I'm studying time series, I don't understand the description below.

The white noise series $w_t$ has $E(w_t) = 0$ and

$$\gamma_w(s, t) = \operatorname{cov}(w_s, w_t)= \begin{cases} \sigma_w^2 & s = t \\ 0 & s\ne t\end{cases} $$

But I don't understand how to get the result.

If $s = t,$ then $\gamma_w(s, t) = \gamma_w(t, t) = E[(w_t- \mu_t)^2] = \operatorname{var}(w_t)$

How does $\operatorname{var}(w_t)$ become $\sigma_w^2$? And why if s and t are not equal, the covariance of two sequence will become 0?


$s$ and $t$ are not two sequences, but two timesteps within one sequence.

If $ s\ne t$ you are comparing two different elements of the sequence. If $s = t$ you are comparing one timestep with itself and in such a case autocovariance reduces to variance. In White Noice different elements of the sequence are independent of each other so their covariance by definition is 0.

| cite | improve this answer | |
  • $\begingroup$ Thank you! I suddenly understood! $\endgroup$ – shihs Sep 12 '19 at 21:00

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.