Abstract:We study the interaction between entropy and Wasserstein distance in free probability theory. In particular, we give lower bounds for several versions of free entropy dimension along Wasserstein geodesics, as well as study their topological properties with respect to Wasserstein distance. We also study moment measures in the multivariate free setting, showing the existence and uniqueness of solutions for a regularized version of Santambrogio's variational problem. The role of probability distributions in these results is played by types, functionals which assign values not only to polynomial test functions, but to all real-valued logical formulas built from them using suprema and infima. We give an explicit counterexample showing that in the framework of non-commutative laws, the usual notion of probability distributions using only non-commutative polynomial test functions, one cannot obtain the desired large-$n$ limiting behavior for both Wasserstein distance and entropy simultaneously in random multi-matrix models.
What problem does this paper attempt to address?
The problems that this paper attempts to solve mainly focus on the interaction between entropy and Wasserstein distance in free probability theory, especially the behavior of random matrix models in the large \(n\) limit. Specifically, the author attempts to answer the following key questions:
1. **What is the correct analogy between entropy and Wasserstein distance in free probability theory?**
2. **How are free entropy and free Wasserstein distance related to each other?**
3. **Under what conditions can free entropy and free Wasserstein distance describe the behavior of classical entropy and Wasserstein distance in the large \(n\) limit of invariant random multi - matrix ensembles?**
### Main Research Contents
- **Lower Bound Estimation of Entropy and Wasserstein Distance**: The author gives several versions of lower bound estimations of the free entropy dimension (free entropy dimension) along Wasserstein geodesics and studies their topological properties under Wasserstein distance.
- **Existence and Uniqueness of Moment Measures**: The author studies moment measures in the multivariate free setting and proves the existence and uniqueness of solutions to the regularized version of the Santambrigo variational problem.
- **The Role of Non - commutative Probability Distributions**: In these results, the role of probability distributions is played by types, that is, functional functionals, which assign values not only to polynomial test functions but also to all real logical formulas constructed from them using suprema and infima.
### Key Challenges
- **Differences between Classical and Free Probability Theories**: Many results in classical probability theory rely on exchangeability, while in free probability theory, this exchangeability no longer holds, so new tools and techniques are required to handle these problems.
- **Monge - Kantorovich Duality**: In the non - commutative setting, a natural class of scalar - valued "non - commutative continuous functions" is not closed under partial suprema and infima operations, which makes the classical Monge - Kantorovich duality difficult to apply directly.
- **Limit Problems**: Even if we know that certain types of random matrix models converge to a definite non - commutative law in the large \(n\) limit, we still cannot ensure that these models can simultaneously reach the correct entropy and Wasserstein distance.
### Conclusions
By introducing the concepts of model theory, the author extends the definition of non - commutative probability distributions, thereby solving the difficulties encountered by classical methods in dealing with the large \(n\) limit. This extension not only helps to understand the information geometry in free probability theory but also provides new directions for future research.
### Formula Examples
Some important formulas involved in the paper include:
- **Wasserstein Distance**:
\[
W_2(\mu, \nu)=\inf_{\pi\in\Pi(\mu, \nu)}\left(\int\|x - y\|^2d\pi(x, y)\right)^{1/2}
\]
where \(\Pi(\mu, \nu)\) represents all couplings with joint distributions \(\mu\) and \(\nu\).
- **Free Entropy**:
\[
\chi(\mu)=\limsup_{n\rightarrow\infty}\frac{1}{k_n}\log\lambda(\Gamma(X_1,\dots,X_m:R,\epsilon))
\]
where \(\Gamma(X_1,\dots,X_m:R,\epsilon)\) represents the microstate space and \(\lambda\) represents the Lebesgue measure.
Through these formulas and theoretical frameworks, the author deeply explores the information geometry problems in free probability theory and lays the foundation for further research in the future.