PMosFM: Preconditioned Manifold Matching for One-Step Physics-Constrained Generation

Abstract

Physics-constrained generative models aim to generate physical fields that match a target distribution and satisfy prescribed constraints. However, enforcing these constraints often increases sampling costs through iterative corrections or training costs through residual optimization and trajectory unrolling. To address this issue, we introduce \textbf{P}reconditioned \textbf{M}anifold \textbf{o}ne-\textbf{s}tep \textbf{F}low \textbf{M}atching (\textbf{PMosFM}), a preconditioned manifold matching framework for one-step physics-constrained generation. By encoding constraints in a manifold decoder, PMosFM learns transport in intrinsic coordinates without separate residual losses or terminal residual unrolling. A geometric preconditioner rescales coordinates using the decoder-induced metric, while a regularized covariance transform approximately whitens the interpolation-state inputs. A finite-interval objective couples velocity supervision with consistency between decoded endpoints in physical space. We show that exact parameterization removes residual-induced Gauss--Newton curvature, that geometric and covariance effects separate in a local conditioning bound, and that physical flow-map error bounds endpoint distributional error. Controlled ablations examine conditioning, and experiments evaluate optimizer-update time and memory footprint. At inference, PMosFM uses one neural transport evaluation followed by physical decoding. Experiments across benchmarks show lower training and sampling time than the multi-step baselines at comparable physical and distributional fidelity. Code and datasets will be released publicly.

The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.

More in Machine Learning

• arXiv preprint · 30 September 2026

  ReGain: Restoring Subject Fidelity in Personalization on Synthetic Images

  Shubhang Bhatnagar, Ishan Bhatnagar, Viraj Shah and 1 other
• arXiv preprint · 1 October 2026

  SmoothOperator: Enhancing Representations for Fine-grained Open-set Recognition via Modulated Label Smoothing

  Thiru Thillai Nadarasar Bahavan, Yu Xia, Sachith Seneviratne and 1 other
• arXiv preprint · 30 September 2026

  Code to Control: Synthesizing Parameterized Reactive Controllers

  Zergham Ahmed, Joshua B. Tenenbaum, Chris Bates and 1 other
• arXiv preprint · 1 October 2026

  Fixing a Model That Learned Worse Cancer Means Lower Risk: Monotonic Constraints in Bladder Cancer Recurrence Prediction

  Saram Abbas, David Thomas, Naeem Soomro and 3 others
• arXiv preprint · 30 September 2026

  Molecular Property Prediction under Structural Shift with Tabular Foundation Models

  Jinmo Lee, Dooho Lee, Minho Jeong and 1 other

All new papers in Machine Learning