Abstract
Uncertainty is a common phenomenon in practice. Due to measurement errors we can hardly expect precise values in real-life linear programming problems.
In this paper, we suppose that we are given lower and upper bounds on the inexact quantities, and the quantities may perturb independently and simultaneously within these bounds. In this model we investigate the problems of optimal value range, basis stability, optimal solutions enclosures, duality etc.
Complexity issues are discussed, too; some tasks are polynomially solvable while another are NP-hard. This approach is more general and powerful than the standard sensitivity analysis.