4
$\begingroup$

We can associate two $\mathbb Q_p$ vector spaces to a $p$-divisible group, and I'm a little confused about the relation between these two groups. First of all, I think part of my problem is that when papers discuss these things, they claim that if $R$ is $p$-complete, sending $x\mapsto p^nx$ is topologically nilpotent on $G(R)$, in particular, it is nilpotent if $p^nR=0$. I don't understand why this is true and I would appreciate an explanation or a reference.

But if we accept the claim it is easy to see that for any connected $R$ such that $p^nR=0$, we have: $$\tilde{G}(R):=\lim_{p:G\to G}(G(R))=\operatorname{Hom}(\mathbb{Q}_p/\mathbb{Z}_p,G(R))[1/p].$$ On the other hand one has $$T_G(R):=\lim(G[n](R))=\operatorname{Hom}(\mathbb{Q}_p/\mathbb{Z}_p,G(R)).$$ So it seems to me that if $p^nR=0$, then $\tilde{G}(R)$ is just the rational Tate module, and they are different only in mixed characteristic. Is my understanding correct?

$\endgroup$
4
  • $\begingroup$ Isn't a $p$-divisible group a forward limit of finite $p$-group schemes? So then multiplication by $p^n$ is nilpotent on each $p$-group scheme and thus topologically nilpotent on the whole group? $\endgroup$
    – Will Sawin
    Commented Oct 9, 2021 at 21:19
  • $\begingroup$ @WillSawin I understand it is topologically nilpotent in this sense(note that the R-points of this p-divisible group is not the limit of the R-points of those finite group, you need a sheafification.) my question was that if $R$ is a p-adic ring. why multiplication by p is topologically nilpotent on $G(R)$ with $p$- adic topology.(where a base is given by $Ker(G(R)\to G(R/p^n))$). $\endgroup$
    – ali
    Commented Oct 10, 2021 at 7:35
  • $\begingroup$ Are we assuming that $R$ mod $p$ is finite? $\endgroup$
    – Will Sawin
    Commented Oct 10, 2021 at 13:49
  • $\begingroup$ @WillSawin No. At least in the context I have seen it( perfectoid space )R mod p) is not even notherian $\endgroup$
    – ali
    Commented Oct 10, 2021 at 14:15

1 Answer 1

1
$\begingroup$

Let $G$ be a $p$-divisible group over any base $S$. In terms of the functor of points we have for any affine scheme $\mathrm{Spec}(R)$ that $G(R):=\mathrm{colim}_n G[p^n](R)$.

Regarding the universal cover etc, let $C(G)=\lim_p G$ and $T(G)=\lim_{p} G[p^n]$. Then you have a short exact sequence (of fpqc sheaves say): $$0\to T(G)\to C(G)\to G\to 0.$$ Then $\mathrm{colim}_p G=\mathrm{colim}_n \mathrm{colim}_p G[p^n]=0$ hence taking the colimit of this short exact sequence under multiplication by $p$ yields an isomorphism $\mathrm{colim}_p T(G)=T(G)\otimes_{\mathbf{Z}_p} \mathbf{Q}_p \stackrel{\sim}{\to} C(G)$.

Regarding topological nilpotence of $p$ on $G$, if $R=\lim_n R/p^n$ is complete then $G(R)=\lim_n G(R/p^n)=\lim_n \mathrm{colim}_m G[p^m](R/p^n)$ and the topology on $G(R)$ is the inverse limit topology where $G(R/p^n)$ has the discrete topology. But $G(R/p^n)=\mathrm{colim} G[p^m](R/p^n)$ is $p$-power torsion and so $p$ is nilpotent on $G(R/p^n)$ and topologically nilpotent on $G(R)=\lim_n G(R/p^n)$.

$\endgroup$

You must log in to answer this question.

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