How can we compute the localization of the ring $\mathbb{Z}/6\mathbb{Z}$ at the prime ideal $2\mathbb{Z}/\mathbb{6Z}$? (or how do we see that this localization is an integral domain)?
-
arent you inverting the zero divisor 3? – yoyo Apr 11 '11 at 20:42
-
4@yoyo: Yes, but $S$ containing a zerodivisor only implies that the canonical map $f:R\rightarrow S^{-1}R$ is not injective, as we would expect here. The map $f$ is only the zero map when $0\in S$. – Zev Chonoles Apr 11 '11 at 20:47
-
See this post for a general approach with $\mathbb{Z}/n\mathbb{Z}$. – Anakhand Jan 27 '24 at 15:05
4 Answers
The localization of a ring $R$ at a prime $P$ is the local ring $R_P$, having maximal ideal $PR_P$. In this case, $R=\mathbb{Z}/6\mathbb{Z}$ and $P=2\mathbb{Z}/6\mathbb{Z}$, so that the maximal ideal $$PR_P=\{\textstyle\frac{r}{s}\mid r\in P, s\notin P\}=\{\frac{0}{1},\frac{2}{1},\frac{4}{1},\frac{0}{3},\frac{2}{3},\frac{4}{3},\frac{0}{5},\frac{2}{5},\frac{4}{5}\}$$ However, note that $\frac{r_1}{s_1}=\frac{r_2}{s_2}$ iff there is a $u\notin P$ such that $u(r_1s_2-r_2s_1)=0$. Thus, for example $\frac{2}{1}=\frac{0}{1}$ because $3(2\cdot1-0\cdot 1)=0$. In fact every element of $PR_P$ is 0, by a similar computation. Thus $PR_P=0$, and a local ring whose maximal ideal is the 0 ideal is a field (and hence in particular an integral domain). Thus $R_P$ is an integral domain.
EDIT: We really also have to check that $R_P$ is not in fact the zero ring (which is not an integral domain or field). We can do this directly, by checking that $\frac{1}{1}\neq\frac{0}{1}$ (because there is no $s\notin P$ such that $s(1\cdot1-0\cdot 1)=s=0$), or we can do it as follows: The canonical map $f:R\rightarrow R_P$ defined by $f(r)=\frac{r}{1}$ can easily be seen to have kernel $\ker(f)=\{r\in R\mid \exists s\notin P: sr=0\}$. For $R=\mathbb{Z}/6\mathbb{Z}$ and $P=2\mathbb{Z}/6\mathbb{Z}$, note that $\ker(f)=P$, so that by the first isomorphism theorem $R/P$ injects into $R_P$, so that $R_P$ is not the zero ring.

- 129,973
We first recall the following well known result.
Let $R$ be a commutative ring with unity, $I$ an ideal in $R$ and $S$ a multiplicative closed set in $R$. Let $\bar{S}$ denote the image of $S$ in the quotient ring $\bar{R}=R/I$. Then $\bar{S}^{-1}\bar{R} = S^{-1}R/IS^{-1}R$. (Check the book here.)
Now take $R=\mathbb{Z}$, $I=6\mathbb{Z}$ and $S=\mathbb{Z}-2\mathbb{Z}$. Then $\bar{\mathbb{Z}}_{(\bar{2})}= \bar{S}^{-1}\bar{R} = S^{-1}R/IS^{-1}R = \mathbb{Z}_{(2)}/ 6\mathbb{Z}_{(2)} = \mathbb{Z}_{(2)}/ 2\mathbb{Z}_{(2)}$ {since 3 is a unit in $\mathbb{Z}_{(2)}$; $2 \mathbb{Z}_{(2)}$ = $6 \mathbb{Z}_{(2)}$} = $\mathbb{Z}/2\mathbb{Z}$.
-
The method allows us to calculate the localization of $\Bbb{Z}/n,$ e.g., the localization at the prime ideal $2\Bbb{Z}/(2^{2}\cdot 3)$ of the ring $\Bbb{Z}/(2^{2}\cdot 3).$ Since $3$ is a unit but $2^{2}$ is not a unit in $\Bbb{Z}{(2)},$ we have $(\Bbb{Z}/(2^{2}\cdot 3)){(2)}=\Bbb{Z}/2^{2}.$ – user623904 Sep 29 '22 at 23:42
One simple way to compute this is to exploit the universal property of localization. By definition $\rm\, L = \mathbb Z/6_{\,(2)} = S^{-1}\, \mathbb Z/6\ $ where $\rm\ S\ =\ \mathbb Z/6 \ \backslash\ 2\ \mathbb Z/6\ =\ \{\bar 1, \bar 3, \bar 5\}.\,$ Since the natural map $\rm\ \mathbb Z/6\ \to\ \mathbb Z/2\ $ maps $\rm\:S\:$ to units, by said universality property it must factor through $\rm\,L,\, $ i.e. $\rm\ \mathbb Z/6\ \to\ L\ \to\ \mathbb Z/2,\, $ so either $\rm\, L = \mathbb Z/6\, $ or $\rm\, L = \mathbb Z/2,\,$ but $\rm\ \bar 3\in S,\ \ {\bar3}^{-1}\!\!\not\in \mathbb Z/6,\, $ so $\rm\,L \ne \mathbb Z/6$.

- 272,048
-
Why is ${\bar3}^{-1}\not\in \mathbb Z/6$? We have ${\bar3}^{-1} = \bar1$, or equivalently $\bar3=\bar1$, because there is $s = \bar3 \in S$ such that $s(\bar3-\bar1) = \bar0$. – Thrash Feb 21 '24 at 13:58
-
1@Thrash It seems you are confusing $\Bbb Z/6,$ with $,S^{-1}\Bbb Z/6.\ \ $ – Bill Dubuque Feb 21 '24 at 15:35
-
-
-
@user26857: I think it is because the kernel of the said natural map $\mathbb Z/6 \to \mathbb Z/2$ consists of 3 elements, but this map also factors through $L$, which means that (because 3 is prime) either the kernel of $\mathbb Z/6 \to L$ or the kernel of $L \to \mathbb Z/2$ consists of 3 elements (while the other map has a trivial kernel, i.e., it is an isomorphism). – Thrash Feb 24 '24 at 23:47
-
I'm not quite sure, but it seems we can also argue as follows: $\bar 3$ must be invertible in $L$, i.e., it exists an $a \in L$ such that $\bar 3a = \bar 1$ (in $L$). Multiplying by 2 yields $\bar 0 = \bar 6a = \bar 2$ (in $L$) and thus $\bar 0 = \bar 2 = \bar 4$ (in $L$) and $\bar 1 = \bar 3 = \bar 5$ (in $L$), i.e., $L = \mathbb Z/2$. – Thrash Feb 25 '24 at 00:01
Here's a much less computational approach to show that the localisation is an integral domain (even a field):
For any (commutative, unital) ring $R$ and prime ideal $P$, recall that $R_P$ is a local ring with unique maximal ideal $P_P$.
In the given case, since there is an element $x \in R\setminus P$ (i.e. $3$) such that $x \in \operatorname{Ann}(P)$ it follows that $P_P$ is $(0)$. Thus $R_P$ is a local ring with unique maximal ideal $(0)$ and so is a field.
Note that this actually gives us the stronger statement that $R_P$ is a field iff $P_P$ is $(0)$.

- 52,094

- 13,034
- 1
- 39
- 77