Tuning tempered transitions

Gundula Behrens, Nial Friel, Merrilee Hurn

Research output: Contribution to journalArticle

  • 12 Citations

Abstract

Abstract: The method of tempered transitions was proposed by Neal (Stat. Comput. 6:353–366, 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 Metropoliscoupled 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.
LanguageEnglish
Pages65-78
Number of pages14
JournalStatistics and Computing
Volume22
Issue number1
Early online date12 Oct 2010
DOIs
StatusPublished - 1 Jan 2012

Fingerprint

Tuning
Simulated Tempering
Tempering
Markov Chain Monte Carlo
Markov processes
Markov chain
State Space
Temperature
Series
Acceptance
Markov chain Monte Carlo
State space

Cite this

Tuning tempered transitions. / Behrens, Gundula; Friel, Nial; Hurn, Merrilee.

In: Statistics and Computing, Vol. 22, No. 1, 01.01.2012, p. 65-78.

Research output: Contribution to journalArticle

Behrens, G, Friel, N & Hurn, M 2012, 'Tuning tempered transitions' Statistics and Computing, vol. 22, no. 1, pp. 65-78. DOI: 10.1007/s11222-010-9206-z
Behrens G, Friel N, Hurn M. Tuning tempered transitions. Statistics and Computing. 2012 Jan 1;22(1):65-78. Available from, DOI: 10.1007/s11222-010-9206-z
Behrens, Gundula ; Friel, Nial ; Hurn, Merrilee. / Tuning tempered transitions. In: Statistics and Computing. 2012 ; Vol. 22, No. 1. pp. 65-78
@article{73771e10497a443a8defc445cecb897e,
title = "Tuning tempered transitions",
abstract = "Abstract: The method of tempered transitions was proposed by Neal (Stat. Comput. 6:353–366, 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 Metropoliscoupled 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.",
author = "Gundula Behrens and Nial Friel and Merrilee Hurn",
year = "2012",
month = "1",
day = "1",
doi = "10.1007/s11222-010-9206-z",
language = "English",
volume = "22",
pages = "65--78",
journal = "Statistics and Computing",
issn = "0960-3174",
publisher = "Springer Netherlands",
number = "1",

}

TY - JOUR

T1 - Tuning tempered transitions

AU - Behrens,Gundula

AU - Friel,Nial

AU - Hurn,Merrilee

PY - 2012/1/1

Y1 - 2012/1/1

N2 - Abstract: The method of tempered transitions was proposed by Neal (Stat. Comput. 6:353–366, 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 Metropoliscoupled 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.

AB - Abstract: The method of tempered transitions was proposed by Neal (Stat. Comput. 6:353–366, 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 Metropoliscoupled 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.

UR - http://www.scopus.com/inward/record.url?scp=81955163114&partnerID=8YFLogxK

UR - http://dx.doi.org/10.1007/s11222-010-9206-z

U2 - 10.1007/s11222-010-9206-z

DO - 10.1007/s11222-010-9206-z

M3 - Article

VL - 22

SP - 65

EP - 78

JO - Statistics and Computing

T2 - Statistics and Computing

JF - Statistics and Computing

SN - 0960-3174

IS - 1

ER -