5
$\begingroup$

Note: We denote by $\mathcal L(U)$ the Lebesgue measure of a set $U$.

Let $\mu: \mathbb R^d \to \mathbb R^d$ and $\sigma: \mathbb R^{d} \to \mathbb R^{d \times d}$ be Borel functions.

Suppose the following assumptions hold (Assumption A):

  • $\mu$ is Lipschitz continuous with Lipschitz constant $L > 0$.

  • $\lvert \mu(0) \rvert < M$ for some constant $M > 0$.

  • $\sigma$ is locally Lipschitz continuous.

  • For some constant $C > 0$, $\xi^{T} \sigma(x) \sigma(x)^{T} \xi \geq C |\xi|^2$ for all $x, \xi \in \mathbb R^d$.

  • $\sup_{ij} |\sigma_{ij}| < D$ for some constant $D > 0$.

Let $W$ be a standard d-dimensional Brownian motion, and let $X$ be the solution to the SDE

$$dX_t = \mu(X_t) \,dt + \sigma(X_t) \,dW_t$$

with initial condition $X_0 = 0$.

Question: Fix $T > 0$. For every $\varepsilon, h > 0$, does there exist a $\delta > 0$ depending only on $\varepsilon, h, L, C, D$ and $M$ satisfying the following property?

For all $\mu, \sigma$ satisfying Assumption A, and all Borel subsets $U$ of $\mathbb R^d \setminus \{0\}$ with $\mathcal L(U) < \delta$, we have

$$\mathbb P\left(\int_{0}^T \mathbf 1_{U} (X_s) ds > h\right) < \varepsilon.$$

$\endgroup$
5
  • $\begingroup$ I think $dX_t = X_tdW_t$ is a perfectly good diffusion started from any point, but it sticks if you start it from 0. $\endgroup$
    – mike
    Commented Mar 12, 2022 at 13:50
  • $\begingroup$ Ah.. I guess I need either $\mu$ or $\sigma$ bounded from below. Hope you’ll allow me to modify the post a little, thanks! $\endgroup$
    – Nate River
    Commented Mar 12, 2022 at 15:12
  • $\begingroup$ @mike I guess for what I want to do with it, its okay if it gets stuck at $0$, so I have modified the condition on $U$ instead. $\endgroup$
    – Nate River
    Commented Mar 12, 2022 at 15:18
  • 2
    $\begingroup$ If you're happy to also impose a uniform bound on $\mu(0)$ and on a global Lipschitz constant for $\sigma$, then the claim is true and not too hard to prove. It's probably even true as stated (and not hard to prove in $d=1$), but the proof I'm thinking of is analytic and would break down. $\endgroup$ Commented Mar 18, 2022 at 17:03
  • $\begingroup$ @Martin Hairer Ah yes, we can have a uniform bound on $\mu(0)$. That $\sigma$ is only locally Lipschitz needs to hold though. $\endgroup$
    – Nate River
    Commented Mar 19, 2022 at 11:29

0

You must log in to answer this question.