In this paper we calculate the asymptotics of various moments of the central values of Rankin-Selberg convolution L-functions of large level, thus generalizing the results and methods of W. Duke, J. Friedlander, and H. Iwaniec and of the authors. Consequences include convexity-breaking bounds, nonvanishing of a positive proportion of central values, and linear independence results for certain Hecke operators.
Abstract. We compute the asymptotics of twisted fourth power moments of modular L-functions of large prime level near the critical line. This allows us to prove some new non-vanishing results on the central values of automorphic L-functions, in particular those obtained by base change from GL2(Q) to GL2(K) for K a cyclic field of low degree.
We prove non-vanishing results for the central value of high derivatives of the complete L-function Λ(f, s) attached to primitive forms of weight 2 and prime level q. For fixed k ≥ 0 the proportion of primitive forms f such that Λ (k) (f, 1/2) = 0 is ≥ p k + o(1) with p k > 0 and p k = 1/2 + O(k −2), as the level q goes to infinity. This result is (asymptotically in k) optimal and analogous to a result of Conrey on the zeros of high derivatives of Riemann's ξ function lying on the critical line. As an application we obtain new strong unconditional bounds for the average order of vanishing of the forms f (i.e. the analytic rank of the Jacobian variety J 0 (q)).
Using the mollification method, we show that, for large q, at least 2Â3&O(k &2 ) of the set [4 (k) (/, 1Â2)] is non-zero, where 4(/, s) is the complete Dirichlet L-function and / runs over all even primitive characters modulo q.
Academic Press
Contents 1. Introduction. Given a large prime N, consider the Jacobian variety J 0 (N) of the modular curve X 0 (N ) = H/ 0 (N). There has been considerable study recently of the rank over Q of these varieties and their quotients, much of it dealing with two questions:• Are there large quotients of J 0 (N) with rank zero over the rationals?• How large a rank can a quotient of J 0 (N) have? See Mazur [9] and Merel [10] for arithmetic approaches to the first question; one consequence of the latter is the existence of a quotient with rank zero and dimension at least N 1/8 (out of a maximum possible dimension of D N = N/12 + ᏻ(1), since N is prime). The progress of Kolyvagin [8] and Gross and Zagier [6] toward the Birch-Swinnerton-Dyer conjecture allows one to investigate these questions through the "analytic rank" of J 0 (N ). This is defined (see, for example, [1]) to be the order of the zero at s = 1 of L(J 0 /Q, s) = f L f (s), where the product is taken over L-functions associated with the Hecke eigenbasis of S 2 ( 0 (N)), the weight-2 holomorphic forms on X 0 (N ) (of which there are D N ). In particular, if one can show that many of the L f (s)'s have order zero or one, then the corresponding quotients provide good answers to each of the questions. Along these lines, Duke [4] has used first and second moments of L-functions to show that at least N/(log N) 2 are nonzero at s = 1, which along with Kolyvagin's results may be used to give a corresponding improvement in the answer to the first question.One can also use Riemann's explicit formula for the L f 's (see Mestre [11]) to control the average order at s = 1, although this approach requires (at least at present) an appeal to the Riemann hypothesis to obtain the desired bounds. In recent papers, Brumer [1] and Murty [12] have used these techniques to show that (assuming the
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.