Rejection control and sequential importance sampling

Jun S. Liu; Rong Chen; Wing Hung Wong

doi:10.1080/01621459.1998.10473764

Rejection control and sequential importance sampling

Jun S. Liu, Rong Chen, Wing Hung Wong

Research output: Contribution to journal › Article › peer-review

70 Scopus citations

Abstract

We discuss ways of combining rejection sampling and importance sampling methods in Monte Carlo computations and demonstrate their usefulness in updating dynamic systems. Specifically, we propose the rejection controlled sequential importance sampling (RC-SIS) algorithm, which is designed to simultaneously reduce Monte Carlo variation and retain independent samples in sequential importance sampling. The proposed method is demonstrated by three examples taken from econometrics, hierarchical Bayes analysis, and digital telecommunications. They all show significant improvements over previous results.

Original language	English (US)
Pages (from-to)	1022-1031
Number of pages	10
Journal	Journal of the American Statistical Association
Volume	93
Issue number	443
DOIs	https://doi.org/10.1080/01621459.1998.10473764
State	Published - Sep 1 1998
Externally published	Yes

All Science Journal Classification (ASJC) codes

Statistics and Probability
Statistics, Probability and Uncertainty

Keywords

Blind deconvolution
Econometric disequilibrium model
Gibbs sampler
Markov chain Monte Carlo
Markov chain Monte Carlo
Sequential imputation
Weighted Markov chain Monte Carlo

Access to Document

10.1080/01621459.1998.10473764

Cite this

@article{8eb16f3e525c42bd8cb97f35268621a7,

title = "Rejection control and sequential importance sampling",

abstract = "We discuss ways of combining rejection sampling and importance sampling methods in Monte Carlo computations and demonstrate their usefulness in updating dynamic systems. Specifically, we propose the rejection controlled sequential importance sampling (RC-SIS) algorithm, which is designed to simultaneously reduce Monte Carlo variation and retain independent samples in sequential importance sampling. The proposed method is demonstrated by three examples taken from econometrics, hierarchical Bayes analysis, and digital telecommunications. They all show significant improvements over previous results.",

keywords = "Blind deconvolution, Econometric disequilibrium model, Gibbs sampler, Markov chain Monte Carlo, Markov chain Monte Carlo, Sequential imputation, Weighted Markov chain Monte Carlo",

author = "Liu, {Jun S.} and Rong Chen and Wong, {Wing Hung}",

note = "Funding Information: Jun S. Liu is Assistant Professor, Department of Statistics, Stanford University, Stanford, CA 94305. Rong Chen is Associate Professor, Department of Statistics, Texas A&M University, College Station, Texas 77843. Wing Hung Wong is Professor, Department of Statistics, UCLA, 8118 Math Sciences, 405 Hilgard Ave., Los Angeles, CA 90095. Liu's research was partially supported by National Science Foundation (NSF) grants DMS 94-04344 and DMS 95-01570 and the Terman fellowship from Stanford. Chen's research was partially supported by NSF grant DMS 96-26113. Wong's research was partially supported by earmarked grant CUHK 473/95p from the Research Grant Council of Hong Kong and by NSF grant DMS-97039 18. The authors are grateful to Jerry Friedman, three referees, and the associate editor for insightful comments.",

year = "1998",

month = sep,

day = "1",

doi = "10.1080/01621459.1998.10473764",

language = "English (US)",

volume = "93",

pages = "1022--1031",

journal = "Journal of the American Statistical Association",

issn = "0162-1459",

publisher = "Taylor and Francis Ltd.",

number = "443",

}

TY - JOUR

T1 - Rejection control and sequential importance sampling

AU - Liu, Jun S.

AU - Chen, Rong

AU - Wong, Wing Hung

N1 - Funding Information: Jun S. Liu is Assistant Professor, Department of Statistics, Stanford University, Stanford, CA 94305. Rong Chen is Associate Professor, Department of Statistics, Texas A&M University, College Station, Texas 77843. Wing Hung Wong is Professor, Department of Statistics, UCLA, 8118 Math Sciences, 405 Hilgard Ave., Los Angeles, CA 90095. Liu's research was partially supported by National Science Foundation (NSF) grants DMS 94-04344 and DMS 95-01570 and the Terman fellowship from Stanford. Chen's research was partially supported by NSF grant DMS 96-26113. Wong's research was partially supported by earmarked grant CUHK 473/95p from the Research Grant Council of Hong Kong and by NSF grant DMS-97039 18. The authors are grateful to Jerry Friedman, three referees, and the associate editor for insightful comments.

PY - 1998/9/1

Y1 - 1998/9/1

N2 - We discuss ways of combining rejection sampling and importance sampling methods in Monte Carlo computations and demonstrate their usefulness in updating dynamic systems. Specifically, we propose the rejection controlled sequential importance sampling (RC-SIS) algorithm, which is designed to simultaneously reduce Monte Carlo variation and retain independent samples in sequential importance sampling. The proposed method is demonstrated by three examples taken from econometrics, hierarchical Bayes analysis, and digital telecommunications. They all show significant improvements over previous results.

AB - We discuss ways of combining rejection sampling and importance sampling methods in Monte Carlo computations and demonstrate their usefulness in updating dynamic systems. Specifically, we propose the rejection controlled sequential importance sampling (RC-SIS) algorithm, which is designed to simultaneously reduce Monte Carlo variation and retain independent samples in sequential importance sampling. The proposed method is demonstrated by three examples taken from econometrics, hierarchical Bayes analysis, and digital telecommunications. They all show significant improvements over previous results.

KW - Blind deconvolution

KW - Econometric disequilibrium model

KW - Gibbs sampler

KW - Markov chain Monte Carlo

KW - Sequential imputation

KW - Weighted Markov chain Monte Carlo

UR - http://www.scopus.com/inward/record.url?scp=0032351643&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0032351643&partnerID=8YFLogxK

U2 - 10.1080/01621459.1998.10473764

DO - 10.1080/01621459.1998.10473764

M3 - Article

AN - SCOPUS:0032351643

SN - 0162-1459

VL - 93

SP - 1022

EP - 1031

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

IS - 443

ER -

Rejection control and sequential importance sampling

Abstract

All Science Journal Classification (ASJC) codes

Keywords

Access to Document

Other files and links

Fingerprint

Cite this