CONTRACTIVE MARKOV SYSTEMS (Q4673208)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2162730
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | CONTRACTIVE MARKOV SYSTEMS |
scientific article; zbMATH DE number 2162730 |
Statements
CONTRACTIVE MARKOV SYSTEMS (English)
0 references
29 April 2005
0 references
invariant probability measure
0 references
attractive measure
0 references
0.8623495101928711
0 references
0.8385739922523499
0 references
0.8261919617652893
0 references
0.8225914835929871
0 references
0.8217182755470276
0 references
Let \((K, d)\) be a (locally compact separable) metric space and let \(K_1, \dots, K_N\) be a partition of \(K\) into non-empty Borel subsets. For each \(i \in \{1, \dots, N\}\), let \(\{w_{i, j}, \, 1 \leq j \leq L_i\}\) be a family of Borel measurable maps from \(K_i\) to \(K\) such that for every \(j\), there is an \(n \in \{1, \dots, N\}\) such that \(w_{i, j}(K_i) \subset K_n\); let \(\{p_{i, j},1 \leq j \leq L_i\}\) be a family of Borel measurable probability measure functions on \(K_i\). The set \(V = \{1, \dots, N\}\) is called the set of vertices and the set \(E = \{ (i, n): i \in V, n \in \{1,\dots, L_i\}\}\) is called the set of edges. When \(e = (i, n) \in E\), write \(p_{i, n}\) and \(w_{i, n}\) as \(p_e\) and \(w_e\), respectively. Then the family \({\mathcal M}= (K_{i(e)}, w_e, p_e)_{e \in E}\) is called a Markov system. Markov systems arise naturally in the so-called graph-directed constructions of fractal sets. A Markov system \((K_{i(e)}, w_e, p_e)_{e \in E}\) is called contractive if there exists \(0 < a < 1\) such that \[ \sum_{e \in E}p_e(x) d(w_e(x), w_e(y)) \leq a \, d(x, y) \] for all \(x, y \in K_i\) and \(i \in V\). The author proves: (i) If \({\mathcal M}= (K_{i(e)}, w_e, p_e)_{e \in E}\) is a contractive Markov system such that \(K_1, \dots, K_N\) are all open and the probability functions \(p_{e}\) are continuous on \(K_{i(e)}\), then \({\mathcal M}\) has an invariant Borel probability measure. (ii) If \({\mathcal M}\) is irreducible and the probability functions \(p_{e}\) are Dini-continuous and bounded below from 0, then the invariant measure is unique. (iii) If, in addition, \({\mathcal M}\) is aperiodic, then the invariant measure is an attractive probability measure of \({\mathcal M}\).
0 references