Assuming that $\{W(t) | t \geq 0\}$ is a Brownian motion, I am trying to find following conditional expectation

$$\mathbb{E}\left[W^{2}(4) | W(1), W(2)\right]$$

My try:

What I think is that I should introduce the conditional terms $W(1)$ and $W(2)$ into $W^{2}(4)$ to solve the problem. I tried


but this introduces too many terms that don't get any simpler, like $(W(4)W(2)^2)$.

Is the approach incorrect?

  • 3
    $\begingroup$ Write $$W^2(4) = (W(4)-W(2)+W(2))^2 = W(4)^2 +2 (W(4)-W(2))W(2) + W(2)^2$$ and then pull out the $W(2)$-terms from the conditional expectation; use that $W(4)-W(2)$ is independent from $W(2)$ to compute the remaining conditional expectation. $\endgroup$ – saz Mar 6 '19 at 7:22

A fundamental property of Brownian motion is that $W(t)-W(s)$ is independent of $\sigma(W(u),u\leqslant s)$ for all $s<t$. Here, in this context, we know that $W(4)-W(2)$ is independent of $\mathcal F:=\sigma(W(1),W(2))$. But unfortunately, we have to compute the conditional expectation of $W(4)^2$ with respect to $\mathcal F$. So in the last expression, we make the term $W(4)-W(2)$ appear like this: $$ W(4)^2=(W(4)-W(2)+W(2))^2= (W(4)-W(2) )^2+2(W(4)-W(2) )W(2)+W(2)^2. $$ Now we take the conditional expectation with respect to $\mathcal F$: $$ \mathbb E\left[W(4)^2\mid\mathcal F\right]= \mathbb E\left[(W(4)-W(2) )^2\mid\mathcal F\right]+\mathbb E\left[2(W(4)-W(2) )W(2)\mid\mathcal F\right]+\mathbb E\left[W(2)^2\mid\mathcal F\right]. $$

  • first term: $(W(4)-W(2) )^2$ is independent of $\mathcal F$ hence $\mathbb E\left[(W(4)-W(2) )^2\mid\mathcal F\right]=\mathbb E\left[(W(4)-W(2) )^2 \right]=2$.
  • second term: $W(2)$ is $\mathcal F$-measurable and $ W(4)-W(2) $ is independent of $\mathcal F$: we get zero.
  • third term: $W(2)^2$ is $\mathcal F$-measurable.

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.