Perturbative-NeuSA: A Structured Spectral Framework for Time-Dependent PDEs

DOI10.48550/arXiv.2607.24345MaRDI QIDQ7284411arXivFDO

Authors Xianli Zhu, Jia Yin

Publication date 27 July 2026

Published in arXiv

Tldr Perturbative-NeuSA: a structured spectral framework improves time-dependent PDE solvers by decomposing target solutions into low-fidelity background and high-resolution perturbation, allowing for separately measurable roles of physical structure and neural closure, and reduces training and extrapolation errors in various numerical experiments.

Full work available at URL https://doi.org/10.48550/arXiv.2607.24345

arXiv classification

cs.LGmath.NA

Cites work

• A survey of projection-based model reduction methods for parametric dynamical systems
• Adam: A Method for Stochastic Optimization
• Frequency Principle: Fourier Analysis Sheds Light on Deep Neural Networks
• Gradient-enhanced physics-informed neural networks for forward and inverse PDE problems
• Iterative Methods for Sparse Linear Systems
• Neural Ordinary Differential Equations
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Reliable extrapolation of deep neural operators informed by physics or sparse observations
• When and why PINNs fail to train: A neural tangent kernel perspective

This page was built for publication: Perturbative-NeuSA: A Structured Spectral Framework for Time-Dependent PDEs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284411)