Hierarchical Iterated Local Search for the Quadratic Assignment Problem

Abstract

Iterated local search is a stochastic local search (SLS) method that combines a perturbation step with an embedded local search algorithm. In this article, we propose a new way of hybridizing iterated local search. It consists in using an iterated local search as the embedded local search algorithm inside another iterated local search. This nesting of local searches and iterated local searches can be further iterated, leading to a hierarchy of iterated local searches. In this paper, we experimentally examine this idea applying it to the quadratic assignment problem. Experimental results on large, structured instances show that the hierarchical iterated local search can offer advantages over using a flat iterated local search and make it a promising technique to be further considered for other applications.