I would like to show that

If $f\in C^0([a,b]),p\geq 1$ then $\int^b_a|f(x)|^pdx=0$ implies $f(x)=0$ $\forall x \in [a,b].$

This is how I argue:

Since $|f(x)|\geq 0$ and $f$ is continuous then the only way the integral can equal $0$ is if $f(x)=0$ for all $x \in [a,b]$

Would this be wrong or not rigorous enough?

  • 1
    $\begingroup$ No, not rigorous. Where did you use continuity? $\endgroup$ – Kavi Rama Murthy May 19 '20 at 12:19
  • $\begingroup$ @KaviRamaMurthy Do I need to use it? It is said in the definition of the function. But could you explain what exactly i'm missing? $\endgroup$ – user634512 May 19 '20 at 12:20
  • $\begingroup$ in your proof what happens when p is 0 $\endgroup$ – Monocerotis May 19 '20 at 12:21
  • 1
    $\begingroup$ If $f(x)=0$ for $x \neq 0$ and $1$ for $x=0$ then $\int |f|^{p}$ exists and equals $0$ but the function is not $0$. Continuity is very important. $\endgroup$ – Kavi Rama Murthy May 19 '20 at 12:23
  • 1
    $\begingroup$ Why is it that the only way the integral can be zero is when $f\equiv 0$? $\endgroup$ – Sam May 19 '20 at 12:25

As mentioned in the comments your argument was not rigorous as you didn't justify your statement: "the only way the integral can equal $0$ is if $f(x)=0$ for all $x \in [a,b]$."

Proof: Suppose there exists $x_0 \in [a,b]$ such that $f(x_0) \neq 0.$ Then $\epsilon=|f(x_0)|^p>0.$ Continuity of $f$ implies there exists $\delta>0$ such that \begin{align*}x\in[a,b],|x-x_0|<\delta &\implies ||f|^p(x)-|f|^p(x_0)|<\frac{\epsilon}{2}\\&\implies ||f|^p(x)-\epsilon|<\frac{\epsilon}{2}\\&\implies|f(x)|^p>\frac{\epsilon}{2}.\end{align*}

Therefore $$\int_a^b |f(x)|^p\,dx\geq \int_{x_0-\delta}^{x_0+\delta}|f(x)|^p\,dx>\epsilon\delta>0$$ which is a contradiction.

(Note: You can always choose $\delta>0$ small enough so that $(x_0-\delta,x_0+\delta)\subseteq [a,b].$)

  • 2
    $\begingroup$ just a small doubt from a student. Here $||f|^p(x)-|f|^p(x_0)|<\frac{\epsilon}{2}$ why have you specifically chosen $\frac{\epsilon}{2}$ please... $\endgroup$ – Monocerotis May 19 '20 at 14:12
  • $\begingroup$ @Monocerotis Oh you could choose anything less than $\epsilon,$ just to obtain the final term is $>0.$ $\endgroup$ – Sahiba Arora May 19 '20 at 16:32
  • 1
    $\begingroup$ got it.thanks a lot! $\endgroup$ – Monocerotis May 20 '20 at 2:32
  • 1
    $\begingroup$ Many compliments for your career. $\endgroup$ – Sebastiano May 20 '20 at 22:46

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.