Dynamic Training Enhances Machine Learning Potentials for Long-Lasting Molecular Dynamics

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Žugec, Ivan, Veljković, Tin Hadži, Alducin, Maite, Juaristi, J. Iñaki
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909692298526720
author Žugec, Ivan
Veljković, Tin Hadži
Alducin, Maite
Juaristi, J. Iñaki
author_facet Žugec, Ivan
Veljković, Tin Hadži
Alducin, Maite
Juaristi, J. Iñaki
contents Molecular Dynamics (MD) simulations are vital for exploring complex systems in computational physics and chemistry. While machine learning methods dramatically reduce computational costs relative to ab initio methods, their accuracy in long-lasting simulations remains limited. Here we propose dynamic training (DT), a method designed to enhance accuracy of a model over extended MD simulations. Applying DT to an equivariant graph neural network (EGNN) on the challenging system of a hydrogen molecule interacting with a palladium cluster anchored to a graphene vacancy demonstrates a superior prediction accuracy compared to conventional approaches. Crucially, the DT architecture-independent design ensures its applicability across diverse machine learning potentials, making it a practical tool for advancing MD simulations.
format Preprint
id arxiv_https___arxiv_org_abs_2504_03521
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Dynamic Training Enhances Machine Learning Potentials for Long-Lasting Molecular Dynamics
Žugec, Ivan
Veljković, Tin Hadži
Alducin, Maite
Juaristi, J. Iñaki
Materials Science
Chemical Physics
Molecular Dynamics (MD) simulations are vital for exploring complex systems in computational physics and chemistry. While machine learning methods dramatically reduce computational costs relative to ab initio methods, their accuracy in long-lasting simulations remains limited. Here we propose dynamic training (DT), a method designed to enhance accuracy of a model over extended MD simulations. Applying DT to an equivariant graph neural network (EGNN) on the challenging system of a hydrogen molecule interacting with a palladium cluster anchored to a graphene vacancy demonstrates a superior prediction accuracy compared to conventional approaches. Crucially, the DT architecture-independent design ensures its applicability across diverse machine learning potentials, making it a practical tool for advancing MD simulations.
title Dynamic Training Enhances Machine Learning Potentials for Long-Lasting Molecular Dynamics
topic Materials Science
Chemical Physics
url https://arxiv.org/abs/2504.03521