In the previous chapter we have seen how R can be easily interfaced with C++ via the Rcpp package. Here we will explore the capabilities of Rcpp more in details. In particular, we will show how Rcpp sugar allows us to write C++ code that looks like R code, but that is often much faster. We will then consider the RcppArmadillo package, which is an interface to the Armadillo C++ linear algebra library, and we will demostrate that it can lead to superior performance, relative to base R, in the context of statistical computing.

The chapter consists of the following sections:

• 1. Rcpp sugar
• 2. RcppArmadillo
• 3. Example on multivariate KDE with RcppArmadillo
• 4. Exercise on local polynomial regression