Metaheuristic Hybridization with Greedy Randomized Adaptive Search Procedures

Abstract

GRASP, or greedyrandomized adaptive search procedure, is a multistart metaheuris- tic that repeatedlyapplies local search starting from solutions constructed bya ran- domized greedy algorithm. In this tutorial, I consider ways to hybridize GRASP to create new and more effective metaheuristics. I consider several types of hybridiza- tions: constructive procedures, enhanced local search, memorystructures, and cost reformulations.