Tuning Tempered Transitions
Preprint
Abstract
The method of tempered transitions was proposed by Neal (1996) for tackling the difficulties arising when using Markov chain Monte Carlo to sample from multimodal distributions. In common with methods such as simulated tempering and Metropolis-coupled MCMC, the key idea is to utilise a series of successively easier to sample distributions to improve movement around the state space. Tempered transitions does this by incorporating moves through these less modal distributions into the MCMC proposals. Unfortunately the improved movement between modes comes at a high computational cost with a low acceptance rate of expensive proposals. We consider how the algorithm may be tuned to increase the acceptance rates for a given number of temperatures. We find that the commonly assumed geometric spacing of temperatures is reasonable in many but not all applications.Keywords
All Related Versions
- Version 1, 2010-10-05, ArXiv
- Published version: Statistics and Computing, 22 (1), 65.
This publication has 13 references indexed in Scilit:
- A path sampling identity for computing the Kullback–Leibler and J divergencesComputational Statistics & Data Analysis, 2010
- Estimating Bayes factors via thermodynamic integration and population MCMCComputational Statistics & Data Analysis, 2009
- Importance temperingStatistics and Computing, 2008
- Marginal Likelihood Estimation via Power PosteriorsJournal of the Royal Statistical Society Series B: Statistical Methodology, 2008
- On population-based simulation for static inferenceStatistics and Computing, 2007
- EXTENDED ENSEMBLE MONTE CARLOInternational Journal of Modern Physics C, 2001
- Annealed importance samplingStatistics and Computing, 2001
- On Bayesian Analysis of Mixtures with an Unknown Number of Components (with discussion)Journal of the Royal Statistical Society Series B: Statistical Methodology, 1997
- Sampling from multimodal distributions using tempered transitionsStatistics and Computing, 1996
- A slowly mixing Markov chain with implications for Gibbs samplingStatistics & Probability Letters, 1993