Generation of pseudorandom numbers.

Computational studies requiring the generation of pseudorandom numbers are becoming increasingly common. These include Monte Carlo methodologies and studies which require the addition of "random" noise to more structured data. Although well-established random number generators exist, many of these are not suitable for implementation on micro- or minicomputer systems. An algorithm for an additive generator that can be used on any computer system is described. The performance of this algorithm as implemented on a microcomputer system is discussed. Extensive testing of the statistical behavior of the resulting number sequence was performed. The algorithm appears to be an appropriate one for a variety of applications.