Analysis, Research and Computation, Inc. P.O. Box 4067, Austin, Texas 78765, U.S.A.;Department of General Business, College of Business Administration, BEB 600, University of Texas, Austin, Texas 78712, U.S.A.
Abstract:
Applying standard transformations of generalized upper bounding (GUB) theory to a pure or generalized network basis is shown to yield a reduced working basis that is itself a basis for a reduced network. As a result, the working basis can be represented via specialized data structures for networks. The resultant GUB based specializations to the network simplex algorithm are described.