Improving the performance of the repeating permutation representation using morphogenic computation and generalised modified order crossover

The permutation with repetition representation, though useful and robust, is one of the more recently developed representations in genetic algorithms. Here we introduce a new genetic operator for this representation and apply it to a problem from the scheduling domain. Furthermore we provide an analysis of the redundancy implicit in the permutation representation, such as is exasperated by the active schedule builder. Moreover, there has been an inherent lack of techniques to minimise the disparity between this representation and the complexity of the objective function used with it. To this end we examine and analyse the role of morphogenic computation, the phenotype to genotype mapping, with a view to improving the evolvability of the GA, as it is largely overlooked in this domain and for this representation.