petitRADTRANS.retrieval.parameter

Classes

Parameter	Parameter
RetrievalParameter

Functions

_cast_distribution_to_jax_dtype(distribution)	Re-create a TFP distribution with floating-point parameters cast to the active JAX dtype.
_build_jaxns_prior(distribution, name)

Module Contents

petitRADTRANS.retrieval.parameter._cast_distribution_to_jax_dtype(distribution)

Re-create a TFP distribution with floating-point parameters cast to the active JAX dtype.

petitRADTRANS.retrieval.parameter._build_jaxns_prior(distribution, name)

class petitRADTRANS.retrieval.parameter.Parameter(name, is_free_parameter, value=0.0, distribution=None, transform_prior_cube_coordinate=None, plot_in_corner=False, corner_ranges=None, corner_transform=None, corner_label=None, units=None)

Parameter This class provides the interface for defining and managing parameters in a retrieval. Each parameter includes a name, which can be used as a reference in the model function, a value, and flag of whether it’s a free parameter, and if it’s free. You can also define a function that translates the unit hypercube into physical space for a pymultinest retrieval, or a tensorflow probability distribution to define the prior for JAX-based retrievals. The remainder of the arguments deal with the corner plots.

Args:

namestring

The name of the parameter. Must match the name used in the model function.

is_free_parameterbool

True if the parameter should be sampled in the retrieval

valuefloat

The value of the parameter. Set using set_param.

transform_prior_cube_coordinatemethod

Transform the unit interval [0,1] to the physical space of the parameter.

plot_in_cornerbool

True if this parameter should be included in the output corner plot

corner_rangesTuple(float,float)

The axis range of the parameter in the corner plot

corner_transformmethod

A function to scale or transform the value of the parameter for prettier plotting.

corner_labelstring

The axis label for the parameter, defaults to name.

unitsastropy.units.UnitBase, optional

The physical units of the parameter value. Defaults to None, in which case the value is assumed to already be in the units expected by the model function (cgs). If set to an astropy unit, the model functions can use it to convert the value to the units required for the radiative transfer calculation (e.g. cm, g).

name

is_free_parameter

value = 0.0

distribution

prior_sample = None

transform_prior_cube_coordinate = None

plot_in_corner = False

corner_ranges = None

corner_transform = None

corner_label = None

units = None

_tree_flatten()

classmethod _tree_unflatten(aux_data, children)

get_param_uniform(cube)

get_flattened_value(value=None)

set_param(value)

set_prior()

class petitRADTRANS.retrieval.parameter.RetrievalParameter(name, prior_parameters, prior_type='uniform', custom_prior=None)

__available_priors = ['log', 'uniform', 'gaussian', 'log_gaussian', 'delta', 'custom']

name

prior_parameters

prior_type = 'uniform'

prior_function

classmethod from_dict(dictionary)

Convert a dictionary into a list of RetrievalParameter. The keys of the dictionary are the names of the RetrievalParameter. The values of the dictionary must be dictionaries with keys ‘prior_parameters’ and ‘prior_type’.

Args:

dictionary: a dictionary

Returns:

A list of RetrievalParameter.

put_into_dict(dictionary=None)

Convert a RetrievalParameter into a dictionary.

Args:

dictionary: a dictionary; if None, a new dictionary is created

Returns:

A dictionary.