Previous Year's PhD Studentships Projects (for September 2022 start)
A list of projects for next year's intake will be available at the end of November 2022. In the meantime, you can find a list of the studentships projects that were available for last year's intake below, which is indicative of the projects that will be avaiable this year. Please do not contact the supervisors about these projects, because a new set of projects will be uploaded shortly. 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 are 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.
COLLIDER/ATLAS - Deep learning the shape of the Higgs potential with ATLAS at the LHC
Supervisor: Prof Nikos Konstantinidis
The LHC Run-3 (2022-26) is promising to be an exciting next step in the exploration of the Higgs properties. This project will focus on the search for Higgs pair production, in HH->bbbb, with data ATLAS will collect in Run-3. Higgs pair production holds the key to understanding the shape of the Higgs potential and its connection to new physics. Through the use of cutting-edge ML techniques in Run-3, we will have a real shot at observing this rare process for the first time ever!
COLLIDER/ATLAS - Modern ML for Long-lived particles
Supervisor: Dr Gabriel Facini
The ATLAS detector was designed to identify particles that have short lifetimes and decay near the collision point. With no hints for new physics, odd signatures such particles with long lifetimes (LLPs) that decay in the middle of the detector are of great interest. These searches, however, often do not use modern techniques. The goal is to create a Graph Neural Network to reconstruct LLPs, dramatically improve LLP reconstruction efficiency and use this tool to discover new physics in Run 3.
COLLIDER/ATLAS - Higgs Hunting at the Energy Frontier
Supervisor: Dr Gabriel Facini
The most energetic collisions created by the LHC are the most challenging to work with but offer the greatest potential for discovery. A Higgs boson is rarely produced and is difficult to identify with the ATLAS detector. The challenge of measuring Higgs bosons created with over 1 TeV of transverse momentum and decaying to two b-quarks is to be met with modern ML techniques i.e. Graph Neural Networks. New physics and/or a legacy measurement that will inform the choices of future colliders await.
COLLIDER/ATLAS - Revolutionising tracking and b-jet identification to explore new regions and understand the fundamental workings of the Universe
Supervisor: Dr Tim Scanlon
The next four years is going to be an exceptionally exciting time to work on the ATLAS experiment at the LHC, with data taking starting again in 2022 and the LHC running for four years. We will be operating at a higher centre of mass and will at least double the dataset we will collect. This project will focus on further boosting our sensitivity to new physics by revolutionising both the tracking and flavour tagging algorithms using deep learning and graph neural networks, before applying these enhanced algorithms to search for new physics using the Higgs boson. By using bleeding-edge machine learning techniques to improve the key algorithms used by almost all analyses at ATLAS, this work will improve the entire physics output of ATLAS and beyond.
COLLIDER/ATLAS - Optimising jet tagging in ATLAS
Supervisor: Prof Mario Campanelli
Hadronic jets are collimated sprays of particles produced by the hadronisation of quarks or gluons. These particles can be produced either directly in the collision, or as a result of a hadronic decay of heavier states like the W, Z or Higgs bosons, or top quarks. This project consists on the optimisation of tagging algorithms to search for the production of W and Z pairs in their hadronic decays, in conjunction with a measured proton in the forward direction.
QUANTUM COMPUTING/ATLAS - Quantum machine learning to reconstruct tracks at the LHC
Supervisor: Dr Sarah Malik
The project will employ machine learning techniques such as deep learning and graph neural nets on a quantum computer, in addition to devising a quantum algorithm to efficiently reconstruct trajectories of charged particles from detector hits in the ATLAS detector at the LHC.
NEUTRINOS/SND@LHC - Event reconstruction with SND@LHC
Supervisor: Prof Mario Campanelli
Various machine learning techniques will be usesd to reconstruct tracks and vertices in the emulsion and electronics detectors of the SND@LHC experiment at CERN, including participating in data analysis.
NEUTRINO/PUEO - Bringing Antarctica neutrino searches into the digital age
Supervisor: Prof Ryan Nichol
Searching for the highest energy particles in the Universe using digital beam forming technology in Antarctica.
NEUTRINO/LEGEND - Towards the discovery of matter creation with the LEGEND project
Supervisor: Dr Matteo Agostini
LEGEND is an experiment designed to search for neutrinoless double beta decay with an unprecedented discovery potential. The observation of this matter creating process will have the most profound implications for particle physics and cosmology, and will shed light on the mechanism behind the dominance of matter over anti-matter in the universe. The experiment’s first phase (LEGEND-200) is about to start and the following stage, with increased discovery power, (LEGEND-1000) is already under preparation. The PhD project will give the unique opportunity to develop state-of-the-art analysis techniques, explore the most advanced machine learning methods, and apply these tools to the data flow from a real experiment. The student will be involved in all data processing steps, including low-level digital signal processing and high-level deep learning, multivariate analysis, and statistical inference. The newly developed techniques will be applied to LEGEND data to search for the first evidence of matter creation in the laboratory.
QUANTUM SENSORS/NEUTRINO MASS - Determining the Neutrino Mass from Cyclotron Radiation Emission Spectroscopy
Supervisor: Prof Dave Waters
You will join a new project at UCL which is exploring a novel method for measuring the neutrino mass, of fundamental importance to particle physics. We use quantum technologies to detect the very faint microwave emissions from beta-decay electrons in a magnetic trap, thereby determining very precisely the electron energy. Advanced signal-processing and machine-learning techniques will need to be developed to extract complex signals from noisy antenna data, in order to achieve our physics goals.
DARK MATTER/LZ - Boosting the physics reach of Dark Matter search experiments using ML and data intensive science techniques (LZ)
Supervisor: Dr Jim Dobson
The LUX-ZEPLIN (LZ) experiment is the largest and most sensitive dark matter (DM) direct detection experiment ever deployed underground. The unprecedented scale and ultra-low background environment of LZ will herald a new era in direct detection. In this PhD project you will apply cutting edge ML and DIS techniques to extract the maximum useful information from the complete LZ dataset and, ultimately, to boost the physics reach of the experiment.
MUONS/g-2+Mu3e - Tracking for Muon Physics
Supervisor: Dr Gavin Hesketh
Precision muon experiments offer the strongest hint of new fundamental physics to date, with the possibility of a huge discovery in the near future. This project will bring machine-learning to the particle tracking algorithms on Muon g-2 and on the upcoming Mu3e experiment, with the potential to significantly increase the sensitivity to physics beyond the Standard Model.
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.
EARTH SCIENCE/EARTHQUAKES - Deep learning for deep earthquakes
Supervisor: Prof Ana M G Ferreira
The occurrence of deep earthquakes in the 70-700 km depth range inside the Earth is one of the greatest outstanding questions in Geosciences. While deep earthquakes provide unique “in situ” information about the Earth’s interior, their detection is challenging due to their large distance to seismic stations at the surface. This project addresses these issues by applying new machine learning tools to greatly expand deep earthquake detection and characterisation with unprecedented robustness.
ASTRO/COSMOLOGY+EARTH SCIENCE - Understanding earthquakes and cosmic structure growth
Supervisor: Prof Benjamin Joachimi
This project will develop surrogate modelling and inference techniques with applications to current problems in both seismology and cosmology. In areas otherwise hampered by complex, computationally expensive models, we will enable the fast and accurate localisation of earthquakes, as well as constraints on the cosmic large-scale structure from current large galaxy surveys.
ASTRO/COSMOLOGY - Geometric deep learning on the celestial sphere for cosmology and beyond
Supervisor: Prof Jason McEwen
The current evolution of our Universe is dominated by dark energy and dark matter but yet an understanding of the dark Universe remains critically lacking. Sensitive statistical and deep learning techniques are required to extract cosmological information from weak observational signatures. Foundations of geometric deep learning will be developed and applied to cosmological data from Euclid and the Rubin Observatory. Further details here: http://www.jasonmcewen.org/opportunities/#phd
ASTRO/COSMOLOGY - Explainable AI to explore galaxy images
Supervisor: Prof Ofer Lahav
Applications of Explainable AI methods to galaxy images from Galaxy Zoo, DES, KiDS, Rubin-LSST, Euclid and simulations.
ASTRO/EXO-PLANETS - Hiding in Gaia; dark planets and companions orbiting white dwarfs
Supervisor: Prof Jay Farihi
The project aims to use the Gaia full DR3 released in 2022 to build a set of likely planetary and stellar companions to white dwarfs stars. These will potentially include the first bona fide exoplanets orbiting white dwarfs as well as the elusive progenitors to type Ia supernovae and related explosive transients. The major challenge will be building a confidence metric based on the expected behavior for single stars within this very large dataset.
ASTRO/EXO-PLANETS - Using deep learning to model complex chemistries of exoplanet atmospheres
Supervisor: Dr Ingo Waldmann
Characterising exoplanet atmospheres allows us to understand their habitability, formation and evolution histories. Their complex chemical modelling is essential yet currently intractable. This project focuses on improving the precision and speed of complex chemical models using Graph Neural Networks and ExplainableAI. The development of such AI solutions has the potential to revolutionise the state-of-the-art and unlock our understanding of these foreign worlds.
ASTRO/PLANETARY - Using Machine Learning for space surveillance using all-sky cameras
Supervisor: Dr Anasuya Aruliah
The number of artificial objects in orbit grows exponentially. There is an urgent need for more accurate tracking systems to minimise collisions that endanger astronauts and lead to loss of critical space services. Radar space surveillance is expensive, resulting in few radar stations globally. All-Sky Cameras are cheap, and available at many observatories to monitor weather. The investigation will apply ML to ASC images as a cheap space surveillance network to augment the existing database.
ASTRO/INSTRUMENTATAION - Genetic algorithm and design evolution of Metamaterial optics for Astrophysics instrumentation and Satellite communication
Supervisor: Prof Giorgio Savini
This projects aims to employ Machine Learning and Evolution algorithm to efficiently evolve the design of antenna receivers for satellite communication and astrophysical instrumentation. The design of optical 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 freedoms involved in these design are too numerous and require DI techniques for optimization.