Struct border::env::py_gym_env::vec::PyVecGymEnv

pub struct PyVecGymEnv<O, A, OF, AF> { /* fields omitted */ }

A vectorized environment using multiprocess module in Python. The code is adapted from tch-rs RL example.

Implementations

impl<O, A, OF, AF> PyVecGymEnv<O, A, OF, AF> where
    O: Obs,
    A: Act,
    OF: PyGymEnvObsFilter<O>,
    AF: PyGymEnvActFilter<A>, 

pub fn new(
    name: &str,
    n_procs: usize,
    obs_filter: OF,
    act_filter: AF,
    atari_wrapper: bool
) -> PyResult<Self>

Constructs a vectorized environment.

This function invl

• name - Name of a gym environment.
• atari_wrapper - If true, wrap_deepmind() is applied to the environment. See atari_wrapper.py.

pub fn close(&self)

Close all subprocesses.

TODO: Consider implementing the method in Drop trait.

Trait Implementations

impl<O: Clone, A: Clone, OF: Clone, AF: Clone> Clone for PyVecGymEnv<O, A, OF, AF>

fn clone(&self) -> PyVecGymEnv<O, A, OF, AF>

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<O: Debug, A: Debug, OF: Debug, AF: Debug> Debug for PyVecGymEnv<O, A, OF, AF>

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

Formats the value using the given formatter.

impl<O, A, OF, AF> Env for PyVecGymEnv<O, A, OF, AF> where
    O: Obs,
    A: Act,
    OF: PyGymEnvObsFilter<O>,
    AF: PyGymEnvActFilter<A>, 

type Obs = O

Observation of the environment.

type Act = A

Action of the environment.

type Info = PyGymInfo

Information in the self::Step object.

fn reset(&mut self, is_done: Option<&Vec<i8>>) -> Result<O, Box<dyn Error>>

Resets the environment, the obs/act filters and returns the observation tensor.

If is_done is None, all environemnts are resetted. Otherwise, is_done is Vec<f32> and environments with is_done[i] == 1.0 are resetted.

fn step(&mut self, a: &A) -> (Step<Self>, Record)

Performes an interaction step.