One of the equivalent forms in definition of a Noetherian Jacobson ring $R$ is that $R$ has no prime ideals $P$ such that $R/P$ is a 1-dimensional semi-local ring.

When $R/P$ has dimension 1, it means that between $R$ and $P$ there is at most one prime ideal, and as $R/P$ is semi-local, there are only finitely many maximal ideals of $R$ containing $P$. Do not these two imply that $P$ lies only in one maximal ideal of $R$?

Incidentally, what is a proof for equivalence of the above-mentioned statement with $R$ being Jacobson? Thanks for any cooperation.


Let $R$ be a noetherian ring. Then $R$ is Jacobson if and only if for every prime ideal $P$ of $R$ such that $\dim R/P=1$ the ring $R/P$ has infinitely many maximal ideals.

"$\Rightarrow$" Actually any homomorphic image of dimension $\ge 1$ of a Jacobson ring has infinitely many maximal ideals. Since homomorphic images of Jacobson rings are also Jacobson, we can start with a Jacobson ring $R$ of dimension $\ge1$, and prove that $R$ has infinitely many maximal ideals. Suppose the contrary, and let $P$ be a prime ideal of $R$ which is not maximal. Then $P$ is a finite intersection of maximal ideals, and therefore $P$ itself is maximal, a contradiction.

"$\Leftarrow$" If $R$ isn't Jacobson, then there is a prime ideal which is not an intersection of maximal ideals. Since $R$ is noetherian we can choose a prime $P$ maximal with this property. In particular, $P$ isn't maximal, so $\dim R/P\ge 1$.
If $\dim R/P=1$, then, by hypothesis, the intersection of its maximal ideals is $(0)$. (Otherwise, taking a non-zero element $x$ in the intersection, and using the primary decomposition of $(x)$ we find that $x$ is contained in a finite number of maximal ideals, contradiction.) Thus $P$ is an intersection of maximal ideals, and this contradicts the choice of $P$.
If $\dim R/P>1$, then there are infinitely many height one prime ideals in $R/P$ and (by a similar argument to the one in the previous paragraph) their intersection is $(0)$, that is, $P$ is an intersection of larger primes. From the choice of $P$ we deduce that $P$ is an intersection of maximal ideals, contradiction.

  • $\begingroup$ An alternative approach for $\Leftarrow$: In a domain $R$, if $x \ne 0$ is contained in every height $1$ prime, then $\{\text{ht 1 primes}\} \subseteq \operatorname{Min}(x)$. If $R$ is Noetherian of dim $> 1$, then $\{\text{ht 1 primes}\}$ is infinite while $\operatorname{Min}(x)$ is finite, and if $\dim R = 1$, then $\{\text{ht 1 primes}\} = \operatorname{mSpec}(R)$. $\endgroup$ – zcn Jun 20 '14 at 18:49
  • $\begingroup$ @zcn Why $Min(x)$ is finte? And what is the contradiction in the case $dim R=1$? $\endgroup$ – karparvar Jun 2 '15 at 8:26
  • $\begingroup$ @karparvar In a noetherian ring the set of minimal primes over an ideal is always finite (they are among the associated primes). (Unfortunately user zcn is no longer active on M.SE.) $\endgroup$ – user26857 Jun 2 '15 at 13:52

There are Noetherian semi-local domains of dimension $1$ that are not local, e.g. $S^{-1}\mathbb{Z}$, where $S = \mathbb{Z} \setminus ((2) \cup (3))$. Probably the line of reasoning you meant is this: a Jacobson domain of dimension $1$ cannot be semilocal. If it were, then $0$ would be an intersection of maximal ideals, and if the intersection is finite, then $0$ would equal one of the maximal ideals, contradicting dimension $1$.

Since quotients of Jacobson rings are Jacobson, this shows that $R$ Jacobson $\implies R$ has no primes $P$ with $R/P$ semi-local of dimension $1$. For the converse, assuming $R$ is Noetherian, one can use the equivalence (Lemma 4.20 in Eisenbud, called the Rabinowitch trick)

i) $R$ is Jacobson
ii) If $P$ is a prime in $R$ with $(R/P)[f^{-1}]$ a field for some $f \in R \setminus p$, then $R/P$ is a field

along with the following proposition (see this question):

Proposition: Let $R$ be a Noetherian domain, not a field. Then $R$ is semi-local of $\dim 1$ iff $R_f$ is a field for some $f \in R$.


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.