1
$\begingroup$

Let $I$ be any ideal in $R=\mathbb C[x,y]$ of height 2. If we know the localization of $I$ at any maximal ideal $m\supseteq I$ is generated by a regular sequence of length two in $R_m$. Is this true that $I$ itself is generated by a regular sequence of length two in $R$?

$\endgroup$
3
  • $\begingroup$ It's pretty late here, so I hope this isn't completely nonsensical. A Gorenstein ring is Cohen-Macaulay and both $R_m$ and $R$ are local Gorenstein and so local Cohen-Macaulay. For such rings any ideal $I$ has height equal to the depth of $I$ with respect to $I$. See e.g. en.wikipedia.org/wiki/Height_(ring_theory). Also, all regular sequences in $I$ have length equal to the depth of $I$, see en.wikipedia.org/wiki/Depth_(algebra). Now, $R$ is already local, so it seems height = depth = length of longest regular sequence even without this bit about $R_m$. How does that sound? $\endgroup$ Aug 10, 2012 at 3:52
  • $\begingroup$ What do you mean by '$R$ is already local'? $\endgroup$
    – J.C. Ottem
    Aug 10, 2012 at 7:49
  • $\begingroup$ Could you give the reference for the claim: If $R$ is local then $I$ generated by a regular sequence? $\endgroup$ Aug 10, 2012 at 12:45

2 Answers 2

4
$\begingroup$

The quickest way to see this is the following. If $I_{\mathfrak{m}}$ is generated by a regular sequence, using Koszul complex, one gets that $Ext^1(I,R)$ is locally isomorphic to $R/I$ and thus by Chinese remainder theorem, globally isomorphic to $R/I$. Then one checks that the extension corresponding to $1\in Ext^1(I,R)$, say $0\to R\to P\to I\to 0$ has $P$ a projective $R$-module of rank 2 and by Seshadri's theorem $P$ is free. Thus $I$ is 2-generated. Easy to arrange the generators to be a regular sequnce.

$\endgroup$
2
  • $\begingroup$ Thanks for your reply. Could you explain in details why the identity $1\in Ext^1(I,R)$ corresponds to a ses $0\rightarrow R\rightarrow P\rightarrow I\to 0$ where $P$ is projective? $\endgroup$
    – Xingting
    Aug 14, 2012 at 0:49
  • $\begingroup$ Let me describe one of may ways of seeing this. By the very nature of it, $P$ is torsion free and hence has depth at least one. That it comes from a generator of $Ext^1(I,R)$ also implies $Ext^1(P,R)=0$. If $P$ is not projective, by Auslander-Buchsbaum formula, $pd P$ must be one. Writing a resolution $0\to A\to B\to P\to 0$ with $A,B$ free, using the fact that $Ext^1(P,R)=0$, one easily checks that $P$ must be projective. $\endgroup$
    – Mohan
    Aug 14, 2012 at 13:31
0
$\begingroup$

Hi. I'm a little bit confused by the above example. The way I looked at this is through Forster-Swan number.

[Forster] Let $R$ be a Noetherian ring and $M$ a finite $R$-module. Then $M$ can be generated by $b(M)$ elements.

Here $b(M) = 0$ if $M = 0$ and $b(M) =$ sup { $\mu(M_p) + $ dim $R/p$ } where $p \in $ supp$(M)$ if $M \neq 0$.

If we take $R = \mathbb{C}[x,y]$ and $M = I$ as in the question. Then $b(M) = 2$. So $I$ can be generated by two elements.

$\endgroup$
2
  • $\begingroup$ In this case $b(M)=3$, since $0\in Supp(M)$ where $M=I$. In fact, Forster-Swan does not use the fact you are working over a polynomial ring and in general, you will not be able to do better than 3 for arbitrary rings of dimension 2. $\endgroup$
    – Mohan
    Aug 11, 2012 at 12:05
  • $\begingroup$ Yes. You are right. It is 3 not 2. Thank you for the clarification. $\endgroup$
    – Youngsu
    Aug 13, 2012 at 2:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.