Benchmarking deterministic optimization algorithms using an outranking approach

The comparison of optimization algorithms, through different performance measures, is not straightforward and can be perceived as a multi-criteria problem. Performance profiles, although widely used, has some difficulties in comparing deterministic algorithms. In this work, an outranking approach is studied on a set of examples. The outranking relations, based on concordance and discordance matrices for given threshold values, can be translated into a graph that explicits these relations. The results indicate that the proposed approach elucidates the merits and the disadvantages of different solvers.This work has been supported by the Portuguese Foundation for Science and Technology (FCT) in the scope of the project UID/CEC/00319/2013 (ALGORITMI R&D Center).