Four-Velocity and Four-Acceleration

Ask a Newtonian physicist for a particle's velocity and they will hand you $\mathrm{d}\mathbf{x}/\mathrm{d}t$ without hesitation. Ask a relativist the same question and they will first ask whose $t$ . Coordinate time is not a property of the particle — it belongs to whichever frame happens to be watching, and time dilation together with the relativity of simultaneity guarantee that two frames disagree about it. Differentiating position by a quantity everyone disagrees on cannot produce a clean geometric object, and indeed the ordinary three-velocity fails to transform as part of a four-vector — it mixes through the awkward velocity-addition rule instead of rotating like the components of a genuine spacetime arrow. The repair is disarmingly simple and it sets the entire agenda of relativistic kinematics. Differentiate the spacetime position $x^\mu$ not by coordinate time but by the proper time $\tau$ ticked off by a clock carried along the worldline itself. Because proper time is built from the invariant interval — this article consumes $c^2\,\mathrm{d}\tau^2 = -\mathrm{d}s^2$ from aq-gr-proper-time and aq-gr-spacetime-interval — every observer agrees on it, and the derivative $u^\mu = \mathrm{d}x^\mu/\mathrm{d}\tau$ comes out as a bona-fide four-vector. That single choice forces two remarkable facts: the four-velocity has a fixed Minkowski length, $u\cdot u = -c^2$ , so a particle is always "moving through spacetime at $c$ "; and its proper-time derivative, the four-acceleration $a^\mu$ , is always perpendicular to it, so it is spacelike and its magnitude is exactly what an accelerometer reads. From these two objects hang the four-momentum, the relativistic $f=ma$ , and — once the ordinary derivative becomes a covariant one — the geodesic equation that turns gravity into geometry.

Key takeaways

The four-velocity $u^\mu = \mathrm{d}x^\mu/\mathrm{d}\tau$ is the tangent to a worldline parametrized by the invariant proper time; differentiating by $\tau$ rather than coordinate time $t$ is what makes it a true four-vector.
In an inertial frame $u^\mu = \gamma(c,\mathbf{v})$ , where the same Lorentz factor $\gamma = \mathrm{d}t/\mathrm{d}\tau$ that governs time dilation appears as the time component.
For a massive particle the four-velocity has fixed normalization $u^\mu u_\mu = -c^2$ (signature $-+++$ ), a single constraint that leaves exactly three independent components — the ordinary velocity.
The four-acceleration $a^\mu = \mathrm{d}u^\mu/\mathrm{d}\tau$ is always Minkowski-orthogonal to $u^\mu$ : $a\cdot u = 0$ , so it is spacelike, and its invariant magnitude $\sqrt{a^\mu a_\mu}$ is the proper acceleration an onboard accelerometer measures.
Constant proper acceleration produces hyperbolic (Rindler) motion — a worldline asymptotic to the light cone whose traveller approaches but never reaches $c$ .
Free fall is exactly $a^\mu = 0$ ; promoting $\mathrm{d}/\mathrm{d}\tau$ to a covariant derivative turns this into the geodesic equation, the gateway to curved spacetime.
Key idea: differentiate the spacetime position by the invariant proper time, not coordinate time. The four-velocity is the unit timelike tangent ( $u\cdot u = -c^2$ ); the four-acceleration is its proper-time derivative and is always orthogonal to it.

The core insight

Differentiate the spacetime position with respect to the invariant proper time, not coordinate time. The four-velocity is the unit timelike tangent ( $u\cdot u = -c^2$ ); the four-acceleration is its proper-time derivative and is always orthogonal to it.

Where this lives today

The kinematics on this page is the daily arithmetic of any laboratory that pushes matter near the speed of light. At the Large Hadron Collider, protons carrying $\gamma \approx 7.5\times10^3$ are bent around dipole magnets of radius $\rho \approx 2.8\ \mathrm{km}$ ; their proper acceleration $\alpha = \gamma^2 c^2/\rho \approx 1.8\times10^{21}\ \mathrm{m/s^2}$ is the largest steady acceleration humans routinely engineer — roughly $10^{20}$ times Earth's gravity — and the resulting synchrotron radiation is set entirely by the four-acceleration's invariant magnitude. The same object dilates the proper time stamped onto unstable particles — a muon's lifetime scales with $u^0/c = \gamma$ , which is why cosmic-ray muons reach the ground. And the constant-proper-acceleration worldline is more than rocketry fantasy — it is the local geometry of a static observer hovering outside a black hole, whose Rindler horizon is the flat-space rehearsal for the Unruh effect and Hawking radiation. Every thread begins with the same decision made here: divide the worldline by its own proper time.

The physical picture

Picture a stopwatch glued to the particle. No matter how the particle darts about, that clock reads the one time everybody agrees it "experiences" — its proper time $\tau$ . If you build velocity by advancing the spacetime position one tick of this clock at a time, the result cannot depend on who is watching, because the ticks themselves do not. That is the whole trick: the naive $\mathrm{d}\mathbf{x}/\mathrm{d}t$ divides by a frame-dependent time, while $\mathrm{d}x^\mu/\mathrm{d}\tau$ divides by an invariant and so glides through Lorentz transformations as a rigid four-dimensional arrow. There is a beautiful constraint hidden here. Every massive particle carries a fixed "spacetime speed" equal to $c$ : at rest it spends its entire budget flowing forward in time, and as it picks up spatial speed, part of that budget merely tilts into space — the total never changes. Geometrically the tip of the four-velocity arrow is confined to a hyperbola, the mass shell $u\cdot u = -c^2$ , and acceleration can only slide it along that hyperbola. That is why the four-acceleration must be orthogonal to the four-velocity: a push that lengthened $u$ would knock the particle off its own mass shell.

Figure

The mass shell of four-velocities: every massive particle sits on the hyperbola u⁰² − u¹² = c² with c drawn as 90 px. At rest the arrow points straight up the time axis, u = (c, 0); at v = 0.6c it tilts to u = γ(c, v) = (1.25c, 0.75c) — yet its Minkowski length stays pinned at u·u = −c²

The four-acceleration is where the felt physics lives. Coordinate acceleration $\mathrm{d}^2\mathbf{x}/\mathrm{d}t^2$ is frame-dependent bookkeeping — it can even be nonzero for a freely-floating observer who merely uses curved coordinates. The four-acceleration's magnitude, by contrast, is invariant: it is precisely the number an accelerometer bolted to the particle displays. Push a rocket with steady thrust and its four-acceleration magnitude stays constant even as its coordinate acceleration fades toward zero; the worldline that results is a hyperbola hugging the light cone, the relativistic replacement for the Newtonian parabola of "uniform acceleration."

Figure

A Rindler worldline: constant proper acceleration α traces the hyperbola x² − (ct)² = (c²/α)², asymptotic to the light cone. The four-velocity u is tangent to it and the four-acceleration a is Minkowski-orthogonal, a·u = 0. For a rocket holding a comfortable 1g the scale is c²/α = 0.97 light-year

The core equations

The starting point is a definition, not a theorem. We build velocity the way Newton did — position divided by time — but we insist the time be the invariant proper time carried along the worldline, so the numerator and denominator both transform cleanly and the quotient is a four-vector.

u^\mu \equiv \frac{\mathrm{d}x^\mu}{\mathrm{d}\tau} \qquad (1)

where: $u^\mu$ — four-velocity, index $\mu = 0,1,2,3$ ; $x^\mu = (ct,\mathbf{x})$ — spacetime position along the worldline; $\tau$ — proper time, the invariant clock ticked off along the curve. Physical reading: $u^\mu$ is the unit tangent to the worldline, pointing "forward" in the particle's own time. To see what $u^\mu$ looks like in a laboratory frame, split the proper-time derivative with the chain rule and feed in the time-dilation relation $\mathrm{d}t/\mathrm{d}\tau = \gamma$ . The spatial part becomes the ordinary velocity scaled by $\gamma$ , and the time part becomes $\gamma c$ .

u^\mu = \gamma\,(c,\,\mathbf{v}), \qquad \gamma = \frac{1}{\sqrt{1 - v^2/c^2}} \qquad (2)

where: $\gamma$ — Lorentz factor, and crucially $\gamma = \mathrm{d}t/\mathrm{d}\tau$ ; $\mathbf{v} = \mathrm{d}\mathbf{x}/\mathrm{d}t$ — the coordinate three-velocity; $c$ — speed of light. Physical reading: the factor that stretches time also sets the four-velocity's time component — kinematics and dilation are one object. The defining structural fact follows by contracting $u^\mu$ with itself using the metric. It is a fixed number — the same in every frame — which is why it deserves to be called a normalization rather than a formula.

u^\mu u_\mu = \eta_{\mu\nu}u^\mu u^\nu = -c^2 \qquad (3)

where: $u_\mu = \eta_{\mu\nu}u^\nu$ — covariant components; $\eta_{\mu\nu} = \mathrm{diag}(-1,+1,+1,+1)$ — Minkowski metric. Physical reading: the four-velocity is timelike with fixed length $c$ ; this one constraint knocks the four components down to the three independent numbers of ordinary velocity. Differentiating once more — again by proper time — defines the four-acceleration. It is the honest second derivative of position along the worldline, the relativistic generalization of $\mathrm{d}^2\mathbf{x}/\mathrm{d}t^2$ .

a^\mu \equiv \frac{\mathrm{d}u^\mu}{\mathrm{d}\tau} = \frac{\mathrm{d}^2 x^\mu}{\mathrm{d}\tau^2} \qquad (4)

where: $a^\mu$ — four-acceleration. Physical reading: whatever bends the worldline away from a straight line shows up here; a straight worldline has $a^\mu = 0$ . Because the length of $u^\mu$ can never change, its derivative can only turn it. Differentiating the normalization (3) makes this exact: the four-acceleration is Minkowski-perpendicular to the four-velocity.

a^\mu u_\mu = 0 \qquad (5)

where: $a^\mu u_\mu$ — Minkowski inner product of the two vectors. Physical reading: since $u^\mu$ is timelike and $a^\mu$ is orthogonal to it, $a^\mu$ must be spacelike (or zero) — there is no such thing as a timelike four-acceleration for a massive particle. The invariant magnitude of that spacelike vector is the quantity with operational teeth: it is the reading of an accelerometer travelling with the particle, independent of any frame.

\alpha \equiv \sqrt{a^\mu a_\mu} \qquad (6)

where: $\alpha$ — proper acceleration (the felt acceleration). Physical reading: unlike coordinate acceleration, $\alpha$ is what you weigh on a bathroom scale; free fall gives $\alpha = 0$ , standing on Earth gives $\alpha \approx 9.81\ \mathrm{m/s^2}$ . Finally, dynamics enters by attaching mass. Multiplying $u^\mu$ by rest mass gives the four-momentum, and Newton's second law becomes a four-vector equation; setting the four-force to zero and passing to curved spacetime yields the geodesic equation.

f^\mu = m\,a^\mu, \qquad \frac{\mathrm{d}^2 x^\mu}{\mathrm{d}\tau^2} + \Gamma^\mu_{\;\nu\rho}\,\frac{\mathrm{d}x^\nu}{\mathrm{d}\tau}\frac{\mathrm{d}x^\rho}{\mathrm{d}\tau} = 0 \qquad (7)

where: $f^\mu$ — four-force; $m$ — rest (invariant) mass; $\Gamma^\mu_{\;\nu\rho}$ — Christoffel symbols encoding the connection (gravity). Physical reading: free particles ( $f^\mu = 0$ ) move so their four-acceleration vanishes; in curved spacetime the extra $\Gamma$ term is gravity.

Derivation AFour-velocity is a unit timelike vector tangent to the worldline

The goal is to prove that the normalization $u\cdot u = -c^2$ is not a lucky feature of one frame's components but a direct, frame-independent consequence of how proper time is defined from the invariant interval. If we can get it straight from the definition of $\tau$ , then it holds in curved spacetime too, and it is the reason the four-velocity carries only three independent numbers.

Step 1. Along any timelike worldline, proper time is defined by $c^2\,\mathrm{d}\tau^2 = -\mathrm{d}s^2 = -\eta_{\mu\nu}\,\mathrm{d}x^\mu\,\mathrm{d}x^\nu$ , the relation consumed from aq-gr-proper-time. In signature $-+++$ the interval $\mathrm{d}s^2$ is negative along timelike separations, so the minus sign makes the right-hand side positive and $\mathrm{d}\tau$ real. Why: this is the only invariant "time" available, so anything built from it will automatically be frame-independent.

Step 2. Divide both sides by $\mathrm{d}\tau^2$ , treating $\tau$ as the curve parameter: $c^2 = -\eta_{\mu\nu}\,(\mathrm{d}x^\mu/\mathrm{d}\tau)(\mathrm{d}x^\nu/\mathrm{d}\tau)$ . Nothing has been assumed about the frame; we have only used that $\tau$ parametrizes the same curve. Why: dividing by the invariant $\mathrm{d}\tau^2$ converts a statement about intervals into a statement about the tangent vector.

Step 3. Recognize $\mathrm{d}x^\mu/\mathrm{d}\tau = u^\mu$ by definition (1), so $c^2 = -\eta_{\mu\nu}u^\mu u^\nu = -u^\mu u_\mu$ , i.e. $u^\mu u_\mu = -c^2$ . Because both $\tau$ and $\mathrm{d}s^2$ are invariants, this holds in every frame and on any spacetime. Why: the fixed length is what reduces four components to three and forbids the four-velocity from ever being null or zero for a massive particle.

Step 4. Cross-check in components: substituting $u^\mu = \gamma(c,\mathbf{v})$ gives $-\gamma^2 c^2 + \gamma^2 v^2 = -\gamma^2 c^2(1 - v^2/c^2) = -c^2$ , using $\gamma^2(1 - v^2/c^2) = 1$ . Why: the abstract and the component routes must agree, and they do — a reassuring closure of the argument. *Verified: $u\cdot u = -c^2$ both from the interval and from $\gamma(c,\mathbf{v})$ ; consistent with MTW §2.4 and Schutz §2.2.*

Go deeper · Derivation B · 1 minThe four-acceleration is orthogonal to the four-velocity and purely spacelike

Derivation BThe four-acceleration is orthogonal to the four-velocity and purely spacelike

The goal is to show, with one line of calculus, that acceleration can only rotate the four-velocity and never stretch it — the geometric content of $a\cdot u = 0$ — and to conclude that the four-acceleration is therefore spacelike, so its magnitude is a real, measurable proper acceleration. This is the fact that separates the frame-independent felt acceleration from the frame-dependent coordinate acceleration.

Step 1. Take the proper-time derivative of the constant $u^\mu u_\mu = -c^2$ . Since the right-hand side is a constant, $\mathrm{d}(u^\mu u_\mu)/\mathrm{d}\tau = 0$ . Why: a quantity that never changes has zero rate of change — this is the entire engine of the proof.

Step 2. Expand the left side with the product rule. In flat spacetime $\eta_{\mu\nu}$ is constant, so $\mathrm{d}(u^\mu u_\mu)/\mathrm{d}\tau = 2\,u_\mu\,(\mathrm{d}u^\mu/\mathrm{d}\tau) = 2\,a^\mu u_\mu$ . Why: the metric passing through the derivative untouched is exactly what fails in curved spacetime, where $\mathrm{d}/\mathrm{d}\tau$ must be promoted to a covariant derivative — but the orthogonality result survives.

Step 3. Combine: $2\,a^\mu u_\mu = 0 \Rightarrow a^\mu u_\mu = 0$ . The four-acceleration is Minkowski-orthogonal to the four-velocity for all time. Why: this is the precise statement that a push cannot change the "spacetime speed" $c$ , only its direction on the mass shell.

Step 4. Conclude spacelike character. Any nonzero vector orthogonal to a timelike vector is spacelike, so $a^\mu a_\mu \ge 0$ , with equality only when $a^\mu = 0$ . Hence $\alpha = \sqrt{a^\mu a_\mu}$ is real and non-negative. Why: only a spacelike vector has a real magnitude that an accelerometer can register; a hypothetical timelike four-acceleration would be physically meaningless here. *Verified: differentiating $u\cdot u=-c^2$ gives $a\cdot u=0$ ; orthogonality to a timelike vector forces $a^\mu a_\mu\ge0$ . Consistent with Wald §4.2, Carroll §1.3.*

Go deeper · Derivation C · 1 minHyperbolic motion under constant proper acceleration

Derivation CHyperbolic motion under constant proper acceleration

The goal is to solve the relativistic equation of motion for one-dimensional travel at constant proper acceleration and recover the Rindler worldline — the exact analogue of "uniformly accelerated motion," but one that respects the speed limit. This is the trajectory of an idealized rocket burning to give its crew a steady felt weight, and it is the local model of a static observer near a horizon.

Step 1. For motion along $x$ , write $u^0 = c\cosh\phi(\tau)$ , $u^1 = c\sinh\phi(\tau)$ . This ansatz satisfies $u\cdot u = -c^2\cosh^2\phi + c^2\sinh^2\phi = -c^2$ identically. Why: rapidity $\phi$ is the natural "angle" of a Lorentz boost, and using it means the mass-shell constraint is built in rather than imposed by hand.

Step 2. Differentiate to get the four-acceleration: $a^\mu = \dot\phi\,(c\sinh\phi,\,c\cosh\phi,\,0,\,0)$ , where $\dot\phi = \mathrm{d}\phi/\mathrm{d}\tau$ . Its Minkowski square is $a^\mu a_\mu = \dot\phi^2(-c^2\sinh^2\phi + c^2\cosh^2\phi) = c^2\dot\phi^2$ , using $\cosh^2\phi - \sinh^2\phi = 1$ . Why: because $a$ is orthogonal to $u$ , its length is captured entirely by the rate of change of the rapidity.

Step 3. Impose constant proper acceleration $\sqrt{a^\mu a_\mu} = \alpha$ : then $c\,\dot\phi = \alpha$ , so $\mathrm{d}\phi/\mathrm{d}\tau = \alpha/c$ and, with $\phi(0)=0$ , $\phi(\tau) = \alpha\tau/c$ . Why: constant felt acceleration means the rapidity grows linearly in proper time — the relativistic replacement for "velocity grows linearly in time."

Step 4. Read off the velocity: $v/c = u^1/u^0 = \tanh\phi = \tanh(\alpha\tau/c)$ , which climbs toward $1$ but never reaches it. Integrate $u^0 = c\,\mathrm{d}t/\mathrm{d}\tau$ and $u^1 = c\,\mathrm{d}x/\mathrm{d}\tau$ to get $ct(\tau) = (c^2/\alpha)\sinh(\alpha\tau/c)$ and $x(\tau) = (c^2/\alpha)\cosh(\alpha\tau/c)$ . Why: the traveller's own clock $\tau$ runs far slower than lab time $t$ at high rapidity — the physical content of the twin-paradox asymmetry for a powered worldline.

Step 5. Eliminate $\tau$ using $\cosh^2 - \sinh^2 = 1$ : $x^2 - (ct)^2 = (c^2/\alpha)^2$ , a hyperbola asymptotic to the light cone $x = \pm ct$ . Why: this is the Rindler worldline — the invariant geometric object behind "constant proper acceleration," drawn in the second figure above. *Verified: $v = c\tanh(\alpha\tau/c) < c$ always; the worldline satisfies $x^2 - c^2 t^2 = (c^2/\alpha)^2$ . Standard Rindler result, MTW §6.1, Carroll §1.3.*

Limiting cases

Case	Result	Meaning
Rest frame / low speed	$u^\mu \to (c,\mathbf{v})$ , $\gamma \to 1$	As $v \ll c$ , $\gamma \approx 1 + \tfrac12 v^2/c^2$ and $u^i = \gamma v^i \to v^i$ recovers the Newtonian three-velocity; proper and coordinate time coincide, $\tau \to t$ .
Ultrarelativistic	$\gamma \to \infty$ , $u^\mu \approx \gamma(c,\mathbf{v})$	Components diverge as $v \to c$ ; the four-velocity tends toward a null direction but a massive particle never gets there.
Massless particle	$u^\mu$ undefined	On a null worldline $\mathrm{d}\tau = 0$ , so $\mathrm{d}x^\mu/\mathrm{d}\tau$ is undefined; one parametrizes by an affine $\lambda$ and uses the null tangent $\mathrm{d}x^\mu/\mathrm{d}\lambda$ instead.
Free fall / inertial	$a^\mu = 0$	Zero four-acceleration defines geodesic motion — straight lines in flat spacetime, the geodesic equation in curved spacetime.
Constant proper acceleration	$\sqrt{a^\mu a_\mu} = \alpha = \text{const}$	Hyperbolic (Rindler) motion asymptotic to the light cone; steady felt "gravity" $\alpha$ , yet $v \to c$ only in the infinite-time limit.

Common trapCoordinate acceleration and proper acceleration are the same thing

They are not. Coordinate acceleration $\mathrm{d}^2\mathbf{x}/\mathrm{d}t^2$ is frame-dependent and can even be nonzero for a freely-floating observer who simply uses curvilinear coordinates — it is bookkeeping, not physics. The four-acceleration magnitude $\alpha = \sqrt{a^\mu a_\mu}$ is invariant and is what an accelerometer actually reads. A particle circling at constant speed has $\mathrm{d}|\mathbf{v}|/\mathrm{d}t = 0$ yet a large, frame-independent proper acceleration, because the direction of $\mathbf{v}$ is turning. And the four-velocity itself is never "sometimes zero" — for a massive particle it is pinned to length $c$ in every frame, never null and never vanishing.

Worked examples

1 · Four-velocity, normalization, and four-momentum at v = 0.8c (confidence builder)

A particle moves along $x$ at $v = 0.8c$ . Write its four-velocity components, verify the normalization explicitly, and form its four-momentum for rest mass $m$ .

Step 1. Compute the Lorentz factor: $\gamma = 1/\sqrt{1 - 0.8^2} = 1/\sqrt{0.36} = 1/0.6 = \mathbf{1.667}$ . This is the same $\gamma$ that would dilate the particle's clock.

Step 2. Assemble the four-velocity from equation (2): $u^\mu = \gamma(c, v, 0, 0) = (1.667c,\ 1.333c,\ 0,\ 0)$ , since $\gamma v = 1.667 \times 0.8c = 1.333c$ . The time component exceeds $c$ — that is normal; only the invariant length is capped.

Step 3. Check the normalization (3) by hand: $u^\mu u_\mu = -(1.667c)^2 + (1.333c)^2 = (-2.778 + 1.778)c^2 = \mathbf{-1.000\,c^2}$ . The constraint $u\cdot u = -c^2$ holds exactly, as it must in every frame.

Step 4. Form the four-momentum $p^\mu = m u^\mu = (1.667\,mc,\ 1.333\,mc,\ 0,\ 0)$ . The time component is $E/c = \gamma m c$ , so the energy is $E = 1.667\,mc^2$ ; the spatial part is the relativistic momentum $p = 1.333\,mc$ . Its invariant norm is $p^\mu p_\mu = m^2(u\cdot u) = -m^2 c^2$ , the mass-shell relation $E^2 = (pc)^2 + (mc^2)^2$ . u = (1.667c, 1.333c, 0, 0), u·u = −c² exactly, E = 1.667 mc².

2 · A relativistic rocket holding 1g (fully numeric)

A rocket maintains constant proper acceleration $\alpha = g = 9.81\ \mathrm{m/s^2}$ , giving its crew Earth-like weight. Using the hyperbolic-motion solution, find the characteristic distance scale, and after $\tau = 1$ year of ship time find the speed, the elapsed lab time, and the distance covered.

Step 1. The natural length and time scale is $c^2/\alpha = c \cdot (c/\alpha)$ . First $c/\alpha = (2.998\times10^8)/9.81 = 3.056\times10^7\ \mathrm{s} = \mathbf{0.968\ yr}$ , so $c^2/\alpha = 0.968$ light-year. This is the "radius" of the Rindler hyperbola — the scale marked on the figure.

Step 2. Form the dimensionless rapidity after one year of proper time: $\phi = \alpha\tau/c = \tau/(c/\alpha) = 1/0.968 = \mathbf{1.033}$ .

Step 3. Speed from Step 4 of the derivation: $v/c = \tanh(1.033) = \mathbf{0.775}$ , so $v = 0.775c$ . The Lorentz factor is $\gamma = \cosh(1.033) = 1.58$ — the crew's clocks now run about $1.58\times$ slower than the launch site's.

Step 4. Elapsed lab time: $t = (c/\alpha)\sinh(\alpha\tau/c) = 0.968\ \mathrm{yr} \times \sinh(1.033) = 0.968 \times 1.226 = \mathbf{1.19\ yr}$ . Distance covered: $x = (c^2/\alpha)[\cosh(\alpha\tau/c) - 1] = 0.968\ \mathrm{ly} \times (1.582 - 1) = \mathbf{0.564\ light\text{-}year}$ . One year of ship time at 1g already carries the crew past half a light-year while only 1.19 years pass at home — and because rapidity grows linearly, a few more ship-years cross galactic distances. After 1 ship-year: v = 0.775c, γ = 1.58, t = 1.19 yr, x = 0.564 ly.

Physics of the portrait

The four-velocity and four-acceleration are less discoveries about particles than about spacetime — and they belong to the man who first drew worldlines through a four-dimensional continuum and insisted that time was a direction, not a backdrop.

Origins · 1 minHermann Minkowski and how this was discovered

Hermann Minkowski

1864–1909 · the geometry of spacetime and the four-vector

Hermann Minkowski was Einstein's former mathematics teacher at the Zürich Polytechnic — he once called the young Einstein a "lazy dog" who "never bothered about mathematics at all." The irony is historic: it was Minkowski who, in his 1908 Cologne address Raum und Zeit, gave Einstein's 1905 relativity its enduring geometric form. "Henceforth space by itself, and time by itself, are doomed to fade away into mere shadows," he declared, "and only a kind of union of the two will preserve an independent reality." In that union a moving particle is a worldline, and the natural way to describe its motion is to differentiate its four-position by the invariant proper time — producing the four-velocity as the unit tangent and the four-acceleration as its bending. By recasting the interval $-c^2\,\mathrm{d}\tau^2 = \mathrm{d}s^2$ as the metric of a genuine geometry, in which Lorentz transformations are just "rotations," he made the normalization $u\cdot u = -c^2$ a statement about lengths rather than an algebraic accident. He did not live to see how far it would run — he died of a ruptured appendix in January 1909, months after the lecture, at forty-four. Within a decade his four-dimensional language was the scaffolding on which Einstein built general relativity, where the same four-acceleration, set to zero, becomes the law of free fall.

History

1905

— Einstein's special relativity fuses space and time and fixes the invariant speed c

1908

— Minkowski's Raum und Zeit introduces spacetime, worldlines, and the four-dimensional velocity

1909

— Max Born analyzes rigid uniformly accelerated (hyperbolic) motion, the Rindler worldline

1911

— Sommerfeld and von Laue systematize four-vector relativistic mechanics and the four-force

1916

— Einstein's general relativity makes vanishing four-acceleration the geodesic law of free fall

Going further

References. Misner, Thorne & Wheeler, Gravitation, §1.5–1.7, §2.4, §6.1 · Wald, General Relativity, §4.2 · Carroll, Spacetime and Geometry, §1.3 · Hartle, Gravity, §5.1–5.3 · Schutz, A First Course in General Relativity, §2.2–2.4.

Provenance. The normalization $u\cdot u = -c^2$ is derived from the proper-time relation $c^2\,\mathrm{d}\tau^2 = -\mathrm{d}s^2$ consumed from aq-gr-proper-time and aq-gr-spacetime-interval; the components $u^\mu = \gamma(c,\mathbf{v})$ use $\gamma = \mathrm{d}t/\mathrm{d}\tau$ recalled from aq-gr-time-dilation. A verifier should confirm: (1) $u\cdot u = -c^2$ by substituting $u^\mu = \gamma(c,\mathbf{v})$ with $\gamma^{-2} = 1 - v^2/c^2$ ; (2) $a\cdot u = 0$ by differentiating the normalization; (3) the Rindler worldline satisfies $x^2 - c^2 t^2 = (c^2/\alpha)^2$ with $v = c\tanh(\alpha\tau/c)$ ; (4) the 1g figures $c/\alpha = 0.968\ \mathrm{yr}$ , $v = 0.775c$ , $t = 1.19\ \mathrm{yr}$ , $x = 0.564\ \mathrm{ly}$ after one ship-year; (5) the LHC estimate $\alpha = \gamma^2 c^2/\rho \approx 1.8\times10^{21}\ \mathrm{m/s^2}$ . Signature is $(-,+,+,+)$ throughout; $g \approx 9.81\ \mathrm{m/s^2}$ is Earth's surface gravity.

Prerequisites. Proper Time · Four-Vectors · Spacetime Interval · Minkowski Metric and Signature · Time Dilation. Leads to. Lorentz Boosts and Rapidity · Causal Structure and Light Cones.

Ask the tutor: Want to see how the four-velocity's fixed normalization becomes the mass-shell $E^2 = (pc)^2 + (mc^2)^2$ , or how zero four-acceleration turns into the geodesic equation in a real metric? Ask and we'll work it through.

Why divide by proper time τ instead of coordinate time t?

Coordinate time belongs to a frame, not to the particle: different inertial observers assign different $t$ to the same events through time dilation and the relativity of simultaneity. If you build velocity as $\mathrm{d}\mathbf{x}/\mathrm{d}t$ , you are dividing an invariant displacement by a frame-dependent number, and the result transforms in the messy way encoded by the velocity-addition law — it is not part of any four-vector. Proper time $\tau$ , ticked by a clock riding the worldline, is something every observer agrees on. Dividing by it keeps both numerator and denominator clean, so $u^\mu = \mathrm{d}x^\mu/\mathrm{d}\tau$ rotates rigidly under Lorentz transformations — the source of everything else on this page.

Why does the four-velocity have only three independent components?

Because the normalization $u^\mu u_\mu = -c^2$ is one algebraic equation relating the four components. Given the three spatial components $u^i = \gamma v^i$ — which is just the ordinary velocity dressed with $\gamma$ — the time component is fixed at $u^0 = \gamma c = \sqrt{c^2 + u^i u_i}$ . So although $u^\mu$ has four entries, only three carry independent information, exactly matching the three numbers of Newtonian velocity. This is the relativistic version of a familiar idea: a unit vector in three dimensions has only two free angles because its length is constrained. Here the "unit" length is $c$ and the constraint surface is the mass-shell hyperboloid.

How can the time component u⁰ = γc exceed c without breaking the speed limit?

The speed limit caps the coordinate velocity $\mathbf{v} = \mathrm{d}\mathbf{x}/\mathrm{d}t$ , which stays below $c$ for any massive particle. The four-velocity is a different object — $\mathrm{d}x^\mu/\mathrm{d}\tau$ , measured per unit proper time — and its components can be arbitrarily large because $\tau$ runs slower than $t$ by the factor $\gamma$ . What is bounded is not the components but the invariant length, fixed at $u\cdot u = -c^2$ : as $v \to c$ both $\gamma c$ and $\gamma v$ diverge, yet their Minkowski combination stays exactly $-c^2$ . Big components, fixed length — no contradiction.

What does an accelerometer actually measure — coordinate or proper acceleration?

Proper acceleration, $\alpha = \sqrt{a^\mu a_\mu}$ , the invariant magnitude of the four-acceleration. An accelerometer is blind to how you have chosen coordinates; it responds only to the physical push that bends its worldline away from free fall. That is why an object in orbit or in free fall reads exactly zero — its four-acceleration vanishes and it follows a geodesic — while a scale under your feet reads $g \approx 9.81\ \mathrm{m/s^2}$ because the floor is pushing you off the geodesic you would otherwise follow. Coordinate acceleration $\mathrm{d}^2\mathbf{x}/\mathrm{d}t^2$ is the wrong quantity: it can be nonzero for a freely-floating observer using curved coordinates, and zero for a genuinely accelerating one at a turning point.

Why must the four-acceleration be perpendicular to the four-velocity?

Because the four-velocity has a fixed length, $u\cdot u = -c^2$ , and a derivative can only change what is allowed to vary. Differentiating the constant gives $2\,a^\mu u_\mu = 0$ , so $a\cdot u = 0$ identically. Geometrically the tip of $u^\mu$ is confined to the mass-shell hyperboloid; any change to $u^\mu$ must run tangent to that surface, which is precisely the orthogonality condition. Since $u^\mu$ is timelike and $a^\mu$ is orthogonal to a timelike vector, $a^\mu$ is necessarily spacelike — that is why its magnitude is a real, measurable number and never a "timelike" quantity. Acceleration turns the four-velocity; it can never stretch it.

Is the geodesic equation really just "zero four-acceleration"?

Yes, with one upgrade. In flat spacetime a free particle has $a^\mu = \mathrm{d}u^\mu/\mathrm{d}\tau = 0$ , which integrates to straight-line motion. In curved spacetime the ordinary derivative is not a tensor — moving a vector along a curve, you must account for how the coordinate basis itself twists, the job of the Christoffel symbols. Replacing $\mathrm{d}/\mathrm{d}\tau$ with the covariant derivative gives $\mathrm{d}^2x^\mu/\mathrm{d}\tau^2 + \Gamma^\mu_{\;\nu\rho}\,u^\nu u^\rho = 0$ . The physical statement is unchanged — "no four-acceleration, no force" — but the $\Gamma$ term now encodes gravity, so free fall looks curved in coordinates while remaining perfectly "straight" in spacetime.