Rationality of the moduli spaces of plane curves of sufficiently large degree (Q2655170)

Let \(G := \text{SL}_3({\mathbb C})\), \(\overline{G} := \text{PGL}_3({\mathbb C})\), \(V = {\mathbb C}^3\) be the standard representation of \(G\), \(e_1,e_2,e_3\) the canonical \(\mathbb C\)-basis of \(V\) and \(x_1,x_2,x_3\) the dual basis of \(V^{\vee}\). For any positive integer \(d\), let \(V(d)\) denote the irreducible representation \(\mathrm{Sym}^d(V^{\vee})\) of \(G\). In the paper under review, the authors show that the quotient space \(\text{Hyp}(d,2) := {\mathbb P}(V(d))^{\text{ss}}/\overline{G}\) is rational for \(d \equiv 1\) (mod 3), \(d \geq 37\) and also for \(d \equiv 2\) (mod 3), \(d \geq 65\). Combining this result with a result of \textit{P. Katsylo} [Math. USSR, Sb. 64, No. 2, 375--381 (1989; Zbl 0679.14028)] asserting that \(\text{Hyp}(d,2)\) is rational for \(d \equiv 0\) (mod 3), \(d \geq 1821\), one gets that \(\text{Hyp}(d,2)\) is rational for \(d \gg 0\). The structure of the proof is similar to that used by \textit{N.I. Shepherd-Barron} [Compos. Math. 67, No. 1, 51--88 (1988; Zbl 0661.14022)] to settle the case \(d \equiv 1\) (mod 9), \(d \geq 19\), but the computational aspects of the proof are quite different. One starts with the fact, proved by Shepherd-Barron, that the quotient map \({\mathbb P}(V(4))^{\text{ss}} \rightarrow {\mathbb P}(V(4))^{\text{ss}}/\overline{G}\) (resp., \({\mathbb P}(V(8))^{\text{ss}} \rightarrow {\mathbb P}(V(8))^{\text{ss}}/ \overline{G}\)) is a principal \(\overline{G}\)-bundle in the Zarisky topology over a non-empty open subset of the quotient variety. Then one constructs, using the \textit{symbolic method}, a \(G\)-equivariant map \(S_d : V(d) \rightarrow V(4)\) for \(d = 3n+1\) (resp., \(T_d : V(d) \rightarrow V(8)\) for \(d = 3n+2\)) defined by homogeneous polynomials of degree 4. \(S_d\) appears already in the paper of Shepherd-Barron. The authors need, actually, from the symbolic method only the following observation: for every homogeneous polynomial \(P\) of degree \(d\) in three indeterminates, there exists a unique \(\mathbb C\)-linear function \({\ell}_P : V(d) \rightarrow {\mathbb C}\) such that \({\ell}_P({\alpha}_x^d) = P(\alpha)\), \(\forall \alpha = ({\alpha}_1,{\alpha}_2,{\alpha _3}) \in {\mathbb C}^3\), where \({\alpha}_x := {\alpha}_1x_1 + {\alpha}_2x_2 + {\alpha}_3x_3 \in V^{\vee}\). More precisely, one has the formula: \[ {\ell}_P = \frac{1}{d!}\, P\left( \frac{\partial}{\partial x_1}, \frac{\partial}{\partial x_2}, \frac{\partial}{\partial x_3} \right)\;. \] For \(\alpha ,\beta ,\gamma ,\delta \in {\mathbb C}^3\), the authors consider the expression: \[ I = I(\alpha ,\beta ,\gamma ,\delta) := (\alpha \beta \gamma)(\alpha \beta \delta)(\alpha \gamma \delta) (\beta \gamma \delta) \] where \((\alpha \beta \gamma)\) is the determinant of the \(3 \times 3\) matrix with rows \(\alpha ,\beta, \gamma\). According to the above observation, there exists, for \(d = 3n+1\), a unique \(\mathbb C\)-linear map \({\widetilde S}_d : V(d)^{\otimes 4} \rightarrow V(4)\) such that: \[ {\widetilde S}_d({\alpha}_x^d \otimes {\beta}_x^d \otimes {\gamma}_x^d \otimes {\delta}_x^d) = I^n{\alpha}_x{\beta}_x{\gamma}_x{\delta}_x\, . \] \(S_d\) is, by definition, \({\widetilde S}_d\) composed with the map \(V(d) \rightarrow V(d)^{\otimes 4}\), \(f \mapsto f^{\otimes 4}\). For the definition of \(T_d\), \(d = 3n+2\), one uses the symbolic expression \(I^n{\alpha}_x^2{\beta}_x^2{\gamma}_x^2{\delta}_x^2\). Now, the authors consider the linear subspaces \(L_S := x_1^{2n+3}{\mathbb C}[x_1,x_2,x_3]_{n-2} \subset V(d)\) for \(d = 3n+1\), and \(L_T := x_1^{2n+5}{\mathbb C}[x_1,x_2,x_3]_{n-3} \subset V(d)\) for \(d = 3n+2\), and notice that \({\mathcal I}_{{\mathbb P}(L_S)}^3\) (resp., \({\mathcal I}_{{\mathbb P}(L_T)}^3\)) contains the ideal sheaf of the base locus of the rational map \(S_d : {\mathbb P}(V(d)) \dashrightarrow {\mathbb P}(V(4))\) (resp., \(T_d : {\mathbb P}(V(d)) \dashrightarrow {\mathbb P}(V(8))\)). They deduce that for \(g \in V(d)\) not in \(L_S\) (resp., \(L_T\)) the restricted map \(S_d\, | \, {\mathbb P}(L_S+{\mathbb C}g) \dashrightarrow {\mathbb P}(V(4))\) (resp., \(T_d \, | \, {\mathbb P}(L_T+{\mathbb C}g) \dashrightarrow {\mathbb P}(V(8))\)) is linear. The key point of the proof consists in showing that, under the above assumptions on \(d\), there exists \(g\) such that the above linear map is \textit{dominant}. The authors reduce the proof of this fact, after a number of elementary but ingenious tricks, to some straightforward computations with a computer algebra program. The proof of the rationality of \(\text{Hyp}(d,2)\) can be now completed as in the paper of Shepherd-Barron.