Struct ClarabelPersistent 

pub struct ClarabelPersistent { /* private fields */ }

A stateful Clarabel handle that keeps the built solver resident across solves.

Created by Clarabel::persistent. When the next model has the same dimensions, cone layout and P/A sparsity pattern, only the numeric data (P, q, A, b) is overwritten in place via Clarabel’s update_data, reusing the KKT symbolic factorization instead of rebuilding it. Any structural change — added/removed rows or columns, a changed sparsity pattern, a new cone, a flipped constraint sense that moves a row between cones — triggers a transparent full rebuild, so every result matches a cold Clarabel::solve.

Clarabel is an interior-point method and does not warm-start from the previous iterate, so iteration counts are unchanged; the saving is the skipped setup (equilibration structure, symbolic factorization, allocations).

In-place updates are rejected by Clarabel when presolve or structural-zero dropping altered the problem, in which case the handle falls back to a rebuild. Set presolve_enable(false) and input_sparse_dropzeros(false) (the defaults leave presolve on) to keep the fast path available.

A failed solve leaves the handle without a resident model; the next call rebuilds from scratch.

Implementations

impl ClarabelPersistent

pub fn new() -> Self

A fresh handle with no model loaded. The first solve builds it.

pub fn reset(&mut self)

Drop the resident solver so the next solve rebuilds from scratch. After this, the next solve behaves exactly like a cold Clarabel::solve.

Trait Implementations

impl Debug for ClarabelPersistent

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

Formats the value using the given formatter.

impl Default for ClarabelPersistent

fn default() -> ClarabelPersistent

Returns the “default value” for a type.

impl Solver for ClarabelPersistent

type Options = ClarabelOptions

Backend-specific options struct. Use () for solvers without any tunables.

fn name(&self) -> &str

fn supports(&self, kind: ModelKind) -> bool

fn solve( &mut self, model: &Model, opts: &ClarabelOptions, ) -> Result<SolverResult, SolverError>

Solves the given Model using this solver.