Inequalities for trace norms of \(2\times 2\) block matrices. (Q1423652): Difference between revisions

Let \[ M=\begin{pmatrix} X&Y\\ Y^*&Z\end{pmatrix} \] be a \(2n\times 2n\) positive semi-definite matrix. The \(p\)-norm of a matrix \(A\) is defined as \(\| A\| _p=\left(\text{Tr}(A^*A)^{p/2}\right)^{1/p}\). The \(2\times 2\) positive semi-definite matrix \(m\) is defined as \[ m=\begin{pmatrix} \| X\| _p&\| Y\| _p\\ \| Y\| _p&\| Z\| _p\end{pmatrix}. \] Then the main theorem of this paper states that \[ \begin{aligned} \| M\| _p&\geq \| m\| _p\qquad \text{for }1\leq p\leq 2,\\ \| M\| _p&\leq \| m\| _p\qquad \text{for }2\leq p\leq \infty.\end{aligned} \] As the inequality had been known for integer values of \(p\), the main contribution here is the extension to all values \(p\geq1\). A weaker inequality which applies also to non-positive matrices is presented. As an application in quantum information theory, the inequality is used to obtain some results concerning \(p\)-norms of product channels.