pybird.inference module

class pybird.inference.Inference(free_cosmo_name, fiducial_cosmo, likelihood_config, cosmo_prior_config=None, boltzmann='class', free_nuisance_name=None, fiducial_nuisance=None, verbose=True)

Bases: object

A class for cosmological parameter inference using EFT of LSS.

The Inference class implements parameter inference for cosmological models using the Effective Field Theory of Large Scale Structure. It handles likelihood calculations, parameter sampling, minimization, and various inference techniques.

cosmo_prior_covmat

Covariance matrix for cosmological parameter priors.

Type:

ndarray

cosmo_prior_dict

Dictionary of cosmological parameter priors.

Type:

dict

likelihood_config

Configuration for the likelihood calculation.

Type:

dict

L

Likelihood instance for calculations.

Type:

Likelihood

l

Dictionary of free parameters and their values.

Type:

dict

M

Boltzmann solver instance (CLASS, Symbolic, or CPJ).

Type:

object

need_cosmo_update

Whether cosmological parameters need updating.

Type:

bool

T

Taylor expansion instance if used.

Type:

Taylor

bias

Debiasing terms if computed.

Type:

ndarray

set_nuisance()

Set nuisance parameters for the likelihood.

set_config_and_boltzmann()

Configure cosmological parameters and Boltzmann solver.

set_need_cosmo_update()

Determine if cosmological parameters need updating.

get_param_name_and_pos()

Get parameter names and positions.

init()

Initialize inference setup.

set_sampler()

Configure parameter sampler (emcee, nuts, mclmc, etc.).

set_minimizer()

Configure parameter minimization.

set_debiasing()

Compute debiasing terms.

set_model_cache()

Cache model at best-fit point.

set_fake()

Generate fake data from best-fit model.

set_taylor()

Set up Taylor expansion around fiducial cosmology.

_set_taylor()

Internal method to compute Taylor expansion.

update_boltzmann()

Update Boltzmann solver with new parameters.

_loglkl()

Compute log-likelihood.

_logm()

Compute log of marginalization term.

_logp()

Compute log-posterior.

set_logp()

Configure log-posterior calculation.

set_nuisance(likelihood_config, free_nuisance_name=None, fiducial_nuisance=None)

set_config_and_boltzmann(free_cosmo_name, free_nuisance_name, fiducial_cosmo, fiducial_nuisance, boltzmann='class')

set_need_cosmo_update()

get_param_name_and_pos(verbose=True)

init(minimize=False, cosmo_prior=False, ext_probe=False, ext_loglkl=None, jax_jit=False, measure=False, taylor_measure=False, debiasing=False, hessian_type=None, vectorize=False, taylor=False, order=3, verbose=True)

set_sampler(sampler='emcee', cosmo_prior=False, ext_probe=False, ext_loglkl=None, jax_jit=False, measure=False, taylor_measure=False, debiasing=False, hessian_type=None, vectorize=False, taylor=False, return_extras=False, options={}, verbose=True)

set_minimizer(minimizer='', cosmo_prior=False, ext_probe=False, ext_loglkl=None, jax_jit=False, taylor=False, options={}, order=3, verbose=True)

set_debiasing(hessian_type='H', cosmo_prior=False, ext_probe=False, ext_loglkl=None, taylor=False, verbose=True)

set_model_cache(verbose=True)

set_fake(sample_fake=False, options={}, verbose=True)

set_taylor(f, bird_correlator=True, log_measure=False, order=3, verbose=True)

update_boltzmann(cosmo)

set_logp(cosmo_prior=False, ext_probe=False, ext_loglkl=None, jax_jit=False, measure=False, taylor_measure=False, hessian_type=None, vectorize=False, taylor=False, verbose=True)