2
$\begingroup$

Fix $n > 1$ and let $\zeta \in \mathbb{C}$ be a primitive $n$-th root of unity. Let $G \subset \text{SL}_2(\mathbb{C})$ be a cyclic subgroup of order $n$ generated by the diagonal matrix $g = \text{diag}(\zeta, \zeta^{-1})$. The group $G$ acts naturally on $\mathbb{C}[x, y]$. Let $\mathbb{C}[x, y]^G$ be the algebra of $G$-invariant polynomials, namely $$\mathbb{C}[x, y]^G = \{f \in \mathbb{C}[x, y] : f(\zeta^{-1}x, \zeta y) = f(x, y)\}.$$What is the description of the algebra $\mathbb{C}[x, y]^G$ in terms of generators and relations?

$\endgroup$
  • 1
    $\begingroup$ That is, roughly, a cone over a rational normal curve. The generators are $u = xy$ and then $v_{l,m} = x^l y^m$ for nonnegative integers $l$, $m$ such that $l + m$ equals $n$. There are some obvious relations among the $v_{l,m}$, namely $v_{l,m}v_{l',m'} = v_{l'',m''}v_{l''',m'''}$ if $l+l'$ equals $l''+l'''$ and $m+m'$ equals $m''+m'''$. If $n$ is even, then there is a relation $u^{n/2} = v_{n/2,n/2}$. If $n$ is odd, then the relation is $v_{l,m}v_{m,l} = u^n$. $\endgroup$ – Jason Starr Dec 5 '15 at 14:41
  • 1
    $\begingroup$ Francesco is correct: some of my relations were incorrect. I was confusing the $(\zeta^{-1},\zeta)$-action with the $(\zeta,\zeta)$-action. $\endgroup$ – Jason Starr Dec 5 '15 at 15:02
  • 1
    $\begingroup$ To correct what I wrote: among the $v_{l,m}$ I wrote down, only $v_{n,0}$ and $v_{0,n}$ are invariant. These satisfy the single relation $v_{n,0}v_{0,n} = u^n$. Sorry about the mistake. $\endgroup$ – Jason Starr Dec 5 '15 at 15:04
  • 1
    $\begingroup$ @JasonStarr: I was confused, too, and my answer was wrong. I've edited it, and now it should be ok. $\endgroup$ – Francesco Polizzi Dec 5 '15 at 19:34
6
$\begingroup$

EDIT. This is a new version of the answer. Since at some point there was confusion between the $(\zeta, \, \zeta)$-action and the $(\zeta^{-1}, \, \zeta)$-action, for the sake of clarity let me discuss both.

The $(\zeta^{-1}, \, \zeta)$-action (the OP case).

In this case there are three invariant monomials, namely $$T_0:= x^n, \quad T_1=xy, \quad T_2=y^n.$$ Then the invariant algebra $\mathbb{C}[x, \, y]^G$ is isomorphic to $$\mathbb{C}[T_0, \, T_1, \, T_2]/(T_0T_2-T_1^n),$$ which is is a Rational Double Point (= Du Val singularity) of type $A_{n-1}$, whose resolution graph is the Dynkin diagram with the same label, corresponding to a chain of $n-1$ smooth rational curves with self-intersection $(-2)$. This is a Gorenstein singularity for all $n \geq 2$, for instance because it is a hypersurface singularity, or because $G \subset \textrm{SL}_2(\mathbb{C})$.

The $(\zeta, \, \zeta)$-action.

In this case the quotient $\mathbb{C}^2/G$ is isomorphic to the affine cone over a rational normal curve $C_n \subset \mathbb{P}^n$.

Therefore $\mathbb{C}[x, \, y]^G$ is the ring of regular functions of such a cone, namely the ring $\mathbb{C}[T_0, \ldots, T_n]/I$, where $I$ is the ideal generated by the $2 \times 2$ minors of the matrix $$\begin{equation} \begin{pmatrix} T_0 & T_1 & \ldots & T_{n-1}\\ T_1 & T_2 & \ldots & T_n \end{pmatrix}. \end{equation}$$

Here the variables $T_i$ correspond to the invariant monomials for the action, namely $T_i:=x^{n-i}y^i$. The singularity at the vertex of the cone is of type $\frac{1}{n}(1, \, 1)$, has embedding dimension $n+1$ and is not Gorenstein for $n \geq 3.$ The resolution is a unique smooth rational curve with self-intersection $(-n)$.

Remark.

The two cases coincide for $n=2$, when the action is simply given by $(x, \, y) \to (-x,\, -y)$. The invariant algebra is generated by the three invariant polynomials $$T_0:=x^2, \quad T_1:=xy, \quad T_2:=y^2$$ subject to the unique relation $T_0T_2-T_1^2=0,$ which is the affine cone over a smooth conic in $\mathbb{P}^2$.

| cite | improve this answer | |
$\endgroup$
  • 2
    $\begingroup$ I'm totally confused; what are your generators $T_i$? Even if this presentation is right, it's extremely redundant: as Jason said, this is a hypersurface in $\mathbb{A}^3$ with the equation $uw=z^n$. $\endgroup$ – Ben Webster Dec 5 '15 at 17:26
  • 1
    $\begingroup$ @BenWebster: I think that was my fault. My comment confused the $(\zeta^{-1},\zeta)$-action and the $(\zeta,\zeta)$-action. Once one person says something wrong, it is hard to realize that is not what the OP asked. $\endgroup$ – Jason Starr Dec 5 '15 at 18:03
  • 1
    $\begingroup$ @BenWebster: you are right. I was confused, too. I will correct the answer. $\endgroup$ – Francesco Polizzi Dec 5 '15 at 19:11
  • 1
    $\begingroup$ @JasonStarr: as I said before, the previous version of the answer was wrong. I hope that now all mistakes are corrected. $\endgroup$ – Francesco Polizzi Dec 5 '15 at 19:36

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.