Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV

Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV | 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 Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV Allen Wang, Alessandro Pau, Cristina Rea, Oswin So, Charles Dawson, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6050911/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 06 Oct, 2025 Read the published version in Nature Communications → Version 1 posted You are reading this latest preprint version Abstract The rampdown in tokamak operations is a difficult to simulate phase during which the plasma is often pushed towards multiple instability limits. To address this challenge, and reduce the risk of disrupting operations, we leverage recent advances in Scientific Machine Learning (SciML) to develop a neural state-space model (NSSM) that predicts plasma dynamics during Tokamak à Configuration Variable (TCV) rampdowns. By integrating simple physics structure and data-driven models, the NSSM efficiently learns plasma dynamics during the rampdown from a modest dataset of 311 pulses with only five pulses in the reactor relevant high performance regime. The NSSM is parallelized across uncertainties, and reinforcement learning (RL) is applied to design trajectories that avoid multiple instability limits with high probability. Experiments at TCV ramping down high performance plasmas show statistically significant improvements in current and energy at plasma termination, with improvements in speed through continuous re-training. A predict-first experiment, increasing plasma current by 20% from baseline, demonstrates the NSSM's ability to make small extrapolations with sufficient accuracy to design trajectories that successfully terminate the pulse. The developed approach paves the way for designing tokamak controls with robustness to considerable uncertainty, and demonstrates the relevance of the SciML approach to learning plasma dynamics for rapidly developing robust trajectories and controls during the incremental campaigns of upcoming burning plasma tokamaks. Physical sciences/Energy science and technology/Nuclear energy/Nuclear fusion and fission Physical sciences/Physics/Plasma physics/Magnetically confined plasmas Full Text Additional Declarations Yes there is potential Competing Interest. This work was funded in part by Commonwealth Fusion Systems. Supplementary Files codeanddata.tar.gz Codebase + Dataset Cite Share Download PDF Status: Published Journal Publication published 06 Oct, 2025 Read the published version in Nature Communications → Version 1 posted 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-6050911","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":427173201,"identity":"c63bf4d4-e84a-44cf-a42b-76d63a6ae4c0","order_by":0,"name":"Allen Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAArklEQVRIiWNgGAWjYDACduYGIGnDwAbi8BClhZkRpCWNgY2NRC2HGRiI1sLfzNj4uLDtfDSffAPjg7dtRGiROMzYbDyz7XZuGxsDs+FcYrQwHGZsk+bdBtbCJs1LjBb5w4ztv3m3nQNpYf9NlBYDoC3MvNsOgG1hJkqLIdAv0rz/koFaEpsl55wjQovc8eaDn3nO2OXObz588MObMiK0IAFwBI2CUTAKRsEooAoAAFDQMGvZsutwAAAAAElFTkSuQmCC","orcid":"","institution":"MIT Plasma Science and Fusion Center","correspondingAuthor":true,"prefix":"","firstName":"Allen","middleName":"","lastName":"Wang","suffix":""},{"id":427173202,"identity":"faa8d322-36a6-4454-9c1b-17415611c327","order_by":1,"name":"Alessandro Pau","email":"","orcid":"https://orcid.org/0000-0002-7122-3346","institution":"École Polytechnique Fédérale de Lausanne (EPFL), Swiss Plasma Center (SPC)","correspondingAuthor":false,"prefix":"","firstName":"Alessandro","middleName":"","lastName":"Pau","suffix":""},{"id":427173203,"identity":"5c6d58e5-0044-4feb-8cb5-d4e33fe39c57","order_by":2,"name":"Cristina Rea","email":"","orcid":"","institution":"MIT Plasma Science and Fusion Center","correspondingAuthor":false,"prefix":"","firstName":"Cristina","middleName":"","lastName":"Rea","suffix":""},{"id":427173204,"identity":"12e115b7-8253-41aa-9ae9-37f4488148f0","order_by":3,"name":"Oswin So","email":"","orcid":"","institution":"MIT LIDS","correspondingAuthor":false,"prefix":"","firstName":"Oswin","middleName":"","lastName":"So","suffix":""},{"id":427173205,"identity":"5886517d-901b-4b7f-83f3-1b9cf9e4d36f","order_by":4,"name":"Charles Dawson","email":"","orcid":"","institution":"MIT LIDS","correspondingAuthor":false,"prefix":"","firstName":"Charles","middleName":"","lastName":"Dawson","suffix":""},{"id":427173206,"identity":"d2fc1afe-2de1-4110-b779-cad636837b58","order_by":5,"name":"Olivier Sauter","email":"","orcid":"https://orcid.org/0000-0002-0099-6675","institution":"École Polytechnique Fédérale de Lausanne (EPFL), Swiss Plasma Center (SPC)","correspondingAuthor":false,"prefix":"","firstName":"Olivier","middleName":"","lastName":"Sauter","suffix":""},{"id":427173207,"identity":"748639be-45c4-4073-99e6-05fde234f483","order_by":6,"name":"Mark Boyer","email":"","orcid":"","institution":"Commonwealth Fusion Systems","correspondingAuthor":false,"prefix":"","firstName":"Mark","middleName":"","lastName":"Boyer","suffix":""},{"id":427173208,"identity":"db51d2c0-b5d0-4e4f-8fe5-8e64a7e549c9","order_by":7,"name":"Anna Vu","email":"","orcid":"","institution":"Swiss Plasma Center, EPFL","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"","lastName":"Vu","suffix":""},{"id":427173209,"identity":"d4a621e1-145c-463d-a957-d1b1176fdb59","order_by":8,"name":"Cristian Galperti","email":"","orcid":"","institution":"École Polytechnique Fédérale de Lausanne","correspondingAuthor":false,"prefix":"","firstName":"Cristian","middleName":"","lastName":"Galperti","suffix":""},{"id":427173210,"identity":"f371df1e-80f5-4e3e-83ce-0d2559b7882a","order_by":9,"name":"Chuchu Fan","email":"","orcid":"","institution":"MIT LIDS","correspondingAuthor":false,"prefix":"","firstName":"Chuchu","middleName":"","lastName":"Fan","suffix":""},{"id":427173211,"identity":"c2c6754f-6ad9-4b6b-b2d7-d39f0322082d","order_by":10,"name":"Antoine Merle","email":"","orcid":"","institution":"Swiss Plasma Center, EPFL","correspondingAuthor":false,"prefix":"","firstName":"Antoine","middleName":"","lastName":"Merle","suffix":""},{"id":427173212,"identity":"72ba8e6b-074a-4b3c-94e0-13c8500ca452","order_by":11,"name":"Yoeri Poels","email":"","orcid":"https://orcid.org/0000-0002-4071-4855","institution":"Swiss Plasma Center, EPFL","correspondingAuthor":false,"prefix":"","firstName":"Yoeri","middleName":"","lastName":"Poels","suffix":""},{"id":427173213,"identity":"99a14b69-898d-402e-be8c-bfd8db46d90f","order_by":12,"name":"Cristina Venturini","email":"","orcid":"","institution":"Swiss Plasma Center, EPFL","correspondingAuthor":false,"prefix":"","firstName":"Cristina","middleName":"","lastName":"Venturini","suffix":""},{"id":427173214,"identity":"2a170056-c50a-4293-9b08-4ae989aba59b","order_by":13,"name":"Stefano Marchioni","email":"","orcid":"","institution":"Swiss Plasma Center, EPFL","correspondingAuthor":false,"prefix":"","firstName":"Stefano","middleName":"","lastName":"Marchioni","suffix":""}],"badges":[],"createdAt":"2025-02-17 21:10:40","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6050911/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6050911/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41467-025-63917-x","type":"published","date":"2025-10-06T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":92921880,"identity":"68d2eec0-a67d-4bb5-9bc1-57fd0231c286","added_by":"auto","created_at":"2025-10-07 07:10:01","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8373188,"visible":true,"origin":"","legend":"Article File","description":"","filename":"TCVRDDraftFINAL.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6050911/v1_covered_4d6dd82f-3942-49df-8e07-79eaca7868da.pdf"},{"id":78412162,"identity":"b4b6094c-2277-4b69-9fc9-9bd3ae56583a","added_by":"auto","created_at":"2025-03-13 03:38:40","extension":"gz","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":359901058,"visible":true,"origin":"","legend":"Codebase + Dataset","description":"","filename":"codeanddata.tar.gz","url":"https://assets-eu.researchsquare.com/files/rs-6050911/v1/4da100b46880c380683876ae.gz"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nThis work was funded in part by Commonwealth Fusion Systems.","formattedTitle":"Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6050911/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6050911/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The rampdown in tokamak operations is a difficult to simulate phase during which the plasma is often pushed towards multiple instability limits. To address this challenge, and reduce the risk of disrupting operations, we leverage recent advances in Scientific Machine Learning (SciML) to develop a neural state-space model (NSSM) that predicts plasma dynamics during Tokamak à Configuration Variable (TCV) rampdowns. By integrating simple physics structure and data-driven models, the NSSM efficiently learns plasma dynamics during the rampdown from a modest dataset of 311 pulses with only five pulses in the reactor relevant high performance regime. The NSSM is parallelized across uncertainties, and reinforcement learning (RL) is applied to design trajectories that avoid multiple instability limits with high probability. Experiments at TCV ramping down high performance plasmas show statistically significant improvements in current and energy at plasma termination, with improvements in speed through continuous re-training. A predict-first experiment, increasing plasma current by 20\\% from baseline, demonstrates the NSSM's ability to make small extrapolations with sufficient accuracy to design trajectories that successfully terminate the pulse. The developed approach paves the way for designing tokamak controls with robustness to considerable uncertainty, and demonstrates the relevance of the SciML approach to learning plasma dynamics for rapidly developing robust trajectories and controls during the incremental campaigns of upcoming burning plasma tokamaks.","manuscriptTitle":"Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-13 03:38:18","doi":"10.21203/rs.3.rs-6050911/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"2eeaeb46-332e-4a3f-a186-c9cc98f05367","owner":[],"postedDate":"March 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":45519503,"name":"Physical sciences/Energy science and technology/Nuclear energy/Nuclear fusion and fission"},{"id":45519504,"name":"Physical sciences/Physics/Plasma physics/Magnetically confined plasmas"}],"tags":[],"updatedAt":"2025-10-07T07:09:22+00:00","versionOfRecord":{"articleIdentity":"rs-6050911","link":"https://doi.org/10.1038/s41467-025-63917-x","journal":{"identity":"nature-communications","isVorOnly":false,"title":"Nature Communications"},"publishedOn":"2025-10-06 04:00:00","publishedOnDateReadable":"October 6th, 2025"},"versionCreatedAt":"2025-03-13 03:38:18","video":"","vorDoi":"10.1038/s41467-025-63917-x","vorDoiUrl":"https://doi.org/10.1038/s41467-025-63917-x","workflowStages":[]},"version":"v1","identity":"rs-6050911","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6050911","identity":"rs-6050911","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}