The limits of lattice inflation: a cautionary tale

This paper demonstrates that approximating the gravitational background as an exact FLRW spacetime in lattice inflation simulations prevents the freezing of superhorizon modes and significantly distorts inflationary observables, particularly during ultra-slow roll, prompting the authors to propose a validity criterion and implement it in CosmoLattice.

The Gist

Imagine the early Universe as a giant, expanding balloon. Scientists use powerful computer simulations (called "lattice simulations") to watch how tiny ripples on this balloon grow into the stars and galaxies we see today. These simulations are like high-speed movies that track the chaotic dance of energy fields right after the Big Bang.

However, the authors of this paper, Will Barker, Benjamin Gladwyn, and Sebastian Zell, have discovered a hidden flaw in how many of these "movies" are made.

The Flaw: Ignoring the "Bumpy Road"

Most of these computer programs assume the Universe expands on a perfectly smooth, flat road (what physicists call an FLRW background). They pretend the road is perfectly flat and ignore any bumps or potholes (which physicists call metric perturbations).

The authors argue that while this "smooth road" assumption works fine for some parts of the story, it breaks the movie during Inflation—the period when the Universe was expanding incredibly fast.

The Problem: The "Freezing" That Never Happens

Here is the core issue explained with an analogy:

Imagine you are trying to freeze a drop of water into an ice cube.

• The Real Universe: When a ripple (a wave of energy) moves far away from the center of the balloon (crossing the "horizon"), it should get "frozen" in place. It stops changing and stays as a permanent mark on the balloon. This is crucial because these frozen marks eventually become the seeds for galaxies.
• The Flawed Simulation: Because the simulation ignores the "bumps" in the road, the ripple never freezes. Instead, it keeps shrinking and fading away, like a sound wave dying out in an empty room.

The authors show that if you ignore these bumps, the simulation predicts that the ripples disappear instead of staying put. This means the simulation gets the final picture of the Universe completely wrong.

Two Scenarios: The Slow Walk vs. The Sprint

The paper tests this idea in two different "speeds" of inflation:

1. Slow Roll (The Slow Walk): Even when the Universe expands at a steady, slow pace, ignoring the bumps causes the ripples to fade away. The simulation thinks the Universe is much quieter than it actually is.
2. Ultra-Slow Roll (The Sprint): Sometimes, the Universe expands even faster in a specific way. In this case, the error is catastrophic. The simulation doesn't just fade the ripples; it makes them grow wildly and unphysically, creating a "monster" universe that doesn't match reality.

The Solution: Adding the "Bumps" Back In

The authors didn't just point out the problem; they fixed it. They updated their simulation software (called CosmoLattice) to include the "bumps" (first-order metric perturbations).

• Before the fix: The ripples faded or grew strangely.
• After the fix: The ripples froze exactly as they should, matching the correct mathematical predictions.

Why Reheating is Different

The paper also looked at Reheating, which is the phase after inflation when the Universe gets hot and forms particles. Surprisingly, the authors found that the "smooth road" assumption works fine here. The ripples don't need to freeze in the same way, so ignoring the bumps doesn't break the simulation for this specific part of the story.

The Takeaway: A New Rule of Thumb

The authors propose a simple rule for scientists using these simulations:

"Check the bumps before you trust the movie."

If the "bumps" (metric perturbations) are small, the simple simulation is probably okay. But if the bumps are significant, the simple simulation is lying to you. In those cases, you need a much more expensive and complex type of simulation (called Numerical General Relativity) that accounts for the bumpy road to get the right answer.

In short: To understand how the Universe grew from a tiny quantum fluctuation into a vast cosmos, you can't pretend the road is perfectly flat. You have to account for the bumps, or your story of the Universe will be a fairy tale, not a fact.

Technical Summary

Technical Summary: The Limits of Lattice Inflation

Problem Statement
Cosmological lattice simulations have become essential for studying non-perturbative dynamics in the early Universe, particularly during reheating. However, the vast majority of widely used codes (e.g., LATTICEEASY, DEFROST, CosmoLattice) approximate the gravitational background as an exact Friedmann–Lemaître–Robertson–Walker (FLRW) spacetime, explicitly neglecting metric perturbations. While this approximation is standard, its validity during the inflationary epoch has not been systematically addressed. The authors identify a critical flaw: neglecting metric perturbations prevents the "freezing" of superhorizon modes, a necessary condition for the persistence of primordial fluctuations until Cosmic Microwave Background (CMB) formation. This leads to significant distortions in inflationary observables, such as the curvature power spectrum and spectral index, particularly in scenarios involving ultra-slow roll (USR).

Methodology
The authors employ a combination of analytical derivation and numerical simulation to quantify the error introduced by the FLRW approximation:

1. Analytical Derivation: Starting from a scalar field minimally coupled to gravity, the authors derive the equation of motion for first-order field perturbations ($\delta\phi$). They compare two cases:

   • Full Treatment ($\sigma=1$): Includes first-order metric perturbations ($A$ and $B$) in the spatially flat gauge, recovering the standard Mukhanov–Sasaki (MS) equation.
   • FLRW Approximation ($\sigma=0$): Neglects metric perturbations, resulting in a "deformed" MS equation.
     They analytically solve the deformed equation during slow-roll and ultra-slow roll phases to determine the behavior of superhorizon modes.

2. Numerical Implementation:

   • Models: The authors test two scenarios: a standard slow-roll model with potential $V(\phi) = \lambda\phi^4$ and a USR model featuring an inflection point designed to enhance perturbations.
   • Codes: They utilize the public lattice code CosmoLattice. They implement a modification to CosmoLattice to include first-order metric perturbations (following the approach of HLATTICE) and compare results against the standard FLRW version.
   • Validation: They solve the MS and deformed MS equations numerically using high-precision ODE solvers (specifically benchmarking Julia's RadauIIA5 and Python's Radau) to serve as analytical benchmarks for the lattice results.

Key Contributions and Results

• Derivation of Superhorizon Decay: The authors demonstrate that in the absence of metric perturbations, the curvature power spectrum $\mathcal{P}_\mathcal{R}$ does not freeze on superhorizon scales. Instead, during slow-roll, it decays as:
  $$\mathcal{P}_\mathcal{R} \sim H^4$$
  where $H$ is the Hubble scale. During ultra-slow roll, this deviation is even more pronounced, leading to unphysical enhancements or drops in the power spectrum depending on the specific evolution of the slow-roll parameters.

• Quantification of Errors:

  • Slow-Roll: In the $\lambda\phi^4$ model, the deformed MS equation yields a power spectrum and spectral index that differ significantly from the correct MS result due to the decay of modes after horizon exit.
  • Ultra-Slow Roll: In the USR model, the FLRW approximation leads to a sudden, unphysical growth of modes that exited the horizon long before the USR phase began. The lattice simulations using the standard FLRW approximation track this deformed analytic behavior, confirming that the lattice error stems directly from the omission of metric perturbations.

• Reheating Insensitivity: In contrast to inflation, the authors find that lattice studies of reheating are considerably less sensitive to the omission of metric perturbations. In their reheating model, including first-order metric corrections had virtually no effect on the evolution of energy densities. This is attributed to the fact that reheating observables depend on total energy densities rather than the freezing of superhorizon spectrum modes, and no modes exit the horizon during the reheating phase.

• Proposed Validity Criterion: The paper proposes a practical, a posteriori criterion for assessing the validity of FLRW simulations:

  1. Compute the first-order metric perturbations ($\Delta A$ and $\Delta \nabla B$) using quantities available on the FLRW lattice.
  2. Re-run the simulation including these first-order metric terms.
  3. Conclusion: If the inclusion of first-order metric perturbations significantly modifies the outcome (e.g., restores freezing or drastically alters the spectrum), the FLRW approximation is invalid, and Numerical General Relativity (NGR) is required. If the correction is negligible, the FLRW simulation is likely reliable (though higher-order perturbations remain a potential limitation).

Significance and Claims
The authors claim that their work initiates a systematic investigation into the validity of lattice simulations for inflation. Their primary significance lies in:

1. Correcting a Fundamental Flaw: They show that the standard FLRW approximation used in many inflationary lattice studies qualitatively changes the behavior of perturbations, rendering extracted observables (like the spectral index) unreliable.
2. Providing a Diagnostic Tool: By implementing first-order metric perturbations in CosmoLattice, they provide a method to diagnose when FLRW simulations fail.
3. Clarifying Applicability: They delineate that while FLRW simulations are robust for reheating, they are fundamentally flawed for inflation unless metric perturbations are included.
4. Defining the Hierarchy of Methods: The paper clarifies the hierarchy of approximation:
   • MS Equation: Linear field perturbations with linear metric perturbations.
   • FLRW Lattice: Non-linear field perturbations with no metric perturbations (found to be invalid for inflation).
   • "Improved" FLRW Lattice: Non-linear field perturbations with linear metric perturbations (valid for testing validity, but limited to first-order accuracy).
   • NGR Lattice: Fully non-linear field and metric perturbations (required for genuinely non-perturbative information when first-order corrections are large).

The authors conclude that while their proposed criterion helps decide when fast FLRW simulations are sufficient, fully non-perturbative information in regimes where first-order corrections are significant requires Numerical General Relativity. They also note that the classical nature of lattice simulations remains an open challenge for future work.