[SciPy-User] Problem with scipy.special.chebyt

Ivo Maljevic ivo.maljevic at gmail.com
Sun Aug 30 14:16:40 EDT 2009


Pauli, Whatever the case is, I would suggest you to consider using
scipy.linalg eigenvalue function inside scipy.special. I see no point
in using numpy.dual version of it. They are clearly differently implemented,
and scipy's eig function solves the problem even in these distributions, and
moves the problem from scipy to numpy.
Ivo

2009/8/30 Pauli Virtanen <pav at iki.fi>

> On 2009-08-30, Ivo Maljevic <ivo.maljevic at gmail.com> wrote:
> [clip]
> > install with both installations. I don't know why nobody from
> > scipy wants to try a fresh install on a virtual machine maybe,
> > as I am sure you would be able to see the problem right away. I
> > agree that this is most likely not numpy/scipy problem, but it
> > would still be helpful for other users out there if somebody
> > actually tried this with a clean distro.
>
> Sorry, but I don't have enough time to spend on this at the
> moment -- maybe later on. Tracking down this kind of
> platform-specific issues often means hours of gruelling work.
>
> Feel free to file a bug ticket to Numpy's Trac so that this won't
> be forgotten. Also, I'd suggest letting the Mandriva/Opensuse
> people know there may be something wrong in their packages.
>
> > In my previous email I specified that using the eig function
> > from scipy.linalg solves the problem, and using the eig from
> > numpy.dual brings it back.
> [clip]
>
> I mistyped scipy.linalg where I meant numpy.linalg. As David
> said, numpy and scipy may possibly use different lower-level
> libraries, which explains the difference.
>
> All this seems to point out that something is wrong on Mandriva
> and Opensuse with how Numpy links to the underlying lapack
> library.
>
> --
> Pauli Virtanen
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User at scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20090830/98b75a9f/attachment.html>


More information about the SciPy-User mailing list