Iterated local search: framework and applications

The key idea underlying iterated local search is to focus the search not on the full space of all candidate solutions but on the solutions that are returned by some underlying algorithm, typically a local search heuristic. The resulting search behavior can be characterized as iteratively building a...

Descripción completa

Detalles Bibliográficos
Autores: Ramalhinho-Lourenço, Helena, Martin, Olivier C., Stützle, Thomas
Tipo de recurso: capítulo de libro
Estado:Versión aceptada para publicación
Fecha de publicación:2019
País:España
Institución:Universitat Pompeu Fabra
Repositorio:Repositorio Digital de la UPF
OAI Identifier:oai:repositori.upf.edu:10230/69330
Acceso en línea:http://hdl.handle.net/10230/69330
http://dx.doi.org/10.1007/978-3-319-91086-4_5
Access Level:acceso abierto
Palabra clave:Optimització combinatòria
Investigació operativa
Informàtica -- Matemàtica
Descripción
Sumario:The key idea underlying iterated local search is to focus the search not on the full space of all candidate solutions but on the solutions that are returned by some underlying algorithm, typically a local search heuristic. The resulting search behavior can be characterized as iteratively building a chain of solutions of this embedded algorithm. The result is also a conceptually simple metaheuristic that nevertheless has led to state-of-the-art algorithms for many computationally hard problems. In fact, very good performance is often already obtained by rather straightforward implementations of the metaheuristic. In addition, the modular architecture of iterated local search makes it very suitable for an algorithm engineering approach where, progressively, the algorithm’s performance can be further optimized. Our purpose here is to give an accessible description of the underlying principles of iterated local search and a discussion of the main aspects that need to be taken into account for a successful application of it. In addition, we review the most important applications of this method and discuss its relationship with other metaheuristics.