Struct RankBasedSelection

pub struct RankBasedSelection { /* private fields */ }

A selection strategy that selects individuals based on their rank in the population.

Rank-based selection assigns a selection probability to each individual based on its rank in the population, rather than its absolute fitness value. This helps prevent premature convergence when there are a few individuals with much higher fitness than the rest of the population.

This strategy works well for both maximization and minimization problems, and can handle negative fitness values.

Examples

use genalg::selection::rank::RankBasedSelection;
use genalg::selection::SelectionStrategy;
use genalg::phenotype::Phenotype;
use genalg::rng::RandomNumberGenerator;
use genalg::error::Result;

#[derive(Clone, Debug)]
struct MyPhenotype {
    value: f64,
}

impl Phenotype for MyPhenotype {
    fn crossover(&mut self, other: &Self) {
        self.value = (self.value + other.value) / 2.0;
    }

    fn mutate(&mut self, _rng: &mut RandomNumberGenerator) {
        self.value += 0.1;
    }
}

fn main() -> Result<()> {
    let population = vec![
        MyPhenotype { value: 1.0 },
        MyPhenotype { value: 2.0 },
        MyPhenotype { value: 3.0 },
        MyPhenotype { value: 4.0 },
        MyPhenotype { value: 5.0 },
    ];
     
    let fitness = vec![0.5, 0.8, 0.3, 0.9, 0.1];
     
    // Create a rank-based selection with default parameters
    let selection = RankBasedSelection::new(1.5, false, true)?;
    let selected = selection.select(&population, &fitness, 3)?;
     
    assert_eq!(selected.len(), 3);
     
    Ok(())
}

Implementations

impl RankBasedSelection

pub fn new( selection_pressure: f64, allow_duplicates: bool, higher_is_better: bool, ) -> Result<Self>

Creates a new RankBasedSelection strategy with default parameters.

By default:

• Selection pressure is set to 1.5 (a balanced middle ground between 1.0 and 2.0)
• Duplicates are not allowed in the selected individuals
• Higher fitness is considered better

pub fn with_pressure(self, selection_pressure: f64) -> Result<Self>

Creates a new RankBasedSelection strategy with the specified selection pressure.

Arguments

• selection_pressure - The selection pressure parameter. Must be in the range [1.0, 2.0]. Higher values increase selection pressure.
  • At 1.0, all individuals have equal selection probability (no selection pressure)
  • At 2.0, selection pressure is at its maximum
  • The default value of 1.5 provides a balanced selection pressure

Returns

A Result containing the RankBasedSelection instance, or an error if the selection pressure is outside the valid range.

Errors

Returns a GeneticError::Configuration error if selection_pressure is not in the range [1.0, 2.0].

pub fn with_duplicates(self) -> Self

pub fn with_lower_is_better(self) -> Self

Trait Implementations

impl Clone for RankBasedSelection

fn clone(&self) -> RankBasedSelection

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for RankBasedSelection

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for RankBasedSelection

fn default() -> Self

Returns the “default value” for a type.

impl<P> SelectionStrategy<P> for RankBasedSelection

where P: Phenotype,

fn select( &self, population: &[P], fitness: &[f64], num_to_select: usize, ) -> Result<Vec<P>>

Selects individuals from the population based on their fitness scores.