Calculating some inverse linear programming problems

From MaRDI portal





For the general linear programming problem \[ \min \{cx/Ax = b,\;x \geq 0\} \] the inverse problem is expressed as \[ \min \bigl\{ |c-\widetilde c|/ \widetilde c\in F(x^0) \bigr\} \] where \(c,x \in \mathbb{R}^n\), \(b \in \mathbb{R}^m\), \(A\) is an \(m \times n\) matrix, \(x^0\) is a feasible solution, \[ F(x^0) = \bigl\{\widetilde c \in \mathbb{R}^n/ \min \{\widetilde cx/Ax=b,\;x\geq 0\} = \widetilde cx^0 \bigr\} \] and \(|\dots|\) is the \(l_1\) norm. A method for solving this inverse problem is suggested which is based on the optimality conditions for linear programming problems. For the application of the given method to inverse minimum cost flow or assignment problems is found that this method yields strongly polynomial algorithms.




Cited in
(75)








This page was built for publication: Calculating some inverse linear programming problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1923457)