In this thesis I investigate galaxy evolution by measuring the luminosity functions of galaxies across a wide range in redshift. I measure the abundances of high redshift galaxies in deep HST imaging of the GOODS-North field from the CANDELS survey. I follow this evolution to lower redshift by measuring the luminosity functions of galaxies in ground-based imaging of the XMM-LSS field from the VIDEO survey with optical data from the CFHTLS. First, at high-redshift, I identify 22 candidate z ≈ 7 and 6 candidate z ≈ 8 − 9 galaxies. By comparing the number of candidate galaxies with those found in the GOODS-South field, I determine that cosmic variance is not the dominant source of uncertainty on the number counts. I constrain the Schechter parameters for the UV luminosity function at z ≈ 7 and z ≈ 8 − 9, finding evidence for evolution in the number density of high redshift galaxies. Next, I present the K_s-band luminosity functions in the 1 degree² and 4.5 degree² overlaps between the VIDEO-XMM field and the CFHTLS-D1 and CFHTLS-W1 fields. I measure the luminosity functions with the 1/V_max method over the range 0.2 < z < 3 in VIDEO-CFHTLS-D1, and over the range 0.2 < z < 1.5 in the shallower VIDEO-CFHTLS-W1 field. I find the evolution of these luminosity functions is best described by luminosity dependent density evolution, where the characteristic magnitude has dimmed at a constant rate since z = 3, while the density has increased since z = 3, first rapidly from z = 3 to z ≈ 1.5 and then more slowly from z ≈ 1.5 to z = 0.2. I measure a significant upturn at the faint end of the luminosity function at low redshift. Finally, I compare the VIDEO-CFHTLS-D1 and VIDEO-CFHTLS-W1 luminosity functions with predicted K-band luminosity functions from the Horizon-AGN simulation. I find both an over-prediction in the numbers of faint galaxies and an under-prediction in the numbers of bright galaxies, implying that the feedback from supernovae is insufficient while the feedback from AGN is over-sufficient.

In recent years the formation of primordial galaxies, cosmic metal enrichment, and hydrogen reionisation have been studied using both refined observations and powerful numerical simulations. High-redshift quasars have become a ubiquitous tool in the study of this era with more than 115 quasars now spectroscopically confirmed at z > 6.0. In this thesis, I use spectra of high-redshift quasars to provide improved observational constraints through a mixture of existing and new techniques. I first investigate the claim of neutral gas around the most distant known quasar, ULASJ1120+0641(J1120), with a cosmological redshift of z=7.1. Its spectrum shows a relatively weak Lyman-α emission line, which has been interpreted as evidence of absorption by neutral gas. Attributing this to a Gunn-Peterson damping wing has led to claims that the intergalactic medium is at least 10% neutral at that redshift. However, these claims rely on a reconstruction of the unabsorbed quasar emission. Initial attempts using composite spectra of lower-redshift quasars mismatched the CIV emission line of J1120, a feature known to correlate with Lyman-α and which is strongly blueshifted in J1120. I attempt to establish whether this mismatch could explain the apparently weak Lyman-α emission line. I find that among a C IV-matched sample the Lyman-α line of J1120 is not anomalous. This raises doubts as to the interpretation of absorbed Lyman-α emission lines in the context of reionisation. I then use a high quality X-Shooter spectrum of the same z=7 quasar to measure the abundances of diffuse metals within one billion years of the Big Bang. I measure the occurrence rates of CIV, CII, SiII, FeII and MgII, producing the first measurement at z > 6 for many of these ions. I find that the incidence of CIV systems is consistent with a continuing decline in the total mass density of highly ionized metals, a trend seen at lower redshifts. The ratio CII/CIV, however, seems to remain constant or increase with redshift, in line with predictions from models which include a decline of the ionising ultraviolet background. The evolution in MgII appears somewhat more complex; while the number density of strong systems continues to decline at high redshift,the number density of weak systems remains high and may even increase. This could signal an increase with redshift in the cross-section of low-ionisation metals. Large numbers of weak MgII systems are also seen at z∼2, suggesting they were already in place when reionisation was ending. I use this X-Shooter spectrum to study metal absorbers associated with the z=7 quasar itself. I find that one such absorber shows signs of only partially covering the line-of-sight, and investigate the possible implications for the quasar’s environment. Finally, I investigate the evolution of the intergalactic medium’s Lyman-α opacity using spectra of quasars at 5.7 < z < 7.1. I assemble a sample of 92 quasar spectra, more than 3 times larger than previous samples. The sample consists of quasars drawn from DES-VHS, SDSS and SHELLQs, new reductions of archival data, and new data. I develop methods to quantify the opacity distribution, providing measurements of the distribution function up to z=6.1. I find that the Lyman-α opacity evolves strongly with redshift. The scatter may be even larger than previously appreciated, posing a serious challenge for models of reionisation.

abstract: Using high-resolution three-dimensional adaptive mesh refinement simulations I study the interaction between primordial minihalo, a clump of baryonic and dark matter with a virial temperature below the atomic cooling limit, and a galaxy outflow. In Chapter 2 I concentrate on the formation of molecular coolants and their effect on the evolution of the minihalo gas. Molecular coolants are important since they allow gas to cool below 10000 K. Therefore, I implement a primordial chemistry and cooling network that tracks the evolution and cooling from these species. I show that the shock from the galaxy outflow produces an abundance of coolants in the primordial gas which allows the gas to cool to below 10000 K. I also show that this interaction produces compact stellar clusters that are ejected from their parent dark matter halos. In Chapter 3 I look at the turbulent mixing of metals that occur between the minihalo and outflow. To do this, I develop a sub-grid model for turbulence that reproduces three primary fluid instabilities. I find that the metals from the outflow are well mixed throughout the minihalo gas. In addition, the metal abundance found roughly corresponds to the observed abundances in halo globular clusters. In Chapter 4, I conduct a suite of simulations that follow this interaction over a wide range of parameters. In almost all cases, the shocked minihalos form molecules and cool rapidly to become compact, chemically homogenous stellar clusters. Furthermore, I show that the unique properties of these clusters make them a prime observational target for study with the next generation of telescopes. Given the unique properties of these clusters there are reasons to suspect that their low-redshift counterparts are halo globular clusters. I outline this comparison in Chapter 5 and give my conclusions in Chapter 6. Finally, I summarize my current work in Chapter 7 and future extensions in Chapter 8. By the end, I hope to convince you that the interaction between a galaxy outflow and a primordial minihalo provides a formation pathway for present day halo globular clusters. / Dissertation/Thesis / Ph.D. Astrophysics 2012

The first billion years of the Universe has been a pivotal time: stars, black holes (BHs), and galaxies formed and assembled, sowing the seeds of galaxies as we know them today. Detecting, identifying, and understanding the first galaxies and BHs is one of the current observational and theoretical challenges in galaxy formation. In this paper we present a population synthesis model aimed at galaxies, BHs, and active galactic nuclei (AGNs) at high redshift. The model builds a population based on empirical relations. The spectral energy distribution of galaxies is determined by age and metallicity, and that of AGNs by BH mass and accretion rate. We validate the model against observations, and predict properties of galaxies and AGN in other wavelength and/or luminosity ranges, estimating the contamination of stellar populations (normal stars and high-mass X-ray binaries) for AGN searches from the infrared to X-rays, and vice versa for galaxy searches. For high-redshift galaxies with stellar ages <1 Gyr, we find that disentangling stellar and AGN emission is challenging at restframe UV/optical wavelengths, while high-mass X-ray binaries become more important sources of confusion in X-rays. We propose a color-color selection in the James Webb Space Telescope bands to separate AGN versus star-dominated galaxies in photometric observations. We also estimate the AGN contribution, with respect to massive, hot, and metal-poor stars, at driving high-ionization lines, such as C IV and He II. Finally, we test the influence of the minimum BH mass and occupation fraction of BHs in low-mass galaxies on the restframe UV/near-IR and X-ray AGN luminosity function.

Galaxies in the redshift range 1 < z < 3 existed during the most vigorous period of star formation in the history of the Universe. In the past 15 years, large rest-frame UV spectroscopic samples of z ~ 3 star-forming galaxies have been assembled. However, this particular redshift range, the so-called Redshift Desert, has only begun to be characterized. Most studies involve low resolution, low signal-to-noise spectra because the small angular size (δ ≤ 1") and faintness (RAB = 24 - 25.5) of high redshift galaxies limit what can be accomplished with a reasonable investment of observing time, even using the world's largest optical telescopes. One way to circumvent these two issues is to study gravitationally lensed galaxies. The magnification boost (up to a factor of 30x) and morphological distortion of a high redshift galaxy by an intervening mass concentration allow for the study of the high redshift Universe in unprecedented detail. I present a detailed analysis of the rest-UV spectrum of two gravitationally lensed galaxies: the 'Cosmic Horseshoe' (zsys = 2.38115) and the 'Cosmic Eye' (zsys = 3.07331). The characterization of the stellar populations and the interstellar gas geometry, kinematics, and composition which I achieve is a preview of the type of information that will be available for unlensed high redshift galaxies with the next generation of optical telescopes. I probe the lower redshift end of the Redshift Desert with a study of Fe ii and Mg ii features in the rest-frame near-UV spectrum of 96 star-forming galaxies in the redshift range 1 < z < 2. Stacked spectra are used to explore average outflow and line profile trends with stellar mass and reddening. I also investigate the phenomenon of emission filling of absorption lines which has implications for the line strength and velocity offset of interstellar absorption lines. Individual galaxies are used to assess the range of outflow velocities as well as the prevalence of emission filling in galaxies from this epoch. This is the first large scale study of fine-structure emission from Feii in high redshift galaxies, both in stacked and individual galaxy spectra. An alternative to investigating galaxies by collecting their light is to study them as seen in absorption against a cosmic backlight, such as a quasar. The Sloan Digital Sky Survey, an imaging and spectroscopic survey which covers about one-quarter of the night sky, has collected many thousands of quasar spectra. I search ~ 44,600 of these spectra, up through Data Release 4, for Mg ii λλ2796,2803 absorption doublets. The final catalog includes ~ 16700 Mgii absorption line systems in the redshift range 0.36 ≤ z ≤ 2.28. Measurements of the absorption redshift and rest equivalent widths of the Mg ii doublet as well as select metal lines are available in the catalog. This is the largest publicly available catalog of its kind and its combination of large size and well understood statistics make it ideal for precision studies of the low-ionization and neutral gas regions of galaxies. I conclude this thesis by suggesting several avenues for extending the studies of high redshift star-forming galaxies presented herein.

This is the second paper in a series on a new luminous z similar to 5 quasar survey using optical and near-infrared colors. Here we present a new determination of the bright end of the quasar luminosity function (QLF) at z similar to 5. Combining. our 45 new quasars with previously known quasars that satisfy our selections, we construct the largest uniform luminous z similar to 5 quasar sample to date, with 99 quasars in the range of 4.7 <= z < 5.4 and -29 < M-1450 <= -26.8, within the Sloan Digital Sky Survey (SDSS) footprint. We use a modified 1/V-a method including flux limit correction to derive a binned QLF, and we model the parametric QLF using maximum likelihood estimation. With the faint-end slope of the QLF fixed as alpha = -2.03 from previous deeper samples, the best fit of our QLF gives a flatter bright end slope beta = -3.58 +/- 0.24 and a fainter break magnitude M-1450(*) = -26.98 +/- 0.23 than previous studies at similar redshift. Combined with previous work at lower and higher redshifts, our result is consistent with a luminosity evolution and density evolution model. Using the best-fit QLF, the contribution of quasars to the ionizing background at z similar to 5 is found to be 18%-45% with a clumping factor C of 2-5. Our sample suggests an evolution of radio loud fraction with optical luminosity but no obvious evolution with redshift.

The ignition of the first sources of light marked the end of the cosmic dark ages, an era when the Universe transitioned from the relatively simple conditions following the Big Bang to the complex tapestry of dark matter, baryons, and pervasive cosmic radiation fields we see today. To better understand this uncharted cosmic epoch, we primarily utilize hydrodynamical, N-body simulations to model the assembly of the first galaxies at redshifts greater than ten and the stars that form within them. These simulations begin from cosmological initial conditions, employ a robust, non-equilibrium chemo-thermodynamic model, and take advantage of adaptive-grid-refinement to probe the multi-scale, complex process of star formation from ab initio principles. We explore the consequences that metal enrichment has on the process of star formation, confirming the presence of a critical metallicity for low-mass star formation. To assess the observational prospects of these primeval stellar populations with next-generation telescopes, like the James Webb Space Telescope, we constrain the star formation efficiency of both metal-enriched and metal-free star formation in a typical first galaxy. We also resolve the formation of individual metal-enriched stars in simulations that ultimately began from cosmological scales, allowing meaningful comparisons between our simulations and the recently discovered ultra-faint dwarf satellite galaxies, the suspected analogs of the first galaxies in the local Universe. / text

In this thesis I used a sample of radio galaxies with 1.4GHz radio luminosity, S1.4 > 10mJy, with the aim of finding high-redshift radio galaxies in the Lockman Hole, ELAIS-N1, ELAISN2, XMM-LSS and CDFS using near-infrared data from the Spitzer Space Telescope, and in the VIMOS4/SSA22 field using near-infrared data from the UKIRT Infrared Deep Sky Survey Deep Extragalactic Survey (UKIDSS-DXS). I used these near-infrared data to filter out low redshift (z < 2) radio galaxies by only including sources with either a very faint identification or non-detection in the K−band or at 3.6μm. I then applied a radio selection criteria based on the compactness of the radio structure in the FIRST survey, to ensure that it was possible to correctly associate a near-infrared identification if detected. Using these filtering critera I reduced the sample from 498 to 220, and I obtained spectroscopic observations for 46 of these. I successfully measured redshifts for 22 of these. Four of these sources were found to have redshifts at z > 3, thus proving the efficiency of the selection. Using these spectroscopically identified sources I investigated the correlation between radio luminosity and Lyα emission line luminosity and showed that these radio sources lie on the same correlations as their more radio-luminous counterparts, suggesting that accretion rate on to the central supermassive black hole determines the power of the AGN, both in terms of photoionizing radiation and radio power. I also investigated the near infrared properties of the host galaxy for these sources. Studying the 3.6μm magnitude versus redshift, I found that many of the radio galaxies in our sample lie on a similar relation as other powerful radio sources studied by different methods, although in my selection there is a bias to the fainter end of the distribution, likely due to the selection bias that I impose. The selection criteria has led us to report on the discovery of the highest redshift radio galaxy ever found, however the robustness of the redshift should be confirmed by follow up spectroscopy as the source has KAB magnitude of ∼ 20.7 which is in an area of the K − z diagram which makes it consistent with both z∼ 1 (in the case of [Oii]) and z ∼ 5 (in the case of Lyα). Finally I determined the number density of high-redshift sources in my sample at different epochs and show that the SKADS simulation is consistent with our observed radio galaxy sample. Using this knowledge I go on to predict the number of sources for current and future near-infrared surveys, where the future appears very bright for this kind of work.

We present initial results from the first systematic survey of luminous z similar to 5.5 quasars. Quasars at z similar to 5.5, the post-reionization epoch, are crucial tools to explore the evolution of intergalactic medium, quasar evolution, and the early super-massive black hole growth. However, it has been very challenging to select quasars at redshifts 5.3 <= z <= 5.7 using conventional color selections, due to their similar optical colors to late-type stars, especially M dwarfs, resulting in a glaring redshift gap in quasar redshift distributions. We develop a new selection technique for z similar to 5.5 quasars based on optical, near-IR, and mid-IR photometric data from Sloan Digital Sky Survey (SDSS), UKIRT InfraRed Deep Sky Surveys-Large Area Survey (ULAS), VISTA Hemisphere Survey (VHS), and Wide Field Infrared Survey Explorer. From our pilot observations in the SDSS-ULAS/VHS area, we have discovered 15 new quasars at 5.3. z. 5.7 and 6 new lower redshift quasars, with SDSS z band magnitude brighter than 20.5. Including other two z similar to 5.5 quasars already published in our previous work, we now construct a uniform quasar sample at 5.3 <= z <= 5.7, with 17 quasars in a similar to 4800 square degree survey area. For further application in a larger survey area, we apply our selection pipeline to do a test selection by using the new wide field J-band photometric data from a preliminary version of the UKIRT Hemisphere Survey (UHS). We successfully discover the first UHS selected z similar to 5.5 quasar.

We report the detection of morphology-dependent stellar age in massive quenched galaxies (QGs) at z similar to 1.2. The sense of the dependence is that compact QGs are 0.5-2 Gyr older than normal-sized ones. The evidence comes from three different age indicators-D(n)4000, H-delta, and fits to spectral synthesis models-applied to their stacked optical spectra. All age indicators consistently show that the stellar populations of compact QGs are older than those of their normal-sized counterparts. We detect weak [O II] emission in a fraction of QGs, and the strength of the line, when present, is similar between the two samples; however, compact galaxies exhibit a. significantly lower frequency of [O II] emission than normal ones. Fractions of both samples are individually detected in 7Ms Chandra X-ray images (luminosities similar to 10(40) - 10(41) erg s(-1)). The 7Ms stacks of nondetected galaxies show similarly low luminosities in the soft band only, consistent with a hot gas origin for the X-ray emission. While both [O II] emitters and nonemitters are also X-ray sources among normal galaxies, no compact galaxy with [O II] emission is an X-ray source, arguing against an active galactic nucleus (AGN) powering the line in compact galaxies. We interpret the [O II] properties as further evidence that compact galaxies are older and further along in. the process of quenching star formation and suppressing gas accretion. Finally, we argue that the older age of compact QGs is evidence of progenitor bias: compact QGs simply reflect the smaller sizes of galaxies at their earlier quenching epoch, with stellar density most likely having nothing directly to do with cessation of star formation.