Struct T3 

pub struct T3 { /* private fields */ }

A 3-dimensional trust tensor with fractal sub-dimension support

Implementations

impl T3

pub fn new() -> Self

Create a new T3 with neutral trust (0.5) and zero confidence

pub fn with_scores(scores: [f64; 3]) -> Result<Self>

Create a T3 with specific initial root scores

pub fn from_parts(scores: [f64; 3], observation_counts: [u64; 3]) -> Self

Reconstruct a T3 from persisted parts: root scores + per-dimension observation counts. Weights are recomputed from the counts using the same logarithmic formula as T3::apply_delta/T3::observe, so a tensor that is saved and re-loaded keeps its confidence without having to replay its observation history. Scores are clamped to [0, 1].

This is the persistence-round-trip constructor used by at-rest stores (e.g. sealed EntityTrust files): the serialized form carries the three root scores plus their observation counts, and this rebuilds the exact tensor state. Sub-dimensions are not part of the persisted parts and are initialized empty.

pub fn score(&self, dimension: TrustDimension) -> f64

Get the score for a root dimension

pub fn weight(&self, dimension: TrustDimension) -> f64

Get the weight (confidence) for a root dimension

pub fn observation_counts(&self) -> &[u64; 3]

Get all per-dimension observation counts.

Exposes the raw evidence counts so persistence layers can serialize and later restore confidence via T3::from_parts. Weights are a pure function of these counts, so this is sufficient to round-trip state.

pub fn scores(&self) -> &[f64; 3]

Get all root dimension scores

pub fn weights(&self) -> &[f64; 3]

Get all weights

pub fn sub_dimensions(&self) -> &HashMap<String, SubDimensionScore>

Get sub-dimensions map

pub fn apply_delta(&mut self, dimension: TrustDimension, delta: f64) -> f64

Record an observation for a root dimension

Uses exponential moving average with decay factor based on observation count Apply a signed reputation delta directly to a dimension, clamped to [0, 1], counting it as one observation. This is how a ReputationDelta from an R7 action outcome — e.g. a missed deadline debiting Temperament — folds into the tensor. Returns the realized change after clamping.

pub fn observe( &mut self, dimension: TrustDimension, observed_score: f64, ) -> Result<()>

pub fn observe_sub_dimension( &mut self, name: &str, parent: TrustDimension, observed_score: f64, ) -> Result<()>

Record an observation for a sub-dimension

Sub-dimensions are keyed by name and linked to a root dimension. Uses the same EMA math as root dimensions.

pub fn aggregate(&self) -> f64

Compute the aggregate trust score (weighted geometric mean)

Geometric mean ensures that a zero in any dimension zeros the total, reflecting that trust requires all dimensions to be positive.

pub fn distance(&self, other: &T3) -> f64

Compute Euclidean distance to another T3

pub fn merge(&self, other: &T3) -> Self

Merge with another T3 using weighted average

The merge is weighted by the observation counts in each tensor

pub fn decay(&mut self, decay_factor: f64)

Apply time decay to the tensor

Trust that isn’t reinforced decays toward neutral (0.5) over time. The decay_factor should be in (0, 1), where smaller = faster decay.

pub fn meets_thresholds(&self, min_scores: &[f64; 3]) -> bool

Check if trust meets minimum thresholds

Trait Implementations

impl Clone for T3

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for T3

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

Formats the value using the given formatter.

impl Default for T3

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for T3

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for T3

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

where __S: Serializer,

Serialize this value into the given Serde serializer.