I have problem with proving the following fact:

Let $\mu$ be regular version of distribution of random variable $X$ with respect to $\sigma$-algebra $\mathcal{B}$. Let $Y$ be $\mathcal{B}$-measurable random variable. Prove that for any Borel function $\phi$ such that $\mathbb{E}|\phi (X,Y)|<\infty$, we have $$ \mathbb{E}(\phi(X,Y)|\mathcal{B})(\cdot) = \int \limits_{\mathbb{R}} \phi(x, Y(\cdot)) \mu(dx, \cdot)\ \ \ \ \ \text{a.s.} $$ How does this equation look like for $X$ independent from $\mathcal{B}$?

I can see that function $\omega \mapsto \int \limits_{\mathbb{R}} \phi(x, Y(\omega)) \mu(dx, \omega)$ is $\mathcal{B}$-measurable, so it would be enough to prove that $$\forall_{B \in \mathcal{B}} \int \limits_{B} \mathbb{E}(\phi(X,Y)|\mathcal{B})(\omega)dP(\omega) = \int \limits_{B} \phi(X(\omega),Y(\omega))dP(\omega) = \\ = \int \limits_{B}\left( \int \limits_{\mathbb{R}} \phi(x, Y(\omega)) \mu(dx, \omega)\right)dP(\omega).$$

First equality is obvious from the definition of conditional expectation, but I have problem with second one.

I would apreciate every help. Thanks!


The proof seems to go by complicating function $\phi$, starting from $\phi=I_{A\times B}$, for $A, B \in \mathcal{B}(\mathbb{R})$.


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.