Lattice polarized \(K3\) surfaces and Siegel modular forms (Q436170)

The paper under review has two goals: One is to present a classification of algebraic \(K3\) surfaces polarized by the lattice \(H\oplus E_8\oplus E_7\), and the other is to give explicit formulas for a Hodge correspondence that relates these \(K3\) surfaces to principally polarized abelian surfaces. For the classification, the key ingredients are normal forms for these \(K3\) surfaces, a coarse moduli space and an explicit description of the inverse period map in terms of Siegel modular forms. The result is formulated as follows. A \(K3\) surface \(X\) is said to be \(N\)-polarized if there is a primitive lattice embedding \(N\hookrightarrow \mathrm{NS}(X)\) where \(\mathrm{NS}(X)\) denotes the Neron-Severi lattice of \(X\). In what follow, \(N\) is taken to be the lattice \(H\oplus E_8\oplus E_7\). Let \((\alpha, \beta, \gamma, \delta)\) be a quadruple of complex numbers. Denote by \(X(\alpha,\beta,\gamma,\delta)\) the minimal resolution of the surface in \({\mathbf{P}}^3(x,y,z,w)\) cut out by the degree-four homogeneous polynomial \[ y^2zw-4x^3z+3\alpha xzw^2+\beta zw^3+\gamma xz^2w-{{1}\over{2}}(\delta z^2w^2+w^4)=0. \] (a) If \(\gamma\neq 0\) or \(\delta\neq 0\), then \(X(\alpha,\beta,\gamma,\delta)\) is a \(K3\) surface endowed with a canonical \(N\)-polarization. (b) Given any \(N\)-polarized \(K3\) surface \(X\), there exists \((\alpha,\beta,\gamma,\delta)\in{\mathbb{C}}^4\), with \(\gamma\neq 0\) or \(\delta\neq 0\), such that surfaces \(X\) and \(X(\alpha,\beta,\gamma,\delta)\) are isomorphic as \(N\)-polarized \(K3\) surfaces. (c) A coarse moduli space for \(N\)-polarized \(K3\) surfaces is given by the open variety \[ {\mathcal{M}}_{K3}^N=\{[\alpha,\beta\,\gamma,\delta]\in {\mathbf{WP}}^3(2,3,5,6)\,|\, \gamma\neq 0,\,\,\text{{or}}\,\, \delta\neq 0\}. \] (d) The period map \(\text{{per}}:{\mathcal{M}}_{K3}^N\to {\mathcal{F}}_2={\mathbb{H}}_2/Sp_4({\mathbb{Z}})\) is an isomorphism (as quasi-projective varieties). (e) The inverse period map is given by \(\text{{per}}^{-1}=[{\mathcal{E}}_4, {\mathcal{E}}_6, 2^{12}3^5{\mathcal{C}}_{10}, 2^{12}3^6{\mathcal{C}}_{12}]\) where \({\mathcal{E}}_4,{\mathcal{E}}_6\) are genus-two Eisenstein series of weight \(4\) and \(6\), and \({\mathcal{C}}_{10},{\mathcal{C}}_{12}\) are Igusa's cusp forms of weight \(10\) and \(12\), respectively. For the Hodge correspondence, the principally polarized abelian surface \(A\) associated to \(X(\alpha,\beta,\gamma,\delta)\) is given explicitly, depending on \(\gamma=0\) and \(\gamma\neq 0\). (f) Suppose that \(\gamma=0\). Then the principally polarized abelian surface \(A\) associated to \(X(\alpha,\beta, 0,\delta)\) is given by \((E_1\times E_2, {\mathcal{O}}_{E_1\times E_2}(E_1+E_2))\) where \(E_1\) and \(E_2\) are complex elliptic curves with \(j\)-invariants satisfying \(j(E_1)+j(E_2)={{\alpha^3-\beta^2}\over{\delta}}+1\) and \(j(E_1)j(E_2)={{\alpha^3}\over{\delta}}\). (g) Suppose that \(\gamma\neq 0\). Then the principally polarized abelian surface \(A\) associated to \(X(\alpha,\beta,\gamma,\delta)\) is given by \((\mathrm{Jac}(C), {\mathcal{O}}_{\mathrm{Jac}(C)}(\Theta))\) where \(C\) is a smooth genus two curve of Igusa-Clebsch invariants \([A,B,C,D]\in{\mathbf{WP}}^3(2,3,5,6)\).