Skip to main content
OSTI.GOVU.S. Department of Energy Office of Scientific and Technical Information
U.S. Department of Energy
Office of Scientific and Technical Information
 

A Performance and Cost Assessment of Machine Learning Interatomic Potentials

Journal Article · · Journal of Physical Chemistry. A, Molecules, Spectroscopy, Kinetics, Environment, and General Theory
 [1];  [1];  [1];  [1];  [1];  [2];  [3];  [4];  [5];  [5];  [1]
  1. Univ. of California, San Diego, CA (United States). Dept. of NanoEngineering
  2. Univ. of Goettingen (Germany). Inst. of Physical and Theoretical Chemistry
  3. Univ. of Cambridge (United Kingdom). Dept. of Engineering
  4. Skolkovo Institute of Science and Technology, Moscow (Russian Federation)
  5. Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
+ Show Author Affiliations
Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors --- Behler-Parrinello symmetry functions, smooth overlap of atomic positions (SOAP), the Spectral Neighbor Analysis Potential (SNAP) bispectrum components, and moment tensors --- using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model, and consequently computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.
Research Organization:
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States); Sandia National Laboratories (SNL-NM), Albuquerque, NM (United States)
Sponsoring Organization:
USDOE National Nuclear Security Administration (NNSA); USDOE Office of Science (SC)
Grant/Contract Number:
AC02-05CH11231; AC04-94AL85000
OSTI ID:
1559244
Alternate ID(s):
OSTI ID: 1596079
Report Number(s):
SAND2019-7998J; ark:/13030/qt0j64t2hz
Journal Information:
Journal of Physical Chemistry. A, Molecules, Spectroscopy, Kinetics, Environment, and General Theory, Journal Name: Journal of Physical Chemistry. A, Molecules, Spectroscopy, Kinetics, Environment, and General Theory Journal Issue: 4 Vol. 124; ISSN 1089-5639
Publisher:
American Chemical SocietyCopyright Statement
Country of Publication:
United States
Language:
English

References (98)

First Principles Neural Network Potentials for Reactive Simulations of Large Molecular and Condensed Systems journal August 2017
FireWorks: a dynamic workflow system designed for high-throughput applications: FireWorks: A Dynamic Workflow System Designed for High-Throughput Applications journal May 2015
Adaptive machine learning framework to accelerate ab initio molecular dynamics journal December 2014
Constructing high-dimensional neural network potentials: A tutorial review journal March 2015
Machine learning for quantum mechanics in a nutshell journal July 2015
Fast Parallel Algorithms for Short-Range Molecular Dynamics journal March 1995
Comparison of theoretical and empirical interatomic potentials journal April 1986
Two-phase solid–liquid coexistence of Ni, Cu, and Al by molecular dynamics simulations using the modified embedded-atom method journal March 2015
Development of bond-order potentials that can reproduce the elastic constants and melting point of silicon for classical molecular dynamics simulation journal April 2007
Python Materials Genomics (pymatgen): A robust, open-source python library for materials analysis journal February 2013
An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO2 journal March 2016
Active learning of linearly parametrized interatomic potentials journal December 2017
Accelerating high-throughput searches for new alloys with active learning of interatomic potentials journal January 2019
Moment tensor potentials as a promising tool to study diffusion processes journal June 2019
DScribe: Library of descriptors for machine learning in materials science journal February 2020
Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials journal March 2015
Optimized Tersoff empirical potential for germanene journal March 2017
Library-Based LAMMPS Implementation of High-Dimensional Neural Network Potentials journal January 2019
SchNetPack: A Deep Learning Toolbox For Atomistic Systems journal November 2018
From Molecular Fragments to the Bulk: Development of a Neural Network Potential for MOF-5 journal April 2019
Modeling the Phase-Change Memory Material, Ge 2 Sb 2 Te 5 , with a Machine-Learned Interatomic Potential journal September 2018
Machine Learning Force Fields: Construction, Validation, and Outlook journal December 2016
Machine Learning Predictions of Molecular Properties: Accurate Many-Body Potentials and Nonlocality in Chemical Space journal June 2015
Proton-Transfer Mechanisms at the Water–ZnO Interface: The Role of Presolvation journal March 2017
Realistic Atomistic Structure of Amorphous Silicon from Machine-Learning-Driven Molecular Dynamics journal May 2018
Embedded Atom Neural Network Potentials: Efficient and Accurate Machine Learning with a Physically Inspired Representation journal August 2019
Analytic Derivatives of Quartic-Scaling Doubly Hybrid XYGJ-OS Functional: Theory, Implementation, and Benchmark Comparison with M06-2X and MP2 Geometries for Nonbonded Complexes journal March 2013
DREIDING: a generic force field for molecular simulations journal December 1990
UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations journal December 1992
ReaxFF:  A Reactive Force Field for Hydrocarbons journal October 2001
Bypassing the Kohn-Sham equations with machine learning journal October 2017
Machine-learned multi-system surrogate models for materials prediction journal April 2019
An electrostatic spectral neighbor analysis potential for lithium nitride journal July 2019
Charting the complete elastic properties of inorganic crystalline compounds journal March 2015
Surface energies of elemental crystals journal September 2016
Comparing molecules and solids across structural and alchemical space journal January 2016
ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost journal January 2017
The TensorMol-0.1 model chemistry: a neural network augmented with long-range physics journal January 2018
Automated calculation of thermal rate coefficients using ring polymer molecular dynamics and machine-learning interatomic potentials with active learning journal January 2018
Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations journal January 2011
Improved tangent estimate in the nudged elastic band method for finding minimum energy paths and saddle points journal December 2000
Atom-centered symmetry functions for constructing high-dimensional neural network potentials journal February 2011
The core structure of dislocations and their relationship to the material γ-surface journal April 2014
Perspective: Machine learning potentials for atomistic simulations journal November 2016
Study of Li atom diffusion in amorphous Li3PO4 with neural network potential journal December 2017
Non-covalent interactions across organic and biological subsets of chemical space: Physics-based potentials parametrized from machine learning journal June 2018
Extending the accuracy of the SNAP interatomic potential form journal June 2018
Comparison of permutationally invariant polynomials, neural networks, and Gaussian approximation potentials in representing water interactions through many-body expansions journal June 2018
Doubly hybrid density functional for accurate descriptions of nonbond interactions, thermochemistry, and thermochemical kinetics journal March 2009
A fast doubly hybrid density functional method close to chemical accuracy using a local opposite spin ansatz journal November 2011
How van der Waals interactions determine the unique properties of water journal July 2016
Ab initio thermodynamics of liquid and solid water journal January 2019
Density functional theory studies of screw dislocation core structures in bcc metals journal January 2003
Generalized stacking fault energy surfaces and dislocation properties of silicon: A first-principles theoretical study journal December 1996
Error Estimates for Solid-State Density-Functional Theory Predictions: An Overview by Means of the Ground-State Elemental Crystals journal October 2013
The Elastic Behaviour of a Crystalline Aggregate journal May 1952
Comparison between classical potentials and ab initio methods for silicon under large shear journal October 2003
Generalized stacking fault energies for embedded atom FCC metals journal February 2000
Highly optimized empirical potential model of silicon journal October 2000
Self-Consistent Equations Including Exchange and Correlation Effects journal November 1965
Embedded-atom-method functions for the fcc metals Cu, Ag, Au, Ni, Pd, Pt, and their alloys journal June 1986
Semiempirical modified embedded-atom potentials for silicon and germanium journal September 1989
Development of an embedded-atom potential for a bcc metal: Vanadium journal February 1990
Comparative study of silicon empirical interatomic potentials journal July 1992
Projector augmented-wave method journal December 1994
Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set journal October 1996
Calculation of theoretical strengths and bulk moduli of bcc metals journal June 1999
Semiempirical atomic potentials for the fcc metals Cu, Ag, Au, Ni, Pd, Pt, Al, and Pb based on first and second nearest-neighbor modified embedded atom method journal October 2003
Tests of a ladder of density functionals for bulk solids and surfaces journal February 2004
Misfit-energy-increasing dislocations in vapor-deposited CoFe/NiFe multilayers journal April 2004
Assessing the performance of recent density functionals for bulk solids journal April 2009
High-dimensional neural-network potentials for multicomponent systems: Applications to zinc oxide journal April 2011
Ab initio based empirical potential used to study the mechanical properties of molybdenum journal June 2012
On representing chemical environments journal May 2013
Accuracy and transferability of Gaussian approximation potential models for tungsten journal September 2014
Learning scheme to predict atomic forces and accelerate materials simulations journal September 2015
Property trends in simple metals: An empirical potential approach journal May 2016
Machine learning based interatomic potential for amorphous carbon journal March 2017
Development of a machine learning potential for graphene journal February 2018
Quantum-accurate spectral neighbor analysis potential models for Ni-Mo binary alloys and fcc metals journal September 2018
Density functional theory based neural network force fields from energy decompositions journal February 2019
Accelerating crystal structure prediction by machine-learning interatomic potentials with active learning journal February 2019
Data-driven material models for atomistic simulation journal May 2019
Gaussian Approximation Potentials: The Accuracy of Quantum Mechanics, without the Electrons journal April 2010
Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning journal January 2012
Molecular Dynamics with On-the-Fly Machine Learning of Quantum-Mechanical Forces journal March 2015
Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics journal April 2018
Data-Driven Learning of Total and Local Energies in Elemental Boron journal April 2018
Density-Functional Theory of the Energy Gap journal November 1983
Generalized Gradient Approximation Made Simple journal October 1996
Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces journal April 2007
Accurate force field for molybdenum by machine learning large materials data journal September 2017
Achieving DFT accuracy with a machine-learning interatomic potential: Thermomechanics and defects in bcc ferromagnetic iron journal January 2018
Machine Learning a General-Purpose Interatomic Potential for Silicon journal December 2018
Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors conference July 2017
Machine learning of accurate energy-conserving molecular force fields journal May 2017
Reproducibility in density functional theory calculations of solids journal March 2016
Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials journal January 2016

Cited By (1)

MAISE: Construction of neural network interatomic models and evolutionary structure optimization journal February 2021

Similar Records

Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials
Journal Article · Sun Mar 15 00:00:00 EDT 2015 · Journal of Computational Physics · OSTI ID:1426894
Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials
Journal Article · Sun Mar 15 00:00:00 EDT 2015 · Journal of Computational Physics · OSTI ID:22465612
Neural Network Interatomic Potentials
Technical Report · Tue Sep 01 00:00:00 EDT 2020 · OSTI ID:1678825

Related Subjects

74 ATOMIC AND MOLECULAR PHYSICS