approximate_double (Covariance Matrix)

Gaussian-to-Dirac approximation using full covariance matrix.

The covariance matrix is internally diagonalized. The optimized support points are automatically transformed back into the original coordinate system.

Parameters

• Sigma: Covariance matrix (N x N)
• L: Number of Dirac components
• N: Dimension
• x: Output buffer (L x N)

Example

Sigma = np.array([[2.0, 0.5],
                  [0.5, 2.0]])
 
g2d = dgs.GaussianToDiracApproximation()
result = g2d.approximate_double(Sigma, L, N, x)