Lorentz Boosts and Rapidity

A Lorentz boost is the answer special relativity gives to the oldest question in kinematics: two observers in uniform relative motion look at the same event — how are their coordinates related? Galileo's answer, $x' = x - vt$ with a universal clock, held for three centuries and is wrong. The correct answer is forced by two facts: no inertial frame is privileged, and every observer measures the same speed of light $c$ . Between them these postulates mix the time coordinate $ct$ with the spatial coordinate along the motion and leave the transverse directions alone, in such a way that the spacetime interval $-(c\,\Delta t)^2 + \Delta x^2$ comes out the same for everyone. That single invariant is the seed of the entire structure. Written in the everyday variables $\beta = v/c$ and $\gamma = (1-\beta^2)^{-1/2}$ , the boost is correct but clumsy: collinear boosts refuse to add their velocities, obeying instead $(\beta_1+\beta_2)/(1+\beta_1\beta_2)$ , and the fact that a boost is a kind of rotation stays hidden. Defining the rapidity $\phi$ through $\beta = \tanh\phi$ dissolves both problems: $\gamma = \cosh\phi$ , $\gamma\beta = \sinh\phi$ , the boost matrix is built from $\cosh\phi$ and $\sinh\phi$ just as a rotation is built from $\cos\theta$ and $\sin\theta$ , and collinear boosts simply add their rapidities. The velocity ceiling at $c$ becomes the statement that $\tanh\phi \to 1$ as $\phi \to \infty$ : you can always add more rapidity, but never reach the speed of light. This is the boost sector of the Lorentz group — the local symmetry that survives in the tangent space of every event in curved spacetime, and the first step toward general relativity.

Key takeaways

A Lorentz boost relates two inertial frames in uniform relative motion along one axis, mixing $ct$ with $x$ while leaving $y$ and $z$ fixed, and preserving the interval $-(ct)^2+x^2$ .
In everyday variables the boost uses $\beta=v/c$ and $\gamma=1/\sqrt{1-\beta^2}$ ; it reduces to the Galilean shift when $\beta\ll1$ and diverges as $\beta\to1$ .
The rapidity $\phi$ , defined by $\beta=\tanh\phi$ , recasts the boost as a hyperbolic rotation in the $t$ – $x$ plane, the Lorentzian twin of an ordinary rotation.
Rapidities add for collinear boosts, $\phi_{\text{tot}}=\phi_1+\phi_2$ , which reproduces relativistic velocity addition and explains why $c$ is an unreachable limit.
A boost is not a rotation: rapidity is unbounded, the matrix is symmetric not orthogonal, and non-collinear boosts pick up an extra Thomas–Wigner rotation.
Key idea: a boost is a hyperbolic rotation of spacetime — parametrize it by rapidity $\phi$ with $\tanh\phi=v/c$ , and the messy velocity-addition rule collapses into ordinary addition $\phi_1+\phi_2$ .

The core insight

A boost is a hyperbolic rotation of spacetime: parametrize it by rapidity $\phi$ with $\tanh\phi = v/c$ , and the awkward velocity-addition rule collapses into ordinary addition $\phi_1+\phi_2$ , exposing the boost as the Lorentzian cousin of an ordinary rotation.

Where this lives today

Rapidity is not a historical curiosity; it is the working coordinate of modern high-energy physics. At the Large Hadron Collider the products of a proton–proton collision are catalogued by their rapidity along the beam axis, precisely because a boost from the laboratory frame to the collision frame merely shifts every particle's rapidity by the same constant. Differences in rapidity are therefore boost-invariant, and the flat "rapidity plateau" of produced particles is one of the cleanest signatures a detector records; experimentalists use the closely related pseudorapidity $\eta = -\ln\tan(\theta/2)$ to map where particles land. The same additivity underlies the hyperbolic worldlines of uniformly accelerated observers, which reappear in the Unruh effect, where an accelerating detector perceives a thermal bath set by its proper acceleration. And in general relativity the boost is the local symmetry that never dies: at every event the tangent space is Minkowskian, and the freedom to boost and rotate a local inertial frame is the gauge freedom of the tetrad formalism, through which spinor fields — electrons, quarks, neutrinos — couple to curved spacetime. Every covariant derivative with a spin connection has the rapidity of a local Lorentz boost quietly at work.

The physical picture

Draw spacetime on paper with $ct$ pointing up and $x$ across. An ordinary rotation swings the two axes toward each other by the same angle in opposite senses, keeping them perpendicular and sliding points around a circle of fixed radius $\sqrt{x^2+y^2}$ . A Lorentz boost looks similar but obeys a different rule: it scissors the $ct$ -axis and the $x$ -axis toward the $45°$ light line by the same rapidity, sliding events not around circles but along hyperbolae of fixed interval $x^2-(ct)^2$ . The faster the motion, the harder the axes squeeze toward the diagonal — but they never cross it, because reaching the light line would take infinite rapidity. The light cone is the one direction every boost leaves fixed, the geometric restatement of "the speed of light is the same for everyone."

Figure

A Lorentz boost as a hyperbolic rotation. Boosting by φ = 0.693 (β = 0.6) scissors the ct and x axes toward the 45° light line by 30.96°; the event on the x-axis slides along the invariant hyperbola x² − (ct)² = 1 to its image, which stays on the new x′-axis

Rapidity is the natural angle of spacetime. You would never describe a spinning flywheel by its rim speed when it is the angular position that accumulates cleanly; likewise you should describe boosts by rapidity, which simply piles up, rather than velocity, which saturates. The velocity $v = c\tanh\phi$ is merely the projection of an ever-growing rapidity onto the bounded window $(-c, c)$ . The cosmic speed limit is therefore not a wall but a horizon in rapidity space: each extra unit of rapidity costs the same effort yet buys less velocity, because $\tanh\phi$ flattens toward $1$ . The second figure shows exactly this, with the point $\phi = 1.0986$ mapping to $\beta = 0.8$ marked on the curve.

Figure

Velocity versus rapidity, β = tanh φ. Rapidity runs to infinity along the horizontal axis but velocity presses against the ceiling β = 1 and never reaches it; the marked point shows φ = 1.0986 mapping to β = 0.8, and near the origin the curve is the line β ≈ φ

The core equations

Start from the boost as it first appears in a course, in terms of $\beta = v/c$ and the Lorentz factor $\gamma$ . The transformation is linear, treats the two frames symmetrically, and is fixed uniquely by demanding that a light pulse $x = ct$ maps to $x' = ct'$ in every frame. Only the direction of motion ( $x$ ) and time mix; the transverse coordinates ride along untouched.

ct' = \gamma\,(ct - \beta x),\qquad x' = \gamma\,(x - \beta\, ct),\qquad y' = y,\quad z' = z \qquad (1)

where: $\beta = v/c$ — relative speed in units of $c$ ; $\gamma = 1/\sqrt{1-\beta^2}$ — the Lorentz factor, $\ge 1$ always; $(ct,x,y,z)$ — event coordinates in $S$ ; $(ct',x',y',z')$ — the same event in $S'$ , moving at $+v$ along $x$ . Physical reading: the $-\beta x$ in the time equation is the relativity of simultaneity — clocks at different $x$ are reset differently — and the whole thing collapses to $x'=x-vt,\ t'=t$ when $\beta\to0$ . The velocity variables are correct but hide the geometry. Introduce the rapidity $\phi$ as the hyperbolic angle whose $\tanh$ is the velocity ratio. The three relations below are one substitution, tied together by a single hyperbolic identity, converting every $\beta$ and $\gamma$ in the boost into a $\cosh$ or a $\sinh$ .

\beta = \tanh\phi,\qquad \gamma = \cosh\phi,\qquad \gamma\beta = \sinh\phi \qquad (2)

where: $\phi = \operatorname{artanh}(v/c)$ — the rapidity, dimensionless and running over the whole real line; $\cosh,\sinh,\tanh$ — hyperbolic functions obeying $\cosh^2\phi - \sinh^2\phi = 1$ . Physical reading: $\gamma^2 - (\gamma\beta)^2 = \gamma^2(1-\beta^2) = 1$ is precisely that identity, so the assignments are forced, not chosen — rapidity is the additive quantity velocity only pretends to be. Feed (2) into (1) and the boost in the $t$ – $x$ plane becomes a matrix built entirely from $\cosh\phi$ and $\sinh\phi$ . This is the central object of the article: a hyperbolic rotation.

\begin{pmatrix} ct' \\ x' \end{pmatrix} = \begin{pmatrix} \cosh\phi & -\sinh\phi \\ -\sinh\phi & \cosh\phi \end{pmatrix} \begin{pmatrix} ct \\ x \end{pmatrix} \qquad (3)

where: $\Lambda(\phi)$ — the $2\times2$ boost matrix in the $(ct,x)$ plane. Physical reading: it is symmetric (a rotation is antisymmetric off the diagonal) with determinant $\cosh^2\phi - \sinh^2\phi = +1$ , marking it proper and orthochronous. Put it beside an ordinary rotation to see how far the analogy runs: replace $(\cosh,\sinh)$ by $(\cos,\sin)$ and the preserved quantity $-(ct)^2+x^2$ by $x^2+y^2$ , and one becomes the other.

\begin{pmatrix} x' \\ y' \end{pmatrix} = \begin{pmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{pmatrix} \begin{pmatrix} x \\ y \end{pmatrix} \qquad (4)

where: $\theta$ — the Euclidean rotation angle, preserving the circle $x^2+y^2$ . Physical reading: a single sign — the minus in the Minkowski norm versus the plus in the Euclidean one — is the entire difference between trigonometric and hyperbolic functions, and hence between a rotation and a boost. Because rapidities add and $\tanh$ obeys its addition formula, composing two collinear boosts reproduces relativistic velocity addition with no extra work — the practical payoff of the rapidity picture.

\phi_{12} = \phi_1 + \phi_2 \quad\Longrightarrow\quad \beta_{12} = \frac{\beta_1 + \beta_2}{1 + \beta_1\beta_2} \qquad (5)

where: $\beta_1,\beta_2$ — two collinear velocities in units of $c$ ; $\beta_{12}$ — their relativistic composite. Physical reading: adding rapidities is exact addition; the velocity formula is just $\tanh(\phi_1+\phi_2)$ rewritten, and setting $\beta_1=\beta_2=1$ gives $\beta_{12}=1$ , so light stays light in every frame. Everything above is a manifestation of one invariant: a Lorentz transformation is defined by preserving the Minkowski metric, equivalently the spacetime interval, and the boost satisfies this exactly.

\eta_{\mu\nu}\,\Lambda^{\mu}{}_{\alpha}\,\Lambda^{\nu}{}_{\beta} = \eta_{\alpha\beta}\qquad\Longleftrightarrow\qquad -(ct')^2 + x'^2 = -(ct)^2 + x^2 \qquad (6)

where: $\eta_{\mu\nu} = \mathrm{diag}(-1,1,1,1)$ — the Minkowski metric; $\Lambda^{\mu}{}_{\nu}$ — the boost. Physical reading: the sign of $-(ct)^2+x^2$ distinguishes timelike, null, and spacelike separations, so its invariance means no boost can change an interval's causal character. One elegant consequence: the longitudinal Doppler shift of light is an exponential in rapidity, so successive shifts multiply exactly as successive boosts add.

\frac{f_{\text{obs}}}{f_{\text{src}}} = e^{-\phi}\ \ (\text{recession}),\qquad e^{+\phi}\ \ (\text{approach}) \qquad (7)

where: $f_{\text{src}},f_{\text{obs}}$ — emitted and observed frequencies; $\phi$ — rapidity of the source along the line of sight. Physical reading: $e^{-\phi} = \sqrt{(1-\beta)/(1+\beta)}$ is the standard relativistic Doppler factor — one boost, one factor of $e^{\pm\phi}$ .

Derivation ABoost as a hyperbolic rotation preserving the interval

The goal is to prove, rather than assume, that a Lorentz boost must be a hyperbolic rotation. We start from the single physical demand that the transformation preserve the spacetime interval $-(ct)^2+x^2$ for every event, impose that on the most general linear map of the $(ct,x)$ plane, and watch the coefficients organize themselves into $\cosh\phi$ and $\sinh\phi$ — nothing hyperbolic is put in by hand; it is forced out by the constraint, with $\beta=\tanh\phi$ emerging as the speed of the boosted origin.

Step 1. Write the invariance condition that defines a Lorentz transformation in the boost plane, $-(ct')^2 + x'^2 = -(ct)^2 + x^2$ , and require it to hold for every $(ct,x)$ , not just special cases. Why: this one equation is the entire physical content of "the interval is the same for all inertial observers" — everything else is algebra unpacking it.

Step 2. Take the most general linear map, $ct' = A\,ct + B\,x$ and $x' = C\,ct + D\,x$ , substitute into the invariance condition, and match the coefficients of $(ct)^2$ , of $x^2$ , and of the cross term $ct\,x$ separately. Why: linearity follows from the homogeneity of spacetime (no point or direction is special), so four constants $A,B,C,D$ capture the whole freedom.

Step 3. Coefficient matching yields three equations, $A^2 - C^2 = 1$ , $D^2 - B^2 = 1$ , and $AB = CD$ . These are three constraints on four unknowns, leaving a one-parameter family of solutions. Why: a boost has exactly one physical parameter — the relative speed — so a one-parameter family is precisely what we should expect.

Step 4. Parametrize the family with a single number $\phi$ by setting $A = D = \cosh\phi$ and $B = C = -\sinh\phi$ . This satisfies all three constraints at once because $\cosh^2\phi - \sinh^2\phi = 1$ , and the common minus sign selects motion in the $+x$ direction. Why: the constraint $A^2-C^2=1$ is the defining identity of $(\cosh,\sinh)$ , so hyperbolic functions are not a clever guess but the unique solution.

Step 5. Read off the velocity. The spatial origin of the primed frame is $x'=0$ , i.e. $x\cosh\phi = ct\sinh\phi$ , so it moves at $x/t = c\tanh\phi$ , giving $\beta = \tanh\phi$ . Why: this anchors the abstract parameter $\phi$ to a measurable speed and confirms $\gamma=\cosh\phi$ , $\gamma\beta=\sinh\phi$ . The boxed result is $ct' = ct\cosh\phi - x\sinh\phi$ and $x' = x\cosh\phi - ct\sinh\phi$ . As a check, small $\phi$ gives $\cosh\phi\to1$ , $\sinh\phi\to\phi$ , reproducing the Galilean limit $x'\approx x - c\phi\,t$ , $t'\approx t$ with $v = c\phi$ ; and $\det\Lambda = \cosh^2\phi - \sinh^2\phi = 1$ exactly.

Go deeper · Derivation B · 1 minAdditivity of collinear rapidities by matrix composition

Derivation BAdditivity of collinear rapidities by matrix composition

The goal is to show that the relativistic velocity-addition law, usually presented as a formula to memorize, is nothing more than the statement that hyperbolic rotations compose by adding their angles. We multiply two boost matrices with rapidities $\phi_1$ and $\phi_2$ , apply the hyperbolic angle-addition identities to find their product is a single boost of rapidity $\phi_1+\phi_2$ , and translate back through $\beta=\tanh\phi$ to recover the composition rule and the invariance of $c$ for free.

Step 1. Write the two boosts in the same $(ct,x)$ plane, each a matrix $\Lambda(\phi_i)$ with diagonal $\cosh\phi_i$ and off-diagonal $-\sinh\phi_i$ . Because they act in the identical plane, they commute and multiply like ordinary $2\times2$ matrices. Why: commuting is special to the collinear case — it is exactly what fails for non-parallel boosts, which is where the Thomas–Wigner rotation appears.

Step 2. Compute the $(1,1)$ entry of the product $\Lambda(\phi_2)\Lambda(\phi_1)$ : it is $\cosh\phi_2\cosh\phi_1 + \sinh\phi_2\sinh\phi_1$ . The two off-diagonal entries are both $-\sinh$ , so their product enters with a plus sign. Why: that plus sign is the whole trick — it is what turns the product into a $\cosh$ of a sum rather than a difference.

Step 3. Apply the identities $\cosh\phi_1\cosh\phi_2 + \sinh\phi_1\sinh\phi_2 = \cosh(\phi_1+\phi_2)$ and, for the off-diagonal, $\sinh\phi_1\cosh\phi_2 + \cosh\phi_1\sinh\phi_2 = \sinh(\phi_1+\phi_2)$ . The product is therefore $\Lambda(\phi_1+\phi_2)$ . Why: this is the entire claim — two boosts make one boost, and their rapidities simply add.

Step 4. Translate to velocities with $\beta=\tanh\phi$ : $\beta_{12} = \tanh(\phi_1+\phi_2) = (\beta_1+\beta_2)/(1+\beta_1\beta_2)$ . Why: the notorious velocity-addition formula is just the $\tanh$ addition identity in disguise, and putting $\beta_1=\beta_2=1$ returns $\beta_{12}=1$ , proving $c$ is the same in every frame. The boxed result is $\Lambda(\phi_2)\,\Lambda(\phi_1) = \Lambda(\phi_1+\phi_2)$ . Numerical check: for $\beta_1=\beta_2=0.5$ , each $\phi = \operatorname{artanh}0.5 \approx 0.5493$ , the sum is $\approx 1.0986$ , and $\tanh(1.0986) = 0.8$ , matching $(0.5+0.5)/(1+0.25) = 0.8$ exactly.

Go deeper · Derivation C · 1 minTime dilation as the projection of a rapidity

Derivation CTime dilation as the projection of a rapidity

The goal is to recover the time-dilation factor directly from a single boost and to see that $\gamma = \cosh\phi$ is a projection, the temporal analogue of $\cos\theta$ in a rotation. We follow two ticks of a clock at rest in the primed frame, transform the events back to the unprimed frame, and read off the elapsed coordinate time: the clock's own time is the proper time, while the frame that watches it move records a longer interval by the factor $\cosh\phi$ .

Step 1. Fix the clock at rest in $S'$ , so between its two ticks $\Delta x' = 0$ and the elapsed primed time $\Delta t'$ is the proper time it physically measures. Why: "at rest in $S'$ " is the defining condition of proper time — the clock is present at both events, so no simultaneity ambiguity enters.

Step 2. Use the inverse boost (flip the sign of $\beta$ ) to express unprimed coordinates in terms of primed ones: $ct = \gamma(ct' + \beta x')$ . Why: we want the time $S$ assigns to the two ticks, so we transform from the clock's frame to the observer's frame.

Step 3. Impose $\Delta x' = 0$ , which kills the $\beta x'$ term and leaves $c\,\Delta t = \gamma\,c\,\Delta t'$ , that is $\Delta t = \gamma\,\Delta t' = \cosh\phi\,\Delta t'$ . Why: with the spatial separation gone, the entire effect is the single factor $\gamma=\cosh\phi\ge1$ — the moving clock is seen to run slow.

Step 4. Interpret geometrically: the clock's worldline is the $ct'$ -axis, and $\cosh\phi$ is the ratio by which its unit tick projects onto the observer's $ct$ -axis, exactly as $\cos\theta$ projects one rotated axis onto another. Why: this reframes time dilation not as a paradox but as an axis projection in a tilted coordinate system. The boxed result is $\Delta t = \gamma\,\Delta t' = \cosh\phi\,\Delta t'$ . Numerical check: at $\beta = 0.6$ , $\gamma = 1/\sqrt{1-0.36} = 1/0.8 = 1.25$ , and independently $\cosh(\operatorname{artanh}0.6) = \cosh(\ln 2) = (2 + \tfrac12)/2 = 1.25$ — the two routes agree.

Limiting cases

The boost interpolates smoothly between the Newtonian world at $\beta\ll1$ and the light cone at $\beta\to1$ , and the rapidity makes each corner transparent.

Case	Result	Meaning
Non-relativistic $\beta\ll1$	$x'\approx x-vt,\quad t'\approx t$	$\gamma\to1$ and the $\beta x/c$ term is negligible; the boost reduces to the Galilean shift, and $\phi\approx\beta$ so rapidity and velocity-in-units-of- $c$ coincide.
Moderate speed $\beta=0.6$	$\gamma=1.25,\ \phi=\ln2\approx0.693$	The regime where relativistic corrections are a clean textbook $25\%$ ; still far from the light cone.
Ultra-relativistic $\beta\to1$	$\gamma=\cosh\phi\to\tfrac12 e^{\phi}\to\infty$	Rapidity grows without bound while velocity saturates; time dilation and length contraction diverge as $v$ presses against $c$ .
Photon / null direction	$\beta=1\ \Longleftrightarrow\ \phi=\infty$	Light has no rest frame; a null worldline cannot be brought to rest by any finite boost, and the light cone is boost-invariant.
Small-rapidity expansion	$\beta=\tanh\phi=\phi-\tfrac{\phi^3}{3}+\cdots$	For everyday speeds $\beta$ and $\phi$ nearly agree, which is exactly why intuition mistakes the additive quantity (rapidity) for velocity.

Common trapVelocities add, rapidities add — but only along a single line

It is tempting to promote "rapidities add" into a universal law and apply it to boosts in different directions. Do not. Rapidities add only for strictly collinear boosts. Two perpendicular boosts do not compose into a single boost at all: the product $\Lambda_y(\phi_2)\Lambda_x(\phi_1)$ is a boost followed by a genuine spatial rotation, the Thomas–Wigner rotation, whose accumulated version (Thomas precession) shifts atomic fine-structure levels by a measurable factor of two. The scalar formula $(\beta_1+\beta_2)/(1+\beta_1\beta_2)$ is likewise collinear-only. The clean geometry — commuting matrices, additive angles — is a privilege of the one-dimensional case, because only then do the two boosts share the same $(ct,x)$ plane.

Worked examples

1 · A spaceship at β = 0.8 — rapidity, dilation, and Doppler (fully numeric)

A spaceship recedes from Earth at $\beta = 0.8$ along the line of sight. Find its rapidity $\phi$ , the Lorentz factor $\gamma$ , the product $\gamma\beta$ , the time dilation of its clock, and the factor by which its radio beacon is Doppler-shifted.

Step 1. Rapidity from velocity: $\phi = \operatorname{artanh}(0.8) = \tfrac12\ln\!\dfrac{1+0.8}{1-0.8} = \tfrac12\ln 9 = \ln 3 \approx \mathbf{1.0986}$ .

Step 2. Lorentz factor: $\gamma = \cosh\phi = 1/\sqrt{1-0.64} = 1/\sqrt{0.36} = 1/0.6 = \mathbf{1.6667}$ , and $\gamma\beta = \sinh\phi = 1.6667\times0.8 = \mathbf{1.3333}$ . As a consistency check $\gamma^2 - (\gamma\beta)^2 = 2.7778 - 1.7778 = 1.000$ , the hyperbolic identity.

Step 3. Time dilation: a $1{:}00\,\mathrm{s}$ tick on the ship is seen from Earth to last $\Delta t = \gamma\,\Delta t' = 1.6667\ \mathrm{s}$ . The ship's clock runs slow by two-thirds of a second every second of proper time.

Step 4. Doppler shift: for recession the beacon frequency scales by $e^{-\phi} = e^{-1.0986} = 1/3 \approx \mathbf{0.333}$ , i.e. the received frequency is one-third of the emitted frequency, equivalently the wavelength is stretched threefold. Check against the textbook form: $\sqrt{(1-\beta)/(1+\beta)} = \sqrt{0.2/1.8} = \sqrt{1/9} = 1/3$ . So β = 0.8 gives φ = 1.0986, γ = 1.667, and a Doppler factor of exactly 1/3.

2 · A three-stage rocket — velocity addition made trivial (fully numeric)

A rocket is built in three stages. Each stage, once the one below it is exhausted, carries the payload $0.6c$ faster relative to the stage below, all collinearly along $x$ . Naively one expects $3\times0.6c = 1.8c$ . What is the true final speed relative to the launch pad, and what is $\gamma$ ?

Step 1. Convert each boost to rapidity. Since $\operatorname{artanh}(0.6) = \tfrac12\ln\dfrac{1.6}{0.4} = \tfrac12\ln 4 = \ln 2 \approx 0.6931$ , every stage contributes exactly $\phi = \ln 2$ .

Step 2. Add the rapidities: $\phi_{\text{tot}} = 3\ln 2 = \ln 8 \approx \mathbf{2.0794}$ .

Step 3. Convert back to velocity: $\beta_{\text{tot}} = \tanh(\ln 8) = \dfrac{8 - 1/8}{8 + 1/8} = \dfrac{63}{65} \approx \mathbf{0.9692}$ . The rocket reaches $0.9692c$ , not $1.8c$ — comfortably below the light cone.

Step 4. Lorentz factor: $\gamma = \cosh(\ln 8) = \dfrac{8 + 1/8}{2} = \dfrac{65}{16} = \mathbf{4.0625}$ . Cross-check by chaining the velocity formula: $0.6\oplus0.6 = 1.2/1.36 = 0.8824$ , then $0.8824\oplus0.6 = 1.4824/1.5294 = 0.9692$ — identical. Three 0.6c boosts give 63/65 c ≈ 0.9692c with γ = 65/16 = 4.0625.

3 · The relativistic rocket — rapidity grows linearly in proper time

An astronaut fires her engine to hold a constant proper acceleration $g = 9.8\ \mathrm{m/s^2}$ (Earth gravity, for comfort). For constant proper acceleration the rapidity accumulates at a constant rate, $\phi = g\tau/c$ , where $\tau$ is her own elapsed time. What speed has she reached after one year of proper time?

Step 1. Accumulate the rapidity. With $g = 9.8\ \mathrm{m/s^2}$ , one year $\tau = 3.156\times10^{7}\ \mathrm{s}$ , and $c = 3.00\times10^{8}\ \mathrm{m/s}$ : $\phi = g\tau/c = (9.8)(3.156\times10^{7})/(3.00\times10^{8}) \approx \mathbf{1.031}$ .

Step 2. Convert to velocity: $\beta = \tanh(1.031) \approx \mathbf{0.775}$ , so she is travelling at about $0.78c$ .

Step 3. Extrapolate. After two years $\phi\approx2.06$ , $\beta\approx0.968$ ; after three, $\phi\approx3.09$ , $\beta\approx0.996$ . Rapidity climbs by the same amount each year forever, yet each increment buys an ever-thinner slice of velocity. Constant thrust adds rapidity linearly and forever, yet β creeps toward 1 and never arrives.

Physics of the portrait

The boost is where relativity stopped being a set of strange formulas and became geometry. The man who made that leap took Einstein's kinematics and Poincaré's group theory and showed the world a picture: spacetime as a single four-dimensional stage on which a boost is nothing but a rotation.

Origins · 1 minHermann Minkowski and how this was discovered

Hermann Minkowski

1864–1909 · the geometry of spacetime and the boost as a hyperbolic rotation

Hermann Minkowski was Einstein's mathematics teacher at the Zurich Polytechnic, where he once called the young Einstein a "lazy dog." A few years later he did more than anyone to reveal the depth of his former student's 1905 paper. In his 1908 Cologne address "Raum und Zeit" he opened with the now-legendary declaration that "henceforth space by itself, and time by itself, are doomed to fade away into mere shadows, and only a kind of union of the two will preserve an independent reality." What Minkowski saw was that the Lorentz boost, which Einstein had derived as a transformation of measurements, is geometrically a rotation of a single spacetime — a hyperbolic rotation that preserves the interval $-(ct)^2 + x^2$ rather than a length. He made the four-dimensional worldline, the light cone, and the invariant interval the fundamental objects, turning special relativity from algebra into a geometry of events. The rapidity $\phi$ is precisely the angle of Minkowski's rotation, and the hyperbolae along which boosts slide events are his calibration curves. Einstein at first dismissed the reformulation as "superfluous learnedness," but soon recognized it as indispensable — without it the leap to the curved spacetime of general relativity, where the metric $\eta_{\mu\nu}$ becomes a dynamical field $g_{\mu\nu}$ , would have been almost unthinkable. Minkowski died suddenly of appendicitis in 1909, only months after his address, at forty-four.

History

1887

— Voigt writes transformations that preserve the wave equation, anticipating the algebra

1904

— Lorentz publishes the transformation equations in essentially their modern form

1905

— Einstein derives the boost from the two postulates of special relativity

1908

— Minkowski recasts the boost as a rotation of spacetime in "Raum und Zeit"

1910

— Varićak interprets the Lorentz boost as a translation in hyperbolic velocity space

1911

— Robb introduces the term "rapidity" for the additive hyperbolic boost parameter

Going further

Texts. Misner, Thorne & Wheeler, Gravitation, §2.9–2.10 (boosts as hyperbolic rotations, rapidity) · Wald, General Relativity, §1.2 · Carroll, Spacetime and Geometry, §1.2 · Hartle, Gravity, Ch. 4 · Schutz, A First Course in General Relativity, Ch. 1–2 · Taylor & Wheeler, Spacetime Physics, Ch. L (rapidity as the natural velocity parameter). Key anchor values used throughout: the hyperbolic identity $\gamma^2-(\gamma\beta)^2=1$ ; the matrix product giving $\Lambda(\phi_1+\phi_2)$ ; $\gamma=1.25$ at $\beta=0.6$ , $\gamma\approx1.667$ at $\beta=0.8$ , with $\operatorname{artanh}0.6=\ln2$ and $\operatorname{artanh}0.8=\ln3$ ; the three-stage result $\beta=63/65$ , $\gamma=65/16$ ; and the Doppler factor $e^{-\operatorname{artanh}0.8}=1/3$ . All are standard results with no nonstandard constants.

Prerequisites. Postulates of Special Relativity · Lorentz Transformations · Spacetime Interval · Minkowski Metric and Signature. Leads to. Time Dilation · Length Contraction · Relativity of Simultaneity · Four-Vectors · Causal Structure and Light Cones.

Ask the tutor: Ask me to boost a specific event, chain any set of collinear velocities through rapidity, or draw the calibration hyperbola for a speed you choose.

Is a Lorentz boost really "just a rotation"

In the $(ct,x)$ plane, yes — a boost is a hyperbolic rotation. Swap the circular functions $\cos\theta,\sin\theta$ for the hyperbolic ones $\cosh\phi,\sinh\phi$ , and swap the preserved quantity from $x^2+y^2$ to $-(ct)^2+x^2$ , and a rotation becomes a boost; the single sign flip is the whole difference. But the analogy has real limits: rapidity is unbounded and non-periodic (there is no "full turn"), the boost matrix is symmetric rather than orthogonal, and two boosts about different axes leave behind an extra spatial rotation. So a boost is a rotation only in the precise, and very useful, sense of a hyperbolic rotation in one spacetime plane.

Why introduce rapidity at all when β and γ already work

Because rapidity turns hard operations into easy ones. Composing collinear boosts, which in velocities needs the fiddly $(\beta_1+\beta_2)/(1+\beta_1\beta_2)$ , becomes plain addition $\phi_1+\phi_2$ . The boost matrix becomes a clean hyperbolic rotation, constant proper acceleration becomes rapidity growing linearly in proper time, and the Doppler shift becomes a simple exponential $e^{\pm\phi}$ . Whenever a problem chains boosts, accelerates steadily, or leans on the geometry, rapidity makes the structure visible. Velocity is what you measure; rapidity is what the mathematics prefers.

How does rapidity explain the cosmic speed limit

The speed limit becomes a horizon rather than a wall. Velocity and rapidity are related by $\beta = \tanh\phi$ , and $\tanh$ climbs from $0$ toward its asymptote $1$ but never reaches it, however large $\phi$ grows. You can always add more rapidity — fire your engine longer, chain another boost — yet each added unit buys a smaller gain in velocity as the $\tanh$ curve flattens. So a massive particle accumulates unlimited rapidity yet asymptotes to $c$ without touching it. Light itself sits at $\phi=\infty$ , which is why it has no rest frame and cannot be caught.

What is the difference between a boost and the full Lorentz transformation

A boost is the part of a Lorentz transformation that relates frames in relative motion — it mixes time with a spatial direction. The full Lorentz group also contains ordinary spatial rotations, which mix spatial directions and leave time alone. Any proper orthochronous transformation factors as a rotation, a boost, and another rotation. Boosts alone do not even form a closed group: combine two non-collinear boosts and you get a boost times a rotation, forcing you back into the larger group. The Lorentz group is six-dimensional — three rotation angles and three boost rapidities — and only the three rapidities describe changes of velocity.

Why does the same rapidity appear in time dilation, length contraction, and Doppler shift

Because all three are shadows of one hyperbolic rotation. Time dilation is the projection factor $\gamma=\cosh\phi$ by which a moving clock's tick lengthens; length contraction is the reciprocal projection $1/\gamma$ of a moving ruler; and the Doppler shift is the exponential $e^{\pm\phi}$ combining time dilation with the light-travel-time geometry. Expressing each in rapidity exposes the shared origin: $\cosh\phi$ , its reciprocal, and $e^{\pm\phi}=\cosh\phi\pm\sinh\phi$ are all built from the single boost angle. The apparent zoo of relativistic effects is really one rotation viewed from different sides.

Do rapidities ever fail to add

Yes — the additivity holds only for boosts along a single common line. The moment two boosts point in different directions, their matrices no longer share a plane and no longer commute, and their product is not a pure boost but a boost combined with a spatial rotation, the Thomas–Wigner rotation. Its physical shadow, Thomas precession, is a real and measured effect: it supplies the famous factor of one-half in the spin–orbit coupling of atomic fine structure. So "rapidities add" is a precise theorem about collinear boosts, not a general law; for skew velocities you must multiply the full matrices and accept the leftover rotation.