Comment #10 on issue 2607 by asmeurer: as_numer_denom() is too slow
http://code.google.com/p/sympy/issues/detail?id=2607
Ah, it's nice to talk to someone other than myself here :)
Yes, Ronan is exactly right. I carefully split apart the code and profiled
it using kernprof/line_profiler (awesome tool, btw), and the bottleneck is
line 97 of mul.py, "coeff *= o". In other words, even though it's just
multiplying 1 together, it's doing it thousands of times (999001 times to
be exact, according to kernprof). If you look at what it's doing, it's
computing multiplications duplicate times.
Re benchmarking: Yes, definitely.
Re preprocessing: I didn't think of this. That would be an additional
improvement. I think the sift() function from utilities/iterables.py would
be useful here. So, we need to benchmark three cases: all denominators the
same (trivial case), all denominators different, and somewhere in between.
By the way, I just realized that when I fixed is_rational_function to be
faster at https://github.com/sympy/sympy/pull/382, it was slow because it
was using is_numer_denom(). It didn't occur to me then that
is_numer_denom() was being too slow. But we should try benchmarking the
expresion I used there too, a = Add(*(exp(i*x) for i in range(1000))).
I'll get to this soon, I promise.
--
You received this message because you are subscribed to the Google Groups
"sympy-issues" 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-issues?hl=en.
