2
$\begingroup$

Hmm, I was just talking to a friend of mine...and I said that

Personally I would like to define the discrete probability function to be $ |event|\over |sample space|$

Then I gave an example about rolling a fair die P(outcome is even)=$|2,4,6| \over |1,2,3,4,5,6|$ which is 50%, and my friend asked me if it worked with infinite discrete set...without thinking too much I said...of course, if the numerator is all even positive integers and the denominator is all positive integers, then I know the probability is going to be 50%

but later, when I tried to prove it mathematically...I failed, since I know that set of even positive integers and all positive integers have the same cardinality...the one to one mapping is just times two

so...my function should return me 1? I know this is not making sense...can someone help me...I think somehow I confused myself :(

$\endgroup$
  • $\begingroup$ are you sure it is 50%? is the die fair? $\endgroup$ – BCLC Jan 31 '16 at 17:51
  • 1
    $\begingroup$ @BCLC, good question I think this probability function will only work when each element in the sample space is equally likely, if this is not the case...maybe we need to introduce a gcd? $\endgroup$ – watashiSHUN Jan 31 '16 at 20:44
5
$\begingroup$

There is no uniform density function on a countable set such as the integers - that is, there is no way to define a probability distribution for which every integer is an equally likely output.

But a related concept is natural density or asymptotic density, which will (in some sense) capture the intuition that the evens and odds pair up as they do for finite sets.

$\endgroup$
  • $\begingroup$ I don't really get the first sentence, are you saying that we can't define a probability function over countable set like integers because of non-uniform density function(what does it mean...)? can you give me a counter example? ie something with a uniform density function and what would the probability function look like? thanks :) $\endgroup$ – watashiSHUN Jan 31 '16 at 21:00
  • $\begingroup$ What I am saying is that there is no way to define a pdf so that every integer has equal weight. We can define a density on the positive integers by something like $p(1) = 1/2$, $p(2) = 1/4$, $p(3) = 1/8$, and so on. $\endgroup$ – user296602 Jan 31 '16 at 23:50

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.