Exercise on Stein's/Sharkachi book chapter 6:

Let $u(x, t)$ be a smooth solution of the wave equation and let $E(t)$ denote the energy of this wave $$E(t) = \int_{\mathbb R^d} \bigg|\frac{\partial u(x,t)}{\partial t}\bigg|^2+\sum_{j=1}^d \int_{\mathbb R^d}\bigg | \frac{\partial u(x,t)}{\partial x_j}\bigg |^2 dx.$$ We have seen that $E(t)$ is constant using Plancherel’s formula. One can give an alternate proof of this fact by differentiating the integral with respect to $t$ and showing that $\frac{dE}{dt} = 0.$

I remember that when we were in the 1-dimensional case, to prove a similar formula (which didn't involved modules back then), we had to multiply $u_{tt}= u_{xx}$ by $u_t$, integrate with respect to $x$ and then realize integration by parts. Noticing that $1/2(u_t^2)_t = u_{tt}u_t $ and doing similarly for $u_{xx}u_t$, we would arrive that: $\frac{d}{dt}(1/2\int u_t^2dx+1/2\int u_x^2dx ) =0 $ which implies the conservation of energy.

So the hint given for the problem in $\mathbb R^d$ is the same: use integration by parts. But I dunno exactly what is the d-dimensional analog of integration by parts. How can I solve this problem?


This is a typical energy method approach to use this to show the wave equation have but one solution. Rewrite your integral as

$$E(t) = \int_{\mathbb{R}^d} u_{t}^2 + |\nabla u|^2 dx$$


$$E'(t) = \int_{\mathbb{R}^d} 2u_{t}u_{tt} + 2(\nabla u \cdot \nabla u_t) dx$$

Here is the step where IBP is performed. Since you're curious, in higher dimension, the formula is

$$\int_U u_{x_{i}}v dx = -\int_U uv_{x_{i}}dx + \int_{\partial U}uv\nu^{i} dS$$

for $i = 1,2,...n$ and $\nu^i$ is the $ith$ component of the outward pointing unit normal vector of $U$, an open set bounded in $\mathbb{R}^n$. Applying it here we get

$$\int_{\mathbb{R}^d}\nabla u \cdot \nabla u_t \,dx = -\int_{\mathbb{R}^d}u_t \Delta u \, dx$$

since the boundary of $\Bbb{R}^n$ is empty. Hence

$$E'(t) = 2\int_{\mathbb{R}^d} u_{t}(u_{tt} - \Delta u) dx = 0$$

  • $\begingroup$ Why the last integral equals 0? $u$ being solution of the wave equation only implies that $u_{tt}= \Delta u$ $\endgroup$ – math.h Nov 20 '18 at 11:42
  • $\begingroup$ Ah, I made an error, I will edit. $\endgroup$ – DaveNine Nov 20 '18 at 12:55

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.