Struct ConjugateGradientSolver 

pub struct ConjugateGradientSolver { /* private fields */ }

Conjugate Gradient solver for symmetric positive-definite sparse systems.

Stores the solver configuration (tolerance, iteration cap, preconditioning). The solve itself is stateless and may be invoked concurrently on different inputs from multiple threads.

Implementations

impl ConjugateGradientSolver

pub fn new( tolerance: f64, max_iterations: usize, use_preconditioner: bool, ) -> Self

Create a new CG solver.

Arguments

• tolerance – Relative residual threshold for convergence. Must be positive and finite.
• max_iterations – Upper bound on CG iterations. Must be >= 1.
• use_preconditioner – Enable diagonal (Jacobi) preconditioning.

pub fn tolerance(&self) -> f64

Return the configured tolerance.

pub fn max_iterations(&self) -> usize

Return the configured maximum iterations.

pub fn use_preconditioner(&self) -> bool

Return whether preconditioning is enabled.

Trait Implementations

impl Clone for ConjugateGradientSolver

fn clone(&self) -> ConjugateGradientSolver

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ConjugateGradientSolver

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl SolverEngine for ConjugateGradientSolver

fn solve( &self, matrix: &CsrMatrix<f64>, rhs: &[f64], budget: &ComputeBudget, ) -> Result<SolverResult, SolverError>

Solve Ax = b using the Conjugate Gradient method.

Errors

• SolverError::InvalidInput – dimension mismatch or invalid params.
• SolverError::NumericalInstability – divergence or non-SPD matrix.
• SolverError::NonConvergence – iteration limit exceeded.
• SolverError::BudgetExhausted – wall-time limit exceeded.

fn estimate_complexity( &self, profile: &SparsityProfile, n: usize, ) -> ComplexityEstimate

Estimate CG complexity from the sparsity profile.

CG converges in O(sqrt(kappa)) iterations, each costing O(nnz) for the SpMV plus O(n) for the vector updates.

fn algorithm(&self) -> Algorithm

Return the algorithm identifier.