UCL's Centre for Doctoral Training in Data Intensive Science


Available PhD Projects

The Centre primarily carries out research in STFC's flagship Data Intensive Science projects, in High Energy Physics and Astronomy, which have been at the forefront of DIS research for several decades and provide the ideal training ground for DIS.

You can find below a provisional list of the studentships projects which will likely be available for the Sept 2023 intake. Applicants should keep in mind the projects outlined below are a starting point for a conversation on the exact research project that will be undertaken during the period of the PhD. Projects will be assigned after you have accepted a place in the CDT, with students able to further discuss their final project choice and topic with perspective supervisors at that point. For further details on any of the projects please contact the project supervisor.

Additional projects may be added to the list prior to the interviews and if there is an area which you would really like to undertake a project on, but you don't see listed below please get in touch with us: dis-cdt-phd-admissions@live.ucl.ac.uk 

COLLIDER/ATLAS - What happened one picosecond after the Big Bang? (or "Using novel ML techniques to search for Higgs pair production")

Supervisor: Prof Nikos Konstantinidis

The Electroweak Phase Transition, one of the most dramatic and defining moments in the evolution of our Universe, happened around 1ps after the Big Bang. The shape of the Higgs field potential (the famous "Mexican hat") is intricately linked to that moment. This project will continue the quest of searching for Higgs pair production, the most sensitive process to give direct access to the Higgs potential, employing novel ML techniques to boost the sensitivity of the analysis for the Run-3 data.

COLLIDER/ATLAS - Graph Neural Networks for High Energy Physics

Supervisor: Dr Gabriel Facini

Design and deploy a tool based on a UCL-designed graph neural network (GNN) to identify b-quarks (1) to look for new physics in the Higgs (H) sector. - Energy frontier. The most energetic Higgs Bosons could be the key to seeing the signs of BSM physics (2). Identify H->bb with pT > 1 TeV with GNN - Long-lived particles (3): Challenging because LHC experiments were not designed with them in mind. Identify H->bbbb with GNN. 1. ATL-PHYS-PUB-2022-027 2. arXiv:2111.08340 3. arXiv:2107.06092

COLLIDER/ATLAS - Feature identification in published Collider data

Supervisor: Prof Jon Butterworth

Measurements from the Large Hadron collider continue to probe the energy frontier. Their growing coverage challenges models for physics beyond the Standard Model, and may throw up anomalies when confronted with the predictions of the Standard Model. This project will use active-learning and related techniques to scan this data set, identify features in the parameter space of new physics models so that efficient constraints can be set, and highlight any anomalies.

COLLIDER/ATLAS - Revolutionising tracking and b-jet identification to explore new regions and understand the fundamental workings of the Universe

Supervisor: Dr Tim Scanlon

With the restart of data taking, the next 4 years is going to be an exceptionally exciting time to work on ATLAS, LHC. The center of mass energy has increased, and we will at least double the dataset we already have. This project will revolutionise the tracking and b-tagging algorithms using deep learning and graph neural networks, before using them to enhance searches for new physics using the Higgs boson. Improving these algorithms will also boost the entire physics output of ATLAS and beyond.

COLLIDER/ATLAS - Use of machine learning in the ATLAS trigger

Supervisor: Prof Mario Campanelli

Develop new machine-learning based algorithms for the current and the future trigger systems of the ATLAS experiment, and study production of hadronically-decaying vector bosons.

DARK MATTER / LZ - The search for Dark Matter with the world-leading LZ experiment

Supervisor: Prof Chamkaur Ghah

LZ is the most sensitive experiment in the search for galactic dark matter and has begun taking primary science data. LZ will explore entirely new electroweak parameter space towards a first discovery or constraints on the leading dark matter theories. The successful applicant will take a highly active role in the analyses from these datasets with genuine discovery potential across a range of dark sector, exotic neutrino and BSM physics.

NEUTRINOS/HIGH ENERGY COLLIDER - Reconstruction of tracks and vertices, and neutrino analysis in the SND@LHC detector

Supervisor: Prof Mario Campanelli

Using neural networks to reconstruct tracks and vertices from particles crossing the emulsions of the SND emulsion target.

HIGH-ENERGY PHYSICS/PROTON BEAM THERAPY - Machine Learning and FPGA optimisation for proton beam therapy

Supervisor: Dr Simon Jolly

Proton therapy is a more precise form of radiotherapy that provides significant benefits over conventional X-ray radiotherapy, particularly for children. At UCL we are developing new Quality Assurance detectors for measuring the size, position, range and dose of the clinical proton beam. This PhD project will focus on the fast reconstruction of proton range and position using FPGA’s, with the application of Machine Learning techniques to improve the speed and accuracy of these QA measurements.

ASTRO/COSMOLOGY/ASTROSTATISTICS - Probabilistic deep learning for cosmology and beyond

Supervisor: Prof Jason McEwen

In the proposed project we will develop probabilistic deep learning approaches, where probabilistic components are incorporated as integral components of deep learning models. Similarly, we will also develop statistical analysis techniques for which deep learning components are incorporated as integral components. For further details see: http://www.jasonmcewen.org/opportunities/#phd-project-1

ASTRO/EARTH SCIENCE - Geometric deep learning for global weather prediction 

Supervisor: Prof Jason McEwen

We will develop deep learning networks that can forecast weather systems natively over the spherical globe, without the need for projections, leveraging the very recent developments for the construction of scalable geometric deep learning approaches on the sphere. Such networks will scalable to sub-kilometre resolution and have the potential to dramatically improve weather predictions. For further details see: http://www.jasonmcewen.org/opportunities/#phd-project-2

ASTRO/EXOPLANETS - Applying Machine Learning techniques to quantum-mechanical calculations required to generate molecular spectroscopic data for models of exoplanets and cool stars

Supervisor: Prof Sergey Yurchenko

UCL is a world-lead at the provision of high quality molecular data obtained by rigorous and accurate ab initio solution of the equations of quantum mechanics (QM) covering a large range of astrophysical and atmospheric applications. This project will use, develop and apply novel machine learning techniques to tune QM procedures required to generate molecular spectroscopic data for models of exoplanets and cool stars.

ASTRO/EXOPLANETS - Understanding TESS exoplanets through machine learning

Supervisor: Dr Vincent Van Eylen

The possibility of planets orbiting other stars has been a topic of fascination for centuries. We are the first generation that has brought these planets from the realm of science-fiction into that of science. Today we know over 5000 exoplanets, and the ongoing NASA TESS mission is expected to increase this number by an order of magnitude. In this project, we address the challenge and seize the opportunity, that novel data sets of increasing magnitude, such as TESS, pose to exoplanet science.

ASTRO/EXOPLANETS - Bayesian methods in planet formation

Supervisor: Dr Mihkel Kama

The impact of Bayesian methods and machine learning is recently evident in exoplanet atmospheric retrievals, where they provide improved quality, scientific nuance, and accessibility. Similar techniques have not yet been widely applied in planet-forming environments. This is limiting our ability to interpret exoplanet data from JWST or Ariel. In this project, you will develop innovative techniques in the study of protoplanetary disks and planet formation, to enable new exoplanet science.

ASTRO/EXOPLANETS - Structure-based machine learning methods for accurate predictions of potential energy surfaces for polyatomic molecules

Supervisor: Prof Jonathan Tennyson

This project will aim at applications and development of efficient ML methods for generating highly accurate molecular PESs, important for quantum mechanics and molecular mechanics calculations, for computing accurate energies of conformations and clusters of organic and inorganic species, producing rotations-vibrational spectra of polyatomic molecules important for atmospheric, astrophysical and fundamental studies.

ASTRO/EXO-PLANETS + MACHINE LEARNING - Investigating performance, interpretability & resource efficiency tradeoffs of machine learning models for exoplanet data 

Supervisor: Dr Nikos Nikolaou

Training Machine Learning (ML) models and using them to perform inference can be very demanding in terms of data & computation. In scientific applications there is need for more resource-efficient approaches for training them while also requiring increased emphasis on model interpretability. The objective of this project is to explore the tradeoffs between predictive performance, model interpretability & resource-efficiency of ML algorithms applied to problems in exoplanetary science.


Supervisor: Prof Thomas Kitching

How do you build a 1000+ meter diameter telescope in space? You don’t, it builds itself! We will explore the possibility of creating autonomous cubesats constellations that can automatically form complex structures in-orbit such as gigantic telescopes; using the cubesats as “Lego” from which big space craft can be formed. By using ML tools developed we will explore three aspects of the problem: optimisation of fuel-efficient strategies, cubesat auto-orientation, and deployment to Edge devices.

ASTRO/GALACTIC - Finding all the Carbon Stars

Supervisor: Dr Jay Farihi

The ambitious aim of this project is to find all the carbon stars in the sky that are sufficiently bright to have Gaia DR3 data including BP/RP spectra. Their distinct molecular bands of carbon are amenable to machine learning techniques and can lead quickly to key publishable results such as a catalog itself, the fraction of dwarf (vs. giant) carbon stars in the Milky Way, and the fraction that are ancient, metal poor halo stars.

ASTRO/PLANETARY - Jupiter's X-ray Stock Market

Supervisor: Dr William Dunn

X-rays offer unique insights into habitability, formation & magnetic fields of planets. NASA & ESA flagship observatories(Chandra & XMM-Newton) awarded UCL the largest planetary X-ray dataset ever acquired, from interstellar comets to Uranus. This data is largely untouched & rich with discovery potential. After applying existing tools to a new observation, students can choose their direction including: stockmarket AI tools for X-ray timeseries; super res CNNs; YOLO algorithms classifying aurora.

ASTRO/GALAXIES - Galaxy properties and anomalies in DESI spectra

Supervisor: Prof Ofer Lahav

The DESI survey has already measured 12M spectra of the planned 35M. The project will focus on the full sample (to be completed during the PhD) in two directions: (1) de-noising galaxy spectra to extract galaxy properties; and (2) identifying anomalies in galaxy spectra. The work will use advanced Machine Learning techniques, including interface to human knowledge.

ASTRO/SPACE WEATHER - CIRCE cubesat mission for monitoring space weather using miniaturised spectrometers and satellite drag

Supervisor: Prof Anasuya Aruliah

The CIRCE mission is a miniaturised suite of space weather monitoring instrumentation carried onboard two 6-unit cube-satellites in a near polar Low Earth Orbit (LEO), with a string-of-pearls flight configuration. It will be the first launch from UK soil, at Spaceport Cornwall in early 2023. This is a cross-faculty collaboration with the Defence Science and Technology Laboratory to study and monitor a region of the upper atmosphere occupied by Low Earth Orbiting satellites.

ASTRO/INSTRUMENTATION - Genetic algorithm and design evolution of Metamaterial optics for Astrophysics instrumentation and Satellite communication

Supervisor: Prof Giorgio Savini

This project aims to employ Machine Learning and Genetic Evolution algorithms to efficiently evolve the design of optics and antenna receivers for satellite communication and astrophysical instrumentation. The design of components at sub-THz frequencies used both for astrophysics as well as for Earth Observing and high speed communication is based on 2D geometrical interacting metal structures. The degree of freedom in these design are numerous and require DI techniques for optimization.

NEUTRINO/COSMO/ASTRO/P-ONE - New frontiers in multi-messenger neutrino astronomy with machine learning imaging

Supervisor: Dr Matteo Agostini

Neutrinos from the edge of the observable universe are revolutionising our understanding of astrophysical systems at the ultimate energy and gravitational frontiers. Giant neutrino telescopes will soon get online, posing extraordinary analysis and computational challenges. This project will address them using statistical and machine-learning methods for image recognition, developed within a cross-disciplinary context and ultimately applied to physics analysis of the P-ONE neutrino experiment.

COLLIDER/COSMOLOGY - Simulation-based inference across accelerator physics and cosmology

Supervisor: Prof Benjamin Joachimi

We will develop and apply simulation-based inference and fast surrogate/forward-modelling techniques for state-of-the-art LHC and cosmological galaxy survey data, exchanging knowledge about methods and best practice between the two fields.