jesterTOV.inference.base.prior.SequentialTransformPrior

class SequentialTransformPrior(base_prior, transforms)

Bases: Prior

Transform a prior distribution by applying a sequence of transforms.

Note: This class follows the Jim/jimgw architecture.

The space before the transform is named as x, and the space after the transform is named as z.

__init__(base_prior, transforms)

Methods

__init__(base_prior, transforms)
add_name(x)	Turn an array into a dictionary.
log_prob(z)	Evaluate the probability of the transformed variable z.
sample(rng_key, n_samples)	Sample from the transformed prior.
transform(x)	Apply forward transforms to x.

Attributes

n_dim
base_prior
transforms
parameter_names
composite

base_prior: Prior

log_prob(z)

Evaluate the probability of the transformed variable z.

This is what flowMC should sample from.

Parameters:

z (dict[str, Float]) – Dictionary of parameter names to values in transformed space.

Return type:

Float

Returns:

log_prob (Float) – The log probability including Jacobian correction.

sample(rng_key, n_samples)

Sample from the transformed prior.

Parameters:

• rng_key (PRNGKeyArray) – A random key to use for sampling.

• n_samples (int) – The number of samples to draw.

Return type:

dict[str, Float[Array, 'n_samples']]

Returns:

samples (dict) – Samples from the transformed distribution.

transform(x)

Apply forward transforms to x.

Parameters:

x (dict[str, Float]) – Dictionary of parameter names to values.

Return type:

dict[str, Float]

Returns:

z (dict[str, Float]) – Transformed dictionary.

transforms: list[BijectiveTransform]