2020/12/23 by Lehmann, Tobias, von Renesse, Max-K., Sambale, Alexander +1 · 1 citation
#49Q22 (Secondary) #65D18 (Primary) #FOS: Mathematics #Numerical Analysis (math.NA) #Optimization and Control (math.OC) #Statistics Theory (math.ST)
paper · doi:10.48550/arxiv.2012.12562
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.