Let Ω be the sample space, J be a semialgebra of subsets. Suppose P:J→[0,1] such that 1. P(∅)=0, P(Ω)=1 2. Finite superadditivity: For A1,…,An∈J disjoint and ⋃i=1nAi∈J then P(i=1⋃nAi)≥i=1∑nP(Ai) 3. Countable Monotonicity: For A,A1,…,An∈J s.t. A⊆⋃i≥1Ai then P(A)≤i≥1∑P(Ai)
\begin{proof} We first begin with some definitions and lemmas:
For (Ω,J) we say the outer measureP∗ on A⊆Ω is defined as follow: P∗(A)=inf{i∑P(Ai):A1,A2,…,∈J,A⊆i⋃Ai}
For any (Bi)i≥1⊆Ω, we have that the outer measure is countably subadditive: P∗(i⋃Bi)≤i∑P∗(Bi)
A⊆Ω is Carathéodory measurable if P∗(E)=P∗(E∩A)+P∗(E∩Ac)∀E⊆Ωand M={A⊆Ω:A Caratheˊodory measurable}is the set of Carathéodory measurable sets.
For A1,A2,⋯∈M disjoint we have that the outer measure is countably additive: P∗(i⋃Ai)=i∑P∗(Ai)
Step 1: prove For A1∈M,A2⊆Ω, such that A1,A2 are disjoint P∗(A1∩A2)=P∗(A1∩(A1∪A2))+P∗(A1c∩(A1∪A2))=P∗(A1)+P∗(A2)Where the first equality is done by applying the definition of and the second by disjoint property of both events. By induction, for A1,…,An∈M disjoint (by and finite superadditivity) P∗(A1∪⋯∪An)=i=1∑nP∗(Ai)Thus P∗(i≥1⋃Ai)≥P∗(i=1⋃nAi)=i=1∑nP∗(Ai)∀n≥1⟹P∗(∪iAi)≥i≥1∑P∗(Ai)But, by we have that P∗(∪iAi)=i≥1∑P∗(Ai)Step 2: Show M is a σ-algebra where M:={A⊆Ω:P∗(E)=P∗(E∩A)+P∗(E∩Ac)∀E⊆Ω}a) Show M is a algebra 1. ∅,Ω∈M ✅ 2. A∈M⟹Ac∈M ✅ 3. Let A,B∈M, E⊆Ω, then P∗((A∩B)∩E)+P∗((A∩B)c∩E)=P∗(A∩B∩E)+P∗((Ac∩B∩E)∪(A∩Bc∩E)∪(Ac∩Bc∩E))≤P∗(A∩B∩E)+P∗(Ac∩B∩E)+P∗(A∩Bc∩E)+Pc(Ac∩Bc∩E)=P∗(B∩E)+P∗(Bc∩E)=P∗(E) where the last two arguments are because A,B∈M. Since this is the the one side of the inequality we wanted to prove (the other is trivial) we have that A∩B∈M.
b) Lemmas… We now state some additional lemmas: >[!lemma|2.3.11] >Let A1,A2,⋯∈M be disjoint. For each n∈N, let Bn=⋃i=1nAi. Then ∀n∈N,∀E⊆Ω we have P∗(E∩Bn)=i=1∑nP∗(E∩Ai)
\begin{proof} Using we just use the fact that for any A1,A2,⋯∈M we have that i⋃Ai=disjointA1∪(A2∖A1)∪(A3∖(A1∪A2))∪…∈M\end{proof}
Step 3: J⊆M:
J⊆M
\begin{proof} Let A∈J. Since J is a semialgebra we can write Ac=J1⊔⋯⊔Jk for some disjoint J1,…,Jk∈J. Also, for any E⊆Ω and ϵ>0 by the definition of : we can find A1,A2,⋯∈J with E⊆⋃nAn and ∑nP(An)≤P∗(E)+ϵ. Then
P∗(E∩A)+P∗(E∩Ac)≤P∗((n⋃An)∩A)+P∗((n⋃An)∩Ac)=P∗(n⋃(An∩A))+P∗(n⋃i=1⋃k(An∩Ji))≤n∑P∗(An∩A)+n∑i=1∑kP∗(An∩Ji)=n∑P(An∩A)+n∑i=1∑kP(An∩Ji)=n∑(P(An∩A)+i=1∑kP(An∩Ji))≤n∑P(An)≤P∗(E)+ϵmonotonicitydefinition of semialgebrasubadditivitysince P∗∣J=Psuperadditivity then Caratheˊodoryby assumption. This is true ∀ϵ>0, hence since ϵ is arbitrary, P∗(E∩A)+P∗(E∩Ac)≤P∗(E),∀E⊆Ω. Since the other direction is trivial we have that the condition holds giving us A∈M, since this is for any A∈J we have J⊆M. \end{proof} All these lemmas prove what we needed to prove ✅
\end{proof} # Extensions of the Extension Theorem
Let J be a semialgebra of subsets of Ω. Let P:J→[0,1] such that: 1. P(Ω)=1, 2. : P(n⋃In)=n∑P(In) for I1,I2,⋯∈J disjoint with n⨆In∈J.
Whereas requires finite superadditivity and countable monotonicity under disjoint sets this one requires countable additivity. The only things we need to prove in this case are Probability Measure and under non-disjoint sets.
Let Ω be the sample space, J a semialgebra of subsets of Ω, and P:J→[0,1] such that (Ω,M,P∗) is a probability space. If (Ω,F,Q) is a prob. space, P=Q on J, and J⊆F⊆M then Q(A)=P∗(A),∀A∈F
Let P and Q be two probability measures defined on the collection B of Borel subsets of R. Suppose P((−∞,x])=Q((−∞,x]),∀x∈R.Then P(A)=Q(A),∀A∈B