PhD projects for Astrophysics studies starting in 2025
This is our current list of possible PhD projects for 2025. If your favourite topic is not in the list, we highly encourage you to contact staff in your fields of interest to discuss PhD projects.
The nature and origin of galactic structure in high resolution simulations
Supervisor: Robert Grand
Please note:
This project is only available for in-person students, not for distance learning students.
Most of the visible starlight in the Universe emanates from spiral galaxies like the Milky Way. A key goal of modern astrophysics is to understand their formation and evolution. Fortunately, the distribution, motions, and chemical compositions of their stellar populations hold vital memory of the Galactic evolutionary history from the very early Universe shortly after the Big Bang down to the present day, presenting a window into their past.
Recent observations from ESA’s Gaia mission and others have revolutionised how we see the Milky Way, garnering several exciting discoveries including an incredibly complex and rich “out-of-equilibrium” dynamical structure. Although much of this structure was completely unexpected, it has likely been shaped by a wide range of physical phenomena, such as the galactic bar and spiral arms, mergers, dark matter, and black holes.
Numerical simulations are some of the most powerful theoretical tools we have at our disposal to progress our understanding: they are necessary to provide crucial interpretation of observations, which are after all just a snapshot in time. The student will work with a new state-of-the-art suite of cosmological, magneto-hydrodynamical simulations of Milky Way-type galaxies. Through a combination of analysis and comparison to observations, these simulations will deliver new capabilities to address many open questions on galaxy formation and Galactic dynamics, including but not limited to:
- The nature and origin of galactic spiral arms: Spiral arms are iconic structures and significantly affect galaxy evolution, however their nature and origin are still unknown. Some open questions are: how and when did they form? how have they moved stars like the Sun around the Galaxy over billions of years? We will make new predictions from chemo-dynamics of disc stars to compare with observations.
- The Milky Way (and galaxies like it) have experienced numerous mergers and fly-by interactions with other galaxies in the past, which imprint chemo-dynamical signatures in the disc. The student will look for novel signatures of these events in order to put new constraints on the most recent and significant interactions and mergers.
There will also be opportunities for the student to develop and run simulations with the state-of-the-art Arepo simulation code.
Explain this: Large Language Models for Galaxy Images
Supervisors: Andreea Font and Rob Lyon (CSM)
Please note:
This project is only available for in-person students, not for distance learning students.
Large volumes of stunningly detailed galaxy images are now being collected by modern ground and spaced-based telescopes. These are revealing intricate features with unprecedented levels of clarity. Such visual detail is exciting and crucially important. It can help improve our understanding of galaxy evolution, supermassive black holes, and even dark matter. However, the sheer scale of data being collected makes analysis increasingly challenging. Forthcoming surveys like Euclid and the Vera Rubin Observatory’ Legacy Survey of Space and Time (LSST) will complicate things even further! This project aims to address this issue, by developing new Machine Learning (ML) tools, that can automatically identify, classify and explain important features in galaxy images. This will help transform how researchers interpret and work with large-scale astronomical datasets.
Central to this effort is the application of Large Language Models (LLMs) - a cutting-edge technology originally designed for understanding and generating human language. LLMs have achieved great success - whether helping software engineers write code more efficiently (GitHub Copilot), allowing people to translate text between different languages in seconds (Google Translate), or even providing new ways to query and synthesise information (ChatGPT, Claude, Co-Pilot etc).
In this project, we will develop and adapt LLMs to process and annotate astronomical image data, translating complex patterns and structures into plain-language summaries, that bridge the gap between raw data and human understanding. An important aspect of this project is ensuring that these tools not only analyse galaxy features correctly, but also explain their findings in ways that are transparent and trustworthy. This interpretability is critical for building confidence in the results and ensuring they are scientifically meaningful.
This is an interdisciplinary research project that will equip you with valuable skills in high performance computing, machine learning, data science and big data. You’ll have the chance to explore ML methods that are currently driving innovation in various sectors and bring those to our wider astronomical community – including to overseas collaborators. You’ll also become acquainted with the state-of-the-art ML packages and libraries via Nvidia Deep Learning Institute accredited training. The expertise you gain during this project will be highly transferable, opening pathways to careers in fields beyond astronomy, to any other data-driven discipline.
Galaxy demographics
Supervisor: Ivan Baldry
Please note:
This project is available for both in-person and distance learning students.
We can analyse galaxies individually or as a population. Focusing on the latter allows us to empirically track galaxy evolution since, if we measure demographics of galaxy populations at different distances, we are viewing the universe at different epochs. Measurements of galaxy populations can also be compared with cosmological-scale simulations. Galaxy demographics are therefore key for empirically describing and understanding galaxy evolution, and can also play a role in constraining cosmological models.
Some of the key demographic measurements are: the galaxy stellar mass function (distribution of galaxy masses), size-mass relation, colour-mass relation,
morphological and dynamical distributions. Related to this are measurements of the properties of the large-scale structure that the galaxies' inhabit: local environment, galaxy groups and clusters. These types of measurements place constraints on the processes affecting galaxy evolution and cosmological-scale models.
Various projects are available in this area, from finding and characterising the lowest-mass galaxies to testing the role of environment on galaxy evolution. The data are from photometric and spectroscopic surveys of galaxies. New deep imaging is available from the Euclid space telescope.
A tango in a crowded ballroom – The binaries inside massive star clusters
Supervisor: Sebastian Kamann
Please note:
This project is only available for in-person students, not for distance learning students.
Binary stars play a key role for the evolution of star clusters. Many of the stellar exotica we find in clusters, such as pulsars, blue stragglers, or cataclysmic variables, can only be explained through binary interactions. Binaries also offer us a unique chance to find black holes, and thereby to understand whether star clusters are indeed gravitational wave factories. Finally, the binding energies stored in binary stars substantially impact the evolution of their entire host clusters.
In this project, we will explore the interplay between binary stars and their host clusters. Using time-series observations, we will detect and characterise binary stars inside a sample of ~30 massive star clusters. For this task, our group is collecting large spectroscopic data sets with instruments like MUSE or FLAMES. In addition, we will exploit the upcoming Gaia data release 4 to complete our picture of binary stars in massive clusters.
Some of the questions we plan to answer in the project are:
What are the period distributions of binaries in clusters of different masses, ages, and densities?
The student will analyse existing MUSE spectroscopic data of Galactic globular clusters and young massive clusters in the Magellanic Clouds. To determine the orbital parameters of the binaries, statistical analysis like Monte Carlo sampling or nested sampling will be used. Where possible, astrometry and photometry from Gaia and the Hubble Space Telescope will be included in the analysis.
Do we find stars orbiting black holes?
Using MUSE and FLAMES spectroscopy, the student will identify binary systems with unseen, yet massive, companions. Follow-up analyses such as spectral disentangling will be performed to determine the nature of the companions.
How well do numerical simulations describe real clusters?
Monte Carlo simulations allow us to study the evolution of entire clusters in a computer. To understand how realistic these simulations are, the student will compare the output of such simulations, generated by collaborating research groups, to the available observational data. Emphasis will be placed on the comparison of the simulated and observed binary populations.
Modelling the next generation of gravitational-wave electromagnetic counterparts
Supervisor: Gavin P Lamb
Please note:
This project is only available for in-person students, not for distance learning students.
Gravitational wave astronomy is making waves in multiple fields of astrophysics, from stellar evolution to chemical enrichment of the Universe. Understanding what the electromagnetic counterparts to gravitational wave detected mergers look like is an essential step in realising the full potential of the next generation of gravitational wave observatories. Building on an existing framework of transient models, improving model precision and physics, and producing output products (sky maps, light curves, spectra) that are directly applicable to future observations is one part of the key required to unlocking new discoveries. Optimised, physically motivated models that can be easily utilised with sophisticated Bayesian sampling tools on astrophysical data is an increasingly important component of modern astrophysical inference.
This project focuses on the development, optimisation, and application of new and state-of-the-art semi-analytic/numerical models for gamma-ray burst transients with a focus on predictions for next generation telescope and gravitational wave observatories.
What are the signatures of lensed gamma-ray bursts?
Supervisor: Gavin P Lamb
Please note:
This project is only available for in-person students, not for distance learning students.
Gravitationally lensed gamma-ray bursts (GRB) have been an elusive object since the earliest days of well localised GRBs and their afterglows – late 1990s. Various claims have been made of detecting lensed GRBs, however, these are often unconvincing or lack critical observations required for a definitive identification. Utilising the recent advances in GRB modelling, we can better investigate observable features that would form the “smoking gun” evidence of lensed GRBs. In additional to the lensed prompt GRB emission, afterglows and any related thermal transients (supernovae, kilonovae, pulsar wind nebulae) would be similarly lensed, if detectable. By developing semi-analytic and numerical models for the study of lensing effects in GRB data, predictions for and identification of lensed GRBs in new and archival data can be made. This project will focus on the development or adaptation of transient models for application to lensed events. Predictions for observables, and statistically motivated searches will be performed giving an indication of how well the new models can explain observations and if any real lensing events have been missed due to redundant assumptions.
Jets from Gamma-ray Bursts and Gravitational Wave Mergers
Supervisor: Shiho Kobayashi
Please note:
This project is only available for in-person students, not for distance learning students.
Gamma-ray bursts (GRBs) are instantaneously the most luminous objects in the universe, associated with relativistic jets. The core-collapses of massive stars or the mergers of binary compact stellar objects are their progenitors. The latter is the primary targets for gravitational wave (GW) observatories (e.g. LIGO, Virgo). In both cases, accretion onto a black hole is likely to power such jets. We will study the structures and other characteristics of relativistic jets using the electromagnetic (EM) counterparts of GW sources and “orphan" GRB afterglows together. This study also has the implications for the measurements of the Hubble constant using GW observations.
Using our numerical models, we will evaluate the light curves of off-axis jet for various jet structures to make predictions and to discuss the EM follow-up strategy for the current and upcoming surveys in optical and radio (e.g. ZTF, LSST, LOFAR, SKA). Although GW 170817 happened at a distance of 40Mpc, such local events seem to be rather rare. We will investigate whether we can statistically give constraints on jet structures by analysing many events together, or how many EM counterparts and orphan afterglows are needed to study jet structures. We will examine whether the structures of jets are universal (e.g. Gaussian), and whether jet structures are similar in the two classes of jets (core-collapse SNe vs compact-stellar mergers).
In the case of GW 170817, the late-time radio detection of superluminal motion played a crucial role to break the degeneracy between two competing models (i.e. jet vs semi-isotropic cocoon), we will numerically evaluate the radio and infrared images of various structured jets, and we will investigate how jet images are affected by the lateral and radial jet structure. We will also study what observable quantities (for example, jet image size, centroid shift) are most sensitive to the viewing angle. We will develop the best scheme to evaluate the viewing angle from EM observations. GW sources accompanied by EM counterparts provides standard “siren” measurements of the Hubble constant. However, the degeneracy in the GW signal between the source distance and the viewing angle induces the uncertainty in its measurement. We will discuss how they uncertainly can be reduced by using our numerical model when jet images are detected by VLBA, JWST and others.
Supernova Studies
Supervisor: Paolo A. Mazzali
Please note:
This project is available for both in-person and distance learning students.
Supernovae are arguably one of the most interesting astrophysical events. They are among the most luminous object in the Universe. They mark the end of the lives of different types of stars, enrich the interstellar medium with nuclearly processed elements, are essential for cosmology as distance indicators, and are linked with extremely powerful events such as gamma-ray bursts.
Students can work on a variety of topics, including observations, interpretation of the data, modelling spectra and light curves as well as developing new and more advanced codes. Projects will be discussed and defined on an individual basis, such as to achieve the best match between the interests, abilities and situation of the student on the one hand and current research topics on the other, and thus can be either full- or part-time.
Possible topics include Type Ia SNe (physics and use as standardizable candles), stripped-envelope (Type Ib/c) SNe (physics and connection with Gamma-ray bursts), Superluminous SNe (physical processes).
Machine learning-based classification of spectra from explosive transients
Supervisors: Matt Darnley, Ivan Olier-Caparroso (CSM), Chris Copperwheat, Fiona Murphy-Glaysher
Please note:
This project is only available for in-person students, not for distance learning students.
The Astrophysics Research Institute has a long-established Time Domain Astrophysics group that specialised in the observation of explosive transients, including novae, supernovae, and gamma-ray bursts, among others. The ARI’s own Liverpool Telescope (LT) plays a pivotal role in the acquisition of spectra of novae and supernovae. The LT archive contains a huge ‘back catalogue’ of such observations. Classification of nova spectra relies upon visual inspection of the observations. This project will develop a machine learning-based approach to nova spectra classification, including utilising unsupervised learning techniques. This will enable exploration of, for example, how nova spectra vary with galactic host location and local stellar population. The PhD student will then apply these techniques to the entire LT spectral archive. This work is particularly important in support of the New Robotic Telescope (NRT) project, which will be obtaining large number of spectra of unknown transients daily, once the NRT comes on-line.
Due to the cross-School nature of this project, it is, unfortunately, not suitable for delivery solely by distance learning. We are happy to discuss other flexible approaches, including potential part-time students.