I have known for a finite extension $E\supseteq Q_p$, the ring of integers $O_E$ is a profinite group as an additive group. And $O_E^\times$ is also profinite as a multiplicative group.

Questions: For the p-adic complex field $C_p=\widehat{\overline{Q_p}}$, is $O_{C_p}$ a profinite group or compact? And what about ${O_{C_p}}^\times$ ?

Thanks in advance!


Hint: The maximal ideal of $O_{\mathbb C_p}$ is $m=\lbrace x \in O_{\mathbb C_p}: \lvert x \rvert_p < 1 \rbrace$. Is $m$ open? How many elements does the quotient $O_{\mathbb C_p} /m$ have? If you answer both questions, you should see a covering of $O_{\mathbb C_p}$ by (actually: a partition of $O_{\mathbb C_p}$ into) infinitely many, mutually disjoint, open sets.

Try an analogous idea for $O_{\mathbb C_p}^\times$ using the multiplicative subgroup $1+m$.

Added: Here's an entirely different argument. One knows that the $p$-adic value on $\mathbb C_p$, normed to $\lvert p \rvert_p=p^{-1}$, has non-discrete value group $\lvert \mathbb C_p^* \rvert_p=p^\mathbb Q$. Now choose a bounded (!) sequence of rationals $(r_n)_n$, all $r_n \le 0$, such that $r_n \neq r_m$ for all $n \neq m$. For each $n$, pick $x_n \in O_{\mathbb C_p}$ with $\lvert x_n \rvert_p = p^{r_n}$. Then show that the sequence $(x_n)_n$ has no convergent subsequence, using that for a sequence $(y_k)_k$ in an ultrametric field to converge to something of absolute value $\lvert y \rvert \neq 0$, one needs $\lvert y_k \rvert = \lvert y\rvert$ for high enough $k$.

Slight generalisations of these two arguments will show that for a field $K$ with non-archimedean valuation $\lvert \cdot\rvert$ to be locally compact, it is necessary that

  • the residue field is finite, and
  • the valuation is discrete (meaning, its value group $\lvert K^* \rvert$ is).

Both conditions are not satisfied for $\mathbb C_p$, and it's a good exercise to come up with many other fields which fail either, or both. -- An obvious third condition, which is met by $\mathbb C_p$, is that

  • the field is complete.

Conversely, if a field with a non-archimedean valuation satisfies all three conditions above, it is locally compact. This is another good exercise; and a last good exercise is to show that such a field is necessarily a finite extension of some $\mathbb Q_p$ or $\mathbb F_p((T))$, so that with all exercises combined, we have proven the non-archimedean part of the theorem mentioned in Lubin's answer. (I am quite sure I have seen this as a combined exercise somewhere in Schikhof's Ultrametric Calculus).

  • $\begingroup$ Thank you, I think I can solve my problems with your hints now. $\endgroup$ – Sssss Jul 2 at 4:42

The answer of @TorstenSchoeneberg is the right one, because it argues from first principles, without calling in any advanced theorem. I take the opposite tack below:

If $\mathcal O_{\Bbb C_p}$ were compact, then $\Bbb C_p$ would be locally compact. But it’s a “well-known” theorem that the only locally compact fields of characteristic zero are $\Bbb C$, $\Bbb R$, and the fields $\Bbb Q_p$ and their finite extensions.

  • $\begingroup$ Yes, you are right, so $O_{C_p}$ is not compact. I remember the well-known theorem is proved in 《Fourier Analysis on Number Fields》by Ramakrishnan or 《Basic Number Theory》by Weil. But for $O_{C_p}^\times$, is there another method to prove the non-compactness? Thanks for your answers! $\endgroup$ – Sssss Jul 2 at 5:19
  • $\begingroup$ Good point to put it in context of course. I've added that to my answer. $\endgroup$ – Torsten Schoeneberg Jul 2 at 16:46

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.