Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $K$ be an algebraically closed field. Assume for simplicity that $K$ has characteristic zero. Let $A/K$ be a semiabelian variety. Let $n$ be an integer coprime to $char(K)$. Denote by $\pi_1(A)$ the etale fundamental group of $A$. Is it true that $Hom(\pi_1(A), {\mathbb{Z}}/n)$ is isomorphic to $A[n]^\vee$ (edit)? What happens, if I replace $A$ by an arbitrary connected algebraic group over $K$?

Remark: The answer is ``yes'' in the case where $A/K$ is proper. (The key ingredient to the proof of this fact is the theorem of Serre-Lang, that every finite etale cover $X/A$ carries the structure of an abelian variety, provided $A/K$ itself is an abelian variety.)

share|improve this question
What is $A^\vee[n]$ for a semiabelian variety $A$? –  Mikhail Borovoi Feb 21 '11 at 17:40
Sorry, I probably meant $A[n]^\vee(K)$, where $-^\vee$ in the Cartier dual. I will edit the question tomorrow accordingly. (I only have the mobile phone at the moment) –  Sebastian Petersen Feb 21 '11 at 20:50
OK, what is $A[n]$? Is it the group of points of order dividing $n$? –  Mikhail Borovoi Feb 21 '11 at 21:35
I know you restricted to char. 0, but it's false in char. $p>0$ already for $\mathbb{G}_m$, which has a ton of $p$-covers. –  Felipe Voloch Feb 21 '11 at 21:54
In my notation $A[n]$ stands for the kernel of multiplication with $n$ on $A$. It is a finite etale group scheme over $K$. I sometimes identify a finite etale group scheme over an alg closed field with its points. So I ask whether $Hom(\pi_1(A), Z/n)$ is $Hom(A[n], Z/n)$, where $A[n]$ are the $n$-torsion points. –  Sebastian Petersen Feb 21 '11 at 22:21

1 Answer 1

I can give a partial answer, or at least a strategy towards my question, myself. I hope that this will also put my question into the right context, and increase its chances of being answered completely. (I admit that it was not so well-formulated at the beginning.)

Let $K$ be algebraically closed of characteristic zero (for simplicity). Let $X/K$ be a separated connected algebraic $K$-scheme. Then we have $H^1(X, \mathbb{Z}/n)=Hom(\pi_1(X), \mathbb{Z}/n)$. On the other hand we have an exact sequence $$1\to \Gamma(X, {\cal{O}}_X)^\times/n\to H^1(X, \mathbb{Z}/n)\to H^1(X, {\mathbb{G}}_m)[n]\to 1\ (*),$$ associated to the short exact sequence of etale sheaves $$1\to \mu_n\to \mathbb{G}_m\to \mathbb{G}_m\to 1$$ (and noting that $\mu_n=\mathbb{Z}/n$ because $K$ is algebraically closed).

Furthermore $H^1(X, \mathbb{G}_m)=Pic(X)$, the group of invertible sheaves on $X$. If $X$ is proper in addition, then $\Gamma(X, {\cal{O}}_X)^\times=K^\times$. Hence there is an isomorphism $$Hom(\pi_1(X), \mathbb{Z}/n)=H^1(X, \mathbb{Z}/n)\cong Pic(X)[n]$$ for every connected proper $K$-scheme.

Now consider the special case where $A/K$ is an abelian variety. Then $$Pic(X)[n]=A^\vee[n]=A[n]^\vee=Hom(A[n], \mathbb{Z}/n);$$ hence we obtain in fact a canonical isomorphism $$Hom(\pi_1(A), \mathbb{Z}/n)=Hom(A[n], \mathbb{Z}/n).$$ This is where the "well-known" isomorphism $\pi_1(A)\cong \prod_\ell T_\ell(A)$ comes from. (The fact that $\pi_1(A)$ is abelian has to be shown in addition.)

One also sees that $H^1(A, Z_{\ell})$ is dual to the Tate module $T_\ell(A)$ ($\ell$ a prime number).

Now let $B$ be a semiabelian variety. I asked the above question, because I wanted to know, whether the situation is similar in the case of a semiabelian variety. For example, I wanted to know:

i) Is $Hom(\pi_1(B), \mathbb{Z}/n)$ (canonically) isomorphic to $Hom(B[n], \mathbb{Z}/n)$?

ii) Is there a useful relation between $\pi_1(B)$ and $\prod_\ell T_\ell(B)$, where $T_\ell(B)=lim_i B[\ell^i]$ is the Tate module (defined in a naive way analoguos to the proper case). Are these groups canonically isomorphic?

iii) Is there a useful relation between $H^1(B, Z_\ell)$ and $T_\ell(B)$? Is the first $\mathbb{Z}_\ell$-module canonically isomorphic to the dual of the second?

I think, this case of semiabelian varieties is somewhat different, because $\Gamma(B, {\cal{O}}_B)^\times/n$ does not vanish any more, unless $B$ is proper.

But nevertheless these questions still make sense to me. Further comments / answers are appreciated very much.

share|improve this answer

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.