On isomorphisms between Siegel modular threefolds (Q266533)

This paper considers two isomorphic finite families of Siegel modular threefolds, and shows that each member of these families carries an endomorphism of degree 8. The complex points of Siegel modular threefolds can be obtained by compactifying the quotient of the Siegel upper half-plane of genus 2 by congruence subgroups of the symplectic group \(\mathrm{Sp}_{4}(\mathbb{Z})\). The compactification is the one defined by Satake using the ring of Siegel modular forms. The paper shows that two (finite) quotients between such congruence subgroups are isomorphic, and the Siegel modular threefolds associated with the intermediate groups carry the endomorphism in question. For presenting more details, we use the notation \(\binom{A\;\;B}{C\;\;D}\) for elements of \(\mathrm{Sp}_{4}(\mathbb{Z})\), where \(A\), \(B\), \(C\), and \(D\) are \(2\times2\) integral matrices satisfying the symplectic conditions. The group \(\Gamma_{0}(2)\) consists of those matrices for which \(C\) has even entries, and in \(\Gamma_{0}^{0}(2)\) the entries of \(B\) must be even as well. In addition, \(\Gamma_{2}^{2}(2,4)\) is defined as the group of those elements in which the entries of \(A-I\), \(D-I\), and \(C\) are even, the diagonal entries of \(B\) are even, and the diagonal entries of \(C\) are divisible by 4. Finally, elements of \(\Gamma_{2}(2,4)\) have even entries in \(A-I\), \(D-I\), \(B\), and \(C\), and the diagonal entries of both \(B\) and \(C\) are divisible by 4. Now, \(\Gamma_{2}^{2}(2,4)\) and \(\Gamma_{2}(2,4)\) are normal subgroups of \(\Gamma_{0}(2)\) and \(\Gamma_{0}^{0}(2)\) respectively, and both quotients are isomorphic to the semi-direct product in which \(S_{3}\) operates on a 4-dimensional vector space over the finite field \(\mathbb{F}_{2}\) (a group of order 96). In order to construct an isomorphism that is compatible with the actions on the Siegel modular threefolds in question, the author investigates the associated rings of Siegel modular forms (as the Proj functor produces from these rings the desired Satake compactifications), including half-integral weights. The paper presents briefly, for this purpose, the theory of theta functions with (integral) characteristics, and the closely related second order theta functions. Reference number 10 (of Runge) has proved that for \(\Gamma_{2}(2,4)\) this ring (with the appropriate characters) is generated by the four second order theta functions, so that in particular the associated Siegel modular threefold is isomorphic to \(\mathbb{P}^{3}\). The author shows that the squares of these theta functions are modular with respect to the larger group \(\Gamma_{2}^{2}(2,4)\), and that they generate the ring of Siegel modular forms of integral weights (with the same characters) with respect to that group. This squaring induces a map of degree 8 between these rational Siegel modular threefolds. Now, ``multiplication of \(B\) by 2'' defines an isomorphism between the two quotients of order 96 described above, which is also compatible with the resulting actions on \(\mathbb{P}^{3}\). Consider now an intermediate group between \(\Gamma_{2}^{2}(2,4)\) and \(\Gamma_{0}(2)\) and the intermediate group between \(\Gamma_{2}(2,4)\) and \(\Gamma_{0}^{0}(2)\) corresponding to the same quotient of our order 96 group. Gathering these results, one obtains the (equivariant) isomorphism between the associated Siegel modular threefolds, as well as the degree 8 endomorphisms. This covers the case of the Igusa quartic (associated with the principal congruence subgroup \(\Gamma(2)\)), and a similar analysis involving the Fricke involution of index 2, which normalizes \(\Gamma_{2}^{2}(2,4)\) and \(\Gamma_{0}(2)\), produces another family of (unirational) Siegel modular threefolds with such an endomorphism. The paper concludes with a few results on Siegel modular varieties of higher dimensions. As another application of these theta functions, a non-vanishing holomorphic 5-form on the 6-dimensional Siegel modular variety associated with the genus 3 equivalent of \(\Gamma_{2}(2,4)\), implying that this variety cannot be unirational. The obstacles in extending the methods of this paper to other higher dimensions are also briefly discussed.