Struct PolicyMemoryStore 

pub struct PolicyMemoryStore<'a> { /* private fields */ }

Policy Memory Store interface for AI agent policy memory

This interface provides Q-learning state-action lookups, contextual bandit policy retrieval, and episodic memory for reasoning.

Example

let policy_store = db.policy_memory();
policy_store.store_policy("state_a", vec![0.1, 0.2], PolicyAction { action: "move_left", reward: 0.8 })?;
let similar = policy_store.retrieve_similar_states(&current_state_embedding, 5)?;

Implementations

impl<'a> PolicyMemoryStore<'a>

pub fn new(db: &'a AgenticDB) -> Self

Create a new policy memory store interface

pub fn store_policy( &self, state_id: &str, state_embedding: Vec<f32>, action: &str, reward: f64, q_value: f64, ) -> Result<String>

Store a policy entry (state-action pair)

pub fn retrieve_similar_states( &self, state_embedding: &[f32], k: usize, ) -> Result<Vec<PolicyEntry>>

Retrieve similar states for policy lookup

pub fn get_best_action( &self, state_embedding: &[f32], k: usize, ) -> Result<Option<String>>

Get the best action for a state based on Q-values

pub fn update_q_value(&self, policy_id: &str, new_q_value: f64) -> Result<()>

Update Q-value for a state-action pair