Article contents
Measures of maximal relative entropy with full support
Published online by Cambridge University Press: 17 November 2010
Abstract
Let π be a factor map from an irreducible shift of finite type X to a shift space Y. Let ν be an invariant probability measure on Y with full support. We show that every measure on X of maximal relative entropy over ν is fully supported. As a result, given any invariant probability measure ν on Y with full support, there is an invariant probability measure μ on X with full support that maps to ν under π. If ν is ergodic, μ can be chosen to be ergodic. These results can be generalized to the case of sofic shifts. We demonstrate that the results do not extend to general shift spaces by providing counterexamples.
- Type
- Research Article
- Information
- Copyright
- Copyright © Cambridge University Press 2010
References
- 3
- Cited by