Effective Application of Guided Local Search

The Guided Local Search (GLS) algorithm is a metaheuristic technique that is a flexible and relatively simple to implement and apply, with few parameters to tune. The versatility of the method has been demonstrated by its several applications to problems with various structures and objectives from routing and scheduling to assignment problems and constraint optimization. In this article, we focus on the effective application of GLS. We do that by describing the details of the main algorithm and the two closely related methods of Fast Local Search (FLS) and Guided Fast Local Search (GFLS). Suggestions and insights on how to efficiently implement the techniques are provided. We also address modeling considerations such as the definition of GLS features and their costs for common classes of problems. Keywords: guided local search; fast local search; λ parameter; routing and scheduling problems; traveling salesman problem