mithi / rusty-genes

Genetic algorithm implementation in Rust with animated visualizations in Python

Rusty Genes

• A Rust implementation of a genetic algorithm to solve the traveling salesman problem with animated visualizations
• See my Medium article for a more detailed discussion
• Check ./citydna/README.md on how to use

$ curl https://sh.rustup.rs -sSf | sh
$ cd citydna
$ cargo test -- --nocapture
$ cargo build
$ cargo run ./data/specs/specs1.csv ./data/cities/cities0.csv
$ cargo run ./data/specs/specsABC.csv ./data/cities/citiesA.csv > ./NEW_OUTPUT.csv

REFERENCES

• Genetic Algo Tech Talk Video
• Genetic Algo Tech Talk Repo
• Coding Train Video
• Coding Train Repo

🐳 ☕️ 🧧

33Mudy961bUk9zz35p68g9fE3uuHLRduRp

rusty-genes's Issues

Clean up git history (squash some commits)

Update License to include author's name

Swap Population Creation and Challenger Selection in run()

https://github.com/mithi/rusty-genes/blob/rusty-genes/citydna/src/simulation.rs

        assert!(skip > 0, "skip must be 1 or larger");

        let mut population = random_population(self.population_size, &self.cities);
        let mut champion = find_fittest(&population);

        for i in 0..self.iterations {

            let challenger = find_fittest(&population);
            population = self.generate_population(population);
            debug_print(debug_level, skip, i + 1, &population, &champion, &challenger, self.number_of_cities);

            if champion.fitness <= challenger.fitness {
                champion = challenger;
            }
        }

        self.fitness = champion.fitness;
        self.dna = champion.dna;

        if debug_level >= 2 { self.print(); }
    }

You may want to swap let challenger = find_fittest(&population); and population = self.generate_population(population);

In that way you are getting the challenger from the newest generated population. As written, in the last iteration, you would be looking at the challenger from the previous iteration.