2
$\begingroup$

Let $\Omega$ be a bounded open connected set in $\mathbb{R}^n$ with $C^1$ boundary and let $0<\alpha<1$. Then there exists a real number $\sigma_0>0$ and a dimensional constant $C>0$ such that $$||Du||_{L^\infty(\Omega)}\leq \sigma^\alpha [|Du|]_{\alpha,\Omega}+\frac{C}{\sigma}||u||_{L^\infty(\Omega)}$$ and $$[u]_{\alpha,\Omega}\leq \sigma[|Du|]_{\alpha,\Omega}+\frac{C}{\sigma^\alpha}||u||_{L^\infty(\Omega)}$$ hold for all $0<\sigma<\sigma_0$ and for all $u\in C^{1,\alpha}(\bar\Omega)$. Here $||u||_{C^{1,\alpha}}=||u||_{L^\infty(\Omega)}+||Du||_{L^\infty(\Omega)}+[|Du|]_{\alpha}$ and $[u]_\alpha=\sup_{x\neq y}\frac{|u(x)-u(y)|}{|x-y|^\alpha}$.

N.B. I have proved the above results for balls and then for domain with $C^2$ boundary. I cant proceed for $C^1$ boundary domain. Any help will be great.

$\endgroup$
  • $\begingroup$ It might help if you showed how you dealt with $C^2$ boundaries. $\endgroup$ – robjohn Mar 28 at 15:20
  • $\begingroup$ $C^2$ boundary have the interior ball property and i have the results for balls. $\endgroup$ – mudok Mar 28 at 18:11
1
+100
$\begingroup$

Since $\Omega$ is bounded, its closure its compact. Also since $u\in C^{1,\alpha}(\overline{\Omega})$, you have that $Du$ is continuous. Hence there exists $x_{0}\in\overline{\Omega}$ such that $$ |Du(x_{0})|=\max_{x\in\overline{\Omega}}|Du(x)|. $$ In particular, $u$ is differentiable $x_{0}$. Thus, $$ \frac{\partial u}{\partial\nu}(x_{0})=Du(x_{0})\cdot\nu, $$ where $\frac{\partial u}{\partial\nu}$ is a directional derivative in an admissible direction $\nu\in\mathbb{R}^{n}$, with $|\nu|=1$. Now since $\partial\Omega$ is of class $C^{1}$, there is a cone $C$ such that every point $x\in\overline{\Omega}$ is the vertex of a cone $C_{x}$ congruent to $C$ and contained in $\Omega$. Hence, if you consider the cone $C_{x_{0}}$, you can find $n$ linearly independent directions $\nu_{1},\ldots,\nu_{n}$ such that the segments $x_{0}+t\nu_{i}$, $t\in\lbrack0,h]$ are contained in the cone $C_{x_{0}}$, where $h>0$. If you consider the system of $n$ equations $$ \frac{\partial u}{\partial\nu_{i}}(x_{0})=Du(x_{0})\cdot\nu_{i}, $$ in the $n$ unknowns $\frac{\partial u}{\partial x_{i}}(x_{0})$, you have that the determinant is different from zero since the vectors are linearly independent. Hence, you can write $$ \frac{\partial u}{\partial x_{i}}(x_{0})=\sum_{j=1}^{n}c_{i,j}\frac{\partial u}{\partial\nu_{j}}(x_{0}), $$ where the numbers $c_{i,j}$ depent only on the directions $\nu_{1},\ldots ,\nu_{n}$.

Along each segment $S_{i}=\{x_{0}+t\nu_{i}$, $t\in\lbrack0,h]\}$ you can apply your inequality for $n=1$ to the function $g_{i}(t):=u(x_{0}+t\nu_{i})$, $t\in\lbrack0,h]$.

Now you have to prove that the coefficients $c_{i,j}$ depend only on $\Omega$. I have to think about this, but if you rotate the cone, your new directions are $R\nu_{1},\ldots,R\nu_{n}$, where $R$ is your rotation, so the determinant should not change.

$\endgroup$
  • $\begingroup$ Very nice argument. What you essentially use is the interior cone property for $C^1$ domain. Which escaped my mind. I also think your $c_{i,j}$ dependence only on domain since it's come from the linear independent direction vectors depending on domain. Am I correct? $\endgroup$ – mudok Mar 30 at 15:56
  • $\begingroup$ yes, you are correct $\endgroup$ – Gio67 Mar 30 at 16:08
  • $\begingroup$ one question : you say " a cone Cx congruent to C and contained in Ω" why congruent? $\endgroup$ – mudok Mar 30 at 17:11
  • $\begingroup$ up to a translation and a rotation. And I should have said contained in the closure of $\Omega$. You will need to rotate the cone. $\endgroup$ – Gio67 Mar 31 at 12:40
  • $\begingroup$ $C^1$ boundary imply uniform cone property or only cone property? $\endgroup$ – mudok Mar 31 at 13:04

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.