Introduction
As part of a project supervised by Claude Bernard University, I developed a program for optimizing delivery vehicle routes in collaboration with Pablo Verchère, Mathieu Ponton and Jules Ginhac.

Features

• Pseudo-random initial solution generation


• Transformation functions: relocate, exchange and 2-OPT


• Functions for calculating the gain generated by each transformation


• Function for generating the best neighbor of a route (shortest route transformed once)


• Function for finding an approximation to the best route


• All of the above, but for several routes simultaneously (each route corresponds to a truck with limited capacity)

Improvement of a journey

Results for multiple journeys

Results for multiple journeys with more points