with Intensity Mapping

Sun participated in project conception, scientific analysis, and manuscript writing. Sun participated in the conception of the project, performed the scientific analysis, and wrote most of the manuscript.

INTRODUCTION

An Overview of the LIM Technique

As shown in Figure 1.2, the LIM measures the 3D fluctuations of the target line signal by acting as an imaging spectrometer. The resulting line intensity map provides a coarse-grained representation of the LSS, which is shown by the distribution of the dark matter halo in Fig. 1.2.

Figure 1.1: A world map of past, ongoing, and planned LIM experiments.

LIM: Scientific Applications

Kittiwisit et al. 2022) and higher-order statistics such as bispectrum and trispectrum are also commonly considered to avoid the loss of information due to non-Gaussianity. 2020a; Greig et al. 2022) was also recently proposed to extract non-Gaussian features from LIM data.

Thesis Outline

In Chapter 8, I describe a new way to constrain the global star formation law of galaxies using LIM measurements of the BAOs. Finally, in Chapter 9, I briefly discuss the prospects for the IM concept in the era of upcoming cosmological surveys with multiple probes.

PROBING COSMIC REIONIZATION AND MOLECULAR GAS GROWTH WITH TIME

Introduction

The coevolution of the cosmic molecular gas density and the SFRD is therefore of great interest (Popping et al. 2014; Decarli et al. 2016). Measurements from the ALPINE study are shown by the hexagons for sources with dust continuum detection (Béthermin et al. 2020).

Figure 2.1: The observed correlation between [C ii] luminosity and the total SFR (UV + IR) of galaxies in the local universe and 𝑧 ≳ 5

Observables for TIME

• Observables Internal to TIME Datasets
• Observables Requiring Ancillary Data

Models

• Tracers of Large-Scale Structure
• Carbon Monoxide and Neutral Carbon Near Cosmic Noon
• Low- 𝑧 NIR-Selected Galaxies
• Ionized Carbon During the EoR
• High- 𝑧 LAEs
• Molecular Gas Content
• Reionization History

We note that [C ii ] luminosity is known to be affected by physical conditions in the PDR in several ways (Ferrara et al. 2019). Theoretical models (e.g. Lagache et al. 2018) are in slight tension with existing constraints on the [C ii ] luminosity function.

Figure 2.2: A comparison of the CO(1–0) auto-correlation power spectra predicted by our fiducial model with results in the literature

Mock Observations

• Survey Strategy
• Sensitivity Analysis

Consequently, the two-point statistic is described by the 2D power spectrum defined in the observed instrument tracking frame, which is related to the theoretical 3D power spectrum defined in equation (2.18) by the survey window function. Here we follow Gong et al. 2012) to provide an overview of sensitivity analysis based on the mode counting method.

Figure 2.6: 2D binned [C ii] auto-power spectra. The 2D power spectra are mea- mea-sured in TIME low- 𝑧 /HF (left) and high- 𝑧 /LF (right) sub-bands and binned in 𝐾 ⊥ (perpendicular to the LOS) versus 𝐾 ∥ (parallel to the LOS) space

Results

• Constraints on [C ii] Intensity

We note that the finite spatial and spectral resolutions of the instrument will also affect the minimum physical scales, or equivalently 𝐾⊥,max and 𝐾∥,max, that can be probed. It is assumed that the probability of the data vector ˆ𝑥 for a given set of model parameters ˆ𝜃 is described by a normal distribution.

TIME-EXT TIME

EoR Constraints Inferred From [C ii] Measurements

To illustrate the information that TIME adds to our understanding of the EoR history, we consider two contrasting cases, namely, whether or not TIME data should be combined with other EoR constraints, including the Thomson integral constraint on optical depth of CMB -photons and constraints at the end scattered. of the EoR from quasar absorption spectra. Specifically, to include these observations as independent constraints in the MCMC analysis, we compare predictions from our reionization model (assuming Gaussian statistics) with 𝜏𝑒 Planck Collaboration et al. 2016a) and 1− 𝑥¯𝑖(𝑧 = 5.5) < 0.1 which represents an updated, albeit conservative, constraint on the IGM neutrality near the end of reionization from quasar observations at 𝑧 ≲ 6 (e.g. et al.

TIME TIME-EXT Carilli+16

CMB+QSO TIME+CMB+QSO

Probing Physics of Molecular Gas Growth with CO and [C I] Intensities Here, we consider two potential applications of in-band cross-correlation to mea-

Top: joint posterior distributions of the free parameters derived from cross-correlation of the CO(4–3), CO(5–4) and [CI] lines at 𝑧 ~1.1. 68% and 95% confidence intervals for the cross-effect spectra, derived from 1000 random samples of the posterior distribution, are shown.

Figure 2.12: Sensitivity to the [C ii]–LAE angular cross-correlation function. The correlation between the [C ii] line intensity and LAEs surveyed by Subaru HSC at 𝑧 = 5

CO–Galaxy Cross-Correlation

The amplitude and detectability of the shot power of CO galaxies is sensitive to the selection threshold of galaxy samples. As shown in the bottom panel of Figure 2-16, as the selection threshold (measured in critical halo mass or stellar mass) increases, the selected galaxies approach the brighter end of the CO luminosity function.

Figure 2.16: Sensitivity to and mean CO luminosities inferred from CO–galaxy power spectra

Foreground Contamination and Mitigation Strategies

• Continuum Emission
• Spectral Line Interlopers

The remaining channels at the high and low frequency edges (16 in total) serve as atmospheric monitors (Hunacek et al. 2016). Several mitigation strategies have been proposed, including the use of cross-correlation (Silva et al. 2015), masking (Breysse et al. 2015;.

Discussion

• Implications and Limitations of Power Spectral Constraints from TIME and TIME-EXT
• Next-generation [C ii] LIM Experiment

Dumitru et al. 2019), we estimate that, for a 10 degree2 survey and a TIME-NG-like capacity with 3000 hours of integration, the [C ii]-LAE cross-correlation with an expected Roman General Observer (GO) survey (Spergel et al. . 2015) of the same size and a depth of 𝑚AB. The bottom panel of Figure 2.17 shows the [C ii]–Ly𝛼 cross-power spectrum, estimated from the Ly𝛼 power spectrum of Heneka et al.

Figure 2.17: Synergies between TIME-NG and surveys of LAEs and Ly 𝛼 intensity fluctuations

Conclusions

Using the [C ii ] and LAE models presented in this work, together with physical models of Ly𝛼 and the 21 cm line motivated by observations (e.g. Gong et al. 2012 ; at lower redshifts, we expect significant detections of cross-power spectra during TIME Maps of CO and galaxies at known redshifts.

Appendix: Modeling the Star Formation Efficiency .1 Dust Correction.1Dust Correction

• The Star Formation Efficiency as a Function of Halo Mass

ATC was supported by a KISS postdoctoral fellowship and a National Science Foundation for Astronomy and Astrophysics under No. Grant. mass dependence at the end of low mass.

Figure 2.18: The star formation efficiency 𝑓 ∗ as a function of halo mass. Curves corresponding to different choices of 𝜉 , as defined in Equation (2.43), are shown to illustrate how our model captures the uncertainty in the mass dependence at the low-mass

Appendix: Window Function

Note that when 𝜉 < 0, we impose a ceiling on 𝑓∗ so that it asymptotes to a constant value instead of blowing up. We note that for the considered line scan we must normalize equation (2.47) by dividing it by𝑉𝑆.

Appendix: Uncertainties of Auto- and Cross-Power Spectra

The corresponding selection function can be specified by a product of top-hat functions, which means a weighting function in the shape space 𝑘. Note that for the Fourier transform ˜𝑓 of a real field, the first terms in the above two expressions vanish due to the Hermitianity condition ˜𝑓∗(𝑘) = 𝑓˜(−𝑘) and the fact that the different modes 𝑘 are statistically independent.

A FOREGROUND MASKING STRATEGY FOR [C II]

INTENSITY MAPPING EXPERIMENTS USING GALAXIES SELECTED BY STELLAR MASS AND REDSHIFT

Introduction

Advances in the investigation of individual galaxies at high redshift in the near-infrared (e.g. Ellis et al. 2013; C ii] is a particularly promising probe for mapping the line intensities of the reionization epoch (e.g. Gong et al.

Methods for Modeling Infrared Galaxies as CO Proxies

• Experiment Overview
• Power Spectrum of CO Foreground
• Masking Strategy
• Residual Foreground Tracers

The entire CO foreground power spectrum can be written as the sum of the clustering and shock noise terms. The shaded bands represent the typical uncertainty in the inferred masking fraction due to errors in fitting 𝐿′.

Table 3.1: Map and catalog information.

Conclusions

Given the uncertainties in [C ii ] signal strength and CO contamination (see Figure 3.6), it is desirable to test the level of residual CO in the foreground after using the voxel masking technique to determine whether the foreground has been sufficiently exposed to remove. 2015) discuss the utility of cross-correlation as a way to constrain the foreground rate after masking. CO–CO cross-correlation can be performed within the experiment's own data set, albeit at the expense of potentially lower sensitivity after masking.

Appendix: Validation of the Stacking Method

The middle panel shows the ratio between the scatter measured by our frame-stacking formalism and that predicted by the flux distribution (0.3 dex) to generate the simulated maps. Sources not in the observed bin differ by 0.3 dex in all three cases.

Figure 3.11: Robustness of measuring the mean flux densities with SIMSTACK.

Appendix: Effect of masking on the [C ii] power spectra

These two models include the expected uncertainty in the [C ii ] signal during the EoR (more precisely at 𝑧=6.5) due to the uncertainty in the SFR powering these emissions. Another important source of uncertainty in the amplitude of the signal [C ii] is the evolution of.

Figure 3.13: Simulated effects of masking on power spectra. Power spectra of CO (projected) and [C ii] emission computed from simulated intensity maps before (solid) and after (dotted) Case A masking as illustrated in Figure 3.7.

According to these CO/[C ii] models, the masking described in this paper would efficiently reduce the CO signal. Different colors indicate cases where the simulated intensity maps contain different combinations of signal and foreground lines. line ratios) of different CO transitions and therefore interpret the cross-correlation measurements of adjacent CO lines.

Figure 3.14: Quantifying residual foregrounds with cross-correlation. Cross-power spectra between observed intensity maps at frequencies 216.1 GHz ( 𝑧 C

A SELF-CONSISTENT FRAMEWORK FOR MULTI-LINE MODELING IN LINE INTENSITY MAPPING EXPERIMENTS

Introduction

Due to sensitive FIR observations from experiments such as Planck (Planck Collaboration XXX 2014) and Herschel (Viero et al. 2013b), the CIB has been the subject of detailed modeling efforts. In particular, analytical models that relate the infrared (IR) luminosity of galaxies to the mass and redshift of their host haloes have been successful in reproducing the statistical properties of the CIB (e.g. Shang et al. 2012; S16; Wu & Doré 2017b ).

A Simple Analytic Model of Mean Halo Properties

• IR luminosity
• Star Formation History
• Dust Mass
• Hydrogen Mass
• Metallicity

We work in the aforementioned framework of the halo model of CIB anisotropies introduced by Shang et al. In Figure 4.2 we plot the redshift evolution of the dust density parameter Ωd, which is compared to a compilation of previous dust abundance measurements by Thacker et al.

Table 4.1: Fiducial Parameters of CIB Model

Models of Emission Lines

• H i 21cm Line

Therefore, in the context of the CIB model, a mass-independent gas metallicity is likely a reasonable approximation. These effects are most pronounced in gas of extreme density and temperature and may be responsible for the breakdown of the 𝐿IR–𝐿[C.

Figure 4.5: The redshift evolution of the metallicity 𝑍 . The 𝑍 evolution derived from our model is compared with semi-analytic estimates of gas-phase 𝑍 from Fu et al

L122 [NII]/L205[NII]

CO(1–0) Line

The CO(1–0) rotational transition (𝜆=2.6 mm) is a powerful tracer of the molecular gas content of individual molecular clouds as well as of galaxies (e.g. Solomon et al. Under these assumptions, we can write the CO brightness directly in terms of the molecular mass of gas 𝑀H.

Intensity Mapping Framework .1 Modeling the Fluctuation Signals.1Modeling the Fluctuation Signals

• Sensitivity Analyses

The total power spectrum (solid curve) rescaled from the one without scattering (filled squares) agrees well with that obtained from averaging over 1000 random realizations (open squares). The (average) power spectrum of thermal noise is scale independent and can be expressed as.

Figure 4.8: The effect of scatter on the power spectrum. The two-halo (dash- (dash-dotted curve) term is rescaled by the correction factor defined by Eq

Comparison to Existing Observational Constraints

In comparison, deep-field results from Switzer et al. 2013) are shown by the teal triangles, which will be interpreted as upper limits when residual foreground is present. The latest observational constraint on 𝑏[C. 95% confidence level) derived from the cross-correlation between Planck maps and galaxy surveys (Yang et al. 2019).

Figure 4.9: Observational constraints on H i density, temperature, and power. Top:

NII] 122

NII] 205

Inferring ISM Properties from Auto/Cross-Correlations

• Case I: Multi-Phase Diagnosis with H i and CO
• Case III: Probing H ii Regions with [N ii] Lines

Left: dummy data sets of the observed [C ii ] auto-power spectrum and [Cii] ×Hi, [Cii] ×CO and CO×Hi cross-power spectra at 𝑧∼ 2. Here we consider two types of measurements, namely, the cross-power spectrum of the two [ N ii] lines and their respective auto-power spectra.

Figure 4.13: Parameter constraints from mock H i and CO data sets. Left: mock data sets of the observed H i and CO auto power spectra at 𝑧 ∼ 2

Discussion and Conclusion

Furthermore, given that this effect appears to be a strong function of galaxy luminosity, this may have important testable implications for the predicted [C ii ] power spectra. In general, the LIMFAST results agree with calculations from other intensity mapping models, especially those that take into account the contribution of small haloes during reionization.

Introduction

As discussed in more detail below, LIMFAST builds on and uses the 21cmFAST code (Mesinger & Furlanetto 2007; Mesinger et al. 2011) to calculate the underlying large-scale, large-volume structure of the universe. We assume a flat,ΛCDM,ℎ=67.8 cosmology that is consistent with recent measurements by Planck Collaboration XIII (Planck Collaboration et al. 2016b).

LIMFAST: the Code

• Galaxy Formation and Evolution Model
• Stellar and Nebular SEDs

This calculation differs from the star formation rate because the metallicity is not an additive quantity, and therefore one cannot integrate the metallicity per hello over HMF. For the calculation of nebular line emission, we use our stellar SEDs as the incident spectrum in the photoionization code Cloudy (version 17.02, Ferland et al. 2017) and the following quantities describing the nebular medium;.

LIMFAST

Ionization Calculation

Because the metallicity of the halos changes with time, the number of ionizing photons also evolves with the redshift. Finally, the ionizing radiation is used to calculate the ionization of the intergalactic medium and the progress of reionization.

IGM Ly 𝛼 Emission

The luminosity of all haloes in a given cell is integrated with the halo mass function calculated by 21cmFAST from the density field and the collapse of structure. Unlike Ly𝛼, H𝛼 would be the brightest of these hydrogen lines, but given the value of its recombination coefficient and the transition probabilities between the atomic energy levels of the hydrogen atom, the expected luminosities for H𝛼 are expected to be about one order of magnitude dimmer than those of Ly𝛼 .

Ly 𝛼 Background

While the Ly𝛼 background is responsible for the WF coupling as mentioned above, the intergalactic gas heating is mainly dominated by the X-ray background field. For the calculation of the 21 cm signal, LIMFAST follows the methodology developed in 21 cmFAST, except for the background derivation Ly𝛼.

Redshift-Space Distortions

The correction shown in the last term of equation 5.20 is the one we will apply to the intensity of the optical and ultraviolet lines calculated by LIMFAST. However, this limit is not required and is therefore used in the calculations of the 21 cm emission.

Results

• Redshift Evolution
• Star Formation Rate Density
• Metallicity
• Neutral Fraction
• Line Emission
• Power Spectra

The solid black lines in Figure 5.6 show the redshift evolution of the line emission luminosity calculated by LIMFAST. For the case of the oxygen lines, the differences between the two approaches again depend strongly on the ionization parameter taken into account in the photoionization calculations.

Figure 5.2: Cosmic star formation rate density from LIMFAST. For comparison, we plot also the evolutions from other works as solid colored lines, and the data from Bouwens et al

Comparison to Previous Work

• Comparing Star Formation Results
• Comparing Line Emission and Power Spectra Results

We show how changing the minimum halo mass parameter in LIMFAST affects the star formation rate density in Figure 5.8. We attribute the differences between the two codes to the different star formation recipes used in each case.

Figure 5.8: Contributions to the cosmic star formation rate density. The different line types consider different minimum halo masses and halo mass ranges able to host star formation

Conclusion

Finally, we note that Comaschi & Ferrara (2016) pointed out an inaccurate treatment of the ionization history of the IGM in Pullen et al. 2014), which may contribute to the observed signals. Finally, the choice of the ionization parameter value has a dramatic impact on the amplitude of the [O ii] 3727Å and [O iii].

Appendix: Luminosity Dependence on 𝑈 and 𝑍

We have also shown that the luminosity and star formation relationships derived from local universe observations commonly used in the literature are not well suited for investigation during reionization, due to the dependence of the emission on the metallicity. In this first article of the LIMFAST series, we introduced the general structure of LIMFAST and some of its capabilities and calculations.

Appendix: Power Spectra of Optical and UV Line Emission

Appendix: 2D Power Spectra with Redshift-Space Distortions

OII]

OIII]

Introduction

Such a generalization allows us to relate specific LIM observables to a fundamental picture of high-𝑧 galaxy formation described by a balance maintained by star formation from the ISM and stellar feedback typically from supernovae (Furlanetto et al. 2017; Furlanetto 2021). Throughout the paper, we assume cosmological parameters consistent with recent measurements from Planck Collaboration XIII (Planck Collaboration et al. 2016b).

The LIMFAST Code

We compare our results with previous work, discuss some limitations and caveats to the analyzes presented, and outline several possible extensions of the current framework in Section 6.5, before summarizing our main conclusions in Section 7.6. Extensions and variations to the basic model presented in Paper I, including an extended model of emission lines originating mainly from the neutral ISM (e.g. [C ii] and CO), are introduced in this work to facilitate the analysis.

Models

• Models of Galaxy Formation and Evolution