A Hecke correspondence theorem for nonanalytic automorphic integrals (Q1581872)

In a previous article [Acta Arith. 90, 155-171 (1999; Zbl 0935.11017)] the author began the study of nonanalytic automorphic integrals on the Hecke groups. These are real-analytic functions \(f\), defined in the upper half-plane by means of exponential series in \(z\) and \(\overline{z}\), and having a generalized modular transformation equation under the inversion \(z\to -1/z\). This equation involves a ``period'' \(q\) of \(f\) analogous to the period polynomials of Eichler integrals. The period is, in fact, what the author terms an ``axial log-polynomial sum'', that is, \(q(iy)\), \(y> 0\), is a finite sum of terms of the form \(\beta(iy)^\alpha \log^t(iy)\), where \(\alpha,\beta\) are complex and \(t\) is a nonnegative integer. The author's 1999 Acta article enunciated and dicussed, but without proof, a ``Riemann-Hecke-Bochner correspondence theorem'' for nonanalytic automorphic integrals. The paper under review presents the proof and derives some consequences. As is the case in earlier correspondence theorems, the author's correspondence associates to an appropriate exponential series \(f(z)\) (in \(z, \overline{z}\)) the Mellin transform \(\Phi_f(z)\) of \(f(z)\) (the sum of the constant terms). The theorem states that \(f(z)\) is a nonanalytic automorphic integral if and only if \(\Phi_f(z)\): (i) has a meromorphic continuation to the entire \(z\)-plane; (ii) has a functional equation of the classical type satisfied by Riemann's \(\zeta (z)\); (iii) remains bounded in vertical strips. The author also establishes a connection between the exponents \(\alpha\) occurring in the finite sum representing \(q(z)\) and the location and orders of the poles of \(f(z)\) (finite in number). Among the applications is a nice counterexample to a 1983 conjecture of the reviewer.