torch_mist.distributions.parametrizations.map

Module Contents

Classes

ParameterMap	Mapping from a list of tensors to a dictionary representing the parametrization of a distribution.
DeltaMap	Delta Parameter mapping: maps a list of one tensor into a delta distribution at the same location
LocScaleMap	Mapping from a list of tensors to a dictionary representing the parametrization of a distribution.
LogitsMap	Mapping from a list of tensors to a dictionary representing the parametrization of a distribution.

Attributes

EPSILON

torch_mist.distributions.parametrizations.map.EPSILON = 1e-06

class torch_mist.distributions.parametrizations.map.ParameterMap

Mapping from a list of tensors to a dictionary representing the parametrization of a distribution. Abstract class.

supported_distributions: List[Type[torch.distributions.Distribution]] = []

n_parameters: int = 0

abstract map_parameters(parameter_list: List[torch.Tensor]) → Dict[str, torch.Tensor]

Abstract method responsible for defining the mapping from the list of tensors to the parameter dictionary.

Parameters:

parameter_list (List[torch.Tensor]) – The list of parameters for the distribution of interest.

Returns:

The named parameters for the distribution of interest.

Return type:

Dict[str, torch.Tensor]

__call__(parameter_list: torch.Tensor | List[torch.Tensor]) → Dict[str, torch.Tensor]

class torch_mist.distributions.parametrizations.map.DeltaMap

Bases: ParameterMap

Delta Parameter mapping: maps a list of one tensor into a delta distribution at the same location

supported_distributions: List[Type[torch.distributions.Distribution]]

n_parameters: int = 1

map_parameters(parameter_list: List[torch.Tensor]) → Dict[str, torch.Tensor]

Abstract method responsible for defining the mapping from the list of tensors to the parameter dictionary.

Parameters:

parameter_list (List[torch.Tensor]) – The list of parameters for the distribution of interest.

Returns:

The named parameters for the distribution of interest.

Return type:

Dict[str, torch.Tensor]

class torch_mist.distributions.parametrizations.map.LocScaleMap(scale: float | None = None, epsilon: float = EPSILON, scale_offset: float = 0.0)

Bases: ParameterMap

Mapping from a list of tensors to a dictionary representing the parametrization of a distribution. Abstract class.

supported_distributions: List[Type[torch.distributions.Distribution]]

map_parameters(parameter_list: List[torch.Tensor]) → Dict[str, torch.Tensor]

Abstract method responsible for defining the mapping from the list of tensors to the parameter dictionary.

Parameters:

parameter_list (List[torch.Tensor]) – The list of parameters for the distribution of interest.

Returns:

The named parameters for the distribution of interest.

Return type:

Dict[str, torch.Tensor]

class torch_mist.distributions.parametrizations.map.LogitsMap

Bases: ParameterMap

Mapping from a list of tensors to a dictionary representing the parametrization of a distribution. Abstract class.

supported_distributions: List[Type[torch.distributions.Distribution]]

n_parameters: int = 1

map_parameters(parameter_list: List[torch.Tensor]) → Dict[str, torch.Tensor]

Abstract method responsible for defining the mapping from the list of tensors to the parameter dictionary.

Parameters:

parameter_list (List[torch.Tensor]) – The list of parameters for the distribution of interest.

Returns:

The named parameters for the distribution of interest.

Return type:

Dict[str, torch.Tensor]