Eisenstein, Daniel JMaksimova, Nina2022-06-0720222022-05-122022-05Maksimova, Nina. 2022. AbacusSummit: Next Generation N-Body Cosmology. Doctoral dissertation, Harvard University Graduate School of Arts and Sciences.29209575https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37372178Cosmological N-body simulations have long been the standard forward modelling tool for interpreting measurements of the large-scale structure (LSS) of the Universe. As Stage IV LSS surveys such as the Dark Energy Spectroscopic Instrument (DESI) map unprecedentedly ambitious volumes of space, so too N-body codes must evolve to deliver larger, more finely resolved simulations. In this dissertation, we meet this need using Abacus, an exceptionally fast and accurate cosmological N-body code that can solve a supercomputer-sized N-body problem on a single commodity workstation. We present a new, massively parallelised version of Abacus, including a custom on-the-fly halo finder, and deploy it on Oak Ridge National Laboratory's Summit supercomputer to create AbacusSummit: the largest suite of high-accuracy N-body simulations produced to date. At nearly 2 PB of data volume, AbacusSummit not only meets and exceeds DESI's N-body simulation requirements but enables a broad range of cosmological applications as well. We describe the suite's design, data products, performance optimisation, and production, then present one of its example use cases, in which we measure the baryon acoustic oscillation scale shift in mock galaxy catalogues generated from AbacusSummit halos. Finally, we seek to improve on Abacus' current time-stepping method to enable yet-finer mass resolution simulations. Abacus and AbacusSummit will be integral parts of interpreting galaxy survey results in the generation of Stage IV and beyond.application/pdfencosmologydark energydark matterhigh-performance computinglarge-scale structureN-body simulationsAstrophysicsAstronomyPhysicsThesis or Dissertation2022-06-070000-0003-4038-8313