An effectiveness analysis of genetic operators for TSP

Although genetic algorithms (GAs) show themselves powerful and robust, the effectiveness of each genetic operator is still vague. To compare the abilities of different operators, a modified GA was designed giving all the operators the same opportunities. Then, to find the better operators, a statistical analysis was done based on the accumulated number of successful searches of each operator. The better operators were selected to form a new GA, which proved to be more powerful. It was also found that the average of the total of successful searches is linearly related to the number of cities, a fact used to estimate an ending condition for the algorithm.