Reactive tabu search in a team-learning problem

In designing an MBA core curriculum, student cohorts are assigned to a number of sections in each term. In a given term, different sections have exclusive schedules, while the cohorts in the same section have an identical schedule. By altering the cohorts in the sections through the terms, any cohort can meet as many other cohorts as possible via the core courses. We model this problem as a quadratic integer programming problem and construct the performance measure to characterize the properties of the problem. We design a reactive tabu search metaheuristic that solves the model effectively. Two other solution methods, constraint logic programming and an improved hill-climbing search heuristic, are also discussed. We analyze the computational results for the cases in practice.