I want to define the covariance of a variable, say $X$, which is the integral over $Y(t)$:

$$ {\rm Cov}[X,X'] = \int\int {\rm Cov} [aY(t), a'Y(t')] dt dt'.$$

Just as an example, let the variable $Y$ be unitless, so that the total covariance has units of ${\rm sec}^2$.

However, in my case, samples over $t$ are independent, so the covariance when $t\neq t'$ is 0. What is the correct way to define my covariance within the integral?

My initial thought is to use ${\rm Cov}[Y(t),Y(t')] = K \delta(t-t').$, however, this has units $1/{\rm sec}$ (unless $K$ has units of ${\rm sec}$?), so that the final result has units of ${\rm sec}$, rather than ${\rm sec}^2$.

Is there a standard way of doing this?

  • $\begingroup$ Did you ever figure it out? $\endgroup$
    – Layman
    Jun 25, 2019 at 11:57


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Browse other questions tagged or ask your own question.