sGDML

Constructing accurate and data efficient molecular force fields using machine learning

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

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We present an optimized implementation of the recently proposed symmetric gradient domain machine learning (sGDML) model. The sGDML model is able to faithfully reproduce global potential energy surfaces (PES) for molecules with a few dozen atoms from a limited number of user-provided reference molecular conformations and the associated atomic forces. Here, we introduce a Python software package to reconstruct and evaluate custom sGDML force fields (FFs), without requiring in-depth knowledge about the details of the model. A user-friendly command-line interface offers assistance through the complete process of model creation, in an effort to make this novel machine learning approach accessible to broad practitioners. Our paper serves as a documentation, but also includes a practical application example of how to reconstruct and use a PBE0+MBD FF for paracetamol. Finally, we show how to interface sGDML with the FF simulation engines ASE (Larsen et al., 2017) and i-PI (Kapil et al., 2019) to run numerical experiments, including structure optimization, classical and path integral molecular dynamics and nudged elastic band calculations.

Original languageEnglish
JournalComputer Physics Communications
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

machine learning
field theory (physics)
Learning systems
gradients
aeroservoelasticity
Potential energy surfaces
documentation
commands
Software packages
Conformations
engines
Molecular dynamics
potential energy
molecular dynamics
Engines
computer programs
Atoms
Molecules
optimization
atoms

Keywords

  • Ab initio molecular dynamics
  • Coupled cluster calculations
  • Gradient domain machine learning
  • Machine learning force field
  • Machine learning potential
  • Molecular property prediction
  • Path integral molecular dynamics
  • Quantum chemistry

ASJC Scopus subject areas

  • Hardware and Architecture
  • Physics and Astronomy(all)

Cite this

sGDML : Constructing accurate and data efficient molecular force fields using machine learning. / Chmiela, Stefan; Sauceda, Huziel E.; Poltavsky, Igor; Muller, Klaus; Tkatchenko, Alexandre.

In: Computer Physics Communications, 01.01.2019.

Research output: Contribution to journalArticle

Chmiela, Stefan ; Sauceda, Huziel E. ; Poltavsky, Igor ; Muller, Klaus ; Tkatchenko, Alexandre. / sGDML : Constructing accurate and data efficient molecular force fields using machine learning. In: Computer Physics Communications. 2019.
@article{11f1807675074bd1ac2e6d1aa540d103,
title = "sGDML: Constructing accurate and data efficient molecular force fields using machine learning",
abstract = "We present an optimized implementation of the recently proposed symmetric gradient domain machine learning (sGDML) model. The sGDML model is able to faithfully reproduce global potential energy surfaces (PES) for molecules with a few dozen atoms from a limited number of user-provided reference molecular conformations and the associated atomic forces. Here, we introduce a Python software package to reconstruct and evaluate custom sGDML force fields (FFs), without requiring in-depth knowledge about the details of the model. A user-friendly command-line interface offers assistance through the complete process of model creation, in an effort to make this novel machine learning approach accessible to broad practitioners. Our paper serves as a documentation, but also includes a practical application example of how to reconstruct and use a PBE0+MBD FF for paracetamol. Finally, we show how to interface sGDML with the FF simulation engines ASE (Larsen et al., 2017) and i-PI (Kapil et al., 2019) to run numerical experiments, including structure optimization, classical and path integral molecular dynamics and nudged elastic band calculations.",
keywords = "Ab initio molecular dynamics, Coupled cluster calculations, Gradient domain machine learning, Machine learning force field, Machine learning potential, Molecular property prediction, Path integral molecular dynamics, Quantum chemistry",
author = "Stefan Chmiela and Sauceda, {Huziel E.} and Igor Poltavsky and Klaus Muller and Alexandre Tkatchenko",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.cpc.2019.02.007",
language = "English",
journal = "Computer Physics Communications",
issn = "0010-4655",
publisher = "Elsevier",

}

TY - JOUR

T1 - sGDML

T2 - Constructing accurate and data efficient molecular force fields using machine learning

AU - Chmiela, Stefan

AU - Sauceda, Huziel E.

AU - Poltavsky, Igor

AU - Muller, Klaus

AU - Tkatchenko, Alexandre

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We present an optimized implementation of the recently proposed symmetric gradient domain machine learning (sGDML) model. The sGDML model is able to faithfully reproduce global potential energy surfaces (PES) for molecules with a few dozen atoms from a limited number of user-provided reference molecular conformations and the associated atomic forces. Here, we introduce a Python software package to reconstruct and evaluate custom sGDML force fields (FFs), without requiring in-depth knowledge about the details of the model. A user-friendly command-line interface offers assistance through the complete process of model creation, in an effort to make this novel machine learning approach accessible to broad practitioners. Our paper serves as a documentation, but also includes a practical application example of how to reconstruct and use a PBE0+MBD FF for paracetamol. Finally, we show how to interface sGDML with the FF simulation engines ASE (Larsen et al., 2017) and i-PI (Kapil et al., 2019) to run numerical experiments, including structure optimization, classical and path integral molecular dynamics and nudged elastic band calculations.

AB - We present an optimized implementation of the recently proposed symmetric gradient domain machine learning (sGDML) model. The sGDML model is able to faithfully reproduce global potential energy surfaces (PES) for molecules with a few dozen atoms from a limited number of user-provided reference molecular conformations and the associated atomic forces. Here, we introduce a Python software package to reconstruct and evaluate custom sGDML force fields (FFs), without requiring in-depth knowledge about the details of the model. A user-friendly command-line interface offers assistance through the complete process of model creation, in an effort to make this novel machine learning approach accessible to broad practitioners. Our paper serves as a documentation, but also includes a practical application example of how to reconstruct and use a PBE0+MBD FF for paracetamol. Finally, we show how to interface sGDML with the FF simulation engines ASE (Larsen et al., 2017) and i-PI (Kapil et al., 2019) to run numerical experiments, including structure optimization, classical and path integral molecular dynamics and nudged elastic band calculations.

KW - Ab initio molecular dynamics

KW - Coupled cluster calculations

KW - Gradient domain machine learning

KW - Machine learning force field

KW - Machine learning potential

KW - Molecular property prediction

KW - Path integral molecular dynamics

KW - Quantum chemistry

UR - http://www.scopus.com/inward/record.url?scp=85062624346&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062624346&partnerID=8YFLogxK

U2 - 10.1016/j.cpc.2019.02.007

DO - 10.1016/j.cpc.2019.02.007

M3 - Article

JO - Computer Physics Communications

JF - Computer Physics Communications

SN - 0010-4655

ER -