The inflationary model of Alan Guth explains the uniformity of the cosmic microwave background and the formation of large-scale structure, predicting a nearly scale-invariant perturbation spectrum. According to Einstein's field equations, these perturbations grow over time, and on small scales quantum fluctuations can become dominant. In this regime, rare but extreme density spikes can collapse into primordial black holes — objects considered as candidates for dark matter and sources of gravitational waves. However, standard methods based on perturbation theory are inapplicable to strongly nonlinear fluctuations. Stochastic inflation offers a non-perturbative approach, treating expansion as a random process, but until now it could not directly compute the compaction function — a critical indicator of collapse.
The method is based on the separate-universe picture, where space is partitioned into causally disconnected regions of Hubble radius. In the stochastic tree formalism, the inflaton's evolution in each region is described by a Langevin equation, and tree branching occurs whenever the physical volume of a region doubles. Tree nodes correspond to branching moments, and leaves to the end of inflation. The compaction function is computed through the volume imbalance between neighboring branches: for node $b_L$ it is proportional to the logarithm of the ratio of volumes of its 'uncle' ($b_R$) and right child ($c_R$). This discretization preserves all symmetries of the continuous limit and automatically accounts for the 'cloud-in-cloud' effect. Numerical simulations were performed using the open-source code FOREST, generating populations of up to 10¹⁰ trees.
Depending on the potential parameters, three regimes are identified. In the classical limit (classicality parameter dμ² ≫ 1) perturbations are nearly Gaussian, and black holes form rarely. Their mass function is narrow, and the vast majority of objects are type I: their horizon forms in the standard way after the perturbation re-enters the Hubble radius. Type II (with a non-monotonic radius function in the metric) is exponentially suppressed. In the quantum regime (dμ² ≪ 1) and near the eternal inflation boundary, the picture changes radically: abundant pumping by quantum fluctuations generates heavy tails in the perturbation distribution. The mass function stretches over several orders of magnitude, and the fraction of type II objects sharply increases, often exceeding that of type I. Also important in this regime are 'cloud-in-cloud' effects, where larger-scale collapses engulf smaller ones.
The discovery of type II black holes as the dominant population in the strong quantum regime challenges established ideas about their rarity. Since such objects may have a fundamentally different horizon structure — a throat connecting the inflating bubble to the external universe — their subsequent evolution and observational signatures (e.g., in gravitational waves) require reexamination. The results also emphasize that the non-Gaussianity naturally arising in stochastic inflation cannot be reduced to simple parameterizations like the local $f_{NL}$.
The proposed method opens the way to more realistic inflation modeling, especially scenarios with an ultra-slow-roll phase, where quantum effects are particularly strong. The next step could be constructing a consistent geometry of the entire spacetime based on the tree topology, enabling studies of primordial black hole clustering and their spatial correlations.
The results will impact cosmology, astrophysics, and gravitational-wave astronomy, providing new guidelines for the search for primordial black holes as a component of dark matter and the interpretation of merger signals.
Detailed numerical simulations of type II perturbation collapse in general relativity are needed to determine their final fate — whether they form black holes, wormholes, or something else. In parallel, the FOREST algorithm should be adapted to more complex potentials, including multi-field models.
The work touches on fundamental problems: the nature of dark matter — primordial black holes remain a viable candidate; the quantum-to-classical transition in the early Universe — where exactly stochastic dynamics gives way to deterministic; and the final stage of gravitational collapse of strongly nonlinear perturbations, at the interface with quantum gravity.
This section is addressed to the author of the paper and their colleagues. Neighbouring works were found by meaning-based search across our archive, not by word overlap. We do not review: we show the field around the work.
What the work does
The work employs the stochastic tree formalism for a nonperturbative analysis of primordial black hole formation during inflation. Branching universes are modeled numerically, and the compaction function is computed, allowing a distinction between two types of collapse. It is found that in the quantum regime, the mass function is stretched, and type II black holes become dominant.
Where its strength is
The stochastic tree method provides direct access to the compaction function, automatically accounting for the 'cloud-in-cloud' effect and preserving all symmetries of the continuum limit. This allows for a quantitative comparison of the contributions from different black hole types.
Where it could go next
Worth trying
Why this may matter
If type II black holes indeed dominate in the quantum regime, this could alter estimates of primordial black hole abundance and their contribution to dark matter, as well as provide a new observational signature for testing inflationary models.
Nearby works in our archive
The closest work in the archive sits at 0.67 by meaning — the usual level for our corpus.
The stochastic tree can be visualized as a family tree of space volumes, where each node inherits the 'state' of the parent Hubble patch. Branching occurs when the volume doubles, and the compaction function turns out to be a measure of 'family inequality' between sibling nodes.