Ultra Fast PDE Solving via Physics Guided Few-step Diffusion

TL;DR

Phys-Instruct introduces a physics-guided distillation framework that significantly accelerates PDE solving with fewer steps and improved physical consistency, outperforming existing diffusion models in speed and accuracy.

Contribution

It presents a novel distillation method that compresses diffusion PDE solvers into few-step generators with explicit physics constraints, enabling ultra-fast and physically consistent PDE solutions.

Findings

Achieves orders-of-magnitude faster inference than baselines.

Reduces PDE error by more than 8 times.

Enables efficient downstream conditional PDE tasks.

Abstract

Diffusion-based models have demonstrated impressive accuracy and generalization in solving partial differential equations (PDEs). However, they still face significant limitations, such as high sampling costs and insufficient physical consistency, stemming from their many-step iterative sampling mechanism and lack of explicit physics constraints. To address these issues, we propose Phys-Instruct, a novel physics-guided distillation framework which not only (1) compresses a pre-trained diffusion PDE solver into a few-step generator via matching generator and prior diffusion distributions to enable rapid sampling, but also (2) enhances the physics consistency by explicitly injecting PDE knowledge through a PDE distillation guidance. Physic-Instruct is built upon a solid theoretical foundation, leading to a practical physics-constrained training objective that admits tractable gradients. Across five PDE benchmarks, Phys-Instruct achieves orders-of-magnitude faster inference while reducing PDE error by more than 8 times compared to state-of-the-art diffusion baselines. Moreover, the resulting unconditional student model functions as a compact prior, enabling efficient and physically consistent inference for various downstream conditional tasks. Our results indicate that Phys-Instruct is a novel, effective, and efficient framework for ultra-fast PDE solving powered by deep generative models.