STL transform + remove_copy for subsetting

December 29, 2012
By

(This article was first published on Rcpp Gallery, and kindly contributed to R-bloggers)

We have seen the use of the STL transform functions in the posts STL transform and Transforming a matrix. We use the same logic in conjuction with a logical (ie boolean) vector in order subset an initial vector.

#include <Rcpp.h>

using namespace Rcpp;
using namespace std;

const double flagval = __DBL_MIN__; // works
//const double flagval = NA_REAL;   // does not

// simple double value 'flagging' function
inline double flag(double a, bool b) { return b ? a : flagval; }

// [[Rcpp::export]]
NumericVector subsetter(NumericVector a, LogicalVector b) {
    // We use the flag() function to mark values of 'a' 
    // for which 'b' is false with the 'flagval'
    transform(a.begin(), a.end(), b.begin(), a.begin(), flag);

    // We use sugar's sum to compute how many true values to expect
    NumericVector res = NumericVector(sum(b));

    // And then copy the ones different from flagval from a into
    // res using the remove_copy function from the STL
    remove_copy(a.begin(), a.end(), res.begin(), flagval);
    return res;    
}

We can illustrate this on a simple example or two:

a <- 1:5
subsetter(a, a %% 2 == 0)
[1] 2 4
subsetter(a, a > 2)
[1] 3 4 5

Casual benchmarking (not shown) shows this to be comparable to and even slightly faster than basic indexing in R itself.

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

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.