Suppose $X_1, X_2, ... X_n$ are independent and uniformly distributed (on $[0,1]$) random variables. Determine $\lim_{n\to\infty} \mathbb{P}(\sum_{i=1}^n X_i \leq \frac{n}{2})$

My thoughts were the following:

I suppose I can say that $$\lim_{n\to\infty} \mathbb{P}\left(\sum_{i=1}^n X_i \leq \frac{n}{2}\right)=\lim_{n\to\infty} \mathbb{P}\left(\frac{1}{n}\sum_{i=1}^n X_i \leq \frac{1}{2}\right)=\lim_{n\to\infty} \mathbb{P}\left(\overline X_n \leq \frac{1}{2}\right)$$ And isn't $\overline X_n$ also uniformly distributed? So the probability equals $\frac{1}{2}$?

  • $\begingroup$ You can get the parentheses to adjust to their content by using \left and \right. $\endgroup$ – joriki Jun 13 '16 at 8:56
  • $\begingroup$ @joriki: And also $\to$ with \to. $\endgroup$ – Asaf Karagila Jun 13 '16 at 8:58

No, $\overline{X_n}$ isn't uniformly distributed; but it's distributed symmetrically about $\frac12$, so you can nevertheless conclude that the probability is $\frac12$ even without the limit.

| cite | improve this answer | |
  • $\begingroup$ So what is the right way to go? I don't fully understand $\endgroup$ – Di-lemma Jun 13 '16 at 9:08
  • 1
    $\begingroup$ If one is not comfortable thinking about this as intuitively as @joriki has described, we can also apply the Central Limit Theorem. In the limit, $\bar{X_n}$ will be distributed normally, and the mean of $\bar{X_n}$ will be $\mathbb{E}(X_i) = \frac{1}{2}.$ Thus, by the symmetry of the normal distribution, this probability would be $1/2.$ $\endgroup$ – David Jun 13 '16 at 9:20
  • $\begingroup$ @Di-lemma: There's no more way to go. The statement is true by symmetry, even if you remove the limit. If you want to make the symmetry explicit, it's $X_i\to1-X_i$. $\endgroup$ – joriki Jun 13 '16 at 9:21
  • 1
    $\begingroup$ @David: I think you meant "normally" instead of "uniformly"? But it would be quite a detour to prove something using the central limit theorem when it's actually true without the limit. $\endgroup$ – joriki Jun 13 '16 at 9:22
  • 1
    $\begingroup$ @David: Then it's a bad exercise :-) $\endgroup$ – joriki Jun 13 '16 at 9:35

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.