Singular values, doubly stochastic matrices, and applications (Q1893086): Difference between revisions

For a complex \(n\times n\) matrix \(A\) there exist doubly stochastic \(n\times n\) matrices \(A_1\), \(A_2\) such that \(\sigma_n(A)^2 A_1\leq A^{O^2}\leq \sigma_1(A)^2 A_2\), where \(\sigma_1(A)\geq \cdots\geq \sigma_n(A)\geq 0\) are the singular values of \(A\), \(A^{O^2}= [a_{jk}\overline a_{jk}]\) (the Hadamard product) and the inequalities are taken entrywise. Several applications of this are given. For a simple graph \(G\) it yields a sufficient condition that \(G\) contain an even cycle. Another result obtained from it is an inequality that can be considered as a generalization of Schur's inequality for nonnegative matrices. A third application concerns the perturbation of eigenvalues (a generalization of the Hoffman-Wielandt inequality).