Does statistical self-similarity preserve covariance?

86 Views Asked by At

Suppose that $X(t)$ is an $\alpha$-fractal Brownian motion.

We know that such a process has the following self-similarity property:

$$X(t) \sim \frac{1}{|a|^{\alpha}} X(at)$$

Where $\sim$ denotes having the same probability distribution.

What I want to understand is whether or not this $\sim$ preserves covariance. For example, does this mean that for any random process $Y(t)$:

$$E[X(t)Y(t)] = E[\frac{1}{|a|^{\alpha}} X(at) Y(t)]$$

The answer isn't clear to me because I'm unsure if the self-similarity means they share all properties including covariance etc, or if they just share the same distribution at time $t$ (ie. Both are $N(0,\sigma^2 t^{2\alpha})$).

1

There are 1 best solutions below

5
On BEST ANSWER

You need to know that the pair $(X(t),Y(t))$ has the same distributions as the pair $(|a|^{-\alpha}X(at),Y(t))$. Your expectation identity will then follow because the expectation is determined by the joint distribution.