A Tutorial for Using 'Rmpi' on the NCAR/Wyoming Supercomputer

The recent growth in the size and complexity of climate data as well as the availability of cluster and supercomputing resources has brought parallel programming to the attention of geostatisticans. The National Center for Atmospheric Research’s (NCAR) supercomputer facility is an accessible and powerful resource for any scientist or statistician working in climate-related fields. This technical report presents a tutorial for users of the statistical language R to quickly utilize the massively parallel computational resources provided by the NCAR Yellowstone supercomputer through the Rmpi package. We provide fully functional scripts on the Yellowstone system that serve as examples of flexible implementations of the capabilities of massively parallel R scripts. Additionally, the tutorial serves as a guide for running parallel R scripts on any cluster or supercomputer with MPI capabilities.