Markov-self-similar sets (Q1191884)

Let \(\{f_ 1,\dots,f_ N\}\) be a family of contractions of a separable, complete metric space. A set \(K\) is called self-similar if it satisfies \(K=f_ 1(K)\cup\cdots\cup f_ N(K)\). This paper considers generalisations of this concept where we have a vector of sets \((K_ 1,\dots,K_ N)\) with each \(K_ k\) satisfying \(K_ k=f_{k1}(K_ 1)\cup\cdots f_{kN}(K_ N)\) for a family of contractions \(\{f_{11},\dots,f_{NN}\}\). This type of generalisation has a rich history, starting with the recurrent sets introduced by the reviewer, the mixed self-similar sets considered by \textit{C. Bandt} [Math. Nachr. 142, 107-123 (1989; Zbl 0707.28004) and Monatsh. Math. 108, No. 2/3, 89-102 (1989; Zbl 0712.58039)], the recurrent iterated function systems considered by \textit{M. F. Barnsley}, \textit{J. E. Elton} and \textit{D. P. Hardin} [Constructive Approximation 5, No. 1, 3-31 (1989; Zbl 0659.60045)], and many others (cf. the references of the paper under review, and the papers mentioned). This entails that some of the results in the first part of the paper are also obtained in one or more of the references given above. The second part gives results concerning the a.s. Hausdorff dimension and Hausdorff measure of random Markov self-similar sets, where the contractions \(f_{jk}\) are chosen randomly.