7
$\begingroup$

I'm trying to work out Exercise 2.6 in Commutative Algebra by Eisenbud, which asks to prove the Chinese Remainder Theorem for commutative rings.

Exercise: Let $R$ be a commutative ring, and let $I_1,\ldots,I_d$ be pairwise comaximal ideals. Prove that $R/\left(\bigcap_{k=1}^d I_k\right) \simeq \prod_{k=1}^d R/I_k$, via the map $\varphi: R\rightarrow \prod R/I_k$ given by $r \mapsto (r+I_1,\ldots,r+I_d)$.

The kernel of $\varphi$ is precisely $\bigcap_{k=1}^d I_k$, so the goal of the argument is to show that $\varphi$ is surjective. Eisenbud hints to use his Corollary 2.9, which states the following.

Lemma: Let $\varphi:M\rightarrow N$ be an $R$-module homomorphism. Then $\varphi$ is surjective if and only if the induced map $\varphi_{\mathfrak{m}} : M_{\mathfrak{m}} \rightarrow N_\mathfrak{m}$ is surjective for every maximal ideal $\mathfrak{m}$ of $R$.

(Specifically, the map $\varphi_\mathfrak{m}$ is given by $m/u \mapsto \varphi(m)/u$.)

I have a few problems with this.

  1. The lemma is a statement about module homomorphisms, but the $\varphi$ appearing in the Chinese Remainder Theorem is not a module homomorphism (it's a ring homomorphism). I tried to change the lemma to work for ring homomorphisms, but ran into the following problem.
  2. Even if $\varphi:R\rightarrow S$ is a ring homomorphism, how does $\varphi_\mathfrak{m} : R_\mathfrak{m} \rightarrow S_\mathfrak{m}$ make sense? I get that you can localize $R$ at a maximal ideals, but what does "$S_\mathfrak{m}$" mean when $\mathfrak{m}$ is a maximal ideal of $R$? (Note this post, which actually gives the failure of the lemma for ring homomorphisms -- except in that context we localize $S$ at a prime ideal of $S$, and then localize $R$ at the preimage of that prime ideal.)

Regarding the second remark above, I was thinking that one could view $S$ as an $R$-module by $rs:=\varphi(r)s$, and then consider $S_\mathfrak{m}$ as a localization of an $R$-module by a maximal ideal. But then in order for $\varphi_\mathfrak{m}:R_\mathfrak{m}\rightarrow S_\mathfrak{m}$ to be a ring homomorphism, one has to decide how $S_\mathfrak{m}$ is a ring and then define how to induce a ring homomorphism on localization. This seems like a lot for Eisenbud to sweep under the rug when he says to use the Lemma to solve the exercise. So as far as I understand, the above lemma isn't even relevant to the exercise.

Am I missing something here? Is there a clever way to indirectly apply the lemma?

$\endgroup$
  • 1
    $\begingroup$ $\varphi$ is a homomorphism of $R$-modules. $\endgroup$ – Zev Chonoles Feb 20 '14 at 5:09
  • $\begingroup$ Oh ... right. Thanks. $\endgroup$ – Ehsaan Feb 20 '14 at 5:18
5
$\begingroup$

Note that the ring $R$, and each of the rings $R/I_k$, is also an $R$-module. It is easy to check that the map $\varphi$ is an $R$-module homomorphism.

By the definition of "comaximal", any given maximal ideal $M\subset R$ will contain at most one $I_k$.

For any given maximal ideal $M\subset R$ and any $I_k$, we have $(R/I_k)_M\cong R_M/(I_k)_M$ because localization commutes with taking quotients. If $M$ contains $I_k$, then $(I_k)_M\neq R_M$, whereas if $M$ does not contain $I_k$, we have $(I_k)_M=R_M$ because $I_k\cap (R\setminus M)\neq\varnothing$.

In either case, we see that the localized map $$\small\varphi_M:R_M\to \left(\prod R/I_k\right)_M\cong \prod R_M/(I_k)_M\cong \begin{cases} R_M/(I_k)_M &\text{if $I_k$ is contained in $M$,}\\ 0 & \text{if none of the $I_k$'s are contained in $M$} \end{cases}$$ is just a quotient map from the ring $R_M$, hence surjective.

(I leave it to you to check that the composition of $\varphi_M$ with all of these isomorphisms actually does result in the quotient map $R_M\to R_M/(I_k)_M$.)

P.S. This is actually the first time I've seen this approach to proving the map in the Chinese remainder theorem is surjective. I like it a lot, though there are proofs which don't need any localization. Consider the approach taken in Atiyah-Macdonald:

  enter image description here

$\endgroup$
  • $\begingroup$ Thanks! Atiyah-McDonald's proof has the advantage of working for noncommutative rings too, right? I've seen that one, but Eisenbud's exercise looked pretty slick. $\endgroup$ – Ehsaan Feb 20 '14 at 18:08
  • $\begingroup$ Could you explain why any given maximal ideal M⊂R will contain at most one Ik.? $\endgroup$ – Bob Feb 12 '17 at 0:20
  • 1
    $\begingroup$ Bob: Since the $I_k$ are comaximal, we have $I_k+I_j=R$ for $k \neq j$. If $I_k, I_j \subset M$ for $i \neq j$, then $R \subset M$, which contradicts the definition of maximal ideal. $\endgroup$ – Massimo Feb 12 '17 at 17:58
  • $\begingroup$ If $M$ does not contain $I_k$, we have $(I_k)_M=R_M$ because $I_k∩(R∖M)\neq∅$. Why does this reason imply $(I_k)_M=R_M$ $\endgroup$ – Bob Feb 13 '17 at 14:03

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.