CoRoPINN: Cognitive region optimized physics-informed neural networks
Jingcong Li, Yu Liao, Ximeng Wang, Jie Deng, Vikash Kumar Sinha · 2026
Read CoRoPINN: Cognitive region optimized physics-informed neural networks by Jingcong Li, Yu Liao, Ximeng Wang, Jie Deng, Vikash Kumar Sinha (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/PMC13592596.txt.
Description
Jingcong Li and four colleagues present a smarter way to train neural networks that solve physics equations. Standard physics informed networks check the equation only at scattered points, leaving gaps, and they train easy and hard regions alike. Their design, CoRoPINN, loops feedback between the two problems: gradient steadiness sets a safe neighborhood size, while each point's equation error gauges its difficulty. Hard points then get wider neighborhoods and heavier training weight.
Tests spanned three one dimensional equations, Poisson problems in two, three, and five dimensions, and a two dimensional flow case, across five network backbones. CoRoPINN posted the lowest average errors on the one dimensional set, cutting the runner up's error by roughly a fifth to three fifths. It also led on most higher dimensional tests.
Catalog record
| CREATOR | Jingcong Li, Yu Liao, Ximeng Wang, Jie Deng, Vikash Kumar Sinha |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |