Simulated annealing with penalization for university course timetabling

In this paper, we present our solver for the new variant of the University Timetabling Problem, which was introduced in the framework of Fourth International Timetabling Competition (ITC2019). This problem is defined on top of previous course timetabling problems in the literature, but introduces several new elements, both in terms of new features like student sectioning and new required and optional elements like distribution constraints. Our approach for solving this problem is based on the simulated annealing metaheuristic and consists of two phases. The first phase focuses on finding a feasible solution, and the second phase attempts to optimize the final score while keeping the solution feasible. Our solver detects local optima and applies gradual penalization to force solutions to new neighborhoods. The solver also detects required constraints which are difficult to satisfy and performs a specialized search on them. These adaptively applied mechanisms allow the solver to find feasible solutions for all problem instances of the competition. Results show that our solver gives good overall results and is competitive against other approaches presented in ITC2019.