Machine learning of accurate energy-conserving molecular force fields

Stefan Chmiela, Alexandre Tkatchenko, Huziel E. Sauceda, Igor Poltavsky, Kristof T. Schütt, Klaus Muller

Research output: Contribution to journalArticle

125 Citations (Scopus)

Abstract

Using conservation of energy-a fundamental property of closed classical and quantum mechanical systems-we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of intermediate-sized molecules with an accuracy of 0.3 kcal mol-1 for energies and 1 kcal mol-1 Å-1 for atomic forces using only 1000 conformational geometries for training. We demonstrate this accuracy for AIMD trajectories of molecules, including benzene, toluene, naphthalene, ethanol, uracil, and aspirin. The challenge of constructing conservative force fields is accomplished in our work by learning in a Hilbert space of vector-valued functions that obey the law of energy conservation. The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods. 2017

Original languageEnglish
Article numbere1603015
JournalScience advances
Volume3
Issue number5
DOIs
Publication statusPublished - 2017 May 1

Fingerprint

Molecular Dynamics Simulation
Uracil
Toluene
Benzene
Aspirin
Ethanol
Learning
Costs and Cost Analysis
Machine Learning

ASJC Scopus subject areas

  • Medicine(all)

Cite this

Chmiela, S., Tkatchenko, A., Sauceda, H. E., Poltavsky, I., Schütt, K. T., & Muller, K. (2017). Machine learning of accurate energy-conserving molecular force fields. Science advances, 3(5), [e1603015]. https://doi.org/10.1126/sciadv.1603015

Machine learning of accurate energy-conserving molecular force fields. / Chmiela, Stefan; Tkatchenko, Alexandre; Sauceda, Huziel E.; Poltavsky, Igor; Schütt, Kristof T.; Muller, Klaus.

In: Science advances, Vol. 3, No. 5, e1603015, 01.05.2017.

Research output: Contribution to journalArticle

Chmiela, S, Tkatchenko, A, Sauceda, HE, Poltavsky, I, Schütt, KT & Muller, K 2017, 'Machine learning of accurate energy-conserving molecular force fields', Science advances, vol. 3, no. 5, e1603015. https://doi.org/10.1126/sciadv.1603015
Chmiela S, Tkatchenko A, Sauceda HE, Poltavsky I, Schütt KT, Muller K. Machine learning of accurate energy-conserving molecular force fields. Science advances. 2017 May 1;3(5). e1603015. https://doi.org/10.1126/sciadv.1603015
Chmiela, Stefan ; Tkatchenko, Alexandre ; Sauceda, Huziel E. ; Poltavsky, Igor ; Schütt, Kristof T. ; Muller, Klaus. / Machine learning of accurate energy-conserving molecular force fields. In: Science advances. 2017 ; Vol. 3, No. 5.
@article{6d027d892fe042c7a96d95baa4ce100c,
title = "Machine learning of accurate energy-conserving molecular force fields",
abstract = "Using conservation of energy-a fundamental property of closed classical and quantum mechanical systems-we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of intermediate-sized molecules with an accuracy of 0.3 kcal mol-1 for energies and 1 kcal mol-1 {\AA}-1 for atomic forces using only 1000 conformational geometries for training. We demonstrate this accuracy for AIMD trajectories of molecules, including benzene, toluene, naphthalene, ethanol, uracil, and aspirin. The challenge of constructing conservative force fields is accomplished in our work by learning in a Hilbert space of vector-valued functions that obey the law of energy conservation. The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods. 2017",
author = "Stefan Chmiela and Alexandre Tkatchenko and Sauceda, {Huziel E.} and Igor Poltavsky and Sch{\"u}tt, {Kristof T.} and Klaus Muller",
year = "2017",
month = "5",
day = "1",
doi = "10.1126/sciadv.1603015",
language = "English",
volume = "3",
journal = "Science advances",
issn = "2375-2548",
publisher = "American Association for the Advancement of Science",
number = "5",

}

TY - JOUR

T1 - Machine learning of accurate energy-conserving molecular force fields

AU - Chmiela, Stefan

AU - Tkatchenko, Alexandre

AU - Sauceda, Huziel E.

AU - Poltavsky, Igor

AU - Schütt, Kristof T.

AU - Muller, Klaus

PY - 2017/5/1

Y1 - 2017/5/1

N2 - Using conservation of energy-a fundamental property of closed classical and quantum mechanical systems-we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of intermediate-sized molecules with an accuracy of 0.3 kcal mol-1 for energies and 1 kcal mol-1 Å-1 for atomic forces using only 1000 conformational geometries for training. We demonstrate this accuracy for AIMD trajectories of molecules, including benzene, toluene, naphthalene, ethanol, uracil, and aspirin. The challenge of constructing conservative force fields is accomplished in our work by learning in a Hilbert space of vector-valued functions that obey the law of energy conservation. The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods. 2017

AB - Using conservation of energy-a fundamental property of closed classical and quantum mechanical systems-we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of intermediate-sized molecules with an accuracy of 0.3 kcal mol-1 for energies and 1 kcal mol-1 Å-1 for atomic forces using only 1000 conformational geometries for training. We demonstrate this accuracy for AIMD trajectories of molecules, including benzene, toluene, naphthalene, ethanol, uracil, and aspirin. The challenge of constructing conservative force fields is accomplished in our work by learning in a Hilbert space of vector-valued functions that obey the law of energy conservation. The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods. 2017

UR - http://www.scopus.com/inward/record.url?scp=85041381183&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041381183&partnerID=8YFLogxK

U2 - 10.1126/sciadv.1603015

DO - 10.1126/sciadv.1603015

M3 - Article

AN - SCOPUS:85041381183

VL - 3

JO - Science advances

JF - Science advances

SN - 2375-2548

IS - 5

M1 - e1603015

ER -