Scale-PINN: Learning Efficient Physics-Informed Neural Networks Through Sequential Correction
Pao-Hsiung Chiu, Jian Cheng Wong, Chin Chun Ooi, Chang Wei, Yuchen Fan, Yew-Soon Ong · 2026
Read Scale-PINN: Learning Efficient Physics-Informed Neural Networks Through Sequential Correction by Pao-Hsiung Chiu, Jian Cheng Wong, Chin Chun Ooi, Chang Wei, Yuchen Fan, Yew-Soon Ong (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-2602-19475.txt.
Description
Physics-informed neural networks solve partial differential equations without meshes, but slow training and modest accuracy have limited their adoption against modern numerical solvers. Pao-Hsiung Chiu and five co-authors introduce Scale-PINN, the Sequential Correction Algorithm for Learning Efficient PINN. It imports the iterative residual-correction principle, a mainstay of numerical solvers, straight into the loss formulation.
That integration changes how PINN losses can be thought up and built. Scale-PINN converges at unprecedented speed across problems from different physics domains. On a challenging fluid-dynamics problem, it cuts training time for a state-of-the-art PINN from hours to under two minutes while keeping superior accuracy. That opens representative problems that were previously out of reach.
Catalog record
| CREATOR | Pao-Hsiung Chiu, Jian Cheng Wong, Chin Chun Ooi, Chang Wei, Yuchen Fan, Yew-Soon Ong |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 79 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-19475