During the first three observing runs of the Advanced gravitational-wave detector network, the LIGO/Virgo collaboration detected several black hole binary (BHBH) mergers. As the population of detected BHBH mergers grows, it will become possible to constrain different channels for their formation. Here we consider the chemically homogeneous evolution (CHE) channel in close binaries, by performing population synthesis simulations that combine realistic binary models with detailed cosmological calculations of the chemical and star-formation history of the Universe. This allows us to constrain population properties, as well as cosmological and aLIGO detection rates of BHBH mergers formed through this pathway. We predict a BHBH merger rate at redshift zero of $5.8 \hspace{1mm} \textrm{Gpc}^{-3} \textrm{yr}^{-1}$ through the CHE channel, to be compared with aLIGO's measured rate of ${53.2}_{-28.2}^{+55.8} \hspace{1mm} \text{Gpc}^{-3}\text{yr}^{-1}$, and find that eventual merger systems have BH masses in the range $17 - 43 \hspace{1mm} \textrm{M}_{\odot}$ below the pair-instability supernova (PISN) gap, and $>124 \hspace{1mm} \textrm{M}_{\odot}$ above the PISN gap. We further investigate the effects of momentum kicks during black hole formation, calculate cosmological and magnitude limited PISN rates and investigate the effects of high-redshift deviations in the star formation rate. We find that momentum kicks tend to increase delay times of BHBH systems, and our magnitude limited PISN rate estimates indicate that current deep surveys should be able to detect such events. Lastly, we find that our cosmological merger rate estimates change by at most $\sim 8\%$ for mild deviations of the star formation rate in the early Universe, and by up to $\sim 40\%$ for extreme deviations.

We present a systematic analysis of the complete set of observations of the neutron star low-mass X-ray binary 4U1608-52 obtained by the Rossi X-ray Timing Explorer's Proportional Counter Array. We study the spectral and fast-time variability properties of the source in order to determine the mass and spin of the neutron star via the relativistic precession model, and find 24 observations containing usable sets of the necessary three quasi-periodic oscillations (triplets) with which to accomplish this task, along with a further 7 observations containing two of the three quasi-periodic oscillations each. We calculate the spin and mass of the source for each of the triplets, and find that they give physically realistic estimates clustering in the spin range $0.19 < a < 0.35$ and mass range $2.15 < M/\textrm{M}_{\odot} < 2.6$. Neutron stars present environments for studying matter under the most extreme conditions of pressure and density; as their equation of state is not yet known, accurate measurements of their mass and spin will eventually allow for the discrimination between various models. We discuss the implications of our findings in the context of equation of state predictions, physically allowed spin ranges, emission proximity to the innermost stable circular orbit and possible model inaccuracies.

We show that multiple machine learning algorithms can match human performance in classifying transient imaging data from the Sloan Digital Sky Survey (SDSS) supernova survey into real objects and artefacts. This is a first step in any transient science pipeline and is currently still done by humans, but future surveys such as the Large Synoptic Survey Telescope (LSST) will necessitate fully machine-enabled solutions. Using features trained from eigenimage analysis (principal component analysis, PCA) of single-epoch g, r and i-difference images, we can reach a completeness (recall) of 96 per cent, while only incorrectly classifying at most 18 per cent of artefacts as real objects, corresponding to a precision (purity) of 84 per cent. In general, random forests performed best, followed by the k-nearest neighbour and the SkyNet artificial neural net algorithms, compared to other methods such as na\"ive Bayes and kernel support vector machine. Our results show that PCA-based machine learning can match human success levels and can naturally be extended by including multiple epochs of data, transient colours and host galaxy information which should allow for significant further improvements, especially at low signal-to-noise.