One-level density estimates for Dirichlet \(L\)-functions with extended support (Q6042784): Difference between revisions

The work of \textit{N. M. Katz} and \textit{P. Sarnak} [Random matrices, Frobenius eigenvalues, and monodromy. Providence, RI: American Mathematical Society (1999; Zbl 0958.11004)] suggests that for any smooth function \(\phi\) and any natural ``family'' of automorphic objects \(\mathcal{F}\) \[ \lim_{\sharp{\mathcal{F}}\to \infty}\frac{1}{\sharp{\mathcal{F}}}\sum_{\pi \in \mathcal{F}}\sum_{\gamma_\pi}\phi\left( \frac{ \log \mathfrak{c}_\pi}{2\pi}\gamma_\pi\right) =\int_{\mathbb{R}}\phi(x)K_\mathcal{F}(x)\, \mathrm{d}x\tag{2} \] under suitable notation. There is a vast literature supporting (2) in all of which a restriction is made on the support of the Fourier transform \(\hat{\phi}\) of \(\phi\) from the limitations of the relevant trace formula. Its application gives rise to the so-called ``diagonal'' and ``off-diagonal'' terms. To extend the supp \(\hat{\phi}\) is a central problem beyond that which is given by trivially bounding the off-diagonal terms. In the paper a first example of a family of (Dirichlet) \(L\)-functions in which the support is unconditionally enlarged beyond the ``diagonal'' range. Theorem. Let \(\Phi\) be a smooth function compactly supported in \(\left[\frac{1}{2},3 \right]\), and \(\phi\) be a smooth function such that \(\operatorname{supp}\hat{\phi} \subset \left(-2-\frac{50}{1093}, 2+\frac{50}{1093}\right)\). Then, as \(Q\to \infty\), \[ \sum_{q}\Phi\left( \frac{q}{Q} \right)\sum_{\substack{\chi \pmod q \\ \text{primitive}}}\sum_{\gamma_\chi}\phi\left( \frac{ \log Q}{2\pi}\gamma_\chi \right)=\hat{\phi}(0)\sum_{q}\Phi\left( \frac{q}{Q} \right)\sum_{\substack{\chi \pmod q\\ \text{primitive}}}1+o(Q^2).\tag{3} \] Here \(\frac{1}{2}+i\gamma_\chi\) correspond to nontrivial zeros of \(L(s,\chi)\) and since we do not assume the generalized Riemann hypothesis we allow the \(\gamma_\chi\) to be complex. As Corollary 2, a non-vanishing result is obtained for the Dirichlet \(L\)-function value \(L\left(\frac{1}{2}.\chi \right)\).