accuracy problem in matrix_mod_pn for Eisenstein extensions of padics
Currently, the accuracy of the matrix returned by matrix_mod_pn()
in padic extension rings is probably too small:
sage: R=ZpCA(3,5)
sage: S.<a> = R[]
sage: L.<a>=R.extension(a^23)
sage: t=a.add_bigoh(2)
sage: t.matrix_mod_pn()
[0 1]
[0 0]
sage: (t*a)._ntl_rep_abs() # but t*a is not zero
([3], 0)
I think it would be better to return the following:
[0 1]
[3 0]
Of course, the "1" is not known to precision O(3^2)
but I think it is better to leave it to the caller to strip away the digits that are not wanted.
The attached patch increases the precision to the maximal precision of any of the entries of the matrix.
Change History (7)
Milestone: 
sage5.11 →
sage5.12

Milestone: 
sage6.1 →
sage6.2

Milestone: 
sage6.2 →
sage6.3

Milestone: 
sage6.3 →
sage6.4

Branch: 
→ u/saraedum/accuracy_problem_in_matrix_mod_pn_for_eisenstein_extensions_of_padics

Commit: 
→ 956f12477e66fa1f2568e910e1f1c6a97d2d7783

Keywords: 
days71 added

New commits:
Trac #13651: fixed a precision problem in _internal_lshift(), _ntl_rep_abs() of padics
Merge branch 'develop' into t/13651/ticket/13651
http://trac.sagemath.org/rawattachment/ticket/13659/trac_13659.patch
Trac #13661: improved precision of matrix_mod_pn() and implemented matrix() for padics