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 $A$ be the event that there is some $t$ such that $B_t=1+t^2$, where $B$ is a Brownian mation. Is there any way to compute the probability of $A$, or to approximate it well?

I ask, because we can calculate the probability that $B$ hits a line explicitly. First we change measure with Cameron-Martin, which allows us to reduce to the problem of a Brownian motion hitting a constant. Then we use the reflection principle, which gives the joint distribution of a Brownian motion and its maximum. This approach doesn't seem to generalize to the quadratic.

Thank you.

share|cite|improve this question

Some insights which maybe helpful: let us assume that $F_\alpha(t)$ is a cdf for the first hitting time of constant level $\alpha>0$ for Brownian motion and $\tau$ is a first hitting time of moving boundary $1+t^2$.

Consider $X_t = B_t - t^2$, i.e. $$ dX_t = -2t\cdot dt+ dB_t $$ with $X_0 = 0$. Let us use Girsanov theorem (generalization of Cameron-Martin one) and put $\theta_s = 2s$, so
$$ Z_t = \exp\left\{-\frac23t^3+2\int\limits_0^ts\cdot dB_s\right\} $$ will be a likelihood ratio process leading to the fact that for any finite $T$ the process $(X_t)_{0\leq t\leq T }$ is a Brownian motion under new measure ${\mathsf Q}(\cdot) = \mathsf E[Z_T 1(\cdot)]$. As a results we have: $$ \mathsf Q\{\tau\leq T\} = \mathsf Q\left\{\sup\limits_{t\leq T}X_t\geq1\right\} =F_1(T) $$ for any finite $T$ since $X_t$ is a Brownian motion under $\mathsf Q$. If we put $\xi_t = \frac{1}{Z_t} = \exp\left\{-\frac23 t^3-2\int\limits_0^ts \,dX_s\right\}$ then $$ \mathsf P\{\tau\leq T\} = \tilde{\mathsf E}\left[\xi_T1\{\tau\leq T\}\right]. $$ As a rough upper-bound we can apply Hölder's inequality to get: $$ \mathsf P\{\tau\leq T\}\leq \sqrt{\tilde{\mathsf E}[\xi_T^2]}\sqrt{F_1(T)} $$ where $\tilde{\mathsf E}[\xi^2_t] = \exp\{4/3\cdot t^3\}$ so you have $\mathsf P\{\tau\leq T\}\leq F_1(T)\exp\{4/3\cdot T^3\}$ which are pretty rough.

On the other hand you can use different method, namely find $f(a,t)$ such that $$ \exp\{aX_t + f(a,t)\} $$ is a martingale for any $a$, then by applying Ito formula you obtain $f(a,t) = at^2-\frac12a^2t$. Then $$ \mathsf E[\exp\{aX_{\tau\wedge T} + f(a,\tau\wedge T)\}] = \mathsf E[\exp\{aX_0 + f(a,0)\}] = 1 $$ for any $T<\infty$, since $\tau = \infty$ with non-negative probability.

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.