Add types to JuMPSolverData

src/modeling/jump_solver.jl

@@ -9,15 +9,15 @@ using TimerOutputs
 
 
 mutable struct JuMPSolverData
-    varname_to_var
+    optimizer_factory
-    optimizer
+    varname_to_var::Dict{AbstractString,VariableRef}
-    instance
+    cname_to_constr::Dict{AbstractString,JuMP.ConstraintRef}
-    model
+    instance::Union{Nothing,PyObject}
-    bin_vars
+    model::Union{Nothing,JuMP.Model}
-    solution
+    bin_vars::Vector{JuMP.VariableRef}
-    cname_to_constr
+    solution::Vector{Float64}
-    reduced_costs
+    reduced_costs::Vector{Float64}
-    dual_values
+    dual_values::Dict{JuMP.ConstraintRef,Float64}
 end
 
 
@@ -74,8 +74,8 @@ function _update_solution!(data::JuMPSolverData)
             end
         end
     else
-        data.reduced_costs = nothing
+        data.reduced_costs = []
-        data.dual_values = nothing
+        data.dual_values = Dict()
     end
 end
 
@@ -85,7 +85,7 @@ function solve(
     tee::Bool=false,
     iteration_cb=nothing,
 )
-    instance, model = data.instance, data.model
+    model = data.model
     wallclock_time = 0
     log = ""
     while true
@@ -144,7 +144,7 @@ function solve_lp(data::JuMPSolverData; tee::Bool=false)
 end
 
 
-function set_instance!(data::JuMPSolverData, instance, model)
+function set_instance!(data::JuMPSolverData, instance, model::JuMP.Model)
     data.instance = instance
     data.model = model
     data.bin_vars = [
@@ -156,9 +156,7 @@ function set_instance!(data::JuMPSolverData, instance, model)
         JuMP.name(var) => var
         for var in JuMP.all_variables(data.model)
     )
-    if data.optimizer !== nothing
+    JuMP.set_optimizer(model, data.optimizer_factory)
-        JuMP.set_optimizer(model, data.optimizer)
-    end
     data.cname_to_constr = Dict()
     for (ftype, stype) in JuMP.list_of_constraint_types(model)
         for constr in JuMP.all_constraints(model, ftype, stype)
@@ -237,8 +235,8 @@ function get_variables(
         )
     end
 
-    rc = data.reduced_costs === nothing ? nothing : Tuple(data.reduced_costs)
+    rc = isempty(data.reduced_costs) ? nothing : Tuple(data.reduced_costs)
-    values = data.solution === nothing ? nothing : Tuple(data.solution)
+    values = isempty(data.solution) ? nothing : Tuple(data.solution)
 
     return miplearn.features.VariableFeatures(
         names=names,
@@ -260,7 +258,7 @@ function get_constraints(
     senses, lhs, rhs = nothing, nothing, nothing
     dual_values = nothing
 
-    if data.dual_values !== nothing
+    if !isempty(data.dual_values)
         dual_values = []
     end
 
@@ -280,7 +278,7 @@ function get_constraints(
             length(name) > 0 || continue
             push!(names, name)
 
-            if data.dual_values !== nothing
+            if !isempty(data.dual_values)
                 push!(dual_values, data.dual_values[constr])
             end
 
@@ -355,18 +353,18 @@ end
 
 
 @pydef mutable struct JuMPSolver <: miplearn.solvers.internal.InternalSolver
-    function __init__(self; optimizer)
+    function __init__(self, optimizer_factory)
         self.data = JuMPSolverData(
-            nothing,  # varname_to_var
+            optimizer_factory,
-            optimizer,
+            Dict(),  # varname_to_var
+            Dict(),  # cname_to_constr
             nothing,  # instance
             nothing,  # model
-            nothing,  # bin_vars
+            [],  # bin_vars
-            nothing,  # solution
+            [],  # solution
-            nothing,  # cname_to_constr
+            [],  # reduced_costs
-            nothing,  # reduced_costs
+            Dict(),  # dual_values
-            nothing,  # dual_values
+        )
-        ) 
     end
 
     add_constraints(self, cf) =
@@ -466,4 +464,4 @@ end
 end
 
 
-export JuMPSolver, solve!, fit!, add!
+export JuMPSolver

test/modeling/jump_solver_test.jl

@@ -12,7 +12,7 @@ miplearn_tests = pyimport("miplearn.solvers.tests")
 traceback = pyimport("traceback")
 
 @testset "JuMPSolver" begin
-    solver = JuMPSolver(optimizer=Gurobi.Optimizer)
+    solver = JuMPSolver(Gurobi.Optimizer)
     try  
         miplearn_tests.run_internal_solver_tests(solver)
     catch e