Abstract
When and why can evolutionary multi-objective optimization (EMO) algorithmscover the entire Pareto set? That is a major concern for EMO researchers andpractitioners. A recent theoretical study revealed that (roughly speaking) ifthe Pareto set forms a topological simplex (a curved line, a curved triangle, acurved tetrahedron, etc.), then decomposition-based EMO algorithms can coverthe entire Pareto set. Usually, we cannot know the true Pareto set and have toestimate its topology by using the population of EMO algorithms during or afterthe runtime. This paper presents a data-driven approach to analyze the topologyof the Pareto set. We give a theory of how to recognize the topology of thePareto set from data and implement an algorithm to judge whether the truePareto set may form a topological simplex or not. Numerical experiments showthat the proposed method correctly recognizes the topology of high-dimensionalPareto sets within reasonable population size.