Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

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

In this book it is listed as an exercise to compute the index $[\mathbb Q_p:\mathbb Q_p^p]$.

This exercise is appended to a section concerning the structure of unit-group filters, investigating some properties of the homomorphism raising to the $p$-th power. The result is that, when $i<e/p-1$, (resp. when $i>e/p-1$), then $U_i/U_{i+1}$ is isomorphic with $U_{p^i}/U_{p^i+1}$ (resp. $U_{pi}/U_{pi+1}$). Furthermore, when $i=e/{p-1}$ then the kernel of raising to $p$-th power is of order $p$.
Therefore, if $p \neq 2$ then the index $[\mathbb Q_p:\mathbb Q_p^p]$ is $p$; while if $p=2$, then the index is $2\cdot 2=4$. Are there any errors in the above reasoning? Thanks for the attention in any case in advance.

share|cite|improve this question
How could the index possibly be 1? At the very least the valuations on $({\mathbf Q}_p^\times)^p$ are the multiples of $p$ and thus its index in ${\mathbf Q}_p^\times$ is at least $p$. You need to calculate $(1+p{\mathbf Z}_p)^p$ and figure out its index in $1 + p{\mathbf Z}_p$. – KCd Jan 21 '13 at 6:48
@KCd Indeed this is where I am attempting to do without success yet. So it depends not upon whether or not $p=2$? Thanks. – awllower Jan 21 '13 at 8:02
%oreover, can we use the binomial expansion to conclude that [1+Z_p:(1+Z_p)^p] is p? – awllower Jan 21 '13 at 8:06
See…. – KCd Jan 21 '13 at 8:44
@KCd Thanks for the link. So the computation was wrong because of the first factor was excluded. As for the p=2 case, again, it should be counted in the first factor, resulting in the index=4. Is there any other error? – awllower Jan 21 '13 at 15:29
up vote 9 down vote accepted

This exercise is a fun way to write down some stuff I've been doing for my $p$-adic class.

Teichmuller characters. Suppose $a\in{\bf Z}_p$. Then using the binomial theorem,

$$\left|\frac{p\cdot p\cdot p\cdots p}{1\cdot 2\cdot 3\cdots k}p^n\cdot(p^n-1)\cdots(p^n-(k-1))\right|_p\le\frac{1}{p^n}, $$

$$\left|(1+pa)^{p^n}-1\right|_p\le \max_{1\le k\le p^n}\left|{p^n\choose k}p^k\right|_p\le\frac{1}{p^n}\to0,$$

and hence $(1+pa)^{p^n}\to1$ (compare with $t^\epsilon\to1$ as $\epsilon\to0$ in $\bf R$ for $t$ in a nbhd of $1$). If $x\in{\bf Z}_p$, then $x^{p-1}\in1+p{\bf Z}_p$ (reduce mod $p$ and invoke Euler's theorem), hence $(x^{p^n})_{n=0}^\infty$ is Cauchy and has a limit in the $p$-adic integers, since $x^{p^{n+1}}-x^{p^n}=x^{p^n}\left((x^{p-1})^{p^n}-1\right)$. In an ultrametric space, the distance between successive terms tending to $0$ is sufficient to be Cauchy. Since multiplication and hence fixed powers are by definition continuous in a topological ring, $x\mapsto x^p$ is continuous, so

$$\omega(x):=\lim\limits_{n\to\infty}x^{p^n}\implies \omega(x)^p=\left(\lim_{n\to\infty} x^{p^n}\right)^p=\lim_{n\to\infty}x^{p^{n+1}}=\omega(x),$$

in which case we have $\omega(x)^{p-1}=1$ when $\omega(x)$ is a unit. Since $x^p\equiv x\bmod p$, and ${\bf Z}_p\to{\bf Z}/p{\bf Z}$ is a continuous topological ring homomorphism (the latter having discrete topology), $\omega(x)\equiv x\bmod p$.

This tells us the image of units under $\omega$ consists of the $(p-1)$th roots of unity. Choosing any representatives of ${\bf F}_p$ inside ${\bf Z}_p$, this also gives us a homomorphism ${\bf F}_p^\times\to{\bf Z}_p^\times$ whose image consists of the Teichmuller representatives.

Unit groups. $U(n):=({\bf Z}/n{\bf Z})^\times$ (not to be confused with unitary groups). The CRT allows us to classify these in terms of $U(p^r)$ for prime powers $p^r$; we have $U(p^r)\cong C(p-1)\oplus C(p^{r-1})$ for $p>2$ and $U(2^r)\cong C(2)\oplus C(2^{r-2})$ for $r>1$, where by $C(n)$ we mean the cyclic group of order $n$.

Primitive roots. Let $p$ be an odd prime. Since ${\bf F}_p^\times$ is cyclic, there is a generator (in fact there are $\phi(p-1)$ generators). Such primitive roots will always lift to primitive roots modulo arbitrarily high powers of $p$ (see lemma 14 here).

Inverse limits. There is a definition ${\bf Z}_p:=\varprojlim\,{\bf Z}/p^n{\bf Z}$ (see here for a good exposition on projective limits in the category of topological rings). Taking inverse limits commutes with direct sums and taking unit groups, so by CRT we obtain

$$\begin{array}{cl} {\bf Z}_p^\times & \cong\left(\varprojlim\frac{\bf Z}{p^n\bf Z}\right)^\times \\ & \cong\varprojlim\left(\frac{\bf Z}{p^n\bf Z}\right)^\times \\ & \cong\varprojlim\left(\frac{\bf Z}{(p-1)}\oplus\frac{\bf Z}{p^{n-1}\bf Z}\right) \\ & \cong\left(\varprojlim\frac{\bf Z}{(p-1)}\right)\oplus\left(\varprojlim\frac{\bf Z}{p^{n-1}\bf Z}\right) \\ & \cong\frac{\bf Z}{(p-1)}\oplus ({\bf Z}_p,+),\end{array}$$ and similarly ${\bf Z}_2\cong{\bf Z}/(2)\oplus({\bf Z}_2,+)$. Observe also ${\bf Q}_l^\times\cong{\bf Z}\oplus{\bf Z}_l^\times$ for any $l$.

Given a sequence of primitive roots $g_n\in{\bf Z}/p^n{\bf Z}$ such that $g_{n+1}\equiv g_n\bmod p^n$, $g:=\lim\limits_{n\to\infty}g_n$ exists as a $p$-adic integer. Then $\tau=g^{p-1}$ has maximal $p$-power order mod every power of $p$, and is in particular an element of $1+p{\bf Z}_p$. There is an explicit isomorphism, then, given by

$${\bf F}_p^\times\oplus{\bf Z}_p\to {\bf Z}_p^\times:(a,x)\mapsto \omega(a)\lim_{n\to\infty}g^{(x~\bmod~p^n)}.$$

Note further that multiplication on the second coordinate of the domain above induces a fairly accessible ${\bf Z}_p$-module structure on $1+p{\bf Z}_p$, and $g$ is a topological generator.

Therefore, in conclusion,

$$\frac{{\bf Z}_p^\times}{({\bf Z}_p^\times)^n}\cong\frac{\bf Z}{r(n,p)}\oplus\frac{\bf Z}{(p^{v_p(n)})},\quad [{\bf Z}_p^\times:({\bf Z}_p^\times)^n]=r(n,p)p^{v_p(n)}$$

$$\frac{{\bf Q}_p^\times}{({\bf Q}_p^\times)^n}\cong\frac{\bf Z}{(n)}\oplus \frac{{\bf Z}_p^\times}{({\bf Z}_p^\times)^n},\quad [{\bf Q}_p^\times:({\bf Q}_p^\times)^n]=n[{\bf Z}_p^\times:({\bf Z}_p^\times)^n],$$

where $\displaystyle r(n,p):=\frac{p-1}{(n,p-1)}$ if $p>2$ and $\displaystyle\frac{2}{(n,2)}$ if $p=2$.

share|cite|improve this answer
Thanks for this particularly interesting answer. It expounds the structure of p-adic numbers in a satisfying manner, using Teichmüller representatives. There is only one small problem: your link (1) appears to miss. It would be appreciated if you supplement the link. Thanks very much. – awllower Feb 13 '13 at 3:37
@awllower The original link was to "p-adic integration and the theory of groups," which had a good couple sections on the topic that I remembered fondly. The system seems to have eaten the link twice, perhaps due to it being a .ps document?, so I have converted it to pdf and uploaded it elsewhere. Another comprehensive resource that I think looks good (and very comprehensive, though focused on groups instead of rings), is Profinite Groups by Ribes and Zalesskii. – anon Feb 13 '13 at 3:53
Thanks very much.I am still processing the content, though. – awllower Feb 13 '13 at 3:58
(-1) for using \mathbf instead \mathbb; but (+2) for knowing all this... :-) – Asaf Karagila Feb 21 '13 at 18:24

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.