Assessing the Evolutionary Algorithm Paradigm to Solve Hard Problems

Evolutionary Algorithms (EAs) are meta-heuristics based on the natural evolution of living beings. We wish to assess their use as intractable problems solver. In this paper, we first describe EAs suited to this task. We put the emphasis on the two main issues that are faced: representation of the solutions (and the related issue of operators), and the combined use of EAs with other search methods, leading to hybrid EAs. Grounded on our experience, we also strongly claim the need of a recognized (and used) framework to assess search methods. Keywords: Evolutionary algorithms, Genetic algorithms, Hybrid algorithms 1 Introduction Evolutionary Algorithms (EAs) have been imagined in the 60's in three places, with different goals in mind: ffl in Germany, both H-P. Schwefel [Sch81] and H. Rechenberg [Rec73] proposed the evolution strategies (ES) to optimize real-valued functions, ffl at UCSD, L.J. Fogel, A.J. Owens, and M.J. Walsh [FOW66] introduced the evolutionary programming (EP) techniqu..