1
$\begingroup$

Let $(B_t)_{t \geq 0}$ be a Brownian motion and $T$ a stopping time such that $T < \infty$ a.s. Then the Strong Markov property (SMP) asserts that under the probability measure $\mathbb{P}( \, \cdot \, \vert \, T < \infty)$, the process $$ B_{t}^{(T)} = \mathbb{1}_{\{T < \infty\}} (B_{T+t} - B_T) $$ is a Brownian motion independent of $$ \mathcal{F}_{T} = \{A \in \mathcal{F}_{\infty} \colon A \cap \{T \leq t\} \in \mathcal{F}_t \text{ for all } t \geq 0\}, $$ where $\mathcal{F}_{\infty} = \sigma(B_s \colon s \geq 0)$ and $\mathcal{F}_t = \sigma(B_s \colon 0 \leq s \leq t)$.

Now I have a question to a remark I found in a book: By the (SMP) $$ T = \sup \{s \in [0,1] \colon B_s = 0\} $$ is not a stopping time. Why is this a consequence of the (SMP), how can we formally prove this?

$\endgroup$
1
$\begingroup$

Choose $r \in (0,1)$ such that $\mathbb{P}(T \leq r) \geq 1/2$. Then

$$W_t(\omega) := B_{T+t}(\omega)-\underbrace{B_T(\omega)}_{0} = B_{T+t}(\omega) \neq 0 \qquad \text{for all $t \in [0,1-r]$, $\omega \in \{T \leq r\}$}.$$

This means that $(W_t)_{t \geq 0}$ cannot be a Brownian motion (see below). By the strong Markov property, this implies that $T$ is not a stopping time.

Lemma Let $(B_t)_{t \geq 0}$ be a Brownian motion. With probability $1$ there exists a sequence $(a_n)_{n \in \mathbb{N}} = (a_n(\omega))_{n \in \mathbb{N}}$ such that $B_{a_n}(\omega)=0$ and $a_n \to 0$ as $n \to \infty$.

$\endgroup$
  • $\begingroup$ Thank you for your answer, but I have some additional questions: 1) Why do we know that such $r$ exists in $(0,1)$? $\endgroup$ – user63841219 Aug 14 '17 at 12:49
  • $\begingroup$ @user63841219 Ah sorry, that was stupid... I got your definition of TT wrong. I'm sorry. I've now edited my answer. $\endgroup$ – saz Aug 14 '17 at 14:43
  • $\begingroup$ Thank you, now its clearer. But still I don't see why such $r$ exists and why we need it. $\endgroup$ – user63841219 Aug 14 '17 at 15:14
  • $\begingroup$ @user63841219 Clearly, $T \leq 1$; on the other hand, it cannot hold that $T=1$ almost surely (since this would imply $B_T=B_1 = 0$ almost surely), and therefore there has to exist some $r \in (0,1)$such that $\mathbb{P}(T \leq r) \geq 1/2$. Why we need it: Because we want that $$W_t \neq 0 \quad \text{on some deterministic interval}$$ with probability $>0$. $\endgroup$ – saz Aug 14 '17 at 15:18

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.