Dataset generator

Samples (N): 150 Noise level (σ): 1.0 True expression programs (K): 4 Genes measured (M): 30

True group legend

Eigendecomposition (live)

Genes × Samples—

PC1 variance explained—

PC2 variance explained—

PC3 variance explained—

Cumulative top-3—

Jacobi sweeps to converge—

Cluster separation ratio—

Drag the 3D view to rotate, scroll to zoom. Each point is one simulated sample projected onto its real top-3 principal components, computed from a genuine Jacobi eigendecomposition of the mean-centered gene covariance matrix — recomputed live whenever you change the controls above. Colour = true generating group (not shown to the algorithm).