3
$\begingroup$

Let $G$ be a semisimple algebraic $\mathbb{Q}$-group and $\Gamma$ an arithmetic subgroup of $G$. In particular $\Gamma$ is finitely generated.

Denote by $\Gamma^{u}$ the set of unipotent elements in $\Gamma$. My main question is the following.

Question 1.: Is the group $\langle \Gamma^{u} \rangle$ generated by $\Gamma^{u}$ also finitely generated?

If the simple factors of $G$ are not of rank $1$, then for example the description of the normal subgroups by Margulis yields a positive answer. However, I am looking for a general answer which preferably doesn't use the latter.

Next, define also the group $N$ generated by the elements of $\Gamma^{u}$ lying in one of the factors of $G$.

Question 2: Is there any general results on the quotient group $\langle \Gamma^{u} \rangle/ N$?

If one defines all the above over $G$ instead of $\Gamma$ then such quotient was investigated and for example is always a finite group.

Assuming the following setup would in fact already suffice for me:

  • $G \cong \prod_{i=1}^q GL_{n_i}(D_i)$ with $D_i$ a finite dimensional division $\mathbb{Q}$-algebra
  • $\Gamma$ a subgroup of finite index in $\prod_{i=1}^q GL_{n_i}(\mathcal{O}_i)$ with $\mathcal{O}_i$ a ($\mathbb{Z}$-)order in $D_i$
$\endgroup$
11
  • 2
    $\begingroup$ For arithmetic Fuchsian groups this is in general false. $\endgroup$ Commented Jul 11, 2023 at 14:08
  • $\begingroup$ Thanks for your comment, however I could not come up with an example. Would it maybe be possible to give one or a reference please? $\endgroup$ Commented Jul 11, 2023 at 14:28
  • 2
    $\begingroup$ Take any arithmetic Fuchsian group with noncompact quotient of positive genus. Then your subgroup will be normal nontrivial of infinite index. Such a subgroup cannot be finitely generated. $\endgroup$ Commented Jul 11, 2023 at 14:38
  • $\begingroup$ math.stackexchange.com/questions/168913/… $\endgroup$ Commented Jul 11, 2023 at 15:10
  • 1
    $\begingroup$ I am not sure about congruence-subgroups, but for arbitrary finite-index subgroups in lattices of $SO(3,1)$ the situation is the same as for Fuchsian groups: Each nonuniform lattice contains a finite-index subgroup such that unipotents generate an normal nontrivial subgroup such that the quotient group is not virtually cyclic. (This is a hard theorem, mostly due to Agol and Wise.) Another nontrivial (but not as hard) theorem ensures that such a subgroup cannot be finitely generated (mostly due to Stallings). $\endgroup$ Commented Jul 18, 2023 at 22:45

0

You must log in to answer this question.

Browse other questions tagged .