$\newcommand{\Q}{\Bbb Q} \newcommand{\N}{\Bbb N} \newcommand{\R}{\Bbb R} \newcommand{\Z}{\Bbb Z} \newcommand{\C}{\Bbb C} \newcommand{\F}{\Bbb F} \newcommand{\p}{\mathfrak{p}} $ Let $A$ be an abelian variety over a number field $F$. It is expected that the $L$-function of $A$ has analytic continuation to $\Bbb C$ and satisfies a functional equation relating $s$ to $2-s$. In that setting, the (generalized) Birch–Swinnerton-Dyer conjecture states that $$\mathrm{ord}_{s=1}(L(A_{/F},s)) = \mathrm{rk}_{\Z}(A(F)) =: r.$$

Originally, the conjecture for an elliptic curve $E$ over $\Q$ was $$\exists C>0,\quad \prod_{p \leq x} \dfrac{|E(\F_p)|}{p} \sim C \;\mathrm{log}(x)^r \qquad (x \to \infty).$$

My question is to know what is the analogue of the original conjecture, in the framework of abelian varieties over number fields.

My first guess would to replace to LHS by $$\prod_{N(\p) \leq x} L_{\p}(A_{/F}, N(\p)^{-1}),$$ where $L_{\p}(A_{/F},s)$ is the local factor of the L-function of $A$ at $\p$. But I'm not sure what the RHS should be. Typically, how does it depend on the dimension of $A$ or on the degree of the number field?


1 Answer 1


$\newcommand{\p}{\mathfrak{p}}$By Theorem 6.3 of this paper by Keith Conrad, strong conjectures about $L(A,s)$ (stronger than GRH for this $L$-function, but still "believable"), imply that

$$ \prod_{N\p\le x}L_{\p}(A,N\p^{-1}) \sim C (\log x)^r $$ where $r$ is the order of vanishing at $s=1$, which by the usual BSD should be the rank of $A(K)$.

See also this question about the relation between the original and the modern formulation of BSD.

  • $\begingroup$ Thank you for your answer. To apply theorem 5.11, I would need to check hypothesis (5.7), wouldn't I? How could I check it (assuming GRH, possibly...)? $\endgroup$
    – Watson
    Sep 4, 2018 at 11:43
  • $\begingroup$ @Watson Sorry, I quoted the wrong theorem! $\endgroup$
    – Aurel
    Sep 4, 2018 at 11:44
  • $\begingroup$ Moreover, do you know whether the constant $C$ depend on $A$ (in particular of $\mathrm{dim}(A)$)? $\endgroup$
    – Watson
    Sep 4, 2018 at 11:44
  • $\begingroup$ @Watson Konrad's paper also relates $C$ to the leading term, at $s=1$: this is Theorem 5.11. For the BSD formula for the leading term, see this question $\endgroup$
    – Aurel
    Sep 4, 2018 at 11:47
  • 2
    $\begingroup$ @Watson I want to point out that the mismatch between the coefficient in the (plausible but unproved!) asymptotic estimate for partial Euler products at the normalized central point $s=1/2$ and the leading Taylor coefficient of the $L$-function at that point is already appearing in the case of Hecke $L$-functions $s=1/2$: see Corollary 5.5 and Corollary 5.6. See Example 5.8 for some data in the case of one Dirichlet $L$-function. $\endgroup$
    – KConrad
    Sep 4, 2018 at 13:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.