Finite Hilbert stability of (bi)canonical curves (Q1943675)

Let us recall briefly the notion of finite Hilbert stability for a scheme. Let \(X\subset \mathbb{P}V\) be a closed subscheme such that the restriction map \[ \rho:H^0(\mathbb{P}V,\mathcal{O}(m))\rightarrow H^0(X,\mathcal{O}(m)) \] is surjective and set \[ W_m=\bigwedge^{h^0(X,\mathcal{O}(m)} H^0(\mathbb{P}V,\mathcal{O}(m))^\vee. \] Thus, from the map \(\rho\) and taking the wedge product, one can consider the \(m\)-Hilbert point \[ [X]_m=\left[\bigwedge^{h^0(X,\mathcal{O}(m)} H^0(\mathbb{P}V,\mathcal{O}(m))\rightarrow \bigwedge^{h^0(X,\mathcal{O}(m)}H^0(X,\mathcal{O}(m))\rightarrow 0\right]\in \mathbb{P}(W_m). \] A polarized variety \((X,L)\) is \(m\)-Hilbert stable (semistable) if the \(m\)-Hilbert point \([X]_m\) associated to \(X\subset\mathbb{P}H^0(X,L)\) is G.I.T stable (semistable). A variety \((X,L)\) is Hilbert stable if it is \(m\)-Hilbert stable for all large \(m>0\).\newline We say that a curve \(C\) is canonically (bicanonically) embedded if there is an embedding \(C\subset \mathbb{P}H^0(C,\omega_C)\) (\(C\subset \mathbb{P}H^0(C,\omega_C^{\otimes 2})\)) associated to the complete linear system given by the canonical bundle \(\omega_C\).\newline The two main results of this paper are the following. A generic canonically or bicanonically embedded smooth curve has its \(m\)-th Hilbert point semistable for \(m\geq 2\). A generic bicanonically embedded smooth curve has its \(m\)-th Hilbert point stable for every \(m\geq 3\).\newline It is remarkable that the authors obtain such precise effective bounds for stability or semistability. \newline The case of embedded bielliptic curves (of genus \(g\)) is also studied in details. They have non-semistable \(m\)-th Hilbert points for \(m\leq (g-3)/2\), and semistable \(m\)-th Hilbert points for \(m\geq (g-1)/2\). Also, various refined results for finite stability of bicanonical curves are proved. For instance, a generic bicanonically embedded smooth curve of genus \(g\geq 4\) has a semistable 2nd Hilbert point. Eventually, the work of the authors leads to some important progress on Morrison's conjecture (a smooth bicanonical curve of genus \(g\geq 3\) has stable \(m\)-th Hilbert point whenever \((g,m)\neq (3,2)\), see \textit{I. Morrison} [Ji, Lizhen (ed.) et al., Geometry of Riemann surfaces and their moduli spaces. Somerville, MA: International Press. Surveys in Differential Geometry 14, 315--369 (2010; Zbl 1215.14026)]). \newline Let's say a few words about the proofs. In the canonical case, they are based on the study of a canonically embedded balanced ribbon and the canonically embedded balanced double \(A_{2k+1}\)-curve which are special singular curves with \(\mathrm{Spec}\, \mathbb{C}[t,t^{-1}]\)-action. In the bicanonical case, the proof is based on showing finite Hilbert stability of special hyperelliptic curves, called Wiman curves. \newline The paper is well written.