The Stacks project

Lemma 10.67.2. Let $R$ be a Noetherian ring. Let $M$ be a finite $R$-module. Consider the set of $R$-submodules

\[ \{ K \subset M \mid \text{Supp}(K) \text{ nowhere dense in } \text{Supp}(M) \} . \]

This set has a maximal element $K$ and the quotient $M' = M/K$ has the following properties

  1. $\text{Supp}(M) = \text{Supp}(M')$,

  2. $M'$ has no embedded associated primes,

  3. for any $f \in R$ which is contained in all embedded associated primes of $M$ we have $M_ f \cong M'_ f$.

Proof. We will use Lemma 10.63.5 and Proposition 10.63.6 without further mention. Let $\mathfrak q_1, \ldots , \mathfrak q_ t$ denote the minimal primes in the support of $M$. Let $\mathfrak p_1, \ldots , \mathfrak p_ s$ denote the embedded associated primes of $M$. Then $\text{Ass}(M) = \{ \mathfrak q_ j, \mathfrak p_ i\} $. Let

\[ K = \{ m \in M \mid \text{Supp}(Rm) \subset \bigcup V(\mathfrak p_ i)\} \]

It is immediately seen to be a submodule. Since $M$ is finite over a Noetherian ring, we know $K$ is finite too. Hence $\text{Supp}(K)$ is nowhere dense in $\text{Supp}(M)$. Let $K' \subset M$ be another submodule with support nowhere dense in $\text{Supp}(M)$. This means that $K_{\mathfrak q_ j} = 0$. Hence if $m \in K'$, then $m$ maps to zero in $M_{\mathfrak q_ j}$ which in turn implies $(Rm)_{\mathfrak q_ j} = 0$. On the other hand we have $\text{Ass}(Rm) \subset \text{Ass}(M)$. Hence the support of $Rm$ is contained in $\bigcup V(\mathfrak p_ i)$. Therefore $m \in K$ and thus $K' \subset K$ as $m$ was arbitrary in $K'$.

Let $M' = M/K$. Since $K_{\mathfrak q_ j}=0$ we know $M'_{\mathfrak q_ j} = M_{\mathfrak q_ j}$ for all $j$. Hence $M$ and $M'$ have the same support.

Suppose $\mathfrak q = \text{Ann}(\overline{m}) \in \text{Ass}(M')$ where $\overline{m} \in M'$ is the image of $m \in M$. Then $m \not\in K$ and hence the support of $Rm$ must contain one of the $\mathfrak q_ j$. Since $M_{\mathfrak q_ j} = M'_{\mathfrak q_ j}$, we know $\overline{m}$ does not map to zero in $M'_{\mathfrak q_ j}$. Hence $\mathfrak q \subset \mathfrak q_ j$ (actually we have equality), which means that all the associated primes of $M'$ are not embedded.

Let $f$ be an element contained in all $\mathfrak p_ i$. Then $D(f) \cap \text{supp}(K) = 0$. Hence $M_ f = M'_ f$ because $K_ f = 0$. $\square$


Comments (2)

Comment #7153 by Xuande Liu on

I think the proof is not very intuitive and it actually does not prove the third conclusion. And there is a better way to rewrite the proof:

Let . Obviously it is a submodule. And since is finitely generated over a noetherian ring, we know is also generated by finitely many such elements. Hence is nowhere dense in . Let be another submodule with support nowhere dense in . Then can be generated by finitely elements with support nowhere dense in . Notice that . Hence . But . Hence the support of is contained in . Therefore we know and thus .

Let . Since we know for all . Hence and have the same support.

Suppose . Let be an element in lying in the inverse image of . Then the support of must contain one of . Since , we know . Hence , which means that all the associated primes of are not embedded.

Let be an element contained in all . Then . Hence . Now let be an element contained in all but not contained in any .

In this proof, we avoid to check that is independent of the choice of . And is obviously the unique maximal element in that set of submodules.

Comment #7301 by on

OK, thanks. I made the changes you suggested. See this commit. I hope others will comment further to improve the exposition even more.


Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.




In order to prevent bots from posting comments, we would like you to prove that you are human. You can do this by filling in the name of the current tag in the following input field. As a reminder, this is tag 02M6. Beware of the difference between the letter 'O' and the digit '0'.