Struct AtomInferenceReport 

pub struct AtomInferenceReport {
    pub atom_index: usize,
    pub atom_name: String,
    pub functionals: Option<AtomFunctionalReport>,
    pub smooth_significance: Option<AtomSmoothSignificance>,
}

The post-PIRLS inference reports for one atom, paired by atom index.

Two reports survive #1115: the descriptive penalty-debiased point summaries of the fitted decoder curve (AtomFunctionalReport, no coverage claim) and the any-n-valid split-LRT smooth-structure e-value (AtomSmoothSignificance, a genuine finite-sample-valid test). The #1099 per-atom curvature confidence interval was removed: its target (a sup-norm extrinsic-curvature BOUND read off the fitted decoder) is not an estimand with a profiled criterion, and its delta-method SE conditioned on the generated latent coordinates as if known. The plug-in curvature point estimate itself survives — as the per-atom kappa_hat entries of crate::manifold::CertificateInputs::per_atom_kappa_hat (the #1008 empirical curved-dictionary report, surfaced to Python as ManifoldSAE.curvature_report), the single source of truth for the bound. It is deliberately not duplicated onto this report: a descriptive geometry bound is a property of the fitted decoder frames, not of the post-PIRLS inner-smooth inference snapshot this type carries.

Fields

atom_index: usize

atom_name: String

functionals: Option<AtomFunctionalReport>

smooth_significance: Option<AtomSmoothSignificance>

Trait Implementations

impl Clone for AtomInferenceReport

fn clone(&self) -> AtomInferenceReport

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for AtomInferenceReport

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

Formats the value using the given formatter.