Online Test-time Adaptation for Interatomic Potentials

TL;DR

This paper introduces TAIP, an online test-time adaptation framework for interatomic potentials that improves generalization and stability in molecular dynamics simulations under distribution shifts without extra data.

Contribution

The paper presents a dual-level self-supervised learning approach for test-time adaptation of MLIPs, enhancing their robustness across diverse molecular datasets.

Findings

TAIP improves test accuracy on multiple benchmarks.

TAIP enables stable MD simulations where baseline models fail.

It effectively reduces domain gap without additional data.

Abstract

Machine learning interatomic potentials (MLIPs) enable more efficient molecular dynamics (MD) simulations with ab initio accuracy, which have been used in various domains of physical science. However, distribution shift between training and test data causes deterioration of the test performance of MLIPs, and even leads to collapse of MD simulations. In this work, we propose an online Test-time Adaptation Interatomic Potential (TAIP) framework to improve the generalization on test data. Specifically, we design a dual-level self-supervised learning approach that leverages global structure and atomic local environment information to align the model with the test data. Extensive experiments demonstrate TAIP's capability to bridge the domain gap between training and test dataset without additional data. TAIP enhances the test performance on various benchmarks, from small molecule datasets to complex periodic molecular systems with various types of elements. Remarkably, it also enables stable MD simulations where the corresponding baseline models collapse.