HEP Colloquia 2024

by Tommaso Ronconi (Sissa)

Thursday 17 Oct 2024, 14:30 → 16:00 Europe/Rome

Sala Consiliare (Department of Physics)

How to build a radio sky: empirical full-sky simulations of radio continuum and HI galaxies with realistic clustering properties for future radio surveys 

The advent of new observational campaigns and upcoming radio surveys marks a transformative era in radio astrophysics and cosmology. As we anticipate first light from the Square Kilometre Array (SKA) Observatory, there is a critical need to prepare for the cosmological interpretation of wide-field data. Mock datasets, which simulate expected observations, play a crucial role in this preparation. Empirical methods provide a flexible approach to investigating the galaxy-halo connection without imposing restrictive assumptions on poorly understood baryonic physics.

 

In this talk, I will detail the algorithm we use to build a mock radio sky, simulating both continuum emission from star-forming galaxies and active galactic nuclei, as well as HI galaxy line emission. The coordinates of these objects are generated by matching the observed two-point correlation function using a dark matter-only full-sky lightcone simulation that spans redshifts from 0 to 8. Our catalogues preserve key observational relations that underpin our models, including the predicted abundances of continuum objects from 150 MHz to 15 GHz, the HI mass function, polarization properties, and clustering across all studied redshifts.

 

I will demonstrate how this method, by coupling three distinct populations of objects to the same simulated lightcone, identifies counterparts without the need to fine-tune models based on additional assumptions. This approach offers flexibility and robustness in the face of future discoveries.

 

The final outcome of this work will be a suite of mock catalogues, as well as open-source software for generating these catalogues. Since our method is computationally light, it can be run even on consumer-grade machines, offering users the ability to adapt and update the models as needed, especially as deeper observational data becomes available.

 

Lastly, I will discuss ongoing investigations into the use of reinforcement learning to improve galaxy assignment techniques beyond traditional sub-halo mass-based models. This work explores the role of secondary halo properties in the galaxy-halo connection and considers the impact of cross-correlations between observables.