MP-ALOE: an r2SCAN dataset for universal machine learning interatomic potentials

作者：Matthew C. Kuner, Aaron D. Kaplan, Kristin A. Persson, Mark Asta, D. C. Chrzan · 发表于：npj Computational Materials · 年份：2025 · DOI：10.1038/s41524-025-01834-9 · 被引用次数：11 · 研究领域：Machine Learning in Materials Science、Model Reduction and Neural Networks、Crystallography and molecular interactions

Abstract We present MP-ALOE, a dataset of nearly 1 million DFT calculations using the accurate r 2 SCAN meta-generalized gradient approximation. Covering 89 elements, MP-ALOE was created using active learning and primarily consists of off-equilibrium structures. We benchmark a machine learning interatomic potential trained on MP-ALOE, and evaluate its performance on a series of benchmarks, including predicting the thermochemical properties of equilibrium structures; predicting forces of far-from-equilibrium structures; maintaining physical soundness under static extreme deformations; and molecular dynamic stability under extreme temperatures and pressures. MP-ALOE shows strong performance on all of these benchmarks and is made public for the broader community to utilize.