nDGPemu

Emulator of the boost factor for the dark matter power spectrum in nDGP gravity models

Installation

It is recommended to install the package in a dedicated python3 environment. The package requires:

• numpy,
• joblib,
• scipy,
• scikit-learn.

To install the package, access the package directory from a terminal window and execute:

pip install .

Using the emulator

The emulator model can be imported in python with

from nDGPemu import BoostPredictor

Instantiating a BoostPredictor object loads the MLP model and the auxiliary data

model = BoostPredictor()

The model can be used to predict the boost factor with the syntax

Bk = model.predict(H0rc, z, cosmo_params)

where H0rc, z, cosmo_params are user defined paramters.

The emulator assumes a flat $\Lambda$-CDM cosmology and neglets the energy density of radiation and neutrinos. The cosmological parameters required in the dictonary cosmo_params to obtain the nDGP boost factors and their allowed ranges are:

• $\Omega_{\rm m} \in [0.28,0.36]$,
• $\Omega_{\rm b} \in [0.04,0.06]$,
• $n_{\rm s} \in [0.92,1]$,
• $A_{\rm s} \in [1.7e-9,2.5e-9]$,
• $h \in [0.61,0.73]$.

Notice that the parameter $\Omega_{\rm m}$ accounts for the sum of CDM and baryonic matter and the amplitude of the primordial power spectrum $A_{\rm s}$ is defined at $k_{\rm pivot} = 0.05 , {\rm Mpc}^{-1}$. Also the modified gravity parameter $H_0r_c$ and the redsfhit $z$ are required and their interpolation ranges are:

• $H_0 r_c \in [0.2,20]$.
• $z \in [0,2]$,

A minimal working example for the emulator is shown in the example notebook.