Error with Solving Pick-Delivery Problem - source Source or target Sink is not in G

I am having trouble solving the following pickup-delivery problem. I keep getting an error saying ‘Source’ or ‘Sink’ node missing, even though I added those in the graph. Does this problem require the use of only one vehicle at a time? As you will see the source for all three deliveries is the same. Thanks.

G = DiGraph()
G.add_edge(1, 2, cost=77.67)
G.add_edge(1, 3, cost=0.0)
G.add_edge(1, 4, cost=96.61)
G.add_edge(1, 5, cost=0.0)
G.add_edge(1, 6, cost=59.03)

G.add_edge(2, 1, cost=77.67)
G.add_edge(2, 3, cost=77.67)
G.add_edge(2, 4, cost=64.85)
G.add_edge(2, 5, cost=77.67)
G.add_edge(2, 6, cost=62.2)

G.add_edge(3, 1, cost=0.0)
G.add_edge(3, 2, cost=77.67)
G.add_edge(3, 4, cost=96.61)
G.add_edge(3, 5, cost=0.0)
G.add_edge(3, 6, cost=59.03)

G.add_edge(4, 1, cost=96.61)
G.add_edge(4, 2, cost=64.85)
G.add_edge(4, 3, cost=96.61)
G.add_edge(4, 5, cost=96.61)
G.add_edge(4, 6, cost=39.82)

G.add_edge(5, 1, cost=0.0)
G.add_edge(5, 2, cost=77.67)
G.add_edge(5, 3, cost=0.0)
G.add_edge(5, 4, cost=96.61)
G.add_edge(5, 6, cost=59.03)

G.add_edge(6, 1, cost=59.03)
G.add_edge(6, 2, cost=62.2)
G.add_edge(6, 3, cost=59.03)
G.add_edge(6, 4, cost=39.82)
G.add_edge(6,5, cost=59.03)

G.add_edge("Source", 1, cost=18.03)
G.add_edge(1, "Sink", cost=18.93)
G.add_edge("Source", 2, cost=61.29)
G.add_edge(2, "Sink", cost=61.29)
G.add_edge("Source",3, cost=18.03)
G.add_edge(3,"Sink", cost=18.03)
G.add_edge("Source", 4, cost=79.92)
G.add_edge(4, "Sink", cost=79.92)
G.add_edge("Source", 5, cost=18.03)
G.add_edge(5, "Sink", cost=18.03)
G.add_edge("Source", 6, cost=44.38)
G.add_edge(6,"Sink", cost=44.38)

G.nodes[1]["request"] = 2
G.nodes[1]["demand"] = 25000
G.nodes[2]["demand"] = -25000

G.nodes[3]["request"] = 4
G.nodes[3]["demand"] = 25000
G.nodes[4]["demand"] = -25000

G.nodes[5]["request"] = 6
G.nodes[5]["demand"] = 10000
G.nodes[6]["demand"] = -10000

v_prob = VehicleRoutingProblem(G, load_capacity = 25000)
v_prob.pickup_delivery = True
v_prob.solve(cspy=False)

The following error is raised:

Traceback (most recent call last):
  File "pdp.py", line 96, in <module>
    v_prob.solve(cspy=False)
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 238, in solve
    self._solve(dive, solver)
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 454, in _solve
    self._column_generation()
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 476, in _column_generation
    self._find_columns()
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 518, in _find_columns
    self._more_routes = self._solve_subproblem_with_heuristic(
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 583, in _solve_subproblem_with_heuristic
    more_columns = self._attempt_solve_best_edges1(vehicle=vehicle,
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\vrp.py", line 625, in _attempt_solve_best_edges1
    self.routes, self._more_routes = subproblem.solve(
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\subproblem_lp.py", line 47, in solve
    self._add_new_route()
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\vrpy\subproblem_lp.py", line 70, in _add_new_route
    "new route %s %s" % (route_id, shortest_path(new_route, "Source", "Sink"))
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\networkx\algorithms\shortest_paths\generic.py", line 160, in shortest_path
    paths = nx.bidirectional_shortest_path(G, source, target)
  File "C:\Users\romain.montagne\AppData\Local\Programs\Python\Python38\lib\site-packages\networkx\algorithms\shortest_paths\unweighted.py", line 221, in bidirectional_shortest_path
    raise nx.NodeNotFound(msg)
networkx.exception.NodeNotFound: Either source Source or target Sink is not in G

[Kuifje02]

@torressa I understand why the error is raised. One of the subproblems is infeasible, and solvers don’t all handle infeasibilities the same way. CBC (the default solver) outputs a value (which is meaningless) while CPLEX (and GUROBI ?) outputs None.

Therefore the following conditions are not strict enough, I will update them:

https://github.com/Kuifje02/vrpy/blob/88d9ea477c2a8d357d7d5a4950536a55d399595d/vrpy/subproblem_lp.py#L42-L45

[ugirumurera]

Ok, thank you both! I will try with CPLEX.