Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

In Eisenbud's Commutative algebra with a view..., he shows that if an $A$-module $M$ has a finite length, then the sum of localization maps at maximal ideals is an isomorphism: $$\phi:M\overset{\sim}{\rightarrow}\oplus_{\mathfrak{m}}M_{\mathfrak{m}}.$$ But isn't this true for any module (i.e., not necessarily finite length)? My argument is that $\phi$ is an isomorphism because $\phi_{\mathfrak{m}}$ is an isomorphism for every maximal ideal $\mathfrak{m}$ since


Am I missing something here? I think only difference that $\ell(M)<\infty$ makes is maybe the sum is finite...

share|cite|improve this question
We certainly know that $\phi:M\to \bigoplus_{m} M_{m}$ is an injective $A$-module homomorphism. How do you propose to prove that $\phi$ is a surjective $A$-module homomorphism? – Amitesh Datta Aug 28 '11 at 23:33
I think I see where I went wrong. I answered my own question below. – ashpool Aug 29 '11 at 0:03
up vote 2 down vote accepted

Let us discuss this in a number of steps:

Exercise 1: Let $A$ be a commutative ring and let $M$ be an $A$-module. Let $\text{MaxSpec}(A)$ be the set of all maximal ideals of $A$. Prove that the natural $A$-module homomorphism $\phi:M\to \bigoplus_{m\in\text{MaxSpec}(A)} M_m$ is injective. (Hint: if $x\in M$, choose a maximal ideal $m$ of $A$ such that $\text{Ann}(x)=\{a\in A:ax=0\}\subseteq m$. Consider the image of $x$ in $M_m$.)

Exercise 2: Let us consider the $\mathbb{Z}$-module $\mathbb{Z}$ and let $\phi:\mathbb{Z}\to \bigoplus_{m\in\text{MaxSpec}(\mathbb{Z})} \mathbb{Z}_m$ be the natural $\mathbb{Z}$-module homomorphism. Prove that $\phi$ is not surjective. (Hint: let us order the positive prime numbers in the manner $p_1,p_2,\dots$. If $n\in \mathbb{Z}$, let us remark that the image of $n$ in $\mathbb{Z}_{p_i}$ is $\frac{n}{1}$. Prove that the element $(1,0,0,\dots)\in \bigoplus_{i=1}^{\infty} \mathbb{Z}_{p_i}$ has no preimage in $\mathbb{Z}$ under $\phi$. If you are stuck, think about the kernel of the localization homomorphism $\mathbb{Z}\to \mathbb{Z}_{p_i}$ for $i\in\mathbb{N}$.)

Problem 1: Let $A$ be an integral domain. Prove that the natural $A$-module homomorphism $\phi:A\to \bigoplus_{m\in\text{MaxSpec}(A)} A_m$ is not surjective unless $A$ is a local domain. Can you generalize to other commutative rings?

Problem 2: Let $A$ be an integral domain. Prove that $A=\bigcap_{m\in\text{MaxSpec}(A)} A_m$.

I hope this helps!

share|cite|improve this answer
Thanks! Very interesting examples. As for Problem 2, I don't immediately see a connection to the current problem. Is there a connection? – ashpool Aug 29 '11 at 0:25
@ashpool Problem 2 should be easy if you understand the answer to your question very well. Indeed, the technique used to solve Problem 2 is not at all dissimilar to the one used to solve Exercise 1. Also, the result of Problem 2 is somewhat important. – Amitesh Datta Aug 29 '11 at 0:29
Dear Amitesh, Problem 1 doesn't seem to be correct: think of a local domain. – Georges Elencwajg Aug 29 '11 at 0:36
@Amitesh Datta I was aware of the result in Problem 2, but dismissed it thinking that it is not related to the current problem. – ashpool Aug 29 '11 at 0:47
@Georges Dear Georges, thanks! A careless mistake on my part. I will edit my answer. Regards, – Amitesh Datta Aug 29 '11 at 0:50

I realized an error in my argument. The equation should be more properly read $$\mbox{Supp}_{A_\mathfrak{m}}M_{\mathfrak{m}}=\mbox{Supp}_AM\cap\mbox{Spec}A_{\mathfrak{m}}.$$ $\phi$ is an $A$-module homomorphism, and $M_{\mathfrak{m}}$ here should be viewed as an $A$-module, not an $A_{\mathfrak{m}}$-module. So $\mbox{Supp}_AM_{\mathfrak{m}}$ may be actaully larger than $\mbox{Supp}_{A_{\mathfrak{m}}}M_{\mathfrak{m}}$. So, even if $\mathfrak{m}\neq\mathfrak{n}$ are two distinct maximal ideals, $(M_{\mathfrak{m}})_{\mathfrak{n}}$ may not be zero. I don't have a definite example, but at least this makes my previous argument invalid.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.