2
$\begingroup$

I've recently started studying the modern theory of PDEs. I studied some basic properties of Sobolev space and then started with linear elliptic PDEs.

I consider the following problem:

For which $\lambda\in\mathbb{R}$ there is a unique weak solution in $W_0^{1,2}((0,1)^2) $ to \begin{align*} -\Delta u=\lambda u\quad &\mathrm{in}\ (0,1)^2,\\ u = 0\quad &\mathrm{on}\ \partial (0,1)^2 \end{align*} ?

The only tool I know for proving the existence and uniqueness is Lax-Milgram theorem.

My thoughts:

If $\lambda\leq 0$ then I can prove the uniqueness easily using Lax-Milgram. So I was wondering, what if $\lambda > 0$. I know Poincaré inequality and the following inequality, holding under some assumptations for every $u\in W^{1,2}(\Omega)$ and $\alpha > 0$, that there exists $C > 0$ such that: \begin{equation} C||u||_2^2\leq||\nabla u ||_2^2 + \alpha\int\limits_{\partial\Omega}{|u|^2\ \mathrm{dS}}. \end{equation} So, from this inequality it follows, using Lax-Milgram, that there is a unique weak solution for every $\lambda < C$, doesn't it?

However, the problem is that I can prove the inequality just by contradiction, so I have no clue how big the constant $C$ is.

Is it possible to find a constant $C > 0$ (as big as possible) explicitly for this particular choice of $\Omega$? I tried googling it but I didn't find anything useful (I probably just don't know what I should look for)...

Thank you for any advice!:)

$\endgroup$
  • $\begingroup$ Well, it seems that I'm looking for smallest possible constant in Poincaré inequality. Using Hölder inequality, I can show that there is a unique weak solution for all $\lambda < 8$. $\endgroup$ – user1321324 Dec 5 '15 at 12:15
2
$\begingroup$

As stated, existence is trivial: $u\equiv 0$ is a solution. When a nontrivial solution exists, it is not unique since one can multiply it by a constant. But this happens only for some specific $\lambda$: those that are in the spectrum of the Dirichlet Laplacian.

You can't infer this kind of structure from Lax-Milgram. Rather, sine Fourier series should be used: expand $u(x,y)=\sum_{m,n}\sin \pi m x\sin \pi n y$ and note that $$-\Delta u = \pi^2 \sum_{m,n} (m^2+n^2)\sin \pi m x\sin \pi n y$$ This gives eigenvalues $\lambda_{m,n} = \pi^2(m^2+n^2)$, the lowest of them being $2\pi^2\approx 19.7$. This set is the entire spectrum of Dirichlet Laplacian on this domain: for $\lambda \notin \{\pi^2(m^2+n^2) : m,n\in\mathbb{Z}\}$, the zero solution is the only one.

Key terms: Dirichlet eigenvalues/spectrum, fundamental frequency/tone.

$\endgroup$

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.