Computational and Statistical Guarantees of the c-Rectified flow
Abstract
Recently, rectified flow has emerged as a fundamental framework for large-scale image generation, powering state-of-the-art systems such as FLUX.1 and Stable Diffusion 3. Despite its remarkable empirical success, the computational and statistical guarantees of iterative rectified flow have remained largely unexplored. We address this problem by studying c-rectified flow, a cost-aware class of rectified flow that projects velocity fields onto a gradient class while preserving endpoint marginals. The ordinary rectified flow can fail to recover the optimal transport coupling: in a Gaussian case study, the iteration converges to the optimal coupling if and only if the source and target covariance matrices commute. More generally, for a Gaussian to log-concave coupling, the ordinary rectified flow can reach a fixed point after one step from independent initialization. In contrast, under suitable regularity and uniform-integrability assumptions, iterative c-rectified flow always converges to the optimal transport coupling. We further establish quantitative one-step contraction and exponential convergence guarantees under projection-stability assumptions for both quadratic and strongly convex displacement costs. Finally, under a Hölder ball assumption, we develop new minimax-optimal score estimation rates and show that, when combined with iterative c-rectified flow, they yield a rate-optimal estimator of the optimal transport for the dimension d ≥ 3 and a nearly parametric rate for d=1,2.
Related
- Setting Things Straight: On the Straightness and Non-Straightness of Two-Step Rectified Flow
- Statistical Guarantees for Covariance Estimation with Adversarial Corruption under Constraints
- Reusing Statistical Guarantees Under Change: Selection Costs and Event Ownership
- Diffusion Flow Matching: Dimension-Improved KL Bounds and Wasserstein Guarantees
- On the Computational and Statistical Efficiency of the Empirical Maximum Entropy on the Mean Method
BibTeX
@misc{wang2026computational,
title = {{Computational and Statistical Guarantees of the c-Rectified flow}},
author = {Leda Wang and Zhehao Xu and Qiang Liu and Harrison H. Zhou},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2608.02487},
eprint = {2608.02487},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.02487},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - Computational and Statistical Guarantees of the c-Rectified flow
AU - Leda Wang
AU - Zhehao Xu
AU - Qiang Liu
AU - Harrison H. Zhou
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2608.02487
UR - https://arxiv.org/abs/2608.02487
AB - Recently, rectified flow has emerged as a fundamental framework for large-scale image generation, powering state-of-the-art systems such as FLUX.1 and Stable Diffusion 3. Despite its remarkable empirical success, the computational and statistical guarantees of iterative rectified flow have remained largely unexplored. We address this problem by studying c-rectified flow, a cost-aware class of rectified flow that projects velocity fields onto a gradient class while preserving endpoint marginals. The ordinary rectified flow can fail to recover the optimal transport coupling: in a Gaussian case study, the iteration converges to the optimal coupling if and only if the source and target covariance matrices commute. More generally, for a Gaussian to log-concave coupling, the ordinary rectified flow can reach a fixed point after one step from independent initialization. In contrast, under suitable regularity and uniform-integrability assumptions, iterative c-rectified flow always converges to the optimal transport coupling. We further establish quantitative one-step contraction and exponential convergence guarantees under projection-stability assumptions for both quadratic and strongly convex displacement costs. Finally, under a Hölder ball assumption, we develop new minimax-optimal score estimation rates and show that, when combined with iterative c-rectified flow, they yield a rate-optimal estimator of the optimal transport for the dimension d ≥ 3 and a nearly parametric rate for d=1,2.
ER -