Primordial Physics in the Nonlinear Universe: signatures of inflationary resonances, excitations, and scale dependence

This paper demonstrates that weak lensing analysis of future LSST data can provide competitive and complementary constraints on primordial non-Gaussianity models, particularly those with features on small scales, by resolving their signatures in the deeply nonlinear late-time density field through a suite of over thirty templates and new simulations of resonant signatures.

The Gist

Imagine the universe as a giant, cosmic ocean. When the universe was born in the Big Bang, it went through a period of rapid expansion called inflation. Think of this like a sudden, violent gust of wind blowing across a calm lake.

Usually, scientists assume the ripples created by this wind are perfectly smooth and random, like gentle waves. But what if the wind didn't just blow randomly? What if it hit a hidden rock, or a specific type of tree, creating a unique, complex pattern of splashes? These unique patterns are called Primordial Non-Gaussianities (PNGs). They are the "fingerprints" of the physics that happened in the very first split-second of the universe.

For decades, scientists have tried to find these fingerprints by looking at the Cosmic Microwave Background (CMB). You can think of the CMB as a baby photo of the universe, taken when it was only 380,000 years old. It's a beautiful picture, but it's a bit blurry and only shows the surface.

This paper is about looking at the universe's "adult" photo instead.

The Problem: The "Baby Photo" vs. The "Adult"

The baby photo (CMB) is great, but it only sees large, smooth waves. It misses the tiny, chaotic splashes that happen when the universe gets older and gravity starts pulling everything together to form stars, galaxies, and massive clusters of galaxies.

The authors of this paper asked: What if we look at the universe today, when it's fully grown and messy? Can we see the fingerprints of those early winds in the tangled web of galaxies?

The Solution: The Cosmic Simulator

To answer this, the team built a super-computer simulation. Imagine a massive digital sandbox where they can recreate the entire history of the universe.

1. The Recipe: They wrote a special computer program (called "Aarambam") that allows them to inject specific, weird patterns (the PNGs) into the "dough" of the universe at the very beginning.
2. The Experiment: They ran over 30 different experiments. In each one, they changed the "recipe" slightly to mimic different theories about how the early universe worked (like particles colliding, or the universe vibrating with a specific rhythm).
3. The Growth: They let the simulation run forward in time, watching gravity turn those tiny ripples into massive clumps of matter (galaxies and dark matter halos).

The Discovery: The "Adult" Photo is Sharper

They compared their simulation results with what we expect to see from the CMB (the baby photo) and what we will see from the Rubin Observatory (a giant new telescope that will take the best "adult" photos of the universe in the next decade).

Here is what they found, using some analogies:

• The Small Scale Advantage: The CMB is like looking at a forest from a helicopter; you see the general shape of the trees. The Rubin Observatory is like walking through the forest; you can see the individual leaves and twigs.
  • The authors found that for certain types of "fingerprints" (specifically those that create patterns on very small scales), the "walking through the forest" approach (lensing) is actually better than the helicopter view. It can detect signals that the baby photo completely misses.
• The "Non-Monotonic" Dance: Some of the early universe theories predicted that the universe would vibrate like a plucked guitar string. When they simulated this, they saw something fascinating: the number of galaxy clusters didn't just go up or down smoothly. Instead, it went up, then down, then up again, depending on the size of the cluster. It was like a dance where the dancers suddenly change steps. This "wiggly" pattern is a unique signature that proves the early universe had these specific vibrations.
• Independent Clues: They also tested a scenario where the "wind" created ripples in two different ways at once (in the density of matter and in the way matter clumped together). They found that these two clues are independent. It's like finding two different fingerprints on a glass; one tells you who touched it, and the other tells you how hard they pressed. You can solve the mystery using either clue, or both together, without them confusing each other.

Why This Matters

This paper is a game-changer because it opens a new window into the early universe.

• Before: We were mostly limited to looking at the smooth, baby universe.
• Now: We have a toolkit to look at the messy, adult universe.
• The Result: We can now test theories about the "micro-physics" of the Big Bang (like particle collisions) that were previously impossible to check.

The authors are essentially saying: "Don't just look at the baby picture. Look at the grown-up universe. It's messy, but it holds the secrets to how the universe was built, and we finally have the tools to read them."

They have made their computer code and simulation data public, so other scientists can use these tools to keep exploring the cosmic ocean.

Technical Summary

1. Problem Statement

While the search for Primordial Non-Gaussianities (PNGs) has historically focused on the Cosmic Microwave Background (CMB) and the quasi-linear regime of large-scale structure (LSS), the impact of these signatures on the deeply nonlinear regime (late-time structure formation, dominated by halos and filaments) remains largely unexplored.

• Limitation of Current Methods: Standard perturbative approaches used in galaxy surveys struggle to model the nonlinear gravitational evolution of structure, limiting their ability to probe small scales where many inflationary signatures (e.g., resonances, excited states) peak.
• Gap: There is a lack of simulation suites capable of consistently propagating arbitrary primordial bispectra (including complex resonant and scale-dependent models) from initial conditions to the nonlinear epoch to test their distinct phenomenology.

2. Methodology

The authors utilize and extend the Ulagam simulation suite (based on the $N$-body code PkdGrav3) to address these gaps.

A. Initial Condition Generation

• Arbitrary Bispectra: Building on their previous work (Paper I), they employ a mode decomposition technique to generate initial density fields with arbitrary primordial bispectra.
• Decomposition: They decompose complex, non-separable bispectra into a sum of factorizable terms using a basis of monomials and modified Legendre polynomials:
  $$B(k_1, k_2, k_3) = \frac{1}{k_1^2 k_2^2 k_3^2} \sum_{i,j,k} \alpha_{ijk} q_i(k_1)q_j(k_2)q_k(k_3)$$
• Validation: The method is validated against numerical accuracy, ensuring the generated fields reproduce the target bispectrum kernels (including Planck templates) within the simulation volume ($V = 1 \text{ Gpc}^3/h^3$, $512^3$ particles).

B. Inflationary Models Studied

The study covers over 30 templates derived from Planck CMB analyses, categorized into:

1. Scale-Dependent Local: Variations of the local shape ($f_{NL}^{local}$) with power-law exponents ($n_{NG}$).
2. Resonant Models:
   • Linear Resonance ($BLinRes$) and Logarithmic Resonance ($BLogRes$) arising from features in the inflaton potential.
   • $K^2 \cos$ models (including damping envelopes).
3. Excited Initial States (Non-Bunch-Davies): Models like $NBD2$ and $NBD_{sin}$ where the inflaton ground state is excited.
4. DBI Inflation: Dirac-Born-Infeld models arising from brane motion.
5. Joint Oscillations: A novel class of models where both the power spectrum and bispectrum contain oscillatory features (resonances), simulating the full physical impact of sharp features in the inflaton potential.

C. Forecasting and Analysis

• Survey Simulation: Forecasts are generated for the Rubin Observatory LSST Year-10 weak lensing survey (14,000 deg$^2$, $n_{gal} = 30$ arcmin$^{-2}$).
• Statistics: The analysis uses the 2nd and 3rd moments (variance and skewness) of the lensing convergence field ($\kappa$) across tomographic bins and smoothing scales.
• Fisher Forecasting: Constraints on the amplitude $f_{NL}$ (and resonance amplitudes $A_{pk}$) are derived using a Fisher matrix approach, comparing against Planck 2018 constraints derived via the CMB-BEST pipeline.

3. Key Contributions

1. First Nonlinear Simulations of 30+ Templates: This is the first study to consistently simulate and analyze the nonlinear signatures of a broad suite of inflationary templates (including resonances and excited states) in the late-time density field.
2. Joint Power/Bispectrum Resonance Simulation: The authors simulate, for the first time, scenarios where resonant features appear simultaneously in both the primordial power spectrum and bispectrum, allowing for the study of their interplay.
3. Public Data Release: The initial condition generator (Aarambam) and the resulting Ulagam simulation suite (including density shells, 3D fields, and halo catalogs) are publicly released.
4. Methodological Validation: The paper provides rigorous validation of the mode-decomposition method for generating initial conditions with arbitrary bispectra, confirming its accuracy for a wide range of models.

4. Key Results

A. Lensing vs. CMB Constraints

• Competitiveness: Weak lensing forecasts for LSST Y10 are competitive with Planck 2018 for many models.
• Small-Scale Advantage: For templates where the signal peaks on small scales ($k \gtrsim 0.2 \, h/\text{Mpc}$), lensing constraints surpass CMB constraints. This is because the CMB is limited to larger scales, whereas lensing can probe the nonlinear regime where these features are amplified.
• CMB Superiority: The CMB remains superior for models peaking on large scales (e.g., $K^2 \cos$ with specific parameters) or logarithmic resonances that overlap heavily with CMB scales.

B. Nonlinear Signatures

• Halo Abundance (HMF): The primary driver of lensing sensitivity is the change in the Halo Mass Function. PNGs alter the tails of the density distribution, significantly boosting or suppressing massive halo counts.
  • Non-Monotonicity: Oscillatory templates (resonances, NBD) induce non-monotonic behaviors in the HMF and halo bias. For example, specific halo masses may be preferentially suppressed or enhanced depending on the oscillation phase, a feature distinct from standard local PNGs.
• Power Spectrum & Bispectrum:
  • Primordial bispectrum features induce oscillatory patterns in the late-time matter power spectrum and bispectrum.
  • These oscillatory features are most prominent at high redshift ($z \sim 2$) and are partially washed out at $z=0$ due to nonlinear structure formation, though the overall signal amplitude grows with time.
• Independence of Amplitudes: In models with oscillations in both the power spectrum ($A_{pk}$) and bispectrum ($f_{NL}$), the constraints on these two parameters are found to be essentially independent (non-degenerate). This suggests that lensing data can constrain power spectrum oscillations without needing to marginalize over bispectrum uncertainties.

C. Correlation Analysis

• Resonant particle production models are orthogonal to scale-dependent and excited-state models in the late-time Fisher posterior. This implies that combining different probes can effectively break degeneracies between different inflationary physics scenarios.

5. Significance

• New Window on Inflation: The work demonstrates that nonlinear structure formation is not just a source of noise but a rich, complementary probe for inflationary physics. It offers a unique window into microphysics (particle interactions, excited states) that is inaccessible to linear probes.
• Synergy of Probes: It establishes that future weak lensing surveys (like LSST) will provide leading constraints on specific classes of inflationary models (particularly those with small-scale features), complementing and extending the reach of CMB experiments.
• Simulation-Based Inference: The paper advocates for a shift toward simulation-based inference for inflationary studies, moving beyond perturbative approximations to fully capture the nonlinear evolution of primordial features.
• Resource for Community: By releasing the Aarambam generator and Ulagam simulations, the authors provide the community with the necessary tools to test a vast landscape of inflationary models against future observational data.

In conclusion, this paper bridges the gap between early-universe inflationary models and late-time nonlinear structure, demonstrating that weak lensing is a powerful, competitive, and complementary tool for constraining the fundamental physics of the early Universe.