{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,8]],"date-time":"2025-12-08T22:37:35Z","timestamp":1765233455487,"version":"3.37.3"},"reference-count":19,"publisher":"IOP Publishing","issue":"1","license":[{"start":{"date-parts":[[2023,1,18]],"date-time":"2023-01-18T00:00:00Z","timestamp":1674000000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2023,1,18]],"date-time":"2023-01-18T00:00:00Z","timestamp":1674000000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2023,3,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Physical systems are commonly represented as a combination of particles, the individual dynamics of which govern the system dynamics. However, traditional approaches require the knowledge of several abstract quantities such as the energy or force to infer the dynamics of these particles. Here, we present a framework, namely, Lagrangian graph neural network (<jats:sc>LGnn<\/jats:sc>), that provides a strong inductive bias to learn the Lagrangian of a particle-based system directly from the trajectory. We test our approach on challenging systems with constraints and drag\u2014<jats:sc>LGnn<\/jats:sc> outperforms baselines such as feed-forward Lagrangian neural network (<jats:sc>Lnn<\/jats:sc>) with improved performance. We also show the <jats:italic>zero-shot<\/jats:italic> generalizability of the system by simulating systems two orders of magnitude larger than the trained one and also hybrid systems that are unseen by the model, a unique feature. The graph architecture of <jats:sc>LGnn<\/jats:sc> significantly simplifies the learning in comparison to <jats:sc>Lnn<\/jats:sc> with \u223c25 times better performance on \u223c20 times smaller amounts of data. Finally, we show the interpretability of <jats:sc>LGnn<\/jats:sc>, which directly provides physical insights on drag and constraint forces learned by the model. <jats:sc>LGnn<\/jats:sc> can thus provide a fillip toward understanding the dynamics of physical systems purely from observable quantities.<\/jats:p>","DOI":"10.1088\/2632-2153\/acb03e","type":"journal-article","created":{"date-parts":[[2023,1,4]],"date-time":"2023-01-04T22:43:29Z","timestamp":1672872209000},"page":"015003","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":11,"title":["Learning the dynamics of particle-based systems with Lagrangian graph neural networks"],"prefix":"10.1088","volume":"4","author":[{"given":"Ravinder","family":"Bhattoo","sequence":"first","affiliation":[]},{"given":"Sayan","family":"Ranu","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1500-4947","authenticated-orcid":true,"given":"N M Anoop","family":"Krishnan","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2023,1,18]]},"reference":[{"year":"2006","author":"LaValle","key":"mlstacb03ebib1"},{"year":"2011","author":"Goldstein","key":"mlstacb03ebib2"},{"key":"mlstacb03ebib3","doi-asserted-by":"publisher","first-page":"422","DOI":"10.1038\/s42254-021-00314-5","article-title":"Physics-informed machine learning","volume":"3","author":"Em Karniadakis","year":"2021","journal-title":"Nat. Rev. Phys."},{"key":"mlstacb03ebib4","first-page":"pp 15379","article-title":"Hamiltonian neural networks","volume":"vol 32","author":"Greydanus","year":"2019"},{"article-title":"Hamiltonian graph networks with ode integrators","year":"2019","author":"Sanchez-Gonzalez","key":"mlstacb03ebib5"},{"article-title":"Lagrangian neural networks","year":"2020","author":"Cranmer","key":"mlstacb03ebib6"},{"key":"mlstacb03ebib7","first-page":"p 33","article-title":"Discovering symbolic models from deep learning with inductive biases","author":"Cranmer","year":"2020"},{"key":"mlstacb03ebib8","first-page":"pp 13880","article-title":"Simplifying Hamiltonian and Lagrangian neural networks via explicit constraints","volume":"vol 33","author":"Finzi","year":"2020"},{"article-title":"Deep Lagrangian networks: using physics as model prior for deep learning","year":"2019","author":"Lutter","key":"mlstacb03ebib9"},{"key":"mlstacb03ebib10","first-page":"pp 1218","article-title":"Benchmarking energy-conserving neural networks for learning dynamics from data","author":"Zhong","year":"2021"},{"key":"mlstacb03ebib11","doi-asserted-by":"publisher","first-page":"9195","DOI":"10.1016\/j.ifacol.2020.12.2182","article-title":"Modeling system dynamics with physics-informed neural networks based on Lagrangian mechanics","volume":"53","author":"Roehrl","year":"2020","journal-title":"IFAC-PapersOnLine"},{"article-title":"Dissipative symoden: encoding Hamiltonian dynamics with dissipation and control into deep learning","year":"2020","author":"Zhong","key":"mlstacb03ebib12"},{"key":"mlstacb03ebib13","article-title":"Extending Lagrangian and Hamiltonian neural networks with differentiable contact models","volume":"vol 34","author":"Zhong","year":"2021"},{"key":"mlstacb03ebib14","first-page":"p 33","article-title":"Unsupervised learning of Lagrangian dynamics from images for prediction and control","author":"Zhong","year":"2020"},{"article-title":"Deconstructing the inductive biases of Hamiltonian neural networks","year":"2021","author":"Gruver","key":"mlstacb03ebib15"},{"key":"mlstacb03ebib16","first-page":"pp 8459","article-title":"Learning to simulate complex physics with graph networks","author":"Sanchez-Gonzalez","year":"2020"},{"key":"mlstacb03ebib17","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/s41524-021-00543-3","article-title":"Accurate and scalable graph neural network force field and molecular dynamics with direct force architecture","volume":"7","author":"Park","year":"2021","journal-title":"npj Comput. Mater."},{"key":"mlstacb03ebib18","first-page":"p 33","article-title":"Jax md: a framework for differentiable physics","author":"","year":"2020"},{"article-title":"Jraph: a library for graph neural networks in jax","year":"2020","author":"Godwin","key":"mlstacb03ebib19"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,1,18]],"date-time":"2023-01-18T11:52:46Z","timestamp":1674042766000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/acb03e"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,1,18]]},"references-count":19,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2023,1,18]]},"published-print":{"date-parts":[[2023,3,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/acb03e","relation":{},"ISSN":["2632-2153"],"issn-type":[{"type":"electronic","value":"2632-2153"}],"subject":[],"published":{"date-parts":[[2023,1,18]]},"assertion":[{"value":"Learning the dynamics of particle-based systems with Lagrangian graph neural networks","name":"article_title","label":"Article Title"},{"value":"Machine Learning: Science and Technology","name":"journal_title","label":"Journal Title"},{"value":"paper","name":"article_type","label":"Article Type"},{"value":"\u00a9 2023 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2022-09-02","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-01-04","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2023-01-18","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}