I am trying to build an R package using mlpack. As suggested in this link I am using the following cpp function
#include <Rcpp/Rcpp>
#include <mlpack.h>
// Two include directories adjusted for my use of mlpack 3.4.2 on Ubuntu
#include <mlpack/core.hpp>
#include <mlpack/methods/kmeans/kmeans.hpp>
#include <mlpack/methods/kmeans/random_partition.hpp>
#include <mlpack/methods/neighbor_search/neighbor_search.hpp>
// [[Rcpp::depends(RcppArmadillo)]]
// [[Rcpp::depends(mlpack)]]
// This is 'borrowed' from mlpack's own src/mlpack/tests/kmeans_test.cpp
// and src/mlpack/tests/kmeans_test.cpp. We borrow the data set, and the
// code from the first test function. Passing data from R in easy thanks
// to RcppArmadillo, 'and left as an exercise'.
// Generate dataset; written transposed because it's easier to read.
arma::mat kMeansData(" 0.0 0.0;" // Class 1.
" 0.3 0.4;"
" 0.1 0.0;"
" 0.1 0.3;"
" -0.2 -0.2;"
" -0.1 0.3;"
" -0.4 0.1;"
" 0.2 -0.1;"
" 0.3 0.0;"
" -0.3 -0.3;"
" 0.1 -0.1;"
" 0.2 -0.3;"
" -0.3 0.2;"
" 10.0 10.0;" // Class 2.
" 10.1 9.9;"
" 9.9 10.0;"
" 10.2 9.7;"
" 10.2 9.8;"
" 9.7 10.3;"
" 9.9 10.1;"
"-10.0 5.0;" // Class 3.
" -9.8 5.1;"
" -9.9 4.9;"
"-10.0 4.9;"
"-10.2 5.2;"
"-10.1 5.1;"
"-10.3 5.3;"
"-10.0 4.8;"
" -9.6 5.0;"
" -9.8 5.1;");
// [[Rcpp::export]]
arma::Row<size_t> kmeansDemo() {
mlpack::kmeans::KMeans<mlpack::metric::EuclideanDistance,
mlpack::kmeans::RandomPartition> kmeans;
arma::Row<size_t> assignments;
kmeans.Cluster((arma::mat) trans(kMeansData), 3, assignments);
return assignments;
}
If I sourceCpp the above in Ubuntu linux Sys.setenv("PKG_LIBS"="-lmlpack") then it compiles successfully. However, I am unable to use it on macOS with Apple M2 architecture. I am getting the following error in macOS
/Library/Frameworks/R.framework/Versions/4.3-arm64/Resources/library/mlpack/include/mlpack.h:52:10: fatal error: mlpack/core.hpp: No such file or directory
52 | #include <mlpack/core.hpp>
| ^~~~~~~~~~~~~~~~~
compilation terminated.
I have installed mlpack R package installed as well as the system mlpack using brew. Seems to me that R cannot link to the mlpack libraries that are located in /opt/homebrew/include/ in my system. Is there a way to link to these? I have tried brew link mlpack which shows linking is successful but still got the same compilation error. Additionally I tried the following in R before sourceCpping but no luck!
Sys.setenv("LDFLAGS"="-L/opt/homebrew/lib")
Sys.setenv("CPPFLAGS"="-I/opt/homebrew/include")
Sys.setenv("PKG_LIBS"="-lmlpack")
Please let me know if there is any way out for this in macOS.
P.S. Both R and Rstudio are installed in my system using brew.
mlpack 4.2.0 is now on CRAN and ships exported headers we can use! A minimally modified version of your example follows.
#include <Rcpp/Rcpp>
#include <mlpack.h>
#include <mlpack/methods/kmeans.hpp>
// -- use C++17
// [[Rcpp::plugins(cpp17)]]
// -- use Armadillo, Ensmallen and mlpack headers
// [[Rcpp::depends(RcppArmadillo)]]
// [[Rcpp::depends(RcppEnsmallen)]]
// [[Rcpp::depends(mlpack)]]
// This is 'borrowed' from mlpack's own src/mlpack/tests/kmeans_test.cpp
// Generate dataset; written transposed because it's easier to read.
arma::mat kMeansData(" 0.0 0.0;" // Class 1.
" 0.3 0.4;"
" 0.1 0.0;"
" 0.1 0.3;"
" -0.2 -0.2;"
" -0.1 0.3;"
" -0.4 0.1;"
" 0.2 -0.1;"
" 0.3 0.0;"
" -0.3 -0.3;"
" 0.1 -0.1;"
" 0.2 -0.3;"
" -0.3 0.2;"
" 10.0 10.0;" // Class 2.
" 10.1 9.9;"
" 9.9 10.0;"
" 10.2 9.7;"
" 10.2 9.8;"
" 9.7 10.3;"
" 9.9 10.1;"
"-10.0 5.0;" // Class 3.
" -9.8 5.1;"
" -9.9 4.9;"
"-10.0 4.9;"
"-10.2 5.2;"
"-10.1 5.1;"
"-10.3 5.3;"
"-10.0 4.8;"
" -9.6 5.0;"
" -9.8 5.1;");
// [[Rcpp::export]]
arma::Row<size_t> kmeansDemo() {
// Originally written to use RandomPartition, and is left that
// way because RandomPartition gives better initializations here.
mlpack::KMeans<mlpack::EuclideanDistance, mlpack::RandomPartition> kmeans;
// mlpack::KMeans<> kmeans; // default arguments as an alternative
arma::Row<size_t> assignments;
kmeans.Cluster((arma::mat) trans(kMeansData), 3, assignments);
return assignments;
}
/*** R
kmeansDemo()
*/
> Rcpp::sourceCpp("~/git/stackoverflow/76336745/answer.cpp")
> kmeansDemo()
[INFO ] KMeans::Cluster(): iteration 1, residual 13.7285.
[INFO ] KMeans::Cluster(): iteration 2, residual 2.51215e-15.
[INFO ] KMeans::Cluster(): converged after 2 iterations.
[INFO ] 186 distance calculations.
[,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13] [,14] [,15] [,16] [,17]
[1,] 2 2 2 2 2 2 2 2 2 2 2 2 2 0 0 0 0
[,18] [,19] [,20] [,21] [,22] [,23] [,24] [,25] [,26] [,27] [,28] [,29] [,30]
[1,] 0 0 0 1 1 1 1 1 1 1 1 1 1
>
> sapply(c("RcppArmadillo", "RcppEnsmallen", "mlpack"), \(x) format(packageVersion(x)))
RcppArmadillo RcppEnsmallen mlpack
"0.12.4.1.0" "0.2.19.0.1" "4.2.0"
>