4
$\begingroup$

Let $B_t$ be a 2D Brownian Motion with $B_0 = (1,0)$. Now, express $B_t$ in polars, that is, $B_t = (r(t), \theta(t))$. Let $\tau = \inf\{t > 0 : \theta(t) \geq 2 \pi \}$. What is $\mathbb{P}[\tau \leq 1]$?

In other words, what is the probability that the winding number is at least one in a time less than one?

$\endgroup$

2 Answers 2

4
$\begingroup$

From the skew-product decomposition of the planar Brownian motion (see for instance Revuz-Yor), it is known that $\theta_t=\beta_{C_t}$ where $C_t=\int_0^t \frac{ds}{\rho_s^2}$ with $\rho_s=\| B_s \|$ and $(\beta_t)_{t \ge 0}$ is a Brownian motion independent from $(\rho_t)_{t \ge 0}$. Consequently we have

$\mathbb{P}(\tau \le 1)=\mathbb{P}(\sup_{t \in [0,1]} \theta_t \ge 2\pi)=\mathbb{P}( \sup_{t \in [0,1]} \beta_{C_t} \ge 2\pi)=\mathbb{P}( \sup_{t \in [0,C_1]} \beta_{t} \ge 2\pi)$ $=\int_0^{\infty}\mathbb{P}( \sup_{t \in [0,\tau]} \beta_{t} \ge 2\pi)p(\tau)d\tau$

where $p(\tau)$ is the density of $C_1$.

The distribution of $\sup_{t \in [0,\tau]} \beta_{t}$ is known and the distribution of $C_1$ has been studied in great details in

MR0576898 Yor, Marc Loi de l'indice du lacet brownien, et distribution de Hartman-Watson. (French) Z. Wahrsch. Verw. Gebiete 53 (1980), no. 1, 71–95.

Putting the pieces together, you will get an integral expression involving the Hartman-Watson distribution.

$\endgroup$
3
$\begingroup$

These two earlier MO questions should shed light on your question:

The former question was answered by Andreas Rüdinger with several relevant references, and this summary: "the root mean square winding number grows logarithmically" with the number of steps.

$\endgroup$
1
  • 1
    $\begingroup$ Yeah, but it looks like the only light that they shed is that the answer is a number strictly between $0$ and $1$, which is almost obvious from the start. This question is about the exact value, not about the order of magnitude. $\endgroup$
    – fedja
    Oct 23, 2013 at 23:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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