Kolmogorov-Sinai entropy via separation properties of order-generated \(\sigma\)-algebras (Q379842)

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 6224707
Language Label Description Also known as
default for all languages
No label defined
    English
    Kolmogorov-Sinai entropy via separation properties of order-generated \(\sigma\)-algebras
    scientific article; zbMATH DE number 6224707

      Statements

      Kolmogorov-Sinai entropy via separation properties of order-generated \(\sigma\)-algebras (English)
      0 references
      0 references
      0 references
      0 references
      11 November 2013
      0 references
      Kolmogorov-Sinai entropy
      0 references
      permutation entropy
      0 references
      multijets transversality theorem
      0 references
      residuality
      0 references
      prevalence
      0 references
      Suppose that \((\Omega,\mathcal{F},\mu,T)\) is a probability-preserving system, not necessarily invertible. Then its Kolmogorov-Sinai entropy may be computed as the supremum of the Kolmogorov-Sinai entropies obtained from any increasing sequence of measurable partitions of \(\Omega\) which generate the whole of \(\mathcal{F}\) modulo \(\mu\). This paper gives conditions under which suitable partitions can be obtained from the order-types of some sequences of random vectors generated by the system. The key point is to show that these partitions together generate the whole of \(\mathcal{F}\) modulo \(\mu\).NEWLINENEWLINEThe rest of the paper has two parts.NEWLINENEWLINEThe technical heart of the first part may be described as follows. Suppose that \((\xi_n)_{n\geq 1}\) is a real-valued one-sided stationary stochastic process. For each \(m\geq 1\), let \(\mathcal{P}_m\) be the partition of the underlying probability space generated by the order-type of the random vector \((\xi_0,\xi_1,\ldots,\xi_m)\). Then the sigma-algebra generated by the partitions \(\mathcal{P}_m\) is equal to the sigma-algebra generated by the whole process, modulo negligible sets. This fact has an easy generalization to the case of an \(\mathbb{R}^d\)-valued process in which one simply treats each of the \(d\) coordinate-processes separately. From this, the authors deduce their first main result:NEWLINENEWLINE Theorem 2.5: If either \(T\) is ergodic or \((\Omega,\mathcal{F})\) is standard Borel, then any \(\mathbb{R}^d\)-valued observable whose \(T\)-shifts together generate the whole of \(\mathcal{F}\) may be used to compute the Kolmogorov-Sinai entropy as described above.NEWLINENEWLINEThe second part of the paper shows that, if \((\Omega,\mathcal{F},\mu)\) consists of an absolutely continuous measure on the Borel sets of a smooth \(m\)-dimensional manifold, and if \(d > m\), then the set of \(\mathbb{R}^d\)-valued observables which have the property above is `large' in the space of all \(\mathbb{R}^d\)-valued smooth functions on \(\Omega\). This is proved in two senses: this set of observables is residual for the strong topology on smooth \(\mathbb{R}^d\)-valued functions, and it is prevalent for the weak topology.
      0 references

      Identifiers