Introduction to Metaheuristics 

Bruce L. Golden and Edward A. Wasil

ABSTRACT

When applied to minimization problems, traditional local improvement methods explore the neighborhood of a current solution and only select new solutions that strictly decrease the total objective function value (e.g., length or distance). Simulated annealing, deterministic annealing, smoothing algorithms, tabu search, genetic algorithms, and other related techniques can select new solutions that increase the total length and, in this way, they can avoid becoming trapped in poor local minima. We refer to these new techniques as metaheuristics and we review them in this article.