Determinant and eigenvalue inequalities involving nonnegative matrices (Q6086436): Difference between revisions

The authors give some algebraic (symmetric) inequalities for \(n\) nonnegative real numbers. These are reformulated as inequalities for some nonnegative matrices with nonnegative eigenvalues \(\lambda_1,\ldots,\lambda_n\) arranged in increasing order. One of the main inequalities states that for \(n\) positive real numbers \(x_1,\ldots,x_n\) we have: \[ \left(\sum_{i=1}^nx_i\right)^2\le n\left(\prod_{i=1}^nx_i\right)^{\frac{2}{n}}+(n-1)\sum_{i=1}^nx_i^2. \] In addition, other bounds for the spectral radius of particular matrices are proved. The authors mainly use properties of positive linear maps along with the variational principle of eigenvalues for \(n\times n\) Hermitian matrices, that is: given \(A\ge B\), we have \(\lambda_i(A)\ge \lambda_i(B)\) for every \(i\), \(i=1,\ldots,n\). The paper is nicely presented and several examples support the results.