Struct PrioritizedReplayBuffer 

pub struct PrioritizedReplayBuffer { /* private fields */ }

Prioritized experience replay buffer backed by a sum-tree.

Implements proportional prioritization with importance-sampling weights.

Implementations

impl PrioritizedReplayBuffer

pub fn new( capacity: usize, obs_dim: usize, act_dim: usize, alpha: f64, beta: f64, ) -> Self

Create a new prioritized replay buffer.

• alpha — prioritization exponent (0 = uniform, 1 = full prioritization)
• beta — importance-sampling correction exponent (1 = full correction)

pub fn len(&self) -> usize

Number of valid transitions stored.

pub fn is_empty(&self) -> bool

Whether the buffer is empty.

pub fn push_slices( &mut self, obs: &[f32], next_obs: &[f32], action: &[f32], reward: f32, terminated: bool, truncated: bool, priority: f64, ) -> Result<(), RloxError>

Push a transition with the given TD-error priority.

pub fn push( &mut self, record: ExperienceRecord, priority: f64, ) -> Result<(), RloxError>

pub fn set_beta(&mut self, beta: f64)

Set the importance-sampling beta parameter.

pub fn sample( &self, batch_size: usize, seed: u64, ) -> Result<PrioritizedSampledBatch, RloxError>

Sample a batch with importance-sampling weights.

pub fn update_priorities( &mut self, indices: &[usize], priorities: &[f64], ) -> Result<(), RloxError>

Update priorities for previously sampled indices.

impl PrioritizedReplayBuffer

pub fn update_priorities_from_loss( &mut self, indices: &[usize], losses: &[f64], epsilon: f64, ) -> Result<(), RloxError>

Update priorities from raw loss values using LAP.

Convenience method: computes priority = |loss| + epsilon for each index, then calls update_priorities.

Trait Implementations

impl Debug for PrioritizedReplayBuffer

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

Formats the value using the given formatter.