public void Step(int MaxThreads = 4)
{
ParallelOptions pOptions = new ParallelOptions
{
MaxDegreeOfParallelism = MaxThreads
};
Parallel.For(0, CurrentGeneration.Height, pOptions, y =>
{
for (int x = 0; x < CurrentGeneration.Width; x++)
{
int AliveNeighbors = CurrentGeneration.GetAliveNeighbours(x, y);
bool isAlive = CurrentGeneration.IsAlive(x, y);
if (isAlive && (AliveNeighbors == 2 || AliveNeighbors == 3))
{
NextGeneration.SetPixel(x, y, true);
}
else if (!isAlive && AliveNeighbors == 3)
{
NextGeneration.SetPixel(x, y, true);
}
else
{
NextGeneration.SetPixel(x, y, false);
}
}
});
}
public static void Initialize()
{
thisEvolution = new Evolution();
thisEvolution.Start();
Loading = true;
Random RNG = new Random();
EvolviMaxAmount = 1f;
EvolviLoadedAmount = 0;
CurrentGeneration.Clear();
for (int i = 0; i < evolviAmount; i++)
{
CurrentGeneration.Add(new Evolvi(new Vector2(RNG.Next(20, 2000 - 20), RNG.Next(20, 2000 - 20))));
}
CurrentFood.Clear();
for (int i = 0; i < foodAmount; i++)
{
CurrentFood.Add(new Food(new Vector2(RNG.Next(20, 2000 - 20), RNG.Next(20, 2000 - 20))));
Thread.Sleep(10);
}
Loading = false;
}
public void NextGeneration()
{
SpeciesCollection.ClearSpecies();
foreach (var genome in CurrentGeneration)
{
SpeciesCollection.Speciate(genome);
}
SpeciesCollection.ClearEmptySpecies();
int population = Population;
var nextGeneration = SpeciesCollection.Selection(population);
LastGenerationWithSpecies = nextGeneration;
// fill up when there is no species or
var remaining = population - nextGeneration.Count;
for (int i = 0; i < remaining; i++)
{
nextGeneration.Add(Genome.CreateDefault(this), null);
}
//Populate(population);
LastFillUp = remaining;
LastGeneration.Clear();
LastGeneration.AddRange(CurrentGeneration);
CurrentGeneration.Clear();
CurrentGeneration.AddRange(nextGeneration.Select(x => x.Key));
Generation++;
}
public string CandidateInfo(Candidate candidate)
{
return($"X1 = {candidate.X1}, \n" +
$"X2 = {candidate.X2}, \n" +
$"X3 = {candidate.X3}, \n" +
$"Fitness = {candidate.Fitness}, \n" +
$"Average of Generation = {CurrentGeneration.Select(x => x.Fitness).Average()}\n");
}
private void InitializePopulation()
{
for (int i = 0; i < PopulationSize; i++)
{
var candidate = new Candidate();
CurrentGeneration.Add(candidate);
}
}
public void Populate(int population)
{
for (var i = 0; i < population; i++)
{
var genome = Genome.CreateDefault(this);
CurrentGeneration.Add(genome);
}
}
private void ToggleCellInNextGeneration(int x, int y)
{
var isAliveInNextGen = CurrentGeneration.IsCellReincarnated(x, y);
if (isAliveInNextGen)
{
NextGeneration[x, y].Toggle();
}
}
private bool CheckPopulationForZero()
{
if (CurrentGeneration.CalculatePopulation() != 0)
{
return(false);
}
Console.WriteLine("All dead! All dead!");
return(true);
}
private void CheckRowGrowth(Borders border)
{
var aliveCellGroupCount
= CurrentGeneration.GetConsecutiveHorizontalCellCount(border);
if (aliveCellGroupCount >= 1)
{
AddRow(border);
}
}
private void CheckColumnGrowth(Borders border)
{
var aliveCellGroupCount
= CurrentGeneration.GetConsecutiveVerticalCellCount(border);
if (aliveCellGroupCount >= 1)
{
AddColumn(border);
}
}
/// <summary>
/// Ends the current generation.
/// </summary>
public virtual void EndCurrentGeneration()
{
CurrentGeneration.End(MaxSize);
if (BestChromosome != CurrentGeneration.BestChromosome)
{
BestChromosome = CurrentGeneration.BestChromosome;
OnBestChromosomeChanged(EventArgs.Empty);
}
}
public void ApplyFitnessJudgement(FitnessJudgement judgement)
{
int weight = _settingsGa.GetWeightOfFitnessJudgement(judgement);
if (weight > 0)
{
_currentCa.Fitness = weight;
CurrentGeneration.Add(_currentCa);
}
GenerateNewCa();
}
/// <summary>
/// Ends the current generation.
/// </summary>
public virtual void EndCurrentGeneration()
{
CurrentGeneration.End(CurrentGeneration.Genomes.Count);
if (BestGenome != CurrentGeneration.BestGenome)
{
BestGenome = CurrentGeneration.BestGenome;
OnBestGenomeChanged(EventArgs.Empty);
}
}
public void CreateEmpty(int universeSize)
{
Size = universeSize;
for (int column = 0; column < universeSize; column++)
{
for (int raw = 0; raw < universeSize; raw++)
{
CurrentGeneration.Add(new Point(raw, column), new Cell());
NextGeneration.Add(new Point(raw, column), new Cell());
}
}
}
public void Reset(int population, int inputs, int outputs)
{
Generation = 0;
InnovationId = 0;
GenomeId = 0;
NodeId = 0;
InitialNodes.Clear();
CurrentGeneration.Clear();
LastGeneration.Clear();
SpeciesCollection.SpeciesItems.Clear();
Connections.Clear();
Initialize(population, inputs, outputs);
}
private bool InBounds(Point startCoords, int figureSize)
{
for (int column = startCoords.Y; column < startCoords.Y + figureSize; column++)
{
for (int raw = startCoords.X; raw < startCoords.X + figureSize; raw++)
{
if (!CurrentGeneration.ContainsKey(new Point(raw, column)))
{
return(false);
}
}
}
return(true);
}
/// <summary>
/// Ends the current generation.
/// </summary>
public void EndCurrentGeneration()
{
CurrentGeneration.End(MaxSize);
if (BestChromosome != CurrentGeneration.BestChromosome)
{
BestChromosome = CurrentGeneration.BestChromosome;
if (BestChromosomeChanged != null)
{
BestChromosomeChanged(this, EventArgs.Empty);
}
}
}
public void StartGame()
{
CurrentGeneration.FillField();
while (true)
{
CurrentGeneration.PrintField();
if (CheckPopulationForZero())
{
break;
}
CurrentGeneration.BornNextGeneration(NextGeneration);
TryToFixStalemate();
ShiftGenerations();
Thread.Sleep(100);
}
}
public void EvaluatePopulation()
{
List <Chromosome> chromosomes = CurrentGeneration.FindAll(x => x.FitnessOutdated);
if (chromosomes.Count == 0)
{
return;
}
int evaluations = (int)((double)chromosomes.Count * EvaluationRate);
List <Chromosome> orderedChromosomes = chromosomes.OrderByDescending(x => x.Priority).ToList();
for (int i = 0; i < orderedChromosomes.Count; i++)
{
Chromosome c = orderedChromosomes[i];
if (i < evaluations) //Chosen for evaluation
{
c.Fitness = FitnessFunction.Evaluate(c);
c.Priority = 0;
c.FitnessOutdated = false;
totalEvaluations++;
}
else
{
c.Priority += NonEvaluationWeigth;
}
//If the maximum number is reached while evaluating the population, the process is interrupted.
if (totalEvaluations == MaxEvaluations)
{
break;
}
}
if (EvaluationRate < 1)
{
EvaluationRate = EvaluationRate + (1 - EvaluationRate) * 0.001;
}
if (EvaluationRate > 1)
{
EvaluationRate = 1;
}
}
/// <summary>
/// Checks square around cell and returns true, if cell is alive.
/// </summary>
/// <param name="cellCoord"></param>
/// <param name="currentGeneration"></param>
/// <returns></returns>
private bool CheckCellState(Point cellCoord)
{
int livesCounter = 0;
var neighbours = new Point []
{
new Point(cellCoord.X - 1, cellCoord.Y - 1),
new Point(cellCoord.X, cellCoord.Y - 1),
new Point(cellCoord.X + 1, cellCoord.Y - 1),
new Point(cellCoord.X - 1, cellCoord.Y),
new Point(cellCoord.X + 1, cellCoord.Y),
new Point(cellCoord.X - 1, cellCoord.Y + 1),
new Point(cellCoord.X, cellCoord.Y + 1),
new Point(cellCoord.X + 1, cellCoord.Y + 1),
};
foreach (var neighbour in neighbours)
{
if (CurrentGeneration.ContainsKey(neighbour) && CurrentGeneration[neighbour].Alive)
{
livesCounter++;
}
}
if ((livesCounter == 3 || livesCounter == 2) && CurrentGeneration[new Point((cellCoord.X + Size) % Size, (cellCoord.Y + Size) % Size)].Alive)
{
return(true);
}
else if (livesCounter == 3)
{
return(true);
}
else if (livesCounter < 2 || livesCounter > 3)
{
return(false);
}
else
{
return(false);
}
}
/// <summary>
/// Tworzy nową generację krzyżując obiekty z obecnej generacji.
/// </summary>
/// <returns>Zwraca nową generację.</returns>
public Generation CreateNewGeneration()
{
if (!CurrentGeneration.WasEvaluated)
{
CurrentGeneration.Evaluate();
}
IDna[] objects = new IDna[GenerationsSize];
for (uint i = 0; i < GenerationsSize; i++)
{
IDna parentA = CurrentGeneration.GetDnaBasedOnFitness();
IDna parentB = CurrentGeneration.GetDnaBasedOnFitness();
objects[i] = evolutionController.Crossover(parentA, parentB);
IDna db = objects[i];
Console.WriteLine(new string(db.GetGenes() as char[]) + " ||| " + db.Fitness);
}
CurrentGeneration = new Generation(evolutionController, objects);
CurrentGeneration.Mutate(MutationChance > 100 ? (byte)100 : MutationChance);
return(CurrentGeneration);
}
private void InitializePopulation()
{
for (var i = 0; i < PopulationSize; i++)
{
CurrentGeneration.Add(new Knapsack(_knapsackCapacity, _maxKnapsackWeight));
}
BestOfRun = CurrentGeneration[0];
foreach (var knapsack in CurrentGeneration)
{
var knapsackFitness = GetKnapsackFitness(knapsack);
var bestKnapsackFitness = GetKnapsackFitness(BestOfRun);
if (knapsackFitness > bestKnapsackFitness)
{
BestOfRun = knapsack;
BestOfRunFitness = knapsackFitness;
}
}
Generations.Add(CurrentGeneration);
}
protected override void OnEndSimulation()
{
var max = CurrentGeneration.First(i => i.Fitness.Equals(CurrentGeneration.Max(j => j.Fitness)));
_data.Genes = max.Chromosome.ToArray();
}
public static void DoUpdate(GameTime gameTime)
{
if ((ElapsedGenerationTime > GenerationTime && GenerationTime != 0) || !AnyoneAlive())
{
// GENETIC ALGORITHM THIS SHIT
thisEvolution.AddGeneration(new Generation(CurrentGeneration));
if (CurrentGeneration.Count > 0)
{
List <Evolvi> sorted = SortFittest();
int root = (int)Math.Sqrt(sorted.Count);
sorted = sorted.OrderByDescending(x => x.Energy).ToList();
List <Evolvi> newGen = new List <Evolvi>();
while (newGen.Count < evolviAmount)
{
Evolvi e = sorted.First().Crossbreed(sorted.First());
e.Mutate(mutationRate);
newGen.Add(e);
Thread.Sleep(10);
}
CurrentGeneration.Clear();
CurrentGeneration = newGen;
}
else
{
Random RNG = new Random();
Evolvi e = new Evolvi(new Vector2(RNG.Next(0, 2000), RNG.Next(0, 2000)));
for (int i = 0; i < evolviAmount; i++)
{
CurrentGeneration.Add(e.Clone());
Thread.Sleep(10);
}
}
ElapsedGenerationTime = 0;
Generation++;
OnNewGeneration(null, EventArgs.Empty);
}
ElapsedGenerationTime += (float)gameTime.ElapsedGameTime.TotalSeconds;
List <Evolvi> deadEvolvis = new List <Evolvi>();
foreach (Evolvi evolvi in CurrentGeneration)
{
if (!evolvi.Dead)
{
if (evolvi.Energy <= 0)
{
evolvi.Dead = true;
}
Input[] inpts = new Input[evolvi.AreasOfVision + 1];
int indexer = 0;
if (Math.Abs(evolvi.AngleDegrees) > 360)
{
evolvi.AngleDegrees = 0;
}
float anglePerAreaOfVision = (float)evolvi.FOV / (float)(evolvi.AreasOfVision + 1);
for (float i = -evolvi.FOV / 2f; i < (evolvi.FOV / 2f); i += anglePerAreaOfVision)
{
inpts[indexer] = new Input();
Vector2 start = evolvi.Position;
Vector2 end;
float xDiff = (float)Math.Cos(evolvi.AngleRadians + MathHelper.ToRadians(i)) * evolvi.VisionRange;
float yDiff = (float)Math.Sin(evolvi.AngleRadians + MathHelper.ToRadians(i)) * evolvi.VisionRange;
end = start + new Vector2(xDiff, yDiff);
inpts[indexer].value = 0;
Vector2 dir = new Vector2(xDiff, yDiff);
dir.Normalize();
Ray ray = new Ray(new Vector3(start, 0), new Vector3(dir, 0));
float angle = (float)Math.Atan2(dir.Y, dir.X);
List <DistanceEvaluator> ends = new List <DistanceEvaluator>();
float intersectionToWall = RayIntersectsAWall(ray);
foreach (Food food in CurrentFood)
{
if (evolvi.CollisionBox.Intersects(food.CollisionBox))
{
evolvi.Energy += food.Nutrition;
evolvi.FoodEaten++;
Random RNG = new Random();
food.Position = new Vector2(RNG.Next(40, 2000 - 80), RNG.Next(40, 2000 - 80));
//food.CollisionBox = new Rectangle(food.Position.ToPoint(), food.TextureFood.Bounds.Size);
//food.BoundingBox = new BoundingBox(new Vector3(food.Position, 0), new Vector3(food.Position + food.CollisionBox.Size.ToVector2(), 0));
bool done = false;
while (!done)
{
bool intersects = false;
foreach (Food secondFood in CurrentFood)
{
if (food != secondFood)
{
if (food.CollisionBox.Intersects(secondFood.CollisionBox))
{
intersects = true;
}
}
}
if (!intersects)
{
done = true;
}
else
{
food.Position = new Vector2(RNG.Next(20, 2000 - 20), RNG.Next(20, 2000 - 20));
//food.CollisionBox = new Rectangle(food.Position.ToPoint(), food.TextureFood.Bounds.Size);
//food.BoundingBox = new BoundingBox(new Vector3(food.Position, 0), new Vector3(food.Position + food.CollisionBox.Size.ToVector2(), 0));
}
}
}
if (ray.Intersects(food.BoundingBox) != 0)
{
if (ray.Intersects(food.BoundingBox) < intersectionToWall)
{
if (ray.Intersects(food.BoundingBox) < evolvi.VisionRange)
{
float?dist = ray.Intersects(food.BoundingBox);
ends.Add(new DistanceEvaluator((1f - (dist.Value / evolvi.VisionRange)) * (food.Nutrition / 50f), new Vector2((float)(Math.Cos(angle) * dist), (float)(Math.Sin(angle) * dist)) + start));
}
}
else
{
if (intersectionToWall < evolvi.VisionRange)
{
float?dist = intersectionToWall;
ends.Add(new DistanceEvaluator(-(1 - (dist.Value / evolvi.VisionRange)), new Vector2((float)(Math.Cos(angle) * dist), (float)(Math.Sin(angle) * dist)) + start));
}
}
}
}
if (ends.Count > 0)
{
List <DistanceEvaluator> sortedByDist = ends.OrderBy(x => x.Distance).ToList();
end = sortedByDist.Last().End;
inpts[indexer].value = sortedByDist.Last().Distance;
}
inpts[indexer].start = start;
inpts[indexer].end = end;
indexer++;
}
evolvi.Update(inpts);
}
}
}
private void AddRow(Borders border)
{
CurrentGeneration.AddRow(border);
NextGeneration.AddRow(border);
}
public static List <Evolvi> SortFittest()
{
List <Evolvi> sorted = CurrentGeneration.OrderBy(x => x.Energy).ToList();
return(sorted);
}
public List <DNA> GetPopulation()
{
return(CurrentGeneration.ToList());
}
public void NewGeneration()
{
if (CurrentGeneration.Count <= 0)
{
return;
}
newGeneration = new List <Specimen>();
oldGeneration = new List <Specimen>();
List <Specimen> bestOfTheBest = new List <Specimen>();
allGenerations.Add(CurrentGeneration);
for (int i = 0; i < CurrentGeneration.Count; i++)
{
for (int j = 0; j < CurrentGeneration.Count; j++)
{
if (i == j)
{
j++;
}
if (j == CurrentGeneration.Count)
{
break;
}
var firstParent = CurrentGeneration[i];
var secondParent = CurrentGeneration[j];
var child = firstParent.CrossoverWith(secondParent);
if (RandomForGA.Generator.NextDouble() < Parameters.MutationRate)
{
child.Mutate(Parameters.MutationAmplitude);
}
newGeneration.Add(child);
}
}
for (int i = 0; i < CurrentGeneration.Count; i++)
{
oldGeneration.Add(CurrentGeneration[i]);
}
CurrentGeneration.Clear();
for (int i = 0; i < oldGeneration.Count; i++)
{
CurrentGeneration.Add(oldGeneration[i]);
}
for (int i = 0; i < newGeneration.Count; i++)
{
CurrentGeneration.Add(newGeneration[i]);
}
CalculateFitness();
CurrentGeneration.Sort((a, b) => population.FittingFunction(a).CompareTo(population.FittingFunction(b)));
for (int i = 0; i < Parameters.PopulationSize; i++)
{
bestOfTheBest.Add(CurrentGeneration[i]);
}
CurrentGeneration = bestOfTheBest;
GenerationNum++;
}
public void RunEpoch()
{
List <Chromosome> newGeneration = new List <Chromosome>();
newGeneration.AddRange(CurrentGeneration.OrderBy(x => x.Fitness).Take((int)(CurrentGeneration.Count * ElitismRate)));
while (newGeneration.Count < CurrentGeneration.Count)
{
//Selection
Tuple <Chromosome, Chromosome> parents = Selection.Select(CurrentGeneration);
Tuple <Chromosome, Chromosome> children;
//Crossover
if (random.NextDouble() < CrossoverRate)
{
children = parents.Item1.Crossover(parents.Item2);
children.Item1.FitnessOutdated = true;
children.Item2.FitnessOutdated = true;
double parentAverage = (parents.Item1.Fitness + parents.Item2.Fitness) / 2;
children.Item1.Fitness = parentAverage;
children.Item2.Fitness = parentAverage;
children.Item1.FitnessOutdated = true;
children.Item2.FitnessOutdated = true;
if (random.NextDouble() < 0.5)
{
children.Item1.Priority = parents.Item1.Priority;
children.Item2.Priority = parents.Item2.Priority;
}
else
{
children.Item1.Priority = parents.Item2.Priority;
children.Item2.Priority = parents.Item1.Priority;
}
double variance = (Math.Pow(parents.Item1.Fitness - parentAverage, 2) + Math.Pow(parents.Item2.Fitness - parentAverage, 2)) / 2;
double stdDev = Math.Sqrt(variance);
double fitnessDifference = Math.Abs(parents.Item1.Fitness - parents.Item2.Fitness);
children.Item1.Priority += variance;
children.Item2.Priority += variance;
}
else
{
children = new Tuple <Chromosome, Chromosome>((Chromosome)parents.Item1.Clone(), (Chromosome)parents.Item2.Clone());
}
//Mutation
if (random.NextDouble() < MutationRate)
{
children.Item1.Mutate();
children.Item1.FitnessOutdated = true;
children.Item1.Priority += MutationWeigth;
}
if (random.NextDouble() < MutationRate)
{
children.Item2.Mutate();
children.Item2.FitnessOutdated = true;
children.Item2.Priority += MutationWeigth;
}
newGeneration.Add(children.Item1);
//Sometimes only one child can be added
if (newGeneration.Count == CurrentGeneration.Count)
{
newGeneration.Add(children.Item2);
}
}
CurrentGeneration = new Generation(newGeneration);
EvaluatePopulation();
}
private void AddColumn(Borders border)
{
CurrentGeneration.AddColumn(border);
NextGeneration.AddColumn(border);
}