Stochastic Inflation in Numerical Relativity

This paper re-derives gauge-invariant stochastic inflation equations incorporating all metric and scalar degrees of freedom, validates their numerical implementation within the BSSN formulation of Numerical Relativity across slow-roll and ultra slow-roll scenarios, and demonstrates their robustness in simulating fully non-linear stochastic dynamics with retained gradients and anisotropic expansion.

The Gist

Imagine the early universe as a giant, expanding balloon being blown up incredibly fast. This period, called "inflation," is where the seeds of all galaxies were planted. For decades, scientists have tried to understand the tiny, random jitters (quantum fluctuations) on this balloon that eventually grew into stars and galaxies.

However, the standard way of studying these jitters has been like looking at the balloon through a very specific, narrow tunnel. Scientists assumed the balloon was perfectly smooth and that every patch of it was evolving independently, ignoring how different parts might tug on each other or how the balloon's shape might get slightly lopsided. This is like trying to understand a storm by only looking at the wind in one single spot, assuming the rest of the sky is calm.

The New Approach: A Full 3D Weather Map
This paper introduces a new, much more powerful way to simulate the universe during inflation. The authors, Yoann L. Launay, Gerasimos I. Rigopoulos, and E. Paul S. Shellard, have built a "numerical weather map" for the early universe that doesn't rely on those narrow tunnels.

Here is the core idea broken down with simple analogies:

1. The "Stochastic" Noise: The Universe's Static

Think of the quantum jitters as a constant, static noise—like the white noise on an old TV. In the standard model, scientists treat this noise as a simple, smooth background.
In this new work, they treat the noise as a living, breathing entity that constantly kicks the universe. They call this "Stochastic Inflation." Instead of just guessing the average effect of the noise, they simulate the actual "kicks" as they happen, allowing the universe to react in real-time.

2. The "Coarse-Graining" Filter: Separating the Big from the Small

Imagine you are watching a movie of the universe expanding.

• The Problem: You can't simulate every single atom (the tiny, high-frequency details) and the whole galaxy (the big, low-frequency details) at the same time on a computer; it's too much data.
• The Solution: The authors use a "filter" (called coarse-graining). They separate the universe into two parts:
  • The Smooth Part (IR): The big, slow-moving waves that have already crossed the "horizon" (the edge of what we can see). These act like the smooth flow of a river.
  • The Choppy Part (UV): The tiny, fast ripples that are still too small to see. These act like the white foam on the river.
• The Magic: As the universe expands, the "choppy" ripples get stretched out and become part of the "smooth" river. The authors' equations mathematically describe this transition, turning the tiny quantum ripples into the large-scale structure of the universe.

3. The "Separate Universe" Myth vs. Reality

Previous methods often used the "Separate Universe" approximation. Imagine a loaf of raisin bread rising in the oven. The old method assumed that each raisin (a patch of the universe) was in its own tiny, separate oven, rising independently without ever touching its neighbors.
This paper says: "No, they are all in the same oven!"
They use Numerical Relativity (a super-complex way of solving Einstein's equations) to simulate the whole loaf rising together. This allows them to see how different patches interact, how the bread might get slightly lopsided (anisotropic expansion), and how the texture of the dough changes in real-time.

4. What They Tested

To prove their new "oven" works, they ran two specific simulations:

• The Smooth Roll (Slow-Roll): A standard, gentle inflation scenario. This was like a control test to make sure their math matched what we already know. It worked perfectly.
• The Bumpy Ride (Ultra Slow-Roll): A more chaotic scenario where the inflation speed changes drastically (like a car hitting a bump). This is where the old "separate universe" methods usually break down. Their new simulation handled this chaos beautifully, showing that the universe can get very "lumpy" and still follow the laws of physics.

5. The Results: A Robust New Tool

The team found that their new equations:

• Keep the Balance: They strictly obey the "Energy and Momentum" rules of the universe (like a bank account that never goes into debt).
• Capture the Chaos: They can simulate the universe getting very "lumpy" without breaking the math.
• See the Shape: For the first time in this type of simulation, they could track not just how fast the universe expands, but also how it stretches in different directions (like a balloon being squeezed into an egg shape).

Why This Matters (According to the Paper)

The authors claim this is a major upgrade. It moves us from a simplified, 2D sketch of the early universe to a full, 3D, non-linear movie. It removes the need for many "shortcuts" scientists previously had to take.

They are now ready to use this tool to study extreme events in the early universe, such as how primordial black holes might form or how gravitational waves (ripples in space-time) are generated, without having to guess or simplify the physics. They have built a more accurate "time machine" to look back at the very beginning of everything.

Technical Summary

1. Problem Statement

Current observational constraints on inflation (via CMB and Large Scale Structure) are limited, particularly regarding non-linear and non-perturbative interactions. Existing theoretical and numerical methods face significant limitations:

• Quantum Field Theory (QFT) on curved spacetime: Typically limited to tree-level computations and truncated perturbative non-linearity.
• Standard Stochastic Inflation: Relies heavily on the Separate Universe Approximation (SUA), 0th-order gradient expansion, and specific gauge choices (e.g., uniform e-folds). It often neglects anisotropic perturbations and gradient couplings between neighboring points.
• Numerical Relativity (NR) & Lattice Cosmology: While capable of non-perturbative studies, previous NR implementations of inflation often treated perturbations as initial conditions rather than continuously evolving stochastic modes crossing the horizon.

There is a lack of a framework that combines full General Relativity (GR), non-perturbative evolution, gradient effects, and continuous quantum diffusion (stochastic noise) without relying on the SUA or specific gauge choices.

2. Methodology

The authors develop a gauge-invariant formulation of stochastic inflation embedded within the BSSN (Baumgarte-Shapiro-Shibata-Nakamura) formulation of Numerical Relativity.

A. Theoretical Derivation (Gauge Invariant Coarse-Graining)

• Master Variable: Instead of fixing a gauge, the authors use the curvature perturbation on comoving hypersurfaces, $R$, as the master dynamical variable.
• Coarse-Graining: They perform a coarse-graining of the Fourier amplitude of $R$ using a window function $W_k$ (separating UV and IR modes). The time-dependent nature of this split introduces a source term ($S_R$) in the evolution equation for the coarse-grained variable $R_>$.
• Source Terms: By expressing all metric and scalar field perturbations in terms of $R_>$ and its derivatives, they derive the stochastic source terms for the full set of ADM (Arnowitt-Deser-Misner) equations.
  • The stochastic noise acts as a "kick" to the system, representing modes continuously crossing the coarse-graining scale (Hubble radius).
  • Crucially, they prove that these source terms are gauge-invariant and valid for any choice of time-slicing and threading.
• Equations: The resulting system consists of:
  • Stochastic ADM inflaton equation.
  • Stochastic evolution equations for the trace ($K$) and traceless part ($\tilde{K}_{ij}$) of the extrinsic curvature.
  • The standard Hamiltonian and Momentum constraints (which receive no stochastic source terms, ensuring constraint preservation).

B. Numerical Implementation (ISTORIz)

• Codebase: The authors utilize GRTeclyn, a modern NR code based on AMReX (supporting GPU offloading and Adaptive Mesh Refinement), extended into a new module called ISTORIz (Inflationary STOchastic Relativity Integrator for zeta).
• Pipeline:
  1. Initialization: Background conditions and spectral initial conditions for $R$ are generated using the STOIIC generator (solving Mukhanov-Sasaki equations).
  2. Real-Space Evolution: The Mukhanov-Sasaki equation is evolved in real space to generate the stochastic noise field $R(t, x)$.
  3. Stochastic Coupling: At each time step (RK4), the noise spectrum is Fourier transformed, integrated, and inverse-transformed to add stochastic kicks to the right-hand side of the BSSN equations.
  4. Backreaction: The framework allows for "level 2 backreaction," where background parameters (like $\phi_b$) are promoted to local spatial values, making the noise non-Gaussian.
• Gauge Choice: Simulations were run in the synchronous gauge (geodesic slicing, $\alpha=1, \beta^i=0$) with periodic boundary conditions.

3. Key Contributions

1. Gauge-Invariant Stochastic Equations: A rigorous derivation of stochastic inflation equations that do not rely on the Separate Universe Approximation, specific gauges, or gradient expansions.
2. Full GR Stochastic NR: The first numerical implementation of stochastic inflation within a full 3+1 Numerical Relativity framework, retaining anisotropic degrees of freedom (shear) and gradient effects.
3. Constraint Preservation: Demonstration that the stochastic source terms satisfy the Hamiltonian and Momentum constraints to the appropriate order in perturbation theory, ensuring the physical validity of the simulation.
4. Generalization: The framework generalizes standard stochastic inflation, inflationary NR, and lattice cosmology, removing the need for the $\Delta N$ formalism to translate field fluctuations to curvature.

4. Results

The authors validated the framework using two distinct scenarios:

Scenario A: Quadratic Inflation (Slow-Roll)

• Dynamics: Standard slow-roll behavior.
• Validation: The simulation successfully reproduced the expected linear physics. The reconstructed power spectrum ($R_{NR}$) matched the analytical Mukhanov-Sasaki solution ($R_{MS}$) perfectly.
• Constraints: Hamiltonian and Momentum constraints were satisfied to $\sim 10^{-10}$ (second-order in noise), confirming the theoretical derivation.
• Anisotropy: Anisotropic modes ($\tilde{K}_{ij}$) were found to be negligible compared to isotropic modes, consistent with the validity of SUA in slow-roll regimes.

Scenario B: Inflection Point Inflation (Ultra-Slow-Roll - USR)

• Dynamics: A potential with an inflection point triggers a transition to Ultra-Slow-Roll ($\epsilon_1 \approx 0, \epsilon_2 \approx 6$), leading to non-perturbative growth of curvature perturbations.
• Performance: The code successfully handled the extreme slow-down of the field and the importance of gradients, a regime where standard gradient-expansion methods often fail.
• Non-Perturbative Growth: The simulation captured the exponential enhancement of the power spectrum ($\Delta[R] \sim 0.1 - 1$) and reached non-perturbative real-space values ($R \sim 10$).
• Anisotropy: Unlike the slow-roll case, the anisotropic and isotropic contributions to the Hamiltonian constraint became comparable in magnitude during USR. This suggests that the SUA truncation of the momentum constraint may be violated in USR regimes, highlighting the necessity of full GR treatment.
• Constraints: Constraints remained satisfied even in this highly non-linear regime, with violations remaining at the expected order of magnitude.

5. Significance

• Theoretical Robustness: The work proves that stochastic inflation can be formulated and solved without the restrictive assumptions of the SUA or specific gauges, providing a more fundamental description of quantum diffusion in inflation.
• Non-Perturbative Predictions: It enables reliable predictions for non-perturbative phenomena, such as Primordial Black Hole (PBH) formation and non-Gaussianities, which are sensitive to the details of the inflationary potential and gradient effects.
• Anisotropic Physics: By retaining anisotropic degrees of freedom, the framework opens the door to calculating scalar-induced gravitational waves and studying the validity of isotropic approximations in extreme regimes (like USR).
• Future Applications: The code (ISTORIz) provides precise initial conditions for subsequent cosmological eras and offers a tool to confront upcoming observational data with a wider, more accurate testable inflationary landscape.

In summary, this paper bridges the gap between stochastic inflation and numerical relativity, offering a powerful, gauge-invariant tool to simulate the fully non-linear, non-perturbative evolution of the early universe.