abstract_stepper

RL4CRN.agent2env_interface.abstract_stepper

Abstract stepper interface.

Defines the minimal interface for a stepper, i.e., a component responsible for updating an environment state given an action. In this project, steppers are used to interface reinforcement learning agents with CRN environments.

A concrete implementation should implement step and encode the rule that transforms a (state, action) pair into an updated state.

AbstractStepper

Bases: Protocol

Protocol for steppers that advance an environment state given an action.

step(state, action)

Advance the environment state by applying an action.

PARAMETER	DESCRIPTION
state	Environment state to be updated.
action	Action to apply to the state.

RETURNS	DESCRIPTION
	Implementations may either mutate state in-place and return None, or return an updated state object. Concrete steppers should document their chosen behavior.