Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling
Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure...
Gespeichert in:
Veröffentlicht in: | arXiv.org 2023-06 |
---|---|
Hauptverfasser: | , , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | |
---|---|
container_issue | |
container_start_page | |
container_title | arXiv.org |
container_volume | |
creator | Daw, Arka Bu, Jie Wang, Sifan Perdikaris, Paris Karpatne, Anuj |
description | Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure modes" of PINNs, although a thorough understanding of the connection between PINN failure modes and sampling strategies is missing. In this paper, we provide a novel perspective of failure modes of PINNs by hypothesizing that training PINNs relies on successful "propagation" of solution from initial and/or boundary condition points to interior points. We show that PINNs with poor sampling strategies can get stuck at trivial solutions if there are propagation failures, characterized by highly imbalanced PDE residual fields. To mitigate propagation failures, we propose a novel Retain-Resample-Release sampling (R3) algorithm that can incrementally accumulate collocation points in regions of high PDE residuals with little to no computational overhead. We provide an extension of R3 sampling to respect the principle of causality while solving time-dependent PDEs. We theoretically analyze the behavior of R3 sampling and empirically demonstrate its efficacy and efficiency in comparison with baselines on a variety of PDE problems. |
format | Article |
fullrecord | <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2685818629</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2685818629</sourcerecordid><originalsourceid>FETCH-proquest_journals_26858186293</originalsourceid><addsrcrecordid>eNqNi0EKwjAURIMgWLR3CLjRRaFNbI1rsbhRSnVfgn41GpOanyDe3hY8gKt5zLwZkIhxniViwdiIxIj3NE1ZsWR5ziNid8qrq_TKXGnlbCt7toaWUungAKkytLp9UJ0wUeZi3RPOdA_BSd2Ff1v3QBqwv9fgpTJJDSifrYYONEgEOqv5nB76rrMmZHiRGiH-5ZhMy81xvU1aZ18B0Dd3G5zppoYVIheZKNiK_2d9AZ3GSls</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2685818629</pqid></control><display><type>article</type><title>Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling</title><source>Free E- Journals</source><creator>Daw, Arka ; Bu, Jie ; Wang, Sifan ; Perdikaris, Paris ; Karpatne, Anuj</creator><creatorcontrib>Daw, Arka ; Bu, Jie ; Wang, Sifan ; Perdikaris, Paris ; Karpatne, Anuj</creatorcontrib><description>Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure modes" of PINNs, although a thorough understanding of the connection between PINN failure modes and sampling strategies is missing. In this paper, we provide a novel perspective of failure modes of PINNs by hypothesizing that training PINNs relies on successful "propagation" of solution from initial and/or boundary condition points to interior points. We show that PINNs with poor sampling strategies can get stuck at trivial solutions if there are propagation failures, characterized by highly imbalanced PDE residual fields. To mitigate propagation failures, we propose a novel Retain-Resample-Release sampling (R3) algorithm that can incrementally accumulate collocation points in regions of high PDE residuals with little to no computational overhead. We provide an extension of R3 sampling to respect the principle of causality while solving time-dependent PDEs. We theoretically analyze the behavior of R3 sampling and empirically demonstrate its efficacy and efficiency in comparison with baselines on a variety of PDE problems.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Boundary conditions ; Collocation ; Neural networks ; Partial differential equations ; Propagation ; Sampling ; Training</subject><ispartof>arXiv.org, 2023-06</ispartof><rights>2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>776,780</link.rule.ids></links><search><creatorcontrib>Daw, Arka</creatorcontrib><creatorcontrib>Bu, Jie</creatorcontrib><creatorcontrib>Wang, Sifan</creatorcontrib><creatorcontrib>Perdikaris, Paris</creatorcontrib><creatorcontrib>Karpatne, Anuj</creatorcontrib><title>Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling</title><title>arXiv.org</title><description>Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure modes" of PINNs, although a thorough understanding of the connection between PINN failure modes and sampling strategies is missing. In this paper, we provide a novel perspective of failure modes of PINNs by hypothesizing that training PINNs relies on successful "propagation" of solution from initial and/or boundary condition points to interior points. We show that PINNs with poor sampling strategies can get stuck at trivial solutions if there are propagation failures, characterized by highly imbalanced PDE residual fields. To mitigate propagation failures, we propose a novel Retain-Resample-Release sampling (R3) algorithm that can incrementally accumulate collocation points in regions of high PDE residuals with little to no computational overhead. We provide an extension of R3 sampling to respect the principle of causality while solving time-dependent PDEs. We theoretically analyze the behavior of R3 sampling and empirically demonstrate its efficacy and efficiency in comparison with baselines on a variety of PDE problems.</description><subject>Boundary conditions</subject><subject>Collocation</subject><subject>Neural networks</subject><subject>Partial differential equations</subject><subject>Propagation</subject><subject>Sampling</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNi0EKwjAURIMgWLR3CLjRRaFNbI1rsbhRSnVfgn41GpOanyDe3hY8gKt5zLwZkIhxniViwdiIxIj3NE1ZsWR5ziNid8qrq_TKXGnlbCt7toaWUungAKkytLp9UJ0wUeZi3RPOdA_BSd2Ff1v3QBqwv9fgpTJJDSifrYYONEgEOqv5nB76rrMmZHiRGiH-5ZhMy81xvU1aZ18B0Dd3G5zppoYVIheZKNiK_2d9AZ3GSls</recordid><startdate>20230607</startdate><enddate>20230607</enddate><creator>Daw, Arka</creator><creator>Bu, Jie</creator><creator>Wang, Sifan</creator><creator>Perdikaris, Paris</creator><creator>Karpatne, Anuj</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20230607</creationdate><title>Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling</title><author>Daw, Arka ; Bu, Jie ; Wang, Sifan ; Perdikaris, Paris ; Karpatne, Anuj</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26858186293</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Boundary conditions</topic><topic>Collocation</topic><topic>Neural networks</topic><topic>Partial differential equations</topic><topic>Propagation</topic><topic>Sampling</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Daw, Arka</creatorcontrib><creatorcontrib>Bu, Jie</creatorcontrib><creatorcontrib>Wang, Sifan</creatorcontrib><creatorcontrib>Perdikaris, Paris</creatorcontrib><creatorcontrib>Karpatne, Anuj</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science & Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Daw, Arka</au><au>Bu, Jie</au><au>Wang, Sifan</au><au>Perdikaris, Paris</au><au>Karpatne, Anuj</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling</atitle><jtitle>arXiv.org</jtitle><date>2023-06-07</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>Despite the success of physics-informed neural networks (PINNs) in approximating partial differential equations (PDEs), PINNs can sometimes fail to converge to the correct solution in problems involving complicated PDEs. This is reflected in several recent studies on characterizing the "failure modes" of PINNs, although a thorough understanding of the connection between PINN failure modes and sampling strategies is missing. In this paper, we provide a novel perspective of failure modes of PINNs by hypothesizing that training PINNs relies on successful "propagation" of solution from initial and/or boundary condition points to interior points. We show that PINNs with poor sampling strategies can get stuck at trivial solutions if there are propagation failures, characterized by highly imbalanced PDE residual fields. To mitigate propagation failures, we propose a novel Retain-Resample-Release sampling (R3) algorithm that can incrementally accumulate collocation points in regions of high PDE residuals with little to no computational overhead. We provide an extension of R3 sampling to respect the principle of causality while solving time-dependent PDEs. We theoretically analyze the behavior of R3 sampling and empirically demonstrate its efficacy and efficiency in comparison with baselines on a variety of PDE problems.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record> |
fulltext | fulltext |
identifier | EISSN: 2331-8422 |
ispartof | arXiv.org, 2023-06 |
issn | 2331-8422 |
language | eng |
recordid | cdi_proquest_journals_2685818629 |
source | Free E- Journals |
subjects | Boundary conditions Collocation Neural networks Partial differential equations Propagation Sampling Training |
title | Mitigating Propagation Failures in Physics-informed Neural Networks using Retain-Resample-Release (R3) Sampling |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-27T08%3A05%3A03IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Mitigating%20Propagation%20Failures%20in%20Physics-informed%20Neural%20Networks%20using%20Retain-Resample-Release%20(R3)%20Sampling&rft.jtitle=arXiv.org&rft.au=Daw,%20Arka&rft.date=2023-06-07&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2685818629%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2685818629&rft_id=info:pmid/&rfr_iscdi=true |