radiate_core/

alter.rs

use std::iter::once;

use crate::stats::ToSnakeCase;
use crate::{Chromosome, Gene, Genotype, Metric, Population, indexes, intern, random_provider};

#[macro_export]
macro_rules! alters {
    ($($struct_instance:expr),* $(,)?) => {
        {
            let mut vec: Vec<Box<dyn Alter<_>>> = Vec::new();
            $(
                vec.push(Box::new($struct_instance.alterer()));
            )*
            vec
        }
    };
}

/// This is the main trait that is used to define the different types of alterations that can be
/// performed on a [Population]. The [Alter] trait is used to define the `alter` method that is used
/// to perform the alteration on the [Population]. The `alter` method takes a mutable reference to
/// the [Population] and a generation number as parameters. The `alter` method returns a vector of
/// [Metric] objects that represent the metrics that were collected during the alteration process.
///
/// An '[Alter]' in a traditional genetic algorithm is a process that modifies the [Population] of
/// individuals in some way. This can include operations such as mutation or crossover. The goal of
/// an alter is to introduce new genetic material into the population, which can help to improve
/// the overall fitness of the population. In a genetic algorithm, the alter is typically
/// performed on a subset of the population, rather than the entire population. This allows for
/// more targeted modifications to be made, which can help to improve the overall performance of
/// the algorithm. The alter is an important part of the genetic algorithm process, as it helps
/// to ensure that the population remains diverse and that new genetic material is introduced
/// into the population. This can help to improve the overall performance of the algorithm and
/// ensure that the population remains healthy and diverse.
///
/// In `radiate` the [Alter] trait performs similar operations to a traditional genetic algorithm,
/// but it is designed to be more flexible and extensible. Because an [Alter] can be of type [Mutate]
/// or [Crossover], it is abstracted out of those core traits into this trait.
pub trait Alter<C: Chromosome>: Send + Sync {
    fn rate(&self) -> f32;
    fn alter(&self, population: &mut Population<C>, generation: usize) -> Vec<Metric>;
}

/// The [AlterResult] struct is used to represent the result of an
/// alteration operation. It contains the number of operations
/// performed and a vector of metrics that were collected
/// during the alteration process.
#[derive(Default)]
pub struct AlterResult(pub usize, pub Option<Vec<Metric>>);

impl AlterResult {
    pub fn empty() -> Self {
        Default::default()
    }

    pub fn count(&self) -> usize {
        self.0
    }

    pub fn merge(&mut self, other: AlterResult) {
        let AlterResult(other_count, other_metrics) = other;

        self.0 += other_count;
        if let Some(metrics) = other_metrics {
            if let Some(self_metrics) = &mut self.1 {
                self_metrics.extend(metrics);
            } else {
                self.1 = Some(metrics);
            }
        }
    }
}

impl From<usize> for AlterResult {
    fn from(value: usize) -> Self {
        AlterResult(value, None)
    }
}

impl From<(usize, Vec<Metric>)> for AlterResult {
    fn from((count, metrics): (usize, Vec<Metric>)) -> Self {
        AlterResult(count, Some(metrics))
    }
}

impl From<(usize, Metric)> for AlterResult {
    fn from((count, metric): (usize, Metric)) -> Self {
        AlterResult(count, Some(vec![metric]))
    }
}

impl From<Metric> for AlterResult {
    fn from(value: Metric) -> Self {
        AlterResult(1, Some(vec![value]))
    }
}

/// The [AlterAction] enum is used to represent the different
/// types of alterations that can be performed on a
/// population - It can be either a mutation or a crossover operation.
pub enum AlterAction<C: Chromosome> {
    Mutate(&'static str, f32, Box<dyn Mutate<C>>),
    Crossover(&'static str, f32, Box<dyn Crossover<C>>),
}

impl<C: Chromosome> Alter<C> for AlterAction<C> {
    fn rate(&self) -> f32 {
        match &self {
            AlterAction::Mutate(_, rate, _) => *rate,
            AlterAction::Crossover(_, rate, _) => *rate,
        }
    }

    #[inline]
    fn alter(&self, population: &mut Population<C>, generation: usize) -> Vec<Metric> {
        match &self {
            AlterAction::Mutate(name, rate, m) => {
                m.update(generation);

                let timer = std::time::Instant::now();
                let AlterResult(count, metrics) = m.mutate(population, generation, *rate);
                let metric = Metric::new(name).upsert(count).upsert(timer.elapsed());

                match metrics {
                    Some(metrics) => metrics.into_iter().chain(once(metric)).collect(),
                    None => vec![metric],
                }
            }
            AlterAction::Crossover(name, rate, c) => {
                c.update(generation);

                let timer = std::time::Instant::now();
                let AlterResult(count, metrics) = c.crossover(population, generation, *rate);
                let metric = Metric::new(name).upsert(count).upsert(timer.elapsed());

                match metrics {
                    Some(metrics) => metrics.into_iter().chain(once(metric)).collect(),
                    None => vec![metric],
                }
            }
        }
    }
}

/// Minimum population size required to perform crossover - this ensures that there
/// are enough individuals to select parents from. If the population size is
/// less than this value, we will not be able to select two distinct parents.
const MIN_POPULATION_SIZE: usize = 3;
/// Minimum number of parents required for crossover operation. This is typically
/// two, as crossover usually involves two parents to produce offspring.
const MIN_NUM_PARENTS: usize = 2;

/// The [Crossover] trait is used to define the crossover operation for a genetic algorithm.
///
/// In a genetic algorithm, crossover is a genetic operator used to vary the
/// programming of a chromosome or chromosomes from one generation to the next.
/// It is analogous to reproduction and biological crossover, upon which genetic algorithms are based.
///
/// A [Crossover] typically takes two parent [Chromosome]s and produces two or more offspring [Chromosome]s.
/// This trait allows you to define your own crossover operation on either the entire population
/// or a subset of the population. If a struct implements the [Crossover] trait but does not override
/// any of the methods, the default implementation will perform a simple crossover operation on the
/// entire population.
pub trait Crossover<C: Chromosome>: Send + Sync {
    fn name(&self) -> String {
        std::any::type_name::<Self>()
            .split("::")
            .last()
            .map(|s| s.to_snake_case())
            .unwrap()
    }

    fn update(&self, _: usize) {}

    fn rate(&self) -> f32 {
        1.0
    }

    fn alterer(self) -> AlterAction<C>
    where
        Self: Sized + 'static,
    {
        AlterAction::Crossover(intern!(self.name()), self.rate(), Box::new(self))
    }

    #[inline]
    fn crossover(
        &self,
        population: &mut Population<C>,
        generation: usize,
        rate: f32,
    ) -> AlterResult {
        let mut result = AlterResult::default();

        for i in 0..population.len() {
            if random_provider::bool(rate) && population.len() > MIN_POPULATION_SIZE {
                let parent_indexes =
                    indexes::individual_indexes(i, population.len(), MIN_NUM_PARENTS);
                let cross_result = self.cross(population, &parent_indexes, generation, rate);
                result.merge(cross_result);
            }
        }

        result
    }

    #[inline]
    fn cross(
        &self,
        population: &mut Population<C>,
        parent_indexes: &[usize],
        generation: usize,
        rate: f32,
    ) -> AlterResult {
        let mut result = AlterResult::default();

        if let Some((one, two)) = population.get_pair_mut(parent_indexes[0], parent_indexes[1]) {
            let cross_result = {
                let geno_one = one.genotype_mut();
                let geno_two = two.genotype_mut();

                let min_len = std::cmp::min(geno_one.len(), geno_two.len());
                let chromosome_index = random_provider::range(0..min_len);

                let chrom_one = &mut geno_one[chromosome_index];
                let chrom_two = &mut geno_two[chromosome_index];

                self.cross_chromosomes(chrom_one, chrom_two, rate)
            };

            if cross_result.count() > 0 {
                one.invalidate(generation);
                two.invalidate(generation);
                result.merge(cross_result);
            }
        }

        result
    }

    #[inline]
    fn cross_chromosomes(&self, chrom_one: &mut C, chrom_two: &mut C, rate: f32) -> AlterResult {
        let mut cross_count = 0;

        for i in 0..std::cmp::min(chrom_one.len(), chrom_two.len()) {
            if random_provider::bool(rate) {
                let gene_one = chrom_one.get(i);
                let gene_two = chrom_two.get(i);

                let new_gene_one = gene_one.with_allele(gene_two.allele());
                let new_gene_two = gene_two.with_allele(gene_one.allele());

                chrom_one.set(i, new_gene_one);
                chrom_two.set(i, new_gene_two);

                cross_count += 1;
            }
        }

        cross_count.into()
    }
}

pub trait Mutate<C: Chromosome>: Send + Sync {
    fn name(&self) -> String {
        std::any::type_name::<Self>()
            .split("::")
            .last()
            .map(|s| s.to_snake_case())
            .unwrap()
    }

    fn update(&self, _: usize) {}

    fn rate(&self) -> f32 {
        1.0
    }

    fn alterer(self) -> AlterAction<C>
    where
        Self: Sized + 'static,
    {
        AlterAction::Mutate(intern!(self.name()), self.rate(), Box::new(self))
    }

    #[inline]
    fn mutate(&self, population: &mut Population<C>, generation: usize, rate: f32) -> AlterResult {
        let mut result = AlterResult::default();

        for phenotype in population.iter_mut() {
            let mutate_result = self.mutate_genotype(phenotype.genotype_mut(), rate);

            if mutate_result.count() > 0 {
                phenotype.invalidate(generation);
            }

            result.merge(mutate_result);
        }

        result
    }

    #[inline]
    fn mutate_genotype(&self, genotype: &mut Genotype<C>, rate: f32) -> AlterResult {
        let mut result = AlterResult::default();

        for chromosome in genotype.iter_mut() {
            let mutate_result = self.mutate_chromosome(chromosome, rate);
            result.merge(mutate_result);
        }

        result
    }

    #[inline]
    fn mutate_chromosome(&self, chromosome: &mut C, rate: f32) -> AlterResult {
        let mut count = 0;
        for gene in chromosome.iter_mut() {
            if random_provider::bool(rate) {
                *gene = self.mutate_gene(gene);
                count += 1;
            }
        }

        count.into()
    }

    #[inline]
    fn mutate_gene(&self, gene: &C::Gene) -> C::Gene {
        gene.new_instance()
    }
}