Let $L_p([a,b];\mathbb{R}) = \{f:[a,b]\to\mathbb{R} | f$ is Riemann-integrable $\}$. I want to prove that $$ d_p(f,g) = \left(\int_a^b |f(x)-g(x)|^p dx\right)^{1/p} $$ defines a metric in $L_p([a,b],\mathbb{R}).$

Given $f,g$ integrable I easily see that $d_p(f,g)$ exists and that $d_p(f,f)=0$ and $d_p(f,g) = d_p(g,f)$. I also proved the triangle inequality by proving the Hölder inequality for positive real functions. I'm just stuck in the condition $f\neq g \to d_p(f,g)>0$. For simplicity I'm assuming $f \neq 0$ and trying to prove $\int |f|^p \neq 0$.

I know that $f\ge 0$ implies $\int f \ge 0$ but I don't see how to get the strict inequality. Any help or hint will be appreciated.

  • 1
    $\begingroup$ Actually the space should be altered a bit, since $f =0 $ almost everywhere on $[a,b]$ would deduce that $\int_a^b |f|^p = 0$. $\endgroup$ – xbh Sep 5 '18 at 9:24
  • $\begingroup$ Will continuos functions work? $\endgroup$ – AnalyticHarmony Sep 5 '18 at 13:03
  • 1
    $\begingroup$ Yes, that would avoid the possibility that $\int |f - g|^p = 0$ but $f \neq g$, since $\int |f|^p = 0$ and $f$ be continuous would deduce that $f \equiv 0$. This claim could be proved by contradiction. $\endgroup$ – xbh Sep 5 '18 at 13:08
  • $\begingroup$ So the usual notation $L_p$ are used for spaces of continuous functions? $\endgroup$ – AnalyticHarmony Sep 5 '18 at 13:09
  • 1
    $\begingroup$ Not exactly. Actually $L$ stands for "Lebesgue" integration. Generally $R$ would be used for Riemann integrals. For continuous functions, try $C$, or $C_p$ to indicate the metric. $\endgroup$ – xbh Sep 5 '18 at 13:12

You can't prove it because it is not true. If$$f(x)=\begin{cases}1&\text{ if }x=a\\0&\text{ otherwise,}\end{cases}$$then $d_p(f,0)=0$, but $f\neq0$. The function $d_p$ is a pseudometric, but not a metric.

  • $\begingroup$ If I change the set for continuous functions then it'll be a metric? Or is the problem in the defining function per se? $\endgroup$ – AnalyticHarmony Sep 5 '18 at 13:04
  • 2
    $\begingroup$ @AnalyticHarmony Yes, if you restrict $d_p$ to the space $\mathcal{C}([a,b])$, then it becomes a metric. $\endgroup$ – José Carlos Santos Sep 5 '18 at 16:31

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.