Decades of advances in observations, simulations, and statistical modeling have converged to enable something unprecedented: a physics-based, data-anchored reconstruction of the actual Universe we observe, not just its statistical properties, but its specific structures across all relevant cosmic scales and times. This “Digital Twin of the Universe” bridges theory and observation by anchoring simulations directly to the measured large-scale structure, overcoming the limitations of traditional analyses that rely on lossy, compressed summary statistics. It serves as a detailed, object-by-object testbed for ΛCDM and its extensions, enabling us to reconstruct and analyze the complex, non-linear formation histories of cosmic structures.
Our framework integrates galaxy survey data and high-fidelity simulations through a Bayesian field-level inference model of cosmic evolution. By directly conditioning on real observations, we jointly infer cosmological parameters, initial density fields, and the non-linear matter and velocity distributions, seamlessly linking the early Universe to present-day structures while preserving the full information content of the data. Recent advances in Bayesian field-level inference have demonstrated its ability to improve cosmological parameter constraints by up to a factor of five, opening new opportunities to test fundamental physics, probe dark matter and neutrino properties, and unravel the origin and evolution of cosmic structure.
In this talk, I will illustrate a fully non-linear extension of this framework through the latest results from the Manticore project, which has produced the most comprehensive physical reconstructions of the Nearby Universe to date. These reconstructions provide precise, dynamically consistent mass estimates for nearby clusters such as Virgo and Coma, and yield physically inferred cosmic velocity fields that outperform state-of-the-art methods based on distance indicators. Furthermore, this approach enables detailed, causal reconstructions of complex non-linear systems like the Milky Way–Andromeda pair, including their orbital dynamics and individual rotation curves.