On Fri, Apr 2, 2010 at 3:39 PM, Ondrej Certik <[email protected]> wrote:
> On Fri, Apr 2, 2010 at 3:35 PM, Ondrej Certik <[email protected]> wrote:
>> On Fri, Apr 2, 2010 at 3:00 PM, Mateusz Paprocki <[email protected]> wrote:
>>> Hi,
>>>
>>> On Fri, Mar 19, 2010 at 08:11:21PM -0700, Ondrej Certik wrote:
>>>> On Fri, Mar 19, 2010 at 7:54 PM, Ted Horst <[email protected]> wrote:
>>>> >
>>>> > On 2010-03-17, at 21:28, Ondrej Certik wrote:
>>>> >
>>>> >> Hi,
>>>> >>
>>>> >> we just fixed the last bugs in the Mateusz's polys branch and pushed
>>>> >> it in our main git repo. All tests pass for all kinds of
>>>> >> configurations that I tried, but if possible, please test it (I didn't
>>>> >> test on Mac nor Windows).
>>>> >>
>>>> >> Just to get some idea, Mateusz did about 150 patches and it implements
>>>> >> all kinds of polynomial algorithms, see e.g. this patch for some
>>>> >> details:
>>>> >>
>>>> >>
>>>> >> http://git.sympy.org/?p=sympy.git;a=commit;h=31c9aa43ffc7bc031493fc4ea27b74dc50c4d035
>>>> >>
>>>> >> It should fix all kinds of bugs and also it's quite fast and robust.
>>>> >> E.g. this failed:
>>>> >>
>>>> >> In [1]: f = expand((x+y+z)**20)
>>>> >>
>>>> >> In [2]: factor(f)
>>>> >> Out[2]:
>>>> >> 20
>>>> >> (x + y + z)
>>>> >>
>>>> >> and now it works.
>>>> >>
>>>> >> Plus the core polynomials are cythonized in pure Cython mode (so far
>>>> >> only about 2x speedup), but I think we'll figure out ways to speed it
>>>> >> up more using Cython.
>>>> >>
>>>> >> Ondrej
>>>> >>
>>>> >
>>>> > I was excited to see the first cython bits in the main repo, but I am
>>>> > having
>>>> > a few problems.
>>>> >
>>>> > I'm on a ppc mac running 10.5.8 and using the system python 2.5.1 and the
>>>> > latest head of cython (12507ec6fa91) from
>>>> > http://hg.cython.org/cython-devel.
>>>> >
>>>> > First, it looks like lambda is not supported by this version of cython
>>>> > (error below). Are you using the cython-closure branch?
>>>> >
>>>> > Second, it looks like the cythonized decorator is not having the desired
>>>> > effect. I'm guessing this is because the cython.locals decorator has to
>>>> > actually be seen by cython (the decorator itself doesn't do anything).
>>>> > Could this be done using a pxd file?
>>>>
>>>> I just tried the same Cython head and I am getting the exact same
>>>> cython error as you do.
>>>>
>>>> Mateusz, how should we get this working? Last time I tried it was at
>>>> the scipy conference and it worked, but it was just some prototype.
>>>>
>>>
>>> this was like half a year ago ;) I fixed this issue in polys8 branch
>>> (http://github.com/mattpap/sympy-polys/commits/polys8). The problem
>>> was mainly with lambda, which still isn't supported, and with
>>> decomposition of tuples in arguments to functions.
>>
>> Indeed, that fixes it for me. Can I push your branch in?
>
>
> I got one failure in documentation:
>
> **********************************************************************
> File "/home/ondrej/repos/sympy/doc/src/modules/polynomials.txt", line
> 216, in polynomials.txt
> Failed example:
> solve_poly_system([y**2 - x**3 + 1, y*x], x, y)
> Expected:
> ___ ___
> I*\/ 3 I*\/ 3
> [(0, -I), (1, 0), (-1/2 - -------, 0), (-1/2 + -------, 0), (0, I)]
> 2 2
> Got:
> ___ ___
> I*\/ 3 I*\/ 3
> [(0, I), (0, -I), (1, 0), (-1/2 + -------, 0), (-1/2 - -------, 0)]
> 2 2
> **********************************************************************
>
>
> Otherwise it's ok.
Except, that for some reason, Cython isn't able to cythonize this:
for i in range(0, n)
into a pure C loop, even though you have added there the decorator:
@cythonized("n,m,i")
so we need to look into that.
Ondrej
Ondrej
--
You received this message because you are subscribed to the Google Groups
"sympy" group.
To post to this group, send email to [email protected].
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.
