Parallel Computing: Statistical and Environmetric Uses†‡

The common use of parallel computing has greatly evolved since the original encyclopedia article of 2001. Multicore processors are now quite common, so many computer users have this readily available on their desktop computers and laptops. Now increasingly large datasets and simulation of complex statistical models are important in the study of many physical systems. Models in various areas, such as environment, biology, and other physical sciences, play an important role in prediction or detection of changes. All these require lots of computing power. Many of these computations can take advantage of parallel or distributed computing. This article discusses some of these ideas and then discusses how these are implemented in one specific language, R. In the present time, one generally no longer has to work at a low-level programming language, as was the case a decade or two ago, but now certain types of parallel computations can be implemented at a relatively higher user-friendly level, even with desktop computing. Parallel computing consists of a computing environment connecting many processors. Instead of the previous generation where dedicated computer architecture was required, a more loose structure of distributed computing is now more common. This article is intended to give the reader an overview of the parallel computing environment, focusing on the statistical uses that can be made as opposed to a more detailed computing or engineering description.