Struct DeterministicPolicy 

pub struct DeterministicPolicy { /* private fields */ }

Deterministic policy wrapper for DDPG/TD3.

Holds only the action bounds and exploration noise configuration; the network parameters (μ_θ(s)) are managed externally.

Implementations

impl DeterministicPolicy

pub fn new(action_dim: usize) -> Self

Create a policy with symmetric action bounds [-1, 1].

pub fn with_bounds(action_dim: usize, action_low: f32, action_high: f32) -> Self

Create a policy with custom action bounds.

pub fn action_dim(&self) -> usize

Number of action dimensions.

pub fn clip_action(&self, action: &[f32]) -> RlResult<Vec<f32>>

Clip action to [action_low, action_high].

Errors

• RlError::DimensionMismatch if action.len() != action_dim.

pub fn exploration_action( &self, action: &[f32], sigma: f32, handle: &mut RlHandle, ) -> RlResult<Vec<f32>>

Add Gaussian exploration noise and clip.

Returns clip(action + N(0, σ²), low, high).

Errors

• RlError::DimensionMismatch if action.len() != action_dim.

pub fn smooth_target_action( &self, action: &[f32], sigma: f32, clip_c: f32, handle: &mut RlHandle, ) -> RlResult<Vec<f32>>

TD3 target policy smoothing: add clipped noise to target actions.

ã = clip(action + clip(N(0, σ²), -c, c), low, high)

Errors

• RlError::DimensionMismatch if action.len() != action_dim.

Trait Implementations

impl Clone for DeterministicPolicy

fn clone(&self) -> DeterministicPolicy

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DeterministicPolicy

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

Formats the value using the given formatter.