If $(B_t)_{t≥0}$ is a standard Brownian motion, show that, as $t \to \infty$, $$ \left(\int_0^t e^{B_s} \, ds\right)^{1/\sqrt{t}} \text{ converges in distribution to} \ e^{M_1}, $$ where $M_1 = \sup_{0 \leq s \leq1} B_s$.

I know that $M_1$ has the same distribution as $|N|$ where $N$ $\sim \mathcal{N}(0,1) $. Any hints on how to get started.

  1. Change variables $s=tu$:

$$\int_0^t e^{B_s} \, ds = \int_0^1 e^{B_{t u} }t \, du=(*)$$

  1. Use scaling: $(B_{tu}:u \ge 0) \overset{\mbox{dist}}{=}\sqrt{t}( B_{u}:i \ge 0)$. Therefore

$$ (*) \overset{\text{dist}}{=} \sqrt{t} \int_0^t e^{\sqrt{t} B_u} \, du $$

  1. Take $\sqrt{t}$-th root:

    $$t^{1/\sqrt{t}} \left(\int_0^1 e^{\sqrt{t} B_u} \, du\right)^{1/{\sqrt{t}}}.$$

  2. Take limits.

a. $t^{1/\sqrt{t}}= \left(\sqrt{t}^{1/\sqrt{t}}\right)^2 \to 1$ as $t\to\infty$.

b. For any continuous nonnegative function $f$ on $[0,1]$ we have $\left(\int f^r(u) \, du \right)^{1/r} \to \max_{u \in [0,1]} f(u)$ as $r\to\infty$. This is basic calculus, or also something you know from measure and integration ($L^p$ norm on a probability space increase to $L^\infty$ norm). Here $f$ is the random function $e^{B_u}$ and $r=\sqrt{t}$.

Result follows.


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.