Metaheuristics – Local Methods
Dan Ştefănoiu, Pierre Borne, Dumitru Popescu, Florin-Gheorghe Filip, Abdelkader El Kamel · 2014
This chapter presents the description of a collection of optimization methods. Such methods actually are inspired either by the behavior of biological entities/systems or by the evolution of some natural phenomena. The chapter focuses on a special class of optimization problems in engineering, more specifically on the class of granular optimization. The methods for solving granular optimization problems should lead to numerical computer algorithms; otherwise, they are not really useful in engineering. The heuristic methods that can be implemented on a computer are referred to as metaheuristics. One of the first approaches related to granularity of numerical problems is known as Monte Carlo method. The overall performance of improved hill climbing algorithm is inferior to other metaheuristics. The metaheuristic described in the chapter belongs to greedy descent local methods class. The chapter also discusses simulated annealing and tunneling algorithms, and greedy randomized adaptive search procedure(s) (GRASP) methods.