SearcharxivSearch

arXiv subjects

William D. Banks

Publications and source records attributed to William D. Banks.

At least 19 recordsLinked to original sources

Zeros of the Dirichlet series of even zeta values

We give a complete unconditional description of the zero set of the Dirichlet series $\mathbf{D}(s) = \sum_{n \ge 1} ζ(2n)n^{-s}$, which continues meromorphically to $\mathbb{C}$ with a single simple pole at $s = 1$. The series possesses neither an Euler product nor a self-dual functional equation, and descriptions with this level of completeness are exceedingly rare for such series. The key input is an exact functional equation of Hecke type, obtained from the Lipschitz summation formula, which expresses $\mathbf{D}$ in the left half-plane as a gamma factor times a dual series over the complex logarithms of the perfect squares; Riemann's functional equation appears as a single column of the dual series. The zeros fall into four families. The half-plane $σ\ge σ_0 = 1.5001\ldots$ is zero-free, and $\mathbf{D}$ has a unique real zero $ρ_0 = 0.2004\ldots$. The zeros in the critical strip are perturbed $a$-points of $ζ$ for values of $a$ near $-(ζ(2)-1)$, and their counting function obeys a Riemann--von Mangoldt law. The remaining zeros form two complex-conjugate strings that recede into the left half-plane along explicit rays, are eventually simple, and satisfy an asymptotic with geometrically decaying error. The string geometry is governed by interference between the two smallest frequencies of the dual series, $2\log 2$ contributed by the entire part of $\mathbf{D}$ and $\pm 2πi$ contributed by $ζ$. No hypothesis of Riemann type is assumed at any point.

math.NT

Pair correlation for sums of two ordinates of zeros of the Riemann zeta function

Assuming the Riemann Hypothesis, we extend Montgomery's pair correlation method to study the distribution of differences between sums $γ_1+γ_2$ of two ordinates of nontrivial zeros of the Riemann zeta function. For the associated pair correlation function we prove that $G_2(α,T)=\{\log T/T^{2α}+4α^3/(3T^α)\} \{1+O(1/\log\log T)\}$ uniformly for $0\leα\le 2/3-2\log\log T/\log T$. In contrast with the conjectured GUE statistics of the ordinates themselves, this result points to an absence of level repulsion among sums of two ordinates, the two-point correlation function of the sums being identically one, as for a Poisson process.

math.NT

Multiple sums with the Möbius function

We establish nontrivial bounds for bilinear sums involving the Möbius function evaluated over solutions to a broad class of equations. Several of our results may be regarded as Möbius-function analogues of the ternary Goldbach problem. By contrast, the binary versions of our results remain out of reach, much like the binary Goldbach problem. Nevertheless, we make partial progress in this direction by restricting the range of the third variable as far as possible.

math.NT

A variant of the Linnik-Sprindzuk theorem for simple zeros of Dirichlet L-functions

For a primitive Dirichlet character $X$, a new hypothesis $RH_{sim}^\dagger[X]$ is introduced, which asserts that (1) all simple zeros of $L(s,X)$ in the critical strip are located on the critical line, and (2) these zeros satisfy some specific conditions on their vertical distribution. We show that $RH_{sim}^\dagger[X]$ (for any $X$) is a consequence of the generalized Riemann hypothesis. Assuming only the generalized Lindelöf hypothesis, we show that if $RH_{sim}^\dagger[X]$ holds for one primitive character $X$, then it holds for every such $X$. If this occurs, then for every character $χ$ (primitive or not), all simple zeros of $L(s,χ)$ in the critical strip are located on the critical line. In particular, Siegel zeros cannot exist in this situation.

math.NT

Shifting the ordinates of zeros of the Riemann zeta function

Let $y\ne 0$ and $C>0$. Under the Riemann Hypothesis, there is a number $T_*>0$ $($depending on $y$ and $C)$ such that for every $T\ge T_*$, both \[ ζ(\tfrac12+iγ)=0 \quad\text{and}\quadζ(\tfrac12+i(γ+y))\ne 0 \] hold for at least one $γ$ in the interval $[T,T(1+ε)]$, where $ε:=T^{-C/\log\log T}$.

math.NT

Comparing zeros of distinct Dirichlet L-functions

For any $θ>\frac13$, we show that there are constants $c_1,c_2>0$ that depend only on $θ$ for which the following property holds. If $χ_1,χ_2$ are two distinct primitive Dirichlet characters modulo $q$, and $T\ge c_1q^θ$, then $L(s,χ_1)$ and $L(s,χ_2)$ do not have the same zeros in the region $$\big\{s=σ+it\in{\mathbb C}:0<σ<1,~T \frac14$.

math.NT

Consecutive primes and IP sets

For an infinite set M of natural numbers, let FS(M) be the set of all nonzero finite sums of distinct numbers in M. An IP set is any set of the form FS(M). Let p_n denote the n-th prime number for each $n \ge 1$. A de Polignac number is any number m such that $p_{n+1}-p_n=m$ for infinitely many n. In this note, we show that every IP set of even natural numbers contains infinitely many de Polignac numbers.

math.NT

The Generalized Riemann Hypothesis from zeros of a single L-function

For each primitive Dirichlet character $χ$, a hypothesis ${\rm GRH}^\dagger[χ]$ is formulated in terms of zeros of the associated $L$-function $L(s,χ)$. It is shown that for any such character, ${\rm GRH}^\dagger[χ]$ is equivalent to the Generalized Riemann Hypothesis.

math.NT

The Riemann Hypothesis via the generalized von Mangoldt Function

Gonek, Graham, and Lee have shown recently that the Riemann Hypothesis (RH) can be reformulated in terms of certain asymptotic estimates for twisted sums with von Mangoldt function $Λ$. Building on their ideas, for each $k\in\mathbb{N}$, we study twisted sums with the \emph{generalized von Mangoldt function} $$ Λ_k(n):=\sum_{d\,\mid\,n}μ(d)\Big(\log\frac{n}{d}\,\Big)^k $$ and establish similar connections with RH. For example, for $k=2$ we show that RH is equivalent to the assertion that, for any fixed $ε>0$, the estimate $$ \sum_{n\leq x}Λ_2(n)n^{-iy} =\frac{2x^{1-iy}(\log x-C_0)}{(1-iy)} -\frac{2x^{1-iy}}{(1-iy)^2} +O\big(x^{1/2}(x+|y|)^ε\big) $$ holds uniformly for all $x,y\in\mathbb{R}$, $x\geq 2$; hence, the validity of RH is governed by the distribution of almost-primes in the integers. We obtain similar results for the function $$ Λ^k:=\mathop{\underbrace{\,Λ\star\cdots\starΛ\,}}\limits_{k\text{~copies}}\,, $$ the $k$-fold convolution of the von Mangoldt function.

math.NT

Towards the Generalized Riemann Hypothesis using only zeros of the Riemann zeta function

For any real $β_0\in[\tfrac12,1)$, let ${\rm GRH}[β_0]$ be the assertion that for every Dirichlet character $χ$ and all zeros $ρ=β+iγ$ of $L(s,χ)$, one has $β\leβ_0$ (in particular, ${\rm GRH}[\frac12]$ is the Generalized Riemann Hypothesis). In this paper, we show that the validity of ${\rm GRH}[\frac{9}{10}]$ depends only on certain distributional properties of the zeros of the Riemann zeta function $ζ(s)$. No conditions are imposed on the zeros of nonprincipal Dirichlet $L$-functions.

math.NT

Sums with the Mobius function twisted by characters with powerful moduli

In their recent work, the authors (2016) have combined classical ideas of A. G. Postnikov (1956) and N. M. Korobov (1974) to derive improved bounds on short character sums for certain nonprincipal characters with powerful moduli. In the present paper, these results are used to bound sums of the Mobius function twisted by characters of the same type, complementing and improving some earlier work of B. Green (2012). To achieve this, we obtain a series of results about the size and zero-free region of $L$-functions with the same class of moduli.

math.NT

On certain zeta functions associated with Beatty sequences

Let $α>1$ be an irrational number of finite type $τ$. In this paper, we introduce and study a zeta function $Z_α^\sharp(r,q;s)$ that is closely related to the Lipschitz-Lerch zeta function and is naturally associated with the Beatty sequence ${\mathcal B}(α):=(\lfloorαm\rfloor)_{m\in{\mathbb N}}$. If $r$ is an element of the lattice ${\mathbb Z}+{\mathbb Z}α^{-1}$, then $Z_α^\sharp(r,q;s)$ continues analytically to the half-plane $\{σ>-1/τ\}$ with its only singularity being a simple pole at $s=1$. If $r\not\in{\mathbb Z}+{\mathbb Z}α^{-1}$, then $Z_α^\sharp(r,q;s)$ extends analytically to the half-plane $\{σ>1-1/(2τ^2)\}$ and has no singularity in that region.

math.NT

Consecutive primes and Beatty sequences

Fix irrational numbers $α,\hatα>1$ of finite type and real numbers $β,\hatβ\ge 0$, and let $B$ and $\hat B$ be the Beatty sequences $$ B:=(\lfloorαm+β\rfloor)_{m\ge 1}\quad\text{and}\quad\hat B:=(\lfloor\hatαm+\hatβ\rfloor)_{m\ge 1}. $$ In this note, we study the distribution of pairs $(p,p^\sharp)$ of consecutive primes for which $p\in B$ and $p^\sharp\in\hat B$. Under a strong (but widely accepted) form of the Hardy-Littlewood conjectures, we show that $$ \big|\{p\le x:p\in B\text{ and }p^\sharp\in\hat B\}\big|=(α\hatα)^{-1}π(x)+O\big(x(\log x)^{-3/2+ε}\big), $$ where $π(x)$ is the prime counting function.

math.NT

On the number of distinct quadratic fields generated by the Shanks sequence

Let $g>1$ be an integer and $f(X)\in{\mathbb Z}[X]$ a polynomial of positive degree with no multiple roots, and put $u(n)=f(g^n)$. In this note, we study the sequence of quadratic fields ${\mathbb Q}(\sqrt{u(n)}\,)$ as $n$ varies over the consecutive integers $M+1,\ldots,M+N$. Fields of this type include Shanks fields and their generalizations. Using the square sieve together with new bounds on character sums, we improve an upper bound of Luca and Shparlinski (2009) on the number of $n \in \{M+1,\ldots,M+N\}$ with ${\mathbb Q}(\sqrt{u(n)}\,) = {\mathbb Q}(\sqrt{s}\,)$ for a given squarefree integer $s$.

math.NT