A Comparison Among Simple Algorithms for Linear Programming

Abstract

This paper presents a comparison between a family of simple algorithms for linear programming and the optimal pair adjustment algorithm. The optimal pair adjustment algorithm improvements the convergence of von Neumann's algorithm which is very attractive because of its simplicity. However, it is not practical to solve linear programming problems to optimality, since its convergence is slow. The family of simple algorithms results from the generalization of the optimal pair adjustment algorithm, including a parameter on the number of chosen columns instead of just a pair of them. Such generalization preserves the simple algorithms nice features. Significant improvements over the optimal pair adjustment algorithm were demonstrated through numerical experiments on a set of linear programming problems.