Struct CmsaEs 

pub struct CmsaEs<B: Backend> { /* private fields */ }

Covariance Matrix Self-Adaptation Evolution Strategy.

Example

use burn::backend::Flex;
use rlevo_evolution::algorithms::cmsa_es::{CmsaEsConfig, CmsaEs};

let strategy = CmsaEs::<Flex>::new();
let params = CmsaEsConfig::default_for(10);
let _ = (strategy, params);

Implementations

impl<B: Backend> CmsaEs<B>

pub fn new() -> Self

Builds a new (stateless) strategy object.

Trait Implementations

impl<B: Clone + Backend> Clone for CmsaEs<B>

fn clone(&self) -> CmsaEs<B>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<B: Copy + Backend> Copy for CmsaEs<B>

impl<B: Debug + Backend> Debug for CmsaEs<B>

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

Formats the value using the given formatter.

impl<B: Default + Backend> Default for CmsaEs<B>

fn default() -> CmsaEs<B>

Returns the “default value” for a type.

impl<B: Backend> Strategy<B> for CmsaEs<B>

where B::Device: Clone,

fn init( &self, params: &CmsaEsConfig, rng: &mut dyn Rng, _device: &<B as BackendTypes>::Device, ) -> CmsaEsState<B>

Initializes m⁰ uniformly in params.bounds (host-RNG convention), C = I, and σ̄ = initial_sigma.

fn ask( &self, params: &CmsaEsConfig, state: &CmsaEsState<B>, rng: &mut dyn Rng, device: &<B as BackendTypes>::Device, ) -> (Tensor<B, 2>, CmsaEsState<B>)

Samples λ offspring with per-individual log-normal step sizes.

For each offspring: σᵢ = σ̄ · exp(τ · N(0,1)), sᵢ = A zᵢ (A the Cholesky factor of C, zᵢ ~ N(0, I)), xᵢ = m + σᵢ · sᵢ. All draws come from one deterministic SeedPurpose::CmaSampling stream; the σᵢ are stashed in state for tell.

fn tell( &self, params: &CmsaEsConfig, population: Tensor<B, 2>, fitness: Tensor<B, 1>, state: CmsaEsState<B>, _rng: &mut dyn Rng, ) -> (CmsaEsState<B>, StrategyMetrics)

Recombines the μ best offspring into the next mean, step size, and rank-μ covariance blend.

fn best(&self, state: &CmsaEsState<B>) -> Option<(Tensor<B, 2>, f32)>

Returns the best-so-far genome and its fitness, or None before the first tell call.

type Params = CmsaEsConfig

Static parameters for a run.

type State = CmsaEsState<B>

Generation-to-generation state.

type Genome = Tensor<B, 2>

Genome container produced by ask.