A note on overrelaxation in the Sinkhorn algorithm

A note on overrelaxation in the Sinkhorn algorithm

We derive an a priori parameter range for overrelaxation of the Sinkhorn algorithm, which guarantees global convergence and a strictly faster asymptotic local convergence. Guided by the spectral analysis of the linearized problem we pursue a zero cost procedure to choose a near optimal relaxation pa...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Optimization letters 2022-11, Vol.16 (8), p.2209-2220
Hauptverfasser: Lehmann, Tobias, von Renesse, Max-K, Sambale, Alexander, Uschmajew, André
Format: Artikel
Sprache:eng
Schlagworte:
Computational Intelligence
Mathematics
Mathematics and Statistics
Matrix scaling
Numerical and Computational Physics
Operations Research/Decision Theory
Optimal transport
Optimization
Original Paper
Overrelaxation
Simulation
Sinkhorn algorithm
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We derive an a priori parameter range for overrelaxation of the Sinkhorn algorithm, which guarantees global convergence and a strictly faster asymptotic local convergence. Guided by the spectral analysis of the linearized problem we pursue a zero cost procedure to choose a near optimal relaxation parameter.
ISSN:1862-4480
1862-4472
1862-4480
DOI:10.1007/s11590-021-01830-0