For a collection of point charges, the charge density is defined as

\begin{equation} \rho(\textbf{x},t) = \sum_{k=1}^{N} q_k \delta^{(3)}\left[\textbf{x}-\textbf{x}_k(t)\right] \end{equation}

while the current density is defined as

\begin{equation} \textbf{J}(\textbf{x},t) = \sum_{k=1}^{N} q_k \textbf{u}_k(t) \delta^{(3)}\left[\textbf{x}-\textbf{x}_k(t)\right] \end{equation}

To show that the expressions above are covariant, I assumed that I need to transform the above expressions under Lorentz boost and show that the continuity equation holds.

The transformation is as such

\begin{equation} \rho'(\textbf{x}',t') = \sum_{k=1}^{N} q_k \delta^{(3)}\left[\textbf{x}'-\textbf{x}'_k(t')\right]-\frac{\gamma}{c^2}\textbf{v}\cdot\sum_{k=1}^{N} q_k \textbf{u}_k(t') \delta^{(3)}\left[\textbf{x}'-\textbf{x}'_k(t')\right] \end{equation}

\begin{equation} \textbf{J}'_{\parallel}(\textbf{x}',t') = \gamma \sum_{k=1}^{N} q_k \textbf{u}_{k,\,\parallel}(t')\delta^{(3)}\left[\textbf{x}'-\textbf{x}'_k(t')\right]+\gamma \textbf{v} \sum_{k=1}^{N} q_k \delta^{(3)}\left[\textbf{x}'-\textbf{x}'_k(t')\right] \end{equation}

\begin{equation} \textbf{J}'_{\perp}(\textbf{x}',t') = \sum_{k=1}^{N} q_k \textbf{u}_{k,\,\perp}(t')\delta^{(3)}\left[\textbf{x}'-\textbf{x}'_k(t')\right] \end{equation}

The continuity equation is given as such

\begin{equation} \frac{\partial}{\partial t'}\rho'(\textbf{x}',t')=-\nabla'\cdot\textbf{J}'(\textbf{x}',t') \end{equation}

However, when I perform the explicit calculations, the steps were quite confusing and tedious. I could not show that the continuity equation holds too.

Is there a more efficient way to show the expressions for charge and current density are covariant under Lorentz boost with explicit calculations?


1 Answer 1


I think you are making your life difficult with considering collections of charges. Does one loose anything by establishing relations for a single point-charge? I don't think so.

Next, you have to be very careful about changing reference frames with 3-delta functions. Delta-functions are 'per-volume', so when you start messing about with reference frames one has to compensate the length contraction effects. The transformation rule would be $\delta^{(3)}\to \gamma\delta^{(3)}$, where $\gamma$ is the Lorentz factor, but I don't really like this approach since it misses the connection between space and time IMHO.

I think it is easier to first prove that four-current $J^\mu=\left(c\rho,\,\mathbf{J}\right)^\mu$ is a four-vector and then charge conservation in any reference frame becomes zero-divergence statement (follows from the anti-symmetry of the electromagnetic tensor and Covariant form of Ampere's law):

$\partial_\mu J^\mu=\partial_{ct}\left(c\rho\right)+\boldsymbol{\nabla}.\mathbf{J}=0$

Where $c$ is the speed of light

Showing that four-current of a single charge is a four-vector

As you said, the charge density of a single charge is:

$$ \rho\left(t, \mathbf{r}\right)=Q\delta^{(3)}\left(\mathbf{r}-\mathbf{r}_Q\left(t\right)\right) $$

Where $Q$ is the charge, and $\mathbf{r}_Q$ is the position of the charge. What I can do next is convert the 3d delta to 4d delta ($c$ is the speed of light)

$$ \rho\left(t, \mathbf{r}\right)=Q\int cdt'\,\delta\left(ct'-t\right)\delta^{(3)}\left(\mathbf{r}-\mathbf{r}_Q\left(t'\right)\right)=\frac{Q}{c}\int cdt'\,\,c\,\delta^{(4)}\left(x-x_Q\right) $$

Where $x^\mu=\left(ct,\,\mathbf{r}\right)^\mu$. The nice thing about this expression is that it now is based on 4-delta function therefore when you change reference frames all you have to multiply the $\delta^{(4)}$ by, is magnitude of the Jacobian of the coordinate transformation. If you stick to reflection, rotations and boosts that Jacobian magnitude will be 1, so you can treat $\delta^{(4)}$ as a 'normal function' (only works in Cartesian coordinates!).

Next, lets deal with integration variable $t'$, which is the temporal coorindate in your lab frame. This is inconvenient. It is easy to show that $dt=\gamma d\tau$, where $\tau$ is the proper time and $\gamma$ is the Lorentz factor. Therefore, swapping the integration variable we find:

$$ c\rho\left(t, \mathbf{r}\right)=Q\int cd\tau\,\,\gamma c\,\delta^{(4)}\left(x-x_Q\left(\tau\right)\right) $$

You can repeat similar machinations with your current density to arrive at:

$$ \mathbf{J}\left(t, \mathbf{r}\right)=Q\int cd\tau\,\,\gamma \mathbf{v}\left(\tau\right)\,\delta^{(4)}\left(x-x_Q\left(\tau\right)\right) $$

Finally note that in the lab-frame the four-velocity of the charge is $u^\mu=\left(\gamma c,\,\gamma\mathbf{v}\right)^\mu$, which is a four-vector. Therefore:

$$ J^\mu = Q\int c d\tau\,\,u^\mu\left(\tau\right)\,\delta^{(4)}\left(x-x_Q\left(\tau\right)\right)=\left(c\rho,\,\mathbf{J}\right)^\mu $$

is a four-vector (since integration with respect to proper time is frame-independent, so is definition of four-velocity, so is 4d delta function).

Divergence of four-current

It is interesting to try to take the divergence of the single-charge four-current

$$ \begin{align} \partial_\mu J^\mu &= Q\int c d\tau\,\,u^\mu\left(\tau\right)\,\partial_\mu\delta^{(4)}\left(x-x_Q\left(\tau\right)\right)=-Q\int c d\tau\,\,u^\mu\left(\tau\right)\,\frac{\partial}{\partial x_Q^\mu}\delta^{(4)}\left(x-x_Q\left(\tau\right)\right) \\ &=-Q\int c d\tau\,\,\frac{dx_q^\mu}{d\tau}\,\frac{\partial}{\partial x_Q^\mu}\delta^{(4)}\left(x-x_Q\left(\tau\right)\right)=-cQ\int d\tau\,\,\frac{d}{d\tau}\delta^{(4)}\left(x-x_Q\left(\tau\right)\right)\\ &=0 \end{align} $$

Expression vanishes because the observer has to be essentially at the beginning or the end of time for integral not be be zero.


Your Answer

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