On 31 May 2013 at 16:35, Colin Rundel wrote:
| I think that I've actually found the issue, the liblapack.so and libblas.so
installed in my /usr/lib/R/lib/ were actually from the revolution-mkl package
and not from r-base-core. Once I purged that package everything seems to be
working. Not sure
I think that I've actually found the issue, the liblapack.so and libblas.so
installed in my /usr/lib/R/lib/ were actually from the revolution-mkl package
and not from r-base-core. Once I purged that package everything seems to be
working. Not sure what criteria the g++ linker was using to chose
Hi Colin,
On 31 May 2013 at 15:31, Colin Rundel wrote:
| For certain operations R does _not_ go to lapack but uses its own. I can
| never remember if chol() was one of them -- but this suggests it. As I
| mentioned in my earlier email you probably really have to follow the
chol()
|
Hi Colin,
I also found an article referring to a blog of Dirk:
http://techyoubaji.blogspot.de/2012/12/r-uses-different-blas-and-lapack.html
Maybe it is interesting for you.
Best
Simon
On May 31, 2013, at 9:55 PM, Simon Zehnder wrote:
> Isn't this actually an option when installing R
> (ht
Isn't this actually an option when installing R
(http://www.cran.r-project.org/doc/manuals/R-admin.html)? If it hasn't been
explicitly stated in the configure stage of R via --with-lapack, R uses its
internal routines.
Best
Simon
On May 31, 2013, at 9:31 PM, Colin Rundel wrote:
>> For cer
Hi Asis,
in addition to what Dirk said I can tell you, that at least Armadillo Objects
do work accurate with OpenMP. I used in my code an '#pragma omp parallel for'
directive on a lot of Armadillo matrices without any problems. Be aware of what
clauses you use in your parallel setting.
There
> For certain operations R does _not_ go to lapack but uses its own. I can
> never remember if chol() was one of them -- but this suggests it. As I
> mentioned in my earlier email you probably really have to follow the chol()
> call all the way down (in the sources).
I've followed both calls to t
On 31 May 2013 at 18:00, Asis Hallab wrote:
| Dear Rcpp Experts,
|
| in my current project I wanted to make use of OpenMP.
That works.
| Unfortunately it seems Rcpp Objects cannot be used inside a block
| executed in parallel.
I think if you're careful it may work. You definitely do not want
On 31 May 2013 at 12:00, Søren Højsgaard wrote:
| Dear all;
|
| In c++ I want to do:
| x<-c(2,4,6,8,10)
| i<- c(2,3,5)
| x[i]
|
| I do:
|
| cat("
| #include
|
| using namespace Rcpp;
| using namespace arma;
|
| // [[Rcpp::export]]
| NumericVector select(NumericVector vec_, NumericVector idx_
Dear Rcpp Experts,
in my current project I wanted to make use of OpenMP.
Unfortunately it seems Rcpp Objects cannot be used inside a block
executed in parallel.
Seg fault errors are the consequence, according to Dirk's reply in this thread
http://lists.r-forge.r-project.org/pipermail/rcpp-devel/20
Dear all;
In c++ I want to do:
x<-c(2,4,6,8,10)
i<- c(2,3,5)
x[i]
I do:
cat("
#include
using namespace Rcpp;
using namespace arma;
// [[Rcpp::export]]
NumericVector select(NumericVector vec_, NumericVector idx_){
arma::vec vv = Rcpp::as(vec_);
arma::uvec ii = Rcpp::as(idx_);
arma::vec
Romain and Dirk:
Thanks for your help and patience!
Søren
-Original Message-
From: Dirk Eddelbuettel [mailto:[email protected]]
Sent: 31. maj 2013 13:24
To: Søren Højsgaard
Cc: Romain Francois; [email protected]
Subject: Re: [Rcpp-devel] Turning a vector into a matrix
On 31 May 2013 at 10:28, Søren Højsgaard wrote:
| Romain,
|
| Perhaps I now see the light at the end of the tunnel (hope it is not a train
comming against me!): If I understand this correctly, the following
|
| > template
| > Matrix vec2matrix2(Vector x, int nrow, int ncol) {
| >return Mat
On 31 May 2013 at 02:05, Colin Rundel wrote:
|
| > reference blas (packages libblas3 and liblapack)
| > atlas (package libatlas3-base or a tuned variant)
| > open-blas (package libopenblas-base)
|
| I had been playing around with update-alternatives to swap between the thre
Romain,
Perhaps I now see the light at the end of the tunnel (hope it is not a train
comming against me!): If I understand this correctly, the following
> template
> Matrix vec2matrix2(Vector x, int nrow, int ncol) {
>return Matrix(nrow, ncol, x.begin()); }
should do the dispatching proper
Another way is to leverage our RCPP_RETURN_VECTOR macro, and use it with
a functor instead of a function. The idea is to encapsulate the
information (nrow and ncol) in the functor class:
#include
using namespace Rcpp ;
class Vec2Matrix {
public:
Vec2Matrix(int nrow_, int ncol_): nrow(nrow
Ah ok, so you want to use this from R, I would suggest to Rcpp::export a
dispatcher, something like this:
#include
using namespace Rcpp ;
// the template
template
Matrix vec2matrix__impl(Vector x, int nrow, int ncol) {
return Matrix(nrow, ncol, x.begin());
}
// the exported dispatcher,used
Romain,
I actually already tried this in my file util.cpp but can't get it to work:
#include
using namespace Rcpp;
// [[Rcpp::export]]
NumericMatrix vec2matrix(NumericVector x, int nrow, int ncol) {
return NumericMatrix(nrow, ncol, x.begin());
}
// [[Rcpp::export]]
template
Matrix vec2ma
Hello Søren,
This is easy templating:
template
Matrix vec2matrix(Vector x, int nrow, int ncol) {
return Matrix(nrow, ncol, x.begin());
}
will work for NumericVector, IntegerVector, CharacterVector, ...
Romain
Le 31/05/13 11:08, Søren Højsgaard a écrit :
Dear all,
Dirk: Thanks for t
Dear all,
Dirk: Thanks for the reply :).
I shall dare to ask a "bonus question": I've played around with templating so
that I get the right output type. I know (thanks to an earlier reply by Romain)
how to do so using SEXP's but if I am constructing a utility function to be
used on c++ object
20 matches
Mail list logo
