Usage

Explore the tutorials in the cloe-org/playground/tutorials/likelihood repository for examples on how to compute cosmological observables and use cloelike.

Basic example: BAO likelihood

import numpy as np
from cloelike.EuclidLikelihood_GCspectro_BAO import EuclidLikelihood_GCspectro_BAO
from cloelib.cosmology.CAMBcosmology import CAMBBackground  # or your preferred Background class

# Load your data dictionary (e.g. from a .pkl or .fits file)
data = {...}  # Check out the expected format in notebooks

# Instantiate the likelihood
likelihood = EuclidLikelihood_GCspectro_BAO(data=data, Background=CAMBBackground)

# Evaluate the log-likelihood for a set of cosmological parameters
parameters = {
    "H0": 67.32,
    "Omega_cdm0": 0.264,
    "Omega_b0": 0.049,
    "Omega_k0": 0.0,
    "w0": -1.0,
    "wa": 0.0,
    "ns": 0.966,
    "sigma8": 0.816,
}
log_like = likelihood.loglike(parameters)

Basic example: Photometric 3×2pt likelihood

from cloelib.cosmology.camb_cosmology import CAMBBackground
from cloelib.cosmology.HMcode2020Emu_cosmology import HMemuLinearPerturbations, HMemuNonLinearPerturbations
from cloelike.EuclidLikelihood_photo_Cls import EuclidLikelihood_3x2pt

# Load data and settings dictionaries
data = {...}
settings = {...}

# Provide Background, linear and non-linear perturbation classes
likelihood = EuclidLikelihood_3x2pt(
    data=data,
    settings=settings,
    Background=CAMBBackground,
    LinPerturbations=HMemuLinearPerturbations,
    NonLinPerturbations=HMemuNonLinearPerturbations,
)

log_like = likelihood.loglike(parameters)

Selecting the photometric theory model

The intrinsic-alignment model, galaxy bias model and nonlinear/baryonic options are chosen in settings. All keys are optional. The photometric Cls, 2PCF and COSEBIs likelihoods accept the same ones (COSEBIs is shear only, so it ignores the galaxy-bias keys).

Key	Options	Parameters read from parameters
ia_model	"NLA" (default), "TATT", None	NLA: AIA, EtaIA, CIA. TATT: AIA, A2IA, bTA, optionally EtaIA, Eta2IA, CIA, z0IA (default 0.62)
galaxy_bias_model	"poly" (default), "per_bin", "per_bin_int", "nonlinear"	poly: b1_photo_poly0..3. per*bin: b1_photo_bin{i}. nonlinear: b1_photo_nl_bin{i}, optionally b2*, bs2*, b3nl*, bk2_photo_nl_bin{i} (i starts at 0)
include_rsd	False (default), True	Not supported with galaxy_bias_model="nonlinear"
nonlinear_param_keys	default ("log10TAGN",)	Passed as keyword arguments to NonLinPerturbations. Use () for backends without baryonic feedback (e.g. EE2)
baryon_param_keys	default ()	Passed as baryon_kwargs to a class built with cloelib.cosmology.cosmology.with_baryon_boost
nonlinear_kwargs	default {}	Fixed keyword arguments for NonLinPerturbations, e.g. {"nonlinear_model": "mead2020"}

TATT and the nonlinear galaxy bias compute their one-loop kernels with FAST-PT (pip install fast-pt) from the linear perturbations. You can supply your own kernels with settings["tatt_loop_computer"] / settings["nl_bias_loop_computer"]: each is a callable that takes the linear perturbations and returns an object with a .compute(name) method.

For example, TATT with nonlinear galaxy bias and the FLAMINGO baryonic response on top of HMcode2020Emu:

from cloelib.cosmology.cosmology import with_baryon_boost
from cloelib.cosmology.FlamingoBaryonResponseEmulator_cosmology import FlamingoBaryonBoostMixin

settings = {
    **settings,
    "ia_model": "TATT",
    "galaxy_bias_model": "nonlinear",
    "nonlinear_param_keys": (),
    "baryon_param_keys": ("fgas_sigma", "Mstar_sigma", "jet_fraction"),
}
likelihood = EuclidLikelihood_3x2pt(
    data=data,
    settings=settings,
    Background=CAMBBackground,
    LinPerturbations=HMemuLinearPerturbations,
    NonLinPerturbations=with_baryon_boost(HMemuNonLinearPerturbations, FlamingoBaryonBoostMixin),
)

Basic example: Spectroscopic GCspectro likelihood

from cloelib.cosmology.camb_cosmology import CAMBBackground
from cloelib.observables.spectro.CometEFT_spectro import CometEFT_SpectroPower
from cloelike.EuclidLikelihood_GCspectro_Pls import EuclidLikelihood_GCspectro_Pls

likelihood = EuclidLikelihood_GCspectro_Pls(
    data=data,
    settings=settings,
    Background=CAMBBackground,
    SpectroPower=CometEFT_SpectroPower,
)

log_like = likelihood.loglike(parameters)