[SciPy-user] Problems testing SciPy 0.3.2

Pearu Peterson pearu at scipy.org
Fri Oct 15 06:44:17 CDT 2004

On Fri, 15 Oct 2004, Francesc Alted wrote:

> I've just installed the newest SciPy 0.3.2, and all tests seems to pass
> well, except 15 (out of 574). All of them are caused by the same reason,
> which ultimately is:
> File "/users/exp/alted/bin-i686/lib/python/scipy/special/orthogonal.py", line 91, in get_eig_func
>    eig = scipy.linalg.eig
> AttributeError: 'module' object has no attribute 'eig'
> I'm attaching the information required in the INSTALL.txt for reporting bugs.

Thanks for the information. The reason for this problem should be looked 
at your note:
[alted at api linalg]$      python -c 'import atlas_version'
Traceback (most recent call last):
   File "<string>", line 1, in ?
ImportError: ./atlas_version.so: undefined symbol: ATL_buildinfo

This does not seem to work, but my ATLAS version is 3.6.0. I've also read
http://math-atlas.sourceforge.net/errata.html#completelp in order to get
complete LAPACK support.

Where your ATLAS-3.6.0 libraries are located? Did you use ATLAS 
environment variable to specify the location?

scipy_distutils seems to  find only ATLAS-3.2.1 or ATLAS-pre3.3.6 
libraries in  /users/exp/alted/bin-i686/lib/atlas/.
The fact that using ATLAS-3.2.1_pre3.3.6 libraries fail with the 
"undefined symbol: ATL_buildinfo" message indicates some bug in scipy (I 
can reproduce it, so I'll fix it).


More information about the SciPy-user mailing list