Recursive Flow Matching

Hugging Face Daily Papers Papers
recursive-flow-matching generative-models spatiotemporal-dynamics physics-systems flow-matching scientific-emulation

Summary

Introduces Recursive Flow Matching (RecFM), a generative framework for forecasting complex spatiotemporal dynamics that achieves high fidelity with fewer steps and improved accuracy and speed, including up to 20x speedup over diffusion-based emulators.

Generative models have emerged as a powerful paradigm for solving physics systems and modeling complex spatiotemporal dynamics. However, achieving high physical accuracy without incurring high computational cost remains a fundamental challenge, as existing approaches face a critical speed-fidelity trade-off. In this work, we introduce Recursive Flow Matching (RecFM), a generative framework for forecasting complex spatiotemporal dynamics. RecFM enforces self-consistency to align trajectories across discretization scales, reducing discretization errors and improving performance across metrics for physics-based tasks. To our knowledge, this is the first method to achieve high-fidelity one- and few-step (2-4 step) dynamic generation for scientific systems with performance comparable to state-of-the-art multi-step solvers. Across challenging scientific benchmarks, RecFM achieves up to a 20times speedup over leading diffusion-based emulators while improving predictive accuracy. Furthermore, RecFM reduces mean squared error by over 15% compared to vanilla flow matching, offering a scalable and efficient solution for real-time scientific emulation.
Original Article
View Cached Full Text

Cached at: 05/27/26, 02:48 AM

Paper page - Recursive Flow Matching

Source: https://huggingface.co/papers/2605.26535

Abstract

Recursive Flow Matching enables high-fidelity, computationally efficient forecasting of complex spatiotemporal dynamics with improved accuracy and speed compared to existing methods.

Generative modelshave emerged as a powerful paradigm for solvingphysics systemsand modeling complexspatiotemporal dynamics. However, achieving high physical accuracy without incurring high computational cost remains a fundamental challenge, as existing approaches face a critical speed-fidelity trade-off. In this work, we introduceRecursive Flow Matching(RecFM), a generative framework for forecasting complexspatiotemporal dynamics. RecFM enforcesself-consistencyto align trajectories across discretization scales, reducingdiscretization errorsand improving performance across metrics for physics-based tasks. To our knowledge, this is the first method to achieve high-fidelity one- and few-step (2-4 step) dynamic generation for scientific systems with performance comparable to state-of-the-art multi-step solvers. Across challenging scientific benchmarks, RecFM achieves up to a 20times speedup over leadingdiffusion-based emulatorswhile improving predictive accuracy. Furthermore, RecFM reduces mean squared error by over 15% compared to vanillaflow matching, offering a scalable and efficient solution for real-time scientific emulation.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2605\.26535

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.26535 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.26535 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.26535 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

SDFlow: Similarity-Driven Flow Matching for Time Series Generation

arXiv cs.AI

This paper introduces SDFlow, a similarity-driven flow matching framework for time series generation that addresses exposure bias in autoregressive models. It achieves state-of-the-art performance and inference speedups by operating in the frozen VQ latent space with low-rank manifold decomposition.

Spatiotemporal Imputation with Graph-Informed Flow Matching

arXiv cs.LG

GiFlow is a graph-informed flow matching framework for spatiotemporal imputation that replaces Gaussian priors with a graph-informed prior, and uses a hybrid vector field model combining spatial attention, temporal attention, and spatiotemporal propagation. It outperforms state-of-the-art methods on synthetic and real-world datasets.