This thesis presents four optical spectroscopic studies of absorption by matter in the diffuse interstellar medium on scales ranging from 10 AU to 10 kpc. The observations investigate two current problems in interstellar medium (ISM) research: small-scale structure (SSS), and the diffuse interstellar bands (DIBs). Very high spectral resolution observations of interstellar Na I, Ca I, Ca II, K I and CH absorption towards kappa Vel are presented. Combined with observations over the last 15 years taken from the literature, the small-scale structure in front of this star is probed on scales of ~ 10 AU. The high resolution and signal-to-noise of the new observations allow detailed modelling of the absorption profiles and the identification of multiple absorption components. For the two narrowest components, the line profile models are used to constrain the temperature, depletion, electron density and total number density within the structures. Diffuse interstellar bands are used as probes of SSS in long-slit observations of lines-of-sight towards three binary/multiple star systems: HD 168075/6, HD 176269/70 and four members of the mu Sgr system, one of which is identified as a member for the first time. The results show clear variations in DIB absorption in the HD 168075/6 and mu Sgr systems over scales of ~ 50,000-200,000 AU, and demonstrate the efficacy of medium-resolution observations of DIBs for identification of small-scale structure in the ISM. Multi-object spectroscopy of 452 stars in the omega Cen globular cluster is also presented, which probes interstellar absorption by Na I, Ca II and DIBs in two dimensions, on scales of ~ 1 pc. The first detections of diffuse interstellar bands in the M33 galaxy are reported. Multi-object spectroscopy of 43 stars is used to derive spectral types and reddenings, and measure DIB absorption across the disk of the galaxy (~ 10 kpc). Very strong DIB absorption per unit E(B-V) is found for one star in the sample, towards which a total of seven DIBs are detected.

The dynamical evolution of the matter content of the universe is modelled throughout this study as that of self and mutually gravitating Lagrangian fluids in the so called ΛCDM-Concordance cosmological framework which leads to the Hierarchical Clustering paradigm for the formation of cosmic structures. As a numerical tool for investigating galaxy formation scenarios in this context, we employed GADGET2 (see Springel 2005) and the more recent GADGET3 (see Springel et al. 2008): we describe the numerical solvers implemented in the code and test their behaviour in both gravitational and hydrodynamical setups of relevance for cosmological calculations (Tasker et al. 2008). Using the outputs of the MILLENNIUM simulation and the relative Semi Analytical galaxy catalogues produced by Croton et al. 2006, we developed an algorithm aimed at the identification of large spherical underdense regions in the simulated Large Scale Structure (LSS), at z = 0. Focusing on this peculiar environment, we found a confirmation in numerical simulations for the observations by Trujillo, Carretero & Patiri (2006). The Tidal Torque Theory can predict the spatial distribution of the orientation of both the angular momentum vector of Milky Way size galaxies located on the surface of large spherical voids, and of their host DM halos. We re–simulated the 5 GIMIC regions (Crain et al. 2009) following the gravitational evolution of the CDM component only. We then applied a Semi Analytical Model (SAM) of galaxy formation (De Lucia & Blaizot 2007) obtaining the galaxy catalogues and merger histories for the 5 different volumes simulated. It is not yet well understood if and how the LSS environment can influence the Star Formation (SF) histories of galaxies. Starting from the stellar mass content of semi–analytical galaxies at z = 0, we defined characteristic epochs for their build up and, as a preliminary study, investigated how these distribute as a function of different LSS environments.

This thesis describes the use of the rarest high-mass and high-redshift galaxy clusters to constrain cosmology, with a particular focus on the methodology of Extreme Value Statistics (EVS). Motivated by the prospect that even a single sufficiently high mass and high redshift cluster can provide strong evidence against a given cosmology, we first use exact EVS to construct the probability density function (PDF) for the mass of the most-massive cold dark matter (CDM) halo within a fixed redshift volume. We find that the approximation of uncorrelated haloes is valid for high mass haloes 10¹⁵ and large volumes 100⁻¹Mpc, which are also required before the shape of the PDF converges to an asymptotic Generalised Extreme Value (GEV) form. Furthermore, we show the GEV shape parameter γ to be a weak discriminant of primordial non-Gaussianity on galaxy cluster scales. We then extend this analysis to real observations, predicting the PDF for the most-massive galaxy cluster within an observational survey, showing no cluster so far observed is significantly larger than the most-massive expected at its redshift in a concordance cosmology. We also show how the predictions for most-massive cluster with redshift are changed in cosmologies with primordial non-Gaussianity or coupled scalar field dark energy. Finally, we consider why this result appears at odds with some previous analyses, reaffirming that they make use of a biased statistic and showing how an equivalent unbiased one may be constructed. This is then used to rank a comprehensive sample of galaxy clusters according to their rareness, with the cluster ACT-CLJ0102-4915 found to be the most extreme object so far observed. However, the observation of this (and all other clusters so far seen) is shown to be a not unusual event in a concordance universe.

In this thesis, I develop statistics capable of detecting peculiar features in current observations of the cosmic microwave background (CMB) radiation. Such tools scrutinise the very foundations of standard cosmological models. Evidence of peculiar features in the CMB may require a reassessment of these building blocks. More likely, any features may be artefacts of some non-cosmological signal. Nevertheless, whether the origin of these strange attributes is primordial or local, their discovery would be instructive. Existing statistical tools focus on the amplitude of the spherical harmonic coefficients, I look instead at their phases. The method I form checks for the uniformity of the distribution of phase angles using a non-parametric descriptor, which is known as Kuiper's statistic. The method is applied to the COBE-DMR and WMAP sky maps, and departures from uniformity are found in both. The results probably reflect Galactic contamination or the known variation of signal-to-noise across the sky rather than primordial non-Gaussianity. Next, the statistic is adjusted to probe the topology of the universe. The new method exploits the existence of correlations in the phases of the CMB temperature pattern associated with matched pairs of circles seen in the CMB in universes with non-trivial topologies. After this, I turn our attention to the issue of Galactic foreground signals. A diagnostic of foreground contamination is developed based around the Faraday rotation measures (RM) of extragalactic sources. Statistically significant correlations of RM with the preliminary WMAP individual frequency maps are found. These correlations remain significant in CMB-only maps. Later, I use catalogues of rotation measures to construct a template of the Galactic sky. The RM maps may be used as templates for CMB foreground analysis. This idea is illustrated with a cross-correlation analysis between the WMAP data and our maps. I find a significant cross-correlation, again indicating the presence of significant residual contamination. Problems and future developments are discussed at the end.

The intricate analysis of a strong gravitational lens is a complex and computationally demanding problem, requiring the lensed source galaxy's extended light profile to be reconstructed simultaneously with non-linear modeling of the lens galaxy's mass and light. When successful, this analysis gives unrivaled insight into dark matter, cosmology and the most distant Universe. However, such studies remain resigned to small samples, simply due to how long this involved analysis takes. To address this, this thesis presents AutoLens, the first automated framework for comprehensive modeling of a strong gravitational lens's light, mass and source. Reconstruction of the lensed source galaxy uses an adaptive pixel-grid, which is derived in a completely stochastic manner such that a unique pixelization is used for every source reconstruction. This removes biases inherent to pixelized methods associated with the discrete nature of the source-plane. Light profile fitting of the lens galaxy is fully integrated into AutoLens, making it the first method to successfully unify modeling of the lens's light, mass and source into one coherent framework. This allows the method to advocate decomposed mass modeling, which treats separately the lens galaxy's light and dark matter. AutoLens is therefore capable of addressing a diverse range of unique science cases, most notably its ability to determine the central density of a lens galaxy's dark matter halo. These features are incorporated into a fully-automated pipeline, such that the analysis requires no input from the user after an initial setup. This pipeline is tested using a suite of simulated strong lens images which span a variety of source morphologies, lens profiles and lensing geometries. Following the completion of AutoLens's development, the method is ready to analyze the hundreds of archival images of strong gravitational lenses that have been amassed over the past decade, and which are still yet to receive a comprehensive lens analysis. With of order one hundred thousand lenses set to be discovered in the next decade, AutoLens's automated philosophy will be paramount to making analysis of the incoming strong lens samples feasible.

In this thesis, I explore the evolution of the brightest cluster galaxies (BCGs) over the last 10 billion years through detailed studies of both local BCGs from SDSS and their high-z progenitors from CANDELS UDS. First, I study a large sample of local BCGs and link their morphologies to their structural properties. We derive visual morphologies for these BCGs and find that ~57% of local BCGs are cD galaxies, ~13% are ellipticals, and ~21% belong to intermediate classes, mostly between E and cD. There is a continuous distribution in the properties of the BCG's envelopes, ranging from undetected (elliptical BCGs) to clearly detected (cD galaxies), with intermediate classes showing increasing degrees of the envelope presence. A minority (~7%) of BCGs have disk morphologies, with spirals and S0s in similar proportions, and the rest (~2%) are mergers. After carefully fitting the galaxy light distributions using Sersic models, I find a clear link between BCG morphology and structure, such that cD galaxies are typically larger than elliptical BCGs, and the visually extended envelope of cD galaxies is a distinct structure differing from the central bulge. Based on this BCG morphology--structure correlation, I develop a statistically robust way to separate cD from non-cD BCGs, by which cD galaxies can be selected with reasonably high completeness and low contamination. Next, I investigate the effect of environment on the properties of local BCGs by studying the relationship between the BCG's internal properties (stellar mass, structure and morphology) and their environment (local density and cluster halo mass). I find that the size of BCGs is determined by the intrinsic BCG stellar mass, with a weak correlation with the cluster environment. Additionally, more massive BCGs tend to inhabit denser regions and more massive clusters than lower mass BCGs. The growth of the BCGs seems to be linked to the hierarchical growth of the structures they inhabit: as the groups and clusters became denser and more massive, the BCGs at their centres also grew. Moreover, I demonstrate that cD galaxies are ~40% more massive than elliptical BCGs, and prefer denser regions and more massive haloes. My results, together with the findings of previous studies, suggest an evolutionary link between elliptical and cD BCGs. I propose that most present-day cDs started their life as ellipticals at z~1, which subsequently grew in stellar mass and size due to mergers. In this process, the cD envelope developed. This process is nearing completion since the majority of the local BCGs have cD morphology. However, the presence of BCGs with intermediate morphological classes suggests that the growth and morphological transformation of BCGs is still ongoing. Finally, I present a new method for tracing the evolution of BCGs from z~2 to z~0. I conclude, on the basis of semi-analytical models, that the best method to select BCG progenitors at z~2 is a hybrid environmental density and stellar mass ranking approach. Ultimately I am able to retrieve 45% of BCG progenitors. Although the selected high-z progenitor sample is a mixture of BCG and non-BCG progenitors, I demonstrate that their properties can be used to trace BCG evolution. Applying this method to the CANDELS UDS data, I construct an observational BCG progenitor sample at z~2. A local BCG comparison sample is constructed using the SDSS data, taking into account the likely contamination from non-BCGs to ensure a fair comparison between high-z and low-z samples. Using these samples I demonstrate that BCG sizes have grown by a factor of ~3.2 since z~2, and BCG progenitors are mainly late-type galaxies, exhibiting less concentrated profiles than their early-type local counterparts. I also find that BCG progenitors have more disturbed morphologies, while local BCGs have much smoother profiles. Moreover, I find that the stellar masses of BCGs have grown by a factor of ~2.5 since z~2, and the SFR of BCG progenitors has a median value of ~14 Msun/yr, much higher than their quiescent local descendants. I demonstrate that at 1 < z < 2 star formation and merging contribute approximately equally to BCG mass growth. However, merging plays a dominant role in BCG assembly at z < 1. I also find that BCG progenitors at high-z are not significantly different from other galaxies of similar mass at the same epoch. This suggests that the processes which differentiate BCGs from normal massive elliptical galaxies must occur at z < 2.

The observed correlations between the masses of supermassive black holes (SMBH), MBH, with a gravitational influence on parsec scales, and properties of the host galaxy, measured on kiloparsec scales, strongly suggest that the SMBH and galaxy co-evolve. These correlations are likely to be a reflection of a more fundamental connection between MBH and the depth of the potential wells that just fail to prevent gas blow-out, due to feedback from rapid accretion during a quasar-phase. The potential wells in question were dominated by dark matter, and a general method is lacking to connect the stellar properties at z = 0 to properties of their dark matter halos, both at z = 0 and higher redshifts. The work presented here develops a method to make these connections self-consistently. Models of two-component spherical galaxies are used to establish scaling relations linking properties of spheroids at z = 0 (stellar masses, effective radii and velocity dispersions) to properties of the dark matter halo (virial masses and circular speeds), also at z = 0. These models are constrained by combining results from the literature connecting the masses and radii of dark matter halos to each other and stellar masses, with data samples for large, early-type galaxies. The z = 0 properties are then connected to dark matter properties at z > 0 by accounting for the halo redshift evolution. A critical SMBH mass prediction, with dependence on the maximum circular-speed in a protogalactic dark matter halo (MBH ∝ V 4 d,pk), is considered. Combining this with the scaling relations between z = 0 properties and halo properties at z > 0 transforms this theoretical relation into predictions for the observable SMBH correlations. A new prediction is also derived, extending on the MBH ∝ V 4 d,pk relation expected from momentum-driven outflows, allowing for the presence of stars and gas not tracing the dark matter. This new prediction is also compared to the observed correlations at z = 0.

In this thesis I took a combined sample of dust-selected galaxies (from Clark et al. 2015) and HI-selected galaxies (from De Vis et al. 2016). The dust selected sample contains a large fraction of intriguing galaxies dubbed the ‘Blue And Dusty Gas Rich Sources’ (BADGRS), while the HI selected sources revealed another population of blue and gas rich systems which are instead dust-poor. I investigated whether the unique properties of these galaxies could be explained by variations in their recent star formation activity. I showed that the BADGRS are younger, and have typically experienced more recent bursts of star formation compared to the non-BADGRS. Splitting the sample into dust-rich and dust-poor sources showed that both subpopulations are of similar age, although the dust-rich sources have experienced a burst of star formation more recently. I took the chemical evolution model of Morgan & Edmunds (2003), used more recently used in Rowlands et al. (2014), and updated many of the functions and libraries in line with recent literature. I then produced a suite of models to investigate the dust and metal properties of 425 Herschel sources. These models showed (i) a delayed star formation history is required to match the observed star formation rates; (ii) inflows and outflows are required to explain the observed metallicities at low gas fractions; (iii) a significantly reduced contribution of dust from supernovae is necessary to explain the dust poor sources with high gas fractions. We also showed the dust-to-metal ratio is not definitively constant in all galaxies, and that there is evidence for a decrease in the dust-to-metals ratio towards lower metallicity. This thesis proposes a model in which the dust, gas, metals and stars can be modelled in a consistent and coherent manner, and gains insight into the dust-to-gas evolution at early epochs.

Outflows are ubiquitous in active galactic nuclei (AGN). They can take the form of either dramatic radio jets, which extend vast distances into the inter-galactic medium, or of much smaller scale winds - whose existence can be inferred only indirectly via their influence on the observed spectra. There is good evidence to show that winds are likely to arise from the accretion disks thought to form the central engine of all AGN; they should therefore be observable in all such systems. The clearest observational signature comes in the form of the broad, blue-shifted absorption features seen in the spectra of broad absorption line quasars (BALQSOs) and the aim of this work is to investigate how the geometry and physical parameters of disk winds affect their absorption spectra. We first discuss the changes made to an existing Monte Carlo radiative transfer code, python, in order to extend its capabilities to include modelling of AGN. These changes include the implementation of an approximate ionization scheme which takes account of arbitrary illuminating spectral energy distributions (SEDs), and the inclusion of heating and cooling effects likely to be important in the presence of high energy photons. Next, we describe the second stage of the project which was to gain insight into the general properties of a wind exhibiting broad absorption line (BAL) features. We did this by performing radiative transfer and ionization calculations on a simple kinematic wind representation using python. We show that BAL features can be produced for plausible AGN parameters. Finally, we present calculations carried out on a more complex wind geometry, generated from a hydrodynamic simulation of a line driven disk wind. This calculation does not produce BAL features at all, and in fact the wind is too highly ionised to permit efficient line driving. This result is significant because it illustrates the importance of detailed radiative transfer in hydrodynamic modelling, and suggests future work.

This thesis probes the role of environment in galaxy evolution, focussing particularly on understanding the links between the truncation of star formation, the transformation of galaxy structure, and environment. This study utilises deep HST imaging, photometric and spectroscopic data for galaxies within ten high-z cluster fields, which form part of the ESO Distant Cluster Survey (EDisCS). I first compare the mass--size relations of cluster and field galaxies, to address the dependence of galaxy size on environment observed from z~2.5, and which seemingly disappear at lower redshifts. I find no significant difference in the size distributions of cluster and field galaxies of a given morphology, or with similar rest-frame B-V colours. I rule out average size differences larger than 10--20 % in both cases. Thus, I conclude that if the size difference at higher-z reported in the literature is real, the growth of field galaxies seem to have caught up with that of cluster galaxies by z~1. Any putative mechanism responsible for galaxy growth has to account for the existence of environmental differences at high redshift and their absence (or weakening) at lower redshifts. I then move on to analyse the effects of the global cluster/field environment on the internal structure of galaxies and their colours. I introduce quantitative non-parametric measures like the residual flux fraction (RFF) and the asymmetry in galaxy residuals (A_res) which measure deviations from symmetric light distributions using HST images, to explore the internal structure of galaxies. I also obtain complementary information on the probable causes of structural disturbances, both internal and external in nature, by performing visual classifications of cluster and field galaxies. Combining these two approaches of measuring galaxy structure, it is found that the RFF is a good proxy for `roughness' in the surface brightness distribution, while A_ res is more sensitive to the causes of the structural disruption. Incorporating visual morphology and environment, it was found that the external causes of disturbances were most often associated with star formation in spiral galaxies. When adding information on the star formation activity of galaxies, I discover two complementary subpopulations of galaxies abundant in clusters: visually undisturbed passive spirals, and undisturbed star-forming lenticulars. In addition to being visually symmetric, these passive cluster spirals are also consistently smoother than their star-forming counterparts. These observations, therefore, strongly advocate gentle physical processes acting on the gas content to modify the star formation properties of galaxies accreted into clusters, without large-scale disturbances in their stellar structure. Considering the variations of quantitative galaxy structure with the star formation history of galaxies, I find that the young, star-forming galaxies are consistently rougher and more asymmetric than the galaxies with older passive stellar populations. Further, the galaxies with different average stellar ages seem to have similar distributions of RFF and A_res over cluster/field environments, thereby emphasising that the star formation history of galaxies is strongly linked to their intrinsic structure alone. Finally, complementing the global cluster/field environment, I explore the projected phase--space as a tool to investigate possible variations in galaxy structure and their stellar ages over the internal cluster environment. The analysis with the projected phase--space shows a decrease in the fraction of galaxies with younger stellar populations in the cluster core when separated by morphology, especially for spirals. This trend, however, is less pronounced in the observed distributions of RFF and A_res across the projected phase--space and the field. All these observations, when put together, signify that the star formation in galaxies is shut down as they get accreted into clusters, while the internal structure of galaxies remains more or less unaffected. The actual morphological change in galaxies, therefore, will follow later after the star formation has already been truncated. Furthermore, the physical mechanisms driving these transformations would, therefore, be gas extinguishing ICM processes like ram-pressure stripping and starvation, which leave the galaxy structure undisturbed.