Coupling control variates for Markov chain Monte Carlo

Jonathan Goodman, Kevin K. Lin

Research output: Contribution to journalArticle

Abstract

We show that Markov couplings can be used to improve the accuracy of Markov chain Monte Carlo calculations in some situations where the steady-state probability distribution is not explicitly known. The technique generalizes the notion of control variates from classical Monte Carlo integration. We illustrate it using two models of nonequilibrium transport.

Original languageEnglish (US)
Pages (from-to)7127-7136
Number of pages10
JournalJournal of Computational Physics
Volume228
Issue number19
DOIs
StatePublished - Oct 20 2009

Fingerprint

Markov chains
Markov processes
Probability distributions

Keywords

  • Control variates
  • Coupling
  • Markov chain Monte Carlo
  • Monte Carlo
  • Nonequilibrium statistical mechanics
  • Variance reduction

ASJC Scopus subject areas

  • Computer Science Applications
  • Physics and Astronomy (miscellaneous)

Cite this

Coupling control variates for Markov chain Monte Carlo. / Goodman, Jonathan; Lin, Kevin K.

In: Journal of Computational Physics, Vol. 228, No. 19, 20.10.2009, p. 7127-7136.

Research output: Contribution to journalArticle

Goodman, Jonathan ; Lin, Kevin K. / Coupling control variates for Markov chain Monte Carlo. In: Journal of Computational Physics. 2009 ; Vol. 228, No. 19. pp. 7127-7136.
@article{a3d17ff2827443e2a68df8593141677a,
title = "Coupling control variates for Markov chain Monte Carlo",
abstract = "We show that Markov couplings can be used to improve the accuracy of Markov chain Monte Carlo calculations in some situations where the steady-state probability distribution is not explicitly known. The technique generalizes the notion of control variates from classical Monte Carlo integration. We illustrate it using two models of nonequilibrium transport.",
keywords = "Control variates, Coupling, Markov chain Monte Carlo, Monte Carlo, Nonequilibrium statistical mechanics, Variance reduction",
author = "Jonathan Goodman and Lin, {Kevin K.}",
year = "2009",
month = "10",
day = "20",
doi = "10.1016/j.jcp.2009.03.043",
language = "English (US)",
volume = "228",
pages = "7127--7136",
journal = "Journal of Computational Physics",
issn = "0021-9991",
publisher = "Academic Press Inc.",
number = "19",

}

TY - JOUR

T1 - Coupling control variates for Markov chain Monte Carlo

AU - Goodman, Jonathan

AU - Lin, Kevin K.

PY - 2009/10/20

Y1 - 2009/10/20

N2 - We show that Markov couplings can be used to improve the accuracy of Markov chain Monte Carlo calculations in some situations where the steady-state probability distribution is not explicitly known. The technique generalizes the notion of control variates from classical Monte Carlo integration. We illustrate it using two models of nonequilibrium transport.

AB - We show that Markov couplings can be used to improve the accuracy of Markov chain Monte Carlo calculations in some situations where the steady-state probability distribution is not explicitly known. The technique generalizes the notion of control variates from classical Monte Carlo integration. We illustrate it using two models of nonequilibrium transport.

KW - Control variates

KW - Coupling

KW - Markov chain Monte Carlo

KW - Monte Carlo

KW - Nonequilibrium statistical mechanics

KW - Variance reduction

UR - http://www.scopus.com/inward/record.url?scp=68549136477&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=68549136477&partnerID=8YFLogxK

U2 - 10.1016/j.jcp.2009.03.043

DO - 10.1016/j.jcp.2009.03.043

M3 - Article

VL - 228

SP - 7127

EP - 7136

JO - Journal of Computational Physics

JF - Journal of Computational Physics

SN - 0021-9991

IS - 19

ER -