Moritz Link, Universität Konstanz
In this talk, we present a framework for computing an enclosure of the nondominated set of multi-objective mixed-integer nonconvex problems. Starting with the introduction of the underlying ideas coming from methods from singleobjective optimization, we display the key ingredients for extending them to the multi-objective setting. Similar as for the related single-objective methods, we are able to prove finite termination of the algorithm for any required accuracy. Furthermore, our approach avoids solving any MINLPs, but only MILP-relaxations which are refined by the method only when necessary. We validate the proposed scheme with numerical experiments.
This is joint work with Stefan Volkwein.