[Trilinos-Users] Anasazi: LOBPCG algorithm

[홍광우]

Dear Trilinos developers and users,


I use Anasazi package, and I have a question about sparse matrix diagonalization algorithm, especially LOBPCG method.


I don't know about the detailed LOBPCG algorithm, so I have the following question.


For a given sparse matrix, Anasazi can compute the eigenvectors and eigenvalues using only 1 processor. However, with multiple processors, Anasazi cannot compute the eigenproblem. Is it general case using LOBPCG method?


Also, do "the number of iterations" and "the block size" have to be modified depending on the number of processors? For example, should the block size be increased when I use multiple processors compared to single processor?


Thank you for your concern.


Sincerely,


Kwangwoo




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://software.sandia.gov/pipermail/trilinos-users/attachments/20140519/9198db7e/attachment.html>