Expand description
genomic.rs
A small crate to help implement genetical algorithms in Rust.
This crate defines two traits:
Chromosome
represents a single trainable parameter. TheChromosome
trait is implemented for a few built-in types, and you can implement it on your own.Genome
represents a set ofChromosome
s.
A few functions are also provided, notably:
mutate
, to mutate an individual’s genomecrossover
, to perform the crossover operation on two individualsreproduce
, to perform sexuated reproduction on two individuals
Implementing the Genome
trait is done in a declarative fashion:
use genomic::prelude::*;
use genomic::chromosome::UniformCh;
pub struct Triple {
first: i32,
second: i8,
third: f32,
}
impl Genome for Triple {
fn mutate(&mut self, mutator: &mut Mutator<impl rand::Rng>) {
mutator
.chromosome(&mut self.first)
.chromosome(&mut self.second)
// For floats, we need to choose a method and bounds for mutating them:
.wrap_ch(UniformCh::new(self.third, 0.0, 1.0), &mut self.third);
}
fn crossover(&mut self, other: &mut Self, crossover: &mut Crossover<impl rand::Rng>) {
crossover
.chromosome(&mut self.first, &mut other.first)
.chromosome(&mut self.second, &mut other.second)
.chromosome(&mut self.third, &mut other.third);
}
fn size_hint(&self) -> usize {
// We have three chromosomes
3
}
}
let mut triple = Triple {
first: 0,
second: 0,
third: 0.0
};
genomic::mutate(
// The genome to mutate
&mut triple,
// The mutation rate - a value of 1.0 means that the chromosomes will be fully scrambled
1.0,
// An RNG
&mut rand::thread_rng()
);
Modules
- A module to quickly import the necessary types for using this library
Functions
- Performs the crossover operation on
genome_left
andgenome_right
. Depending onmethod
, different chromosomes between the two individuals will be swapped, mixing their genetical code. - Mutates all of the chromosomes in
genome
with a rate ofrate
. - Reproduces two parent individuals, creating two children with a crossover of the parent’s chromosomes and some mutation.