Lightweight equivariant model for efficient interatomic potential predictions

Lightweight equivariant model for efficient interatomic potential predictions | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Lightweight equivariant model for efficient interatomic potential predictions Lei Shen, Ziduo Yang, Xian Wang, Yifan Li, Qiujie Lv, Calvin Chen This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3829677/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 26 Feb, 2025 Read the published version in npj Computational Materials → Version 1 posted 11 You are reading this latest preprint version Abstract In modern computational materials science, deep learning has shown the capability to predict interatomic potentials, thereby supporting and accelerating conventional simulations. However, existing models typically sacrifice either accuracy or efficiency. Moreover, lightweight models are highly demanded for offering simulating systems on a considerably larger scale at reduced computational costs. Here, we introduce a lightweight equivariant interaction graph neural network (LEIGNN) that can enable accurate and efficient interatomic potential and force predictions for molecules and crystals. Rather than relying on higher-order representations, LEIGNN employs a scalar-vector dual representation to encode equivariant features. By extracting both local and global structures from vector representations and learning geometric symmetry information, our model remains lightweight while ensuring prediction accuracy and robustness through the equivariance. Our results show that LEIGNN consistently outperforms the prediction performance of the representative baselines and achieves significant efficiency across diverse datasets, which include catalysts, molecules, and organic isomers. Finally, we conduct a comparative analysis of LEIGNN against both classical molecular dynamics (MD) and ab initio MD simulations across solid, liquid, and gas systems. It is found that LEIGNN can achieve the accuracy of ab initio MD and retain the computational efficiency of classical MD across all examined systems, demonstrating its accuracy, efficiency, and universality. Physical sciences/Physics/Condensed-matter physics/Electronic properties and materials Physical sciences/Materials science/Theory and computation/Atomistic models Full Text Additional Declarations (Not answered) Supplementary Files SupplementaryMaterials.pdf Cite Share Download PDF Status: Published Journal Publication published 26 Feb, 2025 Read the published version in npj Computational Materials → Version 1 posted Editorial decision: revise 13 May, 2024 Review # 2 received at journal 10 Apr, 2024 Review # 1 received at journal 07 Apr, 2024 Review # 3 received at journal 04 Apr, 2024 Reviewer # 3 agreed at journal 02 Apr, 2024 Reviewer # 2 agreed at journal 02 Apr, 2024 Reviewer # 1 agreed at journal 04 Mar, 2024 Reviewers invited by journal 28 Feb, 2024 Submission checks completed at journal 03 Jan, 2024 Editor assigned by journal 02 Jan, 2024 First submitted to journal 02 Jan, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3829677","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":264943726,"identity":"6c4050df-9cf6-4894-9e8d-0767635713a6","order_by":0,"name":"Lei Shen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA30lEQVRIiWNgGAWjYDACCSB+wCAhB+GxEaslgUHCmGQtDIkNRGuRn9187EFChUX6don0Bwwfyg4zyPcvwK/F4M6xdIOEMxK5O2fkGDDOOHeYweDGAwJaJHLMJBLbJHI33MhhYOZtA2qROEDAYTPyv0kk/pNIN7iR/oD5L1CL/AwCWhhu5LBJJDZIJBjcSDBgZgRqYTjfQMBhN9LMJBKOSRhuOPPG4GDPuXQegxsELJGfkfxM4kNNnbzB8fSHD36UWcvJ9xNyGDIAqeVhkEggQQsE8JNiyygYBaNgFIwEAADJOUURpcXNRgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0001-6198-5753","institution":"National university of singapore","correspondingAuthor":true,"prefix":"","firstName":"Lei","middleName":"","lastName":"Shen","suffix":""},{"id":264943727,"identity":"96e621bc-40fe-4a90-ae1c-26b773cc79df","order_by":1,"name":"Ziduo Yang","email":"","orcid":"","institution":"National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"Ziduo","middleName":"","lastName":"Yang","suffix":""},{"id":264943728,"identity":"0fbe8acf-ac8d-41a7-b661-856c6d7f3553","order_by":2,"name":"Xian Wang","email":"","orcid":"","institution":"National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"Xian","middleName":"","lastName":"Wang","suffix":""},{"id":264943729,"identity":"76cf096e-59b6-417f-96f8-cf02cb092bb4","order_by":3,"name":"Yifan Li","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yifan","middleName":"","lastName":"Li","suffix":""},{"id":264943730,"identity":"0750e3c8-13da-4a8b-8a91-75be3289e627","order_by":4,"name":"Qiujie Lv","email":"","orcid":"","institution":"National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"Qiujie","middleName":"","lastName":"Lv","suffix":""},{"id":264943731,"identity":"fca40ed9-2144-44e8-b710-20a6448f3912","order_by":5,"name":"Calvin Chen","email":"","orcid":"https://orcid.org/0000-0001-9213-9832","institution":"Peking University Shenzhen Graduate School","correspondingAuthor":false,"prefix":"","firstName":"Calvin","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2024-01-02 14:11:39","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3829677/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3829677/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41524-025-01535-3","type":"published","date":"2025-02-26T05:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":77300047,"identity":"426624d8-4aff-403e-b9af-5e74072e08ff","added_by":"auto","created_at":"2025-02-27 08:16:00","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4280329,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3829677/v1_covered_13ba04dd-cf0a-4a48-acee-de0ca59be3fb.pdf"},{"id":51844095,"identity":"d20c4fb9-103f-41c3-8991-a1876e75aca1","added_by":"auto","created_at":"2024-03-01 06:49:17","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2478531,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"SupplementaryMaterials.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3829677/v1/2805894a2dc80e26f7138de7.pdf"}],"financialInterests":"(Not answered)","formattedTitle":"Lightweight equivariant model for efficient interatomic potential\r\npredictions","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-computational-materials","isNatureJournal":false,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"npjcompumats","sideBox":"Learn more about [npj Computational Materials](http://www.nature.com/npjcompumats/)","snPcode":"41524","submissionUrl":"https://mts-npjcompumats.nature.com/","title":"npj Computational Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3829677/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3829677/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"In modern computational materials science, deep learning has shown the capability to predict interatomic potentials, thereby supporting and accelerating conventional simulations. However,\r\nexisting models typically sacrifice either accuracy or efficiency. Moreover, lightweight models are highly demanded for offering simulating systems on a considerably larger scale at reduced computational costs. Here, we introduce a lightweight equivariant interaction graph neural network (LEIGNN) that can enable accurate and efficient interatomic potential and force predictions for molecules and crystals. Rather than relying on higher-order representations, LEIGNN employs a scalar-vector dual representation to encode equivariant features. By extracting both local and global structures from vector representations and learning geometric symmetry information, our\r\nmodel remains lightweight while ensuring prediction accuracy and robustness through the equivariance. Our results show that LEIGNN consistently outperforms the prediction performance of the representative baselines and achieves significant efficiency across diverse datasets, which include catalysts, molecules, and organic isomers. Finally, we conduct a comparative analysis of LEIGNN against both classical molecular dynamics (MD) and ab initio MD simulations across solid, liquid, and gas systems. It is found that LEIGNN can achieve the accuracy of ab initio MD and retain the computational efficiency of classical MD across all examined systems, demonstrating its accuracy, efficiency, and universality.","manuscriptTitle":"Lightweight equivariant model for efficient interatomic potential\npredictions","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-03-01 06:49:12","doi":"10.21203/rs.3.rs-3829677/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"revise","date":"2024-05-13T09:24:12+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-04-10T10:13:58+00:00","index":2,"fulltext":"This content is not available."},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-04-08T02:41:24+00:00","index":1,"fulltext":"This content is not available."},{"type":"editorInvitedReview","content":"This content is not available.","date":"2024-04-04T16:28:56+00:00","index":3,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-04-02T08:28:08+00:00","index":3,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-04-02T06:51:58+00:00","index":2,"fulltext":"This content is not available."},{"type":"reviewerAgreed","content":"This content is not available.","date":"2024-03-04T16:44:33+00:00","index":1,"fulltext":"This content is not available."},{"type":"reviewersInvited","content":"","date":"2024-02-28T21:04:08+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-01-03T08:56:45+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-01-02T14:06:51+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Computational Materials","date":"2024-01-02T14:06:50+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-computational-materials","isNatureJournal":false,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"npjcompumats","sideBox":"Learn more about [npj Computational Materials](http://www.nature.com/npjcompumats/)","snPcode":"41524","submissionUrl":"https://mts-npjcompumats.nature.com/","title":"npj Computational Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4bd547dd-c28e-40c2-8e3b-d346c247e50e","owner":[],"postedDate":"March 1st, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":27919689,"name":"Physical sciences/Physics/Condensed-matter physics/Electronic properties and materials"},{"id":27919690,"name":"Physical sciences/Materials science/Theory and computation/Atomistic models"}],"tags":[],"updatedAt":"2025-02-27T08:15:54+00:00","versionOfRecord":{"articleIdentity":"rs-3829677","link":"https://doi.org/10.1038/s41524-025-01535-3","journal":{"identity":"npj-computational-materials","isVorOnly":false,"title":"npj Computational Materials"},"publishedOn":"2025-02-26 05:00:00","publishedOnDateReadable":"February 26th, 2025"},"versionCreatedAt":"2024-03-01 06:49:12","video":"","vorDoi":"10.1038/s41524-025-01535-3","vorDoiUrl":"https://doi.org/10.1038/s41524-025-01535-3","workflowStages":[]},"version":"v1","identity":"rs-3829677","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3829677","identity":"rs-3829677","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}