Characteristic Function Measurable iff Set Measurable
Theorem
Let $\left({X, \Sigma}\right)$ be a measurable space.
Let $E \subseteq X$.
Then the following are equivalent:
- $(1): \quad E \in \Sigma$; i.e., $E$ is a $\Sigma$-measurable set
- $(2): \quad \chi_E: X \to \left\{{0, 1}\right\}$, the characteristic function of $E$, is $\Sigma$-measurable
Proof
$(1)$ implies $(2)$
Assume that $E \in \Sigma$.
It is clear that $x \notin \left\{{0, 1}\right\}$ implies $\chi_E^{-1} \left({x}\right) = \varnothing$.
Hence Preimage of Union under Mapping and Characteristic Function Determined by 1-Fiber yield, for any $\alpha \in \R$:
- $\left\{{x \in X: \chi_E \left({x}\right) \ge \alpha}\right\} = \begin{cases}\varnothing & \text{if $1 < \alpha$}\\ E & \text{if $0 < \alpha \le 1$}\\ X & \text{if $\alpha \le 0$}\end{cases}$
Since $\Sigma$ is a $\sigma$-algebra, $X \in \Sigma$.
By assumption, also $E \in \Sigma$.
Lastly, Sigma-Algebra Contains Empty Set ensures $\varnothing \in \Sigma$.
This establishes $(4)$ of Characterization of Measurable Functions.
Hence $\chi_E$ is $\Sigma$-measurable.
$\Box$
$(2)$ implies $(1)$
Assume that $\chi_E$ is $\Sigma$-measurable.
Since for all $x \in X$, it holds that:
- $\chi_E \left({x}\right) > 0 \iff x \in E$
it follows that:
- $E = \chi^{-1} \left({\left({0 \,.\,.\, +\infty}\right)}\right)$
and thus $E \in \Sigma$ as $\chi_E$ is $\Sigma$-measurable.
$\blacksquare$
Sources
- 2005: René L. Schilling: Measures, Integrals and Martingales ... (previous) ... (next): $8.5 \ \text{(i)}$