Document feedback

Question document fragment

When a question document fragment contains a formula, it is displayed as a space.

Submission type

issue

It's a little complicated...

I'd like to ask someone.

PR

Just a small problem.

I can fix it online!

Please select the submission type

Problem type

Specifications and Common Mistakes

- Specifications and Common Mistakes:

- Misspellings or punctuation mistakes,incorrect formulas, abnormal display.

- Incorrect links, empty cells, or wrong formats.

- Chinese characters in English context.

- Minor inconsistencies between the UI and descriptions.

- Low writing fluency that does not affect understanding.

- Incorrect version numbers, including software package names and version numbers on the UI.

Usability

- Usability:

- Incorrect or missing key steps.

- Missing main function descriptions, keyword explanation, necessary prerequisites, or precautions.

- Ambiguous descriptions, unclear reference, or contradictory context.

- Unclear logic, such as missing classifications, items, and steps.

Correctness

- Correctness:

- Technical principles, function descriptions, supported platforms, parameter types, or exceptions inconsistent with that of software implementation.

- Incorrect schematic or architecture diagrams.

- Incorrect commands or command parameters.

- Incorrect code.

- Commands inconsistent with the functions.

- Wrong screenshots.

- Sample code running error, or running results inconsistent with the expectation.

Risk Warnings

- Risk Warnings:

- Lack of risk warnings for operations that may damage the system or important data.

Content Compliance

- Content Compliance:

- Contents that may violate applicable laws and regulations or geo-cultural context-sensitive words and expressions.

- Copyright infringement.

Please select the type of question

Problem description

Describe the bug so that we can quickly locate the problem.

Submit

Document feedback

mindspore_gl.Graph

class mindspore_gl.Graph

Graph class.

This is the class which should be annotated in the construct function for GNNCell class. The last argument in the ‘construct’ function will be resolved into the ‘mindspore_gl.Graph’ whole graph class.

Supported Platforms:

Ascend GPU

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class SrcVertex(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"h": x})
...         return [v.h for v in g.src_vertex]
>>> ret = SrcVertex()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [1.0], [2.0], [0.0], [1.0], [2.0], [3.0], [1.0]]

adj_to_dense()

Get the dense adjacent matrix of the graph.

Note

Due to system limitations, only COO format are supported for build graph and dense format adjacency matrix can be generated.

Returns

Tensor, a tensor with shape $(N,N)$, $N$ is the number of nodes of the graph.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
...
>>> class TestAdjToDense(GNNCell):
...     def construct(self, g: Graph):
...         return g.adj_to_dense()
...
>>> ret = TestAdjToDense()(*graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[0, 1, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0],
 [1, 1, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 1, 0, 0, 0],
 [0, 0, 0, 1, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 1, 0, 1, 0, 0],
 [0, 0, 0, 0, 1, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 0],
 [0, 0, 0, 0, 0, 0, 0, 0, 3]]

avg(neigh_feat)

Aggregating node features from their neighbour and generates a node-level representation by aggregate function ‘avg’.

Parameters

neigh_feat (List[SrcVertex feature or Edge feature]) – a list of SrcVertex or Edge attributes represents the neighbour nodes or edges feature, with shape $(N,F)$, $N$ is the number of SrcVertex or Edge, $F$ is the feature dimension of the SrcVertex or Edge attribute.

Returns

• mindspore.Tensor, a tensor with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

Raises

TypeError – If neigh_feat is not a list of Edge or SrcVertex.

Examples

>>> import math
>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestAvg(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"x": x})
...         for v in g.dst_vertex:
...             v.h = g.avg([u.x for u in v.innbs])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestAvg()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> NAN = 1e9
>>> for row in ret:
...     if math.isnan(row[0]):
...         row[0] = NAN
>>> print(ret)
[[1.0], [1.0], [1000000000.0], [0.0], [1.5], [2.0], [1.0], [1000000000.0], [1.0]]

dot(feat_x, feat_y)

Dot mul operation for two node Tensors.

Parameters

• feat_x (SrcVertex feature or DstVertex feature) – the arttribute of SrcVertex or DstVertex represent feature tensor of graph nodes with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

• feat_y (SrcVertex feature or DstVertex feature) – the arttribute of SrcVertex or DstVertex represent feature tensor of graph nodes with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

Returns

• mindspore.Tensor, a tensor with shape $(N,1)$, N is the number of nodes of the graph.

Raises

• TypeError – If feat_x is not in the ‘mul’ operation support types [Tensor,Number,List,Tuple].

• TypeError – If feat_y is not in the ‘mul’ operation support types [Tensor,Number,List,Tuple].

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestDot(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"src": x, "dst": x})
...         for v in g.dst_vertex:
...             v.h = [g.dot(v.src, u.dst) for u in v.innbs]
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestDot()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[2.0], [1.0], [2.0], [2.0], [0.0], [0.0], [2.0], [0.0], [1.0], [1.0], [1.0]]

property dst_idx

A tensor with shape $(N\mathrm{\_}EDGES)$, represents the destination node index of COO edge matrix.

Returns

• mindspore.Tensor, a list of destination vertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class DstIdx(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.dst_idx
>>> ret = DstIdx()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8]

property dst_vertex

Return a list of destination vertex that only supports iterate its innbs .

Returns

• mindspore.Tensor, a list of destination vertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class DstVertex(GNNCell):
>>>     def construct(self, x, g: Graph):
>>>         g.set_vertex_attr({"h": x})
>>>         return [v.h for v in g.dst_vertex]
>>> ret = DstVertex()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [1.0], [2.0], [0.0], [1.0], [2.0], [3.0], [1.0]]

in_degree()

Get the in degree of each node in a graph.

Returns

Tensor, a tensor with shape $(N,1)$, represent the in degree of each node, $N$ is the number of nodes of the graph.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
...
>>> class TestInDegree(GNNCell):
...     def construct(self, g: Graph):
...         return g.in_degree()
...
>>> ret = TestInDegree()(*graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1], [2], [0], [1], [2], [1], [1], [0], [3]]

max(neigh_feat)

Aggregating node features from their neighbour and generates a node-level representation by aggregate function ‘max’.

Parameters

neigh_feat (List[SrcVertex feature or Edge feature]) – a list of SrcVertex or Edge attributes represents the neighbour nodes or edges feature, with shape $(N,F)$, $N$ is the number of SrcVertex or Edge, $F$ is the feature dimension of the SrcVertex or Edge attribute.

Returns

• mindspore.Tensor, a tensor with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

Raises

TypeError – If neigh_feat is not a list of Edge or SrcVertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestMax(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"x": x})
...         for v in g.dst_vertex:
...             v.h = g.max([u.x for u in v.innbs])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestMax()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [1.0], [0.0], [0.0], [2.0], [2.0], [1.0], [0.0], [1.0]]

min(neigh_feat)

Aggregating node features from their neighbour and generates a node-level representation by aggregate function ‘min’.

Parameters

neigh_feat (List[SrcVertex feature or Edge feature]) – a list of SrcVertex or Edge attributes represents the neighbour nodes or edges feature, with shape $(N,F)$, $N$ is the number of SrcVertex or Edge, $F$ is the feature dimension of the SrcVertex or Edge attribute.

Returns

• mindspore.Tensor, a tensor with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

Raises

TypeError – If neigh_feat is not a list of Edge or SrcVertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestMin(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"x": x})
...         for v in g.dst_vertex:
...             v.h = g.min([u.x for u in v.innbs])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestMin()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0], [0.0]]

property n_edges

An integer, represent the edges count of the graph.

Returns

• int, edges numbers of the graph.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class NEdge(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.n_edges
>>> ret = NEdge()(node_feat, *graph_field.get_graph())
>>> print(ret)
11

property n_nodes

An integer, represent the nodes count of the graph.

Returns

• int, nodes numbers of the graph.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class NNode(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.n_nodes
>>> ret = NNode()(node_feat, *graph_field.get_graph())
>>> print(ret)
9

out_degree()

Get the out degree of each node in a graph.

Returns

Tensor, a tensor with shape $(N,1)$, represent the out degree of each node, $N$ is the number of nodes of the graph.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
...
>>> class TestOutDegree(GNNCell):
...     def construct(self, g: Graph):
...         return g.out_degree()
...
>>> ret = TestOutDegree()(*graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1], [0], [2], [1], [1], [2], [1], [0], [3]]

set_dst_attr(feat_dict)

Set attributes for destination vetices in vertex-centric environment Keys will be attribute’s name, values will be attributes’ data.

Parameters

feat_dict (Dict) – key type: str, value type: recommend tensor of shape $(N\mathrm{\_}NODES,F)$, $F$ is the dimension of the node feature.

Returns

• mindspore.Tensor, the feature of destination vertex.

Raises

TypeError – If feat_dict is not a Dict.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestSetDstAttr(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_dst_attr({"h": x})
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestSetDstAttr()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [1.0], [2.0], [0.0], [1.0], [2.0], [3.0], [1.0]]

set_edge_attr(feat_dict)

Set attributes for edges in vertex-centric environment. Keys will be attribute’s name, values will be attributes’ data.

Parameters

feat_dict (Dict) – key type: str, value type: recommend feature tensor of shape $(N\mathrm{\_}EDGES,F)$, $F$ is the shape of the feature per edge. Recommend the shape of value is $(N\mathrm{\_}EDGES,1)$ when the feature dimension is 1.

Returns

• mindspore.Tensor, the feature of edges.

Raises

TypeError – If feat_dict is not a Dict.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> edge_feat = ms.Tensor([[1], [2], [1], [3], [1], [4], [1], [5], [1], [1], [1]], ms.float32)
...
>>> class TestSetEdgeAttr(GNNCell):
...     def construct(self, nh, eh, g: Graph):
...         g.set_vertex_attr({"nh": nh})
...         g.set_edge_attr({"eh": eh})
...         for v in g.dst_vertex:
...             v.h = g.sum([u.nh * e.eh for u, e in v.inedges])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestSetEdgeAttr()(node_feat, edge_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[2.0], [2.0], [0.0], [0.0], [14.0], [6.0], [1.0], [0.0], [3.0]]

set_graph_attr(feat_dict)

Set attributes for the whole graph in vertex-centric environment. Keys will be attribute’s name, values will be attributes’ data.

Parameters

feat_dict (Dict) – key type: str, value type: recommend feature tensor for the whole graph.

Returns

• mindspore.Tensor, the feature of graph.

Raises

TypeError – If feat_dict is not a Dict.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> g_attr = ms.Tensor([[0.0, 1.0], [0.0, 1.0], [0.0, 1.0]], ms.float32)
>>> v_attr = ms.Tensor([1.0, 1.0], ms.float32)
...
>>> class TestSetGraphAttr(GNNCell):
...     def construct(self, vh, gh, g: Graph):
...         g.set_graph_attr({"x": gh})
...         g.set_vertex_attr({"h": vh})
...         for v in g.dst_vertex:
...             v.h = g.sum([u.h * g.x for u in v.innbs])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestSetGraphAttr()(v_attr, g_attr, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[0.0, 1.0], [0.0, 2.0], [0.0, 0.0], [0.0, 0.0],
 [0.0, 0.0], [0.0, 0.0], [0.0, 0.0], [0.0, 0.0], [0.0, 0.0]]

set_src_attr(feat_dict)

Set attributes for source vertices in vertex-centric environment. Keys will be attribute’s name, values will be attributes’ data.

Parameters

feat_dict (Dict) – key type: str, value type: recommend tensor of shape $(N\mathrm{\_}NODES,F)$, $F$ is the dimension of the node feature.

Returns

• mindspore.Tensor, the feature of source vertex.

Raises

TypeError – If feat_dict is not a Dict.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestSetSrcAttr(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_src_attr({"h": x})
...         return [u.h for u in g.src_vertex]
...
>>> ret = TestSetSrcAttr()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [1.0], [2.0], [0.0], [1.0], [2.0], [3.0], [1.0]]

set_vertex_attr(feat_dict)

Set attributes for vertices in vertex-centric environment. Keys will be attribute’s name, values will be attributes’ data.

Note

set_vertex_attr is equals to set_src_attr + set_dst_attr.

Parameters

feat_dict (Dict) – key type: str, value type: recommend tensor of shape $(N\mathrm{\_}NODES,F)$, $F$ is the dimension of the node feature.

Returns

• mindspore.Tensor, the feature of vertex.

Raises

TypeError – If feat_dict is not a Dict.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestSetVertexAttr(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"h": x})
...         return [v.h for v in g.dst_vertex] * [u.h for u in g.src_vertex]
...
>>> ret = TestSetVertexAttr()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [4.0], [1.0], [4.0], [0.0], [1.0], [4.0], [9.0], [1.0]]

property src_idx

A tensor with shape $(N\mathrm{\_}EDGES)$, represents the source node index of COO edge matrix.

Returns

• mindspore.Tensor, a list of source vertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class SrcIdx(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.src_idx
>>> ret = SrcIdx()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8]

property src_vertex

Return a list of source vertex that only supports iterate with its outnbs .

Returns

• mindspore.Tensor, a list of source vertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
>>> class SrcVertex(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"h": x})
...         return [v.h for v in g.src_vertex]
>>> ret = SrcVertex()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [1.0], [2.0], [0.0], [1.0], [2.0], [3.0], [1.0]]

sum(neigh_feat)

Aggregating node features from their neighbour and generates a node-level representation by aggregate function ‘sum’.

Parameters

neigh_feat (List[SrcVertex feature or Edge feature]) – a list of SrcVertex or Edge attribute represents the neighbour nodes or edges feature, with shape $(N,F)$, $N$ is the number of SrcVertex or Edge, $F$ is the feature dimension of the SrcVertex or Edge attribute.

Returns

• mindspore.Tensor, a tensor with shape $(N,F)$, $N$ is the number of nodes of the graph, $F$ is the feature dimension of the node.

Raises

TypeError – If neigh_feat is not a list of Edge or SrcVertex.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import Graph, GraphField
>>> from mindspore_gl.nn import GNNCell
>>> n_nodes = 9
>>> n_edges = 11
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6, 8, 8, 8], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4, 8, 8, 8], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> node_feat = ms.Tensor([[1], [2], [1], [2], [0], [1], [2], [3], [1]], ms.float32)
...
>>> class TestSum(GNNCell):
...     def construct(self, x, g: Graph):
...         g.set_vertex_attr({"x": x})
...         for v in g.dst_vertex:
...             v.h = g.sum([u.x for u in v.innbs])
...         return [v.h for v in g.dst_vertex]
...
>>> ret = TestSum()(node_feat, *graph_field.get_graph()).asnumpy().tolist()
>>> print(ret)
[[1.0], [2.0], [0.0], [0.0], [3.0], [2.0], [1.0], [0.0], [3.0]]

topk_edges(node_feat, k, sortby=None)

Return a graph-level representation by a graph-wise top-k on node features.

If sortby is set to None, the function would perform top-k on all dimensions independently.

Parameters

• node_feat (Tensor) – A tensor represent the node feature, with shape $(N\mathrm{\_}NODES,F)$. $F$ is the dimension of the node feature.

• k (int) – Represent how many nodes for top-k.

• sortby (int) – Sort according to which feature. If is None, all features are sorted independently. Default is None.

Note

The value participated in the sort by axis (all value if sortby is None) should be greater than zero. Due to the reason that we create zero value for padding and they may cover the features.

Returns

• topk_output (Tensor) - a tensor with shape $(B,K,F)$, where $B$ is the batch size of the input graph. $K$ is the input ‘k’, $F$ is the feature size.

• topk_indices (Tensor), - a tensor with shape $(B,K)$ ( $(B,K,F)$ if sortby is set to None), where $B$ is the batch size of the input graph, $F$ is the feature size.

Raises

• TypeError – If node_feat is not a Tensor.

• TypeError – If k is not an int.

• ValueError – If sortby is not an int.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import GraphField, Graph
>>> from mindspore_gl.nn import GNNCell
>>> node_feat = ms.Tensor([
...     [1, 2, 3, 4],
...     [2, 4, 1, 3],
...     [1, 3, 2, 4],
...     [9, 7, 5, 8],
...     [8, 7, 6, 5],
...     [8, 6, 4, 6],
...     [1, 2, 1, 1],
... ], ms.float32)
...
>>> n_nodes = 7
>>> n_edges = 8
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
...
>>> class TestTopkEdges(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.topk_edges(x, 2, 1)
...
>>> output, indices = TestTopkEdges()(node_feat, *graph_field.get_graph())
>>> output = output.asnumpy().tolist()
>>> indices = indices.asnumpy().tolist()
>>> print(output)
[[9.0, 7.0, 5.0, 8.0], [8.0, 7.0, 6.0, 5.0]]
>>> print(indices)
[3, 4]

topk_nodes(node_feat, k, sortby=None)

Return a graph-level representation by a graph-wise top-k on node features.

If sortby is set to None, the function would perform top-k on all dimensions independently.

Note

The value participated in the sort by axis (all value if sortby is None) should be greater than zero. Due to the reason that we create zero value for padding and they may cover the features.

Parameters

• node_feat (Tensor) – A tensor represent the node feature, with shape $(N\mathrm{\_}NODES,F)$. $F$ is the dimension of the node feature.

• k (int) – Represent how many nodes for top-k.

• sortby (int) – Sort according to which feature. If is None, all features are sorted independently. Default is None.

Returns

• topk_output (Tensor) - a tensor with shape $(B,K,F)$, where $B$ is the batch size of the input graph. $K$ is the input ‘k’, $F$ is the feature size.

• topk_indices (Tensor), - a tensor with shape $(B,K)$ ( $(B,K,F)$ if sortby is set to None), where $B$ is the batch size of the input graph, $F$ is the feature size.

Raises

• TypeError – If node_feat is not a Tensor.

• TypeError – If k is not an int.

• ValueError – If sortby is not an int.

Examples

>>> import mindspore as ms
>>> from mindspore_gl import GraphField, Graph
>>> from mindspore_gl.nn import GNNCell
>>> node_feat = ms.Tensor([
...     [1, 2, 3, 4],
...     [2, 4, 1, 3],
...     [1, 3, 2, 4],
...     [9, 7, 5, 8],
...     [8, 7, 6, 5],
...     [8, 6, 4, 6],
...     [1, 2, 1, 1],
... ], ms.float32)
...
>>> n_nodes = 7
>>> n_edges = 8
>>> src_idx = ms.Tensor([0, 2, 2, 3, 4, 5, 5, 6], ms.int32)
>>> dst_idx = ms.Tensor([1, 0, 1, 5, 3, 4, 6, 4], ms.int32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
...
>>> class TestTopkNodes(GNNCell):
...     def construct(self, x, g: Graph):
...         return g.topk_nodes(x, 2, 1)
...
>>> output, indices = TestTopkNodes()(node_feat, *graph_field.get_graph())
>>> output = output.asnumpy().tolist()
>>> indices = indices.asnumpy().tolist()
>>> print(output)
[[9.0, 7.0, 5.0, 8.0], [8.0, 7.0, 6.0, 5.0]]
>>> print(indices)
[3, 4]