Next-Generation Measurements of Dark Energy: Leveraging Photometric Redshifts for Type Ia Supernova Cosmology

Abstract

Since their use in the ground-breaking discovery of dark energy, the cause of the accelerated expansion of the universe, Type Ia Supernovae (SNe Ia) have remained one of our most powerful tools to measure cosmic distances and constrain the expansion history of our universe. While modern cosmological SNIa analyses have relied on spectroscopy to obtain redshifts, our spectroscopic resources will soon be vastly outpaced by the imminent Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST) and Roman Space Telescope High-Latitude Time Domain Survey (Roman HLTDS). These upcoming datasets have the potential to revolutionize our understanding of dark energy if we are able to address the limitations imposed by requiring spectroscopy. This dissertation presents studies, primarily with our current best data from the Dark Energy Survey (DES), which show that photometric SNIa cosmology using photometric redshifts (photo-z) is a promising avenue for next-generation constraints. First, I present a study of cosmology with DES SNe hosted in redMaGiC galaxies, a selection of Luminous Red Galaxies (LRGs). This approach neatly and simultaneously addresses three major challenges in the SNIa field today: i) obtaining reliable redshifts for every SN, ii) accurately classifying the type of supernovae, and iii) accounting for correlations between supernovae and their host galaxies. Primarily, it is easier to obtain more precise photo-z for LRGs than for the average galaxy. LRGs are also expected to contain low rates of core-collapse SNe, and by focusing on a specific subpopulation of galaxies, we reduce concerns related to host galaxy and SN property correlations. Using redMaGiC photo-z, we find a shift on w, the dark energy equation-of-state parameter, of 0.005 relative to w measured with spectroscopic redshifts (spec-z).Next, I investigate the potential for using average galaxy photo-z's, considering the impact of several different photo-z algorithms. With the DES 5-year photometrically classified SN sample, we evaluate the biases that arise when using redshifts inferred from SN light-curves with host galaxy photo-z priors. For three of the five variants tested, we obtain biases in w of less than 0.02, still subdominant to the overall statistical uncertainty in w. We develop detailed catalog-level simulations to validate the analysis and call attention to remaining open questions to be addressed ahead of SN cosmology analyses with LSST. For the first time, I also evaluate the interplay of photometric classification with photo-z and find that it is a subdominant systematic.Finally, I turn to work in preparation for the Roman Space Telescope HLTDS and describe our characterization of the redshift recovery rate of the Roman slitless grism for SN host galaxies using images of 2D simulated spectra. We provide an approximate picture of the redshifts that will be available for Roman SN cosmology from both the grism, as well as the Roman prism and external ground-based telescopes. We evaluate the size of potential systematics related to the modeling of the spectroscopic redshift efficiency and illustrate the impact that redshift assumptions have on optimizing the HLTDS survey strategy.