Lower eigenvalue bounds for pencils of matrices (Q1088752): Difference between revisions

Under some simple assumptions the author gives upper and lower bounds for the eigenvalues of the matrix pencil A-\(\nu\) B where A is a Stieltjes matrix (i.e. a positive definite matrix with nonpositive off-diagonal entries) and \(B=A+C=LP^{-1}U\) with U upper triangular, \(L=U^ T\), \(P=diag(U)\) and \(C\neq 0\). While upper bounds of the eigenvalues can be determined by closed formulae, lower bounds must be constructed in terms of some auxiliary pencils whose proper choice substantially influences the degree of accuracy in practical situations. Examples of such constructions are also given. The results are related to convergence properties of the iterative scheme \(Bx_{m+1}=Cx_ m+b\).