py3: future division in matrix2

[fchapoton]

part of #15995

found using the "python -3" tests suggested there

Component: python3

Keywords: division

Author: Frédéric Chapoton

Branch: fdf0a68

Reviewer: André Apitzsch, Jeroen Demeyer

Issue created by migration from https://trac.sagemath.org/ticket/23731

[fchapoton]

Branch: u/chapoton/23731

[fchapoton]

Commit: e65a63e

[fchapoton]

New commits:

e65a63e

py3: future division in matrix2

[a-andre]

Reviewer: André Apitzsch

[a-andre]

comment:3

LGTM.

[jdemeyer]

comment:4

I disagree here: why hard-code QQ? Shouldn't we use self._parent._base instead?

@@ -1371,7 +1372,7 @@ cdef class Matrix(Matrix1):
         # now compute the permanental minor of the complement matrix if needed
         if complement:
             a = [one]
-            c1 = 1
+            c1 = QQ.one()
             for k in range(1, mn + 1):
                 c1 = c1*(m-k+1)*(n-k+1)/k
                 c = c1

[fchapoton]

comment:5

Because for this precise variable, we can stay in QQ. Otherwise, more complicated computations are done.

[sagetrac-git]

Changed commit from e65a63e to fdf0a68

[sagetrac-git]

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

fdf0a68

trac 23731 try exact division

[fchapoton]

comment:7

So, what about the new commit ? The first // makes sense (binomial coefficients) but i would not swear that the second one does make sense too.

---

New commits:

fdf0a68

trac 23731 try exact division

[jdemeyer]

Changed reviewer from André Apitzsch to André Apitzsch, Jeroen Demeyer

[vbraun]

Changed branch from u/chapoton/23731 to fdf0a68

[embray]

Changed commit from fdf0a68 to none

[embray]

Changed keywords from none to division