Which physical galaxy-formation model best describes the real Universe

Determine which physical model of galaxy formation and the halo–galaxy connection, such as hydrodynamical simulations or semi-analytic models, best describes the real Universe by assessing consistency with observational data and enabling machine learning methods that generalize across simulation frameworks.

Background

The paper compares hydrodynamical simulations and semi-analytic models (SAMs) as physically motivated approaches to model galaxy formation within dark matter halos, noting both methods produce predictions broadly consistent with observations. However, the authors emphasize that machine learning models trained on simulations must be robust across different frameworks because training data may not match the true Universe.

Within this context, the authors explicitly state that it is not yet known which physical model best represents our Universe. This unresolved question motivates the development of field-level inference methods that can generalize across diverse simulation suites and, ultimately, be applied to real observational data.