Let $\{W_t\}_{t\geq 0}$ be a standard Brownian motion (starting at $0$). For $T$ large enough, I would like to prove that $P(\max_{t\in[0,T]} |W_t| \leq c T^{1/3})$ is bigger than a negative power of $T$ (like $a T^{-1/6}$ for example, for an appropriate constant $a > 0$) or at least bigger than $e^{-b T^{1/3}}$ if $b$ can be made arbitrarily close to $0$.

We know that without the absolute value, i.e. $P(\max_{t\in[0,T]} W_t \leq c T^{1/3})$, this property is true because $\max_{t\in [0,T]} W_t$ has the same law as $|W_T|$.

So the result should be true intuitively because keeping $|W_t|$ in a cylinder should not be much less probable than restricting a standard Brownian motion to be lower than a bound of the same order. I've tried to prove this result without success.

Thanks for any help.

  • $\begingroup$ Why do you think that this holds for any $T>0$? For small $T$, we have $T^{-1/6} \gg 1$, so we cannot expect to bound $\mathbb{P}(\dots)$ from below by $T^{-1/6}$. $\endgroup$ – saz Feb 28 '15 at 17:11
  • $\begingroup$ Sorry, the problem is meant when $T$ is big. I corrected it. $\endgroup$ – Frédéric Ouimet Feb 28 '15 at 17:12

We have a series expansion for the wanted quantity, namely, $$\mathbb P\left(\max_{t\in[0,T]} |W_t| \leqslant c T^{1/3}\right)=\frac 4{\pi }\sum_{n=0}^{+ \infty } \frac{(-1)^n}{2n+1}\exp\left(-\frac{(2n+1)^2\pi^2}{8c^2}T^{1/3} \right) $$ holds (see Feller's An introduction to probability theory (1970) page 342).


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.