Parallel Models for Genetic Algorithms

In the previous chapter we offered a brief introduction to metaheuristics. Now, this chapter is devoted to genetic algorithms (GA) and their parallel models. GAs [4, 20] are stochastic search methods designed for exploring complex problem spaces in order to find optimal solutions, possibly using information of the problem to guide the search. Unlike most other optimization (search, learning) techniques, a population of multiple structures is used by GAs to perform the search along many different areas of the problem space at the same time. The structures composing the population (individuals) encode tentative solutions, which are manipulated competitively by applying them some stochastic operators to find a satisfactory, if not globally, optimal solution.