Crate rlevo_evolution 

Tensor-native classical evolutionary algorithms for rlevo.

This crate ships the classical EA families — Genetic Algorithm (GA), Evolution Strategy (ES), Evolutionary Programming (EP), Differential Evolution (DE), and Cartesian Genetic Programming (CGP) — on top of the Burn tensor abstraction, with GPU acceleration via custom CubeCL kernels on hot paths.

Surface area

• strategy — the central Strategy trait and the EvolutionaryHarness adapter that wraps any strategy into rlevo-core::evaluation::BenchEnv.
• function_set — the shared FunctionSet opcode contract (and the default ArithmeticFunctionSet) used by both Cartesian GP and Gene Expression Programming.
• genome — zero-sized marker types (Real, Binary, Integer, Tree, Permutation) that parameterize the operator set.
• population — Population<B, K>, a thin wrapper around Tensor<B, 2> carrying shape metadata.
• fitness — FitnessFn / BatchFitnessFn, the FromFitnessEvaluable adapter for rlevo-core::fitness::FitnessEvaluable, and the FromLandscape adapter for landscapes that carry their own evaluate method.
• observer — PopulationObserver / PopulationSnapshot / SharedPopulationObserver: structured per-generation callback for recorders that need more than the scalar tracing events (full fitness vector, best-individual index, lineage).
• rng — deterministic seed streams (splitmix64) for reproducibility.
• shaping — fitness shaping transforms (centered rank, z-score).
• ops — selection, crossover, mutation, and replacement operators.
• local_search — host-side, gradient-free refinement (LocalSearch and the four reference searchers) for memetic algorithms.
• probability_model — the ProbabilityModel trait shared by the estimation-of-distribution (EDA) strategies.
• coevolution — competitive / cooperative co-evolution (CompetitiveCoEA, CooperativeCoEA), the CoupledFitness trait, the HallOfFameFitness cycling mitigation, and the CoEvolutionaryHarness BenchEnv adapter.
• algorithms — concrete strategies.

Re-exports

pub use algorithms::eda::BayesianNetwork;

pub use algorithms::eda::BayesianNetworkParams;

pub use algorithms::eda::CompactGenetic;

pub use algorithms::eda::CompactGeneticParams;

pub use algorithms::eda::DependencyChain;

pub use algorithms::eda::DependencyChainParams;

pub use algorithms::eda::EdaParams;

pub use algorithms::eda::EdaState;

pub use algorithms::eda::EdaStrategy;

pub use algorithms::eda::UnivariateBernoulli;

pub use algorithms::eda::UnivariateBernoulliParams;

pub use algorithms::eda::UnivariateGaussian;

pub use algorithms::eda::UnivariateGaussianParams;

pub use algorithms::memetic::CoveragePolicy;

pub use algorithms::memetic::MemeticWrapper;

pub use algorithms::memetic::WritebackPolicy;

pub use algorithms::neuroevolution::ArchNasBuilder;

pub use algorithms::neuroevolution::ArchNasFitnessFn;

pub use algorithms::neuroevolution::ArchNasStrategy;

pub use algorithms::neuroevolution::BatchGraphFitness;

pub use algorithms::neuroevolution::GraphFitnessFn;

pub use algorithms::neuroevolution::NasBuilderConfig;

pub use algorithms::neuroevolution::NasGenome;

pub use algorithms::neuroevolution::NasParams;

pub use algorithms::neuroevolution::NasState;

pub use algorithms::neuroevolution::NeatParams;

pub use algorithms::neuroevolution::NeatState;

pub use algorithms::neuroevolution::NeatStrategy;

pub use algorithms::neuroevolution::VariantEvaluator;

pub use algorithms::neuroevolution::WeightOnly;

pub use coevolution::CoEAMetrics;

pub use coevolution::CoEAState;

pub use coevolution::CoEvolutionaryAlgorithm;

pub use coevolution::CoEvolutionaryHarness;

pub use coevolution::CompetitiveCoEA;

pub use coevolution::CompetitiveCoEAParams;

pub use coevolution::CooperativeCoEA;

pub use coevolution::CooperativeCoEAParams;

pub use coevolution::CooperativeState;

pub use coevolution::CoupledFitness;

pub use coevolution::HallOfFame;

pub use coevolution::HallOfFameFitness;

pub use coevolution::RepresentativePolicy;

pub use function_set::ArithmeticFunctionSet;

pub use function_set::FunctionSet;

pub use function_set::Symbol;

pub use local_search::HillClimbing;

pub use local_search::LocalSearch;

pub use local_search::NelderMead;

pub use local_search::RandomRestart;

pub use local_search::SimulatedAnnealing;

pub use module_eval_fn::ModuleEvalFn;

pub use neuroevolution::ActivationFn;

pub use neuroevolution::BatchPhenotypeEvaluator;

pub use neuroevolution::ConnectionGene;

pub use neuroevolution::DensePaddedEvaluator;

pub use neuroevolution::InnovationId;

pub use neuroevolution::InnovationRegistry;

pub use neuroevolution::InterpretedBuilder;

pub use neuroevolution::InterpretedPhenotype;

pub use neuroevolution::NodeGene;

pub use neuroevolution::NodeId;

pub use neuroevolution::NodeKind;

pub use neuroevolution::NodeSplit;

pub use neuroevolution::Phenotype;

pub use neuroevolution::PhenotypeBuilder;

pub use neuroevolution::Species;

pub use neuroevolution::SpeciesId;

pub use neuroevolution::TopologyGenome;

pub use neuroevolution::compatibility_distance;

pub use observer::PopulationObserver;

pub use observer::PopulationSnapshot;

pub use observer::SharedPopulationObserver;

pub use param_reshaper::ModuleReshaper;

pub use param_reshaper::ParamReshaper;

pub use probability_model::ProbabilityModel;

pub use shaping::ShapingError;

pub use strategy::EvolutionaryHarness;

pub use strategy::Strategy;

pub use strategy::StrategyMetrics;

Modules

algorithms

Concrete evolutionary algorithms.

coevolution

Co-evolutionary algorithms.

fitness

Fitness evaluation traits and adapters.

function_set

Shared opcode contract for tree- and graph-structured genetic programming.

genome

Genome category trait and its zero-sized marker types.

local_search

Host-side local-search refinement for memetic algorithms.

module_eval_fn

ModuleEvalFn — fitness adapter that scores a population of flat parameter rows by reconstructing one Burn module per row.

neuroevolution

Topology-evolving neuroevolution (NEAT) — the host-side graph data model.

observer

Per-generation observer surface for EvolutionaryHarness.

ops

Evolutionary operators.

param_reshaper

Bridge between a Burn Module and a flat parameter vector.

population

Population containers.

probability_model

The ProbabilityModel trait shared by estimation-of-distribution algorithms (EDAs).

rng

Deterministic seed derivation for strategies.

shaping

Fitness shaping transforms.

strategy

Central Strategy trait and the EvolutionaryHarness adapter.