7
0

Beyond Force Metrics: Pre-Training MLFFs for Stable MD Simulations

Main:13 Pages
3 Figures
Appendix:4 Pages
Abstract

Machine-learning force fields (MLFFs) have emerged as a promising solution for speeding up ab initio molecular dynamics (MD) simulations, where accurate force predictions are critical but often computationally expensive. In this work, we employ GemNet-T, a graph neural network model, as an MLFF and investigate two training strategies: (1) direct training on MD17 (10K samples) without pre-training, and (2) pre-training on the large-scale OC20 dataset followed by fine-tuning on MD17 (10K). While both approaches achieve low force mean absolute errors (MAEs), reaching 5 meV/A per atom, we find that lower force errors do not necessarily guarantee stable MD simulations. Notably, the pre-trained GemNet-T model yields significantly improved simulation stability, sustaining trajectories up to three times longer than the model trained from scratch. These findings underscore the value of pre-training on large, diverse datasets to capture complex molecular interactions and highlight that force MAE alone is not always a sufficient metric of MD simulation stability.

View on arXiv
@article{maheshwari2025_2506.14850,
  title={ Beyond Force Metrics: Pre-Training MLFFs for Stable MD Simulations },
  author={ Shagun Maheshwari and Janghoon Ock and Adeesh Kolluru and Amir Barati Farimani and John R. Kitchin },
  journal={arXiv preprint arXiv:2506.14850},
  year={ 2025 }
}
Comments on this paper

We use cookies and other tracking technologies to improve your browsing experience on our website, to show you personalized content and targeted ads, to analyze our website traffic, and to understand where our visitors are coming from. See our policy.