Optimal thinning of MCMC output
Optimal thinning of MCMC output
The use of heuristics to assess the convergence and compress the output of Markov chain Monte Carlo can be sub-optimal in terms of the empirical approximations that are produced. Typically a number of the initial states are attributed to "burn in" and removed, whilst the remainder of the chain is "thinned" if compression is also required. In this paper we consider the problem of retrospectively selecting a subset of states, of fixed cardinality, from the sample path such that the approximation provided by their empirical distribution is close to optimal. A novel method is proposed, based on greedy minimisation of a kernel Stein discrepancy, that is suitable for problems where heavy compression is required. Theoretical results guarantee consistency of the method and its effectiveness is demonstrated in the challenging context of parameter inference for ordinary differential equations. Software is available in the Stein Thinning package in Python, R and MATLAB.
1059–1081
Riabiz, Marina
1ecee25d-b4c9-40ae-b904-819e9410a165
Chen, Wilson Ye
14babbf3-cf1f-46ea-8401-e4455751ecdd
Cockayne, Jon
da87c8b2-fafb-4856-938d-50be8f0e4a5b
Swietach, Pawel
b61b2090-5af9-45a8-9c55-0e8a6cf188ac
Niederer, Steven A.
39de3d6c-0360-455d-93de-776af21ed975
Mackey, Lester
b4468e0b-f801-42ec-bcd9-4f570cbf554a
Oates, Chris J.
3af13c56-dc47-4d2c-867f-e4e933e74619
Riabiz, Marina
1ecee25d-b4c9-40ae-b904-819e9410a165
Chen, Wilson Ye
14babbf3-cf1f-46ea-8401-e4455751ecdd
Cockayne, Jon
da87c8b2-fafb-4856-938d-50be8f0e4a5b
Swietach, Pawel
b61b2090-5af9-45a8-9c55-0e8a6cf188ac
Niederer, Steven A.
39de3d6c-0360-455d-93de-776af21ed975
Mackey, Lester
b4468e0b-f801-42ec-bcd9-4f570cbf554a
Oates, Chris J.
3af13c56-dc47-4d2c-867f-e4e933e74619
Riabiz, Marina, Chen, Wilson Ye, Cockayne, Jon, Swietach, Pawel, Niederer, Steven A., Mackey, Lester and Oates, Chris J.
(2022)
Optimal thinning of MCMC output.
Journal of the Royal Statistical Society: Series B (Statistical Methodology), 84 (4), .
(doi:10.1111/rssb.12503).
Abstract
The use of heuristics to assess the convergence and compress the output of Markov chain Monte Carlo can be sub-optimal in terms of the empirical approximations that are produced. Typically a number of the initial states are attributed to "burn in" and removed, whilst the remainder of the chain is "thinned" if compression is also required. In this paper we consider the problem of retrospectively selecting a subset of states, of fixed cardinality, from the sample path such that the approximation provided by their empirical distribution is close to optimal. A novel method is proposed, based on greedy minimisation of a kernel Stein discrepancy, that is suitable for problems where heavy compression is required. Theoretical results guarantee consistency of the method and its effectiveness is demonstrated in the challenging context of parameter inference for ordinary differential equations. Software is available in the Stein Thinning package in Python, R and MATLAB.
Text
Optimal Thinning of MCMC Output
- Author's Original
Text
jrsssb_84_4_1059
- Version of Record
More information
Accepted/In Press date: 11 July 2021
e-pub ahead of print date: 3 April 2022
Identifiers
Local EPrints ID: 451789
URI: http://eprints.soton.ac.uk/id/eprint/451789
ISSN: 1467-9868
PURE UUID: 6a4c4be6-538f-4d08-8ee4-6952c4c4ce62
Catalogue record
Date deposited: 27 Oct 2021 16:32
Last modified: 13 Jul 2024 02:04
Export record
Altmetrics
Contributors
Author:
Marina Riabiz
Author:
Wilson Ye Chen
Author:
Pawel Swietach
Author:
Steven A. Niederer
Author:
Lester Mackey
Author:
Chris J. Oates
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics
