I have a question on asymptotic behavior of distributions of Brownian hitting times.

Let $B_t$ and $W_t$ be independent one-dimensional Brownian motions starting at the origin. The joint law is denoted by $P$. For $x,y>0$, we set \begin{align*} \sigma_x&=\inf\{t>0 \mid B_t=x\} \\ \tau_y&=\inf\{t>0 \mid W_t \notin (-y,y)\}. \end{align*} Then, we can show that \begin{align*} (1)\quad P(\sigma_r>\tau_{r^{1/2}})=O(r^{1-\varepsilon})\quad \text{as}\quad r \to 0, \end{align*} for any $\varepsilon>1/2.$

For the proof of $(1)$, I used the fact that $P(\sigma_r>t) \lesssim r/\sqrt{t}$ and $P(\tau_{r^{1/2}} \le t) \lesssim \sqrt{t/r}\exp(-r/8t)$ valid for $r,t>0$.

Is my estimate optimistic? I think there is a sharper bound.

  • $\begingroup$ Although I think you meant optimal, your estimate indeed seems optimistic, or even super-optimistic: unless I got something wrong, by scaling, we have $P(\sigma_r > \tau_{r^{1/2}}) = P(\sigma_{r^{1/2}} > \tau_1) \to 1$ as $r \to 0^+$. $\endgroup$ – Mateusz Kwaśnicki Oct 29 '19 at 9:08
  • $\begingroup$ Thank you for your comment. I may have misunderstood something. Why $\lim_{r \to 0}P(\sigma_{r^1/2}>\tau_1) =1$ ? There is the exact formula for the distribution of $\sigma_{r^1/2}$, which means that $P(\sigma_{r^{1/2}}>t) \lesssim (r/t)^{1/2}$. So, as $r \to 0$, $P(\sigma_{r^{1/2}}>t) \to 0$. $\endgroup$ – sharpe Oct 29 '19 at 10:39
  • $\begingroup$ Of course, you are right, sorry. For some reason I was thinking about the limit as $r \to \infty$. But then, it is relatively easy to see that $P(\sigma_{r^{1/2}} > \tau_1)$ is comparable with $r^{1/2}$: this is the value $u(0, r^{1/2})$ of a harmonic function $u$ in $(-1,1) \times (0, \infty)$, with boundary value $0$ on $(-1,1) \times \{0\}$ and $1$ on $\{-1,1\} \times (0, \infty)$. By the boundary Harnack inequality, the decay of $u$ is linear near $u$. In fact, no need to invoke BHI here: extend $u$ so that $u(x,-y)=-u(x,y)$, and observe that $u$ is harmonic in $(-1,1)\times \mathbb{R}$. $\endgroup$ – Mateusz Kwaśnicki Oct 29 '19 at 11:01
  • $\begingroup$ @MateuszKwaśnicki Thank you for your kind reply. I have a question on your comment. We denote by $P_{x,y}$ the joint distribution of independent one-dimensional Brownian motions $B$ and $W$ starting at $(x,y)$. I think $u(x,y)=P_{x,y}(\sigma_{r^{1/2}}>\tau_1)$. Then, $u$ is a harmonic function w.r.t. the two-dimensional Brownian motion $(B,W)$, right? How do you prove the decay of $u$ is linear near $(0,0)$? $\endgroup$ – sharpe Oct 29 '19 at 15:54
  • $\begingroup$ @MateuszKwaśnicki AH... I might understood. The identity $u(x,y)=-u(x,y)$ leads us to the lenearly decay. $\endgroup$ – sharpe Oct 29 '19 at 16:06

OK, here is the extended version of my comment.

Let $u(x,y)$ be the (bounded) harmonic function in $D = (-1,1) \times (0, \infty)$, with boundary value $1$ along $\{-1,1\} \times (0, \infty)$ and $0$ along $(-1,1) \times \{0\}$. Consider a 2-D Brownian motion $(X_t, Y_t)$, started at $(X_0,Y_0) = (x,y)$, and denote the corresponding probability law by $P^{x,y}$. Either by Itô's lemma, or by potential-theoretic results, $u(X_t,Y_t)$ is a martingale up to $\tau_D$, the first time $(X_t,Y_t)$ hits the boundary of $D$. In particular, $$u(x,y) = E^{x,y} u(X_{\tau_D},Y_{\tau_D}).$$ At $t = \tau_D$, we either have $X_t = \pm 1$ or $Y_t = 0$. Let $\sigma$ be the first time $Y_t = 0$, and $\tau$ be the first time $X_t = \pm 1$. Clearly, $\tau_D = \min\{\sigma, \tau\}$. If $\sigma < \tau$, then $\tau_D = \sigma$ and $Y_{\tau_D} = 0$. Thus, $u(X_{\tau_D},Y_{\tau_D}) = 0$. On the other hand, if $\sigma > \tau$, then $X_{\tau_D} = \pm 1$, and hence $u(X_{\tau_D},Y_{\tau_D}) = 1$. It follows that $$u(x,y) = P^{x,y}(\sigma > \tau).$$ By symmetry and translation invariance, with the notation of the problem, $$u(0,y) = P(\sigma_y > \tau_1).$$ By self-similarity (a.k.a. scaling), $$u(0,\sqrt{r}) = P(\sigma_{\sqrt{r}} > \tau_1) = P(\sigma_r > \tau_{\sqrt{r}}).$$

By the boundary Harnack inequality, we know that a finite, positive limit $$ \lim_{y \to 0^+} \frac{u(0, y)}{y} $$ exists. This implies that $$P(\sigma_r > \tau_{\sqrt{r}}) \sim c \sqrt{r}$$ as $r \to 0^+$.

As a side remark: In fact, there is no need to employ BHI here, as soon as one observes that the formula $u(x,-y) = -u(x,y)$ extends $u$ to a harmonic function in $(-1, 1) \times \mathbb{R}$. It is then clear that $u(0,y)$ is a smooth function of $y$, and it remains to show that $\partial_y u(0, y) > 0$. This last stem follows for example from the explicit expression for the Poisson kernel of a strip; however, I think one can easily cook up a soft argument here.

By the way, Hopf's lemma provides a yet another proof of the asymptotic formula for $u(0, y)$.

  • $\begingroup$ Thank you for your very kind reply. But I have one question. Sorry for my lack of study. In your answer, you first take the harmonic function $u$ on $D$ with the following boundary condition; $u=0$ on $(-1,1) \times \{0\}$ and $u=1$ on $\{-1,1\} \times (0,\infty)$. Why $u$ exists ? This may be a basic question. $\endgroup$ – sharpe Oct 30 '19 at 5:32
  • $\begingroup$ I can prove the uniqueness of $u$. This follows from the maximum principle. $\endgroup$ – sharpe Oct 30 '19 at 5:35
  • $\begingroup$ Well, that depends on your point of view. From my probabilistic perspective, you can simply define $u$ as $u(x,y) = P^{x,y}(\sigma > \tau)$. There are, however, alternative analytical arguments. For example, you can prove existence of a solution to the Dirichlet problem with continuous data and then extend it to piecewise continuous boundary values. Or, you can take any smooth function with these boundary values, and solve a Poisson problem instead. $\endgroup$ – Mateusz Kwaśnicki Oct 30 '19 at 6:04
  • $\begingroup$ Thank you for your reply. I understood. An answer of my question follows from some results of the Dirichlet problem., $\endgroup$ – sharpe Oct 30 '19 at 6:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.