Struct StatisticsInfo 

pub struct StatisticsInfo {

    pub iteration_count: i32,
    pub final_objective: f64,
    pub final_scaled_objective: f64,
    pub final_dual_inf: f64,
    pub final_constr_viol: f64,
    pub final_declared_constr_viol: f64,
    pub final_declared_box_viol: f64,
    pub final_compl: f64,
    pub final_kkt_error: f64,
    pub final_kkt_error_above_noise: f64,
    pub num_obj_evals: i32,
    pub num_constr_evals: i32,
    pub num_obj_grad_evals: i32,
    pub num_constr_jac_evals: i32,
    pub num_hess_evals: i32,
    pub total_wallclock_time_secs: f64,
    pub restoration_calls: i32,
    pub restoration_inner_iters: i32,
    pub restoration_outer_iters: i32,
    pub restoration_wall_secs: f64,
    pub quality_escalations: i32,
    pub dual_divergence_signature: bool,
    pub dual_divergence_retry_promoted: bool,
}

Subset of SolveStatistics projected for the report. Mirrors the fields the existing console summary prints.

Fields

iteration_count: i32

final_objective: f64

final_scaled_objective: f64

final_dual_inf: f64

final_constr_viol: f64

final_declared_constr_viol: f64

Primal violation against the model as declared, before the convex arm’s bound_relax_factor widening (qp_extract::BoundRelax, gh #744/#745).

final_constr_viol measures the model the solver was HANDED, whose inequality rows and variable box are widened by min(factor,cap)·|b|. That is the model its convergence test is about and every acceptance gate reads — and it is not how far the returned point sits outside the model the caller wrote. On netlib afiro the point is 4.99e-06 outside a declared row b = 500 (exactly 1e-8·500) while final_constr_viol reads 8.68e-13; 25fv47 reports 2.19e-11 against 1.97e-05.

NaN when the solve applied no widening (the two coincide by construction) or on a path that does not compute it — every NLP-arm solve today. Additive to pounce.solve-report/v1: readers predating it are unaffected.

final_declared_box_viol: f64

How far the returned point sits outside the declared variable box — the box the caller wrote, before the bound_relax_factor widening. Ipopt’s Variable bound violation, and the box half of final_declared_constr_viol reported on its own: maxed together, a box violation and a row violation cannot be told apart.

Variable bounds carry no scaling, so there is one number rather than a scaled/unscaled pair. NaN on a path that does not compute it. Additive to pounce.solve-report/v1: readers predating it are unaffected.

final_compl: f64

final_kkt_error: f64

final_kkt_error_above_noise: f64

The aggregate the strict convergence gate tested (gh #528): as final_kkt_error, but counting each constraint row’s residual only where it exceeds what that row can represent in floating point. Equal to final_kkt_error unless a row is at its own resolution limit.

num_obj_evals: i32

num_constr_evals: i32

num_obj_grad_evals: i32

num_constr_jac_evals: i32

num_hess_evals: i32

total_wallclock_time_secs: f64

restoration_calls: i32

restoration_inner_iters: i32

restoration_outer_iters: i32

restoration_wall_secs: f64

quality_escalations: i32

Successful linear-solver quality escalations over the whole solve, restoration sub-solves included (gh#857). 0 on paths that never escalate. serde(default) so a report written before this field existed still deserializes — it reads as “no escalations”, which is wrong-but-harmless for an old file and correct for every new one.

dual_divergence_signature: bool

The solve saw the gh#884 biactive dual-divergence signature: at one and the same iterate a converged primal, a scale-relative step at zero, and an unscaled dual infeasibility far above dual_inf_tol. Reported whether or not a retry ran or promoted, because the distinction it draws — the multipliers ran away on a settled iterate, as against the iterate itself never settling — is not visible in any other field. serde(default) for the same reason as quality_escalations.

dual_divergence_retry_promoted: bool

A gh#884 dual-divergence retry ran and its answer was returned. false both when no retry ran and when one ran and lost — in the latter case every other field here describes the base attempt.

Trait Implementations

impl Clone for StatisticsInfo

fn clone(&self) -> StatisticsInfo

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StatisticsInfo

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for StatisticsInfo

fn deserialize<__D>( __deserializer: __D, ) -> Result<StatisticsInfo, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for StatisticsInfo

fn serialize<__S>( &self, __serializer: __S, ) -> Result<<__S as Serializer>::Ok, <__S as Serializer>::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.