Academy & Industry Research Collaboration Center
In this paper, a hierarchical population-based memetic algorithm for solving the satisfiability problem is presented. The approach suggests looking at the evolution as a hierarchical process evolving from a coarse population where the basic unit of a gene is composed of cluster of variables that represent the problem to a fine population where each gene represents a single variable. The optimization process is carried out by letting the converged population at a child level serve as the initial population to the parent level. A benchmark composed of industrial instances is used to compare the effectiveness of the hierarchical approach against its single-level counterpart.