{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,26]],"date-time":"2026-02-26T20:31:58Z","timestamp":1772137918675,"version":"3.50.1"},"reference-count":50,"publisher":"IOP Publishing","issue":"3","license":[{"start":{"date-parts":[[2024,7,18]],"date-time":"2024-07-18T00:00:00Z","timestamp":1721260800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2024,7,18]],"date-time":"2024-07-18T00:00:00Z","timestamp":1721260800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/iopscience.iop.org\/info\/page\/text-and-data-mining"}],"funder":[{"DOI":"10.13039\/501100021856","name":"Ministero dell\u2019Universit\u00e0 e della Ricerca","doi-asserted-by":"crossref","award":["2022H77XB7"],"award-info":[{"award-number":["2022H77XB7"]}],"id":[{"id":"10.13039\/501100021856","id-type":"DOI","asserted-by":"crossref"}]},{"DOI":"10.13039\/501100010198","name":"Ministerio de Asuntos Econ\u00f3micos y Transformaci\u00f3n Digital, Gobierno de Espa\u00f1a","doi-asserted-by":"crossref","award":["QUANTUM ENIA"],"award-info":[{"award-number":["QUANTUM ENIA"]}],"id":[{"id":"10.13039\/501100010198","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["iopscience.iop.org"],"crossmark-restriction":false},"short-container-title":["Mach. Learn.: Sci. Technol."],"published-print":{"date-parts":[[2024,9,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>In recent years, machine learning models, chiefly deep neural networks, have revealed suited to learn accurate energy-density functionals from data. However, problematic instabilities have been shown to occur in the search of ground-state density profiles via energy minimization. Indeed, any small noise can lead astray from realistic profiles, causing the failure of the learned functional and, hence, strong violations of the variational property. In this article, we employ variational autoencoders (VAEs) to build a compressed, flexible, and regular representation of the ground-state density profiles of various quantum models. Performing energy minimization in this compressed space allows us to avoid both numerical instabilities and variational biases due to excessive constraints. Our tests are performed on one-dimensional single-particle models from the literature in the field and, notably, on a three-dimensional disordered potential. In all cases, the ground-state energies are estimated with errors below the chemical accuracy and the density profiles are accurately reproduced without numerical artifacts. Furthermore, we show that it is possible to perform transfer learning, applying pre-trained VAEs to different potentials.<\/jats:p>","DOI":"10.1088\/2632-2153\/ad611f","type":"journal-article","created":{"date-parts":[[2024,7,9]],"date-time":"2024-07-09T19:23:58Z","timestamp":1720553038000},"page":"035015","update-policy":"https:\/\/doi.org\/10.1088\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Solving deep-learning density functional theory via variational autoencoders"],"prefix":"10.1088","volume":"5","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9830-3119","authenticated-orcid":true,"given":"Emanuele","family":"Costa","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5617-7436","authenticated-orcid":true,"given":"Giuseppe","family":"Scriva","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4845-6299","authenticated-orcid":false,"given":"Sebastiano","family":"Pilati","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"266","published-online":{"date-parts":[[2024,7,18]]},"reference":[{"key":"mlstad611fbib1","doi-asserted-by":"publisher","first-page":"B864","DOI":"10.1103\/PhysRev.136.B864","article-title":"Inhomogeneous electron gas","volume":"136","author":"Hohenberg","year":"1964","journal-title":"Phys. Rev."},{"key":"mlstad611fbib2","doi-asserted-by":"publisher","DOI":"10.1063\/1.4704546","article-title":"Perspective on density functional theory","volume":"136","author":"Burke","year":"2012","journal-title":"J. Chem. Phys."},{"key":"mlstad611fbib3","first-page":"pp 117","author":"Alexander Wang","year":"2002"},{"key":"mlstad611fbib4","author":"Wesolowski","year":"2013"},{"key":"mlstad611fbib5","doi-asserted-by":"publisher","first-page":"289","DOI":"10.1021\/cr200107z","article-title":"Challenges for density functional theory","volume":"112","author":"Cohen","year":"2012","journal-title":"Chem. Rev."},{"key":"mlstad611fbib6","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.108.253002","article-title":"Finding density functionals with machine learning","volume":"108","author":"Snyder","year":"2012","journal-title":"Phys. Rev. Lett."},{"key":"mlstad611fbib7","doi-asserted-by":"publisher","first-page":"872","DOI":"10.1038\/s41467-017-00839-3","article-title":"Bypassing the Kohn-Sham equations with machine learning","volume":"8","author":"Brockherde","year":"2017","journal-title":"Nat. Commun."},{"key":"mlstad611fbib8","doi-asserted-by":"publisher","first-page":"819","DOI":"10.1002\/qua.25040","article-title":"Understanding machine-learned density functionals","volume":"116","author":"Li","year":"2016","journal-title":"Int. J. Quantum Chem."},{"key":"mlstad611fbib9","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevA.100.022512","article-title":"Deep learning and density-functional theory","volume":"100","author":"Ryczko","year":"2019","journal-title":"Phys. Rev. A"},{"key":"mlstad611fbib10","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.99.075132","article-title":"Machine learning density functional theory for the Hubbard model","volume":"99","author":"Nelson","year":"2019","journal-title":"Phys. Rev. B"},{"key":"mlstad611fbib11","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.125.076402","article-title":"Deep learning the Hohenberg-Kohn maps of density functional theory","volume":"125","author":"Moreno","year":"2020","journal-title":"Phys. Rev. Lett."},{"key":"mlstad611fbib12","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.108.125113","article-title":"Deep learning nonlocal and scalable energy functionals for quantum Ising models","volume":"108","author":"Costa","year":"2023","journal-title":"Phys. Rev. B"},{"key":"mlstad611fbib13","doi-asserted-by":"publisher","first-page":"1102","DOI":"10.1002\/qua.24937","article-title":"Nonlinear gradient denoising: finding accurate extrema from inaccurate functional derivatives","volume":"115","author":"Snyder","year":"2015","journal-title":"Int. J. Quantum Chem."},{"key":"mlstad611fbib14","doi-asserted-by":"publisher","first-page":"5685","DOI":"10.1021\/acs.jctc.0c00580","article-title":"Machine learning approaches toward orbital-free density functional theory: simultaneous training on the kinetic energy density functional and its functional derivative","volume":"16","author":"Meyer","year":"2020","journal-title":"J. Chem. Theory Comput."},{"key":"mlstad611fbib15","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevE.106.045309","article-title":"Deep-learning density functionals for gradient descent optimization","volume":"106","author":"Costa","year":"2022","journal-title":"Phys. Rev. E"},{"key":"mlstad611fbib16","doi-asserted-by":"publisher","DOI":"10.1063\/1.4834075","article-title":"Orbital-free bond breaking via machine learning","volume":"139","author":"Snyder","year":"2013","journal-title":"J. Chem. Phys."},{"key":"mlstad611fbib17","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevB.94.245129","article-title":"Pure density functional for strong correlation and the thermodynamic limit from machine learning","volume":"94","author":"Li","year":"2016","journal-title":"Phys. Rev. B"},{"key":"mlstad611fbib18","doi-asserted-by":"publisher","first-page":"1385","DOI":"10.1126\/science.abj6511","article-title":"Pushing the frontiers of density functionals by solving the fractional electron problem","volume":"374","author":"Kirkpatrick","year":"2021","journal-title":"Science"},{"key":"mlstad611fbib19","doi-asserted-by":"publisher","first-page":"1122","DOI":"10.1021\/acs.jctc.1c00812","article-title":"Toward orbital-free density functional theory with small data sets and deep learning","volume":"18","author":"Ryczko","year":"2022","journal-title":"J. Chem. Theory Comput."},{"key":"mlstad611fbib20","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.10814855)","article-title":"Dataset for solving deep learning density functional theory via variational autoencoder","author":"Costa","year":"2024"},{"key":"mlstad611fbib21","doi-asserted-by":"publisher","DOI":"10.1088\/2516-1075\/ac572f","article-title":"Roadmap on machine learning in electronic structure","volume":"4","author":"Kulik","year":"2022","journal-title":"Electron. Struct."},{"key":"mlstad611fbib22","doi-asserted-by":"publisher","first-page":"A1133","DOI":"10.1103\/PhysRev.140.A1133","article-title":"Self-consistent equations including exchange and correlation effects","volume":"140","author":"Kohn","year":"1965","journal-title":"Phys. Rev."},{"key":"mlstad611fbib23","article-title":"Auto-encoding variational bayes","author":"Kingma","year":"2014"},{"key":"mlstad611fbib24","article-title":"beta-VAE: learning basic visual concepts with a constrained variational framework","author":"Higgins","year":"2017"},{"key":"mlstad611fbib25","first-page":"pp 4047","article-title":"Enhancing variational autoencoders with mutual information neural estimation for text generation","author":"Qian","year":"2019"},{"key":"mlstad611fbib26","doi-asserted-by":"publisher","first-page":"199440","DOI":"10.1109\/ACCESS.2020.3034828","article-title":"Balancing reconstruction error and Kullback-Leibler divergence in variational autoencoders","volume":"8","author":"Asperti","year":"2020","journal-title":"IEEE Access"},{"key":"mlstad611fbib27","article-title":"Understanding disentangling in \u03b2-vae","author":"Burgess","year":"2018"},{"key":"mlstad611fbib28","first-page":"pp 4402","article-title":"Disentangling disentanglement in variational autoencoders","author":"Mathieu","year":"2019"},{"key":"mlstad611fbib29","first-page":"pp 1","article-title":"Improving deep neural networks using softplus units","author":"Zheng","year":"2015"},{"key":"mlstad611fbib30","first-page":"pp 448","article-title":"Batch normalization: accelerating deep network training by reducing internal covariate shift","author":"Ioffe","year":"2015"},{"key":"mlstad611fbib31","first-page":"pp 195","article-title":"The influence of the sigmoid function parameters on the speed of backpropagation learning","author":"Han","year":"1995"},{"key":"mlstad611fbib32","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014"},{"key":"mlstad611fbib33","first-page":"pp 8024","article-title":"Pytorch: an imperative style, high-performance deep learning library","author":"Paszke","year":"2019"},{"key":"mlstad611fbib34","author":"Goodman.","year":"2007"},{"key":"mlstad611fbib35","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevA.73.013606","article-title":"Collective dynamics and expansion of a bose-einstein condensate in a random potential","volume":"73","author":"Modugno","year":"2006","journal-title":"Phys. Rev. A"},{"key":"mlstad611fbib36","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevA.95.013613","article-title":"Localization of interacting Fermi gases in quasiperiodic potentials","volume":"95","author":"Pilati","year":"2017","journal-title":"Phys. Rev. A"},{"key":"mlstad611fbib37","article-title":"Data for deep learning density functionals for gradient descent optimization","author":"Costa","year":"2022"},{"key":"mlstad611fbib38","doi-asserted-by":"publisher","DOI":"10.1063\/1.5025668","article-title":"Can exact conditions improve machine-learned density functionals?","volume":"148","author":"Hollingsworth","year":"2018","journal-title":"J. Chem. Phys."},{"key":"mlstad611fbib39","doi-asserted-by":"publisher","DOI":"10.1016\/j.physleta.2021.127621","article-title":"Highly accurate machine learning model for kinetic energy density functional","volume":"414","author":"Alghadeer","year":"2021","journal-title":"Phys. Lett. A"},{"key":"mlstad611fbib40","doi-asserted-by":"publisher","DOI":"10.1063\/5.0166432","article-title":"Variational principle to regularize machine-learned density functionals: the non-interacting kinetic-energy functional","volume":"159","author":"del Mazo-Sevillano","year":"2023","journal-title":"J. Chem. Phys."},{"key":"mlstad611fbib41","doi-asserted-by":"publisher","DOI":"10.1016\/j.cplett.2020.137358","article-title":"Orbital-free density functional theory calculation applying semi-local machine-learned kinetic energy density functional and kinetic potential","volume":"748","author":"Fujinami","year":"2020","journal-title":"Chem. Phys. Lett."},{"key":"mlstad611fbib42","doi-asserted-by":"publisher","first-page":"3509","DOI":"10.1038\/s41467-020-17265-7","article-title":"Machine learning accurate exchange and correlation functionals of the electronic density","volume":"11","author":"Dick","year":"2020","journal-title":"Nat. Commun."},{"key":"mlstad611fbib43","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevLett.126.036401","article-title":"Kohn-sham equations as regularizer: building prior knowledge into machine-learned physics","volume":"126","author":"Li","year":"2021","journal-title":"Phys. Rev. Lett."},{"key":"mlstad611fbib44","doi-asserted-by":"publisher","first-page":"367","DOI":"10.1038\/s43588-022-00265-6","article-title":"Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation","volume":"2","author":"Li","year":"2022","journal-title":"Nat. Comput. Sci."},{"key":"mlstad611fbib45","doi-asserted-by":"publisher","DOI":"10.1088\/2632-2153\/ace8f0","article-title":"Machine-learning Kohn-Sham potential from dynamics in time-dependent Kohn-Sham systems","volume":"4","author":"Yang","year":"2023","journal-title":"Mach. Learn.: Sci. Technol."},{"key":"mlstad611fbib46","doi-asserted-by":"publisher","first-page":"210","DOI":"10.1038\/s43588-024-00605-8","article-title":"Overcoming the barrier of orbital-free density functional theory for molecular systems using deep learning","volume":"4","author":"Zhang","year":"2024","journal-title":"Nat. Comput. Sci."},{"key":"mlstad611fbib47","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevA.101.050501","article-title":"Machine learning exchange-correlation potential in time-dependent density-functional theory","volume":"101","author":"Suzuki","year":"2020","journal-title":"Phys. Rev. A"},{"key":"mlstad611fbib48","doi-asserted-by":"publisher","first-page":"8487","DOI":"10.1021\/acs.jpca.2c05922","article-title":"Automatic differentiation for the direct minimization approach to the Hartree-Fock method","volume":"126","author":"Yoshikawa","year":"2022","journal-title":"J. Phys. Chem. A"},{"key":"mlstad611fbib49","doi-asserted-by":"publisher","DOI":"10.1063\/5.0138429","article-title":"Automatic differentiation for orbital-free density functional theory","volume":"158","author":"Tan","year":"2023","journal-title":"J. Chem. Phys."},{"key":"mlstad611fbib50","article-title":"Orbital-free density functional theory with continuous normalizing flows","author":"de Camargo","year":"2023"}],"container-title":["Machine Learning: Science and Technology"],"original-title":[],"link":[{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f","content-type":"text\/html","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"am","intended-application":"similarity-checking"},{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f\/pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,9,12]],"date-time":"2024-09-12T05:38:05Z","timestamp":1726119485000},"score":1,"resource":{"primary":{"URL":"https:\/\/iopscience.iop.org\/article\/10.1088\/2632-2153\/ad611f"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,7,18]]},"references-count":50,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2024,7,18]]},"published-print":{"date-parts":[[2024,9,1]]}},"URL":"https:\/\/doi.org\/10.1088\/2632-2153\/ad611f","relation":{"has-review":[{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v1\/review2","asserted-by":"object"},{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v1\/review1","asserted-by":"object"},{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v2\/review1","asserted-by":"object"},{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v2\/response1","asserted-by":"object"},{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v1\/decision1","asserted-by":"object"},{"id-type":"doi","id":"10.1088\/2632-2153\/AD611F\/v2\/decision1","asserted-by":"object"}]},"ISSN":["2632-2153"],"issn-type":[{"value":"2632-2153","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,7,18]]},"assertion":[{"value":"Solving deep-learning density functional theory via variational autoencoders","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 2024 The Author(s). Published by IOP Publishing Ltd","name":"copyright_information","label":"Copyright Information"},{"value":"2024-03-29","name":"date_received","label":"Date Received","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-07-09","name":"date_accepted","label":"Date Accepted","group":{"name":"publication_dates","label":"Publication dates"}},{"value":"2024-07-18","name":"date_epub","label":"Online publication date","group":{"name":"publication_dates","label":"Publication dates"}}]}}