Cooperative co-evolution for school timetabling problem

We address the automatic school timetabling problem and propose a solution based on co-evolution. A school timetable is a weekly schedule for all the teachers of the school and allocates class periods to teachers according to some constraints. Genetic algorithm has been previously used as a powerful tool for timetabling problems but the intrinsic complexity of these problems and the usually huge number of variables makes it slow and complex. The basic idea of this paper is splitting this complexity into several simple algorithms and cooperatively evolving them. We regard teachers´ timetable as various species in the population and aim to evolve them in a way that provide the best fit to the constraints. Our numerical results show that the proposed method is able to solve the school timetabling problem in much less time and with superior performance in comparison with the existing algorithms.