1
$\begingroup$

So I understand the method used in this solution, and I know my method is incorrect, but I was just looking for an explanation why.

I was thinking that if I choose any spot on the circumference, there's a $1/2^{n}$ probability that all $n$ points lie in the semicircle starting from that spot on the circumference. Why couldn't you just integrate this probability from 0 to 1 representing moving the spot around the whole circle?

Again, I know this is the wrong and just looking for reasoning why.

$\endgroup$
0
$\begingroup$

I don't see an obvious way to integrate something to get the right answer. Integrating is basically a form of addition (or (weighted) averaging if you prefer), but the things you're adding are not disjoint events.

E.g. let $E_\theta$ be the event that all the $n$ points are in the semicircle starting at the angle $\theta$ and going clockwise. Then for all $\theta, P(E_\theta) = 1/2^n$. But how do you add $E_\theta$ and $E_{\theta + \epsilon}$ for some small $\epsilon$? Those are overlapping semicircles. I find it hard to imagine any integral like $\int_0^{2\pi} P(E_\theta) \,\,\text{blah} \,\, d \theta$ would give you the right answer.

In particular, your original suggestion is equivalent to saying $\text{blah} = {1\over 2\pi}$, which is the density for picking the semicircle uniformly, so the result is $\int_0^{2\pi} P(E_\theta) {1\over 2\pi} d \theta = 1/2^n = $ Prob that a random semicircle (the one at a random $\theta$) contains all $n$ points (as Brian also pointed out).

$\endgroup$
0
$\begingroup$

Your reasoning would be correct if you wanted the probability that all $n$ points are in a random semicircle. The desired probability, however, is the probability that all $n$ points all fall within the same (but not predetermined) semicircle.

Thus, for example, a single point obviously falls within the same semicircle as itself. However, integration gives the probability that the point falls within a randomly selected semicircle, which is $1/2$.

$\endgroup$
  • $\begingroup$ So I understand the probability all $n$ points are in a random semicircle is $1/2^{n}$, and I know integrating from 0 to 1 doesn't change this. But is there a formulation or way to integrate this probability moving the beginning of the semicircle all the way around the circle? $\endgroup$ – jtanman Mar 25 at 23:31

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.