The function ps_modsym_from_simple_modsym_space calls internally a
randomized method from modular symbols (dual_eigenvector). Here is a
snippet of code that also returns different answers depending on the seed
(on a fresh session, since dual_eigenvector gets cached).
f = Newforms(32, 8, names='a')[1]
A = f.modular_symbols(1)
w = A.dual_eigenvector()
K = w.base_ring()
print K.defining_polynomial()
The easy solution is to change the doctest in pollack_stevens' code. Here
is a solution that would work:
A number field example. Here there are multiple primes above `p`,
and
`\phi` is ordinary at one but not the other.::
sage: f = Newforms(32, 8, names='a')[1]
sage: K = f.hecke_eigenvalue_field()
sage: a = f[3]
sage: from sage.modular.pollack_stevens.space import
ps_modsym_from_simple_modsym_space
sage: phi =
ps_modsym_from_simple_modsym_space(f.modular_symbols(1))
sage: phi.is_ordinary(K.ideal(3, 1/16*a + 3/2)) !=
phi.is_ordinary(K.ideal(3,
1/16*a + 5/2))
True
sage: phi.is_ordinary(3)
Traceback (most recent call last):
...
TypeError: P must be an ideal
On Monday, May 15, 2017 at 3:57:21 PM UTC+2, vdelecroix wrote:
>
> Dear all,
>
> I am worried about a consequence of my cleaning in matrix_rational_dense
> at [1]. It appears that an innocent doctest in
> modular/pollack_stevens/modsym.py depends on the seed of the random
> generators where it should not be, see [2]. I tested a lot more than it
> used to be the code in matrix_rational_dense and I did not detect
> anything wrong there.
>
> If nobody knows how to fix it, I propose to quickly add a stopgap in the
> corresponding functions in sage/modular.
>
> Best,
> Vincent
>
> [1] https://trac.sagemath.org/ticket/22970
> [2] https://trac.sagemath.org/ticket/22970#comment:14
>
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.
