zbMATH — the first resource for mathematics

A practical procedure for computing eigenvalues of large sparse nonsymmetric matrices. (English) Zbl 0605.65027
Large scale eigenvalue problems, Proc. IBM Eur. Inst. Workshop, Oberlech/Austria 1985, North-Holland Math. Stud. 127, 193-240 (1986).
[For the entire collection see Zbl 0595.00022.]
We propose a Lanczos procedure with no reorthogonalization for computing eigenvalues of very large nonsymmetric matrices. (This procedure can also be used to compute corresponding eigenvectors but that issue will be dealt with in a separate paper.) Such computations are for example, central to transient stability analyses of electrical power systems and for determining parameters for iterative schemes for the numerical solution of partial differential equations. Numerical results for several large matrices are presented to demonstrate the effectiveness of this procedure.

65F15 Numerical computation of eigenvalues and eigenvectors of matrices
65F50 Computational methods for sparse matrices