PhD projects

Supervisor: Chris Conselice

Project title: Artificial intelligence and the evolution of galaxies with foundation models

This project develops a foundation model for galaxy evolution, trained on the largest observational database of distant galaxies ever assembled using the combined imaging and photometric datasets of Euclid, the Nancy Grace Roman Space Telescope, and Rubin Observatory/LSST. This project will leverage the complementary depth, resolution, wavelength coverage, and sky area of these three surveys to construct a large-scale, data-driven representation of how galaxies grow and evolve across cosmic time.

At the core of the project is an empirical model of galaxy evolution that captures the dominant physical drivers of galaxy growth, including accretion, in-situ star formation and galaxy merging (both major and minor). In this project we will determine how these features are imprinted in observed galaxy populations, rather than as is done now, derive them as prescribed by theoretical assumptions. The empirical model we develop will learn the statistical connections between galaxy star formation rate, morphology, structure, size, star formation activity, and environment across redshift directly from the data itself.

Part of this will be developing a predictive framework to reconstruct the evolutionary history of any individual galaxy at any redshift. Given a galaxy's observed properties, the AI model will predict how that galaxy would have appeared at earlier (or even later) cosmic epochs. This will include its morphology, structural parameters, size, and star formation rate as well as its underlying dark matter and baryonic mass accretion history. Ultimately this will 'solve' galaxy evolution in terms of broad formation history and mechanisms.

Methodologically, the model will be built as a multi-modal transformer-based foundation model, using vision-transformers to encode imaging and when available spectroscopy. Generative components that are diffusion-based or flow-based will be used to predict plausible evolutionary trajectories and their associated uncertainties, treating redshift as a continuous axis so that a galaxy's appearance, structure, and accretion history can be queried at any epoch rather than only at fixed snapshots as we observe them.

Fundamentally, this approach is entirely empirical and new. It bypasses the ad-hoc assumptions inherent in traditional semi-analytic or hydrodynamical simulation-based models, instead letting the observational data itself define the pathways of galaxy formation and evolution. The result will be a flexible, scalable tool for understanding galaxy assembly that is directly grounded in the observed universe and can be expanded when new data sets become available such as SKA.

Supervisor: Phil Bull

Project title: Building an ultra-high dimensional model of the radio sky

There are many maps and catalogues of the radio sky, observed using many different instruments over several decades, with even more on the way thanks to SKA and its precursors. When these data are combined into multi-frequency models of what the radio sky should look like however, we encounter a variety of problems, ranging from incompatible resolutions, to offsets in the assumed positions, brightnesses, and spectra of compact sources, to large unmodelled calibration errors. These models are nevertheless the cornerstone of many data analysis techniques, such as sky-based calibration of 21cm arrays and foreground removal procedures. Improving them is important for advancing several areas of radio astronomy and cosmology.

In this project, you will address these problems using an ultra-high dimensional framework for analysing radio data that our group has developed. It is called Hydra, and uses a technique called Gibbs sampling to enable it to efficiently explore statistical distributions describing models of complex datasets that can have millions of free parameters. This allows us to work with very detailed “high fidelity” models of the radio sky and the instruments that are observing it, and thus resolve some of the problems that are getting in the way.

There are several aspects to the project, and the relative focus on each is up to the student. One key task is to further develop, test, and run the code on a range of radio datasets, which involves analytic and computational statistics work, including with high-performance computers. Another is to develop better models of instrumental effects, including antenna patterns, gain fluctuations, and missing flux, using Python programming, and potentially EM modelling software like CST. We have access to several cutting-edge datasets covering Cosmic Dawn, Reionisation, and the late-time Universe, including from the HERA, MeerKAT, and uGMRT arrays, and so another task is to analyse and incorporate these data into a more sophisticated sky model with more accurate spectral behaviour.

Finally, new data from “absolutely calibrated” 21cm global signal experiments can be used to re-calibrate the disparate datasets and impose physical consistency on the properties of the model. 

Supervisor: Rene Breton

Contact: rene.breton@manchester.ac.uk

Project title: Caught in the spider’s web: Computational modelling of Black Widow and Redback pulsars

Project overview and objectives:
Spider pulsar binaries - the so-called black widows and redbacks - are remarkable astrophysical laboratories. Each contains a rapidly rotating millisecond pulsar ablating a low-mass companion star, leading to some of the fastest spinning and most massive neutron stars known. Despite extensive observational progress, the physics of how pulsar winds irradiate, heat, and strip their companions remains poorly understood. This project will focus on advancing our theoretical and computational understanding of these extreme systems.

The objectives are to:
Perform hydrodynamic simulations of stars resembling spider companions, in order to study how irradiation alters their internal structure, convection, and energy transport. Investigate surface circulation patterns, irradiation-driven winds, and possible mass-loss mechanisms.
Incorporate these insights into self-consistent binary stellar evolutionary models of spider systems. Explore the Galactic population of spiders, using simulations and state-of-the-art statistical tools (including simulation-based inference and machine learning) to connect theoretical predictions with observations.

Methods and approach:
The student will first develop hydrodynamic simulations of irradiated low-mass stars, quantifying the redistribution of heat and the conditions for surface outflows and ablation. These results will then inform improved prescriptions for binary evolution modelling, which will be applied to track the long-term evolution of spider systems. Parallel work will explore Galactic population synthesis and kinematics, using both traditional simulation methods and modern inference approaches to constrain formation rates and evolutionary channels against observational data.

Student support, training, and environment:
The student will join the Spiders Team led by Prof Rene Breton (University of Manchester), benefiting from close supervision in computational astrophysics and pulsar science. Training will include hands-on experience in hydrodynamic simulations, stellar evolution codes, statistical inference, and machine learning. The astrophysics group hosts a vibrant research environment, including regular colloquia, journal clubs, and specialist meetings on compact objects and data-driven astrophysics. The student will have opportunities to collaborate with international partners, present at conferences, and gain a broad set of transferable skills in computational modelling, data science, and high-performance computing.

Reading list on the general spider pulsar binary topic:

• Spider Pulsars 101

Supervisory team: Dr Patrick Weltevrede

Contact: Patrick Weltevrede

Project title: Characterising the dynamic magnetospheres of neutron stars

Radio pulsars are highly magnetised neutron stars which rotate very rapidly: up to 100s of times per second. During each rotation, the radio emission beamed along the magnetic poles sweeps across the Earth and can be detected by very sensitive radio telescopes as a regular sequence of pulses. The rotation of the neutron stars can be extremely stable, which makes them very accurate clocks allowing tests of the general theory of relativity.

However, for most pulsars, the individual pulses of the observed sequence vary greatly in shape, intensity and polarisation. These variations are caused by largely unknown physical processes in the magnetosphere of these stars. In some cases, these variations happen in a coordinated fashion, which are known as drifting subpulses, indicative of regular dynamical changes in the magnetosphere. On longer timescales (weeks/months/years) the radio emission is known to be correlated to the spin-down rate of the star, showing that the radio emission is linked to powerful magnetosphere-wide processes.

In this project you will explore observational data from the "1000 Pulsar Array" project on the MeerKAT telescope in South Africa (a pre-cursor of the SKA: the Square Kilometre Array, which will be the largest telescope in the world), and Lovell telescope data in Jodrell Bank (one of the largest steerable radio telescopes in the world). In this project you will characterise various types of emission variability seen in the pulse shapes and their polarisation, and explore the implications for magnetospheric theories. Where possible, we will supplement this data with observations from other radio observatories such as the Parkes telescope in Australia (a great instrument, which has discovered more pulsars than any other radio telescope in the world).

Supervisor: Dr Phil Bull

Contact: phil.bull@manchester.ac.uk

Project title: Commissioning the RHINO 21cm global signal experiment at Jodrell Bank

Early in the Universe's history, before the first stars and galaxies had formed, the only significantly detectable EM radiation came from neutral hydrogen, which has a spin-flip transition deep in the radio part of the spectrum, at a rest-frame wavelength of 21cm. As galaxies began to form, the neutral hydrogen was heated and eventually re-ionised. By charting the brightness temperature of the 21cm line over time, we can learn about the magnitude and timing of these early heating processes, and thus learn about the very first stars and galaxies via their impact on their local environment.

The 21cm line is redshifted according to when in cosmic history its emission took place. To probe the time before the reionisation of the Universe, we must observe frequencies in the 50 – 100 MHz range, corresponding to emission from less than a billion years after the Big Bang. Observing this radiation is difficult however; it is faint, while other radio emission processes (such as Galactic synchrotron) occur nearby and can be several orders of magnitude brighter. Radio interference from human activity is also problematic in this part of the spectrum, e.g. FM radio. Instruments designed to observe the 21cm "global" signal (its average over the whole sky) must therefore be calibrated extremely accurately so that these spurious sources of emission can be subtracted from the data to uncover the 21cm signal itself.

In this project, you will work on commissioning and observing with a new 21cm global signal experiment called RHINO, which is currently under construction at Jodrell Bank. The main RHINO telescope is a large (8m high) horn antenna, which is designed to have excellent rejection of many of the systematic effects mentioned above. The aim of this project is to commission the receiver system, develop some of the calibration and analysis code, and perform the first science observing with RHINO. This will require refining some components of the receiver hardware, developing a statistics-based calibration pipeline, and observing and analysing the first seasons of data from the telescope. (Note: Existing knowledge of electronics/RF engineering is not necessary for this project.)

More information:

• RHINO: A large horn antenna for detecting the 21cm global signal

Supervisor: Rebecca Bowler

Project title: A detailed view of the first galaxies with a multi-wavelength approach

Observations with the James Webb Space telescope have revealed a surprising number of luminous galaxies at very high redshift. Furthermore, these galaxies have unusual properties compared to the local universe, showing low chemical enrichment, unusual dust properties, and irregular morphologies.

This project will focus on understanding in detail the properties and formation mechanism of a sample of rare, luminous/massive sources which have detailed follow-up imaging and spectroscopy from the James Webb Space Telescope and ALMA. These data, obtained as part of the REBELS and PHOENIX collaborations, in which the student will become a core member, allow a spatially resolved analysis of stars and dust and provide new insights into how such galaxies form. The project will be primarily observational, and will include the analysis of multi-wavelength data, data reduction and comparison to simulation predictions.

Further reading:

• [1] https://ui.adsabs.harvard.edu/abs/2025MNRAS.539..109F/abstract
• [2] https://ui.adsabs.harvard.edu/abs/2026ApJ..1000..228K/abstract
• Phoenix: https://phoenix.strw.leidenuniv.nl
• REBELS: https://rebelsalma.wordpress.com

Supervisor: Chris Conselice

Project title: Early galaxy evolution and reionization with the Euclid space telescope and Rubin/LSST

The Euclid Space Telescope is a major European mission that will cover one-third of the entire sky to a depth and resolution comparable to the Hubble Space Telescope. LSST/Rubin is a large ground-based telescope taking data on similar parts of the sky. This PhD project will use Euclid’s imaging and spectroscopy to investigate the large-scale structure of the universe during its first 3 billion years, a period that has never been studied in detail as previous data were neither extensive nor deep enough.

With Euclid, the PhD student on this project will identify the first massive and bright galaxies, analyse their spatial distribution, and determine how their number density evolves over time. This is only possible with a Euclid/LSST combination and cannot be done with JWST. We will also use spectroscopy from Euclid to trace the Lyman-alpha line in these galaxies -- a key diagnostic for identifying when and where reionisation occurred (the transition of the universe from neutral hydrogen gas to ionized gas). The project will then measure the strength of the Lyman-alpha line across the sample, thereby determining the topology and structure of how the universe was reionised. This can then be compared with future HI intensity mapping of this epoch from SKA.

Supervisors: David Williams-Baldwin, Rob Beswick

Project title: Expanding the LeMMINGs survey using the VLA to study Active Galactic Nuclei and star formation

Active Galactic Nuclei (AGN) are the central regions of galaxies that include an accreting super massive black hole. While many large black holes are found to produce radio jets that extend into the Inter Galactic Medium, it is the less powerful low-luminosity AGN (LLAGN) that numerically dominate the black hole population. However, LLAGN are much harder to find due to their low bolometric luminosities and smaller scale jets, necessitating high angular resolution and highly sensitive telescopes to not only detect but resolve the emission into individual jet components. The Legacy e-MERLIN Multi-band Imaging of Nearby Galaxies survey (LeMMINGs) was designed to study a statistically-complete sample of low-luminosity AGN in nearby (d < 110Mpc) galaxies, resolving the jet and star formation emission from the compact radio core at the heart of the AGN.

The LeMMINGs project has now been conducted at 1.5 GHz (Baldi et al 2018, 2021a, 2021b) and new results at 5 GHz are being prepared for publication (Williams-Baldwin et al, in prep.). However there is a need to improve the sensitivity of the data to larger scale structures to detect larger scale jets and star forming regions, combining this with the e-MERLIN data to provide a more in-depth study of these LLAGN. The sample is >70% complete with the Very Large Array, probing angular scales slightly larger than e-MERLIN. This project will involve analysing the VLA data on the sample and combining it with the e-MERLIN data to provide a multi-resolution view of LLAGN jets and star formation in nearby galaxies. The candidate should be able to use python and have an interest in radio astronomy on AGN and star formation.

Supervisors: Dr Emmanuel Bempong-Manful, Prof Rob Beswick and Dr Paddy Leahy

Project title: Faraday rotation and the 3-D magnetic field structures in AGN jets

It is widely accepted that magnetic fields play a key role in the bulk acceleration and propagation of jets produced in active galactic nuclei (AGNs) (e.g., Abstract, Polarization of AGN Jets and Polarization Observations of AGN Jets: Past and Future). However, determining the magnetic field structures in AGN jets from polarisation observations is not always straightforward – due to stratification, radiative transfer, among other factors. Whereas the synchrotron radiation polarisation corrected for Faraday rotation is determined by the structure of the magnetic field in the emitting regions, Faraday rotation on the other hand is produced by magnetic field in thermal plasma between us and the emission, almost always in front of the synchrotron emission rather than mixed with it. As a consequence, gradients of rotation measure (RM) are often seen across radio sources, but their relationship to the structure is still very much unclear, because of limited resolution and/or sensitivity of existing facilities. Thus, there exist open key questions in jet physics including; (1) What are the three-dimensional structures of powerful jets? Do they have highly relativistic “spines”? and (2) What are the magnetic field configurations immediately surrounding jets? Is there evidence for confining fields?. Thanks to the wide-band capability of e-MERLIN (an SKA pathfinder instrument) we can now get excellent signal-to-noise ratio, working around interference – allowing us to determine the variation of RM across a jet at high spatial resolution in a single observation, something which has never been possible previously for powerful jets.

To this end, the e-MERLIN Legacy programme on radio jets has been mapping a number of powerful radio galaxies and quasars to obtain the best possible information about the inner structure of the jet outflows. One target of special interest is 3C 273 – the first documented object to be identified as quasar – which lies at a distance of about 700 Mpc, where we have excellent radio and multi-wavelength data. On sub-kpc scales, the jets in 3C 273 is wide and exhibit a rich linear polarisation structure, making it an ideal candidate to study the origin of Faraday rotation and the 3-D magnetic field configuration in AGN jets. In this project the student will map the structure of the inner jets in 3C 273 using the new multifrequency e-MERLIN observations and attempt to investigate the RM and 3-D magnetic field configuration of the jets. The project can be extended by building a model of the radio spectrum in 3C 273, using ancillary data at other frequencies, and by looking at some other sources of similar type.

Supervisor: Chris Conselice

Project title: Hunting the first galaxies, stars, and black holes with the James Webb Space Telescope

Since its launch, the James Webb Space Telescope has started a revolution in our understanding of the first galaxies, black holes, and stars formed within 500 million years after the Big Bang. This major successor to the Hubble Space Telescope is observing, for the first time, the birth of galaxies in the universe. This search is ongoing and we have yet to find these systems. As such, this PhD will search for the first galaxies, black holes, and stars.

The PhD student working on this project will help lead the discovery of the first galaxies which contain the first stars and the first black holes in the universe. These will be identified though the deepest imaging ever taken as well as through detailed spectroscopy. Once these are identified the student will study these first galaxies, including measuring their masses, sizes, and their other properties including their central black hole masses.

This PhD will use the latest JWST data, including the largest programme approved to date, VENUS, which probes blank fields and galaxy clusters. These massive galaxy clusters magnify the background distant galaxies through gravitational lensing, allowing us to detect extremely faint systems.

These observations will then be interpreted in terms of theories of galaxy formation, testing different ideas about how the first generations of galaxies and stars formed. This work will further have a major impact on our understanding of cosmology, dark matter, and early astrophysics and will have broad interest to the general public.

Supervisor: Rebecca Bowler

Project title: Forward modelling early galaxy formation as seen by Euclid, Roman and Rubin

The search for the first galaxies remains challenging even in the era of JWST. When selecting for the highest redshift sources in multi-wavelength images, contaminant populations can vastly outnumber the galaxies of interest. Furthermore, as datasets have increasingly inhomogeneous quality and depth, it is challenging to interpret the resulting samples especially when only photometric redshifts are available. Impure samples and inaccurate modelling of the so-called "selection function" can significantly bias the resulting distribution functions (e.g. size-luminosity, colour-magnitude and luminosity functions), preventing robust insight into the underlying astrophysics.

The goal of this project is to tackle this problem using simulation based inference (SBI). The student will develop a SBI framework to forward model the number and properties of galaxies over a broad range of redshifts (z = 3-15) and different datasets including the Roman, Rubin and Euclid deep fields. Once established this model will be used to fit the observed galaxy samples to infer the evolution of the distribution functions of interest, and compare these to predictions from hydrodynamical cosmological simulations.

As part of the project the student will work closely with international teams (e.g. the Euclid and Rubin science working groups) as well as benefitting from the active group in Manchester who are using these datasets to identify and study high-z galaxies. It is likely to use both traditional and machine learning approaches, including emulation, as tools to make an efficient SBI framework. There is flexibility in the direction of the project in later years.

Supervisors: Clive Dickinson, Stuart Harper, Vasu Shaw, Paddy Leahy, Jens Chubla

Contact: Clive Dickinson

Project title: Galactic radio emission – understanding our Galaxy for future cosmology missions

JBCA has been at the forefront of studying diffuse Galactic radio emission since the very first days of radio astronomy. We've mapped the entire sky at low angular resolution with several "low" radio frequencies (< 5 GHz) and "high" radio frequencies (30-900 GHz) with the Planck space mission. However, we still do not have a full understanding of Galactic emission across the radio/microwave bands. There are many unanswered questions, including what causes the large radio loops that cover large fractions of the sky or the halo bubbles near the Galactic centre, what is the form of dust/molecules that is responsible for anomalous microwave emission – is it due to spinning dust grains? New polarization observations also reveal a very ordered Galactic magnetic field, both locally and across the Galaxy, but a detailed Galactic model is still missing.

In addition to Galactic science, detailed measurements of the Cosmic Microwave Background (CMB) provide the strongest constraints on cosmological parameters. Future CMB polarization missions are aiming to constrain primordial B modes, caused by a background of gravitational waves, which would be a smoking gun signature that inflation happened in the first fractions of a second of the Universe. However, one of the major challenges is in quantifying and removing "foreground" emission, which for B modes, is at least an order of magnitude brighter than the cosmological signal we're trying to detect!

JBCA is involved in several world-leading experiments that are both trying to measure CMB B-modes and quantify the contaminating foreground emission, including:

• C-Band All-Sky Survey – 5 GHz all-sky survey to map synchrotron intensity and polarized emission with high sensitivity and fidelity, to provide a foreground template for future CMB missions. C-BASS is likely to be the key low frequency data for future missions, as it is at the ideal frequency and is able to map the sky with high sensitivity with minimal systematic errors. We have completed the northern survey and the southern survey will be starting in the near future.
• COMAP Galactic Plane Survey – 26-34 GHz survey of the northern Galactic plane with the COMAP instrument at 5 arcmin resolution to study Galactic emission, particularly AME/spinning dust near 30 GHz. Manchester is leading this sub-project for the COMAP collaboration. The survey is well underway and is expected to be completed ~2025.
• LiteBIRD – next generation Japanese-led space mission to provide the ultimate limits on inflationary B-modes. LiteBIRD is the successor to the immensely successful Planck space mission, to be launched ~2030, and could potentially provide the best limits on inflationary B-modes on large scales (r<0.001). As part of LiteBIRD UK, U. Manchester is responsible for analysis pipelines for component separation and systematic error mitigation.

We are looking for a PhD student that is interested in radio data analysis techniques, with the potential to work on both low level (e.g. improving calibration of C-BASS/COMAP data, data reduction etc.) and high level (e.g. CMB component separation, Galactic science) analyses. The exact nature of the PhD will depend on the experience and interests of the student. Much of the work will be aimed at preparation for the LiteBIRD space mission, including simulations of foregrounds and potential systematic effects. Some of the work may also be applicable to the Simons Observatory (SO) project that Manchester is also involved in.

Supervisors: Jonas Latz (Maths), Anna Scaife (JBCA).

Project title: Graph-PDE-based segmentation of astronomical images

Project description (brief): In this project the student will develop a reconstruction-segmentation method using graph-PDE-based segmentation to separate out the components of astrophysical systems within observational data (Joint Reconstruction-Segmentation on Graphs).

Supervisor: Laura Wolz

Project title: HI intensity mapping with MeerKAT

A key goal of cosmology is to understand the accelerated expansion of the Universe, believed to be driven by a force called Dark Energy. Mapping the distribution of galaxies throughout the Universe’s lifetime can measure the expansion history and help us understand the nature of Dark Energy. Historically, cosmologists have successfully used the optical emission of stars located in galaxies to map the cosmic web over time. In the past decade, a new method called intensity mapping has emerged which uses the radio emission of gas (specifically the highly abundant Neutral Hydrogen gas) to trace the galaxy distribution. The future Square Kilometre Array (SKA) and its pre-cursor MeerKAT are enormous radio telescope arrays, capable of higher sensitivities and spatial resolution than any existing radio instrument. Intensity mapping is a unique probe, as it can be observed using the SKA as a single dish array, as well as in interferometric mode which gives much higher spatial resolution in the data. Both datasets are essential if we aim to acquire a complete understanding of how hydrogen traces dark matter and how gas and galaxies evolved with cosmic time.

PhD projects are available to work on the on-going MeerKAT data analysis, both in Single Dish as well as in interferometric mode, as well as the simulation of data including instrumental effects. Topics for exploration include optimisation of the HI and cosmology constraints by combining information from both data types, improvement of the data reduction pipelines as well as preparations and forecasts for SKA observations. Most project work will be computationally and the student will work within the teams of MeerKAT and SKA intensity mapping.

Some background reading can be found at the link below:

• Extracting Hi Astrophysics from Interferometric Intensity Mapping

Supervisor: Patrick Leahy

Co-supervisors: Ian Browne and Peter Wilkinson

Project title: Is there a new radio background? The L-Band All-Sky Survey (L-BASS) 

The aim of the L-BASS project is to map the intensity of the radio sky at ~1.4 GHz with unprecedented absolute accuracy (0.1K) – ten times better than achieved by Penzias and Wilson in their discovery of the cosmic microwave background radiation. There are several reasons to do this, the most exciting being that it should help settle a current astrophysical puzzle about the reality of excess all-sky low frequency emission of unknown origin (the “ARCADE-2 controversy”). In addition, our sky map will have impact on Galactic astrophysics and facilitate the interpretation of future Cosmic Microwave Background observations through having better calibrated maps of galactic synchrotron emission.

During the PhD project the student will produce and interpret the first sky maps made with data from the special purpose L-BASS telescope system which is situated at Jodrell Bank Observatory. The system, which is based on two large horn antennas, is now up and running. This PhD project involves a mixture of hands-on work to optimize the calibration of the system, making precisely calibrated observations, analysing the data followed by the astrophysical interpretation of the results. To achieve the required accuracy (0.1K) requires particularly careful calibration using a cryogenically cooled reference load of known physical temperature; the assembly and testing of this cryogenic calibration load will take place during the next eighteen months and form a significant part of the project.

Supervisors: Dr Michael Keith and Dr Patrick Weltevrede

Contacts: Dr Michael Keith and Dr Patrick Weltevrede

Project title: Mining the Jodrell Bank pulsar timing data archive

The pulsar group at Jodrell Bank has been studying pulsars for over 50 years, and regularly observes around 800 pulsars. These datasets usually stretch back to the discovery of the pulsar, and are therefore the most complete records of pulsar arrival times in the world. Using these data we can track the rotation of the pulsar, and in most cases this means that we unambiguously know when every rotation of the pulsar occurred since it was first observed.

These data are a valuable tool for understanding the complex physics that governs the rotation of pulsars. In particular, it is of great interest to understand how pulsar rotation evolves over time, and how we can characterise and understand the rotational instabilities in the pulsar. We also can use the data to track the position of pulsars over time, and hence get some understanding of the velocity distribution of the pulsars, which we can use to understand the processes leading to the birth of neutron stars.

In this project you will be tasked with extracting deeper understanding of pulsars by applying modern data science techniques such as Bayesian Analysis and Gaussian Processes to the Jodrell Bank pulsar timing database.

Supervisors: Andrina Nicola and Jens Chluba

Project title: A multi-wavelength approach to jointly constrain cosmology and baryonic feedback

Baryonic feedback, i.e. the impact of energetic astrophysical processes on the distribution of matter in the Universe, is one of the largest systematic uncertainties in weak gravitational lensing (WL) cosmology. Although baryons make up only a small fraction of the Universe's total matter, feedback from Active Galactic Nuclei and Supernovae can significantly redistribute matter on small scales. The size of this effect is not yet well constrained, and unless it is brought under control, next-generation surveys such as Euclid, LSST and Roman will not be able to deliver the precision cosmology they were designed for.

Constraining baryonic feedback requires a multi-wavelength approach. WL is sensitive to the total matter distribution, while probes such as the thermal Sunyaev-Zel'dovich (tSZ) effect and X-ray emission trace the baryons specifically. Combining these probes offers a route to tight and robust joint constraints on cosmology and baryonic feedback, thus overcoming one of the biggest challenges for WL cosmology to date.

In this project, you will:

• Develop a self-consistent theoretical model of WL, tSZ and X-ray observables based on the baryonification approach.
• Validate this model against state-of-the-art hydrodynamical simulations that jointly model gravity and gas physics.
• Apply the model to a joint analysis of several surveys JBCA is heavily involved in, to simultaneously constrain cosmology and baryonic feedback. Specifically, you will use WL from Euclid and LSST, tSZ from the Simons Observatory, and X-ray data from eROSITA.

The successful candidate will gain expertise in cosmological simulations, multi-probe statistical inference and Machine Learning, and the analysis of flagship survey data, working at the interface of theoretical modelling and observational cosmology.

This project will require a keen interest in both theoretical and observational cosmology, excellent coding skills and ideally experience with cosmological data analysis and interpretation.

Supervisor: Emmanuel Bempong-Manful

Project title: Numerical simulations of relativistic jet interaction with the intracluster medium

Although relativistic jets launched from active galactic nuclei (AGNs) are implicated in feedback processes, the detailed physics of how they interact with and shape the surrounding intracluster medium (ICM) remains open. In this project, we will employ state-of-the-art relativistic magnetohydrodynamic (RMHD) codes to conduct high-resolution numerical simulations and model the complex fluid dynamics, magnetic field amplification, and shock physics that occur as powerful plasma jets propagate into the hot, magnetised gas of galaxy clusters. Our primary goal is to test evolutionary models of jet propagation in the dense ICM and quantify the mechanical and thermal feedback mechanisms – such as cosmic ray transport, turbulence generation, and bubble cavity formation – that prevent the ICM from undergoing runaway cooling and quenching star formation in central dominant galaxies. The results will have implications for our understanding of the physics of cosmic feedback loops (jet-environment interactions) and galaxy cluster evolution.

Supervisory Team: Dr Paddy Leahy (primary), Professor Anna Scaife (secondary)

Contact: Paddy Leahy

Project title: Probing the Galactic magnetic field with POSSUM

Our Galaxy’s magnetic field plays an important role in the interstellar medium, sometimes dominating the local dynamics and rarely negligible. It helps regulate star formation and accelerates some particles to relativistic energy, forming cosmic rays. These processes are not fully understood, and nor is the structure of the magnetic field, which is often described as turbulent although some organized patterns can also be discerned. There are many observational tracers of the interstellar field, but they all have severe limitations, and so we must make progress by using observational clues to guide theoretical and computational modelling.

One of the most important magnetic tracers is Faraday rotation: the plane of polarization of radio waves change with wavelength, at a rate proportional to the integral the magnetic field component along the line of sight, weighted by the free electron density. This can be relatively easily assessed using extragalactic radio sources, which give us the integrated Faraday rotation along the sight line through the Milky Way. Mapping this across the sky gives a weighted 2D projection of the 3D magnetic pattern. The quality of this information is about to be vastly increased by the Polarization Sky Survey of the Universe’s Magnetism (POSSUM), an international project to measure the polarization of radio sources across most of the sky at 800-1088 MHz, using the Australian SKA Pathfinder (ASKAP), which will give a 10-fold increase in sampling of the Faraday rotation pattern.  In addition to extragalactic sources, POSSUM will detect polarization from synchrotron emission in the interstellar medium, which in principle gives more direct information about the 3D field structure, although there are challenging instrumental issues that have to be overcome.

The aim of this PhD project is to study the statistical structure of the Faraday rotation and the underlying magnetic field. The primary observational input will be the POSSUM Faraday results: about one quarter of the sky will have been observed in time to use,  including fields close to the Galactic plane with a long sight-line through the disk, and at high latitude where we are looking just through the local layer of the Galaxy. You will draw on other observational results as needed, for instance single-dish observations of the Galactic synchrotron emission at short wavelengths, where the polarization traces the field component in the sky plane. You will interpret your results using theories ranging from toy models that can be run on a laptop to the outputs of massive full-disk magneto-hydro-dynamic simulations being run by Rowan Smith and collaborators.

Supervisors: Laura Wolz, Andrina Nicol

Project title: Powering up cosmological probe combination via intensity mapping with galaxy clustering

Neutral hydrogen intensity mapping is a novel, very promising cosmological probe, which uses the integrated 21cm emission from neutral hydrogen to trace the Large-Scale Structure (LSS) of the Universe. Intensity mapping surveys are uniquely suited to probe the largest spatial scales in the Universe and can reach redshifts of up to z~5. These surveys are thus highly complementary to other cosmological probes, in particular galaxy redshift surveys that probe the LSS through the clustering of galaxies, but only reach redshifts up to z~2 and which are systematics-dominated at large spatial scales. 

The next few years will see first data releases from a number of high-precision surveys in the optical, such as the Rubin Observatory Legacy Survey of Space and Time (LSST) and Euclid, which will be accompanied by ongoing intensity mapping surveys such as MeerKLASS and followed by the upcoming SKA telescope in the 2030s.

These surveys are expected to deliver strong constraints on the pillars of our cosmological model, such as Dark Energy and Inflation, but will be dominated by systematic rather than statistical uncertainties. Joint probe analyses offer the redundancy and complementarity needed to obtain both robust and tight constraints on the physics of our Universe. This is particularly important in the light of tensions that have recently appeared within the standard model of cosmology, such as the Hubble tension or recent signs for dynamical Dark Energy. If confirmed, these could constitute a paradigm shift in cosmology, and so it is of utmost importance to ensure that our conclusions are not driven by systematics in the data. In order to optimally benefit from the experimental advances in both optical and radio surveys, it is therefore crucial to develop methods to jointly analyze these high-precision data sets in order to ensure robustness of results.

In this project, you will develop novel methods to combine intensity mapping and galaxy clustering surveys, focusing on power spectra as well as higher-order statistics such as the bispectrum. In a first step, you will investigate optimal ways to combine these surveys, comparing different statistics and survey configurations, with a particular focus on understanding the relative benefits of single dish and interferometric approaches in the radio. A particular focus will be to investigate and optimize the constraining power of these analyses on the physics of Inflation through the non-Gaussianity parameter f_NL, as well as their ability to shed light on dynamical Dark Energy and the Hubble tension. In a second step we aim to apply some of these methods to the combination of intensity mapping from MeerKLASS with early data from LSST and Euclid, thus leading the way towards application of these methods to SKA.

This project will require a keen interest in both theoretical and observational cosmology, excellent coding skills and ideally experience with cosmological theory prediction tools and parameter inference.

Supervisor: Dr Michael Keith

Contact: Dr Michael Keith 

Project title: Pulsar Timing Arrays for the detection of Nanohertz Gravitational Waves

In June 2023 the European Pulsar Timing Array (EPTA) and collaborators around the world announced the first evidence for a background of ultra-low frequency gravitational waves from super-massive black hole binaries in the centres of distant galaxies:

• Astronomers find first evidence for new class of gravitational waves which could unveil origin of the Universe

A pulsar timing array makes use of high precision (<1 microsecond) timing measurements of the rotation of millisecond pulsars to form a galaxy-scale gravitational wave detector. We can directly detect the gravitational waves through the quadrupolar correlated variations in the arrival times of pulses from the pulsars.

The Lovell Telescope at Jodrell Bank has been observing these pulsars for decades, and these observations are a key part of the EPTA and International Pulsar Timing Array datasets. More recently new telescopes have also started contributing highly sensitive observations of additional pulsars. In Manchester we are continuing to provide data from the Lovell Telescopes, as well as from the MeerKAT telescope in South Africa, an important precursor telescope for the upcoming international Square Kilometre Array Telescope. We are also involved in studying the pulsars themselves, including the effort to better understand the 'foreground' pulsar noise which can mask the signal from the gravitational wave background.

This project will involve working with the observations from the Lovell Telescope and MeerKAT for Pulsar Timing Array work, and developing new data analysis techniques, with potential to explore ways to exploit recent developments in machine learning. The overall goal is to better understand the signals that we see in the pulsar timing array data, leading to a clear and unambiguous detection of the gravitational wave background.

Supervisor: Phil Bull

Project title:  High-dimensional Bayesian modelling of the radio sky

Cosmological radio signals, particularly from the 21cm line from neutral hydrogen, are often swamped by much brighter foreground radio emission from our own Galaxy (and others). Separating out the faint signal from all the other contaminants is a delicate task, and is prone to problems such as signal loss (accidentally subtracting some of the cosmological signal) and other systematic effects. We have developed advanced data analysis techniques to perform this separation and model tricky instrumental effects as part of a large Bayesian statistical approach involving up to millions of parameters.

In this project, you will apply our statistical analysis code, called Hydra, to analyse two of the most sensitive datasets in 21cm cosmology – 3D maps from the MeerKLASS survey, which seeks to measure the baryon acoustic oscillations around redshift 1; and interferometer data from the Hydrogen Epoch of Reionization Array (HERA), which targets the 21cm signature from neutral gas around the first stars and galaxies.

The project will involve working with diverse radio telescope data, using Bayesian statistics and high-performance computers, and building models for different effects that crop up in the data as we analyse it. You will be part of a larger team working on related radio data and statistical analysis problems, and will collaborate with the broader MeerKLASS and HERA international collaborations.

Supervisor: Professor Lucio Piccirillo

Project title: Quantum Noise in amplifiers for radio astronomy

Quantum mechanics dictates a fundamental limit to the noise added by any coherent amplifier, such as a device that amplifies an input signal while maintaining its phase. This quantum noise, when expressed in terms of a noise temperature, is of the order of hv/k. High Electron Mobility Transistors (HEMT) based amplifiers today represent the state of the art in term of low noise amplification for radio astronomy or other fields where low noise amplification is needed like, for example, quantum computing. The best HEMT amplifiers today generate noise in the range to 5 to 10 times the fundamental quantum limit when cooled at cryogenic temperatures - typically around 20K. When these amplifiers are further cooled down - even to around 1K as first demonstrated by our group a few years ago - they do not exhibit a further reduction in noise.

Our group has been active for many years in trying to understand the physics behind this “flattening” of the noise temperature as a function of the physical temperature. We currently believe that the extra noise mechanism is due to self heating happening in the 2deg region of the transistor where amplification is achieved but also power is dissipated. HEMTs are designed not keeping in mind that at cryogenic temperatures the phonons generated inside the device do not have an efficient path to get in thermal equilibrium with the cold finger responsible for the cooling. The current research in our group is addressing this problem trying to find mechanisms to facilitate the thermalisation of the phonons with the heat sink.

The PhD student will be part of our group addressing this problem through the following lines of research:

1. Refining the theoretical basis of noise generation versus physical temperature in HEMT transistors;
2. Design and realise new measurement systems to characterise HEMT transistors
3. Processing of HEMT transistors chips to improve the self-heating
4. Full characterisation of HEMT LNA amplifier using improved transistors.

The student will have access to a variety of techniques and instrumentation: HFSS CAD design and simulations; cryogenic systems design, realization and testing; high vacuum and cryogenic techniques; RF measurements. Part of the work involves the usage of the National Graphene Institute where some processing of the HEMT transistors is carried out: etching, evaporation, sputtering, SEM and TEM microscopy, laser and e-beam writing.

Supervisors: Prof Anna Scaife and Dr Emmanuel Bempong-Manful

Project title: Representation learning for radio astronomy in the era of big data

The use of self-supervised deep-learning to build foundation models that learn useful representations from large volumes of data is being developed across multiple branches of astrophysics. As in other applications of AI, these foundation models are expected to transform how AI is used in astronomy. The latent space representations from these models can either be used directly to understand and examine structures within the data - for example finding rare types of galaxy; or they can be fine-tuned to address specific supervised problems where large labelled datasets are not available - for example, classifying radio galaxies into different types. With next generation ground-based facilities, such as the SKA expected to generate data volumes on the order of PB scale, such AI assisted latent space representation models will become a powerful tool for retrieving detailed information from observations.

In this project the student will explore the use of such representations for radio galaxy data from the Radio Galaxy Zoo (RGZ) project. This work will initially build on the RGZ representation learned in Radio Galaxy Zoo: Towards building the first multi-purpose foundation
model for radio astronomy with self-supervised learning, and the student will look at how to define different structures/clusters of objects within the latent space, as well as how to extract samples of different types of radio galaxy for scientific analysis. Further work may include multi-wavelength / multi-scale follow-up of interesting populations and/or further development of the foundation model paradigm.

Supervising team: Michael Brown, Daniel Thomas, Erik Rosenberg

Project title: Studying the beginning of the Universe with Simons Observatory

Simons Observatory (SO) is a Cosmic Microwave Background (CMB) Observatory located in the high Atacama Desert in northern Chile. It is the premier CMB facility in the world. Its primary objective is to search for a signal, originating from the very early Universe (~10^{-34} seconds after the beginning of the Universe). A detection of this signal (termed "primordial B-modes") would represent a phenomenal breakthrough for fundamental physics. There is no other way to explore physics at these energy scales.

The JBCA leads the UK contribution to SO. The UK is providing: (i) two of the seven SO telescopes, (ii) a UK-based data centre for processing the large data volumes and (iii) a program of algorithm development aimed at turning the raw data from the ~100,000 detectors into higher-level data products and scientific results.

We seek high-calibre and highly motivated individuals to join the project team and help deliver this world-leading cosmology project. Opportunities exist across all areas, but we are particularly interested in recruiting students to work on the commissioning of the two UK telescopes and on the analysis of the early data from these cutting-edge instruments.

The project will be a mixture of hand-on experimental work, remote telescope operations, and computer-based data processing and scientific analysis. Candidates must have excellent experimental and computing skills. They will need to undergo a medical assessment for working at extremely high altitude (5,200m) and they must be prepared to deploy to Chile for a significant duration (several weeks at a time), potentially on multiple occasions during their PhD.

Supervisor: Professor Lucio Piccirillo

Project title: Superconducting parametric amplifiers for radio astronomy

Thin films made of superconducting materials exhibit a non-linear inductance caused by the kinetic properties of Cooper pairs and free electrons. When a non linear element is “pumped” with an AC signal, parametric amplification is generated. Our group has been building superconducting resonators optimised to work as narrow-band high gain parametric amplifiers. Being composed exclusively of superconducting material, the resistance of the amplifier is identically zero e this allows the amplifier to work with a near quantum limited noise. Under the proper conditions, sub-quantum noise level can be obtained through the so called “noise squeezing”.

This PhD concern the design, fabrication and testing of superconducting parametric amplifiers. The processing will happen in the National Graphene Institute where the student will be allowed access, after proper induction. Once amplifiers are designed and manufactured, the student will use our sub-K test bed to properly characterise the noise, the gain and other important parameters. These amplifiers will then be used in state-of-the-art astrophysics projects like, for example, our axion search experiment under construction in our laboratories.

Supervising team: Richard Battye, Jens Chluba, Scott Kay

Project title: SZ cluster cosmology with the next generation of CMB surveys

The Sunyaev-Zeldovich (SZ) effect is the spectral distortion of the CMB resulting from the inverse Compton scattering of CMB photons off free electrons. It is a powerful method for constraining cosmological parameters (e.g. from counting galaxy clusters) as well as measuring the distribution and properties of hot gas in groups, clusters and filaments, where most of the Universe's baryons are thought to reside. Recent CMB surveys using the South Pole Telescope, Atacama Cosmology Telescope and the Planck satellite have produced large samples of clusters and sky maps of the SZ effect.

Over the coming decade, the next generation of CMB surveys such as with the Simons Observatory (SO) and LiteBIRD will vastly increase the SZ data volume, allowing us to measure the SZ effect with greater sensitivity and for much larger samples of objects extending to lower masses and higher redshifts.

The JBCA is playing a leading role in the SO project in particular, and will be involved with many aspects of the SO SZ analysis. Novel new areas include using relativistic effects to probe the gas temperature and velocity structure, and improving SZ models guided by state-of-the-art hydrodynamical simulations. In this PhD project, the student will join JBCA's team of SZ scientists to work on the development of theoretical SZ models and analysis of SZ data. The supervisory team of Battye, Chluba and Kay, together, have extensive and complementary experience in SZ theory, modelling (including simulations) and cosmological data analysis. The student will also have the opportunity to interact with relevant international collaborations on both theory and observational sides.

Supervisory Team: Dr Paddy Leahy (primary), Dr Emmanuel Bempong-Manful (co-supervisor)

Contact: Paddy Leahy

Project title: Testing models of relativistic jets with e-MERLIN

Many of the issues in understanding structure formation and black-hole growth are thought to be resolved by suitably-tuned feedback of energy and momentum from AGN activity – from dense to diffuse phases of matter.  Relativistic plasma jets of radio-loud AGNs are particularly thought to be responsible for the production of the most energetic photons and hadrons in the observable Universe, thereby playing a key role in this feedback cycle. However, the physics driving the observed jet structure in these cosmic outflows remains an open question. To resolve them we need to quantify the mass, momentum and energy inputs from jets and to work out how they interact with their environment.

Here at Manchester we are leading an ambitious global effort  (The e-MERLIN Jets Legacy programme) to resolve these and other key questions in extragalactic jet physics. The programme has been mapping a number of powerful radio galaxies and quasars, allowing us to study for example the detailed structure of magnetic field in relativistic jets (the synchrotron radiation polarisation) and in the foreground gas (via Faraday rotation). Deep LOFAR observations have also been acquired for the programme, providing us with the unique opportunity to probe the dynamics and energetics of relativistic jets over broad frequencies.

The goal of this PhD project will be to utilize these new radio observations to map the jet structure of some well known powerful radio galaxies in order to test the relativistic beaming model in jets. The student will join the e-MERLIN Jets Legacy collaboration and become a part of the LOFAR-VLBI Working Group. Depending on their interests, the project could be extended into a multiwavelength campaign with complementary observations at optical and X-ray wavebands, and/or perform numerical MHD simulations to model the jet kinematics and compare theory with observations.

Supervisor: Rachel Cochrane

Project title: Simulating early galaxy evolution in extreme environments

Galaxy protoclusters, the progenitors of modern-day galaxy clusters, are sites of accelerated evolution of galaxies. These overdense large-scale environments are believed to host the most massive galaxies at high redshift, including the early-quenching massive galaxies studied with JWST (e.g. [1], [2]). However, detailed theoretical study of the physics driving rapid star formation and quenching within protoclusters has been limited by the computational challenge of simulating such massive structures at high resolution.

Recently, collaborators have expended >5 million CPU hours modelling a massive galaxy cluster that reaches 10^14 solar masses by z=2. This cluster simulation was run with the FIRE-3 model [3], a state-of-the-art code that models both stellar and AGN feedback within a resolved interstellar medium. The data are now ready to explore. In this project, we will characterise the assembly of central and satellite galaxies within this simulation. We will first explore the evolution of key galaxy properties, such as galaxy stellar masses, star formation rates, gas fractions, and black hole masses, and characterise the differences between the evolution of galaxies within the cluster core, outskirts and field. We will investigate the timing and drivers of satellite galaxy quenching, including the roles of gas starvation, ram pressure stripping, and AGN feedback. The PhD student will be trained in performing detailed radiative transfer forward modelling on the galaxies, to generate predictions for the observable multi-wavelength emission (e.g. [4]). We will hence characterise the observability of the protocluster and its constituent galaxies with different tracers (e.g. the Lyman break, line emission, and sub-millimeter continuum emission) at different epochs. 

References

1. Abstract: A population of faint, old, and massive quiescent galaxies at 3 <z <4 revealed by JWST NIRSpec Spectroscopy
2. Abstract: Explaining Ultramassive Quiescent Galaxies at 3 < z < 5 in the Context of Their Environments
3. FIRE-3: updated stellar evolution models, yields, and microphysics and fitting functions for applications in galaxy simulations
4. Abstract: Predictions for the spatial distribution of the dust continuum emission in 1 < z < 5 star-forming galaxies

Supervisor: Dr Eamonn Kerins

Project title: Towards 100,000 exoplanets with the NASA Roman mission

Microlensing is proving to be the most capable method to find cool low-mass planets, including planets around the most common types of star, and planetary architectures that most resemble that of our own solar system. The demographics of these planets is also crucial for testing planet formation theories.

In October 2026 NASA will launch the $4.3 billion Nancy Grace Roman Space Telescope (Roman), which will undertake a dedicated exoplanet microlensing survey over 5 years. Roman will have a similar sensitivity and resolution to the Hubble Space Telescope, but will have 100x its field of view and 1400x its survey speed. Roman will revolutionize our understanding of exoplanet demographics by discovering up to 200,000 new hot and cool exoplanets over Galactic distances using the transit and microlensing methods. Dr Kerins is the European Space Agency (ESA) appointed scientist to the NASA Roman exoplanet survey, and he leads one of the major working groups investigating exoplanet demography with Roman.

In March 2025 the ESA Euclid space telescope conducted a high resolution imaging survey of the inner Galactic region where Roman will observe. This survey was planned by the ESA Euclid Exoplanet Science Working Group, which Dr Kerins has led. The survey will enable Roman to determine the masses of its detected planets with significantly improved precision. But we are also using the Euclid data to improve modeling for thousands of microlensing events, including many planetary systems, observed in the past by ground-based surveys.

At Manchester we are developing a sophisticated Roman transit and microlensing simulation framework. This is being used to help us determine the ability of Roman to constrain the architecture of exoplanetary systems across the Galaxy, measure the Galactic abundance of planets in the stellar habitable zone, and to aid in the modeling of planetary systems using Euclid and upcoming Roman data.

PhD projects are available to undertake studies in one or more of these areas. The projects are computational in nature, requiring using both existing parallelized python codes and developing new ones. We have a 64-core AMD Threadripper machine dedicated to our exoplanet work. The exoplanet group at Manchester working with Kerins currently comprises five PhD students and one MSc student. This research will contribute to our group's ongoing contribution as part of the NASA Roman Galactic Exoplanet Survey Project Infrastructure Team.

Supervisor: Professor Anna Scaife

Project title: View-based equivariance in multi-source deep-learning for radio astronomy

The volume of labelled data available in astronomy for supervised deep-learning applications is very small compared to other fields. However, given the significant volume of unlabelled data available, the potential of self-supervised learning approaches to underpin the use of AI in astronomy is high. This project will look specifically at the use of view-based self-supervised learning (SSL) for building astronomical foundation models. View-based SSL uses cost functions based on the mutual information between the embeddings for different views (augmentations) of the same data sample. The resultant embedding should be a structured, high-level representation of the training data that is invariant to those views.

Since archival astronomy data are likely to be highly heterogeneous with regard to their sensitivity (noise level), distribution of spectral (frequency) information and, in the case of radio interferometers, the range of spatial scales probed, a potential negative consequence of building foundation models from multi-source databases is that they do not learn the joint distributions of causally-relevant features related to the astrophysical systems under observation, but rather the characteristic parameters of the different types of observation (i.e. sensitivity, resolution, observing frequency etc).

In this project the student will examine whether equivariance to observational characteristics can be enforced in astronomical deep-learning models through the use of multi-source views from archival multi-survey data. They will then go on to examine the sensitivity of various model properties such as generalisation, quantification of dataset-shift and fine-tuning to a multi-source training regime, using test data from SKA pre-cursor telescopes such as MeerKAT and ASKAP.

References:

• Radio Galaxy Zoo: Towards building the first multi-purpose foundation model for radio astronomy with self-supervised learning
• Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style