I have the following proposition:

Proposition: If $R$ is a principal ideal domain (PID) and $M$, $N$ two finitely generated (fg) $R-$modules, then $\text{Hom}_R(M,N)$ is finitely generated.

My idea: since $M$, $N$ are fg $R-$modules over a PID, then each one is $R-$isomorphic to a direct sum of cyclic modules:

$$M \cong R^{r} \oplus R/\left<a_1\right> \oplus R/\left<a_2\right> \oplus \cdots \oplus R/\left<a_m\right>$$ $$N \cong R^{s} \oplus R/\left<b_1\right> \oplus R/\left<b_2\right> \oplus \cdots \oplus R/\left<b_n\right>$$

where $a_i,b_j$ are the invarian factors in $R$ and $r,s$ are the rank of $M$ and $N$ respectively. To simplify define $R_{a_0} := R^r$, $R_{a_i} := R/\left<a_i\right>$, $R_{b_0} := R^s$ and $R_{b_j} := R/\left<b_j\right>$. So, $M\cong\sum_{i=0}^m\oplus R_{a_i}$ and $N\cong\sum_{j=0}^{n}\oplus R_{b_j}$. Therefore, we have the following $R-$isomorphism:

$$\text{Hom}_R(M,N) \cong \text{Hom}_R\left(\sum_{i=0}^m\oplus R_{a_i},\sum_{j=0}^n\oplus R_{b_j}\right) \cong \text{Hom}_R\left(\bigoplus_{i=0}^m R_{a_i},\bigoplus_{j=0}^n R_{b_j}\right) \cong \bigoplus_{i,j}\text{Hom}_R(R_{a_i},R_{b_j}) \cong \sum_{i,j}\oplus\text{Hom}_R(R_{a_i},R_{b_j})$$

Here I have to calculate $(r+m)\times(n+s)$ Hom-groups between cosets of $R$ and copies of $R$ to finally get that $\text{Hom}_R(M,N)$ is $R-$isomorphic to a direct sum of cyclic modules and therefore is fg. But, I think there is a more simple way to do this. I'm right? Any hit? Thanks.

  • $\begingroup$ It is correct for me $\endgroup$ – Federico Fallucca Feb 19 '19 at 15:15

A PID is Noetherian. Let us do the proof more in general for a commutative Noetherian ring $R$:

Let $M,N$ be finitely generated modules. Pick $n,m\in\mathbb{N}$ and $K\leq R^m$, $L\leq R^n$ such that $$M\cong R^m/K, N\cong R^n/L.$$ Let us write these identities as exact sequences: $$0\rightarrow K\rightarrow R^m \rightarrow M\rightarrow 0, \,\,\,\,(1)$$ $$0\rightarrow L\rightarrow R^n \rightarrow N\rightarrow 0. \,\,\,\,\,\,\,\, (2)$$

Let us apply the functor Hom$(\cdot, N)$ to (1). Since it is a left-exact contravariant functor and $R$ is commutative, we get the exact sequence of $R$-modules $$0\rightarrow\text{Hom}(M,N)\rightarrow^f\text{Hom}(R^m,N)\rightarrow\text{Hom}(K,N).$$ From this we see that $f$ is injective, so Hom$(M,N)$ is isomorphic to a submodule of Hom$(R^m,N)$; since $R$ is Noetherian, if Hom$(R^m,N)$ were finitely generated then so would be Hom$(M,N)$. In order to prove this, apply now the functor Hom$(R^m,\cdot)$ to (2). Since it is a left covariant functor, we get the exact sequence $$\text{Hom}(R^m,L)\rightarrow\text{Hom}(R^m,R^n)\rightarrow^g\text{Hom}(R^m,N)\rightarrow 0.$$ From this we see that $g$ is surjective, so Hom$(R^m,N)$ is a quotient of Hom$(R^m,R^n)$. If this last $R$-module were finitely generated, then so would be Hom$(R^m,N)$ and our proof would be complete. But Hom$(R^m,R^n)$ is isomorphic to the $R$-module of $n\times m$ matrices over $R$, which is finitely generated (it is actually isomorphic to $R^{nm}$ as $R$-modules). Therefore Hom$(M,N)$ is finitely generated.

Observe that we have only actually used the sequences $$R^m \rightarrow M\rightarrow 0,$$ $$R^n \rightarrow N\rightarrow 0. $$


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.