The generation of random numbers from various probability distributions.

Abstract

Methods are developed, and Fortran 63 CODAP computer programs are demonstrated, to generate random numbers from the uniform, normal (including multivariate normal), Poisson, and exponential probability distributions. Various statistical tests are described and the results of the application of these tests to the generators are tabulated. A general method for generating random numbers from a large class of distributions is described. The methods of generation are optimized to provide an accurate generator while producing numbers at a maximum rate. The uniform generator that is used as a basis for the other generators is of the congruential type and is capable of generating 1800 numbers per second.