MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Sorry if this is an easy one, I'm a little rusty on my group theory. My first guess was that it's simply the inverse limit of the Aut($\mathbb{Z}/p^i\mathbb{Z})$, with the map when $i\leq j$ given by taking $\sigma\in$ Aut$(\mathbb{Z}/p^j\mathbb{Z})$ to the map $\tilde{\sigma}:\mathbb{Z}/p^i\mathbb{Z}\rightarrow\mathbb{Z}/p^i\mathbb{Z}$ defined by solving $\phi\circ\sigma=\tilde{\sigma}\circ\phi$ where $\phi:\mathbb{Z}/p^j\mathbb{Z}\rightarrow\mathbb{Z}/p^i\mathbb{Z}$ is the reduction map, but that seems too optimistic - I couldn't think of any reason $\tilde{\sigma}$ would be well-defined, much less be an automorphism of $\mathbb{Z}/p^i\mathbb{Z}$.

Also, barring a full answer to my question, I would be interested in whether Aut$(\mathbb{Z}_p)$ is a $p$-group. If not, what can we say about the elements $\sigma\in$ Aut$(\mathbb{Z}_p)$ with order a power of $p$?

share|cite|improve this question
up vote 8 down vote accepted

First the $p$-adic integers are finitely generated (actually cyclic) pro-$p$ group therefore from a result of Serre all automorphisms are continuous. Now as it cyclic it is enough to see what happens to $1$. It has to go to another generator, i.e. any element of the form $a_0+a_1p+a_2p^2+\cdots$, where $0 \leq a_i < p$ for all $i$ and $a_0 \ne 0$. Hence, every element in $\mathbb{Z}_p$ is just multiplied by $a_0+a_1p+a_2p^2+\cdots$. Thus, the automorphism group is the multiplicative group of $\mathbb{Z}_p$.

It is of course not a $p$-group, but it contains a subgroup of index $p-1$ which is pro-$p$ and is actually again a cyclic pro-$p$ group, i.e. isomorphic to $\mathbb{Z}_p$ and thus have no elements of finite order.

share|cite|improve this answer
I'm not sure exactly which result of Serre you're referring to, but this is an elementary question so let's be more explicit. Let $G$ be a profinite group and let $f$ be an automorphism of the underlying abstract group. Then $f$ is a homeomorphism iff it is continuous iff $f^{-1}$ is continuous iff $f$ maps open subgroups to open subgroups. Since every group automorphism preserves the index of subgroups and any open subgroup has finite index, a suficient condition for every group automorphism of $G$ to be a homeo is that every finite index subgroup of $G$ is open. This is the case here... – Pete L. Clark Jul 5 '10 at 3:09
Continued: it was rather recently proven that any finitely generated profinite group has the property that every finite index subgroup is open. I believe this has been known for pro-$p$-groups for a long time: that's probably related to the result of Serre that you're referring to. – Pete L. Clark Jul 5 '10 at 3:11
Indeed, Serre proved that for a finitely generated pro-$p$ group a subgroup of finite index is open. Thus, the topology is determined by the algebra. Hence, every automorphism must be continuous. The questions whether every subgroup of finite index of a finitely generated profinite group is open was known as Serre's problem. It was fairly recently proved to have an affirmative answer by Nik Nikolov and Dan Segal. – Yiftach Barnea Jul 5 '10 at 7:21
Surely the continuity of group automorphisms is obvious for $\mathbb Z_p$: the subgroups $p^n \mathbb Z_p$ are basic open neighbourhoods of the identity, and they are preserved by group homomorphisms. – fherzig Jul 5 '10 at 8:16
fherzig, yes you are right the case of $\mathbb{Z}_p$ is triviel. I am not sure what is the origin of Serre's argument, I think it was in some letter, but I have a terrible memory so don't trust me too much. Luis Ribes knows the history of the problem, so if you are interested you can ask him. – Yiftach Barnea Jul 5 '10 at 9:47

Any automorphism of $\mathbb{Z}_p$ preserves whether an element is divisible by $p^k$, so it is Lipschitz (in particular, continuous) with respect to the $p$-adic norm. On the other hand, any automorphism must preserve $\mathbb{Z}$, which is dense in $\mathbb{Z}_p$.

What I should've said is that any automorphism is determined by its behavior on $\mathbb{Z}$, hence by its behavior on $1$. To make up for that mistake, let me offer a sketch of the description of the structure of $\mathbb{Z}_p^{\ast}$. This group clearly splits up as the direct product of $(\mathbb{Z}/p\mathbb{Z})^{*}$ and the multiplicative group $U = 1 + p \mathbb{Z}_p$. It is now an interesting exercise to show that the exponential map $x \mapsto u^x, x \in \mathbb{Z}, u \in U$ extends to a map from $\mathbb{Z}_p$ to $U$ which, given the right choice of $u$, is an isomorphism for $p > 2$. For $p = 2$, Yiftach Barnea's otherwise excellent answer is slightly wrong and $U$ is in fact isomorphic to $\{ \pm 1 \} \times \mathbb{Z}_2$.

share|cite|improve this answer
The part of the answer about preserving Z seems to refer to automorphisms of the ring of p-adic integers, whereas the (title of the) question asks about automorphisms of the additive group. It seems to me that multiplication by any unit in Z_p is an automorphism of the additive group, and that these will be the only automorphisms. – Andreas Blass Jul 5 '10 at 0:28
Thanks for pointing out my mistake. I always disliked even primes they are just troubles! – Yiftach Barnea Jul 5 '10 at 9:39
Speaking of which:… – Qiaochu Yuan Jul 5 '10 at 10:04

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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