Some Techniques for Testing Pseudo-Random Number Sequences.

Abstract

Statistical tests are presented which may be applied to any type of pseudo-random number generator to ascertain whether its statistics have certain desired properties. Possible pitfalls are discussed, along with a computer program to implement these tests. The program was applied to several types of uniformly-distributed generators, including the mixed-congruence, the exclusive-or-and-shift, and the CDC RANF generator. Each of these was found to exhibit statistical irregularities, indicating that care is necessary in the selection of a generator. A second computer program is described which tests the correlation between two pseudo-random number sequences. A practical example of its application provides insight into the generation of multiple sequences with a single generator. (Author)

Document Details

Document Type
Technical Report
Publication Date
Aug 21, 1974
Accession Number
AD0785379

Entities

People

  • James L. Mannos

Organizations

  • Massachusetts Institute of Technology

Tags

DTIC Thesaurus Topics

  • Computer Programs
  • Computers
  • Computing-Related Activities
  • Data Science
  • Demographic Cohorts
  • Generators
  • Information Science
  • Interdisciplinary Science
  • Mathematical Analysis
  • Random Number Generators
  • Sequences
  • Statistical Analysis
  • Statistical Tests
  • Statistics

Fields of Study

  • Mathematics

Readers

  • Computer Programming and Software Development.
  • Systems Analysis and Design