SK converges linearly at rate 1-σ2 once near the optimum, with explicit burn-in, and can be accelerated by gradient methods on the semi-dual.
Better and simpler error analysis of the sinkhorn–knopp algorithm for matrix scaling.Mathematical Programming, 188(1):395–407, 2021
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Tight Nonasymptotic Local Convergence of Sinkhorn-Knopp
SK converges linearly at rate 1-σ2 once near the optimum, with explicit burn-in, and can be accelerated by gradient methods on the semi-dual.