Struct CoactivationStats 

pub struct CoactivationStats {
    pub n_obs: usize,
    pub n_a: usize,
    pub n_b: usize,
    pub n_joint: usize,
    pub p_a_given_b: f64,
    pub p_b_given_a: f64,
    pub lift: f64,
    pub weight_correlation: f64,
}

Pairwise co-activation summary for two atoms (see SparseAtomCodes::coactivation). All probabilities are empirical popcount ratios over the active-support masks.

Fields

n_obs: usize

Total number of observations the codes cover.

n_a: usize

Rows where atom a is active.

n_b: usize

Rows where atom b is active.

n_joint: usize

Rows where both are active.

p_a_given_b: f64

P(a active | b active); 0 when b is never active.

p_b_given_a: f64

P(b active | a active); 0 when a is never active.

lift: f64

P(a∧b) / (P(a)·P(b)); 1 for independent atoms, 0 when either marginal is empty.

weight_correlation: f64

Pearson correlation of the two atoms’ activation WEIGHTS over the jointly-active rows (see SparseAtomCodes::weight_codependence) — the amplitude/interaction half of the fusion criterion. 0 when the joint support is too small or a weight is constant there.

Implementations

impl CoactivationStats

pub fn dependence(&self) -> f64

Symmetric code dependence min(P(a|b), P(b|a)) — the canonical-order trigger for FUSION proposals (descending). Near 0 for independent or disjoint atoms; near 1 only when the two supports essentially coincide, which is the shattering signature.

pub fn absorption_asymmetry(&self) -> f64

Conditional asymmetry |P(a|b) − P(b|a)| — large when one atom’s support nests inside the other’s (the A⇒B absorption signature, where P(parent|child) ≈ 1 but not conversely). Flags the pair for a targeted within-atom substructure audit; it is never itself an acceptance criterion.

pub fn fusion_evidence(&self) -> f64

#976 — the combined FUSION evidence: dependence · |weight_correlation|. A fusion proposal needs BOTH halves — the atoms must co-activate (support dependence) AND their amplitudes must be dependent on the joint support (the interaction evidence that a single curved family was shattered). Two independent atoms that happen to co-fire score near 0 on the second factor and so are NOT proposed for fusion even at high support overlap; a genuine shattered pair scores high on both. This is the scalar the canonical-order fusion ranking (“fusions by code dependence descending”) should sort on, and the threshold the e-process acceptance gate guards.

Trait Implementations

impl Clone for CoactivationStats

fn clone(&self) -> CoactivationStats

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for CoactivationStats

impl Debug for CoactivationStats

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

Formats the value using the given formatter.

impl PartialEq for CoactivationStats

fn eq(&self, other: &CoactivationStats) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for CoactivationStats