Hello,

  I spent last weekend trying to run this piece of code (and a lot of variations too):

suppressPackageStartupMessages(require(inline))
suppressPackageStartupMessages(require(Rcpp))

vero.empi.cpp <- '
  Rcpp::NumericVector xa(vpf) ;
  Rcpp::NumericVector xb(vpe) ;
  Rcpp::NumericVector xc(tempo) ;
  Rcpp::NumericVector xd(evento) ;
  int n_xa = xa.size(), n_xb = xb.size(), n_xd = xd.size() ;
  Rcpp::NumericMatrix vero(n_xa,n_xb)
  Rcpp::Function d(fun1) ;
  Rcpp::Function S(fun2) ;
  float aux ;
  for (int i = 0; i < n_xa; i++){
    for (int j = 0; j < n_xb; j++){
      aux = 0 ;
      for(int k = 0; k < n_xd; k++){
      aux = aux*(d(xc[k], xa[i], xb[j])^xd[k])*(S(xc[k], xa[i], xb[j])^(1- xd[k])) ;
      }
      vero[i , j] = aux ;
    }
  }
  return vero;
'

vero.empi <- cxxfunction(
  signature(vpf="numeric", vpe="numeric", tempo="numeric", evento="numeric", fun1 = "function", fun2 = "function"),
  vero.empi.cpp, plugin="Rcpp", include = "using namespace Rcpp; ")

The purpose here is to generate a matrix of empirical values of a likelihood. For a first trial, I would like to use the R functions dweibull() and pweibull() in fun1 and fun2, respectively. But it didn't work for some reason I cannot see. Is this possible?


Thanks,
Prof. Hermes
PS.: Sorry for my bad English!
_______________________________________________
Rcpp-devel mailing list
[email protected]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
  • [Rcpp-devel] Begginer ... Antonio Hermes M. da S. Junior - Depto. Estatistica

Reply via email to