RSpectra-compatible SVD shim.

Usage

svds(
  A,
  k,
  nu = k,
  nv = k,
  opts = list(),
  ...,
  Atrans = NULL,
  dim = NULL,
  args = NULL
)

Arguments

A: Matrix, eigencore operator, or a function f(x, args) returning A %*% x (then Atrans and dim are required).
k: Number of singular values to compute.
nu: Number of left singular vectors returned.
nv: Number of right singular vectors returned.
opts: RSpectra options list. tol, center and scale are honoured (center/scale are applied as operators, without densifying); ncv is passed as auto(max_subspace = ncv) (used by the Golub-Kahan and implicit-Gram Lanczos routes; the explicit Gram route has no Krylov subspace). maxitr is accepted but not used, since svd_partial() has no iteration limit yet; other keys raise a warning.
...: Additional arguments passed to svd_partial().
Atrans: Function f(x, args) returning t(A) %*% x when A is a function.
dim: Dimensions of A when A is a function.
args: Extra argument passed to function inputs.

Value

A list compatible with RSpectra::svds(), including d, optional u and v, convergence counts, operation counts, certificate diagnostics, and eigencore diagnostics.

Examples

set.seed(1)
X <- matrix(rnorm(60), 10, 6)
res <- svds(X, k = 2)
res$d
#> [1] 4.728358 3.042304