Title: Big Data in Cosmology
Abstract: Over recent decades cosmology has transitioned from a data-poor to a data-rich field, which has lead to dramatic improvements in our understanding of the cosmic evolution of our Universe. Nevertheless, we remain ignorant of many aspects of the scenario that has been revealed. Little is known about the fundamental physics of structure formation in the early Universe or the formation of the first large-scale structure during the epoch of reionization. A complete understanding of dark energy and dark matter, which dominate the late evolution of our Universe, also remains elusive. In coming decades the field will transition from being data-rich to being overwhelmed by data as next-generation observational facilities come online.
The Large Synoptic Survey Telescope (LSST), currently under construction in Chile and due to achieve first light in 2019, will provide a paradigm shift for cosmological surveys. Previous telescopes have taken a static survey over the lifetime of the experiment. LSST will survey the entire night sky every few days, resulting — for the first time — in a dynamic movie of the celestial sky. LSST is expected to generate 10 million alerts each night, with final data volumes expected to reach the exabyte scale. The Square Kilometre Array (SKA) is an ambitious next-generation radio telescope that will be spread across two continents — Southern Africa and Western Australia — with construction expected to begin in 2018 and first light in 2020. The data rate of the SKA is anticipated to be comparable to world-wide internet traffic today.
The emerging big-data era of cosmology has the potential to lead to another dramatic improvement in our understanding, addressing unanswered fundamental questions about the content and evolution of our Universe – provided that we can make sense of the overwhelming data-sets that will be acquired.
I will review a variety of data intensive science problems in cosmology and the development of new analysis techniques and methodologies to address these problems. In particular, I will discuss the construction and application of analysis techniques defined on spherical manifolds in order to analyse cosmological observations made on the celestial sphere, the prevalence of Bayesian inference in cosmology, the use of compressive sensing techniques to image raw data acquired by radio interferometric telescopes, and the well-established but rapidly growing application of machine learning in cosmology.