RcppArmadillo 0.2.2

June 9, 2010

(This article was first published on dirk.eddelbuettel, and kindly contributed to R-bloggers)

Following the Rcpp 0.8.1 release we made yesterday, we released RcppArmadillo release 0.2.2 this morning. RcppArmadillo uses Rcpp (and some 'glue' code) to provide a transparent interface from R to Conrad Sanderson's impressive Armadillo library for linear algebra.

This release works well with the most recent inline release 0.3.5. One can now employ inlined R code as we generalized how/which headers are included and how library / linking information is added thanks a plugin mechanism. This is the first RcppArmadillo version to provide such a plugin, We also updated the included Armadillo headers to its most recent release 0.9.10, added some more operators and provide a utility function RcppArmadillo:::CxxFlags() to provide include directory information on the fly.

An example of the direct inline approach for the fastLm function:

library(RcppArmadillo) src <- '
	Rcpp::NumericVector yr(ys); // creates Rcpp vector from SEXP
	Rcpp::NumericMatrix Xr(Xs); // creates Rcpp matrix from SEXP
	int n = Xr.nrow(), k = Xr.ncol();

	arma::mat X(Xr.begin(), n, k, false); // reuses memory and avoids extra copy
	arma::colvec y(yr.begin(), yr.size(), false);

	arma::colvec coef = arma::solve(X, y); // fit model y ~ X
	arma::colvec res = y - X*coef; // residuals

	double s2 = std::inner_product(res.begin(), res.end(), res.begin(), double())/(n - k);
 // std.errors of coefficients
	arma::colvec std_err = arma::sqrt(s2 * arma::diagvec( arma::inv(arma::trans(X)*X) ));

	return Rcpp::List::create(Rcpp::Named("coefficients") = coef,
 Rcpp::Named("stderr") = std_err,
 Rcpp::Named("df") = n - k
' fun <- cxxfunction(signature(ys="numeric", Xs="numeric"), src, plugin="RcppArmadillo")

This creates a compiled function fun which, by using Armadillo, regresses a vector ys on a matrix Xs (just how the fastLmPure() function in the package does) --- yet is constructed on the fly using cxxfunction from inline.

More information is on the RcppArmadillo page. Questions, comments etc should go to the rcpp-devel mailing list off the R-Forge page.

To leave a comment for the author, please follow the link and comment on his blog: dirk.eddelbuettel.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...

If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.