Re: Pari Bug?

[Ariel Pacetti <apacetti@math.utexas.edu>]

Actually the output came from "algdep". It is ok for lindep to answer a
real number (with the documentation, I apologice for not reading it), but
should algdep answer a real number?
Let me ask another question. While trying to run "elllseries" the
algorithm doesn't reduce the curve automatically hence the convergence may
be pretty bad. For example:

E121D = [0,1,1,-8,19]

but also

E121D' = [0, 55, 3025, -121000, 34439625]

while computing the L-series values at 1 it answers:

? elllseries(ellinit([0,1,1,-8,19]),1)
%1 = 2.838038282044296194964667434
? elllseries(ellinit([0, 55, 3025, -121000, 34439625]),1)
%2 = 2.835178739088127435329760488

While reducing the curve:

? ellglobalred(ellinit([0, 55, 3025, -121000, 34439625]))
%3 = [275, [11, 22, 0, -847], 1]

which gives the right change of variables. Shouldn't elllseries reduce the
curve first and then compute the L-value so as to make the convergence
better?

Ariel