[Scipy-tickets] [SciPy] #1612: Loss of precision in scipy.linalg.eig().

SciPy Trac scipy-tickets@scipy....
Wed Feb 29 10:40:58 CST 2012


#1612: Loss of precision in scipy.linalg.eig().
-----------------------+----------------------------------------------------
  Reporter:  cgranade  |       Owner:  somebody   
      Type:  defect    |      Status:  closed     
  Priority:  normal    |   Milestone:  Unscheduled
 Component:  Other     |     Version:  0.10.0     
Resolution:  invalid   |    Keywords:             
-----------------------+----------------------------------------------------
Changes (by charris):

  * status:  new => closed
  * resolution:  => invalid


Comment:

 Lapack can be compiled for quad precision using the right flags and the
 Intel compiler, but C support for quad precision is lacking and hardware
 support is rare. What you are calling quad precision is actually IEEE
 extended precision (80 bits) on most architectures. In short, it would be
 nice to support true quad precision at some point, but currently it would
 have to be limited to special builds for special architectures and
 wouldn't be portable. If you google 'lapack quad precision' you will get
 some idea about what is happening in that area.

 I'm going to close this ticked as invalid, but I feel your pain ;)

-- 
Ticket URL: <http://projects.scipy.org/scipy/ticket/1612#comment:1>
SciPy <http://www.scipy.org>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list