Let $f\in L^1(\mathbb R)$ s.t. $$\int_{\mathbb R}|f|=0\implies f=0\ a.e.$$

My attempt

Suppose $f$ continuous and that there is $y$ s.t. $|f(y)|\neq 0$. In particular, there is $\delta >0$ s.t. $f(x)\neq 0$ for all $x\in [y-\delta ,y+\delta ]$. By continuity, there is $m>0$ s.t. $|f(x)|\geq m$ for all $x\in [y-\delta ,y+\delta ]$. Therefore, $0<2m\delta\leq \int_{[y-\delta ,y+\delta ]}|f(x)|\leq \int_{\mathbb R}|f|,$ which is a contradiction. Therefore $f=0$ everywhere.

Since $f\in L^1$. There is a sequence of continuous function s.t. $$\|f_n-f\|_{L^1}\to 0.$$ In particular, $$\int_{\mathbb R}|f_n|\to \int_{\mathbb R}|f|=0.$$

How can I prove that $f_n=0$ for all $n$ ?

  • 2
    $\begingroup$ Why would $f$ be continuous? How can you suppose it? Suppose instead $f > 0$ on non null set. $\endgroup$ – dEmigOd Feb 16 at 10:19
  • 1
    $\begingroup$ @dEmigOd. If you continue reading the proof you will see that he first considers the case where $f$ is continuous, and then approximates a general $f \in L^1$ with continuous functions. $\endgroup$ – md2perpe Feb 16 at 10:51

There is no reason to have $f_n=0$ for all $n$... Your proof would be fine if $$\left\{f\in L^1(\mathbb R)\mid \int_{\mathbb R}|f|=0\right\},$$ would be a dense subspace of $L^1(\mathbb R)$. Since it's not the case, your proof can't work.

For a proof, let $E=\{x\mid |f(x)|>0\}$. Suppose that $m(E)>0$. Let $$E_n=\left\{x\mid |f(x)|\geq\frac{1}{n}\right\}.$$ Then, $$\frac{1}{n}\boldsymbol 1_{E_n}<|f|\boldsymbol 1_{E_n} \leq |f|,$$ and thus, $$\frac{1}{n}m(E_n)\leq \int_{\mathbb R}|f|=0.$$ Therefore, $m(E_n)=0$ for all $n$. Since $\lim_{n\to \infty }m(E_n)=m(E)$, you get a contradiction.


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.