Abstract

When analysing computer programs (especially numerical programs in which arrays are used extensively), one is often confronted with integer programming problems. These problems have three peculiarities: ffl feasible points are ranked according to lexicographic order rather than the usual linear economic function; ffl the feasible set depends on integer parameters; ffl one is interested only in exact solutions. The difficulty is somewhat alleviated by the fact that problems sizes are usually quite small. In this paper we show that: ffl the classical simplex algorithm has no difficulty in handling lexicographic ordering; ffl the algorithm may be executed in symbolic mode, thus giving the solution of continuous parametric problems; ffl the method may be extended to problems in integers. We prove that the resulting algorithm always terminate and give an estimate of its complexity. R'esum'e L'analyse s'emantique des programmes (sp'ecialement des programmes num'eriques utilisant de...

Citations

1232	Theory of Linear and Integer Programming - Schrijver - 1986
638	Linear Programming and Extensions - Dantzig - 1963
169	Outline of an algorithm for integer solutions to linear programs - Gomory - 1958
31	Programmation mathématique : théorie et algorithmes. Dunod Editions - Minoux - 1983
16	Resource-aware programming - Taha - 2004
14	Postoptimal Analysis, Parametric Programming and Related Topics, Mac--Graw - Gal - 1986
8	Tardos: Sensitivity theorems in integer linear programming, Mathem. Programming 34 - Cook, Gerards, et al. - 1986
7	Integer Programming - Greenberg - 1971