Relativistic Dynamics and the Four-Force

By 1945 the cyclotron — the machine that had opened nuclear physics — had hit an invisible wall. Protons circling in a fixed magnetic field began arriving late at the accelerating gap, slipping out of step with the radio-frequency push, and no engineering could fix it. The culprit was a single factor of $\gamma$ : the orbital frequency is not $qB/m$ but $qB/\gamma m$ , and as the protons gained energy their internal clock of revolution slowed. The repair — ramping fields to chase the slipping frequency — created the synchrotron and, eventually, the LHC. Relativistic dynamics is the bookkeeping that makes such machines possible. The deeper problem is that Newton's $\mathbf{F} = m\mathbf{a}$ is not a law all inertial observers can share: it differentiates with respect to one frame's time and presumes velocity-independent inertia. The covariant repair is minimal and beautiful — keep "force equals rate of change of momentum," but let the momentum be the four-momentum of the previous article and let the rate be per tick of proper time. The result, the Minkowski four-force $K^\mu = \mathrm{d}p^\mu/\mathrm{d}\tau$ , transforms identically on both sides, so a law verified in one frame holds in all. Unpacking the components pays immediately: the space part reproduces $\mathbf{F} = \mathrm{d}\mathbf{p}/\mathrm{d}t$ , the time part hands you the work–energy theorem free of charge, and the geometric constraint $K^\mu u_\mu = 0$ — the four-force is always orthogonal to the four-velocity — encodes the conservation of rest mass. Along the way you learn why force and acceleration stop being parallel, why pushing a fast particle forward is $\gamma^2$ times harder than pushing it sideways, and why every accelerator on Earth measures its beams in energy rather than speed.

Key takeaways

Newton's second law survives relativity in the form $K^\mu = \mathrm{d}p^\mu/\mathrm{d}\tau$ : four-vector momentum, differentiated by invariant proper time — the same equation in every inertial frame.
In components, $K^\mu = \gamma\,(\mathbf{F}\cdot\mathbf{v}/c,\ \mathbf{F})$ with $\mathbf{F} = \mathrm{d}\mathbf{p}/\mathrm{d}t$ : the space part is the ordinary force (times $\gamma$ ), the time part is the power — the work–energy theorem is built in, not postulated.
Differentiating the fixed norm $u^\mu u_\mu = c^2$ gives the orthogonality constraint $K^\mu u_\mu = c^2\,\mathrm{d}m/\mathrm{d}\tau$ : pure mechanical forces satisfy $K^\mu u_\mu = 0$ and conserve rest mass; heat-like transfers violate it.
Force and acceleration are not parallel: $\mathbf{F} = \gamma m\,\mathbf{a} + \gamma^3 m\,(\mathbf{v}\cdot\mathbf{a})\,\mathbf{v}/c^2$ . Longitudinal pushes fight inertia $\gamma^3 m$ , transverse ones only $\gamma m$ — the reason speed saturates at $c$ while energy grows without bound.
The electromagnetic Lorentz force is the canonical four-force, $K^\mu = q\,F^{\mu\nu}u_\nu$ ; the antisymmetry of the field tensor makes $K^\mu u_\mu = 0$ automatic, so EM fields can never change a particle's rest mass.
Constant proper acceleration integrates to hyperbolic motion — the relativistic rocket — with speed asymptoting to $c$ along a spacetime hyperbola.
Key idea: Replace $\mathbf{F}=\mathrm{d}\mathbf{p}/\mathrm{d}t$ by the covariant $K^\mu=\mathrm{d}p^\mu/\mathrm{d}\tau$ ; the time component encodes power and the spatial component is $\gamma\mathbf{F}$ , while constancy of rest mass forces $K^\mu u_\mu=0$ .

Where this lives today

Every accelerator complex on Earth is applied four-force bookkeeping. The LHC's 1,232 superconducting dipoles hold 6.8 TeV protons on a 2,804 m bending radius at 8.1 tesla, exactly $B = p/qr$ ; its successor studies (the Future Circular Collider) scale the same formula to 90 km. Medical linacs — thousands of them, in hospital basements — push electrons against $\gamma^3 m$ inertia to a few MeV and slam them into tungsten for therapy X-rays; proton-therapy gantries steer beams with the transverse $\gamma m$ . Beam-dynamics codes do not integrate $\mathbf{F} = m\mathbf{a}$ ; they integrate the four-force, because at $\gamma > 7000$ nothing else is even qualitatively correct. The constant-acceleration problem, meanwhile, became a gateway to deeper physics. Hyperbolic motion defines the Rindler wedge — the natural coordinates of a uniformly accelerating observer — where the mathematics of horizons first appears in special relativity, foreshadowing black-hole thermodynamics and the Unruh effect studied in the quantum field theory sequence. That a mechanics-course exercise about rockets shares its geometry with event horizons is one of the field's better advertisements.

The physical picture

Film a particle and tag each frame with the reading of a clock the particle carries — its proper time. Dynamics in relativity asks how the four-momentum arrow changes per tick of that intrinsic clock. Because the four-velocity has fixed length $c$ (previous article), forces can only turn the arrow in spacetime, never stretch it: that single geometric fact is the orthogonality constraint, and almost everything else in this article is its component-by-component unpacking.

Figure

Pushing a fast particle — the same force applied along the motion fights inertia gamma cubed m, applied across the motion only gamma m; forward pushes mostly buy energy, not speed

The orthogonality $K^\mu u_\mu = 0$ has a vivid rest-frame meaning. In the particle's instantaneous rest frame the four-velocity points purely along time, so the four-force must point purely along space — it is exactly the ordinary push you would feel, the proper force. A component "along the worldline" would pump energy into the particle at rest, changing its rest mass; a pure mechanical force does not do that. When rest mass does change — a rocket burning fuel internally, a particle absorbing radiation — the contraction $K^\mu u_\mu$ measures precisely the rate of that internal energy injection. The asymmetry between pushing forward and pushing sideways is the everyday face of the formalism. A forward push must raise $\gamma$ — expensive, since energy diverges as $v \to c$ — while a sideways push at constant speed leaves $\gamma$ untouched. Hence the longitudinal inertia $\gamma^3 m$ versus the milder transverse inertia $\gamma m$ : the historical "longitudinal and transverse masses" of Lorentz's era, now understood not as two masses but as one invariant $m$ seen through two directions of the same geometry. And the time component? It is power in disguise. Energy and momentum share one four-vector, so a law for momentum change necessarily contains the law for energy change — the work–energy theorem stops being an independent postulate and becomes the zeroth component of geometry. What does a constant force do, run forever? Newton answers: uniform acceleration, speed growing without bound. Relativity answers: hyperbolic motion — a worldline that bends ever closer to the light cone without touching it, the accelerometer reading steady while the lab-frame speed saturates. This one exactly-solvable problem is the cleanest showcase of the whole formalism, and the worked examples fly it to the stars at one gee.

Figure

Hyperbolic motion — a particle under constant proper acceleration traces a spacetime hyperbola that hugs the light cone forever; equal ticks of proper time crowd together in coordinate time as gamma grows

The core equations

Throughout, the metric signature is $(+,-,-,-)$ , matching the four-momentum article. The kinematic actors are the four-velocity and four-momentum built there:

u^\mu = \frac{\mathrm{d}x^\mu}{\mathrm{d}\tau} = \gamma\,(c,\ \mathbf{v}), \qquad u^\mu u_\mu = c^2 \qquad (1)

where: $\tau$ — proper time, $\mathrm{d}\tau = \mathrm{d}t/\gamma$ ; $\gamma = 1/\sqrt{1 - v^2/c^2}$ . The constant norm is the loaded gun of this article: anything whose norm is fixed can only be rotated by dynamics, and differentiating this constancy will hand us the orthogonality constraint.

p^\mu = m\,u^\mu = \left(\frac{E}{c},\ \mathbf{p}\right), \qquad p^\mu p_\mu = m^2c^2 \qquad (2)

where: $m$ — invariant rest mass; $E = \gamma mc^2$ ; $\mathbf{p} = \gamma m\mathbf{v}$ . Nothing new — this is the arrow whose response to forces we now legislate. The law itself is Newton's second law with both of its frame-dependent ingredients repaired — momentum promoted to a four-vector, time promoted to a scalar:

K^\mu = \frac{\mathrm{d}p^\mu}{\mathrm{d}\tau} \qquad (3)

where: $K^\mu$ — the four-force (Minkowski force). Both sides are four-vectors, so a Lorentz transformation maps the equation onto itself: covariance is manifest, not checked after the fact. For constant $m$ this is also $m\,\mathrm{d}u^\mu/\mathrm{d}\tau = m a^\mu$ with $a^\mu$ the four-acceleration. Unpacking into laboratory quantities (Derivation A) exposes what each slot carries:

K^\mu = \gamma\left(\frac{\mathbf{F}\cdot\mathbf{v}}{c},\ \mathbf{F}\right), \qquad \mathbf{F} = \frac{\mathrm{d}\mathbf{p}}{\mathrm{d}t} \qquad (4)

where: $\mathbf{F}$ — the ordinary three-force, still defined as the rate of change of (relativistic) three-momentum; $\mathbf{F}\cdot\mathbf{v}$ — the power delivered. The overall $\gamma$ is the chain rule $\mathrm{d}t/\mathrm{d}\tau$ ; forgetting it is the classic bookkeeping error. Setting $\mathbf{v} = 0$ shows the rest-frame four-force is $(0, \mathbf{F}_{\mathrm{rest}})$ — pure push, no time component. Differentiating the fixed norm of $u^\mu$ (Derivation A again) yields the structural constraint of relativistic dynamics:

K^\mu u_\mu = c^2\,\frac{\mathrm{d}m}{\mathrm{d}\tau} \qquad (5)

where: the right side vanishes for a pure force — one that changes motion but not internal energy. Then $K^\mu u_\mu = 0$ : the four-force is Minkowski-perpendicular to the worldline, and the work–energy theorem $\mathrm{d}E/\mathrm{d}t = \mathbf{F}\cdot\mathbf{v}$ is its component shadow. A four-force with $K^\mu u_\mu \neq 0$ is pumping rest mass — heating, radiating, burning. The canonical pure force of nature realizes all of this automatically:

K^\mu = q\,F^{\mu\nu}\,u_\nu, \qquad F^{\mu\nu} = -F^{\nu\mu} \qquad (6)

where: $q$ — charge; $F^{\mu\nu}$ — the electromagnetic field tensor, with $F^{i0} = E^i/c$ and $F^{ij} = -\epsilon^{ijk}B_k$ in this signature. Derivation C unpacks it into the Lorentz force $q(\mathbf{E} + \mathbf{v}\times\mathbf{B})$ and the power $q\mathbf{E}\cdot\mathbf{v}$ ; the antisymmetry of $F^{\mu\nu}$ makes $K^\mu u_\mu = 0$ an identity — electromagnetic fields cannot change rest mass, by symmetry alone. Finally, expressing $\mathbf{F} = \mathrm{d}(\gamma m\mathbf{v})/\mathrm{d}t$ in terms of acceleration (Derivation B) breaks the Newtonian parallelism of force and acceleration:

\mathbf{F} = \gamma m\,\mathbf{a} + \gamma^3 m\,\frac{(\mathbf{v}\cdot\mathbf{a})}{c^2}\,\mathbf{v} \qquad (7)

where: the second term points along $\mathbf{v}$ and exists only when the speed changes. Along the motion, $F_\parallel = \gamma^3 m\,a_\parallel$ ; across it, $F_\perp = \gamma m\,a_\perp$ . The $\gamma^3$ divergence is the mechanical enforcement of the speed limit. Which formula answers which question:

You want	Reach for	Because
a manifestly frame-independent law	$K^\mu = \mathrm{d}p^\mu/\mathrm{d}\tau$	four-vectors on both sides
lab dynamics from a known force	$\mathbf{F} = \mathrm{d}\mathbf{p}/\mathrm{d}t$ , $\mathrm{d}E/\mathrm{d}t = \mathbf{F}\cdot\mathbf{v}$	components of the same law
acceleration from force (or vice versa)	equation (7), split ∥ and ⊥	inertia differs by direction
to check whether a force conserves mass	compute $K^\mu u_\mu$	zero ⇔ pure force
charged-particle motion	$K^\mu = qF^{\mu\nu}u_\nu$	covariance and mass conservation built in

Derivation AComponents of the four-force and the work-energy theorem

We unpack the covariant law $K^\mu = \mathrm{d}p^\mu/\mathrm{d}\tau$ into laboratory language and show its time component is the work–energy theorem. The strategy has two moves: a chain rule converting proper-time rates to lab-time rates, and one differentiation of the fixed norm $u^\mu u_\mu = c^2$ to pin the time component to the space components. Watch how energy conservation is never assumed — it emerges as the zeroth row of a vector equation.

Step 1. Convert the derivative using time dilation, $\mathrm{d}\tau = \mathrm{d}t/\gamma$ :

K^\mu = \frac{\mathrm{d}p^\mu}{\mathrm{d}\tau} = \gamma\,\frac{\mathrm{d}p^\mu}{\mathrm{d}t}

Why: proper time is the invariant parameter, but measurements happen in lab time; $\gamma$ is the exchange rate between the two clocks.

Step 2. Read off the spatial components with $\mathbf{p} = \gamma m\mathbf{v}$ and the definition $\mathbf{F} = \mathrm{d}\mathbf{p}/\mathrm{d}t$ :

K^i = \gamma\,\frac{\mathrm{d}p^i}{\mathrm{d}t} = \gamma F^i

Step 3. Constrain the time component by differentiating the constant norm $u^\mu u_\mu = c^2$ along the worldline:

\frac{\mathrm{d}}{\mathrm{d}\tau}\left(u^\mu u_\mu\right) = 2\,u_\mu\frac{\mathrm{d}u^\mu}{\mathrm{d}\tau} = 0 \quad\Longrightarrow\quad K^\mu u_\mu = \frac{\mathrm{d}(m u^\mu)}{\mathrm{d}\tau}u_\mu = m\,u_\mu\frac{\mathrm{d}u^\mu}{\mathrm{d}\tau} + \frac{\mathrm{d}m}{\mathrm{d}\tau}\,u^\mu u_\mu = c^2\frac{\mathrm{d}m}{\mathrm{d}\tau}

Why this is the pivot: for constant rest mass the contraction vanishes — the four-force is orthogonal to the four-velocity, so its time component is not independent but determined by the spatial ones.

Step 4. Evaluate the orthogonality in components, $K^\mu u_\mu = K^0\gamma c - \gamma\mathbf{F}\cdot\gamma\mathbf{v} = 0$ :

K^0 = \frac{\gamma\,\mathbf{F}\cdot\mathbf{v}}{c} \quad\Longrightarrow\quad K^\mu = \gamma\left(\frac{\mathbf{F}\cdot\mathbf{v}}{c},\ \mathbf{F}\right)

Step 5. Equate the two expressions for $K^0$ , using $p^0 = E/c$ :

\gamma\,\frac{1}{c}\frac{\mathrm{d}E}{\mathrm{d}t} = \frac{\gamma}{c}\,\mathbf{F}\cdot\mathbf{v} \quad\Longrightarrow\quad \frac{\mathrm{d}E}{\mathrm{d}t} = \mathbf{F}\cdot\mathbf{v}

Why nothing extra was needed: energy and momentum live in one four-vector, so legislating the momentum change fixes the energy change. The work–energy theorem is the shadow of orthogonality. *Independent cross-check: differentiate the mass shell $E^2 = (pc)^2 + (mc^2)^2$ at fixed $m$ to get $E\,\mathrm{d}E = c^2\,\mathbf{p}\cdot\mathrm{d}\mathbf{p}$ , so $\mathrm{d}E/\mathrm{d}t = (c^2\mathbf{p}/E)\cdot\mathbf{F} = \mathbf{v}\cdot\mathbf{F}$ using $\mathbf{v} = c^2\mathbf{p}/E$ — confirming the work–energy theorem by a completely independent route. In the rest frame the result gives $K^\mu = (0, \mathbf{F}_{\mathrm{rest}})$ , the proper force.*

Go deeper · Derivation B · 1 minLongitudinal and transverse inertia from the momentum derivative

Derivation BLongitudinal and transverse inertia from the momentum derivative

We derive the force–acceleration relation and show the effective inertia depends on the direction of the push. The strategy is a single product rule applied to $\mathbf{p} = \gamma m\mathbf{v}$ , remembering that $\gamma$ depends on speed and therefore changes only when the speed changes. The punchline to watch for: the two clean cases (push along, push across) bracket all the messy intermediate geometry.

Step 1. Differentiate the three-momentum with the product rule:

\mathbf{F} = \frac{\mathrm{d}}{\mathrm{d}t}\left(\gamma m\mathbf{v}\right) = m\,\dot{\gamma}\,\mathbf{v} + \gamma m\,\mathbf{a}, \qquad \mathbf{a} = \frac{\mathrm{d}\mathbf{v}}{\mathrm{d}t}

Step 2. Compute $\dot{\gamma}$ by the chain rule, using $\mathrm{d}(v^2)/\mathrm{d}t = 2\,\mathbf{v}\cdot\mathbf{a}$ :

\dot{\gamma} = \frac{\mathrm{d}}{\mathrm{d}t}\left(1 - \frac{v^2}{c^2}\right)^{-1/2} = \gamma^3\,\frac{\mathbf{v}\cdot\mathbf{a}}{c^2}

Why $\gamma^3$ : differentiating the inverse square root brings down a half and raises the power to three-halves — the same $\gamma^3$ that will become the longitudinal inertia.

Step 3. Assemble the general relation:

\mathbf{F} = \gamma m\,\mathbf{a} + \gamma^3 m\,\frac{(\mathbf{v}\cdot\mathbf{a})}{c^2}\,\mathbf{v}

Why force and acceleration now misalign: the second term points along $\mathbf{v}$ regardless of where $\mathbf{a}$ points; only for pushes exactly along or exactly across the motion do the two vectors line up.

Step 4. Push along the motion ( $\mathbf{a} \parallel \mathbf{v}$ , so $\mathbf{v}\cdot\mathbf{a} = va$ ):

F_\parallel = \gamma m\,a + \gamma^3 m\,\frac{v^2}{c^2}\,a = \gamma m\,a\left(1 + \gamma^2\beta^2\right) = \gamma^3 m\,a_\parallel

Why the identity closes: $1 + \gamma^2\beta^2 = (1 - \beta^2 + \beta^2)\gamma^2 = \gamma^2$ .

Step 5. Push across the motion ( $\mathbf{v}\cdot\mathbf{a} = 0$ ):

F_\perp = \gamma m\,a_\perp

Why the price differs: a longitudinal push must inflate $\gamma$ — the diverging energy account — while a transverse push at constant speed leaves $\gamma$ alone and pays only the time-dilated inertia $\gamma m$ . Circular motion in a magnetic field lives entirely in this cheaper case. *As a check: at $\beta \to 0$ both cases collapse to $F = ma$ , as required. The algebraic identity $\gamma + \gamma^3\beta^2 = \gamma^3$ underpins Step 4. And at $v = 0.99c$ , $\gamma = 7.09$ gives $\gamma^3 = 356$ , the acceleration-suppression factor used in the linac example below.*

Go deeper · Derivation C · 1 minThe electromagnetic four-force unpacked

Derivation CThe electromagnetic four-force unpacked

We show that the covariant Lorentz force $K^\mu = qF^{\mu\nu}u_\nu$ contains the familiar three-force and the power, and that its antisymmetric structure conserves rest mass identically. The strategy: write the field tensor components in this signature, contract with $u_\nu = \gamma(c, -\mathbf{v})$ , and compare slots against equation (4). The elegance to watch: one antisymmetric tensor does the work of two separate laboratory laws.

Step 1. Fix the tensor components in signature $(+,-,-,-)$ : $F^{i0} = E^i/c$ and $F^{ij} = -\epsilon^{ijk}B_k$ , with $u_\nu = \gamma(c, -\mathbf{v})$ .

Step 2. Contract the spatial row:

K^i = q\left(F^{i0}u_0 + F^{ij}u_j\right) = q\left(\frac{E^i}{c}\,\gamma c + (-\epsilon^{ijk}B_k)(-\gamma v_j)\right) = \gamma q\left(\mathbf{E} + \mathbf{v}\times\mathbf{B}\right)^i

Why this is the right answer: comparing with $K^i = \gamma F^i$ from the component form of the four-force identifies $\mathbf{F} = q(\mathbf{E} + \mathbf{v}\times\mathbf{B})$ — the Lorentz force, now derived rather than decreed.

Step 3. Contract the time row:

K^0 = q\,F^{0j}u_j = q\left(-\frac{E^j}{c}\right)(-\gamma v_j) = \frac{\gamma}{c}\,q\,\mathbf{E}\cdot\mathbf{v}

Why $\mathbf{B}$ is absent: the magnetic force is always perpendicular to $\mathbf{v}$ , so it steers without working; the time slot records only the electric power $q\mathbf{E}\cdot\mathbf{v} = \mathrm{d}E/\mathrm{d}t$ .

Step 4. Check mass conservation by contracting with the four-velocity:

K^\mu u_\mu = q\,F_{\mu\nu}\,u^\mu u^\nu = 0

Why identically zero: $u^\mu u^\nu$ is symmetric under exchange of the indices while $F_{\mu\nu}$ is antisymmetric; a symmetric object contracted with an antisymmetric one vanishes term by term. Electromagnetic fields are structurally incapable of changing rest mass. *As a check: the $v \ll c$ limit returns the elementary Lorentz force. For a magnetic-only field $\mathrm{d}E/\mathrm{d}t = 0$ , so $\gamma$ is constant and the orbit radius follows from $F_\perp = \gamma m v^2/r = qvB$ , giving $r = \gamma mv/qB = p/qB$ — the accelerator formula used in the worked examples.*

Limiting cases

Case	Result	Meaning
Non-relativistic ( $v \ll c$ , formally $c \to \infty$ )	$K^\mu \to (\mathbf{F}\cdot\mathbf{v}/c,\ \mathbf{F})$ , $\ \mathbf{F} \to m\mathbf{a}$	Newton's second law and the Newtonian work–energy theorem recovered exactly; both inertias collapse to $m$ .
Instantaneous rest frame	$K^\mu = (0,\ \mathbf{F}_{\mathrm{rest}})$	The time component dies (no velocity, no power); what survives is the proper force — what an onboard accelerometer reads.
Pure force, constant $m$	$K^\mu u_\mu = 0$	Force only rotates the four-momentum on its mass shell; heat-like or radiative processes break this and change $m$ .
Ultra-relativistic longitudinal push	$a_\parallel = F/\gamma^3 m \to 0$	Finite force, vanishing acceleration: speed asymptotes to $c$ while energy and momentum grow without bound.
Magnetic field only	$\gamma$ const, $\ \omega = qB/\gamma m$ , $\ r = p/qB$	No work done; orbits at the relativistic cyclotron frequency — the $\gamma$ that broke the cyclotron and built the synchrotron.

Common trapThe missing gamma, and the force that no longer points where the particle accelerates

Two errors account for most wrong answers in relativistic dynamics. First, conflating the two time variables: the four-force is $\mathrm{d}p^\mu/\mathrm{d}\tau = \gamma\,\mathrm{d}p^\mu/\mathrm{d}t$ , so the spatial part of $K^\mu$ is $\gamma\mathbf{F}$ , not $\mathbf{F}$ — track which clock each derivative uses. Second, writing $\mathbf{F} = \gamma m\,\mathbf{a}$ as if $\gamma m$ were "the mass": that formula holds only for transverse pushes. In general $\mathbf{F} = \gamma m\,\mathbf{a} + \gamma^3 m(\mathbf{v}\cdot\mathbf{a})\mathbf{v}/c^2$ , force and acceleration are not parallel, and no single velocity-dependent scalar can play the role of mass — which is precisely why "relativistic mass" was retired. If your calculation needs one number for inertia, first ask in which direction the force acts.

Applications

Synchrotrons and storage rings. In a magnetic field the force is transverse, no work is done, and the orbit obeys $r = p/qB$ with angular frequency $\omega = qB/\gamma m$ . The $\gamma$ in the denominator desynchronized the wartime cyclotrons; phase-stable ramping (McMillan and Veksler, 1945) turned the bug into the synchrotron. Today the arithmetic runs the LHC: bending 6.8 TeV protons around a 2.8 km radius requires $B = p/qr \approx 8.1$ T — the design spec of its superconducting dipoles.

Linear accelerators and the speed wall. A linac pushes longitudinally, against inertia $\gamma^3 m$ . Beyond a few MeV, electrons gain essentially no speed per cavity — only energy and momentum. Designers therefore lock the RF phase to particles moving at effectively $c$ and account the machine entirely in $\mathrm{d}E = \mathbf{F}\cdot\mathrm{d}\mathbf{x}$ : energy per meter, never velocity per meter.

The relativistic rocket and constant proper acceleration. A force constant in the instantaneous rest frame ( $K^\mu K_\mu = -F_{\mathrm{rest}}^2$ fixed) integrates to hyperbolic motion: $v(t) = at/\sqrt{1 + (at/c)^2}$ , a worldline asymptotic to the light cone. This is the twin-paradox travel profile and the classical scaffold beneath Rindler coordinates — worked numerically in Example 2.

Muon storage rings. The Fermilab muon $g-2$ experiment circulated muons at the "magic" $\gamma = 29.3$ , chosen so that electric focusing fields drop out of the spin-precession bookkeeping — a design decision made entirely inside the four-force formalism. The same transverse-inertia physics sets the ring radius from the muon momentum of 3.09 GeV/c.

Worked examples

Example 1 — one force, two directions (confidence builder). An electron ( $m c^2 = 0.511$ MeV, $m = 9.11\times10^{-31}$ kg) moves at $v = 0.99c$ through a field region exerting force $F = 4.8\times10^{-12}$ N (the force of a 30 MV/m accelerating gradient). Find the acceleration if the force is (a) along the motion, (b) across it.

Step 1. Lorentz factor: $\gamma = 1/\sqrt{1 - 0.99^2} = 1/\sqrt{0.0199} = 7.09$ , so $\gamma^3 = 356$ .

Step 2. Longitudinal: $a_\parallel = \dfrac{F}{\gamma^3 m} = \dfrac{4.8\times10^{-12}}{356 \times 9.11\times10^{-31}} = \dfrac{4.8\times10^{-12}}{3.24\times10^{-28}} = 1.48\times10^{16}\ \mathrm{m\,s^{-2}}$ .

Step 3. Transverse: $a_\perp = \dfrac{F}{\gamma m} = \dfrac{4.8\times10^{-12}}{7.09 \times 9.11\times10^{-31}} = 7.43\times10^{17}\ \mathrm{m\,s^{-2}}$ — fifty times larger for the identical force.

Step 4. Newtonian prediction for comparison: $F/m = 5.27\times10^{18}\ \mathrm{m\,s^{-2}}$ . The forward push delivers only $1/356$ of the Newtonian acceleration, the sideways push $1/7.09$ — the ratio $\gamma^2 = 50.3$ between them is pure geometry, no new forces involved.

Example 2 — the one-gee starship (exam level). A rocket maintains constant proper acceleration $a = 9.8\ \mathrm{m\,s^{-2}}$ (comfortable artificial gravity). Using the hyperbolic-motion relations $v/c = \tanh(a\tau/c)$ , $t = (c/a)\sinh(a\tau/c)$ , $x = (c^2/a)[\cosh(a\tau/c) - 1]$ , find the proper time to reach $v = 0.9c$ , the elapsed Earth time, the distance covered, and the crew's $\gamma$ at that moment.

Step 1. The natural unit: $c/a = 3.0\times10^8/9.8 = 3.06\times10^7\ \mathrm{s} = 0.970$ yr — one gee for "one year" is intrinsically relativistic.

Step 2. Proper time: $a\tau/c = \operatorname{artanh}(0.9) = \tfrac12\ln\dfrac{1.9}{0.1} = \tfrac12\ln 19 = 1.472$ , so $\tau = 0.970 \times 1.472 = 1.43$ yr on the ship's clocks.

Step 3. Earth time: $t = (c/a)\sinh(1.472) = 0.970 \times \dfrac{e^{1.472} - e^{-1.472}}{2} = 0.970 \times \dfrac{4.358 - 0.229}{2} = 0.970 \times 2.064 = 2.00$ yr.

Step 4. Distance: $x = (c^2/a)[\cosh(1.472) - 1] = 0.970\,\mathrm{ly} \times (2.294 - 1) = 1.26$ ly.

Step 5. Consistency: the crew's Lorentz factor should be $\gamma = \cosh(a\tau/c) = 2.294$ , and directly $\gamma(0.9c) = 1/\sqrt{1 - 0.81} = 1/\sqrt{0.19} = 2.294$ . ✓ After 1.43 ship-years (2.00 Earth-years) and 1.26 light-years, the rocket reaches $0.9c$ with $\gamma = 2.29$ — and however long it burns, its worldline hugs but never crosses the light cone.

Origins · 1 minHermann Minkowski and how this was discovered

Hermann Minkowski

1864–1909 · Königsberg → Zürich → Göttingen

Minkowski taught Einstein mathematics at the ETH — and famously remembered him as a "lazy dog" — before Göttingen colleagues drew his attention to what his former student had done in 1905. His response reframed the theory forever: in his 1908 Cologne address he declared that "space by itself, and time by itself, are doomed to fade away into mere shadows," and rebuilt relativity as the geometry of a four-dimensional world. The four-force that bears his name completed Newton's program in that geometry — force as the proper-time rate of change of a four-vector momentum, orthogonality to the worldline replacing the separate laws of energy and momentum. He died of appendicitis months later, at forty-four; Einstein, initially dismissive of the "superfluous learnedness," adopted the geometric machinery wholesale when he needed it for general relativity.

History

1905

— Einstein's dynamics paper poses the problem: longitudinal and transverse mass differ, and Newton's law needs relativistic repair

1906

— Planck corrects Einstein's force definitions and gives relativistic momentum its modern form p equals gamma m v

1908

— Minkowski's Cologne lecture recasts relativity as spacetime geometry and defines the covariant four-force

1909

— Born analyses constant proper acceleration; hyperbolic motion becomes the exactly solvable problem of relativistic dynamics

1945

— McMillan and Veksler diagnose the cyclotron's gamma problem and invent phase stability, founding the synchrotron era

2025

— Fermilab's muon g minus 2 program concludes, its storage ring built around four-force kinematics at the magic gamma of 29.3

Going further

References. Morin, Introduction to Classical Mechanics, Ch. 12–13 · Goldstein, Poole & Safko, Classical Mechanics, 3rd ed., Ch. 7 · Landau & Lifshitz, The Classical Theory of Fields, Vol. 2, §§9, 17, 23 · Taylor, Classical Mechanics, Ch. 15 · Rindler, Introduction to Special Relativity, Ch. 5 All definitions and component forms use signature $(+,-,-,-)$ for consistency with the four-momentum article. Key numerical anchors: $u^\mu u_\mu = c^2$ and $K^\mu u_\mu = c^2\,\mathrm{d}m/\mathrm{d}\tau$ hold in this signature; the identity $\gamma + \gamma^3\beta^2 = \gamma^3$ underlies the longitudinal-mass derivation; the antisymmetry argument gives $F_{\mu\nu}u^\mu u^\nu = 0$ ; the field-tensor contraction yields $\mathbf{F} = q(\mathbf{E} + \mathbf{v}\times\mathbf{B})$ and $K^0 = \gamma q\mathbf{E}\cdot\mathbf{v}/c$ with $F^{i0} = E^i/c$ , $F^{ij} = -\epsilon^{ijk}B_k$ ; Example 1 uses $\gamma = 7.09$ , $\gamma^3 = 356$ at $v = 0.99c$ ; Example 2 uses $\operatorname{artanh} 0.9 = 1.472$ , $\sinh(1.472) = 2.064$ , $\cosh(1.472) = 2.294$ ; the LHC dipole estimate gives $B = p/qr \approx 8.1$ T at $pc = 6.8$ TeV, $r = 2804$ m, compared with the 8.33 T design value at 7 TeV.

Scope note. This is the mechanics-course introduction. The full story — covariant electrodynamics and the field tensor in depth, radiation reaction, Thomas precession, spinning particles — lives in the Special Relativity track (migrates later as its own subject).

Builds on: Special Relativity Postulates and the Lorentz Transformation · Relativistic Kinematics: Time Dilation, Length Contraction, and Velocity Addition · Four-Velocity, Four-Momentum, and Mass-Energy Equivalence Leads to: The Relativistic Lagrangian for a Free and Charged Particle · Lagrangian Density and Continuum Lagrangian Formalism · Stress Tensor and Cauchy's Relation

If force and acceleration are not parallel, what happens when I push a fast particle at some odd angle?

Decompose the force along and across the velocity. The longitudinal component produces acceleration $F_\parallel/\gamma^3 m$ ; the transverse component produces $F_\perp/\gamma m$ . Because the two divisors differ by $\gamma^2$ , the acceleration vector rotates away from the force vector, toward the transverse direction — the particle responds much more willingly sideways than forward. Only for pushes exactly along or exactly across the motion do force and acceleration align. This is not an exotic effect: beam-optics codes must apply exactly this decomposition every integration step, or their simulated bunches drift unphysically.

Is the four-force just gamma times the ordinary force? The components look that way.

The spatial part is, and that innocent-looking $\gamma$ is doing real work. $K^i = \gamma F^i$ because the four-force differentiates by proper time while $\mathbf{F}$ differentiates by lab time, and the two clocks disagree by $\gamma$ . But the four-force also carries a time component, $\gamma\mathbf{F}\cdot\mathbf{v}/c$ , which the three-force does not have — the power rides along in the same object. So the honest statement is: the four-force is the ordinary force and the work–energy theorem, packaged covariantly, with $\gamma$ as the clock-conversion fee.

What exactly would break if I used coordinate time instead of proper time in the definition?

The law would change form between frames. $\mathrm{d}p^\mu/\mathrm{d}t$ divides a four-vector by one observer's time — the result transforms messily, mixing in the velocity of the frame, so an equation true in the lab would be false on the moving particle's account. With proper time, both sides are four-vectors and the same Lorentz matrix maps both, making covariance automatic. This mirrors the four-velocity construction: relativity keeps Newton's structure (force = rate of change of momentum) and only replaces the frame-dependent ingredients with invariant ones.

When rest mass changes, what does the four-force contraction actually measure?

$K^\mu u_\mu = c^2\,\mathrm{d}m/\mathrm{d}\tau$ measures the rate at which energy is being deposited in the rest frame — internal energy, which weighs. A particle absorbing radiation, a droplet being heated, a rocket whose fuel tanks are part of "the particle": all have four-forces with a component along the worldline. The clean split is: orthogonal part of $K^\mu$ = pure push (changes motion), parallel part = heating (changes mass). Electromagnetic forces on point charges are purely orthogonal by the antisymmetry of the field tensor, which is why particle physics can treat masses as immutable labels while the four-force acts.

The cyclotron frequency has gamma in it, but the orbit radius formula r equals p over qB does not. Why the asymmetry?

Both come from the same equation, $\gamma m v^2/r = qvB$ , read two ways. Solve for the radius: $r = \gamma mv/qB = p/qB$ — and since $p$ is the natural relativistic momentum, the formula keeps its Newtonian shape with no visible $\gamma$ . Solve for the angular frequency: $\omega = v/r = qB/\gamma m$ — here the $\gamma$ sits in the open. The radius formula hides relativity inside $p$ ; the frequency formula cannot, because frequency compares motion to time, and time is where relativity bites. This is why synchrotrons must ramp their RF frequency and field together as the beam's $\gamma$ grows.

Does the relativistic rocket ever experience its own speed limit? Onboard, the acceleration feels constant forever.

Onboard, nothing ever changes: the accelerometer reads $a$ , artificial gravity holds steady, and each ship-year of burn adds the same rapidity increment $a\tau/c$ . The speed limit lives entirely in the external description: velocities compose through $\tanh$ , so piling up rapidity forever still leaves $v = c\tanh(a\tau/c) < c$ . There is no contradiction because speed is a relation between the rocket and some inertial frame, not a thing the rocket carries. What the crew does notice is the geometry: destination distances contract, Earth clocks redshift, and an event horizon forms behind them — light emitted too late from home never catches a rocket that burns forever.

Why do accelerator physicists quote beam energy instead of beam speed?

Because speed stops being informative almost immediately. An electron is at $0.99c$ by 3 MeV and at $0.9999999c$ by the GeV scale — ten significant figures of nines carry less meaning than simply saying "45 GeV." Energy and momentum keep growing linearly with what the machine does (work per cavity, bending stiffness $p/q$ ), so they are the operationally meaningful labels. The four-force formalism explains why: longitudinal force feeds $\mathrm{d}E/\mathrm{d}t = \mathbf{F}\cdot\mathbf{v}$ efficiently forever, while $\mathrm{d}v/\mathrm{d}t = F/\gamma^3 m$ dies as $\gamma^{-3}$ . Machines buy energy; speed is a saturated by-product.

Is there a relativistic version of Newton's third law between two moving charges?

Not in the naive form, and the failure is instructive. Action–reaction demands the two forces be equal and opposite at the same instant — but simultaneity is frame-dependent, and the electromagnetic influence propagates at $c$ , so the forces two moving charges exert on each other are generally not balanced moment by moment. Momentum is still conserved, because the electromagnetic field itself carries momentum: particle momentum plus field momentum is what balances. This is the conceptual doorway from particle mechanics to field theory — the continuum articles this chapter leads into exist precisely because forces at a distance must be replaced by locally interacting fields.