Relativity: Special Relativity

Special Relativity

Special relativity, published by Einstein in 1905, governs objects moving at constant velocity and forces a radical revision of space and time.

By the late 1800s, Maxwell’s equations predicted a definite speed of light, $c$ — but a speed relative to what? Every other wave (sound, water ripples) travels relative to a medium, and velocities simply add: throw a ball forward on a moving train and the ground sees it go faster. Yet the Michelson–Morley experiment found light always travels at $c$, no matter how fast you chase it. Einstein took this literally: if everyone measures the same light speed, then the rate clocks tick and the length of rulers — assumed absolute — must instead bend so that $c$ stays fixed. Time dilation, length contraction, and $E=mc^2$ are all the logical price of that one stubborn fact.

Postulates of Special Relativity

Principle of relativity — the laws of physics are the same in all inertial reference frames.
Constancy of light speed — the speed of light in vacuum is the same for all observers, regardless of their motion.

Relativity of Simultaneity

Before deriving the Lorentz transformation, isolate the single idea that drives every other relativistic effect: two events simultaneous in one inertial frame are generally not simultaneous in another. Time dilation and length contraction are downstream consequences. The breakdown of absolute simultaneity is what lets the symmetry of relativity — each observer seeing the other’s clocks run slow — be free of contradiction.

The train-and-platform thought experiment. A railway car of proper length moves right past a platform at speed $v$. A lamp sits at the car’s exact midpoint. At the instant the lamp passes a platform observer (Alice), it flashes once, sending light toward the front and rear walls.

In the train's frame (observer Bob, riding at the midpoint): the front and rear walls are equidistant from the lamp and the car is at rest, so the two flashes travel equal distances at the same speed $c$. They strike the front and rear walls simultaneously. For Bob, "front hit" and "rear hit" are the same instant.

In the platform frame (Alice): light still travels at $c$ in her frame too (second postulate), but during the flight the rear wall rushes toward the emission point while the front wall flees away from it. The rearward light therefore meets its wall first; the forward light has to chase a receding target and arrives later. For Alice, the rear event happens before the front event — the very same pair of events is no longer simultaneous.

Neither observer is mistaken. Both correctly apply the same two postulates and reach different — but internally consistent — conclusions about ordering. Simultaneity is a property of a chosen frame, not of the events themselves. (Note that causally connected events, those inside each other's light cones, do keep their order in every frame; only the timing of spacelike-separated events like these two wall-strikes is frame-dependent.)

Leading clocks lag

To make the effect quantitative, replace the single lamp with a row of clocks that Bob has synchronized along the length of his car. Apply the time component of the Lorentz transformation (derived in the next section), $t' = \gamma\left(t - vx/c^2\right)$, to a single instant $t = \text{const}$ in Alice's platform frame. The clock readings Bob's frame assigns differ from place to place purely because of the $-\gamma v x/c^2$ term:

$\Delta t' = -\frac{v\, \Delta x}{c^2}$

Here $\Delta x$ is the spatial separation of two clocks as measured in the platform frame and $\Delta t'$ is the offset between their readings in the train frame at one platform instant. The minus sign carries the physics: of two clocks separated along the direction of motion, the one in the lead (the front clock, at larger $x$) shows an earlier time. This is the "leading clocks lag" rule:

Leading clocks lag. In a frame that sees a row of synchronized clocks moving, the clock that is ahead in the direction of motion reads behind in time, by an amount $vL_0/c^2$, where $L_0$ is the proper separation of the clocks. The trailing clock is the one that appears set forward.

This single asymmetry is the hidden engine behind the apparent paradoxes of special relativity. When Alice insists Bob's clocks are unsynchronized — front behind, rear ahead — and Bob says exactly the same of Alice's clocks, both are right, and the twin- and ladder-paradox "contradictions" dissolve. Length contraction can even be derived from it: because the two ends of a moving ruler are timed using clocks that disagree about "now," a frame measuring the ruler's length records a contracted value $L = L_0/\gamma$.

Worked Example: by how much do the clocks disagree?

A train car of proper length $L_0 = 100\ \text{m}$ moves past a platform at $v = 0.6c$. Bob has synchronized a clock at the front and one at the rear in the train frame. According to Alice on the platform, by how much are they out of step at any single platform instant?

$$\Delta t' = \frac{v L_0}{c^2} = \frac{(0.6c)(100\ \text{m})}{c^2} = \frac{(0.6)(100\ \text{m})}{c} = \frac{60\ \text{m}}{3.0\times10^{8}\ \text{m/s}} = 2.0\times10^{-7}\ \text{s}.$$

Alice finds the leading (front) clock reads about 0.20 µs behind the trailing (rear) clock. The offset is independent of where along the track she looks — it is fixed by the proper length and the speed, not by position — and it grows linearly with both. Stretch the "car" to the diameter of a galaxy and modest speeds produce offsets of years, which is how relativity reconciles wildly different accounts of "now" at cosmic distances.

Spacetime and the Lorentz Transformation

Spacetime Interval

The spacetime interval between two events is invariant:

Convention note: two metric-signature conventions are in common use. This section writes the interval with the (+,−,−,−) ("mostly-minus") convention in its primary algebraic form, then gives the differential form in the (−,+,+,+) ("mostly-plus") convention to match the Minkowski metric $\eta_{\mu\nu}$ below. The two differ only by an overall sign and describe identical physics.

$(\Delta s)^2 = c^2(\Delta t)^2 - (\Delta x)^2 - (\Delta y)^2 - (\Delta z)^2$

In differential form (using the (−,+,+,+) convention):

$ds^2 = -c^2 dt^2 + dx^2 + dy^2 + dz^2 = \eta_{\mu\nu}\, dx^\mu dx^\nu$

Where $\eta_{\mu\nu}$ is the Minkowski metric:

$\eta_{\mu\nu} = \begin{pmatrix} -1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1 \end{pmatrix}$

Light Cone Structure

Derivation of Lorentz Transformations

Starting from the invariance of the spacetime interval and the principle of relativity:

For two reference frames S and S’, where S’ moves with velocity v along the x-axis:

\[c^2t'^2 - x'^2 = c^2t^2 - x^2\]

Assuming linear transformation:

$x' = Ax + Bt$ $t' = Cx + Dt$

From the origin of S’ (x’ = 0) moving at x = vt:

\[0 = Avt + Bt \rightarrow B = -Av\]

From the invariance of light speed (x = ct implies x’ = ct’):

$ct' = Act + Bt = Act - Avt = A(c - v)t$ $x' = Act + Bt = Act - Avt = A(c - v)t$

Therefore: A = γ = 1/√(1 - v²/c²)

Complete Lorentz transformations:

$x' = \gamma(x - vt)$ $y' = y$ $z' = z$ $t' = \gamma(t - vx/c^2)$

Inverse transformations:

$x = \gamma(x' + vt')$ $y = y'$ $z = z'$ $t = \gamma(t' + vx'/c^2)$

Matrix form:

\[\begin{pmatrix} ct' \\ x' \\ y' \\ z' \end{pmatrix} = \begin{pmatrix} \gamma & -\beta\gamma & 0 & 0 \\ -\beta\gamma & \gamma & 0 & 0 \\ 0 & 0 & 1 & 0 \\ 0 & 0 & 0 & 1 \end{pmatrix} \begin{pmatrix} ct \\ x \\ y \\ z \end{pmatrix}\]

Where β = v/c.

Time Dilation

Moving clocks run slower relative to stationary observers:

\[\Delta t = \gamma \Delta t_0\]

where $\gamma = 1/\sqrt{1 - v^2/c^2}$ is the Lorentz factor, $\Delta t_0$ is the proper time (in the rest frame), and $\Delta t$ is the dilated time (in the moving frame).

Time Dilation Calculator

Velocity (as fraction of c): 0.5

Lorentz factor γ = 1.155

1 hour proper time = 1.155 hours observed

GPS example. GPS satellites must account for both special- and general-relativistic effects, which act in opposite directions. Their orbital velocity (~14,000 km/h, $v \approx 3{,}900$ m/s, $\gamma - 1 \approx 8.4\times10^{-11}$) causes a special-relativistic slowing of about −7 µs/day. But the satellites also sit higher in Earth’s gravitational well, where clocks run faster — a general-relativistic gain of about +45 µs/day. The gravitational term dominates, so the net effect makes GPS clocks run fast by roughly +38 µs/day. Left uncorrected, this would introduce navigation errors of about 10 km per day.

Length Contraction

Objects are shorter along the direction of motion:

\[L = \frac{L_0}{\gamma}\]

where $L_0$ is the proper length (in the rest frame) and $L$ is the contracted length (in the moving frame).

Relativistic Velocity Addition

Velocities don’t simply add in special relativity:

\[u = \frac{v + w}{1 + vw/c^2}\]

This ensures that no velocity exceeds the speed of light.

Worked Example: chasing a light beam

Suppose a spaceship moves at $v = 0.9c$ relative to Earth and fires a probe forward at $w = 0.9c$ relative to the ship. Classically you would expect $1.8c$ — faster than light. Relativity gives instead:

$$u = \frac{0.9c + 0.9c}{1 + (0.9)(0.9)} = \frac{1.8c}{1.81} \approx 0.994c$$

The probe still travels below $c$. And if the ship instead fired a light beam ($w = c$), the formula returns exactly $c$ no matter the ship's speed — the second postulate, falling out of the algebra. Speeds combine so that $c$ is an unreachable ceiling, not a wall you can edge past by stacking velocities.

What these effects actually mean. Time dilation and length contraction are not optical illusions or measurement errors — they are real and symmetric. Each observer genuinely sees the other’s clock running slow and ruler shrunk, with no contradiction because “now” is frame-dependent (relativity of simultaneity): two observers disagree on which distant events are simultaneous, so they slice spacetime differently. The one quantity everyone agrees on is the invariant interval $ds^2$ — distances and durations are its shadows cast at different angles.

Mass-Energy Equivalence

Einstein’s most famous equation:

\[E = mc^2\]

Total energy of a particle:

\[E^2 = (pc)^2 + (mc^2)^2\]

Where p is the relativistic momentum:

\[p = \gamma mv\]

Relativistic Dynamics

Relativistic Momentum

\[p = \gamma mv\]

Relativistic Force

\[F = \frac{dp}{dt} = \frac{d(\gamma mv)}{dt}\]

Relativistic Kinetic Energy

\[KE = (\gamma - 1)mc^2\]

Reading the energy-momentum relation. $E^2 = (pc)^2 + (mc^2)^2$ reads like a Pythagorean theorem for energy. For a slow particle ($p \to 0$) it reduces to $E = mc^2$ plus, on Taylor expansion, the Newtonian $\tfrac{1}{2}mv^2$ — classical kinetic energy is just the first correction to the rest energy. For a massless particle like the photon ($m = 0$) it collapses to $E = pc$, which is why light carries momentum despite having no mass. The rest mass $m$ is the invariant “length” of the energy-momentum four-vector: observers disagree on $E$ and $p$ separately but all agree on $m$.

Worked Example: how much energy is locked in one gram?

Mass-energy equivalence says even a stationary object stores energy $E = mc^2$. For $m = 1\ \text{gram} = 10^{-3}\ \text{kg}$:

$$E = (10^{-3}\ \text{kg})(3.0\times10^{8}\ \text{m/s})^2 = 9\times10^{13}\ \text{J}.$$

That is roughly the energy released by 20 kilotons of TNT — comparable to the Hiroshima bomb — from a single gram of matter. The reason chemistry never reveals this is that chemical bonds release a billionth of the rest energy; only nuclear and particle processes tap a meaningful fraction. The mass of a charged battery, a compressed spring, or a hot object is genuinely (if immeasurably) larger than its de-energized state.

Four-Vectors and Tensor Notation

In special relativity, we use four-vectors to unify space and time:

Position four-vector:

\[x^\mu = (ct, x, y, z)\]

Four-momentum:

\[p^\mu = (E/c, p_x, p_y, p_z)\]

Four-velocity:

\[u^\mu = \gamma(c, v_x, v_y, v_z)\]

Invariants:

Spacetime interval: $s^2 = -c^2t^2 + x^2 + y^2 + z^2$
Rest mass: $m^2c^2 = -p^\mu p_\mu / c^2$

Tensor notation conventions. Contravariant indices are written upper ($x^\mu$), covariant indices lower ($x_\mu$), and repeated indices are summed (Einstein summation). The full tensor machinery — covariant derivatives, the Lorentz algebra, spinors — is collected in the Graduate Formalism & Frontiers page.

Continue

Up: Relativity — overview and navigation hub. Next: General Relativity — gravity as the curvature of spacetime.