Struct EvalEngine 

pub struct EvalEngine {
    pub config: EvalConfig,
}

Main evaluation engine for memory quality assessment.

Fields

config: EvalConfig

Configuration.

Implementations

impl EvalEngine

pub fn new() -> Self

Creates a new evaluation engine with default configuration.

pub fn with_config(config: EvalConfig) -> Self

Creates a new evaluation engine with custom configuration.

pub fn evaluate( &self, memories: &[Memory], query_results: &[Vec<QuantumMeasurementResult>], identity_patterns: Option<&[HabituatedPattern]>, ) -> Result<EvalReport, EvalError>

Evaluates a collection of memories across all 7 dimensions.

Parameters

• memories: The stored memories to evaluate
• query_results: Recent recall query results for retrieval evaluation
• identity_patterns: Current habituated patterns from the identity knot

pub fn evaluate_retrieval( &self, query_results: &[Vec<QuantumMeasurementResult>], ) -> Result<DimensionScore, EvalError>

Evaluates retrieval accuracy across query results.

Measures: proportion of queries that returned at least one result above the retrieval threshold.

pub fn evaluate_summarization( &self, memories: &[Memory], ) -> Result<DimensionScore, EvalError>

Evaluates summarization quality by measuring consolidation coverage.

Measures: ratio of semantic-classified memories to total memories. A well-consolidated memory store should have a healthy proportion of semantic (abstracted) vs. raw episodic memories.

pub fn evaluate_isolation( &self, memories: &[Memory], ) -> Result<DimensionScore, EvalError>

Evaluates source isolation quality.

Measures: average isolation score across all memories. High isolation = memories from different sources don’t interfere. Low isolation = source contamination risk.

pub fn evaluate_inference( &self, memories: &[Memory], ) -> Result<DimensionScore, EvalError>

Evaluates inference capability through verifiability distribution.

Measures: proportion of memories at Verified/Durable stage. A memory system with strong inference will have promoted many memories to higher verifiability through reasoning.

pub fn evaluate_reproduction( &self, memories: &[Memory], ) -> Result<DimensionScore, EvalError>

Evaluates reproduction fidelity through procedural memory coverage.

Measures: ratio of procedural-classified memories to total. Procedural memories are the “reproduction” dimension — they encode how to do things, not just what was done.

pub fn evaluate_learning( &self, memories: &[Memory], ) -> Result<DimensionScore, EvalError>

Evaluates learning capability through memory diversity and growth.

Measures: memory class diversity (entropy over class distribution) and the ratio of cross-scope memories (agent vs user vs session).

pub fn evaluate_habituation( &self, patterns: Option<&[HabituatedPattern]>, ) -> Result<DimensionScore, EvalError>

Evaluates habituation through pattern stability analysis.

Measures: proportion of habituated patterns vs. total patterns, and average stability of habituated patterns.

Trait Implementations

impl Default for EvalEngine

fn default() -> Self

Returns the “default value” for a type.