Crate converge_optimization

Expand description

§converge-optimization

Optimization solvers as first-class Suggestors for the Converge Engine.

Every solver is accessed through PackSuggestor – the ONLY public interface. Register it in a formation and let it converge alongside LLM agents, policy gates, and other Suggestors.

§Usage

use converge_pack::{PackSuggestor, ContextKey};
use converge_optimization::packs::budget_allocation::BudgetAllocationPack;

let solver = PackSuggestor::new(
    BudgetAllocationPack,
    ContextKey::Seeds,
    ContextKey::Strategies,
);
engine.register_suggestor(solver);

§Available Packs (21)

LeadRouting, MeetingScheduler, BudgetAllocation, CapacityPlanning, InventoryReplenishment, InventoryRebalancing, AnomalyTriage, PricingGuardrails, ShippingChoice, VendorShortlist, BacklogPrioritization, AssignmentPack, BinPacking, ConstraintProgramming, FacilityLocation, GraphPartitioning, JobShopScheduling, NetworkFlow, StaffRostering, TravelingSalesman, VehicleRouting

§Feature Flags

sat - Varisat SAT solver for constraint programming
full - All native optimization features

Modules§

assignment: Linear Assignment Problem solvers
graph: Graph algorithms
knapsack: Knapsack problem solvers
packs: Domain Packs for Solver Gate
provider: Converge platform integration
scheduling: Scheduling algorithms and constraints
setcover: Set Cover problem solvers
suggestors

Structs§

PackSolveResult: Result of pack solving
PackSuggestor: Wraps any Pack as a Converge Suggestor.
SolverParams: Common solver parameters
SolverStats: Statistics from a solver run

Enums§

Error: Errors that can occur during optimization
SolverStatus: Solver status after optimization

Traits§

Pack: A domain pack for the solver gate
SolveBudgetsExt: Extension trait for SolveBudgets that bridges to optimization-specific SolverParams.

Type Aliases§

Cost: Cost type for optimization problems (signed to allow negative costs)
Index: Index type for nodes/variables
Result: Result type alias using our Error type
Value: Value type for objective functions
Weight: Weight type for capacity constraints