doctest fix for: substitute exp(-x) with x infinity fails

[videlec]

sage: exp(-Infinity)
0

is fine, but

sage: exp(-x).subs(x=Infinity)
Traceback (most recent call last):
...
RuntimeError: exp_eval(): exp^(unsigned_infinity) encountered 

Original report thread:

https://groups.google.com/forum/#!topic/sage-devel/tqDSbZ499ME

Depends on #19948

CC: @rwst

Component: symbolics

Keywords: bug

Author: Benjamin Hackl

Branch/Commit: 45420f5

Reviewer: Ralf Stephan

Issue created by migration from https://trac.sagemath.org/ticket/19918

[jdemeyer]

comment:2

This is very likely a bug in pynac.

[behackl]

comment:3

As far as I can tell, there is some sort of sign confusion occurring in pynac. This is now pynac/pynac#123.

[behackl]

Upstream: Reported upstream. Developers acknowledge bug.

[rwst]

comment:4

The reason is

sage: hash(SR(-oo))
2886518159
sage: hash((-x).subs(x=oo))
81231252
sage: hash((x).subs(x=-oo))
2886518159

See #19927 and #19928 for related bugs.

[behackl]

Dependencies: pynac-0.6.0

[behackl]

Changed upstream from Reported upstream. Developers acknowledge bug. to Fixed upstream, in a later stable release.

[behackl]

comment:5

This will be fixed in pynac-0.6.0; I'll add a doctest on this ticket after the upgrade.

[behackl]

Changed dependencies from pynac-0.6.0 to pynac-0.6.1

[behackl]

comment:6

Replying to @behackl:

This will be fixed in pynac-0.6.0; I'll add a doctest on this ticket after the upgrade.

... typo; I meant the next release (pynac-0.6.1 or so...).

[behackl]

New commits:

5c88141	19448: update Pynac pkg data
b76a8ad	19948: add stieltjes() interface; fix doctest
4857402	19949: fix type of error raised by stieltjes()
be7423b	add doctest for exp (substitution with oo)
f2546c0	add doctest for arctan (substitution with oo)

[behackl]

Changed dependencies from pynac-0.6.1 to #19948

[behackl]

Branch: u/behackl/symbolics/oo_subs

[behackl]

Commit: f2546c0

[behackl]

Changed commit from f2546c0 to a9719e7

[behackl]

New commits:

8a9b6ac	Trac #19948: Upgrade to pynac-0.6.1
bd1c11d	add doctest for exp (substitution with oo)
a9719e7	add doctest for arctan (substitution with oo)

[behackl]

Changed branch from u/behackl/symbolics/oo_subs to u/behackl/symbolics/subs_oo

[jdemeyer]

comment:9

I recommend adding a test also for [comment:4], which is the real underlying reason of this ticket.

[behackl]

comment:10

I agree. Where is the best place to add tests for the equality of these hashes? Probably in the __hash__-methods in rings/infinity.py introduced in #19927?

[rwst]

comment:11

Replying to @behackl:

I agree. Where is the best place to add tests for the equality of these hashes? Probably in the __hash__-methods in rings/infinity.py introduced in #19927?

No, this is SR-specific, so Expression.__hash__ would be the place.

[jdemeyer]

comment:12

Indeed. There are already several doctests of the form hash(x) == hash(SR(x)) and that would fit there as well.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

00b7c17

add doctest for equality of hashes

[sagetrac-git]

Changed commit from a9719e7 to 00b7c17

[behackl]

comment:14

Thanks for pointing me there! I've added some tests for this; there might be some overlap with #19928. Do you want to have an explicit test like

sage: hash(x.subs(x=-oo)) == hash((-x).subs(x=oo))
True

too?

[rwst]

comment:15

What you added is fine already, thanks! I'll just wait for #19948 with the review.

[rwst]

Changed upstream from Fixed upstream, in a later stable release. to none

[rwst]

Reviewer: Ralf Stephan

[rwst]

Author: Benjamin Hackl

[vbraun]

comment:17

doctest fails

[rwst]

Changed branch from u/behackl/symbolics/subs_oo to u/rws/symbolics/subs_oo

[rwst]

New commits:

aba1d30	Merge branch 'develop' into t/19918/symbolics/subs_oo
45420f5	19918: fix typo

[rwst]

Changed commit from 00b7c17 to 45420f5

[vbraun]

Changed branch from u/rws/symbolics/subs_oo to 45420f5