Expand description
Exact-aware evolutionary and proposal-search carriers.
hyperevolution owns candidate encodings, populations, fitness reports,
exact comparison policies, archives, and replay policies for the Hyper
ecosystem. It treats stochastic search as proposal generation: accepted
candidates must replay through exact/certified predicates, residuals, or
domain reports.
This follows Yap, “Towards Exact Geometric Computation,” Computational Geometry 7(1-2), 1997 (https://doi.org/10.1016/0925-7721(95)00040-2) at the optimization boundary. Search heuristics such as Holland’s genetic algorithms (“Adaptation in Natural and Artificial Systems”, 1975) and Kirkpatrick, Gelatt, and Vecchi’s simulated annealing (“Optimization by Simulated Annealing”, Science 220(4598), 1983) can propose candidates, but exact ranking and replay evidence remain first-class data.
Re-exports§
pub use domain::DomainReplayManifest;pub use domain::DomainReplayReport;pub use domain::DomainReplayTarget;pub use domain::domain_replay_manifest;pub use fitness::FitnessComparison;pub use fitness::FitnessDirection;pub use fitness::FitnessInterval;pub use fitness::FitnessIntervalComparison;pub use fitness::FitnessReport;pub use fitness::FitnessValue;pub use fitness::ParetoRelation;pub use gp::GpRealExpr;pub use gp::GpValidationIssue;pub use gp::GpValidationLimits;pub use gp::GpValidationReport;pub use gp::eval_gp_batch;pub use identity::CandidateId;pub use oracle::BlackBoxEvaluationReport;pub use oracle::ConstructionDependency;pub use oracle::EvaluationCacheKey;pub use oracle::EvaluationCost;pub use oracle::FitnessOracle;pub use oracle::ReplayHook;pub use oracle::SurrogateDecision;pub use oracle::SurrogateScreen;pub use oracle::SurrogateScreenReport;pub use oracle::SurrogateStage;pub use oracle::evaluate_candidate_with_oracle;pub use oracle::surrogate_allows_archive_promotion;pub use search::AnnealingAcceptance;pub use search::Archive;pub use search::Candidate;pub use search::CrossoverPolicy;pub use search::DiversityRelation;pub use search::DiversityReport;pub use search::Genome;pub use search::HillClimbPolicy;pub use search::HillClimbReport;pub use search::HillClimbStopReason;pub use search::HillClimbStrategy;pub use search::MutationPolicy;pub use search::Population;pub use search::ReplayPolicy;pub use search::ReplayStatus;pub use search::SelectionError;pub use search::SelectionPolicy;pub use search::SelectionReport;pub use search::SimulatedAnnealingPolicy;pub use search::VariationError;pub use search::classify_simulated_annealing_neighbor;pub use search::crossover_one_point;pub use search::exact_structural_diversity;pub use search::hill_climb_exact;pub use search::mutate_exact_delta;pub use search::select_exact_best;pub use search::select_tournament_by_indices;
Modules§
- domain
- Domain replay integration manifests.
- fitness
- Exact fitness values and comparisons.
- gp
- Typed genetic-programming expression genomes.
- identity
- Stable search identifiers.
- oracle
- Black-box objective and surrogate-screening surfaces.
- search
- Candidate, population, archive, and proposal-policy carriers.
Structs§
- Real
- (More) Real numbers