Abstract: How to apply differential evolution to discrete field is currently a hot research problem in this area. Estimation of distribution algorithms build a probability model which characterizes the distribution of the current promising solutions in the search space and generates new solutions according to the model. Using the global learning of estimation of distribution algorithms, a discrete differential evolution with learning mechanism was proposed for multidimensional knack problem. Simulation results show that the proposed algorithm has good performance.

Key words: Discrete differential evolution,Learning mechanism, Estimation of distribution algorithm, Multidimensional knack problem