private Evolution initEvolution(ProgressReport pReport)
{
var e = new Evolution();
e.Iteration = pReport.Iteration;
e.MaxFitness = pReport.IterationStatistics.MaximumFitness;
e.AvgFitness = pReport.IterationStatistics.AverageFitness;
e.Status = pReport.IterationStatus;
//
return(e);
}
//default constructor
public Factory()
{
m_progresReport = new ProgressReport()
{
BestIteration = -1,
BestSolution = null, Iteration = -1, IterationStatus = IterationStatus.Initialize,
IterationStatistics = new IterationStatistics()
{
AverageFitness = -1, MaximumFitness = -1
}
};
}
/// <summary>
/// Constructor
/// </summary>
/// <param name="param"></param>
/// <param name="funSet"></param>
/// <param name="terSet"></param>
public Factory(Parameters param, Function[] funSet, int[] terSet, CancellationToken token)
{
//create population
m_Population = new Population();
m_Population.Token = token;
m_History = new List <Evolution>();
//prepare progress report for the first time.
m_progresReport = new ProgressReport();
m_progresReport.Iteration = 0;
m_progresReport.IterationStatus = IterationStatus.Initialize;
m_progresReport.IterationStatistics = new IterationStatistics();
m_progresReport.Message = "GPFactory successfully created algorithm.";
m_progresReport.BestSolution = new Chromosome();
m_progresReport.IterationStatistics.IterationSeconds = 0;
//initialize main GP components parameters, function and terminals
initMainComponents(param, funSet, terSet);
}
/// <summary>
/// Calculate best chromosome in the current population (generation)
/// </summary>
/// <param name="chromosomes"></param>
/// <param name="pr"></param>
private void calculatePopulation(List <IChromosome> chromosomes, ProgressReport pr)
{
// calculate basic stat of the population
float fitnessMax = float.MinValue;
float fitnessSum = 0;
if (chromosomes != null && chromosomes.Count > 0)
{
ProgresReport.BestSolution = setBestSolution(chromosomes[0]);
}
else
{
return;
}
int counter = 0;
for (int i = 0; i < chromosomes.Count; i++)
{
if (float.IsNaN(chromosomes[i].Fitness))
{
continue;
}
counter++;
float fitness = chromosomes[i].Fitness;
// sum calculation
fitnessSum += fitness;
// cal the best chromosome
if (fitness > fitnessMax)
{
fitnessMax = fitness;
pr.BestSolution = setBestSolution(chromosomes[i]);
}
}
float fitnessAvg = (float)Math.Round(fitnessSum / counter, 4);
pr.IterationStatistics.AverageFitness = fitnessAvg;
pr.IterationStatistics.MaximumFitness = fitnessMax;
}
/// <summary>
/// perform one iteration evolution for the GP
/// </summary>
/// <param name="pr"></param>
private void performIteration(ProgressReport pr, CancellationToken ct)
{
//mating preparation offspring population
m_Population.PrepareForMating();
Stopwatch sw = Stopwatch.StartNew();
string str = $"It={pr.Iteration}; ";
//1. operaton
m_Population.Crossover(m_Parameters, m_FunctionSet, m_TerminalSet);
str += $"Crossover ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
sw.Restart();
//2. operation
m_Population.Mutate(m_Parameters, m_FunctionSet, m_TerminalSet);
str += $"Mutation ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
sw.Restart();
//3. operation
m_Population.Reproduction(m_Parameters, m_FunctionSet, m_TerminalSet);
str += $"Reproduction ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
sw.Restart();
//4. operation
m_Population.EvaluatePopulation(m_Parameters, pr.Iteration);
str += $"FitnessCalc ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
sw.Restart();
//
m_Population.CreateNewGeneration(m_Parameters, m_FunctionSet, m_TerminalSet);
str += $"NewGen ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
sw.Restart();
//
calculatePopulation(m_Population.m_Chromosomes, pr);
str += $"Calculation ={sw.Elapsed.TotalSeconds.ToString("F3")} sec| ";
pr.Message = str;
}
public void GPFactoryFromString(string strFactory)
{
if (string.IsNullOrEmpty(strFactory))
{
return;
}
try
{
Evolution[] hist = null;
JsonSerializerSettings sett = new JsonSerializerSettings();
sett.NullValueHandling = NullValueHandling.Ignore;
var obj = JsonConvert.DeserializeObject(strFactory, sett);
var param = ((dynamic)obj)["m_Parameters"] as Newtonsoft.Json.Linq.JObject;
var ter = ((dynamic)obj)["m_TerminalSet"] as Newtonsoft.Json.Linq.JArray;
var fun = ((dynamic)obj)["m_FunctionSet"] as Newtonsoft.Json.Linq.JArray;
var prog = ((dynamic)obj)["m_progresReport"] as Newtonsoft.Json.Linq.JObject;
var tc = ((dynamic)obj)["m_TerminationCriteria"] as Newtonsoft.Json.Linq.JObject;
var hi = ((dynamic)obj)["history"] as Newtonsoft.Json.Linq.JArray;
var pop = ((dynamic)obj)["popStr"] as Newtonsoft.Json.Linq.JValue;
if (param != null)
{
m_Parameters = param.ToObject <Parameters>();
}
if (ter != null)
{
m_TerminalSet = ter.ToObject <int[]>();
}
if (fun != null)
{
m_FunctionSet = fun.ToObject <Function[]>();
}
if (prog != null)
{
m_progresReport = prog.ToObject <ProgressReport>();
}
if (tc != null)
{
m_TerminationCriteria = tc.ToObject <TerminationCriteria>();
}
if (hi != null)
{
hist = hi.ToObject <Evolution[]>();
}
//create history
if (hist != null)
{
m_History = new List <Evolution>(hist);
}
m_Population = new Population();
if (pop != null)
{
var popStr = pop.ToObject <string>();
m_Population.PopulationFromString(popStr);
}
}
catch (Exception)
{
throw;
}
}