Cosmology

Supervisors: Will Coulton, David Alonso

A wealth of cosmological observations are providing unprecedentedly precise measurements of the 3D structure of our Universe. Buried in these detailed maps of the cosmos are tiny statistical signatures that encode the physics of the processes that shaped the evolution of the Universe and, hopefully, smoking gun signatures from the Universe’s beginning. A key challenge is how to disentangle the primordial signals from the dominant, late-time signals to ensure our understanding of the primordial universe is robust. This PhD project will focus on developing new statistical tools to address that problem and applying them to new datasets.

The Oxford cosmology group has expertise in many on-going surveys, including the Simons Observatory and the Vera Rubin Observatory and this project will use a mixture of analytical, computational and statistical tools to search these datasets for primordial signals. The goal will be to drive forward our understanding of this key epoch in the history of the Universe.

References:
Coulton et al (2023)  https://arxiv.org/abs/2206.01619
Yap, Coulton, Iršič (2026)  https://arxiv.org/abs/2609.35713
Yap, Coulton, Hou, Gratton (2026) https://arxiv.org/abs/2610.08671

Supervisors: Maria Vincenzi, Stephen Smartt, Matt Jarvis

It’s an exciting time for cosmology. Over the past two years, numerous independent cosmological studies have been published using very different techniques (Type Ia supernovae, baryon acoustic oscillations and the cosmic microwave background). Remarkably, taken together, these findings seem to point to the same conclusion: the properties of dark energy are more complex than previously understood.

Type Ia supernovae (SNe Ia) are among the most well-established tools in modern cosmology, with dozens of successful cosmological experiments conducted over the past three decades. After 15 years of design and construction, the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) began in summer 2026. LSST is expected to transform the field: in its first year of operations alone, it will discover more SNe Ia than have been observed since the field began.

Our team is heavily involved in the analysis of the first LSST transient data. We work on a wide range of projects, both observation-oriented and simulation-oriented. They revolve around developing and testing transient discovery and classification pipelines, optimising LSST spectroscopic follow-up programmes, studying the host galaxies where these SNe explode, and designing new approaches to SN cosmology to minimise sources of systematic biases.

With other UK scientists, we are also involved in the TiDES spectroscopic follow-up programme. It will use the 4MOST spectrograph to follow up tens of thousands of LSST transients and their host galaxies, providing crucial insights into their properties and redshifts.

Besides LSST, which will mainly discover high-redshift (z > 0.1) SNe Ia, we are leading the cosmological analysis of the ATLAS SN Ia sample, the largest low-redshift (z < 0.1) SN Ia sample ever compiled for cosmology.

The DPhil project will focus on preparing for future SN Ia cosmological measurements using LSST, TiDES and ATLAS data, with an emphasis on addressing the key sources of uncertainty that affect these measurements.

Supervisors: David Alonso, Matt Jarvis

The spatial distribution of galaxies and other tracers of the large-scale structure (LSS) are crucial observables in our quest to understand the origin and evolution of our Universe, as well as the complex physics that governs the formation of galaxies within the cosmic web of dark matter. Key open questions are:

1. The nature of dark matter and dark energy, and their impact in the evolution of cosmic structure.
2. The physics of feedback from supermassive black holes and supernovae, and their impact on the formation and evolution of stars and galaxies.
3. How these processes and galaxy properties are related to their dark matter halo environments 

In this project, we will combine state-of-the-art LSS datasets with novel theoretical and statistical approaches to address these questions. On the observational side, we will leverage Oxford's access to deep-field galaxy survey data, providing pristine measurements of clustering on small scales, with wide-field probes of the large-scale structure across multiple wavelengths. These will include the first galaxy and weak lensing data from the Vera C. Rubin Observatory and Euclid, and the first maps of the microwave sky from the Simons Observatory. On the theory side, we will combine analytical approaches describing the physics of the small-scale clustering of matter, gas, and galaxies, with simulation-based techniques providing an accurate connection with the large-scale cosmic web.

The project itself can evolve in various directions depending on the interests of the student, with the overarching aim to bring about a deeper understanding of galaxy formation and evolution to gain new insights into the process of structure formation at all scales. The project is well suited for students with an interest in combining both theory and observations. It will provide experience in multiple areas of data analysis, from images and maps to science-ready observables, and it will introduce the student to the latest theoretical tools to model the formation of cosmic structures and the physics of galaxies and the inter-galactic medium.

Further reading:

• Hamlett et al. 2026. https://arxiv.org/abs/2610.23567 
• Pellejero-Ibanez et al. 2026. https://arxiv.org/abs/2607.09571 
• Boreiko et al. 2026. https://arxiv.org/abs/2601.07799 
• Verdiani et al. 2026. https://arxiv.org/abs/2510.17796
• Stylianou et al. 2026. https://arxiv.org/abs/2610.19030 
• Wayland et al. 2026. https://arxiv.org/abs/2609.31774 

Supervisors: Deaglan Bartlett

The next generation of astronomical surveys, including Euclid and the Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST), will map the distribution of millions to billions of galaxies across the Universe. These observations offer unprecedented opportunities to understand the nature of dark matter and dark energy, test our theories of gravity, and uncover how cosmic structure formed. However, extracting this information requires accurate predictions for the complex, nonlinear processes governing the evolution of matter and the formation of galaxies.

Cosmological simulations provide a powerful way to model these processes, but traditional statistical methods struggle to exploit their full information content. Moreover, the computational cost of generating sufficiently accurate simulations across a wide range of cosmological and astrophysical models presents a major challenge.

In this project, we will develop and apply new machine-learning and statistical methods to overcome these limitations. A particular focus will be simulation-based inference, which enables cosmological parameters to be inferred directly from simulations without requiring an explicit likelihood function. We will investigate how to combine approximate simulations, machine-learning models, and high-resolution numerical calculations to generate accurate predictions efficiently, while accounting for uncertainties in galaxy formation and other astrophysical processes.

The student will have opportunities to develop new computational and statistical methods, investigate novel ways of extracting information from the spatial distribution of galaxies, and apply these techniques to observations from major galaxy surveys, including Euclid. The project will combine theoretical cosmology, numerical simulations, Bayesian statistics, and machine learning, with the precise direction tailored to the student's interests.

Supervisors: Julien Devriendt, Adrianne Slyz

Stars in the Universe form predominantly in clusters within giant galactic molecular clouds. This clustered mode of star formation has profound implications for the formation and evolution of galaxies, as it concentrates stellar feedback in both space and time. Ionising radiation bubbles and supernova explosions emanating from massive stars within clusters overlap, changing how stellar feedback energy couples to the interstellar medium and driving powerful, mass-loaded gas outflows.

Whilst observations of star clusters have traditionally been the preserve of the local Universe, JWST, combined with gravitational lensing, now allows us to probe their formation at cosmic dawn. Magnifications of tens to hundreds resolve individual star clusters in galaxies from z=6 to beyond z=10. In the Cosmic Gems arc at z=10.2 for example, Adamo et al (2024) find five clusters younger than 50 Myr with masses around 10⁶ M₈ (see Fig 1.). These objects appear unusually bright for their host galaxies. Why the first clusters should be so luminous is an open question. They are also so dense that the stars at their very centre are believed to collide with one another, eventually fusing together to form intermediate mass black holes.

For this project, the DPhil student will perform numerical studies of the initial cluster mass function (ICMF) within the explicit framework of a cold dark matter dominated Universe, using the adaptive mesh refinement radiative hydrodynamics code RAMSES-RT (Rosdahl et al 2013). At first, they will be tasked 

with developing a stellar wind feedback module, building on the previous star cluster modelling work carried out within the group (Thompson et al 2026; see Fig. 2).

They will then move on to designing and running their own high-resolution simulations of galaxy formation, building on the newly developed module and radiative hydrodynamics, to incorporate both magnetic fields (Fromang et al 2006, Martin-Alvarez et al 2018) and cosmic rays (Dubois & Commercon 2015; Rodriguez-Montero et al. 2024). Scientific goals include assessing which feedback processes most strongly influence the ICMF, and in which way; investigating how these processes shape molecular clouds at both local and galactic scales particularly through their impact on the cloud mass function; assessing whether simulated ICMFs produce clusters bright enough to account for the JWST observations; and exploring how these ICMFs impact the initial black hole mass function predicted by stellar collisions within these clusters, as well as the resulting gravitational wave signals and rates of tidal disruption events.

The DPhil student will be part of an international collaboration (US, France, South Korea, UK) developing and exploiting state-of-the-art, high-resolution, and physics rich cosmological simulations.

Supervisors: Pedro Ferreira, Tariq Yasin, Harry Desmond

One of the open questions in physics is whether General Relativity and collisionless dark matter are really the whole story. Dark matter detectors have steadily closed off much of the parameter space for the most popular particle candidates without finding anything, and the cosmic expansion history on its own cannot distinguish between the many theories proposed to explain cosmic acceleration. Both fields have been looking under the same few lampposts for a while now, and it is time to look under new ones. Galaxies are an extremely promising hunting ground. Each occupies a different gravitational environment, ambient dark matter density and velocity field, providing a population of natural experiments in regimes inaccessible in the laboratory. For dark matter in particular, the phenomenology is rich. In the standard picture, a galaxy's stars, gas, dark matter and central black hole all fall together, but if dark matter interacts with itself, it can pull them apart. A halo dragged back as its galaxy moves through the surrounding dark matter can shift the stars off-centre, bend the disk into a warp, thicken it, or leave it lopsided. Each of these distortions points in a direction set by the galaxy's environment. The catch is that galaxies are messy systems, and ordinary astrophysics can affect the same observables. The project will tackle this with what we call "Messy Bayesian Forward Modelling": reconstructing each galaxy's latent environment and properties, predicting the effect of the dark matter interaction object by object, and marginalising over baryonic physics, halo structure, selection and measurement uncertainty. Rather than committing to a specific theory, we will develop general, phenomenological prescriptions for the effect of novel dark-sector physics, allowing us to cast our net as widely as possible in the search for the particle nature of dark matter.

Related works:

• SIDM constraints from galaxy warps: - https://arxiv.org/abs/1911.04497
• Messy Bayesian Forward Modelling - https://arxiv.org/abs/2009.08743v2
• Effective description of dark matter microphysics for structure formation: https://arxiv.org/abs/1512.05344
• Review of astrophysical tests of SIDM - https://arxiv.org/abs/2207.10638

Supervisors: Angela Taylor, Mike Jones, Jamie Leech

The search for the signature of inflation in the polarization of the cosmic microwave background (CMB)  is one of the most important topics in observational cosmology. The primordial B-mode (curl-like) polarization of the CMB contains information about the high-energy physics of the very early universe that cannot be obtained any other way. Detecting it will answer questions about the nature of inflation and the fundamental physics of the Big Bang. 

A major international effort is underway to detect the B-modes and determine other cosmological parameters, such as the mass of the neutrino. As part of this, the UK has joined the Simons Observatory (https://simonsobservatory.org), an international collaboration to detect the CMB B-mode polarization from a high-altitude site in Chile. We are building two Small-Aperture Telescopes (SATs) to join the three SATs already built in the US and operating on site, and one being built in Japan. There is also a Large-Aperture Telescope (LAT) to provide higher resolution data essential for removing the effects of gravitational lensing from the CMB. The UK SATs are being deployed in late 2026, with commissioning and then first science observations beginning in 2027.

This DPhil offers the opportunity to join the experiment at a particularly exciting stage, as the UK telescopes begin commissioning and make their first scientific observations. We are looking for a student to take part in the commissioning of the UK SATs, travelling to the site and developing the calibration and data analysis procedures, and taking part in the operations and science exploitation. Oxford has a unique role in the project -  we are currently analysing data from the existing SO telescopes and have designed and built the data readout system for the UK telescopes. The team therefore have a unique perspective on the data flow from raw detectors to final science results. We have also been leading an independent project to characterize the Galactic foreground signals that will tend to mask any primordial signal, and there will be the opportunity for the student to work on these data sets as well. The student will be joining a large and well-organised international collaboration, as well as the established team in Oxford and will have the opportunity to make significant individual contributions to this major project.

Supervisor: Aprajita Verma

Strong gravitational lenses are versatile probes of many open questions in cosmology and extragalactic astrophysics. While their rarity has limited their use to date, using exquisite high-spatial resolution wide field data from ESA’s Euclid space telescope we have increased the number of known strong lenses already by a factor of 15 with the promise of hundreds of thousands of systems to come with further Euclid data and the surveys carried out by Vera C. Rubin Observatory’s Legacy Survey of Space and Time and the Nancy G. Roman Telescope. The Euclid DR1 sample is an unparalleled treasury which we are collecting ancillary and follow-up data on. With this sample in the DPhil project, we can perform detailed studies of structural evolution of the deflector lenses and investigate how processes such as baryonic feedback and merger history have shaped galaxies. The near-infrared imaging provided by Euclid and Roman will produce large samples of lensed background galaxies that are at high redshift and/or obscured by dust. The magnification and amplification due to lensing allows us to understand the nature and structure of such galaxies at fainter intrinsic luminosities than would be possible with wide field surveys. These samples include strongly lensed quasars and compound lenses that provide a complementary probe to constrain cosmological models. The project will be tailored to the interests of the student. The student will have the opportunity to contribute to lens searches in new datasets (including citizen science), join and collaborate with large international consortia and apply for follow-up observations on interesting systems.