Implicit Delta Learning of High Fidelity Neural Network Potentials

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Thaler, Stephan, Gabellini, Cristian, Shenoy, Nikhil, Tossou, Prudencio
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917862600343552
author Thaler, Stephan
Gabellini, Cristian
Shenoy, Nikhil
Tossou, Prudencio
author_facet Thaler, Stephan
Gabellini, Cristian
Shenoy, Nikhil
Tossou, Prudencio
contents Neural network potentials (NNPs) offer a fast and accurate alternative to ab-initio methods for molecular dynamics (MD) simulations but are hindered by the high cost of training data from high-fidelity Quantum Mechanics (QM) methods. Our work introduces the Implicit Delta Learning (IDLe) method, which reduces the need for high-fidelity QM data by leveraging cheaper semi-empirical QM computations without compromising NNP accuracy or inference cost. IDLe employs an end-to-end multi-task architecture with fidelity-specific heads that decode energies based on a shared latent representation of the input atomistic system. In various settings, IDLe achieves the same accuracy as single high-fidelity baselines while using up to 50x less high-fidelity data. This result could significantly reduce data generation cost and consequently enhance accuracy and generalization, and expand chemical coverage for NNPs, advancing MD simulations for material science and drug discovery. Additionally, we provide a novel set of 11 million semi-empirical QM calculations to support future multi-fidelity NNP modeling.
format Preprint
id arxiv_https___arxiv_org_abs_2412_06064
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Implicit Delta Learning of High Fidelity Neural Network Potentials
Thaler, Stephan
Gabellini, Cristian
Shenoy, Nikhil
Tossou, Prudencio
Chemical Physics
Machine Learning
Neural network potentials (NNPs) offer a fast and accurate alternative to ab-initio methods for molecular dynamics (MD) simulations but are hindered by the high cost of training data from high-fidelity Quantum Mechanics (QM) methods. Our work introduces the Implicit Delta Learning (IDLe) method, which reduces the need for high-fidelity QM data by leveraging cheaper semi-empirical QM computations without compromising NNP accuracy or inference cost. IDLe employs an end-to-end multi-task architecture with fidelity-specific heads that decode energies based on a shared latent representation of the input atomistic system. In various settings, IDLe achieves the same accuracy as single high-fidelity baselines while using up to 50x less high-fidelity data. This result could significantly reduce data generation cost and consequently enhance accuracy and generalization, and expand chemical coverage for NNPs, advancing MD simulations for material science and drug discovery. Additionally, we provide a novel set of 11 million semi-empirical QM calculations to support future multi-fidelity NNP modeling.
title Implicit Delta Learning of High Fidelity Neural Network Potentials
topic Chemical Physics
Machine Learning
url https://arxiv.org/abs/2412.06064