Analysis on the Convergence of Quantum-inspired Evolutionary Algorithms
Ming Chen, Lixin X. Ding · International Journal of Advancements in Computing Technology · 2011
This article deals with the convergence of quantum-inspired evolutionary algorithms with one quantum individual and multiple observations. Applying the theory and analytical techniques in non-homogeneous Markov chain, we obtain the conclusion that quantuminspired evolutionary algorithms could converge in probability under some mild conditions imposed on the probability amplitude of the Q-bit individual. We also analyze three cases from both theoretical and experimental viewpoints. The QEA with rotation gate can not guarantee convergence in theory, while others with modified Q-gates meet the convergence conditions. Numerical results further illustrate feasibleness and effectiveness of the improved algorithms.