Reliable Numerical ComputationM. G. Cox, S. J. Hammarling A collection of research papers based on talks given in honour of Jim Wilkinson, a pioneer of numerical analysis. These papers provide information on new techniques in numerical analysis, written by his colleagues and collaborators. |
Contents
The Lanczos algorithm for a pure imaginary Hermitian matrix | 25 |
Computational aspects of the Jordan canonical form | 57 |
Some aspects of generalized QR factorizations 73 | 73 |
Copyright | |
11 other sections not shown
Common terms and phrases
A₁ A₁₁ application approximation backward error BLAS Cholesky decomposition clustering complete pivoting complex condition number constraints convergence defective matrix defined Demmel denote diagonal Dongarra Dooren Duff efficiency eigenvalue problem eigenvectors EISPACK elements error analysis example floating-point FORTRAN function Gaussian elimination given Golub ill-conditioned implementation iterative refinement Jordan KKT system Lanczos algorithm least squares problems Lemma linear algebra linear equations linear systems LINPACK machine Math mathematical matrix memory method minimum-degree node non-singular non-zero null space numerical analysis numerically stable obtained optimization orthogonal P₁ Paige parallel parameters performed perturbation polynomial precision processors QR decomposition QR factorization quadratic programming R₁ rank residuals Ritz values rounding errors Schur complement Section semi-definite sequence SIAM single precision singular value solving sparse sparse matrix step structure symmetric techniques Theorem tion transformations U₁ updating variables vector Wilkinson zero