The paper describes two examples of pairs of ergodic processes \(P\) and \(Q\) (with \(Q\) being a \(B\)-process and \(P\) the binary symmetric coin- tossing process), with bad divergence-rate properties. More precisely, in the first example, the lower-divergence rate of \(P\) with respect to \(Q\) is 0, whereas the upper-divergence rate is \(\infty\). In the second example, the upper-divergence rate is 0, yet \(P\neq Q\).
Recommendations
Cites work
- A counterexample to Perez's generalization of the Shannon-McMillan theorem
- Cutting and independent stacking of intervals
- Cutting and stacking: a method for constructing stationary processes
- Entropy and isomorphism theorems for actions of amenable groups
- Entropy and prefixes
- scientific article; zbMATH DE number 3463931 (Why is no real title available?)
- Sanov property, generalized I-projection and a conditional limit theorem
- Stationary coding of processes
- String matching: The ergodic case
- The positive-divergence and blowing-up properties
Cited in
(9)- The positive-divergence and blowing-up properties
- An ergodic process of zero divergence-distance from the class of all stationary processes
- Asymptotic equivalence of probability measures and stochastic processes
- A quantum version of Sanov's theorem
- Typical support and Sanov large deviations of correlated states
- On stability of probability laws with respect to small violations of algorithmic randomness
- scientific article; zbMATH DE number 679787 (Why is no real title available?)
- Equivalence and nonequivalence of ensembles: thermodynamic, macrostate, and measure levels
- Discrimination between B-processes is impossible
This page was built for publication: Two divergence-rate counterexamples
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q685736)
