New Pivoting Rule for the Dual Network Simplex Method

摘要

A new pivoting rule for the dual simplex method for solving the linear programming problem is applied to the network flow problem. The results in a new pivoting rule for the dual network simplex method. The rule significantly reduces the number of required pivoting steps; the use of appropriate data structures gives rise to a significant reduction in computational times. Computational experiments are reported, including a comparison with the fastest dual and primal network codes existing today, namely DUALINC and ARCNET respectively. (Copyright (c) 1987 by Faculty of Mathematics and Informatics, Delft, The Netherlands.)