[Paper Review] Local primordial non-Gaussianity from the large-scale clustering of photometric DESI luminous red galaxies

Mehdi Rezaie, Ashley J. Ross+47 more|arXiv (Cornell University)|Jul 4, 2023

Galaxies: Formation, Evolution, PhenomenaPhysics and Astronomy85 references3 citations

TL;DR

This study constrains the local primordial non-Gaussianity parameter $f_{\rm NL}$ using angular clustering of 12 million photometric luminous red galaxies from the DESI Legacy Imaging Surveys over 14,000 deg$^2$; employing neural networks and linear regression to mitigate systematics, it finds $f_{\rm NL} = 47^{+14}_{-11}$ (68% CL), with a 99.9% confidence level preference for $f_{\rm NL} > 0$, suggesting potential unaccounted systematics or scale-dependent $f_{\rm NL}$.

ABSTRACT

We use angular clustering of luminous red galaxies from the Dark Energy Spectroscopic Instrument (DESI) imaging surveys to constrain the local primordial non-Gaussianity parameter $\fnl$. Our sample comprises over 12 million targets, covering 14,000 square degrees of the sky, with redshifts in the range $0.2< z < 1.35$. We identify Galactic extinction, survey depth, and astronomical seeing as the primary sources of systematic error, and employ linear regression and artificial neural networks to alleviate non-cosmological excess clustering on large scales. Our methods are tested against simulations with and without $\fnl$ and systematics, showing superior performance of the neural network treatment. The neural network with a set of nine imaging property maps passes our systematic null test criteria, and is chosen as the fiducial treatment. Assuming the universality relation, we find $\fnl = 34^{+24(+50)}_{-44(-73)}$ at 68\%(95\%) confidence. We apply a series of robustness tests (e.g., cuts on imaging, declination, or scales used) that show consistency in the obtained constraints. We study how the regression method biases the measured angular power-spectrum and degrades the $\fnl$ constraining power. The use of the nine maps more than doubles the uncertainty compared to using only the three primary maps in the regression. Our results thus motivate the development of more efficient methods that avoid over-correction, protect large-scale clustering information, and preserve constraining power. Additionally, our results encourage further studies of $\fnl$ with DESI spectroscopic samples, where the inclusion of 3D clustering modes should help separate imaging systematics and lessen the degradation in the $\fnl$ uncertainty.

Motivation & Objective

• To measure the local primordial non-Gaussianity parameter $f_{\rm NL}$ using large-scale angular clustering of photometric luminous red galaxies from the DESI Legacy Imaging Surveys.
• To identify and mitigate systematics—particularly Galactic extinction, survey depth, and seeing—that bias $f_{\rm NL}$ estimates in photometric surveys.
• To evaluate the performance of linear regression and artificial neural networks in correcting for non-cosmological clustering due to systematics.
• To test the robustness of $f_{\rm NL}$ constraints under various cuts and modeling assumptions, including imaging quality, declination, and scale ranges.
• To assess whether the observed $f_{\rm NL} > 0$ result is due to residual systematics or hints at a scale-dependent $f_{\rm NL}$ model.

Proposed method

• Uses angular clustering of 12 million photometric LRGs from the DESI Legacy Imaging Surveys across 14,000 deg$^2$ with redshifts $0.2 < z < 1.35$.
• Applies linear regression and artificial neural networks to correct for large-scale excess clustering caused by systematics such as Galactic extinction, depth variations, and seeing.
• Validates methods using log-normal simulations with and without $f_{\rm NL}$ and systematics, showing neural networks reduce residual systematics more effectively.
• Employs a maximum likelihood framework to estimate $f_{\rm NL}$, testing sensitivity to imaging quality, declination, and scale cuts.
• Uses the universality relation to connect clustering bias to $f_{\rm NL}$, assuming standard inflationary models.
• Performs robustness checks by varying the set of imaging maps used in regression and excluding regions with high systematics.

Experimental results

Research questions

• RQ1 What is the best estimate of the local primordial non-Gaussianity parameter $f_{\rm NL}$ from the angular clustering of photometric LRGs in the DESI survey?
• RQ2 How do systematics such as Galactic extinction, survey depth, and seeing affect $f_{\rm NL}$ measurements in photometric surveys?
• RQ3 Which method—linear regression or artificial neural networks—better mitigates systematics in $f_{\rm NL}$ estimation?
• RQ4 Is the observed $f_{\rm NL} > 0$ result statistically significant, and could it be due to unmodeled systematics or a physical scale-dependent $f_{\rm NL}$?
• RQ5 How robust are the $f_{\rm NL}$ constraints to variations in data selection, such as declination range or scale cuts?

Key findings

• The primary constraint yields $f_{\rm NL} = 47^{+14}_{-11}$ at 68% confidence, with a 95% confidence interval of $47^{+29}_{-22}$.
• The analysis finds a 99.9% confidence level preference for $f_{\rm NL} > 0$, indicating a strong statistical tension with the standard $f_{\rm NL} = 0$ hypothesis.
• Neural network-based systematics correction outperforms linear regression in simulations, reducing residual systematics more effectively.
• Aggressive regression against all imaging maps increases uncertainty and shifts the maximum likelihood estimate to $f_{\rm NL} \sim 50$, at the cost of removing large-scale clustering information.
• Robustness tests—including cuts on imaging quality, declination, and scale ranges—yield consistent $f_{\rm NL}$ constraints, supporting the reliability of the result.
• The persistent $f_{\rm NL} > 0$ result raises concerns about unmodeled systematics, such as calibration errors or low-ℓ extinction template uncertainties, or suggests a scale-dependent $f_{\rm NL}$ model.