Non-vanishing of Taylor coefficients and Poincaré series (Q1937986): Difference between revisions

One of the most important aspects of modular forms is that they can be expanded into a Fourier series. The Fourier coefficients of many examples of modular forms are related to specific objects in number theory, arithmetic geometry or other branches of mathematics. However, as holomorphic functions on the upper half plane modular forms can also be expanded into a Taylor series. In [\textit{J. H. Bruinier} et al., The 1-2-3 of modular forms. Lectures at a summer school in Nordfjordeid, Norway, 2004. Universitext. Berlin: Springer (2008; Zbl 1197.11047)] Zagier considers the coefficients of a variant of the usual Taylor series of a modular forms at a CM point, see [loc. cit.], p. 52. It turns out that these coefficients have also number theoretic applications. They can be computed by some recursive formula. Proposition 28 of [loc. cit.] gives the coefficients of the Eisenstein \(E_4\) at \(i\). The present paper deals with the same type of Taylor expansions of modular forms at CM points. The authors provide a similar formula of the Taylor coefficients of the Delta function at a CM point in terms of certain polynomials, see Theorem 1.2. Studying this sequence of polynomials, the authors are able to prove that all Taylor coefficients of \(\Delta\) at particular CM points (of a finite set of small discriminants) are non-zero (Theorem 1.3). The authors point out that the tempting conjecture that the Taylor coefficients of \(\Delta\) are non-zero at all CM points is false. In fact, they prove that there exist Taylor coefficients of \(\Delta\) at \(\approx 1.344i\) which are zero. In Theorem 1.5 a more general vanishing result for an arbitrary cusp form is presented. As an application, the authors show that Poincaré series \(P_{z, m}\in S_{12}(\Gamma)\) of elliptic type, which are associated to particular CM points in the upper half, plane don't vanish. This result is obtained using the relation \[ \langle f, P_{z,m}\rangle = c_{z}(f,m)\left[\frac{\pi(k-2)!m!}{2^{k-3}(m+k-1)!}\right] \] which is analogous to the well known relation for the classical Poincaré series.