New Applications of Ergodic Theory to Sets of Differences

Kabir Belgikar, Vitaly Bergelson, Gabriel Black, David Kruzel · 2024

Read New Applications of Ergodic Theory to Sets of Differences by Kabir Belgikar, Vitaly Bergelson, Gabriel Black, David Kruzel (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-2412-01185.txt.

Introduction

Kabir Belgikar, Vitaly Bergelson, Gabriel Black, and David Kruzel bring ergodic theory, the study of long-run averages in moving systems, to a classic topic in number theory: sets of differences. Their 2024 paper revisits known results by Stewart, Tijdeman, and Ruzsa, three researchers who studied how difference sets behave inside sets of whole numbers.

The authors show the ergodic approach both simplifies the old proofs and extends their reach. A notable feature is their use of pointwise ergodic theory, which tracks averages along single paths rather than only on average. The paper offers number theorists a sharper set of tools for difference problems. The work also gives simpler proofs of some of those older results.

Catalog record

CREATORKabir Belgikar, Vitaly Bergelson, Gabriel Black, David Kruzel
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE95 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-01185

Browse the full library · About the Tower of Babel

Full text

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

New Applications of Ergodic Theory to Sets of Differences Kabir Belgikar, Vitaly Bergelson, Gabriel Black, David Kruzel Abstract Abstract. We apply the methods of ergodic theory to both simplify and significantly extend some classical results due to Stewart, Tijdeman, and Ruzsa. One of the notable features of our approach is the utilization of pointwise ergodic theory. 1 Introduction Given a set S\subseteq\mathbb{N}=\{1,2,3,\ldots\} , the upper density of S , denoted \overline{d}(S) , is defined by \overline{d}(S)=\limsup_{N\rightarrow\infty}\frac{|S\cap\{1,\ldots,N\}|}{N}, and the lower density of S , denoted \underline{d}(S) , is defined by \underline{d}(S)=\liminf_{N\rightarrow\infty}\frac{|S\cap\{1,\ldots,N\}|}{N}. If \overline{d}(S)=\underline{d}(S) , then the common value is denoted by d(S) and is called the natural density of S . Following Rusza, we introduce three types of difference sets; note that these sets are symmetric subsets of \mathbb{Z} . \Delta_{1}(S)=\{n\in\mathbb{Z}:S\cap(S-n)\neq\emptyset\},\quad\quad\Delta_{2}(S)=\{n\in\mathbb{Z}:\overline{d}(S\cap(S-n))>0\}, \Delta_{3}(S)=\{n\in\mathbb{Z}:|S\cap(S-n)|=\infty\}. Suppose that S_{1},\ldots,S_{k}\subseteq\mathbb{N} and \overline{d}(S_{i})>0 for 1\leq i\leq k . In [ 27 ] , Stewart and Tijdeman answered in the affirmative a question of Erdős by showing that the set C\vcentcolon=\bigcap_{i=1}^{k}\Delta_{3}(S_{i}) has positive lower density. Moreover, they showed that C is syndetic , meaning that there are integers m_{1},\ldots,m_{\ell} such that \bigcup_{i=1}^{\ell}\left(C+m_{i}\right)=\mathbb{Z}. (1.1) Stewart and Tijdeman’s result was amplified by Rusza as follows. Theorem 1.1 ( [ 24 , Theorem 1 and Theorem 2] ) . If S_{1},\ldots,S_{k}\subseteq\mathbb{N} satisfy \overline{d}(S_{i})>0 for each i=1,\dots,k , then there exists S\subseteq\mathbb{N} such that d(S)\geq\prod_{i=1}^{k}\overline{d}(S_{i}) and \Delta_{1}(S)\subseteq D\vcentcolon=\bigcap_{i=1}^{k}\Delta_{2}(S_{i}) . Furthermore, there are m_{1},\ldots,m_{\ell}\in\mathbb{Z} , where \ell\leq\prod_{i=1}^{k}1/\overline{d}(S_{i}), (1.2) such that \bigcup_{i=1}^{\ell}(D+m_{i})=\mathbb{Z} . The goal of this paper is to offer an ergodic approach to Theorem 1 which not only provides an alternative proof, but also leads to various generalizations which range from establishing versions of Theorem 1 in the setup of countably infinite amenable cancellative semigroups 4 4 4 Amenable cancellative semigroups form a natural framework where notions of density can be defined. A precise definition of amenability is given in subsection 4.2 . to variants of Theorem 1 which involve modified versions of the sets \Delta_{1}(S) and \Delta_{2}(S) such as \Delta_{1,c}(S)=\{n\in\mathbb{N}:S\cap(S-[n^{c}])\neq\emptyset\}\quad\text{ and }\quad\Delta_{2,c}(S)=\{n\in\mathbb{N}:\overline{d}(S\cap(S-[n^{c}]))>0\} (1.3) with c>0 , or, more generally, \Delta_{1,g}(S)=\{n\in\mathbb{N}:S\cap(S-[g(n)])\neq\emptyset\}\quad\text{ and }\quad\Delta_{2,g}(S)=\{n\in\mathbb{N}:\overline{d}(S\cap(S-[g(n)]))>0\} (1.4) for certain functions g:\mathbb{N}\rightarrow[1,\infty) (in formulas ( 1.3 ) and ( 1.4 ), [\cdot] denotes the integral part). An interesting new variant of Theorem 1 is provided by the following theorem, which deals with the function g(n)=n^{c} and is a special case of the much more general Theorem 3.2 , to be proved in subsection 3.2 . Theorem 1.2. Let c>0 , c\not\in\mathbb{N} . If S_{1},\dots,S_{k}\subseteq\mathbb{N} satisfy \overline{d}(S_{i})>0 for each i=1,\dots,k , then there is S\subseteq\mathbb{N} with d(S)\geq\prod_{i=1}^{k}\overline{d}(S_{i}) and \Delta_{1,c}(S)\subseteq D_{c}\vcentcolon=\bigcap_{i=1}^{k}\Delta_{2,c}(S_{i}) . Furthermore, we have that \overline{d}(\Delta_{1,c}(S))\geq\prod_{i=1}^{k}\overline{d}(S_{i}) . In addition, the set D_{c} is thick, meaning that for all n\in\mathbb{N} there is an a\in\mathbb{N} such that \{a,a+1,\dots,a+n\}\subseteq D_{c} . The reader may wonder why we require c\notin\mathbb{N} in the statement of Theorem 1 . The reason for this (to be revealed in subsection 3.1 ) is that for c\notin\mathbb{N} , the sequence ([n^{c}])_{n=1}^{\infty} is pointwise ergodic (see Definition 3.1 ), which in turn leads to the rather sharp inequality d(S)\geq\prod_{i=1}^{k}\overline{d}(S_{i}) appearing in the statement of Theorem 1 . However, even when c\in\mathbb{N} , we still have the following nontrivial statement which naturally complements Theorem 1 and is a special case of Theorem 3.2 . Theorem 1.3. Let c\in\mathbb{N} . If S_{1},\ldots,S_{k}\subseteq\mathbb{N} satisfy \overline{d}(S_{i})>0 for each i=1,\ldots,k , then there is S\subseteq\mathbb{N} with d(S)>0 , and \Delta_{1,c}(S)\subseteq D_{c}\vcentcolon=\bigcap_{i=1}^{k}\Delta_{2,c}(S_{i}) . Furthermore, the set D_{c} is syndetic. Theorem 1 is weaker than Theorem 1 in the sense that it does not provide a bound for d(S) . But, unlike Theorem 1 , Theorem 1 guarantees syndeticity of the set D_{c} . Remark 1.4. An interesting feature of Theorem 1 is that for any c>1 with c\not\in\mathbb{N} the set D_{c}=\bigcap_{i=1}^{k}\Delta_{2,c}(S_{i}) is thick. One can show that if c\not\in\mathbb{N} , unlike in Theorem 1 , then the set D_{c} may not be syndetic. Also when c\in\mathbb{N} the syndetic set D_{c}=\bigcap_{i=1}^{k}\Delta_{2,c}(S_{i}) in Theorem 1 may not be thick. A detailed discussion regarding syndeticity and thickness of sets of the form D_{g}=\bigcap_{i=1}^{n}\Delta_{2,g}(S_{i}) for g:\mathbb{N}\to[1,\infty) , that appear in Theorems 3.2 and 3.2 (which contain Theorems 1 and 1 as rather special corollaries), is presented in subsection 3.3 . As mentioned above, one of the major goals of this paper is to demonstrate that ergodic methods allow one to amplify and generalize Theorem 1 to the setup of countable cancellative amenable semigroups. Theorem 1 , formulated below and proved in Section 4 , generalizes Theorem 1 in more than one way. First, Theorem 1 extends the setup from \mathbb{Z} to a general countably infinite amenable group G . Second, for the sets S_{1},\dots S_{k} , which appear in the formulation of Theorem 1 , it is assumed \overline{d}_{\mathbf{F}_{i}}(S_{i})>0 for i=1,\dots k , where the Følner sequences \mathbf{F}_{1},\dots,\mathbf{F}_{k} are arbitrarily chosen. Finally, the tempered Følner sequence \mathbf{G} , which appears in the formulation of Theorem 1 , leads even in the case G=\mathbb{Z} to a nontrivial generalization of Theorem 1 . We will discuss Følner and tempered Følner sequences in detail in subsection 4.2 . Theorem 1.5 (Theorem 4.2 , subsection 4.2 ) . Let G be a countably infinite amenable group with left Følner sequences \mathbf{G},\mathbf{F}_{1},\ldots,\mathbf{F}_{k} , where \mathbf{G} is left tempered. Let S_{1},\ldots,S_{k}\subseteq G such that \overline{d}_{\mathbf{F}_{i}}(S_{i})>0 for each i=1,\ldots,k . Then there is S\subseteq G such that d_{\mathbf{G}}(S)\geq\prod_{i=1}^{k}\overline{d}_{\mathbf{F}_{i}}(S_{i}) and \Delta_{1}(S)\subseteq D=\bigcap_{i=1}^{k}\Delta_{2}(\mathbf{F}_{i},S_{i}) . Furthermore, there are m_{1},\ldots,m_{\ell}\in G , where \ell\leq\prod_{i=1}^{k}1/\overline{d}_{\mathbf{F}_{i}}(S_{i}), (1.5) such that \bigcup_{i=1}^{\ell}(m_{i}D)=\bigcup_{i=1}^{\ell}(Dm_{i}^{-1})=G . The structure of this paper is as follows. The goal of Section 2 is to provide a short ergodic proof of Theorem 1 and to set up some ideas that will be further developed and amplified in subsequent sections. In Section 3, we obtain Theorems 3.2 and 3.2 dealing with the difference sets defined in ( 1.4 ), which contain, respectively, Theorems 1 and 1 as instantaneous corollaries. In Section 4, we will prove Theorem 1 and its more general version, Theorem 4.3 , which pertains to cancellative amenable semigroups. Finally, in Section 5, we give examples of tempered Følner sequences in various semigroups such as (\mathbb{N},+) , (\mathbb{N},\cdot) , the Heisenberg group, and locally finite groups. Acknowledgments: The authors thank John Griesmer and Saúl Rodríguez Martín for helpful communications. 2 Ergodic proof of Ruzsa’s Theorem The proof of Theorem 1 we present in this section will utilize two facts from ergodic theory. The first of them is an extension of the classical Poincaré recurrence theorem. 5 5 5 Poincaré’s recurrence theorem states that if (X,\mathcal{B},\mu) is a probability space, T a \mathcal{B} -measurable map X\to X such that \mu(T^{-1}A)=\mu(A) for all A\in\mathcal{B} , and B\in\mathcal{B} with \mu(B)>0 , then there is n\in\mathbb{N} such that \mu(B\cap T^{-n}B)>0 . Theorem 2.1 ( [ 5 , Theorem 1.2] ) . Let (X,\mathcal{B},\mu,T) be a measure-preserving system 6 6 6 A measure-preserving system is a quadruple (X,\mathcal{B},\mu,T) , where (X,\mathcal{B},\mu) is a probability space and T is a \mathcal{B} -measurable map X\to X such that for any A\in\mathcal{B} , one has \mu(T^{-1}A)=\mu(A) . A measure-preserving system (X,\mathcal{B},\mu,T) is invertible if the map T is invertible. and let B\in\mathcal{B} with \mu(B)>0 . Then there is a set P\subseteq\mathbb{N} with d(P)\geq\mu(B) such that for any n_{1},\ldots,n_{r}\in P we have \mu\left(B\cap T^{-n_{1}}B\cap\cdots\cap T^{-n_{r}}B\right)>0 . Remark

Read the complete text