My overarching goal is to infer the initial conditions of the local Universe – the primordial fluctuations from which all nearby structure grew – and to develop new field-level tests of galaxy formation and cosmology along the way.
For a full list of publications, see my CV or ADS library.
Digital twins and the initial conditions of the local Universe
Digital twins of the local Universe are simulations whose initial conditions are inferred to reproduce the specific structures – galaxies, clusters, voids, filaments – that we observe in our cosmic neighbourhood, rather than a random patch of the Universe. I am interested in exploiting these to learn about cosmology and galaxy formation: mapping large-scale flows, measuring the expansion rate to high precision, testing galaxy formation models object by object with semi-analytical modelling, and ultimately inferring the initial conditions and cosmological parameters at the field level.
The Velocity Field Olympics
The Manticore Project I: a digital twin of our cosmic neighbourhood
Revisiting the Great Attractor
Learning the Universe: Constrained simulations of the Coma galaxy cluster – I. Radial X-ray and Compton-y signatures
Distance-ladder cosmology
I develop new statistical frameworks for distance-ladder cosmology based on rigorous Bayesian forward modelling. A particular focus is exploiting relatively small samples of distance indicators, such as Cepheids, tip of the red giant branch (TRGB) stars, and masers, that are otherwise dominated by cosmic variance, using digital twins of the local Universe to suppress it and deliver precision measurements of the Hubble constant. Upcoming goals include delivering a forward model of the entire distance ladder – from Milky Way stars to distant supernovae – and developing a novel scalable framework for the LSST era.
Forward-modelling Milky Way Cepheids: selection effects and physical priors in the Gaia–HST calibration
1.8 per cent measurement of H₀ from Cepheids alone
The subtle statistics of the distance ladder: On the distance prior and selection effects
Forward-modelling the Tolman and distance-duality tests with IllustrisTNG
Peculiar velocities
Peculiar velocities – deviations from the smooth Hubble flow – can be inferred by comparing a galaxy’s observed redshift with an independent distance estimate, obtained from a scaling relation such as Tully–Fisher or the fundamental plane, or from supernova standardisation. I use these surveys to constrain cosmological parameters such as fσ₈, test the cosmological principle through searches for anisotropies in the local expansion rate, and ultimately to build the next generation of digital twins of the local Universe.
S₈ from peculiar velocities: agreement with Planck for Tully–Fisher and supernovae, tension for the fundamental plane
No evidence for local H₀ anisotropy from Tully–Fisher or supernova distances
Testing cosmic anisotropy with cluster scaling relations
Galaxy–halo connection
I am interested in all aspects of the galaxy–halo connection, from empirical to semi-analytic models. On the empirical side, I extend traditional models such as abundance matching and test them against diverse observational samples, including optical and HI-selected populations. I also use semi-analytic models in conjunction with digital twins to predict properties of nearby structures on an object-by-object basis.
The dependence of subhalo abundance matching on galaxy photometry and selection criteria
Learning the Universe with cosmological rescaling of merger trees and semi-analytic galaxy formation models
Introducing sapphire: Towards Hybrid Physics-Informed, Data-Driven Modeling of Galaxy Formation
Galaxy dynamics
I am interested in the empirical correlations governing the dynamics of disc galaxies, such as the radial acceleration relation (RAR) and the baryonic Tully–Fisher relation (BTFR). A particular focus has been establishing the RAR – the tight correlation between baryonic and total dynamical accelerations – as the fundamental organising principle of late-type galaxy dynamics, and exploring what this implies for galaxy formation and modified gravity.
On the fundamentality of the radial acceleration relation for late-type galaxy dynamics
Testing subhalo abundance matching with galaxy kinematics
Machine learning methods
I am interested in graph-based methods and geometric deep learning to capture physical structure in cosmological datasets, and simulation-based inference (SBI) to perform implicit likelihood inference where traditional likelihoods are intractable – including applications to JWST data to infer the ionising photon contributions of high-redshift galaxies. I commonly employ normalising flows, Gaussian processes, neural networks, and tree-based models to study, for example, the scatter in the galaxy–halo connection and to quantify uncertainties in astrophysical models. I also work extensively with Hamiltonian Monte Carlo and gradient-based samplers for scalable Bayesian inference.
The scatter in the galaxy–halo connection: a machine learning analysis
Inferring the ionizing photon contributions of high-redshift galaxies to reionization with JWST NIRCam photometry
CosmoBench: A Multiscale, Multiview, Multitask Cosmology Benchmark for Geometric Deep Learning
Gravitational-wave astronomy
During my master’s degree, I worked on gravitational-wave data analysis and strong-field lensing. This included testing the isotropy of binary black hole mergers with LIGO/Virgo data, exploring transdimensional parameter estimation, and studying frequency- and polarisation-dependent lensing in the gravitational spin Hall effect.