| In this paper, we focus on chaotic metaheuristic methods which solve continuous and discrete global optimization problems having many local minima. Those methods exploit the sensitive dependence on initial conditions of the chaotic dynamics and search for a solution extensively in the feasible region. Then, the performance of the chaotic generator used in those methods is very important to find a desirable solution. However, the conventional chaos generator based on the steepest descent method has a drawback, though being used widely. A sequence generated by the method tends to accumulate to the boundary of the feasible region. Thus, in some cases, it is difficult to obtain a satisfactory solution by the existing methods. Therefore, in this paper, we propose a new chaos generator based on the affine scaling method which can overcome the drawback. We apply the chaotic metaheuristic method with the proposed chaos generator to the minimization problem of a concave function and the quadratic assignment problem, and verify the efficiency of the proposed method through some numerical experiments. |
|
