End-to-end Symmetry Preserving Inter-atomic Potential Energy Model for Finite and Extended Systems

TL;DR

This paper introduces DeepPot-SE, a machine learning model that accurately and efficiently predicts potential energy surfaces for various systems while preserving symmetries and scaling linearly with system size.

Contribution

The paper presents DeepPot-SE, a novel end-to-end PES model that is symmetry-preserving, extensive, differentiable, and scalable for diverse finite and extended systems.

Findings

Accurately models PES for organic molecules, metals, semiconductors, and insulators.

Scales linearly with system size, enabling large-scale simulations.

Maintains all natural symmetries of the physical systems.

Abstract

Machine learning models are changing the paradigm of molecular modeling, which is a fundamental tool for material science, chemistry, and computational biology. Of particular interest is the inter-atomic potential energy surface (PES). Here we develop Deep Potential - Smooth Edition (DeepPot-SE), an end-to-end machine learning-based PES model, which is able to efficiently represent the PES for a wide variety of systems with the accuracy of ab initio quantum mechanics models. By construction, DeepPot-SE is extensive and continuously differentiable, scales linearly with system size, and preserves all the natural symmetries of the system. Further, we show that DeepPot-SE describes finite and extended systems including organic molecules, metals, semiconductors, and insulators with high fidelity.