Machine learning potentials for modeling alloys across compositions
Killian Sheriff, Daniel Z. Xiao, Yifan Cao, Lewis R. Owen, Rodrigo Freitas · 2026
Read Machine learning potentials for modeling alloys across compositions by Killian Sheriff, Daniel Z. Xiao, Yifan Cao, Lewis R. Owen, Rodrigo Freitas (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/PMC13281821.txt.
Description
Killian Sheriff and colleagues taught software to model metal mixes at any recipe in a 2026 study. A metal's traits swing with its recipe, from neat ordered mixes to jumbled ones. Simulating each recipe from scratch is slow and costly. Learning models can help, but they often fail outside the recipes they trained on. The team used information theory to pick the most telling atom patterns to teach. The model then learned the full landscape of each mix.
They tested it on four mixes, from two-metal pairs to a four-metal blend. It foretold fault energies, local atom order, heat storage, and phase maps across recipes. Its calls matched lab data closely throughout. The authors say smart sampling, not brute data, is the key step. The method could speed the hunt for alloys built for heat, strength, or wear.
Catalog record
| CREATOR | Killian Sheriff, Daniel Z. Xiao, Yifan Cao, Lewis R. Owen, Rodrigo Freitas |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63847 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |