Struct SparseSolver 

pub struct SparseSolver { /* private fields */ }

The one linear solver Carapace ships — a concrete struct, not a trait or enum, since there is exactly one implementation to commit to.

faer’s sparse LU (with built-in COLAMD/AMD fill-reducing ordering), not a dense nalgebra solve. A dense O(n^3) LU doesn’t scale to real models. It builds and solves correctly under wasm32.

Implementations

impl SparseSolver

pub fn new() -> Self

pub fn factor( &self, k: &SparseColMat<usize, f64>, ) -> Result<SparseFactorization, AnalysisError>

Factor k, without solving — the half of a solve that TangentStrategy::ReuseAtStepStart/Initial (docs/algorithms.md §3.1) let a caller skip repeating across iterations. faer‘s Lu is a fully owned value (no lifetime tied back to k), so the result outlives the matrix it was formed from — see SparseFactorization. Symbolic analysis is reused while the full CSC pattern matches; values (including zeros) never participate in cache invalidation. Numeric factorization is always recomputed here, independently of callers’ numeric tangent-reuse caches.

pub fn solve( &self, k: &SparseColMat<usize, f64>, r: &DVector<f64>, ) -> Result<DVector<f64>, AnalysisError>

Factor-then-solve in one call — for Algorithm::Linear and anywhere else a single ad-hoc solve is enough. Equivalent to self.factor(k)? .solve(r), kept as its own method so a one-off caller doesn’t have to name SparseFactorization just to discard it immediately.

Trait Implementations

impl Default for SparseSolver

fn default() -> Self

Returns the “default value” for a type.