Energy landscapes of some matching-problem ensembles

The maximum-weight matching problem and the behavior of its energy landscape is numerically investigated. We apply a perturbation method adapted from the analysis of spin glasses. This gives inside into the complexity of the energy landscape of different ensembles. Erd\"os-Renyi graphs and ring graphs with randomly added edges are considered and two types of distributions for the random edge weighs are used. For maximum-weight matching, fast and scalable algorithms exist, such that we can study large graphs of more than $10^5$ nodes. Our results show that the structure of the energy landscape for standard ensembles of matching is simple, comparable to the energy landscape of a ferromagnet. Nonetheless, for some of the here presented ensembles our results allow for the presence of complex energy landscapes in the spirit of Replica-Symmetry Breaking.
Comment: 9 pages, 5 figures