Evolutionary programming using the Levy probability distribution
Chang‐Yong Lee, Yoonseon Song · 1999
In this work, an evolutionary programming, using the mutation operator based on the Levy probability distribution, was studied. The Levy stable probability distribution has an infinite second moment and has been widely studied in conjunction with the fractal structure of nature. Because of the infinite second moment of the Levy probability distribution, the Levy mutation operator is more likely to generate an offspring that is farther away from its parents than the conventional Gaussian mutation operator. As a result, one can search wider area in the variable space with the Levy mutation. The proposed algorithm was applied to the multivariate functional optimization and we showed empirically that in the case of functions having many local optima, the performance of the proposed algorithm was far better than that of the conventional evolutionary programming.