I have seen multiple definitions for what a measurable set is (all of which come together to form a sigma algebra). I was wondering if they are all equivalent and if not what situation would one be used over another?
- Definition 1
Let $(X, \Sigma)$ be a measurable space, then any set $S \in \Sigma$ is a measurable set.
Measurable Space: The pair $(X, \Sigma)$ where $X$ is a set and $\Sigma$ is a $\sigma$-algebra on $X$
- Definition 2
Given a space $X$ let there exist an outer measure $\mu : 2^{X} \to [0, \infty]$ (where $2^{X} = \mathcal{P} \left( X \right) = $ all the subsets of $X$) then a set $S$ is measurable iff for every $A \in 2^{X}$ $$ \mu(A) = \mu(A \cap S) + \mu(A \cap S^{c}) = \mu(A \cap S) + \mu(A \setminus S) $$
- Definition 3 (I'm drawing this one from memory from baby rudin, and it's defined on $\mathbb{R}$)
Begin by defining a (outer$_1$) measure $\mu$. Next put $\mathcal{M}_f$ to be the set of countable unions of intervals. Then a set $S$ is measurable iff $$ \exists \{ S_n \}_{n=0}^{\infty} \, s.t. \mu(S_n) \to \mu(S) \text{ as } n \to \infty $$
Now I know that the sets described in definition 2 form a sigma algebra on $X$, and likewise I know the sets $S$ in definition 3 form a sigma algebra on $\mathbb{R}$, but definition 1 seems to imply that any possible sigma algebra can be used.
In definition 3 I cannot remember if he defines this using the outer measure (I believe he does).
The only conclusion I have been able to draw from this is definition 2 and 3 must be equivalent, but we call these sets $\boldsymbol\mu$-measurable (a specific measure is defined). However in definition 1 we call any set like that just measurable, and that given any sigma algebra $\Sigma \, \exists \mu$ an outer measure s.t. the sigma algebra formed by definition 2 with this outer measure is $\Sigma$. Is this correctly put?