3
$\begingroup$

I am trying to understand the derivation of the solution of the wave equation (homogeneous case) while using the Fourier Transformation:

$$\left(\frac 1 {c^2} \frac {\partial^2}{\partial t^2}- \triangle\right)\phi=0$$

The expression for $\phi(\vec r,t)$ is:

$$\phi(\vec r,t)= \iint \tilde \phi(\vec k, \omega)e^{i(\vec k \vec r - \omega t)}d^3kd\omega$$

which I assume is the reverse Fourier transformation, when one knows the $\tilde \phi(\vec k, \omega)$.

Then by substituting the 2nd expression in the first one, we get :$\omega= \pm kc$

Then for $\tilde \phi(\vec k, \omega)$ we have:

$$\tilde \phi(\vec k, \omega)=2\omega f(\vec k, \omega)\delta(\omega^2-c^2k^2)$$

Now, while I am aware that $2\omega$ is a convention, I don't know where does the above expression comes from? I recently learned about the Fourier Transformation and Series, but I didn't come across this expression. Can anyone explain this to me?

$\endgroup$

1 Answer 1

5
$\begingroup$

I'll compare this to a less rigorous way of solving the wave equation that you may be used to.

First let's start by guessing that the solution is a plane wave with $\omega, \vec k$ to be determined. $$\phi(\vec x,t)=A e^{i(\omega t-\vec k\cdot \vec x)}$$ Now plugging this in the wave equation gives $$\left(-\frac{\omega^2}{c^2}+|\vec k|^2\right)Ae^{i(\omega t-\vec k\cdot \vec x)}=0.$$ So this ansatz solves the wave equation provided that $\omega^2=c^2|\vec k|^2\implies\omega=\pm c|\vec k|$. So we are free to choose $\vec k$ and $A$ as long as we replace $\omega$ with $\omega(\vec k)=c|\vec k|$ or $\omega(\vec k)=-c|\vec k|$. We can also take a superposition of these plane waves for different values of $\vec k$: $$\phi(\vec x,t)=\int\mathrm d^3k\, A(\vec k)e^{i(\omega(\vec k) t-\vec k\cdot \vec x)}$$ Your solution is the same as this solution up to some relabelling. Your solution is given by

$$\phi(\vec x,t)=\int\mathrm d^3k\,\mathrm d\omega\, 2\omega f(\vec k,\omega)e^{i(\omega t-\vec k\cdot \vec x)}\delta(\omega^2-c^2k^2)$$

There is an identity for integrating delta functions that have functions in them: $$\int\mathrm d x\, f(x)\delta(g(x))=\sum_i\frac{f(x)}{|g'(x_i)|}$$ where $x_i$ are the solutions to $g(x)=0$. So integrating $\delta(\omega^2-c^2k^2)$ gives $$\int\mathrm d \omega\, f(\omega)\delta(\omega^2-c^2k^2)=\frac{f(ck)}{2ck}+\frac{f(-ck)}{-2ck}$$ since $\omega=\pm ck$ solves $\omega^2-c^2k^2=0$. You can check for yourself that the two solutions now coincide.

$\endgroup$
2
  • $\begingroup$ In my eq. I have it $e^{i(kr - \omega t)}$ while you have it $e^{i(\omega t - kr )}$. Is there a difference ? In order to find the solution in the time domain and position space I need to know $\phi(\vec k, \omega)$. But in your solution I couldn't understand the expression : $$\tilde \phi(\vec k, \omega)=2\omega f(\vec k, \omega)\delta(\omega^2-c^2k^2)$$. This is what I initially don't understand. $\endgroup$
    – imbAF
    Commented Dec 29, 2021 at 16:39
  • $\begingroup$ @imbAF Whether you pick $e^{i(kr-\omega t)}$ or$e^{i(\omega t-kr)}$ doesn't matter if you're consistent, sorry I was a bit sloppy here. You need to know $\tilde\phi(\vec k,\omega)$: you already know $\tilde\phi$. If you plug in any function $f(\vec k,\omega)$ you will get a solution that solves the wave equation. You can then plug it in your expression for $\phi(x,t)$ and perform the integral. If you want a specific function for $f$ you need to include boundary conditions. $\endgroup$ Commented Dec 30, 2021 at 11:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.