Surrogate Analysis on Searching Characteristics of Chaotic Dynamics for Combinatorial Optimization

摘要

An algorithm for solving combinatorial optimization problems by a chaotic neurodynamics has already been proposed. Although the algorithm shows good performance numerically, its fundamental characteristics are not clarified yet: why the chaotic search is so effective and what is the most essential characteristics of the chaotic search for combinatorial optimization. In this paper, we analyzed searching characteristics of the searches for solving combinatorial optimization problems by the method of surrogate data, which is frequently used in the nonlinear time series analysis. As the results, we revealed that its high searching ability strongly depends on a statistical property of refractory effects.