Inequalities related to \(2\times 2\) block PPT matrices (Q2814829)

A positive semidefinite \(2\times2\) block matrix \(\begin{bmatrix} A&X\\X^\ast &B\end{bmatrix}\) is called PPT if it has a positive partial transpose meaning that also \(\begin{bmatrix} A&X^*\\X&B\end{bmatrix}\) is positive semidefinite (the star means Hermitian transpose and all blocks are \(n\times n\)). These play a role in quantum information theory. This paper gives all kinds of inequalities for such matrices like for the traces \(\mathrm{tr}(X^\ast X)\leq \mathrm{tr}(AB)\), or for any unitary invariant norm \(2\|X\|\leq\|A+B\|\), or for singular values NEWLINE\[NEWLINE\prod_{j=1}^k s_j(X)\leq\prod_{j=1}^k s_j(A^{1/2}B^{1/2}), \quad k=1,\dots,nNEWLINE\]NEWLINE and more. The Hua matrix NEWLINE\[NEWLINE\begin{bmatrix} (I-A^\ast A)^{-1} & (I-B^\ast A)^{-1} \\ (I-A^\ast B)^{-1} & (I-B^\ast B)^{-1} \end{bmatrix}NEWLINE\]NEWLINE is PPT if \(A\) and \(B\) are contractive. This matrix gets some special attention in this paper.