1
$\begingroup$

I'm trying to grasp the intuition of conditional expectations. On a probability space $(\Omega,\mathcal{F},\mathbb{P})$, let $\mathcal{G}\subset\mathcal{F}$ be a sub-sigma-algebra. Denote $X:\Omega\to\mathbb{R}$ to be a random variable. If we have a Lipschitz continuous function $f:\mathbb{R}\to\mathbb{R}$ with Lipschitz constant $C$, is it true that $$ \left|\mathbb{E}[f(X):\mathcal{G}] - f(0)\right| \leq C \mathbb{E}\left[||X||:\mathcal{G}\right] $$ holds?

So in the above, both sides a random variables. So I just applied the definition of Lipschitz continuity and I am unsure if the above is valid since we are conditioning on a sub-sigma-algebra.

$\endgroup$
2
  • $\begingroup$ Conditional expectation inherit the monotonicity from the normal expectation. $\endgroup$ Oct 4 '17 at 10:48
  • $\begingroup$ Thank you quallenjager $\endgroup$
    – jerom
    Oct 4 '17 at 10:58
1
$\begingroup$

Since $\left\lvert \mathbb E\left[Y\mid\mathcal G \right] \right\rvert\leqslant \mathbb E\left[\left\lvert Y\right\rvert\mid\mathcal G \right] $ for any integrable random variable $Y$, we have $$ \left\lvert \mathbb E\left[f(X)\mid\mathcal G \right] -f(0) \right\rvert =\left\lvert \mathbb E\left[f(X)-f(0) \mid\mathcal G \right] \right\rvert \leqslant \mathbb E\left[\left\lvert f(X)-f(0)\right\rvert \mid\mathcal G \right].$$ If $Y_1$ and $Y_2$ are two integrable random variables such that $Y_1\leqslant Y_2$ almost surely, then $\mathbb E\left[Y_1\mid\mathcal G \right] \leqslant \mathbb E\left[Y_2\mid\mathcal G \right]$ almost surely. Apply this to $Y_1:=\left\lvert f(X)-f(0)\right\rvert$ and $Y_2:=C\left\lvert X\right\rvert $.

$\endgroup$

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.