Hausdorff dimension of self-similar measures and sets with common fixed point structure

Balázs Bárány, Manuj Verma · 2025

Read Hausdorff dimension of self-similar measures and sets with common fixed point structure by Balázs Bárány, Manuj Verma (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-2507-05835.txt.

Introduction

Balázs Bárány and Manuj Verma compute the fractal dimension of self-similar measures whose generating maps share a fixed point. Self-similar sets and measures live on the real line here. When several contraction maps fix the same point, overlaps pile up in a structured way. Dimension formulas usually demand a separation condition to tame those overlaps.

The authors prove a weak exponential separation holds outside an exceptional parameter set of codimension one. That exception is far thinner than in earlier results, strengthening prior work by the first author with Szvák. As an application, they give the dimension of self-affine measures on a generalized four-corner set. Shared fixed points occur naturally in many constructions, so removing them from the exceptional pile matters. The result widens the territory where fractal dimension is exactly known.

Catalog record

CREATORBalázs Bárány, Manuj Verma
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-05835

Browse the full library · About the Tower of Babel

Full text

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

Hausdorff dimension of self-similar measures and sets with common fixed point structure Thanks: BB and MV acknowledge support from the grants NKFI K142169 and NKFI KKP144059 ”Fractal geometry and applications” Research Group. Balázs Bárány, Manuj Verma Abstract . In this paper, we study the Hausdorff dimension of self-similar measures and sets on the real line, where the generating iterated function system consists of some maps that share the same fixed point. In particular, we will show that out of a Hausdorff co-dimension one exceptional set of natural parameters, such systems satisfy a weak exponential separation. This significantly strengthens the previous result of the first author and Szvák [ 6 ] . As an application, we give the Hausdorff dimension of self-affine measures supported on the generalised 4-corner set. 1991 Mathematics Subject Classification Primary 28A80 Secondary 28A78 1. Introduction and Statements For N\geq 2 , the system \mathcal{I}=\big\{f_{1},f_{2},\ldots,f_{N}\big\} on \mathbb{R}^{d} is called an iterated function system (IFS) , if each f_{i} is a contraction map on \mathbb{R}^{d} for i\in\{1,2,\ldots,N\} . Hutchinson [ 16 ] proved that there exists a unique non-empty compact set A\subset\mathbb{R}^{d} such that A=\bigcup_{i=1}^{N}f_{i}(A). Let \mathbf{p}=(p_{1},p_{2},\dots,p_{N}) be a probability vector. Hutchinson [ 16 ] also proved that there exists a unique Borel probability measure \mu supported on A such that \mu(B)=\sum_{i=1}^{N}p_{i}\mu(f_{i}^{-1}(B)) for all Borel sets B\subset\mathbb{R}^{d} . The set A is called the attractor of IFS \mathcal{I} and the measure \mu is known as the stationary measure corresponding to IFS \mathcal{I} with the probability vector \mathbf{p} . This paper focuses on the dimension theory of self-similar IFSs on the line. The IFS \mathcal{I} is called a self-similar iterated function system on \mathbb{R} if each f_{i}:\mathbb{R}\to\mathbb{R} is a similarity map for i\in\{1,2,\dots,N\} , i.e. f_{i}(x)=r_{i}x+c_{i} where 0<|r_{i}|<1 and c_{i}\in\mathbb{R} . Then, the attractor A is called a self-similar set corresponding to the self-similar IFS \mathcal{I} . The stationary measure \mu is known as the self-similar measure corresponding to the self-similar IFS \mathcal{I} with probability vector \mathbf{p}=(p_{1},p_{2},\dots,p_{N}) . For later purposes, let us recall the definition of the Hausdorff dimension here. Let F\subseteq\mathbb{R}^{d} . We say that \{U_{i}\}_{i=1}^{\infty} is a \delta cover of F if F\subset\bigcup\limits_{i=1}^{\infty}U_{i} and 0<|U_{i}|\leq\delta for each {i} , where |U_{i}| denotes the diameter of the set U_{i} . For each \delta>0 and s\geq 0 , we define \mathcal{H}_{\delta}^{s}(F):=\inf\Big\{\sum_{i=1}^{\infty}|U_{i}|^{s}:\{U_{i}\}\text{ is a }\delta\text{-cover of }F\Big\}\text{ and }\mathcal{H}^{s}(F):=\lim_{\delta\to 0+}\mathcal{H}_{\delta}^{s}(F). We call \mathcal{H}^{s}(F) the s -dimensional Hausdorff measure of the set F . Using this, the Hausdorff dimension of the set F is defined by \dim_{H}(F)=\inf\{s\geq 0:\mathcal{H}^{s}(F)=0\}. For the basic properties, we refer the reader to Falconer’s book [ 11 ] . Another widely used dimension concept is the box-counting (or Minkowski) dimension; however, Falconer [ 9 ] pointed out that these dimension concepts are equal for self-similar sets, so we omit them. For a Borel probability measure \mu in \mathbb{R}^{d} , the lower- and upper local dimension of the measure \mu at the point x\in\mathbb{R}^{d} defined as \underline{d}_{\mu}(x)=\liminf_{r\to 0}\frac{\log\mu(B(x,r))}{\log r},\quad\overline{d}_{\mu}(x)=\limsup_{r\to 0}\frac{\log\mu(B(x,r))}{\log r}, where B(x,r) denotes the open ball of radius r with center x . We say that the measure \mu is exact dimensional if there is a constant c such that \underline{d}_{\mu}(x)=\overline{d}_{\mu}(x)=c for \mu -almost every x\in\mathbb{R}^{d} . We will denote this value by \dim_{H}(\mu) . Feng and Hu [ 12 ] proved that every self-similar measure \mu is exact dimensional. For further properties of the dimension of measures, see Przytycki and Urbański [ 21 , Section 7.6] . Let us define the similarity dimension for the self-similar IFS \mathcal{I} to be the unique number s_{0}\in\mathbb{R} such that \sum_{i=1}^{N}|r_{i}|^{s_{0}}=1. Furthermore, the quantities h_{\mathbf{p}}=-\sum_{i=1}^{N}p_{i}\log p_{i}\quad\text{and}\quad\chi(\mathbf{p})=-\sum_{i=1}^{N}p_{i}\log|r_{i}| are known respectively as the entropy and the Lyapunov exponent of the self-similar measure \mu corresponding to the self-similar IFS \mathcal{I} with probability vector \mathbf{p} . It is well known that \dim_{H}(A)\leq\min\{1,s_{0}\}\quad\text{and}\quad\dim_{H}(\mu)\leq\min\bigg\{1,\frac{h_{\mathbf{p}}}{\chi(\mathbf{p})}\bigg\} regardless of any further assumptions. However, equalities appear in the above expressions under some suitable separation conditions. Theorem 1.1 (Hutchinson [ 16 ] , Cawley and Mauldin [ 8 ] ) . Let the self-similar IFS \mathcal{I} on \mathbb{R} . If \mathcal{I} satisfies the open set condition (OSC) , that is, there is a nonempty and bounded open set O\subset\mathbb{R}^{d} with f_{i}(O)\subset O and f_{i}(O)\cap f_{j}(O)=\emptyset for i\neq j\in\{1,2,\dots,N\} , then \dim_{H}(A)=s_{0}\quad\text{and}\quad\dim_{H}(\mu)=\frac{h_{\mathbf{p}}}{\chi(\mathbf{p})}, (1.1) where A is the self-similar attractor of the self-similar IFS \mathcal{I} and \mu is the self-similar measure corresponding to the self-similar IFS \mathcal{I} with probability vector \mathbf{p} . Hochman [ 14 ] introduced the Exponential Separation Condition for the self-similar IFSs, which is a much weaker separation condition than the OSC. We say that the self-similar IFS \mathcal{I} on \mathbb{R} satisfies the Exponential Separation Condition (ESC) if there exists a constant \eta>0 such that for infinitely many values of n\in\mathbb{N} for all i_{1},\ldots,i_{n},j_{1},\ldots,j_{n}\in\{1,\ldots,N\} with (i_{1},\ldots,i_{n})\neq(j_{1},\ldots,j_{n}) and r_{i_{1}}\cdots r_{i_{n}}=r_{j_{1}}\cdots r_{j_{n}} , we have |f_{i_{1}}\circ\cdots\circ f_{i_{n}}(0)-f_{j_{1}}\circ\cdots\circ f_{j_{n}}(0)|\geq{\eta}^{n}. In a proper sense, the exponential separation is a typical property for self-similar IFSs. For example, suppose the contraction ratios and translations are analytic maps of some one-dimensional parameter, and the system is not degenerate. In that case, ESC holds out of a zero-dimensional set of exceptional parameters. In particular, for fixed contraction ratios with \max_{i\neq j}\{|r_{i}|+|r_{j}|\}<1 , the dimension of translation vectors (c_{i})_{i=1}^{N}\in\mathbb{R}^{N} , for which \mathcal{I} does not satisfy ESC, is at most N-1 , see [ 4 , Corollary 6.8.12] 1 1 1 We note that the assumption \max_{i\neq j}\{|r_{i}|+|r_{j}|\}<1 is missing incorrectly in the statement. . Hochman [ 14 ] proved the following groundbreaking result for the dimension theory of self-similar IFSs. Theorem 1.2 . [Hochman [ 14 ] ] If the self-similar IFS \mathcal{I} on \mathbb{R} satisfies the ESC, then \dim_{H}(A)=\min\{1,s_{0}\}\quad\text{and}\quad\dim_{H}(\mu)=\min\left\{1,\frac{h_{\mathbf{p}}}{\chi(\mathbf{p})}\right\}. (1.2) The ESC implies that the self-similar IFS \mathcal{I} has no exact overlap, i.e. there are no finite sequences of symbols (i_{1},\ldots,i_{n})\neq(j_{1},\ldots,j_{m}) such that f_{i_{1}}\circ\cdots\circ f_{i_{n}}\equiv f_{j_{1}}\circ\cdots\circ f_{j_{m}} . When exact overlaps exist, a better bound can be obtained by considering the number of exact overlaps. Lau and Ngai [ 18 ] and Zerner [ 23 ] introduced the Weak Separation Condition (WSC), which allows exact overlaps. Still, the not-exactly overlapping maps must be relatively far from each other, comparable with the contraction ratios of the maps. This allows us to calculate the dimension of the attractor as a limit of the similarity dimensions of higher iterates. In some special cases, when the contraction ratios are reciprocals of integers, it is possible to calculate the dimension of the corresponding self-similar measures, see Ruiz [ 22 ] . The situation becomes significantly more difficult if the overlapping structure of the self-similar IFS \mathcal{I} is complicated, for instance, when some of the maps of the IFS share the same fixed point. Let (t_{i})_{i=1}^{N}\in\mathbb{R}^{N} be a vector with pairwise different entries. For every i\in\{1,\ldots,N\} , let n_{i}\geq 1 be an integer. For every j\in\{1,\ldots,n_{i}\} , let \lambda_{i,j}\in(0,1) . Consider the following self-similar IFS on the real line: \mathcal{S}=\{f_{i,j}(x)=\lambda_{i,j}x+t_{i}(1-\lambda_{i,j})\}_{1\leq i\leq N,1\leq j\leq n_{i}}. (1.3) If n_{i}\geq 2 for some i , then the maps f_{i,j} and f_{i,j^{\prime}} with j\neq j^{\prime} share the same fixed point, which causes many exact overlaps, for example f_{i,j}\circ f_{i,j^{\prime}}\equiv f_{i,j^{\prime}}\circ f_{i,j} . Furthermore, if \frac{\log\lambda_{i,j}}{\log\lambda_{i,j^{\prime}}}\notin\mathbb{Q} then the IFS \mathcal{S} does satisfy the WSC, see Fraser [ 13 ] , and if \frac{\log\lambda_{i,j}}{\log\lambda_{i,j^{\prime}}}=\frac{p}{q}\in\mathbb{Q} then it causes further exact

Read the complete text