On the connection between zero-free regions and the error term in the prime number theorem

Frederik Broucke · 2025

Read On the connection between zero-free regions and the error term in the prime number theorem by Frederik Broucke (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-13780.txt.

Introduction

Frederik Broucke published a mathematics paper in 2025 linking zero-free regions of zeta functions to the error term in the prime number theorem. Where a zeta function provably has no zeros controls how sharply primes can be counted. For a wide class of such regions, the author derives an improved upper bound on the counting error. The method covers not just the classic zeta function but generalized Beurling versions.

Broucke also builds Beurling zeta functions with endlessly many zeros on a chosen curve and none beyond it. The matching oscillation result for the error term shows the new bound is nearly best possible. The paper tightens a classic bridge between zeros and primes.

Catalog record

CREATORFrederik Broucke
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE92 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-13780

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the connection between zero-free regions and the error term in the prime number theorem Thanks: F. Broucke was supported by a postdoctoral fellowship (grant number 12ZZH23N) of the Research Foundation – Flanders Frederik Broucke Abstract . We provide for a wide class of zero-free regions an upper bound for the error term in the Prime Number Theorem, refining works of Pintz (1980), Johnston (2024), and Révész (2024). Our method does not only apply to the Riemann zeta function, but to general Beurling zeta functions. Next we construct Beurling zeta functions having infinitely many zeros on a prescribed contour, and none to the right, for a wide class of such contours. We also deduce an oscillation result for the corresponding error term in the Prime Number Theorem, showing that our aforementioned refinement is close to being sharp. 2020 Mathematics Subject Classification Primary 11M26, 11N05; Secondary 11M41, 11N80 1. Introduction In his 1859 paper, Riemann communicated his revolutionary insight that the distribution of prime numbers is intimately connected with the distribution of zeros of the Riemann zeta function \zeta(s) . His ideas led to the first proofs of the Prime Number Theorem (PNT) by Hadamard and de la Vallée Poussin, which crucially make use of the fact that \zeta(s) does not vanish for \operatorname{Re}s\geq 1 . Later, de la Vallée Poussin proved that \zeta(s) does not vanish in the region \sigma>1-c/\log(\lvert t\rvert+2) for some c>0 (where we write s=\sigma+\mathrm{i}t ). From this he deduced the following form of the PNT with remainder: \psi(x)\coloneqq\sum_{n\leq x}\Lambda(n)=x+O\Bigl(x\exp\bigl(-c^{\prime}\sqrt{\log x}\bigr)\Bigr),\quad\text{for some }c^{\prime}>0. In his 1932 monograph [ 11 ] , Ingham posed the question of what error term would follow from a general zero-free region \sigma>1-\eta(\lvert t\rvert) for a certain function \eta . His initial result was greatly improved by János Pintz in 1980 [ 18 ] . Before stating Pintz’s theorem we define, for a given function \eta:[0,\infty)\to[0,1/2] , the function \omega_{\eta}(x)\coloneqq\inf_{t\geq 1}\bigl(\eta(t)\log x+\log t\bigr). When \eta is clear from the context, we will often omit the subscript and simply write \omega(x) . Theorem 1.1 (Pintz) . Suppose that \eta:[0,\infty)\to(0,1/2] is continuously decreasing, and suppose that \zeta(s) has no zeros in the domain \sigma>1-\eta(\lvert t\rvert) . Then for every \varepsilon>0 , \psi(x)-x\ll_{\varepsilon}x\exp\bigl(-(1-\varepsilon)\omega(x)\bigr). In the same paper, Pintz also showed the following converse. Theorem 1.2 (Pintz) . Suppose that \zeta(s) has infinitely many zeros in the domain \sigma\geq 1-\eta(\lvert t\rvert) , where f(u)\coloneqq\eta(\mathrm{e}^{u}) is such that f^{\prime}(u) is non-decreasing with limit 0 and strictly increasing to 0 if f\to 0 . Then for every \varepsilon>0 , \psi(x)-x=\Omega_{\pm}\Bigl(x\exp\bigl(-(1+\varepsilon)\omega(x)\bigr)\Bigr). These two theorems provide a close to definitive answer to Ingham’s question. Note that the Riemann hypothesis (RH) implies that \psi(x)-x\ll_{\varepsilon}x^{1/2+\varepsilon} , so that both theorems follow trivially from RH: the first because the conclusion is always true, the second because the premise is always false (at least if \eta<1/2 ). A natural broader context to study Ingham’s question is that of Beurling zeta functions attached to Beurling generalized number systems . Such zeta functions need not satisfy RH, and it is conceivable that a wide range of functions \eta can occur as describing the asymptotic best zero-free region 1 1 1 Avoiding a precise definition, one can imagine the curve 1-\eta(t)+\mathrm{i}t as connecting the extreme points of the convex hull of the set of all zeros. Instead of using a function \eta bounding a zero-free region, it is also possible to state versions of Pintz’s theorems which refer to the zeros directly. For this we refer to [ 23 ] , in particular to Section 8 regarding some subtleties for the converse Pintz theorem. of some Beurling zeta function. In fact, it is one of the aims of this paper to show this. In contrast, the Riemann zeta function has a fixed zero distribution and the error term in the PNT has a fixed asymptotic behavior. Hence, if \eta_{\mathrm{RZ}} is a function asymptotically describing this zero-distribution, the Pintz theorems for functions \eta which are asymptotically larger than or smaller than \eta_{\mathrm{RZ}} are trivial consequences of the one for \eta=\eta_{\mathrm{RZ}} . Recently, Révész [ 23 ] generalized Pintz’s theorems to the Beurling context (see Theorems 1.3 and 1.4 below), further confirming that 1-\eta(\lvert t\rvert)\longleftrightarrow x\mathrm{e}^{-\omega_{\eta}(x)} is the “true connection” between zero-free regions and error terms in the PNT. The aim of this paper is twofold. First, we prove a refinement of the first Pintz–Révész theorem replacing \varepsilon by a function tending to 0 . We note that in the case of the Riemann zeta function, such a refinement was recently obtained by Johnston [ 13 ] . Our refinement is slightly better and applies to general Beurling zeta functions, while Johnston’s result applies to a wider class of zero-free regions. Nonetheless, our theorem is sufficient to capture any “reasonable” choice of \eta . Second, we show that our obtained refinement is sharp, up to the value of the constant of the secondary term \varepsilon(x)\omega(x) in the exponential. We do this by constructing for every “reasonable” choice of \eta , a Beurling number system (\mathcal{P},\mathcal{N}) whose zeta function \zeta_{\mathcal{P}}(s) has infinitely many zeros on the contour \sigma=1-\eta(\lvert t\rvert) and none to the right, and for which we also obtain an oscillation result for \psi_{\mathcal{P}}(x)-x , matching the error term predicted by the refined Pintz–Révész theorem (up to the value of the aforementioned constant). 1.1. Beurling generalized number systems Beurling generalized number systems were introduced in [ 3 ] in order to investigate the minimal assumptions needed to prove the PNT. They form an abstraction of the multiplicative structure of the primes and integers, and typically lack any additive structure. A Beurling generalized number system is a pair of sequences (\mathcal{P},\mathcal{N}) where \mathcal{P} , the sequence of generalized primes , consists of real numbers p_{k} satisfying 1<p_{1}\leq p_{2}\leq\dotso and p_{k}\to\infty . The sequence of generalized primes \mathcal{N} consists of all possible products of generalized primes n_{j}=p_{1}^{\nu_{1}}\dotsm p_{k}^{\nu_{k}} (including the empty product n_{0}=1 ) ordered in a non-decreasing fashion. We allow repeated elements in both sequences; if a generalized integer has multiple factorizations into generalized primes, its multiplicity in the sequences \mathcal{N} equals the number of such distinct factorizations. It can be useful to think of these as different generalized integers with the same numerical value. (Alternatively, one may define a generalized number system as an abstract multiplicative semigroup with a multiplicative norm function.) Given a Beurling number system (\mathcal{P},\mathcal{N}) , we set for x\geq 1 \pi_{\mathcal{P}}(x)\coloneqq\#\{p_{k}\leq x\},\quad N_{\mathcal{P}}(x)\coloneqq\#\{n_{j}\leq x\}. Note that \pi_{\mathcal{P}}(1)=0 and N_{\mathcal{P}}(1)=1 . In this article, we will assume that (1.1) N_{\mathcal{P}}(x)=Ax+O(x^{\theta}), for some A>0 and \theta\in[0,1) . This assumption is referred to as the integers being “ \theta -well-behaved”; sometimes it is also called Axiom A. It guarantees that the associated Beurling zeta function , defined as \zeta_{\mathcal{P}}(s)\coloneqq\sum_{j\geq 0}\frac{1}{n_{j}^{s}}=\prod_{k\geq 1}\frac{1}{1-p_{k}^{-s}} and initially holomorphic on \operatorname{Re}s>1 due to the absolute convergence, is such that \zeta_{\mathcal{P}}(s)-\frac{A}{s-1} admits analytic continuation to the half-plane \operatorname{Re}s>\theta . Finally we define the weighted prime-counting functions \Pi_{\mathcal{P}}(x)\coloneqq\sum_{p_{k}^{\nu}\leq x}\frac{1}{\nu},\quad\psi_{\mathcal{P}}(x)\coloneqq\sum_{p_{k}^{\nu}\leq x}\log p_{k}=\int_{1}^{x}\log u\mathop{}\!\mathrm{d}\Pi_{\mathcal{P}}(u), which have Mellin–Stieltjes transform \int_{1}^{\infty}x^{-s}\mathop{}\!\mathrm{d}\Pi_{\mathcal{P}}(x)=\log\zeta_{\mathcal{P}}(s),\quad\int_{1}^{\infty}x^{-s}\mathop{}\!\mathrm{d}\psi_{\mathcal{P}}(x)=-\frac{\zeta_{\mathcal{P}}^{\prime}(s)}{\zeta_{\mathcal{P}}(s)}. More background on Beurling generalized number systems can be found in the book [ 9 ] . We mention that ( 1.1 ) implies the existence of an absolute constant c>0 such that \zeta_{\mathcal{P}}(s) has no zeros in the

Read the complete text