Merge branch 'branch-24.10' into pyg-neg-sampling

python/nx-cugraph/nx_cugraph/classes/graph.py

@@ -689,6 +689,14 @@ def _get_plc_graph(
             src_indices = src_indices.astype(index_dtype)
             dst_indices = dst_indices.astype(index_dtype)
 
+        # This sets drop_multi_edges=True for non-multigraph input, which means
+        # the data in self.src_indices and self.dst_indices may not be
+        # identical to that contained in the returned pcl.SGGraph (the returned
+        # SGGraph may have fewer edges since duplicates are dropped). Ideally
+        # self.src_indices and self.dst_indices would be updated to have
+        # duplicate edges removed for non-multigraph instances, but that
+        # requires additional code which would be redundant and likely not as
+        # performant as the code in PLC.
         return plc.SGGraph(
             resource_handle=plc.ResourceHandle(),
             graph_properties=plc.GraphProperties(
@@ -702,6 +710,7 @@ def _get_plc_graph(
             renumber=False,
             do_expensive_check=False,
             vertices_array=self._node_ids,
+            drop_multi_edges=not self.is_multigraph(),
         )
 
     def _sort_edge_indices(self, primary="src"):

python/nx-cugraph/nx_cugraph/tests/test_pagerank.py

@@ -0,0 +1,36 @@
+# Copyright (c) 2024, NVIDIA CORPORATION.
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import networkx as nx
+import pandas as pd
+from pytest import approx
+
+
+def test_pagerank_multigraph():
+    """
+    Ensures correct differences between pagerank results for Graphs
+    vs. MultiGraphs generated using from_pandas_edgelist()
+    """
+    df = pd.DataFrame({"source": [0, 1, 1, 1, 1, 1, 1, 2],
+                       "target": [1, 2, 2, 2, 2, 2, 2, 3]})
+    expected_pr_for_G = nx.pagerank(nx.from_pandas_edgelist(df))
+    expected_pr_for_MultiG = nx.pagerank(
+        nx.from_pandas_edgelist(df, create_using=nx.MultiGraph))
+
+    G = nx.from_pandas_edgelist(df, backend="cugraph")
+    actual_pr_for_G = nx.pagerank(G, backend="cugraph")
+
+    MultiG = nx.from_pandas_edgelist(df, create_using=nx.MultiGraph, backend="cugraph")
+    actual_pr_for_MultiG = nx.pagerank(MultiG, backend="cugraph")
+
+    assert actual_pr_for_G == approx(expected_pr_for_G)
+    assert actual_pr_for_MultiG == approx(expected_pr_for_MultiG)