private async void RunOptimizer([NotNull] object sender, [NotNull] RoutedEventArgs args)
{
await Task.Run(() =>
{
var parameters = new GeneticAlgorithmParameters
{
CrossoverRate = 0.85,
ElitismRate = 0.1,
Generations = 300,
MutationRate = 0.7,
PopulationSize = 50
};
var optimizer = new GeneticCompoundFDPOptimizer(parameters);
optimizer.GenerationExtincts += OnGenerationExtincts;
optimizer.Run();
optimizer.GenerationExtincts -= OnGenerationExtincts;
PrintChromosomes(optimizer);
void OnGenerationExtincts(int generation)
{
Dispatcher.Invoke(() =>
{
FeedbackLabel.Text = $"Generation {generation}/{parameters.Generations}";
GenerationResult.Text = PrintChromosomes(optimizer);
});
}
});
}
/// <summary>
/// Request <seealso cref="OptimizationManagerGeneticAlgorithm"/> to start the Optimization Process
/// </summary>
private void StartGeneticOptimization()
{
// Reset Value
RoundsCompleted = 0;
// Array of object to be used as parameters for given strategy
var parameterValues = GetParameterValues(SelectedStrategy.ParameterDetails).ToArray();
var optimizationParametersDetail = new SortedDictionary <int, OptimizationParameterDetail>();
foreach (var optimizationParameterDetail in OptimizationParameters)
{
optimizationParametersDetail.Add(optimizationParameterDetail.Index, optimizationParameterDetail);
}
var strategyDetails = new GeneticAlgorithmParameters(SelectedStrategy.StrategyType, parameterValues,
optimizationParametersDetail, _iterations, _populationSize, _rounds);
// Change Status
Status = OptimizationStatus.Working;
// Clear existing data
OptimizedParameters.Clear();
// Raise Event to notify listeners to start Optimization Process
Task.Factory.StartNew(() => { EventSystem.Publish <GeneticAlgorithmParameters>(strategyDetails); });
}
public void ButtonStartOnClick()
{
if (!IsFormValid())
{
ValidationMessage.enabled = true;
SetTimestampToNow();
return;
}
var geneticAlgorithmParameters = new GeneticAlgorithmParameters()
{
PopulationCount = 2 * int.Parse(PopulationPairCount.text),
MaxGenerationCount = int.Parse(MaxGenerationCount.text),
NeuralNetworkWeightInitializationDistribution = new Normal(0f, double.Parse(WeightInitStdDev.text)),
MinSensorCount = int.Parse(MinSensorCount.text),
MaxSensorCount = int.Parse(MaxSensorCount.text),
MinHiddenLayerNeuronCount = int.Parse(MinHiddenLayerNeuronCount.text),
MaxHiddenLayerNeuronCount = int.Parse(MaxHiddenLayerNeuronCount.text),
EvolveNetworkTopology = EvolveNetworkTopology.isOn,
NeuralNetworkHiddenLayerActivationFunction = GetHiddenLayerActivationFunction(),
NeuralNetworkOutputLayerActivationFunction = x => (float)Math.Tanh(x),
SelectionStrategy = GetSelectionStrategy(),
CrossoverStrategy = GetCrossoverStrategy()
};
geneticAlgorithmParameters.MutationStrategy = GetMutationStrategy(geneticAlgorithmParameters);
IFitnessAccumulatorFactory fitnessAccumulatorFactory = GetFitnessAccumulatorFactory();
Scenes.Data.Add(DataTags.GeneticAlgorithmParameters, geneticAlgorithmParameters);
Scenes.Data.Add(DataTags.FitnessAccumulatorFactory, fitnessAccumulatorFactory);
Scenes.Load(SceneId.MainScene);
}
/// <summary>
/// Handles incoming strategy optimization requests
/// </summary>
/// <param name="strategyInfo">Information to optimized specified strategy</param>
private void OptimizeStrategy(GeneticAlgorithmParameters strategyInfo)
{
// Save constructor arguments
_ctorArguments = strategyInfo.CtorArgs;
// Save user strategy type
_strategyType = strategyInfo.StrategyType;
// Save optimization parameters info
_optimizationParameters = strategyInfo.OptimzationParameters;
//save iterations
_iterations = strategyInfo.Iterations;
//save population size
_populationSize = strategyInfo.PopulationSize;
int roundCount = 0;
// Execute GA for the specified No. of times
while (++roundCount <= strategyInfo.Rounds)
{
// Executes a single Round of Genetic Algorithm optimization
ExecuteGeneticAlgorithmRound();
}
}
public GeneticService(GeneticAlgorithmParameters parameters, Network network, PathFinder paths)
{
_parameters = parameters;
_network = network;
_pathFinder = paths;
_isNetworkFromFile = false;
_outputWriter = new OutputWriter();
}
public GeneticService(GeneticAlgorithmParameters parameters, Network network, PathFinder paths, string fileName)
{
_parameters = parameters;
_network = network;
_pathFinder = paths;
_currentFileName = fileName;
_isNetworkFromFile = true;
_outputWriter = new OutputWriter();
}
static void Main(string[] args)
{
var parser = new InputFileParser();
// get input network from user
Console.WriteLine("Please specify network topology which you want to solve.");
Console.Write("How many nodes are in the network? ");
var nodesCountInput = Console.ReadLine();
if (nodesCountInput is null)
{
Console.WriteLine("Invalid input.");
return;
}
var nodesCount = int.Parse(nodesCountInput);
var nodeConnections = new List <string>(nodesCount);
for (var i = 1; i <= nodesCount; i++)
{
Console.WriteLine($"Enter nodes reachable from node {i} separated by comma: ");
var currentNodeConnectionsInput = Console.ReadLine();
nodeConnections.Add($"{i}:{currentNodeConnectionsInput}");
}
var inputNetwork = parser.ReadNetworkFromInput(nodeConnections);
// find all node pairs
PathFinder pathFinder = new PathFinder();
for (int i = 1; i < inputNetwork.Nodes.Count; i++)
{
for (int j = i + 1; j < inputNetwork.Nodes.Count + 1; j++)
{
pathFinder.FindAllPaths(i, j, inputNetwork);
}
}
// all paths generated. Finding perfect combination.
GeneticAlgorithmParameters parameters = new GeneticAlgorithmParameters
{
InitialPopulationSize = 100,
CrossoverProbability = (float)0.2,
MutationProbability = (float)0.1,
RandomSeed = 4253,
LimitValue = 30,
StoppingCriteria = StoppingCriteria.NoImprovement
};
new GeneticService(parameters, inputNetwork, pathFinder).Solve();
}
static void Main(string[] args)
{
InputFileParser parser = new InputFileParser();
Console.WriteLine("Choose network to optimize:");
Console.WriteLine("1: Network_1.txt");
Console.WriteLine("2: Network_2.txt");
string choice = Console.ReadLine();
Network inputNetwork = new Network();
if (choice.Equals("1") || choice.Equals("2"))
{
inputNetwork = parser.ReadNetwork("Network_" + choice + ".txt");
}
else
{
return;
}
//Find all node pairs
AllPathFinder pathFinder = new AllPathFinder();
for (int i = 1; i < inputNetwork.Nodes.Count; i++)
{
for (int j = i + 1; j < inputNetwork.Nodes.Count + 1; j++)
{
pathFinder.FindAllPaths(i, j, inputNetwork);
}
}
// all paths generated. Finding perfect combination.
GeneticAlgorithmParameters parameters = new GeneticAlgorithmParameters();
parameters.InitialPopulationSize = 100;
parameters.CrossoverProbability = (float)0.2;
parameters.MutationProbability = (float)0.1;
parameters.RandomSeed = 4253;
parameters.LimitValue = 30;
parameters.StoppingCriteria = StoppingCriteria.NoImprovement;
string fileName = "Network_" + choice;
GeneticService geneticService = new GeneticService(parameters, inputNetwork, pathFinder, fileName);
geneticService.Solve();
}
public IEnumerable <ValidationResult> Validate(ValidationContext validationContext)
{
// Check the selected algorithm.
if (Algorithm == AnalysisAlgorithm.Greedy.ToString())
{
// Check if the parameters don't match the algorithm.
if (GreedyAlgorithmParameters == null)
{
// Return an error.
yield return(new ValidationResult("The parameters do not match the chosen algorithm.", new List <string> {
string.Empty
}));
}
// Get the validation results for the parameters.
var validationResults = GreedyAlgorithmParameters.Validate(validationContext);
// Go over each validation error.
foreach (var validationResult in validationResults)
{
// Return an error.
yield return(new ValidationResult(validationResult.ErrorMessage, validationResult.MemberNames.Select(item => $"Input.{nameof(GreedyAlgorithmParameters)}.{item}")));
}
}
else if (Algorithm == AnalysisAlgorithm.Genetic.ToString())
{
// Check if the parameters don't match the algorithm.
if (GeneticAlgorithmParameters == null)
{
// Return an error.
yield return(new ValidationResult("The parameters do not match the chosen algorithm.", new List <string> {
string.Empty
}));
}
// Get the validation results for the parameters.
var validationResults = GeneticAlgorithmParameters.Validate(validationContext);
// Go over each validation error.
foreach (var validationResult in validationResults)
{
// Return an error.
yield return(new ValidationResult(validationResult.ErrorMessage, validationResult.MemberNames.Select(item => $"Input.{nameof(GeneticAlgorithmParameters)}.{item}")));
}
}
}
private IMutationStrategy GetMutationStrategy(GeneticAlgorithmParameters geneticAlgorithmParameters)
{
switch (MutationStrategy.value)
{
case 0:
double mutationRangeD2 = double.Parse(MutationRange.text) / 2;
var parameters = new StandardMutationStrategyParameters()
{
MinSensorCount = geneticAlgorithmParameters.MinSensorCount,
MaxSensorCount = geneticAlgorithmParameters.MaxSensorCount,
MinHiddenLayerNeuronCount = geneticAlgorithmParameters.MinHiddenLayerNeuronCount,
MaxHiddenLayerNeuronCount = geneticAlgorithmParameters.MaxHiddenLayerNeuronCount,
EvolveNetworkTopology = EvolveNetworkTopology.isOn,
NeuralNetworkWeightInitializationDistribution = geneticAlgorithmParameters.NeuralNetworkWeightInitializationDistribution,
MutationRate = float.Parse(MutationRate.text),
NeuralNetworkWeightMutationDistribution = new ContinuousUniform(-mutationRangeD2, mutationRangeD2)
};
return(new StandardMutationStrategy(parameters));
default: throw new Exception("Invalid dropdown value.");
}
}
private static bool RunGeneticAlgorithm <T>(AbstractGeneticAlgorithm <T> algo,
bool verbose = true,
GeneticAlgorithmParameters param = null) where T : IChromosome
{
bool interrupted = false;
if (verbose)
{
param = GetParamGA(out interrupted);
if (interrupted)
{
return(false);
}
//param = new RecuitSimuleParameters(qap.N);
Console.WriteLine("Tout les paramètres sont entrées, commencer l'exécution ? ( o/n, y/n )");
Console.WriteLine(param.ToString());
string str = "";
if (!GetCorrectString(out str, (s) => IsValidation(s)))
{
return(false);
}
if (!IsYes(str))
{
return(false);
}
}
//Lancer l'exécution
algo.Verbose = true;
algo.WithLogs = true;
bestFitness = algo.Run(param).Fitness;
Console.WriteLine("Paramètres :");
Console.WriteLine(param.ToString());
Console.WriteLine("Résultats :");
Console.WriteLine(algo.Logs.FinalLog.ToString());
paramString = param.ToString();
resultString = algo.Logs.FinalLog.ToString();
//Save Results
SaveResults();
Console.WriteLine("Résultats sauvegardées dans " + GetResultPath() + "!");
//Save Logs
algo.Logs.SaveLogsTo(GetCSVPath());
Console.WriteLine("Logs sauvegardées dans " + GetCSVPath() + "!\n");
if (verbose)
{
Console.WriteLine("\nAppuyez sur une touche pour revenir au menu.");
Console.ReadKey();
Console.WriteLine();
}
return(true);
}
static void Main()
{
seed = new Random().Next(0, 1000); //random default seed
RandomSingleton.Instance.Seed = seed;
string[] line;
qap = new QuadraticAssignment(GetInstancePath(filename));
bestKnownSolution = new QuadraticAssignmentSolution(GetSolutionPath(filename), true);
List <string> allNames = new List <string>();
bool interrupted = false;
string allText;
while (state != State.Quit)
{
Console.WriteLine("------------");
Console.WriteLine(WelcomeMessage());
Console.WriteLine(GetInstanceInfo());
Console.WriteLine("------------");
line = GetLine();
RandomSingleton.Instance.ResetCurrentAlgoRandom();
switch (line[0].ToLower())
{
case "q":
case "quit":
state = State.Quit;
break;
case "s":
case "seed":
Console.WriteLine("Veuillez entrer un nouveau seed > 0:");
int newSeed = -1;
if (!GetCorrectInt(out newSeed, val => val > 0))
{
break;
}
seed = newSeed;
RandomSingleton.Instance.Seed = seed;
Console.WriteLine("Le nouveau seed est : " + seed);
break;
case "t":
case "tai":
bool found = false;
while (!found)
{
Console.WriteLine("Tapez le nom de l'instance de taillard à charger (sans son .dat):");
line = GetLine();
//early exit or quit
if (TryExitOrQuit(line[0]))
{
break;
}
found = FindTaillardInstance(line[0]);
if (!found)
{
Console.WriteLine("Fichier non trouvé ! Réessayer ou quitter (x or exit):");
}
else
{
//Update QAP
filename = line[0];
qap = new QuadraticAssignment(GetInstancePath(filename));
bestKnownSolution = new QuadraticAssignmentSolution(GetSolutionPath(filename), true);
}
}
break;
case "r":
case "rec":
case "recuit":
RunRecuit();
break;
case "tab":
case "tabou":
RunTabou();
break;
case "ga":
case "genetic":
algo = Algo.GA;
GeneticAlgorithmQAP ga = new GeneticAlgorithmQAP(qap);
Console.WriteLine("Vous avez sélectionné l'Algorithme Génétique !");
RunGeneticAlgorithm(ga);
break;
case "recga":
case "recuitga":
case "garecuit":
case "garec":
algo = Algo.RecuitGA;
RecuitSimule recuit = new RecuitSimule(qap);
recuit.Verbose = false;
GeneticAlgorithmRecuit garec = new GeneticAlgorithmRecuit(recuit);
Console.WriteLine("Vous avez sélectionné l'Algorithme Génétique pour RECUIT SIMULE !");
RunGeneticAlgorithm(garec);
break;
case "alltab":
case "allt":
allNames = GetAllInstances();
interrupted = false;
TabouParameters paramTabou = GetParamTabou(out interrupted);
allText = paramTabou.ToString();
if (interrupted)
{
break;
}
foreach (string instanceName in allNames)
{
ChangeInstance(instanceName);
paramTabou.InitialSol = new QuadraticAssignmentSolution(qap.N);
RunTabou(false, paramTabou);
allText += "\n\n" + instanceName + ":\n";
allText += "Resultats:\n" + resultString + ComparisonWithBest();
allText += "\n----";
}
allText = allText.Replace("\r\n", "\n").Replace("\r", "\n").Replace("\n", "\r\n");
string fullNameAllTabou = Path.Combine(resultPath, "all_" + algo.GetString() + "_s" + seed + ".txt");
System.IO.File.WriteAllText(fullNameAllTabou, allText);
break;
case "allrec":
case "allr":
allNames = GetAllInstances();
interrupted = false;
RecuitSimuleParameters paramRec = GetParamRecuit(out interrupted);
allText = paramRec.ToString();
if (interrupted)
{
break;
}
foreach (string instanceName in allNames)
{
ChangeInstance(instanceName);
paramRec.InitialSol = new QuadraticAssignmentSolution(qap.N);
RunRecuit(false, paramRec);
allText += "\n\n" + instanceName + ":\n";
allText += "Resultats:\n" + resultString + ComparisonWithBest();
allText += "\n----";
}
allText = allText.Replace("\r\n", "\n").Replace("\r", "\n").Replace("\n", "\r\n");
string fullNameAllRec = Path.Combine(resultPath, "all_" + algo.GetString() + "_s" + seed + ".txt");
System.IO.File.WriteAllText(fullNameAllRec, allText);
break;
case "allga":
case "allg":
allNames = GetAllInstances();
interrupted = false;
algo = Algo.GA;
GeneticAlgorithmParameters paramGA = GetParamGA(out interrupted);
allText = paramGA.ToString();
if (interrupted)
{
break;
}
foreach (string instanceName in allNames)
{
ChangeInstance(instanceName);
GeneticAlgorithmQAP geneticAlgo = new GeneticAlgorithmQAP(qap);
//paramGA. = new QuadraticAssignmentSolution(qap.N);
RunGeneticAlgorithm <QuadraticAssignmentSolution>(geneticAlgo, false, paramGA);
allText += "\n\n" + instanceName + ":\n";
allText += "Resultats:\n" + resultString + ComparisonWithBest();
allText += "\n----";
}
allText = allText.Replace("\r\n", "\n").Replace("\r", "\n").Replace("\n", "\r\n");
string fullNameAllGA = Path.Combine(resultPath, "all_" + algo.GetString() + "_s" + seed + ".txt");
System.IO.File.WriteAllText(fullNameAllGA, allText);
break;
case "autoga":
string instanceCurrent = "tai12a";
ChangeInstance(instanceCurrent);
algo = Algo.GA;
List <GeneticAlgorithmParameters> paramsGa = new List <GeneticAlgorithmParameters>();
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 0, 0));
paramsGa.Add(new GeneticAlgorithmParameters(300, 25, 0.005, 0, 0));
paramsGa.Add(new GeneticAlgorithmParameters(100, 50, 0.005, 0, 0));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.1, 0, 0));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 5, 0));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 20, 0));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 0, 20));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 0, 50));
paramsGa.Add(new GeneticAlgorithmParameters(100, 25, 0.005, 2, 20));
GeneticAlgorithmQAP geneticAlgoAll = new GeneticAlgorithmQAP(qap);
allText = instanceCurrent + "\n";
foreach (GeneticAlgorithmParameters parGA in paramsGa)
{
RunGeneticAlgorithm <QuadraticAssignmentSolution>(geneticAlgoAll, false, parGA);
allText += parGA.ToString();
allText += "\nResultats:\n" + resultString + ComparisonWithBest();
allText += "\n----";
}
allText = allText.Replace("\r\n", "\n").Replace("\r", "\n").Replace("\n", "\r\n");
string fulenamega = Path.Combine(resultPath, "all_" + algo.GetString() + "_s" + seed + ".txt");
System.IO.File.WriteAllText(fulenamega, allText);
break;
default:
Console.WriteLine("\nCommande invalide!");
break;
}
}
}
public GeneticAlgorithmViewModel(GeneticAlgorithmParameters gap)
{
this.Parameters = gap;
this.CreateGACommand = new CreateGACommand(this);
}
