arXiv cs.AI· Shogo Sato, Tomoyasu Shimada, Kazuhiko Murasaki, Takashi Sano·· 10 小时前AI 评分27
Physics2Weight：用大语言模型从物理定律直接生成神经网络架构与权重
Physics2Weight: Direct Neural Network Architecture and Weight Generation from Physical Laws Using Large Language Models
AI 导读
Physics2Weight（PtW）框架将权重空间编程（WSP）扩展到偏微分方程求解，基于控制方程与初始、边界条件直接生成神经网络权重并编辑架构，无需梯度优化。在热传导、平流、Burgers 和 Allen–Cahn 四个 PDE 上，其解误差低于传统物理信息神经网络（PINN），但在非线性 PDE 上仍存在难以满足控制方程的局限。
正文 · 原文
Abstract:Recent advances in large language models (LLMs) have enabled weight-space programming (WSP), which directly generates neural network weights without gradient-based optimization. However, extending WSP to physical problems remains challenging, particularly in achieving both solution accuracy and physical consistency. In this paper, we propose Physics2Weight (PtW), a framework that extends WSP to solving partial differential equations by directly generating neural network weights and editing architectures based on governing equations and initial and boundary conditions. PtW introduces physics-constrained prompting and physics-guided architecture editing, combined with iterative weight refinement from WSP, to improve solution accuracy and physical consistency without gradient-based optimization. Experiments on four PDEs, namely the heat, advection, Burgers', and Allen--Cahn equations, demonstrate lower solution errors than conventional physics-informed neural networks, while revealing limitations in satisfying the governing equations for nonlinear PDEs.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.12954 [cs.AI] |
| (or arXiv:2611.12954v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.12954 arXiv-issued DOI via DataCite |
Submission history
From: Shogo Sato [view email]
[v1]
Sat, 7 Nov 2026 05:39:53 UTC (2,704 KB)
来源：arXiv cs.AI · arxiv.org