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

[Ivo Maljevic]

Well, I've tried, and this is the best I could come up with.
Inside gen_roots_and_weights() function I added few lines for debugging:

    nn = np.arange(1.0,n)
    sqrt_bn = sqrt_bn_func(nn)
    an = an_func(np.concatenate(([0], nn)))

        print 'Step A (before eig)'
    print 'mat=', np.diagflat(an) +np.diagflat(sqrt_bn,1) +
np.diagflat(sqrt_bn,-1)
    print 'Step B (before eig)'

    x, v = eig((np.diagflat(an) +
                np.diagflat(sqrt_bn,1) +
                np.diagflat(sqrt_bn,-1)))

    print 'Step C (after eig)'

and step B is the last point where I get before the program breaks. But, if
I load the same matrix (result between A and B), eig() doesn't break.
Anyway, since this works on other distributions, I guess it is not scipy
error. You are probably correct to guess that it's the eignevalue function,
even though I'm not sure why doesn't it break consistently.

Ivo



2009/8/28 Pauli Virtanen <pav at iki.fi>

> Fri, 28 Aug 2009 07:15:49 -0400, Ivo Maljevic kirjoitti:
> >>>> import scipy.special
> >>>> scipy.special.chebyt(12)(-0.5)
> > *** glibc detected *** python: malloc(): memory corruption:
> > 0x0000000000afdf40 ***
>
> This particular glibc error does not necessarily mean that the problem is
> in chebyt -- glibc just notices the problem at this point, but the
> corruption may have occurred earlier. Can you try running eig() etc.
> multiple times?
>
> The point is that scipy.special.chebyt is mostly pure-python code, which
> calls, among others, eig() and special.gamma(). It is very likely the
> error is in these lower-level routines. It'd be helpful if you were able
> to find out which routine bugs out.
>
> --
> 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/20090828/e0bb926a/attachment.html>