Searcharxiv⌕ Search

arXiv subjects

Kieran G. O'Grady

Publications and source records attributed to Kieran G. O'Grady.

At least 19 recordsLinked to original sources

HK manifolds of Type $K3^{[a^2+1]}$ as moduli spaces of projective bundles on HK manifolds of Type $K3^{[2]}$

We prove results on moduli spaces of slope stable bundles of projective spaces on a hyperkähler manifold of Type $K3^{[2]}$. Let $X$ be projective of Type $K3^{[2]}$ and $h$ be a (generic) ample class. We prove that the moduli space $M_{\overline{\bf w}_a}(X,h)$ parametrizing $h$ slope stable bundles with a suitable mock Mukai vector ${\overline{\bf w}}_a$ contains an irreducible component $M_{\overline{\bf w}_a}(X,h)^{*}$ whose normalization $\widetilde{M}_{\overline{\bf w}_a}(X,h)^{*}$ is a (projective) HK manifold of Type $K3^{[a^2+1]}$, and that conversely every projective HK manifold $W$ of Type $K3^{[a^2+1]}$ is isomorphic to $\widetilde{M}_{\overline{\bf w}_a}(X,h)^{*}$ for a suitable $(X,h)$ as above. Moreover the universal bundle of projective spaces on $X\times \widetilde{M}_{\overline{\bf w}_a}(X,h)^{*}$ defines a vector bundle whose $2nd$ Chern class defines a rational Hodge isometry $H^2(X)\to H^2(\widetilde{M}_{\overline{\bf w}_a}(X,h)^{*})$. From this and a result of Markman one gets that the analogue of the Shafarevich conjecture (a special case of the Hodge conjecture) holds for rational Hodge isometries $H^2(W_1) \to H^2(W_2)$ between projective hyperkähler manifolds $W_1,W_2$ of Types $K3^{[a_1^2+1]}$ and $K3^{[a_2^2+1]}$ respectively. We prove results also for $(X,ω)$ a general HK manifold of Type $K3^{[2]}$. In fact one ingredient in our proof is Verbistsky's theory of projectively hyperhomolorphic vector bundles.

math.AG↗

The geometry of antisymplectic involutions, II

We continue our study of fixed loci of antisymplectic involutions on projective hyper-Kähler manifolds of $\mathrm{K3}^{[n]}$-type induced by an ample class of square 2 in the Beauville-Bogomolov-Fujiki lattice. We prove that if the divisibility of the ample class is 2, then one connected component of the fixed locus is a Fano manifold of index 3, thus generalizing to higher dimensions the case of the LLSvS 8-fold associated to a cubic fourfold. We also show that, in the case of the LLSvS 8-fold associated to a cubic fourfold, the second component of the fixed locus is of general type, thus answering a question by Manfred Lehn.

math.AG↗

Moduli of sheaves on hyperkähler manifolds

We survey recent advances in the theory of moduli spaces of stable sheaves on hyperkähler manifolds of dimension greater than $2$. We start by recalling the well-known theory in dimension $2$, i.e.~for $K3$ surfaces, emphasizing the techniques which can be extended to higher dimensions.

math.AG↗

Modular sheaves with many moduli

We exhibit moduli spaces of slope stable vector bundles on general polarized HK varieties $(X,h)$ of type $K3^{[2]}$ which have an irreducible component of dimension $2a^2+2$, with $a$ an arbitrary integer greater than $1$. This is done by studying the case $X=S^{[2]}$ where $S$ is an elliptic $K3$ surface. We show that in this case there is an irreducible component of the moduli space of stable vector bundles on $S^{[2]}$ which is birational to a moduli space of sheaves on $S$. We expect that if the moduli space of sheaves on $S$ is a smooth HK variety (necessarily of type $K3^{[a^2+1]}$) then the following more precise version holds: the closure of the moduli space of slope stable vector bundles on $(X,h)$ in the moduli space of Gieseker-Maruyama semistable sheaves with its GIT polarization is a general polarized HK variety of type $K3^{[a^2+1]}$.

math.AG↗

Rank 4 stable vector bundles on hyperkähler fourfolds of Kummer type

We partially extend to hyperkähler fourfolds of Kummer type the results that we have proved regarding stable rigid vector bundles on hyperkähler (HK) varieties of type $K3^{[n]}$. Let $(M,h)$ be a general polarized HK fourfold of Kummer type such that $q_M(h)\equiv -6\pmod{16}$ and the divisibility of $h$ is $2$, or $q_M(h)\equiv -6\pmod{144}$ and the divisibility of $h$ is $6$. We show that there exists a unique (up to isomorphism) slope stable vector bundle $\cal F$ on $M$ such that $r({\cal F})=4$, $ c_1({\cal F})=h$, $Δ({\cal F})=c_2(M)$. Moreover $\cal F$ is rigid. One of our motivations is the desire to describe explicitly a locally complete family of polarized HK fourfolds of Kummer type.

math.AG↗

The geometry of antisymplectic involutions, I

We study fixed loci of antisymplectic involutions on projective hyperkähler manifolds of $\mathrm{K3}^{[n]}$-type. When the involution is induced by an ample class of square 2 in the Beauville-Bogomolov-Fujiki lattice, we show that the number of connected components of the fixed locus is equal to the divisibility of the class, which is either 1 or 2.

math.AG↗

Theta groups and projective models of hyperkähler varieties

We define the theta group associated to a simple coherent sheaf $\cal F$ on a hyperkähler manifold $X$ of Kummer type or OG6 type, provided $g^{*}({\cal F})$ is isomorphic to $\cal F$ for every automorphism $g$ of $X$ acting trivially on $H^2(X)$. Note that this condition is satisfied if $\cal F$ is invertible, if $\cal F$ is one of the rank $4$ stable vector bundles on general polarized HK fourfolds with certain discrete invariants constructed in arXiv:2203.03987, or if $\cal F$ is the tangent bundle. We compute the commutator pairings of theta groups of line bundles and the rank $4$ modular vector bundles of arXiv:2203.03987 (the commutator pairing of the tangent bundle is trivial). We have been motivated by the quest for an explicit description of locally complete families of polarized varieties of Kummer (or OG6) type.

math.AG↗

Moduli of sheaves on $K3$'s and higher dimensional HK varieties

We review a proof of the well know result stating that moduli spaces of stable sheaves with fixed Chern character on a polarized $K3$ surface are deformations of a hyperkähler variety of Type $K3^{[n]}$ (if a suitable numerical hypothesis is satisfied). In a recent work we have adapted that proof in order to prove results on moduli of vector bundles on polarized hyperkähler varieties of Type $K3^{[2]}$ - this is the content of the second part of the paper.

math.AG↗

Modular sheaves on hyperkähler varieties

A torsion free sheaf on a hyperkähler variety $X$ is modular if the discriminant satisfies a certain condition, for example if it is a multiple of $c_2(X)$ the sheaf is modular. The definition is taylor made for torsion-free sheaves on a polarized hyperkähler variety (X,h) which deform to all small deformations of (X,h). For hyperkählers deformation equivalent to $K3^{[2]}$ we prove an existence and uniqueness result for slope-stable modular vector bundles with certain ranks, $c_1$ and $c_2$. As a consequence we get uniqueness up to isomorphism of the tautological quotient rank $4$ vector bundles on the variety of lines on a generic cubic $4$-dimensional hypersurface, and on the Debarre-Voisin variety associated to a generic skew-symmetric $3$-form on a $10$-dimensional complex vector space. The last result implies that the period map from the moduli space of Debarre-Voisin varieties to the relevant period space is birational.

math.AG↗

Compact tori associated to hyperkaehler manifolds of Kummer type

Let X be a hyperkaehler manifold of Kummer type. We study the 4 dimensional intermediate Jacobian J^3(X) constructed out of the 3rd cohomology of X, motivated by the desire to understand the Kuga-Satake torus associated to X. We prove that there is an embedding of Hodge structures of H^2(X) into H^2(J^3(X)). It follows that if X is projective with polarization L, the projective Kuga-Satake torus of (X,L) is isogenous to the fourth power of J^3(X). By studying the cohomology ring of X, we find that there is a natural smooth quadric Q(X) in the projectivization of H^3(X), with a natural choice of one of the two irreducible components of the variety parametrizing maximal linear subspaces of Q(X). Let Q^{+}(X) be the chosen irreducible component; then it is a smooth quadric in the projectivization of S^{+}(X), one of the two spinor representations of Spin H^3(X), where H^3(X) is equipped with the quadratic form whose zero locus is Q(X). Gauss-Manin parallel transport identifies the set of projectivizations of H^{2,1}(Y), for Y a deformation of X, with an open subset of a linear section of Q^{+}(X). This amounts to an effective description of the intermediate Jacobian J^3(X) in terms of the H.S. on H^2(X). A simple consequence is a new proof of a result of Mongardi giving (necessary) conditions for the action of monodromy on H^2(X). Lastly, we show that if X is projective, then J^3(X) is an abelian fourfold of Weil type.

math.AG↗

Birational geometry of the moduli space of quartic K3 surfaces

By work of Looijenga and others, one has a good understanding of the relationship between GIT and Baily-Borel compactifications for the moduli spaces of degree 2 K3 surfaces, cubic fourfolds, and a few other related examples. The similar-looking cases of degree 4 K3 surfaces and double EPW sextics turn out to be much more complicated for arithmetic reasons. In this paper, we refine work of Looijenga to allow us to handle these cases. Specifically, in analogy with the so-called Hassett-Keel program for the moduli space of curves, we study the variation of log canonical models for locally symmetric varieties of Type IV associated to D lattices. In particular, for the dimension 19 case, we conjecturally obtain a continuous one-parameter interpolation (via a directed MMP) between the GIT and Baily-Borel compactifications for the moduli of degree 4 K3 surfaces. The predictions for the similar 18 dimensional case, corresponding geometrically to hyperelliptic degree 4 K3 surfaces, can be verified by means of VGIT -- this will be discussed elsewhere.

math.AG↗

Decomposable cycles and Noether-Lefschetz loci

We prove that there exist smooth surfaces of degree d in projective 3-space such that the group of rational equivalence classes of decomposable 0-cycles has rank at least the integer part of (d-1)/3.

math.AG↗

Computations with modified diagonals

Beauville and Voisin proved that the third modified diagonal of a complex K3 surface X represents a torsion class in the Chow group of X^3. Motivated by this result and by conjectures of Beauville and Voisin on the Chow ring of hyperkaehler varieties we prove some results on modified diagonals of projective varieties and we formulate a conjecture.

math.AG↗

Periods of Double EPW-sextics

We study the indeterminacy locus of the period map for double EPW-sextics. We recall that double EPW-sextics are parametrized by lagrangian subspaces of the third wedge-product of a 6-dimensional complex vector-space. The indeterminacy locus is contained in the set of lagrangians containing a decomposable vector. The projectivization of the 3-dimensional support of such a decomposable vector contains a degeneracy subscheme which is either all of the plane or a sextic curve. We show that the period map is regular on any lagrangian A such that for all decomposables in A the corresponding degeneracy subscheme is a GIT-semistable sextic curve whose closure (in the semistable locus) does not contain a triple conic.

math.AG↗

Moduli of double EPW-sextics

We study the GIT quotient of the symplectic grassmannian parametrizing lagrangian subspaces of \bigwedge^3{\mathbb C}^6 by the natural action of SL_6, call it M. This is a compactification of the moduli space of smooth double EPW-sextics. We determine the stable points, the irreducible components of the GIT boundary and their dimensions. There are strong analogies with the moduli space of cubic 4-folds: we prove a result which is analogous to a theorem of Laza stating that cubic 4-folds with simple singularities are stable. Our final goal (not achieved in the present paper) is to understand the period map from M to the Baily-Borel compactification of the relevant period domain modulo an arithmetic group. With this goal in mind we analyze the locus in the GIT-boundary of M where the period map is not regular: we prove that it has two irreducible components, one of dimension 3, the other of dimension 1. Our results suggests that M is isomorphic to Looijenga's compactification associated to 3 specific hyperplanes in the period domain.

math.AG↗

Double covers of EPW-sextics

EPW-sextics are special 4-dimensional sextic hypersurfaces (with 20 moduli) which come equipped with a double cover. We analyze the double cover of EPW-sextics parametrized by a certain prime divisor in the moduli space. We associate to the generic sextic parametrized by that divisor a K3 surface of genus 6 and we show that the double epw sextic is a contraction of the Hilbert square of the K3. This result has two consequences. First it gives a new proof of the following result of ours: smooth double EPW-sextics form a locally complete family of hyperkaehler projective deformations of the Hilbert square of a K3. Secondly it shows that away from another prime divisor in the moduli space the period map for double EPW-sextics is as well-behaved as it possibly could be.

math.AG↗

Moduli of sheaves and the Chow group of K3 surfaces

Let X be a projective complex K3 surface. Beauville and Voisin singled out a 0-cycle c_X on X of degree 1: it is represented by any point lying on a rational curve in X. Huybrechts proved that the second Chern class of a rigid simple vector-bundle on X is a multiple of the Beauville-Voisin class c_X if certain hypotheses hold and he conjectured that the additional hypotheses are unnecessary. We believe that the following generalization of Huybrechts' conjecture holds. Let M and N be moduli spaces of stable pure sheaves on X (with fixed cohomological Chern characters) and suppose that they have the same dimension: then the set whose elements are second Chern classes of sheaves parametrized by the closure of M (in the corresponding moduli spaces of semistable sheaves) is equal to the set whose elements are second Chern classes of sheaves parametrized by the closure of N after a translation by a suitable multiple of c_X (so that degrees match). We will prove that the above statement holds under some additional assumptions.

math.AG↗

Pairwise incident planes and hyperkaehler four-folds

We address the following question: what are the cardinalities of maximal finite families of pairwise incident planes in a complex projective space? One proves easily that the span of the planes has dimension 5 or 6. Up to projectivities there is one such family spanning a 6-dimensional projective space - this is an elementary result. Maximal finite families of pairwise incident planes in a 5-dimensional projective space are considerably more misterious: they are linked to certain special (EPW) sextic hypersurfaces which have a non-trivial double cover, generically a hyperkaehler 4-fold. We prove that the cardinality of such a set cannot exceed 20. We also show that there exist such families of cardinality 16 - in fact we conjecture that 16 is the maximum.

math.AG↗