Bypassing the Kohn-Sham equations with machine learning
Max Planck Institute of Microstructure Physics · Technische Universität Berlin · +6 more institutions
Abstract
Last year, at least 30,000 scientific papers used the Kohn-Sham scheme of density functional theory to solve electronic structure problems in a wide variety of scientific fields. Machine learning holds the promise of learning the energy functional via examples, bypassing the need to solve the Kohn-Sham equations. This should yield substantial savings in computer time, allowing larger systems and/or longer time-scales to be tackled, but attempts to machine-learn this functional have been limited by the need to find its derivative. The present work overcomes this difficulty by directly learning the density-potential and energy-density maps for test systems and various molecules. We perform the first molecular…
Citation impact
- FWCI
- 28.65
- Percentile
- 100%
- References
- 61
Authors
6- FBFelix BrockherdeCorresponding
Max Planck Institute of Microstructure Physics, Technische Universität Berlin
- LVLeslie Vogt
New York University
- LLLi Li
University of California, Irvine
- MEMark E. Tuckerman
New York University Shanghai, Courant Institute of Mathematical Sciences, New York University
- KBKieron Burke
University of California, Irvine
Topics & keywords
- Density functional theory
- Variety (cybernetics)
- Scheme (mathematics)
- Work (physics)
- Computational learning theory
- Dynamical systems theory
- Transfer of learning
- Time-dependent density functional theory
Funding
- NSNational Science FoundationAwards: 1464795, CHE-1464795, W911NF-13
- UOUniversity of California, Irvine
- IFInstitute for Information and Communications Technology PromotionAward: 2017-0-00451
- DODivision of ChemistryAward: CHE-1464795
- ARArmy Research OfficeAwards: W911NF-13-1-0387, W911NF-13-1-, W911NF-13, W911NF