This is the problem I want to solve :

Show that for any finitely generated projective module $P$ over a ring $R $, $\mathrm{Hom}(P,M)$ is isomorphic with $\mathrm{Hom}(P,R)\otimes M $.

This is what I've done:

I define a map : $Hom (P,R) × M \rightarrow Hom (P,M) $, with the law: $(\phi,m)\rightarrow \phi_m $, $\phi_m (x):=\phi(x) m $.

This map is obviously bilinear, so we have a map $Hom (P,R) \otimes M \rightarrow Hom (P,M) $, with the law $\phi\otimes m \rightarrow \phi_m $, $\phi_m (x):=\phi(x) m $.

Now I have to construct the inverse homomorphism to end the proof.

Because $P $ is finitely generated, we can assume $P=R^n $. I define a map:

$Hom (P,M) \rightarrow Hom (P,R)\otimes M $ that sends $f $ to $f'\otimes f (1) $, when $1$ is the identity element of $ P=R^n $. I don't know how to define $f'$.

Is there any hint?

If you don't agree with these, do you have another idea?


  • $\begingroup$ You can't just assume that $P = R^n$; in general, finitely generated projective modules need not be free. $\endgroup$ – lokodiz Oct 28 '16 at 5:47
  • $\begingroup$ A special case is here: math.stackexchange.com/questions/679584 $\endgroup$ – Watson Sep 26 '18 at 14:42

Hint: This is where you need to use finite generation and projectiveness of $P$. Since $P$ is finitely generated, there is a surjective $R$-map $p : R^n \to P$ for some $n > 0$. Projectiveness of $P$ yields a section $s : P \to R^n$ of $p$ (so that $p\circ s = \operatorname{id}_P$). For $1\le i\le n$, let $s_i$ be the composition $P \xrightarrow{s} R^n \xrightarrow{\pi_i} R$, where $\pi_i$ is the projection onto the $i$th coordinate. Now we define an $R$-map $$\operatorname{Hom}(P,M) \to \operatorname{Hom}(P,R) \otimes M$$ by sending a $g\in \operatorname{Hom}(P,M)$ to $\sum s_i \otimes g(p(e_i))$, where $e_1,\ldots, e_n$ is the standard basis of $R^n$.

| cite | improve this answer | |
  • $\begingroup$ thank you a lot!but how do you know that $\pi_i$'s are well defined? $\endgroup$ – user115608 Oct 28 '16 at 12:51
  • $\begingroup$ @user115608 actually they weren't well-defined unless the sum was direct. I've made an edit to my answer. $\endgroup$ – kobe Oct 28 '16 at 16:20

Correct me if anything worng. If we consider left modules. Then $P=Re$ for some idempotent $e \in R$. Hence $Hom_R(P,M)\cong eM$ as (eRe,R)-modules. Also $Hom_R(P,R) \otimes_R M\cong eR\otimes_R M \cong eM$. So $Hom_R(P,M)\cong Hom_R(P,R)\otimes_RM$.

| cite | improve this answer | |

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.