Genetic Travelling Salesman

Overview

• A simple genetic algorithm for the Travelling salesman problem.
• < 100 lines of code
• Achieves reasonable performance for 25, 50, 100 cities
• Optimal solution isn't guaranteed as the algorithm can get stuck in local minima / be slow to converge, but empirically was found to work well

25 cities

• Convergence is fast

• Snapshot of each generation (with freeze on finish)

50 cities

• Convergence is fast

• Snapshot of every 10th generation (with freeze on finish)

100 cities

• Convergence is reasonable

• Snapshot of every 10th generation (with freeze on finish)