Flexible Cutoff Learning: Optimizing Machine Learning Potentials After Training
Rick Oerder, Jan Hamaekers · 2026
Read Flexible Cutoff Learning: Optimizing Machine Learning Potentials After Training by Rick Oerder, Jan Hamaekers (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-10205.txt.
Description
Rick Oerder and Jan Hamaekers introduce Flexible Cutoff Learning for machine learning interatomic potentials whose cutoff radii can change after training. Standard potentials fix the cutoff during training, locking in one accuracy cost balance. Their models train with cutoffs sampled randomly per atom, so deployment can tune each atom's radius to the task. On molecular crystals, optimised cutoffs cut compute cost by over half with almost no rise in force error.
Catalog record
| CREATOR | Rick Oerder, Jan Hamaekers |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51172 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.10205