Local mean dimension theory for sofic group actions

Felipe García-Ramos,Yonatan Gutman
2024-09-17
Abstract:Using a local perspective, we introduce \textit{mean dimension pairs} and give sufficient conditions of when every non-trivial factor of a continuous group action of a sofic group $G$ has positive mean dimension. In addition we show that the mean dimension map is Borel, and that the set of subshifts with completely positive mean dimension of $[0,1]^G$, the full $G$-shift on the interval, is a complete coanalytic set in the set of all subshifts (hence not Borel). Our results are new even when the acting group is $\Z$.
Dynamical Systems
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is about the local mean - dimension theory under continuous group actions, especially for sofic group actions. Specifically, the author introduced the concept of mean - dimension pairs and gave sufficient conditions for a nontrivial factor of a continuous group action to have positive mean - dimension. In addition, they also proved that the mean - dimension map is Borel, and under certain conditions, the set of sub - shifts with fully positive mean - dimension is a complete co - analytic set rather than a Borel set. These results are new contributions even for the case where the acting group is Z. In short, the goals of this paper are: 1. **Introduce mean - dimension pairs**: This is to better understand and analyze dynamical systems under sofic group actions. 2. **Give sufficient conditions for nontrivial factors to have positive mean - dimension**: This helps to identify which systems have complex structures locally. 3. **Prove the properties of the mean - dimension map**: In particular, prove that it is Borel, which is very important for further theoretical research. 4. **Explore the complexity of the set of sub - shifts with fully positive mean - dimension**: By proving that these sets are not Borel sets, it shows that they have high complexity. These achievements not only extend the existing mean - dimension theory but also provide new tools and perspectives for understanding dynamical systems under sofic group actions.