ADAPTIVE GIBBS SAMPLERS AND RELATED MCMC METHODS

• 成果类型: Article
• 署名作者: Latuszynski, Krzysztof; Roberts, Gareth O.; Rosenthal, Jeffrey S.
• 署名单位: University of Warwick; University of Toronto
• 刊物名称: ANNALS OF APPLIED PROBABILITY
• ISSN/ISSBN: 1050-5164
• DOI: 10.1214/11-AAP806
• 发表日期: 2013
• 页码: 66-98
• 关键词: chain monte-carlo; geometric ergodicity; metropolis algorithms; WEAK-CONVERGENCE; STABILITY; hastings; rates

摘要

We consider various versions of adaptive Gibbs and Metropolis-within-Gibbs samplers, which update their selection probabilities (and perhaps also their proposal distributions) on the fly during a run by learning as they go in an attempt to optimize the algorithm. We present a cautionary example of how even a simple-seeming adaptive Gibbs sampler may fail to converge. We then present various positive results guaranteeing convergence of adaptive Gibbs samplers under certain conditions.