Minimum deviation, quasi-LU factorization of nonsingular matrices

From MaRDI portal
Publication:2383020





The authors are concerned with the following problem. If the matrix \(A\) is nonsingular and has no \(LU\) factorization, how may \(A\) be factored into \(\widetilde L\) and \(\widetilde U\) that are ``nearly lower and upper triangular, respectively? They give an explicit measure of nearness to a matrix being triangular, and they introduce a measure, the deviation \(\text{dev}(\widetilde L,\widetilde U)\) of the quasi-\(LU\) factorization \(A=\widetilde L\widetilde U\) from an \(LU\) factorization. It is known that \(A\) can be written as \(LPU\), where \(L\) is unit lower triangular, \(P\) is a permutation matrix, and \(U\) is upper triangular. One obtains a quasi-\(LU\) factorization of \(A\) by putting \(\widetilde L=LP\) and \(\widetilde U=U\). The authors calculate \(\text{dev}(\widetilde L,\widetilde U)\) and establish that the deviation of any quasi-\(LU\) factorization cannot be better than the one mentioned above.











This page was built for publication: Minimum deviation, quasi-LU factorization of nonsingular matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2383020)