THE MOVE-TO-FRONT RULE - A CASE-STUDY FOR 2 PERFECT SAMPLING ALGORITHMS

Authors
FILL JA
Citation
Ja. Fill, THE MOVE-TO-FRONT RULE - A CASE-STUDY FOR 2 PERFECT SAMPLING ALGORITHMS, Probability in the engineering and informational sciences, 12(3), 1998, pp. 283-302
Citations number
21
Categorie Soggetti
Statistic & Probability","Operatione Research & Management Science","Engineering, Industrial","Statistic & Probability","Operatione Research & Management Science
Journal title
Probability in the engineering and informational sciences → ACNP
ISSN journal
02699648
Volume
12
Issue
3
Year of publication
1998
Pages
283 - 302
Database
ISI
SICI code
0269-9648(1998)12:3<283:TMR-AC>2.0.ZU;2-6
Abstract
The elementary problem of exhaustively sampling a finite population wi thout replacement is used as a nonreversible test case for comparing t wo recently proposed MCMC algorithms for perfect sampling, one based o n backward coupling and the other on strong stationary duality. The ba ckward coupling algorithm runs faster in this case, but the duality-ba sed algorithm is unbiased for user impatience. An interesting by-produ ct of the analysis is a new and simple stochastic interpretation of a mixing-time result for the move-to-front rule.