Likelihood Interfaces

BOBE provides interfaces for different types of likelihood functions.

Base Likelihood

class BOBE.likelihood.Likelihood(loglikelihood, param_list, param_labels=None, param_bounds=None, name=None, minus_inf=-10000000000.0)[source]

Bases: object

Base class for log-likelihoods with common evaluation logic.

Parameters:

  • loglikelihood (Callable) – Log-likelihood function that takes parameter array and returns float.

  • param_list (Optional[List[str]]) – List of parameter names.

  • param_labels (Optional[List[str]]) – LaTeX labels for parameters. Default is None (uses param_list).

  • param_bounds (Union[List, ndarray, None]) – Parameter bounds, shape (2, ndim). Default is None (unit cube).

  • name (Optional[str]) – Name for this likelihood. Default is “loglikelihood”.

  • minus_inf (float) – Value to return for failed evaluations. Default is -1e5.

__init__(loglikelihood, param_list, param_labels=None, param_bounds=None, name=None, minus_inf=-10000000000.0)[source]
__call__(X)[source]

Evaluate the likelihood function at a single point.

This method is designed to be called by pool.run_map_objective, which handles distributing tasks across MPI processes in parallel mode or iterating in serial mode.

Parameters:

X (Union[ndarray, List[float]]) – Single input parameter vector.

Returns:

val – Log-likelihood value at the input point.

Return type:

float

Cobaya Likelihood

Interface for Cobaya cosmological parameter estimation.

class BOBE.likelihood.CobayaLikelihood(input_file_dict, confidence_for_unbounded=0.9999995, minus_inf=-10000000000.0, name='CobayaLikelihood')[source]

Bases: Likelihood

Likelihood wrapper for Cobaya models.

Parameters:

  • input_file_dict (Union[str, Dict[str, Any]]) – Cobaya input YAML file path or input dictionary.

  • confidence_for_unbounded (float) – Confidence level for unbounded priors. Default is 0.9999995.

  • minus_inf (float) – Value to return for failed/minus infinity evaluations. Default is -1e10.

  • name (str) – Name for this likelihood. Default is “CobayaLikelihood”.

__init__(input_file_dict, confidence_for_unbounded=0.9999995, minus_inf=-10000000000.0, name='CobayaLikelihood')[source]
__call__(X)[source]

Evaluate Cobaya likelihood with logprior volume correction. This is added to match Cobaya behaviour.

Return type:

float