This is an unproven proposition I've come across in multiple places.

Suppose $A$ is a commutative Noetherian ring, and $S$ a multiplicative subset of $A$. Then $S^{-1}A$ is Noetherian.

Why is this? I thought about taking some chain of submodules $$ S^{-1}M_1\subset S^{-1}M_2\subset\cdots $$ and pulling back to a chain $$ M_1\subset M_2\subset\cdots $$ of submodules of $A$ which must eventually stablize. Is there more to it than this? I kind of wary of assuming all submodules of $S^{-1}A$ have form $S^{-1}M$ for $M\leq A$.


This is a standard property of localizations:

Theorem. Let $R$ be a commutative ring, and let $S\neq\emptyset$ be a multiplicative subset. Let $M$ be an $R$-module. Let $\varphi\colon R\to S^{-1}R$ be the canonical map ($\varphi(r) = \frac{rs}{s}$), and likewise, by abuse of notation, let $\varphi\colon M\to S^{-1}M$ be the natural map $(\varphi(m) = \frac{sm}{s}$ with $s\in S$).

  1. For every submodule $N$ of $M$, $S^{-1}N = \{\frac{a}{s}\mid a\in N\}$ is a submodule of $S^{-1}M$.

  2. If $L$ is a submodule of $S^{-1}M$, then $\varphi^{-1}(L) = \{m\in M\mid \varphi(m)\in L\}$ is a submodule of $M$.

  3. If $N$ is a submodule of $M$, then $N\subseteq \varphi^{-1}(S^{-1}N)$. Moreover, if $N=\varphi^{-1}(L)$ for some submodule $L$ of $S^{-1}M$, then $L=S^{-1}N$.

Proof. $S^{-1}N$ is nonempty, as it contains $\frac{0}{s}$; it is closed under differences, since $\frac{a}{s}-\frac{b}{t} = \frac{ta-sb}{st}\in S^{-1}N$ if $a,b\in N$. And it is closed under scalar multiplication, since $a\in N$ implies $ra\in N$ for all $r\in R$, so $\frac{r}{t}(\frac{a}{s}) = \frac{ra}{ts}\in S^{-1}N$ if $a\in N$.

Now let $L$ be a submodule of $S^{-1}M$; since $\varphi$ is a module homomorphism, the pullback of a submodule is a submodule, so 2 is immediate.

Again, let $N$ be a submodule of $M$. Then for every $a\in N$ we have $\varphi(a) = \frac{sa}{s}\in S^{-1}N$, since $sa\in N$, hence $a\in \varphi^{-1}(S^{-1}N)$. Now assume that $N=\varphi^{-1}(L)$. If $a\in N$ and $s\in S$, then $\frac{a}{s} = \frac{ssa}{sss} = \frac{s}{ss}\frac{sa}{s} =\frac{s}{ss}\varphi(a)\in L$ (since $\varphi(a)\in L$), hence $S^{-1}N\subseteq L$. Conversely, let $\frac{m}{t}\in L$. Then $\frac{tt}{t}\frac{m}{t} = \frac{(tt)m}{tt}=\varphi(m)\in L$, hence $m\in \varphi^{-1}(L) = N$; thus, $\frac{m}{t}\in S^{-1}N$, proving that $L\subseteq S^{-1}N$. Therefore, $L=S^{-1}N$, as claimed. $\Box$

Corollary. Every submodule of $S^{-1}M$ is of the form $S^{-1}N$ for some submodule $N$ of $M$.

  • $\begingroup$ Thanks, these properties and the detail are very helpful. $\endgroup$ – Buble Feb 18 '12 at 23:28
  • 4
    $\begingroup$ @ArturoMagidin I swear to god I've learned so much from your posts on Math.SE :D $\endgroup$ – user38268 Feb 19 '12 at 1:00

Let $f\colon A \to S^{-1}A$ be the canonical homomorphism. It is true that every ideal $\mathfrak b$ of $S^{-1}A$ is of the form $S^{-1}\mathfrak a = f(\mathfrak a)(S^{-1}A)$ for some ideal $\mathfrak a$ of $A$. We can even take $\mathfrak a = f^{-1}(\mathfrak b)$, and this should help you prove that $S^{-1}A$ is Noetherian using increasing chains.

You can find related facts in and around Proposition 6.4 of Milne.

  • $\begingroup$ Thanks for your help again, Dylan. btw, should it say some ideal $\mathfrak{a}$ of $A$? in the second line, not $S^{-1}A$? $\endgroup$ – Buble Feb 18 '12 at 23:27
  • $\begingroup$ @Buble Yes indeed. I was toying with the wording and this mixup, inevitably, happened. $\endgroup$ – Dylan Moreland Feb 18 '12 at 23:47

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.