作者: Buddhima Gamlath , Michael Kapralov , Andreas Maggiori , Ola Svensson , David Wajc
关键词: Online algorithm 、 Combinatorics 、 Blossom algorithm 、 Competitive analysis 、 Randomized algorithm 、 Bipartite graph 、 Vertex (geometry) 、 Existential quantification 、 Computer science
摘要: The online matching problem was introduced by Karp, Vazirani and nearly three decades ago. In that seminal work, they studied this in bipartite graphs with vertices arriving only on one side, presented optimal deterministic randomized algorithms for setting. comparison, more general arrival models, such as edge arrivals vertex arrivals, have proven challenging positive results are known various relaxations of the problem. particular, even basic question whether randomization allows to beat trivially-optimal competitive ratio 1/2 either these models open. paper, we resolve both natural show following.1) For does not help --- no algorithm is better than competitive. 2)For helps there exists a (1/2+Ω(1)) -competitive algorithm.