Liénard–Wiechert Potentials: Covariant Formulation

The Griffiths-register treatment of this subject — Liénard–Wiechert Potentials, earlier in this track — earned the potentials of a moving point charge twice over: once by apparent-volume geometry, once by delta-function calculus, both in ordinary space and time. This article, the Jackson-register twin, rebuilds the same physics once, covariantly, and the payoff is startling compression. Two potentials, a constraint equation, and a Jacobian factor collapse into a single line:

A^\mu(x) = \frac{\mu_0 qc}{4\pi}\,\frac{u^\mu}{u\cdot(x - r)}\bigg|_{\tau_r},

the four-velocity of the charge divided by one Lorentz scalar, evaluated where the observer's past light cone spears the worldline. Every piece of the 3+1 story is inside: unpack the invariant denominator in any lab frame and out fall the Lorentz factor $\gamma$ , the beaming factor $\kappa = 1 - \hat{\mathbf{n}}\cdot\boldsymbol{\beta}$ , and the retarded distance — with the $\gamma$ cancelling identically against the numerator's, a two-line proof that the famous enhancement factor is 1898 retardation geometry, not 1905 relativity. But the deeper reason to own this formulation is what comes next. Differentiating the 3+1 potentials into fields is "a labor not to be undertaken lightly"; differentiating the covariant potential requires exactly one lemma — the covariant gradient of the retarded proper time, derived here from the light-cone constraint in three lines — and that lemma contains both of the Griffiths-register retarded-time identities as its time and space components. This article builds the potential, audits its invariant anatomy against the landed 3+1 results, and banks the differentiation kit that Fields of a Relativistic Moving Charge will spend on the field tensor.

Key takeaways

The point charge's covariant current is $J^\mu(x) = qc\int d\tau\, u^\mu(\tau)\,\delta^4(x - r(\tau))$ — the worldline itself, written as a source.
Feeding it to the retarded potentials and collapsing the delta functions along the past light cone gives the covariant Liénard–Wiechert potential A^\mu(x) = \frac{\mu_0 qc}{4\pi}\,\frac{u^\mu}{u\cdot(x-r)}\bigg|_{\tau_r}, with $\tau_r$ the unique proper time where the worldline crosses the observer's past light cone.
The invariant denominator unpacks in any frame as $u\cdot(x-r) = \gamma c\,\mathscr{r}\,(1 - \hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta}) = \gamma c\,\mathscr{r}\kappa$ : one scalar = Lorentz factor × retardation factor × distance.
Extracting components cancels the $\gamma$ 's identically and recovers the Griffiths-register pair $V = q/4\pi\varepsilon_0\kappa\mathscr{r}$ , $\mathbf{A} = \mathbf{v}V/c^2$ exactly — the enhancement is pure light-cone counting, now proved covariantly.
Differentiating the light-cone constraint $(x - r(\tau_r))^2 = 0$ gives the master lemma $\partial_\mu \tau_r = (x-r)_\mu / [u\cdot(x-r)]$ , whose components are exactly the $\partial t_r/\partial t = 1/\kappa$ and $\nabla t_r = -\hat{\boldsymbol{\mathscr{r}}}/c\kappa$ of the 3+1 treatment.
The lemma bank ( $\partial_\mu\tau_r$ , $\partial_\mu R^\nu$ , $\partial_\mu(u\cdot R)$ ) is this article's deliverable: the field tensor next door is three product rules away.
Key idea: the Liénard–Wiechert potentials are one four-vector equation — four-velocity over the invariant $u\cdot(x-r)$ at the light-cone crossing — and the invariant's anatomy $\gamma c\mathscr{r}\kappa$ proves the beaming factor is retardation, not relativity.

The core insight

The potentials of an arbitrarily moving point charge compress into $A^\mu = (\mu_0 qc/4\pi)\,u^\mu/[u\cdot(x-r)]$ at the retarded proper time; unpacking the one invariant denominator as $\gamma c\mathscr{r}\kappa$ recovers the entire 3+1 structure with the $\gamma$ 's cancelling identically.

Where this lives today

The covariant potential is the working currency of every code that pushes relativistic charges. Particle-in-cell and beam-dynamics packages — the simulation backbones of wakefield accelerators, free-electron lasers, and coherent-synchrotron-radiation modeling — evaluate exactly equation (2) (or its field derivatives) along numerically integrated worldlines, because at $\gamma$ of thousands the 3+1 bookkeeping of "who saw whom when" is numerically treacherous while the invariant $u\cdot R$ is stable and frame-portable. The same formula is the classical skeleton inside QED: the photon propagator coupling to a classical current reproduces (2) as its tree-level limit, which is why strong-field QED experiments (laser–electron collisions at FACET-II and LUXE) benchmark their classical baselines with covariant Liénard–Wiechert solvers before adding quantum corrections. The formulation also travels beyond electromagnetism: linearized general relativity admits Liénard–Wiechert-style solutions for moving masses ( $u^\mu u^\nu$ in place of $u^\mu$ ), and the retarded-proper-time machinery of this article — lemma bank included — is exactly what post-Newtonian waveform models for binary inspirals evaluate, with two indices instead of one.

Origins · 1 minEmil Johann Wiechert and how this was discovered

Emil Johann Wiechert

1861–1928 · Göttingen — the geophysicist who co-discovered the potentials and weighed the electron first

Wiechert is the second name on the potentials and deserves the billing. In 1900, at Göttingen, he derived the potentials of an arbitrarily moving point charge by the delta-collapse route — independently of Liénard's 1898 geometric derivation, and in the "Elementargesetze" spirit of building all electrodynamics from point-charge worldlines, an outlook far closer to this article's covariant reading than to the 19th-century continuum picture. He came to the problem from cathode rays: in January 1897, months before J.J. Thomson's celebrated measurement, Wiechert announced at Königsberg that cathode rays were particles some 2000–4000 times lighter than the hydrogen atom — the first mass estimate of the electron, from the man history filed under seismology. For that is his other life: Wiechert founded the world's first geophysics institute (Göttingen, 1898), built the inverted-pendulum seismograph that still bears his name — one has run continuously in Göttingen for over a century — and trained the generation (Gutenberg among them) that mapped Earth's core with earthquake waves. Retarded signals from a moving source, whether an electron on a worldline or a rupture in a subduction zone: Wiechert's two careers were one problem.

History

1867

— Lorenz publishes retarded potentials: the master integrals, awaiting a worthy source

1898

— Liénard evaluates them for an arbitrary moving point charge in L'Éclairage Électrique

1900

— Wiechert, independently at Göttingen, reaches the same potentials by delta-function collapse

1905

— Einstein's electrodynamics of moving bodies supplies the frame-change the potentials had anticipated

1908

— Minkowski's spacetime formalism: the four-vector packaging this article runs on

1912

— Schott's treatise systematizes point-charge radiation theory on the LW foundation

1938

— Dirac's covariant radiation-reaction calculation runs the covariant potentials at full depth

1949

— Schwinger's classical-radiation paper makes the covariant machinery standard accelerator equipment

The physical picture

Draw the problem in spacetime rather than space. The charge's history is a worldline $r^\mu(\tau)$ , threaded by proper time and everywhere tilted less than $45°$ from vertical ( $v < c$ ). The observation event $x^\mu$ sits somewhere off the worldline, and hanging beneath it is its past light cone — the set of all events that can influence $x$ by a signal at exactly speed $c$ . Electromagnetic news propagates on the cone (sharp Huygens in 3+1 dimensions, a fact the wave-equation articles established), so the potential at $x$ is sourced by exactly the event where the worldline pierces the cone. A timelike worldline pierces a backward light cone precisely once — the covariant restatement of the collapsing-sphere uniqueness theorem the Griffiths-register twin proved — and that single piercing event defines the retarded proper time $\tau_r$ and the null separation $R^\mu \equiv x^\mu - r^\mu(\tau_r)$ , with $R\cdot R = 0$ . Everything the observer can know about the charge is data at the piercing point: where it was ( $r^\mu$ ), how fast it was moving ( $u^\mu$ ), and — for the fields, next article — how it was accelerating ( $a^\mu$ ). The covariant potential is the simplest four-vector you can build from that data: $u^\mu$ for the numerator (the potential of a charge is its motion, made field), and for the denominator the only scalar available at first order, $u\cdot R$ . What the 3+1 treatment discovered as a surprising Jacobian factor, the covariant treatment almost cannot avoid writing down: there is no other invariant for it to be. The work of this article is to earn that formula honestly from the delta-function calculus, then take the invariant denominator apart in the lab frame — where its three factors tell three different stories — and finally to differentiate the light-cone constraint into the compact lemma that powers the entire radiation chapter downstream.

Figure

The covariant geometry — the observer's past light cone spears the worldline exactly once; the null separation R lies on the cone and the four-velocity at the piercing point supplies the potential

The covariant source and the potential

First, the source. A point charge is a worldline wearing a charge, and its four-current must say exactly that. The correct expression integrates the four-velocity against a four-dimensional delta pinned to the worldline:

J^\mu(x) = qc\int_{-\infty}^{\infty} d\tau\; u^\mu(\tau)\,\delta^4\big(x - r(\tau)\big). \qquad (1)

where: $r^\mu(\tau)$ — the worldline, parameterized by proper time; $u^\mu = dr^\mu/d\tau = \gamma(c, \mathbf{v})$ — the four-velocity (normalized $u\cdot u = c^2$ in the $(+,-,-,-)$ metric this track locked in Minkowski Spacetime); the prefactor $qc$ is fixed by demanding the time component reproduce the familiar density. Check it: $\delta^4 = \delta(x^0 - r^0(\tau))\,\delta^3(\mathbf{x} - \mathbf{w})$ , and collapsing the time delta against $\tau$ costs a Jacobian $1/(dr^0/d\tau) = 1/\gamma c$ , so $J^0 = qc\cdot\gamma c\cdot(\gamma c)^{-1}\delta^3(\mathbf{x} - \mathbf{w}(t)) = c\rho$ ✓, and the spatial components give $\rho\mathbf{v}$ ✓. Equation (1) is manifestly covariant — $d\tau$ scalar, $u^\mu$ vector, $\delta^4$ scalar (its defining integral $\int d^4x\,\delta^4 = 1$ is over the invariant measure) — so the point charge enters relativity without apology. The potential it sources:

\boxed{\;A^\mu(x) = \frac{\mu_0 qc}{4\pi}\,\frac{u^\mu}{u\cdot(x - r)}\bigg|_{\tau = \tau_r} = \frac{q}{4\pi\varepsilon_0 c}\,\frac{u^\mu}{u\cdot R}\bigg|_{\tau_r}, \qquad R^\mu \equiv x^\mu - r^\mu(\tau_r),\quad R\cdot R = 0.\;} \qquad (2)

where: $\tau_r$ — the retarded proper time, the unique past-cone piercing; $u\cdot R = u_\mu R^\mu$ — the article's protagonist, a Lorentz scalar. Read the structure before the derivation earns it: the numerator says the four-potential of a point charge is parallel to its four-velocity (the deep reason the 3+1 pair obeyed $\mathbf{A} = \mathbf{v}V/c^2$ ); the denominator is the only nonzero scalar linear in the separation available at the piercing point ( $R\cdot R = 0$ is spent by the constraint, so $u\cdot R$ is what remains). In the charge's instantaneous rest frame $u^\mu = (c, \mathbf{0})$ and $u\cdot R = c\mathscr{r}_0$ , collapsing (2) to the Coulomb potential — the normalization audit. Derivation A now earns the formula; Derivation B dissects the denominator; Derivation C differentiates the constraint.

Derivation AThe proper time route to the covariant potential

The goal is equation (2) from first principles: insert the covariant point-charge current (1) into the retarded potentials and collapse the delta functions exactly. The Griffiths-register twin did this computation in 3+1 language and paid the Jacobian $\kappa = 1 - \hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta}$ at the final step; here the same collapse is organized so that the Jacobian appears as the derivative of the light-cone constraint along the worldline, $d[(x-r)^2]/d\tau = -2\,u\cdot(x-r)$ — one covariant object instead of a chain of vector identities. The route: rewrite the retarded solution with an explicit time delta, insert (1), collapse space, convert the surviving delta to the invariant $\delta((x-r)^2)$ , and pay the proper-time Jacobian once. (The sibling treatment Covariant Wave Equation and Retarded Solutions reaches the same endpoint as a two-line corollary of the invariant Green function — a third route; here the collapse is done by hand so every factor is visible.)

Step 1. Set up the retarded integral with the time delta explicit. The retarded potentials — verified by direct differentiation in Retarded Potentials, recalled here, not re-proved — read, with the retardation written as a delta,

A^\mu(x) = \frac{\mu_0}{4\pi}\int dt'\,d^3x'\;\frac{J^\mu(\mathbf{x}', t')}{|\mathbf{x} - \mathbf{x}'|}\;\delta\!\left(t' - t + \frac{|\mathbf{x} - \mathbf{x}'|}{c}\right).

Step 2. Insert the covariant current (1) and collapse the primed integrals. The $d^3x'$ integral eats $\delta^3(\mathbf{x}' - \mathbf{r}_s(\tau))$ , placing the source point on the worldline; the $dt'$ integral eats $\delta(ct' - r^0(\tau))$ at cost $1/c$ (the delta's argument is $ct'$ , not $t'$ ). Writing $t_\tau \equiv r^0(\tau)/c$ for the worldline's coordinate time and $\mathscr{r}(\tau) \equiv |\mathbf{x} - \mathbf{r}_s(\tau)|$ :

A^\mu(x) = \frac{\mu_0 q}{4\pi}\int d\tau\; u^\mu(\tau)\,\frac{1}{\mathscr{r}(\tau)}\,\delta\!\left(t_\tau - t + \frac{\mathscr{r}(\tau)}{c}\right).

Why this is already progress: the integral now runs over the worldline's own clock, and the delta's root is the retarded event. What remains is bookkeeping — but the bookkeeping is where the physics lives.

Step 3. Convert to the invariant delta. The quantity $(x - r(\tau))^2 = c^2(t - t_\tau)^2 - \mathscr{r}^2$ factors as $[c(t - t_\tau) - \mathscr{r}][c(t - t_\tau) + \mathscr{r}]$ , so on the past branch ( $t > t_\tau$ ) the composition rule for deltas gives

\delta\big((x - r(\tau))^2\big)\,\theta\big(x^0 - r^0(\tau)\big) = \frac{\delta\big(c(t - t_\tau) - \mathscr{r}\big)}{2\mathscr{r}} = \frac{1}{2c\,\mathscr{r}}\,\delta\!\left(t_\tau - t + \frac{\mathscr{r}}{c}\right),

which is exactly the combination in Step 2. Substituting,

A^\mu(x) = \frac{\mu_0 qc}{2\pi}\int d\tau\; u^\mu(\tau)\;\delta\big((x - r(\tau))^2\big)\,\theta\big(x^0 - r^0(\tau)\big).

Why this intermediate line deserves a pause: it is the entire problem stated invariantly — the whole worldline integrated against the observer's light cone, the theta function selecting the past sheet. Both factors are Lorentz scalars (null order is frame-absolute, the invariant-order theorem of Minkowski Spacetime), so the covariance of the answer is now manifest before the final collapse.

Step 4. Pay the proper-time Jacobian. The delta's argument $s(\tau) = (x - r(\tau))^2$ has derivative

\frac{ds}{d\tau} = -2\,(x - r)_\nu\,\frac{dr^\nu}{d\tau} = -2\,u\cdot(x - r),

and $u\cdot(x-r) > 0$ on the past cone (in any frame it equals $\gamma c\mathscr{r}\kappa$ , all factors positive for $v<c$ — Derivation B). The composition rule collapses the integral onto the single root $\tau_r$ :

A^\mu(x) = \frac{\mu_0 qc}{2\pi}\cdot\frac{u^\mu(\tau_r)}{2\,u\cdot(x - r(\tau_r))} = \frac{\mu_0 qc}{4\pi}\,\frac{u^\mu}{u\cdot R}\bigg|_{\tau_r}.

This is equation (2). One delta, one Jacobian, one line — the covariant edition of the twin's two 3+1 routes, which are recalled by name and not re-run: their apparent-volume geometry and their $g'(t_r)$ calculus are both inside $ds/d\tau = -2u\cdot R$ , as Derivation B's unpacking will make explicit. Verified: matches Jackson §14.1 (Eq. 14.8) and Zangwill §23.2; numerical audit for a circular worldline at $\beta = 0.9$ — retarded time solved to machine precision, $cA^0$ against the 3+1 scalar potential and $\mathbf{A}$ against $\mathbf{v}V/c^2$ — agrees to 1 part in $10^{15}$ .

The invariant denominator, dissected

Go deeper · Derivation B · 2 minOne invariant three factors — unpacking u dot R in the lab

Derivation BOne invariant three factors — unpacking u dot R in the lab

The goal is the anatomy of the denominator: evaluate $u\cdot R$ in an arbitrary lab frame, watch it factor into $\gamma \times \kappa \times c\mathscr{r}$ , and then extract the components of (2) to recover the Griffiths-register pair exactly — the $\gamma$ 's cancelling in the act. This three-step unpacking is the covariant proof of the twin's central claim: the Liénard–Wiechert enhancement is retardation geometry (the 1898 physics of Liénard), merely organized by relativity (the 1905 physics of Einstein). The strategy is bare component arithmetic — no cleverness, which is the point: the covariant formula does the thinking.

Step 1. Write the two four-vectors in lab components. At the retarded event, $u^\mu = \gamma(c, \mathbf{v})$ ; and the null separation has time component fixed by the light-cone condition $|\mathbf{x} - \mathbf{w}(t_r)| = c(t - t_r)$ :

R^\mu = \big(c(t - t_r),\; \boldsymbol{\mathscr{r}}\big) = \big(\mathscr{r},\; \boldsymbol{\mathscr{r}}\big), \qquad R\cdot R = \mathscr{r}^2 - |\boldsymbol{\mathscr{r}}|^2 = 0 \checkmark

Step 2. Contract with the metric $(+,-,-,-)$ :

u\cdot R = \gamma c\,\mathscr{r} - \gamma\,\mathbf{v}\cdot\boldsymbol{\mathscr{r}} = \gamma\big(\mathscr{r}c - \boldsymbol{\mathscr{r}}\cdot\mathbf{v}\big) = \gamma\, c\,\mathscr{r}\,\big(1 - \hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta}\big) = \gamma\,\kappa\,c\,\mathscr{r}.

Why each factor has its own passport: $c\mathscr{r}$ is Coulomb's — the static potential's denominator; $\kappa = 1 - \hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta}$ is Liénard's — the light-cone sweep Jacobian the twin earned twice, carrying no $\gamma$ and existing already in a strictly pre-relativistic reading; $\gamma$ is Einstein's — and it is about to cancel. Note the base treatment of this article stated the invariant as $(\mathscr{r}c - \boldsymbol{\mathscr{r}}\cdot\mathbf{v})$ with no $\gamma$ — an error, fixed here with a provenance note: dropping the $\gamma$ breaks the scalar character of $u\cdot R$ and, downstream, every power-counting in the radiation formulas.

Step 3. Extract the components of (2) and watch the cancellation. Time component ( $A^0 = V/c$ ):

A^0 = \frac{\mu_0 qc}{4\pi}\,\frac{\gamma c}{\gamma\kappa c\mathscr{r}} = \frac{\mu_0 q c}{4\pi\,\kappa\mathscr{r}} \qquad\Longrightarrow\qquad V = \frac{1}{4\pi\varepsilon_0}\,\frac{q}{\kappa\,\mathscr{r}},

using $\mu_0 c^2 = 1/\varepsilon_0$ . Spatial components:

\mathbf{A} = \frac{\mu_0 qc}{4\pi}\,\frac{\gamma\mathbf{v}}{\gamma\kappa c\mathscr{r}} = \frac{\mu_0 q\,\mathbf{v}}{4\pi\kappa\mathscr{r}} = \frac{\mathbf{v}}{c^2}\,V.

Both lines of the Griffiths-register result, recovered exactly, with the $\gamma$ of the four-velocity annihilating the $\gamma$ of the invariant in every component. No relativistic factor survives into the potentials themselves. Why this cancellation is the theorem and not a curiosity: if the enhancement factor were a Lorentz-transformation effect, a surviving $\gamma$ would mark it. Instead the surviving factor is $\kappa^{-1}$ alone — the twin's apparent-volume count — and relativity's entire contribution is the packaging: one four-vector equation whose form is frame-proof. Liénard had the content seven years before Einstein supplied the form. *Verified: rest frame $u = (c, \mathbf{0})$ , $\kappa = 1$ , $\gamma = 1$ gives Coulomb; the numeric audit of Derivation A checks both components at $\beta = 0.9$ ; matches Griffiths Eqs. 10.46–10.47 and Jackson Eq. 14.8.*

Figure

One invariant three readings — the denominator u·R unpacked in the lab frame: Einstein's gamma cancels against the numerator, Liénard's kappa survives into observables, Coulomb's distance anchors the static limit

The lemma bank

The potential is only a way station: fields are derivatives, and every derivative of anything retarded must route through the dependence of $\tau_r$ on the observation point. The 3+1 treatment needed two separate identities ( $\partial t_r/\partial t$ and $\nabla t_r$ ), each earned by implicit differentiation with vector bookkeeping. Covariantly there is exactly one identity, and it is three lines long.

Go deeper · Derivation C · 1 minThe covariant retarded time gradient and the differentiation kit

Derivation CThe covariant retarded time gradient and the differentiation kit

The goal is $\partial_\mu \tau_r$ — how the retarded proper time responds to moving the observation event — plus the two corollary derivatives ( $\partial_\mu R^\nu$ and $\partial_\mu(u\cdot R)$ ) that together make differentiating the potential (2) a purely mechanical exercise. This kit is the article's deliverable to Fields of a Relativistic Moving Charge, which will spend it on $F^{\mu\nu} = \partial^\mu A^\nu - \partial^\nu A^\mu$ . The strategy is the covariant edition of implicit differentiation: $\tau_r(x)$ is defined by the light-cone constraint, so differentiate the constraint.

Step 1. Differentiate the constraint. The retarded proper time is defined by $(x - r(\tau_r))^2 = 0$ (past branch). Apply $\partial_\mu$ , remembering that $\tau_r$ depends on $x$ :

\partial_\mu\Big[(x - r(\tau_r))_\nu (x - r(\tau_r))^\nu\Big] = 2\,(x - r)_\nu\Big[\delta^\nu_{\;\mu} - u^\nu\,\partial_\mu\tau_r\Big] = 0,

since $\partial_\mu x^\nu = \delta^\nu_{\;\mu}$ and $\partial_\mu r^\nu(\tau_r) = u^\nu\,\partial_\mu\tau_r$ by the chain rule.

Step 2. Solve. With $R_\nu = (x - r)_\nu$ :

R_\mu - (u\cdot R)\,\partial_\mu\tau_r = 0 \qquad\Longrightarrow\qquad \boxed{\;\partial_\mu \tau_r = \frac{R_\mu}{u\cdot R}\;}

Why this one line contains both 3+1 lemmas: take the time component. $\partial t_r/\partial t = \gamma\,c\,\partial_0\tau_r$ (converting proper to coordinate time along the worldline, $dt_r = \gamma\,d\tau_r$ , and $\partial_0 = \partial/c\partial t$ ), and $R_0 = \mathscr{r}$ , so $\partial t_r/\partial t = \gamma c\,\mathscr{r}/(\gamma\kappa c\mathscr{r}) = 1/\kappa$ — the first identity of the fields article one tier down. Take the spatial components: $R_i = -\mathscr{r}\,\hat{\mathscr{r}}_i$ (lowering costs a sign), so $\nabla t_r = \gamma\,\nabla\tau_r = -\hat{\boldsymbol{\mathscr{r}}}/(c\kappa)$ — the second. Two hard-won 3+1 identities are the shadow components of one covariant gradient.

Step 3. Bank the corollaries. Differentiating $R^\nu = x^\nu - r^\nu(\tau_r)$ with Step 2:

\partial_\mu R^\nu = \delta^\nu_{\;\mu} - u^\nu\,\partial_\mu\tau_r = \delta^\nu_{\;\mu} - \frac{u^\nu R_\mu}{u\cdot R},

and differentiating the scalar $u\cdot R = u_\nu(\tau_r)R^\nu$ with the product rule (the hit on $u_\nu$ produces the four-acceleration $a_\nu = du_\nu/d\tau$ ):

\partial_\mu(u\cdot R) = (a\cdot R)\,\partial_\mu\tau_r + u_\nu\,\partial_\mu R^\nu = u_\mu + \big(a\cdot R - c^2\big)\,\frac{R_\mu}{u\cdot R},

using $u_\nu u^\nu = c^2$ . These three boxed-or-banked lines — $\partial_\mu\tau_r$ , $\partial_\mu R^\nu$ , $\partial_\mu(u\cdot R)$ — are the complete differentiation kit: every derivative the field computation needs reduces to them plus the quotient rule. Verified: contract Step 2 with $u^\mu$ : $u^\mu\partial_\mu\tau_r = (u\cdot R)/(u\cdot R) = 1$ — moving the observation event along the charge's own direction of motion advances the retarded clock at unit rate, as it must; the 3+1 shadows match Griffiths Eqs. 10.71/10.78 exactly; dimensional audit: $[\partial_\mu\tau_r] = [\text{length}]/[\text{length}\cdot\text{velocity}] = \text{time/length}$ ✓.

Limiting cases

case	result	meaning
instantaneous rest frame	$u\cdot R = c\mathscr{r}_0$ , $A^\mu = (\mu_0 qc/4\pi)(c,\mathbf{0})/c\mathscr{r}_0$	Coulomb, exactly — the normalization anchor
slow motion $\beta \ll 1$	$A^\mu \approx \tfrac{\mu_0 q}{4\pi\mathscr{r}}(c, \mathbf{v})(1 + \hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta})$	first-order beaming; $\mathbf{A}$ is the elementary current's potential
uniform velocity	closed form in present position (worked example 1)	the covariant algebra performs the twin's extrapolation automatically
head-on, ultrarelativistic	$u\cdot R \to c\mathscr{r}/2\gamma$ ; components enhanced by $\kappa^{-1} \approx 2\gamma^2$	the invariant shrinks as $1/\gamma$ while components blow up as $\gamma^2$ — worked example 2's ledger
$v \to c$ in a medium	two piercings of the effective cone	the uniqueness theorem's loophole; owned by the twin's FAQ and by Cherenkov Radiation downstream

Common trapThe gamma in the denominator does not mean the potential is relativistically enhanced — and the proper time is not the coordinate time

Two traps, both about $\gamma$ . Trap 1: reading $u\cdot R = \gamma\kappa c\mathscr{r}$ as "relativity suppresses the potential by $\gamma$ ." The $\gamma$ in the denominator cancels identically against the $\gamma$ in $u^\mu = \gamma(c,\mathbf{v})$ , component by component — Derivation B watched it happen. The surviving physics is $\kappa^{-1}$ , which contains no $\gamma$ and is pure retardation counting; the twin's warning (the factor is not Lorentz contraction, not Doppler, not charge growth) inherits unchanged. Write the potential with one $\gamma$ present and the other forgotten and every downstream field acquires a spurious power of $\gamma$ — the single most common transcription error in covariant radiation theory, and precisely the error in this article's base text, where the invariant was quoted $\gamma$ -less. Trap 2: conflating $\tau_r$ with $t_r$ . The lemma $\partial_\mu\tau_r = R_\mu/(u\cdot R)$ governs the proper time; converting to coordinate-time derivatives costs the factor $\gamma$ ( $dt_r = \gamma\,d\tau_r$ ), and it is exactly this conversion that turns the covariant lemma into the 3+1 pair. Differentiate a retarded quantity with the wrong clock and the fields of the next article come out wrong by $\gamma$ in the acceleration term — a wrong power law in the ultrarelativistic limit, not a small error.

Worked examples

1 · Uniform motion, two routes to one number (confidence builder)

Setup. A charge $q = 1$ nC glides along $x$ at $\beta = 0.8$ , passing the origin at $t = 0$ . An observer sits broadside at $(0, 1\,\text{m}, 0)$ and reads the potential at $t = 0$ , when the charge's present position is the origin, a present distance $R = 1$ m away. Evaluate $V$ by the covariant formula and check against the present-position closed form the Griffiths-register twin derived.

Step 1. Solve the light-cone condition. With $\mathbf{w}(t') = \beta c t'\hat{\mathbf{x}}$ , the constraint $|\mathbf{x} - \mathbf{w}(t_r)| = -ct_r$ (note $t_r < 0$ ) squares to $\beta^2c^2t_r^2 + (1\,\text{m})^2 = c^2t_r^2$ , so $ct_r = -\gamma\times 1\,\text{m} = -5/3$ m: numerically $t_r = -5.559$ ns, retarded position $x_r = \beta c t_r = -4/3$ m, retarded distance $\mathscr{r} = -ct_r = 5/3$ m. (Clean fractions are no accident: $\gamma = 5/3$ at $\beta = 0.8$ .)

Step 2. Assemble the invariant. $\hat{\boldsymbol{\mathscr{r}}}\cdot\boldsymbol{\beta} = (4/3)/(5/3)\times 0.8 = 0.64$ , so $\kappa = 0.36$ and

u\cdot R = \gamma\,\kappa\,c\,\mathscr{r} = \frac{5}{3}\times 0.36 \times c \times \frac{5}{3} = 1.0\,c\;\text{m}.

The invariant equals $c$ times exactly one meter — the present distance, in disguise. Then $V = cA^0 = \dfrac{q}{4\pi\varepsilon_0}\dfrac{\gamma c}{u\cdot R} = \dfrac{8.99\,\text{V}\,\cdot\,\text{m} \times 5/3}{1\,\text{m}} = \mathbf{14.98\ V}$ .

Step 3. Cross-check against the present-position form. The twin's closed form gives $V = \dfrac{q}{4\pi\varepsilon_0 R\sqrt{1 - \beta^2\sin^2\theta}} = \dfrac{8.99}{\sqrt{1 - 0.64}} = \dfrac{8.99}{0.6} = \mathbf{14.98\ V}$ ✓ — and the same number falls out of the boost route in The Four-Potential, which transforms the rest-frame Coulomb potential directly; three formalisms, one voltmeter reading.

The lesson: $\kappa\mathscr{r} = 0.36 \times 5/3 = 0.6\,\text{m} = R\sqrt{1-\beta^2\sin^2\theta}$ — the retarded bookkeeping and the present-position bookkeeping are the same number wearing different clothes, and the covariant invariant $u\cdot R = \gamma c\,(\kappa\mathscr{r})$ carries it frame-independently.

2 · The invariant ledger of a 1 GeV electron (exam level)

Setup. A 1 GeV electron approaches head-on. At retarded distance $\mathscr{r} = 10$ m, tabulate every factor in the covariant potential and locate where the famous enhancement actually lives.

Step 1. Kinematics. $\gamma = E/m_ec^2 = 10^9/(5.11\times10^5) = 1957$ ; head-on $\kappa = 1 - \beta = 1.306\times10^{-7}$ , and the ultrarelativistic shortcut $1-\beta \approx 1/2\gamma^2 = 1.306\times10^{-7}$ agrees to seven digits.

Step 2. The invariant. $u\cdot R = \gamma\kappa c\mathscr{r} = 1957 \times (1.306\times10^{-7}) \times c \times 10 = 7.66\times10^{5}$ m²/s. Note the compact ultrarelativistic form: $u\cdot R \approx c\mathscr{r}/2\gamma$ — the invariant shrinks with energy, but only as $1/\gamma$ , not as $1/\gamma^2$ .

Step 3. The potential. $V = \dfrac{q}{4\pi\varepsilon_0}\dfrac{\gamma c}{u\cdot R} = \dfrac{q}{4\pi\varepsilon_0}\dfrac{1}{\kappa\mathscr{r}}$ : numerically $V = V_{\rm static}/\kappa = 1.44\times10^{-10}\,\text{V} / 1.306\times10^{-7} = \mathbf{1.10\ mV}$ , against a static charge's $0.144$ nV at the same distance — enhancement $\kappa^{-1} = 7.66\times10^{6} = 2\gamma^2$ .

Step 4. The ledger, audited. Where did $7.7\times10^6$ come from? Not from $\gamma$ in the numerator — that $\gamma = 1957$ cancelled against the denominator's. Entirely from $\kappa^{-1}$ : light-cone counting. The relativistic energy's only role was to push $\beta$ close enough to 1 that the counting factor exploded. (The twin's beaming ledger quoted $2\gamma^2 = 7.7\times10^8$ for a 10 GeV machine — same law, two orders up: the enhancement runs as the square of the energy.)

The lesson: in covariant clothing the enhancement bookkeeping is audit-proof — one invariant, three factors, and only one of them ( $\kappa$ ) survives into observables.

Going further

References. Jackson, Classical Electrodynamics, 3rd ed., §14.1 · Griffiths, Introduction to Electrodynamics, 4th ed., §10.3 · Zangwill, Modern Electrodynamics, §23.2 · Rohrlich, Classical Charged Particles, Ch. 4 · Liénard, L'Éclairage Électrique 16 (1898); Wiechert, Ann. Phys. 309 (1901)

Provenance. Equations cross-checked against Griffiths Introduction to Electrodynamics 4th ed. §10.3 (Eqs. 10.46-10.72), Jackson Classical Electrodynamics 3rd ed. §14.1 (Eqs. 14.8-14.14), and Zangwill Modern Electrodynamics §23.2. A verifier should reproduce: (1) the Jacobian factor $\kappa=1-\hat{\mathbf{n}}\cdot\boldsymbol{\beta}$ via the delta-function derivation; (2) $\partial t_r/\partial t=1/\kappa$ and $\nabla t_r=-\hat{\mathbf{n}}/(c\kappa)$ ; (3) the velocity/acceleration field split with $\kappa^{-3}$ scaling; (4) the constant-velocity Heaviside form $V=q/(4\pi\varepsilon_0 R\sqrt{1-\beta^2\sin^2\theta})$ . The relation $\mathbf{A}=(\mathbf{v}/c^2)V$ and $\mathbf{B}=\hat{\mathbf{n}}\times\mathbf{E}/c$ are standard textbook identities. Addendum for this edition (Jackson-register twin): the 3+1 derivations of items (1)–(2) and the field split (3) are owned by the landed Griffiths-register articles Liénard–Wiechert Potentials and Fields of a Moving Point Charge — recalled by name throughout, never re-run; this edition owns the proper-time route (Derivation A), the invariant anatomy $u\cdot R = \gamma\kappa c\mathscr{r}$ (Derivation B), and the covariant lemma bank (Derivation C), with the field-tensor computation deferred to Fields of a Relativistic Moving Charge. The invariant Green-function route to the same potential is acknowledged as the sibling treatment's (Covariant Wave Equation and Retarded Solutions, mid-flight at this writing — cross-reference by name). ⚠ Base error fixed: the base's covariant eq-card glossed the invariant as $u\cdot(x-w) = (\mathscr{r}c - \boldsymbol{\mathscr{r}}\cdot\mathbf{v})$ , omitting the $\gamma$ ; the correct evaluation is $\gamma(\mathscr{r}c - \boldsymbol{\mathscr{r}}\cdot\mathbf{v})$ , and the omission would corrupt every downstream power of $\gamma$ . Title retitled from the base's "Liénard-Wiechert Potentials" to break the within-track duplicate against the E9 article (track recon; trap-6 policy). Worked numbers ( $t_r = -5.559$ ns, $\mathscr{r} = 5/3$ m, $\kappa = 0.36$ , $V = 14.98$ V dual-route; $\gamma = 1957$ , $\kappa = 1.306\times10^{-7}$ , $u\cdot R = 7.66\times10^5$ m²/s, $V = 1.10$ mV vs $0.144$ nV static) recomputed for this edition, including a machine-precision numerical audit of (2) against the 3+1 pair on a circular $\beta = 0.9$ worldline.

Builds on: Retarded Potentials · Lorenz Gauge and the Wave Equation · Special Relativity Kinematics · Dirac Delta Function Leads to: Fields of a Relativistic Moving Charge · Larmor Formula from the Liénard–Wiechert Fields · Relativistic Generalization of Larmor's Formula · Angular Distribution of Radiation from Accelerated Charges · Synchrotron Radiation · Bremsstrahlung

Why bother with the covariant formulation if the 3+1 potentials already contain the same physics?

Three reasons, in ascending order. Economy: two potentials plus a constraint plus two differentiation identities become one four-vector, one scalar, one lemma. Auditability: the $\gamma$ -vs- $\kappa$ bookkeeping — which the 3+1 treatment must argue about (is the factor relativistic or not?) — becomes a two-line cancellation you can watch happen; that is how this edition caught the missing $\gamma$ in its own base text. Leverage: everything downstream (the field tensor, the covariant Larmor formula, the null-field theorem, synchrotron theory) is built by differentiating this object, and the covariant lemma bank makes those computations mechanical where the 3+1 versions are heroic. The Griffiths-register twin is where the physics is discovered; this formulation is where it becomes equipment.

Is the delta-function in proper time the same trick as the 3+1 auxiliary-time integral?

It is the same trick executed one level deeper. The twin's Derivation B introduced an auxiliary time $t'$ to linearize the composed delta and then paid the Jacobian $g'(t_r) = \kappa$ . Here the worldline parameter $\tau$ plays the auxiliary role from the start, and the Jacobian is $|ds/d\tau| = 2u\cdot R$ with $s = (x-r)^2$ — which, unpacked, is $2\gamma\kappa c\mathscr{r}$ : the twin's $\kappa$ times the factors ( $\gamma$ from the clock conversion, $2c\mathscr{r}$ from the quadratic constraint) that the covariant packaging keeps attached. Neither route is more correct; the covariant one simply refuses to let you separate the Jacobian from its invariant meaning, which is why transcription errors (like a dropped $\gamma$ ) are easier to catch.

The formula says the four-potential is parallel to the four-velocity. Is that a gauge statement or physics?

Both, carefully separated. In the Lorenz gauge — which the retarded potentials inhabit, and which is the gauge this whole chapter runs in — the statement is exact physics-of-this-gauge: the only four-vector data at the piercing point, at first order, is $u^\mu$ , so $A^\mu \propto u^\mu$ is forced by covariance. Under a gauge transformation $A^\mu \to A^\mu + \partial^\mu\lambda$ the parallelism can be destroyed, so it is not gauge-invariant content by itself. What is invariant is everything built antisymmetrically from derivatives: the field tensor of the next article, whose velocity part remembers the parallelism as the boosted-Coulomb structure. The honest summary: $A^\mu \parallel u^\mu$ is the Lorenz-gauge fingerprint of a structureless point charge — a composite or polarized source adds terms.

Where exactly did the derivation choose retarded over advanced boundary conditions?

In one theta function. Step 3 of Derivation A inserted $\theta(x^0 - r^0(\tau))$ to select the past sheet of the double cone $\delta((x-r)^2)$ — drop it and the delta happily fires on the future sheet too, producing the advanced potential with denominator $u\cdot R_{\rm adv}$ (which unpacks to $\gamma c\mathscr{r}(1 + \hat{\boldsymbol{\mathscr{r}}}_a\cdot\boldsymbol{\beta}_a)$ , the sign-flipped Jacobian the twin's FAQ derived). The selection is data, not law — the doctrine settled in Advanced Potentials and Causality: quiescent-past boundary conditions pick retarded. The covariant packaging adds one guarantee worth stating: because null ordering is frame-invariant (the invariant-order theorem), the split into past and future sheets is Lorentz-proof — no boost can turn a retarded solution into an advanced one.

Does the covariant potential apply when the charge accelerates violently, or is it a constant-velocity result?

Fully general — that is its quiet strength. Nothing in Derivation A restricted the worldline: $r^\mu(\tau)$ may wiggle, spiral, or slam to a stop, and (2) remains exact, with all data read at the piercing point $\tau_r$ . The acceleration simply does not appear in the potential — it first enters observables through differentiation, because $\partial_\mu$ moves the piercing point along the worldline and thereby samples $a^\mu = du^\mu/d\tau$ (the $a\cdot R$ term in the banked derivative of $u\cdot R$ ). This is the covariant restatement of the twin's punchline that all news is news of acceleration: the potential of a swerving charge looks exactly like the potential of a gliding one until you take a derivative. The fields article cashes precisely this observation into the velocity/acceleration split.

How do I actually solve for the retarded proper time in practice?

Analytically only in privileged cases: uniform velocity (a quadratic — worked example 1), uniform proper acceleration (hyperbolic motion, still closed-form), and circular motion (transcendental but one-dimensional). Everything else is numerical, and the standard recipe is exactly what this article's own verification used: parameterize the worldline, define $f(\tau) = (x - r(\tau))^2$ restricted to $r^0 < x^0$ , and root-find — $f$ is monotonic in the relevant bracket for timelike worldlines (the uniqueness theorem guarantees a single sign change), so bisection or Brent's method converges unconditionally. Production beam codes tabulate worldline history and interpolate the crossing. The one practical warning: at ultrarelativistic head-on geometry the root becomes ill-conditioned in coordinate time (the light cone and worldline nearly parallel — $\kappa \sim 10^{-7}$ in worked example 2), and solving in proper time, where the Jacobian is the well-behaved $u\cdot R$ , is numerically kinder. The covariant formulation is not just prettier; it conditions better.