> On 4 Mar 2016, at 08:51, MAURICE Jean - externe <jean-externe.maur...@edf.fr> > wrote: > > Hi Berend, > > >> The question belongs on the R-devel mailinglist. > I try to find this mailing-list ... >
See https://www.r-project.org/mail.html >> You are calling your Fortran routines directly from an R file with .Fortran, >> I presume? > Yes. Is there another solution (possibility ?) > Yes. Use C to interface with Fortran. See for example packages: minpack.lm, QZ, expm, nleqslv. And more: rootSolve, ucminf. There are probably even more. >> Declare them with (:,:) or (*,*), allocate and the return the array? A >> Fortran95 pointer? > That was what I wanted to do ... > >> I think all of that is just a no no. > It would be a big problem with memory management but R could copy the result > before the memory attached to the FORTRAN subroutine is cleared at the end of > the routine. R does so strange things (for an old programmer as I am). > >> Is it possible to calculate the required size of the array in a separate >> routine given some parameters (as can often be done with >Lapack routines) >> before calling the main subroutine? >> If so, call the sizing routine and then declare the correctly sized array in >> your R code and pass it to the main subroutine. > > It's a good idea : a 'two shots' routine. I'll dig that (I'll work in this > direction). > Commercial: Have a look at (my) packages nleqslv and geigen for that approach. Berend ______________________________________________ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
