Computational Cosmology and Artificial Intelligence

点击翻译按钮获取中文摘要

Over the past few decades, the standard model of cosmology has had tremendous success at explaining a vast array of observations spanning an immense range of scales both in space and in time. However, the origin and nature of all 3 key components of this model remain, to this day, unknown. The physical nature of the field(s) responsible for the period of primordial inflation in the early Universe, the source of the apparent accelerated expansion of the Universe (dark energy), and the particle(s) that make up dark matter are the biggest outstanding mysteries of modern cosmology. Their understanding constitutes the main goal of modern cosmology, and will most likely cause a revolution in fundamental physics. In the next decade, a large number of new observatories and experiments will attempt to shed light on these mind-boggling questions. However, while the data cosmologists expect will be of unprecedented quality, the volumes of these new data will pose serious challenges when it comes to their analysis through traditional statistical methods. Artificial intelligence and machine learning will offer alternative analysis methods that promise great increase not only in speed, but also, in some cases, in accuracy.The goal of the proposed Tier 2 Canada Research Chair program is to lead cutting-edge observational and theoretical research on the use of machine learning for the simulation and analysis of cosmological data with the aim of enabling rapid, precise, and accurate analysis of upcoming observations from these large sky surveys, paving new ways for major breakthrough discoveries in the field of astrophysics.More specifically, the research program will focus on two central questions:1. Measuring the Hubble constant with strong gravitational lensing and machine learning, with the hope of solving an emerging crisis in cosmology and potentially uncovering new physics.2. Reconstructing the initial conditions of the Universe with survey data and artificial intelligence, creating a map of the initial 3-dimensional Universe only 400 000 years after its birth. This will open new avenues for answering questions about the nature of inflation and probing fundamental properties of dark energy.