Hybrid Metaheuristics: Powerful Tools for Optimization (Artificial Intelligence: Foundations, Theory, and Algorithms)

This book explains the most prominent and some promising new, general techniques that combine metaheuristics with other optimization methods. A first introductory chapter reviews the basic principles of local search, prominent metaheuristics, and tree search, dynamic programming, mixed integer linear programming, and constraint programming for combinatorial optimization purposes. The chapters that follow present five generally applicable hybridization strategies, with exemplary case studies on selected problems: incomplete solution representations and decoders; problem instance reduction; large neighborhood search; parallel non-independent construction of solutions within metaheuristics; and hybridization based on complete solution archives.

The authors are among the leading researchers in the hybridization of metaheuristics with other techniques for optimization, and their work reflects the broad shift to problem-oriented rather than algorithm-oriented approaches, enabling faster and more effective implementation in real-life applications. This hybridization is not restricted to different variants of metaheuristics but includes, for example, the combination of mathematical programming, dynamic programming, or constraint programming with metaheuristics, reflecting cross-fertilization in fields such as optimization, algorithmics, mathematical modeling, operations research, statistics, and simulation. The book is a valuable introduction and reference for researchers and graduate students in these domains.

本書解釋了最重要的和一些有前途的新的、結合元啟發式與其他優化方法的一般技術。首先介紹了基本的局部搜索原則、著名的元啟發式方法，以及用於組合優化目的的樹搜索、動態規劃、混合整數線性規劃和約束規劃。接下來的章節介紹了五種通用的混合策略，並以選定問題的實例案例進行了研究：不完整解表示和解碼器；問題實例簡化；大鄰域搜索；元啟發式中非獨立解構建的並行；以及基於完整解存檔的混合化。

作者是元啟發式與其他優化技術混合化領域的領先研究人員之一，他們的工作反映了從以算法為導向轉向以問題為導向的廣泛趨勢，使得在實際應用中能夠更快速、更有效地實施。這種混合化不僅限於不同變體的元啟發式方法，還包括數學規劃、動態規劃或約束規劃與元啟發式方法的結合，反映了在優化、算法、數學建模、運籌學、統計學和模擬等領域的交叉學習。本書對於這些領域的研究人員和研究生來說是一本有價值的介紹和參考資料。