Say $\Omega\subseteq \mathbb{R}^n$ is a bounded open set and $0<\alpha<1$. I need some $C^{2, \alpha}(\overline\Omega)$ regularity result for elliptic equations with Neumann boundary conditions but that is what I find in Jurgen Jost' book:

If $u$ is a weak solution of $\Delta u=f$ in $\Omega$ and $f\in C^\alpha(\Omega)$, then $u\in C^{2,\alpha}(\Omega)$..........(1)

But it says that we can identify $C^{2, \alpha}(\Omega)$ with $C^{2,\alpha}(\overline\Omega)$ here:

However, Jost used both $C^{2, \alpha}(\Omega)$ and $C^{2,\alpha}(\overline\Omega)$ (when he is doing regularity for Dirichlet boundary) in his book, which indicated these two spaces are not the same?

Or do we have $u\in C^{2,\alpha}(\overline\Omega)$ in (1)? Otherwise where can I find some $C^{2,\alpha}(\overline\Omega)$ regularity result with Neumann boundary conditon?


Let $\Omega\subset\mathbb{R}^N$ be a open bounded domain and assume that $f:\Omega\to\mathbb{R}^N$ satisfies $$\frac{|f(x)-f(y)|}{|x-y|^\alpha}\leq M,\ \forall\ x,y\in\Omega\tag{1}$$

for some fixed $\alpha\in (0,1)$. We can conclude from $(1)$ that $f$ is uniformly continuous in $\Omega$.

Because $f$ is uniformly continuous in $\Omega$, we can extend it in a unique way to the boundary of $\Omega$ and we get a function (still denoted by $f$) $f:\overline{\Omega}\to\mathbb{R}$ which is continuus. The question is: Is this function Holder continuous with exponente $\alpha$?

The answer is yes: take $x,y\in\partial\Omega$ with $x\neq y$ and choose $x_n,y_n\in\Omega$ such that $f(x_n)\to f(x)$ and $f(y_n)\to f(y)$. Note that $$\frac{|f(x)-f(y)|}{|x-y|^\alpha}=\lim_{x_n,y_n\to x,y}\frac{|f(x_n)-f(y_n)|}{|x_n-y_n|^\alpha}\leq M$$

The last equality is true because of the continuity of the function $g(x,y)=\frac{|f(x)-f(y)|}{|x-y|^\alpha}$.

We conclude that (in terms of identification) $C^\alpha(\Omega)=C^\alpha(\overline{\Omega})$.

Now here, come the crucial observation. If $\Omega$ does not have a nice boundary, then it's nonsense to talk about $C^1(\overline{\Omega})$ or $C^2(\overline{\Omega})$, therefore, if we ask some regularity on the boundary, to wit, $\partial\Omega \in C^{2,\alpha}$, hence, the same argument given above we have that $C^{2,\alpha}(\Omega)=C^{2,\alpha}(\overline{\Omega})$

Take a look in Theorem 11.2.2. of the book you mentioned from Jost. Note that the boundary regularity is required.

  • $\begingroup$ Thank you very much for your reply. It helps me a lot. $\endgroup$ – user100356 Oct 12 '13 at 19:00
  • $\begingroup$ You are welcome @user100356 $\endgroup$ – Tomás Oct 12 '13 at 19:03

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.