0
$\begingroup$

Here is Liouville's Theorem

enter image description here

Suppose that $u \colon \mathbb{R}^n \to \mathbb{R}$ is harmonic and $u \geq 0$. Prove that $u$ is constant. (In this problem , instead of $u$ is bounded now $u \geq 0$ .)

Harnack 's inequality proof in Evans :

enter image description here

Can I use Harnack 's inequality and show that $u(x) < 2^n u(0)$ ? How can I create an upper bound like the above proof that can converge to 0 as r goes to infinity ?

$\endgroup$
2
$\begingroup$

If you have Harnack's inequality then you are essentially done already. Here's a way to continue:

Let $x$ be arbitrary, $\| x \| = r$, and $R>r$. Then

$$\frac{1-r/R}{1+(r/R)^{n-1}} f(0) \leq f(x) \leq \frac{1+r/R}{1-(r/R)^{n-1}} f(0).$$

Now send $R \to \infty$.

$\endgroup$
  • $\begingroup$ Unless I'm mistaken, I don't think that inequality is done explicitly in Evans' book; for the record, it can be proved using Poisson's formula and the mean value theorem (there's a proof on Wikipedia at en.wikipedia.org/wiki/Harnack's_inequality). $\endgroup$ – Matt Rigby Sep 14 '14 at 16:09
  • $\begingroup$ The hint I was given : Look at the proof of Harnack’s inequality. Show that for any radius r, if x ∈ B(0,r), then u(x) ≤ $2^n$ u(0), and note that this bound doesn’t actually depend on r. $\endgroup$ – Peter Sep 14 '14 at 16:17
  • 1
    $\begingroup$ Following that hint, you get that $u$ is bounded, and then you can apply the unmodified Liouville theorem. $\endgroup$ – Ian Sep 14 '14 at 16:35
  • $\begingroup$ I got stuck at the part for "any radius r" , since that Harnack 's proof above in my post, r is 1/4 distance(V,dV). $\endgroup$ – Peter Sep 14 '14 at 16:53
  • $\begingroup$ That's only to make sure things are defined where they need to be; $r$ can be anything you like when it's defined on all of $\mathbb{R}^n$. $\endgroup$ – Matt Rigby Sep 14 '14 at 17:11
1
$\begingroup$

I would say that Evan's proof is not sharp. Actually, if $u \geq 0$ is harmonic, by the divergence theorem you can write $$ \frac{\partial u}{\partial x_i}(x_0)=\frac{n}{\omega_n R^n} \int_{\partial B_R(x_0)} u(y) \, d\Sigma. $$ Hence $$ \left| \frac{\partial u}{\partial x_i}(x_0) \right| \leq \frac{n}{R} u(x_0), $$ and you let $R \to +\infty$. This "proof" comes from the book Elliptic differential equations by Lin and Lin, AMS.

$\endgroup$

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.