Struct QLearningOptimizer 

pub struct QLearningOptimizer { /* private fields */ }

Q-learning agent for circuit optimization

Implementations

impl QLearningOptimizer

pub fn new( learning_rate: f64, discount_factor: f64, initial_epsilon: f64, ) -> Self

Create a new Q-learning optimizer

Arguments

• learning_rate - How quickly the agent learns (0.0 to 1.0)
• discount_factor - How much future rewards matter (0.0 to 1.0)
• initial_epsilon - Initial exploration rate (0.0 to 1.0)

pub fn choose_action( &self, state: &CircuitState, available_actions: &[OptimizationAction], ) -> OptimizationAction

Choose action using epsilon-greedy policy

Arguments

• state - Current circuit state
• available_actions - List of actions that can be taken

pub fn update_q_value( &mut self, state: &CircuitState, action: OptimizationAction, reward: f64, next_state: &CircuitState, next_actions: &[OptimizationAction], )

Update Q-value based on observed reward

Arguments

• state - Previous state
• action - Action taken
• reward - Reward received
• next_state - New state after action
• next_actions - Actions available in next state

pub fn calculate_reward( &self, old_state: &CircuitState, new_state: &CircuitState, ) -> f64

Calculate reward for a state transition

Reward is based on improvements in circuit metrics:

• Reduced depth (+reward)
• Reduced gate count (+reward)
• Increased fidelity (+reward)

pub fn finish_episode(&mut self, episode: OptimizationEpisode)

Complete an optimization episode

pub fn get_statistics(&self) -> OptimizationStatistics

Get optimization statistics

pub fn save_q_table(&self, path: &str) -> QuantRS2Result<()>

Save Q-table to file

pub fn load_q_table(&mut self, path: &str) -> QuantRS2Result<()>

Load Q-table from file

Trait Implementations

impl Default for QLearningOptimizer

fn default() -> Self

Returns the “default value” for a type.