Suppose $ X $ is a random variable. Then there is an associate characteristic function $\phi_X(t)=E(e^{itX})$. Then we know that the moments can be expressed as derivatives of $\phi_X$, $$EX=\frac{d\phi_X}{idx}\biggr|_{t=0},\quad EX^2=\frac{d^2\phi_X}{i^2dx^2}\biggr|_{t=0} \text{ and so on}.$$

My question is, is there such a relationship between the autocorrelation function $$E(X(x)X(x'))$$ or higher order versions $$ E(X(x)X(x')X(x''))$$ and the characteristic function $\phi_X$?

  • 2
    $\begingroup$ $t$ in your question is used are argument of $\phi_X(t)$ and also as parameter/argument of $X(t)$. That is confusing. $\endgroup$ – drhab May 24 '18 at 11:33
  • $\begingroup$ Why would you expect this? The characteristic function $\phi_X$ sees only the distribution of $X$ whereas the the autocorrelation function depends on the joint distribution of $(X(x),X(x'))$. $\endgroup$ – saz May 24 '18 at 15:20
  • $\begingroup$ Does it mean that the autocorrelation is related to the characteristic function of products of random variables? $\endgroup$ – adamG May 24 '18 at 15:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.