Solution transfer rates in graph based evolutionary algorithms

Combinatorial graphs have recently been used to control the rate of information spread in evolutionary algorithms, allowing for the preservation of diversity found necessary as the fitness landscape grows in complexity. This paper examines the combined effect of graph type and population size on the transmittal of a solution using graph based evolutionary algorithms. This study identifies a correlation between population size, graph, and time for a good solution to spread. While no numerical relationships are introduced here, it is readily apparent that the required number of mating events for a solution to spread across an entire graph is proportional to the graph diameter, population size, and the fitness difference of the individuals.