Quality evaluation of some combinations of unit uniform random number generators and unit normal transformation algorithms

Abstract

This study is concerned with evaluation of the goodness of fit and randomness of unit normal variates generated from combinations of several unit uniform generators and several normal transformation algorithms. The basic conclusion is that use of a good, theoretically exact transformation in conjunction with a theoretically good unit uniform generator does not necessarily guarantee variates with good statistical properties. In many cases generators which aren't quite so good when combined with transformation algorithms which also aren't quite so good, will yield variates with statistical properties which are superior to those of the “better” generators.