Struct StructureBook 

pub struct StructureBook { /* private fields */ }

Structural rank analysis of the equality constraint Jacobian, after the Dulmage–Mendelsohn decomposition used by IDAES’s DiagnosticsToolbox. The Hessian-free, iterate-independent sparsity pattern alone tells us whether a subset of equations is over-determined — more equations than the variables they jointly touch — which forces at least one of them to be redundant or mutually inconsistent (a structurally singular Jacobian, LICQ failure).

The payoff is naming those rows. The solver’s δ_c dual regularization and wrong-inertia flags detect rank deficiency but report it as a scalar; this book maps the dependent rows back to the model’s equation names so diagnose can say mass_balance instead of “equation 13”. Tracing a singular system to named equations is exactly the roadblock Lee et al. (2024) identify for equation-oriented model debugging. See https://doi.org/10.69997/sct.147875.

Implementations

impl StructureBook

pub fn new( inc: EqualityIncidence, con_names: Vec<String>, var_names: Vec<String>, ) -> Self

Build from the equality incidence graph plus the model’s constraint and variable name vectors (original order). The incidence rows index into con_names via inc.eq_row_inner_idx; the incidence columns index var_names directly.