Dear Scalapack gurus,
I would like to compute all eigenvalues of a large nonsymmetric matrix (perhaps
1 million x 1 million). The matrix is dense but block offdiagonal:
0 A
M = where A and B are full.
B 0
As I understand, in Scalapack I have to first use PDGEHRD to reduce the matrix
to upper Hessenberg form and then PDLAHQR to compute the eigenvalues.
But these subroutines cannot take advantage of the block structure, can they?
Is there another possibility how to compute the eigenvalues efficiently?
Thank you for your answers, Tomas Dohnal.
 next part 
An HTML attachment was scrubbed...
URL:
http://lists.cs.utk.edu/private/scalapack/attachments/20060715/eecfeea7/attachment.htm
