Assessment of MCMC convergence: A time series and dynamical systems approach

Rodney C. Wolff; Darfiana Nur; Kerrie L. Mengersen

Assessment of MCMC convergence: A time series and dynamical systems approach

Rodney C. Wolff, Darfiana Nur, Kerrie L. Mengersen

Research output: Contribution to conference › Conference presentation/ephemera › peer-review

Abstract

Important in the application of Markov chain Monte Carlo (MCMC) methods is the determination that a search run has converged. Given that such searches typically take place in high-dimensional spaces, there are many pitfalls and difficulties in making such assessments. In the present paper, we discuss the use of phase randomisation as tool in the MCMC context, provide some details of its distributional properties for time series which enable its use as a convergence diagnostic, and contrast its performance with a selection of other widely used diagnostics. Some brief comments on analytical results, obtained via Edgeworth expansion, are also made.

Original language	English
Pages	46-49
Number of pages	4
Publication status	Published - 2001
Externally published	Yes
Event	2001 IEEE Workshop on Statitical Signal Processing Proceedings - Singapore, Singapore Duration: 6 Aug 2001 → 8 Aug 2001

Conference

Conference	2001 IEEE Workshop on Statitical Signal Processing Proceedings
Country/Territory	Singapore
City	Singapore
Period	6/08/01 → 8/08/01

Cite this

@conference{af853fa1935c43269c97faab405b6322,

title = "Assessment of MCMC convergence: A time series and dynamical systems approach",

abstract = "Important in the application of Markov chain Monte Carlo (MCMC) methods is the determination that a search run has converged. Given that such searches typically take place in high-dimensional spaces, there are many pitfalls and difficulties in making such assessments. In the present paper, we discuss the use of phase randomisation as tool in the MCMC context, provide some details of its distributional properties for time series which enable its use as a convergence diagnostic, and contrast its performance with a selection of other widely used diagnostics. Some brief comments on analytical results, obtained via Edgeworth expansion, are also made.",

author = "Wolff, {Rodney C.} and Darfiana Nur and Mengersen, {Kerrie L.}",

year = "2001",

language = "English",

pages = "46--49",

note = "2001 IEEE Workshop on Statitical Signal Processing Proceedings ; Conference date: 06-08-2001 Through 08-08-2001",

}

TY - CONF

T1 - Assessment of MCMC convergence

T2 - 2001 IEEE Workshop on Statitical Signal Processing Proceedings

AU - Wolff, Rodney C.

AU - Nur, Darfiana

AU - Mengersen, Kerrie L.

PY - 2001

Y1 - 2001

N2 - Important in the application of Markov chain Monte Carlo (MCMC) methods is the determination that a search run has converged. Given that such searches typically take place in high-dimensional spaces, there are many pitfalls and difficulties in making such assessments. In the present paper, we discuss the use of phase randomisation as tool in the MCMC context, provide some details of its distributional properties for time series which enable its use as a convergence diagnostic, and contrast its performance with a selection of other widely used diagnostics. Some brief comments on analytical results, obtained via Edgeworth expansion, are also made.

AB - Important in the application of Markov chain Monte Carlo (MCMC) methods is the determination that a search run has converged. Given that such searches typically take place in high-dimensional spaces, there are many pitfalls and difficulties in making such assessments. In the present paper, we discuss the use of phase randomisation as tool in the MCMC context, provide some details of its distributional properties for time series which enable its use as a convergence diagnostic, and contrast its performance with a selection of other widely used diagnostics. Some brief comments on analytical results, obtained via Edgeworth expansion, are also made.

UR - http://www.scopus.com/inward/record.url?scp=0035555695&partnerID=8YFLogxK

M3 - Conference presentation/ephemera

AN - SCOPUS:0035555695

SP - 46

EP - 49

Y2 - 6 August 2001 through 8 August 2001

ER -

Assessment of MCMC convergence: A time series and dynamical systems approach

Abstract

Conference

Other files and links

Fingerprint

Cite this