Applying Integer Programming Techniques to Find Minimum Integer Weights of Voting Games

Using concepts from computer science and mathematics I develop three algorithms to find the minimum integer weights for voting games. Games with up to at least 17 players can be solved in a reasonable amount of time. First, coalitions are mapped to constraints, reducing the problem to constraint optimization. The optimization techniques used are Gomory’s all integer simplex algorithm and a variant of the popular integer programming method branch and bound. Theoretical results include that minimum integer weights are not unique and a confirmation of a prior result that minimum integer weights are proportional to a priori seat share. Thus, these algorithms can be useful for researchers evaluating the differences between proportional bargaining models and formateur models. The running times of the different algorithms are contrasted and analyzed for potential improvements. Thesis Supervisor: Stephen Ansolabehere Title: Professor, Department of Political Science