14
$\begingroup$

I know this must be easy. I'm new to modules, so perhaps I must be missing something important.

Let $M$ be an $R$-module. Show that if there exists a submodule $N$ such that $N$ and $M/N$ are finitely generated, then $M$ is finitely generated.

If the exact sequence $0 \longrightarrow N \longrightarrow M \longrightarrow M/N \longrightarrow 0 $ (where the second and third arrows are inclusion and projection respectively) splits then $M$ would be isomorphic to the direct sum of $M$ and $M/N$ and the conclusion would follow (I think). But I can't show that it splits.

Thank you.

$\endgroup$
4
  • 3
    $\begingroup$ Can you do it for vector spaces? $\endgroup$ Jan 18 '12 at 2:06
  • 2
    $\begingroup$ By the way, the sequence does not split in general! It is therefore good that you were not able to prove it does :) $\endgroup$ Jan 18 '12 at 4:03
  • 1
    $\begingroup$ @Mariano: thank you. What's an example that it doesn't split? $\endgroup$ Jan 18 '12 at 12:35
  • 1
    $\begingroup$ @Dylan: Thank you. The first map, though, is not inclusion... $\endgroup$ Jan 18 '12 at 22:22
25
$\begingroup$

Perhaps this works?

Suppose $M/N$ is finitely generated with generators $a_1+N,\dots,a_m+N$, and $N$ is finitely generated with generators $b_1,\dots,b_n$. Take any $m\in M$. Then $m+N=\sum_{i=1}^m r_i(a_i+N)=\left(\sum_{i=1}^m r_ia_i\right)+N$ viewing $M/N$ as an $R$-module.

This implies $m-\sum_{i=1}^m r_ia_i\in N$, and thus $m-\sum_{i=1}^m r_ia_i=\sum_{j=1}^n s_jb_j$ for some $r_i,s_j\in R$. Finally, $m=\sum_{i=1}^m r_ia_i+\sum_{j=1}^n s_jb_j$. So $M$ is generated by $a_i, b_j$ for $i=1,\dots, m$ and $j=1,\dots n$

$\endgroup$
2
  • $\begingroup$ Doesn't that only show that the element $m \in M$ can be expressed with those generators? Did you show that those generators $i= 1,....,m$ and $j=1,....,n$ generate ALL of $M$? $\endgroup$ Nov 15 '18 at 19:28
  • 4
    $\begingroup$ @MichaelVaughan He has chosen arbitrary element $m\in M$ and showed that it can be represented by $R-$linear combination of $\{a_1,\dots,a_n,b_1,\dots, b_m\}$ which is already specified. He does not say it only applied to specific $m$. $\endgroup$
    – Lev Bahn
    Nov 21 '18 at 0:01
5
$\begingroup$

Take a generating set $\{\bar z_1, \ldots, \bar z_m\}$ for $M/N$. If $x$ is an element of $M$, then there exist $a_1, \ldots, a_n \in R$ such that the image $\bar x$ of $x$ in $M/N$ is \[ \bar x = a_1\bar z_1 + \cdots + a_n\bar z_n. \] Since the map on the right is surjective, you can choose lifts $\{z_1, \ldots, z_m\}$ in $M$ of the aforementioned generating set. Now, the element \[ x' = a_1z_1 + \cdots + a_nz_n \] has the same image as $x$ in $M/N$. Do you see a way to use the exactness in the middle?

Note that the sequence need not split: consider the $\mathbf Z$-submodule $2\mathbf Z$ of $\mathbf Z$.

$\endgroup$

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.