I Is this conditional expectation identity true?

AI Thread Summary
The discussion revolves around the validity of the conditional expectation identity E(f(X,Y) | Y=y) = E(f(X,y) | Y=y). The author expresses concern about this identity, referencing Breiman's work which suggests that E(f(X,y) | Y=y) may not be well defined. Breiman's Proposition 4.36 indicates that while E(f(X,Y) | Y=y) can be expressed through a regular conditional distribution, the substitution with E(f(X,y) | Y=y) is problematic. The author is seeking clarification on how to approach proving the identity given these concerns. The thread highlights the complexities of defining conditional expectations in probabilistic contexts.
psie
Messages
315
Reaction score
40
TL;DR Summary
I am working an exercise to show a conditional expectation identity, but I'm not sure if it is true at all.
I'm working through an exercise to prove various identities of the conditional expectation. One of the identities I need to show is the following $$E(f(X,Y)\mid Y=y)=E(f(X,y)\mid Y=y).$$ But I am a little concerned about this identity from things I've read elsewhere. I am paraphrasing from Breiman's book Probability:
Proposition 4.36: Let ##X## and ##Y## be random variables. Let ##f(X,Y)## be a real valued, random variable such that the expectation of the absolute value of ##f(X,Y)## is finite. If ##Q(\cdot\mid Y=y)## is a regular conditional distribution for ##X## given ##Y=y##, then, $$E(f(X,Y)\mid Y=y)=\int f(x,y) \, dQ(\cdot \mid Y=y)\quad \text{a.s. with respect to the law of }Y.\tag{4.37}$$
Again, paraphrasing Breiman in section 4.3, page 80:
It is tempting to replace the right-hand side of (4.37) by ##E(f(X,y)\mid Y=y)##. But this object cannot be defined through the standard definition of conditional expectation (4.18) [defined below].
Definition 4.18: ##E(X\mid Y=y)## is any random variable on ##\mathbb R##, where ##Q(B)=P(Y \in B)##, satisfying $$ \int_B E(X\mid Y=y) dQ = \int_{Y \in B} X dP,$$ for all Borel sets ##B##.
Thus, does ##E(f(X,Y)\mid Y=y)= E(f(X,y)\mid Y=y)## make sense? Breiman seems to suggest that ##E(f(X,y)\mid Y=y)## is not well defined, so I'm not sure how to proceed.
 
Physics news on Phys.org
I think this pdf clarified things a bit.
 
Back
Top