MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $A$ an abelian variety over a field $k$ and $A^{*}$ the dual abelian variety.

How can we relate the étale cohomology of $A$ with etale cohomology of $A^{*}$?

share|cite|improve this question
up vote 4 down vote accepted

The Weil pairing induces a perfect pairing $T_\ell A \times T_\ell A^\vee \to \mathbf{Z}_\ell(1)$, and $T_\ell A = H^1_{et}(\bar{A},\mathbf{Z}_\ell)^\vee$ (here, $\bar{A}$ denotes $A \times_k \bar{k}$). Further, $H^i_{et}(\bar{A},\mathbf{Z}_\ell) = \wedge^iH^1_{et}(\bar{A},\mathbf{Z}_\ell)$. One also has $A^\vee = \mathcal{E}\mathrm{xt}_k^1(A,\mathbf{G}_m)$.

share|cite|improve this answer
Perhaps it should be said that $\overline{A}$ means $A_{\overline{k}}$ (and similarly in Joel's answer below, one has to take cohomology of the geometric fiber). – user28172 May 30 '13 at 14:39
Yes, thank you. I have added this. – Timo Keller May 30 '13 at 15:40

Let me add something to what Timo Keller says rightly. What he said can be reformulated as $H^i_{et}(A,{\bf Z_\ell})$ and $H^i_{et}(A^*,{\bf Z_\ell})(1)$ are canonically dual of each other. This should be understand that there are dual not only as $\bf Z_\ell$-modules, but also as $Gal(\bar k / k)$-representation, and here the (1) takes its significance: it is a twist by the $\ell$-adic cyclotomic character.

Now, more can be said. Every abelian variety $A$ over $k$ as a polarization $p$ over $k$, which is in particular a symmetric isogeny $p: A \rightarrow A^\ast$. Now any isogeny induces isomorphism between the $V_\ell$'s (that is $T_\ell \otimes \bf Q_\ell$), hence taking again the wedge power, we get for all $i$ an isomorphism $$H^i_{et}(A,{\bf Z_\ell}) \otimes {\bf Q_\ell} \rightarrow H^i_{et}(A^*,\bf Z_\ell) \otimes \bf Q_\ell.$$ This isomorphism preserves other structures (the $Gal(\bar k/k)$-action), but is not canonical, since it depends on the choice of $p$. But this is certainly a relation between the stale cohomology of $A$ and $A^\ast$ that is worth mentioning, and that is logically independent of the one given by Timo.

share|cite|improve this answer
Addition: Moreover, when $i=1$, this isomorphism is alternate, in the sense that the Weil pairing $\langle\cdot,\cdot\rangle$ induces a pairing $(a,b)\mapsto \langle a,p(b)\rangle$ which is alternate. Hence the Galois group $\mathop{\rm Gal}(\bar k/k)$ acts on $H^1_{\rm et}(A,\mathbf Q_\ell)$ by symplectic automorphisms. – ACL May 30 '13 at 19:43

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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