Relativistic Uniform Acceleration and the Rindler Hyperbola

Strap an accelerometer to a rocket and hold the reading at a steady $1\,g$ . The crew feel Earth-normal weight; nothing about their situation changes from one year to the next. Newton says the answer is immediate: $v = gt$ , a parabola, and after about a year the ship is doing the speed of light. After two years it is doing twice the speed of light. That is not a small quantitative error to be patched with a correction factor. It is a different curve. Constant proper acceleration in relativity does not trace a parabola in spacetime; it traces a hyperbola, $x^2 - c^2t^2 = (c^2/\alpha)^2$ — a curve of constant spacetime interval from a fixed event, exactly as a circle is a curve of constant Euclidean distance. The hyperbola runs forever, its speed climbing toward $c$ and never touching it, tucking itself asymptotically under the light cone. The accelerometer never wavers. The engine never throttles back. And the ship never reaches $c$ , because the curve it is on physically cannot cross the line it approaches. The whole of accelerated motion in flat spacetime hangs off this one worldline. The asymptote is a horizon; the family of nested hyperbolae is a coordinate system; the constant $c^2/\alpha$ is a length scale that turns out to be about a light-year at $1\,g$ . And none of it requires general relativity. Special relativity handles acceleration perfectly well.

Key takeaways

Signature $(+,-,-,-)$ : $\eta_{\mu\nu} = \mathrm{diag}(+1,-1,-1,-1)$ , $ds^2 = c^2dt^2 - dx^2$ , timelike intervals positive, $U\!\cdot\!U = +c^2$ and $A\!\cdot\!A = -\alpha^2$ (the four-acceleration is spacelike).
Constant proper acceleration $\alpha$ means constant accelerometer reading — the acceleration in the momentarily comoving inertial frame. It is not constant coordinate acceleration: $d^2x/dt^2 = \alpha/\gamma^3$ decays to zero as $v\to c$ .
Owned result: the worldline is the hyperbola $x^2 - c^2t^2 = (c^2/\alpha)^2$ , parametrised by proper time as $x = \frac{c^2}{\alpha}\cosh\frac{\alpha\tau}{c}$ , $ct = \frac{c^2}{\alpha}\sinh\frac{\alpha\tau}{c}$ . Rapidity grows linearly: $\varphi = \alpha\tau/c$ . Hyperbolic motion is to a boost what uniform circular motion is to a rotation.
Owned result: the Newtonian parabola crosses the light cone at exactly $t = c/\alpha$ — and at that very instant the true speed is exactly $c/\sqrt2 = 0.707c$ , and Newton has overstated the distance by exactly $(1+\sqrt2)/2 = 1.2071$ . Both are closed forms, not approximations: this is where the two theories part company, and it is a factor of $\sqrt2$ , not a factor of $1.000001$ .
The scale $c^2/\alpha$ is the ship's distance from the vertex of its hyperbola, and the light cone through that vertex is a Rindler horizon: signals from behind it never catch up. At $\alpha = g$ that distance is $c^2/g = 9.16\times10^{15}\ \mathrm{m} = 0.97$ light-year, and $c/g = 3.06\times10^{7}\ \mathrm{s} = 0.97$ year.
Key idea: the Minkowski metric's minus sign turns circles into hyperbolae. A curve of constant distance from the origin is a circle; a curve of constant spacetime interval is a hyperbola — and that is literally the trajectory. Everything else (the horizon, Rindler coordinates, the equivalence-principle model of a uniform field) is a consequence of this single curve.

The core insight

Rapidity, not velocity, is the natural measure of "how hard you have been accelerating". Velocities do not add; rapidities do, $\varphi = \operatorname{artanh}(v/c)$ . Constant proper acceleration says exactly one thing: you add rapidity at a constant rate, $d\varphi/d\tau = \alpha/c$ . Integrate that and you get $\varphi = \alpha\tau/c$ , and every kinematic quantity is then a hyperbolic function of it — $\beta = \tanh\varphi$ , $\gamma = \cosh\varphi$ , $\gamma\beta = \sinh\varphi$ . The speed saturates because $\tanh$ saturates, not because anything mysterious happens to the engine.

Where this lives today

Every particle accelerator is a hyperbolic-motion machine in the small. In a linac the field is applied in the lab frame, so what stays fixed is not quite the proper acceleration — but the essential fact, that the lab-frame acceleration dies as $\gamma^{-3}$ while the energy climbs linearly, is what forces the design. Past a few MeV an electron's speed is essentially $c$ and stops changing; every subsequent cavity buys energy, not velocity, and the beam optics are built on that. The hyperbola is also the flat-spacetime laboratory for horizon physics. The asymptote $x = ct$ is a genuine causal boundary for the accelerating observer: events beyond it can never send a signal that catches up, no matter how long the observer waits, and the entire apparatus of black-hole thermodynamics — Rindler coordinates, the Unruh temperature, the analogy to Hawking radiation — is first learned here, in a spacetime that is exactly flat. `aq-sr-rindler-coordinates` builds the coordinate system from the family of these curves; `aq-sr-rindler-horizon` treats the asymptote; `aq-sr-unruh-effect` treats the temperature. And `aq-sr-relativistic-rocket-equation-s6` asks what it would cost in fuel to fly one.

The physical picture

Newton's rocket is easy and wrong. Constant force, constant mass, constant coordinate acceleration: $x = \tfrac12 at^2$ , a parabola whose slope in the $(x, ct)$ plane grows without bound. At $t = c/a$ that slope reaches the light cone; beyond it, the worldline is spacelike, and the ship is outrunning its own light. No amount of care about the definition of force rescues this. The curve is wrong. The right question is: which quantity should be held constant? Coordinate acceleration is frame-dependent and not the thing an engineer controls. What the accelerometer reads — the acceleration measured in the frame momentarily comoving with the ship — is an invariant, it is what the crew feel underfoot, and it is the natural idealisation of "steady thrust". Call it the proper acceleration $\alpha$ ; its definition and its four-vector home are `aq-sr-proper-acceleration-comoving-frame` and `aq-sr-four-acceleration`, and we take from them the one fact we need: $\alpha$ is the invariant magnitude of the four-acceleration, and for collinear motion $\alpha = \gamma^3\,dv/dt$ . That factor of $\gamma^3$ is the whole story. It says the coordinate acceleration is $\alpha/\gamma^3$ — it dies as the ship speeds up, even though the accelerometer never budges. At $\beta = 0.9$ ( $\gamma = 2.294$ ) the ship gains coordinate speed at only $g/\gamma^3 = 0.813\ \mathrm{m/s^2}$ , a twelfth of what the crew feel. All the "effort" is going into rapidity that shows up less and less as extra coordinate velocity. This is the relativistic tax that keeps $v$ below $c$ , and it is levied automatically. Integrate it and the worldline is a hyperbola. That is not a coincidence of the algebra; it is geometry. In Euclidean space, a curve at constant distance from a point is a circle, and the natural motion at constant "turning" is uniform circular motion, with $x = R\cos\theta$ , $y = R\sin\theta$ and $\theta$ growing linearly. In Minkowski space, a curve at constant interval from an event is a hyperbola, and the natural motion at constant "boosting" is hyperbolic motion, with $x = X\cosh\varphi$ , $ct = X\sinh\varphi$ and $\varphi$ growing linearly. Cosine becomes cosh, angle becomes rapidity, and the minus sign in $ds^2 = c^2dt^2 - dx^2$ does everything else.

Figure

PUNCHLINE — the two curves part company by a factor of √2, not a factor of 1.000001. The Newtonian parabola x = ½gt² and the relativistic hyperbola x² − c²t² = (c²/g)² are indistinguishable at early times; but the parabola CROSSES the light cone at exactly ct = c²/g, and at that very instant the true worldline is doing exactly c/√2 = 0.707c and has gone only 0.401 ly to Newton's 0.484 ly — an overstatement of exactly (1+√2)/2 = 1.2071. Beyond that point the parabola is spacelike: the ship is outrunning its own light. The hyperbola never crosses; it asymptotes to the light cone forever, and that asymptote is the Rindler horizon, sitting c²/g = 0.97 ly behind the ship

Figure

The speed curve, and why the engine never notices. LEFT — Newton's v = gt is a straight line that walks through c at t = c/g and keeps going; the true v = gt/√(1+(gt/c)²) bends over and saturates. They are indistinguishable below about 0.3c and then part company completely. The marked point is the punchline: at t = c/g exactly, Newton says c and the truth says c/√2 = 0.707c. RIGHT — the accelerometer reading is FLAT at 9.81 m/s² forever, while the coordinate acceleration g/γ³ collapses: by β = 0.9 it is 0.813 m/s², a twelfth of what the crew feel, and it keeps falling. Both curves describe the same rocket. Nothing throttles the engine — the proper velocity γv = gt is still climbing linearly and without bound

The core equations

Take the defining condition from the comoving frame — the derivation of $\alpha = \gamma^3\,dv/dt$ belongs to `aq-sr-proper-acceleration-comoving-frame`:

\alpha \;=\; \gamma^{3}\,\frac{dv}{dt}\,, \qquad \gamma = \frac{1}{\sqrt{1-v^2/c^2}} \qquad\text{(collinear motion, } \alpha = \text{const)} \qquad (1)

where: $\alpha$ is the accelerometer reading, an invariant. Reading it backwards gives the sentence that this whole article exists to make: $\dfrac{d^2x}{dt^2} = \dfrac{\alpha}{\gamma^3}$ — the coordinate acceleration decays to zero as $v \to c$ . Integrating (1) from rest gives the lab-frame kinematics:

v(t) = \frac{\alpha t}{\sqrt{1 + (\alpha t/c)^2}}\,, \qquad x(t) = \frac{c^{2}}{\alpha}\left(\sqrt{1 + \left(\frac{\alpha t}{c}\right)^{2}} - 1\right) + x_0 \qquad (2)

where: $x_0$ is the starting position. Note the shape of $v(t)$ : it is $\alpha t$ for small $t$ and saturates at $c$ , and it does so without any adjustable parameter. Relativity does not "impose" a speed limit here — the limit falls out of the integral. Putting the vertex at the origin, so that $x_0 = 0$ and the ship starts at $x = c^2/\alpha$ , the worldline collapses to the article's central object:

\boxed{\;x^{2} - c^{2}t^{2} \;=\; \left(\frac{c^{2}}{\alpha}\right)^{2}\;} \qquad (3)

where: the right-hand side is a constant spacetime interval, so the worldline is a curve of fixed invariant separation from the vertex event — the Minkowski analogue of a circle. This is the Rindler hyperbola. Its asymptotes are $x = \pm ct$ . Parametrise by the ship's own clock. With rapidity $\varphi = \operatorname{artanh}(\beta)$ :

\boxed{\;\varphi = \frac{\alpha\tau}{c}\;} \qquad\Longrightarrow\qquad \beta = \tanh\frac{\alpha\tau}{c}, \quad \gamma = \cosh\frac{\alpha\tau}{c}, \quad x = \frac{c^{2}}{\alpha}\cosh\frac{\alpha\tau}{c}, \quad ct = \frac{c^{2}}{\alpha}\sinh\frac{\alpha\tau}{c} \qquad (4)

where: $\tau$ is proper time. This is the deepest form. Rapidity is additive under boosts (`aq-sr-rapidity-hyperbolic-boosts`), so "constant proper acceleration" simply means "rapidity accumulates at a constant rate", and the hyperbolic functions do the rest. Inverting the $t$ equation gives $\tau = \frac{c}{\alpha}\operatorname{arsinh}\!\big(\frac{\alpha t}{c}\big)$ , the ship's clock in terms of the lab's. And the scale that governs everything:

\frac{c}{g} = 3.06\times10^{7}\ \mathrm{s} = \mathbf{0.97\ yr}\,, \qquad \frac{c^{2}}{g} = 9.16\times10^{15}\ \mathrm{m} = \mathbf{0.97\ light\text{-}year} \qquad (5)

where: $g = 9.81\ \mathrm{m/s^2}$ . Physical reading: a $1\,g$ ship becomes relativistic on a timescale of a year and has a horizon about a light-year behind it. That coincidence — that both numbers are $0.97$ in their natural units — is why $1\,g$ is the canonical example.

Where the hyperbola comes from

Derivation AIntegrating constant proper acceleration into the Rindler hyperbola

The goal is to get from the accelerometer's reading to equation (3) with no hand-waving, and in particular to see where the speed limit enters. It does not enter as an assumption; it enters as the behaviour of an integral. We integrate the defining ODE twice in the lab frame, obtain $v(t)$ and $x(t)$ , and then eliminate $t$ to expose the invariant hyperbola. The payoff is that the ship's inability to reach $c$ turns out to be the same fact as the hyperbola's inability to cross its asymptote.

Step 1. Write the ODE. $\alpha = \gamma^3\,dv/dt = \big(1-v^2/c^2\big)^{-3/2}\,dv/dt$ , so $\big(1-v^2/c^2\big)^{-3/2}dv = \alpha\,dt$ . Why: this is the only place the physics enters. Everything after it is calculus, which is exactly the property a good formulation should have.

Step 2. Integrate once, from rest at $t=0$ . The antiderivative of the left side is $\dfrac{v}{\sqrt{1-v^2/c^2}} = \gamma v$ , so $\gamma v = \alpha t$ , giving $v(t) = \dfrac{\alpha t}{\sqrt{1+(\alpha t/c)^2}}$ . Why: the clean statement is $\gamma v = \alpha t$ — the proper velocity grows linearly and without bound, forever. It is only the ordinary velocity $v = \gamma v/\gamma$ that saturates, because $\gamma$ grows just as fast. Nothing is being suppressed; the bookkeeping is simply not what Newton assumed.

Step 3. Integrate again. $x(t) = \displaystyle\int_0^t \frac{\alpha t'\,dt'}{\sqrt{1+(\alpha t'/c)^2}} + x_0 = \frac{c^2}{\alpha}\left(\sqrt{1+(\alpha t/c)^2} - 1\right) + x_0$ . Why: for small $\alpha t/c$ the square root expands as $1 + \tfrac12(\alpha t/c)^2$ , and $x - x_0 \to \tfrac12\alpha t^2$ — Newton, recovered exactly, as he must be. The relativistic answer contains the parabola as its first term.

Step 4. Put the vertex at the origin. Choose $x_0 = c^2/\alpha$ , so $x(t) = \frac{c^2}{\alpha}\sqrt{1+(\alpha t/c)^2}$ . Squaring: $x^2 = \frac{c^4}{\alpha^2}\big(1 + \alpha^2t^2/c^2\big) = \big(\frac{c^2}{\alpha}\big)^2 + c^2t^2$ , which is equation (3). Why: the choice of origin is a convenience, but the invariant content is not — the quantity $x^2 - c^2t^2$ is a spacetime interval, the same number in every inertial frame. The worldline is a curve of constant interval from the vertex, and "curve of constant interval" is exactly what a Minkowski circle is.

Step 5. Read the speed limit off the geometry. As $t \to \infty$ , $x \to ct\sqrt{1 + (c/\alpha t)^2} \to ct + \frac{c^3}{2\alpha^2 t}$ : the worldline approaches the asymptote $x = ct$ from the right, and $\beta = dx/d(ct) = \alpha t/\sqrt{c^2 + \alpha^2t^2} \to 1$ but never equals it. Why: this is the whole speed-limit argument, and it is geometric rather than dynamical. A hyperbola cannot cross its asymptote — not because of a force, but because $x^2 - c^2t^2$ is a positive constant and $x = ct$ would require it to be zero. The ship is not being held back. It is on a curve that does not go there. *Verified in verify_aq-sr-relativistic-uniform-acceleration.py, symbolically (sympy) and in 50-digit arithmetic (mpmath): $x(t)$ from (2) satisfies (3) identically; $\gamma v = \alpha t$ ; and the small- $t$ expansion of $x(t)$ reproduces $\tfrac12\alpha t^2$ — computed with the cancellation-safe form, not by subtracting nearly-equal floats.*

Go deeper · Derivation B · 2 minRapidity grows linearly and hyperbolic functions do the rest

Derivation BRapidity grows linearly and hyperbolic functions do the rest

Equation (4) is the form that makes the subject easy, and it deserves a derivation rather than an assertion, because the statement "rapidity is linear in proper time" is the one sentence a reader should carry away. The goal is to show that constant proper acceleration is exactly the condition $d\varphi/d\tau = \alpha/c$ , and that once this is known, the entire kinematics can be written down without integrating anything at all.

Step 1. Recall the definitions. $\beta = \tanh\varphi$ , so $\gamma = \cosh\varphi$ and $\gamma\beta = \sinh\varphi$ — the identity $\cosh^2 - \sinh^2 = 1$ is $\gamma^2 - \gamma^2\beta^2 = 1$ . Why: rapidity is not a change of notation; it is the parameter in which boosts compose by addition (`aq-sr-rapidity-hyperbolic-boosts`). That is why it, and not $v$ , is the right thing to accumulate.

Step 2. Differentiate. $\dfrac{d\beta}{d\varphi} = \operatorname{sech}^2\varphi = 1/\gamma^2$ , and $dt = \gamma\,d\tau$ , so $\dfrac{dv}{dt} = c\,\dfrac{d\beta}{d\varphi}\dfrac{d\varphi}{d\tau}\dfrac{d\tau}{dt} = \dfrac{c}{\gamma^{3}}\dfrac{d\varphi}{d\tau}$ . Why: two factors of $\gamma$ come from $d\beta/d\varphi$ and one from time dilation. That is the $\gamma^3$ , assembled from its parts rather than quoted.

Step 3. Compare with (1). Since $\alpha = \gamma^3\,dv/dt$ , the last line gives $\alpha = c\,d\varphi/d\tau$ , hence $\boxed{d\varphi/d\tau = \alpha/c}$ and, from rest, $\varphi = \alpha\tau/c$ . Why: this is the theorem. "Constant accelerometer reading" and "rapidity accumulating at a constant rate" are the same statement. The $\gamma^3$ that looked like an ugly relativistic correction was the Jacobian between $v$ and $\varphi$ all along.

Step 4. Write everything down. $\beta = \tanh(\alpha\tau/c)$ , $\gamma = \cosh(\alpha\tau/c)$ , and integrating $dx/d\tau = \gamma v = c\sinh\varphi$ and $c\,dt/d\tau = c\gamma = c\cosh\varphi$ with $d\varphi = (\alpha/c)d\tau$ gives $x = \frac{c^2}{\alpha}\cosh\frac{\alpha\tau}{c}$ and $ct = \frac{c^2}{\alpha}\sinh\frac{\alpha\tau}{c}$ — equation (4), and squaring recovers (3) at once via $\cosh^2-\sinh^2=1$ . Why: not one integral was harder than $\int\cosh$ . That is the payoff of choosing the right parameter, and it is precisely the analogy with uniform circular motion, where choosing the angle turns everything into $\sin$ and $\cos$ .

Step 5. Check the four-acceleration. From (4), $U^\mu = c(\cosh\varphi, \sinh\varphi)$ and $A^\mu = dU^\mu/d\tau = \alpha(\sinh\varphi, \cosh\varphi)$ , so $U\!\cdot\!U = c^2(\cosh^2-\sinh^2) = +c^2$ and $A\!\cdot\!A = \alpha^2(\sinh^2 - \cosh^2) = -\alpha^2$ , with $A\!\cdot\!U = 0$ . Why: $A\!\cdot\!A$ is negative in $(+,-,-,-)$ — the four-acceleration is spacelike, and $\alpha = \sqrt{-A\!\cdot\!A}$ . A base that prints $\alpha = \sqrt{+A\!\cdot\!A}$ has silently changed signature. The magnitude is constant, which confirms that this worldline really does have the property we demanded of it. *Verified symbolically in verify_aq-sr-relativistic-uniform-acceleration.py: on the worldline (4), $U\!\cdot\!U = +c^2$ , $A\!\cdot\!U = 0$ and $A\!\cdot\!A = -\alpha^2$ exactly, in $\eta = \mathrm{diag}(+1,-1,-1,-1)$ , for symbolic $\tau$ .*

Go deeper · Derivation C · 2 minWhere Newton breaks: exactly c over root two, and exactly one plus root two over two

Derivation CWhere Newton breaks: exactly c over root two, and exactly one plus root two over two

Textbooks say the parabola "fails at relativistic speeds" and leave it there, which teaches nothing. The goal of this card is to make the failure quantitative and exact. There is a distinguished moment — the instant the Newtonian worldline crosses the light cone — and at that moment both the true speed and the ratio of the two predicted distances are closed forms with square roots in them, not messy decimals. This is the article's owned structural result, and it is the sharpest possible answer to "how wrong is Newton".

Step 1. Locate the crossing. Newton's worldline is $x_N = \tfrac12\alpha t^2$ , with speed $v_N = \alpha t$ . It reaches $v_N = c$ at $t_* = c/\alpha$ , and at that instant its slope in the $(x,ct)$ plane is exactly $1$ : it is tangent to the light cone, and for $t > t_*$ it is spacelike. Why: this is not a place where the approximation gets bad. It is where the curve stops being a physically possible worldline at all. For $g = 9.81$ , $t_* = c/g = 3.06\times10^{7}\ \mathrm{s} = 0.97$ years.

Step 2. Evaluate the truth at the same instant. From (2) with $\alpha t_*/c = 1$ : $v(t_*) = \dfrac{c}{\sqrt{1 + 1}} = \boxed{\dfrac{c}{\sqrt2}} = 0.7071\,c$ . Why: exact, and startling. At the very moment Newton says "light speed", the ship is doing seventy per cent of it. The disagreement is not a small correction — it is a factor of $\sqrt2$ in $\gamma\beta$ , and the two theories are simply describing different curves.

Step 3. Compare the distances. Newton: $x_N(t_*) = \tfrac12\alpha t_*^2 = \tfrac12\,c^2/\alpha$ . Truth, from (2): $x(t_*) = \dfrac{c^2}{\alpha}\big(\sqrt{2}-1\big)$ . The ratio is $\dfrac{x_N}{x} = \dfrac{1/2}{\sqrt2 - 1} = \dfrac{1}{2(\sqrt2-1)} = \boxed{\dfrac{1+\sqrt2}{2}} = 1.2071$ . Why: rationalising the denominator gives a clean closed form. Newton overstates the distance covered by exactly $20.71\%$ — again a $\sqrt2$ -flavoured number, not an accident of arithmetic.

Step 4. Put it in light-years. With $\alpha = g$ : $x_N(t_*) = \tfrac12(9.16\times10^{15}) = 4.58\times10^{15}\ \mathrm{m} = 0.484$ ly, while the true distance is $(\sqrt2-1)(9.16\times10^{15}) = 3.79\times10^{15}\ \mathrm{m} = 0.401$ ly. Why: a real ship, on a real trip, and a discrepancy of 0.08 light-years after less than a year of flight. This is why the relativistic rocket problem cannot be done with a Newtonian sketch and a fudge factor.

Step 5. Confirm they agree where they must. Expand (2) for $\alpha t \ll c$ : $x(t) = \frac{c^2}{\alpha}\big(\sqrt{1+(\alpha t/c)^2}-1\big) \to \frac{c^2}{\alpha}\cdot\tfrac12\big(\tfrac{\alpha t}{c}\big)^2 = \tfrac12\alpha t^2$ , and $v \to \alpha t$ . Why: the two theories are not rivals at low speed; the parabola is the leading term of the hyperbola. What the exact result adds is the knowledge of exactly where and by how much the leading term stops being enough — and the answer, at $t = c/\alpha$ , is "by a factor of $\sqrt2$ ". *Verified in verify_aq-sr-relativistic-uniform-acceleration.py with mpmath at 50 digits: $v(c/\alpha) = c/\sqrt2$ and $x_N/x = (1+\sqrt2)/2 = 1.2071$ exactly (symbolically in sympy, and to 40 significant figures numerically), independent of $g$ and $c$ — and the script perturbs both to confirm the printed light-year figures do move, so they are results and not floating-point artefacts.*

Limiting cases

Case	Result	Meaning
$\alpha\tau \ll c$	$x \to \tfrac12\alpha t^{2}$ , $\;v \to \alpha t$ , $\;\tau\to t$	the Newtonian parabola, recovered exactly as the leading term
$t = c/\alpha$	$\beta = 1/\sqrt2$ ; Newton is high by $(1+\sqrt2)/2$	the crossing point — the owned result
$\alpha\tau \gg c$	$\gamma \approx \tfrac12 e^{\alpha\tau/c}$ ; $\;x - ct \to \frac{c^2}{\alpha}e^{-\alpha\tau/c}$	the worldline hugs the light cone exponentially; the gap never closes
$c \to \infty$	$c^2/\alpha \to \infty$ ; hyperbola $\to$ parabola	the vertex and the horizon recede to infinity; Newton is the flat-cone limit
$\alpha \to 0$	$x^2 - c^2t^2 \to \infty$	the hyperbola degenerates into a straight, inertial worldline
photon	$x = ct$	the asymptote itself: an infinite proper acceleration would be needed to ride it

Common trapConstant proper acceleration is NOT constant coordinate acceleration

This is the single most common error in the subject and it produces the forbidden parabola. The accelerometer reads $\alpha$ , a constant, forever. The lab sees $\dfrac{d^2x}{dt^2} = \dfrac{\alpha}{\gamma^{3}} = \alpha\big(1 - v^2/c^2\big)^{3/2}$ , which decays to zero. At $\beta = 0.9$ ( $\gamma = 2.294$ ) a $1\,g$ ship is gaining lab speed at only $0.813\ \mathrm{m/s^2}$ ; at $\beta = 0.99$ it is a hundred times less again. Both statements are true and they are answers to different questions. If you hold $d^2x/dt^2$ constant instead, you are describing a rocket whose crew are crushed harder and harder — and which still overtakes light. What does grow without bound, linearly and forever, is the proper velocity $\gamma v = \alpha t$ . Nothing is capped except the ratio $v = \gamma v/\gamma$ .

Common trapYou do not need general relativity for this — and the sign of the four-acceleration is not optional

Every result on this page lives in flat Minkowski spacetime, with the Riemann tensor identically zero. The folklore that "accelerated frames require GR" is false: general relativity is needed for gravity, that is, for curvature and for tidal effects that no change of coordinates can remove. Acceleration is not curvature, and hyperbolic motion is not a gravitational problem. Separately, in $(+,-,-,-)$ the four-acceleration is spacelike: $A\!\cdot\!A = -\alpha^2 < 0$ , so $\alpha = \sqrt{-A_\mu A^\mu}$ . Sources written in $(-,+,+,+)$ print $\alpha = \sqrt{+A_\mu A^\mu}$ , and transcribing that into our convention asks for the square root of a negative number. The magnitude is the same; the formula is not.

Worked examples

1 · One year at 1 g

A ship holds a steady $1\,g$ from rest. After one year on the ship's own clock, how fast is it going, how much time has passed on Earth, and how far has it come?

Step 1 — identify. Constant proper acceleration, so use the proper-time form (4). Everything is a hyperbolic function of the single dimensionless number $\varphi = g\tau/c$ .

Step 2 — the rapidity. With $c/g = 3.0560\times10^{7}\ \mathrm{s}$ and $\tau = 1\ \mathrm{yr} = 3.1558\times10^{7}\ \mathrm{s}$ (Julian year):

\varphi = \frac{g\tau}{c} = \frac{3.1558\times10^{7}}{3.0560\times10^{7}} = \mathbf{1.0326}.

Step 3 — apply. $\beta = \tanh(1.0326) = \mathbf{0.775}$ , so $\gamma = \cosh(1.0326) = \mathbf{1.582}$ . Earth's clock reads $t = \frac{c}{g}\sinh\varphi = (3.0560\times10^{7})(1.2262) = 3.747\times10^{7}\ \mathrm{s} = \mathbf{1.19\ yr}$ . The distance travelled is $x = \frac{c^2}{g}(\cosh\varphi - 1) = (9.16\times10^{15})(0.5823) = 5.33\times10^{15}\ \mathrm{m} = \mathbf{0.564\ ly}$ .

Step 4 — the boundary number. In one year of ship time the crew have reached $77.5\%$ of light speed and are already $0.56$ light-years out, and Earth is running only $19\%$ fast. The relativistic effects are large but not yet dramatic — and the whole regime was reached with an engine you could survive. That is what makes $1\,g$ the canonical case: it is the acceleration at which a human can live, and the timescale on which relativity takes over is a year, not a geological age.

2 · Exactly how wrong is Newton

At what moment does the Newtonian parabola become impossible, and what is the true state of the ship at that instant?

Step 1 — identify. Newton's speed is $v_N = gt$ , which equals $c$ at $t_* = c/g$ . This is the crossing point of Derivation 3.

Step 2 — apply. $t_* = c/g = 3.06\times10^{7}\ \mathrm{s} = \mathbf{0.97\ yr}$ . The true speed there is $v(t_*) = \dfrac{gt_*}{\sqrt{1 + (gt_*/c)^2}} = \dfrac{c}{\sqrt2} = \mathbf{0.7071\,c}$ — exactly.

Step 3 — the distances. Newton: $x_N = \tfrac12 g t_*^2 = \tfrac12\,c^2/g = 4.58\times10^{15}\ \mathrm{m} = \mathbf{0.484\ ly}$ . Truth: $x = \frac{c^2}{g}(\sqrt2 - 1) = 3.79\times10^{15}\ \mathrm{m} = \mathbf{0.401\ ly}$ . The ratio is exactly $\dfrac{1+\sqrt2}{2} = \mathbf{1.2071}$ .

Step 4 — the boundary number. Newton overstates the distance by $20.71\%$ and the speed by $41.4\%$ at the very instant his own theory becomes self-contradictory — and both errors are exact multiples of $\sqrt2$ , not fitted numbers. Push on to a ship-clock reading of $\tau = 5c/g$ (about $4.8$ years) and the truth is $\gamma = \cosh 5 = \mathbf{74.2}$ with $\beta = \tanh 5 = 0.99991$ ; Newton, who does not distinguish $\tau$ from $t$ at all, would simply report $v = g\tau = \mathbf{5c}$ . The two theories are not close, and they never converge again.

Physics of the portrait

Hyperbolic motion is the only motion in special relativity that can be sustained forever with a finite, constant effort. Everything else either coasts or runs out.

Origins · 2 minHyperbolic motion and how this was discovered

Hyperbolic motion

1909–1966 · Born names the curve and finds it in the rigid electron · Minkowski's geometry makes it inevitable · Rindler turns the family of hyperbolae into a coordinate system and a horizon

The curve has a longer history than the name. Hermann Minkowski's 1908 lecture "Raum und Zeit" had already made hyperbolae the natural geometry of spacetime — his "calibration hyperbola" $x^2 - c^2t^2 = 1$ is the locus of events at unit invariant interval from the origin, the Minkowski replacement for the unit circle — and once you have that, the worldline of constant proper acceleration is sitting there waiting to be noticed. It was noticed the following year by Max Born, in the same 1909 paper on the rigid electron ("Die Theorie des starren Elektrons in der Kinematik des Relativitätsprinzips", Ann. Phys. 335, 1–56) that gave us Born rigidity; it is Born who called the trajectory hyperbolische Bewegung — hyperbolic motion — and Arnold Sommerfeld who, shortly after, showed how cleanly the whole of it falls out of hyperbolic functions of the rapidity. The idea then lay comparatively quiet for half a century, useful but unglamorous, until Wolfgang Rindler made the family of hyperbolae into a frame. In 1956 (MNRAS 116, 662) he had coined the terms "event horizon" and "particle horizon" for cosmology; in 1966 (Am. J. Phys. 34, 1174, "Kruskal Space and the Uniformly Accelerated Frame") he drew the explicit parallel between the uniformly accelerated rigid rod in flat spacetime and Kruskal's chart of the Schwarzschild black hole — and the wedge, the coordinates and the horizon have carried his name since. That is the arc of the subject in one sentence: a curve found while thinking about electrons became the standard laboratory for thinking about black holes.

History

1908

— Minkowski's "Raum und Zeit" makes the invariant hyperbola the calibration curve of spacetime geometry.

1909

— Born, in his paper on the rigid electron, identifies the worldline of constant proper acceleration and names it hyperbolic motion (Ann. Phys. 335, 1-56).

1910

— Sommerfeld recasts the kinematics in hyperbolic functions of the rapidity, showing that constant proper acceleration means rapidity linear in proper time.

1956

— Rindler coins the terms "event horizon" and "particle horizon" in Visual Horizons in World Models (MNRAS 116, 662).

1966

— Rindler, in "Kruskal Space and the Uniformly Accelerated Frame" (Am. J. Phys. 34, 1174), maps the accelerated frame onto the black-hole chart; the wedge and its horizon take his name.

1976

— Unruh shows that a detector on this worldline sees the vacuum as a thermal bath, tying hyperbolic motion to horizon thermodynamics.

Going further

Texts. Rindler, Introduction to Special Relativity, 2nd ed., §14 (hyperbolic motion) · Taylor & Wheeler, Spacetime Physics, 2nd ed., ch. on accelerated observers · Misner, Thorne & Wheeler, Gravitation, §6 · Born, M., Ann. Phys. 335, 1–56 (1909) · Rindler, W., MNRAS 116, 662 (1956); Am. J. Phys. 34, 1174 (1966).

Provenance. Signature $(+,-,-,-)$ ; $U\!\cdot\!U = +c^2$ , $A\!\cdot\!A = -\alpha^2$ (negative, spacelike). Every number above is recomputed in verify_aq-sr-relativistic-uniform-acceleration.py, which greps this article for each printed value and evaluates every cancellation-prone expression in 50-digit arithmetic rather than float64. Base status: this article's AllQFT base is one of the sound ones — all eight of its equation cards are on-topic and correct, and its scale figures ( $c/g \approx 0.97$ yr, $c^2/g \approx 0.97$ ly) check out. The corrections here are of emphasis, not of fact: the base states $\alpha = \sqrt{-a^\mu a_\mu}$ correctly for our signature, and we have kept it.

Scope. The definition of proper acceleration and the derivation of $\alpha = \gamma^3\,dv/dt$ belong to `aq-sr-proper-acceleration-comoving-frame`; the four-acceleration as an object is `aq-sr-four-acceleration`; rapidity and its additivity are `aq-sr-rapidity-hyperbolic-boosts`. The horizon is named here and treated in `aq-sr-rindler-horizon`; the coordinate system built from these curves is `aq-sr-rindler-coordinates`. The Unruh temperature is mentioned as a signpost and is not derived.

Builds on. Proper Acceleration and the Comoving Frame · The Equivalence Principle. Leads to. Rindler Coordinates and the Uniformly Accelerated Frame · The Rindler Horizon · The Relativistic Rocket Equation and Traveler Time · Unruh Effect and Thermal Radiation for Accelerated Observers.

If the accelerometer never changes its reading, why does the ship stop gaining speed?

It does not stop gaining something. What grows linearly and without bound, forever, is the proper velocity $\gamma v = \alpha t$ — the distance covered per unit of the ship's own time. It is only the ratio $v = (\gamma v)/\gamma$ that saturates, because $\gamma$ climbs at exactly the same rate. So the ship keeps buying exactly as much motion per second of engine burn as it always did; what changes is how much of that purchase shows up as extra coordinate speed in the lab. Nothing throttles the engine and nothing resists it. The bookkeeping is simply not Newton's.

Is the hyperbola an approximation, valid only for small acceleration or short times?

No — it is exact, for any $\alpha$ and for all time. Equation (3) is the complete solution of the exact relativistic equation of motion for constant proper acceleration, with no expansion anywhere. It is the parabola that is an approximation: expand the hyperbola for $\alpha t \ll c$ and its first term is $\tfrac12\alpha t^2$ . The relationship is the usual one between an exact result and its leading behaviour, and the exact result is the simpler of the two to write down.

Does the ship need general relativity, since it is accelerating?

No, and this is the most persistent myth in the subject. Every equation on this page lives in flat Minkowski spacetime, with the Riemann tensor identically zero everywhere. Special relativity is perfectly capable of handling accelerated objects and even accelerated frames — Rindler coordinates are a coordinate change on flat spacetime and nothing more. General relativity is needed for gravity: for curvature, and for tidal effects that no choice of coordinates can remove. Acceleration is not gravity, and the fact that the equivalence principle relates them locally is precisely a statement about how small a region you have to stay in.

What exactly is at the vertex of the hyperbola, and why is it a horizon?

The vertex is the event at $x = 0$ , $t = 0$ — a point the ship never visits, sitting $c^2/\alpha$ behind its starting position (about a light-year for a $1\,g$ ship). The light cone through that vertex, $x = \pm ct$ , is the hyperbola's pair of asymptotes. A light signal emitted from just behind the future asymptote can never catch the ship: the ship's worldline approaches that null line but stays strictly to the right of it forever, so the signal chases it eternally at a shrinking but never-vanishing deficit. That is a genuine causal boundary, and it is called a Rindler horizon. It is not a place where anything is physically wrong — an inertial observer sails straight through — but for this particular family of accelerated observers it is as impenetrable as a black hole's. That is why the wedge is the standard model system for horizon physics, and it is developed in `aq-sr-rindler-horizon`.

Why is the analogy with uniform circular motion more than a pun?

Because it is the same mathematics with one sign changed. Circular motion is the curve of constant distance from a point, $x^2 + y^2 = R^2$ , traversed with the angle growing linearly: $x = R\cos\theta$ , $y = R\sin\theta$ . Hyperbolic motion is the curve of constant interval from an event, $x^2 - c^2t^2 = X^2$ , traversed with the rapidity growing linearly in proper time: $x = X\cosh\varphi$ , $ct = X\sinh\varphi$ . The metric's minus sign swaps $\cos,\sin$ for $\cosh,\sinh$ and the angle for rapidity; the "radius" becomes the invariant $c^2/\alpha$ ; the rotation group becomes the boost group. Even the constant $|A|$ carries over — one points at the centre, the other at the vertex.

Both numbers in the 1 g example come out at 0.97. Is that a coincidence?

Yes, and a useful one. The two ratios are necessarily equal — a light-year is $c$ times a year, and $c^2/g$ is $c$ times $c/g$ — so there is only one coincidence, not two: that $g$ on this particular planet puts $c/g$ within three per cent of the Earth's orbital period. Nothing deep. But it makes $1\,g$ an extraordinarily convenient unit, because "one year of ship time" and "one light-year" are, to three per cent, the natural units of the problem.