BALANCINGREGULARMATRIXPENCILS1-CiteSeer.PDFVIP

BALANCING REGULAR MATRIX PENCILS DAMIEN LEMONNIER AND PAUL VAN DOOREN Abstract. In this paper we present a new diagonal balancing technique for regular matrix pencils , which aims at reducing the sensitivity of the corresponding generalized eigenvalues. It is inspired from the balancing technique of a square matrix and has a comparable complexity. Upon convergence, the diagonally scaled pencil has row and column norms that are balanced in a precise sense. We also show that balancing a pencil boils down to making it closer to some standardized normal pencil. We give numerical examples illustrating that the sensitivity of generalized eigenvalues of a pencil may significantly improve after balancing. Key words. generalized eigenvalues, normal pencils, balancing AMS subject classifications. 15A18, 15A22, 65F15, 65F35 1. Introduction. A matrix with a norm that is several orders of magnitude larger than the modulus of its eigenvalues typically has eigenvalues that are sensitive to perturbations in the entries of . It is shown in [4] that the Frobenius norm of a matrix can then often be reduced via a diagonal scaling of the type 1. Such a scaling can be performed in exact arithmetic if the diagonal elements are constrained to be integer powers of the base of the finite precision arithmetic (typically 2 or 10). As a consequence the eigenvalues do not change, but their sensitivity can significantly be reduced. Such a diagonal scaling is therefore typically used before running any eigenvalue algorithm. In this paper we introduce a similar scaling method for square pencils with a determinant det( ) that is not identically zero for all values of . For such pencils – which are called regular – one can define generalized eigenvalues via the zeros of