If you take the homogenous wave equation:

$$-\Delta_x u(x,t) + \frac{1}{c^2} \frac{\partial^2 u}{\partial^2 t} (x,t) \ = \ 0 \ \ \mathrm{in} \ \Omega \times (0, \infty),$$

with some proper initial- and boundary conditions and make the ansatz:

$$u(x,t):= e^{-i \omega t} v(x),$$

i.e. we seek a time-harmonic wave to the angular frequency $\omega \geq 0 $.
Inserting this ansatz into the equation, we get the eigenvalue problem:

$$-\Delta v(x) \ = \ \underbrace{\frac{1}{c^2} \omega^2}_{:= \ \lambda} v(x) \ \ \mathrm{in} \ \Omega.$$

Now, if i solve this eigenvalue problem and got a eigenvalue $\lambda$, i can retrieve the angular frequency by $\omega = \sqrt{\lambda \ c^2}$.
Here, i'm trying to model the 2D case of acoustic waves ($\Omega \subset \mathbb{R}^2$), but the 3D-case could be also interesting. Doing some research, i found out (for example here), that $c$ is in my case the speed of sound, i.e. for example $c$ = 343 m/s in air at 20°C. I assume, since $c$ is dealing with meters, my domain $\Omega$ has also the dimension m$^2$ (or m$^3$ in 3D-case).
But, by the upper formula, $\omega = \sqrt{\lambda \ c^2}$, the angular frequency would have dimension "m/s" (meters per second) - that doesn't sound right, shouldn't this be of dimension s$^{-1}$?

I would be glad if anyone, who's more experienced in physics, could explain this. Thanks in advance!


What I am missing in your question, is the dimension of $\lambda$. You can see from your second formula, that the dimension of $\lambda$ is $\text{m}^{-2}$ (because of the Laplacian).

Further, to add a bit more information,

Normally we would write your Eigenvalue problem as

$$\Delta v = - k^2 v.$$

We use $k^2$ here mostly because it will make the answer look nicer (no square roots). We then call $k$ the wavenumber

  • $\begingroup$ Thanks! I always thought of eigenvalues as dimensionless numbers. But what exactly do you mean with "because of the laplacian"? I know that $u(x,t)$ is the pressure and should be of dimension $\frac{kg}{m s^2}$, but how exactly do you know what dimension this $v(x)$ has? I mean, the time dimension (1/s) could be in that $e^{-i \omega t}$-part after doing that ansatz, but what about the remaining part? $\endgroup$ – Rfield Aug 24 '14 at 9:31
  • $\begingroup$ @Rfield The dimension of $v$ does not matter, because it is on both sides. $\Delta=d^2/dx^2$ and $dx$ has dimension of meter, hence $[\lambda]=\text{m}^{-2}$. $\endgroup$ – Bernhard Aug 24 '14 at 10:00
  • $\begingroup$ @Rfield Let's just look at your definition $\lambda:=\omega^2/c^2$. It is easily seen from it that dimension of $\lambda$ is $\mathrm{m}^{-2}$. $\endgroup$ – firtree Aug 24 '14 at 11:05

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.