3
$\begingroup$

Let $k$ be a field and $I$ an ideal of $k[x,y]$ generated by two polynomials. Let $\mathcal{E}$ be the set defined by $$ \mathcal{E}=\{(P,Q)\in k[x,y]^2 | \langle P,Q\rangle =I\}.$$

The group $G=GL_2(k[x,y])$ acts naturally on $\mathcal{E}$. What can be said about the quotient?

For instance, if $I$ is a maximal ideal, then the quotient reduces to a point.

$\endgroup$

1 Answer 1

1
$\begingroup$

Disclaimer. This answer is far from complete. Some pointers are provided (see [1] and [2]) that could help go beyond the results presented below.

Rings are supposed to be commutative and unital. Given such a ring $R$, we denote by $R^{\times}$ its unit group. For $r, s \in R$, we denote by $(r:s)$ the ideal of $R$ consisting of the elements $\lambda$ such that $r$ divides $\lambda s$.

Claim. Let $R$ be a ring. Let $I$ be an ideal of $R$ be generated by two elements $a$ and $b$ which are not zero divisors. Let $$\operatorname{V}_2(I) = \{(r, s) \in R^2 \, \vert \, Rr + Rs = I\}$$ and let $J := \operatorname{Fitt}_1(I)$, that is, the second Fitting ideal of $I$. Then the following hold.

  1. If $I$ is principal, then $\vert \operatorname{V}_2(I)/\operatorname{SL}_2(R) \vert = 1$.
  2. The exterior product $\Lambda^2(I)$ is $R$-isomorphic to $R/J$ so that there is a unique $R$-alternating map $\det_{(a,b)}: I \times I \rightarrow R/J$ such that $\det_{(a, b)}(a, b) = 1 + J$.
  3. The set $\operatorname{V}_2(I)/\operatorname{SL}_2(R)$ is equipotent with $\det_{(a, b)}(\operatorname{V}_2(I)) \subseteq (R/J)^{\times}$. The set $\det_{(a, b)}(\operatorname{V}_2(I))$ contains the image of $R^{\times}$ under the natural map $R \twoheadrightarrow R/J$.
  4. If the natural map $(R/ (a:b))^{\times} \rightarrow (R/J)^{\times}$ is surjective, then $\det_{(a, b)}(\operatorname{V}_2(I)) = (R/J)^{\times}$.
  5. If $R$ is a GCD domain, e.g. $R$ is unique factorization domain (UFD), then $J = \frac{1}{\gcd(a, b)} I$.

The above claim proved useful for rings of Krull dimension at most $1$ [1]. It has nonetheless some modest bearing on the case $R = k[x, y]$; see the propositions below.

Proof.

  1. This follows from the straightforward and well-known identity $\vert \operatorname{V}_2(R)/\operatorname{SL}_2(R) \vert = 1$.
  2. By Exercise 20.9.i of [3], the ideal $J$ is the annihilator of $\Lambda^2(I)$, which is clearly a cyclic $R$-module. Hence there is a unique $R$-isomorphism $\phi: \Lambda^2(I) \rightarrow R/J$ such that $\phi(a \wedge b) = 1 + J$. The map $\det_{(a, b)}$ defined by $\det_{(a, b)}(c, d) = \phi(c \wedge d)$ where $c, d \in I$ has the desired properties.
  3. Apply [1, Theorem A].
  4. Apply [1, Lemma 4.1].
  5. Observe that $J = (a:b) + (b:a)$ and that $(a:b) = Ra$ if $\gcd(a, b) = 1$.

Proposition 1. Let $k$ be an algebraically closed field, let $R = k[x, y]$ and let $I$ be a maximal ideal of $R$. Then $\vert \operatorname{V}_2(I)/\operatorname{GL}_2(R) \vert = 1$.

Proof. Reason as in [1, Example 3.9] to infer that $\operatorname{V}_2(I)/\operatorname{SL}_2(R)$ identifies with $k^{\times}$.

Proposition 2. Let $k$ be a field and let $R = k[x, y]$. Let $f \in k[y]$ and let $I$ be the ideal of $R$ generated by $x$ and $f$. Then $\operatorname{V}_2(I)/\operatorname{SL}_2(R)$ is equipotent with $(R/I)^{\times} = (k[y]/(f))^{\times}$.

Proof. Use the assertions (4) and (5) of the above claim.

Example. Let $R = \mathbb{R}[x, y]$ and let $I$ be the ideal of $R$ generated by $x$ and $y^2 + 1$. By Proposition 2, the orbit set $\operatorname{V}_2(I)/\operatorname{SL}_2(R)$ is equipotent with the unit group of $R/I$. As $R/I \simeq \mathbb{R}[y]/(y^2 + 1) \simeq \mathbb{C}$, we infer that $\operatorname{V}_2(I)/\operatorname{SL}_2(R) \simeq \mathbb{C}^{\times}$ and it easily follows that $$\operatorname{V}_2(I)/\operatorname{GL}_2(R) \simeq \mathbb{C}^{\times} / \mathbb{R}^{\times}.$$


  • [1] L. Guyot, "Equivalent generating pairs of an ideal of a commutative ring", 2020.
  • [2] M. Murthy, "Generators of a general ideal", 2003.
  • [3] D. Eisenbug, "Commutative Algebra", 1995.
$\endgroup$
1
  • $\begingroup$ @L. Guyot : Thanks a lot for your answer and the references ! I will have a close look at it. Hope it will be enough for my purpose :). Yohann $\endgroup$ Commented Apr 14 at 13:41

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.