The \(GL(2)\) Rankin-Selberg method for functions not of rapid decay (Q1273702)

In [J. Fac. Sci., Univ. Tokyo, Sect. I A 28, 415-437 (1981; Zbl 0505.10011)] \textit{D. Zagier} introduced a method to carry over the ``Rankin-Selberg convolution'' to the case of functions \(F(z)\) invariant with respect to the full modular group \(SL(2,\mathbb{Z})\) but not of rapid decay as \(y=\text{Im} z\to+\infty\). The invariant functions \(F(z)\) to which Zagier's result applies are those with an expansion at \(i\infty\) of the form \[ F(z)= \sum^\infty_{-\infty} a_n(y)e^{2\pi inx}, z=x+iy,\;y>0, \] and such that \(F(z)= \psi(y)+ O(y^{-N})\), for all \(N>0\), as \(y\to+ \infty\), with \(\psi(y)\) a ``log-polynomial sum'', that is, \(\psi(y)= \sum^\ell_{i=1} C_iy^{\alpha_i} (\log y)^{n_i}\). Here, the \(C_i\) and \(\alpha_i\) are complex numbers and the \(n_i\) are nonnegative rational integers. (Note that the log-polynomial sums play a key role in Bochner's generalization of the Riemann-Hecke correspondence. See the reviewer's article [\textit{M. I. Knopp}, Invent. Math. 117, 361-374 (1994; Zbl 0816.11032), \S II].) While Zagier and, following him, others have derived this result by truncating the standard fundamental domain of \(SL(2,\mathbb{Z})\), the article under review avoids the truncation of the domain, and thereby achieves some simplification of the proof. The authors express the belief that in more general settings their modified perspective will be easier to apply than Zagier's more geometric approach.