10
$\begingroup$

While reading the probability space in Wikipedia, I'd found the usual formulation is a triplet, which is ${\displaystyle (\Omega ,{\mathcal {F}},P)}$.

Upon my understanding, the middle ${\mathcal {F}}$ is a power set of $\Omega$ which will be allocated with real-valued probabiilty by $P$.

If every set in this nature has power set, there might be no necessity of introduction of ${\mathcal {F}}$ I guess however, I've never thought of a set which doesn't have its power set.

Is there any set that doesn't have power set? or if not, which means every set has its power set, is there any plausible reason that ${\mathcal {F}}$ is introduced in probability formulation?

|cite|improve this question
$\endgroup$
  • 2
    $\begingroup$ Could you say more specifically where you read that $\mathcal F$ is the power set of $\Omega$? This is not true in general, and I can't find where the Wikipedia article on probability spaces says that. $\endgroup$ – joriki Sep 20 '18 at 10:49
  • $\begingroup$ @joriki your link is what I exactly intended to refer in OP. as you mentioned, I thought myself intuitively reasonable to regard it as a power set.. but it's not. Thanks for the correction point! $\endgroup$ – Beverlie Sep 20 '18 at 10:51
  • 1
    $\begingroup$ Since this question is a good example of the XY problem, and the question is therefore rather more specific than the title suggests, should the title be edited to reflect this? Perhaps by adding "and is the set of events always the power set of the sample space?" $\endgroup$ – Silverfish Sep 20 '18 at 21:03
27
$\begingroup$

In standard mathematics, every set has a power set. This is encoded in the Axiom of Power Set. However, your confusion lies with the definition of a probability space, not with set theory.

The set $\mathcal F$ in a probability space $(\Omega, \mathcal F, P)$ is not necessarily the power set of $\Omega$. The set $\mathcal F$ is a subset of the power set $\mathcal P(\Omega)$. This $\mathcal F$ is required to be a so-called sigma algebra, which tells you that it shares some properties in common with the full power set, but it need not be the full power set at all.

In particular, for any $\Omega$, you can take $\mathcal F = \{\emptyset, \Omega\}$, and this will be a sigma algebra on $\Omega$. Unless $|\Omega| \leq 1$, it will not be the power set.

|cite|improve this answer
$\endgroup$
  • $\begingroup$ thanks for letting me the point that every set has a power set is an axiom.. never known that. $\endgroup$ – Beverlie Sep 20 '18 at 10:53
  • 7
    $\begingroup$ I think it is worth pointing out that in many cases we do not have a good way of using the power set of $\Omega$ for the set $\mathcal F$, so the practice of allowing the use of a subset of the power set is not just something we arbitrarily choose to do, it's practically a necessity. $\endgroup$ – David K Sep 20 '18 at 12:30
  • $\begingroup$ In finite spaces $\Omega$ we almost always take the power set. It's also pretty comon for discrete probabilities on a counatbel $\Omega$. It really pays to be precise for event spaces in the reals... $\endgroup$ – Henno Brandsma Sep 20 '18 at 13:20
  • $\begingroup$ @HennoBrandsma Of course, for finite sets the probability is uniquely determined by its values on singletons. $\endgroup$ – Acccumulation Sep 20 '18 at 15:39
7
$\begingroup$

This $\mathcal{F}$ is actually not the power set, but a sigma algebra. The power set is a sigma algebra and is often used, but sometimes probability theory requires smaller subsets of the power set in order to properly define the problem.

|cite|improve this answer
$\endgroup$
5
$\begingroup$

I think the notation here is due to the fact that a probability space is, is particular, a finite measure space. So, in the more general sense, the set $\mathcal{F}$ does not need to be the power set, it is sufficient that the set $\mathcal{F}$ satisfy the axioms of a $\sigma$-algebra of subsets of $\Omega$. And yes, every set has a power set, it is guaranteed by the power set axiom in set theory. If you're interested, study some measure theory, which provides the mathematical tools for developing rigorous probability.

|cite|improve this answer
$\endgroup$

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.