Abstract:Neural network potentials (NNPs) are an innovative approach for calculating the potential energy and forces of a chemical system. In principle, these methods are capable of modeling large systems with an accuracy approaching that of a high-level ab initio calculation but with a much smaller computational cost. Due to their training to density-functional theory (DFT) data and neglect of long-range interactions, some classes of NNPs require an additional term to include London dispersion physics. In this perspective, we discuss the requirements for a dispersion model for use with an NNP, focusing on the MLXDM (Machine Learned eXchange Hole Dipole Moment) model developed by our groups. This model is based on the DFT-based XDM dispersion correction, which calculates interatomic dispersion coefficients in terms of atomic moments and polarizabilities, both of which can be effectively approximated using neural networks.

What problem does this paper attempt to address?

The paper primarily aims to address the limitations of Neural Network Potentials (NNPs) in describing intermolecular interactions, particularly the inadequate handling of long-range interactions such as London dispersion forces. Specifically, the paper addresses the following key issues: 1. **Improving the accuracy of NNPs**: Although NNPs can simulate the potential energy and forces of chemical systems at a relatively low computational cost, they are typically trained on Density Functional Theory (DFT) data and often neglect long-range interactions. This leads to deficiencies in NNPs when describing certain types of intermolecular forces, such as London dispersion forces. 2. **Incorporating dispersion effects**: To enhance the ability of NNPs to describe long-range interactions, especially London dispersion forces, researchers have developed various methods to correct these models. The paper discusses a machine learning-based XDM (Exchange Hole Dipole Moment) dispersion correction method (MLXDM), which can effectively capture intermolecular dispersion effects. 3. **Achieving dynamic response**: The MLXDM model not only supplements NNPs in describing dispersion effects but also dynamically responds to changes in the atomic environment. This means it can adjust dispersion coefficients based on different molecular structures. 4. **Comparison with other methods**: The paper also compares different molecular mechanics force fields, quantum mechanical methods, and existing NNPs, demonstrating the potential of MLXDM to maintain high accuracy while improving computational efficiency. In summary, the core objective of this paper is to improve the description of intermolecular interactions, particularly London dispersion forces, by introducing the MLXDM model to NNPs, thereby enhancing their accuracy and applicability in simulating chemical systems.

Modeling Intermolecular Interactions With XDM Dispersion Corrections to Neural Network Potentials

Modeling Intermolecular Interactions with Exchange-Hole Dipole Moment Dispersion Corrections to Neural Network Potentials

Machine Learning Interatomic Potentials and Long-Range Physics

Ab initio Accuracy Neural Network Potential for Drug-like Molecules

Modeling of Metal Nanoparticles: Development of Neural-Network Interatomic Potential Inspired by Features of the Modified Embedded-Atom Method

Neural Network Potential with Multi-Resolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution

Incorporating Neural Networks into the AMOEBA Polarizable Force Field

Combining the D3 dispersion correction with the neuroevolution machine-learned potential

The Potential of Neural Network Potentials

Global Neural Network Potential with Explicit Many-Body Functions for Improved Descriptions of Complex Potential Energy Surface.

Generalized Many-Body Dispersion Correction through Random-phase Approximation for Chemically Accurate Density Functional Theory

Spline-based neural network interatomic potentials: blending classical and machine learning models

Molecular Dynamics with Neural-Network Potentials

MLRNet: Combining the Physics-Motivated Potential Models with Neural Networks for Intermolecular Potential Energy Surface Construction.

Molecular dynamics simulation with finite electric fields using Perturbed Neural Network Potentials

Improving the reliability of machine learned potentials for modeling inhomogenous liquids

A deep potential model with long-range electrostatic interactions

Capturing long-range interaction with reciprocal space neural network

Extending the atomic decomposition and many-body representation, a chemistry-motivated monomer-centered approach for machine learning potentials

Fast and Sample-Efficient Interatomic Neural Network Potentials for Molecules and Materials Based on Gaussian Moments