Modeling dual-range atomic interactions with physicochemical principles for molecular force fields
Honghao Wang, Zunlong Liu, Xiangxiang Zeng, Zhaohui Song, Chen Lin, Virginie Uhlmann · 2026
Read Modeling dual-range atomic interactions with physicochemical principles for molecular force fields by Honghao Wang, Zunlong Liu, Xiangxiang Zeng, Zhaohui Song, Chen Lin, Virginie Uhlmann (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/PMC13516912.txt.
Description
Wang and colleagues build a machine learning force field, a computer model that predicts the forces between atoms so simulations of molecules can run fast. Their 2026 system, called GeoNet, aims to capture both short range forces, such as bonds between near atoms, and long range forces that act across a molecule. Past models often missed the direction and distance of long range pulls and wobbled when a molecule changed shape.
GeoNet uses a form of attention, a method that weighs which parts matter most, over a graph linking atoms to larger molecular fragments. Extra training steps keep its answers steady across shape changes, and a fusion module shifts the balance between short and long range forces to suit each atom's local setting. Across the test sets, GeoNet beat ten leading rival models while also being the smallest model and the fastest to train.
Catalog record
| CREATOR | Honghao Wang, Zunlong Liu, Xiangxiang Zeng, Zhaohui Song, Chen Lin, Virginie Uhlmann |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |