Dynamic Behavior Analysis Of The Evolutionary Algorithm Based On Real Coding In The General Search Space

This paper researches on the dynamic behavior of the evolutionary algorithm based on real coding in the general search space, mainly including convergence and time complexity. The paper introduces the existing results of convergence and time complexity of the evolutionary algorithm based on the binary coding and real coding under the homogeneous and inhomogeneous case. Using the theoretical knowledges and methods of theory of probability and random process and other areas, we obtain further conclusions of the convergence and time complexity based on real coding in the general search space. In this paper,we have two research results. On the one hand to generalize the convergence results in homogeneous case of the evolutionary algorithm based on the real coding, we get the convergence theorems in inhomogeneous case; On the other hand, using the drift analysis method, we get the upper bound of the expected first hitting time to the optimal solution based on real number coding in the general search space under homogeneous and inhomogeneous cases.As the application of the theoretical results, we analyzes the convergence and time complexity of the elitist selection strategy evolutionary algorithm for the specific optimization problems in the last part of the paper.