Consecutive pure fields of the form ℚ ⁡ ( a l ) \mathbb{Q}\left(\sqrt[l]{a}\right) with large class numbers

Jishu Das, Srilakshmi Krishnamoorthy · 2025

Read Consecutive pure fields of the form ℚ ⁡ ( a l ) \mathbb{Q}\left(\sqrt[l]{a}\right) with large class numbers by Jishu Das, Srilakshmi Krishnamoorthy (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-2504-03904.txt.

Introduction

Jishu Das and Srilakshmi Krishnamoorthy study pure number fields with large class numbers in a two thousand twenty-five number theory paper. A pure field here is formed by adjoining an l-th root of an integer to the rationals.

Fix a prime l at least three and any positive integer k. The authors show that runs of k consecutive pure fields with arbitrarily large class numbers are common. Counting such starting points up to a bound x gives at least roughly x to the power one over l of them. The proof is conditional: it assumes a conjecture of Langlands and an upper bound on the regulators of these fields, which the authors state openly.

Catalog record

CREATORJishu Das, Srilakshmi Krishnamoorthy
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-03904

Browse the full library · About the Tower of Babel

Full text

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

Consecutive pure fields of the form ℚ ⁡ ( a l ) \mathbb{Q}\left(\sqrt[l]{a}\right) with large class numbers Jishu Das, Srilakshmi Krishnamoorthy Abstract . Let l l be a rational prime greater than or equal to 3 3 and k k be a given positive integer. Under a conjecture due to Langlands and an assumption on upper bound for the regulator of fields of the form ℚ ⁡ ( a l ) \mathbb{Q}\left(\sqrt[l]{a}\right) , we prove that there are atleast x 1 / l − o ⁡ ( 1 ) x^{1/l-o(1)} integers 1 ≤ d ≤ x 1\leq d\leq x such that the consecutive pure fields of the form ℚ ⁡ ( d + 1 l ) , … , ℚ ⁡ ( d + k l ) \mathbb{Q}\left(\sqrt[l]{d+1}\right),\dots,\mathbb{Q}\left(\sqrt[l]{d+k}\right) have arbitrary large class numbers. 2020 Mathematics Subject Classification Primary:11R29, Secondary: 11R21 1. Introduction The class number of an algebraic number field plays a vital role in number theory. An important problem concerning class number of a number field is to have an understanding of the size of the class number. Let h_{K} denote the class number of the number field K. Let d be the fundamental discriminant of the number field \mathbb{Q}(\sqrt{d}) . Assuming the Generalized Riemann Hypothesis (GRH), Littlewood [ 12 ] proves (1.1) h_{\mathbb{Q}(\sqrt{-d})}\leq\left(\frac{2e^{\gamma}}{\pi}+\text{o}(1)\right)\sqrt{|d|}\log\log(|d|), where \gamma denotes the Euler-Mascheroni constant. Under the same hypothesis, Littlewood shows the existence of infinitely many fundamental discriminants d such that h_{\mathbb{Q}(\sqrt{-d})}\geq\left(\frac{2e^{\gamma}}{\pi}+\text{o}(1)\right)\sqrt{|d|}\log\log(|d|). For positive discriminants, one can show that the bound (1.2) h_{\mathbb{Q}(\sqrt{d})}\leq\left(4e^{\gamma}+\text{o}(1)\right)\sqrt{d}\frac{\log\log(d)}{\log d}, holds under GRH. Montgomery and Weinberger [ 13 ] show that this bound cannot be improved for real quadratic fields, apart from the value of the constant and an unconditional result without GRH. They prove that there exist infinitely many real quadratic fields \mathbb{Q}(\sqrt{d}) such that (1.3) h_{\mathbb{Q}(\sqrt{d})}\gg\sqrt{d}\frac{\log\log(d)}{\log d}. Lamzouri [ 11 , Theorem 1.2] improved the result by Montgomery and Weinberger by showing the following. Let x be large. There are at least x^{\frac{1}{2}-\frac{1}{\log\log x}} and at most x^{\frac{1}{2}+\text{o}(1)} real quadratic fields \mathbb{Q}(\sqrt{d}) with discriminant d\leq x such that (1.4) h_{\mathbb{Q}(\sqrt{d})}\geq\left(2e^{\gamma}+\text{o}(1)\right)\sqrt{d}\frac{\log\log(d)}{\log d}. Assuming GRH and Artin’s conjecture for Artin L-functions, Duke [ 6 , Theorem 1] proves an analogue lower bound like equation ( 1.3 ) for more general number fields. To be precise, Duke considers the set \mathcal{K}_{n} consisting of totally real number fields of degree n whose Galois closures have S_{n} as their Galois group. [ 6 , Theorem 1] states that there is a constant c>0 depending only on n such that there exist K\in\mathcal{K}_{n} with arbitrarily large discriminant d for which h_{K}>c\sqrt{d}\left(\frac{\log\log d}{\log d}\right)^{n-1}. Daileda [ 5 , Theorem 1] proves an unconditional version of Duke’s result for \mathcal{K}_{3}. He further [ 5 , Theorem 3] shows that there is an absolute constant c>0 such that h_{\mathbb{Q}(\sqrt[3]{d})}\geq c\sqrt{|d|}\frac{\log\log|d|}{\log|d|}. In 2023, Cherubini, Fazzari, Granville, Kala and Yatsyna [ 4 ] prove that for a given positive integer k, there are at least x^{1/2-\text{o}(1)} integers 1\leq d\leq x such that for all j=1,\dots,k, h_{\mathbb{Q}(\sqrt{d+j})}\gg_{k}\frac{\sqrt{d}}{\log d}\log\log d. To generalize the above, recently Byeon and Yhee [ 3 ] show the following: Given a positive integer k, there are at least x^{1/3-\text{o}(1)} integers 1\leq d\leq x such that the consecutive pure cubic fields \mathbb{Q}(\sqrt[3]{d+1}) , …, \mathbb{Q}(\sqrt[3]{d+k}) have arbitrary large class number. To be precise, we have the following theorem. Theorem 1 ( [ 3 , Theorem 1.1] ) . Let k be a fixed positive integer. There are atleast x^{1/3-\text{o}(1)} integers 1\leq d\leq x such that the class number h_{\mathbb{Q}(\sqrt[3]{d+j})} of the pure cubic field \mathbb{Q}(\sqrt[3]{d+j}) satisfy h_{\mathbb{Q}(\sqrt[3]{d+j})}\gg_{k}\frac{\sqrt{d}}{\log d}\log\log d, for all j=1,\dots,k. Let l be a given prime throughout the article. We wish to answer the following in this article: Given a positive integer k and an prime l>3, do there exist at least x^{1/l-\text{o}(1)} integers 1\leq d\leq x such that the pure fields \mathbb{Q}(\sqrt[l]{d+1}) , \mathbb{Q}(\sqrt[l]{d+2}) , …, \mathbb{Q}(\sqrt[l]{d+k}) have arbitrary large class numbers? We start with a hypothesis that will be useful. Hypothesis 1 . Let a=n^{l}+r where n,r are positive integers and r|ln^{l} with l>3. Let R_{\mathbb{Q}\left(\sqrt[l]{a}\right)} and D_{\mathbb{Q}\left(\sqrt[l]{a}\right)} denote the regulator and the absolute value of the discriminant of the field \mathbb{Q}(\sqrt[l]{a}). Then R_{\mathbb{Q}\left(\sqrt[l]{a}\right)}=\text{o}\left(\sqrt{D_{\mathbb{Q}\left(\sqrt[l]{a}\right)}}\log\log D_{\mathbb{Q}\left(\sqrt[l]{a}\right)}\right). In Section 2.1 , we briefly discuss Hypothesis 1 with a heuristic argument (see Section 2.1 and Remark 5 ), along with some computational evidence in Section 2.2 . Let \pi be a finite dimensional complex representation of \text{Gal}(\mathbb{Q}(\omega,\sqrt[l]{b})/\mathbb{Q}), where \omega is a primitive l -th root of unity and b is a l -th powerfree positive integer. By Langlands conjecture ( [ 2 , Conjecture 1.8.1] ), there is an automorphic representation \tilde{\pi}_{D} with conductor D such that L(s,\mathbb{Q}(\omega,\sqrt[l]{b})/\mathbb{Q},\pi)=L(s,\mathbb{Q}(\omega,\sqrt[l]{b})/\mathbb{Q},\tilde{\pi}_{D}), where D denotes the absolute value of the discriminant of \mathbb{Q}(\sqrt[l]{b}), and L(s,\mathbb{Q}(\omega,\sqrt[l]{b})/\mathbb{Q},\pi) , L(s,\mathbb{Q}(\omega,\sqrt[l]{b})/\mathbb{Q},\tilde{\pi}_{D}) denote the Artin L-function associated to \pi and \tilde{\pi}_{D} respectively. The following holds under the Langlands conjecture( [ 2 , Conjecture 1.8.1] ). Theorem 2 . Let k be a fixed positive integer and l be a rational prime greater than or equal to 3 . Let D_{j} be the absolute value of the discriminant of the pure field \mathbb{Q}\left(\sqrt[l]{\Delta(m)+j}\right) (see Section 3 for \Delta(m) ). Let \omega be a primitive l -th root of unity and 0<\epsilon<1 . Suppose that \sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv\,1(\text{mod}\,\,l)\\ \left(\frac{D_{j}}{\pi}\right)_{l}=1\end{subarray}}\frac{l-1}{p}+\sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv 1\,(\text{mod}\,\,l)\\ \left(\frac{D_{j}}{\pi}\right)_{l}=1\end{subarray}}^{\prime}\ \frac{\lambda(p)}{p}\gg_{k}\log\log\log x, where \lambda(p) corresponds to the Dirichlet series given by ( 2.7 ) for the Artin L-function L\left(s,\mathbb{Q}\left(\omega,\sqrt[l]{\Delta(m)+j}\right)/\mathbb{Q},\tilde{\pi}_{d_{\mathbb{Q}\left(\sqrt[l]{\Delta(m)+j}\right)}}\right), and \sum^{\prime} denotes the summation over p\leq(\log D_{j})^{\epsilon} such that any one of the conditions given below p\leq(\log D_{j})^{\epsilon} does not get satisfied. Suppose \mathbb{Q}\left(\sqrt[l]{\Delta(m)+j}\right) satisfy Hypothesis 1 for all j=1,\dots,k. Then there are at least x^{1/l-\text{o}(1)} integers 1\leq d\leq x such that the pure fields \mathbb{Q}(\sqrt[l]{d+1}) , …, \mathbb{Q}(\sqrt[l]{d+k}) have arbitrary large class numbers. Remark 1 . The hypothesis \sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv\,1(\text{mod}\,\,l)\\ \left(\frac{D_{j}}{\pi}\right)_{l}=1\end{subarray}}\frac{l-1}{p}+\sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv 1\,(\text{mod}\,\,l)\\ \left(\frac{D_{j}}{\pi}\right)_{l}=1\end{subarray}}^{\prime}\ \frac{\lambda(p)}{p}\gg_{k}\log\log\log x is not vacuous. For instance, when l=3, \lambda(p)=0 for p\equiv 2\,(\text{mod}\,\,3) (see [ 3 , Section 2] and proof of [ 5 , Lemma 9] ). This makes \sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv\,1(\text{mod}\,\,3)\\ \left(\frac{D_{j}}{\pi}\right)_{3}=1\end{subarray}}\frac{2}{p}+\sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv 1\,(\text{mod}\,\,3)\\ \left(\frac{D_{j}}{\pi}\right)_{3}=1\end{subarray}}^{\prime}\ \frac{\lambda(p)}{p}=\sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv\,1(\text{mod}\,\,3)\\ \left(\frac{D_{j}}{\pi}\right)_{3}=1\end{subarray}}\frac{2}{p}\gg_{k}\log\log\log x. Let us consider l=5 . We have \lambda(p)=\sum_{i=1}^{4}\alpha_{i}(p) (see equation ( 2.4 )), where \alpha_{i}(p) are 20 -th roots of unity (as |\text{Gal}(\mathbb{Q}(e^{\frac{2\pi i}{5}},\sqrt[5]{b})/\mathbb{Q})|=20 ) or 0. Hence \lambda(p)=-4c_{p} , where c_{p} is very close to 1 is not possible. Based on this argument, we suppose that \sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv\,1(\text{mod}\,\,5)\\ \left(\frac{D_{j}}{\pi}\right)_{5}=1\end{subarray}}\frac{4}{p}+\sum_{\begin{subarray}{c}p\leq(\log D_{j})^{\epsilon},\\ p\equiv 1\,(\text{mod}\,\,5)\\ \left(\frac{D_{j}}{\pi}\right)_{5}=1\end{subarray}}^{\prime}\ \frac{\lambda(p)}{p}\gg_{k}\log\log\log x. The structure of the paper is as follows. In Section 2 , we prove Proposition 2.1 which deals with an approximation of L(1,\tilde{\pi}_{d_{K}}) (see equation ( 2.7 )). In Section 3 , we prove two lemmas leading to the main theorem’s proof. The proof of Theorem 2 is covered in Section 4 . 2. A Proposition Let F be a number field and E be a degree n extension of F with Galois closure \hat{E} . Let G=\text{Gal}(\hat{E}/F) and \pi be a finite dimensional complex representation of G. For a prime ideal \mathfrak{p} in F , let I_{\mathfrak{p}} and \sigma_{\mathfrak{p}} denote the inertia group and Frobenius element for an ideal \mathfrak{P} of \hat{E} lying over \mathfrak{p}. Let the space corresponding to \pi is V and V^{I_{\mathfrak{p}}} denote the subspace of V fixed by I_{\mathfrak{p}}. We define L_{\mathfrak{p}}(s,\hat{E}/F,\pi)=\det\left(I-\pi(\sigma_{\mathfrak{p}})|_{V^{I_{\mathfrak{p}}}}N(\mathfrak{p})^{-s}\right)^{-1}.

Read the complete text