Comparison of Two Constraint Solvers for University Timetable Creation with Focus on Algorithm Choices

Abstract: The problem of finding an optimal schedule for universities based on their numerous constraints is NP-complete and thus requires heuristics to get good solutions. Our study has compared two constraint solvers that employ different combinations of heuristics to determine which is more suitable for universities. Our results indicate that CPSolver performed better at all intervals (20, 60, and 300 seconds) for optimizing the soft constraints, however the default algorithm configuration used with Optaplanner was discovered to be sub optimal during the result analysis. CPSolver used a combination of IFS, Hill Climbing, Great Deluge and Simulated Annealing while Optaplanner’s configuration only used a version of Local Search. Therefore we conclude that when comparing the two solvers running with their default configurations, CPSolver performs better by a wide margin, but further research is required to determine which solver is better with ideal algorithm choices.