Unexplained error in mac_lane_step

[swewers]

The following gives an error message:

sage: from mclf import sage: R. = QQ[] sage: v_2 = QQ.valuation(2) sage: v = GaussValuation(R, v_2).augmentation(x+1, 1/2) sage: f = x^4 - 30x^2 - 75 sage: v.is_equivalence_unit(f) sage: v.mac_lane_step(f) False

AssertionError Traceback (most recent call last)

in () 5 f = x**Integer(4) - Integer(30)*x**Integer(2) - Integer(75) 6 v.is_equivalence_unit(f) ----> 7 v.mac_lane_step(f)

/home/stefan/comp/SageMath/local/lib/python2.7/site-packages/sage/rings/valuation/inductive_valuation.pyc in mac_lane_step(self, G, principal_part_bound, assume_squarefree, assume_equivalence_irreducible, report_degree_bounds_and_caches, coefficients, valuations, check) 843 ret.append((w, degree_bound, multiplicities[slope], w_coefficients, new_valuations)) 844 --> 845 assert ret 846 if not report_degree_bounds_andcaches: 847 ret = [v for v,,,,_ in ret]

AssertionError: sage:

[saraedum]

Ok, my code does not assume that somebody would set the valuation of a key to the "wrong" value .augmentation(x+1, 1/2). The "correct" value would have been 3/4. In other words, I only tested mac_lane_step starting from Gauss valuations.

[saraedum]

This is now https://trac.sagemath.org/ticket/26066.

[saraedum]

Should we monkey patch this?

[swewers]

This would be good. The error occurs only very rarely (otherwise I would have found it already), but the option of using mac_lane_step starting from the "wrong value" is used a lot in mclf.

[swewers]

Thanks!

[swewers]

I wrote a positive review, but again, there seems to be a problem with the patchbot, unrelated with the ticket.

[saraedum]

This is now in the latest 8.4 beta.