0
$\begingroup$

Let $R$ be a Dedekind domain and $M$ be a nonzero finitely generated torsion $R$-module. In Curtis and Reiner's Methods of Representation Theory it states that $M$ has a composition series and if its factors are given by $R/P_i$ with $P_i$ maximal $R$-ideals, then the ideal $\mathrm{ord}(M)=\prod P_i$ is well defined.

I don't know why a composition series for $M$ exists as I'm not sure if $M$ is an Artinian module, and also it's not clear why $\mathrm{ord}(M)$ is well defined.

I'd be very grateful for some help.

$\endgroup$
1
$\begingroup$

A finitely generated torsion module $M$ over a Dedekind domain $R$ is artinian because it is a finitely generated $(R/\operatorname{Ann}M)$-module, and $\operatorname{Ann}M\ne0$. This shows that $M$ is an $R$-module of finite length.
Moreover, by Jordan-Holder theorem one knows that every two composition series are equivalent. But the quotient modules of a composition series are of the form $R/P$, with $P$ a maximal ideal, and then these maximal ideals are uniquely determined by $M$.

$\endgroup$
  • $\begingroup$ I've been trying to understand why $M$ is artinian if it is finitely generated over $R/\operatorname{Ann} M$ if $\operatorname{Ann}M$ is nonzero. Also I am not sure why the maximal ideals $P$ are uniquely determined. Is it possible that $R/P_1\cong R/P_2$ for $P_1\neq P_2$? regards $\endgroup$ – eddie Apr 11 '16 at 13:07
  • $\begingroup$ @eddie This holds since the quotient ring $R/\operatorname{Ann}M$ is zero-dimensional and noetherian, that is, artinian. For the next question: no, it's not. $\endgroup$ – user26857 Apr 11 '16 at 13:07
  • $\begingroup$ For more details see here. $\endgroup$ – user26857 Apr 11 '16 at 13:15
  • 1
    $\begingroup$ Last question: yes, that's what I've meant. First question: you misread my comment; I've talked about the quotient ring $R/I$ where $I=\operatorname{Ann} M$. Such a quotient ring is zero-dimensional whenever $I\ne 0$, and this happens for $R$ is one-dimensional integral domain, so all its non-zero prime ideals are maximal. $\endgroup$ – user26857 Apr 11 '16 at 13:28
  • 1
    $\begingroup$ @eddie 1. Modules which are artinian and noetherian are of finite lenght, so they have a composition series. 2. $\operatorname{Ann}M \neq 0$ because $M$ is finitely generated and torsion. (This has nothing to do with the dimension of the ring.) 3. Why need a reference for showing that $\dim R/I=0$ whenever $\dim R=1$ and $I\ne 0$? The prime ideals of $R/I$ are of the form $P/I$ with $P$ prime in $R$ containing $I$. Since $I\ne 0$ then $P$ is maximal and you are done. $\endgroup$ – user26857 Apr 11 '16 at 13:47

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.