Speaker: Denis Trystram, researcher at Laboratoire d'Informatique de Grenoble, Grenoble, France
The purpose of this talk is to give an overview of the domain of multi-objective optimization. The motivation is that most optimization problems involve many actors but only reflect one point of view. It is likely to have algorithms that provide good trade-offs between various objectives. We will first present the basic notions (dominancy, Pareto set) and discuss the main difficulties. In particular, the Pareto set may be very large. This leads to design approximation algorithms. The talk is based on the case study of resource management in extreme scale parallel and distributed platforms.
More information on XRCE seminars