Enum InferenceMethod 

pub enum InferenceMethod {
    DirectLookup,
    MajorityVote,
    CitationLink,
    AnalogyInference,
    PatternSummarize,
    ArchitecturalChain,
    DominanceAnalysis,
    EntityCount,
    IntervalCalc,
    FeedbackConsolidation,
    QualitativeInference,
    ProvenanceChain,
    MultiSourceConsensus,
    ConflictReconciliation,
}

One of the 14 registered inference methods. Every Inferential memory’s method field resolves to a symbol whose canonical name maps to one of these variants.

Variants

DirectLookup

Pass-through from a single parent. output.conf = parent.conf.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let out = InferenceMethod::DirectLookup
    .compute(&[Confidence::try_from_f32(0.73).unwrap()])
    .unwrap();
assert!((out.as_f32() - 0.73).abs() < 0.001);

MajorityVote

Majority vote across an odd number of parents (N ≥ 3). Per the spec v1 convention (librarian-pipeline.md § 5.1) the write surface carries only voters-in-favor in derived_from, so votes_for == N and the formula collapses to min(parents.conf).

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::MajorityVote.compute(&[p(0.9), p(0.6), p(0.7)]).unwrap();
assert!((out.as_f32() - 0.6).abs() < 0.001);

CitationLink

Citation-linked pair of parents. output.conf = min(parents.conf) * 0.9.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::CitationLink.compute(&[p(1.0), p(1.0)]).unwrap();
assert!((out.as_f32() - 0.9).abs() < 0.001);

AnalogyInference

Analogical mapping from source to target. output.conf = product(parents.conf) * 0.7.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::AnalogyInference.compute(&[p(1.0), p(1.0)]).unwrap();
assert!((out.as_f32() - 0.7).abs() < 0.001);

PatternSummarize

Geometric-mean summarization over N ≥ 2 parents. output.conf = geomean(parents.conf) * 0.8.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
// geomean(0.5, 0.5) = 0.5; * 0.8 = 0.4.
let out = InferenceMethod::PatternSummarize.compute(&[p(0.5), p(0.5)]).unwrap();
assert!((out.as_f32() - 0.4).abs() < 0.001);

ArchitecturalChain

Chained architectural derivation. output.conf = product(parents.conf).

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
// 0.5 * 0.5 * 0.5 = 0.125.
let out = InferenceMethod::ArchitecturalChain
    .compute(&[p(0.5), p(0.5), p(0.5)]).unwrap();
assert!((out.as_f32() - 0.125).abs() < 0.001);

DominanceAnalysis

Dominance / ranking analysis. output.conf = min(parents.conf) * 0.6.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
// min(0.9, 0.5) * 0.6 = 0.3.
let out = InferenceMethod::DominanceAnalysis.compute(&[p(0.9), p(0.5)]).unwrap();
assert!((out.as_f32() - 0.3).abs() < 0.001);

EntityCount

Cardinality / count over N ≥ 1 parents. output.conf = min(parents.conf) * 0.8.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::EntityCount.compute(&[p(0.7)]).unwrap();
assert!((out.as_f32() - 0.56).abs() < 0.001); // 0.7 * 0.8.

IntervalCalc

Interval calculation between two endpoint parents. output.conf = min(parents.conf) * 0.9.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::IntervalCalc.compute(&[p(0.8), p(0.6)]).unwrap();
assert!((out.as_f32() - 0.54).abs() < 0.001); // min=0.6; *0.9.

FeedbackConsolidation

Feedback consolidation over N ≥ 1 parents. output.conf = min(parents.conf) * 0.85.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::FeedbackConsolidation.compute(&[p(0.6)]).unwrap();
assert!((out.as_f32() - 0.51).abs() < 0.001); // 0.6 * 0.85.

QualitativeInference

Qualitative / narrative inference. output.conf = min(parents.conf) * 0.5.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::QualitativeInference.compute(&[p(0.8)]).unwrap();
assert!((out.as_f32() - 0.4).abs() < 0.001);

ProvenanceChain

Provenance chain through N ≥ 2 parents. output.conf = product(parents.conf).

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
let out = InferenceMethod::ProvenanceChain.compute(&[p(0.9), p(0.8)]).unwrap();
assert!((out.as_f32() - 0.72).abs() < 0.001);

MultiSourceConsensus

Noisy-OR consensus across independent parents. output.conf = 1 - product(1 - parents.conf).

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
// 1 - (0.5)(0.5) = 0.75.
let out = InferenceMethod::MultiSourceConsensus.compute(&[p(0.5), p(0.5)]).unwrap();
assert!((out.as_f32() - 0.75).abs() < 0.001);

ConflictReconciliation

Conflict reconciliation over contested peers. output.conf = max(parents.conf) * 0.8.

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;
let p = |f| Confidence::try_from_f32(f).unwrap();
// max(0.3, 0.9) * 0.8 = 0.72.
let out = InferenceMethod::ConflictReconciliation.compute(&[p(0.3), p(0.9)]).unwrap();
assert!((out.as_f32() - 0.72).abs() < 0.001);

Implementations

impl InferenceMethod

pub fn symbol_name(self) -> &'static str

The canonical symbol name (with leading @) for this method.

pub fn from_symbol_name(name: &str) -> Option<Self>

Resolve a canonical method name (with or without leading @) to a variant, or None if the name is not a registered method.

pub fn parent_count_rule(self) -> ParentCountRule

Parent-count rule expected by this method.

pub fn staleness_rule(self) -> StalenessRule

Staleness predicate per spec § 5.1.

pub fn compute( self, parents: &[Confidence], ) -> Result<Confidence, InferenceMethodError>

Compute the output confidence for this method given parent confidences. See variant docs for each formula.

Errors

• InferenceMethodError::WrongParentCount if the supplied count violates the method’s ParentCountRule.
• InferenceMethodError::TooManyParents if parents.len() > 8 for methods whose formulas require a joint product (ArchitecturalChain, ProvenanceChain, AnalogyInference, PatternSummarize, MultiSourceConsensus). The u128 intermediate accommodates up to eight 16-bit factors; larger N is reserved for a log-table follow-up.

Example

use mimir_core::inference_methods::InferenceMethod;
use mimir_core::Confidence;

let parents = [Confidence::ONE, Confidence::ONE];
let out = InferenceMethod::CitationLink.compute(&parents).unwrap();
// min(1.0, 1.0) * 0.9 ≈ 0.9; allow ±1 fixed-point step of drift.
let expected = (f32::from(u16::MAX) * 0.9) as u16;
assert!((i32::from(out.as_u16()) - i32::from(expected)).abs() <= 1);

Trait Implementations

impl Clone for InferenceMethod

fn clone(&self) -> InferenceMethod

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for InferenceMethod

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

Formats the value using the given formatter.

impl Hash for InferenceMethod

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for InferenceMethod

fn eq(&self, other: &InferenceMethod) -> 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 Copy for InferenceMethod

impl Eq for InferenceMethod

impl StructuralPartialEq for InferenceMethod