3
$\begingroup$

Let $R$ be a finitely generated $\mathbb{Z}$-algebra with an [edit: linear algebraic] action of $G(\mathbb{Z})$ where $G$ is a split simply-connected semisimple group.

Then for any prime $p$ we have a map $R^{G(\mathbb{Z})} \otimes \mathbb{F}_p \rightarrow (R \otimes \mathbb{F}_p)^{G(\mathbb{F}_p)}$. Is this map necessarily surjective for sufficiently large $p$?

Comments: (1) The simply-connectedness assumption may seem weird; it is made to ensure that $G(\mathbb{Z}) \rightarrow G(\mathbb{F}_p)$ is surjective so that there is a map at all.

(2) If $G$ is a finite group, then the answer is yes by an averaging argument.

(3) If $G$ is unipotent, then the answer is no. For example, take $x \mapsto x+1$ acting on $k[x]$; there are many invariants in positive characteristic (Artin-Schreier covers!).

$\endgroup$
6
  • 1
    $\begingroup$ Do you mean just an abstract action of $G(\mathbb Z)$, or an algebraic action of $G$? $\endgroup$
    – LSpice
    Commented Jul 2, 2020 at 3:56
  • $\begingroup$ Also, how does the averaging in (2) work for orbits of size divisible by $p$? $\endgroup$
    – LSpice
    Commented Jul 2, 2020 at 3:56
  • $\begingroup$ Algebraic action. Regarding (2), I'm still referring to the statement for "sufficiently large $p$". $\endgroup$
    – user125639
    Commented Jul 2, 2020 at 4:16
  • 1
    $\begingroup$ Compare mathoverflow.net/questions/350590/…. $\endgroup$ Commented Jul 2, 2020 at 7:17
  • $\begingroup$ If a reductive group is split over $\mathbb{F}_p$ and acts as such on a module, then a root subgroup fixes a weight vector if and only if the $\mathbb{F}_p$ points of the root subgroup fix the vector. $\endgroup$ Commented Jul 2, 2020 at 7:22

1 Answer 1

5
$\begingroup$

No. Let $G=SL_n$, acting on its defining representation $V$, with $n\geq2$. Let $R=\mathbb{Z}[X_1,\dots,X_n]$ be the obvious $\mathbb{Z}$-form of the ring of polynomial functions on $V$. Let $p$ be a prime. For any $f\in R/pR$ the product over all $g\in SL_n(\mathbb{F}_p)$ of $f\circ g$ is invariant under $SL_n(\mathbb{F}_p)$. But there are no nontrivial $G(\mathbb{Z})$-invariants in $R$ because $G(\mathbb{Z})$ has a Zariski dense orbit in $V$.

$\endgroup$

You must log in to answer this question.

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