/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weight difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> ea = new NeatEvolutionAlgorithm <NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
BoxesVisualDiscriminationEvaluator evaluator = new BoxesVisualDiscriminationEvaluator(_visualFieldResolution);
// Create genome decoder. Decodes to a neural network packaged with an activation scheme that defines a fixed number of activations per evaluation.
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder(_visualFieldResolution, _lengthCppnInput);
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator <NeatGenome> innerEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
// Wrap the list evaluator in a 'selective' evaluator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determined by examining each genome's evaluation info object.
IGenomeListEvaluator <NeatGenome> selectiveEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
// Initialize the evolution algorithm.
ea.Initialize(selectiveEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(ea);
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome2 population and their associated/parent genome2 factory.
/// </summary>
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> ea = new NeatEvolutionAlgorithm <NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create genome2 decoder.
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = new NeatGenomeDecoder(_activationScheme);
// Create a genome2 list evaluator. This packages up the genome2 decoder with the genome2 evaluator.
IGenomeListEvaluator <NeatGenome> genomeListEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, PhenomeEvaluator, _parallelOptions);
// Wrap the list evaluator in a 'selective' evaulator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determiend by examining each genome2's evaluation info object.
if (!EvaluateParents)
{
genomeListEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(genomeListEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
}
// Initialize the evolution algorithm.
ea.Initialize(genomeListEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(ea);
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weight difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> ea = new NeatEvolutionAlgorithm <NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
PreyCaptureEvaluator evaluator = new PreyCaptureEvaluator(_trialsPerEvaluation, _gridSize, _preyInitMoves, _preySpeed, _sensorRange, _maxTimesteps);
// Create genome decoder.
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder();
// TODO: evaluation scheme that re-evaluates existing genomes and takes average over time.
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator <NeatGenome> genomeListEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
// Initialize the evolution algorithm.
ea.Initialize(genomeListEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(ea);
}
private void GetGenomeListEvaluator()
{
var innerEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(_GenomeDecoder, _PhenomeEvaluator, _ParallelOptions);
_GenomeListEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
}
NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(bool load)
{
// Create a genome2 factory with our neat genome2 parameters object and the appropriate number of input and output neuron genes.
var genomeFactory = new NeatGenomeFactory(TetrisEvaluator.NumInputs, TetrisEvaluator.NumOutputs);
// Create an initial population of randomly generated genomes.
List <NeatGenome> genomeList = null;
if (load)
{
try
{
using (var reader = XmlReader.Create("SavedProgress.xml"))
genomeList = NeatGenomeXmlIO.ReadCompleteGenomeList(reader, true, genomeFactory);
Console.WriteLine("Loaded network!");
}
catch
{
load = false;
}
}
if (!load)
{
genomeList = genomeFactory.CreateGenomeList(150, 0);
}
var parallelOpts = new ParallelOptions()
{
MaxDegreeOfParallelism = -1
};
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
var distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
var speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, parallelOpts);
// Create the evolution algorithm.
var ea = new NeatEvolutionAlgorithm <NeatGenome>(new NeatEvolutionAlgorithmParameters {
SpecieCount = 10
}, speciationStrategy, new DefaultComplexityRegulationStrategy(ComplexityCeilingType.Absolute, 50));
// Create genome2 decoder.
var genomeDecoder = new NeatGenomeDecoder(NetworkActivationScheme.CreateCyclicFixedTimestepsScheme(2));
// Create a genome2 list evaluator. This packages up the genome2 decoder with the genome2 evaluator.
IGenomeListEvaluator <NeatGenome> genomeListEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, tetrisEvaluator, parallelOpts);
// Wrap the list evaluator in a 'selective' evaulator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determiend by examining each genome2's evaluation info object.
//if (!EvaluateParents)
//genomeListEvaluator = new SelectiveGenomeListEvaluator<NeatGenome>(genomeListEvaluator, SelectiveGenomeListEvaluator<NeatGenome>.CreatePredicate_OnceOnly());
ea.UpdateEvent += Ea_UpdateEvent;
// Initialize the evolution algorithm.
ea.Initialize(genomeListEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(ea);
}
private NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm_private(IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList, HyperNEAT_Args args = null)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1, 0, 10);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> retVal = new NeatEvolutionAlgorithm <NeatGenome>(NeatEvolutionAlgorithmParameters, speciationStrategy, complexityRegulationStrategy);
// Genome Decoder
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = null;
if (args == null)
{
genomeDecoder = CreateGenomeDecoder();
}
else
{
genomeDecoder = CreateGenomeDecoder(args);
}
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator <NeatGenome> genomeEvaluator = null;
if (_phenomeEvaluator != null)
{
IGenomeListEvaluator <NeatGenome> innerEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, _phenomeEvaluator, _parallelOptions);
// Wrap the list evaluator in a 'selective' evaluator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determined by examining each genome's evaluation info object.
genomeEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
}
else if (_phenomeEvaluators != null)
{
// Use the multi tick evaluator
genomeEvaluator = new TickGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, _phenomeEvaluators, _phenometickeval_roundRobinManager, _phenometickeval_worldTick);
}
else
{
throw new ApplicationException("One of the phenome evaluators needs to be populated");
}
// Initialize the evolution algorithm.
retVal.Initialize(genomeEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(retVal);
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> ea = new NeatEvolutionAlgorithm <NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
DoublePoleBalancingEvaluator evaluator;
switch (_variantStr)
{
case "DoublePole":
evaluator = new DoublePoleBalancingEvaluator();
break;
case "DoublePoleNv":
evaluator = new DoublePoleBalancingEvaluatorNv();
break;
case "DoublePoleNvAntiWiggle":
evaluator = new DoublePoleBalancingEvaluatorNvAntiWiggle();
break;
default:
throw new SharpNeatException(string.Format("DoublePoleBalancing experiment config XML specifies unknown variant [{0}]", _variantStr));
}
// Create genome decoder.
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator <NeatGenome> innerEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
// Wrap the list evaluator in a 'selective' evaulator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determiend by examining each genome's evaluation info object.
IGenomeListEvaluator <NeatGenome> selectiveEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
// Initialize the evolution algorithm.
ea.Initialize(selectiveEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(ea);
}
/*
* List<NeatGenome> CreateNewGenome(IGenomeFactory<NeatGenome> genomeFactory)
* {
* Console.WriteLine("Saved genome not found, creating new files.");
* return genomeFactory.CreateGenomeList(_populationSize, 0);
* }
*/
/// <summary>
/// Creates and returns a NeatEvolutionAlgorithm object ready for running
/// the NEAT algorithm/search. Various sub-parts of the algorithm are also
/// constructed and connected up. This overload accepts a pre-built genome2
/// population and their associated/parent genome2 factory.
/// </summary>
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(
IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Creates distance metric. Mismatched genes have a fixed distance of 10;
// for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
//ISpeciationStrategy<NeatGenome> speciationStrategy = new KMeansClusteringStrategy<NeatGenome>(distanceMetric);
ISpeciationStrategy <NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, _parallelOptions);
IComplexityRegulationStrategy complexityRegulationStrategy =
ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Creates the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> evolAlgorithm = new NeatEvolutionAlgorithm <NeatGenome>(
_eaParams, speciationStrategy, complexityRegulationStrategy, userName);
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = new NeatGenomeDecoder(_activationScheme);
// Creates a genome2 list evaluator. This packages up the genome2 decoder with the genome2 evaluator.
IGenomeListEvaluator <NeatGenome> genomeListEvaluator =
new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, PhenomeEvaluator, _parallelOptions);
//To use single-thread evaluator:
//IGenomeListEvaluator<NeatGenome> genomeListEvaluator =
// new SerialGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, PhenomeEvaluator, false);
// Wraps the list evaluator in a 'selective' evaulator that will only
// evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes).
// This is determiend by examining each genome's evaluation info object.
/*
* int reevaluationPeriod = 1;
* genomeListEvaluator = new SelectiveGenomeListEvaluator<NeatGenome>(
* genomeListEvaluator,
* SelectiveGenomeListEvaluator<NeatGenome>.CreatePredicate_PeriodicReevaluation(reevaluationPeriod));
*/
genomeListEvaluator = new SelectiveGenomeListEvaluator <NeatGenome>(
genomeListEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
// Initializes the evolution algorithm.
evolAlgorithm.Initialize(genomeListEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return(evolAlgorithm);
}
public NeatEvolutionAlgorithm <NeatGenome> CreateEvolutionAlgorithm(
IGenomeFactory <NeatGenome> genomeFactory, List <NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10;
// for matched genes the distance is their weight difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy <NeatGenome> speciationStrategy =
new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = new NullComplexityRegulationStrategy();
// Create the evolution algorithm.
NeatEvolutionAlgorithm <NeatGenome> ea =
new NeatEvolutionAlgorithm <NeatGenome>(this.eaParams, speciationStrategy, complexityRegulationStrategy);
// Create genome decoder.
IGenomeDecoder <NeatGenome, IBlackBox> genomeDecoder = this.CreateGenomeDecoder();
var evaluator = new Evaluator();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
//IGenomeListEvaluator<NeatGenome> innerEvaluator =
// new SerialGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, evaluator);
IGenomeListEvaluator <NeatGenome> innerEvaluator =
new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, evaluator);
// Wrap the list evaluator in a 'selective' evaluator that will only evaluate new genomes.
// That is, we skip re-evaluating any genomes that were in the population in previous
// generations (elite genomes). This is determined by examining each genome's evaluation info object.
IGenomeListEvaluator <NeatGenome> selectiveEvaluator =
new SelectiveGenomeListEvaluator <NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator <NeatGenome> .CreatePredicate_OnceOnly());
// Initialize the evolution algorithm.
ea.Initialize(selectiveEvaluator, genomeFactory, genomeList);
ea.UpdateScheme = new UpdateScheme(1);
// Finished. Return the evolution algorithm
return(ea);
}
private static void Train()
{
File.WriteAllText($"{NeatConsts.experimentName}/fitness.csv", "generation,firness\n");
var neatGenomeFactory = new NeatGenomeFactory(NeatConsts.ViewX * NeatConsts.ViewY * NeatConsts.typeIds.Count, 1);
var genomeList = neatGenomeFactory.CreateGenomeList(NeatConsts.SpecCount, 0);
var eaParams = new NeatEvolutionAlgorithmParameters
{
SpecieCount = NeatConsts.SpecCount
};
//var distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
var distanceMetric = new ManhattanDistanceMetric();
var parallelOptions = new ParallelOptions();
var speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, parallelOptions);
//var speciationStrategy = new KMeansClusteringStrategy<NeatGenome>(distanceMetric);
//var speciationStrategy = new RandomClusteringStrategy<NeatGenome>();
var complexityRegulationStrategy = new NullComplexityRegulationStrategy();
//var complexityRegulationStrategy = new DefaultComplexityRegulationStrategy(ComplexityCeilingType.Relative, 0.50);
var ea = new NeatEvolutionAlgorithm <NeatGenome>(eaParams, speciationStrategy, complexityRegulationStrategy);
var activationScheme = NetworkActivationScheme.CreateCyclicFixedTimestepsScheme(1);
var genomeDecoder = new NeatGenomeDecoder(activationScheme);
var phenomeEvaluator = new GameEvaluator();
var genomeListEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(genomeDecoder, phenomeEvaluator, parallelOptions);
ea.Initialize(genomeListEvaluator, neatGenomeFactory, genomeList);
ea.UpdateScheme = new UpdateScheme(NeatConsts.LogRate);
ea.StartContinue();
ea.UpdateEvent += Ea_UpdateEvent;
while (ea.RunState != RunState.Paused)
{
}
ea.Stop();
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm<NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory<NeatGenome> genomeFactory, List<NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy<NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy<NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm<NeatGenome> ea = new NeatEvolutionAlgorithm<NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
InvertedDoublePendulumEvaluator evaluator = new InvertedDoublePendulumEvaluator();
// Create genome decoder.
IGenomeDecoder<NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator<NeatGenome> innerEvaluator = new ParallelGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
//IGenomeListEvaluator<NeatGenome> innerEvaluator = new SerialGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, evaluator);
// Wrap the list evaluator in a 'selective' evaulator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determiend by examining each genome's evaluation info object.
IGenomeListEvaluator<NeatGenome> selectiveEvaluator = new SelectiveGenomeListEvaluator<NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator<NeatGenome>.CreatePredicate_OnceOnly());
// Initialize the evolution algorithm.
ea.Initialize(selectiveEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return ea;
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm<NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory<NeatGenome> genomeFactory, List<NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy<NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy<NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm<NeatGenome> ea = new NeatEvolutionAlgorithm<NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
DoublePoleBalancingEvaluator evaluator;
switch(_variantStr)
{
case "DoublePole":
evaluator = new DoublePoleBalancingEvaluator();
break;
case "DoublePoleNv":
evaluator = new DoublePoleBalancingEvaluatorNv();
break;
case "DoublePoleNvAntiWiggle":
evaluator = new DoublePoleBalancingEvaluatorNvAntiWiggle();
break;
default:
throw new SharpNeatException(string.Format("DoublePoleBalancing experiment config XML specifies unknown variant [{0}]", _variantStr));
}
// Create genome decoder.
IGenomeDecoder<NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator<NeatGenome> innerEvaluator = new ParallelGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
// Wrap the list evaluator in a 'selective' evaulator that will only evaluate new genomes. That is, we skip re-evaluating any genomes
// that were in the population in previous generations (elite genomes). This is determiend by examining each genome's evaluation info object.
IGenomeListEvaluator<NeatGenome> selectiveEvaluator = new SelectiveGenomeListEvaluator<NeatGenome>(
innerEvaluator,
SelectiveGenomeListEvaluator<NeatGenome>.CreatePredicate_OnceOnly());
// Initialize the evolution algorithm.
ea.Initialize(selectiveEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return ea;
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search. Various sub-parts
/// of the algorithm are also constructed and connected up.
/// This overload accepts a pre-built genome population and their associated/parent genome factory.
/// </summary>
public NeatEvolutionAlgorithm<NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory<NeatGenome> genomeFactory, List<NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy<NeatGenome> speciationStrategy = new ParallelKMeansClusteringStrategy<NeatGenome>(distanceMetric, _parallelOptions);
// Create complexity regulation strategy.
IComplexityRegulationStrategy complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStr, _complexityThreshold);
// Create the evolution algorithm.
NeatEvolutionAlgorithm<NeatGenome> ea = new NeatEvolutionAlgorithm<NeatGenome>(_eaParams, speciationStrategy, complexityRegulationStrategy);
// Create IBlackBox evaluator.
PreyCaptureEvaluator evaluator = new PreyCaptureEvaluator(_trialsPerEvaluation, _gridSize, _preyInitMoves, _preySpeed, _sensorRange, _maxTimesteps);
// Create genome decoder.
IGenomeDecoder<NeatGenome, IBlackBox> genomeDecoder = CreateGenomeDecoder();
// TODO: evaulation scheme that re-evaulates existing genomes and takes average over time.
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator<NeatGenome> genomeListEvaluator = new ParallelGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, evaluator, _parallelOptions);
// Initialize the evolution algorithm.
ea.Initialize(genomeListEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return ea;
}
public NeatEvolutionAlgorithm<NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory<NeatGenome> genomeFactory, List<NeatGenome> genomeList)
{
var parallelOptions = new ParallelOptions();
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
var distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
var speciationStrategy = new ParallelKMeansClusteringStrategy<NeatGenome>(distanceMetric, parallelOptions);
// Create complexity regulation strategy.
var complexityRegulationStrategy = ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStrategy, _complexityThreshold);
// Create the evolution algorithm.
var evolutionAlgorithm = new NeatEvolutionAlgorithm<NeatGenome>(NeatEvolutionAlgorithmParameters, speciationStrategy, complexityRegulationStrategy);
// Create genome decoder.
var genomeDecoder = CreateGenomeDecoder();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
var genomeListEvaluator = new ParallelGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, _neuromonPhenomeEvaluator, parallelOptions);
// Initialize the evolution algorithm.
evolutionAlgorithm.Initialize(genomeListEvaluator, genomeFactory, genomeList);
return evolutionAlgorithm;
}
public static void Main(string[] args)
{
var random = new Random();
var circuits = circuitsFilePaths().ToArray();
var perceptionStep = TimeSpan.FromSeconds(0.1);
var simulationStep = TimeSpan.FromSeconds(0.05); // 20Hz
var maximumSimulationTime = TimeSpan.FromSeconds(60);
var tracks = circuits.Select(circuitPath => Track.Load($"{circuitPath}/circuit_definition.json"));
var worlds = tracks.Select(track => new StandardWorld(track, simulationStep)).ToArray();
var inputSamplesCount = 3;
var maximumScanningDistance = 200;
ILidar createLidarFor(ITrack track)
=> new Lidar(track, inputSamplesCount, Angle.FromDegrees(135), maximumScanningDistance);
var settings = new EvolutionSettings
{
PopulationSize = 1000,
SpeciesCount = 30,
ElitismProportion = 0,
ComplexityThreshold = 50
};
// prepare simulation
var parameters = new NeatEvolutionAlgorithmParameters
{
ElitismProportion = settings.ElitismProportion,
SpecieCount = settings.SpeciesCount
};
var distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
var parallelOptions = new ParallelOptions {
MaxDegreeOfParallelism = 4
};
var speciationStrategy = new ParallelKMeansClusteringStrategy <NeatGenome>(distanceMetric, parallelOptions);
var complexityRegulationStrategy = new DefaultComplexityRegulationStrategy(ComplexityCeilingType.Absolute, settings.ComplexityThreshold);
var evolutionaryAlgorithm = new NeatEvolutionAlgorithm <NeatGenome>(
parameters,
speciationStrategy,
complexityRegulationStrategy);
var phenomeEvaluator = new RaceSimulationEvaluator(
random,
simulationStep,
perceptionStep,
maximumSimulationTime,
worlds,
createLidarFor);
var genomeDecoder = new NeatGenomeDecoder(NetworkActivationScheme.CreateAcyclicScheme());
var genomeListEvaluator = new ParallelGenomeListEvaluator <NeatGenome, IBlackBox>(
genomeDecoder,
phenomeEvaluator);
evolutionaryAlgorithm.Initialize(
genomeListEvaluator,
genomeFactory: new NeatGenomeFactory(
inputNeuronCount: inputSamplesCount,
outputNeuronCount: 2,
DefaultActivationFunctionLibrary.CreateLibraryNeat(new BipolarSigmoid()),
new NeatGenomeParameters
{
FeedforwardOnly = true,
AddNodeMutationProbability = 0.03,
DeleteConnectionMutationProbability = 0.05,
ConnectionWeightMutationProbability = 0.08,
FitnessHistoryLength = 10,
}),
settings.PopulationSize);
var lastVisualization = DateTimeOffset.Now;
evolutionaryAlgorithm.UpdateEvent += onUpdate;
evolutionaryAlgorithm.StartContinue();
Console.WriteLine("Press enter to stop the evolution.");
Console.ReadLine();
Console.WriteLine("Finishing the evolution.");
evolutionaryAlgorithm.Stop();
Console.WriteLine("Evolution is stopped.");
// simulate best individual
Console.WriteLine("Simulating best individual...");
evaluate(evolutionaryAlgorithm.CurrentChampGenome);
Console.WriteLine("Done.");
void onUpdate(object sender, EventArgs e)
{
Console.WriteLine($"Generation #{evolutionaryAlgorithm.CurrentGeneration}");
Console.WriteLine($"- max fitness: {evolutionaryAlgorithm.Statistics._maxFitness}");
Console.WriteLine($"- mean fitness: {evolutionaryAlgorithm.Statistics._meanFitness}");
Console.WriteLine();
if (DateTimeOffset.Now - lastVisualization > TimeSpan.FromSeconds(35))
{
lastVisualization = DateTimeOffset.Now;
Console.WriteLine("Simulating currently best individual...");
evaluate(evolutionaryAlgorithm.CurrentChampGenome);
}
}
void evaluate(NeatGenome genome)
{
var worldId = random.Next(0, worlds.Length - 1);
var world = worlds[worldId];
var bestIndividual = genomeDecoder.Decode(genome);
var agent = new NeuralNetworkAgent(createLidarFor(world.Track), bestIndividual);
var simulation = new Simulation.Simulation(agent, world);
var summary = simulation.Simulate(simulationStep, perceptionStep, maximumSimulationTime);
File.Copy($"{circuits[worldId]}/visualization.svg", "C:/Users/simon/Projects/racer-experiment/simulator/src/visualization.svg", overwrite: true);
IO.Simulation.StoreResult(world.Track, world.VehicleModel, summary, "", "C:/Users/simon/Projects/racer-experiment/simulator/src/report.json");
}
}
/// <summary>
/// Create and return a NeatEvolutionAlgorithm object ready for running the NEAT algorithm/search based on the given
/// genome factory and genome list. Various sub-parts of the algorithm are also constructed and connected up.
/// </summary>
/// <param name="genomeFactory">The genome factory from which to generate new genomes</param>
/// <param name="genomeList">The current genome population</param>
/// <returns>Constructed evolutionary algorithm</returns>
public NeatEvolutionAlgorithm<NeatGenome> CreateEvolutionAlgorithm(IGenomeFactory<NeatGenome> genomeFactory,
List<NeatGenome> genomeList)
{
// Create distance metric. Mismatched genes have a fixed distance of 10; for matched genes the distance is their weigth difference.
IDistanceMetric distanceMetric = new ManhattanDistanceMetric(1.0, 0.0, 10.0);
ISpeciationStrategy<NeatGenome> speciationStrategy =
new ParallelKMeansClusteringStrategy<NeatGenome>(distanceMetric, _parallOptions);
// Create complexity regulation strategy.
var complexityRegulationStrategy =
ExperimentUtils.CreateComplexityRegulationStrategy(_complexityRegulationStrategy, _complexitythreshold);
// Create the evolution algorithm.
var ea = new NeatEvolutionAlgorithm<NeatGenome>(NeatEvolutionAlgorithmParameters, speciationStrategy,
complexityRegulationStrategy);
// Create IBlackBox evaluator.
var mazeNavigationEvaluator = new MazeNavigationEvaluator();
// Create genome decoder.
var genomeDecoder = CreateGenomeDecoder();
// Create a genome list evaluator. This packages up the genome decoder with the genome evaluator.
IGenomeListEvaluator<NeatGenome> listEvaluator =
new ParallelGenomeListEvaluator<NeatGenome, IBlackBox>(genomeDecoder, mazeNavigationEvaluator,
_parallOptions);
// Initialize the evolution algorithm.
ea.Initialize(listEvaluator, genomeFactory, genomeList);
// Finished. Return the evolution algorithm
return ea;
}
