Multivariate initial sequence estimators in Markov chain Monte Carlo

Ning Dai, Galin Jones

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Markov chain Monte Carlo (MCMC) is a simulation method commonly used for estimating expectations with respect to a given distribution. We consider estimating the covariance matrix of the asymptotic multivariate normal distribution of a vector of sample means. Geyer (1992) developed a Monte Carlo error estimation method for estimating a univariate mean. We propose a novel multivariate version of Geyer's method that provides an asymptotically valid estimator for the covariance matrix and results in stable Monte Carlo estimates. The finite sample properties of the proposed method are investigated via simulation experiments.

Original languageEnglish (US)
Pages (from-to)184-199
Number of pages16
JournalJournal of Multivariate Analysis
Volume159
DOIs
StatePublished - Jul 1 2017

Fingerprint

Covariance matrix
Markov Chain Monte Carlo
Markov processes
Estimator
Normal distribution
Error analysis
Multivariate Normal Distribution
Sample mean
Error Estimation
Simulation Methods
Simulation Experiment
Univariate
Valid
Experiments
Estimate
Markov chain Monte Carlo

Keywords

  • Central limit theorem
  • Covariance matrix estimation
  • Gibbs sampler
  • Markov chain Monte Carlo
  • Metropolis–Hastings algorithm

Cite this

Multivariate initial sequence estimators in Markov chain Monte Carlo. / Dai, Ning; Jones, Galin.

In: Journal of Multivariate Analysis, Vol. 159, 01.07.2017, p. 184-199.

Research output: Contribution to journalArticle

@article{cb982986ee884d93abd70e8c0114d782,
title = "Multivariate initial sequence estimators in Markov chain Monte Carlo",
abstract = "Markov chain Monte Carlo (MCMC) is a simulation method commonly used for estimating expectations with respect to a given distribution. We consider estimating the covariance matrix of the asymptotic multivariate normal distribution of a vector of sample means. Geyer (1992) developed a Monte Carlo error estimation method for estimating a univariate mean. We propose a novel multivariate version of Geyer's method that provides an asymptotically valid estimator for the covariance matrix and results in stable Monte Carlo estimates. The finite sample properties of the proposed method are investigated via simulation experiments.",
keywords = "Central limit theorem, Covariance matrix estimation, Gibbs sampler, Markov chain Monte Carlo, Metropolis–Hastings algorithm",
author = "Ning Dai and Galin Jones",
year = "2017",
month = "7",
day = "1",
doi = "10.1016/j.jmva.2017.05.009",
language = "English (US)",
volume = "159",
pages = "184--199",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Multivariate initial sequence estimators in Markov chain Monte Carlo

AU - Dai, Ning

AU - Jones, Galin

PY - 2017/7/1

Y1 - 2017/7/1

N2 - Markov chain Monte Carlo (MCMC) is a simulation method commonly used for estimating expectations with respect to a given distribution. We consider estimating the covariance matrix of the asymptotic multivariate normal distribution of a vector of sample means. Geyer (1992) developed a Monte Carlo error estimation method for estimating a univariate mean. We propose a novel multivariate version of Geyer's method that provides an asymptotically valid estimator for the covariance matrix and results in stable Monte Carlo estimates. The finite sample properties of the proposed method are investigated via simulation experiments.

AB - Markov chain Monte Carlo (MCMC) is a simulation method commonly used for estimating expectations with respect to a given distribution. We consider estimating the covariance matrix of the asymptotic multivariate normal distribution of a vector of sample means. Geyer (1992) developed a Monte Carlo error estimation method for estimating a univariate mean. We propose a novel multivariate version of Geyer's method that provides an asymptotically valid estimator for the covariance matrix and results in stable Monte Carlo estimates. The finite sample properties of the proposed method are investigated via simulation experiments.

KW - Central limit theorem

KW - Covariance matrix estimation

KW - Gibbs sampler

KW - Markov chain Monte Carlo

KW - Metropolis–Hastings algorithm

UR - http://www.scopus.com/inward/record.url?scp=85033401268&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85033401268&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2017.05.009

DO - 10.1016/j.jmva.2017.05.009

M3 - Article

AN - SCOPUS:85033401268

VL - 159

SP - 184

EP - 199

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

ER -