eisenstein_series_qexp broken over finite fields

[aghitza]

sage: eisenstein_series_qexp(10, 15, GF(17))
15 + q + 3*q^2 + 15*q^3 + 7*q^4 + 13*q^5 + 11*q^6 + 11*q^7 + 15*q^8 + 7*q^9 + 5*q^10 + 7*q^11 + 2*q^12 + 12*q^13 + 12*q^14 + O(q^15)

is wrong, whereas

sage: eisenstein_series_qexp(10, 15).change_ring(GF(17))
15 + q + 3*q^2 + 15*q^3 + 7*q^4 + 13*q^5 + 11*q^6 + 11*q^7 + 15*q^8 + 7*q^9 + 5*q^10 + 7*q^11 + 3*q^12 + 14*q^13 + 16*q^14 + O(q^15)

is right. We tracked this down to a change in the polynomials over finite fields constructor when check=False. We'll quickly fix this at the cost of making it slower; better fix will come soon.

CC: @craigcitro

Component: modular forms

Issue created by migration from https://trac.sagemath.org/ticket/5102

[aghitza]

comment:1

Attachment: trac_5102.patch.gz

Credit to Craig Citro and Alex Ghitza.

[mwhansen]

comment:2

This gets a positive review from William.

[sagetrac-mabshoff]

comment:3

Merged in Sage 3.3.alpha4.

Cheers,

Michael