Struct ObservationNormalizer 

pub struct ObservationNormalizer { /* private fields */ }

Observation normalizer with running statistics.

Implementations

impl ObservationNormalizer

pub fn new(obs_dim: usize) -> Self

Create a normalizer for observations of dimension obs_dim.

Default: enabled, clip = 5.0, update_stats = true.

pub fn with_no_update(self) -> Self

Disable statistics update (useful at evaluation time).

pub fn with_clip(self, clip: f32) -> Self

Set the clip range.

pub fn disable(&mut self)

Disable normalisation (pass-through mode).

pub fn enable(&mut self)

Enable normalisation.

pub fn count(&self) -> u64

Number of observations seen.

pub fn process(&mut self, batch: &[f32]) -> RlResult<Vec<f32>>

Process a batch of observations: optionally update stats, normalise, clip.

• batch — [B × obs_dim] flat slice.

Returns a normalised [B × obs_dim] vector.

Errors

• RlError::DimensionMismatch if batch.len() % obs_dim != 0.

pub fn process_one(&mut self, obs: &[f32]) -> RlResult<Vec<f32>>

Process a single observation.

Errors

• RlError::DimensionMismatch if obs.len() != obs_dim.

pub fn stats(&self) -> &RunningStats

Access the underlying running statistics.

Trait Implementations

impl Clone for ObservationNormalizer

fn clone(&self) -> ObservationNormalizer

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ObservationNormalizer

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

Formats the value using the given formatter.