Finding density functionals with machine learning

John C. Snyder, Matthias Rupp, Katja Hansen, Klaus Muller, Kieron Burke

Research output: Contribution to journalArticle

182 Citations (Scopus)

Abstract

Machine learning is used to approximate density functionals. For the model problem of the kinetic energy of noninteracting fermions in 1D, mean absolute errors below 1kcal/mol on test densities similar to the training set are reached with fewer than 100 training densities. A predictor identifies if a test density is within the interpolation region. Via principal component analysis, a projected functional derivative finds highly accurate self-consistent densities. The challenges for application of our method to real electronic structure problems are discussed.

Original languageEnglish
Article number253002
JournalPhysical Review Letters
Volume108
Issue number25
DOIs
Publication statusPublished - 2012 Jun 19

Fingerprint

machine learning
functionals
education
principal components analysis
interpolation
fermions
kinetic energy
electronic structure
predictions

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Snyder, J. C., Rupp, M., Hansen, K., Muller, K., & Burke, K. (2012). Finding density functionals with machine learning. Physical Review Letters, 108(25), [253002]. https://doi.org/10.1103/PhysRevLett.108.253002

Finding density functionals with machine learning. / Snyder, John C.; Rupp, Matthias; Hansen, Katja; Muller, Klaus; Burke, Kieron.

In: Physical Review Letters, Vol. 108, No. 25, 253002, 19.06.2012.

Research output: Contribution to journalArticle

Snyder, JC, Rupp, M, Hansen, K, Muller, K & Burke, K 2012, 'Finding density functionals with machine learning', Physical Review Letters, vol. 108, no. 25, 253002. https://doi.org/10.1103/PhysRevLett.108.253002
Snyder JC, Rupp M, Hansen K, Muller K, Burke K. Finding density functionals with machine learning. Physical Review Letters. 2012 Jun 19;108(25). 253002. https://doi.org/10.1103/PhysRevLett.108.253002
Snyder, John C. ; Rupp, Matthias ; Hansen, Katja ; Muller, Klaus ; Burke, Kieron. / Finding density functionals with machine learning. In: Physical Review Letters. 2012 ; Vol. 108, No. 25.
@article{7f167bbdcd7e4f989a3a5094ab024802,
title = "Finding density functionals with machine learning",
abstract = "Machine learning is used to approximate density functionals. For the model problem of the kinetic energy of noninteracting fermions in 1D, mean absolute errors below 1kcal/mol on test densities similar to the training set are reached with fewer than 100 training densities. A predictor identifies if a test density is within the interpolation region. Via principal component analysis, a projected functional derivative finds highly accurate self-consistent densities. The challenges for application of our method to real electronic structure problems are discussed.",
author = "Snyder, {John C.} and Matthias Rupp and Katja Hansen and Klaus Muller and Kieron Burke",
year = "2012",
month = "6",
day = "19",
doi = "10.1103/PhysRevLett.108.253002",
language = "English",
volume = "108",
journal = "Physical Review Letters",
issn = "0031-9007",
publisher = "American Physical Society",
number = "25",

}

TY - JOUR

T1 - Finding density functionals with machine learning

AU - Snyder, John C.

AU - Rupp, Matthias

AU - Hansen, Katja

AU - Muller, Klaus

AU - Burke, Kieron

PY - 2012/6/19

Y1 - 2012/6/19

N2 - Machine learning is used to approximate density functionals. For the model problem of the kinetic energy of noninteracting fermions in 1D, mean absolute errors below 1kcal/mol on test densities similar to the training set are reached with fewer than 100 training densities. A predictor identifies if a test density is within the interpolation region. Via principal component analysis, a projected functional derivative finds highly accurate self-consistent densities. The challenges for application of our method to real electronic structure problems are discussed.

AB - Machine learning is used to approximate density functionals. For the model problem of the kinetic energy of noninteracting fermions in 1D, mean absolute errors below 1kcal/mol on test densities similar to the training set are reached with fewer than 100 training densities. A predictor identifies if a test density is within the interpolation region. Via principal component analysis, a projected functional derivative finds highly accurate self-consistent densities. The challenges for application of our method to real electronic structure problems are discussed.

UR - http://www.scopus.com/inward/record.url?scp=84862560607&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84862560607&partnerID=8YFLogxK

U2 - 10.1103/PhysRevLett.108.253002

DO - 10.1103/PhysRevLett.108.253002

M3 - Article

AN - SCOPUS:84862560607

VL - 108

JO - Physical Review Letters

JF - Physical Review Letters

SN - 0031-9007

IS - 25

M1 - 253002

ER -