Action Space

reinforcement-learning

Definition

Action Space

An action space $A$ is the set of all actions an agent is allowed to choose from in an environment.

At each time step $t$, the agent observes a state $s_t\in S$ and selects an action

$$a_t\in A.$$

More generally, if the available actions depend on the current state, the action space may be state-dependent:

$$A(s)\subseteq A,$$

where $A(s)$ is the set of actions available in state $s$.