Nat Comput SciJClub
NEP89: universal neuroevolution potential for inorganic and organic materials across 89 elements.
Nature Computational Science · · Journal Article · Open access
Liang, Xu + more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
NEP89, a neuroevolution-potential foundation model, achieves near-empirical-potential speed with high accuracy across 89 elements, enabling large-scale atomistic simulations of inorganic and organic systems.
- Why it matters: Existing machine-learned interatomic potentials are often material-specific or too computationally intensive, restricting broader, large-scale atomistic modeling.
- What they did: The model was trained on a compact but comprehensive inorganic-and-organic dataset curated via descriptor-space subsampling and iterative refinement across multiple datasets, using a neuroevolution potential architecture.
- The result: NEP89 matches representative foundation-model accuracy while being three to four orders of magnitude more computationally efficient, supporting million-atom alloy compression, ion diffusion, rocksalt dissolution, methane combustion, and protein-ligand dynamics, plus rapid fine-tuning.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.
- Open access
- 12 cites