Evolution and “Natural” Selection of Green Sheets

In this thesis, we propose an efficient solution to the course scheduling problem at the Bard College at Simon’s Rock. We need to reconcile different time preferences and overlapping constraints, which is hard to do manually. We use a method of stochastic optimization called a Genetic Algorithm to solve this problem. In this method, we generate a population of random solutions and let them evolve for several generations, selecting the fittest and increasing the diversity of the genetic pool with mutations and crossover. We show that our solution produces high quality schedules, which satisfy multiple preferences and constraints. We also discuss how this method can be used to optimize different college tasks