Jon Loveday


Constraining cosmological and galaxy formation models from galaxy surveys

 Funding: STFC quota studentship or other

 Observations appear to show that the Universe is not only expanding, but that it is doing so at an accelerating rate.  Two explanations have been proposed to understand this: (i) the accelerated expansion is caused by some form of 'dark energy' (including a cosmological constant), or (ii) Einstein's law of gravity (GR) needs to be modified on cosmological scales.  Many surveys have been attempting to measure the equation of state, w, of dark energy; all results so far are consistent with w = -1, as expected for a cosmological constant. 

In this project you will explore alternative ways of constrainingcosmological and galaxy formation models utilising existing and future galaxy surveys.  The goal is to design observational tests that can distinguish between different models, verified using numerical simulations.  These tests might include galaxy dynamics, as inferred from redshift-space distortions of the clustering signal, cross-correlation of optimally-weighted galaxy samples, and cross-correlation with other surveys, such as radio surveys for neutral hydrogen.  You will then apply these tests to existing survey data, such as the Galaxy and Mass Assembly survey (GAMA), and inform the design of future surveys to be carried out with major new facilities such as the 4-metre Multi-Object Spectroscopic Telescope (4MOST) and the Large Synoptic Survey Telescope (LSST).

This project will give you experience in numerical simulations and state of the art observational data-sets as well as getting closely involved in major new surveys.

For more information/to apply for this project, please contact

Automated galaxy photometry in the era of Euclid and LSST

 Funding: DISCnet

Obtaining reliable photometry of resolved galaxy images is an unsolved and extremely complicated problem, due to the presence of extended low surface-brightness features, foreground stars, close companions, irregular variations in surface brightness due to spiral arms and star-forming regions, and contamination of the background sky level by scattered light.  Doing so is, however, vital for fully exploiting the next generation of deep, wide imaging surveys to be conducted by the Euclid satellite and the Large Synoptic Survey Telescope (LSST).

 In this project you will exploit machine-learning techniques to develop an automated method for analysis of astronomical images and to reliably associate detected flux with their sources.  The goal is to simultaneously model all overlapping images and the sky background, employing empirical Bayesian priors based on previous morphological catalogues to minimize parameter degeneracies and to avoid unphysical models.  Colour, as well as morphology, will be important for discriminating between different source types, and so you will be simultaneously fitting in several passbands.  Your techniques will be developed using both simulated data, and deep imaging data from the Hyper Suprime-Cam (HSC) Subaru Strategic Program.  They will then be applied to upcoming, major new imaging surveys conducted by Euclid and LSST.

While some knowledge of astronomy would be an advantage, it is not essential for this project.  Far more important is an aptitude for coding and a strong interest in machine learning.

For more information/to apply for this project, please contact