The main scientific goals of the CAMELS project are:

​

•  Provide theory predictions for statistics, or fields, as a function of cosmology and astrophysics.

​

•  Extract cosmological information while marginalizing over baryonic effects.

​

•  Find the mapping between N-body and hydrodynamic simulations.

​

•  Quantify the dependence of galaxy formation and evolution on astrophysics and cosmology.

​​

•  Use machine learning to efficiently calibrate subgrid parameters in cosmological hydrodynamic simulations to match a set of observations.​

 

​

CAMELS have been designed to achieve these goals by making using of machine learning techniques. With more than 4,000 numerical simulations, both N-body and (magneto-)hydrodynamic, CAMELS follow the evolution of more than 100 billion dark matter particles and fluid elements in a combined volume of ~ (400 Mpc/h)^3. CAMELS span thousands of different cosmological and astrophysical models; the simulations represent thus a large dataset to train machine learning algorithms.

​

Some simple machine learning applications to the CAMEL simulations are these:

​

• Non-linear interpolation

• Parameter estimation

• Symbolic regression

• Generative models

• Anomaly detection