2
$\begingroup$

I was reading the proof of Lemma 10.12 in this paper. In the second sentence, the following fact is used implicitly:

Let $(R,\mathfrak{m})$ be a commutative local ring. Let $\widehat{R}$ be its $\mathfrak{m}$-adic completion, which is a local ring whose maximal ideal we denote by $\widehat{\mathfrak{m}}$. If $\widehat{\mathfrak{m}}$ is finitely generated, then $\widehat{R}$ is noetherian.

The first idea one could have is that in a local ring, it suffices to check that the maximal ideal is finitely generated to see that the ring is noetherian. However, this is not true, as the following example due to Clark shows. One could try to prove this under the stronger assumption that the local ring is complete with respect to its maximal ideal. However, we also don't know if $\widehat{R}$ is $\widehat{\mathfrak{m}}$-adically complete, for $\mathfrak{m}$ may not be finitely generated.

Does anyone know a proof for the claim in the box (in case it is true)?

Thank you for reading, any help will be highly appreciated :)

Remark. As Keerthi Madapusi recalls and follows from part of the proof of the Lemma, the vector space $\mathfrak{m}/\mathfrak{m}^2$ is finite-dimensional if and only if $\widehat{\mathfrak{m}}$ is finitely generated.

$\endgroup$
7
  • $\begingroup$ The vector space begin finite dimensional is equivalent (under completeness) to the maximal ideal being finitely generated. As for your question, see here: stacks.math.columbia.edu/tag/05GH $\endgroup$ Commented Sep 30 at 9:15
  • 1
    $\begingroup$ Hi, @KeerthiMadapusi . I do agree that $\mathfrak{m}/\mathfrak{m}^2$ is finite-dimensional if and only if $\widehat{\mathfrak{m}}$ is finitely generated (if that is what you meant), so the last remark is unnecessary. However, I don't see how the linked result solves the problem, for $\mathfrak{m}$ is not finitely generated, and the $\widehat{\mathfrak{m}}$-adic completion of $\widehat{R}$ is not necessarily $\widehat{R}$. $\endgroup$
    – Don
    Commented Sep 30 at 10:49
  • $\begingroup$ Apply Corollary 10.25, p. 113 of Atiyah -- MacDonald, "Introduction to commutative algebra." $\endgroup$ Commented Sep 30 at 12:42
  • 2
    $\begingroup$ Dear @JasonStarr . I'd say that we cannot apply this result because neither $R$ nor $\widehat{R}$ are complete wrt $\mathfrak{m}$ and $\widehat{\mathfrak{m}}$. $\endgroup$
    – Don
    Commented Sep 30 at 13:04
  • 1
    $\begingroup$ @JasonStarr Look at stacks.math.columbia.edu/tag/05JA for the subtlety here. Of course in that case $\mathfrak m/\mathfrak m^2$ is not finitely generated. $\endgroup$
    – Will Sawin
    Commented Sep 30 at 13:48

1 Answer 1

4
$\begingroup$

We can just follow the proof from the stacks tag linked from the one suggested by Keerthis Madapusi but under your hypothesis.

Let $a_1,\dots,a_r$ in $R$ generate $\mathfrak m/\mathfrak m^2$. Then the degree $k$ monomials in $a_1,\dots,a_r$ generate $\mathfrak m^k/\mathfrak m^{k+1}$.

Hence the homomorphism $R^{ \binom{k+r-1}{r-1}}\to \mathfrak m^k$ that sends each generator to a different monomial is surjective modulo $\mathfrak m$. Thus by the same lemma cited in the stacks tag above, except part (1) this time, the induced map on completions is surjective. The completion of $\mathfrak m^k$ is the kernel of $\hat{R} \to R/\mathfrak m^k$ and so the kernel of $\hat{R} \to R/\mathfrak m^k$ is generated by degree $k$ monomials in $a_1,\dots,a_r$ and thus is equal to $\hat{\mathfrak m}^k$ (of course it contains $\hat{\mathfrak m}^k$, and this shows it is generated by members of $\hat{\mathfrak m}^k$). So the limit topology is the $\hat{\mathfrak m}$-adic topology and thus $\hat{R}$ is complete for the $\hat{\mathfrak m}$-adic topology.

A comment on the tag references [Fujiwara. K and Kato. F, The foundation of Rigid Geometry I, Def. 7.2.6 and Prop. 7.2.7] as giving some kind of more general statement. I didn't check, but it's possible that general version is also what you need.

$\endgroup$
1
  • 1
    $\begingroup$ Dear Will, thank you for your answer, I really appreciate it. It looks correct to me, although I'll go through it in depth to avoid further misconceptions (nonnoetherian rings are delicate, indeed...). $\endgroup$
    – Don
    Commented Sep 30 at 16:46

You must log in to answer this question.

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