2

Let $f$ be continuous on $[a,b]$ with $f(x) \geq 0$ for all $x \in [a,b]$. Prove

$$\lim_{n \to \infty} \left( \int_{a}^{b} (f(x))^n \, dx\right)^{\frac{1}{n}} = \max_{x \in [a,b]} \{f(x)\}.$$

Attempted Solution:

Note first that it suffices to prove the above for $\max_{x \in [a,b]} \{f(x)\} = 1$, since we may scale accordingly. Let $f$ attain it's maximum at a point $c$. Let $\epsilon >0$. Since $f$ is uniformly continuous, there exists $\delta > 0$ so that $f(x) \in (1-\epsilon,1]$ whenever $x \in (c-\delta,c+\delta)$. If there exist multiple points at which $f(x) = 1$, let $$D = \bigcup_{i \in I} B_\delta(c_i),$$ an open set with finite measure. Now, partition $D$ into disjoint subsets, $D_i$, each having finite measure. We have on each $D_i$, $(f(x))^n \in ((1-\epsilon)^n,1]$. Now, $$(1-\epsilon)(\mu(D))^{\frac{1}{n}} \leq \left( \int_{a}^{b} (f(x))^n \, dx \right)^\frac{1}{n} \leq (b-a)^\frac{1}{n}$$ and since $\lim \limits_{n \to \infty} x^\frac{1}{n} = 1$,

$$(1-\epsilon) \leq \lim_{n \to \infty} \left( \int_{a}^{b} (f(x))^n \, dx \right)^{\frac{1}{n}} \leq 1$$

as desired.

Any issues here? Anything to make it more succinct? Moreover, where does continuity come into play to make this an essentially simpler proof than the general $L^p$ norm limit proof I've just found here?

  • 1
    I don't think your proof is essential simpler than the general proof. The ideas are very much the same. Your $\leq |f|_\infty$ part is simpler because you have a finite total measure. You are using the supremum norm instead of the essential supremum norm (see https://en.wikipedia.org/wiki/Essential_supremum_and_essential_infimum). They are identical for continuous functions on $\mathbb{R}$. I think this is the only role that continuity plays here. – Ningxin Feb 20 '16 at 05:06
  • You're also using continuity when you select the maximising point $c$ - a general $L^p([a,b])$ function doesn't necessarily attain its (essential) supremum. –  Feb 20 '16 at 11:43

0 Answers0