close gap option in column generation

Author: BatyLeo

src/FixedSizeShortestPath/FixedSizeShortestPath.jl

@@ -146,5 +146,6 @@ function Utils.generate_statistical_model(bench::FixedSizeShortestPathBenchmark)
 end
 
 export FixedSizeShortestPathBenchmark
+export generate_dataset, generate_maximizer, generate_statistical_model
 
 end

src/PortfolioOptimization/PortfolioOptimization.jl

@@ -124,5 +124,6 @@ function Utils.generate_statistical_model(bench::PortfolioOptimizationBenchmark)
 end
 
 export PortfolioOptimizationBenchmark
+export generate_dataset, generate_maximizer, generate_statistical_model
 
 end

src/StochasticVehicleScheduling/StochasticVehicleScheduling.jl

@@ -100,15 +100,22 @@ function Utils.generate_dataset(
     ]
 end
 
+"""
+$TYPEDSIGNATURES
+"""
 function Utils.generate_maximizer(bench::StochasticVehicleSchedulingBenchmark)
     return vsp_maximizer
 end
+
+"""
+$TYPEDSIGNATURES
+"""
 function Utils.generate_statistical_model(bench::StochasticVehicleSchedulingBenchmark)
     return Chain(Dense(20 => 1; bias=false), vec)
 end
 
 export StochasticVehicleSchedulingBenchmark
-
+export generate_dataset, generate_maximizer, generate_statistical_model
 export compact_linearized_mip,
     compact_mip, column_generation_algorithm, evaluate_solution, is_feasible
 

src/StochasticVehicleScheduling/solution/exact_algorithms/column_generation.jl

@@ -90,7 +90,7 @@ function column_generation(
 
     c_low = objective_value(model)
     columns = unique(cat(initial_paths, new_paths; dims=1))
-    return columns
+    return columns, c_low, value.(λ)
 end
 
 """
@@ -155,18 +155,38 @@ function column_generation_algorithm(
     bounding=true,
     use_convex_resources=true,
     silent=true,
+    close_gap=false,
 )
     Ω = isnothing(scenario_range) ? (1:get_nb_scenarios(instance)) : scenario_range
-    columns = column_generation(
+    columns, c_low, λ_val = column_generation(
         instance;
         bounding=bounding,
         use_convex_resources=use_convex_resources,
         scenario_range=Ω,
         silent=silent,
     )
-    _, _, sol = compute_solution_from_selected_columns(
+    c_upp, _, sol = compute_solution_from_selected_columns(
         instance, columns; scenario_range=Ω, silent=silent
     )
+
+    if close_gap && abs(c_upp - c_low) > 1e-6
+        (; vehicle_cost, delay_cost, graph, slacks) = instance
+        intrinsic_delays = instance.intrinsic_delays[:, Ω]
+        slacks = deepcopy(instance.slacks)
+        for edge in edges(graph)
+            slacks[src(edge), dst(edge)] = slacks[src(edge), dst(edge)][Ω]
+        end
+        threshold = (c_upp - c_low - vehicle_cost) / delay_cost
+        additional_paths, _ = stochastic_routing_shortest_path_with_threshold(
+            graph, slacks, intrinsic_delays, λ_val ./ delay_cost; threshold
+        )
+        columns = unique(cat(columns, additional_paths; dims=1))
+
+        _, _, sol = compute_solution_from_selected_columns(
+            instance, columns; scenario_range=Ω, silent=silent
+        )
+    end
+
     col_solution = solution_from_paths(sol, instance)
     return col_solution
 end

src/SubsetSelection/SubsetSelection.jl

@@ -88,5 +88,6 @@ function Utils.generate_statistical_model(bench::SubsetSelectionBenchmark; seed=
 end
 
 export SubsetSelectionBenchmark
+export generate_dataset, generate_maximizer, generate_statistical_model
 
 end