Struct StructureLedger 

pub struct StructureLedger { /* private fields */ }

The dictionary’s claim ledger: every structural claim the discovery stack makes, each with its own e-process. Serializable — evidence resumes across corpus shards (#973) by persisting the ledger, not by refitting. Calling StructureLedger::certify at ANY data-dependent stopping time yields a valid certificate; that is the entire point.

Implementations

impl StructureLedger

pub fn new() -> Self

pub fn register(&mut self, kind: ClaimKind) -> usize

Register a claim and return its ledger index. Idempotent on the claim kind: re-registering an existing claim (a resumed shard loop re-announcing its claim surface) returns the existing index and PRESERVES its accumulated evidence — a fresh e-process here would silently discard the stream’s history.

pub fn absorb_log(&mut self, idx: usize, log_e_value: f64) -> Result<(), String>

Absorb one conditionally-valid log e-value for claim idx (a universal-inference shard ratio, a frozen-prior log-BF — the caller’s contract is per-source, documented on the producing gate).

pub fn claims(&self) -> &[StructuralClaim]

pub fn absorb_probe_outcome( &mut self, idx: usize, log_lik_alt_on_outcome: f64, log_lik_null_on_outcome: f64, ) -> Result<(), String>

The likelihood half of the probe-design loop (work-plan step 4): after running a planned probe (ProbePlan → crate::inference::steering::steer_delta), evaluate the REALIZED outcomes under both hypotheses’ predictive densities and absorb the log-ratio into the contested claim’s e-process.

Validity contract: both predictive densities must be FROZEN before the probe outcome is observed — which the design loop satisfies by construction, since both hypotheses’ dictionaries were fitted before the probe was even chosen. For a composite null, the null density must be the honest constrained fit (the same rule as split_likelihood_log_e_value, which this delegates to); for a simple null the predictive density is the sup. Probe outcomes are new data by construction (the model was steered to produce them), so they compound validly with the claim’s prior shard evidence.

pub fn certify(&self, alpha: f64) -> StructureCertificate

The dictionary certificate: e-BH over the ledger’s CURRENT e-values at level α. FDR ≤ α over the confirmed set under arbitrary dependence — atoms sharing every token is fine — and valid at any stopping time because each entry is an e-process. Claims not confirmed are CONTESTED, never rejected (demote-never-reject); they keep their evidence and are the inputs to the probe-design loop.

Trait Implementations

impl Clone for StructureLedger

fn clone(&self) -> StructureLedger

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StructureLedger

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

Formats the value using the given formatter.

impl Default for StructureLedger

fn default() -> StructureLedger

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for StructureLedger

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for StructureLedger

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.