Most subrings of ℤ n \mathbb{Z}^{n} have large corank

Kelly Isham, Nathan Kaplan · 2024

Read Most subrings of ℤ n \mathbb{Z}^{n} have large corank by Kelly Isham, Nathan Kaplan (2024) 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-2412-18692.txt.

Introduction

The authors study quotients of integer lattices by subrings, in this 2024 paper. A sublattice quotient forms a finite group whose shape has known statistical habits. Adding ring structure changes the story.

When the sublattice is also a subring and the dimension reaches seven, the quotient is very rarely a simple cyclic group. In large dimensions its rank is typically big instead. Proofs mix analytic number theory with counting arguments, via special zeta functions and matrix forms whose columns span a subring. The head byline names Gautam Chinta alongside the catalog authors. The team frame the result as a sharp break from lattice habits.

Catalog record

CREATORKelly Isham, Nathan Kaplan
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE77 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-18692

Browse the full library · About the Tower of Babel

Full text

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

Most subrings of ℤ n \mathbb{Z}^{n} have large corank Kelly Isham, Nathan Kaplan Abstract . If Λ ⊆ ℤ n \Lambda\subseteq\mathbb{Z}^{n} is a sublattice of index m m , then ℤ n / Λ \mathbb{Z}^{n}/\Lambda is a finite abelian group of order m m and rank at most n n . Several authors have studied statistical properties of these groups as we range over all sublattices of index at most X X . In this paper we investigate quotients by sublattices that have additional algebraic structure. While quotients ℤ n / Λ \mathbb{Z}^{n}/\Lambda follow the Cohen-Lenstra heuristics and are very often cyclic, we show that if Λ \Lambda is actually a subring, then once n ≥ 7 n\geq 7 these quotients are very rarely cyclic. More generally, we show that once n n is large enough the quotient typically has very large rank. In order to prove our main theorems, we combine inputs from analytic number theory and combinatorics. We study certain zeta functions associated to ℤ n \mathbb{Z}^{n} and also prove several results about matrices in Hermite normal form whose columns span a subring of ℤ n \mathbb{Z}^{n} . (With an Appendix by Gautam Chinta) 1. Introduction The focus of this article is about a family of random finite abelian groups that arise in number theory. There has recently been extensive interest in this subject; see for example the survey of Wood [ 22 ] . We show that quotients of \mathbb{Z}^{n} by random subrings do not look like quotients of \mathbb{Z}^{n} by random sublattices. This has an interpretation in terms of the distribution of cokernels of families of random integer matrices and the Cohen-Lenstra heuristics, a research topic that has been quite active in recent years. In order to describe our results, we introduce some notation. A sublattice \Lambda\subseteq\mathbb{Z}^{n} is a finite index subgroup, that is, we reserve this term for full-rank sublattices of \mathbb{Z}^{n} . For vectors u=(u_{1},\ldots,u_{n}) and w=(w_{1},\ldots,w_{n}) in \mathbb{Z}^{n} , we write u\circ w for the vector given by the componentwise product, u\circ w=(u_{1}w_{1},\ldots,u_{n}w_{n}) . A sublattice \Lambda\subseteq\mathbb{Z}^{n} is multiplicatively closed if u,w\in\Lambda implies u\circ w\in\Lambda . A multiplicatively closed sublattice is a subring if it also contains the multiplicative identity (1,1,\ldots,1) . In this article, we address questions of the following type. Question 1.1 . Let X be a positive real number and R\subseteq\mathbb{Z}^{n} be a subring of index at most X chosen uniformly at random. What does the quotient \mathbb{Z}^{n}/R ‘look like’? For example, as X\rightarrow\infty how often is this group cyclic? A main point of this article is to show that the additional algebraic structure possessed by subrings leads these quotients to very often have large rank. In order to set up the contrast with quotients of random sublattices, we highlight some results about \mathbb{Z}^{n}/\Lambda where \Lambda\subseteq\mathbb{Z}^{n} is a sublattice of index at most X chosen uniformly at random. A finite abelian group G can be written uniquely in terms of its invariant factors, G\cong\mathbb{Z}/\alpha_{1}\mathbb{Z}\oplus\mathbb{Z}/\alpha_{2}\mathbb{Z}\oplus\cdots\oplus\mathbb{Z}/\alpha_{n}\mathbb{Z}, where \alpha_{i+1}\mid\alpha_{i} for 1\leq i\leq n-1 . The rank of G is the largest i for which \alpha_{i}>1 . The corank of a sublattice \Lambda\subseteq\mathbb{Z}^{n} is the rank of \mathbb{Z}^{n}/\Lambda . A sublattice \Lambda\subseteq\mathbb{Z}^{n} is cocyclic if either \Lambda=\mathbb{Z}^{n} or \Lambda has corank 1 . Nguyen and Shparlinski compute the proportion of sublattices of \mathbb{Z}^{n} that are cocyclic [ 17 ] , a result which also follows by a different method from earlier work of Petrogradsky [ 18 ] . Chinta, Kaplan, and Koplewitz generalize this result [ 4 ] , determining the proportion of sublattices of \mathbb{Z}^{n} that have corank at most k for any k\in[1,n] . Throughout this paper we use p to denote a prime number and write \prod_{p} for a product over all primes. Theorem 1.2 . [ 4 , Corollary 1.2] Let n,k be positive integers satisfying 1\leq k\leq n . Then, \displaystyle p_{n,k} \displaystyle:= \displaystyle\lim_{X\rightarrow\infty}\frac{\#\left\{\text{Sublattices }\Lambda\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon\Lambda]<X\text{ and }\Lambda\text{ has corank at most }k\right\}}{\#\left\{\text{Sublattices }\Lambda\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon\Lambda]<X\right\}} \displaystyle= \displaystyle\prod_{p}\left(\prod_{j=1}^{n}(1-p^{-j})^{2}\cdot\sum_{i=0}^{k}\frac{1}{p^{i^{2}}\prod_{j=1}^{i}(1-p^{-j})^{2}\prod_{j=1}^{n-i}(1-p^{-j})}\right). These probabilities arise in the famous Cohen–Lenstra heuristics, which were developed during the study of statistical questions about Sylow p -subgroups of class groups of families of number fields [ 5 ] . Let P_{n} be the distribution on finite abelian p -groups of rank at most n that chooses a finite abelian p -group G of rank r\leq n with probability (1) P_{n}(G)=\frac{1}{\#\Aut(G)}\left(\prod_{i=1}^{n}(1-p^{-i})\right)\left(\prod_{i=n-r+1}^{n}(1-p^{-i})\right). Theorem 1.2 states that the proportion of \Lambda\subseteq\mathbb{Z}^{n} that have corank at most k is equal to the product over all primes of the probability that a finite abelian p -group chosen from the distribution P_{n} has rank at most k . These connections are explained in [ 4 ] . Even for small values of k , when n goes to infinity these probabilities are quite large. For example, for large n the proportion of cocyclic sublattices \Lambda\subseteq\mathbb{Z}^{n} is approximately 85\% , the proportion with corank at most 2 is approximately 99.4\% , and the proportion with corank at most 3 is approximately 99.995\% . The main theorem of this paper is that for large n , while the vast majority of random sublattices of \mathbb{Z}^{n} have small corank, it is extremely rare for a random subring of \mathbb{Z}^{n} to have small corank. Theorem 1.3 . Let n and k be positive integers satisfying 1\leq k<n . Define p^{R}_{n,k}=\lim_{X\rightarrow\infty}\frac{\#\left\{\text{Subrings }R\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon R]<X\text{ and }R\text{ has corank at most }k\right\}}{\#\left\{\text{Subrings }R\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon R]<X\right\}}. (1) If k\in\{1,2,3\} and n\geq 6 , then p^{R}_{n,k}=0 . (2) If n\geq 7 , then p^{R}_{n,4}=0 . (3) If k\leq(6-4\sqrt{2})n+2\sqrt{2}-\frac{8}{3} and n\geq 7 , then p^{R}_{n,k}=0 . For example, while approximately 85\% of sublattices of \mathbb{Z}^{7} are cocyclic, the proportion of subrings of \mathbb{Z}^{7} that are cocyclic is 0 . We prove Theorem 1.3 by studying the asymptotic rate of growth of the function H_{n,k}(X)=\#\left\{\text{Subrings }R\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon R]<X\text{ and }R\text{ has corank at most }k\right\} and comparing it to the asymptotic growth rate of the function N_{n}(X)=\#\left\{\text{Subrings }R\subseteq\mathbb{Z}^{n}\colon[\mathbb{Z}^{n}\colon R]<X\right\}. We use techniques from the theory of zeta functions of rings to prove upper bounds for the growth rate of H_{n,k}(X) and compare these results to lower bounds of Isham for N_{n}(X) [ 10 , Theorem 1.6] . A key part of our argument involves counting special classes of matrices in Hermite normal form. Every finite index sublattice \Lambda\subseteq\mathbb{Z}^{n} is the column span of a unique n\times n matrix H(\Lambda) in Hermite normal form. Moreover, \mathbb{Z}^{n}/\Lambda is isomorphic to the cokernel of this matrix. We discuss matrices in Hermite normal form and their cokernels at the start of Section 2 and in Proposition 4.2 . Theorem 1.2 says that the probability \mathbb{Z}^{n}/\Lambda has rank at most k is what one would expect if these random groups were distributed according to the Cohen-Lenstra heuristics. In [ 4 ] , the authors also show that Sylow p -subgroups of \mathbb{Z}^{n}/\Lambda are distributed according to the distribution P_{n}(G) from ( 1 ). These results can be interpreted by saying that cokernels of random integer matrices in Hermite normal form follow the Cohen-Lenstra heuristics. There is a general philosophy that cokernels of families of random integer and p -adic matrices should be distributed according to the Cohen-Lenstra heuristics, except when there is additional algebraic structure that must be taken into account; see for example

Read the complete text