Determining if a random process is first order stationary and mean ergodic

29 Views Asked by At

Let me start out and say I am very new to statistics and am trying to do some self-learning. Some definitions that may seem trivial, I am still learning.

Given some random process:

$$X_t = cos(t\omega_k + \theta_k)$$

If $w_k = 8\pi$, and $\theta_k$ is randomly sampled from $[0,\pi]$, is the process first order stationary and mean ergodic?


  1. Finding out if it is stationary

I am not having the easiest time with the definition I have found on Wikipedia:

Let $F_X(x_{t1+\tau},...x_{t_n+\tau})$ represent the cumulative distribution function for the unconditional, if $F_X(x_{t1+\tau},...x_{t_n+\tau}) = F_X(x_{t1},...x_{t_n})$, the process is stationary.

How do we find the cumulative distribution function of a process that has several variables like this?

$$\int_0^\pi cos(8\pi t + \theta_k) d\theta$$

I assume once I have this I just need to see if the CDF produces different results for different $t$.

While I can see easily how we could disprove stationarity, what would we do in order to prove it. I.e. we need to somehow show for all values of $t$ the values are the same above.


  1. Finding out if it is mean ergodic

For this we would need to see if the mean ($E[X(t)]$) remains constant for all t.

In this case, would the mean calculation be done by:

$$\int_0^\pi \theta\ cos(8\pi t + \theta_k) d\theta$$

?

For this part I am not actually sure where to go with it

Thank you for your time