Understanding Zadimoghaddam's Edge-weighted Online Matching Algorithm: Weighted Case

10/08/2019

∙

by   Zhiyi Huang, et al.

∙

0

∙

share

This article presents a simplification of Zadimoghaddam's algorithm for the edge-weighted online bipartite matching problem, under the online primal dual framework. In doing so, we obtain an improved competitive ratio of 0.514. We first combine the online correlated selection (OCS), an ingredient distilled from Zadimoghaddam (2017) by Huang and Tao (2019), and an interpretation of the edge-weighted online bipartite matching problem by Devanur et al. (2016) which we will refer to as the complementary cumulative distribution function (CCDF) viewpoint, to derive an online primal dual algorithm that is 0.505-competitive. Then, we design an improved OCS which gives the 0.514 ratio.