Initial Population for Genetic Algorithms: A Metric Approach

Besides the difficulty of the application problem to be solved with Genetic Algorithms (GAs), an additional difficulty arises because the quality of the solution found, or the computational resources required to find it, depends on the selection of the Genetic Algorithm’s characteristics. The purpose of this paper is to gain some insight into one of those characteristics: the difficult problem of finding a good initial population. We summarize previous approaches and metrics related to this problem and we suggest the center of mass as an alternative metric to measurement diversity at the population level. This theoretical approach of analysis and measure of the diversity of the initial random population is important and could be quite necessary for the design of GAs because of the relation of the initial population to other GA parameters and operators and because of its relation to the problem of premature convergence.