Microlensed Supernovae as Probes of Astrophysics and Cosmology

Abstract

Gravitational lensing is poised on the brink of a data explosion from future large area surveys such as Euclid, the Vera Rubin Observatory (LSST), and the Roman Space Telescope. One of the most eagerly awaited discoveries from these observatories are gravitationally lensed supernovae (glSNe). The transient nature of supernovae (SNe) makes them an exciting prospect for cosmological studies such as time delay cosmography, and for astrophysical studies that probe the density profiles and stellar content of the lens galaxy. Gravitationally lensed Type Ia supernovae (glSNe Ia) are of particular interest due to their ‘standardizable’ candle nature, potentially allowing us to break lens modeling degeneracies or better constrain astrophysical parameters of the lens.
The small physical extent of supernovae photospheres means that they are unequivocally subject to microlensing, a stochastic gravitational lensing effect by the individual stars in the lens galaxy. While viewed as a source of noise in many studies, microlensing is an exquisite probe of the stellar content
in the lens galaxy, allowing us to infer parameters such as the stellar mass-to-light ratio Υ★ and typical stellar mass. While quasars have long been the only source of microlensing studies, that will change in the next decade; the estimated rates of glSNe guarantee that they will eventually outnumber quasars for microlensing studies.
As every galaxy-scale lensed supernova is subject to microlensing, it is vital to ensure that we have the proper tools to fully analyze future data in a timely manner while also taking advantage of all the microlensing information present, be it single epoch flux ratio anomalies or time-sampled caustic crossings. Incorporating microlensing into fully hierarchical Bayesian analyses has historically been a computationally challenging problem, but it is crucial for future analyses of large samples of glSNe data.
In this thesis, we first describe a series of improvements to computational microlensing that takes advantage of graphics processing unit (GPU) speedups. The usage of GPUs both speeds up numerical computations by orders of magnitude, while simultaneously allowing us to incorporate additional microlensing information into our analyses that are either novel or had previously been computationally intractable.
We then explore how such improvements can be used to better constrain cosmological and astrophysical parameters. By including in our analyses the presence or absence of caustic crossing events in the lightcurves of glSNe Ia, we demonstrate how the microlensing (de)magnification can be better constrained, thus allowing the standardizable candle nature of Type Ia supernovae (SNe Ia) to be used to break the mass-sheet degeneracy (MSD) and better constrain the Hubble constant.
If we instead assume a cosmology, microlensed supernovae allow us to measure Υ★ in the lens galaxy. We explore the level to which Υ★ can be constrained with single epoch data, finding that 50 well-modeled glSNe Ia can be used to constrain Υ★ within lens galaxies to within 15%, discriminating between an average Salpeter or Chabrier initial mass function (IMF). Time series
data of the microlensing lightcurves additionally allows us to constrain the typical stellar mass in the lens, providing information on the low mass end of the IMF.
Future glSNe will thus be essential for establishing independent measurements of the Hubble constant and testing the standard cosmological model, but they will also prove to be one of the best methods for constraining stellar mass in lensing galaxies, with minimal assumptions related to the underlying relationship between the stellar mass and the observed light profiles.

Date of Award	7 May 2026
Original language	English
Awarding Institution	University of Portsmouth
Supervisor	Thomas Collett (Supervisor), Coleman Krawczyk (Supervisor) & David Bacon (Supervisor)