The derivatives of sinkhorn-knopp converge

Edouard Pauwels; Samuel Vaiter

doi:10.1137/22M1512703

The derivatives of sinkhorn-knopp converge

Edouard Pauwels
, Samuel Vaiter

Centre national de la recherche scientifique
Institut Universitaire de France
Université de Nice

Research output: Contribution to journal › Article › peer-review

3 Citations (Scopus)

Abstract

We show that the derivatives of the Sinkhorn-Knopp algorithm, or iterative proportional fitting procedure, converge towards the derivatives of the entropic regularization of the optimal transport problem with a locally uniform linear convergence rate.

Original language	English
Pages (from-to)	1494-1517
Number of pages	24
Journal	SIAM Journal on Optimization
Volume	33
Issue number	3
DOIs	https://doi.org/10.1137/22M1512703
Publication status	Published - 1 Jan 2023
Externally published	Yes

Keywords

Sinkhorn algorithm
algorithmic differentiation
optimal transport

Access to Document

10.1137/22M1512703

Cite this

@article{abf24ba203734f51ba097565cae0a25b,

title = "The derivatives of sinkhorn-knopp converge",

abstract = "We show that the derivatives of the Sinkhorn-Knopp algorithm, or iterative proportional fitting procedure, converge towards the derivatives of the entropic regularization of the optimal transport problem with a locally uniform linear convergence rate.",

keywords = "Sinkhorn algorithm, algorithmic differentiation, optimal transport",

author = "Edouard Pauwels and Samuel Vaiter",

year = "2023",

month = jan,

day = "1",

doi = "10.1137/22M1512703",

language = "English",

volume = "33",

pages = "1494--1517",

journal = "SIAM Journal on Optimization",

issn = "1052-6234",

publisher = "Society for Industrial and Applied Mathematics Publications",

number = "3",

}

TY - JOUR

T1 - The derivatives of sinkhorn-knopp converge

AU - Pauwels, Edouard

AU - Vaiter, Samuel

PY - 2023/1/1

Y1 - 2023/1/1

N2 - We show that the derivatives of the Sinkhorn-Knopp algorithm, or iterative proportional fitting procedure, converge towards the derivatives of the entropic regularization of the optimal transport problem with a locally uniform linear convergence rate.

AB - We show that the derivatives of the Sinkhorn-Knopp algorithm, or iterative proportional fitting procedure, converge towards the derivatives of the entropic regularization of the optimal transport problem with a locally uniform linear convergence rate.

KW - Sinkhorn algorithm

KW - algorithmic differentiation

KW - optimal transport

UR - https://www.scopus.com/pages/publications/85171542378

U2 - 10.1137/22M1512703

DO - 10.1137/22M1512703

M3 - Article

AN - SCOPUS:85171542378

SN - 1052-6234

VL - 33

SP - 1494

EP - 1517

JO - SIAM Journal on Optimization

JF - SIAM Journal on Optimization

IS - 3

ER -

The derivatives of sinkhorn-knopp converge

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this