15

I've got to prove that an ideal $Q$ whose radical is a maximal ideal is a primary ideal. That is, I want to prove that if $xy\in Q$, then $x\in Q$ or $y^n\in Q$ for some $n>0$.

I've been trying for a while and I'm not sure where to begin. All I've got is that if $\text{Rad}(Q)$ is maximal and by definition it's the intersection of all prime ideals $P_i$ containing $Q$, then it must be equal to each of these $P_i$. So there is only 1 prime ideal containing $Q$, namely $\text{Rad}(Q)$.

Could anyone point me in the right direction? Thanks for any replies.

Lammey
  • 1,363

4 Answers4

19

There's not too much to the proof:

Let $\sqrt{Q}$ be a maximal ideal $\mathfrak{m}$. Then $Q$ is $\mathfrak{m}$-primary.

$\it{Proof}$: Suppose that $\alpha \beta \in Q$ and $\beta$ is not in $\sqrt{Q}=\mathfrak{m}$. Then by the maximality of $\mathfrak{m}$, it follows that $\mathfrak{m} +R\beta=R$ where $R$ is a ring. Then, for some $m \in \mathfrak{m}$ and $r \in R$ we have $m+r\beta=1$. Now $m \in \mathfrak{m}= \sqrt{Q}$, hence $m^{n} \in Q$ for $n \geq 1$. Thus, $1=1^{n}=(m+r\beta)^{n}=m^{n} +s\beta$ for some $s \in R$. Multiply by $\alpha$ to get $\alpha=\alpha m^{n}+s\alpha \beta \in Q$.

user 3462
  • 1,493
  • 7
  • 16
  • What does the notation $R\beta$ mean? Sorry I'm a bit new to this notation. – Lammey Jan 23 '14 at 20:27
  • It usually means the set ${ r\beta \mid r \in R }$, which is also the ideal generated by $\beta$. See https://math.stackexchange.com/a/1801984/577979 – eatfood Feb 03 '20 at 11:20
9

Hints:

Reflect the problem in the quotient ring $R/Q$. Prove (if you don't already believe) that the definition of $Q$ being primary is equivalent to $R/Q$ satisfying "$ab=0$ implies $a=0$ or $b$ is nilpotent."

The condition that $\sqrt{Q}$ is maximal implies that $R/Q$ has exactly one maximal ideal, which is $\sqrt{Q}$. Thus it's a local ring.

Let $ab=0$ in $R/Q$ and suppose $b\notin \sqrt{Q}$. What can you say that is special about $b$ if it is outside the maximal ideal of a local ring? Conclude that $a=0$.

rschwieb
  • 153,510
  • I thought $R/Q$ was the set of ${r+Q|r\in R}$? I'm not sure what you mean by" $b$ isn't nilpotent in $R/Q$." Is $b$ an element of $R/Q$? THanks for the reply by the way. – Lammey Jan 23 '14 at 20:26
  • Yeah, $R/Q$ is the quotient ring of $R$ by the ideal $Q$. There is a little to unpack here :) Basically "no power of $b$ is in $Q$" is equivalent to "no power of $b+Q$ is $0+Q$ in $R/Q$." That's what I mean by "$b$ isn't nilpotent in $R/Q$." I should rather say that $b+Q$ isn't nilpotent in $R/Q$. – rschwieb Jan 23 '14 at 21:15
  • Oh ok I see, thanks for clarifying! I don't really understand the first line though - I'm not sure how the hypothesis implies what you say it does. I mean I believe you but I can't write it down as a proof without understanding it, because I'm going to be questioned on it! Thanks again for the reply. – Lammey Jan 23 '14 at 22:27
  • @JamesMachin If there were two different maximal ideals containing $Q$, then the radical would be contained in their intersection (the radical is the intersection of all prime ideals containing $Q$.) That would be smaller than a maximal ideal, so it's not possible that there is more than one maximal ideal above $Q$. – rschwieb Jan 23 '14 at 23:54
  • Sorry I meant I'm not sure what $R/Q$ has to doing with anything! As in, why does the hypothesis imply that $R?Q$ has a unique maximal ideal that corresponds to the radical of Q. I understand that there is only one prime ideal containing Q though! – Lammey Jan 24 '14 at 00:20
  • @JamesMachin But maximal ideals are prime! If you believe there's only one prime ideal containing $Q$, then you believe there's only one maximal ideal containing $Q$. – rschwieb Jan 24 '14 at 01:17
5

Here is a different proof. Assume $xy \in Q$ but $x \notin \sqrt{Q}$. Since no maximal ideal contains both $Q$ and $(x)$, we have $Q + (x) = R$. Thus $(y) = y(Q + (x)) \subseteq Q + (xy) = Q$.

Jason Juett
  • 1,125
2

Assume that $R$ is a commutative ring and $I⊆R$ is an ideal, such that the radical $r(I)$ of I is a maximal ideal. Then $I$ is a primary ideal. Proof. We will show, that every zero divisor in $R/I$ is nilpotent First of all, recall that $r(I)$ is an intersection of all prime ideals containing $I$. Since $r(I)$ is maximal, it follows that there is exactly one prime ideal $P=r(I)$ such that $I⊆P$. In particular the ring $R/I$ has only one prime ideal (because there is one-to-one correspondence between prime ideals in $R/I$ and prime ideals in $R$ containing $I$). Thus, in $R/I$ an ideal $r(0)$ is prime.Now assume that $α∈R/I$ is a zero divisor. In particular $α≠0+I$ and for some $β≠0+I∈R/I$ we have $αβ=0+I.$ But $0+I∈r(0)$ and $r(0)$ is prime. This shows, that either $α∈r(0)$ or $β∈r(0)$. Obviously $α∈r(0)$ (and $β∈r(0)$), because $r(0)$ is the only maximal ideal in $R/I$ (the ring $R/I$ is local). Therefore elements not belonging to $r(0)$ are invertible, but $α$ cannot be invertible , because it is a zero divisor. On the other hand $r(0)=\{x+I∈R/I: (x+I)^n=0 \ \ for \ some \ \ n∈\mathbb{N}\}$. Therefore $α$ is nilpotent and this completes the proof.

mrMath
  • 57