2
$\begingroup$

Let $K$ be an extension field of $\mathbb{Q}_p$, let $O$ be the ring of integers of $K$, and let $P$ be the maximal ideal of $O$.

If $K$ is a finite extension of $\mathbb{Q}_p$, there is the well-known algebraic isomorphism

$$O[[X]]\cong\varprojlim O[X]/((1+X)^{p^n}-1).$$

A proof of this fact can be found, for example, in Lang's "Cyclotomic fields", in Washington's "Introduction to Cyclotomic Fields", and in Neukirch et al's "Cohomology of number fields". In all of this proofs, at some point one needs a generator of the maximal ideal $M$, that is, a uniformizer element of $K$. Hence, a priori, this proofs do not work if we let $K$ to be a non discretely valued extension.

Now, in page 96 of Lang's "Cyclotomic fields I and II", he claims (and uses) that this isomorphism is true when one let $K=\mathbb{C}_p$, the completion of the algebraic closure of $\mathbb{Q}_p$. My question is: where can I find a proof of this?, or, why is this obviously true?, since he only proves this isomorphism for $K=\mathbb{Q}_p$.

Note: Since I do not know too much about (the foundations of) Iwasawa theory, I do not know if this is an elementary question. If it is, please let me know and I will ask it in MSE.

$\endgroup$
4
  • 2
    $\begingroup$ Why shouldn't the proof work if $O/M$ is infinite? Any discretely valued field has a uniformiser; this is just another name for an element of least positive valuation. $\endgroup$
    – LSpice
    Commented Jun 2, 2018 at 17:04
  • $\begingroup$ @LSpice Thanks for the remark. It was a stupidity of mine. I had a lapsus and confused two different things. I will edit the question. $\endgroup$
    – efs
    Commented Jun 2, 2018 at 17:18
  • $\begingroup$ @LSpice Note that $\mathbb{C}_p$ is not discretely-valued. $\endgroup$ Commented Jun 3, 2018 at 8:47
  • $\begingroup$ @FilippoAlbertoEdoardo, indeed. I did not mean to suggest that the issue did not arise for $\mathbb C_p$, only that it did not come (as the post originally suggested) from the infinitude of $O/M$. $\endgroup$
    – LSpice
    Commented Jun 3, 2018 at 20:31

1 Answer 1

6
$\begingroup$

There is unique division with remainder by a monic polynomial in $O[X]$, where $O$ is any commutative ring. When $O$ is a $p$-adically complete ring, meaning the natural ring homomorphism $O \to \varprojlim O/(p^n)$ is an isomorphism, the Weierstrass division theorem tells us there is unique division with remainder by a polynomial in $O[[X]]$ that is distinguished: monic with lower degree coefficients in the maximal ideal of $O$. Therefore, since $(1+X)^{p^n}-1$ is both monic and distinguished when $O$ is $p$-adically complete (as its non-leading coefficients are all divisible by $p$), the natural composite mapping $O[X] \rightarrow O[[X]] \rightarrow O[[X]]/((1+X)^{p^n}-1)$ is surjective with kernel the ideal $((1+X)^{p^n}-1)$ in $O[X]$, so we have a natural ring isomorphism $$ O[X]/((1+X)^{p^n}-1) \cong O[[X]]/((1+X)^{p^n}-1). $$ This is compatible with the inverse system built on both sides from the ideals $((1+X)^{p^n}-1)$ for $n = 0, 1, 2, \ldots$, so we get an isomorphism of inverse limits $$ \varprojlim O[X]/((1+X)^{p^n}-1) \cong \varprojlim O[[X]]/((1+X)^{p^n}-1). $$ The ring $O[[X]]$ is complete with respect to the $(p,X)$-adic topology and let's show $(1+X)^{p^n}-1 \subset (p,X)^n$. The ideal is $(p^n,p^{n-1}X,\ldots,pX^{n-1},X^n)$, the polynomial is $\sum_{k=1}^{p^n} \binom{p^n}{k}X^k$, and ${\rm ord}_p\binom{p^n}{k} = n - {\rm ord}_p(k) \geq n - k$, so for $1 \leq k \leq n$ we have $\binom{p^n}{k}X^k \in (p^{n-k}X^k)$ and for $k > n$ we have $\binom{p^n}{k}X^k \in (X^n)$, so $(1+X)^{p^n} - 1 \in (p,X)^n$. Thus $(1+X)^{p^n} - 1$ tends to $0$ in the $(p,X)$-adic topology on $O[[X]]$, so $\varprojlim O[[X]]/((1+X)^{p^n}-1) \cong O[[X]]$ for the same reason $\varprojlim \mathbf Z_p/(p^{n_i}) \cong \mathbf Z_p$ for any increasing sequence $n_i$. (Note: the polynomials $(1+X)^{p^n} - 1$ do not tend to $0$ in the $X$-adic topology or in the $p$-adic topology.)

$\endgroup$
2
  • 1
    $\begingroup$ I meant $p$-adically complete: $O \cong \varprojlim O/(p^n)$. I do not want to say $P$-adically complete when $O$ is the integers of $\mathbf C_p$ since that $P$ does not have its powers shrink to $0$: $P^k = P$ for all $k \geq 1$. $\endgroup$
    – KConrad
    Commented Jun 3, 2018 at 14:47
  • $\begingroup$ Thanks. I deleted my questions thinking I was in another thread. Sorry. $\endgroup$
    – efs
    Commented Jun 3, 2018 at 14:49

You must log in to answer this question.

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