The Four-Wavevector and Phase Invariance

Statement

For a monochromatic plane wave the phase φ = ωt − k·r is a Lorentz scalar. Consequently the angular frequency and the wavevector assemble into a single four-vector k^μ = (ω/c, k), and for light propagating in vacuum, where ω = c|k|, this four-vector is null: k^μk_μ = ω²/c² − |k|² = 0.

Why it matters

The phase of a wave counts crests. Between two spacetime events, the number of wavefronts that pass is a physical fact that every observer must agree on — you cannot Lorentz-transform away a crest. Turning that intuitive statement into a covariant object is what forces frequency and wavevector to travel together, and it hands us the relativistic Doppler effect and stellar aberration as two components of one transformation law.

The null condition is the deepest payoff. It is the wave-mechanical shadow of the light cone: a plane wave whose four-wavevector is lightlike propagates at exactly c. Multiplying by ℏ turns k^μ into the photon four-momentum, so k^μk_μ = 0 is the same equation as “the photon is massless.”

Assumptions

Derivation

Result

Limiting cases

• Low velocity (β → 0): the Lorentz transform of k^μ reduces to the classical (non-relativistic) Doppler shift ω′ ≈ ω(1 − β cos θ), with the transverse (θ = 90°) shift disappearing at this order.
• Transverse observation (θ = 90°): a purely relativistic redshift ω′ = ω/γ survives with no classical counterpart — time dilation seen directly in the frequency.
• Medium, n → 1: the phase four-vector k^μ = (ω/c, nω/c n̂) approaches the null vacuum case as n → 1⁺; for n > 1 its norm is negative (spacelike four-vector).
• Quantum limit: multiply by ℏ to get p^μ = ℏk^μ; the null condition becomes E² = p²c², the energy–momentum relation of a massless particle.

Breaks when

Failure modes

• Frequency in the time slot. Writing k^μ = (ω, k) instead of (ω/c, k): the components then have mismatched units and the norm is dimensionally wrong. The factor of c is not cosmetic — it pairs with ct in x^μ.
• Sign/metric slips. Using diag(−,+,+,+) but keeping φ = ωt − k·r without adjusting k_μ, giving a spurious sign in the contraction. Fix the metric and the raised/lowered forms together.
• Confusing phase invariance with amplitude invariance. The amplitude A is not a scalar (it transforms with the fields); only the phase φ is. Students assume “the wave looks the same” — it does not, only the crest-count does.
• Believing phase velocity is a physical speed limit. ω/|k| can exceed c in a medium without violating relativity; this tempts a wrong claim that k^μ must be null in a medium too.
• Treating a wave packet as one mode. Applying the null condition to a pulse and concluding it travels at exactly c, ignoring group velocity and dispersion.

Discussion

The logical engine here is the quotient theorem: if a contraction k^μx_μ is a scalar for every four-vector x^μ, then k^μ must itself be a four-vector. We never had to know how ω and k transform — we deduced it from a single physical input, the invariance of phase. This is a recurring move in relativity: promote a known scalar, pair it with a known four-vector, and let covariance do the bookkeeping. The same reasoning turns proper time into four-velocity and charge density into the four-current.

Because k^μ is a four-vector, the entire relativistic optics of point sources follows from one Lorentz transformation. The time component gives the Doppler shift; the direction of the spatial part gives aberration. Historically these were separate empirical laws; here they are two faces of the same rotation-boost acting on a null vector. The null constraint is preserved by the transformation, which geometrically is why a boost cannot bring light to rest — there is no frame on the light cone’s vertex.

The null character of k^μ is intimately tied to gauge invariance and masslessness in field theory. Promoting p^μ = ℏk^μ, the on-shell condition p^μp_μ = m²c² = 0 identifies the photon as massless, and the little group of a null momentum is ISO(2) rather than SO(3), which is precisely why the photon carries two helicity states rather than three spin projections. That a classical crest-counting argument reaches into the representation theory of the Poincaré group is one of the quiet unifications in physics: the geometry of a plane wave already knows about the spin of light.

Common misconceptions. (i) “Frequency is absolute” — no; only phase is invariant, and frequency is merely its time component, frame-dependent. (ii) “The four-wavevector is null for all waves” — only in vacuum; in matter it is timelike. (iii) “Null means zero” — a null four-vector has vanishing Minkowski norm but nonzero components; it is very much not the zero vector.

Worked examples

Example 1 — Longitudinal relativistic Doppler shift. A source at rest in frame S emits light of frequency f = 5.00 × 10¹⁴ Hz travelling in the +x direction. An observer’s frame S′ moves at β = v/c = 0.500 in the same +x direction (receding from the source). Find the observed frequency.

Example 2 — Null four-vector and photon four-momentum. Verify that a 500 nm vacuum plane wave has a null four-wavevector, then compute the photon four-momentum and confirm E = pc. Take c = 2.998 × 10⁸ m s⁻¹, ℏ = 1.055 × 10⁻³⁴ J s.

Problems

1. A star recedes from Earth at β = 0.100. A spectral line emitted at 656.3 nm (hydrogen-α) is observed. Using the longitudinal Doppler formula from the four-wavevector, find the observed wavelength.
   Solution

   λ′ = λ √[(1 + β)/(1 − β)] for recession (wavelength stretches, inverse of the frequency factor). = 656.3 nm × √(1.100/0.900) = 656.3 × √(1.2222) = 656.3 × 1.1055 = 725.5 nm. A redshift of about 69 nm into the near-infrared, consistent with Δλ/λ ≈ β = 0.10 at leading order.

2. Show explicitly that for a boost of speed v along x, the transformed four-wavevector of a light wave initially along +x is still null, i.e. (ω′/c)² − k′_x² = 0.
   Solution

   Boost: ω′/c = γ(ω/c − β k_x), k′_x = γ(k_x − β ω/c). With k_x = ω/c: ω′/c = γ(ω/c)(1 − β) and k′_x = γ(ω/c)(1 − β). The two are equal, so (ω′/c)² − k′_x² = 0. Null is preserved — as it must be, since k^μk_μ is a Lorentz scalar.

3. Light travels through glass of refractive index n = 1.50, so the phase four-vector is k^μ = (ω/c, nω/c n̂). Compute k^μk_μ in terms of ω/c and classify the four-vector.
   Solution

   k^μk_μ = (ω/c)² − (nω/c)² = (ω/c)²(1 − n²) = (ω/c)²(1 − 2.25) = −1.25 (ω/c)². The norm is negative, so the four-vector is spacelike in the (+,−,−,−) convention. There exists a frame in which the wave is a pure standing spatial oscillation — the medium picks out a rest frame, breaking the vacuum light-cone structure.

4. A gamma-ray photon has energy 1.00 MeV. Compute its angular frequency ω, wavenumber |k|, and verify the four-wavevector is null. (1 eV = 1.602 × 10⁻¹⁹ J, ℏ = 1.055 × 10⁻³⁴ J s.)
   Solution

   E = 1.00 × 10⁶ × 1.602 × 10⁻¹⁹ = 1.602 × 10⁻¹³ J. ω = E/ℏ = 1.602 × 10⁻¹³/1.055 × 10⁻³⁴ = 1.519 × 10²¹ rad s⁻¹. |k| = ω/c = 1.519 × 10²¹/2.998 × 10⁸ = 5.067 × 10¹² rad m⁻¹. Then (ω/c)² − |k|² = (5.067 × 10¹²)² − (5.067 × 10¹²)² = 0. Null, as required for vacuum propagation.

5. Starting from the invariance of phase alone, argue why two co-located observers in relative motion, both looking at the same star, generally disagree on both its colour and its position in the sky — and identify which components of k^μ encode each disagreement.
   Solution

   Phase invariance forces k^μ = (ω/c, k) to be a four-vector, so a boost mixes its components. The time component ω/c transforms under the boost — a change in frequency, hence colour (Doppler). The direction of the spatial part k rotates because its transverse and longitudinal pieces transform differently (k′_⊥ = k_⊥, k′_∥ = γ(k_∥ − βω/c)), tilting the apparent arrival direction — that is aberration (position). Both effects are the single geometric fact that a Lorentz transformation slides the null vector k^μ along the light cone; they cannot be separated at a fundamental level because they are components of one object. Only the phase itself — the crest count — is agreed upon.