Thoughts on pseudorandom number generators

From MaRDI portal





The author of this survey paper summarizes the basic requirements on ``good random number generators. He points out the necessity of the period greater than \(2^{50}\). Then he discusses the properties of seven commonly used congruential generators, the principles of (and problems of) shift-register generators, especially the initialization and the problem of shuffling. He does not mention the shuffling scheme when the members from the table are reintroduced into the generator. Timings for some computers (PC and workstations) and generators are given. The author concludes that ``for all but the most demanding problems the Marsaglia generator suffices. This statement is not, however, generally accepted. The paper contains the FORTRAN program that initializes and runs a shift-register generator with the period \(2^{521}-1\).




Cited in
(28)








This page was built for publication: Thoughts on pseudorandom number generators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q914322)