Bounds for probabilistic integer programming problems (Q1850114)

Let \(s\), \(m\), and \(n\) be positive integers, \(A\) be a real matrix of type \(m\times n\), \(T\) be an integer matrix of type \(s\times n\), \( b\) an \(m\)-dimensional real vector, \(c\) be an \(n\)-dimensional real vector, \(\xi\) be a random \(s\)-dimensional integer vector and \(p\in <0,1>\). The probabilistic integer programming seeks an \(n\)-dimensional nonnegative integer vector \(x\) such that \(\text{Prob}(Tx\geq\xi )\geq p\), \(Ax\geq b\) and \(x\) minimizes \(c^Tx\). The paper presents lower and upper bounds on the vector \(x\). The lower bound is based on the cone generation method. A simple illustrating example is given.