Abstract:We study a diffusion process with random space-time dependent coefficients. Moreover the diffusion matrix is allowed to degenerate. An invariance principle is proved provided that the diffusion coefficient is controlled by a time independent one.
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
This paper aims to study the homogenization problem of diffusion processes (i.e., random flows) with spatio - temporal - dependent random coefficients. Specifically, the author hopes to establish an invariance principle for diffusive particles in the following stochastic differential equation (SDE):
\[ X_t^\omega = x + \int_0^t b(r, X_r^\omega, \omega) \, dr + \int_0^t \sigma(r, X_r^\omega, \omega) \, dB_r \]
where \( B \) is a \( d \)-dimensional Brownian motion, and \( \sigma \) and \( b \) are stationary random fields. In particular, the generator operator \( L^\omega \) coincides with the following on smooth functions:
\[ L^\omega=\frac{1}{2} e^{2V(x, \omega)} \text{div}_x\left(e^{-2V(x, \omega)}[a(t, x, \omega)+H(t, x, \omega)] \nabla_x\right) \]
Here \( a(t, x, \omega)=\sigma \sigma^*(t, x, \omega) \), \( V \) and \( H \) are stationary random fields, with \( V \) being bounded and \( H \) being skew - symmetric.
Furthermore, the author hopes to study the effective diffusion properties of the advection - diffusion equation on the macroscopic scale through these results:
\[ \partial_t z(t, x, \omega)=\frac{1}{2} \text{Trace}[a \Delta_{xx} z](t, x, \omega)+b \cdot \nabla_x z(t, x, \omega) \]
and prove that in the sense of probability, when \( \varepsilon \to 0 \),
\[ \lim_{\varepsilon \to 0} z\left(\frac{t}{\varepsilon^2}, \frac{x}{\varepsilon}, \omega\right)=z(t, x) \]
where \( z \) is the solution of the deterministic equation:
\[ \partial_t z(t, x)=\text{Trace}[A \Delta_{xx} z](t, x) \]
Here \( A \) is a constant matrix, called the effective coefficient matrix.
### Key innovation points
1. **Spatio - temporal - dependent random flows**: This work deals with spatio - temporal - dependent random flows, which is more challenging than previous studies that only considered time - independent or periodic flows.
2. **Degenerate diffusion coefficients**: Allows the diffusion coefficient \( a \) to be degenerate in some directions and can even vanish on a zero - measure subset, but cannot be completely reduced to zero on an open subset.
3. **Ergodicity and regularization properties**: Introduces new ergodicity and regularization conditions, making these assumptions close to the minimum conditions, so that a wider range of problems can be dealt with.
Through these studies, the author hopes to better understand the diffusion behavior in complex random environments and provide theoretical support for applications in related fields.