Premium
Analytic Comparison of Three General‐Purpose Methods for the Computer Generation of Discrete Random Variables
Author(s) -
Peterson Arthur V.,
Kronmal Richard A.
Publication year - 1983
Publication title -
journal of the royal statistical society: series c (applied statistics)
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.205
H-Index - 72
eISSN - 1467-9876
pISSN - 0035-9254
DOI - 10.2307/2347950
Subject(s) - statistics , mathematics , computer science , calculus (dental) , medicine , dentistry
SUMMARY Three general‐purpose methods for generating large numbers of discrete random variables are compared: the alias method, the indexed‐search method, and the Marsaglia table method. Unlike previous comparisons, analytic comparisons are emphasized. Results indicate that for any distribution the alias method has fewer operations and thus is faster than the indexed search method. Conclusions about relative speed of the table method, however, are highly dependent on the distribution. For speed and accuracy, the alias method is recommended as a general‐purpose method for generating large numbers of discrete random variables. Timing results are also presented and discussed, and compared with others in the recent literature.