CORTEXA
← Browse
arxivmath.APstat.ML2026-07-14

Wasserstein gradient flows for Coulomb discrepancies

Antonin Chodron de Courcel, Matthew Rosenzweig

We study the long-time behavior of the Wasserstein gradient flow of the squared Maximum Mean Discrepancy (MMD) between a probability measure $ρ$ and a target measure $μ$, where the underlying kernel is given by a Coulomb potential. First, we establish the existence of global weak solutions starting from arbitrary Borel probability measures and prove an ultracontractive estimate, showing that the density $ρ_t$ becomes instantly bounded in $L^\infty$ for $t>0$. We also investigate the regularity of these solutions, showing that the Hölder norm can grow exponentially in time. Second, on the flat torus $\mathbb T^\mathsf{d}$, we prove exponential decay of the squared MMD along the flow toward a uniformly positive target $μ$, without requiring a lower bound on the initial data. This result is based on a ''defective Polyak-Lojasiewicz (PL) inequality'' whose defect term accounts for possible vacuum regions in the evolving density. We also prove that the usual PL inequality may fail when the target vanishes only at one point, and, in dimensions at least two, that no coercivity constant can depend only on a prescribed positive lower bound for the target. Finally, on $\mathbb R^\mathsf{d}$, we identify an obstruction at spatial infinity. For a compactly supported target, uniformly localized sources initially separated from the target by distance $D$ retain a fixed fraction of their initial squared MMD for times of order $D$. Consequently, neither a multiplicative squared-MMD decay modulus uniform over the initial datum nor a global PL inequality can hold on the unrestricted whole-space class. By contrast, under radial symmetry, source-support inclusion, and target-positivity assumptions, we establish a PL inequality and exponential convergence.

View free PDFSource page

Related papers

arxivstat.MLmath.AP2026-06-26

Local Fokker--Planck Geometry for Score Estimation: Heat-Ball Mean-Value Representations and Exact High-Dimensional Sampling

Jiayao Bai, Lang Deng, Yi Du, Yifei Jia

Score-based generative models and Langevin samplers rely on estimating the score function $\nabla_x\log p_t(x)$ of a forward diffusion. Classically this is tractable when the drift is linear: the marginal density is Gaussian and the score is a global conditional expectation. For…

View free PDFSource page
arxivstat.MLcs.AIcs.LG2026-07-06

Wasserstein Residuals: Learning Gradient Flows from Population Dynamics

Markus Heinonen, Yair Shenfeld, Ricardo Baptista, Daniel Waxman, Dmitry Batenkov, Tim Cooijmans, et al.

Reconstructing population dynamics is a central problem in the physical and data sciences. Often, the dynamics are modeled as a Wasserstein gradient flow (WGF): a curve of distributions driven by an energy functional. Though there are multiple mathematical characterizations of a…

View free PDFSource page
arxivcs.LGstat.ML2026-07-04

A Gradient Flow Perspective on Minimum MMD Estimation

Sophia Seulkee Kang, Louis Sharrock, Xiaoyuan Cheng, François-Xavier Briol, Zonghao Chen

Minimum maximum mean discrepancy (MMD) estimation has emerged as a robust and likelihood-free alternative to maximum likelihood estimation for parameter estimation. Yet, despite its practical success, the associated optimization problem remains poorly understood, with theoretical…

View free PDFSource page
arxivcs.LGstat.ML2026-07-23

Zero-Flow Two-Sample Tests

Yakun Wang, Leyang Wang, Song Liu, Taiji Suzuki

We propose a new approach to two-sample testing for deciding whether two sets of samples are drawn from the same distribution. The test is built on a statistical discrepancy based on the zero-flow criterion, termed zero-flow discrepancy (ZFD). We prove the validity of ZFD and pro…

View free PDFSource page
arxivmath.PRstat.ML2026-06-30

Uniform-in-time Propagation-of-Chaos for Stein Variational Gradient Descent

Krishnakumar Balasubramanian, Sayan Banerjee, Anna Korba

We study uniform-in-time propagation-of-chaos for continuous-time Stein Variational Gradient Descent (SVGD). Classical finite-time propagation-of-chaos estimates for mean-field systems typically deteriorate rapidly with time and therefore do not directly explain the long-time rel…

View free PDFSource page