private void _CalcCorr90()
{
_corr90 = new List <CorrRes>(_cntPoint90 - _cntItem90);
for (int i = 0; i < (_cntPoint90 - _cntItem90); i++)
{
CorrRes tmp = new CorrRes();
double[] res = new double[(_cntPoint90 - _cntItem90)];
tmp.i = i;
for (int j = 0; j < (_cntPoint90 - _cntItem90); j++)
{
res[j] = Covariance.Correlation(_A90, _resDict90[j]);
}
tmp.max = res.Max();
tmp.j_max = Array.IndexOf(res, tmp.max);
tmp.min = res.Min();
tmp.j_min = Array.IndexOf(res, tmp.min);
_corr90.Add(tmp);
}
int[] arr_i = (from el in _corr90 orderby el.max select el.i).ToArray <int>();
int[] arr_j_max = (from el in _corr90 orderby el.max select el.j_max).ToArray <int>();
int j_max = arr_j_max.Last();
Print("(90) j_max = " + j_max);
_DrawGraphArray90(_resDict90[j_max], _xTime90, _resDict90[j_max + _cntItem90 - 1]);
}
private void EvaluateCovariance(IObjectiveModel objective)
{
objective.EvaluateAt(objective.Point); // Hessian may be not yet updated.
var Hessian = objective.Hessian;
if (Hessian == null || objective.DegreeOfFreedom < 1)
{
Covariance = null;
Correlation = null;
StandardErrors = null;
return;
}
Covariance = Hessian.PseudoInverse() * objective.Value / objective.DegreeOfFreedom;
if (Covariance != null)
{
StandardErrors = Covariance.Diagonal().PointwiseSqrt();
var correlation = Covariance.Clone();
var d = correlation.Diagonal().PointwiseSqrt();
var dd = d.OuterProduct(d);
Correlation = correlation.PointwiseDivide(dd);
}
else
{
StandardErrors = null;
Correlation = null;
}
}
private static void covariance_initialize(int par_num)
//****************************************************************************80
//
// Discussion:
//
// Note that VALUE.FACTOR is the upper triangular Cholesky factor of the
// covariance matrix VALUE_ARRAY, so that
//
// VALUE.ARRAY = VALUE.FACTOR' * VALUE.FACTOR
//
// Licensing:
//
// This code is distributed under the GNU LGPL license.
//
// Modified:
//
// 26 June 2013
//
// Author:
//
// John Burkardt
//
{
int i;
c = new Covariance
{
//
// Set ORDER
//
order = par_num
};
//
// Set ARRAY.
//
covariance_set(c.order);
//
// Set FACTOR.
//
c.factor = typeMethods.r8mat_pofac(c.order, c.array);
//
// Set INV
//
c.inv = typeMethods.r8mat_poinv(c.order, c.factor);
//
// Set DET.
//
c.det = typeMethods.r8mat_podet(c.order, c.factor);
//
// Set MEAN.
//
c.mean = new double[c.order];
for (i = 0; i < c.order; i++)
{
c.mean[i] = 0.0;
}
}
public PearsonsCorrelation(Covariance covariance)
{
RealMatrix covarianceMatrix = covariance.GetCovarianceMatrix();
if (covarianceMatrix == null)
{
//throw new NullArgumentException(LocalizedFormats.COVARIANCE_MATRIX, new Object[0]);
}
this.nObs = covariance.getN();
this.correlationMatrix = CovarianceToCorrelation(covarianceMatrix);
}
public void Maximize(LLNASufficientStatistics sufficientStatistics)
{
for (int i = 0; i < K - 1; i++)
{
Mu[i] = sufficientStatistics.MuSufficientStatistic[i] / sufficientStatistics.NData;
}
for (int i = 0; i < K - 1; i++)
{
for (int j = 0; j < K - 1; j++)
{
Covariance[i, j] =
(1 / sufficientStatistics.NData) *
(sufficientStatistics.CovarianceSufficientStatistic[i, j] +
sufficientStatistics.NData * Mu[i] * Mu[j] -
sufficientStatistics.MuSufficientStatistic[i] * Mu[j] -
sufficientStatistics.MuSufficientStatistic[j] * Mu[i]);
}
}
if (_parameters.CovEstimate == SHRINK)
{
ShrinkCov(Covariance, sufficientStatistics.NData);
}
InvCovariance = Covariance.Inverse();
LogDeterminantInvCovariance = InvCovariance.ComputeLNDeterminant();
for (int i = 0; i < K; i++)
{
double sum = 0;
for (int j = 0; j < LogBeta.ColumnCount; j++)
{
sum += sufficientStatistics.BetaSufficientStatistic[i, j];
}
if (sum == 0)
{
sum = MathHelper.SafeLog(sum) * LogBeta.ColumnCount;
}
else
{
sum = MathHelper.SafeLog(sum);
}
for (int j = 0; j < LogBeta.ColumnCount; j++)
{
LogBeta[i, j] = MathHelper.SafeLog(sufficientStatistics.BetaSufficientStatistic[i, j]) - sum;
}
}
}
public void Test()
{
//Covariance<Base> coB = new CoContraVarianceCollection();
//Covariance<Derived> coD = coB;
Covariance <Derived> coD = new CoContraVariance();
Covariance <Base> coB = coD;
//Contravariance<Derived> contraD = new CoContraVarianceCollection();
//Contravariance<Base> contraB = contraD;
Contravariance <Base> contraB = new CoContraVariance();
Contravariance <Derived> contraD = contraB;
}
public bool Equals(IDistribution other)
{
var d = other as MultivariateDiscreteDistribution;
if (d == null)
{
//throw new ApplicationException("Incompatable types, other is not UnivariateDiscreteDistribution");
return(false);
}
return(VectorExtentions.EqualsTo(Variance, d.Variance) &&
VectorExtentions.EqualsTo(Mean, d.Mean) &&
Covariance.EqualsTo(d.Covariance) &&
Dimension == d.Dimension);
}
static void varianceInDelegates()
{
//with variance, the method does not need to match with the delegate
//Covariance | return type | Delegate return Parent | Method return a Child
//ContraVariance | parameter type | Delegate recive as parameter a Child | Method recive as parameter Parent
Covariance conv = new Covariance(covarianceMethod);
conv();
Contravariance cont = new Contravariance(contraVarianceMethod);
Child c = new Child();
contraVarianceMethod(c);
}
public void TestMethod001()
{
// Arrange
IEnumerable <Cat> cats = new List <Cat>
{
new Cat("Troublemaker"),
new Cat("Alpithar")
};
Covariance covariance = new Covariance();
// Act
IEnumerable <string> result = covariance.GetAnimalNames(cats);
// Assert
Assert.IsTrue(result.Any(a => a.Equals("TroubleMaker", StringComparison.OrdinalIgnoreCase)));
}
public void Merge(Covariance Group)
{
// Put your code here
List <double> firstListToCopy = Group.FirstList;
List <double> secondListToCopy = Group.SecondList;
this.sumOfFirst += Group.SumOfFirst;
this.sumOfSecond += Group.SumOfSecond;
foreach (double d in firstListToCopy)
{
this.firstList.Add(d);
}
foreach (double d in secondListToCopy)
{
this.secondList.Add(d);
}
}
public void Merge(Covariance Group)
{
// Put your code here
List<double> firstListToCopy = Group.FirstList;
List<double> secondListToCopy = Group.SecondList;
this.sumOfFirst += Group.SumOfFirst;
this.sumOfSecond += Group.SumOfSecond;
foreach (double d in firstListToCopy)
{
this.firstList.Add(d);
}
foreach (double d in secondListToCopy)
{
this.secondList.Add(d);
}
}
/// <summary>
/// 显示波段统计量
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
private void Comparetwobands_Load(object sender, EventArgs e)
{
this.textBox1.Text += "波段" + (i + 1);
this.textBox1.Text += "\t均值:";
this.textBox1.Text += average1.ToString();
this.textBox1.Text += "\t标准差:";
this.textBox1.Text += standard1.ToString();
this.textBox1.Text += "\r\n\r\n\r\n";
this.textBox1.Text += "波段" + (j + 1);
this.textBox1.Text += "\t均值:";
this.textBox1.Text += average2.ToString();
this.textBox1.Text += "\t标准差:";
this.textBox1.Text += standard2.ToString();
this.textBox1.Text += "\r\n\r\n\r\n";
this.textBox1.Text += "协方差:";
this.textBox1.Text += Covariance.ToString();
this.textBox1.Text += "\r\n相关系数:";
this.textBox1.Text += Correlation.ToString();
}
public static void test()
{
ProblemSize ps = new()
{
chain_num = 10,
cr_num = 3,
gen_num = 10,
pair_num = 3,
//par_num = 100,
par_num = 5
};
c = new Covariance(ps.par_num);
ProblemValue pv = new()
{
chain_filename = "problem1_chain00.txt",
gr_filename = "problem1_gr.txt",
gr_threshold = 1.2,
jumpstep = 5,
printstep = 10,
restart_read_filename = "",
restart_write_filename = "problem1_restart.txt",
limits = new double[ps.par_num * 2]
};
for (int j = 0; j < ps.par_num; j++)
{
pv.limits[0 + j * 2] = 9.9;
pv.limits[1 + j * 2] = +10.0;
}
Dream.dream(ref ps, ref pv, prior_sample, prior_density, sample_likelihood);
}
private static Dream.DensityResult prior_density(int par_num, double[] zp, int zpIndex = 0)
public override string ToString()
{
return("LeastSquareResults [chiSq=" + ChiSq + ", fit parameters=" + FitParameters.ToString() +
", model parameters= " + ModelParameters.ToString() + ", covariance="
+ Covariance.ToString() + "]");
}
Covariance(Covariance cc1, double v1, Covariance cc2, double v2)
{
vi = cc1.vi + cc2.vi;
vj = cc1.vj + cc2.vj;
value = v1 + v2;
}
public static void Main(string[] args)
{
Console.WriteLine("Hello World!");
// TODO: Implement Functionality Here
Animal a = new Animal();
SubCat c = new SubCat();
Dog d = new Dog();
Covariance cov = new Covariance();
Console.WriteLine("---------------- Test generic animal");
a.Out();
cov.feed(a);
Console.WriteLine("---------------- Func now");
cov.funcfeed(() => a);
Console.WriteLine("---------------- and Lazy!");
var l_a = new Lazy<Animal>();
cov.lazyfeed(l_a);
Console.WriteLine("created {0}",l_a.IsValueCreated);
Console.WriteLine("---------------- Func+Lazy");
cov.funcfeed(() => l_a.Value);
Console.WriteLine("created {0}",l_a.IsValueCreated);
l_a.Value.Out();
Console.WriteLine("---------------- CovLazy!!");
var cl_a = new CovLazy<Animal>();
cov.covlazyfeed(cl_a);
Console.WriteLine("fired {0}",cl_a.IsFired);
cl_a.MyVal.Out();
Console.WriteLine("---------------- Test cat");
c.Out();
cov.feed(c);
Console.WriteLine("---------------- Func now");
cov.funcfeed(() => c);
Console.WriteLine("---------------- and Lazy!");
var l_c = new Lazy<SubCat>();
cov.lazyfeed(l_c);
Console.WriteLine("created {0}",l_c.IsValueCreated);
Console.WriteLine("---------------- Func+Lazy");
cov.funcfeed(() => l_c.Value);
Console.WriteLine("created {0}",l_c.IsValueCreated);
l_c.Value.Out();
Console.WriteLine("---------------- CovLazy!!");
var cl_c = new CovLazy<SubCat>();
cov.covlazyfeed(cl_c);
Console.WriteLine("fired {0}",cl_c.IsFired);
cl_c.MyVal.Out();
Console.WriteLine("---------------- Test dog");
d.Out();
cov.feed(new Dog());
Console.WriteLine("---------------- Func now");
cov.funcfeed(() => d);
Console.WriteLine("---------------- and Lazy!");
var l_d = new Lazy<Dog>();
cov.lazyfeed(l_d);
Console.WriteLine("created {0}",l_d.IsValueCreated);
Console.WriteLine("---------------- Func+Lazy");
cov.funcfeed(() => l_d.Value);
Console.WriteLine("created {0}",l_d.IsValueCreated);
l_d.Value.Out();
Console.WriteLine("---------------- CovLazy!!");
var cl_d = new CovLazy<Dog>();
cov.covlazyfeed(cl_d);
Console.WriteLine("fired {0}",cl_d.IsFired);
cl_d.MyVal.Out();
Console.Write("Press any key to continue . . . ");
Console.ReadKey(true);
}
private static void BaseCovarianceRunner()
{
NewSubject.WriteLine("Base Covariance");
Covariance.Run();
}
public void Update <TMeasurements>(Vector <double> motion, TMeasurements measurements) where TMeasurements : IEnumerable <LandmarkMeasurement>
{
// expand kalman filter to accomodate new landmarks, if necessary
var prevMeanCount = Mean.Count;
var maxIndex = measurements.Max(m => (int?)m.Index);
if (maxIndex != null)
{
var m = LandmarkDim * maxIndex.Value + StateDim;
if (m >= prevMeanCount)
{
var mean = new DenseVector(m + LandmarkDim);
var covariance = new DenseMatrix(m + LandmarkDim);
covariance.SetDiagonal(DenseVector.Create(m + LandmarkDim, 10000000));
Mean.CopySubVectorTo(mean, 0, 0, Mean.Count);
covariance.SetSubMatrix(0, Covariance.RowCount, 0, Covariance.ColumnCount, Covariance);
Mean = mean;
Covariance = covariance;
}
}
var N = (Mean.Count - StateDim) / LandmarkDim;
var Fx = new DenseMatrix(StateDim, StateDim + LandmarkDim * N);
for (int i = 0; i < StateDim; i++)
{
Fx[i, i] = 1;
}
var motionModel = new DenseVector(new[] { motion[0] * Math.Cos(Mean[2]), motion[0] * Math.Sin(Mean[2]), motion[1] });
var motionJacobian = DenseMatrix.OfArray(new[, ]
{
{ 0, 0, -motion[0] * Math.Sin(Mean[2]) },
{ 0, 0, motion[0] * Math.Cos(Mean[2]) },
{ 0, 0, 0 }
});
// Integrate motion
var identity = DenseMatrix.CreateIdentity(Mean.Count);
Mean += Fx.TransposeThisAndMultiply(motionModel.ToColumnMatrix()).Column(0);
var jacobian = identity + Fx.TransposeThisAndMultiply(motionJacobian) * Fx;
Covariance = jacobian * Covariance.TransposeAndMultiply(jacobian) + Fx.TransposeThisAndMultiply(MotionNoise) * Fx;
// Auxiliary variables
var positionEstimate = new DenseVector(new[] { Mean[0], Mean[1] });
var Fxj = new DenseMatrix(StateDim + LandmarkDim, LandmarkDim * N + StateDim);
for (int i = 0; i < StateDim; i++)
{
Fxj[i, i] = 1;
}
// Integrate measurements
foreach (var measurement in measurements)
{
// m is the corrected index of the landmark
var m = LandmarkDim * measurement.Index + StateDim;
if (m >= prevMeanCount)
{
// landmark never seen before
Mean[m] = Mean[0] + measurement.Measurement[0] * Math.Cos(measurement.Measurement[1] + Mean[2]);
Mean[m + 1] = Mean[1] + measurement.Measurement[0] * Math.Sin(measurement.Measurement[1] + Mean[2]);
}
var landmarkEstimate = new DenseVector(new[] { Mean[m], Mean[m + 1] });
var landmarkDelta = landmarkEstimate - positionEstimate;
var q = landmarkDelta.DotProduct(landmarkDelta);
var sqrtQ = Math.Sqrt(q);
var measurementModel = new DenseVector(new[] { sqrtQ, Math.Atan2(landmarkDelta[1], landmarkDelta[0]) - Mean[2] });
for (int i = 0; i < LandmarkDim; i++)
{
Fxj[StateDim + i, m + i] = 1;
}
var measurementJacobian = (1 / q) * DenseMatrix.OfArray(new[, ]
{
{ -sqrtQ * landmarkDelta[0], -sqrtQ * landmarkDelta[1], 0, sqrtQ * landmarkDelta[0], sqrtQ * landmarkDelta[1] },
{ landmarkDelta[1], -landmarkDelta[0], -q, -landmarkDelta[1], landmarkDelta[0] }
}) * Fxj;
var kalmanGain = Covariance * measurementJacobian.TransposeThisAndMultiply((measurementJacobian * Covariance.TransposeAndMultiply(measurementJacobian) + MeasurementNoise).Inverse());
Mean += kalmanGain * (measurement.Measurement - measurementModel);
Covariance = (identity - kalmanGain * measurementJacobian) * Covariance;
for (int i = 0; i < LandmarkDim; i++)
{
Fxj[StateDim + i, m + i] = 0;
}
}
}
public Controller()
{
FF = new FormingFilter(3, 5000, 0.01, 11.427, 3.464, 0.286, 4.414);
CM = new Covariance(3, 5000, 1.621, 0.542, 1.732, 2.903, 1.973, 0.01, 11.427, 3.464, 0.286, 4.414);
KF = new KalmanFilter(3, 5000, FF.OutputSignal, CM);
}