Inverse period mappings of \(K3\) surfaces and a construction of modular forms for a lattice with the Kneser conditions (Q2204831)

The lattice \(A=2U\oplus A_2(-1)\) is the orthogonal complement of \(M=U\oplus E_8(-1)\oplus E_6(-1)\) in the K3 lattice and is thus associated with a family of lattice-polarised K3 surfaces. The purpose of this paper is to use the geometry of that family to study the ring of modular forms for the group \(\Gamma=\tilde{\mathop{\mathrm {O}}\nolimits}^+(A)\), which is the subgroup of the stable orthogonal group that preserves the period domain \(\mathcal D\). \par The outcome is that the ring is generated by modular forms \(t_i\) of trivial character of weights \(i=4\), \(6\), \(10\), \(12\) and \(18\), and a form of weight \(54\) and character det. The latter can be seen as a product \(s_9s_{45}\) of functions that satisfy \(s_9^2=t_{18}\) and \(s_{45}^2=d_{90}\) where \(d_{90}\) is a certain weight 90 homogeneous polynomial in the \(t_i\), but \(s_9\) and \(s_{45}\) come not from the family of K3 surfaces but from a double cover of it (defined away from a subset of codimension~2 in \(\mathcal D\)), and are not themselves modular forms for \(\Gamma\). \par The lattice \(A\) satisfies the Kneser conditions and \(\Gamma\) is therefore generated by \((-2)\)-reflections, so there is a natural subgroup \(\Gamma'\) consisting of the products of even numbers of reflections, and this is what gives rise to the double cover. \par The K3 surfaces arise as hypersurfaces in weighted projective space and the coefficients of the equations are essentially the modular forms \(t_i\): in other words, in this case it is possible to write down an inverse to the period map more or less explicitly. Analogous results for the case of Siegel modular forms (signature \((2,3)\) rather than \((2,4)\) as here) are known, and other subgroups of \({\mathop{\mathrm {O}}\nolimits}(2,4)\) have been studied in a similar way: see especially [\textit{A. Clingher} and \textit{C. F. Doran}, Adv. Math. 231, No. 1, 172--212 (2012; Zbl 1245.14036)]. The author explains carefully the similarities and differences between those cases and the one studied here. \par There is also an intriguing connection with complex reflection groups (specifically, with number~33 on the Shepherd-Todd list): subsequent work of the author (see [``Sequence of families of lattice polarized K3 surfaces, modular forms and degrees of complex reflection groups'', Preprint, \url{arXiv:2108.08168}]) extends this in a non-trivial way to some higher rank groups and to more of the complex reflection groups.