Please use this identifier to cite or link to this item:
Main Title: Machine learning of accurate energy-conserving molecular force fields
Author(s): Chmiela, Stefan
Tkatchenko, Alexandre
Sauceda, Huziel E.
Poltavsky, Igor
Schütt, Kristof T.
Müller, Klaus-Robert
Type: Article
Language Code: en
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 Å̊−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.
Issue Date: 2017
Date Available: 19-Apr-2018
DDC Class: 500 Naturwissenschaften und Mathematik
Subject(s): force field
machine learning
gradient field
potential-energy surface
energy conservation
kernel regression
path integrals
molecular dynamics
Sponsor/Funder: BMBF, 01IS14013A, Verbundprojekt: BBDC - Berliner Kompetenzzentrum für Big Data
DFG, MU 987/20-1
Journal Title: Science Advances
Publisher: American Association for the Advancement of Science (AAAS)
Publisher Place: Washington, DC [u.a.]
Volume: 3
Issue: 5
Article Number: e1603015
Publisher DOI: 10.1126/sciadv.1603015
EISSN: 2375-2548
ISSN: 2375-2548
Appears in Collections:FG Maschinelles Lernen » Publications

Files in This Item:
File Description SizeFormat 
e1603015.full.pdf1.18 MBAdobe PDFThumbnail

This item is licensed under a Creative Commons License Creative Commons