Final year PhD student at the Institute of Astronomy, University of Cambridge.
Cosmologist and astrophysicist developing simulation-driven Bayesian frameworks to constrain cosmology with the Lyman-α forest.
Research
Projected dark matter density distribution on a slice through Sherwood hydrodynamical simulation of size 20 h−1 Mpc.
My main (PhD) work has involved using numerical simulations together with Lyman-alpha forest data to test cosmological models. I would like to push this further along three directions:
Theory: Constraining the dark sector: dark matter models across a broad mass range, from light dark matter candidates through the ultra-light regime, exploring the implications for the early Universe and inflation. Dark energy: early or late-time modifications that modify the expansion history or growth of structure. Or interacting dark matter and dark energy — can these elucidate the fate of our Universe? I am very interested in model-building to incorporate the effect of these and other beyond ΛCDM models into (1) Boltzmann solvers and (2) numerical simulations.
Neural network architecture for emulator used to constrain Cold+Warm Dark Matter.Data & Statistics: Future high-resolution Lyman-α forest data will be (available soon) with improved statistics that will allow tighter constraints to be set. Moving beyond 1d flux power spectrum (P1D) analyses will allow to break parameter degeneracies (between e.g. cosmology and astrophysical parameters), for which robust higher-order estimators need to be developed. Beyond summary statistics, I am particularly interested in constructing forward models for field-level inference, connecting the underlying matter density field to the observed transmitted flux.
Emulation: More datasets and more models will require a robust tool that leverages expensive hydrodynamical simulations and efficiently marginalizes over astrophysical uncertainties. I am working on enhancing the emulator for P1D studies to support automatic hyperparameter optimization across simulation suites, thereby improving accuracy while substantially accelerating cosmological inference.
