4. Carathéodory's Extension Theorem  PDF TEX

Carathéodory’s extension theorem

Outer measures induced by premeasures

Definition of a premeasure . Recall that a premeasure on an algebra \(\mathcal A \subseteq \mathcal{P}(X)\) of sets on \(X\) is a function \(\mu_0:\mathcal A \to [0,\infty]\) satisfying

  1. \(\mu_0(\varnothing)=0\),

  2. if \((A_n)_{n\in \mathbb N}\) is a sequence of disjoint sets in \(\mathcal A\) such that \(\bigcup_{n=1}^{\infty}A_n \in \mathcal A\), then \[\mu_0\bigg(\bigcup_{n=1}^{\infty}A_n\bigg)=\sum_{n=1}^{\infty}\mu_0(A_n).\]

Outer measures induced by premeasures.

If \(\mu_0\) is a premeasure on \(\mathcal A \subseteq \mathcal{P}(X)\), it induces an outer measure on \(X\): \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\mu_0(E_j): (E_j)_{j\in\mathbb N} \subseteq \mathcal A \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}, \qquad A\subseteq \mathcal P(X).\]

Proposition

Proposition. Let \(X\neq\varnothing\) be a set and let \(\mu_0\) be a premeasure on an algebra \(\mathcal A\subseteq \mathcal P(X)\) and let \(\mu^{*}\) be an outer measure defined as before \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\mu_0(E_j): (E_j)_{j\in\mathbb N} \subseteq \mathcal A \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}, \qquad A\subseteq \mathcal P(X).\] Then one has that

  1. \(\mu_0(E)=\mu^{*}(E)\) for all \(E \in \mathcal A\),

  2. \(\mathcal A\subseteq \mathcal M(\mu^*)\). In other words, every set in \(\mathcal A\) is \(\mu^{*}\)-measurable.

Proof of (a). Suppose \(E \in \mathcal A\), we will show that \(\mu_0(E)=\mu^{*}(E)\).

  • Taking \(E_1=E\) and \(E_2=E_3=\ldots=\varnothing\) we see that \(E \subseteq \bigcup_{j=1}^{\infty}E_j\) and consequently \(\mu^{*}(E) \leq \mu_0(E)\).

  • We now reverse this inequality and show that \(\mu_0(E) \leq \mu^{*}(E)\).

    If \(E \subseteq \bigcup_{j=1}^{\infty}A_j\) with \(A_j \in \mathcal A\), then we let \[B_n=E \cap \bigg(A_n \setminus \bigcup_{j=1}^{n-1}A_j\bigg)\in\mathcal A.\] Then \(B_n \cap B_m =\ \varnothing\) if \(n \neq m\) and \(E\subseteq \bigcup_{j=1}^{\infty}B_j=\bigcup_{j=1}^{\infty}A_j\). Now \[\begin{aligned} \mu_0(E)\le \sum_{j=1}^{\infty}\mu_0(B_j) \leq \sum_{j=1}^{\infty}\mu_0(A_j). \end{aligned}\] It follows that \(\mu_0(E) \leq \mu^{*}(E)\) as claimed.

Proof of (b). We now prove that \(\mathcal A\subseteq \mathcal M(\mu^*)\).

  • Let \(C \in \mathcal A\) and \(E \subseteq X\) it suffices to show that \[\mu^{*}(E) \geq \mu^{*}(E \cap C)+\mu^{*}(E \cap C^c).\] Let \(\varepsilon>0\), then there is a sequence \((B_n)_{n=1}^{\infty}\subseteq \mathcal A\) such that \[E \subseteq \bigcup_{n =1}^{\infty}B_n \quad \text{ and } \quad \sum_{n=1}^{\infty}\mu_0(B_n) \leq \mu^{*}(E)+\varepsilon.\] \(\mu_0\) is additive on \(\mathcal A\), thus \[\begin{aligned} \mu^{*}(E)+\varepsilon \geq \sum_{j=1}^{\infty}\mu_0(B_j)= \sum_{j=1}^{\infty}(\mu_0(B_j \cap C)+\mu_0(B_j \cap C^c)). \end{aligned}\]

  • Since \(\mu_0 \geq \mu^{*}\) and \(B=\bigcup_{j=1}^{\infty}B_j\) we obtain \[\begin{aligned} \mu^{*}(E)+\varepsilon &\geq\sum_{j=1}^{\infty}(\mu_0(B_j \cap C)+\mu_0(B_j \cap C^c))\\& \geq \sum_{j=1}^{\infty}(\mu^{*}(B_j \cap C)+\mu^{*}(B_j \cap C^c))\\&\geq (\mu^{*}(B \cap C)+\mu^{*}(B \cap C^c)) \\&\geq (\mu^{*}(E \cap C)+\mu^{*}(E \cap C^c)) \end{aligned}\] by subadditivity of \(\mu^*\). Letting \(\varepsilon \to 0\) we conclude \[\begin{aligned} \mu^{*}(E) \geq \mu^{*}(E \cap C)+\mu^{*}(E \cap C^c), \end{aligned}\] thus \(C\) is \(\mu^{*}\)-measurable and \(\mathcal A\subseteq \mathcal M(\mu^*)\) as desired.$$\tag*{$\blacksquare$}$$

Carathéodory’s extension theorem

Theorem. Let \(X\neq\varnothing\) be a set, let \(\mathcal A \subseteq \mathcal{P}(X)\) be an algebra, \(\mu_0\) a premeasure on \(\mathcal A\), and \(\mathcal M= \sigma(\mathcal A)\). There exists a measure \(\mu\) on \(\mathcal M\) whose restriction to \(\mathcal A\) is \(\mu_0\) and \(\mu(E)=\mu^{*}(E)\) for all \(E \in \mathcal M\), where \(\mu^{*}\) is given by \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\mu_0(E_j): (E_j)_{j\in\mathbb N} \subseteq \mathcal A \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}, \qquad A\subseteq \mathcal P(X).\]

  1. If \(\nu\) is another measure on \(\mathcal M\) that extends \(\mu_0\), then \(\nu(E) \leq \mu(E)\) for all \(E \in \mathcal M\), with equality when \(\mu(E)<\infty\).

  2. If \(\mu_0\) is \(\sigma\)-finite, then \(\mu\) is the unique extension of \(\mu_0\) to a measure on \(\mathcal M\).

Proof. For the first part of the theorem note that:

  • \((X, \mathcal M(\mu^*), \mu^*)\) is a complete measure space, which follows from the Carathéodory theorem.

  • \(\mathcal A\subseteq \mathcal M(\mu^*)\) by the previous proposition and consequently we obtain that \(\mathcal M=\sigma(\mathcal A)\subseteq \mathcal M(\mu^*)\).

  • Thus it suffices to define \(\mu\) to be the restriction of \(\mu^*\) to \(\mathcal M=\sigma(\mathcal A)\). Then we see that \(\mu(E)=\mu_0(E)\) for all \(E\in \mathcal A\).

Proof of (i). Assume that \(\nu\) is another measure on \(\mathcal M=\sigma(\mathcal A)\) such that \(\nu(E)=\mu_0(E)\) for all \(E \in \mathcal A\). We show that

\(\nu(E) \leq \mu(E)\) for all \(E \in \sigma(\mathcal A)\);

\(\nu(E)=\mu(E)\) if \(E \in \sigma(\mathcal A)\) and \(\mu(E)<\infty\).

For (a): If \(E \in \mathcal M\) and \(E \subseteq \bigcup_{j \in \mathbb{N}}A_j\), where \((A_j)_{j\in\mathbb N} \in \mathcal A\) then \[\nu(E) \leq \sum_{j=1}^{\infty}\nu(A_j)=\sum_{j=1}^{\infty}\mu_0(A_j)\] and we conclude that \({\color{red}\nu(E) \leq \mu(E)}\).

For (b): If we set \(A=\bigcup_{j=1}^{\infty}A_j\), then we have \[\color{blue} \nu(A)=\lim_{n \to \infty}\nu\big(\bigcup_{j=1}^{n}A_j\big) =\lim_{n \to \infty}\mu_0\big(\bigcup_{j=1}^{n}A_j\big)=\lim_{n \to \infty}\mu\big(\bigcup_{j=1}^{n}A_j\big)=\mu(A).\]

If \(\mu(E)<\infty\), let \(\varepsilon>0\) and choose \(A_j\)’s from \(\mathcal A\) such that \[\mu(A) \leq \mu(E)+\varepsilon,\] hence \(\mu(A \setminus E)<\varepsilon\) and \[\mu(E) \leq {\color{blue}\mu(A) =\nu(A)}=\nu(E)+{\color{red}\nu(A \setminus E)} \leq \nu(E)+{\color{red}\mu(A \setminus E)} \leq \nu(E)+\varepsilon.\]

Letting \(\varepsilon \to 0\) we see \(\mu(E) \leq \nu(E)\) if \(E \in \mathcal M=\sigma(\mathcal A)\) and \(\mu(E)<\infty\) and we conclude in this case that \[\nu(E)=\mu(E).\]

Proof of (ii). Finally suppose that \(X=\bigcup_{j=1}^{\infty}A_j\), with \(\mu_0(A_j)<\infty\) and \(A_i \cap A_j=\varnothing\) if \(i \neq j\). Then for any \(E \in \mathcal M\) we have \[\mu(E)=\sum_{j=1}^{\infty}\mu(E \cap A_j)=\sum_{j=1}^{\infty}\nu(E \cap A_j)=\nu(E).\] This competes the proof of Carathéodory’s extension theorem. $$\tag*{$\blacksquare$}$$

Outer and inner regularity of outer measures

Outer measures induced by general set functions

Definition. Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For any \(A\subseteq X\)

\[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}.\] defines an outer measure, which we call an outer measure induced by \(\rho\).

  • By Carathéodory’s theorem the collection of \(\mu^*\)-measurable sets \[\mathcal M(\mu^*) =\big\{A\subseteq X: \mu^{*}(E)=\mu^{*}(E \cap A)+\mu^{*}(E \cap A^c) \ \text{ for all } \ E \subseteq X\big\}.\] is a \(\sigma\)-algebra, which will be called the Carathéodory \(\sigma\)-algebra.

  • Moreover, \((X, \mathcal M(\mu^*), \mu^*)\) is a complete measure space, and the outer measure \(\mu^*\) restricted to the Carathéodory \(\sigma\)-algebra \(\mathcal M(\mu^*)\) will be called the Carathéodory measure.

The sets \(F_{\sigma}, G_{\delta}, F_{\sigma\delta}, G_{\delta\sigma}, \dots\)

Definition. Let \(X\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\). We define

  • \(\mathcal E_{\sigma}=\{\bigcup_{n\in\mathbb N}E_n: (E_n)_{n\in\mathbb N}\subseteq \mathcal E\}\) - the collection of countable unions of sets from \(\mathcal E\).

  • \(\mathcal E_{\delta}=\{\bigcap_{n\in\mathbb N}E_n: (E_n)_{n\in\mathbb N}\subseteq \mathcal E\}\) - the collection of countable intersections of sets from \(\mathcal E\).

  • \(\mathcal E_{\sigma\delta}=(\mathcal E_{\sigma})_\delta\) - the collection of countable intersections of sets from \(\mathcal E_\sigma\).

  • \(\mathcal E_{\delta\sigma}=(\mathcal E_{\delta})_\sigma\) - the collection of countable unions of sets from \(\mathcal E_\delta\).

  • \(\mathcal E^c=\{E^c: E\in \mathcal E\}\) - the collection of complements of sets from \(\mathcal E\).

Note that \((\mathcal E_{\sigma})^c=(\mathcal E^c)_{\delta}\) and \((\mathcal E_{\delta})^c=(\mathcal E^c)_{\sigma}\).

Outer regularity of outer measures

Proposition. Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For \(A\subseteq X\) let \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}. \] Suppose that \(\mu^*=\rho\) on \(\mathcal E\). Then

  • for any \(A \subseteq X\) and \(\varepsilon>0\) there exists a set \(U \supseteq A\) such that \[U\in \mathcal E_\sigma \quad\text{ and } \quad \mu^{*}(A) \leq \mu^{*}(U) \leq \mu^{*}(A)+\varepsilon.\]

  • for any \(A \subseteq X\) there exists a set \(B \supseteq A\) such that \[B\in \mathcal E_{\sigma\delta} \quad\text{ and } \quad \mu^{*}(A)=\mu^{*}(B).\]

Proof

  • Let \(A \subseteq X\), then by the definition of \(\mu^{*}\) for every \(\varepsilon>0\) there exists a sequence \((E_j)_{j=1}^{\infty}\subseteq \mathcal E\) such that \[A \subseteq \bigcup_{j \in \mathbb{N}}E_j=U\in \mathcal E_{\sigma} \qquad \text{ and } \qquad \sum_{j \in \mathbb{N}}\rho(E_j) \leq \mu^{*}(A)+\varepsilon.\] Since \(\mu^*=\rho\) on \(\mathcal E\) we obtain that \[\mu^{*}(A) \leq \mu^{*}(U)\le \sum_{j \in \mathbb{N}}\mu^*(E_j)= \sum_{j \in \mathbb{N}}\rho(E_j) \leq \mu^{*}(A)+\varepsilon.\]

  • Let \(A \subseteq X\), for every \(n \in \mathbb{N}\) there is a set \(U_n \in \mathcal E_{\sigma}\) such that \(A \subseteq U_n\) and \[\mu^{*}(A) \leq \mu^{*}(U_n) \leq \mu^{*}(A)+1/n, \qquad {\color{blue} \text{by (a) with }\ \varepsilon=1/n.}\] Define a set \(B=\bigcap_{n \in \mathbb{N}} U_{n}\in \mathcal E_{\sigma\delta}\) and note that \(A\subseteq B\) and \[\mu^{*}(A) \leq \mu^{*}(B) \leq \mu^{*}(U_n) \leq \mu^{*}(A)+1/{n},\] hence \(\mu^{*}(A)=\mu^{*}(B)\) as desired.$$\tag*{$\blacksquare$}$$

Outer regularity of outer measures

Theorem. Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For \(A\subseteq X\) let \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}.\] Suppose that \(\mu^*=\rho\) on \(\mathcal E\).

Then the outer measure is outer regular with respect to the family \(\mathcal E_{\sigma}\), which means that for every \(E\subseteq X\) one has \[\mu^*(E)=\inf\{\mu^*(U): U\supseteq E \text{ and } U\in\mathcal E_{\sigma}\}.\]

Proof

  • Let \[\lambda(E)=\inf\{\mu^*(U): U\supseteq E \text{ and } U\in\mathcal E_{\sigma}\}.\] we have to prove that \[\mu^*(E)=\lambda(E)\] for every \(E\in\mathcal P(X)\).

  • Fix \(E\in\mathcal P(X)\). Then for any \(U\in \mathcal E_{\sigma}\) such that \(E\subseteq U\) we have \(\mu^*(E)\le \mu^*(U)\), which yields that \(\mu^*(E)\le \lambda(E)\).

  • We now prove that \(\lambda(E)\le \mu^*(E)\). Let \(\varepsilon>0\), then by the previous proposition we can find \(U\in\mathcal E_{\sigma}\) such that \(U\supseteq E\) and \[\lambda(E)\le \mu^*(U)\le \mu^*(E)+\varepsilon.\] Since \(\varepsilon>0\) was arbitrary we obtain \[\lambda(E)\le \mu^*(E)\] as desired. $$\tag*{$\blacksquare$}$$

Regularity of the Carathéodory measures

Theorem (assumptions). Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For \(A\subseteq X\) let \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}.\] Let \(\mathcal M(\mu^*)\) be the Carathéodory \(\sigma\)-algebra and suppose that

  • \(\mu^*=\rho\) on \(\mathcal E\);

  • \(\mathcal E\subseteq \mathcal M(\mu^*)\);

  • \(\mu^*\) is a \(\sigma\)-finite measure on \((X, \mathcal M(\mu^*))\).

Regularity of the Carathéodory measures

Theorem (conclusions). Under the assumptions from the previous slide the following are equivalent:

  • \(E\in \mathcal M(\mu^*)\).

  • For every \(\varepsilon >0\) there exists a set \(U\supseteq E\) such that \[U\in \mathcal E_{\sigma} \quad\text{ and } \quad \mu^*(U\setminus E)\le \varepsilon.\]

  • There exists and a set \(G\supseteq E\) such that \[G\in\mathcal E_{\sigma\delta} \quad\text{ and } \quad \mu^*(G\setminus E)=0.\]

  • For every \(\varepsilon >0\) there exists and a set \(C\subseteq E\) such that \[C\in (\mathcal E^c)_{\delta} \quad\text{ and } \quad \mu^*(E\setminus C)\le \varepsilon.\]

  • There exists a set \(F\subseteq E\) such that \[F\in (\mathcal E^c)_{\delta\sigma} \quad\text{ and } \quad \mu^*(E\setminus F)=0.\]

Proof (i) \(\Longrightarrow\) (ii). Assume that \(E\in \mathcal M(\mu^*)\).

  • By the previous proposition for every \(\varepsilon>0\) there is a set \(U\supseteq E\) such that \(U\in \mathcal E_{\sigma}\) and \(\mu^{*}(E) \leq \mu^{*}(U) \leq \mu^{*}(E)+\varepsilon.\) If \(\mu^*(E)<\infty\) we obtain that \(\mu^{*}(U\setminus E)\le \varepsilon\), since by the Carathéodory condition \[\mu^{*}(E)+\mu^{*}(U\setminus E)=\mu^{*}(U\cap E)+\mu^{*}(U\cap E^c)=\mu^{*}(U) \leq \mu^{*}(E)+\varepsilon.\]

  • Assume now that \(\mu^*(E)=\infty\). Since \(\mu^*\) is \(\sigma\)-finite one can write \(X=\bigcup_{n\in \mathbb N}X_n\) for a sequence \((X_n)_{n\in \mathbb N}\subseteq \mathcal M(\mu^*)\) of disjoint sets such that \(\mu^*(X_n)<\infty\). Setting \(E_n=E\cap X_n\) for \(n\in\mathbb N\) one sees that \((E_n)_{n\in \mathbb N}\subseteq \mathcal M(\mu^*)\) and \(E=\bigcup_{n\in \mathbb N}E_n\) and \(\mu^*(E_n)<\infty\).

  • Applying the previous result to \(E_n\) for each \(n\in \mathbb N\) we obtain that for every \(\varepsilon>0\) there exists a set \(U_n\supseteq E_n\) such that \(U_n\in \mathcal E_{\sigma}\) and \[\mu^*(U_n\setminus E_n)\le \varepsilon2^{-n}.\]

  • We take \(U=\bigcup_{n\in \mathbb N}U_n\in \mathcal E_{\sigma}\) and note that \[\begin{aligned} \mu^*(U\setminus E)&=\mu^*\bigg(\bigcup_{n\in \mathbb N}U_n\cap \Big(\bigcup_{n\in \mathbb N}E_n\Big)^c\bigg)\\ &\le \sum_{n\in\mathbb N}\mu^*\bigg(U_n\cap \Big(\bigcup_{n\in \mathbb N}E_n\Big)^c\bigg)\\ &\le \sum_{n\in\mathbb N}\mu^*(U_n\setminus E_n)\\ &\le \sum_{n\in\mathbb N}\varepsilon2^{-n}\le \varepsilon, \end{aligned}\] since \(E_n\subseteq \bigcup_{n\in \mathbb N}E_n\) and \(\mu^*(U_n\setminus E_n)\le \varepsilon2^{-n}\). $$\tag*{$\blacksquare$}$$

Proof (ii) \(\Longrightarrow\) (iii). Assume that for every \(\varepsilon >0\) there exists a set \(U\supseteq E\) such that \(U\in \mathcal E_{\sigma}\) and \(\mu^*(U\setminus E)\le \varepsilon.\)

  • For each \(n\in \mathbb N\) with \(\varepsilon=1/n\) we find a set \(U_n\supseteq E\) such that \(U_n\in \mathcal E_{\sigma}\) and \(\mu^*(U_n\setminus E)\le 1/n\). Define a set \(G=\bigcap_{n\in \mathbb N}U_n\in \mathcal E_{\sigma\delta}\) then \(E\subseteq G\) and \(\mu^*(G\setminus E)\le \mu^*(U_n\setminus E)\le 1/n\). Thus \(\mu^*(G\setminus E)=0\). $$\tag*{$\blacksquare$}$$

Proof (iii) \(\Longrightarrow\) (i). Assume that there exists and a set \(G\supseteq E\) such that \(G\in \mathcal E_{\sigma\delta}\) and \(\mu^*(G\setminus E)=0\).

  • Since \(\mathcal E\subseteq \mathcal M(\mu^*)\) we have that \(\mathcal E_{\sigma\delta}\subseteq\sigma(\mathcal E)\subseteq \mathcal M(\mu^*)\) giving that \(G\in \mathcal M(\mu^*)\). Also \(G\setminus E\in \mathcal M(\mu^*)\) since \(\mu^*(G\setminus E)=0\) and \(\mu^*\) is complete on \((X, \mathcal M(\mu^*))\). Hence \(G\setminus(G\setminus E)\in \mathcal M(\mu^*)\), which implies \(E\in \mathcal M(\mu^*)\), since \(G\setminus(G\setminus E)=G\cap (G\cap E^c)^c=G\cap E=E\). $$\tag*{$\blacksquare$}$$

Proof (i) \(\Longrightarrow\) (iv). Assume that \(E\in \mathcal M(\mu^*)\).

  • Equivalently \(E^c\in \mathcal M(\mu^*)\), since (i) \(\Longleftrightarrow\) (ii) \(\Longleftrightarrow\) (iii), we may apply (ii) to \(E^c\). Then for every \(\varepsilon >0\) we find a set \(U\supseteq E^c\) such that \(U\in \mathcal E_{\sigma}\) and \(\mu^*(U\setminus E^c)\le \varepsilon\). But \(U\setminus E^c=E\setminus U^c\) so it suffices to take \(C=U^c\), then \(C\in (\mathcal E^c)_{\delta}\) and \(C\subseteq E\) and \(\mu^*(E\setminus C)\le \varepsilon\).$$\tag*{$\blacksquare$}$$

Proof (iv) \(\Longrightarrow\) (v). Assume that for every \(\varepsilon >0\) there exists a set \(C\subseteq E\) such that \(C\in (\mathcal E^c)_{\delta}\) and \(\mu^*(E\setminus C)\le \varepsilon.\)

  • For each \(n\in \mathbb N\) with \(\varepsilon=1/n\) we find a set \(C_n\subseteq E\) so that \(C_n\in (\mathcal E^c)_{\delta}\) and \(\mu^*(E\setminus C_n)\le 1/n\). Define a set \(F=\bigcup_{n\in \mathbb N}C_n\in (\mathcal E^c)_{\delta\sigma}\) then \(F\subseteq E\) and \(\mu^*(E\setminus F)\le \mu^*(E\setminus C_n)\le 1/n\). Thus \(\mu^*(E\setminus F)=0\). $$\tag*{$\blacksquare$}$$

Proof (v) \(\Longrightarrow\) (i). Assume that there exists and a set \(F\subseteq E\) such that \(F\in (\mathcal E^c)_{\delta\sigma}\) and \(\mu^*(E\setminus F)=0\).

  • Since \(\mathcal E\subseteq \mathcal M(\mu^*)\) we have that \((\mathcal E^c)_{\delta\sigma}\subseteq\sigma(\mathcal E)\subseteq \mathcal M(\mu^*)\) giving that \(F\in \mathcal M(\mu^*)\). Also \(E\setminus F\in \mathcal M(\mu^*)\) since \(\mu^*(E\setminus F)=0\) and \(\mu^*\) is complete on \((X, \mathcal M(\mu^*))\). Hence \(E=F\cup (E\setminus F)\in \mathcal M(\mu^*)\). $$\tag*{$\blacksquare$}$$

Remark

  • In fact we have proved that \[{\bf (i)\ \Longleftrightarrow \ (ii)\ \Longleftrightarrow \ (iii)} \qquad\text{ and } \qquad {\bf (i)\ \Longleftrightarrow \ (iv)\ \Longleftrightarrow \ (v).}\]

  • In practice, this theorem will be applied with various families \(\mathcal E\) of open sets, like for instance open intervals or open rectangles.

Outer and inner regularity of the Carathéodory measures

Theorem. Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For \(A\subseteq X\) let \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}.\] Let \(\mathcal M(\mu^*)\) be the Carathéodory \(\sigma\)-algebra. Suppose that \((X, \mathcal M(\mu^*), \mu^*)\) is a \(\sigma\)-finite measure space, \(\mathcal E\subseteq \mathcal M(\mu^*)\), and \(\mu^*=\rho\) on \(\mathcal E\). Then

  • \(\mu^*\) is outer regular with respect to the family \(\mathcal E_{\sigma}\), i.e. \[\mu^*(E)=\inf\{\mu^*(U): U\supseteq E \text{ and } U\in\mathcal E_{\sigma}\} \ \text{ for any }\ E\in \mathcal M(\mu^*).\]

  • \(\mu^*\) is inner regular with respect to the family \((\mathcal E^c)_{\delta}\), i.e. \[\mu^*(E)=\sup\{\mu^*(C): C\subseteq E \text{ and } C\in(\mathcal E^c)_{\delta}\}\ \text{ for any }\ E\in \mathcal M(\mu^*).\ \]

Proof. Part (a) has been proved, see outer regularity for outer measures.

  • To prove part (b) let \[\lambda(E)=\sup\{\mu^*(C): C\subseteq E \text{ and } C\in(\mathcal E^c)_{\delta}\}.\] we have to prove that \[\mu^*(E)=\lambda(E)\] for every \(E\in\mathcal M(\mu^*)\).

  • Fix \(E\in\mathcal M(\mu^*)\). Then for any \(C\in (\mathcal E^c)_{\delta}\) such that \(C\subseteq E\) we have \(\mu^*(C)\le \mu^*(E)\), which yields that \(\lambda(E)\le \mu^*(E)\).

  • We now prove that \(\mu^*(E)\le \lambda(E)\). Let \(\varepsilon>0\), then by the previous proposition we can find \(C\in(\mathcal E^c)_{\delta}\) such that \(C\subseteq E\) and \[\mu^*(E\setminus C)\le \varepsilon.\] This implies \(\mu^*(E)\le \lambda(E)\), since \(\varepsilon>0\) is arbitrary and \[\mu^*(E)=\mu^*(C)+ \mu^*(E\setminus C)\le \lambda(E)+ \varepsilon. \tag*{$\blacksquare$}\]

Approximations

Theorem. Let \(X\neq\varnothing\) be a set and let \(\mathcal E\subseteq \mathcal P(X)\) be a collection containing \(\varnothing, X\). Let \(\rho:\mathcal E\to [0, \infty]\) be a set function such that \(\rho(\varnothing)=0\). For \(A\subseteq X\) let \[\mu^{*}(A)=\inf\bigg\{\sum_{j=1}^{\infty}\rho(E_j):(E_j)_{j\in\mathbb N} \subseteq \mathcal{E} \text{ and } A \subseteq \bigcup_{j=1}^{\infty}E_j\bigg\}.\] Let \(\mathcal M(\mu^*)\) be the Carathéodory \(\sigma\)-algebra. Suppose that \((X, \mathcal M(\mu^*), \mu^*)\) is a finite measure space, \(\mathcal E\subseteq \mathcal M(\mu^*)\), and \(\mu^*=\rho\) on \(\mathcal E\). Then for every \(E \in \mathcal M(\mu^*)\) and any \(\varepsilon>0\) there exists \(A \in \mathcal{E}_{\sigma}\) such that \[\mu^*(E \triangle A)<\varepsilon.\] In fact, \(A\) is a finite union of elements from \(\mathcal E\), which is very important.

Proof.
  • Fix \(E \in \mathcal M(\mu^*)\) and \(\varepsilon>0\). By the definition of \(\mu^{*}\) there is a sequence \((E_n)_{n \in \mathbb{N}}\subseteq\mathcal{E}\) so that \[E\subseteq \bigcup_{n \in \mathbb{N}}E_n \quad \text{ and } \quad \mu^{*}(E) \leq \sum_{n \in \mathbb{N}}\rho(E_n) \leq \mu^{*}(E)+\frac{\varepsilon}{2}.\]

  • Since \(\rho=\mu^*\) on \(\mathcal{E}\) thus \[\mu^{*}(E) \leq \sum_{n \in \mathbb{N}}\mu^*(E_n) \leq \mu^{*}(E)+\frac{\varepsilon}{2}.\]

  • For every \(n \in \mathbb{N}\) we have \(\bigcup_{k=1}^{n}E_k \in \mathcal{E}_{\sigma}\) and \(\mathcal{E}_{\sigma}\subseteq \sigma(\mathcal E)\subseteq \mathcal M(\mu^*)\), and consequently by the continuity of measure \[\lim_{n \to \infty}\mu^*\bigg(\bigcup_{k=1}^n E_k\bigg)=\mu^* \bigg(\bigcup_{n \in \mathbb{N}}E_n\bigg).\]

  • Since \(\mu^*\big(\bigcup_{n \in \mathbb{N}}E_n\big) \leq \mu^*(X)<\infty\) so there exist \(N \in \mathbb{N}\) so that \[\mu^*\bigg(\bigcup_{n \in \mathbb{N}}E_n\bigg) -\frac{\varepsilon}{2}<\mu^*\bigg(\bigcup_{n=1}^{N}E_n\bigg).\]

  • Then \[\mu^*\bigg(\bigcup_{n \in \mathbb{N}} E_n \setminus \bigcup_{k=1}^{N}E_k\bigg) =\mu^*\bigg(\bigcup_{n \in \mathbb{N}}E_n\bigg)-\mu^*\bigg(\bigcup_{k=1}^{N}E_k\bigg)<\frac{\varepsilon}{2}.\]

  • Let \(A = \bigcup_{n=1}^{N}E_n \in \mathcal{E}_{\sigma}\) and since \(E \subseteq \bigcup_{n \in \mathbb{N}}E_n\) we have \[\mu^*(E \setminus A) \leq \mu^* \bigg(\bigcup_{n \in \mathbb{N}}E_n \setminus A\bigg)<\frac{\varepsilon}{2}.\]

  • Similarly since \(A \subseteq \bigcup_{n \in \mathbb{N}}E_n\) and \(E \subseteq \bigcup_{n \in \mathbb{N}}E_n\) we have \[\begin{aligned} \mu^*(A \setminus E) &\leq \mu^* \left(\bigcup_{n \in \mathbb{N}}(E_n \setminus E)\right)\\ &=\mu^* \bigg(\bigcup_{n \in \mathbb{N}}E_n\bigg)-\mu^*(E) \\ &\leq \sum_{n \in \mathbb{N}}\mu^*(E_n)-\mu^*(E) \leq \frac{\varepsilon}{2}. \end{aligned}\]

  • Thus \[\mu^*(E \triangle A)=\mu^*(E \setminus A)+\mu^*(A \setminus E)<\frac{\varepsilon}{2}+\frac{\varepsilon}{2}=\varepsilon.\] as desired. $$\tag*{$\blacksquare$}$$

Top