Working on MIC 2024 paper

examples/NSGAIIAsAnEvolutionaryAlgorithmExample.jl

@@ -9,22 +9,25 @@ include("../src/utils.jl")
 using Dates
 
 # NSGA-II algorithm configured from the evolutionary algorithm template
-problem = zdt4Problem()
+problem = zdt1Problem(100)
 
 solver::EvolutionaryAlgorithm = EvolutionaryAlgorithm()
 solver.name = "NSGA-II"
 
 solver.problem = problem
 solver.populationSize = 100
-solver.offspringPopulationSize = 1
+solver.offspringPopulationSize = 100
 
 solver.solutionsCreation = DefaultSolutionsCreation((problem = solver.problem, numberOfSolutionsToCreate = solver.populationSize))
 
 solver.evaluation = SequentialEvaluation((problem = solver.problem, ))
 
-solver.termination = TerminationByEvaluations((numberOfEvaluationsToStop = 25000, ))
+solver.termination = TerminationByEvaluations((numberOfEvaluationsToStop = 200000, ))
 
 mutation = PolynomialMutation((probability=1.0/numberOfVariables(problem), distributionIndex=20.0, bounds=problem.bounds))
+
+#mutation = UniformMutation((probability=1.0/numberOfVariables(problem), perturbation=20.0, bounds=problem.bounds))
+
 """
 crossover = BLXAlphaCrossover((probability=1.0, alpha=0.5, bounds=problem.bounds))
 """

examples/NSGAIIExampleWithExternalCrowdingDistanceArchive.jl

@@ -16,7 +16,7 @@ solver::EvolutionaryAlgorithm = EvolutionaryAlgorithm()
 solver.name = "NSGA-II"
 solver.problem = problem
 solver.populationSize = 100
-solver.offspringPopulationSize = 1
+solver.offspringPopulationSize = 100
 
 solver.solutionsCreation = DefaultSolutionsCreation((problem = solver.problem, numberOfSolutionsToCreate = solver.populationSize))
 

examples/NSGAIISolvingAConstrainedProblem.jl

@@ -10,7 +10,7 @@ using Dates
 
 # NSGA-II algorithm example configured from the NSGA-II template
 
-problem = golinskiProblem()
+problem = tanakaProblem()
 
 solver::NSGAII = NSGAII()
 solver.problem = problem

src/bounds.jl

@@ -16,7 +16,7 @@ function createBounds(lowerBounds::Vector{T}, upperBounds::Vector{T}) where {T <
 
 end
 
-function restrict(value::Number, bounds::Bounds)::Number
+function restrict(value::T, bounds::Bounds{T}) where {T <: Number}
     result = value
     if (value < bounds.lowerBound)
         result = bounds.lowerBound 
@@ -26,6 +26,7 @@ function restrict(value::Number, bounds::Bounds)::Number
     return result 
 end
 
+
 function restrict(values::Vector{T}, bounds::Vector{Bounds{T}}) where {T <: Number}
     for i in 1:length(values)
         values[i] = restrict(values[i], bounds[i])

src/constrainedProblem.jl

@@ -93,7 +93,7 @@ function tanakaProblem()
   return tanaka
 end
 
-function ozyczka2Problem()
+function osyczka2Problem()
   osyczka2 = ContinuousProblem{Float64}("Osyczka2")
 
   # Setting the variable bounds

src/continuousProblem.jl

@@ -65,6 +65,8 @@ function evaluate(solution::ContinuousSolution{T}, problem::ContinuousProblem{T}
     solution.objectives[i] = problem.objectives[i](solution.variables)
   end
 
+  #print(solution.objectives)
+
   #solution.objectives = [f(solution.variables) for f in problem.objectives]
   #map!(x -> problem.objectives[x](solution.variables), solution.objectives, [_ for _ in 1:length(problem.objectives)])
 
@@ -285,7 +287,7 @@ function zdt4Problem(numberOfVariables::Int=10)
     return g + constant
   end
 
-  function evalH(v::Real, g::Float64)
+  function evalH(v::Float64, g::Float64)
     return 1.0 - sqrt(v/g)
   end
 
@@ -363,6 +365,8 @@ end
 # DTLZ benchmark
 #######################
 
+
+
 struct DTLZ1 <: AbstractContinuousProblem{Float64}
   bounds::Vector{Bounds{Float64}}
   numberOfObjectives::Int
@@ -394,8 +398,11 @@ function evaluate(solution::ContinuousSolution{Float64}, problem::DTLZ1)::Contin
   k = numberOfVariables(problem) - numberOfObjectives(problem) + 1
   #println("x: " , solution.variables)
 
-  g = sum([(x[i]-0.5)*(x[i]-0.5) - cos(20.0 * pi * (x[i]-0.5)) for i in range((numberOfVariables(problem)-k+1),numberOfVariables(problem))])
-
+  g = 0.0
+  for i in (numberOfVariables(problem) - k + 1):numberOfVariables(problem)
+      g += (x[i] - 0.5) * (x[i] - 0.5) - cos(20.0 * π * (x[i] - 0.5))
+  end
+
   #println("G: ",g)
   g = 100 * (k + g)
   #println("G: ",g)
@@ -408,32 +415,7 @@ function evaluate(solution::ContinuousSolution{Float64}, problem::DTLZ1)::Contin
   #println("F: ", f)
 
 
-  """
-    M = 3
-    n = 7
-    k = 5
 
-    f1 = 
-    f2 = 
-    f3 =
-
-    i = 1
-      j = 1     
-         f1 = x1 x2 (M - 1 = 3 - 1 = 2) 
-
-    i = 2
-      j = 1
-          f2 = x1
-          aux = 1?
-    i = 3
-          aux = 0?
-      
-
-    f1 = 0.5 x1 x2
-    f2 = 0.5 x1 (1 - x2) aux = 2
-    f3 = 0.5 (1 - x1) aux = 1
-
-  """
   for i in 1:numberOfObjectives(problem)
     #println("i: ",i)
     for j in 1:(numberOfObjectives(problem) - i)
@@ -442,7 +424,13 @@ function evaluate(solution::ContinuousSolution{Float64}, problem::DTLZ1)::Contin
       #println("F[",i,"] = ", f[i])
     end
     if i != 1
-      aux = numberOfObjectives(problem) - i + 1
+      #println("I: ", i)
+      #println("Objs: ", numberOfObjectives(problem))
+      aux = numberOfObjectives(problem) + 1 - i
+      """
+      2 : 2
+      3 : 1
+      """
       #println("AUX: ", aux)
       f[i] *= 1 - x[aux]
       #println("F[",i,"] = ", f[i])
@@ -454,6 +442,39 @@ function evaluate(solution::ContinuousSolution{Float64}, problem::DTLZ1)::Contin
   return solution
 end
 
+"""
+function dtlz1Problem(numberOfVariables::Int=7, numberOfObjectives::Int=3)
+  dtlz1 = ContinuousProblem{Float64}("DTLZ1")
+
+  # Add variables
+  for _ in 1:numberOfVariables
+      addVariable(dtlz1, Bounds{Float64}(0.0, 1.0))
+  end
+
+  # Function to evaluate g
+  function evalG(x::Vector{Float64})
+      return 100 * (length(x) - numberOfObjectives + 1 + sum((xi - 0.5)^2 - cos(20 * π * (xi - 0.5)) for xi in x[numberOfObjectives:end]))
+  end
+
+  # Objective functions
+  for m in 1:numberOfObjectives
+      addObjective(dtlz1, x -> begin
+          g = evalG(x)
+          f = 0.5 * (1 + g)
+          for i in 1:(numberOfObjectives - m)
+              f *= x[i]
+          end
+          if m > 1
+              f *= (1 - x[numberOfObjectives - m])
+          end
+          return f
+      end)
+  end
+
+  return dtlz1
+end
+"""
+
 function UF1Problem(numberOfVariables::Int = 30)
   uf1 = ContinuousProblem{Float64}("UF1")
 
@@ -500,3 +521,4 @@ end
 
 
 
+

src/operator.jl

@@ -108,7 +108,7 @@ function polynomialMutation(x::Vector{T}, parameters)::Vector{T} where {T<:Real}
       x[i] = y
     end
   end
-  x = randomRestrict(x, bounds)
+  x = restrict(x, bounds)
   return x
 end