Hi Florent,
On Tue, 22 Nov 2011 09:04:13 +0100
Florent Hivert <florent.hiv...@lri.fr> wrote:
> Follow up from thread "indexed symbolic variables".
>
> > > > Do we have to call Maxima to compute the numerator of a rational
> > > > fraction ? Is seems to me that Pynac/Ginac should be able to do
> > > > it by themselves.
> > > >
> > > Unfortunately, we still call maxima for a lot of trivial
> > > operations.
> >
> > This is now #12068
>
> To speedup some computation I'm trying to wrap Ginac numer and denom
> method. Thanks to Cython, this is very easy. However, Maxima and
> Ginac have a different semantic for those two functions:
>
> - Maxima: Does not change the expression; if the expression is not a
> quotient, then this will return the expression itself.
> - Ginac: try to normalize the expression to put it in the form N/D
> and then return N;
>
> For example, with an expression such as "x + y/(x + 2)",
>
> - Maxima: numer = x + y/(x + 2), denom = 1
> - Ginac: numer = (x^2 + 2x + y), denom = x + 2
>
> I can more or less emulate Maxima's behavior with Ginac, but I'm not
> sure what's best to do. I see several options:
>
> 1 - Forget about Maxima and follows Ginac semantic;
> 2 - Follows Maxima semantic as close as possible using Ginac;
> 3 - Have a parameter 'algorithm' to select the needed one. Depending
> one the algorithm the semantic differ;
> 4 - Have a parameter 'normal' to select if we normalize or nor the
> result before computing the numerator, always use Ginac however.
> 5 - 3 and 4, have two parameters, one for selecting Maxima or Ginac,
> one for performing a normalization before.
> 6 - other options ???
>
> In case 3, 4, 5, we also should decide what is the default.
>
> What do you think ?
I like 4 best, though I suggest calling the parameter 'normalize.'
How do you plan to handle normalize=False? I suggest
if the expression is not a mul:
return self
else:
separate the positive and negative exponents
Thank you.
Cheers,
Burcin
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
