Machine learning based energy-free structure predictions of molecules, transition states, and solids

The computational prediction of atomistic structure is a long-standing problem in physics, chemistry, materials, and biology. Conventionally, force-fields or ab initio methods determine structure through energy minimization, which is either approximate or computationally demanding. This accuracy/cos...

Full description

Saved in:
Bibliographic Details
Published in:Nature communications Vol. 12; no. 1; pp. 4468 - 10
Main Authors: Lemm, Dominik, von Rudorff, Guido Falk, von Lilienfeld, O. Anatole
Format: Journal Article
Language:English
Published: London Nature Publishing Group UK 22.07.2021
Nature Publishing Group
Nature Portfolio
Subjects:
119/118
639/638/563/606
639/638/563/758
639/638/563/983
639/638/630
Accounting
Computer applications
Datasets
Humanities and Social Sciences
Interatomic distance
Learning algorithms
Machine learning
Molecular structure
multidisciplinary
Optimization
Organic chemistry
Predictions
Quantum computing
Science
Science (multidisciplinary)
Training
ISSN:2041-1723, 2041-1723
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The computational prediction of atomistic structure is a long-standing problem in physics, chemistry, materials, and biology. Conventionally, force-fields or ab initio methods determine structure through energy minimization, which is either approximate or computationally demanding. This accuracy/cost trade-off prohibits the generation of synthetic big data sets accounting for chemical space with atomistic detail. Exploiting implicit correlations among relaxed structures in training data sets, our machine learning model Graph-To-Structure (G2S) generalizes across compound space in order to infer interatomic distances for out-of-sample compounds, effectively enabling the direct reconstruction of coordinates, and thereby bypassing the conventional energy optimization task. The numerical evidence collected includes 3D coordinate predictions for organic molecules, transition states, and crystalline solids. G2S improves systematically with training set size, reaching mean absolute interatomic distance prediction errors of less than 0.2 Å for less than eight thousand training structures — on par or better than conventional structure generators. Applicability tests of G2S include successful predictions for systems which typically require manual intervention, improved initial guesses for subsequent conventional ab initio based relaxation, and input generation for subsequent use of structure based quantum machine learning models. Accurate computational prediction of atomistic structure with traditional methods is challenging. The authors report a kernel-based machine learning model capable of reconstructing 3D atomic coordinates from predicted interatomic distances across a variety of system classes.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:2041-1723
2041-1723
DOI:10.1038/s41467-021-24525-7