Conserve curvature only when its hypotheses hold
Separate uniform-density geometry, entropy pressure, gradient terms, and clock degeneracies.
On this page
The question for today
What you’ll understand
- Construct the curvature perturbation on uniform-density slices.
- Identify the nonadiabatic pressure source in its large-scale evolution.
- State the regularity and scale assumptions behind superhorizon conservation.
A useful primordial variable should retain information while a mode is much larger than the expansion scale. The curvature perturbation on uniform-density slices often does so, allowing an early calculation to be connected to later radiation and matter evolution. Its conservation is conditional. Entropy pressure, non-negligible gradients, or a singular choice of time label can spoil an argument that simply says outside the horizon means frozen. The geometric definition and the evolution equation should therefore be checked together rather than inferred from a plot of one gauge-dependent potential.
Define the geometry of a density slice
Use ζ=−ψ−ℋδρ/ρ′ with the preceding metric signs. A passive time shift to uniform density removes δρ while changing ψ; its negative is precisely ζ. At first order the spatial metric on that slice has the scalar conformal factor one plus twice ζ after a suitable spatial threading choice. Equivalently, its long-wavelength local scale factor is a times expζ. Positive ζ in this convention means a locally enlarged spatial scale, not a positive Newtonian lapse by definition.
The background continuity equation gives ρ′=−3ℋ(ρ+p) for a conserved total stress tensor. Thus a barotropic fluid with constant w has ζ=−ψ+δ/[3(1+w)], where δ is its density contrast in the original slicing. This rewrite assumes the background evolves and ρ+p is nonzero. A pure cosmological constant has no usable uniform-density clock of this kind, so dividing by its vanishing density derivative cannot define an independent propagating scalar mode.
Derive the source from energy conservation
Define the adiabatic derivative ca²=p′/ρ′ and split the pressure perturbation into ca²δρ plus a remainder δpnad. This remainder is gauge invariant because the background time-shift pieces cancel. It measures failure of the perturbed pressure to follow the background pressure-density relation; ca² is not automatically the physical propagation speed of every field or mixture. Multiple components can create such a remainder even if each component separately has a simple equation of state.
Rewriting the perturbed energy-conservation equation yields ζ′=−ℋδpnad/(ρ+p) plus spatial-gradient terms. On scales k/ℋ much smaller than one, those terms can be neglected only for regular perturbations whose velocity/shear variables do not compensate the small gradients. With negligible nonadiabatic pressure and that regular large-scale limit, ζ is approximately constant. This energy-conservation argument is broader than one specific gravitational field equation, but still presupposes a meaningful background-plus-perturbation expansion.
Distinguish a Hubble scale from a causal theorem
The phrase superhorizon is commonly used here for k≪aH, although the Hubble radius is not identical to every particle or event horizon. A comoving curvature variable can coincide with ζ at leading large-scale order for regular adiabatic solutions under the relevant constraints; its sign depends on how that variable was defined. A non-attractor scalar background can make a comoving-curvature mode evolve, so one must not replace the stated conservation hypotheses with the phrase single field alone.
The heat lab illustrates the importance of a rate hierarchy rather than primordial dynamics. Reset and set mode n=1 and diffusivity κ=0.1, then Step twice to time 1; predict amplitude exp(−0.1). Reset and set n=1,κ=1, then Step twice; predict exp(−1). A slowly varying mode can look nearly constant over a short interval without an exact conservation law. This analytic periodic diffusion model has no expansion, density slicing, entropy pressure, or inflationary curvature equation.
Work it through
Check radiation curvature across a slicing change
For radiation, use ψ=0.002 and δρ/ρ=0.012 at one time. Compute ζ. Then shift the time label by ℋT=0.004 and verify ζ using the new density and curvature coefficients.
Radiation has w=1/3, so ζ=−ψ+δ/4=−0.002+0.003=0.001. The factor four comes from the radiation background dilution law.
The transformed curvature is ψtilde=0.006. Because ρ′/ρ=−4ℋ, the transformed density contrast is δtilde=0.012+4×0.004=0.028.
The recombined value is −0.006+0.028/4=0.001. This verifies gauge invariance at one time; conservation between different times would additionally require the pressure and gradient conditions.
The same event has ζ=0.001 in both slicings, despite different density contrasts.
The living notebook
Explore the model
Interactive lab · heat
Distinguish slow change from conservation
Reset and set n=1,κ=0.1, then Step twice to t=1; predict amplitude exp(−0.1). Reset and set n=1,κ=1, then Step twice; predict exp(−1).
- Current amplitude
- 1
- Spatial L² norm
- 1.772
- Decay rate · κn²
- 3.6 /s
The exact periodic heat-equation mode is u(x,t) = exp(−κn²t) sin(nx), on [0,2π]. Playback advances physical time. The initial sine is shown separately, and the unnormalized L² norm is √π exp(−κn²t). This analytic instrument does not run a finite-difference solver or test discretization stability. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forThis exact periodic heat mode is a rate comparison only. It contains no curvature perturbation, inflationary expansion, or entropy source.
Pause & reason
Check your understanding
Your turn
Try it yourself
Neglect gradients and suppose δpnad/(ρ+p)=0.002 stays constant for three background e-folds. Starting from ζ=0.010, find the final ζ and state why being outside the Hubble scale did not freeze it.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Integrate dζ/dN=−0.002 rather than setting its source to zero.
Compare with the solution
The change is −0.002×3=−0.006, so the final value is 0.004. The nonadiabatic pressure remains a source even after gradient terms have been neglected. This calculation assumes the supplied small constant source is dynamically realizable over the interval; it is not a claim that every entropy mode is constant or that a particular inflationary model produces it.
Keep the idea
Take it with you
- Uniform-density curvature has a declared spatial sign and requires a usable density clock.
- Large-scale conservation follows after checking entropy and gradient/shear sources.
- Gauge invariance at one event and conservation through time are different statements.
Read further
Original sources and deeper explanations to follow at your own pace.
- Malik and Wands — Cosmological perturbations, §§7.6,8.2.2 and 9: uniform-density curvature and its conservation sources
- Daniel Baumann — TASI Lectures on Inflation, §9.2.4: superhorizon evolution and its qualifications
- Namjoo, Firouzjahi and Sasaki — Violation of non-Gaussianity consistency relation in a single field inflationary model: non-attractor large-scale mode
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.