Struct ConvergedState 

pub struct ConvergedState {
    pub status: ApplicationReturnStatus,
    pub x: Vec<Number>,
    pub obj_val: Number,
    pub bound_relax_factor: Number,
    pub exact_hessian: bool,
    /* private fields */
}

State captured at convergence: the user-visible iterate plus the PdSensBacksolver that wraps the converged KKT factor.

Read this via Solver::converged.

Fields

status: ApplicationReturnStatus

IPM return status of the most recent solve.

x: Vec<Number>

Final primal iterate x* (length n_x), in the user’s own units: a user-scaling change of variables is undone here, so this is x, never the algorithm’s x̃ = d ⊙ x (gh#486).

obj_val: Number

Final objective value f(x*).

bound_relax_factor: Number

bound_relax_factor as the solve that produced this state ran with it, not as the application’s options read today. The bounds were relaxed (or not) once, during this solve; a later set_numeric_value cannot change what the held slacks were measured against, so post-solve calls whose validity depends on unrelaxed bounds must guard on this value. See Solver::classify_activity.

exact_hessian: bool

Whether the solve computed exact Hessians, as it ran. A limited-memory solve’s IpoptData::w is the quasi-Newton matrix, and there is no exact Hessian to evaluate at another point, so the corrector keeps that matrix instead of refreshing it at the predicted iterate.

Implementations

impl ConvergedState

pub fn block_dims(&self) -> [usize; 8]

Block dimensions of the compound KKT vector in (x, s, y_c, y_d, z_l, z_u, v_l, v_u) order.

pub fn kkt_dim(&self) -> usize

Total dimension of the compound KKT vector (sum of block_dims).