The expectation of the random variable (Poisson process)

252 Views Asked by At

I have to find the following expectation of a Poisson process with parameter $\lambda$ "in terms of independence":

$E(X(t)X(t+s))$

I supposed that "in terms of independence" means that both random variables are independent, so I tried the following:

$E(X(t)X(t+s)) = E(X=t)E(X=t+s)= \lambda E(X=t+s)$

Here is where I start struggling. I do not know how to work with $E(X(t+s))= \sum_{k=0}^{\infty} k P(X=t+s)$. I tried using conditional probability to see if I could convert the formula into something like $ \sum_{n=s}^{\infty}kP(....) $ but that left me confused.

Also, how would you compute the original expectation if the random variables were dependent?

1

There are 1 best solutions below

1
On BEST ANSWER

Poisson process has independent increments property. In terms of independence means, use independent increment property to obtain the required expectation. \begin{align*} E\left\{X(t)\cdot X(t+s)\right\}& = E\left[X(t)\left\{\overbrace{X(t+s)-X(t)}+X(t)\right\}\right]\\ & = E\left[X(t)\left\{X(t+s)-X(t)\right\}\right] + E\left[X(t)\cdot N(t)\right]\\ & = E\left[X(t)\right]\cdot E\left[X(t+s) - X(t)\right] + E\left[X^{2}(t)\right]\\ &\qquad\mbox{(because of independent increment property}\\ & \qquad\mbox{of the Poisson Process.)}\\ & = \lambda t\cdot E\left[X(s)\right] + \left[\lambda t + (\lambda t)^{2}\right]\\ & \qquad \mbox{(on using the stationarity property) }\\ & = \lambda t\cdot \lambda s + \lambda t + (\lambda t)^{2}\\ & = \lambda t + \lambda t\cdot(\lambda t + \lambda s). \end{align*}