Struct EqpReport 

pub struct EqpReport {
    pub outcome: Result<QpStatus, QpError>,
    pub working_set_changes: u32,
    pub refactors: u32,
    pub schur_updates: u32,
    pub rank_guard_calls: u32,
    pub time: Duration,
    pub linsol_times: LinsolTimes,
}

What the §7 step-3 QP did (gh#1061).

CrossoverDecline::QpFailed covers every way that QP can fail to reach Optimal — an iteration cap, an infeasible verdict, a hard linear-solver error — and on a large model those call for different responses. This is the detail behind it, kept on accepted reports too, so a slow crossover can be attributed: working-set changes, refactorizations and rank tests against the wall time they cost.

Fields

outcome: Result<QpStatus, QpError>

The QP’s status, or the hard error it returned instead of one.

working_set_changes: u32

Working-set changes (adds + drops). 0 when the QP returned an error.

refactors: u32

Base-KKT refactorizations. 0 when the QP returned an error.

schur_updates: u32

Schur-complement updates between refactorizations. 0 when the QP returned an error.

rank_guard_calls: u32

Linear-independence rank tests, counted on errors too.

time: Duration

Wall time of the QP call.

linsol_times: LinsolTimes

How that wall time split between factorization, back-solves and rank tests; the rest of time went outside the linear solver (gh#1061).

Trait Implementations

impl Clone for EqpReport

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EqpReport

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

Formats the value using the given formatter.