I am reading through Le Gall's book on Brownian Motion, Martingales, and Stochastic Calculus. I just read through the chapter on optional stopping of martingales, but I cannot solve the first exercise:

  1. Let $M$ be a martingale on a probability space with a complete filtration such that $M$ has continuous stopping times, and $M_0 = x \in \mathbb{R}_+$. We assume $M_t \geq 0$ for every $t \geq 0$, and that $M_t \to 0$ when $t \to \infty \; $ a.s. Show that, for every $y > x,$ $$P(\sup_{t \geq 0} M_t \geq y) = \frac{x}{y}$$

  2. Give the law of $\sup_{t \leq T_0} B_t$, when $B$ is a Brownian motion started from $x > 0$ and $T_0 = \inf \{t \geq 0 : B_t = 0 \}$

  3. Assume now that $B$ is a Brownian motion started from $0$, and let $\mu > 0$. Using an appropriate exponential martingale, show that $\sup_{t\geq 0}(B_t - \mu t)$ is exponentially distributed with parameter $2\mu$.

Now, I am able to upper bound the probability $P(\sup_{t \geq 0} M_t \geq y) \leq \frac{x}{y}$ by using one of the Doob martingale inequalities. However, I am not sure how to finish off the proof of 1. I am assuming that I would have to use optional stopping, but I cannot make a useful stopping time.

  • $\begingroup$ If your question concers only the first part of the exercise, then why post the other two unrelated parts of the exercise? The first part is solved here $\endgroup$
    – saz
    Apr 15 '19 at 17:28
  1. Let $T_y=\inf\left\{t\ge0\;;\; M_t\ge y\right\}$, we have : $$ \mathbb{E}\left[M_0\right] = \mathbb{E}\left[M_{T_y}.\mathbf{1}_{\left\{T_y<\infty\right\}} + M_\infty.\mathbf{1}_{\left\{T_y=\infty\right\}}\right] $$ but $M_\infty=0$, so $$ x = y . \mathbb{E}\left[\mathbf{1}_{\left\{T_y<\infty\right\}}\right] = y . \mathbb{P}\left(T_y<\infty\right) = y . \mathbb{P}\left(\sup_{t\ge 0} M_t\ge y\right) $$ i.e. $$ \mathbb{P}\left(\sup_{t\ge 0} M_t\ge y\right) = \frac{x}{y} $$

  2. For a standard Brownian Motion, the reflexion principle gives $$ \mathbb {P} \left(\sup _{0\leq s\leq t}W(s)\geq a\right) = 2\mathbb {P} \left(W(t)\geq a\right) $$ from which you can get the density probability function by derivating : $$ f_{\sup _{0\leq s\leq t}W(s)}(a) = \sqrt{\frac{2}{\pi t}} \mathbf{e}^{-\frac{a^2}{2t}}. $$ Now, $B = x + W$ so for $a\ge x$ $$ f_{\sup _{0\leq s\leq T_0}B(s)}(a) = \sqrt{\frac{2}{\pi T_0}} \mathbf{e}^{-\frac{(a-x)^2}{2T_0}}. $$

  3. We need to prove that $\mathbb{P}\left(\sup_{t\ge 0} B(t)-\mu t \geq y\right) = \mathbf{e}^{-2\mu y}$. But \begin{alignat*}{2} \mathbb{P}\left(\sup_{t\ge 0} B(t)-\mu t \geq y\right) & = \mathbb{P}\left(\sup_{t\ge 0} \left(2\mu B(t)-\frac{1}{2}4\mu^2 t \right) \geq 2\mu y\right) \\ & = \mathbb{P}\left(\sup_{t\ge 0} \mathbf{e}^{2\mu B(t)-\frac{1}{2}4\mu^2 t }\geq \mathbf{e}^{2\mu y}\right), \end{alignat*}

and $\left(\mathbf{e}^{2\mu B(t)-\frac{1}{2}4\mu^2 t}\right)_{t\ge 0}$ is a martingale that verifies the conditions in question 1. Thus $$ \mathbb{P}\left(\sup_{t\ge 0} \mathbf{e}^{2\mu B(t)-\frac{1}{2}4\mu^2 t }\geq \mathbf{e}^{2\mu y}\right) = \mathbf{e}^{-2\mu y} $$ which is the result.


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.