Sage fails to compute a (simple) symbolic expression

[EmmanuelCharpentier]

See this sage-devel thread. In short:

sage: t=((1/10)*I/pi)^(3/2)
sage: t
(1/10*I/pi)^(3/2)
sage: t^2

This doesn't return(!), and seems hard to interrupt.

Depends on #30446

CC: @dimpase @videlec

Component: symbolics

Reviewer: Dave Morris, Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/28620

[sheerluck]

comment:2

with some additional prints:

sage: t=((1/10)*I/pi)**3

Expression._pow_ 1/10*I/pi 3
class NumberFieldElement.__pow__ 1/10*I 3 None
generic_power_long', 1/10*I, 3
generic_power_long', 1/10*I, 3
generic_power_pos', 1/10*I, 3
apow = -1/100
n = 0
res = -1/1000*I

sage: t**2

Expression._pow_ -1/1000*I/pi^3 2
class NumberFieldElement.__pow__ -1/1000*I 2 None
generic_power_long, -1/1000*I, 2
generic_power_long, -1/1000*I, 2
generic_power_pos, -1/1000*I, 2
apow = -1/1000000
n = 1

-1/1000000/pi^6

sage: t=((1/10)*I/pi)**(3/2)

Expression._pow_ 1/10*I/pi 3/2

sage: t**2

Expression._pow_ (1/10*I/pi)^(3/2) 2
class NumberFieldElement.__pow__ 1/10*I 4294967297 None
generic_power_long', 1/10*I, 4294967297
generic_power_long', 1/10*I, 4294967297
generic_power_pos', 1/10*I, 4294967297
apow = -1/100
n = 1073741824
res = 1/10*I
apow = 1/10000
n = 536870912
res = 1/10*I
apow = 1/100000000
n = 268435456
res = 1/10*I
apow = 1/10000000000000000
n = 134217728
res = 1/10*I
apow = 1/100000000000000000000000000000000
n = 67108864
res = 1/10*I
...

[EmmanuelCharpentier]

comment:3

Replying to @sheerluck:

with some additional prints:

How did you got them ? I wasn't aware of the availability of this steb-by-step tracing...

[ SNIP... ]

Illuminating...

[sheerluck]

comment:4

no, it's not steb-by-step tracing, I just inserted lines in

• symbolic/expression.pyx
• rings/number_field/number_field_element.pyx
• arith/power.pyx
  and just reinstalled sage (I use Gentoo Linux)

[embray]

comment:5

Ticket retargeted after milestone closed

[mkoeppe]

comment:6

Still unfixed in 9.1rc2

[DaveWitteMorris]

Dependencies: #30446

[DaveWitteMorris]

comment:9

This will be fixed by the patch to pynac in #30446. When that ticket is closed, we can add this as another doctest.

[DaveWitteMorris]

comment:10

This can be closed as a duplicate of #30446. The pull request at #30786 adds the appropriate doctest.

[DaveWitteMorris]

Reviewer: Dave Morris

[dimpase]

Changed reviewer from Dave Morris to Dave Morris, Dima Pasechnik