Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm studying advanced probability theorem by myself and have encountered a exercise:

Let $A>0$ be a constant, $\xi$ be a $r.v.$ such that $E|\xi|<\infty$ and

$P(\xi\leq x) = P(-\xi\leq x),\quad x\in\mathbb{R}$

Compute the conditional expectation $E(\xi\ |\ \xi I_{\{|\xi|\leq A\}})$.

I have no idea of this problem.

Any help would be appreciated.


I written a solution (more mathematical in my opinion) follow did's. Tell me if it has anything wrong.

For any $B\in \mathscr{B}_{\mathbb{R}}$, we have

  • If $0\in B$, then $\{\xi I_{\{|\xi|\leq A\}}\in B\} = \{\xi\in B,|\xi|\leq A\}\cup\{|\xi|>A\}$. Hence

$\begin{eqnarray*} & &\int_{\{\xi I_{\{|\xi|\leq A\}}\in B\}}\xi I_{\{|\xi|> A\}} dP \\ &=& \int_{ \{\xi\in B,|\xi|\leq A\}}\xi I_{\{|\xi|> A\}} dP +\int_{\{|\xi|>A\}}\xi I_{\{|\xi|> A\}} dP\\ &=& 0. \end{eqnarray*}$

The last integral is zero due to the symmetric of $\xi$.

  • If $0\notin B$, it is easy to see that

$ \int_{\{\xi I_{\{|\xi|\leq A\}}\in B\}}\xi I_{\{|\xi|> A\}} dP = 0. $

Consequently, we have $E(\xi I_{\{|\xi|> A\}}\ |\ \xi I_{\{|\xi|\leq A\}}) = 0, a.s.$. Therefor, $\begin{eqnarray*} E(\xi\ |\ \xi I_{\{|\xi|\leq A\}}) &=& E(\xi I_{\{|\xi|\leq A\}}\ |\ \xi I_{\{|\xi|\leq A\}}) + E(\xi I_{\{|\xi| > A\}}\ |\ \xi I_{\{|\xi|\leq A\}})\\ &=& \xi I_{\{|\xi|\leq A\}}. \end{eqnarray*}$

share|improve this question
add comment

1 Answer

up vote 2 down vote accepted

Let $\xi_A=\xi\,\mathbf 1_{\{|\xi|\leqslant A\}}$. By definition, $\mathrm E(\xi\mid\xi_A)=u(\xi_A)$ for some measurable function $u$ such that, for every Borel set $B$, $\mathrm E(\xi:\xi_A\in B)=\mathrm E(u(\xi_A):\xi_A\in B)$.

  • For every $x\ne0$ such that $|x|\leqslant A$, $[\xi_A=x]=[\xi=x]$ hence $u(x)=x$.
  • For $x=0$, $[\xi_A=0]=[\xi=0]\cup[|\xi|\gt A]$ and the distribution of $\xi$ is symmetric hence the best predictor of $\xi$ on $[\xi_A=0]$ is $0$.
  • For every $x$ such that $|x|\gt A$, $[\xi_A=x]=\varnothing$ hence the value of $u(x)$ is irrelevant.

To sum up, a solution is $\mathrm E(\xi\mid\xi_A)=\xi_A$ almost surely.

share|improve this answer
    
Thanks very much! –  Eastsun Jun 11 '12 at 16:06
add comment

Your Answer

 
discard

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.