A Hierarchical Genetic Algorithm (HGA) is an algorithmic technique of artificial intelligence that converges on a solution at both the atomic and structural levels. Complex adaptive and emergent systems (CAESs) are ubiquitous in our world— appearing as beehives, to social systems, to the brains in our heads that contemplate them. The modeling techniques used for understanding and predicting behavior of these systems in the past been ordinary differential equations, cellular automata, evolutionary game theory, agent based models, and networks [Ahmed2005]. Because of their implicit nested structure, self-organizing and iterative nature, however, we contend that HGAs show great promise to provide an alternative method of modeling CAESs. Moreover, by using HGA’s to intentionally generate systemic simulations of CAESs, we expect to gain insight as to how CAESs are triggered, internally controlled, and sustained.
This research includes the author’s taxonomy of techniques and implementations based on the nesting of chromosomes and of the overarching generate-measure- select algorithm canonical to genetic algorithms.
The thrust of this work lies in the simultaneous evolution of a solution at two different abstraction layers of any given system: the atomic and the structural. How the simultaneity is executed – via interleaving or as an emergent by-product of a finer level’s evolution – is largely what defines each of the algorithms presented.