Let $f: \Omega \to \mathbb{R}$ be a measurable function defined on a canonical probability space $(\Omega, \mathcal{F}, \mathbb{P})$. Here, we can set $\Omega := \mathbb{R}^n$, and the measurability always refers to Borel measurability.

It is well-know that if $f \geq 0$ $\mathbb{P}$-a.e., then $\int_\Omega f (x) \, \mathbb{P}(\mathrm{d} x) \geq 0$.

I am curious that could we get a stricter result such that $\int_\Omega f (x) \, \mathbb{P}(\mathrm{d} x) > 0$ whenever $f(x) > 0$ for each $ x \in \Omega$?

The answer seems to be YES, but I am not sure how to prove it rigorously. Could anyone help me out please?

My attempt:

I proved the case where $f$ is continuous (rather than just measurable) on $\Omega$. To see this, pick any compact subset $K \subset \Omega$, and then $f(x) \geq \eta= \max_{x \in K} f(x) >0$ on $K$.

It then follows that $\int_{\Omega} f(x) \, \mathbb{P}(\mathrm{d} x) \geq \int_{K} f(x) \, \mathbb{P}(\mathrm{d} x) \geq \int_{K} \eta \, \mathbb{P}(\mathrm{d} x) = \eta >0.$

However, I did not know how to show such result for the case of a measurable function $f$ on $\Omega$.

Would you mind giving me some guidance or hint on it please? Thank you so much in advance!


If $f(x)>0$ for all $x$, then for some positive integer $n$ the set $A_n=\{x:f(x)>1/n\}$ has positive measure (since $\bigcup_n A_n=\Omega$). We thus have $$\int_\Omega f dP\geq \int_{A_n} f dP\geq \int_{A_n} \frac{1}{n}dP=\frac{P(A_n)}{n}>0.$$

  • $\begingroup$ Thank you very much @Eric Wofsey. $\endgroup$ – Paradiesvogel May 29 '18 at 1:32

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.