Fast Compression of MCMC Output

Fast Compression of MCMC Output

We propose cube thinning, a novel method for compressing the output of an MCMC (Markov chain Monte Carlo) algorithm when control variates are available. It allows resampling of the initial MCMC sample (according to weights derived from control variates), while imposing equality constraints on the av...

Full description

Bibliographic Details
Main Authors: Nicolas Chopin, Gabriel Ducrocq
Format: Article
Language:English
Published: MDPI AG 2021-08-01
Series:Entropy
Subjects:
control variates
Markov chain Monte Carlo
thinning
Online Access:https://www.mdpi.com/1099-4300/23/8/1017
id doaj-878b72b8a99d4c45a91343f5d4b93224
record_format Article
spelling doaj-878b72b8a99d4c45a91343f5d4b932242021-08-26T13:44:11ZengMDPI AGEntropy1099-43002021-08-01231017101710.3390/e23081017Fast Compression of MCMC OutputNicolas Chopin0Gabriel Ducrocq1Institut Polytechnique de Paris, ENSAE Paris, CEDEX, 92247 Malakoff, FranceInstitut Polytechnique de Paris, ENSAE Paris, CEDEX, 92247 Malakoff, FranceWe propose cube thinning, a novel method for compressing the output of an MCMC (Markov chain Monte Carlo) algorithm when control variates are available. It allows resampling of the initial MCMC sample (according to weights derived from control variates), while imposing equality constraints on the averages of these control variates, using the cube method (an approach that originates from survey sampling). The main advantage of cube thinning is that its complexity does not depend on the size of the compressed sample. This compares favourably to previous methods, such as Stein thinning, the complexity of which is quadratic in that quantity.https://www.mdpi.com/1099-4300/23/8/1017control variatesMarkov chain Monte Carlothinning
collection DOAJ
language English
format Article
sources DOAJ
author Nicolas Chopin
Gabriel Ducrocq
spellingShingle Nicolas Chopin
Gabriel Ducrocq
Fast Compression of MCMC Output
Entropy
control variates
Markov chain Monte Carlo
thinning
author_facet Nicolas Chopin
Gabriel Ducrocq
author_sort Nicolas Chopin
title Fast Compression of MCMC Output
title_short Fast Compression of MCMC Output
title_full Fast Compression of MCMC Output
title_fullStr Fast Compression of MCMC Output
title_full_unstemmed Fast Compression of MCMC Output
title_sort fast compression of mcmc output
publisher MDPI AG
series Entropy
issn 1099-4300
publishDate 2021-08-01
description We propose cube thinning, a novel method for compressing the output of an MCMC (Markov chain Monte Carlo) algorithm when control variates are available. It allows resampling of the initial MCMC sample (according to weights derived from control variates), while imposing equality constraints on the averages of these control variates, using the cube method (an approach that originates from survey sampling). The main advantage of cube thinning is that its complexity does not depend on the size of the compressed sample. This compares favourably to previous methods, such as Stein thinning, the complexity of which is quadratic in that quantity.
topic control variates
Markov chain Monte Carlo
thinning
url https://www.mdpi.com/1099-4300/23/8/1017
work_keys_str_mv AT nicolaschopin fastcompressionofmcmcoutput
AT gabrielducrocq fastcompressionofmcmcoutput
_version_ 1721193613493272576

Similar Items