Estimation of distribution algorithm based on nested Archimedean copulas constructed with Lévy subordinators

This paper proposes an improved estimation of distribution algorithm(EDA) based on a class of nested Archimedean copulas which is constructed with Lévy subordinators(LNAcopula-EDA). Utilizing of Lévy subordinators, a class of nested Archimedean copulas has been conveniently constructed. In order to exploit EDA to solve high-dimensional continuous optimization problem, we fully use of the capability of nested Archimedean copula in modeling high-dimensional joint distribution of multivariate with complex rank correlation structure to construct the probability distribution model of promising individuals in EDA. Then, the procedure of LNAcopula-EDA has been presented. And comparing with other EDAs that based on copula functions for the benchmark functions in the experiments, the obtained results demonstrated the effectiveness of the proposed algorithm.