Struct SnapshotBase 

pub struct SnapshotBase<const R: usize, ObservationType: Observation<R>, RewardType: Reward> {
    pub observation: ObservationType,
    pub reward: RewardType,
    pub status: EpisodeStatus,
}

Default snapshot implementation for standard reinforcement learning observations.

SnapshotBase stores an observation, reward, and EpisodeStatus. Construct via the named constructors running, terminated, or truncated.

Type Parameters

• R - The observation tensor rank
• ObservationType - The type of observation (must implement Observation<R>)
• RewardType - The type of reward (must implement Reward)

Fields

observation: ObservationType

The observation derived from the state.

reward: RewardType

The reward received from the last action.

status: EpisodeStatus

Episode lifecycle status.

Implementations

impl<const R: usize, ObservationType: Observation<R>, RewardType: Reward> SnapshotBase<R, ObservationType, RewardType>

pub fn running(observation: ObservationType, reward: RewardType) -> Self

Snapshot for a step where the episode is still running.

pub fn terminated(observation: ObservationType, reward: RewardType) -> Self

Snapshot for the step on which the MDP reached a terminal state.

pub fn truncated(observation: ObservationType, reward: RewardType) -> Self

Snapshot for the step on which an external step limit was reached.

Trait Implementations

impl<const R: usize, ObservationType: Clone + Observation<R>, RewardType: Clone + Reward> Clone for SnapshotBase<R, ObservationType, RewardType>

fn clone(&self) -> SnapshotBase<R, ObservationType, RewardType>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<const R: usize, ObservationType: Debug + Observation<R>, RewardType: Debug + Reward> Debug for SnapshotBase<R, ObservationType, RewardType>

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

Formats the value using the given formatter.

impl<const R: usize, ObservationType: Observation<R>, RewardType: Reward> Snapshot<R> for SnapshotBase<R, ObservationType, RewardType>

type ObservationType = ObservationType

The observation type exposed to the agent at each step.

type RewardType = RewardType

The type of reward contained in this snapshot.

fn observation(&self) -> &Self::ObservationType

Access the observed state.

fn reward(&self) -> &Self::RewardType

Access the reward received.

fn status(&self) -> EpisodeStatus

Episode lifecycle status for this step.

fn is_done(&self) -> bool

true when the episode loop should stop.

fn is_terminated(&self) -> bool

true only for intrinsic MDP termination.

fn is_truncated(&self) -> bool

true only for extrinsic step-limit truncation.

fn metadata(&self) -> Option<&SnapshotMetadata>

Optional named reward components and position data.