Algebraic formulas for the coefficients of mock theta functions and Weyl vectors of Borcherds products (Q515587)

The fact that many interesting holomorphic and weakly holomorphic modular forms have rational (or algebraic) Fourier coefficients has been well known for a very long time. It is related to the objects being sections of line bundles defined over number fields. On the other hand, Fourier coefficients of mock modular forms conjecturally tend to contain many transcendental numbers (see the conjecture in [Ann. Math. (2) 172, No. 3, 2135--2181 (2010; Zbl 1244.11046)] of \textit{K. Ono} with the first author of this paper), except in the case where the shadow of the mock modular form is a unary theta series. This remaining case, existing only in weights \(\frac{1}{2}\) and \(\frac{3}{2}\), is the main topic of interest of this paper, which in particular gives explicit algebraic formulae for the Fourier coefficients of two mock theta functions defined by Ramanujan. There are, roughly speaking, two non-trivial ways to obtain modular forms of half-integral weight from a modular form \(F\) of weight 0. One is based on traces of singular moduli: For every discriminant \(D\) there are finitely many classes of singular moduli of discriminant \(D\), and by taking the corresponding traces of \(F\), the resulting generating series produce modular forms in many cases. The other way is via so-called theta lifts, which are integrals (sometimes regularized) of \(F\) against appropriate theta kernels. The first main result of this paper is an expression for the Fourier coefficients of Ramanujan's mock theta functions \(f\) and \(\omega\). Indeed, there is a weakly holomorphic modular function \(F\) on \(\Gamma_{0}(6)\), defined as the quotient of an Eisenstein series of weight 4 over an eta product, whose traces over correctly chosen singular moduli produce the desired Fourier coefficients. The proof is based on comparing the Millson lift of \(F\) with a modification of the vector-valued harmonic Maaß\ form defined by \textit{S. P. Zwegers} in [Contemp. Math. 291, 269--277 (2001; Zbl 1044.11029)]. Next, the paper addresses the construction of special pre-images of unary theta functions, both of weight \(\frac{1}{2}\) and of weight \(\frac{3}{2}\), under the corresponding \(\xi\)-operators, for any level \(N\). The idea is defining an analogue of the modular function \(F\) from above for level \(N\), and investigating its Millson and Kudla-Millson theta lifts. The construction of this general analogue of \(F\) is based on a Serre duality argument, combined with a delicate analysis of Fourier expansions at the various cusps in order to make sure that the only non-zero principal part of \(F\) is at the cusp \(\infty\). Using this the authors deduce the algebraicity and integrality of the traces of \(F\), under some mild assumptions on \(F\). It follows that pre-images of unary theta functions have algebraic Fourier coefficients, up to the addition of a cusp form in weight \(\frac{3}{2}\). As an application, Petersson pairings of such modular forms with theta functions are evaluated, and the rationality of Weyl vectors of certain Borcherds products at the various cusps is established (including the explicit expression for these vectors in certain cases). The paper is organized as follows. Section 1 is an introduction, including the defining expressions for some of the modular forms that are later used. Section 2 reviews the required parts of the theory of Weil representations, unary theta series, pairings of modular forms, traces of singular moduli, and theta lifts. In Section 3, the authors prove their formulae for the Fourier coefficients of \(f\) and \(\omega\). Section 4 constructs the modular function \(F\) (of general level), analyzes its Millson and its Kudla-Millson lifts, and proves the theorems about the rationality of its traces as well as of the Fourier coefficients of appropriate \(\xi\)-pre-images of unary theta series. Finally, Section 5 presents the results about pairings with theta series and Weyl vectors.