6
$\begingroup$

I have this problem in my assignment:

Show that there do not exist uncountably many independent, non-constant random variables on $ ([0,1],\mathcal{B},\lambda) $, where $ \lambda $ is the Lebesgue measure on the Borel $ \sigma $-algebra $ \mathcal{B} $ of $ [0,1] $.

Can someone please help me to solve this?

$\endgroup$
11
$\begingroup$

Assume that $(X_i)_{i \in I} $ are independent random variables. Let $Y_i := X_i \cdot 1_{|X_i|\leq C_i} $, where $C_i>0$ is chosen so large that $Y_i \not\equiv c_i $ for some constant $c_i $. This is possible since the $X_i $ are non-constant.

Then the $(Y_i - \Bbb{E}(Y_i))_i $ form a family of independent and hence orthogonal random variables. Note that the $Y_i $ are bounded and thus contained in $L^2$.

But the separable (!) space $L^2 ([0,1],\lambda) $ can only contain countably many elements which are mutually orthogonal. Hence, $I $ is countable.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ thanks a lot for the answer. That was a great help!! $\endgroup$ – Raghava G D Nov 30 '15 at 6:05
  • $\begingroup$ @ PhoemueX: Are there other approaches to solve this problem that you are aware of ? $\endgroup$ – Raghava G D Nov 30 '15 at 7:23
  • $\begingroup$ @Raghava: I was thinking about that myself. At the moment I don't see a substantially different method. One could probably argue somehow using the cardinality of the Borel sigma algebra, but this idea is somewhat imprecise right now. $\endgroup$ – PhoemueX Nov 30 '15 at 9:18
  • $\begingroup$ Does this also imply that there do not exist uncountably many orthogonal, non-constant random variables on $([0,1], \mathcal{B}, \lambda)$? $\endgroup$ – gigalord Jan 9 at 21:01
  • $\begingroup$ @gigalord: Yes, exactly. $\endgroup$ – PhoemueX Jan 9 at 22:39

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.