Past exam of the mathematics course of the University of Cambridge 2016 iii Paper 310 2 c Solution Created 2026-10-03 Updated 2026-10-06
Because density is a scalar field, . Write both sides as background plus perturbation and expand the background time shift:The coordinate shift acting on an already first-order perturbation contributes only at second order. Thus the density perturbation gauge transformation isThe uniform-density curvature perturbation in the paper's sign convention isThe derivative on is essential and is present in the PDF; the TeX transcription drops it. The scalar-curvature combination transforms aswhile the density term changes by . Hence the two time-slicing changes cancel:On a uniform-density slice, and this variable is the signed spatial-curvature perturbation. The construction assumes ; a pure constant-density cosmological constant does not define such a time slicing.
In Newtonian gauge in cosmology, and . The background cosmological perfect-fluid continuity equation gives , so, writing ,Set the cosmological adiabatic sound speed . Then . Differentiate the previous expression and insert the perturbed energy-conservation equation:Consequently the curvature evolution equation in this sign convention isThe last equality defines the total energy-frame velocity by . For adiabatic cosmological perturbations, the non-adiabatic pressure perturbation vanishes. On superhorizon scales, the stated suppression of the velocity divergence makes the remaining gradient term negligible. Thus is conserved to leading order in . The positive sign of the pressure source follows from the paper's definition of and must not be replaced by a formula using the opposite curvature convention.
Superhorizon conservation of uniform-density curvature lets one carry a primordial perturbation from its inflationary generation through otherwise complicated eras and predict the initial conditions for later Cosmic microwave background anisotropy and structure growth. It relies on adiabaticity: an isocurvature perturbation can source curvature evolution, so the same conclusion is not automatic in an arbitrary multifield model.