I'm looking through the proof on Brownian motion with drift carried out by my professor and I can't get my head around a passage.

Suppose I have a Brownian motion with drift $\mu$ and an absorbing barrier in $a$, that is

$$\bar{B}^\mu(t)=\left\{\begin{matrix}B^\mu(t) &\text{ for t} <T_a\\ a &\text{ for }t\geq T_a \end{matrix}\right.$$

where $T_a = \text{inf}\{t:B^\mu(t)\geq a\}$.

If I want to find the cdf of such process, we can see that, for $y<a,$

$P(\bar{B}^\mu(t) \leq y) = P(B^\mu(t) \leq y, T_a > t)=P(B^\mu(t) \leq y) - P(B^\mu(t) \leq y, T_a \leq t)$.

The first part of the last difference can be found with the transition probability of a free Brownian motion with a drift. For the second part, I can say that

$P(B^\mu(t) \leq y, T_a \leq t) = \mathbb{E}[I_{B^\mu(t) \leq y},I_{T_a \leq t}] = \mathbb{E}[\mathbb{E}[I_{B^\mu(t) \leq y},I_{T_a \leq t}|\mathcal{F}_{T_a}]]$

where $I$ is the indicator function and $\mathcal{F}_{T_a}$ is the sigma algebra stopped at the random time $T_a$. Since $T_a$ is a stopping time, the indicator of the event $T_a < t$ is $\mathcal{F}_{T_a}$-measurable and I can take the indicator random variable out of the conditional expected value.

$P(B^\mu(t) \leq y, T_a \leq t) = \mathbb{E}[I_{T_a \leq t}\mathbb{E}[I_{B^\mu(t) \leq y)}|\mathcal{F}_{T_a}]]$

So far so good. The problem comes with the remaining conditional expectation. If I take it, I can rewrite it as

$\mathbb{E}[I_{B^\mu(t) \leq y}|\mathcal{F}_{T_a}] = P(B^\mu(t) \leq y|\mathcal{F}_{T_a})$

and this is where the problem starts. The next passage would be to say that since Brownian motion (even with a drift) is a Markov process, then conditioning on the $\sigma$-algebra $\mathcal{F}_{T_a}$ is the same as conditioning on the position of a standard Brownian motion at random time $T_a$.

I guess that not understanding this passage shows I have not fully grasped what is contained inside $\mathcal{F}_{T_a}$ in this case.

I know that if $T_a$ is a stopping time with respect to the filtration $\{\mathcal{F}_t,t\geq 0\}$ on the probability space $(\Omega, \mathcal{F},P)$ then

$$\mathcal{F}_{T_a}=\{A\in\mathcal{F}:A \cap (T_a<t) \in \mathcal{F}_t \text{ }\forall t\}$$

So does $\mathcal{F}_{T_a}$ contain the positions of all the possible trajectories of the Brownian motion that have first passage time through $a$ before time $t$?

If this is the case I'm a happy man. Otherwise, could somebody clear up my confused mind? Thank you.


Your Answer

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

Browse other questions tagged or ask your own question.