Effective divisors on M_g associated to curves with exceptional secant planes

From MaRDI portal
Publication:2428554



Abstract: This paper is a sequel to cite{C}, in which the author studies secant planes to linear series on a curve that is general in moduli. In that paper, the author proves that a general curve has no linear series with exceptional secant planes, in a very precise sense. Consequently, it makes sense to study effective divisors on ovmcMg associated to curves equipped with secant-exceptional linear series. Here we describe a strategy for computing the classes of those divisors. We pay special attention to the extremal case of (2d−1)-dimensional series with d-secant (d−2)-planes, which appears in the study of Hilbert schemes of points on surfaces. In that case, modulo a combinatorial conjecture, we obtain hypergeometric expressions for tautological coefficients that enable us to deduce the asymptotics in d of our divisors' virtual slopes.


An \(s\)-dimensional linear series of degree \(m\) on a smooth curve \(C\) is said to have a \(d\)-secant \(k\)-plane if there is a \(k\)-dimensional subspace in \({\mathbb P}^s\) which intersects the image of the \(g^s_m\) in \(d\) points, counted with multiplicity. The existence of \(d\)-secant \(k\)-planes to general genus \(g\) curves was studied in [\textit{E. Cotterill}, Math. Z. 267, No. 3--4, 549--582 (2011; Zbl 1213.14064)]. In particular, a Brill--Noether-type result holds in this case, ensuring that there are no \(d\)-secant \(k\)-planes on a general curve if the Brill--Noether number \(\rho\) is \(0\) and the expected dimension \(\mu\) of the space of \(d\)-secant \(k\)-planes is \(-1\). Therefore, in this case the condition of having a \(d\)-secant \(k\)-plane defines a divisor inside the moduli space \(\overline{\mathcal M}_g\) of stable genus \(g\) curves. The paper under review describes a strategy for computing the classes of these divisors, by computing the number of exceptional series along judiciously chosen test families. The strongest results are obtained in the extremal case of linear series of dimension \(2d-1\) with \(d\)-secant \((d-2)\)-planes, which appears in the study of Hilbert schemes of points on surfaces [\textit{M. Lehn}, Invent. Math. 136, No. 1, 157--207 (1999; Zbl 0919.14001)]. In that case, there is a (comparatively) simple generating function for the expected number of \(d\)-secant \((d-2)\)-planes. This formula, together with a Gysin-type calculation of \textit{D. Khosla} [``Tautological classes on moduli spaces of curves with linear series and a push-forward formula when \(\rho=0\), \url{arXiv:0704.1340}], allows to determine the class of any secant plane divisor on \(\overline{\mathcal M}_g\) modulo the boundary classes \(\delta_i\) (\(i\geq1\)) corresponding to reducible curves. In particular, this yields asymptotic results on the virtual slope of secant plane divisors for \(d\rightarrow\infty\).











This page was built for publication: Effective divisors on \({\overline{\mathcal{M}}}_g\) associated to curves with exceptional secant planes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2428554)