Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Preconditioned iterations to calculate extreme eigenvalues

Conference ·
OSTI ID:223844
;  [1]
  1. Institut fuer Angewandte Mathematik, Leoben (Austria)

Common iterative algorithms to calculate a few extreme eigenvalues of a large, sparse matrix are Lanczos methods or power iterations. They converge at a rate proportional to the separation of the extreme eigenvalues from the rest of the spectrum. Appropriate preconditioning improves the separation of the eigenvalues. Davidson`s method and its generalizations exploit this fact. The authors examine a preconditioned iteration that resembles a truncated version of Davidson`s method with a different preconditioning strategy.

Research Organization:
Front Range Scientific Computations, Inc., Boulder, CO (United States); USDOE, Washington, DC (United States); National Science Foundation, Washington, DC (United States)
OSTI ID:
223844
Report Number(s):
CONF-9404305--Vol.1; ON: DE96005735
Country of Publication:
United States
Language:
English