Doctest fix for: dilog(RR) should return an element of RR

[zimmermann6]

in Sage 6.10:

sage: dilog(-1.0)
-1/12*pi^2
sage: dilog(1)
1/6*pi^2
sage: dilog(0.5)
1/12*pi^2 - 1/2*log(2)^2

we should not get symbolic output with floating-point input.

Reported as pynac/pynac#219

Upstream: Reported upstream. Developers acknowledge bug.

Component: basic arithmetic

Issue created by migration from https://trac.sagemath.org/ticket/19906

[rwst]

Upstream: Reported upstream. Developers acknowledge bug.

[rwst]

Branch: u/rws/dilog_rr__should_return_an_element_of_rr

[rwst]

Commit: 896bdc7

[rwst]

Dependencies: pynac-0.7.2

[rwst]

comment:3

This is fixed in Pynac git master. This ticket has doctests but they depend on a pynac-0.7.2 upgade.

---

New commits:

896bdc7

19906: doctest fix for: dilog(RR) should return an element of RR

[rwst]

Changed dependencies from pynac-0.7.2 to #21963

[videlec]

comment:5

You need to set your name in the Authors field.

[videlec]

comment:6

Could you be more careful and actually check the parent of the output? In other words something like

sage: y = dilog( RealField(13)(1.0))
sage: assert parent(y) is RealField(13)
sage: y = dilog( RealField(100)(1.0))
sage: assert parent(y) is RealField(100)

I don't know to which extent this function is available on intervals, balls and complexes but they should be tested as well with various precision.

[rwst]

Changed branch from u/rws/dilog_rr__should_return_an_element_of_rr to u/rws/19906

[rwst]

comment:9

Note also

sage: polylog(3,RBF(1.1))      # there is a polylog RBF member!
1.37625299668538 - 0.0142691615444952*I
sage: dilog(RBF(1.1))
1.96199910130557 - 0.299425760685590*I

---

New commits:

4f451a4

19906: doctest fix for: dilog(RR) should return an element of RR

[rwst]

Changed commit from 896bdc7 to 4f451a4

[zimmermann6]

comment:10

For me, the last commit 4f451a4 fixes the issue raised in comment [comment:6].
We could proceed and give a positive review. (The case of intervals, balls and complexes can be dealt with in a different ticket.)

Paul

[rwst]

Changed branch from u/rws/19906 to none

[rwst]

comment:11

Replying to @zimmermann6:

For me, the last commit 4f451a4 fixes the issue raised in comment [comment:6].
We could proceed and give a positive review. (The case of intervals, balls and complexes can be dealt with in a different ticket.)

No, actually the given doctest fails (that's why I didn't set needs review). But the notion that dilog(RealField(13)(1.0)) should return anything real I commented on in #18386 comment:8. More obviously, demanding that dilog(RealField(13)(1.1)) should have RealField(13) as parent is a red herring because the result just is complex.

However, there are still issues with polylog (and thus dilog) , I just reopened pynac/pynac#219 to fix them. I believe #18386 is more fitting to be the ticket to add a doctest once it's fixed so I propose to make this a duplicate. As you seem to be of similar attitude I take the liberty to set positive all at once.

[rwst]

Changed dependencies from #21963 to none

[rwst]

Changed commit from 4f451a4 to none

[zimmermann6]

comment:12

since the initial issue in the description of this ticket is now fixed (here with Sage 7.6):

sage: dilog(-1.0)
-0.822467033424113
sage: dilog(0.5)
0.582240526465012

I would simply add those two tests in the doctests, maybe with one test checking the parent:

sage: y = dilog(RealField(13)(0.5))
sage: parent(y)
Real Field with 13 bits of precision

and indeed defer to #18386 the other issues (branch point, complex values).

[embray]

comment:13

Closing tickets in the sage-duplicate/invalid/wontfix module with positive_review (i.e. someone has confirmed they should be closed).