Accession Number : ADA195734

Title :   An Averaging Algorithm For Modes. Revision,

Corporate Author : GEORGIA INST OF TECH ATLANTA PRODUCTION AND DISTRIBUTION RESEARCH CENTER

Personal Author(s) : Iyer, Ananth V ; Jarvis, John J ; Ratliff, H D

PDF Url : ADA195734

Report Date : 05 Apr 1988

Pagination or Media Count : 16

Abstract : The MODES model consists of the LIFT-CAP and MRMATE models. LIFTCAP is a network flow model. MODES uses Benders' decomposition as a solution technique that separates MRMATE and LIFTCAP and adds constraints at each iteration to the LIFTCAP model. Benders' thus represents a finitely convergent algorithm. This report presents an alternative procedure that maintains problem structure of both LIFTCAP and MRMATE. Problem sizes are maintained at each iteration of the solution procedure. It ensures asymptotic convergence to a global optimal solution. Due to its nature it is termed the averaging method. The steps involved in implementing this procedure are discussed and it is compared to Benders' decomposition technique. Computational results are also presented. These reflect the performance of averaging versus Benders' decomposition technique. The procedure is asymptotically optimal and maintains problem size and structure at each iteration. Some preliminary results were presented which demonstrate the effect of this procedure over Benders' decomposition technique. (jhd)

Descriptors :   *NETWORK FLOWS, ALGORITHMS, COMPUTATIONS, CONVERGENCE, FLOW, GLOBAL, MEAN, MODELS, NETWORKS, OPTIMIZATION, SIZES(DIMENSIONS), SOLUTIONS(GENERAL)

Subject Categories : Operations Research

Distribution Statement : APPROVED FOR PUBLIC RELEASE