For an integrable random variable and a sub-sigma-algebra , the conditional expectation is an integrable, -measurable random variable satisfying
for every .
Conditional expectation is unique up to almost sure equality. If and both satisfy the definition, then for every ,
The events and belong to . Testing on them, or first on and , shows that both have probability zero. Hence almost surely.
If , independence makes independent of itself, so
Thus every event in has probability zero or one. The intersection is trivial modulo null sets, and the independent sigma-algebras have trivial intersection result gives
almost surely.
Put . By symmetry,
Their sum is , which is measurable with respect to , so
Therefore
This also follows from the Gaussian conditional expectation formula because .
If , then is already -measurable, so
If , the tower property of conditional expectation gives
Finally, if and are independent, the -measurable variable is independent of . Its conditional expectation given is its mean . Part c shows that the right side is also .
The equation fails for general nonnested sigma-algebras. On the four-point space with uniform probability, let
and take . The intersection is trivial, so
But is -measurable and
which is zero on and is not almost surely . This exhibits the failure of iterated conditional expectation over nonnested sigma-algebras.

Articles by others on the same topic (0)

There are currently no matching articles.