2
$\begingroup$

I am considering the following question.

Question 1. Let $G$ be a reductive algebraic group over $\mathbb{R}$, can we find finitely many reductive $\mathbb{R}$-subgroups $H_1,...,H_m$ such that for any reductive $\mathbb{R}$-subgroup $H\subset G$, there is a $g\in G(\mathbb{R})$ such that $H=gH_ig^{-1}$ for some $i$?

We know that if one replace "reductive" in the above question to "semi-simple", then we get a positive answer, see this question.

Unfortunately, the answer to my Question 1 is definitely "No". For example, take $G=\mathbb{G}_{m,\mathbb{R}}^2$, and for each $i\in \mathbb{N}$, define the subtorus $H_i=\{(x,x^i);x\in \mathbb{R}^*\}$, then these $H_i$ are not conjugate to each other.

The reason why we have the above counter-example is that in a torus, conjugation is a trivial action. So in order to get non-trivial conjugations, I am trying to enlarge the ambient algebraic group. I modify my Question 1 as follows.

Question 2. Let $G$ be a reductive algebraic group over $\mathbb{R}$, can we find a faithful representation $\rho: G\rightarrow \mathrm{GL}_{n,\mathbb{R}}$ and finitely many reductive $\mathbb{R}$-subgroups $H_1,...,H_m\subset \mathrm{GL}_{n,\mathbb{R}}$ such that for any reductive $\mathbb{R}$-subgroup $H\subset G$, there is a $g\in \mathrm{GL}_n(\mathbb{R})$ such that $H=gH_ig^{-1}$ for some $i$?

In my new question, the conjugation element $g$ can be chosen more flexibly. So I would like to know whether Question 2 is true. Thanks very much!

$\endgroup$
4
$\begingroup$

No. Again the 2-torus $T$ is a counterexample. Indeed, first an embedding into $\mathrm{GL}_n$ is given by $n$ characters, hence $n$ pairs $(k_1,m_1)$,... $(k_n,m_n)$ of integers such that the corresponding matrix defines a surjective endomorphism $\mathbf{Z}^n\to\mathbf{Z}^2$. Namely it's conjugate to the embedding $$(t,u)\mapsto \mathrm{diag}(t^{k_1}u^{m_1},\dots,t^{k_n}u^{m_n}).$$ For $\ell$, let $T_\ell$ be the subtorus of $T$ of those $(t,u)$ with $u=t^\ell$. Then it maps to $\mathrm{diag}(t^{k_1+\ell m_1},\dots,t^{k_n+\ell m_n})$ [note that for every $\ell$, $\mathrm{gcd}(k_1+\ell m_1,\dots,k_n+\ell m_n)=1$]. For $\ell_1,\ell_2$, the images of $T_{\ell_1}$ and $T_{\ell_2}$ are conjugate in $\mathrm{GL}_n$ if and only if the $n$-tuple $(k_1+\ell_2 m_1,\dots,k_n+\ell_2 m_n)$ is a signed permutation of $(k_1+\ell_1 m_1,\dots,k_n+\ell_1 m_n)$. For each given $\ell_1$ this can hold only for finitely many $\ell_2$, and hence these fall into infinitely many conjugacy classes.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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