Convergence of the Sinkhorn Riemannian metric for finitely supported measures

Abstract

The Sinkhorn Riemannian metric characterizes the local behavior of the Sinkhorn divergence, a debiased version of entropy-regularized optimal transport. Although the Sinkhorn divergence is known to approximate the squared Wasserstein-2 distance as the regularization parameter ε goes to $0$, the behavior of the Sinkhorn Riemannian metric as ε→ 0 remains largely open. To the best of our knowledge, the only existing argument is for measures with a density and is still a formal computation. In this work, we prove the convergence of the Sinkhorn Riemannian metric as ε→ 0 in the case of finitely supported measures undergoing horizontal perturbations. Under our assumptions, our techniques additionally allow us to prove the convergence of higher order derivatives of the Sinkhorn Riemannian metric, thereby ensuring the convergence of the associated Riemann curvature tensor and Christoffel symbols as ε→ 0. We also compare with the absolutely continuous setting and show where our approach fails for measures with a density.

Exponential Convergence of the Sinkhorn Algorithm for the Schrödinger Bridge with Regime Switching
Katharina Eichinger, Anna Kazeykina, Zhenjie Ren +1
HAL21 Jul 2026PreprintLicence not statedGeneral mathematicsHAL
Singularity of harmonic measure for finitely supported random walks
Petr Kosenko, Giulio Tiozzo
arXiv16 Sep 2026PreprintarXiv licenceFull textDynamical systems ProbabilityarXiv
Tight Nonasymptotic Local Convergence of Sinkhorn-Knopp
Wenzhi Gao, Zhaonan Qu, Yinyu Ye +1
arXiv28 Sep 2026PreprintCC BY-NC-SA 4.0Full textOptimisation & control StatisticsarXiv
Exponential convergence of Sinkhorn algorithm for entropy martingale optimal transport
Anna Kazeykina, Zhenjie Ren, Hecheng Wang
arXiv16 Sep 2026PreprintarXiv licenceFull textOptimisation & control ProbabilityarXiv
Uniform Statistical Convergence of Empirical Sinkhorn Potentials with Exponential and Polynomial Dependence on the Regularization Parameter
Denis Belomestny
arXiv29 Aug 2026PreprintCC BY 4.0Full textOptimisation & control StatisticsarXiv

Cite

BibTeX RIS

BibTeX

@misc{mordant2026convergence,
  title = {{Convergence of the Sinkhorn Riemannian metric for finitely supported measures}},
  author = {Gilles Mordant and Liane Xu},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.14651},
  eprint = {2609.14651},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.14651},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - Convergence of the Sinkhorn Riemannian metric for finitely supported measures
AU  - Gilles Mordant
AU  - Liane Xu
PY  - 2026
DA  - 2026/09/13
PB  - arXiv
DO  - 10.48550/arxiv.2609.14651
UR  - https://arxiv.org/abs/2609.14651
AB  - The Sinkhorn Riemannian metric characterizes the local behavior of the Sinkhorn divergence, a debiased version of entropy-regularized optimal transport. Although the Sinkhorn divergence is known to approximate the squared Wasserstein-2 distance as the regularization parameter ε goes to $0$, the behavior of the Sinkhorn Riemannian metric as ε→ 0 remains largely open. To the best of our knowledge, the only existing argument is for measures with a density and is still a formal computation. In this work, we prove the convergence of the Sinkhorn Riemannian metric as ε→ 0 in the case of finitely supported measures undergoing horizontal perturbations. Under our assumptions, our techniques additionally allow us to prove the convergence of higher order derivatives of the Sinkhorn Riemannian metric, thereby ensuring the convergence of the associated Riemann curvature tensor and Christoffel symbols as ε→ 0. We also compare with the absolutely continuous setting and show where our approach fails for measures with a density.
ER  -

Abstract

Related

Cite