Enrico, I was in a similar position with a small likelihood function which was not vectorizable (is that a word?) and in one morning I had a working function using Rcpp which was 60 times faster than my reference function in R. I *strongly* recommend that you use C++ through the Rcpp interface which really simplifies things, refer to
http://adv-r.had.co.nz/Rcpp.html http://rcpp.org/ One possible suggestion for a tempate for your function might be found on http://gallery.rcpp.org/articles/gibbs-sampler/ The Rcpp-devel mailing group is very very helpful and responsive. Kind regards, Sean On 8 July 2014 09:33, Enrico Bibbona <enrico.bibb...@gmail.com> wrote: > I would like to write a fast version of a function which contains slow > loops that I cannot vectorize. > I would like to rewrite it in C, but in the loops I need to call a few > (two) R functions. > Which is the best way to do so? I have read that one can call a C library > which import all the R functions (R.h, is it?). Is that efficient? I tried > to have a look into the internal C code to find just the two functions I > need, but it is quite complicated and I did not succeeded. > Any suggestion? > Thanks a lot, Enrico Bibbona > > [[alternative HTML version deleted]] > > ______________________________________________ > R-devel@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel > [[alternative HTML version deleted]] ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
