Crate evolutionary

Evolutionary

A fully extensible Rust framework for using paralyzed genetic algorithms to solve problems.

Currently, it supports coding in Bin, Real, Permuted Integers, Integers and any other coding you may want to implement. It also has built in implementation of the following genetic operators:

• Selection
• Crossover
• Mutation

You can code your own selection, crossover or mutation implementing the traits and passing them to the EvolutionBuilder.

Example:

use evolutionary::prelude::*;

// First you'll need to code your Fitness function:
#[derive(Clone)]
pub struct MaxFitness;

// To do this you need to implement the Fitness trait.
impl Fitness<Bin> for MaxFitness {
    fn calculate_fitness(&self, individual: &Bin) -> f64 {
        let mut sum = 0.;

        for i in 0..individual.get_chromosome().len() {
            if individual.get_gene(i) {
                sum += 1.;
            }
        }

        sum
    }
}

// Then you will be able to build a evolution object using the `EvolutionBuiler`
// and setting all the required parameters:
fn main() {
    let mut evolution = EvolutionBuilder::new(30, 10, GeneCod::Bin, ())
        .with_fitness(MaxFitness)
        .with_selection(TournamentSelection::default())
        .with_crossover(NPointsCrossover::default())
        .with_mutation(BitFlipMutation::default())
        .with_title("Max".to_string())
        .with_stop_condition(move |_, iterations, _| iterations >= 100)
        .build().unwrap();

    evolution.run();

    assert_eq!(*evolution.current_best().get_chromosome(), vec![true; 10]);
    assert_eq!(evolution.current_best_fitness(), 10.0);
}

Find this and other examples in the examples folder.

Modules

• config_read
• experiment_runner
• metrics
• plotting
• prelude
• utils

Structs

• Evolution
  This is the struct that does the evolution magic. It has the methods needed to start and iterate through the evolution.
• EvolutionBuilder
  This is the helper struct to create a new Evolution object. The fitness, selection, crossover, mutation and stop_condition are required.

Traits

• Crossover
  Trait that defines the crossover method. You can implement your own crossover method by implementing this trait.
• Fitness
  Trait that defines the fitness function and that will be used by the evolutionary algorithm. You need to implement your own fitness function for each problem you’re trying to solve. The fitness function is directly related to the type of individual you’re using and how your mapping the answers.
• Individual
  Trait that must be implemented by a struct to be considered a individual.
• Mutation
  Mutation Trait
• Selection
  Trait that defines the selection method. You can implement your own selection method by implementing this trait.