime)); /* sum yi at each time*/ yi = REAL(yi2); PROTECT(dni2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ dni = REAL(dni2); PROTECT(dnisisq2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ dnisisq = REAL(dnisisq2); /*initialize Si values*/ for (i=0; i= times[j]){ // if still at risk - this is the same throughout the time intervals - the crude fine intervals are at event and censoring times. Spi must be calculated also for those entering later (period...) /* ** initialize */ for (k=0; k0) {*/ //this loop is needed if changes can happen between the interval points. et2 = pystep2(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, fthiscell, 1); lambdapi = expect[indx]; lambdapi2 = expect[indx2]; if(ys[i]<=times[j]){ //he has entered before the crude interval - this guy is at risk for the whole interval - contributes to the values on this interval fyidlisi+= lambdapi/si[i]; fyidlisi2+= lambdapi/(si[i]*exp(-fthiscell* lambdapi)); fyisi+=1/si[i]; fyisi2+=1/(si[i]*exp(-fthiscell* lambdapi)); if (wt <1) hazard+= fthiscell*(wt*lambdapi +(1-wt)*lambdapi2); else hazard+= fthiscell* lambdapi; //length of the time interval * hazard on this interval } // if start of observation before this time /*for (k=0; k #include "survprotomoj.h" /* using thernau's habit: name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP cmpfast( SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2,SEXP ys2, SEXP status2, SEXP times2) { int i,j,k,kt; int n, edim, ntime; double **x; double *data2, *si, *sitt; double *dLambdap, *dLambdae, *dLambdao, *sigma, *sigmap, *sigmae, *So, *Soprej; double **ecut, *etemp; double hazard, hazspi; /*cum hazard over an interval, also weigthed hazard */ double thiscell, etime, time, et2; int indx, indx2; double wt; int *efac, *edims, *status; double *expect, *y,*ys, *times; SEXP rlist, rlistnames; /*my declarations*/ SEXP yidli2, dnisi2,yisi2,yidlisi2,yi2,dni2,dnisisq2,yisitt2, cumince2,cumincp2,ve2,vp2,areae2,areap2; double *yidli, *dnisi,*yisi,*yidlisi,*yi,*dni,*dnisisq, *yisitt,*cumince, *cumincp, *ve, *vp, *areae, *areap; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ ys = REAL(ys2); status = INTEGER(status2); /* status */ times = REAL(times2); ntime = LENGTH(times2); /*length of times for reportint */ /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); si = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - this is a pointer, the values are called using s[i]*/ sitt = (double *)ALLOC(n, sizeof(double)); /*Si at the beg. of the interval for each individual */ dLambdap = (double *)ALLOC(ntime, sizeof(double)); dLambdae = (double *)ALLOC(ntime, sizeof(double)); dLambdao = (double *)ALLOC(ntime, sizeof(double)); sigma = (double *)ALLOC(ntime, sizeof(double)); sigmap = (double *)ALLOC(ntime, sizeof(double)); sigmae = (double *)ALLOC(ntime, sizeof(double)); So = (double *)ALLOC(ntime, sizeof(double)); Soprej = (double *)ALLOC(ntime, sizeof(double)); /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(yidli2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ yidli = REAL(yidli2); PROTECT(dnisi2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ dnisi = REAL(dnisi2); PROTECT(yisi2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisi = REAL(yisi2); PROTECT(yisitt2 = allocVector(REALSXP, ntime)); /*add tt*/ yisitt = REAL(yisitt2); PROTECT(yidlisi2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ yidlisi = REAL(yidlisi2); PROTECT(yi2 = allocVector(REALSXP, ntime)); /* sum yi at each time*/ yi = REAL(yi2); PROTECT(dni2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ dni = REAL(dni2); PROTECT(dnisisq2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ dnisisq = REAL(dnisisq2); PROTECT(cumince2 = allocVector(REALSXP, ntime)); /*add cumince*/ cumince = REAL(cumince2); PROTECT(cumincp2 = allocVector(REALSXP, ntime)); /*add cumincp*/ cumincp = REAL(cumincp2); PROTECT(ve2 = allocVector(REALSXP, ntime)); /*add ve*/ ve = REAL(ve2); PROTECT(vp2 = allocVector(REALSXP, ntime)); /*add vp*/ vp = REAL(vp2); PROTECT(areae2 = allocVector(REALSXP, ntime)); /*add areae*/ areae = REAL(areae2); PROTECT(areap2 = allocVector(REALSXP, ntime)); /*add areap*/ areap = REAL(areap2); /*initialize Si values*/ for (i=0; i= times[j]){ // if still at risk /* ** initialize */ for (k=0; k0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); hazspi+= et2* expect[indx]/(si[i]*exp(-hazard)); //add the integrated part if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k0){ So[j]=So[j-1]*(1-dLambdao[j]); Soprej[j]=So[j-1]; } else { So[j]=1-dLambdao[j]; } if(j>0){ cumince[j]=cumince[j-1] + Soprej[j]*dLambdae[j]; cumincp[j]=cumincp[j-1] + Soprej[j]*dLambdap[j]; } else{ cumince[j]=Soprej[j]*dLambdae[j]; cumincp[j]=Soprej[j]*dLambdap[j]; } for (kt=0; kt<=j; kt++) { // ve[j]+= (cumince[j] - cumince[kt])*(cumince[j] - cumince[kt])*sigma[kt] + So[kt]*sigmae[kt]*(So[kt]-2*(cumince[j]-cumince[kt])); // vp[j]+= (cumincp[j] - cumincp[kt])*(cumincp[j] - cumincp[kt])*sigma[kt] + So[kt]*sigmap[kt]*(So[kt]-2*(cumincp[j]-cumincp[kt])); ve[j]+= So[kt]*So[kt]*(1-(cumince[j] - cumince[kt])/So[kt])*(1-(cumince[j] - cumince[kt])/So[kt])*sigma[kt]; vp[j]+= (cumincp[j] - cumincp[kt])*(cumincp[j] - cumincp[kt])*sigma[kt]; } areae[j] = thiscell*cumince[j]; areap[j] = thiscell*cumincp[j]; time += thiscell; }// loop through times for (j=0; j #include "survprotomoj.h" /* using thernau's habit: name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP netfastp( SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2,SEXP ys2, SEXP status2, SEXP times2) { int i,j,k; int n, edim, ntime; double **x; double *data2, *si; double **ecut, *etemp; double hazard; /*cum hazard over an interval */ double thiscell, etime, time, et2; int indx, indx2; double wt; int *efac, *edims, *status; double *expect, *y,*ys, *times; SEXP rlist, rlistnames; /*my declarations*/ SEXP yidli2, dnisi2,yisi2,yidlisi2,yi2,dni2,dnisisq2; double *yidli, *dnisi,*yisi,*yidlisi,*yi,*dni,*dnisisq; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ ys = REAL(ys2); status = INTEGER(status2); /* status */ times = REAL(times2); ntime = LENGTH(times2); /*length of times for reportint */ /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); si = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - this is a pointer, the values are called using s[i]*/ /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(yidli2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ yidli = REAL(yidli2); PROTECT(dnisi2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ dnisi = REAL(dnisi2); PROTECT(yisi2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisi = REAL(yisi2); PROTECT(yidlisi2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ yidlisi = REAL(yidlisi2); PROTECT(yi2 = allocVector(REALSXP, ntime)); /* sum yi at each time*/ yi = REAL(yi2); PROTECT(dni2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ dni = REAL(dni2); PROTECT(dnisisq2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ dnisisq = REAL(dnisisq2); /*initialize Si values*/ for (i=0; i= times[j]){ // if still at risk /* ** initialize */ for (k=0; k0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k #include "survprotomoj.h" /* using thernau's habit: name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP netfastpinter( SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2,SEXP ys2, SEXP status2, SEXP times2) { int i,j,k; int n, edim, ntime; double **x; double *data2, *si, *sitt; double **ecut, *etemp; double hazard, hazspi; /*cum hazard over an interval, also weigthed hazard */ double thiscell, etime, time, et2; int indx, indx2; double wt; int *efac, *edims, *status; double *expect, *y,*ys, *times; SEXP rlist, rlistnames; /*my declarations*/ SEXP yidli2, dnisi2,yisi2,yidlisi2,yi2,dni2,dnisisq2,yisitt2,yidlisitt2,yidlisiw2; double *yidli, *dnisi,*yisi,*yidlisi,*yi,*dni,*dnisisq, *yisitt,*yidlisitt,*yidlisiw; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ ys = REAL(ys2); status = INTEGER(status2); /* status */ times = REAL(times2); ntime = LENGTH(times2); /*length of times for reportint */ /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); si = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - this is a pointer, the values are called using s[i]*/ sitt = (double *)ALLOC(n, sizeof(double)); /*Si at the beg. of the interval for each individual */ /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(yidli2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ yidli = REAL(yidli2); PROTECT(dnisi2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ dnisi = REAL(dnisi2); PROTECT(yisi2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisi = REAL(yisi2); PROTECT(yisitt2 = allocVector(REALSXP, ntime)); /*add tt*/ yisitt = REAL(yisitt2); PROTECT(yidlisi2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ yidlisi = REAL(yidlisi2); PROTECT(yidlisitt2 = allocVector(REALSXP, ntime)); /*add tt*/ yidlisitt = REAL(yidlisitt2); PROTECT(yidlisiw2 = allocVector(REALSXP, ntime)); /*add w*/ yidlisiw = REAL(yidlisiw2); PROTECT(yi2 = allocVector(REALSXP, ntime)); /* sum yi at each time*/ yi = REAL(yi2); PROTECT(dni2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ dni = REAL(dni2); PROTECT(dnisisq2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ dnisisq = REAL(dnisisq2); /*initialize Si values*/ for (i=0; i= times[j]){ // if still at risk /* ** initialize */ for (k=0; k0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); hazspi+= et2* expect[indx]/(si[i]*exp(-hazard)); //add the integrated part if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k #include "survprotomoj.h" /* using thernau's habit: name a S object "charlie2" and the pointer ** to the contents of the object "charlie"; the latter is ** used in the computations */ SEXP netwei( SEXP efac2, SEXP edims2, SEXP ecut2, SEXP expect2, SEXP x2, SEXP y2, SEXP status2, SEXP times2) { int i,j,k; int n, edim, ntime; double **x; double *data2, *si; double **ecut, *etemp; double hazard; /*cum hazard over an interval */ double thiscell, etime, time, et2; int indx, indx2; double wt; int *efac, *edims, *status; double *expect, *y, *times; SEXP rlist, rlistnames; /*my declarations*/ SEXP yidli2, dnisi2,yisi2,yidlisi2,yi2,dni2,sidli2,dnisisq2,yisisq2,sis2,yisidli2,yisis2,yidsi2,sit2; double *yidli, *dnisi,*yisi,*yidlisi,*yi,*dni,*sidli,*dnisisq,*yisisq,*sis,*yisidli,*yisis,*yidsi,*sit; /* ** copies of input arguments */ efac = INTEGER(efac2); edims = INTEGER(edims2); edim = LENGTH(edims2); expect= REAL(expect2); n = LENGTH(y2); /*number of individuals */ x = dmatrix(REAL(x2), n, edim); y = REAL(y2); /*follow-up times*/ status = INTEGER(status2); /* status */ times = REAL(times2); ntime = LENGTH(times2); /*length of times for reportint */ /* scratch space */ data2 = (double *)ALLOC(edim+1, sizeof(double)); si = (double *)ALLOC(n, sizeof(double)); /*Si for each individual - to je zdaj pointer, vrednosti klicem s s[i]*/ /* ** Set up ecut index as a ragged array */ ecut = (double **)ALLOC(edim, sizeof(double *)); etemp = REAL(ecut2); for (i=0; i1) etemp += 1 + (efac[i]-1)*edims[i]; } /* ** Create output arrays */ PROTECT(yidli2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ yidli = REAL(yidli2); PROTECT(dnisi2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ dnisi = REAL(dnisi2); PROTECT(yisi2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisi = REAL(yisi2); PROTECT(yidlisi2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ yidlisi = REAL(yidlisi2); PROTECT(yi2 = allocVector(REALSXP, ntime)); /* sum yi at each time*/ yi = REAL(yi2); PROTECT(dni2 = allocVector(REALSXP, ntime)); /*sum Yi dLambdai for each time* - length=length(times2)*/ dni = REAL(dni2); PROTECT(sidli2 = allocVector(REALSXP, ntime)); /*sum dNi/Si for each time* - length=length(times2)*/ sidli = REAL(sidli2); PROTECT(yisisq2 = allocVector(REALSXP, ntime)); /*sum Yi/Si for each time* - length=length(times2)*/ yisisq = REAL(yisisq2); PROTECT(dnisisq2 = allocVector(REALSXP, ntime)); /*sum yi/Si dLambdai for each time* - length=length(times2)*/ dnisisq = REAL(dnisisq2); PROTECT(sis2 = allocVector(REALSXP, ntime)); /* sum of Si at each time*/ sis = REAL(sis2); PROTECT(yisidli2 = allocVector(REALSXP, ntime)); /* sum of Si*dLambdai*Yi at each time*/ yisidli = REAL(yisidli2); PROTECT(yisis2 = allocVector(REALSXP, ntime)); /* sum of Si*Yi at each time*/ yisis = REAL(yisis2); PROTECT(sit2 = allocVector(REALSXP, n)); /* Si for each individual*/ sit = REAL(sit2); PROTECT(yidsi2 = allocVector(REALSXP, ntime)); /* sum of dSi*Yi at each time*/ yidsi = REAL(yidsi2); /*initialize Si values*/ for (i=0; i0) { et2 = pystep(edim, &indx, &indx2, &wt, data2, efac, edims, ecut, etime, 1); //sit[i]+=1/expect[indx]*(si[i]* exp(-hazard)- si[i]* exp(-hazard + et2*expect[indx])); if(expect[indx]==0) expect[indx]=0.000000001; if (wt <1) hazard+= et2*(wt*expect[indx] +(1-wt)*expect[indx2]); else hazard+= et2* expect[indx]; for (k=0; k= times[j]){ yidsi[j]+=exp(-hazard); yidli[j]+=hazard; yisidli[j]+=hazard*si[i]; yi[j]+=1; yisi[j]+=1/si[i]; yisisq[j]+=1/(si[i]*si[i]); yisis[j]+=si[i]; yidlisi[j]+=hazard/si[i]; if(y[i]==times[j]){ dnisi[j]+=status[i]/si[i]; dni[j]+=status[i]; dnisisq[j]+=status[i]/(si[i]*si[i]); } } } time += thiscell; } /* ** package the output */ PROTECT(rlist = allocVector(VECSXP, 14)); SET_VECTOR_ELT(rlist,0, yidli2); SET_VECTOR_ELT(rlist,1, yidsi2); SET_VECTOR_ELT(rlist,2, dnisi2); SET_VECTOR_ELT(rlist,3, yisi2); SET_VECTOR_ELT(rlist,4, yidlisi2); SET_VECTOR_ELT(rlist,5, sidli2); SET_VECTOR_ELT(rlist,6, yi2); SET_VECTOR_ELT(rlist,7, dnisisq2); SET_VECTOR_ELT(rlist,8, yisisq2); SET_VECTOR_ELT(rlist,9, dni2); SET_VECTOR_ELT(rlist,10, sis2); SET_VECTOR_ELT(rlist,11, yisidli2); SET_VECTOR_ELT(rlist,12, yisis2); SET_VECTOR_ELT(rlist,13, sit2); PROTECT(rlistnames= allocVector(STRSXP, 14)); SET_STRING_ELT(rlistnames, 0, mkChar("yidli")); SET_STRING_ELT(rlistnames, 1, mkChar("yidsi")); SET_STRING_ELT(rlistnames, 2, mkChar("dnisi")); SET_STRING_ELT(rlistnames, 3, mkChar("yisi")); SET_STRING_ELT(rlistnames, 4, mkChar("yidlisi")); SET_STRING_ELT(rlistnames, 5, mkChar("sidli")); SET_STRING_ELT(rlistnames, 6, mkChar("yi")); SET_STRING_ELT(rlistnames, 7, mkChar("dnisisq")); SET_STRING_ELT(rlistnames, 8, mkChar("yisisq")); SET_STRING_ELT(rlistnames, 9, mkChar("dni")); SET_STRING_ELT(rlistnames, 10, mkChar("sis")); SET_STRING_ELT(rlistnames, 11, mkChar("yisidli")); SET_STRING_ELT(rlistnames, 12, mkChar("yisis")); SET_STRING_ELT(rlistnames, 13, mkChar("sit")); setAttrib(rlist, R_NamesSymbol, rlistnames); unprotect(16); /*kolk mora bit tu stevilka?? kolikor jih je +2??*/ return(rlist); } relsurv/src/init.c0000644000176200001440000000202413400221670013643 0ustar liggesusers#include #include #include // for NULL #include /* FIXME: Check these declarations against the C/Fortran source code. */ /* .Call calls */ extern SEXP cmpfast(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP); extern SEXP expc(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP); extern SEXP netfastpinter(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP); extern SEXP netfastpinter2(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP); extern SEXP netwei(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP); static const R_CallMethodDef CallEntries[] = { {"cmpfast", (DL_FUNC) &cmpfast, 9}, {"expc", (DL_FUNC) &expc, 6}, {"netfastpinter", (DL_FUNC) &netfastpinter, 9}, {"netfastpinter2", (DL_FUNC) &netfastpinter2, 10}, {"netwei", (DL_FUNC) &netwei, 8}, {NULL, NULL, 0} }; void R_init_relsurv(DllInfo *dll) { R_registerRoutines(dll, NULL, CallEntries, NULL, NULL); R_useDynamicSymbols(dll, FALSE); } relsurv/src/dmatrix.c0000644000176200001440000000062713400221670014357 0ustar liggesusers/* $Id: dmatrix.c 11357 2009-09-04 15:22:46Z therneau $ ** ** set up ragged arrays, with #of columns and #of rows */ #include "survprotomoj.h" double **dmatrix(double *array, int ncol, int nrow) { register int i; register double **pointer; pointer = (double **) ALLOC(nrow, sizeof(double *)); for (i=0; i“¤úJæ7¯QVÃ8\Ò¡ÁÍ¥ûœ[Nû5CP rÆ$ÀLàÃiÿÛ•Œ Tzgµ—÷ºÍÉÿþ±îß½-³ôW/X˜Òš?%´­‘N|¹ŠR ¬eel{ƒ…Oqe…7_ýlZã@tk„´gñä"\àçr >š>ÙWÊÁìÿspxîÏ:…ÑåE¤ZÀÏbÕxé üc ïŸ®RÁ#wÖk1hyGu7éš1’xÝÀÁíÇmëóÙÝgåÖ¡,d .6_¸>û¨eúè t«öØq  ¶ß0îN3¹j«, ‚#úV×Mg.=×þ›©ƒA4ůï*éXr²'g¥Yï•CÍvèM›Ýú$²yÕyðзÖ<(½\zÞ84jàÿ”Ç9gT`‰9Ô9Yâw½L¬ªI‹—óÔóª—û…‰3Ú¨}£"•OÚxWN+™Ïîž,–Tã«aÅ8Äã¹ßNÊþU÷ åËhIò2ÔùK~ÆP£ã(ô]جÓ'H³k€Ò@ªoRs™¼â¿ÞÌáÁ·±–¾£x Í«|  ‰0¸‰Ù%d6Õkýpÿ¢ØLXô@ƒ~[·Óòä|e’B Å—#©+íÚáÉÕR1H=KëY±a˹ û“”ëz3$Fi–[é´ã³nþg…ó†p;“1ÿb¥)8¯øEþå¹Ò2ÜÍ g.•&ಀ!:v—¹NÿÂíOWähKˆÇ<Ü÷xd€³³AؔգH·ã|ˆµÁU>Há#:˜³ ~h21ÜéY¤å-èzÿƒ®¥ ÊK¸ÓzøH‘ý`´>äjnP›%‹áüçòþ4ßÍÊ&«úʼ©§‚ñW¬öèK&¥Ëè3â­œð+­ÆH®qVšçùwW/M-ú¯]½ÛUÀ3 _”TU²ónr]0IóÀö‰^èrB?¹pR½ GD5v£‚Ué,.Pîæ!BµS¨»•åBp­ûüúדΣñ‚ÿ± é z›*<„Àp Kѵÿ‰$Lø0ˆÉw»R:Œs—ñ´Q¢N߃E=ßá6jÉÚj©¿è7ýé‘õNj>j¬›%Q",!X RËö¤€ô+ÌÆNú7†À.ÞŠÉYñØøœ][ÃF›0³ãlÜr* Dë>OÊ8w•¾¯bØÔ"_7V4å#BÊw­î ˈ½†¿”êNÀ—¿‰°—¼º3ä7º¤·µÿE;碬œØ5ŽM²‘,„–ƒÈO¸²äPÀãv"H.?\b*µèùc³;ùzÉlƒ2F…Ýñ’1æeùw1°œäÚMa»”/S»î剗éLÕF—FÉ'ì´HT'j&ÀÄ$wŠÝrýZÚ ÒßOª È¡¦Êgøþªú´ óîÓú7þz1OçhNQ®aŒÊ”Õ_ÐÛ‹—F›o4$kžxάw„•çd¯ýe¢Í¨y“)Ï.áÓS‘?ÃüKz¬Ö5áÙ³;¬Pݵ¸ËžïÀ¦Uº £ô--ç\pPMŒ";@Žú0jù‘Øø&¼¿\ÚŸisïÞ{N†¼öþ(4ö­Ë¼³„ÊÁË_2JuΠn0íÛG/àq‰ nr¶rSȪçâygÎf’›4œíJP^ÙÚF²è×–äKO>HÅF£Pùª Úý\@,e eÆš9ÛŽÆŒÆø:V¾tv@çíÖ?ò/ÏîÚkéû7AÅo¨ÉVoQê0h¦.ËÓ ÍxGmœÕ žEó¯bÛs:Ë2‹WÚI.Ío3T¾òoZ'%U ‰ÜÜ(”t*”ɦbøùXÀ‡í˜qó}8éH Áqrjšµ8R®+ñošYJ'øá€A#ô/7bLô|aÅoˆ´8]c¸î“qÓ«k…öù.çà}«’Œ¤ÇÓyIipÜaA(ä„«3±»Ç^àÑ¿Å2“ kª—T-ø>šh=5ðÍqvì:‹õf~m#1‡Qì(åÙvKâä[ÃML[C,«ýÑ6jtžÍ/,1_¿,ºm³.›Õù{.‰›‹(å%Ÿ/a>»òhï Gƒ!L²G+=ì–Ú8¶­rØ !› •õ\c_ǹ¯[WÔ–=öŠC©)Ù>b/±wà@-”ë±ÎåR>—;÷É3xû7ð‰F†mëø4Q‡GÛƒ¥µ„¨Ñx«Zèýk¤XiÊ x“Ë´ð0³”käLž´p+5¹Ø”÷{øÙ‚sØŠk¸ÆMŸð‡Ôݪnï¾3ùŸwð`¾ø´ ¬”WÐ úQ˜k'Ÿ6¡¸kY“5û9tpiyÍÐ ©ûšP½éQÌÇ Í¦·õÀ-äJºç‚0' þð]g\²›Àþ<mo¢Ä#ä¼–@sJÂÊey…ö´*`4±)I aæèÙjŠ?æ]qãêm5¿“¤8bHé©›ÿD°ÜT¹´~}Ú0M•ÏÆ,¶n¡ÍÊ ,Ï€ö P—eÈy¡¤ôf7´ÛHãÚãÞ€}£ûûÒépñy® ûÑÔƇŸ WûÇÄþ=õ…7Ylz*«ƒ«ÏõqènÚ\Nƒ3ÝcC”€Tþ…nËÌP6>çYu_LéÎQØ:}?þ ¸1f©ÏbÚ•G7cØéW x`UD1ûqÈÔb”ê€`ÕÿØXN¥yªâÊÕø­Õµ2E_mm”v÷ÔÒ¹Âá䀲TçóC'×û¶›oHN¢ü/±…¯Ó»ù€üÒ ¬dz rÛH¢]5e±á Â-xFEä(Ò`œèV4D¤e0œÓaïÌ\s4³ÓØ)Nx"p®³^-^Fð”} }r§G“nò𑥕14»T^nyü5ØÑ^FdOèæGy³Óõžäøf?Ž¢’(Á¢Ä$X5¦Ýu†ˆ)u¾¹ØvߥïLcÑGî.2ÿ²R¨—eü~Yžø`õ÷ùÈÿº8z»nq¡&ö63Ùp˜*Fõ`‹ḵ¿’•…baÞœÏëN´«Å!±¸@ÐÆ'«ÖJC¶œü™à ý> séÖîßTŸäÙ2® nßì¶þüãÀŠaµ1âÚ9 ݽV±£Ú ¤÷›1\’¬lž›®¬ù ˜(ª†PÉ2dX?¸ªœ—’GIõ&wÿ‚pãγTäw}vbÒm5xÜ´¯2ûÌÆog[YÓF¹ž~‘S¢?wp½zû³¨{0!K_ÝSI®Ì˜ý¬´ýÛÄÎs›[Twý\´=hŒžémv›%<Ͼ9RC”²ºo‚”åm-th{LÒ¦H¢ì@m<ûñUÒ„¸—jÈ?ÔÕŠ¬0 Æx¾Îµ¦É1 b‘–1è±–Ú;dzºÈ¡( žd as²þ˜•hn¡hË¥ŒX7G˜ÔÛús(KˆpÕ}ÓpÌj¬‹±xOèCÝv®š¸ö­R2U~ñÆ®è?>ý’„a›‚*WŸ6­û´|rÓM âÀ„÷Â³Ž™î‚üìDª‹/’£V~®:*OfRÏpsj|9b>Ú¢vE!kAF»€|Zvë=*<û7÷™Ü©Ð¦ (%¾}ލ(Ê›_ìO`w_ÂZ·ÐÝ÷j°Ñ HY’Ÿ«[Ô”½'ÚÁFÕë^ìcýå#;.Ü»½¹’–á„€æ£Zü‡:àž%F*ŧTMÍR,”n±OçUÝ#írLp5gô=¿ aÏE\{צëØDóÝŒm´ì¦ñÓ©¹Ze<§jÈ¿1‰^šë±÷×ná‚Ô»ø]b¼9¦$£B6¿P`Œ¥6Ž)p’#:9ö¹‘\ÈNJN½;DA Bh)´¥Y¾0U¬.„¦“å1{Ç”j[GOHîá`òÇd‚eQ{ æýú?Ö¥ƒf Ë"\.LWQ¥iHž!{}ü!øôÞÄ—æÏŽ™ãi%ð£Šq/ Þsùá¦bû¹¼°ò ò™otÆCж“Yæqàdel=nÎædÚ‚áýe¨+›òcÉRÇš›òŃ!èÆ%ăҥP W! ÖÁ.ˆê~)£!T9Æ€»K䣹t%!È‹ýÖ–ë?z˯žƒóÞ‰Þl.Dßïêhfåà¶´…ñø/-’•(½;á˜uìâ¤^<>Dך+©cÐæ[¼L½gÚUY¹½vw{cŽÌ6Hë5e;Y3ÒÁ‘:›Ë2 ´p •óÍ#»¹] ßn/|VìɪõÝàT¾?® ÐÇòy®0’óZ× ´4,óˆ |,µWóop¾Á›Aœ—|P °Ýâ!­£yw®QQ>Ã/QU‹Ç¬ym}ål"T¥Nvé¨kj , , CRP»Lt#˜O@& ð Àï¹áêHõN…Tïß—1;)“ãù¾l©ñ–ص}·ÕÉÃް/~Úѧ~FèrZ1ÑÚUo ‰r©AÊ’åÇR¢/Ì}ŠüNÊë†:¹éN¦WÊ®;øîÚ­o‡/¡Å 88~4¶xüÊò*É­Rr+ZOïbþˆÁ¥`Ôè{C^ડ°= ¾!¹Ìl2ø“5½ÕdfÁÔF &L£Â¹•Ša~š½´ óN‡½3D`AY«ÿ·CɵHO¢‰ó5WmÝ÷bÒ?¿9‚oà+ÙÓNÞ P]Ñ 3´Êh¥øëX4NÐÂ7'1ök"Îö >2õ$søÂd·c((½#?ƒžd±›¾=ÔR4ÊÂÀµªDÑÊ„Á‚á¶wI0ðU9_u4„…¹jíYÀ”õ '?ÔŠvýɮƿ[Vû(|ƒ}ü:œÍa%h`3œ ¿¡CÁÀŠ ü”3Wñã{nÏ«êSõu\±[–6yq¨hƒuš ÷È¿‡?§†¦Mª;`ÏЋ¶¡£¨ 'X¦ ­_¾÷›: o yE´l²dÚ-ç°œÈú໓C‰"‰Ü´Ó˜ò¸‚R‚'dr‹ä:ù@Ïhe´5ÝÛÒ5(§Åå $ûœ.±O%6h¬¯,:‡1QiÇ{ñ¡dš›hxÞ›r”b«tŽÂk¸¨ÏUèþRŽ‡ÒÆƒ£YPö­üy[SŠ cBà ;ñ97®„ 1lOïÐ ÕQ„žSÕ­øt^Ö—WôÂ0[éŽò2 !«候?Èãýmœ¢¼"k ?ýxl_/UzvO2× TüŒwãHÅÛä 1&3€ê©æ¿ÕC›š…Ùàܾ™À î™huÞ°ÚIW7Hñ´=}X iÝ%ŽõU P¦@³’bOÂ3Â÷—¢¿/¹ŽñÞˆ.ïSÝÖž¦‘°^—x³^ñZ[f+ò¯2DÞ?vüù“ò'!%]‹¿€î{wÍÒ„9òöà_H[+j`Nßôú—‹ßoÙ18ùÿ4„h?ä6úAŸåkª‚Ø^Ó÷#Ô}ÅlÄuËB6¼0{Ýv­Ð´pnÁÍ„ÃÈ’hs“Ò‚ÿæ^c×%_ÀÃÈ ÃEwFPtû8S,3>­TèTK„*·^;œ`å”fI•u­0ŽÌu'6o¼‘ždµ¤¼ÀX5UŽÎny½Ž7V"Ž7·{hÉóß­`Öi.':vX]¯?Z“F̰=Ö¥qGc‰ÀÞeÎÕÌêhQ4.Œ„¨.0S»Åå5]Ø%^!ðO>Py²–ì><1¼jq¤ÿý+(E V{UÏ6ËEâ5}Ý]Ï:¯[58^|âq0/bž'·ðÙˆ€óë~™Ì…šqfIÛZ؉ 㼉Öa©°ÉªŒX Tºˆ={·^Õ{¾Þe†h>÷Ú¸£ûƒÜ­;ËÛ£žÑó â¹(\—ºŠöÅð+SèÝ8!dù–GÙx¶;—Â+;Ø2ŸøôH0?]É|ðÖê—Ðò°l× ÿF?t’º‰ÕòÍwJbE9@„®q1ë›ÞO ¿@L4Pô‘TDfùWRy,à¯gÄΘFQ¤vZb»®¯Œó>Ý÷}£“>q²šøœâ˜5ï9ˆŠJÍÞü%›%C ¨é9w±Ø³ÐB¬£ôÐVÐs5èuœR,Ô8_Ix/ß¾ú¢×p=‚ã0N;Ãàª{ é6öcìÔ¥°oÙ˜|€gVKÍ X¤-ÿKrè+«A'A<ç_Â/÷”ñ¾€ÓŸ÷¦*+¬ ;³;:Õc‹$d)Ìð§‰:Š1yÆé…² ’è!!¢£¥Êu1}÷)u·¢6Ïß|(‘#ÛPe7»Î-nD½ÿu ¹iáoS`âv ™×"™®ëjr§0`åÍýý3dø/BØË–›ñƒÂÆë3éÒÄx¯¿§ÉsŽ[fR@õØÝÓ”WåæŠD„ÂÌ& ,¿jÌœ•ÀŠI)±vaV£(@e g|ÒmgG¯ùƒ{¼sÃ$ÄIý')à@ng`KØEL·šuó¦žùý²;‘Ó†èƒ-m”·Ì /ÌL/$Äh;‡ÐgåÉÁ€~75Œ³@«-Òéàÿd7±Óyè[ Ÿå}O«€”†òM,„;ÇÐÛ›WðÓÃø0ŒHƒ!4+5ÝLÑÁ¸t$|)Π–ó‰Q˜,Eäcâ‹­ÖÕ^Íůü1r,-åÉ®9pÞ£ùnþo[WÙn¤gÝ:~õÈ)* 4E! Ì€Òšôûª«ŒéZ`Ð3SÚ躾˜Fâ¨`Þ¢:Å#}Ó'ÁÏ A—éÑåMD€‚GÄMÈŸ…¤)aY£5GC/$oP:D-ñ¿¾D¯÷á§¶°<&Ÿ‚[)1bh4ÌLG‡¦«5¨|%M=&™чÖYxZÜìžB¬ü¿,‡¸¶cÞ˜¥ò7$•rb\ÓÆïy¶ìöÇi;¿%Í B§?©ž½¬ëV@èlLv½ƒŒæ¯´ÌÚ6}ÔqŽ6 &E!`³š‰Î.¨!‚›òØLš6Ó`Í¿÷Á\(8¡E6gXk¹¥7h>öê HñÄ!1`ˆ­ðÝ:Û˜km¸3˜†}h‹I¾boÂÔJ~ͼ„ŒLq¥¹ò¬•·æÄÐp‡+½ 8úF,u,¼”tª(‘vF©S¯`]v—N4Ȩ‹5GëFÈým$wWå,< ŹÊMé´úšùåmHjæ«À-vŸµ+®¶d«ºþܶ¸›¨m¸È3ÁŸòÒg]…%3¬Û岕ÜÿÌü—¾4Âò)£ê .W­¶­FE+4'ÈÃú4;Nž{ê>Ö1UËYƒøùÄŸvjd0 ų)L»Ü1 fÑdÍwçº[Šð£ŠÇ95sn K©¤ãPM+—\Í™tÉ™¯Ë·'”û9d|·5v 8@hj0›ñ f–ë:„#ˆŒ{ñ¶ç¥tLÃÔeí´I´ Æ$äwc×[x GÎýNreÉ1l$>tIÒä6¥Eóš&bXnÀK·~·#ªž@³“ÞÁ1žú²‰”rÿTÃ6yñô²‚åZ¹±0I‡°®ÃÉÇáÙ±?JþMOƒœÓÚ{ÂUuD'¦„Ý{¶«:¯g•îÌrz«º§±ºK „H˘me°u Q…›¬yiAÂMx+[Ápša×è;Ò±2Çú'uœ¾­«>ùvíº§ÏÑ„ö°ÄkÌ~dÙ£ f3 ¿rð†zЉítüwó·É¢X‹bdäô™³Ÿ¼RöâqfS7†Ð½Åqù]FÀ^i/ï Ï“†^ÉûnaâbÐ)¥ '‘KN'(ù‚=áù?ŠDä«*aC4•žVÑ=~i~v‘t ýõ9vÇ }±Êf*Âps*w>¤È%YWðJ‚ÕfÂû«s€mÇ!îìRR¯óGöâçÖ€Ãif“:8:o;³õðµÒŸ7YŸ9, ë6X¥½-M¬BūDz¤ÆöPâRð¡uÓ©êì½²é©ÛqöëÏ›R׸ Òzïuo½»¦¥>þõÞïÍ4‡VÂ:³¸“ÉlnVÕ'€Aš')c†>Ü ‚NÊ©ò%sg´ö}­q ˜Îš]ÖÖ*ÙäM–ŒcÇ`PÛ‘%hoÂöÖ:ƒ^§ õê²`£ñãFHÖy7|@ñ1Þ NÝ #É/[îç!쪪„ó:"Ô`}iB;w`ÁÃÍ_ w8³£4 ##ê]·Õ­¼U#›¿õ©©X™1`öšPä—ßÒçf«žBè¬ZÄ®°â@YØzÒ¿u…Ý;¡‹Ê–¶£cWÍYÛ›êjòzœ‹$EãïÚ¼Ž€£XáŒ^ŸÊ7ÖR…#å—'Ý€²ø(qèhè¥K.F:eàJÆï«Xä¾é@2U–³0ë±Dà]wÃ>*&´­DoŽ'ßQ´ˆøöñÕyÃn¦ÍŸ¦žp:ÃrX`ä¼ïÞ<Äç?”`˜¸ø°ä;‡iä/5º¢-póþâò~ôÝy^ϵM:9àø’†÷Ÿ Ëçá8°v`'Ç·*U½F¶Z’Ì@kdñèÌÝÁ0 ÁXàñpéÙçìoYsd…‡ð Y&‚Zî{^ª¢DÎ3L^ŸîþÙR+YÚônŽ)3ÂŽ/3ÉØR*³v[’ªe“dgÁì^U£œèw¹‚z¶L­±ЮI…ë\7Üv¡`^Lãj›Þ8.“ ÐÙ%^`ñ£Ÿ×@Oïo"š9¢zж`œ¯Bÿ±F ð(çáó°+ìt–ÜiÍ[Ök¶Bˆ–·™–OÊQW@4Rd:W–óšFû¹ö¹¥ô0)•¢‡ÇV6GX€”pÛ¾†(]ÆÆü¨„w&ã–Ë/«ìZ}:DgaÈB} ; "Ѽ‘õFÜ[À9õ¶N5³N—mG ¯º÷3nû¦)Uîû¸N: µÈ×$ð Yjxê?¼SU¹3¯±‰1ÿvé ‰W¾+‹Ò­T¥—\ƒëI²)ã£{Õ¯)µ–_2å_N:ò—È=:X”iÉß« °çi¥Æ'Þ—§½#++L¦Hpjhèw¥7W…%íNTæyåèê»&ïÃÞ,Æ ^.ÿÿýÙ°KzÌAT GÐE kQ»)z•É‹}îÖãJx½o>ß;·7›‚²¸UXØ ÕŒ-d÷IrÒ}vHDoSµm¤zJÐñ(sW´¢8*± OÂ9ñ[Xà7E­;É"Ñ7¦:$œÑpAQœ$ªü:¬ÓWc#'¬¢J(–°v%¦'µðåªCBâ=¶¨Õ§¤L±*„Ù)Ž„;¶nŒÚ®‚ç€ÞQôÉɦèMX¥=½“XsqöX›lx‚A2˜æÚ=6u[VoÝž®ÿÛ¬•Þ–Õœe¢¶sß*v·â_Š\…z¸Ð3r 9)ZªØ[**"d'õÃÔzÒaž¬kHÒæš³½Ã«òañ6Ùf+éý'jRpôK— ëß߆~ônC—ÄO¬ÒÇ3FUnjÜ—³‹†j‰hš`‰?(È2¸¶Œ6ûFÍO³±4ÚNtÄ„U1¾üÞ:h)f Yì‚Æg‘ï-Â)M( ;Ž“ÛTrVox2)IÜŸ,¯êœ€äð5«dh $ǨPõîÞE÷(%jq…QÛF¡ sï&ÿ²Úµá(¡7áä¾&ÈMq“$ý«¦÷½‰J£t¢§fÁYšÒùLøƒã4ñ„T—ÜÆõ{<–ð¹øÁÍ›~Éï¾¢vÃ¥œ·hº°ezZ»_~Üu%©úC ÌGh›ìØÑ‰Äš¯QPf } 4NŚξV¤KÖžG¢D‰?Œ”‰äJ >{mØ›º r6;™ k ÷ý`FA¾~óù±ŽÐ HL{ðx3P>¶¨}èó‚—µ^û9´‡§oÓ„3Åç >¼2_®Ñ  ýßÀi¢Þžå*Ìf´&pâUËK0-äu1Á^”WáÌîøÑŸÒ»®üq׆tužÉ¦¶¸–Ø]ª›‹éLûÍ–u”¤âù¤Œ*\ ýÇ›Ar×—ìG Œ(Z‘ç5ñ8Å>iuΆö¡ÆO,q@ŸvÅ®9ho>…?EÆdIDôù(Ã+1_í‰ç…{ÝÀasŸJè B¬.ÝZ^×:hN)nå- åã{±á‚7}S.±ã/ÁưMhA ¦o¥Ý‹)žöýø‹¬ðXÛ\©ª#2ÐR ‰—Äâ\/7.§4:å†2ͽÇ!ZV)Ä%×^Ü…“e‡Œ#íD/ ¨ÿD¾‰ä&-שcES3R°«8w›k_Ñßuì-?Ó„E%JÉÑÅ—©°Ãæä„Í ±«Ö" ¶ ýI︋Cáß:gØõ_»_/ȣ쉘žÊÝ©úk˜Ðú·_âdTç?.Uð=åïÁJixò¿«±š6sÐäH›¢ŠIU§A‡UûßBÛörÏY±EõLõo?îè>PVëÆ~Ëã^( ôÈ5™mí¦ I¦ð¿h°MÜ ãGàGØ,ŽÖ•Zž?sÕ)ȹå–ÏZIÁ‚§Û¶õL ˜Ç¨ åX€ îr:jæ1å­lÇpžgçȲx*3Õ?MK!‹TïnÄL€¼¥”4¥„_uÌõ¦åe*YxAû¤B†Q šœ;Ÿ8ù ¡ËȆrêþ¾4‚³ïÚ‚[ç¢2×M>¾XàÇŸÕ„aÛª%:§¤7¦™ÃðîŒÁ«)Õ í*á½D] €xûïcÆ?©Š¸Ÿ=é?^:ª¸SОK!,|ƒi.Å*ÞÊ;Ź$ ì‘)1½ÚÞ0ÂÛßþ¬î†XUR"\ÓS¬˜áÏp H´(©ôþÜB{¢uözFk’»ø Ÿ  œxGµL|/Ä ¯Ðæmõ·sëàn…¨aØa½õ'E¯ˆ¢¶¥ùFv *Ψ”È‘5Ëö™IµíG#,Šš=§  >\ÅfV"Y£D2@víAÓV´¼×nŽw â=ÌÛ¯`-í¯•{8UÂë^­1I„EÔ¡ùÜÊ,»"ÜcšSòÁû5k/KðÒßÙs0!öhðÏ#·+ÆK!q/Äîì,=fn(¥=%çWb$5§ýk¯°),nÝû•l š‡ÜI‚K`ìB›Íêøùª8¯’Ó²9ìØÄ|Ä>€óÄHéŸÂÿd¤&0a.[è@†Ã@úbÇ¡EwÛ)aÖ8Ï:ä³Ö(y¶64Â-êÚ¡A›>8X‡>§`·U.[3uݦóR©Å8í‚·ÁŸW±ç@ÍV€ûM›î¿oŠ$r‘×¹Ú¶ce90ëÂu‰Áì">Xâáj÷Å‘#Æaª†_ë½N:CãÌË_{ÓòÄ>ܰŒµ ¬á˜{¿Ç2 tçxB“‹ËàY÷ ¢@ÌZÓÈ:+D6Њf`”wzЙL6n›ñ„ïS |ÈòE-Κ3€†nnï0ª=JuJm‰×^{O•Ý·o=^ÈXòÚâJ\޼³ÞÇD †¥=ù:¼ÚÉ»'„é…VY?;êAùÅC v£ì}HÚÏòðÝ5І×Ç?·½4÷“ª<þŸayœŽÇK©½ç–rœY\w$+ Â8› cdýÆœ¥ê­ˆ[é¸þ}#•…MoÊ~bçS=5í=WSbËG”(yæʪ^~˜K¥$ì石}ö!F±à¡,º'l /êðÆ´áð/CϳòqbKKÞ©´È.îYx»ÉEJmbÄñ€þ•×]3>lÈ·>¿rM6£Ö}$m "êôß? +©M9"(T÷;M(ÕÊ €à¼šºÞ¬G^˜N\ø{h0YGð™QÖ8Í¿ ó‚QJD"9FX÷ÝÖìj-iL‘Ô뤆ý:Í›é0ý6 *Q»ÂÈä¶{ €Íc³•°?žlÞ“RæÇ¹£ rå1ÊšÚ%Z™) [b`ÈÄTgÞdCQçfþº™4NŸ“ u}Ý>“žk;NÅtoÉ^™}þÓâZ(/ZþïÜŠ¾j¦cFtU($ÞM‘CËêÆž?¦`Sû©Ú.G¾÷ü÷¦ ²ö…³Š1™¡ö˜T¦ž7¸pW³$.iVacûÚ+J˜sDqB‡ÿö½|+8¿Sµ@l6+-Í Çé©`-ìñ]2CbC†s¿ÈDä"¨¼LþÃÎõ$’xûŒƒ¦§Õ°œ‡’îI¸íJ¬Ð™3ìËŸØvÙÃ3ŽþÄØ6näX£¡µ:ðurv®½‘È¥‹tœgìÿû˜T~w?(ESé€#çúŠbž>'˜‡i(LI‰•nhÔš4ß§•åÎÍù?¢r¿Vá©8çÙVª¡b6ã{@„­¾nd@`u‡×ôø¯Ö®½Ð¸f‰Ë°9Ýj hÎààSÄ— æöìµ%IŠyqeÑØ9ÄÆ‘n´š/’oB[¿ÈN\DP©f¾j@§j#hißcÞè)ÝüÚ×ÇÜÀ*•çS4×Ïâ…H¤¾ÕCÒÐã¢ßÐNYi˜Xû·äýƒRúÑìE„'Þ3 Cò ç²Ö,Ùyùñbwpá}ˆíGÚpZœ0(M) ñ=N¸M#Lï1óïkQ$ˆh¶|þ…ûhÖÝq´ò$hyÓ±ÖV¼šîG$TîG¯"ªîÂ1ýÞv] %‘Èš¶‚–™DÙ÷Œ½^«é¦MöÐ måçù0yöÛß½W”‰¸8À.Þµ`ô@Ý’ Æ“œöã5HŠ´$þãY-áÄ/(Ô#YH‡;ÿjË6Üï}S=P^jY}ÙV §ÖVxZ>C”/ |4ñ" nØÁZÔìPÅ21‹ú‘Ä›‰'¶†fbv˱œsÿæùò*G sÍ.HÚûV@Ùá?™›{e§»¿ÊQ§º“•üÜÇ4ëfÿðˆôû*»ÀãÇÃãREoÞ})Z'ŒïX?Ôþà„ ¿[ù…•sòÓ”}£í¯4Þ÷–R.+QÕÚ‡~tŽU—-\™5(U9£ 'Rêt¶ÊÑ‹Dì¿ sˉ ·Ö*Ê4и(Øï¬™œ&ûÕ…1ìɸæÜc[ ã8>ÖŸù tcM­IPæ'ˆ2ƒ—°ø ©è˜vòn(÷éðA2ªRË‘_½Z¾p¾vöï˜ÃÔ²Ôè ~„Y/»5«Ð•“ŽrsPŶ‹—¼€¬©¢.¼}‹§è°G Ъ “ Á^'î³elçµ#Æ9Á¼ G§¢T×ìN£-¿³±ºûRû´&'JA¥N`ßurCZÐ0Uª‘w_z¯òý¾$þJ”c±…ß‹rÄ£ÿ¨W“[æûÀš§(Ûä@ZÆk?0‘_GÔ¯Ì|Ä7Äæ¨X"Õ†BCüâ.ýu ®@ÉV\×$UÓØD5qk……ôI/3Í ÝÍ+N±¢Õ€ñ¦â äïl$ÁÞ0+„{‡úžO~lIL¸ÆdÝ¢­›1Ô<L:EŒæ¤ä€rRO˜G›Dr–©%¬­¹OùOyž †‘¹„“?㞟ýÍþ™·ó¶èÛ ‡ó™ÇèUÇKm\œ+AÑúop7œ÷­³ÉT8N‚ïÉŒGþ@æ4gB;g‡Ù*Í#àÕ!èOlÿi”)ø‹ Ç‹.ɵCו…M»µw×ÌZaï©Z>¡Rz¬ÛIcÛ2c·¤ÿ›^îåk@I31}—Ž'̓"œã#3rÿ¤ÄTì8ñ—,ý;ã8˜¬@Ú~ì0GVXj q1‰{ÌÎ lÇó ¥¼3§á Þó_ü ·kã•g( Ó¡gÌNñ÷íùfh£2Ö®t°Öýi±¦ßF¶˜ºzX¾`$[9lE5KÊ Ô:îÓæº+±^fl2©šÎ4Ÿy\å¸Ý/ûý5ÊÖ’´º4Ê™ðöhƒÍiUƒƒà‚ù,T(¸ ˆ{Z<;ŒHÿ£¡}º³`g`Öx r@eò‹’LPÇŸ¿+›ó|U©”„)m2JÆ¢ÅèÊ*U³èyÛëz®5]²;I"m2¬|}‹ô”bjh.¼ ã.  MѹžÝ‰ Vè;9»ÄÄÿ0CS¡›Äa¼9¿/ˆ Ш†¦%é>ƒ½W.i®½\’d¦¢5>P¶=ô3s¤ÊF«m´yúö[Õj®`³’$°Ãý6ÎK h,Wô*Ya,q·²5Þó ªPxÉÎÂ:°åð‰™ä|­`v{ròENJ“-›Ž;míŠk )5÷›hP‘=ók ,J¸(hâσj‹‚Ø>ÃÃ¥ ×ù½sï}&þé`t]ÔÖÊZM!Ë{aºÛ6bjtæ(לòhF~…‡XÒ>g¥‡CYÂÅQn$ }«”õBj&5\Ú¬«¦„£9Äd(Ž9TK•b±eLñûJ#Öð#ý^é$“zê¦ ç>?ãó/²ÆA,—äúÔ__<•[`4Ötr‚ªÁÚøÓ¨)Œ‚â@¿¢"7y4<%ñ›]²-¤Úßk̹aa´m ú¨£/?À“†|6‰ |¡°ës¤_Þ¤­@œ* saþTó¤ä"̺DÝM±ìUBÿöv ¨Z7òÞ S5l³ŒDàš'gš rÊýfµ¹´ëxZ‰FJóüÆÏ†€Î˜ ¦Ð¡q´ ]=;†VNX1Zò²¼íý~qáÀ üåÝÉfÂYZ¸¹&Q€¾eàĸ‹YPþ0$£l}kÖ!5õ×3-Ÿ\›3¿×Žª¥ÖïöïÛ|…xäJ¾}(ðà†dΛvDxæ¬ùžø#*9úØ¢êù“U¯ë´0³bb bëÁŠîI7]yÙ™g~çq8…f‚ fÍ^ø¯í#Þ%æIuÚM½ùú!¢—]$b¦ßâ¯ñí±Íú²~­=ÎQ.·ìä'ˆ}²©÷¡ÐsëhòŸ+ |6è$‹!n-aü¢”r#⃨šÉ!ŸÛÒÔ&qÒå¯Z:íPs¼ãµ~g¹aÆWÀ³èµÁR75ÔŸoŸ‚+£g¡ód_UO-IqŸˆ+›œ†O--Hô,¤ÑKÞ³‰«ª:Þ³,Ö¾dá2oXî4–[¬i¸}==q Ùðy‹Ò¡—¤…EÎ34g×µÅÝŽ`ÿÃ@¯U€~Å£8I4I‹ÓD£ÙÛÖ±D‹¢mÃ_CC¾µ„MI™ ƒ&éZd\]öd\8ma5ß"GJUÈf`µí·Ò#ìf‹˜ÄôÌd-| ^ýׇÕjÚ§o¹ó6ád}±È—5ç}kGÚJæ_ª¡$@ÀàÅ—Qž¿°Pùzæd£ˆQа˜:Â}I«Õř竓bž¯096Y—ýæ”°ªúáŠÇ5δ¾Áõ 2š ¹3"%´®ÂÆ›êÿ„xÛAdlš#þ¥A?É…ÍìD‡#áÿS¨P®ˆ°òîN¯š†ÓÞ%½x"8: R3¦¦–®ÊãWüýC¦¬ÚÄ–÷·ÕßS…o¬Ì½­€N÷OYC‘’Í}­#Ï¥©~(ñ—’aÿÅ ûncM# ƒ,;n9Ÿb×EËÐ(TX‹¦OsH»]\û{hAfy»ÍBשlˆ'Z7´l/kcˆŸÛ*hgÉ‘ ßÔú½=jÃßálq§dÈ¿|oOM3a Óèêºý²þ¦oò_Sï–ðµ·´k.©f曊¹î‰i‹דÈ"„FåkX.t’C€0#”OÁè ®ÆO雌l`íÄ$4!¿•€åBÆ^ïõ¢Yg/Ò÷ôøH:]ŠŠ{Ýž.¤“>£}AWL­·ó§1¦È Pû¬1Ôâ…˜©„ÚÀšÎ@£Ïiý*ÄQÏbsëÐ’fƒSeøZ¿ ,²£—öækšIÇ¥ebe\iîY‚§qš‚ö i²z _3Æ@ºžŠ#s@e¦Ý|\Ÿ½1¯À))ÁŽ&nRüÕ¶[Gƒ…±Âá™úML¬X€€çT•ª¸´KÒwÂD¸Ž1~N§7”Ïxbé›5ÛYíè„ÎJ{Kz _(áa€" [6,õÍ\Ó+ÁlÅKh¼·î'vŸ,ô ™‘<Ó—jóòCeus¢Ì ´è’íTé°ƒ7ð3ÂU¬¹ËQÞ•.Iùƒü}­€áð¿‘èe"p©è*Í L{³;„ K÷lßËìB®˜ Óž¡qUIT¯øÊzEH"œ@½e3”LÒµ´3X­H‰€5Cþm@¤ë«=Ko™0²€l61þ×Ýé6. ÿ“‡uÄ36›µ]Ô‚I_C/ž1hBvH…9ß׿âïúT•7>m݃à¹l×Û÷ÔǬ’VŽ×K+vv–2bÀ®˜p ‘¹_|v™_:ë—ÆÐˆœîo™ê†/e}¿Þ§­eÞ tÓ…+y½ó£©56…¼f*¯ÓRm¦ñ!Ü©K ¹ìGÍÌqþ£m±ŽMpþI2|àsK¶híæ ÷7«ÔNŸÝ³ïù¤+GA5‡™7€ðX¸hnäuŸ÷–}C¡¨t\›AÈ5¿¼ pÆð%*¾JgûŒ ˆŽgïFWìæÏb¾ìcëd²°ï„Ò¿¥bRއ„SÙÜàI 5”­yLQy%?µ€Ju´˜‚khãõHä6'qþ›<¨Ñˆv}‘—ÜBÁ¤„õæíî…pĽ<:‚膩Bè’‚¤6x¾'/éÐAèV9s–¦%;|Ûwý»€75 Rj&¬qìVÂN…©âÉA*ó]rQØÝBu‘GÀÑ9éÊxÀS˜Ä­¥î„ǯO%ÀÖéºÆqÕ>sÙPy]dPe  Tº20‚õÿðCEC±¡ºö×m!†ê|¤ƒœ[¡ óV’q-LÙ’e\ÿEH :(qQÈ—ÅçR{H†#!€µ‘x0mÂâËs9»PvœŸRš0Jl¿8ÒÿJ£½¯®¼Äj­ENÌÜ=k*–}Ê) ›/oiÆAÌòÓ—gݺÀ»öÇ|ø“{ï4NßáWÀUÕR}O£ÿ¼¹D¢„ÓÐû¼¡‚ò!våÂS±£«¼ýÛ'k§T¾3µþ+Ñoºm1S- ™wq£7m£éÃ¥½çDUew™¤" D{c‰CØ”Dýj´·r`à ä&§¸ÒU1H¸õ‰÷¢\Jä6rÿKùûû_9`¥â˜Ý£îàv¸oÀ…¾Ân­%\ãÅvƒ‘G¥WDž¤48cê¿íµëÃÞt–_\2µá3pU½ñ‘µÝ•Çvúa€ÿfhÊÉɇÑHs‰•.öLÉÄ7äezáù5p„þ›_¸ÌôâëY{Xøí²8ÙPÏwÞÍ-þ˜nŸ¦«°Ic©Þ½1!`ÁöÜ€¾OÆ”-`rRNYøß*’Ê z¤…íŽ+”‘vaµ¯g?‘)ã°Z`[á·è¢º*yº³hã5_éòЈ•(P°¨RÚ Í•uS‚(¦o¥>?¼DRï=‚ÁG—ß∊ô”š#p]àw&}¼ñ6¿†“T=œøªÒ°u×—½¯dtºî§a¤-•¨e±Q™Ö´g¨V˜±ÿR”‚i"bx¦£CÀ~õæ¹’í[Ÿ"^[€Ç7n{ng\³´Ô¹„ÝC5”†‰Ìãy«Åß_²E·¨^f­ˆ”ŒÐiŽ€óñ%ý†cÓ·æbæã»’¡ÖÜû‰jW`ø“Ó'( nØ`y³åÇÝÈEJ£ËVŠ¥Ç)®¥ #ª©ºÑ¦î¡Fð­g}Ü‚±ÈIŠÏÓÇXó0ºÑ×q²9“¢&x§s¯üîóUE¾y3kEñÓQr19w P~ž„×Ý!liõÂeÕ0ÜQž`‹é¶¢Q2âÊaÑòTê) )x<&¤P˜z±”°¶=eˆïTˆ-²†jè­å>¶|ºˆ1ΰ”Ès²=µñ _÷Þ­UjÂ)ùÜ\z D%Åv-jd.…Õ„ãUÌïÔ¡”i±sx#‹9éÍꉨ_Öo.{3ݱò©µv |Öš… åFê04>†<ók «‡€CéQ›†ŽBvSV®ÞxŒãS¦Ð‘z±«?%_ˆ”á“ô\áZI 76¦lã.70Â$—»UüßÈšFÝ¥öå ûc6?pµI4ÏÖ©Jp±°Ý<òMÖíï¸4W³‡0,¾=ݘâÿÉûÞŽßsO£·å!¯µ-ò7À³5ݘ…\›•Š'Ìò`1GB Ñ~ý¿îl™ÒwSÔ-½+9eXõÃnü§Høp:¨‚my)Ú6c€•ÄÍK‚ÌGbwNLÐp×g .3Ù ù²öÇу4ŽŒI²("4ðùar§þ“yMˆêJŽl³NcÝhEKÇé=–hòKzO¥¿bK›½ÂAžÎx‡åR(4&D-{Ä÷¢Kú˜û7Ô*Å'r ÷Ý_nHÖ$õzuãÁÉ–äñ|Q*Áã_×k§ g¢ö!ëÁŠÈK_ò‚/‹Ž…e°ýÝkÀ–Ã1 kaÕ5®R’‰ÈµúÜ~ð˜Êˆ.ü{Å8–Õ:rû³ÿéßUÇ™X)K¶äÆš„‰•Áá¼¾ó]¯è¸ºÙðW­vÿ¾ÑESÿøL(yö¤øf›4‡ãŒdxèE1¼K.˜”·í«Ø…µ«QÆl€²˜¶ï4ZÇÃb½nfëûÝõÚ˜ÖŸÇ×b ç, ¥*IÙ& Dòþú^%lJTœôXwîY±p7Xhqe¬Ì[)A~í°êÐÇVÔD²#%zS« ¼¬ëAtÝÒ°×ÕFØ­çå)²†µûW¢g?±e¦Ê§2vWX"1¦Ûà>û+›ð%\Ê‚×c}®F™G6X9˜é„·15«jí6çð“?¿ ù× ²ie|@Œm¢ë¾ÙXRdT`ˆw;üï­G‹.„)mäÝÞj¾iü¦lð7n\R¶Œ¶ûSzj µ½Ihu}ñ+MËÜq 2h:Ÿ¶ëÊ\A&‹]òë¨]ÜÏÊèûAÅ–MèŠÝµîR 5á›ECM ³Ž¹‚ÕáXcÃvù’ïÃí|Sîò7Û×ÛÚASúòFÚ«gdÛÌeæ}$ È|@½è;¯ÍgO—ú­8ÜŠdÇåbVp`ÏóñÔùì g ™ò{¯.ÑËb‡‚Ô÷ïðºÊ)%\(’§‡¹ìˆ „Êv±hü«t*#"I7›jí^´Ḭ̈ìþåû`6-F«=vhå&n¼ôéØaª÷]ñE’9ˆF Ïv.ÿ‚á£O·¸Íyuúá 4ЬH¬-Ì#³âqéÌK2F¦ÔÃ6:Êñ둊 ¼Ûâ7øxp/§—­vR^%eq18û‘â½nìDÁ¶Rƒ}h0øMš>)B]¨±_ÐK¡»ã›™ýo  .Ý„ µR?hJ© ­|î1MãwkçùÙ±î$¬¬‘ä©‹þ&ìL0›XÛtìË òûÕK™£E\ *òB6@ÍÔIº«n’ö™aR3nf¡)ɨJ“”˜çÉï¸ñÍvMÚ‰K»Ëáa=á³YãâU¹È*p3×ÇÈ¡ª'/ ÇÞ߉–¨²êŸöÿB|cy §¬2ϨUæôh¥¥…¬,â`õ®±Õy!¡Æk–ÔRç†2ñ»Jj—eïËú…›Üm"9†?æ7n¦¿ ÖíùS ×çðˆò_´E®gu°nqÂñä±Ât#…6ª—HDe)ø:åéF×üþÀûÜQÉ“M M¿vËàU_Só¢ù_Þ¬÷Àˆ7·r­O…@§¾Ü- ?¸f"k%¾¢hï¨//—çì§Ág$\-µH»0½<³fñ¼×Öýsüµ¨à4^j×’" í·Ÿ¿í‰£áåO~Áç>Ù*Î2bω#“ÿZƒe¡¬N'}r&1aVã i1p1ºê>8߬ī ~¼`àEs¿½y cÓŸõ3/)ûae'T~[k“Yà ãD'Æìš’Á!>mÅx³:!ÏC¿ùŽlPÏÕh_êâcKÀ/8øw*¼¶A¯®yé/$ƒzžtŽBSÇ•;“ß&M4Ò1‘å)¦RpD<&€qKv3ˆÛÓ*˜«m5õnFV“ÈúG3a$J]¦N ¤fp„RüÁ”S–ø¹„Ì#×ÐûóøÆŒ6 ¨ÓJŒ0,:Éê&ùc W®‹`RwÕN@^Š*™JÉÓJ!lnjªS6åCÿïF=™HŒ{<ÂyøseOøuˆÙÜ«Ivœ=îZâØôFCæ‚ïξ.Lˆå12TuJÈ­)1¯Ô"©/Z¬@]Kk­ˆG*¥¯R4QBÓl.–†¬¬ßG¯y±LÂ-"Ï”_Wy}i—…7 èLŽ=ˆâ&Ôô ;3ßkÄ',ìG/iv£wÉ›ÑcD¦j: :/ìzñÀ!qUš‹„UOE(Õ‚pëÌíÛyßO ìM 7‘TñOC•R(ÏïMõ-î1Á$)íe 1ì OѬ¼÷¢±áU±¬WóD­“ö¡`ɹ¹y™wqúHÒµW ho” ou˜"D®Éëwüö¯Šp=f3ÚláÐcæíõ…hÖ›ë s3ª'¯õ¶1“%úr=$7 g¡PZæý¥ÕÊY¬'¾¿“Ø9h’ ïœËÝ€Jç¦Ôn›­Ìr+™:[ðmh)õ©ëÀ¦!þ+æäþ_¸@E§÷¹\~Œ`6Õ`\æ£yqžôñÕ¤e]sºýCDgZ>:ƒ¶u?E¨–29_šÏ“hAõ`ÈrÒ0v'sä“z0Âö½‰úoðÍšI¬±”$¢È‡‘jiÄî fêrb@òâÕmP3,Äðƒ€fC~O‰‹þ¸Ac½øoEkºûºõÍsZ¨›ð¦dÞ¸Üòë¿× ¯ðþQË¡,B/öçCkEºTû\IþÏØ£”qXJ¢S™l™. ÆÂ<Ê;8@yÞ£_r³ 2ì[n¶ŸÍ²ŸÌuX?8’ÌYTQó¿§Oö„…Ôù“q¸ˆ£¶ûÞŽú>¯ÞÑãh…aIa~P0㲄¹v³a²  ¨Ä¹²±ñ ÝÆÿÞÑPá€Lšé43tÓx2:¸á"ø¯Cp1B;aNK4Uö³L˜#Âáþu!l´Ë6á,Œú0U¸1+ÖKQ“¨ï‰Ù+'ñôCYãŒÄË&µ×Ècü‹ åše¬°j3S{„MÓ‡n°š}H”ß°çtH]KŠgDAô}ÇÏ‹oñ– |ÁëÌò¢C¦X–|ä?*Øÿ4rÊY~$mqŠÊãÂ=ßRí…D.­ïŠÚ\ûÛó®ï©Oú(˜]Ôzêz}˜2ûà–—Ù V.¿¬´8K6ÝuhÉ9çšË•@°‹ëÜÿwV]ìÿ׺œ#›œi[gD§u’UPÇtU•K\Ï4ÊþùváQWŽ©B-“é%„mÿ@ƒke¡™›dæýœ|Éb»h_ätnñó'¤ ad‘ä²Ùp}bqă9üU¸ùƬu–qøÚª>ÝÒSí”Ha‚ÄqX4²3€%ßw=Ë€”6œG(­¼›Ì”SBMu;mä” Ú¶~x+¯a4RÒÜ߇ö“˜L8MLáülè»AùW‘z »ÚAøqÙÚd’ëñÕ²4[I_sôþ‘ê¤o䓤*ÑF”Øß†ŠpTÿÆp±ž‰z!¿°–^ÍçüŒË=‘“\¼Ÿ `e'V¸Eï–€€M=·áƒ@XŒÚÇSx=džI§ü |'C¦G½—ç¡!K——fE±Ø'Ũ(ÆyOÑ’·)\Á<ÙÞÃJK`Û@êAJÐæŽÏY ÒyqŸ¤cD¢–ÒRÛ›“ÒÚæè’XÒ><¸² ýÞ#õ*òJ#ù·ßä‰lCo)mÃr ÷7§Ð6§5FèÁkÕ¨“pˆ[1N|%¨ïáÇ\sÁ’—¦J>ÚõÒŒ˜¨£¿æ{j̓w=H iÓy[ý:úìIm+4& }‡Qœùv•,)YŽïŽ5¢³Í1†½rÉì¤ã„—džaPåJ•Xl¿RGáï«Þ|ärCÅ#¸çÔ/Ú‘²ÖJ¡K¿™wdpŒîx>h‚øÍh¤w.tÀìË+ÿþ융÷ÏŒEÅÀ•H+HéñŒ~ZÊu(€ýž Xà²5s[4±}µÔ_êNV4Äêê™ÌÄ«pþ溴eÿäªNËû¸ÈÐu}>¡µÄ*Íê±áLOƒ”é7M~¡èœË‹t“I&¢š¼ùÍ<t¼é1½ƒ»Á4ÎÃvëg½Ôr¸ƒ5S@¾¸¢½úù`ˆæWÁQA*ÆKî•]B4îÔžsþ9OÈÆ¢šËïÜ—Bö]é/‡5 SaÓ8éºKáØ—÷w-ÎÓ‰ÀÁðÏasRe™(Dz´ ŠŒáŸª|9ójÞÕ>µ5z‹qÒžZ[õaT°’X¸æ&]Ôx&W½€<ñLåp»â å{«œf–Ö:CÛAŽ•æë̦…8ŒœOÕqX¼«pg$7Žù8á6Zy›Ë òþšƒÌéâ®Å`ç‡M8å“$0IVÐé:GÈɳOa¿¬ˆ$!J"`cHÕ¡Óór‡H†|Gð:¨5û[kÿ1>[Âè¶ßC’‘ä´ïÕ”§´¹¬^fêº-ç²EMO½`„cNbNgÞω8€QAK_âøo»È"º%ß$2б>MÞ¤³¨Ð§U¸V…~Oà œ”ž’ßgA^–³ðò¯ÎtÃM רä±ËÂàÇzWÉuø˜ +Ø? Ižl#™ VÑPµªÜò>.st4 V\ijé *†¶ðï§¿ Ä|ØF1I@±E~6<Íæ Oº’%»ŽÏÞ|ü›–îÁXbG„*‰\ JÌ>ÃÝ-d-‘¼IçÍx2ßê¯/óhZ¤·¢ûvÞ‰üÏOè›rØý¶ï™ˆ ËMH2 .æ§êý¸vµAd£Ñ£Rt8}FÍ$±2ÓÓ%¬aFSÍàh”Ø9•[›îjÆ(Lôó¶qñÂ,ÿA eOhÉåê"Ë þ³¬I M›€<¦WñY üyÌC­K‡ã¤ÍüÅ£&B†‚[o‚&ñ÷<œêE69ƒˆ?Ë>œyÿ·{¼+²Èg.j_ §öV¬®e|«ΕsÂg‹ût‡ÍÇÍ/çþ%¿„}.œ¯ÝêÏœ­…=X<4æ¥FÉùå íVSø2ü¿Y(Qè¶žTï$\¹e&†*l[¶˜ £ÌR8€×ê2}ƒ:¾ÈnZßà³1†8’ºÃ‘i9Þ¬Y+ŽXÓJ ÓFÒBP{0„IT>kÚ{I8ɼî=—>æež´ÁÁ¤BmwiETpAY€øL7Ú½IŒÜÚbŠ9lă_7É/vîŽ;pxœ ­¬÷,縢 ‚·5£©îw¦§,t¬MN’ JøÚ¥W×Km$…ÛŒqFQO=ÊbF ¼ýÍ]X "½&ƒµMR?—Äà‹‚9àãUmf­¥ÝEb}Š Ã¹*'@W;⤟°h@òF´ˆ­¸D¿×´ ŽÕ$ãä°œ+3Äðã+¦vv=N™ #JïáÕ”üT®ô‘có™ ;‚_„pÒaÉì00ïl‰žÎ'A&.Tì±î¯_á3 ö3-7?oÆoÀ›&`âgj#Ss±/ã\è•o[{´»ˆHµ#GÙe³Š„×ð´{Îæe%•J ÿĨ/4ùÀgjjTR–M›„€±vÅö(>±%š$…[ê„®uÂn¸lâà‚ñÕ€}§Ù³WÄÏbˆj=Ùä'áÅÊGa¬äïW ð¦U”ÏfØ54KˆP·€$¬Ý2ƒBöhpX~;yæÎÙf›˜cëQŠRíQ«pÅGÔû ùm[ ‚ùÅáÚÉTž°óEèIô‡²˜Ó‡ã¡Ñ^ŒÑŠÓxû”ßV£–…°GäÍr Ý Áœ¥ED–% È X_yªìYdÛ…@½n}&⃈@;)N‚ûo¤‚GÝÓØ•¾Ï!”½dGë\WàÑn^%ÃÐ"ÇÈ5òëå݇Ý,x6Œæ’rWƒ?R­É+’xÖj¹­’4yS°Úfeým ´9 1…²ÍþÑuÆV\ÐþyÔnÎÒøÜšÚ»çÒ©·Â†ÿa¤„†~×72òÁ.,™Á½6Åd,eÙy&6þ¸%Ÿ«x…år˜ˆ!K¯‚ý¿J)³ð¶)fìÆ= ÜA¡Fp_²Õ» †ëELÓÁ8Bì4z‰J=d£ØUº7ꪮ»›¡}7ðë“,¢w>ËœX“øéÔ+&páõ¹]8[=Ì(ÉeŸ…Óìg çÒÀâT$]0åpÃáOdÒ;„­A(Vœ¬ÄŬڪB—ù ’ßæCº­Ð¿²"ÖèÍÙÙ´û@ÀÜéLµ}3>³èΊ§`UwFÍ×=ÚÞB²B¬­/Z7™Nc>ö“Ò%òóhŽÎ¹ôo•ùªÌ¶q„ÈzdäŒãWßËB<Öƒ·È“̓Ù×ú/fŠ' 7A}H™ý;ÿ±îÚaËÛ}¸«zÚC¾\RÎ4Tb:éàaV>‰J!tÊÞŽxô…uÞÎYÈËn)yXzZ yÙÔ¹L^c–‡Æ HJ)?™Fþ@ü9ƒrõ>oòIr-ÚåðêÏTê\P sÇ3Øà+ƒóš|,€vWú› T½;Áƒ®×-DVÂ\KÛ@3gxníÃFÜq3™dÌ™ÿh¹hyœÒ Ÿ³akË€4ë9¤ÅA3'+0íœß ºC¡9­¾Û6Öö„>/2Uð8*I–g ˜ˆ)^‰ÛøaÑÉ¡K;Ä;Lú¯Ïæ{%\ž£c'md3‰_y¢îXæó¯ä¬°Ö Âz}ͦ:3ám×õ\óñ¥©%ÆNš¯¨n¼é‹rX0”3 ¢‹K9yظ;”‚úˆ6ƒgYO„Ú (–sȪ°Ó^q&ËE÷T’¸êükÔn7´Æ–×°ŠÊæÞA?¾8èR‘"Ø X£ ŸR«b`²_ó{ ‘VîU”]@`\±JI\#Ø«c"Öƒ¸Ê¹h 3è•vb|sz,Ö61/?á×r‡2‘ g˜&P¶³%at7NÖвº=½¾¢‘>u´Ð8ð–'þ ~–-|°z¡›îi3(&D¶k\R&<_m)LQ+·7ʘ|tîx'RÛÿ8(Ò€ÆÓWqÉIÐýÑ)L˜!Ê+jÞ6ʦÿu JÞㆠ£+‘[NÄ‚²ÿx5“P›^÷ÐÎ)1VÑÎO8ŽA¿‡ ,úVh™#-Òq»«unRÝ^âLD¨Gàáå<‰‰û½l˜¡—¥Ä Núªû‰¶q!HhZ.°„¸$*µËˆ", ¿!Ö˜q(SÙžíeVÞÌ?Á6>Õ݈“ÏÈ hqøy}Q Ã-V1½®¨šÎÿ×DâèV¶ô>]¨“_àé#W/±=þŠñ³ÆL¬pÇ}`S>¶y.kš®û°9fغ`¼œ&üáí'Å^{í2ÎÖ Ýª¸ÒA8ÖŽÊB§Sn–ÑŠžˆŸØÔ˜‰’#É-ií£hþŠ`däì(“&tVg{±E”Ù»¤«Š;Á­bàÜ1:ww}€óìVá`HbèxŒ2ŠfT̨›…ˆC"cZ&‹ºDy‘ã†z2iñ ?ßcáæÊÉqé䯰]“&fà/Ì}ÊUþøæw qJéw¦Ê™ËÕ­´ N”óSÙe ÓOYmŠj똶ÊCX”èK¸HíÔÜC„Îh*dÞ¡‹Ñ/jàL¤‡f `v‚ÅDž ¨¿XP½á%àe^¾zÄ,™ëRØ3z$~i‰WPŒ!âOÍj¤Óu‚ åúçÈp¸:ô¬CjVÿ”ÐÀý¢ š:(§±›`1Hš^:è¶>תR³Å¤8Œø33“ŸÑ ¸ª¬ ƒ†Ä­ˆÇ6 A^lX³Úà’¤î€e™¾ØM2=Nð†{Iäöôó­®ç?ƒþa•Ž,­é¾É’‘S%† 2^òü±nNó̱ë/ž„*èÐü hvû”àØ6)‰oÙeK/÷܆](Iojí(,ƒFðäTÇÔŽ>j?ׯW¶«Š—×â©‚áp ÉKްð+€­Ûà~Ü„Û^(ðç0žòÙ©‰¶ÅóZƒ–AçE–/ù'˜{éOÜ÷¥ù…p ’üMI€ðGPýNÙ¹G>]a–hÅã…ôÇ]ùè¦TÉ „Õ`¯Ô1¯;TŸT¨ZÇ'Èàt•þl¼‡³©'dïìæxnOVáQ®·)«¾’ÈÒ8bßí„Í9:Òº×Y^?†/,> ‹wN# Ýß`êË$×½mEu9c>9¶`±À¢AóIJ²®up9ßùiEð©c»RV¦]ëm>%šq°íhLÚfQ%㦇Õù¦•7¬©~´/ÏÕ;Á™Ø£(ç"Ñy~ÞôK.ŸbPÁk@-bîVËŒ}ʹhÝÂu܉ßtÿ2m„ZÒ÷Á™Ó*ð|鸎]N<¸Ï`Ûÿ¤¤I´úAE{y…-ÐÚöÄÔ4âØ)ë AÛá^uM«¼x‰á¾]LÞRhå鶨¸S<;T»_–ëS Íí*X«ßÜŽÖ·ðŠÏãê(à# ƒù½Ù œá¤ßV§xÀÚ@ªiËZe`BFÎiÚ„>jªÄ›½L|æ<(gê4ªÙL+ÑÑÈ<ã'Å“'oæ€î± ̳‰è:ŽßJb×û­pÕ8„þ†ÞÃYøGËsq—9ŠzSIãOhC¹ì@Pb݇Xµô§½fÙ`ƆÊuà ØÓïžûk2Àª2xú$j³è)64h1g¿XW» ²Š2IŸáÀ"tn‘©%¥Fu¿âÈŸdý?…·Eäßï¢ò(d(‚þ¶íóáRò&¸TýÅûN8,¼–-„#c}HœŽ Ó|¥ËxÀ?QPS§|ðÁÍ+FtÚ…)J3Â:ýj¦h‰ºÜ–‘«Š`MO±3†wÃ/K¼¤¶Ãº޳U2šx4×÷Œ…Ñ–ÖKàG. 郙«Ü¢Ã>+u {wª¨ô‹åôɬTß‘ó~—¾+1É»sE_˜Ñu• úÅCã?ê3É`dŠ>À$SKp"yß-ä"ò™eÐîXAŽ<™X§ùª SB;—rd6´#¿Ìʺã‡ݘ$\MAŠš%ŠÜÛãlà‰"qˆ ¡º¸â‘}&¥îrð„(ŠjC:„r'W®-úܹ §€Þ•N5{Eºkq[(y!k/Xÿªå¼C&V@G¥âúmêÍ>×tîSë7ñR\ó…ËûÊÀC³âH¤9"Šd÷»8? Û…žL‡rŠRöØþ«´V©¬ÿë6êý±WÓ†ÝEN;g'”DZXìvùg’£Ýãº,×볞õbS™¿UOeÕ Ùýƒ!H|Úk‘~ØM*È€§v“?yKV‚‡)Aá@¡·Ð³ŽˆZû¸ÍƒJ±;NO¨è',Ù녹ƪÀý~@r>£òÎä]Hêù~9ã[§5Øzª¦áqÌõ¢V‘ð´‡4ŠŒÅð\Ên^àÁ#ŽNã<4y¢Ï$Xú»ŸÔ´v-"!n’kO åpUÇ¢Fé¯`üvSLfO ±}÷y }l×¹“ײH¬‡Õª3k öñÐŽNl…y¥, a4œÏ‰õ WV°óôI¥B6ù¡c…íˆuöxôGÃ3qŽ»æ€K®·Ð»³·))eÐÈðîð7Ýë& Owåg¬H™z¦·Ýš8ÿš¸Š¨€(æ)* ÐQÁ Öºm;Ÿ]&¥Ùš+´ÁÆSx_+½´Z¾FˆÆ£…<>˜Ê¡M ƒ÷Cjg‡£b‡ÏÕØðd§s/ìbŒrøînúóšHÔô¼–}.†õÍ oø…Ðâ'Q? è‚­D';Á:½­È"Ñ+Fô*¤Q~FgƒÙ;îÈù÷No1øþ*îÛ)ùàê ðnùþJ†õ¦±\¹£ÿÕÄ(¤¯ô¦Ì/wH€ŸŸ©¿ y¥ôüæ’Jû? è¥þq>Џ°L¥&9 ¥i0ðlí¤'ׇH>”0ã+¦qòdaRšVóöŸ\QÙ•©=Á¶œÎIATÚ/bElMjWtì Š‰¹Õ]Z³o ïë÷ã.zÒjI/P„-—Ê©g AÈ„’ƒñѬþ!Ž~2N•Æu0œkÁ&¬ƒS‹|ìSMâ!\8aUa}Xq|nH ¡àí(1JÖElV --§AÈÜÊ€%Dr^UÔr¥ÉŽR’5½¡'\à¢ÅóýM]âßva²Í2äTé¯'f8¾d(l¼ç^ÿsB†jÔû“éÐKEP‰·Bi~)€Mܶ‘öÔ7~ž9ˆ°àöç:ãÙoù­QW9 µZRnÖbÊæ¼Oxžv’‰Hèƒz^¦“SÆå'`Fü´ 9·çÒH•™_Òú¾†L¦ìGérËé'ÜŸ7.I'W¹RµíMa¯y3G=~¹ç£`ì[£ä¼ý‘°í·u33*VŽe#êØØ#᭖ΜAžÀž¹BAõš@hE+4¤ÅP8¼‚#ò¿5¯6΀ࡡôN¾(“i dÀóhZJ™Åõú”-·üîÁö'à¿»™Ÿ}ç©[„·öì Ô–Ü`–f7ºŽ?>ÂZ+þ‹pIRsà àø0tƒLÁŽ fØ%ƒ[XÞU£® qbŒ“Úá1tòi^@ñÐÜÙÍZ‰2ñ+¢C£ã)‹~³ÀQ/¸³Òû|öTáí…=ÖÝfÝ"ôi?$‡ K²a8¡‰<æ)™(Êi'Z”a ôú:– |NÉÍÜT”â÷å€%NYË%:${!}2`m}*95õñpr`k&&ÃÕÒFjZÕµR¤êÁ 僧L…œdôzŽ!»?¢dRñ€›íéßðéWÍsóÃï»-(¥ { ê²«$w ¿Ÿ)ÞYŒïlË£Õ«¾ØÀ*ýóý¢îÚ«ôÓo[ Úª(·'Âa'¬ì…6ƒø¢a‘ÛðMjb‡ùSï—R©qй|õíúÌÜê¡CHqEjÅ5&‚˜G*†ûWšLFËí³XË](4¾Ð‡hÿo>ÆV Cˆ€*A9%óÞG™ItO$‚— EبOo‹Õ¥äü&VmÃ;ÄÎyÌѝÅ®–`;Øì´xÄ]Æ.‹Ÿ_#xš»­êààŒ)Ü|C¢q>º_Ö¢Т¨ǵÑCB‚ü57y(ž›\ö–Nz ñ¬¨7\„²Nv¶’qY,ÉF<®üë‚Dö¼ó**æYÄ›Ô~ÞpfÇÌϾ{Õ;»‹ËE8+T’ãžæQ}RrC6È›/#$,E&E‰¿º½1N‹f]%+¶jú’Òš]«kºø-J“ïÎÒê bûÕ½ Ië¥M÷¼ú öq³\–A:%˜uȼQùQ¨Ç„ rUEgb7îòvi>¦øÜM¯Œ}Vˆ8JÜë\±åõN–k©}÷)€èéíÈhº¼‰I¦í}¤Õkk‰'ð„äbÛ6Êá¥:ǹüdª$ \Xài~@ƒ5/î4¥Œ‹Â·ÿá%gSTŸ‡š)oÀhäÍíB{`𠞟ÿa­kò, 5˜ï[ŽêЍb,Ö&®2ýÂ7 ²Ó³’W†@£³# ¼òg-x'ùBvNœ,aàZšYHq¡0½¡sLiç™t¢¸r A9„\$:潿Ê\Ç¡Ø8ÐL,YYíKAÆûSÝCÛI"·ÇS?³ À¤†æ|c7™K²#V*âú´¬fµS_Ñ„¥ ‡=I.œTJN¢T‹ ²sÁµ¯i÷9sbû`áSgZü¿þ!ù_gÍEïðÝê9tšJgbÝ^äªÏúØö“Nª¬@SËì,÷”å뤌'-¾õ› áO'™o.pž]Nq‡6@‘´óø¡‘ÅÏú )2˜‘²mæ:{-G(÷X¶ó ü¾§ZDºÔÏæNÁnŠ<è@’£çG€»þ† kwŸp7ùX›yÂþK7£âUÇPmë$"§½À¶úåyzÞâû=ÚŠß0Ó÷4ˆ? SÇè i)Õmª1ì: š±qí¤ó‰ÝÞòÿLW¢1û€ª: ÓˆY)̘=ü‡SÍÏrÉQÿa©/ˆÏGãO̓‘˜«'«²Ù?ÂöyŒŠ@ƒ Œ”ÖÕéÈ´/ @l<ïÊÅ¢†BQÄ•7€]=ÇXíLÏÉÒ]Ò[¡ø9ñí¡)ÚtŸÝ¹8_Ëc¼Vö¥•¼ãIlÜâ:ß‹]É÷«G{1ï)3`/û¢çwý{ßÙtn,®þ…îÝï=#ЯxœàbW†NÕÆòÉ&æ~TýÒ#ȳ É“Ò-î>Œ‹šˆÉNlil7¦ {ü,¶e 6yß§˜.«0…›6àÞ){ëÈâO 1  â‘´™Fm€Ø(Ê£ôè>”újªvòÐ_zR–Ö¿ ;|¨Û±ÐVLÊ VúúÊ‘®×nGòŒoœ×X· g6žqÂñA µç÷kªÏd"ŒD1ÕNí,]ZWàU" _ìýÍ['DuۨΠã%s‰sÒ™‘ÈpC¿4¶vM`Œ¾háBþÃEU·Q«Ìí^Íd7͸—ik$ ûÙ/IçÈÖÓ~ÔïËi±e€;õý‰]:ÊÒŸÚVØØ•°x©køçœ…9îîûü^¤ˆO:]ß=–ô ¤¸F®¼DÄ_‚o5r¿@gÉvAþB†Èe0ܑ΢×ÅŨJÑö€ÞÒÎÌìgž>°ë±GÿÀ¯‹‚ꆬ6‹ã% 4„¡¥qбˆ ôcØ×<8«ÞyO‚Í饗0¾Ÿ(Aýû¼C¢ÃF÷å,¨ˆ,˜¾âÛØPŒÒ$õèZC¾§…-TÛëàOéÓï6pjâ’»RêµØ€cî–&uͲj•æûoofÞ2ã×!€x3QŸY ý¢î´7)w Þç鉂Zbðb¦ÐÝöœrÅ ª± nõƒ­›¢í÷)wMúù‘çÓDïãFEdéxØ”T>ü7èLiÖwމ q×ú­Ÿ¾­r†AŒ˜¬T$“lwIO¿™æ)•|­‹¦9ƒg%Ă٠·öú±T,i*d…å¸UùgŠ(nMᚈ±¿õ…&U¾ð 0ôðC n§µárFº©Œ¢X”x[ñÁQåÓ³’Ò'ìšæu÷ÕDsaþFå²Áï¡÷T©œÿá÷°Á]Ôó^dÀûø`ö ú0-8•0ß } ®¬„rœ.h ªŠ90SR,³-×+1«èXr㺨RºG†1J;2:-â.‡qÀíkV»LSW^*q|æ„Ä Ðj9ˆ0€ Svž¿ã½/Ð…ç} 0LJbÞ¡€«‡Ûáî‚vÉÈåì?éÆî— U'“c¨âPÌ”òƒÊmáÒÂvP)VØ´.;WúÈa#”·Cû¡g“† 3Ê™%Þ$òs“öÂþ»ñ%Çe«“ðj¿Ë¢´uáV]¹ ˜UŸÀ§Øjnà á˜SÈhõçí=7Èõÿºó Ù¥ï7‡PgÐñœÚwörI¿YîæÑÚÝ&V¤¨‡½¬¡GC7g^2Ê [XÌJtT9ñ¸–ÉùÈ eVèFÁ@ÂC“)kãáî¡¥À<Ü4”9Û›S’ä髹´`øô×j‡‘øu\hZÅš©×W0>çä©c­b 0PË>˜Sc7ÕŸàSQ Hï£jš\¢ëo²£!îöid¢= Î|2>±Žbïž¶·!“¡ÊÈÕ}:= <4KE#‹<VO©Ò!)OSlÄqÛ´]dÁÓ(ma 1Ó|ÑÝ_8'-R 3ÐÝx¿‡y@÷Z´o*«ÏW‹£)ßžþœÅíôúqë²·µ|U×6!1[;:·É‚Lí Œc볘u²€¿ë Lò{BBÕ@Jàèr;à¦F‘‘„ó°Ì!Ûå´X…츂kE°õ&lI¼Ä¥ð7>'÷hºèéÅߦY\+VŸW›¾‚V½ ´à“3ÿÎøÒI‡Ô½ÜD/)…üçiŒ+§ßx»„Ûàþz9b¦Ó8ÑBIJÉo€®šÄ¶’ÔϦ_ñhÖ~7¿+|¹¨Ga¬Àëœ9Ñ—›†“óLddŒeëÿÉu+1¶-k Ó‰÷2¿ù+ìE‘Ö`ªu®Ó ÊÂà1Us3 $Gð’‘ügÛ]’1Aƒø_^N¸Efìå†åànVzÄ-‡e©Ô$z”ÜV“°Ü$ßÕü}»§~q!ç9µbp&ÊÝySÌô‘9’[Pv­¶A_Z!ZTvX"™‹£Š‹&Þ0?Ùs¶Ð mdMv±[äm|Ì£O#ÿSÁø—Ƹ‚õ×_+ù%sç §,­4|ƒ:@ìl}ãTDÀ&‘P æ~Ú€z_Qé2-éRÕ%Þ¤ƒZm÷ÍwÓxĪ„‡N'ùÞl ÕâO|–s ÷"a7Õ·ìJjÚÉ*f·Y49- b­xÖ+8Ky¦²´kŠÁ½0¥z5'!?CŠÎ.,ÿŽ{×—¤@ÍËÕÖøf*qüØ"ÄÝ(lè¿W²§b áömUkzžDG›ó›—†³5”TóÂK%KÙñhùç£;¨ú´ë^JÙ(nå4RØ}hƒwD,®Ã¼>º‘WyÒOjÇL‘gFô§¶Ûoni@M2:RTûæè(¥x.E<§ÿ\±}É6¯Bûç¤1ì5€ö]éDi0ªÃDë9—p"ž•šOŸ åV›È~zíü›¡E¼JKí4ø }ÚQ¯ØRß·xµ¦7VÙ6éÙ‚TÚƒâ2ñ,á«!%HºŽ¾Šs1zŒ¦¦H ƒÎqN)šw‹ƒ6h÷í´Õy½ñÚÛö}}¸®Î‰}å¹²xêåÅ(ÔüLXdY™eusò: âg”1gïmµ€“v¼?ò6gÔÿ+šA$‰Î>» Óï¯|T˜vK±»·ÂCHˆk”»žS1ª¢[ó$çxüBúžH…á”äXïö;œå éßÊl&ÄoR~òŠïÇeÃêä6G@Úñ‘õ3¹Lи„ºOìC~‘¬Ñ¬ŠL¢í*‚:‘ÞS úÿ‡KÊBSܦ9=H‹ÏhÄÃ×mmÈ ¼I5 ? Õxò’*º×`Ž”9:‰žX«è²qŸ+‹v3µZÙ=ˆT.Š(b{‹¦Î'9V“ÄŽ¦çhá5¹¬ZO«°½xìÌvJ‰¿Bö›‚?n@À/ލçׇ³åN<”ÏÝ ÿNU¤GìŸÕFB>ž­«Æˆ\§í•Uè£cð9ÍC8´P-žÝÞjCEóµ+SŸF¸àãét•IÒÎ5Ÿ9MY¯0Eér"çzéYó†N“Ý;~€å*Óé\— _º³â]<¾PñpçpÍÈ閭>=D9Õé>·àˆ¹1-˃#اî`OÚóGçy˵ïÒà‘bkǯÚêÀé´Np„èdopUs pyâ”bóN€[;€"Tcw°ìq*l˜öÿØPZu‰¯Þs–Üž»ñz·PǶeHî;ÁpI,3S,"ÉíxévoÒ-ÓL½™¥?Ôë\­F‚UeD"];|ÜP9¥O”ƒ'*÷Z(fä0d,í2DAnyßsNGŸ2Bð=ýGþ‘ø³$ ÏÑYú’@ûG_¸†Ìo Ûà$ÂÜ­°ÿÆ(š>£È†žeêaO ^ã²62×§@‰È1‘qsß±å?Äw6o@މV'nWÇù“ú(€´£ØyÉ.gFVQYäàkcú»¤Â8ÝÊçÚ’ìsý@µÉ@4¶óþ¡ÀÙíoͳQ£öñÝàüku÷½ö~„í›Á朗I|ñ@Òº®ßXG—«Õ¾cFÌYD&c6Z#q^”VÞc za«þˆ;ÆÔ)dÝÇ %BÈörIúws†.{1CÂRµW¨h¹G`V„Â\Ä-ŸHˆA:è½`‡6ŒµÛõÄØdù2ñ¢B“µ++NG‡G·œz«¨}K¾jØdH''YÁp&4ñ½`QV·$*¦F^OJÂ;Ç 9åx`(ïžeSQ곌çVÝe;×w‚¶ÞÕ~Yl¬Ï ½?8ï?H†›ÛZ‹Ì¹P1ž{;6Oöq€¯€Þ·isjÁ†iØ(œ%\#ºgâsX Âî[ø‡þ˜àTÚ"PyHh•¨Cïæ²y¢0üä?PòÔ}IúÊ ½Ž}!\Ã@K#«ŒWé­±®å¿­ø8á #»¥ùAVm1ï²+<Lc ^¥ÆËšhö8Z_©5f²Î”¿\ >ët­Óør dÁBw”âo#QÕä‹ÌN°Dò‡cïrŨºªcÆ` –£˜OÙÙï©5îÝ81äž;ÕCXb"Ç ^@r;ºK&>ki,ˆ•žLv“ê2fN}ü‹·›âÄ >¬Èh`r.û&…ð‹S@ˆmü0èCE÷É쎥Ïñð´^Æž@t:åʽZW"ÆÇ-e&×_»Z[%.A{‘áu´*hô–>]ÝÖ(G£nP' ܼ‘b°EsÄ™Ž0ä¤nº–›J]ˆtÅÙß e~j™;(ÈzÝTžÚн`¾ >>˜ï„ÅæÃ±Êîa)b©K¹Œ t,ÖG-–¼dà §QÞ:ƼB?¬Ç°t.”¯– ‡Í«\÷žøïU¶Ù¡27~·~2äVèE ù9Ÿìˇ¦ÿ»îvz‡qI³«šº“>Eu›Jpoyû$Ùä ã&6cÉ#|J¶$ˆfÑ-ÉÖjè?·EÏSŽWyìb>B­œË’Í Çmk±ìWýtâU·še#çÇk®s‘´á¥Ð 8§Ž¬7üˆÒ/¶#˜ÑŽGÐTëMöï¸Á<"O¿±§-èÐ4š §÷’kUšMYš³¾Hw8»ÉûG¦c$:ÕIÆ ‰ïÁæ,qQRT­«Üu¥ÎF˜¢à”?‹Ë`k§¢Î‰[³FñAOô)Ö‡ãï++éé,¹Ï¡£$ÁYè¡ÁÄ7ÁPè22?¯4)Z$î¥òú‰¥$nQƒÜ›ÀÌdƒkÇÑÑ ÙF0ç[VK™,|Å卵Üä³v¤Em+Û‹ùz±¦¹‚0Ð3…&ïÒ,ëdØ>¸eȧ÷•ás‡‰–¶øcnñŸkãs ˆê,&,þ kùl˜µ%JïO¸¿Þ&xøíTtv†¨HG.ò@ñyŽóÆßD¡PêÏÆ«oÄ€Ý õÆÜÌûÞ<¨ƒ”; ¾¦º}àrcàœSÙÙðˆ¡7fØõÄŠÒa¼Êù³†Ð’¿eh›nƒg¡Œ<" (A:@ ýö:(:Y`s£NJ}Fé –©5;†^5…¤0Ð^0}žéÙ ¨þµ´¿.µ¥ú(jÏ—Ïg,å-Äq¾˜•ÊßÎ Å>Â’Í_~‰‡÷g’Ò¨@îbN_é>DF»áN®­>ô΢VÁ—¬ Gxqô«Ÿ%;¾¬ã2È&×㦢Õ1»ïܲò¦>Ê^ !]Y¶Öú±i¾0#jþaE§Že²Ë&ضl8"/ tFœÅæ@€—Æi÷…Ð[‰h:qË<†¢‡ÇqÁü™Y©iº¯°z@¼NÑÆ:¯ïhkP®ßïphµ¥zeÒÖUJ´ÅÁ=¬NûI¤qˆ¿êWŒ!â½KÒ¥/0á²gäI/3ˆPiäqMÒQ³–œCCž)iG„Ϊ6·»z?j+Š¥ˆsË9ã|GžŽIÞªáDV¹oÈí³Î u‡njɆÒH‚^ó×Ñã×ðá§Vì21R=ÙÂÝö ?ñ?)kÒ›sã)Ðëz^#FD†QΚ&%ö|ƒÊ@bÓÒ“{jÿÑúY íΆú`ŽY"@Ç¡S”LGDÄWšƒ«±’!R~‡¨PÉœ@çÄy¿4À ÃZˆc‰’«.Rd05J“CNŒbþÊà/R\1KÌOé§fÎ1wPûq{aaô»«åü­˜Ä¶?2î1«k[óØãelbEI5góÍ\~s¯"u¯*ì×%p!aDÈ7CÊî¬;©“ògÁKË•íG§ÁùsèüzZϵ4A"æÈ¨²ÚÇì4Å‹]†~l¥SˆãUdª¨÷?ÐP’¬rãéÝÜö„e°—¸³ŸõN…­Æ2±ëÑ ”¡ý=ú)ïYk â²°"ð¯Á‰†âv}c¯WÓ:$/ˆg Üù»°rºÞ"ñ|ÑÍæG]öMäH­ÖNÄ °w{<+ êø@bZuÁÖ©£¶¹ œ«å´ Á¸ Vg<úÃ:ùñŽÜª4ï.eÁZv“m‘Éíí?0¤GÏ’ÂõVN­¯³—e”·C\ôG›aÛÔÊO áÒe“AtÇsɆ¥ëû;º³ˆ‰$Wjò0,_£Ž“Uô¼ü‡ì GܸÐ&FŸ˜/!°ÂâqFþB &3»§¤Ù¥=Üm= €6 œdiáÉ‹­’ž¡m/?Ò¶ó¢EOŸ—?öè ¢£¼"3ÕºÅ΢ÿ6Öw ‡€GûðÑm:žCÆLøï·t‚,@P7¶sw£mTÔcwÑYì?ÄDVB`IŽ‹Å8/˜¡f¶‚Ô†s´<ç–B”°Žb–D¿\CÀ†ßmùìî`JÁ3ÀYì 1fñØßHÇsÎâY†þ5,IñúÊï1ù†^Îß/Þ¢ÇßX•8ZËz÷ ¨,pW6*à™ëu¹ Ù9Ñw‡DBôK‡iËh î @”Ü;³Õk‚1úÏ:WÎŽ{£¢xbªøU¤^1Ç¡ˆïšhË¥T*Î~¢Qïr ^FWûý6‹#[ð&£+ Q„¹ßÞ×6¾m@Ðb•ÅåUX›@ñ]B±®+s¿ûF¡Dá2ÓÌ™ µ”’—pfq•)´•XºúvbûðIóþ!¨¤'C¿œÜF•8ŽªL'Þè¿Vͽ:foŸö fÏÆh30Bx„òkúYEåÉ0Kôú7½nÆãïc Æ@’mŽqƒ³)Y}¡R6¤¸B½…$¯D†¦‚ùŠä•Tú˰V¡×ÔAŇT˜˜ u!ñ ¢¯{Dð²–Ëæ5Îø }îÑØÝN”V¤õÆ€Þæ|¿L$7Žz„ˆ±nc[›ql-¸HX^œ˜0šóFæuÎîŠì«;ƻѳœlBöŒë¨ô”5Xs*¬ø ç…Æ±™Æ£~F£²¨¯Þ3ÄùH¦R` Š<ÂîÇ DâØZ…·Ž 7Š<ÞüzÖ_ëºádÀ}u…8§×^Ü#<ˆ ·.¤ÿ+~‡àÐb¬¾áo*VëÔ»ÜGµžÜ¢$ûõ}½OЄ¹GùšømÏåOí"vÛ«¸waB‘å|¿wÏD&ddÜV3‘"( îûÅ»>vNÛ=ÿùUåpiE‹½è í_2£mÝâ=hïFÃ35<—oEŠ£í\Fâ#C¯N€•)EÐ ÃäqÐ¥‘—¡pê/¤5­ð¸È"F—wúšŽm4-ÆQrØA-ÂþZ•Å^€×ÝrG½)ýüuõYj¢•{ü‘ˆõÙC%!TèÄÛZ1Ù—Ièîm ËTü`¥øh³pY8ýò1¡ %åìO^“‰ˆ¼™zÔ ¡Û™Dµ~›mVÕ.úbÐrB$qFîC× eÝPš#˜?¸—’H&—ÅQ?ªÎ¹ çÞì±âF(̇{Ž•ŠíË€E2kX!‘þ|+þl§AÆÀ;pªè÷Ï{Þc•ý ðÀ#Q»Š‘/¼ôWrµ8ä§Â~EÄìjÈ [–R†\ÆIÑ“¶t “jKàO^ß3=¦j÷2kÏ9£öPuiëÆ§½áÛÈÃy¿î” Í °åþŒ|ìXk¾~"(£-—.*}²ÔÓ­›#ܲM‰„¹ °´D$fÝjZvŽ€l"™§Ï°Â÷ŽúÆÓBˆ³¢ê)íïgIdø§ícîs„oœ¤×ÕÅÂAɘr±ÅC»†ÑÌ•îñÁ´’^Ÿš•³Ùa mœd¿ˆ<‡¿0F„ú¨Üº e¨XêWýÅ8èó€ëô‚F8.C~ß«0œ_¥õ¢<ãÖSãŒÉ+6;¡hPàªs* hŒt›Ä4:zÀ’W'f0Äq>Æê|8Ϙ:,N¢™±)Üa«m°0@Ê”8á>¯ÖÎiq]0+êrùuÿ¬ñêô6Dlðs/颉«ªB€¾NÒÛÛëÃEô ålF<Í ª®îdNšþ Æ# !úŽ[°PêΡCª’¢­ˆ6vjOÀÄ‘†ß“ì“ÐÆ÷ajYàÂû'÷7shcÙ©cy‰ÉÆdzdzNxnÝ@nò»ÚÆ:8ä4%k8ä¹Á-ùbƒ¤K=V‚•„@˜# Þ(ÌÔcÒ‘¦µ¥;F×¾äíp=­cÍn±d±Xr¿`æ¢X“MFêba°ˆÒ¾4€ž0}b¨ŠíE<|dà ŸŽgª<-ì¹ ›Çmc¬~•€¨¯Ã÷ŽéðÙÅNh2öèœÀˆòâ/‰i"G˜œQH‹v6¹Fû‹£½Êfšåÿ!A‘Šs@ ~ÖòÅTX×Jœ“ÓP¹CÔA–ù$UCðTQ‹'÷¯¡_XaBàÔÏQesQx&: ‰$lÀ¢TØÂ¡1YñŸ%Õ‘ÂÅòSÙÍ;á)»þ'ò´ŒöÁ”‚žÙÝQ¾ß“ËKáôe÷£ÃÜNxõé˜'M¶„5ÚДJMÒáµ.Mh#æ&´Ô5zÆÓ?  †$ßÚ¼éÊ\ÂçÕñÆäíPCtUñ\±ûÄ5ùq x8é—"/.¹–™¹æÅû¨ÑuÁ¶¨3êä"a4jÝN9æ.Šdât?ª‹‰i†Hw…J\`›h,‹ØÁr‹˜ugÂÏ6ÍþFs+$p´ô]ó\—B¸0à~ÆËçÒùâ¤55®Ámãú,Á’‹:Òª×^4²òåß1ÇS—rîI×Õ yÅBÇô§.`™ÝZÜ× EPn³OD\T¢LáRfçÓ´É¡}MNˆ¶¢ î|Ö„À›py»Ó+N³ðõTëå™%WEt Èãh”þdî0Ùy'¤4áÑ÷þ&Ù½°¾tä…®ÁÙ % ÙXçvÊ7¸ßƒñ(IÑn`ïÇ~úµ*sîAO]½Ÿ>=M$ Zìˑǂ7³:¼r'-Êþ…ž‰õpú’¨Â˜ù°Ð»FŠ~¥ÕÿÍ©áï÷¸ªÐÁkÌ6UKO‚˜Y­GøâX§Å}3¢H߇Ó>ææpÄctϹõ6ŽÕ¸vw”äÕf6Œ?6ÏzMŽDþ®güƒGE ,ºA¯£ÚZê rbµžvqAãçóõs®L‹ðzL> F_“fÈT÷Ïóƒ‡nñ´ž1÷çœCª yªþµå'Vª¬¹h¦B‚²Úêm½¥ó1•sh5Á’þšÀsLŸ5v3@sðyl!2cä3½IWaˆS5þáRiõt2‘IšEq]¦POq_íô%qЧ-Å¿Ë $÷ÓA‹v Aeà±³h|r¥OR«C~ã*v¼>ÎËGU·üž¤•þŠòV³Â«G‹2$ì;P0 "q:Š£©I+Ç ¡z[ŠÁÀ-UÕÄìÞѼW³WGÔ;"—¸P¸ñ‡MÅ%ձɎÔ_&(|Ká˜è´œÕ¤#J(¦zªæÞ[aX;¦\ºqš¡‹#ˆwf}ß;šô53?nåÈ\#G¼”ù„õvæ¯øÊ:±àì›h]f‹a»Û?…怆“1ÀóµK¯Ï=i‡WÁq7À‰0‹gXñÀ@jæºE7*jÄƪÞ+UÚ¡Dá,º8ëˆÐ~Êd´k:6{OL—Úœ²UƒG¶+Âæ¾Ô ßÈÈëýù»6ØôDBйx¬˜žà°kE4ÏíÏû\¡íZõ;#"Š,@Âv”†EpS¯\B|Žq†=Ø4`Þqô¶Ø}(Bmxé p2ÿzuÆ¡ãXkqˆÛÖQ3Éq?ÚPk:÷WÜÜ“zwÅ 8||É+´1b˜BdGÕ/Qu\ì‰Ä?SÖsòi7þhq,i–H#r¨k¤íM•'X 5˜-ébZsÖâ<2” _èq­s=+ä“ýÛÇÚAÓ¡^·!é?@Ñf&mº†K %¥^Óž4áxâa»‡Ò7šç”á9œ©È™>dQæû;¦(H±I£CëÑÖ̾(­sݧŠX_¯Ç…Užèô$ˆeÅu2!ðÊH0gÉ¥5ÐSãõ8êOgÇ!¸rXn\E²H¸¨°¸ºlQêîݯ Îàr³ý¾Û õzÂpÊS6Ä“ ÿÞÆ;O²N3_ŒŒƒæËkl›Hís\ò„•úѯ- ÆT˵‰F½çŸ,‚J\%@ê´´8?aÀ¯§jP®èÂÙ°°ŠHäȉÝåf¼%¥MPºÛWQÔØu׬åu’V÷³Ga@û¨<ÓåGïKäÏ ¦Êý’Éú~ªö½«ŒÂùïëÏ]Ž'çÃ(ic+HUá“ÞÏ]‰.Î’„£$o×àtê©jÕÔꎪ¹òzöñràšÓ#Ó1OÍÆw&ÈÕ@ŒV®h2Š@ÝŒ/Ë%eGò2—Ä:t×È{ëÆ7i6$Öãű<ª60Òëé’ òAOJè5LP;Þ0 ù•A­|Z«jâöüA)"ó-`–ì²Cø¸á»©à?îm…YÄ~¥þüNl€±Hmks£²Ùù;:;Rô·»º€pnüyxbl¿aZµCnCW÷´_§¾Š¸ºûê~Ñp‡jAëŒùuH€DØ-æ#›Æµ‡én¨ ¦x¶7ÄåŒlšN‰G†Fû›ÍÌqÕ»h¨Î±öïŒh1F ¶¾fRuKýQ‡M;ç?‚K<2–¥>1fpŒº+%c+§ö™+/Tðã ñ˜Ý§¦sÀnÈ (« x(Ù{®ÉÂÆ¹†Þç£ËÇûUä:BbU¼ö“+©“¡Š=–ûaQ(,x_²˜Ç€AÊk³§2jÚ}r,õqm˜b°—Ø'L5厽·G•%ý¦¶ï¯®ÉZÂï³gòïß œ;édIr’L“K-{5²Ï„†ÂdÞŽþQ`nuiTßÃ?—íB”p14S«39¤¸U<“ˆ@§z÷“‘DÊïå H¹qõ04VßxbºŒô!7³º$u‚ÝFž"”Öe4æÞÚ )c6rÄhÌšŒf°¸+?ú›ZI¶êóÀ¡n’0l^ÌL1‰V¦é©&~ÑF:úb]JòøÈÃÊDs'³ÂW_„D‰zY—žŸnݫ׹V}ÆÄ-)‚–Š6¸”.EÉU¯JX·b"ÖfDÇeõrÀ¹åªè’.ÌoÉã­-B£2:þ»‹)ÑËcfÙÅhøø&àO£}fÏA6›Ä·]}J`Ø£SpýÝ¿™s[‡ –×¢â<•¬ ÀoãʨûY¨ŒÛÂKjëÈ D}O/š"ƬwÁh‚iI'X®¾®â~äþ¯6…fÞÙî x ‹ñu–ºIÆâõÞW }-M´o¾·Öó7ô,uÜƤ‘ï&k‚øïrTcX?¨\U¥³ê‡ËË»?«Ÿ€Êùl/S‘FûgTcÉÈTs°þ"okzZ/nˆº8m@o’pé+Ô:øœE u}Ö Ç½yAÄ0 ¬“túk³33ÊÃÅš– ·¨Z¬ €$ÆÒC RÕ×ÉÈðŸ27÷<},BýK9ÕK£ì3(_0,§…·G(1ãf³ÌóN}¬¬‰¤ñgîé% ¦Êwgu ñT½Ý“ÝwÞuŸú6`R=—<ãªÄfíÙ„$¥+ãöÇ…eÉÓ÷W¡»|W>{âòru V@àÐ,v–<àròÓ²€^ÏCúÀœÀَ΀UzV•XÝÖo¥ëÙ%/ð5buÔ;&Ø r}÷ßyOfe\—®”Ñ2zÀÑ ×ß Ç|Ç„€PZ}tôÎ?KÒm¢ì`MU±Õ÷½£¤Iö•™…Fþö¯b’ެçUUy&+P5³£= † p“}wMwˆ9jZö£ÈD#W§ñJ"•=CÝâq–7%#3ô¼áªPÌhs™L¦°—©) 8øDí#ˆô1QÖ5„à5gV#Wÿí_Úóïè:¸ELÊWJ”8>øþMíIÚKÀ¥^ÈçèäÉ¢“ñjõ’e–ª v÷ çŸIí4“ºÁþ²Ñ²€“—ã E-æN‰!^ÛÝòÌteaRWÝ¡JB éyòx䄤_Èz’úÆ*-¡ðÐ)ÂQ!/>róT¢ˆ_z!P7¼ÕRªüV›³ogßáCÿTM ¹ZÇ%èÅ)l¾*KXÔÇ…Iɤr„ýiw‹=µT«:$¼E^³hóÉJgÐa°´¯í-·ÆÅ²ÛF–÷ ¬hE3Ÿþ?Œ8͘D„ ZO ,"ÿcƒÄæšïŒü,‘ÿÛAßÅ›2’hÊÞ (r<ÐHºÿ”A5ÞjL¾#dupº¤ÊÒêžy?ç˜o8Ç›ÛzÞH& \[¢~•QT‘£Ä)K¦8†ø_„™p¡Ô™œÚL¢ôƒ²öÕ i­HÅ8ýB߉ÃÝö›oÄ[I0Âì/žÿq?g0Î!P ÆÓœLó9ê7d±d´¦Øc7·»ç¹íŸke„ÔºåkË숬¨|Ão‘Ó-/übÅT÷uPUuü!ÕæLåƒÊ7ÏV7òY4ªÇ‹>Õžç·)žt-dËÓB®ê˜ÝH' Ü[Ues  …tp'iïv£ãI/ÿêÑx ÿàãEU$yµ^+£xà¯vÙnV<`À+B Ó³NÁ8ÃMxƒùñ•=ãÕx_Šûˆ£/s†¼¹‚«!îŸL8FðŽT–ϲ%î Rû/']³2•þÿ…tVpžá›V8µ°-¨+ôؽ,ð Mé©î;a#¨÷’!ÄoŸJ.üöIw"S WX.¨³^_™ïkÅÉqÑ X!p›þ*Ñ]Q÷¶Ž%ïž³›¡ò]ðU¾Û\UÇ©T‹:@w*}4ß&‚î[Îi]$‡‘ë?†èÝã8b§Ð÷jŽ`‚©8{2mÝNd-\'󎱼 ]ërÍu‰ýB5é¡€qh{LT,‹3ïLh¢'íõI¦€' ¸;åe(dKEWEôjÒŽ´"Xà¹^Û@Å3–óÙÁ@»*ŠEo–®‘Qõ÷«KëÜbF?{äõ»"Ž@ÊÀì6£’Þ‡;yY™ÿAG€)-žôtzÀϼC ÿÿÒ>zUrŠêmo,DìÔÈ:Ëhç[—ê¢ùKú0†€=o‰ ‹Åãc¸‹$…nëg«k£jM% '0àj¢«®‚ß)sËPJU7b­NôQ ®ÔÁ¹sgUºT/%o]·ŽÌA¨Út´ÝÏÅ[†bÁYoüÒêò,dP› ±6¢ü§¼5¿þð`@aàÇÈ4£×륢ƒ°µÄe«yÛqob${o_–oµCªà(ûÂnžŽÎXgzi…õ±ÔYž¬_š+ϧÃ]±e6œ‚W¿~PÛГ›˜¡,qŠ!3œÌsÊþ'Û È®ÛûŠPy¤©×WDhÿÿ:üÞÿZsVÇ‹ -{ðÚØSõa=‡„âp=  ®IÜÜ‘Pt\Fô^§ÐCÖz󂆈hŽ©³8K?È—UøìS:Dš˜A{Áñ~ˆñGîwzéYl¹DQß·[Ê¡!MÒ¥Æ2¥rÏY<½ÆLSø í:ÚPNQWí[Qþ÷ñ‘ ôúe<—Ëc¾1l^ `ǦޒI¿›ÔŽûY‹»åi”l4‘bai"!Œ{4‹´|wU°â{©ÓÔeZÀ”¾vBl-åP_Y†4© ŽÖ¹EªPP˜Å"ÝFÈ0~$>g3÷ü¯”"wÞum˜X)p·îÜ-h€mj1ב©ˆÄâÀ«ª´–ÆW›µB*Yà046 !fB  L¯j*ðÀ£ŠÜÜÛB&Oshà>“l‰qï@sç”3L˜ÁHצTêÔI:.C’«ÂB³a‰{Ëô)–‚Zœûrß8rùžgœjx O&ìR‰ Õvò6²š>áK¯ ›!Út‘PHe»Ô)J«˜¨LËáÏVªOƒ CTʵç *†¦)Uµ{ÇÙÓrC …ŽÊ’ XtA[Hé¨|8å+[Àsˆw+Q5 “;œ”ni$ñýº¿4ÚQrõwQËòIÓ¾œ'ÓúÐË ƒöq›8\Ë7ÙšÒ©ØLÎrXôÐl"å•&} þ*}·EP)ž!¹ÜÔŒµz™³´Ô_“è]¨~b_¯ í}C+¤ïCe¦S ˆ´UªÕвm62}ˆ–œ.)Jç7 •À~*ÓÂY™¬ÕWÙEýÃ[$BUƒºÝɹ`ƃ“ ŘuiH±'ä’Ïd2QTJ9®÷¶°ÍR-J#Š‹qy`冥”Іß@w‰ÁŒÊ©ØÞHžé|ZÒ§õ°·Ñ;¦·´“èýAi_ÓYy¸G^Ø¥eЬµ9´Gƒ‹½ƒKiϦ ÷ÉJ ‡¥\¤->ÌZ/á~ä}+QŽ®;/\3¥ŽyÇÓì€52xx ÍÇö´Åˆì¾’ca w(Ehǹԓf¾m¿Dn“ù/Áa¾‰jÐlWIÕƒ÷‡§óE¿e™Îòp¾OÛÿ=Zän¿„Õ;šü¡XŽ*†y`ƒÛ˜3Z͸±òøÙøÔQ·ÂÏÎy–棞°U4’~‘Y¬FŠ©ÇNî“SæF³+é€Þ“ºš7¹÷•n‹0#9\+=g毤ب[¬Ú‹¨ícº‹–Xn:­3‡›;¥çZž Òm½žaþN-4'RÙ„1ò^. !O1™·iï!ÕÙÚh*M[í–%÷urIv )ØdpF'Ç sáJoVU‹$6¥¹üÂñ^­Ë›ý}ª¯•‹=ØÅøHBy 'JÔß‘3 ¸#ëÛ_Ì]dïUˆ¡êþŸ £©ôº}΀ºÔû¹xoÌŒëýØÙI~DÈ“¾Ó Oh@Å¢ŽYJ?ürÞXw‡cÝÒw ÷ìÝlqË Å÷ ……uC5-[)í8—ªƒU–Þ¥ìýØç]íe6³a¢q ó<\§>ŽÂ­ò›®žo¾¤5uÑgˆÄ½»ÞûßCàΚñf6ÈJ½Ã´À©záh"MBµ@ׯ9ä²ð›®0jÊ2tùª®2Ά™âvþiÊ'ùsÀ+’ƒë~wÐ.ÇH$º’BC9É¡íàjæï¯¹ÄÒ‚yƒ<¤U(ÞHRáËÈâÉ>ñÌ9’¸Ø~8˜û(n—Šœ8_.” zü´Šx|aÿçD¹ÄP“C½>DÞ[&Qû¹Äº¬zR&L,i5€nuÿjüd'æàww`‚ª©NÍÑå·ô ð‡PÖõ¡ö±iï÷¸Êb÷wTòèIh„®ps‡58Ú¥¼ƒa2YM«é¶ûb¤8p)ÐvS0“צ“,:oµ?5úçîúÙxM/ÃJî ï”öÔ&QW«Fê—W€áŠ­Xj„[¼JN†¦H…È®½œÌWX8ç$÷»™¨â)/˜žv×ad¤XJÈØ5æH#’ê“õ Ö â’ÒŽD>¹zÐ#¢!ôf Q)R Å#;‘}Ò¯Q<õÁes_'ÒCó9VÂï{ž¾ž“hmØQŽâѵœþÝÅÕ’z–ÏZ3Õ”òyJ'‚îkïµ˲ò¡)÷—´x#^'ÕK-a`»eÎo‹ÿ#mçiADþ§´°ßÊoCPX sSCõnê38+¶›8'þ¿³ùV”¢’ rxñÝz\ÏNù$³d8œ.Ãu¸°€°6JÚ*‘€‹™ ÛõâõBç¡K¿8YÒ“j‚SÒ‹áp)ڥЦpVRÿõ!)Âyp²M±½‚ó\\[‡¬ò\“8/š‚1üE€½ƒùæÍ°É<—)0²ýå2¸<¾o,¿ÿ5½¿m0…»#\;§›Ð}y îH`²Ö Ûç©°½%½©œk¢òÀGH8"¤öD6|¹ÞA{hÛÍýûNíX=ßçiÔ„¡,ô¬YÐÓdÉUç‰Ùi ‘•ŽG”„ ~¿ïáY4ñ §¥úà’O.Ól!§ÈMøÙ\=—±²á×rEÞ`X¶gü†Ïüǰê(€Ï¤ ÂsRRw(}¹ôNˆ¨˜&”–ZV‡p‹ý5Å~סo˜›N/‹º´ÆR>,€i:ÝÔ`‘léÛ9Eýœk>[Ã)JƒüæZ¯;Þ}ïT3øÝû5ë[pî`mq»Ãä(‘dSfϚģÚ/¯Å]ƒÆœ¢—É;Õæ9÷ëšs0k9ÒÜNL}a·í“Jßs¤1]á´ÂVë·è¬„ßcíìëGü×ðÉÁ€S·?qi ft™³*0aß\¯»ÏŠG>õ@ˆ©®‘ƒ&MÖ ¥×Ù€2îO÷293Âp‚G|$ª¯ÅÔês²zÒñü ;à‹~ça‘rL /6ÿ;ûqÙùSß=\6d»´­]œ`=‡hÀl½!št­$F ®a]._3u_‚çε®l´™1¹×Xøo¹Ct‡-Þ$›ˆp¥PÊ×p’¦ÍŸÑ>f‰h¥¨°‹Q¬(Ôh”d ¹‰Õ¡91Åå(è{ôßÚ϶±Â7#êÀ)0.Ø&ëݳýÛJÙmÊÜ2k¹ãθRóƒL?.óŠg¯^K˜>93ìƒ]–ŽPâà»Ù`Þ]¶y‘íÂŽèqõÅ–œwnµæH=´†÷é;fF¥Ì‘&ß[ºt" (úNç¯Ýqƒ2kÖœ—{R `Æ õ\§g<8]¾DlÝšTեóñ„ÒxéŸÈÔã‰5NsUd¦åmy=A¥g¬™¶{¥é¶8ÇÄhªúäaÒ' å—`Âù*±»jåäšv‚´â„æR•—«²E_WA@z“ŽuL×yP÷ï{Œã¤áw’?ßm²z¿Yê'¢‹0R,'aA©®„ ýNlå×Ôhsao 8p5ZvÔ?1žÇKþ¼eÖåí”ýHU„+ïn­Ø+L€o&kvý8Ë#Í'¢XPÝú s†’V.uÊ%=ÿœY«·ÿPÌ1¦%«^ù9 pÆtšßŠq“Ûç¿_Ï 0¬åa,ÍÆÍX "õŸ¿VòÁag¯ î‰þ]í8š%)³ÈvE˳n^"˜…©fìvÕ½£K6P¡äȘPl]¡„FâAeú¦hªÀ4é¶ ¾8‡Rú¸±Pjå½÷/SɈu²œ^A)<€¡]XhØYÔš àÌTfP—wáÁ?ÓÑMãÁïLÑìóøw¿)‡‰ý²Ëý·Ómÿ§‡cKN˜PP¦ºb·¨{ì~½%:‰­ŽpE¶š¶¿n9XéI¶ Рɂ…§YeD? Ž÷9.8’´)Þ+Qº â¾sSžpÁÞ•MÙèkõ"$ÂÕJ/%jh]߃ÏÍšwÒ“ÐV5›É9S­Šöµ«¥œ€b+¡ûêÈfásë©Õß¼#ß®ÍÑçR1‹Öe×çä½·=Ÿ·.?ä«g6¡x¥ÆU…ûò %E¸ÝÀOUÁÉñè(5SÌe|®XŒ[êÖ÷LX•Šax¥¼’ŠÜ /ˆÝSêÕ¹+¤|@æ£ýL3›ßhÞóñ¤z8áÆuvh ó*&Çðª[ǨEtÐoíð5êôª~êÅ!É&˜(Û»Ë ¯è,xÞ†0BÍë8S¿H$ òQlŸùœ j Ô_€{2Ë Qñalr°¬½ñ"j‹h_&ð-'ª:Ó·²0ÔÓUÞ’8§zõd«$xOÌw†ÌZuIQŠ–0oÏ*V£/K­”7 ·Ë ×;@ýZ3©”ªö™¯'på`íÙÓàƒ™Ò0²íaqêàöøÆÉ¡Ùàƒ Ýþ}mж¸å¯6q]h’¸“¢ ÐÔÏ´iÌ!âÊÎl¶í‰þÕ ÑYÆÅ6/k%µ‘… § à‡J%—(×Ú4t%©2úœ3Øeég=<Ÿ éd/í凉îæCô&@¨´RQ)O½4Ç;ÙG%mUM~¤(?èfWo„31!E·î L÷I:Ë ×UšmkQ£áÒg;ÏÔ' ÍÜ^ażߒ>úɾBÛêæ™û Xq濨‰V ìH"¡F/éÅ[gò#̳…Yý´ß ‰EÃ$ Í` „p˜íCI'¥.=ÿÄ<ÿÓž„£<§¸_®•œ[7YE¨ª7ÞG±zsâáììÂ|ñCƒ×A¿6+ꢋ,yJç#´˜¡Îî÷¦†üÔSó–Ù“!A÷[!–. äH\}Ãq0–œAb½MDÏ¡{»ÎÍÎ{ÁûÏ6_NŸÉÍ8Þ 'í6ùJR³%üYøÁõt YV:ìË­jD51Ƹ(®ÉaµÑë‹g™œ=ÀÞ(õYJ}LåM$$"VtÖ3ñER&~µ:­‘0¤:>5áÕ‚{î<ì1e¤íUÿqEL+–±’AÁ¨â™iãÿÑ;âà[Z€ÃYG[¬ÁÍQâ%¹”O+'‚[B+p•”,©Í¥Uè¥×GwŸ±PH“mšC=IWçLIR[¥‹ ž…f” úû›Jõ[•µ} p%¨Å„õ¬}u._VxÎëägZ%†>ˆ,z€‹Ü€j‚¯çÞðûí_·Ÿ«Ê M€­æFeÊØ¹!.Œd:«ô~×s®³fÑâôÐÖLr•yئtÓlz—âMœµNÁ·J‹Uí)YKS#kp©æÑéî­ ówÂÛ¤®ÇÂLåw7ðŒeŽIDAà ¥>9¾r­’£"À8p>NQ4 ‡ëmt!#×—h%ŒWωxÕL§g{0Òò†Œ”qŠ’ÔDè¼ÂàgýÁüC_€«»ˆÒ>@ÏEæ ÐLó–0ã@ïGÚå™ o ¾:m”íሸÛH3šê&&4ï¾&_‹”soZ’8Ö^D’;uãªðTxؾ©<"ƒYûñJýóJ0B3Ùìk„+©Õ‚Iÿ'í.håõ†l$;¿ô!æsTi>«\À÷Ô–à]º§ìôrp‰S£p©±HýÞ™<î<ã݇pÓöÐý÷¡…ö<Û%Ó< 0Š =œh·F€³Ô±‰Ôê%8ýTü>úpªŠ9™°Syù½ˆ­˜}Æl‘¢å±_i^ýŠ'1tœÑíü¡A iz5²ËÆëÐ ~óx†ßtŠLfàvòS]6Îܵv—æxe[eÎEÝÿ;ØÖ«TLÐŒ3©ElÔlÃnŸúû %¶ÞÇÂÌ2¶”€•¬ó†!Zàb: v#´O÷ªy'\V­§ÿ—†¯‡)Qh’u5“P*/¨qÞ÷oÏqyÃÆÔ“ñÙ‘f/É(øêßÛŽoy³"ŸÐ' j[]÷IÜÓï7¯šÌË™;å4£)æmÝmþOGŽØí¼3e2÷.¤$ßžZ£¼bØ— n½Ï>ñ }v’ÝDb%áµt+ŸJ™+>t‹§ÈžŸª ês<âç6 ¬¿Ÿ¢Â±¢1Qã^+z$²»h,•â÷ªã`#ëõÇ ðO[”_a‹kWç„"Á ¿\²×ž^Ê l~¢ÂÇŽêÖÃóÉ‘Ì7¹¯°f%:»ºÏc¡v§‡ÿÜó¹2 ='Óê<ÄåyFUâðFÞÿ`$×öÇ#ª3Ð Eä lŸDÆ”—Ó"ŠÆà»´TDkÍ\Fº¿Y…PÐãqÅVß+¶‹~7 Y±Û І’D•¬àŸŸõLT7 ÖÄç+ãÑÔðAò1u„™õ‚[Føþ&¢QI~'Ň#Uù| c™_„ul‹‚}Ün?ã»+Ç®ëü ·Æ^&Õ‘Ðúö¬¦/f1‘¬x‚âÐ^'ª]êg¶ÄQ¢¿I{¾½’ìÒÙÙã‹Âãy—³Ôò.ÆTÏEÕ2H³åÖЛøE™K ¯ZÚé^:1ÀTEÎkÞÚxØF«÷«<…6Ñ:÷Þ1d?ÖlGEFTû·úœNÓõ®±!Θ¼ö Ï‚0GôÜø«RÊ*‡¦Ë`Î÷x ›ôX ñÄÄ)씦Å–ϧiÏŽ"œßé5EÂW Gr%T‡4täÛÂô$¸‘¿\¶¢¸˜_”Ñ\:ˆ®nÊÁ;DaÒTUÙ‡äLNcÍ,,!ÜS=#Û`¤Ø}:àaxC\D÷×1>#G4‰©ß¤=>KtòˆéfMg?"§CY~˜2Ú wq1’+GÎãø÷Ï©¿²Æ¼¡Jfû1ˇ!û§µ€ í+þõíªÚlÀ#€ÐÛ°:´3›žg!ʽiUšÉ)ñ0è8{hðßÄÛκ¦+7RpóXWeó¤¾j9Ä3ŠFKtõ ,æòçØp¶Bî¶æŸß€‡Hjy–þΰ§Œ6iBÙI,NóÑ“ÞÌžýÆ1/Ÿ7 `·U8l9¶·Õ„ÿNoTpšjAc¸AŠ# ž „ÒÇrÝoÍ Á-w¨Óg)™‚K!![qb52zï´×ò›mù,o"‹´ø“ÈêŸPn0˜bî>lËd( ©Êsê³È²Œî…"w]5²ä·åÆIxAÊ'̱)?Ë#Sb0zfÛJãäàÐY&õË:&=ÂÖj JÌþ®òXîÝ'ß—‰àƒh÷‘ÓÔÛâ§ û‡ñ9Lœ6{u„´½±æ\Ða&i}º‰uÇpd¸ …XK'Ì2S`Û½|ÄF8sÚß0JÔ¨#ßþ]ë¸ëãØÿ6²:õ$ÕaµY‘y([Þ„ÖÆXíqçØ†ôÙ}f˜îÄC ™J;&ßG}\@3…Š5Ó5~ Í*r©0ÌÁw Yå+$Š‹B½HñÎG£P pÎúHÑÉeæ/9íD¯gñCsB 켫IÄdç„q„ív­‚I§BßsMÀl'óZ³ðîi¾A¾JêÆ Ð]¨oeÛÉe:*¼kà‡¹öÑ0zÄ‚öM q_ì õY¶wÛ):Ùx£¸ŽÌס6A˜Ù šÚ£ÃU äë=|—T°+VX’½«F˜F¥‡Öc,¨GøÑ0cH>yUéçä»Kj|#óz¼§¥½úR½‹ëeoã6 (Î+¤ƒí51S¿ÿ˜àå5PSB³¿‡„B¨zÑÜa½ `ÐumaǤ½‹-) Ú–{t¿“G¬wîϾ·Í5ìÆc¤1’ ðþì–Þ5ß«ÉÂXÈ’Ž5ôª¸ÍTÍ/&(!-+‚b%熤T%ÙÏæ4˜Ò1E¡¸oÒïn^ékP‡\ é{3$ ÝÇñ1ÓdóŠLgøgÖùe4 ¡”‡ ôòë6§ŒŸM"í»¹y–ªÕåpÐ~ýc¦1Y­n÷ªÀ(»å õ¾u5èN«0ñî=ô̯ê½Ð ~ ín:ù0¹Ïl¸ryS³ßãš8ž%š|á ýR Q˜ãs­Ò£Z@†(tâ>8.Œé_ ñ¡ùžÉ‡Nšy9bßïiÓÈ|baò 9†(rø¡Kìso*ªepxG˜ñ‘|ø§>6äæ' €=?ço£ðŸùöy–²ØˆõÈH=R$Ÿ– Wf¹jœfC€Ã:l8‘U¾œHÝ‹ÿ¿ö.$þxëèndwA “ô‹a£‘` —à˜m®ÅA‰S³QuFhbüIa?'ÈAþpQü‡VÞªö#š“ϪºOW¿‡ZŒ6Tê$p2Îqjk>´jò¶øò’ êeðô;ý5}åÞ¹OÖo‚ÞC;ŠU>R{©´ô_Œ/d,¥˜Ž£ì.bRãÕ2È£{D½ÚµŠBÂ:™üGXÿ;ÿbx‚Žùd+áÕ´ÝBäås üî@¢Ûëþ})Ù®Nðˆ÷b˜-IF¥Q£Á1ކW90ké;¬hAÑji–lBÕÕ#6¢<Û Ð<²ÕÊEýè›óK~rŸ§?Už´ÅCí–N4Gœƒ¹ÍªaÝCé0=“–¨‘´Ûb=úZK÷1^i`>õ÷´«p¾;4ŠwGééiþ±;X¬&¯AªúËòòÝ¡¡3® ª%1·µ[†wãŸL«­ÍØv, ˆ¦Ïsæ âz!@pؤ4å$Ç ­Ómµ$­-¢QžZa@« c?5 L‡@wLÿI@³˜u+éÈü… ^:‰‘BÊ ŠÐç{µF¤Ôžö2›#Û±ÙxR(þ„ê1a†F¼[ÅßQáfâ‘ ]éK ण»Û¦+W)s\í/OèæEà×<ÿ§j5=ÿÛÑq®½ÎÚ´YïoÒ3.aK}÷“Ûß‚91¨½éZ'æûÅûl3ç”wo¯M‹H÷¸­ w¢ô'x’RMAÞâ-ßSVlêf?×oHUj™ 1‚<²rÔ ‚¹,Æ«œi„Õ÷1š 58XiCL­ÑÒžœX(!'åD=ã÷5ºK_ºhsÚ-ÈW?õüJ³¤XJ9,Ð+x×6]¸ÞEóïãkã©uu|;‰2æ À˜\–º}—´fîÍž:eÉþÃ;qŸñGGwɉÄî„îãGè§_·¨Ìh2_‡1žÒ¦Àq¿d*[Ö@vTÈïöèa€ÔѱX´Û9ì MaI±ñßáÕZ6•¬ 8^áph@–°»«7t-H‚ Ón¡ESêêšçC‰£Q/Ë„Ž/Iñ;ô@ý»6ª8ÍoˆXQ˜›gz–pß=7}3-ñzíK‘ Œ  «‰ª»M_[N0¡­ºá§oÐ*ÑFµSÃìQîjJáúÑìþ‡ô³9‚8Ø6@&ÓRŒÀ$J(È-¿>`]ž h|ß|ÿì3»nƒIp¤áÖ{ {~Bj7§[C¶v• "œµ²@-4ÂD¢^4§nˆ€€qꊓ‹$ÉN¥Â„ÂëˆzxSÂ6ÆMåÅ7Ö},*Œ™E½_ùSj'gZøØ²µŠ Ö®À•„$ÒH¸['¸'×fþhhŒ0ä|Þ!¹žA–?1ÛqÊ¥‚+'çÚQÃÇ{N8sâ|”á-!Ÿ·^J>xá+·øàäÓÍÓ6S„½æâLéA2±êZßÃ)·—e•#”‡ƒh½™!ã€7ž31 “òkHE`ßïø873§yaÓ% ãJgo8Ô)rCU~L¸Œ¸&éŽËlm?šZmß.Œ°0…¿~B#…œq¾È÷¢LC?ù¿³$§–õ ëÐl\_ØÞÇ*|BÿÅ]—Ëí·]™PD`oÀš×÷5“J/'a{ÀÍ5‡©ƒŽ-ÁÜä'ÈmhœÆ‹*XÁ¨õuI–jQRÆmžô Ó‚6èc-ÇßÛKvº|5Qc‘•#$W(¨Ïq ê\p7_ÝèU Ï-uœ„ð°o™4Þ;â µ[Nz9 Ô.ŽºîaÕzÊ. {+,Jó2þ ÒúŸ¤{¤­xoo’+ÏéjýÕE\~â`¬õš¯Ó\¥&O÷³´—|mƒ {xÍ­z§ ºƒàÔ•8_^_Øä¸0Ú£¥©·" õã Lbb8§lb‡Tö#†‚›ƒ Ôk@U„ûv#ÌnûE^pðM¹;™ó w=¼ÂÏ’öÝA= m ±·[î$XN\É´|Øj*ô h[_šÛ&!ŒØßÏ0(ƺD{ÖTßÍ0õ Û=`…Â¥¥—Ê5Ct¬¡ý/+-¤¨nGtó™¶~ñ³Ì)0ZnÐ)]¢é$T+(q¸Qf!ìäG˜µ%PÈ*Mzt€¡PÜ4JÃxŸª€º¤ ñN³—HIyû©eÚO´\á˜öé’ ;žãeÞŒëã?³ÜÍÇú¨]]b £Ø*vîbãa?p%ÁÒÁº÷P Ûy{»’Ǭú{—áÿ÷#qÛ<ã ]ÏgYdš„æ_LÔ1d0Pn3ÐÄYQôðïýt½mÝYì&I_$V˜Ä@ˆéeÃu_là!1Pâ Ó7‰Å5v†·u½js.‚¥nÆç©ÚËXsLAº#Á®GRç+Óþp –Q·ÒnšU‚¾÷ÔÝBo°J“ðÁaÒÕ®[µ=Ñ׬4´Æ‚1uì#}sXÂôÌ«s¢¢c"IÁ”eÕxÄ_­ o]’vÂÑ>¥ÁM\'xD¸p…¾>éøE‰ÿÑJíB2˜0œ_›æ‡¤zn„îÃC8‘D+s¼•£ô쪫•qõeQÐcX­/Vp‰ûâƒ|| ˜¸=ýçÍ;js7ÓŠo*,ßÅ× ÌC^L”r¢ß[ã»=´bÏ»UdͨX²4’EFCÔCõ’ CÜ2š^9´8#‡Ôš×´‘²‚d¹nÊê3”;ƪÓ"¢´á4 q6ì¤VÆßz„ñTî¤GØ>~Êfˆ¶ñÕõ,›TØAÎî!*ë È̲1óT…ù Ò±ùCÞù-v§‰•@Õ|ã´¶#×—âM­}—§ÈÊ Od„0’e º²r¼ OŽpßâÇÊ©ûé•þ”‚¯eÜÓA¦úÎ 2Í'õ}à“…ÂÒ,[OZ -jêao¯ˆæ ãý¾é?Òj²RÆþôKø”mE Æá2Pc©ÿ&¿eÀ)s"úˆ(;ùyTvc'ˆó~):Ócñp´—"þ±Ü¬pGö(­{]Vý{èä\ã¨굂ÄSÙ0tæUn)°E<‹Å`¸ô•ÏØQמ>ˆ±‰N;ø…Ý`râ£~žëNmT °2ÁÍŽ§èDק޵3ÏfÀ¤–®}–\³KºOü\3òÓ˜™æ;ä¾tg®uye ³ì•ƒàû +rb)ª/‰F<Ï8ÇèÑœ_³¤ ı¥4ÞçñÈëøõœƒ‰®IˆÍN]æÍÜ}{äÓîx*l Ò¶ŸG\ à‘`—˜-Ê—£7ku,$v¶ñ,¨Ì˜µ–ô‰¹ ¦G”±:É~)Påb§º tÖ†‰ìtñÔói •‘7×$üèO8ýH¸Œè¢°VíH tl¶’×hŽˆ~{à:a S-È0q-ñµ+Éi¼€ØÂ¹œ(£M^=‰]±îv¸çhŠìeöàO- ¹V¶]!gyš÷­ÇdŸµ ‚¹c×»s<âa×z5º3¤Ò©ýó™èíü îû’AýI2®Æ{Z™E\´T˯Z#d±…«´ ­’~H8’@]æÇå¥~¢ÀÙ.…튌®âÊn˜òbSOüàp§±õ˜jñ[yþ‡dfË*@T¿z!Çv†`‹¡Lê ¯„C>gc|@i5’RëÐ{/Ë6`ƈtlü3K«Œ«sQXÝLoùbÀ¡,>Jš!‚µ|ÖÞûõO‹ÁÖÿ{Qç`ñYiÂq§9zÇð4 9kÔþ„lÖiÇI¨e0ñ¤¡ƒ±`®õ¥$¸m¶©j6ˆUªóÁ²ï©^šý*bÒÓoVÅsW’RÈÑð÷=¾_Ÿw4ÖCÅÍšˆ'Nf Ñ6¶õåxXqtg-í%aY³Výؽ`«R€þ!ÉPVØ'ºÆ¯×ý{þªh{ÃÖ{ûL¡L‚¾{÷…ŠÒù:yµÇCfHao „[ßUö.Æ^@W[²È¶™q+¶œÞBéšñéëZ(™çïî0„fÎçt§BHKì Ã@ë!ÝH¤ÔVÆAzÓáŸ÷IŒ¸;˘•…R°?ÀtHèZÙ!蔆~¡!òdE {‚æY 9aÙÈ×ãáöhÒî½u’²‰’Áp…ÒýÙ¶ÒÒh9Ÿu‘¢`€Hb%Ò¦}7NÙ(±¼u†…9í¿p=ñH æ0&õ~N¿Í‚Ù+l‰%Yžˆ¦çã‹„Áa(ÛâsnëÀøVȤP Š>~q˜X¨Žf‡â«>™í÷v`¨å|» Ú 5íu$)h¥CÆ žÕÊa¨ÂÞ÷о'GѦM“v_F?ØŸýeU™Šk÷„¶ýÔ£º&ó’ãŽsðå\úÁ^º ¿õ@ˆèŒ'V~†£"!ìô\ZyÍÁã7ýÙ9â¦Å UÄØšÖxŒ‰€`ä(°¨Kå<*ÈÖ8^L<6ÏàÒR¯$OÁeØ,{/M›ÔG¸æáºB Y: è3VëØG©Íz»M]Àh\™ñÀ?µ’ ‚Ú͇¹…æI—Ë•èÛéX)¾ÁÁÚxþK·Ò .ýèÖ1ýj¬a#ñž?¤Êžœ'y\±UµçÁ!˜ø8h-'®Â¦i¡¹dc?ìÒ…_Â⮢E'CUóz8 ¡)·ÊP-“(L‘‘®h}A:­c†1¹‘à¿—õ_¥ÐêH‘´8¦\€2æÀš>ñ ªÈ¢HqAFª æÚšŽ¬ N˜ŸóbÈ4Ý@UzÈòjA…{½{HMh݆I{‚Ÿ ç¹”jíŽTÞÎû~ÝFõ6U–ó§Å,s§‚« _ÑÄÞ¡º^óÈÁZ¼y˜öÄ©f«…Gÿ§(*Þê`ÿ».”B¨•€ß̰fÌ5"óÖE2çœLyúKŸ<žÕ,Œ¼ï¬¢ŠÙiÅÈZH("0"÷½Î¬Ðï CŒÊˆµž[›Þðàèp$[­‘·BKBh¿g©T•OÖ†§ož“GŽ á>{c–v%WwÒéC¦š€ˆŸÅ+hRnóå£|M-ÀKÒ{JûrzGŽÿFå’¢FÃ\ßãݰGËfý™<@‚X†Ø»Ž,™`ÞŠY¬”Hûª¥;&à³JJ(Ïgö„Æ«ï‘Bï?ö¼H쩲ߒ@d°%=†’þ°$¤³†  ı—­£ ©"˜kË1ïèʤI¡?ãÛo |‰'—º¨ÇJV!œ[ªÒŽÀÝúüðŽŒ9ñì!RÞÒ4ÚÍmk{#ûÊÞ6Dlù¥m}Ý@ÿ‡S*4)<ˆn@šL/€ U¬b EWÍè„DH´meÎ:ngº$í³ŽÍd½•Üfƒº ø©ÊÚx‡ALà70Í›”Â=AH€=ÀP]ñÜ£ž©@v”©ÊkH}¾8¼2 ƪ£eÎ?çgKµˆjï_'L.óC)b¿_{z3ΠýÔ ÜŠÚœ ‚YNÝš!pCå×})ÅHŸb{ˆLBј<‚pÂÎŽeBHœ&¹’'/ª[FŒÍ¤×ð à  ßªñ‘V÷lŽt&×aƼ^àEóÖ׿™poÌ£ncöfën ß Añ¦@ÉÏwô™) J‰«2ä À£Æ]àHßùߪêu&çž”E¼c²½·Û¡ó÷=Þ*ïãјÈÍJð¹æ¶ vÖl£.T£Ù0>"]3®pˆ¿å¼Z[WîÚ_'“ÞƒÂg¸ç4n¾L¤¡&Œîóe¹žO$(5ßœŽ<šý‚@¨Fµ¸_fÖP‘úìÛmÄÅÛŒ½ÐM'›}ÔáV$ßûʵ«ÌrêÈBÅkÓ?«RX¡[EúÇ8ê»ì§ %üöÏ))UØ•‹ÕÙ-–vûÄê³J°É±Fè¬áAWÆÖ,<çŽ'wª•g¹I8Úµë$˜ëóÅ& 7¥¸_ ïøncÚ4Ñ`“™££h _jòJŠé¤ºÕzkJP  fSùÈS“ú[Õ9(û aÏ´¸Øw™ÕÝŸ¤’1îÎ]-¾†±ËÏæèÄ¿–ì]ÒÁSbCJ£)u¢?ë„«¡¸/ÙÑÞÓcå‘¶¥“¬‚@ùæû.šØÚ\ÚÄL¨Z£ŠžKµ#ú­}ºÎà¯× ›²òÛGè‹Sø†ƒèômîˆöd2OOú'RD“$oIE–öWl–ß;”{àu3ó§ØNºü¶Ÿ€ì€w”'çÃeŽø&©‡?N-( nDÇæ¢WzÍú¤¾ [m´qÒÂ÷5„¸ìo¥þ(˜8Q³e“à7èÀ­¢•Á€÷H0 —õ…2ËýÑšäQ TBØ\I˜+©¾òÑ9xþBC„v?ÕLÀ ÃeˆbÏØ®Ò8¼šTó±iIÓÐR>0šU·t»Êø ¶†á„¹Š>ä’-ª~Tzh§i3+6š—½ #½K DüyÅø}AŒ >—¡d£òY ¦(iÐú(aÂÑÐdŽá™N©²;E@i¸f¿ÀJß“ƒÉVtÕ×Sƒ:˜èO‘më×yצª“ýÄ$b`õÉH‰"'<Ê£´iL’P-ÍÍ\í’æÏTà«k|¦sÄèbÅdáÆG-þf822Lõ_`3œ\e)Z‹ÿDJ?¥NÍÉfƒÛÕèFR`d†$«ùÍP}xm1îÁe´UÆ:øßìoÒ)wZ°¸÷îŠ^VQwè¥æ¬Ù& SP,ñ±„êFúB}³¤pºh¹/M0åòh9[—•U¾'`Ѥã|Ø.Çèä¨3V×Aý{šQá|k…ïÛvK«VH_ÜŸì3ŸÙ (ý_Ä×é©!gR7H«Þô¿¡2ÐJq‡ŒØo¥Æ`}ŽWûê--Àêcãòâÿœ»Ö+4=@ `´¼í3”HB!CDzèÞ½¼E¬ô‘ùQ]öY¡ßëj*bÖË~2æno©˜q¤{¬WWÞec‹,½þÆÚy½á5M«mñü¥=ù,BÒ{¡—a› &æ5öúþžÿ»å?bcB@¢qï ñ ¨ñ'-UóPÝ 6ìÞ{?Šô‘Wê[›š˜`_ÃøBÜei¸ 3ÓÓµëüF>—2ܼ¨ ÒºÞRžÛÛ…M e•ÓÈä”Ḋ`V^®É1æ’6Æy÷4çuf-ÔÞö–=GÃÎ÷œ&ƒÈ†ß‘u<>mp×@F!O5G™Ë’AáÙkâ%\ã,}ƒUÝ mÜûÅŒ—þÑá•­Nþ q.5?ïðØ`ïÇ- ˜_½²ÇFœ “•E<>Ǧ$Å ·úØ®­ º7›/·º|$)$‘Áy~ [\…ÎÂm¬ï-:@—gCE\‰²ƒ ¢W¶Uµ'a ákv«Š {p—_•Iâ|]\‚Ȱ²k‡CÕ.Y’Fý>:—ì0Y)A©Û¸S»®-ÜÐô³C;¥ûá †RÃ_׬#íŸGJV‚¡¦¢ þ6<ýwPÝ­ÝÒËÅ=Ö‚°°ù!+†¿TÿgÑ–Æî ø¸Ôn•ñô¡!&š C›ÂÿOjÁñª[u ¸˜ƒp‚¼0H.ã‘#.;‚¦°í­Ó‘Ú£8Õ}´_zˆÀ©äþárocò³eÂm.æ¢kÑAÝ¿*:HÛg.‹”3º.*Fݸ˜_ú¿ïM «‘^Ь†<ÎÁF€y“ñ¹o•F¾8weÿÓœSÓWñûæÛ¾dµ‘MbîÝ™’4\Ux‰ùÊ/¡¤âÑ–žHÎÜýT=j”BÛSµ/ú±|2‡‰ŸM÷@óÌDæb»Ž¢Œg¶§.lGû=¹ŠƒåÇŽM¿I>äÔ[êÈ"ÿaôàJq•üV j£»ãMñóX¥Lʆoþ¸šˆ*ýþbZRsÉÕÖ¦)§©øÂÑš´u;´@]W¯(þ§LqA6ßÜûxfG¬ÒÆeE‚÷‹}+PôŽöU×g¡y9?YHf]lªc7vöŸ×Âøf< ^S¨¼’C2FÂ)>íôf”èµ80DsŸœpiÏ…ë$CÂö.1ºª&ýüÆûz1¬{“w"}¬ýS ìÊûAÝËŠ`æµVq±RŽ6qUšçMÍÒŶk#âö"Ì9øõpÖóÎø‚½SFË’†l1Rb q÷—à»ÀÈv=È‹ö…=õ Ïít9n&¥¿_“µOj 3£Ç1æIE¸#Œ(GùOìåá?ö\n…¿)ë‹.¦Z¬Mq“0v/CqdxÝ ùècB¾Æê •„<Õ¹%ñ‘·ž„i·."ž Æ¡:Ÿ±_sæÉ®×yÂŽÌ_=ëPª{ß3–ú{ì¦ÝjŒ3WqrðU:K·¬€ˆ¾ûÜ}¢c/o=ùw–´ù|Pª¤§È9œž=ZÔè™k­|R±†áÜ>Sœpoº ·ƒ6¼CIøÒ븨!ð1Õn5þráPÔ¯—¡.,p4I¯¾Àårõ&]¿M˜|Èû ؈ý?šŽ*y98ĺ¡j …DX¢n·³*VÃìÛš+ŸÂáÀnµû 5m, z#îý‘OŠ·ÏÐÌ©6{ ‹^×=úd´wB¾ûD±Ó`:–H›íbNÃFCœ1Eú«¿¦o›wï[0ù:´úøb¨TÚ[ÊîálïÖ8µÁ¡±²êW±#+èéÒÜ?õÀc8ø@ÑUû¶ï¦ð²eÄÖŽ“ʉç·"ߪ%e¢=lÐÅ×CTÏ]á'5ærø'Ç aÇ„ ¸FåÂÖ‡î¾%êŽ b`zàb”c{…y½bk£ßúqÏZxÓ²n´öûÞr­cgDœ0¨Í]¤,¹-Ø+²âÖBœë-yTX°ÃE´HöŸ1Zš^›~S%bš=›Òÿ1½ë™Ñ^Ðq M4Dá+Q0/šdRÔ6Y½l9|óÄ%Ç3¥·Ï/ ñÏå{"ó‰™özÇýùðº_Öæ[œí‡`]ØÎÈÇ>ºæ·Ì£¥:Æ›ó¼ÒÏñ;íZ)ò—&†›“pûðå|Àë(e§bj6Û $ƒÎ&èNËS0óuTEo2³ô/Çè„/ÓÈwMÀ·“hvE Ÿý0(ȹš& œVZyÊÑ¿aÇm)éƒ÷…á…‰b «XN]%°KSÝ.3hØ3Ÿö©ù_£Rþ/Qeð™×T…HòMnôögæ÷¦Òi #Á ~ˆåÿ²™ÐHwpýŸ ¸õ.KÜäØSsìfè€mÝ ý‰Nãnz$¢;6’Ÿ³(„äØ}!WÏíD†oc×ä]ÛîØqå½aI2ö@Mô»#‚ðÆÔ™Bö ä—$@<œ1ÃAóç¾3YtkîÒG‘Vqi© VE’Á¡¬âg áoïÃSèÎ$EåÑ÷†(ƒ4ñA*5€ÚsôÚ厨7h¦9ŸGˆâÝç5/od hA‚Ÿ{$uçàj¤©Eë&¿˜GÏ8ç(ƒ;#’gÛÜÏÜí¸ww¼ÔЙKôè~s,ÿÁ$íÝ M¡HÙÇÒ×4…w³[k?V‘?fJ¶Õþ»²ndzz5 –è_aáÈóú°É:Bdˆä^󈿩…¹e,x+%TžÎ n`Û[Ö¡É nä¼7”/3‚ÿº”×OcGBÁm{%©F+È4Ë}Æ”ªßÍÖ,’¿£h»<„I?üŽAè}lÚ­àUÒ%F'W(tÕ"pY©$7Œ¢ Uq‘} få{Ënd°+^èb Lóãüƒ:Íõ+Š:‡ìÉ‚æ$ÎËcDŽ„nálÿ“†Þaã6æ„6hXñ$-ß0ã+á{Ú‹`‹XºÚA@ȾèUò›Ã_©W˸4Ú$×gÓââœ2 øÃ’nôO™…±QêZ9IÍY÷¨z—é÷ªž$Ë\•!“)»¡¾R@#„Ð:€"Ç<Ç{:ô²dÔ^Ö<¶D;‹$›D x5bÀŠmÓÒ;ëuþº¦Á(?=‡"$D‘áuùæFôéÉî ØÀ{Vs.?Y‰è,küÂc³¼,¢&îèÕsIá`«"NLn§*¹ZõÚ7ÞƒJXy~ûË… Gþ#:ÏéÎW D*EÀ‘âèÚŽuº}ÉÉÀâþ’?ææ¾1%ï²Qvè?Õa 6çæÂËËõ{‚‹&—G]ÉŸ`õLúì@®[]«ÎˆÞ‹\3ÖÞ‹3­6WOSaÔæÆÐ‡ÖÓhDÄ®S´Y‰­6ÞTß¿KAôÝí^^Ýé±2꨹5îæ‰pÌZÛ¸' €ÏåõqA¹.eÔ Ÿèkíù>æ1µ¸€hHfi‡‚ÍÇznGÆÌîÁ¯VáßÍĤ+c—QœÐð:tÁ´üJÛ;A5™ý{Mgi<nFª>» ?rMfaM]†®%ñáฆË<˜]kßÃwƒ» ŠŒ´@dp€ñ Jèä 'V ¨€¸%Ö´ûrŸ,éQÁ>Ìi¿:•¸óÜЛ@££Ôf]á–å4·ý‘ªa(r¨Ž fý<¯9ˆïÛ0/¡l`e#¼Q‰G¡ä‡‘†³Ón; '§ç6G­ söB¸Q©ÖÎã¤<¦v˜Ü÷]0˜·ü´‹Ð´X±uY5³Ü¿¥Ó“VàM/ia‚bÏê[}òÊîß²+’ç8i$W±uÕ‚¥»e‘ö­C‚d_Ù£¤¦Ê’M‚ï‘RìXŸD–¯„Þ©ËÐÁm·Ýbé"6¬A×p7­"8¼>³zd^q„ïìôzPò¼Y ¹d‹¾òÝ0øRαd;.ïâçŠkÃ7­¯è"ãéx׳Î,^.y,tªk=jÛ[íkÀÒnA-QÏ‹øSôœ;Ñt=ÎìH¯,Ðnf†Y™?ÈÈ™†Šã¨l¸Ò:ÕÊÜÈõ]L‚Rá¤Bö˜ÞÄic×9˜XÛÕL_ ƒ *ì8÷æº;2RŠOU=#¡½\:;ü1ÿïÁÛþÝà.~ Óp׉‰±©àÂV\朣úÓÅîátÉå#*”_I³¢,*k%ÇS”‘Aºßð@Ú2þ?Ü­ru)Þ9õI}ù#{M¿ï•°HÚX˜¯Ø"R%ÿ82-ÆäO&¨ÏúkÓ³dO@GãùGîïÏt“Ó9ì b.ûRCr¯úÚ@µÖå¨íÉ.Ùàò¿‚…×tàõyÃ6Ÿïù O¥+ 0Òá—0[õ!¯«ÜK!D÷½JâÀÔäƒIhãiMg&Ž.N†š³ì¨â-L”ö1u¾IÞWól?¨5ï¾Ïm]ÿvÿXcÑ(ÜsjÌ] EI=¹nÚáÜä;yô+Æ %¿eĨs ?IFyÃÖ ©t–çT2JafC'lÇùÄ•,&Hi7s'‘á¦6ÿà2zŸbûŸ™ŒÄg@Pdå^ƒ¡ÆLÝ{-›&YÙ‰h›qÏ©ÀÙ•ãéèϳó°x”¹¨Å´M'Ob£7µgç%7š[v Ã-å„uðOXmãg^¼Ô4ã“·ÇÉk ¿ZòLãÜ)À`^Hgÿ‘¥þÏš‡;ñù†°ã¾ª<Äc¥¤3îh%ŠN\N‡€Ä†g’Ý¡=} BØ+åàxÐ!Gt@t¾%Õíô‹ó†Ëê:‰hÖ!‰˜Z+>¡º©1^yACa~Ó¼ƒKÝ»ÕUi=ÉÃx, g)ŽÉöÄ:FØmâNY¥¾ÙÑ£—¾FÑ+¨òÊ ä+zfè"°ÜAÉv=ivb®ˆ&s< A#Úƒ„Ù•QbaÝ þ*×Qb¨ì=þ‚.zºÐ5¾–¯‹ù/dKêˆV׋ڽëCVVýdzˆmbfN5q ¡—òACi¶WB gõ£¯5ÚgäÜRß«þ09C=Ï}l· h{V²0QX)¬=Vqm™ŽXóCYóBxñG°¢7±Kñn"hú6¹=÷ûj”é^LÅÙíÖòl·éå01hò§•©ûí® š“á ‚u¡‡ßÁâ®9îæ{‘žÆ³JªÌt6[ÜÝ»N¶‡d”¿¾ßÝ»©ÜŽæ5Fu1KóÁÆ1ñz€¯/ø–PL£LrHu~Ò< d.ê™$r?\þíàòóúêf) Åh‰W ˜z\«a+òX:¸•2ò7¶®ê 8Ê÷LØÍ"œ-êÀ§À×:§è¾H<â1bÈøe"Ö[qãÉ|Эhähc;DyÕ•# þ™ghM !çYLK¹—í®{= :ú)µãÓ/@.[Ís(]ÓºÌïF…$<Êûfš{2 sŸá1û÷ž·/ôCñw!pU¨î™ûbY€ÌõÁ‡È4—}B;ä( Ôs?]Xø-mù ­ÂX ZŠ&:¯*ÅibA¼åËŽö¸u30Ÿk;pQýÝe:&]¢*ÀcšuüÌØMbdç80§Ðˆ6K{bY Ô ? 98Qu™b‡oùüׄ¢¨£ë9T±oPþo]©M‡Cn·ìkÌëÊöé¢3o1ü²1ÙAî;€ì*yaðpˆ¾äµ4 nª¥„Hð‹ò˜I¸=$]=œU_Øx“V ‡——.Âã¡$–úÐEÃÄþÌ C"@¨¦!çGå?êO¶wIeê»:‚ð’Ú±^53${‡qöN8Åö¥ƒ#pƒº[ƒ½ ~æb˜š¿%Ô¥ ºq8@AsGû÷&NŽKîD¥ø±À#ý’ÎO¿†‹²¯¥‘ÞS§>-Ô£MÆd;vkvñ_&ÊÀÌ9’—"4÷£§ÍnC猾†zŸ8xƒ†z>aDnŠaØüy±~ï,ègNm‚fb°W’ß¶ìq.Ën{g÷Ù(¯¬øÉõØdœbÄÐWñÎÛE¤DÑÒƒâçu…×—¶ÎOž1\IØi+Ó q]í Ùº*š8gSbe¦ÝZÊÀ–Äp~¾“Â{{ûdNJgd˜w¾ä1,«_žJóoŽ=ý,q»”ØüAµò íÈÒ!ð"÷_ˆ/õÁ ß#Re¢•”eîOÿÔJyîPˆž×ãD·sðÀcå¥÷Ù¯@”ÙU? èêÐUðŸ3ÜÆc1<´¼£èˆ¶.–Wç­‡Ð`ÔçbãÀï·„Â:m(UF=á)*ÁJœï[ºKi&7Ê#Á5‡©´Jül{'½)°_ì­ó±ó‚ÁºtêC¡N˜ÛÆLäi2è›éÜ™¥†mÛU¬Ÿ¾†9IT¬)Úm.uzÉd™Ë<;¡wÿTuÇ<mgøk´(_4‰øŸ¬§}­ í"ˆÃ·ÂRWí±M´ ù÷Ÿˆ·†óí(IV­ãÎÛÛŒG¥ò™Ô¦\à :â3£Ø?èCCµ¬\E[J?©€tæ6;ú3‡=m†"¸! ¶”ì?û^ ¤Ç1åž¡ôÇe79º¿ë(cÒÝWÕ#A…ïÝ›Ñ Á¦¹›„ˆ©`]P»a.ƒ?‡#È]`vCHéhb 'Ë4fxUåáqñùWíf̳³)K’{}Eõ»Àô5Niœójûtà üUYk/›aôç|‹bbp²õ4ü¬ c¬F&q—ÓùÀLò½Ìu°,ÈhMN‰2¹~Ó,_ƒÇÐè4›¯tT°œ…º˜Ë­¼”c¬Ÿ «\C™L^ïH?d3sëiÇ<Æ×„½Õqo0´ÌÉ:ÅHÑ8éË&Œ(χ0@ªœa³X…êõm°9äg!æ6âBÎõמårëú.Tšœ½ É_Ŷ.*ý¤˜Rw¬nò!™©lk•@kô«Qùïi€eïªéD×ÿl¬_˜ËJ$bIÙŒ€Kò.¼×ýLI¹Xâ ú2CÙh ¾f2iý…‡hf¬™¨|¢‹ ÈCSâ·ìiÙºöî¯ðë¾Æ´›N”þÖÀU‰)%½š˜ØÎü"v‰ŽIB4—¾·ý‹~Í*<Ô­ $Âî0a¡„Læ0‡õîBUÁjŠäÝ6Ó+PxÞ=l_’¡Å®-êÁ¸@j×M~YÓ/> ƒ0XqI+8“¢—ÝÀê6 ÖñöGÝD›Mz7\îÞ?T-͇QµÊ]ú×¹äÒeèŽ&õ¾c~‘)¹ 3EÒúrÎÍ©“nÍÔVp6±®™˜ÛUIy•n¯ä1éì $ÈÇøºû+ ÂyÉüí à9&Ç#–]<›çdÁçh¹Úeìð´Æxˆôœ4'è]ù;Ï™a„•–׆í·k¶Uõ?:(ß=îîö‡Î-JŠ2zm‡Ùš‹ÍmiÑöåï®@]ÏOx+ZA$´­’$0Þ¾‡Åïþº†‹šÚZâ1R¾Ò~7Ó0Ìã;æØ{l‘›KtŸBÓðótòʹÕiÐ/éìúàP­á?ÃJ:6GøR‰[¤;?»qFÂøeÇW†ä*»[ïï6nùªŠOo¯â&Ýâ  e2¨q*ž^ê/>–mzïÑiÚJÔ Iü}DF56¢ÿŒ¸¥û{ľFŽø¿WΤ+__‹–Óu*Ÿ£Š²;å@´¥N‘µÁÈšÚr›t /Œƒ²Êpg`­yÙ*2ú…ì d’([v]‚Î×Å#¿€¥³¨O:û-â§øí[lS)$RÁý¡%ñåÿO‹ùïH`˜Õô~ý^AèD3rí·vð¯ØÜôŸŸUÀXã•Üœõ ŸàtB1æ»/¬ìd§NJÚ¦,y€Þ×"Ú’~’9ÑÁÎK×Ûû±·D‚‚`À ÊrÏü¢™ä—ßµ=a9E7!¤È"žZ± Ù`mh‡‡:ñ™'2Æ¿µÑƒR§½b)þ}7¯EZrËh ŸóÝóÁøàúd]†û,†XÕ;žá‘)\>6}”ý[“D VèÀ¸Îà2'õò˜ïPŸ'ÅÈ+»naÂîÐJbHSO ”ªdy_{.jƒ#úžØp½O}¯Í/lK4nº¤ë„+¢^»ïíÕ¢&GÊÊh&ßnjÇ• º1D÷Ò–C¼öœaRõù”ö*Œìvå¯N\éĶý\Ìè"Ú ¶ªúˆ`·„Pì9{bæD¯k[C°r^régÆv4€%2øÂÍgÅ`fÁ2(7*ïu²6øÔ.;ƒ¨Áñ@‰õ1›kXªî0Ô{Xºì’À ÉP´û}¹/þ¼¸)':†]ÃôzðÚ5r®üLp}}þa±û§â$×1²WUéß åHh7¤6›l /=^d‚d}ßDÔ_ W-Ÿ&mEç¶\W$¦‘ or‘uâ€s¼”9\/¯+_]“å ûs㼡`€Üä~ ¾>}ƒäÿEÓÊš›:Çzì¢EÛœ]ŠHê¬Ñ)g'XéNàŽ¬Ð+‡ÖŒÑ >=vÂ:Ç60•­;¯S®([wÖ¥îE¬† s£MI¼Ø îìæ¼–Ô×ïçtÙŽMû=†b¼u7“d,œ6.ZpG‹":­4ͱðÞ ¥fÁH '“JàG]í +â+cðVi(øï™}ÖƒktÎ"ÜʃÍkæxÁÜÎ~ïØ‰áŸæ-Œ¶ÂУ$sĤô„à!#T^Lù¤ØJÂbÍ‹2we÷ï¦.ˆUCQˆý?/05²ÁÈhΡ„ BÝ핱>ªÆW]‹JLK{NQÌ’(Xâ ¸‹„ÐDhO¸=' [OÔß·™‘QijïTÁ@Æ~ 乤y »½ØÐÝ0§m1Óx]섟EŽP ƒFV¸]<·4AbÈOü›ö ?pžÚìÓ=1Ï%£/\õüWËtqÉ„øÕ{Òÿë Až†vIØoéøWX>RÅ)·Š¡"à*u–{x>qà xýF»;•¨ˆä;I§E¬_i :¶V¢$jÖsJo…wÿûÜãW>~Õo†¿´“¶QàáÙfÅQªfH«O¸&8>ƒ,ÅX†æÆ<ª½6ö¨à÷\ê×:t`˜àÐ/A‹GQø€È ¯á+…TûËlªSù$Hä R=¶T§-°ã4½?üAMjÑä6±ží/E¥FKT½`(~{ËþÛÆÁèÖ¼W\?n6Àž?w•R|/î]'²L|%ïyÎWΡ½YÙdܘW¿sb«Â;KëÉèeó`#qmÞ…,uó¨ YO;Ë,À¦ytCL¯c¸ÕƒÃćŃ sé/ÕÜÝÿ«šs$ôÇÛXPÝä=©ä[Ž¿˜ä“µÛ¿¢$~åɸ\Lj”@P4,U÷œü"6Á–0¾2]ÉI¶HV¯9cW2· ’L¨šCÃ)ÞÈŽvu–e” ÌÖEE“ ´×å%œÊ9±e¤>äᤢn&·ñ·Yd ãäuŠí½Í»ûˆ™|Å"Ú/5jè¿À¢ý±-d7ÚK–>(<㮞Ø=¿@L›F2¬7µ_òÜyÚ¶M‚›ZBðþÝê”NCk.„±uhÐbåH×OÄè¯Økq×g¯k3Þ-g?ýUª ¼\¶¬ì,KLVН=Ñ xOèÀUHv"vè#‡ùa¯tE“é‹Tyå|ÜNu(<ØOXEKm^bÃ̘yµ«Kc^ns pW‘F"´ê&YyýF?Q¦xa%}5à…Òv*RÄRFtAíõPûW…æ/ PꀗIõä[΋ñ+ú·ö%^Á*5PArkk›¨{­\²‘¿-r¥Á¥²ºg¦,Iúü3BÑü¢…÷»ˆS!#ä 1³¼ý..:”æ¯tœž'×°#4¼KoÄ)× ¤ƒ×fTpF°±bƒ±n¢=öÅ ÞÝE„p}¥–‘¢ƒ{1­„o{6ó¡ š2w"xàµ-:èD¶µ3¯ŽÆŠ¸äà)Ýr:‹àÒ(qõà…]‘B*!K¨²­æbè_µáÝuFîä„:½Ï”&Žhu?LC?H>¹»„6P¾‘ô+xE(ÞÀÖðøô-5™† Ô8xûÆ“8P‘} ûŽ]`~?¬'÷ø©];º¿áÞ¶ -ݾØÂ'ÔÆ{…˜×›JÉž5«fì=“ Μ#ŸÒ§íOÒz*¬ó7‡¡¹0û'P+ª uñÁäyÎ}¸=0Ì\>¾ ¼ÿÊ…V³üýô¯ Ôcìd±gl:°¾ÔeÉ?Yù£3«-å÷u¼ø€†B#d¶Ô¶å7Ÿ~•ù'«†ëíY«é÷!@xqê²0·îÙ§æ™ä[jmT’ !¿lɯÎ/ÑYM^‘+»KËb»»ÑXVU§~…”•H¬2"{eîôÕ¬®,„ÓU9ðãL¬UÍü¶SzX˜BæÿÀ‡{LÈKßÛ¹ÿazh h§:ÝSÃ.\CŠÛ H\ Árè„8xj•®ÒvyØwY1 ˜2è±êa»Âf- Jhú`»â‘|ŒT°–ú‹ƒ²ï_b&¢5š\q0-óÈȇvºY]Ø1i]ŒøsÃ¥cŸÒM4ç- Ø}—íúL¤F¯ÁÆ{sà á‡ùuO[+ª‡ávT×DåÊ Bm5©;jÉF¬ÅÕuèÚ6´¹›_6{ºU·Ð'ßWBå`ŽR^A|`üø¼þ¾ Ór‡?/|µÈõOgBýP×M +ŒIñ—Z—aæàô1à«^)·þk‚¹{D:.L󴿃ú×Ge"×*æÛ`ÄI Nݬ^´æWB¦þ8–½iŽØçFŒÈËŸ šmr|¯áÜGa{3E5« ?ztw €2üzÇ~¬“Ú±úa}<КôîáX2<|OumÌÜGwô ⬎ØrcÎ 3Ònnvûîlc½¡¿‹À‘ÈùH¡äëÝ@4TRœm„$0#éJ{tj ìihÔžæìQþiÎ_%ŒFæ¹qÄxyN; Ú3¦Óa¯h7Íç®ïÿp|^¡’Ô73úÚS îq¾v'É;htwíÙGä½ë<ìƒYL¦ÂÑ"þ©¯Zàu³ñy¥|­tN%Ág’1ÿáiØ0ú&@hJŒýâ€oøc4Q»6uV Õá@ÜLꥢÇõêzý¥ïz O&¡™O‰sZ»Û¹7ìŒ.ôS(OvßÈ…Œ¡ï¾¦‡øHá+gíøóá…9©™lf”„½rq˜ð\›ˆ†?¢RÎ*É- =°®ê7>§“Ä}´œFSPQVa ”jo7¤K럵.ìól,`û»š8ò/†Wÿü¢ßhj$Ð(Ùä¯=Ø4ïð®…s)—#à݃ÞE¡«]|s•3æÁÛGwâVÏ©8`º¥¾GS¯=æV] É’•¸Â³·­â¼4A Ùíë±±Ò‚7­Åû³#´÷y“ úžA¾';o™äÃd^rU±›ÞxTîsÙdÝùÒÐ?ÚAºŒ±„°Þµâ·P¿éz»g?Þ‰€«+U¡2—èÄ¿6}ˆÛaÊwEˆ¿[Òè÷D§«!3/-Þ«²‘(ËïDh0YR_ñŽK]3î­°ž@œ·UmÛá3;(¤|’6i)à–{NSîs”¡IòðÂð?‡Cº^øŒ±ŽRîï­7Å8 gZ‚Óö]‹…øû£íé ˜cbnéYïêðЉ¥t¾Oí¡.g °ÚB7´.ܡ䊌5ÏÈûè§µdcMwc[ªœ˜r¿¹EªßT~ßã©Fx<àFU½Ë¦À“£ºÊ°ü³-3ê0‚É2*Û~-’Æ@¹‘’ÙÖ=ß0ê? Æb}†Ò;Èà4y–HÿZŒéÛ-’åà—Å]|=õQP^÷@¼¸;Ñ}TÊs_KéÚઠ“Jìÿ&Zªõ¢^h¤(?ÄoÊíÔžï]£­ãÖŽL¬*ß¿'Ð}3¯AOvUІÒêt¯„ŽQpáõ‘¶F¨žžòQ³œ b5q ÿ9­üsߢ^_*9ɪÞË‚T\?M¨´a§ÃͶÖ¹ñ7"83pxƒ`êú±Y(…ø_¦I «*¨{¾3M¥üâ›YÍí,Œ›óqSkFˆ±‰3¹UiLóAÑÐ)%™&ïŠkç0­Ü-lï6I®ö©ÂUzœm©ì¿Ã¦Ñ|·]ç>"tÎߥ ã<²á)Ïù¸µv¹B#xªj¸êMQŠT¬ dKÌp­¶šY¡FÎVûÑe¯V eí ¶YⶬÀN-Ž'<¯ÀnO¤<×Qª!T©Á…G®gŠ‹Šò‡I!¦À<‰•Õ:IÊpý–dê*Ê·†$¿ã.ù:ȵe0XÙªO¬cÿuØP ÿâÛîR_5ÞãÃUQg>¦§;ˆ€y‘Þ¿ÿUr5Äz¤‘æ½:}˜ô9Q—£Òèd§A÷p[<€ábÓ†PÝŽ:L=ÞH<šîšÔÇ)iÁ’pv‰íˆ•jÙWĺp£1ÅñÚTê€(ú–¢¥4´¤‡ä!ã¯_q”«új &¾c¸ü_M»ïÒL'-åÜ\^`h Ü %«´$íŸ~ £†­ y3hZÓ†³DúKÅÌYïkÚˆNt@í¤ˆøk¯9ªrB&ÂD“¤Y^ºõäõJoØÏHóÑñúfÿ¿½Ò}ñ˜?“b_"¬·a%|"Úcœ¯˜6çnqq丮5åá‚k ý.œ.Sðç¸t:ªI-O¾€8 Ïc·üˆ0—¢ùoÒ¡™™NT¤¶P=óþyäžÊ}™¡Ó¯?pÿÀš™}MbÎÑKß|Üÿ‘Ç®]piØsŸ˜¾†æzC1_Á —1Ó×n™¼8[ì•€’Ä›\ìþh¤ÝüçÝÂÁL¦¼4…]5 ÔÎÒL{.øÅeþ}ÄÁa_=Üô?|KÓ§Ë–kÍVT,ÂbW7× fkÜ(ÚXøå^3ÂuÌKÝç«í!¥™¦¯û /Šr9[¢.ù]û”'ÓÄkEßúÞ®¬Dœ ¹»2Ld¾IÒÈCi/˜º*ÓN>»¨±Ê³@¶âô鳬Å6Ó £pÈûÿ¬šy­]0~0®.XTè»ó¡~ " '@¶!„E¸©¸ü#ŽÍÚ‡…îýÄé*x2÷cKqNËÄÄV‰¯¼íg|¥¶|=ŒÜ®œ2R½óØÇZÃd†ÛÄ8mÝeWÅìB\nÙC°Qð|lt¡“Ϙ"ƒ§À¯E0T\&cB‡þÚï+ñð€m0‘¡PÇY T Ÿ6¬ñqíÇXÔž©ùPnЋõwO˜òôÏ~×ñQðÓ|£] Û2yAmI•ýiC2ž%½>d|º$ú‡×b>šIZÞlSLOZܘ€d¯%ÛÈMg ~kþ“[Âë¡€\+h¦×6IÒ-T8"O¬–jËOÐÓIÀ*^*ÃMŠLrpxÎY§·«} ãž)gå—p"ëà§Q“ù£tm‹òü8¾Æ 4¨S‹HvI¨*Q/æGF¶Ê…Yƒçuå^¾:—«Î–»'Ù@Ãø¢Wzü¤ ‘ˆ¯ŒÉ.'ÀXï;>ùRbÛ¹-?R±:¢ŒƒˆûÏs¾­ëR•{µK75þ™)±ÿßÖFŠeíP.[ÜjäŽBðøw}’hFž™½±/AëŠxähóÃñ+A¹‹Túë€ëç‰èÕ<ƒp¹SëàŸ\ï–µŸ—@\bÃz”O´¤˜!W1AoTµÿúþÐåÒPÒç;çL‚ ²¼t‰­ÜmðÕòÈÝØ©*p½5X˜¸êÍþ±~âv¿×¾Æš!˜ùS€2f®g Á7ëXÜ£^¢ \)'Q!ãšÜò*/‰_Ò¨mÁc@¹mvKô‰ëv.¬À^eši{nœXp­=Ù kUž`b^ŽþÒ ºÃTÏ7âÄV>Rc†”½œ&Ö-ã­[bµ8÷ÑèXâ5ŠrÃã]Iÿ†ÎŠP~(jÑ ´N,ùi2eF— Ȩ®ê°™i¦¼Ôw\»^Æún†¾œ.2^gX±ë³›À9x¤>t+ÇëwÞ[/Ç-TœÒœYKWkjŒ«rÈ=†d7àæÞù•׆vV5›|)lúYÄz2pß C¨^¯Ph,åØjJ¿XBY¢ð-#Ã*ûÚ÷Ê$ßgÐ}­uã&Õ›w"%"‹öµO<Ò Ù×}a;â3v’ïÖº½œar´wúZ;ˆžº‹Wÿ¡ÅÄ.Ž–•ZØä:6- Ä; ~o^~gKï³hlM`¦ð=(êk|8åx𙶝ŠožT¬Ž5SÅ„ YKÏ#Ø1è”g§DZFàð¿ºYBª,gV®ÀiOÅšˆýõAO®ô8æPÊø8•\Öç ZR{¯C2ÍÄh…šŸ;›'¢ïa½¿É{">…wÌ"•Æø“MÙìŸÌŠgÉe/=™3ÛŽx!HdÙwŒc»v¢&)rÚÁêÿH¡úØ!z:!2Æ+¸;6ßÄ´ô'ÎÌôp·Ò,­°Ñ)—óšÇ³¥=›:ä.ûÌB"K?‡{¨ °uðç yÕôÁpR¬ü’@+ôgý“L¨qušA³z'Ï,QN캺~{,Dë€}I†P2ãl)È^Û:誨Iáêå½m ¨gEì*¼&ÇÕy ¥nœ&‚aF¦…_b„Nxœm÷ÛCÁ]m¶!4]ì-aNrZž8ªá1Èê‘Áâ¿kT| +‰¸g f~ÜûyÕô…õ>4·fýðg9жD4Ú UtqPÕÿU9Îz„|u ÀŒð÷ «œgô‚Ž ãÊ[7{m.*L÷àŸï¡×ç±þN»i®œÿVäÌÈéƒä¶SYW¤™†º}¯±å)jy9õL]i¹$iž¢›n±¯EˆèS`ž @püf¼ºïñŒvv!2,ƒAÅýyÜ äG—˜F`$@½'²°lPVkÉt¯u?—8Aõ³!#”„œšk,°»)2£p{I ŸeMêÏn1w9¡ d?U eÀr‰ð×p—˜…u•ƒžq´ÿ»MÉw•5ðô…™1¢ÏEê?)ç\ßI3rOq,+…üg°2µMœ1b ñ¨êáB$ð[>-‰åmj ­pŸEƒpÏCJì÷>Nv®¬T,뜊²ëÅ\ BÒÝáØgâ+M±"C}}å„Ú éÌ£ rÂ×Û¦„Û´Rço¥lÎÎ;aÔ\ÿEI>ÆÖžùG`èÇ Ù©ì} gX{M{õ!w¤“ËËw°…ñæQ#™9Ã\ ƒÐJêu=–âKbR«·EUœ#‘r"³Ë•þ(®±ƾÎ×Éu¸þO|‡(oï•<›RFÞ3رèÿ|)püò=~„;„Ï.Ö!rãåqE—Îbô¯ãÄdwòñýíd —*ßäÿûã4N¼·I7%@^5æ¿gSe_wÇÌž£”§sÈ—|­aSÚ“5¯Ð¿•¹ÒSŠäƪü,5°na!é(ê!;öCôÑ}¾Û"‡$ÁŽ Ö·v¡„öñº×`¼ÜÒeÄŒ°èÔf²ÓTs)q”îCÓK¨Á«–‘tŠPÛ`Û° ItÓv´a¹a™‘¸XîÕk콑R”²jåÚŽð¨/ÕCOzÓ…Ô "ã†ÈEßÙ×ßÒb vw»õÏP{ÏZ72ÿ¡H]p{õŒµk"x0ßyC=0l>Q0Þ±õö­ÎtHŸ9­¨5Ùò9ΆžÂ ‹GcÊŸ\¢Œœä#÷‹Èþ–Q6o ˜K5gêt 3_“ÓÈ3PÓQZÁ²ÙÝR“>ãÛæÆ^aßF E'ýQ+"Kñ#/V?q©ˆqAëÙ—Šù¬¥6 üð,¿ÿ\úÈVÏðÞ-ÉÍ«P°4מÇÌ“sš}PòKÇ[WO'‡ÙªÃA¥Êvåu9cëÈ=ÉykæÈoΕî¬sÞ¥EšÇ'÷e§´àëúg¾¬j|^…R|Ä‚ëÛÊF| p×"s?† , §yÞîCn´½¯Ý¢ªšù·íß:˜œ&³Ç³ñú5Y´“„©×íƒ|¸í[æ7d¢n˜t…˜Í"¦Ö(лɲ+&Dîš wθ+ ;øf^+ŸÀóY:\Ÿ,—þšØÙRhÇEÔ ˜•×YÚ>´ÔÖñù»Põï`åýqÛü»ñ÷ÿ™^í‚­ôù:…¨¶\Qü‚³…Ùh† çVq¼Œ*&z¼%j©Í$¬Õõ²PØG ö`ÊA¨Ñ†}AþH]!åg¢€²‚©P¹@=®Ã ·ò"gdñըů(®‚!ùÔR ®ó®“li‘Ú"•~ƒ5öò1!®sj]Ô¥B1[$&ÍÛD<8x§ú¦Ÿ…ˆ7*^;…¬µÇ-üû(Mååã®t)]·\bƲ"Kö~™ÃS Ò<ý9ñ“‡4 É_vÞ5?£„ã¢DY÷Œ€þèž^Zñ…å£Î‘r¡iù ,™Ë~Ûï?S'3IŽ¢—SÎ| ï—YþYÀËÂ×_(* ci"B¦#„±=–¢új¸Û=>2³âw]ʺ±f•ÆZKPçƒ`Gµ>ëÜ 6I?t>WÒ»Ršþ ¡ÎNv|à¡‘o3!µI" 7RÇLœ6槴ͼ:}:ؘçÂÖ ‹eR ¯¨a|Ç5¯QÍÑ-ÇFÿùœá×Hæî¼bNÆÊ¨¸ü…½IoaK:B¡C"±šÀ–6̃G”À ¨€Fd燳Ðmìd„Ä4¬€J†#Ëïî0 ÓÙžji'f•ÃþLWÆH^Ê\DŒ¿ÉfH_ ÒæIƘÍ.ûCÅÛõ«£4Å²ð‚©ž¯ŽÈËò$ö°»ßêBsL0¢TñïpHáM¤3â#"(¿§fnô×îCVŠèäÉ<‡ º|c ýb½Ø<ÇZ…¬½*õ7ˆ°+°ÝÿËø·:KÚ ˆ2ŸóÀ6g±ºk# Þû¨j-@–±rQ)°­\õR$†:Yú[6ÊH.6€¿Is¾ß«:+ÿ– Š £ùÙoXV}Rmçt—Ð- [‹w8 è'Ù(N±mÉØqLJh½Bë³Ï@Ôæ¡Ã—ÈDjÈ${_Ùy%•™nð%TYêöjÃ…y4OÍèjA±.‰^–¨$(•͵©å+½*xg"¹âZK ÙÚ«Ênº=˜*<»…6aß}¤N»U­—›bõü]*ôëy 1a Ò+Å”·7‹ ¨@ÚV§…®,ÂÇï5ŒÉò÷ÐÆ;²”ªæÁ¬T¿\  Ô@ o<Ììo΀†(Jܾ¢–#·Ae‚9jQ‚âT r(nH ZÇð LG£ù륆_ñjôÂ9µœn}Šš¬°2tZH-=“¦u.?ÉÄÒ³únFâ8W{ÚVÉ Ë¶Æ 'ÒuÉ´6"Ý·"Kè+ô•.Eš’ŠîH&)²þùž$jØ2¢lЪˆÄlèrý´f´È6•—0ý‚wn$ù yh /B»HMÿ Ì-#‚¯‘Ps_½ñG0ÂúálÎ&ŒÔã=i'ƒï:ΪHi©Ì¤té†zÊçç{á{ PB ´Z¿þŲ…6ßÐÆT„eJ±¦gs¯l±XˆžðaÈ£«!üÌìÃùJŠáÅ¿ú©E¼i„ù's¹U©½5â–)~ñ2ˈãÙSK› ËÕÝ_åÿtï*ÛAîc…ç§­æòü0‘…[JŽLAÈÃÒ†¿¶Qú–Ãí$l0©>Çn@à•o¡„éÅ ²T â|üV®³Ì"¨ÜžH·g!dʰgÔ‡5€î]ÑkD²†ßa°{T¾Pƒy¼õEXAÇï&*Ôçl)9Ç“’Ö•$–»$‰Xü¤]&àë9™zOxqº(Ì~m'ËŠܳüK9M÷™ͳð£ SÕ÷uÃðéØA»tc¥Ú þX„’÷A²èIk«më…ôa,ÀKô‹—(ÐÓðsߦ´6ÌPü²7ñÀD§ –<Ým"”H“k§s`Ú†Ç‰Ø ÿØ‘w*b+Iïÿ4°Ä©VOÊAˆ‚*®®¨IµÏ¬ÍaͧʪŸè™o§lißï[~˜‹WŸ‚hÍÈ{#JíhH+¤{¬1o|•)å«sì^€å›À-¹™£¸Àâ›Ö/úÿÑL-Ÿr,¥'L"iûÕR‹ªógųöA˜`‡¸R÷nrÛ½¬Ò}cäNý°”ÿ̯Β¡©L=¤~–%VÛÝÜ"y™\´ü§ÔlxC̱ԸiK2ÊhR…Ô÷šºîR«¡jG)Ã6 ªU'qæ§ÔíIR±FÆüyþ5Ô}Ñu t‡²Óà‰ñ¹ÈZÒfßL‡°ÝâlM÷Ò˜FQÆ©T]“ì6öNST>Fƒ³Ëš¨ÉR“5±õÓýW¶ƒð W÷zíW%3ï>W§Õuu¯ÿÜŠvŸ›}^íà Îg0ç" ¨góÀ@óz¤u6? °…fþr™óËÒéxÈDÔ%Š;Ó¨Ê/·íBNÎp1• ÄC›jAˆú3ji¥-E+uðœY2þ+˜urê˜Þ[“˜Ñ¿vªü³ÃB(ÁIË“XñIh‰G&PÀÔ¬z†U‘LHfä szÐ3 Å8˜‰2b¸Uð øx[ׄ§ÁbÍ!šåiU~t”†„?Î'øë€$ð‡ÌU„’lų¿9=§BÕ^|Í^ü¾FnÃ9\CðN%XX þ( Íü{ß#é^ìVdBO5n蛤|P™ÑºÎe£ºì ü˜¼f±ÿæ66°Tí¯ƒÚs- Ù}Í#È¢"øSØÖ¢[‡fòà™“nÏEl–œÌæ¤Uie.ëŠF1]±øÈ¦Š U •wrèr(4ÒP:Qã܈¸o¶‡Ï Ðåz³˜}ô'²³œ×fdŸG£¦¹@ãšØ8ñ~õ>ºéئê J{{µÐ´0z¬€3=BKvoýÔRþ5ᇛѻ‰”·æÍ@¾JË‘áì$§ü¶Êg«ÖõÖ5©!Ë”œwÚ¦#ß´ñûƒªt,‡ˆ^nw›îûú¢ä M—5|ç´`fÁþН,ñg ÂwuÛ­Qþæê–Ž}V²ÕÚ\þ°èÕmÑnžàC“ `¹P@š¤œí[°Öró°‹ÎÁd'=8ªôBIªq·ÿ±Öþxm+=¿8½¤ê`ía¸%-ãîÂvÙ4lSIù”-‹þ‡àÛ:csQ·mÓ|øðÆdKÆåÿ6Í3§Ç»ŸZ®Œ3X´`äAâ’«Z7Ýž~ÔòŒC»ƒzlÁã$i Ç{Ïš 9z(akG#-Ú“~†U~Úž0ûÁŠÁÉcðм£Î9Äû»Iyyd:Tµvjt¨'+ѤÓzŒ#ƒËb׌xèO(f]-p¯Þ#û4p2 ªA»ìÀ‘?~0PñÄšÿ^5 çÿPŒsġ!ó@û¢ӃM¯c·TnMÁy•ƒÏq)eÌG°;¸=làB¼ƒ!œ¥Ç!ÿ}Ò„®Q`*ÃküÂÌ÷}Ë {A¶©.‘ª8³J£… Èþ®sÐA1Þ @ËÄcûPéQ 㦖&qkt4/^´MdÌ éV¥Æãé"ÇTâ[ OIž.40°†ºÿIÙ* ŸgùèkŽè4Ý—r¨  HrŸEãĨ6˜Ïw*9Þ(ó=50Ü<–·a‰qó6¥ä)`¯ôбL¬yŠÃ¬Á<£µÝG¥WÀÂ{¢§ü$Ùü½ëô _M§³Ì_XÏÔ¡-HÕŒy|m@L®×X±]`J-5ÜÄA¸¹Ñe~gêŸsí˜À^ýauÅy"tÓü %©òËÃËvzïüóó¼¿ú°ÎeU:X`ÕÅ¥¨ ·GX‰ŽªöÉ."qñr8O/#–ΚôÐcšï'->Eο8Óõ¸h»PŸ#`‹–{䃗îêm©ú³êÃ=‹sp(~‰æÍV }L‚pôÒÛԮ⾉qAßãLæ¹NÀ_Õ ¯ÖQ8Ñ~Ft`åÀ8‘Ňà;*aKrÓ¦"ñnÆ€QåÂéCC:½” î×÷%!Œó}ŽÞÓR#d°9ÚŻኴSMº÷˜ÂÊ}sJÕz dµp€Üͱ¶jÜ,]†•7e)ÕÙiݶzI²Z{*~Ä ù•m€ íÐ{ƒVÓY÷ò›Çfõb<±®û™Û‡¶dì¢ÀÐÂÒ®°“’ößás¡ŸXi¹°ör&`•éqd@é("–SŒ—®®ÈF¹”çJ¦Ž]tì†ðM])œæ„yX‘o2?%Ýš»)6dkÂQ}VpË&¢ø«)Ë3m’*¡wüøôémtWÒ»¡ú ñŽÛˆÕí*3Ò7l#L½ ˜ÒõH° £µ½–41a>ò*¥û¡ô:ÞSâtúAúe’€îзlb¿I™·)á3lÑ‚E¼Êì ¡œÌ”¡O«¶aQ§"¦mÈâWbùØÁŠ,6hè Ùº‰6œXg8óAÂD;ký<|­Â?ŸÚŒ‡1CuöÔ"•iÎhÎn>“FË»üÝ9«“æí·<ÒùÍÇÂåìð/è{¸?,=KK‡÷˜ž¯?œ~j¨3‡$ZÚþúi *²ã3\¸ù„u†ôA½qùn)U£<ÏàSVùV°_?‡xðŒX±¦pz|UT»fZÞRì ±kiÆN?#IÖ †l·ô+©ì¼ü4•&é±K<‹ïè=Ôt½ÂôcK¥¬Ñ–'‘m8gã ;÷ÿÃÍ "SQò‚ÉõgÒh¾MȪ\µˆ×Ò]•Q’F1OÄPbF&­ ³(¹fõiqøËAЛQ¹)i÷¨¦*wí,·P£ÓÒëÞµ°r§ïÌ!]͉'[{øéFlY.4çÏ7B‚hŽ5庑ìÜÿ÷Ø’œX›7;±>îÙô‚QEƒâÕX"QÿHÂû({#0qqú«0qÓryþŽcUÖB·|ÞEOO)èÌšûñw½Ð|érË®½aÄZÄ·=]ËO<0V-» ×ãŸTtý9 ¯ï!þÌÌÅJ§œp|â²í˜P·+ÓAU:‡ê±A«ÅíPïËLO;X˜ïøp:ùZÖ¼0÷ww…C4ÔJ”ýÀ #bú¡x¬CÚ-1€*Y×XƒR™h ñ ò¦T–£0S· @Ãå ¦ Ü\D<1c‘·¥òo¬‹¥?ëÿƒ¾Ç$ÔÉ ‚ŸÜ‚½>«©¤µížzÖy\"ù+åÍfð\,rxªPÁEú·VõqF ÈŠ£y²  (øÅd6æÇœqëèa .ü,ÖÉ÷#W¨Ù‡`.^áø c¹C°` sO©Å-¶ˆZt•Æ`÷‡„òƒ¶Å$ÜG0§ÁÕèÍîH~-‰×oV^ÒŽXw(‹‡®™ã•ñ>[RYwæ¢æµs{ÉàE·èWãü #Jt'ÆOH2úÛíjÒê #^ˆxz¥¹_2œÐžF¨ÚbýtÎ6A"VâjT,²µê€À×o@_ðìV¹^UN'-‡g`Œ|’“½"°Á3ꌠֻéÊ.‚°v?i•Îü yéâlÐôòÒ(äXn™Û\÷‘LúûÜÒgæÎ¼;yŽÊþ¾Lô¯Xg`t*®‘™ú¯ÜäæÒêêâ xs Åí˜Ùù?ÍÚx'ŽG¨•i‰xy- feê=6ªÉ:BoÍŒÄEÕ.½ÈBÒ9ögn’q? ±bW‰·ËF÷ß•Ä4!ˆ-Ü#­ ˜üö®úÇÒrÉ2Ùµ+t„_úÊ‹·`ºéë=f‘^ WÆDÜîa iå'VaFÂÃÐðÑ8k†PGU¯î îIÙe£âãÁ”(±2ü‘¾‘Û›¹ªÄ7ùLQ 8Ý»çPÿóóñgÐ3 žŠoø®iÁ>‘NëÞÿD!uØ—£û„X¬ï°+fQ9¸3<n] ¥ÌÒkÁÆÛ¯Þøú­S‘R.Ü|ËNÚ¾µ«åÂ?¶™±r@'‰55:i}ŽÅ(Â/ó>랪›1ô=2ãþ4$ȵ¦QØrZm'Ÿ]_Ý5«k"¨XÉ«aAþ·®Oüê"Uä}pXü>Wu{6E‹$1<›×íÝpî+«”fK®8E#F`µK…fVÖ˳zobéé·ÿ'¸S¶´{ÂpŒÄŸk] ùjj ´“FLZÞD¥ZÐÿ)„ì¾m¨Çè # ¾ÉØ:XPª´9O0ò.=T³¦s\ÕÓ*»X¯—ÝNò±úÑ´¶g¢œÓÓ|ߜؖ(¹èAH*D’Àég¹Ë”oF›¦ÃÏh¥áD×À5<Ü Ÿ¯mVÂJGr`' ¾ø§š—õ‰«OD»LhõXü”CéÂ+Cdì‡xËøpM+Ïb]ç@>·•¥D Ú¹*…i‚±.ÔVLÇ•hRKuêçÚóRó:´²å}ƒeÇ:^­†uyÚ«´;ÔÁÄùÒþÊ/ Q•`Jaßjµ-¸É^V¤ µþ@ª£ÜVrŸ@+Eâ×óŠ ËÙI©ºP6è¸ð¡$HxnZ˜-o'bpù eAÛ“Jî½ïLX`lD+8$Už¨Ó"¹Ú¤bF~››ö¢FB ŠôbŒBŽß!ân™ðZö%‹!âDÇ÷ByªŠ‹"Ææ—/Á€7ˆ®g)ü^·£Ë=7dbM«îJ?st=ï|º&4ZÞ¤¨s–©ø 룘OôöRõmª{d÷öLmµEjË#R-pã+Nµ*i e*kž¶ËÐyèÎ5ðj5Ëéz¨lèÃ>«sÀÊôQh5/ëqä0×ÖØ‹ã&‹ƒ ›…}н_M½ª¼{Öp?މPãíGæW%¹~*–¢M¨hGxjfS;¿òR–ç4ÔoÈŸ»òTB±¨ î³â"+° ¯uCÜ8x!ÚÕÙ+çNò%U3,–û*i A¤ß |è×? så–’q*WÝHý»†~Ä»÷I‡·…Y‡0…L-´88#ºu€´’Ô~qÊÀ&Ç‹èd—d[ÐxÓ§Û-Ö{ʃx×<ÕP$¶g8¥Ã¹cgÚ/—)Šz}sLľKcIKBEóëî·À{u€ã†…ùÌ#»f®ÐôoÚÒίÔt,Q$_ʼ䈘”y´¢eK/\ ‰èæž*ó/NIÑ7`«Ì‘q·‰<áçU²¡.\àÆi“iR;/rø)=Dc*-ËèfÔ@åÿy•~fô 2Ã%Š-v€²‰57°G;]œÝW§ê¦dó U©"q"}&¾†g)þ­É”3åáÀŽóÚSt‘‡¾ã¤åÙO0›a†¨¢#ýý%{è\RuÁ4!‰š7ævø²(->}íÓg{÷”øß#RvoŠ%s#"_SjŒGýQ˜S‡U,ú>ü…¦ûG¸•R¯w;×Uä êÄ5fþûËc=bKÒwêtœ4™HÐnõqÕmŸU skôà‹³9ó\íì„,¹ø<øY!¹ÉåxÓ4ãÄù{ŒÚ¦âg`»¾mt†ë ÓNßÊ;=@œc¬¡zÀÀN^,¾=üˆŸ°²vììù›ã4ÉŠ˜²9N¡áXi–ÔâH9 e*vz2¾Ã "‡Ø/çUãÆíb`ò†žæKFú‡<ËÆ"{%bS*6“¼Ai¹ïû?°µ¼!t/é5& 1Ö0,nÊ`ºª£4‡tŠ‹¬Åtx•o–ÆFÌ}æ”ïÁ넰ײøûì/‡”­î®R¿ieÍq<(bá‚:ðOÀù/k;°ìü—E$<“"ŠÃI€æð-A:¥ ˜sgõ#TV ;U^YvÑt†ÀßF»?sAT Ýf+ øKÚK.rüU¯´/ Žå»õ²Ëøízý*ctzš¨¾“ãÜ''_Ub#<à¨)Åž=l¬Z¼S›åóRÇ©"tò¨¢[¹írÇ©HȲÔãNÓò‰Žoõ¤ÿÓ}—Š(kùZ»|ÎPÿ ˜’µŒ¤B•†§ªÌá¼Üà1-gûõ²“ÁÇ+ß·ÜqøÚb˜zê .2³¼ `dë3û¹DŽ\,ì $þxggvî%8ŽökÅ·N$²’¾JɲR5†ãÍ“­Ì‘€¦J¬[ ?ÀJèdžâƒË<8·!užG°@¬üö%!G™wŽx"ùÆßˆ½Y°ž*bSÛ¹ÝfW'3Aù¾ÙÃE Ylôb˜‡—õÔÂ< WØ‚8O,Â}lÓ¨yy,ß䛫¸w3L«xVœ²Ó²_O«Úq®¥#ȶ A3=PU‰¨;œ&0äÂQC®o 0Õ›ÿ´ªüg¼8 CI®‘²GâðE†2‚Èoz¼ó´Ä i+Å]&ˆ?ÒòaÀPPå&Çw=FbzWÃÐ+éɵдfÅ`µ û¡«û„†ýê<ÖÿSˆÛ „MÁÐÏWâ â·Õ7ŒÜnvs-X29°»ÎfyOi…úíÚêvóì:¢ÅbFš%(º ³º¡ŸDïc&ñL?wšÎ§wSg˜ð4 ÷dëV€¿<ìÊöA_?iv““™®$D€G³U¾ÛLÞV`תmYùXÒÔ£éÐàbƒ²Íz×ËìDÁêo§máÚ<ïëN^o›-œÁÏ_›fñï.‰Ähpe¦‹`L ,„+­¤ €Æþ(ë­o'z;ñ¬Éæ~¸#‡±0ô7BêÅ‹hzÌa®ù¾¥—6 ç8ÈäŠ-ÔF'fÒ*aßÐÔ‚É})\YÚioÊJ‘-‡Í[;zÅ)ŸÆjˆS’£*†ƒ˜ ×BG1Šl㦠µö2iÏßR•ƒ^Û>.йf™?s ˜×ò«œù|XÍÓ‚°îª¬ ™ËëD‹™hí™×Öc³Ú𤜚íhŽÅ—œ|íëWÙ¨£yüœÙÎ:x4Me2êw¡J@·G¿ÖgV" Þ\>Ò íÌ^å‡î<)BÇd—šËKmþm|ð²ØðCE’Á{Ù<ûbÕ ©?ôC6',ü%OgžKºÈ±—»ŽÊï’ʬjÿñgw¸qœ »íË–ð—?çœ ´Ë<œvñÇ*ÄÜÔ4 KwŠcå9  †)¶dªTP*¬µÜëA›œ‹"¹aÊÀÊâ½ õkx/Lz£˜ZÃr¨žIšîØÜÌZ#q aÄûÿM´ßž’^‹¼v½']"u”ëÖ‡¢ˆZÞ)°ß1·Q“_wäk°M&M02±L,óQ¦²IµÚ©hð!½*fÍ7ó5Öü)SQS†p=éÌt=eJ‹Ö&ëf ­”0ãÌø’yk-±‡ûÉ©Šäù“Ͱ·Y¬²X›…áXމâÒúÁÍd*¦g—&´,)×–P´þël/rƒºØ>–qW|ÜHAÄ kº÷0þg ý¿ ˆ—ó¸Žo}ƒÕrÒéŠH}9V걋F&Ṉn„·y—mÖzÈ´+B·†ù¸BdmE0nâuVerÓ€ˆ5I1 9cuÍæŠ$ltL…I`¶ÜR.å' ²¹LÊØþj›RÑuÙ&RXÞDò¥B0S¾èé9àô®fQ‰6ir¥‡¶?¡ån—›xâF€ÿˆ…Hë¦úÛ)^_cú~áÉ#ákìG“¢v€ÊÛÞñ«-ÓÌo“½¥µRçyçÄGh"ÌŸŒ÷ž°ˆYýD&éæŽ®¶àMáb®Ÿsòù9_mä:ç¤p˜!¹A,ís‘;¢²Ã+‚}œÏEÛb|ÃvŠ*¨l2èÕó]äClê¡›Û*Ô*Öȯ4-:œ­…ù¦¦Jí6xå×'~!c ¡ÈH[è¬wb ¨@ª¾ÿók•ø™N‰ë¶Áé5ï…‰ bÍZæI2Qu24yÀ¬ÞṲ̀;b²Ìi·6áípK™¼ÀDÚf*ö‹)8b@ö)7Põ½ïG<ê©c4·¦=ÂÍš@ë~ü6œn7£®ÞE·DxUEÿ‹W4w`|ÞÂFtX}ºtqs÷Žƒ “à›Y'¿}W?•aÂE⇿«I¿F 0¡Ã뤇ÃÜ©D—Z‘xØTxtcªpŒÓ\dûáN’ÅhôÚñÿ1ŽÀ¹œ•u8PÂfr:ûd³Ç(Áb—#VBØÛ¢èž%8í¸Ar¾Ú‡ÙÞ0§'V‡œÈ ŠÖÑáQmâÎÂeKMº9×¾‘UIÕr©lŠQ=¢|ðl©‰V;½˜@D!Sàc|åe¿ÙEÁ<Æä”Š`6ÑaÿUa—Ó)ùÜG<âËu–e™ ¤cõ>V†\Çóµ@Ë)°Ë|X™pƒþñBÃ1Mó¾€fI MÀGÛY¬uêÕÄt¨ƒ@«k­8Ó.:Æ~Pü_ ^ÿ!œò“$ݪråN˜|÷GµÏ¡;ã‘¡­Ôõ²Ór\Ï1ÃŒSªÀs¡|zù”¸³ˆ˜9Ã<­8i°ð1Ìȱ“É ö‰Õ¡gÐ8Çüéãá=ªäƒë©rk…Ch{‡ü6爠a¡Êß™gH¸u. 8 ÝÅp>ªJÝ×D÷l‡µS…½ªÂÜâ8÷Ï¢ÉN>ÁT|RÆÔf¡°9ÊîrÕkûíâHI”¬@YïÕAùÙ¨Hc:²©ÿ‘9>„2a©ÈŠAø>9j§BÒa†%Ǫ^Â!Á™‰e'f‚ìpÄUË1‚$¶~Ê97½òÿØ$I€ÖP¼œ«Dz ÿ7n  Nm¸UüáŒØÎC‹ÎŒ%òño8QÕV8ÇoÔ¼XÚp/¯¯Uû1‰¿Råô±þŸŠN 2˜WþX z"£!‰¾Öži$õþ]¹6HœåJç»óݶxzVç6ZÒHoT oØÕ Ò'•ŒÈí†d*:Uö‡ßF¦4›íMgB^ðÊ–u–ŠC½ÌvSWŸ?{Ýeu:I"ºÍì§»a¯ •DI°1ë‡C,4=$>¢=™6ë‹BÄàs/ئá…M¿¨8Øõ ó>2FŒñEm™ "~PÍÆ^¹ÕJ°P8âtg!Œš¸ý©È^zU•"?ßÚ„fyD2Ü öd[\šŒœx5"aâ¦-y ~‘ضHs†Ë¥G÷‡|ëˈ{bA“| ùQ'˜Y?kJ) Y§6×u»Í TåÈ{ß³>"-ç±Û‚e]a˜½þtÃûÆ[ÿ[q¤’™|ª ip uF©Š"½¥Ë<>+ÅKö,ûÉæˆª\–_º] >ÀÅ v ‰Ëý+"$Ÿcn\ÓÇØÃ&‡Iä®åj$‡`Ô‘_ì Àõö6û $598[Zœ\št?«ØöZ¯®Å ÃD¥VlÍp[(€wöÑÊhCJµ£K'—¾ûÀ ¾jïÓõ„;Ùÿ£mLÈ Ð6pdî{ ×Aó€éV ©2¹3Ò0í÷.Íq3¶L€DïÖ{gœ¯ßºkC#f¾vß³Jð4s3…®J¼ N[d‹8Âm¥)ÁŠH’ÓZä[qœö9ö¦8ß{‘„øýì‚i Ö÷ˆ‚A„`ߊ¸ÔG‹d•­Ûª®~ÆÐ=¬Ñ}t.ËÀâqÕ6d(ÅTÅ D+Ðý ¤üä碋²Èa&ªÇ°0 ‹ê°P~*9Œ:›!2Àˆ¾°r:„´Ô€¹ôñ…sWœñ® Õ 4´ÙhK”|¯¡ƒwEªLi¯â;mžãvp ‚˜x ëÐn$ ‰Ý«¶ïR`rÞÎbÂÆì­Û¤rþÙI·¿KjFÛËžœ‡E9z¥? •ê-Ýø$<Ç-hv‰Ï±ÒÍ{ê0?Øïqžÿ«DÍëŒjli$~ h>+xí¦Œ66Z«ÏÓ{6fÃ!öô™ÞÖˆÞ'ÅEï ×T8¬\ ì’íÍÉA¦z$×Óêq£¸À?ƒ˜Ý.l ³AL%ž)ȼ´•Ó aü ð<"ÒLKü)L÷í[÷DJ@oºmú¹’Ø¥Oßÿö(6àaÄ#ýšêP^Ö£¤ìaYL¶ Õ<‹Ð:¬ ¯Æ½üªJŽ“íùwöêˆ1è.ôüK¡jƒ‘ÿhÆËõ,+cñkÿ±ä½wÃTÏ'q5BÍmýp—ã»Ëyé„ÎŒ™~J8:@¦þ¾– ÖÑ22ChѵUb!ä^ƒ$ÖÔ³G° ÌAÚÞ'$Yœ}‹j/GòéÓÌêG¡“"Y¾q8ÆN¯ˆlÿFn¬_ÒN䯂Þ÷¡™™å«Ô?³+P¿ba*‚·È%@ÄyŠ—˜et)°Ë]íª"ëιë0{¡C%Æúÿ{ š)@l‡Ÿ¶‚ÝøÃ˜ðæ¼?lÓB¾Ufâ”w!§\F—xýw6Üò^¬¶?æyé;MÜ‹vVægkv$ši°Úœ§·ü´ýADvÂj® ޳rÒñùŠòéI·sø¶ÆÂä< (bÜð¦[­šlÿm`–é ªŒj­±c‘5L¶^6C'D6nà§äW–¾ÂäòÂCY^‘ã‡b7öØöºŠº'ßà?B"›$¸‹{“ògh±|FÃ7DΆVÿl~uœu šÀ8lOùcx»kälöµvö³T*C¸Ò!‹ï¨´K Àeì x¤gÖ>ñÜi~¥PŽß5çe°ùhEDÄì9—£¶P^²­¦á JËkkVËö&Ú?ŸL1× ÚpSÈJÁ×MÌMîQ»îe°ÄX(‘KIE7š¥^"ŽÇuK’¨ù|•ãí&H•uSüzÞ¾IQCäO–Þ.Ó‹$z/ùÛ*•ØnÉÄ6œ}ɬ= -3¦ãðµUzò¿V^f½-ó¶ý¥`¼i¨Gû×z \ÊL$ÏÂpoäÑ¥l%ódc´BÍcÅŃr¨PL¯“hk*¾ ˜ü’™²Ê±ýÁUc¶XÆ^7ï%t!òf¥=gAûFLoØÃÏ1EŒ‹ÎÙ<‹„¬HË J)וð}”5j»E¾{Rªèê¨EdŠÜ*wiÈpÃV±°Bg-ÀÉB‡Äó(ø©5–ÊÄÑÛˆºÑøR°´ìÇu@[…µà¢ÍNîeÆpx#ÿ‚ÀŸ‚I6.|Ñ"Å*„ÛJÛ°ì%öx1ÖÂàJöðk*‹´Ù¦d:Z`œ;…²Ü!­<‰?{„7!Ôöl» õàBKžs¿ço”.~W¸tÉük<– ³?%˜àú›õ(ÜMå^Ègcúƒ©åßX×õq¥ ÔÉcñ8=bP¢–ÄØøua‰46:ˆË‡ò;€´ï¼bÙJr᪠­§xΔ©BövÈt+èÛTUc,íaËuxëò»ÉÊ@ 1\¾©ªMAÅúíN R½£:ÜÚ_7—¡4ðÀ}×=2¦ ø’»©d´yÖÕ<Ú‰Èj÷k¯Ë‡ÑJklžoa¢|ö®ycMïé¾§‹ò…;N¨ þÄwb}-áκ*^õuâW*¹è‚<û ¡ÀqÓFæû°„lV[=êj³/Kˇá­²[GÈ¿–ï5-§¸ì'¬SÛþ>ð®vŠÍ:`½œÓój<‚ Ž’ÅÊœ°¸ Lyº¤…,wûÓ}lˆ@\ߥæ`…G£eEË}ö~q±ÜXÐXêÔ~Oyß7ùôcH:©)p7θZX–„J,GÆyÏ'QF„¯I}Õ2åÆÒlbZk]om§™-+ŽƒY}VÓ¸YF w%ë÷‘â0¿”?rp$Š®êW$,-á“ÁÑ»®då,=ªÐæÀõÉeÎ¥Þ%ºÃ]ÕU$Ü Ñk°ŸS0D”UiOpRIW1‡· ™\±{ª>B#yš%žºíÙ]…¦ë .ÑÑ;M c\zrQ}Œqý7®¦µš´<ØëѸ,‚hÇÐ=„¢Qù5frït¶?´Ã¿³y4fù@©<¡ÐÉ“ÊÞs ªÔ¦ÑªÌܨɣ֚¢¬¿ïHµ¼HÍqühÕ\xÏ ŽBç|÷~¡Ã4î|LKÑɘî°›òK߃߰ÎÜ”yc¦ of„„;%8fø+¶M5^/›Cör×oo;ó´É‹Å\,ÿÿ» 2¹¦ óû6°‹›Ÿѽ„Ū/ûº³qÜzñÙ=™&7wëìÎ"¤5£;Ü òsd¸Z³nÁ(ý¨;õåêì}¼7|B5´€QWËÃsBO{öíàùö 'Ö8'I$ÖЩ±¸ÜÄüÓëJ9I%P@5 RäªQ !­Ôsô) n–Õ­w$ ©¹{Pr޼^KG‡Dt`D…ž¾õ÷ ­>·{è„9„ϸû1œ((JS%pÖ)yå>î¼IÓã8:ó%ã„Æ ÒðÁþ‚òú¤ ÿuê…E¨D¨5O#ÀÁŒ¼cç‹Q÷×›Ú¿uiÅ‘­¦Ï0œb’”Îh‹Ç§p­Ecž<ˆˆm„Ý«“ôÉÒ X\Î4ýØ!\ܘº9^ÃÓ6&ßh\üÍPìnM²áÝÈ'“ÓyúÖ\~ôˆˆÿß›%Lj4TH„y`‹¹= Aüñq¼ƒ›ex{]9!&¯MÏ\;ý\G€âˆ¶"dÆ3Ïaû»éžÈÔ«]Õ+ÝKþ…á;+Ó-ÃÛm¥q$FÎö¥¨ÝvŠG %XoÖö˜ñ;öà kòí >²qaÉ+.c€æQ†Ö½%ÉT6pU3²Í•ý~ɘÁÚ­÷¢1M¹5m ]ÊÐ&— ÆyN5U[ÕÄôŸ˜ƒA3¸o;DGÝ` ý6Á»ð ¨î ÓwÞ)sƒ¤Ze£±‹lºAƬnÍ0zô˜«•nñɶòáB ’â©6™æÕOýÞ]NÀÓ×´¾ÎX\²ÑN±¬ƒ@`Ã… î¨ǽ›3‹±²0 ·Œ˜bÌ‘üa`ô\ts¾(ØBpèKÞÝ2…Gó4°ý¤å®§ -0â#Z e¿r”u-Æ ¶ähp°H/KNu'0ËfÙEò¤ uJ8ü>¼Ç¿sw|TŽœõ˜Îl¿ö K¨HRè¨ÿ­Z <3ù–Ï™°êÒÞîD¥.ÙÎ#ín¸F‚vW%YýßÇÈŸk¼)F{é\q.šLØ›çœÛ˜ÎŽú 0†S°û(ç‚%Ï Ì8˜A"aÞü§q¹œA¾Á›Rè(û¤+Ôí¤õzª°Ía  ÉÛß©;çôt9q-7ô›´™¤¨_›ÏM˜—·N%‡2OÁkgjê†äVçh;à¨ß“sÜ˵ûҰó©dìÙ6O›óÿ¿õ_ú¶Aê%Bƒ7Uï<-¡ÿ ±HÆó…ÝÈbFºømȽv@ l!Æï·/pIDâB·žø³ëò¦‘<Æc¯’RmÖ_˜6ì%ÛM{È›Úì â×›÷ÍÑao•ŒóIº<ãÝ…ïØtç®ÃçÎ:ñO·ƒþ†ôA×Þ1ÿ†ß­È n5»U B…åÍ£I”#LÒ»ÃR‰”Ôõü+jðl˜ …v»ö~@6Z‰¿oy#:–fYI¢’?µ pªªŸß *sÊÅ£†ˆ@Ù³¨nêä¤@÷=𜥦¼‡ÃúÄñé{ÚE³¿˜PdjfÒîajÊždZ ÃÃè'½é&'’ütô@S_»ª–C„<ØŽ_KHà÷5]Â˦ù+$årztÍÎ>tùNË­PSpyßÝFû=÷ý’{c!¯KA¶Å׎* g1ˆ >­Õ{‹óÏzß“k¦D]‡`$¶p˜ß­å>²åóWÐZä«ÉŒ¡¢{ŽÃÔj %*žƒ%%(ä(Éeƒ7²­Q‹žŽ9¼ïk¢4|[/„_å#³t(òºdL»”ÚüåÌv*|0ªÕ*5¼nfüÖ<¬-°k)€kJÞáj/f2\ûÒÙ˜¥¿4n2È´l|‡퉎zIÕ?¥¿OÚwX÷ëTš"óÞ> ݇ÓŒ–ýŠ¿\ûg6Çù‡DGN í`èÌg–T„§ª®©ÁŒ~¯_¦XWÛwÁ•=dÇß’÷‚-Ó@`*DK˜õº «l»!‚š)‹Ê ˜¡†„9¶q2ëT'%•ÏP‚*˜bNN¶4ãÞN“~o:€É76`Äãø"Ú÷‡0Þ°@·óú ¾²[ÝS^V;ÀƒÐXDµe•\{^xa$ød ãNY«Dñ=d70Üà×ñ5ÜÂDÕTĺX Iÿ/¸u D$ÕÁî| ¯¼»Vy!¿ódÓ…by-¢+Q¢ÔRrÀA¢^vnƒ=ôÝT†$ö/Ž7Yäiº}õˆ%éðÞ¡š=˜O/líköKt„êÿ7Ĝں˜âÈÌrÉR28ö/¢À%cû¬¢÷6¨/Jvè?–(Hd½ëK{$pDRS(“ÑÅØeÙwµ»Žµ¯Q³Öµy©U(GôIl–*f HpqHynã–ï¾5åŒôâÔÜ%yDXì,¸IýÇìZÞ–ÓYd`+9cZj¡|‡œÍv§TÆ¢Óàß¡,tvâ,Ž ß›GFמš#ÝSPJÒ³×Âð4þÇûfÔ“÷õ{¥þ¹BwôÚ*Œ$dB—óø-8D×&óÈbúɦÓu³që -= ŽÍÿV/"ImHg±~HNKTœöwÌæÞ¾ò ¤ ÂI|Ôp¥+AÓÂÎQ\¨$kPÒjˆÓÑd”*¯£þo_‘©•ðN°Û÷sÝp´Ãný(rK©#êÐùÒ&é'Ô³ë-Ð/×ýå“úô 3f[£ø¸–Úub)S¢èîÍ·ÿ)òÜýKJ–’å¢ïèCÌXNÙI†§ß⸄¥&U…)¡xåH¡P Z¿E¬ï>Šžð_ú©©}aΡœŠµs{³ç¡c ù2ß5"nÔ.v[Àç 9|ftÔtn8Ž<•}7¿¶’:€‚œu7é( ¢{™¬k2l^˜?!öÝt0Ä錄•6í3Ì ú¯{Aæ:ëŸ\­òc Öz#)mÇS¼ðKoñ­X/. û"ÁˆZ•²ôuj-ƒOß™7-&¸bjÓ¦™Î;ºùˆø«‰`5í3AÈÀc³v¡5q„fˆþ¥zÕ‚û4š4(†­0‹þ±}móà# Óîü:ÁL9ìHš†¯B&ígà’}v•°„sÜ Àó!j dC˽\1»â¦^B»-DËäT'Ù£ÿóR®V.|Û ‹¦Ï*g×¶¿à}½ž[ØÇ5ÞŠQWÂp»N/>öÊÖbt½Æ³^F¾KMØ8 }¬Š_æa÷îͲÉ΢Àû ˆ‹Ákf#ua/'tí¼|ƘÝ#åÑ0¤jùCj1ݧþv©^C³¥tØâµdÔÝÀ‰[Ã;3  ÿ_¿Ý³Ž8îfÈ ¹ÜãÑdÚvnÝ… „ƒ[R+!B™ø§ÒÔL¡Ì‚ö–È=-2]Ï£ÒXxr¡TN ^{OÂF.ÑܶjÔxÏ#Á_¦GxQæ8Dl©j 5ˆ+ý$s×±‘ÿ,W2¤»Ç•l·ð·î÷åÞlzro»E¹•â/*¿žqEwÑ”’ÚèLÔSgi Þ‡ŸQüñHߜࣼnå1­I9=g -–9ç1]8‘¡>ŵÀÿ¹"/ÒÞ͹‚tëßÀ[aú»:ÂûŸä×øc¦†Yè¤È ‹ÿ¬Š&OƱúÜù@<õ»¡ô/Á’nE<†@†?NOÖç—VrþЈ·ï^Ráà´Ñ@4¥¨‡äL¯‘oüO8'j©½ðbüèG¬ìkñ½×/‚küMÀæTmeOÄÁí_E`dåÑÞ½¾YØŠœwæYÁŸ”tº$?ÏÝÂÙ&ˆ,¡PJ"hTžOæò›3(¡÷ü~Ä…~ÆÔRv®dLÄðÂå#NÂ^‡æÅùäÀ½H”Çå×C·H,”fpJìÓú¤.á/pË“,ÜêÊ€— ö½D‚æ‹e)¬A/í5ŒKQÓÒ_Ч*ËV=ij§‰A7Ñ}æ|h”dG(GHóöõcû8­§þvø’ô3Ø@å#è΀žç¸@}׿Š8ÛÖYJ*ØìɉÃA¦Ï ‚¥Áéë½oÜ÷f¹~ŒüÛþò sü ß óN)'d­€ù4—&_ª·÷¦À±æ-è×E¡TÇP<ë[˜Ì?ó‘éYªm,»òÜ«^Áµ–^-Û’Zöc3?Orçx^Ë*oŒMJ7ÎÆätÖÀøÅeœH»¯vËð5{ãz˜8-¿µÚ¸f¥]GªD=TŠŠ>À'¯™Àçʯ?Úl0ð÷ÇÏÞNYÍ—öþh³ÐAÄÙ©ãUæ"'Ñcé] ¼#Vê3C±-íà XuÃxVJ °“êD+Z«þ½w´kO‡žðCæÂn¼?ô1”lj ‘Ë* ŠÈȤÂÇÔ·©æ·H^ØÙWÔºqÅîD2ä D=×%Ýoâ•`í-ªÀÒðbÖŠCYo¤yª0ÚÇ—¤"SÀÀ'ÚnÏ)ÛIwâ96Ÿnšˆ‘u5nó%R^s…F Òœjó,:q+×¾âF—f6|#äšw¿¸MbîE°zÐW®Šê˜Ã0®T2õt£ýZLc¶Úhò¦´™Q›ä‹´GóâyHD¥l˜½o©”8§ 1ºÞ ?ꯖ|üe6tÓ¸>‹ÌçŠÀ?{µÝ¨›÷¶Dë@nƒ«.¼æ‰mé…ÏÀ6fYB%•ªR£ðƒje¯ø_'ò/MàM†ö«·nµÇ¹dïy¬!ÍÛ•¿KAX½×tÑmÄ€aç¹JþÑ2ër©º­Œ`3”>é·:„ ŸVAN;òM<˜ß‰Œ·‘‰‰kT[­c/ÙLë½µdˆL>Ù©k‡ç&†ó]è*Œ¥AÓÆ²ÁC=×¼¾S=®¾O$+)›Ò¸CHzÔ6½0-‚0õº£½÷»h­òÒbýQÇWaת2¶ìZ¨¢/÷DÏÙ#6V£`j±$j'¬×¡@¨²-}€[(ð‰ o?5Ì¡Ç9jƒ† bOȶýú§½µ´aXkL¼¯Kø’Á×6D¯%¢ó冽÷/æ@ü¯Ñ˜D›µnÔO†¥¨%dÄÖ•ÅL× Ë¹S«ôrÕ(0¥¨%È–O~#Ï‘-³qå’;9'‘Hirvm”ém3ÙLÎ1YÎ2*Qåõ9Ô¾zém]ªž/Ð~l©CE­Ýñ˜Kôá]–¹''—HÇòâ©_–šNÛÐãNÇì1œˆz˜˜ºšVðh]ëetJN… øØ<†‹7Þv8{kùgÓ˜ÆÑü¥ö÷%GôñµÂ%ì,aç—žOrZËÙÎïpµQ$,¡Ž¸‹?È”IfLK¸]Ó } Cˆó¿îy£ÝÝÊþ±ß‡a·€‘è;F{¼‚ík «Ïø-Åp“¾ªÐ$€,ðzžb™ó-9g£Â­W.@å")\âH®‰d5“@RwäÇŠ /pð*Ýø¾¦í’ºL#ضO8W~7ïÁ¥ÏoË'Јnäý*Ñ Í”ï9’PcwÁü^©º+©Ã±Nž%Šr ý;â«ýMú˜};Oó‡Îâùšû„@MCHÖ>×þ`ï (գљ!.ž¶ïa&—Õåp]‚5²¸Ð;8¨RB„÷ïÄ8¨¹V‘j¥k"«ü¾dàZƒ †2g§Š­]þqXÛ'ø&–Ÿ…ç*QMVx?Zºö¼P^HH÷È_ûä¸ß ,·«[¶kCy±[bqç4¤@\œ§<ºC) ˜¹öÔÔEáÁ¬Yøø×ÔmÙes›ˆdx ýy©7Ô¥™dsð»á«KšÉ„såe=`Æ_Eˆh³ÙùðvGìi‹S ^ºnž}\/C uI» Ç1J:Æôâ…_ZøAØ¿”ºÀìð£Ûsb#îîÙܽú~ëØVà)üÕSØ}ø{|ÿ­*4ŽŽ}ˆŠÆ8Vˆ¶g§ö…”™àFôq¥=ÂÃV¥HÒÁÀ D*ƶ¥£‰ÌN¯¼c^ÝŠ-óf<²…1·øi"ªVr˜.2Nò¼¡þHÓ¸|B‡TOÙº9^õ:£?\æmc´vð-—ƒ®BÎt?Q,À*>U}Îz³R_Mxà·?_1Eß·ÇvÕ ò‡­¿*Œƒé…ë +7mÎÍðêÄäR£?ñÞ"󰸺è~2„‡!fþN–"‘Ä{Ø¢9É\@¡÷Ç£ïý鼯]úG¬Àä¨Ó’LlÈJÖœS»N-ûãZ2;TGXæmaËÏHe ¡{šÖPV 7Aú.ŸVŽÁå3J}§hÂþú(Å MÁóƒN$<}®¡öê×h¨îŒ“},©ì¯y¡¿y«ÐEoǦ¬½¦t!%?Õ/?kñâåX®ê£{Ù·”ý.»*(ú[ ¿gy^òƽï‘Ñ r4Xx/{s&æBÏ1B¸¥bàö+×øÆ«x‹0ùÁˆûüîq˜ÄÀ¿Úÿ™¬iKö–m`¶n¿šÒ夤Œ‡x™` {ÃL;ϰC“<_ {¤!/Ä•c“ˆ¢øs¼€Â ìqä›]í0OCøbÎjšºÂ«-=hÑ“V‹¶0uŠòÀÉ­TøÃï--I©ÙÃ’ÜGÉŠ/ä/=ãâý =[5O]¬Çõ*ÊoØÂÊyµ—=/NÖ ¦µoAø³fØHâamSÚ¦Á]σ eKqoˆw³{­.(Ô±c›Ïä´Ÿ…íF‚Sw—í®ÕŽÝΦÃ9C@©êIŠz!ÔÛÚý9’o€d3cqµ&åõ|—»NÉÊbLзenhË<4×o-3Ò6Œ‘à/ჿ Ú}Sà£)ö3ÄTG-Ì柂¾·ub²ypút§+ìÄݬ{¹ÁU1.•)Â=çJÎ(¡¨pÇý]3Oߨj¼œ©1d¼¨žå¼v¥Ü´}ƒm§HëPsÍ™¡àÈw®ºê¹À-XúkØÎ|j¹U$©ú¤VëôŽ’¨ž×ÆÏVB ›Öt# ãLBQhNGyk«µ6eÖ²5½Ü£³9!ŒúÎWJd÷Y«õµXH¸uOß”€1çXv`—Ärûœê› ­„Vv®0Iun-¤4,nl3RQ·oPýÍû³ZÉœjÎ”ÂØ·°Û ÅÏ U~è¢4B:á™59JtùŽb3 sUa•åÛ-ÓfÝŠƒ/Òן×å›l9ó(¥Î.‡K=7}±:*Z›Ým\zé8·!jÍîÁ¢Ž¾È-¬˜ivt³ì">æX´Ø½ûÎh´ÖjGCUÿÙÉ(Ï1[*tèŒÝÀ‘ôÖ‹[zÁòãÀP—3ËâDBHW|Ú’&ÈžÁü__›•Óɉ9§jËÞþ©L¶{½Tvk±Ç‚0Ä´úX>àÓ]ïÌ€q‰¯µ† nBÈÈ^¢“ÏÓÓˆÇÌ•H‡]½§M¹p]Ÿº¼-9C}–q³Rù„/0îÏxŒ±4&o3ÜðƒÅÊÄnq{0ƒ¨bLìý Ö&a»ÕZºÒ`º}vÁ”Ì£‘±PÿxWHí°›Ga΀éæÈÞ9<*ßÃ.YŒ§ïc·ùç0 Ö¦M%›­ø ‚ð†ãH­Uò^ÕW(±JÉ!™`-EÁ¨‡Ú™ ÂÀ»KUQè£ØëÚÄÜ,NqÚw¿Œ¨0g[¤Ì4•EFÂk h•èHÿB%¿¾„9˜rÈ©Þ Šò·è‘C[ÃqŽPÍ®Eœ§Æ–üJz&cÛ®vOWì‰f’M| îš­Üì_'‹CI½úpé4DZµ¿ÑÆéÙâYÂî^¸'Sw”z=ŸR‰@án齕ì?aÇGœÓØO·tÑ*ýð*Sº4>:%P½ýAŸÛÏÊ¢Ÿ4€<F²5 ýo Ïi „¶ž9*‚¦##Þ*Ó{¦ol¹6¤ÒÃ]½‘¯¬DkÄkšü”⧆0“±7Zgv?EMÚ~Àøjê}Ù;lé3w•Å&S^‹'cY¹!0–^¾žUòxÈ«ä²Ðozã¥)Â)­þí,§îÏ)ì:Õë\ÆÅš˜‹Ï?ëÇ–•õ^àƒ®nnÃ-6퓨·FëT˜LY« dJPtXã˜çn˜}N.xü²]^ŠcJ—Hæ*Ó› /\O¨á!lE}üÿ±$Ìj„ÙéXÍüG¹®úÃsNà Øõãb'Äÿ¬ìÏd0ö­io¯ÊÔçAq˜tÝÄÃfMÜmÖç1I¨ÇŒ5ƒ¥ª’²³„9<ë/&ÕB‹3¤8äræl uÁ)Ò`v7%©céf·Þ7Äž{uIáµc7#ïË€©E)âw*¬­J1‘Ÿ;_‘/XÈÿ¥wþãË1B[©¥0™û\£|ËÁœ²Mé *Ïà?:—Ü[ŠrK¥÷TŒ”–‘ÂÙ·ºruEüPßmµEÀ{$x‘©Y¨™‰^¡+ÅU¦®06ÊÛv ¡&—ùí_“@â8‘Ó‰Uê³bíTŒõ™·×¿ÞÃùT Cd‘a¦qó=t,úX*æåê`8øž†š ±•=_½Õà7ÕžÂTh4¥G%a ½¸{GÞh´Ó{(¶¾§ÄVytŽ1—Kº”càТeHŽ^:©$ó3¥1n¶*í«¡ŠHÀàbš¸8ˆÓ!„sX¹¨;uÛ]˜ TKK°9¸nkE¦£;?›Õç_Y8b‘žâsl‚-»ˆ¹ƒ«Y˜ë{ _š£pi„}2µ¬æ1€Ú“— *eüó Mɺq4o©Ln²Õ2¹î%*Äß^¥Ýó5žiü[Zq‰TÚSš¬îÅ:%ó°^øŽíy|æã¬ööj?l{,*öx8{#m`i#ž¡­‹»áÉ›k6“Î*[²ÍwÁtÝ->#AÅá5qa™’3ß«WÑœ¨¬PTBV8ôZpÉ!l¥w[,—M”Î8ÆÜÅ!DϽԇà*ÿ0ŠqŒot¨üÁvRd?X[Ž*]s8®w^7Ydœtk9[ØéÆ‘3Ú´áC¸>£?†>L8 ÎF~ªî’vÿ€*ÀPǯ±ÉBBš--Ý쓺«œ‘³Šs4× ’*&€ŽÈƒviè‡fÕgNx²qaŽëá]{|Š=ú$òYýr¯Ø—xgã6ðegöSà7Ì” œ¥oí40<ðÉœí„Æ6Óì 4rÙpjÒay—óžBúßÐÙÓE w}qáµ8^äjÖHO6ÚLnYý°MÐé?^†Î¤T‹t#“c/CŸÙ¤1ؾU·ªBúEäåe4·ßž»E ¡íg#«1¸Hw€Àwã:¨2Âð°Ú\—W”AòÌ‚`6CÁíÇ ûረ$( tb5t¾Ì-djÃ'Z S]íþñÚò°â ÍŸ—‚Ê9¨zµ`Ó²n4÷úîÈúu²Œ³mœ¥ P2³U}¨Ü"·"Ä&ÖÆÂ/ –Ù^‘½;c«³ Åà!¨¼ †ÎÌlí–@²¤–£2Í7õÍf_.Vlù:Nf°YÅÛPr’¼óú7 žïôÁóã¨ÈœìÖ‰ðšFåLC_Ž|(ƒàCy*Μ4/ÌË n6ÖGY¼â‚v#qX¤|Bš²^<½FÈ­ø¢Cm.kqZÈb®È”ÚAîô¢Ç‡×ša¢Ì*ß'ÕY1Ñ?w¡øî{A¼“RŽ™Ñ›÷’¼¯Yk/XÊÍxÚ]R÷@Õä—*îþݳIöüÉ­´È* |r "Š~ÄÔ,ÌN‡‘uˆXÔ­ŽêùÏ[dßqÀ™’…µ•²jùÖ9.zþì¦(9âÿbÓâLªâe’›‰$×Ñg,âêB8v¸ÎèzjwK@ |üÁ¾A­L<©áo.ïK¯h pFâÛfµKY¹³œm˜ˆ¶jÓ´–g‰wR¾gß|ëžuÿ%²K‚Y1ë HµM¶2lPŽ˜®ÿØmqTŽª>ŒGèî¤êx†zeçA `B“q³·)eÍùüúêÑ€råв, LPÔ¾«òfn¤„ÙÃ;:v»ð{Z=ûHñ“4™âØCÖõ:!Ér„:9„ 訃«Øé®ñVµMyº¬‹ù·Ýç”U†Ñ$UI4Zójì1›dwÅ +dtº…ó ŒT‘½vvZ÷éØžX–¥XØ:!§L°&(ï½N ±Œ[jÖƒøª)ÛNù/YîÊî$`I‡×Á Ø@a Ê™úÒ]ßX™•Ì5Sâ¹²ª¤ Nú<&5%]6•·8Ð<_¹\ÖŸþ7;ÁÂ"D·eÀÆâÆÒJlƒVEãäU9UÓŒz*ú¿S0„úSN§³Š¸"r»Ûh5eh.m &úUÏá>Ò„éÝ/%4äÇÁÕ+A*®ÑpÍ‹Z±ACÛ¹Žô¦±Ö?ÞÊrSþ gýãÿ?ÕÑÆ—m,¿ç”îï­å»ãv¬í-ÜýÿJ@бû ?>|ÌÝèHúFÞ:®æ ä“p5zoÁme…V ­ÚÍsÉ”ÝþBÝ'¬Ó:fð³4oÎTÓõ¡¥ÌL¿ñ*äÑ„i§ÞÍÑý[wnè§‘°,ÊMŸê¡¤>N}s§Ÿ ¾.D[kÖö€‡l3PD^SRÈèÂa>´jâi¼#ºÞiڒ׫жÚo°b‰òœÐ›²ÄaÑ\"ëK5Y˜(ÚæL%ºZàu›ÎàÁ%å£Ë9ÞmÐ̰Íoz=Pʶ]—çðHL“äƒûwyÚá•£ÃÓ\¦t{—â8Ž_ØŠ÷»¡(F½mt¬s* êílj*}ù_ío°f÷¶ã£ÄF>ºyö¦x8-7Ýí·ÍÃuãÿe4˜‡w)×TN‹¯ö‡ñ%9[CÑX‰`»ùGÔ­C¸Œ‹M©dqôrR^ô¿D½ƒµÛHÿN«7‡"½LUcŸ¯ æræ²Sõ‚²¢5LJժ®%sÝåy«<õ£‡‰ÕÜil@R ¬S9ºl§ª~{º&ÿ/äK”… øÙSØY£Í‘µ¼»Ñé– µ0O‰ûâ1út«SµË¡õÖl2‚e5švLçq=/zªJгýÖs›`óUÚ¾€¹—©¢Û3×+\ZpÍõíñᎊ;>FàÕ†6Ïœ=FŠf ”³‡Ú…—¹sg4«É5!"Aüw„ïµ–‰­DÙ«ãÐuž²ÆÌ[DpacÿI×/ Q×bÛ}³rþ=Ûnjj?Ë[¿“z°®%|AÓÃÌçåömû6BÃ)ª%­­?ĺ¾ѱ‰=;EgEY;Fþˆ *P‚j—˜`òÛ:ý¸¹b–xÅŽ`TÓ¢”÷‹9¼áé{ãn£øœn0K¡(´I»£1z%v"ÇF‚îÀ¿¥™*ræ><ÒaqPè¤Ìʧw' ]ˆWj'~$W~Æ­ª™ЈiÙ€¢*ù¢ ebÅÿšØ¡7ðB¡Cæ;|íµ.ÉͦÆâYðX"«]’ ¤V)¦¬íïY»ûïÐú4óíÃr9âOsš¤Ö5!\=£Xj%E]å–©[a¸»Á´4nüshò`«ê@ù@Ó†>õ:1M皥Π¨HTót,áOzõ—/Zd"Rl!-Œ=4¡lä^×mrs‰+Iìt?ƒúK¶ìRŸâ³¦g±(qcí¢¸¥eð†ޝwx~½iŘ`ÐjÂq Îˆâ†õ.ûEÌw†M@mÍáŠ铹d®Jg":Î}Ý×£2þÀ ¢†>š"o—ȨŠ;VºeçÆÔ§JŒÑ†Ê…@Â"0¢(R( 7¼¬}K¨³ï`¢0-&g˜ê]ýTï½ìgA8æ.=!?ñù#j‰ˆÇÙsJ2È!ªðñâü3´Ò¥Y3vã_åéÏ[sP½5õY„Øý“™54"š.͇˜NNæZÃ/±Lý]±ŒQ°¤ßïyñ ¾ødôÉ(=¡v¥€fªÚ1Æ-wMb!,†Pbª‰Ñ*,{ìIQBÀʰ¥ïÃîò¨žfr’§ˆË]ÚS œ·ÍÜ…ôy(VŸR^x_…šXL2Ñ Ðzùt÷äÔh©t’±-g,*·Å¯»@$êPñ‘B,wqpÐ3CŽ>;|La•™/ô|fÌ–Y,òB•>öüúü\-€DÆûºc‹ œNK¸ýÖåèÁ¦Óm° "”çÖdWÁCI«Hn-÷»tQè—­cøA¸º££ªÌ©d;íy‹å«…0þõÊi€ V´…ɘÿÑmJc5ÅÆqSŒ•ÜR)âMÜ ®Jò󞌠…j*wiònW`š(D¿è6=cx¬L þR•†q¤$g’#(tèh$ÉäQ.T´d$8T®æ–µ%´8ÍiüPÍ5Ç@ðÅ)¨äÀÛ—\ÍîÔˆ ©qˆ,},G«|’\ÄÎ÷,¨ÞkXÈÕfß°2ì=]=dq”g*)ÙLòͦ?*ô9ÁºŽˆÞ]è_{%\·húáF%(-÷Ç=jœBžF8©êa¥ +%½F“‹šþ“Á`Fä'çXÝÒ “¾).ĬàíÿðñifÝjq(¤:[C ¿­ߦÊ#„ø˜~²bKpãŤ<ô—EÀ SF7¦MÛöŠïoá1—?ð hH5ÕºV™P¼…-¿TiںؖGâ/ð‰ÑªûÈŸM?Y ï§˜Ùx´Õ ±Ox5­E@K®~ÈGú#ÀÂ>Sñ*«¿%–¾T ÙÜX¤:2~ÛÇ0ý£WiRc2ýÂ3ݤÇ7@üKXtÔO2¢ŽÛ§G2lñõ‹9±[•m‹ÎÌ ÇE£ÁMÆhüŸ ¦=Ô¥³¬¸sFé¾ÂenŽJƽpHLb‰âF¢™aÊü„ó¿ÄOÝ8«€»‡¬^íèº ë¯p“‡Qq ¨îj5bðJ¼>®´‹ÏP¢-èš\ÿþñ (J·-ªÃÓà ¡pF¾/«¾c›á àææÆÝSß íeãÀªÍ¹¶s«¿í@•À7Žšè'‰Z<Æ©ÖŨ÷òôk Z¾V[näÏ–*ZO8D©h¦TÊmÒ^ZRÐÞ†[¸ŒæÔQþObÕnŸáqÙõ‚Ò97<ÜytR¸8ŒîzR7ø‰ç%ÐÖ^LtáIá“Wv+¤Æü§¾tñ&û—ÑwfýÕ™9Õn<ä̇ž°¡àâ1ÃÅaŒÖ&]5Ó£ºߤ‹ŽÜq¤)=~b btÀz½LAo÷eÒC¬#!/h޼˜W8=×õRKÃ`ÁÅj»NV–Ñm×c7WãBNšöå¹Uv}ñX¥£Ïïš´¤\l-÷ÔÑ­«|2 °úЩ êóŽH9ÙðÕýuó¬Zuäa¢„[ÜŒDGh2Ê& ¹ÎV0í9ÊÞ,¡xU†o'6ÙØÅâŽT2Ø6înA¾Øí&YrÓØ~݃ü!±öüŽDL¢Œ0þˆ‹oêëiH ÏãqPþÂ][X;°]0@‹VV—L~„.NJ`¸÷Œˆm×ùcz9Ÿ¿,©0.Ý?,†1ч@Ò†DTõZ'Vëò)cÐtwÕÞÆuíL¸=õ`—g!šñ€¯åÒôøVFT˜¯‰¾ÉU¨Zãíe¹°Í"?µ^©yˆ@tA2¥&XQv*tøLeêü [:× ']º%ÌiÜM6•ñì\*Ä»ïç¿EvÊÐ Ö8ntÌÈ1Â,G:ŠÈ7T‚ÏÜ… ÉPûs©Rå"ð¦Sû€Ñ!ç´úóÅNo:“Tj"‚éëÚneD$¢ŠŽ ”›Q| }«çÿ*0KZãcŸü“aëõ^y.ÒwsìŸìÛ] hcußðÇgÖ0J}“<žéAäP«©ÕÉY¦ß‰4ÅÀÍ¿ž—”[ߙӨZ"Yë‘ Ï¼~GL±ë7Zse|t0(ÅB\ãî9˜ØŒúìI4^3H™ØšÇ¶5Lã–xÑwÅ ´(焇\†Ju"1Z{ Ÿmyò¯т&ua@Óõ“`ÂËVw@é&°>\¢S’WdÁŸù}T&s춯Áµ˜×cZ¸Âÿy«©ÁÖÇÁX»LòÑqNÑGª³,=p2æ”åJê/‘­/¤¿÷—Óú÷=lÓº¿` 1á&‘Ýøˆ¶(é`„b²0É#O'ÿÀ̾ðJÎe4yo† î«³‰ú^мxЕµ÷08žvÉÊHö}Ôä ûëŒ\ZÌêòþÓà‚=‡› êäu ËÜr|ÂH(Ù>ë—ð]Hº•L+]l¿¿6’7d3y;ê9¬¶4ÅÉ»èvÇ¡ÚîÄ ÍÇ/Ï}ÈrãW!2?ÎJè(Ñg› UÜÏ|+Ÿ³l™ÈE‘+Pó\ظ[$…úŒ¹àïžwª»ª‘'â™ æ¼Ú¶¬"𨻴Ë+;5R¢…LD4ÛWÇÊäkõô:“£°}BõLy¤Â°žFlñ¯—¬C¾ˆ+·-qâìÚ,(ÊåF?Éõ‡¾™ÅDÊnúŽ zÓù$Š£ÃwD=¥È‚|OÑ80É þ‘Jª}/²‹M"½! æãUS56@°žÚ‚ñEÒ‚Yw8¦yÿ'Ås;H0Úît¸p)¡ÊMÔÔ°G6ðU¯½õ»ïT¼[†·Àà×4ëFC P,H-0iRâaéŸUÉ íiÕ9OlŒ›ùh5š¿~/‘yr¼ÌÀŠL»Í’¤TXZ-îQ8ˆåÆ ´iªÅ­až„‹wïgbI÷ŒUpU!W±V,øÃ­-˜–íÕxÝE§!ø'Þj癩Çs/†’X5…+õÐn4€AÙÎ|ÓÙe¾ÒxwäsH­ ±SÌ¿öî¦MTlŠO</Q³œæ ­ŽY5ØÜµT Txwà€wÕY¯“½ù˜š˜Ã}ú…¸UŒ‚U'ÀŒ¿ëF48%†UÿWúܰº DÚ–¢?ì¤öÖ¦å”ÈÆeJ¦Lvh@ÍûS{A6·?¹ Èÿ»ùo /©·ã£åû3üÔdÿ–¸wÓoAðfôk;ÛúÖäSÑì±ýSÔµÙX'†YnƒU9;èýò/k{S£Îc n¸ÏßtmßòH•ܺ/£¼lU÷žŒ›«{œÍLÍRó¸gù³&ñ’åêÁ’»Ž²©Ì?øÞð[HcbT!^/ÐŒp ”è[B YÆÞÝôÿÐÄûÂ¥H2ZÝ£u§Õ€Šß:Ð÷¬&ôø¬Ñµë«ŠÉ°ÎÞz''cB<MéÅÞJdòbƒæÕƒåEßÐö;³qæ»ëRU}ÓºXëÚÓ˜´æ‹RCôäv¯¬]„¿šÁåŠWdw3P"•E¡çÑ}³kQÇ9 ÇÞgŸ:œü¾ÂKNWôÃÖdÖ~µÉ“§û mmaxãÝ Ï;†ù•…êšVu´dÕQ@XMmxfÕ»‚fÖzblÛÎFRL¦À+QЙÍ3>˜:ÒËÒ?E¯â„XëÓôdö§ZÀEÅÙ0þ:!¹éâT¶!ý×Ðh”ZìC>!0T¸ Ä"aJú2OʼfæfÛ`žñ{¡ñS«9]ëê?—G ·\Ühh¹RЉ¯½bõEˆs‹þǓ٠i4÷®x~úá[æw? ß^AlÐx ¨=§_ÕCiáïAHÐuŠ>3eþGð Þ¦µ­ïÐm@§.¸R+ogð¬DCÍ‚v–«0®~hÄî6ÎMfkõÐW;ÆÁdž7¾[ŠîÄ ƒ›¢¡\ÐÆÌ6ì2y6ÁÝÑê=%r FXÑdpý#”òpÙªÏÉïɦmVÏ)«Âm êÐÿˆ«ð6Z$K~à Ü²Žœ>X×,Q3 áÓ'­½4Ò°»¼4×vCž z9ð.ÍÓÀ ÜX:Bï9œÚg9M'ˆßS\¨ŸraÅ¢HyTy´@Ÿ2ï œa³H{ÔxÁð‹€\{ä:íÏ ª»ª·‡}j>j÷ýôðRP¢)Mc·„¼)WÿV’«\,g'œ¢ÉØ#pQí¯Rq·Ý©ÑÎô-°‘–ÊÆãÞÒ\PÝ‘0}»“@êZüßKú¡ïÜÜü! ~>Hyµ¾?S wö én»³E³Óí.ªNE‹Îž€7^Ä÷œ`YEˆ]ÕgÙ­Oiü â\Ù'±Ñ\ EØw‚<"™rÞ¡JcÓ2“}ˆûqOšw4=ìÌV÷CìAÍÁÝ+¢±k×Ô­+Ì×"(†‚tVuËÐâ$J¯$ø¶¦¢þBzŒHê¿–(aAm>RjsœÏè‹mC¥;{Åuý…t"¾ÆÔøXzÌ[©ó_n¢ˆ`dÓ†KrA3×1 Ú15¢²Èß§Ë©[3oÅ´á´¬ð¥òO㢓yÕŠ’æ±ù¡.”lTî§E÷.Ù’èAõ:)>ø&I\“#¾Íjó©¼íAž„ •èÏ凰Sš>å»ðwg3À\¨m}{Ä!¿_‘/‹Ë\ƒççî"„„uŒ—JüM]²„\ÎF®UVU·4YFKsÌ¿XZ3)Xx^ù 2oÀ³’8c®Ç¨÷@m·Ú„jÁñlor¨už¾¿3ùp&.¬êY³wÆÚÕ‡vV¡ä×™C"Ÿ»îP±ÎÂòïÿ …Ìg+tXÂÌ–â- rBWÖB©æÇêU×6çÎZŽ$<¬ÆØÄ(7’‚ò ô=*ÇLô9i[ß¾ï~S—æì<šl\Õ]ˆ&š5,N è•k ÈÖ”BØè>ìlÅÛÍŒI»HàÏXô;Á¦ñ®+sVñ¦ ÅL~»ÐOS¿7¶ºhÂÞQÓFeuÊâ´JÎdîã®Ñ59LÒ?YÂ[v‚Í2o2 2'2øôóçºàÎUÔË»óWÀ—œÌâ^*S]û{ˆ˜Œ×!mwŸ†òöV;ß~ŒÿvÚ„‚á¸~èËfSõ®¾YÚ¸¶<¨]\ |æ)<ÜÕ† ‰þ¯ý Òë¬Ô)wu=JU§³þbU AóÒ ‘“‰¿L_Ó´wÕeë±e¢8}˜"Aíýžwäfdë=S´ñ¨)ÇLÕÖ[þC_º®¡`³ÚšËì|" ñhALM Ä-ùÇR§³Ñ**•™[oÊv’Ùì*òÓüwé±ï˜¸¥ KYl¦Ñi3($¨×ï½[¡ŽÀ–ª†4œü0ø¹C1ßl?vhº Ûq¹^r!¬ Bç¤X&ø*(ãšÞ«l˜|/;‹9šŽ&”>CªØŸL"¯Aù²ÇÎ`…o÷)(8Ç}Cò»± öDí[Ôb>„#¬Ç|Û ´œÑ赿,_ˆ6Z¿4$©_íâåÈÌnLt~êJ• ¡¾ÍßÍCJ ƒÐ ¦þ¯v_ƒ@ÐÎcÿ÷À:…‰HÍC…à¥!ã÷_‹ rd»œÞA˜rÃR?‹ {€Á~É(S³„¡² „qøjÎ0Ê®»2T­ƒàÕ:#žh¸#á6ÙËW+ÿ±2Ë<ØhÃ,&d¡£ð­T3N[ÙÚ™áûqQŒß‹ßP5œÀüm¹0Ñcœ?oçŠ/uŠÒ~ãÒ< ˪I\GT+WŽíbùû˜ÔÀisÀ:Ÿ©AdØɉÚ.µÿ–øÏIgT¥yÜ]ç*ÑO€ ·ÊÔÚjü 7îXç?ÙUXsµFXD≉€,ñ|®®áÙL°é.ìÆš0¥Èj¢ö&¢÷{¤aô”_:«É¸ê2^Ç-·hºn„‚š6‚äIöC†ëýN;ìNbæ&Ú>®s(J\"ÓÇhš>TEfš‹!7Î4&EzÈjf‡ç±VÀv÷úàmÆóy†ïëÞtY$²2§ôÁ[‚ZT;ÔY‹:ëHÛMƒ)hR¬Ÿ<ÏìcÈbç|Aä\…dà6a”Gû­>‚‹si°èô4ÏPÇ7 >̬÷ïî.esiÅÇp%tö[¶Þ=R “õ?+¾‘³é nŠ®+Ú¥?8“yˆœìR§ªù2×Ù³e`-vŸ¸oÀ¹s²‚Eu9óTznÿ°Þtxû5V\‡õ$îÛ š{?2­lªß@Ž–˜zk™îæj`—Òmý!ªRÊ…n DÈL±ÓkFq4_| eeqn^kÐÏÄ‘fLÿ^²ë…J`Ðé¿§x Šà·JùfÕ‰;³Ê'ÿ’ÖáUé¼·\—t|ÞÕRGþßeNÃJâ™à |ۗ΢ËòáLfXëÝ“9¨¥÷€àÞxÌáÐ5=Œ…ÔÒþîT i~´~¡òNÂä½× Eô{!%÷?€è0Ïà³¶Ú#Ÿ¼C甼ÌÂ~8&‹½à ®¨j]н«™é«ÐÂWDL;10©ØÑŒß9;ó›ŽØ‰TM Ù•ñŒ¹ñêæÙ4®x2‹ÿ)7¢¦,]'ëÚ"{FQ´C”Õâ17Ö¨lš‚ãÞG ¯4ÁbÉÆì4_‰†¦“ÈCV’ FÂ9 L0’üš%¢¨„»C‡ê0QðþÖå6Q¤Ä§$ÁÙ¼ƒŽ@ ¡5ïeòíã¨ÌzázhCi0yãj’mÁchá³RÊ j ½“Ùɱ­>ÆaêÙÞ1zÝJ¤Elˆ™¬†CŽÇ ñIkA¨p4t»+“œÞ?°¯3Ïb 'ÀI©¦Œ3æ}æê®Ùð.„Å4屯HÞ§³öÌÙA«ëšÜu…zÙ›ShbôÞ†TyR¦€€¨Y}ire¸‡ð­5(í{£5è‘Ú¨ ®…­ð´eøƒßÂŽŠ<9p§‹>e><¼+öŠž<âB†üéúàŽ€óÃ)E $“Ìt|^ŠY›qávkôžÂ‘êµçš¨¯‘z.‹ºlk<¶›–jWÖëƒÌÏU וÉ*t#¹‡0‰I([e›e±~úû¿=§¤FeÚê†;ˆQ/ÖkN8­eD2ÙÆâ¡}V”Áå’É&¹Ýú{öÂÛ~èÔ<¡–„?1M‡ÅH]§÷¢ë…ÂÁ/JÉrÖlÿ•MUµ)úÓzÌÃL—9õdÝ#hD×O¡Ù¨óV5)g%ÜÿNÓäOœw»ç÷ÓˆE˜âûnëR(šä$¼1%$êéð÷}@xm§RÝl〠?¤ƒ¨l"tøÆ†W{â‘q©;%Èã„ü,UûW›l¨ä¼•¾œ@åYYo£äò^…e óø¹6¯Å­6ÞÂ?¥Átš ä힌<ˆð¶v‡Qè⹩OƒÈ|ú0Úà`ÖÅ¢EЦåÕè¡‹ÙÔÃçõQÔ&ÃÌaµÍóWÓjRxGâR§oª"Uá”Ò»«jû4 `©ú`|–g—Kˆ·|Ea˜Q?sw‘0_±†%B:×¼¬¥´¢|>@É€”¨wTúféIl{g ÛE†|¬Dw;~ÌÐðÏFí•"—ò£è“|Vyä—QQnWŽ™ðúÌ:ÙÏU“ÐGb]fûb¨Ìæ†Zg;T–>=«{â4¹bç±%.åÜn‘æËºQ²‡>”ÌpâÉô`ôßóBxÚ†”5JM–œæÕ™zfþþχ°–¯ÄV z\ö!;~é *šr¥lô—Kãuþ¹îE+O¥^ G…mðN ¯ ”—¯©¦B:\’Ú¿Êlà†£};Éhtn…šJÅ`oÁ d™I`²a€—Á**µöÿ?xË jEeNö¢FZÕûp< µöÙ⥰óx`?Ÿúãõ69!ÔÚÖp“*üJüï]ªsÛ¯¯ªe‘ ›ÝÍâ¼r¸§¡EÙêù×ÛG5*ò ÷É­R~XurDDû‚$§—_igÒêÈ¢šô7ò ½Õj‚cÅ*a‰Ê%ˆ°MDüIî’‰¯{ Ý®ž®Þ yÅÇBžeÍ!GF~{,´óg“:òùSïÑž¯ £ó­Dއø!;V«$Á–øÌ˵|Vh“aOÍOyZI ît9¼KÈÂÑê?{OÀ¦RnpˆÓà8ÜVù zu;òq±„= ³þF,aÍNŸM¸p”?‚ŠÛçh×k!ßéEÕ –8Îð ©»¿ç1‡ÐóŸ‰ý¬‰ûBÓÃvÃtÕùâô.'‚×nãÛÓ—“ðþD¹áB)Sø”£’[eT+N¥0Ö®ô‚ïm°õ>m)ë³DžfÃQ;!RüÜ®«Í$Öx´ÕÇ(æ:ÇÚ'ë@®§GúƒÜÖ‰ Á-UºõŸ@…˜ *F„ç§ /®ÃႽãm.äNî|åbõGñoŠÍÄ-¸véÊ7) ˆ$ ……ÁúóæÔÿû(ÎEÌD O?,«wï½ÑÔð„±EÄAs0yÝXÐJ̶™ôµÀä/ŽE)ØMÕ°†Þ¥xPX™o“y%cçÜ÷°Y2·®¹Uš¬Xý{乄9¾—C†¹®B^[Àº3M‘åf\€ó™ßÔp,fSßµ\°4ƒßz ”‡é…7¢¿@+—ÁÉè«ßVsO”ÌÝ%Gñ£Ùsø"lÐÕuÂ5º¬›gÄ[4,Øý—åÒ\n™=Qj§Eλ„`¢UßPÄUðŸÏ5–uUœzarãÎò™ èslëä–¿âŽñÏ TGž±œöÞ=v&öA¼…,]ÊÑZ8J.ëî8èpZe[òn*Ä‹›L7 žS2ÊØç7¨syZ0%Kjx€hЕ¬§M%rе¥Y€ñùPóYñÝ$„zÅÎ÷ÓÏÀÄž0…t÷Ò&Åœ¢Ç—ýCñU P£Òeªx÷òsoíEsò·1ZÓ}ˆ²üÎëô ¹ï.ï÷¦n^‘š (àHØ1bq6‰mÚLmwÀɲà¯d”&î¦Û‘Ø„ÈÙüÃÌBDD¼Ber¨”{›DKm@àõ‡‚o&]¢Ü¦@§6k´àš¹†Þ»?Û>’ýT¿Â*[âNö¡|ÌËEó¤óÓ²Bq¥[Œßl–‚e÷Õ¸à¨àä§;§1wB%›«~ðhKŽ!1ú?7Úï­ÑHŽ_¼[Û+Rƒº%¢(ö›ª gAÿØ ¼»Å8jKõ/CÇG×u )ÍUé¼B5ÃeµîeîrXö½`i Z€5µ˜VøVÍÄYósâCAšÔ_MXWš›5òó§çD”L{këÔk“¢ÙÓ¿Õ£}­gUc ÔûÞ„Ö¥Èw\UϤ®4‚h,ÙÙ™ÉøY)üô!‘ a6LÜ1©F$a"Ò6´È¡Ôy ¢Jš6¾úL Ì<Ǧʌ­éàóâ©B Êoa¼˜æîNë~bôêoŠZi|Ÿý'3©ýik¨¦Í@;ììÂÓ{­À'ž"!Ã¥Yä/¾k¿è}À»BwBìZÒH«pSrIÞ*—7â‚›w²,ì<í÷‚ý,?®i°wydž wÒ .Y†Á0íPOîàÍË Ižº 5±¶)‹RÙÀZ+|ß v½?”=‰ç—†«ãÀ,g497Ÿ–`NyiŸôvçjGš"V<^!QA~?˜Äð Æ>Q,>0ÌÒE »JøS)Î¥4T};tqñt¬µ~Ý}«'è›§C̺mœÇù1Æ×$Ñ™o·*{cf,•º³<”­€RDç™JÛûŰìƧë³æ·yr9C57÷ãþ^á§%sŠo::[±¬™s§oÏ"¬¯Ï¡tJcFu›Ø‘ n!žU¢˜e/,¿:ÁÆa€y˜Á ‘K) c’^?`¯2­Š~_Û[{Þ—›Je10êTè²2tÖ(³ý~óÁØTð UŒžCò-7“CÀ–\÷åäô°$¯µalnmE'±Ê§ÍdÃŒF•Û«à8–rãøJŠÇí°”(í0aÜút dÕïDÇ¡¢{?Eu÷1¿égë@7Loq÷e‹—rµ½‚›[ⳈþðŠáÿ˜m—l.AÁçÔ uÞ5½s‰w¨ÞC%R'£àQ0‡”œùR'·G]¦”€øª6£6ùmÍ»hCÿwi^/=¼uù ?{ÑrÌ®å#ÜKnઠêV*ËôB8xè F|yÙæ‘OiŸfF¢`åRSzYx_BN±Ÿ ¶]HV,©!Ùã¨üÈò²Q­Àq™ýUo_±3E}ü‚¸U2¦ßqÂÝiXKï ʈ1êü7µ}ŒU‡*¯ìmžiS*žþ÷4é³à;u ¼A"|IÞ &°j6¿ÆX)„ \°í1¦*wO/–fž¢EV€–ÆVœ/›Ó™g;ë‚$‹GýP);l=³‚ÄgÄL4k«¨@ÆÐmfµÖ çVÌÿå9ùæÆQc ¾$TÅHž×½Mºd¹Âï#ÍÅ€³Dn¸û!I/Á&ÀñÕø:ÄÁ¨F™ÉuuîÆÖ¹®ç(›`J@›R7Ao bAјÍ#híçËÂe…Ýž]Rp2†$ÙñÒ@5"7=¨Át†ûž­¶¡©<öp!Yù6¨;š`°|”†»|Z•žÖRɹÄnë½Ú›t…9ò  j®Ha2M9sí¥ÀgÁgªÿù_ýS„îå×úÇ¥Åü-!uuÒê,XA–JUG4#g»D\œ*‘+8,g,w¶T¦êVÉžcÌ:ö#4¹LÃ~;UÖUÛèè(Ü\tÒ¥*!UÃJêò&êV§u en†ôÂŒD vR¾cÎìçäÈÌj@Ä™“Õ”‚öÕ…×åé5)5;mÕ^³¹Ç*¤By]¥NX5¿c ÷HQU!/§Éø`É“²{Q[ÒO‹jËwà÷Å@³NEƒ(ÎÔ9²>¼Îm¿º”,A-ŠAÚå§ £y˜‹I”óCúV|¦cTGãðý¾Éymÿ›Uo¬(ç Û†Òö|Âç¬#‘«›‹U‰8LQÇù~ ï‘'gÀ÷'ËRÝú1¥þ³ø¾Ib9'cçˆã|ÂIxV®&|«ë¦´®c¿…F*Ù”F ½ÿ6uýÃK±ÜÍõÆKÀ.äJ ˆ: üY'K=É®.~ÚÁÜ4ÿ°³ïÚYKþ€ýŽ çgQÌ^Ä%ÿ÷“€MkÏgŸH®›¿›Ç õ?ñ2ògfZâ Díñ$ÃÂЗàÙQ¹3«’B`Ú‘p1Í„œ-:» üéœø*ÛJ"|.*<Ƈۂãù´ñ˜_‘XÄyvñ2ïxr›‘ÒøþV›OëÊöð%D$¨ü÷1´.”eëÒ®”=¬%µ žåŽÁëá‘jú±ùr¢<½Ø©iqž|÷;$£O ÏyÕ»šPP\+V–«gþ4qe‹L⃔Ž\G÷Rö|ÇQê×m·ÔÙ=ì:'³lAãŸÈ‹Jy¤³õNùõ¬•ˆ· nH¼ç^ï%¸hÑÏTB‹ÅEET”ÿg¿¢b=4FÞr5Æ;¤àñÍ\Tw¼~h°ž==ëÞ¡Ìó>Ç Þ˜Ôµi ,BãØÏŸív 4çø-Fñæ æ‡»ô¥›•xò½_ÙvOŠgLl›0›Ï]Œ¯Õ’.ˆL61G‹T€šLŒ¢úÜX¿~" oê§EV‰9GœÌ^OAÀ,«¶p/롱øh´SЧn»|–Ž4š0#`ÖP󀓌5`áoTô{ÑL³—ŸÐ™]‡s؆3œªPÎ3ÿ‹HLï0 @n,3hÆ"®š‘n GU –RNâñu§ýý™ÞÊâJš¾È[E‚šÜ´&Ûá¦;¶ ú«(%‰Ø–ÅÔœ}¾®­MÖ«Œ|Ü”·›MîËXî߯Z&ø1@ÆzÛÜRÏT[!(¨M-\ék¾ö–Ïu¯Øi}²`ŸÃ+qµ›ÿ‹ñqr¢Æ,ë…ûjç<)•U ŠbŠq*þÓ“¡œg¤Û1¹ÍRÅ–×Vë£+e}Š¿ñè9ç àÌGÿ¡‹}kʼ¥*ÕU v0$ìíÿL—¤'–v~Ï+ĤÇÓí܆Ê~J`߯e‰Xr,Ób×=Ú— N!QUÖ´Y‚â&Ï¿}¢ ŸsÇF÷ç„Àý_‚x¥Õ:ç>$m’àzÕÍy¼‘Ðf˜-†ù5‡¡±úåÎn‘….¸w@ÿažÇÝS¬ÂeÜ™"ÝÁ1=âi[íHu=b%u0«j‚?ô?rqŒöžw9÷ŒÛM·rK»j 4K£+d9!ÒJÆáÀrR)`z¤`îXŒäÔZjfîGÜšm×ñ~–ÚÂ5Ðᄇȕ3|/¬µbá`à…I’å‘*2êÕˆ9ºŽS*ÀQ‘MŠˆô‰«ý #å¢ÉË»j-À2mGOý¸uèÓ«ªyĦÙôÏ!xXºˆ)«k)¾uÿ¿ÞÐûN?óPy’­½e …ÚæÕîOµ}AR}°þ(øèÔcDp6.cÎ9ýÄFnʹ–Ÿ?Ì;Hu!_’®Þ>JU±ŽÈ`±vjDMØ‚§«0NäëaÜÍC‡CqB sÏÀEóánk5ˆÀ³ÚFHµôÑlˆ׉ß8”š·ÃÌØÜÉ6««ƒ› à‹¿µ¸¢'×»Ü-M& Ow;ßñ˜u½cÂGn·æ½ä¥Ne'T@P`6ø|«]WÝ 0¢3 ,ÜË<‰§ý†V?)•w5©iÑ8Ëô8 gtÇ }2Áæð;!Êz0šKµŸ%6o€e[>‚5UzH3w8Äúú’õÁñ'×dx ªNÞaO׸Þý'ç X8Á]³ ˜ŽJzCX1ëÖ´ÿhñ¶°Ç–?ze£±XLjIv Ūv”ßJHæÅ³.ŒÒ•Û»ž☿.täø¸m¡m?º ˜@ÿVlË©èS ´~ägKU}úÂÀåïI:âÁ­B7R¶ÖV¶Ê_ÙÑ‚C‚øÌ¢R‚fÁ#0… ’Œ@ßæ³a+tõ«'z©ÑùmìÏ#ùùgÒ!ã °zw;~ÀɇµúËGÜ¡g}ûæwºUSàWúÇì*=óÜtjÕ°(‡Ë,6¬ LêTéG0!ÈçVkÞuV–œäV(þ.`OÜé+[ 6Ñ ŸÄèPPÐHŸø óGJ®Àt]0KõŒ•µy<¨R2ŽS­çIE¦x:ø¡/²¶^PÌÔL(pÕ6†Ìšÿ†ÓÜi!µûÖðþ\ JƒèË+X *¿A³v×1Ù `aÑå‹’`NÝÒV©mžì#äßAªµýˆ9²V›Ù?m{Ñ•ÁvÐ}eU/³´*Ð\×b¹|Ý…}Mõ4 §<ÆW%Pm8cׂi¦Gæã‘eÝæÜzâØùjÚkúB7ÿè·|F«@ž]¤½M:ÕÕ~+'à}í©`yÆ,8m¢Û:Òä‰)>=÷Äfô•ŽS§_nâ+^³CÍÕ’sÌÎh@ñ& E€Â²FÝQʶRâ¢AB£kÛ½ ˆ°Ä7„Œˆ˜YékHè|¿†c„aÚß¹¬ùÀoÏ#²‹$]CC¼Œ@A7àž\š¼u“¯A¬CLÍí‹I9\¡ õ£ÊòYùfدõƒ549gž'ú,Q$Åì?$ Ca‘1,‹Ñ%Gåsœ4R ³ÓÆùnÈ ^2]™Áw=3‹ÚÓ‡›¿²‚Í„“ž:µý+oõ¤tT…ž˜¿ §ÎPD7×ņ“€Ó†©Z««(aÞþÖ)sè *}mKq>Féû¨nÔæ¶’ÿ#Å›@ur–  #S±gJÕéÄ÷’:Msì;qÕ„q³ aêKfœï$=×çR™·ˆqîi1@?YTˆ{~Gšoc"û¿hƒþBÜH”óhÿ||-Àæ‹…=ˆ~m™¥*o›ÙTp|püʶ!%e8*òJˆ<{mËŒÊhŒ_M ^!–î¦ødo"ƒ'qFÓ$*äã7-„»p"Æ3 ìò3wL¹Ùo6Œ’8ˤ24(-Ÿ÷þø±'ÜÙÇ~/Ýw÷…vÈLµn›ÏœelyÇyxQAwA‹te<ð­l/ñQ÷¿Ò‹p¬Þ£êX¹´0bZ††fà&à %¤?-– bÛÃ~ËûòM“s²›s»D_|g;ÀžÇÕŸfhw½ƒ ƒ¸NÄSŽƒ‘Î’=I.nË aR*LŠ¼Ã¼Â𶉪uò‰¦P‹ÚØE·}ÊŽ‘Ñ6m•-O¾Ó7>ß!ÿcŒˆw·6á@9·)§«âíÈäýÛV ¬ƒ~ßâÑM{V²ñUº®n<ªê´¶ÕqÍã&[vå†Ü!láUêÛxæƒ)of7ŠA»ÈHÚhŸ|9¢æ‡Ëœ ”Q!ﵯC–bHŽcuŽ%6jà×”ôàﱊbˆ(Üjù?k6žªœó¯9xÜ á¨É›j¹¹…Ö+8!þÖ¥ P^¡R³‚hÆ «š ÂØ¡N2ö —‹"cËgj+¯JŠeÜ 5àùR®¹Õ•p1ù k>)õÚ1á…¾ ;HƒÎ [;«<"A î¯N0ýyŒZNƒâËþ­æÝ9‰†#¹ ¶šÄòòï\έqØ×'Uò­ºíG".@i/ÓëuÚ8Eê&bšµÎ¦‘hÝØgcµÿçv—±ÃÙk”|8†{Ã\!*ýZ˜¸úú`–Pxù#ÂécÉT©f!ÂU¥/AWcÝ@ÒpËv–ë»ckb4sKœ2¿ (šLoúþ$óÌEJ/9Ü%(ô,Ãà)¼+¤sK‘|‚sVÚƒ…2^ãâZü€F^ظC” è´´áì-ü“`ZjK4MsßY±[ƒã—ノÙÝ¡9²ÙÄ-¼i³iÉM¿gÐö†ª^ ù[ÇT·¾ö:²¦+%˜)¤'ÒOi8øp¨_Zèf£;¼äMiÙ‘nˆ¬ &¡q½qȱ ÍÔ¦˜9«u 5Ú›ä ¯ ´MMüy;@çþÊ3ÿˆZšõf`DQ“‰¸Í·©S .îI³°áЖ*e…¹rᎶvØ Jô ®‘œ‰öÁA`ï5*‡ìxhÄ2]<6è‘•›BÐÚ ãZUÀתªZ§#q%ÜÕG·¸9|J¨š9ÌÓŽ¾5kC‰+«!\ÀCØ&ˆ ÇàQ®äwZ‡¼À°GÆ~ì ‘ÕŸaô$œ#í2Ã<’ÚÖû5ÔÅ5(Ç £dv[,Ímg=°”èGlëüϱcß³ÓôôƶÀg "ÿb2Ä gª§ÉŠæñ)¢ ˜Sr2šyçàmšIB$Æ2òÇž37æË q“‡Rj »a§H _å¬Ì0C¥·õ®å±:Yã_¬xË™g ë¬È=^R;µ©|èj>\îŽAÝêçÙ–H‰#æ§6‚[Nšý ‹3ÖúÔNŸÒµìDTå š²MÐw{ :k§NK>E¼Læè¡°“ÔVô`²•Û«{A™YŸ}¬É›HÜ{'ÐL1ò]P½ˆ}FI ·›½»ð¯Çƒl¦Mo95FÖóH“yß×ï@>9˜®ÕPc®àµ)*¬Œþ‘‚‚~_?7\Ný ×íq>¥Üö¤‚×6‹¨ª/æãdzR†íf–—ÚLä¡%,Dß7zÝ_¾ ™©\\ Špt±Ÿx{¹[‚hux[T.>G//o±þ2']ùݸÐô½þú‰Y$~É s"ŸG¥&)¬Õê+ÉôüìŸ7ºHé‡qâX&lk£ö0ÈU(9¯k OÅ’õ"&A~åJ{Ñò¢D䎚Ôú¦[q¯alÌéï³}iSÊ£€”¢$UUœBæíX¨G X/¤ãès؃žÖ¾lŒv½Ã× ÆÝÕœ¦¡2Mù½oO$ð™¿A(27+¸- §DÍ_ Tݼ ÆQ“™¸ŒÅǧÖ.ð5]]U¯›¾OoP9UZºÖ ÄÜD™âºÀt ,ˆ}¸&«áf=/ÆKÆëÄÈT„AäÜ w#¬Ú)ö(÷ÇZ9ëëõfÚŠÿŽ3øšèçlfóåÛXÄFj§ò²É–åÙÍ.•¿¹“‡ºxަKlq?.¨Üçn훾J²¤'ä*…o»cŽ âzÕ×pCÉÖ“.#ÚD¸¨RíÀOÇû2Š-–ñKÍÍj³ô˜mèM¬<#¥x‚¶Fš`kv,€b#ö·gÜ#Š“ àú®W×™ë/‘í·?%ÈÓÚ¾&ðj3SiŽE0Ý»ædùþ{tÃDŒþû¹Ç¡9RòÚi Uþ@̓È!AÜãöŠÿ üwœm’ ˜"¤u€©*þTšPÛí5Ù›µÜûZªÚµÚîTCƒhju¶yÞqë½àGНJDq§bmì";  ”±›¤Ôm…R-7;e²³(inÄ<õ½œXp™zíP@;è‰/¤Ø,ü¤CáØ½âqÂ3ZS â~Šƒo ²!&ä*?bÇ®Ô{U¤ÕÕø&ŽW˜PÓš(ÞÍKòª4šS×ÉW!ƤS€) 0ÿ·Þ!K-tˆÊñh–mDÔ[†5ìu w:w¹ CS™Ôùíýº„ # Šªî€†NªŠ·å­ Ž3ÀìJF!3v•·(nÓ5ˆùŸ¨_·• 1¿ªOžØ HÝôbœ<Ý*1‹¨þ_»üÞž¹·ä“`ÊrVC +¼ißÀöÏ?BüÅJ[#^¾®¢$ãõã~ÙjÖQؾÔÀwÒ#èd©ä_‚DmƼPÁ‹GgÐõ§Ô¨í¤£Š|fki¼\çß&óä '—iæ?kÐ{ô^žR[ŽUÏ'Þüæh“oÈߺ-Ä–ù¿æ­4ˆÕà -s¬ŠfÏ7âÏšeê®ÉoW¯‹§ËyÄî^iâbÒV2ï‚N/ƒxfa‰ Ú¦BäEëKÞS‰C)Wùb¾3çsÕ>ð„Ÿ;·{:X wÁ¿.<Çh.ìºôz§X©»!ËôžkøoŠjÔs.c[m uÙpH"3Ö6A®€Vã–ÀŸKõl™‡ò‘0-p‹øÆpn´èϬ~5HXè €`õÍoâýyŸ*â\›‘íØSH ¡FÙbñå/îÐêŽ)`Ò^›´ynýh¡X4òõO† t;­œuÁÚ X$v÷ûíU< ¸šÞkíô•¤`„AÌœäæTg·ãg¢Ùü:Do^[LgIä·xveŸ¥8ž›‡î§Úφ俦ؽO!Fè—ÒˆqÙk2~ž>#KË©‚g~4•› è÷ÆÇJÏB çØn'8Âͯò37Ø™àÌ/å,,Ìb?lÒdÅ4²œxûáY_£)ω*²ÝfP`r~¢Éu;²º›ˆºÖSGYÚŸ$Í3MR¯ý¾›p_]êÞ¹ÚÿÆTólÒ¤wÏǧß}æ—ÑÒÚþô‘@$îó|Uè`ÕÿÙn:èOÌïöJÔÞ‡´KÀ'½ ÷1é–Àžï$NÇ‘úUè# (©eÉ:½.áò‘ãÏ=ú4Ȉ‡«s yÔL/¨"DÙu‚JêzÂìÀ`Í:VÇ|©è©ˆ©£ äí¼X ~>Ǽ•áõ­Ç›a’²ƒ¯ÝÊ'ß7 á:Tãb²w_Ýl´77 n÷¥é‘õžÕ94ŒÙà`zb;æö¹â°ÐùiÆõ%ÒG5ÿdæ†Üø®-«ÕŸ”‰‹DÎçµX#õ³<:v¥~&cýkš^†ä ŸmÕªvŠGgÙwhÐVŽ&^ƒ^dûa‘Öëß?âÜìÈËVíiØ$Ö ¹ˆ9RèDUDºþÉyš ÿáÞp¾^Í4®ÐRÙCSú*tEL¶£s(†Öd3T î¡ÞúlÃAQó=â„õNâ#Ö‹¦WìÉÌA3IFœ†úôQu`loJ,Í,[­Çá-ƒ’ÐiE°áÒ¤•³°©”kr/$ù…"Øk¨uN Á(É©d7% ”|k"hŸ±· )ÕD\¯tw<`ãX϶0‘Ðêã3¬) ºG (dkÖ,¥´œï™މ=¿Ó½7¤Àùu:Žã»;&/}#vãB%¨a;òŽû³âÞì™Gv9éRRÅÏ‹’:"õ–‹ Qe(Œ¥H”¹ø!,—tvŽ Sßaö*â»,öͱî’ÌÕÜ15µ¨¸L¶°{kýø\ÿëEHm¥•襒Vª¥{-‹ªhÎ&y?± hÖQuÀÇdµMÌH‹»)É& ‚QÚi¡fxmt€ÊJ¢eYÞCÊ¿¿‡6/³oÊ9 #N0n_kýõ=¥v\<«4ؘ~EÝ‹ŽB=‘ <©=©Sºñ AŽÐ¯ W¥AEnB¢®5´þ¤fg—ÿ¼iSIðºÓ¯/õ†Šþ0­m¿ '­ÑhŽçG)¢/¥Õ¬ã~‰*z­­ˆâGŽ«¬ïƺ¶¢ô¸µeàe¿ô´ÌåìkœÓï9Åš²ÿ‹í\£“ïs=‘º#‚«ï­Š 9UéDi”tp„E­Ð_Ú|ýìË_P@Ë`9·¥ ß>oâ\ë»Ö emXI½³O”% ¦7n/¨õ’ Qz#_HÄ@Œ€8x§SK{“ôS" ‚]M æa‚%)j€™8¤Wx;ý8q|¬çÕ¾•S8ÄCÝ€ËÖì|ÓÓâàÍ%ŽË1“1Ù Q¤ø+ÓÓÙæ—£øàX´ÝR€á-SƒÏ‰æ» Ꟍ¦» ^ QL½f÷'/’íÕKË÷ä¥QÇæUÆv‡¾¼’Üù™a4T†µÍ«ÊÕÿõW„ºí}äìz(µÂîýîlTR° >‚Viý]|PqgZm*ÐäR8¡ú6OI~›8»÷“‹P%@ÚD ‘%Ç|ßÞ‚ôFøº¼Šý|Ü\ÈËqcÄ•þícÑï*¾o4‡]¶Î‡]¦âqnˆ’;qFÉH¯ë=+ Z%Å4™QüUÁkº˜ÏFŒ åÁ¸²&Öú¿¬:;­U62=Sï)&tÿç Ypˆú/g}ø´[—ÃΗ§q‘ó>™G4 “5vO½3lº¬Tr+&”µ‰GŠ\:ÙŽCÓGÿnÔΖ=h¾‹ëWò.3Òe2çéì”o= c¶©Ÿ=sÀ”âüÛ¬§O[ùígøoäÇ‹’³ªÇàžÖz°ÙÝÍB;Øs?±˜™×ŠÙA¯™S›©Ë}´|ÁåÀO-(0¤¸Ý άWây…‹¢4r©óY†åÔn£cܺñlìy'4éŠ+@Î`’J@†rãu[:ùÌ8§,zG™ìcÒSï‚ÜFbèëå«–ýQÞh*à/ÈèõfЊ–ÄŸ'ÙxËý{jÍ´5–°jÒ#Qåóö3w¤Sg¥êd÷Žß³()â9)vY(TxÙ7ÿt{9ܘk†¤Hheгx É[<Þ<&³\3¶f‹'ӢŜ –C`z}à,0U¢8p¡£‡t‡ŸY¿Ï•§rÉ­T‘ˆ;¢A÷At,´â¾âìçp„¾Ç}¢F\ œƒTbX¡Tvjαêä„YÚËL|š\_)ìÒZ~á…_þƒõYÓ Uhá5 •ÑŽlÛ¤Vå¾ûþãðCصuøµduÇ9=âU¢*8E,(êQ'èÁ%YJ,qc‹¥ÒŒ›c1ó…ždÂ.uõôÿfW#˜~'Q<ªSód `ƒæVVK·ž^ØËP Óú9(SËn«–¦æ‚sL¾v·±…ÉVGå1öf_Kdü»²Ðପ%àÇáV‹äZ15ë=ÎÌ7Äõßk£Ð‡à§ìh¾ˆþ«%pÞ¬­Œí“óãf1Ù¿šÌ¿We" ›¬#Öã?¯vuÓ r»KHÁªü±äñŠÇ>õë¦ Íì2!é.k¼Þúy’É]dѧ¿ôz[ÅB7ý`}"Â[Ûß«K¼f€.¯•Ñ;Æ tC dšõŒp öBwñqð¶A»Ì–Ñ7=üåþ@ Ò¼½­ìÄ|—²âÑ<€—äéG’cÑ„˜4leåÊ’ç~šæ*¡¨õ$%h ßâ×é¯óCùT­“DÒ2–E‘Ô^—Q$$YÒ냌Æj‹rp¦q/Qg-{EèD¾j¦]ó@}UîÖ`4JÅVÖ˜ëíçkúRG­{æsK84G7 âE$’# …Sº M%h”¡G§µ¼G;ñkî•H·÷ìÄiQ»ùr.ÛiÒœƒN>—òÉ!Ãg*;|mAÛIâÍbe'ѳƒ×ñE"F>ÌDˆG‚.†{ê¶«‰ ÄÅöݶEfGØ­4•ÇV¼æÊ4׫Ä@Ê Œ3ãV0bz]½‹.}’éNŒh²¯Ã¤—ÀeEAœ&*”(Å¿‹ˆuñ ùIp¦:ƒ½çKkö[¸X›â.±üû| !Ì.lâSŸÂ¿ ª·ßÇ€çý¡M¡Üû½›ç Š‹ÛiŽm‰³íRÒ™1¹… Ûnòqî×ÐN ØÑÓ«¢{Q"÷àVÌc°6± t1ÞPžõíG2&8WgõAºV˜]„¼V{9bO]d›‹.4ʳòùe¥ •‹‚j^#q ù¬¡qÝlüºµÀû³ÁÜM¿ÜŠ^Ø”å©Å0ÿý *û‘'±}”ÞâÛzÌí‘¿2ëU¤Rš´äxëTÿægð ü¾†ÃD²BCKcÝ—æ¥þ„eâ¼Î÷ ›^WôßµÊäÛ´y±µTZŒ³ m·9ïŠZ“c#ÿ.ç,;ÝûBœªîÅ`&:}mÉÉ»¼Hs!©ªös@>AÜ3q.—!¹™¤$ ¥ØÆ>Du¾Œï•ZmüM ÷ VRžA.ÐSKiIh–.kwë¡FºKXOˆÿSÛˆ û˃|Òº2SMÃ3nî¿MKË”Fûcfo¯Æ][Ãù*Ôò÷¤þÓ}™l -‹xø¨NÉøÉhXØàb? ™´¶§x ®©*ê4-x@ÿŒq{àöCEÈNT~qÒ—òòšKœVâÉ„q<îZùÜM'Â82‡ZfòAöÓíÁ²«È®4²ê¼XŒÐAäñüjÙÔLB‘§î+õç3±+7åAèÓÓ§LSž»²õÕ Í[Zˆ`í9</ï!’V  ¡#_¬dZ$Agí› ͈õ`¸À!ñ°fbµ¥H p'>y†+£ cÓÛà?›z}ã€pyþ¥tÅǦh%nTÞß÷“¸ —$úã×dwÆ^ŽB_³‘¯W"‘ˆÐ?^Â^ŸÚò…~²aTž<ðõ ‰‰Ší©9¨P¦¯ëÉ#¡´‘îq9}¹å†6â›yôÃ.4^‰ÍGc¦iØ„ bOî)š(Ø*>ª3Í76«g¡+ú†OËXíÃ8«†8b©íW¨|÷!±çRZĨ®oØ|pSëAa9noÑúsÏÀçÝT1eHuˆÐý g2a´Pþ¯)HWŠŠ™ýVDÃÜÀ r ü`¬³‰p> oTN‰3 ã½Ìfö+Ò°ê=•ÜûÉúúäbÈ.ž/¡ãW_:ª’ m÷aÍH.LF8˜DtèËâUÀ¤Ø‹O[}%Ø•\«s T¿KnSdä†ç6ìa¥F4ìÎ}äÒfâ»»ôïI1ÿ:é‹ê´ç·º¨îl ¥þÚ£bó… íÒøöa $|{ø<×Юj_œJÌȯV¬‹¦NðÖªt(»ÓÍÖ¢ýÞ3“¤:#øèåiE=Í.³ým?µWµ°-ꆢ¹¡{Óú«AŽ]òÃã>.ó央鑿Ù%VÇŠÖ$³ñ Ž)©Br1‚žœ÷1ÿÏ'œ#ç0¯tG’¹Kæ³]î’K%ºa±|ÁÀþI äˆU¯_[m1_N£ºO«M9L‡¡þ?Êç'uAØñ˜5®2ÊTÓCOû·‹¸î§N^ІFZ´¨D 75ŸC«~ÑW~×#|:BqHajº€ ×cE|ÇÅÐ:S)USÆwÜn“ôf`:‹Z`6ÃV {/°±l;GC’é;1µc!f.nfÜKLç^G¤Õÿô¿ }°@¢ƒ~vÚN “§†R BŠOÓGÚÌôž¦v‚ˆžŸùN Hŵ OGp%cí=WNI¿“ûBî~Z)´¤=q ï1qdtå–‰õí2ÆkØÜM;<Ò»kš>+´ª^GÏK}%GuxÜî8þGÜU¼Ú¹=Ò_çs,C——˜N‡+¯ŽbSOÞÑÜzrHÈ¥ñà¬)$t¡&y±`çÄ™Ù&Kð|P» ˜%cýCgꯠÐê£gu±ý7ÆÌ"EÔ,P¢ý=»g‘C<Ÿ3³M£Ù+̆RÄæIÞþÞ ýÇè-’Õ*uêPF†›Ãݼž“AèÑËê$pã…–f±¸‚†ÙÁœÆ\¡°V4ÿO‘Á\†f…Ð6iÄÌUU!G)oê½cœ[è á7ö¡õøÎkjoæàËfËK{â­NF&ž]²!uy†1岓A«@÷ *2.M¥ÚZkë²C¶Ò夻ôâ <í¡šÖ;n»Q'£9¬R­JËߪ]úÅ& 3Þ§­TN?²&ñ±Ð¹ŸéñÀ¿=cÚ\zRU:Ï"¿ªÎZjÃý¶‹ ³9°Ð‰Ìè‹â›º˜r¥ë9_Ýàùºf'Ã5GÇOçï7°_¶S÷:~φ$ ²\O¡30ŠM2÷ÿÅ"@úÚ™·x€„+÷q2.—TsÉ]vý¿Â?\䵤\ã§/™ÂœÜßôp‰1¾9õ`ÎM¢fiAýbÃ?EIÏ¡FgDñÆÛ~Ëö‰]Ðç_º^u:ŒÙy;hŒÉ{ç*4ëã.ech¯[sl»\ÊÉä½4EaNɾ8›D†Ýv:ôýÊ!¾«=:Å=Tåx9ÿ£þŒÊS4Ú²_8e:³ªvì$(@UÂ}'¤ ñ-Løß&®vi»éfú)ØNêÊÄ2VeóË4>   Ö¸!øÔñéè~iÍܼáý`×Î UúÖ5L¬¸špp°"Éãùá¬A–u]QfDîÙ©Ü»kÎ4]Ìt% Ì»È„†éš8Ï®WªLu ¼¢®“„ò`6ºñO‡{Òû»jµ¥*Ù>Â|À<}žçŒæƒe„ÿÒ_ñ¢„Ýæõùõsb‰úúšºôIµ€Êfß?vGž0A•7‹ì´0–d(má6J~ÆqmÊñM[†À?M¨S+|E|Ãw9T‹d2r³i<ú%7¦/"ÑlÍû\?³bæmZŒ Ì™ã½v~øî†Qv B‡% î…¨ÎM-1޹¡…[¦Pº„ÃWl)ÆY+‡;î ÿÅaânœ=,ˆgåÑ<©ÚŠâ©(Ë4’ô{‚3•œäˆîÃnãeÅ©ybÖŽü^øßÞ Ó¬ôèדn—À"=Á?©Áëá<5ŸX¤%Ç-¾€.ºÊÊlÀ¼aÚê]ÛVÿI2N¶öxÕK‡Ïá¢#>•¯H°κ9}öñ3ÇrJU¦1øeíéætêkyn2j-@ pHÜÄ´Ï Æ(ÞKD“Z»†VI 0p_iøê=qÀk#,žä}Õ¦K0æ+DnÞXHsÂn©ùÀ“‘4Å©b·Þ¯ mr– qÓÂBœÖ p ü¥(—†yçœ{1¼˜£—t˜aÇšwþÅþp@!~vÃ!.i •8ÑyèIXÜ.cDM´–H'›S†b çó½('Lm·3“J½Œè¿u<Ӹ$ÝuøìEÏÎÂ’ _±º^òÈgM³ÆräIšÞÎRëœG¤Á=Uÿú\VN%?סµuq‹a`·GÁÕ>m]WNòƒ(^D*?–?+¥, ‘ 1š+†¸Èû~0TT9Lq¦:¼Næ!·ËöähõÕÆU—NÝUé……Ù 8J Éœ}¤ø‘Å@—†|öÚ©E–Ñöú¬`ýïo÷ªÉë@¡×¦¤ÿiBù¢Jnáz˜Â÷¥>S îóPËãc[gyq [½‰jS«àÙçHnMVC(šªI?{‹1´o Ž€ý´éNÂéÇÐç?]Ò3Lõ´@ó£#O–7v¹ÉއBþgßß™YÀ æz¦¬vÛÆc»;GƒÐO íá·Z7@çÓbû½ódshþkxxͰNJ‡áßž83í¸Ní;Õ1ý1 Œ ¸t$ä𻵌± +²¢,m£X6‹åPoað Éð•Z‚Â,¤l²\bÑ-eâ¿E†ÉÏ>ÇÒûÖØ+Š…hî’ð3C¹z— ®É,àÒ8¡gK, |ŒUs¤¥Êþµdª0×êIè$EhéfÉùçZu;íS³Â­sHè† ¯¥ß{æf"S›n€ÒHù”æ.’×ÉÃõ{=:ó­pæñùÅFˆ£æE3¶H¾ê.Ù<;Vh¿NøË ^3"ÞM°îƒáÁÀM,;^bÒåóXÅ'€Ó¿€9†È_¿0„áU°Ö .(c²à1.^% ËÄÍ,üTƒ¯„} 0›ÛÙÝ^¢òðÄDÜ7{¯?>6˜.C‰¦ý@ ¬!âð‚áì ÿWÛƒÒúÅ_˲¿Q²±K&ƒhûR`öJ~0¡¦–©¸x µ¾:Ôolß*Ö Ág†IÞ¸o•¨‚xmÞîÍ6ëÒ5,H½‘‚em7KvoÏ)Àƒî»•¹ýÔ{#?o|‚At:ÉeSw‡øÎÕ„ÏÚúÎýBAÛ¨;Ë4Û[i´Bt̘þcìléfMåz|E+uõø#ŽG -=‡šÝ²·Ø>‹Ì½¨ŠÈÈvá@~ïvjKp#¸•l(1gIýˆKÿoîÑÐó¸H_É×ë¥j)Ù"¯››·å<²—foþ¾®EZ5éǬÜiDW°ÀºýÂ%ĈlµâIš¸BScáßµjyÄŽ-FÕ/½Mü~݉•…•ãØ2Ö :PÕ#šwÃôÜ` “ddþáOìPÅŽ-|“W­Ð=­ÃŒÎð´M÷8ƒMTŽn8`üÝ+{¶so×eÑR i*Ö@e9¢Gw¨WQÇÔ†Õ}SÖS|ßC[dø²•¯Ã­×úünQ˜XÓ©²0ÛŒ³=c‘˜2kµ2GΔ)‰†¼luØ–îÁòÖ¯ {Ï`“€x¦ç<¢gyßÛ–Šó'͉^œZÖÃGKD [+w”dAü  „·Ü—È‹¤ƒnö/,{í?g<¬K²ì<¶Óv8.êœX!ŽI˵Ïw<Ò-Cå^ïã9›]Ö6 ™á5Ô#húA˜”Þ ƒ€¨ôÏGoR¹\f”®^SW°¼¯—Ù|áî5;w-A¥oÖ­ÂO:üWò&Źsü5¬çq\ÉÔ¾ª¿¬S€i‘^ï—’?úŽe.²0®¤rvQó€þ©„ꊛ”:ga œû-ƒ÷gb˜³A¸eù0Љ·2짘:ÇÝæâï½4Vlr°Œõ/â¥ð—-[NнíÞ—ò Ö( ¾VÆYßÃ÷›M^õåRu×ê(ZâDÅ÷ÿ·g|ª Ôj|Â厪tüûÆ‚Êáœ×çúóˆm Ø*Í¿}{_ÑþÔ_q 52}F£ñc"´¶=¼×Ò˜ø7g•þô¼ô)Àg—åÌŒ¹#}>¸F˜öï‚­'’ÒmÙˆ]’ˆ˜òGì”D¹aè‹·­xó—„"ÛÙN o Ð&€Á‹ï'¢”¸¦ÖßÏäößUñáä7B;LjÓ"ÎIqA†mN÷µmñŠü H ã÷€K tšiŒVO2ÑpIŠœSñ¦¥;n?s¬‹u¬zUÁ/îB‡A¢Ë‘½£ÅÏ«GJ²9†E; 1Ãør<<ø1ÕI7W]ò"ë ôšç>{ŠoM8cÉy±Ÿz¢’ÌB×è*+*ÉuM -&c1Uд£–¤¸Bzó¿È>+Z¹¢Ús4½ø·ï!fß!YsZƨ^=Å:BZ߈öÀ-Õ@± à©ýX:BÛq5‡F=iÒÝ}[ªg;§R2K«ÒYJmô5¿Œ\™Æð/7µ²3Å,-i”ø~®{‰\D Ï9êOµ²IRÇwtè^ÎÖFòpF™9 ˜÷ ¢¡ßwE?Ó„ “}v ü©™˜\¯)ýe'E!o6ÌÝëÍ(×´œ54(3ôUò?0?’_RkšŸ3¿<›C2È/yo~-t|êbÂVøÐ© ´õfþ‘É> ã‡p¸v?+;½@)6\ëò:»™”oeZj “¬‰Èfô¶³Î ëtù«ÒBòêºk¡a)Ÿ¨íx"ì8VÉhÚ}mê‡ gûÆ¡­´ÝS—Ѫ jCÀ˜± eTÝ G+%&"­wšâKVj‚Ê‚føw÷k/ÿœlŸ#þUD´hÕeÚþ¹Åšw2Ÿ™ÿ’+K>ñòâõk…h˜=ž3Ü@lÆËÍ'7;ÏÿÐÖ„ç²ÄÚÃÈceR#HL£‚&"Îñ#Hb ØÃÄSÉm‡³~»>¡½r=2lÑaÜêœi4ÁµÖõŸ¯Y¸ž[EaV‰g¾»ôýÝܪvp–Êʰ{¥žëSJu~*X?2 ·}»$Çè ûÊ«ôfw-­ªÏ?“g+jzá äö`ú x†¹¶ö4xV?Îxƒšu‘šÖ F ƯDš¿ÌÐð‰6¥óKÈÜÛZ•~~¼k…¿ª[P~éÓü»1+õaZàL‡"|Ónw#5be#mÖ+ívX×]¯‚“Œfhj) ¼»`õ»w3÷ÆQV4 )±s|Ž˜wFÁ×®#ú#âøÞù;‰k¤6‹èaÂð6 àï©ú´²Uãtîbüëgåä¹xA÷*|K›ŠfG¥¥¹ºú¯h)4Cl‚ôŽâ• TûûU—^­N±¡2ym“'°®…×v˜’ïÎÿâÏQ &z'À]ý¿h§Å9ZˆÍs“HŒ,È?¢eÓ÷b÷Êjrz a]Ðhp!‚Ý>i9 #bÇFÊàå:ù€bIÒ@[¸ô©Ôä\úŽ…UöõÊè¨Ãt ¬~=¯JdÇ*õëg:h¤Å‚Ûñö×LÐ Xq æ¤A¡“lLËO”ª3ã`šz–-©´E ýåÞ.4ÇD=w¶Áƒ½­»-8´ Pw™¡1@<ñº­}_¾V>/¯ÊHRÊš ¥¤Ü„¶?—:œ[›È¸â Íï³SïàvV~/ UãIOÏömìÄ?"ö/¿,O%•i~ê4‹[ÞfÏQÐö1`!ë3KsÉÛbR$¯¥iü>I¾¼iӆ薿z3nž¢tù§K7’?øö¹g±C‰qÐÙ£á|PÎÊükNµry¯¿–¼u3Ç,¿è­~lGÁ»¹jÒŽÂ¥€Õ“¶M¥ §õΨÊç.s½ÛN1ÕqK=C™Ýà5÷|E‡xPf÷6Ÿ€g5y§(Qy:;]Fcj“färùâˆY®„ è/Z· „íîÆ²g¡™F•4 å„+8 5æÕÙ#d dí/Úd”Ëô7D¬Ö»^·!2ý›–®Î„j(Fþi4Ñ‘0ã ÌSª>À“=•2ÍîÙ²CSrJWú|Tð‰mËû ¤.)á?ô@_ðO?¹µžGix÷Ú/"e—4ùô¢ÊÄ´So@RŒøc­kbNËH ™0Ï Ëf wôv-óý{±¦ð@ÌU$‚™.të’¤‡*<ƒªÀÖš}p*Ù×ß™¡TÏùc pÜ/eoÏX¸’Oiç0…—1%Á·ï@ /ÐÝ—9‡ÃK §JÙ a0W?òl¹mü¬¸I‰ì.\é3í°’.·@ƒç)nTêÆ2ËiõÙ²2çËñTc¯Ã2ÖW¢ÒZ58å.ßóµèŽXgˆå܉3œ;Íí‰൉‘”û3 hI‡veŠçFÃà®^!.‘å¯E]Z$«‚ÏÑø;5rJî%Z¨Ù;–"ã„ÿs”i5w üõüMÿÁP€{µý vqEœŠÕŽ?öÌlö#ý%š^w3¥K‰o-kDÕz31èm äÛû÷ÇéõHgj6%$.iíuÅ|…Y8œwÝ1ïü7­b4‘ŒÖ§”€W[È´{Åkñ¶Öv`A )ЫÙÑr=IÅH?¶CZVêºuÅ{Ïf®ím®b³‡ ‘fxæ%¬ÊAoÙ¾8ÞÎO`…ÿGæ ŒÑÕB7t=à äŒ‚ܨ8Ú ¼‘O‡nr[ø`ÁâñÆ2fVGóÝ&¶¦&½)òŽjráñ­Q€ÜØQȉÍèXÐ=Õ#°>¢ðŸÝ•P­´š±ñ¯ÈeÆÚ©0¢ ‰L"^lö»ß·p’×ÙñíBc,êüÁnEy¨™ Þ$Iø÷{‚¹1ÛM7ÛÛIsh ò>¢ù³;1s{YÆúÙíi³ä†¨ò&Ö=²?8ÛÍâÛË„[Ä™1K°è:M^‹þ÷ŠÆ¸ Zˆ ’!VÄqJ'&iFÿ.œÂ~—x=òÕ¥ƒî­]^½Ö»}3`Ê8ûã1"˜âgt ?Ú5ÆÅu^êIÛí’Çš@ºÄïJÞCdàÃPyÓIŒw2â§`iˆ³\¸)dúé¸yÚ×燼P_»¨+Þ<ÓÀ@.oE´ßzûˆýàhVŒ¦ìÓ¥%¤Óž|œÍôù:ŒY…¢sÖYýC&ãžr1×BE%>Û:ìë?Oà'ø/9¾hä„äap¹ÆF×@©æƒÇñ^d†+‚|¿gÂm?v‹sÁX.õ  8³ck5w¼nÙ)¤5 §¸X°ÁÒíÇ«IL§Ú`éÌ‘'$©ïòT¿Å¨&d«kΫ ö©bö#Í~ÛŠx.†[SZ]šøÓu«ÇàÕÖIŠjé‚`Ðx‘¨KeñF½ ¾”ÇP1‰,3CZ1[òê’\§hO'ηь’As˜[E4 ¸ªƒyÚʇÄquåäj´+û9’òÞ¡yo¶á/1ñVš20{:9Qcøeâ&ŠãÖ‘ –c±Õ‡‰‹eÞ!£…!älåv·ÌuI?àÓ‡L–ä:dð­ÐXâ6£}"”Ê[µd£pÿdÛ F-ÿfp›£¶‘VüL\Éì玢méÈŽÃ/Óa½±‚X»¯½µrŠÕV“óhïO¦äÚ$ÕòGÿäVÂRÃÈ…ª—¾Þ’ÆÁTÒõâS6E¤Mícú}"b‚Ë‘&Qpp¾¶¯†y[з Ú$ ªG”Êȸ²ðHðN‘iìƒ ÉÞ1·â@Á×÷J÷S ¯D#û5Q,Í70$öÇ0¡DmEÈ"d¹¥…ÅÄr‡‰Êh^Οç""}tЦ®GÊ™?è¸ H{üfËž ùÿm05𑤔&l°Å ÒvwK4*d‘„lx§v,¹ešm§J¾af¾–•HÍßæˆÔ›ŒâRpFB!ÌY?Å`ÜÜÔÚ#*÷,=pS`U®…b.OÿÛv Cž¤‡69yIÁÄ—}žuûÜ2ñ³`_vQ{†1ë$@¯åkÔíÌ´F+.§9æŒ[X cÕ„7DÃsôSî™Rk©o-K¡õÉV_XU2ŠEŸî*º’EžÂ)˨:\ùÎy±_e_|¼½ÖñÄSƒjë U‡2ðsqSë_üeò®ˆ¸?©Äô±RÝÓU›“‰\ð]-‚hg˜»13k“.“o´|¸Õ¾›œíåM²™x1•ÏÂúš—&53ŠÏ9ø ŸB—J±^!Nóà¢@1“ï<ï²DG£Ã89ÒwÞµ:û§°·`iÞyÀ¾ÙY)ê©·›¯` À6r¥ÍõŠB™ƒÞ œYj×GicUó¯_ê;K2¿«ðíyöé5ÐXl#¯61hOW^P®„à1ƒÅ£SÒ9ËHI0âÄe­>œ¸Ä/v3\ ‘—E­E²vœ‡þ=Iþ ¡¹×ñ÷¤VíÉ_:ߣ]ûÚE™þë“(ÉÌC)û¹%÷¾¥:ŽüøÕcE1ãVA¸zÈVgO¡ÖãvOsr«çŸËl uu‚p¯+Ñ0."tŒ±&ò»À”†Ü{ZYÏUÔVécŽ•4M쑎¾šåB¥51âe¥à dQŸLÈ_³åü gÉǸB)"…’a5‰ŸJ¶÷3ûý9ó _FƖ΀ë“'弃+*ð"ø•rhyqX¼‰x¼·la¯  ¬SÞîÔK­œMûŽÌ9ÀpR·-Ç.‹5 {QÙH­Ñjñu£¦$ìm;+ñ÷ãiß-+C_.ápWÀ±¼]áwÒ"x¿¦­{yà+m-Öòbo>¦¿‡z!ú›Áõv_×þ †Àµl|º 6ìP©¸ÍÇÇþø„¡\ÞªA¥søê{ ùóí7‘Ö$3dúw ÊÎÎtUíׂóÇwúz6°‚G[¾cÓ×᪀1l1 ûÈ5Ï'ÙúâZÁÚÕ¾:î*–`41ÕÖþø8¯A˜Žü#ß²6Œà©>å6ŒêŸ/£Vñ[àÂÛ IÊ%ª Ò€Ü5Xü"¥ûx:$·äÙ> ¼¨ýœéÉâ Wë‹gµÁè"ÌTvЛG*pÆëfqãK¶Ä„˜€¥ W«tÄíÕëÆgvÆ$뮹èEúʓκ&ÃûÅ}ÙnzakBkÈÜ.ñ³Z§ü™}ÒÐ.Q1P4Hx¬3àp«¨( ¼Å£JhÀ?¬ËúLÑÓÈ¡à<ƒIó›±KAþRN.¹ý[ͼ«+G|”¥·`+TO[›äS?h ßè°mn#núý„®¡êG«å˜öô€Ñ—V)=sɼós©u#ýà;®r°µ„f5&töËÒ€å¶Þ¹Àr—¨“¨ËèYnð)5mžì-=&£Ç™UNü¨]rmèé†j ¢‰ÅøíuÖ.­·lí—ç»êÆ">¼Óü‡ß»®î S ùN[òE²Á?_nz~:·Ðî ˜Û‚GhpyÜ1ª®—Ö¨Yö‹ÞŠ“N«ä)X:AšZ *ãƒ[,\+“†±Ù±VN?Ž{Åsþ 2 –[‹±y³·Ñžî¶I«)þ]=# ößè1ÜÀqGH'V)—Eö›³Îöû®“¤BÓ_B.¬€ê®ù¡7çIPXQ¢n–â6:Qw&¿'°dv8fñwÀ?îd“޹ŒÈk­(-ªm®–ÞŠõS1ðÝO^£™l[Éj5(üçDH;jÔ Ð%ç(–Uïm 2L){ÕûëH˽d<Ó\%V·;Fñ°G7ˆ÷Ž8Œ–ì½Ëxα¤5ðßùÆ,⡽ýP;f ¼Oò̆d«f$¿ï7’9ü 2axùÛgëXgSý?¶´ŒÿCðO̘_0YíBLí^få<Ö÷° š&´ðXÛ õµ.˜m•bñOÕ.â«s¹ÀaT×fê?˧ÝpÍ)ms?‘’ÝÚøȯ{ñ¾xR¼=ãÓš®Ð^ÛZÎIX‘Ù(’õèY¾W4nû.Ù‡Ñ:RÛ9î™]väí—àÄŽÛâeIÐío$üÔ—Yå½€µ„^B/TùÍž9‚¢úvHjžlÎ*ØðÚµ ¬”»“úêØC‡t†`šb¿K½DHÃ]øíaÿŸ XÖþ¶³e$!Y„£žÜì¶!ØñÜ?äTö øJ{êîÏE%‰‰Æ¥kOšôÆ@ Bƒdb¼å!#:žºÎMÒÏ*—àëû³¥›Ó†Â™è n‰aAK¶Œç”>ÁÙ“dUÛ¡‰9‚ý)+ñ [Ó?ùo‡MÀR¡'W\[L“æ#Yžõ€ï#ñ"Œºç5ô‘xÎéÒðn}Þ+CógéÙŒUÅ"Ó1ä/G櫬"sµÞ‚jbº½s`^¿K½<³AU.Às»Ä»NjK9™Xè@±ásœð™|`x£Š¯„øŸm{¾ÉÎ2#Ý^ˆŸlLíVœèj¿s¯¿8ȬĂ1ÂõF¦Cp_öÍôМ^¾±À‚·Ae’)þ§-¯iJî’Ê MÿîýK쥉~Ùš¹Â‹ˆÜÂÈUÿd´LX4ü5³óväh>Ðøéu÷†–¯`:ùqÓ—µ‰ü‡!B‰fUdj²¶+„@@‚´MípþîZ!Þ¢ämç%U>"Ñ8ÜE¹ûXKÃ: öé:Ž_XZó”Bi5®7ŽÄ³’X ж…×+Çó\šÑ¤Rû ™µÏýÀ˜Ñmð‹4½Ua‚xO.üÓDFåaLÃkéµâ¤É­mn¡ÔhÇÒ¤B^ŒØ®ÜÿØL:(¹QÖ‚Pa®Gì#ÿEĤ¸Z’ˆ­»´ø1«­»Ø´µ²É¬ßyK×^¸fY$ou‘¯DA&Úy—ô­<Æ$AeV¸Í«=(~hƒK)ÒP™¦VøJdC ×Ñf3ÄÉɇY)»· ÌðLBÞ:ˤtÑò–þ´ÏsG œ-²ã©†Ô\­ˆJ­5óЮž"Yì-Kh \@—uá#ÂOä–>>ÞˆÔû-JÓŸ#“=‚˜çܦ€Ä—†üíN¥H”zQŠº+Xž/샶šCñ²x*Â9´wZ|ÊYJï›ÜýR«[A“Žºm£Tí±]ëÅ~ò¿!øad±x„ºÑ¡©‹A¥ˆ«Þ þ›8Ž6Ï«~ ±ª(’m¥¾ihÈ)Êþ$;àÒÂ5·l9Z¥Ò&aƒ{ô²| HØ'o[³ÓAPÎwÈRûîhóëFÞÇ¿F#]È+s©7VØZD»xÞÅ^¬Wd8'®Ù cOb; “”ͰÔz ÕYÚ›aÓšÊ@KM¤nü^ƒ0sª¾Q>8>Ó½>cJí.¢–lïÚ¦›j´6Ž—6M±‰T¢+eÌ–àÈ|]+H[fûG`Äý*@‹}ØØê5mœ⌫¤é\×>Ðú„K+º?Ó¢ÿÔiC³‘;“p×lógí#¦š^*‚Œï¸[Uˆv¢6ók M€xmñ9jvžycçõÐ~cu´T² C%äÐt7œÌ;¢ 15£= ó_~AÄ?$‚*áΉ+UŠûhFÝä¨èLÉví$h}~ ‚–.NÒ4ª‚ ‹R¬ÜØQ©Ø2f#OIäu÷o:Óíº úvÜ®gáñµd|h±-;b7VÔ ‚2掫«áUÏä;ÜÊ$æÚãæÎú7LÕøpž÷W¾¸æŸùTƒA¾VèhÄ&„Ç(ƒéÜ]g@Ó§±Ù°s „“Î'–ðTeqÎÚžCgšç‡ÞîÝŒ ¿ï Nd^–ˆ:·6Ã@[DG¶÷UHØÓqH¸øÇÄjCê¼kT;MÞ~Ƴ%< 9ìÀ²*dùp/vq쥌måo½«ÿ êÏÎ3D,*ï'“mÍxF HPÒZº×Ç­Iq½cOÍÓ®¹‚*Ê_˩ߵ_0kA>Ô·E…b8›ú˜&H̸¤9h”†¶œ“>égÄwQfB<=CA9ð ië®Æ7ûOs`R++ßOãhI¸s¥— S4O Ý´?ÁA\m="E‰e™(Ê“¸*>4nZnŠPÒ{Do‚¸Qé­³zËHb3‹q.ÇÄ7Ü4·Aȃ¨bžƒm2ס®hاFÞm.³t “ÚéÜP]F'c# ¼"Ùxòôq‚¹•é ¶::tP²T¹|ÀR"n­Ø…4ׂ/fÔo죊_5FþˆðKcçµÞœÀú÷Œ¼uj“Å‘ì‹ bñÈ‚ùñlœ–üJŽNLÞ-n5Üž&œÄ§ɤ¹nyþX‰z_Q9ÉÞ³œchš×n5wgB>ÈIè|4pØ%ºfCŸ>@%~‰ YWYÿ¿Õ_urgB·©=©ÝÌd—nù,ºž>[¬lÀ½Àœá AdvË"OÙ¨¦ø:wߨuv@dÆ¢;õgT\¯$g륲²7íà èN®´Æ©Aô“CËšDÞsi³%WÁ » š¶žÉ‰T°›f·ºx¡YåpVw´ß™ËŸq\é•e&™v¥Î WW¥éõ6ô.ªr‘¯Cø¶BŠ`R\{~cß“Že;þMYõŽÕ#ÅÂäa |>2?ÅGK<Øþö‰EÁ3…0uk ^Yùå&vPò½/:÷¬×Å]|óœŽ#>ôVÂ/Î ,´õ|åÉÐ;2!Ð׉‹­V(H0³¼oµ¬'`âaOßd‘ä‰Gn#Æ 1ÐÒ¼à¬Ìq{›–)öN^ )?¸¸V·Ç!ã—[¤sNŸZ5è&WOW?°fwýsÁ"òó ©jyY ‘ÐÖïþX7Œ®ãIJôêÐDôùp%MÈt±ƒoÏ÷§æº¹Yd%›‰æ\èãò€{Œ÷"W²ôS@µ NÇÿ·Hïòt¾Õààúk%<;#Ô•è¶‘EŠoÓ@­Ðê]°y/SRÚµäÉ1 MÐÈ`AÍÑJø´’‘ëʤ´Åvhn‰*­¤ë%ùğ‡Ã­¯ çÏ]† 5®µF¾8íÏpÜ©à\ÝzOä3‰æëêß@c9}·«óz €Å Ù÷u$y¤[ûª2©]àGâ« yfp`)tÚÑ,èói7gjþ!ˆ6]˜~ݱ³WêÄ}ÔR€j6 Ìά,*o£× 97]”øýRӾѵºÀáB¥¯9Aoozg{‚´Á>¤ 7³teTz›!±Æ×ÎBû:w)sŠ2>«Ó3šÑ>Éü! l[jâ‰éw2‡2âq0×{©8ø”zß=Oæ­ßÆèߢKƒðâ_ Òš3#ßGa.¥½$ngþADCAtæ2ƒèeõHu]mF›'Ýœ³$ë¡qÞý$¡Í»ZDLT¦Éˆ›±êzš`½ò[V…Lõ)Ow)“Kè$}yêÒâ[MŒÛ*Е{#YÔ¶˜¨y8x.¿!H¡yY!Iè©ÎSMßáÞW eÚ0d“!¹æ‘uFÈn«Ø J„˜PþÙå÷µ˜ì`+yIELµøÇjçoA¢Ö³• `qåñ N!´YÌØ/Gàô&¨8q*d&:E®ûI @\h"ßñpXÅmk-T}Iç1ãUM1…ûÙJ¦ÌI¦gÎ ë·øTõÙ‰ùc†Îpvšô‘Žù„-ø7.ÏØŒ¤Ä#LuýëOŠ>…;ÃÂaÉŽÏUÂKrn5’0 ÈŠ`ÅÓÌÜ–´g4©s5ai‘ZöböYàµÑâ,O¬Ûu´¦éûnðH€d’­ÒŸK@äDòJïž™S ŠÎuŽ˜’MY10 ÊRìE¸ƒ­7,oJW€Ü>ÛI"Ö&Ò%$Uö–]½=¯;ËîZ÷÷€ˆÈL톳bªð¾æN½ˆU>™þÍÉém2nN>=›* àõhî.Ç#ª)˜Ì”‰BÐûøÓâÒ÷¼âùË‹öu Õôî¾S¯EâòFü@ø²7zdeÞ2ˆ’C½÷:FLï$[ ìÒ>–›ñ,Óz7s›F©ÊØÀUÁt3~5{bx®u¿ç­¢L£‚pe[íZg% ™S$//Äh'PŒ«óáÇ£Nò›ƒn²ÜQ™9lis?Á=4žWýÙG©tàù— NÜbžïGB+Jýß6«ÇŒŒXÎËôÕ±åM_NQF¥õÇç«€K-¹Du <À·Ë`“õM1ÿ¾*Ó1ÑT*ãúÄí¤"UðE ö#´GMé(“Ê´qVïöaí?Ã#Þ\·gd°È^¢TÑ×jºçÿxÕnõwš—^Z²£ª²QJyË<ÔqV»tÓ²yýÕÔÈÅ@œù2 E¡{Éây¡1nKutFæöŒ”(/ªÿIqphOéUÁ¼3Õzž3—æÐMÀKŒdÞOóÅX×gDLÔ.`’âÈ Qa^T*4DÃhqb-°B‘0Á+šÿER#Ôê°,nH`K—aƒ„ª>qáš ¬ÔG&Ò…aá™#qc€cïyÎôYKÇîn±N2Nïl¯éJbOp½o/j¸À2ŸµíHµ²$3ØéžpEWºñ$ú~ŠE ë“̨'zK~ô]D:º=Z* |T²úœSi   ©žA³KÅÊ üËNõç,ÁçvÎà,+SlQÇ ÆÆÊ]€w Ö}˜º©o%|VINzŠ7^Ì¢'W“_Ö>ˆ€Ðò©†Ž.ѧx´ülâö[ji‹óosnÆŒ¶XMë§›Ý,85<ÊtÚÞÀÞv6ñ[:éQÍär˜,ç2rú‡ pºýgŽäÆšòr£Ä„ 8*Etq¹¿cý¶Ð¨ê³Û¥Fz¼à6CŸ{&Ó¨QdÄh{Š žÝ3š Á øT.kfýH™‡{„ùìØD x™¯Òïû‹–»'Œ‚`aoû'_‹’í3¹j·‘÷,§wûŽÎܬ?8ÂÂRl·c§–U õéÏñ>Of0Ãf9sÏ…h±ŽÇÙ´?ÔÜ7ç€æž6˜­Û7½ZŠ.[ÎëfÚèáž+÷Á‡åWa¶Ä:\¸gÒs+¥©U3¨§f($62èl_Äušo|þnÁcCœ•)ë‹ 3É9ý¯œ·²b¯­¯õp‹ž»åQ3ù \7“è|ôO)³1’W½P.q_Vu§ …T‰ göÍJ\\ä>ôæ~ñ)DYŠi åß…ÿÛ&!“6ÌtàÞHæ Èý´ßfƒ¿E3_&6Ș—Ò ­ò2üQù‰âÀl·±0Fì¾ð yÀùsM¾Ž®.Ô募°PÄÆëÈ|4*É–¼g9æØ*†=büÝ<w±;œHñFHŒÂ„}Øþ0}Èv¡ôUõvO"`Ê’EEö :²ˆË³MGuK„upaðR†M^ËÖ­ÒšÅúD-b?âSê/aY›,rass4A„méPqSFkýEòįwRCά:¥^†¶…2¯»òzÎ /–¬Äß/pa|¿Aö@Ó °ˆ*'˜¤ìÑ1DB©]º2ðˆžwû>-Ís}:õЬ#!1™½-÷. *}–{ô·{ê½væþ†`@‚hæHüõÞTzEnýRUÀêÁ¥be à„ò@¦µÉdòA°;>¥Fmz çµEŸù¹Du[˜k0úW\Ó:ˆí·Rtœ£ÐUˆq ú2׊9¨ÂçkyJ¾è¢ÔÆÛ±Š-}L¯«9¥ï…ž63iØG+ŒÈq½Ö€x2x¹9Ùi¥ð˜‘îª :–¢Pò)Š#`°6hꉳy¢j«éC-ò¨ å¡÷ßq¹è ÿhTJâ¤Ù·@íx+³w®äÝ=ˆ Mx4G94âBOóèåà»+¿ ²œ7O„ÝöyïD*ŽòóäÀ/ Õ­¿ϰ†ƒãÓxM,”±yë'Ý[I‹·õ¨’iMМ|µ]øÿìd@óûp…-§¦åHÿ3ô/î Å߈Hqc€IÐÝ5; xi¸õ†¾÷ŒåsÆù‘­|#vÅùL€VèzÂÐ#Is UxEK pr˜†^?‚ÃàmÁTy–øR!¥·w,Ø^z´l‡z ¹2©Ìxe ± p¹ õyaÖ`IHlk,äY—¯=ÃWq§³{VÃj˲=)Oº@‡¿€&›òbŒèy´á”ÀV1‰¿–‹˜¾J~bùòÕÍ5)m3Çsú t vTw1ߌÐjDd½µè¼´åwÄóéÔ ÿõ°»îD ìGÐUtYÏÑܳ¤<ƒÛùf5×㔵ù@ÄŠöÀ{Ç̓ÚO$êå0Z½ŸQf.IÄÐÿÊò{õ¾4º¹† {Á‚s#~+‚Ècj `ÞŽ¢ç…¤{Ò½F¬­8´§oÜÍÇu±‡yA¡úæO§-‘X­"‘ž+)qì5â z÷.ÛQÉè‰R_HŠŽO¶°&pkvUc™ˆûüœZ•»ÛH,ÎÕê^8$Ì › „ÅTªâÎõ¼Æ‘¯ž¶j= ±Å¼ŒìùìÉKvØÎl»}²|¸ŸÞ¤ù¸wNG>·ïÓµË?4ø,˜ÆOHþhß éçØñó³¥Óê!Ôü@×{" |bú#}€€÷þM²´a,Þ³nMR׬å£NµKäUv_,áZ?ºÑÃ{ù°&Ê3ŠÝ8Z^‰y™Ä”Õ> ) »·n¿Ï£ËžÆ%™(oÝý¸¦ôÃQÃò2d|/{Š*UùÔö´ Ü€:Õ„·o¶!¤Áá¾R•Yûqï6èÇý œ6BbµâsvMc²pPü` ¥ÃŽ'rÊÀ$1óøcÙ®T(Í ÔðHvŠü>¶¼“öý“æè›˜bøy/‘þùYŠežsà"‡Ò•&…!¨0þe³ ò£°!ôÚ)£áæ2ÜT[Óž¶ÁZ¼®ïøœ —–îÒÅ+é“.çF~L¥žö#4`úÇ…Z ÂÑ¥ë_ÈÁ‰PÃox/\Òü¶x1aŸòuê¾õDPc­ù7*oN¨R=;6ûw¹^†4Ó|Ñ”»_·‰€å­©ç¦ªaÙ›§49ž-ÃÖLÊÑÄ2"àY¾¡Ä8+kè§¥Eç åUT»ë>{Ž“´={¦#/³Ë&µwÝ$k8³µ9« EŬr÷(ð,ŸUbí6Åw¢ &¡¶É{/·ÊŒ[ì;è½[V§BrÉq5“‰N•Øt­ &Â5 'š4µlƒ€ú,€}Ù{÷Þ…ÿ´Y4ÉX$ÊhmXMôÚGœ‰–‚mÝãLç𽓪Ú[ˆ•ô¨,Xð¦Æ»ÖçÀ¨9ÝÜI”:Ä`È"S~2¼…Ýúê€e¢¬G–”ÛØdùNÂN&È4w+®{ã(üÕÓ =¡ `âžÖ¶W´õ(œÏÑ5•œˆày>‚lôË?555vHŸÁtxDŽÓ¦y+U/Ï›Á­Bȉ? ìèE#ʦ¤£ðÝ*T­êÿ†]T y›x*‰06°ÌÁ§Wjrî ýje&™N°§-Ü(¬‹ÍŒ5Ê"œX¾ÃjYÌþJó‡êy y»Ã3+I šÁñFdª¼ÿò&Ƀ6Âñ±r)ùþüÅ *âíc˜,xÕ&œhäõl››”ÍUvJmqöTµÁ¿ldqZûlÈsþôíTéøòCÖšÝ Û³¼Šîmëü‘l§C£g™Âx{Êg¿Ñ1Lj!~PüžÚS»óìHv ÿj´` sÕ×[¤Qm´v wذÚU‰DlW…PŸñž%'8zÃHWA©‡ˆn†C<7§MÉÌ~´ûÅJÑî‡j¬f9`w5Þ›°ŸfÂ<ÂRÐä;HÐ@ʪušÌmPƒ´ˆê:;V(×(q%Ãî‰B½IݧßrLËaÙ]äsè¸𛹠°Ò€!è_ œ¿8ÄZÉ÷XB% ñcÐ  Ù’jÄ¡<°8DªÞÖw—Lòé Æ7zdùnºÝŒœ›] ŠŠ‰LŒÙ ±SJ…ù¼Ÿ°á<̧.2¶äm’9üîóÁ`ë÷ è‚%(|°*ÊjB9Üë!Õ1YG§RçõÚÂK>$c3#˜NOzÂsŠN|#1]0Ìip‹ÏàýÈîR¸r–xÔÆƒŒ[•.m*nPÂ>ÎÚ†ÕQÎàîýcWÀ+íàïɾµÿï¹j#´™©Šc ßp'ŠÑ<Ê£Àäü䟟É‚Û,ßáèÃ(22ýú#QØÒ$Nñhþ$[œêHâî…Oˆ­_¬{\™Šbß8ÂT·î'¹Fµ?†fòFJihªjWtÙªÀ3@tø R¬Û/ŽËJ\áêÜ8DÛ) dcöfªaÙn©çuG¼/8‡{ô&EÉøîÎŒ@Ëo~{Gí —ñå{ê¾HG {0Ns+¼«–Ñ;ù‰E¼ÙC¡2¶2uÒpwÛ Äç‹õŽe Yáùþ¤,›ZÅ¢z¨ºkÆÄG™iÑ*TT¿b·c >0‰T,x¸zy£:ÞàLÙ©mmËôüÐÆáÿtêÝùuÇ”¼?Å|ZL[ HŽ, y²úâ½ðX”ñ@ßß;ÒKU:ŒiD—*š%ú“L\a° }) P”ƒ.es‚·:‰ ¨ „úa7ÁºfˆsOÙÒíjÑší-àIõû·†Ëƒø|Ò7*'6Ùr¨¼Qhܳ»Ñ;®÷Þ=f1ø$J}ê%…ê¿×t ¯2¥¡üžÜÇ›©ä Óù² °­Zÿ7e‡³\Ó¿.£¦6¼y†å;@ Ñ-n—œWЯ^_58a‡Ä®û § Â"©aF¶@Z‰ÛU zÐaÙ8æ§Ì¶8<»×ŠëšcÚRâ‘7”FP,d v=%¿ÏF£+/Ùã+Í”;ž¹"Õ»dÇ¿€?kðcÑyÊ=\¼J ¡ÄæIQäzæM¼¨Œ©FFè4J\ìj þÙ÷#ëTXÔ+A2Œ÷µ«)Ú¯»…P'Š®Ô&ã@©í˸á{‘õó°Jú'÷Ò™g:ô¢ªÀ}—7 #3@úzܤëd,fþE)t‚¯ª(ósö$ŽV^¢RÂ^Æ_?pOd°Úž10~§ÊîIb©ô–íóï1Ý 9‚ði]â¡TnÖ·œ{¦u?^ Oé¹å2ð±r83PJ§õ›H-¢ÿõع¶l\NÃdŠ lä†~’NÐVLUT0dš­¤´7¸¦ŠÂ tú´5ówÛ®¬tâ¬_ˆ°t%kz¹óäÿJ«9üñ½r£Ì<ÝUšFPÓW_Y£vMhH% t³ÆÎÇêOp{¤“qÕ]¡Ÿ27eJ§»>èíSiœÛs{˜³4Wè󈞮+§œ'‡³˳/!« ÇXóÕÝ‘žGM¦¡à†Ð—àÆwõL†NhlüRò-FX¥½6V“°s“_û¸ Æ?–ÇÆè«ò„fjXÀ¹n”¡h“³Øâ$é£É ¬ ÍgɧdvF¨„÷áÃ¥á›ã" àqÉ!SËsí|¬T á+D‘Æ`×%5‡…WËÑnN+ûT†^{Ù;^™ ñO¹§\ì>Î7èîüb­öª¶ïhÒ‰²¿ex­|{Õäµv…_5x홞žŠKõgùÞúc¥yxßSsmDûœ±¡UæW…¹¤´ôïù®(¯ˆÝKù?ö°Ÿð¿Õ”}Eöï‹ „!&µ]§Š@˜å/|üŸÏ%y²T“‹iœ¹wÑÝÐe OP¾uo/·ÿOQ[Kq9ËÕMrèfÁ‘6T9õÏûɇÉçmuGžüÓÉVáB{‡íðŰÁCE9×™Ïo@¦écÑÅZüÿVÁ¶á±N—p±ª ¾L°8ý÷GõžëkäNŸèɹ#¬ÃÔ4ë-fγ|ÌMÿOC¤qÃ#Û`†™‘j0@tõøÚ¿êº£õ„¹›*0¼9 Qe¼o-"ˆˆ—MVÿ\’õ?ÊQW/¤9à¾zî)7çýišMÀ!µYþµR¾f°è.4}£s IÝÅâIÍAm’3³?éU¬Gg/ÙÜo_“<ìPø<Ñ—9–og[ý»›k ¥PZ1éû™ºÅ½®©C«CïœÞ¶tý|Ñ|ÿÌ®Aì,#Ý;Õã¬RB G{”GÒ›‹ ¨W¢ÔcªOn8ïn·j5^Å"2âTª;¡Œ?ñ½ ñ›‰[æ0JñH€m™ØûTÐsÁ‚‡2å•ÅâÂUW}èææ L‰uˆ`¶«ÐŸú„²#­TiQ$—fõ5^Fv!»¯N-º¦ô$·‡L_‡;¢f­Œ¶Y²#{&ºjžŒêKÄ \÷”î"‘{lÚ­œ>ŠÿT¡dƒkÔ‡Ûïž\†feAßLzîÞa\bÀZ¡P™ Âsîf >{ú¼ÖëLñºs¨38ÿ«Bh¥6é³{bŸbÖJ/h/iËŒ’óì6á€~5ËÊL»* ºÌê‚Æ_*É’“)H›)Ï–-ž„±£v±+0!Ù¥Ðk}×gæY›`n÷ùžˆ5ÍwŸCÇŸBÙ¢}èHåšì†&À- ÀÛ…T’ƒ®«‘¢Ì%ewpuc3鼃ï¨XÀ[ôÖGsÆØD¥TãÜ!ðù¾§Óמí׬%ÞòQØ(z [#å‰4Ñ­a‘G4xØ%Ü«`T ·~pU!áÍÞ=­ŽÜðWöd¹!¥i‹4úå0﫵ÁºŸ6Ð7“îVt°ªŒúÜ¿9•±, &ïšWnG‡D^8‹Ó-·wëÙ)Æ$]£nÃ!ÐÛããØOR‰ÎßÁ+Í#ÿê³è>d‚ʳNÎ¥Tí mÖ+:×Èc‰bBx6á¾dj¬bK^ÏÔC—›ŽD† ‰º/íµÑÀZ˜M‹‡eæ3æ[ôÄN6?»ýš€·r5¸9ìŽïÃLfœôÍ-s’—ĤËÍ6sa"æÿx)a ¦'ê º°Ñê ©JÂÐÝ—•ž? K‡ZÃE[2ßî;Eƒœ<¡èì“uvª<¢53ÞT#Í; §<@ìE¾1d4°ó³45Ð6Ç5»Åash`É-c!ãj¨ ˆz J´Ô\ò·ã6líÐ3Þ¢¾Åã'€][x~K»ÏÿíX`ÜâT^4º€ó.™y¿K4€‘·.©š[Ÿí˜š(ão+ZöIîŸ×ÔËø´r *¯èÞz7嫯@¯³,-©ø€Û1Ë Ýß«ƒn»eàdƒÉ]UN9Ý5©}/sù F2(ÈnÈeUb=“˜~ï*‡õÑùr¾E ‚N ‹äÅŲ Q2o(ÇßÏ^”IJc£@ž\¤'®6¸]MsÉÏ–jœ¤½ h¥žW’k—Á-ø)7lè"îU‡Ñj^EHž|ì@¡‡ªÖ…J --G(*‚½Ž—üGN<,€÷¡œÏΈ®)JRa%<ð] Ù”ubœ¶Þé!iY‹–>}ì§ÛƒïT‰ml+ôw&Ëò‰•,uh•QÞˆ÷q KmKdP€ÉgºÛ•Y´¶×åU|B[þáµ¹©áÑ…è,ä…çC7ñkz™•Uã1ªWù|¾Ž3¥Øz9òl›7çGe[ûÜ‚p»@QºÿáÄ®ªèóö¶«AùtÃ@Иæsé V?¨ÕMb:ú¹ÄPt¹°'=ÉÐ'Þ±)äiêJÏÄ!QúË]ê–ú+Îr¿2› †6šƒÔ•‰>¾v,ôƒFaä…N q[„5ò²Œ½öé£È ’R±·%&ùSéû5ù Îk ?Mô³yUºþ  ††vôË}Å=zæEà*33»Mç ÀfX›2âçhU YE¨9ÍÚj“óŠ^‹®aw&%¯ÿÄÕ¡,†¿½ûç€0;m¿E¨JÊñ‹Ññ¨£mã¤þY3f«öSáRNG¡täÆÎqŽX„SÝÈ*,1K` ×o긼SZûú=äM€PîÎ9£žKåÕË™\SI2ÖánÛ^©Yå=v!9£P‡J©"«àË'ÁqPU=EDêŸäQ‹|&»w ÿ¸°ÉÎ8± hôvšñ³¶þñÛ{–˜ŠõG¾•iiúׇ«ØÆÖG†™ÊޏÉÏÖ^̃¢Ù[sÒu´V‚ÿ²Bt_«8Y¸ ½œP„éaà‘?LC ðq”Ä–áà:YXø)2Hv1ʽéÛî?Z®0·®1eøU¢s P|9Ä|óP=é#™mG/uÄtl9|ev"H’bßZ\Ñ­¢@£}\×Ù!qMÛÅÿFb1X|g ŸÃÅA¹b^S žd'sœéÖ¹5Þ@ºLJ²Î24†OßÐv!x†“;ÎpßÿÙHÖ¥”5½aØ:jv&Ðúó8|ö£Êà*¢Œ€9ìaî[õ(uG@Gü?Ï|¦’Ì~§¬ûKÚvêö þ¶«­Ô¡1»Ñ•ô#‡@98Á?O‡UC%msá¹W[ÝB= ¥êfú™]ƒì|òuöÿ÷ذ›òòB>è°˜pì1«þ17^:1öKú¯æÝ¬ïîYnkÞQ’sÛr´zOϘBþíÖÈdrÂÊÄwYVÈhÄbêzt2ج°r¯x¯Î•äen!ÝV«~ WX@€üé2;p¸Ëoî°/FËú|/·Êé³Òc3ëeåÈO7Y4RŒ™ 6NqVlŠÞ÷–Žb§Ú¿Òcá<Ä@Síp¨_dšewHÏ éHà«-0ì¦Æ¦é´®äþ݈¥"=ÿ£/¤„S<üXu÷ $C¬âÙº –ºDÐnÝŸ(Ö@òÞ¯§™RI ÷ÜÊé©Ä—N£VGüÌL6CH$kÑr49_kôHý©ô@ŒAêpá©@.áwßGpÑ®ï~ªç™¢œÐ¦x*΢Û¢Þ#(š"€â×±ë™zUŽ`,k<-†÷N¥«&YÙD¶êéIŒEÛ~* IÛÔ<„OBœ·þÚÏqj¬'!óPXØ«¡£ÂZãkÕª¬‚[üëŠ>7ñ„ZjAV¶õJ'éu$D‘æS¨Ùx:{¯ˆÒœ–4×Á”!ñ¦d?Å4a0OO,–iì”j«ïÜ%Ûø%§á£r êœUYI!Ö~@]̬äW…– 9‘+“™äHÒÉ9»û8ÒcõÂÞ•Š*Óö%É\“É–=ÍWÃŒûçLrç’ˆÆÃ‡+¯HðUç}¤äœIh8…r¦ö¼H@ãhæWÈîëUSþT²DÙfžHj@sëÊG$°”0³f ëÝšhvÊ5°ÅÈo›rV̈oû®î]3m»ú§È Ëdûã3c9ÛùŠnÊ Uüx)‚‘™]Žž^fA8¿ Ýu­z³Ý'ˆ2ºqøù¢÷q·˧«/ È%ŽòkYkÈeA|·Ùx¾±lÍÔOJB”£6*eÏé4î+‚¼*ôœ„à‰áp…µë IJг-<¥ïÓ#3bnóoMeøªdèŽÚÃÜ«CX•É,|0QNÛ±`S]>¢ÃÙðI%p#+“ ëpÌm\s'mÝáHó¸À2€ùå,e\Ömï»Ä 3Ö± Ù·/Nzå]MRs¿º2VÐwf6™’Áög Ìv®E š¡’ÕÒ$ü½Ø±é"‰¼˜°qúŸ•‰Ÿå'|ªàç¹7¨^ú”zí?FŸo(âûð'²Þ=ñK‰É—¶uÁ óhÊâØD` :I>Іmß®1,˜Jìº28ΨH߃>ÆðŽ÷Z£®+÷™dˆ'º.\ÆÌ=oÊ3†-ßHŒ{=žòð;Þ%þí¦ˆS#°à„¿è'ƒ1ì“RÛ_/¿ª¼’¬«K®k¾ÈàÈ(‰<Ȫ¤)!é¿ô²¶ º{s–*' VzÍWú\tÖ$Ô‹“‡Í¢åøÏÕ™n35ö§…éïôäN žã£–b‹ò¥3Žõ»é‡Ã<;¶êµä#™AdL]|¦B,Ó\ù+Q”Ò;uàÆntžoeËГE"0áæÏòl3\Œ)7nJ0}|Ö(Q×ÍDo«G¿ªˆôù&¤â@ÂYÛRc7=w FXÛ‡¹‡VQèÊG„6XµU-vñƒ˜‚~À7ÈQÉuÑ]vçÞ†7$am,f!¿Nè³U ±±ê? 5{~׉a’V6"™4¡–ïטÅã§mتáA8X倹¡J€÷_´‘”Êe·;SïáØä&Ü©Ddõ}qÓP\çD·Ë&Tû‘`kH1eÌs§ü¶pCšK»êónÒ>Êž€cú¡[Q{dH”wÚ»Ô³Ôy]±Rk^ëyӌ̸zSêDO‚îŸÌNÂuòi0çϬ1^¤Õ \œ¤ýyÆ{Ei\Ž69×1ø…NQëB4M|·Åa,KFà„ä¡’¿B™›—l?öd«b³áéVÝ„±Óâ)ØcNB‘ 0Ç:‘ZCTs¤Ž¥ðMÆWFÛ<^{ßyòìߟèîwWHÏʸÁ#@ôÒ0Ä}ÄQwÊ ˆŠ0_Ñ$÷Ÿ"Ü„b’8¿Çƒw³Eo&†$UÞ‚k%.®6\1›ÔCËÒ×7.ú悚®®EêŽÀjмUtˆÅ“ÐÄùªI°,öãVUJ‡Õ¨6OS$FØGË'¯#A\9ÔP–ð§ãÿxjþ³J-*þÈ)å–G`¡ÇGD¸©ÝîìÇ×JÅñ`ŠGM·Þìe—Ÿ“»nÙégËbÒ²—í»xE§i ²|)æ}q1b}!74¯ÓOŠÍá_ÄaáÜrzý9 ˆý€JÄ7B¼úØ)¼%Hßm½Ï–ŸdÈ*u÷£Ä Åe>©0àŽ¶vüŠÞÞ¤!̾X ¹µ»UˆÙ¦©?ü¬žÖ°)[£ï~Œ‹œêSóD )¶×s$Iyó-þ´Û›šEÇ&Ø)ä1‰olØ!¼|ŽæöÌt¤J÷ ×͆ìçrbÔ`&¿‡É M±£Í•›±nÁ"Ñö'Wv‰ü˜\!Nñ#ÕUÞIt³Õ%ïbHÁ:#DŠZè"ì\ów vKª…^ÔÒ0%léñÑ‚Sçï`)‹ürôˆt'­j­Œ ¡CÐrqk¾ù¿ÑÊöÃ_ýó¼4÷d:Svæ)Àå6hn³ÏýÏDLvéOïh|ó¸°üídÍvY~Ì|Þ0îhY‡‡ÆHìlå…SÝK ¸*¤¥VIèМǮïGjC®†Ô˜IަJ W†•7)OÅñ@ è’A-÷$È^:¹ÓžMàËÔ"B›<àwæÏáÜvŽ?à#g–¡€,Qh0~¿ë°wÆA®x¹VFç”DÝCÿ9`ýƨ…á‡þþn½HrÐ5ù3id`Œ4·æÚXü±8Ôƒ½6 k§ðI ±²b¿PC‘6ê Šê¿¦å#R•C6}e¯aúäa ó5d½²ñ÷ÓJ³Ô·Ý°bõË©ÛÛ8êQoü )ª¤yXsi„¢æóÉ‹ÌU«K|bmëF‹@å9 "?"7Þ¦ý KµctŸl§æ5,Ù‡Ö-Àµ-§•=OD$C? ·;£éè;™ ƒÃ`žx»;ËyÍS–Û+YwÃsE†}!¤”—³€\A¹”î!¡<œßì€9:‚ Ř2ÀâÖøeí Ùå_[Æ[Ô_«ã‰z.®¦ÏŸ½Ôò“2¬É0q”ëÕ™‰aÚm #mu™õšu]ß]ÍÿUŠE8±©êfû5[ºU’ÖT6ùÊÏo|G:)wS,Ë •Úî/ÈAbx)ïxˆÏNª¾œõî9INuÏuî<¥DµuPœÐáQá¹y‘¤ô†ü÷Å ¨½#ƒ¬ùšÄ4Lüå{2|z•üâ’­2•‹SÜφö>øp ÙÑE¾”«xªËà«ÝÚ„ÃÃ*­1ô3­ÿp(åéráVø @³ª§zlÓ>~¿‘[tø?h{ÓTUÔ©tÑà¤, ½+,rÅuâAE½¿y¼~ÑOË)ÝLÿh”cž2úó™hÈ2áÉ€_ÏD¯ï.ÍÈÆú„W\=Û4ðú:û2Í2ê°¬âJóúxïzÞkðT•‡õ:\ay§7íð¢$â³9Gx¢cj"¤^MÄ•¹ ZÙR×}Þ‡CReaV™_½½ä`. ^•²1þއ´ÿ¶#<èRÃÕ_Ucð[€;lO‚»W“ÊIpÚ¯äàè=—¾Ê|x.Ñ´UФÌ~Ät|•”–Müpº0¥9¥"þŸg¶KN8Å)г’é§4ÒÕ›BË@ó2Øù GÐsÍ»‚ Ó‡‡:¾IL¥¼ ·Å‹úÚØ.CYÌP]Ÿ`Ê ß-S¸_°îG>‘Íîß3™#hwÅR±Û]C%€@¶1›Ed/5/†q»kœøµÕ)°´É uodœ7!‰#x4à ÏP£Ôô ÓèP³Õûo’€Zi) =ùgþ&"'÷€Rj‡þ®J¬<„õôMRv2q5ñet3üîŸÁ¬êò6¬€ MDn%œ«èêî`éÎýÄ~Ñ9Kh+>aî7PÈ´©w€Dèoܧ-Q¥wE¸MæÏ¿@}°Âi%ŽÊ`sBDY[ºó0 „–ûÌŒ«8B¾“w^ÏQ»Š¶ ©b,Ûå}cCÃQàðm,VQ2×°oáÒàÿZÃãtm€nK~ Ë£å¸Só ÉY-Tб±X÷5m.d?«":wR4´éeŠúùJÄ1Ü®_ÇM—iÝ,¼C‡ !ıæhŒàF’b—7[á&e^Ü æÔ7¾RÖ -3ù#þ?¢âT»¡È ûÉŠ'«!†=žPaÁ_, ¸JFô*äå¬Óñ8Þ—ïa‰žP ¹£Æúp @6»”ŠêÙ»#Y qTˆ5|¹ {ºb<7ä ¸©qÖí(5™gÇDHa…Ã,îÄè1(‰H,ØÔ{}j›¹€ÄŸ“¨è¦¶ªªÒ.šƒkÏÓÄNG‡¤¼Ü_S¦‘!ù@2ˆ@—bÃlj̨>vöf0¤Òü>ŒFnz6U´©;˜G¡N¹»Ã{?›Yr†nŒß·¹Y¬½ŸÛÕ²•ËaØ/T?¼  “YýxŸ3Ï!bOUxý[ å^^lâ·üvTó$–ü1 ./aÅ×¢—;áð}õcD™}„×éO ù¦ÁEÝJ·ªêéçÎ'—mw¢bøòfZÅ•¹Šôvœä?‚•šTê3#-ºw2‹f¦l &pºÂNV²çá-¢˜÷TvõNëÞü’¢V?“ê23ïN˻Յ™ÓPÀX˜#À°áŒÀ(•§_E)Ci>¾—®Sýõ.›\…þW™ÊIxV²¸»ƒGqH 7¯ÑPá›ä¦þŽª.8HÒPÇh²ã²ë1%& `)𬶚¡1Ø&¦±®bø3/„˜L.¡³a™™œ¥û½¸pýä–Àç,Ó{6ÏŒbNOçYhþ\­eš2/BÅÈl2psjÝi;‰(,õÁè.JŠEf(˜Aémü­Ù‡¹r÷¡h§]ví‚RKv‡ºBVÊg^;[… ™ª³_7»±ï(ÞjÀ ŽõbýwÒXëß×͉y$ÍØŸqî‰Ž_hAÖ ²XÈM*ž()ÚïlåËq”B*Jò ì)qìrËÞÑ­E$&Ø#Ézw Äå„”ÄÇÏ«R™RšR“I¼Ù×פ¡³ú¨Ê…Ÿ "£`Ýjk$Oy‚,â†]§…Éi†W ®¥ð‡¯ž±Àlå¾îÝ/ÞðGb;{p„Š4ü×u 5(¸fòI;¼yq¾6Äq{:3'Ž3¦µ,yíðßèèó8ÿ²µ{ÀóO\tbdÆæ¥"›(Áò|p-/’$«4ãrd«\ +Õ<‡•Û®æyêY˜}DêŸ_…¦rû5%Š´ùTÙ $cÛ;؃{óM,ªçQ¼»³,¦Gì •2!o#pÄ“¦ì¤ñdØ m±‰h©²?ç»cÞWÒQ®b ‘óE´¬®P˜¤ï„§½ž‹$KÔF¶"‚v˜úæêñV;ÛdÕ7“§´»Žp]蚿ݗ5­fí±Bxl¼è¨Ü€ÏpÇ—zà}·æOL‚×ýb^öaN¹Î܉@c»ú ˆ<›×µÐOç „£ éÀÝ€œrÊyŽ~Ê*"—®í¬ª}$Î\¨,­¡Š.üAšˆÉZÁ3'Ÿ°¸ Ð-ª(5\Uñ¿Kyª¸þv7:Þ@1¹si06¯þOA*Æ‚wézX+…î œnjdÉNÆ:»2iÁÞÍFÇLt®f °.BRÇê,-øš¡3¨1ÓÒÔO¦4tžc_EZ”ÓmGLÙq7zBÍ¢j!óYðÍ<¬ã!ø+3S ßD2AææKq2T2\fe>ºöî/5Ê¢¸¼ë|1‡€î›_hú`¡4xµØ?ù©ƒ Õt³%3#}Øæ‚‰ºÛÕ|Dó"ƒ…úMNkŒV k·p¡¤z4ãŒÒ3er±]ÚÐÌ@€J &±ñË+²¶ßqo¨Lõsõæ“%ó?WÿGH+!“é•{O ž”„ó,n° bPOWø {M9xïÝý†TÂ¥A\ZY¯u³5ËÛ¯iÓé¥LY €zˆV,%áFM±\æ¾VÜ,Ü2ôÒ²’Ö¡î—X!€Ê'.3l.üæ BzÓf¡im!Þ®•¼É<|!`·=›¸©8O†fIM¼IEðÄN”Î%JÇ^¹ôëúñ¹ÎÒÕü ©ðõn*™à‚jNû55ïù)_ñû‹/á‘5Ô}àå$ZÛz‘¿œÕÌÍ•fü`‡+Ø´ý䦎•‡‘Êüµ@õ‘ ¬N˜äwg/lñ[V P^t}0 àä××÷0ò%¾¤:2ï×W¥s-¾MggÂ1!@Þ+ ³t¡.U¿E©¸•'éo]1§’«G §ßܨR­ÏÐ Àc4Ã`ߥ¸äV£ Ê_ÉÔ·yœÕB…‹N÷ÏqæAC¯ìlæÉd®ÐuŠœ@ê™Æw\A‰ÉÎÍ~¸`ºdþ$ø*ô‰sQÐ{Âö­£§ê-l~}´6¸²‹hÚ~LC€1\ü|LRÈÿQÝœ˜/÷ç–: ˜O¦ '+ bgçÐbîØt¬¿}oUK2^~xgÑÅ™XÁGÜÆ–÷Ŧz0×G¾v©>+L%{Œ„«Ï2Z!Dß¹8‹òyÔ#9ïÿÖ,s†-8N¾Ð Yåzkíô›È“ê:)#*™¹­cÀGýš¼Že—jsœjP1|ÑLÆ6B˜héC½ÈLRrªÞ<­ð²w‚” <ã|¥ 2x—¢õž¤=Â`}uÄdaú‰d^4AåŲË>- <+!ø@îaÖf”Jcg6ÿýš_`µ‰ L½zYj¨Hƒãgÿˆ¶“ýÆ Eצ_î.U—§þ©nP!ÃYÈ"€ ÏâŠÙh¬îCH†x¸‚… %Ógx©[ÅΨåY…­tmŒ•ŠcDJ ¹]€ŸT@ÕÓž ;ñe40>P‡ÒXæIŠ_2GœË®;žÆ-0›o…¿Ò~õ ù9 `EafY;i´ñBTPò —“.îÑ“½=ÀLÝQ›i8nØÓ¶•œD-·¢¾hx&ë¾|©€²ËŽqR(ZnD8CDEX]Õë™S…²WdÜÕä«ÞWÒ"dÒF‘ó† Œ¬ÒUî}Þ+'w5„C¹à{ê…ï€Ò½d / mÀXd–åAÖŠeïhP{¶tõÂt~¹äÅïVšØL#Óa{û~7wDºK”é¯q?R‰0‚öðPdç?ÂúJé kï¼Øí)h­Ç8=<ÛÜ7\K¼?*æ^=â÷ºiËM ¯çœðãU]dBòQÀ¥ Ötk¡³)ªþ›b³·DÚåuß*ÇlßHnh£"Kæ—E Õ|¥Ø1Äü§|ì<Øá|´K£µÀ½©7VMö-à0nEu²léÁ"ØyÔ݇ ÐãÁøáù n݈‹›§yÍ{¼.1mÛ-nýÎëÂ;ÂŽ\4T…Þ¼Ö«]M²òL2vÍWCQ£ÿÑaÜï° Uaîä‘Ç‹Œ ‚Ó›ž2íñü·`ÕN„ ù²C ©$ jûG 7¹}÷ÞüNü­™®6 Ë2“ö‰d ªb9qŠúnm¦ñk´ˆ–¬¿]õó)¿%ñš½Á¨G°GÞ‘-Õ·@Å&MNˆöúñBpnÑoaµai胛Öm2VÆuq;$€Ü+,LHiI|à¥î-ðSÀÙ¨èZª g¦í<aKÝ9#ê‘ݶ˜ÃÍÓ{Fl#--ÊÍ&íß)á—¹·ÿöØ"ÇB¨2Õ-9¢#ö8C}gxsTSyùY›ÇÀƒ’ôtmcé«ÂѨÈÄ“d.RtðW;PY3ÐÖ?›Ô4„…Ò-$‚¤Àâ‚yÂ3ˆ²FˆÒ\ÅÙDÚœ$–£ÚHÌárP¦W[·e:«%ñtf„Á»N-êhA¸ïC°ÿV׸¶ȶAËÖ|ï*þÐX¸“²”¶Ÿ4ôËørŸI²þ6±Ò"ÔÑ¿}×9óà·ü{¨OVéˆóç§7~ØX_õ¦&}~>Å&é $¾è§³ÛJ¯^%[8‰Ï*áýªr= _WŠÿ}–Ì‚§³Guçá”ã€;〵•{ú!«Z³êMÑ©Êþ=â·£~ܱuýs`û¿Z…ü„ÅóiŒz}ù²'“D:gYòª¨SÊ{"ù×sâïZC©¬Chñ½\.SÏ‚‹q!Jö3jžˆ‰NÁ¯^Œ [Uš¬ yƒíó‰$žÔ ´ðWAÔ@—¤Ãç¹³ˆJ¼ª6²îúÿŸ³xLÀŸàþü½ “úgYz˜‘â5ǼýG•hv ‰oÿzÓ9n¥‚%í‚Çe9ÞÛ~×ΡyÎ6Yþºcä~ÛKk‰‘‰hcÛ 1 -दô Nù{ZÆégvD] €BþQSÑpåu³0ĺÄ:„Fùø;r Ïúï‰QÒxÖ‰ÐAìß6Fò~ˆá“±"Û•Y*Ì↜xƒÆdø9ËXÀ§°4‰4$„Lz‘L Î+‡¼¿óƒ[¡£€tÛ­³žŸã áÝ_ŒûƒÚm\þ…#/—7Ta—›ÒŠlVÍŠ0‰¢dá·`BÍ M=Ç̦<=hzLóqøí{ñÕ]V„«¥‡gþö:¹7.X'Vxà×Ì샵¼æ¬]î 2pI×Û‘Ë'êëÎ/I+«YØó ]§ñsè=ò£ØJ?3oj{?Ú÷„ºL+áþÑ%YǮǩ1g8§§Ã&«1ä²V_ Øy&̺€¬$÷¼_£U Ãp¾ˆ‡ûä8Õª~:×yMË7³8^iæØã5êG”; ¥¤Ê'«ît}Öa ÷LMÊ›å I>©–ìxÞQ´›§È~–4eD"ëzí{õ0\e$¹R÷üÐ(ñ'Í€”½R…m8|ÅoïËòÈËäžM’9ª°Ÿ)?$_âMê2]èÓ(˜àø­/ì8ÐHa½?œS·y¢8Ó¢åX@aIër¤‹äc.þQ˜[†…iŒ”@*u:ù‘§Ífó?í”™±TsÝXXP ŠT|øòÿžÐÉZ?ø²2´ìÀ8jJ 2>‚"Å!çXYcÆb´éеV Áz†XPŒèGòå> þBVÒª Ò7Œ­ˆð^úwZ0jd®£A<ªø†rìXµR”ÓUšÅ¶SÓꃉp6–P/"r® ’sʘ6â\n˜£¥VÎU²±¿ÄÜ<вç8ÿ´3ÅP9 ’²‘P™êJÁÒ^†lÿüU¢„ÚW@ÀöYiÀ=ð©xèS‘ÁAÔNö3}Œ“†Ó­Oºòt÷²ÐZbïÓbÕSs=7<CÄŒÚ2Q{)ü‹¯Ï³DèY0 , ˆwRD¢Ý¨Æq;qÞ9Y[n„,&LõÙä<äå^ûºâÇðЕ…“Ä ŠëÂcÜbÚ+:.ÜâxFeZ3ë.”·„µœ×…ÜŠ#‚ëvàTggè%P1Ãë¬EUþ<ŸÉ£µ£e7 '»1ßæó蕊YÍN­™'µ#? â<¹ñ÷¬Tø:9/³4v©-¿ÂøS‰ž½ åOÖîv]\))bë\ Û[ðìIÕŽ'fTÌ\ƒ>ùxMG-\†¥ÿÿ1‚’û¥ßÉÈ8Ŧ{6ýÿPsºLâé» ø2+U¶Ž©B Ë¡àÞ¤H9àF–=_"°ohí7¿ÈïHT9‚Z*ù|vZýÊ 7?4oKÔc¶=ü÷GÓI Àt@¡Ít>?ªo´Œš’]¾kÄ@p]î!5òá>o<—@C‹¦<§ñ^ìV¿Íš|°f¿ìE}#°i'Ý)‚…~ý£êO ’o—È_Ñ3Â&*þ·Øþb–}•Ø÷„ô\;œÛ2îndûÅ“uý‡/NnþWÃ&®Êñ„ð ÛM«u Þ–‚ÃhµŠòHå‚ÊœŽšës׉æî—“#ýæ}:Ì 4Êäáùcÿ-½2OR±Ìž¡Ð¹‚·À_ØGÐÄ¥ô>ÊU®Am7M:Þ 5k"u+%´ú’#E!½TL¥´—åö‹¾µ!l¾>ãépsC|˜w—b4œáÈfç4þWšå±;rÌ÷ʹuK R¾mP.3ݘ1^.âLÒ÷ÝPñVTéüƒ37%!?^DÙâã?ªHhç•0~«å§"¢;K0~~:ÆY¶=ÞƒXT¬—0ÉÒE¾!:vEp³Ô;sŽ^þ0àÿ?ë£ë›¿¨(ÁƒJPiÈXÇ/T'¥¾xs† þúbœríÍèÒ íNÑ€$Z1ª„Íñã( ì|½Êí¯‹Ðáµ­!á¸l€KK-\ˆRéßa¢¼g=&‚^­õh]ÏöìŽÉP'ÜÖðC´ÊÆè§G‹“á%]‡5Zމ´£&´V÷Ïáv®Þn›|«áŽ;¶0c¨]Ic¢XÀà Úô ÙŸ±5—#ä?6î¨&h•dƒQ8Ö ƒ¡—ÿ‚+*F½î0r8¦æœ1¦“#°Ê0kjÚë7x'âs»b€—@R©-–½¾¹Y:´|>Åæ1¡ág¹`’Nu_QZ”83íäž! ¶|Õ²€;,^0p€–¶«Ô­¬g!/‹Š35QN9¥ùîßMGÏn„‹ŒÎ™õó› ’Ú+0’ø—‚¶õ¼óèí"/ì{ÕØR\Ò¡™ù¶çk³²»ÛcÀ¬6ùýÆQ–ßB˰€T•:mïM^ó':§<Ö"Q-á\5‹q>(îQ Ùpí8÷ŒÑ`"Î#ÉÝw|÷BHšïE€¾ÂRê ØˆÎL1ÆC7EiwÔ˜Y$~¿}ª…Æ‘èØ—Z{xò›VØ7( ”}ïØÀ«»˜’cV ÖQI”ç‘RÎæ¦UB^¯3‡,') kV×Ò”#࿽-â`~Cï~ŠÄÔï(T›BÁ›˜Ð©Dãb_ú‡T¼]ð*ÉäIu͸=á„—üÙ¸(>¶'D!Ç®~  vÎeFR õ ˧îN‘ gÌ®ó»bÞÑ^Ð,–znGB¦ f,š¦ù¿ˆúvggìFî>àMWÛ3@*Ú6ÊÂo´Ë¿Jk­Ù¼…!çãK=؎錔¡ ‰±{–I¨kÿ¡C–NêsÉá>ïïÔÕ ¼¦åëßáѵäM›es_åÀ ˆ¶ÿàY-™äÓ˜wœ8û ­½R7nÜw°Öµ€êÌÈ•2ŽØ"ài$›/EÆå©xÙxÈßYíuZÈjhŒYãå7Áô¤Žà[î{‘ÈÕŸh‹~9.é+|­ìï õ6‹\²7™žŒUWÊÅVGíF~W­Ì&¡”«<*¶),•Ô@jQEfâc/°<¶¿ÃÅl¡àok…æEƒ”Þ$²¦éÍ£ ö*Ç šÃvnxp¶Ñ†k}µ ¹ûîëórpäF—›֥Ŕ #P¯¡·W/tÀwÄ€Ô{+ñ©ŒG”îU-“ú>1Á`?Iþ ³ùÚ‰·¾Ôâb’•}Tp(BE®¹²ŽŽî`Zÿý(øcZ[üØæÕ«](/Ã|2·®…¨üb ÆÄ…—|?é[vaøX&KÏïÒ%WHÔI*D…+ΰ˰ªK44$YZ®DK?kÙÙ u1òôÒ@ÊÜ`J2bÛë<ÛB¸1ÂÑ:ÔÂ<-î…ž(<QÔµ:RPücOæ™FÁ¼4«S+V™ìÑŸï¦R¤WalSÀ ,ïc¼™=j[iOÕö’LÏmAž{Â'‰ä\6b+Ìh 9ªG/1=.¶b!î º‘F*WëEâÄœ¹(hk l¢t?¯dž©}ô‹Ø* ¸SûÈM ;;£¯“ñ¦~1¾ÄþOŸÌ×-Æúȵé¡&ºŒŒ$é3ƒ)"šg¿êpô5(Q‰¯ÂF²mÛtcœ”e¤HØg”ôãÒZñùèƒù_ú©í½×!axon°aòÙŠ›Óm,¥ ±Õ6ËB‰»(o€Aƒ©E¹ õêèJ N¢V{Eó9%‡Ãg9šÏê{zÙË*3Ý¿rˆËõi:AŘ( ¦ïÕÓ¼ƒ²³°Ìû"¼žž¾'}wÞƒ£½š¾ÒüÛêdñi¥(Mðby6€èG¿Öù¸k‰­–Û °{MI¥qKýÛ b™xNÛ”ÑÄ—›[E†nÁ›ÿ“­’KöN×aEÓX½A?F³È„¨ˆ!\){÷Cͱüø€ä<´Õô˜£X>°òްèvµOq“…'D@ vP¡èƒ-±³À1š@–ÙÝÙTþÞûœ¦‰¼Ý:™%ŽNô[ ׋þ•)ä1ýe½ïyÊO/Ѩѭð`ïÒøjó=Éî›ÓÊô%`‚´Ÿ¨¨9áV<÷…nœò¨EÛt ã²·Åî,FÉx’sc¥ŒmÈ€åPëÛvK ¿©® L_·Ìù@²xûQHs¥‚ÐÛª?_ÒŸý4cÚÓÞ如méCfÅ ›h_i«ŸåWwËŠBK ®³æ™tHéÖ7Yè&fr†;D–w~÷<·2‘®"T„ÍHök}dTñ º¤Å!îqŽIé+ºT‡0I¨#Î/ÝÑU ¹p¸¢ôçÞf¡›ÿâ·ï¢âÑ<<áqš†]΄Œk7uªtašê™{†„úÌ¿•ˆ† „”qÞä³€3[Nï"‡1>z«Jê„;E-ñw¹1$.ºø_€ë‚‰Xx°Ç‡Î•ZüÒ0‚äß=Ó»!@öÔüáâÌö†ùðÃÎë.”÷þÁž&Û¯€È} ô(ŒVN¶¾·¥å%C¯D(½Û‚§µ™·¹ð#ÔÐO4ì³ðôþc°†(úñK¤­D• ŸÐ©í$Ûi 滘û~åÝÖ2Ï tR©PèC³’::ˆ¬ËÌÒ™´lV ç©ÖƢΗ>¨°F+®C=0€ñooûÜ(èÚImí`&×C͈Ô5º¼ý\`š÷§[ëœ:ƒi¾Ý/s®ç›Ç<þ¶ÓÈ2dT9cÊUoà’„Òƒ|°íº–zf2ýñ„Òð·m'H6*:ºªSØá"ø;?Ù S ƒËÙšÑ÷-À”NãsyDvØ¢ùÛÜM3Ä×S\£[î‹™Åã‡H‡¸Rh= uØ0hoŒPë,Ú|®óÜÅH¨æÐÔï³ÚNÇW';9Ïb–ÿd¢<¡JB°ƒLèkлš7Cw—“¦õ_cøÐ·@ñÄÛû—Êl˜ú3LzuÛwR<*ÿJÂE¼—²ïˆÛ ñ”Û¦áu1¡>(Qf”¸ F¡FÿWŽªÿ` gG¸PÓ¥€ác;5-†RBÝŠhÂ~žÍ åå:Ï P>6 CÈ Ç„l)éDöh»×nÇ+Oéçƒ$BU»„±‡0 íö n  Kƒmp°8’¾Ž6û¼”üšŸ³ÿ(k,Püb=Šm°Àƒöâ9¢V~ 7RÛDõ k„H“»ÓR^ô<7zÛøÓ‚³¤Ÿ¢]ÐCª(v«!…o\̇?¥Y[–m ­xSŒ~tÑÏõL‰äËo¿£4 &??$ˆÌ-\1óÍÞhQ$QåJÆ 'aÞq}|tq®é¢+xØ"þ1èzUNáT&˜g^(>æza¢L¤NÇ v¤ˆòžsf‡¨´: ‡ùÒ'†Ý/ȑťV¸á­íñôHÔ|å}ç ¥ò1F%ò¯£«jÄ%vÌZ>½ GàS8¸y²BtÂÕÌAµ2owgGz=i ;¿ØþyyÇ|«&e~Ï5sŠ¥9ø“÷&Ide!ƒZNKH§¯µÉ× ¬t¾Ùüdgÿ"»jš(ù6«ÚîjJ¸³ÁÌœ€ú¶k8 íÑÙ`=÷d/ˆG¸~šzÖ\ ÷""ì¦ö‡*Ø1–ÍÀóþc»k »™÷d‚ÕÄW²±_(xˆ¹Öìl¿jí—• 1Pa&ýÅfšhDϲ‹ ˜,õŒÙÕ½3Ü*×—¨¼Wd½A(¬ÉОܺj­Ô9u¬Ùñ,]Sç nkâXÁWÁ“bB†ó,–>qµÝ©½ßVDNfÊm{ì(ÈÒcÅ/·gq:»’ý‹L*…= Ä-à·²&ÝóO>à:Þ{½*½ÎÕpù…°´‘l{¢j>»—%-Y•Ïèþ ªO¼ 3ео9¯*2û%ª_ÂÂ#S}4çàÐ<{§’É×Ç¥@ UÝc—ÎJ’Ë Œlˆ.âN•İŒ8ƒóæÇ¶Úv)·Špð­˜GxX /º–p©v…iè³ä”;™šïZÐí·öNs`"ky »áXNê1äóa©x5s+dz¡”Ðkb9íª–Ù{¤&?4˼E’<¹ íøÌ)uòŬuÎØ­«ã„kømüÂîRÁ*¤â 7>á0èÃ@GEÕ<š/ÔĦ$*r|ÿ Û²Û…L °¡gÅ|š9ÊЛQ>š¡_h³Ô”Êa&äð [òKßͯÿ£//Ð{£ÓLàÓ E#€Øqê>ÛTÞ¼«ê¢S3‘TË-†O±Í­FyÛb¢»¡ß¡ÿ¡op óáYÍLàrè*$ÔG«l9´8Ûnœ Üè`•Úpï5ò_Sðp¾ÒJD¾tÇóÒÿ´3ŠBÝþë:2ªè^›a#ݮ߭÷¡Ô¯ê!­óx"FAñx öž@^ŸeÁ¿z¤í´¼Üfˆ$2C(`¹çô¡2žV¥Âj á%ŠÆ£§‰«×r…r„naŠÉo.èŠdŽò¬ÆÿÐ1uÐæÏEœØfT‹_Z–9n|eç°¥˜ö U›ÆØ­!}„Bx#èažzåL}OXÚ´-røô.£\³êNÑ™? !‡Òìö—p\ÿ[ÑÅTVwÇïå"ºì8_}¥»T %æNÛÚ" {Q2!A=:}÷'&Ìô‰„U÷ *.ä v»}pÇÞ£Ç#…ã‰Iî'VÙ‘[š²ìNÐÜÙÂiÖbr¶„FÀaÏ(Š$nRïH!¶Ö0D)ó;Ì«{Ý#nö.­©:H—Kb/ûЫĦ’1=/“PÆØê·Ðëkü”á$:ÙÅÝÏÑhÉ…Í–3ëYõ PŽ8B€ï÷Ï%M9>ð¾6îcƒlغ° ×ç:ÅYñcÔ£åÔO¤•]Çᢪm)q¹´¥ï±®{ÄSHé›C¢=«U Ø{M¶“-EžnâGÔsTÖ¡:•pxkcø·AÍt ú¨â¨æ¨.U¹ùĶ™¢ãÉ ?~½±âÞÛ¤¶ðc×4Þüý‹–¨¬¢ÓèºÁS¤+{¸vVmöšŸmzáou¥ÓmŠDP{ul;å9FÊ´*VŸ8}3 aÞß;™3ëûa   váù1PrÇïsc½?Ù°ù1–¸ À÷µ½-3Í'4aš©+$vƒ†ÎÊåð‰O@Àq+ Ãïº#aõÞu‹”NÃVà™Ì(ò”dÕ [Ö B-xÃ{£2î`ƒ±³(rO†ÇÚ¥F?aÛ7w5z2t3´Y =1;EHAÑeŸwÕ™fl ìåuã¹¥°( óÉ[i¸7—ŽX„ʽ·ò^Y7Í–ÓC[}'Ùí‰/^-M1äÜå‹ý“éÈLjkT nò$‰M;в7nª… 0&3é¿í2l’AWŠ©0±ù‚˜’ÛºT/R—#3"€Å¯+Á»Èñ>`N8g*ðeUå£æÙ½:8©Šx³1^éPY]1ÐLXóUoÍ—{Ü´ªc”ûc]Ls þkÐB›+Cä $ ¡@ ú‹ e_Ûd£¡k| ÇœÏ3»å45&ª‰ÒDe!<¨¦‹G«ü¶œjÁDî Î ÜœT£¼)%frÔ|ä=;pÎRx¿u9á'sÛí=ë¢TÞu»PÊÀŽà´ûÆšcÔ×{Í·¸cg?vÔÌßʪë.§%w·£·CUžN@ã[0Ï¥=ü`¥CG«Q‹øÃJ¼:‹"qÃJ5³x|'ÆÀÜŽ¬A‰Ã×}a&uÃn¼¯¤êâ¤6ýYô›Ž-輩€p©L‡ÁÔý#ÂôñZ2í:Òrò¯ðÀ\fÇâœÞÛ|G0€z95exɾÓSz•VœÌnöù¯ØŸì›1.4ú^<]¾ªÕûèñõŠÈ¿"¬j6”cGé»L‹Ëq,Ë*%꣫1ûÁ8Äì4¦ëƒÆó}ªÇíÉ–ìAbDUùO¹ oÕ™r:ZiÉØ.I¹§è©x/”è̬òqµD´¾ 9ó·X¢÷Ad®#Â/…GAR#[øa[¿:ãp>–ÞL+Ž,[é7 ÏB³f {‡Õ+·˜?»GcÖÛòN\Ή>´í–G –yïÕþpOèòhj‘ˆ—þbuV[´4­ú>$joP_Ú9PnÉŠÁ[.@l–×@¨:‰™Gtœ§ýö°!BŠ¢þRâå@ºXLºjê¨Mýœ_…q–'*’.^Kìåd”áÎk-ôAŒ¡ÉMR‹ñeê·5Œzë#J¿œQW··&Ç‹ŸÝ˜åD£Æ­»Lû;•YKвá>¢ÊZ«ëLÛ7òéBBCú– Eé袕a¨ø4Þ ²ã—zH铉gžëÕ¡Êh€9P FqAî‹h¬~㣥˜â(dØÒg]Ð8Íí¶®ÏÖ—â²!7YôáçÖ~®0 ¥* ¢ŒJûAœTÝ»][Ñ~qB° k©…¢¶r­jðÀÆ+~Ì~0RÄÅùðËðæž§)#P9Ÿ…ƃï>qªŽ³ù ü;ɯöƒ“ªÓ@qÃÄñz¯Ôèb7àw‡ÿî©á·¾Š õuÛ[¹íÓ£€10Ñeá1e;‘­Yâ¾®ßThM²5yž×ª_Ùv”` Múw—L:dÔ<EsÔbß”äxyý¤V ?ÞûÍñ€+óo†¤=žzm¡Â¤'³E Ç Ÿ†=¬iF… ˜q/H¤"pïm=D jù§h‘YÞDƒóî*­ŸÞ%xæê·"4Úbù³ˆ´·/Ï$<ÔU Ìÿб.ô3æ ü IÇÚÚårÀ“~!]‡Øá€!à죢¬uÄp§£ˆ %‘•6è”϶ÜÞÿjuVÐÆ§Â‚µ²tûW‚Õ78z6ô¸d§Ïù !ï#ÛTx @îšÙ£ü áñ§”­ü{hýl— È364Ÿtõ°ÈO¾ÌÀA–÷‡ûžg°b‹!ã•-Ci<¤8.öñ›²ðÿXB^ëÐg4A{^_æxÁ†ÈAª®}NM°:Ø\¢7™’k”†ç”ôN½ÜFÚaQ\oƽ¸%kDB‹cSë‰F¾P=ùTæ3²±û« b1•øº p2ô¶®5ÒWÉa›"D™šéC ®(0ÜÁy†¨I#…}ßÖ£b•Oj+Íq˜î”øïŽ©ÊQÖ™àúë§C¥LIcåìÓÓoÒ7M/~½•‚iç’°ÞB·Ì Nj#nÉèª2 »¤‚2öLØhµw†;>[¡Íª‰Acé’fÕú~ ½X‘Ø)p}ÝdzQjg¡u*¦¬f—³Y¡¾bóÿâ“MÔ'ûWX7¦hØÒ™¢ åÇô‚rÄuÁÃîÏ-*AÂV²ÇÒ'Ëb×ÁyÍøæZ?0“ƒ',Ðíj{ÊzJ\7U©kÒÉÜ ZNjÏ™£K¥ÕKß1y †ÄM öÿñ5d<íLÉhÆÄ.JÝÊ¿wŸÒc¦Ì'SÖ_Z¤×¬+ú%"êÓ3¢|ëE_•`Ó€†‰i}û;AøŠ š¼¥Aà¼óÌ%‚™ü¸C*±›€zÔÏØmƒ¿Šßp´ÖѤâ"'þ+¥³{É~ O¡Éó4n’C·l²Ù2BO[&_¡§P;'HS­Tû`G¢€\f››ÈÕ¬W7}#k_¯ö7-Òo“­³îú¬ „'ùdû+‚Õ–±A9‰= -\Ð3àA}Èûpþ0vÂIï üÐYÊ?PDáãùÅý!“éĦ4t/D~þù‹Ö6&0Æ;jlÜì®ÔiŽZƒ¸Âõ®6w)Z‰Är jÿ1™¯ $ìþBô•jͯ£¤¬ƒé<¸Á¼¯´è‡¿!šÃV—ÃcCD¬:•ßì¶d‘x‰ö’¦|y«µ†²&M8ΊŷMDßéËÒµßÛQËCqÆ[h†s4Líð¢èPÓSïžOòFQí!ÝdÓ– óçGû¼ &êΖÞî^Ý0R¹»Ã½ b°æªOºY\ôÞìvÇ\žTÛ³•î}ŽŠ@€ï&ÅUÔ{Uns¨¬~Í ^ÐçÓ”#cïBOSŒ¬Hòñ)ZäŒ`1Qòb¬ƒ2ü™ï2ßcö†-ñùã{GÂiûOö´Ö½£Mß%0Øß‹B_E3pYåb± Üm¸Sbú¨M]ùjò›5Àá<Ê­»2=èB¥h2 ãltk­WÞ¹Ókà|EKŽó’æµc*FŒbç«Ïq? }¬Â¹ à¹ÿ/ÞʼnÚ9Yäßã)~A[zOþݱقTz J,^ý[hÂ-*½ÃÊâÌ™UF°°Q·ö?˜Ø5q@VØŒ«x¸µuëé?¦¾°§×îÝ´6Ò6qÎbŽû»Ã«x´÷cÈNðÔê()F‘ÁB¾ÿÚ 1 ¶ø"ì Vþ#!‚Œg6hÿ"¹Ãí¿–U mX¡Ia°#[4Tusø%hÈy-ÓH??áaˆ—ÃM±ûI*{Çb 1˜ë´ðóúÕŠÞZRi%îÙNÔl‡fÌq§·e‚â;ã•óNj™*ÈóH÷Óö=ùV¿©çîÍEûë€g‚Oÿ§Ötæ^>gCi÷yXŠÐ+÷‰í¢7ã P®½ŸþDÌi3v¡î߸~JíK$\{|°QÌU5YÑßpBR³-È×ÂMM[ù•‡mB?<<Â:q˜Í£;d Š4'RûÜ!Á”ËuYuÉŸc4ôq ¹ŠŽ=šô,ƒËMÚY&2ÉQì¼$PvˆI̸†|GðX>â<¿! vEgGo›Ae¯Åæ#õòøË­í`Ü[L‹SRßUC˜´# %˜V〯¢à™ò¥ý¤,›ýí3µ¥¥²àŠ…¼eånw=¸ÍÏ8öúef¨´ÅÀ‰ZÞVK4‰™ÿü_3‹E‹;RVtJݽK0øe\“8ªå 2b_Iª›T³«|a¸K{@Õö9Ž8µ½H­yoºX|Š%仫/´£k=£ àÓÊFßÂÝá´t‡Ùd`ñ?–<ƒ²~éJbœ×‚ÖåBIBˆm$Çbb90»2ô¿°wZ>ñHX Ç $_«Í9n¾v»¦ÑF­”SëéZ&¹|ì ,Qá)åTü‘üÁÍOôö<à]r§¨¨ÐúÐB,ÿM¨¿GœzRMUóG!Õ{i²}NOÙ¼ëëç„â==%ýá eSäÕêbíõ8L•¤´Má¨Qº|Mâ³k°^Çoª£ªõàÕ8ð_f 1”ˆ bºöê§£"¹‚ë—¸"¾‹ii5% B°B‚ýÿ4ºž©ÇšÐQL|¾>$TG29ýR\ˆÃ;7nþ&ÂoÂþ”ã2ô.è¯UªŽ9§ ïD_Mö12Osõ¿ÿû íˆRÊðe$§_¿zøu‚4Ž“\`¾¶Q1w@j™0Ê `jä>S;':äÝÛ G¸¿d¨Õ—ð\í&é ‰  §øC[òÍvw:û Zöè½¶•×WÔÌÐö³Ë“ûæØQ˜ýP¤’7¨o½d ‚ÚeœDùÈRd(ÎÉØöbT]5ýX0×¶‰ ¢‰×¥Uí™…çÜäêÁŽ'ƒiw`îµÚ9kk 3õ^q^0½>Â…[vùà’{^µqýDôÃãì@Ì¥™ñ¢=å7îêÐIµˆAf±cҙΎ(½›0awý:²·*#Å»hï,ÊG]úze'<+ Ã ¤>½ÛžùëTê>¸Ø5Æ:2-®W«ãî;Ÿ;t©ÔhFg¾ƒâPýÁœbL™d³A£ÂÊdCì+Ì™‘(cmÃûT°þŠ3½—}Kq¨ãžE ú“äÏkñcˆÛóµôS¦Û[ZtK%ÉšBƒ±ÑKYÿ÷6äÛÛÉZæ9$f;[“ º{wKH`Î ûF‰4|"dÿ§nþ5U°–P¢&m3zq窥UYv3Bxe]Ür:‰µ‚ü/ÿ´Ìàª^kdZã·ògÑj(›é½^…õmv‘ŒM˜êÃCï Û¹¡ã_ü©2ì| ©ÚýÜÁojÞXw›JŽ­í¤0pO¨ýpJÈ6ª„ú=Ö„<¬“þ´Sž‚ù´¢Ä…Iã1Ît?å\¾!O¾E­½Å|ÿœè_7~é‡~—&ñ¨~fo‚#ÉUVÆ¢B ©ióÿów±4­Z)ÓfuÌ|=Îå4BØŒ& cQÝIOmÖóžïÁmßÛ»sÔÕóš½ÀÀÈT©Cêý—Ò±u­ß bÚ\ Ó" ’4áš›q´k£š²tºrª-ïpåÿ2ø¦ùF ˆlÄÂÄGåN¡=·j–Î+(Ùò[|€¹ÇëVRÑ(0«¤D7AŽ«+ —mÙÇ”úÅïà¥T_hf’¢ U8Ú ´-ŠO¹É¨Ò_ÔcŽKýZq°]YœŠ>*®ññeâÞ€Q€Ó6È—zõEZª_®ˆn° PBâÁAdJÏq x’’¦7Û9Õ%:¼g¦”=lFÎUÀDjDT`9‘rËŽ³Ñ×âIFw½òÒ¹‹gQÊt¤Ï»`"Ž¥öž˜hqÖp¹é^=Ê7 v¸PE÷èF^Ô1<çPÂÃ t{|¾ÌØÄû‚>ÈOSWY4w .¬b´;=^5/ƒ£>b™ðKA®Jsü|Ö oš„›Auˆ*sN:Ïã(\ÐÇð&9å2ˆ§õŸ5‹ÒÏPI"bWjRY¡úÜfÖùRüˆK@smfûÐEâ–… Z˜H9õêRvðCØÛ4°ãÍ%;ºƒB=Ty<±ï#œ™ +¬dç‹-µW^nò!lÉ´…—žü)Û¤Ü/¦Œ¡õûSµ:y݉c¦yÇ–¶ð/tM…ê%m8d À´ÄHÒèðֵí¤l e̵QB•”Zxácþ‘}º°©‘J&ÍÊñ™º#!쳨J]‡‚·±y?ÒîõÛnÝQ| 0ˆ&ó æV„OZ²–âPçþÍá*5ä5¸z!)m7tÌ𸠎Ñ`GµÏjž ÷ ±3f.ð¤Í3k‡ ñ­îʧÖD0’¸eçK(zA–Z‹‡ÇÅL¿…NˆŠ92Õ¡ø~›7#“\eq ø¬åvëæ®N¹óȱN/Þ -¾ˆÌ/,R¨YªíGR€¸¢‚o Bå,˜~ }òˆ™ü-5o@uylüÈ# òúh.x{ÝTz[v¿ ä‰Q£þŽFè+¯ ru£à|Ì\ |£zew“5–*Üþâ‘Ë¢ÀDaBd9ÍØKÄ®j›¥I?´íÕäÆ6f7Ô˜ñfƒñ±|'à½_I£ !àö–PhÅrñ €™ ‘äïÍDÑ ýí ~ZS¨ð&G2èË©¾ý*¤mcŽ¡¼†æçΔ҂GîæZ´7›¨¬¦¢½x Žp¹Ð‘½ª¢[íq”ÖL‰ŸƒÚ šÊÆõÏš„Н v¸üsÜîPüËdëOd35;FƒAiÊ8p^ ˜1Ó':½&€‘‘t·Éð§UÒVÏñ~€ÚXÑPÔôXnÒùYÇ)e‹;«¤`.Åœ$Åy.¶ë¤7X¾øW°@ÜàïÑ$4á;·ï»Ÿº+4dx ö\¬ÅìǪ*á/…f'oWÆòÿ4_!øD"B» kÖ áŠKdÿÙ Öžä`J³fj3ï‚DÞÞ?Í~ ܶøÛ C}Ï’æ}MÛ HŽM_‘§^ÒŽïVÍŒƤH·­¿Dwbá„+¶5æ­ûM˺ê#eÔn¸£em_>æ§)7ª7¼FµdZ]±31¸*oÐΛ}w<ÒZºÿ{çÁDGƒ¥ ¬½g¶~×·ª{ï÷±9’0¸O…Q»|C*!E9¹wõ'6i3FÅ#Ã[à â¹ÜË» B%C«’ ˜=þN€Ì3g ºFµÓf“ U>üÝZyu9ð§-'9Ìíž…Ä.£wrâP&ïzÜG%eu»‹‰xgaÁmp5g¥Y±JºÅ.µØ¯¢ÿz8´ÀÄ@[¦eðÙR çg,6„¾Î33² 먟/ LŠSãQâ5òüÿÓX…RÂÙø°‡W¾Ï~+õ$»ÇÀ1Èd|ôŒöjÅY\þ7î8dÔ&ÖŸ2 cÅ[2x€ ‰/3LQ ö:aƒê¸$f;µi£Tžr(_erE‚ây¤%6n£‹’ YÎî‰ÅŒ6@êAA©z$=ØY¶ ‰" =¦¤ÍþŽYa;PÞÇ4£šb}PÞ+l(])½ÖÅu(:¨ p9›ÉI²Ó»a&VÎܹkpì˘^G&D¯6ÍáZè¡öb9[?x¾eï¥N.Û­×h‰Má2ñ¯Â8°°V°“秃Gø3rö&ô±*ºüÇžôVE®ŒšÒ'â'wÜŽÐÞhEü=7kry˜'6]k@ÔŠ1:¬ïùØ\Ðd\ZwôÎ÷ã˜Ù&e:j®`F²ÅjAGÿF#ÌäBêÅØL,óðOl?ã1†9ݯª+ï-¼¼°˜ÖO‚:Uwøcs^oø?ÙøìcGD½MïÞåíAÞÿx_`|f]D–ÄT5·hâÌÈK©âȃi#öý¿+¹AbQ²Ösx£ÛþùZªÚ[jšãØyùú²NLÇã &¦þ>6+.ìÌ¢h §»H4F}‹Ûò0àv‘ذ<ŽÌ®¬`#~ñ©LO ¬tIöq"ç,ðëg¸ì¾«îË“øÇ*Œ[}ø·†&ºÆë‚ééÛÝþ±™O)ù¾H-‰Ou(éåÞÁéM'éÛü;_ùb€¨†©|ðnãüXÚÆLü*¾ÕAÍëôsË%P²ºØ=³\Ym­Šù3¸øŠø­ä”ia³ÈZHá$¿Ï[ ÚÔl‰J@WfëɪÌxw†`»•!À\²¬€ùc¼má.­¼‰wTšÒmŒÞvÉdoôºFŸ¼ R¬¹ø"ëæ´M‚ºüDð1n éªA™ûÊÜ^ÍÕh%ïÊ ±ê¿É`1£ÿѺዬ!«Ò(|÷As³V‚/Ù&RôfH·5QVuã£î17k|Wɧ”ÎH­jz•*×ÜÍ+{$¿v;8Ï^òʆ‰3I*ãN¥ˆ4o·“œMËä$ ¬.«\1Â?Ôš&† ²Ã»@d ÃÚªâ8ƒ:篳oD!~]X„·g¬ì2ê"Œ“&LfI¸¢(/E_ÎÊåJoq5äE¬1 ~.H¶³¶ÄfŠÐXèÕkÛGg&l+Ö€ôøa,gßðX‹ûzHhÌewA Yb㹋Äg[À3øµøˆ—Ùgaú/=Žà­BÖ̆®Œ8Ç1R1ÆéÔG ×$‘f¦÷aOf¸"ÄÞsÇ'c@¢q€MüMÀ¤ìÊ~jÜ3FqÛ—9«z²–yI]UPô±ˆôWõ1Ó@©@ÌT‡õC9y÷î}ûɪ)|ŒÀ3ˆæà'ß‚ÐÁÝŠ5¯"”:<{eõ®dðqƒˆãF`~ŸH íº§ ©”0Öªà=àÛô¼ø;è}£ÿŽP©C[Y¿©ØTíhÊå5µïrä\Pa*ÛYX1œó þæ¥æ×v>å!þ|Y×ä[»k¶"[]”7Gͦ[ h»¼o:7q7֦ŠËÂã“2E’wi@o²ÝÆ}· /.¼#a·ÛbI¾#O½dÀ$<*<æ°ˆE$_öýbl3軋ûÜ`n¦†,–•1Õ]m0„ûƒ¡ˆ‡§5'½cf>@î_\vPeòž†ûï­½*€GMêÆÙ. 6ê¶EþVî R ’ºÊpmòh(T[#Î},:ûKWºýu §÷õlâ•Wyòñîl3 ³]WšoÇãÍlBŸë©`TÄz’yfû¿Z¥•t$$X JYH¥îˆ‡¢Ì¿IYà– ¶°V¬×Ÿ‰} É:Z_PŠŽÉag*9÷(:”5<ыڿNÜôžj¹fæÐr8¡ssïÊHeô@Å©šr0©¸Óz VÐ ÞhCYì'ÞG:#2Íœ²à¼z¢*š€dPéãa/I¿ÿš8rŸ¸|=!e¸«l¬Z1Gä/>Mqç›|Z¡6úºÑõt¥‡"us¿D~׿Žj‰¢EúæþÖ·|ŸåNLNzÀ£-¿Š:¡ys ˆD¥ãífõøÈ©ÿ½õVÁC à›„ÿŸA=FåŠÎ>’ê‡7­ÈêII§PNýùW꼑zˆ_<¸¹ðëA°´UC¡ŽB€Ã`ª¡—¬>u¶¬7›(°N)ùHwê®èvBÈ©~NU¹È?˜åVi* ëÙR;=~ù3vŽR®ÁJŽ’þ+ØozÖH V‹Y6šx=ÒLú&Ì£vI³Xãó{r¨täeÌroâ1Ö"ÅŸW¨Q¾–\–={Ã`€.=ª~ê˜ã¿èªT[ }x 9(/ýf¸Ý5µl[WL™GÃk™ÃOBã×AÖœ¦)­8ÕqRÅpšW5ôi>6Ctœ«ýk&‹% bÃÕK³ÿ©P©h)1'Õºª1ÞûMDá"ìñ›Üá™È¨ïŽjÒ6¾Û"•°·œ¯¬Û«{+~dÄáùË Køùà¯Ð>aÁ»m6Òðx®Ê§Œ}Ï¢ïTïV1Ìáu1ÅLô’ EuÌ>#F968ôjy´701abÓ@z¸Ë0{¤MLÞ Î([öoHgp LsX¾¸¸;+ÁÖè9­Lê[L!žØ¬…G´þP C…ÿëŽ/‡mÝÿ@úýUÜ%v°Hp"ˆ³´ë¡×ªé•Œ÷äJ¶Yl‚ãu Ø™ÕÛRí©¹ú݃p\|Ou*ECžçÇ”±_•“³Dy– ‡©IÒü~Bº=‘q•vׯ2¹U³°esxD€¤†£DKH@@qÐìfÖ†âN8ö]œ-lïyv4 OuF¨6»+×!Ђos»=„(ëÚÞ‡ å'·¿<¯#}!Ø+>÷Žïû»/©UYÙ¡ö_ÛÊC©˜hª;ÉüJ‰ˆ’ ~»z:L·Dñ‡$Zî³Ïl†&%§7Ç6ë¹¶\\¯d ´³6Èýf1Áæ”=­‘â²ÃÆAW æE úê°)>šËý¿ÐªÙÂç"”´ýªÚ%hâ/ÿàÕÜ99tV<ÿˆH$QåeÍzK?ÐaìLqwÔ&©­*ÙIƒÊ¦H¤ûz?‘Ãî |)J^²>K+Y}üÊv°d3qÀÌ¥S—°©…˜Lbën£ó¥lߪÀA[$¨‘ Ý °®NB¡&ÉNwÛÚ­£œYC3¾ê:‘sHmîŠÈ ýHHþ^M:09‘ÑŘ!ÛÐÁ¨žüŽR~4®œ*Ø[ÈŸ°ØeÈXÑ :‹ÛòÁ[î÷hE$ANh…ï 1v½å*&¥ vë# üV k’Túªz®³üK+Pí-|žÞ•Çòx˼ûs†ê†{¯Úˆ†#Ïyʼïm .VS/HÊ”Meí™mô{:X|3²ù{ùœ,r¼¦bO}gŽž‘BBiÖÔàʌكú“ξs%ë–uä»YØÆì_Œ„Éû@t³æ‰¾G}:véæx:1œ²ÌÕ°8_ ñð`Àe¨›¤ÞhqaI®˜6Œ’_4×uÞ# ¢Û'J-#¤sÐü5RªßÂRá]NÁÁÞø…q:j%êT©»Ý:æRÈ9›óº<[MflüÕéÆv#kÂ/^ÎÅ& ¥™•ähSU!&GwÂË›TSÏ âfJ¬YÀ“•&°ïGƒß³ÅV’tš%^=^àS äÞ%âŠpÊ È?(§wá&ÅgÍK·| uÅ^–’ÅñœåôÃÍ“Û`oõHì”…+”)-À½äîEeÍP+Åüb8lñ e³vºq-ZRã‹~$Sæ ‘T6éÎ ‡½ˆ¬Ü¾Ö`ᨠ./¾Ñºuà {îT×˸ n¬›ŒÍ¤³pb“ÖJÿN÷{ÙôØSA_Ý)7-v×ísiÀ$‚2‚ׇà…i£8öN/¹öÆŒàßËeÙRúé…ÿ‘H }ܾêp™pzÌ5]p;ï™\Cº5#V­µî¦6¦R¿ªßÁ쾈Ægm•!ª¼.›ËýPæ;°&SÌ´ê 9mfGŒÚcCb)®È¹tÞÖ‰„À¬IX„j‹ñÚbϰ`ŠQX3Ö; MŒ½ºü®¨O»^ÜËwÞ™…<˜¯í T5g†]j–îÉbã%Ç®wZip–Q|K>C]k_ùSÄDœlï V¾™Xñ"$1 ‘‘þÚäßHŒuL Þ¼S ÈVX6!vu q6‰CE37~[—$õ1"} ï½G}ÅMkJã^ÇÑÁîWH°fv^áÐ>ª*ôÏÈbÉç¶{0Ü%˜]g c@¦üÉz?gý·úŒÂØúTíqj —Ø{²yÆ'à(³ƒdUqÓŒÀÒ˜iu¢dVLHAy¨3Ë:ÑïBûPƒwÛ?ŸœF‹ S¤¦{ÆŸÞ`ÞwÔìg[K,ÒCMqõX¹û+Ü<é5޳H“èhNÿUµÁ–Õ[ea+Ó,Ç(ºûPÝl›äI÷ÐqWº` ì'‘J`k{â}r|³éY4•_iyMÎ3jýFJ'T%>¤|cýݬ‡¹uEo¶ßY"‚DˆáK~u¼_ÞÀNÛ ‚¨þi«éP5•\¶SžªFZæVÿYKÀo·bWáwÝy|EΰÃFF´ÝEŒ“Ô³ë`ÎW[¢Ÿ åí»/H Žûî!\WøÀÝ‘øR SOJ«ØÌ.b×Uh9Úûe)$Ó•"TÞg ,¦Ô«&– ÛøN¸Ü>i íV&KŸ>^úHßéÌf›Y›uÒjòj!½ÉFèDâjµËÙKyß ],ªÚ² LÿÎ’ºÀ)²É]¥¬Ã’Òîþ |™æ¼íÂ¥ÍzÙuŸß˜¼O&w@<J›tôÐål œwÖ9(ž“¿Üš° ‚µºøðgB%=$°¡OD¯–(§ŠÔŽeÜ *LY"Õ}O~qeÿclËn…×å«oa†n.d+3,O¯„ åKî” í7BAÓ¸}°2 ?ý¤bmºd6p;¼‰®c)õ5§é¸h™©ÿ|AËÔ óIk`µ­” ª’~Ÿ„5§üªö}<ûÝ‚‘¤kB¸Ú«7N5àQuÉÖzQ¹“O$¤Z,ª!ÊØsÏ#ÙÉ”Õk†ü]åW牢aÝàM $"oHD×P‡¶œÅ€€ðWïwc ÑÄ^)ërˆ7®_/ü Dµàèñ¼ ×ã CÃ*µ{ÂÙž U£ÿÂKݲœì7/qObYáÊÿ²;D|¿]®¬a—z¾…;®íæ3âàî$SFR¦z¸¢ÏÀ.b—m–wÖ—ž „KeA™ôXø‚¼}©Ùô `9Œ$ð+"a#™*ý¥•ö Ô©þ'Å7ª €ÍÎzp¡-7I±å€Î²ÓL¾‘¤÷ã]¦Ý•‘ãq§³«yã¸h|­y2l¡hªŠVâ ‹f©°ÐΫæO‡qQB¼½7ÍÁ4XE°X§ù}ɹ –°±!Ûº€ FgÎv?´üI ‘ÓûA™›gTíàm»†X|­<zHi[½ÊEšù”æ_׆Ý;…<|ÌŽyh9®eXN½ØÜìŽn»>HÇô‘'b~„u+”<…ý`°ï,,&ù¨>2'S†§±¸DsaR,TÔ(ù }Ù÷Þè"½Gü´Yw¼ ÔRVæBìÚqÇDž Œ#Zb)UîWm«ô™'À£Ÿhf0#¶–ýtšŒ*5BTWýÈo²ã]’ÖXV]7ƒ¶"WÁðý 'O›wo µÞ?EïÂ>´è(G>¾”xµíà ÊdŒ=Wú4\ºS"?^ïp¥Ôp=°Í¤à»á^Ή ò½ûËÒ.×Õ@ü*ºdÁ‹#¹2Ù]ÌŒÒUË ƒ}’ºy~ź[þ-S»{iß»g8—c” >Æ\‡ãE2¦¶WK:JÕbû©Óöb¢~åÔÒÕ*B+?ÌÑÒ031êM bgû í´&I2m—ƒÃ6!µ#.‚ôl®aÃ\øŸ…Y&¥F KŸÝúmR|Å}*µìò8Á®Ö“»#ÖC¸Jî 'Ï—Æ=ÆdÎì:Oc¿„U©K‘µùÐM€è˜%0+(žÚZ+bˆ1ç§Ýì€AF†r†Zx eÿ`ÒVl³‚R”x.Ñi ‡W!ÌÑAÏ$¬ Æ Gá×$Œ¨ÕΦ©cP-bd@J†¥&N1¬ï½§˜Øâ9íŒÕž”<ìd¤Á5˜öCÙª…„ª>9œäéš¡nbÒá^¬$݆)ø3k:ÄÒ3Ø¿˜ý+1ªŸÆ¥õc’žì*ñÛ§hí™üÁ=þ!ÐzÖ”²_%’üpéÔâ¨(ãVðß‚&É­ôé—– ¨:—gÈ$eMñçÙg“¹ÅɂȠÎÏy#’êíÁLX¡åµau4¦Ããûã™ojzÍ÷Q‰LA²"lXÕ;¨9*vŽiÓ ãQ¢€ô¨p®ð:‘±ÔD÷plÍ+¾yƒ©ßÖ6XÚryy.4™å/²Bìlm–'Uß^mg†–FÌØ%î|[±àÜ$H#¦6GžÔï$Pv‡`°ÜDífD7UÜCÇ 5à -K‚œÔ¾kÇl*9¶“ñn7¤= wÐ;¤˜+"&~<“¡3¿ÚZ ”P8Õä[1Püõ?Ù_ƉÂ?½Áõ±÷„-¢Å M·×ZÓ@ ˆ¥Ç P'-öXö…hþžw‡6Pµÿ”@×¢ér¢RnpQ8‹9úÖà-‘Çx}¡åù%A[ÇŠ] ÂÎ$-«ÔèU̇Ëý¹h_,íßCƒ*¤äÇ]ÿgÿq2 ·•V5ÁŸÀ§œ0\ŸkJU¼B>D#’óóHé =D7œ¬v´ï©ȯ€/ɶŒÖM¥=éë÷,Qi1v ðšF3&#Ô Wùœ"£®ulaûO>äîM`¦_w ¦Œ2ˆsŒ@Ái¯X°V}FŽ»‰°j´Wš’fik1øuŸ†Æå®"‡Zû{I#D O¼¡Ñ—m+r@¢§¼H—9âö׳ˆÆ‹±AALÀØž'ó8¦GQíäðoÉâ—Û•äwI_»Hí†ãÉ[¹Jzz\å^¢ ¿‹¦AÏ2þÏv•ðTÇ Ôè^¸Kº~ì-¡KØ Iµ¨‹:/sBòlÑEPö*4†ÂY}´pɆ"Мò o]†Î5@^éêVÁ’"¨'éØè1$)`Ý.:@÷§&ï€&ù\lq¨Ô•óqÕLy:7DÏeË#Ô¿¦16}ÁcŸ·ƒi^é§h´˜Úµ?áþË…ÝÍ÷Ej«7Úä-qÓ»"¿GÞ÷´ç8mnC™î¢ÖíÏe˜Éƒ»C/¢a=uÌÖ$–©_!l§ Õùžk1Q[ÀúÃ߈wG²‰Ü3!Ƿ׬e†zºÈ%dÅÚ/1ûó´cÐjŸñ{6.=Ëay†K3,fûÊC:=x¢ÐÆc§sfb’óÊ0Þ3é¢Ux X”÷,vr–éïÂíØ`òŸÌœi„€^µNI¸Ø÷ŒÐãÚ'5Õüpmœ–Ò1l­ÃغD¦GÀ©…Öµ~{vMý·~¨ôî-ÝIRYs¦LX;Ÿ4V™Í¹XÐiæãUCr+¾<œ?(¹¿)å ù[wÆᘧºÞ2­_±¤ë®À€M“´*$RY¬Z#Ÿ<4>M—uêåÉŽR;Ñfåkš¸1à¹ùtãJóUMÿM ¬YøÙ“ô'™Ô˜†‹ÿ¶]  ú@Ù1›«íàÌ‘›:8Ò5ñŸ¸ÉiÚœ¢àå3P/cJïªÉhwZƒ‹™ÏjŽíÍ´:ÿ{›BèJl'óRˆˆ"Z¥ktл¢3Š}ÀúlaÕ诽n,6¾É'‰›ï=«f›ï”Eñ,dEàÙÌûº ±£ì¦ÊGNiMpbíÄ>1Pæ®KPZóEý‚ÜÔÙÒHN …ñ’7ê¨õ~áËìïÞÜ?ú#¤”[Çt–w™¬¬ÿïÖ¾œ3dhN öîÎÜ…™`£¥ÉŽé/ås/²}%fÁ*VÒ¡>b-“HžøíŠ‚NÈ.¢D;½ož´1®LíyÐËXˆC»¢ÆHó½8I^ÊPëfö%ÇJH¹p{)ÕöX åµQ»ž² \3KIÅçƒôT‰uËqÍr 2‡ÚÓåúzf9Ä1v|];aƒôÑÎ_$vAìí™ .Ë›oÜä—kæ›ýŠ–2­þŒ1*¾+å³Pm‰6¶®L :¬¿O²¤Xó£ÉÇO€·ì} \÷s{JÎìùÌgú¬P>v$¥T9g[ M*¸+3„¶]åãÇÉe´b·oœQ•Ì8#¥SM‚ZæËò¬·J ŠæWÌ»Rîs¹† |™}@Ó)Ê Y»‚B-àíƒýÕèјr[颕¤°±ŽA{¼Íñ¦"4“^\8EÿC /SZ"%øÉïb™>ûùó|oVÀ;¶ôS™œî-”í}¤¨©RæRë„.…¨<{QpwÒmN)ëÃ+˜Ã\Ü\¡=Û\¦FöÅ%¥Ê%~_ ±)è3 йƒÅŸrgê* ÚÒ¦±œ(T8”IñÄŽúHôƒ¥”Τfƒ¾[œÊFP¿¹-‘¼¼ù…õcŸÓâž½Itw®Ëô*¡tbø§]R+1dG2Ù Žuå„ß-Ñà©fOõB »ˆˆ5†éÿòd&„ÁÙ¨ólEà´Ûî$ã`s›ë1"Äú÷Öìfü«•_ÓZz¨Ã 23’6îî4ʼnרœ¥çóÖK9s*–Äû°þ È(ÉaÐS:\ú3\:3¸¾3¨ö]8ÝÌ«‰Æ›t…žm6¹ ÇäSÖÈÕ‚ž.õ´&šç££Xì¤E½™Ð§ñའ5…I»Â<ŸpÙüˆÛé®Ý¡79É››”´ü“[ùn¸ÀêsäDÕ©Q7™»ªs¸Íi™˜Õ–$ʆúÓ‰x’*ª¥¸j9´ Zxè’eR³±ñ+åªàu^£¹Q—™…-£ÏÄÍ|LªíZnµŽðDf¡ñ+ˆô{ö‘D}lŠO¢¬E·µðBì!jðé™ M"–·„öapH›m¨©R•tÂãUzy¯kh"Òu¿Ê¾,𒆟5Öç¯¸Š†¾@Oi )Vøˆêvããõ€C:í9Ó_*(£—K1I·ñš»vUŒ¡Æ€MÝÁùL-ïp5æ-@|Jò0’l•»ºá¶ ËÕá‚¥½†&=fêÈÓ9 ÔX#D¡¬Ï•…KžDD¬ëͲÜÖž_ŽWÈZÁU§xIm"xþ†köׄõÎgBóßEºmÅùÈÂÓ°9ZRû1LÍÓ±î{ÄÕhÅ™“Ú’&¥ ž%–pZˆfYeœf¿ž8ëà£#k^~ø€_Ô†ÌS±'°¼Qð…«ýþÓ51M|ÝMdjŒJuµE—Á­’Qz*'ƒmhNLác`w£Ti€¤X%>=\Ù”/[Uw:5ŸÄÚÚŸ?¯:4ù*¿÷r²§"dÄ* °E(‘Pš 轩VÁyÉ'ph6@gýŸ C÷ž*þŸs·I¤¼N3›óœ|½Á§yÐæ{¾ã'\Ѻ÷9uýºÊŸqªŒ³‚úuÇ0'ÏÒƒ&‹Àa CßmáÂ6äÅ 6 ýö49 T¨÷„½kioÅóáž~ZÔÇX>qŽ¢]7vkkÞsÊ? €¤¦yž5ÐY”¿ÁÛcŠLíØ/šQ._©D<*öÙ€&µd¬þ¼9 Í#Âiz"âHm·7Eám±§ª“kã*?½B@‚x2Þáf.§‚.S†‡¢ê²äõpÛï%DþjÔ¦pìÍν©— k¯Þ;VåÁÕ,ÚòibÈ´Lç»­…-û¶ÂEa0*I:(Ô¦M¢üJj…‘û¤ÅW\ÄÞ;T#Úßõðœ@FÝÆäרÖÜtáˆ\fÐ6ØK(ÃE|u±¹D};£/‹µSägšË(û=`(d'âfÿš ¥Â%Œûuí÷,\÷HÏEw((äÍì |y9¶ÕŠâÌì(n™Ú§‚õ?¸Ê¤~‰Ôd+wFE“úÍ,“ÎÌŸs9ÝúŽÇ—BÊB‹b‚â!|R½t)h¢mŠ0Ùƒ×E†žB›Q[á›q˜l“5û;ç8žeû!5AmëÒºª"^®É(W j.sýºI#Rz¸db×|ô¡Aa‚õüÂèBüe„¾±®¤9ðiZÄÎ(­„´éQGh³‰Z þö8Ûj{eßÿÔ.ô5¶8ËpÌ­¦¢šy´÷ q߀‡†£N¦#PM±ªycRýa‡„Ä­ž£"\?!;|ï…xIUúù`œ±nªë?îØÂ¿šÿŽ\­¯y!<ÜR’÷Ùú Ù|«µá Ÿ`Áw½‰ƒ ¹›ð:YqUŸ2¡Š¤Áœ­2ªô„h>+Í@bâן¾®|1Ÿ·1npž§#_ì(Ë…¼·‹_„„ÏÙùÑ:ÿ‚,°ê°~§]=dÖæ)Ðî8zb‚çÐj}h̦›V¼fˆWc&È߸Gm¬nZÃ3oŸ]± ìz4lÒ#±ADÍSãîcºWŸ õ7ÉÒǨ7´Šãf‰T€Ñ¼h¾Rpçîqg ŒÁKV_—Ãt0|yÁîÏúv²=ŸäšiuQ7ïFÞ+ØŸmë%%=â­õÖQÞæÉcž±¨_›¤BÌLD²ñ*:©O¡šº™‘\àú(¸ žGÎÍm„´‚èq”ß û UðÀƒ¼h%’ˆ¶ÄÎ7þj“œÓÞýÉ`ãÈdÓ°Ú&‡V[lCŽá•âçÍf¶Ùæü*ÅDÄ™*…c:o•z'{N–üÂxÄñFܨ_p×zúßUãCÓƒb“›'ÁóƒyGÔ×¶eê™Äsè|¼œr56ù|«­²ägË÷­¼÷è?9Äš)»’ª–ò Ϙ›çBÊ:¯¶êx³®7šUmÚwH !ˆò°j1‹½l1{Z‡–AF;ÀzÞtÖø°óê’Ä<ë¨IGæÐÛ–m1ëÞf“GµJWȼ—LéiºÑK‡œìÎ÷#ÓÏT•,¦Nß–Ù¦¦®l 5Ì%Ü*{úûæ«üX™ ¹I«ð·¤þ tá;˜4žIúôtý¡Æßmh Ý·ærOZV_1øÞFPƒiß÷ŽVг2®Ò·³ú(=ˆÔÙíì&Sϲ˜Ëu%WDbSÌÔFgºI÷$bÃ`˜þ·SÜD;7zÉ4¢Ã!HábñþÀ›¡èCYí_ÎèÏ$¾œmg84¾øi¼§—c-!&ôZšŒ PA7á±ZX¯z-L ²_ŽóMFÄfSU[T¢(d&üýí#ê)9¬ÓöÌâd·¥æ°Bwþ…Û3Ynpc’zRÞ~YEÜ̘,šBŸéÞŠ¦_oÊØ£³f\P½óÀ»ß¾?ÎÔÔe¢(£‚ù…-9ÿün’èwßèÅ7®‚óõ,& ­€…¡üù0ý“PóCMܬðKæ“ôú÷‘F_ :—̾òÕa¤Ãã õ,Ý;ˆu\A·®Îè4 á«~òGŒg¦…„d]Á Düõ‘¢¼a]Ñà(CS/´'7Äm|ZÙOkcUÀ­–)ÀÚÅœ/yeÅâ¬(”4óþ‡þÕ*,mÀ•=à”‚äˆ‚¾ž\©ûŸšÕÑB)T¨¦sÂ’ÛmV°ãý$±eau ”§‰ÿmRú,ë±ÀÞêå@À#Øâƒ8¿¢ƒ‡¬Lçs0k¸¿€ïâçæ]ý„®%g42^À(²1÷Ô2(™Ä½¾¯‹”¼™ý=”jwÅŽoVNM²m¯èô¯Ñ ‡£kù+” {E¾f§ü®äÁLĶcå–¤ÅužêÏÙ²U™u%`ZâD›"IÅË–9®W›-9¬“-2q|¯½¬(s‰²Žnk¿7Î-r9ƒ‰B†–+N ¡YË‚D@=ñí ƒ#ì»›Œ·w¹¯¾*»›R@øv;€çMÆùà’ öfÍv³í»÷ÍË]þRþ})#æ4µ•/J9·T©A£!ƒ°•¬¯íˆ3€cOð²3;- \#½>äx™"¿é-³,Uð#æ ÉDÔÅHyí{¦g1I=Tõr¸âñÇØëœÿe_^c*[}ß¿\sŒXzS0¯†× /ícØ„‚)~F÷-ÄúypHX¿ÜçÁfk –ÀÁà[¾áçHú‚휕ä-k€fL?ЋÆÚmávä(«ogf¦·©tOBZ,ÈK¶)°ÁFˆò~¥ôš’•ædÿoÓ¹Q¢ìŸ*„ŠA]{Ú½lr¢öÆšãrnŽxêì¦Ç{F…õNƒN&‹´,¿•Ÿ¯!ÿÉÍXœõo-v/Y$Ww·ÝAq Cõ/Í›¿21÷À<ì…d`,XaQf‹63 T85£:*’ͨ¥Ejç[_äšÐß㢽(3ÒbJGKfËÖ#ä ÷pÐ4.ÞtÞÎ0ð aHø#×ûbw~”–d'ç)v$ʃ²¢ŽÀÅœélŠ&YU]}5 • sù¹\KØNX"*×qH[Æng>¯Š½þº—VÑžà .S›láŸÎº²êβ§ûT£å8‚@û’³ë/5tcì9z4Ex9”¡ÔëôE5î‡;y±º žQ¼E&rå«Mä™ÿ”j`Љ}ØÝ {~µù¸ç8þN½1´eÓ‹xÌáóZó^…¾cÖ±$ÌÖ*0Þ°e6;°ÝNÈ~È…@ÅiŒÞ²Œ ú1uØ!PvLõê¿M¡Ã±Úü?÷Ê¡›Îã–3éZÙ¤XIáMÚ]ÍYRiÙ“˜”Ø•’„L(‘õ¥U* Ý<0¯É-_ª¼Ò(å€õX`L3,[·'Joã€Ø±û¢ß,À§}sŒžße%—hˆFÊW ½õPý7EÎÉ]hä¿ÔÉŠbåG}Ÿaõ¯ãMßúŒ¨ 2Т™ÉvTËáLV‡¥!6ì+rø7ƒz‚¹`»+Æœ)eÎsÛäŠßÿ@À”1Á‚ì;E”J.ó“ÙÉZgÔçÈ &Y||ÁÁ-®ï·ùø¥¼ûÝüDààXw¶|‰9AŠLüŸ^AÙmqÐV „lŽúÊà=¡  ©ÜiÉ=¢¢L`1àÿ¤ DkÆmÍC[³{–ñ:‹¬y³þºâ×ò˜(—fm¼I×n˜É¼¨š+8àÊ?:ò # x?(Aô(/vÓÀʲ€o ð2ò§‹7©[¼j(ô7Q:Ú2lb$maÚÅH6£2u+›á÷=&Õ=j³OÂHø'zy»UºR×™2=Kò^4}gc àÞ R¸†C,Gˆb_Þ¬Bp҉꾬ý|­¿îMÞ´õçW(,®7"ÐñâŠ÷}º-*Ér ûÂí0.6¥Ù4ágãÝ^n ôô¾yLÅö0˜°Uä7³B.wÞ7AÈ£ú/»h-Ò@}ùÇf™ië…n«Þ ÃG™Ìm)9=¼[ Vá_¶€dÏÇéøÙÝøJà+J“º·©ß¯Îp¾‹qzW^ãj‰4J;C\©ïçÿ}¸g}DCFlÄbk5äfÊÍô¤1+þÃ5«>¿I ŒüÈ­ÀÉmàã W²5ÿO®À• p/Ë®rWiÑÀì]ú¢?öOrý__c*^V2Ѽ;lmÌ@ Y ÌÔ¶MÀ‡|Iep´ÌÉÚŠ<ˆ û†"e‚BTŽ8X»ž‹LžÇ¾´ÇFoúp`¿öRÁ|ÄP1ÒðçCÂÿN!7¶ËxаE£D¹²ƒz¾Ì×ù‚æ¯*䙣‹°Åìmݬfö®ëE6“9°-«þ ÄOPlèOJ…ý}+2:WºÂÃY<;Ë’Â0𹩰+õ×{((ØÒ$ªõ?8uì¹^xî (^§R9~EµZ 5ÂÌíÚ ¸l”ÏS¬ýñ.XcÌl&fÊ.•/Òû‚½{Ü2š OoJv²è+ݶK€HgDPðôõŠZ®úŠ?!ÐR ‰k!$úÇlÌj¥­×D”µn¢¾ÜB„/Øqy޾¾¿-7=+q%‚¢¸'ß§¸gagÛè1<¡<² c“o¹_ojRŠ©ã´'ç%!2²ugÉW$BÇU—ÇÝ™ãµÂÅ} ­ª›¿ÛSœ%&zZÞYLW§öÄg²aRúAÔ¯¦æ›Ï¶2o_1æe­Î×E”S§kV¤ý„c×?€¬­È¤÷6ABü0°¯7"–“¬ô¥ Dœ50K¾P‡°‚ÈÈ8¿ï.hP4®Z·ŠãO§ô8è¦;g½ÈYc‹*/ôüa×F÷‘kY,keeºŸÜ‡?Þܬɸ£·„a.;±£EFÃïk†Dei/d™ûÓ1 Ø·À$à<ŒeíÖžà8ö¦QÛý–Ô–MKügQ¼ÅC2ojRÈ€QM÷,ˆàÿÈÞ âVNZE‚q¿E+2ÙÎýgà”Õ0® Ôì2öб½õ‘‚ß{ЫGmk0wyÕS‰ýï"ÝþùrîQ˜>ç&Dª¯©¬rÊ Ò õþü¼¯1-oâY Ùâãô©š1ûÔÕz;k#“Òà œ[3$úð°rèBF80gGs©&ÔÊNÿ9²“ å‚ëN‚  sŽŽ“ ¢Uí4^,[áç‘QÄ÷ –0|Wkãt®.ã«d æP¶ÕaICq3–.)Æ'¥`¦ÒÇCêgGÍ—úÌ4µEá‘çÒ¬.œb|fsÏhÉQ]~¸9ò EÝ*n’ŸêG×s/ÇO™Æ†•~Ÿ]‘…ëÈN¤$’ðÄ4Gÿ”aºüš%%=e–’3ž 7€WïwJ›p-Ä]ëe»¤ûØ&¨lÇkµ²³Èרܡ úeØ#™â­s08ÉCS… DVNLù÷ÄAÝ5Ý£*v&Ú"4˹‹ÀfsôLWbÐW>S&0`o² }îü1ñäÎ s;Ô7<™ˆg¸I~ñ•òr1æyY?iÖq(âQ%itjžÇ€€©¯"Ì,gSöÆy}¶<¦ù3`…+^ŸFw› GlN)ËÛÕq÷=¢’v›á½zÈÖ)]1ŠÉi¡|˜ÒàÉsíƒ$Uã§æ }F éIŸ-8KÄ _Yv¦S².Xí®LœÚÒ°bɸÝObÐÆÜ5Öû)š<>VøÕÝz{p*FµÍ®*Ip}Ìe‹âƒû•qÿIÑ0€†8Sår__ ‡¤¡7AËz¸ëˆUf %±~µïBÿ@‚+W= •bí™pD5øÿŸü¯.ž ‚ÙÜ0ûòX‹ƒ8t|dz¢ö]W,ëfcJç ½a@ýŒH¦“MKODcšZ$"µTsæœT%Y ¦&ÐÒ¹%Qĸì|Î mãø+ 3¶¼®¼£ê˜yÉ)cþΑùNÖ1,?²Çû§”_버ÊÕWðRоrBÖK¯nÎÐÞ¼ïb¹‘Ð&l¹ÌSéH á²8°6ÄÞä3j#±@+Ë¢¶ ¬™äÛ X¾B *æ…ØßD˜¦ˆÙ5¡¡#* ›Î‰q×ÏWØ‚›ÇJ­’fZ]¥pw½¾Ó‡’ì³uQ)l¹ß[w<À¿sã_›‚ôÝ‘Ånøü°ï7^ó5†×JÝ%jwô𓉷_ìžpì`©ÀÈ"Rº ŒËK_2¯í ÛÇdüàPºe3 ‰© }œQŽ\$`KšßuP[^bB'üµ'è Á&C‘~ÏMçnGÒ ßd@7ïQß—)çwñäªÂ‡ÎWÁã †E{ï&€³Š®Z;\ø‘\01l3ÃÃ=4eE‚Áë”?rÑß•i©>¼¼x"60‡ÄA×LRÂ\/þˆ–>6Ò˜¥ŒpœD(ŠÞéUpwIƒ²0ž!ÚKUtï¸:¿âl{‹‘싚=¨šÅéd«.ý- œÉaq1• ¦×+OŒìak°< ¯d®•zŸ@4þ_ig¼3 qEÛÖ‘ù 0Õºð"¥úÁX‹ú"gZZÚQW-4w‰J×"§ù"xÂìyÆÂ`·žI0˜w¿í¤Ì 5†;¨Þz‡Ÿ0O;åÿÛf…Øâ`=wÏ|>ˆo¶ À.Œkå^µ£þ¦wý˜÷žïUšŠ$œaÆ¢ŠÄœëãàÅ·ÈÇÿT¢ÅBRéfÏ ö+R? xçS%žc›Pø³ß5³è¤Vìf¸ªÅ…Kô%'^‚éÂS¥9ÌÄ’÷¤à|%&ßöŒÛ$#SÚS¢f†p}ÖFhãŒ2­ˆJ¦1½©Këà>HD¡]±8mªB˜vfƶRòöõ¼ð§ ¦CÎî¹§D/&W"rNP&%‰ïG~a˜Fé°ôkÅ76¡É±Ž—&ðߕаøÏÎVƒgFºC|겪œ}7¾5ìCŠ¡AJt„ëÕ<—§ @õù%'@‰yƽQNœäakØsÝ•zÈB£õ""ZæÐXQ¶Â7|Ôjs±´ù1°`^:KÖ²û@â`døù^pó—㸠»æ“S­ãjèê®Ûp®Ãž7cÛ*°kjÞ'…˜Y‘ÒÓŨñ2ÛºÍë(H÷ ›¯ (ø²fµ”²{x—ÀÝ”ìz ʬ2SÑ EVÍò§#㉣'{uz>eÿâܹÃHÀ-a8wT § ÃÊÉa^tÔ¥Jµ9ÔÌçKq'ÂuÜ»À‡Œ]òCL´4ÛçCøÐû U+¥Cø.x'À s¯x«†/1ÔWsy'9Æ.@F™ÓýÞFEÞ Ì¿ælúÝD¸B u$«Í­=ödŽÈ‡)ÚHÆZE…yz­T0ÜÎ劳Íö1½ͤpíw³oá#o+!C±cËØ|+Lx[ë ÚQ`›¸?»2Á'žÿLI&¿óoÎ7ZËÍŸºû-ŸL½j-2Ì>s“š –¡õ¬µã¸Ñ­¦' w}U_‹MÊÏfbß.ä-wíÑ΃Bòs˜ŠM®QÃþ“xÒ 2A$Õ¼S­Fîüž©ª$cDü÷Zd&s›Àꪎ’–º]—A 7'Á^a!þ£~†â öÞíÓ³W1¦… dæ;ªÛö0±kW ûj\Ä£À%àPkv„ÔeY·¹†s·alhsAZª}aQ¯FXq‡ÍÓå±±óþ_F² ÙužÈ­.ž› ¼Ä§Y%|\ù^q,cÈ|"ò™ôD ?x ¦j"v^ÔK%_ © вrC”Ë ÇÚŒÊïÈjQ Rvã~é>%pˆLÕ5Œø©µ,’9B¡ "ZúD­jû²x_¹!Ùÿ+ŸÈß@q¼Fuñ5'«Ÿõ;®ùÍvzùaópÜTAí?–xÍ¢>€JHxÕÏŠèØhñ¾õÄæ~AüvôËšßã!ÜPùè~x{"“ã9`F[1™ÏäÉ…|ðçY4’×+_!ßmï¯ kâ–<ŒÇ-TMø‘æî…,~éeÖ´ €í±?‚JcÑcš4¤‡v K³x®}4e$â·Ükõ¶Ÿþ[Wñ^ÞÂ>ÕѰê‚7o¶¢ ±öé¥K³'gÂx4Â(éP0 e!•oí0³æ²Ç3t×¹¹¹£è.½2_ ëÚ€j?’tÑaœ~шj”¨+ÝèŒKû Òs5ƒœƒ/5“Û±“7Œœ=o}!$ Q¹”–‘ šDŸŸ.aœ0Pã\9ã›`—`|a´Š7Å<’ˉLóý¬Ä).ÃÇ_„¸¸íqÛñH„ÇT£ "‹7w¢ØûCäJ}œÁWiu7ê<ȹ¸i1DÊ.¡ô£¤×Ô.™xG)f†–É3?&_ ‚Ñ¥)ÉȾøç XH ®À¨_˜‚ë%¦õdDé-ÇîD°äN…7NÈ@Y‹Þ„=¼¢Ø«Ê€ƒ#¤¼NGDùVN³U»;sÜ øW˜ðœùÜð»‚.™ìzŪº8b^†® óèô+^“Pjö+d½êæÔ"8ªMµ®Åo†ÎM÷©!¡ÁÛðñSR3ò@îвÎ]¬ÿÕpyÌla²æbNJO$˜*#Ñ<ÎHä6Rvà4B[ x qofWÔf£oÊÕ#™Ço3{{lÆ‘{ˤ\‚&&a¨!D6‹$Öª¹÷ ƒ²?LA_s3*ýZ’î£ÉoÔFNÿаڳ¥ž™®ð`l‘V;ºa a}¾L1z"'Ã4%é‰Õö‘‚C{©®ÖÙëšÿpàÅ`/ØçxVûû0ž€×ì,".ú¤³û¹øÜ¡ÍŒ…0†lvÀX_éu~»7⸆_+ùµA¼fàrþf„f¼ði|ï·˜T äõëï+è%CÓó×âEIÔÛ´r{LðÁ&ß9vŸ¾¥\Î|tNjNãâtÎ÷?|5D<%ÿ¨ªJØ ¨p½~XK¨x&tË<‡äÆ=†]yP"Û43xÇÔЙaSmí Öä›w‹&ì²Qªàê’ÉT!äOˆÎn¼À­$}I2ä:æÍH¾¥àÐbf4³CoE„‚³EŒ¢z;î­»A]PÉðÉ•D¿Ÿ!¤v"Š6Mt:ÕôÕ£2¤Æò§N Þ¯d4¬ýæ::˜ÓOžÚÆtH‡x)‡û cÆÒYŽä©7Ð&qGŽm´ˆEAŽ4®-G>• æ9S|!S ,¶œ¼6÷mgN í"jxvxç­•íV®m°À®4¶@u„ CpAÑ.ÇQ‹‡†T~ÚZæí™W?,•©)ù£+êÉ¢ÛIGH&Ù‹RèŒä×.P]„¿Ñp蔂Á  ¬-˜`éÄöÌ ÜWŽaé,ì$)ÇË5 .!̰(ÏÂ…„“£Ã%­eU:Yz¦Ù £ˆÛ Ý“œh]®ž+=:=¥Ôòyº:ë^BrTkxüZ+ÜAÖ)Ðuñ}=à°H3'0 ñí­UŠÌ¿­ÃS('\ÄÜþm\ëdA÷濇§U„ʱO5Ñg:$1Sñvæú¶ Q–·¦á \ž“¦L+ÚCjó_Ž8îËÈ#5ALµKwÆ–n5¦P+Øq鵄~S—N)ýD+¼OÙXüUœ“ æÿßY_Êñ‹öc‡¦Îô4t>'˜£€ºÑä±Üï3]Ú' 7ìXD)’Sw™“Ùi¡îMK†äNšç›Â»^ñZ°,yL“ýir¦£h”‘ÔÅxÁXYP ¸DTàGö(pÛvM½$‹¦ö¹Y—Ýg1 Ôª6„Û‰6åïSIð[Á<>N$Þ G ¬ž)éU´ýî^¥™*£è=õ®«RK3Ió¹øªß±RK¤ê]Œƒf¸)/:—MªÍâsËâ-¼/EŽ3žó=º¿—EñT©®g,eÇâÁê ®©~e1Ñû¾¬¡Ÿá»ð»#¯0š€Z”„Äñ~”ðY#dʺ–vŒ€q;/ÑxÁ(ꤶŒç¬Ÿóeä„16+¯ºm 3Æ¢òÑ,›׌°¤-?³?üÑlÁ¡ZâžÒ˜3<Õ 5AúÆ[³nž5öq}¢£ÌAåžs ù0)ØM{È6¢Ó©JMs#Ž;›]B4–CPqM®ÏŽ >´Z¾«ý„ ;®÷¡Ì-i}xH Mc=‹Í­’K7L€Ú=0êž\bü%¼g·‰Ó°‰»­’W'ªbq…+ëmË^*Ç|RX·Xáër31ì7¸j~¹ó½n!àJͯYÛÐ`ŽßVY€ôÇ!-9þt+Œ±wÇáŒbfàÄÁ3ý%Ã8ôÞë–ÈÉ `°EJ¥ æ4ö& qWJ5ð”·\Ëk ‡NH qF%cr'yivõdÝWé莜#/œ½›RÈQ8lCB½|Ò?sÏUÍš)F½ >ãóH_Í´›ú„æ!É5±E¯²‡xFn~%ÿ,ÈØÑHéÅÕ°!I(&/ú¡^aûúŒuKmIÑO5ålÙ[ÈM‰¾¿W4Â7˜–­ÛtL2q.™êóºŠ6s~¹Á;àitOÕ£02ƒ³ ·Ç‰hÆRõi‰nTôs§^• ŠWeãʦÿz´:qWè°‡›Ñ›®JRUSvu Ò²ÂË.æ¾õÈPœé¼¶Ûu–@ÜB8mð3V[ƒ\u„æ¾&GBÃJc–f¤DŒ'űû+ñÈ'éˆPY"ïÿ ÕU~«€7òH ò»˜¢Ï¾¢ûÿÁ*JÓû[£ ÷­#·Þ i8ûJ—§³ëªéò?‚5öŽÁ r]}ÕÍE>Å~.~B-fñ«Sgéº~µú 8]p»ßaÈãi¯¹(£HòƒÛlCdSkë­úýŸîݯ07-v +#ãçµ; ¼Ýa`ʹ†¥:ü†g[˜b›Š"õK¬]Œ£hŽ:ÏØþ"¢Š‹çª • ÅL1þ ).^ôEhP<ë …¥AÎ|ÎÍòD#?ÙÈ4¬)!÷¿l÷Àjy›ä-m7s;-2ɾXÂÍR:²¨üaäÍ㤠ÞiQŸ›Öæ0_°ò¡¢\ÒÞi (T7ˬö‰lÚ¦Õn˜ºíO´†VVÌÕqrGÌ»Y㟖†Ô¶”ÌU<ÿɾø‘¼1xjºÿC+~þSO*¦Þ=­‡ùä+º-ü&šÿ[ßÑ‹ÿ$Xø¼¥eŽð5宬I2w'ÂX—ÜÎ]Üêú8BsªõµÊƒŒÜA(ÑÎlµšƒ‰˜‚½ù{>êkfÔ_þ+ÉпÁlwŸ†È;\`Yö`?AsûRœ·mkÜb¦*#Ú”š8ƒJìIDú¢€×Š6>¼O7F*~>"æ¡’ è-åòt¥^Á;ÄWóuÎÛ%êþÁwV üØÚ®r ã+R°ß_\eoäT!M=ÔL¤Ûçüv¤/=--RBĸ,sħϼ}†z?’q’»_——(]ù‘˜A«ú륷™º£KQF]¿„nwõr<ÙèŒ}W³¥í/Ýxg$V ìçÂÞ¤œ3P3e2QÖîi· ìà™Ë}™×8WœŒ×&6”£ñ`’*³Ê6®»‰ûé!÷fW`ûOÿ¼È—•IlD˜#W1Z²´ÛœÉ’)Äýº0õA(eÚÉZÆ£æ¯Ï+v ¬ NÛ¶Ø0xI½8 Ú\˜€5nö%†2)û¼:‘¬:C rÓª ™Œ·E¬c{}¡ý‰Ó Öj¡I›¢C4¼þ;ÿg;[>w¹KÓˆêwTØ 0U’ƒª8T~Q©OÙûå÷{KÙÓDZ,÷'p-Óãà2ûP5ÐÞ¹H<´·:Ÿ²Á}¡ƒa/¿£Š»°L‰€®¹lglw\⸎×7æ²øp äšâ1X“£I›ÏôO+Í{¶’­rÇÐo“R²**þªü¾ªo#~û!ü‰r<¬DO· )€FdîSxxÈ0€‡ÿªª­‚ˆ´©áoF>€ ^[¹,ô”b» ﳯ›Õž$F˜4Æpt¼tÖÝù}N†ÂDNûà±y#î¼Æ‡\“öú9‚„;—o%lFõ·½@×}ó@™>ë“VºnÊnRòÁdúA:x]ÝŸ)2I Œî³dñ*N0ƒ²ã9 :_ThÁ„¸Z¹¬²»þÙ² ÛÔËÊv„Ìu¡É`WÍTPц:LC‰jC›±’ÆY¨üÖ‚yÂÀ}‡CZ ‹ªv‹½|øÝS -=ƾw4±ý]YºÝFOÛ8x‡WœMáž÷3öeËYBi(µJ ™ j!j¼B¸Û-Ú-•–éÏ!k`:MR.½í Ž’ºÐsE¤àE­zëƒÙäÍþ/÷Äé™Ñ©òã:­ô\󥿇'ð:'XJ"0Çeüê} ‚OŠP‚KùY·…°] jÁÌ—–ž)ƒÌg–@^2jªo·°`Ý7ŠúPJ‘•àöÇHöåÀä·• œÑ÷ eö@QZgV‰¸C¿7Ð$ûo}˲JýÀ‰H6Böe¼KŽ™ ¤ ö§Çø0 Ð%òRŒRëáwÌ¿i(I‰_Oª˜0èŸÆøN ÁŒ¼ ¸ž_øƒ¶ƒ=ÀDÜ^5¡$ß Fv¾ ¦_ÞÁ#»ûúÃ?”¦ñP!’ªÈjh¡H®z $ Ž%GŸl¾8­!Bñ¥¹»ƒ•è`ÒÑ_Îå6¶§¿¯oça«ŽÞ+ÙqߤƒÿîtT[ÝgÙ˜NCPŒFD ,)ªpÛÝ8Ö ̱}ûáJÈÌ…M‰bë+GYçÀÏÿ€~K}W„1«A¦Y¶1;r!1t€H@ ¥àƒÈ8bS…ô©Éس1‘–SGKÚÀÓ(º»Ä>R6¥ïx1ÓJõée;fÄêÊ[ÌJ8ÇÕ®ÂY¾A-ïwÞjOJÆXN]°æ@Ã`Œö<0á!ša|D©' g ÿœðHÛC×§øŒ6®óèîx{Àø?ÉETv1–j¨vÙ®Ÿ†øçZW ç+Äv>d1&µÂ°hƒC—ä Ðuþ*köŽ*ÞT…5œÚ$ 6Vt騤Á·ÊUËËeÙ‘J ö7$ãÛß‚ô"Nw]̉¬ËïÕÕâ[1}¨Q»×ö# ^4=G¾"Óžôžç’Ñ4üRèérê4œ×-¢ƒìùQÈÏ>»Ô”»¤25ÇcC:ûA¿ Q4N¯ÃÀqcáÈÜéˆäw–ý„ZXŽ™‘¼ÜÔ·ª{»v2ãb’ñíµ1+‚-2A¤®e}G—úá)MPŠEv–ÞnÊRöÂÙBÝmº÷\ÏGXš’ë_a$âž).nþ oÉ|HàÏBÊvÜ1ñŒ¹&¹Ð÷Û¬çà/¤ßwb4¶APHYî‡YM„,ÑôZ/Âåòˆ'Y¾ÂË’iÂô¶rC{§ ö1&ðB€Wiú×cX¥½`4œ7­’G›Š¢fdÒF#\ú;ʦ\ÏPkñë/iÓ©s¯ÑË €b@?²¥ƒpPx² m›„MŸ~‚ã¹r¸7ÀèTÚØÖZ`A·¯´I»õ‚‹óµ—™ ‰e¼ïÍ?€Ï׃U$OÆܼÞèt¬g-Üô}¹OûPbEÃ{¼¢F¦KõÒÎ|Ÿ"_£YXÔ|qÑL¼U¢q —×¿Ž©£&DsÊN¯Œ|u TŸÞ`I™cº"å>‚Ó`s‘Ì'êfΉ¾Âß?ŒûçŽ=½AghnikäÑŽÐdŽ¥¹:‘ƒ _³¾oÓ¶ =¾Û?ü:Vª˜—{¢Ó‡ŸÕؽþV ÝFçº:åeÑPž$BTdalûT! ±aò§M÷“n¥UêÎèïIج%eaÛ˜.Ü)LZ/bÀA,ªÂjÕžìÜóÞ2Ç"öpx¾G1,Få­X°’Ð$2}x‘´“'ŒcnýÖàk‘N0j÷QÐèÑÒÏ*‚Õf»zO—æøìMhŠÿ`B"¬ù¾¾Š°BôÐ9Û¶5H‡YëØ ±}X2HÖ tЇìZûøß|1Ý_åkÑkh/ÝäWV/“_O©¦"–+0srþ8SEÙ³ðKáöû–šïÌû‹¨õ—`@Fߨ¦HÉÞh÷Lƒ!^fô(ÍÛ>EQV&œ¾D͇AÓ†è/ζä†5ºYíŒ\°*a´õU7ý^j‰ÂÙ[q7MÐp}ûøøÓÌG ݧY ÷ò3ä>h§˜ÓOØŸÈ ônÂØ€¡ Ì´ l„ö¬Ì$Rä•Á«ú}¡52^˺D€t¶Mŵ"gÒ¥ÕÌU$„ÓÚ ‹ÍD Äg×Õ–lí®2Èù€_º¼C™uúÀk¨8é¹Ä¸ÿÞÅÇ Õ†­k?W¶¡›^ϫ鯈¬2â+®§( Û½LARw»k‹W„îæf ÓÂlZÎyfð| •IÉTÞàjŒFöã¤û-`tB"YãE o! ªç˜tÆI?M„´w>̤¢šm™YóŠײL€á/¹±j‚ý·šh²]³.’²vÒ°„ò ­³ÑL6q4|¨ÁÍJù¼¬%æ2 ­wñëÐNƒD/4=ˆÎÞtª?w‘EÅDºKÔ2bmÙýë“ݘÉC‰D/~hëxPgqþ# ê˜Ü®®®üÞ;]3ºüL|àÍò]€ªès[ôˆB¹ðåÓà†<‘ßpWdB!ƒ¬K~xÂψŽô]PE*“ÆçÏ"—>æXqÏ¥1AÀ@¸­J1˜oJ¦SÍ„ióN æ743±3 Y \Ñ ß±9@oÖ›ky3o\ x<Íɦ×2Ÿþ}”¡<<ò ¡.¸¨¼&Z ü‹\µ,UqgŸåE™©?îÇ ¾IÚ…%"I¢G3Biq¬;cÅC‰±§û¦*)%±Ø]²jåJõå€1¤ þþÛv …eSõDº'ˆÈ(Çÿ,ö¹5Ý„9&áu§ìnƒ>šö¡9ãA·‘ •² =/ÿ«¥<\ÝqHoñ˜æT%Æšµ)YP„eN/£3 0ßÚîm€;dö4â„]Oûö ÀªÈ­¦%Ð ²ƒ7î”91-ºx6–o¸P£µïQ5F í«dÛEÛÈžn Æf§»š´n+Ï®)ß/yïP4‘ŒÙ¡˜48bí+t2¤Á{Š#Ùžsd¸¿:Í]דè¿bÉÓŸÊväÏdp ^Û~ Äü÷ìÇÄ „wC iõ~Œy%GÌ:a¯´™{”üXøŽ™èFÉé‡ñDtòƒï’>jŠá0ñºäÃ’熓n±‰•mC@­ü&Ç-5BÉ8)² Aä ²Ìàt¬lÙ#›ù(Ü?¯ûC×KÒ@¨&a]wǔǤӅe§­ôR^èJöºo¡:!¥@¨è?ì×Ã!˜ê­¢ú¤¢à$ K¬šŒÍë‹§«†¹ù§s¡8—¥0ü‰O ƒr…|pø;ºÔu‰Ùû}_a:]—Lä«QøºÜ³Òók6zà–‘XÂ4ÊŽä~`Wtƒ»µ´ <…_©zôHKd0NC?ÇWŒpòæ¹)6~•ÿÇ’(Q¡C×ðÏU† ¸,ôÂcìl8€[.*ŒÌ³*M0I9/ˆ+r3ºR€àbâTŽ2z]õiÎF¤[…J¼F’úp鋳ÂÉ›sž¹ièZ²šš“?Nö/&=Œ©I0y9ê:ÒIä_~Û€îŸÀ–°Ÿ}Í0ÚŸnÉf†Þ¿:/ú‚@C-N’‡‹í©‹ó©DLÔlm§¨)_ø‹‘´K¥æŒ Òîï\˜0Ôu®IÇ- ø™t†{‰“ËÖíºßcát‰6N@D g(MbÅî‘J¶Ò¤JT ‘r\|Vë±|µ4³Ø+eó¯³Ò;ÜW¯vq.Ñ#ºŠ1Ï 'Ö‹¤p¯±Z8ÂÀŒÎ%ZÅLÄ™*¨¾\A‹ñW0•Øz|Iu÷62›aj´\3…‹â=¦„ñnËÕ,0gª˜n Ö=œV9hk{ŒH ¿y/ìFRÿN‰ð'u"MçKÃÙ·•ð)®€ŸCw:Sð›tô0ÆhÅóH©¤I³õÊ•²ÜÜ<¯Â`½¿ÐSÕJ`F>…qL×&YÑ•’ô%SµêÈïæ™ábU·_r‘g¦ ´¦Ør1@…‘€\t·ß`î÷8y'¯ËCVËfk­Ñ­ Wøå&2~â¨07»Ó~i†ªóÓô”±ÖH¢dÎĶUábÝWâfÎ m¦Smî jÖš1ðʘ±¤ŸÞ…=-3SQ¨Ë!ª"§¬Úêʧ¾g”`«Ø¦…z†¿.”Nk‚Q…ÔÖ¾1x©Éù¬Î……$íÑ/a7õTÛÁÿÃâï’Ù,àžý(™¸ {vŒËY5Õv˜e(Z§1h¯““‡$ÈR56d¹3é­^§Ù–Š…¸[&1»€¡ö¥Á…mÚYhÇ( ”1Gk“í}¥©×ʼˆä&‘Ÿq:$èm›è™Ò'¶*‘ ÛO"õYk¦=ˆ”2è·“õ€*µS_–alÌTÝù卑,×/ùF v:c¼K>ì»ßãÂ'…“‰Á´k¥!$¹·5Ã{ß7Ú ›ùt"Qö>°~z{zá-|Oºjëõ%èMNćÏ^ÀPëïÖ³ÖÖ¤¢Œ›ö–×±ÓT‚-¾Ìu•¸èÂí ÕáÂêYáX{ Å}Ê›yiÅÅ*+ð3è¨tiË6ËkÈNeÚÚ“3¦aà ºNN=w‹òLÊX^ÝÊÆâ`yûÍÕM?Û][fõêy:FTY‘DîPPg°è¡C;ÝWüA®1®„97ôÝÞÔ t/®ÅŒ´}=ÿ<®Ñ,êØBÍi/C&O!b»’rºÅu ù%7•*/ÁÝRª&°r­¶È¡Ë! ”á¼-5((–˜Ô­Ox=§L¥›Uˆš™\ìKÚ̃ÓÐh‘B—Ý`—9¶ö ¬]BI€¢È#X}K» $aJ˜™‡ ÐQYð>ÆZÖ(ì$;$-w8#¬½ïi1“+}= ‹8·èiά &à†‰ºï;­QI¦x©We@Xß»LŽßG¶èûÜÿµËgÆ`e¸ªÌ[Έl¥õg ]Ýo8œf¼iònUuèçÍù£Óv•·ÈÛòJr]äü¹q/ÏŒÐaŒ_õYÏÔ=ú¤©ÁSÎÓ´Zúw¸Æ@ÔäÞ¯ ^ ¢–ÊÔz^Á9f“½'u©©Ujà6áË\à£|”%ɟ̸YÆû-Ðci¬ö\´5¦ËDƒâc‰/ngÞÖß/Dï?DK&CçL ñ9Õ3ê!±«UÖ¬`M€‚?¯Mó9CôoÀÔ—`4öõS•Q(³})–ÕLÅÛ>äv²6󘚫™ËëKÊVs38· 布Îâ£æ">ã㣲œW}Ýz•Øhôçù~É£üëq²Íõýê•NyÓñ&èEÄ9Ì*÷½ÿÌèwì™÷†?º;˜Éšýë¢[¯GÓð ÷ÝQª©78µñÐß|¦-B¿¬'€[ûªì²›ÒD- Z~àê•Ös up위LŒ ìYĵTJÌçýv¢Dó)a.FsÑjî^xÌR‹àšâ’@I2M6HĬ>"¿àõ—-ÀÐIúF¿û%Z„?û.äL7 -?–ºSZz£€¢„ˆ$0Yp0DRñËà5–•d@M¥ÔÙ?›WYÛÁ̳ŠKlY›k¿0µÄW[:`ÝOó ·XÐ °(¦_&s‰xQ'ÆVÓ »yÔÊa£羆+LÛɬ0³Û–Þ[›Æc/¾‚èí߯À;Dyó—Ïó´ùÖSoÅ! CõHºLtƒ[Üÿˆ{ç3€CϹ]ÿc°nË]Don¬î cÐö}0•1'ÔÙ©›k¤V¯#FGÇbÑôÈÊJ×Ïrsà·z…ƒcü®OB6B]xVòí¤6ÖÓŸ"jw¥ÚŸg|Ü !Ô¾j³,hÛn‹×o€j X)eå™;ç*HvÂÉ—AÀ6ûALÚÏ@ÙV •ÁU7PåNµ*wwó°Uâ7Û¢©% p ]jÄÁñœ‚:+ã<œ_Eëípí»äÇèAuÛø®cŸGèd´ÅjÛ%³ÇL<)N–œ³CYOñ­à¾·¶îm¤<^`ü;+WN‚U¬¿û$›ƒ$üTˆDXŒï¯ª85Nbè÷‰r,ˆE8œ„luÝP­€¥>ívm™7Á+´à1š¿`4EN³*|{Ñ&WŸÃ–èLøßˆ±MâKÕ³ÓRÆJ%’nZ“ùòÝ‘Æ «Ru´´)|Y…pÝÌ¢$N³•y·—'w¹s¹®‰ ~áò{ œ‰žáÞ«jSiHš°>ÐAÍBmpø~G·(í¹od—Ææ´%oHœj2{‡2B¹©<¬Í%k5æAióÅ•ö¥ƒÛ;Y‡¸távÜè¿X9ž7¿¡¡ƒ2Sj?ì3°ÆðÌŒ}¿RIµRºö­î ´?TÓØƒöEÚ‘ÖÅ„¡‰ÌmÜòg‹Ù«…€°)”Z¡W`9r9½b± ¯„((.?k!`©ûåý+ßZsåcTò9Nïã$Ÿ­<\ ™ª-Pí`Y}o ¾Ø’…¤‚ˆžÞH±ÜZœ×§…KVú“ }€ d™†¥Ü¶­¿švôk¼£Ô}ƒÅ²DôX*ú´0ïõ•ÓµWý$ ²<\§vôÁå7äC^ÊäÞ üSÛÈ$ϱ“Èô1Àúÿ<±9^óŒçpyÞ‡­š¢¥àÃhƒû™i>íTb<ìÏt §8‚V‰Šo&Ò/ýOþÑ¥-´¿å|m˜iy늵—ÿ°+¾Ø$µÑTFH_àÚ9"âW‡5ÝÕ–erA*ho4nï«H¹ >Jïm³’Ã8Dþóz⣂–vÎï°1×FÉñw¸Ço—zHŸøÀ"»ˆn'¸üù‹‡ÒÔ߃BT£7¬ßQìÉò®Xšçûžiƒ*Å6Üá0ßáXkÂÃQ|!·SºÑ0}„Y‚ÿ`‘hmú¢™ëNË«>MÀº®×…i1¥ÙFQY ³ó ¯,ƒŽúxâ‹ „£™ƒHwƒáá¤#­ýáе:L,Uó Oí5×ÄèÛ¾òŽÐNDÔÈæ\òs(éú¥>tY_Óhz“ÿô± ’|m¼Ï‹©° U¦1¾C¡ºèçŒöL-ðu¯Ä)LTÜá?÷åSøc>¹è$Ùèù;é¦^ž Ä&RÞ”ûtH‡¼9q96Ážº,c8gËm½Ÿ«ðä?æð›E5Lòa7ò Î˜£ ¶ =!ÝMæP;×á<1N•ØtÃ42úÝ,‰GEÎȳàJŒ Xú2š†ÆÚØÔÛAüñŒÊ–¹’V ,(úx#þÙ¸[ Á'6gE ?´‰ÊBŽ`˼²ë –û?¨A-cu\ë5.þøæƒVfjÂø7ÓÙß;ƒ[ZbD+s”`KH»ÑÜCÔË#ž%ëÔî•-ˆ»í¶÷´PÑ ; <Í{o¯í³I˜Žt}|ùÑ TQ–ëñHOºqnÔ Ï…Ýxîé+rúç¿û.ÒI?“ëDuVL…©˜œ>¦`Fr…ÈÿþÀÔ”QláÊkMÌ~ó :X6䦖ÍÕ@|‹ Žhâ«4_žus!„“Ò¹ÕOõàþø¡‰T±ä2)µM•~­…Ú”0BS#÷¤-àËQñ—5b`“ø¸^S§è•]Ì$ŸhQôOóB ·dhÝzDÑæ÷n(S{SFKæ âŸwÊh]óo‰7òñîA’YGŒ‚Ƚ¯íHdƒxA0þ3À¾?DZ.‰Žæ[&Q€ `ö3“Ê ð¹ÙRH=g¿Óqû+†®,y=>:7]â寶wà¨Mtzœê+Ø~{[LÍ„vGô„Š^ÛŽ ¢|›i ƒ/#z=ò”›¤M/b|wŽ¿#)ljyóúY¸Sƒ>v›+2Á|Ë¡/¶óþ Ž‚j¡q}‘‚*„‡}åÝç?ˆ%í)]ýhZ¸ôeÍïzi”61¡UÍÜUxt–ü%N·c ËyPŠðjjh¥ q:$¢Œ§Ž§{MEľ%­²§ÿ…ŽøN¿Db+;VÈ£žì†®xt~˜º@üÄ’$»Ïî¶±¸JPo›îuA¸3—äËð¢‰P7œŠ’c© ˜½¿HTÓåaMÃñâ¾J¸ÞowöôC­³n/‡ÙöÀ…ÆÕx?™Ú{1qp2LAKÀ§w€+2@Â/gF›†æq[å©­?ßz¸ºÒ$X¤Ç_¸‹ö°‰{u ÷¦«r7Â%›ºñyá>÷Vö—OnR• >Ö/`…Ÿó‡LNCÿñŸ8VZº®ÝÊœ½¨½,.›{ƒ9­G¸°®±—fpþ¡?"V®`W­›ï“1 FÀÚØPìWåUN®¼,åÖL•ûÖÛpŽ¢å¾^ŒÃˆ¿ ^Wÿ·í0lÑLž™ãÍ#)´Ÿ7oû·âf[I*(ÇÆeLe(?Ù¨â\S¬¨´ZÏ Ó< ÀBöt'¸¯«·jãŸOÅ'›i$0éùv5Y¸Q’7 ¼4u2pÌSk·)Bsþæ}$Ë,b… ùžÐ噹`ô*OÏï~Âv~_»¶/ñv*„àǼ|€h-&õbzxÉîí:ë2ê¿þ\õ[•Þc>¿‡3[€Ù}Ì]ƒàã  ÃH6p°ð<#3ÔPuÒž>F¿ 2Þö晜tÀ¡o‰*àÁÛÛ•I6yûœ4¬¤¯r+$ÇÞµv §ÌÿI6`„à›g]ÛËøâ]sL6ÏJ‹¾Îª~›ý$ÕIæù=ͽþùÖÃŒÎ\D*û­Å™_\ß…ø}¼ >!Ä5»í¾¥Zá|ìË6PŸHùEBàF4yÓ•ƒ&UȯqÀœlÙEÏã%s­„OEo§‡›¨TXó› f1,8ˆÇÞ;ïJŠì½7Çf•C—[áÅmcG;>C‘|ÕÁZ÷óê0)§ 9s¿(={C¨§ª¼vþtd7zûÒ{@ª cÚ•°Ü¥È£xHÛZŠ:×»EÖý h6Y…^ªÝNA9«¿¶²û‘åg—ðwÛèó§{ÃÛ8«}³eÁ¡n'¾ØÈø³+á•)PÜòL)ŽEÉY§, Î8V‹€'- F/Œåz þQPPÝ'¶øð÷P§p”wÇCJ¯Æ)ÍK/ ʵòÛ³C/Ùn“$ §ôqŒ¢2ã¼½aʯì§y…zN ‰ 3ž£KK%#ÀÀR¥·†Ãq¿…RƒuϘw×¢ÍGY„:OZÒXœÃ!¥!’ùà2c 2¶õçöcØ õ»·#*Éͼ®Jdv YÜe¹"Óü|òáoo˜6Ú”¼'…Êaç^OSNÆÓ´ËÖ%lTúpIÝü&Ñ©}5S†ÇÏ–cáw«Úþ™Ž"»Rf9¥-‘Ý俣!i¸U ‘|ubóÂ=\³Áí®Hâžò~†¦ò.»æãü éZÅu8ðWñá&J´~´ül'ftðb°ã¦ê°ÒÇüX”±A‡îýTfaWb•ÐlYúìóB9ßß9³ à…’oì/‰,>vs}×NÁ‡€-±ÊF™A¸ésýkh¨°Q²û ´TÛ1ÁÂ~š¯üVR»o­èÝJ§ˆLÁ‘›ëÅô8T¥˜"ËšES ‘ÇóÍÜ&±2a¿uEÆvlv9ohN+ýºé9ëÏK—ºtž>YJžuæ±õÇ}#tŸ“Ñéɹ*­ª)óf³'åncÜôžeÿ};BááiÔ† >-Z}5–j,ðîÔ¥é-Í¢M-`XTå‰I/Û#¥n¦·]磋إnÉѦe-ü|{¥Ö³B4 ÙsÂÄms…S¼cE=]ŸÇûàú—ý?Ô‰¬XGO‰¼WÉO þ“Ô"’+ÆÈäÈæL}G!xŸJÏ·ÖéWæ·ŠŸ%Rœå³„ºÚ‘1eÌlH¨ß`´ÞS¾vŸýüùÄöûð°ˆÙíšëdÝÏæh†Ö Ø÷ÔNâ_iÄKÇÝ´Øm®º@ûôÌ!$N7¥G{ÅööØ "µ˜i*Ô!ëì®Ã3ÃLš«W:º  Þÿy§“Ž™Á)É•–M÷þè.¨´?á´xaué ×Üü±/Á‰…¸¥µHŠÓÊ´„qìÓÐ-–5ߎX;6õšMó-Ó¢•‘ŠT0j@TÚIᑼtß„pTî÷vŸª:fWû›tή£'»,äµäëÉ'H¹ñÎè›5äÍžÄí›ìF–·s #»«ú˜-^ù´’e·^^:Ö¼a6Ú˜]ªu ŒÉ :M½Mkˆw­‹ã5–>ÇE'‡º5¥f³Ó¦àø[qÆ!Ì*ÜëE ç-̧O¦R¸°g^Ѓc ÷5}HµÝ´«vräûxŒj¤yÕ)Ïœ¤ 9Â}í¾y†K’2Ym\fãÚ™R1s)§Ì¹/eþÙ)4SÖßp8B¿ªUT¶Cï1#°yÌž²äïÿÒÒ…Km¾ÕñþFIÝlW p³oŸá…’pÍ W¸³c\F<5®=ÞûãO‘ÅPif÷ÿÌ÷iãd ݶZ—µõÄ*ŸÏÊÍ.×÷¶ }U ¸õÿþV1¯LT#ÆwE=…@š›1ñ@º=›,>/¼¥A "s³éI¡[êpÆÓ'»üwxsh4(šQ².›^WÏ® ËûSDÕ„R9žŽbE©r)‹Ãɉ—¥¨ä¼Ïžß’‡…`qƒt³e¹:䔊ÊÚF´¼¤«i¬2EVw¿`Ð F<Ö4ëJ[nÉŸ2³™Ö•6ª€u|4ŒžÌ§[1d²–ó]>ùu ã¹WOÙY¸»;ìá¶z10†&’vƇrÑ®B{é€i´]/“Ò‡†y hën¹±µÀX=qMêßáìêÙuzÅ-ÄÙÔÔ9ÿPk~cxÏåû2Ž:âéÁh*ª b;­ÅMÌIo©üž>¥Ý¤°>Ϻ’ÚªcÌÙÖôþáa·OŠ¿¡ORÏG×@nG@ ½cÂÉÉÓ9©¸Ûå¶ì¸OŒ‡=c%-Ñ%F~æ®G lÅ–½?œ…ìô€¯þ©Û­„£‘¤é®3‡µ}Ý8ªÑ;âöüB#ãíÙvÒ=B]m ÷Î1‹¹:Õ%ç…øúÐÆË’áŸ3¿Æ…±8>Ÿê˜çÅá¢%ZgÂûDIrSB¿>­(‚sð9GÈ%WqRÐb¿ æ%ç@%`pÁypƒØö°»Ù`æ.òM¹È^pQàeÛ‹´''°´´ÈÛûëâî¼²“uŽšÖ`ùJ¯)Ë›1k?\¿Á[c"ÆF72L`ö£h4ö:ØŠm·9ÌÙ˜PŽ2§:Š‹sí¡ª¹¢¬í¡‹i‰Ù5d_UµHþ<‚ °’ʾtÂN™ÿt¥D + õÁœ´Š„˜r [ÒÈbdr%³ÓŽÃ!6¾EŽÍSd‰ŠÏ¢4ÿó6ó³7ˆA×’eÂLÒý}{gÑí»eR¿ùf€PO€aY^ëa¿äë·€¸/`y%€ ŠzÅ@—@‰Ê N^b#¥¹ÚÐI1kâ4YF¬y>ä§.†èi! Æ2͆ ÷?YwÔ¶ÅÇå9õ¼YÊû†;bÚnE.w—€ºUܤÂA¡Î3ë„…ÍPÐmg$BsÖ¥Hç4-š›‰Ó‚O`]‚Ì­7!*}¨n½À½’óCïÃí‹ÔDXázÜvG£ê±fõ3†e¸¶Œ m¥©:Õlx|œH„crÏõÚFþš^úøzÈ)›|vÎ/PøÙ¥,—çæÈt¾ñy£ ¨P·Î[®B—Ò+ýÕC]dFMÑ3ÚÚÉ‘B¢»‹Û FÝŽ1>ÞAÛñ•ï„7)Pžœâ€pË õ¢Dù 2XܯmÄÀºQíîiÈ”X¤ÐX°¿%ë/1Ò1í³¬–€µ¤Ir±tzè¼³[¥Dð’KK¿si¹;ö‡j[¡îЛÞúô²2×Û¹uáR3*¾ûc|š*Ë„3È!ø{N‚>sgûß½‹Q7´f‘©i¸EpM!¿…V¤ Bœð’€µ0t·Å~ Šh¬I­fÿ‹¿¸D»æÒŸü’ĆЉ­JÙtä«r„NöDŠÆh‡÷S®íA¯ã¿WarW™A0-¦ò.kð$ýN@φˆÛ‡£˜kÛüŒñö¾F¬Úòí÷Ó¦Ö.ÍØúú×(SáLï€'úóYD¼õ_¨O: àÝ&JŽÒ¤6ëá‰D;¨žbG¶ÑŠÕXníø)Jõ‘¨¸Ñ)Úš8$ä/PP´±…¿ó§CGH‚øã“6ƒ´. …s y‹+Ün/&ü¥õ¼XX“–²ƒkŸ€Wœ±}0¯sCÛ'È¿@žÄY»}Lÿ-?ã\‚B…³û—¿k#ÿgÀW¥l‰0QF—ÊðÇ@pƒyì\Áœf/w8ñiôÍ`è¶ð˜{q"Có/¤å©ÿùÔå+î¶Z=©âÚ¼¬†ŒIú;Xâþ†G“ºéj,mób[ÉIÃÌ<ç½Û:j' T £P†¾.?Ôˆ5 ¹7þ—¤ØR¬2±0ÉФsm<·íû,©ÒŽ"ï¶1ˆ64J¦p-sfú|—1%¬IYW§[…0Ûã1 ÃíUèø ùhÿSv×`¢ºvp”#Zñ¼¥ Ç”f_¾ºÁ›Ý9 ~¿Ý ×JŸ°½È䉃\W¶O˜qêTÂ:ˆEîB^_.“K›ŽT„³Ò7Fÿ“Ä^Ôì—Æ†·Ã[ÿP‚ zÇ%ãœ3¨{ý½d,»²‡KØZPdßñ Ò|lº•<ª¿Mlá% -T¹QuÝeÓ÷=Õ´ï£ÖQq_f%EcuŒT9ýöÄ×!Ë„–ã-v"T–[&P É¿ælœð¯s7ŽŸEÓemâÅ,`ˆraÆ…¡Ù² ½%—üå‹%ü»™ËngãXS°öþ‰H, ¼÷ÉkÃçµ;õBºÉX&zê”g)ÆÑžýM ½´Ã8`Ø„„|8q`~—‘Ô –‡ÇN%–Ûnge¦m^{£ÆŒm 1q&éu¶ ºÐ4‚O¾@æmD(z ~ªwçAbcÂ(Ñ ”,=#.Çߦ¬1‡0B˜ÌíXüù5Xd>ª³   ñyn꽸HN«ËÝåòDR#5æR»„äTÝ02‘µÛÎkæ©H‰D§Fâϱ4þ0õyÖ€h¢¡žÿCÙë­4\ä'Ýçr­¿»ü™³ôI ¬>™—¶¼|Y4ïÀl[H¥NûWÁz½þ#>±Ž¤³ýtîˆFí4Î<]‡ Í=SŽ7†M8ì÷Žæ»€b:ä³ÁgZ?|“…ÏfTáU%Çõƒ£­"~\VófÂȇp”ž óXŸÜ%=îч .(#U2ÁN(5gÿ¿cò¬…ÏÀ;Ë1ZCF"ƒÜ”=…S}<GSV1ø å°8C’n÷:Ř&̯â‹öõį7íª-è=èJ£Ž0³0‚4 BýϨzÀ³â …B ÈÏ*ž¢5ʾ_ûìáx(eÆÐMQPæG"xK8òÕ¥_Å9bUZât•þÙu^Ìúy#s\E€Ûû@ÔõÊÂûYžwïÈ1·05k>Q—·¹·°ãÙ‚ZÖSk1?‡èíÂc ¸5QvØ|3q'å`°TÀx-}Ç,šXº"­ƒ]m’ë¬WÊ“sjgg¾_9ÚsÇ)3›‰/Œ9SùÚl‡)È äZ0ôê‹SE¬½t*¬”%?Þ7'𼩅ÚBH˜¿Óõh5ª¬qá…[ɾH lŸÆÕïÔ6NÕØ¿ïº¶"›€r¡Äú€3Bh×MÇ/ÐÏÓmQé›Õm3ò;”JãÐ`ìúã~¨T(ÊË“%-ÐR…Ñû´Î¦oöƒ"TMœ¨šÞñû†>Ë <˜È> ÕV-Ül*e¬yz˜‘)EuRÖ ½Ùð„Ú¦J¶c=Gƒq€cÀŠLÓ"Hì†(ø:Ãf*Ä|,o¦€à íØ6¦‡L±GdÂjV+c׺YgäÙ‰ÚGY±$ØííbG_<Îò‡mµÂrÍRèƒ|nw’ub.AѸUEßÑìY)kœs| N>¦ûà]úýRê-åo¹€Ø´ø3m=³ØGX-FÊc­ÔìÅ‚$!Ö'‚ÕŽÒDXS ð1ÍŸE¤µ¬Z÷‚ÔæÔ]ÀEkGñ‹µ¥¯<ÏGpÞË5û(Hãè¹é^/[7Ä-‰ì"A}òåKã© ;]‘:Ü„'ÖÍB[tHÐÞÿ>ƒ4Ž«LMdxp‚ê^ÚvÄ©Ã1tvŒ¥ÿ²Ùøæ,ÀÆÆ0|]|ÿ±€úFq§¶Ô¥$…do8Ž%ƒùšý AAtÐ^’–=Oo@>RI"wöÎWh»c°ë²¶ê!H“¶gðDÚÛ$8qÂc_x³  O-úõ£p„2 ¼ƒ}g ÆçK S«çƒÙ8Ö-m ~MÞ%`„H!p*¤“‰úÓRcJ­o§|îڛݚR"ÜÁ™¥·¾A‡‘{„9+8·IÒnú™Üøô4NùàsɃ¤1á×oݲ²uÅöa‘ýU‚Y9æætN¤Mž|MÞTò ãŠÛ$Ž–ÿ•¨ƒÈIÆÄ)Èï4/uÂ[lO¡â0óIÍ^óË4%›ˆUïÞºÁ1=Þ‡82øÏG™¡h••?½ÛÓó}Ûô_’ÈŒG…Ø ”Ü ­ß\'>Ðî3&õ “Úž¢#7Î&]D;µzʼnŽLZñË–úöwUDò‡ð¥sR*ŽÙŸ3Ë5½ë?†ô”MTç^3š¯m6âÝøÜjÓz'ò©]rˆ§B¬˜ðÅ«{ƒÓɬ¼8`«_˜®s4ð¹ÚØj@šˆ×J•Î;ä-º ¿ÃÂIG¸õÁNµ½»›»[D‹€!L`Wâjû·oUJlÖcwŸ¯œ»º»±ïä%ï‚TÊ‹é?X½'ÑE«ÅºS‹IÄÍo`Ø«=&Q\Vû¾øï‰ÆÓD_ÑÛÿGÏBbGvãoÍi1¢”_Ï©dÿ1€¥ië~üøìþ0G¿ô:ÿçξj î$ªÚrùygw“=zZ;à2(¤H6f¯lÞPÅ›žüÂ2eRZáU¶·l\øPÀÞÝŽy–„rà{O.1kÿ6VµÍ®Îy5±ñõP6,êÝÕ”áj»ú6…ÇwäˆÃE=^ÑCF3DJ;Ô.ÐD·òõ‚?ùEãoi¼ÖµÄYC2°sÈ"w$F½—ÔÈ—ØZ¦²q5I·þÙ\ÜÍh}""ÿŠjmžüþ(•ƒ;­j‚hÐȪl’—Ö¢.‡>h:ƒ|®·ý°õ+×j•OÌ#°ÚFñz®Ó€À„×ÎÖ‘nÝ×@à>ù^ÛE—¶Äí³ûRžm¾ºxg°‰â†ÀôY|TÕZÙjú^IÖîÝ® î;Æîâ|·Ì«Sˆ^“(Vlºëàìâ•SHaF86¶F’<âáAgË:¾Jk$ãn—’D»ÞÐØxø»;éÃr&à°iõ)‰`ã,cFÛ””Yµz0‡Ð^o|º(ä±;?m!¨ºÈÙ+—7RW¯ð(šúV$ËÜ`’ÔÈ!/°™vk+ˆÂ‚X6´Iø˜òCàg c£ƒƒŠ6õ#QQug1l}·úwå…¤Ç>ï§ÐlçeBAÑa¯óx/Dü–qJ(ÆSîžò6Ãú—ŽéÛú¾pŠ­`ØÆ¢™>ý?ñ–ºFŸ†{ån?,컪&”|œ ó>¤…ÊÙÇ*u¯ ?úlk(ü‰½ÃMG¨|ÚÉZÙë$ÙE:h®da/ä|*·§ÍÛµ¯&!ȹˆà+cÚk•εÓÔÚ”öº >ªÊ¶Ã9…²ù‰[¦#ú9µIƒAÀ§…ýDäc  ú$Sˆ¤ Wgaè«*«†¦äÄÑŽR1„UԽ߄ËKÚË„lN™¤Ijú%½öEÐ[ˆ!4`·ŒêY`ñ+ç«ôûÆ,°p“$l¥Û宯¤ Ó§é ›žDJ¿àÿ|1 ÙfUR,ÊÉ 9cUíîIu}½T®ÍBk9 ‡~ôD sƒPLº qIs ¿á¢+’â×OÙ|Ýf-@jNš_l:ZÛ"'bŽKÚKÉös ¤ÿ܈ À|;1QÊ Câ^9-p7¦0ªé[úÜ*‹[ÜÊ£‰sÛÊÈ E÷Ô¥r­‹ë£µçvôè ñmé…Þ¥…u*ÍUýTó³"˼݈ à ý딚’e)ŠØ½h(Øä€ae°Q`Žr¡XÙr³ƒ«4"”íÅ,ôÆ. qˆÂ7#¢r6ü<Ÿ~Ӈʱ б×Þ‹f·7Æ•€šÊ ¥’µ‘*.±ûãvypÁÖ:æJÚÄiì’·^è„þ¦mçEHãw®Í¾"¢ÅUÿ“?0‡Wâ¥tµdÿ×½Lªç\‘ü¬9X½ûÅ'íõ-²k¢íì$,ö±q‰…J½Ï±Â´~'ºÐw .p±ëû’GÉ—H Éá—0Q¡Žâ'|ýþM¢` gTÅ%|«¾â,c¿,’Ü-/çÿØ)GêVÇœ:2w®I³$ i$aêö«3?gáÇw†¾‡a8ŠSmš!Á6æ<¤ÊyVûÿ;~GçæKæ w§Ð_”È ŒÇ ®¦sPü’š2ß%σ 1+æI È¿OïÞ‹OÎáADƸ”OHpY¿pð3ê”5F¦¹äÂasqWHØ•?Â(…yAI˜ÉZmu|Ô4Ïá{?4sÆÍi sÅ'¶³©pwHt.žÞ|VØy@îå·fw¢ïdOø__÷È:‡•ýRª½æªã¤—êšç’ª!¾£µaVëXoŽHMeNÕd,ÄüÄ™“$œ„àYNYß×ÑÌéßétØLÑxê6-öUÅZaÅ&_8Ùe(op¯LûìÐùQUøJ%º»(uD3ˆ/£hŒÅ*@E÷S¶S j䀙ó5Ù@8çÉ™û¬ËD\ô¤9_ï áõ+KÁ*ÑžK?KòrÃnpš$jÐÞ‘OÝüÆjXZ¼Û úðž[±AšÇ³ºù Ùa…”ë‰Ëáìà³;LkaË7§J-&Á[xYVB{X^u`öMmò˜—!݈oðš!>{ÓÃrXTõ›)T¸*dðø¶V¿Ú¸H7Àê´å§öRçùû»Fñ‘ÙéWÍÁDÒÛÂ_àš/³Ul!P$…Êd£k–ɽ.nÌ,U j›^N½£ÃMŽÐÀÕé'HÊ)xP>XKý¬‘o Rdÿ(—¡ãy ¯ ˜{uVR›-÷m%×-¿[¼¡¢}¯Wÿë`Hlφy0‘ô^f§¯m,&SÈ=Á{—z ÏáC—»ñå˜Oô„ŽfÐWß›CJ<‘åWÀºDASûõOåöÓ£Úñpº5óô`V 4HøÌ®ë]¯m'D]‚Y›­ÖæwI.u4èÄc¡ü(Þ‰õÔöµÄ'¸˜zCè-“"ð¿ùÐ ðñ*JnÉ– v½Ô¯Œ:–ÅÚÜ»O÷>Æ\¸ú×»ö4 =iæ—9µƒ¯l~ErP‰{mèÑw6‚ºèñº a‡Z ߇…<Ô¬ÞºKö)ñş뚇ýàeu¼x~{÷hàOcððmì¢p¸°¿E¢mÆ#¹?•nx„ä‡x<}™þ>uÏÜûàœàÙ’X7ÿ؅ɱWK7”A÷‰¬ MšâRô‡ Yx4§ä½;\pbðu€ýL¾•á!ÿÕo¼A®Ñ0%¨ŠK;=¥”Ö70˜^tö+TeÆ5òì”F¢üËïüéÿYû^>Ó f*•¿™~ŒúÊΈ9¾÷†ðzðƒñ-t˜´`ÇéTjð¡<U³,£UÞåu»¡íÚÀ-XâX:÷Y0X9·íÌB=©7I›Û”y=yÝMå0yõ8Ž1·Ì8ˆ²êŒdQÀbؽ‡e•˜2éÇtô~†¸Ý…Ç=víÀ…L¯<‡XRŠW²ã7wÍÁ1¸Ä·î{KrwäæÌ§¯ô ³tdZýèy(¡"‡WÓcKèöS‚{ª›“4•jÅ­<ဠtßë¦Ò©Z~𠼝…Ýø U$Ð17I4wÒ.0x9€1¸{\¯ûGªtdz-0Wxá©i‚*ÍäR¢63µyó!à* O¬;¾3ÂL[sžcÕ­Àüü‡ZúúCÂN{ð´òÊiÍ.w`õjôð`houˆVPNÇB@½D ¨:y,ïªù,|²•=) Ìæ¦ YOÎý?kó,d•Ø®d,‹5›Åúž\€|½UÙ>VŠÁã*@ȃ¨xiu•Ïzø!¦¹(Ô|¬´† C¤â̼ª çŠßúâ@'dY̱‰ìŠMLæ3C;៚ –žžå§:-%j?ÖÖ~ÿƒ¦üoÏÁË$tj¥s—I;ºß'Ièf§ d“Dø}” y,/D?µ­ÍøÄ˜Q]«Ú#õûáo®Mï»[DÃI’ãS,E¨ÏßR·Aê˜1i";zåqù„M`ßë÷€Ö¡úwù/õ3=©0DÂV5zÐ*¢v³´…F› »Qœbë—•ôšàÌ&Í’ÇàòôÊ9ºIZ,6|¨å4ƒ0W¨«J6háµø;P;Ú#îÁ…WäJR<@+wB€ ÖZÄö1“$JT1 ܹwæ±ÝEfçaR»ß¦˜øm<Êâ»;}8ŠiœÒC¼}‚JþÝhh“ÛÁ¿ ÇövQ/¿Öaâú§ZTÚÆˆä ÿ¬s}Vr>0 ‹YZrelsurv/data/rdata.rda0000644000176200001440000001562413400221670014453 0ustar liggesusersBZh91AY&SYr«]L6ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿà @à,ô;à€’`>$lĉ6bD߀M ÀA Š 2h&M11i‰¦#@›§§Ñ£L`žF#&&™'¦Ié Æ©SÚOL ÈdmOSLŒÅOa'‰= lÓL¨5@! ™OS&§¤›Tx¡  MÒÀhÑ 24hdd4dbcHÑÓM &' e =M &€À•Jzž¡þÿ*•CÚ›Jd )SýTñOÿýT¦¤Ÿ¶ªŸåHõh@ 2 @zªªOPÀ0 ‰€€`& @A4¤¦™=F#ô bÔÉ Ð4Èd4h4 100Ójhõ1&4ÄiÄÐÐÁ05€Ó ‹@E“N9ؘxÕv&Åâû|¼7V|%ö“°0°•¾Ú*nšÞ¯3ÃbŽÓqfÇ*œøÛ¬Qf0•µ_`°ÏÆÓ)ÖY0ØÌï&Ô]*Öµ­±/-±—6%{Ûß7Ý1ÐÞV3ðœçïâÎõÜc£JŒì˜–•Ì r•ë!°ëÌJ!²™îס¢‚É¡ -OÆ¡¤§© ˆz¥Mm_èçp¨ò,æJ,ûqG¬Þ^h/¸Ûdh}­¾Ë/½)yì%FGçXóEÒft‚M-¤rQ‡6,™/åc«U.ŸÈ'®„†˜•†Cw®÷ú/7ÄMN³ìsá¥äý\r¯’{Š¢5z kjÊ<ïá¡ÉDŠYüg;Ô~§XÈŒ_R·B©èpâêpj‹9'ãÖ=Ö~åb9=úßÿ@œù*¨ y€'ôLo'ˆü<©ºÃÄrúΕLK}ôû ʃ±^ÉK(ÐèÍ;ô´àLŠ¥Ò"KÀL(Ÿ4ˆ-æùQÑ@ŸùÍŽGXõQ€ùÔ~Dž³Õ1B•è#GY8*}–èÝ îÃya‰ÒtÇÛ‹?sz§ _??爋½ò,güúS>xtÑÊT8I"ðèñ ŠL¦)ÓÞ²céÑ$0i’½gKv®Ÿ7ãñéß…ãíãÄûÍ‹å5«»÷ŽŽiKF¢h   ÞPÀ]<8õŸì“û¼oLI #ôz”€ô€X’þ”„$˜ %ë˜ì'ô8£-°°ŽÛ)¶Úí‘•Ç ZåvÉ"v&8ä,®QŽ ¶Óƒƒ²Y#$¤n1ÚÈì£%j·Keƒ±ŽWKR²ªÔŽ:¤‘Ôån;$’9 2FÈÛn•¹ ›Þ É ,sÏ:Û"m×*d²”q;$+®BÙ)!]Ž6W+qFã’»HÙ Hã¤qYcŽV¤”²×,V×hâv9]nŒ­¹"UÉdVÊ2ÈË%RÊä–)[¬r7bmÒÆEeûü³ÒHI`™Œ‘L£®’ ·,W$+´Ž9)mP’ZÙ$²Zšµ¶2Ye¶a+„c„M–d$¯Ip!€ØÛ…ep®Y+$ Û#ˆ²ÖX£©ËYƒmËÇ%€ €h&ÌlnY®Iq¹#S%nl–IG$²ÈHäeq‘º8ái*²6V:5#–Ú5£â…¬ŠA•Éj2Ç…²Æ2˲¸5\V[a Ë"Û$Éå’:m.La€„³Ã—%íǦ•OÕª‰¹ ‹Ì#?yÑ¢_öb_­0"¢p)HŽz¼i„·@fÒ>?¶›—Ät»¿{°Ÿ,ž>Ü»±a—Å ÞCÙÍì;iX=oñ厽é|_×ÿŽl×c½ÿ³ýŸ÷ŸÄí£mó ü‘|ô’|³Yæ7+oeÚÓâjߊi¾n‡òúé¯âBŠ)öÛõSøß—@áfJíÎ}½¿Tôc0Ã6›ºzʲT2çÃÏS”ü½”ãvKZË3Ûù3¶ùº<û"½Úské%ù¸«¯k>}/±[EúzŠrhXàžÉ:kà7rãOO9ѵ7[ÄãÑ?OŒqéN)õ[©¸t4Ùô§ÿ‘>ÜS\þ²Iäμû¦Ñ»¡®þ d̲ƤŸ`š¶:vcæçöw®é65hÿ3äáÍÀš“S«Í£ZÙ°ëß®½¬Û«‡UZáGM›/û>«š dK XÙìVmƒ ä«%XµŸe¤${².ÔâåáØ{e •wé¯wC2•YÆëÞ8ìFiZ–bÒÙèêø;êÝ… uuõ˜tsoÚYÏQ Z'F+¿ýï®Í³‡™+Ÿg[W"¦Br\à‘50ß‹= s*lœ]nkowS{pʹ+sÕ³·bหnìÙºˆ½ÖðôåÐZ9ri«ŽyQ¹ çæ„ÇÕ7&ÎÃ×­ÑU\´mÔ47‹6N¿¥ zÆ{°Û±»¶óñÙ»×'Y›ÂêÝK¸ü$Ûp{Ýû3aGmÔßJ‡.\ôî`âjË^Æf~]MÇÙåÝ˸5= ôTòÊ·Ó§ëã'ÒŽ7ßÕºûa£/%õõDÃô8vãÚ˜;6MPdÇÁux³3ŠÍøøl©™[–É‘ÝìÃÒÏÙÛ¬ç+å Ôå¯7éÛ£~™û*lxÝ4¹3Æ<,—~èjoEÎʺ±biÂӗGçâ´~–Ùm‘YAæWLj´Ó”ßï“Ó÷5¿—Ö,ŸLå}Ëœ]Ú$ÑÎhÒlLVa³8ݲoϼèzàØþÏ€¯ªú ‚‚¨ Qå¾ã×3oô]¤š\Db‡|WÞXÁ”ä——î!e[S¼JH•ïÑíqw~ç‹qYÒX6‡î}¨ùŽúá«å¸ßHå\¥á6­Û®ÒµUn Ù¿‡å¾·”ÿ¿QÈü·çåžO¤b îV­ñ·PÜ/bû×Âó}ë¡ x*Òrg&^úrˆþj@…@ wƒ¬¾.°~‰±®Zpy7ï¸}þ[“…ã2IHœ¶gÎúš!ǨÁ¬ 0‹V¨§Îš׋Ä.Ë¥º ½4x§‹ ~·‡—ÿ3Ñ$‚ܹ}èëÁg«÷k”Wd;îÝÅåÿ7‹Tð¬™?3Qß’×F&œÅ¸¡Á›vô þÈÀùdÚFpù¥¸PðpVì ‡˜Ÿb&óú²Ô‘ØÿY»÷Ç”‹aÅx'ƒ`ÊOÆ—*slº4vb^Z}ÊɶªI–REn|X’sž‰!ê@ "Oï'×àšJ†U~©Õ‘çŸE܈iÖ ®¢`Úiçz¢¯Þ3­p.Ì6³õy-Î/Ì{Oê–ÒPÖÎ%Ò——,c·ú2 “QîEw®Ÿqå°K—XåÄ ]öKùn6ÛÞ½š§!öÅûïs+'3(YGÒÿm1óÎF».‚G>Ó£•Üø·Ð»7±õ—u®Ù–üJùרµ\¦sk¢p€ ‘HøQ›ËŠ™8I=5ìlÆ‹•ÏÛHË}z±¦rªt¿2/@ÓÍgÏÿTK[û*©îºŠö9›¹¨°ƒMSø{H’vùz̪D­ËVÙr¼SÝãU³Ê”ññç::8¬ m.†²ÎNõ"9D¿–¦¾d~KgùÈV”aaQÊtjcÛèåxdÇෲ\·8Sey¤¥,«oê1ÿã㧃¯¾×ê¬i§¼3Èó•Þ23y©d©$=ó1ÛžQFfÆæßïk•å½ãƒÜŸr\'~Ñß}¡/s ªŸ%˜mÚê_yŒe9j}5û9ÍÂDD±Kkóc«6óe6Q„”Í×+|JÆvÓ^^š~7(ºÎêܳR¸€\X©kBûn³w6Ë”o¦Á3_ç¾ZdíéY¬v‘EêÑÞM4kyZýóš®_SðU–wÜÆ êƒ©©Ïð¹R8 Þ[óÒºÂñã8Ñ&9û×n·7¥"pû»<×·RŸ»;BN“ =‹Gå´XdP’WçQÆmjfQÔÙ3$ÄØè[ñåûidº=–ÖTQk޼Vÿ‡ ·Ñ¢W«! ³ÜÒ‰‡Ñ󵯩ñÞ´{=g%Äœ[nðëˆ&/±†VÈÖü=PMýkd¹.îÝ0À¦ÝÒ©x#l2iUuîüpÞzÍÜ. –}òŸÛ]ïš‘Á¯ÚÍö¦*G¨·ì0Öÿ Ù*=Hâ;«¬ÿÔ…aÍ6ë›å¯K'¾¡|AI$uÌ’AçBHí½÷ñ»ßú>·«üÍëu×]u×]uE•4ÓM4ÓM4€•Í_ØS{ú]Æ3„cF0ª„!’{äûÁRATŸ4Ï ÐÜsä:ªO«˜âloå³x6ÿ¾ÆÖ6Úm f2°‡k£ÆR:1¶Ñ›˜9„•É™8Á˜`¬Ë+pÄ¥•ÖÓa“ŠL© Å™<“@lÂÁå ì— 3 æO'Fž.&ÌÝuaN:giY…¶6ë´++ÁÄÆ`äq·Z-’`5×ijÀhlm6bÊä†l0i– ˜18Æ!Œ¶Ý4¸<"‘Á´Ú`ÓŒpÆ‘•*ÆëFºÔŽ1©…µÇm¶ ‘$€'e ?° qº$q! žtBÒLI$¾ìïºy~\R@‚I7n‰ He§QÛ8Ï Í MG »pBœ¾w›M›sæhi¶ &¶m¦»ºÅ¡e—C¡·…Ì[u€gíüÙ!âKMÿºq†0JdeÕôº\|›µM‹]`–ɶ峚‚-¸ïÁ :«©†oƒ9F…Ž0HDÇ£–ÝØ‰#n±$¦Ã,tâˆfdvDÚ{ø¯rñÜRM¶Õ‰“‡6Aµ¡ãqâçᘄ³Ï<ÍØò=tâ($šGÇÅ¿„ÄÀøJ…ˆÌ(`:$µ¸Š R-Uö<žçÀè©ø@Ê]³ *²4Æb©Ó‡—ª=qš s‚šêÈ-eQé|.iš[øþ›Ÿ.^†«~8˜pÇa iŽ8Èù^YŽ®¶ð7›ö=ùª˜eÆòÏ] ’Ç£´Ý³~{‹jh0€¹µ¬,çY€qÄŽŽòT Ðvˆ¨Æ2çêÇ©í Õ:‡ú?3ŽÓ ±U}!‚vÀF”Yyê¯z÷ïß¿€îŠ(aÀŒ`F0#a-ÞÆ1†}]UUUUUUUòóuuu)^^JR½É!ZµÂÕ¨F¯%KËËÊ–Ûm¶Ûm¶Ûk**ªªÊŠªª²¢ªª¬¨ªª«**ªªÊŠªª²¢ªª¬¨ªª«*`ã3]„c FM‡®sœç9Îs,²Ë,²Ë,²Ì“»}}|¸’­ÙQUUVTUUU•UUeEUUYQUUVTUUU•UUeEUUYQaB„!afCŠÌ`*QT¥)JR”¤ç9Îsœç9ΪŠ(”¥)JUB0·nÝ»jªÊŠªª²¢ªª¬¨ªª«**ªªÊŠªª²¢ªª¬¨ªª«**ªªÊ—&-R¤Èo/r —/ïïïïïî\¹p1û—×ׯ$‹J€J€J€J€J€J€J€J€J€J˜l$»†Â`T¨B¼ e–Ye–YMøË,²Ë,²Ë,³ {{zYP P P P=|=²¥¶Ûm¶Ûm¶’    J†=…Œcû™DèÃZ¾®ºë®ºë®ø Q)JR•Œ%)JR”¥.=•UVý•UT•••••{Þ÷½ï{Á…³xYœö ò”¥)JR±Îsœç:(¢R”¥*ò(F*ª­ÖTUURTTTTTTTT»0¸2¥L‡‹Ý»råË—.\¹U`¦Š(„••••••••••0˜Ö’L>&ÀƯ'9ÎsœçXe–]½½½ ²    ¦2ªª¬¨¨¨¨©ƒƒˆL5Æ0¾­`ëÌÀ l D£k£±Ä cdàŒáž`—`Ë,pH:ùÔí¼[³Û·#àÐØ6Ø“blobŠ1"“)¢"1D”EbL)1&"ÌÑFD’ɉ6€“fÀ5Œ•A¥"1bƈÔZ4EŠ ÐF"1%Y*(¦,`¦h´’DbÁ£D…LÐQ£bB4b(0‹ Çø AQ†!5A™%!$Ê€£(¡I4P¢ ‚˜ š2bÒ†˜Á63X i6Øj’SIxctφpí¸Á#d#„ð\É$$F1ŠFRR™L•™AŒLÁ‘ 2 ŒÅ bFaÌ̓–2%$ÑAL` Q JL”˜$S3Šde“)™d"4Ó!"LÉL!™ 2!© £&i™ ™™¢ HФBJeÍ,’ŠZ@Jc¢SK $c0ˆ#2ÄE1`ÀdYÈFe£ šá|ÁD ¶4Á±ŒmŒ46e²ˆXÓ,Ö. ÇP-üšîÙªÉ$»Zoäë2HŽ<ÜÜ}ÛÙ ZØEu¬r-²Îp‚ó¶D–@ÀJÀI 9  ¡ºa‚Ð’Q#A¼LLqH7i¦³y¦†ƒz øå†8¤.uOU“x‰c·Mú™d$v=g¡žáÌ1„ó´ˆàà€\]äœs¬LR”-9ªÛ"SOE/3Â¥Kç –î<+®âôÌIÆ{¦jÖT“l.T1J-‰×ËðaÖø‘.':Rxè´Î°‚ŽH*ǃ”©L_Ÿ•\Þütþ¢4©Õ55h̃ÏÕ9î uqœù?j»$=BÿyUûŠuÓÎsëɢ͟Ûy…Uƒà3F ëFÀàˆåJÇFj—6Ä æÉU3aE-yܱ§=~ ጽqƒcï#okç;;ËáA¿‘ÆMbLv9¡ûâsObÉusò£ÂÿÈ2zëÞœ”›¼Ç9öA‡{k2Z¾ñ8f2NÛOÑ=Ÿ`4bÌ;Ա݂77:‰ÿ\tÕy‰/ áZA^ª§ÎæÛPmÝu`=ÓRBŠžü¿Þ@rüóÊ ±bžÁ~Ôô ¢¡û鋳@œ9’Í¡ùwø¯qòÆúD¡c&G*¬²F\Œ¡Iydtöþ.g‰Ÿ~;\?Q; ø°@yAÿSíM> Å£hG¢ á_U Å}ùÕ#züáPÁ'B üèòO[L$.I™¼á½ QÝKåÞu-“µ }hØH )øm §³E2é~eï¦ÍÍœÐU¡A\DךËdE"OjaÎe2#bÂŽ­‚¥„{UÔQƒ4JÔíÉ Ks”¾ÂÀsµu?Nmm7â}@i¡•Õ¤?x/¹“dèµ¾5&ò²ÕN:|ðZfÀîäJxuašÅN9{w¥¥G |…êÀÑè!F]–D˜tu=6¼iÊlkåO¢+úx«Qj ”J°~O,¬3YjŒÞ“oñNòLY؇Ã௲k;Â’¨ÅÝ¡ whIȵφ2Ij|Ô|£¿|áá­ U‰2n¹Èí œS°0- *è[ÁʹÃ![(;X”&M*Ôš–ö4Fr¬SEs Žëcb‚§„šDbS=iC”‘(SYNå!-ŠÍJI](b•k!;ÃsY;ãDC+RZúÀC&%ÈtY\ É‘ T2`b9˜ÛàjOE”¬ÑªaD*¶Ej ‰ÑA9ÒJÖЄiÝkF²Ê'‚]GI’Bw(*iíµ”°Á”F1¥l8RÝižfÚÊ®=5+4U'Õ9ÅKAX»˜†,Í4âµ¶¼²åÜiÇÏÅ“Ý.ã0çæåÅ–B»¶òrìÑ$Í‚3;Ã!«$ &2…Ð` ¹ÖÄ N“U3ݪ¦4=µÉR™I+óÑs3£:dÚl[’¶Æy§ZðŠÃí’ ze‰J®X¬Šz”’RL{c[¨{EÔ¨(ŠYˆ­ƒ'©í©Zìc ƒÔ¥ ÜÙœ cܱudABÙœëö´È;n¤(‰0¦Ëhó¬`s‚ <)X"]K™J ±BšqsîÁg–•[‡QÈk³æsãŽ:w óÜc©Ñá³fF}. åËf|}.}†~ÜnåyxäcëÉ$$…ð|øšÀã9 ˆJOJX•„0YŠÆ¶M¤o)Ž!…’á#¸NMZ{CXÈ»ŠÇí~X¨]O¡±§‹4ýe+¿G¢„œ5[6 ‚Qצhÿy¤¿Qn—qªƒFµÂ4mëJåŠÌ£ù~8›®ì¼Í&ìƒÈ·…UÆ´‹²ƒµ„rÇ1ó™©Å] >/¸çô­îØáäfñm\ËQ Þ·Šž²¾ÊM¸ÖåLDà] “¬’í¯[O<É(Œ·¶ŽÉ÷Ü[H=o"98®i¶ØuåtÆVŠŠ™ÿ‡Ëþr¥ý&:¦ œs¶1øøÖCuÕAt¥¬ÉôÃ*†ƒÂƒÈë†Á`¿"ú7ìãV]žBë±6`KÈ1³š~7+àHËç_K›ÿd‡Ó×HÇ¡cì¾]òá›Rï¿r»öªÔ¥¢þ““©omÍs2Eaж‘Y¼mó"ð$‰aÜKÑ^l ØÐ¢¶ï¼'Ÿ ‚üüµ/C¬³­,Òa omp͘;íÙô±u×°[ÉÞ‹C…[A£¹µ çÆ‘ÃTCRRñiË<ÉöžðJ~Ôí ’²²cj¨I°‘$ZÓ©\–ÏS»äN_Ä#E]»7åóœ';1™ \®EN肽®ÕÀà Øû¼¿E+m`z‘‘ü’6ÂÁ0þJ ¹ì-?]ÛÖFhøµár1sî1’à !Ú?lÌfã—²ÍøáyŠ[$rpE²åYã•π߽l¤h#­6C”£5ëí¯e|̪ò_JÚ+H©#H@"f‰~ÒÆ¯.ÑÎsâ²sÀNÅÛ ›?çzbw½Y·ð/Ä+y•¢–7~°§å*§;é1dÇ`·X·\ç¼ÑvQ\ )Ÿc!ëô}AQ¼rò˜/i±5!™–৉¤>r®§…´m”茘پà@Golqšö;‚Y&CΠHSq™à¨äx];È«9êÆüÎÈJ ,‚ÌI 1w“Ë¡‚WÈ@:,©4üʵƒ*aRk¿z‚å‡Â†PW`v>âc°ZJˆ0Çk5hº]ž”ô}®4Ô©PxÍÑ­ƒ<°žóYêN It[¤ª«2ö!o7±—#2¼ñíJh•Ž-­ˆYÙ0óÀÿî»×Ó*º¥Ú¬Ù2ã;=Û̱¡VÙˆ*c„iqw•íBm³½ÚÏí2¬È´¨/øs-º°*“JâΤ™±¼á¼¢—š#éi•Ù9ÆÙ…ð$ù@ŠŒ$?`)$(Ï;ž¶ ô“ŠzàÑK¾w;À‘½8–ó?«6µô¼_$ Ú‰’v—" >ÝᇡÄ1.Ñ,× •š#Íh]š^Þ¾®BâæªÂyŸ<‘µ0áÒKȰ*ññï^ri.)ÔÞŠúœ犡»¹Ûe½ÁÆ´iÎß¾Z¨]${‘|lÑ…*æDÝöš(b­E6^ò±Öª0»ŠøMÇø!èZŸÞÏ* ˜ÙêDzy#¼¦¥:Á¸Q?ã}×û¬s|fcl FwÍÐ8ñ6ÍÉ$Þw+MŒ´’ÚµöaÇ„ —M!2"4ä²ÏÚ¼ ¥øŽ”tO'¸NÐ.ù¹ùÒ{©Èr†ªmÔ£–jk’Œ-ø4”êszÆêlåÂÑœf.ÁÅÛ¨ÔfÚÔú~/Iж‘Véešê-!µB`Ö®.G±\­fË7²x‚*ª÷{yVÑŸ"JØAé­Âó.}Ò‚I¿µÃ®Uáâ›Ï»]¹`@s®åŠsˆl†í“y÷­<8Ë?è}˜æ÷æcHGݘ ¤ ®ý_ûYTú¹sÝm©ÑU‹s½g™Ä3Mà„¢ ÂæóiOªßçÇÀÓU!”ÒÔƒõæD[DþÎζÂäAèàÔ7fIIpΙ;)¨"Þª_!zA 5W‘æ]¡ºDÕf(è>Zé Ÿ¦NSßžµ ;„j„oÒ7Ëd‘…9´+dü+ë¸ÀdG1Á|t¸àlûIAºÇb.Û N5GZ~HW!'’ðUÞ ÕŒSÃzªÍçúZʨžóv\”‡…÷þ:Ά;âíH´IHÞÀ²{냛W¦R Gë—uóïk·¤·Àé€uFšêŒPN»Mß@áíì­æO0VkäÂÕPƒ÷)®LhÚÿgD0¶ˆÖâßÑçTk¯ÊÛ&,|9¹dö{ßÖ\¹!ÀÐ2Ö|ʇ@Wu£ƒ;-ÂÑÅ—êØD#é–ü?5Ð.@軃rPÏYm»¿“Ê7 qr÷±o¼d`1aÏ}Í`6ûá±õ;"vãø‹LUk…—%€ŒÇÞF”yÎ<”ŽS–óB>Îö N"›ÃV²‡‡±„Ø&#BKíαlöÈ«Ô>c³AòöÒ ó‘Ì:´‹$JÉë(¶|äxWGZo‡Ü¢ QÒDDíðqØ´Çazׂu½ù sEpW”]”3Ý9ž¿[„ø~ä2~ðÚĽ' UÙ5ŽM{)y¡¨Æ"× m ò¦@©Û¢2æWF[®Ëµû*dtŠÏjŒXùìNcà1Îu*hjC=û‹-˜(ç0æ§Óá'Ñ´È m–•_¯«-ÛR5lÀS‚BõI7kÚD0k!M\Š›ïTØs?4W ÑÌЪ]E³ôމÆ}*6óìžHïz*^*H§*IP«F&#´ƒÁN»´,þIôi ±ÅÆHc<ÑŒÌ"ƒ^lÌP-oWÍõÊÂ>s`þoÒƒä)r8ÿ¥ »¤W/•Ú Þ©€r7ÁÜô¹"#b0æàè÷'ŠÚ°ôÒš~’¬y¥áµ ´-s\ÐRžÆ6+ ù…õ}¥^mˆÒ×>¹BL~›…4]N'‚„˜ØwH³C¨ž%_ƒ?¯F—KzXzYW£IѪÁ{ýÇÕh§¬'üò9ùZ &âJijǺ°æ Œìi³ž2ø_‹¨òtTã»xˆÛÃõšä·Áöí{ª¹Å™Ÿù5x]{~_‚¹x†¤…ÜêGß“n¿°6²©o¢ÜX¿}Ûº;ˆ»P]è[ãHÏFÔÝîÔaØU“Å?^ÐbŸî‘ÙÆÌ ¨g®[äŽÌÑ/‚˜‘ ÕtèfKð¿õ®ÛZÍ<ß·NtäÎG­:ݺ§¡¿g£’ôúåc|Šï)¶ú.öT¢m£5©uŠcïr$2žqã%ðÉÌÃm6x… ö ëâ A}lE²#B ×ÊÖ*uWöå°÷/°ZÖŽÙgƵ¡m´oÆ&©;â&6Òv½Ã¶q$#Ðù®Ut·ÉNÃb ù‡û¶)«*ã„ñyÛla•¾D.¤òÀƒÍ[öï^jX >XÁêÜ,ßb#B.$åž’œmNR¤¥Õ<@òõpÝ!ãKMeG‹ýØí¯½®äŸ ÖTÒ{ãææø'8U ˆw™ýíƒ=,@9™Z_Z<þ|"‹ªÇ¥išOòªTéá:ƒ–Œ¥é!]`q”0 8ùˆç5ú°û¶×U™XAZ@9;Òƒ Ü%º¢NÁp>™ôwˆŽ“=!äO«r[]öî¤è¦@âir·s ¡—M&vFWb) ”ô‡˜ò»xE5©Ê#ö€äÂÓ 3eB6h( ö’É\7ä–+‡5G[¥´¶_ÙÉëºi  …E—õ8oU „åÉÞŠM_R#Ö|ÿ’3„(õD-?—äèXŒ†–Ä%ÔZ¡f¤ãò8ÿîèÚ;ï·H…JýrЏ{Qˆ¸ªŽ20 g\GVØ[4þûxïý}ùĹu*ûÞO/Š6á9.ÒUèÁ|ãùy@u‘á¦!mŒ,úçÝb~÷Õ__Ÿ·Ð ì“z–™t"?[AWèo9¾‡l¤.£=ªŽá¨Ä½£*ôÁ&uTñïÆ/l4]Írâ¦àÅÅs]ûu¶™ª•P}koû·ê yP‹¥~3Áj^Ç Ž‰×MÒ}Ý€¡P…|ÚmùÌÑÙå'E–òfÙÙ€¹@Æ.<©'à'å ˜QW+—P¾Ò ¬O¿œN25áò»Ìžºú\0€/Nß_éÌè+1íÊJ/ÿ_:¯®YR¶ú¸D½8xùŽìÚLGéÔ˲+Q7ÆûÞK­ÌýÏßì<¯¢‰¯èø^ÖoÞîNk–úÿÏínèSù?ƒ‚ÿÑÐÔ Æ'g×Õ#¶V™¸^Ö5pB.øße#-ü5’äM/üÞÍß³ì8úÇ!°#­k/5¹øâÛÙo¨U”Ö€W·• ÜÑ‚h†Ù¼ ÔÖécïºï-ìa¬¨ô$þ—6A `£I–‡—²pUÉ¿»K–¤‘$î©ú!Zç}~+Iu>=¿˜ïQ£{œ‚lú.õ×½[ƒ—€×ë°yÔ³|ºÑÛjûqÕœ¢#üåŠÙ”yÅíå¢ÜeòFt.#Âi¶‚dµÿ•’én›Ò«>P`ÿ¡T¡Á9NÐ5’GJ—ýqàÈÍ0ð³fÌ"*ŽÉ}½X‘]&›ÿ é·òæÓ¾+¦h DþßU¬Öy?‚fô+è\¹S£Å ‡œ¸æ ÔÍ>?6od.+{¿õæ~í˜j—œ˜½¥Ü˜Ç3ciƦH1ޱ”=•'D¨h/W”Ì,í§Ì1"]¤cŠ7i‡U›4YfÆÆ‹­šñ”'¯‰^û¦ kʆ¸[B¥ò-~Þ¨*[~¶í(ìÄCF,UNÛÑ:»¸‘|»Õ/)+Žv'ó-À{”B»!¢žjK¡óûÛ~K gXI;éüNѰ8O`寭~åÈÚßC¦¹d£‹mE?îÑ ãbd£ðHVv²pì VK¸uÞÃâK_ò‰‡8¾¯×ÑL œ~ãÿõîÙYµìÎ:’ç-©Vð×ÜýPì _k.(¨„Úáïçð¶:Rhø¿ºÈ¥Ÿòsø|ŠË݈Qî¾{Æþd ‚qÅ]«à"¦ÛÌØ¤P\«ç¯úÒY¹ýbùöíÌNI <™×…d“hz0Æ6¾2¥¯§.‚}~ª÷1 2ŒÃ‡ÅS”Ù†|cJ†B4OL­ss ¹qsÈd\ã8Û† (¼œ0¼-ˆPœ²j…"o}[ÚîItì’@BÙ éïqFæÊOo³öjT\ÈÑDsÍ„6u<ŽîCÔ»ôD¿31Å~C‘}:y't­[˜q~µ˜1:ƒÁw赊;åü×¢ZújïHPDq©‚Áùìne;FÙÆàKS?5Þ(Ê· WÛ_¢X’\Ëíi^¼²›û±+oºu¦äû[6Æ…ò^B´íš³õMß™P3âÉDZ’y1lØCt¸Ã,@¸ ¨,†Æ“Oš´VNÃÕŽ—ð­áyY§Çº®uõË<§àÈ@OÔ:ÀN‹w»™ý½«@Ƶ÷OnÏ/Wð“­°JÂÆ{Ú®Î7Ä/“;™¹/ãù ©¥+fí–P„/ ˜zÐ h°T5Œ vzòZd‡L¹Á«²RY gÌwF¼S×Ô%æõ%x% +k8ÕSbR»qel\NE³Û„ºÌо «–Sí’SX#B;B¥÷Þ•ª=màb£ØuiÊè«Qò½½÷a7Eo{âB&jC*…îÈTCÞ‹ž?œC„ièÒSnŒ1·J€ÚÐÕ`ÜFÓIÈÙ°ŠßÏ^DBs†Õþ™ÑÕ‰PÕó–ÎRÝS§„u.;‰“ÞŒJÌW–À"´)p`M>•Œ¨ýSoÌcÃ*5)¹}ñ#À¼5è{îýJüHy^¥Á€“â€D‰Ó­%)TyȆjgñ÷}•&OûÈsìr1êo%ïfÁ›µR÷á¤ËWÛA&~ᦎ~¾»‘I—q»ûàŒU°¥+9gåÂøkAæîç2Õ|ÆTc6Ä,†ôÊ ŃAÊ¥órª<õN"°¾w1䶦ǟ*LÍÒ¿º:Încmö/ìõ Ü¿7M÷("@õ€rLüeÓ/‚t!ûF'ï:ÓãˆI|°M'‰Éø:S·•ak½N*v5—X á°0cÃ#|ÿÍoØX"Õàþ±Ci6ûvvœ³íPôP?¥´õNeªM8H!C%ÀNd"Óîã©­%ÍÝì¾ß?*ý–Çî$õœkÝùg/šê¯ÿ&Ì÷>FfºšÙÈ´¼$x¶V*š©Í b'÷‹_˵‹E—ÖZãnè9V¾œ’Í€×›Ö–?f­—3<‰±¾žÊ ¾ÛLþ¹pѵ4måâV°o hL—ÖUçÝGÔ¸ÅÑA÷ÜrP7à'GÆæjÇÄž—¦+[D,§ ­yèeaS4X² 4‰O‡¢ú©´zH¾aÛq |¥NŽ})=õ[F…ß'º„F~yùCÃ(µ¦ ÔÆÞp] 訇٧o†±ØœŠ×•…ÉåmBÌ… õ;7ÍÓ-^óZE֭ß ÿ\ª¢-:;ÎuÁÞ3[&(‹“Ù"{ž²–fýÝ«³­¦ÖrÄõ›{Ñ9«`o)ì¾f$ßÎùv§ÍšÞRÙæå¹?¢pldɆ„¡G‡¦a¸Sÿà„(ˆÎ„‘ü6ž-’XñeYƒ‘‘ìÈ.…A¢µÄî‡fg»ç …Ó¢[„”2½(-XsôsM£&IÇÓ\á1•u8wBÛ[ûªM+ fyn¥mŠs…¯Q;Ík·Rn«AÂi<% Z»ŒäT§­· 2×!´Ñ9ña¿]àS²u Z÷ÒîéxÉ &âe«œ,ƒNfè2EÜmºæ1™@ƒåÌa PX.S.g€ÞdW"{“œ£0ž«‹ê6ä- b&û‹ç˜ëãåQo½“Uy¾ƒÂ³ý»ê¤MHØ· ô™Ÿw½ýÎ〽&Hú‰kT\/ÛÿË‚ýÎ(¾¤Ô@°º-{% JYR Ñ(òN^dÔ]UWù_•në4v!k±ã±·9`:]½DéëÃÕÖþ;±"Õ(6ó;ÈÏ=?f®wƒ‰wc}zîûC[Õ˜ÐTåÚ-ç胺w­µ3‚'y­iñ©kb*¸ª-¢L^´º!^êÀE·<´TÄeG-n™n»b:Õºo<‰|ÐæïôO|NƾFÐᢒh3ÊâÕšíTÂLWþ2‡0»ôÝ[] ˜~TÎ@‰Ÿï?k’vå`¬wåädIŠeÉYMÈËR¯ã¦¤VL”€ÉÅ`|È„vÐßh¼ºø[´yUõ£­±v§Â ‡H$Ä||ÆkÆàXÄ"¸Bã0?j†­‡%älÜÍh€KDö¬þæ¶QÊúK 3»tB¿ï5©F4ÐÀt»ó÷ºÔªŸÂЂ‰C7 ,m `IS¬NZ?níúøú¼—½È(!×7úƒê£Š2}¦,‰ôÝË6¾Úl0Æ–VÌr`¯#ÙF#RòáðÑòöœÂ@•T;@ÖhzE,2§ìûœÇø¨Ãc3;3¦ù<[v7®Þ#aIáj½³ŠÒ1ŽÚ¾Òëb@2—åßX"=/Þíõ;#5Мÿ»_hý]dµœQÀm¯§7¦þbÚ¼/WÐÉ„ä½i¿ÙŸ‡*È0˦œ× FÒ$« öZ­úd ¬1_–LèÎNýE"µÎz~|°°ºfóôMÈÎ6ÊH]´¢S@P.‡mñ<Öà¨Ã@¸Û€Ü˜!Ó¸§NîXßC*‘%æ}ØÉÍ\1äË9%±XLÔ‰¥b¿¼=” ŒñJžUieï´Êo¿røs&(Á†íc¡°!죭‰öü²–‰¨T’3ÿ$¯k}ï$µèØè;®}[%È3ÖДÿd)›Ñ®„ØåVö.6~Ð+ ôx}Mî>™®ØW«—¬ Ë<•ÇB#uºÃ(dE6W¦ãtJ«;,|V=@´DÜ–*ÊTQ/5±µLys Å¢ ñ1p†"…§P&®äØÚÞòÓ¹ ójî­‰zÁˆ†Q_–nÅs}5`ˆ ¥ ýxPiß4—ý(þ«î ›!}»í·¹Z¹t>œ„þ¤Œ&Rº¸ƒ‡‘B[0~¯ÛS§úàÝŸXvð¶B‘yÃ)¬ –讃à&r½Ãªø!-(ÚX\É´^NöðpOZÂTYcÚ¯€A&™àÉ.àƒ†ñ÷cvìü¢2äèèšØÍ’Ū 6a×7i®¦ª›h”ÿ—‘\ õ¶Ul‚¿pâ|¶’¢eSù™âýÂN›*íj[zƒÕv¶D2,Mˆ@_gï~²PR_=:”= VS5 áÔº4æýa‘bŤÁÂÈIŠÄ´¶Öîû˜Ç¤„žï¼®¯´uÓ!Ìõ3!j¢x¬ãt¡JàØBóp×Sv-µ× °¸0lRñœ¤ãJf>}3™Rçü`kVA ú _i,5.+]J,7mD4À”γ*ȇgÖ[V‹ìQ€6—!Ûü/|úT_ÍÛ)éÃø×Ÿ>‡Ý)Tq;6ï+x8‹öÖš6Ìü½”Ç}FµŽYÈn áMr•D õ ð†(¥å %¤nÊT÷c;o¨FƒËž[NÞ‹Ñ·ªnŽŒ­û݆Uá}ŽÓµH•íœBÊáÇ%í2püœ—Ju¼’C-*Ñ”zûûZþ¿é™Æ"07Æê9¨óðÖ8&þd$ƒ—ŠC½Â?u‘W˜ ¼nàžå௶N^:Jz•زÖã¾þ³«ë†ÄäÓc{ˆÇÆ{‘ôý0òÈ„1H7 Cš-Ä6”>] vàïoXÄ¡qï™ÿi¾þsê´9ñêŸ!ôMöƒÒ”5}”Xõ%[ŽÇ%ŽÉ5Ëá!œEõ/«-È»TCiî -þ…D'ÿwóê*t©"ñ{Å–ŽÃZ- Vߨ«Î=L‡ü¦G…„&³…j¤,tu–%f†"O´™þÐvÿ%+¾ ËY¡~©´ó\†)Ú~Hßž,^weÃò~Øn9Sÿ–l‡ rLzÁµ5ñ#Á¼ÎgüËà·‡ š¼êÖ:äõBÐÄ[û•™(ãý×á©•á"²Í*œ8Œ¢Q©¦Áˆ¹7¿¾1¯èPÍP†ýÚLý:úß7";üÙ-Ëëqì÷Ta³º€íSŽ[¦ßV]ò­Ûwd•©‡ÅcUG`~ ås¯´ñ-çÔ¦_-TÌ{›x2óŠ¢LsI8¦jL3Úr2-ÒrX‚HÕYf‚ðƒ0rôü„Iþ ðM‘l©mDŸ÷Ð($o[/GcO÷ÍTÔ¼ñÄA(ÿ•¶Ái-Ð @Ì»ÌÇ•qGöe?¬ðsÛ6ÿšïë—BS=H¾-ü!õ‡¡•ehG|0µIâñ8¨ú:¶Æ^z ¶Ž®ÛÊ5;–ŒWd’*»çy*×ê|9…Çìt{]R¦©+ìèèÖ/::ŠWa¼` Ù$ä†ÑÙb³p"ŠV",È#Etˆ†ï¾ü”§‰¸£up |Î|@¸Ò¬ÊÊAcÕœ©¤•©F OMQI½ý.ÊL%ðÌ^ ¦ì!¾¬\’æaëá/Rù“Ðÿä™'c‘µEaWŠ´ÂÙ•Áñi‡‡4‰àAÂæŸ“"š¨À²¾0Q.Ý‚=ÇvƒN×¶˜7òö”¨E[o”ïÛÔn> a?ö‘ûxöt•3>C—6Ù~Ŭ ÒL°°–!ϰ ­ˆ¥.ÔV¼ŸÛ4|°‚¡v;⸾L™EJ‰:,©XY99δPK壬{™…6ÆUv“HU¾È`¦ù±¤Â”Èq¦ž†òC—Æöc툀Ϲ$—lŽ–”‰Áßr ³µr¶õÚd¤ÓõØßIôÛ‹GGšÛ\é)Ä&fù°/·£{­¿€÷ù@‡ ¬B-²ÕÒöâ+/vF5ßäúCló¿ËïTYrSd9å"1èé¤üóDì˜ØÀ¡©óÄᙂ¸%éV׎ˠU¡ÀTwó#vÍ…ùÒ²2Ù›!¤GÌ92ªœM¥ê}þvÇ{«@‰¨ù:q”ÙòÞÐgEý8!¹O‡ÒÙú¾šÕÛ§¿ˆ'iDöÀI‘øòÄדñÀS iƒdè-œ$º_éÄIô9q“ì–t)&ÐÉf²f¢+EwpÌe‡Zü¸äiÆß–eg}éǽ™V[Æ-éÃC†ÍHÉ¥×]-š)DÃ3ÒD>-•2q' [_Æf ˜#m‰d×דœEÈï•í^Oç÷âþ¡ÚšÛþðÑ“¸ò‘æ…ãž–)3å¿?-–I÷AK°O£ApÔtÚÓ¯zm×°Ô·ë{R5‡`º*œb!Ø®WÖCŒ$ªqF€"‘§ Óò…Þ¦&”Y1j“$÷np0r²“ÛÁkC_)ý!ÝfL¢6ArmÌiÀ˜G^û®Ï—ý¨}–·¶IîÇ>›1Ã.ûq+–7õOÝeÝéæøÔqF±4Lµh\M¸}_ Ç6Ö…ø-f; !}Žv6‚<ˆ'Î`T™^¿Vúòäœ1ùVM²Ò<öd©ây/TŽ^…c’¹w¦ˆúÍ•×~†ò~ã©§¥XaA€þ^´ koï{{1jn6GyÕg¨2£{V½zhºÝj^#:ÇèÂñb¤™ùíR¨ìi1}\PC±3»x]8•™'Ó½U¡…³N° E¼ˆgfµ8®A>àÜ…£2°¦Êƒ®³ŠqdÎ^¸p(ÒãÁµ€ßàéÏ<Ϧ°Ãô¡EüËÏÂ$£BTabhºßWº7ä}Aßï{Ëa"!„daþ`þgm\£¢ªí±iH…y¦V#1ê ªD«!ðÎÿ/‘—.ïbáŽW˼¾]ÆAÖR8ž£;Œ(é Ö8{¬Øà4b¢þðg/žÒ¦ëy6rH«î·>eåà~Ço,¼*Rz™U+c¢îÿ£¡)7ç6¼ó9wërx÷•±0¦Ýo˜úßÕ2΋•ÔÈÅÎb6¿ßºB#[©»fÍôëýë-.+IL„V¾›É• %2ùí0™’õOu½ËCüB÷$'aî:ïu§&>Ú[™ÌÆ@gž`Zeª±4¶Ê}š¤Q´€X ú sà”Y…›¯?“ L=„–a©¢¤”©¾Ìêï6ô·ÒX3ìr òŒm‹n‰!Êoè2Œ;ùœ]Lþ íc˜sÁIèÑ"WùÁø]Ñ+LŠHÉ¥é12FÃ^«/YðØrGìçzÚWSßmnEí‰Ñ“GCŽ5ÁÍ^oŽñéáNƒWË·C¢åýxY±Ð¨_'Þ_`#OËVQRû•˜ˆ%ûÎØM»ÿ 褪 ™LQ†½Q¥óeXª4OˆÞ•?3»5w²¢ßa{7RÜRÔÎy¥å°jÌûòÈ»-ŒK Gج”!Y›ÞqWb‹Ó>3ÏÐMËý*Ößﯯ{Ŷ¾í ríXZ€¾¾Ã‘ú¹•Zr”ÂCÈ/…•+"ÿŒ<2ˆî”Å»*¢Xú8LÖª‰½{9;'ÃÏAñ¾tŸãv2‹~µàï‚CÔ¾~…2’üŒóÍӨ‡|¤4’@;Wé»/\/ãCÙÇ´OŸ-xÖë:ÃZß1’Ίë¡‘-O/4_9»GQ¿Kí*£!Ç+eOa4«[Oß3³Þ™|lï$±ÜY-Œ‚Þ‹ÅRWt j••×Lp¾' Aå“àøU—Ã)8²’MêÉ5 z†ñþiãè&©5˜Â¡Í{ö”½óëâîõ7M°3zí‚­z‡þ&IÍ$jòþ¨YGC3YáœÏ¢ê¹eÑè‰t±dgçm¨•¿;áAÏ8‘°dfÿ«D ±(áù¶IÅÊÔ¨\Ò:¾s@aò\;Ú_!«Ni ïJöÁhŒjJ^K¾“1ïȨ |°%"6bF¥Á¤cÞeK Dõq·£©½’_¯Ý؃kÏîû®Âåkd~^1Øk¤j8fÄû¾}[Tf84xµë—hŸ ny2#Ìí¾)MYW¢õÌ5ŠËid„6_·[ùë4Yý¸«æèðj‘¿dÍÑUxS¼p Pš>…'÷õ·ÕDÒ}}!9'¦=ðþ ˆùÌ –¼¢œíâzÀ+Ðm[½6ƒ·ãú:lUöôt¤wÕDï<ŽæÏµ¦ |J†Ç­Ÿª4¸ a,מŠ3}(ýµ‰kGŸ…Rÿ,Ež?îsŠºRìB&R˜–é5ˆJÝÝ…—eSIô'ªn*Q¼oº°›ñüW·VA/Óƒ ùñJ°Gk|SªoðYX®Á±¡¾—2ìf#}¿§;@B¬ÚXM ýûL|ƒÄ¬oωù—Ðû?S$l^J\íýŒÙšè5e ÆáoÖøÑÁ‚z¾h3Äù9Òü$¼ñìW_œ’%·™Ÿ%½[Ù¢[prYf…Qä3ÀÏvÆ_OWÎ¥Fá†Ùœ¦œÍî(V•ÓÜÅüÍ ˜/l§_æ0΄¬’‰"dðѾzô©ô©gWF<Ïáw{¨£¯jRЪ‹]Ø`[V69¦ë[&êäTBW2Únó&°Ô·êtá:½†ß Ð!±zS5bó*mQmôr“Dxha`d2ƒ¿=i R¤Xæîµª«GÂI´kî_8©µsvhJBRT·¦Ž-Ó©ûhZ¬½ é$‘T–ˆ¶?ŒkT,A³o'ïe˜eêiÊ·ýÜÎ9ÒèºTL]ÊîäB¤‹$™:VV¾ðbÊ Óµ`çÖ,¸7`, bW,àsˆáП f1j6mT,E:fm0‰aã•ÇÙväC÷)v5%æä¹§3*72¶2XwnY+{hg³ƒy‰ºÌ'Ò ©;=j>©D¢Á#¦\î]‚›p ScGŒ¸ºÚµ£²näïį«VÏ cÛª§ü½Çý)O¿¥E^ØO!>ÿImòÔÁÈ"ZÿË/†cQ‘p "AÊc2ΡKóªÀ°1Þ7Q"CUŪµ»žÜ “=ee„ßþšþµÂ1žÝýÈfgÝfy°9’ù›©x…Ë _ÝìsD-ökM?X‘¨ Î‘Z7ã“Uøk/ý¡¤±Çwpʇq»(muÞKØú]¥ÙËú›‰1}‚A “é÷ñ^ˆ´Ž=þlgëãü†vð“4Øçpž&]Ψ ¶Ã·Ì“Ð*º‡Þ#ÈDCƒCY¹8¨ÄŦ-0I¨š™ýû(ïû´-4H¼Uç±{Ô G+Yf§²®á°CçMÖjüŸ ¯kê:{:N©<©H×g$ýþÎÉãÍ¡Î­ê¤æÄ±σ‹ßÀš%ž«6,w(ŧ~â[­|åÜy¾©E“&ž;ÓK§™ 4˜¨ŸÏB«®ë¢ºAÓ–wA2掖)°Ô‚HÖ$õfy¿.˜dšrÉK°Á–f£Ð¸!¤¤&•¥·ß}îü•víX¡|¿¶fË2$Ü®'P †|n»ÁÊGòI±ß¾´Ñ?s(±¨å®Œ¢üŽ]ë±éŽŸ€­¹ÃÞO~ï5û,0—&,Ñw«6”+]‡.5œc=’¦9öðÓJô»üj•БЋU‹ôá,Â|}@@ÃJ~îÌ î²_ôQ™­kõ m6ým:<‚‹Ž×CÄbe­8­^ÙE—•¤ ¡G`Íp•önľá¶.¯µËeÊè5¢¹ sôƒw¢zU+ø|øB¥˜¹ïÔĶ{øÅ@^*Dçk橃6Çá‘Å5Ré,¨,¦™5ÿí³™ÒDiy$TFÐFþØ´¦2£*)y·—®m,öžåé¡·G (†Éül·/š=bÄY%Šâ.⻋Üõ@éý•L@˜…|Œº@Œ”>›1üÕóÚ6šÝþépûæ\eUzjÜ+$Î:ÐÑè€p„fjÃÖ)…ÝúªaËÀÞ×ÿÐP¿·Š]ÜhÛþ¼!Õܬ.Û¬Õåü&^5¹àš˜Jœ‰íŽ29P‹çí ªl¤,Å*‡”@Ú(‚ŠO”u¤–j÷E>ý`Áä<¸5­¢wNѹ‚[à ?/ñPc³$š è ”‰ò`QF]OCmñŒ÷k›2§•Ã^r:—lW6JucŽ9<>򫭬“Ï~÷è™70%Tú'äáï_ŸN!í•M(èRÃWKqØ_æ>Bë–S~dCc§óÓKc®eÙNêYøŽÜì\DOÉeøaw±¥ÞY¸§E_ P#¶Xر¸rd@„dºœŠôÆ]  ©V Q·¿Ãh[éhWQÛHçO•$)«ñqŠ®a­1ü‘åÁ™¶00a7*-ØÉ ¤‡äÃ0Šà¿tj…‰ kg©æÃ˾T5 g;EêVwØ‚œÕ2uçiŽiDÖš$u˜3Ñgt×7­‡€Ðp’“³,Óí¢þÙÖ/t,Éþû˜fcØå'3Q¶ç(ñÔ Ã÷I}Bœâ%ƒþûùO®Aè¯~ Ë ~‚•íJ$̯Îix÷šü€=;K•ýžt¯?ÆŒ„“è,!@"r´–Œz'Ïí%c¢#š£0×xJöÓL Þ =~+àr%8‘ŸÉÝûç3?òÈóÕNûUÏ‹ë _Chô¬æ†)уL9S BMˆ“¤â¼é“ŠKLs«¶6Ç€à *šH}5¾²U­ü )GîŸ3ÍöÃaØà#99ïiüì•ÃÕíh<€V¸4„öÏõ¯ùÎ@¥ŸN5¸Ç+(‚kŽ:Zù/›°½`<7G|jÕ{ ýótlÕwh^ñ‰ë‘œ×ÊLxapõ—‹Þ ×Áèÿ¼ÛYОûíé‚Ô, …ͽ€›lD?Âã94Ï%…—õçÒ’ÇÉãs¨é¬rw>Àù6TŽ–ð–ûÒ Œr ++Î)R“Z]Ãb²õ³d¬Ï‘]K=r¡SÚɽ7¼,iâ†Ç­üòXü­âÙ'ÉaEF: ¤¤­õdšÚÆIRæ>YþÒ=ßEéy¨x¤Ý›ì|ñ ./œ8&Fi5¢ñ‡ÕǵÄ+ uK<;¦ÁYw ¢äËdWǶÌ–±óŒXµ SE—]u†xP–Áƒ·êPz¹‚öYlIczSßn³²ö]a‡CЕ»lÕîrERû@]ŸËf®e.gë‚gB•H‹–´’#üˆi<ædþ Á»îÀTÂLro]ðQ|ô[`4o~Z’Ùžø´ÙVÖXº_šQôk![–5gNbdv“ŠÈxt½ÆÙ3¥]d?k©ÖGëÃ2|˜ìú‰¡l•ÐG ?iBä¦yË5š¼mžÁïU¦ÖRµ ˆÃ×£EϺŸÝ\á©¥·Ø!ƒ~OP5‚ªz5\ƒ@¦‹ò»°kïf’‰)Ü`É¢ò1íêÈAþRÕ¯rU8–ßep126JnÎGkr+B(+ý'£­àaT'v>fëfå ]öxˆž†á úµÁm vRuùŽ\<˜écÓKØäÅ‹´eÞ{ÍÑŠu9÷¥s€îDF«ÓsѾü¦ôÞ$ß^ /Çú±Ž+÷BU©;‘Âç8é_4äá¸9¡ˆÅG,cP'³l¡ —–èS*‘,lx&׿Ñy7o,ÝéÊ<Çgñ<]?í­Cƒ•ð²ê¤$άñ%Š‘r¼ìʼ\ì[|§¸ËV~à[)žiÁ2åB\«·Ñï0ƒçN^@|ÿîf¸„€¯íˆ “ñöY’xúq·ð§&éÈ-}æ&[ Æ¿¼l·`Á ü7(›æ«iÁ&j‡¼ Êf×ÞK”([¶ïŸ=;`ÒCã¯N#*¹à¡0ÁOæýúD“‘·ÅNTÉ›ü2N^ÃŒq¦§|5äȧTË53‡Ù³&Ã×U…k"&ZD./«ƒœŸªú¬…ôM‘B+?¯ '8ܹ€TÈ kçOOqÀôl±«*჆ÖÁ6éâ/~È•<…Œyi°´ÃTŸ//hÐç˜))?¢€ê!L*ú¬ËÝ­”ãÝÔGÙñ¶}šÆÇr3_ÉP’Hd\¯ËŠKû‹Ôž3±: ð¿®^!G¹P7Ší8Ôeú&1ïðò- z€3lùæK°>­×3®âeã~("® ¯dØvùÿ‹ÍȪi®˜7ˆ¿ÇM/ëâugÚrq$°³ •íŽØ„®¤¯ÿãisÜ_ª[ü*£Ý÷¢½D¥ÿxÓ¬AUD¯CVx+bCŠr l®ÛλAP O%²‡áÔï·Ê:ÁÝY&Íü:0ä^—ÎZùþÎÄRw‹ýéú $.õ8w¸ƒ«ùÑü%µ¢ö~ä°”ßUV;kkM{Û¹ÒÆ·.h)±|‹iYjÁbOððÒú*`@‹P´0Ó Ý>5…ÿLïлÆ8º/wÃ%Öüw\-s|­J½Ã*gÑR‘¾Ëmßï Ä4¶1Ï Ъ ¸Ñ×Í k1zM‰"e¡DuSÇ/[I‘T VAFÛAíoäq¯°Û“ådAm÷×3s˹oMK‹0ÓQL Š9@:X¤z/\§Ë"þRóT‰"%ÀyðsÙøcTT&(taM›¬Þ©- ûkš±ƒÛâ…³ž»ÛõOówæ!¼OCt§±£…:áB³°Ú0)‚5R‚ÓÅZ@ÌqÇÿ/Ot¿È,:"íæ;ý8u˜:s%s|Òä5áy¯!™°¬•9d?± DÜ|ÕèÎrùö‡¾šrAý“QvŸÆ`és¸¼læÓwŽ"œƒã]ðÆ4í©v·ÙP%m˜§zæêTtxõ§ã}³!»³û•ºtíÓøFÊLª›"òúÌãQ*~®'Œ «‚ª)ƒn…Rƒ5›þ>‡±®SC’Á±ƒJV«Aî·‡Ýpw½P¯mj¤±Ç€\øß'ùq\)m'~=!²" 4üÓ}†6:Xs‡O]ˉzÁþ_QJÖö%Íðd0„e°~ôÄߪšÓôg4³³Óñ¤ÖD#JÔ.6ÙœøScÜÀ›B8¹‰ñ¼GL–ô†ÃÈ Ç%9Xº¥wX¡E}8–£êþ „ÓþÝ’7:(¸µÊܶâÀO:î'÷]ò¦í0•åøXƒiƒÆT­}·!šPC 5,Îô÷×eWÔ”^Ó)BwÌ ¼àvñâ<¯“" ŽÍ«ÛŽmþdC_N ‚Ð4CtijÝìêeIÉœ"Îë-ö Z[ærNeH×É\§tR˜äI`Üÿ~ˆþêSÕæ«h¤à-r]›BÔÝèa³ S{g$³ñ“…(s•ªè£?+NÌðÞB±ÜÐܘk«¡Á¿¨Ä61v"‚oÜ„nêvxL;Š $ˆÍ?[׃µÛ¶•zå\TÃÊÓk9ø#ú™mlMß&q<ú*Ú>¥‹0Gi¦H|êè?äûÝ´Ó)ã¹Çf5+õ“žàOõÚMD]´ÓN#9Ýl™¡oL½ÀÌ~ø >¶q˳—„[5¡…ŸI™8ò»¹eWTÏPø¯œRhíƒu{ÕHŒ"Ð~óˆ¬J@Añå%øáð9 údìÈU6Ë?$mln·  ÷ùëÄ€íô)˜e»œÍ©xfY㌖zãÅŸŽT?ul¼?0Qñ èâs¬º°†¯r GnMƒ(.øœßì9Ž{ Ú'(U÷‘JŠ)Z>åþ°ã¦çIé(î œBŽì ‘'C›²¬§¸MÃÁ¼ß.`І 7qãÚÒ«Ù§¾îm)”…N!& Vrü$ö~Wµ}7 ‰ÀzCV@Ô‘…öóDU;KT, €¨ï”°k„?rpñäÉ©à2lÜaë—áDŸLss"znç|Qq(x/£ÊßXje*¦ ç¤0ÚMŽ©XÐ!µ¶à…™zÉ—û˜’þ´[ Ç48ˆøÔ­–­ÇQ¥Ê˜nçN¨‡M̉þvæRö6t¢xw—,6ï¸'®PmŠÍ/YäY^êo] l5iêæ~¶ò2 r_¯fÄò}6–`è 8îO2o*°å~êè¨\ ^ï”Ä$P[ ´¿SRêGü˜ôí?ù [Ô¤ Úv«å—v(v‰8¬m{L¼,ÙPØs@,íşȸ‚¶Ù{Ú¦ƒ –A^øe^†)äDZAhÖå×.+xX¯Ðæ?®=çÓÕ¨qZ¶€ce¨æíI÷ŽLUUŽ5ßÚøê­{„™(aöxÒvóÉ ¸X]xê Š2´WŒV9Cƒÿ$@$%O ÞùfAì긒K¥à¯"}lëÏûz¼œ´¸„fªÒÙD醫øH®©QÑ¡Xâà2²âLØ8þfÏ0¾R“¯)>…㹺a ów3µW¾ÈŸb‰àü«ŸÖh¦aüz(Ž»NKº. Q d0þùxì¤À¢6JU8¸oµ_ÀÎcƒµ²Ìûˆut<¦ž¢#µ±ì€}mÄä¹s çps5¬Ï KÓE%e*{cõw ¾VøŸqe‹¼¡68‹b€’ˆÍoríXNéñ$ÄÅ‘QŸvæ#ÉÚŠvNV=+‰í/ÞìBÃLíÁf€ý®¶0X©Š¥º¦Ú¸í•ú?‹ÖñA±øÙÊÚ£øŸÛ‚Ï‚¼˜×›¨tWà ¤ªŽ}¦—Å> …ªšÇPK¸Lû.„g6üKK~z;oƒèyÐdúôlvÒ‚ :”(CXŠÎ"Œª×ÞÔqIiã¡[ï§Ï>W1$æU±ç3@w{Øâ[Ë1¹Èº®bw|Uƒj¬:¾Ž.(qo‘ePȸ°ÌX:­B K„ð™I’hœvAA‹ÝIQ-5؞؅«Y*¥¯9%S:þ‰Î`@6Ä Úx¿? ÙÞ1Ä虳UyG À£z;ÞôÞµ²àƒRaW;,•žn*ǬL‰³\óø,æ<G¬•<¡QË.Õ1]@eÉ@n;w´•ÝOºìV„©ªÖ)A(Ó+XÎv°‹ÍÇÐíañ¤åosdhù 4lV9Å#ÅièžpîÙa×+5€Û@—“VÆ—E:Fêå.¾ía¾óÊeU \†²‚‹Ìpû%Âó•QTcm£¾*÷sµ®V¬„„$C¶ë^Ú|­:ñÞš A…ž*v¨NàphOÞâKŽ”éKî@Ø«IŸž=Èsô’àêd7×”Äù™é¢/“‘·dsýúcØÒ0“ꉵԖÞ4½ûv §¯o.#M)2oñ+9$ï ‘3ÇÓa¯¶wkœ¨Ù' &ƒÛ„D3ÓŸ¨ðáºö^ð6šìñÕ”>•zêæJüï H¡O[àoÀ½O”_ÜFÖÂøñ±z)¬²ørçs]šÆ’Pe™NPÀŠú¢ãòD:¥.ó_ÅäE¡±ÔN ™X+¶xBmÖU0 ~9@Ïßi¢“9r¨)Ë@0½hÞaN#8›øIüž÷ø¸€ŸÈQÙ˜qOJLØZõc¥>*Q¾µ"Á¸×€5zA U‚²zú{GC}à¾4Sµ*æ„ìê`<œèƒ&Ý'¿0ÏgÚbŽŸreDB~„§Rì>œcë: —˜b¯³¬ð:kC ”´¹ùŽùaR²¡Â³å¦uO+ÐæÐý•vŠ'câjŽ/s|â`ˆ}ƘntlØ2Œõ¥Z Aè6æÂÕQ©è98±+ÁeóÃ÷Ý£eIñÐ;Ã6Äç)èÆ ¬¡ÖdF6üÔ›ùÆ1gƒ^˜aÈrü÷•`ÒºªËzág^6¥“YLUŽÜùÊvBÜ,š6aºÇI þg®J Ÿåbµ³ÓÜ*ß¶»ìk΂]7ŽMi;Ñæ´¨@‘Ã2yüÛL©q‹¨LµÏ)Òy¾@Vä6Ÿü@´ˆµæð²†@9èߺQšËÜÓI÷Ôik[‰ÏWÆfã?Û¡}•¤ôk›IÚ-a£B ú˧ìù~Žƒimc‹vqû‹ú‰ì(§²6í«e€÷žlˆJèšû¶ÞjÝÄy òÔ N¨2ï¼ 97mdÓNP® Èí¹*ÇC9&$€ ççoˆb¸W0å X#\Éÿ,ƒÛ“Û"(€Iš¯±jhØc;¸¯˜ 1©¾[PK£e;î#z=gÄEÚ4,ø˜wÛO}$.þß‚RÊÏ,!ªÖäérCŸFp[ LAåãÓeT b½ï€ÆÌ.¥-çðž¦S+a¤¦Pr³41IÓóðtGH(1­Î»Æ¸±/ó•~˜¬®Å±;p¹Ÿ«³Ur°áZ'ºœ80#éá@ ¡²š>ŠøCJÙE"ˆ`®,ä焟Šâ^[‰æº‹`v‹éxkEogÓ¸Ué&ªI¢"eϋȇ:aÅ;Ž&ÂI^œ;%ðø¼(2ºwa-TöËÆo}óæ€ÚO¥ÒUu­s¬$t¾nØÆtÿgúѽ•×:V£KË™gëhoÒKÛŒyO]Íb§ï’Z`‚˜ …Í”T$ä5›ªä_‹{9VI`/RÐ)~ÓÄãSõE¹<ÝÏå7”ê¥èÛ,$ÑÇ’€=Œg™¤[é‡[}Ì%Å=Çk½êÌ+À éülÍÚ‹a˜µÇ®RT%êk(,cx®?@›«IîxQÒîîá#£í¿pž.ÄÀ‚­#,ÈTn†_AüVº¹‡¬‚¨ÑP"å ù,@¯¼…W÷³2í 4öŸF!±"  —o7³ÿÅS÷ª—l ¥tåŽPºU Þ3ƒ »A„Í“V¢ª,P×Óµ¦¼)†× ¨Dïk»vmo&Ê’+ó‘D|mVí³ßØ‚-FhNæiЊà©sì­šÙ·H·†ÙÂ{n¨üyÞ9½sÆn|’ÉÄSs¤Y¡úUž Þ‚ Ë#¹´Ùü(˘üÏnHÁ›êûûÓÞĈ¯>Ê‘—ruo[„݃CÑ+ 5 ›Ì“¯°›Ü@óL>Ý‘òSÛ䃫àÖv†þM·K§½÷ìîÑuzŒ>­Žm†„W*Ó¹Šëó/a­r“0a:vÍñÞ6`Ȫ^»8áý4Ò&ñÙÒ²F‚úúè t¯”ËNË…c1É`­Uò·ôÙg ŠÂÿ}ÐHªMþ‹`§bÁ»pKžî0Sº¬²Ä§¯ ägÓdHQ¦˜äÑyc•æROs|A….ß­xÙ[VÖÃ<ŠâYò»¦ÞˆÁ¼³˜:þ@‰F}nXrÞ-“‘³4Vï‰q\˜Œ~ÇšŒª1uê9ykßt{Ñ‚|-¹I#jƒ&´mvÙ.·÷NƪCÀ\ªvH‚7ƒð®A¦Õ¹CK ï†!û!I¿šŸÇgŒ…x\HBÕ J6ìö–È5úbÓ›Ó—êŒéâê„óŽõò!fÙâ‰YÛâJÓ îPû% šKh4ØZ"rf‹Y¶nëÔi^ë^¤“B†(T 2HWk¼CoÞÖ7]ð\†d`,pÝçõsÛiÜP(!ŸDªÏ\ '¥<-ÿËî ,ÀÝÐqŠ:Û%µá†E!®¤65Å<Þ[Kíõäuˆ™êìš^‚½BÐHqѸú Vˆ&ÓÆžô9‰/a,ƒ3°ë/­tú8ø­úÍñDßÔ–šV‰é¡âuökFö›ë¿¹ Sl‡9ÕÅlK}À†­0ÍÓÃ/Ð êÛvLcb˜R€mB§6 íÉ&Ä”Jy»ˆöápÖ·JvdꯗWØê~6(h«tK+&X£.,8¦õ¢îÂ8ãÿYp*$¢ µ¹y*½6çÃ;o¬áðçÀWZ^<™îõ•U‡õf4DŽÉZ¨Ó=¿"iŸ¦)êÏìPæ%¶ú:¼›si”ý1¯â¸$k•zu plB€rÖØšð‡ÿï¿«ƒÃÃÂO¦’^Æc-°FT¼àÅ£ŸÁ{>M|bì}ýœOÔ÷lÚÀ‡ü­’¿è´”îÚÌxrÿkñ,—Ò)³xª‡wH·SÖ¯ŠÈÊØÈ*j‡``»iØ„~þŸø+hÚ‰ùë[ §ò%l ƘªQÿÜçÔ¨²È›YšY&Fë[¦W}?H¸^À]1›˜ùF G).&ì€Sϧu¡Eq¼ÏapéÒ¾åâ,ø0£í:„cÏ袚¢C¥Ýò©äí–^/=!ã[~«ÔCœ–|Øñ©‰ìÒ£w}Û5n‘>pFí!„Ï®íg‹xøû‘®À› èîûD3 jô~Ãb´ú@±añzP´YVàÏ¿ì‘Pùî=Ï…92ó.1Eq .èFbâZw´ŽØY̧§c…R^Cz4([±ü,ªîÃblŽuVÄaÁ,ž$q*_ oõ2÷­›m_èêG`ûvÁÉXžðÅÇH¯÷– ÷±jE¶ú>³6IšÍ³}¼¼Øb2n9*/3VckÑë{œ®ÞN]ç’v mcï†dœútòqAúȤðK8Hÿúáñûõqê™d^1;…›°¾X«w.Á¦ «. ¡-+ è)T±7Óv¸_©êÈF4–nÔ¦TB¾ös«ïkSáÍw7­›àßÁ² înN ëM.*I¿ì‰NQb)†y¦àÀuk†5¨Îzðên` .GqåM"—EESÕ0äëNŽ‚=ÖG˜– .ÊeJ^‡ ùÖ.XïÉ|cŒI |Õ¾‘€É4öyì½Ø•>¦PeÆ !c¬IÔY´#é£CF¹6 ,>²%¤%Ëwtc¬Îãääß(DLa`zêç‘—»œñø,Õ»Õá޲PndsA†gÕÄÃGxÞÉ buê7ƽ—‹åÏBT{¶Põ=wÒÕn¬µïæ"ØîVtWïAA6`ŒFkŸò,@!0ê’=Š@¬s/ 7º´ì?ìæATí×t‹}*W±U,™µ2 Îáܵþ?]kgÿìÙ‰¯¶ô« ÂÇ‚¬ ßÝ£³ˆ(e}ùÛ× ‰—ýãyÔú`Ë *n{¼ÐË)Ê^«ÜÊVÃDUL/»,S·s9ô9mDx6Íâ±4`åÈËd(a‚€ÏˆLp-i’c” ÂSŽš&ËÐʘÓ`“;®SãïHMÞ²5„n}–š¯lÞ­j%»§I)3<†•V’²­ßÔHüiBnJL QxÊ o’Þ~õ¯_1­š—'uÜóåo9f·>dd—3õx‡ ,g Ñ¥n7hh½°aÐX+[¢+ŒÒ‹«ªÊ,(ªÑn]Ï^豬bOÂ'ôV±‚ÐX¶¤„w/Æh·x”ÑÕ|‰’m{gçýŒFéêpˆ¿I0 •{EÊurƒ’Q*€«u€ªP õÚšS;Þm&>³×&ç|ÂÖ‘‡ìâ=r˜A8[êð¡³×@†–Ù4]½wÒO–RIþ]øTážÍ®D½×#´Ba"ÿû¢ó—C ƒ#«n û¨rÃø.#PÊ>h‹Ù ‡–^•ñwŒGEijñó’!o¹ìÚªàNݧŒûB*<”µßáicÃÓȶë˜ò笩S¾VpõÉÍÇ}=„’Æv;_A&þx έȢÀÞÎ*—ç]‹_ÇöŽêXlüÚ¡Úú‰–Ȉ¤­¨wµaÿô,°ÌD³¦j׳ ê€×„¤Ià§Kl«4Ð[Õ’YT»(Ü}EÁv½ç< y‚”€Ü'úZÀ¨aHºCKÓ ã´#˜]¯nR6`ßR|I^ÓK/BiJ¢šsaþ¨…¨`Œëv0»õê g™Æôó|.Q`³{(ú Š {eÓùý´Ó½=ÃRú1Á”&K-i]ËÍk„‘Ú#qó¿ô™9òàm¼DÙÁ„ež€1~qÍsd+Á©W99èNko‡–J30ÍVÝ9Y/>2Ìç¦>~†³jT­ЪOƒ,®ªÛÇH¢þ^æàg‹û¬/lÁ¥ÕZ'jÞ›ºNAé€\¸8Àâù£Ê«VŒ!zgù1m ŽŸ²È® ²EOwäNS¿\ßä=ÜE-¿hå‡Vª"ÐÕv."ËPš¹Vïj+?Ýðš!Ðæën.ªrü¸B«gx2°êQèš-ÚÎ>¹<Ì“±«#C|í‰R£BvJ¡]Ìl¹©SòÊYh°¼ÍT*¬ì!ýÙ %3ütÃþÂIz=ø ¶ f@€Ý‡§7aœòÓ4J™º/ jP#g—‹3áj 9Dã›T ‹jÝ 8µß%9mK¶w~I«&¨Õò‚ Ì|‘È-J—4ëô²&¸õrA,ÿ´e1|]/ŒÓîÐgÐÛ‚gQgçXi`ªçš’ÓBj?h "NhÛò‹-Y¦…°˜`R¢^ܰ,3᥃¿O5c®RŒ‹/n¯9šÐs#Lr£E,í=oçù“ì…ìIÇSKqŒÌøEÕ è(¦Ýø§šÖãY³ºhã6¢P+r /nwÆ{RP甫[×^@uÞ6ÆH¤ƒO¥ëºÜá•Fða@ƒz©~Ô×B6~Ð÷}æE"³kj ì BÖ»Û¸y®é퇉)ðèôüUDc[2Ý"øM”æUiOú4;HA'7ª¦ÂŒ…ÐÜ‚è°v‚¹D]Ÿ9ð‘b)…»Õ—&»nZ0²q†Ú”síùŠl¤Zµ_k4‡ÐÄû‰þÛ|ÀÊ)>L÷QÜ9±/Ç-0¸£Ì@µé-x@,UŠ%ÆK&`p!7§sëÌ›rÎÉÉãpBêꎡpOH›I#äiÃrаþÓZ§œä©Op·0‚W‘9KAù´©ü˜Ìj¿”Δ“ Y£VtCšÌg<’d¨Îí6üE<§Ìà¢)¾[¿2'@É7æ”G"P,Rs”¢|it)'ƒÿŒÍ€R–½rcû×%Ùe}ð¯èüG÷ ù",˜ üUÞE.[ 3Ä%Çäo¼Eº­F·Žƒ,â–”:8‰fʘÛäéÝNþ RþB'²¯~Â^Ñâšo=›_¸ ´U1¸;³¯6OùÄÐþ¦¿i×ë>¹'ä0¦ž¿ÿš£¼k^ùÚÛ{p昼;žÙÖdV§³±¯WG¥ßåܧˆ›ëÁ‰}«dÝq¬Ê,7ûƒ„† »˜ZÂÜ.r]“QðW€Ð1ÝÎé $¹Ž†˜ÚK-e«æ±›í´è0=–t 6¡œ6¦Š·-vQ|?bׯ»[Ò³/ 1»N?ÈF;¸ ×CÎÛv,›þ„vµeŽ¡âÓûᥥP§l³‰‚nãm¹å z6$c†±ùje ÖL`•ñ8ÕØúåý wLà“Ãð2—î˜)‹¸ù’fä2î}Ý[—DH{²à›6^ÓTñv´ ¥¤™V?Àž- ,ñtí&IV:ÌM„Úæ…BuÍBs›‚®’øá²êðPqhR%Éãp— ÃPqù’LÖ×;t.r¦ø(àš†Vœw"®ÛQTR­Œˆ‚ÂÒT„ ^´¹ñtð—GA?².¹_,Ÿ=ÙÈ „ÑÝt™uòS¦òáÇf¶æ~ôDRqÊÓ ˜B{wÓ~*ôGޥߕñ|½rmúæì¼ú6ŸtÑ›ÈXœN°øôë«~„Aç±#´ðeõçÓˆ€Õ22!<ëaˆ¾‘4[Ëha‘ƒ¦@(è+t~ðHHuAóáþ-é…(Jo¾R·yñ}ßua󬛠çÝ­ÏŸIZN õH:Í-  B0!È"™–<$¨«©"‘¬¹j8'&«ŠU¶> ÙÉ^[[S)’‹ WÁg2÷ÖÔ/-xˆÍ™G݇Œ *â%¹¥æ|åp©ÐiF㛤é)öOí³'9æRÉñž:Øð%åqÓ«œí ‰F)M›4ˆ¢§Q#G›â –tÎ>ÝP¿•a2Ç ªað6§.^·g`ÊBÇÝ~¤"`,º]l³¯ß¹†*Š˜jép²—QÛÅ ¿?AÏy:†CR¹2Õà˜>XØïNX½AZ"ï?_ØA9Ä*ÿ¡– ë³WÈ]çÂʄnÎ9­ŽÕŒù(÷†} â£òó¯Áî‡ÙéŸaµìlð"„Q­¤&1ŒyÍs>õ›.”ïŒ2áh k@ÂM9ÖüÞ›AUËøòèu9±Þ ΢@‰K¤kŸ?a †LLˆÝ&Ùi™Ud·°h¬uLÄ%£~—%Ô(ÂòÐø‹ë=æ¦íÔáôé“èý”¹¬NQÅÑ“M'\I˜6à %—Û©m Ñ‘¹Ü¼hpa{‚=õÕ 'árGýªO.PM~!’÷µJÕr'®Ã€ IPŒ 51ršÍÔœ¹P*{-žn“«þa˜+¨sŒ`¹o-ÜóÏYÕĘØNõ€@0/n!(2®œ±³YÐòÂnñ÷MOÚèhñê­øw-OÚ𨜠žcXg˜w~(ÁxY?6肪¬W" 8nQç§a°ígw U—kkœwXÀŠÌ„Kû‡èG…æ8mU(À¯ ùÌöáÓøµà[Û6ê+Š>—‘%åAdˆ–ïæLW“¨H¹îJ÷îÇ[Âr‚šŒØ"< Àz#ë9 b>"}<É_+ó £T·ùdÊ-_…»ŠÞ6Òo§åv÷Çx+EH š6<!Ý ƒ•Uƒ~,}‰1µXá,/ ~™ePÆfvëLJU”ð[{Œ:jjË´"Äû²Ág³Ë­›$gyøQâ§Åèò°­˜"Åä=>+œV~aýbœÃÒ‚Æû’÷t;Ë꟦,Jï—rowßô¹ÆnÉ67Üö?%ŽA¿Qòé!ÄèšíVÞÖÁ;ʉ4(ËAõ¥îæwêïF=½‚ŸÁ0¦™r0tã Ýë>ÏAåæ­•´ç«°iQ§ŽSÍž™ßÛÀ…üEòˆ¦áõK pé>§\I´5Ç_ï4håÉ_ð“'®R.i›zvYÁú žE5A„N9žªËJ›µëàÛ©j'ßP˜{Qœã•!šÎyCà3X)¾o"¾3lx~…œj|Z¥†¹Ü–/ÌÈÜûò7÷~±çí¾Gƒ2jøÔšZB¤ŽtIÛÌê'p§aªÔß\`5Y"­í†K³M1ô˜BØd_%ÄJ\ªfS¯XcZ¢#ig¯,kÆïÎ"%»O©Þ- 9ôGïju9š:=„’Êžï‹¿Ër)«þ¥²IeŒf±[ž+qóȽ0,ðQO#Þ¸¨îr¯Ö] íùé“jB÷ßbÎzÜžˆÆ`Ñ+#WÒúç¹'q,*ÄlwÿüE*càÀ óolZ:š.½_S %‡¡KÛîödÑã8Ú¢ÛÔä†h‡¹E»ÐŠÜOBÄtø¥R¦´õèÂî€z,a œð)ŸŸ\èµa÷¦‘QVŠžŠwòòV²R‚4»±—P®Eoä‹ÀÝ} É£fn×øENx‡¿a(¢¼ù{›v4þËwÛÔ¾€¶†}ïz{±¥NPßÅNÜzR¤‡Œ xx…A]ÇƵG„5a­¿CSQDã–ËܹOö³Žž~ ©ÇÇ+Ò b«‹)9âëLÿXÛœÉ1eu¶˜=ž¿$/ løQXØBÊ")aíÀÄX@u[_kÚ_dmçÔjøn0Á£#»¦šoxC^ëeG„( ò_”É6k]*Õ´aYü°G5FjÌ2GÇ"8ÒÐ’ îïëlWóNÈ ²#ųxûÐüš£iC1WøzóçÌmEm‘A¯‚Å$ÂKÖ^§\L›®Ñ¥&Ç*ÇÐÖ¥~þKÙaŠóï •²\i¿´`ø ÃÔ`x0S”ɹhMtÇ–äjîSªð‘„p‘ÑÍ7„ýÀ”c!hžÄÞÜ¢GQâ²Nm %L˜›IìoÙö‡o%KMæ½xßXªLÊ ¡H©Þ5›ˆÿ#ã¿ðHW“¥©,<:¹Üçù§ÖØu»,ž”éróÓ_^©}À= hdT³Ö¬Lùø“và›È+G¼BiÒ]#ŒÊZñ©2Ð smµ7‚û9Ë ;«Ó&M‘ƒ&_º?Tò­C„ï*,„OÄÞ*RL@š(õäÏŽD¾8<æ,·îšÅСw˲©`úa”«n­óséiä65ç­G\Iù”n)üq/ÛàFTÇWŸ[¹Þ¸¿ŽÝ73'“­)£WЄ¯(@AzÌ4ŒQl’†0x~ñ€‰n|‡h*ÊyÀƒ•˜âö¦ô\“l])W›6ƒ?Æ^óõšZ EI$ q´4’° ûqù”RnUÕ*ñ 9 jjTñ¸*´Ÿÿ÷EŸ+½J¢ýíåvæ]Ay4õU?à wBÖ µÏ;Ðâûì‹ÄÓ?€3…0÷ [è®’ ³qÉ-!ªì )L:Þâ-.Œ¯ê‡ïЕ,'‡¹¶YÂÐÁ ɫޤ ·ð7iß’•/Ü4DÔ¢¾Š¡ç%5ŒTÜ1 |mVðin®+¬—« Ó<¯íÜ©ô\T“Á r•[!³…—7Žz óʦd*1GMÓ"Z±Ùâ|sùˆrî<œíÙ‚’î÷êÐ0d 3èë$b$WGjBGhŒéÀl³¼_!ãªÜ4cU–uÜzõÁ¸0ÛcdèCf)…çéìqœþ¨ƒó1›)¿xÙò92ÛäŸ]¡dØÿ/»ôWàÏ"ìvêº8‰³vþyƒv+Õ&U…NrŒpYš(| ÇL]¤)¦žÐ_›†5Âtg‹Å Øp¬DÍæ‚ë?<7ƒyø _°Ç¸[ÿ¹Šr0ÒIí;­®lKÑZ,cOE¹÷<Òå=Nö¢Þp”n†¤%¤åg¼´À¸ë©¡cŽ6a‹»îØ»´ Tm Q Ÿ~¿³¶â´ÅëxZ›n,Ü(õO9e]•{qTÆ£¡/IF¦çˆv­š†]ÑÔµ»=,Φ£s:ïPy¦ø€»;0©Qµ¢U*ª•mñ Ǻ9è ÖúQωfg“Ì.x8º°¢­çuËWó€”zkGmÈ*ÀuV>Ï×侃7Iº“%oE¬•¡·Ãþ ¦q\Ôö«‡’=Ž,Úªyù>¿µ±öU÷¯Ä#…5ßM±V¯S»/"@ 1Á3Ô@¯ñÍ?NßÈ•e^g·äœdâêÀ1 *e–ø‹š\Ó6Y·/{)°¥µ"´™ì?ƒ®xúaQcrÝí1uVaþÖ]L ÔÏŸÅ£Žöd[=IÒ¬Æ% Ôÿôîõ©v±yåo^éð-n䱂µ›XöX|H¤5„Žö3%ˆ²–¦­tUµÔÆt;ã‰*yD­ÕÅ7¼•½šKDq*_^+3nœŠÄÐéØa‚ú®¶B?¶ä³+†ëä¿Wœ4>¥é–-ºfèû$€¬Š>eöø· /SÛðËærÂß劮Žî ¥J1Žðfªñ-+¨5ÑÙ©1²çÑ E,àˆ-™0×êGéæþú·ÚVô°»ˆŒÈLõ lÉw ©…««Pg|Ê‘B¾õ“å2רK´àBwm× (zª¥? ôwÐȧ˜ƒ•`g¡“a'Ûöƒ¼¹Ë%?e¼0úépðá2 9–ÚçæŠì0€yÖ q:‘­ƒ¿) ÒôìW¨žJÆy¡ö¬[€ªušÀö}‹!ÍÀ±¢ec÷)",b¢³)g¾ÅÉ• Lv‹E§Þ…I1Ú:#´£geÃÇ{3„‚€'ËT ÃíÞ‰uÉ©2a×ç»Ø~ ˆÕ%Lˆñ ƒ‡ú;íW<ÝùˆŠj‡vTªµ  ÀdÙåDBö<«Ñd©˜AC"â4dŸ"n¥ð¨×ð4«Ç®¾Ôc_‡i/pØM§˜¼À6>€s/ØÞ¿qœãB¬q ƒî\T;· mÒ¤½:6V šýã^Öyy^™Ò¬Ø.ôïræ±£bsu/$ͤɀá€ÃHE?ÈL\ä×wXyë·‚mu¾x¦‹@,›lgêy¤“L·R<2ªùúߥ)wj¬™=ØçEMT÷»Eò[C8©EìlÓõ xYSLÃÅЈí\ÅÒË+ñE _È»±5nb©õéø?, |4»X¦0‘ú¯Ÿ&ÇÅK :D_… ¿¬uAÔ4cPúúw‹µašpɪ$œ€båÆ7-ܤik7ºäÅÅNV9j¨¹/ųHE5ï=YháËEâ(Ì[ð ³À—Gñ[ ;º÷“¯ÈÀ\nuÍÔnÈ-%ÿØAÀ%Y“T"Qi¿º ñNÄI¬µ“ê<Çý!ÀßQPÉ4¡¨~l¶52@lì_"œÆaŒòÉrÏUeŸIÚƒjJR>³×gbòÙ¨§Þå0uuΨ`’x4ˆþÝÞXÀdž՘IX±hsß!Æé&yÑÃÂå”eï÷ªšaÀÖ=L¢“ø±¶^\5Ò=yãŸ>Þ~'Ô† x¯Nê¸hƒ«‰xæ².ᦎ I ¥IÃçNaý]ë³ýÓ5¨C¼m¶ÝgôU+Eeѯ\­ÝGó8>lP¬&¡\7l Ç´‰½Â|X® Ë‹™Ñ#†ÍÃËwÞ‰¥c¬Kr+¬'WQ?V…*^{;—qg\‘j•ª3@c÷ÈÛíSÇ u—«ˆuh°ôx¾0$ Û¡ïy%ŸDÆ;‚EÁ·»!éþZ‚Ê¥B˜7[¶‹êÐÙ(Ú9æŽg¦hTÏ•Mœ‚†PÌò=—õ;}§Rvß3â3QÛhŠÓøÅÿy»ÏVY´Ù}Ñ•üî]B$ø£_4 Njš<DΔv´aÌzÑÔ(R?<*z¶Ýìά:eöübRÈPÎU +k6Œ¶H©GÉá¼XBBGpñÍ+ÌÊ•g~xÖ*X$Šôbß!ƒsNp) Ög¼žKMTV'¥`ˆ{낵hòF×ÓœãxϦ Va;,å¯};ùHN79Ý»5)\yóÏð¥×¿XügÀúvðÖ;Û"4úƒ²Ë—áaOÎyhö(V·q”P¶J¡ í’3üq‘g.¹ƒã„óðª­t—;Ð:ß.7ø­gÊA.=HTKMg%{kLýˆÊZ¸Ö iêÔƒzÈZkœ-Ø ÎÒ"Q¿(ÐëD _àõ<ËÌ'µöƒdí&°(æ+WÍ6{C.òGFu„7åš=¸=œþ4†³¿;)‹ ×Q“ë}²ßÊ5“°EnB8­¨mB?yxgìͲÎàOÔ\1À .Òküâ*»1N‹cˆhÚãŒÕ&Z.Ûµù¾ŸF„·.(”3ÒJ ãO6稵R·’¦èýp3 K~ó"ÖÚëAëegÍ`^)ò<]îu5TÛº):½—[·<×çß‚Ê~æ4ÀYFæøSŸ8¡aˆO¯Fè¶~ÍS¦Îj°€ù"O¥ÒCFÆd‰púh’R3ó+a—>~DZx9i¦¥ÓÄÍ`RUÍuŒîf¸È˜ŽÔE]ëFu@¥è’XÚ%Ob‡=1€/lÙžÐQ(¬ö–uFbµÄp„1·ÙÏ{ãÏsÎwY' AOi›&îO¦ŸbQ%ÙÓè=±ˆ$åƒAT ?_tI™Îžß…|èËßH£˜/´°Ÿ¤Q…Ói⛉Éî©þªŸQ®m8´Á-z "ƒ‹óÏ2tö&ùðÄ”Ît™±0Ÿ¾Á¤Q=Ú}PöÞÙŸœÔ‘½ôIë}þkJm¾,æoÈÞŸBä±OÅÐu–6hE¦;ׇ]ø˜é=$êÊÀ…;šyÿ`Ê‹xN++ä]Ž?1%o1=M8fF\~ý¼\±0žj÷ÿgÐðõÖ”¾ VÞiDâo¡á¥f k'|'¡úаâÜ¿®R|ŒØ63 Å0{^{š¾—å’¶¿rígè3´)À1Qwœ‘Ad‹~ï`ŬŠÍ%385qÄO;|*Åq\k¬+²Ô$xi„ ¬6,&Œé¦{í:OUy4L_ɘá?xûˆØš)^Å=(þg;ùŒöÙxË|w‘‘…İ)YÒÆK¿Q(ï!ž"PÄ!4¾Röü¯QŽ÷gh‰)*;å2+ñßá³äñv¸áŸ¨›µ£ýŽäQx$™)[À: mS0”9]Mc› *^Ó6ü3ÇGV¯¹WçïdW÷ƒM¶ÆÝñ؆ê¿ÓIK{9+åg†Òeø»Ã´ó£ëü:¡2zÅ/ôêF+RÍ4Úš–•zš-ªce£§ªœ\RgË@)­75w±^Ó·ËáýmÕ¸éÚÇû¼e÷[–| dìdz©à–{-ü > ÍÛì‰=PÐ:ËJ‘Nʤ ŸÎóì8`bæc“å;Ÿ ºlMI–é ?§•vI”æ£â7¥M"îo×ü=‡À>KUN¤]¶Üiy#Úçä¼»öi"~8NQn¶VÔó?€8cÏõ4Ôq¶÷”Â.Í®s4Œlï*i°Ã îVìÇ´3Vóv F} Ý›[…d!ƒKËŸ]]£øß·0‰Ò??¯»F›Ž®ìK;~Õµ¡»RzáÝå[:T¢.RÕeSë-¹'râù«k¬1ÍA–rôè×ÿÿˆNOûz§ÚÛ˜’UßY“´N)ÂÊß)Vu^íÑe)£â•(Âßr&³ñ‘×*‘ýÑjâDç`e€Ü\¢&ÿÚGS‘–§ý‡»SÀÉSñ1Ÿêè¬3VZr¨ð»'F[C…ÜL³lkgæ]D5Gžñx½Ÿ^† ´¬ù4‹±w¿2GkÄŽ+è–ðWF_Ž’2$þÃ$䃿Z5v'ǵÃ+ë§eÿŸÈ˜r†1a~,ºC0#ºT“ ·ðqÅÔûŸó;cëq´nŸ³–§‰q©¡òœâ¬ªŸÛˆ‡wö;3ïæ²pŠÿX²Ï3™¯91xú—{¯L0W—Q‡ê¾¿œú£#?jxÚiÉÓO"Ÿ´ÈB_2”l\“bâ1¬©IK·hCj<òcóÎ’ÁWºÜû!á]Ì”¹ #. kB*5ò‘´¬ücâÈsb¸ô¬ K}z•„F¨çïç0ˆC'„È8¤šfi}Í÷ n¡vô ¿ÿ2ÚFÕÙ8°J’ÕêOÎ9ð¸Â­j¨võïË“m’<ŠàþHû‚Úî^§‰nÛ…˜âlûPƒÀ3ò–¯W@v7ì#wL/"Báw~Ý1;BØd]4qÉ Û?¸X0µ¯^(·ç¹>'ÓOü’Ž%Ìt•ÐSìî,îù#©—<ßG¬Õ{¤›¡Î3:Ãn([¤gÓ HxǨK›Üh`ˆ¹ðÖTüÍÃØ õÎKÂÙk¦8†„ß”|Û2Ý<ÕJ™N$ÓŠ·´/[´l¿³Úî‹Wñ]0 àF*±ƒ¨ôAõÆ,âÊÛs~r9Ði0frÀšŠ>•»?‚ÝßÖŒÏ fÛ¢¥ÒÞÂ87ÓôåKï›hxØ›c!Ns«;ûF-Øàäšþ€uó:þ³Êþ{ÅþÒÞÈËÜ­q<é¼zÕÊbæäë#úÔ4çÑzÁµ„r$žC"Ðb£2:½±Nžº0%­§ËîçZ^U¤Ðø¦¸¹£Íò} .T6eLG , b@ßk4¾âª˜¥2ðVèæ&Iâó.rÎu…Y4ÑÑÚ ftô”š…µ)ÌÀÁ=kobËhæ®:1wóšÈ@#4ŒE$ÖˆÖ.ì¶y•òp“¶‹êr©ÜB*ûù¢^¹® Lsì·è;À&1Eø„=î\È9Ù ò¶ßþi€ú-Y½®w„`#ÓÙ\ÂÎÒTç #Rþ;6ôîòߎ7Bjn²'Ô¶P5ÿ£J”ÜÞGm}Ë¥pVââ~v*­od,êŽ|aŸTý-ÄÔÂPÖœˆHú¨>0òò[s0L¤’c³|²õ¸´hÇÁ³Fp{ûwÈ=”£ýx‰¦÷“¼eøËê]ç¿ú($Rø¯¸ÛUËJ™òXx%¦²±K*Ì¡]áá!šÆºf[)ËcãŸèZꀟ9À¹dš³£¼~­Ö‚N0…mF}ò˜Üš'ŒœÔ $õó÷¼¹&cF¹\*"yÛ2ÐóŠ”U-¸èð¦u~ýð‚¾‘Ç·ç"ŒH£Ff™?ÓI-;ÕòÈ V9(£bÛ+”’Šh¦i&®”mË}7sÊn  °:å~Ëw+2šÕ¾>EŒ¸Wžsž-> ™KÆ;ÐÍ/¡mW‘è!ti!iIÿ¤P†’ý²ˆW=L©ò\”žk#3o€8Odx†AT‚êGùÞ,ó2z!¹¾I6ç]á\á&µ] OÉ;¼-¶ZeíºAi^#uùÏÂE–i€^Ï¢Äÿ]Hï²æ´í/f!å ‚ “êU(ÁÌ×U—jÓ)ÇIE¡ÿ'(ž»ElimÜú êé'$·¾º†çÙQÓ† .ŵQßÈhÊþ4=¥8U¼XÆösGZÉèô2ûÙVßmørrÑ•ÑÌÂRÏ.C«â~A»Ü¤àÜÈuËl -€æ¶‹™?*‚E¿3ÐÛ}ÅÜ©>¨3 Ü9l[š,Ô{\¢@ÞÛ ­iLG:Á¯dðsGá!ïÖmœÖ4oã›ÉÙ£Ô£–2%ºmMâ;RiÆ}çðPi׋Úþ@U_ÃMª!Æšµ©Kõ^G]<Š…ÞëWûþ˜±ìí“Ì?¶g in€…R‚¤²ó²| •b§uìiõVÔX˜‘T ¤ÌAbêö‰iìEOËy´;F%K&æT€£§Éa,@DÌ”-Q‡ÄWËž”² R¾Eª°ð?s©áGø<šv+[fÍž¨LØU3%·iÖxMM¸éÒW‡uóLÍT~-“ž?hËLBûΛŒöÓ¸Ýä›kœû ÿ‰©y‚iôÔ©( š£[— ŽÁ~«T§‡^ÉÈìæx”»"ÚÖ$ö‡ }Ë~ÆPž£•¹e-³éÌãHfÎâNâr}¦sÃoÍj¸Z/ÁÒ—¯vGÉ7½“Rœ²˜Õfdžo4AñôNn\þèÕÜžx.ž–‡_}wtë7Q.@ïŒVDsJÈ®²½lÂVz¦,ÍÎì>·÷ChÝåZâ3qÙëGre[½í¸Œ¹a¨ÄÄ<‡üiØšó7ijD'¿M”ËÇcRG+ñëpò„ÞI‰jŠÙb’Û 3ðR§•Å(1B.µ2"yù Ï@|æ`ˆ`coÆŸtuj®n §Ø‰¾ƒHx:‚tIaÓ¡|=Š48»Ñ0WΑ‘ã9*‘/×áî¸xT;ëë ³ZÉnâb×*àKŒVë£M“rÌïÏD¡ûA8š!ù0 L´¤“òW_4µluð?Ûñ‘3«<çð::2Q<ªnmOöü/â:lòì³{BÈ»÷/òןÇÝœÂ`Íì}ÀÚ¾9Ÿ†5n|œz±I1 þ§ãUÿÑPïéë;F!֡ᤔ&…o9ä~ßvJ1!¤ñùçgÉpÝ…$’uŸ7²I‘¸¥Ð©â‘ÿCßðÐÒ÷Ñ åbÛb ƒ·A>EÈ\ÿ~%LuŸR¦‰¾]’F%‘H¡!cïý ÿÏmçª8Cž¥5¹à8ªî½J¤Ü£C<ƒæ“Ê@Þ–×Ú¿b«“òÏgñ?ƒ[êôì™…³¶ÏñÔçLi®Ñ%¯RVFšÂ€‚¢A Å«0ðÿâ„î}X~™,µzÓ Ö dˆSÿJZx³ôÏ­’Ê•—³»ORf§™åžÁ|jœ”¼­ 9Ó/B¹l¬¯è³uWòíRÌà^¹ÀcbÜþØT÷Ñg£w.ÇÔëyºíõ{)q³ˆÔ{R\‡c8°@;éöj„(§¬pìEÕi‹FjýC¹“Rÿ±êÕÑuÍDPQ1É1 ¶dÔÀ¹f¨”À«-#à 3´Yµ!ïwU!\úcRç"Ç }·Á£ÚH~.-»× (öC^–MÔV;æªË>VDöþ9ÕX'00œúvPÃNï#¤M'U„fˆQ“Ù›(5iãÃNžVN€WŽtEºžór„62ï«•PÞ”Ú®ã’+޲ ÿáXƒ¢v+äŽ}h]ñ#ž¯Ù]ƒ¾Z†2Ì£µ4 ùþyŒ©0[œ¸ƒñѶ‡ŠwÞjãmþEfD îvì¬X¤*:$Õ9ÃuCf5Ø!Ýy»? Þ4h><‚áJí=ºé–ÉLkTz>Ù#×W̫؜¶§èøm, ]gýŠ,#§º§’ °¶ûI· …ÖÑÞTÛðÚÆ‡òÓŠ¶G³P•Œ0sY¬ˆ\û­H%4Uí%dK”Ö굟”§Ú¼W¤\º[¦» -‰ÀKâ¿vXVÞð7#$áÃg‘knÜáMøPVÏB9"Ëàf:whÛñÒaöÅÇ.ÌÜ2 ¤¹'Öæ–^Ù¾a; É pܲP²ó;÷›E¶µ¾mHª Ùk|Ùst“Í»±¼PKÑÃŽ_ñåõN•xã…@‹=åàsÜ·Ãs!#ü2ý $ØmßÇ\Ã1_›®•釶[q&v ÏëØŒþ °‡È4ë#ä’9ƒ(bjñ$ðÔaàú×e8»PsR7¹¢¥&åŒ.‹/RcbX”û¢ñŽñƒñýó³ _XªË†*ß©^ÑQ .7?™½‰?j(nš²Z%T§8r?],§š´§Ú mê0°9߀ÇÔ}ÑÛá:™`à ÷¶~ŒÿØY{½÷@Ò` »ä §$*¯§IœYŒúd¼Oü2ÒÇÁöSìã=­gc è361ÕÜ”bߦÁ4ˆCmuã.óówA´êE½¤8lõXPï—‰mQ€K×[i˜²šÇÓ@ÉnÈçIÔn-Z…]ùí€ÝÑc¹Õ„äßZÊa$Ìñ>…c:ïÄ!@ÌqùƳVu‚É HÞò½º+¢t¥ôÛ£C6qy³b~\*±Û°×…Ù0`×B€Sš¬i/q‡ë£_hKº¨Ç3}"¦”š¬\œ"¸ ’&|ÎAb2r9²asÑZóáë\¿Í9gÕÈ+€v…vy–VŒaÿ2‘¦zÁ`ãP•çÉåK–“¸"QMÇg,.‘úîx}¯+ˆ!|\ŒÊ&}{¯.¸Ü›G¹÷é3íûpû¬Éß0½øº¨Ñu@"f“¾£se'È!t£àcsù¿ öûGô®çÑ’-× ËRXÎ*ç¼uXC¹gxP˜Vv„ŠÛl˜Q)Ùé”ý6˜X¹7ËjÅÞ´1ð@u `?‘Ú[N‰O^‘yvdõùíL.Q76º<Çô 5bΛH¼¾ÆíÇ¢Ä2œL„°"׳¯;–ùROàÖ\rm Ó{O'¿ç´ZÉ:ÏæuGM6¯—¤­¿ëã¨IÅ-Šeëˆ?,ó¤Å¢!„ù ;|ÛHï!`Á¯#DÃ5è`N=¼% ìóîqpAÆw£A¦t榌MJɶ­ˆ{à˜ßÇÅwmVÓ£©r Q DñÝLx3pàˆ™3Õó•(¼"eso%«A¬Vgû΃]ÍP Þz.´Pi`üžo"3ÂÖü,›.{…3–-|$E&Í„ð¬C³´3ø45CkÍÿìéÕí(SâÊqŠ 1x`Qa“JA䪈AY‹ E —sa=x£G£Š ƒ7OÅŬ¯9eI dÊÉÓçªÒ¶‰IvÅÚE¸d)؆sfbDìäìO.*“B’¼¯ÿ•{(J?Ó‚ø8±k,Ï“r}xÁü?ËÄ…D«Y΀®–Œ”¶Hœ×·ã÷L~‹kCÃߊ?¦üýÊö±ÏWh.1²oˆ^œE€¬TfõØ•_0ÍSýe•ø¶Úƒ“‘¿e\Ñø2t¿óP(ÆÑ¡ÓŽÓ4F€¦ÖV·ó1뻫™Û@“Óg~·ä¹•%Øõð*XÔ§ÿ@¥TY|óÍ;HÀ‚•v_ Ù>gXjzöíÿXsîp áˆs"›eYlÜ.êÂéû~Ÿ¡ã5j^ÕŽ?|¸_qý€´ýi«TbìŽÂÌÀô”AŽh9Ãäéú›2 #IÂWr~}šlÿA\vÈÕcÇ´>·–Ö?áJ Y;rNCxáŽâgpKàú%˜m\Àwh^ Ú?øQxÇÉYÉa‹vÞ?—l­N©p§nŽÖ)jZw> œç§tÆñOL–ÈFßµÔ#²¹IÁ¤àxÔJŽÕú;„ñzhí½öâÕðT@0b@ľav !‰0jû k#¹H6—cË1À’x4»ùÁ=ƒJC‚¤‹D4¹iç6÷h öDóÆ ÓVÖACùrHè2šS¬É^F‹S€Å+±ðY'k–•m² ‰‰QFš=gÃj ¬8Mìc÷î²m\ÎhÑ ÕSÔ/s[„ö¢½u”c2¤‹äR{Y.VeO7RP¾‡•c/ù3®f•­`#ïPñYt«éî¶£ÙYÌ´ºÆð{øwaÜêO „ÖÈ=»½ÀK~_ÁŒö÷e‰4²=˜Dk"ï‘`­¡#ïnM{Ž Êl! P!Ž/w& 3ÿñ†”‹lòÛm̨j¹Y´+ðˆøø™ü4ë¿2dá]×ôGEÿuih©ÝZ.泡Oó–^ڥ͙7D›>†‡êCu-‘’ž¾ÛPqnn3ŒKñH”ÏÀRVO£¥íâ´¼ŸÚw2®'¾óB{Þ‚Õv˜ÞfG²ZÆ~'…‰‰Ú¸æoilV(©”äg+îÖˆkRÃÒM‹—¡¾„Ú׺iU93ïPð ,Ç4ä©X=I‡i‚ù~±.{n"Ú³ÝO‡b ÛŒûŠÓÄç$Íó»FX<Vy¡ªîx·j~¹†eñÛÀ²üO¥ÅXÏ‘V.µÞÜG©|B‚„q¿Û”óÂ0 q4 ÇÙ—Æm?›+Ó^'ƒûÌå=´b¿É‡…µƒ²|Hn²:˜/ßP‰œø/ø}‰£‡b}ßÀ(†Þ„Q”Š0¿¸R½Ä’¯]À§aßã(ÍæÉ‘8ˆnsÕÙpF¼Äáo¨Vñ8‘–©¦™r® )6cÑ«¹ž°²Ž¢us×Dÿäð&åÝ(X2³"Ëèad”Zƒˆ7J¨žY¦}h+‡ì¶1`ä0;¼ÍWj×Hül¥ 6/ÝvÛÈ4ÏvÑŒ4:5tFý‚Ê-९0ÊÓ•ŽŒ4£Çñøá!~Œù‡˜i’¨CO×uŒçÚÉ ^C’ÒÛý©Ã|¨Àô§:“l/ÎÀ– œàW¹à3ÏÕk¦/kmvÈ6É'”ĸ™¡&12çkÀ'¡’Ô8ŠšFõ/<(˳§KuÊXV‚“Ó<ì‘;£d†a}ŽüĆAc\|Ìe© ½A¬V[+SÏó¤Ì‡MÕ ™)ަ(»à o#o§êˆÜ–¬r] v%Þ;Ûø_“‰øQ¤<6 âÜLUÍq <…‡†^FHò£3RÑð4Vq"ëÿÂÄ;: vü„"btކ+IÜð®H{gêµÈ„Puzj{Z®á‡×j{Ù, £p; ’;ꦘ(š¡uÀÆjî¾xÕOB³Ñ¤<¢²ýñ=æ!IŠj›cc ¥ ƒ^ÿ¿®½j'Ùò6f|ûq RÃÿÇ1ç(\¨g½PNvPÝiUå4xœˆäŸ -“OÂH¬ÓÑ,Å ?6÷·ØeüCáè¸/YöÝÞÞQvAù…Tëë=ƒn¡×ïõÁc2Q:ç®ìOR ’!œî]nÞ¢ÎÖ>“Ðf·‰ÑÇÍ&gc›=Îo(`¡)³¡é%õ4baê²Rb/ñi©`0>GM\>s 9ÎñH³[c S"zSšRÿÞs!q:7KëUù]kiáaÉé }˜Óïs^—4ŽÚŽ#G6þÆ‘a,D¼u±ØããþDMæ,R—¦ðÔA·-Þ:Ìňô±”ßEb§Ö"¾8]ÛdÖxÀrb¥>­¥b‘b9ºj á¿ÝÝ-WxXWí%+·;ëâП&»!:„÷l—Óz£UöDÉæñ/†f…qIyŒ‘…W”„ º”Ày즈öÈŸVÂâ4áHÜe4GRSíS*üdpÒíhj¬ŠBé,ŽTÞE^ëCÓAÒÎãª2Ì€ »ý¤"h„C)Ìa΄ûмVöòþ8p’ßñ¬ˆ`¦MATùúÆRDm6cóT0fýïäc´£Å…Pq9:×úk¢Ü-Ëû‹û÷‡û"h’.Л-7p„ä ©g6ãW·H°ÐÞ!h€5Íé.ñ>X-®G¹(@UºYB½óÁpÇý"•¼C{1¢ùÑbPÒO3Eï.™©­¹‹äAÙ6O!1"n£>YÉ: ¸å¹ÜªjA‹!™Ûœ&-þËWÄgàí¸Cˆ@î‰ӢȖÚ[(Í$á³…ŒÜlOyÉ(L¼-:\ÉIreE€™W­^BC2dÇwm…Ÿu™¥vÌ(¨Æûe]ÞRÌUš¶Öõ‹ íѸS²{ßÎ1«LiÍÿ%lzYÖÞ¾É*>󱵉㲠3µ† é‹ ÌNÇ—øYvŒ˜0+t± ǹ_Ä+ðîc[±Ô[¨z¸Ô@ø«û-q%Ùš‚9ÿw7QžS3RãG§Ö¿m_ë¹ZENÐ2‘nÕB sY‘Lf»§qâ úˆg•ÀW¢N§Vѽ ¸ØÎRFw£ÏŒ)›]§îgQ%Ÿ=ÇSè@ºÐ£3lI‚P°‘Á¶bbÁÐ{Ï«÷‚³Ti øÄ_å³ak±XÚ§Ü´þGÚU"{þ"š·ûm`ËduJ`G}#{9îHÞ›QâwÕ¼Ñ,½ß,.ïV­TÒiêBH1ó/«épM¾¬2ÀiÒ­ óY–ÆáGdsØ\Éž&JjßC¢g<1˜ÿk·Ž´(j«­úJ-ÍtËh=ø‡ó Ãù&Ú ¬xlFPî=ð™ÕÔÃÐ1/,§§ê¿»"Vkk¥MOB~º/6O¶š)¾ZG=ÝÕÙma~XÀß¶X>pmoÒä”8]Š|Ý…]é†Q°Œ¿BufQE5ìgc0Й1øË>žƒ%ü£5Kv¸oÂ!3Å“µ‰˜þŠ.ÐIÓTì±›wxH’£ÒS"¾³˜XàLKù_µêoÐmÕù°¢OæûmµÛ‡`ƒ¡¹·ÜQ¹³—¹@¸V“¼ÂßÔjÔµêËo¤]RÖ|¡CóÉïª÷ÏëîÑe= úà.l' #'Â~¤÷œG®¬f¨$6iÃà*Ĩ&°;W¨·ãú˜ž*ˆãkhö¹í$!v㾟Šív"þ;X¬"å™;ô[qïb?ÑÇÙÏ?ã›&!a–Šªw*- Ä¿\é×iâöù!£ªDy›éÖ 6w‡“²0¸b]„·1%OD£DZç%•Éí}pÚFG–§|}hS›àŸ§ÀT vÞ’¸,ˆ[Pjñ$¯„s’øZŸYgù’°Ó˜ŠÚÃïDÆ–¦Å#÷ƒ÷îz‰µñ€2%Þ\ #™@¡òçp¬”ܯ) åã”Î[Æ%“<ºXî(]hξ’çÿïæ €›ðf{F'¢7°+r^`)È í~EÖQy¸Vô ²Gž ‘6xÿ8\Ößÿƒê™B\ß"Vk˜ ìêñÖ #BŠõ4Éi¹üE2iFr;OFû­¨&Ó1D™Éå×.¡B½úyò"`ÜR CEÅàPzkØ>Œu“aeT›ñj¸sˆ½¸šó À&¬;U=@öiE)5åÖAÕ3´AÁ݃«)©¼‚¬¿ VúY&­Þ=MrQIá/»ÎŸA5´­³µ‘Õ) ykǩƥå:qh\€*!Ç– yU÷œ‘&pë-ü–Þ‘OÔ暞HÍ¥Í5{Bâ…„+Wˆêv“-0Œ,·Å@¨¥G¯z<~ºä•+È›z·‡:…é¦Ë'Br¤ÃÔ¢^àíMØ¿€xß®”Ë‘!þ·#ØÝæP×Õ–rcUàÉâ[0fÀU¾7Kš/> Ójjq Ïr$6y¯òÏ¢Wæ~f°EN('v…ÌK^•—Ò–hyWýÉ“–å‰áD{d3÷úº—ii#ÏdáûAšµâhN¥°i-.xÄÞ%y„¾Êtå[0—ì{?(.Ô) ¹˜ðŒo›þ±‹ÃE¢°S“‘…}ëÉíÀF‹E€ñ¡ÛJ%0®ê}ÎÈ)&¼¹TJ)ìSZ¶@:0ñZq·¦¿ÇÏîóõÉÌU¤ª«þÙØîø¸›xf™°°e¸yÙ&ý ÅÐJ¸;!³V«SýõSd`Ü—Lžð¸&ŸÓg‰)Iû—p%ÔË ÌEf$ ¡,cWÖô#¯hô±f÷Í™ýÊËÓÇÖ&ª„¬²â=Ÿf;¿òçi&•‰O!\jD;U‰µ`™ %J“Rqü“UÑ}}ÊǼ]E"½­)r½tÞ ò2._¤.·Šéß“-NA? 32ávÍ$<ÚõãMP6dÈÀÉ–OÓޤ㚠»`º®Î@='1ïÞXˆ¾gb)…L-1V=m!õìÓÄ+F­qÄ¥i´ÚY·Ëà±ïü'ú ÚªzDÇÝž`´ås³%Ý÷U¨J;Qk %Ó+å’´á’ŽÝ¢Ú˜ Ö›èúõ6vª ?Gb¦Øï´€¿ÌËBMiH·+êõÄ »pž¯  ߪî0eCÿ—NÝhÂôê=FÂx~Tûj›B Iî­FÛ sY5×<:ÝVª=›QÇ›™îöühå(› ht¯d ˜}éŽÙ¸¸¿¸ °63þÍr ×ô·ÝKI¯êI!î2›Íî!ˆ¹™‚7Hg¶” ¯~Û‚ÙY¬Û bRFð—N`‚±2#`š~›In-êMƒ“ ‰î €}j,S謅 SÁG–Ô¶äø¢IgÕL\õS¼Æ£àÄù³&½!d ܲÞZí–úMÖR°§ôì€ì[I5*˜Ký—”ÃGWo2öD.l6/8¸8îçõojcl=/•&ˆæ½A,Edì1ÿFËVxÏ©e,Æ™¢ã׫‹±WÙ­Žòò^ÞßVˆß°Çf è̃hüÚ²áµÔ…›v¹è—òƒä³ïJàçRô2ù ¬+Û†LÚ(=‚%8IÕø£Êœl¯_K€o6îÊå ‘@îo´#÷Fl`g|”ZEönJÆž›¤{f|¿&øCÛܘb—Lpýœ8äC%šäóˆÜ½ÈPf4ö+UÒ’ð¸!EèÞ`~§–qèo„È»¥Ljäm›Û>0 ‹YZrelsurv/R/0000755000176200001440000000000013400221670012150 5ustar liggesusersrelsurv/R/rformulate.r0000644000176200001440000002060113357572312014527 0ustar liggesusers# This is a version with suggested updates by T Therneau # All updates are stolen from survexp in the survival package, with comments. # Most changes are used, some further corrections were required. rformulate <- function (formula, data = parent.frame(), ratetable, na.action, rmap, int, centered, cause) { call <- match.call() m <- match.call(expand.dots = FALSE) # keep the parts of the call that we want, toss others m <- m[c(1, match(c("formula", "data", "cause"), names(m), nomatch=0))] m[[1L]] <- quote(stats::model.frame) # per CRAN, the formal way to set it Terms <- if (missing(data)) terms(formula, specials= c("strata","ratetable")) else terms(formula, specials=c("strata", "ratetable"), data = data) Term2 <- Terms #sorting out the ratetable argument - matching demographic variables rate <- attr(Terms, "specials")$ratetable if (length(rate) > 1) stop("Can have only 1 ratetable() call in a formula") #matching demographic variables via rmap if (!missing(rmap)) { # use this by preference if (length(rate) >0) stop("cannot have both ratetable() in the formula and a rmap argument") rcall <- rmap if (!is.call(rcall) || rcall[[1]] != as.name('list')) stop ("Invalid rcall argument") } #done with rmap else if (length(rate) >0) { #sorting out ratetable stemp <- untangle.specials(Terms, 'ratetable') rcall <- as.call(parse(text=stemp$var)[[1]]) # as a call object rcall[[1]] <- as.name('list') # make it a call to list Term2 <- Term2[-stemp$terms] # remove from the formula } else rcall <- NULL # A ratetable, but no rcall or ratetable() # Check that there are no illegal names in rcall, then expand it # to include all the names in the ratetable if (is.ratetable(ratetable)) { israte <- TRUE dimid <- names(dimnames(ratetable)) if (is.null(dimid)) dimid <- attr(ratetable, "dimid") # older style else attr(ratetable, "dimid") <- dimid #put all tables into the old style temp <- match(names(rcall)[-1], dimid) # 2,3,... are the argument names if (any(is.na(temp))) stop("Variable not found in the ratetable:", (names(rcall))[is.na(temp)]) if (any(!(dimid %in% names(rcall)))) { to.add <- dimid[!(dimid %in% names(rcall))] temp1 <- paste(text=paste(to.add, to.add, sep='='), collapse=',') if (is.null(rcall)) rcall <- parse(text=paste("list(", temp1, ")"))[[1]] else { temp2 <- deparse(rcall) rcall <- parse(text=paste("c(", temp2, ",list(", temp1, "))"))[[1]] } } } else stop("invalid ratetable") # Create a temporary formula, used only in the call to model.frame, # that has extra variables newvar <- all.vars(rcall) if (length(newvar) > 0) { tform <- paste(paste(deparse(Term2), collapse=""), paste(newvar, collapse='+'), sep='+') m$formula <- as.formula(tform, environment(Terms)) } m <- eval(m, parent.frame()) n <- nrow(m) if (n==0) stop("data set has 0 rows") Y <- model.extract(m, "response") offset <- model.offset(m) if (length(offset)==0) offset <- rep(0., n) if (!is.Surv(Y)) stop("Response must be a survival object") Y.surv <- Y if (attr(Y, "type") == "right") { type <- attr(Y, "type") status <- Y[, 2] Y <- Y[, 1] start <- rep(0, n) ncol0 <- 2 } else if (attr(Y, "type") == "counting") { type <- attr(Y, "type") status <- Y[, 3] start <- Y[, 1] Y <- Y[, 2] ncol0 <- 3 } else stop("Illegal response value") if (any(c(Y, start) < 0)) stop("Negative follow up time") if(max(Y)<30) warning("The event times must be expressed in days! (Your max time in the data is less than 30 days) \n") # rdata contains the variables matching the ratetable rdata <- data.frame(eval(rcall, m), stringsAsFactors=TRUE) rtemp <- match.ratetable(rdata, ratetable) #this function puts the dates in R and in cutpoints in rtabledate R <- rtemp$R cutpoints <- rtemp$cutpoints if(is.null(attr(ratetable, "factor"))) attr(ratetable, "factor") <- (attr(ratetable, "type") ==1) attr(ratetable, "dimid") <- dimid rtorig <- attributes(ratetable) nrt <- length(rtorig$dimid) #checking if the ratetable variables are given in days wh.age <- which(dimid=="age") wh.year <- which(dimid=="year") if(length(wh.age)>0){ if (max(R[,wh.age])<150 & median(diff(cutpoints[[wh.age]]))>12) warning("Age in the ratetable part of the formula must be expressed in days! \n (Your max age is less than 150 days) \n") } # TMT -- note the new class if(length(wh.year)>0){ if(min(R[,wh.year])>1850 & max(R[,wh.year])<2020& class(cutpoints[[wh.year]])=="rtdate") warning("The calendar year must be one of the date classes (Date, date, POSIXt)\n (Your variable seems to be expressed in years) \n") } #checking if one of the continuous variables is fixed: if(nrt!=ncol(R)){ nonex <- which(is.na(match(rtorig$dimid,attributes(ratetable)$dimid))) for(it in nonex){ if(rtorig$type[it]!=1)warning(paste("Variable ",rtorig$dimid[it]," is held fixed even though it changes in time in the population tables. \n (You may wish to set a value for each individual and not just one value for all)",sep="")) } } #NEW in 2.05 (strata) # Now create the X matrix and strata strats <- attr(Term2, "specials")$strata if (length(strats)) { temp_str <- untangle.specials(Term2,"strata",1) if (length(temp_str$vars) == 1) strata.keep <- m[[temp_str$vars]] else strata.keep <- strata(m[,temp_str$vars],shortlabel=TRUE,sep=",") Term2 <- Term2[-temp_str$terms] } else strata.keep <- factor(rep(1,n)) # zgoraj ze definirano n = nrow(m) if (!missing(cause)) strata.keep <- factor(rep(1,n)) attr(Term2, "intercept") <- 1 # ignore a "-1" in the formula X <- model.matrix(Term2, m)[,-1, drop=FALSE] mm <- ncol(X) if (mm > 0 && !missing(centered) && centered) { mvalue <- colMeans(X) X <- X - rep(mvalue, each=nrow(X)) } else mvalue <- double(mm) cause <- model.extract(m, "cause") if(is.null(cause)) cause <- rep(2,nrow(m)) #NEW: ce cause manjka #status[cause==0] <- 0 keep <- Y > start if (!missing(int)) { int <- max(int) status[Y > int * 365.241] <- 0 Y <- pmin(Y, int * 365.241) keep <- keep & (start < int * 365.241) } if (any(start > Y) | any(Y < 0)) stop("Negative follow-up times") if (!all(keep)) { X <- X[keep, , drop = FALSE] Y <- Y[keep] start <- start[keep] status <- status[keep] R <- R[keep, ,drop=FALSE] strata.keep <- strata.keep[keep] # dodano za strato #NEW in 2.05 offset <- offset[keep] Y.surv <- Y.surv[keep, , drop = FALSE] cause <- cause[keep] n <- sum(keep) rdata <- rdata[keep,] } # I do not want to preserve variable class here - so paste R onto here, give it names temp <- R names(temp) <- paste0("X", 1:ncol(temp)) # with the right names #if variable class needs to be preserved, use this instead # variable class. So paste on rdata, but with the right order and names #temp <- rdata[,match(dimid, names(rdata))] # in the right order #names(temp) <- paste0("X", 1:ncol(temp)) # with the right names data <- data.frame(start = start, Y = Y, stat = status, temp) if (mm != 0) data <- cbind(data, X) # we pass the altered cutpoints forward, keep them in the date format (could be changed eventually to get rid of the date package dependence) attr(ratetable, "cutpoints") <- lapply(cutpoints, function(x) { if (class(x) == 'rtabledate') class(x) <- 'date' x}) out <- list(data = data, R = R, status = status, start = start, Y = Y, X = as.data.frame(X), m = mm, n = n, type = type, Y.surv = Y.surv, Terms = Terms, ratetable = ratetable, offset = offset, formula=formula, cause = cause, mvalue=mvalue, strata.keep=strata.keep) # dodano za strato #NEW in 2.05 na.action <- attr(m, "na.action") if (length(na.action)) out$na.action <- na.action out } relsurv/R/cmprel.r0000644000176200001440000002270013332277470013633 0ustar liggesuserscmp.rel <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, na.action,tau,conf.int=0.95,precision=1,add.times,rmap) #formula: for example Surv(time,cens)~1 #not implemented for subgroups - DO IT! #data: the observed data set #ratetable: the population mortality tables #conf.type: confidence interval calculation (plain, log or log-log) #conf.int: confidence interval #tau: max. cas do katerega racuna { call <- match.call() if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula, data, ratetable, na.action,rmap) #get the data ready data <- rform$data #the data set se.fac <- sqrt(qchisq(conf.int, 1)) #factor needed for confidence interval if(missing(tau)) tau<-max(rform$Y) p <- rform$m #number of covariates if (p > 0) #if covariates data$Xs <- strata(rform$X[, ,drop=FALSE ]) #make strata according to covariates else data$Xs <- rep(1, nrow(data)) #if no covariates, just put 1 tab.strata <- table(data$Xs) #unique strata values ntab.strata <- length(tab.strata) #number of strata dtemp <- list(NULL) out <- as.list(rep(dtemp,ntab.strata*2)) for (kt in 1:ntab.strata) { #for each stratum inx <- which(data$Xs == names(tab.strata)[kt]) #individuals within this stratum extra <- as.numeric(seq(1,max(rform$Y[inx]),by=precision)) if(!missing(add.times)) extra <- c(extra,as.numeric(add.times)) tis <- sort(unique(pmin(tau,union(rform$Y[inx],extra))) ) #1-day long intervals used - to take into the account the continuity of the pop. part #if(!all.times)tis <- sort(unique(pmin(rform$Y[inx],tau))) #unique times #else{ # tis <- sort(union(rform$Y[inx], as.numeric(1:floor(max(rform$Y[inx]))))) #1-day long intervals used - to take into the account the continuity of the pop. part # tis <- unique(pmin(tis,tau)) #} k <- length(tis) out[[2*kt-1]]$time <- out[[2*kt]]$time <- c(0,tis) temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=tis,fast=TRUE,cmp=T) #calculate the values for each interval of time areae <- sum(temp$areae)/365.241 # sum(diff(c(0,tis))*temp$cumince)/365.241 areap <- sum(temp$areap)/365.241 #sum(diff(c(0,tis))*temp$cumincp)/365.241 options(warn=-1) out[[2*kt-1]]$est <- c(0,temp$cumince) out[[2*kt-1]]$var <- c(0,temp$ve) out[[2*kt-1]]$lower <- temp$cumince-se.fac*sqrt(temp$ve) out[[2*kt-1]]$upper <- temp$cumince+se.fac*sqrt(temp$ve) out[[2*kt-1]]$area <- areae out[[2*kt]]$est <- c(0,temp$cumincp) out[[2*kt]]$var <- c(0,temp$vp) out[[2*kt]]$lower <- temp$cumincp-se.fac*sqrt(temp$vp) out[[2*kt]]$upper <- temp$cumincp+se.fac*sqrt(temp$vp) out[[2*kt]]$area <- areap options(warn=0) ne <- sum(temp$ve<0) if(ne>0) warning(paste(names(tab.strata)[kt],": The estimated variance of crude mortality is negative in ", ne, " out of ", length(temp$ve)," intervals"), call. = FALSE) if(!missing(add.times)){ out[[2*kt-1]]$index <- out[[2*kt]]$index <- unique(c(1,which(tis %in% c(rform$Y[inx],add.times,tau)))) out[[2*kt-1]]$add.times <- out[[2*kt]]$add.times <- add.times } else out[[2*kt-1]]$index <- out[[2*kt]]$index <- unique(c(1,which(tis %in% c(rform$Y[inx],tau)))) } if(p>0)names(out) <- paste(rep(c("causeSpec","population"),ntab.strata),rep(names(tab.strata),each=2)) else names(out) <- c("causeSpec","population") out$tau <- tau class(out) <- "cmp.rel" out } plot.cmp.rel <- function (x, main = " ", curvlab, ylim = c(0, 1), xlim, wh = 2, xlab = "Time (days)", ylab = "Probability", lty = 1:length(x), xscale=1, col = 1, lwd = par("lwd"), curves, conf.int, all.times=FALSE,...) { #wh= upper left coordinates of the legend, if of length 1, the legend is placed top left. tau <- x$tau x$tau <- NULL nc <- length(x) #number of curves if (length(lty) < nc) #if not enough different line types lty <- rep(lty[1], nc) else lty <- lty[1:nc] if (length(lwd) < nc) #if not enough different line widths lwd <- rep(lwd[1], nc) else lwd <- lwd[1:nc] if (length(col) < nc) #if not enough different colors col <- rep(col[1], nc) else col <- col[1:nc] if (missing(curvlab)) { #if no curve labels desired if (mode(names(x)) == "NULL") { #and no curve labels prespecified curvlab <- as.character(1:nc) } else curvlab <- names(x)[1:nc] #use prespecified if they exist } if (missing(xlim)) { #if no limits desired xmax <- 0 for (i in 1:nc) { xmax <- max(c(xmax, x[[i]][[1]]/xscale)) #take max time over all strata } xlim <- c(0, xmax) } if(all.times){ for(it in 1:nc){ x[[it]]$index <- 1:length(x[[it]][[1]]) } } if(missing(curves))curves <- 1:nc if(missing(conf.int))conf.int <- NULL curves <- unique(curves) conf.int <- unique(conf.int) if(any((curves %in% 1:nc)==FALSE)) stop(paste("The curves argument should be specified as a vector of integers from 1 to", nc,sep=" ")) if(any((conf.int %in% 1:nc)==FALSE)) stop(paste("The conf.int argument should be specified as a vector of integers from 1 to", nc,sep=" ")) if(any((conf.int %in% curves)==FALSE)) stop("Confidence interval may only be plotted if the curve is plotted, see argument curves") col_nums <- floor(seq(from=95,to=50,length.out=length(conf.int)+2)) col.conf.temp <- sapply(col_nums,function(x)paste("gray",as.character(x),sep="")) col.conf.int <- rep("white",nc) col.conf.int[conf.int] <- col.conf.temp[-c(1,length(col.conf.temp))] plot((x[[1]][[1]]/xscale)[x[[1]]$index], (x[[1]][[2]])[x[[1]]$index], type = "n", ylim = ylim, xlim = xlim, main = main, xlab = xlab, ylab = ylab, bty = "l", ...) #plot estimates [[1]]=time, [[2]]=est if (length(wh) != 2) { wh <- c(xlim[1], ylim[2]) } u <- list(...) if (length(u) > 0) { i <- pmatch(names(u), names(formals(legend)), 0) do.call("legend", c(list(x = wh[1], y = wh[2], legend = curvlab[curves], col = col[curves], lty = lty[curves], lwd = lwd[curves], bty = "n", bg = -999999), u[i > 0])) } else { do.call("legend", list(x = wh[1], y = wh[2], legend = curvlab[curves], col = col[curves], lty = lty[curves], lwd = lwd[curves], bty = "n", bg = -999999)) } for(i in conf.int){ if(i%%2==0)with(x[[i]],polygon(c(time[index][!is.na(lower[index])],rev(time[index][!is.na(upper[index])]))/xscale,c(lower[index][!is.na(lower[index])],rev(upper[index][!is.na(upper[index])])),col = col.conf.int[i] , border = FALSE)) else with(x[[i]],my.poly(time[index][!is.na(lower[index])]/xscale,time[index][!is.na(upper[index])]/xscale,lower[index][!is.na(lower[index])],upper[index][!is.na(upper[index])],col = col.conf.int[i] , border = FALSE)) } for (i in curves) { tip <- "s" if(i%%2==0)tip <- "l" lines((x[[i]][[1]]/xscale)[x[[i]]$index], (x[[i]][[2]])[x[[i]]$index], lty = lty[i], col = col[i], lwd = lwd[i], type=tip, ...) } } my.poly <- function(x1,x2,y1,y2,...){ x1 <- rep(x1,each=2)[-1] y1 <- rep(y1,each=2)[-(2*length(y1))] x2 <- rep(x2,each=2)[-1] y2 <- rep(y2,each=2)[-(2*length(y2))] polygon(c(x1,rev(x2)),c(y1,rev(y2)),...) } print.cmp.rel <- function (x, ntp = 4, maxtime,scale=365.241, ...) { tau <- x$tau x$tau <- NULL nc <- length(x) if (missing(maxtime)) { maxtime <- 0 for (i in 1:nc) maxtime <- max(maxtime, x[[i]]$time) } tp <- pretty(c(0, maxtime/scale), ntp + 1) tp <- tp[-c(1, length(tp))] if(length(x[[1]]$add.times)>0 & length(x[[1]]$add.times)<5){ tp <- sort(unique(c(tp,round(x[[1]]$add.times/scale,1)))) } cat("Estimates, variances and area under the curves:\n") x$tau <- tau print(summary(x, tp,scale,area=TRUE), ...) invisible() } summary.cmp.rel <- function (object, times,scale=365.241,area=FALSE,...) { tau <- object$tau object$tau <- NULL ng <- length(object) times <- sort(unique(times))*scale nt <- length(times) storage.mode(times) <- "double" storage.mode(nt) <- "integer" ind <- matrix(0, ncol = nt, nrow = ng) oute <- matrix(NA, ncol = nt, nrow = ng) outv <- oute outa <- matrix(NA,ncol=1,nrow=ng) storage.mode(ind) <- "integer" slct <- rep(TRUE, ng) for (i in 1:ng) { if (is.null((object[[i]])$est)) { slct[i] <- FALSE } else { z <- rep(NA,nt) for(kt in 1:nt)z[kt] <- rev(which(object[[i]][[1]]<=times[kt]))[1] ind[i, ] <- z oute[i, ind[i, ] > 0] <- object[[i]][[2]][z] outa[i,] <- object[[i]][[6]] if (length(object[[i]]) > 2) outv[i, ind[i, ] > 0] <- object[[i]][[3]][z] } } dimnames(oute) <- list(names(object)[1:ng], as.character(times/scale)) dimnames(outv) <- dimnames(oute) rownames(outa) <- rownames(oute) colnames(outa) <- paste("Area at tau =",tau/scale) if(area)list(est = oute[slct, , drop = FALSE], var = outv[slct, , drop = FALSE], area=outa[slct,,drop=FALSE]) else list(est = oute[slct, , drop = FALSE], var = outv[slct, , drop = FALSE]) }relsurv/R/rsdiff.r0000644000176200001440000001355313332277513013632 0ustar liggesusersrs.diff <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, na.action,precision=1,rmap) #formula: for example Surv(time,cens)~sex #data: the observed data set #ratetable: the population mortality tables { call <- match.call() if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula, data, ratetable, na.action,rmap) #get the data ready data <- rform$data #the data set p <- rform$m #number of covariates if (p > 0) #if covariates data$Xs <- strata(rform$X[, ,drop=FALSE ]) #make groups according to covariates else data$Xs <- rep(1, nrow(data)) #if no covariates, just put 1 # Xs is a vector of factors determining the groups we wish to compare strats <- rform$strata.keep # added for strata str_num <- length(levels(strats)) # number of strata out <- NULL out$n <- table(data$Xs) #table of groups out$time <- out$n.risk <- out$n.event <- out$n.censor <- out$surv <- out$std.err <- out$groups <- NULL #TIMES ARE EQUAL FOR ALL GROUPS if(!precision)tis <- sort(unique(rform$Y)) #unique times else{ extra <- as.numeric(seq(1,max(rform$Y),by=precision)) tis <- sort(union(extra,rform$Y)) #1-day long intervals used - to take into the account the continuity of the pop. part } # start working kgroups <- length(out$n) #number of groups if (kgroups == 1) stop("There is only one group in your data. You should choose another variable.") w.risk <- w.event <- dnisisq <- array(NA,dim=c(length(tis),length(out$n),str_num)) #MATRIX - COLUMNS ARE GROUPS, ROWS ARE TIMES,levels are strata #numOfSmallGrps <- 0 numOfFewEvents <- 0 for (s in 1:str_num){ # added for strata for (kt in 1:kgroups) { #for each group inx <- which(data$Xs == names(out$n)[kt] & strats == levels(strats)[s]) #individuals within this group #if (length(inx)<10)numOfSmallGrps <- numOfSmallGrps + 1 temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=tis,fast=TRUE) #calculate the values for each interval of time out$time <- c(out$time, tis) #add times out$n.risk <- c(out$n.risk, temp$yi) #add number at risk for each time out$n.event <- c(out$n.event, temp$dni) #add number of events for each time if (sum(temp$dni) < 10) numOfFewEvents <- numOfFewEvents + 1 out$n.censor <- c(out$n.censor, c(-diff(temp$yi),temp$yi[length(temp$yi)]) - temp$dni) #add number of censored for each time w.risk[,kt,s] <- temp$yisi #Y_h^w w.event[,kt,s] <- temp$dnisi - temp$yidlisi #dN_eh^w dnisisq[,kt,s] <- temp$dnisisq #dN/S_p^2 out$groups <- c(out$groups, length(tis)) #number of times in this group } } #if (numOfSmallGrps > 0) warning(numOfSmallGrps, " out of ", kgroups*str_num, " groups is/are smaller than 10.") if (numOfFewEvents > 0) warning("In ", numOfFewEvents, " out of ", kgroups*str_num, " groups there are less than 10 events.") w.risk.total <- apply(w.risk,c(1,3),sum) #sum over all individuals at each time point ## Y_{.,s}^w w.event.total <- apply(w.event,c(1,3),sum) #sum over all individuals at each time point ## dN_{E,.,s}^w zs <- rep(0,kgroups) # added for strata for (s in 1:str_num){ # znotraj danega stratuma inx_str <- which(w.risk.total[,s] > 0) zhst <- w.event[inx_str,,s,drop=FALSE] - w.risk[inx_str,,s,drop=FALSE]/w.risk.total[inx_str,s]*w.event.total[inx_str,s] #value under the integral of zh # integriramo po casu - sestejemo po casih dogodkov zhs <- apply(zhst,2,sum) # the vector of test statistics zs <- zs + zhs } # cat("vektor testnih statistik je = \n") # print(zs) #covariance matrix: covmats <- matrix(0,nrow=kgroups,ncol=kgroups) d <- diag(kgroups) #identity matrix of groups size (for the kronecker deltas) for (s in 1:str_num){ underint <- 0 inx_str <- which(w.risk.total[,s] > 0) for(kt in 1:kgroups){ #matrix calculation through the groups ys <- matrix(d[kt,],nrow=length(inx_str),ncol=kgroups,byrow=T) - w.risk[inx_str,,s]/w.risk.total[inx_str,s] #preparing the matrix for the first two terms #yslist <- apply(apply(ys,1,list),unlist) #a list, each row of ys (each time point) represents one item yslist <- as.list(data.frame(t(ys))) #a list, each row of ys (each time point) represents one item yprod <- lapply(yslist,function(x)outer(x,x)) #a list of matrices with y products through all the time points, yproda <- array(unlist(yprod),dim=c(kgroups,kgroups,length(inx_str)))#y terms transformed to an array dnisisqa <- array(rep(dnisisq[,kt,s],each=kgroups^2),dim=c(kgroups,kgroups,length(inx_str))) #dnisisq terms transformed into an array of equal size underint <- underint + yproda * dnisisqa #the terms under the integral } covmat <- apply(underint,1:2,sum) #summing down the array covmats <- covmats + covmat } # cat("kovariancna matrika je = \n") # print(covmats) # del za testiranje zs <- zs[-kgroups] # the last one is deleted zs <- matrix(zs,nrow=1) # print(covmats) covmats <- covmats[-kgroups,-kgroups,drop=F] # print(covmats) test.stat <- zs %*% solve(covmats) %*% t(zs) p.value <- 1-pchisq(test.stat,df=kgroups-1) names(out$groups) <- names(out$n) if (p == 0) out$groups <- NULL #if no covariates out$n <- as.vector(out$n) out$call <- call #class(out) <- c("survdiff", "rs.surv") #cat(zh) out$zh <- zs out$covmat <- covmats out$test.stat <- test.stat out$p.value <- p.value out$df <- kgroups-1 class(out) <- "rsdiff" out } print.rsdiff <- function(x,...){ invisible(cat("Value of test statistic:", x$test.stat, "\n")) invisible(cat("Degrees of freedom:", x$df, "\n")) invisible(cat("P value:", x$p.value, "\n")) } relsurv/R/survfitrsadd.r0000644000176200001440000000455713332561302015071 0ustar liggesuserssurvfit.rsadd <- function (formula, newdata, se.fit = TRUE, conf.int = 0.95, individual = FALSE, conf.type = c("log", "log-log", "plain", "none"),...) { call <- match.call() Terms <- terms(formula) #to rabis, ce je model mal bl smotan - as.factor ali splines ali svasta Terms <- delete.response(Terms) popdata <- newdata newdata <- model.frame(Terms,newdata) resp <- list(y=formula$y,x=newdata) n <- formula$n nvar <- length(formula$coef) nx <- nrow(newdata) nt <- length(formula$times) temp <- list(n=formula$n,time=formula$times,call=call,type="right") Lambda0 <- formula$Lambda0 Lambda0 <- matrix(Lambda0,ncol=nt,nrow=nrow(newdata),byrow=TRUE) rate <- attr(Terms, "specials")$ratetable #rat <- attributes(formula$ratetable)$dimid rat <- names(attributes(formula$ratetable)$dimnames) #mein <- attributes(newdata[,rate])$dimnames[[2]] mein <- names(popdata) x <- match(rat,mein) #R <- as.matrix(newdata[, rate, drop = FALSE]) R <- as.matrix(popdata) R <- R[,x,drop=FALSE] R <- data.frame(R) names(R) <- rat #newdata <- newdata[,1:(rate-1),drop=FALSE] labeli <- attr(attr(newdata,"terms"),"term.labels") colnami <- colnames(newdata) if(length(rate>0)){ labeli <- labeli[-rate] colnami <- colnami[-rate] } newdata <- newdata[,match(colnami,labeli),drop=F] if(any(formula$mvalue)>0)newdata <- newdata - matrix(formula$mvalue,nrow=nrow(newdata),byrow=TRUE) nx <- ncol(newdata) #getl <- function(times,data=R,ratetable=formula$ratetable){ # -log(srvxp.fit(data,times,ratetable)) #} #Lambdap <- sapply(formula$times, getl) # Lambdap <- NULL # for(it in 1:nt){ # Lambdap <- cbind(Lambdap,-log(srvxp.fit(R,formula$times[it],formula$ratetable))) # } Lambdap <- NULL for(it in 1:nrow(newdata)){ Lambdap <- rbind(Lambdap,-log(survexp(~1,data=R[it,,drop=FALSE],times=formula$times,ratetable=formula$ratetable)$surv)) } ebx <- exp(as.matrix(formula$coef %*%as.numeric(newdata))) ebx <- matrix(ebx,ncol=nt,nrow=length(ebx)) Lambda <- Lambdap + Lambda0*ebx temp$surv <- t(exp(-Lambda)) temp$n.event <- rep(1,nt) temp$n.risk <- n+1 - cumsum(temp$n.event) class(temp) <- c("rs.surv.rsadd", "rs.surv","survfit") temp }relsurv/R/rssurvrsadd.r0000644000176200001440000000271411551273116014730 0ustar liggesusersrs.surv.rsadd <- function (formula, newdata) { call <- match.call() Terms <- terms(formula$formula) #to rabis, ce je model mal bl smotan - as.factor ali splines ali svasta Terms <- delete.response(Terms) newdata <- model.frame(Terms,newdata) n <- formula$n if(formula$method=="max.lik"){ nvar <- length(formula$coef) - length(formula$int)+1 formula$coef <- formula$coef[1:nvar] } nvar <- length(formula$coef) nx <- nrow(newdata) nt <- length(formula$times) temp <- list(n=formula$n,time=formula$times,call=call,type="right") Lambda0 <- formula$Lambda0 Lambda0 <- matrix(Lambda0,ncol=nt,nrow=nx,byrow=TRUE) rate <- attr(Terms, "specials")$ratetable R <- as.matrix(newdata[, rate,drop=FALSE]) rat <- attributes(formula$ratetable)$dimid mein <- attributes(newdata[,rate])$dimnames[[2]] x <- match(rat,mein) R <- R[,x,drop=FALSE] newdata <- newdata[,1:nvar,drop=FALSE] if(any(formula$mvalue)>0)newdata <- newdata - matrix(formula$mvalue,nrow=nx,byrow=TRUE) R <- data.frame(R) names(R) <- rat ebx <- exp(data.matrix(newdata)%*%as.vector(formula$coef)) ebx <- matrix(ebx,ncol=nt,nrow=length(ebx)) Lambdae <- Lambda0*ebx temp$surv <- t(exp(-Lambdae)) temp$n.event <- rep(1,nt) temp$n.risk <- n+1 - cumsum(temp$n.event) temp$time <- formula$times class(temp) <- c("rs.surv.rsadd", "rs.surv","survfit") temp }relsurv/R/plotrssurv.r0000644000176200001440000002345412700667377014632 0ustar liggesusersplot.rs.surv <- function (x, conf.int, mark.time = TRUE, mark = 3, col = 1, lty = 1, lwd = 1, cex = 1, log = FALSE, xscale = 1, yscale = 1, firstx = 0, firsty = 1, xmax, ymin = 0, fun, xlab = "", ylab = "", xaxs = "S", ...) { dotnames <- names(list(...)) if (any(dotnames == "type")) stop("The graphical argument 'type' is not allowed") if (is.logical(log)) { logy <- log logx <- FALSE if (logy) logax <- "y" else logax <- "" } else { logy <- (log == "y" || log == "xy") logx <- (log == "x" || log == "xy") logax <- log } if (missing(firstx)) { if (!is.null(x$start.time)) firstx <- x$start.time else { if (logx || (!missing(fun) && is.character(fun) && fun == "cloglog")) firstx <- min(x$time[x$time > 0]) else firstx <- min(0, x$time) } } firstx <- firstx/xscale if (missing(xaxs) && firstx != 0) xaxs <- par("xaxs") if (!inherits(x, "survfit")) stop("First arg must be the result of survfit") if (missing(conf.int)) { if (is.null(x$strata) && !is.matrix(x$surv)) conf.int <- TRUE else conf.int <- FALSE } #if (all.times == FALSE & x$method == 1){ #if (is.null(x$strata0)){ # nstrat <- 1 # stemp <- rep(1, length(x$index)) # length(x$time[x$index]) == length(x$index) # } # else { # nstrat <- length(x$strata0) # stemp <- rep(1:nstrat,x$strata0) # } #} #else { if (is.null(x$strata)) { nstrat <- 1 stemp <- rep(1, length(x$time)) } else { nstrat <- length(x$strata) stemp <- rep(1:nstrat, x$strata) } #} ssurv <- x$surv stime <- x$time supper <- x$upper slower <- x$lower #if (all.times == FALSE & x$method == 1){ # ssurv <- ssurv[x$index]; stime <- stime[x$index]; supper <- supper[x$index]; slower <- slower[x$index] #} if (!missing(xmax) && any(x$time > xmax)) { keepx <- keepy <- NULL yzero <- NULL tempn <- table(stemp) offset <- cumsum(c(0, tempn)) for (i in 1:nstrat) { ttime <- stime[stemp == i] if (all(ttime <= xmax)) { keepx <- c(keepx, 1:tempn[i] + offset[i]) keepy <- c(keepy, 1:tempn[i] + offset[i]) } else { bad <- min((1:tempn[i])[ttime > xmax]) if (bad == 1) { keepy <- c(keepy, 1 + offset[i]) yzero <- c(yzero, 1 + offset[i]) } else keepy <- c(keepy, c(1:(bad - 1), bad - 1) + offset[i]) keepx <- c(keepx, (1:bad) + offset[i]) stime[bad + offset[i]] <- xmax x$n.event[bad + offset[i]] <- 1 } } stime <- stime[keepx] stemp <- stemp[keepx] x$n.event <- x$n.event[keepx] if (is.matrix(ssurv)) { if (length(yzero)) ssurv[yzero, ] <- firsty ssurv <- ssurv[keepy, , drop = FALSE] if (!is.null(supper)) { if (length(yzero)) supper[yzero, ] <- slower[yzero, ] <- firsty supper <- supper[keepy, , drop = FALSE] slower <- slower[keepy, , drop = FALSE] } } else { if (length(yzero)) ssurv[yzero] <- firsty ssurv <- ssurv[keepy] if (!is.null(supper)) { if (length(yzero)) supper[yzero] <- slower[yzero] <- firsty supper <- supper[keepy] slower <- slower[keepy] } } } stime <- stime/xscale if (!missing(fun)) { if (is.character(fun)) { tfun <- switch(fun, log = function(x) x, event = function(x) 1 - x, cumhaz = function(x) -log(x), cloglog = function(x) log(-log(x)), pct = function(x) x * 100, logpct = function(x) 100 * x, stop("Unrecognized function argument")) if (fun == "log" || fun == "logpct") logy <- TRUE if (fun == "cloglog") { logx <- TRUE if (logy) logax <- "xy" else logax <- "x" } } else if (is.function(fun)) tfun <- fun else stop("Invalid 'fun' argument") ssurv <- tfun(ssurv) if (!is.null(supper)) { supper <- tfun(supper) slower <- tfun(slower) } firsty <- tfun(firsty) ymin <- tfun(ymin) } if (is.null(x$n.event)) mark.time <- FALSE if (is.matrix(ssurv)) ncurve <- nstrat * ncol(ssurv) else ncurve <- nstrat mark <- rep(mark, length.out = ncurve) col <- rep(col, length.out = ncurve) lty <- rep(lty, length.out = ncurve) lwd <- rep(lwd, length.out = ncurve) if (is.numeric(mark.time)) mark.time <- sort(mark.time) if (xaxs == "S") { xaxs <- "i" tempx <- max(stime) * 1.04 } else tempx <- max(stime) tempx <- c(firstx, tempx, firstx) if (logy) { tempy <- range(ssurv[is.finite(ssurv) & ssurv > 0]) if (tempy[2] == 1) tempy[2] <- 0.99 if (any(ssurv == 0)) { tempy[1] <- tempy[1] * 0.8 ssurv[ssurv == 0] <- tempy[1] if (!is.null(supper)) { supper[supper == 0] <- tempy[1] slower[slower == 0] <- tempy[1] } } tempy <- c(tempy, firsty) } else tempy <- c(range(ssurv[is.finite(ssurv)]), firsty) if (missing(fun)) { tempx <- c(tempx, firstx) tempy <- c(tempy, ymin) } plot(tempx, tempy * yscale, type = "n", log = logax, xlab = xlab, ylab = ylab, xaxs = xaxs, ...) if (yscale != 1) { if (logy) par(usr = par("usr") - c(0, 0, log10(yscale), log10(yscale))) else par(usr = par("usr")/c(1, 1, yscale, yscale)) } dostep <- function(x, y) { if (is.na(x[1] + y[1])) { x <- x[-1] y <- y[-1] } n <- length(x) if (n > 2) { dupy <- c(!duplicated(y)[-n], TRUE) n2 <- sum(dupy) xrep <- rep(x[dupy], c(1, rep(2, n2 - 1))) yrep <- rep(y[dupy], c(rep(2, n2 - 1), 1)) list(x = xrep, y = yrep) } else if (n == 1) list(x = x, y = y) else list(x = x[c(1, 2, 2)], y = y[c(1, 1, 2)]) } i <- 0 xend <- NULL yend <- NULL for (j in unique(stemp)) { who <- (stemp == j) xx <- c(firstx, stime[who]) nn <- length(xx) if (x$type == "counting") { #if (all.times == FALSE & x$method == 1){deaths <- c(-1,x$n.censor[x$index][who])} #else { deaths <- c(-1, x$n.censor[who]) #} zero.one <- 1 } else if (x$type == "right") { #if (all.times == FALSE & x$method == 1){deaths <- c(-1,x$n.censor[x$index][who])} #else { deaths <- c(-1, x$n.censor[who]) #} zero.one <- 1 } if (is.matrix(ssurv)) { for (k in 1:ncol(ssurv)) { i <- i + 1 yy <- c(firsty, ssurv[who, k]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) if (is.numeric(mark.time)) { indx <- mark.time for (k in seq(along.with = mark.time)) indx[k] <- sum(mark.time[k] > xx) points(mark.time[indx < nn], yy[indx[indx < nn]], pch = mark[i], col = col[i], cex = cex) } else if (mark.time && any(deaths >= zero.one)) { points(xx[deaths >= zero.one], yy[deaths >= zero.one], pch = mark[i], col = col[i], cex = cex) } xend <- c(xend, max(xx)) yend <- c(yend, min(yy)) if (conf.int && !is.null(supper)) { if (ncurve == 1) lty[i] <- lty[i] + 1 yy <- c(firsty, supper[who, k]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) yy <- c(firsty, slower[who, k]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) } } } else { i <- i + 1 yy <- c(firsty, ssurv[who]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) if (is.numeric(mark.time)) { indx <- mark.time for (k in seq(along = mark.time)) indx[k] <- sum(mark.time[k] > xx) points(mark.time[indx < nn], yy[indx[indx < nn]], pch = mark[i], col = col[i], cex = cex) } else if (mark.time == TRUE && any(deaths >= zero.one)) { points(xx[deaths >= zero.one], yy[deaths >= zero.one], pch = mark[i], col = col[i], cex = cex) } xend <- c(xend, max(xx)) yend <- c(yend, min(yy)) if (conf.int == TRUE && !is.null(supper)) { if (ncurve == 1) lty[i] <- lty[i] + 1 yy <- c(firsty, supper[who]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) yy <- c(firsty, slower[who]) lines(dostep(xx, yy), lty = lty[i], col = col[i], lwd = lwd[i]) } } } invisible(list(x = xend, y = yend)) } relsurv/R/mystrata.r0000644000176200001440000000607412531603441014212 0ustar liggesusersmy.strata <- function (..., nameslist, sep = ", ") { #nameslist = lista imen spremenljivk words <- as.character((match.call())[-1]) #ime podatkov allf <- list(...) #podatki if (length(allf) == 1 && is.list(ttt <- unclass(allf[[1]]))) #so samo eni podatki allf <- ttt #ohranim le podatke (ne listo podatkov), v obliki list nterms <- length(allf) #nterms= st. spremenljivk +1 (row.names) if (is.null(names(allf))) #ce ni imen argname <- words[1:nterms] #jih dam else argname <- ifelse(names(allf) == "", words[1:nterms], #ce so prazna jih dam names(allf)) #imena so v argname varnames <- names(nameslist) #1. iteracija what <- allf[[1]] #prva spremenljivka for(it in 1:length(varnames)){ if (length(grep(varnames[it],names(allf)[[1]]))) break #poiscem ji mesto v svojem poimenovanju } if (is.null(levels(what))) what <- factor(what) #ce se ni, jo prisilimo v faktorsko levs <- unclass(what) - 1 #nastavim prvi level = 0 wlab <- levels(what) #imena faktorjev labs <- paste(argname[1], wlab, sep = "=") #prvo ime = 0/1 labsnow <- 1 allab <- NULL dd <- length(nameslist[[it]]) if(dd!=2) { mylabs <- rep(argname[1],length(wlab)) mylabs[wlab==0] <- "" } else mylabs <- labs for (i in (1:nterms)[-1]) { if(length(grep(varnames[labsnow],names(allf)[[i]]))==0){ #ce je zdaj to nova spremenljivka, moram najprej ustimat prejsnjo mylabs[mylabs==""] <- nameslist[[labsnow]][1] if(!any(allab!=""))allab <- paste(allab,mylabs,sep="") #the first time - do not separate by comma else allab <- paste(allab,mylabs,sep=",") mylabs <- rep("",length(mylabs)) labsnow <- labsnow+1 } what <- allf[[i]] if (is.null(levels(what))) what <- factor(what) wlev <- unclass(what) - 1 wlab <- levels(what) labsnew <- format(paste(argname[i], wlab, sep = "=")) levs <- wlev + levs * (length(wlab)) a <- rep(labs, rep(length(wlab), length(labs))) b <- rep(wlab, length(labs)) mya <- rep(mylabs, rep(length(wlab), length(labs))) allab <- rep(allab,rep(length(wlab), length(labs))) myb <- rep(argname[i],length(labs)*length(wlab)) for(it in 1:length(varnames)){ #it se ustavi pri trenutni spremenljivki if (length(grep(varnames[it],names(allf)[[i]]))) break } dd <- length(nameslist[[it]]) if(dd==2)myb <- paste(myb,rep(wlab,length(labs)),sep="=") else myb[rep(wlab,length(labs))==0] <- "" mylabs <- paste(mya,myb,sep="") labs <- paste(a,b, sep = sep) } mylabs[mylabs==""] <- nameslist[[labsnow]][1] if(!any(allab!=""))allab <- paste(allab,mylabs,sep="") else allab <- paste(allab,mylabs,sep=",") levs <- levs + 1 ulevs <- sort(unique(levs[!is.na(levs)])) levs <- match(levs, ulevs) labs <- labs[ulevs] allab <- allab[ulevs] factor(levs, labels = allab) } relsurv/R/zzz.R0000644000176200001440000000036011716722606013144 0ustar liggesusers#.First.lib <- function(lib, pkg) library.dynam("runproba", pkg, lib) # use .onLoad instead of .First.lib for use with NAMESPACE and R(>= 1.7.0) .onLoad <- function(lib, pkg) { library.dynam("relsurv", pkg, lib) }#end of .onLoad relsurv/R/Rcode.r0000644000176200001440000030621513361314117013403 0ustar liggesusersrsfitterem<-function(data,b,maxiter,ratetable,tol,bwin,p,cause,Nie){ pr.time<-proc.time()[3] if (maxiter<1) stop("There must be at least one iteration run") n<-nrow(data) m <- p dtimes <- which(data$stat==1) #the positions of event times in data$Y td <- data$Y[dtimes] #event times ntd <- length(td) #number of event times utimes <- which(c(1,diff(td))!=0) #the positions of unique event times among td utd <- td[utimes] #unique event times nutd <- length(utd) #number of unique event times udtimes <- dtimes[utimes] #the positions of unique event times among data$Y razteg <- function(x){ # x is a 0/1 vector, the output is a vector of length sum(x), with the corresponding rep numbers n <- length(x) repu <- rep(1,n) repu[x==1] <- 0 repu <- rev(cumsum(rev(repu))) repu <- repu[x==1] repu <- -diff(c(repu,0))+1 if(sum(repu)!=n)repu <- c(n-sum(repu),repu) #ce je prvi cas censoring, bo treba se kej narest?? repu } rutd <- rep(0,ntd) rutd[utimes] <- 1 rutd <- razteg(rutd) #from unique event times to event times rtd <- razteg(data$stat) #from event times to data$Y a <- data$a[data$stat==1] if(bwin[1]!=0){ #the vector of change points for the smoothing bandwidth nt4 <- c(1,ceiling(c(nutd*.25,nutd/2,nutd*.75,nutd))) if(missing(bwin))bwin <- rep(1,4) else bwin <- rep(bwin,4) for(it in 1:4){ bwin[it] <- bwin[it]*max(diff(utd[nt4[it]:nt4[it+1]])) } while(utd[nt4[2]]0){ whtemp <- data$stat==1&cause==2 dataded <- data[data$stat==1&cause==2,] #events with unknown cause datacens <- data[data$stat==0|cause<2,] #censorings or known cause datacens$cause <- cause[data$stat==0|cause<2]*data$stat[data$stat==0|cause<2] databig <- lapply(dataded, rep, 2) databig <- do.call("data.frame", databig) databig$cause <- rep(2,nrow(databig)) nded <- nrow(databig) databig$cens <- c(rep(1,nded/2),rep(0,nded/2)) datacens$cens <- rep(0,nrow(datacens)) datacens$cens[datacens$cause<2] <- datacens$cause[datacens$cause<2] names(datacens) <- names(databig) databig <- rbind(databig,datacens) cause <- cause[data$stat==1] #NEW IN 2.05 (next 4 lines) fk <- (attributes(ratetable)$factor != 1) nfk <- length(fk) varstart <- 3+nfk+1 #first column of covariates varstop <- 3+nfk+m #last column of covariates #model matrix for relative survival xmat <- as.matrix(data[,varstart:varstop]) #NEW IN 2.05 #ebx at initial values of b ebx <- as.vector(exp(xmat%*%b)) #model matrix for coxph modmat <- as.matrix(databig[,varstart:varstop]) #NEW IN 2.05 varnames <- names(data)[varstart:varstop] #NEW IN 2.05 } else{ cause <- cause[data$stat==1] ebx <- rep(1,n) } #for time-dependent data: starter <- sort(data$start) starter1<-c(starter[1],starter[-length(starter)]) #the values of interest in the cumsums of the obsolete values (there is at least one value - the 1st) index <- c(TRUE,(starter!=starter1)[-1]) starter <- starter[index] #the number of repetitions in each cumsum difference - needed for s0 calculation val1 <- apply(matrix(starter,ncol=1),1,function(x,Y)sum(x>=Y),data$Y) val1 <- c(val1[1],diff(val1),length(data$Y)-val1[length(val1)]) eb <- ebx[data$stat==1] s0 <- cumsum((ebx)[n:1])[n:1] ebx.st <- ebx[order(data$start)] s0.st <- ((cumsum(ebx.st[n:1]))[n:1])[index] s0.st <- rep(c(s0.st,0),val1) s0 <- s0 - s0.st #s0 only at times utd s0 <- s0[udtimes] #find the corresponding value of Y for each start!=0 - needed for likelihood calculation start <- data$start if(any(start!=0)){ wstart <- rep(NA,n) ustart <- unique(start[start!=0]) for(its in ustart){ wstart[start==its] <- min(which(data$Y==its)) } } #tale del je zelo sumljiv - kako se racuna likelihood za ties??? difft <- c(data$Y[data$stat==1][1],diff(td)) difft <- difftu <- difft[difft!=0] difft <- rep(difft,rutd) a0 <- a*difft if(sum(Nie==.5)!=0)maxit0 <- maxiter else maxit0<- maxiter - 3 for(i in 1:maxit0){ #Nie is of length ntd, should be nutd, with the values at times being the sum nietemp <- rep(1:nutd,rutd) Nies <- as.vector(by(Nie,nietemp,sum)) #shorter Nie - only at times utd lam0u <- lam0 <- Nies/s0 #the smooting of lam0 if(bwin[1]!=0)lam0s <- krn%*%lam0 else lam0s <- lam0/difftu #extended to all event times lam0s <- rep(lam0s,rutd) #compute Nie, only for those with unknown hazard Nie[cause==2] <- as.vector(lam0s*eb/(a+lam0s*eb))[cause==2] } if(maxit0!=maxiter & i==maxit0) i <- maxiter #likelihood calculation - manjka ti se likelihood za nicelni model!!! #the cumulative hazard Lam0 <- cumsum(lam0) #extended to all event times Lam0 <- rep(Lam0,rutd) if(data$stat[1]==0) Lam0 <- c(0,Lam0) #extended to all exit times Lam0 <- rep(Lam0,rtd) #for time dependent covariates: replace by the difference if(any(start!=0))Lam0[start!=0] <- Lam0[start!=0] - Lam0[wstart[start!=0]] lam0 <- rep(lam0,rutd) likely0 <- sum(log(a0 + lam0*eb)) - sum(data$ds + Lam0*ebx) likely <- likely0 tempind <- Nie<=0|Nie>=1 if(any(tempind)){ if(any(Nie<=0))Nie[Nie<=0] <- tol if(any(Nie>=1))Nie[Nie>=1] <- 1-tol } if(p>0)databig$wei <- c(Nie[cause==2],1-Nie[cause==2],rep(1,nrow(datacens))) if(maxiter>=1&p!=0){ for(i in 1:maxiter){ if(p>0){ b00<-b if(i==1)fit <- coxph(Surv(start,Y,cens)~modmat,data=databig,weights=databig$wei,init=b00,x=TRUE,iter.max=maxiter) else fit <- coxph(Surv(start,Y,cens)~modmat,data=databig,weights=databig$wei,x=TRUE,iter.max=maxiter) if(any(is.na(fit$coeff))) stop("X matrix deemed to be singular, variable ",which(is.na(fit$coeff))) b <- fit$coeff ebx <- as.vector(exp(xmat%*%b)) } else ebx <- rep(1,n) eb <- ebx[data$stat==1] s0 <- cumsum((ebx)[n:1])[n:1] ebx.st <- ebx[order(data$start)] s0.st <- ((cumsum(ebx.st[n:1]))[n:1])[index] s0.st <- rep(c(s0.st,0),val1) s0 <- s0 - s0.st #Nie is of length ntd, should be nutd, with the values at times being the sum nietemp <- rep(1:nutd,rutd) Nies <- as.vector(by(Nie,nietemp,sum)) #shorter Nie - only at times utd #s0 only at times utd s0 <- s0[udtimes] lam0u <- lam0 <- Nies/s0 #the cumulative hazard Lam0 <- cumsum(lam0) #extended to all event times Lam0 <- rep(Lam0,rutd) if(data$stat[1]==0) Lam0 <- c(0,Lam0) #extended to all exit times Lam0 <- rep(Lam0,rtd) #for time dependent covariates: replace by the difference if(any(start!=0))Lam0[start!=0] <- Lam0[start!=0] - Lam0[wstart[start!=0]] #the smooting of lam0 if(bwin[1]!=0)lam0s <- krn%*%lam0 else lam0s <- lam0/difft #extended to all event times lam0s <- rep(lam0s,rutd) #compute Nie, only for those with unknown hazard Nie[cause==2] <- as.vector(lam0s*eb/(a+lam0s*eb))[cause==2] #likelihood calculation - manjka ti se likelihood za nicelni model!!! lam0 <- rep(lam0,rutd) likely <- sum(log(a0 + lam0*eb)) - sum(data$ds + Lam0*ebx) if(p>0){ tempind <- Nie<=0|Nie>=1 if(any(tempind)){ if(any(Nie<=0))Nie[Nie<=0] <- tol if(any(Nie>=1))Nie[Nie>=1] <- 1-tol #if(which(tempind)!=nev)warning("Weights smaller than 0") #if(any(is.na( match(which(tempind),c(1,nev)) )))browser() } if(nded==0) break() databig$wei[1:nded] <- c(Nie[cause==2],1-Nie[cause==2]) bd <- abs(b-b00) if(max(bd)< tol) break() } #early stopping time for no covariates??? } } iter <- i #if (maxiter > 1& iter>=maxiter) # warning("Ran out of iterations and did not converge") if(p>0){ if(nded!=0){ resi <- resid(fit,type="schoenfeld") if(!is.null(dim(resi)))resi <- resi[1:(nded/2),] else resi <- resi[1:(nded/2)] swei <- fit$weights[1:(nded/2)] if(is.null(dim(resi))) fishem <- sum((resi^2*swei*(1-swei))) else { fishem <- apply(resi,1,function(x)outer(x,x)) fishem <- t(t(fishem)*swei*(1-swei)) fishem <- matrix(apply(fishem,1,sum),ncol=m) } } else fishem <- 0 fishcox <- solve(fit$var) fisher <- fishcox - fishem fit$var <- solve(fisher) names(fit$coefficients)<-varnames fit$lambda0 <- lam0s } else fit <- list(lambda0 = lam0s) fit$lambda0 <- fit$lambda0[utimes] fit$Lambda0 <- Lam0[udtimes] fit$times <- utd fit$Nie <- Nie fit$bwin <- bwin fit$iter <- i class(fit) <- c("rsadd",class(fit)) fit$loglik <- c(likely0,likely) fit$lam0.ns <- lam0u fit } em <- function (rform, init, control, bwin) { data <- rform$data n <- nrow(data) p <- rform$m id <- order(data$Y) rform$cause <- rform$cause[id] data <- data[id, ] fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) nev <- length(data$Y[data$stat == 1]) data$a <- rep(NA, n) xx <- exp.prep(data[, 4:(nfk + 3),drop=FALSE], data$Y - data$start, rform$ratetable) data$ds <- -log(xx) data1 <- data data1[, 4:(nfk + 3)] <- data[, 4:(nfk + 3)] + data$Y %*% t(fk) xx <- exp.prep(data1[data1$stat == 1, 4:(nfk + 3),drop=FALSE], 1, rform$ratetable) data$a[data$stat == 1] <- -log(xx) if (p > 0) { if (!missing(init) && !is.null(init)) { if (length(init) != p) stop("Wrong length for inital values") } else init <- rep(0, p) beta <- matrix(init, p, 1) } pr.time<-proc.time()[3] Nie <- rep(.5,sum(data$stat==1)) Nie[rform$cause[data$stat==1]<2] <- rform$cause[data$stat==1][rform$cause[data$stat==1]<2] #NEW IN 2.05 varstart <- 3+nfk+1 #first column of covariates varstop <- 3+nfk+p #last column of covariates if(missing(bwin))bwin <- -1 if(bwin<0){ if(p>0)data1 <- data[,-c(varstart:varstop)] #NEW IN 2.05 else data1 <- data nfk <- length(attributes(rform$ratetable)$dimid) names(data)[4:(3+nfk)] <- attributes(rform$ratetable)$dimid expe <- rs.surv(Surv(Y,stat)~1,data,ratetable=rform$ratetable,method="ederer2") esurv <- -log(expe$surv[expe$n.event!=0]) if(esurv[length(esurv)]==Inf)esurv[length(esurv)] <- esurv[length(esurv)-1] x <- seq(.1,3,length=5) dif <- rep(NA,5) options(warn=-1) diter <- max(round(max(data$Y)/356.24),3) for(it in 1:5){ fit <- rsfitterem(data1,NULL,diter,rform$ratetable,control$epsilon,x[it],0,rform$cause,Nie) dif[it] <- sum((esurv-fit$Lambda0)^2) } wh <- which.min(dif) if(wh==1)x <- seq(x[wh],x[wh+1]-.1,length=5) else if(wh==5)x <- c(x, max(data$Y)/ max(diff(data$Y))) if(wh!=1) x <- seq(x[wh-1]+.1,x[wh+1]-.1,length=5) dif <- rep(NA,5) for(it in 1:5){ fit <- rsfitterem(data1,NULL,diter,rform$ratetable,control$epsilon,x[it],0,rform$cause,Nie) dif[it] <- sum((esurv-fit$Lambda0)^2) } options(warn=0) Nie <- fit$Nie bwin <- x[which.min(dif)] } fit <- rsfitterem(data, beta, control$maxit, rform$ratetable, control$epsilon, bwin, p, rform$cause,Nie) Nie <- rep(0,nrow(data)) Nie[data$stat==1] <- fit$Nie fit$Nie <- Nie[order(id)] fit$bwin <- list(bwin=fit$bwin,bwinfac=bwin) fit } rsadd <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, int, na.action, method = "max.lik", init, bwin, centered = FALSE, cause, control, rmap, ...) { call <- match.call() if (missing(control)) control <- glm.control(...) if(!missing(cause)){ #NEW: ce cause ne manjka, ga preverim in dodam kot spremenljivko if (length(cause) != nrow(data)) stop("Length of cause does not match data dimensions") data$cause <- cause rform <- rformulate(formula, data, ratetable, na.action, int, centered, cause) } else{ #no cause if (!missing(rmap)) { rmap <- substitute(rmap) #rform <- rformulate(formula,data, ratetable, na.action, rmap,int, centered) #get the data ready } #else rform <- rformulate(formula,data, ratetable, na.action, rmap, int, centered) } if (method == "EM") { if (!missing(int)) { if (length(int) > 1 | any(int <= 0)) stop("Invalid value of 'int'") } } else { if (missing(int)) int <- c(0,ceiling(max(rform$Y/365.241))) if (length(int) == 1) { if (int <= 0) stop("The value of 'int' must be positive ") int <- 0:int } else if (int[1] != 0) stop("The first interval in 'int' must start with 0") } method <- match.arg(method,c("glm.bin","glm.poi","max.lik","EM")) if (method == "glm.bin" | method == "glm.poi") fit <- glmxp(rform = rform, interval = int, method = method, control = control) else if (method == "max.lik") fit <- maxlik(rform = rform, interval = int, init = init, control = control) else if (method == "EM") fit <- em(rform, init, control, bwin) fit$call <- call fit$formula <- formula fit$data <- rform$data fit$ratetable <- rform$ratetable fit$n <- nrow(rform$data) if (length(rform$na.action)) fit$na.action <- rform$na.action fit$y <- rform$Y.surv fit$method <- method if (method == "EM") { if (!missing(int)) fit$int <- int else fit$int <- ceiling(max(rform$Y[rform$status == 1])/365.241) fit$terms <- rform$Terms if(centered)fit$mvalue <- rform$mvalue } if (method == "max.lik") { fit$terms <- rform$Terms } if (rform$m > 0) fit$linear.predictors <- as.matrix(rform$X) %*% fit$coef[1:ncol(rform$X)] fit } maxlik <- function (rform, interval, subset, init, control) { data <- rform$data max.time <- max(data$Y)/365.241 if (max.time < max(interval)) interval <- interval[1:(sum(max.time > interval) + 1)] fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) data <- cbind(data, offset = rform$offset) data <- survsplit(data, cut = interval[-1] * 365.241, end = "Y", event = "stat", start = "start", episode = "epi", interval = interval) del <- which(data$start==data$Y) if(length(del)) data <- data[-del,] offset <- data$offset data$offset <- NULL d.int <- diff(interval) data[, 4:(nfk + 3)] <- data[, 4:(nfk + 3)] + data$start %*% t(fk) data$lambda <- rep(0, nrow(data)) nsk <- nrow(data[data$stat == 1, ]) xx <- exp.prep(data[data$stat == 1, 4:(nfk + 3),drop=FALSE] + (data[data$stat == 1, ]$Y - data[data$stat == 1, ]$start) %*% t(fk), 1, rform$ratetable) data$lambda[data$stat == 1] <- -log(xx) * 365.241 xx <- exp.prep(data[, 4:(nfk + 3),drop=FALSE], data$Y - data$start, rform$ratetable) data$epi <- NULL data$ds <- -log(xx) data$Y <- data$Y/365.241 data$start <- data$start/365.241 data <- data[, -(4:(3 + nfk))] intn <- length(interval[-1]) m <- rform$m p <- m + intn if (!missing(init) && !is.null(init)) { if (length(init) != p) stop("Wrong length for inital values") } else init <- rep(0, p) if(m>0){ init0 <- init[-(1:m)] data1 <- data[,-(4:(3+m))] } else{ init0 <- init data1 <- data } fit0 <- lik.fit(data1, 0, intn, init0, control, offset) if(m>0){ init[-(1:m)] <- fit0$coef fit <- lik.fit(data, m, intn, init, control, offset) } else fit <- fit0 fit$int <- interval class(fit) <- "rsadd" fit$times <- fit$int*365.241 #dodano za potrebe rs.surv.rsadd fit$Lambda0 <- cumsum(c(0, exp(fit$coef[(m+1):p])*diff(fit$int) )) fit } lik.fit <- function (data, m, intn, init, control, offset) { n <- dim(data)[1] varpos <- 4:(3 + m + intn) x <- data[, varpos] varnames <- names(data)[varpos] lbs <- names(x) x <- as.matrix(x) p <- length(varpos) d <- data$stat ds <- data$ds h <- data$lambda y <- data$Y - data$start maxiter <- control$maxit if (!missing(init) && !is.null(init)) { if (length(init) != p) stop("Wrong length for inital values") } else init <- rep(0, p) b <- matrix(init, p, 1) b0 <- b fit <- mlfit(b, p, x, offset, d, h, ds, y, maxiter, control$epsilon) if (maxiter > 1 & fit$nit >= maxiter) { values <- apply(data[data$stat==1,varpos,drop=FALSE],2,sum) #NEW: deluje tudi, ce je ratetable eno-dimenzionalen problem <- which.min(values) outmes <- "Ran out of iterations and did not converge" if(values[problem]==0)tzero <- "" else tzero <- "only " if(values[problem]<5){ if(!is.na(strsplit(names(values)[problem],"fu")[[1]][2]))outmes <- paste(outmes, "\n This may be due to the fact that there are ",tzero, values[problem], " events on interval",strsplit(names(values)[problem],"fu")[[1]][2],"\n You can use the 'int' argument to change the follow-up intervals in which the baseline excess hazard is assumed constant",sep="") else outmes <- paste(outmes, "\n This may be due to the fact that there are ",tzero, values[problem], " events for covariate value ",names(values)[problem],sep="") } warning(outmes) } b <- as.vector(fit$b) names(b) <- varnames fit <- list(coefficients = b, var = -solve(fit$sd), iter = fit$nit, loglik = fit$loglik) fit } survsplit <- function (data, cut, end, event, start, id = NULL, zero = 0, episode = NULL, interval = NULL) { ntimes <- length(cut) n <- nrow(data) p <- ncol(data) if (length(interval) > 0) { ntimes <- ntimes - 1 sttime <- c(rep(0, n), rep(cut[-length(cut)], each = n)) endtime <- rep(cut, each = n) } else { endtime <- rep(c(cut, Inf), each = n) sttime <- c(rep(0, n), rep(cut, each = n)) } newdata <- lapply(data, rep, ntimes + 1) eventtime <- newdata[[end]] if (start %in% names(data)) starttime <- newdata[[start]] else starttime <- rep(zero, length = (ntimes + 1) * n) starttime <- pmax(sttime, starttime) epi <- rep(0:ntimes, each = n) if (length(interval) > 0) status <- ifelse(eventtime <= endtime & eventtime >= starttime, newdata[[event]], 0) else status <- ifelse(eventtime <= endtime & eventtime > starttime, newdata[[event]], 0) endtime <- pmin(endtime, eventtime) if (length(interval) > 0) drop <- (starttime > endtime) | (starttime == endtime & status == 0) else drop <- starttime >= endtime newdata <- do.call("data.frame", newdata) newdata <- newdata[!drop, ] newdata[, start] <- starttime[!drop] newdata[, end] <- endtime[!drop] newdata[, event] <- status[!drop] if (!is.null(id)) newdata[, id] <- rep(rownames(data), ntimes + 1)[!drop] fu <- NULL if (length(interval) > 2) { for (it in 1:length(interval[-1])) { drop1 <- sum(!drop[1:(it * n - n)]) drop2 <- sum(!drop[(it * n - n + 1):(it * n)]) drop3 <- sum(!drop[(it * n + 1):(length(interval[-1]) * n)]) if (it == 1) fu <- cbind(fu, c(rep(1, drop2), rep(0, drop3))) else if (it == length(interval[-1])) fu <- cbind(fu, c(rep(0, drop1), rep(1, drop2))) else fu <- cbind(fu, c(rep(0, drop1), rep(1, drop2), rep(0, drop3))) } fu <- as.data.frame(fu) names(fu) <- c(paste("fu [", interval[-length(interval)], ",", interval[-1], ")", sep = "")) newdata <- cbind(newdata, fu) } else if (length(interval) == 2) { fu <- rep(1, sum(!drop)) newdata <- cbind(newdata, fu) names(newdata)[ncol(newdata)] <- paste("fu [", interval[1], ",", interval[2], "]", sep = "") } if (!is.null(episode)) newdata[, episode] <- epi[!drop] newdata } glmxp <- function (rform, data, interval, method, control) { if (rform$m == 1) g <- as.integer(as.factor(rform$X[[1]])) else if (rform$m > 1) { gvar <- NULL for (i in 1:rform$m) { gvar <- append(gvar, rform$X[i]) } tabgr <- as.data.frame(table(gvar)) tabgr <- tabgr[, 1:rform$m] n.groups <- dim(tabgr)[1] mat <- do.call("data.frame", gvar) names(mat) <- names(tabgr) tabgr <- cbind(tabgr, g = as.numeric(row.names(tabgr))) mat <- cbind(mat, id = 1:rform$n) c <- merge(tabgr, mat) g <- c[order(c$id), rform$m + 1] } else g <- rep(1, rform$n) vg <- function(X) { n <- dim(X)[1] w <- sum((X$event == 0) & (X$fin == 1) & (X$y != 1)) nd <- sum((X$event == 1) & (X$fin == 1)) ps <- exp.prep(X[, 4:(nfk + 3),drop=FALSE], t.int, rform$ratetable) ld <- n - w/2 lny <- log(sum(X$y)) k <- t.int/365.241 dstar <- sum(-log(ps)/k * X$y) ps <- mean(ps) if (rform$m == 0) data.rest <- X[1, 7 + nfk + rform$m, drop = FALSE] else data.rest <- X[1, c((3 + nfk + 1):(3 + nfk + rform$m), 7 + nfk + rform$m)] cbind(nd = nd, ld = ld, ps = ps, lny = lny, dstar = dstar, k = k, data.rest) } nint <- length(interval) if (nint < 2) stop("Illegal interval value") meje <- interval my.fun <- function(x) { if (x > 1) { x.t <- rep(1, floor(x)) if (x - floor(x) > 0) x.t <- c(x.t, x - floor(x)) x.t } else x } int <- apply(matrix(diff(interval), ncol = 1), 1, my.fun) if (is.list(int)) int <- c(0, cumsum(do.call("c", int))) else int <- c(0, cumsum(int)) int <- int * 365.241 nint <- length(int) X <- cbind(rform$data, grupa = g) fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) Z <- X[X$start >= int[2], ] nz <- dim(Z)[1] Z$fin <- rep(0, nz) Z$event <- rep(0, nz) Z$fu <- rep(0, nz) Z$y <- rep(0, nz) Z$origstart <- Z$start Z$xind <- rep(0, nz) if (nrow(Z) > 0) Z[, 4:(nfk + 3)] <- Z[, 4:(nfk + 3)] + matrix(Z$start, ncol = nfk, byrow = FALSE, nrow = nrow(Z)) * matrix(fk, ncol = nfk, byrow = TRUE, nrow = nrow(Z)) X <- X[X$start < int[2], ] X$fin <- (X$Y <= int[2]) X$event <- X$fin * X$stat ford <- eval(substitute(paste("[", a, ",", b, "]", sep = ""), list(a = meje[1], b = meje[2]))) X$fu <- rep(ford, rform$n - nz) t.int <- int[2] - int[1] X$y <- (pmin(X$Y, int[2]) - X$start)/365.241 X$origstart <- X$start X$xind <- rep(1, nrow(X)) gr1 <- by(X, X$grupa, vg) grm1 <- do.call("rbind", gr1) X <- X[X$fin == 0, ] X$start <- rep(int[2], dim(X)[1]) X <- rbind(X, Z[Z$start < int[3], ]) Z <- Z[Z$start >= int[3], ] temp <- 0 if (nint > 2) { for (i in 3:nint) { ni <- dim(X)[1] if (ni == 0) { temp <- 1 break } X$fin <- X$Y <= int[i] X$event <- X$fin * X$stat l <- sum(int[i - 1] >= meje * 365.241) if(l==1) ftemp <- eval(substitute(paste("[", a, ",", b, "]", sep = ""), list(a = meje[l], b = meje[l + 1]))) else ftemp <- eval(substitute(paste("(", a, ",", b, "]", sep = ""), list(a = meje[l], b = meje[l + 1]))) ford <- c(ford, ftemp) X$fu <- rep(ford[i - 1], ni) t.int <- int[i] - int[i - 1] index <- X$origstart < int[i - 1] index1 <- as.logical(X$xind) if (sum(index) > 0) X[index, 4:(nfk + 3)] <- X[index, 4:(nfk + 3)] + matrix(fk * t.int, ncol = nfk, byrow = TRUE, nrow = sum(index)) X$xind <- rep(1, nrow(X)) X$y <- (pmin(X$Y, int[i]) - X$start)/365.241 gr1 <- by(X, X$grupa, vg) grm1 <- rbind(grm1, do.call("rbind", gr1)) X <- X[X$fin == 0, ] X$start <- rep(int[i], dim(X)[1]) if (i == nint) break X <- rbind(X, Z[Z$start < int[i + 1], ]) X <- X[X$start != X$Y, ] Z <- Z[Z$start >= int[i + 1], ] } l <- sum(int[i - temp] > meje * 365.241) interval <- meje[1:(l + 1)] } else interval <- meje[1:2] grm1$fu <- factor(grm1$fu, levels = unique(ford)) if (method == "glm.bin") { ht <- binomial(link = cloglog) ht$link <- "Hakulinen-Tenkanen relative survival model" ht$linkfun <- function(mu) log(-log((1 - mu)/ps)) ht$linkinv <- function(eta) 1 - exp(-exp(eta)) * ps ht$mu.eta <- function(eta) exp(eta) * exp(-exp(eta)) * ps .ps <- ps <- grm1$ps #assign(".ps", grm1$ps, envir = .GlobalEnv) # ht$initialize <- expression({ # n <- y[, 1] + y[, 2] # y <- ifelse(n == 0, 0, y[, 1]/n) # weights <- weights * n # mustart <- (n * y + 0.01)/(n + 0.02) # mustart[(1 - mustart)/data$ps >= 1] <- data$ps[(1 - mustart)/data$ps >= # 1] * 0.9 # }) if (any(grm1$ld - grm1$nd > grm1$ps * grm1$ld)) { n <- sum(grm1$ld - grm1$nd > grm1$ps * grm1$ld) g <- dim(grm1)[1] warnme <- paste("Observed number of deaths is smaller than the expected in ", n, "/", g, " groups of patients", sep = "") } else warnme <- "" if (length(interval) == 2 & rform$m == 0) stop("No groups can be formed") if (length(interval) == 1 | length(table(grm1$fu)) == 1) grm1$fu <- as.integer(grm1$fu) y <- ifelse(grm1$ld == 0, 0, grm1$nd/grm1$ld) #weights <- weights * grm1$ld mustart <- (grm1$ld * y + 0.01)/(grm1$ld + 0.02) mustart[(1 - mustart)/grm1$ps >= 1] <- grm1$ps[(1 - mustart)/grm1$ps >= 1] * 0.9 if (!length(rform$X)) local.ht <- glm(cbind(nd, ld - nd) ~ -1 + fu + offset(log(k)), data = grm1, family = ht,mustart=mustart) else { xmat <- as.matrix(grm1[, 7:(ncol(grm1) - 1)]) local.ht <- glm(cbind(nd, ld - nd) ~ -1 + xmat + fu + offset(log(k)), data = grm1, family = ht,mustart=mustart) } names(local.ht[[1]]) <- c(names(rform$X), paste("fu", levels(grm1$fu))) } else if (method == "glm.poi") { pot <- poisson() pot$link <- "glm relative survival model with Poisson error" pot$linkfun <- function(mu) log(mu - dstar) pot$linkinv <- function(eta) dstar + exp(eta) #assign(".dstar", grm1$dstar, envir = .GlobalEnv) if (any(grm1$nd - grm1$dstar < 0)) { pot$initialize <- expression({ if (any(y < 0)) stop(paste("Negative values not allowed for", "the Poisson family")) n <- rep.int(1, nobs) #mustart <- pmax(y, .dstar) + 0.1 }) } if (any(grm1$nd - grm1$dstar < 0)) { n <- sum(grm1$nd - grm1$dstar < 0) g <- dim(grm1)[1] warnme <- paste("Observed number of deaths is smaller than the expected in ", n, "/", g, " groups of patients", sep = "") } else warnme <- "" dstar <- grm1$dstar if (length(interval) == 2 & rform$m == 0) stop("No groups can be formed") if (length(interval) == 1 | length(table(grm1$fu)) == 1) grm1$fu <- as.integer(grm1$fu) mustart <- pmax(grm1$nd, grm1$dstar) + 0.1 if (!length(rform$X)) local.ht <- glm(nd ~ -1 + fu, data = grm1, family = pot, offset = grm1$lny,mustart=mustart) else { xmat <- as.matrix(grm1[, 7:(ncol(grm1) - 1)]) local.ht <- glm(nd ~ -1 + xmat + fu, data = grm1, family = pot, offset = grm1$lny,mustart=mustart) } names(local.ht[[1]]) <- c(names(rform$X), paste("fu", levels(grm1$fu))) } else stop(paste("Method '", method, "' not a valid method", sep = "")) class(local.ht) <- c("rsadd", class(local.ht)) local.ht$warnme <- warnme local.ht$int <- interval local.ht$groups <- local.ht$data return(local.ht) } residuals.rsadd <- function (object, type = "schoenfeld", ...) { data <- object$data[order(object$data$Y), ] ratetable <- object$ratetable beta <- object$coef start <- data[, 1] stop <- data[, 2] event <- data[, 3] fk <- (attributes(ratetable)$factor != 1) nfk <- length(fk) n <- nrow(data) scale <- 1 if (object$method == "EM") scale <- 365.241 m <- ncol(data) rem <- m - nfk - 3 interval <- object$int int <- ceiling(max(interval)) R <- data[, 4:(nfk + 3)] lp <- matrix(-log(exp.prep(as.matrix(R), 365.241, object$ratetable))/scale, ncol = 1) fu <- NULL if (object$method == "EM") { death.time <- stop[event == 1] for (it in 1:int) { fu <- as.data.frame(cbind(fu, as.numeric(death.time/365.241 < it & (death.time/365.241) >= (it - 1)))) } if(length(death.time)!=length(unique(death.time))){ utimes <- which(c(1,diff(death.time))!=0) razteg <- function(x){ # x is a 0/1 vector, the output is a vector of length sum(x), with the corresponding rep numbers n <- length(x) repu <- rep(1,n) repu[x==1] <- 0 repu <- rev(cumsum(rev(repu))) repu <- repu[x==1] repu <- -diff(c(repu,0))+1 if(sum(repu)!=n)repu <- c(n-sum(repu),repu) #ce je prvi cas censoring, bo treba se kej narest?? repu } rutd <- rep(0,length(death.time)) rutd[utimes] <- 1 rutd <- razteg(rutd) #from unique event times to event times } else rutd <- rep(1,length(death.time)) lambda0 <- rep(object$lambda0,rutd) } else { pon <- NULL for (i in 1:(length(interval) - 1)) { width <- ceiling(interval[i + 1]) - floor(interval[i]) lo <- interval[i] hi <- min(interval[i + 1], floor(interval[i]) + 1) for (j in 1:width) { fu <- as.data.frame(cbind(fu, as.numeric(stop/365.241 < hi & stop/365.241 >= lo))) names(fu)[ncol(fu)] <- paste("fu", lo, "-", hi, sep = "") if (j == width) { pon <- c(pon, sum(fu[event == 1, (ncol(fu) - width + 1):ncol(fu)])) break() } else { lo <- hi hi <- min(interval[i + 1], floor(interval[i]) + 1 + j) } } } m <- ncol(data) data <- cbind(data, fu) rem <- m - nfk - 3 lambda0 <- rep(exp(beta[rem + 1:(length(interval) - 1)]), pon) fu <- fu[event == 1, , drop = FALSE] beta <- beta[1:rem] } if (int >= 2) { for (j in 2:int) { R <- R + matrix(fk * 365.241, ncol = ncol(R), byrow = TRUE, nrow = n) xx <- exp.prep(R, 365.241, object$ratetable) lp <- cbind(lp, -log(xx)/scale) } } z <- as.matrix(data[, (4 + nfk):m]) out <- resid.com(start, stop, event, z, beta, lp, lambda0, fu, n, rem, int, type) out } resid.com <- function (start, stop, event, z, beta, lp, lambda0, fup, n, rem, int, type) { le <- exp(z %*% beta) olp <- if (int > 1) apply(lp[n:1, ], 2, cumsum)[n:1, ] else matrix(cumsum(lp[n:1])[n:1], ncol = 1) ole <- cumsum(le[n:1])[n:1] lp.st <- lp[order(start), , drop = FALSE] le.st <- le[order(start), , drop = FALSE] starter <- sort(start) starter1 <- c(starter[1], starter[-length(starter)]) index <- c(TRUE, (starter != starter1)[-1]) starter <- starter[index] val1 <- apply(matrix(starter, ncol = 1), 1, function(x, Y) sum(x >= Y), stop) val1 <- c(val1[1], diff(val1), length(stop) - val1[length(val1)]) olp.st <- (apply(lp.st[n:1, , drop = FALSE], 2, cumsum)[n:1, , drop = FALSE])[index, , drop = FALSE] olp.st <- apply(olp.st, 2, function(x) rep(c(x, 0), val1)) olp <- olp - olp.st olp <- olp[event == 1, ] olp <- apply(fup * olp, 1, sum) ole.st <- cumsum(le.st[n:1])[n:1][index] ole.st <- rep(c(ole.st, 0), val1) ole <- ole - ole.st ole <- ole[event == 1] * lambda0 s0 <- ole + olp sc <- NULL zb <- NULL kzb <- NULL f1 <- function(x) rep(mean(x), length(x)) f2 <- function(x) apply(x, 2, f1) f3 <- function(x) apply(x, 1:2, f1) ties <- length(unique(stop[event == 1])) != length(stop[event == 1]) for (k in 1:rem) { zlp <- apply((z[, k] * lp)[n:1, , drop = FALSE], 2, cumsum)[n:1, , drop = FALSE] zlp.st <- (apply((z[, k] * lp.st)[n:1, , drop = FALSE], 2, cumsum)[n:1, , drop = FALSE])[index, , drop = FALSE] zlp.st <- apply(zlp.st, 2, function(x) rep(c(x, 0), val1)) zlp <- zlp - zlp.st zlp <- zlp[event == 1, , drop = FALSE] zlp <- apply(fup * zlp, 1, sum) zle <- cumsum((z[, k] * le)[n:1])[n:1] zle.st <- cumsum((z[, k] * le.st)[n:1])[n:1][index] zle.st <- rep(c(zle.st, 0), val1) zle <- zle - zle.st zle <- zle[event == 1] zle <- zle * lambda0 s1 <- zle + zlp zb <- cbind(zb, s1/s0) kzb <- cbind(kzb, zle/s0) } s1ties <- cbind(zb, kzb) if (ties) { s1ties <- by(s1ties, stop[event == 1], f2) s1ties <- do.call("rbind", s1ties) } zb <- s1ties[, 1:rem, drop = FALSE] kzb <- s1ties[, -(1:rem), drop = FALSE] sc <- z[event == 1, , drop = FALSE] - zb row.names(sc) <- stop[event == 1] out.temp <- function(x) outer(x, x, FUN = "*") krez <- rez <- array(matrix(NA, ncol = rem, nrow = rem), dim = c(rem, rem, sum(event == 1))) for (a in 1:rem) { for (b in a:rem) { zzlp <- apply((z[, a] * z[, b] * lp)[n:1, , drop = FALSE], 2, cumsum)[n:1, , drop = FALSE] zzlp.st <- (apply((z[, a] * z[, b] * lp.st)[n:1, , drop = FALSE], 2, cumsum)[n:1, , drop = FALSE])[index, , drop = FALSE] zzlp.st <- apply(zzlp.st, 2, function(x) rep(c(x, 0), val1)) zzlp <- zzlp - zzlp.st zzlp <- zzlp[event == 1, , drop = FALSE] zzlp <- apply(fup * zzlp, 1, sum) zzle <- cumsum((z[, a] * z[, b] * le)[n:1])[n:1] zzle.st <- cumsum((z[, a] * z[, b] * le.st)[n:1])[n:1][index] zzle.st <- rep(c(zzle.st, 0), val1) zzle <- zzle - zzle.st zzle <- zzle[event == 1] zzle <- zzle * lambda0 s2 <- zzlp + zzle s20 <- s2/s0 ks20 <- zzle/s0 s2ties <- cbind(s20, ks20) if (ties) { s2ties <- by(s2ties, stop[event == 1], f2) s2ties <- do.call("rbind", s2ties) } rez[a, b, ] <- rez[b, a, ] <- s2ties[, 1] krez[a, b, ] <- krez[b, a, ] <- s2ties[, 2] } } juhu <- apply(zb, 1, out.temp) if (is.null(dim(juhu))) juhu1 <- array(data = matrix(juhu, ncol = a), dim = c(a, a, length(zb[, 1]))) else juhu1 <- array(data = apply(juhu, 2, matrix, ncol = a), dim = c(a, a, length(zb[, 1]))) varr <- rez - juhu1 kjuhu <- apply(cbind(zb, kzb), 1, function(x) outer(x[1:rem], x[-(1:rem)], FUN = "*")) if (is.null(dim(kjuhu))) kjuhu1 <- array(data = matrix(kjuhu, ncol = rem), dim = c(rem, rem, length(zb[, 1]))) else kjuhu1 <- array(data = apply(kjuhu, 2, matrix, ncol = rem), dim = c(rem, rem, length(zb[, 1]))) kvarr <- krez - kjuhu1 for (i in 1:dim(varr)[1]) varr[i, i, which(varr[i, i, ] < 0)] <- 0 for (i in 1:dim(kvarr)[1]) kvarr[i, i, which(kvarr[i, i, ] < 0)] <- 0 varr1 <- apply(varr, 1:2, sum) kvarr1 <- apply(kvarr, 1:2, sum) if (type == "schoenfeld") out <- list(res = sc, varr1 = varr1, varr = varr, kvarr = kvarr, kvarr1 = kvarr1) out } rs.br <- function (fit, sc, rho = 0, test = "max", global = TRUE) { test <- match.arg(test,c("max","cvm")) if (inherits(fit, "rsadd")) { if (missing(sc)) sc <- resid(fit, "schoenfeld") sresid <- sc$res varr <- sc$varr sresid <- as.matrix(sresid) } else { coef <- fit$coef options(warn = -1) sc <- coxph.detail(fit) options(warn = 0) sresid <- sc$score varr <- sc$imat if (is.null(dim(varr))) varr <- array(varr, dim = c(1, 1, length(varr))) sresid <- as.matrix(sresid) } if (inherits(fit, "coxph")) { if(is.null(fit$data)){ temp <- fit$y class(temp) <- "matrix" if(ncol(fit$y)==2)temp <- data.frame(rep(0,nrow(fit$y)),temp) if(is.null(fit$x))stop("The coxph model should be called with x=TRUE argument") fit$data <- data.frame(temp,fit$x) names(fit$data)[1:3] <- c("start","Y","stat") } } data <- fit$data[order(fit$data$Y), ] time <- data$Y[data$stat == 1] ties <- (length(unique(time)) != length(time)) keep <- 1:(ncol(sresid)) options(warn = -1) scaled <- NULL varnova <- NULL if (ncol(sresid) == 1) { varr <- varr[1, 1, ] scaled <- sresid/sqrt(varr) } else { for (i in 1:ncol(sresid)) varnova <- cbind(varnova,varr[i,i,]) scaled <- sresid/sqrt(varnova) } options(warn = 0) nvar <- ncol(sresid) survfit <- getFromNamespace("survfit", "survival") temp <- survfit(fit$y~1, type = "kaplan-meier") n.risk <- temp$n.risk n.time <- temp$time if (temp$type == "right") { cji <- matrix(fit$y, ncol = 2) n.risk <- n.risk[match(cji[cji[, 2] == 1, 1], n.time)] } else { cji <- matrix(fit$y, ncol = 3) n.risk <- n.risk[match(cji[cji[, 3] == 1, 2], n.time)] } n.risk <- sort(n.risk, decreasing = TRUE) varnames <- names(fit$coef)[keep] u2 <- function(bb) { n <- length(bb) 1/n * (sum(bb^2) - sum(bb)^2/n) } wc <- function(x, k = 1000) { a <- 1 for (i in 1:k) a <- a + 2 * (-1)^i * exp(-2 * i^2 * pi^2 * x) a } brp <- function(x, n = 1000) { a <- 1 for (i in 1:n) a <- a - 2 * (-1)^(i - 1) * exp(-2 * i^2 * x^2) a } global <- as.numeric(global & ncol(sresid) > 1) table <- NULL bbt <- as.list(1:(nvar + global)) for (i in 1:nvar) { if (nvar != 1) usable <- which(varr[i, i, ] > 1e-12) else usable <- which(varr > 1e-12) w <- (n.risk[usable])^rho w <- w/sum(w) if (nvar != 1) { sci <- scaled[usable, i] } else sci <- scaled[usable] if (ties) { if (inherits(fit, "rsadd")) { sci <- as.vector(by(sci, time[usable], function(x) sum(x)/sqrt(length(x)))) w <- as.vector(by(w, time[usable], sum)) } else { w <- w * as.vector(table(time))[usable] w <- w/sum(w) } } sci <- sci * sqrt(w) timescale <- cumsum(w) bm <- cumsum(sci) bb <- bm - timescale * bm[length(bm)] if (test == "max") table <- rbind(table, c(max(abs(bb)), 1 - brp(max(abs(bb))))) else if (test == "cvm") table <- rbind(table, c(u2(bb), 1 - wc(u2(bb)))) bbt[[i]] <- cbind(timescale, bb) } if (inherits(fit, "rsadd")) { beta <- fit$coef[1:(length(fit$coef) - length(fit$int) + 1)] } else beta <- fit$coef if (global) { qform <- function(matrix, vector) t(vector) %*% matrix %*% vector diagonal <- apply(varr, 3, diag) sumdiag <- apply(diagonal, 2, sum) usable <- which(sumdiag > 1e-12) score <- t(beta) %*% t(sresid[usable, ]) varr <- varr[, , usable] qf <- apply(varr, 3, qform, vector = beta) w <- (n.risk[usable])^rho w <- w/sum(w) sci <- score/(qf)^0.5 if (ties) { if (inherits(fit, "rsadd")) { sci <- as.vector(by(t(sci), time[usable], function(x) sum(x)/sqrt(length(x)))) w <- as.vector(by(w, time[usable], sum)) } else { w <- w * as.vector(table(time)) w <- w/sum(w) } } sci <- sci * sqrt(w) timescale <- cumsum(w) bm <- cumsum(sci) bb <- bm - timescale * bm[length(bm)] if (test == "max") table <- rbind(table, c(max(abs(bb)), 1 - brp(max(abs(bb))))) else if (test == "cvm") table <- rbind(table, c(u2(bb), 1 - wc(u2(bb)))) bbt[[nvar + 1]] <- cbind(timescale, bb) varnames <- c(varnames, "GLOBAL") } dimnames(table) <- list(varnames, c(test, "p")) out <- list(table = table, bbt = bbt, rho = rho) class(out) <- "rs.br" out } rs.zph <- function (fit, sc, transform = "identity", var.type = "sum") { if (inherits(fit, "rsadd")) { if (missing(sc)) sc <- resid(fit, "schoenfeld") sresid <- sc$res varr <- sc$kvarr fvar <- solve(sc$kvarr1) sresid <- as.matrix(sresid) } else { coef <- fit$coef options(warn = -1) sc <- coxph.detail(fit) options(warn = 0) sresid <- as.matrix(resid(fit, "schoenfeld")) varr <- sc$imat fvar <- fit$var } data <- fit$data[order(fit$data$Y), ] time <- data$Y stat <- data$stat if (!inherits(fit, "rsadd")) { ties <- as.vector(table(time[stat==1])) if(is.null(dim(varr))) varr <- rep(varr/ties,ties) else{ varr <- apply(varr,1:2,function(x)rep(x/ties,ties)) varr <- aperm(varr,c(2,3,1)) } } keep <- 1:(length(fit$coef) - length(fit$int) + 1) varnames <- names(fit$coef)[keep] nvar <- length(varnames) ndead <- length(sresid)/nvar if (inherits(fit, "rsadd")) times <- time[stat == 1] else times <- sc$time if (is.character(transform)) { tname <- transform ttimes <- switch(transform, identity = times, rank = rank(times), log = log(times), km = { fity <- Surv(time, stat) temp <- survfit(fity~1) t1 <- temp$surv[temp$n.event > 0] t2 <- temp$n.event[temp$n.event > 0] km <- rep(c(1, t1), c(t2, 0)) if (is.null(attr(sresid, "strata"))) 1 - km else (1 - km[sort.list(sort.list(times))]) }, stop("Unrecognized transform")) } else { tname <- deparse(substitute(transform)) ttimes <- transform(times) } if (var.type == "each") { invV <- apply(varr, 3, function(x) try(solve(x), silent = TRUE)) if (length(invV) == length(varr)){ if(!is.numeric(invV)){ usable <- rep(FALSE, dim(varr)[3]) options(warn=-1) invV <- as.numeric(invV) usable[1:(min(which(is.na(invV)))-1)] <- TRUE invV <- invV[usable] sresid <- sresid[usable,,drop=FALSE] options(warn=0) } else usable <- rep(TRUE, dim(varr)[3]) } else { usable <- unlist(lapply(invV, is.matrix)) if (!any(usable)) stop("All the matrices are singular") invV <- invV[usable] sresid <- sresid[usable, , drop = FALSE] } di1 <- dim(varr)[1] di3 <- sum(usable) u <- array(data = matrix(unlist(invV), ncol = di1), dim = c(di1, di1, di3)) uv <- cbind(matrix(u, ncol = di1, byrow = TRUE), as.vector(t(sresid))) uv <- array(as.vector(t(uv)), dim = c(di1 + 1, di1, di3)) r2 <- t(apply(uv, 3, function(x) x[1:di1, ] %*% x[di1 + 1, ])) r2 <- matrix(r2, ncol = di1) whr2 <- apply(r2<100,1,function(x)!any(x==FALSE)) usable <- as.logical(usable*whr2) r2 <- r2[usable,,drop=FALSE] u <- u[,,usable] dimnames(r2) <- list(times[usable], varnames) temp <- list(x = ttimes[usable], y = r2 + outer(rep(1, sum(usable)), fit$coef[keep]), var = u, call = call, transform = tname) } else if (var.type == "sum") { xx <- ttimes - mean(ttimes) r2 <- t(fvar %*% t(sresid) * ndead) r2 <- as.matrix(r2) dimnames(r2) <- list(times, varnames) temp <- list(x = ttimes, y = r2 + outer(rep(1, ndead), fit$coef[keep]), var = fvar, transform = tname) } else stop("Unknown 'var.type'") class(temp) <- "rs.zph" temp } plot.rs.zph <- function (x,resid = TRUE, df = 4, nsmo = 40, var, cex = 1, add = FALSE, col = 1, lty = 1, xlab, ylab, xscale = 1, ...) { #require(splines) xx <- x$x if(x$transform=="identity")xx <- xx/xscale yy <- x$y d <- nrow(yy) df <- max(df) nvar <- ncol(yy) pred.x <- seq(from = min(xx), to = max(xx), length = nsmo) temp <- c(pred.x, xx) lmat <- splines::ns(temp, df = df, intercept = TRUE) pmat <- lmat[1:nsmo, ] xmat <- lmat[-(1:nsmo), ] qmat <- qr(xmat) if (missing(ylab)) ylab <- paste("Beta(t) for", dimnames(yy)[[2]]) if (missing(xlab)) xlab <- "Time" if (missing(var)) var <- 1:nvar else { if (is.character(var)) var <- match(var, dimnames(yy)[[2]]) if (any(is.na(var)) || max(var) > nvar || min(var) < 1) stop("Invalid variable requested") } if (x$transform == "log") { xx <- exp(xx) pred.x <- exp(pred.x) } else if (x$transform != "identity") { xtime <- as.numeric(dimnames(yy)[[1]])/xscale apr1 <- approx(xx, xtime, seq(min(xx), max(xx), length = 17)[2 * (1:8)]) temp <- signif(apr1$y, 2) apr2 <- approx(xtime, xx, temp) xaxisval <- apr2$y xaxislab <- rep("", 8) for (i in 1:8) xaxislab[i] <- format(temp[i]) } for (i in var) { y <- yy[, i] yhat <- pmat %*% qr.coef(qmat, y) yr <- range(yhat, y) if (!add) { if (x$transform == "identity") plot(range(xx), yr, type = "n", xlab = xlab, ylab = ylab[i],...) else if (x$transform == "log") plot(range(xx), yr, type = "n", xlab = xlab, ylab = ylab[i],log = "x", ...) else { plot(range(xx), yr, type = "n", xlab = xlab, ylab = ylab[i],axes = FALSE, ...) axis(1, xaxisval, xaxislab) axis(2) box() } } if (resid) points(xx, y, cex = cex, col = col) lines(pred.x, yhat, col = col, lty = lty) } } plot.rs.br <- function (x, var, ylim = c(-2, 2), xlab, ylab, ...) { bbt <- x$bbt par(ask = TRUE) if (missing(var)) var <- 1:nrow(x$table) ychange <- FALSE if (missing(ylab)) ylab <- paste("Brownian bridge for", row.names(x$table)) else { if (length(ylab) == 1 & nrow(x$table) > 1) ylab <- rep(ylab, nrow(x$table)) } if (missing(xlab)) xlab <- "Time" for (i in var) { timescale <- bbt[[i]][, 1] bb <- bbt[[i]][, 2] plot(c(0, timescale), c(0, bb), type = "l", ylim = ylim, xlab = xlab, ylab = ylab[i], ...) abline(h = 1.36, col = 2) abline(h = 1.63, col = 2) abline(h = -1.36, col = 2) abline(h = -1.63, col = 2) } par(ask = FALSE) } Kernmatch <- function (t, tv, b, tD, nt4) { kmat <- NULL for (it in 1:(length(nt4) - 1)) { kmat1 <- (outer(t[(nt4[it] + 1):nt4[it + 1]], tv, "-")/b[it]) kmat1 <- kmat1^(kmat1 >= 0) kmat <- rbind(kmat, pmax(1 - kmat1^2, 0) * (1.5/b[it])) } kmat } kernerleftch <- function (td, b, nt4) { n <- length(td) ttemp <- td[td >= b[1]] ntemp <- length(ttemp) if (ntemp == n) nt4 <- c(0, nt4[-1]) else { nfirst <- n - ntemp nt4 <- c(0, 1:nfirst, nt4[-1]) b <- c(td[1:nfirst], b) } krn <- Kernmatch(td, td, b, max(td), nt4) krn } invtime <- function (y = 0.1, age = 23011, sex = "male", year = 9497, scale = 1, ratetable = relsurv::slopop, lower, upper) { if (!is.numeric(age)) stop("\"age\" must be numeric", call. = FALSE) if (!is.numeric(y)) stop("\"y\" must be numeric", call. = FALSE) if (!is.numeric(scale)) stop("\"scale\" must be numeric", call. = FALSE) temp <- data.frame(age = age, sex = I(sex), year = year) if (missing(lower)) { if (!missing(upper)) stop("Argument \"lower\" is missing, with no default", call. = FALSE) nyears <- round((110 - age/365.241)) tab <- data.frame(age = rep(age, nyears), sex = I(rep(sex, nyears)), year = rep(year, nyears)) vred <- 1 - survexp(c(0, 1:(nyears - 1)) * 365.241 ~ ratetable(age = age, sex = sex, year = year), ratetable = ratetable, data = tab, cohort = FALSE) place <- sum(vred <= y) if (place == 0) lower <- 0 else lower <- floor((place - 1) * 365.241 - place) upper <- ceiling(place * 365.241 + place) } else { if (missing(upper)) stop("Argument \"upper\" is missing, with no default", call. = FALSE) if (!is.integer(lower)) lower <- floor(lower) if (!is.integer(upper)) upper <- ceiling(upper) if (upper <= lower) stop("'upper' must be higher than 'lower'", call. = FALSE) } lower <- max(0, lower) tab <- data.frame(age = rep(age, upper - lower + 1), sex = I(rep(sex, upper - lower + 1)), year = rep(year, upper - lower + 1)) vred <- 1 - survexp((lower:upper) ~ ratetable(age = age, sex = sex, year = year), ratetable = ratetable, data = tab, cohort = FALSE) place <- sum(vred <= y) if (place == 0) warning(paste("The event happened on or before day", lower), call. = FALSE) if (place == length(vred)) warning(paste("The event happened on or after day", upper), call. = FALSE) t <- (place + lower - 1)/scale age <- round(age/365.241, 0.01) return(list(age, sex, year, Y = y, T = t)) } rsmul <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, int, na.action, init, method = "mul", control,rmap, ...) { #require(survival) if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula,data, ratetable, na.action,rmap,int) U <- rform$data if (missing(int)) int <- ceiling(max(rform$Y/365.241)) if(length(int)!=1)int <- max(int) fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) if (method == "mul") { U <- survsplit(U, cut = (1:int) * 365.241, end = "Y", event = "stat", start = "start", episode = "epi") fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) U[, 4:(nfk + 3)] <- U[, 4:(nfk + 3)] + 365.241 * (U$epi) %*% t(fk) nsk <- dim(U)[1] xx <- exp.prep(U[, 4:(nfk + 3),drop=FALSE], 365.241, rform$ratetable) lambda <- -log(xx)/365.241 } else if (method == "mul1") { U$id <- 1:dim(U)[1] my.fun <- function(x, attcut, nfk, fk) { intr <- NULL for (i in 1:nfk) { if (fk[i]) { n1 <- max(findInterval(as.numeric(x[3 + i]) + as.numeric(x[1]), attcut[[i]]) + 1, 2) n2 <- findInterval(as.numeric(x[3 + i]) + as.numeric(x[2]), attcut[[i]]) if (n2 > n1 & length(attcut[[i]] > 1)) { if (n2 > length(attcut[[i]])) n2 <- length(attcut[[i]]) intr <- c(intr, as.numeric(attcut[[i]][n1:n2]) - as.numeric(x[3 + i])) } } } intr <- sort(unique(c(intr, as.numeric(x[2])))) intr } attcut <- attributes(rform$ratetable)$cutpoints intr <- apply(U[, 1:(3 + nfk)], 1, my.fun, attcut, nfk, fk) dolg <- unlist(lapply(intr, length)) newdata <- lapply(U, rep, dolg) stoptime <- unlist(intr) starttime <- c(-1, stoptime[-length(stoptime)]) first <- newdata$id != c(-1, newdata$id[-length(newdata$id)]) starttime[first] <- newdata$start[first] last <- newdata$id != c(newdata$id[-1], -1) event <- rep(0, length(newdata$id)) event[last] <- newdata$stat[last] U <- do.call("data.frame", newdata) U$start <- starttime U$Y <- stoptime U$stat <- event U[, 4:(nfk + 3)] <- U[, 4:(nfk + 3)] + (U$start) %*% t(fk) nsk <- dim(U)[1] xx <- exp.prep(U[, 4:(nfk + 3),drop=FALSE], 1, rform$ratetable) lambda <- -log(xx)/1 } else stop("'method' must be one of 'mul' or 'mul1'") U$lambda <- log(lambda) if (rform$m == 0) fit <- coxph(Surv(start, Y, stat) ~ 1 + offset(lambda), data = U, init = init, control = control, x = TRUE, ...) else { xmat <- as.matrix(U[, (3 + nfk + 1):(ncol(U) - 2)]) fit <- coxph(Surv(start, Y, stat) ~ xmat + offset(lambda), data = U, init = init, control = control, x = TRUE, ...) names(fit[[1]]) <- names(U)[(3 + nfk + 1):(ncol(U) - 2)] } class(fit) <- c("rsmul",class(fit)) fit$basehaz <- basehaz(fit) #NEW 2.05 fit$data <- rform$data fit$call <- match.call() fit$int <- int if (length(rform$na.action)) fit$na.action <- rform$na.action fit } rstrans <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, int, na.action, init, control,rmap, ...) { if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula, data, ratetable, na.action, rmap, int) if (missing(int)) int <- ceiling(max(rform$Y/365.241)) fk <- (attributes(rform$ratetable)$factor != 1) nfk <- length(fk) if (rform$type == "counting") { start <- 1 - exp.prep(rform$R, rform$start, rform$ratetable) } else start <- rep(0, rform$n) stop <- 1 - exp.prep(rform$R, rform$Y, rform$ratetable) if(any(stop==0&rform$Y!=0))stop[stop==0&rform$Y!=0] <- .Machine$double.eps if(length(int)!=1)int <- max(int) data <- rform$data stat <- rform$status if (rform$m == 0) { if (rform$type == "counting") fit <- coxph(Surv(start, stop, stat) ~ 1, init = init, control = control, x = TRUE, ...) else fit <- coxph(Surv(stop, stat) ~ 1, init = init, control = control, x = TRUE, ...) } else { xmat <- as.matrix(data[, (4 + nfk):ncol(data)]) fit <- coxph(Surv(start, stop, stat) ~ xmat, init = init, control = control, x = TRUE, ...) names(fit[[1]]) <- names(rform$X) } fit$call <- match.call() if (length(rform$na.action)) fit$na.action <- rform$na.action data$start <- start data$Y <- stop fit$data <- data fit$int <- int return(fit) } transrate <- function (men, women, yearlim, int.length = 1) { if (any(dim(men) != dim(women))) stop("The men and women matrices must be of the same size. \n In case of missing values at the end carry the last value forward") if ((yearlim[2] - yearlim[1])/int.length + 1 != dim(men)[2]) stop("'yearlim' cannot be divided into intervals of equal length") if (!is.matrix(men) | !is.matrix(women)) stop("input tables must be of class matrix") dimi <- dim(men) temp <- array(c(men, women), dim = c(dimi, 2)) temp <- -log(temp)/365.241 temp <- aperm(temp, c(1, 3, 2)) cp <- as.date(apply(matrix(yearlim[1] + int.length * (0:(dimi[2] - 1)), ncol = 1), 1, function(x) { paste("1jan", x, sep = "") })) attributes(temp) <- list(dim = c(dimi[1], 2, dimi[2]), dimnames = list(age=as.character(0:(dimi[1] - 1)), sex=c("male", "female"), year=as.character(yearlim[1] + int.length * (0:(dimi[2] - 1)))), dimid = c("age", "sex", "year"), factor = c(0, 1, 0),type=c(2,1,3), cutpoints = list((0:(dimi[1] - 1)) * (365.241), NULL, cp), class = "ratetable") attributes(temp)$summary <- function (R) { x <- c(format(round(min(R[, 1])/365.241, 1)), format(round(max(R[, 1])/365.241, 1)), sum(R[, 2] == 1), sum(R[, 2] == 2)) x2 <- as.character(as.date(c(min(R[, 3]), max(R[, 3])))) paste(" age ranges from", x[1], "to", x[2], "years\n", " male:", x[3], " female:", x[4], "\n", " date of entry from", x2[1], "to", x2[2], "\n") } temp } transrate.hld <- function(file, cut.year,race){ nfiles <- length(file) data <- NULL for(it in 1:nfiles){ tdata <- read.table(file[it],sep=",",header=TRUE) if(!any(tdata$TypeLT==1)) stop("Currently only TypeLT 1 is implemented") names(tdata) <- gsub(".","",names(tdata),fixed=TRUE) tdata <- tdata[,c("Country","Year1","Year2","TypeLT","Sex","Age","AgeInt","qx")] tdata <- tdata[tdata$TypeLT==1,] #NEW - prej sem gledala tudi AgeInt, izkaze se, da ni treba. pri q(x) bi bilo vseeno tudi, ce bi gledala TypeLT=3. tdata <- tdata[!is.na(tdata$AgeInt),] #NEW - vrzem ven zadnji interval, ki gre v neskoncnost in vsi umrejo (inf hazard) if(!missing(race))tdata$race <- rep(race[it],nrow(tdata)) data <- rbind(data,tdata) } if(length(unique(data$Country))>1)warning("The data belongs to different countries") data <- data[order(data$Year1,data$Age),] data$qx <- as.character(data$qx) options(warn = -1) data$qx[data$qx=="."] <- NA data$qx <- as.numeric(data$qx) options(warn = 0) if(missing(cut.year)){ y1 <- unique(data$Year1) y2 <- unique(data$Year2) if(any(apply(cbind(y1[-1],y2[-length(y2)]),1,diff)!=-1))warning("Data is not given for all the cut.year between the minimum and the maximum, use argument 'cut.year'") } else y1 <- cut.year if(length(y1)!=length(unique(data$Year1)))stop("Length 'cut.year' must match the number of unique values of Year1") cp <- as.date(apply(matrix(y1,ncol=1),1,function(x){paste("1jan",x,sep="")})) dn2 <- as.character(y1) amax <- max(data$Age) a.fun <- function(data,amax){ mdata <- data[data$Sex==1,] wdata <- data[data$Sex==2,] men <-NULL women <- NULL k <- sum(mdata$Age==0) mind <- c(which(mdata$Age[-nrow(mdata)] != mdata$Age[-1]-1),nrow(mdata)) wind <- c(which(wdata$Age[-nrow(wdata)] != wdata$Age[-1]-1),nrow(wdata)) mst <- wst <- 1 for(it in 1:k){ qx <- mdata[mst:mind[it],]$qx lqx <- length(qx) if(lqx!=amax+1){ nmiss <- amax + 1 - lqx qx <- c(qx,rep(qx[lqx],nmiss)) } naqx <- max(which(!is.na(qx))) if(naqx!=amax+1) qx[(naqx+1):(amax+1)] <- qx[naqx] men <- cbind(men,qx) mst <- mind[it]+1 qx <- wdata[wst:wind[it],]$qx lqx <- length(qx) if(lqx!=amax+1){ nmiss <- amax + 1 - lqx qx <- c(qx,rep(qx[lqx],nmiss)) } naqx <- max(which(!is.na(qx))) if(naqx!=amax+1) qx[(naqx+1):(amax+1)] <- qx[naqx] women <- cbind(women,qx) wst <- wind[it]+1 } men<- -log(1-men)/365.241 women<- -log(1-women)/365.241 dims <- c(dim(men),2) array(c(men,women),dim=dims) } if(missing(race)){ out <- a.fun(data,amax) dims <- dim(out) attributes(out)<-list( dim=dims, dimnames=list(as.character(0:amax),as.character(y1),c("male","female")), dimid=c("age","year","sex"), factor=c(0,0,1),type=c(2,3,1), cutpoints=list((0:amax)*(365.241),cp,NULL), class="ratetable" ) } else{ race.val <- unique(race) if(length(race)!=length(file))stop("Length of 'race' must match the number of files") for(it in 1:length(race.val)){ if(it==1){ out <- a.fun(data[data$race==race.val[it],],amax) dims <- dim(out) out <- array(out,dim=c(dims,1)) } else{ out1 <- array(a.fun(data[data$race==race.val[it],],amax),dim=c(dims,1)) out <- array(c(out,out1),dim=c(dims,it)) } } attributes(out)<-list( dim=c(dims,it), dimnames=list(age=as.character(0:amax),year=as.character(y1),sex=c("male","female"),race=race.val), dimid=c("age","year","sex","race"), factor=c(0,0,1,1),type=c(2,3,1,1), cutpoints=list((0:amax)*(365.241),cp,NULL,NULL), class="ratetable" ) } attributes(out)$summary <- function (R) { x <- c(format(round(min(R[, 1])/365.241, 1)), format(round(max(R[, 1])/365.241, 1)), sum(R[, 3] == 1), sum(R[, 3] == 2)) x2 <- as.character(as.date(c(min(R[, 2]), max(R[, 2])))) paste(" age ranges from", x[1], "to", x[2], "years\n", " male:", x[3], " female:", x[4], "\n", " date of entry from", x2[1], "to", x2[2], "\n") } out } transrate.hmd <- function(male,female){ nfiles <- 2 men <- try(read.table(male,sep="",header=TRUE),silent=TRUE) if(class(men)=="try-error")men <- read.table(male,sep="",header=TRUE,skip=1) men <- men[,c("Year","Age","qx")] y1 <- sort(unique(men$Year)) ndata <- nrow(men)/111 if(round(ndata)!=ndata)stop("Each year must contain ages from 0 to 110") men <- matrix(men$qx, ncol=ndata) men <- matrix(as.numeric(men),ncol=ndata) women <- try(read.table(female,sep="",header=TRUE),silent=TRUE) if(class(women)=="try-error")women <- read.table(female,sep="",header=TRUE,skip=1) women <- women[,"qx"] if(length(women)!=length(men))stop("Number of rows in the table must be equal for both sexes") women <- matrix(women, ncol=ndata) women <- matrix(as.numeric(women),ncol=ndata) cp <- as.date(apply(matrix(y1,ncol=1),1,function(x){paste("1jan",x,sep="")})) dn2 <- as.character(y1) tfun <- function(vec){ ind <- which(vec == 1 | is.na(vec)) if(length(ind)>0)vec[min(ind):length(vec)] <- 0.999 vec } men <- apply(men,2,tfun) women <- apply(women,2,tfun) men<- -log(1-men)/365.241 women<- -log(1-women)/365.241 nr <- nrow(men)-1 dims <- c(dim(men),2) out <- array(c(men,women),dim=dims) attributes(out)<-list( dim=dims, dimnames=list(age=as.character(0:nr),year=as.character(y1),sex=c("male","female")), dimid=c("age","year","sex"), factor=c(0,0,1),type=c(2,3,1), cutpoints=list((0:nr)*(365.241),cp,NULL), class="ratetable" ) attributes(out)$summary <- function (R) { x <- c(format(round(min(R[, 1])/365.241, 1)), format(round(max(R[, 1])/365.241, 1)), sum(R[, 3] == 1), sum(R[, 3] == 2)) x2 <- as.character(as.date(c(min(R[, 2]), max(R[, 2])))) paste(" age ranges from", x[1], "to", x[2], "years\n", " male:", x[3], " female:", x[4], "\n", " date of entry from", x2[1], "to", x2[2], "\n") } out } joinrate <- function(tables,dim.name="country"){ nfiles <- length(tables) if(is.null(names(tables))) names(tables) <- paste("D",1:nfiles,sep="") if(any(!unlist(lapply(tables,is.ratetable))))stop("Tables must be in ratetable format") if(length(attributes(tables[[1]])$dim)!=3)stop("Currently implemented only for ratetables with 3 dimensions") if(is.null(attr(tables[[1]],"dimid")))attr(tables[[1]],"dimid") <- names((attr(tables[[1]],"dimnames"))) for(it in 2:nfiles){ if(is.null(attr(tables[[it]],"dimid")))attr(tables[[it]],"dimid") <- names((attr(tables[[it]],"dimnames"))) if(length(attributes(tables[[it]])$dimid)!=3)stop("Each ratetable must have 3 dimensions: age, year and sex") mc <- match(attributes(tables[[it]])$dimid,attributes(tables[[1]])$dimid,nomatch=0) if(any(mc)==0) stop("Each ratetable must have 3 dimensions: age, year and sex") if(any(mc!=1:3)){ atts <- attributes(tables[[it]]) tables[[it]] <- aperm(tables[[it]],mc) atts$dimid <- atts$dimid[mc] atts$dimnames <- atts$dimnames[mc] atts$cutpoints <- atts$cutpoints[mc] atts$factor <- atts$factor[mc] atts$type <- atts$type[mc] atts$dim <- atts$dim[mc] attributes(tables[[it]]) <- atts } } list.eq <- function(l1,l2){ n <- length(l1) rez <- rep(TRUE,n) for(it in 1:n){ if(length(l1[[it]])!=length(l2[[it]]))rez[it] <- FALSE else if(any(l1[[it]]!=l2[[it]]))rez[it] <- FALSE } rez } equal <- rep(TRUE,3) for(it in 2:nfiles){ equal <- equal*list.eq(attributes(tables[[1]])$cutpoints,attributes(tables[[it]])$cutpoints) } kir <- which(!equal) newat <- attributes(tables[[1]]) imena <- list(d1=NULL,d2=NULL,d3=NULL) for(jt in kir){ listy <- NULL for(it in 1:nfiles){ listy <- c(listy,attributes(tables[[it]])$cutpoints[[jt]]) } imena[[jt]] <- names(table(listy)[table(listy) == nfiles]) if(!length(imena[[jt]]))stop(paste("There are no common cutpoints for dimension", attributes(tables[[1]])$dimid[jt])) } for(it in 1:nfiles){ keep <- lapply(dim(tables[[it]]),function(x)1:x) for(jt in kir){ meci <- which(match(attributes(tables[[it]])$cutpoints[[jt]],imena[[jt]],nomatch=0)!=0) if(it==1){ newat$dimnames[[jt]] <- attributes(tables[[it]])$dimnames[[jt]][meci] newat$dim[[jt]] <- length(imena[[jt]]) newat$cutpoints[[jt]] <- attributes(tables[[it]])$cutpoints[[jt]][meci] } if(length(meci)>1){if(max(diff(meci)!=1))warning(paste("The cutpoints for ",attributes(tables[[1]])$dimid[jt] ," are not equally spaced",sep=""))} keep[[jt]] <- meci } tables[[it]] <- tables[[it]][keep[[1]],keep[[2]],keep[[3]]] } dims <- newat$dim out <- array(tables[[1]],dim=c(dims,1)) for(it in 2:nfiles){ out1 <- array(tables[[it]],dim=c(dims,1)) out <- array(c(out,out1),dim=c(dims,it)) } mc <- 1:4 if(any(newat$factor>1)){ wh <- which(newat$factor>1) mc <- c(mc[-wh],wh) out <- aperm(out,mc) } newat$dim <- c(dims,nfiles)[mc] newat$dimid <- c(newat$dimid,dim.name)[mc] newat$cutpoints <- list(newat$cutpoints[[1]],newat$cutpoints[[2]],newat$cutpoints[[3]],NULL)[mc] newat$factor <- c(newat$factor,1)[mc] newat$type <- c(newat$type,1)[mc] newat$dimnames <- list(newat$dimnames[[1]],newat$dimnames[[2]],newat$dimnames[[3]],names(tables))[mc] names(newat$dimnames) <- newat$dimid attributes(out) <- newat out } mlfit <- function (b, p, x, offset, d, h, ds, y, maxiter, tol) { for (nit in 1:maxiter) { b0 <- b fd <- matrix(0, p, 1) sd <- matrix(0, p, p) if (nit == 1) { ebx <- exp(x %*% b) * exp(offset) l0 <- sum(d * log(h + ebx) - ds - y * ebx) } for (it in 1:p) { fd[it, 1] <- sum((d/(h + ebx) - y) * x[, it] * ebx) for (jt in 1:p) sd[it, jt] = sum((d/(h + ebx) - d * ebx/(h + ebx)^2 - y) * x[, it] * x[, jt] * ebx) } b <- b - solve(sd) %*% fd ebx <- exp(x %*% b) * exp(offset) l <- sum(d * log(h + ebx) - ds - y * ebx) bd <- abs(b - b0) if (max(bd) < tol) break() } out <- list(b = b, sd = sd, nit = nit, loglik = c(l0, l)) out } print.rs.br <- function (x, digits = max(options()$digits - 4, 3), ...) { invisible(print(x$table, digits = digits)) if (x$rho != 0) invisible(cat("Weighted Brownian bridge with rho=", x$rho, "\n")) } print.rsadd <- function (x, digits = max(3, getOption("digits") - 3), ...) { cat("\nCall: ", paste(deparse(x$call), sep = "\n", collapse = "\n"), "\n\n", sep = "", "\n") if (length(coef(x))) { cat("Coefficients") cat(":\n") print.default(format(x$coefficients, digits = digits), print.gap = 2, quote = FALSE) } else cat("No coefficients\n\n") if(x$method=="EM") cat("\n", "Expected number of disease specific deaths: ",format(round(sum(x$Nie),2))," = ",format(round(100*sum(x$Nie)/sum(x$data$stat),1)),"% \n" ,sep="") if(x$method=="EM"|x$method=="max.lik"){ chi <- 2*max((x$loglik[2]-x$loglik[1]),0) if(x$method=="EM")df <- length(x$coef) else df <- length(x$coef)-length(x$int)+1 if(df>0){ p.val <- 1- pchisq(chi,df) if(x$method=="max.lik")cat("\n") cat("Likelihood ratio test=",format(round(chi,2)),", on ",df," df, p=",format(p.val),"\n",sep="") } else cat("\n") } cat("n=",nrow(x$data),sep="") if(length(x$na.action))cat(" (",length(x$na.action)," observations deleted due to missing)",sep="") cat("\n") if (length(x$warnme)) cat("\n", x$warnme, "\n\n") else cat("\n") invisible(x) } summary.rsadd <- function (object, correlation = FALSE, symbolic.cor = FALSE, ...) { if (inherits(object, "glm")) { p <- object$rank if (p > 0) { p1 <- 1:p Qr <- object$qr aliased <- is.na(coef(object)) coef.p <- object$coefficients[Qr$pivot[p1]] covmat <- chol2inv(Qr$qr[p1, p1, drop = FALSE]) dimnames(covmat) <- list(names(coef.p), names(coef.p)) var.cf <- diag(covmat) s.err <- sqrt(var.cf) tvalue <- coef.p/s.err dn <- c("Estimate", "Std. Error") pvalue <- 2 * pnorm(-abs(tvalue)) coef.table <- cbind(coef.p, s.err, tvalue, pvalue) dimnames(coef.table) <- list(names(coef.p), c(dn, "z value", "Pr(>|z|)")) df.f <- NCOL(Qr$qr) } else { coef.table <- matrix(, 0, 4) dimnames(coef.table) <- list(NULL, c("Estimate", "Std. Error", "t value", "Pr(>|t|)")) covmat.unscaled <- covmat <- matrix(, 0, 0) aliased <- is.na(coef(object)) df.f <- length(aliased) } ans <- c(object[c("call", "terms", "family", "iter", "warnme")], list(coefficients = coef.table, var = covmat, aliased = aliased)) if (correlation && p > 0) { dd <- s.err ans$correlation <- covmat/outer(dd, dd) ans$symbolic.cor <- symbolic.cor } class(ans) <- "summary.rsadd" } else if (inherits(object, "rsadd")) { aliased <- is.na(coef(object)) coef.p <- object$coef var.cf <- diag(object$var) s.err <- sqrt(var.cf) tvalue <- coef.p/s.err dn <- c("Estimate", "Std. Error") pvalue <- 2 * pnorm(-abs(tvalue)) coef.table <- cbind(coef.p, s.err, tvalue, pvalue) dimnames(coef.table) <- list(names(coef.p), c(dn, "z value", "Pr(>|z|)")) ans <- c(object[c("call", "terms", "iter", "var")], list(coefficients = coef.table, aliased = aliased)) if (correlation && sum(aliased) != length(aliased)) { dd <- s.err ans$correlation <- object$var/outer(dd, dd) ans$symbolic.cor <- symbolic.cor } class(ans) <- "summary.rsadd" } else ans <- object return(ans) } print.summary.rsadd <- function (x, digits = max(3, getOption("digits") - 3), symbolic.cor = x$symbolic.cor, signif.stars = getOption("show.signif.stars"), ...) { cat("\nCall:\n") cat(paste(deparse(x$call), sep = "\n", collapse = "\n"), "\n\n", sep = "") if (length(x$aliased) == 0) { cat("\nNo Coefficients\n") } else { cat("\nCoefficients:\n") coefs <- x$coefficients if (!is.null(aliased <- x$aliased) && any(aliased)) { cn <- names(aliased) coefs <- matrix(NA, length(aliased), 4, dimnames = list(cn, colnames(coefs))) coefs[!aliased, ] <- x$coefficients } printCoefmat(coefs, digits = digits, signif.stars = signif.stars, na.print = "NA", ...) } if (length(x$warnme)) cat("\n", x$warnme, "\n") correl <- x$correlation if (!is.null(correl)) { p <- NCOL(correl) if (p > 1) { cat("\nCorrelation of Coefficients:\n") if (is.logical(symbolic.cor) && symbolic.cor) { print(symnum(correl, abbr.colnames = NULL)) } else { correl <- format(round(correl, 2), nsmall = 2, digits = digits) correl[!lower.tri(correl)] <- "" print(correl[-1, -p, drop = FALSE], quote = FALSE) } } } cat("\n") invisible(x) } epa <- function(fit,bwin,times,n.bwin=16,left=FALSE){ #bwin ... width of the window, relative to the default (1) #fit ... EM fit #times... times at which the smoothed plot is calculated #n.bwin ... number of different windows #left ... only predictable smoothing utd <- fit$times if(missing(times))times <- seq(1,max(utd),length=100) if(max(times)>max(utd)){ warning("Cannot extrapolate beyond max event time") times <- pmax(times,max(utd)) } nutd <- length(utd) nt4 <- c(1,ceiling(nutd*(1:n.bwin)/n.bwin)) if(missing(bwin))bwin <- rep(length(fit$times)/100,n.bwin) else bwin <- rep(bwin*length(fit$times)/100,n.bwin) for(it in 1:n.bwin){ bwin[it] <- bwin[it]*max(diff(utd[nt4[it]:nt4[it+1]])) } while(utd[nt4[2]]tvs[nt4[it]] & t<=tvs[nt4[it + 1]]] if(length(cajti)){ q <- min( cajti/b[it],1,(Rb-cajti)/b[it]) if(q<1 & length(cajti)>1){ jc <- 1 while(jc <=length(cajti)){ qd <- pmin( cajti[jc:length(cajti)]/b[it],1,(Rb-cajti[jc:length(cajti)])/b[it]) q <- qd[1] if(q==1){ casi <- cajti[jc:length(cajti)][qd==1] q <- 1 jc <- sum(qd==1)+jc } else{ casi <- cajti[jc] jc <- jc+1 } kmat1 <- outer(casi, tv, "-")/b[it] #z - to je ok if(q<1){ if(casi>b[it]) kmt1 <- -kmat1 vr <- kt(q,kmat1)*(kmat1>=-1 & kmat1 <= q) } else vr <- pmax((1 - kmat1^2) * .75,0) kmat <- rbind(kmat, vr/b[it]) totcajti <- c(totcajti,casi) } } else{ kmat1 <- outer(cajti, tv, "-")/b[it] #z - to je ok q <- min( cajti/b[it],1) if(q<1)vr <- kt(q,kmat1)*(kmat1>=-1 & kmat1 <= q) else vr <- pmax((1 - kmat1^2) * .75,0) kmat <- rbind(kmat, vr/b[it]) totcajti <- c(totcajti,cajti) }#else }#if }#for kmat } kern <- function (times,td, b, nt4) { n <- length(td) ttemp <- td[td >= b[1]] ntemp <- length(ttemp) if (ntemp == n) nt4 <- c(0, nt4[-1]) td <- c(0,td) nt4 <- c(1,nt4+1) b <- c(b[1],b) krn <- Kern(times, td, b, max(td), nt4) krn } exp.prep <- function (x, y,ratetable,status,times,fast=FALSE,ys,prec,cmp=F) { #function that prepares the data for C function call #x= matrix of demographic covariates - each individual has one line #y= follow-up time for each individual (same length as nrow(x)!) #ratetable= rate table used for calculation #status= status for each individual (same length as nrow(x)!), not needed if we only need Spi, status needed for rs.surv #times= times at which we wish to evaluate the quantities, not needed if we only need Spi, times needed for rs.surv #fast=for mpp method only x <- as.matrix(x) if (ncol(x) != length(dim(ratetable))) stop("x matrix does not match the rate table") atts <- attributes(ratetable) cuts <- atts$cutpoints if (is.null(atts$type)) { rfac <- atts$factor us.special <- (rfac > 1) } else { rfac <- 1 * (atts$type == 1) us.special <- (atts$type == 4) } if (length(rfac) != ncol(x)) stop("Wrong length for rfac") if (any(us.special)) { if (sum(us.special) > 1) stop("Two columns marked for special handling as a US rate table") cols <- match(c("age", "year"), atts$dimid) if (any(is.na(cols))) stop("Ratetable does not have expected shape") if (exists("as.Date")) { bdate <- as.Date("1960/1/1") + (x[, cols[2]] - x[, cols[1]]) byear <- format(bdate, "%Y") offset <- as.numeric(bdate - as.Date(paste(byear, "01/01", sep = "/"))) } else if (exists("date.mdy")) { bdate <- as.date(x[, cols[2]] - x[, cols[1]]) byear <- date.mdy(bdate)$year offset <- bdate - mdy.date(1, 1, byear) } else stop("Can't find an appropriate date class\n") x[, cols[2]] <- x[, cols[2]] - offset if (any(rfac > 1)) { temp <- which(us.special) nyear <- length(cuts[[temp]]) nint <- rfac[temp] cuts[[temp]] <- round(approx(nint * (1:nyear), cuts[[temp]], nint:(nint * nyear))$y - 1e-04) } } if(!missing(status)){ #the function was called from rs.surv if(length(status)!=nrow(x)) stop("Wrong length for status") if(missing(times)) times <- sort(unique(y)) if (any(times < 0)) stop("Negative time point requested") ntime <- length(times) if(missing(ys)) ys <- rep(0,length(y)) # times2 <- times # times2[1] <- preci if(cmp) temp <- .Call("cmpfast", as.integer(rfac), #fast=pohar-perme or ederer2 - data from pop. tables only while under follow-up as.integer(atts$dim), as.double(unlist(cuts)), ratetable, x, y, ys,as.integer(status), times,PACKAGE="relsurv") else if(fast&!missing(prec)) temp <- .Call("netfastpinter2", as.integer(rfac), #fast=pohar-perme or ederer2 - data from pop. tables only while under follow-up as.integer(atts$dim), as.double(unlist(cuts)), ratetable, x, y, ys,as.integer(status), times,prec,PACKAGE="relsurv") else if(fast&missing(prec)) temp <- .Call("netfastpinter", as.integer(rfac), #fast=pohar-perme or ederer2 - data from pop. tables only while under follow-up as.integer(atts$dim), as.double(unlist(cuts)), ratetable, x, y, ys,as.integer(status), times,PACKAGE="relsurv") else temp <- .Call("netwei", as.integer(rfac), as.integer(atts$dim), as.double(unlist(cuts)), ratetable, x, y, as.integer(status), times,PACKAGE="relsurv") } else{ #only expected survival at time y is needed for each individual if(length(y)==1)y <- rep(y,nrow(x)) if(length(y)!=nrow(x)) stop("Wrong length for status") temp <- .Call("expc", as.integer(rfac), as.integer(atts$dim), as.double(unlist(cuts)), ratetable, x, y,PACKAGE="relsurv") temp <- temp$surv } temp } rs.surv <- function (formula = formula(data), data = parent.frame(),ratetable = relsurv::slopop, na.action, fin.date, method = "pohar-perme", conf.type = "log", conf.int = 0.95,type="kaplan-meier",add.times,precision=1,rmap) #formula: for example Surv(time,cens)~sex #data: the observed data set #ratetable: the population mortality tables #conf.type: confidence interval calculation (plain, log or log-log) #conf.int: confidence interval { call <- match.call() if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula,data, ratetable, na.action,rmap) data <- rform$data #the data set type <- match.arg(type, c("kaplan-meier", "fleming-harrington")) #method of hazard -> survival scale transformation type <- match(type, c("kaplan-meier", "fleming-harrington")) method <- match.arg(method,c("pohar-perme", "ederer2", "hakulinen","ederer1")) #method of relative surv. curve estimation method <- match(method,c("pohar-perme", "ederer2", "hakulinen","ederer1")) conf.type <- match.arg(conf.type,c("plain","log","log-log")) #conf. interval type if (method == 3) { #need potential follow-up time for Hak. method R <- rform$R coll <- match("year", attributes(ratetable)$dimid) year <- R[, coll] #calendar year in the data if (missing(fin.date)) fin.date <- max(rform$Y + year) #final date for everybody set to the last day observed Y2 <- rform$Y #change into potential follow-up time if (length(fin.date) == 1) #if final date equal for everyone Y2[rform$status == 1] <- fin.date - year[rform$status == 1]#set pot.time for those that died (equal to censoring time for others) else if (length(fin.date) == nrow(rform$R)) Y2[rform$status == 1] <- fin.date[rform$status == 1] - year[rform$status == 1] else stop("fin.date must be either one value or a vector of the same length as the data") status2 <- rep(0, nrow(rform$X)) #stat2=0 for everyone } p <- rform$m #number of covariates if (p > 0) #if covariates data$Xs <- strata(rform$X[, ,drop=FALSE ]) #make strata according to covariates else data$Xs <- rep(1, nrow(data)) #if no covariates, just put 1 se.fac <- sqrt(qchisq(conf.int, 1)) #factor needed for confidence interval out <- NULL out$n <- table(data$Xs) #table of strata out$time <- out$n.risk <- out$n.event <- out$n.censor <- out$surv <- out$std.err <- out$strata <- NULL #out$index <- out$strata0 <- NULL # out$index = indices of the original times from the data among the times used for calculations # out$strata0 = the same as out$strata but only on the original times from the data for (kt in 1:length(out$n)) { #for each stratum inx <- which(data$Xs == names(out$n)[kt]) #individuals within this stratum tis <- sort(unique(rform$Y[inx])) #unique times #if (method == 1 & all.times == TRUE) tis <- sort(union(rform$Y[inx],as.numeric(1:max(floor(rform$Y[inx]))))) #1-day long intervals used - to take into the account the continuity of the pop. part if (method == 1 & !missing(add.times)){ #tis <- sort(union(rform$Y[inx],as.numeric(1:max(floor(rform$Y[inx]))))) #1-day long intervals used - to take into the account the continuity of the pop. part add.times <- pmin(as.numeric(add.times),max(rform$Y[inx])) tis <- sort(union(rform$Y[inx],as.numeric(add.times))) #1-day long intervals used - to take into the account the continuity of the pop. part } if(method==3)tis <- sort(unique(pmin(max(tis),c(tis,Y2[inx])))) #add potential times in case of Hakulinen #out$index <- c(out$index, which(tis %in% rform$Y[inx])+length(out$time)) temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=tis,fast=(method<3),prec=precision) #calculate the values for each interval of time out$time <- c(out$time, tis) #add times out$n.risk <- c(out$n.risk, temp$yi) #add number at risk for each time out$n.event <- c(out$n.event, temp$dni) #add number of events for each time out$n.censor <- c(out$n.censor, c(-diff(temp$yi),temp$yi[length(temp$yi)]) - temp$dni) #add number of censored for each time if(method==1){ #pohar perme method #approximate1 <- (temp$yidlisi/temp$yisi +temp$yidlisitt/temp$yisitt)/2 #approximate <- (temp$yidlisiw/temp$yisi +temp$yidlisiw/temp$yisitt)/2 #approximation for integration approximate <- temp$yidlisiw #haz <- temp$dnisi/temp$yisi - temp$yidlisi/temp$yisi #cumulative hazard increment on each interval haz <- temp$dnisi/temp$yisi - approximate #cumulative hazard increment on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dnisisq/(temp$yisi)^2))) #standard error on each interval } else if(method==2){ #ederer2 method haz <- temp$dni/temp$yi - temp$yidli/temp$yi #cumulative hazard increment on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } else if(method==3){ #Hakulinen method temp2 <- exp.prep(rform$R[inx,,drop=FALSE],Y2[inx],ratetable,status2[inx],times=tis) #calculate the values for each interval of time popsur <- exp(-cumsum(temp2$yisidli/temp2$yisis)) #population survival haz <- temp$dni/temp$yi #observed hazard on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } else if(method==4){ #Ederer I popsur <- temp$sis/length(inx) #population survival haz <- temp$dni/temp$yi #observed hazard on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } if(type==2)survtemp <- exp(-cumsum(haz)) else survtemp <- cumprod(1-haz) if(method>2){ survtemp <- survtemp/popsur } out$surv <- c(out$surv,survtemp) out$strata <- c(out$strata, length(tis)) #number of times in this strata #out$strata0 <- c(out$strata0, length(unique(rform$Y[inx]))) } if (conf.type == "plain") { out$lower <- as.vector(out$surv - out$std.err * se.fac * #surv + fac*se out$surv) out$upper <- as.vector(out$surv + out$std.err * se.fac * out$surv) } else if (conf.type == "log") { #on log scale and back out$lower <- exp(as.vector(log(out$surv) - out$std.err * se.fac)) out$upper <- exp(as.vector(log(out$surv) + out$std.err * se.fac)) } else if (conf.type == "log-log") { #on log-log scale and back out$lower <- exp(-exp(as.vector(log(-log(out$surv)) - out$std.err * se.fac/log(out$surv)))) out$upper <- exp(-exp(as.vector(log(-log(out$surv)) + out$std.err * se.fac/log(out$surv)))) } names(out$strata) <- names(out$n) #names(out$strata0) <- names(out$n) if (p == 0){ out$strata <- NULL #if no covariates #out$strata0 <- NULL } #if (method != 1) out$index <- out$strata0 <- NULL # if method != pohar-perme out$n <- as.vector(out$n) out$conf.type <- conf.type out$conf.int <- conf.int out$method <- method out$call <- call out$type <- "right" class(out) <- c("survfit", "rs.surv") out } nessie <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop,times,rmap) #formula: for example Surv(time,cens)~sex #data: the observed data set #ratetable: the population mortality tables #times: the times at which to report NESS, if no default, then all unique times { call <- match.call() if (!missing(rmap)) { rmap <- substitute(rmap) } na.action <- NA #set the object just to be able to execute the rformulate call rform <- rformulate(formula, data, ratetable,na.action, rmap) #get the data ready templab <- attr(rform$Terms,"term.labels") if(!is.null(attr(rform$Terms,"specials")$ratetable))templab <- templab[-length(templab)] #delete the last term in the formula if the ratetable argument is there nameslist <- vector("list",length(templab)) for(it in 1:length(nameslist)){ valuetab <- table(data[,match(templab[it],names(data))]) nameslist[[it]] <- paste(templab[it],names(valuetab),sep="") } names(nameslist) <- templab data <- rform$data #the data set p <- rform$m #number of covariates if (p > 0) { #if covariates data$Xs <- my.strata(rform$X[,,drop=F],nameslist=nameslist) #make strata according to covariates #data$Xs <- factor(data$Xs,levels=nameslist) #order them in the same way as namelist } else data$Xs <- rep(1, nrow(data)) #if no covariates, just put 1 if(!missing(times)) tis <- times else tis <- unique(sort(floor(rform$Y/365.241))) #unique years of follow-up tis <- unique(c(0,tis)) tisd <- tis*365.241 out <- NULL out$n <- table(data$Xs) #table of strata out$sp <- out$strata <- NULL # for (kt in 1:length(out$n)) { #for each stratum for (kt in order(names(table(data$Xs)))) { #for each stratum inx <- which(data$Xs == names(out$n)[kt]) #individuals within this stratum temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=tisd,fast=FALSE) #calculate the values for each interval of time out$time <- c(out$time, tisd) #add times out$sp <- c(out$sp, temp$sis) #add expected number of individuals alive out$strata <- c(out$strata, length(tis)) #number of times in this strata temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=(seq(0,100,by=.5)*365.241)[-1],fast=FALSE) #calculate the values for each interval of time out$povp <- c(out$povp,mean(temp$sit/365.241)) } names(out$strata) <- names(out$n)[order(names(table(data$Xs)))] if (p == 0) out$strata <- NULL #if no covariates mata <- matrix(out$sp,ncol=length(tis),byrow=TRUE) mata <- data.frame(mata) mata <- cbind(mata,out$povp) row.names(mata) <- names(out$n)[order(names(table(data$Xs)))] names(mata) <- c(tis,"c.exp.surv") cat("\n") print(round(mata,1)) cat("\n") out$mata <- mata out$n <- as.vector(out$n) class(out) <- "nessie" invisible(out) } rs.period <- function (formula = formula(data), data = parent.frame(), ratetable = relsurv::slopop, na.action, fin.date, method = "pohar-perme", conf.type = "log", conf.int = 0.95,type="kaplan-meier",winst,winfin,diag.date,rmap) #formula: for example Surv(time,cens)~sex #data: the observed data set #ratetable: the population mortality tables #conf.type: confidence interval calculation (plain, log or log-log) #conf.int: confidence interval #winst: start of the period window (inclusive) #winfin: end of the period window (inclusive) { call <- match.call() if (!missing(rmap)) { rmap <- substitute(rmap) } rform <- rformulate(formula, data, ratetable, na.action,rmap) #get the data ready data <- rform$data #the data set type <- match.arg(type, c("kaplan-meier", "fleming-harrington")) #method of hazard -> survival scale transformation type <- match(type, c("kaplan-meier", "fleming-harrington")) method <- match.arg(method,c("pohar-perme", "ederer2", "hakulinen","ederer1")) #method of relative surv. curve estimation method <- match(method,c("pohar-perme", "ederer2", "hakulinen","ederer1")) conf.type <- match.arg(conf.type,c("plain","log","log-log")) #conf. interval type #machinations needed for period survival: R <- rform$R coll <- match("year", attributes(ratetable)$dimid) year <- R[, coll] #calendar year in the data ys <- as.numeric(winst - year) yf <- as.numeric(winfin - year) relv <- which(ys <= rform$Y & yf>0) #relevant individuals -> live up to the period window and were diagnosed before window end centhem <- which(yf < rform$Y) #censor these - their event happens outside of the period window rform$status[centhem] <- 0 rform$Y[centhem] <- yf[centhem] rform$Y <- rform$Y[relv] rform$X <- rform$X[relv,,drop=F] rform$R <- rform$R[relv,,drop=F] rform$status <- rform$status[relv] data <- data[relv,,drop=F] ys <- ys[relv] yf <- yf[relv] year <- year[relv] if (method == 3) { #need potential follow-up time for Hak. method if (missing(fin.date)) fin.date <- max(rform$Y + year) #final date for everybody set to the last day observed Y2 <- rform$Y #change into potential follow-up time if (length(fin.date) == 1) #if final date equal for everyone Y2[rform$status == 1] <- fin.date - year[rform$status == 1]#set pot.time for those that died (equal to censoring time for others) else if (length(fin.date[relv]) == nrow(rform$R)) { fin.date <- fin.date[relv] Y2[rform$status == 1] <- fin.date[rform$status == 1] - year[rform$status == 1] } else stop("fin.date must be either one value of a vector of the same length as the data") status2 <- rep(0, nrow(rform$X)) #stat2=0 for everyone } p <- rform$m #number of covariates if (p > 0) #if covariates data$Xs <- strata(rform$X[, ,drop=FALSE ]) #make strata according to covariates else data$Xs <- rep(1, nrow(data)) #if no covariates, just put 1 se.fac <- sqrt(qchisq(conf.int, 1)) #factor needed for confidence interval out <- NULL out$n <- table(data$Xs) #table of strata out$time <- out$n.risk <- out$n.event <- out$n.censor <- out$surv <- out$std.err <- out$strata <- NULL for (kt in 1:length(out$n)) { #for each stratum inx <- which(data$Xs == names(out$n)[kt]) #individuals within this stratum tis <- sort(unique(rform$Y[inx])) #unique times if(method==3)tis <- sort(unique(pmin(max(tis),c(tis,Y2[inx])))) #add potential times in case of Hakulinen ys <- pmax(ys,0) #tis <- sort(unique(c(tis,ys[ys>0]-1,ys[ys>0]))) tis <- sort(unique(c(tis,ys[ys>0]))) tis <- sort(unique(c(tis,tis-1,tis+1))) #the day after exiting, the day before entering tis <- tis[-length(tis)] #exclude the largest since it is beyond observation time (1 day later) temp <- exp.prep(rform$R[inx,,drop=FALSE],rform$Y[inx],rform$ratetable,rform$status[inx],times=tis,fast=(method<3),ys=ys) #calculate the values for each interval of time out$time <- c(out$time, tis) #add times out$n.risk <- c(out$n.risk, temp$yi) #add number at risk for each time out$n.event <- c(out$n.event, temp$dni) #add number of events for each time out$n.censor <- c(out$n.censor, c(-diff(temp$yi),temp$yi[length(temp$yi)]) - temp$dni) #add number of censored for each time if(method==1){ #pohar perme method haz <- temp$dnisi/temp$yisi - temp$yidlisi/temp$yisi #cumulative hazard increment on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dnisisq/(temp$yisi)^2))) #standard error on each interval } else if(method==2){ #ederer2 method haz <- temp$dni/temp$yi - temp$yidli/temp$yi #cumulative hazard increment on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } else if(method==3){ #Hakulinen method temp2 <- exp.prep(rform$R[inx,,drop=FALSE],Y2[inx],rform$ratetable,status2[inx],times=tis,ys=ys) #calculate the values for each interval of time popsur <- exp(-cumsum(temp2$yisidli/temp2$yisis)) #population survival haz <- temp$dni/temp$yi #observed hazard on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } else if(method==4){ #Ederer I popsur <- temp$sis/length(inx) #population survival haz <- temp$dni/temp$yi #observed hazard on each interval out$std.err <- c(out$std.err, sqrt(cumsum(temp$dni/(temp$yi)^2))) #standard error on each interval } if(type==2)survtemp <- exp(-cumsum(haz)) else survtemp <- cumprod(1-haz) if(method>2){ survtemp <- survtemp/popsur } out$surv <- c(out$surv,survtemp) out$strata <- c(out$strata, length(tis)) #number of times in this strata } if (conf.type == "plain") { out$lower <- as.vector(out$surv - out$std.err * se.fac * #surv + fac*se out$surv) out$upper <- as.vector(out$surv + out$std.err * se.fac * out$surv) } else if (conf.type == "log") { #on log scale and back out$lower <- exp(as.vector(log(out$surv) - out$std.err * se.fac)) out$upper <- exp(as.vector(log(out$surv) + out$std.err * se.fac)) } else if (conf.type == "log-log") { #on log-log scale and back out$lower <- exp(-exp(as.vector(log(-log(out$surv)) - out$std.err * se.fac/log(out$surv)))) out$upper <- exp(-exp(as.vector(log(-log(out$surv)) + out$std.err * se.fac/log(out$surv)))) } names(out$strata) <- names(out$n) if (p == 0) out$strata <- NULL #if no covariates out$n <- as.vector(out$n) out$conf.type <- conf.type out$conf.int <- conf.int out$method <- method out$call <- call out$type <- "right" class(out) <- c("survfit", "rs.surv") out } relsurv/MD50000644000176200001440000000474613400227643012300 0ustar liggesusers3d4d54a0f11c27771250aa45682378d2 *DESCRIPTION 89b7b40368351c34e6f1c543ad5902a3 *NAMESPACE 4486ca7c2174692c2a3036e5deca59c6 *R/Rcode.r a9af69aa6f278a631bb39a3593358ece *R/cmprel.r 476d5effd3aff473f5ce65717dd4551a *R/mystrata.r e8c5220a2b671c39c0be09d5991f5aae *R/plotrssurv.r a68b2a2e02b73cfbbd7a99710a01f69a *R/rformulate.r 87179b85428844d4add5e733cae955d6 *R/rsdiff.r ceca1cde417d54a95d3e057ffda85405 *R/rssurvrsadd.r b4637a419c60f04bf206fc517dcc3cd2 *R/survfitrsadd.r 3820354bc630edaa5d29f0225798b4da *R/zzz.R 3d33b58409d59c1f05de22766960f039 *data/colrec.rda da512257141fedce138de599b5e9997a *data/rdata.rda 6590e14f71cc425e679350e43cb6ee53 *data/slopop.rda 538a7487bf697ab8b5e9038679ba76c9 *inst/CITATION b083e2e24c6734aca25cc23a8f4eee40 *inst/news.Rd 809ff6dfcec2e0c759b3b364980c24ee *man/cmprel.Rd e9e43bae88190dd04b902a6e3e4e702c *man/colrec.Rd 9107f531f6c42fb0b11f6f607d2ca0ca *man/epa.Rd cd16fc0741fa54b6ce43293678704531 *man/invtime.Rd 185f03cd09bda2acfecbde763fce4209 *man/joinrate.Rd 612ee95ccbd481320a22c018d52675cb *man/nessie.Rd 156dbf848608b8af7cf409bef4924044 *man/plotcmprel.Rd 795d3ace0f78a662461dd2007bf818df *man/plotrszph.Rd c28a88ecfa1f891e98c8fc9ef7566e55 *man/rdata.Rd 1c651d6afbdea7eaaffd550251fe595e *man/resid.Rd 64a1f206fc6690cc3b0b3a0f3466b860 *man/rsadd.Rd da1c79e8bf94acf641eee22e9d81f007 *man/rsbr.Rd cede103aa3b59c0917961e3272ed1e76 *man/rsdiff.Rd 45932d98be554ad05806c9074aae80fd *man/rsmul.Rd d5f87476191d46898da890c023802909 *man/rssurv.Rd b6a1d43f1e15891a82c35ddc9dc5fec7 *man/rssurvrsadd.Rd 3f5027f1683a5d943c101e89a9320eac *man/rstrans.Rd 676f6bc1797b7082598e96fb488ee2e6 *man/rszph.Rd be655c8edf2364de4cc91c03c6e61487 *man/slopop.Rd b4255e5550219ff1c8213cd714370b70 *man/summary.cmp.rel.Rd 4ec64e929b076aed0e24384a9febd263 *man/survfitrsadd.rd 43de679849893f7f8847efb130cd7af5 *man/survsplit.Rd 713ee65969b8de3669824feed81846cf *man/transrate.Rd ce2365d59b077324be90c798f79379de *man/transratehld.Rd ade1cfda41d4ceb8f5ec749753a9ef78 *man/transratehmd.Rd 3165b274aef1ec5d5313b1c357ff23c7 *src/cmpfast.c 6853ad4d02cc6b1ff9e2e786f7dad4b5 *src/dmatrix.c 35fe86cf308d11c704de3aaf3e58a629 *src/exps.c 610b2b35cbcc528d6263bb1dbe7f856f *src/init.c f1508f17270fdc2fef10788b7ca034be *src/netfastp.c 835a0c78fa3f88bf27b963fee8867c3c *src/netfastpinter.c 9da637dbb10add8b072da100f842ecf4 *src/netfastpinter2.c f5c736b5d98722344c3a72e32a424959 *src/netwei.c 2d85cafe474de3c4d9cb2df04eafb8bb *src/pystep.c cf7cd04af54914c338bf99663db7e2d2 *src/pystep2.c fffb5d75a0a72415aa59db882ea7933f *src/survprotomoj.h relsurv/DESCRIPTION0000644000176200001440000000146713400227643013473 0ustar liggesusersPackage: relsurv Title: Relative Survival Date: 2018-11-28 Version: 2.2-3 Authors@R: c(person(c("Maja","Pohar","Perme"),role=c("aut","cre"),email="maja.pohar@mf.uni-lj.si")) Author: Maja Pohar Perme [aut, cre] Maintainer: Maja Pohar Perme Description: Contains functions for analysing relative survival data, including nonparametric estimators of net (marginal relative) survival, relative survival ratio, crude mortality, methods for fitting and checking additive and multiplicative regression models, transformation approach, methods for dealing with population mortality tables. Depends: R (>= 2.10), survival (>= 2.42), date Imports: splines License: GPL LazyData: true NeedsCompilation: yes Repository: CRAN Packaged: 2018-11-30 11:49:12 UTC; majap Date/Publication: 2018-11-30 12:40:03 UTC relsurv/man/0000755000176200001440000000000013400221667012530 5ustar liggesusersrelsurv/man/survsplit.Rd0000644000176200001440000000257610221543070015075 0ustar liggesusers\name{survsplit} \alias{survsplit} \title{Split a Survival Data Set at Specified Times} \description{ Given a survival data set and a set of specified cut times, the function splits each record into multiple records at each cut time. The new data set is be in \code{counting process} format, with a start time, stop time, and event status for each record. More general than \code{survSplit} as it also works with the data already in the \code{counting process} format. } \usage{ survsplit(data, cut, end, event, start, id = NULL, zero = 0, episode = NULL,interval=NULL) } \arguments{ \item{data}{data frame. } \item{cut}{vector of timepoints to cut at.} \item{end}{character string with name of event time variable. } \item{event}{character string with name of censoring indicator. } \item{start}{character string with name of start variable (will be created if it does not exist). } \item{id}{character string with name of new id variable to create (optional). } \item{zero}{If \code{start} doesn't already exist, this is the time that the original records start. May be a vector or single value. } \item{episode}{character string with name of new episode variable (optional).} \item{interval}{this argument is used by \code{max.lik} function} } \value{New, longer, data frame.} \seealso{\code{\link{survSplit}}.} \keyword{survival} relsurv/man/epa.Rd0000644000176200001440000000414313332517176013575 0ustar liggesusers\name{epa} \alias{epa} \title{Excess hazard function smoothing} \description{ An Epanechnikov kernel function based smoother for smoothing the baseline excess hazard calculated by the \code{rsadd} function with the \code{EM} method. } \usage{ epa(fit,bwin,times,n.bwin=16,left=FALSE) } \arguments{ \item{fit}{ Fit from the additive relative survival model using the \code{EM} method.} \item{bwin}{ The relative width of the smoothing window (default is 1). } \item{times}{ The times at which the smoother is to be evaluated. If missing, it is evaluated at all event times. } \item{n.bwin}{ Number of times that the window width may change. } \item{left}{If \code{FALSE} (default) smoothing is performed symmetrically, if \code{TRUE} only leftside neighbours are considered. } } \details{ The function performs Epanechnikov kernel smoothing. The follow up time is divided (according to percentiles of event times) into several intervals (number of intervals defined by \code{n.bwin}) in which the width is calculated as a factor of the maximum span between event times. Boundary effects are also taken into account on both sides. } \value{ A list with two components: \item{lambda}{the smoothed excess baseline hazard function} \item{times}{the times at which the smoothed excess baseline hazard is evaluated.} } \examples{ data(slopop) data(rdata) #fit an additive model with the EM method fit <- rsadd(Surv(time,cens)~sex+age,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,int=5,method="EM") sm <- epa(fit) plot(sm$times,sm$lambda) } \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. EM algorithm: Pohar Perme M., Henderson R., Stare, J. (2009) "An approach to estimation in relative survival regression." Biostatistics, \bold{10}: 136--146. } \seealso{ \code{\link{rsadd}}, } \keyword{survival} relsurv/man/survfitrsadd.rd0000644000176200001440000000536513332561443015613 0ustar liggesusers\name{survfit.rsadd} \alias{survfit.rsadd} \title{Compute a Predicited Survival Curve} \description{ Computes a predicted survival curve based on the additive model estimated by rsadd function. } \usage{ \method{survfit}{rsadd}(formula, newdata, se.fit = TRUE, conf.int = 0.95, individual = FALSE, conf.type = c("log", "log-log", "plain", "none"),...) } \arguments{ \item{formula}{a rsadd object} \item{newdata}{a data frame with the same variable names as those that appear in the rsadd formula. The curve(s) produced will be representative of a cohort who's covariates correspond to the values in newdata.} \item{se.fit}{a logical value indicating whether standard errors should be computed. Default is \code{TRUE}.} \item{conf.int}{the level for a two-sided confidence interval on the survival curve(s). Default is 0.95.} \item{individual}{a logical value indicating whether the data frame represents different time epochs for only one individual (T), or whether multiple rows indicate multiple individuals (F, the default). If the former only one curve will be produced; if the latter there will be one curve per row in newdata.} \item{conf.type}{One of \code{none}, \code{plain}, \code{log} (the default), or \code{log-log}. The first option causes confidence intervals not to be generated. The second causes the standard intervals curve +- k *se(curve), where k is determined from conf.int. The log option calculates intervals based on the cumulative hazard or log(survival). The last option bases intervals on the log hazard or log(-log(survival)). } \item{...}{Currently not implemented} } \details{ When predicting the survival curve, the ratetable values for future years will be equal to those of the last given year. The same ratetables will be used for fitting and predicting. To predict a relative survival curve, use \code{rs.surv.rsadd}. } \value{ a \code{survfit} object; see the help on \code{survfit.object} for details. The \code{survfit} methods are used for \code{print}, \code{plot}, \code{lines}, and \code{points}. } \references{ Package: Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine,\bold{81}: 272--278. Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \examples{ data(slopop) data(rdata) #BTW: work on a smaller dataset here to run the example faster fit <- rsadd(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata[1:500,],method="EM") survfit.rsadd(fit,newdata=data.frame(sex=1,age=60,year=17000)) } \seealso{ \code{survfit}, \code{survexp}, \code{\link{rs.surv}} } \keyword{survival} relsurv/man/invtime.Rd0000644000176200001440000000413713332256161014500 0ustar liggesusers\name{invtime} \alias{invtime} \title{Inverse transforming of time in Relative Survival } \description{ This function can be used when predicting in Relative Survival using the transformed time regression model (using \code{rstrans} function). It inverses the time from Y to T in relative survival using the given ratetable. The times Y can be produced with the \code{rstrans} function, in which case, this is the reverse function. This function does the transformation for one person at a time. } \details{ Works only with ratetables that are split by age, sex and year. Transforming can be computationally intensive, use lower and/or upper to guess the interval of the result and thus speed up the function. } \usage{ invtime(y, age, sex, year, scale, ratetable, lower, upper) } \arguments{ \item{y}{time in Y.} \item{age}{age of the individual. Must be in days.} \item{sex}{sex of the individual. Must be coded in the same way as in the \code{ratetable}.} \item{year}{date of diagnosis. Must be in a date format} \item{scale}{ numeric value to scale the results. If \code{ratetable} is in units/day, \code{scale = 365.241} causes the output to be reported in years. } \item{ratetable}{a table of event rates, such as \code{survexp.us}. } \item{lower}{the lower bound of interval where the result is expected. This argument is optional, but, if given, can shorten the time the function needs to calculate the result. } \item{upper}{ the upper bound of interval where the result is expected. See \code{lower} } } \value{ A list of values \item{T}{the original time} \item{Y}{the transformed time} } \references{ Package: Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272-278. Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741-1749. } \examples{ data(slopop) invtime(y = 0.1, age = 23011, sex = 1, year = 9497, ratetable = slopop) } \seealso{ \code{\link{rstrans}} } \keyword{ survival } relsurv/man/transrate.Rd0000644000176200001440000000307612705402360015026 0ustar liggesusers\name{transrate} \alias{transrate} \title{Reorganize Data into a Ratetable Object} \description{ The function assists in reorganizing certain types of data into a ratetable object. } \usage{ transrate(men,women,yearlim,int.length=1) } \arguments{ \item{men}{ a matrix containing the yearly (conditional) probabilities of one year survival for men. Rows represent age (increasing 1 year per line,starting with 0), the columns represent cohort years (the limits are in \code{yearlim}, the increase is in \code{int.length}. } \item{women}{ a matrix containing the yearly (conditional) probabilities of one year survival for women. } \item{yearlim}{the first and last cohort year given in the tables.} \item{int.length}{the length of intervals in which cohort years are given.} } \details{ This function only applies for ratetables that are organized by age, sex and year. } \value{An object of class \code{ratetable}.} \examples{ men <- cbind(exp(-365.241*exp(-14.5+.08*(0:100))),exp(-365*exp(-14.7+.085*(0:100)))) women <- cbind(exp(-365.241*exp(-15.5+.085*(0:100))),exp(-365*exp(-15.7+.09*(0:100)))) table <- transrate(men,women,yearlim=c(1980,1990),int.length=10) } \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \seealso{\code{\link{ratetable}}.} \keyword{survival} relsurv/man/transratehmd.Rd0000644000176200001440000000315411536726416015530 0ustar liggesusers\name{transrate.hmd} \alias{transrate.hmd} \title{Reorganize Data obtained from Human Mortality Database into a Ratetable Object} \description{ The function assists in reorganizing the .txt files obtained from Human Mortality Database (http://www.mortality.org) into a ratetable object. } \usage{ transrate.hmd(male,female) } \arguments{ \item{male}{ a .txt file, containing the data on males. } \item{female}{ a .txt file, containing the data on females. } } \details{ This function works automatically with tables organised in the format provided by the Human Mortality Database. Download Life Tables for Males and Females separately from the column named 1x1 (period life tables, organized by date of death, yearly cutpoints for age as well as calendar year). If you wish to provide the data in the required format by yourself, note that the only two columns needed are calendar year (Year) and probability of death (qx). Death probabilities must be calculated up to age 110 (in yearly intervals). } \value{An object of class \code{ratetable}.} \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \examples{ \dontrun{ auspop <- transrate.hmd("mltper_1x1.txt","fltper_1x1.txt") } } \seealso{\code{\link{ratetable}}, \code{\link{transrate.hld}}, \code{\link{joinrate}}, \code{\link{transrate}}.} \keyword{survival} relsurv/man/resid.Rd0000644000176200001440000000325513332504453014133 0ustar liggesusers\name{residuals.rsadd} \alias{residuals.rsadd} \title{Calculate Residuals for a "rsadd" Fit} \description{ Calculates partial residuals for an additive relative survival model. } \usage{ \method{residuals}{rsadd}(object,type="schoenfeld",...) } \arguments{ \item{object}{ an object inheriting from class \code{rsadd}, representing a fitted additive relative survival model. Typically this is the output from the \code{rsadd} function. } \item{type}{ character string indicating the type of residual desired. Currently only Schoenfeld residuals are implemented. } \item{...}{other arguments.} } \value{ A list of the following values is returned: \item{res}{a matrix containing the residuals for each variable.} \item{varr}{the variance for each residual} \item{varr1}{the sum of \code{varr}.} \item{kvarr}{the derivative of each residual, to be used in \code{rs.zph} function.} \item{kvarr1}{the sum of \code{kvarr}.} } \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. Goodness of fit: Stare J.,Pohar Perme M., Henderson R. (2005) "Goodness of fit of relative survival models." Statistics in Medicine, \bold{24}: 3911--3925. } \examples{ data(slopop) data(rdata) fit <- rsadd(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,int=5) sresid <- residuals.rsadd(fit) } \seealso{\code{\link{rsadd}}.} \keyword{survival} relsurv/man/cmprel.Rd0000644000176200001440000001117313400170766014307 0ustar liggesusers\name{cmp.rel} \alias{cmp.rel} \alias{print.cmp.rel} \title{Compute crude probability of death} \description{ Estimates the crude probability of death due to disease and due to population reasons } \usage{ cmp.rel(formula, data, ratetable = relsurv::slopop, na.action,tau, conf.int=0.95,precision=1,add.times,rmap) } \arguments{ \item{formula}{ a formula object, with the response as a \code{Surv} object on the left of a \code{~} operator, and, if desired, terms separated by the \code{+} operator on the right. If no strata are used, \code{~1} should be specified. NOTE: The follow-up time must be in days. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, organized as a \code{ratetable} object, such as \code{slopop}. } \item{na.action}{a missing-data filter function, applied to the model.frame, after any subset argument has been used. Default is \code{options()$na.action}.} \item{tau}{the maximum follow-up time of interest, all times larger than \code{tau} shall be censored. Equals maximum observed time by default } \item{conf.int}{the level for a two-sided confidence interval on the survival curve(s). Default is 0.95.} \item{precision}{the level of precision used in the numerical integration of variance. Default is 1, which means that daily intervals are taken, the value may be decreased to get a higher precision or increased to achieve a faster calculation. The calculation intervals always include at least all times of event and censoring as border points.} \item{add.times}{specific times at which the value of estimator and its variance should be evaluated. Default is all the event and censoring times.} \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details below.} } \details{ NOTE: The follow-up time must be specified in days. The \code{ratetable} being used may have different variable names and formats than the user's data set, this is dealt with by the \code{rmap} argument. For example, if age is in years in the data set but in days in the \code{ratetable} object, age=age*365.241 should be used. The calendar year can be in any date format (date, Date and POSIXt are allowed), the date formats in the \code{ratetable} and in the data may differ. Note that numerical integration is required to calculate the variance estimator. The integration precision is set with argument \code{precision}, which defaults to daily intervals, a default that should give enough precision for any practical purpose. The area under the curve is calculated on the interval [0,\code{tau}]. Function \code{summary} may be used to get the output at specific points in time. } \value{ An object of class \code{cmp.rel}. Objects of this class have methods for the functions \code{print} and \code{plot}. The \code{summary} function can be used for printing output at required time points. An object of class \code{cmp.rel} is composed of several lists, each pertaining the cumulative hazard function for one risk and one strata. Each of the lists contains the following objects: \item{time}{the time-points at which the curves are estimated} \item{est}{the estimate} \item{var}{the variance of the estimate} \item{lower}{the lower limit of the confidence interval} \item{upper}{the upper limit of the confidence interval} \item{area}{the area under the curve calculated on the interval [0,\code{tau}]} \item{index}{indicator of event and censoring times among all the times in the output. The times added via paramater \code{add.times} are also included} \item{add.times}{the times added via parameter \code{add.times}} } \references{ Package: Pohar Perme, M., Pavlic, K. (2018) "Nonparametric Relative Survival Analysis with the {R} Package {relsurv}". Journal of Statistical Software. 87(8), 1-27, doi: "10.18637/jss.v087.i08" } \examples{ data(slopop) data(rdata) #calculate the crude probability of death #note that the variable year must be given in a date format and that #age must be multiplied by 365.241 in order to be expressed in days. fit <- cmp.rel(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,tau=3652.41) fit plot(fit,col=c(1,1,2,2),xscale=365.241,xlab="Time (years)") #if no strata are desired: fit <- cmp.rel(Surv(time,cens)~1,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,tau=3652.41) } \seealso{ \code{rs.surv}, \code{summary.cmp.rel} } \keyword{survival} relsurv/man/nessie.Rd0000644000176200001440000000422713400170743014310 0ustar liggesusers\name{nessie} \alias{nessie} \title{Net Expected Sample Size Is Estimated} \description{ Calculates how the sample size decreases in time due to population mortality } \usage{ nessie(formula, data, ratetable = relsurv::slopop,times,rmap) } \arguments{ \item{formula}{ a formula object, same as in \code{rs.surv}. The right-hand side of the formula object includes the variable that defines the subgroups (a variable of type \code{factor}) by which the expected sample size is to be calculated. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, organized as a \code{ratetable} object, such as \code{slopop}. } \item{times}{Times at which the calculation should be evaluated - in years!} \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details of the \code{rs.surv} function.} } \details{ The function calculates the sample size we can expect at a certain time point if the patients die only due to population causes (population survival * initial sample size in a certain category), i.e. the number of individuals that remains at risk at given timepoints after the individuals who die due to population causes are removed. The result should be used as a guideline for the sensible length of follow-up interval when calculating the net survival. The first column of the output reports the number of individuals at time 0. The last column of the output reports the conditional expected (population) survival time for each subgroup. } \value{ A list of values. } \references{ Pohar Perme, M., Pavlic, K. (2018) "Nonparametric Relative Survival Analysis with the {R} Package {relsurv}". Journal of Statistical Software. 87(8), 1-27, doi: "10.18637/jss.v087.i08" } \examples{ data(slopop) data(rdata) rdata$agegr <-cut(rdata$age,seq(40,95,by=5)) nessie(Surv(time,cens)~agegr,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,times=c(1,3,5,10,15)) } \seealso{ \code{rs.surv} } \keyword{survival} relsurv/man/rsmul.Rd0000644000176200001440000000724713332517377014205 0ustar liggesusers\name{rsmul} \alias{rsmul} \title{Fit Andersen et al Multiplicative Regression Model for Relative Survival} \description{ Fits the Andersen et al multiplicative regression model in relative survival. An extension of the coxph function using relative survival. } \usage{ rsmul(formula, data, ratetable = relsurv::slopop, int,na.action,init, method,control,rmap,...) } \arguments{ \item{formula}{ a formula object, with the response as a \code{Surv} object on the left of a \code{~} operator, and, if desired, terms separated by the \code{+} operator on the right. NOTE: The follow-up time must be in days. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, such as \code{slopop}. } \item{int}{ the number of follow-up years used for calculating survival(the data are censored after this time-point). If missing, it is set the the maximum observed follow-up time. } \item{na.action}{a missing-data filter function, applied to the model.frame, after any subset argument has been used. Default is \code{options()$na.action}. } \item{init}{vector of initial values of the iteration. Default initial value is zero for all variables. } \item{method}{the default method \code{mul} assumes hazard to be constant on yearly intervals. Method \code{mul1} uses the ratetable to determine the time points when hazard changes. The \code{mul1} method is therefore more accurate, but at the same time can be more computationally intensive.} \item{control}{a list of parameters for controlling the fitting process. See the documentation for \code{coxph.control} for details. } \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details below.} \item{...}{Other arguments will be passed to \code{coxph.control}.} } \value{ an object of class \code{coxph} with an additional item: \item{basehaz}{Cumulative baseline hazard (population values are seen as offset) at centered values of covariates.} } \details{ NOTE: The follow-up time must be specified in days. The \code{ratetable} being used may have different variable names and formats than the user's data set, this is dealt with by the \code{rmap} argument. For example, if age is in years in the data set but in days in the \code{ratetable} object, age=age*365.241 should be used. The calendar year can be in any date format (date, Date and POSIXt are allowed), the date formats in the \code{ratetable} and in the data may differ. } \references{ Method: Andersen, P.K., Borch-Johnsen, K., Deckert, T., Green, A., Hougaard, P., Keiding, N. and Kreiner, S. (1985) "A Cox regression model for relative mortality and its application to diabetes mellitus survival data.", Biometrics, \bold{41}: 921--932. Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \examples{ data(slopop) data(rdata) #fit a multiplicative model #note that the variable year is given in days since 01.01.1960 and that #age must be multiplied by 365.241 in order to be expressed in days. fit <- rsmul(Surv(time,cens)~sex+as.factor(agegr),rmap=list(age=age*365.241), ratetable=slopop,data=rdata) #check the goodness of fit rs.br(fit) } \seealso{\code{\link{rsadd}}, \code{\link{rstrans}}. } \keyword{survival} relsurv/man/summary.cmp.rel.Rd0000644000176200001440000000246113332517527016065 0ustar liggesusers\name{summary.cmp.rel} \alias{summary.cmp.rel} \title{Summary of the crude probability of death} \description{ Returns a list containing the estimated values at required times. } \usage{ \method{summary}{cmp.rel}(object, times, scale = 365.241,area=FALSE,...) } \arguments{ \item{object}{output of the function \code{cmp.rel}.} \item{times}{the times at which the output is required.} \item{scale}{The time scale in which the times are specified. The default value is \code{1}, i.e. days.} \item{area}{Should area under the curves at time \code{tau} be printed out? Default is \code{FALSE}.} \item{...}{Additional arguments, currently not implemented} } \details{ The variance is calculated using numerical integration. If the required time is not a time at which the value was estimated, the value at the last time before it is reported. The density of the time points is set by the \code{precision} argument in the \code{cmp.rel} function. } \value{ A list of values is returned. } \examples{ data(slopop) data(rdata) #calculate the crude probability of death and summarize it fit <- cmp.rel(Surv(time,cens)~sex,rmap=list(age=age*365), ratetable=slopop,data=rdata,tau=3652.41) summary(fit,c(1,3),scale=365.241) } \seealso{ \code{cmp.rel} } \keyword{survival} relsurv/man/rssurvrsadd.Rd0000644000176200001440000000377013332517445015416 0ustar liggesusers\name{rs.surv.rsadd} \alias{rs.surv.rsadd} \title{Compute a Relative Survival Curve from an additive relative survival model} \description{ Computes the predicted relative survival function for an additive relative survival model fitted with maximum likelihood. } \usage{ rs.surv.rsadd(formula, newdata) } \arguments{ \item{formula}{ a \code{rsadd} object (Implemented only for models fitted with the code{max.lik} (default) option.) } \item{newdata}{ a data frame with the same variable names as those that appear in the \code{rsadd} formula. a predicted curve for each individual in this data frame shall be calculated } } \details{Does not work with factor variables - you have to form dummy variables before calling the rsadd function.} \value{ a \code{survfit} object; see the help on \code{survfit.object} for details. The \code{survfit} methods are used for \code{print}, \code{plot}, \code{lines}, and \code{points}. } \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 } \examples{ data(slopop) data(rdata) #fit a relative survival model fit <- rsadd(Surv(time,cens)~sex+age+year,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,int=c(0:10,15)) #calculate the predicted curve for a male individual, aged 65, diagnosed in 1982 d <- rs.surv.rsadd(fit,newdata=data.frame(sex=1,age=65,year=as.date("1Jul1982"))) #plot the curve (will result in a step function since the baseline is assumed piecewise constant) plot(d,xscale=365.241) #calculate the predicted survival curves for each individual in the data set d <- rs.surv.rsadd(fit,newdata=rdata) #calculate the average over all predicted survival curves p.surv <- apply(d$surv,1,mean) #plot the relative survival curve plot(d$time/365.241,p.surv,type="b",ylim=c(0,1),xlab="Time",ylab="Relative survival") } \seealso{ \code{survfit}, \code{survexp} } \keyword{survival} relsurv/man/joinrate.Rd0000644000176200001440000000265712531542107014643 0ustar liggesusers\name{joinrate} \alias{joinrate} \title{Join ratetables} \description{ The function joins two or more objects organized as \code{ratetable} by adding a new dimension. } \usage{ joinrate(tables,dim.name="country") } \arguments{ \item{tables}{ a list of ratetables. If names are given, they are included as \code{dimnames}. } \item{dim.name}{ the name of the added dimension. } } \details{ This function joins two or more \code{ratetable} objects by adding a new dimension. The cutpoints of all the rate tables are compared and only the common intervals kept. If the intervals defined by the cutpoints are not of the same length, a warning message is displayed. Each rate table must have 3 dimensions, i.e. age, sex and year (the order is not important). } \value{An object of class \code{ratetable}.} \references{ Package: Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272-278. Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741-1749. } \examples{ #newpop <- joinrate(list(Arizona=survexp.az,Florida=survexp.fl, # Minnesota=survexp.mn),dim.name="state") } \seealso{\code{\link{ratetable}}, \code{\link{transrate.hld}}, \code{\link{transrate.hmd}}, \code{\link{transrate}}.} \keyword{survival} relsurv/man/colrec.Rd0000644000176200001440000000150512705412213014263 0ustar liggesusers\name{colrec} \alias{colrec} \docType{data} \title{Relative Survival Data} \description{ Survival of patients with colon and rectal cancer diagnosed in 1994-2000. } \usage{data(colrec)} \format{ A data frame with 5971 observations on the following 7 variables: \describe{ \item{sex}{sex (1=male, 2=female).} \item{age}{age (in days).} \item{diag}{date of diagnosis (in date format).} \item{time}{survival time (in days).} \item{stat}{censoring indicator (0=censoring, 1=death).} \item{stage}{cancer stage. Values 1-3, code \code{99} stands for unknown.} \item{site}{cancer site. } } } \references{ Provided by Slovene Cancer Registry. The \code{age}, \code{time} and \code{diag} variables are randomly perturbed to make the identification of patients impossible. } \keyword{datasets} relsurv/man/rszph.Rd0000644000176200001440000000457613332517511014201 0ustar liggesusers\name{rs.zph} \alias{rs.zph} \title{Behaviour of Covariates in Time for Relative Survival Regression Models} \description{ Calculates the scaled partial residuals of a relative survival model (\code{rsadd}, \code{rsmul} or \code{rstrans})} \usage{ rs.zph(fit,sc,transform="identity",var.type="sum") } \arguments{ \item{fit}{ the result of fitting an additive relative survival model, using the \code{rsadd}, \code{rsmul} or \code{rstrans} function. In the case of multiplicative and transformation models the output is identical to \code{cox.zph} function, except no test is performed. } \item{sc}{ partial residuals calculated by the \code{resid} function. This is used to save time if several tests are to be calculated on these residuals and can otherwise be omitted. } \item{transform}{ a character string specifying how the survival times should be transformed. Possible values are \code{"km"}, \code{"rank"}, \code{"identity"} and \code{log}. The default is \code{"identity"}. } \item{var.type}{ a character string specifying the variance used to scale the residuals. Possible values are \code{"each"}, which estimates the variance for each residual separately, and \code{sum}(default), which assumes the same variance for all the residuals. } } \value{ an object of class \code{rs.zph}. This function would usually be followed by a plot of the result. The plot gives an estimate of the time-dependent coefficient \code{beta(t)}. If the proportional hazards assumption is true, \code{beta(t)} will be a horizontal line. } \references{ Goodness of fit: Stare J.,Pohar Perme M., Henderson R. (2005) "Goodness of fit of relative survival models." Statistics in Medicine, \bold{24}: 3911--3925. Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \examples{ data(slopop) data(rdata) fit <- rsadd(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata,int=5) rszph <- rs.zph(fit) plot(rszph) } \seealso{\code{\link{rsadd}}, \code{rsmul}, \code{rstrans}, \code{\link{resid}}, \code{\link{cox.zph}}.} \keyword{survival} relsurv/man/slopop.Rd0000644000176200001440000000035510063271174014336 0ustar liggesusers\name{slopop} \alias{slopop} \docType{data} \title{Census Data Set for the Slovene Population} \description{ Census data set for the Slovene population. } \usage{data(slopop)} \examples{ data(slopop) } \keyword{datasets} relsurv/man/plotrszph.Rd0000644000176200001440000000443613332504432015071 0ustar liggesusers\name{plot.rs.zph} \alias{plot.rs.zph} \title{Graphical Inspection of Proportional Hazards Assumption in Relative Survival Models} \description{ Displays a graph of the scaled partial residuals, along with a smooth curve. } \usage{ \method{plot}{rs.zph}(x, resid=TRUE, df = 4, nsmo = 40,var,cex=1,add=FALSE,col=1, lty=1,xlab,ylab,xscale=1,...) } \arguments{ \item{x}{ result of the \code{rs.zph} function. } \item{resid}{a logical value, if \code{TRUE} the residuals are included on the plot, as well as the smooth fit.} \item{df}{ the degrees of freedom for the fitted natural spline, \code{df=2} leads to a linear fit. } \item{nsmo}{ number of points used to plot the fitted spline. } \item{var}{ the set of variables for which plots are desired. By default, plots are produced in turn for each variable of a model. Selection of a single variable allows other features to be added to the plot, e.g., a horizontal line at zero or a main title. } \item{cex}{a numerical value giving the amount by which plotting text and symbols should be scaled relative to the default. } \item{add}{logical, if \code{TRUE} the plot is added to an existing plot} \item{col}{a specification for the default plotting color.} \item{lty}{the line type.} \item{xlab}{x axis label.} \item{ylab}{y axis label.} \item{xscale}{units for x axis, default is 1, i.e. days.} \item{...}{Additional arguments passed to the \code{plot} function. } } \references{ Goodness of fit: Stare J.,Pohar Perme M., Henderson R. (2005) "Goodness of fit of relative survival models." Statistics in Medicine, \bold{24}: 3911-3925. Package: Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272-278. Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741-1749, 2007. } \examples{ data(slopop) data(rdata) fit <- rsadd(Surv(time,cens)~sex+as.factor(agegr),rmap=list(age=age*365.241), ratetable=slopop,data=rdata,int=5) rszph <- rs.zph(fit) plot(rszph) } \seealso{\code{\link{rs.zph}}, \code{\link{plot.cox.zph}}.} \keyword{survival} relsurv/man/rstrans.Rd0000644000176200001440000000730313332517464014525 0ustar liggesusers\name{rstrans} \alias{rstrans} \title{Fit Cox Proportional Hazards Model in Transformed Time} \description{ The function transforms each person's time to his/her probability of dying at that time according to the ratetable. It then fits the Cox proportional hazards model with the transformed times as a response. It can also be used for calculatin the transformed times (no covariates are needed in the formula for that purpose). } \usage{ rstrans(formula, data, ratetable, int,na.action,init,control,rmap,...) } \arguments{ \item{formula}{ a formula object, with the response as a \code{Surv} object on the left of a \code{~} operator, and, if desired, terms separated by the \code{+} operator on the right. NOTE: The follow-up time must be in days. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, such as \code{slopop}. } \item{int}{ the number of follow-up years used for calculating survival(the rest is censored). If missing, it is set the the maximum observed follow-up time. } \item{na.action}{a missing-data filter function, applied to the model.frame, after any subset argument has been used. Default is \code{options()$na.action}. } \item{init}{vector of initial values of the iteration. Default initial value is zero for all variables. } \item{control}{a list of parameters for controlling the fitting process. See the documentation for \code{coxph.control} for details. } \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details below.} \item{...}{other arguments will be passed to \code{coxph.control}.} } \value{ an object of class \code{coxph}. See \code{coxph.object} and \code{coxph.detail} for details. \item{y}{ an object of class \code{Surv} containing the transformed times (these times do not depend on covariates). } } \details{ NOTE: The follow-up time must be specified in days. The \code{ratetable} being used may have different variable names and formats than the user's data set, this is dealt with by the \code{rmap} argument. For example, if age is in years in the data set but in days in the \code{ratetable} object, age=age*365.241 should be used. The calendar year can be in any date format (date, Date and POSIXt are allowed), the date formats in the \code{ratetable} and in the data may differ. A side product of this function are the transformed times - stored in teh \code{y} object of the output. To get these times, covariates are of course irrelevant. } \references{ Method: Stare J., Henderson R., Pohar M. (2005) "An individual measure for relative survival." Journal of the Royal Statistical Society: Series C, \bold{54} 115--126. Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \examples{ data(slopop) data(rdata) #fit a Cox model using the transformed times #note that the variable year is given in days since 01.01.1960 and that #age must be multiplied by 365.241 in order to be expressed in days. fit <- rstrans(Surv(time,cens)~sex+as.factor(agegr),rmap=list(age=age*365.241, sex=sex,year=year),ratetable=slopop,data=rdata) #check the goodness of fit rs.br(fit) } \seealso{\code{\link{rsmul}}, \code{\link{invtime}}, \code{\link{rsadd}}, \code{\link{survexp}}.} \keyword{survival} relsurv/man/rsdiff.Rd0000644000176200001440000000572613400170664014306 0ustar liggesusers\name{rs.diff} \alias{rs.diff} \alias{print.rsdiff} \title{Test Net Survival Curve Differences} \description{ Tests if there is a difference between two or more net survival curves using a log-rank type test. } \usage{ rs.diff(formula, data, ratetable = relsurv::slopop, na.action,precision=1,rmap) } \arguments{ \item{formula}{ A formula expression as for other survival models, of the form \code{Surv(time, status) ~ predictors}. Each combination of predictor values defines a subgroup. A \code{strata} term may be used to produce a stratified test. NOTE: The follow-up time must be in days. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, organized as a \code{ratetable} object, such as \code{slopop}. } \item{na.action}{a missing-data filter function, applied to the model.frame, after any subset argument has been used. Default is \code{options()$na.action}.} \item{precision}{Precision for numerical integration. Default is 1, which means that daily intervals are taken, the value may be decreased to get a higher precision or increased to achieve a faster calculation. The calculation intervals always include at least all times of event and censoring as border points. } \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details below.} } \details{ NOTE: The follow-up time must be specified in days. The \code{ratetable} being used may have different variable names and formats than the user's data set, this is dealt with by the \code{rmap} argument. For example, if age is in years in the data set but in days in the \code{ratetable} object, age=age*365.241 should be used. The calendar year can be in any date format (date, Date and POSIXt are allowed), the date formats in the \code{ratetable} and in the data may differ. } \value{ a \code{rsdiff} object; can be printed with \code{print}. } \references{ Package: Pohar Perme, M., Pavlic, K. (2018) "Nonparametric Relative Survival Analysis with the {R} Package {relsurv}". Journal of Statistical Software. 87(8), 1-27, doi: "10.18637/jss.v087.i08" Theory: Graffeo, N., Castell, F., Belot, A. and Giorgi, R. (2016) "A log-rank-type test to compare net survival distributions. Biometrics. doi: 10.1111/biom.12477" Theory: Pavlic, K., Pohar Perme, M. (2017) "On comparison of net survival curves. BMC Med Res Meth. doi: 10.1186/s12874-017-0351-3" } \examples{ data(slopop) data(rdata) #calculate the relative survival curve #note that the variable year is given in days since 01.01.1960 and that #age must be multiplied by 365.241 in order to be expressed in days. rs.diff(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata) } \seealso{ \code{rs.surv}, \code{survdiff} } \keyword{survival} relsurv/man/rdata.Rd0000644000176200001440000000122211203231600014072 0ustar liggesusers\name{rdata} \alias{rdata} \docType{data} \title{Survival Data} \description{ Survival data. } \usage{data(rdata)} \format{ A data frame with 1040 observations on the following 6 variables: \describe{ \item{time}{survival time (in days).} \item{cens}{censoring indicator (0=censoring, 1=death).} \item{age}{age (in years).} \item{sex}{sex (1=male, 2=female).} \item{year}{date of diagnosis (in date format).} \item{agegr}{age group.} } } \references{ Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272-278. } \keyword{datasets} relsurv/man/transratehld.Rd0000644000176200001440000000365411203233051015507 0ustar liggesusers\name{transrate.hld} \alias{transrate.hld} \title{Reorganize Data obtained from Human Life-Table Database into a Ratetable Object} \description{ The function assists in reorganizing the .txt files obtained from Human Life-Table Database (http://www.lifetable.de -> Data by Country) into a ratetable object. } \usage{ transrate.hld(file,cut.year,race) } \arguments{ \item{file}{ a vector of file names which the data are to be read from. Must be in .tex format and in the same format as the files in Human Life-Table Database. } \item{cut.year}{ a vector of cutpoints for years. Must be specified when the year spans in the files are not consecutive. } \item{race}{a vector of race names for the input files.} } \details{ This function works with any table organised in the format provided by the Human Life-Table Database, but currently only works with TypeLT 1 (i.e. age intervals of length 1). The age must always start with value 0, but can end at different values (when that happens, the last value is carried forward). The rates between the cutpoints are taken to be constant. } \value{An object of class \code{ratetable}.} \examples{ \dontrun{ finpop <- transrate.hld(c("FIN_1981-85.txt","FIN_1986-90.txt","FIN_1991-95.txt")) } \dontrun{ nzpop <- transrate.hld(c("NZL_1980-82_Non-maori.txt","NZL_1985-87_Non-maori.txt", "NZL_1980-82_Maori.txt","NZL_1985-87_Maori.txt"), cut.year=c(1980,1985),race=rep(c("nonmaori","maori"),each=2)) } } \references{ Package. Pohar M., Stare J. (2006) "Relative survival analysis in R." Computer Methods and Programs in Biomedicine, \bold{81}: 272--278 Relative survival: Pohar, M., Stare, J. (2007) "Making relative survival analysis relatively easy." Computers in biology and medicine, \bold{37}: 1741--1749. } \seealso{\code{\link{ratetable}}, \code{\link{transrate.hmd}}, \code{\link{joinrate}}, \code{\link{transrate}}.} \keyword{survival} relsurv/man/rssurv.Rd0000644000176200001440000001316413400221657014367 0ustar liggesusers\name{rs.surv} \alias{rs.surv} \title{Compute a Relative Survival Curve} \description{ Computes an estimate of the relative survival curve using the Ederer I, Ederer II method, Pohar-Perme method or the Hakulinen method } \usage{ rs.surv(formula, data,ratetable=relsurv::slopop,na.action,fin.date, method="pohar-perme", conf.type="log",conf.int=0.95,type="kaplan-meier", add.times,precision=1,rmap) } \arguments{ \item{formula}{ a formula object, with the response as a \code{Surv} object on the left of a \code{~} operator, and, if desired, terms separated by the \code{+} operator on the right. If no strata are used, \code{~1} should be specified. NOTE: The follow-up time must be in days. } \item{data}{ a data.frame in which to interpret the variables named in the \code{formula}. } \item{ratetable}{ a table of event rates, organized as a \code{ratetable} object, such as \code{slopop}. } \item{na.action}{a missing-data filter function, applied to the model.frame, after any subset argument has been used. Default is \code{options()$na.action}.} \item{fin.date}{the date of the study ending, used for calculating the potential follow-up times in the Hakulinen method. If missing, it is calculated as \code{max(year+time)}. } \item{method}{the method for calculating the relative survival. The options are \code{pohar-perme}(default), \code{ederer1}, \code{ederer2} and \code{hakulinen}.} \item{conf.type}{one of \code{plain}, \code{log} (the default), or \code{log-log}. The first option causes the standard intervals curve +- k *se(curve), where k is determined from conf.int. The log option calculates intervals based on the cumulative hazard or log(survival). The last option bases intervals on the log hazard or log(-log(survival)). } \item{conf.int}{the level for a two-sided confidence interval on the survival curve(s). Default is 0.95.} \item{type}{defines how survival estimates are to be calculated given the hazards. The default (\code{kaplan-meier}) calculates the product integral, whereas the option \code{fleming-harrington} exponentiates the negative cumulative hazard. Analogous to the usage in \code{survfit}. } \item{add.times}{specific times at which the curve should be evaluated. } \item{precision}{Precision for numerical integration. Default is 1, which means that daily intervals are taken, the value may be decreased to get a higher precision or increased to achieve a faster calculation. The calculation intervals always include at least all times of event and censoring as border points. } \item{rmap}{an optional list to be used if the variables are not organized and named in the same way as in the \code{ratetable} object. See details below.} } \details{ NOTE: The follow-up time must be specified in days. The \code{ratetable} being used may have different variable names and formats than the user's data set, this is dealt with by the \code{rmap} argument. For example, if age is in years in the data set but in days in the \code{ratetable} object, age=age*365.241 should be used. The calendar year can be in any date format (date, Date and POSIXt are allowed), the date formats in the \code{ratetable} and in the data may differ. The potential censoring times needed for the calculation of the expected survival by the Hakulinen method are calculated automatically. The times of censoring are left as they are, the times of events are replaced with \code{fin.date - year}. The calculation of the Pohar-Perme estimate is more time consuming since more data are needed from the population tables. The old version of the function, now named \code{rs.survo} can be used as a faster version for the Hakulinen and Ederer II estimate. Numerical integration is required for Pohar-Perme estimate. The integration precision is set with argument \code{precision}, which defaults to daily intervals, a default that should give enough precision for any practical purpose. Note that even though the estimate is always calculated using numerical integration, only the values at event and censoring times are reported. Hence, the function \code{plot} draws a step function in between and the function \code{summary} reports the value at the last event or censoring time before the specified time. If the output of the estimated values at other points is required, this should be specified with argument \code{add.times}. } \value{ a \code{survfit} object; see the help on \code{survfit.object} for details. The \code{survfit} methods are used for \code{print}, \code{summary}, \code{plot}, \code{lines}, and \code{points}. } \references{ Package: Pohar Perme, M., Pavlic, K. (2018) "Nonparametric Relative Survival Analysis with the {R} Package {relsurv}". Journal of Statistical Software. 87(8), 1-27, doi: "10.18637/jss.v087.i08" Theory: Pohar Perme, M., Esteve, J., Rachet, B. (2016) "Analysing Population-Based Cancer Survival - Settling the Controversies." BMC Cancer, 16 (933), 1-8. doi:10.1186/s12885-016-2967-9. Theory: Pohar Perme, M., Stare, J., Esteve, J. (2012) "On Estimation in Relative Survival", Biometrics, 68(1), 113-120. doi:10.1111/j.1541-0420.2011.01640.x. } \examples{ data(slopop) data(rdata) #calculate the relative survival curve #note that the variable year must be given in a date format and that #age must be multiplied by 365.241 in order to be expressed in days. rs.surv(Surv(time,cens)~sex,rmap=list(age=age*365.241), ratetable=slopop,data=rdata) } \seealso{ \code{survfit}, \code{survexp} } \keyword{survival} relsurv/man/plotcmprel.Rd0000644000176200001440000000627213332504412015203 0ustar liggesusers\name{plot.cmp.rel} \alias{plot.cmp.rel} \title{Plot the crude probability of death} \description{ Plot method for cmp.rel. Plots the cumulative probability of death due to disease and due to population reasons } \usage{ \method{plot}{cmp.rel}(x, main, curvlab, ylim=c(0, 1), xlim, wh=2, xlab="Time (days)", ylab="Probability", lty=1:length(x), xscale=1,col=1, lwd=par('lwd'), curves, conf.int, all.times=FALSE,...) } \arguments{ \item{x}{a list, with each component representing one curve in the plot, output of the function \code{cmp.rel}.} \item{main}{the main title for the plot.} \item{curvlab}{Curve labels for the plot. Default is \code{names(x)}, or if that is missing, \code{1:nc}, where \code{nc} is the number of curves in \code{x}.} \item{ylim}{yaxis limits for plot.} \item{xlim}{xaxis limits for plot (default is 0 to the largest time in any of the curves).} \item{wh}{if a vector of length 2, then the upper right coordinates of the legend; otherwise the legend is placed in the upper right corner of the plot.} \item{xlab}{X axis label.} \item{ylab}{y axis label.} \item{lty}{vector of line types. Default \code{1:nc} (\code{nc} is the number of curves in \code{x}). For color displays, \code{lty=1}, \code{color=1:nc}, might be more appropriate. If \code{length(lty)