In this paper, we present new error bounds for the Lanczos method and the shift-and-invert Lanczos method for computing e−τAv for a large sparse symmetric positive semidefinite matrix A. Compared with the existing error analysis for these methods, our bounds relate the convergence to the condition numbers of the matrix that generates the Krylov subspace. In particular, we show that the Lanczos method will converge rapidly if the matrix A is well-conditioned, regardless of what the norm of τA is. Numerical examples are given to demonstrate the theoretical bounds.

Document Type


Publication Date


Notes/Citation Information

Published in SIAM Journal on Numerical Analysis, v. 51, no. 1, p. 68-87.

© 2013, Society for Industrial and Applied Mathematics. Unauthorized reproduction of this article is prohibited.

The copyright holder has granted permission for posting the article here.

Digital Object Identifier (DOI)


Funding Information

This research was supported in part by NSF grant DMS-0915062.

Included in

Mathematics Commons