In this paper, we investigate the convergence properties of two very different optimization methods, Lagrangean relaxation and decomposition technique and Genetic Algorithm optimization, in integer programming. Using ANSI C codes designed by us, we applied these two methods on the famous facility location problem for optimality. We find that these two methods can perform quite differently, either in terms of convergence rates, or in terms of their final solutions. We also confirmed that initial population used in Genetic Algorithm optimization should be chosen carefully, as different initial population can give out quite different outcomes. As a result, there is no general result guaranteeing the convergence of a genetic algorithm to the global optimum.