Struct TSymLinearSolver 

pub struct TSymLinearSolver { /* private fields */ }

Driver wrapping a SparseSymLinearSolverInterface (and optionally a TSymScalingMethod).

Implementations

impl TSymLinearSolver

pub fn new( backend: Box<dyn SparseSymLinearSolverInterface>, scaling_method: Option<Box<dyn TSymScalingMethod>>, linear_scaling_on_demand: bool, ) -> Self

Build a driver around backend. Pass Some(scaling) to enable symmetric scaling. linear_scaling_on_demand=true matches upstream’s default and keeps scaling off until Self::increase_quality turns it on.

pub fn initialize_structure( &mut self, dim: Index, airn: &[Index], ajcn: &[Index], ) -> ESymSolverStatus

Pin the triplet sparsity pattern. Must be called once before the first Self::multi_solve. airn / ajcn are 1-based. Mirrors the bulk of TSymLinearSolver::InitializeStructure.

pub fn multi_solve( &mut self, vals: &[Number], new_matrix: bool, nrhs: Index, rhs_vals: &mut [Number], check_neg_evals: bool, number_of_neg_evals: Index, ) -> ESymSolverStatus

Solve A x = b (or multiple RHS).

vals is the new triplet-format value array (length nonzeros_triplet). new_matrix=true requests a refactor (and a fresh scaling-factor computation if scaling is on); new_matrix=false reuses the existing factor and just runs back-substitution.

rhs_vals packs nrhs columns, each length dim, in column-major layout. Solutions overwrite rhs_vals.

Trait Implementations

impl Debug for TSymLinearSolver

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

Formats the value using the given formatter.

impl SymLinearSolver for TSymLinearSolver

fn increase_quality(&mut self) -> bool

Mirrors upstream’s IncreaseQuality: switching scaling on at the wrapper level (linear_scaling_on_demand=true path) takes precedence over asking the backend for tighter pivoting.

fn number_of_neg_evals(&self) -> Index

Most recent factorization’s negative-eigenvalue count.

fn provides_inertia(&self) -> bool

Whether this solver reports inertia.