2

I ask this question two days ago, but no one answer or comment. Thus, I reedit my question and edit more about mine thought. Hope someone can help me prove that. Here the question:

Let $E$ be a measurable set in $\mathbb{R^n}$ with $|E| <\infty$. Suppose that $f > 0$ a.e. in $E$ and $f, \log f \in L^1(E)$. Prove that $$ \lim_{p\rightarrow 0^+}\left(\frac{1}{|E|}\int_E f^p\right)^{\frac{1}{p}}=\exp\left(\frac{1}{|E|}\int_E \log f\right) $$

First, I think to show that $$\frac{f^p-1}{p} \rightarrow \log f$$ Whlie I don't know what should I do next that can help to get the result can I take $\frac{1}{p}$ down by $\log$? or first step I think should be change. I think to use LDCT, but I can't find $\phi$ to control $f^p.$


Another question is let $f$ be measurable, nonnegative, and finite a.e. in a set $E$. Prove that for any nonnegative constant $c$, $$\int_E e^{cf(x)}dx=|E|+c\int_0^\infty e^{c\alpha}\omega_f(\alpha)d(\alpha)$$

I think that deduce that $e^{cf} \in L(E)$ if $|E| < \infty$ and there exist constants $C_1$ and $c_1$ such that $c_1>c$ and $\omega_f(\alpha) \leq C_1e^{-c_1\alpha}$ for all $\alpha>0$. I try to use $$\int_E \phi(f)=-\int_{-\infty}^{\infty}\phi(\alpha)d\omega(\alpha)$$ Thus, $$\int_E e^{cf}=-\int_{-\infty}^{\infty}e^{c\alpha} d\omega(\alpha)=-\int_{-\infty}^{0}e^{c\alpha} d\omega(\alpha)-\int_{0}^{\infty}e^{c\alpha} d\omega(\alpha)$$ What can I do next to get the result?

How can I prove this two questions?

Gary
  • 31,845
Apple
  • 79
  • Does this answer your question? Limit of $L^p$ norm when $p\to0$ – Mittens Nov 24 '23 at 03:55
  • The two questions are not related. I suggest you post two different questions. Your first question has been already answered before in MSE (hence the link I put in my comment). The second can be solved as an excessive on Fubini's theorem. – Mittens Nov 24 '23 at 04:35
  • Yes, it includes two different question. For second question I write a little above but stuck what to do next. I haven't learn Fubini's theorem, so I think it can do it without it.Do you have some idea about mine prove above of second question. – Apple Nov 24 '23 at 05:09

1 Answers1

1

The first question has already been answered in MSE, see here for example.


As for the second question, assuming that $f\geq0$ and $\mu(E) <\infty$, one can use Fubini's theorem to get

\begin{align} \int_Ee^{cf}d\mu&=\int_E\big(e^{cf}-1+1\big)d\mu=\mu(E)+\int_E(e^{cf}-1)\,d\mu\\ &=\mu(E)+ c \int_E\Big(\int^f_0 e^{ct}\,dt\Big)\,d\mu\\ &\stackrel{Fubini}{=}\mu(E)+ c\int^\infty_0\Big(\int_E e^{ct}\mathbb{1}_{\{f >t\}}\,d\mu\Big)\,dt\\ &=\mu(E)+ c\int^\infty_0e^{ct}\mu(f>t)\,dt \end{align}

Comment:

There are some treatments of integration theory in which the Lebegue integral of a measurable nonnegative function $f$ is defined as $$\int_E f\,d\mu:=\int^\infty_0\mu(x\in E: f(x)>t)\,dt$$ where the integral in the right-hand-side is taken in the sense of improper Riemann integral. See Lieb, E. and Loss, M., Analysis, AMS, 2001 for example. In such case, an alternative solution can be obtained along the following lines:

\begin{align} \int_E e^{cf}\,d\mu&=\int^\infty_0\mu(e^{cf}>t)\,dt=\int^\infty_0\mu(f>\frac1c\log t)\,dt\\ &\stackrel{u=\frac1c\log t}{=}c\int^\infty_{-\infty}\mu(f>u)e^{cu}\,du\\ &=c\int^0_{-\infty}\mu(f>u)e^{cu}\,du+ c\int^\infty_0\mu(f>u)e^{cu}\,du\\ &=\mu(E)c\int^0_{-\infty}e^{cu}\,du+ c\int^\infty_0\mu(f>u)e^{cu}\,du\\ &=\mu(E)+ c\int^\infty_0\mu(f>u)e^{cu}\,du \end{align}

Mittens
  • 39,145
  • From my thought above, can I get the result? Use $\int_E \phi(f)=-\int_{-\infty}^{\infty}\phi(\alpha)d\omega(\alpha)$ ,then separate in two parts. – Apple Nov 24 '23 at 05:23
  • @Apple: I just added an alternative method based on a definition of Lebesgue integral develop in Kiev and Loss for example. – Mittens Nov 24 '23 at 06:47
  • Sorry, why can the last line be deduced from the second to last line? – Apple Nov 24 '23 at 08:32
  • Because for any negative number $t$, ${x\in E: f(x)>t}=E$ on account that $f\geq0$. Thus $\mu(f>t)=\mu(E)$ on $-\infty<t<0$. – Mittens Nov 24 '23 at 14:43