This is a corollary to a proof in Bass, but I don't understand why it follows from the proof he gives. I follow everything up until the last statement. Why is it that proving that the integral is $0$ for all Borel measurable sets guarantees it is true for all Lebesgue measurable sets? Am I missing something vital about the relationship between Lebesgue measurable sets and Borel measurable sets?

Statement: Let $m$ be Lebesgue measure and $a \in\mathbb R$. Suppose that $f : \mathbb R \to\mathbb R$ is integrable, and $\int_a^xf(y) \, dy = 0$ for all $x$. Then $f = 0$ a.e.

Proof: We prove the conclusion first for intervals, then for unions of intervals, then for open sets, then for Borel measurable sets. For any interval $(c,d)$, we have $$\int_c^df = \int_a^d f - \int^c_a f = 0.$$ By linearity, if $G$ is the finite union of disjoint open intervals, then $\int_G f = 0$. Now, if $G$ is open, then $G$ is the countable union of disjoint, open intervals, $G = \bigcup^n_{i=1}I_i$. We have $$\int_G f = \int \lim_{n\to\infty} f \cdot \chi_{\bigcup^n_{i=1} I_i}$$ and since $f_n$ is integrable for all $n$, and $|f_n| \le f$ for all $n$, by the dominated convergence theorem, $$\int_G f = \lim_{n\to\infty} \int_{\bigcup^n_{i=1}I_i} f =0.$$ Now, if $G_n$ is a decreasing sequence of open sets converging to $H$, since $|f \cdot \chi_{G_n}| \le f \cdot \chi_{G_1}$ for all $n$, we again have by the DCT that $\int_H f = 0 = \lim_{n\to\infty} \int_{G_n}f = 0$.

Finally, if $G$ is a Borel measurable set, then we know that there exists a sequence $G_n$ of decreasing, open sets, that converge to some set $H$, where $H \setminus G$ is a null set. Thus, $$\int_G f = \int_H f = 0.$$ We have found that for every Borel measurable set $G$, $\int_G f =0$. Since $f$ is real valued and measurable, $f = 0$ a.e.

  • 1
    $\begingroup$ You need the characterization of measurable sets: they are the union of a Borel set and a set of measure zero. $\endgroup$ – Crostul Aug 5 '15 at 18:32
  • $\begingroup$ Perfect, that's exactly what I was looking for! Thank you. You are welcome to make your comment an answer. $\endgroup$ – poppy3345 Aug 5 '15 at 18:34
  • $\begingroup$ This question has been asked before on this site. Maybe someone can find the original and mark as duplicate. $\endgroup$ – Nate Eldredge Aug 5 '15 at 20:18

You need the characterization of measurable sets:

Let $G \subset \Bbb{R}^k$ a bounded set. Then the following are equivalent:

  1. $G$ is Lebesgue measurable.

  2. There exist $B, N \subset \Bbb{R}^k$ such that $G = B \cup N$, $B$ is Borel, $N$ has zero Lebesgue measure.

Now, to conclude your exercize call $G= \{ x : f(x) >0\}$ and write $G=\bigcup_n (G \cap [-n,n])$ as a union of bounded measurable sets. Use the characterization to write $G \cap [-n,n] = B_n \cup N_n$. Finally $$\int_{G \cap [-n,n]} f = \int_{B_n} f = 0$$ So $f=0$ a.e. in $G \cap [-n,n]$. Since $n$ is arbitrary, you have $f=0$ a.e. in $G$.

Do the same thing with $H=\{ x : f(x) <0\}$ and you are done.


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.