TY - JOUR

T1 - Finding Universal Relations in Subhalo Properties with Artificial Intelligence

AU - Shao, Helen

AU - Villaescusa-Navarro, Francisco

AU - Genel, Shy

AU - Spergel, David N.

AU - Angles-Alcazar, Daniel

AU - Hernquist, Lars

AU - Dave, Romeel

AU - Narayanan, Desika

AU - Contardo, Gabriella

AU - Vogelsberger, Mark

N1 - Funding Information:
We thank Rachel Somerville for useful conversations. This work has made use of the Tiger cluster of Princeton University and the Iron and Popeye clusters at the Flatiron Institute, which is supported by the Simons Foundation. F.V.N. acknowledges funding from the WFIRST program through NNG26PJ30C, NNN12AA01C, and from NSF through grant No. AST-2108078. D.A.A. was supported in part by NSF grant Nos. AST-2009687 and AST-2108944. Details on the CAMELS simulations can be found in https://www.camel-simulations.org .
Publisher Copyright:
© 2022. The Author(s). Published by the American Astronomical Society.

PY - 2022/3/8

Y1 - 2022/3/8

N2 - We use a generic formalism designed to search for relations in high-dimensional spaces to determine if the total mass of a subhalo can be predicted from other internal properties such as velocity dispersion, radius, or star formation rate. We train neural networks using data from the Cosmology and Astrophysics with MachinE Learning Simulations project and show that the model can predict the total mass of a subhalo with high accuracy: more than 99% of the subhalos have a predicted mass within 0.2 dex of their true value. The networks exhibit surprising extrapolation properties, being able to accurately predict the total mass of any type of subhalo containing any kind of galaxy at any redshift from simulations with different cosmologies, astrophysics models, subgrid physics, volumes, and resolutions, indicating that the network may have found a universal relation. We then use different methods to find equations that approximate the relation found by the networks and derive new analytic expressions that predict the total mass of a subhalo from its radius, velocity dispersion, and maximum circular velocity. We show that in some regimes, the analytic expressions are more accurate than the neural networks. The relation found by the neural network and approximated by the analytic equation bear similarities to the virial theorem.

AB - We use a generic formalism designed to search for relations in high-dimensional spaces to determine if the total mass of a subhalo can be predicted from other internal properties such as velocity dispersion, radius, or star formation rate. We train neural networks using data from the Cosmology and Astrophysics with MachinE Learning Simulations project and show that the model can predict the total mass of a subhalo with high accuracy: more than 99% of the subhalos have a predicted mass within 0.2 dex of their true value. The networks exhibit surprising extrapolation properties, being able to accurately predict the total mass of any type of subhalo containing any kind of galaxy at any redshift from simulations with different cosmologies, astrophysics models, subgrid physics, volumes, and resolutions, indicating that the network may have found a universal relation. We then use different methods to find equations that approximate the relation found by the networks and derive new analytic expressions that predict the total mass of a subhalo from its radius, velocity dispersion, and maximum circular velocity. We show that in some regimes, the analytic expressions are more accurate than the neural networks. The relation found by the neural network and approximated by the analytic equation bear similarities to the virial theorem.

U2 - 10.3847/1538-4357/ac4d30

DO - 10.3847/1538-4357/ac4d30

M3 - Article

VL - 927

SP - 1

EP - 19

JO - Astrophysical Journal

JF - Astrophysical Journal

SN - 0004-637X

IS - 1

M1 - 85

ER -