public override void Setup(IEvolutionState state, IParameter paramBase)
{
base.Setup(state, paramBase);
if (!(Species is FloatVectorSpecies))
{
state.Output.Error("PSOSubpopulation requires that its Species is ec.vector.FloatVectorSpecies or a subclass. Yours is: "
+ Species.GetType(), null, null);
}
if (!(Species.I_Prototype is DoubleVectorIndividual))
{
state.Output.Error("PSOSubpopulation requires that its Species' prototypical individual be is ec.vector.DoubleVectorSpecies or a subclass. Yours is: " + Species.GetType(), null, null);
}
NeighborhoodBests = new DoubleVectorIndividual[Individuals.Count];
PersonalBests = new DoubleVectorIndividual[Individuals.Count];
PreviousIndividuals = new DoubleVectorIndividual[Individuals.Count];
NeighborhoodSize = state.Parameters.GetInt(paramBase.Push(P_NEIGHBORHOOD_SIZE), null);
ClampRange = state.Parameters.GetBoolean(paramBase.Push(P_CLAMP_RANGE), null, false);
InitialVelocityScale = state.Parameters.GetDouble(paramBase.Push(P_INITIAL_VELOCITY_SCALE), null, 0);
VelocityMultiplier = state.Parameters.GetDouble(paramBase.Push(P_VELOCITY_MULTIPLIER), null, 0.1);
pFactor = state.Parameters.GetDoubleWithDefault(paramBase.Push(P_P_FACTOR), null, 1);
nFactor = state.Parameters.GetDoubleWithDefault(paramBase.Push(P_N_FACTOR), null, 1);
gFactor = state.Parameters.GetDoubleWithDefault(paramBase.Push(P_G_FACTOR), null, 1);
}
public void DoubleVectorIndividualWriteAndRead()
{
// First we'll set up a Fitness for the Individual
var rand = new MersenneTwisterFast(0);
var f = new SimpleFitness();
f.SetFitness(null, float.MaxValue, true);
const int n = 10;
f.Trials = new List <double>(n);
for (var i = 0; i < n; i++)
{
f.Trials.Add(rand.NextDouble());
}
// Now we can create and initialize the Individual
var ind = new DoubleVectorIndividual();
var ind2 = new DoubleVectorIndividual(); // We'll read back into this instance
ind.Genome = new double[10]; // This is the set of genes
for (var i = 0; i < 10; i++)
{
ind.genome[i] = rand.NextDouble() * double.MaxValue; // some random genes
}
ind.Fitness = f;
ind.Evaluated = true;
using (var ms = new MemoryStream())
{
var writer = new BinaryWriter(ms);
ind.WriteIndividual(null, writer);
ms.Position = 0;
var reader = new BinaryReader(ms);
ind2.Fitness = new SimpleFitness();
ind2.ReadIndividual(null, reader);
Assert.IsTrue(ind.Fitness.EquivalentTo(ind2.Fitness));
Assert.IsTrue(ind2.Fitness.IsIdeal);
Assert.IsTrue(ind.Equals(ind2)); // Genetically equivalent
for (var i = 0; i < 10; i++)
{
Assert.AreEqual(ind.genome[i], ind2.genome[i]); // check each gene
}
}
}
public void Evaluate(IEvolutionState state, Individual ind, int subpopulation, int threadnum)
{
if (ind.Evaluated)
{
return;
}
DoubleVectorIndividual _ind = (DoubleVectorIndividual)ind;
int vertexSkip = NumVertices * 2 + 4; // for four colors
pic.Clear();
for (int i = 0; i < _ind.genome.Length; i += vertexSkip)
{
pic.AddPolygon(_ind.genome, i, NumVertices);
}
double error = pic.Error();
((SimpleFitness)_ind.Fitness).SetFitness(state, (float)(1.0 - error), error == 0);
ind.Evaluated = true;
}
/** Revises the CMA-ES distribution to reflect the current fitness results in the provided subpopulation. */
public void UpdateDistribution(IEvolutionState state, Subpopulation subpop)
{
// % Sort by fitness and compute weighted mean into xmean
// [arfitness, arindex] = sort(arfitness); % minimization
// xmean = arx(:,arindex(1:mu))*weights; % recombination % Eq.39
// counteval += lambda;
// only need partial sort?
((List <Individual>)subpop.Individuals).Sort();
SimpleMatrixD artmp = new SimpleMatrixD(GenomeSize, mu);
SimpleMatrixD xold = xmean;
xmean = new SimpleMatrixD(GenomeSize, 1);
for (int i = 0; i < mu; i++)
{
DoubleVectorIndividual dvind = (DoubleVectorIndividual)subpop.Individuals[i];
// won't modify the genome
SimpleMatrixD arz = new SimpleMatrixD(GenomeSize, 1, true, dvind.genome);
arz = (arz.minus(xold).divide(sigma));
for (int j = 0; j < GenomeSize; j++)
{
xmean.set(j, 0, xmean.get(j, 0) + weights[i] * dvind.genome[j]);
artmp.set(j, i, arz.get(j, 0));
}
}
// % Cumulation: Update evolution paths
SimpleMatrixD y = xmean.minus(xold).divide(sigma);
SimpleMatrixD bz = invsqrtC.mult(y);
SimpleMatrixD bz_scaled = bz.scale(Math.Sqrt(cs * (2.0 - cs) * mueff));
ps = ps.scale(1.0 - cs).plus(bz_scaled);
double h_sigma_value =
((ps.dot(ps) / (1.0 - Math.Pow(1.0 - cs, 2.0 * (state.Generation + 1)))) / GenomeSize);
int hsig = (h_sigma_value < (2.0 + (4.0 / (GenomeSize + 1)))) ? 1 : 0;
SimpleMatrixD y_scaled = y.scale(hsig * Math.Sqrt(cc * (2.0 - cc) * mueff));
pc = pc.scale(1.0 - cc).plus(y_scaled);
// % Adapt covariance matrix C
c = c.scale(1.0 - c1 - cmu);
c = c.plus(pc.mult(pc.transpose()).plus(c.scale((1.0 - hsig) * cc * (2.0 - cc))).scale(c1));
c = c.plus((artmp.mult(SimpleMatrixD.diag(weights).mult(artmp.transpose()))).scale(cmu));
// % Adapt step-size sigma
sigma = sigma * Math.Exp((cs / damps) * (ps.normF() / chiN - 1.0));
// % Update B and D from C
if ((state.Generation - lastEigenDecompositionGeneration) > 1.0 / ((c1 + cmu) * GenomeSize * 10.0))
{
lastEigenDecompositionGeneration = state.Generation;
// make sure the matrix is symmetric (it should be already)
// not sure if this is necessary
for (int i = 0; i < GenomeSize; i++)
{
for (int j = 0; j < i; j++)
{
c.set(j, i, c.get(i, j));
}
}
// this copy gets modified by the decomposition
DMatrixRMaj copy = c.copy().getMatrix();
EigenDecomposition <DMatrixRMaj> eig = DecompositionFactory_DDRM.eig(GenomeSize, true, true);
if (eig.decompose(copy))
{
SimpleMatrixD dinv = new SimpleMatrixD(GenomeSize, GenomeSize);
for (int i = 0; i < GenomeSize; i++)
{
double eigrt = Math.Sqrt(eig.getEigenValue(i).real);
d.set(i, i, eigrt);
dinv.set(i, i, 1 / eigrt);
CommonOps_DDRM.insert(eig.getEigenVector(i), b.getMatrix(), 0, i);
}
invsqrtC = b.mult(dinv.mult(b.transpose()));
CommonOps_DDRM.mult(b.getMatrix(), d.getMatrix(), bd);
}
else
{
state.Output.Fatal("CMA-ES eigendecomposition failed. ");
}
}
CommonOps_DDRM.scale(sigma, bd, sbd);
// % Break, if fitness is good enough or condition exceeds 1e14, better termination methods are advisable
// if arfitness(1) <= stopfitness || max(D) > 1e7 * min(D)
// break;
// end
if (useAltTermination && CommonOps_DDRM.elementMax(d.diag().getMatrix()) >
1e7 * CommonOps_DDRM.elementMin(d.diag().getMatrix()))
{
state.Evaluator.SetRunCompleted("CMAESSpecies: Stopped because matrix condition exceeded limit.");
}
}
public override Individual NewIndividual(IEvolutionState state, int thread)
{
Individual newind = base.NewIndividual(state, thread);
IMersenneTwister random = state.Random[thread];
if (!(newind is DoubleVectorIndividual)) // uh oh
{
state.Output.Fatal(
"To use CMAESSpecies, the species must be initialized with a DoubleVectorIndividual. But it contains a " +
newind);
}
DoubleVectorIndividual dvind = (DoubleVectorIndividual)(newind);
DMatrixRMaj genome = DMatrixRMaj.wrap(GenomeSize, 1, dvind.genome);
DMatrixRMaj temp = new DMatrixRMaj(GenomeSize, 1);
// arz(:,k) = randn(N,1); % standard normally distributed vector
// arx(:,k) = xmean + sigma*(B*D*arz(:,k));
int tries = 0;
while (true)
{
for (int i = 0; i < GenomeSize; i++)
{
dvind.genome[i] = random.NextGaussian();
}
CommonOps_DDRM.mult(sbd, genome, temp); // temp = sigma*b*d*genome;
CommonOps_DDRM.add(temp, xmean.getMatrix(), genome); // genome = temp + xmean;
bool invalid_value = false;
for (int i = 0; i < GenomeSize; i++)
{
if (dvind.genome[i] < MinGenes[i] || dvind.genome[i] > MaxGenes[i])
{
if (useAltGenerator && tries > altGeneratorTries)
{
// instead of just failing, we're going to select uniformly from
// possible values for this particular gene.
dvind.genome[i] = state.Random[thread].NextDouble() * (MaxGenes[i] - MinGenes[i]) +
MinGenes[i];
}
else
{
invalid_value = true;
break;
}
}
}
if (invalid_value)
{
if (++tries > MAX_TRIES_BEFORE_WARNING)
{
state.Output.WarnOnce(
"CMA-ES may be slow because many individuals are being generated which\n" +
"are outside the min/max gene bounds. If an individual violates a single\n" +
"gene bounds, it is rejected, so as the number of genes grows, the\n" +
"probability of this happens increases exponentially. You can deal\n" +
"with this by decreasing sigma. Alternatively you can use set\n" +
"pop.subpop.0.alternative-generation=true (see the manual).\n" +
"Finally, if this is happening during initialization, you might also\n" +
"change pop.subpop.0.species.covariance=scaled.\n");
}
continue;
}
return(newind);
}
}
