Implementation and analysis of a variant of the dual method for the capacitated transshipment problem

This paper presents a variant of the dual simplex method for the capacitated pure network problem and a computational analysis of this algorithm. This work includes the considerations of different list structures to store the original problem data and the basis and the testing of various procedures to select the leaving basic variable. This study also examines the sensitivity of the implementation to changes in problem parameters. The results show that the algorithm which is presented here is superior to earlier dual implementations.