Hilbert theta series and invariants of genus 2 curves (Q898809)

The coarse moduli space of smooth genus \(2\) curves over \(\mathbb{C}\) can analytically be described as the quotient \(\mathrm{Sp}_4 (\mathbb{Z}) \backslash \mathbb{H}_2\), where \(\mathbb{H}_2\) is a Siegel upper half space of dimension \(3\). The (absolute) Igusa invariants that furnish coordinates for this moduli space [\textit{J-I. Igusa}, Ann. Math. (2) 72, 612-649 (1960; Zbl 0122.39002)] can accordingly be lifted to \(\mathrm{Sp}_4 (\mathbb{Z})\)-invariant complex analytic functions on \(\mathbb{H}_2\), as was done by Igusa himself in [\textit{J-I. Igusa}, Am. J. Math. 89, 817--855 (1967; Zbl 0159.50401)]. It is useful to express these invariants in terms of Siegel theta constants, which are modular forms of weight \(1/2\) for the subgroup \(\Gamma (2) \subset \mathrm{Sp}_4 (\mathbb{Z})\), since such theta constants have good convergence properties. A similar construction is available for curves with extra endomorphisms. Let \(F\) be a real quadratic field, with different ideal \(\partial_F = \sqrt{D} \mathcal{O}_F\). Then by using both embeddings of \(F\) into \(\mathbb{R}\), the group \(\mathrm{SL}_2 (F)\) acts on \(\mathbb{H}^2\). Let \(\mathrm{SL}_2 (\mathcal{O}_F \oplus \partial_F^{-1})\) be the arithmetic subgroup of \(\mathrm{SL}_2 (F)\) that stabilizes the integral submodule \(\mathcal{O}_F \oplus \partial_F^{-1}\) of \(F \oplus F\). The authors describe an embedding map \(\mathbb{H}^2 \to \mathbb{H}_2\) under which Siegel's theta constants pull back to Hilbert theta constants, which are modular for a congruence subgroup of \(\mathrm{SL}_2 (\mathcal{O}_F \oplus \partial_F^{-1})\). It is shown how the Igusa invariants, as well as their level \(2\) refinements, the Rosenhain invariants, can be expressed in terms of these Hilbert theta constants. When \(D = 5\), several further variants are possible, the most notable of which involves the use of Gundlach invariants [\textit{K.B. Gundlach}, J. Reine Angew. Math. 220, 109--153 (1965; Zbl 0166.34002)]. These invariants generate the function field of the Hilbert modular surface that corresponds to the construction above. The authors express the Gundlach invariants in terms of Hilbert theta constants, and do the same for the Eisenstein series of weight \(2\). Some of these constructions generalize to another handful of discriminants. The authors apply their results by efficiently obtaining class polynomials for primitive quartic CM fields \(K\) that contain \(F\). Such class polynomials can be expressed in terms of either Igusa, Rosenhain or Gundlach invariants. The precision needed for provable results can be derived by combining variants of analytic estimates of \textit{A. Weng} [Math. Comput. 72, No. 241, 435-458 (2003; Zbl 1013.11023)] with height bounds due to [\textit{K. Lauter} and \textit{B. Viray}, Am. J. Math. 137, No. 2, 497--533 (2015; Zbl 1392.11033)] in the Igusa and Rosenhain case and to Bruinier--Yang [\textit{J. Bruinier} and \textit{T. Yang}, Invent. Math. 163, No. 2, 229--288 (2006; Zbl 1093.11041)] in the Gundlach case. The class polynomials for the Rosenhain invariants typically have the smallest height, and these can be used immediately to furnish equations of the corresponding CM curves (which in general will admit a further descent).