public ElitismOpr(OptimisationTemplate Template, GAParameters gaParameters, Population.Population parentPopulation, Population.Population childPopulation)
{
this.Template = Template;
this.gaParameters = gaParameters;
this.parentPopulation = parentPopulation;
this.childPopulation = childPopulation;
}
public GeneticAlgorithmTravelSalesman(CitiesCollection cities, GAParameters gap)
: this(cities)
{
//SetParameters(gap);
_gap = gap;
tmrTimer = new System.Timers.Timer(10); //инициализация таймера
//tmrTimer.Elapsed += new ElapsedEventHandler(tmrTimer_Elapsed); //подписка на событие таймера "Elapsed"
}
/// <summary>
/// Создать новое поколение
/// </summary>
/// <param name="gap">Параметры генетического алгоритма</param>
/// <param name="cities">Города для расчёта</param>
public Generation(GAParameters gap, CitiesCollection cities)
{
_gap = gap;
_cities = cities;
_liAgents = new List <Agent>();
for (int i = 0; i < _gap.Population; i++)
{
_liAgents.Add(new Agent(_cities));
}
}
public void Awake()
{
// Do all the GetComponent calls upfront - Justin said it's much more efficient!
penManagers = new List <GameObject>(GameObject.FindGameObjectsWithTag("PenManager"));
penManagerScripts = new List <ManagePen>();
frogs = new List <GameObject>();
frogPlayerInfo = new List <PlayerInfo>();
neuralNetSteering = new List <NeuralNetSteering>();
// Get references to the frogs and some of their components
for (int i = 0; i < penManagers.Count; i++)
{
penManagerScripts.Add(penManagers[i].GetComponent <ManagePen>());
GameObject frog = penManagers[i].GetComponent <ManagePen>().frog;
frogs.Add(frog);
frogPlayerInfo.Add(frog.GetComponent <PlayerInfo>());
neuralNetSteering.Add(frog.GetComponent <NeuralNetSteering>());
}
// Set up directory for saving neural nets, etc
saveDataPath = "SaveData/" + System.DateTime.Now.ToString("yyMMddHHmmss");
Directory.CreateDirectory(saveDataPath);
if (LoadPath != "")
{
// Load the population from a saved file
LoadPopulation();
populationSize = population.Count;
}
else
{
// Create a new population from scratch
currentParams = parameters[parameterIndexToUse];
population = new List <NeuralNet>();
populationSize = currentParams.NumberOfBatches * FrogsOnScreen;
for (int i = 0; i < populationSize; i++)
{
population.Add(new NeuralNet(currentParams.neuralNetSettings.NumFlyPositions,
currentParams.neuralNetSettings.NumSnakePositions,
currentParams.neuralNetSettings.NumObstaclePositions,
currentParams.neuralNetSettings.FeedObstacleInfo,
currentParams.neuralNetSettings.FeedOwnVelocity,
currentParams.neuralNetSettings.FeedLakePosition,
currentParams.neuralNetSettings.FeedWaterLevel,
currentParams.neuralNetSettings.HiddenNeurons,
currentParams.neuralNetSettings.inputTransformation,
currentParams.neuralNetSettings.inputSmoothingSegments));
}
}
// Create snakes if need be
ResetSnakes();
}
public SingleObjectivePopulation(OptimisationTemplate Template, GAParameters gaParameters, int populationSize, bool empty)
: base(Template, gaParameters, populationSize)
{
//initialising population chromosomes
chromosomes = new SingleObjectiveChromosome[populationSize];
for (int i = 0; i < Size; i++)
{
chromosomes[i] = new SingleObjectiveChromosome(Template, gaParameters, empty);
}
//initialising best chromosome
bestChromosome = new SingleObjectiveChromosome(Template, gaParameters, true);
}
private void OnExecuteMainWindow()
{
est = new Estimation(dataContext.Inputs);
if (dataContext.Inputs.IsOptimizing)
{
gaView = new GAParameters();
gaView.DataContext = dataContext;
gaView.Show();
}
else
{
ResultsView resView = new ResultsView(est.Estimate());
resView.Show();
}
}
// Load the entire population from a binary file
private void LoadPopulation()
{
BinaryFormatter bf = new BinaryFormatter();
FileStream paramsFile = File.Open("SaveData/" + LoadPath + "/params.bin", FileMode.Open);
currentParams = (GAParameters)bf.Deserialize(paramsFile);
paramsFile.Close();
CurrentEpoch = LoadEpoch;
string saveFilename = "SaveData/" + LoadPath + "/population" + LoadEpoch + ".bin";
FileStream popFile = File.Open(saveFilename, FileMode.Open);
population = (List <NeuralNet>)bf.Deserialize(popFile);
popFile.Close();
}
public frmMain()
{
InitializeComponent();
// Начальная инициализация параметров расчета. Используются параметры по умолчанию.
_paramAnt = new AntParameters();
_paramBnB = new BnBParameters();
_paramGA = new GAParameters();
// Отправляем ссылку на Лист Маршрутов контроллу прорисовки
liRoute = new List <Route>();
ucCP.ListRoute = liRoute;
//инициализация списков экземпляров алгоритмов
_prAntList = new List <ProcessAnt>();
_prNNList = new List <ProcessNearestNeighbour>();
_prBnBList = new List <ProcessBranchAndBound>();
_prGAList = new List <ProcessGeneticAlgorithm>();
}
public MultiObjectivePopulation(OptimisationTemplate Template, GAParameters gaParameters, int populationSize, bool empty)
: base(Template, gaParameters, populationSize)
{
//initialising population chromosomes
chromosomes = new MultiObjectiveChromosome[size];
for (int i = 0; i < size; i++)
{
chromosomes[i] = new MultiObjectiveChromosome(Template, gaParameters, empty);
}
//initialising the seven statistics variables
int noOfObjectives = Template.Objectives.Count;
bestObjectives = new double[noOfObjectives];
worstObjectives = new double[noOfObjectives];
averageObjectives = new double[noOfObjectives];
maximumFitnesses = new double[noOfObjectives];
minimumFitnesses = new double[noOfObjectives];
averageFitnesses = new double[noOfObjectives];
fitnessVariances = new double[noOfObjectives];
}
public Chromosome(OptimisationTemplate Template, GAParameters gaParameters, bool empty)
{
this.Template = Template;
this.gaParameters = gaParameters;
int noOfDesignVariables = Template.DesignVariables.Count;
genes = new double[noOfDesignVariables];
for (int i = 0; i < noOfDesignVariables; i++)
{
if (Template.DesignVariables[i].Type == DesignVariableType.Integer)
{
if (!empty)
{
genes[i] = GARandom.BoundedRandomInteger((int)Template.DesignVariables[i].LowerBound, (int)Template.DesignVariables[i].UpperBound);
}
else
{
genes[i] = 0.0;
}
}
if (Template.DesignVariables[i].Type == DesignVariableType.Double)
{
if (!empty)
{
genes[i] = GARandom.BoundedRandomDouble(Template.DesignVariables[i].LowerBound, Template.DesignVariables[i].UpperBound);
}
else
{
genes[i] = 0.0;
}
}
}
constraintValues = new double[Template.Constraints.Count];
constraintViolationValues = new double[Template.Constraints.Count];
}
public Population(OptimisationTemplate Template, GAParameters gaParameters, int size)
{
this.Template = Template;
this.gaParameters = gaParameters;
this.size = size;
}
public Population(OptimisationTemplate Template, GAParameters gaParameters)
: this(Template, gaParameters, gaParameters.PopulationSize)
{
}
protected SelectionOpr(OptimisationTemplate Template, GAParameters gaParameters)
{
this.Template = Template;
this.gaParameters = gaParameters;
}
public SimulatedBinaryCrossoverOpr(OptimisationTemplate Template, GAParameters gaParameters, double genewiseSwapProb, int polynomialOrder)
: base(Template, gaParameters)
{
this.genewiseSwapProb = genewiseSwapProb;
this.polynomialOrder = polynomialOrder;
}
public TournamentSelectionOprWithoutReplacement(OptimisationTemplate Template, GAParameters gaParameters)
: this(Template, gaParameters, 2)
{
}
public ProportionalElitismOpr(OptimisationTemplate Template, GAParameters gaParameters, Population.Population parentPopulation, Population.Population childPopulation)
: base(Template, gaParameters, parentPopulation, childPopulation)
{
}
protected double crowdingDistance; //will be initialised to 0.0 upon object creation
public MultiObjectiveChromosome(OptimisationTemplate Template, GAParameters gaParameters, bool empty)
: base(Template, gaParameters, empty)
{
objectiveValues = new double[Template.Objectives.Count];
fitnessValues = new double[Template.Objectives.Count];
}
public CrossoverOpr(OptimisationTemplate Template, GAParameters gaParameters)
{
this.Template = Template;
this.gaParameters = gaParameters;
}
public SelectiveMutationOpr(OptimisationTemplate Template, GAParameters gaParameters)
: base(Template, gaParameters)
{
}
public MutationOpr(OptimisationTemplate Template, GAParameters gaParameters)
{
this.Template = Template;
this.gaParameters = gaParameters;
}
//public Treatment Treatment { get; }
public override bool Execute()
{
bool status = true;
/*
#region Atif
* string databaseFileName = Path.Combine(AircadiaProject.ProjectPath, "Studies\\" + this.Name + "\\" + this.Name + ".sdf"); // Microsoft SQL server compact edition file
*
* // Create database for optimisation results
* string connectionString;
*
* connectionString = string.Format("Data Source = " + databaseFileName + ";Persist Security Info=False");
* SqlCeEngine engine = new SqlCeEngine(connectionString);
*
* // Create tables
* SqlCeConnection connection = new SqlCeConnection(connectionString);
* if (connection.State == ConnectionState.Closed)
* {
* connection.Open();
* }
*
*
*
*
*
* // Create SQL statement command for "SQL Server Compact Edition"
* SqlCeCommand command1 = new SqlCeCommand(sqlStatement1, connection);
* SqlCeCommand command2 = new SqlCeCommand(sqlStatement2, connection);
* // Execute the SQL command
* try
* {
* command1.ExecuteNonQuery();
* command2.ExecuteNonQuery();
* }
* catch (SqlCeException sqlexception)
* {
* Console.WriteLine(sqlexception.Message);
* }
* catch (Exception ex)
* {
* Console.WriteLine(ex.Message);
* }
#endregion Atif
*/
if (Treatment is NSGAIIOptimiser opt)
{
//throw new NotImplementedException("NSGAII or SGA used a GUI to define some parameters, removing this dependencies needs to be implemented");
//opt.GAParameters.SelectionOprMethod = Treatments.Optimisers.GA.Selection.SelectionOprMethods.TournamentSelectionWithoutReplacement;
//opt.GAParameters.CrossoverOprMethod = Treatments.Optimisers.GA.Crossover.CrossoverOprMethods.UniformCrossover;
//opt.GAParameters.MutationOprMethod = Treatments.Optimisers.GA.Mutation.MutationOprMethods.PolynomialMutation;
//opt.GAParameters.EvaluatedSolutionsFile = Path.Combine(AircadiaProject.ProjectPath, "evaluatedSolutions.txt");
// UI Code
// NoOfGenerations = (int)numericUpDownNoOfGenerations.Value;
//GAParameters.NoOfGenerations = NoOfGenerations;
//PopulationSize = (int)numericUpDownPopulationSize.Value;
//GAParameters.PopulationSize = PopulationSize;
//if ((comboBoxSelectionOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Roulette Wheel")
// GAParameters.SelectionOprMethod = SelectionOprMethods.RouletteWheelSelection;
//else if ((comboBoxSelectionOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Tournament Without Replacement")
// GAParameters.SelectionOprMethod = SelectionOprMethods.TournamentSelectionWithoutReplacement;
//else if ((comboBoxSelectionOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Tournament With Replacement")
// GAParameters.SelectionOprMethod = SelectionOprMethods.TournamentSelectionWithReplacement;
//if ((comboBoxCrossoverOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "One-Point")
// GAParameters.CrossoverOprMethod = CrossoverOprMethods.OnePointCrossover;
//else if ((comboBoxCrossoverOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Two-Point")
// GAParameters.CrossoverOprMethod = CrossoverOprMethods.TwoPointCrossover;
//else if ((comboBoxCrossoverOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Uniform")
// GAParameters.CrossoverOprMethod = CrossoverOprMethods.UniformCrossover;
//else if ((comboBoxCrossoverOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Simulated Binary")
// GAParameters.CrossoverOprMethod = CrossoverOprMethods.SimulatedBinaryCrossover;
//if ((comboBoxMutationOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Genewise")
// GAParameters.MutationOprMethod = MutationOprMethods.GenewiseMutation;
//else if ((comboBoxMutationOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Selective")
// GAParameters.MutationOprMethod = MutationOprMethods.SelectiveMutation;
//else if ((comboBoxMutationOpr.SelectedItem as RadComboBoxItem).Content.ToString() == "Polynomial")
// GAParameters.MutationOprMethod = MutationOprMethods.PolynomialMutation;
// GA setup
GASetup gaSetup = new GASetup(this.OptimisationTemplate);
if (gaSetup.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
GAParameters gaParameters = gaSetup.GAParameters;
//gaParameters.EvaluatedSolutionsFile = Path.Combine(AircadiaProject.ProjectPath, "evaluatedSolutions.txt");
((NSGAIIOptimiser)(Treatment)).GAParameters = gaParameters;
}
}
else if (Treatment is SGAOptimizer)
{
throw new NotImplementedException("NSGAII or SGA used a GUI to define some parameters, removing this dependencies needs to be implemented");
}
Treatment.Result = ActiveResult;
status = Treatment.ApplyOn();
return(status);
}
public SingleObjectiveChromosome(OptimisationTemplate Template, GAParameters gaParameters, bool empty)
: base(Template, gaParameters, empty)
{
}
public TournamentSelectionOprWithoutReplacement(OptimisationTemplate Template, GAParameters gaParameters, int tournamentSize)
: base(Template, gaParameters, tournamentSize)
{
}
public TournamentSelectionOpr(OptimisationTemplate Template, GAParameters gaParameters, int tournamentSize)
: base(Template, gaParameters)
{
this.tournamentSize = tournamentSize;
}
public SimulatedBinaryCrossoverOpr(OptimisationTemplate Template, GAParameters gaParameters)
: this(Template, gaParameters, 0.5, 10)
{
}
public MultiObjectivePopulation(OptimisationTemplate Template, GAParameters gaParameters, bool empty)
: this(Template, gaParameters, gaParameters.PopulationSize, empty)
{
}
public PolynomialMutationOpr(OptimisationTemplate Template, GAParameters gaParameters)
: base(Template, gaParameters)
{
}
public OnePointCrossoverOpr(OptimisationTemplate Template, GAParameters gaParameters)
: base(Template, gaParameters)
{
}
public SGAOptimizer(string name, string description, GAParameters gaParameters)
: base(name, description, gaParameters)
{
}
public GASetup(OptimisationTemplate template)
{
InitializeComponent();
gaParameters = new GAParameters(template);
}
