Redundancy And Linear Programs

During the Spring of 1979 one of us (Zionts) was invited to visit Erasmus University in Rotterdam, The Netherlands. It was there that Zionts met another of us (Telgen) who was then in the process of completing a dissertation on redundancy in linear programming. At that time, Telgen proposed an extended visit to Buffalo, during which time he and Zionts would do an extensive study on redundancy. Redundancy, hardly an exciting or new topic, does have numerous applications. Telgen and Zionts planned the project for the Summer of 1980, and enlisted the support of all the contributors as well as the other two members of our team (Karwan and Lotfi). Lotfi was then a Ph. D. student in Industrial Engineering searching for a thesis topic. Redundancy became his topic. Karwan and Zionts served as his thesis co-chairmen, with Telgen serving as an outside reader of the thesis. We initially had hoped to complete the study during Telgen's stay in Buffalo, but that was far too optimistic. Lotfi completed his dissertation during the late Spring-early Summer of 1981. As the project took shape, we decided that we had more than enough for an article, or even several articles. Accordingly, not wanting to produce redundant papers, we decided to produce this volume --- a state-of-the-art review of methods for handling redundancy and comprehensive tests of the various methods, together with extensions and further developments of the most promising methods.

Inequalities; Equalities; Implicit equalitites; Minimal representation; Existing theory; Implicit equalities; Redundant constraints; Minimal representation; Existing methods; Results from literature; Experimental results; Nonbinding constraints; Primal-dual relations; The simplex method; The complexity of linear programming; LP-equivalent problems; The ellipsoidal method.

This analysis considers the optimum allocation of redundancy in a system of serially connected subsystems in which each subsystem is of the k-out-of-n type. Redundancy is optimally allocated when: (1) reliability is maximized for given costs; or (2) costs are minimized for given reliability. Several techniques are presented for achieving optimum allocation and their relative merits are discussed. Approximate solutions in closed form were attainable only for the special case of series-parallel systems and the efficacy of these approximations is discussed.

Postoptimal Analyses, Parametric Programming, and Related Topics: Degeneracy, Multicriteria Decision Making Redundancy.