Method Families

Method Families

The public strategy catalog is organized by what the user actually needs to choose, not by historical release packs.

Exact algorithms

Direct algorithms that solve ULS without delegating the mathematical problem to an external MILP solver. The current library includes dynamic programming, geometric acceleration, planning-horizon methods, network methods, branch-and-bound and parallel variants.

Use these when you want a self-contained algorithm implementation with a proven optimum.

Mathematical optimization

Exact formulations translated to a portable linear/MILP model and solved by an external optimization engine. Automatic discovery follows:

CPLEX -> Gurobi -> Xpress -> COIN-OR CBC

Use these when formulation choice, solver comparison or mathematical-model integration matters.

Cutting planes

Exact solver-backed methods that strengthen the root model with classical (l,S) inequalities before the final exact MILP solve. Generated, rejected and added cuts remain traceable.

Use these when you want polyhedral strengthening and convergence information.

Heuristic strategies

Fast construction rules that return a feasible plan but do not claim optimality. Families include baseline rules, average-cost rules, part-period methods, marginal-cost rules and look-ahead/look-back variants.

Use these when speed, warm starts or comparison against classical planning rules matters.

Not currently present

The documentation does not display empty categories. Metaheuristics or other future method families will appear only when the repository contains actual public implementations.