2
$\begingroup$

Let $\mathbb{B}$ denote the set of Borel sets of $\mathbb{R}$. Let $\Omega = [0,1]$.

Let $A_1 = \sigma \text{(open subsets of }\Omega)$, that is, sigma algebra generated by open subset of $\Omega$.

Let $A_2 = \{K\subset \mathbb{R}: K = B \cap \Omega, B\in \mathbb{B}\}$.

Then, I would like to show that $A_1 = A_2$.

I was able to show that $A_2$ is a sigma algbra containing $A_1$, hence $A_1\subset A_2$.

But, how do I show $A_2 \subset A_1$?

$\endgroup$
3
$\begingroup$

To prove that $A_2 \subset A_1$, first, recall that $\mathbb{B} = \sigma(\text{open subsets of }\mathbb{R}).$ We already know that the intersection of any open subset of $\mathbb{R}$ with $\Omega$ is contained in $A_1$. From here, to prove that the intersection of any Borel set of $\mathbb{R}$ is contained in $A_1$, it suffices to prove the following two claims:

(1) If $\{E_n \cap \Omega\}_{n \in \mathbb{N}} \subset A_1$, then $$ \left(\cup_{n \in \mathbb{N}}E_n \right) \cap \Omega \in A_1 \text{,}$$ i.e. if $\{E_n\}$ is an indexed collection of sets so that the intersection of each with $\Omega$ is in $A_1$, then the intersection of their union with $\Omega$ is contained in $A_1$.

(2) If $E \cap \Omega \in A_1$, then $E^c \cap \Omega \in A_1$.

$$ $$ Proof of (1):

Let $\{E_n\}$ be as in the statement. Then $$\left( \cup_{n \in \mathbb{N}} E_n \right) \cap \Omega = \cup_{n \in \mathbb{N}} (E_n \cap \Omega).$$ Since $A_1$ is a $\sigma$-algebra on $\Omega$ and $E_n \cap \Omega \in A_1$ for all $n$, $\cup_{n \in \mathbb{N}} (E_n \cap \Omega) \in A_1$.

$$ $$ Proof of (2): Let $E$ be a subset of $\mathbb{R}$ so that $E \cap \Omega \in A_1$. Since $A_1$ is a $\sigma$-algebra on $\Omega$ (and hence closed under complementation relative to $\Omega$), we have that $\Omega \backslash(E \cap \Omega) \in A_1$. We claim that $E^c \cap \Omega = \Omega \backslash(E \cap \Omega)$. The inclusion $E^c \cap \Omega \subset \Omega \backslash(E \cap \Omega)$ is obvious. To prove the reverse inclusion, suppose that $x \in \Omega \backslash(E \cap \Omega)$. Then $x \in \Omega \wedge (x \notin (E \cap \Omega))$, i.e. $x \in \Omega \wedge (x \notin E \lor x \notin \Omega)$. The only way for this statement to be true is to have $x \in \Omega \wedge x \notin E$.

$$ $$ To apply these claims, let $\mathcal{C}$ be the collection of sets $E \subset \mathbb{R}$ so that $E \cap \Omega$ is in $A_1$. The claims (1) and (2), combined with the fact that $A_1$ contains any set that is open relative to $\Omega$, show that $\mathcal{C}$ is a $\sigma$-algebra that contains the open sets of $\mathbb{R}$. Hence, by definition of $\mathbb{B}$, $\mathbb{B} \subset \mathcal{C}$. As a result, the intersection of any Borel set with $\Omega$ is contained in $A_1$.

$\endgroup$
  • $\begingroup$ With the claims (1) and (2) proved, how do you get $A_2 \subset A_1$? $\endgroup$ – mononono Jun 15 '15 at 18:49
  • $\begingroup$ If I pick a set, say $K \in A_2$, I do not see how the claims (1) and (2) give $K \in A_1$ $\endgroup$ – mononono Jun 15 '15 at 18:51
  • $\begingroup$ I have added another paragraph at the end. I hope it helps. The gist is that the collection of sets whose intersection with $\Omega$ is in $A_1$ is a sigma algebra which contains all the open sets of $\mathbb{R}$. $\endgroup$ – Jordan Green Jun 16 '15 at 16:14

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.