Hello, Here is a code snippet where I try to use the shortest_path function with a precomputed predecessor_map. targets = range(1,10) source = 0 distances_in, pred_map = shortest_distance(graph, source=source, target=targets, weights=graph.ep[EDGE_LEN], pred_map=True) inward_path = shortest_path(graph, source=source, target=targets[5], pred_map=pred_map) To run that code ends in: File "/usr/lib/python2.7/dist-packages/graph_tool/topology/__init__.py", line 1348, in shortest_path if pred_map[target] == int(target): # no path to target File "/usr/lib/python2.7/dist-packages/graph_tool/__init__.py", line 438, in __getitem__ return self.__map[self.__key_trans(k)] Boost.Python.ArgumentError: Python argument types in VertexPropertyMap<int64_t>.__getitem__(VertexPropertyMap<int64_t>, numpy.int64) did not match C++ signature: __getitem__(graph_tool::PythonPropertyMap<boost::checked_vector_property_map<long, boost::typed_identity_property_map<unsigned long> > > {lvalue}, graph_tool::PythonVertex) How could i circumvent that issue ? Thanks! François.