Luminosity Distance

A distance you can read off brightness

Most distances in astronomy are infuriating to measure. You cannot lay a tape measure across the cosmos, and parallax — the gold standard — runs out of precision beyond a few thousand parsecs. But there is one trick that reaches across billions of light-years: if you already know how intrinsically bright an object is, you can measure how bright it looks, and the difference tells you how far away it is. That is the whole idea behind luminosity distance.

The definition is built directly on the inverse-square law. Light from a source spreads out over an expanding sphere; double the distance and the same energy is smeared over four times the area, so the flux you collect drops by a factor of four. Write that as an equation and you get the defining relation:

F = L / (4π d_L²)

Here F is the energy flux you measure (watts per square meter), L is the source's intrinsic luminosity (watts), and d_L is the luminosity distance — by definition, whatever value makes the inverse-square law come out right. In a static, flat, Euclidean universe, that value is just the true distance. The interesting physics is everything that happens when the universe is none of those things.

What expansion does to the light

In an expanding universe, two extra effects degrade the flux on top of the ordinary 1/area dilution, and both scale with one power of (1+z), where z is the redshift.

Energy loss per photon. Every photon is stretched by the expansion of space. A photon emitted with wavelength λ arrives with wavelength (1+z)λ, so its energy E = hc/λ is reduced by a factor (1+z). The light literally arrives carrying less energy than it left with.

Arrival-rate time dilation. The expansion also stretches the interval between successive photons. If the source emits one photon per second, the cosmic stretching means you receive one photon every (1+z) seconds. The rate at which energy lands on your detector is therefore reduced by another factor (1+z). This is the same time dilation that makes distant supernova light curves evolve more slowly — a high-z supernova that fades in 20 days intrinsically takes 20(1+z) days to fade as seen from Earth, an effect measured directly and a beautiful confirmation that the redshift is genuine expansion, not tired light.

Flux is energy per photon times photon rate, divided by collecting area. The area term, for a spatially flat universe, is set by the comoving distance d_C: the proper area of the sphere over which the photons are spread today is 4π d_C². Combining all three pieces:

F = L / [ 4π d_C² (1+z)² ]

Comparing this to the definition F = L / (4π d_L²) and reading off the luminosity distance gives the headline result for a flat cosmology:

d_L = (1 + z) · d_C        (flat universe)

The luminosity distance is the comoving distance multiplied by a single, clean factor of (1+z). Both factors of (1+z) live inside the flux; one of them gets "absorbed" into d_C — because the comoving distance already accounts for the geometric area — leaving exactly one power of (1+z) in d_L itself. (For a non-flat universe, d_C is replaced by the transverse comoving distance D_M, but the (1+z) prefactor is unchanged.)

Where the comoving distance comes from

The comoving distance d_C is the part of luminosity distance that carries all the cosmological model dependence. It is the integral of the light-travel increment c/H(z) over the redshift the photon traversed:

d_C = ∫₀ᶻ c / H(z') dz'

H(z) = H₀ √[ Ω_m (1+z)³ + Ω_Λ + Ω_r (1+z)⁴ ]   (flat ΛCDM)

The Hubble parameter H(z) depends on the present-day density fractions: Ω_m for matter, Ω_Λ for dark energy (the cosmological constant), and Ω_r for radiation. A universe with more matter decelerates faster, packing distant objects closer in comoving distance; a universe with more dark energy lets the comoving distance — and hence d_L — grow larger at a given z. This is exactly the lever that supernova surveys pull on. The full luminosity distance is therefore

d_L(z) = (1+z) · ∫₀ᶻ c / H(z') dz'

and the d_L versus z curve rises steeply, faster than linearly, because the (1+z) prefactor multiplies an integral that is itself growing.

Worked example: a Type Ia at z = 0.5

Take a Type Ia supernova at z = 0.5 in a standard flat ΛCDM cosmology with H₀ = 70 km/s/Mpc, Ω_m = 0.3, Ω_Λ = 0.7. The Hubble distance is the natural scale:

D_H = c / H₀ = (3×10⁵ km/s) / (70 km/s/Mpc) = 4283 Mpc

Numerically integrating c/H(z) from 0 to 0.5 gives a comoving distance of d_C ≈ 1920 Mpc. The luminosity distance is then

d_L = (1 + 0.5) × 1920 Mpc ≈ 2880 Mpc

Now translate that into what an observer actually measures. A Type Ia near peak has an absolute magnitude M ≈ −19.3. The distance modulus is

μ = 5 log₁₀(d_L / 10 pc) = 5 log₁₀(d_L / Mpc) + 25
  = 5 log₁₀(2880) + 25 ≈ 42.3

m = M + μ = −19.3 + 42.3 = 23.0

So this supernova peaks at apparent magnitude ≈ 23 — visible to an 8-meter telescope but hopelessly beyond the naked eye. Here is the crux: in a decelerating, matter-only universe (Ω_m = 1, Ω_Λ = 0) the same redshift gives a smaller luminosity distance, about 2500 Mpc, and a brighter predicted magnitude m ≈ 22.7. The 1998 supernova teams measured the fainter value. Distant supernovae were systematically ~0.25 magnitudes dimmer — about 12% farther in d_L — than the decelerating prediction. The only way to push d_L outward at fixed z is to let the expansion accelerate: dark energy.

The supernova Hubble diagram

The plot that decided the matter is the Hubble diagram: distance modulus μ on the vertical axis, redshift z (usually log z) on the horizontal axis. Because μ = 5 log₁₀(d_L) + 25, the curve is just luminosity distance in disguise, scaled logarithmically. At low z it is a straight line — Hubble's law, cz ≈ H₀ d_L — but its curvature at z ≳ 0.3 encodes the cosmology.

Expanding d_L for small z makes the leverage explicit:

d_L ≈ (c z / H₀) [ 1 + ½ (1 − q₀) z + ... ]

The first deviation from the linear Hubble law is governed by the deceleration parameter q₀ = Ω_m/2 − Ω_Λ. In a matter-only universe q₀ = +0.5 (decelerating); in our universe q₀ ≈ −0.55 (accelerating). The sign flip is precisely what the supernova residuals revealed. Today, samples like Pantheon+ (over 1500 spectroscopically confirmed Type Ia supernovae) trace the d_L–z curve from z ≈ 0.01 out to z ≈ 2.3, pinning the dark-energy equation of state to w = −1.03 ± 0.03 — consistent with a cosmological constant.

The family of cosmological distances

Luminosity distance is one member of a family of distances that all coincide in a static Euclidean universe and diverge once expansion enters. They are tied together by the (1+z) factors and by the Etherington reciprocity relation, which holds in any metric gravity theory with photon-number conservation:

d_L = (1+z) · D_M           (D_M = transverse comoving distance)
d_A = D_M / (1+z)           (angular diameter distance)
⇒  d_L = (1+z)² · d_A        (Etherington reciprocity)

Distance	Defined by	Flat-universe form	Behaviour at high z
Comoving d_C	Fixed coordinate grid expanding with the universe	∫ c/H dz	Grows, asymptotes to particle horizon
Luminosity d_L	Flux: F = L/(4π d_L²)	(1+z) d_C	Grows fastest — steepest curve
Angular diameter d_A	Angular size: θ = size / d_A	d_C / (1+z)	Peaks near z ≈ 1.6, then decreases
Proper (now)	Ruler distance at the present epoch	d_C (flat)	Equal to comoving for flat universe
Light-travel	c × lookback time	∫ c/[(1+z)H] dz	Bounded by age of universe × c
Hubble distance	c / H₀ (scale only)	4283 Mpc (H₀=70)	Constant reference length

The startling entry is the angular diameter distance d_A. Because of the (1+z) in its denominator, d_A actually turns over: beyond z ≈ 1.6 a galaxy of fixed physical size looks angularly larger the farther away it is. At z = 1, d_L is already four times d_A. This is not a paradox — it reflects the fact that we see the distant object as it was when the universe was smaller and the light had less far to spread.

Standard candles and the distance ladder

Luminosity distance is only useful if you know L. The objects for which we do are called standard candles, and they form a calibrated ladder:

• Cepheid variables. The period of a Cepheid's pulsation predicts its luminosity (the Leavitt law), making them standard candles out to ~30 Mpc with HST. They are the rung that calibrates the supernovae.
• Tip of the red giant branch (TRGB). The peak luminosity of the brightest red giants is nearly fixed (M_I ≈ −4.0), giving an independent calibration that sidesteps Cepheid crowding.
• Type Ia supernovae. The workhorse for cosmology. After standardizing peak brightness using the light-curve width (Phillips relation: broader light curves are brighter) and color, their scatter shrinks to ~0.12–0.15 mag, about 6–7% in d_L. A single Type Ia can be seen to z > 1.5.
• Standard sirens. Gravitational-wave events like the binary neutron-star merger GW170817 deliver d_L directly from the waveform amplitude — no calibration ladder needed — providing a completely independent route to the Hubble constant.

The supernovae must be tied to Cepheids or TRGB in nearby galaxies that host both, anchoring the absolute scale; only then does d_L from a distant supernova become a calibrated number rather than a relative one.

Common pitfalls and misconceptions

• Thinking d_L is a "real" distance. At z = 1, d_L is roughly 6600 Mpc in standard ΛCDM, but the object's actual proper distance today is only about 3300 Mpc. Luminosity distance is a flux-bookkeeping device, not a ruler length — it is inflated by the (1+z) factor precisely to keep the inverse-square law formally true.
• Using only one factor of (1+z) for flux. A frequent error is dimming the source by a single (1+z) (energy only) and forgetting the arrival-rate time dilation. The flux dilution is (1+z)², which is why d_L = (1+z) d_C carries one factor and the bolometric flux carries two.
• Confusing d_L with d_A. They differ by (1+z)². Use d_L for flux/magnitudes; use d_A for angular sizes, gravitational-lens geometry, and the baryon-acoustic-oscillation standard ruler. Plugging d_L into an angular-size calculation overestimates physical sizes by (1+z)².
• Forgetting the K-correction. A redshifted spectrum shifts flux out of your observing band. Converting a band-limited magnitude to the d_L relation requires a K-correction; ignoring it biases distances, especially at high z.
• Assuming the (1+z) prefactor is model-dependent. It is not — the (1+z) is pure kinematics/geometry. All the cosmology lives in the comoving-distance integral d_C, through H(z).
• Treating Type Ia supernovae as identical out of the box. Raw peak magnitudes scatter by ~0.4 mag. Only after the light-curve-shape and color standardization do they become the precise candles the Hubble diagram relies on.

Observational status

Luminosity distance is the backbone of modern observational cosmology. The Pantheon+ compilation and the Dark Energy Survey 5-year supernova sample have driven the statistical precision on the dark-energy equation of state below a few percent. Standard sirens give a parallel, fully independent d_L measurement: GW170817 yielded H₀ ≈ 70 km/s/Mpc with no distance ladder at all. These independent d_L routes have sharpened the Hubble tension — local d_L-ladder measurements (SH0ES, H₀ ≈ 73) disagree at ~5σ with the value inferred from the early-universe physics of the cosmic microwave background (H₀ ≈ 67). Whether that tension is a systematic in some rung of the ladder or a crack in ΛCDM is one of the central open questions in cosmology, and every line of attack flows through how we measure luminosity distance.