Trait relearn::envs::StructuredEnvironment

pub trait StructuredEnvironment: EnvStructure + Environment<Observation = <Self::ObservationSpace as Space>::Element, Action = <Self::ActionSpace as Space>::Element, Feedback = <Self::FeedbackSpace as Space>::Element> { }

A reinforcement learning Environment with consistent EnvStructure.

Design Discussion

EnvStructure is not a supertrait of Environment because knowing the observation and action spaces is not necessary for simulation, only the observation and action types must be known.

Implementors

impl<T> StructuredEnvironment for T where
    T: EnvStructure + Environment<Observation = <Self::ObservationSpace as Space>::Element, Action = <Self::ActionSpace as Space>::Element, Feedback = <Self::FeedbackSpace as Space>::Element> + ?Sized,