Spaces of modular forms generated by eta-quotients (Q2642515): Difference between revisions

Let \(\eta(z)\) be the usual Dedekind function: \[ \eta(z)= e^{\pi i/z}\prod^\infty_1 (1- e^{2\pi inz}),\quad z\in \mathcal H,\tag{1} \] \(\mathcal H\) denotes the upper half-plane. A function of the form \[ f(z)= \prod_{\delta\mid N} \eta(\delta z)^{r_\delta}\quad\text{with }N\in \mathbb Z^+,\;r_\delta\in \mathbb Z,\tag{2} \] is called an eta-quotient; if the \(r_\delta\geq 0\) for all \(\delta\mid N\), we call \(f(z)\) an eta-product. Clearly, \(f(z)\) of the form (2) is a modular form on the congruence subgroup \(\Gamma_0(N)\) in \(\text{SL}(2,\mathbb Z)\), with the weight \(\tfrac 12\sum_{\delta\mid N} r_\delta\). There is, furthermore, a multiplier system (MS) attached to \(f(z)\), consisting of (at worst) 24th roots of unity. The study of eta products/quotients goes back at least as far as the well-known work of \textit{M. Newman} [Proc. Lond. Math. Soc. (3) 7, 334--350 (1957; Zbl 0097.28701) and ibid. 9, 373--387 (1959; Zbl 0178.43001)]. Now, more than half a century later, interesting questions remain, often of the form: which spaces of modular forms are generated by eta products/quotients: The answers of course depend upon the group, weight and MS defining the space and, as well, upon the meaning of the term, ``generated by''. The article under review addresses this question when the group is \(\Gamma_0(N)\), for some special values of \(N\), the weight is an even integer, the \(\text{MS}\equiv 1\), and the modular forms are holomorphic in \(\mathcal H\), but not necessarily at the cusps. Here, ``generated by'' means ``expressible as a rational function in''. (In this case there is no distinction to be made between eta-products and eta-quotients.) The results are stated in two theorems: Theorem 1. Let \(N= 1-10,12,13,16,18,25\) (the \(N\) for which \(\Gamma_0(N)\) has genus \(0\)). Then every modular form on \(\Gamma_0(N)\) can be written as a rational function of eta-quotients. Theorem 2. Let \(p\) be a prime; consider the eta-quotient \[ \phi_p(z)= (\eta(z)/\eta(pz))^h,\quad h:={p-1\over \gcd(12,p-1)}. \] Let \(f(z)\) be a modular form of weight \(12m\), \(m\in \mathbb Z\), and \(\text{MS}\equiv 1\) on \(\Gamma_0(p)\). Then, \[ f(z)= {\Delta(pz)^{2m}\over \Delta(z)^m}\,\Biggl\{F(\phi_p(z), j(z))+ {g(z)\over\Delta(z)^h}\Biggr\}, \] where \(F(X,Y)\in \mathbb C[X,Y]\), \(j(z)\) is Klein's modular invariant on \(\text{SL}(2,\mathbb Z)\) and \(g(z)\) is an entire modular form of weight \(12h\). The principal tool used in the proofs of these results is the usual formula for the dimension of the space of entire modular forms of even integral weight and \(\text{MS}\equiv 1\) on \(\Gamma_0(N)\) (cf. 2.1). The dimension formula is, in turn, a consequence of the Riemann-Roch theorem, when applied to the compact Riemann surface \({}_{\Gamma_0(N)}\backslash{}^{ \mathcal H^+}\).