Increment property of Brownian Motion

86 Views Asked by At

I'm trying to prove an increment property for the Brownian motion, but I'm unable to figure it out, maybe someone can help me out.

Consider two sequences $(a_n)_{n \in \mathbb{N}}$, $(b_n)_{n \in \mathbb{N}}$ of positive real numbers and let $\mathbb{B}$ be a standard Brownian motion on a probability space $(\Omega, \mathcal{A}, \mathbb{P})$.

The statement I'm trying to prove is the following.

Assume that $a_n - b_n = o(n)$ as $(n\to\infty)$. Or in other words $\lim_{n \to \infty} \frac{\vert a_n - b_n\vert}{n} = 0$, then already $\max_{1 \leq k \leq n} \vert \mathbb{B}(a_k) - \mathbb{B}(b_k) \vert = o\left(\sqrt{n \log(n)}\right)$. Or equivalently

$\lim_{n \to \infty} \frac{\max_{1 \leq k \leq n} \vert \mathbb{B}(a_k) - \mathbb{B}(b_k) \vert}{\sqrt{n \log(n)}} = 0$

I initially thought that I might have to use a result like Levy's modulus of continuity for large increments (Something like this), but I still wasn't able to figure it out. I'm thankful for any hints.

1

There are 1 best solutions below

0
On BEST ANSWER

Let $t_k=|b_k-a_k|$ and $\epsilon_k=\sqrt{2t_k/k}$. We are given that $\epsilon_k \to 0$ as $k \to \infty$. Also, denote $\psi(k)=\sqrt{2k \log k}$. Then by the tail bound for a standard Normal variable, $P(|Z|>r) \le e^{-r^2/2}$, applied to $Z_k=[B(a_k)-B(b_k)]t_k^{-1/2}$, we have $$P\bigl(|B(a_k)-B(b_k)|>\epsilon_k \psi(k)\bigr) \le \exp\Bigl(\frac{-\epsilon_k^2 \psi^2(k)}{2t_k}\Bigr)=k^{-2}\,.$$

By Borel Cantelli, with probability 1 the inequality $$|B(a_k)-B(b_k)| \le \epsilon_k \psi(k)$$ holds for all sufficiently large $k$, so $$P\Bigl(\lim_{k \to\infty} \frac{|B(a_k)-B(b_k)|}{\psi(k)} =0\Bigr)=1 \,, $$ whence $$P\Bigl(\lim_{n \to\infty} \max_{k \in [1,n]}\frac{|B(a_k)-B(b_k)|}{\psi(n)} =0\Bigr)=1 \,. $$