Endless symbolic computation

[sagetrac-ipasquinelli]

I tried calculating

a=e^(I*pi/4)+1
b=1-e^(I*pi/4)
a*b

and both expressions a*b and a/b don't stop computing.

I tried both on sage-8.1 for Windows and on sage-8.2.beta6 on Ubuntu (native Ubuntu desktop on Windows10).

Depends on #24838

CC: @rwst @videlec

Component: symbolics

Keywords: days93, days94

Author: Irene Pasquinelli

Branch/Commit: 4d0c51e

Reviewer: Ralf Stephan, Travis Scrimshaw

Issue created by migration from https://trac.sagemath.org/ticket/24883

[videlec]

comment:1

Could you add your architecture/sage version in the ticket description? I can confirm the behavior on archlinux with compiled sage-8.2.beta6.

[videlec]

comment:2

Alternative computation that terminates:

UCF = UniversalCyclotomicField()
a = UCF.zeta(8) + 1
b = 1 - UCF.zeta(8)
a * b

[sagetrac-sdrewitz]

comment:4

I can confirm it on archlinux with compiled sage-8.2.beta6, sage-8.2.beta5 and sage-8.1.
However, with the sagemath 8.1-11 from the archlinux community repository it does work.

sage: a = 1 + e^(I*pi/4)
sage: b = 1 - e^(I*pi/4)
sage: a*b
1/4*((I + 1)*sqrt(2) - 2)*(-(I + 1)*sqrt(2) - 2)
sage: a/b
-1/2*(-(I + 1)*sqrt(2) - 2)/(-(1/2*I + 1/2)*sqrt(2) + 1)

[rwst]

comment:5

Confirmed in beta6. The loop is in Pynac. Thanks for the report.

[rwst]

comment:6

Actually there were changes in pynac-0.7.17 that appear to have fixed it. With #24838:

sage:  sage: a = 1 + e^(I*pi/4)
....:  sage: b = 1 - e^(I*pi/4)
....:  sage: a*b
....: 
1/4*((I + 1)*sqrt(2) - 2)*(-(I + 1)*sqrt(2) - 2)

We might doctest this in this ticket, though.

[videlec]

comment:7

Replying to @rwst:

Actually there were changes in pynac-0.7.17 that appear to have fixed it. With #24838:

Cool

We might doctest this in this ticket, though.

+1

[videlec]

Author: Irene Pasquinelli

[videlec]

comment:8

Ralf, Irene is working on this ticket (she is learning how to develop). We will have a branch in a minute. Thanks for pointing #24838.

[sagetrac-ipasquinelli]

Branch: u/ipasquinelli/24883

[sagetrac-ipasquinelli]

New commits:

e470685	version / chkum / rm patch
d3511ce	24838: doctest fixes
efe5f14	24883: adding doctest for symbolic

[sagetrac-ipasquinelli]

Commit: efe5f14

[sagetrac-ipasquinelli]

Dependencies: #24838

[rwst]

Reviewer: Ralf Stephan

[rwst]

comment:11

I think this is fine, but we may have to wait for setting positive until #24838 gets positive.

[rwst]

Changed branch from u/ipasquinelli/24883 to public/24883

[rwst]

Changed commit from efe5f14 to 699c0a2

[rwst]

comment:13

Okay, the dependency #24838 is now in the develop branch. Your branch no longer merges because there were further changes in the dependency. Instead of merging them I created a fresh branch with only your commit (using git cherry-pick).

---

New commits:

699c0a2

24883: adding doctest for symbolic

[tscrim]

Changed keywords from days93 to days93, days94

[tscrim]

Changed reviewer from Ralf Stephan to Ralf Stephan, Travis Scrimshaw

[tscrim]

comment:14

LGTM.

[vbraun]

comment:15

see patchbot

[tscrim]

comment:16

Problem actually comes from an earlier doctest:

sage: e = x+1 <= x-2

(I tested this by copy/pasting the doctest).
So we probably need to use exp.

[sagetrac-git]

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

cb24f9a

24883: improve usage of ambigous symbol in doctest

[sagetrac-git]

Changed commit from 699c0a2 to cb24f9a

[videlec]

comment:19

update milestone 8.3 -> 8.4

[tscrim]

comment:20

Forgot about this. LGTM.

[sagetrac-git]

Changed commit from cb24f9a to 4d0c51e

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

4d0c51e

Merge with sage-8.3

[bryangingechen]

comment:22

Replying to @sagetrac-git:

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

4d0c51e

Merge with sage-8.3

I fixed a trivial merge conflict with 8.3.

[vbraun]

Changed branch from public/24883 to 4d0c51e