static void Main(string[] args)
{
//Thread.CurrentThread.CurrentCulture = new CultureInfo("en-us");
//CultureInfo.DefaultThreadCurrentUICulture = CultureInfo.GetCultureInfo("en-US");
//Thread.CurrentThread.CurrentUICulture = new CultureInfo("en-US");
Thread.CurrentThread.CurrentCulture = CultureInfo.InvariantCulture;
var experimentParameters = new ExperimentParameters(
numberOfDimensions: 2,
basePopulationSize: 10,
offspringPopulationSize: 100,
numberOfGenerations: 100,
//seed: new Random().Next(),
seed: 1,
//seed: 1971749862,
//oneFifthRuleCheckInterval: 10,
//individualLearningRate: 0.1, //0.47287080450158792
//globalLearningRate: 0.1, //0.31622776601683794
typeOfParentsSelection: ParentsSelectionType.Even,
typeOfStdDeviationsRecombination: RecombinationType.Intermediate,
typeOfRotationsRecombination: RecombinationType.Intermediate,
useRedundantConstraintsRemoving: true,
useDataNormalization: true,
allowQuadraticTerms: true,
useRecombination: false,
trackEvolutionSteps: true,
useSeeding: true,
numberOfParentsSolutionsToSelect: 5,
numberOfPositivePoints: 500,
numberOfNegativePoints: 1500,
//ballnBoundaryValue: 10,
//simplexnBoundaryValue: 10,
typeOfBenchmark: BenchmarkType.Simplexn);
var enginesFactory = new EnginesFactory();
var engine = enginesFactory.Create(experimentParameters);
var distanceCalculator = new CanberraDistanceCalculator();
var positivePointsGenerator = new PositivePointsGenerator();
var positiveTrainingPoints = positivePointsGenerator.GeneratePoints(experimentParameters.NumberOfPositivePoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
var negativePointsGenerator = new NegativePointsGenerator(positiveTrainingPoints, distanceCalculator, new NearestNeighbourDistanceCalculator(distanceCalculator));
//var negativeTrainingPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark.Domains);
var negativeTrainingPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
var trainingPoints = positiveTrainingPoints.Concat(negativeTrainingPoints).ToArray();
var testPointsGenerator = new TestPointsGenerator();
var testPoints = testPointsGenerator.GeneratePoints(experimentParameters.NumberOfTestPoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
Console.WriteLine("Evolution starts!");
var mathModel = engine.SynthesizeModel(trainingPoints);
var statistics = engine.EvaluateModel(testPoints);
var visualization = PrepareVisualisation(experimentParameters, engine, positiveTrainingPoints, negativeTrainingPoints,
testPoints);
visualization.Show();
Console.WriteLine("==========================================");
Console.WriteLine("=============== PARAMETERS ===============");
Console.WriteLine("==========================================");
Console.WriteLine(experimentParameters.ToPrintableString());
Console.WriteLine();
Console.WriteLine("==========================================");
Console.WriteLine("============= REFERENCE MODEL ============");
Console.WriteLine("==========================================");
Console.WriteLine(mathModel.ReferenceModelInLpFormat);
Console.WriteLine();
Console.WriteLine("==========================================");
Console.WriteLine("============ SYNTHESIZED MODEL ===========");
Console.WriteLine("==========================================");
Console.WriteLine(mathModel.SynthesizedModelInLpFormat);
Console.WriteLine();
Console.WriteLine("==========================================");
Console.WriteLine("=============== STATISTICS ===============");
Console.WriteLine("==========================================");
Console.WriteLine(statistics.ToPrintableString());
Console.WriteLine();
Console.ReadKey();
}
private static void RunExperiment(ExperimentParameters experimentParameters)//, DatabaseContext databaseX)
{
var database = new DatabaseContext(DatabaseFullPath);
var version = new Version(DateTime.Now);
var stoper = new Stopwatch();
//if (database.Exists(experimentParameters))
//{
// Console.WriteLine("BYLEM WIDZIALEM");
// //database.Dispose();
// return;
//}
database.Insert(version);
database.Insert(experimentParameters);
database.Insert(experimentParameters.EvolutionParameters);
IDictionary <int, EvolutionStep> evolutionSteps = null;
try
{
var enginesFactory = new EnginesFactory();
var engine = enginesFactory.Create(experimentParameters);
var distanceCalculator = new CanberraDistanceCalculator();
var positivePointsGenerator = new PositivePointsGenerator();
stoper.Restart();
var positiveTrainingPoints = positivePointsGenerator.GeneratePoints(experimentParameters.NumberOfPositivePoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
stoper.Stop();
engine.Statistics.PositiveTrainingPointsGenerationTime = stoper.Elapsed;
var negativePointsGenerator = new NegativePointsGenerator(positiveTrainingPoints, distanceCalculator, new NearestNeighbourDistanceCalculator(distanceCalculator));
//var negativeTrainingPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark.Domains);
stoper.Restart();
var negativeTrainingPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
stoper.Stop();
engine.Statistics.NegativeTrainingPointsGenerationTime = stoper.Elapsed;
//Console.WriteLine("Evolution starts!");
var trainingPoints = positiveTrainingPoints.Concat(negativeTrainingPoints).ToArray();
stoper.Restart();
var mathModel = engine.SynthesizeModel(trainingPoints);
stoper.Stop();
engine.Statistics.TotalSynthesisTime = stoper.Elapsed;
database.Insert(mathModel);
var testPointsGenerator = new TestPointsGenerator();
stoper.Restart();
var testPoints = testPointsGenerator.GeneratePoints(experimentParameters.NumberOfTestPoints, engine.Benchmark.Domains, engine.Benchmark.Constraints);
stoper.Stop();
engine.Statistics.TestPointsGenerationTime = stoper.Elapsed;
var statistics = engine.EvaluateModel(testPoints);
database.Insert(statistics);
evolutionSteps = engine.CoreEvolutionSteps;
//Logger.Instance.Log(experimentParameters);
//Logger.Instance.Log(mathModel);
//Logger.Instance.Log(statistics);
//database.Insert(Logger.Instance.GetLogAsString());
}
catch (Exception exception)
{
database.Insert(exception);
}
database.Save();
if (evolutionSteps != null && experimentParameters.TrackEvolutionSteps)
{
Logger.Instance.Log(evolutionSteps);
var logsFullPath = Path.GetFullPath(LogsDirPath + experimentParameters.GetFileName("Log", ".cmplog"));
File.WriteAllText(logsFullPath, StringCompressor.CompressString(Logger.Instance.GetLogAsString()));
}
//GlobalStoper.Stop();
//Console.WriteLine(GlobalStoper.ElapsedMilliseconds);
//Console.ReadKey();
database.Dispose();
}
static void Main(string[] args)
{
CultureInfo.DefaultThreadCurrentCulture = CultureInfo.InvariantCulture;//CultureInfo.GetCultureInfo("en-US");
var dbPath = Path.GetFullPath("Database test.db");
var databaseContext = new DatabaseContext(dbPath);
//var experimentParameters = Arguments.GetExperimentParameters();
var experimentParameters = new ExperimentParameters(2, 10,
typeOfMutation: ExperimentParameters.MutationType.Correlated,
typeOfBenchmark: ExperimentParameters.BenchmarkType.Simplexn,
stepThreshold: 0.1, numberOfGenerations: 1,
basePopulationSize: 15,
//basePopulationSize: 3,
offspringPopulationSize: 100,
//offspringPopulationSize: 20,
globalLerningRate: 1 / Math.Sqrt(2 * 2),
//globalLerningRate: 0.7,
individualLearningRate: 1 / Math.Sqrt(2 * Math.Sqrt(2)),
//individualLearningRate: 0.8,
numberOfPositivePoints: 10,
numberOfNegativePoints: 10,
ballnBoundaryValue: 2,
//simplexnBoundaryValue: 2,
seed: 1,
useRecombination: false,
numberOfParentsSolutionsToSelect: 5
//domainSamplingStep: 1
);
if (databaseContext.Exists(experimentParameters))
{
return;
}
if (experimentParameters.TypeOfBenchmark == ExperimentParameters.BenchmarkType.Other)
{
experimentParameters.ConstraintsToPointsGeneration = ReferenceModelsExamples.Example1;
}
var version = new Version(DateTime.Now, experimentParameters.GetHashString());
var stoper = new Stopwatch();
var distanceCalculator = new CanberraDistanceCalculator();
var engine = EngineFactory.GetEngine(experimentParameters);
stoper.Restart();
var positivePointsGenerator = new PositivePointsGenerator();
var positiveTrainingPoints = positivePointsGenerator.GeneratePoints(experimentParameters.NumberOfPositivePoints, engine.Benchmark);
stoper.Stop();
engine.Statistics.PositiveTrainingPointsGenerationTime = stoper.Elapsed;
stoper.Restart();
var negativePointsGenerator = new NegativePointsGenerator(positiveTrainingPoints, distanceCalculator, new NearestNeighbourDistanceCalculator(distanceCalculator));
var negativeTrainingPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark);
stoper.Stop();
engine.Statistics.NegativeTrainingPointsGenerationTime = stoper.Elapsed;
var trainingPoints = positiveTrainingPoints.Concat(negativeTrainingPoints).ToArray();
var mathModel = engine.SynthesizeModel(trainingPoints);
stoper.Restart();
var positiveTestPoints = positivePointsGenerator.GeneratePoints(experimentParameters.NumberOfPositivePoints, engine.Benchmark);
var negativeTestPoints = negativePointsGenerator.GeneratePoints(experimentParameters.NumberOfNegativePoints, engine.Benchmark);
var testPoints = positiveTestPoints.Concat(negativeTestPoints).ToArray();
stoper.Stop();
engine.Statistics.TestPointsGenerationTime = stoper.Elapsed;
var statistics = engine.EvaluateModel(testPoints);
databaseContext.Insert(version);
databaseContext.Insert(experimentParameters);
databaseContext.Insert(statistics);
databaseContext.Insert(mathModel);
databaseContext.Save();
databaseContext.Dispose();
TryGetVisibleResults(engine, positiveTrainingPoints, negativeTrainingPoints);
}