Representation

class sakkara.relation.representation.Representation#

Abstract class for a representation, i.e., class that transforms variables and other elements between different shapes.

abstract get_member_array(group: Group) → ndarray[Any, dtype[Any]]#: Get an array shaped as a variable with this representation, where each value correspond to the member of the input group.

abstract get_member_tuples() → List[Tuple[Any, ...]]#: Get a (flat) list of all member tuples of this representation

abstract get_members() → Tuple[ndarray[Any, dtype[ScalarType]], ...]#: Get member arrays for all groups of this representation. For documentation of a member array, see Representation.get_member_array().

abstract get_shape() → Tuple[int, ...]#: Get the shape of a variable corresponding to a component with this representation

abstract map(element: Any, target_representation: Representation) → Any#

Map/transform an element (variable or equivalent) corresponding to this representation to a target representation.

Parameters:

element – The object/variable to transform.
target_representation – The representation that corresponds to the shape the variable should be in.

Returns:

The element but transformed to target representation.

class sakkara.relation.representation.UnrepeatableRepresentation#

Simple representation for elements that always have a single value and that can not be repeated.

get_member_array(group: Group) → ndarray[Any, dtype[Any]]#: Get an array shaped as a variable with this representation, where each value correspond to the member of the input group.

get_member_tuples() → List[Tuple[Any, ...]]#: Get a (flat) list of all member tuples of this representation

get_members() → Tuple[ndarray[Any, dtype[ScalarType]], ...]#: Get member arrays for all groups of this representation. For documentation of a member array, see Representation.get_member_array().

get_shape() → Tuple[int, ...]#: Get the shape of a variable corresponding to a component with this representation

map(element: object, target_representation: Representation)#

Map/transform an element (variable or equivalent) corresponding to this representation to a target representation.

Parameters:

element – The object/variable to transform.
target_representation – The representation that corresponds to the shape the variable should be in.

Returns:

The element but transformed to target representation.

class sakkara.relation.representation.TensorRepresentation(*groups: Group)#

Class for representation shapes corresponding to sequneces of groups.

Parameters:: *groups – Group objects to include in representation

abstract add_group(group: Group)#: Add group to the representation.

get_group_mapping(target: Representation) → Dict[Group, Tuple[Group, ...]]#

Check if representation is mappable to another representation, i.e., groups of the other representation have twins or parents among the groups of this representation.

Returns:: True is self is mappable to target

get_member_array(group: Group) → ndarray[Any, dtype[Any]]#: Get an array shaped as a variable with this representation, where each value correspond to the member of the input group.

get_member_tuples() → List[Tuple[Any, ...]]#: Get a (flat) list of all member tuples of this representation

get_members() → Tuple[ndarray[Any, dtype[ScalarType]], ...]#: Get member arrays for all groups of this representation. For documentation of a member array, see Representation.get_member_array().

get_shape() → Tuple[int, ...]#: Get the shape of a variable corresponding to a component with this representation

map(element: Any, target: Representation) → Any#

Map/transform an element (variable or equivalent) corresponding to this representation to a target representation.

Parameters:

element – The object/variable to transform.
target_representation – The representation that corresponds to the shape the variable should be in.

Returns:

The element but transformed to target representation.

class sakkara.relation.representation.MinimalTensorRepresentation(*groups: Group)#

Tensor representation that always keeps the minimal set of groups needed. E.g., groups will not be added to this representation if there already is a twin or a child in the representation. When adding a group that has a parent in the representation, the group will be added and the parent will be removed. This class is currently the only implementation of TensorRepresentation.

add_group(group: Group) → None#: Add group to the representation.

Representation

Contents