Merge pull request #899 from SebastianM-C/smc/cleanup

Cleanup BBO and MOI

Author: Vaibhavdixit02

lib/OptimizationBBO/Project.toml

@@ -1,7 +1,7 @@
 name = "OptimizationBBO"
 uuid = "3e6eede4-6085-4f62-9a71-46d9bc1eb92b"
 authors = ["Vaibhav Dixit <[email protected]> and contributors"]
-version = "0.4.0"
+version = "0.4.1"
 
 [deps]
 BlackBoxOptim = "a134a8b2-14d6-55f6-9291-3336d3ab0209"

lib/OptimizationBBO/src/OptimizationBBO.jl

@@ -83,6 +83,13 @@ function __map_optimizer_args(prob::Optimization.OptimizationCache, opt::BBO;
     return mapped_args
 end
 
+# single objective
+map_objective(obj) = obj
+# multiobjective
+function map_objective(obj::BlackBoxOptim.IndexedTupleFitness)
+    obj.orig
+end
+
 function SciMLBase.__solve(cache::Optimization.OptimizationCache{
         F,
         RC,
@@ -109,20 +116,19 @@ function SciMLBase.__solve(cache::Optimization.OptimizationCache{
         P,
         C
 }
-    local x, cur, state
-
     function _cb(trace)
         if cache.callback === Optimization.DEFAULT_CALLBACK
             cb_call = false
         else
             n_steps = BlackBoxOptim.num_steps(trace)
             curr_u = decompose_trace(trace, cache.progress)
+            objective = map_objective(BlackBoxOptim.best_fitness(trace))
             opt_state = Optimization.OptimizationState(;
                 iter = n_steps,
                 u = curr_u,
-                objective = x[1],
+                objective,
                 original = trace)
-            cb_call = cache.callback(opt_state, x...)
+            cb_call = cache.callback(opt_state, objective)
         end
 
         if !(cb_call isa Bool)
@@ -139,13 +145,7 @@ function SciMLBase.__solve(cache::Optimization.OptimizationCache{
     maxtime = Optimization._check_and_convert_maxtime(cache.solver_args.maxtime)
 
     _loss = function (θ)
-        if isa(cache.f, MultiObjectiveOptimizationFunction)
-            x = (cache.f(θ, cache.p),)
-            return x[1]
-        else
-            x = cache.f(θ, cache.p)
-            return first(x)
-        end
+        cache.f(θ, cache.p)
     end
 
     opt_args = __map_optimizer_args(cache, cache.opt;
@@ -157,8 +157,6 @@ function SciMLBase.__solve(cache::Optimization.OptimizationCache{
 
     opt_setup = BlackBoxOptim.bbsetup(_loss; opt_args...)
 
-    t0 = time()
-
     if isnothing(cache.u0)
         opt_res = BlackBoxOptim.bboptimize(opt_setup)
     else
@@ -170,13 +168,11 @@ function SciMLBase.__solve(cache::Optimization.OptimizationCache{
         Base.@logmsg(Base.LogLevel(-1), "", progress=1, _id=:OptimizationBBO)
     end
 
-    t1 = time()
-
     # Use the improved convert function
     opt_ret = Optimization.deduce_retcode(opt_res.stop_reason)
     stats = Optimization.OptimizationStats(;
         iterations = opt_res.iterations,
-        time = t1 - t0,
+        time = opt_res.elapsed_time,
         fevals = opt_res.f_calls)
     SciMLBase.build_solution(cache, cache.opt,
         BlackBoxOptim.best_candidate(opt_res),

lib/OptimizationBBO/test/runtests.jl

@@ -26,13 +26,20 @@ using Test
     @test 10 * sol.objective < l1
 
     fitness_progress_history = []
+    fitness_progress_history_orig = []
+    loss_history = []
     function cb(state, fitness)
-        push!(fitness_progress_history, [state.u, fitness])
+        push!(fitness_progress_history, state.objective)
+        push!(fitness_progress_history_orig, BlackBoxOptim.best_fitness(state.original))
+        push!(loss_history, fitness)
         return false
     end
     sol = solve(prob, BBO_adaptive_de_rand_1_bin_radiuslimited(), callback = cb)
     # println(fitness_progress_history)
     @test !isempty(fitness_progress_history)
+    fp1 = fitness_progress_history[1]
+    fp2 = fitness_progress_history_orig[1]
+    @test fp2 == fp1 == loss_history[1]
 
     @test_logs begin
         (Base.LogLevel(-1), "loss: 0.0")
@@ -77,6 +84,39 @@ using Test
             @test sol_1.objective[2]≈1.7763568e-15 atol=1e-3
         end
 
+        @testset "Sphere and Rastrigin Functions with callback" begin
+            function multi_obj_func_1(x, p)
+                f1 = sum(x .^ 2)  # Sphere function
+                f2 = sum(x .^ 2 .- 10 .* cos.(2π .* x) .+ 10)  # Rastrigin function
+                return (f1, f2)
+            end
+
+            fitness_progress_history = []
+            fitness_progress_history_orig = []
+            function cb(state, fitness)
+                push!(fitness_progress_history, state.objective)
+                push!(fitness_progress_history_orig,
+                    BlackBoxOptim.best_fitness(state.original))
+                return false
+            end
+
+            mof_1 = MultiObjectiveOptimizationFunction(multi_obj_func_1)
+            prob_1 = Optimization.OptimizationProblem(mof_1, u0; lb = lb, ub = ub)
+            sol_1 = solve(prob_1, opt, NumDimensions = 2,
+                FitnessScheme = ParetoFitnessScheme{2}(is_minimizing = true),
+                callback = cb)
+
+            fp1 = fitness_progress_history[1]
+            fp2 = fitness_progress_history_orig[1]
+            @test fp2.orig == fp1
+            @test length(fp1) == 2
+
+            @test sol_1 ≠ nothing
+            println("Solution for Sphere and Rastrigin: ", sol_1)
+            @test sol_1.objective[1]≈6.9905986e-18 atol=1e-3
+            @test sol_1.objective[2]≈1.7763568e-15 atol=1e-3
+        end
+
         # Test 2: Rosenbrock and Ackley Functions
         @testset "Rosenbrock and Ackley Functions" begin
             function multi_obj_func_2(x, p)

lib/OptimizationMOI/Project.toml

@@ -1,7 +1,7 @@
 name = "OptimizationMOI"
 uuid = "fd9f6733-72f4-499f-8506-86b2bdd0dea1"
 authors = ["Vaibhav Dixit <[email protected]> and contributors"]
-version = "0.5.2"
+version = "0.5.3"
 
 [deps]
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"

lib/OptimizationMOI/src/nlp.jl

@@ -559,7 +559,19 @@ function SciMLBase.__solve(cache::MOIOptimizationNLPCache)
         minimum = NaN
         opt_ret = SciMLBase.ReturnCode.Default
     end
-    stats = Optimization.OptimizationStats()
+
+    # check if the solver supports BarrierIterations
+    iterations = try
+        MOI.get(opt_setup, MOI.BarrierIterations())
+    catch e
+        if !(e isa MOI.GetAttributeNotAllowed)
+            rethrow(e)
+        end
+        0
+    end
+
+    stats = Optimization.OptimizationStats(; time = MOI.get(opt_setup, MOI.SolveTimeSec()),
+        iterations)
     return SciMLBase.build_solution(cache,
         cache.opt,
         minimizer,

lib/OptimizationMOI/test/runtests.jl

@@ -77,7 +77,7 @@ end
     # cache interface
     cache = init(prob, Ipopt.Optimizer())
     sol = solve!(cache)
-    @test 10 * sol.minimum < l1
+    @test 10 * sol.objective < l1
 
     optprob = OptimizationFunction(rosenbrock, Optimization.AutoZygote())
     prob = OptimizationProblem(optprob, x0, _p; sense = Optimization.MinSense)
@@ -88,16 +88,30 @@ end
     sol = solve(prob, opt) #test reuse of optimizer
     @test 10 * sol.objective < l1
 
+    # test stats
+    @test sol.stats.time > 0
+    @test sol.stats.iterations > 0
+
     sol = solve(prob,
         OptimizationMOI.MOI.OptimizerWithAttributes(Ipopt.Optimizer,
             "max_cpu_time" => 60.0))
     @test 10 * sol.objective < l1
 
+    # test stats with AbstractBridgeOptimizer
+    sol = solve(prob,
+        OptimizationMOI.MOI.OptimizerWithAttributes(Ipopt.Optimizer,
+            "max_cpu_time" => 60.0, "max_iter" => 5))
+
+    @test 60 > sol.stats.time > 0
+    @test sol.stats.iterations == 5
+
     sol = solve(prob,
         OptimizationMOI.MOI.OptimizerWithAttributes(NLopt.Optimizer,
             "algorithm" => :LN_BOBYQA))
     @test 10 * sol.objective < l1
 
+    @test sol.stats.time > 0
+
     sol = solve(prob,
         OptimizationMOI.MOI.OptimizerWithAttributes(NLopt.Optimizer,
             "algorithm" => :LD_LBFGS))
@@ -161,6 +175,7 @@ end
         res = solve(optprob, minlp_solver)
         @test res.u == [0.0, 0.0, 1.0, 0.0]
         @test res.objective == -4.0
+        @test res.stats.time > 0
     end
 
     @testset "Integer Domain" begin

lib/OptimizationNOMAD/Project.toml

@@ -1,7 +1,7 @@
 name = "OptimizationNOMAD"
 uuid = "2cab0595-8222-4775-b714-9828e6a9e01b"
 authors = ["Vaibhav Dixit <[email protected]> and contributors"]
-version = "0.3.0"
+version = "0.3.1"
 
 [deps]
 Optimization = "7f7a1694-90dd-40f0-9382-eb1efda571ba"
@@ -10,7 +10,7 @@ Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 
 [compat]
 julia = "1"
-NOMAD = "2"
+NOMAD = "2.4.1"
 Optimization = "4"
 Reexport = "1.2"