The variability of emission from active galactic nuclei (AGN), both intrinsic, due to changes in the emitting regions, and extrinsic, due to absorption and scattering from material near to the black hole, can reveal a wealth of information about their geometry and accretion behaviour. In this thesis, I use analysis of the spectral variability of the Xray, UV/optical emission from AGN to probe the properties of the material surrounding supermassive black holes at their centres. Firstly, I present constraints on the variability of the photon index of the X-ray spectrum of 24 Swift AGN from the Palomar sample of galaxies. The change in the behaviour of AGN from ‘softer-brighter’ to ‘harder-when-brighter’ at a critical accretion rate, seen previously in other samples of AGN, is found. ‘Harder-when-brighter’ behaviour is measured in several sources individually, achieved previously in only one AGN. I present a study of the long-term (months-years) X-ray spectral variability of three Seyfert AGN known to exhibit strong absorption variations. Trends in the long-term behaviour of the absorbers, in particular a strong anti-correlation between the observed absorbing column and the intrinsic luminosity of NGC 1365, are explained using a variable wind model. In this model, the radius at which the wind arises is dependent on the luminosity of the central engine, leading to correlated long-time-scale changes in the observed absorption properties. Finally, time lags between the X-ray and UV/optical variability of the Seyfert AGN NGC 4395 are measured, using cross-correlation techniques. The lags fit the ?4/3 dependence expected from reprocessing of X-rays in the accretion disc. Modelling of the expected lags from a standard Shakura-Sunyaev thin disc with the previously-measured physical parameters of the system is then carried out. The results are consistent with the data, further supporting reprocessing as the principle source of UV/optical variability in NGC 4395.

Outflows are ubiquitous in accreting systems across 10 orders of magnitude in mass, and there is good evidence that mass-loaded winds are launched from the accretion discs of quasars and cataclysmic variables (CVs). Perhaps the most spectacular evidence for accretion disc winds is the blue-shifted, broad absorption lines (BALs) in UV resonance lines, seen in CVs and the BAL quasars. As well as imprinting absorption features, disc winds may affect the line and continuum emission from accreting objects. They thus offer a natural way to unify much of the phenomenology of CVs and active galactic nuclei (AGN). In this thesis, I use a state-of-the-art Monte Carlo radiative transfer (MCRT) code, Python, to conduct a series of simulations designed to test simple biconical disc wind models. I outline the MCRT techniques used, before describing a series of code validation exercises. Having tested my methods thoroughly, I explore whether the winds that are responsible for the UV BALs in high-state CVs could also have an effect on the optical spectrum. I find that the wind produces strong emission in the Balmer series, He ii 4686 A and a series of He i lines. The model shows the observed trends with inclination and in some cases produces sufficient recombination continuum emission to fill in the Balmer photoabsorption edge intrinsic to disc atmospheres. The results suggest that disc winds could have a significant impact on the optical spectra of high-state CVs. The next step was to apply the techniques to quasar winds in a test of disc wind unification models. In previous e?fforts, the ouflow tended to become `over-ionized', and BAL features were only present if the X-ray luminosity was limited to around 1043 erg s-1. The out ow also failed to produce significant line emission. Motivated by these problems, I introduce a simple treatment of clumping and find that it allows BAL features to form in the rest-frame UV at more realistic X-ray luminosities. The fiducial model shows good agreement with AGN X-ray properties and the wind produces strong line emission. Despite these successes, the model cannot reproduce all emission lines seen in quasar spectra with the correct equivalent width (EW) ratios, and I find the emission line EWs have a strong dependence on inclination. Informed by the quasar wind modelling, I examine the emission line EW distributions of quasars in the context of geometric unification. I find that the observed distributions are not consistent with a model in which an equatorial BAL outflow rises from a foreshortened accretion disc. I discuss this finding in the context of other observational orientation indicators. Finally, I summarise my findings and suggest avenues for future work. Overall, the work presented here suggests that disc winds matter. They not only act as a spectral 'filter' for the underlying accretion continuum, but may actually dominate the emergent spectrum from accreting objects. As a result, unveiling their driving mechanisms, mass-loss rates and ionization structure is an important goal for the astronomical community.

This thesis contains the first systematic X-ray investigation of the relationships between the properties of different types of radio-loud AGN and their large-scale environments, using samples at two distinct redshifts to isolate the effects of evolution. I used X-ray observations of the galaxy clusters hosting the radio galaxies to characterise the properties of the environments and compared them with the low-frequency radio properties of the AGN. I found a strong relationship between radio luminosity and ICM richness for lowexcitation radio galaxies (LERGs) but not for high-excitation radio galaxies (HERGs), and tentative evidence of evolution of the environment for HERGs, but not for LERGs. I then used the X-ray emission from the lobes of the FRII galaxies in the samples to calculate their internal pressures, finding them to be at slightly higher pressures than predicted by equipartition and near pressure balance with their external environment at the lobe mid-points. I made the first estimates of lower limits for lobe-tip Mach numbers for a large sample of lobes; suggesting that about half the lobe tips were driving strong shocks. Finally I verified that the correlation between LERG radio luminosity and cluster X-ray luminosity is not caused by a common correlation with black hole mass, and also found no evidence that HERG properties are affected by ICM richness, adding evidence for theories of local fuelling for HERGs. I found evidence that HERGs should have a strong relationship between jet power and radio luminosity, but that LERG jet power must be subject to factors additional to black hole mass. I also found evidence that the same mass black hole is associated with lower radio luminosities in FRI morphology galaxies than in FRIIs, perhaps as a result of the higher proton content of FRI jets leading to lower radiative efficiency.

Correlations of the observed sizes and luminosities of galaxies can be used to estimate the magnification that arises through weak gravitational lensing. However, the intrinsic properties of galaxies can be similarly correlated through local physical effects, and these present a possible contamination to the weak lensing estimation. In this thesis we model these intrinsic correlations using the halo model, assuming that both the sizes and luminosities of galaxies reflect the mass in the associated halo, assuming the observed galaxy properties correlate closely with the mass of the haloes and sub-haloes. Larger and more luminous galaxies live in more massive haloes, and even if the sub-halo population is largely independent of the halo mass, the sizes and luminosities of the largest sub-haloes will still be limited by the total halo mass. We use this simple model to predict what would be observed for a magnification estimator based on galaxy sizes and magnitudes, and how the intrinsic signal correlates with the true lensing convergence. Additionally, we include in our analysis the effects of cuts in the sample and we model the size-magnitude distribution and study how the cuts in the survey affect the intrinsic mean size and magnitude as well as the inferred convergence power spectrum when using galaxy sizes and magnitudes. Studying these correlations is important both to improve our understanding of galaxy properties and because they are a potential systematic for weak lensing size magnification measurements. Our model assumes that the density field drives these intrinsic correlations and we also model the distribution of satellite galaxies. We calculate the possible contamination to measurements of lensing convergence power spectrum from galaxy sizes and luminosities, and show that the cross-correlation of intrinsic properties with convergence is potentially an important systematic. We also explore how these intrinsic correlations may affect surveys with different redshift depth. We find that, in this simple approach, intrinsic size and luminosities correlations cannot be neglected in order to estimate lensing convergence power spectrum for constraining cosmological parameters.

An accepted explanation for the accelerated expansion of the late-time Universe is to modify the Einstein equation, either by adding a component to the energy-momentum tensor via dark energy, or to the Einstein tensor via a modification to gravity. The second of these options often involves the introduction of a scalar field, which couples to the matter components of the Universe and gives rise to a fifth force, of the same order of magnitude as gravity. Through a variety of experiments and astronomical observations, this fifth force has been demonstrated to be negligible on Terrestrial and Solar System scales. Therefore if it does act on large scales, it must be suppressed, or `screened', on small scales. In this thesis, I place constraints upon one of these screening methods, chameleon gravity. Chameleon gravity postulates the existence of a scalar field that couples with matter to mediate a fifth force. If it exists, this fifth force would influence the hot X-ray emitting gas filling the potential wells of galaxy clusters. However, it would not influence the cluster's weak lensing signal. Therefore, by comparing X-ray and weak lensing profiles, upper limits can be placed on the strength of a fifth force. To do so I first present two hydrodynamical simulations, one evolved under CDM+GR and the other under f(R). From these two simulations I generate X-ray surface brightness and weak lensing profiles for a number of simulated clusters. Using these profiles I test many of the assumptions of the technique used to constraint If_R0I. I then use these profiles to test the analytic pipelines developed to constrain f(R) gravity by applying a full MCMC analysis. From doing so I find constraints on the modified gravity parameters of If_R0I< 8:3 x10^-5. Next I outline the creation of a sample of 58 clusters, including 12 new to the literature, with high quality weak lensing data from CFHTLenS and X-ray data from XCS. By stacking these clusters I use a multi-parameter MCMC analysis to constrain the two chameleon gravity parameters ( ß and φ_∞). The fits are consistent with general relativity, not requiring a fifth force. In the special case of f(R) gravity (where ß = √1/6), I set an upper limit on the background field amplitude today of If_R0I< 6 x 10-5 (95% CL). This is one of the strongest constraints to date on If_R0I on cosmological scales. These fits are also found to be consistent with those recovered from the f(R) simulations. Finally I look at the future of this method, beginning with forecasting the constraints that this technique will be able to place on f(R) gravity using the Dark Energy Survey, finding If_R0I> 2 x 10^-5. Next I discuss how the X-ray surface brightness profiles might be improved by removing contaminating point sources from the X-ray images and find that doing so leads to a reduction in the error bars of 5%. I end this thesis by detailing how the techniques discussed within can be applied to constrain other modified gravity theories, namely the Vainshtein mechanism. Doing so I am able to place competitive constraints upon Vainshtein gravity, including the first ever constraint on a particular parametrisation.

Galactic bars play in an important role in the evolution of disk galaxies. Bars also act as a tracer for dynamically cool, `mature' disk galaxies. Therefore, the observational study of bars, especially at higher redshifts, provides an invaluable insight into how the disk galaxy population has evolved as the Universe has aged. In this thesis, I present research that explores the redshift evolution of the bar fraction since z = 1, and the first look at the fraction of disk galaxies hosting bars out to z = 2. My research combines optical and infrared observations, taken by the Advanced Camera for Surveys and Wide Field Camera 3 instruments on board the Hubble Space Telescope, with visual morphological classifications made by citizen scientists as part of the Galaxy Zoo project. After a careful analysis of the visual morphologies provided and the possible effects of a range of potential observational biases, I measure the redshift evolution of the bar fraction in the COSMOS field. For a volume limited sample (0.4 ≤ z ≤ 1.0; log(M_★/M_⊙) ≥10) of 2,380 disk galaxies, of which 317 are barred (f_bar∼13%), I find that the bar fraction has increased from 11±2% at z = 1, to 22±5% at z = 0.4. Furthermore, when splitting this sample into three stellar mass bins, I find that, whilst an increasing bar fraction with decreasing redshift is observed for all stellar masses, it is the most massive disk galaxies that drive the evolution observed. Building on this work, I present the first observations of the bar fractions beyond z = 1. Using visual morphological classifications from Galaxy Zoo: CANDELS, a sample of 876 disk galaxies, with 123 being barred disks (f_bar ∼ 14%) are identified across the redshift range 0.5 ≤ z ≤ 2.0. Selecting a sub-sample within the same region of the evolving galaxy luminosity function (brighter than L), we find that the bar fraction across the redshift range 0.5 ≤ z ≤ 2 (f_bar = 10.7^+6:3_-3:5% after correcting for incompleteness) does not significantly evolve. I present a recipe that describes how to explore the effects observational biases have on visual morphological classifications, before working through this recipe in detail in order to debias the galaxies in the Galaxy Zoo: Hubble catalogue. Of the 46,703 galaxies found within the redshift range 0.3 ≤ z ≤ 1.0, 24% (15,855) can be debiased, with a further 33% (15,338) having lower and upper limits placed on their original p_features vote-fractions. I use the newly debiased classifications to re-explore the redshift evolving bar fraction, finding that the trend is greater within these galaxies, increasing from f_bar = 4±1% at z = 1, to f_bar = 24±2% at z = 0.3. In my exploration of barred disk galaxies in the COSMOS field, up to z = 1, I identified a sample of 98 quiescent disk galaxies whose bar fraction is considerably higher (f_bar ∼ 45±5%) than the overall disk sample. To investigate this observation further, I explore the evolving demographics of the red sequence since z = 1. Using rest-frame colours from the UltraVISTA catalogue (McCracken et al., 2012), I find that, when using debiased p_features classifications, the fraction of all disk galaxies that are part of the red sequence decreases from 26±2% at z = 1, to 11±2% at z ∼0. Additionally, I also find that the fraction of all barred disk galaxies that are part of the red sequence increases from 0% at z = 1, to 22±4% at z ∼ 0. I conclude that these results show a maturing disk galaxy population as the universe ages.

We present the search for Superluminous Supernovae (SLSNe) within the Dark Energy Survey (DES), and using the Dark Energy Camera (DECam) in general. We construct empirical selection criteria for the selection of SLSNe from the first year of DES data followed by visual inspection of candidates. We present DES13S2cmm, the first spectroscopically-confirmed SLSN from DES, as a hydrogen-poor SLSN at a redshift of z = 0.663 ± 0.001. We updated our SLSNe search methodology based on our experiences and knowledge gained from this first season of DES data. In combination with the increase of the spectroscopic follow-up resources in the second year of DES operations, we have found 22 SLSNe candidates, 7 of which where spectroscopically confirmed as SLSNe events in the redshift range of 0.47 ≤ z ≤ 1.5. We present in detail the seven confirmed SLSNe and we draw attention to three SNe as the highlights of our SLSNe quest in DES Y2. Firstly, DES14S2qri was discovered at a redshift of z = 1.5, as themost luminous SN (M_z = −22.90) up to the writing of this thesis. DES14C1rhg was discovered at a redshift of z = 0.47 as the least luminous SLSN-I (M_r= −19.80), a finding that supports the hypothesis of a continuum that fills the gap between the peak brightness of SNe Ic and SLSNe-I. Lastly, DES14X3taz was detected at a redshift of z = 0.6 and is the second ever recorded SLSN to possess a well constrained pre-peak bump, which could be caused by a pre-explosion outburst of the SN. Using the DES Y2 sample of SLSNe we provide a basic calculation of the volumetric SLSNe rate, reporting a value of ∼ 206 events Gpc⁻³ yr⁻¹ h³₇₁. This value is compatible with other estimates in the literature, considering the simplicity of our method. We fit the bolometric light curves of the entire SLSNe sample discovered by DES with two competing models for SLSNe-I - the radioactive decay of ⁵⁶Ni, and a magnetar - and find that the magnetar is formally a better fit overall. Using the Magnetar model, we investigate possible cor-relations between the model parameters. We discovered moderate correlations between the explosion phase (t_exp) and the rise-time (τ_m), and between the rise-time (τ_m) and the spin-down time (P ). Moreover, we compare the bolometric light-curves of the entire SLSNe sample, without clear evidence of common behaviour, that would point towards a standardization. We find that the short DES observing seasons (∼ 5 months per calendar year) are limiting the discerning between the two models which, at late times (beyond 100 rest frame days), behave significantly differently. We present the Survey Using DECam for Superluminous Supernovae (SUDSS) as a complementary effort to DES SN survey, which operates since April 2014 and aims at obtaining a large and homogeneous sample of SLSNe out to redshift z ∼ 3, by maximizing its observing seasons in collaboration with DES SN survey.

The highly successful standard model of cosmology is built upon two fundamental assumptions: that structure formation proceeds hierarchically through the gravitational collapse of cold dark matter (CDM), and that the late-time expansion of the Universe is dominated by dark energy in the form of the cosmological constant, Λ. While predictions of the ΛCDM model have survived stringent tests spanning a wide range of scales, the true nature of the dark matter and dark energy remains a mystery. Here, we investigate structure formation in well-motivated, alternative scenarios. In the first half, we consider dark matter in the form of sterile neutrinos rather than CDM. We quantify the abundance, formation times and internal structure of sterile neutrino dark matter haloes, before making a detailed comparison of the properties of their substructures compared to their CDM counterparts. Using a semi-analytic model of galaxy formation, we compare observable differences between sterile neutrino and CDM cosmologies and find that future observations of the high redshift Universe and faint dwarf galaxies in the Local Group can place strong constraints on the sterile neutrino scenario. In the second half, the dark matter is assumed to be CDM, but we modify the underlying theory of gravity according to the f(R) model as an alternative theory for accelerated expansion. We test the commonly-assumed quasi-static approximation in f(R) gravity simulations, confirming its validity for a wide choice of model parameters. We then propose a new method for solving the equations of motion in f(R) gravity simulations. Using a suite of high resolution simulations, we find that the new method greatly accelerates the convergence rate of the solutions, improving the efficiency of these simulations by more than a factor of 20 compared to previous methods. This new method will bring us to a new era for precision cosmological tests of gravity.

We study the large-scale structure of alternative cosmologies, which intrinsically avoid the theoretical problems of the cosmological constant in the standard $\Lambda$ cold dark matter ($\Lambda$CDM) cosmology. We focus on early dark energy (EDE) models and $f(R)$ gravity, which are the typical examples of dark energy and modified gravity models respectively. We constrain the EDE model using measurements of cosmic microwave background and baryon acoustic oscillations, then use large-volume N-body simulations to study the structure formation and growth. EDE and $\Lambda$CDM can be distinguished through the shape of matter power spectrum on large scales, as well as the halo mass function at high redshift. We run high-resolution N-body simulations of $f(R)$ gravity in order to study in detail the properties of haloes and their environment. We find that halos less massive than $10^13 M_\odot/h$ have a more compact inner structure in $f(R)$ gravity than in $\Lambda$CDM. These low-mass haloes grow faster and contain substantially more subhaloes in $f(R)$ gravity. We also measure the correlation between different halo environment definitions used in observations and the fifth force potential in $f(R)$ gravity. Although the different ways to define environment do not agree with one another perfectly, they can provide useful guidance about how well a dark matter halo is screened. We also find that the screening of subhaloes in dark matter haloes is primarily determined by the environment, with the subhalo mass playing a minor role. Finally, we investigate and improve the H$\alpha$ luminosity function predicted by semi-analytical galaxy formation model, GALFORM. The result is important for accessing the performance of Euclid redshift survey, which is one of the missions cosmological probe of accelerating cosmic expansion.

The behaviour of atoms and molecules on interstellar dust grain analogue surfaces has been probed with surface science techniques such as reflection-absorption infrared spectroscopy (RAIRS) and temperature programmed desorption (TPD). Chemical systems under investigation include O atoms, O2, CO, H2O and N2O as deposited either on a silica (SiO2) surface representative of an interstellar dust grain or on H2O surfaces. As H2O is the dominant solid state chemical species in the interstellar medium (ISM) preliminary experiments have been conducted to investigate one of its formation pathways. Such experiments involved bombarding SiO2 or H2O surfaces with O atoms, producing O2 and O3 molecules as determined by RAIRS and TPD. During the initial stages of H2O growth on dust grains in the ISM, only small quantities will be found in the solid state. The de-wetting behaviour of such quantities was investigated by directly dosing H2O onto the SiO2 surface and observing the νOH vibrational band with time and temperature. Through such experiments, H2O has been observed to de-wet even at cryogenic temperatures of 17 K with an activation energy of about 2 kJ mol-1. The effect of this in the ISM is that bare dust grain surfaces will be left exposed for other molecules, such as CO, to interact with. Such interactions may be stronger than those of the molecules directly interacting with a water surface, meaning molecules will reside in the solid state for longer times. As icy mantles develop throughout the lifetime of a molecular cloud, adsorbed CO becomes a large part of the icy mantle. Multilayers of CO were investigated and shown to exhibit a bulk potential. This charge is due to a spontaneous electric (spontelectric) field arising as certain dipolar molecules align in the solid state. To investigate the spontelectric effect in CO, N2O was first examined and shows a correlation between IR features and the direct measurement of the spontelectric potential of N2O. Such a link was found through the temperature dependent shifts in the LO-TO splitting of the N2O. CO exhibits the same temperature dependence and spontelectric parameters have been extracted to show that a spontelectric potential of 6.7 mV per monolayer is created as multilayers adsorb on a surface. The effect of this in the ISM would be to reduce the gas-phase charge and ionisation fractions in molecular clouds of the ISM which in itself has the potential to have wide-reaching implications.