Large values of derivatives of the Riemann zeta function on vertical homogeneous progressions

Qiyu Yang 1, Shengbo Zhao 2 · 2025

Read Large values of derivatives of the Riemann zeta function on vertical homogeneous progressions by Qiyu Yang 1, Shengbo Zhao 2 (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-2510-18576.txt.

Introduction

Qiyu Yang and Shengbo Zhao prove a lower bound in a short paper. They study the Riemann zeta function. They look at its derivatives along a vertical line. The points on that line are spaced evenly. The zeta function lies at the heart of deep questions about primes. The authors use the resonance method. They show a clear result for a fit range of the real part. At these discrete points, the largest values grow just as in the continuous case.

Catalog record

CREATORQiyu Yang 1, Shengbo Zhao 2
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-18576

Browse the full library · About the Tower of Babel

Full text

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

Large values of derivatives of the Riemann zeta function on vertical homogeneous progressions Qiyu Yang 1, Shengbo Zhao 2 Abstract . In this paper, we establish lower bounds for the maximum of derivatives of the Riemann zeta function on vertical homogeneous progressions. When the real part σ \sigma lies within a suitable range, we show that the discrete case has a similar order of magnitude to the continuous case, using the resonance method. 2020 Mathematics Subject Classification Primary 11M06, 11N37. 1. Introduction The study of the Riemann zeta function \zeta(s) has always been an important topic in analytic number theory, and the investigation of large values plays a crucial role in describing the value distribution of the Riemann zeta function. Here and throughout, we write \log_{k} for the k -th iterated logarithm. On the critical line, i.e. \sigma=1/2 , Soundararajan [ 17 ] introduced a highly effective technique, known as the resonance method, and proved the following lower bound \max_{T\leq t\leq 2T}\Big|\zeta\Big(\frac{1}{2}+it\Big)\Big|\geqslant\exp\bigg((1+o(1))\sqrt{\frac{\log T}{\log_{2}T}}\bigg) as T\to\infty . This method originates Voronin’s work [ 18 ] , which was later refined by Bondarenko and Seip [ 4 , 5 ] , who established the following bound for sufficiently large T, \max_{\sqrt{T}\leq t\leq T}\Big|\zeta\Big(\frac{1}{2}+it\Big)\Big|\geq\exp\bigg((1+o(1))\sqrt{\frac{\log T\log_{3}T}{\log_{2}T}}\bigg). Notably, this result was later slightly improved by de la Bretèche and Tenenbaum [ 9 ] using optimized GCD sums. Specifically, the leading term in the exponent was improved from (1+o(1)) to (\sqrt{2}+o(1)) . For \sigma=1 , Aistleitner, Mahatab and Munsch [ 2 ] established the following currently best known lower bound when T\to\infty , \max_{\sqrt{T}\leq t\leq T}|\zeta(1+it)|\geq e^{E}(\log_{2}T+\log_{3}T-c) for some constant c , where E is the Euler–Mascheroni constant. They developed the long resonance method, refining Soundararajan’s original approach. For the case where 1/2<\sigma<1 , there are also \Omega -results established for \zeta(s) in [ 1 , 6 , 16 ] and so on. Large values of derivatives of \zeta(s) have been extensively studied by Yang [ 20 , 21 ] in recent years. In this paper, let \mathbb{N} denote the set of all non-negative integers. As \sigma=1/2 , in [ 20 ] he proved that if T is sufficiently large, \max_{t\in[0,T]}\Big|\zeta^{(j)}\Big(\frac{1}{2}+it\Big)\Big|\geqslant\exp\bigg(\sqrt{2}\cdot\sqrt{\frac{\log T\log_{3}T}{\log_{2}T}}\bigg) for all j\in\mathbb{N} . Furthermore, he also obtained extreme values of \zeta^{(j)}(\sigma+it) in the critical strip 1/2<\sigma<1 , see [ 20 , Theorem 2.(B)] . The above results showed that large values of derivatives of \zeta(s) have the same order of magnitude as the currently known extreme values of \zeta(s) itself, up to constants in the exponent. Later, Yang [ 21 ] showed that uniformly for all positive integers j\leq(\log_{3}T)/(\log_{4}T) , \max_{T\leq t\leq 2T}|\zeta^{(j)}(1+it)|\geq(Y_{j}+o(1))(\log_{2}T)^{j+1}, where Y_{j}=\int_{0}^{\infty}u^{j}\rho(u)\mathrm{d}u and \rho(u) is the Dickman function. Including Yang’s work, almost all above results utilize either Soundararajan’s resonance method or its modified versions. For more results and details about the resonance method, we recommend [ 3 , 8 , 19 , 22 ] and the references therein. Building on these works, we study large values of derivatives of \zeta(s) on vertical arithmetic progressions, focusing particularly on homogeneous progressions. Our study is inspired by Li and Radziwiłł [ 13 ] and Minelli and Sourmelidis [ 14 ] . Let \mathbb{R} denote the set of all real numbers in this paper, and let \alpha>0 and \beta\in\mathbb{R} . Li and Radziwiłł showed that infinitely many positive integers \ell satisfy \bigg|\zeta\Big(\frac{1}{2}+i(\alpha\ell+\beta)\Big)\bigg|\gg\exp\bigg((1+o(1))\sqrt{\frac{\log\ell}{6\log_{2}\ell}}\bigg). Subsequently, Minelli and Sourmelidis continued their research on related problems. To obtain larger values, they restricted \beta=0 and considered lower bounds for the maximum of |\zeta(1/2+i\alpha\ell)| and |\zeta(1+i\alpha\ell)| on homogeneous progressions, see [ 14 , Remark 4.1] for reasons. Using approaches in [ 4 , 5 ] and [ 2 ] , they obtained following conclusions respectively. If N\geq 1 is sufficiently large, they showed that \max_{\sqrt{N}\leq\ell\leq N\log N}\bigg|\zeta\Big(\frac{1}{2}+i\alpha\ell\Big)\bigg|\geq\exp\bigg(\frac{1}{\sqrt{2}}\cdot\sqrt{\frac{\log N\log_{3}N}{\log_{2}N}}\bigg) and \max_{\sqrt{N}\leq\ell\leq N}|\zeta(1+i\alpha\ell)|\geq e^{E}(\log_{2}N+\log_{3}N+O(1)). Motivated by [ 10 , 14 , 20 , 21 ] , we aim to obtain large values of derivatives of \zeta(s) on vertical homogeneous progressions near the lines \sigma=1/2 and \sigma=1 . Our results demonstrate that large values on homogeneous progressions achieve the same order of magnitude as those in the continuous case. Henceforth, we assume that \ell\in\mathbb{Z} , where \mathbb{Z} is the set of all integers. Theorem 1.1 . Let j\in\mathbb{N} and fix \alpha>0 . Let A be any positive real number and N be sufficiently large. Set \sigma_{A}\coloneqq 1/2+A/(\log_{2}N) . We have \max_{\sqrt{N}\leq\ell\leq N}|\zeta^{(j)}(\sigma_{A}+i\alpha\ell)|\geq\exp\bigg((\lambda(A)+o(1))\sqrt{\frac{\log N\log_{3}N}{\log_{2}N}}\bigg), where the leading term \lambda(A)=\frac{1}{\sqrt{2}(e-1)e^{A}}. Compared with [ 20 , Theorem 2.(A)] , our Theorem 1.1 shows that when 0<\sigma-1/2\ll(\log_{2}N)^{-1} , the large value of \zeta^{(j)}(\sigma+i\alpha\ell) can reach a similar order of magnitude as those of \zeta^{(j)}(1/2+it) as N\to\infty , while the leading constant in the exponent is slightly smaller. Remark . In fact, we can obtain the large values over any interval [N^{\theta},N],0<\theta<1 by an almost identical method. In this case, the leading term in the exponent \lambda(A)=\lambda(A,\theta) will depend on \theta . For simplicity, we consider only the case \theta=1/2. Then, we state the following result. Compared with [ 10 , Theorem 3] , Theorem 1.2 yields a same lower bound when 0<1-\sigma\ll(\log_{2}N)^{-1} . This also indicates that the currently known large values in the continuous case can also be realized on vertical homogeneous progressions. Theorem 1.2 . Let \alpha>0 be fixed. Let A be any positive real number and N be sufficiently large. Set \sigma^{\prime}_{A}\coloneqq 1-A/(\log_{2}N) . We have \max_{\sqrt{N}\leq\ell\leq N}|\zeta^{(j)}(\sigma^{\prime}_{A}+i\alpha\ell)|\geq(D_{j}(A)+o(1))(\log_{2}N)^{j+1} uniformly for all positive integers j\leq(\log_{3}N)/(\log_{4}N), where D_{j}(A)=\int_{0}^{\infty}e^{Au}u^{j}\rho(u)\mathrm{d}u. Remark . In [ 21 , Theorem 1] , Yang showed that the number of positive integers j for which |\zeta^{(j)}(1+it)| takes larger values in the range t\in[T,2T] is smaller compared to [ 20 , Theorem 1] . In [ 21 ] , he removed the term -\log_{3}T and improved the leading term from e^{\gamma}\cdot j^{j}\cdot(j+1)^{-(j+1)} to Y_{j}=D_{j}(0). Moreover, Dong and Wei [ 12 ] further refined Yang’s result in [ 20 ] . A similar argument applies to |\zeta^{(j)}(\sigma^{\prime}_{A}+i\alpha\ell)| . By slightly modifying the proof in [ 20 , Section 3] , specifically by adjusting the choices of x and b , we can almost obtain a result similar to [ 20 , Theorem 1] . Although this result yields a smaller lower bound, it holds for more positive integers j . We do not pursue this further, as Theorem 1.2 suffices to illustrate the point discussed in the introduction. Finally, we introduce some notations. Let p be a prime number, and let \mathbb{C} denote the set of all complex numbers. Let \varepsilon>0 be an arbitrarily small number. We remark that each occurrence of \varepsilon may represent a different value. Finally, we denote the Fourier transform of a function f\in L^{1}(\mathbb{R}) as \widehat{f}(\xi):=\int_{\mathbb{R}}f(x)e^{-2\pi i\xi x}\mathrm{d}x. 2. Preliminary lemmas In this section, we present several lemmas that will be used subsequently. We begin with the following asymptotic formula for derivatives of \zeta(s) . Lemma 2.1 ( [ 21 ] , Lemma 1) . Fix \sigma_{0}\in(0,1) . If T is sufficiently large, then uniformly for \varepsilon>0 , t\in[T,2T] , \sigma\in[\sigma_{0}+\varepsilon,\infty) and all j\in\mathbb{N} , we have (-1)^{j}\zeta^{(j)}(\sigma+it)=\sum_{n\leqslant T}\frac{(\log n)^{j}}{n^{\sigma+it}}+O\Big(\frac{j!}{\varepsilon^{j}}\cdot T^{-\sigma+\varepsilon}\Big), where the implied constant only depends on \sigma_{0} . The next lemma is a key tool the subsequent calculations, providing an effective upper bound for mean squares. Lemma 2.2 ( [ 15 ] , Lemma 1.4) . If 0<W<V-2

Read the complete text