def solve(
instance: CVRPInstance,
params: Optional[ORToolsParams] = None,
) -> Optional[CVRPSolution]:
"""Solves a CVRP instance using ORTools"""
# Initialize parameters if not provided.
params = params or ORToolsParams.get_baseline()
# Number of points is the number of deliveries + the origin.
num_points = len(instance.deliveries) + 1
logger.info(f"Solving CVRP instance of size {num_points}.")
# There's no limit of vehicles, or max(vehicles) = len(deliveries).
num_vehicles = params.max_vehicles or len(instance.deliveries)
manager = pywrapcp.RoutingIndexManager(
num_points,
num_vehicles,
0, # (Number of nodes, Number of vehicles, Origin index).
)
model = pywrapcp.RoutingModel(manager)
# Unwrap the size index for every point.
sizes = np.array([0] + [d.size for d in instance.deliveries],
dtype=np.int32)
def capacity_callback(src):
src = manager.IndexToNode(src)
return sizes[src]
capacity_callback_index = model.RegisterUnaryTransitCallback(
capacity_callback)
model.AddDimension(capacity_callback_index, 0, instance.vehicle_capacity,
True, "Capacity")
# Unwrap the location/point for every point.
locations = [instance.origin] + [d.point for d in instance.deliveries]
# Compute the distance matrix between points.
logger.info("Computing distance matrix.")
distance_matrix = (calculate_distance_matrix_m(locations) * 10).astype(
np.int32)
def distance_callback(src, dst):
x = manager.IndexToNode(src)
y = manager.IndexToNode(dst)
return distance_matrix[x, y]
distance_callback_index = model.RegisterTransitCallback(distance_callback)
model.SetArcCostEvaluatorOfAllVehicles(distance_callback_index)
search_parameters = pywrapcp.DefaultRoutingSearchParameters()
search_parameters.first_solution_strategy = params.first_solution_strategy
search_parameters.local_search_metaheuristic = (
params.local_search_metaheuristic)
if params.solution_limit:
search_parameters.solution_limit = params.solution_limit
search_parameters.time_limit.FromTimedelta(
timedelta(microseconds=1e3 * params.time_limit_ms))
logger.info("Solving CVRP with ORTools.")
assignment = model.SolveWithParameters(search_parameters)
# Checking if the feasible solution was found.
# For more information about the type error:
# https://developers.google.com/optimization/routing/routing_options
if not assignment:
return None
def extract_solution(vehicle_id):
# Get the start node for route.
index = model.Start(vehicle_id)
# Iterate while we don't reach an end node.
while not model.IsEnd(assignment.Value(model.NextVar(index))):
next_index = assignment.Value(model.NextVar(index))
node = manager.IndexToNode(next_index)
yield instance.deliveries[node - 1]
index = next_index
routes = [
CVRPSolutionVehicle(
origin=instance.origin,
deliveries=list(extract_solution(i)),
) for i in range(num_vehicles)
]
# Return only routes that actually leave the depot.
return CVRPSolution(
name=instance.name,
vehicles=[v for v in routes if len(v.deliveries)],
)
def to_tsplib(instance: CVRPInstance,
file_name: Optional[str] = None) -> Optional[str]:
"""Convert instance into TSPLIB format
Parameters
----------
instance
CVRP instance
file_name
File with extension ".vrp" to save the instance. If not provided,
return a string with the same data
Returns
-------
A string with instance data in the TSPLIB format if no ``file_name`` was
especified, or nothing otherwise
"""
# Header information
tspfile = (
f"NAME : {instance.name}\n"
"TYPE : ACVRP\n"
f"DIMENSION : {len(instance.deliveries) + 1}\n" # + 1 for the origin
f"CAPACITY : {instance.vehicle_capacity}\n"
"NODE_COORD_TYPE : TWOD_COORDS\n"
"EDGE_WEIGHT_TYPE : EXPLICIT\n"
"EDGE_WEIGHT_FORMAT : FULL_MATRIX\n")
# Nodes section
tspfile += (
"NODE_COORD_SECTION\n"
f"1 {instance.origin.lng} {instance.origin.lat}\n" # origin
)
tspfile += "\n".join(
f"{i} {delivery.point.lng} {delivery.point.lat}"
for i, delivery in enumerate(instance.deliveries, start=2))
tspfile += "\n"
# Demand section
tspfile += ("DEMAND_SECTION\n" "1 0\n")
tspfile += "\n".join(
f"{i} {delivery.size}"
for i, delivery in enumerate(instance.deliveries, start=2))
tspfile += "\n"
# Depot section: ensure node 1 is the depot (-1 to terminate the list)
tspfile += ("DEPOT_SECTION\n" "1\n" "-1\n")
# Edge section:
# Compute distance matrix
locations = [instance.origin
] + [delivery.point for delivery in instance.deliveries]
distance_matrix = (calculate_distance_matrix_m(locations) * 10).astype(
np.int)
tspfile += "EDGE_WEIGHT_SECTION\n"
def print_row(row):
return " ".join(str(el) for el in row)
tspfile += "\n".join(print_row(row) for row in distance_matrix)
if not file_name:
return tspfile
with open(file_name, "w") as f:
f.write(tspfile)