Function conjugate_gradient

pub fn conjugate_gradient(
    a: &SparseMatrix,
    b: &[f64],
    initial: Option<&[f64]>,
    options: ConjugateGradientOptions,
) -> NumericResult<ConjugateGradientSolution>

Solve A x = b for a symmetric positive definite sparse A by the Jacobi-preconditioned conjugate-gradient method, starting from initial (zero when None).

A run that exhausts its budget still returns its last iterate and residual with Status::BudgetExhausted; call ConjugateGradientSolution::converged to refuse it. A zero b returns the exact zero solution without iterating.

Errors

• NumericError::DimensionMismatch for a non-square matrix or a wrongly sized b or initial.
• NumericError::NonFiniteInput for a non-finite b or initial, NumericError::InvalidArgument for a negative or non-finite tolerance.
• NumericError::NotSymmetric for an asymmetric matrix.
• NumericError::NotPositiveDefinite when a diagonal entry is not positive (pivot is its row) or when the iteration meets a direction p with p^T A p <= 0, which proves A is not positive definite (pivot is the iteration).