We train neural networks to perform likelihood-free inference from
$(25\,h^-1Mpc)^2$ 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 ($100\,h^-1kpc$) while
performing a robust marginalization over baryonic physics at the field level:
the model can infer the value of $Ømega_m (4\%)$ and $\sigma_8 (\pm
2.5\%)$ from simulations completely different to the ones used to train it.
Description
Robust marginalization of baryonic effects for cosmological inference at the field level
cite arxiv:2109.10360Comment: 7 pages, 4 figures. Second paper of a series of four. The 2D maps, codes, and network weights used in this paper are publicly available at https://camels-multifield-dataset.readthedocs.io
%0 Generic
%1 villaescusanavarro2021robust
%A Villaescusa-Navarro, Francisco
%A Genel, Shy
%A Angles-Alcazar, Daniel
%A Spergel, David N.
%A Li, Yin
%A Wandelt, Benjamin
%A Thiele, Leander
%A Nicola, Andrina
%A Matilla, Jose Manuel Zorrilla
%A Shao, Helen
%A Hassan, Sultan
%A Narayanan, Desika
%A Dave, Romeel
%A Vogelsberger, Mark
%D 2021
%K library
%T Robust marginalization of baryonic effects for cosmological inference at
the field level
%U http://arxiv.org/abs/2109.10360
%X We train neural networks to perform likelihood-free inference from
$(25\,h^-1Mpc)^2$ 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 ($100\,h^-1kpc$) while
performing a robust marginalization over baryonic physics at the field level:
the model can infer the value of $Ømega_m (4\%)$ and $\sigma_8 (\pm
2.5\%)$ from simulations completely different to the ones used to train it.
@misc{villaescusanavarro2021robust,
abstract = {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 $\Omega_{\rm m} (\pm 4\%)$ and $\sigma_8 (\pm
2.5\%)$ from simulations completely different to the ones used to train it.},
added-at = {2021-09-23T12:12:18.000+0200},
author = {Villaescusa-Navarro, Francisco and Genel, Shy and Angles-Alcazar, Daniel and Spergel, David N. and Li, Yin and Wandelt, Benjamin and Thiele, Leander and Nicola, Andrina and Matilla, Jose Manuel Zorrilla and Shao, Helen and Hassan, Sultan and Narayanan, Desika and Dave, Romeel and Vogelsberger, Mark},
biburl = {https://www.bibsonomy.org/bibtex/2e004184e9670fe064b42daa6341d3cfa/gpkulkarni},
description = {Robust marginalization of baryonic effects for cosmological inference at the field level},
interhash = {dfce6206cb6482ae3c70bb01036825d5},
intrahash = {e004184e9670fe064b42daa6341d3cfa},
keywords = {library},
note = {cite arxiv:2109.10360Comment: 7 pages, 4 figures. Second paper of a series of four. The 2D maps, codes, and network weights used in this paper are publicly available at https://camels-multifield-dataset.readthedocs.io},
timestamp = {2021-09-23T12:12:18.000+0200},
title = {Robust marginalization of baryonic effects for cosmological inference at
the field level},
url = {http://arxiv.org/abs/2109.10360},
year = 2021
}