Approximation algorithms for maximum matchings in undirected graphs

International audienceWe propose heuristics for approximating the maximum cardinality matching on undirected graphs. Our heuristics are based on the theoretical body of a certain type of random graphs, and are made practical for real-life ones. The idea is based on judiciously selecting a subgraph of a given graph and obtaining a maximum cardinality matching on this subgraph. We show that the heuristics have an approximation guarantee of around 0.866 − log(n)/n for a graph with n vertices. Experiments for verifying the theoretical results in practice are provided