{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,5]],"date-time":"2026-05-05T20:17:41Z","timestamp":1778012261987,"version":"3.51.4"},"reference-count":47,"publisher":"IOP Publishing","issue":"2","license":[{"start":{"date-parts":[[2022,5,30]],"date-time":"2022-05-30T00:00:00Z","timestamp":1653868800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2022,5,30]],"date-time":"2022-05-30T00:00:00Z","timestamp":1653868800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/501100001659","name":"German Research Foundation","doi-asserted-by":"crossref","award":["EXC 2046\/1"],"award-info":[{"award-number":["EXC 2046\/1"]}],"id":[{"id":"10.13039\/501100001659","id-type":"DOI","asserted-by":"crossref"}]},{"name":"BMBF","award":["01GQ0850"],"award-info":[{"award-number":["01GQ0850"]}]},{"name":"Arti\ufb01cial Intelligence Graduate School Program, Korea University"},{"name":"Korea Government"},{"name":"Institute of Information"}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2022,6,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:p>Molecular dynamics (MD) simulations are a cornerstone in science, enabling the investigation of a system\u2019s thermodynamics all the way to analyzing intricate molecular interactions. In general, creating extended molecular trajectories can be a computationally expensive process, for example, when running <jats:italic>ab-initio<\/jats:italic> simulations. Hence, repeating such calculations to either obtain more accurate thermodynamics or to get a higher resolution in the dynamics generated by a fine-grained quantum interaction can be time- and computational resource-consuming. In this work, we explore different machine learning methodologies to increase the resolution of MD trajectories on-demand within a post-processing step. As a proof of concept, we analyse the performance of bi-directional neural networks (NNs) such as neural ODEs, Hamiltonian networks, recurrent NNs and long short-term memories, as well as the uni-directional variants as a reference, for MD simulations (here: the MD17 dataset). We have found that Bi-LSTMs are the best performing models; by utilizing the local time-symmetry of thermostated trajectories they can even learn long-range correlations and display high robustness to noisy dynamics across molecular complexity. Our models can reach accuracies of up to 10<jats:sup>\u22124<\/jats:sup> \u00c5 in trajectory interpolation, which leads to the faithful reconstruction of several unseen high-frequency molecular vibration cycles. This renders the comparison between the learned and reference trajectories indistinguishable. The results reported in this work can serve (1) as a baseline for larger systems, as well as (2) for the construction of better MD integrators.<\/jats:p>","DOI":"10.1088\/2632-2153\/ac6ec6","type":"journal-article","created":{"date-parts":[[2022,5,11]],"date-time":"2022-05-11T22:42:35Z","timestamp":1652308955000},"page":"025011","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":12,"title":["High-fidelity molecular dynamics trajectory reconstruction with bi-directional neural networks"],"prefix":"10.1088","volume":"3","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1354-4715","authenticated-orcid":false,"given":"Ludwig","family":"Winkler","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3861-7685","authenticated-orcid":false,"given":"Klaus-Robert","family":"M\u00fcller","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6091-3408","authenticated-orcid":true,"given":"Huziel E","family":"Sauceda","sequence":"additional","affiliation":[]}],"member":"266","published-online":{"date-parts":[[2022,5,30]]},"reference":[{"key":"mlstac6ec6bib1","author":"Tuckerman","year":"2010"},{"key":"mlstac6ec6bib2","doi-asserted-by":"publisher","first-page":"361","DOI":"10.1146\/annurev-physchem-042018-052331","volume":"71","author":"No\u00e9","year":"2020","journal-title":"Annu. Rev. Phys. Chem."},{"key":"mlstac6ec6bib3","doi-asserted-by":"publisher","DOI":"10.1021\/acs.jpclett.1c02954","volume":"12","author":"Sadeghi","year":"2021","journal-title":"J. Phys. Chem. Lett."},{"key":"mlstac6ec6bib4","doi-asserted-by":"publisher","DOI":"10.1002\/anie.201703114","volume":"56","author":"Behler","year":"2017","journal-title":"Angew. Chem., Int. Ed."},{"key":"mlstac6ec6bib5","first-page":"pp 991","author":"Sch\u00fctt","year":"2017"},{"key":"mlstac6ec6bib6","doi-asserted-by":"publisher","DOI":"10.1063\/1.5019779","volume":"148","author":"Sch\u00fctt","year":"2018","journal-title":"J. Chem. Phys."},{"key":"mlstac6ec6bib7","doi-asserted-by":"publisher","first-page":"629","DOI":"10.4208\/cicp.OA-2017-0213","volume":"23","author":"Han","year":"2018","journal-title":"Commun. Comput. Phys."},{"key":"mlstac6ec6bib8","doi-asserted-by":"publisher","DOI":"10.1126\/sciadv.1603015","volume":"3","author":"Chmiela","year":"2017","journal-title":"Sci. Adv."},{"key":"mlstac6ec6bib9","doi-asserted-by":"publisher","first-page":"3887","DOI":"10.1038\/s41467-018-06169-2","volume":"9","author":"Chmiela","year":"2018","journal-title":"Nat. Commun."},{"key":"mlstac6ec6bib10","doi-asserted-by":"publisher","DOI":"10.1088\/2632-2153\/abc9fe","volume":"2","author":"Novikov","year":"2021","journal-title":"Mach. Learn.: Sci. Technol."},{"key":"mlstac6ec6bib11","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.104.136403","volume":"104","author":"Bart\u00f3k","year":"2010","journal-title":"Phys. Rev. Lett."},{"key":"mlstac6ec6bib12","doi-asserted-by":"publisher","DOI":"10.1063\/1.5024558","volume":"148","author":"Zeni","year":"2018","journal-title":"J. Chem. Phys."},{"key":"mlstac6ec6bib13","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.97.184307","volume":"97","author":"Glielmo","year":"2018","journal-title":"Phys. Rev. B"},{"key":"mlstac6ec6bib14","doi-asserted-by":"publisher","DOI":"10.1021\/acs.chemrev.0c01111","volume":"121","author":"Unke","year":"2021","journal-title":"Chem. Rev."},{"key":"mlstac6ec6bib15","doi-asserted-by":"publisher","first-page":"347","DOI":"10.1038\/s41570-020-0189-9","volume":"4","author":"von Lilienfeld","year":"2020","journal-title":"Nat. Rev. Chem."},{"key":"mlstac6ec6bib16","doi-asserted-by":"publisher","first-page":"4125","DOI":"10.1038\/s41467-020-17844-8","volume":"11","author":"Tkatchenko","year":"2020","journal-title":"Nat. Commun."},{"key":"mlstac6ec6bib17","doi-asserted-by":"publisher","first-page":"9759","DOI":"10.1021\/acs.chemrev.1c00021","volume":"121","author":"Musil","year":"2021","journal-title":"Chem. Rev."},{"key":"mlstac6ec6bib18","doi-asserted-by":"publisher","first-page":"9816","DOI":"10.1021\/acs.chemrev.1c00107","volume":"121","author":"Keith","year":"2021","journal-title":"Chem. Rev."},{"key":"mlstac6ec6bib19","doi-asserted-by":"publisher","first-page":"7273","DOI":"10.1038\/s41467-021-27504-0","volume":"12","author":"Unke","year":"2021","journal-title":"Nat. Commun."},{"key":"mlstac6ec6bib20","author":"Sauceda","year":"2021"},{"key":"mlstac6ec6bib21","article-title":"Graph neural networks accelerated molecular dynamics","author":"Li","year":"2021"},{"key":"mlstac6ec6bib22","doi-asserted-by":"publisher","first-page":"442","DOI":"10.1038\/s41467-020-20212-1","volume":"12","author":"Sauceda","year":"2021","journal-title":"Nat. Commun."},{"key":"mlstac6ec6bib23","doi-asserted-by":"publisher","first-page":"5115","DOI":"10.1038\/s41467-020-18959-8","volume":"11","author":"Tsai","year":"2020","journal-title":"Nat. Commun."},{"key":"mlstac6ec6bib24","author":"Thaler","year":"2021"},{"key":"mlstac6ec6bib25","article-title":"Multiscale simulations of complex systems by learning their effective dynamics","author":"Vlachas","year":"2021"},{"key":"mlstac6ec6bib26","article-title":"Accelerated simulations of molecular systems through learning of their effective dynamics","author":"Vlachas","year":"2021"},{"key":"mlstac6ec6bib27","author":"Kadupitiya","year":"2020"},{"key":"mlstac6ec6bib28","volume":"31","author":"Chen","year":"2018"},{"key":"mlstac6ec6bib29","first-page":"pp 15379","author":"Greydanus","year":"2019"},{"key":"mlstac6ec6bib30","author":"Cranmer","year":"2020"},{"key":"mlstac6ec6bib31","doi-asserted-by":"publisher","DOI":"10.1088\/1367-2630\/ac3261","volume":"23","author":"Ullah","year":"2021","journal-title":"New J. Phys."},{"key":"mlstac6ec6bib32","doi-asserted-by":"publisher","first-page":"38","DOI":"10.1016\/j.cpc.2019.02.007","volume":"240","author":"Chmiela","year":"2019","journal-title":"Comput. Phys. Commun."},{"key":"mlstac6ec6bib33","doi-asserted-by":"publisher","first-page":"2355","DOI":"10.1021\/acs.jctc.0c01343","volume":"17","author":"Doerr","year":"2021","journal-title":"J. Chem. Theory Comput."},{"key":"mlstac6ec6bib34","doi-asserted-by":"publisher","first-page":"448","DOI":"10.1021\/acs.jctc.8b00908","volume":"15","author":"Sch\u00fctt","year":"2019","journal-title":"J. Chem. Theory Comput."},{"key":"mlstac6ec6bib35","doi-asserted-by":"publisher","DOI":"10.1063\/1.5078687","volume":"150","author":"Sauceda","year":"2019","journal-title":"J. Chem. Phys."},{"key":"mlstac6ec6bib36","doi-asserted-by":"publisher","first-page":"1735","DOI":"10.1162\/neco.1997.9.8.1735","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural Comput."},{"key":"mlstac6ec6bib37","doi-asserted-by":"publisher","first-page":"19","DOI":"10.1016\/0771-050X(80)90013-3","volume":"6","author":"Dormand","year":"1980","journal-title":"J. Comput. Appl. Math."},{"key":"mlstac6ec6bib38","first-page":"pp 1310","author":"Pascanu","year":"2013"},{"key":"mlstac6ec6bib39","author":"Kingma","year":"2014"},{"key":"mlstac6ec6bib40","doi-asserted-by":"publisher","DOI":"10.1063\/5.0023005","volume":"153","author":"Sauceda","year":"2020","journal-title":"J. Chem. Phys."},{"key":"mlstac6ec6bib41","doi-asserted-by":"publisher","first-page":"3865","DOI":"10.1103\/PhysRevLett.77.3865","volume":"77","author":"Perdew","year":"1996","journal-title":"Phys. Rev. Lett."},{"key":"mlstac6ec6bib42","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.102.073005","volume":"102","author":"Tkatchenko","year":"2009","journal-title":"Phys. Rev. Lett."},{"key":"mlstac6ec6bib43","first-page":"pp 5998","author":"Vaswani","year":"2017"},{"key":"mlstac6ec6bib44","author":"Bahdanau","year":"2014"},{"key":"mlstac6ec6bib45","author":"Wu","year":"2016"},{"key":"mlstac6ec6bib46","first-page":"pp 277","author":"Sauceda","year":"2020"},{"key":"mlstac6ec6bib47","first-page":"pp 9377","author":"Sch\u00fctt","year":"2021"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,5,30]],"date-time":"2022-05-30T11:17:17Z","timestamp":1653909437000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ac6ec6"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,5,30]]},"references-count":47,"journal-issue":{"issue":"2","published-online":{"date-parts":[[2022,5,30]]},"published-print":{"date-parts":[[2022,6,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ac6ec6","relation":{},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,5,30]]},"assertion":[{"value":"High-fidelity molecular dynamics trajectory reconstruction with bi-directional 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 2022 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2022-01-06","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2022-05-11","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2022-05-30","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}