1
$\begingroup$

I know the path of Brownian mothion is continuous in probability if and only if ,for every $\delta>0$ and $t>0$, $$\lim_{\Delta t\to 0}P(|B(t+\Delta t)-B(t)|\ge \delta)=0$$ I can't continue it . I have seen some proofs of this theorem in this site but I want to prove it by definition.

So thanks

$\endgroup$
5
$\begingroup$

Since $\frac{B(t+\Delta t) -B(t)}{\sqrt{\Delta t}} \sim N(0, 1)$, then \begin{align*} &\ P(|B(t+\Delta t) -B(t)|\ge \delta) \\ =&\ P\left(\frac{B(t+\Delta t) -B(t)}{\sqrt{\Delta t}}\ge \frac{\delta}{\sqrt{\Delta t}}\right) + P\left(\frac{B(t+\Delta t) -B(t)}{\sqrt{\Delta t}}\le \frac{-\delta}{\sqrt{\Delta t}}\right)\\ =&\ 1-\Phi\left(\frac{\delta }{\sqrt{\Delta t}}\right) + \Phi\left(\frac{-\delta }{\sqrt{\Delta t}}\right), \end{align*} where $\Phi$ is the cumulative distribution function of a standard normal random vaiable. It is now obvious that \begin{align*} \lim_{\Delta t \rightarrow 0}P(|B(t+\Delta t) -B(t)|\ge \delta) = 0. \end{align*}

$\endgroup$
  • $\begingroup$ Can you please explain it $\endgroup$ – Crisis2008 Aug 19 '16 at 13:19
  • $\begingroup$ @Crisis2008: see updates. $\endgroup$ – Gordon Aug 19 '16 at 13:24
  • $\begingroup$ Nice answer (+1 ) $\endgroup$ – Behrouz Maleki Aug 19 '16 at 13:43
  • $\begingroup$ So Thanks Gordon $\endgroup$ – Crisis2008 Aug 19 '16 at 13:48
1
$\begingroup$

Other way

We know $B_t$ is martingale thus $$\lim_{\Delta t\to 0}\mathbb{P}(|B_{t+\Delta t}-B_{t}|\ge \delta)=\lim_{\Delta t\to 0}\mathbb{P}(|B_{t+\Delta t}-\mathbb{E}[B_{t+\Delta t}|\mathcal{F}_t]\,|\ge \delta)$$ By application of Chebyshev's inequality, we have $$\mathbb{P}(|B_{t+\Delta t}-B_{t}|\ge \delta)\le\frac{\operatorname{Var}(B_{t+\Delta t}\,|\mathcal{F}_t)}{\delta^2}\\ \quad\qquad\qquad\qquad\qquad\qquad\qquad=\frac{\operatorname{Var}(B_{t+\Delta t}+B_t-B_t\,|\mathcal{F}_t)}{\delta^2}\\ \quad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\,\,\,=\frac{\operatorname{Var}(B_{t+\Delta t}-B_t\,|\mathcal{F}_t)}{\delta^2}+\frac{\operatorname{Var}(B_t\,|\mathcal{F}_t)}{\delta^2}\\ \quad\qquad\quad=\frac{\Delta t}{\delta ^2}$$ since Brownian motion has independent increments .On the other hand $B_t$ is $\mathcal{F}_t$ measurable , hence $\operatorname{Var}(B_t\,|\mathcal{F}_t)=0$. As a result $$\lim_{\Delta t\to 0}\mathbb{P}(|B_{t+\Delta t}-B_{t}|\ge \delta)=0$$

$\endgroup$

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.