scientific article; zbMATH DE number 1145309

From MaRDI portal
Publication:4384602





Several matrix inequalities are announced here. For \(n\times n\) \(A\) and \(B\), \(A\circ B\) denotes the Hadamard product, and \(A\geq B\) means \(A-B\) is positive semidefinite. Theorem 1: Let \(A,B\) be \(n\times n\) positive definite and \(X,Y\) arbitrary \(n \times n\) matrices. Then \((X^*A^{-1} X)\circ (Y^*B^{-1} Y)\geq (X\circ Y)^* (A\circ B)^{-1} (X\circ Y)\) and \(X^*A^{-1} X+Y^* B^{-1} Y\geq (X+Y)^* (A+B)^{-1} (X+Y)\). This generalizes two inequalities of Berenstein and Veinstein for vectors \(x\) and \(y\) to matrices \(X\) and \(Y\). Theorem 3: Let \(\| \cdot \|\) be a unitarily invariant matrix norm. For all \(m\times n\) \(X\) and positive definite \(n\times n\) \(A\) and \(m\times m\) \(B\) such that \(A\geq I\) and \(B\geq I\), the function \(f(p)= \| A^pX +XB^p \|\) is nondecreasing on \([0,\infty)\).




Cited in
(23)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4384602)