I'm trying to simplify the following expectation so that I can later solve a maximization problem: $max_k E[(A - kB)^2]$, where $A$ ~ $N(0,\sigma^2_1)$, $B = A+ \epsilon$ and $\epsilon$ ~ $N(0,\sigma^2_2)$.

I thought I could linearize the expectation by expanding the square, and then plugging in $B$ in terms $A + \epsilon$, but each term I get depends on $A$ in this case, which has an expected value of 0, so I feel that I'm thinking about this the wrong way. What's the best way to simplify? Am I solving a conditional expectation here, since $B$ depends on $A$?


Expanding the expectation yields:


Here, if $A$ and $\epsilon$ are independent, $E[A\epsilon]=E[A]E[\epsilon]=0$, and we'll have $f(k)=(1-k)^2\sigma_1^2+k^2\sigma_2^2$. We just take derivative wrt $k$ and have $$f'(k)=-2(1-k)\sigma_1^2+2k\sigma_2^2=0\rightarrow k=\frac{\sigma_1^2}{\sigma_2^2-\sigma_1^2}$$

  • $\begingroup$ Ah I think the step I was missing was that the expected of $A^2$ and such is the variance. Thanks! $\endgroup$ – John Alperto Mar 14 '19 at 12:25

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.