0
$\begingroup$

Let $A$ be the power series ring $\mathbb{C}[[x,y]]$.

Assume we are given two ideals $I,J$ of finite length in $A$ such that:

  • $xJ\subseteq I\subseteq J$

Is it possible to find ideals of finite length $B,C$ in $A$ such that:

  • $xC\subseteq B\subseteq C$
  • $B\subset I$ and $I/B\cong R/J$
  • $C\subset J$ and $J/C\cong R/I$

Here are my thoughts so far: we see that $x$ annihilates $J/I$ so that this is a module of finite length over $\mathbb{C}[[y]]$, hence also some power of $y$ annihilates $J/I$. I thought maybe we should write $I=<f_1,\ldots,f_n>$ and $J=<h_1,\ldots,h_m>$ and see if we can use these generators to construct/find the wanted ideals, but this seems quite hard.

Is it even possible to find such ideals $B$ and $C$? Any ideas how to approach this problem are welcome. One could also ask more genrally:

Assume we are given $n$ ideals $I_r$ of finite length in $A$ such that:

  • $xI_r\subseteq I_1\subseteq I_r$ for all $r\geq 2$
  • for all $r\geq 2$: $xI_r\subseteq I_s\subseteq I_r$ for all $s\geq r+1$

Is it possible to find ideals of finite length $J_r\subset I_r$ in $A$ for $r=1\ldots n$ which satisfy:

  • $xJ_r\subseteq J_1\subseteq J_r$ for all $r\geq 2$
  • for all $r\geq 2$: $xJ_r\subseteq J_s\subseteq J_r$ for all $s\geq r+1$
  • $I_r/J_r\cong R/I_{r-1}$ ($r\geq 2$) and $I_1/J_1\cong R/I_n$
$\endgroup$
6
  • $\begingroup$ What do you mean when you speak of an ideal in $\mathbb{C}[[x,y]]$ of finite length? Do you mean finite colength, i.e., the quotient module has finite length? $\endgroup$ Jun 16, 2015 at 14:07
  • $\begingroup$ @Jason Starr: Yes, exactly. An ideal $I$ of finite length in $A$ is an ideal such that $A/I$ has finite length. $\endgroup$
    – Bernie
    Jun 16, 2015 at 15:18
  • 1
    $\begingroup$ In general, no. Take $I=J=(x,y)^2$. Then your third bulleted requirement want a surjection from $J$ to $R/I=R/J$. An elementary calculation will show that this is not possible. $\endgroup$
    – Mohan
    Jun 17, 2015 at 0:17
  • $\begingroup$ @Mohan: Do you mean something like this: every morphism $f: J \rightarrow R/J$ must have image in $(x,y)/J$, because we can make things like $[y]f(x^2)=f(yx^2)=[x]f(xy)$ and so on? So no morphism can be surjective in this case. $\endgroup$
    – Bernie
    Jun 17, 2015 at 11:35
  • $\begingroup$ Yes, the image of any such $f$ will be contained in $(x,y)$ and thus will not be onto. $\endgroup$
    – Mohan
    Jun 17, 2015 at 12:39

1 Answer 1

1
$\begingroup$

As I commented, in general, the answer is no. For an easy example, take $I=J=(x,y)^2$. Then you require a surjection $f:J\to R/I$. Letting $f(x^2)=a, f(xy)=b, f(y^2)=c$, since $y.x^2=x.xy$, we get $ya=xb$ and similarly, $yb=xc$, using $y.xy=x.y^2$. Since $ya, xb\in (x,y)/I$, which is a $k$ vector space with basis $x,y$, we see that $a,b\in (x,y)$. Similarly, we get $b,c\in (x,y)$. Thus $a,b,c\in (x,y)/I$ and then the map $f$ can not be onto.

$\endgroup$

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.