I found the following definition of stochastic process: A stochastic process on a countable state space V is a sequence of random variables $(S_n)_{n \in \mathbb{N}}$ taking values in V and defined on a probability space $(\Omega, \mathcal{F}, P)$.

Why is in this definition the $\sigma$-algebra associated to $V$ omitted? In order $S_n$ to be a random variable, then shouldn't $V$ be a measurable space? What would be the $\sigma$-algebra then?


For a countable set $V$, the typical way to make it into a measurable space is to equip it with the discrete $\sigma$-algebra, i.e. its power set $2^V$, so that every subset of $V$ is measurable.

For most purposes, this is the only sensible thing to do. Usually one wants singleton sets to be measurable. In a countable space, every subset is a countable union of singletons, so this forces every subset to be measurable.

So most people would consider this the "default" $\sigma$-algebra for a countable set; if no $\sigma$-algebra is explicitly stated, then you are almost certainly meant to assume $2^V$. Similarly, if someone talks about $\mathbb{R}$ or $\mathbb{R}^n$ in a context where a measurable space is wanted, you should normally assume that they mean to use the Borel $\sigma$-algebra.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.