public int CompareTo(object obj)
{
CMAESCandidate other = (CMAESCandidate)obj;
return(this.ObjFunctionValue.CompareTo(other.ObjFunctionValue)); //sorts in acsending, array index of 0 with be highest.
// returns 0 if equal, -1 if less, and 1 if greater.
}
public VariantData(int coreNumIN, int currentEvalIN, int dimensionNumberIN, int lambdaValIN, Matrix <double> xMeanIN, Matrix <double> dMatrixIN, Matrix <double> bMatrixIN, double sigmaIN, Matrix <double> weightsIN, Matrix <double> psIN, double csIN, double mueffIN, double ccIN, Matrix <double> invSqrtCIN, Matrix <double> matrixCIN, double muIN, Matrix <double> pcIN, double c1IN, double cmuIN, double dampsIN, double chiNIN, double eigenValIN, CMAESCandidate globalBestIN)
{
coreNumber = coreNumIN;
currentEval = currentEvalIN;
dimensionNumber = dimensionNumberIN;
lambdaVal = lambdaValIN;
xMean = xMeanIN;
dMatrix = dMatrixIN;
bMatrix = bMatrixIN;
weights = weightsIN;
ps = psIN;
cs = csIN;
mueff = mueffIN;
cc = ccIN;
pc = pcIN;
invSqrtC = invSqrtCIN;
C = matrixCIN;
mu = muIN;
c1 = c1IN;
cmu = cmuIN;
damps = dampsIN;
chiN = chiNIN;
eigenVal = eigenValIN;
globalBest = globalBestIN;
sigma = sigmaIN;
}
//Attack different XOR APUFs (same type) in parallel
//public static void RepeatAttackOnePUFType()
//{
// int attackNumber = 40;
// double[] currentAccuracies = ClassicalAttackXORAPUFMulti(bitNumber, xorNumber, AppConstants.CoreNumber);
//}
//int bitNumber = 128;
//int xorNumber = 4;
//Runs attack multiple times, each time it is on a DIFFERENT XOR APUF
public static double[] ClassicalAttackXORAPUFMulti(int bitNumber, int numXOR, int attackRepeatNumber)
{
//Generate a PUF
double aPUFMean = 0.0;
double aPUFVar = 1.0;
double aPUFMeanNoise = 0.0;
double aPUFNoiseVar = 0.0;
//Create the XOR APUF for parallel runs
XORArbiterPUF xPUF = new XORArbiterPUF(numXOR, bitNumber, aPUFMean, aPUFVar, aPUFMeanNoise, aPUFNoiseVar);
XORArbiterPUF[] xArray = new XORArbiterPUF[attackRepeatNumber];
for (int i = 0; i < xArray.Length; i++)
{
xArray[i] = new XORArbiterPUF(numXOR, bitNumber, aPUFMean, aPUFVar, aPUFMeanNoise, aPUFNoiseVar);
}
sbyte[][] trainingData = new sbyte[AppConstants.TrainingSize][]; //these will be phi vectors
sbyte[][] allTrainingResponses = new sbyte[attackRepeatNumber][]; //first index PUF, second index sample
for (int i = 0; i < attackRepeatNumber; i++)
{
allTrainingResponses[i] = new sbyte[AppConstants.TrainingSize];
}
Random[] rGenArray = new Random[AppConstants.CoreNumber];
for (int i = 0; i < AppConstants.CoreNumber; i++)
{
rGenArray[i] = new Random((int)DateTime.Now.Ticks);
System.Threading.Thread.Sleep(10); //prevent the random number generators from being the same
}
DataGeneration.GenerateTrainingDataParallel(xArray, trainingData, allTrainingResponses, rGenArray);
Console.Out.WriteLine("Data Generation Complete.");
//create the objective function for parallel runs
ObjectiveFunctionResponseXOR[] rObjArray = new ObjectiveFunctionResponseXOR[attackRepeatNumber];
for (int i = 0; i < rObjArray.Length; i++)
{
rObjArray[i] = new ObjectiveFunctionResponseXOR();
}
double[][] solutionList = new double[attackRepeatNumber][];
Random[] randomGeneratorArray = new Random[attackRepeatNumber];
for (int r = 0; r < attackRepeatNumber; r++)
{
randomGeneratorArray[r] = new Random((int)DateTime.Now.Ticks);
System.Threading.Thread.Sleep(10); //prevent the random number generators from being the same
}
//time to save the invariant data
//if (AppConstants.IsLargeData == false)
//{
InvariantData invD = new InvariantData(trainingData, allTrainingResponses, xArray);
string dayString = System.DateTime.Today.ToString();
dayString = dayString.Replace(@"/", "-");
dayString = dayString.Replace(" ", string.Empty);
dayString = dayString.Replace(":", string.Empty);
string invariantDataFileName = "InvariantData" + dayString;
string fName = AppConstants.SaveDir + invariantDataFileName;
FileInfo fi = new FileInfo(fName);
Stream str = fi.Open(FileMode.OpenOrCreate, FileAccess.Write);
BinaryFormatter bf = new BinaryFormatter();
invD.Serialize(bf, str);
str.Close();
var watch = System.Diagnostics.Stopwatch.StartNew();
Parallel.For(0, attackRepeatNumber, a =>
{
Random randomGenerator = randomGeneratorArray[a]; //remove the dependences for parallelization
int dimensionNumber = (bitNumber + 1) * xArray[a].GetPUFNum(); //the weights of all the XOR APUFs
sbyte[] trainingResponse = allTrainingResponses[a];
//Generate the first solution randomly for CMA-ES
double[] firstSolution = new double[dimensionNumber];
for (int i = 0; i < firstSolution.Length; i++)
{
firstSolution[i] = randomGenerator.NextDouble();
}
Console.Out.WriteLine("Beginning CMA-ES run # " + a.ToString());
//CMAESCandidate solutionCMAES = CMAESMethods.ComputeCMAES(dimensionNumber, rObjArray[a], trainingData, trainingResponse, firstSolution, randomGenerator);
CMAESCandidate solutionCMAES = CMAESMethods.ComputeCMAESRecoverable(dimensionNumber, rObjArray[a], trainingData, trainingResponse, firstSolution, randomGenerator, a);
double solutionVal = solutionCMAES.GetObjectiveFunctionValue();
solutionList[a] = solutionCMAES.GetWeightVector(); //store the solution in independent memory
Console.Out.WriteLine("CMA-ES on core " + a.ToString() + " finished.");
});
watch.Stop();
Console.Out.WriteLine("Elapsed Time is " + watch.ElapsedMilliseconds.ToString());
//measure the accuracy
Random randomGenerator2 = new Random((int)DateTime.Now.Ticks);
double averageAccuracy = 0;
double[] solutionAccuracies = new double[attackRepeatNumber];
for (int a = 0; a < solutionList.Length; a++)
{
sbyte[][] testingData = new sbyte[AppConstants.TestingSize][]; //these will be phi vectors
sbyte[] testingResponse = new sbyte[AppConstants.TestingSize];
DataGeneration.GenerateTrainingData(xArray[a], testingData, testingResponse, randomGenerator2);
double accMeasures = rObjArray[0].ObjFunValue(solutionList[a], testingData, testingResponse);
solutionAccuracies[a] = accMeasures;
averageAccuracy = averageAccuracy + accMeasures;
}
averageAccuracy = averageAccuracy / (double)attackRepeatNumber;
Console.Out.WriteLine("The average accuracy for the XOR APUF is " + averageAccuracy.ToString());
return(solutionAccuracies);
}
//Runs the attack on one PUF model, the cores are used to evaluate the CRPs of one model (in parallel) so the method will run as fast as possible
public static double ClassicalAttackXORAPUFSingle(int bitNumber, int numXOR)
{
//Generate a PUF
double aPUFMean = 0.0;
double aPUFVar = 1.0;
double aPUFMeanNoise = 0.0;
double aPUFNoiseVar = 0.0;
//Create the XOR APUF
XORArbiterPUF xPUF = new XORArbiterPUF(numXOR, bitNumber, aPUFMean, aPUFVar, aPUFMeanNoise, aPUFNoiseVar);
//Arrays for storing the training data
sbyte[][] trainingData = new sbyte[AppConstants.TrainingSize][]; //these will be phi vectors
sbyte[][] allTrainingResponses = new sbyte[1][]; //first index PUF, second index sample
allTrainingResponses[0] = new sbyte[AppConstants.TrainingSize];
Random[] rGenArray = new Random[AppConstants.CoreNumber];
for (int i = 0; i < AppConstants.CoreNumber; i++)
{
rGenArray[i] = new Random((int)DateTime.Now.Ticks);
System.Threading.Thread.Sleep(10); //prevent the random number generators from being the same
}
DataGeneration.GenerateTrainingDataParallel(xPUF, trainingData, allTrainingResponses, rGenArray);
Console.Out.WriteLine("Data Generation Complete.");
//create the objective function for parallel runs
ObjectiveFunctionResponseXOR rObj = new ObjectiveFunctionResponseXOR();
//Start the attack run
var watch = System.Diagnostics.Stopwatch.StartNew();
Random randomGenerator = new Random((int)DateTime.Now.Ticks);;
int dimensionNumber = (bitNumber + 1) * xPUF.GetPUFNum(); //the weights of all the XOR APUFs
sbyte[] trainingResponse = allTrainingResponses[0];
//Generate the first solution randomly for CMA-ES
double[] firstSolution = new double[dimensionNumber];
for (int i = 0; i < firstSolution.Length; i++)
{
firstSolution[i] = randomGenerator.NextDouble();
}
Console.Out.WriteLine("Beginning CMA-ES");
//The next line uses maximum parallelism on a single run
CMAESCandidate solutionCMAES = CMAESMethods.ComputeCMAES(dimensionNumber, rObj, trainingData, trainingResponse, firstSolution, randomGenerator);
double solutionVal = solutionCMAES.GetObjectiveFunctionValue();
double[] computedSolution = solutionCMAES.GetWeightVector(); //store the solution in independent memory
Console.Out.WriteLine("CMA-ES finished.");
watch.Stop();
Console.Out.WriteLine("Elapsed Time is " + watch.ElapsedMilliseconds.ToString());
//turn off parallelism
AppConstants.UseParallelismOnSingleCMAES = false;
//measure the accuracy
Random randomGenerator2 = new Random((int)DateTime.Now.Ticks);
sbyte[][] testingData = new sbyte[AppConstants.TestingSize][]; //these will be phi vectors
sbyte[] testingResponse = new sbyte[AppConstants.TestingSize];
DataGeneration.GenerateTrainingData(xPUF, testingData, testingResponse, randomGenerator2);
double accMeasure = rObj.ObjFunValue(computedSolution, testingData, testingResponse);
Console.Out.WriteLine("The accuracy for the XOR APUF is " + accMeasure.ToString());
return(accMeasure);
}
public static double[] ClassicalAttackXORAPUFMultiRecovered(int bitNumber, int attackRepeatNumber, InvariantData invData, VariantData[] variantDataArray)
{
//Create the XOR APUF for parallel runs
XORArbiterPUF[] xArray = new XORArbiterPUF[attackRepeatNumber];
for (int i = 0; i < xArray.Length; i++)
{
xArray[i] = (XORArbiterPUF)invData.GetPUFatIndex(i);
}
sbyte[][] trainingData = invData.GetTrainingData(); //these will be phi vectors
sbyte[][] allTrainingResponses = invData.GetTrainingResponseAll(); //first index PUF, second index sample
//create the objective function for parallel runs
ObjectiveFunctionResponseXOR[] rObjArray = new ObjectiveFunctionResponseXOR[attackRepeatNumber];
for (int i = 0; i < rObjArray.Length; i++)
{
rObjArray[i] = new ObjectiveFunctionResponseXOR();
}
double[][] solutionList = new double[attackRepeatNumber][];
Random[] randomGeneratorArray = new Random[attackRepeatNumber];
for (int r = 0; r < attackRepeatNumber; r++)
{
randomGeneratorArray[r] = new Random((int)DateTime.Now.Ticks);
System.Threading.Thread.Sleep(10); //prevent the random number generators from being the same
}
var watch = System.Diagnostics.Stopwatch.StartNew();
Parallel.For(0, attackRepeatNumber, a =>
{
Random randomGenerator = randomGeneratorArray[a]; //remove the dependences for parallelization
int dimensionNumber = (bitNumber + 1) * xArray[a].GetPUFNum(); //the weights of all the XOR APUFs
sbyte[] trainingResponse = allTrainingResponses[a];
//Generate the first solution randomly for CMA-ES
double[] firstSolution = new double[dimensionNumber];
for (int i = 0; i < firstSolution.Length; i++)
{
firstSolution[i] = randomGenerator.NextDouble();
}
Console.Out.WriteLine("Beginning CMA-ES run # " + a.ToString());
//CMAESCandidate solutionCMAES = CMAESMethods.ComputeCMAES(dimensionNumber, rObjArray[a], trainingData, trainingResponse, firstSolution, randomGenerator);
CMAESCandidate solutionCMAES = CMAESMethods.RecoveredCMAES(randomGenerator, a, rObjArray[a], invData, variantDataArray[a]);
double solutionVal = solutionCMAES.GetObjectiveFunctionValue();
solutionList[a] = solutionCMAES.GetWeightVector(); //store the solution in independent memory
Console.Out.WriteLine("CMA-ES on core " + a.ToString() + " finished.");
//Console.Out.WriteLine("Final training value is "+solutionVal.ToString());
//}
});
watch.Stop();
Console.Out.WriteLine("Elapsed Time is " + watch.ElapsedMilliseconds.ToString());
//measure the accuracy
Random randomGenerator2 = new Random((int)DateTime.Now.Ticks);
double averageAccuracy = 0;
double[] solutionAccuracies = new double[attackRepeatNumber];
for (int a = 0; a < solutionList.Length; a++)
{
sbyte[][] testingData = new sbyte[AppConstants.TestingSize][]; //these will be phi vectors
sbyte[] testingResponse = new sbyte[AppConstants.TestingSize];
DataGeneration.GenerateTrainingData(xArray[a], testingData, testingResponse, randomGenerator2);
double accMeasures = rObjArray[0].ObjFunValue(solutionList[a], testingData, testingResponse);
solutionAccuracies[a] = accMeasures;
averageAccuracy = averageAccuracy + accMeasures;
}
averageAccuracy = averageAccuracy / (double)attackRepeatNumber;
Console.Out.WriteLine("The average accuracy for the XOR APUF is " + averageAccuracy.ToString());
return(solutionAccuracies);
}
//standard CMA-ES with maximum parallelization
public static CMAESCandidate ComputeCMAESBecker(int dimensionNumber, PUFObjectiveFunction pFunction, sbyte[][] trainingData, sbyte[] targets, double[] initialWeightVector, Random randomGenerator)
{
int n = dimensionNumber + 1; //number of weights + epsilon
Matrix <double> xMean = Matrix <double> .Build.Dense(n, 1);
CMAESCandidate c11 = new CMAESCandidate(initialWeightVector, trainingData, targets, pFunction);
Console.Out.WriteLine("Starting Point Accuracy");
Console.Out.WriteLine(c11.GetObjectiveFunctionValue().ToString());
CMAESCandidate globalBestCandidate = null;
double sigma = 0.5;
//double stopfitness = 1e-10;
//double stopfitness = 0.99;
double stopfitness = double.MaxValue;
int stopeval = AppConstants.MaxIterationNumberCMAES;
//double stopeval = AppConstants.MaxEvaluations
//Strategy parameter setting: Selection
int lambdaVal = (int)(4.0 + Math.Floor(3.0 * Math.Log(n))); //population size, note lambda keyword reserved by python so lambda->lambdaVal
//lambdaVal = AppConstants.PopulationSizeCMAES #change by KRM cause I think we need more sampling
double mu = lambdaVal / 2.0;
Matrix <double> weights = Matrix <double> .Build.Dense((int)mu, 1); //weights = numpy.matrix(weightsPrime, dtype = float).reshape(mu, 1)
for (int i = 0; i < (int)mu; i++)
{
weights[i, 0] = Math.Log(mu + 0.5) - Math.Log(i + 1); //weights[i, 0] = math.log(mu + 0.5) - math.log((i + 1))#use i+1 instead of i in this case to match matlab indexing
}
mu = Math.Floor(mu);
double sumWeights = 0.0;
for (int i = 0; i < weights.RowCount; i++)
{
sumWeights = sumWeights + weights[i, 0];
}
//Divide by the sum
for (int i = 0; i < weights.RowCount; i++)
{
weights[i, 0] = weights[i, 0] / sumWeights;
}
//Computation for the mueff variable
double mueffNum = 0.0;
double mueffDem = 0.0;
for (int i = 0; i < weights.RowCount; i++)
{
mueffNum = weights[i, 0] + mueffNum;
mueffDem = weights[i, 0] * weights[i, 0] + mueffDem;
}
mueffNum = mueffNum * mueffNum;
double mueff = mueffNum / mueffDem;
// Strategy parameter setting: Adaptation
double cc = (4.0 + mueff / n) / (n + 4.0 + 2.0 * mueff / n); //#time constant for cumulation for C
double cs = (mueff + 2.0) / (n + mueff + 5.0); //#t-const for cumulation for sigma control
double c1 = 2.0 / ((n + 1.3) * (n + 1.3) + mueff); //#learning rate for rank-one update of C
double cmu = Math.Min(1.0 - c1, 2.0 * (mueff - 2.0 + 1.0 / mueff) / ((n + 2) * (n + 2) + mueff)); //# and for rank-mu update
double damps = 1.0 + 2.0 * Math.Max(0, Math.Sqrt((mueff - 1) / (n + 1)) - 1) + cs; //# damping for sigma
//Initialize dynamic (internal) strategy parameters and constants
//evolution paths for C and sigma
Matrix <double> pc = Matrix <double> .Build.Dense(n, 1);
Matrix <double> ps = Matrix <double> .Build.Dense(n, 1);
Matrix <double> D = Matrix <double> .Build.Dense(n, 1);
for (int i = 0; i < n; i++)
{
pc[i, 0] = 0;
ps[i, 0] = 0;
D[i, 0] = 1.0;
}
//Create B Matrix
Matrix <double> B = Matrix <double> .Build.Dense(n, n);
for (int i = 0; i < n; i++)
{
for (int j = 0; j < n; j++)
{
if (i == j)
{
B[i, j] = 1.0;
}
else
{
B[i, j] = 0.0;
}
}
}
//Create C Matrix
Matrix <double> dSquare = Matrix <double> .Build.Dense(n, 1);
for (int i = 0; i < n; i++)
{
dSquare[i, 0] = D[i, 0] * D[i, 0];
}
Matrix <double> C = B * Diagonalize1DMatrix(dSquare) * B.Transpose(); //C = B * self.diag(DSquare) * numpy.transpose(B)
//Create invertsqrtC Matrix
Matrix <double> oneOverD = Matrix <double> .Build.Dense(n, 1);
for (int i = 0; i < n; i++)
{
oneOverD[i, 0] = 1.0 / D[i, 0];
}
Matrix <double> invsqrtC = B * Diagonalize1DMatrix(oneOverD) * B.Transpose();
double eigeneval = 0; //track update of B and D
double chiN = Math.Pow(n, 0.5) * (1.0 - 1.0 / (4.0 * n) + 1.0 / (21.0 * Math.Pow(n, 2.0)));
int counteval = 0;
//the next 40 lines contain the 20 lines of interesting code
//CMAESCandidate[] candidateArray = new CMAESCandidate[lambdaVal];
List <CMAESCandidate> candidateArray = new List <CMAESCandidate>();
for (int i = 0; i < lambdaVal; i++)
{
candidateArray.Add(new CMAESCandidate());
}
while (counteval < stopeval)
{
Matrix <double> arx = Matrix <double> .Build.Dense(n, lambdaVal);
//fill in the initial solutions
for (int i = 0; i < lambdaVal; i++)
{
Matrix <double> randD = Matrix <double> .Build.Dense(n, 1);
for (int j = 0; j < n; j++)
{
randD[j, 0] = D[j, 0] * GenerateRandomNormalVariableForCMAES(randomGenerator, 0, 1.0);
}
Matrix <double> inputVector = xMean + sigma * B * randD;
double[] tempWeightVector = new double[inputVector.RowCount];
for (int k = 0; k < inputVector.RowCount; k++)
{
tempWeightVector[k] = inputVector[k, 0];
}
candidateArray[i] = new CMAESCandidate(tempWeightVector, trainingData, targets, pFunction);
counteval = counteval + 1;
}
candidateArray.Sort(); //This maybe problematic, not sure about sorting in C#
candidateArray.Reverse();
Matrix <double> xOld = xMean.Clone();
//Get the new mean value
Matrix <double> arxSubset = Matrix <double> .Build.Dense(n, (int)mu); //in Maltab this variable would be "arx(:,arindex(1:mu))
//This replaces line arxSubset[:, i] = CandidateList[i].InputVector
for (int i = 0; i < mu; i++)
{
for (int j = 0; j < n; j++)
{
arxSubset[j, i] = candidateArray[i].GetWeightVector()[j];
}
}
xMean = arxSubset * weights; //Line 76 Matlab
//Cumulation: Update evolution paths
ps = (1 - cs) * ps + Math.Sqrt(cs * (2.0 - cs) * mueff) * invsqrtC * (xMean - xOld) / sigma;
//Compute ps.^2 equivalent
double psSquare = 0;
for (int i = 0; i < ps.RowCount; i++)
{
psSquare = psSquare + ps[i, 0] * ps[i, 0];
}
//Compute hsig
double hSig = 0.0;
double term1ForHsig = psSquare / (1.0 - Math.Pow(1.0 - cs, 2.0 * counteval / lambdaVal)) / n;
double term2ForHsig = 2.0 + 4.0 / (n + 1.0);
if (term1ForHsig < term2ForHsig)
{
hSig = 1.0;
}
//Compute pc, Line 82 Matlab
pc = (1.0 - cc) * pc + hSig * Math.Sqrt(cc * (2.0 - cc) * mueff) * (xMean - xOld) / sigma;
//Adapt covariance matrix C
Matrix <double> repmatMatrix = Tile((int)mu, xOld); //NOT SURE IF THIS IS RIGHT IN C# FIX repmatMatrix = numpy.tile(xold, mu)
Matrix <double> artmp = (1.0 / sigma) * (arxSubset - repmatMatrix);
// C = (1-c1-cmu) * C + c1 * (pc * pc' + (1-hsig) * cc*(2-cc) * C) + cmu * artmp * diag(weights) * artmp' #This is the original Matlab line for reference
C = (1.0 - c1 - cmu) * C + c1 * (pc * pc.Transpose() + (1 - hSig) * cc * (2.0 - cc) * C) + cmu * artmp * Diagonalize1DMatrix(weights) * artmp.Transpose();
//Adapt step size sigma
//sigma = sigma * Math.Exp((cs / damps) * (numpy.linalg.norm(ps) / chiN - 1))
sigma = sigma * Math.Exp((cs / damps) * (ps.L2Norm() / chiN - 1)); //NOT SURE IF THIS IS RIGHT FIX IN C#
//Update B and D from C
if ((counteval - eigeneval) > (lambdaVal / (c1 + cmu) / n / 10.0))
{
eigeneval = counteval;
//C = numpy.triu(C) + numpy.transpose(numpy.triu(C, 1)) #enforce symmetry
C = C.UpperTriangle() + C.StrictlyUpperTriangle().Transpose(); //NOT SURE IF THIS IS RIGHT FIX IN C#
//eigen decomposition
Evd <double> eigen = C.Evd();
B = eigen.EigenVectors;
Vector <System.Numerics.Complex> vectorEigenValues = eigen.EigenValues;
for (int i = 0; i < vectorEigenValues.Count; i++)
{
D[i, 0] = vectorEigenValues[i].Real;
}
//take sqrt of D
for (int i = 0; i < vectorEigenValues.Count; i++)
{
D[i, 0] = Math.Sqrt(D[i, 0]);
}
for (int i = 0; i < n; i++)
{
oneOverD[i, 0] = 1.0 / D[i, 0];
}
Matrix <double> middleTerm = Diagonalize1DMatrix(oneOverD); //#Built in Numpy function doesn't create the right size matrix in this case (ex: Numpy gives 1x1 but should be 5x5)
invsqrtC = B * middleTerm * B.Transpose();
}
globalBestCandidate = candidateArray[0];
//Termination Conditions
//bestCandidate = candidateArray[0]; //Array index 0 has the smallest objective function value
//if (bestCandidate.GetObjectiveFunctionValue() < currentBestValue)
//{
// currentBestValue = bestCandidate.GetObjectiveFunctionValue();
// globalBestCandidate = (CMAESCandidate)bestCandidate.Clone();
//}
//if (bestCandidate.GetObjectiveFunctionValue() >= stopfitness)
//{
// return globalBestCandidate;
//}
//Add in extra termination conditions
//if (bestCandidate.GetObjectiveFunctionValue() == currentBestValue)
//{
// repeatCount = repeatCount + 1;
//}
//else
//{
// repeatCount = 0;
//}
////we have repeated too many times
//if (repeatCount > 10)
//{
// Console.Out.WriteLine("CMA-ES terminated to due to repeated best.");
// return globalBestCandidate;
//}
Console.Out.WriteLine("Iteration #" + counteval.ToString());
//Console.Out.WriteLine("Best Value=" + (1.0 - currentBestValue).ToString());
Console.Out.WriteLine("Current Value=" + (candidateArray[candidateArray.Count - 1].GetObjectiveFunctionValue()).ToString());
}//end while loop
return(globalBestCandidate); //just in case everything terminates
}
//Run the CMA-ES from a recovered file
public static CMAESCandidate RecoveredCMAES(Random randomGenerator, int coreNumberForSaving, PUFObjectiveFunction pFunction, InvariantData invData, VariantData varData)
{
double stopFitness = 0.98;
//non-specific variables (used by each core)
int stopeval = invData.GetMaxEval();
sbyte[][] trainingData = invData.GetTrainingData();
sbyte[] targets = invData.GetTrainingResponseForPUF(coreNumberForSaving);
//core specific variables (specific to each run of CMA-ES)
int counteval = varData.GetCurrentEval();
int n = varData.GetDimensionNum();
int lambdaVal = varData.GetLambda();
Matrix <double> xMean = varData.GetXMean();
Matrix <double> D = varData.GetD();
Matrix <double> B = varData.GetB();
double sigma = varData.GetSigma();
Matrix <double> weights = varData.GetWeights();
Matrix <double> ps = varData.GetPS();
double cs = varData.GetCS();
double mueff = varData.GetMueff();
double cc = varData.GetCC();
Matrix <double> pc = varData.GetPC();
Matrix <double> invsqrtC = varData.GetInvSqrtC();
Matrix <double> C = varData.GetMatrixC();
double mu = varData.GetMu();
double c1 = varData.GetC1();
double cmu = varData.GetCmu();
double damps = varData.GetDamps();
double chiN = varData.GetChiN();
double eigeneval = varData.GetEigenVal();
CMAESCandidate globalBestCandidate = varData.GetGlobalBest();
Matrix <double> oneOverD = Matrix <double> .Build.Dense(n, 1); //Don't need to copy this because we get it from D
//just a sanity check
if (varData.coreNumber != coreNumberForSaving)
{
throw new Exception("The saved core number and current core number don't match.");
}
//the next 40 lines contain the 20 lines of interesting code
//CMAESCandidate[] candidateArray = new CMAESCandidate[lambdaVal];
List <CMAESCandidate> candidateArray = new List <CMAESCandidate>();
for (int i = 0; i < lambdaVal; i++)
{
candidateArray.Add(new CMAESCandidate());
}
while (counteval < stopeval)
{
Matrix <double> arx = Matrix <double> .Build.Dense(n, lambdaVal);
//fill in the initial solutions
for (int i = 0; i < lambdaVal; i++)
{
Matrix <double> randD = Matrix <double> .Build.Dense(n, 1);
for (int j = 0; j < n; j++)
{
randD[j, 0] = D[j, 0] * GenerateRandomNormalVariableForCMAES(randomGenerator, 0, 1.0);
}
Matrix <double> inputVector = xMean + sigma * B * randD;
double[] tempWeightVector = new double[inputVector.RowCount];
for (int k = 0; k < inputVector.RowCount; k++)
{
tempWeightVector[k] = inputVector[k, 0];
}
candidateArray[i] = new CMAESCandidate(tempWeightVector, trainingData, targets, pFunction);
counteval = counteval + 1;
}
candidateArray.Sort(); //This maybe problematic, not sure about sorting in C#
candidateArray.Reverse();
Matrix <double> xOld = xMean.Clone();
//Get the new mean value
Matrix <double> arxSubset = Matrix <double> .Build.Dense(n, (int)mu); //in Maltab this variable would be "arx(:,arindex(1:mu))
//This replaces line arxSubset[:, i] = CandidateList[i].InputVector
for (int i = 0; i < mu; i++)
{
for (int j = 0; j < n; j++)
{
arxSubset[j, i] = candidateArray[i].GetWeightVector()[j];
}
}
xMean = arxSubset * weights; //Line 76 Matlab
//Cumulation: Update evolution paths
ps = (1 - cs) * ps + Math.Sqrt(cs * (2.0 - cs) * mueff) * invsqrtC * (xMean - xOld) / sigma;
//Compute ps.^2 equivalent
double psSquare = 0;
for (int i = 0; i < ps.RowCount; i++)
{
psSquare = psSquare + ps[i, 0] * ps[i, 0];
}
//Compute hsig
double hSig = 0.0;
double term1ForHsig = psSquare / (1.0 - Math.Pow(1.0 - cs, 2.0 * counteval / lambdaVal)) / n;
double term2ForHsig = 2.0 + 4.0 / (n + 1.0);
if (term1ForHsig < term2ForHsig)
{
hSig = 1.0;
}
//Compute pc, Line 82 Matlab
pc = (1.0 - cc) * pc + hSig * Math.Sqrt(cc * (2.0 - cc) * mueff) * (xMean - xOld) / sigma;
//Adapt covariance matrix C
Matrix <double> repmatMatrix = Tile((int)mu, xOld); //NOT SURE IF THIS IS RIGHT IN C# FIX repmatMatrix = numpy.tile(xold, mu)
Matrix <double> artmp = (1.0 / sigma) * (arxSubset - repmatMatrix);
// C = (1-c1-cmu) * C + c1 * (pc * pc' + (1-hsig) * cc*(2-cc) * C) + cmu * artmp * diag(weights) * artmp' #This is the original Matlab line for reference
C = (1.0 - c1 - cmu) * C + c1 * (pc * pc.Transpose() + (1 - hSig) * cc * (2.0 - cc) * C) + cmu * artmp * Diagonalize1DMatrix(weights) * artmp.Transpose();
//Adapt step size sigma
//sigma = sigma * Math.Exp((cs / damps) * (numpy.linalg.norm(ps) / chiN - 1))
sigma = sigma * Math.Exp((cs / damps) * (ps.L2Norm() / chiN - 1)); //NOT SURE IF THIS IS RIGHT FIX IN C#
//Update B and D from C
if ((counteval - eigeneval) > (lambdaVal / (c1 + cmu) / n / 10.0))
{
eigeneval = counteval;
//C = numpy.triu(C) + numpy.transpose(numpy.triu(C, 1)) #enforce symmetry
C = C.UpperTriangle() + C.StrictlyUpperTriangle().Transpose(); //NOT SURE IF THIS IS RIGHT FIX IN C#
//eigen decomposition
Evd <double> eigen = C.Evd();
B = eigen.EigenVectors;
Vector <System.Numerics.Complex> vectorEigenValues = eigen.EigenValues;
for (int i = 0; i < vectorEigenValues.Count; i++)
{
D[i, 0] = vectorEigenValues[i].Real;
}
//take sqrt of D
for (int i = 0; i < vectorEigenValues.Count; i++)
{
D[i, 0] = Math.Sqrt(D[i, 0]);
}
for (int i = 0; i < n; i++)
{
oneOverD[i, 0] = 1.0 / D[i, 0];
}
Matrix <double> middleTerm = Diagonalize1DMatrix(oneOverD); //#Built in Numpy function doesn't create the right size matrix in this case (ex: Numpy gives 1x1 but should be 5x5)
invsqrtC = B * middleTerm * B.Transpose();
}
globalBestCandidate = candidateArray[0];
//Stopping condition
if (globalBestCandidate.GetObjectiveFunctionValue() > stopFitness)
{
return(globalBestCandidate);
}
//Time to save the state
VariantData varDataUpdated = new VariantData(coreNumberForSaving, counteval, n, lambdaVal, xMean, D, B, sigma, weights, ps, cs, mueff, cc, invsqrtC, C, mu, pc, c1, cmu, damps, chiN, eigeneval, globalBestCandidate);
string fname = AppConstants.SaveDir + counteval.ToString() + "VariantData" + coreNumberForSaving.ToString();
FileInfo fi = new FileInfo(fname);
Stream str = fi.Open(FileMode.OpenOrCreate, FileAccess.Write);
BinaryFormatter bf = new BinaryFormatter();
bf.Serialize(str, varDataUpdated);
str.Close();
Console.Out.WriteLine("Iteration #" + counteval.ToString());
Console.Out.WriteLine("Current Value=" + (candidateArray[candidateArray.Count - 1].GetObjectiveFunctionValue()).ToString());
}//end while loop
return(globalBestCandidate); //just in case everything terminates
}
