probability that use measure spaces, but what I beleive this is
trying to say is something like:
A discrete random vector is a random variable which can take values
in a finite or countably infinite subset of R^n
I expect someone will be along shortly who can give an kosher measure
theoretic interpretaion for you.