Module extendr_api::prelude::sparse::linalg::qr::supernodal

Supernodal factorization module.

A supernodal factorization is one that processes the elements of the QR factors of the input matrix by blocks, rather than by single elements. This is more efficient if the QR factors are somewhat dense.

Structs

• SupernodalQrRef
  QR factors containing both the symbolic and numeric representations.
• SymbolicSupernodalHouseholder
  Symbolic structure of the Householder reflections that compose $Q$,
• SymbolicSupernodalQr
  Symbolic structure of the QR decomposition,

Functions

• factorize_supernodal_numeric_qr
  Computes the numerical QR factorization of $A$.
• factorize_supernodal_numeric_qr_req
  Computes the size and alignment of the workspace required to compute the numerical QR factorization of the matrix whose structure was used to produce the symbolic structure.
• factorize_supernodal_symbolic_qr
  Computes the symbolic QR factorization of a matrix $A$, given a fill-reducing column permutation, and the outputs of the pre-factorization steps.
• factorize_supernodal_symbolic_qr_req
  Computes the size and alignment of the workspace required to compute the symbolic QR factorization of a matrix with dimensions (nrows, ncols).