# Plot the orders
plotOrders(roads, prob.orders)
plt.title("Showing orders. Click to show path")
plt.title("Showing orders. Click to show path")
plt.show(block=False)
plt.waitforbuttonpress()
colors = ['red', 'blue', 'green', 'orange', 'grey']
plt.title("Showing paths")
totalDistance = 0
# Plot each order at a time
for i,order in enumerate(prob.orders):
# Create a sub-problem to solve with A* (getting the optimal path for each order separately)
subProblem = MapProblem(roads, order[0], order[1])
subPath, distance, _, _ = mapAstar.run(subProblem)
print("Shortest distance for order from #{} to #{}: {:.2f}km".format(order[0], order[1], distance / 1000))
totalDistance += distance
# Plot the path
plotPath(Path(roads, [s.junctionIdx for s in subPath]), color=colors[i])
plt.show(block=False)
plt.waitforbuttonpress()
print("Total distance: {:.2f}km".format(totalDistance / 1000 ))
scorer = L2DistanceCost(roads)
solver = GreedyStochasticSolver(roads, mapAstar, scorer,
Consts.STOCH_INITIAL_TEMPERATURE,
Consts.STOCH_TEMPERATURE_DECAY_FUNCTION,
Consts.STOCH_TOP_SCORES_TO_CONSIDER)
REPEATS = 200
results = [solver.solve(prob).getDistance() / 1000 for _ in range(REPEATS)]
print("Stochastic ({} repetitions): {}km".format(REPEATS, min(results)))
############ A* solver ############
# Run A* with the zero heuristic
busAstar = AStar(NullHeuristic(), cost=ActualDistanceCost(roads, mapAstar))
_, gBus, hVal, developed = busAstar.run(prob)
print(
"A* (null heuristic):\tg(G)={:.2f}km, h(I)={:.2f}km, developed: {} states".
format(gBus / 1000, hVal / 1000, developed))
# Run A* with the custom heuristic
customH = TSPCustomHeuristic(roads, prob.initialState)
busAstar = AStar(customH, cost=ActualDistanceCost(roads, mapAstar))
_, gBus, hVal, developed = busAstar.run(prob)
print(
"A* (Custom heuristic):\tg(G)={:.2f}km, h(I)={:.2f}km, developed: {} states"
.format(gBus / 1000, hVal / 1000, developed))
# Run A* with the MST heuristic
tspH = MSTHeuristic(roads, prob.initialState,
ActualDistanceCost(roads, mapAstar))
