Some quenched and annealed limit theorems of superprocesses in random environments

Zeteng Fan,Jieliang Hong,Jie Xiong
2024-06-11
Abstract:Let $X=(X_t, t\geq 0)$ be a superprocess in a random environment described by a Gaussian noise $W=\{W(t,x), t\geq 0, x\in \mathbb{R}^d\}$ white in time and colored in space with correlation kernel $g(x,y)$. When $d\geq 3$, under the condition that the correlation function $g(x,y)$ is bounded above by some appropriate function $\bar{g}(x-y)$, we present the quenched and annealed Strong Law of Large Numbers and the Central Limit Theorems regarding the weighted occupation measure $\int_0^t X_s ds$ as $t\to \infty$.
Probability
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the quenched and annealed limit theorems of superprocesses in a random environment. Specifically, the author focuses on the Strong Law of Large Numbers (SLLN) and the Central Limit Theorem (CLT) of the weighted occupation measure $\int_0^t X_s \, ds$ when the superprocess is in a random environment described by Gaussian noise in high - dimensional space $d \geq 3$. ### Research Background and Problems 1. **Super - Brownian Motion (SBM)**: - The classical SBM is a measure - valued branching process, which emerges as the scaling limit of a critical branching particle system. - In the classical setting, all underlying particles move and branch independently, and their behavior is completely determined by the spatial motion and branching mechanism. 2. **Superprocesses in a Random Environment**: - This paper studies a variant of superprocesses in a random environment proposed by Mytnik. - In this case, the branching probability depends on the random environment, which makes the problem more complicated. - The random environment is described by Gaussian noise $W = \{W(t,x), t \geq 0, x \in \mathbb{R}^d\}$, which is white noise in time and colored noise in space, with a correlation kernel $g(x,y)$. ### Main Results - **Quenched Strong Law of Large Numbers (QSLLN)**: - When $d \geq 3$ and $\alpha > 2$, for almost all random environments $W$, we have \[ \lim_{T \to \infty} \frac{1}{T} Y_T(\varphi)=\langle\lambda, \varphi\rangle, \] where $Y_T(\cdot)=\int_0^T X_t(\cdot) \, dt$, $\lambda$ is the Lebesgue measure, and $\varphi$ is a continuous function with compact support. - **Annealed Strong Law of Large Numbers (ASLLN)**: - For almost all initial conditions, the above result still holds. - **Quenched Central Limit Theorem (QCLT)**: - When $d \geq 5$ and $\alpha > 2$, for almost all random environments $W$, we have \[ \frac{Y_T(\varphi)-\langle\lambda, V_\varphi^1(T)\rangle}{\sqrt{T}}\Rightarrow\sigma(W, \varphi) Z, \] where $Z \sim N(0,1)$ is a standard normal distribution, and $\sigma(W, \varphi)$ is a positive random variable. - **Annealed Central Limit Theorem (ACLT)**: - For almost all initial conditions, the above result still holds. ### Conclusion This paper proves the Strong Law of Large Numbers and the Central Limit Theorem of the weighted occupation measure of superprocesses in high - dimensional space through strict mathematical derivation and analysis. These results not only extend the research on classical super - Brownian motion but also provide new insights into understanding the influence of random environments on branching processes.