Unbiased cosmology inference from biased tracers using the EFT likelihood

Abstract

We present updates on the cosmology inference using the effective field theory (EFT) likelihood presented previously in Schmidt et al., 2018, Elsner et al., 2019 \cite{paperI,paperII}. Specifically, we add a cutoff to the initial conditions that serve as starting point for the matter forward model. We show that this cutoff, which was not employed in any previous related work, is important to regularize loop integrals that otherwise involve small-scale, non-perturbative modes. We then present results on the inferred value of the linear power spectrum normalization σ8 from rest-frame halo catalogs using both second- and third-order bias expansions, imposing uniform priors on all bias parameters. Due to the perfect bias-σ8 degeneracy at linear order, constraints on σ8 rely entirely on nonlinear information. The results show the expected convergence behavior when lowering the cutoff in wavenumber, L. When including modes up to k ⩽ Λ = 0.1 h Mpc−1 in the second-order case, σ8 is recovered to within ≲ 6% for a range of halo masses and redshifts. The systematic bias shrinks to 4% or less for the third-order bias expansion on the same range of scales. Together with additional evidence we provide, this shows that the residual mismatch in σ8 can be attributed to higher-order bias contributions. We conclude that the EFT likelihood is able to infer unbiased cosmological constraints, within expected theoretical systematic errors, from physical biased tracers on quasilinear scales.

A rigorous EFT-based forward model for large-scale structure

“…Perturbation-theory predictions of the matter power spectrum and other statistics typically do not employ an explicit cutoff as done in this forward model (but see [34,35,58]). Instead, the cutoff is sent to infinity, or at least large values, so that loop integrals run over arbitrarily large momenta.…”

Galaxy bias from forward models: linear and second-order bias of IllustrisTNG galaxies

“…(1.1)) to construct the galaxy distribution out of the final matter distribution; and (iii) a likelihood function to compare the forward-evolved galaxy field with the simulated one. Forward models are still in their infancy, but rapid progress is being made [44][45][46][47][48][49][50][51][52][53][54][55][56][57][58][59]. A key advantage of forward models is that the inference is made directly at the level of the actually observed galaxy density field, which maximizes the extraction of the information available up to a given order in perturbation theory, in contrast to relying for example on a handful of summary statistics like N -point functions.…”

“…In this paper we build specifically on the recent developments made in Refs. [55][56][57][58] on forward models, including the derivation of the crucial likelihood function using the EFT formalism [55,60,61]. This paper is accompanied by another one using the same methodology, but focused on the assembly bias signal of b 2 and b K 2 of dark matter halos [62] (i.e.…”

Optimal machine-driven acquisition of future cosmological data

“…Recent extensions to the Borg framework have led to substantial improvements in cosmological parameter inference (Kodi Ramanah et al 2019;Elsner et al 2020;Schmidt et al 2020), Lyman-α (Porqueres et al 2019a(Porqueres et al , 2020 and cosmic shear (Porqueres et al 2021) reconstructions, with the field-level treatment transcending the capabilities of conventional cosmological analyses. Novel sophisticated additions to the forward model include a robust likelihood to account for unknown foreground contamination and systematics (Porqueres et al 2019b) and machine learning-based galaxy bias models (Charnock et al 2020).…”