Some inequalities for sums of matrices

From MaRDI portal





From a generalized matrix version of the Cauchy-Schwarz and Fouch-Kantorovich inequalities NEWLINE\[NEWLINE (X^{\ast}AX)^{-1}\leq X^{\ast}A^{-1}X \leq \frac{(\lambda_1+ \lambda_2)^2}{4\lambda_1\lambda_2}(X^{\ast}AX)^{-1} NEWLINE\]NEWLINE where \(A\) is an \(n\times n\) Hermitian positive definite matrix with eigenvalues \(\lambda_1 \geq \lambda_2\geq \cdots\lambda_n>0\), and the \(n\times p\) complex matrix \(X\) and its conjugate transpose \(X^{\ast}\) satisfy \(X^{\ast}X=I_p\), some inequalities are given involving the Moore-Penrose inverse. In the above inequality the Löwner partial ordering \(M\geq N\) means \(M-N\) is Hermitian nonnegative definite. NEWLINENEWLINENEWLINERelated to the first inequality a result is given for sums of \(k\) Hermitian nonnegative definite matrices of order \(n\) and \(k\) complex matrices of size \(n\times p\). In addition, the equality is characterized. Finally, another inequality corresponding to the right hand side of the above inequality is obtained for the sum of \(k\) Moore-Penrose matrices of nonnull Hermitian nonnegative definite matrices.











This page was built for publication: Some inequalities for sums of matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2758096)