1
$\begingroup$

I am trying to prove the following:

Suppose $u \in C^2(\Omega)$. For some $x \in \Omega$ we have that

\begin{align} \Delta u(x) = \lim_{r \to 0} \frac{2n}{r^2} \left[ \frac{1}{w_n} \int_{\partial B_{1}} u(x + ry)\mathrm{d}S_y - u(x) \right]. \end{align}

If $u$ is harmonic, then $u$ has the Mean Value Property.


I believe that we basically need to see this limit in some smart way, but I still couldnt do that. Can someone give me a hint on this? But it must use the limit above.

Thanks in advance.

$\endgroup$
0
$\begingroup$

Version 1 Maybe stepping back one step actually make things clearer. You want to proof that if $\Delta u(x) \equiv 0$, then for any $x \in \Omega$ and $r > 0$ we have $$ u(x) = \frac{1}{|\partial B_r(x)|} \int_{\partial B_r(x)} u(y) dS(y). $$

Now, the right hand side turns with transformation of variables into $$ \frac{1}{\omega_n r^{n-1}} \int_{\partial B_1} u(x+ry)dS(ry) = \frac{1}{\omega_n} \int_{\partial B_1} u(x+ry) dS(y). $$ Now lets differentiate the right hand side with respect to $r$ (where by dominated convergence we can differentiate under the integral): $$ \frac{d}{dr}\;\frac{1}{\omega_n} \int_{\partial B_1} u(x+ry) dS(y) = \frac{1}{\omega_n} \int_{\partial B_1} \frac{d}{dr}\;u(x+ry) dS(y) = \frac{1}{\omega_n} \int_{\partial B_1} (\nabla u)(x+ry)\cdot y\, dS(y) $$ Now, changing the varibales back again, we get: $$ \frac{1}{\omega_n} \int_{\partial B_1} (\nabla u)(x+ry)\cdot y\, dS(y) = \frac{1}{\omega_n r^{n-1}} \int_{\partial B_r(x)} \nabla u(x)\cdot \frac{(x-y)}{r}\, dS(y) $$ which with the divergence theorem and the fact that $u$ is harmonic turns into $$ \frac{d}{dr}\;\frac{1}{\omega_n} \int_{\partial B_1} u(x+ry) dS(y) = \frac{1}{\omega_n r^{n-1}} \int_{\partial B_r(x)} \Delta u dy = 0. $$

From this we can infer that the mean value representation actually is constant for $r$. So, it is left to show that the right hand side in the very first equation tends to $u(x)$ when radius $r$ shrinks (from above) to zero. But this is also not to difficult to see.

Version 1b

I am aware that I didn't use your cited formula and to be honest I am a bit baffled by it. But isn't it just this argument: Suppose $u$ is harmonic then $\Delta u(x) \equiv 0$. Now, this means we have $$ 0 = \lim_{r\to 0} \frac{2n}{r^2}\left(\frac{1}{\omega_n} \int_{\partial B_1} u(x+ry)dS(y) - u(x)\right). $$ But for this to be zero we need $$ \lim_{r\to 0} \frac{1}{r}\left(\frac{1}{\omega_n} \int_{\partial B_1} u(x+ry)dS(y) - u(x)\right) = 0 $$ and $$ \lim_{r\to 0} \left(\frac{1}{\omega_n} \int_{\partial B_1} u(x+ry)dS(y) - u(x)\right) = 0 $$ to be zero and now that I typed it, I think that is actually the same argument I used :). Because that is the statement.

I would love some feedback !

$\endgroup$

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.