Genetic Heuristic for Search Space Exploration

This paper deals wi th the way dual genetic algorithms ( D G A ) , an extension of the standard ones, explore the search space. After a brief introduction presenting genetic algorithms and dualism, the fitness distance correlation is discussed in the context of dualism. From this discussion, a conjecture is made about the genetic: heuristic used by dual genetic algorithms to explore the search space. This conjecture is reinforced by the visualization of the population centroid trajectories in the plane fitness distance. These trajectories help to point out "leg-up" behaviors, which allow the dual genetic algorithm to reach the global optimum from walks on deceptive paths.