Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices

Crude and quasi Monte Carlo (MC) sampling techniques are common tools dedicated to estimating statistics (expectation, variance, covariance) of a random quantity of interest.We focus here on the uncertainty quantification framework where the quantity of interest is the output of a numerical simulato...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:SIAM/ASA journal on uncertainty quantification 2019-01, Vol.7 (4), p.1323-1348
Hauptverfasser: Mycek, Paul, De Lozzo, Matthias
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 1348
container_issue 4
container_start_page 1323
container_title SIAM/ASA journal on uncertainty quantification
container_volume 7
creator Mycek, Paul
De Lozzo, Matthias
description Crude and quasi Monte Carlo (MC) sampling techniques are common tools dedicated to estimating statistics (expectation, variance, covariance) of a random quantity of interest.We focus here on the uncertainty quantification framework where the quantity of interest is the output of a numerical simulator fed with uncertain input parameters.Then, sampling the output involves running the simulator for different samples of the inputs, which may be computationally time-consuming.To reduce the cost of sampling, a first approach consists in replacing the numerical simulator by a surrogate model that is cheaper to evaluate, thus making it possible to generate more samples of the output and therefore leading to a lower sampling error.However, this approach adds to the sampling error an unavoidable model error.Another approach, which does not introduce any model error, is the so-called multilevel MC (MLMC) method.Given a sequence of levels corresponding to numerical simulators with increasing accuracy and computational cost, MLMC combines samples obtained at different levels to construct an estimator at a reduced cost compared to standard MC sampling.In this paper, we derive and analyze multilevel covariance estimators and adapt the MLMC convergence theorem in terms of the corresponding covariances and fourth order moments. We propose a multilevel algorithm driven by a target cost as an alternative to typical algorithms driven by a target accuracy.These results are used in a sensitivity analysis context in order to derive a multilevel estimation of Sobol' indices, whose building blocks can be written as covariance terms in a pick-and-freeze formulation.These contributions are successfully tested on an initial value problem with random parameters.
doi_str_mv 10.1137/18M1216389
format Article
fullrecord <record><control><sourceid>hal_cross</sourceid><recordid>TN_cdi_hal_primary_oai_HAL_hal_01894503v2</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>oai_HAL_hal_01894503v2</sourcerecordid><originalsourceid>FETCH-LOGICAL-c301t-df06366204a3fe9e29b9fa594a5a9a2c7affa19cca5fbcbbeef0d40c69767ff3</originalsourceid><addsrcrecordid>eNpNkE9LAzEQxYMoWGovfoLcRGE1fzbZ5liWaoUWQXtfZtMMXUk3JUkLfnu3VNS5zOPxm-HxCLnl7JFzWT3x6YoLruXUXJDRIHQhlFCX__Q1maT0yYYxJZeKj8j76uBz593ReboKfXa0hugDrcMRYge9dXSecreD3IWeYog0bwcm7PaHfPYC0o_QBn9HX_tNZ126IVcIPrnJzx6T9fN8XS-K5dvLaz1bFlYynosNMi21FqwEic44YVqDoEwJCgwIWwEicGMtKGxt2zqHbFMyq02lK0Q5Jvfnt1vwzT4OEeNXE6BrFrNlc_IYn5pSMXkUA_twZm0MKUWHvwecNafumr_u5DdxyGE3</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices</title><source>SIAM Journals Online</source><creator>Mycek, Paul ; De Lozzo, Matthias</creator><creatorcontrib>Mycek, Paul ; De Lozzo, Matthias</creatorcontrib><description>Crude and quasi Monte Carlo (MC) sampling techniques are common tools dedicated to estimating statistics (expectation, variance, covariance) of a random quantity of interest.We focus here on the uncertainty quantification framework where the quantity of interest is the output of a numerical simulator fed with uncertain input parameters.Then, sampling the output involves running the simulator for different samples of the inputs, which may be computationally time-consuming.To reduce the cost of sampling, a first approach consists in replacing the numerical simulator by a surrogate model that is cheaper to evaluate, thus making it possible to generate more samples of the output and therefore leading to a lower sampling error.However, this approach adds to the sampling error an unavoidable model error.Another approach, which does not introduce any model error, is the so-called multilevel MC (MLMC) method.Given a sequence of levels corresponding to numerical simulators with increasing accuracy and computational cost, MLMC combines samples obtained at different levels to construct an estimator at a reduced cost compared to standard MC sampling.In this paper, we derive and analyze multilevel covariance estimators and adapt the MLMC convergence theorem in terms of the corresponding covariances and fourth order moments. We propose a multilevel algorithm driven by a target cost as an alternative to typical algorithms driven by a target accuracy.These results are used in a sensitivity analysis context in order to derive a multilevel estimation of Sobol' indices, whose building blocks can be written as covariance terms in a pick-and-freeze formulation.These contributions are successfully tested on an initial value problem with random parameters.</description><identifier>ISSN: 2166-2525</identifier><identifier>EISSN: 2166-2525</identifier><identifier>DOI: 10.1137/18M1216389</identifier><language>eng</language><publisher>ASA, American Statistical Association</publisher><subject>Computational Engineering, Finance, and Science ; Computer Science ; Mathematics ; Statistics</subject><ispartof>SIAM/ASA journal on uncertainty quantification, 2019-01, Vol.7 (4), p.1323-1348</ispartof><rights>Distributed under a Creative Commons Attribution 4.0 International License</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c301t-df06366204a3fe9e29b9fa594a5a9a2c7affa19cca5fbcbbeef0d40c69767ff3</citedby><cites>FETCH-LOGICAL-c301t-df06366204a3fe9e29b9fa594a5a9a2c7affa19cca5fbcbbeef0d40c69767ff3</cites><orcidid>0000-0002-6919-112X ; 0009-0002-2694-3587</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,314,780,784,885,3184,27924,27925</link.rule.ids><backlink>$$Uhttps://hal.science/hal-01894503$$DView record in HAL$$Hfree_for_read</backlink></links><search><creatorcontrib>Mycek, Paul</creatorcontrib><creatorcontrib>De Lozzo, Matthias</creatorcontrib><title>Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices</title><title>SIAM/ASA journal on uncertainty quantification</title><description>Crude and quasi Monte Carlo (MC) sampling techniques are common tools dedicated to estimating statistics (expectation, variance, covariance) of a random quantity of interest.We focus here on the uncertainty quantification framework where the quantity of interest is the output of a numerical simulator fed with uncertain input parameters.Then, sampling the output involves running the simulator for different samples of the inputs, which may be computationally time-consuming.To reduce the cost of sampling, a first approach consists in replacing the numerical simulator by a surrogate model that is cheaper to evaluate, thus making it possible to generate more samples of the output and therefore leading to a lower sampling error.However, this approach adds to the sampling error an unavoidable model error.Another approach, which does not introduce any model error, is the so-called multilevel MC (MLMC) method.Given a sequence of levels corresponding to numerical simulators with increasing accuracy and computational cost, MLMC combines samples obtained at different levels to construct an estimator at a reduced cost compared to standard MC sampling.In this paper, we derive and analyze multilevel covariance estimators and adapt the MLMC convergence theorem in terms of the corresponding covariances and fourth order moments. We propose a multilevel algorithm driven by a target cost as an alternative to typical algorithms driven by a target accuracy.These results are used in a sensitivity analysis context in order to derive a multilevel estimation of Sobol' indices, whose building blocks can be written as covariance terms in a pick-and-freeze formulation.These contributions are successfully tested on an initial value problem with random parameters.</description><subject>Computational Engineering, Finance, and Science</subject><subject>Computer Science</subject><subject>Mathematics</subject><subject>Statistics</subject><issn>2166-2525</issn><issn>2166-2525</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><recordid>eNpNkE9LAzEQxYMoWGovfoLcRGE1fzbZ5liWaoUWQXtfZtMMXUk3JUkLfnu3VNS5zOPxm-HxCLnl7JFzWT3x6YoLruXUXJDRIHQhlFCX__Q1maT0yYYxJZeKj8j76uBz593ReboKfXa0hugDrcMRYge9dXSecreD3IWeYog0bwcm7PaHfPYC0o_QBn9HX_tNZ126IVcIPrnJzx6T9fN8XS-K5dvLaz1bFlYynosNMi21FqwEic44YVqDoEwJCgwIWwEicGMtKGxt2zqHbFMyq02lK0Q5Jvfnt1vwzT4OEeNXE6BrFrNlc_IYn5pSMXkUA_twZm0MKUWHvwecNafumr_u5DdxyGE3</recordid><startdate>20190101</startdate><enddate>20190101</enddate><creator>Mycek, Paul</creator><creator>De Lozzo, Matthias</creator><general>ASA, American Statistical Association</general><scope>AAYXX</scope><scope>CITATION</scope><scope>1XC</scope><scope>VOOES</scope><orcidid>https://orcid.org/0000-0002-6919-112X</orcidid><orcidid>https://orcid.org/0009-0002-2694-3587</orcidid></search><sort><creationdate>20190101</creationdate><title>Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices</title><author>Mycek, Paul ; De Lozzo, Matthias</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c301t-df06366204a3fe9e29b9fa594a5a9a2c7affa19cca5fbcbbeef0d40c69767ff3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computational Engineering, Finance, and Science</topic><topic>Computer Science</topic><topic>Mathematics</topic><topic>Statistics</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Mycek, Paul</creatorcontrib><creatorcontrib>De Lozzo, Matthias</creatorcontrib><collection>CrossRef</collection><collection>Hyper Article en Ligne (HAL)</collection><collection>Hyper Article en Ligne (HAL) (Open Access)</collection><jtitle>SIAM/ASA journal on uncertainty quantification</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Mycek, Paul</au><au>De Lozzo, Matthias</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices</atitle><jtitle>SIAM/ASA journal on uncertainty quantification</jtitle><date>2019-01-01</date><risdate>2019</risdate><volume>7</volume><issue>4</issue><spage>1323</spage><epage>1348</epage><pages>1323-1348</pages><issn>2166-2525</issn><eissn>2166-2525</eissn><abstract>Crude and quasi Monte Carlo (MC) sampling techniques are common tools dedicated to estimating statistics (expectation, variance, covariance) of a random quantity of interest.We focus here on the uncertainty quantification framework where the quantity of interest is the output of a numerical simulator fed with uncertain input parameters.Then, sampling the output involves running the simulator for different samples of the inputs, which may be computationally time-consuming.To reduce the cost of sampling, a first approach consists in replacing the numerical simulator by a surrogate model that is cheaper to evaluate, thus making it possible to generate more samples of the output and therefore leading to a lower sampling error.However, this approach adds to the sampling error an unavoidable model error.Another approach, which does not introduce any model error, is the so-called multilevel MC (MLMC) method.Given a sequence of levels corresponding to numerical simulators with increasing accuracy and computational cost, MLMC combines samples obtained at different levels to construct an estimator at a reduced cost compared to standard MC sampling.In this paper, we derive and analyze multilevel covariance estimators and adapt the MLMC convergence theorem in terms of the corresponding covariances and fourth order moments. We propose a multilevel algorithm driven by a target cost as an alternative to typical algorithms driven by a target accuracy.These results are used in a sensitivity analysis context in order to derive a multilevel estimation of Sobol' indices, whose building blocks can be written as covariance terms in a pick-and-freeze formulation.These contributions are successfully tested on an initial value problem with random parameters.</abstract><pub>ASA, American Statistical Association</pub><doi>10.1137/18M1216389</doi><tpages>26</tpages><orcidid>https://orcid.org/0000-0002-6919-112X</orcidid><orcidid>https://orcid.org/0009-0002-2694-3587</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 2166-2525
ispartof SIAM/ASA journal on uncertainty quantification, 2019-01, Vol.7 (4), p.1323-1348
issn 2166-2525
2166-2525
language eng
recordid cdi_hal_primary_oai_HAL_hal_01894503v2
source SIAM Journals Online
subjects Computational Engineering, Finance, and Science
Computer Science
Mathematics
Statistics
title Multilevel Monte Carlo Covariance Estimation for the Computation of Sobol' Indices
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-25T20%3A22%3A48IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-hal_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Multilevel%20Monte%20Carlo%20Covariance%20Estimation%20for%20the%20Computation%20of%20Sobol'%20Indices&rft.jtitle=SIAM/ASA%20journal%20on%20uncertainty%20quantification&rft.au=Mycek,%20Paul&rft.date=2019-01-01&rft.volume=7&rft.issue=4&rft.spage=1323&rft.epage=1348&rft.pages=1323-1348&rft.issn=2166-2525&rft.eissn=2166-2525&rft_id=info:doi/10.1137/18M1216389&rft_dat=%3Chal_cross%3Eoai_HAL_hal_01894503v2%3C/hal_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true