This is a problem from Billingsley's text Probability and measure.

Suppose that $f$ is nonnegative on a $\sigma$-finite measure space $(\Omega, \mathscr{F}, \mu)$. Show that $$\int_\Omega f d \mu = (\mu \times \lambda)[(\omega, y) \in \Omega \times \mathbb{R}^1: 0 \leq y \leq f(\omega)]. \tag{1}$$ Prove that the set on the right is measurable.

My Attempt:

Denote the set on the right by $G$. First, by condition, the product measure $\mu \times \lambda$ exists on the $\sigma$-field $\mathscr{F} \times \mathscr{R}^1$. I am able to show that $G$ is measurable and $(1)$ holds if $f$ are simple functions. It looks naturally that for general nonnegative $f$, let $\{f_n\}$ be a sequence of simple functions such that $f_n \uparrow f$. Then define $G_n = [(\omega, y) \in \Omega \times \mathbb{R}^1: 0 \leq y \leq f_n(\omega)]$ accordingly for every $n$. It is expected that $G_n \uparrow G$ so that the measurability of $G$ follows from the measurability of $G_n$. However, this seems unjustified, since in fact $$G_n \uparrow [(\omega, y) \in \Omega \times \mathbb{R}^1: 0 \leq y \color{red}{<} f(\omega)].$$ So we have to show that $$[(\omega, y) \in \Omega \times \mathbb{R}^1: f(\omega) = y, y \geq 0] = [(\omega, f(\omega)): f(\omega) \geq 0] \tag{2}$$ is measurable. And here is where I got stuck. Can someone show me why $(2)$ is measurable or to show the assertion through other ways? Thank you very much.


First of all you need to assume that $f$ is $(\cal F, B)$-measurable, where $\cal B$ is the Borel sigma algebra on $\Bbb R$.

Define a map $g: \Omega \times \Bbb R \to \Bbb R$ as $(\omega, y) \mapsto f(\omega)-y$. Note that $g$ is $(\cal F \otimes B, B)$-measurable, since it is the difference of two $(\cal F \otimes B, B)$-measurable functions, namely $h_1(\omega, y) := f(\omega)$ and $h_2(\omega, y) := y$.

It follows that the set $A:=g^{-1}[0,\infty)$ is measurable. Hence $A \cap B$ is measurable, where $B=\Omega \times [0,\infty) \in \cal F \otimes B$. But $A \cap B$ is precisely the set you are trying to prove is measurable.

  • $\begingroup$ This looks good to me, thank you. But in your way, is it also easy to show that $(1)$ also holds? $\endgroup$ – Zhanxiong Oct 24 '15 at 21:02
  • $\begingroup$ I got it, $(1)$ then follows from the Fubini's theorem. $\endgroup$ – Zhanxiong Oct 24 '15 at 21:05
  • $\begingroup$ Exactly, just let $D$ be that set. And then notice that $f(\omega) = \int_{\Bbb R} 1_D(\omega,y) dy$ and integrate both sides w.r.t $\mu$. Then apply Fubini. $\endgroup$ – Shalop Oct 24 '15 at 21:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.