A proof of purely singular splitting conjecture

Ka Hin Leung, Tao Zhang · 2026

Read A proof of purely singular splitting conjecture by Ka Hin Leung, Tao Zhang (2026) 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-2605-09871.txt.

Introduction

Leung and Zhang proved the purely singular splitting conjecture, posed by Woldar in 1995. A set M of nonzero integers splits a finite abelian group G if some subset S satisfies M times S equals G without zero. The splitting is purely singular when every prime divisor of the group's order divides some element of M. The conjecture concerned splittings by the set {1, 2, ..., k}.

The paper proves the finite abelian groups admitting such a splitting are precisely the cyclic groups of orders 1, k+1, and 2k+1. The problem connects to tilings of space by cube clusters and to lattice tilings, where translation vectors form an additive subgroup. Settling a thirty-year-old conjecture, the proof pins down exactly which groups the initial segment can split purely singularly.

Catalog record

CREATORKa Hin Leung, Tao Zhang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE17 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.09871

Browse the full library · About the Tower of Babel

Full text

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

A proof of purely singular splitting conjecture Ka Hin Leung, Tao Zhang Abstract A set M M of nonzero integers is said to split a finite abelian group G G if there exists a subset S ⊆ G S\subseteq G such that M ⋅ S = G ∖ { 0 } M\cdot S=G\setminus\{0\} . Such a splitting is called purely singular if every prime divisor of | G | |G| divides some element of M M . In 1995, Woldar [ 32 ] conjectured that the finite abelian groups admitting a purely singular splitting by the set { 1 , 2 , … , k } \{1,2,\dots,k\} are precisely the cyclic groups of orders 1 1 , k + 1 k+1 , and 2 ​ k + 1 2k+1 . In this paper, we prove this conjecture. Keywords : Purely singular splitting, lattice tiling, semi-cross. AMS subject classifications : 52C22, 05A18, 11H71, 20K01. † † footnotetext: E-mail addresses: [email protected] (K. Leung), [email protected] (T. Zhang). 1 Introduction Problems involving tilings of \mathbb{R}^{n} by clusters of cubes have a long and rich history, dating back at least to the early work of Minkowski [ 20 ] . In order to formalize this setting, let Q=\{(x_{1},\dots,x_{n}):0\leq x_{i}<1,\ x_{i}\in\mathbb{R}\} denote the unit cube, which we regard as being placed at the origin. Given a vector e\in\mathbb{R}^{n} , a unit cube placed at e is defined as the translate e+Q=\{e+x:x\in Q\}. Building on this notion, a cluster of cubes is defined as a finite (or, more generally, discrete) union of pairwise disjoint translates of Q , namely \mathcal{C}=\mathcal{B}+Q=\{e+Q:e\in\mathcal{E}\}, where \mathcal{B}\subset\mathbb{R}^{n} . With this terminology in place, we can now describe packings and tilings. A set of pairwise disjoint translates of \mathcal{C} is called a packing of \mathbb{R}^{n} by \mathcal{C} . If, in addition, these translates cover the entire space \mathbb{R}^{n} , then the packing is said to be a tiling. Furthermore, when the set of translation vectors forms an additive subgroup of \mathbb{Z}^{n} , the tiling is called a lattice tiling. This algebraic structure allows for a useful discrete reformulation: if \mathcal{B}\subset\mathbb{Z}^{n} , then a tiling of \mathbb{R}^{n} by \mathcal{C} is equivalent to a tiling of \mathbb{Z}^{n} by \mathcal{B} . Motivated by applications to error-correcting codes for flash memories, Wei and Schwartz [ 31 ] introduced the notion of a limited-magnitude error ball. For integers n\geq t\geq 1 and k_{+}\geq k_{-}\geq 0 , the (n,t,k_{+},k_{-}) -error ball is defined as \displaystyle\mathcal{B}(n,t,k_{+},k_{-}):=\{{\bf a}=(a_{1},a_{2},\dots,a_{n})\in\mathbb{Z}^{n}:\ a_{i}\in[-k_{-},k_{+}],\ \mathrm{wt}({\bf a})\leq t\}, where \mathrm{wt}({\bf a}) denotes the Hamming weight of {\bf a} . As a first observation, when n=t , the set \mathcal{B}(n,t,k_{+},k_{-}) coincides with the cube [-k_{-},k_{+}]^{n} . Therefore, in what follows, we restrict our attention to the nontrivial case 1\leq t\leq n-1 . The study of tilings (and packings) of \mathbb{Z}^{n} by \mathcal{B}(n,1,k_{+},k_{-}) has attracted considerable attention in recent years, both for its intrinsic combinatorial interest and for its applications to flash memory. In particular, extensive research has been devoted to the cases of the cross \mathcal{B}(n,1,k,k) and the semi-cross \mathcal{B}(n,1,k,0) (see, for example, [ 4 , 7 , 8 , 10 , 15 , 16 , 17 , 24 , 25 , 27 , 28 , 29 , 30 , 32 ] ). Subsequently, Schwartz [ 22 ] generalized these shapes to the quasi-cross \mathcal{B}(n,1,k_{+},k_{-}) , which has since been the subject of extensive investigation (see [ 23 , 33 , 34 , 35 , 36 , 37 , 39 , 42 ] and the references therein). In contrast, for the case t\geq 2 , only a limited number of results are known (see [ 1 , 6 , 16 , 18 , 26 , 41 ] and the references therein). This highlights a significant gap in our understanding of higher-weight error balls. It is also worth noting that the cross \mathcal{B}(n,1,1,1) coincides with the Lee sphere of radius 1 . More generally, the existence of tilings of \mathbb{Z}^{n} by Lee spheres is a long-standing open problem, known as the Golomb–Welch conjecture [ 5 ] . Except for the cases of radius 1 or dimension 2 , it is widely conjectured that such tilings do not exist. We refer the reader to [ 2 , 11 , 12 , 13 , 19 , 21 , 38 , 40 , 43 ] , as well as the works [ 3 , 14 ] , for surveys on the current status of this conjecture. In this paper, we focus on lattice tilings of \mathbb{Z}^{n} by semi-cross \mathcal{B}(n,1,k,0) . It is well known that this geometric problem can be reduced to an algebraic one, namely the problem of splitting finite abelian groups. To this end, let G be a finite abelian group, written additively. For s\in G and a nonnegative integer m\in\mathbb{Z} , we denote by ms the sum s+\cdots+s , where s appears m times. Definition 1.1 . A splitting of G consists of a pair of sets M\subset\mathbb{Z}\setminus\{0\} , called the multiplier set , and S=\{s_{1},s_{2},\dots,s_{n}\}\subseteq G , called the splitter set , such that every nonzero element of G admits a unique representation of the form ms , where m\in M and s\in S . Let k be a positive integer, and define S(k):=\{1,2,\dots,k\}. The connection between tilings and group splittings is given by the following classical result. Theorem 1.2 . [ 10 , 27 ] There exists a lattice tiling of \mathbb{Z}^{n} by the semi-cross \mathcal{B}(n,1,k,0) if and only if there exists a finite abelian group G of order nk+1 such that S(k) splits G . Moreover, the above problem can be further reduced to the cyclic case. Theorem 1.3 . [ 9 , 27 ] If S(k) splits a finite abelian group G , then it also splits the cyclic group \mathbb{Z}_{|G|} . Consequently, the existence of lattice tilings by semi-crosses is closely tied to the structure of splittings of cyclic groups. In particular, the following necessary condition is known. Theorem 1.4 . [ 27 , Chapter 4, Theorem 3] Let n\geq 3 and k be integers. If S(k) splits a finite abelian group G of order nk+1 , then k\leq n-2 . To further investigate splittings of finite abelian groups, we recall the following definition from [ 27 ] . Definition 1.5 . Let G be a finite abelian group, and let M and S be the multiplier and splitter sets forming a splitting of G . The splitting is called nonsingular if \gcd(|G|,m)=1 for every m\in M . Otherwise, it is called singular . Moreover, if for every prime divisor p of |G| there exists some m\in M such that p\mid m , then the splitting is called purely singular . This terminology provides a useful framework for the study of splittings, since the classification problem for splittings of finite abelian groups can be reduced to the classification of nonsingular and purely singular splittings [ 27 ] . Although a number of results are known for nonsingular splittings, the purely singular case remains much less understood. One result established by Woldar [ 32 ] is the following. Theorem 1.6 . [ 32 ] If S(k) splits a finite abelian group G purely singularly, then either \gcd(|G|,6)=1 or G is one of \mathbb{Z}_{1} , \mathbb{Z}_{k+1}

Read the complete text