Forces are not Enough: Benchmark and Critical Evaluation for Machine Learning Force Fields with Molecular Simulations

TL;DR

This paper introduces a new benchmark suite for evaluating machine learning force fields in molecular dynamics, emphasizing realistic simulation metrics over traditional force prediction errors, and highlights stability as a key area for improvement.

Contribution

The paper presents a comprehensive benchmark suite with evaluation metrics aligned to scientific objectives, and critically assesses the performance of state-of-the-art ML force fields in realistic MD simulations.

Findings

Force accuracy does not correlate well with simulation quality.

Many SOTA ML force fields struggle with stability during simulations.

The benchmark suite and metrics facilitate targeted improvements.

Abstract

Molecular dynamics (MD) simulation techniques are widely used for various natural science applications. Increasingly, machine learning (ML) force field (FF) models begin to replace ab-initio simulations by predicting forces directly from atomic structures. Despite significant progress in this area, such techniques are primarily benchmarked by their force/energy prediction errors, even though the practical use case would be to produce realistic MD trajectories. We aim to fill this gap by introducing a novel benchmark suite for learned MD simulation. We curate representative MD systems, including water, organic molecules, a peptide, and materials, and design evaluation metrics corresponding to the scientific objectives of respective systems. We benchmark a collection of state-of-the-art (SOTA) ML FF models and illustrate, in particular, how the commonly benchmarked force accuracy is not well aligned with relevant simulation metrics. We demonstrate when and how selected SOTA methods fail, along with offering directions for further improvement. Specifically, we identify stability as a key metric for ML models to improve. Our benchmark suite comes with a comprehensive open-source codebase for training and simulation with ML FFs to facilitate future work.