FastIca.MixingMatrix Property

Gets the estimated mixing matrix (M) with dimensions (nFeatures × nComponents).

Definition

Namespace: Numerics.NET.Statistics.Multivariate
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 9.1.5
C#
public Matrix<double> MixingMatrix { get; }

Property Value

Matrix<Double>

Remarks

The mixing matrix represents the linear transformation from independent sources to the observed mixed signals: X ≈ S·M + μ, where S are the sources (nSamples × nComponents), M is the mixing matrix, and μ are the feature means.

Each column of M represents how one independent component contributes to the observed features.

This property corresponds to sklearn's mixing_ attribute and satisfies the relationship: MixingMatrix ≈ pinv(UnmixingMatrix) (pseudoinverse), ensuring that UnmixingMatrix @ MixingMatrix ≈ I in component space.

Exceptions

InvalidOperationException Thrown if accessed before calling Fit().

See Also