0
$\begingroup$

I'm reading the following proof:

Suppose $\mu(X) < \infty$ and $0 < p< q \leq \infty$. If $q=\infty$ then $L^{q}(\mu) \subseteq L^{p}(\mu)$.

Proof: If $q=\infty$ then $\int_X |f|^{p}d\mu \leq (||f||_{\infty})^{p} \int_X 1d\mu$.

Why we have the above inequality??

  • 0
    What ways do you know of bounding an integral such as $\int|f|^p$?2010-11-10
  • 0
    Certainly we always have $ |f| \leq ||f||_{\sup}$. But here the $||.||_{\infty}$ denotes the seminorm of essentially bounded functions not the usual supremum norm. So that's why I'm confused.2010-11-10
  • 0
    @user: The first thing we prove after define the norm $\| \cdot \|\_{\infty}$ is that $|f(x)| \le \| f \|_{\infty}$ almost everywhere. With that at hand we prove that $\| f \|\_{\infty} = 0 $ if and only if $f(x) = 0$ almost everywhere.2010-11-10
  • 0
    Isn't this inequality true for all p and q?2010-11-10
  • 0
    @user1736: consider X = (0,1) with \mu a standard Lebesgue measure on X. Then x^{-1/2} is certainly in L^1 but not in L^2.2010-11-10
  • 0
    @user1736: I just realized that I provided counter-example for the reversed relation. If I am reading it correctly this time, then yes: the relation holds, assuming 1 \leq p < q.2010-11-10
  • 0
    @user1736: Do you mean the inclusion $L^q(\mu) \subseteq L^p(\mu)$ if $1 \le p < q \le \infty $? If $\mu$ is a finite measure then it works for all p and q, but to prove the inclusion with $q \ne \infty$ we need Hölder's inequality. If $\mu$ isn't a finite measure you can find functions $f \in L^q(\mu)$ and $f \not \in L^p(\mu)$. You can probably find an example by yourself considere $\mu$ the Lebesgue measure.2010-11-10

2 Answers 2