missing Sage interface for SymPy's RisingFactorial

[zimmermann6]

Minimal test case:

sage: import sympy
sage: SR(sympy.RisingFactorial(x,x))
AttributeError: 'RisingFactorial' object has no attribute '_sage_'
sage: SR(sympy.FallingFactorial(x,x))
AttributeError: 'FallingFactorial' object has no attribute '_sage_'

Original ticket description:

This is related to #14437:

sage: from sympy import Function, Symbol
sage: u = Function('u')
sage: n = Symbol('n', integer=True)
sage: from sympy import rsolve
sage: f = u(n+2) - u(n+1) + u(n)/4
sage: rsolve(f,u(n))
2**(-n)*C0*RisingFactorial(C0/C1 + 1, n)/RisingFactorial(C0/C1, n)

The result returned by rsolve is a SymPy object:

sage: s = rsolve(f,u(n))
sage: type(s)
<class 'sympy.core.mul.Mul'>

Ideally, it should be automatically converted to Sage.

However, if we try to convert it manually, we get:

sage: s._sage_()
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)

/users/caramel/zimmerma/svn/sagebook/tex/<ipython console> in <module>()

/usr/local/sage-5.1-linux-64bit-ubuntu_12.04_lts-x86_64-Linux/local/lib/python2.7/site-packages/sympy/core/mul.py in _sage_(self)
   1192         s = 1
   1193         for x in self.args:
-> 1194             s *= x._sage_()
   1195         return s
   1196 

/usr/local/sage-5.1-linux-64bit-ubuntu_12.04_lts-x86_64-Linux/local/lib/python2.7/site-packages/sympy/core/power.py in _sage_(self)
    846 
    847     def _sage_(self):
--> 848         return self.args[0]._sage_()**self.args[1]._sage_()
    849 
    850 from add import Add

AttributeError: 'RisingFactorial' object has no attribute '_sage_'

Paul

CC: @kcrisman

Component: interfaces

Reviewer: Paul Zimmermann

Issue created by migration from https://trac.sagemath.org/ticket/14446

[kcrisman]

comment:5

Just as a point of information, nowadays

sage: rsolve(f,u(n))
2**(-n)*(C0 + C1*n)

so apparently there was some extra simplification that should have happened. Doesn't mean this ticket is invalid, of course.

[rwst]

comment:8

This is now fixed in the proposed sympy pull request sympy/sympy#8592
but we will upload a branch with the specific sympy patch next.

[rwst]

comment:9

This was resolved long ago by the latest SymPy update.

[zimmermann6]

Reviewer: Paul Zimmermann

[zimmermann6]

comment:10

fixed indeed in Sage 8.0 (or before).

[kcrisman]

comment:11

I've noticed a few recent closed tickets which were apparently resolved -- but no one mentioned doctesting these resolutions. We should do that to avoid regressions. (Unless they are already tested, of course.)

[zimmermann6]

comment:12

I tried adding a doctest with those commands:

$ git clone git://git.sagemath.org/sage.git
$ cd sage
$ git branch u/zimmerma/14446
$ git co u/zimmerma/14446
$ patch -p1 -i /tmp/patch
$ git commit -a
$ git push --set-upstream origin u/zimmerma/14446

but it failed with:

$ git push --set-upstream origin u/zimmerma/14446
fatal: remote error: access denied or repository not exported: /sage.git

Am I doing something wrong?

For the record:

$ cat /tmp/patch
commit ef0235ff466d40c787681e3e483342e091247392
Author: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Date:   Tue Oct 3 13:03:18 2017 +0200

    added doctest for #14446

diff --git a/src/sage/calculus/test_sympy.py b/src/sage/calculus/test_sympy.py
index 996e417..15b1d7e 100644
--- a/src/sage/calculus/test_sympy.py
+++ b/src/sage/calculus/test_sympy.py
@@ -196,4 +196,15 @@ This was fixed in Sympy, see :trac:`14437`::
     sage: rsolve(f,u(n))
     2**(-n)*(C0 + C1*n)
 
+This was fixed in Sympy, see :trac:`14446`::
+
+    sage: import sympy
+    sage: SR(sympy.RisingFactorial(x,x))
+    gamma(2*x)/gamma(x)
+    sage: SR(sympy.RisingFactorial(x,3))
+    (x + 2)*(x + 1)*x
+    sage: SR(sympy.FallingFactorial(x,x))
+    gamma(x + 1)
+    sage: SR(sympy.FallingFactorial(x,3))
+    (x - 1)*(x - 2)*x
 """

[rwst]

comment:13

Actually this was a SymPy fix which should be tested in SymPy:
https://github.com/sympy/sympy/blob/master/sympy/functions/combinatorial/factorials.py#L559
https://github.com/sympy/sympy/blob/master/sympy/external/tests/test_sage.py

[kcrisman]

comment:14

That is very interesting. Is it possible to have this sympy file doctested whenever Sage runs doctests? (I don't know how many people would run the package doctests, and in any case this probably isn't automatically tested as easily within sympy itself.) That would be very helpful.

[rwst]

comment:15

Replying to @kcrisman:

That is very interesting. Is it possible to have this sympy file doctested whenever Sage runs doctests?

Just copy it at SymPy installation time into src/sage/tests.

this probably isn't automatically tested as easily within sympy itself

You're right. Their Travis doesn't have a Sage installation AFAIK.

That would be very helpful.

So I opened #23960.