Adjust tests, and fix errors

Author: BatyLeo

src/DynamicVehicleScheduling/anticipative_solver.jl

@@ -92,14 +92,14 @@ function anticipative_solver(
     job_indices = 2:nb_nodes
     epoch_indices = T
 
-    @variable(model, y[i=1:nb_nodes, j=1:nb_nodes, t=epoch_indices]; binary=true)
+    @variable(model, y[i = 1:nb_nodes, j = 1:nb_nodes, t = epoch_indices]; binary=true)
 
     @objective(
         model,
         Max,
         sum(
-            -duration[i, j] * y[i, j, t] for i in 1:nb_nodes, j in 1:nb_nodes,
-            t in epoch_indices
+            -duration[i, j] * y[i, j, t] for
+            i in 1:nb_nodes, j in 1:nb_nodes, t in epoch_indices
         )
     )
 
@@ -171,14 +171,12 @@ function anticipative_solver(
         routes = epoch_routes[i]
         epoch_customers = epoch_indices[i]
 
-        y_true =
-            VSPSolution(
-                Vector{Int}[
-                    map(idx -> findfirst(==(idx), epoch_customers), route) for
-                    route in routes
-                ];
-                max_index=length(epoch_customers),
-            ).edge_matrix
+        y_true = VSPSolution(
+            Vector{Int}[
+                map(idx -> findfirst(==(idx), epoch_customers), route) for route in routes
+            ];
+            max_index=length(epoch_customers),
+        ).edge_matrix
 
         location_indices = customer_index[epoch_customers]
         new_coordinates = env.instance.static_instance.coordinate[location_indices]
@@ -202,7 +200,8 @@ function anticipative_solver(
             is_must_dispatch[2:end] .= true
         else
             is_must_dispatch[2:end] .=
-                planning_start_time .+ epoch_duration .+ @view(new_duration[1, 2:end]) .> new_start_time[2:end]
+                planning_start_time .+ epoch_duration .+ @view(new_duration[1, 2:end]) .>
+                new_start_time[2:end]
         end
         is_postponable[2:end] .= .!is_must_dispatch[2:end]
         # TODO: avoid code duplication with add_new_customers!
@@ -223,7 +222,7 @@ function anticipative_solver(
             compute_features(state, env.instance)
         end
 
-        return DataSample(; info=(; state, reward), y, x)
+        return DataSample(; info=(; state, reward), y=y_true, x)
     end
 
     return obj, dataset

src/DynamicVehicleScheduling/plot.jl

@@ -208,9 +208,9 @@ The returned dictionary contains:
 This lets plotting code build figures without depending on plotting internals.
 """
 function build_plot_data(data_samples::Vector{<:DataSample})
-    state_data = [build_state_data(sample.instance.state) for sample in data_samples]
-    rewards = [sample.instance.reward for sample in data_samples]
-    routess = [sample.y_true for sample in data_samples]
+    state_data = [build_state_data(sample.info.state) for sample in data_samples]
+    rewards = [sample.info.reward for sample in data_samples]
+    routess = [sample.y for sample in data_samples]
     return [
         (; state..., reward, routes) for
         (state, reward, routes) in zip(state_data, rewards, routess)
@@ -273,8 +273,8 @@ function plot_epochs(
     # Create subplots
     plots = map(1:n_epochs) do i
         sample = data_samples[i]
-        state = sample.instance.state
-        reward = sample.instance.reward
+        state = sample.info.state
+        reward = sample.info.reward
 
         common_kwargs = Dict(
             :xlims => xlims,
@@ -292,7 +292,7 @@ function plot_epochs(
         if plot_routes_flag
             fig = plot_routes(
                 state,
-                sample.y_true;
+                sample.y;
                 reward=reward,
                 show_route_labels=false,
                 common_kwargs...,
@@ -351,7 +351,7 @@ function animate_epochs(
     kwargs...,
 )
     pd = build_plot_data(data_samples)
-    epoch_costs = [-sample.instance.reward for sample in data_samples]
+    epoch_costs = [-sample.info.reward for sample in data_samples]
 
     # Calculate global xlims and ylims from all states
     x_min = minimum(min(data.x_depot, minimum(data.x_customers)) for data in pd)
@@ -393,12 +393,12 @@ function animate_epochs(
     anim = @animate for frame_idx in 1:total_frames
         epoch_idx, frame_type = frame_plan[frame_idx]
         sample = data_samples[epoch_idx]
-        state = sample.instance.state
+        state = sample.info.state
 
         if frame_type == :routes
             fig = plot_routes(
                 state,
-                sample.y_true;
+                sample.y;
                 xlims=xlims,
                 ylims=ylims,
                 clims=clims,

src/StochasticVehicleScheduling/StochasticVehicleScheduling.jl

@@ -208,7 +208,7 @@ function plot_solution(
 )
     (; tasks, district_width, width) = sample.info.city
     ticks = 0:district_width:width
-    solution = Solution(sample.y_true, sample.info)
+    solution = Solution(sample.y, sample.info)
     path_list = compute_path_list(solution)
     fig = plot(;
         xlabel="x",

src/Utils/policy.jl

@@ -44,7 +44,7 @@ function evaluate_policy!(
         features, state = observe(env)
         state_copy = deepcopy(state)  # To avoid mutation issues
         reward = step!(env, y)
-        sample = DataSample(; x=features, y_true=y, instance=(; state=state_copy, reward))
+        sample = DataSample(; x=features, y=y, info=(; state=state_copy, reward))
         if @isdefined labeled_dataset
             push!(labeled_dataset, sample)
         else

src/Warcraft/Warcraft.jl

@@ -91,14 +91,14 @@ The keyword argument `θ_true` is used to set the color range of the weights plo
 function Utils.plot_data(
     ::WarcraftBenchmark,
     sample::DataSample;
-    θ_true=sample.θ_true,
+    θ_true=sample.θ,
     θ_title="Weights",
     y_title="Path",
     kwargs...,
 )
     x = sample.x
-    y = sample.y_true
-    θ = sample.θ_true
+    y = sample.y
+    θ = sample.θ
     im = dropdims(x; dims=4)
     img = convert_image_for_plot(im)
     p1 = Plots.plot(

src/Warcraft/utils.jl

@@ -40,7 +40,7 @@ function create_dataset(decompressed_path::String, nb_samples::Int)
     ]
     Y = [BitMatrix(terrain_labels[:, :, i]) for i in 1:N]
     WG = [-terrain_weights[:, :, i] for i in 1:N]
-    return [DataSample(; x, y_true, θ_true) for (x, y_true, θ_true) in zip(X, Y, WG)]
+    return [DataSample(; x, y=y_true, θ=θ_true) for (x, y_true, θ_true) in zip(X, Y, WG)]
 end
 
 """

test/argmax.jl

@@ -18,11 +18,13 @@
     @test gap >= 0
 
     for (i, sample) in enumerate(dataset)
-        (; x, θ_true, y_true) = sample
+        x = sample.x
+        θ_true = sample.θ
+        y_true = sample.y
         @test size(x) == (nb_features, instance_dim)
         @test length(θ_true) == instance_dim
         @test length(y_true) == instance_dim
-        @test isnothing(sample.instance)
+        @test isnothing(sample.info)
         @test all(y_true .== maximizer(θ_true))
 
         θ = model(x)

test/argmax_2d.jl

@@ -21,11 +21,14 @@
     @test figure isa Plots.Plot
 
     for (i, sample) in enumerate(dataset)
-        (; x, θ_true, y_true, instance) = sample
+        x = sample.x
+        θ_true = sample.θ
+        y_true = sample.y
+        instance = sample.info
         @test length(x) == nb_features
         @test length(θ_true) == 2
         @test length(y_true) == 2
-        @test !isnothing(sample.instance)
+        @test !isnothing(instance)
         @test instance isa Vector{Vector{Float64}}
         @test all(length(vertex) == 2 for vertex in instance)
         @test y_true in instance

test/dynamic_assortment.jl

@@ -2,7 +2,7 @@
     const DAP = DecisionFocusedLearningBenchmarks.DynamicAssortment
 end
 
-@testitem "DynamicAssortment - Benchmark Construction" setup=[Imports, DAPSetup] begin
+@testitem "DynamicAssortment - Benchmark Construction" setup = [Imports, DAPSetup] begin
     # Test default constructor
     b = DynamicAssortmentBenchmark()
     @test b.N == 20
@@ -13,7 +13,7 @@ end
     @test !is_exogenous(b)
 
     # Test custom constructor
-    b_custom = DynamicAssortmentBenchmark(N=10, d=3, K=2, max_steps=50, exogenous=true)
+    b_custom = DynamicAssortmentBenchmark(; N=10, d=3, K=2, max_steps=50, exogenous=true)
     @test b_custom.N == 10
     @test b_custom.d == 3
     @test b_custom.K == 2
@@ -28,8 +28,8 @@ end
     @test DAP.max_steps(b) == 80
 end
 
-@testitem "DynamicAssortment - Instance Generation" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=5, d=3, K=2)
+@testitem "DynamicAssortment - Instance Generation" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=5, d=3, K=2)
     rng = MersenneTwister(42)
 
     instance = DAP.Instance(b, rng)
@@ -53,8 +53,8 @@ end
     @test DAP.prices(instance) == instance.prices
 end
 
-@testitem "DynamicAssortment - Environment Initialization" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=5, d=2, K=2, max_steps=10)
+@testitem "DynamicAssortment - Environment Initialization" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=5, d=2, K=2, max_steps=10)
     instance = DAP.Instance(b, MersenneTwister(42))
 
     env = DAP.Environment(instance; seed=123)
@@ -80,8 +80,8 @@ end
     @test DAP.prices(env) == instance.prices
 end
 
-@testitem "DynamicAssortment - Environment Reset" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=3, d=1, K=2, max_steps=5)
+@testitem "DynamicAssortment - Environment Reset" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=3, d=1, K=2, max_steps=5)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -107,8 +107,8 @@ end
     @test env.features ≈ expected_features
 end
 
-@testitem "DynamicAssortment - Hype Update Logic" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=5, d=1, K=2)
+@testitem "DynamicAssortment - Hype Update Logic" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=5, d=1, K=2)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -135,8 +135,8 @@ end
     @test all(hype .== 1.0)  # Should not affect any item hype
 end
 
-@testitem "DynamicAssortment - Choice Probabilities" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=3, d=1, K=2)
+@testitem "DynamicAssortment - Choice Probabilities" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=3, d=1, K=2)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -167,8 +167,8 @@ end
     @test probs[4] ≈ 1.0  # Only no-purchase available
 end
 
-@testitem "DynamicAssortment - Expected Revenue" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=3, d=1, K=2)
+@testitem "DynamicAssortment - Expected Revenue" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=3, d=1, K=2)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -183,8 +183,8 @@ end
     @test revenue == 0.0  # Only no-purchase available with price 0
 end
 
-@testitem "DynamicAssortment - Environment Step" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=3, d=1, K=2, max_steps=5)
+@testitem "DynamicAssortment - Environment Step" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=3, d=1, K=2, max_steps=5)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -219,9 +219,9 @@ end
     @test_throws AssertionError step!(env, assortment)
 end
 
-@testitem "DynamicAssortment - Endogenous vs Exogenous" setup=[Imports, DAPSetup] begin
+@testitem "DynamicAssortment - Endogenous vs Exogenous" setup = [Imports, DAPSetup] begin
     # Test endogenous environment (features change with purchases)
-    b_endo = DynamicAssortmentBenchmark(N=3, d=1, K=2, exogenous=false)
+    b_endo = DynamicAssortmentBenchmark(; N=3, d=1, K=2, exogenous=false)
     instance_endo = DAP.Instance(b_endo, MersenneTwister(42))
     env_endo = DAP.Environment(instance_endo; seed=123)
 
@@ -232,7 +232,7 @@ end
     @test any(env_endo.d_features .!= 0.0)  # Delta features should be non-zero
 
     # Test exogenous environment (features don't change with purchases)
-    b_exo = DynamicAssortmentBenchmark(N=3, d=1, K=2, exogenous=true)
+    b_exo = DynamicAssortmentBenchmark(; N=3, d=1, K=2, exogenous=true)
     instance_exo = DAP.Instance(b_exo, MersenneTwister(42))
     env_exo = DAP.Environment(instance_exo; seed=123)
 
@@ -243,8 +243,8 @@ end
     @test all(env_exo.d_features .== 0.0)  # Delta features should remain zero
 end
 
-@testitem "DynamicAssortment - Observation" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=3, d=2, max_steps=10)
+@testitem "DynamicAssortment - Observation" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=3, d=2, max_steps=10)
     instance = DAP.Instance(b, MersenneTwister(42))
     env = DAP.Environment(instance; seed=123)
 
@@ -266,10 +266,10 @@ end
     @test obs1 != obs2  # Observations should differ after purchase
 end
 
-@testitem "DynamicAssortment - Policies" setup=[Imports, DAPSetup] begin
+@testitem "DynamicAssortment - Policies" setup = [Imports, DAPSetup] begin
     using Statistics: mean
 
-    b = DynamicAssortmentBenchmark(N=5, d=2, K=3, max_steps=20)
+    b = DynamicAssortmentBenchmark(; N=5, d=2, K=3, max_steps=20)
 
     # Generate test data
     dataset = generate_dataset(b, 10; seed=0)
@@ -307,8 +307,8 @@ end
     @test sum(greedy_action) == DAP.assortment_size(env)
 end
 
-@testitem "DynamicAssortment - Model and Maximizer Integration" setup=[Imports, DAPSetup] begin
-    b = DynamicAssortmentBenchmark(N=4, d=3, K=2)
+@testitem "DynamicAssortment - Model and Maximizer Integration" setup = [Imports, DAPSetup] begin
+    b = DynamicAssortmentBenchmark(; N=4, d=3, K=2)
 
     # Test statistical model generation
     model = generate_statistical_model(b; seed=42)
@@ -317,7 +317,7 @@ end
 
     # Test integration with sample data
     sample = generate_sample(b, MersenneTwister(42))
-    @test hasfield(typeof(sample), :instance)
+    @test hasfield(typeof(sample), :info)
 
     dataset = generate_dataset(b, 3; seed=42)
     environments = generate_environments(b, dataset)

test/dynamic_vsp.jl

@@ -26,7 +26,7 @@
     @test mean(r_lazy) <= mean(r_greedy)
 
     env = environments[1]
-    instance = dataset[1].instance
+    instance = dataset[1].info
     scenario = generate_scenario(b, instance)
     v, y = generate_anticipative_solution(b, env, scenario; nb_epochs=2, reset_env=true)
 
@@ -49,6 +49,8 @@
 
     anticipative_value, solution = generate_anticipative_solution(b, env; reset_env=true)
     reset!(env; reset_rng=true)
-    cost = sum(step!(env, sample.y_true) for sample in solution)
+    cost = sum(step!(env, sample.y) for sample in solution)
+    cost2 = sum(sample.info.reward for sample in solution)
     @test isapprox(cost, anticipative_value; atol=1e-5)
+    @test isapprox(cost, cost2; atol=1e-5)
 end