Let
be a weight
newform on
with
Nebentypus character
(recall that
is a form on
if and
only if
), and let
be
the corresponding optimal quotient of
, as in [Shi73].
Shimura proved in [Shi94, Ch. 7]
that the local Euler factor of
at
is
If
is odd, standard facts about
formal groups imply that the reduction map
is injective, so
The upper bound given above is the same for every abelian variety isogenous to
,
so it is not surprising that it is not sharp in general. For example,
let
(resp.,
) be
the elliptic curve labeled 30A1 (resp. 30A2)
in Cremona's tables [Cre97]. Then
and
are isogenous,
, and
, so
Consider the abelian variety quotient tex2html_wrap_inline$A_f=A_20$ of tex2html_wrap_inline$J_0(389)$,
as in Section
.
Each tex2html_wrap_inline$G_p(x)$ is a polynomial of degree tex2html_wrap_inline$20$. We
find that tex2html_wrap_inline$G_3(3+1) = 2^2·97 ·2586967$ and
tex2html_wrap_inline$G_5(5+1) = 5^2·97 ·183880478851$.
Thus tex2html_wrap_inline$97$ is an upper bound on tex2html_wrap_inline$#A(Q)_$.
We will see in Example 3.6
that in fact tex2html_wrap_inline$#A(Q)_=97$.