Relativistic Doppler Effect

Where the formula comes from

Consider a source emitting at rest-frame frequency f. In the source frame, the period between wave-crests is T₀ = 1/f. Two events: emission of crest n and emission of crest n+1, separated by T₀ in source-frame time and zero in source-frame space. Lorentz-transform these events to the observer frame:

Δt_obs = γ·T₀,    Δx_obs = γ·v·T₀

The observer measures the time between receiving the crests by accounting for the extra propagation distance. If the source velocity makes angle θ_o (in the observer's frame) with the line of sight, the second crest has to travel an extra v·T₀·cos θ_o ·γ at speed c, adding γ·v·T₀·cos θ_o / c to the reception interval:

Δt_reception = γ·T₀ · (1 − β·cos θ_o)

Frequency is the reciprocal:

f' = 1/Δt_reception = f / [γ(1 − β·cos θ_o)]

The bracketed combination of γ and the cos-θ_o factor is what distinguishes relativistic Doppler from its classical cousin. The classical formula has only the (1 − β·cos θ_o) piece; relativity multiplies by an extra 1/γ.

Three canonical cases

Geometry	θ_o	f'/f formula	At β = 0.5	At β = 0.9
Pure approach	0	√((1+β)/(1−β))	√3 ≈ 1.732	√19 ≈ 4.359
Pure recession	π	√((1−β)/(1+β))	1/√3 ≈ 0.577	1/√19 ≈ 0.229
Transverse (received perpendicular)	π/2	1/γ	√(0.75) ≈ 0.866	√(0.19) ≈ 0.436
Classical analog (transverse)	π/2	1 (no shift)	1.000	1.000

The classical-vs-relativistic split is starkest in the transverse case, where classical predicts no shift but relativity gives a non-trivial redshift purely from time dilation.

Worked example — quasar at z = 6

A distant quasar has cosmological redshift z = 6.0, defined by (1 + z) = f_emit / f_obs. If we (incorrectly) treat the redshift as purely kinematic Doppler:

1. (1 + z) = √((1+β)/(1−β)) → 7² = (1+β)/(1−β) → 49(1−β) = 1+β → β = 48/50 = 0.96.
2. Equivalent recession speed v ≈ 0.96c ≈ 2.88 × 10⁸ m/s.
3. γ = 1/√(1−0.96²) = 1/√0.0784 ≈ 3.57. Time-dilation factor.
4. The observed Lyman-α line at rest 121.6 nm appears at 121.6 × 7 = 851 nm (near-infrared).

In reality, cosmological redshift mixes Doppler with metric expansion of space. The kinematic interpretation is an instructive limit but breaks down at large z — there, the relation between z and β depends on the cosmological model.

Ives-Stilwell — the transverse measurement

In 1938 Ives and Stilwell accelerated H₂⁺ and H₃⁺ ions through ~7 kV (β ≈ 0.005) and observed the Balmer-Hα line emitted forward and backward along the beam, plus a third reference at rest. The two longitudinal shifts to first order in β differ from the rest wavelength by ±β·λ. Relativity adds a second-order γ·λ shift to both, in the same direction (redshift). Classical theory predicts no second-order shift.

Defining

⟨λ_obs⟩ ≡ (λ_forward + λ_backward) / 2

the first-order Doppler cancels, leaving a pure transverse shift proportional to β². Ives and Stilwell measured ⟨λ_obs⟩ − λ_rest ≈ (β²/2)·λ_rest to within 1%. The transverse Doppler effect, predicted by time dilation, was confirmed three decades after Einstein wrote down special relativity.

Real-world relativistic-Doppler applications

System	What's measured
Cosmological redshift	Galaxy distances and recession velocities; basis of modern cosmology
Exoplanet radial-velocity	m/s-level Doppler shifts in stellar spectra reveal orbiting planets
Pulsar timing	Binary-pulsar orbital Doppler tests GR to 14 decimal places (Hulse-Taylor, Kramer 2021)
GPS clock corrections	Transverse Doppler offsets satellite clocks by ~7 µs/day
Mössbauer rotor experiments	γ-ray frequency shift in spinning systems verifies SR time dilation
Free-electron lasers	Forward-Doppler-boosted radiation from relativistic electrons
Quark-jet kinematics	Reconstruction of decay vertices uses Lorentz-boosted Doppler-shifted spectra
Compton scattering	Photon-electron kinematics — energy shift is two successive Doppler factors

JavaScript — relativistic Doppler in code

// Relativistic Doppler factor (observer frame)
function dopplerFactor(beta, theta_observer_rad) {
  const gamma = 1 / Math.sqrt(1 - beta * beta);
  return 1 / (gamma * (1 - beta * Math.cos(theta_observer_rad)));
}

// Longitudinal cases
function approachFactor(beta) { return Math.sqrt((1 + beta) / (1 - beta)); }
function recessionFactor(beta) { return Math.sqrt((1 - beta) / (1 + beta)); }
function transverseFactor(beta) { return Math.sqrt(1 - beta * beta); }

console.log(`Approach β=0.5: ${approachFactor(0.5).toFixed(4)}`);   // 1.7321
console.log(`Recession β=0.5: ${recessionFactor(0.5).toFixed(4)}`);  // 0.5774
console.log(`Transverse β=0.5: ${transverseFactor(0.5).toFixed(4)}`);// 0.8660

// Cosmological redshift to β (kinematic interpretation)
function redshiftToBeta(z) {
  const r = (1 + z) ** 2;
  return (r - 1) / (r + 1);
}
console.log(`z = 6 → β ≈ ${redshiftToBeta(6).toFixed(4)}`); // 0.9592

// Wavelength shift for a moving source
function observedLambda(lambdaRest_m, beta, theta_observer_rad = 0) {
  return lambdaRest_m / dopplerFactor(beta, theta_observer_rad);
}
console.log(`Lyman-α @ z=6: ${(observedLambda(121.6e-9, 0.9592) * 1e9).toFixed(1)} nm`); // 851

// GPS satellite: orbital v = 3874 m/s, transverse-Doppler clock rate
function gpsTransverseRateOffset(v_m_s) {
  const c = 3e8;
  const beta = v_m_s / c;
  return -0.5 * beta * beta; // fractional rate (-) → clock slower
}
console.log(`GPS rate offset: ${(gpsTransverseRateOffset(3874) * 86400e6).toFixed(2)} µs/day`);
// ≈ -7.2 µs/day from velocity time dilation

// Compute γ explicitly
function lorentzGamma(beta) { return 1 / Math.sqrt(1 - beta * beta); }
console.log(`γ at β=0.9: ${lorentzGamma(0.9).toFixed(3)}`); // 2.294
console.log(`γ at β=0.99: ${lorentzGamma(0.99).toFixed(2)}`); // 7.09
console.log(`γ at β=0.999: ${lorentzGamma(0.999).toFixed(1)}`); // 22.4

// Ives-Stilwell mean shift (mean of forward and backward = transverse contribution)
function ivesStilwellMeanShift(lambda_rest_m, beta) {
  const lf = observedLambda(lambda_rest_m, beta, 0);        // approach: blueshift
  const lb = observedLambda(lambda_rest_m, beta, Math.PI);  // recession: redshift
  const mean = (lf + lb) / 2;
  return { lambda_forward: lf, lambda_backward: lb, mean, shift: mean - lambda_rest_m };
}
const out = ivesStilwellMeanShift(656.3e-9, 0.005);
console.log(`Mean λ shift: ${(out.shift * 1e12).toFixed(2)} pm`); // ~8 pm — what Ives & Stilwell measured

Where relativistic Doppler matters

• Astronomy and cosmology. Every spectroscopic redshift survey — SDSS, JWST, DESI — uses the relativistic Doppler relation, sometimes combined with metric expansion in modern frameworks.
• Exoplanet detection. Radial-velocity instruments (HARPS, ESPRESSO) detect Doppler signatures down to 0.1 m/s — Earth-class planets around nearby stars.
• Tests of relativity. Modern ion-trap experiments measure γ to parts in 10⁹, far beyond Ives-Stilwell's original 1%.
• Particle physics. Relativistic kinematics of decay photons, Compton scattering, free-electron-laser radiation — Doppler factors of 10⁴ are routine.
• Engineering. GPS, very-long-baseline interferometry, satellite-based time transfer — all carry transverse-Doppler corrections in their clock budgets.
• Pulsar timing arrays. Doppler-induced phase modulation of pulsar signals probes nanohertz gravitational waves (NANOGrav 2023).
• Beam diagnostics. Thomson and inverse-Compton scattering deduce relativistic electron-beam energy from photon Doppler shifts.

Common mistakes

• Using the classical Doppler at relativistic speeds. Beyond β ≈ 0.1 the differences become percent-level; at β = 0.5 they're 30+%.
• Confusing source-frame angle with observer-frame angle. They differ by relativistic aberration: tan(θ_s/2) = √((1+β)/(1−β))·tan(θ_o/2). Using the wrong one slips a γ factor.
• Treating cosmological redshift as pure Doppler. Local Doppler works for nearby galaxies; at z > 1 the metric expansion of space dominates, and z does not map cleanly to a recession velocity.
• Missing the transverse shift. Even when nothing approaches you, a fast-passing emitter is redshifted by 1/γ at closest approach. Forgetting this misses time dilation entirely.
• Sign errors in β. Use β > 0 for the magnitude and cos θ_o for direction — don't double-count.
• Forgetting frequency-energy linkage. A photon's Doppler shift is also an energy shift: E' = ℏω·D, where D is the Doppler factor. Number of photons is invariant.