Course scheduling problem is hard and time-consuming to solve which is commonly faced by academic administrator at least two times every year. This problem can be solved using search and optimization technique with many constraints. This problem has been well studied in the past, and still becomes favorite subject for researchers. We will briefly discuss the convergence difficulty in our initial work on this subject using a modified hill-climbing search technique[8].  In this paper, an evolutionary algorithm is applied to solve the course scheduling problem and studying mutation techniques involved in the algorithm.

No abstract

Course scheduling problem is a combinatorial optimization problem which is defined over a finite discrete problem whose candidate solution structure is expressed as a finite sequence of course events scheduled in available time and space resources. This problem is considered as non-deterministic polynomial complete problem which is hard to solve. Many solution methods have been studied in the past for solving the course scheduling problem, namely from the most traditional approach such as graph coloring technique; the local search family such as hill-climbing search, taboo search, and simulated annealing technique; and various population-based metaheuristic methods such as evolutionary algorithm, genetic algorithm, and swarm optimization. This article will discuss these various probabilistic optimization methods in order to gain the global optimal solution. Furthermore, inclusion of a local search in the population-based algorithm to improve the global solution will be explained rigorously.