Proof of Existence of Integers Excluding Two Residue Values in a Specific Range

Liang Zhao · 2025

Read Proof of Existence of Integers Excluding Two Residue Values in a Specific Range by Liang Zhao (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-2501-15707.txt.

Introduction

Liang Zhao, writing alone in 2025, proved a claim about whole numbers that dodge two set remainders. Working modulo primes up to a chosen prime, he studied the span between that prime squared and the next prime squared. Using deep counting tools from analytic number theory, he bounded how much of the span the banned remainder classes can cover. The work adds to the picture of how residue classes spread over long intervals, with links to wider questions in number theory.

Catalog record

CREATORLiang Zhao
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE14 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-15707

Browse the full library · About the Tower of Babel

Full text

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

Proof of Existence of Integers Excluding Two Residue Values in a Specific Range Liang Zhao Abstract This paper investigates the existence of integers that exclude two specific residence values modulo primes up to p k p_{k} within the interval [ p k 2 , p k + 1 2 ] [p_{k}^{2},p_{k+1}^{2}] . Using asymptotic results from analytic number theory, we establish bounds on the proportion of integers excluded by the union of residue classes. The findings highlight the density of residue class coverage in large intervals, contributing to the understanding of modular systems and their implications in number theory and related fields. Introduction The study of residue classes of integers modulo prime numbers has long been a cornerstone of analytic number theory. A significant focus in this domain is to understand the distribution of integers that satisfy specific residue class conditions within bounded intervals. This problem finds relevance in cryptography, random number generation, and the analysis of modular systems. In this paper, we investigate the existence of integers n within the interval [p_{k}^{2},p_{k+1}^{2}] such that for any prime p\leq x , n does not have two specific residue values r_{1} and r_{2} when divided by p . The motivation for focusing on this interval arises from its mathematical structure and the inherent growth rate of prime numbers with special properties. By leveraging the principles of inclusion-exclusion and known results on prime number distributions, we aim to establish bounds on the proportion of integers that can be covered by the union of such residue classes and their complements. Our main objective is to demonstrate that no matter how we select residue classes modulo primes up to p_{k} , there always exists at least one integer in the specified interval that does not fall into any of the predefined residue classes. This result emphasizes the sparsity of residue class coverings over large intervals, providing new insights into the interplay between prime numbers and modular arithmetic. A Mertens–Type Formula for \prod_{2<p<x}(1-2/p) , Excluding p=2 Theorem 1 . Let P(x)\;:=\;\prod_{\begin{subarray}{c}2<p<x\\ p\,\mathrm{prime}\end{subarray}}\Bigl(1-\tfrac{2}{p}\Bigr). Then there is a positive constant \widetilde{C}_{2} such that, for x\to\infty , P(x)\;=\;\widetilde{C}_{2}\,\frac{1}{(\ln x)^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln x}\bigr)\Bigr]. Moreover, \widetilde{C}_{2}\;=\;\exp\!\Bigl(1.5\;-\;2\,[\,M_{1}+S\,]\Bigr)\;\approx\;1.07, where \,M_{1}\approx 0.2614972 is the usual Meissel–Mertens constant for all primes and S=\sum_{p}1/p^{2}\approx 0.4522474 . Proof. Define \,P(x)=\prod_{2<p<x}(1-2/p) . Taking logarithms: \ln P(x)\;=\;\sum_{\begin{subarray}{c}2<p<x\end{subarray}}\ln\Bigl(1-\tfrac{2}{p}\Bigr). For each prime p>2 , we have \ln\Bigl(1-\tfrac{2}{p}\Bigr)\;=\;-\;\tfrac{2}{p}\;-\;\tfrac{(2/p)^{2}}{2}\;-\;\tfrac{(2/p)^{3}}{3}\;-\;\cdots\;=\;-\;\tfrac{2}{p}\;-\;\tfrac{2}{p^{2}}\;+\;O\!\bigl(\tfrac{1}{p^{3}}\bigr). Hence \ln P(x)\;=\;\sum_{2<p<x}\Bigl[-\tfrac{2}{p}\;-\;\tfrac{2}{p^{2}}+O\!\bigl(\tfrac{1}{p^{3}}\bigr)\Bigr]\;=\;-2\,\sum_{2<p<x}\tfrac{1}{p}\;\;-\;2\,\sum_{2<p<x}\tfrac{1}{p^{2}}\;+\;O(1). Now use two standard expansions, but each for all primes p\leq x (including p=2 ) [ 1 ] : \sum_{p\leq x}\frac{1}{p}\;=\;\ln\ln x+M_{1}+O\!\bigl(\tfrac{1}{\ln x}\bigr),\quad\sum_{p\leq x}\frac{1}{p^{2}}\;=\;S+O\!\bigl(\tfrac{1}{x}\bigr), where M_{1}\approx 0.2614972,\quad S:=\sum_{p}\tfrac{1}{p^{2}}\approx 0.4522474. Since our product omits p=2 , we must remove that term from these sums: \sum_{\begin{subarray}{c}2<p<x\end{subarray}}\tfrac{1}{p}\;=\;\sum_{p\leq x}\tfrac{1}{p}\;-\;\tfrac{1}{2}\;=\;\ln\ln x+M_{1}-\tfrac{1}{2}+O\!\bigl(\tfrac{1}{\ln x}\bigr), \sum_{\begin{subarray}{c}2<p<x\end{subarray}}\tfrac{1}{p^{2}}\;=\;S-\tfrac{1}{4}\;+\;O\!\bigl(\tfrac{1}{x}\bigr). Thus \ln P(x)\;=\;-2\Bigl(\ln\ln x+M_{1}-\tfrac{1}{2}\Bigr)\;-\;2\Bigl(S-\tfrac{1}{4}\Bigr)\;+\;O\!\bigl(\tfrac{1}{\ln x}\bigr) (since 1/x and constant terms are easily absorbed into O(1/\ln x) for large x ). Combine constants: -\,2\Bigl(M_{1}-\tfrac{1}{2}\Bigr)-2\Bigl(S-\tfrac{1}{4}\Bigr)\;=\;-2(M_{1}+S)\;+\;1.5. Therefore \ln P(x)\;=\;-2\,\ln\ln x\;-\;2\,(M_{1}+S)\;+\;1.5\;+\;O\!\bigl(\tfrac{1}{\ln x}\bigr). Exponentiating, P(x)\;=\;\exp[-2\,\ln\ln x]\,\exp\bigl[\,1.5-2(M_{1}+S)\bigr]\,\Bigl[1+O\!\bigl(\tfrac{1}{\ln x}\bigr)\Bigr]. Since \exp[-2\,\ln\ln x]=(\ln x)^{-2} , we set \widetilde{C}_{2}\;:=\;\exp\!\Bigl[\,1.5-2\,(M_{1}+S)\Bigr]. That yields P(x)\;=\;\widetilde{C}_{2}\,\frac{1}{(\ln x)^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln x}\bigr)\Bigr]. Numerically, M_{1}+S\approx 0.7137446 , so -\,2(M_{1}+S)\approx-1.4274892 whose exponential is about 0.239; multiplying by \,e^{1.5}\approx 4.481689 gives \widetilde{C}_{2}\approx 1.07. ∎ Remark. A naive inclusion of p=2 in the product would yield the factor 1-2/2=0 , making the product zero for all x>2 . Hence it is essential to exclude p=2 . Once excluded, one must also subtract \tfrac{1}{2} and \tfrac{1}{4} from the classical prime sums \sum_{p\leq x}1/p and \sum_{p\leq x}1/p^{2} , respectively. That extra shift eventually multiplies the constant by \,e^{1.5}\approx 4.48 , changing it from \approx 0.239 (a naive guess) to \approx 1.07 . A Telescoping Difference for p_{k+1}^{2}P(p_{k+1}^{2})\;-\;p_{k}^{2}P(p_{k}^{2}) , Excluding p=2 We derive an asymptotic formula for the difference \Delta_{k}\;:=\;p_{k+1}^{2}\,P\bigl(p_{k+1}^{2}\bigr)\;-\;p_{k}^{2}\,P\bigl(p_{k}^{2}\bigr), where p_{i} is the i th prime number. P(x)=\prod_{2<p<x}(1-2/p) excludes p=2 and we assume a fixed gap p_{k+1}-p_{k}=h\geq 2 . The resulting main term is proportional to p_{k}/(\ln p_{k})^{2} , with an explicit constant \widetilde{C}_{2}\approx 1.07 . We also discuss how residue–class counting in the interval [p_{k}^{2},p_{k+1}^{2}] may include both even and odd integers, and how one might reduce the tally if only odd or prime candidates are needed. Let p_{k} be the k th prime. Define P(x)\;=\;\prod_{\begin{subarray}{c}2<p<x\\ p\,\mathrm{prime}\end{subarray}}\Bigl(1-\tfrac{2}{p}\Bigr). Then for \Delta_{k}\;:=\;p_{k+1}^{2}\,P\bigl(p_{k+1}^{2}\bigr)\;-\;p_{k}^{2}\,P\bigl(p_{k}^{2}\bigr), we establish: Theorem 2 . Assume there are infinitely many k with p_{k+1}-p_{k}=h\geq 2 . Then for those k , \Delta_{k}\;=\;\frac{\widetilde{C}_{2}\,h}{2}\;\frac{p_{k}}{(\ln p_{k})^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln p_{k}}\bigr)\Bigr], where \widetilde{C}_{2}\approx 1.07 is the constant from P(x)\;=\;\widetilde{C}_{2}\,\frac{1}{(\ln x)^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln x}\bigr)\Bigr]. Numerically, \;\widetilde{C}_{2}=\exp\!\bigl(1.5-2(M_{1}+S)\bigr)\approx 1.07, where M_{1}\approx 0.2614972 is the Meissel–Mertens constant (including p=2 ) and S=\sum_{p\geq 2}1/p^{2}\approx 0.4522474. Evaluating \texorpdfstring p^{2}\,P(p^{2}) Setting x=p^{2} and noting that \ln(p^{2})=2\ln p , we get p^{2}\,P\bigl(p^{2}\bigr)\;=\;\widetilde{C}_{2}\,\frac{p^{2}}{4\,(\ln p)^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln p}\bigr)\Bigr]. Let \widetilde{R}(p)\;=\;\frac{\widetilde{C}_{2}}{4}\,\frac{p^{2}}{(\ln p)^{2}}. Hence p^{2}\,P\bigl(p^{2}\bigr)\;=\;\widetilde{R}(p)\Bigl[1+O\!\bigl(\tfrac{1}{\ln p}\bigr)\Bigr]. Telescoping Difference \texorpdfstring \Delta_{k} Δₖ We define \Delta_{k}\;=\;p_{k+1}^{2}\,P\bigl(p_{k+1}^{2}\bigr)\;-\;p_{k}^{2}\,P\bigl(p_{k}^{2}\bigr). Assuming p_{k+1}=p_{k}+h with fixed h\geq 2 , we have: p_{k+1}^{2}\,P\bigl(p_{k+1}^{2}\bigr)\;=\;\widetilde{R}\bigl(p_{k+1}\bigr)\Bigl(1+O(\tfrac{1}{\ln p_{k+1}})\Bigr),\quad p_{k}^{2}\,P\bigl(p_{k}^{2}\bigr)\;=\;\widetilde{R}\bigl(p_{k}\bigr)\Bigl(1+O(\tfrac{1}{\ln p_{k}})\Bigr). Since p_{k+1}\sim p_{k} , the O(\tfrac{1}{\ln p}) terms combine, yielding \Delta_{k}\;=\;\bigl[\widetilde{R}(p_{k+1})-\widetilde{R}(p_{k})\bigr]\Bigl(1+O(\tfrac{1}{\ln p_{k}})\Bigr). Hence it suffices to expand \widetilde{R}(p+h)-\widetilde{R}(p),\quad\text{where }\widetilde{R}(p)=\frac{\widetilde{C}_{2}}{4}\,\frac{p^{2}}{(\ln p)^{2}}. Discrete difference of \texorpdfstring \widetilde{R}(p) At first, we aim to prove, for fixed integer h\geq 1 and large p , \frac{(p+h)^{2}}{(\ln(p+h))^{2}}\;-\;\frac{p^{2}}{(\ln p)^{2}}\;=\;\frac{2hp}{(\ln p)^{2}}\;+\;O\!\Bigl(\tfrac{p}{(\ln p)^{3}}\Bigr). Expand (p+h)^{2} , (p+h)^{2}\;=\;p^{2}+2hp+h^{2}\;=\;p^{2}+2hp+O(1), since h is fixed, and h^{2}=O(1) as p\to\infty . Also, we have p+h=p\bigl(1+\tfrac{h}{p}\bigr) , so \ln(p+h)\;=\;\ln p\;+\;\ln\!\Bigl(1+\frac{h}{p}\Bigr). For large p , we use \ln(1+u)=u+O(u^{2}) . Here u=h/p , so \ln\!\Bigl(1+\tfrac{h}{p}\Bigr)\;=\;\frac{h}{p}\;+\;O\!\Bigl(\tfrac{1}{p^{2}}\Bigr). Thus \ln(p+h)\;=\;\ln p\;+\;\frac{h}{p}\;+\;O\!\bigl(\tfrac{1}{p^{2}}\bigr). Expand (\ln(p+h))^{2} and invert it. (\ln(p+h))^{2}\;=\;\Bigl(\ln p+\tfrac{h}{p}+O(\tfrac{1}{p^{2}})\Bigr)^{2}\;=\;(\ln p)^{2}\;+\;2\,\frac{h\,\ln p}{p}\;+\;O\!\bigl(\tfrac{1}{p}\bigr). Hence \frac{1}{(\ln(p+h))^{2}}\;=\;\frac{1}{(\ln p)^{2}}\Bigl[1+O\!\Bigl(\tfrac{1}{p\,\ln p}\Bigr)\Bigr]\;=\;\frac{1}{(\ln p)^{2}}\;+\;O\!\Bigl(\tfrac{1}{p(\ln p)^{3}}\Bigr). (One may view this as a standard binomial expansion: if (\ln(p+h))^{2}=(\ln p)^{2}+O(\tfrac{1}{p}) , then 1/(\ln(p+h))^{2}=1/(\ln p)^{2}+O(\tfrac{1}{p(\ln p)^{2}}) , which is O(\tfrac{1}{p(\ln p)^{3}}) to match notation.) We have (p+h)^{2}\;=\;p^{2}+2hp+O(1),\quad\frac{1}{(\ln(p+h))^{2}}\;=\;\frac{1}{(\ln p)^{2}}\;+\;O\!\Bigl(\tfrac{1}{p(\ln p)^{3}}\Bigr). Thus \frac{(p+h)^{2}}{(\ln(p+h))^{2}}\;=\;\Bigl[p^{2}+2hp+O(1)\Bigr]\Bigl[\frac{1}{(\ln p)^{2}}+O\!\bigl(\tfrac{1}{p(\ln p)^{3}}\bigr)\Bigr]. Distribute: • The product p^{2}\cdot\tfrac{1}{(\ln p)^{2}} is the leading \tfrac{p^{2}}{(\ln p)^{2}} . • The product (2hp)\cdot\tfrac{1}{(\ln p)^{2}} is \tfrac{2hp}{(\ln p)^{2}} . • Terms like p^{2}*O(\tfrac{1}{p(\ln p)^{3}})=O(\tfrac{p}{(\ln p)^{3}}) . • The O(1) factor times \tfrac{1}{(\ln p)^{2}} is O\bigl(\tfrac{1}{(\ln p)^{2}}\bigr) , which is smaller or comparable to O(\tfrac{p}{(\ln p)^{3}}) for large p . Hence \frac{(p+h)^{2}}{(\ln(p+h))^{2}}\;=\;\frac{p^{2}}{(\ln p)^{2}}\;+\;\frac{2hp}{(\ln p)^{2}}\;+\;O\!\Bigl(\tfrac{p}{(\ln p)^{3}}\Bigr). Therefore, \frac{(p+h)^{2}}{(\ln(p+h))^{2}}\;-\;\frac{p^{2}}{(\ln p)^{2}}\;=\;\frac{2hp}{(\ln p)^{2}}\;+\;O\!\Bigl(\tfrac{p}{(\ln p)^{3}}\Bigr). Multiplying by \tfrac{\widetilde{C}_{2}}{4} yields \widetilde{R}(p+h)-\widetilde{R}(p)\;=\;\frac{\widetilde{C}_{2}}{4}\Bigl[\tfrac{2hp}{(\ln p)^{2}}+O\!\bigl(\tfrac{p}{(\ln p)^{3}}\bigr)\Bigr]\;=\;\frac{\widetilde{C}_{2}\,h}{2}\,\frac{p}{(\ln p)^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln p}\bigr)\Bigr]. Substituting p=p_{k} and recalling \Delta_{k}=[\,\widetilde{R}(p_{k+1})-\widetilde{R}(p_{k})\,][\,1+O(1/\ln p_{k})\,] gives \Delta_{k}\;=\;\frac{\widetilde{C}_{2}\,h}{2}\,\frac{p_{k}}{(\ln p_{k})^{2}}\Bigl[1+O\!\bigl(\tfrac{1}{\ln p_{k}}\bigr)\Bigr]. Thus Theorem 2 is proved. ∎ Remarks: 1. Residue Classes in \texorpdfstring [\,p_{k}^{2},p_{k+1}^{2}) and Parity Considerations In many related applications, one counts how many integers in the interval [p_{k}^{2},\,p_{k+1}^{2}) (or (p_{k}^{2},p_{k+1}^{2}) ) satisfy certain residue–class constraints. For instance, if the density of solutions is \tfrac{2}{\pi} among all integers, one might expect about \frac{2}{\pi}\;\bigl[\,(p_{k+1}^{2}-p_{k}^{2})\bigr] solutions. However, those solutions could include both even and odd integers. If we only want to consider odd integers (such as primes >2 ), one typically divides that raw count by 2 , plus lower–order terms, because about half of all integers in large ranges are even. Hence, in prime–gap settings or other contexts where we impose p>2 , the integral count in certain residue classes might be halved to focus on odd solutions. This is a common bookkeeping tactic: the fraction \tfrac{2}{\pi} , or other fractions from residue distributions, often initially refers to all integers, but restricting to odd solutions imposes a further factor \tfrac{1}{2} (or more refined partition arguments). 2. Comparison to average gap. On average, p_{k+1}-p_{k}\sim\ln p_{k} , which would make \Delta_{k} on the order of p_{k}/\ln p_{k} . By contrast, a fixed h\geq 2 yields \Delta_{k} of order p_{k}/(\ln p_{k})^{2} . Thus the fixed–gap scenario is asymptotically smaller by a factor of \ln p_{k} . 3. Extensions to (1-c/p) . For real c>0 , the analogous

Read the complete text