Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $(X_t, Y_t)$ be a stationary 2D Gaussian process, therefore $\mathbb{E}\left(X_t\right) = \mathbb{E}(Y_t) = 0$.

I am looking for an explicit example of a valid auto-covariance matrix, i.e: $$ R(h) = \begin{pmatrix} \mathbb{E}\left(X_t X_{t+h}\right) & \mathbb{E}\left(X_t Y_{t+h}\right) \\ \mathbb{E}\left(Y_t X_{t+h}\right) & \mathbb{E}\left(Y_t Y_{t+h}\right) \end{pmatrix} $$ such that $R(h)$ is not symmetric for $h\not=0$. Thank you.

share|improve this question
2  
I fail to see why this was down-voted... If I am being signaled about a flaw in the question, please speak up. I'll try to improve it. –  Sasha Dec 21 '12 at 22:07
    
Somebody downvoted this question? Wow. –  Did Dec 21 '12 at 22:17

1 Answer 1

up vote 2 down vote accepted

Consider your favorite stationary Gaussian process $(X_t)_t$ and define $(Y_t)_t$ by $Y_t=X_{t+\ell}$, for some nonzero $\ell$. Then $\mathbb E(X_tY_{t+h})=\mathbb E(X_0X_{\ell+h})$ is the covariance at distance $|\ell+h|$, while $\mathbb E(Y_tX_{t+h})=\mathbb E(X_0X_{h-\ell})$ is the covariance at distance $|\ell-h|$.

share|improve this answer
    
Thanks a lot! Very useful for me. –  Sasha Dec 21 '12 at 22:03
    
This construction leads to a Gaussian process with degenerate covariance of some distinct-positive-time slices of the process, that is covariance of $(X_{t_1}, Y_{t_1}, \ldots, X_{t_n}, Y_{t_n} )$. I guess I need to revise the question. –  Sasha Dec 29 '12 at 6:07
    
I actually found that a vector valued ARMA process fits the bill. –  Sasha Dec 29 '12 at 6:24

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.