alumbreras / nbmf Goto Github PK

Hello! Great package and nice paper, congratulations.

I think the current way of loading your package (from your example scripts) is to use devtools::load_all(). Is it possible to do any of the following?

1. Remove the object files (*.o) which is OS-specific? I had to delete them, and use R CMD INSTALL to recompile the cpp files and install the package.
2. Make a tar gz source file so the package can be installed from there?

Or preferably, list installation instructions for the package? e.g. using devtools::install_github()

Thanks!

Hello,

I'm getting a "Cube::init(): requested size is too large" error when using BernoulliNMF. I have tried rebuilding Rcpp and RcppArmadillo with the ARMA_64BIT_WORD feature enabled which I did by adding

PKG_CPPFLAGS = -DARMA_64BIT_WORD=1

to my Makevars file and then re-building both Rcpp and RcppArmadillo. I then reinstalled

devtools::install_github('alumbreras/NBMF')

after rebuilding Rcpp and RcppArmadillo. Unfortunately, I ended up getting the same error message.

I'm not the most experienced R user, but based on what searching I did it sounds like the code is having trouble allocating a large matrix? For what it's worth, The input dataset is a 11451 x 971 matrix. Is this too large of a matrix to use in BernoulliNMF? Is there a way around this problem?

Thanks!