Multi-dimensional uniformity of portable pseudo-random number generators

Abstract

Multipliers used with portable multiplicative pseudorandom number generators for 16-b computers are considered. Certain multipliers that are able to produce uniform random variables with good one-dimensional statistical properties have already been found. The tests implemented were the chi-square goodness-of-fit test and the Kolmogorov-Smirnov and modified Kolmogorov-Smirnov tests. These tests have been adapted for the multidimensional case in order for statistical tests to be performed on two- and three-dimensional uniform random vectors generated by these same multipliers. These tests show statistical independence between samples, as well as uniformity.<>