7
$\begingroup$

Let $p:X\rightarrow Y$ be a double cover of curves, denote by $$SU_n:=(p_*SL_n(\mathcal O_X))^{\tilde{\sigma}}$$ i.e. the $\tilde{\sigma}-$invariant part, the action of $\tilde{\sigma}$ is given by $$\tilde{\sigma}(g)=\,^t(g\circ\sigma)^{-1}$$ where $\sigma$ is the involution induced by the double cover. $SU_n$ is well knowing to be a parahoric group shceme in the sens of Bruhat-Tits.

My question: What is $\pi_1(({SU_n})_\eta)$ ? (the algebraic fundamental group). where $\eta$ is the generic point of $Y$.

Thanks

$\endgroup$
2
  • $\begingroup$ Do you really mean to take $g \circ \sigma$ and not $\sigma(g)$? I don't see why $\sigma$ should lie in $p_* SL_n$. $\endgroup$
    – Will Sawin
    Nov 23, 2015 at 17:24
  • $\begingroup$ I see $g$ as an automorphism of $SL_r\times X$ and that's why I used the notation $g\circ \sigma$, $\endgroup$
    – Z.A.Z.Z
    Nov 23, 2015 at 17:45

1 Answer 1

2
$\begingroup$

There is an exact sequence

$$ \pi_1 ((SU_n)_{\overline{\eta}}) \to \pi_1((SU_n)_{\eta}) \to \operatorname{Gal}(\overline{\eta}|\eta)$$

Over $\overline{\eta}$, $p_* SL_n$ is just $SL_n \times SL_n$. The involution $\tilde{\sigma}$ acts by switching the two $SL_n$s and then doing an performing some automorphism, so the involution invariants $SU_n$ are just the diagonal $SL_n$ embedded by the graph of the automorphism.

In characteristic $0$, $\pi_1(SL_n)$ is trivial, so the fundamental group is just the Galois group of the base field. In characteristic $p$ things are more complicated as the fundamental group is infinite, as witnessed by the family of etale covers $g \to g^{-1} \operatorname{Frob}_q(g)$ which is Galois of group $SL_n(\mathbb F_q)$ for each $q$.

$\endgroup$
8
  • $\begingroup$ I don't see why over $\bar \eta$ the fiber shoud be two copies of $SL_r$? what is $X_{\bar\eta}$? $\endgroup$
    – Z.A.Z.Z
    Nov 23, 2015 at 18:01
  • 1
    $\begingroup$ @Z.h $X_{\overline{\eta}}$ is just the pullback of $X$ along the map $\overline{\eta} \to X$, which is just $\overline{\eta}$ - a single point. $Y_{\overline{\eta}}$ is the pullback of $Y$ along the same map, which is two points. Thus the pushforward of $SL_r$ is two cope of $SL_r$. $\endgroup$
    – Will Sawin
    Nov 24, 2015 at 1:48
  • $\begingroup$ I considred the 2:1 cover as $\pi:X\rightarrow Y$, so I think you mean $Y_{\bar\eta}$ is one point and $X_{\bar\eta}$ is two points? right? $\endgroup$
    – Z.A.Z.Z
    Nov 24, 2015 at 19:14
  • 1
    $\begingroup$ @Z.h yes, precisely that. $\endgroup$
    – Will Sawin
    Nov 24, 2015 at 21:34
  • 1
    $\begingroup$ @Z.h For the other map, use the nontrivial automorphism of $K(X)$ over $K(Y)$ and then apply the embedding into $\overline{K(X)}$. Alternately, use that $X_{\overline{\eta} } = \operatorname{Spec} K(X) \otimes_{K(Y)} \overline{K(X)} = \operatorname{Spec} \overline{K(X)} \times \overline{K(X)}$ $\endgroup$
    – Will Sawin
    Nov 28, 2015 at 13:08

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.