On Sun, Oct 19, 2008 at 12:22 PM, Fredrik Johansson
<[EMAIL PROTECTED]> wrote:
>
> I found the problem. In evalf.py line 626
>
> re, im, re_acc, im_acc = evalf(func, prec+15, {'subs':{x:t}})
>
> should be
>
> re, im, re_acc, im_acc = evalf(func, mp.prec, {'subs':{x:t}})
>
> It needs to be mp.prec for oscillatory quadrature because mpmath
> significantly increases the precision when quadosc is called.
>
> This has been broken in SymPy all along. It only worked in the
> previous tests by accident. The change of quadrature algorithm in
> mpmath 0.10 exposed the problem.
Thanks, I fixed everything and sent patches to sympy-patches. Could
you please review them? Alternatively, you can get them from here:
http://github.com/certik/sympy/tree/mpmath5
In fact, I'd appreciate if you could try that on windows, here I wrote how:
http://wiki.sympy.org/wiki/Git_hg_rosetta_stone#how_to_checkout_remote_branch
i.e. instead of using
git checkout mpmath5
use:
git checkout -b mpmath5 origin/mpmath5
and that should do the trick. I checked it on windows using:
http://code.google.com/p/msysgit/
and it works for me.
>
> It's good to pay attention to performance regressions; quite often
> there are real bugs hiding beneath...
Exactly, my words.
Ondrej
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sympy" group.
To post to this group, send email to sympy@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sympy?hl=en
-~----------~----~----~----~------~----~------~--~---
