Recent trends in random number and random vector generation (Q1176851): Difference between revisions

This is an expository paper on recent work concerning the generation of uniformly distributed pseudorandom numbers and vectors. All the methods mentioned in the paper are clearly exposed and illustrated and often compared each other so that the reader can recover an exhaustive knowledge of the matter. The distribution uniformity is generally evaluated by considering the asymptotic behaviour of discrepancy. The most relevant methods mentioned are the following: the linear congruential method which goes back to \textit{D. H. Lehmer} [Proc. 2nd Symposium Large-Scale Digital Calculating Machines, 141--146 (1951; Zbl 0045.40001)]\ but is also object of recent studies, e.g. by \textit{G. S. Fishman} and \textit{L. R. Moore III} [SIAM J. Sci. Stat. Comput. 7, 24--45 (1986; Zbl 0603.65003); Erratum: ibid. 7, 1058 (1986; Zbl 0615.65004)]; the nonlinear congruential method, introduced by \textit{J. Eichenauer}, \textit{H. Grothe}, and \textit{J. Lehn} [Metrika 35, No. 3--4, 241--250 (1988; Zbl 0653.65006)]; the inversive congruential method, introduced by \textit{J. Eichenauer}, \textit{H. Grothe} and \textit{J. Lehn} [Stat. Hefte 27, 315--326 (1986; Zbl 0607.65001)]; the shift register methods, whose first diffusion goes back to \textit{R. C. Tausworthe} [Math. Comput. 19, 201--209 (1965; Zbl 0137.34804)], but widely studied also in recent times. Two final chapters concern the generation of pseudorandom vectors (utilizing the methods of matrix generators) and the generation of quasirandom vectors. In the references eighty five papers are quoted.