Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $\tau_a=\inf\{t: B_t=a\}$, the hitting time of the standard Brownian motion to reach the boundary $a$.

This is easily derived

$$E(e^{-\lambda \tau_a})=e^{-|a|\sqrt{2\lambda}}$$

But I am having a problem of using this formula to get the moments of $\tau_a$ by matching the coefficients of terms for each power of $\lambda$, specifically, the LHS, if expanded, has all integer powers of $\lambda$; however, the RHS has the various terms of $\sqrt{\lambda}$. How to reconcile the difference here? Could somebody please help? Thanks.

share|cite|improve this question
up vote 6 down vote accepted

Well, in fact $E(\tau_a)$ is infinite...

A classic argument to see that $\tau_a$ does not have finite mean is to consider $\tau_{2a}$. On the one hand, to hit $2a$, one has to hit $a$ and then, starting from $a$, to hit $2a$. By the strong Markov property, this second duration is independent of $\tau_a$ and, by invariance by translations, distributed like $\tau_a$. On the other hand, by the scale invariance of Brownian motion, $(2B_t)$ is distributed like $(B_{4t})$, hence $\tau_{2a}$ is distributed like $4\tau_a$. Finally, $\tau_a$ solves a distribution equality: the distribution of $T+T'$ and the distribution of $4T$ coincide, where $T$ and $T'$ are independent copies of $\tau_a$. This is enough to prove that $\tau_a$ is not integrable (and almost enough to deduce its Laplace transform, for that matter).

share|cite|improve this answer
right!... forgot to see this. So higher moments will also be infinite then! there is no point to do the matching. correct? :) – Qiang Li Feb 22 '11 at 21:41

To add on Didier's answer, note that the PDF of $\tau_a$, for $a > 0$, is given by $$ f_{\tau _a } (t) = \frac{a}{{\sqrt {2\pi } }}t^{ - 3/2} e^{ - a^2 /(2t)} ,\;\; t > 0. $$ From this you immediately see that ${\rm E}(\tau_a) = \infty$.

EDIT: The distribution function of $\tau_a$ can be derived as follows (cf. p.1 here). First note that, for any $t > 0$, $$ {\rm P}(\tau _a < t) = {\rm P}(\tau _a < t,B_t < a) + {\rm P}(\tau _a < t,B_t > a). $$ The right-hand side is equal to $2{\rm P}(\tau _a < t,B_t > a)$, hence to $2{\rm P}(B_t > a)$. Thus, $$ {\rm P}(\tau _a < t) = 2 {\rm P}(B_t > a) = \frac{2}{{\sqrt {2\pi t} }}\int_a^\infty {e^{ - \frac{{y^2 }}{{2t}}} \,{\rm d}u} = \frac{2}{{\sqrt \pi }}\int_{\frac{a}{{\sqrt {2t} }}}^\infty {e^{ - y^2 } \,{\rm d}y}. $$ Differentiating with respect to $t$ gives the above formula for $f_{\tau _a } (t)$. (For arbitrary $a \in \mathbb{R}$, just put $|a|$ instead of $a$ in that formula.)

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.