Computing two-person zero sum games at multiple times the speed of Linear Programming solvers

It has been proposed in the paper ‘Reducing the complexity of computing the values of a Nash Equilibrium’ by the authors that the optimal row and column strategies and value of a two-person, zero sum game can be more efficiently computed than by linear programming optimization and other existing methods. That paper had described the algorithm for achieving such a result. The present paper presents a computational validation of the claimed result of the algorithm by test of the speed of its performance in comparison with the speed achieved by the GLOP linear programming solver. Evidence is also adduced in favor of the algorithm’s performance much exceeding that of other open-source and commercial LP solvers. While the framework of the Colonel Blotto game has been adopted to carry out the performance tests, the results achieved by the algorithm is generalizable and valid for all two-person, zero sum games.