Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

$\mathcal{F}$ and $\mathcal{G}$ are two sigma algebras on the same space, neither a subset of the other.

If $X$ is $\mathcal{F}$-measurable but not $\mathcal{G}$-measurable is:

$$ E[E[X\mid \mathcal{G}]\mid \mathcal{F}] = X? $$

share|cite|improve this question
up vote 3 down vote accepted

Let $\mathcal{G}=\{\Omega,\emptyset\}$ and $X$ be not almost surely constant. Then $E[X|\mathcal{G}]$ is simply the expectation and constant almost surely. So the answer is no.

In the following example, the two $\sigma$-algebras are not comparable:

Let the underlying probability space be $([0,1],\mathcal{B},\mu)$, with $\mathcal{B}$ being the Borel $\sigma$-algebra and $\mu$ being the uniform distribution (Lebesgue measure). Let $\mathcal{G}$ consists of all subsets that are countable or a have countable complement. Let $\mathcal{F}=\{\emptyset,\Omega,[0,1/2],(1/2,1]\}$. Let $X$ be a random-variable that is $\mathcal{F}$-measurable, but not $\mathcal{G}$-measurable. Then $E[x|\mathcal{G}]$ is equal to a constant function random variable almost surely. This is easily shown. For each positive $n$, there is a closed subinterval of $\mathbb{R}$ with length at most $1/n$ and probability $1$ under the distribution $\mu\circ X^{-1}$ since all sets in $\mathcal{G}$ have measure $0$ or $1$. Pick such an interval for each $n$. Their intersection contains a single point $r$, so $\mu\circ X^{-1}\{r\}=1.$

share|cite|improve this answer
Thanks, but in that scenario G is a subset of F... I'm wondering in the case where it is not a subset. – Dirk Calloway Nov 18 '12 at 18:36
Does that make a difference? – Dirk Calloway Nov 18 '12 at 18:36
@Dirk I've added a counterexample satisfying your restriction. – Michael Greinecker Nov 18 '12 at 19:18

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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