private void performDialogue(Agent speaker, Agent listener, SocialNetwork socialNetwork)
{
Dialogue dialogue = new Dialogue();
string utterance = dialogue.utterWord(speaker);
bool isSuccess = false;
if (!(listener.getVocabulary().getVocabulary() == null) && listener.getVocabulary().getVocabulary().ContainsKey(utterance))
{
isSuccess = true;
speaker.updateDialogs(true);
double i = speaker.getGenome().getNormalisedGenome()[0] + speaker.getGenome().getNormalisedGenome()[8] + speaker.getGenome().getNormalisedGenome()[9] - speaker.getGenome().getNormalisedGenome()[4] - speaker.getGenome().getNormalisedGenome()[1] - speaker.getGenome().getNormalisedGenome()[7];
speaker.getVocabulary().updateVocabulary(utterance, Math.Max(0, i));
double i2 = listener.getGenome().getNormalisedGenome()[0] + listener.getGenome().getNormalisedGenome()[8] + listener.getGenome().getNormalisedGenome()[9] - listener.getGenome().getNormalisedGenome()[4] - listener.getGenome().getNormalisedGenome()[1] - listener.getGenome().getNormalisedGenome()[7];
listener.getVocabulary().updateVocabulary(utterance, Math.Max(0, i2));
succcessfullDialogues++;
}
else
{
double i = speaker.getGenome().getNormalisedGenome()[0] + speaker.getGenome().getNormalisedGenome()[8] + speaker.getGenome().getNormalisedGenome()[9] - speaker.getGenome().getNormalisedGenome()[4] - speaker.getGenome().getNormalisedGenome()[1] - speaker.getGenome().getNormalisedGenome()[7];
double i2 = listener.getGenome().getNormalisedGenome()[0] + listener.getGenome().getNormalisedGenome()[8] + listener.getGenome().getNormalisedGenome()[9] - listener.getGenome().getNormalisedGenome()[4] - listener.getGenome().getNormalisedGenome()[1] - listener.getGenome().getNormalisedGenome()[7];
speaker.updateDialogs(false);
speaker.getVocabulary().updateVocabulary(utterance, Math.Min(0, -1 * i));
listener.getVocabulary().updateVocabulary(utterance, Math.Max(0, i2));
}
speaker.updatepersonality(listener, isSuccess);
listener.updatepersonality(speaker, isSuccess);
socialNetwork.setConnection(speaker, listener, getWeight(speaker, isSuccess));
socialNetwork.setConnection(listener, speaker, getWeight(listener, isSuccess));
}
public void addDiscreteGraph(List <Agent> pop, SocialNetwork n, int generation)
{
string s = "";
string sep = ",";
List <string> addedEdges = new List <string>();
foreach (var a in pop)
{
s += a.getID().ToString() + sep;
if (n.getAgentsConnections(a) != null)
{
foreach (var b in n.getAgentsConnections(a))
{
if (b.Value >= 0.0) //&& !(addedEdges.Contains(a.getID().ToString() + b.Key.getID().ToString())) )
{
s += b.Key.getID().ToString() + sep;
addedEdges.Add(a.getID().ToString() + b.Key.getID().ToString());
}
}
s += "\n";
}
}
string filename = "C:/Users/andrl/Desktop/masterStuff/MasterData/Figures/Experiment 8/Graph" + generation.ToString() + ".txt";
System.IO.File.WriteAllText(@filename, s);
}
public List <Agent> survivalSelection(List <Agent> population, SocialNetwork socialNetwork)
{
List <Agent> deadAgents = new List <Agent>();
List <Agent> survivingAgents = new List <Agent>();
while (survivingAgents.Count < populationSize * k)
{
int i = RandomInt(0, population.Count);
survivingAgents.Add(population[i]);
population.RemoveAt(i);
}
List <Agent> sortedList = survivingAgents.OrderBy(agent => agent.getFitness()).ToList();
//Console.WriteLine("\n\npopSize: " + populationSize + "population: " + sortedList.Count);
sortedList.Reverse();
//Console.WriteLine("fitnesses: " + sortedList[0].getFitness() + " " + sortedList[sortedList.Count - 1].getFitness());
removeDeadAgents(socialNetwork, population, populationSize);
List <Agent> survivors = new List <Agent>();
int counter = 0;
while (counter < sortedList.Count * n)
{
survivors.Add(sortedList[counter]);
counter++;
}
//Console.WriteLine("Number of survivors: "+ survivors.Count);
return(survivors);
}
public List <Agent> breed(List <Agent> pop, SocialNetwork socialNetwork)
{
List <Agent> children = new List <Agent>();
for (int i = 0; i < 1; i++)
{
Agent parent1 = tournamentSelection(pop);
Agent parent2 = tournamentSelection(pop);
Agent child = crossover(parent1, parent2);
if (RandomDouble() <= (child.getGenome().getNormalisedGenome()[5] * speakToParentsConstant))
{
breedMut.WaitOne(); // MUTEX start
performDialogue(parent1, child, socialNetwork);
performDialogue(parent2, child, socialNetwork);
performDialogue(parent1, child, socialNetwork);
performDialogue(parent2, child, socialNetwork);
performDialogue(parent1, child, socialNetwork);
performDialogue(parent2, child, socialNetwork);
performDialogue(parent1, child, socialNetwork);
performDialogue(parent2, child, socialNetwork);
breedMut.ReleaseMutex(); // MUTEX end
}
children.Add(child);
}
return(children);
}
public void performDialogues(SocialNetwork socialNetwork, List <Agent> speakerPool, List <Agent> population)
{
diaMut.WaitOne();
Dialogue dialogue = new Dialogue();
Agent speaker; Agent listener;
if (speakerPool.Count == 0)
{
//Console.WriteLine("forced extrovert");
speaker = population[RandomInt(0, population.Count)];
listener = population[RandomInt(0, population.Count)];
}
else
{
//Console.WriteLine("Introvert");
speaker = dialogue.selectSpeaker(speakerPool);
listener = dialogue.selectListener(speaker, socialNetwork, population);
}
if (listener == null)
{
while (listener == null || listener == speaker)
{
Console.WriteLine("new listener");
listener = population[RandomInt(0, population.Count)];
}
}
performDialogue(speaker, listener, socialNetwork);
diaMut.ReleaseMutex();
}
private void addFitnessDegreeData(List <Agent> population, SocialNetwork socialNetwork, DataCollector d)
{
double C = 0.5;
double allConnections = 0;
double AvgWeight = 0;
double learnRateSum = 0;
double totalVocLen = 0;
double speakToParentsGene = 0;
double extrovertProb = 0;
foreach (Agent a in population)
{
double agentsAvgWeight = 0;
double agentsConnections = 0;
speakToParentsGene += a.getGenome().getNormalisedGenome()[5] * speakToParentsConstant;
List <double> genome = a.getGenome().getValuesGenome();
extrovertProb += ((genome[3] + (100 - genome[6]) / 200) * C) / 100;
totalVocLen += a.getVocabulary().getVocabulary().Count;
List <double> normGenome = a.getGenome().getNormalisedGenome();
learnRateSum += (normGenome[0] + normGenome[8] + normGenome[9] - normGenome[4] - normGenome[1] - normGenome[7]);
if (socialNetwork.getAgentsConnections(a) != null && socialNetwork.getAgentsConnections(a).Count > 0)
{
foreach (var i in socialNetwork.getAgentsConnections(a))
{
if (i.Value > 0.0)
{
agentsConnections++;
agentsAvgWeight += i.Value;
}
}
if (agentsConnections > 0)
{
AvgWeight += (agentsAvgWeight / agentsConnections);
allConnections += agentsConnections;
}
}
}
Console.WriteLine("Speak to parents prob: " + (speakToParentsGene / population.Count));
Console.WriteLine("Extrovert probability: " + (extrovertProb / population.Count));
Console.WriteLine("average vocabulary length: " + (totalVocLen / population.Count));
Console.WriteLine("Fittest agent: " + population[0].getFitness());
Console.WriteLine("Average connections: " + (allConnections / population.Count));
Console.WriteLine("Agents average weight: " + (AvgWeight / population.Count));
Console.WriteLine("Average learn rate: " + (learnRateSum / population.Count) + "\n\n");
d.addExtrovertData((extrovertProb / population.Count));
d.addSpeakToParentsGenome((speakToParentsGene / population.Count));
d.addAvgVocLen((totalVocLen / population.Count));
d.addFittest(population[0].getFitness());
d.setDegree(allConnections / population.Count);
d.addLearnRate((learnRateSum / population.Count));
}
public void dialogueThreads(SocialNetwork socialNetwork, List <Agent> speakerPool, List <Agent> population)
{
List <Thread> ts = new List <Thread>();
for (int i = 0; i < population.Count * d; i++)
{
Thread t = new Thread(new ThreadStart(() => performDialogues(socialNetwork, speakerPool, population)));
t.Name = String.Format("t{0}", i + 1);
t.Start();
ts.Add(t);
}
foreach (var t in ts)
{
t.Join();
}
}
private void fitnessOfPopulation(List <Agent> population, SocialNetwork socialNetwork)
{
foreach (Agent a in population)
{
if (a != null && socialNetwork.getAgentsConnections(a) != null)
{
double fitness = a.calculateFitness(socialNetwork.getAgentsConnections(a));
a.setFitness(fitness);
}
else
{
a.setFitness(0);
}
//Console.WriteLine("Vocabulary size: "+ a.getVocabulary().getVocabulary().Count + "\nFitness: "+ fitness+"\n");
}
population = population.OrderBy(agent => agent.getFitness()).ToList();
population.Reverse();
}
private void removeDeadAgents(SocialNetwork socialNetwork, List <Agent> deadAgents, int populationSize)
{
foreach (var a in socialNetwork.socialNetwork.ToList())
{
if (deadAgents.Contains(a.Key))
{
socialNetwork.socialNetwork.Remove(a.Key);
}
else
{
foreach (var b in socialNetwork.socialNetwork[a.Key].ToList())
{
if (deadAgents.Contains(b.Key))
{
socialNetwork.socialNetwork[a.Key].Remove(b.Key);
}
}
}
}
}
public List <Agent> makeChildren(List <Agent> population, SocialNetwork socialNetwork)
{
EALoop ea = new EALoop();
List <Agent> Children = new List <Agent>();
List <Thread> ts = new List <Thread>();
//Console.WriteLine("making " + (populationSize - n )+ " children.");
for (int i = population.Count; i < populationSize; i++)
{
Thread t = new Thread(new ThreadStart(() => Children.AddRange(ea.breed(population, socialNetwork))));
t.Name = String.Format("t{0}", i + 1);
t.Start();
ts.Add(t);
}
foreach (var t in ts)
{
t.Join();
}
//Console.WriteLine("Number of children: " + Children.Count);
return(Children);
}
public Agent selectListener(Agent agent, SocialNetwork net, List <Agent> population)
{
var genome = agent.getGenome().getValuesGenome();
double P_extrovert = ((genome[3] + (100 - genome[6]) / 200) * C) / 100;
Dictionary <Agent, double> connections = net.getAgentsConnections(agent);
//if ((EALoop.RandomDouble() <= P_extrovert && agent.getZ() < maxExtroConv) || net.getAgentsConnections(agent) == null)
if (EALoop.RandomDouble() <= P_extrovert)
{
// Extrovert
//System.Console.WriteLine("EXTROVERT");
Agent listener = population[EALoop.RandomInt(0, population.Count)];
while (listener == agent)
{
listener = population[EALoop.RandomInt(0, population.Count)];
}
agent.incrementZ();
return(listener);
}
// Introvert
double sum = 0;
foreach (var friend in connections)
{
sum += friend.Value;
}
double random = EALoop.RandomDouble();
double to = 0;
foreach (var friend in connections)
{
to += friend.Value / sum;
if (random <= to)
{
return(friend.Key);
}
}
return(null);
}
static void Main(string[] args)
{
EALoop ea = new EALoop();
Console.WriteLine("STARTING");
SocialNetwork socialNetwork = new SocialNetwork();
DataCollector data = new DataCollector();
List <Agent> population = new List <Agent>();
Dialogue dialogue = new Dialogue();
int generations = 1;
for (int i = 0; i < populationSize; i++)
{
population.Add(new Agent(ea.AgentIdCounter));
ea.AgentIdCounter++;
}
List <Agent> speakerPool = new List <Agent>();
foreach (Agent a in population)
{
if (socialNetwork.getAgentsConnections(a) != null)
{
speakerPool.Add(a);
}
}
Console.WriteLine("Speaker pool: " + speakerPool.Count);
ea.dialogueThreads(socialNetwork, speakerPool, population);
ea.fitnessOfPopulation(population, socialNetwork);
population = ea.survivalSelection(population, socialNetwork);
data.addDiscreteGraph(population, socialNetwork, generations);
data.addFitnessData(population);
data.setDialogues(population);
ea.addFitnessDegreeData(population, socialNetwork, data);
data.setUniqueWords(population);
ea.updateAges(population);
while (generations < Totalgenerations)
{
generations++;
Console.WriteLine("\nGeneration number: " + generations);
//Console.WriteLine("Nodes in the socialnetwork: " + socialNetwork.socialNetwork.Count);
//-- MAKE CHILDREN --//
//Console.WriteLine("Size of population before children is made: " + population.Count);
population.AddRange(ea.makeChildren(population, socialNetwork));
//Console.WriteLine("Size of population after children was made: " + population.Count);
//-- PERFORM DIALOGUES --//
//Console.WriteLine("socialnetwork count "+socialNetwork.socialNetwork.Count);
speakerPool = new List <Agent>();
for (int q = 0; q < population.Count; q++)
{
//Console.WriteLine("index: " + q);
if (population[q] == null)
{
population.RemoveAt(q);
}
else if (socialNetwork.getAgentsConnections(population[q]) != null && socialNetwork.getAgentsConnections(population[q]).Count > 0)
{
speakerPool.Add(population[q]);
}
}
Console.WriteLine("Speaker pool: " + speakerPool.Count);
ea.dialogueThreads(socialNetwork, speakerPool, population);
//ea.succcessfullDialogues = 0;
//Console.WriteLine("Dialogues performed");
//-- CALCULATE FITNESS --//
ea.fitnessOfPopulation(population, socialNetwork);
//-- SURVIVAL SELECTION --//
population = ea.survivalSelection(population, socialNetwork);
//-- DATA GATHERING --//
data.setDialogues(population);
data.setUniqueWords(population);
data.addFitnessData(population);
if (generations % 5 == 0)
{
data.addDiscreteGraph(population, socialNetwork, generations);
}
ea.addFitnessDegreeData(population, socialNetwork, data);
ea.updateAges(population);
//Console.WriteLine("Size of population after survival selection: " + population.Count);
}
Console.WriteLine("fitness data: " + data.getFitnessdata().Count + "\ndialogue data: " + data.getDialogues().Count + "\nUnique words data: " + data.getUniqueWords().Count);
data.writeToFiles();
data.addDiscreteGraph(population, socialNetwork, generations);
Console.Write("Press any button to end");
}
