Struct EmlCoherenceModel 

pub struct EmlCoherenceModel { /* private fields */ }

Depth-4 multi-head EML master formula for O(1) coherence prediction.

This is a thin wrapper around eml_core::EmlModel that provides the WeftOS-specific GraphFeatures and CoherencePrediction types.

Backward compatible: supports both depth-3 (34 params, single output) and depth-4 (50 params, multi-head) architectures.

Implementations

impl EmlCoherenceModel

pub fn new() -> Self

Create a new untrained depth-4 multi-head model with zeroed parameters.

pub fn new_v1() -> Self

Create a new untrained depth-3 legacy model (34 params).

pub fn drain_events(&mut self) -> Vec<EmlEvent>

Drain accumulated EML lifecycle events for ExoChain forwarding.

pub fn is_multi_head(&self) -> bool

Whether this model uses the depth-4 multi-head architecture.

pub fn is_trained(&self) -> bool

Whether the model has been trained to convergence.

pub fn training_sample_count(&self) -> usize

Number of training samples collected so far.

pub fn mean_error(&self) -> f64

Mean of the recent error history (empty => 0.0).

pub fn predict(&self, features: &GraphFeatures) -> CoherencePrediction

O(1) multi-head coherence prediction from graph features.

Returns a CoherencePrediction with lambda_2, fiedler_norm, and uncertainty. Falls back to a density-based estimate if untrained. For depth-3 legacy models, fiedler_norm and uncertainty are synthetic estimates derived from lambda_2.

pub fn predict_lambda2(&self, features: &GraphFeatures) -> f64

Convenience: returns only the primary lambda_2 value.

Use this when you only need the algebraic connectivity scalar (backward compatible with callers that expected f64).

pub fn record(&mut self, features: GraphFeatures, lambda_2: f64)

Record a training point (called after every exact Lanczos computation).

Only records lambda_2; use [record_full] to also supply Fiedler norm and uncertainty ground truth.

pub fn record_full( &mut self, features: GraphFeatures, lambda_2: f64, fiedler_norm: Option<f64>, uncertainty: Option<f64>, )

Record a full training point with optional Fiedler norm and uncertainty.

pub fn train(&mut self) -> bool

Train the model when enough data is collected.

Uses random restart + coordinate descent (gradient-free optimization suitable for 50 parameters).

Returns true if the model converged (MSE < 0.01).

Trait Implementations

impl Clone for EmlCoherenceModel

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 EmlCoherenceModel

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

Formats the value using the given formatter.

impl Default for EmlCoherenceModel

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for EmlCoherenceModel

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for EmlCoherenceModel

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

where __S: Serializer,

Serialize this value into the given Serde serializer.