2
$\begingroup$

we are given the problem $u_{tt}-c^2\Delta u=0$ with conditions $u(x,0)=u_0(x)$ and $u_t(x,0)=u_1(x)$ where $x\in\mathbb{R}^n$ and $u_0,u_1\in\mathcal{C}^1$ having compact supports. If a solution exists, is it possible to conclude that this solution has a compact support, too? Thanks.

$\endgroup$
  • $\begingroup$ The solution will not have global compact support for $t \in [0,\infty)$, but it will for every finite time because of finite wave propagation speed. Is that what you mean? $\endgroup$ – Lukas Geyer Jan 21 '15 at 15:48
  • $\begingroup$ Thanks for your answer! Basically, I want to prove uniqueness of the solution if the intial data have a compact support. Therefore, my idea was to use energy conservation which my professor proved in the case when the solution has a compact support. Unfortunately, this seems not to be true in general and I have no nice tool such as D'Alembert's formula as in the one-dimensional case available. Is there another way to prove uniqueness of the solution having compactly supported data? $\endgroup$ – JohnSmith Jan 21 '15 at 23:35
  • $\begingroup$ Hint: Consider two solutions say $u,v$ of your wave equation. The difference, $w$ will be a solution too, but with null data. Now write down the weak formulation (with null data) and use as test function $w_t$. Since you can write the term $(w_{tt},w_t)$ as $\frac{d}{2dt}(||w||^2_{L^2})$, Gronwall's inequality should provide you uniqueness. This is a very standard routine, see e.g. Evans, Partial Differential Equations for further details. $\endgroup$ – alemou Jan 22 '15 at 13:39
2
$\begingroup$

Both uniqueness and finite propagation speed follow by an energy method just like in the one-dimensional case. For the sake of simplicity the following is for $c=1$, the same argument works for arbitrary $c>0$.

Fix some point $x_0 \in \mathbb{R}^n$ and time $t_0 > 0$, and define $$ E(t) = \int_{B(x_0, t_0-t)} \left( u_t(x,t)^2 + |\nabla u (x,t)|^2 \right) \, dx $$ for $0 \le t \le t_0$. Using integration by parts (Green's identities) you should get $$ E'(t) = \int_{\partial B(x_0, t_0-t)} \left( 2 u_t \frac{\partial u}{\partial n} -u_t^2 - |\nabla u|^2\right) \, dx $$ Then, using the inequality $$ \left| 2u_t \frac{\partial u}{\partial n} \right| \le 2|u_t| \, |\nabla u| \le u_t^2 + |\nabla u|^2 $$ you get that $E'(t) \le 0$. So if you know $u(x,0)\equiv u_t(x,0)\equiv 0$ for $x \in B(x_0, t_0)$, then $E(0)=0$, and so $E(t)=0$ for $0 \le t \le t_0$, implying that $u_t(x,t) = 0$ and $\nabla u (x,t) = 0$ for $0\le t \le t_0$ and $|x-x_0| \le t_0 - t$, which implies $u(x,t) = 0$ for the same parameters. This directly implies uniqueness and propagation speed $\le 1$.

$\endgroup$

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.