2009/8/12 Fabian Pedregosa <fab...@fseoane.net>
>
> Vinzent Steinberg wrote:
> >
> > Some questions:
> >
> > You did remove some tests. Are these things tested elsewhere? Things like
> >
> > >>> ask(2*n+1, Q.odd, Assume(n, Q.integer))
> > True
>
> I think that's in sympy/queries/tests/test_query.py line 797
>
> >
> > You seem to break (2*n).is_even, because it's no longer possible to tie
> > to n that it's an integer. So I would rather remove it altogether.
>
> You can still tie integer to n, it's just that method .is_even is
> missing, although not removed (it does not raise an exception when you
> try to access it) because in file sympy/core/assumptions.py zero is
> defined as:
>
> 'zero -> infinitesimal & even',
>
> but removing that line makes breaks a lot of things.
>
> I wonder if it's better to remove the old system altogether (which will
> be a huge chunk of code) or in several steps.
I would not remove it as long as the new system can replace it without
regressions in performance or functionality. I also would not remove it
before 0.7.
>
>
> >
> > Also performance got much worse, you can even watch the difference
> > between the old
> >
> > >>> n = Symbol('n', integer=True)
> > >>> (2*n).is_even
> > True
> >
> > and the new
> >
> > >>> ask(2*n, Q.even, Assume(n, Q.integer))
> > True
> >
> > But that's not a problem, because I think you did not yet look for
> > performance. It's better to have the functionality first, but we will
> > need to care about the performance, before releasing at the latest.
>
>
> Yes, current bottleneck is the dpll algorithm, which I call naively
> couple of times in ask(). I have some notes on how to improve speed in
> doc/src/modules/queries.txt.
>
>
> Also, using integers for internal data representation which would not be
> too difficult to implement (in logic.algorithms.dpll) should make it a
> lot faster, see
>
> http://code.google.com/p/sympy/issues/detail?id=1545
This sounds good! Looking forward.
Vinzent
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sympy-patches" group.
To post to this group, send email to sympy-patches@googlegroups.com
To unsubscribe from this group, send email to
sympy-patches+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sympy-patches?hl=en
-~----------~----~----~----~------~----~------~--~---
