Definition of measurable set: A set $E$ is measurable if $$m^*(T) = m^*(T \cap E) + m^*(T \cap E^c)$$ for every subset of $T$ of $\mathbb R$.

Definition of Lebesgue measurable function: Given a function $f: D \to \mathbb R ∪ \{+\infty, -\infty\}$, defined on some domain $D \subset \mathbb{R}^n$, we say that $f$ is Lebesgue measurable if $D$ is measurable and if, for each $a\in[-\infty, +\infty]$, the set $\{x\in D \mid f(x) > a\}$ is measurable.

Definition of Lebesgue integral of simple function: We say that a simple function $\psi$ is Lebesgue integrable if the set $\{\psi \ne 0\}$ has finite measure. In this case, we may write the standard representation for $\psi$ as $\psi = \sum_{i=0}^n a_i \chi_{A_i}$, where $a_0 = 0, a_1, \ldots , a_n$ are distinct real numbers, where $A_0 = \{\psi = 0\}, A_1, \ldots , A_n$ are pairwise disjoint and measurable, and where only $A_0$ has infinite measure, Once $\psi$ is so written, there is an obvious definition for $\int \psi$, namely, $$\int \psi = \int_{\mathbb R} \psi = \int_{-\infty}^{+\infty} \psi(x) \ \mathsf dx = \sum_{i=1}^n a_i m(A_i).$$ In other words, by adopting the convention that $0 \cdot \infty = 0$, we define the Lebesgue integral of $\psi$ by $$\int \sum_{i=0}^n a_i \chi_{A_i} = \sum_{i=0}^n a_i m(A_i).$$ Please note that $a_im(A_i)$ is a product of real numbers for $i \ne 0$, and it is $0 \cdot \infty = 0$ for $i = 0$; that is, $\int \psi$ is a finite real number.

Definition of Lebesgue integrable of non-negative function: If $f: \mathbb R \to [0, +\infty]$ is measurable, we define the Lebesgue integral of $f$ over $\mathbb R$ by $$\int f = \sup \left\{\int \psi: 0 \le \psi \le f, \psi\text{ simple function and integrable }\right\}.$$

Suppose $f(x)$ and $g(x)$ are non-negative Lebesgue measurable functions defined on $E \subset \mathbb R$. How to prove $$\left(\int_{E} f(x)\ \mathsf dx\right)^{\frac{1}{2}} \left(\int_{E} g(x)\ \mathsf dx\right)^{\frac{1}{2}} \ge \int_{E} {\sqrt{f(x)g(x)}}\ \mathsf dx$$?

Besides, can anyone introduce some mathematical books about inequalities to me such as basic inequalities, integral inequalities, norm inequalities, convex inequalities and so on? I'm not sure whether there is a book covering such a board area of mathematical inequalities. So it will be pretty helpful even if your introduced book merely covering one specific area of them. Thanks in advance.


I notice that if I treat $a = \sqrt {f(x)}$ and $b = \sqrt {g(x)}$ then it will be much looked like a Holder inequality in Riemann integral.

  • $\begingroup$ Look carefully - there's an extra $1/2$ in your formula. As long as we're talking about non-math, note what happens if you say \left(\int_E\right) instead of (\int_E). Anyway, this is exactly Holder's inequality. You've evidently seen a proof for the Riemann integral - the same proof works here. (At least I can't imagine a proof for the Riemann integral that doesn't work for the Lebesgue integral; check back if you find a proof for the Riemann integral but have problems adapting it...) $\endgroup$ – David C. Ullrich Aug 5 '15 at 22:36
  • $\begingroup$ Note every \left( has to have a matching \right) or it won't work... $\endgroup$ – David C. Ullrich Aug 5 '15 at 22:36
  • $\begingroup$ @DavidC.Ullrich: Sorry, my typo. $\endgroup$ – Bear and bunny Aug 5 '15 at 22:40
  • $\begingroup$ @DavidC.Ullrich: Sorry, I can't figure out what's the difference between \left(\int_E\right) and (\int_E). I typed in and latex returned correspondingly two almost same symbols(actually, exactly same in my eyes). Can you explain more? $\endgroup$ – Bear and bunny Aug 5 '15 at 22:50
  • $\begingroup$ @Bearandbunny I edited it in. \left and \right make parentheses/brackets/braces/etc. resize to match the height of their contents. $\endgroup$ – Math1000 Aug 5 '15 at 23:30

When $||f||_p > 0, ||g||_{p'}>0$ and $p, p' < +\infty$, then$$a^{\frac{1}{p}}b^{\frac{1}{p'}} \le \frac{a}{p} + \frac{b}{p'}, a>0, b>0$$ and $a = \frac{|f(x)|^p}{||f||_{p}^{p}}, b = \frac{|g(x)|^{p'}}{||g||_{p'}^{p'}}$, then get $$\frac{|f(x)g(x)|}{||f||_p ||g||_{p'}} \le \frac{1}{p}\frac{|f(x)|^p}{||f||_p^p} + \frac{1}{p'} \frac{|g(x)|^{p'}}{||g||_{p'}^{p'}}$$. Integrate both side, then $$\int_E \frac{|f(x)g(x)|}{||f||_p ||g||_{p'}}dx \le \frac{1}{p} \int_E \frac{|f(x)|^p}{||f||_p^p}dx + \frac{1}{p'} \int_E \frac{|g(x)|^{p'}}{||g||_{p'}^{p'}}dx = \frac{1}{p} + \frac{1}{p'} = 1$$ where $$\int_E {|f(x)|^p} dx =||f||_p^p, \int_E {|g(x)|^{p'}} dx =||g||_{p'}^{p'}$$.

| cite | improve this answer | |

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.