Pass the Salt: Light-Curve Models for the Next Generation of Supernova Cosmology

Abstract

Nearly three decades ago, two teams of astronomers used a sample of just ~40 Type Ia supernovae (SNe Ia) to reveal a ground-breaking mystery: the accelerating expansion of the Universe. The quest to understand the underlying physics (dubbed dark energy) behind this accelerating expansion, which currently makes up ~70% of the energy density in our Universe, has captivated the field of modern cosmology ever since.

Since that initial discovery, dedicated surveys have observed thousands of SNe Ia, enabling precise measurements of the dark energy equation-of-state parameter w through their redshift-independent distance measurements. Recent results are consistent with w=-1 (indicating that dark energy is an inherent vacuum energy of space-time), but most are limited in precision by statistical and systematic uncertainties in roughly equal measure. However, the upcoming Dark Energy Survey 5-Year Supernova Analysis (DES-SN5YR) is set to analyse the largest ever single-survey SN sample (~1700 SNe) and will be statistics-limited thanks to a thorough curtailing of systematic errors throughout the analysis. This thesis contributes to those efforts by making a pivotal deep dive into the modelling that underlies all SN~Ia-based measurements of w.

Every modern dark energy analysis over the past decades has relied on the SALT (Spectral Adaptive Light curve Template) model framework to fit their SN Ia light curves, which is a critical step in estimating the distances required for cosmological modelling. The precision and accuracy of w is thus directly limited by that of SALT. In this thesis, I apply key updates to the inputs for the ubiquitous SALT2 model, providing the first publicly-available updated version of SALT2 since 2014 along with a measurement of its impact on SN cosmology (Delta w = 0.015 +/- 0.004). I develop a SALT2 training framework to estimate the end-to-end photometric calibration uncertainty in SN Ia cosmology measurements - which represents one of the largest sources of systematic uncertainty in current analyses - and implement this framework into the popular SNANA software package.

I use this training framework to produce a new version of SALT2, using a re-calibrated training sample according to the state-of-the-art "SuperCal-Fragilistic" calibration standard. I also use the framework to produce the SALT2 model suite needed to estimate the photometric calibration uncertainty underlying the Pantheon+ Supernova Analysis.

I provide the first independent validation of a new "SALT3" model, finding a negligible systematic uncertainty of Delta w = 0.001 +/- 0.005 in SN+CMB cosmology arising from the choice of SALT framework. Based on this result, I develop the SALT3 model and calibration suite used for the DES-SN5YR analysis.

Finally, I use the SALT3 framework to investigate dependencies of SALT models (and moreover, SN Ia cosmology) on the underlying training data, particularly focusing on understanding the effects of host galaxy mass and dust. This is a novel method of exploring the persistent ~0.06 mag "mass step" problem, demonstrating SALT's extended usefulness for investigating interesting astrophysical phenomena beyond cosmological distances.

This thesis emphasises SALT's place as a versatile tool for the foundational success of the next era of supernova cosmology, where the anticipated statistics of surveys such as the Vera C. Rubin Observatory's Legacy Survey of Space and Time (~1,000,000 SNe) will demand momentous reductions to the level of systematic uncertainties. If this is achieved, the next generation of SN cosmology analyses are set to measure w precisely enough to distinguish between physical explanations for the acceleration. Show more