Extending Search Phases in the Micali-Vazirani Algorithm

Michael Huang, Clifford Stein

Abstract: The Micali-Vazirani algorithm is an augmenting path algorithm that offers the best theoretical runtime of O(n^{0.5} m) for solving the maximum cardinality matching problem for non-bipartite graphs. This paper builds upon the algorithm by focusing on the bottleneck caused by its search phase structure and proposes a new implementation that improves efficiency by extending the search phases in order to find more augmenting paths. Experiments on different types of randomly generated and real world graphs demonstrate this new implementation's effectiveness and limitations.