GENETIC SEARCH ALGORITHMS AND THEIR RANDOMIZED OPERATORS

Genetic algorithms are stochastic search procedures based on randomized operators such as crossover and mutation. Many of the earlier works that have attempted to analyse the behaviour of genetic algorithms have marginalized the role of these operators through gross abstractions. In this paper, we place in perspective these operators within the genetic search strategy. Certain important properties of these operators have been brought out through simple formalisms. The search behaviour of the genetic algorithm has been modeled in a Markovian framework and strong convergence proved. The limiting analysis of the algorithm reveals the vital interplay of the randomized operators