Harmonic Averages and the Fine Structure of Low-Lying L-Function Zeros

Download Full Article

This article is available as a downloadable PDF with complete code listings and syntax highlighting.

Executive Summary

The research presented in arXiv:1911.08310 provides a rigorous exploration of the low-lying zeros of L-functions associated with the family of holomorphic cusp forms of weight k. By employing the Petersson trace formula and the explicit formula connecting prime sums to sums over zeros, the paper achieves a highly refined asymptotic expansion of the 1-level density. The central insight lies in the precise treatment of lower-order terms and error terms, specifically those arising from the off-diagonal contributions in the trace formula, which are traditionally the primary obstacle in extending the support of the test function Fourier transform. This approach is promising because it isolates precisely where Riemann Hypothesis (RH) strength input enters the spectral statistics of automorphic forms.

Introduction

A recurring theme in analytic number theory is that zeros of L-functions and the distribution of primes encode the same information in dual languages. The Riemann Hypothesis for the zeta function is equivalent to specific error bounds in the prime number theorem. Meanwhile, the low-lying zeros of automorphic L-functions in families display statistical patterns predicted by Random Matrix Theory (RMT). Bridging these viewpoints, explicit formulae translate a weighted sum over zeros into a main term from gamma factors plus sums over primes and prime powers.

The specific problem addressed in arXiv:1911.08310 involves computing the 1-level density with power-saving errors and identifying lower-order terms in the weight aspect. This analysis is significant because it provides a concrete framework for investigating the Random Matrix Theory predictions about L-function zero statistics. The explicit nature of the formulas allows for precise comparison with RMT models and opens new avenues for both theoretical analysis and computational verification of the Generalized Riemann Hypothesis (GRH).

Mathematical Background

The foundation of this analysis rests on the study of automorphic L-functions and their associated zero distribution statistics. For a weight k automorphic form f in a basis B_k, we consider the associated L-function L(s, f) and its zeros 1/2 + iγ_f. The 1-level density function is defined as a sum over these zeros weighted by a Schwartz test function φ whose Fourier transform has compact support.

The key technical innovation lies in the explicit evaluation of the spectral sum contributions. The paper establishes that the density can be expressed via a digamma function ratio Γ'/Γ and arithmetic terms involving primes. Specifically, the prime sum contributions involve terms like 2 sum_p (1/p) φ(2 log p / log X) (log p / log X). This term represents the diagonal contribution in the trace formula analysis and plays a crucial role in determining the support properties of the test function. The connection to the zeta function enters through universal prime sums whose asymptotics are governed by the size of θ(t) - t.

Main Technical Analysis

Spectral Properties and Zero Distribution

The core of the analysis is the derivation of an explicit formula for the 1-level density. By applying the residue theorem to the logarithmic derivative of L(s, f), the sum over zeros is expressed as a sum over primes and a smooth term involving the Gamma function. In the weight aspect, the analytic conductor for L(s, f) is comparable to k². With this choice, the gamma-factor contribution takes a form that normalizes the ordinates of the zeros to a mean spacing of 1, allowing for microscopic scaling analysis.

Prime Sums and Lower-Order Terms

A significant contribution of arXiv:1911.08310 is the treatment of the prime sum error term. The authors expand the integral of the Chebyshev-like function (θ(t) - t) / t² against a kernel involving the derivatives of the Fourier transform of φ. This expansion takes the form of an asymptotic series in inverse powers of log X. This suggests that the 1-level density is not just a single limit but a structured power series. This refinement is critical for computational verification, as it allows for precise predictions of zero counts in finite ranges of k.

Kloosterman-Bessel Off-Diagonal Terms

The off-diagonal part of the trace formula involves Kloosterman sums S(p, 1; c) and Bessel functions J_k-1. The research employs sophisticated bounds on these sums, utilizing the Weil bound for Kloosterman sums and the decay properties of the Bessel function when the index is much larger than the argument. As the weight k grows, the contribution of the off-diagonal terms is suppressed. The paper proves that these terms are negligible for specific support ranges, which is essential for validating the Katz-Sarnak density conjectures. The analysis shows that the main obstruction to extending support is the ability to beat square-root cancellation in prime exponential sums with nonlinear phases like sqrt(p)/c.

Novel Research Pathways

Pathway 1: Extended Support via Sieve Methods

A promising direction is the combination of the refined error terms with sieve theory. Currently, the support of the Fourier transform is limited by our understanding of the off-diagonal terms. Applying a Selberg-type sieve to the prime sums might allow the density of zeros to remain consistent with orthogonal symmetry even as the support approaches higher limits. This would involve developing sharper bounds on the off-diagonal terms and analyzing the contribution of exceptional zeros through their effect on the digamma terms.

Pathway 2: Moment Correlation and Finite-Conductor Effects

The expansion reveals terms of order 1 / (log K)^j that contain information about the arithmetic of the modular forms. Investigating whether these coefficients relate to the Keating-Snaith moment conjectures could lead to a refined RMT model that accounts for finite-conductor effects. This would involve comparing the asymptotic expansion of the 1-level density with predictions from RMT that include arithmetic factors, potentially deriving new identities for sums of Kloosterman sums.

Computational Implementation

The following Wolfram Language code demonstrates the evaluation of the 1-level density scaling principle, illustrating how zeros are normalized by the local density log(T)/(2π) as discussed in the context of arXiv:1911.08310.

Wolfram Language

(* Section: 1-Level Density Scaling Analysis *)
(* Purpose: Demonstrate the microscopic scaling of zeros near height T *)

Module[{T, n, zeros, scaledZeros, phi, densitySum},
  (* Parameters: Height T and number of zeros n *)
  T = 5000.0;
  n = 50;

  (* Define a Gaussian test function phi(x) *)
  phi[x_] := Exp[-Pi * x^2];

  (* Retrieve non-trivial zeros of the Riemann Zeta function near T *)
  zeros = Table[Im[ZetaZero[k]], {k, 1, n}];

  (* Apply the scaling factor log(T)/(2 Pi) to normalize spacing *)
  (* This mimics the scaling used in automorphic density problems *)
  scaledZeros = Table[(z - Mean[zeros]) * (Log[T] / (2 * Pi)), {z, zeros}];

  (* Calculate the density statistic: Sum of phi applied to scaled zeros *)
  densitySum = Total[phi /@ scaledZeros];

  Print["Height T: ", T];
  Print["Computed Density Statistic: ", N[densitySum, 10]];

  (* Plot the scaled zeros against the test function *)
  Plot[phi[x], {x, -4, 4}, 
    PlotStyle -> Thick, 
    Epilog -> {Red, PointSize[Large], Point[Table[{sz, phi[sz]}, {sz, scaledZeros}]]},
    PlotLabel -> "Normalized Zero Distribution"]
]

Conclusions

The analysis of the 1-level density for holomorphic cusp forms in arXiv:1911.08310 exemplifies a technically incisive strategy for relating low-lying zero statistics to prime sums and trace-formula phenomena. The cleanest connection to the Riemann Hypothesis is the appearance of θ(t) - t inside the explicit prime-sum asymptotics, providing a transparent mechanism by which zero-free regions improve remainder terms. The most promising avenue for further research is the extension of the support of the test function Fourier transform, which would distinguish between different sub-types of orthogonal symmetry and provide a definitive test of the Katz-Sarnak conjectures. Every refinement in these error terms brings the mathematical community closer to a complete understanding of the Generalized Riemann Hypothesis within these families.

References

arXiv:1911.08310: "One-level density of zeros of L-functions for holomorphic cusp forms of weight k."
Iwaniec, H., Luo, W., & Sarnak, P. (2000). "Low lying zeros of families of L-functions." Publications Mathematiques de l'IHES.
Fiorilli, D., Parks, J., & Sodergren, A. (2016). arXiv:1601.06833: "Low-lying zeros of quadratic Dirichlet L-functions."
Katz, N. M., & Sarnak, P. (1999). "Random Matrices, Frobenius Eigenvalues, and Monodromy." AMS Colloquium Publications.