NASA Logo

NTRS

NTRS - NASA Technical Reports Server

Back to Results
On polynomial preconditioning for indefinite Hermitian matricesThe minimal residual method is studied combined with polynomial preconditioning for solving large linear systems (Ax = b) with indefinite Hermitian coefficient matrices (A). The standard approach for choosing the polynomial preconditioners leads to preconditioned systems which are positive definite. Here, a different strategy is studied which leaves the preconditioned coefficient matrix indefinite. More precisely, the polynomial preconditioner is designed to cluster the positive, resp. negative eigenvalues of A around 1, resp. around some negative constant. In particular, it is shown that such indefinite polynomial preconditioners can be obtained as the optimal solutions of a certain two parameter family of Chebyshev approximation problems. Some basic results are established for these approximation problems and a Remez type algorithm is sketched for their numerical solution. The problem of selecting the parameters such that the resulting indefinite polynomial preconditioners speeds up the convergence of minimal residual method optimally is also addressed. An approach is proposed based on the concept of asymptotic convergence factors. Finally, some numerical examples of indefinite polynomial preconditioners are given.
Document ID
19910023500
Acquisition Source
Legacy CDMS
Document Type
Contractor Report (CR)
Authors
Freund, Roland W.
(Wuerzburg Univ. Germany, F.R. , United States)
Date Acquired
September 6, 2013
Publication Date
August 1, 1989
Subject Category
Computer Programming And Software
Report/Patent Number
RIACS-TR-89-32
NASA-CR-188857
NAS 1.26:188857
Report Number: RIACS-TR-89-32
Report Number: NASA-CR-188857
Report Number: NAS 1.26:188857
Accession Number
91N32814
Funding Number(s)
CONTRACT_GRANT: NCC2-387
Distribution Limits
Public
Copyright
Work of the US Gov. Public Use Permitted.
No Preview Available