Struct RedHessianResult 

pub struct RedHessianResult {
    pub var_indices: Vec<usize>,
    pub hr: Vec<Number>,
    pub eigenvalues: Option<Vec<Number>>,
    pub eigenvectors: Option<Vec<Number>>,
}

Outputs of try_compute_red_hessian: the column-major n × n reduced Hessian (hr), the variable indices var_indices that label its rows/cols (so a downstream JSON consumer can map back to AMPL var names), and the optional eigendecomposition.

Fields

var_indices: Vec<usize>

var-x indices (algorithm-side, length n) that label the rows/cols of hr, ordered by the 1..n slot from the AMPL red_hessian suffix. Fixed variables are skipped (they cannot participate in the reduced Hessian).

hr: Vec<Number>

Column-major n × n reduced Hessian.

eigenvalues: Option<Vec<Number>>

Optional ascending eigenvalues (length n).

eigenvectors: Option<Vec<Number>>

Optional column-major eigenvectors (length n²).