Integer Programming with Bounded Variables via Canonical Separation

Abstract

This paper presents a new algorithm for integer programming with bounded variables which is efficient when m < n and when the upper bounds on the variables are small. The main idea is the application of the Balas and Jeroslow canonical hyperplanes and the systematic search of integer points over certain faces of the feasible region. During each iteration the integer points on a certain face are examined, and then this whole face is discarded from the feasible region of a linear programming problem. After a bounded number of iterations, the optimal integer solution is found, if one exists.