Theoretical Aspects of Evolutionary Multiobjective Optimization

Evolutionary multiobjective optimization (EMO), the optimization of problems with multiple objectives by means of evolutionary computation methods, has become one of the main approaches to tackle real-world problems in recent years. Although theory in EMO is less established than for single-objective randomized search heuristics or the classical field of deterministic multiobjective optimization, several important theoretical results have been accomplished in recent years. This chapter gives a broad overview over those theoretical studies obtained in the field while focusing on the topics performance assessment, hypervolume- based search, and rigorous runtime analyses and convergence results.