Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Explore how cutting-edge Bayesian field-level inference techniques are revolutionizing cosmology by creating physics-based reconstructions of the actual Universe we observe in this 20-minute conference lecture. Discover how decades of advances in observations, simulations, and statistical modeling have converged to enable the creation of a "Digital Twin of the Universe" that bridges theory and observation by anchoring simulations directly to measured large-scale structure data. Learn about a comprehensive framework that integrates galaxy survey data with high-fidelity simulations through Bayesian field-level inference models of cosmic evolution, enabling joint inference of cosmological parameters, initial density fields, and non-linear matter and velocity distributions while preserving the full information content of observational data. Examine how this approach overcomes the limitations of traditional analyses that rely on lossy, compressed summary statistics, instead providing detailed, object-by-object testbeds for ΛCDM and its extensions. Understand how recent advances in Bayesian field-level inference have demonstrated the ability to improve cosmological parameter constraints by up to a factor of five, opening new opportunities to test fundamental physics and probe dark matter and neutrino properties. Delve into the latest results from the Manticore project, which has produced the most comprehensive physical reconstructions of the Nearby Universe to date, including precise mass estimates for nearby clusters like Virgo and Coma, and physically inferred cosmic velocity fields that outperform state-of-the-art methods based on distance indicators. Investigate how this methodology enables detailed, causal reconstructions of complex non-linear systems such as the Milky Way–Andromeda pair, including their orbital dynamics and individual rotation curves, representing a paradigm shift in how we study cosmic structure formation and evolution.
