Get rid of KeyErrors in to_line_graph

tests/convert/test_line_graph.py

@@ -34,7 +34,14 @@ def test_to_line_graph(edgelist1, hypergraph1):
 
     assert L.number_of_nodes() == 0
     assert L.number_of_edges() == 0
+
+def test_557():
+    H = xgi.Hypergraph({0: [(6.019592999999995, 47.2350647), (6.019592999999853, 47.2350647), (6.01959299999999, 47.2350647), (6.019592999999993, 47.2350647), (6.0195929999999755, 47.2350647)], 
+                        1: [(6.019592999999995, 47.2350647), (6.019592999999853, 47.2350647), (6.01959299999999, 47.2350647), (6.019592999999993, 47.2350647), (6.0195929999999755, 47.2350647)]})
+    L = xgi.to_line_graph(H)
 
+    assert L.number_of_nodes() == 2
+    assert L.number_of_edges() == 1
 
 def test_abs_weighted_line_graph(edgelist1, hypergraph1, hypergraph2):
     H = xgi.Hypergraph(edgelist1)

xgi/algorithms/centrality.py

@@ -58,7 +58,7 @@ def clique_eigenvector_centrality(H, tol=1e-6):
     if not is_connected(H):
         return {n: np.nan for n in H.nodes}
     W, node_dict = clique_motif_matrix(H, index=True)
-    _, v = eigsh(W.asfptype(), k=1, which="LM", tol=tol)
+    _, v = eigsh(W.astype(float), k=1, which="LM", tol=tol)
 
     # multiply by the sign to try and enforce positivity
     v = np.sign(v[0]) * v / norm(v, 1)

xgi/convert/line_graph.py

@@ -53,28 +53,25 @@ def to_line_graph(H, s=1, weights=None):
         )
     LG = nx.Graph()
 
-    edge_label_dict = {tuple(edge): index for index, edge in H._edge.items()}
+    LG.add_nodes_from([(k, {'original_hyperedge':v}) for k,v in H._edge.items()])
 
-    LG.add_nodes_from(H.edges)
-
-    for edge1, edge2 in combinations(H.edges.members(), 2):
+    for e1, e2 in combinations(H._edge, 2):
         # Check that the intersection size is larger than s
-        intersection_size = len(edge1.intersection(edge2))
+        intersection_size = len(H._edge[e1].intersection(H._edge[e2]))
         if intersection_size >= s:
             if not weights:
                 # Add unweighted edge
                 LG.add_edge(
-                    edge_label_dict[tuple(edge1)], edge_label_dict[tuple(edge2)]
+                    e1, e2
                 )
             else:
                 # Compute the (normalized) weight
                 weight = intersection_size
                 if weights == "normalized":
-                    weight /= min([len(edge1), len(edge2)])
+                    weight /= min([len(H._edge[e1]), len(H._edge[e2])])
                 # Add edge with weight
                 LG.add_edge(
-                    edge_label_dict[tuple(edge1)],
-                    edge_label_dict[tuple(edge2)],
+                    e1, e2,
                     weight=weight,
                 )