Active Learning of Uniformly Accurate Inter-atomic Potentials for Materials Simulation

TL;DR

The paper introduces DP-GEN, an active learning framework for creating accurate, transferable machine learning models of potential energy surfaces in materials, demonstrated on Al and Mg alloys.

Contribution

It presents a novel active learning method that efficiently constructs accurate inter-atomic potentials with minimal reference data.

Findings

DP-GEN produces uniformly accurate PES models.

The method requires fewer reference data points.

Successful application to Al, Mg, and Al-Mg alloys.

Abstract

An active learning procedure called Deep Potential Generator (DP-GEN) is proposed for the construction of accurate and transferable machine learning-based models of the potential energy surface (PES) for the molecular modeling of materials. This procedure consists of three main components: exploration, generation of accurate reference data, and training. Application to the sample systems of Al, Mg and Al-Mg alloys demonstrates that DP-GEN can produce uniformly accurate PES models with a minimal number of reference data.