English

Stabilizing Physics-Informed Consistency Models via Structure-Preserving Training

Machine Learning 2026-02-11 v1 Numerical Analysis Numerical Analysis

Abstract

We propose a physics-informed consistency modeling framework for solving partial differential equations (PDEs) via fast, few-step generative inference. We identify a key stability challenge in physics-constrained consistency training, where PDE residuals can drive the model toward trivial or degenerate solutions, degrading the learned data distribution. To address this, we introduce a structure-preserving two-stage training strategy that decouples distribution learning from physics enforcement by freezing the coefficient decoder during physics-informed fine-tuning. We further propose a two-step residual objective that enforces physical consistency on refined, structurally valid generative trajectories rather than noisy single-step predictions. The resulting framework enables stable, high-fidelity inference for both unconditional generation and forward problems. We demonstrate that forward solutions can be obtained via a projection-based zero-shot inpainting procedure, achieving consistent accuracy of diffusion baselines with orders of magnitude reduction in computational cost.

Keywords

Cite

@article{arxiv.2602.09303,
  title  = {Stabilizing Physics-Informed Consistency Models via Structure-Preserving Training},
  author = {Che-Chia Chang and Chen-Yang Dai and Te-Sheng Lin and Ming-Chih Lai and Chieh-Hsin Lai},
  journal= {arXiv preprint arXiv:2602.09303},
  year   = {2026}
}
R2 v1 2026-07-01T10:28:59.399Z