provide wrapper for psi function of GiNaC

[burcin]

From sage-support:

On Fri, 18 Sep 2009 15:47:45 -0700 (PDT)
The_Fool <masterfulet@yahoo.com> wrote:

> While working with the derivative of the Gamma function, the digamma
> function is obviously involved.  The sage "diff" function does show Γ
> '(x) == Γ(x)ψ(x) like it should, however, the digamma function (called
> psi in sage) is not defined whenever I try to do anything with it.  It
> seems as if only the output of "diff" can use this function.

psi() is defined in GiNaC, but we don't provide wrappers for it.

Defining a function psi, similar to the way arctan2 is defined in line 422 of sage/functions/trig.py should fix this.

The sage-support thread is here:

​http://groups.google.com/group/sage-support/browse_thread/thread/1ad313c921b7dbc0

[zimmerma]

More precisely:

sage: diff(gamma(x), x)
gamma(x)*psi(x)

but "psi" is unknown to Sage...

[burcin]

wrappers for psi function from pynac

[burcin]

It turns out defining this function wasn't as straightforward as I thought. Pynac defines 2 functions with the same name psi. :)

attachment:trac_6961-psi.patch​ wraps these 2 functions, and provides a global psi() to call the appropriate one. Unpickling these correctly requires changes in pynac. I'll make a new pynac package available later this week.

[burcin]

New pynac package available here:

​http://sage.math.washington.edu/home/burcin/pynac/pynac-0.1.11.spkg

The package contains fixes for #7822, #6961, #7876, #7363, #6465 and #6559. From these #7876 and #7363 contain printing fixes. Doctests should be run at least with patches from these two tickets.

[mvngu]

The patch on this ticket conflicts with #6207. If I first apply the patches on #6207 to Sage 4.3.1, and then ​trac_6961-psi.patch, I would get a hunk failure:

[mvngu@sage sage-main]$ hg qimport http://trac.sagemath.org/sage_trac/raw-attachment/ticket/6207/trac_6207.patch && hg qpush
adding trac_6207.patch to series file
applying trac_6207.patch
now at: trac_6207.patch
[mvngu@sage sage-main]$ hg qimport http://trac.sagemath.org/sage_trac/raw-attachment/ticket/6207/trac_6207-part2.patch && hg qpush
adding trac_6207-part2.patch to series file
applying trac_6207-part2.patch
now at: trac_6207-part2.patch
[mvngu@sage sage-main]$ hg qimport http://trac.sagemath.org/sage_trac/raw-attachment/ticket/6207/trac_6207-review.patch && hg qpush
adding trac_6207-review.patch to series file
applying trac_6207-review.patch
now at: trac_6207-review.patch
[mvngu@sage sage-main]$ hg qimport http://trac.sagemath.org/sage_trac/raw-attachment/ticket/6961/trac_6961-psi.patch && hg qpush
adding trac_6961-psi.patch to series file
applying trac_6961-psi.patch
patching file sage/symbolic/pynac.pyx
Hunk #2 FAILED at 1566
1 out of 3 hunks FAILED -- saving rejects to file sage/symbolic/pynac.pyx.rej
patch failed, unable to continue (try -v)
patch failed, rejects left in working dir
errors during apply, please fix and refresh trac_6961-psi.patch

Perhaps you want to rebase ​trac_6961-psi.patch on top of #6207?

[burcin]

rebased

[kcrisman]

The unrebased patch is fine; I can't get the other patches in the symbolics queue to work off of the rebase, not yet anyway.

If there is time, would it be possible to make an alias polygamma = psi2? Not just because Mma uses it, but because a lot of references call it that (such as the ones a students of mine is using for some research) in written form. This is great functionality to add, though.

[kcrisman]

Replying to burcin:

New pynac package available here:

​http://sage.math.washington.edu/home/burcin/pynac/pynac-0.1.11.spkg

Now apparently at boxen.math instead.

[rossk]

sage: diff(gamma(x),x)/gamma(x) # definition of digamma (or "psi")
psi(x)

sage: diff(gamma(x),x).subs(x=1)
-euler_gamma

sage: psi(1)
-euler_gamma

# analytical result
sage: psi(1/2)
-euler_gamma - 2*log(2)

# numerical examples
sage: psi(.5)
-1.96351002602142

sage: psi(.5r)
-1.9635100260214233

sage: psi(complex(.5))
(-1.9635100260214233+0j)

# 1st argument means 0'th derivative (so psi(0, x) = psi(x))
sage: psi(0, .5)
psi(0.500000000000000)

sage: N(psi(0, .5))
-1.96351002602142

[mvngu]

fix doctest failure

[mvngu]

I applied patches in the following order against Sage 4.3.3.alpha0, together with the updated package ​pynac-0.1.11.spkg:

1. #6961: ​trac_6961-psi.rebased.patch
2. #7822: ​trac_7822-py_log.take2.patch
3. #7876: ​trac_7876-pynac_print.take2.patch
4. #7363
5. #7957
6. #7916: ​trac_7916-same_name_method.take2.patch
7. #6465:
   1. ​trac_6465-chain_rule.take2.patch
   2. ​trac_6465-moves-integration-into-sfunction-subclass.take3.patch
   3. ​trac_6465-integral.take4.patch
8. #6559:
   1. ​trac_6559-domain-and-latex_name-for-variable.take2.3.patch
   2. ​trac_6559-referee.take2.patch

All doctests passed except for this trivial failure:

[mvngu@sage sage-4.3.3.alpha0.1]$ ./sage -t -long devel/sage/sage/rings/arith.py
sage -t -long "devel/sage/sage/rings/arith.py"              
**********************************************************************
File "/dev/shm/mvngu/sandbox/sage-4.3.3.alpha0.1/devel/sage/sage/rings/arith.py", line 287:
    sage: factorial(-32)
Expected:
    Traceback (most recent call last):
    ...
    ValueError: factorial -- self = (-32) must be nonnegative
Got:
    Traceback (most recent call last):
      File "/dev/shm/mvngu/sandbox/sage-4.3.3.alpha0.1/local/bin/ncadoctest.py", line 1231, in run_one_test
        self.run_one_example(test, example, filename, compileflags)
      File "/dev/shm/mvngu/sandbox/sage-4.3.3.alpha0.1/local/bin/sagedoctest.py", line 38, in run_one_example
        OrigDocTestRunner.run_one_example(self, test, example, filename, compileflags)
      File "/dev/shm/mvngu/sandbox/sage-4.3.3.alpha0.1/local/bin/ncadoctest.py", line 1172, in run_one_example
        compileflags, 1) in test.globs
      File "<doctest __main__.example_3[10]>", line 1, in <module>
        factorial(-Integer(32))###line 287:
    sage: factorial(-32)
      File "/dev/shm/mvngu/sandbox/sage-4.3.3.alpha0.1/local/lib/python/site-packages/sage/rings/arith.py", line 315, in factorial
        raise ValueError, "factorial -- must be nonnegative"
    ValueError: factorial -- must be nonnegative
**********************************************************************
1 items had failures:
   1 of  11 in __main__.example_3
***Test Failed*** 1 failures.
For whitespace errors, see the file /dev/shm/mvngu/dot_sage/tmp/.doctest_arith.py
	 [50.3 s]

The failure is fixed via ​trac_6961-doctest.patch, so only this latter patch needs some reviewing love. I'm happy with both #6465 and #6559. If my patch is OK, then these 8 tickets can be closed: #6961, #7822, #7876, #7363, #7957, #7916, #6465, #6559.

[burcin]

Hi Minh,

Your patch looks ok to me. I'm switching this to positive review.

It's be truly awesome if all those tickets can get merged in this release. :)

Cheers,
Burcin

[mvngu]

Merged in this order:

1. #6961: ​trac_6961-psi.rebased.patch
2. #7822
3. #7876
4. #7363
5. #7957
6. #7916
7. #6465
8. #6559
9. #6961: ​trac_6961-doctest.patch