Let $I$ be an ideal in a Noetherian quasi-unmixed local ring $(R,\mathfrak m)$ of dimension $d.$

Let $q(I)=\overline{I}\cap I^{sat}$ where $I^{sat}=\cup_{n\geq 0}I:\mathfrak m^n.$ Then $q(I)$ is called relative integral closure of $I.$

Question Is there any example of a principal ideal $I$ (i.e. $I=(a)$) such that $I\neq q(I)$ where $d\geq 2?$

I know that in normal domains, $I=q(I)$ for all principal ideals $I.$

  • $\begingroup$ It looks to me like you already get an example for $R=(k[x,y]/\langle \ y^2-x^3 \rangle)_{\langle \overline{x},\overline{y} \rangle}$ and $I=\langle \overline{x} \rangle$. $\endgroup$ – Jason Starr Jan 20 '17 at 8:48
  • 4
    $\begingroup$ Wait a minute: was that question supposed to be a riddle involving your username? $\endgroup$ – Jason Starr Jan 20 '17 at 8:50
  • $\begingroup$ @JasonStarr actually I am trying to find an example in higher dimensional rings $\endgroup$ – Cusp Jan 20 '17 at 9:14
  • $\begingroup$ When you say that you are trying to find an example in higher dimensional rings, are you trying to find an example that satisfies $R1$ but not $S2$? $\endgroup$ – Jason Starr Jan 20 '17 at 9:19
  • $\begingroup$ @JasonStarr I want an example where the ring is such that the principal ideal is not integrally closed and I is different from its relative integral closure. $\endgroup$ – Cusp Jan 20 '17 at 9:27

There are examples in arbitrary dimension. Begin with the polynomial ring $S=k[N]=k[s_1,\dots,s_d]$, i.e., the semigroup $k$-algebra on the semigroup $N=(\mathbb{Z}_{\geq 0})^d$ of exponent vectors for monomials in $S$. Now for any subsemigroup $M\subset N$, let $R=k[M]\subset k[N]$ be the corresponding $k$-subalgebra of $S$. These give plenty of examples.

For instance, let $R$ be the $k$-subalgebra generated by monomials of total degree $\geq d+1$. Let $\mathfrak{m}$ be the maximal ideal generated by all monomials of total degree $\geq d+1$. Let $I$ be the principal ideal generated by $x=s_1^2s_2\cdots s_d$. Then the element $y=s_1^3s_2\cdots s_d$ in $R$ is in $q(I)$, but it is not in $I$. Indeed, $y^{d+1}$ equals $x^{d+1}\cdot s_1^{d+1}$, so $y$ is in the integral closure of $I$. Also, for every monomial $m$ of degree $\geq d+1$, then $my$ equals $(s_1m)x$, where $s_1m$ is another monomial in $R$. Thus $y$ is also in the saturation of $I$ with respect to the maximal ideal $\mathfrak{m}$. Please note also: $R$ is regular in codimension $\leq d-1$, but it is definitely not $S2$.

There are some exercises in Eisenbud's "Commutative Algebra" that discuss these types of rings and their integral closures.

  • $\begingroup$ Here by codimension did you mean embedding dimension (maximal homogeneous ideal ) minus dim R? $\endgroup$ – Cusp Jan 20 '17 at 11:30
  • $\begingroup$ By codimension I really meant the height of the corresponding prime ideal. You are correct that I should localize the ring $R$ at the maximal ideal $\mathfrak{m}$. Since the finitely generated $R$-module $S/R$ is annihilated by $\mathfrak{m}$, for every prime ideal $\mathfrak{p}$ with $\mathfrak{p}\subsetneq \mathfrak{m}$, the localization $R_{\mathfrak{p}}$ equals $S_{\mathfrak{p}}$. Thus $R_{\mathfrak{p}}$ is regular for all prime ideals $\mathfrak{p}R_{\mathfrak{m}}$ of $R_\mathfrak{m}$ except the maximal ideal. $\endgroup$ – Jason Starr Jan 20 '17 at 12:10

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.