First-hand information for everyone
Researchers tested whether a local bump in the inflaton field can seed cosmic inflation when the rest of space is not inflating. Using numerical relativity in spherically symmetric, asymptotically flat spacetimes, they evolved initial field bumps and the full gravitational constraints to see which ones grow into inflation and which do not. Their main result echoes earlier work by Goldwirth and Piran: a fluctuation must span several times the characteristic inflationary length scale, (G ρ_infl)^{-1/2}, for inflation to succeed.
To set up the problem the authors pick a simple inflation model with a quadratic potential V(φ)=½ m^2 φ^2 and a Gaussian initial field profile φ(r)=φ0 exp(−r^2/σ^2). They choose vanishing initial field momentum and impose asymptotically flat boundary conditions so the spacetime becomes ordinary (non‑inflating) far away. The initial gravitational data must satisfy Einstein’s constraint equations, so they solve for the conformal factor of the spatial metric and the extrinsic curvature numerically. To explore different ways that the same energy can source spatial curvature, they introduce a parameter ε that splits the potential energy between intrinsic curvature (shape of the spatial slice) and extrinsic curvature (how the slice sits in spacetime).
Because the constraint equations are nonlinear, the authors find families of allowed initial data only up to a critical overdensity. Below that critical point there are two branches of solutions, labeled “weak‑field” and “strong‑field,” that join at the maximum allowed overdensity. Working in the asymptotically flat setting lets them follow the evolution through strong gravity events such as black‑hole formation and the pinch‑off of small inflating regions, sometimes described as baby universes.
This work matters because many previous simulations used periodic boundary conditions that effectively model closed or tiled universes. Those periodic setups force certain averages of the spatial curvature or extrinsic curvature to take specific values, which can bias whether inflation appears to start. By using asymptotically flat boundaries and a more general treatment of initial curvature profiles, the new simulations test a wider and arguably more generic set of starting conditions.
Important caveats remain. The study assumes spherical symmetry, a particular quadratic potential, and zero initial field momentum, so the thresholds they find may change for other shapes, models, or fully three‑dimensional initial data. The authors also point out that their choice of how to split the energy between intrinsic and extrinsic curvature (the parameter ε) is a practical one; they argue the qualitative behaviour is generic but it is still a choice. Finally, the paper emphasizes model dependence seen in earlier work: models with shorter inflationary plateaus need larger homogeneous patches to inflate.
Overall, the simulations support the view that inflation does not naturally begin from arbitrarily small, localized bumps in the inflaton. Instead, a patch that is several times the characteristic inflationary length scale is needed, and the trusted numerical set‑up and boundary conditions matter when assessing how generic inflation is from given initial data.