On c-norms and c-antinorms on cones (Q755849): Difference between revisions

A norm \(\Phi\) on \(R^{m\times n}\) is a c-norm if there exist norms \(\theta\) and \(\phi\) on \(R^ m\) and \(R^ n\), respectively, such that \(\Phi (XY^ T)=\theta (X)\phi (Y)\) for all \(X\in R^ m\), \(Y\in R^ n\). Let \({\mathcal H}\subset R^ m\) and \({\mathcal K}\subset R^ n\) be proper cones and \(A\in \pi ({\mathcal K},{\mathcal H})\), i.e., \(A\in R^{m\times n}\) with A\({\mathcal K}\subset {\mathcal H}\). An antinorm \(\Phi\) on \(\pi\) (\({\mathcal K},{\mathcal H})\) is a c-antinorm if there exist antinorms \(\theta\) and \(\phi\) on \({\mathcal H}\) and \({\mathcal K}^*\) (the dual cone of \({\mathcal K})\), respectively, such that \(\Phi (XY^ T)=\theta (X)\phi (Y)\) for all \(X\in {\mathcal H}\), \(Y\in {\mathcal K}^*\), where an antinorm is a real-valued function which is positive, homogeneous, superadditive, and continuous. The author discusses the properties of c-antinorms on cones and discovers that many properties of a c-antinorm on a cone turn out to be analogous to well known properties of a c-norm on the whole space.