https://github.com/phoenixsmaug/tsp

A comparison of various solvers for the Travelling Salesman Problem.
https://github.com/phoenixsmaug/tsp

backtracking branch-and-bound integer-programming tsp tsp-solver

Last synced: about 1 month ago
JSON representation

A comparison of various solvers for the Travelling Salesman Problem.

• Host: GitHub
• URL: https://github.com/phoenixsmaug/tsp
• Owner: PhoenixSmaug
• License: mit
• Created: 2025-04-25T11:59:25.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2026-01-26T13:42:00.000Z (2 months ago)
• Last Synced: 2026-01-27T02:56:27.223Z (about 2 months ago)
• Topics: backtracking, branch-and-bound, integer-programming, tsp, tsp-solver
• Language: Julia
• Homepage:
• Size: 12.7 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Various solvers for the Travelling Salesman Problem

The [Travelling Salesman Problem (TSP)](https://en.m.wikipedia.org/wiki/Travelling_salesman_problem) is a famously hard combinatorial optimization problem. Here two solvers are implemented:

* An efficient solver, which uses the [Miller-Tucker-Zemlin Encoding](https://phabe.ch/2021/09/19/tsp-subtour-elimination-by-miller-tucker-zemlin-constraint/) to translate TSP into an Integer Programming problem and solves that with the commercial state-of-the-art solver Gurobi

* A standalone solver which implements the [Held-Karp algorithm](https://en.wikipedia.org/wiki/Held%E2%80%93Karp_algorithm) to solve TSP with dynamic programming

* A simple branch-and-bound solver for educational purposes, who can prune backtracking branches as soon as their lower bound is bigger than the current maximum

Using `benchmark()`, one can compare the performances on the [TSPLIB95 Dataset](http://comopt.ifi.uni-heidelberg.de/software/TSPLIB95/). `benchmark.log` also provides pre-recorded results measured with an Intel i9-10980HK and 32 GB of memory.

(c) Mia Müßig