topobench.transforms.liftings.liftings module#

This module implements the abstract classes for lifting graphs.

class AbstractLifting(feature_lifting=None, **kwargs)#

Bases: BaseTransform

Abstract class for topological liftings.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting=None, **kwargs)#

forward(data)#

Apply the full lifting (topology + features) to the input data.

Parameters:

datatorch_geometric.data.Data: The input data to be lifted.

Returns:

torch_geometric.data.Data: The lifted data.

abstract lift_topology(data)#

Lift the topology of a graph to higher-order topological domains.

Parameters:

datatorch_geometric.data.Data: The input data to be lifted.

Returns:

dict: The lifted topology.

class CellComplexLifting(feature_lifting='ProjectionSum', **kwargs)#

Bases: AbstractLifting

Abstract class for lifting cell complexes to other domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', **kwargs)#

class CombinatorialLifting(feature_lifting='ProjectionSum', **kwargs)#

Bases: AbstractLifting

Abstract class for lifting combinatorial complexes to other domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', **kwargs)#

class GraphLifting(feature_lifting='ProjectionSum', preserve_edge_attr=False, **kwargs)#

Bases: AbstractLifting

Abstract class for lifting graph topologies to other domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
preserve_edge_attrbool, optional: Whether to preserve edge attributes. Default is False.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', preserve_edge_attr=False, **kwargs)#

class HypergraphLifting(feature_lifting='ProjectionSum', **kwargs)#

Bases: AbstractLifting

Abstract class for lifting hypergraphs to other domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', **kwargs)#

class PointCloudLifting(feature_lifting='ProjectionSum', **kwargs)#

Bases: AbstractLifting

Abstract class for lifting point clouds to other topological domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', **kwargs)#

class SimplicialLifting(feature_lifting='ProjectionSum', **kwargs)#

Bases: AbstractLifting

Abstract class for lifting simplicial complexes to other domains.

Parameters:

feature_liftingstr, optional: The feature lifting method to be used. Default is ‘ProjectionSum’.
**kwargsoptional: Additional arguments for the class.

__init__(feature_lifting='ProjectionSum', **kwargs)#

is_undirected(edge_index, edge_attr=None, num_nodes=None)#

Returns True if the graph given by edge_index is undirected.

Parameters:

edge_index (LongTensor) – The edge indices.
edge_attr (Tensor or List[Tensor], optional) – Edge weights or multi- dimensional edge features. If given as a list, will check for equivalence in all its entries. (default: None)
num_nodes (int, optional) – The number of nodes, i.e. max(edge_index) + 1. (default: None)

Return type:

bool

Examples

>>> edge_index = torch.tensor([[0, 1, 0],
...                         [1, 0, 0]])
>>> weight = torch.tensor([0, 0, 1])
>>> is_undirected(edge_index, weight)
True

>>> weight = torch.tensor([0, 1, 1])
>>> is_undirected(edge_index, weight)
False

to_undirected(edge_index, edge_attr='???', num_nodes=None, reduce='add')#

Converts the graph given by edge_index to an undirected graph such that \((j,i) \in \mathcal{E}\) for every edge \((i,j) \in \mathcal{E}\).

Parameters:

edge_index (LongTensor) – The edge indices.
edge_attr (Tensor or List[Tensor], optional) – Edge weights or multi- dimensional edge features. If given as a list, will remove duplicates for all its entries. (default: None)
num_nodes (int, optional) – The number of nodes, i.e. max(edge_index) + 1. (default: None)
reduce (str, optional) – The reduce operation to use for merging edge features ("add", "mean", "min", "max", "mul"). (default: "add")

Return type:

LongTensor if edge_attr is not passed, else (LongTensor, Optional[Tensor] or List[Tensor]])

Warning

From :pyg:`PyG >= 2.3.0` onwards, this function will always return a tuple whenever edge_attr is passed as an argument (even in case it is set to None).

Examples

>>> edge_index = torch.tensor([[0, 1, 1],
...                            [1, 0, 2]])
>>> to_undirected(edge_index)
tensor([[0, 1, 1, 2],
        [1, 0, 2, 1]])

>>> edge_index = torch.tensor([[0, 1, 1],
...                            [1, 0, 2]])
>>> edge_weight = torch.tensor([1., 1., 1.])
>>> to_undirected(edge_index, edge_weight)
(tensor([[0, 1, 1, 2],
        [1, 0, 2, 1]]),
tensor([2., 2., 1., 1.]))

>>> # Use 'mean' operation to merge edge features
>>>  to_undirected(edge_index, edge_weight, reduce='mean')
(tensor([[0, 1, 1, 2],
        [1, 0, 2, 1]]),
tensor([1., 1., 1., 1.]))