public void CholeskyDecompositionConstructorTest()
{
// Based on tests by Ken Johnson
double[,] value = // positive-definite
{
{ 2, -1, 0 },
{ -1, 2, -1 },
{ 0, -1, 2 }
};
double[,] expected =
{
{ 1.4142, 0.0000, 0.0000 },
{ -0.7071, 1.2247, 0.0000 },
{ 0.0000, -0.8165, 1.1547 }
};
CholeskyDecomposition chol = new CholeskyDecomposition(value);
double[,] L = chol.LeftTriangularFactor;
Assert.IsTrue(Matrix.IsEqual(L, expected, 0.0001));
// Decomposition Identity
Assert.IsTrue(Matrix.IsEqual(L.Multiply(L.Transpose()), value, 0.001));
Assert.AreEqual(new LuDecomposition(value).Determinant, chol.Determinant, 1e-10);
Assert.AreEqual(true, chol.PositiveDefinite);
Assert.AreEqual(true, chol.Symmetric);
}
/// <summary>
/// Constructs a multivariate Gaussian distribution
/// with given mean vector and covariance matrix.
/// </summary>
public NormalDistribution(double[] mean, double[,] covariance)
: base(mean.Length)
{
int k = mean.Length;
this.mean = mean;
this.covariance = covariance;
variance = covariance.Diagonal();
double detSqrt = System.Math.Sqrt(System.Math.Abs(covariance.Determinant()));
constant = 1.0/(System.Math.Pow(2.0*System.Math.PI, k/2.0)*detSqrt);
chol = new CholeskyDecomposition(covariance, true);
if (chol.Determinant == 0)
{
// The covariance matrix is singular, use pseudo-inverse
svd = new SingularValueDecomposition(covariance);
}
}
public void InverseTestNaN()
{
int n = 5;
var I = Matrix.Identity(n);
for (int i = 0; i < n; i++)
{
for (int j = 0; j < n; j++)
{
double[,] value = Matrix.Magic(n);
// Make symmetric
value = Matrix.Multiply(value, value.Transpose());
value[i, j] = double.NaN;
value[j, i] = double.NaN;
Assert.IsTrue(value.IsSymmetric());
bool thrown = false;
var target = new CholeskyDecomposition(value);
try
{
target.Solve(I);
}
catch (Exception)
{
thrown = true;
}
Assert.IsTrue(thrown);
}
}
}
public void SolveTest2()
{
double[,] value = // not positive-definite
{
{ 6, -1, 2, 6 },
{ -1, 3, -3, -2 },
{ 2, -3, 2, 0 },
{ 6, -2, 0, 0 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, true);
double[,] L = chol.LeftTriangularFactor;
double[,] B = Matrix.Identity(4);
double[,] expected =
{
{ 0.4000, 1.2000, 1.4000, -0.5000 },
{ 1.2000, 3.6000, 4.2000, -2.0000 },
{ 1.4000, 4.2000, 5.4000, -2.5000 },
{ -0.5000, -2.0000, -2.5000, 1.0000 },
};
double[,] actual = chol.Solve(B);
Assert.IsTrue(Matrix.IsEqual(expected, actual, 0.0000000000001));
}
public void SolveTest4()
{
double[,] value = // not positive-definite
{
{ 6, -1, 2, 6 },
{ -1, 3, -3, -2 },
{ 2, -3, 2, 0 },
{ 6, -2, 0, 0 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, true);
double[,] L = chol.LeftTriangularFactor;
double[] B = new double[] { 1, 2, 3, 4 };
double[] expected = { 5, 13, 16, -8 };
double[] actual = chol.Solve(B);
Assert.IsTrue(Matrix.IsEqual(expected, actual, 0.0000000000001));
}
/// <summary>Cholesky decomposition.</summary>
protected static double[,] chol(double[,] a)
{
var chol = new CholeskyDecomposition(a);
return chol.LeftTriangularFactor;
}
private void initialize(double[] m, double[,] cov, CholeskyDecomposition cd)
{
int k = m.Length;
this.mean = m;
this.covariance = cov;
this.chol = cd;
this.invCovariance = covariance.Inverse();
// Original code:
// double det = chol.Determinant;
// double detSqrt = System.Math.Sqrt(System.Math.Abs(det));
// constant = 1.0 / (System.Math.Pow(2.0 * System.Math.PI, k / 2.0) * detSqrt);
// Transforming to log operations, we have:
double lndet = cd.LogDeterminant;
// Let lndet = log( abs(det) )
//
// detSqrt = sqrt(abs(det)) = sqrt(abs(exp(log(det)))
// = sqrt(exp(log(abs(det))) = sqrt(exp(lndet)
//
// log(detSqrt) = log(sqrt(exp(lndet)) = (1/2)*log(exp(lndet))
// = lndet/2.
//
//
// Let lndetsqrt = log(detsqrt) = lndet/2
//
// constant = 1 / ( ((2PI)^(k/2)) * detSqrt)
//
// log(constant) = log( 1 / ( ((2PI)^(k/2)) * detSqrt) )
// = log(1)-log(((2PI)^(k/2)) * detSqrt) )
// = -log( ((2PI)^(k/2)) * detSqrt) )
// = -log( ((2PI)^(k/2)) * exp(log(detSqrt))))
// = -log( ((2PI)^(k/2)) * exp(lndetsqrt)))
// = -log( ((2PI)^(k/2)) * exp(lndet/2)))
// = -log( ((2PI)^(k/2))) - log(exp(lndet/2)))
// = -log( ((2PI)^(k/2))) - lndet/2)
// = -log( 2PI) * (k/2) - lndet/2)
// =(-log( 2PI) * k - lndet ) / 2
// =-(log( 2PI) * k + lndet ) / 2
// =-(log( 2PI) * k + lndet ) / 2
// =-( LN2PI * k + lndet ) / 2
//
// So the log(constant) could be computed as:
lnconstant = -(Special.Log2PI * k + lndet) * 0.5;
}
/// <summary>
/// Constructs a multivariate Gaussian distribution
/// with given mean vector and covariance matrix.
/// </summary>
///
/// <param name="mean">The mean of the distribution.</param>
/// <param name="covariance">The covariance for the distribution.</param>
///
public MultivariateNormalDistribution(double[] mean, double[,] covariance)
: base(mean.Length)
{
int rows = covariance.GetLength(0);
int cols = covariance.GetLength(1);
if (rows != cols)
throw new DimensionMismatchException("covariance", "Covariance matrix should be square.");
if (mean.Length != rows)
throw new DimensionMismatchException("covariance", "Covariance matrix should have the same dimensions as mean vector's length.");
CholeskyDecomposition chol = new CholeskyDecomposition(covariance, false, true);
if (!chol.PositiveDefinite)
throw new NonPositiveDefiniteMatrixException("Covariance matrix is not positive definite.");
initialize(mean, covariance, chol);
}
/// <summary>
/// Gets the probability density function (pdf) for
/// this distribution evaluated at point <c>x</c>.
/// </summary>
///
/// <param name="x">A single point in the distribution range.
/// For a matrix distribution, such as the Wishart's, this
/// should be a positive-definite matrix or a matrix written
/// in flat vector form.
/// </param>
///
/// <returns>
/// The probability of <c>x</c> occurring
/// in the current distribution.
/// </returns>
///
/// <remarks>
/// The Probability Density Function (PDF) describes the
/// probability that a given value <c>x</c> will occur.
/// </remarks>
///
public double ProbabilityDensityFunction(double[,] x)
{
var chol = new CholeskyDecomposition(x);
double det = chol.Determinant;
double[,] Vx = chol.Solve(inverseScaleMatrix);
double z = -0.5 * Vx.Trace();
double a = Math.Pow(det, power);
double b = Math.Exp(z);
return constant * a * b;
}
public void InverseTest1()
{
double[,] value = // positive-definite
{
{ 2, -1, 0 },
{ -1, 2, -1 },
{ 0, -1, 2 }
};
var chol = new CholeskyDecomposition(value, robust: false);
Assert.IsTrue(chol.IsPositiveDefinite);
var L = chol.LeftTriangularFactor;
float[][] expected =
{
new float[] { 0.750f, 0.500f, 0.250f },
new float[] { 0.500f, 1.000f, 0.500f },
new float[] { 0.250f, 0.500f, 0.750f },
};
double[,] actual = chol.Inverse();
Assert.IsTrue(actual.IsEqual(expected, 1e-6));
double[,] inv = chol.Solve(Matrix.Identity(3));
Assert.IsTrue(inv.IsEqual(expected, 1e-6));
}
public void CholeskyDecompositionConstructorTest3()
{
double[,] value = // not positive-definite
{
{ 6, -1, 2, 6 },
{ -1, 3, -3, -2 },
{ 2, -3, 2, 0 },
{ 6, -2, 0, 0 },
};
double[,] expected =
{
{ 1.0000, 0, 0, 0 },
{ -0.1667, 1.0000, 0, 0 },
{ 0.3333, -0.9412, 1.0000, 0 },
{ 1.0000, -0.3529, 2.5000, 1.0000 },
};
double[,] diagonal =
{
{ 6.0000, 0, 0, 0 },
{ 0, 2.8333, 0, 0 },
{ 0, 0, -1.1765, 0 },
{ 0, 0, 0, 1.0000 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, true);
double[,] L = chol.LeftTriangularFactor;
double[,] D = chol.DiagonalMatrix;
Assert.IsTrue(Matrix.IsEqual(L, expected, 0.001));
Assert.IsTrue(Matrix.IsEqual(D, diagonal, 0.001));
// Decomposition Identity
Assert.IsTrue(Matrix.IsEqual(L.Multiply(D).Multiply(L.Transpose()), value, 0.001));
Assert.AreEqual(new LuDecomposition(value).Determinant, chol.Determinant, 1e-10);
Assert.AreEqual(false, chol.PositiveDefinite);
Assert.AreEqual(true, chol.Symmetric);
}
public void SolveTest()
{
double[,] value = // positive-definite
{
{ 2, -1, 0 },
{ -1, 2, -1 },
{ 0, -1, 2 }
};
CholeskyDecomposition chol = new CholeskyDecomposition(value);
double[,] L = chol.LeftTriangularFactor;
double[,] B = Matrix.ColumnVector(new double[] { 1, 2, 3 });
double[,] expected = Matrix.ColumnVector(new double[] { 2.5, 4.0, 3.5 });
double[,] actual = chol.Solve(B);
Assert.IsTrue(Matrix.IsEqual(expected, actual, 1e-10));
Assert.AreNotEqual(actual, B);
actual = chol.Solve(B, true);
Assert.AreEqual(actual, B);
Assert.IsTrue(Matrix.IsEqual(expected, B, 1e-10));
Assert.IsTrue(Matrix.IsEqual(chol.Reverse(), value, 1e-6));
}
public void InverseTest()
{
double[,] value = // not positive-definite
{
{ 6, -1, 2, 6 },
{ -1, 3, -3, -2 },
{ 2, -3, 2, 0 },
{ 6, -2, 0, 0 },
};
double[,] expected =
{
{ 0.40000, 1.20000, 1.40000, -0.50000 },
{ 1.20000, 3.60000, 4.20000, -2.00000 },
{ 1.40000, 4.20000, 5.40000, -2.50000 },
{ -0.50000, -2.00000, -2.50000, 1.00000 },
};
var chol = new CholeskyDecomposition(value, robust: true);
double[,] L = chol.LeftTriangularFactor;
Assert.IsFalse(chol.IsPositiveDefinite);
double[,] actual = chol.Inverse();
Assert.IsTrue(Matrix.IsEqual(expected, actual, 1e-6));
double[,] inv = Matrix.Inverse(value);
Assert.IsTrue(Matrix.IsEqual(expected, inv, 1e-10));
Assert.IsTrue(Matrix.IsEqual(chol.Reverse(), value, 1e-6));
}
/// <summary>
/// Creates a new object that is a copy of the current instance.
/// </summary>
/// <returns>
/// A new object that is a copy of this instance.
/// </returns>
///
public object Clone()
{
var clone = new CholeskyDecomposition();
clone.L = (double[,])L.Clone();
clone.D = (double[])D.Clone();
clone.n = n;
clone.robust = robust;
clone.positiveDefinite = positiveDefinite;
clone.symmetric = symmetric;
return clone;
}
/// <summary>
/// Creates a new Cholesky decomposition directly from
/// an already computed left triangular matrix <c>L</c>.
/// </summary>
/// <param name="leftTriangular">The left triangular matrix from a Cholesky decomposition.</param>
///
public static CholeskyDecomposition FromLeftTriangularMatrix(double[,] leftTriangular)
{
CholeskyDecomposition chol = new CholeskyDecomposition();
chol.n = leftTriangular.GetLength(0);
chol.L = leftTriangular;
chol.symmetric = true;
chol.positiveDefinite = true;
chol.robust = false;
chol.D = Matrix.Vector(chol.n, 1.0);
return chol;
}
static void Main(string[] args)
{
Console.WriteLine("Please take a look at the source for examples!");
Console.ReadKey();
#region 1. Declaring matrices
// 1.1 Using standard .NET declaration
double[,] A =
{
{1, 2, 3},
{6, 2, 0},
{0, 0, 1}
};
double[,] B =
{
{2, 0, 0},
{0, 2, 0},
{0, 0, 2}
};
{
// 1.2 Using Accord extension methods
double[,] Bi = Matrix.Identity(3).Multiply(2);
double[,] Bj = Matrix.Diagonal(3, 2.0); // both are equal to B
// 1.2 Using Accord extension methods with implicit typing
var I = Matrix.Identity(3);
}
#endregion
#region 2. Matrix Operations
{
// 2.1 Addition
var C = A.Add(B);
// 2.2 Subtraction
var D = A.Subtract(B);
// 2.3 Multiplication
{
// 2.3.1 By a scalar
var halfM = A.Multiply(0.5);
// 2.3.2 By a vector
double[] m = A.Multiply(new double[] { 1, 2, 3 });
// 2.3.3 By a matrix
var M = A.Multiply(B);
// 2.4 Transposing
var At = A.Transpose();
}
}
// 2.5 Elementwise operations
// 2.5.1 Elementwise multiplication
A.ElementwiseMultiply(B); // A.*B
// 2.5.1 Elementwise division
A.ElementwiseDivide(B); // A./B
#endregion
#region 3. Matrix characteristics
{
// 3.1 Calculating the determinant
double det = A.Determinant();
// 3.2 Calculating the trace
double tr = A.Trace();
// 3.3 Computing the sum vector
{
double[] sumVector = A.Sum();
// 3.3.1 Computing the total sum of elements
double sum = sumVector.Sum();
// 3.3.2 Computing the sum along the rows
sumVector = A.Sum(0); // Equivalent to Octave's sum(A, 1)
// 3.3.2 Computing the sum along the columns
sumVector = A.Sum(1); // Equivalent to Octave's sum(A, 2)
}
}
#endregion
#region 4. Linear Algebra
{
// 4.1 Computing the inverse
var invA = A.Inverse();
// 4.2 Computing the pseudo-inverse
var pinvA = A.PseudoInverse();
// 4.3 Solving a linear system (Ax = B)
var x = A.Solve(B);
}
#endregion
#region 5. Special operators
{
// 5.1 Finding the indices of elements
double[] v = { 5, 2, 2, 7, 1, 0 };
int[] idx = v.Find(e => e > 2); // finding the index of every element in v higher than 2.
// 5.2 Selecting elements by index
double[] u = v.Submatrix(idx); // u is { 5, 7 }
// 5.3 Converting between different matrix representations
double[][] jaggedA = A.ToArray(); // from multidimensional to jagged array
// 5.4 Extracting a column or row from the matrix
double[] a = A.GetColumn(0); // retrieves the first column
double[] b = B.GetRow(1); // retrieves the second row
// 5.5 Taking the absolute of a matrix
var absA = A.Abs();
// 5.6 Applying some function to every element
var newv = v.Apply(e => e + 1);
}
#endregion
#region 7. Vector operations
{
double[] u = { 1, 2, 3 };
double[] v = { 4, 5, 6 };
var w1 = u.InnerProduct(v);
var w2 = u.OuterProduct(v);
var w3 = u.CartesianProduct(v);
double[] m = { 1, 2, 3, 4 };
double[,] M = Matrix.Reshape(m, 2, 2);
}
#endregion
#region Decompositions
{
// Singular value decomposition
{
SingularValueDecomposition svd = new SingularValueDecomposition(A);
var U = svd.LeftSingularVectors;
var S = svd.Diagonal;
var V = svd.RightSingularVectors;
}
// or (please see documentation for details)
{
SingularValueDecomposition svd = new SingularValueDecomposition(A.Transpose());
var U = svd.RightSingularVectors;
var S = svd.Diagonal;
var V = svd.LeftSingularVectors;
}
// Eigenvalue decomposition
{
EigenvalueDecomposition eig = new EigenvalueDecomposition(A);
var V = eig.Eigenvectors;
var D = eig.DiagonalMatrix;
}
// QR decomposition
{
QrDecomposition qr = new QrDecomposition(A);
var Q = qr.OrthogonalFactor;
var R = qr.UpperTriangularFactor;
}
// Cholesky decomposition
{
CholeskyDecomposition chol = new CholeskyDecomposition(A);
var R = chol.LeftTriangularFactor;
}
// LU decomposition
{
LuDecomposition lu = new LuDecomposition(A);
var L = lu.LowerTriangularFactor;
var U = lu.UpperTriangularFactor;
}
}
#endregion
}
public void InverseTest()
{
double[,] value = // not positive-definite
{
{ 6, -1, 2, 6 },
{ -1, 3, -3, -2 },
{ 2, -3, 2, 0 },
{ 6, -2, 0, 0 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, true);
double[,] L = chol.LeftTriangularFactor;
double[,] expected = Matrix.Inverse(value);
double[,] actual = chol.Inverse();
Assert.IsTrue(Matrix.IsEqual(expected, actual, 1e-10));
}
public void LogDeterminantTest()
{
CholeskyDecomposition chol = new CholeskyDecomposition(bigmatrix);
Assert.AreEqual(0.0, chol.Determinant);
Assert.AreEqual(-2224.8931093738875, chol.LogDeterminant);
Assert.IsTrue(chol.PositiveDefinite);
Assert.IsTrue(chol.Symmetric);
Assert.IsTrue(chol.Nonsingular);
}
public void CholeskyDecompositionConstructorTest2()
{
double[,] value = // positive-definite
{
{ 2, -1, 0 },
{ -1, 2, -1 },
{ 0, -1, 2 }
};
double[,] expected =
{
{ 1.0000, 0.0000, 0.0000 },
{ -0.5000, 1.0000, 0.0000 },
{ 0.0000, -0.6667, 1.0000 }
};
double[,] diagonal =
{
{ 2.0000, 0.0000, 0.0000 },
{ 0.0000, 1.5000, 0.0000 },
{ 0.0000, 0.0000, 1.3333 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, true);
double[,] L = chol.LeftTriangularFactor;
double[,] D = chol.DiagonalMatrix;
Assert.IsTrue(Matrix.IsEqual(L, expected, 0.001));
Assert.IsTrue(Matrix.IsEqual(D, diagonal, 0.001));
// Decomposition Identity
Assert.IsTrue(Matrix.IsEqual(L.Multiply(D).Multiply(L.Transpose()), value, 0.001));
Assert.AreEqual(new LuDecomposition(value).Determinant, chol.Determinant, 1e-10);
Assert.AreEqual(true, chol.PositiveDefinite);
Assert.AreEqual(true, chol.Symmetric);
}
public void LogDeterminantTest2()
{
double[,] value =
{
{ 6, -1, 2, 1 },
{ -1, 9, -3, -2 },
{ 2, -3, 8, 1 },
{ 1, -2, 1, 7 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value);
Assert.AreEqual(2232, chol.Determinant, 1e-12);
double expected = System.Math.Log(2232);
double actual = chol.LogDeterminant;
Assert.AreEqual(expected, actual, 1e-10);
}
public void CholeskyDecompositionConstructorTest4()
{
// Based on tests by Ken Jhonson
double[,] value = // positive-definite
{
{ 2, 0, 0 },
{ -1, 2, 0 },
{ 0, -1, 2 }
};
double[,] expected =
{
{ 1.4142, 0.0000, 0.0000 },
{ -0.7071, 1.2247, 0.0000 },
{ 0.0000, -0.8165, 1.1547 }
};
CholeskyDecomposition chol = new CholeskyDecomposition(value, false, true);
double[,] L = chol.LeftTriangularFactor;
Assert.IsTrue(Matrix.IsEqual(L, expected, 0.0001));
Assert.AreEqual(4, chol.Determinant, 1e-10);
Assert.AreEqual(true, chol.PositiveDefinite);
Assert.AreEqual(true, chol.Symmetric);
double[,] expected2 =
{
{ 1.0000, 0.0000, 0.0000 },
{ -0.5000, 1.0000, 0.0000 },
{ 0.0000, -0.6667, 1.0000 }
};
chol = new CholeskyDecomposition(value, true, true);
L = chol.LeftTriangularFactor;
Assert.IsTrue(Matrix.IsEqual(L, expected2, 0.0001));
Assert.AreEqual(4, chol.Determinant, 1e-10);
Assert.AreEqual(true, chol.PositiveDefinite);
Assert.AreEqual(true, chol.Symmetric);
}
/// <summary>
/// Gets the probability density function (pdf) for
/// this distribution evaluated at point <c>x</c>.
/// </summary>
///
/// <param name="x">A single point in the distribution range.</param>
///
/// <returns>
/// The probability of <c>x</c> occurring
/// in the current distribution.
/// </returns>
///
/// <remarks>
/// The Probability Density Function (PDF) describes the
/// probability that a given value <c>x</c> will occur.
/// </remarks>
///
public override double ProbabilityDensityFunction(double[] x)
{
// http://www.buch-kromann.dk/tine/nonpar/Nonparametric_Density_Estimation_multidim.pdf
// http://sfb649.wiwi.hu-berlin.de/fedc_homepage/xplore/ebooks/html/spm/spmhtmlnode18.html
double sum = 0;
CholeskyDecomposition chol = new CholeskyDecomposition(smoothing);
double[] delta = new double[Dimension];
for (int i = 0; i < samples.Length; i++)
{
for (int j = 0; j < x.Length; j++)
delta[j] = (x[j] - samples[i][j]);
sum += kernel.Function(chol.Solve(delta));
}
return sum / (samples.Length * determinant);
}
public void DeterminantTest()
{
double[,] value =
{
{ 6, -1, 2, 1 },
{ -1, 9, -3, -2 },
{ 2, -3, 8, 1 },
{ 1, -2, 1, 7 },
};
CholeskyDecomposition chol = new CholeskyDecomposition(value);
Assert.AreEqual(2232, chol.Determinant, 1e-12);
Assert.IsTrue(chol.Nonsingular);
Assert.IsTrue(chol.Symmetric);
}
public void CholeskyDecompositionConstructorTest4()
{
// Based on tests by Ken Johnson
double[,] value = // positive-definite
{
{ 2, 0, 0 },
{ -1, 2, 0 },
{ 0, -1, 2 }
};
double[,] expected =
{
{ 1.4142, 0.0000, 0.0000 },
{ -0.7071, 1.2247, 0.0000 },
{ 0.0000, -0.8165, 1.1547 }
};
var chol = new CholeskyDecomposition(value, false, valueType: MatrixType.LowerTriangular);
double[,] L = chol.LeftTriangularFactor;
Assert.IsTrue(Matrix.IsEqual(L, expected, 1e-4));
Assert.AreEqual(4, chol.Determinant, 1e-10);
Assert.IsTrue(chol.IsPositiveDefinite);
Assert.IsTrue(Matrix.IsEqual(chol.Reverse(), value.GetSymmetric(type: MatrixType.LowerTriangular), 1e-4));
double[,] expected2 =
{
{ 1.0000, 0.0000, 0.0000 },
{ -0.5000, 1.0000, 0.0000 },
{ 0.0000, -0.6667, 1.0000 }
};
chol = new CholeskyDecomposition(value, robust: true, valueType: MatrixType.LowerTriangular);
L = chol.LeftTriangularFactor;
Assert.IsTrue(Matrix.IsEqual(L, expected2, 1e-4));
Assert.IsTrue(Matrix.IsEqual(chol.Reverse(), value.GetSymmetric(type: MatrixType.LowerTriangular), 1e-6));
Assert.IsTrue(chol.IsPositiveDefinite);
Assert.AreEqual(4, chol.Determinant, 1e-10);
}
/// <summary>
/// Creates a new Inverse Wishart distribution.
/// </summary>
///
/// <param name="degreesOfFreedom">The degrees of freedom v.</param>
/// <param name="inverseScale">The inverse scale matrix Ψ (psi).</param>
///
public InverseWishartDistribution(double degreesOfFreedom, double[,] inverseScale)
: base(inverseScale.Length)
{
if (inverseScale.GetLength(0) != inverseScale.GetLength(1))
throw new DimensionMismatchException("inverseScale", "Matrix must be square.");
this.inverseScaleMatrix = inverseScale;
this.size = inverseScale.GetLength(0);
this.v = degreesOfFreedom;
if (degreesOfFreedom <= size - 1)
throw new ArgumentOutOfRangeException("degreesOfFreedom", "Degrees of freedom must be greater "
+ "than or equal to the number of rows in the inverse scale matrix.");
var chol = new CholeskyDecomposition(inverseScale);
if (!chol.PositiveDefinite)
throw new NonPositiveDefiniteMatrixException("scale");
if (!chol.Symmetric)
throw new NonSymmetricMatrixException("scale");
double a = Math.Pow(chol.Determinant, v / 2.0);
double b = Math.Pow(2, (v * size) / 2.0);
double c = Gamma.Multivariate(v / 2.0, size);
this.constant = a / (b * c);
this.power = -(v + size + 1) / 2.0;
}
public void LogDeterminantTest()
{
var chol = new CholeskyDecomposition(bigmatrix);
Assert.AreEqual(0.0, chol.Determinant);
Assert.AreEqual(-2224.8931093738875, chol.LogDeterminant, 1e-10);
Assert.IsTrue(chol.IsPositiveDefinite);
Assert.IsTrue(chol.Nonsingular);
Assert.IsTrue(Matrix.IsEqual(chol.Reverse(), bigmatrix, 1e-6));
}
/// <summary>
/// Gets the probability density function (pdf) for
/// this distribution evaluated at point <c>x</c>.
/// </summary>
///
/// <param name="x">A single point in the distribution range.
/// For a matrix distribution, such as the Wishart's, this
/// should be a positive-definite matrix or a matrix written
/// in flat vector form.
/// </param>
///
/// <returns>
/// The probability of <c>x</c> occurring
/// in the current distribution.
/// </returns>
///
/// <remarks>
/// The Probability Density Function (PDF) describes the
/// probability that a given value <c>x</c> will occur.
/// </remarks>
///
public double LogProbabilityDensityFunction(double[,] x)
{
var chol = new CholeskyDecomposition(x);
double det = chol.Determinant;
double[,] Vx = chol.Solve(inverseScaleMatrix);
double z = -0.5 * Vx.Trace();
return Math.Log(constant) + power * Math.Log(det) + z;
}
public void DeterminantTest2()
{
double[,] value =
{
{ 6, -1, 2, 1 },
{ 0, 9, -3, -2 },
{ 0, 0, 8, 1 },
{ 0, 0, 0, 7 },
};
var chol = new CholeskyDecomposition(value);
//var L = chol.LeftTriangularFactor;
//var det = L.Determinant();
Assert.IsTrue(chol.IsPositiveDefinite);
Assert.AreEqual(2232, chol.Determinant, 1e-12);
Assert.IsTrue(chol.Nonsingular);
//Assert.IsTrue(chol.Symmetric);
}
/// <summary>
/// Creates a new Wishart distribution.
/// </summary>
///
/// <param name="degreesOfFreedom">The degrees of freedom n.</param>
/// <param name="scale">The positive-definite matrix scale matrix V.</param>
///
public WishartDistribution(double degreesOfFreedom, double[,] scale)
: base(scale.Length)
{
if (scale.GetLength(0) != scale.GetLength(1))
throw new DimensionMismatchException("scale", "Matrix must be square.");
this.scaleMatrix = scale;
this.n = degreesOfFreedom;
this.size = scale.GetLength(0);
if (degreesOfFreedom <= size - 1)
throw new ArgumentOutOfRangeException("degreesOfFreedom", "Degrees of freedom must be greater "
+ "than or equal to the number of rows in the scale matrix.");
this.chol = new CholeskyDecomposition(scale);
if (!chol.PositiveDefinite)
throw new NonPositiveDefiniteMatrixException("scale");
if (!chol.Symmetric)
throw new NonSymmetricMatrixException("scale");
double a = Math.Pow(chol.Determinant, n / 2.0);
double b = Math.Pow(2, (n * size) / 2.0);
double c = Gamma.Multivariate(n / 2.0, size);
this.constant = 1.0 / (a * b * c);
this.lnconstant = Math.Log(constant);
this.power = (n - size - 1) / 2.0;
}
public void SolveTest3()
{
double[,] value = // positive-definite
{
{ 2, -1, 0 },
{ -1, 2, -1 },
{ 0, -1, 2 }
};
CholeskyDecomposition chol = new CholeskyDecomposition(value);
double[,] L = chol.LeftTriangularFactor;
double[] B = new double[] { 1, 2, 3 };
double[] expected = new double[] { 2.5, 4.0, 3.5 };
double[] actual = chol.Solve(B);
Assert.IsTrue(Matrix.IsEqual(expected, actual, 0.0000000000001));
actual = chol.Solve(B, true);
Assert.AreEqual(actual, B);
Assert.IsTrue(Matrix.IsEqual(expected, B, 0.0000000000001));
}
