I am trying to complete Exercise 1.4.27 page 78 of Terence Tao's book "An Introduction to Measure Theory", but is caught up in some troublesome details. The exercise is as follows:
$\bullet$ Show that the Lebesgue measure space $(\mathbb{R}^d,\mathcal{L}[\mathbb{R}^d],m)$ is the completion of the Borel measure space $(\mathbb{R}^d,\mathcal{B}[\mathbb{R}^d],m)$
I know that the completion of a measure space $(X,\mathcal{B},\mu)$ is the measure space $(X,\overline{\mathcal{B}},\overline{\mu})$ where $\overline{\mathcal{B}}=\{B\cup N |B \in \mathcal{B} \land N \in \mathcal{N} \}$, where $\mathcal{N}$ is the collection of subsets of the null sets of $\mathcal{B}$, and $\overline{\mu}:B\cup N \mapsto \mu(B)$.
Claim: $\mathcal{L}[\mathbb{R}^d] \subseteq \{B \cup N | B \in \mathcal{B}[\mathbb{R}^d] \land N \in \mathcal{N}\}$, where $\mathcal{N}:$ Borel subnull-sets
Proof attempt:
Part 1: We show that any $S \in \mathcal{L}[\mathbb{R}^d]$ is of the form $B \cup N$, where $B$ is a Borel set and $N$ is a (Lebsegue) null set.
Let $S \in \mathcal{L}[\mathbb{R}^d]$. Suppose that $m(S)=\delta > 0$. Then there is a closed set $E \subseteq S$ such that $m(S \setminus E)\leq \delta$. Since any closed set is a Borel-set, this proves that $\Lambda=\{U\subseteq S| U \in \mathcal{B}[\mathbb{R}^d]\}$ is nonempty. Therefore, let $B=\bigcup_{A \in \Lambda} A$ be the largest Borel set contained in $S$. Then we also have $B \in \mathcal{L}[\mathbb{R^d}]$, and accordingly $S \setminus B \in \mathcal{L}[\mathbb{R^d}]$. We claim that $m(S \setminus B)=0$. Suppose not, and that $m(S \setminus B)=\gamma >0$. Then we can once again find a closed set $F \subseteq S \setminus B$, contradictiong the maximality of $B$. Therefore $m(S\setminus B)=0.$
Part 2: We show that any (Lebesgue) null set is a Borel subnull set.
Let $N$ be any (Lebsegue) null set. Then, given any $n \in \mathbb{N}$, there is an open set $U_n$ containing $N$, such that $m(U_n \setminus N)\leq 2^{-n}$. Let $U=\cap_{n=1}^{\infty}U_n$. Then $U$ is an open set containing $N$, furtheremore $m(U)\leq 2^{-n}$ for any $n$, so $m(U)=0$. Thus, $U$ is a (Borel) null set, showing that $N \in \mathcal{N}$.
The problem however, is in part 1 of my attempted proof, as $B$ is defined as an (possibly uncountable) union of Borel sets, so I cannot guarantee that it is even well defined. I tried to change the definition of $B$ to the union of all open sets contained in $S$, i.e. $B:=S^\circ$, but this makes matters worse, as $m(S)>0 $ does not imply that $ S^\circ \neq \emptyset$. Can anyone see a way around this problem?