New Codification Schemas for Scheduling with Genetic Algorithms

Codification is a very important issue when a Genetic Algorithm is designed to dealing with a combinatorial problem. In this paper we introduce new codification schemas for the Job Shop Scheduling problem which are extensions of two schemas of common use, and are worked out from the concept of underlying probabilistic model. Someway the underlying probabilistic model of a codification schema accounts for a tendency of the schema to represent solutions in some region of the search space. We report results from an experimental study showing that in many cases any of the new schemas results to be much more efficient than conventional ones due to the new schema tends to represent more promising solutions than the others. Unfortunately the selection in advance of the best schema for a given problem instance is not an easy problem and remains still open.