[Paper Review] Limits on primordial non-Gaussianities from BOSS galaxy-clustering data

Guido D’Amico, Matthew Lewandowski+2 more|arXiv (Cornell University)|Jan 27, 2022

Astronomy and Astrophysical ResearchPhysics and Astronomy61 citations

TL;DR

This paper analyzes BOSS DR12 galaxy clustering with one-loop EFTofLSS to constrain f_NL for equilateral, orthogonal, and local shapes; finds no evidence for primordial non-Gaussianity and provides competitive bounds.

ABSTRACT

We analyze the power spectrum and the bispectrum of BOSS galaxy-clustering data using the prediction from the Effective Field Theory of Large-Scale Structure at one-loop order for $ extit{both}$ the power spectrum $ extit{and}$ the bispectrum. With $Λ$CDM parameters fixed to Planck preferred values, we set limits on three templates of non-Gaussianities predicted by many inflationary models: the equilateral, the orthogonal, and the local shapes. After validating our analysis against simulations, we find $f_{ m NL}^{ m equil.}= 207 \pm 292\, , f_{ m NL}^{ m orth.}= -68 \pm 73\, , f_{ m NL}^{ m loc.}= 52 \pm 34$, at $68\%$ confidence level. These bispectrum-based constraints from Large-Scale Structure, not far from the ones of WMAP, suggest promising results from upcoming surveys.

Motivation & Objective

• Motivate the search for primordial non-Gaussianities in large-scale structure data.
• Apply the EFTofLSS framework at one-loop to both power spectrum and bispectrum with non-Gaussian initial conditions.
• Calibrate the model against simulations and derive constraints on f_NL for three standard shapes.
• Quantify theoretical and observational uncertainties and compare with CMB results.

Proposed method

• Model galaxy clustering with EFTofLSS at one-loop for power spectrum and bispectrum.
• Incorporate non-Gaussian initial conditions via f_NL and the leading B_phi_p shapes.
• Use biased tracers in redshift space with a perturbative bias expansion.
• Convolve theory with survey window functions and estimate covariances from mocks (Patchy, Nseries).
• Perform likelihood analyses with Planck-fixed LCDM parameters using MontePython and PyBird.
• Compute and compare constraints against simulations to assess theoretical error.

Experimental results

Research questions

• RQ1 What are the one-loop EFTofLSS constraints on primordial non-Gaussianity (f_NL) for equilateral, orthogonal, and local shapes from BOSS DR12?
• RQ2 Do BOSS DR12 measurements show evidence for primordial non-Gaussianity in the power spectrum and bispectrum, given Planck LCDM parameters?
• RQ3 How do simulation-based calibrations and theoretical errors affect the robustness of the f_NL bounds?
• RQ4 How do one-loop bispectrum analyses compare to tree-level predictions in tightening f_NL constraints?

Key findings

• No significant evidence of primordial non-Gaussianity is found from BOSS data for any shape.
• 68% CL constraints: f_NL^equil = 245 ± 293, f_NL^orth = -60 ± 72, f_NL^loc = 7 ± 31.
• Compared to WMAP/Planck, BOSS bounds are competitive, with the orthogonal constraint surpassing WMAP and approaching Planck in some cases.
• Inclusion of the one-loop bispectrum improves equilateral and orthogonal constraints by factors of ~1.5–1.9 over tree-level analyses.
• The analysis validates against 84 N-series Gaussian mocks and shows theoretical errors are subdominant to statistical errors under their modeling.
• Forecasts suggest upcoming surveys (e.g., DESI) could further tighten these bounds by substantial factors.