Learning intermolecular forces at liquid-vapor interfaces

TL;DR

This paper explores how neural network potentials can be trained to accurately model liquid-vapor interfaces by incorporating explicit long-range interactions, addressing limitations of local models in inhomogeneous systems.

Contribution

It introduces a method to improve neural network potentials for interfaces by explicitly modeling long-range interactions and training only on short-range components.

Findings

Local neural networks struggle with long-range interactions at interfaces.

Explicit electrostatics improve training and accuracy.

Models with explicit electrostatics better capture interfacial properties.

Abstract

By adopting a perspective informed by contemporary liquid state theory, we consider how to train an artificial neural network potential to describe inhomogeneous, disordered systems. We find that neural network potentials based on local representations of atomic environments are capable of describing some properties of liquid-vapor interfaces, but typically fail for properties that depend on unbalanced long-ranged interactions which build up in the presence of broken translation symmetry. These same interactions cancel in the translationally invariant bulk, allowing local neural network potentials to describe bulk properties correctly. By incorporating explicit models of the slowly-varying long-ranged interactions and training neural networks only on the short ranged components, we can arrive at potentials that robustly recover interfacial properties. We find that local neural network models can sometimes approximate a local molecular field potential to correct for the truncated interactions, but this behavior is variable and hard to learn. Generally, we find that models with explicit electrostatics are easier to train and have higher accuracy. We demonstrate this perspective in a simple model of an asymmetric dipolar fluid where the exact long-ranged interaction is known, and in an ab initio water model where it is approximated.