Physically-informed artificial neural networks for atomistic modeling of materials

Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the potential energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The eme...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2019-01
Hauptverfasser: Purja Pun, G P, Batra, R, Ramprasad, R, Mishin, Y
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 Purja Pun, G P
Batra, R
Ramprasad, R
Mishin, Y
description Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the potential energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation between the energies in a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. The network adjusts the parameters of the physics-based model on the fly during the simulations according to the local environments of individual atoms. This approach, called the physically-informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. The potential provides a DFT-level accuracy of energy predictions and excellent agreement with experimental and DFT data for a wide range of physical properties. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.
doi_str_mv 10.48550/arxiv.1808.01696
format Article
fullrecord <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_1808_01696</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2092746963</sourcerecordid><originalsourceid>FETCH-LOGICAL-a523-ee9319fedc661f830e556f50954298cc7be8517000187135e9c9cec32aa6d5b33</originalsourceid><addsrcrecordid>eNotj0tLxDAURoMgOIzzA1wZcN2aR5MmSxl8wYAisy-Z9EYzts2YtGr_vZ2Oq7M59_IdhK4oyQslBLk18dd_51QRlRMqtTxDC8Y5zVTB2AVapbQnhDBZMiH4Ar29fozJW9M0Y-Y7F2ILNTax985bbxrcwRBn9D8hfiY8Gdj0ofWp9xa3oYbGd-84ONyaHuJ0ki7RuZsAq38u0fbhfrt-yjYvj8_ru01mBOMZgOZUO6itlNQpTkAI6QTRomBaWVvuQAlaTlupKikXoK22YDkzRtZix_kSXZ_ezsHVIfrWxLE6hldz-GTcnIxDDF8DpL7ahyF206aKEc3K4ijxPyAmXDk</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2092746963</pqid></control><display><type>article</type><title>Physically-informed artificial neural networks for atomistic modeling of materials</title><source>Freely Accessible Journals</source><source>arXiv.org</source><creator>Purja Pun, G P ; Batra, R ; Ramprasad, R ; Mishin, Y</creator><creatorcontrib>Purja Pun, G P ; Batra, R ; Ramprasad, R ; Mishin, Y</creatorcontrib><description>Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the potential energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation between the energies in a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. The network adjusts the parameters of the physics-based model on the fly during the simulations according to the local environments of individual atoms. This approach, called the physically-informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. The potential provides a DFT-level accuracy of energy predictions and excellent agreement with experimental and DFT data for a wide range of physical properties. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.1808.01696</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Artificial neural networks ; Computer simulation ; Interpolation ; Machine learning ; Mathematical models ; Neural networks ; Parameters ; Physical properties ; Physics ; Physics - Materials Science ; Potential energy ; Predictions ; Regression analysis ; Simulation</subject><ispartof>arXiv.org, 2019-01</ispartof><rights>2019. 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><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,784,885,27916</link.rule.ids><backlink>$$Uhttps://doi.org/10.1038/s41467-019-10343-5$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.48550/arXiv.1808.01696$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Purja Pun, G P</creatorcontrib><creatorcontrib>Batra, R</creatorcontrib><creatorcontrib>Ramprasad, R</creatorcontrib><creatorcontrib>Mishin, Y</creatorcontrib><title>Physically-informed artificial neural networks for atomistic modeling of materials</title><title>arXiv.org</title><description>Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the potential energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation between the energies in a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. The network adjusts the parameters of the physics-based model on the fly during the simulations according to the local environments of individual atoms. This approach, called the physically-informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. The potential provides a DFT-level accuracy of energy predictions and excellent agreement with experimental and DFT data for a wide range of physical properties. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.</description><subject>Artificial neural networks</subject><subject>Computer simulation</subject><subject>Interpolation</subject><subject>Machine learning</subject><subject>Mathematical models</subject><subject>Neural networks</subject><subject>Parameters</subject><subject>Physical properties</subject><subject>Physics</subject><subject>Physics - Materials Science</subject><subject>Potential energy</subject><subject>Predictions</subject><subject>Regression analysis</subject><subject>Simulation</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><sourceid>GOX</sourceid><recordid>eNotj0tLxDAURoMgOIzzA1wZcN2aR5MmSxl8wYAisy-Z9EYzts2YtGr_vZ2Oq7M59_IdhK4oyQslBLk18dd_51QRlRMqtTxDC8Y5zVTB2AVapbQnhDBZMiH4Ar29fozJW9M0Y-Y7F2ILNTax985bbxrcwRBn9D8hfiY8Gdj0ofWp9xa3oYbGd-84ONyaHuJ0ki7RuZsAq38u0fbhfrt-yjYvj8_ru01mBOMZgOZUO6itlNQpTkAI6QTRomBaWVvuQAlaTlupKikXoK22YDkzRtZix_kSXZ_ezsHVIfrWxLE6hldz-GTcnIxDDF8DpL7ahyF206aKEc3K4ijxPyAmXDk</recordid><startdate>20190122</startdate><enddate>20190122</enddate><creator>Purja Pun, G P</creator><creator>Batra, R</creator><creator>Ramprasad, R</creator><creator>Mishin, Y</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><scope>GOX</scope></search><sort><creationdate>20190122</creationdate><title>Physically-informed artificial neural networks for atomistic modeling of materials</title><author>Purja Pun, G P ; Batra, R ; Ramprasad, R ; Mishin, Y</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a523-ee9319fedc661f830e556f50954298cc7be8517000187135e9c9cec32aa6d5b33</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Artificial neural networks</topic><topic>Computer simulation</topic><topic>Interpolation</topic><topic>Machine learning</topic><topic>Mathematical models</topic><topic>Neural networks</topic><topic>Parameters</topic><topic>Physical properties</topic><topic>Physics</topic><topic>Physics - Materials Science</topic><topic>Potential energy</topic><topic>Predictions</topic><topic>Regression analysis</topic><topic>Simulation</topic><toplevel>online_resources</toplevel><creatorcontrib>Purja Pun, G P</creatorcontrib><creatorcontrib>Batra, R</creatorcontrib><creatorcontrib>Ramprasad, R</creatorcontrib><creatorcontrib>Mishin, Y</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; 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><collection>arXiv.org</collection><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Purja Pun, G P</au><au>Batra, R</au><au>Ramprasad, R</au><au>Mishin, Y</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Physically-informed artificial neural networks for atomistic modeling of materials</atitle><jtitle>arXiv.org</jtitle><date>2019-01-22</date><risdate>2019</risdate><eissn>2331-8422</eissn><abstract>Large-scale atomistic computer simulations of materials heavily rely on interatomic potentials predicting the potential energy and Newtonian forces on atoms. Traditional interatomic potentials are based on physical intuition but contain few adjustable parameters and are usually not accurate. The emerging machine-learning (ML) potentials achieve highly accurate interpolation between the energies in a large DFT database but, being purely mathematical constructions, suffer from poor transferability to unknown structures. We propose a new approach that can drastically improve the transferability of ML potentials by informing them of the physical nature of interatomic bonding. This is achieved by combining a rather general physics-based model (analytical bond-order potential) with a neural-network regression. The network adjusts the parameters of the physics-based model on the fly during the simulations according to the local environments of individual atoms. This approach, called the physically-informed neural network (PINN) potential, is demonstrated by developing a general-purpose PINN potential for Al. The potential provides a DFT-level accuracy of energy predictions and excellent agreement with experimental and DFT data for a wide range of physical properties. We suggest that the development of physics-based ML potentials is the most effective way forward in the field of atomistic simulations.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.1808.01696</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2019-01
issn 2331-8422
language eng
recordid cdi_arxiv_primary_1808_01696
source Freely Accessible Journals; arXiv.org
subjects Artificial neural networks
Computer simulation
Interpolation
Machine learning
Mathematical models
Neural networks
Parameters
Physical properties
Physics
Physics - Materials Science
Potential energy
Predictions
Regression analysis
Simulation
title Physically-informed artificial neural networks for atomistic modeling of materials
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-15T02%3A59%3A44IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_arxiv&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Physically-informed%20artificial%20neural%20networks%20for%20atomistic%20modeling%20of%20materials&rft.jtitle=arXiv.org&rft.au=Purja%20Pun,%20G%20P&rft.date=2019-01-22&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.1808.01696&rft_dat=%3Cproquest_arxiv%3E2092746963%3C/proquest_arxiv%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2092746963&rft_id=info:pmid/&rfr_iscdi=true