Calling R Functions from C++

Dirk Eddelbuettel — written Jan 5, 2013 — source

At its very essence, Rcpp permits easy access to native R objects at the C++ level. R objects can be

simple vectors, list or matrices;
compound data structures created from these;
objects of S3, S4 or Reference Class vintage; or
language objects as for example environments.

Accessing a function object is no different. And calling a function can be very useful. Maybe to pick up parameter initializations, maybe to access a custom data summary that would be tedious to recode, or maybe even calling a plotting routine. We already have examples for just about all of these use case in the Rcpp examples or unit tests shipping with the package.

So here were a just providing a simple example of calling a summary function, namely the Tukey fivenum().

But before we proceed, a warning. Calling a function is simple and tempting. It is also slow as there are overheads involved. And calling it repeatedly from inside your C++ code, possibly buried within several loops, is outright silly. This has to be slower than equivalent C++ code, and even slower than just the R code (because of the marshalling of data). Do it when it makes sense, and not simply because it is available.

set.seed(42)
x <- rnorm(1e5)
fivenum(x)

[1] -4.043276 -0.682384 -0.002066  0.673325  4.328091

Now via this C++ code:

#include <Rcpp.h>

using namespace Rcpp;

// [[Rcpp::export]]
NumericVector callFunction(NumericVector x, Function f) {
    NumericVector res = f(x);
    return res;
}

And unsurprisingly, calling the same function on the same data gets the same result:

callFunction(x, fivenum)

[1] -4.043276 -0.682384 -0.002066  0.673325  4.328091

tags: basics function

Extending R with C++ and Fortran — Dirk Eddelbuettel and JBrandon Duck-Mayr
Benchmarking Rcpp code with RcppClock — Zach DeBruine
Simulation Smoother using RcppArmadillo — Tomasz Woźniak