Neural Network Potential Energy Surfaces for Small Molecules and Reactions

被引:233
作者
Manzhos, Sergei [1 ]
Carrington, Tucker, Jr. [2 ]
机构
[1] Inst Natl Rech Sci, Ctr Energie Mat Telecommun, Quebec City, PQ J3X 1S2, Canada
[2] Queens Univ, Chem Dept, Kingston, ON K7L 3N6, Canada
基金
加拿大自然科学与工程研究理事会;
关键词
LEAST-SQUARES METHODS; COMPUTING VIBRATIONAL-SPECTRA; ELECTRONIC GROUND-STATE; SELF-CONSISTENT-FIELD; EMBEDDED-ATOM METHOD; QUANTUM DYNAMICS; UNIVERSAL APPROXIMATION; PERTURBATION-THEORY; FORCE-FIELD; NUMERICAL IMPLEMENTATION;
D O I
10.1021/acs.chemrev.0c00665
中图分类号
O6 [化学];
学科分类号
0703 ;
摘要
We review progress in neural network (NN)-based methods for the construction of interatomic potentials from discrete samples (such as ab initio energies) for applications in classical and quantum dynamics including reaction dynamics and computational spectroscopy. The main focus is on methods for building molecular potential energy surfaces (PES) in internal coordinates that explicitly include all many-body contributions, even though some of the methods we review limit the degree of coupling, due either to a desire to limit computational cost or to limited data. Explicit and direct treatment of all many-body contributions is only practical for sufficiently small molecules, which are therefore our primary focus. This includes small molecules on surfaces. We consider direct, single NN PES fitting as well as more complex methods that impose structure (such as a multibody representation) on the PES function, either through the architecture of one NN or by using multiple NNs. We show how NNs are effective in building representations with low-dimensional functions including dimensionality reduction. We consider NNbased approaches to build PESs in the sums-of-product form important for quantum dynamics, ways to treat symmetry, and issues related to sampling data distributions and the relation between PES errors and errors in observables. We highlight combinations of NNs with other ideas such as permutationally invariant polynomials or sums of environment-dependent atomic contributions, which have recently emerged as powerful tools for building highly accurate PESs for relatively large molecular and reactive systems.
引用
收藏
页码:10187 / 10217
页数:31
相关论文
共 354 条
[1]   PES-Learn: An Open-Source Software Package for the Automated Generation of Machine Learning Models of Molecular Potential Energy Surfaces [J].
Abbott, Adam S. ;
Turney, Justin M. ;
Zhang, Boyi ;
Smith, Daniel G. A. ;
Altarawy, Doaa ;
Schaefer, Henry F., III .
JOURNAL OF CHEMICAL THEORY AND COMPUTATION, 2019, 15 (08) :4386-4398
[2]   Molecular dynamics investigations of the dissociation of SiO2 on an ab initio potential energy surface obtained using neural network methods [J].
Agrawal, PM ;
Raff, LM ;
Hagan, MT ;
Komanduri, R .
JOURNAL OF CHEMICAL PHYSICS, 2006, 124 (13)
[3]   MOLECULAR MECHANICS - THE MM3 FORCE-FIELD FOR HYDROCARBONS .1. [J].
ALLINGER, NL ;
YUH, YH ;
LII, JH .
JOURNAL OF THE AMERICAN CHEMICAL SOCIETY, 1989, 111 (23) :8551-8566
[4]   2ND-ORDER PERTURBATION-THEORY WITH A CASSCF REFERENCE FUNCTION [J].
ANDERSSON, K ;
MALMQVIST, PA ;
ROOS, BO ;
SADLEJ, AJ ;
WOLINSKI, K .
JOURNAL OF PHYSICAL CHEMISTRY, 1990, 94 (14) :5483-5488
[5]  
[Anonymous], 2020, Python Language Reference
[6]  
[Anonymous], 2017, R LANG ENV STAT COMP
[7]  
[Anonymous], 1994, Multidimensional Scaling
[8]  
[Anonymous], 2019, MATLAB 2019a
[9]  
[Anonymous], 2015, ACS SYM SER
[10]  
[Anonymous], 2011, CONICAL INTERSECTION, P768