Guardado en:
Detalles Bibliográficos
Autores principales: An, Kang, Si, Chenhao, Ma, Shiqian, Yan, Ming
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2604.15392
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911599291269120
author An, Kang
Si, Chenhao
Ma, Shiqian
Yan, Ming
author_facet An, Kang
Si, Chenhao
Ma, Shiqian
Yan, Ming
contents Physics-Informed Neural Networks (PINNs) often suffer from slow convergence, training instability, and reduced accuracy on challenging partial differential equations due to the anisotropic and rapidly varying geometry of their loss landscapes. We propose a lightweight curvature-aware optimization framework that augments existing first-order optimizers with an adaptive predictive correction based on secant information. Consecutive gradient differences are used as a cheap proxy for local geometric change, together with a step-normalized secant curvature indicator to control the correction strength. The framework is plug-and-play, computationally efficient, and broadly compatible with existing optimizers, without explicitly forming second-order matrices. Experiments on diverse PDE benchmarks show consistent improvements in convergence speed, training stability, and solution accuracy over standard optimizers and strong baselines, including on the high-dimensional heat equation, Gray--Scott system, Belousov--Zhabotinsky system, and 2D Kuramoto--Sivashinsky system.
format Preprint
id arxiv_https___arxiv_org_abs_2604_15392
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Lightweight Geometric Adaptation for Training Physics-Informed Neural Networks
An, Kang
Si, Chenhao
Ma, Shiqian
Yan, Ming
Machine Learning
Artificial Intelligence
Physics-Informed Neural Networks (PINNs) often suffer from slow convergence, training instability, and reduced accuracy on challenging partial differential equations due to the anisotropic and rapidly varying geometry of their loss landscapes. We propose a lightweight curvature-aware optimization framework that augments existing first-order optimizers with an adaptive predictive correction based on secant information. Consecutive gradient differences are used as a cheap proxy for local geometric change, together with a step-normalized secant curvature indicator to control the correction strength. The framework is plug-and-play, computationally efficient, and broadly compatible with existing optimizers, without explicitly forming second-order matrices. Experiments on diverse PDE benchmarks show consistent improvements in convergence speed, training stability, and solution accuracy over standard optimizers and strong baselines, including on the high-dimensional heat equation, Gray--Scott system, Belousov--Zhabotinsky system, and 2D Kuramoto--Sivashinsky system.
title Lightweight Geometric Adaptation for Training Physics-Informed Neural Networks
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.15392