It is well known that the crossover is an important operator in the genetic algorithm (GA). We select two individuals, i.e., parents, among from a population, and apply the crossover using the parents to create a new individual (offspring). In the process of the crossover, it is important to he the offspring which is created by using the genetic information of the parents. In this paper, we give the keynote of crossover: 1) it inherits as many good genetic information as possible because they are worth preserving for offspring, and 2) the offspring created by it are guaranteed for a given problem. By using the framework of genetic local search (GLS), we compare uniform crossover (UX) and greedy crossover (GX) implemented according to the keynote for the graph bi-partitioning problem (GBP).
展开▼