5
$\begingroup$

Does there exist a reductive group $G$ of type $E_7$ over a given totally real field $L^+$ such that for every embedding $\tau:L^+\to \mathbb{R}$ except one , $G_{\tau,\mathbb{R}}(\mathbb{R})$ is compact? Is there a way to construct a group with a given number of compact forms?

$\endgroup$
4
  • 2
    $\begingroup$ Yes such a group should exist. For example it can be constructed as an inner form of the split adjoint group $G$ of type $E_7$ using the short exact sequence of pointed sets $H^1(L^+, G) \to \bigoplus_{v} H^1(L^+_{v},G) \to \pi_1(G)_{\Gamma_{L^+}}$. Here $\pi_1(G)$ is Borovoi's algebraic fundamental group and $\Gamma_{L^+}$ is the absolute Galois group of $L^+$.... $\endgroup$ Commented 2 days ago
  • 2
    $\begingroup$ ... We need to show that there is a class in $H^1(L^+,G)$ whose image in $H^1(L^+_{v},G)$ for infinite places $v$ of $L^+$ corresponds to the compact inner form of $G_{\mathbb{R}}$ for all but one $v$. This can always be done using the short exact sequence abpve, by choosing nontrivial classes at some auxiliary finite places. In fact, a single finite place should be enough since I think that $\pi_1(G)=\mathbb{Z}/2\mathbb{Z}$ with trivial Galois action. $\endgroup$ Commented 2 days ago
  • $\begingroup$ Thank you very much. $\endgroup$
    – ALi1373
    Commented 2 days ago
  • 2
    $\begingroup$ @PolvanHoften, of course it would be more stable, here, if you wrote your excellent "comments" as "an answer". :) $\endgroup$ Commented yesterday

1 Answer 1

2
$\begingroup$

I'm posting my comments here as an answer.

Yes such a group should exist. For example it can be constructed as an inner form of the split adjoint group $G$ of type $E_7$ using the short exact sequence of pointed sets

\begin{equation} H^1(L^+,G) \to \bigoplus_v H^1(L^+_{v},G) \to \pi_1(G)_{\Gamma_{L^+}}. \end{equation}

Here $\pi_1(G)$ is Borovoi's algebraic fundamental group and $\Gamma_{L^+}$ is the absolute Galois group of $L^+$.

We need to show that there is a class in $H^1(L^+,G)$ whose image in $H^1(L^+_v,G)$ for infinite places $v$ of $L^+$ corresponds to the compact inner form of $G_{\mathbb{R}}$ for all but one $v$. This can always be done using the short exact sequence above, by choosing nontrivial classes at some auxiliary finite places. In fact, a single finite place should be enough since I think that $\pi_1(G)=\mathbb{Z}/2\mathbb{Z}$ with trivial Galois action.

In the same way, one can construct an inner form of $G$ which is compact at a chosen set of infinite places.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .