Description
Abstract: The npRmpi library written by Tristen Hayfield and Jeffrey Racine is a collection of functions that perform a variety of nonparametric and semiparametric estimation techniques in the statistical software environment R. Specifically, npRmpi combines the R packages Rmpi (Yu, 2002) with np (Hayfield and Racine, 2008) and therefore it is a wrapper to the Message Passing Interface standard (MPI) providing an interface within R to low-level MPI functions that allows parallelization of various np native functions without placing a major programming burden on the researcher.1 In this way, the practitioner can use np commands without the knowledge of MPI implementation details. A major advantage of the package is its ability to be deployed on personal dual-core or quad-core laptop or desktop machines running any of the commonly used operating systems such as Windows, Linux, or Mac OS X in either interactive or in batch mode. This package will be of interest to applied econometricians who previously found the computational burden of applying kernel methods to be too costly to justify their use.