Theta blocks related to root systems (Q2089705)

A theta block is a function \[ \Theta_f(\tau,z) = \eta^{f(0)}(\tau)\prod_{a=1}^\infty (\vartheta(\tau,az)/\eta(\tau))^{f(a)} \] where \(f\colon \mathbb{Z}_{\ge 0}\to \mathbb{Z}\) is a function with finite support, \(\eta\) is the Dedekind eta function and \(\vartheta\) is the Jacobi theta function. It is a meromorphic Jacobi form. If \(f(a) \ge 0\) for all \(a\) then \(\Theta_f\) is called a pure theta block. Gritsenko, Poor and Yuen [\textit{V. Gritsenko} et al., J. Number Theory 148, 164--195 (2015; Zbl 1380.11050)] conjectured that if \(\Theta_f\) is a holomorphic Jacobi form of weight \(k\) and index \(m\), of vanishing order \(1\) in the variable \(q=e^{2\pi i \tau}\), then the Gritsenko lift of \(\Theta_f\) (a modular form for the paramodular group) is equal to the Borcherds lift of \(\Psi_f=-(\Theta_f|T_-(2))/\Theta_f\). This is known as the theta block conjecture. \par This paper introduces a many-variable version of this, starting with a Jacobi form \(\vartheta_R\) of lattice index associated to a root system \(R\). The lifts of Gritsenko and Borcherds are available for such forms and one may again ask about the forms that arise as lifts of both kinds. For these special Jacobi forms, they prove the analogue of the theta block conjecture, in which \(\Theta_f\) is simply replaced by \(\vartheta_R\). \par The method is first to determine the root systems \(R\) for which \(\vartheta_R\) satisfies the order \(1\) vanishing condition. There are eight such root systems: \(A_4\), \(A_1\oplus B_3\), \(A_1\oplus C_3\) and \(B_2\oplus G_2\) giving weight \(2\); \(3A_2\), \(3A_1\oplus A_3\) and \(2A_1\oplus A_2\oplus B_2\) giving weight \(3\); and \(8A_1\) giving weight \(4\). For each of these, the authors then show that the divisor of the Borcherds lift is contained in the divisor of the Gritsenko lift: for three of the eight, the computation has already been done by Gritsenko or by Gritsenko and the first author. The Borcherds product occurs elsewhere, in work of Scheithauer: it is a twisted denominator identity associated with \({\operatorname{Co}}_0\).