A posteriori stochastic correction of reduced models in delayed-acceptance MCMC, with application to multiphase subsurface inverse problems

Tiangang Cui, Colin Fox, Michael J. O'Sullivan

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Sample-based Bayesian inference provides a route to uncertainty quantification in the geosciences and inverse problems in general but is very computationally demanding in the naïve form, which requires simulating an accurate computer model at each iteration. We present a new approach that constructs a stochastic correction to the error induced by a reduced model, with the correction improving as the algorithm proceeds. This enables sampling from the correct target distribution at reduced computational cost per iteration, as in existing delayed-acceptance schemes, while avoiding appreciable loss of statistical efficiency that necessarily occurs when using a reduced model. Use of the stochastic correction significantly reduces the computational cost of estimating quantities of interest within desired uncertainty bounds. In contrast, existing schemes that use a reduced model directly as a surrogate do not actually improve computational efficiency in our target applications. We build on recent simplified conditions for adaptive Markov chain Monte Carlo algorithms to give practical approximation schemes and algorithms with guaranteed convergence. The efficacy of this new approach is demonstrated in two computational examples, including calibration of a large-scale numerical model of a real geothermal reservoir, that show good computational and statistical efficiencies on both synthetic and measured data sets.

Original languageEnglish
Pages (from-to)578-605
Number of pages28
JournalInternational Journal for Numerical Methods in Engineering
Volume118
Issue number10
DOIs
Publication statusPublished - 8 Jun 2019

Keywords

  • adaptive MCMC
  • delayed acceptance
  • geothermal reservoir modeling
  • inverse problem
  • reduced model

Cite this

@article{b3565085a9c14ba2a9ab48e7cadb8c87,
title = "A posteriori stochastic correction of reduced models in delayed-acceptance MCMC, with application to multiphase subsurface inverse problems",
abstract = "Sample-based Bayesian inference provides a route to uncertainty quantification in the geosciences and inverse problems in general but is very computationally demanding in the na{\"i}ve form, which requires simulating an accurate computer model at each iteration. We present a new approach that constructs a stochastic correction to the error induced by a reduced model, with the correction improving as the algorithm proceeds. This enables sampling from the correct target distribution at reduced computational cost per iteration, as in existing delayed-acceptance schemes, while avoiding appreciable loss of statistical efficiency that necessarily occurs when using a reduced model. Use of the stochastic correction significantly reduces the computational cost of estimating quantities of interest within desired uncertainty bounds. In contrast, existing schemes that use a reduced model directly as a surrogate do not actually improve computational efficiency in our target applications. We build on recent simplified conditions for adaptive Markov chain Monte Carlo algorithms to give practical approximation schemes and algorithms with guaranteed convergence. The efficacy of this new approach is demonstrated in two computational examples, including calibration of a large-scale numerical model of a real geothermal reservoir, that show good computational and statistical efficiencies on both synthetic and measured data sets.",
keywords = "adaptive MCMC, delayed acceptance, geothermal reservoir modeling, inverse problem, reduced model",
author = "Tiangang Cui and Colin Fox and O'Sullivan, {Michael J.}",
year = "2019",
month = "6",
day = "8",
doi = "10.1002/nme.6028",
language = "English",
volume = "118",
pages = "578--605",
journal = "International Journal for Numerical Methods in Engineering",
issn = "0029-5981",
publisher = "Wiley-Blackwell",
number = "10",

}

A posteriori stochastic correction of reduced models in delayed-acceptance MCMC, with application to multiphase subsurface inverse problems. / Cui, Tiangang; Fox, Colin; O'Sullivan, Michael J.

In: International Journal for Numerical Methods in Engineering, Vol. 118, No. 10, 08.06.2019, p. 578-605.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - A posteriori stochastic correction of reduced models in delayed-acceptance MCMC, with application to multiphase subsurface inverse problems

AU - Cui, Tiangang

AU - Fox, Colin

AU - O'Sullivan, Michael J.

PY - 2019/6/8

Y1 - 2019/6/8

N2 - Sample-based Bayesian inference provides a route to uncertainty quantification in the geosciences and inverse problems in general but is very computationally demanding in the naïve form, which requires simulating an accurate computer model at each iteration. We present a new approach that constructs a stochastic correction to the error induced by a reduced model, with the correction improving as the algorithm proceeds. This enables sampling from the correct target distribution at reduced computational cost per iteration, as in existing delayed-acceptance schemes, while avoiding appreciable loss of statistical efficiency that necessarily occurs when using a reduced model. Use of the stochastic correction significantly reduces the computational cost of estimating quantities of interest within desired uncertainty bounds. In contrast, existing schemes that use a reduced model directly as a surrogate do not actually improve computational efficiency in our target applications. We build on recent simplified conditions for adaptive Markov chain Monte Carlo algorithms to give practical approximation schemes and algorithms with guaranteed convergence. The efficacy of this new approach is demonstrated in two computational examples, including calibration of a large-scale numerical model of a real geothermal reservoir, that show good computational and statistical efficiencies on both synthetic and measured data sets.

AB - Sample-based Bayesian inference provides a route to uncertainty quantification in the geosciences and inverse problems in general but is very computationally demanding in the naïve form, which requires simulating an accurate computer model at each iteration. We present a new approach that constructs a stochastic correction to the error induced by a reduced model, with the correction improving as the algorithm proceeds. This enables sampling from the correct target distribution at reduced computational cost per iteration, as in existing delayed-acceptance schemes, while avoiding appreciable loss of statistical efficiency that necessarily occurs when using a reduced model. Use of the stochastic correction significantly reduces the computational cost of estimating quantities of interest within desired uncertainty bounds. In contrast, existing schemes that use a reduced model directly as a surrogate do not actually improve computational efficiency in our target applications. We build on recent simplified conditions for adaptive Markov chain Monte Carlo algorithms to give practical approximation schemes and algorithms with guaranteed convergence. The efficacy of this new approach is demonstrated in two computational examples, including calibration of a large-scale numerical model of a real geothermal reservoir, that show good computational and statistical efficiencies on both synthetic and measured data sets.

KW - adaptive MCMC

KW - delayed acceptance

KW - geothermal reservoir modeling

KW - inverse problem

KW - reduced model

UR - http://www.scopus.com/inward/record.url?scp=85062999319&partnerID=8YFLogxK

U2 - 10.1002/nme.6028

DO - 10.1002/nme.6028

M3 - Article

VL - 118

SP - 578

EP - 605

JO - International Journal for Numerical Methods in Engineering

JF - International Journal for Numerical Methods in Engineering

SN - 0029-5981

IS - 10

ER -