randu and randn hang on 0.6.5 about rcpparmadillo HOT 2 CLOSED

Ok, a few things:

1. Try formatting your posts. It makes it easier for other people to read your posts, and hence more likely they will help you. I formatted it for you.
2. You were using the older inline approach. Rcpp Attributes is nicer. A new version is below.
3. RcppArmadillo is used from R. R has very good and extremely well-tested statistical algorithms. I would use R's RNG. In fact, that is what we do here as I recall...
4. For me it all works.

New file:

#include <RcppArmadillo.h>

// [[Rcpp::depends(RcppArmadillo)]]

// [[Rcpp::export]]
arma::mat foo(int n) {
    arma::mat f = arma::join_rows(arma::randu(n),  
                                  arma::randn(n));
    return f;
}

/*** R
set.seed(123)
foo(5)
*/

which we simple feed to sourceCpp() which will also run the code:

R> set.seed(123)

R> foo(5)
          [,1]        [,2]
[1,] 0.2875775 -0.61079930
[2,] 0.7883051  0.03777546
[3,] 0.4089769  0.95877582
[4,] 0.8830174  0.12566834
[5,] 0.9404673 -0.05538708
R> 

Best of all, the seed is linked to R as we (as I recall) use R's RNG by default. Calling again gives different numbers; resetting the seed gives (as expected) what we just got:

R> foo(5)
          [,1]       [,2]
[1,] 0.4533342 -0.1193901
[2,] 0.6775706  0.3153385
[3,] 0.5726334  0.6708789
[4,] 0.1029247  0.3320531
[5,] 0.8998250  0.9669271
R> set.seed(123)
R> foo(5)
          [,1]        [,2]
[1,] 0.2875775 -0.61079930
[2,] 0.7883051  0.03777546
[3,] 0.4089769  0.95877582
[4,] 0.8830174  0.12566834
[5,] 0.9404673 -0.05538708
R> 

Strictly no issue here. I suspect that the cause may have been your use of cxxfunction() AND the random generator WITHOUT setting RNGState (see Writing R Extensions). The Rcpp Attributes approach does that automatically for you, and I no longer recall if we did that for inline too.

from rcpparmadillo.

Superb! Thanks for the detailed correction and broad lesson. This now works on RcppArmadillo 0.6.5.

from rcpparmadillo.