I am trying to prove if the process $M_t = e^{W_t^2-t}$ is a submartingale ($W_t$ is the Wiener Process). The proof becomes a bit difficult, to the point where I am unsure how to move forward.

Let me show you what I have done so far:

\begin{align*} \mathbb E\left[e^{W_t^2-t} \middle| \mathscr{F}_s\right] = & \mathbb E\left[e^{(W_t-W_s+W_s)^2-t} \middle| \mathscr{F}_s\right] = &\\ & \mathbb E\left[e^{(W_t-W_s)^2 + 2W_s(W_t-W_s)+W_s^2-t} \middle| \mathscr{F}_s\right] = \\ & e^{W_s^2-t}\mathbb E(e^{(W_t-W_s)^2})\mathbb E\left[e^{2W_s(W_t-W_s)} \middle| \mathscr{F}_s\right] \end{align*}

And here is where I am unsure about how to solve the different expectations in order to prove that this is indeed a submartingale: I know I am nearly there, but I have tried different things from this point on without succeding. Can somebody perhaps give me a hint?


In the last line, you need $$e^{W_s^2-s}\mathbb E(e^{(W_t-W_s)^2 - (t-s)})\mathbb E\left[e^{2W_s(W_t-W_s)} \middle| \mathscr{F}_s\right] $$

which is equal to $$e^{W_s^2-s}\mathbb E(e^{W_h^2 - h})\mathbb E\left[e^{k W_h} \right]$$ where $k = 2 W_s$ and $h=t-s$. Now you can use Jensen to show that each of the expectations is $\ge 1$.



Let $(M_t)_{t \geq 0}$ be a martingale and $f: \mathbb{R} \to \mathbb{R}$ a convex function. Then it follows directly from Jensen's inequality that $(f(M_t))_{t \geq 0}$ is a submartingale. Apply this to the martingale $$M_t := W_t^2-t$$ and $$f(x) := \exp(x).$$

  • $\begingroup$ Would the downvoter care to comment...? $\endgroup$ – saz Jul 10 '14 at 21:09

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.