Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning

Using supervised and unsupervised machine learning (ML) on features generated from nonadiabatic (NA) molecular dynamics (MD) trajectories under the classical path approximation, we demonstrate that mutual information with the NA Hamiltonian can be used for feature selection and model simplification....

Full description

Saved in:
Bibliographic Details
Main Authors: How, Wei Bin, Wang, Bipeng, Chu, Weibin, Tkatchenko, Alexandre, Prezhdo, Oleg V.
Other Authors: School of Physical and Mathematical Sciences
Format: Article
Language:English
Published: 2022
Subjects:
Online Access:https://hdl.handle.net/10356/163142
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-163142
record_format dspace
spelling sg-ntu-dr.10356-1631422022-11-25T01:33:02Z Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning How, Wei Bin Wang, Bipeng Chu, Weibin Tkatchenko, Alexandre Prezhdo, Oleg V. School of Physical and Mathematical Sciences Science::Chemistry Dimensionality Reduction Machine-Learning Using supervised and unsupervised machine learning (ML) on features generated from nonadiabatic (NA) molecular dynamics (MD) trajectories under the classical path approximation, we demonstrate that mutual information with the NA Hamiltonian can be used for feature selection and model simplification. Focusing on CsPbI3, a popular metal halide perovskite, we observe that the chemical environment of a single element is sufficient for predicting the NA Hamiltonian. The conclusion applies even to Cs, although Cs does not contribute to the relevant wave functions. Interatomic distances between Cs and I or Pb and the octahedral tilt angle are the most important features. We reduce a typical 360-parameter ML force-field model to just a 12-parameter NA Hamiltonian model, while maintaining a high NA-MD simulation quality. Because NA-MD is a valuable tool for studying excited state processes, overcoming its high computational cost through simple ML models will streamline NA-MD simulations and expand the ranges of accessible system size and simulation time. The work was supported by U.S. National Science Foundation Grant CHE-1900510. 2022-11-25T01:33:02Z 2022-11-25T01:33:02Z 2021 Journal Article How, W. B., Wang, B., Chu, W., Tkatchenko, A. & Prezhdo, O. V. (2021). Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning. Journal of Physical Chemistry Letters, 12(50), 12026-12032. https://dx.doi.org/10.1021/acs.jpclett.1c03469 1948-7185 https://hdl.handle.net/10356/163142 10.1021/acs.jpclett.1c03469 34902248 2-s2.0-85121587009 50 12 12026 12032 en Journal of Physical Chemistry Letters © 2021 American Chemical Society. All rights reserved.
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Science::Chemistry
Dimensionality Reduction
Machine-Learning
spellingShingle Science::Chemistry
Dimensionality Reduction
Machine-Learning
How, Wei Bin
Wang, Bipeng
Chu, Weibin
Tkatchenko, Alexandre
Prezhdo, Oleg V.
Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
description Using supervised and unsupervised machine learning (ML) on features generated from nonadiabatic (NA) molecular dynamics (MD) trajectories under the classical path approximation, we demonstrate that mutual information with the NA Hamiltonian can be used for feature selection and model simplification. Focusing on CsPbI3, a popular metal halide perovskite, we observe that the chemical environment of a single element is sufficient for predicting the NA Hamiltonian. The conclusion applies even to Cs, although Cs does not contribute to the relevant wave functions. Interatomic distances between Cs and I or Pb and the octahedral tilt angle are the most important features. We reduce a typical 360-parameter ML force-field model to just a 12-parameter NA Hamiltonian model, while maintaining a high NA-MD simulation quality. Because NA-MD is a valuable tool for studying excited state processes, overcoming its high computational cost through simple ML models will streamline NA-MD simulations and expand the ranges of accessible system size and simulation time.
author2 School of Physical and Mathematical Sciences
author_facet School of Physical and Mathematical Sciences
How, Wei Bin
Wang, Bipeng
Chu, Weibin
Tkatchenko, Alexandre
Prezhdo, Oleg V.
format Article
author How, Wei Bin
Wang, Bipeng
Chu, Weibin
Tkatchenko, Alexandre
Prezhdo, Oleg V.
author_sort How, Wei Bin
title Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
title_short Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
title_full Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
title_fullStr Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
title_full_unstemmed Significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
title_sort significance of the chemical environment of an element in nonadiabatic molecular dynamics: feature selection and dimensionality reduction with machine learning
publishDate 2022
url https://hdl.handle.net/10356/163142
_version_ 1751548584498561024