Consider the drift Brownian motion $X_t:=1+bt+W_t$, where $(W_t)_{t\ge 0}$ is a Brownian motion starting at zero. Set $\tau:=\inf\{t\ge 0: X_t=0\}$. Assume $b>0$, then $\mathbb P[\tau=\infty]>0$. What is the conditional law of $X_{\infty}$ knowing $\tau=\infty$?


1 Answer 1


By the law of Large numbers, $X_t/t \to b$ almost surely as $t \to +\infty$, hence $X_t \to +\infty$ almost surely as $t \to +\infty$. Therefore $X_\infty = +\infty$ almost surely under $P$ and also under $P[\cdot|\tau = \infty]$.

  • $\begingroup$ I see that $X_\infty = +\infty$ almost surely under $P$ — but I don’t think that implies much about $P[\cdot|\tau = \infty]$ since $\tau=\infty$ is a measure-0 condition. $\endgroup$
    – Matt F.
    Jun 23 at 16:22
  • 2
    $\begingroup$ @Matt F. You forget the drift! Brownian motion with constant drift is transient. $\endgroup$ Jun 23 at 16:48
  • 1
    $\begingroup$ @GJC20 The density and the distribution function of $X_t$ under $P$ and under $P[\cdot|\tau>t] $converge pointwise to 0 as $t \to +\infty$. $\endgroup$ Jun 23 at 16:52
  • 1
    $\begingroup$ @GJC20 : As Christophe Leuridan noted, the conditional density of $X_t$ given $\tau>t$ will go to $0$ pointwise as $t\to\infty$. However, it is not hard to find a simple expression for the conditional density of the standardized version $Z_t:=(X_t-EX_t)/\sqrt{Var\,X_t}$ given $\tau>t$ and show that it will go to the standard normal density pointwise as $t\to\infty$. $\endgroup$ Jun 23 at 20:57
  • 1
    $\begingroup$ Previous comment continues: At least intuitively, this should be clear, as $X_t$ for large $t$ will "forget" whether the process previously reached $0$ or not. So, for large $t$, the conditional distribution of $X_t$ given $\tau>t$ will be close to the unconditional distribution of $X_t$. $\endgroup$ Jun 23 at 20:58

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.