A Comparative Study of Two Methods for Staircase Linear Programs

Publisher Website
Google Scholar
Abstract
Thin paper considers the important class of large-scale statrcase linear programs arising from dynamic or multistage models The two major approaches to large LPs--problem decomposition and basis factonzatlon--are discussed. Two methods for statrcase LPs, one from each approach, that have recently been developed and maplemented are compared in both algorithmic and data-structural aspects Computational results of an emptrical comparison are presented. The study demonstrates that both specml techniques can be more efficient than the dtrect simplex approach. It also identifies certain classes of problems for whmh a partmular techmque is especially promising.