Skip to contents

Compute a Laplacian spectral basis (eigenpairs) from an adjacency matrix

Usage

compute_laplacian_basis(
  A,
  k,
  normalized = TRUE,
  which = "SA",
  eig_tol = 1e-12,
  extra = 4L,
  seed = 1L,
  tol = 1e-06,
  maxitr = 1000
)

Arguments

A

Square adjacency/affinity matrix (dense or sparse).

k

Integer number of non-trivial eigenpairs to return.

normalized

Logical; use the normalized Laplacian (default TRUE).

which

Which eigenpairs to request (passed to the eigensolver). Default is `"SA"` (smallest algebraic) for Laplacian bases.

eig_tol

Positive threshold; eigenvalues `<= eig_tol` are treated as trivial and dropped.

extra

Integer number of extra eigenpairs requested to survive multiple near-zero modes on disconnected graphs.

seed

Integer seed used to build a deterministic initial vector for RSpectra (helps reproducibility).

tol

Numeric tolerance passed to RSpectra/PRIMME when available.

maxitr

Maximum iterations passed to RSpectra when available.

Value

A list with fields `values` (length k) and `vectors` (n x k).