The number of Hecke eigenvalues of same signs (Q1038701): Difference between revisions

The authors give the best possible lower bounds in order of magnitude for the number of positive and negative Hecke eigenvalues thereby improving upon a recent work of \textit{W. Kohnen, Y.-K. Lau} and \textit{I. Shparlinski} [J. Aust. Math. Soc. 85, No. 1, 87-94 (2008; Zbl 1241.11053)]. Also, we study an analogous problem for short intervals. Let \(k\geq 2\) be an even integer and \(N\geq \) be squarefree. Among all holomorphic cusp forms of weight \(k\) for the congruence subgroup \(\Gamma_0(N)\), there are finitely many of them whose Fourier coefficients in the expansion at the cusp \(\infty\), \[ f (z) =\sum_{n=1}^\infty \lambda_f(n)n^{(k-1)/2}e^{2\pi inz}\qquad(\text{Im}\,z > 0), \] are the Hecke eigenvalues. Up to scalar multiples, these forms are the only simultaneous eigenfunctions of all Hecke operators, called primitive forms. Let \(H_k^*(N)\) denote the set of all primitive forms of weight \(k\) for \(\Gamma_0(N)\). One central problem in modular form theory is to study the Hecke eigenvalues \(\lambda_f(n)\). In this paper, the authors investigate the Hecke eigenvalues of the same sign. Kohnen, Lau and Shparlinski [loc. cit., Theorem 1] proved \[ \mathcal N^\pm_f (x):=\sum_{{n\leq x,\,(n,N)=1}\atop{\lambda_f(n){>\atop<} 0}}1\gg_f\frac x {(\log x)^{17}} \] for \(x\geq x_0(f)\). Recently \textit{J. Wu} [Acta Arith. 137, No. 4, 333--344 (2009; Zbl 1232.11054), Corollary] improved this result by reducing the exponent 17 to \(1-1/\sqrt 3\), as a simple application of his estimates on power sums of Hecke eigenvalues. The exponent \(1-1/\sqrt 3\) can be improved to \(2-16/(3\pi)\) if one assumes Sato-Tate's conjecture. The author's first result is to remove the logarithmic factor by the \(\mathcal B\)-free number method, which is the best possible in order of magnitude. Theorem 1: Let \(f\in H_k^*(N)\). Then there is a constant \(x_0(f)\) such that the inequality \[ \mathcal N^\pm_f (x)\gg_f x \] holds for all \(x\geq x_0(f)\). Moreover they prove that there are plenty of eigenvalues of the same signs in intervals of length about \(x^{1/2}\) as follows. Theorem 2: Let \(f\in H_k^*(N)\). There is an absolute constant \(C>0\) such that for any \(\varepsilon > 0\) and all sufficiently large \(x\geq N^2x_0(k)\), we have \[ \mathcal N^\pm_f (x + C_N x^{1/2}) - \mathcal N^\pm_f (x) \gg_\varepsilon (Nx)^{1/4-\varepsilon}, \] where \[ C_N := CN^{1/2}\Psi(N)^3,\quad \Psi(N):=\sum_{d\mid N}d^{-1/2} \log(2d) \] and \(x_0(k)\) is a suitably large constant depending on \(k\) and the implied constant in \(\gg_\varepsilon\) depends only on \(\varepsilon\). Note that the result in Theorem 2 is uniform in the level \(N\), and its method of proof is based on [\textit{D. R. Heath-Brown} and \textit{K.-M. Tsang}, J. Number Theory 49, 73--83 (1994; Zbl 0810.11046)]. The exponent of \(\Psi(N)\) in \(C_N\) can be easily reduced to any number bigger than \(3/2\), which however may not be essential as \(\Psi(N)\) is already very small -- \(\log\Psi(N) = o(\sqrt{\log N})\). The range of \(x\geq N^2x_0(k)\) can also be refined to \(x\geq N^{1+\varepsilon}k ^A\) for some constant \(A > 0\).