Abstract:Let $X$ be a full-shift on the alphabet $[0, 1]^a$ and let $(Y, S)$ be an arbitrary dynamical system. We prove that any equivariant continuous map from $X$ to $Y$ has conditional metric mean dimension not less than $a-\mathrm{mdim}(Y, S)$. This solves a problem posed in a paper of Shi--Tsukamoto (2023).
What problem does this paper attempt to address?
### What problem does the paper attempt to solve?
The paper aims to address the following issues:
1. **Background and Motivation**:
- The paper is a continuation of the previous work [ST23], which aims to solve an unresolved problem posed in [ST23, Problem 1.6].
- In [ST23], the authors studied the application of the waist inequality in dynamical system entropy and mean dimension. The waist inequality is a fundamental inequality in geometry and topology discovered by Gromov.
- [ST23] proposed a theorem regarding the waist inequality and applied it to the entropy and mean dimension of dynamical systems. However, there was an unsatisfactory factor of 2 in the theorem's conclusion.
2. **Specific Problem**:
- [ST23, Problem 1.6] asks whether this factor of 2 can be removed.
- The main goal of this paper is to affirmatively answer this question and improve the results in [ST23].
3. **Main Results**:
- Theorem 1.2 (Main Theorem): For any dynamical system (Y, S), if there is a continuous and dynamics-preserving map π from ([0,1]^a)^Z to (Y, S), then:
\[
\text{mdim}_M(\pi, \sigma, D) \geq a - \text{mdim}(Y, S).
\]
- This result improves the conclusion in [ST23] by removing the original factor of 2, making the estimate more precise.
- If \(a > \text{mdim}(Y, S)\), then any continuous and dynamics-preserving map π has infinite topological conditional entropy.
4. **Further Questions**:
- The paper also poses a further question (Problem 1.3), which asks whether every continuous and dynamics-preserving map π has infinite topological conditional entropy when two dynamical systems (X, T) and (Y, S) satisfy \(\text{mdim}(X, T) > \text{mdim}(Y, S)\).
- The main theorem addresses this question in the case where (X, T) is the full shift on the alphabet [0,1]^a, but it remains open in the general case.
Through this work, the paper not only improves previous research results but also provides new directions for further research.