Abstract:It is known that a properly rescaled version of Sinai's random walk converges in distribution to Brox's diffusion. In this article we quantify this convergence by considering a specific coupling between Sinai's walk and Brox's diffusion. Our method relies on convergence results for martingale problems considered in the rough path setting.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to quantify the convergence between Sinai random walks and Brox diffusions. Specifically, the paper focuses on how to quantify the convergence rate between Sinai random walks in discrete time (appropriately scaled) and Brox diffusions in continuous time under a specific coupling.
### Problem Background
1. **Sinai Random Walks**
- A Sinai random walk is a random walk in a random environment. Its state space is the set of integers \( \mathbb{Z} \), and the environment is given by a set of independent and identically distributed random variables \( \{\omega^+_x; x \in \mathbb{Z}\} \), which represent the probability of jumping to the right at each position \( x \).
- Under certain assumptions (such as the recurrence assumption), Sinai random walks exhibit highly nontrivial asymptotic behavior.
2. **Brox Diffusions**
- A Brox diffusion is an evolution process of Brownian motion in a Brownian environment. It is a continuous - time process, and the environment is given by a two - sided Brownian motion \( W(x) \).
- The Brox diffusion satisfies a stochastic differential equation with a distribution - type drift term:
\[
X^c_t = -\frac{1}{2} \int_0^t \dot{W}(X^c_s) \, ds + B_t,
\]
where \( B_t \) is a standard Brownian motion.
### Core Problem of the Paper
The main contribution of the paper is to provide a specific coupling method and quantify the convergence rate between Sinai random walks \( X^\delta \) and Brox diffusions \( X^c \) through this method. Specifically:
- **Theorem 1.3**: For any \( t \in [0, T] \) and \( \delta \in (0, 1) \), there exists a coupling \( (X^\delta, X^c) \) such that for any function \( h \in C_b^3 \), we have
\[
\left| E_\omega[h(X^c_t)] - E_\omega[h(X^\delta_t)] \right| \leq C_{h,T}(\omega) \delta^{\frac{1}{17}},
\]
where \( C_{h,T}(\omega) \) is a random constant that only depends on \( h \), \( T \) and the environment \( \omega \).
### Key Points of the Solution
1. **Rough Path Analysis**: The paper uses rough path theory to deal with the coupling problem between discrete and continuous processes. In particular, by considering the weak solutions of discrete PDEs and the corresponding martingale problems, the paper provides detailed convergence estimates.
2. **Extension of Donsker - type Theorems**: Traditional Donsker - type theorems only provide results on distributional convergence, while this paper further quantifies the speed of this convergence.
3. **Path Coupling**: Through the Komlos - Major - Tusnady coupling theorem, the paper couples the discrete increments \( \dot{U}^\delta \) and the continuous environment \( \dot{W} \) together, thus achieving the coupling of Sinai random walks and Brox diffusions.
### Summary
The main purpose of this paper is to fill the gap in quantitative research in this field by introducing a new coupling method to quantify the convergence rate between Sinai random walks and Brox diffusions. This not only deepens our understanding of the relationship between these two stochastic processes but also provides a powerful tool for further research.