Skip to content

graph

TemporalGraph

Bases: pathpyG.Graph

Source code in src/pathpyG/core/temporal_graph.py 
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
class TemporalGraph(Graph):
    def __init__(self, data: Data, mapping: IndexMap | None = None) -> None:
        """Creates an instance of a temporal graph from a `TemporalData` object.

        Args:
            data: PyG `Data` object containing edges saved in `edge_index` and timestamps in `time`.
            mapping: Optional mapping from node IDs to indices.

        Example:
            ```py
            from pytorch_geometric.data import TemporalData
            import pathpyG as pp

            d = Data(edge_index=[[0,0,1], [1,2,2]], time=[0,1,2])
            t = pp.TemporalGraph(d, mapping)
            print(t)
            ```
        """
        self.data = data
        if not isinstance(self.data.edge_index, EdgeIndex):
            self.data.edge_index = EdgeIndex(
                data=self.data.edge_index.contiguous(), sparse_size=(self.data.num_nodes, self.data.num_nodes)
            )

        # reorder temporal data
        # Note that we do not use `torch_geometric.self.data.Data.sort_by_time` because it cannot sort numpy arrays`
        sorted_idx = torch.argsort(self.data.time)
        for edge_attr in set(self.data.edge_attrs()).union(set(["time"])):
            if edge_attr == "edge_index":
                self.data.edge_index = self.data.edge_index[:, sorted_idx]
            else:
                self.data[edge_attr] = self.data[edge_attr][sorted_idx]

        if mapping is not None:
            self.mapping = mapping
        else:
            self.mapping = IndexMap()

        # create mapping between edge index and edge tuples
        self.edge_to_index = {
            (e[0].item(), e[1].item()): i for i, e in enumerate(self.data.edge_index.t())
        }
        self.tedge_to_index = {
            (e[0].item(), e[1].item(), t.item()): i for i, (e, t) in enumerate(zip([e for e in self.data.edge_index.t()], self.data.time))
        }

        self.start_time = self.data.time[0].item()
        self.end_time = self.data.time[-1].item()

    @staticmethod
    def from_edge_list(edge_list, num_nodes: Optional[int] = None, device: Optional[torch.device] = None) -> TemporalGraph:  # type: ignore
        """Create a temporal graph from a list of tuples containing edges with timestamps."""
        edge_array = np.array(edge_list)

        # Convert timestamps to tensor
        if isinstance(edge_list[0][2], int):
            ts = torch.tensor(edge_array[:, 2].astype(np.int_), device=device)
        else:
            ts = torch.tensor(edge_array[:, 2].astype(np.double), device=device)

        index_map = IndexMap(np.unique(edge_array[:, :2]))
        edge_index = index_map.to_idxs(edge_array[:, :2].T, device=device)

        if not num_nodes:
            num_nodes = index_map.num_ids()

        return TemporalGraph(
            data=Data(
                edge_index=edge_index,
                time=ts,
                num_nodes=num_nodes,
            ),
            mapping=index_map,
        )

    @property
    def temporal_edges(self) -> list:
        """Return all temporal edges as a list of tuples (source, destination, timestamp).

        Returns:
            list: A list of tuples representing temporal edges in the format (source, destination, timestamp).

        Examples:
            Get the list of temporal edges:

            >>> g = pp.TemporalGraph.from_edge_list([('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)])
            >>> print(g.temporal_edges)
            [('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)]

            Iterate over temporal edges:
            >>> for edge in g.temporal_edges:
            >>>     print(edge)
            ('a', 'b', 1)
            ('b', 'c', 2)
            ('c', 'a', 3)
        """
        return [(*self.mapping.to_ids(e), t.item()) for e, t in zip(self.data.edge_index.t(), self.data.time)]

    def to(self, device: torch.device) -> TemporalGraph:
        """Moves all graph data to the specified device (CPU or GPU).

        Args:
            device: The target device to move the graph data to.

        Returns:
            TemporalGraph: A new TemporalGraph instance with data on the specified device.
        """
        self.data.edge_index = self.data.edge_index.to(device)
        self.data.time = self.data.time.to(device)
        for attr in self.node_attrs():
            if isinstance(self.data[attr], torch.Tensor):
                self.data[attr] = self.data[attr].to(device)
        for attr in self.edge_attrs():
            if isinstance(self.data[attr], torch.Tensor):
                self.data[attr] = self.data[attr].to(device)
        return self

    @property
    def order(self) -> int:
        """Return order 1, since all temporal graphs must be order one."""
        return 1

    def shuffle_time(self) -> None:
        """Randomly shuffle the temporal order of edges by randomly permuting timestamps."""
        self.data.time = self.data.time[torch.randperm(len(self.data.time))]

    def to_static_graph(self, weighted: bool = False, time_window: Optional[Tuple[int, int]] = None) -> Graph:
        """Return weighted time-aggregated instance of [`Graph`][pathpyG.Graph] graph.

        Args:
            weighted: whether or not to return a weighted time-aggregated graph
            time_window: A tuple with start and end time of the aggregation window

        Returns:
            Graph: A static graph object
        """
        if time_window is not None:
            idx = (self.data.time >= time_window[0]).logical_and(self.data.time < time_window[1]).nonzero().ravel()
            edge_index = self.data.edge_index[:, idx]
        else:
            edge_index = self.data.edge_index

        n = edge_index.max().item() + 1

        if weighted:
            i, w = torch_geometric.utils.coalesce(
                edge_index.as_tensor(), torch.ones(edge_index.size(1), device=self.data.edge_index.device)
            )
            return Graph(Data(edge_index=EdgeIndex(data=i, sparse_size=(n, n)), edge_weight=w), self.mapping)
        else:
            return Graph.from_edge_index(EdgeIndex(data=edge_index, sparse_size=(n, n)), self.mapping)

    def to_undirected(self) -> TemporalGraph:
        """Return an undirected version of a directed graph.

        This method transforms the current graph instance into an undirected graph by
        adding all directed edges in opposite direction. It applies [`ToUndirected`](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.transforms.ToUndirected.html#torch_geometric.transforms.ToUndirected)
        transform to the underlying [`torch_geometric.Data`](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.data.Data.html#torch_geometric.data.Data) object, which automatically
        duplicates edge attributes for newly created directed edges.

        Example:
            ```py
            import pathpyG as pp
            g = pp.TemporalGraph.from_edge_list([('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)])
            g_u = g.to_undirected()
            print(g_u)
            ```
        """
        rev_edge_index = self.data.edge_index.flip([0])
        edge_index = torch.cat([self.data.edge_index, rev_edge_index], dim=1)
        times = torch.cat([self.data.time, self.data.time])
        return TemporalGraph(data=Data(edge_index=edge_index, time=times), mapping=self.mapping)

    def get_batch(self, start_idx: int, end_idx: int) -> TemporalGraph:
        """Return an instance of the TemporalGraph that captures all time-stamped
        edges in a given batch defined by start and (non-inclusive) end, where start
        and end refer to the index of the first and last event in the time-ordered list of events."""

        return TemporalGraph(
            data=Data(edge_index=self.data.edge_index[:, start_idx:end_idx], time=self.data.time[start_idx:end_idx]),
            mapping=self.mapping,
        )

    def get_window(self, start_time: int, end_time: int) -> TemporalGraph:
        """Return an instance of the TemporalGraph that captures all time-stamped
        edges in a given time window defined by start and (non-inclusive) end, where start
        and end refer to the time stamps"""

        return TemporalGraph(data=self.data.snapshot(start_time, end_time), mapping=self.mapping)

    def __getitem__(self, key: Union[tuple, str]) -> Any:
        """Return node, edge, temporal edge, or graph attribute.

        Args:
            key: name of attribute to be returned
        """
        if not isinstance(key, tuple):
            if key in self.data.keys():
                return self.data[key]
            else:
                raise KeyError(key + " is not a graph attribute")
        elif key[0] in self.node_attrs():
            return self.data[key[0]][self.mapping.to_idx(key[1])]
        elif key[0] in self.edge_attrs():
            # TODO: Get item for non-temporal edges will only return the last occurence of the edge
            #       This is a limitation and should be fixed in the future.
            if len(key) == 3:
                return self.data[key[0]][self.edge_to_index[self.mapping.to_idx(key[1]), self.mapping.to_idx(key[2])]]
            else:
                return self.data[key[0]][self.tedge_to_index[self.mapping.to_idx(key[1]), self.mapping.to_idx(key[2]), key[3]]]
        else:
            raise KeyError(key[0] + " is not a node or edge attribute")

    def __str__(self) -> str:
        """
        Return a string representation of the graph
        """
        s = "Temporal Graph with {0} nodes, {1} unique edges and {2} events in [{3}, {4}]\n".format(
            self.data.num_nodes,
            self.data.edge_index.unique(dim=1).size(dim=1),
            self.data.edge_index.size(1),
            self.start_time,
            self.end_time,
        )

        attr = self.data.to_dict()
        attr_types = {}
        for k in attr:
            t = type(attr[k])
            if t == torch.Tensor:
                attr_types[k] = str(t) + " -> " + str(attr[k].size())
            else:
                attr_types[k] = str(t)

        from pprint import pformat

        attribute_info = {"Node Attributes": {}, "Edge Attributes": {}, "Graph Attributes": {}}
        for a in self.node_attrs():
            attribute_info["Node Attributes"][a] = attr_types[a]
        for a in self.edge_attrs():
            attribute_info["Edge Attributes"][a] = attr_types[a]
        for a in self.data.keys():
            if not self.data.is_node_attr(a) and not self.data.is_edge_attr(a):
                attribute_info["Graph Attributes"][a] = attr_types[a]
        s += pformat(attribute_info, indent=4, width=160)
        return s

order property

Return order 1, since all temporal graphs must be order one.

temporal_edges property

Return all temporal edges as a list of tuples (source, destination, timestamp).

Returns:

Name Type Description
list list

A list of tuples representing temporal edges in the format (source, destination, timestamp).

Examples:

Get the list of temporal edges:

>>> g = pp.TemporalGraph.from_edge_list([('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)])
>>> print(g.temporal_edges)
[('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)]

Iterate over temporal edges:

>>> for edge in g.temporal_edges:
>>>     print(edge)
('a', 'b', 1)
('b', 'c', 2)
('c', 'a', 3)

__getitem__

Return node, edge, temporal edge, or graph attribute.

Parameters:

Name Type Description Default
key typing.Union[tuple, str]

name of attribute to be returned

 required
Source code in src/pathpyG/core/temporal_graph.py 
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
def __getitem__(self, key: Union[tuple, str]) -> Any:
    """Return node, edge, temporal edge, or graph attribute.

    Args:
        key: name of attribute to be returned
    """
    if not isinstance(key, tuple):
        if key in self.data.keys():
            return self.data[key]
        else:
            raise KeyError(key + " is not a graph attribute")
    elif key[0] in self.node_attrs():
        return self.data[key[0]][self.mapping.to_idx(key[1])]
    elif key[0] in self.edge_attrs():
        # TODO: Get item for non-temporal edges will only return the last occurence of the edge
        #       This is a limitation and should be fixed in the future.
        if len(key) == 3:
            return self.data[key[0]][self.edge_to_index[self.mapping.to_idx(key[1]), self.mapping.to_idx(key[2])]]
        else:
            return self.data[key[0]][self.tedge_to_index[self.mapping.to_idx(key[1]), self.mapping.to_idx(key[2]), key[3]]]
    else:
        raise KeyError(key[0] + " is not a node or edge attribute")

__init__

Creates an instance of a temporal graph from a TemporalData object.

Parameters:

Name Type Description Default
data torch_geometric.data.Data

PyG Data object containing edges saved in edge_index and timestamps in time.

 required
mapping pathpyG.core.index_map.IndexMap | None

Optional mapping from node IDs to indices.

 None
Example 
from pytorch_geometric.data import TemporalData
import pathpyG as pp

d = Data(edge_index=[[0,0,1], [1,2,2]], time=[0,1,2])
t = pp.TemporalGraph(d, mapping)
print(t)
Source code in src/pathpyG/core/temporal_graph.py 
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
def __init__(self, data: Data, mapping: IndexMap | None = None) -> None:
    """Creates an instance of a temporal graph from a `TemporalData` object.

    Args:
        data: PyG `Data` object containing edges saved in `edge_index` and timestamps in `time`.
        mapping: Optional mapping from node IDs to indices.

    Example:
        ```py
        from pytorch_geometric.data import TemporalData
        import pathpyG as pp

        d = Data(edge_index=[[0,0,1], [1,2,2]], time=[0,1,2])
        t = pp.TemporalGraph(d, mapping)
        print(t)
        ```
    """
    self.data = data
    if not isinstance(self.data.edge_index, EdgeIndex):
        self.data.edge_index = EdgeIndex(
            data=self.data.edge_index.contiguous(), sparse_size=(self.data.num_nodes, self.data.num_nodes)
        )

    # reorder temporal data
    # Note that we do not use `torch_geometric.self.data.Data.sort_by_time` because it cannot sort numpy arrays`
    sorted_idx = torch.argsort(self.data.time)
    for edge_attr in set(self.data.edge_attrs()).union(set(["time"])):
        if edge_attr == "edge_index":
            self.data.edge_index = self.data.edge_index[:, sorted_idx]
        else:
            self.data[edge_attr] = self.data[edge_attr][sorted_idx]

    if mapping is not None:
        self.mapping = mapping
    else:
        self.mapping = IndexMap()

    # create mapping between edge index and edge tuples
    self.edge_to_index = {
        (e[0].item(), e[1].item()): i for i, e in enumerate(self.data.edge_index.t())
    }
    self.tedge_to_index = {
        (e[0].item(), e[1].item(), t.item()): i for i, (e, t) in enumerate(zip([e for e in self.data.edge_index.t()], self.data.time))
    }

    self.start_time = self.data.time[0].item()
    self.end_time = self.data.time[-1].item()

__str__

Return a string representation of the graph

Source code in src/pathpyG/core/temporal_graph.py 
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
def __str__(self) -> str:
    """
    Return a string representation of the graph
    """
    s = "Temporal Graph with {0} nodes, {1} unique edges and {2} events in [{3}, {4}]\n".format(
        self.data.num_nodes,
        self.data.edge_index.unique(dim=1).size(dim=1),
        self.data.edge_index.size(1),
        self.start_time,
        self.end_time,
    )

    attr = self.data.to_dict()
    attr_types = {}
    for k in attr:
        t = type(attr[k])
        if t == torch.Tensor:
            attr_types[k] = str(t) + " -> " + str(attr[k].size())
        else:
            attr_types[k] = str(t)

    from pprint import pformat

    attribute_info = {"Node Attributes": {}, "Edge Attributes": {}, "Graph Attributes": {}}
    for a in self.node_attrs():
        attribute_info["Node Attributes"][a] = attr_types[a]
    for a in self.edge_attrs():
        attribute_info["Edge Attributes"][a] = attr_types[a]
    for a in self.data.keys():
        if not self.data.is_node_attr(a) and not self.data.is_edge_attr(a):
            attribute_info["Graph Attributes"][a] = attr_types[a]
    s += pformat(attribute_info, indent=4, width=160)
    return s

from_edge_list staticmethod

Create a temporal graph from a list of tuples containing edges with timestamps.

Source code in src/pathpyG/core/temporal_graph.py 
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
@staticmethod
def from_edge_list(edge_list, num_nodes: Optional[int] = None, device: Optional[torch.device] = None) -> TemporalGraph:  # type: ignore
    """Create a temporal graph from a list of tuples containing edges with timestamps."""
    edge_array = np.array(edge_list)

    # Convert timestamps to tensor
    if isinstance(edge_list[0][2], int):
        ts = torch.tensor(edge_array[:, 2].astype(np.int_), device=device)
    else:
        ts = torch.tensor(edge_array[:, 2].astype(np.double), device=device)

    index_map = IndexMap(np.unique(edge_array[:, :2]))
    edge_index = index_map.to_idxs(edge_array[:, :2].T, device=device)

    if not num_nodes:
        num_nodes = index_map.num_ids()

    return TemporalGraph(
        data=Data(
            edge_index=edge_index,
            time=ts,
            num_nodes=num_nodes,
        ),
        mapping=index_map,
    )

get_batch

Return an instance of the TemporalGraph that captures all time-stamped edges in a given batch defined by start and (non-inclusive) end, where start and end refer to the index of the first and last event in the time-ordered list of events.

Source code in src/pathpyG/core/temporal_graph.py 
189
190
191
192
193
194
195
196
197
def get_batch(self, start_idx: int, end_idx: int) -> TemporalGraph:
    """Return an instance of the TemporalGraph that captures all time-stamped
    edges in a given batch defined by start and (non-inclusive) end, where start
    and end refer to the index of the first and last event in the time-ordered list of events."""

    return TemporalGraph(
        data=Data(edge_index=self.data.edge_index[:, start_idx:end_idx], time=self.data.time[start_idx:end_idx]),
        mapping=self.mapping,
    )

get_window

Return an instance of the TemporalGraph that captures all time-stamped edges in a given time window defined by start and (non-inclusive) end, where start and end refer to the time stamps

Source code in src/pathpyG/core/temporal_graph.py 
199
200
201
202
203
204
def get_window(self, start_time: int, end_time: int) -> TemporalGraph:
    """Return an instance of the TemporalGraph that captures all time-stamped
    edges in a given time window defined by start and (non-inclusive) end, where start
    and end refer to the time stamps"""

    return TemporalGraph(data=self.data.snapshot(start_time, end_time), mapping=self.mapping)

shuffle_time

Randomly shuffle the temporal order of edges by randomly permuting timestamps.

Source code in src/pathpyG/core/temporal_graph.py 
138
139
140
def shuffle_time(self) -> None:
    """Randomly shuffle the temporal order of edges by randomly permuting timestamps."""
    self.data.time = self.data.time[torch.randperm(len(self.data.time))]

to

Moves all graph data to the specified device (CPU or GPU).

Parameters:

Name Type Description Default
device torch.device

The target device to move the graph data to.

 required

Returns:

Name Type Description
TemporalGraph pathpyG.core.temporal_graph.TemporalGraph

A new TemporalGraph instance with data on the specified device.
Source code in src/pathpyG/core/temporal_graph.py 
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
def to(self, device: torch.device) -> TemporalGraph:
    """Moves all graph data to the specified device (CPU or GPU).

    Args:
        device: The target device to move the graph data to.

    Returns:
        TemporalGraph: A new TemporalGraph instance with data on the specified device.
    """
    self.data.edge_index = self.data.edge_index.to(device)
    self.data.time = self.data.time.to(device)
    for attr in self.node_attrs():
        if isinstance(self.data[attr], torch.Tensor):
            self.data[attr] = self.data[attr].to(device)
    for attr in self.edge_attrs():
        if isinstance(self.data[attr], torch.Tensor):
            self.data[attr] = self.data[attr].to(device)
    return self

to_static_graph

Return weighted time-aggregated instance of Graph graph.

Parameters:

Name Type Description Default
weighted bool

whether or not to return a weighted time-aggregated graph

 False
time_window typing.Optional[typing.Tuple[int, int]]

A tuple with start and end time of the aggregation window

 None

Returns:

Name Type Description
Graph pathpyG.Graph

A static graph object
Source code in src/pathpyG/core/temporal_graph.py 
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
def to_static_graph(self, weighted: bool = False, time_window: Optional[Tuple[int, int]] = None) -> Graph:
    """Return weighted time-aggregated instance of [`Graph`][pathpyG.Graph] graph.

    Args:
        weighted: whether or not to return a weighted time-aggregated graph
        time_window: A tuple with start and end time of the aggregation window

    Returns:
        Graph: A static graph object
    """
    if time_window is not None:
        idx = (self.data.time >= time_window[0]).logical_and(self.data.time < time_window[1]).nonzero().ravel()
        edge_index = self.data.edge_index[:, idx]
    else:
        edge_index = self.data.edge_index

    n = edge_index.max().item() + 1

    if weighted:
        i, w = torch_geometric.utils.coalesce(
            edge_index.as_tensor(), torch.ones(edge_index.size(1), device=self.data.edge_index.device)
        )
        return Graph(Data(edge_index=EdgeIndex(data=i, sparse_size=(n, n)), edge_weight=w), self.mapping)
    else:
        return Graph.from_edge_index(EdgeIndex(data=edge_index, sparse_size=(n, n)), self.mapping)

to_undirected

Return an undirected version of a directed graph.

This method transforms the current graph instance into an undirected graph by adding all directed edges in opposite direction. It applies ToUndirected transform to the underlying torch_geometric.Data object, which automatically duplicates edge attributes for newly created directed edges.

Example 
import pathpyG as pp
g = pp.TemporalGraph.from_edge_list([('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)])
g_u = g.to_undirected()
print(g_u)
Source code in src/pathpyG/core/temporal_graph.py 
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
def to_undirected(self) -> TemporalGraph:
    """Return an undirected version of a directed graph.

    This method transforms the current graph instance into an undirected graph by
    adding all directed edges in opposite direction. It applies [`ToUndirected`](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.transforms.ToUndirected.html#torch_geometric.transforms.ToUndirected)
    transform to the underlying [`torch_geometric.Data`](https://pytorch-geometric.readthedocs.io/en/latest/generated/torch_geometric.data.Data.html#torch_geometric.data.Data) object, which automatically
    duplicates edge attributes for newly created directed edges.

    Example:
        ```py
        import pathpyG as pp
        g = pp.TemporalGraph.from_edge_list([('a', 'b', 1), ('b', 'c', 2), ('c', 'a', 3)])
        g_u = g.to_undirected()
        print(g_u)
        ```
    """
    rev_edge_index = self.data.edge_index.flip([0])
    edge_index = torch.cat([self.data.edge_index, rev_edge_index], dim=1)
    times = torch.cat([self.data.time, self.data.time])
    return TemporalGraph(data=Data(edge_index=edge_index, time=times), mapping=self.mapping)