Hi,
Thanks Harald for inviting us (in SymPy) to participate under Sage
this year. One big goal is to make Sage use CSymPy
(https://github.com/sympy/csympy) as the main symbolic engine instead
of Pynac. CSymPy is written in C++ with optional Python wrappers. The
motivation is simple, it is much faster, e.g.:
certik@redhawk:~/repos/csympy/benchmarks(master)$ sage legendre1_sage.py
import...
done.
0 1
1 x
2 3/2*x^2 - 1/2
3 5/2*x^3 - 3/2*x
4 35/8*x^4 - 15/4*x^2 + 3/8
5 63/8*x^5 - 35/4*x^3 + 15/8*x
6 231/16*x^6 - 315/16*x^4 + 105/16*x^2 - 5/16
7 429/16*x^7 - 693/16*x^5 + 315/16*x^3 - 35/16*x
8 6435/128*x^8 - 3003/32*x^6 + 3465/64*x^4 - 315/32*x^2 + 35/128
9 12155/128*x^9 - 6435/32*x^7 + 9009/64*x^5 - 1155/32*x^3 + 315/128*x
Total time for legendre(500, x): 7.2570168972 s
certik@redhawk:~/repos/csympy/benchmarks(master)$ python legendre1.py
0 1
1 x
2 -1/2 + 3/2*x**2
3 5/2*x**3 + -3/2*x
4 3/8 + -15/4*x**2 + 35/8*x**4
5 63/8*x**5 + 15/8*x + -35/4*x**3
6 -5/16 + 105/16*x**2 + 231/16*x**6 + -315/16*x**4
7 -693/16*x**5 + 429/16*x**7 + 315/16*x**3 + -35/16*x
8 35/128 + -315/32*x**2 + -3003/32*x**6 + 3465/64*x**4 + 6435/128*x**8
9 12155/128*x**9 + -6435/32*x**7 + 315/128*x + -1155/32*x**3 + 9009/64*x**5
Total time for legendre(500, x): 0.983409166336 s
You can check that the symbolic code is equivalent, it just evaluates
the Rodrigues' formula (obviously this is not the fastest way to get
Legendre polynomials, it's just for benchmarking general symbolic
code).
You can run more benchmarks yourself
(https://github.com/sympy/csympy/tree/master/benchmarks).
And CSymPy's community is much more active than GiNaC, e.g. compare
GiNaC and SymPy for the last year:
csympy$ git shortlog -ns --since="1 year ago"
336 Ondřej Čertík
271 Thilina Bandara Rathnayake
238 Sushant Hiray
109 Isuru Fernando
10 Peter Brady
9 Shivam Vats
4 Sumith
3 Thomas Hisch
1 Vinzent Steinberg
ginac$ git shortlog -ns --since="1 year ago"
6 Richard Kreckel
6 Vladimir V. Kisil
5 Alexei Sheplyakov
3 Stefan Weinzierl
and even all time (CSymPy from 2012, GiNaC from 1999) is really quite
comparable:
csympy$ git shortlog -ns
1039 Ondřej Čertík
338 Thilina Bandara Rathnayake
238 Sushant Hiray
109 Isuru Fernando
10 Peter Brady
9 Shivam Vats
5 Dale Lukas Peterson
4 Sumith
3 Thomas Hisch
1 Christopher Dembia
1 Julien Rioux
1 Vinzent Steinberg
ginac$ git shortlog -ns
669 Richard Kreckel
650 Christian Bauer
279 Jens Vollinga
201 Alexei Sheplyakov
66 Chris Dams
32 Alexander Frink
20 Vladimir V. Kisil
9 Stefan Weinzierl
7 Jan Rheinländer
7 Richard B. Kreckel
3 welzel
1 Matthias Dellweg
1 Paul Irofti
Pynac itself also doesn't seem to be very active:
http://hg.pynac.org/pynac/commits/all
Compared to GiNaC, CSymPy is missing specialized polynomial
manipulation and series expansion (we are working on both now) and
pattern matching (we also plan to do it soon), hopefully as other
GSoC projects.
I was thinking that a GSoC project could:
* Write Sage wrappers (the SymPy wrappers are already done and we can
have arbitrary sympy expression as an object in the C++ expression
tree), so most of this work is done --- then anybody in Sage can
easily try it out
* Figure out what needs to be done to hook it up just like Pynac is
(this might be quite a bit of work, hopefully it's manageable)
* Identify what features in CSymPy are still missing (and start
implementing those)
At the end of the summer I don't expect Sage to be switched, but I
think there could be a clear path forward, and it should be possible
to use it as an optional package in Sage, side by side, with good
interoperability. People can try it out, provide feedback, etc., which
would provide further motivation for us to finish it up and so on.
I was hoping to propose this project after we have all the features
already in there, but since SymPy didn't get accepted as an
organization this year, this GSoC project would really help us out,
and I think this is equally beneficial to both Sage and SymPy. We have
students that can apply for this. Let me know what you think, I'll be
happy to help them with the application and look at more details.
Ondrej
--
You received this message because you are subscribed to the Google Groups
"sage-gsoc" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-gsoc.
For more options, visit https://groups.google.com/d/optout.
