The GIGANTES Data Set: Precision Cosmology from Voids in the Machine-learning Era

Abstract

We present GIGANTES, the most extensive and realistic void catalog suite ever released—containing over 1 billion cosmic voids covering a volume larger than the observable universe, more than 20 TB of data, and created by running the void finder VIDE on QUIJOTE’s halo simulations. The GIGANTES suite, spanning thousands of cosmological models, opens up the study of voids, answering compelling questions: Do voids carry unique cosmological information? How is this information correlated with galaxy information? Leveraging the large number of voids in the GIGANTES suite, our Fisher constraints demonstrate voids contain additional information, critically tightening constraints on cosmological parameters. We use traditional void summary statistics (void size function, void density profile) and the void autocorrelation function, which independently yields an error of 0.13 eV on ∑ m ν for a 1 h −3 Gpc3 simulation, without cosmic microwave background priors. Combining halos and voids we forecast an error of 0.09 eV from the same volume, representing a gain of 60% compared to halos alone. Extrapolating to next generation multi-Gpc3 surveys such as the Dark Energy Spectroscopic Instrument, Euclid, the Spectro-Photometer for the History of the Universe and Ices Explorer, and the Roman Space Telescope, we expect voids should yield an independent determination of neutrino mass. Crucially, GIGANTES is the first void catalog suite expressly built for intensive machine-learning exploration. We illustrate this by training a neural network to perform likelihood-free inference on the void size function, giving a ∼20% constraint on Ωm. Cosmology problems provide an impetus to develop novel deep-learning techniques. With GIGANTES, machine learning gains an impressive data set, offering unique problems that will stimulate new techniques.

The Halo Bias inside Cosmic Voids

“…Recently, many cosmological statistics of cosmic voids have been explored, for instance the void size function and void clustering, which can probe the underlying cosmological model of the universe (Hamaus et al 2015;Pisani et al 2015;Cai et al 2016;Sahlen et al 2016;Achitouv et al 2017;Chuang et al 2017;Hamaus et al 2017;Hawken et al 2017;Sahlen & Silk 2018;Achitouv 2019;Kreisch et al 2019;Nadathur et al 2019;Pisani et al 2019;Verza et al 2019;Aubert et al 2020;Hamaus et al 2020;Hawken et al 2020;Nadathur et al 2020;Bayer et al 2021;Kreisch et al 2021;Contarini et al 2022;Moresco et al 2022;Woodfinden et al 2022).…”

“…Cosmic voids are large underdense regions in the large-scale structure of the universe, span a large range of scales, and constitute the largest observable objects in the universe. Their size and underdense nature make them particularly suited to probe dark energy and modified gravity (Lee & Park 2009;Biswas et al 2010;Lavaux & Wandelt 2010;Li & Efstathiou 2012;Clampitt et al 2013;Spolyar et al 2013;Cai et al 2015;Pisani et al 2015;Zivick et al 2015;Achitouv 2016;Pollina et al 2016;Sahlen et al 2016;Falck et al 2018;Sahlen & Silk 2018;Paillas et al 2019;Perico et al 2019;Verza et al 2019;Contarini et al 2021Contarini et al , 2022, massive neutrinos (Massara et al 2015;Banerjee & Dalal 2016;Kreisch et al 2019;Sahlen 2019;Schuster et al 2019;Zhang et al 2020;Kreisch et al 2021;Contarini et al 2022), primordial non-Gaussianity (Chan et al 2019), and physics beyond the standard model (Peebles 2001;Reed et al 2015;Yang et al 2015;Baldi & Villaescusa-Navarro 2016;Lester & Bolejko 2021;Arcari et al 2022).…”

The Disordered Heterogeneous Universe: Galaxy Distribution and Clustering Across Length Scales

“…We recall that our H V void statistic follows a somewhat different definition to the void size function often used in cosmology, and is restricted to comparatively large scales (r > 20h −1 Mpc, as for the other statistics), with only spherical voids. This differs from the approach used in several cosmological studies [e.g.,10,128,129] and explains the reduced utility found herein, and the different correlation properties seen in Fig.11.…”

Why Cosmic Voids Matter: Nonlinear Structure & Linear Dynamics

“…This would open up the possibility to maximize the number of observable linear modes of the density field of large-scale structure that can be exploited for the purpose of cosmological inference, far beyond the previously imposed limits. Moreover, these data sets will contain hundreds of thousands of voids with numerous individual properties, providing a rich playground for the latest machine learning applications [51,131].…”