Abraham-Lorentz Radiation Reaction Force

Two articles in the radiation chapter told the Griffiths story of the self-force: Radiation Reaction Force extracted $\mathbf{F}_{\rm rad} = m\tau\dot{\mathbf{a}}$ from energy balance with the Schott fine print honest, and Abraham–Lorentz Equation and Its Pathologies solved the third-order equation, cataloged its pathologies, and executed the Landau–Lifshitz cure. This article is the Jackson systematic layer above that pair, and it owns three things the Griffiths treatment only gestures at. First, the validity-domain theorem: the single time $\tau = 2r_e/3c = 6.266\times10^{-24}\,$ s is promoted from a curiosity to a ruler, and one estimate — run for bounded motion and for collisions alike — proves that radiation reaction is a perturbation of relative size $\omega\tau$ everywhere classical electrodynamics is itself valid. The scale where the perturbation would reach unity is $\hbar\omega = \tfrac{3}{2}m_ec^2/\alpha \approx 105$ MeV — two hundred electron masses deep into quantum territory. The pathologies live outside the theory that generated them. Second, the covariant equation: the Lorentz–Abraham–Dirac four-force, assembled here by the cleanest route available — covariantize the jerk, watch it fail the orthogonality constraint $u_\mu F^\mu = 0$ , and let the constraint force the famous second term $u^\mu(a\cdot a)/c^2$ upon you. The anatomy that falls out (a Schott four-momentum plus a Larmor drain along the four-velocity) turns the E9 energy ledger into one covariant statement. Third, the Landau–Lifshitz reduction, covariant edition — the equation strong-field codes actually integrate — with the honest QED handoff boundary.

Key takeaways

The characteristic time $\tau = \mu_0q^2/6\pi mc = 2r_e/3c = 6.266\times10^{-24}$ s (electron) is the article's ruler: radiation reaction is a perturbation of relative size $\omega\tau$ , where $\omega$ is the fastest frequency in the motion.
The validity-domain theorem: $\omega\tau \to 1$ requires $\hbar\omega \approx \tfrac{3}{2}m_ec^2/\alpha = 105$ MeV $= 206\,m_ec^2$ — but QED replaces classical electrodynamics already at $\hbar\omega \sim m_ec^2$ . Classical radiation reaction never exceeds parts-per-thousand strength inside its own domain; the runaway/preacceleration pathologies live outside it.
The covariant Lorentz–Abraham–Dirac (LAD) force is $F^\mu_{\rm rad} = m\tau\left[\dot{a}^\mu + u^\mu(a\cdot a)/c^2\right]$ ; the second term is forced by the constraint $u_\mu F^\mu_{\rm rad} = 0$ , because $u\cdot\dot{a} = -a\cdot a \neq 0$ .
Anatomy: $F^\mu_{\rm rad} = -\,dp_S^\mu/d\tau_p - (P/c^2)\,u^\mu$ with Schott four-momentum $p_S^\mu = -m\tau\,a^\mu$ and covariant Larmor power $P = -m\tau\,(a\cdot a)$ — the E9 Schott ledger, upgraded to one four-vector identity.
Hyperbolic motion (uniform proper acceleration) gives $F^\mu_{\rm rad} = 0$ exactly — the charge radiates $P = m\tau g^2$ while feeling nothing; the Schott account pays covariantly.
The covariant Landau–Lifshitz force replaces $\dot{a}^\mu$ using the external field: second-order, causal, pathology-free, identical to LAD to first order in $\tau$ — the working equation of storage-ring and strong-field codes, valid until the quantum parameter $\chi = \gamma E_{\rm rest}/E_S \sim 1$ .
Key idea: the self-force on a point charge, made covariant, is $m\tau[\dot{a}^\mu + u^\mu(a\cdot a)/c^2]$ — the jerk term covariantized plus the unique completion along $u^\mu$ that orthogonality demands — and it is a small, well-behaved perturbation everywhere classical physics holds.

The core insight

The self-force on a point charge, to leading order, is $\mathbf{F}_{\rm rad} = m\tau\dot{\mathbf{a}}$ with $\tau = \mu_0q^2/6\pi mc$ — proportional to the jerk, not the acceleration — and its unique covariant completion, $F^\mu_{\rm rad} = m\tau[\dot{a}^\mu + u^\mu(a\cdot a)/c^2]$ , is fixed by the requirement that a four-force keep $u\cdot u = c^2$ .

Where this lives today

The covariant Landau–Lifshitz equation (4) is production code: storage-ring suites and plasma-acceleration codes integrate it — or Monte-Carlo quantum extensions — for every electron, and its damping rates set the equilibrium emittance of every light source (systematics: the relativistic Larmor articles' territory). At the strong-field frontier the equation became an experimental subject: all-optical collisions at Gemini-class petawatt lasers (2018) steered GeV-scale bunches into $a_0 > 10$ pulses and watched the energy distribution respond to the self-force within one optical cycle — the first direct dynamical sighting of radiation reaction, at $\chi \sim 0.1$ where classical and quantum-corrected models begin to disagree (the data lean quantum). ELI, FACET-II, and LUXE are built to walk up the $\chi$ dial deliberately. Astrophysics runs the equation at values no laboratory reaches: in pulsar and magnetar magnetospheres ( $B \sim 10^8$ – $10^{11}$ T) particles settle onto radiation-reaction-limited trajectories — the drain balances the accelerating field, and the emitted curvature and synchrotron radiation is the observed pulse. The "Aristotelian electrodynamics" of modern gap simulations is the limit where the LL drain dominates inertia — this article's perturbation promoted, at last, to the leading term.

Origins · 1 minMax Abraham and how this was discovered

Max Abraham

1875–1922 · Göttingen — the vector that bears his name found its true home in the theory he fought

This track has met Abraham twice: as the fiercest craftsman of the rigid-electron program (E9's radiation-reaction article, where his 1902–03 electromagnetic momentum and the Kaufmann episode are told), and as one pole of the photon-in-matter momentum controversy. Here he appears for the piece of him that is in the equation: the four-vector $\Gamma^\mu = m\tau[\dot{a}^\mu + u^\mu(a\cdot a)/c^2]$ is called the Abraham vector, because the 1905 edition of his Theorie der Elektrizität — the textbook that trained a generation of European theorists — already contained the relativistically correct self-force structure, derived by a man who did not believe in relativity. When Dirac's 1938 worldtube analysis produced the covariant equation from stress-energy conservation alone, the expression that emerged was Abraham's — orthogonality term and all, now wearing proper time. The irony is exact: the completion term exists because four-velocity is normalized, a constraint of the very kinematics Abraham fought to his death in 1922, yet his algebra survived the change of worldview untouched. The equation is named fairly for all three: Lorentz's shell, Abraham's vector, Dirac's derivation.

History

1902

— Abraham computes the rigid electron's self-force and electromagnetic momentum; the program of electromagnetic mass peaks

1905

— Abraham's Theorie der Elektrizität contains the full velocity-dependent reaction term — the Abraham vector, pre-relativity

1909

— Born solves uniform proper acceleration; hyperbolic motion becomes the self-force's sharpest test case

1912

— Schott names the acceleration energy; the reversible account in the radiation ledger gets its owner

1938

— Dirac derives the covariant LAD equation from stress-energy conservation on a worldtube, renormalizing the mass covariantly

1951

— the Landau-Lifshitz Classical Theory of Fields canonizes the reduction of order as the working equation

2018

— Gemini-laser collision experiments observe radiation reaction dynamically at chi of order 0.1, on the classical-quantum fence

The physical picture

Every question in this article is a question about clocks. The self-force has one intrinsic time, $\tau \approx 6.3\times10^{-24}$ s — light crossing two-thirds of a classical electron radius. Any motion has its own fastest clock: a cyclotron period, an optical cycle, a collision duration. Radiation reaction compares the two. When the motion's clock ticks slowly against $\tau$ — and everything classical does — the self-force is a parts-per-billion correction riding on ordinary dynamics. Making the two clocks comparable needs hundred-MeV photons, and classical electrodynamics ended two decades of energy earlier. This is the effective-theory reading, and it reframes the famous pathologies. The E9 pathology article showed that the Abraham–Lorentz equation, taken literally, offers a choice between runaway solutions and acausal preacceleration, both confined to the scale $\tau$ . The systematic view adds the missing theorem: no experiment inside the classical domain can ever probe that scale, because the domain boundary (set by $\hbar$ ) sits at $\omega\tau \sim \alpha \ll 1$ . The equation misbehaves only where it was never entitled to speak — the same logic by which modern physics trusts effective field theories up to their cutoff and not beyond. The second picture is geometric. In relativity a legitimate four-force must be orthogonal to the four-velocity — otherwise it changes $u\cdot u = c^2$ , which is to say it changes the particle's rest mass. The naive covariantization of the jerk force fails this test by exactly the amount $-a\cdot a$ (a positive number: four-acceleration is spacelike). The fix is unique: add the right multiple of $u^\mu$ . That forced correction term turns out to be the covariant Larmor power flowing along the worldline — the equation's energy bookkeeping is welded into its geometry.

Figure

The clock ladder — every classical motion ticks far slower than tau, and the quantum boundary arrives two decades before omega tau reaches one

The objects: the ruler, the force, the constraint

Three imports and one new requirement organize everything. From the E9 pair (by name, never re-derived): the self-force and its clock,

\mathbf{F}_{\rm rad} = \frac{\mu_0 q^2}{6\pi c}\,\dot{\mathbf{a}} = m\tau\,\dot{\mathbf{a}}, \qquad \tau = \frac{\mu_0 q^2}{6\pi m c} = \frac{2}{3}\frac{r_e}{c} = 6.266\times10^{-24}\ \text{s (electron)}. \qquad (1)

where: the energy-balance extraction of (1), the Schott identity $\mathbf{F}_{\rm rad}\cdot\mathbf{v} = -P + \tfrac{d}{dt}(m\tau\,\mathbf{v}\cdot\mathbf{a})$ , and the shell calculation with its mass renormalization all belong to Radiation Reaction Force; the runaway/preacceleration solutions and the 3+1 Landau–Lifshitz cure belong to Abraham–Lorentz Equation and Its Pathologies. This article never re-runs them. From Relativistic Generalization of Larmor's Formula: the covariant radiated power,

P = -\frac{\mu_0 q^2}{6\pi c}\,a_\nu a^\nu = -m\tau\,(a\cdot a), \qquad a^\mu = \frac{du^\mu}{d\tau_p}, \qquad (2)

where: $a\cdot a = a_\nu a^\nu < 0$ on any worldline (four-acceleration is spacelike, metric $(+,-,-,-)$ ), so $P > 0$ always; $\tau_p$ is proper time. And from Relativistic Dynamics of a Charged Particle: the constraint. Differentiating $u\cdot u = c^2$ gives $u\cdot a = 0$ ; a four-force that respects the rest mass must satisfy $u_\mu F^\mu = 0$ (the Lorentz force passes by antisymmetry). Whatever the covariant self-force is, it must pass too — and that single requirement writes the equation's second term without any appeal to models. An ownership table keeps the imports straight:

quantity	formula	belongs to
NR self-force	$m\tau\dot{\mathbf{a}}$	E9 Radiation Reaction Force (derivation) — used here as the reduction target
covariant power	$P = -m\tau(a\cdot a)$	Relativistic Generalization of Larmor's Formula — used here as the drain term
covariant self-force	$m\tau[\dot{a}^\mu + u^\mu(a\cdot a)/c^2]$	this article (Derivation B)
working equation	covariant Landau–Lifshitz	this article (Derivation C, working level)

Derivation AThe validity domain of classical radiation reaction

The goal is Jackson's §16.1 estimate promoted to a theorem: radiation reaction is a relative perturbation of size $\omega\tau$ (bounded motion) or $\tau/T$ (collisions), and the strength-one boundary $\omega\tau = 1$ corresponds to $\hbar\omega = \tfrac{3}{2}m_ec^2/\alpha \approx 105$ MeV — far beyond the Compton scale $\hbar\omega \sim m_ec^2$ where QED has already replaced the classical theory. The strategy: compare radiated energy to mechanical energy on the motion's own timescale, in the two canonical regimes, then locate the boundary with $\hbar$ inserted only at the last step.

Step 1. Bounded motion. Let the charge oscillate with amplitude $d$ and frequency $\omega$ : typical velocity $v \sim \omega d$ , acceleration $a \sim \omega^2 d$ . Over one period $T = 2\pi/\omega$ the Larmor loss is $E_{\rm rad} \sim m\tau a^2\,T \sim m\tau\omega^3 d^2\cdot 2\pi$ , while the mechanical energy is $E_{\rm mech} \sim m\omega^2d^2$ . The ratio:

\frac{E_{\rm rad}}{E_{\rm mech}} \sim 2\pi\,\omega\tau.

Why this is also the force ratio: $|\mathbf{F}_{\rm rad}|/|\mathbf{F}_{\rm ext}| \sim m\tau\omega^3 d / m\omega^2 d = \omega\tau$ — the same dimensionless dial, whether you compare energies per cycle or forces at an instant. One parameter governs the whole subject.

Step 2. Collisions. Let an external force act for a duration $T$ , producing velocity change $\Delta v \sim aT$ . Radiated energy $E_{\rm rad} \sim m\tau a^2 T$ ; kinetic energy delivered $\sim m(aT)^2$ . Ratio $\sim \tau/T$ : the same statement with the collision time in the denominator. Both regimes unify into: radiation reaction matters only when the motion changes appreciably within a time $\tau$ .

Step 3. Price the ladder. Electron in a 1 T field: $\omega_c = eB/m_e = 1.76\times10^{11}$ s⁻¹ gives $\omega\tau = 1.1\times10^{-12}$ . Optical transition (600 nm): $\omega\tau = 2.0\times10^{-8}$ — the classical linewidth scale the E9 article computed as spectroscopy. A 1 MeV gamma frequency: $\omega\tau = 9.5\times10^{-3}$ . Even at the Compton frequency $m_ec^2/\hbar = 7.8\times10^{20}$ s⁻¹ — the largest frequency classical electron physics can name — $\omega\tau = 4.9\times10^{-3}$ : half a percent.

Step 4. The boundary theorem. Set $\omega\tau = 1$ and insert $\hbar$ :

\hbar\omega_{\rm crit} = \frac{\hbar}{\tau} = \frac{6\pi\varepsilon_0\hbar m_e c^3}{q^2} = \frac{3}{2}\,\frac{m_ec^2}{\alpha} = 105.0\ \text{MeV} = 206\,m_ec^2,

using $\alpha = q^2/4\pi\varepsilon_0\hbar c$ . The frequency where classical radiation reaction reaches order one exceeds the frequency where classical electrodynamics itself dies ( $\hbar\omega \sim m_ec^2$ : pair creation, Compton recoil) by the factor $3/2\alpha \approx 206$ . Conclusion, stated as the domain theorem: within the entire domain of validity of classical electrodynamics, radiation reaction is a small perturbation — bounded above by $\sim\alpha$ , and in every laboratory situation far smaller. The pathologies at scale $\tau$ can never be classically observed; the effective-theory reading of the E9 pathology article is thereby derived, not just adopted. Verified: $\tau = 6.266\times10^{-24}$ s from CODATA ( $\mu_0q^2/6\pi m_ec$ and $2r_e/3c$ agree to 9 digits); $\hbar/\tau = 105.05$ MeV vs $\tfrac{3}{2}m_ec^2/\alpha = 105.04$ MeV; ladder values recomputed. Matches Jackson §16.1 (Eqs. 16.3–16.8) in structure and numbers.

The covariant equation — orthogonality writes the second term

The nonrelativistic force (1) cannot be the whole story: it is not a four-vector statement, and storage rings, strong-field lasers, and pulsar magnetospheres all operate at $\gamma \gg 1$ . The systematic route to the relativistic completion — Dirac's 1938 stress-energy computation on a worldtube — is deferred by name to Runaway Solutions and Preacceleration, where its machinery earns its keep. Here we take the structural route, which is shorter, honest about what it assumes, and delivers the identical equation: covariantize, then repair.

Go deeper · Derivation B · 2 minThe Lorentz-Abraham-Dirac four-force from the orthogonality constraint

Derivation BThe Lorentz-Abraham-Dirac four-force from the orthogonality constraint

The goal is the LAD self-force $F^\mu_{\rm rad} = m\tau[\dot{a}^\mu + u^\mu(a\cdot a)/c^2]$ , with the second term derived as the unique completion demanded by $u_\mu F^\mu_{\rm rad} = 0$ , followed by its anatomy (Schott four-momentum plus Larmor drain) and the nonrelativistic reduction audit. The strategy: write the obvious covariant guess, test it against the constraint using two differentiation lemmas, and fix it minimally. All dots denote $d/d\tau_p$ ; metric $(+,-,-,-)$ .

Step 1. Two lemmas from the normalization of four-velocity. Differentiating $u\cdot u = c^2$ along the worldline: $u\cdot a = 0$ . Differentiating again:

\frac{d}{d\tau_p}(u\cdot a) = a\cdot a + u\cdot\dot{a} = 0 \qquad\Longrightarrow\qquad u\cdot\dot{a} = -\,a\cdot a.

Why this is the engine: the jerk four-vector is never orthogonal to $u$ on an accelerating worldline — its projection along $u$ is $-(a\cdot a) > 0$ whenever the charge accelerates. Any force built from $\dot{a}^\mu$ alone is illegal.

Step 2. The naive guess and its failure. The obvious covariantization of (1) is $F^\mu_{\rm guess} = m\tau\,\dot{a}^\mu$ — in the instantaneous rest frame its spatial part is $m\tau\dot{\mathbf{a}}$ , as required. But by Step 1, $u_\mu F^\mu_{\rm guess} = -m\tau\,(a\cdot a) = +P \neq 0$ : the guess injects four-work at the covariant Larmor rate, steadily pumping the rest mass. A self-force that changes the electron's mass is a bookkeeping error, not a force.

Step 3. The unique repair. Add a multiple of the only vector available along the worldline that can absorb the projection — $u^\mu$ itself: $F^\mu_{\rm rad} = m\tau[\dot{a}^\mu + \lambda\,u^\mu]$ . Enforce the constraint:

u_\mu F^\mu_{\rm rad} = m\tau\left[-a\cdot a + \lambda\,c^2\right] = 0 \qquad\Longrightarrow\qquad \lambda = \frac{a\cdot a}{c^2},

\boxed{\;F^\mu_{\rm rad} = \frac{\mu_0 q^2}{6\pi c}\left[\frac{d^2u^\mu}{d\tau_p^2} + \frac{u^\mu}{c^2}\left(\frac{du_\nu}{d\tau_p}\frac{du^\nu}{d\tau_p}\right)\right] = m\tau\left[\dot{a}^\mu + \frac{u^\mu}{c^2}(a\cdot a)\right].\;} \qquad (3)

Why "unique": any additional orthogonal piece would survive the constraint, but it would also survive into the rest frame and spoil the reduction to (1), which is fixed by the E9 energy-balance derivation. Requiring (i) four-vector, (ii) rest-frame reduction to $m\tau\dot{\mathbf{a}}$ , (iii) orthogonality, pins (3) exactly. Dirac's worldtube computation lands on the same expression with the additional dividend that the divergent self-energy is renormalized covariantly — his machinery is the advanced pathology article's subject, deferred by name.

Step 4. Anatomy: Schott four-momentum plus Larmor drain. Define $p_S^\mu \equiv -m\tau\,a^\mu$ (the Schott four-momentum) and recall $P = -m\tau(a\cdot a)$ from (2). Then (3) rearranges identically into

F^\mu_{\rm rad} = -\frac{dp_S^\mu}{d\tau_p} \;-\; \frac{P}{c^2}\,u^\mu.

The second piece is an irreversible drain of four-momentum along the worldline — radiation leaving at the covariant Larmor rate, exactly as Relativistic Generalization of Larmor's Formula prices it. The first is a total proper-time derivative: a reversible loan account whose rest-frame time component reproduces the E9 Schott identity, now revealed as the $\mu = 0$ shadow of one four-vector statement. The near-field reservoir of the E9 ledger is $p_S^\mu$ ; borrowing and repayment are frame-independent bookkeeping.

Step 5. Reduction audit. In the instantaneous rest frame $u^\mu = (c, \mathbf{0})$ , $a^\mu = (0, \mathbf{a})$ , $a\cdot a = -a^2$ , and $\dot{a}^\mu = (a^2/c,\, \dot{\mathbf{a}})$ (the time component follows from Step 1's lemma). The spatial part of (3): $m\tau\dot{\mathbf{a}}$ — equation (1) exactly, with the $u^\mu$ term contributing nothing spatially at $v = 0$ . The time component: $m\tau[a^2/c - a^2/c] = 0$ — no rest-mass pumping, by construction. Verified (sympy): on a generic numeric worldline, $u\cdot a = 0$ , $u\cdot\dot a = -a\cdot a$ , and $u_\mu F^\mu_{\rm rad} = 0$ hold to stencil precision; on the hyperbolic worldline $u^\mu = c(\cosh(g\tau_p/c), \sinh(g\tau_p/c), 0, 0)$ the force (3) evaluates to the zero four-vector exactly (symbolic). Coefficient matches Jackson Eq. 16.55 and the base card; rest-frame reduction matches Griffiths Eq. 11.80.

Figure

The covariant ledger — the self-force splits into a reversible Schott loan account and an irreversible Larmor drain along the worldline; hyperbolic motion is the case where the force is zero and the loan account pays the whole radiation bill

The working equation — Landau–Lifshitz, covariant edition

Equation (3) inherits the third-derivative disease of its nonrelativistic shadow — worse, since runaways now come in Lorentz-boosted families; their solution theory belongs to Runaway Solutions and Preacceleration. What this article owns is the reduction of order, covariant edition: the tool accelerator and laser-plasma codes actually integrate.

Go deeper · Derivation C · 2 minThe covariant Landau-Lifshitz force at working level

Derivation CThe covariant Landau-Lifshitz force at working level

The goal is the Landau–Lifshitz (LL) four-force: replace the jerk inside the small self-force term using the zeroth-order equation of motion $ma^\mu = qF^{\mu\nu}u_\nu$ , producing a second-order, causal equation identical to LAD to first order in $\tau$ . The E9 pathology article executed this reduction in 3+1 for the oscillator and cyclotron touchstones; here the covariant edition is assembled at working level — every term identified physically, the full index bookkeeping quoted where it is pure algebra. Validity: the substitution is legitimate precisely because Derivation A proved the self-force term is always small in the classical domain.

Step 1. The substitution principle. In $m\,da^\mu/d\tau_p = F^\mu_{\rm ext} + F^\mu_{\rm rad}$ the self-force is $O(\omega\tau)$ relative to the external force. To first order in $\tau$ it is therefore consistent — not an approximation on top of the theory but an approximation matching its accuracy — to evaluate $F^\mu_{\rm rad}$ on the zeroth-order motion: $a^\mu = (q/m)F^{\mu\nu}u_\nu$ with $F^{\mu\nu}$ the external field tensor.

Step 2. Differentiate along the worldline. The jerk becomes

\dot{a}^\mu = \frac{q}{m}\left[(\partial_\alpha F^{\mu\nu})\,u^\alpha u_\nu + F^{\mu\nu}a_\nu\right] = \frac{q}{m}(\partial_\alpha F^{\mu\nu})u^\alpha u_\nu + \frac{q^2}{m^2}F^{\mu\nu}F_{\nu\alpha}u^\alpha,

substituting the zeroth-order $a_\nu$ again in the second piece. The first term is the field-gradient contribution — the charge feels its field change because it moves through an inhomogeneous $F^{\mu\nu}$ ; the second is the field-squared contribution — the field re-accelerates the already-field-driven velocity.

Step 3. The Larmor completion term. In $u^\mu(a\cdot a)/c^2$ , substitute once in each factor: $a\cdot a = (q/m)^2(F^{\alpha\nu}u_\nu)(F_{\alpha\beta}u^\beta)$ . Assembling all three pieces:

F^\mu_{\rm LL} = \frac{\mu_0q^2}{6\pi c}\left[\frac{q}{m}(\partial_\alpha F^{\mu\nu})u_\nu u^\alpha + \frac{q^2}{m^2}\left(F^{\mu\nu}F_{\nu\alpha}u^\alpha + \frac{1}{c^2}(F^{\alpha\nu}u_\nu)(F_{\alpha\beta}u^\beta)\,u^\mu\right)\right]. \qquad (4)

where: no derivative of the motion higher than first appears — (4) is an ordinary velocity-dependent force. Standard initial data $(x^\mu, u^\mu)$ suffice; no runaway mode exists because the homogeneous equation has no self-exciting solution; causality is manifest because only the local external field and its gradient enter. When $F^{\mu\nu} = 0$ , the self-force is identically zero — a free charge coasts, with no memory of the pathologies.

Step 4. When LAD and LL differ — and the quantum fence. The two equations differ at relative order $O(\tau^2\omega^2)$ — by Derivation A at most $\sim\alpha^2 \sim 5\times10^{-5}$ anywhere classical physics applies. Observable differences would need a non-small self-force term, which the domain theorem forbids classically. The interesting boundary is quantum: define $\chi = \gamma E_\perp/E_S$ (rest-frame field over the Sauter–Schwinger field $E_S = m_e^2c^3/e\hbar = 1.32\times10^{18}$ V/m). A Gemini-class collision — an $a_0 = 13.7$ laser ( $4\times10^{20}$ W/cm², 800 nm) against a 500 MeV electron — reaches $\chi \approx 0.08$ : classical LL already needs quantum corrections (stochastic photon recoil), and at $\chi \gtrsim 1$ the deterministic self-force dissolves into strong-field QED. The 2018 experiments sit exactly on this fence — the first dynamical sighting of radiation reaction, and the first data preferring quantum-corrected models. The handoff belongs to Bridge from Classical Electrodynamics to QED. Verified: (4) matches the base card and Landau & Lifshitz §76 (their Eq. 76.3 in SI translation) term by term; boosted-frame check $\omega'\tau = 2\gamma\omega\tau = 2.9\times10^{-5}$ for the Gemini parameters (correctly small — LL self-consistent there); $E_S$ and $\chi$ recomputed from CODATA.

Limiting cases

case	result	meaning
$v \ll c$	$F^\mu_{\rm rad} \to (0,\, m\tau\dot{\mathbf{a}})$	Abraham–Lorentz force recovered; E9 pair's regime
uniform proper acceleration	$F^\mu_{\rm rad} = 0$ exactly ( $\dot{a}^\mu = -u^\mu(a\cdot a)/c^2$ identically)	hyperbolic motion: radiates $P = m\tau g^2$ , feels nothing; Schott account pays
$F^{\mu\nu} = 0$ (LL)	$F^\mu_{\rm LL} = 0$ identically	free particles coast: no runaway in the working equation
slow variation, $\omega\tau \ll 1$	LAD $=$ LL $+\,O(\tau^2\omega^2)$	all classical physics identical in the two equations
$\chi = \gamma E/E_S \to 1$	stochastic photon recoil; classical overestimates loss	quantum radiation reaction: hand off to QED
circular ultrarelativistic	drain term dominates: $F_{\rm rad} \approx (P/c^2)\,u$ , $P \propto \gamma^4$	storage-ring damping; Relativistic Generalization of Larmor's Formula prices it

Common trapThe completion term is not optional bookkeeping — and tau-scale structure in solutions is not physics

Trap 1. Dropping $u^\mu(a\cdot a)/c^2$ from (3) "because it is small" is never legitimate: it is the same order in $\tau$ as the jerk term, and without it the self-force does four-work $u\cdot F = P \neq 0$ — pumping the particle's rest mass at the Larmor rate. Truncated-force simulations show mass drift, not damping. The term is the equation's energy conservation, welded in by geometry. Trap 2. Features of LAD solutions that live at the scale $\tau$ — the runaway growth time, the preacceleration window, sub- $\tau$ transients — are artifacts of using an effective theory beyond its domain (Derivation A: the classical domain ends two orders of magnitude before $\omega\tau \sim 1$ ). Report them as properties of the equation, never as predictions about electrons — those scales are QED's business; the systematic solution theory belongs to Runaway Solutions and Preacceleration.

Worked examples

1 · The half-percent self-force — LEP at full power (confidence builder)

Setup. At LEP's final energy ( $E = 100$ GeV, bending radius $\rho = 3096$ m) synchrotron loss was famously brutal. How large was the radiation-reaction force compared to the magnetic bending force — the largest ratio any classical machine has produced?

Step 1. Kinematics. $\gamma = E/m_ec^2 = 1.96\times10^5$ ; the covariant Larmor power for circular motion (owned by the relativistic Larmor articles, quoted): $P = e^2c\gamma^4/6\pi\varepsilon_0\rho^2 = 7.05\times10^{-6}$ W per electron — per turn, $2\pi\rho P/c = \mathbf{2.86}$ GeV, matching the E9 anchor.

Step 2. The two forces. The self-force is dominated by the drain term: $|\mathbf{F}_{\rm rad}| \approx P/c = 2.35\times10^{-14}$ N. The external (bending) force: $F_{\rm ext} = \gamma m_e c^2/\rho = 5.17\times10^{-12}$ N. Ratio:

\frac{F_{\rm rad}}{F_{\rm ext}} = \mathbf{4.5\times10^{-3}}.

Step 3. Read it against the domain theorem. Half a percent — the largest self-force-to-external-force ratio in accelerator history, achieved by pushing $\gamma^4$ as hard as economics allowed, and still parts-per-thousand: Derivation A's bound in action. Yet the cumulative effect is enormous — 3% of the beam energy per turn, replaced by RF — because a tiny force applied every one of $10^4$ turns per second compounds. Radiation reaction in the classical domain is always this: instantaneously negligible, secularly decisive (the damping bonus is priced in Relativistic Generalization of Larmor's Formula).

The lesson: "perturbation" does not mean "ignorable" — it means the force never competes with the dynamics locally, while its time-integral can dominate the energy budget.

2 · Hyperbolic motion — the covariant Schott payment, audited (exam level)

Setup. An electron rides a uniform-gradient linac field $E = 30$ MV/m, undergoing (idealized) uniform proper acceleration $g = eE/m_e = 5.28\times10^{18}$ m/s². Show the LAD force vanishes exactly, compute the radiated power, and audit who pays.

Step 1. The worldline. Hyperbolic motion: $u^\mu = c(\cosh(g\tau_p/c), \sinh(g\tau_p/c), 0, 0)$ . Then $a^\mu = g(\sinh, \cosh, 0, 0)$ , and the invariants: $a\cdot a = g^2\sinh^2 - g^2\cosh^2 = \mathbf{-g^2}$ , constant — the defining property.

Step 2. The force vanishes exactly. $\dot{a}^\mu = (g^2/c)(\cosh, \sinh, 0, 0) = (g^2/c^2)\,u^\mu$ , so

F^\mu_{\rm rad} = m\tau\left[\frac{g^2}{c^2}u^\mu + u^\mu\frac{(-g^2)}{c^2}\right] = \mathbf{0} \quad\text{(exactly, all components, sympy-verified)}.

Step 3. The radiation is real anyway. $P = -m\tau(a\cdot a) = m\tau g^2 = \mathbf{1.59\times10^{-16}}$ W — about $3.3\ \mu$ eV radiated per meter of linac: real photons, detectable in principle, negligible in practice (the linac loophole the E9 relativistic Larmor article adjudicated).

Step 4. The audit: who pays, covariantly. The Schott four-momentum $p_S^\mu = -m\tau a^\mu$ has time component $p_S^0 = -m\tau g\sinh(g\tau_p/c)$ , growing ever more negative. Its drain rate: $-c\,dp_S^0/d\tau_p = m\tau g^2 c\cosh/c = \gamma P$ — per unit coordinate time, exactly $P$ . The near-field loan account runs a deficit at precisely the radiated rate, forever, while the particle's mechanical four-momentum evolves as if the self-force did not exist. The E9 puzzle — "uniform acceleration radiates but feels no jerk force; who pays?" — is settled here in one covariant line: the Schott four-vector pays, at rate $P$ , in every frame.

The lesson: the split $F^\mu_{\rm rad} = -\dot{p}_S^\mu - (P/c^2)u^\mu$ is not decoration — on hyperbolic motion the pieces cancel as forces while the ledger underneath moves real energy into the field.

Going further

References. Jackson, Classical Electrodynamics, 3rd ed., §16.1–16.5 · Griffiths, Introduction to Electrodynamics, 4th ed., §11.2.2–11.2.3 · Zangwill, Modern Electrodynamics, §23.5–23.6 · Landau & Lifshitz, The Classical Theory of Fields, §75–76 · Rohrlich, Classical Charged Particles (2007) · Dirac, "Classical theory of radiating electrons," Proc. R. Soc. A 167 (1938)

Provenance. Equations cross-checked against Griffiths 4th ed. §11.2.2-11.2.3 (AL force, $\tau$ , runaway/pre-acceleration), Jackson 3rd ed. §16.2-16.5 (energy balance derivation, LAD equation), and Landau-Lifshitz §76 (Landau-Lifshitz reduction). Verifier should recompute: (1) $\tau=\mu_0 q^2/6\pi m c=(2/3)(r_e/c)$ giving $\approx6.26\times10^{-24}\,\mathrm{s}$ for the electron using $r_e=2.818\times10^{-15}\,\mathrm{m}$ , $c=2.998\times10^8\,\mathrm{m/s}$ ; (2) the integration-by-parts step transforming $\int a^2\,\mathrm{d}t$ ; (3) the force-free runaway $a\propto e^{t/\tau}$ . The classical electron radius and $\tau$ are standard textbook values; no numbers fabricated. Addendum for this edition: per the chapter's division of labor, item (2) — the energy-balance derivation — and the Schott identity are owned by the E9 article Radiation Reaction Force, and item (3) — the runaway/preacceleration solutions and the 3+1 Landau–Lifshitz reduction — by Abraham–Lorentz Equation and Its Pathologies; both are recalled by name and never re-derived. This edition owns: the validity-domain theorem with the $\hbar/\tau = \tfrac{3}{2}m_ec^2/\alpha = 105.0$ MeV boundary (identity verified to four digits against CODATA), the covariant LAD assembly from the orthogonality constraint (both differentiation lemmas derived; force verified symbolically to vanish on the hyperbolic worldline and to satisfy $u_\mu F^\mu = 0$ on generic numeric worldlines), the Schott-four-momentum anatomy $F^\mu_{\rm rad} = -\dot p_S^\mu - (P/c^2)u^\mu$ , and the covariant Landau–Lifshitz force at working level. Worked numbers (LEP: $P = 7.05\times10^{-6}$ W, 2.86 GeV/turn, force ratio $4.5\times10^{-3}$ · linac 30 MV/m: $g = 5.28\times10^{18}$ m/s², $P = 1.59\times10^{-16}$ W · Gemini-class: $a_0 = 13.7$ , $\chi = 0.08$ , $\omega'\tau = 2.9\times10^{-5}$ · validity ladder: $\omega\tau = 1.1\times10^{-12}$ at 1 T, $2.0\times10^{-8}$ optical, $4.9\times10^{-3}$ at the Compton frequency) all recomputed for this edition. Deep runaway/preacceleration systematics including Dirac's worldtube machinery are deferred by name to Runaway Solutions and Preacceleration; the natural linewidth to Radiative Frequency Shift and Line Width; the electromagnetic-mass and 4/3 story to Classical Electron Theory and Self-Energy; quantum radiation reaction to Bridge from Classical Electrodynamics to QED.

Builds on: Larmor Formula from the Liénard–Wiechert Fields · Liénard–Wiechert Potentials: Covariant Formulation · Relativistic Generalization of Larmor's Formula · Fields of a Relativistic Moving Charge Leads to: Synchrotron Radiation · Relativistic Generalization of Larmor's Formula · Angular Distribution of Radiation from Accelerated Charges

What exactly do the two E9 articles own, and what does this one add?

Radiation Reaction Force (E9) owns the case for a self-force — the energy-balance derivation of $m\tau\dot{\mathbf{a}}$ , the boundary-term fine print, the Schott identity, the Lorentz shell with its renormalization. Abraham–Lorentz Equation and Its Pathologies (E9) owns the equation as a dynamical system — runaway mode, step-force preacceleration solved, 3+1 Landau–Lifshitz reduction. This article adds the layer above both: the validity-domain theorem (the E9 articles state the smallness; this one derives its universality), the covariant LAD equation with the completion term forced by orthogonality, the Schott four-momentum anatomy, and the covariant LL working equation with the quantum fence. The classic story lives there; the working covariant equation and its trust theorem live here.

Does the covariant LAD equation still have runaways, or did covariance cure them?

Still: the force-free covariant equation admits self-accelerating worldlines (proper acceleration growing as $e^{\tau_p/\tau}$ ), and boosting a runaway gives another runaway — the disease comes in Lorentz families. Covariance was never the problem; the third derivative was, and (3) still contains $\dot{a}^\mu$ . What kills the pathologies in practice is the reduction of order (Derivation C), which is available covariantly and is exactly as accurate as the original equation within its domain. The systematic solution theory — the covariant general solution, Dirac's asymptotic conditions, why excising runaways buys preacceleration of duration $\tau$ , and what the modern integro-differential reformulations do about it — is the next article's territory: Runaway Solutions and Preacceleration, deferred by name.

If radiation reaction is at most a parts-per-thousand force, how does it dominate the design of every electron storage ring?

Through time, not strength. The domain theorem bounds the instantaneous ratio $F_{\rm rad}/F_{\rm ext}$ ; it says nothing about the time-integral. A LEP electron lost 3% of its energy every turn (worked example 1) — a 0.45% force applied along $2\times10^4$ m of arc, $10^4$ times per second: without RF replacement the beam dies in milliseconds. The same accumulation with the sign turned useful is radiation damping — the drain removes transverse momentum, the RF restores only longitudinal, and oscillations shrink exponentially (systematics priced in Relativistic Generalization of Larmor's Formula). Rings are machines for harvesting a perturbation: locally negligible, globally in charge.

Is the Schott four-momentum real, or just a term moved to the other side of the equation?

As real as the E9 Schott energy it generalizes — real, locatable in the near field, spendable, but reversible. The identity $F^\mu_{\rm rad} = -\dot{p}_S^\mu - (P/c^2)u^\mu$ is a rearrangement, but a canonical one: it separates the total derivative (nets to zero whenever the acceleration returns to its initial value) from the secular drain (never nets to zero). The physical content: $p_S^\mu = -m\tau a^\mu$ is the four-momentum parked in the velocity-acceleration cross terms of the charge's bound field. Hyperbolic motion is the regime that makes it undeniable — the force vanishes exactly while photons flow to infinity, and the audit (worked example 2) shows the Schott account draining at precisely the Larmor rate in every frame. An accounting fiction cannot pay real bills indefinitely; this one does.

Why does the Landau-Lifshitz equation get to modify the equation of motion — is not that just changing the theory to avoid inconvenient solutions?

It changes no physics at the order the theory is defined. The LAD force was derived as the leading term of an expansion in $\tau$ (the E9 shell story: divergent inertia renormalized, first finite term kept). An equation trusted only to $O(\tau)$ does not own its $O(\tau^2)$ consequences — and the runaways are precisely non-perturbative-in- $\tau$ artifacts, with essential singularity $e^{t/\tau}$ at $\tau \to 0$ . Substituting the zeroth-order motion into the $O(\tau)$ term (Derivation C) produces an equation identical through $O(\tau)$ with no spurious solutions: same theory, consistently truncated. Whether an exact classical point-charge dynamics exists to be approximated is open and moot — Derivation A puts the scales where it would matter inside QED's territory. The move is standard effective-theory discipline, four decades early.

Where did the divergent self-energy go in this article — the shell calculation had to subtract an infinite mass?

Into $m$ , where it always goes: every equation here uses the observed mass — bare mass plus electromagnetic self-inertia — the subtraction the E9 shell derivation executed at structure level and Dirac's worldtube analysis performs covariantly (the divergent piece isolates as a term $\propto a^\mu$ , absorbable by definition of mass). The self-force (3) is what survives: finite, universal, structure-independent. The divergence's full story — electromagnetic mass, the 4/3 scandal, Poincaré stresses, the QED comparison — is the next-door article's entire subject: Classical Electron Theory and Self-Energy.