LAPACK Archives

[Lapack] What's the best lapack routine to use?

I agree with Julien's advice. But you should know that if you
have a zero or very tiny eigenvalue, roundoff may cause these
routines to return a (tiny) negative eigenvalue. The easiest
thing to do is simply replace these negative values by zero
(as long as you are certain your input matrix is positive
semidefinite).
Jim Demmel

On 1/15/15, 8:08 AM, Langou, Julien wrote:
Do you want all the eigenvalues or just a  few of them?
Is the matrix dense or is the matrix sparse?

If you have a dense symmetric matrix and want all of its eigenvalue then
you can use LAPACK subroutines: DSYEV, DSYEVD, DSYEVR, or DSYEVX.
Which is the best?
DSYEV is fine enough if you want only the eigenvalues (and not the
eigenvectors).
DSYEVR or DSYEVD are good for eigenvector computation.
Which one is better kind of depends on the platform and the problem at
hand. Both are good enough.

Cheers,
Julien.


On 1/14/15, 6:38 PM, "Kevin Cahill" <kevinecahill@Domain.Removed> wrote:

What's the best lapack routine to use to find the eigenvalues of a huge
real symmetric matrix whose eigenvalues are positive or zero?

Best wishes,
Kevin

_______________________________________________
Lapack mailing list
Lapack@Domain.Removed
http://lists.eecs.utk.edu/mailman/listinfo/lapack


<Prev in Thread] Current Thread [Next in Thread>


For additional information you may use the LAPACK/ScaLAPACK Forum.
Or one of the mailing lists, or