TheoremComplete

Sigma-Algebras and Measures - Main Theorem

TheoremMonotone Class Theorem

Let A\mathcal{A} be an algebra of subsets of XX. Let M\mathcal{M} be a monotone class containing A\mathcal{A}. Then M\mathcal{M} contains the sigma-algebra σ(A)\sigma(\mathcal{A}) generated by A\mathcal{A}.

A collection M\mathcal{M} is a monotone class if:

  1. Whenever A1A2A_1 \subseteq A_2 \subseteq \cdots with AiMA_i \in \mathcal{M}, then i=1AiM\bigcup_{i=1}^{\infty} A_i \in \mathcal{M}
  2. Whenever A1A2A_1 \supseteq A_2 \supseteq \cdots with AiMA_i \in \mathcal{M}, then i=1AiM\bigcap_{i=1}^{\infty} A_i \in \mathcal{M}

The Monotone Class Theorem is a powerful tool for proving that two measures agree on a sigma-algebra. It reduces the problem to checking agreement on a simpler algebra that generates the sigma-algebra.

ExampleApplication to Uniqueness

Suppose μ\mu and ν\nu are two measures on (X,F)(X, \mathcal{F}). If they agree on an algebra A\mathcal{A} that generates F\mathcal{F} (i.e., σ(A)=F\sigma(\mathcal{A}) = \mathcal{F}), and if there exists a sequence XnAX_n \in \mathcal{A} with XnXX_n \uparrow X and μ(Xn)=ν(Xn)<\mu(X_n) = \nu(X_n) < \infty, then μ=ν\mu = \nu on F\mathcal{F}.

This is particularly useful when A\mathcal{A} is the collection of all finite unions of intervals, which generates the Borel sigma-algebra.

Remark

The theorem's power lies in its ability to extend results from algebras to sigma-algebras. An algebra A\mathcal{A} of sets satisfies:

  1. XAX \in \mathcal{A}
  2. If AAA \in \mathcal{A}, then AcAA^c \in \mathcal{A}
  3. If A,BAA, B \in \mathcal{A}, then ABAA \cup B \in \mathcal{A} (finite unions only)

The difference between algebras and sigma-algebras is that algebras only require closure under finite unions, while sigma-algebras require closure under countable unions.

The Monotone Class Theorem is instrumental in establishing uniqueness results for measures, particularly for extending measures from simple sets to more complex sigma-algebras. It bridges the gap between algebraic simplicity and the full power of sigma-algebras needed for measure theory.