PhD projects in astrophysics for our STFC studentships 2024 are listed below.

- Probabilistic deep learning for cosmology and beyond (Prof. Jason McEwen)
- Understanding exoplanets and their host stars using large data sets and AI (Dr. Vincent Van Eylen)
- Studying the Universe with neutrinos (Prof. Kinwah Wu)
- Multi-Messenger Astrophysical Modelling of Compact Object Systems (Dr. Ziri Younsi)
- Simulations of polarized emission from black hole binaries (Prof. Silvia Zane)

### Probabilistic deep learning for cosmology and beyond

#### Primary Supervisor: Prof. Jason McEwen

The current evolution of our Universe is dominated by the influence of dark energy and dark matter, which constitute 95% of its content. However, an understanding of the fundamental physics underlying the dark Universe remains critically lacking. Forthcoming experiments have the potential to revolutionalise our understanding of the dark Universe. Both the ESA Euclid satellite and the Rubin Observatory Legacy Survey of Space and Time (LSST) will come online imminently, with Euclid successfully launched in July 2023 and the Rubin LSST Observatory having recently achieved first light. Furthermore, the Simons Observatory is in advanced stages of construction. Sensitive statistical and deep learning techniques are required to extract cosmological information from weak observational signatures of dark energy and dark matter.

The classical approach of deep learning is to make single predictions. A single estimate of a quantity of interest, such as an image, is typically made. For robust scientific studies, however, single estimates are not sufficient and a principled statistical assessment is critical in order to quantify uncertainties. Bayesian inference provides a principled statistical framework in which to perform scientific analyses. In cosmology, in particular, Bayesian inference is the bedrock of most cosmological analyses. While such approaches provide a complete statistical interpretation of observations, which is critical for robust and principled scientific studies, they are typically computationally slow, in many cases prohibitively so. Furthermore, in such analyses prior information typically cannot be injected by a deep data-driven approach.

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. This deep hybrid approach, where statistical and deep learning components are tightly coupled in integrated approaches, rather than considered as add-ons, will allow us to realise the complementary strengths of these different approaches simultaneously. For some examples of related research, please see the following recent papers: McEwen et al. 2021, arXiv:2111.12720; Spurio Mancini et al. 2022, arXiv:2207.04037; Polanska et al. 2023, arXiv:2307.00048 (although in this PhD project we will go beyond the Bayesian model comparison focus of these works).

Specifically, in this project we will develop novel probabilistic deep learning models, variational inference techniques and simulation-based inference approaches. These new methodologies will be applied to various cosmological problems and probes, focusing on the cosmic microwave background and weak gravitational lensing, and will include generative models for emulation and inference approaches for the estimation of not only the parameters of cosmological models but also to assess the most effective models and physical theories for describing our Universe.

The student should have a strong mathematical background and be proficient in coding, particularly in Python. The student will gain extensive expertise during the project in deep learning, going far beyond the straightforward application of existing deep learning techniques, instead focusing on the construction on novel probabilistic deep learning approaches and their application to novel problems in cosmology and beyond. The expertise gained in foundational deep learning will prepare the student well for a future career either in academia or industry. In particular, the emerging field of probabilistic deep learning is a speciality highly sought after in industry by many companies, such as Google/DeepMind, Facebook, Amazon and many others.

#### Desired Knowledge and Skills

- Undergraduate in physics, mathematics, computer science or statistics (or related field) Strong computational skills
- Strong mathematical skills
- Some background in astrophysics is desirable but not essential
- Some background in machine learning is desirable but not essential

### Understanding exoplanets and their host stars using large data sets and AI

#### Primary 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 – now known as exoplanets – from the realm of science-fiction into that of science. An important milestone was the discovery of several planets orbiting a pulsar (Wolszczan & Frail, 1992), followed by the first planet orbiting a star more similar to our Sun (Mayor & Queloz, 1995), an achievement awarded the 2019 Nobel Prize in Physics. The 25+ years since have been filled with an abundance of exciting discoveries and today we know over 5000 exoplanets. These planets exhibit an incredible diversity of properties. Why do so many planets have tiny orbits – often much smaller than that of Mercury? What causes planets to become rocky, gaseous, or something in between? Why do some planets have orbits that are strongly eccentric, or misaligned with the rotation of their host stars? What happens to planets when stars evolve away from the main sequence? Which planets are the most favourable and interesting targets for studies of their atmospheres? How unique is our solar system – are we alone?

Exoplanet science is a young field of research and there is great potential for many ground-breaking new discoveries. A PhD project is available that will seek to link the discovery of thousands of exoplanets to planet formation models. In particular, a key ingredient to understand the properties of exoplanet systems is the stars around which they orbit. In this project, we will attempt to link stellar properties, such as the planet birth environment and the stellar age, to the architecture of planetary systems. To do so we will use state-of-the-art statistical techniques, ranging from Bayesian modelling to machine learning and AI. The successful student will have the opportunity to shape the direction of the project.

During this project a motivated student will sharpen their analytical background and physical knowledge, while developing strong data science skills that will be valuable both in an academic career and outside of academia. Furthermore, there will be ample opportunity to travel to other universities and present new findings in international conferences, as well as the potential to conduct novel observations at telescopes around the world. The successful applicant will join a vibrant and diverse research team (see www.vincentvaneylen.com for details on ongoing research projects), with ample opportunity to develop communication and other soft skills.

#### Desired Knowledge and Skills

- Undergraduate in astrophysics, planetary science, computer science, or related degree.
- A background in physics and/or data science is helpful but lack thereof can be overcome with strong motivation.
- Alternatively, a motivated student with a strong background in computer science or data science rather than astrophysics will also be considered.
- Excellent writing and presentation skills are a bonus, as is evidence of motivation, leadership and creativity.

### Studying the Universe with neutrinos

#### Primary Supervisor: Kinwah Wu

With the discovery of ultra-high-energy neutrinos from the blazar TXS 0506+056 and the star-forming active galaxy NGC1068 by IceCube, we may now study high-energy astrophysical sources with neutrinos as we use photons. Recently, IceCube released a neutrino image (with resolution of about 5-10 degree scale) of the Galactic Plane. This piece of work has shifted the paradigm of astronomical research, and neutrinos become a new means for imaging the Universe. Here in UCL, we work on high-energy particle astrophysics, and one ongoing project is neutrino imaging of the region around the Galactic Centre, at scales down to about 0.1 degree, using the next generation large neutrino telescopes, P-ONE (a large undersea neutrino observatory in the Northern hemisphere, in construction, with capability similar to IceCube). Neutrinos are particularly useful to probe fundamental physics and astrophysics processes in violent and dense environments, because they are practically not absorbed by media along the line-of-sight. In addition to Galactic Centre neutrino imaging astrophysics, we are also working on projects of the following high-energy neutrino sources: (i) the inner core region of active galactic nuclei (AGN), where the black hole interfaces with accretion flows and jets, (ii) jet-star and jet-cloud interactions inside AGN jets, (iii) particle acceleration and hadronic interactions, when stars are tidally disrupted upon encountering a black hole, and (iv) high-energy particle physics in gravitational wave sources such as mergers of neutron-star and stellar-mass black hole, and (v) gamma-ray bursters and fast-radio busters. The student will choose one of the above systems as the prime thesis research, depending on the student’s interest and technical background. It will be a theoretical and modelling project, with the objective of providing the physical inputs for the next generation neutrino observatories, such as P-ONE and IceCube Gen 2, and understanding the fundamental physics and astrophysics that operate in the neutrino emitting systems. The student will be guided to build scenarios and construct phenomenological models, in addition to conduct theoretical calculations. The student will also learn the experimental aspects of neutrino “observations”. UCL (MSSL/P&A[HEP]) is the lead UK institution in P-ONE, among other core institutions in Germany, Canada and USA.

#### Desired Knowledge and Skills

- undergraduate or MSc in physics or astrophysics. (Required)
- a solid grounding in electromagnetism, classical mechanics and quantum physics. (Required)
- comfortable with analytical calculation and computation. (Required)
- open-minded and resourceful. (Required)
- knowledge in astronomy is desirable. (Desirable)
- numerical computation experience. (Desirable)

### Multi-Messenger Astrophysical Modelling of Compact Object Systems

#### Primary Supervisor: Dr Ziri Younsi

Since the first detection of gravitational waves (GWs) from the black hole (BH) binary GW150914 by the LIGO and Virgo collaborations, a new window has been opened onto the Universe. In addition to being able to ‘see’ astrophysical phenomena using electromagnetic (EM) radiation, i.e., photons, we may now ‘listen’ for hitherto undetectable events via GWs. Over 90 GW events in the stellar-mass-range have been detected to-date, from binaries comprising BHs and neutron stars (NSs). These detections provide a unique opportunity to test gravity in the highly nonlinear and dynamical regime, providing new insights into the properties of matter and radiation under extreme physical conditions. LISA, a future spaced-based GW experiment, will transform our view of the GW Universe by detecting a plethora of new GW events, including binary supermassive BHs (SMBHs). Three years after the birth of GW astronomy, the Event Horizon Telescope (EHT) Collaboration published the first ever images of solitary SMBHs, first in M87 and later the Milky Way. These images reveal a bright emission ring enclosing the purported event horizon of the SMBH, providing new constraints on BH mass and spin, the mechanisms powering accretion and relativistic jet collimation, and even enable new tests of gravity. Modelling EM and GW emissions from compact object systems requires solving the equations of magnetohydrodynamics (MHD) for relativistic plasmas in strong gravity. In particular, the counterpart detection and characterisation of GW sources requires accurate modelling of their EM signatures. Multi-frequency EM observations from compact object systems are essential to understand their fundamental properties and governing physical processes.

The student will begin by learning to perform MHD simulations of BH accretion, calculating the multi-frequency EM emissions from the vicinity of the event horizon using covariant radiative transfer. Such calculations may be leveraged to investigate the following research directions: flaring events in the Galactic Centre, particle production and acceleration processes around BHs, and strong-field tests of gravity. The student will select one of these three directions as the focus for the first part of the project. The second part of the project will model the emission (EM and GW) from merging compact objects. The student will learn to use a numerical relativity (NR) code to simulate merging BHs (and NSs, depending on progress). The student will interface the results from their NR simulations with covariant radiative transfer, calculating contemporaneous EM and GW signal models. These models will provide new observable signatures which will help guide future LISA and EHT observations. Throughout the project the student will learn experimental techniques underpinning GW detections and Very-Long-Baseline-Interferometry (VLBI) imaging of BHs. UCL is a core member of both the LISA Consortium and the EHT Collaboration.

#### Desired Knowledge and Skills

- Undergraduate or MSc in physics or astrophysics. (Required)
- firm foundation in general relativity, radiative transfer and magnetohydrodynamics. (Required)
- knowledge in astronomy, numerical relativity and interferometric imaging. (Desirable)
- computational code deployment on shared and distributed memory architectures. (Desirable)
- strong background (and demonstrable prior experience) in analytic and numerical computational approaches. (Required)
- creativity and determination. (Required)

### Simulations of polarized emission from black hole binaries

#### Primary Supervisor: Prof. Silvia Zane

Electromagnetic and multi-messenger emission from magnetized neutron stars and black hole binaries carry a huge amount of information on the properties of the source. The launch of the NASA IXPE mission in Dec. 2021 has opened a new window of opportunity to study this problem, adding for the first time X-ray polarimetric information to spectral and timing data. Specifically, the signal observed from accretion disks around black holes is strongly affected by the properties of the matter orbiting near the black hole horizon and its study may reveal exquisite constraints on the black hole spin, one of the major open issues in compact objects astrophysics. The interpretation of magnetar’s X-ray emission, instead, can provide the only method to test phase transitions of plasma in strong magnetic fields and still unproved QED effects as the birifrengency of the magnetized vacuum. In parallel, the discovery of gravitational waves from merging compact objects has given now a new torch on discovering the status of the poorly understood interior of the neutron star, where matter reaches supranuclear densities, and which description is still unknown.

Goal of this project is to simulate the X-ray emission from strongly magnetized neutron stars (aka magnetars or XDINS) or black hole binaries in the soft state, producing numerical models of atmosphere, magnetosphere and, for the binary case, of the accretion disks. The latter will include the effects of scattering and absorption on the emitted and reflected radiation, through the modelling of the disk ionization structure and surface albedo profile. To this aim, the student will use and modify existing numerical codes based on ray-tracing, finite difference and Monte Carlo methods, together with the public photo-ionization code CLOUDY. The final goal is to make state-of-the art simulations of the IXPE detections for neutron stars and black holes, and produce detectability predictions to be used for driving the design of other future proposals for X-ray polarimetric observatories. It is expected that the student will interact closely with international teams.

Using existing codes is also possible to simulate the density stratification in the neutron star crust and envelope. This be largely affected by strong magnetic stresses that may result in a non axisymmetric distribution of matter. Would this be the case, then the neutron star will behave as a rotating body with a substantial non zero quadrupole mass moment, that may result in the emission of detectable gravitational waves (GW). This scenario can be investigated by making use of numerical methods to infer the density distribution of the star, with the goal to make prediction for the signal detectability with the LIGO/KAGRA/ET/LISA GW experiments and/or the space gamma-ray observatories.

#### Desired Knowledge and Skills

- Have solid coding experience (essential).
- Knowledge in neutron stars, black hole X-ray binaries, accretion disks, radiative transfer and general high energy astrophysics or multimessengers.
- Creative and critical thinking.
- Strong undergraduate training in physics/astrophysics and in mathematical methods.
- Comfortable with both analytical and heavy numerical calculations.