2
$\begingroup$

Let $n\in \mathbb{Z}$ be a non-zero integer. Let us define: $$\mathbb Z\left[\frac{1}{n}\right] := \left\{\left.\frac{a}{n^r}\, \right\rvert\, a,r\in \mathbb{Z}, r \geq 0\right\}.$$ What are the prime and maximal ideals?

Hi I asked a question with this set and I was wondering the prime and maximal ideals.

I remember seeing a theorem in a textbook that says that if $R$ is a commutative ring with identity, then the prime ideals are the maximal ideals.

So in this case, this set is a subring of $\mathbb{Q}$ so that means the prime ideals in this set are the maximal ideals, correct?

So the definition of a prime ideal is that if we have $a,b \in R$, if $ab \in P$ then $a\in P$ or $b \in P$.

So what I started with was, suppose that $P$ is a prime ideal for $\mathbb{Z}[\frac{1}{n}]$. Let $\frac{a}{n^r}, \frac{b}{n^s} \in \mathbb{Z}[\frac{1}{n}]$.

Suppose that: $\frac{ab}{n^{r+s}} \in P$, that means $\frac{a}{n^r} \in P$ or $\frac{b}{n^s} \in P$.

But I'm stuck here. Any tips or hints to progress further?

$\endgroup$
  • 4
    $\begingroup$ Your claim is false: maximal ideals are always prime but not vice versa. $\endgroup$ – Zhen Lin Mar 6 '14 at 20:24
  • $\begingroup$ Ah, woops. I see, thanks. $\endgroup$ – user133458 Mar 6 '14 at 20:31
  • 1
    $\begingroup$ That's the localized $\mathbb{Z}_n$, no? In any book of commutative algebra, you 'll find all what you need about localisation. $\endgroup$ – Léo Mar 6 '14 at 20:37
  • $\begingroup$ Wait, if that's the localized $\mathbb{Z_{n}}$, that would just mean the prime ideal is just $\mathbb{Z_{(p)}}$ = {$\frac{a}{m^r}$ | $a,r \in \mathbb{Z}$ | $p \nmid m^r$}? $\endgroup$ – user133458 Mar 6 '14 at 20:47
1
$\begingroup$

First of all, as Zhen Lin observed, a prime ideal is not necessary maximal. Think to the ideal $(p) \subseteq \mathbb{Z}[x]$.

We can ask if this holds in some particular categories of rings.

It is not true if we chose UFD rings, in fact $\mathbb{Z}[x]$ is UFD and we have ever seen a counterexample. The assert is true for the PID Rings with only one exception: every non zero prime is maximal!(otherwise, if the $(0)$ ideal is prime the ring is a field...)

Now, your ring is the localization $S^{-1}\mathbb{Z}$ for the multiplicative set $S=\{n^i | i >0\}$.

Holds the following

Proposition Let $R$ be a ring and $S$ a multiplicative set of $R$. Then there's a bijection between the primes $Q$ of $S^{-1}R$ and the primes $P$ of $R$ such that $P \cap S = \emptyset$

This proposition is easy to prove and you can find it in any introductive book of Commutative Algebra. The idea is that, if an ideal $I$ meets $S$, its image in $S^{-1}R$ contains an invertible element and then is equal to all $S^{-1}R$.

Now we come to your question: what about the maximal ideals of $S^{-1}\mathbb{Z}$ fwhere $S=\{n^i | i >0\}$?

Well, the primes are all images of the prime ideals $P$ of $\mathbb{Z}$ such that $n^i \notin P$ The ideals of this forms are the image of $(0)$ or all the images of the ideals $(q)$ where $q$ is not a prime factor of $n$.

Which of this are maximals? Answer: every image of a non zero prime of $\mathbb{Z}$ as above.

There are a lot of metods to prove this, my favourite uses the following

Theorem If $0 \rightarrow I \rightarrow R \rightarrow R/I \rightarrow 0$ is an exact sequence and $S$ is a multiplicative subset of $R$, then $0 \rightarrow S^{-1}I \rightarrow S^{-1}R \rightarrow S^{-1}(R/I) \rightarrow 0$

Now you can easily prove that the ideals of the form above are maximal observing that $$ S^{-1}R /S^{-1}I \simeq S^{-1}(R/I) $$.

$\endgroup$
  • $\begingroup$ Your explanation was very thorough and straightforward! Thank you so much for your help. :) $\endgroup$ – user133458 Mar 7 '14 at 3:04
1
$\begingroup$

Sketch for a low-tech solution: show $\Bbb Z[\frac{1}{n}]$ is a PID ($\Leftarrow$ every ideal $(a,b)$ is principal $\Leftarrow \Bbb Z$ is a PID) and irreducibles $\pi\in\Bbb Z[\frac{1}{n}]$ are associate to irreducibles $p\in\Bbb Z$ (but which $p$ are units in $\Bbb Z[1/n]$?).

$\endgroup$

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.