In this notebook, we show a few usage examples for how to read and write Cosmology objects to disk.
from pyccl import Cosmology
cosmo = Cosmology(Omega_c=0.25, Omega_b=0.05, sigma8=0.8, h=0.7, n_s=0.96)
print(cosmo.__doc__)
A cosmology including parameters and associated data (e.g. distances,
power spectra).
.. note:: Although some arguments default to `None`, they will raise a
ValueError inside this function if not specified, so they are not
optional.
.. note:: The parameter ``Omega_g`` can be used to set the radiation density
(not including relativistic neutrinos) to zero. Doing this will
give you a model that is physically inconsistent since the
temperature of the CMB will still be non-zero.
.. note:: After instantiation, you can set parameters related to the
internal splines and numerical integration accuracy by setting
the values of the attributes of
:obj:`Cosmology.cosmo.spline_params` and
:obj:`Cosmology.cosmo.gsl_params`. For example, you can set
the generic relative accuracy for integration by executing
``c = Cosmology(...); c.cosmo.gsl_params.INTEGRATION_EPSREL = 1e-5``.
See the module level documentation of `pyccl.core` for details.
Args:
Omega_c (:obj:`float`): Cold dark matter density fraction.
Omega_b (:obj:`float`): Baryonic matter density fraction.
h (:obj:`float`): Hubble constant divided by 100 km/s/Mpc; unitless.
A_s (:obj:`float`): Power spectrum normalization. Exactly one of A_s
and sigma_8 is required.
sigma8 (:obj:`float`): Variance of matter density perturbations at
an 8 Mpc/h scale. Exactly one of A_s and sigma_8 is required.
n_s (:obj:`float`): Primordial scalar perturbation spectral index.
Omega_k (:obj:`float`): Curvature density fraction.
Defaults to 0.
Omega_g (:obj:`float`): Density in relativistic species
except massless neutrinos. The default of `None` corresponds
to setting this from the CMB temperature. Note that if a non-`None`
value is given, this may result in a physically inconsistent model
because the CMB temperature will still be non-zero in the
parameters.
Neff (:obj:`float`): Effective number of massless
neutrinos present. Defaults to 3.044.
m_nu (:obj:`float` or `array`):
Mass in eV of the massive neutrinos present. Defaults to 0.
If a sequence is passed, it is assumed that the elements of the
sequence represent the individual neutrino masses.
mass_split (:obj:`str`): Type of massive neutrinos. Should
be one of 'single', 'equal', 'normal', 'inverted'. 'single' treats
the mass as being held by one massive neutrino. The other options
split the mass into 3 massive neutrinos. Ignored if a sequence is
passed in m_nu. Default is 'normal'.
w0 (:obj:`float`): First order term of dark energy equation
of state. Defaults to -1.
wa (:obj:`float`): Second order term of dark energy equation
of state. Defaults to 0.
T_CMB (:obj:`float`): The CMB temperature today. The default of
is 2.725.
transfer_function (:obj:`str` or :class:`~pyccl.emulators.emu_base.EmulatorPk`):
The transfer function to use. Defaults to 'boltzmann_camb'.
matter_power_spectrum (:obj:`str` or :class:`~pyccl.emulators.emu_base.EmulatorPk`):
The matter power spectrum to use. Defaults to 'halofit'.
baryonic_effects (:class:`~pyccl.baryons.baryons_base.Baryons` or `None`):
The baryonic effects model to use. Options are `None` (no baryonic effects), or
a :class:`~pyccl.baryons.baryons_base.Baryons` object.
mg_parametrization (:class:`~pyccl.modified_gravity.modified_gravity_base.ModifiedGravity`
or `None`):
The modified gravity parametrization to use. Options are `None` (no MG), or
a :class:`~pyccl.modified_gravity.modified_gravity_base.ModifiedGravity` object.
Currently, only :class:`~pyccl.modified_gravity.MuSigmaMG` is supported.
extra_parameters (:obj:`dict`): Dictionary holding extra
parameters. Currently supports extra parameters for CAMB.
Details described below. Defaults to None.
T_ncdm (:obj:`float`): Non-CDM temperature in units of photon
temperature. The default is 0.71611.
Currently supported extra parameters for CAMB are:
* `halofit_version`
* `HMCode_A_baryon`
* `HMCode_eta_baryon`
* `HMCode_logT_AGN`
* `kmax`
* `lmax`
* `dark_energy_model`
Consult the CAMB documentation for their usage. These parameters are passed
in a :obj:`dict` to `extra_parameters` as::
extra_parameters = {"camb": {"halofit_version": "mead2020_feedback",
"HMCode_logT_AGN": 7.8}}
.. note :: If using camb to compute the non-linear power spectrum with HMCode
to include baryonic effects, you should not include any extra
baryonic effects (i.e. set `baryonic_effects=None`).
Cosmology objects can be saved to a YAML format using the write_yaml method. This format is not currently very robust -- the exact order of the parameters must be maintained or the object cannot be read back in.
cosmo.write_yaml('example_params.yaml')
!cat example_params.yaml
Omega_c: 0.25
Omega_b: 0.05
h: 0.7
n_s: 0.96
sigma8: 0.8
A_s: null
Omega_k: 0.0
Omega_g: null
Neff: 3.044
m_nu: 0.0
mass_split: normal
w0: -1.0
wa: 0.0
T_CMB: 2.7255
T_ncdm: 0.71611
extra_parameters: {}
transfer_function: boltzmann_camb
matter_power_spectrum: halofit
The parameters can be read back in using the read_yaml class method. Note that this must be called on the Cosmology class itself, as shown below, and not an instance of the class.
cosmo2 = Cosmology.read_yaml("example_params.yaml")
print(cosmo2)
<pyccl.cosmology.Cosmology> Neff = 3.044 Omega_b = 0.05 Omega_c = 0.25 h = 0.7 n_s = 0.96 sigma8 = 0.8 extra_parameters = HASH_ACCURACY_PARAMS = 0xb6e46c24158c0e30
This Cosmology object can then be used to obtain cosmological predictions. See the other examples in this directory, for example Distance Calculations Example.ipynb or the more comprehensive demo SLAC Feb2018 Demo.ipynb.
Cosmology objects are also pickle-able, to make them easy to store on disk and to pass around in MPI environments.
import pickle
with open('cosmo.pkl', 'wb') as fp:
pickle.dump(cosmo2, fp)
with open('cosmo.pkl', 'rb') as fp:
cosmo3 = pickle.load(fp)
print(cosmo3)
<pyccl.cosmology.Cosmology> Neff = 3.044 Omega_b = 0.05 Omega_c = 0.25 h = 0.7 n_s = 0.96 sigma8 = 0.8 extra_parameters = HASH_ACCURACY_PARAMS = 0xb6e46c24158c0e30