Learning Smooth and Expressive Interatomic Potentials for Physical Property Prediction
This addresses a critical issue for computational materials scientists by improving the reliability of MLIPs for practical applications, though it is incremental as it builds on existing models.
The paper tackled the problem that low test errors in machine learning interatomic potentials (MLIPs) do not always improve physical property predictions, by proposing a test based on energy conservation in molecular dynamics to better correlate errors with performance. The resulting eSEN model achieved state-of-the-art results on tasks like materials stability and thermal conductivity prediction.
Machine learning interatomic potentials (MLIPs) have become increasingly effective at approximating quantum mechanical calculations at a fraction of the computational cost. However, lower errors on held out test sets do not always translate to improved results on downstream physical property prediction tasks. In this paper, we propose testing MLIPs on their practical ability to conserve energy during molecular dynamic simulations. If passed, improved correlations are found between test errors and their performance on physical property prediction tasks. We identify choices which may lead to models failing this test, and use these observations to improve upon highly-expressive models. The resulting model, eSEN, provides state-of-the-art results on a range of physical property prediction tasks, including materials stability prediction, thermal conductivity prediction, and phonon calculations.