Topology of univoque sets in double-base expansions

Vilmos Komornik, Yichang Li, Yuru Zou · 2025

Read Topology of univoque sets in double-base expansions by Vilmos Komornik, Yichang Li, Yuru Zou (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-2504-21374.txt.

Introduction

Vilmos Komornik and co-authors study numbers that can be written in exactly one way when two different bases share the writing. In a double-base expansion, each digit's weight depends on which base was chosen at every earlier step. The set of numbers with a unique such expansion is called univoque, and its shape is a delicate fractal question.

The equal-base case has been probed for years. With unequal bases, many old results survive, but a crowd of new phenomena appear from the added freedom. The authors chart the set's topological properties: where it is closed, scattered, or riddled with gaps. Pure as it sounds, unique expansions matter for coding and for understanding redundancy in number systems.

Catalog record

CREATORVilmos Komornik, Yichang Li, Yuru Zou
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE130 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-21374

Browse the full library · About the Tower of Babel

Full text

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

Topology of univoque sets in double-base expansions Vilmos Komornik, Yichang Li, Yuru Zou Abstract . Given two real numbers q 0 , q 1 > 1 q_{0},q_{1}>1 satisfying q 0 + q 1 ≥ q 0 ​ q 1 q_{0}+q_{1}\geq q_{0}q_{1} and two real numbers d 0 ≠ d 1 d_{0}\neq d_{1} , by a double-base expansion of a real number x x we mean a sequence ( i k ) ∈ { 0 , 1 } ∞ (i_{k})\in\{0,1\}^{\infty} such that x = ∑ k = 1 ∞ d i k q i 1 q i 2 ⋯ q i k . x=\sum_{k=1}^{\infty}\frac{d_{i_{k}}}{q_{{i_{1}}}q_{{i_{2}}}\cdots q_{{i_{k}}}}. We denote by U q 0 , q 1 \pazocal{U}_{{q_{0},q_{1}}} the set of numbers x x having a unique expansion. The topological properties of U q 0 , q 1 \pazocal{U}_{{q_{0},q_{1}}} have been investigated in the equal-base case q 0 = q 1 q_{0}=q_{1} for a long time. We extend this research to the case q 0 ≠ q 1 q_{0}\neq q_{1} . While many results remain valid, a great number of new phenomena appear due to the increased complexity of double-base expansions. 1. Introduction The study of non-integer base expansions started with the pioneering papers of Rényi [ 31 ] and Parry [ 29 ] . Since then hundreds of papers have been devoted to the study of expansions of real numbers of the form (1.1) x=\pi_{q,D}((d_{i})):=\mathop{\sum}\limits_{i=1}^{\infty}\dfrac{d_{i}}{q^{i}}, where q>1 is a given real number, and (d_{i}) is a sequence of digits , belonging to a finite alphabet D of real numbers. Many remarkable results have been discovered, revealing deep connections to various fields of mathematics, including number theory [ 33 , 22 ] , topology [ 12 , 13 ] , ergodic theory [ 19 ] , Diophantine approximation and dynamical systems [ 7 ] . Concerning the original alphabet \{0,1\} , Erdős et al. [ 14 , 15 ] discovered in the 1990’s that for each k\in\mathbb{N}\cup\{\aleph_{0}\}\cup\{2^{\aleph_{0}}\} there exist infinitely many bases q\in(1,2) such that x=1 has exactly k different expansions of the form ( 1.1 ). Subsequently the unique expansions have been intensively studied, and a surprisingly rich theory has emerged [ 16 , 24 , 17 , 30 , 22 , 10 , 21 , 11 , 20 , 5 , 12 , 6 , 23 , 1 , 2 , 3 , 34 , 35 , 13 ] . An essentially complete theory was presented in the papers [ 12 , 23 , 2 , 13 ] ; it was also shown that the theory remains valid for the more general alphabets \left\{0,1,\ldots,M\right\} , where M is an arbitrary positive integer. The paper [ 12 ] was devoted to the study of bases in which the number 1 has a unique expansion. Based on these results, the papers [ 23 , 2 , 13 ] were devoted to the sets of numbers having unique expansions in a fixed base. In the past few years the expansions ( 1.1 ) have been generalized by Neuhäuserer [ 28 ] , Li [ 27 ] and in [ 25 ] to multiple-base expansions of the form (1.2) x=\pi_{S}((i_{k})):=\sum_{k=1}^{\infty}\frac{d_{i_{k}}}{q_{{i_{1}}}q_{{i_{2}}}\cdots q_{{i_{k}}}},\quad(i_{k})\in\{0,1,\ldots,M\}^{\infty}, where S=\left\{(d_{0},q_{0}),(d_{1},q_{1}),\ldots,(d_{M},q_{M})\right\} is a given finite digit-base system of pairs of real numbers satisfying q_{0},q_{1},\ldots,q_{M}>1 . Although these generalized expansions have a much higher complexity (see, e.g., [ 26 ] ), most theorems of [ 12 ] could be generalized in [ 18 ] to all double-base expansions , i.e., to expansions of the form ( 1.2 ) with M=1 . A lot of new phenomena have appeared that do not occur in the equal-base case q_{0}=\cdots=q_{M} . The purpose of this paper is to similarly extend many theorems of [ 13 ] to this more general framework. Before stating the main results of this paper, let us recall the theorems of [ 13 ] that we are going to generalize. We need some definitions and notations. Unless stated otherwise, in this paper by a sequence we always mean an element of \left\{0,1\right\}^{\infty} , i.e., a sequence of zeros and ones. We systematically use the notations of symbolic dynamics for sequences (x_{i}) like x_{1}x_{2}\cdots, 0^{\infty} , 1^{\infty} , (10)^{\infty} or (10)^{k}1^{\infty} . We systematically use the lexicographical order between sequences: we write (x_{i})\prec(y_{i}) or (y_{i})\succ(x_{i}) if there exists an index n\in\mathbb{N} such that x_{i}=y_{i} for all i<n , and x_{n}<y_{n} . Furthermore, we write (x_{i})\preceq(y_{i}) or (y_{i})\succeq(x_{i}) if (x_{i})\prec(y_{i}) or if (x_{i})=(y_{i}) . The reflection of a sequence (x_{i}) is defined by the formula \overline{(x_{i})}:=(1-x_{i}) , i.e., we exchange the digits 0 and 1 . We denote by \sigma the right shift of sequences, so that \sigma^{n}(x_{1}x_{2}\cdots)=x_{n+1}x_{n+2}\cdots\quad\text{for every integer}\quad n\geq 0. We also consider the lexicographical order between finite words of digits of the same length, and the reflection of a word is defined similarly to the reflection of sequences. A sequence (x_{i}) is called • finite if it ends with 10^{\infty} , and infinite otherwise; • co-finite if its reflection is finite, i.e., if it ends with 01^{\infty} , and co-infinite otherwise; • doubly infinite if it is both infinite and co-infinite, i.e., if it contains infinitely many zero digits and infinitely many one digits. Remark 1.1 . There are only countably many finite or co-finite sequences, so that “most” sequences are doubly infinite. Now we consider the expansions of the form (1.3) x=\pi_{q}((d_{i})):=\mathop{\sum}\limits_{i=1}^{\infty}\dfrac{d_{i}}{q^{i}},\quad(d_{i})\in\left\{0,1\right\}^{\infty} with a given base q>1 on the alphabet \{0,1\} . Observe that if x has an expansion, then x\in J_{q}:=[0,\frac{1}{q-1}] . The converse is not true in general: \left\{\pi_{q}((d_{i}))\ :\ (d_{i})\in\left\{0,1\right\}^{\infty}\right\}=J_{q}\Longleftrightarrow q\in(1,2]. Moreover, if q\in(1,2] , then every x\in J_{q}:=[0,\frac{1}{q-1}] has a lexicographically largest expansion b(x,q)=(b_{i}(x,q)) , and a lexicographically largest infinite expansion a(x,q)=(a_{i}(x,q)) , called the greedy and quasi-greedy expansions of x in base q , respectively. Following [ 22 ] and [ 12 ] we introduce the sets \displaystyle\pazocal{U}:=\left\{q\in(1,2]\ :\ 1\text{ has a unique expansion in base }q\right\}, \displaystyle\pazocal{V}:=\left\{q\in(1,2]\ :\ 1\text{ has a unique doubly infinite expansion in base }q\right\}. Then the topological closure of \pazocal{U} has an analogous characterization: \displaystyle\overline{\pazocal{U}}=\left\{q\in(1,2]\ :\ 1\text{ has a unique infinite expansion in base }q\right\}. We recall that \pazocal{U}\subsetneqq\overline{\pazocal{U}}\subsetneqq\pazocal{V}\quad\text{with}\quad\left\lvert\pazocal{V}\setminus\overline{\pazocal{U}}\right\rvert=\left\lvert\overline{\pazocal{U}}\setminus\pazocal{U}\right\rvert=\aleph_{0}; here and in the sequel \left\lvert A\right\rvert denotes the cardinality of a set A . Furthermore, \pazocal{V} is compact and \overline{\pazocal{U}} is a Cantor set , i.e., a non-empty compact set having neither isolated, nor interior points. Their smallest elements are the Golden ratio and the Komornik–Loreti constant , respectively, and their largest element is 2 , also belonging to \pazocal{U} . As in [ 10 ] and [ 13 ] we introduce the following sets for each fixed base q\in(1,2] : \displaystyle\pazocal{U}_{q}:=\left\{x\in J_{q}\ :\ x\text{ has a unique expansion in base }q\right\}, \displaystyle\overline{\pazocal{U}}_{q}\text{ is the topological closure of }\pazocal{U}_{q}, \displaystyle\pazocal{V}_{q}:=\left\{x\in J_{q}\ :\ x\text{ has at most one doubly infinite expansion in base }q\right\}. Then \pazocal{V}_{2}:=J_{2}=[0,1] , and \displaystyle\pazocal{V}_{q}:=\left\{x\in J_{q}\ :\ x\text{ has a unique doubly infinite expansion in base }q\right\}\text{ if }q\in(1,2). We recall that \pazocal{U}_{q}\subseteqq\overline{\pazocal{U}}_{q}\subseteqq\pazocal{V}_{q}\quad\text{with}\quad\left\lvert\pazocal{V}_{q}\setminus\pazocal{U}_{q}\right\rvert\leq\aleph_{0}, and that \pazocal{V}_{q} is compact. Finally, we introduce the following partition of \pazocal{V}_{q}\setminus\pazocal{U}_{q} , where \alpha(q)=a(1,q) denotes the quasi-greedy expansion of 1 in base q : (1.4) \begin{split}&A_{q}:=\left\{x\in\pazocal{V}_{q}\setminus\pazocal{U}_{q}:\sigma^{i}(a(x,q))=\alpha(q)\text{ for at least one digit }a_{i}(x,q)=0\right\},\\ &B_{q}:=\left\{x\in\pazocal{V}_{q}\setminus\pazocal{U}_{q}:\sigma^{i}(a(x,q))\prec\alpha(q)\text{ for all $i$ with }a_{i}(x,q)=0\right\}.\\ \end{split} Equivalently, A_{q} and B_{q} are

Read the complete text