Solution

Solution classes for the three Python-based PS-BHLP models.

Each solution class extracts variable values from the Gurobi model, formats them as tables, and infers prices for models without explicit price variables.

PC_HLPSolution

Bases: PriceMixin, DecisionMixin, HubLocationMixin, LinearWeightMixin, InferredPricingMixin, RecursiveDecisionMixin, BaseModelSolution

Solution for the PC-HLP (Fast Lagrange) model.

Since PC-HLP has no explicit price variable, price is inferred from the budget-to-weight ratio of the marginal client on each route. Recursive decisions are unrolled to the original client indices.

Mixin composition:

• PriceMixin — price table
• DecisionMixin — active routes
• HubLocationMixin — hub status
• LinearWeightMixin — weights
• InferredPricingMixin — computes prices from marginal clients
• RecursiveDecisionMixin — unrolls aggregated decisions to original client keys

PPC_HLPSolution

Bases: PriceMixin, DecisionMixin, HubLocationMixin, LinearWeightMixin, InferredPricingMixin, BaseModelSolution

Solution for the PPC-HLP (Lagrange) model.

Like PC-HLP, price is inferred post-solve from the marginal client on each route. No recursive unrolling is needed here — PPC-HLP uses flat (non-aggregated) client indices.

Mixin composition:

• PriceMixin — price table
• DecisionMixin — active routes
• HubLocationMixin — hub status
• LinearWeightMixin — weights
• InferredPricingMixin — computes prices from marginal clients

PS_HLPSolution

Bases: PriceMixin, DecisionMixin, HubLocationMixin, BaseModelSolution

Solution for the PS-HLP (Big M) model.

Unlike the Lagrange-based models, PS-HLP has price as an explicit Gurobi variable, so no inference is needed. The price values are extracted directly from the solver.

Mixin composition:

• PriceMixin — price table
• DecisionMixin — active routes
• HubLocationMixin — hub status
• [BaseModelSolution][bilevelpy.solution.core.BaseModelSolution] — base variable extraction