Let $R$ be a commutative ring with $1\neq 0$. $R$ is said to be von Neumann regular if for all $a\in R$, there is some $x\in R$ such that $a^2x=a.$ Prove that if $R$ is von Neumann regular and $P$ a prime ideal, then $P$ is maximal.

My idea: We know that $P$ is a prime ideal and $R$ is a commutative ring, so $R/P$ is an integral domain. If we can show that $R/P$ is a field, then $P$ is maximal. Further, every finite integral domain is a field, although I am not sure it will be helpful here.

Any suggestions/comments/answers are welcome. Thanks.

  • $\begingroup$ This doesn't jive with the usual definition of regular, right? $\endgroup$ Jul 8, 2012 at 2:52
  • 5
    $\begingroup$ @Dylan: these are the von Neumann regular rings: see e.g. en.wikipedia.org/wiki/Von_Neumann_regular. As the article says, they are precisely the absolutely flat rings, i.e., the rings for which every module is flat. Maybe this terminology is safer... $\endgroup$ Jul 8, 2012 at 3:31
  • $\begingroup$ @PeteL.Clark Wonderful, thanks! $\endgroup$ Jul 8, 2012 at 3:32
  • $\begingroup$ This is proved without using quotient rings in this answer $\endgroup$
    – Xam
    Sep 10, 2017 at 16:45

3 Answers 3


This answer is the same as Zev's, but perhaps stated more "conceptually". For what it's worth:

A commutative ring is von Neumann regular if for all $a \in R$, there is $x \in R$ such that $a^2 x = a$.

Here are two straightforward facts:

Fact 1: Every quotient of a von Neumann regular ring is von Neumann regular.

[The defining condition is an identity, and if an identity holds in a ring it holds in any quotient.]

Fact 2: An integral domain which is von Neumann regular is a field.

[Fact 2 is literally the first thing that springs to mind when I see the somewhat strange defining condition. What does $a^2 x = a$ mean? Well, if we're allowed to cancel the $a$'s, it means $ax = 1$!]

Thus if $\mathfrak{p}$ is a prime ideal in a von Neumann regular ring, $R/\mathfrak{p}$ is a von Neumann regular domain, hence a field, so $\mathfrak{p}$ is maximal.

  • 1
    $\begingroup$ Damn, this is also nice, smooth and beautiful! Twice such answers to the same question...lucky OP! +1 , of course. $\endgroup$
    – DonAntonio
    Jul 8, 2012 at 10:28

You're on the right track! For any prime ideal $P\in R$ and $a\notin P$, we have $a^2x=a$ in $R$ for some $x\in R$, so that in $R/P$, we have $\bar{a}^2\bar{x}=\bar{a}$ (where $\bar{s}$ means the equivalence class $s+P$). Do you see how to proceed?

Rest of solution (mouse over to reveal):

Rewriting, we have$$\bar{a}^2\bar{x}-\bar{a}=\bar{a}(\bar{a}\bar{x}-\bar{1})=\bar{0}.$$Because $a\notin P$, we have $\bar{a}\neq\bar{0}$, so that because $R/P$ is an integral domain, we can conclude $\bar{a}\bar{x}-\bar{1}=0$. Thus any $\bar{a}\neq\bar{0}$ in $R/P$ has an inverse, so $R/P$ is a field.

  • $\begingroup$ @ZevChonoles Thanks! That seems simple after I look at the solution:) $\endgroup$
    – Lyapunov
    Jul 8, 2012 at 2:16
  • $\begingroup$ That mouse-over thing is cool! I had no idea you could do that. $\endgroup$ Jul 8, 2012 at 4:01
  • $\begingroup$ @Keenan: It's quite useful for making complete answers that don't give everything away all at once :) The Markdown help page describes the code for it (towards the bottom). $\endgroup$ Jul 8, 2012 at 6:06

Let's show that every $a + P\in \dfrac{R}{P}$ such that $a + P \neq 0 + P$ is an invertible element. Since $R$ is a von Neumann regular ring, for such an $a \in R$ there must exist some $b \in R$ such that $aba = a$, so $a+P = aba + P$, so $a+P = (ab + P)\cdot (a + P)$, so $ab + P = 1 + P$, hence $(a + P)^{-1} = b+P$. Since every nonzero element of $\dfrac{R}{P}$ is invertible, it follows that $\dfrac{R}{P}$ is a field, so $P$is a maximal ideal of $R$.


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.