Metaheuristic Hybrids

Over the last years, so-called hybrid optimization approaches have become increasingly popular for addressing hard optimization problems. In fact, when looking at leading applications of metaheuristics for complex real-world scenarios, many if not most of them do not purely adhere to one specific classical metaheuristic model but rather combine different algorithmic techniques. Concepts from different metaheuristics are often hybridized with each other, but they are also often combined with other optimization techniques such as branch-and-bound and methods from the mathematical programming and constraint programming fields. Such combinations aim at exploiting the particular advantages of the individual components, and in fact well-designed hybrids often perform substantially better than their “pure” counterparts. Many very different ways of hybridizing metaheuristics are described in the literature, and unfortunately it is usually difficult to decide which approach(es) are most appropriate in a particular situation. This chapter gives an overview on this topic by starting with a classification of metaheuristic hybrids and then discussing several prominent design templates which are illustrated by concrete examples. Günther R. Raidl Institute of Computer Graphics and Algorithms, Vienna University of Technology, Austria, e-mail: [email protected] Jakob Puchinger NICTA Victoria Laboratory, University of Melbourne, Australia, e-mail: [email protected] Christian Blum ALBCOM Research Group, Universitat Politècnica de Catalunya, Barcelona, Spain, e-mail: [email protected]