Skip to content
#small language model Open access

Primitive Roots Break the Independent Estimate Only When the Square-Free Base Leaves 1 on Division by 4 ── for a = 5, being a quadratic non-residue and the remainder of p on division by 5 are tied together by reciprocity, and the share of primes with 5 as a primitive root becomes 1 + 1/19 times the independent product ── the separator is whether a leaves 1 or 3 on division by 4 ── [Paper 1020]

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)
Analytic Number Theory Research

Abstract

An integer a is a primitive root of a prime p when its powers produce every nonzero remainder on division by p. Whether infinitely many primes have a as a primitive root (Artin’s conjecture) is open for every a that is neither a square nor −1. This paper counts the share of primes for which a is a primitive root and lines it up against the estimate that multiplies a probability for each prime. No new theorem or law is claimed, and the conjecture is not proved. Scope of this paper (scope note): No new theorem or law is claimed──Artin’s conjecture (1927), the estimate of the share of primes with a as a primitive root and its correction, that the estimate holds under the generalized Riemann hypothesis (Hooley, 1967), quadratic reciprocity and Euler’s criterion are all known. How the estimate was found to fail in numerical computation and was corrected is set out in Moree’s survey (2012). The conjecture is not proved──counting up to 2×10⁶ proves nothing, and the generalized Riemann hypothesis is not entered. Negative a and a with square factors are not treated──of a with square factors, only the square a = 4 is looked at. Relation to earlier papers: Paper 513 checked quadratic reciprocity by counting lattice points──here it is counted that the same reciprocity ties two conditions together and makes the product of probabilities fail. Paper 603 applied Euler’s criterion (a to the power (p−1)/2 modulo p) to composites and counted the bases that are fooled──here the same criterion decides whether a is a non-residue, and the count is split by the remainder of p on division by a. Paper 640 counted that the number of ways to write an even number as a sum of two primes carries multipliers set by divisibility by small primes──here the multiplier is set by the prime factors of a itself and appears only when a leaves 1 on division by 4. What is added is counting the share for ten values of a and lining it up against the estimate, counting the source of the correction directly as conditional shares of non-residues, showing as a contrast that for a leaving 3 on division by 4 the conditions are not tied, and lining up how the size of the correction is set by the prime factors of a. First, and this is the core. The independent estimate fails only in cases like a = 5, and the correction fills the failure exactly──the product of probabilities, Artin’s constant, is 0.373955826; counting over the 148932 primes up to 2×10⁶, a = 2, 3, 6 and 7 gave 0.374292, 0.373623, 0.373958 and 0.374422. Only a = 5 gave 0.393578, differing from a = 2 by 10.82 standard errors and matching the corrected estimate 0.393638 (1 + 1/19 times the constant) (Section 2). Second, the failure comes from two conditions not being independent──among primes leaving 1 on division by 5, the share for which 5 is a quadratic non-residue was 0.000000, and among the others 0.666914. If independent, both would be one half (Section 3). Third, for a leaving 3 on division by 4, the two conditions are not tied──for a = 3 the two shares were 0.500349 and 0.499685, and for a = 7 0.500807 and 0.500415, both one half. For a = 13, which leaves 1, they were 0 and 0.545794, matching the share of non-residue remainders among remainders other than 1, 6/11 = 0.545455 (Section 3). Fourth, the size of the correction is set by the prime factors of a and shrinks for large factors──for a = 13 it is 1 + 1/155, with estimate 0.376368 against 0.376611, and for a = 21 it is 1 − 1/205, with 0.372132 against 0.372873; in this range they cannot be told from the uncorrected constant (Section 2). Fifth, a square a is never a primitive root──the share for a = 4 was 0, because a square is a quadratic residue for every p (Section 2). Sixth, the separator is whether a leaves 1 or 3 on division by 4──quadratic reciprocity, only for a leaving 1 on division by 4, fixes whether a is a non-residue by the remainder of p on division by a. There the two conditions become tied and the product of probabilities fails (Section 4). Primitive roots break the independent estimate only when the square-free base leaves 1 on division by 4. Counting the primes up to 2×10⁶, a = 2, 3, 6 and 7 matched the constant 0.373955826, and only a = 5, at 0.393578, lay 10.82 standard errors away, matching the corrected estimate 0.393638──among primes leaving 1 on division by 5, the share for which 5 is a non-residue is 0, and among the others 0.666914, so the two conditions are not independent. For a = 3 and 7 the share is one half whatever the remainder, and they are not tied. a = 13 has the same form (0 and 0.545794), but its correction, 1 + 1/155, is small and cannot be told apart in this range. The separator is whether a leaves 1 or 3 on division by 4. Placed among the earlier papers──Paper 513 counted reciprocity by lattice points, Paper 603 counted the bases fooled by Euler’s criterion, and Paper 640 counted multipliers set by small primes. Here it was counted that reciprocity ties two conditions together and makes the product of probabilities fail. To be honest──whether infinitely many primes have a as a primitive root has not been proved for a single a. On the making of this work: The ideas and content of this work stem from the author's own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and checking the algebra. Any remaining errors or misinterpretations are solely the author's. Feedback and corrections are sincerely appreciated. Keywords: primitive root, Artin's conjecture, Artin's constant, quadratic reciprocity, Euler's criterion, quadratic non-residue. ----- 整数 a の累乗が p で割った余りをすべて作るとき、a は素数 p の原始根である。 a が原始根になる素数が無数にあるか(アルティン予想)は、平方数でも −1 でもない a について未解決である。本稿は、原始根になる素数の割合を数え、素数ごとの確率を掛けた見立てと並べる。新しい定理も法則も主張せず、予想を証明もしない。 本稿の射程(射程注記):新しい定理も法則も主張しない──アルティン予想(1927)、原始根になる割合の見立てとその補正、一般化リーマン予想を仮定すれば見立てが正しいこと(フーリー、1967)、平方剰余の相互法則、オイラーの規準は、いずれも既知である。見立てが数値計算で外れて補正された経緯は、モリーの概説(2012)にまとめられている。予想を証明しない──2×10⁶ まで数えたことは証明にならず、一般化リーマン予想にも立ち入らない。負の a と平方因子のある a の補正を扱わない──平方因子のある a は、平方数の a=4 だけを見る。既刊との関係:論文513 は、平方剰余の相互法則を格子点を数えて確かめた──ここでは、同じ相互法則が二つの条件を結びつけ、確率の積を外すことを数える。論文603 は、オイラーの規準(a の (p−1)/2 乗を p で割った余り)を合成数に当てて、騙される底を数えた──ここでは、同じ規準で平方非剰余かどうかを決め、p を a で割った余りごとに数える。論文640 は、偶数を二つの素数の和で表す仕方の数が、小さな素数で割り切れるかで決まる倍率を持つことを数えた──ここでは、倍率が a 自身の素因数で決まり、a を 4 で割った余りが 1 のときだけ現れることを数える。加えたのは、十個の a で割合を数えて見立てと並べたこと、補正の出どころを、条件つきの平方非剰余の割合で直接数えたこと、4 で割って 3 余る a では条件が結びつかないことを対照として示したこと、補正の大きさが a の素因数で決まることを並べたことである。 第一に、これが本稿の芯である。独立の見立てが外れるのは a=5 のような場合だけで、その外れは補正でちょうど埋まる──素数ごとの確率を掛けたアルティン定数は 0.373955826 で、2×10⁶ までの素数 148932 個で数えると a=2・3・6・7 は 0.374292・0.373623・0.373958・0.374422 だった。 a=5 だけは 0.393578 で、a=2 との差は標準誤差の 10.82 倍あり、補正した見立て 0.393638(定数の 1+1/19 倍)と合った(第2節)。 第二に、外れの出どころは、二つの条件が独立でないことである──p を 5 で割って 1 余る素数では、5 が平方非剰余になる割合は 0.000000 で、それ以外の素数では 0.666914 だった。独立なら、どちらも半分のはずである(第3節)。 第三に、a を 4 で割って 3 余る a では、二つの条件は結びつかない──a=3 では二つの割合が 0.500349 と 0.499685、a=7 では 0.500807 と 0.500415 で、どちらも半分だった。4 で割って 1 余る a=13 では 0 と 0.545794 で、1 以外の余りのうち非剰余の余りの割合 6/11=0.545455 に合った(第3節)。 第四に、補正の大きさは a の素因数で決まり、大きな素因数では小さくなる──a=13 では 1+1/155 倍で見立て 0.376368 に対して 0.376611、a=21 では 1−1/205 倍で 0.372132 に対して 0.372873 で、この範囲の数では補正の無い定数と見分けられない(第2節)。 第五に、平方数の a は原始根にならない──a=4 の割合は 0 だった。平方数はどの p でも平方剰余だからである(第2節)。 第六に、分離子は、a を 4 で割った余りが 1 か 3 か──平方剰余の相互法則は、4 で割って 1 余る a についてだけ、a が平方非剰余であることを p を a で割った余りで決めてしまう。そこで二つの条件が結びつき、確率の積が外れる(第4節)。 平方因子の無い a が原始根になる素数の割合は、a を 4 で割って 1 余るときだけ独立の見立てから外れる。2×10⁶ までの素数で数えると、a=2・3・6・7 は定数 0.373955826 と合い、a=5 だけが 0.393578 と標準誤差の 10.82 倍離れて、補正した見立て 0.393638 と合った──p を 5 で割って 1 余る素数では 5 が平方非剰余になる割合が 0、それ以外では 0.666914 で、二つの条件が独立でない。 a=3・7 では余りによらず半分で、結びつかない。 a=13 も同じ形(0 と 0.545794)だが、補正は 1+1/155 倍と小さく、この範囲では見分けられない。分離子は、a を 4 で割った余りが 1 か 3 かである。既刊との位置──論文513 は相互法則を格子点で数え、論文603 はオイラーの規準で騙される底を数え、論文640 は小さな素数で決まる倍率を数えた。ここでは、相互法則が二つの条件を結びつけ、確率の積を外すことを数えた。正直に言えば──a が原始根になる素数が無数にあるかどうかは、一つの a についても証明していない。 作成にあたって:本稿の着想と内容は、著者自身の考察に基づくものです。文章の構成整理や英訳、数式の確認には AI(大規模言語モデル)の助力を得ました。最終的な内容の解釈や誤りがあれば、それらはすべて著者の責に帰します。お気づきの点があれば、ご教示いただければ幸いです。 キーワード:原始根、アルティン予想、アルティン定数、平方剰余の相互法則、オイラーの規準、平方非剰余。

View source

Similar papers

#small language model Dataset Open access Oct 2026

Socratic guiding questions in synthetic arithmetic data: matched LoRA runs (revision v2)

Supporting data, adapters, predictions and code for the article *Low-Cost LoRA Fine-Tuning of Small Language Models for Multi-Step Arithmetic Reasoning* by Jake O'Grady, Asena Isik Gürhan, Chee Fong Ting and Effirul Ramlan (University of Galway). We generated 20,000 GSM8K-derived arithmetic problems with step-by-step s...

O'Grady, Jake, Gürhan, Asena Isik, Chee, Fong Ting et al. · 465 citations
#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6

Related blog posts

Microsoft Research Blog Oct 12, 2026

Simulating student learning for better AI tutoring

StudentSim is a research system for evaluating and improving AI tutoring tools before they are tested with learners. It models how individuals may engage with tutors, giving feedback to improve tutor performance across subjects like chess, writing, and mathematics. The post Simulating student learning for better AI tutoring appeared first on Microsoft Research.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.