Robust marginalization of baryonic effects for cosmological inference at the field level
This work addresses the challenge of accounting for baryonic physics in cosmological analyses for astrophysicists, offering a method to improve parameter constraints, though it is incremental as it builds on existing simulation-based inference techniques.
The authors tackled the problem of cosmological inference from 2D mass maps by training neural networks on hydrodynamic simulations to perform likelihood-free inference, achieving robust marginalization over baryonic effects and inferring parameters like Ω_m (±4%) and σ_8 (±2.5%) on unseen simulations.
We train neural networks to perform likelihood-free inference from $(25\,h^{-1}{\rm Mpc})^2$ 2D maps containing the total mass surface density from thousands of hydrodynamic simulations of the CAMELS project. We show that the networks can extract information beyond one-point functions and power spectra from all resolved scales ($\gtrsim 100\,h^{-1}{\rm kpc}$) while performing a robust marginalization over baryonic physics at the field level: the model can infer the value of $Ω_{\rm m} (\pm 4\%)$ and $σ_8 (\pm 2.5\%)$ from simulations completely different to the ones used to train it.