EquiNO: A physics-informed neural operator for multiscale simulations
Hamidreza Eivazi, Jendrik-Alexander Tröger, Stefan Wittek, Stefan Hartmann, Andreas Rausch · 2025
Read EquiNO: A physics-informed neural operator for multiscale simulations by Hamidreza Eivazi, Jendrik-Alexander Tröger, Stefan Wittek, Stefan Hartmann, Andreas Rausch (2025) 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-2504-07976.txt.
Description
Many physics tasks span two scales at once, and a full sim at fine scale costs far too much for repeat runs. Hamidreza Eivazi and co-authors build a fast stand in in 2025 that keeps one law by design: force balance. Their tool, EquiNO, learns from data yet can never break that law.
The trick is to build each answer from a set of flow free base shapes, drawn from past sims. Since each base shape is free of net force, any mix of them is too. The team sets the tool in a frame that weds grid maths with learned maps. On stiff solid tasks, it stayed true on small train sets and ran more than eight thousand times faster than the old route.
Catalog record
| CREATOR | Hamidreza Eivazi, Jendrik-Alexander Tröger, Stefan Wittek, Stefan Hartmann, Andreas Rausch |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 84 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-07976