Integer cohomology of compact Hyperkähler manifolds

Transcription

1 Integer cohomology of compact Hyperkähler manifolds Simon Kapfer University of Augsburg Géométrie Algébrique en Liberté Trieste, 27 June 204 Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds

2 Motivation Let X be a compact Hyperkähler manifold of complex dimension 2m or, equivalently, an IHS manifold. Why should we be interested in H (X, Z)? It feels more geometric. Comparing H (X, Z) with H (X, C) gives us information about X, e.g. on projectivity. We obtain restrictions to possible automorphisms of our manifold X.... Question Which constructions in H (X, R/C) carry over to H (X, Z)? Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds / 0

3 Beauville Bogomolov form As an example, consider the quadratic Beauville Bogomolov form q X : H (X, R) R Theorem (Fujiki) q X (α) m = c X α2m for some c R. Corollary q X can be renormalized to yield a primitive integral quadratic form: H (X, Z) Z. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 2 / 0

4 Hodge numbers for K3 surfaces For S a compact Hyperkähler manifold of complex dimension two, i.e. a K3 surface, we have h p,q (S) h k (S, Z) and the intersection pairing on H 2 is isomorphic to U 3 E 8 ( ) 2, where U and E 8 are the bilinear forms corresponding to the hyperbolic resp. E 8 lattice. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 3 / 0

5 Hodge numbers for Hilbert schemes Denote S [n] the Hilbert scheme of n points on the K3 surface S. Then, the Hodge decomposition for S [2] is given by: h p,q (S [2] ) h k (S [2], Z) and there are formulae for all S [n] due to Göttsche. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 4 / 0

6 Betti numbers k S [] S [2] S [3] S [4] S [5] S [6] For k fixed, these numbers stabilize for n big enough. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 5 / 0

7 Cohomology of Hilbert schemes Theorem (Nakajima) For each m and each α H j (S, Q), there is an operator a m (α) : H i (S [n], Q) H i+j+2m 2 (S [n+m], Q) and these operators, applied to H (S [0], Q) = Q, span the entire cohomology of all Hilbert schemes S [n]. Construction of the operators: Use the incidence scheme I S [n] S S [n+m] and define, using Poincaré duality, a m (α)β P.D. = p 3 ((p β p 2α) [I]). Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 6 / 0

8 Integral basis for H (S [n], Z) This also works in integer cohomology! Theorem (Qin, Wang) Let H 0 (S, Z) be the canonical generator. The operators z λ a λ () : H i (S [n], Z) H i+2m 2k (S [n+m], Z) a m (α) : H i (S [n], Z) H i+j+2m 2k (S [n+m], Z) span the integer cohomologies H (S [n], Z). Here, the composition of several operators a mi () is denoted via a partition λ and z λ denotes some constant depending on λ. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 7 / 0

9 Algebra generated by H 2 (S [n], C) Theorem (Verbitsky) The subalgebra generated by H 2 (S [n], C) in H (S [n], C) is equal to Sym H 2 (S [n], C) α n+ q(α) = 0, where q is the Beauville-Bogomolov Form. In fact, this holds for any compact Hyperkähler manifold of complex dimension 2n. What about cohomology with integral coefficients? Lehn, Sorger and Vasserot developed formulae for the cup product on H (S [n], Q). We can use them also for computations in H (S [n], Z). Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 8 / 0

10 Ring structure of H (S [n], Q) Rough idea of the algebraic model: To any composition of operators a m (α )... a mk (α k ) one associates a conjugacy class of the symmetric group S n, given by the partition λ = (m,..., m k ) The cup product is built with the product in S n : E.g. ( 2 3)(4 5) ( 4) = ( ), but ( 2 3)(4 5) ( 2) = ()(2 3)(4 5). If two cycles ν i, ν j are joined together by a transposition, multiply the corresponding classes, e.g. ( 2 3) αi (4 5) αj ( 4) αl = ( ) αi α j α l If a cycle is split in two by a transposition, use a map adjoint to the multiplication in H (S). Sum up all such possibilities and put the constants in the right way. Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 9 / 0

11 Algebra generated by H 2 (S [n], Z) For integer cohomology, we get torsion for small n, e.g. H 4 (S [2], Z) = Sym 2 H 2 (S [2], Z) Z H 4 (S [3], Z) = Sym 2 H 2 (S [3], Z) Z 23 Z 3 H 4 (S [n], Z) = Sym 2 H 2 (S [n], Z) Z 24, n 4. Question Is there a geometric interpretation, e.g. for n = 3? Thank you for your attention! Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds 0 / 0

12 References S. Boissière, M. Nieper-Wisskirchen, and A. Sarti. Smith theory and irreducible holomorphic symplectic manifolds. arxiv e-prints, April 202. M. Gross, D. Huybrechts, and D. Joyce. Calabi-Yau Manifolds and Related Geometries: Lectures at a Summer School in Nordfjordeid, Norway, June, 200. Universitext (979). Springer Berlin Heidelberg, M. Lehn and C. Sorger. The cup product of the Hilbert scheme for K3 surfaces. arxiv Mathematics e-prints, December Zhenbo Qin and Weiqiang Wang. Integral operators and integral cohomology classes of Hilbert schemes. Mathematische Annalen, Simon Kapfer, University of Augsburg Integer cohomology of compact Hyperkähler manifolds