Struct SpaiPreconditioner 

pub struct SpaiPreconditioner<F> { /* private fields */ }

Sparse Approximate Inverse (SPAI) preconditioner

This preconditioner computes a sparse approximate inverse M ≈ A^(-1) using a minimization approach: minimize ||I - AM||_F subject to sparsity constraints on M.

Implementations

impl<F: Float + NumAssign + Sum + Debug + 'static> SpaiPreconditioner<F>

pub fn new(matrix: &CsrMatrix<F>, options: SpaiOptions) -> SparseResult<Self>

Create a new SPAI preconditioner from a sparse matrix

Trait Implementations

impl<F: Float + NumAssign + Sum + Debug + 'static> LinearOperator<F> for SpaiPreconditioner<F>

fn shape(&self) -> (usize, usize)

The shape of the operator (rows, columns)

fn matvec(&self, x: &[F]) -> SparseResult<Vec<F>>

Apply the operator to a vector: y = A * x

fn has_adjoint(&self) -> bool

Check if the operator supports adjoint operations

fn rmatvec(&self, x: &[F]) -> SparseResult<Vec<F>>

Apply the adjoint of the operator to a vector: y = A^H * x Default implementation returns an error

fn matmat(&self, x: &[Vec<F>]) -> SparseResult<Vec<Vec<F>>>

Apply the operator to a matrix: Y = A * X where X is column-major (each column is a vector)

fn rmatmat(&self, x: &[Vec<F>]) -> SparseResult<Vec<Vec<F>>>

Apply the adjoint of the operator to a matrix: Y = A^H * X Default implementation calls rmatvec for each column