Adds Cycle Motif Support

CHANGELOG.md

@@ -5,6 +5,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ## [2.3.0] - 2023-MM-DD
 ### Added
+- Added `CycleMotif` motif generator to generate `n`-node cycle shaped motifs ([#6256](https://github.com/pyg-team/pytorch_geometric/pull/6256))
 - Added the `InfectionDataset` to evaluate explanations ([#6222](https://github.com/pyg-team/pytorch_geometric/pull/6222))
 - Added `characterization_score` and `fidelity_curve_auc` explainer metrics ([#6188](https://github.com/pyg-team/pytorch_geometric/pull/6188))
 - Added `get_message_passing_embeddings` ([#6201](https://github.com/pyg-team/pytorch_geometric/pull/6201))

test/datasets/motif_generator/test_cycle_motif.py

@@ -0,0 +1,15 @@
+from torch_geometric.datasets.motif_generator import CycleMotif
+
+
+def test_cycle_motif():
+    motif_generator = CycleMotif(5)
+    assert str(motif_generator) == 'CycleMotif(5)'
+
+    motif = motif_generator()
+    assert len(motif) == 2
+    assert motif.num_nodes == 5
+    assert motif.num_edges == 10
+    assert motif.edge_index.tolist() == [
+        [0, 0, 1, 1, 2, 2, 3, 3, 4, 4],
+        [1, 4, 0, 2, 1, 3, 2, 4, 0, 3],
+    ]

torch_geometric/datasets/motif_generator/__init__.py

@@ -1,9 +1,11 @@
 from .base import MotifGenerator
 from .custom import CustomMotif
 from .house import HouseMotif
+from .cycle import CycleMotif
 
 __all__ = classes = [
     'MotifGenerator',
     'CustomMotif',
     'HouseMotif',
+    'CycleMotif',
 ]

torch_geometric/datasets/motif_generator/cycle.py

@@ -0,0 +1,31 @@
+import torch
+
+from torch_geometric.data import Data
+from torch_geometric.datasets.motif_generator import CustomMotif
+
+
+class CycleMotif(CustomMotif):
+    r"""Generates the cycle motif from the `"GNNExplainer:
+    Generating Explanations for Graph Neural Networks"
+    <https://arxiv.org/pdf/1903.03894.pdf>`_ paper.
+
+    Args:
+        num_nodes (int): The number of nodes in the cycle.
+    """
+    def __init__(self, num_nodes: int):
+        self.num_nodes = num_nodes
+
+        row = torch.arange(num_nodes).view(-1, 1).repeat(1, 2).view(-1)
+        col1 = torch.arange(-1, num_nodes - 1) % num_nodes
+        col2 = torch.arange(1, num_nodes + 1) % num_nodes
+        col = torch.stack([col1, col2], dim=1).sort(dim=-1)[0].view(-1)
+
+        structure = Data(
+            num_nodes=num_nodes,
+            edge_index=torch.stack([row, col], dim=0),
+        )
+
+        super().__init__(structure)
+
+    def __repr__(self) -> str:
+        return f'{self.__class__.__name__}({self.num_nodes})'