Projective orbifolds of Nikulin type (Q6071807)

This paper is concerned with projective irreducible symplectic orbifolds of dimension four that are deformations of partial resolutions of quotients of hyperkähler manifolds of \(K3^{[2]}\)-type by symplectic involutions which they denote as orbifolds of Nikulin type. Recall that manifolds of \(K3^{[n]}\)-type are deformations of Hilbert schemes \(W^{[n]}\) of \(n\) points on a \(K3\) surface \(W\). The quotient \(X/σ\) of a fourfold \(X\) of \(K3^{[2]}\)-type by a symplectic involution \(σ\) is singular along a \(K3\) surface and \(28\) points. Since this quotient does not admit a crepant resolution, we consider the partial resolution by blowing up the \(K3\) surface, which gives an irreducible symplectic orbifold with \(28\) terminal points. This is defined to be Nuklin orbifold. Orbifolds of Nikulin type are just deformations of such orbifolds. The paper firstly focuses on Nuklin orbifolds. They show that (in Theorem~1.2) there is a (generical) bijection between \begin{itemize} \item the pair \((T_X, NS(X))\) of the transcendental lattice and the Néron-Severi group of a projective fourfold \(X\) of \(K3^{[2]}\)-type admitting a symplectic involution \(\sigma\), \item the transcendental lattice of the corresponding Nuklin orbifold \(Y\) obtained from \((X,\sigma)\). \end{itemize} The proof is based on a connection between \(T_S\) and \(T_Y\) where \(S\) is the \(K3\) surface in the fixed locus of \(\sigma\). The authors conjecture that they are isomorphic even with integer coefficients. They classify (in Theorem~1.1) the pair \((T_X, NS(X))\) of the transcendental lattice and the Néron-Severi group of a general projective fourfold \(X\) of \(K3^{[2]}\)-type admitting a symplectic involution \(\sigma\). There are only two possibilities of \(NS(X)\) \begin{itemize} \item \(\Lambda_{2d}\), \item \(\tilde{\Lambda}_{2d}\). \end{itemize} They prove a result in the converse direction showing that once a projective orbifold \(X\) of \(K3^{[2]}\)-type is has an \(NS(X)\) lying in the above two possibilities, it has a simpletic resolution. Actually, they also give a similar result (Corollary~2.3) for very genral (possibly not projective) fourfolds \(X\) of \(K3^{[2]}\)-type admitting a symplectic involution \(\sigma\). The \(NS\) groups will be \(E_8(-2)\). Conversely, if a fourfold \(X\) has an \(NS\) group isomorphic to \(E_8(-2)\), then \(X\) is non-projective and it admits a symplectic involution. This classification is based on fine analysis of lattices and previous results using \(E_8(-2)\) (resp.\(\Lambda_{2d}\)) to characterize fourfolds of \(K3^{[2]}\) type with symplectic involutions (resp. projective fourfolds). In Section~4, the authors also give an orbifold Rimann-Roch formula using the BBF quadratic form. The proof uses the smooth resolution of \(Y\) (thus \(X/\sigma\)) as a bridge. In Section~5, the authors focus on a family of projective orbifolds of Nikulin type polarized by a class of BBF degree \(2\). By the Rimann-Roch formula obtained, their projective models are fourfolds in \(\mathbb{P}^6\). In the end, they show that the general element \(Y\) in a family of orbifolds of Nikulin type with a polarization of BBF degree \(2\) and divisibility \(1\) is a double cover of a special complete intersection \((3, 4)\) in \(\mathbb{P}^6\) branched along a surface of degree \(48\). And this is the first known description of a locally complete family of polarized singular irreducible symplectic varieties. The characterization by lattices of (general and projective) \(K3^{[2]}\)-type fourfolds with symplectic resolutions and the Riemann-Roch formula are very elegant. The reviewer is curious whether the description of the locally complete family given could be generalized to an abstract and systematical way to give desriptions of more families without very explicit calculations.