Geometry of the Winger pencil (Q2049058)

There is essentially one irreducible action of the icosahedral group \(A_5\) on \(\mathbb C^3\), which induces an \(A_5\)-action on \(\mathbb P^2\). \textit{F. Klein} determined the invariants of this action [Lectures on the Icosahedron and the solution of equations of the fifth degree, 2nd ed. New York: Dover (1956; Zbl 0072.25901)]. The sextic invariants form a pencil of genus 10 curves invariant under the \(A_5\)-action, which can be written \(Q^3 + \lambda F, \lambda \in \mathbb P^1\) with \(Q\) the equation of a conic \(K\) and \(F\) the equation of a union \(C_\infty\) of \(6\) lines, no three collinear. \textit{R. M. Winger} showed that the pencil has exactly \(4\) singular members (\(3K, C_\infty\) and two nodal curves) and found equations for them [Math. Ann. 93, 210--216 (1925; JFM 51.0120.01)], so we call it \textit{the Winger pencil}. After reviewing the construction of the Winger pencil both from the irregular orbits of \(A_5\) acting on \(\mathbb P^2\) and invariant theory, the author uses cohomological arguments to show that the pencil is locally universal at each of its smooth members. The Winger pencil is not a moduli space because the triple conic is unstable and the outer automorphism group of \(A_5\) has order \(2\), so precomposing the \(A_5\)-action with such an automorphism gives a second copy of the family. Therefore the author replaces the triple conic with a smooth algebraic curve \(C_K\) of genus \(10\) (it lies on the cone over \(K\) and projection from the vertex makes it a triple cover of \(K\)) and takes the disjoint union of two copies of the family to construct a complete family of \(A_5\)-curves of genus 10 with stable degeneration. The new family is not universal because \(C_K\) admits an \(A_5\)-automorphism cyclic of order \(3\). The author remedies this by showing that a smooth projective genus \(10\) curve with faithful \(A_5\)-action has corresponding orbifold of genus zero with four orbifold points, one of type \(\mu_5\) and three of type \(\mu_2\), and then setting up a moduli functor for such objects using the approach of \textit{B. Farb} and \textit{E. Looijenga} [Geom. Dedicata 208, 197--220 (2020; Zbl 1455.14052)] to obtain a fine moduli space with base \(\mathcal M_{10,4}^{A_5}\) a non-singular quasi-projective variety with two connected components and an \(S_3\)-action. The action of \(S_3\) has one irregular orbit and the \(S_3\)-orbit space \(\mathcal M_{10}^{A_5}\), which is also the coarse moduli space of smooth projective genus 10 curves with faithful \(A_5\)-action. Furthermore, \(\mathcal M_{10}^{A_5}\) admits a Deligne-Mumford compactification \(\overline{\mathcal M}_{10}^{A_5}\) parametrizing stable genus 10 curves with faithful \(A_5\)-action which are smoothable as \(A_5\)-curves. The \(A_5\)-orbifold formation extends to a morphism \(\overline{\mathcal M}_{10}^{A_5} \to \overline{\mathcal M}_0 (5;2,2,2)\), where \(\overline{\mathcal M}_0 (5;2,2,2)\) is defined by allowing two fo the three orbifold points of order \(2\) to coalesce, giving the full Winger pencil.