Opened 11 years ago

Closed 10 years ago

#7930 closed defect (fixed)

strange bug for elliptic curves over number fields

Reported by: wuthrich Owned by: cremona
Priority: major Milestone: sage-4.5.2
Component: elliptic curves Keywords:
Cc: cremona, was, robertwb Merged in: sage-4.5.2.alpha0
Authors: Chris Wuthrich Reviewers: John Cremona
Report Upstream: N/A Work issues:
Branch: Commit:
Dependencies: Stopgaps:


The following code

E = EllipticCurve('99d1')

R.<X> = QQ[]
K.<t> = NumberField(X^3 + X^2 - 2*X - 1)
L.<s> = NumberField(X^3 + X^2 - 36*X - 4)

EK = E.base_extend(K)
toK = EK.torsion_order()
da = EK.local_data()

EL = E.base_extend(L)
da = EL.local_data()

produces a TypeError. Having played around with this for hours, I believe that all lines in the above code are necessary to produce the bug. Especially both the torsion and local data computations over another field. I have no idea whatsoever of where to look for the error.

Attachments (1)

trac_7930.patch (1.6 KB) - added by wuthrich 10 years ago.
exported against 4.4.4.alpha0

Download all attachments as: .zip

Change History (12)

comment:1 Changed 11 years ago by wuthrich

More precisely it produces

Traceback (most recent call last):    
  File "", line 1, in <module>
  File "/tmp/tmpJwKYhV/", line 23, in <module>
    da = EK.local_data()
  File "", line 1, in <module>
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/schemes/elliptic_curves/", line 437, in local_data
    return [self._get_local_data(pr, proof) for pr in primes]
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/schemes/elliptic_curves/", line 487, in _get_local_data
    self._local_data[P, proof] = EllipticCurveLocalData(self, P, proof)
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/schemes/elliptic_curves/", line 159, in __init__
    self._Emin, ch, self._val_disc, self._fp, self._KS, self._cp, self._split = self._tate(proof)
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/schemes/elliptic_curves/", line 704, in _tate
    cp = 1 + _pcubicroots(b, c, d)
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/schemes/elliptic_curves/", line 536, in _pcubicroots
    return sum([rr[1] for rr in PolynomialRing(F, 'x')([d, c, b, 1]).roots()],0)
  File "parent.pyx", line 538, in sage.structure.parent.Parent.__call__ (sage/structure/parent.c:4956)
  File "coerce_maps.pyx", line 82, in sage.structure.coerce_maps.DefaultConvertMap_unique._call_ (sage/structure/coerce_maps.c:3109)
  File "coerce_maps.pyx", line 77, in sage.structure.coerce_maps._call_ (sage/structure/coerce_maps.c:3000)
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/rings/polynomial/", line 312, in _element_constructor_
    return C(self, x, check, is_gen, construct=construct, **kwds)
  File "/usr/local/sage/local/lib/python2.6/site-packages/sage/rings/polynomial/", line 604, in __init__
    Polynomial_generic_dense.__init__(self, parent, x, check, is_gen)
  File "polynomial_element.pyx", line 5111, in sage.rings.polynomial.polynomial_element.Polynomial_generic_dense.__init__ (sage/rings/polynomial/polynomial_element.c:34951)
  File "residue_field.pyx", line 975, in sage.rings.residue_field.ResidueFiniteField_givaro.__call__ (sage/rings/residue_field.c:9062)

comment:2 Changed 11 years ago by cremona

  • Cc robertwb added

In the middle of Tate's algorithm it needs to know how many roots a cubic mod P has, where P is the prime, and in doing this it constructs the polynomial ring over the residue field on the fly. These poly rings are constructed in such a way that for each field the rings is only constructed once (I think). the errors a re something which goes wrong in that construction.

I am CC'ing robertwb who I hope will be able to help!

comment:3 Changed 11 years ago by robertwb

No time to look t it now, but I've added this to my bug hit list for this upcoming week.

comment:4 Changed 11 years ago by wuthrich

Note that in #7935 we have added a #random to line 1026 in, because the output of the minimal_global_model produced two differnt answers. Maybe this is linked to this bug and the #random can be removed once the issue here is fixed.

This is very strange. For that curve E2, I sometimes get

Elliptic Curve defined by y^2 + a*x*y + (a+1)*y = x^3 + (a+1)*x^2 + (12289755603565800754*a-75759141535687466985)*x + (51556320144761417221790307379*a-317814501841918807353201512829) over Number Field in a with defining polynomial x^2 - 38

but sometimes I get

Elliptic Curve defined by y^2 + a*x*y + (a+1)*y = x^3 + (a+1)*x^2 + (368258520200522046806318444*a-2270097978636731786720859345)*x + (8456608930173478039472018047583706316424*a-52130038506793883217874390501829588391299) over Number Field in a with defining polynomial x^2 - 38

comment:5 Changed 11 years ago by cremona

Note that E has bad reductions at 3 and 11, which are both inert in both those cubic fields. That means that when we are finding the local data for each field, we construct the residue fields which are of order 3^3 and 11^3 in each case, but the generating polynomial (called the "modulus" in the finite field constructors) will be different, namely X^3 + X^2 - 2*X - 1 and then X^3 + X^2 - 36*X - 4.

This might help debugging the session will have more than one field of the same order, with different defining polynomials.

comment:6 Changed 11 years ago by wuthrich

Any news on this ?

comment:7 Changed 10 years ago by cremona

This problem seems to have gone away now -- in 4.4.4 the above script works fine. Can we close the ticket?

comment:8 Changed 10 years ago by wuthrich

  • Authors set to Chris Wuthrich

Almost. Here is a patch that adds the example as a doctest (in a _function so as not to appear in the documentation) to make sure that the bug does not appear again in the future. Because we are not certain what caused the bug in the first place.

comment:9 Changed 10 years ago by wuthrich

  • Status changed from new to needs_review

Changed 10 years ago by wuthrich

exported against 4.4.4.alpha0

comment:10 Changed 10 years ago by cremona

  • Reviewers set to John Cremona
  • Status changed from needs_review to positive_review

Looks good & tests fine on 4.4.4.alpha0.

comment:11 Changed 10 years ago by mpatel

  • Merged in set to sage-4.5.2.alpha0
  • Resolution set to fixed
  • Status changed from positive_review to closed
Note: See TracTickets for help on using tickets.