A Schur-Horn theorem for symplectic eigenvalues (Q2176130)

The authors present a symplectic version of the famous Schur-Horn theorem about the relationship between diagonal entries and the eigenvalues of a \(n \times n\) Hermitian matrix \(A\). The new result reads as follows: Theorem. Let \(d_s(A)\) be a \(n\)-vector of symplectic eigenvalues of \(A\), \(\Delta(A)\) be the diagonal of \(A\) and \(\Delta_s(A) = d_s(\Delta(A))\). Then for any \(2n \times 2n\) positive definite matrix \(A\) one has the weak majorisation \[ \Delta_s(A) \prec^w d_s(A). \] Conversely, if \(x\), \(y\) are two \(n\)-vectors with positive entries such that \(x \prec^w y\), then there exists a \(2n \times 2n\) positive definite \(A\) such that \(x =\Delta_s(A)\) and \(y=d_s(A)\). Here, by saying that \(x\) is weakly supermajorised by \(y\), i.e., \(x \prec^w y\), one means that \[ \sum\limits_{j=1}^k x_j^\uparrow \geq \sum\limits_{j=1}^k y_j^\uparrow \,, {{\mbox { for }}} 1\leq k \leq n\,, \] with coordinates of \(x\) and \(y\) rearranged in increasing order, i.e., \(x_1^\uparrow \leq \cdots x_n^\uparrow\) and \(y_1^\uparrow \leq \cdots y_n^\uparrow\).