Supermassive black holes are leading candidates for the regulation of galaxy growth and evolution over cosmic time, via ‘feedback’ processes, whereby outflows from the Active Galactic Nuclei (AGN) halt star formation within the galaxy. AGN feedback is generally thought to occur in two modes, high-excitation (HERG, or ‘quasar-mode’) and low-excitation (LERG or ’radio-mode’) each having a different effect on the host galaxy. LERGs curtail the growth of the most massive galaxies, whereas HERGs are thought to be activated by mergers/interactions, switching off star formation at high redshift. A critical problem in current extragalactic astrophysics lies in understanding the precise physical mechanisms by which these feedback processes operate, and how they evolve over cosmic time. Radio-loud AGN are an essential tool for studying major feedback mechanisms, as they are found within the largest ellipticals, and hence are beacons for the most massive black holes across the bulk of cosmic time. In this thesis I develop and study existing complete radio samples with extensive new multi-wavelength data in the radio, optical and infrared, aiming to investigate the evolution of AGN feedback modes, and methods to locate and study such systems at the very highest redshifts. This will serve to inform further studies of radio-AGN planned with next generation radio instruments such as the LOw Frequency ARray (LOFAR). Very few radio-loud AGN systems are currently known at high redshifts, and the effectiveness of traditional high redshift selection techniques, such as selection based on steep spectral index, have not been well quantified. A purely evidence-based approach to determining the efficiency of various high redshift selection techniques is presented, using nine highly spectroscopically complete radio samples; although weak correlations are confirmed between spectral index and linear size and redshift, selection first of infrared-faint radio sources remains by far the most efficient method of selecting high-z radio galaxies from complete samples. Radio spectral curvature in four of the complete samples is analysed and the effect of radio spectral shape on the measurement of the radio luminosity function (RLF) of steep-spectrum radio sources is investigated. Below z=1, curvature has negligible effect on the measurement of the RLF, however at higher redshifts, where source numbers are low, the shape of the radio spectrum should be taken into account, as individual source luminosities can change up to 0.1-0.2 dex, and this can in some cases introduce errors in space density measurements of up to a factor of 2-3 where source numbers are low. Building upon these samples, the very first independent determinations of the separate RLFs for high and low excitation radio sources across the bulk of cosmic time are made, out to z=1. Here it is shown that HERGs show very clear signs of strong evolution, in line with theoretical predictions. LERGs also show some very weak evolution with redshift, showing increases in space density of typically around a factor of 2. These measurements are also used to estimate the contribution of LERGs, which typically show weak or no emission lines to the ‘missing redshift’ population, which are sources within the complete samples not identifiable spectroscopically. Complementary to this, a pilot study is presented in selecting ‘missing redshift’ sources which are classed as infra-red faint (IFRS), which show no optical or near-IR identification, and are compact in the radio. Follow up spectroscopy on these candidate high z sources detected no line emission. Finally, work carried out towards the testing and commissioning of the new LOFAR telescope is presented. The findings from this thesis will serve to both streamline and inform high redshift radio-AGN searches and studies planned to be carried out with LOFAR and other multi-wavelength complementary surveys in the near future, and help to open up an as yet unexplored epoch in radio-AGN formation and evolution.

Masters of Science / Upcoming and ongoing 21cm surveys, such as the Square Kilometre Array (SKA), Hydrogen Epoch of Reionization Array (HERA) and Low Frequency Array (LOFAR), will enable imaging of the neutral hydrogen distribution on cosmological scales in the early Universe. These experiments are expected to generate huge imaging datasets that will encode more information than the power spectrum. This provides an alternative unique way to constrain the astrophysical and cosmological parameters, which might break the degeneracies in the power spectral analysis. The global history of reionization remains fairly unconstrained. In this thesis, we explore the viability of directly using the 21cm images to reconstruct and constrain the reionization history. Using Convolutional Neural Networks (CNN), we create a fast estimator of the global ionization fraction from the 21cm images as produced by our Large Semi-numerical Simulation (SimFast21). Our estimator is able to efficiently recover the ionization fraction (xHII) at several redshifts, z = 7; 8; 9; 10 with an accuracy of 99% as quantified by the coefficient of determination R2 without being given any additional information about the 21cm maps. This approach, contrary to estimations based on the power spectrum, is model independent. When adding the thermal noise and instrumental effects from these 21cm arrays, the results are sensitive to the foreground removal level, affecting the recovery of high neutral fractions. We also observe similar trend when combining all redshifts but with an improved accuracy. Our analysis can be easily extended to place additional constraints on other astrophysical parameters such as the photon escape fraction. This work represents a step forward to extract the astrophysical and cosmological information from upcoming 21cm surveys.