Struct CompactGenetic 

pub struct CompactGenetic;

Compact Genetic Algorithm for binary spaces (cGA).

Implements ProbabilityModel with a per-gene probability vector updated by winner/loser competition from the truncation-selected subset. Samples are raw {0, 1} f32 values; EdaParams::bounds clamps are no-ops for this model.

See the module docs for the update rule, the deviation from classic cGA, and the reference.

Trait Implementations

impl Clone for CompactGenetic

fn clone(&self) -> CompactGenetic

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for CompactGenetic

impl Debug for CompactGenetic

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

Formats the value using the given formatter.

impl Default for CompactGenetic

fn default() -> CompactGenetic

Returns the “default value” for a type.

impl<B: Backend> ProbabilityModel<B> for CompactGenetic

fn fit( &self, params: &Self::Params, prev: Option<&Self::State>, population: Tensor<B, 2>, fitness: Tensor<B, 1>, device: &<B as BackendTypes>::Device, ) -> Self::State

Update the per-gene probability vector by winner/loser competition.

When prev = None returns the uniform-0.5 prior; population and fitness are ignored on that path. Otherwise finds the argmax (winner) and argmin (loser) of the fitness vector (canonical maximise: higher is better), then nudges each gene where winner and loser disagree by ±1 / virtual_pop_size (toward the winner), clamped to [0, 1].

Panics

Panics if the population or fitness tensor cannot be read back as f32 (.expect("population tensor must be readable as f32") / .expect("fitness tensor must be readable as f32")). Also panics with an out-of-bounds index if the population column count d exceeds prev.prob.len() (the per-gene probability vector carried in prev).

fn sample( &self, state: &Self::State, n: usize, rng: &mut dyn Rng, device: &<B as BackendTypes>::Device, ) -> Tensor<B, 2>

Draw n binary genomes from the per-gene Bernoulli probabilities.

Each gene is sampled independently as 1.0 with probability prob[j], 0.0 otherwise, using the supplied host RNG (never Tensor::random / B::seed). The returned tensor has shape (n, D) and contains only 0.0 and 1.0 values.

Panics

Panics if the n * d element count overflows allocation or TensorData capacity (Vec::with_capacity(n * d) / the (n, D) TensorData).

type Params = CompactGeneticParams

Static, per-run model configuration (learning rates, priors, …).

type State = CompactGeneticState

Evolving fitted statistics carried generation to generation.