Fix SimulatorSamplingScheme for deterministic nodes and update docs

docs/src/tutorial/example_milk_producer.jl

@@ -161,8 +161,13 @@ model = SDDP.PolicyGraph(
         x_stock.in + ω_production + u_spot_buy - x_forward[1].in - u_spot_sell
     )
     ## The random variables. `price` comes from the Markov node
+    ##
+    ## !!! warning
+    ##     The elements in Ω MUST be a tuple with 1 or 2 values, where the first
+    ##     value is `price` and the second value is the random variable for the
+    ##     current node. If the node is deterministic, use Ω = [(price,)].
     Ω = [(price, p) for p in Ω_production]
-    SDDP.parameterize(sp, Ω) do ω::Tuple{Float64,Float64}
+    SDDP.parameterize(sp, Ω) do ω
         ## Fix the ω_production variable
         fix(ω_production, ω[2])
         @stageobjective(

src/plugins/sampling_schemes.jl

@@ -481,8 +481,9 @@ which returns a `Vector{Float64}` when called with no arguments like
 This sampling scheme must be used with a Markovian graph constructed from the
 same `simulator`.
 
-The sample space for [`SDDP.parameterize`](@ref) must be a tuple in which the
-first element is the Markov state.
+The sample space for [`SDDP.parameterize`](@ref) must be a tuple with 1 or 2
+values, value is the Markov state and the second value is the random variable
+for the current node. If the node is deterministic, use `Ω = [(markov_state,)]`.
 
 This sampling scheme generates a new scenario by calling `simulator()`, and then
 picking the sequence of nodes in the Markovian graph that is closest to the new
@@ -508,7 +509,7 @@ julia> model = SDDP.PolicyGraph(
            @variable(sp, x >= 0, SDDP.State, initial_value = 1)
            @variable(sp, u >= 0)
            @constraint(sp, x.out == x.in - u)
-           # Elements of Ω must be a tuple in which `markov_state` is the first
+           # Elements of Ω MUST be a tuple in which `markov_state` is the first
            # element.
            Ω = [(markov_state, (u = u_max,)) for u_max in (0.0, 0.5)]
            SDDP.parameterize(sp, Ω) do (markov_state, ω)
@@ -559,7 +560,8 @@ function sample_scenario(
         noise_terms = get_noise_terms(InSampleMonteCarlo(), node, node_index)
         noise = sample_noise(noise_terms)
         @assert noise[1] == node_index[2]
-        push!(scenario_path, (node_index, (value, noise[2])))
+        ω = length(noise) == 1 ? (value,) : (value, noise[2])
+        push!(scenario_path, (node_index, ω))
     end
     return scenario_path, false
 end

test/plugins/sampling_schemes.jl

@@ -318,6 +318,70 @@ function test_OutOfSampleMonteCarlo_initial_node()
     end
 end
 
+function test_SimulatorSamplingScheme()
+    function simulator()
+        inflow = zeros(3)
+        current = 50.0
+        Ω = [-10.0, 0.1, 9.6]
+        for t in 1:3
+            current += rand(Ω)
+            inflow[t] = current
+        end
+        return inflow
+    end
+    graph = SDDP.MarkovianGraph(simulator; budget = 8, scenarios = 30)
+    model = SDDP.PolicyGraph(
+        graph,
+        lower_bound = 0.0,
+        direct_mode = false,
+    ) do sp, node
+        t, price = node
+        @variable(sp, 0 <= x <= 1, SDDP.State, initial_value = 0)
+        SDDP.parameterize(sp, [(price,)]) do ω
+            return SDDP.@stageobjective(sp, price * x.out)
+        end
+    end
+    sampler = SDDP.SimulatorSamplingScheme(simulator)
+    scenario, _ = SDDP.sample_scenario(model, sampler)
+    @test length(scenario) == 3
+    @test haskey(graph.nodes, scenario[1][1])
+    @test scenario[1][2] in ((40.0,), (50.1,), (59.6,))
+    return
+end
+
+function test_SimulatorSamplingScheme_with_noise()
+    function simulator()
+        inflow = zeros(3)
+        current = 50.0
+        Ω = [-10.0, 0.1, 9.6]
+        for t in 1:3
+            current += rand(Ω)
+            inflow[t] = current
+        end
+        return inflow
+    end
+    graph = SDDP.MarkovianGraph(simulator; budget = 8, scenarios = 30)
+    model = SDDP.PolicyGraph(
+        graph,
+        lower_bound = 0.0,
+        direct_mode = false,
+    ) do sp, node
+        t, price = node
+        @variable(sp, 0 <= x <= 1, SDDP.State, initial_value = 0)
+        SDDP.parameterize(sp, [(price, i) for i in 1:2]) do ω
+            return SDDP.@stageobjective(sp, price * x.out + i)
+        end
+    end
+    sampler = SDDP.SimulatorSamplingScheme(simulator)
+    scenario, _ = SDDP.sample_scenario(model, sampler)
+    @test length(scenario) == 3
+    @test haskey(graph.nodes, scenario[1][1])
+    @test scenario[1][2] isa Tuple{Float64,Int}
+    @test scenario[1][2][1] in (40.0, 50.1, 59.6)
+    @test scenario[1][2][2] in 1:3
+    return
+end
+
 end  # module
 
 TestSamplingSchemes.runtests()