Let $R$ be a ring and let $A_1, \ldots, A_n$ be $R$-modules.

I have to show that there are $R$-homomorphisms $e_i : \prod\limits_{i =1}^n A_i \rightarrow \prod\limits_{i =1}^n A_i$ for $i = 1, \ldots, n$ satisfying the following conditions :

(I) $\quad$ $e_i \circ e_i = e_i$ for all $i = 1, \ldots ,n$

(II) $\ \ \ \ e_i \circ e_j = 0$ for all $1 \leq i, j \leq n$ and $i \neq j$

(III) $\ \ \ id_A = e_1 + \cdots + e_n$

My problem is that I have no idea how to construct such a map $e_i$.

Furthermore, I have to prove that $\varphi : A \rightarrow \prod\limits_{i =1}^n A_i$ is an $R$-isomorphism, where $A$ is an $R$-module, $e_1, \ldots, e_n \in Hom_R(A, A)$ a complete set of orthogonal idempotents of $M$ and $A_i = e_i(A)$.

It is clear to me how to prove that a map between modules is an $R$-isomorphism, but in this case I may need some hints how $\varphi$ is defined.

Thanks for your help.

  • 1
    $\begingroup$ My guess is projection maps $e_i : (x_1, x_2, \cdots, x_i, \cdots, x_n)\to (0, 0, \cdots, x_i,\cdots,0)$ $\endgroup$ – Bumblebee Nov 22 '17 at 17:22
  • 1
    $\begingroup$ The obvious candidate (which Bumblebee points out) works, and the verification of $I-III$ is all routine after that. It's also trivial to show that $\phi = \sum e_i$ is the map you seek in the second part. $\endgroup$ – rschwieb Nov 22 '17 at 17:31
  • $\begingroup$ @Bumblebee and rschwieb. Thank you for your helpful hints :). Finally, I could solve the task. $\endgroup$ – Crystal Nov 23 '17 at 16:01
  • $\begingroup$ @Crystal: That is great. I am happy to up see your answer. You can post the answer for your own question which would help for many students using this site. $\endgroup$ – Bumblebee Nov 24 '17 at 3:39
  • 1
    $\begingroup$ @Bumblebee : I apologize for the late answer. Of course I can post my solution. In fact, the first part is trivially verified if one just considers your comment. I am sure that every student at my level will succeed in proving the first part. So I only post an answer to the second part. $\endgroup$ – Crystal Dec 1 '17 at 22:42

Since the first part is obvious, I answer the second part of the question.

By the universal property of $\prod\limits_{i = 1}^n A_i$, there exists a unique homomorphism $\phi : A \rightarrow \prod\limits_{i = 1}^n A_i$ such that $\pi_i \circ \phi = e_i$ for all $i \in \{1, \ldots ,n\}$. By condition III we have $(e_1 + \cdots + e_n) \circ \phi = id_A$.

It remains to show that $\phi$ is a bijection.

Let $x \in ker(\phi)$. Then $(\pi_1 + \cdots + \pi_n) \circ \phi(x) = x$ implies that $x = 0$. Hence, $\phi$ is injective.

We have $A_i = e_i(A)$. Since $\pi_i$ is surjective, $\pi_i(\prod\limits_{i = 1}^n A_i) = A_i$. Moreover, $(\pi_i \circ \phi)(A) = e_i(A)$. It follows that $\pi_i(\prod\limits_{i = 1}^n A_i) = \pi_i(\phi(A))$ for all $i = 1, \ldots,n$.

Finally, $\prod\limits_{i = 1}^n A_i = \phi(A) \Rightarrow \phi$ is surjective.


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.