Merge pull request #229 from marcfranquesa/main

Fix `d_separated` deprecation warning

Author: kunwuz

causallearn/utils/DAG2PAG.py

@@ -3,7 +3,6 @@
 
 import numpy as np
 import networkx as nx
-from networkx.algorithms import d_separated
 
 from causallearn.graph.Dag import Dag
 from causallearn.graph.Edge import Edge
@@ -66,7 +65,7 @@ def dag2pag(dag: Dag, islatent: List[Node], isselection: List[Node] = []) -> Gen
             for Z in combinations(observed_nodes, l):
                 if nodex in Z or nodey in Z:
                     continue
-                if d_separated(dg, {nodes_ids[nodex]}, {nodes_ids[nodey]}, set(nodes_ids[z] for z in Z) | set([nodes_ids[s] for s in isselection])):
+                if nx.is_d_separator(dg, {nodes_ids[nodex]}, {nodes_ids[nodey]}, set(nodes_ids[z] for z in Z) | set([nodes_ids[s] for s in isselection])):
                     if edge:
                         PAG.remove_edge(edge)
                     sepset[(nodes_ids[nodex], nodes_ids[nodey])] |= set(Z)

causallearn/utils/PCUtils/Helper.py

@@ -419,7 +419,7 @@ def draw_graph(nx_graph):
 def is_dsep(nx_graph, x, y, Z):
     "Return True if x and y are d-separated by the set Z in nx_graph (networkx graph object) and False otherwise"
     S = set([str(i) for i in Z])
-    return nx.d_separated(nx_graph, {str(x)}, {str(y)}, S)
+    return nx.is_d_separator(nx_graph, {str(x)}, {str(y)}, S)
 
 
 #######################################################################################################################

causallearn/utils/cit.py

@@ -529,8 +529,8 @@ def __init__(self, data, true_dag=None, **kwargs):
     def __call__(self, X, Y, condition_set=None):
         Xs, Ys, condition_set, cache_key = self.get_formatted_XYZ_and_cachekey(X, Y, condition_set)
         if cache_key in self.pvalue_cache: return self.pvalue_cache[cache_key]
-        p = float(nx.d_separated(self.true_dag, {Xs[0]}, {Ys[0]}, set(condition_set)))
-        # pvalue is bool here: 1 if is_d_separated and 0 otherwise. So heuristic comparison-based uc_rules will not work.
+        p = float(nx.is_d_separator(self.true_dag, {Xs[0]}, {Ys[0]}, set(condition_set)))
+        # pvalue is bool here: 1 if is_d_separator and 0 otherwise. So heuristic comparison-based uc_rules will not work.
 
         # here we use networkx's d_separation implementation.
         # an alternative is to use causal-learn's own d_separation implementation in graph class: