Py3: Minor enhancements in rings.polynomial.multi_polynomial.pyx

[vklein]

• Replace a doctest with #random with a doctest returning the same result in py2 and py3.
• Remove unused imports.

[vklein]

New commits:

​2e3024d

Trac #27791: Fix multi_polynomial.pyx doctests ...

[jhpalmieri]

I'm not sure it makes sense to sort list(ff). In principle, you could reconstruct the function from this list, but not if you sort it. In particular, once you sort, the previous comment "Currently, we use a fairly unoptimized method that evaluates one monomial at a time, with no sharing of repeated computations and with useless additions of 0 and multiplications by 1" is hard to make sense of.

[vklein]

Replying to jhpalmieri:

I'm not sure it makes sense to sort list(ff). In principle, you could reconstruct the function from this list, but not if you sort it.

Does that mean even if the list comes in a different order in py2 and py3 you will get an equivalent function by reconstruction ?

If this is the case what do you think is the best fix ? #py2 #py3 tags.

[jhpalmieri]

Replying to vklein:

Replying to jhpalmieri:

I'm not sure it makes sense to sort list(ff). In principle, you could reconstruct the function from this list, but not if you sort it.

Does that mean even if the list comes in a different order in py2 and py3 you will get an equivalent function by reconstruction ?

This all guesswork on my part, but I think so. The list is just giving the summands in two different orders.

'push 0.0', 'push 12.0', 'load 1', 'load 2', 'dup', 'mul', 'mul', 'mul', 'add',

says to store 0, then store 12, take a factor of y ("load 1"), take a factor of z ("load 2"), duplicate it to get two factors of z, multiply (so you have z^2), multiply again (so you have y z^2), multiply again (to incorporate the factor of 12), then add to the 0 from the start.

With Python 3, it starts with the summand 9z^4 instead. The ordering in Python 3 seems to match the order in which the polynomial is actually printed, for what that's worth.

If this is the case what do you think is the best fix ? #py2 #py3 tags.

How about a new doctest: instead of list(ff), do something like this after the comment:

sage: g = (x*y**2*z)._fast_float_()
sage: list(g)
[ insert output here ]

Since there is only one summand, there should be no issues with orderings, and I think it still illustrates the same issue.

[git]

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

​9567cc9

Trac #27791: Implement reviewer's comment.

[vklein]

It works ! Thanks for your help.

[vklein]

Merge failed with 8.8.beta5

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​c532b8d

Trac #27791: Fix multi_polynomial.pyx doctests ...

[vklein]

Rebased with 8.8.beta5 which already fix the py3 doctests.

[jhpalmieri]

Looks good to me.