1
$\begingroup$

Let $R$ be a Dedekind domain. $A$ be a finitely generated R-module. Then $A = A_1 ⊕ A_2$,for some torsion module A 1 and torsion-free module $A_2.$ Prove that $A_1$ has finite composition length.

I can see why $A_1$ has ACC, but I can not prove $A_1$ has DCC.
This is from Module Theory, Extending Modules, and Generalizations. Theorem 4.12.

$\endgroup$

1 Answer 1

2
$\begingroup$

Let $R$ be a Dedekind domain and $M$ a torsion module over $R$. Then $M$ is a finitely generated module over the ring $S = R/ \operatorname{Ann}_R(M)$, and any composition series of $M$ over $S$ is a composition series over $R$. Note that $\operatorname{Ann}_R(M)$ is non-zero because $M$ is torsion, and according to Wikipedia it follows that $S$ is a (principal) Artinian ring. As described on the same page, it follows from Hopkins' theorem that $M$ is of finite length over $S$, and therefore over $R$.


Here is an alternative approach.

Proposition. Let $R$ be a noetherian ring, and $M$ a finite $R$ module. Then $M$ is of finite length if and only if $\operatorname{Ass}_R(M) \subseteq \operatorname{Spm} R$.

[Recall that the set of associated primes of $M$ is $\operatorname{Ass}_S(M) = \lbrace \mathfrak{p} \in \operatorname{Spec} R : \mathfrak{p} = \operatorname{Ann}_S(m) \text{ for some } m \in M \rbrace$.]

Proof. By 00L0, there exists a filtration by submodules $$ 0 = M_0 \subset M_1 \subset \cdots \subset M_n = M $$ such that each quotient $M_i /M_{i-1} \cong R/ \mathfrak{p}_i$ for some $\mathfrak{p}_i \in \operatorname{Spec} R$. Now, there are two cases:

  1. Each $\mathfrak{p}_i$ is maximal. In this case, each quotient $M_i / M_{i-1}$ is simple, so the chain $M_\bullet$ is maximal. By 00J3, it follows that the length of $M$ is $n < \infty$.
  2. There is some $\mathfrak{p}_i$ that is not maximal. Then $R/ \mathfrak{p}_i$ is an integral domain that is not a field. Take any non-zero non-unit $x\in R/ \mathfrak{p}_i$. Then $R/ \mathfrak{p}_i \neq (x) \nsupseteq (x^2) \nsupseteq (x^3) \cdots \neq 0$ is an infinite chain of submodules of $R/\mathfrak{p}$, which corresponds to an infinite chain of submodules between $M_{i-1}$ and $M_i$. So the chain $M_\bullet$ can be refined to a chain of arbitrary length. In particular, $M$ does not have finite length.

Now we need to make a connection between the above and $\operatorname{Ass}_R(M)$. By 02CE, $\operatorname{Ass}_R(M)$ is precisely the set of minimal primes among $\lbrace \mathfrak{p}_1,\dots,\mathfrak{p}_n \rbrace$. So $\operatorname{Ass}_R(M) \subseteq \operatorname{Spm}R$ if and only if each $\mathfrak{p}_i$ is maximal. $\Box$

Corollary. Let $R$ be a noetherian domain of Krull dimension 1, and $M$ an $R$-module that is finite and torsion. Then $M$ has finite length.

Proof. Since $R$ has dimension 1, any associated prime of $M$ is either 0 or maximal. $M$ being torsion, $0 \notin \operatorname{Ann}_R(M)$, so $\operatorname{Ann}_R(M) \subseteq \operatorname{Spm}R$. $\Box$

$\endgroup$
3
  • $\begingroup$ The "according to wikipedia" obscures a lot, because it's not clear what you're looking at to decide it's Artinian. In fact, it is true that $S$ is Artinian (it's semisimple Artinian, even) so I believe you. I'm just curious to see where that "according to wikipedia" part shows up. $\endgroup$ Commented Dec 15, 2021 at 14:31
  • $\begingroup$ I was referring to the following bullet point under "Examples and Counterexamples": "If $I$ is a nonzero ideal of a Dedekind domain $A$, then $A/I$ is a principal Artinian ring." $\endgroup$ Commented Dec 15, 2021 at 14:36
  • 1
    $\begingroup$ Thanks for clarifying! good job on this post, too. $\endgroup$ Commented Dec 15, 2021 at 15:06

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.