Physics-Guided Conditional Flow Matching with Energy Regularization for Robust PDE Inverse Problems

DOI10.48550/arXiv.2609.15536MaRDI QIDQ7345569arXivFDO

Authors Shuo Lu, Xinghui Zhong, Yongsheng Chen, Wei Guo

Publication date 14 September 2026

Published in arXiv

Tldr We develop physics-guided conditional flow matching with energy regularization for solving PDE inverse problems from sparse noisy observations by incorporating strong-form PDE information through residual regularization along generative trajectories and assigning each observation a physics-data energy score, achieving state-of-the-art accuracy on several benchmarks.

Full work available at URL https://doi.org/10.48550/arXiv.2609.15536

Mathematics Subject Classification ID

Numerical methods for inverse problems for initial value and initial-boundary value problems involving PDEs (65M32) Numerical methods for inverse problems for boundary value problems involving PDEs (65N21) Artificial neural networks and deep learning (68T07)

arXiv classification

math.NA

Cites work

• B-PINNs: Bayesian physics-informed neural networks for forward and inverse PDE problems with noisy data
• Modern regularization methods for inverse problems
• Neural Ordinary Differential Equations
• PDEformer-2: A Versatile Foundation Model for Two-Dimensional Partial Differential Equations
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Solving inverse problems using data-driven models
• Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks

This page was built for publication: Physics-Guided Conditional Flow Matching with Energy Regularization for Robust PDE Inverse Problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7345569)