Countably generated means it is a $\sigma$-algebra generated by a countable collection of sets, say $\{A_{n}\}$. I want to show that for a countably generated $\sigma$-algebra $\mathscr{F}$, there exists a countable collection $\{F_{n}\}\subset \mathscr{F}$ s.t. $\forall \varepsilon>0, \forall B\in\mathscr{F}$, $\exists F_{m}\in\{F_{n}\}$ s.t. $P\left(F_{m} \Delta B\right)<\varepsilon$. $P$ is a probability measure but I think this works for any measure space?

My idea is to show that if there is no such countable dense collection, then it contradicts the fact that $\mathscr{F}$ is the smallest $\sigma$-algebra generated by a countable collection, but I am really not sure about it.


This follows from two standard facts both of which can be found in Halmos' book.

1). If an algebra $\mathcal A$ generates a sigma algebra $\mathcal F$ and $P$ is a finite measure on $\mathcal F$ then, for any $\epsilon>0$ and any $A\in \mathcal F$, there exists $B \in \mathcal A$ such that $P(A\Delta B)<\epsilon$.

2) The algebra generated by a countable collection of sets is countable.

  • $\begingroup$ Halmos actually proved the second for ring, will the proof work for algebra? $\endgroup$ – The R Apr 8 at 3:04
  • 1
    $\begingroup$ @TheR Yes the arguments in Halmos' book can be used for algebras too. $\endgroup$ – Kavi Rama Murthy Apr 8 at 5:17

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.