Is there a choice of a $\sigma$-algebra which makes the conditional expectation of an random-variable $X \in L^1_{\mathbb{P}}(\mathcal{F},\mathbb{P})$, on a probability space $(\Omega,\mathcal{F},\mathbb{P})$ equal to the expectation itself?
2 Answers
If $\mathcal F=\{\Omega,\varnothing\}$ then random variables measurable wrt to $\mathcal F$ must be constant.
$\mathbb E[X\mid\mathcal F]$ is by definition measurable wrt to $\mathcal F$ (so must be constant) and must also satisfy the condition $\mathbb EX=\mathbb E[\mathbb E\mid X]]$.
This together leads to the conclusion that $\mathbb E[X\mid\mathcal F](\omega)=\mathbb EX$ for every $\omega\in\Omega$.
By identifying $\mathbb{R}$ with the set of constants in $L^1_{\mathbb{P}}(\mathcal{F})$, you get $\mathbb{E}[X]$, (or at-least a copy of it after the identification is made).
-
$\begingroup$ Cool but, can you please provide some more details...since this is just the same point I'm at....thank you though :) $\endgroup$– user355356Commented Feb 5, 2019 at 17:05