Saving Flops in LU Based Shift-and-Invert Strategy - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports (Research Report) Year : 2008

Saving Flops in LU Based Shift-and-Invert Strategy

Laura Grigori
Desire W. Nuentsa
  • Function : Author
  • PersonId : 849624
Hua Xiang
  • Function : Author
  • PersonId : 848046


The shift-and-invert method is very efficient in eigenvalue computations, in particular when interior eigenvalues are sought. This method involves solving linear systems of the form $(A-\sigma I)z=b$. The shift $\sigma$ is variable, hence when a direct method is used to solve the linear system, the LU factorization of $(A-\sigma I)$ needs to be computed for every shift change. We present two strategies that reduce the number of floating point operations performed in the LU factorization when the shift changes. Both methods perform first a preprocessing step that aims at eliminating parts of the matrix that are not affected by the diagonal change. This leads to $43\%$ and $50\%$ flops savings respectively.
Fichier principal
Vignette du fichier
RR-6553.pdf (273.62 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

inria-00286417 , version 1 (09-06-2008)
inria-00286417 , version 2 (10-06-2008)


  • HAL Id : inria-00286417 , version 2


Laura Grigori, Desire W. Nuentsa, Hua Xiang. Saving Flops in LU Based Shift-and-Invert Strategy. [Research Report] RR-6553, INRIA. 2008, pp.15. ⟨inria-00286417v2⟩
202 View
354 Download


Gmail Facebook X LinkedIn More