Struct DeliberationEngine 

pub struct DeliberationEngine {
    pub target_confidence: f32,
    pub max_rounds: usize,
    pub alpha: f32,
}

Multi-round evidence accumulation engine.

Models how an AI agent should reason under uncertainty: instead of forcing a binary guess from thin evidence, hold at State 0 and keep gathering signals until the confidence threshold is crossed or rounds run out.

Each round adds new evidence (a slice of f32 signals). The engine uses an exponential moving average so recent evidence weighs more than stale data.

Fields

target_confidence: f32

Confidence required to declare convergence (0.0–1.0).

max_rounds: usize

Maximum rounds before returning with whatever confidence was reached.

alpha: f32

Recency weight (0 < α ≤ 1). Lower α = more memory of past rounds.

Implementations

impl DeliberationEngine

pub fn new(target_confidence: f32, max_rounds: usize) -> Self

pub fn with_alpha(self, alpha: f32) -> Self

pub fn run(&self, rounds_evidence: Vec<Vec<f32>>) -> DeliberationResult

Run deliberation. rounds_evidence[i] is the evidence for round i. Missing rounds receive no new evidence (engine holds).