So I am trying to solve the Klein-Gordon equation using a Fourier transform of the spatial components only. The Klein-Gordon equation reads:

$$ (\partial ^2 + m^2)\phi(x) = 0. $$

If I let

$$ \phi(x) = \phi(t, \mathbf{x}) = \frac{1}{(2\pi)^{3/2}}\int \widetilde{\phi}(t,\mathbf{k})e^{-i \mathbf{k} \cdot \mathbf{x} } \mathrm{d}^3k,$$

I substitute this into the K-G equation to get

$$\frac{\partial^2 \widetilde{\phi}}{\partial t^2} +(\mathbf{k}^2+m^2)\widetilde{\phi}=0. $$

This is the differential equation for a simple harmonic oscillator so I can immediately write

$$ \widetilde{\phi}(t,\mathbf{k})=A(\mathbf{k})e^{i \omega_{\mathbf{k}}t}+B(\mathbf{k})e^{-i \omega_{\mathbf{k}} t}, $$

and therefore I find

$$ \phi(x)= \frac{1}{(2\pi)^{3/2}}\int \bigg(A(\mathbf{k})e^{i \omega_{\mathbf{k}}t-\mathbf{k} \cdot \mathbf{x}}+B(\mathbf{k})e^{-i \omega_{\mathbf{k}} t -\mathbf{k} \cdot \mathbf{x}}\bigg) \mathrm{d}^3k. $$

Now at this stage I get slightly confused. I have seen the general solution to the Klein-Gordon equation and it has a factor of $1/2 \omega_{\mathbf{k}}$ inside the integrand, any hints as to how I can proceed to this?

I have also read that in order to tidy notation up I would set $\mathbf{k} \rightarrow -\mathbf{k} $ in the second term of the integrand, but wouldn't this change $ \mathrm{d}^3k \rightarrow -\mathrm{d}^3k$ as the Jacobian determinant is $-1$, so surely it should be a difference of two terms?

The solution I am trying to get to:

$$ \phi(x)= \frac{1}{(2\pi)^{3/2}}\int \frac{1}{2\omega_{\mathbf{k}}} \bigg(A(\mathbf{k})e^{i \omega_{\mathbf{k}}t-\mathbf{k} \cdot \mathbf{x}}+B(\mathbf{k})e^{-i \omega_{\mathbf{k}} t +\mathbf{k} \cdot \mathbf{x}}\bigg) \mathrm{d}^3k. $$

  • 3
    $\begingroup$ Redefine your $A({\bf k})$ and $B({\bf k})$ to $\frac{1}{2\omega_{\bf k}} A({\bf k})$ and $\frac{1}{2\omega_{\bf k}} B({\bf k})$ respectively. $\endgroup$
    – Prahar
    Dec 19, 2017 at 6:30

1 Answer 1


The general linear combination of solutions is $$ \phi(x)= \int\!\frac{\mathrm{d}^4k}{(2\pi)^4}\,\delta(k^2-m^2)\,\Theta(k^0)\,\left[A(\vec k)\,\mathrm{e}^{-\mathrm{i}\,k\cdot x}+B(\vec k)\,\mathrm{e}^{\mathrm{i}\,k\cdot x}\right]\;. $$ In contrast to what you wrote, this is (manifestly) Lorentz invariant. After all you want to describe a scalar field. The next observation is that $$ \frac{\mathrm{d}^4k}{(2\pi)^4}\,\delta(k^2-m^2)\,\Theta(k^0)= \frac{\mathrm{d}^3k}{(2\pi)^3}\,\frac{1}{2\omega_{\vec k}} \quad\text{with}~\omega_{\vec k}=\sqrt{\vec k^2+m^2}\;.$$ This explains the factor $1/\omega_{\vec k}$. In this argument there is also no need to play with minus signs.

  • $\begingroup$ This is arguably the most straightforward way to get a manifestly Lorentz invariant expression. $\endgroup$
    – user178876
    Dec 19, 2017 at 14:00
  • $\begingroup$ Why do you have the $\Theta(k^0)$? I mean, why do you restrict the $k^0$ to positive values? $\endgroup$
    – user171780
    May 27, 2018 at 3:34
  • 1
    $\begingroup$ @user171780 Well, our conventions are such that energies of physical objects, i.e. particles and antiparticles, are positive. $\endgroup$
    – user178876
    May 27, 2018 at 14:58
  • $\begingroup$ Ok, thanks. And what about the extra $\pi$ that disappears in the denominator? I am doing the math and cannot see when it cancels out. $\endgroup$
    – user171780
    May 27, 2018 at 15:21
  • $\begingroup$ An expression written in a particular coordinate system (in which e.g. the component $k^0$ is evaluated) is not manifestly Lorentz invariant, although it is indeed implicitly Lorentz invariant $\endgroup$ Nov 22, 2019 at 18:51

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.