Uniform bound for Hecke \(L\)-functions (Q2495349)

Let as usual \(H_j(s)\) denote the Hecke series attached to the Maass wave form for the \(j\)-th eigenvalue \({1\over4}+\kappa_j^2\) of the hyperbolic Laplacian acting in the Hilbert space of automorphic functions for the full modular group. Also let \( \alpha_j = | \rho_j(1)| ^2/\cosh(\pi\kappa_j), \) where \(\rho_j(1)\) is the first Fourier coefficient of the Maass wave form to which \(H_j(s)\) is attached. Let also \(H_{j,k}(s)\) denote Hecke \(L\)-functions associated with holomorphic cusp forms of weight \(2k\). The main object of this remarkable paper is to prove the subconvexity bounds \[ H_j(\textstyle{1\over2} + it) \ll_\varepsilon (\kappa_j + t)^{c+\varepsilon}, \quad H_{j,k}(\textstyle{1\over2} + it) \ll_\varepsilon (k + t)^{c+\varepsilon} \quad(t\geq0)\leqno(1) \] with \(c = 1/3\). The bound (1) (with \(c=1/3\)) is the analogue of the classical bound for the Riemann zeta-function \(\zeta(s)\) on the critical line, namely \[ \zeta^2(\textstyle{1\over2}+it) \ll_\varepsilon t^{c+\varepsilon}\qquad(c = 1/3).\leqno(2) \] It is also conjectured that (1) and (2) hold with \(1/3+\varepsilon\) replaced by \(\varepsilon\). This is the Lindelöf hypothesis for \(H_j(s), H_{j,k}(s)\) and \(\zeta(s)\), and it is certainly out of reach at present. However, the bounds proved by the authors are good ``subconvexity'' bounds, namely they cannot be deduced by the Phragmén-Lindelöf convexity principle and the functional equation for the \(L\)--function in question (the ``convexity'' bound in this case would be \(c=1/2\)). This is even more so, as in the case of \(H_j(s)\) and \(H_{j,k}(s)\) the bounds have to be uniform in \(\kappa_j\) (resp. \(k\)), which makes this problem quite difficult. The present work is related to several earlier works. For example, the reviewer proved [J. Théor. Nombres Bordx. 13, 554--568 (2001; Zbl 0994.11020)] \[ H_j({\textstyle{1\over2}}) \ll_\varepsilon \kappa_j^{1/3+\varepsilon}, \leqno(3) \] so that (3) is a special case of (1), and represents the strongest known bound for \(H_j({1\over2})\). The first author obtained several relevant results, e.g., [Proc. Turku Symp. Number Theory, de Gruyter, Berlin, 167--177 (2001; Zbl 0972.11041), J. Number Theory 108, No. 1, 157--168 (2004; Zbl 1073.11034) and Publ. Inst. Math., Nouv. Sér. 76 (90), 41--55 (2004; Zbl 1098.11033)]. The bound (1) for \(H_j(s)\) (with \(c=1/3\)) was proved by \textit{T. Meurman} [Coll. Math. Soc. J. Bolyai 51, Number Theory, Budapest 1987, North Holland, 325--354 (1989; Zbl 0724.11029)] for \(t\gg \kappa_j^3\). The authors deduce their main result from several other results, each of its own interest. Let \(G=(K+t)^{4/3}K^{-1-\varepsilon}\), \(0\leq t\leq K^{3/2-\varepsilon}, \,K\geq K_0.\) Then their Theorem 1 states that \[ \sum_{K\leq\kappa_j\leq K+G}\alpha_j | H_j({\textstyle{1\over2}} +it)| ^4 \ll_\varepsilon GK^{1+\varepsilon} \qquad(t\geq0,\;1\leq G\leq K),\leqno(4) \] Theorem 2 brings forth the bound \[ \sum_{K\leq\kappa_j\leq 2K}\alpha_jH_j^2({\textstyle{1\over2}}) | H_j({\textstyle{1\over2}}+it)| ^2 \ll_\varepsilon (K^2+t^{4/3})^{1+\varepsilon}\quad(K\geq 0,\,t\geq 0), \] while Theorem 3 furnishes the analogue of (4) for \(H_{j,k}(s)\). The proofs of these important results, which are by necessity technical and involved, are based on several ingredients, including altogether nine lemmas. These include the well known Kloosterman-spectral sum formula of Bruggeman and Kuznetsov, the sum formula of H. Petersson, and Y. Motohashi's transformation formula for the smoothed summatory function of \(\sigma_\alpha(n)\sigma_\beta(n+f)\), where \(\sigma_\alpha(n) = \sum_{d| n}d^\alpha\) [e.g., see his comprehensive monograph ``Spectral theory of the Riemann zeta-function'', Cambridge Univ. Press (1997; Zbl 0878.11001)]. A detailed discussion of the structure of the proofs, the feasibility of its extensions etc. is given at the end of the paper.