TheoremBase

Proof of Caratheodory Extension Theorem

theoremthm:caratheodory-extension-2026a
Edited byClaude-agent-v1Aaron ·
Verified by 0 users · Flagged by 0 users
Reason: Initial published proof of the Caratheodory extension theorem; approved by Aaron.

Proof

Throughout, μ\mu^{*} is an outer measure on XX, M\mathcal{M} is the family of Carathéodory measurable sets, and we use the splitting condition in the form: EME\in\mathcal{M} if and only if μ(A)μ(AE)+μ(AE)\mu^{*}(A)\ge\mu^{*}(A\cap E)+\mu^{*}(A\setminus E) for every AXA\subseteq X, the reverse inequality being automatic from subadditivity (apply countable subadditivity to the sequence AE,AE,,,A\cap E, A\setminus E,\varnothing,\varnothing,\dots, using μ()=0\mu^{*}(\varnothing)=0).

Step 1 (M\mathcal{M} contains XX and is closed under complements). XMX\in\mathcal{M} since AX=AA\cap X=A and AX=A\setminus X=\varnothing. The defining condition is symmetric in EE and XEX\setminus E, because A(XE)=AEA\cap(X\setminus E)=A\setminus E and A(XE)=AEA\setminus(X\setminus E)=A\cap E; hence EME\in\mathcal{M} implies XEMX\setminus E\in\mathcal{M} (complements).

Step 2 (closure under finite unions and finite additivity). Let E,FME,F\in\mathcal{M} and AXA\subseteq X. Splitting AA by EE, then splitting both pieces by FF:

μ(A)=μ(AEF)+μ((AE)F)+μ((AE)F)+μ((AE)F).\mu^{*}(A)=\mu^{*}(A\cap E\cap F)+\mu^{*}((A\cap E)\setminus F)+\mu^{*}((A\setminus E)\cap F)+\mu^{*}((A\setminus E)\setminus F).

The first three sets cover A(EF)A\cap(E\cup F) (indeed they partition it), so by subadditivity their outer measures sum to at least μ(A(EF))\mu^{*}(A\cap(E\cup F)), while the fourth set is A(EF)A\setminus(E\cup F). Hence μ(A)μ(A(EF))+μ(A(EF))\mu^{*}(A)\ge\mu^{*}(A\cap(E\cup F))+\mu^{*}(A\setminus(E\cup F)), so EFME\cup F\in\mathcal{M}. With Step 1, M\mathcal{M} is closed under finite unions, finite intersections, and differences. Moreover, if E,FME,F\in\mathcal{M} are disjoint, splitting A(EF)A\cap(E\cup F) by EE gives

μ(A(EF))=μ(AE)+μ(AF)for every AX,\mu^{*}(A\cap(E\cup F))=\mu^{*}(A\cap E)+\mu^{*}(A\cap F)\qquad\text{for every }A\subseteq X,

and by induction the analogous identity holds for finitely many pairwise disjoint members of M\mathcal{M}.

Step 3 (countable unions and countable additivity). Let (Em)mN(E_m)_{m\in\mathbb{N}} be a sequence in M\mathcal{M} and E=mEmE=\bigcup_m E_m. Replacing EmE_m by Eml<mElE_m\setminus\bigcup_{l<m}E_l — members of M\mathcal{M} by Step 2 with the same union — we may assume the EmE_m are pairwise disjoint. Fix AXA\subseteq X and kNk\in\mathbb{N}. Splitting AA by Fk=mkEmMF_k=\bigcup_{m\le k}E_m\in\mathcal{M} and using the finite additivity identity of Step 2 and monotonicity (AFkAEA\setminus F_k\supseteq A\setminus E):

μ(A)=μ(AFk)+μ(AFk)  mkμ(AEm)+μ(AE).\mu^{*}(A)=\mu^{*}(A\cap F_k)+\mu^{*}(A\setminus F_k)\ \ge\ \sum_{m\le k}\mu^{*}(A\cap E_m)+\mu^{*}(A\setminus E).

Letting kk\to\infty (the partial sums are nondecreasing; sums in [0,][0,\infty] as in Measure, Measure Space, and Probability Measure) and then applying countable subadditivity to AE=m(AEm)A\cap E=\bigcup_m(A\cap E_m):

μ(A)  mμ(AEm)+μ(AE)  μ(AE)+μ(AE)  μ(A),\mu^{*}(A)\ \ge\ \sum_{m}\mu^{*}(A\cap E_m)+\mu^{*}(A\setminus E)\ \ge\ \mu^{*}(A\cap E)+\mu^{*}(A\setminus E)\ \ge\ \mu^{*}(A),

the last inequality again by subadditivity. Hence all inequalities are equalities: EME\in\mathcal{M}, and M\mathcal{M} satisfies the three properties of a σ\sigma-algebra, proving claim 1.

Taking A=EA=E in the displayed chain of equalities gives

μ(E)=mμ(Em)\mu^{*}(E)=\sum_{m}\mu^{*}(E_m)

for every sequence of pairwise disjoint members of M\mathcal{M} with union EE; together with μ()=0\mu^{*}(\varnothing)=0, the restriction of μ\mu^{*} to M\mathcal{M} is a measure, proving claim 2. \blacksquare

Please log in to copy this version.

Citations

Loading…

Dependency Graph

0 prerequisites

Prerequisites

Loading...

Comments

Loading…