I'm having trouble deducing the value for the problem in the title.

Here is what I have done so far. (Given a standard Brownian motion (BM) $W_t, t\geq0 $ with $W_0 = 0$ and $\sigma^2=1$)

The standard covariance formula is given by $cov(W_x,W_y) = E(W_xW_y) - E(W_x)E(W_y)$ and because BM is also a Markov process $E(W_x)E(W_y) = 0$ so the covariance formula can be simplified to $cov(W_x,W_y)=E(W_xW_y)$

BM is also a Gaussian process so $cov(W_iW_j) = E(W_i,W_j) = min(W_i,W_j)$ where $i,j\in[0,\infty]$ so e.g. as $\sigma^2=1$, $ cov(W_7,W_9) = 7$

Going back to my original question of finding $cov(5W_7+6W_9,W_7)$, is the following computation correct?

$cov(5W_7+6W_9,W_7) = cov(5W_7,W_7) + cov(6W_9,W_7)$

$=E(5W_7,W_7) + E(6W_9,W_7)$

$=5E(W_7,W_7) + 6E(E_9,W_7)$

$=5*7 + 6*7$


  • 1
    $\begingroup$ Your title says $cov(5W_7+6W_9,W_3)$ but your calculations use $cov(5W_7+6W_9,W_7)$. $\endgroup$ – Henry Sep 14 '14 at 16:42
  • $\begingroup$ Good point. I've amended the title to match. thanks $\endgroup$ – Paul P M Sep 14 '14 at 17:14
  • 1
    $\begingroup$ Unless I missed something, that calculation looks okay now $\endgroup$ – Glen_b Sep 14 '14 at 17:47
  • $\begingroup$ Thanks @Glen_b, I just wasn't sure and couldn't find any sources that went through this kind of problem even though it is really just a combination of simpler problems ... still so much to learn. $\endgroup$ – Paul P M Sep 15 '14 at 0:16

Your calculations look correct, and your approach is probably the one intended.

Another way of getting the same result is to note that $W_9-W_7$ is independent of $W_7$ so $$cov(5W_7+6W_9,W_7) = cov(11W_7+6(W_9-W_7),W_7)$$ $$= 11 var(W_7)+6cov(W_9-W_7,W_7) =11\times 7 +6\times 0.$$

| cite | improve this answer | |
  • $\begingroup$ Thanks. btw, which rule did you use to get from $cov(5W_7+6W_9,W_7)$ to $cov(11W_7+2(W_9-W_7),W_7)$ ? $\endgroup$ – Paul P M Sep 15 '14 at 0:33
  • $\begingroup$ I think the "2" is wrong, actually (it's plainly 6), though since it gets multiplied by zero, the error doesn't change the answer. $\endgroup$ – Glen_b Sep 15 '14 at 0:41

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.