[docs] fixes to ARMA and objective state tutorials (odow#561)

docs/src/tutorial/arma.jl

@@ -198,9 +198,11 @@ model = SDDP.LinearPolicyGraph(
     @variable(sp, ε[1:2])
     ## The equation describing our statistical model
     A = [0.8 0.2; 0.2 0.8]
-    inflow_in = [inflow[i].in for i in 1:2]
-    inflow_out = [inflow[i].in for i in 1:2]
-    @constraint(sp, inflow_out .== A * inflow_in .+ ε)
+    @constraint(
+        sp,
+        [i = 1:2],
+        inflow[i].out == sum(A[i, j] * inflow[j].in for j in 1:2) + ε[i],
+    )
     ## The new water balance constraint using the state variable
     @constraint(sp, x.out == x.in - g_h - s + inflow[1].out + inflow[2].out)
     ## Assume we have some empirical residuals:

docs/src/tutorial/objective_states.jl

@@ -192,8 +192,13 @@ model = SDDP.LinearPolicyGraph(
         lower_bound = (50.0, 50.0),
         upper_bound = (150.0, 150.0),
     ) do fuel_cost, ω
-        fuel_cost′ = fuel_cost[1] + 0.5 * (fuel_cost[1] - fuel_cost[2]) + ω.fuel
-        return (fuel_cost′, fuel_cost[1])
+        ## fuel_cost is a tuple, containing the (fuel_cost[t-1], fuel_cost[t-2])
+        ## This function returns a new tuple containing
+        ## (fuel_cost[t], fuel_cost[t-1]). Thus, we need to compute the new
+        ## cost:
+        new_cost = fuel_cost[1] + 0.5 * (fuel_cost[1] - fuel_cost[2]) + ω.fuel
+        ## And then return the appropriate tuple:
+        return (new_cost, fuel_cost[1])
     end
 
     Ω = [
@@ -202,7 +207,7 @@ model = SDDP.LinearPolicyGraph(
     ]
 
     SDDP.parameterize(subproblem, Ω) do ω
-        (fuel_cost, fuel_cost_old) = SDDP.objective_state(subproblem)
+        fuel_cost, _ = SDDP.objective_state(subproblem)
         @stageobjective(subproblem, fuel_cost * thermal_generation)
         return JuMP.fix(inflow, ω.inflow)
     end