Primordial black holes from inflation: on the decoupling between large and small scales

This paper demonstrates that within single-field inflation models, large-scale modes constrained by CMB observations decouple from small-scale enhancements required for primordial black hole production, ensuring that 1-loop back-reaction effects remain unobservable and do not disrupt large-scale predictions.

The Gist

The Big Picture: Tiny Black Holes and the "Echo" Problem

Imagine the early universe as a giant, smooth ocean. During a period called "inflation," this ocean expanded incredibly fast. Usually, the waves on this ocean are tiny and gentle. However, to create Primordial Black Holes (PBHs)—tiny black holes that formed right after the Big Bang—you need a few massive, rogue waves.

To get these rogue waves, the rules of the ocean had to change briefly. The universe had to switch from a smooth, predictable expansion to a chaotic, "ultra-fast" expansion for a moment, creating a huge spike in wave activity on very small scales (where the black holes form).

The Problem:
Scientists were worried about a "ripple effect." If you create a massive wave on a small scale, does it send a shockwave back to the rest of the ocean? In physics terms, they worried that the intense activity creating the black holes might "back-react" and mess up the statistics of the huge, gentle waves we see today (which we measure using the Cosmic Microwave Background, or CMB).

If this back-reaction were real, it would mean our current understanding of the early universe is broken, because the tiny black holes would have ruined the big picture.

The Investigation: The "Separate Universe" Tool

The author, L. Iacconi, and their colleagues wanted to check if this "echo" from the small waves actually ruins the big waves.

To do this, they used a clever mental tool called the "Separate Universe" framework.

• The Analogy: Imagine the universe is a giant patchwork quilt. Instead of trying to calculate how every single thread interacts with every other thread at once (which is impossible), you treat each patch of the quilt as its own tiny, separate universe.
• You look at a "long wave" (a big patch) and ask: "How does this patch change if the tiny, chaotic waves inside it shift slightly?"

They used this method to calculate what happens when you add up all the tiny interactions (loops) between the big waves and the small waves.

The Discovery: The "Decoupling"

The paper's main finding is surprisingly comforting: The small waves and the big waves don't actually talk to each other in a way that causes damage.

Here is how they broke it down:

1. The Two Types of "Noise":
   When they did the math, they found two ways the small waves could theoretically mess up the big waves:

   • Type A (Bad Start): The small waves started with a weird "initial condition" that was already messed up.
   • Type B (Bad Evolution): The small waves grew strangely while they were outside the "horizon" (the point where they could communicate with us).

2. The "Total Derivative" Trick:
   When they added up all the contributions from the small waves, they found a mathematical pattern called a "total derivative."

   • The Analogy: Imagine you are walking along a beach and counting how many seashells you pick up. If you only care about the total number of shells you have at the end, it doesn't matter how many you picked up in the middle of the beach. It only matters how many you picked up at the very start and the very end of your walk.
   • In this paper, the "middle" is the huge, chaotic peak of waves that creates the black holes. The math showed that all the messy details of that peak cancel each other out. The only things that matter are the edges of the peak.

3. The Result:
   Because the "middle" cancels out, the intense activity creating the black holes does not change the statistics of the large-scale waves.

   • The big waves (CMB) remain calm and predictable.
   • The small waves (PBHs) can go wild without disturbing the big picture.
   • The author calls this "decoupling." The two scales are like two separate radio stations; one can play heavy metal (black holes) without static interfering with the other playing classical music (CMB).

Why This Matters

• Reassurance: It confirms that we can have a theory where tiny black holes exist without breaking our current models of the early universe. The "tree-level" predictions (the simple, first-order math) are safe.
• The Catch: The author notes this only works if the "long waves" are adiabatic (meaning they are smooth and uniform, like a steady breeze). If the long waves themselves are chaotic or if we are looking at the black holes' own internal corrections, this "decoupling" might not happen. But for the standard scenario of single-field inflation, the universe is safe.

Summary in One Sentence

The paper proves that even if the early universe had a violent, chaotic moment that created tiny black holes, that chaos stays local and doesn't send a shockwave back to ruin the smooth, large-scale patterns we observe in the cosmic background radiation.

Technical Summary

Technical Summary: Primordial Black Holes from Inflation – On the Decoupling between Large and Small Scales

Problem Statement
Primordial Black Holes (PBHs) can be generated during inflation if the primordial curvature power spectrum, $\mathcal{P}_\zeta(k)$, is significantly enhanced on small scales ($k_{PBH}$) compared to the scales probed by Cosmic Microwave Background (CMB) experiments ($k_{CMB}$). In single-field inflation models, this enhancement typically requires a transient departure from slow-roll attractor dynamics, such as an Ultra-Slow-Roll (USR) phase where $\epsilon_2 \simeq -6$.

A critical theoretical concern arises in these scenarios: the potential for "back-reaction." Large-scale modes ($p$), which are tightly constrained by CMB observations, interact with the amplified small-scale modes ($q$) through non-linear couplings. Previous analyses (e.g., Ref. 6) suggested that these interactions could generate significant 1-loop corrections to the large-scale power spectrum, proportional to the peak amplitude of the small-scale spectrum ($\mathcal{P}_\zeta(k_{peak})$). If true, this would imply that the very mechanism required to produce PBHs would disrupt the successful tree-level predictions for large-scale statistics, creating a tension with perturbativity and observational data.

Methodology
This work addresses the back-reaction problem by applying the separate-universe framework and its generalization using multi-point propagators (MPP). The authors aim to compute the 1-loop corrections to the long-mode power spectrum more transparently than the standard In-In formalism used in prior literature.

The methodology relies on two fundamental assumptions:

1. Adiabaticity of the long mode: The long-wavelength mode $\zeta_p$ is frozen shortly after horizon crossing.
2. Separation of scales: There is a significant hierarchy between the long mode ($p$) and the enhanced short modes ($q$), such that $p \ll q$.

The calculation proceeds by:

• Initial Conditions: Defining the curvature perturbation $\zeta$ via the $\delta N$ formalism, where $\zeta$ is the variation in the duration of inflation ($\delta N$) across different patches. The initialization time $t_i$ is chosen carefully to ensure all relevant scales (including the transition scale $k_{tr}$ into the non-attractor phase) are super-horizon, allowing gradients to be neglected.
• Loop Expansion: Decomposing the 1-loop contribution to the 2-point function $\langle \zeta_p \zeta_{-p} \rangle$ into two distinct categories based on the $\delta N$ expansion:
  • 11-type loops: Arising from 1-loop corrections to the initial conditions of phase-space variables, linearly evolved to the final time.
  • $\delta N$ loops (12, 13, 22 types): Arising from the non-linear evolution of tree-level initial conditions by the separate-universe dynamics.
• Multi-point Propagators: For 11-type loops, the authors employ a multi-point propagator expansion anchored at an earlier time $t_*$ (during the first slow-roll phase) to handle the evolution through the non-attractor phase without relying on the separate-universe approximation for the initial conditions themselves.

Key Contributions and Results
The paper demonstrates that the 1-loop back-reaction on large scales is not determined by the detailed properties of the small-scale peak (such as its amplitude), but rather by boundary terms in the momentum integration.

1. Decomposition of Back-Reaction:

   • Volume-Suppressed Terms: The 22-type $\delta N$ loops are suppressed by the volume factor $(p/q)^3$ due to the averaging of uncorrelated Gaussian noise over large scales. These are neglected.
   • Non-Volume-Suppressed Terms: The 12- and 13-type loops (representing squeezed couplings) and the 11-type loops (initial condition corrections) are the primary focus.

2. The Total Derivative Structure:
   Under the assumptions of adiabaticity and scale separation, the authors show that the integrands for both the 11-type loops and the $\delta N$ loops (12+13) can be expressed as total derivatives with respect to the logarithm of the short-scale momentum $q$.
   $$\mathcal{P}_\zeta(p)_{\text{loops}} \propto \int_{q_{min}}^{q_{max}} d \ln q \, \frac{d}{d \ln q} [\dots]$$
   Consequently, the integral reduces to contributions solely from the boundaries ($q_{min}$ and $q_{max}$).

3. Independence from Peak Amplitude:
   The resulting back-reaction depends on the difference between the power spectrum at the boundaries of the integration range (e.g., $\mathcal{P}_\zeta(q_{max}) - \mathcal{P}_\zeta(q_{min})$). Crucially, this result is scale-invariant and does not depend on the amplitude of the peak $\mathcal{P}_\zeta(k_{peak})$.

   • For USR models, the back-reaction is proportional to the equilateral non-Gaussianity amplitude $f_{NL,eq}$ and the difference in power spectra across the transition, not the peak height.
   • Since the effect is degenerate with UV modes that were still in the vacuum at the end of inflation, the authors conclude that the 1-loop back-reaction is not observable.

4. Decoupling:
   The primary result is that an adiabatic long mode decouples at 1-loop from the collective effect of enhanced small-scale perturbations. Therefore, the production of PBHs via single-field inflation with a transient non-attractor phase does not disrupt the large-scale predictions consistent with CMB observations.

Significance and Scope
The paper claims to resolve the tension between PBH production and large-scale cosmological constraints within single-field inflation. By utilizing the separate-universe framework, the authors provide a transparent organization of the calculation that confirms the decoupling of scales.

However, the authors explicitly note the limitations of their results:

• Self-Corrections: The results do not apply to the "self-corrections" of the peak scales themselves (where $p \sim q$), which are exponentially sensitive to the peak characteristics and crucial for PBH abundance calculations.
• Non-Adiabatic Long Modes: The decoupling relies on the long mode being adiabatic. In multi-field models, where the long mode may be non-adiabatic, this result may not hold, potentially allowing observable back-reaction.

In summary, the work establishes that for single-field models with adiabatic long modes, the large-scale statistics remain robust against the non-linear back-reaction from the small-scale enhancements required for PBH formation.