On the loci of morphisms from P^1 to G(r,n) with fixed splitting type of the restricted universal sub-bundle or quotient bundle

From MaRDI portal
Publication:2042955



Abstract: Let ngeq4, 2leqrleqn−2 and egeq1. We show that the intersection of the locus of degree e morphisms from mathbbP1 to G(r,n) with the restricted universal sub-bundles having a given splitting type and the locus of degree e morphisms with the restricted universal quotient-bundle having a given splitting type is non-empty and generically transverse. As a consequence, we get that the locus of degree e morphisms from mathbbP1 to G(r,n) with the restricted tangent bundle having a given splitting type need not always be irreducible.


A well known theorem of Grothendieck says that every vector bundle \(\mathcal E\) on \(\mathbb P^1\) is a direct sum of line bundles, so one can consider it's splitting type, the tuple \((a_1, \dots, a_r)\) of integers in non-decreasing order for which \(\mathcal E \cong \bigoplus_{i=1}^r {\mathcal O}_{\mathbb P^1} (a_i)\), where \(r\) is the rank of \(\mathcal E\). There has been plenty of interest in the behavior of the splitting type in a family of rational curves, especially the tangent and normal bundles of embedded smooth rational curves. For example, \textit{J. L. Verdier} [Lect. Notes Phys. 180, 136--141 (1983; Zbl 0528.58008)] and \textit{L. Ramella} [C. R. Acad. Sci., Paris, Sér. I 311, No. 3, 181--184 (1990; Zbl 0721.14014)] showed that the locus of nondegenerate rational curves in \(\mathbb P^n\) with given splitting type of the restricted tangent bundle is irreducible of the expected dimension. By contrast, \textit{I. Coskun} and \textit{E. Riedl} showed that the locus of rational normal curves of fixed degree in \(\mathbb P^n\) can be reducible for \(n \geq 5\) [Math. Z. 288, No. 3--4, 803--827 (2018; Zbl 1391.14067); Commun. Contemp. Math. 21, No. 2, Article ID 1850011, 29 p. (2019; Zbl 1436.14057)]. The author studies the space of morphisms \(f: \mathbb P^1 \to G(r,n)\), where \(G(r,n)\) is the Grassmann variety of \(r\)-dimensional subspaces of an \(n\)-dimensional vector space, which comes equipped with a universal exact sequence \(0 \to \mathcal S \to {\mathcal O}_{G(r,n)}^{\oplus n} \to \mathcal Q \to 0\). He shows that the locus \(M(b_\bullet)\) for which \(f^*\mathcal Q\) has splitting type \(b_\bullet\) is smooth of expected dimension and the same for the locus \(M^\prime (a_\bullet)\) for which \(f^*\mathcal S\) has splitting type \(a_\bullet\). For \(n \geq 4\) and \(2 \leq r \leq n-2\) he shows that the intersection of \(M (b_\bullet)\) and \(M^\prime (a_\bullet)\) is non-empty and generically transverse along at least one component of the intersection. Using the fact that the tangent bundle \(T_{G(r,n)}\) is isomorphic to \(\mathcal S^* \otimes \mathcal Q\), he deduces that the locus of morphisms \(f\) for which \(f^* T_{G(r,n)}\) has fixed splitting type has at least one irreducible component of expected dimension arising from the splitting types of \(f^*\mathcal S\) and \(f^*\mathcal Q\). Examples are given to show that this locus can be reducible.











This page was built for publication: On the loci of morphisms from \(\mathbb{P}^1\) to \(G(r,n)\) with fixed splitting type of the restricted universal sub-bundle or quotient bundle

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2042955)