Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise

In this study, we propose a scalable batch sampling scheme for optimization of simulation models with spatially varying noise. The proposed scheme has two primary advantages: (i) reduced simulation cost by recommending batches of samples at carefully selected spatial locations and (ii) improved scal...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of mechanical design (1990) 2021-03, Vol.143 (3)
Hauptverfasser: van Beek, Anton, Ghumman, Umar Farooq, Munshi, Joydeep, Tao, Siyu, Chien, TeYu, Balasubramanian, Ganesh, Plumlee, Matthew, Apley, Daniel, Chen, Wei
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue 3
container_start_page
container_title Journal of mechanical design (1990)
container_volume 143
creator van Beek, Anton
Ghumman, Umar Farooq
Munshi, Joydeep
Tao, Siyu
Chien, TeYu
Balasubramanian, Ganesh
Plumlee, Matthew
Apley, Daniel
Chen, Wei
description In this study, we propose a scalable batch sampling scheme for optimization of simulation models with spatially varying noise. The proposed scheme has two primary advantages: (i) reduced simulation cost by recommending batches of samples at carefully selected spatial locations and (ii) improved scalability by actively considering replicating at previously observed sampling locations. Replication improves the scalability of the proposed sampling scheme as the computational cost of adaptive sampling schemes grow cubicly with the number of unique sampling locations. Our main consideration for the allocation of computational resources is the minimization of the uncertainty in the optimal design. We analytically derive the relationship between the “exploration versus replication decision” and the posterior variance of the spatial random process used to approximate the simulation model’s mean response. Leveraging this reformulation in a novel objective-driven adaptive sampling scheme, we show that we can identify batches of samples that minimize the prediction uncertainty only in the regions of the design space expected to contain the global optimum. Finally, the proposed sampling scheme adopts a modified preposterior analysis that uses a zeroth-order interpolation of the spatially varying simulation noise to identify sampling batches. Through the optimization of three numerical test functions and one engineering problem, we demonstrate (i) the efficacy and of the proposed sampling scheme to deal with a wide array of stochastic functions, (ii) the superior performance of the proposed method on all test functions compared to existing methods, (iii) the empirical validity of using a zeroth-order approximation for the allocation of sampling batches, and (iv) its applicability to molecular dynamics simulations by optimizing the performance of an organic photovoltaic cell as a function of its processing settings.
doi_str_mv 10.1115/1.4049134
format Article
fullrecord <record><control><sourceid>asme_cross</sourceid><recordid>TN_cdi_crossref_primary_10_1115_1_4049134</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1091481</sourcerecordid><originalsourceid>FETCH-LOGICAL-a250t-5f88c9ff2ecc7b5e7ed8d4b5aeb7bb04b73ad2b725d150314f71e9206aec5e763</originalsourceid><addsrcrecordid>eNotkLtPwzAYxC0EEqUwsDN4ZUjxl9hNMrblUaQKhvIYo8_2l9ZVHlXsIvHfY9ROdzr9dDodY7cgJgCgHmAihSwhk2dsBCotklIIOI9eKJEImaeX7Mr7XQyhkGrEvtYGG9QN8ZnFfXA_xOcYzJavsd03rttw1_G1aw8NBtd3yRw9Wf5I3m06_u3Cli8p0NB7QxZ9cIa_9c7TNbuosfF0c9Ix-3x--lgsk9X7y-titkowVSIkqi4KU9Z1SsbkWlFOtrBSKySday2kzjO0qc5TZUGJDGSdA5WpmCKZSE-zMbs_9po4wQ9UV_vBtTj8ViCq_0MqqE6HRPbuyKJvqdr1h6GL0yJYgiwg-wOHtFyL</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise</title><source>Alma/SFX Local Collection</source><source>ASME Transactions Journals (Current)</source><creator>van Beek, Anton ; Ghumman, Umar Farooq ; Munshi, Joydeep ; Tao, Siyu ; Chien, TeYu ; Balasubramanian, Ganesh ; Plumlee, Matthew ; Apley, Daniel ; Chen, Wei</creator><creatorcontrib>van Beek, Anton ; Ghumman, Umar Farooq ; Munshi, Joydeep ; Tao, Siyu ; Chien, TeYu ; Balasubramanian, Ganesh ; Plumlee, Matthew ; Apley, Daniel ; Chen, Wei</creatorcontrib><description>In this study, we propose a scalable batch sampling scheme for optimization of simulation models with spatially varying noise. The proposed scheme has two primary advantages: (i) reduced simulation cost by recommending batches of samples at carefully selected spatial locations and (ii) improved scalability by actively considering replicating at previously observed sampling locations. Replication improves the scalability of the proposed sampling scheme as the computational cost of adaptive sampling schemes grow cubicly with the number of unique sampling locations. Our main consideration for the allocation of computational resources is the minimization of the uncertainty in the optimal design. We analytically derive the relationship between the “exploration versus replication decision” and the posterior variance of the spatial random process used to approximate the simulation model’s mean response. Leveraging this reformulation in a novel objective-driven adaptive sampling scheme, we show that we can identify batches of samples that minimize the prediction uncertainty only in the regions of the design space expected to contain the global optimum. Finally, the proposed sampling scheme adopts a modified preposterior analysis that uses a zeroth-order interpolation of the spatially varying simulation noise to identify sampling batches. Through the optimization of three numerical test functions and one engineering problem, we demonstrate (i) the efficacy and of the proposed sampling scheme to deal with a wide array of stochastic functions, (ii) the superior performance of the proposed method on all test functions compared to existing methods, (iii) the empirical validity of using a zeroth-order approximation for the allocation of sampling batches, and (iv) its applicability to molecular dynamics simulations by optimizing the performance of an organic photovoltaic cell as a function of its processing settings.</description><identifier>ISSN: 1050-0472</identifier><identifier>EISSN: 1528-9001</identifier><identifier>DOI: 10.1115/1.4049134</identifier><language>eng</language><subject>Design for Manufacturing</subject><ispartof>Journal of mechanical design (1990), 2021-03, Vol.143 (3)</ispartof><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-a250t-5f88c9ff2ecc7b5e7ed8d4b5aeb7bb04b73ad2b725d150314f71e9206aec5e763</citedby><cites>FETCH-LOGICAL-a250t-5f88c9ff2ecc7b5e7ed8d4b5aeb7bb04b73ad2b725d150314f71e9206aec5e763</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,776,780,27903,27904,38499</link.rule.ids></links><search><creatorcontrib>van Beek, Anton</creatorcontrib><creatorcontrib>Ghumman, Umar Farooq</creatorcontrib><creatorcontrib>Munshi, Joydeep</creatorcontrib><creatorcontrib>Tao, Siyu</creatorcontrib><creatorcontrib>Chien, TeYu</creatorcontrib><creatorcontrib>Balasubramanian, Ganesh</creatorcontrib><creatorcontrib>Plumlee, Matthew</creatorcontrib><creatorcontrib>Apley, Daniel</creatorcontrib><creatorcontrib>Chen, Wei</creatorcontrib><title>Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise</title><title>Journal of mechanical design (1990)</title><addtitle>J. Mech. Des</addtitle><description>In this study, we propose a scalable batch sampling scheme for optimization of simulation models with spatially varying noise. The proposed scheme has two primary advantages: (i) reduced simulation cost by recommending batches of samples at carefully selected spatial locations and (ii) improved scalability by actively considering replicating at previously observed sampling locations. Replication improves the scalability of the proposed sampling scheme as the computational cost of adaptive sampling schemes grow cubicly with the number of unique sampling locations. Our main consideration for the allocation of computational resources is the minimization of the uncertainty in the optimal design. We analytically derive the relationship between the “exploration versus replication decision” and the posterior variance of the spatial random process used to approximate the simulation model’s mean response. Leveraging this reformulation in a novel objective-driven adaptive sampling scheme, we show that we can identify batches of samples that minimize the prediction uncertainty only in the regions of the design space expected to contain the global optimum. Finally, the proposed sampling scheme adopts a modified preposterior analysis that uses a zeroth-order interpolation of the spatially varying simulation noise to identify sampling batches. Through the optimization of three numerical test functions and one engineering problem, we demonstrate (i) the efficacy and of the proposed sampling scheme to deal with a wide array of stochastic functions, (ii) the superior performance of the proposed method on all test functions compared to existing methods, (iii) the empirical validity of using a zeroth-order approximation for the allocation of sampling batches, and (iv) its applicability to molecular dynamics simulations by optimizing the performance of an organic photovoltaic cell as a function of its processing settings.</description><subject>Design for Manufacturing</subject><issn>1050-0472</issn><issn>1528-9001</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><recordid>eNotkLtPwzAYxC0EEqUwsDN4ZUjxl9hNMrblUaQKhvIYo8_2l9ZVHlXsIvHfY9ROdzr9dDodY7cgJgCgHmAihSwhk2dsBCotklIIOI9eKJEImaeX7Mr7XQyhkGrEvtYGG9QN8ZnFfXA_xOcYzJavsd03rttw1_G1aw8NBtd3yRw9Wf5I3m06_u3Cli8p0NB7QxZ9cIa_9c7TNbuosfF0c9Ix-3x--lgsk9X7y-titkowVSIkqi4KU9Z1SsbkWlFOtrBSKySday2kzjO0qc5TZUGJDGSdA5WpmCKZSE-zMbs_9po4wQ9UV_vBtTj8ViCq_0MqqE6HRPbuyKJvqdr1h6GL0yJYgiwg-wOHtFyL</recordid><startdate>20210301</startdate><enddate>20210301</enddate><creator>van Beek, Anton</creator><creator>Ghumman, Umar Farooq</creator><creator>Munshi, Joydeep</creator><creator>Tao, Siyu</creator><creator>Chien, TeYu</creator><creator>Balasubramanian, Ganesh</creator><creator>Plumlee, Matthew</creator><creator>Apley, Daniel</creator><creator>Chen, Wei</creator><scope>AAYXX</scope><scope>CITATION</scope></search><sort><creationdate>20210301</creationdate><title>Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise</title><author>van Beek, Anton ; Ghumman, Umar Farooq ; Munshi, Joydeep ; Tao, Siyu ; Chien, TeYu ; Balasubramanian, Ganesh ; Plumlee, Matthew ; Apley, Daniel ; Chen, Wei</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a250t-5f88c9ff2ecc7b5e7ed8d4b5aeb7bb04b73ad2b725d150314f71e9206aec5e763</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Design for Manufacturing</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>van Beek, Anton</creatorcontrib><creatorcontrib>Ghumman, Umar Farooq</creatorcontrib><creatorcontrib>Munshi, Joydeep</creatorcontrib><creatorcontrib>Tao, Siyu</creatorcontrib><creatorcontrib>Chien, TeYu</creatorcontrib><creatorcontrib>Balasubramanian, Ganesh</creatorcontrib><creatorcontrib>Plumlee, Matthew</creatorcontrib><creatorcontrib>Apley, Daniel</creatorcontrib><creatorcontrib>Chen, Wei</creatorcontrib><collection>CrossRef</collection><jtitle>Journal of mechanical design (1990)</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>van Beek, Anton</au><au>Ghumman, Umar Farooq</au><au>Munshi, Joydeep</au><au>Tao, Siyu</au><au>Chien, TeYu</au><au>Balasubramanian, Ganesh</au><au>Plumlee, Matthew</au><au>Apley, Daniel</au><au>Chen, Wei</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise</atitle><jtitle>Journal of mechanical design (1990)</jtitle><stitle>J. Mech. Des</stitle><date>2021-03-01</date><risdate>2021</risdate><volume>143</volume><issue>3</issue><issn>1050-0472</issn><eissn>1528-9001</eissn><abstract>In this study, we propose a scalable batch sampling scheme for optimization of simulation models with spatially varying noise. The proposed scheme has two primary advantages: (i) reduced simulation cost by recommending batches of samples at carefully selected spatial locations and (ii) improved scalability by actively considering replicating at previously observed sampling locations. Replication improves the scalability of the proposed sampling scheme as the computational cost of adaptive sampling schemes grow cubicly with the number of unique sampling locations. Our main consideration for the allocation of computational resources is the minimization of the uncertainty in the optimal design. We analytically derive the relationship between the “exploration versus replication decision” and the posterior variance of the spatial random process used to approximate the simulation model’s mean response. Leveraging this reformulation in a novel objective-driven adaptive sampling scheme, we show that we can identify batches of samples that minimize the prediction uncertainty only in the regions of the design space expected to contain the global optimum. Finally, the proposed sampling scheme adopts a modified preposterior analysis that uses a zeroth-order interpolation of the spatially varying simulation noise to identify sampling batches. Through the optimization of three numerical test functions and one engineering problem, we demonstrate (i) the efficacy and of the proposed sampling scheme to deal with a wide array of stochastic functions, (ii) the superior performance of the proposed method on all test functions compared to existing methods, (iii) the empirical validity of using a zeroth-order approximation for the allocation of sampling batches, and (iv) its applicability to molecular dynamics simulations by optimizing the performance of an organic photovoltaic cell as a function of its processing settings.</abstract><doi>10.1115/1.4049134</doi></addata></record>
fulltext fulltext
identifier ISSN: 1050-0472
ispartof Journal of mechanical design (1990), 2021-03, Vol.143 (3)
issn 1050-0472
1528-9001
language eng
recordid cdi_crossref_primary_10_1115_1_4049134
source Alma/SFX Local Collection; ASME Transactions Journals (Current)
subjects Design for Manufacturing
title Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T20%3A28%3A12IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-asme_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Scalable%20Adaptive%20Batch%20Sampling%20in%20Simulation-Based%20Design%20With%20Heteroscedastic%20Noise&rft.jtitle=Journal%20of%20mechanical%20design%20(1990)&rft.au=van%20Beek,%20Anton&rft.date=2021-03-01&rft.volume=143&rft.issue=3&rft.issn=1050-0472&rft.eissn=1528-9001&rft_id=info:doi/10.1115/1.4049134&rft_dat=%3Casme_cross%3E1091481%3C/asme_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