Regenerative Markov Chain Importance Sampling. Issue 5 (28th May 2017)

• Record Type: Journal Article
• Authors: Nguyen, Andrew L.
• Is Part Of: Communications in statistics. Volume 46:Issue 5(2017)
• Page Start: 3892
• Page End: 3906
• Publication Date: 2017-05-28
• Subjects: Importance Sampling -- MCMC -- Regenerative; Stochastic Processes; Mathematical statistics -- Periodicals; Mathematical statistics -- Data processing -- Periodicals; Digital computer simulation -- Periodicals; 519.5
• DOI: 10.1080/03610918.2015.1043383
• Languages: English
• ISSNs: 0361-0918

Abstract

ABSTRACT: We introduce Markov Chain Importance Sampling (MCIS), which combines importance sampling (IS) and Markov Chain Monte Carlo (MCMC) to estimate some characteristics of a non-normalized multi-dimensional distribution. Especially, we introduce some importance functions whose variates are regeneratively generated by MCMC; these variates then are used to estimate the quantity of interest through IS. Because MCIS is regenerative, it overcomes the burn-in problem associated with MCMC. It could also speed up the mixing rate in MCMC.