private GeneticAlgorithm PrepareGeneticAlgorithm(string[][] board)
{
double[][] parsedBoard = this.ParseBoardToDouble(board);
var chromosome = new TicTacToeChromosome(parsedBoard);
var population = new Population(50, 100, chromosome);
var fitness = new TicTacToeFitness
{
Minimum = BOARD_SIZE
};
var selection = new EliteSelection();
var crossover = new UniformCrossover(0.5f);
var mutation = new UniformMutation();
var termination = new OrTermination(new ITermination[]
{
new FitnessStagnationTermination(3000),
new FitnessThresholdTermination(4),
new FitnessThresholdTermination(3),
new FitnessThresholdTermination(2)
});
var geneticAlgorithm = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = termination
};
#if DEBUG
geneticAlgorithm.GenerationRan += (sender, e) =>
{
var bestChromosome = geneticAlgorithm.BestChromosome as TicTacToeChromosome;
Console.WriteLine($"Generation { geneticAlgorithm.GenerationsNumber }: Position ({ bestChromosome.Position.X }, { bestChromosome.Position.Y }) = { bestChromosome.Fitness }");
};
#endif
return(geneticAlgorithm);
}
public void AddTermination_Null_Exception()
{
var target = new OrTermination();
ExceptionAssert.IsThrowing(new ArgumentNullException("termination"), () =>
{
target.AddTermination(null);
});
}
public ITermination Or()
{
var target = new OrTermination(_falseTermination, _trueTermination);
var ga = Substitute.For <IGeneticAlgorithm>();
target.HasReached(ga);
return(target);
}
public void AddTermination_Null_Exception()
{
var target = new OrTermination();
var actual = Assert.Catch <ArgumentNullException>(() =>
{
target.AddTermination(null);
});
Assert.AreEqual("termination", actual.ParamName);
}
public void ToString_NoArgs_State()
{
var t1 = new AndTermination();
var t2 = new OrTermination();
var t3 = new AndTermination();
var target = new AndTermination(t1, t2, t3);
Assert.AreEqual("AndTermination (AndTermination (), OrTermination (), AndTermination ())", target.ToString());
}
public void HasReached_LessThan2Terminations_Exception()
{
var target = new OrTermination();
target.AddTermination(Substitute.For <ITermination>());
Assert.Catch <InvalidOperationException>(() =>
{
target.HasReached(Substitute.For <IGeneticAlgorithm>());
}, "The OrTermination needs at least 2 terminations to perform. Please, add the missing terminations.");
}
public void HasReached_LessThan2Terminations_Exception()
{
var target = new OrTermination();
target.AddTermination(MockRepository.GenerateMock <ITermination>());
ExceptionAssert.IsThrowing(new InvalidOperationException("The OrTermination needs at least 2 terminations to perform. Please, add the missing terminations."), () =>
{
target.HasReached(MockRepository.GenerateMock <IGeneticAlgorithm>());
});
}
private IList <CheckersMove> GetNewMoves()
{
int attempts = m_maxMoveAttempts;
var termination =
new OrTermination(
new FitnessStagnationTermination(10),
new AndTermination(
new FitnessThresholdTermination(0.1),
new GenerationNumberTermination(GA.GenerationsNumber + 20)
)
);
GA.Termination = termination;
Debug.Log("Running GA...");
if (GA.State == GeneticAlgorithmState.NotStarted)
{
GA.Start();
}
else
{
GA.Resume();
}
Debug.Log("Fitness: " + GA.BestChromosome.Fitness);
Debug.Log("Generations: " + GA.GenerationsNumber);
if (GA.BestChromosome.Fitness <= 0)
{
Debug.LogError("Lower than zero.");
}
attempts--;
if (GA.BestChromosome.Fitness == 0)
{
HudController.IsGameOver = true;
}
return((GA.BestChromosome as CheckersChromosome).Moves);
}
public void HasReached_OnlyOneTerminationsHasReached_True()
{
var target = new OrTermination();
var ga = Substitute.For <IGeneticAlgorithm>();
var t1 = Substitute.For <ITermination>();
t1.HasReached(ga).ReturnsForAnyArgs(false);
target.AddTermination(t1);
var t2 = Substitute.For <ITermination>();
t2.HasReached(ga).ReturnsForAnyArgs(true);
target.AddTermination(t2);
var t3 = Substitute.For <ITermination>();
t3.HasReached(ga).ReturnsForAnyArgs(false);
target.AddTermination(t3);
Assert.IsTrue(target.HasReached(ga));
}
public void HasReached_OnlyOneTerminationsHasReached_True()
{
var target = new OrTermination();
var ga = MockRepository.GenerateMock <IGeneticAlgorithm>();
var t1 = MockRepository.GenerateMock <ITermination>();
t1.Expect(t => t.HasReached(ga)).IgnoreArguments().Return(false);
target.AddTermination(t1);
var t2 = MockRepository.GenerateMock <ITermination>();
t2.Expect(t => t.HasReached(ga)).IgnoreArguments().Return(true);
target.AddTermination(t2);
var t3 = MockRepository.GenerateMock <ITermination>();
t3.Expect(t => t.HasReached(ga)).IgnoreArguments().Return(false);
target.AddTermination(t3);
Assert.IsTrue(target.HasReached(ga));
}
public Test UseAlgorithm(List <int> oblasti, List <double> zastupljenost, int Test_Length)
{
Random r = new Random(10);
IDataAccess dataAccess = DataAccess.DataAccess.GetInstance();
List <Question> qs = new List <Question>();
for (int i = 0; i < Test_Length; i++)
{
qs.Add(dataAccess.GetQuestionById(1 + r.Next(1000)));
}
Test adamTest = new Test(qs, Test_Length);
List <double> zastupljenost_kao_br_pitanja = new List <double>();
foreach (var vrednost in zastupljenost)
{
zastupljenost_kao_br_pitanja.Add(Math.Round(vrednost * Test_Length + 0.01));
}
var fitness = new TestFitness(oblasti, zastupljenost_kao_br_pitanja, 0);
IGenerationStrategy generationStrategy = new PerformanceGenerationStrategy(3);
var population = new TestPopulation(20, 50, adamTest, oblasti)
{
GenerationStrategy = generationStrategy
};
ISelection selection = new EliteSelection();
ICrossover crossover = new TestCrossover(oblasti, zastupljenost_kao_br_pitanja, 8, 4);
IMutation mutation = new TestMutation(oblasti, zastupljenost_kao_br_pitanja);
ITermination termination = new OrTermination(new ITermination[] { new TestTermination(100, oblasti),
new FitnessThresholdTermination(oblasti.Count),
new GenerationNumberTermination(3000) });
GeneticAlgorithm ga = new GeneticAlgorithm(population, fitness, selection, crossover, mutation)
{
Termination = termination,
CrossoverProbability = 1f,
MutationProbability = 0.7f
};
double oldFitness = 0;
List <Test> tests = new List <Test>();
//Callback
ga.GenerationRan += (sender, arg) =>
{
if (ga.BestChromosome.Fitness >= oldFitness)
{
oldFitness = ga.BestChromosome.Fitness.Value;
tests.Add(ga.BestChromosome as Test);
//Console.WriteLine($"Current fitness: {ga.BestChromosome.Fitness} for generation : {ga.GenerationsNumber}");
}
};
ga.Start();
if (oldFitness == 0)
{
return(null);
}
var lista = tests.Where(x => !x.HasDuplicate()).OrderByDescending(x => x.Fitness.Value).ToList();
Test bestChromosome = lista.FirstOrDefault();
foreach (var q in bestChromosome.questions)
{
q.Izabrano = true;
}
return(bestChromosome);
}
/**
*
* Genetic Algorithm with Revit API
*
* This sample is using GeneticSharp Library (github.com/giacomelli/GeneticSharp)
* The MIT License (MIT)
* Copyright (c) 2013 Diego Giacomelli
*/
public void GridObjectPlacement(DesignAutomationData data)
{
m_doc = data.RevitDoc;
m_app = data.RevitApp;
InputData inputParameters = JsonConvert.DeserializeObject <InputData>(File.ReadAllText("params.json"));
// Family Symbol
Document familyProjectDoc = m_app.OpenDocumentFile("family.rvt");
string tempFamilyName = Path.GetFileNameWithoutExtension(inputParameters.FamilyFileName) + ".rfa";
ModelPath tempFamilyModelPath = ModelPathUtils.ConvertUserVisiblePathToModelPath(tempFamilyName);
FamilySymbol tempFamilySymbol = null;
FilteredElementCollector familyInstanceCollector
= new FilteredElementCollector(familyProjectDoc)
.WhereElementIsNotElementType()
.OfCategory(BuiltInCategory.OST_Furniture)
.OfClass(typeof(FamilyInstance));
foreach (Element familyInstanceElem in familyInstanceCollector)
{
FamilyInstance fi = familyInstanceElem as FamilyInstance;
Element superComponent = fi.SuperComponent;
if (superComponent == null)
{
tempFamilySymbol = fi.Symbol;
Family family = tempFamilySymbol.Family;
Document familyDoc = familyProjectDoc.EditFamily(family);
Family loadedFamily = familyDoc.LoadFamily(m_doc);
ISet <ElementId> familySymbolIds = loadedFamily.GetFamilySymbolIds();
foreach (ElementId familySymbolId in familySymbolIds)
{
FamilySymbol familySymbol = m_doc.GetElement(familySymbolId) as FamilySymbol;
m_familySymbol = familySymbol;
}
break;
}
}
if (!m_familySymbol.IsActive)
{
using (Transaction tx = new Transaction(m_doc))
{
tx.Start("Transaction Activate Family Symbol");
m_familySymbol.Activate();
tx.Commit();
}
}
// Room
m_room = m_doc.GetElement(inputParameters.RoomUniqueId) as Room;
// Level
m_level = m_doc.GetElement(m_room.LevelId) as Level;
// View
ElementId viewId = m_level.FindAssociatedPlanViewId();
if (viewId != null)
{
m_view = m_doc.GetElement(viewId) as View;
}
// Selected Placement Method
m_selectedPlacementMethod = int.Parse(inputParameters.GridTypeId);
// Construct Chromosomes with 3 params, m_objectDistanceX, m_objectDistanceY, m_selectedPlacementMethod
var chromosome = new FloatingPointChromosome(
new double[] { double.Parse(inputParameters.DistanceXMinParam), double.Parse(inputParameters.DistanceYMinParam), double.Parse(inputParameters.DistanceWallMinParam) },
new double[] { double.Parse(inputParameters.DistanceXMaxParam), double.Parse(inputParameters.DistanceYMaxParam), double.Parse(inputParameters.DistanceWallMaxParam) },
new int[] { 32, 32, 32 },
new int[] { 2, 2, 2 });
// Population Settings
//
// The population size needs to be 'large enough'.
// The question of when a population is large enough is difficult to answer.
// Generally, it depends on the project, the number of genes, and the gene value range.
// A good rule of thumb is to set the population size to at least 3x the number of inputs.
// If the results don't start to converge to an answer, you may need to increase the population size.
//
// by www.generativedesign.org/02-deeper-dive/02-04_genetic-algorithms/02-04-02_initialization-phase
var population = new Population(8, 12, chromosome);
// Fitness Function Settings
//
// Call CreateObjectPlacementPointList() and get count of points.
// This sample maximize a number of objects to place in a room.
//
// A fitness function is used to evaluate how close (or far off) a given design solution is from meeting the designer�fs goals.
//
// by www.generativedesign.org/02-deeper-dive/02-04_genetic-algorithms/02-04-03_evaluation-phase
var fitness = new FuncFitness((c) =>
{
var fc = c as FloatingPointChromosome;
var values = fc.ToFloatingPoints();
m_objectDistanceX = values[0];
m_objectDistanceY = values[1];
m_minimumDistanceFromObjectToWall = values[2];
List <XYZ> objectPlacementPointList = CreateObjectPlacementPointList();
return(objectPlacementPointList.Count);
});
var selection = new EliteSelection();
var crossover = new UniformCrossover(0.5f);
var mutation = new FlipBitMutation();
// Termination Condition Settings
//
// To finish the process in half an hour, this sample sets 2 conditions.
var termination = new OrTermination(
new GenerationNumberTermination(20),
new TimeEvolvingTermination(TimeSpan.FromMinutes(10)));
// Construct GeneticAlgorithm
var ga = new GeneticAlgorithm(
population,
fitness,
selection,
crossover,
mutation);
ga.Termination = termination;
Console.WriteLine("Generation: objectDistanceX, objectDistanceY, minimumDistanceFromObjectToWall = objectCount");
var latestFitness = 0.0;
// Callback Function of Generation Result
ga.GenerationRan += (sender, e) =>
{
var bestChromosome = ga.BestChromosome as FloatingPointChromosome;
var bestFitness = bestChromosome.Fitness.Value;
if (bestFitness != latestFitness)
{
latestFitness = bestFitness;
var phenotype = bestChromosome.ToFloatingPoints();
m_objectDistanceX = phenotype[0];
m_objectDistanceY = phenotype[1];
m_minimumDistanceFromObjectToWall = phenotype[2];
Console.WriteLine(
"Generation {0,2}: objectDistanceX: {1}, objectDistanceY: {2}, minimumDistanceFromObjectToWall: {3} = objectCount: {4}",
ga.GenerationsNumber,
m_objectDistanceX,
m_objectDistanceY,
m_minimumDistanceFromObjectToWall,
bestFitness
);
List <XYZ> objectPlacementPointList = CreateObjectPlacementPointList();
using (Transaction tx = new Transaction(m_doc))
{
tx.Start("Transaction Create Family Instance");
m_view.SetCategoryHidden(new ElementId(BuiltInCategory.OST_Furniture), false);
foreach (XYZ point in objectPlacementPointList)
{
FamilyInstance fi = m_doc.Create.NewFamilyInstance(point, m_familySymbol, m_level, Autodesk.Revit.DB.Structure.StructuralType.NonStructural);
m_doc.Regenerate();
BoundingBoxXYZ fiBB = fi.get_BoundingBox(m_view);
double xDiff = fiBB.Max.X - fiBB.Min.X;
double yDiff = fiBB.Max.Y - fiBB.Min.Y;
if (m_objectDistanceX / m_objectDistanceY >= 1.2 && yDiff / xDiff >= 1.2)
{
LocationPoint location = fi.Location as LocationPoint;
if (null != location)
{
XYZ fiLocationPoint = location.Point;
XYZ axisPoint = new XYZ(fiLocationPoint.X, fiLocationPoint.Y, fiLocationPoint.Z + 10);
Line axis = Line.CreateBound(fiLocationPoint, axisPoint);
location.Rotate(axis, Math.PI / 2.0);
}
}
else if (m_objectDistanceY / m_objectDistanceX >= 1.2 && xDiff / yDiff >= 1.2)
{
LocationPoint location = fi.Location as LocationPoint;
if (null != location)
{
XYZ fiLocationPoint = location.Point;
XYZ axisPoint = new XYZ(fiLocationPoint.X, fiLocationPoint.Y, fiLocationPoint.Z + 10);
Line axis = Line.CreateBound(fiLocationPoint, axisPoint);
location.Rotate(axis, Math.PI / 2.0);
}
}
}
DWGExportOptions dwgOptions = new DWGExportOptions();
ICollection <ElementId> views = new List <ElementId>();
views.Add(m_view.Id);
m_doc.Export(Directory.GetCurrentDirectory() + "\\exportedDwgs", m_level.Name + "_" + m_room.Name + "_Gen " + ga.GenerationsNumber + "_" + DateTime.Now.ToString("yyyyMMddHHmmss"), views, dwgOptions);
tx.RollBack();
}
}
};
ga.Start();
}
