Let $A$ be a regular ring and $\mathfrak q$ be an ideal, such that $\sqrt{\mathfrak q}$ is prime. Further assume that $\mathfrak q$ is locally principal (i.e. $\mathfrak q$ is an irreducible divisor which is not necessarily reduced).

Now we assume that $(A/\mathfrak q)_{\sqrt{\mathfrak q}}$ is regular.

Under these hypotheses, we can deduce that $A/\mathfrak q$ is reduced (equivalently $\mathfrak q$ itself is prime) because it is a generically reduced (since $(A/\mathfrak q)_{\sqrt{\mathfrak q}}$ is a zero-dimensional regular local ring, hence a field) Cohen-Macaulay-Ring (since $A$ is regular and $\mathfrak q$ is locally generated by non-zerodivisors).

The geometric interpretation is: If an irreducible component of a divisor $D$ in a regular variety contains a regular point of $D$, then it is a reduced component.

My 2 questions:

1) Can anyone come up with a direct proof (for example not using words like Cohen-Macaulay), which is more elementary?

2) What if we drop the assumption that $\mathfrak q$ is locally principal? Then the question comes down to: Let $B$ be a ring (which is a quotient of a regular local Ring) with prime nilradical such that the localization at the nilradical is a field. Does this imply that $B$ is reduced, hence a domain? If not, can someone come up with a counterexample?

All rings can assumed to be Noetherian.


Ok, $R=k[x,y]/(x^2,xy)$ should be a counterexample to question 2). We have $nil(R)=(x)/(x^2,xy)$ and $R/nil(R) = k[y]$, so $nil(R)$ is prime. Since we have $xy=0$ in $R$, we deduce $x=0$ in $R_{nil(R)}$, thus the maximal ideal of $R_{nil(R)}$, which is generated by $x$, is zero.

Question 1) is now less interesting because there is no hope to come up with a proof that shows that the "locally principal"-assumption is not necessary at all.


A Noetherian ring is reduced if and only if it is $(R_0)$ and $(S_1)$. (See, for example, http://stacks.math.columbia.edu/tag/031R ) This condition can be used to provide a simple proof that your ring $A/\mathfrak{p}$ is reduced.

A ring $R$ is $(R_k)$ iff $A_P$ is a regular local ring for all primes $P$ of height at most $k$. So $R$ is $(R_0)$ iff $R_P$ is regular for all minimal primes $P$. This is a simple condition and clearly holds for your ring $R = A/{\mathfrak q}$ since it has a unique minimal prime and its localization at that prime is a field. The assumption that $\mathfrak{q}$ is locally principal is not required here.

A ring $R$ is $(S_k)$ iff $PR_P$ contains a regular sequence in $R_P$ of length at least $\min(k,\mbox{ht} P)$ for every prime $P$ of $R$. So $R$ is $(S_1)$ iff $PR_P$ contains a nonzerodivisor in $R_P$ for every non-minimal prime $P$ of $R$. This condition also clearly holds for your ring $R = A/\mathfrak{q}$. In general the condition required beyond $(R_0)$ is precisely the condition $(S_1)$.

| cite | improve this answer | |
  • $\begingroup$ Thank You for your answer. Actually I am aware of this criteria for a Noetherian ring to be reduced and it was precisely what I had in mind when giving the proof. $(R_0)$ is just another term for "generically reduced" and $(S_1)$ is kind a weaker version of Cohen-Macaulay. Actually in my counterexample $A/\mathfrak q$ was one-dimensional, hence Cohen-Macaulay and $(S_1)$ coincide in this ring. Maybe this is the remaining the question now: Is there another sufficient condition in terms of the ideal $\mathfrak q$, that suffices to show that $A/\mathfrak q$ is reduced? $\endgroup$ – MooS Jan 22 '15 at 8:09
  • $\begingroup$ If you're looking for a condition on $\mathfrak{q}$, then one such necessary and sufficient condition is that $\mathfrak{q}$ be primary. (Indeed, $\mathfrak{q}$ is primary if and only if the map $A/\mathfrak{q} \longrightarrow (A/\mathfrak{q})_{\mathfrak{p}}$ is injective, which since the image is a field implies that $\mathfrak{q}$ is prime.) Under the assumption that $A/\mathfrak{q}$ is $(R_0)$, the ideal $\mathfrak{q}$ is prime iff it is primary iff it is radical. I'm not sure what else can be said. $\endgroup$ – Jesse Elliott Jan 22 '15 at 10:32
  • $\begingroup$ This clarifies everything. Thanks a lot! $\endgroup$ – MooS Jan 22 '15 at 13:16

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.