Given a line bundle $L$ on an abelian variety $A/k$, there is an associated Weil pairing $e_L\colon\bigwedge^2V_pA\to\mathbb Q_p(1)$, where $p$ is a prime different from the residue characteristic of the base field $k$ and $V_pA$ is the $\mathbb Q_p$-linear Tate module of $A$. This is usually constructed by explicitly constructing a pairing between $A[p^n]$ and $A^\vee[p^n]$ using the interpretation of the latter as classes of divisors, and then pulling back along the polarisation induced by $L$.

There is, however, another way to construct such a pairing, using the fact that $V_pA$ is dual to the etale cohomology $\mathrm H^1_{et}(A_{\bar k},\mathbb Q_p)$. Namely, the first etale Chern class $c^{et}_1(L)$ is an element of $\mathrm H^2_{et}(A_{\bar k},\mathbb Q_p)(1)=\mathrm{Hom}(\bigwedge^2V_pA,\mathbb Q(1))$, and we can just take the pairing corresponding to this element.

What I want to know (and ideally would like a reference for) is whether these two pairings are the same. In other words, is the element of $\mathrm H^2_{et}(A_{\bar k},\mathbb Q_p)(1)$ corresponding to the Weil pairing equal to the first etale Chern class of $L$?


1 Answer 1


It seems that one of the pairings is the negative of the other (in char 0 this assertion is actually Lemma 2.6 of https://arxiv.org/pdf/1809.01440.pdf ).

  • $\begingroup$ Great! The minus sign that intervenes here is a good cautionary tale: not all naturally-defined pairings are the same as one another. In any case, the instance I care about is in characteristic zero, so I will accept this answer. $\endgroup$ Commented May 15, 2021 at 15:34
  • $\begingroup$ This is the same minus sign as that in Section 24, Theorem 1, of Mumford's book on Abelian Varieties (over the complex numbers). $\endgroup$
    – user166831
    Commented May 16, 2021 at 12:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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