Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials

DOI10.1016/J.JCP.2014.12.018zbMATH Open1351.81109arXiv1409.3880OpenAlexW26088913WikidataQ108299833 ScholiaQ108299833MaRDI QIDQ729092FDOQ729092

Garritt J. Tucker, C. R. Trott, S. M. Foiles, Aidan P. Thompson, L. P. Swiler

Publication date: 20 December 2016

Published in: Journal of Computational Physics (Search for Journal in Brave)

Abstract: We present a new interatomic potential for solids and liquids called Spectral Neighbor Analysis Potential (SNAP). The SNAP potential has a very general form and uses machine-learning techniques to reproduce the energies, forces, and stress tensors of a large set of small configurations of atoms, which are obtained using high-accuracy quantum electronic structure (QM) calculations. The local environment of each atom is characterized by a set of bispectrum components of the local neighbor density projected on to a basis of hyperspherical harmonics in four dimensions. The bispectrum components are the same bond-orientational order parameters employed by the GAP potential [arXiv:0910.1019]. The SNAP potential, unlike GAP, assumes a linear relationship between atom energy and bispectrum components. The linear SNAP coefficients are determined using weighted least-squares linear regression against the full QM training set. This allows the SNAP potential to be fit in a robust, automated manner to large QM data sets using many bispectrum coefficients.

Full work available at URL: https://arxiv.org/abs/1409.3880

Recommendations

• A mortar spectral element method for full-potential electronic structure calculations
• Spectral quadrature method for accurate \(\mathcal{O}(N)\) electronic structure calculations of metals and insulators
• Automated parameterization of intermolecular pair potentials using global optimization techniques
• Proper orthogonal descriptors for efficient and accurate interatomic potentials
• Constructing atom-molecule potential surfaces from ab initio data: A method combining quadrature and interpolation
• A spectral scheme for Kohn-Sham density functional theory of clusters
• Self-consistent finite-difference electronic structure calculations
• A Framework for a Generalization Analysis of Machine-Learned Interatomic Potentials

zbMATH Keywords

machine learningsnapmolecular dynamicsinteratomic potentialgaussian approximation potentialsspectral neighbor analysis potential

Mathematics Subject Classification ID

Spectral, collocation and related methods for boundary value problems involving PDEs (65N35) Computational methods for problems pertaining to quantum theory (81-08) Atomic physics (81V45)

Cites Work

• Fast parallel algorithms for short-range molecular dynamics
• Title not available (Why is that?)
• Title not available (Why is that?)
• The Gaussian Approximation Potential
• Multipole expansions in four-dimensional hyperspherical harmonics

Cited In (8)

• Accuracy evaluation of different machine learning force field features
• Proper orthogonal descriptors for multi-element chemical systems
• Atomic cluster expansion: completeness, efficiency and stability
• FLAME: a library of atomistic modeling environments
• DP-GEN: a concurrent learning platform for the generation of reliable deep learning based potential energy models
• Proper orthogonal descriptors for efficient and accurate interatomic potentials
• LAMMPS -- a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales
• Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials

Uses Software

• DAKOTA
• LAMMPS
• SRIM

This page was built for publication: Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q729092)