Entrywise perturbation theory for diagonally dominant M-matrices with applications (Q1348917): Difference between revisions

Consider a weakly diagonally dominant M-matrix \(A= [a_{ij}]\in \mathbb{R}^{n\times n}\) and set \(d_i:= a_{ii}- \sum_{j\neq i}|a_{ij}|\). Let \(\widetilde A= [\widetilde a_{ij}]\in \mathbb{R}^{n\times n}\) be another weakly diagonally dominant M-matrix with \(|a_{ij}- \widetilde a_{ij}|\leq \varepsilon|a_{ij}|\), \(i\neq j\), and \(|d_i- \widetilde d_i|\leq \varepsilon|d_i|\), where \(\varepsilon> 0\) is a small parameter. It is shown that \[ (1-\varepsilon)^n(1+ \varepsilon)^{1-n} A^{-1}\preceq\widetilde A^{-1}\preceq (1+\varepsilon)^n (1-\varepsilon)^{1- n}A^{-1} \] and \[ (1- \varepsilon)^{n-1}(1+ \varepsilon)^{-n} \lambda\leq \widetilde\lambda\leq (1- \varepsilon)^{n-1}(1- \varepsilon)^{-n} \lambda, \] where \(\preceq\) is the component-wise order relation in \(\mathbb{R}^{n\times n}\) and \(\lambda\) is the smallest eigenvalue of \(A\). Applications of these results to the theory of digital electrical circuits are considered.