Métaheuristiques pour l'optimisation difficile : recuit simulé, recherche avec tabous, algorithmes évolutionnaires et algorithmes génétiques, colonies de fourmis... / Johann Dréo, Alain Pétrowski, Patrick Siarry,... [et al.]

Trad. de : " Metaheuristics for hard optimization : methods and case studies". This is one of the very few remarkable books on metaheuristics, which aims at common features and at a unifying framework for modeling, choice of adequate methods and at first attempts to a more rigorous scientific evaluation and comparison of various approaches. In the first part of the book four of the most widely used basic approaches (simulated annealing, tabu search, evolutionary algorithms and ant colony algorithms) are introduced, compared and evaluated with respect to some basic ingredients and generic schemes. Applications in various fields are mentioned in a more or less detailed way, conclusions are drawn and practical suggestions for the beginners as well as annotated references are added. In the second part the authors are focusing on variants, extensions and methodological advices. Less well-known metaheuristics are illustrated. Extensions to continuous, multimodal and multi-objective optimization problems are shown and the integration of constraints into evolutionary optimization is discussed in detail. First, but not sufficient steps in comparing iterative optimization methods are undertaken, using estimation of probabilities (with respect to given confidence levels) on the base of Mann-Whitney tests et al. The third part of this volume illustrates three case studies of complex industrial problems in an intelligible clear manner. The appendix contains a complete implementation of tabu search for the quadratic assignment problem and the Markov chain formalism for simulated annealing. (Zentralblatt)

Bibliogr. p. 335-352. Index