License

The VRPSolverEasy package itself is open-source and free to use. It includes compiled libraries of BaPCod, its VRPSolver extension, and COIN-OR CLP solver. These libraries are also free to use.

For better performance, it is possible to use VRPSolverEasy together with CPLEX MIP solver. This combination called academic version requires an access to the source code of BaPCod available with an academic-use-only license. The academic version of VRPSolverEasy additionally includes a MIP-based (slow) heuristic which is useful for finding feasible solutions in the absence of an external heuristic solver.

Accompanying paper

The paper presents the motivation to create VRPSolverEasy, the interface of the package, the solution approach (optional to read), the computational results for the three classic VRP variants (CVRP, VRPTW, HFVRP), and possible future extensions of the model. For the moment, the paper is available as a preprint :

N. Errami, E. Queiroga, R. Sadykov, E. Uchoa. “VRPSolverEasy: a Python library for the exact solution of a rich vehicle routing problem”, Technical report HAL-04057985, 2023.

Please cite it if you use VRPSolverEasy in your research.

Installation

VRPSolverEasy requires a version of python >= 3.6

There is two different way to install VRPSolverEasy :

The first way is to install it with pip:

python -m pip install VRPSolverEasy

The second way is to follow these steps:

• Download the package and extract it into a local directory

• Move to this local directory and enter :

  python pip install .

Installation instructions for Mac computers with Apple ARM processors, as well as for the academic version, are given in the documentation.

Copy binaries

Once the package is installed you will need to request the Bapcod distribution here: https://bapcod.math.u-bordeaux.fr/ Once you have downloaded the distribution. You just have to go to the VRPSolverEasy folder and copy the system folder corresponding to your computer and copy it into the lib folder of the VRPSolverEasy python package. For example if your computer is a Mac you will copy and replace the Darwin folder, you will then have VRPSolverEasy/lib/Darwin.

Example

A simple example that shows how to use the VRPSolverEasy package:

import VRPSolverEasy as vrpse
import math

def compute_euclidean_distance(x_i, x_j, y_i, y_j):
  """compute the euclidean distance between 2 points from graph"""
     return round(math.sqrt((x_i - x_j)**2 +
                            (y_i - y_j)**2), 3)

# Data
cost_per_distance = 10
begin_time = 0
end_time = 5000
nb_point = 7

# Map with names and coordinates
coordinates = {"Wisconsin, USA": (44.50, -89.50),  # depot
               "West Virginia, USA": (39.000000, -80.500000),
               "Vermont, USA": (44.000000, -72.699997),
               "Texas, the USA": (31.000000, -100.000000),
               "South Dakota, the US": (44.500000, -100.000000),
               "Rhode Island, the US": (41.742325, -71.742332),
               "Oregon, the US": (44.000000, -120.500000)
               }

# Demands of points
demands = [0, 500, 300, 600, 658, 741, 436]

# Initialisation
model = vrpse.Model()

# Add vehicle type
model.add_vehicle_type(
    id=1,
    start_point_id=0,
    end_point_id=0,
    name="VEH1",
    capacity=1100,
    max_number=6,
    var_cost_dist=cost_per_distance,
    tw_end=5000)

# Add depot
model.add_depot(id=0, name="D1", tw_begin=0, tw_end=5000)

coordinates_keys = list(coordinates.keys())
# Add customers
for i in range(1, nb_point):
    model.add_customer(
        id=i,
        name=coordinates_keys[i],
        demand=demands[i],
        tw_begin=begin_time,
        tw_end=end_time)

# Add links
coordinates_values = list(coordinates.values())
for i in range(0, 7):
    for j in range(i + 1, 7):
        dist = compute_euclidean_distance(coordinates_values[i][0],
                                          coordinates_values[j][0],
                                          coordinates_values[i][1],
                                          coordinates_values[j][1])
        model.add_link(
            start_point_id=i,
            end_point_id=j,
            distance=dist,
            time=dist)

# solve model
model.solve()
model.export()

if model.solution.is_defined():
    print(model.solution)

Documentation

Documentation, explanation of demos (CVRP, VRPTW, HFVRP, and MDVRP), and the solver API are accessible here: https://vrpsolvereasy.readthedocs.io/en/latest/.

You can also build the documentation locally by following this instructions from the source folder :

cd docs
python -m pip install -r requirements.txt
cd ..
make html

The HTML pages will be in the folder build\html.