Perfect forward simulation via simulated tempering

被引：4

作者：

Brooks, Stephen P.

Fan, Yanan

Rosenthal, Jeffrey S.

机构：

[1] Univ Cambridge, Stat Lab, Cambridge CB3 0WB, England

[2] Univ New S Wales, Dept Math & Stat, Kensington, NSW 2033, Australia

[3] Univ Toronto, Dept Math, Toronto, ON M5S 1A1, Canada

来源：

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION | 2006年 / 35卷 / 03期

关键词：

autoregressive time series; band-return data; coupling from the past; regeneration; reversible jump MCMC; small sets;

D O I：

10.1080/03610910600716753

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

Several authors discuss how the simulated tempering scheme provides a very simple mechanism for introducing regenerations within a Markov chain. In this article we explain how regenerative simulated tempering schemes provide a very natural mechanism for perfect simulation. We use this to provide a perfect simulation algorithm, which uses a single-sweep forward-simulation without the need for recursively searching through negative times. We demonstrate this algorithm in the context of several examples.

引用

页码：683 / 713

页数：31

共 51 条

[1]

[Anonymous], 1993, ANN APPL PROBAB, DOI [10.1214/aoap/1177005366, DOI 10.1214/AOAP/1177005366]

[2]

BREYER LA, 2000, CATALYTIC PERFECT SI

[3]

BROCKWELL A, 2002, PRACTICAL REGENERATI

[4] Convergence assessment techniques for Markov chain Monte Carlo [J].