jesterTOV.inference.base.transform.BijectiveTransform

class BijectiveTransform(name_mapping)

Bases: NtoNTransform

Bijective (invertible) N-to-N parameter transform with Jacobian corrections.

Note: This class follows the Jim/jimgw architecture. Used for sample transforms where parameters are transformed during MCMC sampling and Jacobian corrections are applied to the prior.

__init__(name_mapping)

Parameters:

name_mapping (tuple[list[str], list[str]]) – Tuple of (input_names, output_names) for the transform.

Methods

__init__(name_mapping)
backward(y)	Pull back the input y to original coordinate x (without Jacobian).
forward(x)	Push forward the input x to transformed coordinate y.
inverse(y)	Inverse transform the input y to original coordinate x.
propagate_name(x)	Propagate parameter names through the transform.
transform(x)	Transform the input x to transformed coordinate y and return the log Jacobian determinant.

Attributes

n_dim
inverse_transform_func
transform_func
name_mapping

backward(y)

Pull back the input y to original coordinate x (without Jacobian).

Parameters:

y (ParamDict) – The transformed dictionary.

Return type:

dict[str, Float]

Returns:

x (ParamDict) – The original dictionary.

inverse(y)

Inverse transform the input y to original coordinate x.

Parameters:

y (ParamDict) – The transformed dictionary.

Return type:

tuple[dict[str, Float], Float]

Returns:

• x (ParamDict) – The original dictionary.

• log_det (Float) – The log Jacobian determinant.

inverse_transform_func: Callable[[dict[str, Float]], dict[str, Float]]