betaFeedback

WorksMathematics

Flow Matching from Non-Gaussian Sources: A Theory of Efficient Sampling

Abstract

Flow matching is a versatile framework for generative modeling that transports a source distribution to a target data distribution. A key advantage is its flexibility in the choice of source distribution, such as heavy-tailed and multimodal sources. Yet finite-step sampling guarantees for non-Gaussian sources remain limited. We establish a nonasymptotic convergence theory for Euler discretization of flow matching with linear interpolation. Suppose the source distribution on Rd\mathbb{R}^d has an everywhere positive, continuously differentiable density, finite total variance TotVarbase\mathsf{TotVar}_{\mathsf{base}}, and an LL-Lipschitz score. Using the population velocity field and a suitable time grid, this sampler achieves total variation error at most ε\varepsilon in O~(LdTotVarbase/ε)\widetilde{O}(Ld\mathsf{TotVar}_{\mathsf{base}}/\varepsilon) iterations, provided the target distribution satisfies a mild second-moment condition. These assumptions cover, among many examples, source distributions such as multivariate Student's tt distributions with more than two degrees of freedom and finite Gaussian mixtures with shared covariance, while accommodating target distributions with neither a smooth density nor bounded support.

Introduction

Flow matching ([34, 1, 35]) has established itself as a powerful and versatile framework for generative modeling, achieving compelling performance in image generation ([17]), speech synthesis ([26]), and even molecule generation ([54]) and protein structure generation ([8]). In a nutshell, flow matching seeks to transport a source distribution toward a target data distribution by means of an ordinary differential equation (ODE):

ddtXt=vt(Xt),t∈[0,1),X0∼Pbase,(1)\frac{\mathrm{d}}{\mathrm{d}t}X_t=v_t(X_t),\quad t\in[0,1),\quad X_0\sim P_{\mathrm{base}}, \tag*{(1)}

where PbaseP_{\mathrm{base}} represents a source distribution on Rd\mathbb{R}^{d} that is easy to sample from, and vt:Rd→Rdv_t:\mathbb{R}^{d}\to\mathbb{R}^{d} stands for a time-dependent vector field (often referred to as the velocity field). Ideally, the distribution of XtX_t converges to the target data distribution — denoted by PdataP_{\mathrm{data}} — as tt approaches 1. Flow matching is typically composed of two stages: during training, the population vector field vt(⋅)v_t(\cdot) is learned by regression using samples drawn from the source and target distributions; during sampling, one draws X0∼PbaseX_0\sim P_{\mathrm{base}} and numerically solves the ODE ((1)) from this initialization X0X_0, using a standard approach such as Euler’s method, to obtain an approximate sample from PdataP_{\mathrm{data}}.

The popularity of flow matching stems in part from its simulation-free training, empirical effectiveness, conceptual simplicity, and flexibility in choosing the source distribution and, in some formulations, the source-target coupling [35, 57, 45]. Given the widespread adoption of flow matching in generative modeling, an important practical consideration is its sampling efficiency. As with ODE-based samplers for diffusion models, generating a sample with flow matching requires numerically solving an ODE driven by a learned velocity field, typically by discretizing time into a grid and evaluating the field at each step. A finer time grid can reduce discretization error and improve sampling accuracy, but it also requires more evaluations of the velocity field, each typically involving a forward pass through a large neural network. To understand this tradeoff and guide the choice of time grid in practice, the present paper seeks to characterize the iteration complexity of the flow matching sampler, namely, the number of steps needed to achieve a prescribed sampling accuracy.

Notably, while substantial work has been devoted to establishing convergence theory for diffusion models — whose probability flow ODEs can be formulated as flow matching with Gaussian sources [2] — the analyses and insights therein do not directly carry over to flow matching with non-Gaussian sources. This theoretical gap matters in practice because non-Gaussian sources can be highly versatile: they support sampling from heavy-tailed or multimodal distributions, data-to-data transport, and the generation of non-Euclidean data. This motivates a closer examination of the role of non-Gaussian sources and the theoretical support for them.

Non-Gaussian sources in flow matching

Conventionally, Gaussian sources are the standard choice in diffusion models: the forward process of a diffusion model gradually adds Gaussian noise until the data distribution approaches a Gaussian source, while preserving tractable forward transition distributions [22, 52, 53]. However, Gaussian sources may be less well suited to target distributions with heavy tails or to data with discrete or geometric structure. Alternative source distributions have therefore been explored, with tractability as a key consideration; examples include Gamma and α\alpha-stable noise with suitable closure properties under addition [40, 62, 46, 50], and discrete corruption schemes that allow multi-step transition probabilities to be computed tractably [4].

By contrast, flow matching offers greater flexibility in the choice of source distribution. Given samples from any source and target distributions, the velocity field can be learned through regression. For heavy-tailed targets, prior work explored multivariate tt distributions as sources, reporting improved performance over Gaussian-source baselines, while other work proposed to learn quantile-based source distributions that adapt to heavy-tailed data [43, 19, 10]. Gaussian mixture sources have also been used effectively to generate multimodal targets [38, 58]. Other work employed non-Euclidean source distributions to generate discrete data or data lying on a manifold, such as DNA sequences, proteins, crystals, and geographic locations [56, 12, 39, 8, 55]. In addition, flow matching has been used for data-to-data transport, such as translating images between domains, where the source is itself a non-Gaussian data distribution [35, 57]. Despite these promising applications of non-Gaussian sources, the effects of source choice on training and sampling remain under debate [15, 27]. It is largely unclear to what extent existing convergence results for standard Gaussian sources carry over to more general sources.

A few recent studies began to analyze the convergence behavior of flow matching and its variants with non-Gaussian sources. Some assumed the learned velocity field is bounded [20], an assumption that may be restrictive for source distributions with unbounded support. Others imposed tail conditions on the target data distribution [19] or introduced Gaussian smoothing along the flow trajectory through stochastic interpolants [36, 37, 51]. This motivates the following question:

For commonly used non-Gaussian sources, can we establish nonasymptotic convergence guarantees without introducing Gaussian smoothing or imposing stronger assumptions on the target?

This paper

Main contributions. In this paper, we address the above question by analyzing the Euler sampler for flow matching when the true velocity field is available. Our results hold under the following conditions on the source and target distributions, which are satisfied by a broad class of commonly used distributions:

  • The source distribution PbaseP_{\mathrm{base}} has an everywhere positive and continuously differentiable density on Rd\mathbb{R}^{d}, finite total variance, and an LL-Lipschitz score function.

  • The target distribution PdataP_{\mathrm{data}} has a finite second moment that is not overly large relative to the source’s second moment.

Under these conditions, we prove that the Euler sampler for flow matching achieves a total-variation-based error of at most ε\varepsilon using at most

O~(Ld TotVar⁡baseε)(2)\widetilde{O}\left(\frac{Ld\,\operatorname{TotVar}_{\mathrm{base}}}{\varepsilon}\right) \tag*{(2)}

evaluations of the velocity field, where TotVar⁡base:=EZ∼Pbase[∥Z−E[Z]∥2]\operatorname{TotVar}_{\mathrm{base}} := \mathbb{E}_{Z \sim P_{\mathrm{base}}}\left[\left\|Z-\mathbb{E}[Z]\right\|^{2}\right] represents the total variance of the source distribution.

Our convergence theory is established under substantially milder assumptions on the source and target distributions than those imposed in prior work. For the source distribution, the assumptions of a Lipschitz score and a finite second moment are satisfied, for example, by multivariate Gaussian distributions with positive definite covariance matrices, multivariate Student’s tt distributions with more than two degrees of freedom, and finite Gaussian mixtures whose components share a positive definite covariance matrix; note that both Student’s tt distributions and Gaussian mixtures have been used as non-Gaussian sources in flow matching applications [43, 38]. For the target distribution, we require only a finite second moment that is not too large relative to that of the source, with no additional tail assumptions. These results demonstrate that flow matching samplers can efficiently transport a broad class of non-Gaussian source distributions, including heavy-tailed distributions, to fairly general target distributions.

From a technical standpoint, our analysis rests on carefully characterizing the nonlinear score of the source distribution and establishing a restricted injectivity property for the Euler map. Both of these technical steps are more involved than in the Gaussian case, the latter of which features an affine source score and a globally injective Euler map.

Organization. The remainder of this paper is organized as follows. § 2 formally introduces the flow matching sampler. § 3 presents our main convergence theory and its consequences, followed by additional related work in § 4. § 5 outlines the proof strategy and highlights the key steps for handling non-Gaussian source distributions. § 6 proves the main theorem, with proofs of auxiliary results deferred to the appendix. § 7 concludes the paper by discussing directions for future work.

Notation. We use the following notational conventions throughout. Capital letters X⋆,Z,Xt,YnX_{\star}, Z, X_{t}, Y_{n} denote random vectors, while lowercase letters x⋆,z,x,ynx_{\star}, z, x, y_{n} denote fixed values or integration variables. For any random vector WW, let PWP_{W} denote its law and pWp_{W} its Lebesgue density when it exists. We write pX⋆∣Xt(⋅∣x)p_{X_{\star} \mid X_{t}}(\cdot\mid x) for the density of the conditional law of X⋆X_{\star} given Xt=xX_{t}=x, with respect to Pdata=PX⋆P_{\mathrm{data}}=P_{X_{\star}}. For probability laws PP and QQ, define their total-variation (TV) distance as

TV⁡(P,Q)=12∫∣dP−dQ∣.\operatorname{TV}(P,Q)=\frac{1}{2}\int\lvert\mathrm{d}P-\mathrm{d}Q\rvert.

We also adopt the Jacobian convention [∇xu]ij=∂ui/∂xj[\nabla x u]_{ij}=\partial u_{i}/\partial x_{j}. For a symmetric matrix Σ\Sigma, λmin⁡(Σ)\lambda_{\min}(\Sigma) denotes the smallest eigenvalue of Σ\Sigma.

We use f(⋅)=O(g(⋅))f(\cdot)=O(g(\cdot)) to mean that f≤C0gf\le C_{0}g for some universal constant C0>0C_{0}>0. The notation O~(⋅)\widetilde{O}(\cdot) has the same meaning as O(⋅)O(\cdot), except that it hides the logarithmic factors. Unless stated otherwise, C,c>0C,c>0 denote numerical constants whose values may change from line to line.

Preliminaries

In this section, we formalize the flow-matching setup and the Euler sampler studied in this paper.

Flow matching setup. Recall that PbaseP_{\mathrm{base}} and PdataP_{\mathrm{data}} denote the source and target distributions on Rd\mathbb{R}^{d}, respectively. Flow matching models the transport from a source distribution PbaseP_{\mathrm{base}} to PdataP_{\mathrm{data}} through an ODE driven by a time-dependent velocity field:

dXtdt=vt(Xt),X0∼Pbase,(3)\frac{\mathrm{d}X_t}{\mathrm{d}t}=v_t(X_t), \qquad X_0\sim P_{\mathrm{base}}, \tag*{(3)}

where the velocity field {vt(⋅)}t∈[0,1)\{v_t(\cdot)\}_{t\in[0,1)} is assumed to be sufficiently regular to ensure existence and uniqueness of solutions to this ODE. Let ϕt:Rd→Rd\phi_t:\mathbb{R}^{d}\to\mathbb{R}^{d} denote the corresponding flow map, which maps each initial condition at time 00 to the solution at time tt, i.e.,

Xt=ϕt(X0).(4)X_t=\phi_t(X_0). \tag*{(4)}

Throughout this paper, we let Pt=PXtP_t=P_{X_t} denote the law of XtX_t for 0≤t<10\le t<1. We assume that PbaseP_{\mathrm{base}} and each PtP_t (0≤t<10\le t<1) admit densities pbasep_{\mathrm{base}} and ptp_t, respectively; in particular, p0=pbasep_0=p_{\mathrm{base}}. For an ideal velocity field, PtP_t converges weakly to PdataP_{\mathrm{data}} as t↑1t\uparrow1.

Flow matching trains a continuous normalizing flow by regressing a parameterized velocity field onto a target field that generates a probability path {Pt}t∈[0,1]\{P_t\}_{t\in[0,1]} connecting the source and target distributions. A core idea behind flow matching is to construct this path using conditional flows, which we introduce next.

Conditional flows. A useful way to construct a probability path from PbaseP_{\mathrm{base}} to PdataP_{\mathrm{data}} is to first specify how a source sample moves toward a given target sample, then average over target samples. More precisely, let X⋆∼PdataX_{\star}\sim P_{\mathrm{data}} and Z∼PbaseZ\sim P_{\mathrm{base}} be independent samples from the target and source distributions. Conditional on X⋆=x⋆X_{\star}=x_{\star}, consider the flow defined by

dXtdt=vt(Xt∣x⋆),X0=Z,vt(x∣x⋆)=x⋆−x1−t,t∈[0,1).(5)\frac{\mathrm{d}X_t}{\mathrm{d}t}=v_t(X_t\mid x_{\star}), \qquad X_0=Z, \qquad v_t(x\mid x_{\star})=\frac{x_{\star}-x}{1-t}, \qquad t\in[0,1). \tag*{(5)}

In words, vt(⋅∣x⋆)v_t(\cdot\mid x_{\star}) is the conditional velocity field that points from the current state toward the fixed target x⋆x_{\star}. Solving this ODE (5) for each target value yields the linear interpolant

Xt=(1−t)Z+tx⋆,t∈[0,1),(6)X_t=(1-t)Z+tx_{\star}, \qquad t\in[0,1), \tag*{(6)}

where X1=x⋆X_1=x_{\star} by continuous extension. For t<1t<1, the resulting conditional density obeys

pXt∣X⋆(x∣x⋆)=1(1−t)dpbase(x−tx⋆1−t).(7)p_{X_t\mid X_{\star}}(x\mid x_{\star})=\frac{1}{(1-t)^d}p_{\mathrm{base}}\left(\frac{x-tx_{\star}}{1-t}\right). \tag*{(7)}

This straight-line interpolation corresponds to the conditional optimal transport path in Lipman et al. [34]; it was also used by Liu et al. [35] to construct 1-rectified flow.

Marginal flow. Marginalizing the conditional densities over X⋆∼PdataX_{\star}\sim P_{\mathrm{data}} yields the marginal probability path

References

  1. [1]Michael S Albergo and Eric Vanden-Eijnden. Building normalizing flows with stochastic interpolants. In International Conference on Learning Representations, 2023.arxiv.org/abs/2209.15571
  2. [2]Michael S. Albergo, Nicholas M. Boffi, and Eric Vanden-Eijnden. Stochastic interpolants: A unifying framework for flows and diffusions. Journal of Machine Learning Research, 26(209):1–80, 2025.
  3. [3]Jason M Altschuler and Sinho Chewi. Shifted composition II: Shift Harnack inequalities and curvature upper bounds. IEEE Transactions on Information Theory, 72(1):42–57, 2026. doi: 10.1109/TIT.2025.3631770.DOI
  4. [4]Jacob Austin, Daniel D. Johnson, Jonathan Ho, Daniel Tarlow, and Rianne van den Berg. Structured denoising diffusion models in discrete state-spaces. In M. Ranzato, A. Beygelzimer, Y. Dauphin, P.S. Liang, and J. Wortman Vaughan, editors, Advances in Neural Information Processing Systems, volume 34, pages 17981–17993. Curran Associates, Inc., 2021. URL https://proceedings.neurips.cc/paper_files/paper/2021/file/958c530554f78bcd8e97125b70e6973d-Paper.pdf.arxiv.org/abs/2107.03006
  5. [5]Iskander Azangulov, George Deligiannidis, and Judith Rousseau. Convergence of diffusion models under the manifold hypothesis in high-dimensions. arXiv preprint arXiv:2409.18804, 2024.arxiv.org/abs/2409.18804
  6. [8]Avishek (Joey) Bose, Tara Akhound-Sadegh, Guillaume Huguet, Kilian Fatras, Jarrid Rector-Brooks, Cheng-Hao Liu, Andrei Cristian Nica, Maksym Korablyov, Michael Bronstein, and Alexander Tong. SE(3) stochastic flow matching for protein backbone generation. In International Conference on Learning Representations, volume 2024, pages 22590–22621, 2024.arxiv.org/abs/2310.02391
  7. [10]Jannis Chemseddine, Gregor Kornhardt, Richard Duong, and Gabriele Steidl. Adapting noise to data: Generative flows from learned 1D processes. In Proceedings of the 43rd International Conference on Machine Learning, volume 306 of Proceedings of Machine Learning Research, pages 13400–13430. PMLR, 2026. URL https://proceedings.mlr.press/v306/chemseddine26a.html.arxiv.org/abs/2510.12636
  8. [11]Hongrui Chen, Holden Lee, and Jianfeng Lu. Improved analysis of score-based generative modeling: User-friendly bounds under minimal smoothness assumptions. In International Conference on Machine Learning, pages 4735–4763. PMLR, 2023a.arxiv.org/abs/2211.01916
  9. [12]Ricky T. Q. Chen and Yaron Lipman. Flow matching on general geometries. In International Conference on Learning Representations, 2024.arxiv.org/abs/2302.03660
  10. [13]Sitan Chen, Giannis Daras, and Alex Dimakis. Restoration-degradation beyond linear diffusions: A non-asymptotic analysis for DDIM-type samplers. In International Conference on Machine Learning, pages 4462–4484. PMLR, 2023b.arxiv.org/abs/2303.03384
  11. [15]Hamza Cherkaoui, Hélène Halconruy, and Antonio Ocello. Do heavy tails help diffusion? On the subtle trade-off between initialization and training. In Advances in Neural Information Processing Systems, 2026. URL https://arxiv.org/abs/2605.13175. Forthcoming.
  12. [16]Sinho Chewi, Alkis Kalavasis, Anay Mehrotra, and Omar Montasser. DDPM score matching and distribution learning (extended abstract). In Proceedings of the 39th Conference on Learning Theory, volume 336 of Proceedings of Machine Learning Research, pages 1434–1435. PMLR, 2026. URL https://proceedings.mlr.press/v336/chewi26a.html. Full version: arXiv:2504.05161.arxiv.org/abs/2504.05161
  13. [17]Patrick Esser, Sumith Kullal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, and Robin Rombach. Scaling rectified flow transformers for high-resolution image synthesis. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 12606–12633. PMLR, 2024. URL https://proceedings.mlr.press/v235/esser24a.html.arxiv.org/abs/2403.03206
  14. [18]Kenji Fukumizu, Taiji Suzuki, Noboru Isobe, Kazusato Oko, and Masanori Koyama. Flow matching achieves almost minimax optimal convergence. In The Thirteenth International Conference on Learning Representations, 2025. URL https://openreview.net/forum?id=20WyAFijIJ.arxiv.org/abs/2405.20879
  15. [19]Yunrui Guan, Krishnakumar Balasubramanian, and Shiqian Ma. Mirror flow matching with heavy-tailed priors for generative modeling on convex domains. In International Conference on Learning Representations, volume 2026, pages 130098–130124, 2026a.
  16. [20]Yunrui Guan, Krishnakumar Balasubramanian, and Shiqian Ma. Total variation rates for Riemannian flow matching. In Advances in Neural Information Processing Systems, 2026b. URL https://arxiv.org/abs/2602.05174. Forthcoming.
  17. [21]Ianis Hammani and Claire Boyer. From global to tail convergence: Dynamics of α-stable score-based generative models, 2026. URL https://arxiv.org/abs/2610.19773. arXiv:2610.19773.
  18. [22]Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems, 33:6840–6851, 2020.arxiv.org/abs/2006.11239
  19. [23]Zhihan Huang, Yuting Wei, and Yuxin Chen. Denoising diffusion probabilistic models are optimally adaptive to unknown low dimensionality. Mathematics of Operations Research, 2026.arxiv.org/abs/2410.18784
  20. [24]Yuchen Jiao, Na Li, Changxiao Cai, Yuting Wei, and Gen Li. Forward-value discretization: Faster stochastic diffusion sampling without high-order approximation. arXiv preprint arXiv:2611.14328, 2026a.
  21. [25]Yuchen Jiao, Yuchen Zhou, and Gen Li. Optimal convergence analysis of DDPM for general distributions. In Advances in Neural Information Processing Systems, 2026b. URL https://arxiv.org/abs/2510.27562. Forthcoming.
  22. [26]Matthew Le, Apoorv Vyas, Bowen Shi, Brian Karrer, Leda Sari, Rashel Moritz, Mary Williamson, Vimal Manohar, Yossi Adi, Jay Makhodaekar, and Wei-Ning Hsu. Voicebox: Text-guided multilingual universal speech generation at scale. Advances in Neural Information Processing Systems, 36:14005–14034, 2023.
  23. [27]Junho Lee, Kwanseok Kim, and Joonseok Lee. Is there a better source distribution than Gaussian? Exploring source distributions for image flow matching. Transactions on Machine Learning Research, 2025. URL https://openreview.net/forum?id=sev0GtV1fc.
  24. [30]Gen Li, Yu Huang, Timofey Efimov, Yuting Wei, Yuejie Chi, and Yuxin Chen. Accelerating convergence of score-based diffusion models, provably. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 27942–27954. PMLR, 2024a. URL https://proceedings.mlr.press/v235/li24ad.html.arxiv.org/abs/2403.03852
  25. [31]Gen Li, Yuting Wei, Yuejie Chi, and Yuxin Chen. A sharp convergence theory for the probability flow ODEs of diffusion models. arXiv preprint arXiv:2408.02320, 2024b.
  26. [32]Gen Li, Yuchen Zhou, Yuting Wei, and Yuxin Chen. Faster diffusion models via higher-order approximation. arXiv preprint arXiv:2506.24042, 2025.arxiv.org/abs/2506.24042
  27. [33]Jiadong Liang, Zhihan Huang, and Yuxin Chen. Low-dimensional adaptation of diffusion models: Convergence in total variation. arXiv preprint arXiv:2501.12982, 2025.arxiv.org/abs/2501.12982
  28. [34]Yaron Lipman, Ricky T. Q. Chen, Heli Ben-Hamu, Maximilian Nickel, and Matthew Le. Flow matching for generative modeling. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=PqvMrdCJT9t.arxiv.org/abs/2210.02747
  29. [35]Xingchao Liu, Chengyue Gong, and Qiang Liu. Flow straight and fast: Learning to generate and transfer data with rectified flow. In International Conference on Learning Representations, 2023.arxiv.org/abs/2209.03003
  30. [36]Yuhao Liu, Yu Chen, Rui Hu, and Longbo Huang. Finite-time analysis of discrete-time stochastic interpolants. In Proceedings of the 42nd International Conference on Machine Learning, volume 267 of Proceedings of Machine Learning Research, pages 38119–38143. PMLR, 2025. URL https://proceedings.mlr.press/v267/liu25d.html.arxiv.org/abs/2502.09130
  31. [37]Yuhao Liu, Yu Chen, Rui Hu, and Longbo Huang. Finite-time convergence analysis of ODE-based generative models for stochastic interpolants. In International Conference on Learning Representations, volume 2026, pages 127011–127042, 2026.arxiv.org/abs/2508.07333
  32. [38]Gaoxiang Luo, Frank Cole, Sihang Zhang, Yuxiang Wan, Yulong Lu, and Ju Sun. Flow matching for multimodal distributions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 23260–23271, 2026.
  33. [39]Benjamin Kurt Miller, Ricky T. Q. Chen, Anuroop Siram, and Brandon M. Wood. FlowMM: Generating materials with Riemannian flow matching. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 35664–35686. PMLR, 2024. URL https://proceedings.mlr.press/v235/miller24a.html.arxiv.org/abs/2406.04713
  34. [40]Eliya Nachmani, Robin San Roman, and Lior Wolf. Non gaussian denoising diffusion models. arXiv preprint arXiv:2106.07582, 2021.arxiv.org/abs/2106.07582
  35. [42]Kazusato Oko, Shunta Akiyama, and Taiji Suzuki. Diffusion models are minimax optimal distribution estimators. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 26517–26582. PMLR, 2023. URL https://proceedings.mlr.press/v202/oko23a.html.arxiv.org/abs/2303.01861
  36. [43]Kushagra Pandey, Jaideep Pathak, Yilun Xu, Stephan Mandt, Michael Pritchard, Arash Vahdat, and Morteza Mardani. Heavy-tailed diffusion models. In International Conference on Learning Representations, 2025.arxiv.org/abs/2410.14171
  37. [45]Aram-Alexandre Pooladian, Heli Ben-Hamu, Carles Domingo-Enrich, Brandon Amos, Yaron Lipman, and Ricky TQ Chen. Multisample flow matching: Straightening flows with minibatch couplings. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 28100–28127. PMLR, 2023. URL https://proceedings.mlr.press/v202/pooladian23a.html.arxiv.org/abs/2304.14772
  38. [46]Vadim Popov, Assel Yermekova, Tasnima Sadekova, Artem Khrapov, and Mikhail Kudinov. Improved sampling algorithms for Lévy–Itô diffusion models. In International Conference on Learning Representations, volume 2025, pages 76407–76422, 2025.
  39. [47]Peter Potaptchik, Iskander Azangulov, and George Deligiannidis. Linear convergence of diffusion models under the manifold hypothesis. arXiv preprint arXiv:2410.09046, 2024.arxiv.org/abs/2410.09046
  40. [48]C. R. Rao. Information and accuracy attainable in the estimation of statistical parameters. Bulletin of the Calcutta Mathematical Society, 37(3):81–91, 1945.
  41. [49]Saptarshi Roy, Alessandro Rinaldo, and Purnamrita Sarkar. Low-dimensional adaptation of rectified flow: A new perspective through the lens of diffusion and stochastic localization. arXiv e-prints, pages arXiv–2601, 2026.arxiv.org/abs/2601.15500
  42. [50]Dario Shariatian, Umut Simsekli, and Alain Durmus. Heavy-tailed diffusion with denoising Lévy probabilistic models. In International Conference on Learning Representations, 2025.arxiv.org/abs/2407.18609
  43. [51]Marta Gentiloni Silveri, Giovanni Conforti, and Alain Oliviero Durmus. Diffusion flow matching: Dimension-improved KL bounds and Wasserstein guarantees. In Proceedings of the 43rd International Conference on Machine Learning, volume 306 of Proceedings of Machine Learning Research, pages 113117–113149. PMLR, 2026. URL https://proceedings.mlr.press/v306/silveri26a.html.arxiv.org/abs/2606.16610
  44. [52]Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. In International Conference on Learning Representations, 2021.arxiv.org/abs/2010.02502
  45. [53]Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456, 2020.arxiv.org/abs/2011.13456
  46. [54]Yuxuan Song, Jingjing Gong, Minkai Xu, Ziyao Cao, Yanyan Lan, Stefano Ermon, Hao Zhou, and Wei-Ying Ma. Equivariant flow matching with hybrid probability transport for 3d molecule generation. Advances in Neural Information Processing Systems, 36:549–568, 2023.DOI
  47. [55]Anuroop Sriram, Benjamin K Miller, Ricky T Chen, and Brandon Wood. FlowLLM: Flow matching for material generation with large language models as base distributions. Advances in Neural Information Processing Systems, 37:46025–46046, 2024.arxiv.org/abs/2410.23405
  48. [56]Hannes Stark, Bowen Jing, Chenyu Wang, Gabriele Corso, Bonnie Berger, Regina Barzilay, and Tommi Jaakkola. Dirichlet flow matching with applications to DNA sequence design. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 46495–46513. PMLR, 2024. URL https://proceedings.mlr.press/v235/stark24b.html.arxiv.org/abs/2402.05841
  49. [57]Alexander Tong, Kilian Fatrás, Nikolay Malkin, Guillaume Huguet, Yanlei Zhang, Jarrid Rector-Brooks, Guy Wolf, and Yoshua Bengio. Improving and generalizing flow-based generative models with minibatch optimal transport. Transactions on Machine Learning Research, 2024.arxiv.org/abs/2302.00482
  50. [58]Junli Wang, Yinan Zheng, Xueyi Liu, Zebin Xing, Pengfei Li, Kun Ma, Hangjun Ye, Guang Chen, Guang Li, Long Chen, et al. Meanfuser: Fast one-step multi-modal trajectory generation and adaptive reconstruction via meanflow for end-to-end autonomous driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17884–17893, 2026a.arxiv.org/abs/2602.20060
  51. [60]Yuchen Wu, Yuxin Chen, and Yuting Wei. Stochastic Runge–Kutta methods: Provable acceleration of diffusion models. arXiv preprint arXiv:2410.04760, 2024.arxiv.org/abs/2410.04760
  52. [62]Eun Bi Yoon, Keehun Park, Sungwoong Kim, and Sungbin Lim. Score-based generative models with Lévy processes. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems, volume 36, pages 40694–40707. Curran Associates, Inc., 2023. doi: 10.52202/075280-1772. URL https://proceedings.neurips.cc/paper_files/paper/2023/file/8011b23e1dc3f57e1b6211ccad498919-Paper-Conference.pdf.DOI

Paper details

vlnc:W_xnx6nf3q
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents