Try this:
library(RcppArmadillo)
library(inline)
if( require( RcppArmadillo ) ){
fx <- cxxfunction( signature(x = "vector", n = "integer") , '
Rcpp::NumericVector XX(x);
Rcpp::IntegerVector dim(n);
arma::cube AY(XX.begin(), dim[0], dim[1], dim[2]);
return(wrap(AY));
', plugin = "RcppArmadillo" )
}
set.seed(10)
x = array(rnorm(25*25*10), dim=c(25,25,10))
y = fx(x, c(25,25,10))
all.equal(x,y)
Regards,
Alexios
On 06/03/2013 01:15, Dirk Eddelbuettel wrote:
On 6 March 2013 at 00:50, Silkworth,David J. wrote:
| I am helping a student build some R code in Rcpp for his first time. He is
| working with 3-dimensional arrays, so I am hopeful to be working with
| arma::cube objects.
|
| When I simply try to bring the array in as an argument I can?t seem to use the
| same as<[TEMPLATE]> construction as other objects.
|
| arma::cube X=Rcpp::as<arma::cube>(arg1);
|
| This line fails to compile.
Yes. Nobody wrote such a converter. Maybe you and your student want to take a
stab?
| I can build this cube from an input vector using explicit nested loops. I
also
| notice that wrap() does a fine job of returning the array back to R.
|
| Is there anything else I need to know to more simply bring the array in as an
| SEXP and land it in the arma::cube.
Array plus dim attribute may work.
I used arma::cube quite extensively in one project, but only at the C++ level
as a container. Results back to R were always of lesser dimension.
Dirk
_______________________________________________
Rcpp-devel mailing list
[email protected]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
