⚛️ phenomenology

Stochastic Bubble Completion in de Sitter Space: Power Spectrum, Bispectrum, and Infrared Moment Hierarchies

This paper investigates the stochastic completion time of first-order phase transitions in de Sitter space, demonstrating how rare ancient bubbles create an infrared-sensitive hierarchy in the power spectrum and bispectrum that is regulated by finite starting times or temporal localization, ultimately establishing a general framework for understanding curvature perturbations and non-Gaussianity.

Original authors: Hyukjung Kim, Alp Kagan Ozmen, Heeseung Zoe

Published 2026-10-07

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Hyukjung Kim, Alp Kagan Ozmen, Heeseung Zoe

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the earliest moments of our universe, space itself may have undergone a dramatic transformation, shifting from one state of energy to another, much like water freezing into ice. This process, known as a first-order phase transition, did not happen everywhere at once. Instead, it began in scattered pockets where the new state of matter, or "true vacuum," spontaneously formed and began to expand. These pockets are called bubbles. As they grew, they collided and merged, eventually consuming the entire universe. Because this process is driven by quantum randomness, the exact moment a bubble reaches any specific point in space is unpredictable. This creates a patchwork history where different regions of the universe finish their transformation at slightly different times.

Physicists have long been interested in how these timing differences might leave a mark on the cosmos. If the universe expanded rapidly while these bubbles were forming, the timing variations could stretch out and influence the large-scale structure of space, potentially seeding the galaxies we see today. The key question is how the randomness of bubble formation interacts with the rapid expansion of space. Does the expansion smooth out these differences, or does it amplify them into distinct patterns that we might one day detect? Understanding this requires looking at the statistical fingerprints left behind by the bubbles, specifically how the timing of their arrival correlates across vast distances.

A team of researchers has now mapped out these statistical fingerprints, revealing a hidden hierarchy in how the universe's expansion affects the randomness of bubble formation. They focused on a specific type of expansion called de Sitter space, which describes a universe growing at an accelerating rate, similar to the inflationary period of the early universe or the current dark-energy-dominated era. By simulating the formation of bubbles in this expanding environment, they discovered that the expansion creates a unique "ancient-bubble tail." This is a phenomenon where bubbles that formed extremely long ago, in the distant past, grow so large due to the expansion that they can connect points in space that are now incredibly far apart. These giant, ancient bubbles act as a bridge, linking distant regions and creating long-range correlations that would not exist in a static universe.

The researchers found that these ancient bubbles leave a distinct signature in the data, but the strength of this signature depends on how fast the universe is expanding relative to how fast the bubbles are forming. They identified specific thresholds where the behavior of the universe changes dramatically. For the simplest measure of correlation, known as the power spectrum, the influence of these ancient bubbles becomes significant only when the expansion rate reaches a certain fraction of the bubble formation rate. However, for a more complex measure involving three points in space, known as the bispectrum, the universe becomes sensitive to these ancient bubbles at a much lower expansion rate. This means that the three-point pattern is more fragile and reacts to the ancient bubbles sooner than the two-point pattern does.

This difference in sensitivity creates a fascinating window where the universe behaves in a mixed way. In this intermediate zone, the simple two-point pattern might still look smooth and predictable, while the three-point pattern has already begun to show the jagged, long-range influence of the ancient bubbles. The researchers calculated that this creates a specific, predictable variation in the "non-Gaussianity" of the universe—a technical term describing how much the distribution of matter deviates from a perfectly random, bell-curve shape. When the two patterns are out of sync, the universe exhibits a unique scale-dependent signature that could, in theory, be detected in future observations of the cosmic microwave background or gravitational waves.

To ensure their findings were robust, the team also examined what happens when the phase transition does not stretch back infinitely into the past, as in their idealized model, but begins at a specific, finite time. In a realistic scenario, the universe has a starting point for these events. They found that this finite start acts as a natural cutoff, slicing off the influence of the most ancient, giant bubbles. When this happens, the strange, long-range correlations disappear, and the universe returns to a smooth, predictable state where all statistical patterns follow standard rules. This confirms that the unusual effects they observed are indeed caused by the extreme duration of the expansion and the accumulation of ancient bubbles, rather than an error in their calculations.

The study also explored a second way to regulate these ancient bubbles: by making the rate of bubble formation change over time. If the formation of bubbles is concentrated in a specific window rather than happening continuously, the contribution of the ancient bubbles is naturally suppressed. This acts as a smooth regulator, gradually turning off the influence of the distant past without the sharp cutoff of a finite start time. The researchers showed that even with this smooth suppression, the universe passes through a temporary phase where the ancient-bubble tail is visible before settling back into the standard, smooth behavior. This provides a flexible framework for understanding how different physical conditions in the early universe could lead to different observable outcomes.

Ultimately, this work establishes a clear framework for understanding how the timing of cosmic phase transitions fluctuates across space. It clarifies that the expansion of the universe does not just stretch space; it fundamentally alters the statistical nature of randomness within it. The researchers demonstrated that the universe's expansion can generate distinct hierarchies of patterns, where some statistical measures become sensitive to the deep past while others remain unaffected. These findings offer a new way to interpret potential signals from the early universe, suggesting that if we can measure the subtle differences between two-point and three-point correlations, we might be able to tell whether the universe underwent a long, drawn-out phase transition or a quick, sharp one. The results provide a solid theoretical foundation for future searches for the echoes of these primordial bubbles, turning abstract mathematical concepts into concrete predictions for what we might observe in the sky.

Technical Summary: Stochastic Bubble Completion in de Sitter Space

Problem Statement
First-order cosmological phase transitions are intrinsically stochastic processes where the local completion time—the moment a spatial point is first reached by a true-vacuum bubble—forms a random spatial field. While the power spectrum and bispectrum of these completion-time fluctuations are well-understood in non-expanding (Minkowski) spacetime, the impact of de Sitter expansion on the infrared (IR) structure of these statistics remains to be fully characterized. Specifically, the paper addresses how the interplay between bubble growth and exponential expansion generates an "ancient-bubble tail" in the correlators when the nucleation history is extended arbitrarily far into the past. The central problem is to determine how this tail affects the hierarchy of spatial moments, the analyticity of the power spectrum and bispectrum at $k \to 0$ , and the resulting scale dependence of non-Gaussianity, particularly in the context of finite-duration transitions and temporal localization of the nucleation rate.

Methodology
The authors employ a stochastic first-arrival formalism based on Poisson nucleation statistics. The nucleation rate per unit physical four-volume is parameterized as $\Gamma(t) = \Gamma_f \exp(\beta t - \frac{1}{2}\gamma^2 t^2)$ , where $\beta$ sets the transition scale and $\gamma$ controls temporal localization. The dimensionless expansion rate is defined as $h \equiv H/\beta$ .

Formalism: The completion time $t_c(\mathbf{x})$ is treated as a first-arrival time. The survival probability that a point remains in the false vacuum is determined by the integrated nucleation intensity over its past light cone (or bubble influence region). Connected correlators are derived using the inclusion-exclusion principle for overlapping influence regions, leading to explicit integral expressions for the two-point function $C_2(r)$ and three-point function $C_3(r, s, \mu)$ .
Analytical Derivation: The authors analytically derive the asymptotic behavior of these correlators in the limit of large spatial separation ( $r \to \infty$ ) for an eternally extended de Sitter background. This involves identifying the power-law tails generated by bubbles nucleated in the remote past.
Moment Analysis: The convergence of spatial moments $M_{2n}$ (for the power spectrum) and $M_{B,2n}$ (for the bispectrum) is analyzed. The divergence of these moments in the eternal limit dictates the breakdown of the standard analytic IR expansion ( $P_\tau \propto k^3$ , $B_\tau \propto k^6$ ).
Numerical Evaluation: For the three-point function, the full triple-overlap kernel is evaluated numerically to determine the precise logarithmic enhancements and to verify the analytical scaling predictions.
Finite-Duration and Localization: The study extends to transitions with a finite start time $t_i$ and non-zero temporal localization ( $g = \gamma^2/\beta^2 > 0$ ). Numerical scans are performed to track the duration dependence of moment coefficients and the crossover scales between analytic IR regimes and tail-dominated regimes.

Key Contributions and Results

Ancient-Bubble Tail and Moment Hierarchies: In an eternally extended de Sitter background, the completion-time correlators develop a power-law tail $C_2(r) \sim (Hr)^{-1/h} \ln^2(Hr)$ . This tail causes spatial moments to diverge when the expansion rate $h$ exceeds critical thresholds.
- Power Spectrum: The leading zero mode ( $M_0$ ) diverges at $h = 1/3$ , and the first quadratic correction ( $M_2$ ) becomes sensitive at $h = 1/5$ .
- Bispectrum: The hierarchy is more sensitive. The leading zero mode diverges at $h = 1/6$ , and the first quadratic correction becomes sensitive at $h = 1/8$ .
- Logarithmic Enhancements: The two-point tail exhibits a $\ln^2(Hr)$ enhancement, while the three-point tail exhibits a $\ln^4(H\lambda)$ enhancement, confirmed by numerical evaluation of the full kernel.
Finite-Duration Transitions: For any transition with a finite start time, the ancient-bubble tail is causally cut off, restoring analyticity ( $P_\tau \propto k^3$ , $B_\tau \propto k^6$ ) in the strict $k \to 0$ limit. However, the eternal-background thresholds govern the duration dependence of the moment coefficients. As the duration increases, coefficients grow according to power laws of the number of e-folds $N_{dS}$ , with exponents determined by the critical values (e.g., $M_2 \propto e^{(5-1/h)N_{dS}}$ for $1/5 < h < 1/3$ ).
Crossover and Scale-Dependent Non-Gaussianity: The paper identifies a regime where the power spectrum and bispectrum may enter their tail-dominated behaviors at different scales ( $\xi_{P,tail}$ vs. $\xi_{B,tail}$ ). Because the bispectrum becomes tail-sensitive at smaller $h$ than the power spectrum, a "mixed" regime can exist where one spectrum is tail-dominated while the other remains analytic. This leads to a characteristic scale dependence in the reduced equilateral non-Gaussianity $f_{NL}^{eq}$ , which can scale as $k^{1/h - 6}$ or $k^{6 - 2/h}$ depending on the ordering of the crossover scales.
Temporal Localization as a Regulator: Introducing a quadratic term in the nucleation rate exponent ( $g > 0$ ) provides a smooth regulator. This Gaussian-like localization suppresses the contribution of ancient bubbles via a log-normal factor, ensuring all spatial moments remain finite even in the infinite-duration limit. The spectra recover the analytic $k^3$ and $k^6$ laws in the deep IR, with the tail behavior appearing only as an intermediate regime.
Curvature Perturbations: For a vacuum-dominated transition that completes successfully and reheats suddenly, the curvature perturbation is related to the completion time by $\zeta = H \tau$ . In the fast-transition limit ( $h \ll 1$ ), the intrinsic reduced non-Gaussianity is large ( $f_{NL}^{bubble} \propto h^{-1}$ ), though the power is suppressed on large scales. The paper notes that this interpretation relies on the existence of a common post-transition hypersurface and does not apply to non-percolating regimes like old inflation.

Significance
The paper establishes a general first-arrival framework for analyzing completion-time fluctuations in expanding spacetimes. Its primary significance lies in clarifying how de Sitter expansion generates distinct infrared hierarchies for the power spectrum and bispectrum, specifically demonstrating that the bispectrum is more sensitive to the ancient-bubble tail than the power spectrum. This hierarchy implies that the reduced non-Gaussianity can exhibit unique scale-dependent features depending on the relative timing of the crossover into tail-dominated regimes. Furthermore, the work provides a rigorous treatment of how finite-duration effects and temporal localization act as physical regulators, restoring analyticity while preserving the imprint of expansion in the duration dependence of correlation coefficients. These results offer a theoretical basis for understanding how stochastic phase transitions could source curvature perturbations and non-Gaussianity in the early universe.

More like this