NII] 205
5.1 Introduction
C h a p t e r 5
LIMFAST. I. A SEMI-NUMERICAL TOOL FOR LINE INTENSITY MAPPING
Mas-Ribas, L., Sun, G., Chang, T.-C., Gonzalez, M. O., Mebane, R. H. (2022).
“LIMFAST. I. A Semi-Numerical Tool for Line Intensity Mapping”, preprint, arXiv:2206.14185. https://arxiv.org/abs/2206.14185.
Abstract
We present LIMFAST, a semi-numerical code for computing the progress of reion- ization and line intensity mapping signals self-consistently, over large cosmological volumes and in short computational times. LIMFAST builds upon and extends the 21cmFAST code by implementing modern galaxy formation and evolution models.
Furthermore, LIMFAST makes use of pre-computed stellar synthesis and photoion- ization results to obtain ensemble ionizing and line emission fields on large scales that vary with redshift, following the evolution of galaxy properties. We show LIMFAST calculations for the redshift evolution of the cosmic star formation rate, hydrogen neutral fraction, and metallicity in galaxies during reionization, which agree with current observational constraints. We also display the average signal with redshift, as well as the auto-power spectra at various redshifts, for the 21 cm line, the Ly𝛼intergalactic and background emission, and the Ly𝛼, H𝛼, H𝛽, [O ii]
3727 Å, and [O iii] 5007 Å line emission from star formation. Overall, the LIMFAST results agree with calculations from other intensity mapping models, especially with those that account for the contribution of small halos during reionization. We fur- ther discuss the impact of considering redshift-space distortions, the use of local luminosity and star formation relations, and the dependence of line emission on the ionization parameter value. LIMFAST aims at being a resourceful tool for a broad range of intensity mapping studies, enabling the exploration of a variety of galaxy and reionization scenarios and frequencies over large volumes in a short time scale.
tectable bright sources, LIM takes into account the emission produced by all the star formation present in large areas of the sky; its data, therefore, contains information of the entire galaxy population (Madau et al. 1997; Suginohara et al. 1999; Visbal
& Loeb 2010; Kovetz et al. 2019). This characteristic is especially relevant for high redshift studies, including the epochs of reionization and cosmic dawn. The LIM approach can prove beneficial here because the faint end of the galaxy population may have played a major role during these early times, but this is difficult to explore directly (e.g., Fontanot et al. 2012; Choudhury & Ferrara 2007; Robertson et al.
2015; Yue et al. 2018).
A number of emission lines resulting from different radiative processes and envi- ronments are taken into account for LIM studies, the usual ones being the [C ii] line at 158 𝜇m (e.g., Gong et al. 2012; Silva et al. 2015; Yue et al. 2015; Dumitru et al.
2019; Yue & Ferrara 2019; Sun et al. 2021b), those of the CO molecule (e.g., Righi et al. 2008; Gong et al. 2011; Lidz et al. 2011; Pullen et al. 2013; Li et al. 2016;
Chung et al. 2019; Ihle et al. 2019), the hydrogen 21cm spin-flip transition (e.g., Scott & Rees 1990; Madau et al. 1997; Furlanetto et al. 2006; Chang et al. 2008;
Visbal et al. 2009; Chang et al. 2010; Pritchard & Loeb 2012; Switzer et al. 2013;
Liu & Shaw 2020), and the potentially bright rest-frame optical, ultraviolet lines such as H𝛼, H𝛽, Ly𝛼, He 2, [O ii], and [O iii] among others (e.g., Silva et al. 2013;
Pullen et al. 2014; Visbal et al. 2015; Comaschi & Ferrara 2016; Heneka et al. 2017;
Gong et al. 2017; Visbal & McQuinn 2018; Mas-Ribas & Chang 2020; Heneka &
Cooray 2021; Kannan et al. 2022b; Padmanabhan et al. 2021; Parsons et al. 2021).
Numerically modeling LIM data is of major importance to guide future missions and experiments, but it is computationally challenging because the statistical power of LIM resides in the analysis of emission over large areas of the sky and at several frequencies that are sensitive to small-scale physics. Simulations need to include both, detailed processes related to star formation, as well as the emission and trans- port of radiation in large cosmological volumes. This combination of a broad range of dynamical scales and redshifts is demanding: numerical simulations accounting for resolved galaxy physics typically only allow for the simulation of a small number of galaxies (e.g., Hopkins et al. 2018; Katz et al. 2019; Pallottini et al. 2019; Kan- nan et al. 2020). On the other hand, simulations covering large volumes typically lack numerical precision at the smallest scales, and the processes connected to star formation are often modeled by means of sub-grid prescriptions (e.g., Vogelsberger et al. 2014; Eide et al. 2018, 2020; Shen et al. 2020; Kannan et al. 2022a,b; Lewis
et al. 2022; Shen et al. 2022, and the review by Vogelsberger et al. 2020). Overall, in all cases these simulations typically require very long running times (and/or the need of super computers), on the order of weeks or months and, therefore, they are not well-suited for an exploration of parameter values or of different scenarios in a very short time scale.
With these limitations and constraints in mind, we present here LIMFAST, a semi- numerical tool designed for flexible high-redshift intensity mapping modeling. LIM- FAST aims at self-consistently simulating line emission from galaxies and the in- tergalactic medium, over scales of several hundreds of Mpc and spanning the epoch of cosmic reionization, in a matter of hours with a current personal computer. The fast calculations are possible by using analytical prescriptions and pre-computed radiative processes, and they allow the user to perform large-scale computations following different parameterizations and scenarios in a short time.
As discussed in more detail below, LIMFAST builds upon and uses the 21cmFAST code (Mesinger & Furlanetto 2007; Mesinger et al. 2011) to compute the underlying large-scale structure in large volumes of the universe. This computational step is rapidly achieved because 21cmFAST uses perturbation theory and analytical approaches to the evolution of the density field and formation of collapsed objects.
LIMFAST will then inherit this collapsed structure and will apply the modern galaxy formation and evolution models of Furlanetto (2021) to it to ultimately compute the radiation fields for a number of emission lines, in addition to the original 21cm emission from 21cmFAST. The progress of reionization is simultaneously computed also following the approach in 21cmFAST, but with extensions that self-consistently include the ionizing emission from galaxy populations that co-evolve with redshift.
This paper introduces and details the main structure of the LIMFAST code, and it is the first of a series that will implement further extensions and use LIMFAST to address science questions related to the epoch of cosmic reionization. For example, the second paper, Sun et al. (2022; Paper II hereafter) presents the computations to include the [C ii] 158 𝜇m and CO line emission, and explores the effects of different feedback and star-formation models beyond the fiducial signals presented here. Furthermore, a progenitor version of LIMFAST was presented and used to address the use of intensity mapping to measure the average He 2/H𝛼line ratio for inferring the initial mass function of Population III stars in Parsons et al. (2021).
That code is publicly available under request to the authors. The structure of this present paper is as follows: the models and calculations of LIMFAST are detailed
in Section 5.2, and the fiducial results are shown in Section 7.3. In Section 7.5 we discuss and compare the LIMFAST results with those from the literature. We finally summarize and conclude in Section 5.5.
We assume a flat,ΛCDM,ℎ=67.8 cosmology consistent with recent measurements by Planck Collaboration XIII (Planck Collaboration et al. 2016b) throughout.