On the cardinality and complexity of the set of codings for self-similar sets with positive Lebesgue measure (Q907027)

From MaRDI portal
scientific article
Language Label Description Also known as
English
On the cardinality and complexity of the set of codings for self-similar sets with positive Lebesgue measure
scientific article

    Statements

    On the cardinality and complexity of the set of codings for self-similar sets with positive Lebesgue measure (English)
    0 references
    0 references
    1 February 2016
    0 references
    \textit{P. Erdős} et al. [Bull. Soc. Math. Fr. 118, No. 3, 377--390 (1990; Zbl 0721.11005)] showed that if \(\lambda \in (\gamma, 1)\), where \(\gamma\) denote the reciprocal of the Golden mean, then every \(x \in (0, \lambda/(1-\lambda))\) has a continuum of \(\lambda\)-expansions. Namely, there exists a continuum of infinite sequences \((\epsilon_{i})_{i = 1}^{\infty}\) with \(\epsilon_{i} \in \{ 0, 1\}\), for all \(i \in \mathbb{N}\), and \[ x = \sum_{i = 1}^{\infty} \epsilon_{i} \lambda^{i}. \] The lower bound \(\gamma\) was shown to be sharp by \textit{Z. Daróczy} and \textit{I. Kátai} [Publ. Math. 42, No. 3--4, 397--407 (1993; Zbl 0809.11008)] in the following sense: if \(\lambda \in (1/2, \gamma)\), then there exists an \(x\) with a unique \(\lambda\)-expansion. The size of the set of points with a unique \(\lambda\)-expansion has been further studied by \textit{P. Glendinning} and \textit{N. Sidorov} [Math. Res. Lett. 8, No. 4, 535--543 (2001; Zbl 1002.11009)], amongst other things it was shown that the set of \(x \in (0, \lambda/(1-\lambda))\) with a unique \(\lambda\)-expansion has positive Hausdorff dimension when \(\lambda \in (0, \lambda^{*})\). Here \(\lambda^{*}\) denotes the reciprocal of the Komornik-Loreti constant. In [Am. Math. Mon. 110, No. 9, 838--842 (2003; Zbl 1049.11085)] \textit{N. Sidorov} complemented this result by showing that Lebesgue almost every \(x \in (0, \lambda/(1-\lambda))\) has a continuum of \(\lambda\)-expansions, for any \(\lambda \in (1/2, 1)\). There is a strong relationship between \(\lambda\)-expansions and iterated function systems. In the setting of the paper in review, an iterated function system is a finite collection of maps \(\{ f_{1}, f_{2}, \dots, f_{n} \}\) where \(f_{i}: \mathbb{R}^{d} \to \mathbb{R}^{d}\) is a contraction, that is, there exists a \(\lambda \in (0, 1)\) with \(\| f(x) - f(y) \| \leq c \| x - y \|\) for all \(x, y \in \mathbb{R}^{d}\). It is a well-know fact that to each iterated function system there exists a unique non-empty compact set \(\Lambda \subseteq \mathbb{R}^{d}\), called the attractor, which satisfies, \(\Lambda = f_{1}(\Lambda) \cup f_{2}(\Lambda) \cup \dots \cup f_{n}(\Lambda)\). Setting \(d = 1\), \(n = 2\), \(f_{1}(x) = \lambda x\) and \(f_{2} = \lambda x + \lambda\) one can show that the attractor of \(\{ f_{1}, f_{2} \}\) is the closed interval \([0, \lambda/(1-\lambda)]\) and that \((\epsilon_{i})_{i = 1}^{\infty}\) is a \(\lambda\)-expansion of \(x \in (0, \lambda/(1-\lambda))\), if and only if, \[ \lim_{N \to \infty} f_{\epsilon_{1} + 1} \circ f_{\epsilon_{2} + 1} \circ \dots \circ f_{\epsilon_{N} + 1}(0) = x. \] Given a collection of real numbers \(\lambda_{1}, \lambda_{2}, \dots, \lambda_{n}\) in the open unit interval \((0, 1)\) and a collection of points \(p_{1}, p_{2}, \dots, p_{n}\) in \(\mathbb{R}^{d}\), the author of the article in review considers the size and complexity of the set of codings of a generic \(x\) in the attractor of the iterated function system \(\{ f_{i}: x \to \lambda_{i}x + p_{i} \}_{i = 1}^{n}\). Here a coding of a point \(x\), is an infinite sequence \(\{ \epsilon_{i} \}_{i= 1}^{\infty}\), where \(\epsilon_{i} \in \{ 1, 2, \dots, n \}\) and \[ \lim_{N \to \infty} f_{\epsilon_{1}} \circ f_{\epsilon_{2}} \circ \dots \circ f_{\epsilon_{N}}(0) = x. \] Further, a coding \((\epsilon_{i})_{i = 1}^{\infty}\) is called universal if for any given finite word \((\delta_{1}, \delta_{2}, \dots, \delta_{N})\), with \(\delta_{i} \in \{ 1, 2, \dots, n \}\), there exists a \(k \in \mathbb{N}\) such that \(\epsilon_{k + i} = \delta_{i}\) for all \(i \in \{ 1, 2, \dots, N \}\). Specifically, the author of the article in review proves the following results (Theorems 1.2, 1.3 and 1.4), which generalises and strengthens the results of \textit{N. Sidorov} [Nonlinearity 20, No. 5, 1299--1312 (2007; Zbl 1122.60066); Period. Math. Hung. 47, No. 1--2, 221--231 (2003; Zbl 1049.11009)]. Here, analogous results are obtained, but for homogeneous iterated function systems where \(\Lambda\) is assumed to be equal to its convex hull. (An iterated function systems is said to be homogeneous if \(\lambda_{1} = \lambda_{2} = \dots = \lambda_{n}\).) Theorem 1.2. If \(\lambda_{1}^{d} + \lambda_{2}^{d} + \dots + \lambda_{n}^{d} \neq 1\) and the \(d\)-dimensional Lebesgue measure of \(\Lambda\) is positive, then almost every \(x \in \Lambda\) has a continuum of codings. Theorem 1.3. If \(f_{k}(\Lambda) \cap f_{l}(\Lambda) \neq \emptyset\) for some \(1 \leq k < l \leq n\), then the set of points that do not have a continuum of codings has Hausdorff dimension strictly less than \(d\). Theorem 1.4. If the \(d\)-dimensional Lebesgue measure of \(\Lambda\) is positive, then almost every \(x \in \Lambda\) has a universal coding At the heart of the proof of Theorems 1.2 and 1.4, is an application of Lebesgue's density theorem. The article also contains a section on applications to \(\lambda\)-expansions with deleted digits and concludes with a collection of interesting open problems.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    iterated function systems
    0 references
    beta-expansions
    0 references
    Hausdorff dimension
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references