Refactor unit tests (#141)

* refactor tests

* Clean test project

* reduce nlpmodeltest size

Project.toml

@@ -37,14 +37,11 @@ julia = "^1.6.0"
 
 [extras]
 ADNLPModels = "54578032-b7ea-4c30-94aa-7cbd1cce6c9a"
-BenchmarkTools = "6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
-NLPModelsIpopt = "f4238b75-b362-5c4c-b852-0801c9a21d71"
 NLPModelsTest = "7998695d-6960-4d3a-85c4-e1bceb8cd856"
-OptimizationProblems = "5049e819-d29b-5fba-b941-0eee7e64c1c6"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 SolverTest = "4343dc35-3317-4c6e-8877-f0cc8502c90e"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["ADNLPModels", "BenchmarkTools", "ForwardDiff", "NLPModelsIpopt", "NLPModelsTest", "OptimizationProblems", "Random", "SolverTest", "Test"]
+test = ["ADNLPModels", "ForwardDiff", "NLPModelsTest", "Random", "SolverTest", "Test"]

test/nlpmodelstest.jl

@@ -1,11 +1,11 @@
 function test_fps_model(T, σ, vivi, qds_type)
-  nlp1 = ADNLPModel(x -> dot(x, x), zeros(T, 10), x -> [sum(x)], ones(T, 1), ones(T, 1))
+  nlp1 = ADNLPModel(x -> dot(x, x), zeros(T, 2), x -> [sum(x)], ones(T, 1), ones(T, 1))
   qds = FletcherPenaltySolver.eval(qds_type)(nlp1, T(0))
   return FletcherPenaltyNLP(nlp1, T(σ), vivi, qds = qds)
 end
 
 function test_fps_lin_model(T, σ, vivi, qds_type, use_linear = false)
-  nvar = 10
+  nvar = 2
   nlp1 =
     ADNLPModel(x -> dot(x, x), zeros(T, nvar), sparse(ones(T, 1, nvar)), ones(T, 1), ones(T, 1))
   qds = FletcherPenaltySolver.eval(qds_type)(nlp1, T(0), explicit_linear_constraints = use_linear)

test/runtests.jl

@@ -1,16 +1,10 @@
-using LinearAlgebra, LinearOperators, Random, SparseArrays
+using LinearAlgebra, Logging, Random, SparseArrays, Stopping, Test
 
 #JSO packages
-using ADNLPModels, Logging, NLPModels, NLPModelsTest, SolverCore, SolverTest
-using JSOSolvers, NLPModelsIpopt, NLPModelsKnitro
+using ADNLPModels, JSOSolvers, NLPModels, NLPModelsTest, SolverCore, SolverTest
 
 Random.seed!(1234)
 
-using Stopping, StoppingInterface
-
-#For Test only
-using Test
-
 #This package
 using FletcherPenaltySolver
 
@@ -38,11 +32,15 @@ using FletcherPenaltySolver
   @test stats.iter == 4
 end
 
-include("restart.jl")
+@testset "Test restart" begin
+  include("restart.jl")
+end
 
 include("nlpmodelstest.jl")
 
-include("unit-test.jl")
+@testset "Unit tests" begin
+  include("unit-test.jl")
+end
 
 #Test the solvers:
 mutable struct DummyModel{S, T} <: AbstractNLPModel{S, T}
@@ -65,70 +63,33 @@ end
   )
 end
 
-@testset "Problem using KKT optimality" begin
-  f(x) = (x[1] - 1)^2 + 4 * (x[2] - x[1]^2)^2
-  c(x) = [x[1]^2 + x[2]^2 - 2]
-  T = Float64
-  x0 = T[-1.2; 1.0]
-  ℓ, u = zeros(T, 2), 2 * ones(T, 2)
-  nlp = ADNLPModel(f, x0, ℓ, u, c, zeros(1), zeros(1))
-
-  ϵ = eps(T)^T(1 / 4)
-
-  ng0 = norm(grad(nlp, nlp.meta.x0))
-
-  stp = NLPStopping(nlp)
-  stats = fps_solve(stp)
+@testset "Test usual problems and difficult cases" begin
+  @testset "Problem using KKT optimality" begin
+    f(x) = (x[1] - 1)^2 + 4 * (x[2] - x[1]^2)^2
+    c(x) = [x[1]^2 + x[2]^2 - 2]
+    T = Float64
+    x0 = T[-1.2; 1.0]
+    ℓ, u = zeros(T, 2), 2 * ones(T, 2)
+    nlp = ADNLPModel(f, x0, ℓ, u, c, zeros(1), zeros(1))
+
+    ϵ = eps(T)^T(1 / 4)
+
+    ng0 = norm(grad(nlp, nlp.meta.x0))
+
+    stp = NLPStopping(nlp)
+    stats = fps_solve(stp)
+
+    @test eltype(stats.solution) == T
+    @test stats.objective isa T
+    @test stats.dual_feas isa T
+    @test stats.primal_feas isa T
+    @test stats.dual_feas < ϵ * ng0 + ϵ
+    @test isapprox(stats.solution, ones(T, 2), atol = ϵ * ng0 * 10)
+  end
 
-  @test eltype(stats.solution) == T
-  @test stats.objective isa T
-  @test stats.dual_feas isa T
-  @test stats.primal_feas isa T
-  @test stats.dual_feas < ϵ * ng0 + ϵ
-  @test isapprox(stats.solution, ones(T, 2), atol = ϵ * ng0 * 10)
+  include("test-2.jl")
+  include("rank-deficient.jl")
 end
 
-#On usual test problems
-include("test-2.jl")
-
-#Rank-deficient problems
-include("rank-deficient.jl")
-
 # Solver tests
 include("solvertest.jl")
-
-@testset "Problems with explicit linear constraints" begin
-  nlp = ADNLPModels.ADNLPModel(
-    x -> 0.0,
-    [-1.2; 1.0],
-    [1],
-    [1],
-    [-1.0],
-    x -> [10 * (x[2] - x[1]^2)],
-    [-1.0, 0.0],
-    [-1.0, 0.0],
-    name = "mgh01feas";
-  )
-  stats = fps_solve(nlp, explicit_linear_constraints = true)
-  @test norm(cons(nlp, stats.solution) - get_lcon(nlp)) ≤ 1e-10
-  @test stats.dual_feas ≤ 1e-10
-  @test stats.primal_feas ≤ 1e-10
-  @test stats.status == :first_order
-
-  nlp = ADNLPModels.ADNLPModel(
-    x -> 0.0,
-    [-1.2; 1.0],
-    [1],
-    [1],
-    [-1.0],
-    x -> [10 * (x[2] - x[1]^2)],
-    [-1.0, 1.0],
-    [-1.0, 1.0],
-    name = "mgh01feas-bis";
-  )
-  stats = fps_solve(nlp, explicit_linear_constraints = true)
-  @test norm(cons(nlp, stats.solution) - get_lcon(nlp)) ≤ 1e-10
-  @test stats.dual_feas ≤ 1e-10
-  @test stats.primal_feas ≤ 1e-10
-  @test stats.status == :first_order
-end

test/test-2.jl

@@ -286,3 +286,39 @@ CUTEst FLT problem
   @test primal < 1e-6 * max(norm(nlp.meta.x0), 1.0)
   @test status == :first_order
 end
+
+@testset "Problems with explicit linear constraints" begin
+  nlp = ADNLPModels.ADNLPModel(
+    x -> 0.0,
+    [-1.2; 1.0],
+    [1],
+    [1],
+    [-1.0],
+    x -> [10 * (x[2] - x[1]^2)],
+    [-1.0, 0.0],
+    [-1.0, 0.0],
+    name = "mgh01feas";
+  )
+  stats = fps_solve(nlp, explicit_linear_constraints = true)
+  @test norm(cons(nlp, stats.solution) - get_lcon(nlp)) ≤ 1e-10
+  @test stats.dual_feas ≤ 1e-10
+  @test stats.primal_feas ≤ 1e-10
+  @test stats.status == :first_order
+
+  nlp = ADNLPModels.ADNLPModel(
+    x -> 0.0,
+    [-1.2; 1.0],
+    [1],
+    [1],
+    [-1.0],
+    x -> [10 * (x[2] - x[1]^2)],
+    [-1.0, 1.0],
+    [-1.0, 1.0],
+    name = "mgh01feas-bis";
+  )
+  stats = fps_solve(nlp, explicit_linear_constraints = true)
+  @test norm(cons(nlp, stats.solution) - get_lcon(nlp)) ≤ 1e-10
+  @test stats.dual_feas ≤ 1e-10
+  @test stats.primal_feas ≤ 1e-10
+  @test stats.status == :first_order
+end