Rank-Inspired Neural Network for Solving Partial Differential Equations
SIAM Journal on Scientific Computing · 30.06.2026 · 48 (3) · C453-C478
DOI: 10.1137/25m1769995 ↗Kurz gefasst
Fachartikel in der Zeitschrift SIAM Journal on Scientific Computing, online erschienen am 06.05.2026. Verfasst von Wentao Peng und 2 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Gefördert von: Scientific Research Fund of the Hunan Provincial Science and Technology Department, National Natural Science Foundation of China.
Abstract (Originalsprache)
This paper proposes a rank-inspired neural network (RINN) to tackle the initialization sensitivity issue of physics-informed extreme learning machines (PIELMs) when numerically solving PDEs. Unlike PIELM, which randomly initializes the parameters of its hidden layers, RINN incorporates a preconditioning stage. In this stage, covariance-driven regularization is employed to optimize the orthogonality of the basis functions generated by the last hidden layer. The key innovation lies in minimizing the off-diagonal elements of the covariance matrix derived from the hidden-layer output. By doing so, pairwise orthogonality constraints across collocation points are enforced, which effectively enhances both the numerical stability and the approximation ability of the optimized function space. The RINN algorithm unfolds in two sequential stages. First, it conducts a nonlinear optimization process to orthogonalize the basis functions. Subsequently, it solves the PDE constraints using the linear least-squares method. Extensive numerical experiments demonstrate that RINN significantly reduces performance variability due to parameter initialization compared to PIELM. Incorporating an early stopping mechanism based on PDE loss further improves stability, ensuring consistently high accuracy across diverse initialization settings. Reproducibility of computational results. This paper has been awarded the “SIAM Reproducibility Badge: Code and data available” as a recognition that the authors have followed reproducibility principles valued by SISC and the scientific computing community. Code and data that allow readers to reproduce the results in this paper are available at https://github.com/Lambda-None/RINN and in the supplementary materials ( RINN-main.zip [5.63MB]). [Formula: see text]
Förderung
- Scientific Research Fund of the Hunan Provincial Science and Technology Department
- National Natural Science Foundation of China
Zitieren
APA
Peng, W., Huang, Y., & Yi, N. (2026). Rank-Inspired Neural Network for Solving Partial Differential Equations. SIAM Journal on Scientific Computing, 48(3), C453-C478. https://doi.org/10.1137/25m1769995
Vancouver
Peng W, Huang Y, Yi N. Rank-Inspired Neural Network for Solving Partial Differential Equations. SIAM Journal on Scientific Computing. 2026;48(3):C453-C478. doi:10.1137/25m1769995
BibTeX
@article{Peng2026,
author = {Peng, Wentao and Huang, Yunqing and Yi, Nianyu},
title = {Rank-Inspired Neural Network for Solving Partial Differential Equations},
journal = {SIAM Journal on Scientific Computing},
year = {2026},
volume = {48},
number = {3},
pages = {C453--C478},
doi = {10.1137/25m1769995}
}
EUVIMED