Spectral Graph Theory Lab

Change a graph continuously and watch its Laplacian spectrum respond. Connectivity, cuts, clusters and spanning trees can all be read from the eigenvalues; each classic theorem is checked live.

Space play the morph · hover a node or a point to link the views

Graph

1.00

Spectrum and colouring

Matrix
Colour nodes by
Edge width shows Wij. Dashed edges are cut by the current partition.

Eigenvalue trajectories

Spectral embedding

node i ↦ (u₂(i), u₃(i))

Nodes that are close here are well connected in the graph.

Theorems, checked live

Notation and equations

Every symbol is defined in the first card

Sources: Fiedler (1973) algebraic connectivity; Kirchhoff (1847) matrix-tree theorem; Cheeger (1970), Alon & Milman (1985) and Chung (1997) for the graph Cheeger inequality; Shi & Malik (2000) normalised cuts; Ng, Jordan & Weiss (2002) and von Luxburg (2007) spectral clustering; Spielman, Spectral and Algebraic Graph Theory (2019).