Let be i.i.d. r.v.'s such that and a.s., and let .
(a) Find a number such that is a martingale and justify the martingale property.
(b) Define . Compute .
(c) Compute .
Each is clearly -measureable and finite a.s. (Hence ). Therefore we only need to verify the martingale property. That is, we want to show
We can assert that exists and is finite since each almost surely. Therefore, in order to make a martingale, we must have .
Let be independent Poisson processes with respective parameters , where is an unspecified positive real number. For each , let . Show that does not depend on and find explicitly.
First let us find the distribution of :
Thus by the chain rule, our random variable has probability density function
Now integrate the remaining integral by parts letting . We get:
Repeat integration by parts another times and we get