[Scipy-tickets] [SciPy] #1159: Use ARPACK to implement eigs? (wrong eigenvalues generated by linalg.eigh/DSYEVR) (was: Wrong eigenvalues generated by scipy.linalg.eigh)

SciPy Trac scipy-tickets@scipy....
Thu Apr 22 11:20:08 CDT 2010


#1159: Use ARPACK to implement eigs? (wrong eigenvalues generated by
linalg.eigh/DSYEVR)
------------------------------------------------+---------------------------
 Reporter:  ysr                                 |       Owner:  somebody
     Type:  defect                              |      Status:  reopened
 Priority:  normal                              |   Milestone:  0.8.0   
Component:  scipy.linalg                        |     Version:  0.7.0   
 Keywords:  eigenvalue, hermitian, eigenvector  |  
------------------------------------------------+---------------------------
Changes (by pv):

  * status:  needs_decision => reopened


Comment:

 ARPACK is available as `scipy.sparse.linalg.eigen.arpack.eigen` (in Scipy
 0.7.1), but not used by the `eigs`-type routines at the moment. Perhaps it
 should be, at least optionally, since ARPACK is probably better tested as
 computing a few eigenvalues is more or less its main purpose.

 (However, for some reason I often get exceptions when calling the ARPACK
 -- probably needs a separate bug report for that.)

 I think it's unlikely at the moment that there will be any change in 0.7.2
 this late in the release schedule. 0.8 is a more realistic target for
 this...

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1159#comment:5>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list