VRPTW-gurobipy: Solving the Solomon VRPTW with direct MIP formulation using gurobipy

VRPTW stands for "Capacitated Vehicle Routing Problem with Time Windows".

Setting up

pip3 install numpy gurobipy matplotlib tabulate

Executable scripts are solve.py and plot.py. Read through README before running them.

dataset

This folder stores the Solomon dataset downloaded from http://www.vrp-rep.org/variants/item/vrptw.html, and a simple dataset representing small scenarios. The small scenarios are also solved in Excel.

result

This directory contains solutions in the format of text files. pretty* files are human-friendly and raw* files store flattened decision matrices. In visual.plot_solution(), fileutil.pretty_print() is called to generate pretty* from raw*.

You can find visualization of results in result/fig.

result/comparison.txt shows the performance of the implemented solver, compared to the SOTA results 2005.

core.py

MIP formulation and solver using gurobipy.

fileutil.py

Functions related to file operation.

plot.py

Running this script generates figures from the raw* results, updates pretty* results, shows figures of small scenarios, and saves the comparison file.

solve.py

It calls the solver to solve all problems in the dataset. solve_solomon() needs a full gurobi liscense to work properly, while solve_simple_test() does not.

SOTA2005.py

Stores the results of the state-of-the-art solutions to the Solomon problems.

visual.py

Function for visualizing results.

Reference

http://alvarestech.com/temp/vrptw/Vehicle%20Routing%20Problem%20with%20Time%20Windows.pdf