Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I have trouble proving the following problem (Evans PDE textbook 5.10. #15). Could anyone kindly help me solving the problem? I know that I should somehow use Poincare inequality but I still cannot solve it.

Fix $\alpha>0$ and let $U=B^0(0,1)\subset \mathbb{R}^n$. Show there exists a constant $C$ depending only on $n$ and $\alpha$ such that $$ \int_U u^2 dx \le C\int_U|Du|^2 dx, $$ provided that $u\in W^{1,1}(U)$ satisfies $|\{x\in U\ |\ u(x)=0\}|>\alpha$.

As Henry pointed out, this was asked before but no answer was given. I hope someone will kindly give an answer this time. Thank you very much.

share|improve this question
1  
The same question was posted here: math.stackexchange.com/questions/235296/… –  Henry T. Horton Nov 29 '12 at 23:55
    
I should have mentioned the post but I am aware of it. It's closed without answer. Thank you for pointing out that. –  Zheng Nov 30 '12 at 1:10

4 Answers 4

Setting: $A\equiv \{x\in U: u(x)=0\}$. By using the Poincaré-Witinger inequality, we have

\begin{eqnarray*} C(n)\times \|\nabla u\| _{L^2(U)} &\ge& \|u - (u)_U\| _{L^2(U)} \\ &\ge & \|u\| _{L^2(U)} - \|(u)_U\| _{L^2(U)} \\ &=& \|u\| _{L^2(U)} - \|(u)_U\| _{L^2(A^c)} \\ &=& \|u\| _{L^2(U)} - \left|(u)_U\right| \times \left|A^c \right|^{1/2} \\ &\ge& \|u\| _{L^2(U)} - {1\over |U|} \times\left[ \int_{U} |u| dx\right] \times |A^c|^{1/2} \\ &=& \|u\| _{L^2(U)} - {1\over |U|}\times \left[ \int_{A^c} |u| dx\right]\times |A^c|^{1/2} \end{eqnarray*} By Holder inequality, this last factor is equal to the largest \begin{eqnarray*} &\ge& \|u\| _{L^2(U)} - {1\over |U|}\times \left|A^c\right|^{1/2} \times \|u\|_{L^2(U)} \times|A^c|^{1/2} \\ &=& \|u\| _{L^2(U)} - {1\over |U|}\times \left|A^c\right| \times \|u\|_{L^2(U)} \\ &\stackrel{\boxed{-|A^c|\ge \alpha- |U|}}{\ge}& \|u\| _{L^2(U)} + {1\over |U|}\times (\alpha - |U|)\times \|u\|_{L^2(U)} \\ &=& \left[1+ {1\over |U|} \ \left(\alpha - |U|\right)\right]\times \|u\|_{L^2(U)} \end{eqnarray*} Since $\alpha>0 \Rightarrow |U|>|U|-\alpha \Rightarrow 1>{|U|-\alpha \over |U|} \Rightarrow 1+ {\alpha -|U| \over |U|}>0.$

Denoting the constant $C_1(\alpha) \equiv 1+ {\alpha -|U| \over |U|} >0$, inequality obtained above we obtain that $$ \|u\|_{L^2(U)} \le {C(n)\over C_1(\alpha)} \times \|\nabla u\|_{L^2(U)}, $$ concluding the proof.$_\blacksquare$

I hope I have helped.

A hug.

share|improve this answer
    
$|u| _{L^2(U)} - |(u)_U| _{L^2(U)} = |u| _{L^2(U)} - |(u)_U| _{L^2(A^c)}$. Why this is equal? I don't see how this equality is true? –  Sherry Feb 22 at 7:10
    
@Sherry I answered this same question here. –  Yes Feb 22 at 7:22
1  
@FamousBlueRaincoat My teacher gives me hint like this, but I don't understand how the hint is related to the problem: –  Sherry Feb 22 at 23:03
1  
@FamousBlueRaincoat The hint is suppose $w$ is a smooth function on the ball $B$ and $S$ is a set of positive measure. For $x\in B$ and $y\in S$, $|w(x)-w(y)|\le C\int_0^1 |Dw(tx+(1-t)y)|dt$. and so $\int\int|w(x)-w(y)|dydx\le C\int_0^1 \int \int |Dw(tx+(1-t)y)|dydxdt$. Then split the integration in the $t$ variable into two parts $0 \le t \le \frac{1}{2}$ and $\frac{1}{2} \le t \le 1$ and estimate. In general, use approximation of $u$ by smooth functions. Could you please help me with some further explanation about how the hint is related to the problem? Thanks! –  Sherry Feb 22 at 23:08

Read the proof of Poincare inequality in Evans' book carefully. At some point one conclude that $u-\tilde{u}$ is constant. That is where you will need your $\alpha$ in your problem. Hope this will help you solve the problem.

share|improve this answer

a hint: write the poincaré inequality and study the integral : $$ \frac{\int_{U} |u(x)| dx}{|U|} = \frac{\int_{U - \{ x \in U ; u(x) = 0\}} |u(x)| dx}{|U|} $$

Use Holder in the last integral .

share|improve this answer

Ok, this question can be proved by using general version of Poincare inequality.

This is Theorem 12.23 in Leoni's book. Let me copy it here:

Let $1\leq p<\infty$ and let $\Omega\subset\mathbb R^N$ be a connected extension domain for $W^{1,p}(\Omega)$ with finite measure. Let $E\subset \Omega$ be a lebesgue measureable set with positive measure. Then there exists a constant $C=C(p,\Omega,E)>0$ such that for all $u\in W^{1,p}(\Omega)$, $$ \int_\Omega|u(x)-u_E|^pdx\leq C\int_\Omega |\nabla u(x)|^pdx,\tag 1 $$ where $$ u_E:=\frac{1}{|E|}\int_E u(x)dx $$

What you need to do is set $E:=\{x\in\Omega,\,u(x)=0\}$ and notice that $|E|>0$. Then $u_E=0$ and we done by $(1)$.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.