On Optimality Condition of Complex Systems: Computational Evidence

Abstract

A general condition determining the optimal performance of a complex system has not yet been found and the possibility of its existence is unknown. To contribute in this direction, an optimization algorithm as a complex system is presented. The performance of the algorithm for any problem is controlled as a convex function with a single optimum. To characterize the performance optimums, certain quantities of the algorithm and the problem are suggested and interpreted as their complexities. An optimality condition of the algorithm is computationally found: if the algorithm shows its best performance for a problem, then the complexity of the algorithm is in a linear relationship with the complexity of the problem. The optimality condition provides a new perspective to the subject by recognizing that the relationship between certain quantities of the complex system and the problem may determine the optimal performance.