Divisors of modular forms on \(\Gamma_{0}(4)\) (Q1775573)

Let \(f(z)\) be meromorphic on the upper half plane. Define \(\theta(f(z))= {1\over 2\pi i}f'(z)\). From the paper: ``If \(f\) is a meromorphic modular form of weight \(k\in\mathbb{Z}\) on \(\text{SL}_2(\mathbb{Z})\), then it is known that \[ {\theta f\over f}= {k\over 12} E_2- f_\theta, \] [here \(E_2\) is the weight 2 Eisenstein series] where \(f_\theta\) is a meromorphic modular form of weight two. \textit{J. H. Bruinier}, \textit{W. Kohnen} and \textit{K. Ono} [Compos. Math. 140, No. 3, 552--566 (2004; Zbl 1060.11019)] determined \(f_\theta\) explicitly in terms of the values of a sequence of modular functions. This gives interesting results regarding the exponents of product expansions of modular forms, and recurrence relations for their Fourier coefficients. \textit{Y. Choie} and \textit{W. Kohnen} [Abh. Math. Semin. Univ. Hamb. 73, 289--295 (2003; Zbl 1046.11022)] examined some corresponding problems for functions modular on Hecke groups, and \textit{S. Ahlgren} [Math. Res. Lett. 10, No. 5--6, 787--798 (2003; Zbl 1149.11308)] considered the genus zero congruence subgroups \(\Gamma_0(p)\) for \(p\in \{2,3,5,7,13\}\). After this series of papers (the general outline for studying such questions was provided by \textit{J. H. Brunier} et al. [loc. cit.]) it is natural to consider the analogous problem for \(\Gamma_0(4)\). This case is of interest because it is the first where half-integral weight forms arise. This approach will clearly extend to other genus zero subgroups. We consider some applications to natural number-theoretic functions; for example, representations of integers as sums of squares. The paper's analog for \(\Gamma_0(4)\) is given below. Note that the constants \(\alpha_f\), \(\beta_f\) and \(\gamma_f\) are explicitly given in the text as an \(f_\theta\) and the slash operator \(| V(m)\). Theorem 3. Suppose that \(f(z)\in{\mathcal M}_k\). \(\exists \alpha_f\), \(\beta_f\), \(\gamma_f\) [and \(a\)] weight two meromorphic modular form \(f_\theta\) with \[ {\theta f\over f}=-f_\theta- \alpha_f E_2|V(2)- \beta_f E_2| V(4)- \gamma_f E_2. \] Here is an application to \(r_s(n)= \#\) representations of \(n\) as a sum of \(s\) squares: Corollary 1. Suppose \(s\) is a positive integer. Then for \(n\geq 1\), \(r_s(n)\) satisfies the recurrence relation \[ \begin{multlined} r_s(n)=\\ \sum_{\substack{ m_1+ 2m_2+\cdots+ (n-1)m_{n-1}= n\\ m_1\cdots m_{n-1}\geq 0}} (-1)^{m_1+\cdots+ m_{n-1}} {(m_1+\cdots+ m_{n-1}- 1)!\over m_1!\cdots! m_{n-1}!} r_s(1)^{m_1}\cdots r_s(n-1)^{m_{n-1}}\\ +(-1)^{n+1} {2s\over n}[\sigma(n)- \sigma(n/2)].\end{multlined} \] As usual, \(\sigma(n)= \sum_{d|n}d\).